{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.999982186436753, "eval_steps": 500, "global_step": 28068, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 3.562712649411262e-05, "grad_norm": 27.79722746179553, "learning_rate": 2.3724792408066433e-08, "loss": 1.4398, "step": 1 }, { "epoch": 7.125425298822524e-05, "grad_norm": 21.195375529370004, "learning_rate": 4.7449584816132866e-08, "loss": 1.3401, "step": 2 }, { "epoch": 0.00010688137948233786, "grad_norm": 17.3766687006042, "learning_rate": 7.117437722419929e-08, "loss": 1.4305, "step": 3 }, { "epoch": 0.00014250850597645047, "grad_norm": 20.22205324226869, "learning_rate": 9.489916963226573e-08, "loss": 1.4024, "step": 4 }, { "epoch": 0.0001781356324705631, "grad_norm": 23.752746665580418, "learning_rate": 1.1862396204033216e-07, "loss": 1.3616, "step": 5 }, { "epoch": 0.0002137627589646757, "grad_norm": 23.590268689650852, "learning_rate": 1.4234875444839858e-07, "loss": 1.3262, "step": 6 }, { "epoch": 0.0002493898854587883, "grad_norm": 21.155313918949382, "learning_rate": 1.66073546856465e-07, "loss": 1.3875, "step": 7 }, { "epoch": 0.00028501701195290095, "grad_norm": 25.327032504873717, "learning_rate": 1.8979833926453146e-07, "loss": 1.3898, "step": 8 }, { "epoch": 0.00032064413844701354, "grad_norm": 18.70291669438463, "learning_rate": 2.135231316725979e-07, "loss": 1.3442, "step": 9 }, { "epoch": 0.0003562712649411262, "grad_norm": 23.961599024211587, "learning_rate": 2.3724792408066432e-07, "loss": 1.3692, "step": 10 }, { "epoch": 0.0003918983914352388, "grad_norm": 19.658999231583845, "learning_rate": 2.6097271648873075e-07, "loss": 1.3548, "step": 11 }, { "epoch": 0.0004275255179293514, "grad_norm": 23.300155328494853, "learning_rate": 2.8469750889679715e-07, "loss": 1.3739, "step": 12 }, { "epoch": 0.000463152644423464, "grad_norm": 19.44716671758718, "learning_rate": 3.084223013048636e-07, "loss": 1.312, "step": 13 }, { "epoch": 0.0004987797709175766, "grad_norm": 24.535073271948455, "learning_rate": 3.3214709371293e-07, "loss": 1.3433, "step": 14 }, { "epoch": 0.0005344068974116893, "grad_norm": 21.723142027469734, "learning_rate": 3.558718861209965e-07, "loss": 1.2934, "step": 15 }, { "epoch": 0.0005700340239058019, "grad_norm": 21.397743488276745, "learning_rate": 3.795966785290629e-07, "loss": 1.2556, "step": 16 }, { "epoch": 0.0006056611503999145, "grad_norm": 23.55720266287107, "learning_rate": 4.0332147093712933e-07, "loss": 1.2326, "step": 17 }, { "epoch": 0.0006412882768940271, "grad_norm": 20.197348707140616, "learning_rate": 4.270462633451958e-07, "loss": 1.3161, "step": 18 }, { "epoch": 0.0006769154033881398, "grad_norm": 18.355288423550192, "learning_rate": 4.507710557532622e-07, "loss": 1.2523, "step": 19 }, { "epoch": 0.0007125425298822524, "grad_norm": 22.59010545479618, "learning_rate": 4.7449584816132864e-07, "loss": 1.16, "step": 20 }, { "epoch": 0.000748169656376365, "grad_norm": 20.867448439266216, "learning_rate": 4.982206405693951e-07, "loss": 1.0655, "step": 21 }, { "epoch": 0.0007837967828704776, "grad_norm": 15.315303189039632, "learning_rate": 5.219454329774615e-07, "loss": 1.0287, "step": 22 }, { "epoch": 0.0008194239093645903, "grad_norm": 17.524581455100773, "learning_rate": 5.456702253855279e-07, "loss": 1.0301, "step": 23 }, { "epoch": 0.0008550510358587028, "grad_norm": 14.640899906995571, "learning_rate": 5.693950177935943e-07, "loss": 1.1444, "step": 24 }, { "epoch": 0.0008906781623528154, "grad_norm": 16.724946822084046, "learning_rate": 5.931198102016608e-07, "loss": 1.0094, "step": 25 }, { "epoch": 0.000926305288846928, "grad_norm": 14.819363207093549, "learning_rate": 6.168446026097272e-07, "loss": 1.0571, "step": 26 }, { "epoch": 0.0009619324153410406, "grad_norm": 12.643230253712904, "learning_rate": 6.405693950177936e-07, "loss": 0.9412, "step": 27 }, { "epoch": 0.0009975595418351532, "grad_norm": 8.519440140221665, "learning_rate": 6.6429418742586e-07, "loss": 0.9412, "step": 28 }, { "epoch": 0.0010331866683292658, "grad_norm": 7.183691595792046, "learning_rate": 6.880189798339265e-07, "loss": 0.8454, "step": 29 }, { "epoch": 0.0010688137948233786, "grad_norm": 8.745266208309936, "learning_rate": 7.11743772241993e-07, "loss": 0.9804, "step": 30 }, { "epoch": 0.0011044409213174912, "grad_norm": 5.657226629555957, "learning_rate": 7.354685646500593e-07, "loss": 0.8213, "step": 31 }, { "epoch": 0.0011400680478116038, "grad_norm": 6.9731889932233, "learning_rate": 7.591933570581258e-07, "loss": 0.939, "step": 32 }, { "epoch": 0.0011756951743057164, "grad_norm": 6.156904147669122, "learning_rate": 7.829181494661923e-07, "loss": 0.7822, "step": 33 }, { "epoch": 0.001211322300799829, "grad_norm": 6.772298286645846, "learning_rate": 8.066429418742587e-07, "loss": 0.9644, "step": 34 }, { "epoch": 0.0012469494272939416, "grad_norm": 6.706084049371431, "learning_rate": 8.303677342823251e-07, "loss": 0.9437, "step": 35 }, { "epoch": 0.0012825765537880542, "grad_norm": 6.504614040466318, "learning_rate": 8.540925266903916e-07, "loss": 0.8896, "step": 36 }, { "epoch": 0.0013182036802821667, "grad_norm": 5.412541243854199, "learning_rate": 8.778173190984579e-07, "loss": 0.9148, "step": 37 }, { "epoch": 0.0013538308067762796, "grad_norm": 5.221802857224159, "learning_rate": 9.015421115065244e-07, "loss": 0.8768, "step": 38 }, { "epoch": 0.0013894579332703921, "grad_norm": 4.24598161548031, "learning_rate": 9.252669039145908e-07, "loss": 0.7087, "step": 39 }, { "epoch": 0.0014250850597645047, "grad_norm": 4.592868263291016, "learning_rate": 9.489916963226573e-07, "loss": 0.739, "step": 40 }, { "epoch": 0.0014607121862586173, "grad_norm": 5.702556175233735, "learning_rate": 9.727164887307238e-07, "loss": 0.891, "step": 41 }, { "epoch": 0.00149633931275273, "grad_norm": 3.775961641888677, "learning_rate": 9.964412811387902e-07, "loss": 0.7895, "step": 42 }, { "epoch": 0.0015319664392468425, "grad_norm": 4.254743191215041, "learning_rate": 1.0201660735468566e-06, "loss": 0.8973, "step": 43 }, { "epoch": 0.001567593565740955, "grad_norm": 4.518591227160171, "learning_rate": 1.043890865954923e-06, "loss": 0.9344, "step": 44 }, { "epoch": 0.0016032206922350677, "grad_norm": 4.546504106754028, "learning_rate": 1.0676156583629894e-06, "loss": 0.9332, "step": 45 }, { "epoch": 0.0016388478187291805, "grad_norm": 3.603557085836606, "learning_rate": 1.0913404507710558e-06, "loss": 0.8451, "step": 46 }, { "epoch": 0.001674474945223293, "grad_norm": 4.373316243480188, "learning_rate": 1.1150652431791222e-06, "loss": 0.9544, "step": 47 }, { "epoch": 0.0017101020717174057, "grad_norm": 3.8475715444561858, "learning_rate": 1.1387900355871886e-06, "loss": 0.7488, "step": 48 }, { "epoch": 0.0017457291982115183, "grad_norm": 3.611346773323566, "learning_rate": 1.1625148279952552e-06, "loss": 0.7532, "step": 49 }, { "epoch": 0.0017813563247056309, "grad_norm": 3.640063886270524, "learning_rate": 1.1862396204033216e-06, "loss": 0.733, "step": 50 }, { "epoch": 0.0018169834511997435, "grad_norm": 3.505042803672965, "learning_rate": 1.209964412811388e-06, "loss": 0.7979, "step": 51 }, { "epoch": 0.001852610577693856, "grad_norm": 3.688741959602845, "learning_rate": 1.2336892052194544e-06, "loss": 0.7091, "step": 52 }, { "epoch": 0.0018882377041879686, "grad_norm": 3.671636436072009, "learning_rate": 1.257413997627521e-06, "loss": 0.7885, "step": 53 }, { "epoch": 0.0019238648306820812, "grad_norm": 3.0253223070807365, "learning_rate": 1.2811387900355872e-06, "loss": 0.7286, "step": 54 }, { "epoch": 0.001959491957176194, "grad_norm": 3.177706187428551, "learning_rate": 1.3048635824436536e-06, "loss": 0.7029, "step": 55 }, { "epoch": 0.0019951190836703064, "grad_norm": 3.2027112903892614, "learning_rate": 1.32858837485172e-06, "loss": 0.737, "step": 56 }, { "epoch": 0.0020307462101644192, "grad_norm": 3.4354122301281977, "learning_rate": 1.3523131672597867e-06, "loss": 0.7743, "step": 57 }, { "epoch": 0.0020663733366585316, "grad_norm": 3.9817423027228225, "learning_rate": 1.376037959667853e-06, "loss": 0.8911, "step": 58 }, { "epoch": 0.0021020004631526444, "grad_norm": 2.971150792523345, "learning_rate": 1.3997627520759195e-06, "loss": 0.7485, "step": 59 }, { "epoch": 0.0021376275896467572, "grad_norm": 3.3885601442799853, "learning_rate": 1.423487544483986e-06, "loss": 0.8719, "step": 60 }, { "epoch": 0.0021732547161408696, "grad_norm": 3.4213310570541067, "learning_rate": 1.4472123368920523e-06, "loss": 0.8678, "step": 61 }, { "epoch": 0.0022088818426349824, "grad_norm": 3.831612106700712, "learning_rate": 1.4709371293001187e-06, "loss": 0.789, "step": 62 }, { "epoch": 0.0022445089691290948, "grad_norm": 2.8541512826768005, "learning_rate": 1.494661921708185e-06, "loss": 0.669, "step": 63 }, { "epoch": 0.0022801360956232076, "grad_norm": 3.382594832781971, "learning_rate": 1.5183867141162517e-06, "loss": 0.9974, "step": 64 }, { "epoch": 0.00231576322211732, "grad_norm": 3.0845493586370654, "learning_rate": 1.542111506524318e-06, "loss": 0.7726, "step": 65 }, { "epoch": 0.0023513903486114328, "grad_norm": 2.951070804405524, "learning_rate": 1.5658362989323845e-06, "loss": 0.8308, "step": 66 }, { "epoch": 0.0023870174751055456, "grad_norm": 3.348823024999136, "learning_rate": 1.5895610913404507e-06, "loss": 0.8129, "step": 67 }, { "epoch": 0.002422644601599658, "grad_norm": 2.8749155910410655, "learning_rate": 1.6132858837485173e-06, "loss": 0.8562, "step": 68 }, { "epoch": 0.0024582717280937708, "grad_norm": 2.720669499010511, "learning_rate": 1.6370106761565837e-06, "loss": 0.7265, "step": 69 }, { "epoch": 0.002493898854587883, "grad_norm": 2.8511261510145594, "learning_rate": 1.6607354685646501e-06, "loss": 0.6966, "step": 70 }, { "epoch": 0.002529525981081996, "grad_norm": 2.9037093460040615, "learning_rate": 1.6844602609727167e-06, "loss": 0.742, "step": 71 }, { "epoch": 0.0025651531075761083, "grad_norm": 3.6283177341684585, "learning_rate": 1.7081850533807831e-06, "loss": 0.7939, "step": 72 }, { "epoch": 0.002600780234070221, "grad_norm": 2.8357793209028164, "learning_rate": 1.7319098457888495e-06, "loss": 0.6245, "step": 73 }, { "epoch": 0.0026364073605643335, "grad_norm": 3.427738169295175, "learning_rate": 1.7556346381969157e-06, "loss": 0.7736, "step": 74 }, { "epoch": 0.0026720344870584463, "grad_norm": 2.789867367791748, "learning_rate": 1.7793594306049826e-06, "loss": 0.677, "step": 75 }, { "epoch": 0.002707661613552559, "grad_norm": 3.9714451484302327, "learning_rate": 1.8030842230130487e-06, "loss": 0.6939, "step": 76 }, { "epoch": 0.0027432887400466715, "grad_norm": 2.693678496779125, "learning_rate": 1.8268090154211151e-06, "loss": 0.6312, "step": 77 }, { "epoch": 0.0027789158665407843, "grad_norm": 3.5809934495435916, "learning_rate": 1.8505338078291815e-06, "loss": 0.688, "step": 78 }, { "epoch": 0.0028145429930348967, "grad_norm": 3.4677545428352863, "learning_rate": 1.8742586002372482e-06, "loss": 0.7915, "step": 79 }, { "epoch": 0.0028501701195290095, "grad_norm": 3.057695231560799, "learning_rate": 1.8979833926453146e-06, "loss": 0.8262, "step": 80 }, { "epoch": 0.002885797246023122, "grad_norm": 3.349958138824004, "learning_rate": 1.921708185053381e-06, "loss": 0.8672, "step": 81 }, { "epoch": 0.0029214243725172347, "grad_norm": 2.583818495220837, "learning_rate": 1.9454329774614476e-06, "loss": 0.6539, "step": 82 }, { "epoch": 0.002957051499011347, "grad_norm": 2.2956073717052363, "learning_rate": 1.9691577698695138e-06, "loss": 0.6915, "step": 83 }, { "epoch": 0.00299267862550546, "grad_norm": 3.2476081898468405, "learning_rate": 1.9928825622775804e-06, "loss": 0.8253, "step": 84 }, { "epoch": 0.0030283057519995727, "grad_norm": 2.7440601903707216, "learning_rate": 2.0166073546856466e-06, "loss": 0.6706, "step": 85 }, { "epoch": 0.003063932878493685, "grad_norm": 2.270856519192938, "learning_rate": 2.040332147093713e-06, "loss": 0.6608, "step": 86 }, { "epoch": 0.003099560004987798, "grad_norm": 3.08184618506475, "learning_rate": 2.0640569395017794e-06, "loss": 0.7262, "step": 87 }, { "epoch": 0.00313518713148191, "grad_norm": 2.2897688114168755, "learning_rate": 2.087781731909846e-06, "loss": 0.6334, "step": 88 }, { "epoch": 0.003170814257976023, "grad_norm": 2.598639878735302, "learning_rate": 2.111506524317912e-06, "loss": 0.7163, "step": 89 }, { "epoch": 0.0032064413844701354, "grad_norm": 2.2671935240672285, "learning_rate": 2.135231316725979e-06, "loss": 0.6922, "step": 90 }, { "epoch": 0.003242068510964248, "grad_norm": 3.848945414138539, "learning_rate": 2.1589561091340454e-06, "loss": 0.8004, "step": 91 }, { "epoch": 0.003277695637458361, "grad_norm": 2.647375684431088, "learning_rate": 2.1826809015421116e-06, "loss": 0.6574, "step": 92 }, { "epoch": 0.0033133227639524734, "grad_norm": 2.3674650711663525, "learning_rate": 2.2064056939501782e-06, "loss": 0.6234, "step": 93 }, { "epoch": 0.003348949890446586, "grad_norm": 3.3842930505155646, "learning_rate": 2.2301304863582444e-06, "loss": 0.7287, "step": 94 }, { "epoch": 0.0033845770169406986, "grad_norm": 3.4561556843920145, "learning_rate": 2.253855278766311e-06, "loss": 0.8137, "step": 95 }, { "epoch": 0.0034202041434348114, "grad_norm": 2.888779138053944, "learning_rate": 2.2775800711743772e-06, "loss": 0.8197, "step": 96 }, { "epoch": 0.0034558312699289238, "grad_norm": 2.103332267092534, "learning_rate": 2.301304863582444e-06, "loss": 0.6578, "step": 97 }, { "epoch": 0.0034914583964230366, "grad_norm": 3.148616715925409, "learning_rate": 2.3250296559905105e-06, "loss": 0.6321, "step": 98 }, { "epoch": 0.003527085522917149, "grad_norm": 2.5274337505472784, "learning_rate": 2.3487544483985766e-06, "loss": 0.6399, "step": 99 }, { "epoch": 0.0035627126494112617, "grad_norm": 2.755375460184002, "learning_rate": 2.3724792408066433e-06, "loss": 0.7444, "step": 100 }, { "epoch": 0.0035983397759053746, "grad_norm": 2.201173781992292, "learning_rate": 2.3962040332147095e-06, "loss": 0.6434, "step": 101 }, { "epoch": 0.003633966902399487, "grad_norm": 2.233617108610439, "learning_rate": 2.419928825622776e-06, "loss": 0.6596, "step": 102 }, { "epoch": 0.0036695940288935997, "grad_norm": 2.406545841850302, "learning_rate": 2.4436536180308423e-06, "loss": 0.6824, "step": 103 }, { "epoch": 0.003705221155387712, "grad_norm": 6.650632472608422, "learning_rate": 2.467378410438909e-06, "loss": 0.859, "step": 104 }, { "epoch": 0.003740848281881825, "grad_norm": 2.1863064846178673, "learning_rate": 2.4911032028469755e-06, "loss": 0.6635, "step": 105 }, { "epoch": 0.0037764754083759373, "grad_norm": 3.298615493492129, "learning_rate": 2.514827995255042e-06, "loss": 0.7821, "step": 106 }, { "epoch": 0.00381210253487005, "grad_norm": 2.560381268383419, "learning_rate": 2.5385527876631083e-06, "loss": 0.6712, "step": 107 }, { "epoch": 0.0038477296613641625, "grad_norm": 3.3045058606666773, "learning_rate": 2.5622775800711745e-06, "loss": 0.6907, "step": 108 }, { "epoch": 0.0038833567878582753, "grad_norm": 3.3640431720794925, "learning_rate": 2.586002372479241e-06, "loss": 0.8279, "step": 109 }, { "epoch": 0.003918983914352388, "grad_norm": 2.8158075198097303, "learning_rate": 2.6097271648873073e-06, "loss": 0.8186, "step": 110 }, { "epoch": 0.003954611040846501, "grad_norm": 3.302253920763101, "learning_rate": 2.633451957295374e-06, "loss": 0.6684, "step": 111 }, { "epoch": 0.003990238167340613, "grad_norm": 2.251908353442914, "learning_rate": 2.65717674970344e-06, "loss": 0.6311, "step": 112 }, { "epoch": 0.004025865293834726, "grad_norm": 2.369914042881292, "learning_rate": 2.680901542111507e-06, "loss": 0.7095, "step": 113 }, { "epoch": 0.0040614924203288385, "grad_norm": 2.984414128040217, "learning_rate": 2.7046263345195733e-06, "loss": 0.783, "step": 114 }, { "epoch": 0.004097119546822951, "grad_norm": 2.1967582152420064, "learning_rate": 2.7283511269276395e-06, "loss": 0.6689, "step": 115 }, { "epoch": 0.004132746673317063, "grad_norm": 2.1555266809433973, "learning_rate": 2.752075919335706e-06, "loss": 0.5614, "step": 116 }, { "epoch": 0.004168373799811176, "grad_norm": 3.538278791994623, "learning_rate": 2.7758007117437723e-06, "loss": 0.8269, "step": 117 }, { "epoch": 0.004204000926305289, "grad_norm": 3.136955153821265, "learning_rate": 2.799525504151839e-06, "loss": 0.7836, "step": 118 }, { "epoch": 0.004239628052799402, "grad_norm": 3.0265987890380415, "learning_rate": 2.823250296559905e-06, "loss": 0.7828, "step": 119 }, { "epoch": 0.0042752551792935144, "grad_norm": 3.440665887010993, "learning_rate": 2.846975088967972e-06, "loss": 0.7746, "step": 120 }, { "epoch": 0.004310882305787626, "grad_norm": 2.016036067609392, "learning_rate": 2.8706998813760384e-06, "loss": 0.6092, "step": 121 }, { "epoch": 0.004346509432281739, "grad_norm": 2.26366120718496, "learning_rate": 2.8944246737841046e-06, "loss": 0.6673, "step": 122 }, { "epoch": 0.004382136558775852, "grad_norm": 4.131408549621545, "learning_rate": 2.918149466192171e-06, "loss": 0.7549, "step": 123 }, { "epoch": 0.004417763685269965, "grad_norm": 3.7658039098152245, "learning_rate": 2.9418742586002374e-06, "loss": 0.7751, "step": 124 }, { "epoch": 0.004453390811764077, "grad_norm": 2.2655089516256433, "learning_rate": 2.965599051008304e-06, "loss": 0.6327, "step": 125 }, { "epoch": 0.0044890179382581896, "grad_norm": 3.534956265715494, "learning_rate": 2.98932384341637e-06, "loss": 0.7311, "step": 126 }, { "epoch": 0.004524645064752302, "grad_norm": 2.5950873684482425, "learning_rate": 3.0130486358244364e-06, "loss": 0.6452, "step": 127 }, { "epoch": 0.004560272191246415, "grad_norm": 4.653114150625334, "learning_rate": 3.0367734282325034e-06, "loss": 0.7422, "step": 128 }, { "epoch": 0.004595899317740528, "grad_norm": 2.1268122686878654, "learning_rate": 3.0604982206405696e-06, "loss": 0.5925, "step": 129 }, { "epoch": 0.00463152644423464, "grad_norm": 2.2328938474243802, "learning_rate": 3.084223013048636e-06, "loss": 0.6442, "step": 130 }, { "epoch": 0.004667153570728753, "grad_norm": 3.481457361958673, "learning_rate": 3.1079478054567024e-06, "loss": 0.7911, "step": 131 }, { "epoch": 0.0047027806972228655, "grad_norm": 3.057087738198068, "learning_rate": 3.131672597864769e-06, "loss": 0.7854, "step": 132 }, { "epoch": 0.004738407823716978, "grad_norm": 2.2043078968537317, "learning_rate": 3.155397390272835e-06, "loss": 0.6268, "step": 133 }, { "epoch": 0.004774034950211091, "grad_norm": 2.2204683110508365, "learning_rate": 3.1791221826809014e-06, "loss": 0.6007, "step": 134 }, { "epoch": 0.004809662076705203, "grad_norm": 3.1263783091929866, "learning_rate": 3.2028469750889684e-06, "loss": 0.7938, "step": 135 }, { "epoch": 0.004845289203199316, "grad_norm": 2.070726530640572, "learning_rate": 3.2265717674970346e-06, "loss": 0.7028, "step": 136 }, { "epoch": 0.004880916329693429, "grad_norm": 3.3865684040427433, "learning_rate": 3.2502965599051012e-06, "loss": 0.8358, "step": 137 }, { "epoch": 0.0049165434561875415, "grad_norm": 2.4586834133560798, "learning_rate": 3.2740213523131674e-06, "loss": 0.6791, "step": 138 }, { "epoch": 0.0049521705826816535, "grad_norm": 3.5353257211924434, "learning_rate": 3.297746144721234e-06, "loss": 0.8318, "step": 139 }, { "epoch": 0.004987797709175766, "grad_norm": 3.9128938833216593, "learning_rate": 3.3214709371293002e-06, "loss": 0.6607, "step": 140 }, { "epoch": 0.005023424835669879, "grad_norm": 2.135847368393073, "learning_rate": 3.3451957295373664e-06, "loss": 0.6573, "step": 141 }, { "epoch": 0.005059051962163992, "grad_norm": 3.7701112867263493, "learning_rate": 3.3689205219454335e-06, "loss": 0.7074, "step": 142 }, { "epoch": 0.005094679088658105, "grad_norm": 2.075221934963425, "learning_rate": 3.3926453143535e-06, "loss": 0.6217, "step": 143 }, { "epoch": 0.005130306215152217, "grad_norm": 3.1836144215349362, "learning_rate": 3.4163701067615663e-06, "loss": 0.7763, "step": 144 }, { "epoch": 0.0051659333416463294, "grad_norm": 3.4259362015185015, "learning_rate": 3.4400948991696325e-06, "loss": 0.6901, "step": 145 }, { "epoch": 0.005201560468140442, "grad_norm": 3.610821964267663, "learning_rate": 3.463819691577699e-06, "loss": 0.7617, "step": 146 }, { "epoch": 0.005237187594634555, "grad_norm": 2.2120248305335735, "learning_rate": 3.4875444839857653e-06, "loss": 0.642, "step": 147 }, { "epoch": 0.005272814721128667, "grad_norm": 3.7939007034961505, "learning_rate": 3.5112692763938315e-06, "loss": 0.7377, "step": 148 }, { "epoch": 0.00530844184762278, "grad_norm": 3.2187236294197614, "learning_rate": 3.5349940688018985e-06, "loss": 0.7431, "step": 149 }, { "epoch": 0.005344068974116893, "grad_norm": 2.0936130111648676, "learning_rate": 3.558718861209965e-06, "loss": 0.5871, "step": 150 }, { "epoch": 0.005379696100611005, "grad_norm": 3.6466458473127052, "learning_rate": 3.5824436536180313e-06, "loss": 0.7888, "step": 151 }, { "epoch": 0.005415323227105118, "grad_norm": 3.4006584333968406, "learning_rate": 3.6061684460260975e-06, "loss": 0.7393, "step": 152 }, { "epoch": 0.00545095035359923, "grad_norm": 2.3587385463408532, "learning_rate": 3.629893238434164e-06, "loss": 0.6258, "step": 153 }, { "epoch": 0.005486577480093343, "grad_norm": 3.35243315738266, "learning_rate": 3.6536180308422303e-06, "loss": 0.7561, "step": 154 }, { "epoch": 0.005522204606587456, "grad_norm": 2.958667692921445, "learning_rate": 3.6773428232502965e-06, "loss": 0.6731, "step": 155 }, { "epoch": 0.005557831733081569, "grad_norm": 2.044471927570652, "learning_rate": 3.701067615658363e-06, "loss": 0.593, "step": 156 }, { "epoch": 0.0055934588595756805, "grad_norm": 3.7251585551380586, "learning_rate": 3.72479240806643e-06, "loss": 0.8534, "step": 157 }, { "epoch": 0.005629085986069793, "grad_norm": 2.0574425260270517, "learning_rate": 3.7485172004744963e-06, "loss": 0.6643, "step": 158 }, { "epoch": 0.005664713112563906, "grad_norm": 4.148713750270547, "learning_rate": 3.7722419928825625e-06, "loss": 0.629, "step": 159 }, { "epoch": 0.005700340239058019, "grad_norm": 3.3538889149112507, "learning_rate": 3.795966785290629e-06, "loss": 0.6896, "step": 160 }, { "epoch": 0.005735967365552132, "grad_norm": 2.5576998940531226, "learning_rate": 3.819691577698696e-06, "loss": 0.6404, "step": 161 }, { "epoch": 0.005771594492046244, "grad_norm": 2.367220618045496, "learning_rate": 3.843416370106762e-06, "loss": 0.6147, "step": 162 }, { "epoch": 0.0058072216185403565, "grad_norm": 3.2681400525939073, "learning_rate": 3.867141162514828e-06, "loss": 0.7343, "step": 163 }, { "epoch": 0.005842848745034469, "grad_norm": 2.7714100436158353, "learning_rate": 3.890865954922895e-06, "loss": 0.7566, "step": 164 }, { "epoch": 0.005878475871528582, "grad_norm": 4.870693620851394, "learning_rate": 3.914590747330961e-06, "loss": 0.7631, "step": 165 }, { "epoch": 0.005914102998022694, "grad_norm": 2.6630447510850876, "learning_rate": 3.9383155397390276e-06, "loss": 0.6333, "step": 166 }, { "epoch": 0.005949730124516807, "grad_norm": 6.085365135157088, "learning_rate": 3.962040332147094e-06, "loss": 0.6729, "step": 167 }, { "epoch": 0.00598535725101092, "grad_norm": 3.1600520154144665, "learning_rate": 3.985765124555161e-06, "loss": 0.6607, "step": 168 }, { "epoch": 0.0060209843775050325, "grad_norm": 2.2745467347661723, "learning_rate": 4.009489916963227e-06, "loss": 0.5893, "step": 169 }, { "epoch": 0.006056611503999145, "grad_norm": 1.9868744361685164, "learning_rate": 4.033214709371293e-06, "loss": 0.637, "step": 170 }, { "epoch": 0.006092238630493257, "grad_norm": 2.139112859976985, "learning_rate": 4.05693950177936e-06, "loss": 0.6269, "step": 171 }, { "epoch": 0.00612786575698737, "grad_norm": 3.2842748732537315, "learning_rate": 4.080664294187426e-06, "loss": 0.6808, "step": 172 }, { "epoch": 0.006163492883481483, "grad_norm": 3.9647094146455846, "learning_rate": 4.104389086595493e-06, "loss": 0.791, "step": 173 }, { "epoch": 0.006199120009975596, "grad_norm": 3.669939788264286, "learning_rate": 4.128113879003559e-06, "loss": 0.6053, "step": 174 }, { "epoch": 0.006234747136469708, "grad_norm": 2.1894529972697585, "learning_rate": 4.151838671411626e-06, "loss": 0.6051, "step": 175 }, { "epoch": 0.00627037426296382, "grad_norm": 2.214052078417976, "learning_rate": 4.175563463819692e-06, "loss": 0.6258, "step": 176 }, { "epoch": 0.006306001389457933, "grad_norm": 3.1609309033735435, "learning_rate": 4.199288256227758e-06, "loss": 0.6985, "step": 177 }, { "epoch": 0.006341628515952046, "grad_norm": 2.3201957238531357, "learning_rate": 4.223013048635824e-06, "loss": 0.6162, "step": 178 }, { "epoch": 0.006377255642446159, "grad_norm": 3.0055925851726153, "learning_rate": 4.2467378410438914e-06, "loss": 0.6191, "step": 179 }, { "epoch": 0.006412882768940271, "grad_norm": 3.7394028946109006, "learning_rate": 4.270462633451958e-06, "loss": 0.7449, "step": 180 }, { "epoch": 0.006448509895434384, "grad_norm": 3.2959216177793884, "learning_rate": 4.294187425860024e-06, "loss": 0.7166, "step": 181 }, { "epoch": 0.006484137021928496, "grad_norm": 2.162379506419001, "learning_rate": 4.317912218268091e-06, "loss": 0.6576, "step": 182 }, { "epoch": 0.006519764148422609, "grad_norm": 3.4247210273821755, "learning_rate": 4.341637010676157e-06, "loss": 0.6836, "step": 183 }, { "epoch": 0.006555391274916722, "grad_norm": 4.654609068053908, "learning_rate": 4.365361803084223e-06, "loss": 0.6654, "step": 184 }, { "epoch": 0.006591018401410834, "grad_norm": 2.291497316273739, "learning_rate": 4.389086595492289e-06, "loss": 0.671, "step": 185 }, { "epoch": 0.006626645527904947, "grad_norm": 3.453159320720734, "learning_rate": 4.4128113879003565e-06, "loss": 0.7612, "step": 186 }, { "epoch": 0.00666227265439906, "grad_norm": 2.453084213059725, "learning_rate": 4.436536180308423e-06, "loss": 0.6134, "step": 187 }, { "epoch": 0.006697899780893172, "grad_norm": 1.9904272312093738, "learning_rate": 4.460260972716489e-06, "loss": 0.6326, "step": 188 }, { "epoch": 0.006733526907387284, "grad_norm": 2.2108900913010987, "learning_rate": 4.483985765124556e-06, "loss": 0.634, "step": 189 }, { "epoch": 0.006769154033881397, "grad_norm": 3.7965513410674, "learning_rate": 4.507710557532622e-06, "loss": 0.7766, "step": 190 }, { "epoch": 0.00680478116037551, "grad_norm": 3.2745756841248213, "learning_rate": 4.531435349940688e-06, "loss": 0.7991, "step": 191 }, { "epoch": 0.006840408286869623, "grad_norm": 2.91697028752993, "learning_rate": 4.5551601423487545e-06, "loss": 0.6498, "step": 192 }, { "epoch": 0.0068760354133637356, "grad_norm": 2.2000724018679216, "learning_rate": 4.5788849347568215e-06, "loss": 0.6296, "step": 193 }, { "epoch": 0.0069116625398578475, "grad_norm": 3.044943200954558, "learning_rate": 4.602609727164888e-06, "loss": 0.6739, "step": 194 }, { "epoch": 0.00694728966635196, "grad_norm": 3.6660042258421544, "learning_rate": 4.626334519572954e-06, "loss": 0.5861, "step": 195 }, { "epoch": 0.006982916792846073, "grad_norm": 2.8696610988664597, "learning_rate": 4.650059311981021e-06, "loss": 0.7371, "step": 196 }, { "epoch": 0.007018543919340186, "grad_norm": 2.1282062301009486, "learning_rate": 4.673784104389087e-06, "loss": 0.5901, "step": 197 }, { "epoch": 0.007054171045834298, "grad_norm": 2.047030228327888, "learning_rate": 4.697508896797153e-06, "loss": 0.6266, "step": 198 }, { "epoch": 0.007089798172328411, "grad_norm": 2.1551616058155796, "learning_rate": 4.7212336892052195e-06, "loss": 0.6555, "step": 199 }, { "epoch": 0.0071254252988225235, "grad_norm": 4.424834419608751, "learning_rate": 4.7449584816132865e-06, "loss": 0.7444, "step": 200 }, { "epoch": 0.007161052425316636, "grad_norm": 2.601092765910125, "learning_rate": 4.768683274021353e-06, "loss": 0.586, "step": 201 }, { "epoch": 0.007196679551810749, "grad_norm": 3.5920768263223635, "learning_rate": 4.792408066429419e-06, "loss": 0.7034, "step": 202 }, { "epoch": 0.007232306678304861, "grad_norm": 2.0842474420807955, "learning_rate": 4.816132858837486e-06, "loss": 0.6317, "step": 203 }, { "epoch": 0.007267933804798974, "grad_norm": 3.1128132611029633, "learning_rate": 4.839857651245552e-06, "loss": 0.7006, "step": 204 }, { "epoch": 0.007303560931293087, "grad_norm": 3.51892545037501, "learning_rate": 4.863582443653618e-06, "loss": 0.7376, "step": 205 }, { "epoch": 0.0073391880577871995, "grad_norm": 2.9607242761121895, "learning_rate": 4.8873072360616845e-06, "loss": 0.5834, "step": 206 }, { "epoch": 0.007374815184281311, "grad_norm": 3.912372274200243, "learning_rate": 4.911032028469751e-06, "loss": 0.6257, "step": 207 }, { "epoch": 0.007410442310775424, "grad_norm": 2.125002455952471, "learning_rate": 4.934756820877818e-06, "loss": 0.663, "step": 208 }, { "epoch": 0.007446069437269537, "grad_norm": 2.9388143649845033, "learning_rate": 4.958481613285884e-06, "loss": 0.7251, "step": 209 }, { "epoch": 0.00748169656376365, "grad_norm": 3.2466701370590294, "learning_rate": 4.982206405693951e-06, "loss": 0.6815, "step": 210 }, { "epoch": 0.007517323690257763, "grad_norm": 1.951054321249849, "learning_rate": 5.005931198102017e-06, "loss": 0.5956, "step": 211 }, { "epoch": 0.007552950816751875, "grad_norm": 3.0002771844267717, "learning_rate": 5.029655990510084e-06, "loss": 0.745, "step": 212 }, { "epoch": 0.007588577943245987, "grad_norm": 3.4887527169952364, "learning_rate": 5.05338078291815e-06, "loss": 0.723, "step": 213 }, { "epoch": 0.0076242050697401, "grad_norm": 2.7542655666596954, "learning_rate": 5.077105575326217e-06, "loss": 0.6011, "step": 214 }, { "epoch": 0.007659832196234213, "grad_norm": 3.6250741105783226, "learning_rate": 5.100830367734283e-06, "loss": 0.7049, "step": 215 }, { "epoch": 0.007695459322728325, "grad_norm": 2.5354644124567898, "learning_rate": 5.124555160142349e-06, "loss": 0.7175, "step": 216 }, { "epoch": 0.007731086449222438, "grad_norm": 3.49180702681559, "learning_rate": 5.148279952550416e-06, "loss": 0.7478, "step": 217 }, { "epoch": 0.0077667135757165506, "grad_norm": 2.2970942448540224, "learning_rate": 5.172004744958482e-06, "loss": 0.6046, "step": 218 }, { "epoch": 0.007802340702210663, "grad_norm": 4.766010988114086, "learning_rate": 5.195729537366548e-06, "loss": 0.8129, "step": 219 }, { "epoch": 0.007837967828704776, "grad_norm": 2.3632448609613905, "learning_rate": 5.219454329774615e-06, "loss": 0.5751, "step": 220 }, { "epoch": 0.007873594955198888, "grad_norm": 4.242427304152151, "learning_rate": 5.243179122182681e-06, "loss": 0.7089, "step": 221 }, { "epoch": 0.007909222081693002, "grad_norm": 2.763511589072624, "learning_rate": 5.266903914590748e-06, "loss": 0.7259, "step": 222 }, { "epoch": 0.007944849208187114, "grad_norm": 2.7628230526835313, "learning_rate": 5.290628706998814e-06, "loss": 0.5944, "step": 223 }, { "epoch": 0.007980476334681226, "grad_norm": 4.471320036053781, "learning_rate": 5.31435349940688e-06, "loss": 0.6685, "step": 224 }, { "epoch": 0.00801610346117534, "grad_norm": 3.82421305406958, "learning_rate": 5.338078291814946e-06, "loss": 0.6951, "step": 225 }, { "epoch": 0.008051730587669451, "grad_norm": 2.118254973188766, "learning_rate": 5.361803084223014e-06, "loss": 0.6221, "step": 226 }, { "epoch": 0.008087357714163565, "grad_norm": 3.540847711334209, "learning_rate": 5.3855278766310805e-06, "loss": 0.5553, "step": 227 }, { "epoch": 0.008122984840657677, "grad_norm": 6.466106164969915, "learning_rate": 5.409252669039147e-06, "loss": 0.7327, "step": 228 }, { "epoch": 0.008158611967151789, "grad_norm": 2.10652321069356, "learning_rate": 5.432977461447213e-06, "loss": 0.6527, "step": 229 }, { "epoch": 0.008194239093645903, "grad_norm": 3.172891016311093, "learning_rate": 5.456702253855279e-06, "loss": 0.6201, "step": 230 }, { "epoch": 0.008229866220140014, "grad_norm": 2.5547660195323876, "learning_rate": 5.480427046263346e-06, "loss": 0.6026, "step": 231 }, { "epoch": 0.008265493346634126, "grad_norm": 2.3480696239241055, "learning_rate": 5.504151838671412e-06, "loss": 0.5918, "step": 232 }, { "epoch": 0.00830112047312824, "grad_norm": 2.4261073865087246, "learning_rate": 5.5278766310794785e-06, "loss": 0.5699, "step": 233 }, { "epoch": 0.008336747599622352, "grad_norm": 4.402228961018783, "learning_rate": 5.551601423487545e-06, "loss": 0.6235, "step": 234 }, { "epoch": 0.008372374726116466, "grad_norm": 2.79037615963722, "learning_rate": 5.575326215895611e-06, "loss": 0.547, "step": 235 }, { "epoch": 0.008408001852610578, "grad_norm": 2.2754494743087847, "learning_rate": 5.599051008303678e-06, "loss": 0.5969, "step": 236 }, { "epoch": 0.00844362897910469, "grad_norm": 3.2392300007323604, "learning_rate": 5.622775800711744e-06, "loss": 0.6307, "step": 237 }, { "epoch": 0.008479256105598803, "grad_norm": 3.659518127243994, "learning_rate": 5.64650059311981e-06, "loss": 0.7471, "step": 238 }, { "epoch": 0.008514883232092915, "grad_norm": 3.1205631054306355, "learning_rate": 5.6702253855278765e-06, "loss": 0.7235, "step": 239 }, { "epoch": 0.008550510358587029, "grad_norm": 3.387902733021182, "learning_rate": 5.693950177935944e-06, "loss": 0.6804, "step": 240 }, { "epoch": 0.00858613748508114, "grad_norm": 3.355698366552274, "learning_rate": 5.7176749703440105e-06, "loss": 0.671, "step": 241 }, { "epoch": 0.008621764611575253, "grad_norm": 2.840357426734429, "learning_rate": 5.741399762752077e-06, "loss": 0.7727, "step": 242 }, { "epoch": 0.008657391738069366, "grad_norm": 2.6011650906945456, "learning_rate": 5.765124555160143e-06, "loss": 0.5405, "step": 243 }, { "epoch": 0.008693018864563478, "grad_norm": 4.515271763858659, "learning_rate": 5.788849347568209e-06, "loss": 0.6944, "step": 244 }, { "epoch": 0.008728645991057592, "grad_norm": 2.177031112245255, "learning_rate": 5.812574139976276e-06, "loss": 0.6108, "step": 245 }, { "epoch": 0.008764273117551704, "grad_norm": 4.3343697103876435, "learning_rate": 5.836298932384342e-06, "loss": 0.7215, "step": 246 }, { "epoch": 0.008799900244045816, "grad_norm": 3.279280677053888, "learning_rate": 5.8600237247924085e-06, "loss": 0.7724, "step": 247 }, { "epoch": 0.00883552737053993, "grad_norm": 3.123804606578186, "learning_rate": 5.883748517200475e-06, "loss": 0.7163, "step": 248 }, { "epoch": 0.008871154497034042, "grad_norm": 3.3988740543696623, "learning_rate": 5.907473309608541e-06, "loss": 0.6399, "step": 249 }, { "epoch": 0.008906781623528153, "grad_norm": 2.2719005271426886, "learning_rate": 5.931198102016608e-06, "loss": 0.6114, "step": 250 }, { "epoch": 0.008942408750022267, "grad_norm": 2.5156869949221035, "learning_rate": 5.954922894424674e-06, "loss": 0.6283, "step": 251 }, { "epoch": 0.008978035876516379, "grad_norm": 2.4257891340136952, "learning_rate": 5.97864768683274e-06, "loss": 0.6741, "step": 252 }, { "epoch": 0.009013663003010493, "grad_norm": 2.3502774917934106, "learning_rate": 6.0023724792408065e-06, "loss": 0.5445, "step": 253 }, { "epoch": 0.009049290129504605, "grad_norm": 2.1243614694286044, "learning_rate": 6.026097271648873e-06, "loss": 0.5547, "step": 254 }, { "epoch": 0.009084917255998717, "grad_norm": 4.059534925701807, "learning_rate": 6.049822064056941e-06, "loss": 0.7034, "step": 255 }, { "epoch": 0.00912054438249283, "grad_norm": 2.3658957233317097, "learning_rate": 6.073546856465007e-06, "loss": 0.5806, "step": 256 }, { "epoch": 0.009156171508986942, "grad_norm": 2.309246966936275, "learning_rate": 6.097271648873073e-06, "loss": 0.6191, "step": 257 }, { "epoch": 0.009191798635481056, "grad_norm": 2.999250702540632, "learning_rate": 6.120996441281139e-06, "loss": 0.5674, "step": 258 }, { "epoch": 0.009227425761975168, "grad_norm": 3.5226390846933224, "learning_rate": 6.144721233689206e-06, "loss": 0.6305, "step": 259 }, { "epoch": 0.00926305288846928, "grad_norm": 2.6808748198194103, "learning_rate": 6.168446026097272e-06, "loss": 0.6099, "step": 260 }, { "epoch": 0.009298680014963394, "grad_norm": 2.207781197140729, "learning_rate": 6.192170818505339e-06, "loss": 0.6076, "step": 261 }, { "epoch": 0.009334307141457505, "grad_norm": 2.7281715685660854, "learning_rate": 6.215895610913405e-06, "loss": 0.5712, "step": 262 }, { "epoch": 0.009369934267951619, "grad_norm": 4.095934805541487, "learning_rate": 6.239620403321471e-06, "loss": 0.6327, "step": 263 }, { "epoch": 0.009405561394445731, "grad_norm": 2.30644426604409, "learning_rate": 6.263345195729538e-06, "loss": 0.5751, "step": 264 }, { "epoch": 0.009441188520939843, "grad_norm": 4.005511269265285, "learning_rate": 6.287069988137604e-06, "loss": 0.622, "step": 265 }, { "epoch": 0.009476815647433957, "grad_norm": 3.4669762594739946, "learning_rate": 6.31079478054567e-06, "loss": 0.6283, "step": 266 }, { "epoch": 0.009512442773928069, "grad_norm": 2.752757996885337, "learning_rate": 6.334519572953737e-06, "loss": 0.6238, "step": 267 }, { "epoch": 0.009548069900422182, "grad_norm": 3.712582937431813, "learning_rate": 6.358244365361803e-06, "loss": 0.7096, "step": 268 }, { "epoch": 0.009583697026916294, "grad_norm": 2.297626599494038, "learning_rate": 6.381969157769871e-06, "loss": 0.5996, "step": 269 }, { "epoch": 0.009619324153410406, "grad_norm": 2.072485784291643, "learning_rate": 6.405693950177937e-06, "loss": 0.6017, "step": 270 }, { "epoch": 0.00965495127990452, "grad_norm": 2.2182546555960805, "learning_rate": 6.429418742586003e-06, "loss": 0.5696, "step": 271 }, { "epoch": 0.009690578406398632, "grad_norm": 2.9983261793640055, "learning_rate": 6.453143534994069e-06, "loss": 0.7259, "step": 272 }, { "epoch": 0.009726205532892744, "grad_norm": 4.36892346743237, "learning_rate": 6.476868327402136e-06, "loss": 0.7085, "step": 273 }, { "epoch": 0.009761832659386857, "grad_norm": 2.3553867725581346, "learning_rate": 6.5005931198102025e-06, "loss": 0.5918, "step": 274 }, { "epoch": 0.00979745978588097, "grad_norm": 2.2782206048189253, "learning_rate": 6.524317912218269e-06, "loss": 0.5539, "step": 275 }, { "epoch": 0.009833086912375083, "grad_norm": 3.5829914025043443, "learning_rate": 6.548042704626335e-06, "loss": 0.6599, "step": 276 }, { "epoch": 0.009868714038869195, "grad_norm": 2.3635442657722368, "learning_rate": 6.571767497034401e-06, "loss": 0.6004, "step": 277 }, { "epoch": 0.009904341165363307, "grad_norm": 4.451638560483535, "learning_rate": 6.595492289442468e-06, "loss": 0.7569, "step": 278 }, { "epoch": 0.00993996829185742, "grad_norm": 2.1741425443101043, "learning_rate": 6.619217081850534e-06, "loss": 0.5789, "step": 279 }, { "epoch": 0.009975595418351533, "grad_norm": 2.0283493128811645, "learning_rate": 6.6429418742586005e-06, "loss": 0.6063, "step": 280 }, { "epoch": 0.010011222544845646, "grad_norm": 2.357658345835138, "learning_rate": 6.666666666666667e-06, "loss": 0.5639, "step": 281 }, { "epoch": 0.010046849671339758, "grad_norm": 3.690731073298382, "learning_rate": 6.690391459074733e-06, "loss": 0.6642, "step": 282 }, { "epoch": 0.01008247679783387, "grad_norm": 3.096498423035901, "learning_rate": 6.7141162514828e-06, "loss": 0.7377, "step": 283 }, { "epoch": 0.010118103924327984, "grad_norm": 2.052009823218646, "learning_rate": 6.737841043890867e-06, "loss": 0.5661, "step": 284 }, { "epoch": 0.010153731050822096, "grad_norm": 3.9428448076896196, "learning_rate": 6.761565836298933e-06, "loss": 0.6889, "step": 285 }, { "epoch": 0.01018935817731621, "grad_norm": 2.107201861419919, "learning_rate": 6.785290628707e-06, "loss": 0.5525, "step": 286 }, { "epoch": 0.010224985303810321, "grad_norm": 2.8908589529570112, "learning_rate": 6.809015421115066e-06, "loss": 0.7514, "step": 287 }, { "epoch": 0.010260612430304433, "grad_norm": 3.5292663397374837, "learning_rate": 6.8327402135231325e-06, "loss": 0.651, "step": 288 }, { "epoch": 0.010296239556798547, "grad_norm": 2.312951927271289, "learning_rate": 6.856465005931199e-06, "loss": 0.5629, "step": 289 }, { "epoch": 0.010331866683292659, "grad_norm": 3.4170193485119453, "learning_rate": 6.880189798339265e-06, "loss": 0.6674, "step": 290 }, { "epoch": 0.01036749380978677, "grad_norm": 2.006431238500953, "learning_rate": 6.903914590747331e-06, "loss": 0.5067, "step": 291 }, { "epoch": 0.010403120936280884, "grad_norm": 2.949304786248961, "learning_rate": 6.927639383155398e-06, "loss": 0.6055, "step": 292 }, { "epoch": 0.010438748062774996, "grad_norm": 2.678405365550337, "learning_rate": 6.951364175563464e-06, "loss": 0.6356, "step": 293 }, { "epoch": 0.01047437518926911, "grad_norm": 2.9821391920783142, "learning_rate": 6.9750889679715305e-06, "loss": 0.7669, "step": 294 }, { "epoch": 0.010510002315763222, "grad_norm": 2.7130909343786076, "learning_rate": 6.998813760379597e-06, "loss": 0.6839, "step": 295 }, { "epoch": 0.010545629442257334, "grad_norm": 2.2867037269780464, "learning_rate": 7.022538552787663e-06, "loss": 0.6472, "step": 296 }, { "epoch": 0.010581256568751448, "grad_norm": 2.343508337997501, "learning_rate": 7.04626334519573e-06, "loss": 0.6338, "step": 297 }, { "epoch": 0.01061688369524556, "grad_norm": 2.4684702726112273, "learning_rate": 7.069988137603797e-06, "loss": 0.6179, "step": 298 }, { "epoch": 0.010652510821739673, "grad_norm": 3.1827335760802047, "learning_rate": 7.093712930011863e-06, "loss": 0.6766, "step": 299 }, { "epoch": 0.010688137948233785, "grad_norm": 2.637167863221858, "learning_rate": 7.11743772241993e-06, "loss": 0.5915, "step": 300 }, { "epoch": 0.010723765074727897, "grad_norm": 3.4342252738459433, "learning_rate": 7.141162514827996e-06, "loss": 0.6731, "step": 301 }, { "epoch": 0.01075939220122201, "grad_norm": 5.618725730625092, "learning_rate": 7.164887307236063e-06, "loss": 0.7982, "step": 302 }, { "epoch": 0.010795019327716123, "grad_norm": 4.006030852718342, "learning_rate": 7.188612099644129e-06, "loss": 0.6299, "step": 303 }, { "epoch": 0.010830646454210236, "grad_norm": 3.598731586076928, "learning_rate": 7.212336892052195e-06, "loss": 0.6386, "step": 304 }, { "epoch": 0.010866273580704348, "grad_norm": 3.552304261692546, "learning_rate": 7.236061684460261e-06, "loss": 0.6987, "step": 305 }, { "epoch": 0.01090190070719846, "grad_norm": 3.4014680508126753, "learning_rate": 7.259786476868328e-06, "loss": 0.6019, "step": 306 }, { "epoch": 0.010937527833692574, "grad_norm": 3.088272536570978, "learning_rate": 7.283511269276394e-06, "loss": 0.608, "step": 307 }, { "epoch": 0.010973154960186686, "grad_norm": 34.59523043118503, "learning_rate": 7.307236061684461e-06, "loss": 0.7437, "step": 308 }, { "epoch": 0.011008782086680798, "grad_norm": 3.144475737374394, "learning_rate": 7.330960854092527e-06, "loss": 0.682, "step": 309 }, { "epoch": 0.011044409213174912, "grad_norm": 2.711891467606018, "learning_rate": 7.354685646500593e-06, "loss": 0.6169, "step": 310 }, { "epoch": 0.011080036339669024, "grad_norm": 2.3980382311934356, "learning_rate": 7.37841043890866e-06, "loss": 0.6269, "step": 311 }, { "epoch": 0.011115663466163137, "grad_norm": 2.3758368693758403, "learning_rate": 7.402135231316726e-06, "loss": 0.5476, "step": 312 }, { "epoch": 0.011151290592657249, "grad_norm": 3.215321058370194, "learning_rate": 7.425860023724793e-06, "loss": 0.7636, "step": 313 }, { "epoch": 0.011186917719151361, "grad_norm": 3.1045481927350385, "learning_rate": 7.44958481613286e-06, "loss": 0.7049, "step": 314 }, { "epoch": 0.011222544845645475, "grad_norm": 2.6565164880201717, "learning_rate": 7.4733096085409265e-06, "loss": 0.596, "step": 315 }, { "epoch": 0.011258171972139587, "grad_norm": 6.650111383568342, "learning_rate": 7.497034400948993e-06, "loss": 0.7025, "step": 316 }, { "epoch": 0.0112937990986337, "grad_norm": 2.516933539767502, "learning_rate": 7.520759193357059e-06, "loss": 0.6532, "step": 317 }, { "epoch": 0.011329426225127812, "grad_norm": 2.590630537699977, "learning_rate": 7.544483985765125e-06, "loss": 0.5202, "step": 318 }, { "epoch": 0.011365053351621924, "grad_norm": 3.4202947680772464, "learning_rate": 7.568208778173192e-06, "loss": 0.6544, "step": 319 }, { "epoch": 0.011400680478116038, "grad_norm": 2.783568426573461, "learning_rate": 7.591933570581258e-06, "loss": 0.6346, "step": 320 }, { "epoch": 0.01143630760461015, "grad_norm": 3.7798380508382383, "learning_rate": 7.6156583629893245e-06, "loss": 0.7364, "step": 321 }, { "epoch": 0.011471934731104264, "grad_norm": 3.13662005915425, "learning_rate": 7.639383155397391e-06, "loss": 0.709, "step": 322 }, { "epoch": 0.011507561857598375, "grad_norm": 3.516188067964708, "learning_rate": 7.663107947805457e-06, "loss": 0.665, "step": 323 }, { "epoch": 0.011543188984092487, "grad_norm": 3.2102918833890373, "learning_rate": 7.686832740213524e-06, "loss": 0.723, "step": 324 }, { "epoch": 0.011578816110586601, "grad_norm": 4.276203438908076, "learning_rate": 7.71055753262159e-06, "loss": 0.6514, "step": 325 }, { "epoch": 0.011614443237080713, "grad_norm": 3.2031270679161183, "learning_rate": 7.734282325029656e-06, "loss": 0.635, "step": 326 }, { "epoch": 0.011650070363574827, "grad_norm": 2.39693297649667, "learning_rate": 7.758007117437723e-06, "loss": 0.5841, "step": 327 }, { "epoch": 0.011685697490068939, "grad_norm": 2.3177526518859057, "learning_rate": 7.78173190984579e-06, "loss": 0.5497, "step": 328 }, { "epoch": 0.01172132461656305, "grad_norm": 3.506343070568429, "learning_rate": 7.805456702253856e-06, "loss": 0.7201, "step": 329 }, { "epoch": 0.011756951743057164, "grad_norm": 16.22725326980985, "learning_rate": 7.829181494661923e-06, "loss": 0.7132, "step": 330 }, { "epoch": 0.011792578869551276, "grad_norm": 3.263876841789549, "learning_rate": 7.85290628706999e-06, "loss": 0.6651, "step": 331 }, { "epoch": 0.011828205996045388, "grad_norm": 4.786579532300427, "learning_rate": 7.876631079478055e-06, "loss": 0.6841, "step": 332 }, { "epoch": 0.011863833122539502, "grad_norm": 3.0510051154486626, "learning_rate": 7.900355871886122e-06, "loss": 0.6059, "step": 333 }, { "epoch": 0.011899460249033614, "grad_norm": 3.265795075011091, "learning_rate": 7.924080664294187e-06, "loss": 0.7313, "step": 334 }, { "epoch": 0.011935087375527727, "grad_norm": 3.110871845132752, "learning_rate": 7.947805456702255e-06, "loss": 0.7255, "step": 335 }, { "epoch": 0.01197071450202184, "grad_norm": 3.117008369650712, "learning_rate": 7.971530249110322e-06, "loss": 0.6018, "step": 336 }, { "epoch": 0.012006341628515951, "grad_norm": 2.9965148295931203, "learning_rate": 7.995255041518387e-06, "loss": 0.6844, "step": 337 }, { "epoch": 0.012041968755010065, "grad_norm": 5.5634329329540275, "learning_rate": 8.018979833926454e-06, "loss": 0.6809, "step": 338 }, { "epoch": 0.012077595881504177, "grad_norm": 4.7625129803604365, "learning_rate": 8.04270462633452e-06, "loss": 0.7769, "step": 339 }, { "epoch": 0.01211322300799829, "grad_norm": 2.490167591199567, "learning_rate": 8.066429418742586e-06, "loss": 0.6401, "step": 340 }, { "epoch": 0.012148850134492403, "grad_norm": 2.8369306264016565, "learning_rate": 8.090154211150653e-06, "loss": 0.5914, "step": 341 }, { "epoch": 0.012184477260986514, "grad_norm": 3.359786135526255, "learning_rate": 8.11387900355872e-06, "loss": 0.8532, "step": 342 }, { "epoch": 0.012220104387480628, "grad_norm": 3.429091702622392, "learning_rate": 8.137603795966786e-06, "loss": 0.6224, "step": 343 }, { "epoch": 0.01225573151397474, "grad_norm": 3.050292311196783, "learning_rate": 8.161328588374853e-06, "loss": 0.7091, "step": 344 }, { "epoch": 0.012291358640468854, "grad_norm": 2.4796258379887246, "learning_rate": 8.18505338078292e-06, "loss": 0.6236, "step": 345 }, { "epoch": 0.012326985766962966, "grad_norm": 2.1227790525771595, "learning_rate": 8.208778173190985e-06, "loss": 0.5502, "step": 346 }, { "epoch": 0.012362612893457078, "grad_norm": 2.2516135347099104, "learning_rate": 8.232502965599052e-06, "loss": 0.505, "step": 347 }, { "epoch": 0.012398240019951191, "grad_norm": 3.185163529809676, "learning_rate": 8.256227758007118e-06, "loss": 0.6392, "step": 348 }, { "epoch": 0.012433867146445303, "grad_norm": 2.1693362471974074, "learning_rate": 8.279952550415185e-06, "loss": 0.6361, "step": 349 }, { "epoch": 0.012469494272939415, "grad_norm": 2.070694701826702, "learning_rate": 8.303677342823252e-06, "loss": 0.5955, "step": 350 }, { "epoch": 0.012505121399433529, "grad_norm": 3.042728047204649, "learning_rate": 8.327402135231317e-06, "loss": 0.6702, "step": 351 }, { "epoch": 0.01254074852592764, "grad_norm": 2.3036749957589726, "learning_rate": 8.351126927639384e-06, "loss": 0.5657, "step": 352 }, { "epoch": 0.012576375652421755, "grad_norm": 3.392130616258841, "learning_rate": 8.37485172004745e-06, "loss": 0.6676, "step": 353 }, { "epoch": 0.012612002778915866, "grad_norm": 3.186434458571833, "learning_rate": 8.398576512455516e-06, "loss": 0.7249, "step": 354 }, { "epoch": 0.012647629905409978, "grad_norm": 3.5435255327377924, "learning_rate": 8.422301304863583e-06, "loss": 0.7421, "step": 355 }, { "epoch": 0.012683257031904092, "grad_norm": 2.3685375098720876, "learning_rate": 8.446026097271649e-06, "loss": 0.6209, "step": 356 }, { "epoch": 0.012718884158398204, "grad_norm": 2.8919894111590625, "learning_rate": 8.469750889679716e-06, "loss": 0.7694, "step": 357 }, { "epoch": 0.012754511284892318, "grad_norm": 3.981502329023644, "learning_rate": 8.493475682087783e-06, "loss": 0.5797, "step": 358 }, { "epoch": 0.01279013841138643, "grad_norm": 6.430374992852741, "learning_rate": 8.51720047449585e-06, "loss": 0.6833, "step": 359 }, { "epoch": 0.012825765537880542, "grad_norm": 2.2445074155305664, "learning_rate": 8.540925266903915e-06, "loss": 0.5573, "step": 360 }, { "epoch": 0.012861392664374655, "grad_norm": 4.807518781392831, "learning_rate": 8.564650059311982e-06, "loss": 0.7438, "step": 361 }, { "epoch": 0.012897019790868767, "grad_norm": 3.824069763271676, "learning_rate": 8.588374851720048e-06, "loss": 0.7068, "step": 362 }, { "epoch": 0.012932646917362881, "grad_norm": 2.038099068052751, "learning_rate": 8.612099644128115e-06, "loss": 0.5989, "step": 363 }, { "epoch": 0.012968274043856993, "grad_norm": 2.4279887980491486, "learning_rate": 8.635824436536182e-06, "loss": 0.5347, "step": 364 }, { "epoch": 0.013003901170351105, "grad_norm": 2.141086023799668, "learning_rate": 8.659549228944247e-06, "loss": 0.5983, "step": 365 }, { "epoch": 0.013039528296845218, "grad_norm": 4.301100261507106, "learning_rate": 8.683274021352314e-06, "loss": 0.5705, "step": 366 }, { "epoch": 0.01307515542333933, "grad_norm": 4.7693211944582075, "learning_rate": 8.70699881376038e-06, "loss": 0.7355, "step": 367 }, { "epoch": 0.013110782549833444, "grad_norm": 2.2201102454339035, "learning_rate": 8.730723606168446e-06, "loss": 0.5616, "step": 368 }, { "epoch": 0.013146409676327556, "grad_norm": 4.366647144498267, "learning_rate": 8.754448398576513e-06, "loss": 0.6485, "step": 369 }, { "epoch": 0.013182036802821668, "grad_norm": 1.8690012968176541, "learning_rate": 8.778173190984579e-06, "loss": 0.5666, "step": 370 }, { "epoch": 0.013217663929315782, "grad_norm": 3.4700258288402357, "learning_rate": 8.801897983392646e-06, "loss": 0.6471, "step": 371 }, { "epoch": 0.013253291055809894, "grad_norm": 4.049277928017355, "learning_rate": 8.825622775800713e-06, "loss": 0.6283, "step": 372 }, { "epoch": 0.013288918182304005, "grad_norm": 2.2809415158103064, "learning_rate": 8.84934756820878e-06, "loss": 0.612, "step": 373 }, { "epoch": 0.01332454530879812, "grad_norm": 2.101029489564298, "learning_rate": 8.873072360616845e-06, "loss": 0.5933, "step": 374 }, { "epoch": 0.013360172435292231, "grad_norm": 2.064382102705697, "learning_rate": 8.896797153024912e-06, "loss": 0.5602, "step": 375 }, { "epoch": 0.013395799561786345, "grad_norm": 1.9625445610170982, "learning_rate": 8.920521945432978e-06, "loss": 0.5891, "step": 376 }, { "epoch": 0.013431426688280457, "grad_norm": 3.09624983254485, "learning_rate": 8.944246737841045e-06, "loss": 0.6669, "step": 377 }, { "epoch": 0.013467053814774569, "grad_norm": 3.7177163430777562, "learning_rate": 8.967971530249112e-06, "loss": 0.6446, "step": 378 }, { "epoch": 0.013502680941268682, "grad_norm": 2.163303262859076, "learning_rate": 8.991696322657177e-06, "loss": 0.5422, "step": 379 }, { "epoch": 0.013538308067762794, "grad_norm": 2.0678303474668085, "learning_rate": 9.015421115065244e-06, "loss": 0.5799, "step": 380 }, { "epoch": 0.013573935194256908, "grad_norm": 2.827556916255906, "learning_rate": 9.03914590747331e-06, "loss": 0.7232, "step": 381 }, { "epoch": 0.01360956232075102, "grad_norm": 3.4089414169523597, "learning_rate": 9.062870699881377e-06, "loss": 0.7496, "step": 382 }, { "epoch": 0.013645189447245132, "grad_norm": 2.142355273176486, "learning_rate": 9.086595492289444e-06, "loss": 0.5957, "step": 383 }, { "epoch": 0.013680816573739246, "grad_norm": 3.306426733266018, "learning_rate": 9.110320284697509e-06, "loss": 0.6442, "step": 384 }, { "epoch": 0.013716443700233357, "grad_norm": 2.646387605987424, "learning_rate": 9.134045077105576e-06, "loss": 0.6007, "step": 385 }, { "epoch": 0.013752070826727471, "grad_norm": 3.664813872799944, "learning_rate": 9.157769869513643e-06, "loss": 0.5942, "step": 386 }, { "epoch": 0.013787697953221583, "grad_norm": 3.441351351921276, "learning_rate": 9.18149466192171e-06, "loss": 0.687, "step": 387 }, { "epoch": 0.013823325079715695, "grad_norm": 2.240286415043619, "learning_rate": 9.205219454329775e-06, "loss": 0.5645, "step": 388 }, { "epoch": 0.013858952206209809, "grad_norm": 1.9583329111354986, "learning_rate": 9.228944246737842e-06, "loss": 0.5728, "step": 389 }, { "epoch": 0.01389457933270392, "grad_norm": 2.22313074179019, "learning_rate": 9.252669039145908e-06, "loss": 0.6174, "step": 390 }, { "epoch": 0.013930206459198033, "grad_norm": 4.555282929280806, "learning_rate": 9.276393831553975e-06, "loss": 0.5717, "step": 391 }, { "epoch": 0.013965833585692146, "grad_norm": 2.0200621317788463, "learning_rate": 9.300118623962042e-06, "loss": 0.5607, "step": 392 }, { "epoch": 0.014001460712186258, "grad_norm": 2.0494635961604044, "learning_rate": 9.323843416370107e-06, "loss": 0.6117, "step": 393 }, { "epoch": 0.014037087838680372, "grad_norm": 2.8375300699912187, "learning_rate": 9.347568208778174e-06, "loss": 0.6842, "step": 394 }, { "epoch": 0.014072714965174484, "grad_norm": 2.0871052368634313, "learning_rate": 9.37129300118624e-06, "loss": 0.6262, "step": 395 }, { "epoch": 0.014108342091668596, "grad_norm": 2.869279973484368, "learning_rate": 9.395017793594307e-06, "loss": 0.6259, "step": 396 }, { "epoch": 0.01414396921816271, "grad_norm": 2.9102586936316954, "learning_rate": 9.418742586002374e-06, "loss": 0.6202, "step": 397 }, { "epoch": 0.014179596344656821, "grad_norm": 3.89793381245094, "learning_rate": 9.442467378410439e-06, "loss": 0.6282, "step": 398 }, { "epoch": 0.014215223471150935, "grad_norm": 1.9887439714010082, "learning_rate": 9.466192170818506e-06, "loss": 0.6061, "step": 399 }, { "epoch": 0.014250850597645047, "grad_norm": 3.59612644026893, "learning_rate": 9.489916963226573e-06, "loss": 0.7701, "step": 400 }, { "epoch": 0.014286477724139159, "grad_norm": 2.1200326060162253, "learning_rate": 9.51364175563464e-06, "loss": 0.6135, "step": 401 }, { "epoch": 0.014322104850633273, "grad_norm": 2.000630724091526, "learning_rate": 9.537366548042705e-06, "loss": 0.5953, "step": 402 }, { "epoch": 0.014357731977127385, "grad_norm": 1.8719611834177292, "learning_rate": 9.561091340450772e-06, "loss": 0.604, "step": 403 }, { "epoch": 0.014393359103621498, "grad_norm": 2.1753762680759037, "learning_rate": 9.584816132858838e-06, "loss": 0.59, "step": 404 }, { "epoch": 0.01442898623011561, "grad_norm": 2.8816062773539515, "learning_rate": 9.608540925266905e-06, "loss": 0.6055, "step": 405 }, { "epoch": 0.014464613356609722, "grad_norm": 1.9737273512280982, "learning_rate": 9.632265717674972e-06, "loss": 0.5616, "step": 406 }, { "epoch": 0.014500240483103836, "grad_norm": 3.0057049877954594, "learning_rate": 9.655990510083037e-06, "loss": 0.6122, "step": 407 }, { "epoch": 0.014535867609597948, "grad_norm": 1.7968987354882324, "learning_rate": 9.679715302491104e-06, "loss": 0.5145, "step": 408 }, { "epoch": 0.01457149473609206, "grad_norm": 2.740621321668158, "learning_rate": 9.70344009489917e-06, "loss": 0.7473, "step": 409 }, { "epoch": 0.014607121862586173, "grad_norm": 2.015512557594076, "learning_rate": 9.727164887307237e-06, "loss": 0.566, "step": 410 }, { "epoch": 0.014642748989080285, "grad_norm": 3.1201588579260133, "learning_rate": 9.750889679715304e-06, "loss": 0.6452, "step": 411 }, { "epoch": 0.014678376115574399, "grad_norm": 2.1782923022772107, "learning_rate": 9.774614472123369e-06, "loss": 0.5689, "step": 412 }, { "epoch": 0.014714003242068511, "grad_norm": 2.002810728794198, "learning_rate": 9.798339264531436e-06, "loss": 0.5577, "step": 413 }, { "epoch": 0.014749630368562623, "grad_norm": 4.369046214963852, "learning_rate": 9.822064056939501e-06, "loss": 0.7346, "step": 414 }, { "epoch": 0.014785257495056736, "grad_norm": 1.9791131223032286, "learning_rate": 9.84578884934757e-06, "loss": 0.5723, "step": 415 }, { "epoch": 0.014820884621550848, "grad_norm": 1.9989708622892834, "learning_rate": 9.869513641755635e-06, "loss": 0.6225, "step": 416 }, { "epoch": 0.014856511748044962, "grad_norm": 2.150660414587732, "learning_rate": 9.893238434163703e-06, "loss": 0.5913, "step": 417 }, { "epoch": 0.014892138874539074, "grad_norm": 2.021897744053346, "learning_rate": 9.916963226571768e-06, "loss": 0.5361, "step": 418 }, { "epoch": 0.014927766001033186, "grad_norm": 3.4766455818675435, "learning_rate": 9.940688018979835e-06, "loss": 0.7197, "step": 419 }, { "epoch": 0.0149633931275273, "grad_norm": 2.77085686335957, "learning_rate": 9.964412811387902e-06, "loss": 0.6506, "step": 420 }, { "epoch": 0.014999020254021412, "grad_norm": 3.026201867068394, "learning_rate": 9.988137603795967e-06, "loss": 0.6775, "step": 421 }, { "epoch": 0.015034647380515525, "grad_norm": 2.6805176440589213, "learning_rate": 1.0011862396204034e-05, "loss": 0.6635, "step": 422 }, { "epoch": 0.015070274507009637, "grad_norm": 2.1246928774500478, "learning_rate": 1.0035587188612101e-05, "loss": 0.6015, "step": 423 }, { "epoch": 0.01510590163350375, "grad_norm": 3.825292395010679, "learning_rate": 1.0059311981020168e-05, "loss": 0.6367, "step": 424 }, { "epoch": 0.015141528759997863, "grad_norm": 2.1696588379471753, "learning_rate": 1.0083036773428234e-05, "loss": 0.5458, "step": 425 }, { "epoch": 0.015177155886491975, "grad_norm": 3.4064190933975813, "learning_rate": 1.01067615658363e-05, "loss": 0.6532, "step": 426 }, { "epoch": 0.015212783012986088, "grad_norm": 1.890221137685816, "learning_rate": 1.0130486358244366e-05, "loss": 0.6016, "step": 427 }, { "epoch": 0.0152484101394802, "grad_norm": 1.8762547225361894, "learning_rate": 1.0154211150652433e-05, "loss": 0.5616, "step": 428 }, { "epoch": 0.015284037265974312, "grad_norm": 4.33075120280541, "learning_rate": 1.01779359430605e-05, "loss": 0.6388, "step": 429 }, { "epoch": 0.015319664392468426, "grad_norm": 2.1215872674341, "learning_rate": 1.0201660735468566e-05, "loss": 0.6413, "step": 430 }, { "epoch": 0.015355291518962538, "grad_norm": 3.5144806012806367, "learning_rate": 1.0225385527876633e-05, "loss": 0.6684, "step": 431 }, { "epoch": 0.01539091864545665, "grad_norm": 2.76710521082718, "learning_rate": 1.0249110320284698e-05, "loss": 0.6485, "step": 432 }, { "epoch": 0.015426545771950764, "grad_norm": 2.950479740553892, "learning_rate": 1.0272835112692765e-05, "loss": 0.6979, "step": 433 }, { "epoch": 0.015462172898444876, "grad_norm": 2.4425103099937187, "learning_rate": 1.0296559905100832e-05, "loss": 0.5236, "step": 434 }, { "epoch": 0.01549780002493899, "grad_norm": 1.906787859387233, "learning_rate": 1.0320284697508897e-05, "loss": 0.5191, "step": 435 }, { "epoch": 0.015533427151433101, "grad_norm": 3.318575585480275, "learning_rate": 1.0344009489916964e-05, "loss": 0.7132, "step": 436 }, { "epoch": 0.015569054277927213, "grad_norm": 2.1356007077895165, "learning_rate": 1.036773428232503e-05, "loss": 0.5327, "step": 437 }, { "epoch": 0.015604681404421327, "grad_norm": 3.263196319942567, "learning_rate": 1.0391459074733097e-05, "loss": 0.7075, "step": 438 }, { "epoch": 0.01564030853091544, "grad_norm": 1.8562426182975478, "learning_rate": 1.0415183867141164e-05, "loss": 0.5514, "step": 439 }, { "epoch": 0.015675935657409552, "grad_norm": 2.596397446833789, "learning_rate": 1.043890865954923e-05, "loss": 0.6732, "step": 440 }, { "epoch": 0.015711562783903663, "grad_norm": 1.8892261468073261, "learning_rate": 1.0462633451957296e-05, "loss": 0.5876, "step": 441 }, { "epoch": 0.015747189910397776, "grad_norm": 1.987379504993165, "learning_rate": 1.0486358244365362e-05, "loss": 0.6401, "step": 442 }, { "epoch": 0.01578281703689189, "grad_norm": 2.0462718975261054, "learning_rate": 1.0510083036773429e-05, "loss": 0.5768, "step": 443 }, { "epoch": 0.015818444163386004, "grad_norm": 2.228980270134512, "learning_rate": 1.0533807829181496e-05, "loss": 0.5624, "step": 444 }, { "epoch": 0.015854071289880114, "grad_norm": 2.342262595771382, "learning_rate": 1.0557532621589561e-05, "loss": 0.5872, "step": 445 }, { "epoch": 0.015889698416374227, "grad_norm": 2.010260512189931, "learning_rate": 1.0581257413997628e-05, "loss": 0.5886, "step": 446 }, { "epoch": 0.01592532554286834, "grad_norm": 1.9572775851267903, "learning_rate": 1.0604982206405693e-05, "loss": 0.5718, "step": 447 }, { "epoch": 0.01596095266936245, "grad_norm": 2.3865390294867663, "learning_rate": 1.062870699881376e-05, "loss": 0.5744, "step": 448 }, { "epoch": 0.015996579795856565, "grad_norm": 2.314102629935604, "learning_rate": 1.0652431791221827e-05, "loss": 0.5951, "step": 449 }, { "epoch": 0.01603220692235068, "grad_norm": 2.622382165096803, "learning_rate": 1.0676156583629893e-05, "loss": 0.7676, "step": 450 }, { "epoch": 0.01606783404884479, "grad_norm": 7.56096264208087, "learning_rate": 1.0699881376037962e-05, "loss": 0.6121, "step": 451 }, { "epoch": 0.016103461175338903, "grad_norm": 3.276862439796909, "learning_rate": 1.0723606168446029e-05, "loss": 0.6323, "step": 452 }, { "epoch": 0.016139088301833016, "grad_norm": 1.9782449966420088, "learning_rate": 1.0747330960854094e-05, "loss": 0.5996, "step": 453 }, { "epoch": 0.01617471542832713, "grad_norm": 5.6900853705281875, "learning_rate": 1.0771055753262161e-05, "loss": 0.6472, "step": 454 }, { "epoch": 0.01621034255482124, "grad_norm": 8.500427039073601, "learning_rate": 1.0794780545670226e-05, "loss": 0.6528, "step": 455 }, { "epoch": 0.016245969681315354, "grad_norm": 2.2106710126016402, "learning_rate": 1.0818505338078293e-05, "loss": 0.5886, "step": 456 }, { "epoch": 0.016281596807809467, "grad_norm": 4.490833055314061, "learning_rate": 1.084223013048636e-05, "loss": 0.7109, "step": 457 }, { "epoch": 0.016317223934303578, "grad_norm": 2.8762516448688515, "learning_rate": 1.0865954922894426e-05, "loss": 0.7124, "step": 458 }, { "epoch": 0.01635285106079769, "grad_norm": 1.9596589364051729, "learning_rate": 1.0889679715302493e-05, "loss": 0.5739, "step": 459 }, { "epoch": 0.016388478187291805, "grad_norm": 3.3169298031109244, "learning_rate": 1.0913404507710558e-05, "loss": 0.631, "step": 460 }, { "epoch": 0.016424105313785915, "grad_norm": 2.57787054333587, "learning_rate": 1.0937129300118625e-05, "loss": 0.6577, "step": 461 }, { "epoch": 0.01645973244028003, "grad_norm": 2.6261616878418486, "learning_rate": 1.0960854092526692e-05, "loss": 0.6055, "step": 462 }, { "epoch": 0.016495359566774143, "grad_norm": 1.8744324905474397, "learning_rate": 1.0984578884934757e-05, "loss": 0.5593, "step": 463 }, { "epoch": 0.016530986693268253, "grad_norm": 2.947101771753413, "learning_rate": 1.1008303677342825e-05, "loss": 0.6522, "step": 464 }, { "epoch": 0.016566613819762366, "grad_norm": 3.247832681499135, "learning_rate": 1.103202846975089e-05, "loss": 0.6884, "step": 465 }, { "epoch": 0.01660224094625648, "grad_norm": 1.87325443963057, "learning_rate": 1.1055753262158957e-05, "loss": 0.5873, "step": 466 }, { "epoch": 0.016637868072750594, "grad_norm": 2.560275462093063, "learning_rate": 1.1079478054567024e-05, "loss": 0.6085, "step": 467 }, { "epoch": 0.016673495199244704, "grad_norm": 2.0702420159404467, "learning_rate": 1.110320284697509e-05, "loss": 0.5896, "step": 468 }, { "epoch": 0.016709122325738818, "grad_norm": 3.0861085643024824, "learning_rate": 1.1126927639383156e-05, "loss": 0.7107, "step": 469 }, { "epoch": 0.01674474945223293, "grad_norm": 3.71992100132832, "learning_rate": 1.1150652431791222e-05, "loss": 0.6826, "step": 470 }, { "epoch": 0.01678037657872704, "grad_norm": 2.2073805778877196, "learning_rate": 1.1174377224199289e-05, "loss": 0.6205, "step": 471 }, { "epoch": 0.016816003705221155, "grad_norm": 2.0720758099714423, "learning_rate": 1.1198102016607356e-05, "loss": 0.6285, "step": 472 }, { "epoch": 0.01685163083171527, "grad_norm": 3.0501461012720177, "learning_rate": 1.1221826809015421e-05, "loss": 0.6055, "step": 473 }, { "epoch": 0.01688725795820938, "grad_norm": 2.9130178017806805, "learning_rate": 1.1245551601423488e-05, "loss": 0.7263, "step": 474 }, { "epoch": 0.016922885084703493, "grad_norm": 2.1595138847890722, "learning_rate": 1.1269276393831553e-05, "loss": 0.5327, "step": 475 }, { "epoch": 0.016958512211197607, "grad_norm": 2.777099833588526, "learning_rate": 1.129300118623962e-05, "loss": 0.7477, "step": 476 }, { "epoch": 0.01699413933769172, "grad_norm": 1.9897445936211369, "learning_rate": 1.1316725978647688e-05, "loss": 0.5645, "step": 477 }, { "epoch": 0.01702976646418583, "grad_norm": 2.0939117051622915, "learning_rate": 1.1340450771055753e-05, "loss": 0.5543, "step": 478 }, { "epoch": 0.017065393590679944, "grad_norm": 2.0186105326724753, "learning_rate": 1.136417556346382e-05, "loss": 0.5385, "step": 479 }, { "epoch": 0.017101020717174058, "grad_norm": 1.9200933679211485, "learning_rate": 1.1387900355871889e-05, "loss": 0.517, "step": 480 }, { "epoch": 0.017136647843668168, "grad_norm": 2.0475067899872306, "learning_rate": 1.1411625148279954e-05, "loss": 0.5722, "step": 481 }, { "epoch": 0.01717227497016228, "grad_norm": 2.806845195340794, "learning_rate": 1.1435349940688021e-05, "loss": 0.6772, "step": 482 }, { "epoch": 0.017207902096656395, "grad_norm": 3.1073501185824264, "learning_rate": 1.1459074733096086e-05, "loss": 0.7194, "step": 483 }, { "epoch": 0.017243529223150506, "grad_norm": 3.075821049437625, "learning_rate": 1.1482799525504153e-05, "loss": 0.7038, "step": 484 }, { "epoch": 0.01727915634964462, "grad_norm": 3.028239769610335, "learning_rate": 1.150652431791222e-05, "loss": 0.6517, "step": 485 }, { "epoch": 0.017314783476138733, "grad_norm": 2.32050332614693, "learning_rate": 1.1530249110320286e-05, "loss": 0.6147, "step": 486 }, { "epoch": 0.017350410602632843, "grad_norm": 1.8348362824925926, "learning_rate": 1.1553973902728353e-05, "loss": 0.5501, "step": 487 }, { "epoch": 0.017386037729126957, "grad_norm": 2.3540605313124794, "learning_rate": 1.1577698695136418e-05, "loss": 0.5924, "step": 488 }, { "epoch": 0.01742166485562107, "grad_norm": 5.583019355953897, "learning_rate": 1.1601423487544485e-05, "loss": 0.6976, "step": 489 }, { "epoch": 0.017457291982115184, "grad_norm": 3.0846722957750927, "learning_rate": 1.1625148279952552e-05, "loss": 0.5039, "step": 490 }, { "epoch": 0.017492919108609294, "grad_norm": 3.1156735814890224, "learning_rate": 1.1648873072360618e-05, "loss": 0.5801, "step": 491 }, { "epoch": 0.017528546235103408, "grad_norm": 3.5322076452389632, "learning_rate": 1.1672597864768685e-05, "loss": 0.7169, "step": 492 }, { "epoch": 0.01756417336159752, "grad_norm": 2.5784463391418395, "learning_rate": 1.169632265717675e-05, "loss": 0.6291, "step": 493 }, { "epoch": 0.017599800488091632, "grad_norm": 4.223527106812036, "learning_rate": 1.1720047449584817e-05, "loss": 0.5653, "step": 494 }, { "epoch": 0.017635427614585746, "grad_norm": 3.2114746247865273, "learning_rate": 1.1743772241992884e-05, "loss": 0.6462, "step": 495 }, { "epoch": 0.01767105474107986, "grad_norm": 2.1363681496931632, "learning_rate": 1.176749703440095e-05, "loss": 0.5974, "step": 496 }, { "epoch": 0.01770668186757397, "grad_norm": 1.852911078366921, "learning_rate": 1.1791221826809016e-05, "loss": 0.5431, "step": 497 }, { "epoch": 0.017742308994068083, "grad_norm": 2.2179353834719424, "learning_rate": 1.1814946619217082e-05, "loss": 0.5662, "step": 498 }, { "epoch": 0.017777936120562197, "grad_norm": 2.0115935656314, "learning_rate": 1.1838671411625149e-05, "loss": 0.5239, "step": 499 }, { "epoch": 0.017813563247056307, "grad_norm": 3.282075774327233, "learning_rate": 1.1862396204033216e-05, "loss": 0.5983, "step": 500 }, { "epoch": 0.01784919037355042, "grad_norm": 2.0614960839484753, "learning_rate": 1.1886120996441281e-05, "loss": 0.5568, "step": 501 }, { "epoch": 0.017884817500044534, "grad_norm": 2.3699697130937953, "learning_rate": 1.1909845788849348e-05, "loss": 0.631, "step": 502 }, { "epoch": 0.017920444626538648, "grad_norm": 3.8082604936699758, "learning_rate": 1.1933570581257414e-05, "loss": 0.6665, "step": 503 }, { "epoch": 0.017956071753032758, "grad_norm": 3.409640247416419, "learning_rate": 1.195729537366548e-05, "loss": 0.7607, "step": 504 }, { "epoch": 0.017991698879526872, "grad_norm": 2.268896414707566, "learning_rate": 1.1981020166073548e-05, "loss": 0.6208, "step": 505 }, { "epoch": 0.018027326006020986, "grad_norm": 3.0333818957257614, "learning_rate": 1.2004744958481613e-05, "loss": 0.624, "step": 506 }, { "epoch": 0.018062953132515096, "grad_norm": 3.0780579218752666, "learning_rate": 1.202846975088968e-05, "loss": 0.5956, "step": 507 }, { "epoch": 0.01809858025900921, "grad_norm": 2.021085997421653, "learning_rate": 1.2052194543297745e-05, "loss": 0.5643, "step": 508 }, { "epoch": 0.018134207385503323, "grad_norm": 2.2023476973401444, "learning_rate": 1.2075919335705814e-05, "loss": 0.5772, "step": 509 }, { "epoch": 0.018169834511997433, "grad_norm": 2.939265531511997, "learning_rate": 1.2099644128113881e-05, "loss": 0.6263, "step": 510 }, { "epoch": 0.018205461638491547, "grad_norm": 2.089423784720356, "learning_rate": 1.2123368920521947e-05, "loss": 0.5949, "step": 511 }, { "epoch": 0.01824108876498566, "grad_norm": 1.8677543741422513, "learning_rate": 1.2147093712930014e-05, "loss": 0.5935, "step": 512 }, { "epoch": 0.018276715891479774, "grad_norm": 1.7893967090938703, "learning_rate": 1.217081850533808e-05, "loss": 0.5343, "step": 513 }, { "epoch": 0.018312343017973885, "grad_norm": 2.228248823373795, "learning_rate": 1.2194543297746146e-05, "loss": 0.6408, "step": 514 }, { "epoch": 0.018347970144467998, "grad_norm": 2.147890149501066, "learning_rate": 1.2218268090154213e-05, "loss": 0.6501, "step": 515 }, { "epoch": 0.018383597270962112, "grad_norm": 1.92324549743403, "learning_rate": 1.2241992882562278e-05, "loss": 0.563, "step": 516 }, { "epoch": 0.018419224397456222, "grad_norm": 2.3529102050006037, "learning_rate": 1.2265717674970345e-05, "loss": 0.6014, "step": 517 }, { "epoch": 0.018454851523950336, "grad_norm": 2.658935580141421, "learning_rate": 1.2289442467378412e-05, "loss": 0.6856, "step": 518 }, { "epoch": 0.01849047865044445, "grad_norm": 2.4216797442101865, "learning_rate": 1.2313167259786478e-05, "loss": 0.7021, "step": 519 }, { "epoch": 0.01852610577693856, "grad_norm": 3.036224896872677, "learning_rate": 1.2336892052194545e-05, "loss": 0.6122, "step": 520 }, { "epoch": 0.018561732903432673, "grad_norm": 2.2286246038989956, "learning_rate": 1.236061684460261e-05, "loss": 0.573, "step": 521 }, { "epoch": 0.018597360029926787, "grad_norm": 2.3208795733290875, "learning_rate": 1.2384341637010677e-05, "loss": 0.5626, "step": 522 }, { "epoch": 0.018632987156420897, "grad_norm": 2.8745974394042757, "learning_rate": 1.2408066429418744e-05, "loss": 0.7501, "step": 523 }, { "epoch": 0.01866861428291501, "grad_norm": 4.026778662458885, "learning_rate": 1.243179122182681e-05, "loss": 0.6691, "step": 524 }, { "epoch": 0.018704241409409125, "grad_norm": 1.9189702631532728, "learning_rate": 1.2455516014234877e-05, "loss": 0.5924, "step": 525 }, { "epoch": 0.018739868535903238, "grad_norm": 2.592740133869991, "learning_rate": 1.2479240806642942e-05, "loss": 0.7327, "step": 526 }, { "epoch": 0.01877549566239735, "grad_norm": 3.4602748244336654, "learning_rate": 1.2502965599051009e-05, "loss": 0.6953, "step": 527 }, { "epoch": 0.018811122788891462, "grad_norm": 3.3481305898261784, "learning_rate": 1.2526690391459076e-05, "loss": 0.6726, "step": 528 }, { "epoch": 0.018846749915385576, "grad_norm": 2.474749170102465, "learning_rate": 1.2550415183867141e-05, "loss": 0.5528, "step": 529 }, { "epoch": 0.018882377041879686, "grad_norm": 2.0098631123439477, "learning_rate": 1.2574139976275208e-05, "loss": 0.5751, "step": 530 }, { "epoch": 0.0189180041683738, "grad_norm": 2.798066893513132, "learning_rate": 1.2597864768683274e-05, "loss": 0.6855, "step": 531 }, { "epoch": 0.018953631294867913, "grad_norm": 2.9533138038122133, "learning_rate": 1.262158956109134e-05, "loss": 0.6539, "step": 532 }, { "epoch": 0.018989258421362024, "grad_norm": 2.675066262780007, "learning_rate": 1.2645314353499408e-05, "loss": 0.7052, "step": 533 }, { "epoch": 0.019024885547856137, "grad_norm": 2.5309659122172636, "learning_rate": 1.2669039145907473e-05, "loss": 0.651, "step": 534 }, { "epoch": 0.01906051267435025, "grad_norm": 2.427303000783682, "learning_rate": 1.269276393831554e-05, "loss": 0.6689, "step": 535 }, { "epoch": 0.019096139800844365, "grad_norm": 3.5730175038157115, "learning_rate": 1.2716488730723606e-05, "loss": 0.7821, "step": 536 }, { "epoch": 0.019131766927338475, "grad_norm": 3.140148159091097, "learning_rate": 1.2740213523131673e-05, "loss": 0.7533, "step": 537 }, { "epoch": 0.01916739405383259, "grad_norm": 2.539501332268524, "learning_rate": 1.2763938315539741e-05, "loss": 0.75, "step": 538 }, { "epoch": 0.019203021180326702, "grad_norm": 4.833785241140187, "learning_rate": 1.2787663107947808e-05, "loss": 0.6693, "step": 539 }, { "epoch": 0.019238648306820812, "grad_norm": 2.620023423809915, "learning_rate": 1.2811387900355874e-05, "loss": 0.5737, "step": 540 }, { "epoch": 0.019274275433314926, "grad_norm": 2.2955765818527043, "learning_rate": 1.283511269276394e-05, "loss": 0.6477, "step": 541 }, { "epoch": 0.01930990255980904, "grad_norm": 2.6778555933039763, "learning_rate": 1.2858837485172006e-05, "loss": 0.6059, "step": 542 }, { "epoch": 0.01934552968630315, "grad_norm": 2.1631154298911337, "learning_rate": 1.2882562277580073e-05, "loss": 0.5408, "step": 543 }, { "epoch": 0.019381156812797264, "grad_norm": 3.195502092700627, "learning_rate": 1.2906287069988138e-05, "loss": 0.6424, "step": 544 }, { "epoch": 0.019416783939291377, "grad_norm": 2.301034826879702, "learning_rate": 1.2930011862396206e-05, "loss": 0.6099, "step": 545 }, { "epoch": 0.019452411065785487, "grad_norm": 4.692575634649482, "learning_rate": 1.2953736654804273e-05, "loss": 0.6586, "step": 546 }, { "epoch": 0.0194880381922796, "grad_norm": 1.8004439473855651, "learning_rate": 1.2977461447212338e-05, "loss": 0.591, "step": 547 }, { "epoch": 0.019523665318773715, "grad_norm": 1.866246857284383, "learning_rate": 1.3001186239620405e-05, "loss": 0.6158, "step": 548 }, { "epoch": 0.01955929244526783, "grad_norm": 1.9594365075331772, "learning_rate": 1.302491103202847e-05, "loss": 0.6267, "step": 549 }, { "epoch": 0.01959491957176194, "grad_norm": 4.539019985654841, "learning_rate": 1.3048635824436537e-05, "loss": 0.7137, "step": 550 }, { "epoch": 0.019630546698256052, "grad_norm": 3.7604758930713067, "learning_rate": 1.3072360616844604e-05, "loss": 0.7014, "step": 551 }, { "epoch": 0.019666173824750166, "grad_norm": 1.9239877514314232, "learning_rate": 1.309608540925267e-05, "loss": 0.6102, "step": 552 }, { "epoch": 0.019701800951244276, "grad_norm": 1.95546664948337, "learning_rate": 1.3119810201660737e-05, "loss": 0.5859, "step": 553 }, { "epoch": 0.01973742807773839, "grad_norm": 1.8810099098513655, "learning_rate": 1.3143534994068802e-05, "loss": 0.5412, "step": 554 }, { "epoch": 0.019773055204232504, "grad_norm": 1.8209406644281918, "learning_rate": 1.3167259786476869e-05, "loss": 0.5438, "step": 555 }, { "epoch": 0.019808682330726614, "grad_norm": 3.9831522834955497, "learning_rate": 1.3190984578884936e-05, "loss": 0.6994, "step": 556 }, { "epoch": 0.019844309457220728, "grad_norm": 2.9049124843392455, "learning_rate": 1.3214709371293001e-05, "loss": 0.6334, "step": 557 }, { "epoch": 0.01987993658371484, "grad_norm": 5.141494100481329, "learning_rate": 1.3238434163701069e-05, "loss": 0.6125, "step": 558 }, { "epoch": 0.01991556371020895, "grad_norm": 2.785890545936064, "learning_rate": 1.3262158956109134e-05, "loss": 0.6658, "step": 559 }, { "epoch": 0.019951190836703065, "grad_norm": 3.4892391340572506, "learning_rate": 1.3285883748517201e-05, "loss": 0.7217, "step": 560 }, { "epoch": 0.01998681796319718, "grad_norm": 2.083588094621621, "learning_rate": 1.3309608540925268e-05, "loss": 0.623, "step": 561 }, { "epoch": 0.020022445089691292, "grad_norm": 2.7798688052431135, "learning_rate": 1.3333333333333333e-05, "loss": 0.6886, "step": 562 }, { "epoch": 0.020058072216185403, "grad_norm": 2.053151060277579, "learning_rate": 1.33570581257414e-05, "loss": 0.5806, "step": 563 }, { "epoch": 0.020093699342679516, "grad_norm": 1.9763925483793512, "learning_rate": 1.3380782918149466e-05, "loss": 0.575, "step": 564 }, { "epoch": 0.02012932646917363, "grad_norm": 3.1632902050497784, "learning_rate": 1.3404507710557533e-05, "loss": 0.5619, "step": 565 }, { "epoch": 0.02016495359566774, "grad_norm": 2.800927024445949, "learning_rate": 1.34282325029656e-05, "loss": 0.6909, "step": 566 }, { "epoch": 0.020200580722161854, "grad_norm": 1.8718986900227004, "learning_rate": 1.3451957295373668e-05, "loss": 0.5474, "step": 567 }, { "epoch": 0.020236207848655968, "grad_norm": 2.4834424004113353, "learning_rate": 1.3475682087781734e-05, "loss": 0.6254, "step": 568 }, { "epoch": 0.020271834975150078, "grad_norm": 1.781611727622813, "learning_rate": 1.3499406880189801e-05, "loss": 0.5899, "step": 569 }, { "epoch": 0.02030746210164419, "grad_norm": 1.6975214027564343, "learning_rate": 1.3523131672597866e-05, "loss": 0.5438, "step": 570 }, { "epoch": 0.020343089228138305, "grad_norm": 1.7702948951622777, "learning_rate": 1.3546856465005933e-05, "loss": 0.5435, "step": 571 }, { "epoch": 0.02037871635463242, "grad_norm": 1.957323248765506, "learning_rate": 1.3570581257414e-05, "loss": 0.5999, "step": 572 }, { "epoch": 0.02041434348112653, "grad_norm": 1.8962850062689598, "learning_rate": 1.3594306049822066e-05, "loss": 0.5718, "step": 573 }, { "epoch": 0.020449970607620643, "grad_norm": 2.699437898785684, "learning_rate": 1.3618030842230133e-05, "loss": 0.5815, "step": 574 }, { "epoch": 0.020485597734114756, "grad_norm": 2.013434873788181, "learning_rate": 1.3641755634638198e-05, "loss": 0.5771, "step": 575 }, { "epoch": 0.020521224860608867, "grad_norm": 1.9017218645039455, "learning_rate": 1.3665480427046265e-05, "loss": 0.5566, "step": 576 }, { "epoch": 0.02055685198710298, "grad_norm": 2.0839821671386165, "learning_rate": 1.368920521945433e-05, "loss": 0.59, "step": 577 }, { "epoch": 0.020592479113597094, "grad_norm": 2.944005582145907, "learning_rate": 1.3712930011862397e-05, "loss": 0.6561, "step": 578 }, { "epoch": 0.020628106240091204, "grad_norm": 1.9601168868897718, "learning_rate": 1.3736654804270464e-05, "loss": 0.5863, "step": 579 }, { "epoch": 0.020663733366585318, "grad_norm": 1.9384075178541795, "learning_rate": 1.376037959667853e-05, "loss": 0.5193, "step": 580 }, { "epoch": 0.02069936049307943, "grad_norm": 1.9734817193716396, "learning_rate": 1.3784104389086597e-05, "loss": 0.5629, "step": 581 }, { "epoch": 0.02073498761957354, "grad_norm": 2.3683425405459495, "learning_rate": 1.3807829181494662e-05, "loss": 0.5381, "step": 582 }, { "epoch": 0.020770614746067655, "grad_norm": 1.8804972945061125, "learning_rate": 1.383155397390273e-05, "loss": 0.5429, "step": 583 }, { "epoch": 0.02080624187256177, "grad_norm": 1.757775921661659, "learning_rate": 1.3855278766310796e-05, "loss": 0.5681, "step": 584 }, { "epoch": 0.020841868999055883, "grad_norm": 1.7163834682640495, "learning_rate": 1.3879003558718862e-05, "loss": 0.5627, "step": 585 }, { "epoch": 0.020877496125549993, "grad_norm": 2.518775178307271, "learning_rate": 1.3902728351126929e-05, "loss": 0.6695, "step": 586 }, { "epoch": 0.020913123252044107, "grad_norm": 2.483721434780657, "learning_rate": 1.3926453143534994e-05, "loss": 0.7241, "step": 587 }, { "epoch": 0.02094875037853822, "grad_norm": 3.1317366462367766, "learning_rate": 1.3950177935943061e-05, "loss": 0.608, "step": 588 }, { "epoch": 0.02098437750503233, "grad_norm": 1.9750611684605002, "learning_rate": 1.3973902728351128e-05, "loss": 0.5827, "step": 589 }, { "epoch": 0.021020004631526444, "grad_norm": 3.8914091197549743, "learning_rate": 1.3997627520759193e-05, "loss": 0.7913, "step": 590 }, { "epoch": 0.021055631758020558, "grad_norm": 2.8182603003091473, "learning_rate": 1.402135231316726e-05, "loss": 0.5788, "step": 591 }, { "epoch": 0.021091258884514668, "grad_norm": 1.8125177812461082, "learning_rate": 1.4045077105575326e-05, "loss": 0.5583, "step": 592 }, { "epoch": 0.02112688601100878, "grad_norm": 2.8496290030719797, "learning_rate": 1.4068801897983393e-05, "loss": 0.5896, "step": 593 }, { "epoch": 0.021162513137502895, "grad_norm": 1.7962797995087232, "learning_rate": 1.409252669039146e-05, "loss": 0.5315, "step": 594 }, { "epoch": 0.02119814026399701, "grad_norm": 2.047113043457379, "learning_rate": 1.4116251482799525e-05, "loss": 0.5873, "step": 595 }, { "epoch": 0.02123376739049112, "grad_norm": 2.455751028940288, "learning_rate": 1.4139976275207594e-05, "loss": 0.5005, "step": 596 }, { "epoch": 0.021269394516985233, "grad_norm": 2.5797001374933712, "learning_rate": 1.4163701067615661e-05, "loss": 0.6064, "step": 597 }, { "epoch": 0.021305021643479347, "grad_norm": 2.6042154938046087, "learning_rate": 1.4187425860023726e-05, "loss": 0.6901, "step": 598 }, { "epoch": 0.021340648769973457, "grad_norm": 1.9486695915749428, "learning_rate": 1.4211150652431793e-05, "loss": 0.5825, "step": 599 }, { "epoch": 0.02137627589646757, "grad_norm": 2.3433480135542144, "learning_rate": 1.423487544483986e-05, "loss": 0.6177, "step": 600 }, { "epoch": 0.021411903022961684, "grad_norm": 3.761376810816491, "learning_rate": 1.4258600237247926e-05, "loss": 0.716, "step": 601 }, { "epoch": 0.021447530149455794, "grad_norm": 2.8214571797734633, "learning_rate": 1.4282325029655993e-05, "loss": 0.612, "step": 602 }, { "epoch": 0.021483157275949908, "grad_norm": 1.8958223498233195, "learning_rate": 1.4306049822064058e-05, "loss": 0.5479, "step": 603 }, { "epoch": 0.02151878440244402, "grad_norm": 1.96575085133209, "learning_rate": 1.4329774614472125e-05, "loss": 0.6093, "step": 604 }, { "epoch": 0.021554411528938132, "grad_norm": 2.435982839312655, "learning_rate": 1.4353499406880192e-05, "loss": 0.7071, "step": 605 }, { "epoch": 0.021590038655432246, "grad_norm": 1.6118536338114222, "learning_rate": 1.4377224199288258e-05, "loss": 0.5437, "step": 606 }, { "epoch": 0.02162566578192636, "grad_norm": 2.536754949988605, "learning_rate": 1.4400948991696325e-05, "loss": 0.6628, "step": 607 }, { "epoch": 0.021661292908420473, "grad_norm": 1.7893854386836012, "learning_rate": 1.442467378410439e-05, "loss": 0.5779, "step": 608 }, { "epoch": 0.021696920034914583, "grad_norm": 2.4369341393233155, "learning_rate": 1.4448398576512457e-05, "loss": 0.6882, "step": 609 }, { "epoch": 0.021732547161408697, "grad_norm": 2.1289024884554664, "learning_rate": 1.4472123368920522e-05, "loss": 0.5887, "step": 610 }, { "epoch": 0.02176817428790281, "grad_norm": 5.366879083682593, "learning_rate": 1.449584816132859e-05, "loss": 0.6718, "step": 611 }, { "epoch": 0.02180380141439692, "grad_norm": 2.7097047225562307, "learning_rate": 1.4519572953736656e-05, "loss": 0.7023, "step": 612 }, { "epoch": 0.021839428540891034, "grad_norm": 2.088675428742109, "learning_rate": 1.4543297746144722e-05, "loss": 0.5879, "step": 613 }, { "epoch": 0.021875055667385148, "grad_norm": 2.003241656071187, "learning_rate": 1.4567022538552789e-05, "loss": 0.6167, "step": 614 }, { "epoch": 0.021910682793879258, "grad_norm": 1.995708351392957, "learning_rate": 1.4590747330960854e-05, "loss": 0.645, "step": 615 }, { "epoch": 0.021946309920373372, "grad_norm": 1.8717359519181906, "learning_rate": 1.4614472123368921e-05, "loss": 0.5718, "step": 616 }, { "epoch": 0.021981937046867486, "grad_norm": 2.705264234535648, "learning_rate": 1.4638196915776988e-05, "loss": 0.647, "step": 617 }, { "epoch": 0.022017564173361596, "grad_norm": 1.9204884150806207, "learning_rate": 1.4661921708185054e-05, "loss": 0.5836, "step": 618 }, { "epoch": 0.02205319129985571, "grad_norm": 1.9479462274983155, "learning_rate": 1.468564650059312e-05, "loss": 0.6004, "step": 619 }, { "epoch": 0.022088818426349823, "grad_norm": 2.8208813406464155, "learning_rate": 1.4709371293001186e-05, "loss": 0.6546, "step": 620 }, { "epoch": 0.022124445552843937, "grad_norm": 3.1773470788051075, "learning_rate": 1.4733096085409253e-05, "loss": 0.5638, "step": 621 }, { "epoch": 0.022160072679338047, "grad_norm": 1.7313859119322084, "learning_rate": 1.475682087781732e-05, "loss": 0.5552, "step": 622 }, { "epoch": 0.02219569980583216, "grad_norm": 1.8902111780163078, "learning_rate": 1.4780545670225385e-05, "loss": 0.5567, "step": 623 }, { "epoch": 0.022231326932326274, "grad_norm": 1.8552073140695315, "learning_rate": 1.4804270462633452e-05, "loss": 0.6183, "step": 624 }, { "epoch": 0.022266954058820385, "grad_norm": 1.8528715932556765, "learning_rate": 1.4827995255041521e-05, "loss": 0.5477, "step": 625 }, { "epoch": 0.022302581185314498, "grad_norm": 1.665328653603288, "learning_rate": 1.4851720047449586e-05, "loss": 0.5454, "step": 626 }, { "epoch": 0.022338208311808612, "grad_norm": 1.6516473153836306, "learning_rate": 1.4875444839857654e-05, "loss": 0.616, "step": 627 }, { "epoch": 0.022373835438302722, "grad_norm": 1.7448502139714441, "learning_rate": 1.489916963226572e-05, "loss": 0.5924, "step": 628 }, { "epoch": 0.022409462564796836, "grad_norm": 2.638209247394455, "learning_rate": 1.4922894424673786e-05, "loss": 0.6968, "step": 629 }, { "epoch": 0.02244508969129095, "grad_norm": 3.105015777243513, "learning_rate": 1.4946619217081853e-05, "loss": 0.6679, "step": 630 }, { "epoch": 0.022480716817785063, "grad_norm": 1.7860761180372495, "learning_rate": 1.4970344009489918e-05, "loss": 0.6185, "step": 631 }, { "epoch": 0.022516343944279173, "grad_norm": 1.8993481124913352, "learning_rate": 1.4994068801897985e-05, "loss": 0.5889, "step": 632 }, { "epoch": 0.022551971070773287, "grad_norm": 1.8234285181590257, "learning_rate": 1.5017793594306052e-05, "loss": 0.5857, "step": 633 }, { "epoch": 0.0225875981972674, "grad_norm": 1.9750227545754104, "learning_rate": 1.5041518386714118e-05, "loss": 0.5711, "step": 634 }, { "epoch": 0.02262322532376151, "grad_norm": 3.158006127296657, "learning_rate": 1.5065243179122185e-05, "loss": 0.6654, "step": 635 }, { "epoch": 0.022658852450255625, "grad_norm": 3.028074988908904, "learning_rate": 1.508896797153025e-05, "loss": 0.7859, "step": 636 }, { "epoch": 0.02269447957674974, "grad_norm": 1.8614233784835796, "learning_rate": 1.5112692763938317e-05, "loss": 0.5518, "step": 637 }, { "epoch": 0.02273010670324385, "grad_norm": 5.5969376738172345, "learning_rate": 1.5136417556346384e-05, "loss": 0.7012, "step": 638 }, { "epoch": 0.022765733829737962, "grad_norm": 1.8941144782797814, "learning_rate": 1.516014234875445e-05, "loss": 0.598, "step": 639 }, { "epoch": 0.022801360956232076, "grad_norm": 3.0845153191324792, "learning_rate": 1.5183867141162517e-05, "loss": 0.5834, "step": 640 }, { "epoch": 0.022836988082726186, "grad_norm": 3.1754998001515595, "learning_rate": 1.5207591933570582e-05, "loss": 0.7226, "step": 641 }, { "epoch": 0.0228726152092203, "grad_norm": 1.7949989751369502, "learning_rate": 1.5231316725978649e-05, "loss": 0.5627, "step": 642 }, { "epoch": 0.022908242335714413, "grad_norm": 2.560297409233319, "learning_rate": 1.5255041518386714e-05, "loss": 0.633, "step": 643 }, { "epoch": 0.022943869462208527, "grad_norm": 3.0484046068605433, "learning_rate": 1.5278766310794783e-05, "loss": 0.6966, "step": 644 }, { "epoch": 0.022979496588702637, "grad_norm": 2.874074089071478, "learning_rate": 1.5302491103202847e-05, "loss": 0.6745, "step": 645 }, { "epoch": 0.02301512371519675, "grad_norm": 1.7211660925433936, "learning_rate": 1.5326215895610914e-05, "loss": 0.5796, "step": 646 }, { "epoch": 0.023050750841690865, "grad_norm": 2.649228007848976, "learning_rate": 1.534994068801898e-05, "loss": 0.6695, "step": 647 }, { "epoch": 0.023086377968184975, "grad_norm": 2.1325077375004797, "learning_rate": 1.5373665480427048e-05, "loss": 0.5268, "step": 648 }, { "epoch": 0.02312200509467909, "grad_norm": 1.8924507925976572, "learning_rate": 1.5397390272835115e-05, "loss": 0.5879, "step": 649 }, { "epoch": 0.023157632221173202, "grad_norm": 2.0044781448959195, "learning_rate": 1.542111506524318e-05, "loss": 0.5705, "step": 650 }, { "epoch": 0.023193259347667312, "grad_norm": 3.1627915838294243, "learning_rate": 1.5444839857651245e-05, "loss": 0.7237, "step": 651 }, { "epoch": 0.023228886474161426, "grad_norm": 2.385266956605564, "learning_rate": 1.5468564650059313e-05, "loss": 0.6365, "step": 652 }, { "epoch": 0.02326451360065554, "grad_norm": 2.0209793344294833, "learning_rate": 1.549228944246738e-05, "loss": 0.5776, "step": 653 }, { "epoch": 0.023300140727149653, "grad_norm": 3.226494562188484, "learning_rate": 1.5516014234875447e-05, "loss": 0.6367, "step": 654 }, { "epoch": 0.023335767853643764, "grad_norm": 1.681481267269964, "learning_rate": 1.5539739027283514e-05, "loss": 0.5662, "step": 655 }, { "epoch": 0.023371394980137877, "grad_norm": 1.8258622289295061, "learning_rate": 1.556346381969158e-05, "loss": 0.5699, "step": 656 }, { "epoch": 0.02340702210663199, "grad_norm": 3.1773206337170588, "learning_rate": 1.5587188612099648e-05, "loss": 0.6062, "step": 657 }, { "epoch": 0.0234426492331261, "grad_norm": 1.792115300516738, "learning_rate": 1.561091340450771e-05, "loss": 0.5633, "step": 658 }, { "epoch": 0.023478276359620215, "grad_norm": 2.6789414427323512, "learning_rate": 1.563463819691578e-05, "loss": 0.6032, "step": 659 }, { "epoch": 0.02351390348611433, "grad_norm": 2.012005746967658, "learning_rate": 1.5658362989323845e-05, "loss": 0.5518, "step": 660 }, { "epoch": 0.02354953061260844, "grad_norm": 3.69938631512234, "learning_rate": 1.5682087781731912e-05, "loss": 0.6534, "step": 661 }, { "epoch": 0.023585157739102552, "grad_norm": 2.4659187548477153, "learning_rate": 1.570581257413998e-05, "loss": 0.6977, "step": 662 }, { "epoch": 0.023620784865596666, "grad_norm": 2.552798346830395, "learning_rate": 1.5729537366548043e-05, "loss": 0.7088, "step": 663 }, { "epoch": 0.023656411992090776, "grad_norm": 2.1715153466077526, "learning_rate": 1.575326215895611e-05, "loss": 0.5688, "step": 664 }, { "epoch": 0.02369203911858489, "grad_norm": 2.2022991990156084, "learning_rate": 1.5776986951364177e-05, "loss": 0.6629, "step": 665 }, { "epoch": 0.023727666245079004, "grad_norm": 1.965480021952743, "learning_rate": 1.5800711743772244e-05, "loss": 0.5664, "step": 666 }, { "epoch": 0.023763293371573117, "grad_norm": 2.323703567402514, "learning_rate": 1.582443653618031e-05, "loss": 0.5923, "step": 667 }, { "epoch": 0.023798920498067228, "grad_norm": 2.7959153986792193, "learning_rate": 1.5848161328588375e-05, "loss": 0.657, "step": 668 }, { "epoch": 0.02383454762456134, "grad_norm": 3.168517769124534, "learning_rate": 1.5871886120996442e-05, "loss": 0.6893, "step": 669 }, { "epoch": 0.023870174751055455, "grad_norm": 1.9750705104014128, "learning_rate": 1.589561091340451e-05, "loss": 0.5493, "step": 670 }, { "epoch": 0.023905801877549565, "grad_norm": 3.5563743558456227, "learning_rate": 1.5919335705812576e-05, "loss": 0.7002, "step": 671 }, { "epoch": 0.02394142900404368, "grad_norm": 2.3662202276795106, "learning_rate": 1.5943060498220643e-05, "loss": 0.6403, "step": 672 }, { "epoch": 0.023977056130537792, "grad_norm": 2.5758423123573406, "learning_rate": 1.5966785290628707e-05, "loss": 0.6875, "step": 673 }, { "epoch": 0.024012683257031903, "grad_norm": 2.3639260643730395, "learning_rate": 1.5990510083036774e-05, "loss": 0.5443, "step": 674 }, { "epoch": 0.024048310383526016, "grad_norm": 1.733189012936582, "learning_rate": 1.601423487544484e-05, "loss": 0.5703, "step": 675 }, { "epoch": 0.02408393751002013, "grad_norm": 2.5326792044287996, "learning_rate": 1.6037959667852908e-05, "loss": 0.7628, "step": 676 }, { "epoch": 0.024119564636514244, "grad_norm": 2.5994027466494782, "learning_rate": 1.6061684460260975e-05, "loss": 0.5717, "step": 677 }, { "epoch": 0.024155191763008354, "grad_norm": 1.5744492360335844, "learning_rate": 1.608540925266904e-05, "loss": 0.5839, "step": 678 }, { "epoch": 0.024190818889502468, "grad_norm": 1.8570418708284382, "learning_rate": 1.6109134045077106e-05, "loss": 0.5974, "step": 679 }, { "epoch": 0.02422644601599658, "grad_norm": 1.8973856277693204, "learning_rate": 1.6132858837485173e-05, "loss": 0.5604, "step": 680 }, { "epoch": 0.02426207314249069, "grad_norm": 3.1488555510980727, "learning_rate": 1.615658362989324e-05, "loss": 0.6704, "step": 681 }, { "epoch": 0.024297700268984805, "grad_norm": 1.6999810720776154, "learning_rate": 1.6180308422301307e-05, "loss": 0.5538, "step": 682 }, { "epoch": 0.02433332739547892, "grad_norm": 4.646754898234997, "learning_rate": 1.620403321470937e-05, "loss": 0.7352, "step": 683 }, { "epoch": 0.02436895452197303, "grad_norm": 2.0442971073446814, "learning_rate": 1.622775800711744e-05, "loss": 0.5705, "step": 684 }, { "epoch": 0.024404581648467143, "grad_norm": 3.0324723285841673, "learning_rate": 1.6251482799525508e-05, "loss": 0.604, "step": 685 }, { "epoch": 0.024440208774961256, "grad_norm": 2.4798415320075193, "learning_rate": 1.627520759193357e-05, "loss": 0.6223, "step": 686 }, { "epoch": 0.024475835901455367, "grad_norm": 1.90492685836387, "learning_rate": 1.629893238434164e-05, "loss": 0.5395, "step": 687 }, { "epoch": 0.02451146302794948, "grad_norm": 2.1340761140294187, "learning_rate": 1.6322657176749706e-05, "loss": 0.5619, "step": 688 }, { "epoch": 0.024547090154443594, "grad_norm": 1.7107676734099126, "learning_rate": 1.6346381969157773e-05, "loss": 0.5593, "step": 689 }, { "epoch": 0.024582717280937708, "grad_norm": 1.9009777897062843, "learning_rate": 1.637010676156584e-05, "loss": 0.5967, "step": 690 }, { "epoch": 0.024618344407431818, "grad_norm": 1.7791598934954707, "learning_rate": 1.6393831553973903e-05, "loss": 0.5969, "step": 691 }, { "epoch": 0.02465397153392593, "grad_norm": 1.853504729482397, "learning_rate": 1.641755634638197e-05, "loss": 0.588, "step": 692 }, { "epoch": 0.024689598660420045, "grad_norm": 1.9969786857254581, "learning_rate": 1.6441281138790037e-05, "loss": 0.5148, "step": 693 }, { "epoch": 0.024725225786914155, "grad_norm": 2.7445477944839904, "learning_rate": 1.6465005931198104e-05, "loss": 0.7112, "step": 694 }, { "epoch": 0.02476085291340827, "grad_norm": 2.8349761761011414, "learning_rate": 1.648873072360617e-05, "loss": 0.6775, "step": 695 }, { "epoch": 0.024796480039902383, "grad_norm": 2.68249886967038, "learning_rate": 1.6512455516014235e-05, "loss": 0.7411, "step": 696 }, { "epoch": 0.024832107166396493, "grad_norm": 2.255295793010976, "learning_rate": 1.6536180308422302e-05, "loss": 0.5589, "step": 697 }, { "epoch": 0.024867734292890607, "grad_norm": 2.0778607161656764, "learning_rate": 1.655990510083037e-05, "loss": 0.6022, "step": 698 }, { "epoch": 0.02490336141938472, "grad_norm": 3.5036554752522164, "learning_rate": 1.6583629893238436e-05, "loss": 0.7409, "step": 699 }, { "epoch": 0.02493898854587883, "grad_norm": 1.6915235865376872, "learning_rate": 1.6607354685646503e-05, "loss": 0.5458, "step": 700 }, { "epoch": 0.024974615672372944, "grad_norm": 1.6662775407315638, "learning_rate": 1.6631079478054567e-05, "loss": 0.5513, "step": 701 }, { "epoch": 0.025010242798867058, "grad_norm": 2.54510992961774, "learning_rate": 1.6654804270462634e-05, "loss": 0.7165, "step": 702 }, { "epoch": 0.02504586992536117, "grad_norm": 1.586622058906353, "learning_rate": 1.66785290628707e-05, "loss": 0.5399, "step": 703 }, { "epoch": 0.02508149705185528, "grad_norm": 2.286595955878133, "learning_rate": 1.6702253855278768e-05, "loss": 0.7199, "step": 704 }, { "epoch": 0.025117124178349395, "grad_norm": 1.7050676754409555, "learning_rate": 1.6725978647686835e-05, "loss": 0.5204, "step": 705 }, { "epoch": 0.02515275130484351, "grad_norm": 1.7912733517702017, "learning_rate": 1.67497034400949e-05, "loss": 0.5729, "step": 706 }, { "epoch": 0.02518837843133762, "grad_norm": 2.760741635720398, "learning_rate": 1.6773428232502966e-05, "loss": 0.6647, "step": 707 }, { "epoch": 0.025224005557831733, "grad_norm": 2.243256170463119, "learning_rate": 1.6797153024911033e-05, "loss": 0.7007, "step": 708 }, { "epoch": 0.025259632684325847, "grad_norm": 3.832732231819308, "learning_rate": 1.68208778173191e-05, "loss": 0.6517, "step": 709 }, { "epoch": 0.025295259810819957, "grad_norm": 2.7162352205459688, "learning_rate": 1.6844602609727167e-05, "loss": 0.6511, "step": 710 }, { "epoch": 0.02533088693731407, "grad_norm": 2.0129063836388696, "learning_rate": 1.686832740213523e-05, "loss": 0.5674, "step": 711 }, { "epoch": 0.025366514063808184, "grad_norm": 1.8660416760289253, "learning_rate": 1.6892052194543298e-05, "loss": 0.5857, "step": 712 }, { "epoch": 0.025402141190302298, "grad_norm": 2.4549511529527077, "learning_rate": 1.6915776986951368e-05, "loss": 0.6053, "step": 713 }, { "epoch": 0.025437768316796408, "grad_norm": 1.7322243241639128, "learning_rate": 1.693950177935943e-05, "loss": 0.5619, "step": 714 }, { "epoch": 0.02547339544329052, "grad_norm": 2.2192898531531373, "learning_rate": 1.69632265717675e-05, "loss": 0.6631, "step": 715 }, { "epoch": 0.025509022569784635, "grad_norm": 2.2134122841180814, "learning_rate": 1.6986951364175566e-05, "loss": 0.6124, "step": 716 }, { "epoch": 0.025544649696278746, "grad_norm": 1.7088448175503839, "learning_rate": 1.7010676156583633e-05, "loss": 0.6275, "step": 717 }, { "epoch": 0.02558027682277286, "grad_norm": 3.0817650723554038, "learning_rate": 1.70344009489917e-05, "loss": 0.6947, "step": 718 }, { "epoch": 0.025615903949266973, "grad_norm": 2.116405223469558, "learning_rate": 1.7058125741399763e-05, "loss": 0.6352, "step": 719 }, { "epoch": 0.025651531075761083, "grad_norm": 3.3847583968988775, "learning_rate": 1.708185053380783e-05, "loss": 0.6296, "step": 720 }, { "epoch": 0.025687158202255197, "grad_norm": 2.442925376802581, "learning_rate": 1.7105575326215898e-05, "loss": 0.7036, "step": 721 }, { "epoch": 0.02572278532874931, "grad_norm": 2.639267934216283, "learning_rate": 1.7129300118623965e-05, "loss": 0.6573, "step": 722 }, { "epoch": 0.02575841245524342, "grad_norm": 1.9763466337951776, "learning_rate": 1.715302491103203e-05, "loss": 0.6034, "step": 723 }, { "epoch": 0.025794039581737534, "grad_norm": 1.7261705788714792, "learning_rate": 1.7176749703440095e-05, "loss": 0.5721, "step": 724 }, { "epoch": 0.025829666708231648, "grad_norm": 1.5666747869253013, "learning_rate": 1.7200474495848162e-05, "loss": 0.514, "step": 725 }, { "epoch": 0.025865293834725762, "grad_norm": 1.7686912258090912, "learning_rate": 1.722419928825623e-05, "loss": 0.6021, "step": 726 }, { "epoch": 0.025900920961219872, "grad_norm": 1.8122382607473686, "learning_rate": 1.7247924080664296e-05, "loss": 0.5438, "step": 727 }, { "epoch": 0.025936548087713986, "grad_norm": 6.246640664084044, "learning_rate": 1.7271648873072363e-05, "loss": 0.6417, "step": 728 }, { "epoch": 0.0259721752142081, "grad_norm": 4.324310726008264, "learning_rate": 1.7295373665480427e-05, "loss": 0.7699, "step": 729 }, { "epoch": 0.02600780234070221, "grad_norm": 2.0107424524404824, "learning_rate": 1.7319098457888494e-05, "loss": 0.5728, "step": 730 }, { "epoch": 0.026043429467196323, "grad_norm": 1.7565948558557012, "learning_rate": 1.734282325029656e-05, "loss": 0.5899, "step": 731 }, { "epoch": 0.026079056593690437, "grad_norm": 1.695264196323913, "learning_rate": 1.7366548042704628e-05, "loss": 0.5669, "step": 732 }, { "epoch": 0.026114683720184547, "grad_norm": 1.8537158372342477, "learning_rate": 1.7390272835112695e-05, "loss": 0.6184, "step": 733 }, { "epoch": 0.02615031084667866, "grad_norm": 2.560522762060512, "learning_rate": 1.741399762752076e-05, "loss": 0.6993, "step": 734 }, { "epoch": 0.026185937973172774, "grad_norm": 2.333442388320042, "learning_rate": 1.7437722419928826e-05, "loss": 0.6655, "step": 735 }, { "epoch": 0.026221565099666888, "grad_norm": 2.7447905519522364, "learning_rate": 1.7461447212336893e-05, "loss": 0.6932, "step": 736 }, { "epoch": 0.026257192226161, "grad_norm": 2.018244306918077, "learning_rate": 1.748517200474496e-05, "loss": 0.5803, "step": 737 }, { "epoch": 0.026292819352655112, "grad_norm": 1.9890134259267045, "learning_rate": 1.7508896797153027e-05, "loss": 0.5164, "step": 738 }, { "epoch": 0.026328446479149226, "grad_norm": 3.2485325150449498, "learning_rate": 1.753262158956109e-05, "loss": 0.6971, "step": 739 }, { "epoch": 0.026364073605643336, "grad_norm": 1.7777037216361231, "learning_rate": 1.7556346381969158e-05, "loss": 0.5431, "step": 740 }, { "epoch": 0.02639970073213745, "grad_norm": 1.695734583838836, "learning_rate": 1.7580071174377225e-05, "loss": 0.5395, "step": 741 }, { "epoch": 0.026435327858631563, "grad_norm": 2.1598461864339518, "learning_rate": 1.7603795966785292e-05, "loss": 0.5539, "step": 742 }, { "epoch": 0.026470954985125673, "grad_norm": 1.6412826936418554, "learning_rate": 1.762752075919336e-05, "loss": 0.5418, "step": 743 }, { "epoch": 0.026506582111619787, "grad_norm": 3.4972438227356837, "learning_rate": 1.7651245551601426e-05, "loss": 0.642, "step": 744 }, { "epoch": 0.0265422092381139, "grad_norm": 2.7240895851423144, "learning_rate": 1.7674970344009493e-05, "loss": 0.6514, "step": 745 }, { "epoch": 0.02657783636460801, "grad_norm": 1.8233516479173422, "learning_rate": 1.769869513641756e-05, "loss": 0.6032, "step": 746 }, { "epoch": 0.026613463491102125, "grad_norm": 2.279517259563635, "learning_rate": 1.7722419928825624e-05, "loss": 0.6951, "step": 747 }, { "epoch": 0.02664909061759624, "grad_norm": 1.7395255487178742, "learning_rate": 1.774614472123369e-05, "loss": 0.5202, "step": 748 }, { "epoch": 0.026684717744090352, "grad_norm": 2.000035125387126, "learning_rate": 1.7769869513641758e-05, "loss": 0.6663, "step": 749 }, { "epoch": 0.026720344870584462, "grad_norm": 1.7389222377256417, "learning_rate": 1.7793594306049825e-05, "loss": 0.5679, "step": 750 }, { "epoch": 0.026755971997078576, "grad_norm": 2.5968340161105745, "learning_rate": 1.7817319098457892e-05, "loss": 0.7573, "step": 751 }, { "epoch": 0.02679159912357269, "grad_norm": 1.9167188112970102, "learning_rate": 1.7841043890865955e-05, "loss": 0.5496, "step": 752 }, { "epoch": 0.0268272262500668, "grad_norm": 3.001594687039781, "learning_rate": 1.7864768683274022e-05, "loss": 0.6437, "step": 753 }, { "epoch": 0.026862853376560913, "grad_norm": 1.9506415973798537, "learning_rate": 1.788849347568209e-05, "loss": 0.6154, "step": 754 }, { "epoch": 0.026898480503055027, "grad_norm": 4.719467759600568, "learning_rate": 1.7912218268090156e-05, "loss": 0.7287, "step": 755 }, { "epoch": 0.026934107629549137, "grad_norm": 1.7503172212474913, "learning_rate": 1.7935943060498224e-05, "loss": 0.616, "step": 756 }, { "epoch": 0.02696973475604325, "grad_norm": 2.31469979056566, "learning_rate": 1.7959667852906287e-05, "loss": 0.6491, "step": 757 }, { "epoch": 0.027005361882537365, "grad_norm": 2.66922770703968, "learning_rate": 1.7983392645314354e-05, "loss": 0.6635, "step": 758 }, { "epoch": 0.027040989009031475, "grad_norm": 1.749914963294943, "learning_rate": 1.800711743772242e-05, "loss": 0.5869, "step": 759 }, { "epoch": 0.02707661613552559, "grad_norm": 4.305114831736421, "learning_rate": 1.8030842230130488e-05, "loss": 0.6021, "step": 760 }, { "epoch": 0.027112243262019702, "grad_norm": 2.246091861896535, "learning_rate": 1.8054567022538555e-05, "loss": 0.6166, "step": 761 }, { "epoch": 0.027147870388513816, "grad_norm": 2.4856410162910287, "learning_rate": 1.807829181494662e-05, "loss": 0.6838, "step": 762 }, { "epoch": 0.027183497515007926, "grad_norm": 1.9264944798068888, "learning_rate": 1.8102016607354686e-05, "loss": 0.602, "step": 763 }, { "epoch": 0.02721912464150204, "grad_norm": 3.283227960873826, "learning_rate": 1.8125741399762753e-05, "loss": 0.7168, "step": 764 }, { "epoch": 0.027254751767996153, "grad_norm": 1.6023320595183153, "learning_rate": 1.814946619217082e-05, "loss": 0.5056, "step": 765 }, { "epoch": 0.027290378894490264, "grad_norm": 1.8423944108746404, "learning_rate": 1.8173190984578887e-05, "loss": 0.5211, "step": 766 }, { "epoch": 0.027326006020984377, "grad_norm": 1.6311509125242927, "learning_rate": 1.819691577698695e-05, "loss": 0.5898, "step": 767 }, { "epoch": 0.02736163314747849, "grad_norm": 2.0015071438904544, "learning_rate": 1.8220640569395018e-05, "loss": 0.5669, "step": 768 }, { "epoch": 0.0273972602739726, "grad_norm": 1.8424332702452721, "learning_rate": 1.8244365361803085e-05, "loss": 0.5398, "step": 769 }, { "epoch": 0.027432887400466715, "grad_norm": 1.701357262357989, "learning_rate": 1.8268090154211152e-05, "loss": 0.5266, "step": 770 }, { "epoch": 0.02746851452696083, "grad_norm": 3.1154139215649646, "learning_rate": 1.829181494661922e-05, "loss": 0.7096, "step": 771 }, { "epoch": 0.027504141653454942, "grad_norm": 1.9116844561106514, "learning_rate": 1.8315539739027286e-05, "loss": 0.5648, "step": 772 }, { "epoch": 0.027539768779949052, "grad_norm": 1.5701716322111394, "learning_rate": 1.8339264531435353e-05, "loss": 0.5406, "step": 773 }, { "epoch": 0.027575395906443166, "grad_norm": 2.4399317257141093, "learning_rate": 1.836298932384342e-05, "loss": 0.7441, "step": 774 }, { "epoch": 0.02761102303293728, "grad_norm": 1.869854975435682, "learning_rate": 1.8386714116251484e-05, "loss": 0.5504, "step": 775 }, { "epoch": 0.02764665015943139, "grad_norm": 3.315896522735423, "learning_rate": 1.841043890865955e-05, "loss": 0.6847, "step": 776 }, { "epoch": 0.027682277285925504, "grad_norm": 1.8447479892366334, "learning_rate": 1.8434163701067618e-05, "loss": 0.488, "step": 777 }, { "epoch": 0.027717904412419617, "grad_norm": 3.0968744935457115, "learning_rate": 1.8457888493475685e-05, "loss": 0.6846, "step": 778 }, { "epoch": 0.027753531538913728, "grad_norm": 2.5200490659282475, "learning_rate": 1.8481613285883752e-05, "loss": 0.6294, "step": 779 }, { "epoch": 0.02778915866540784, "grad_norm": 3.3929827986698293, "learning_rate": 1.8505338078291815e-05, "loss": 0.7172, "step": 780 }, { "epoch": 0.027824785791901955, "grad_norm": 2.0404928817372796, "learning_rate": 1.8529062870699883e-05, "loss": 0.588, "step": 781 }, { "epoch": 0.027860412918396065, "grad_norm": 2.8560875649119932, "learning_rate": 1.855278766310795e-05, "loss": 0.5912, "step": 782 }, { "epoch": 0.02789604004489018, "grad_norm": 2.8316099682318985, "learning_rate": 1.8576512455516017e-05, "loss": 0.6851, "step": 783 }, { "epoch": 0.027931667171384292, "grad_norm": 2.2666696634343895, "learning_rate": 1.8600237247924084e-05, "loss": 0.5693, "step": 784 }, { "epoch": 0.027967294297878406, "grad_norm": 2.311098844044447, "learning_rate": 1.8623962040332147e-05, "loss": 0.6745, "step": 785 }, { "epoch": 0.028002921424372516, "grad_norm": 2.507099431785896, "learning_rate": 1.8647686832740214e-05, "loss": 0.6685, "step": 786 }, { "epoch": 0.02803854855086663, "grad_norm": 1.7489250801130132, "learning_rate": 1.867141162514828e-05, "loss": 0.5663, "step": 787 }, { "epoch": 0.028074175677360744, "grad_norm": 1.8170466564593741, "learning_rate": 1.869513641755635e-05, "loss": 0.5544, "step": 788 }, { "epoch": 0.028109802803854854, "grad_norm": 2.334603407396125, "learning_rate": 1.8718861209964415e-05, "loss": 0.6664, "step": 789 }, { "epoch": 0.028145429930348968, "grad_norm": 2.227259753992637, "learning_rate": 1.874258600237248e-05, "loss": 0.6305, "step": 790 }, { "epoch": 0.02818105705684308, "grad_norm": 1.9116877657880573, "learning_rate": 1.8766310794780546e-05, "loss": 0.5274, "step": 791 }, { "epoch": 0.02821668418333719, "grad_norm": 2.793985621485211, "learning_rate": 1.8790035587188613e-05, "loss": 0.6494, "step": 792 }, { "epoch": 0.028252311309831305, "grad_norm": 2.954145993116308, "learning_rate": 1.881376037959668e-05, "loss": 0.6156, "step": 793 }, { "epoch": 0.02828793843632542, "grad_norm": 1.8105365506552489, "learning_rate": 1.8837485172004747e-05, "loss": 0.6079, "step": 794 }, { "epoch": 0.028323565562819532, "grad_norm": 1.8259247633494076, "learning_rate": 1.886120996441281e-05, "loss": 0.5884, "step": 795 }, { "epoch": 0.028359192689313643, "grad_norm": 1.6717665596088767, "learning_rate": 1.8884934756820878e-05, "loss": 0.5418, "step": 796 }, { "epoch": 0.028394819815807756, "grad_norm": 3.6620772630569554, "learning_rate": 1.8908659549228945e-05, "loss": 0.6614, "step": 797 }, { "epoch": 0.02843044694230187, "grad_norm": 2.4485895212986035, "learning_rate": 1.8932384341637012e-05, "loss": 0.5488, "step": 798 }, { "epoch": 0.02846607406879598, "grad_norm": 1.8034566798991962, "learning_rate": 1.895610913404508e-05, "loss": 0.6082, "step": 799 }, { "epoch": 0.028501701195290094, "grad_norm": 1.641735210013732, "learning_rate": 1.8979833926453146e-05, "loss": 0.5965, "step": 800 }, { "epoch": 0.028537328321784208, "grad_norm": 1.9134426853151116, "learning_rate": 1.9003558718861213e-05, "loss": 0.5808, "step": 801 }, { "epoch": 0.028572955448278318, "grad_norm": 3.0575950961714953, "learning_rate": 1.902728351126928e-05, "loss": 0.6417, "step": 802 }, { "epoch": 0.02860858257477243, "grad_norm": 1.9845122746093358, "learning_rate": 1.9051008303677344e-05, "loss": 0.5709, "step": 803 }, { "epoch": 0.028644209701266545, "grad_norm": 1.7924949888065336, "learning_rate": 1.907473309608541e-05, "loss": 0.5681, "step": 804 }, { "epoch": 0.028679836827760655, "grad_norm": 2.6270638035261658, "learning_rate": 1.9098457888493478e-05, "loss": 0.6435, "step": 805 }, { "epoch": 0.02871546395425477, "grad_norm": 1.9338959538958662, "learning_rate": 1.9122182680901545e-05, "loss": 0.5995, "step": 806 }, { "epoch": 0.028751091080748883, "grad_norm": 1.717831279790205, "learning_rate": 1.9145907473309612e-05, "loss": 0.5664, "step": 807 }, { "epoch": 0.028786718207242996, "grad_norm": 2.0080335413731456, "learning_rate": 1.9169632265717676e-05, "loss": 0.616, "step": 808 }, { "epoch": 0.028822345333737107, "grad_norm": 2.9861478870275033, "learning_rate": 1.9193357058125743e-05, "loss": 0.7558, "step": 809 }, { "epoch": 0.02885797246023122, "grad_norm": 1.862868081731416, "learning_rate": 1.921708185053381e-05, "loss": 0.5532, "step": 810 }, { "epoch": 0.028893599586725334, "grad_norm": 3.17377272403992, "learning_rate": 1.9240806642941877e-05, "loss": 0.7347, "step": 811 }, { "epoch": 0.028929226713219444, "grad_norm": 2.2947566006729003, "learning_rate": 1.9264531435349944e-05, "loss": 0.678, "step": 812 }, { "epoch": 0.028964853839713558, "grad_norm": 2.9471513864885295, "learning_rate": 1.9288256227758007e-05, "loss": 0.6651, "step": 813 }, { "epoch": 0.02900048096620767, "grad_norm": 3.1372009072867795, "learning_rate": 1.9311981020166074e-05, "loss": 0.6386, "step": 814 }, { "epoch": 0.02903610809270178, "grad_norm": 3.0971972365545457, "learning_rate": 1.933570581257414e-05, "loss": 0.6525, "step": 815 }, { "epoch": 0.029071735219195895, "grad_norm": 1.794480979929958, "learning_rate": 1.935943060498221e-05, "loss": 0.5953, "step": 816 }, { "epoch": 0.02910736234569001, "grad_norm": 1.683407210718618, "learning_rate": 1.9383155397390276e-05, "loss": 0.541, "step": 817 }, { "epoch": 0.02914298947218412, "grad_norm": 2.9519323122495496, "learning_rate": 1.940688018979834e-05, "loss": 0.6814, "step": 818 }, { "epoch": 0.029178616598678233, "grad_norm": 2.3752045363830905, "learning_rate": 1.9430604982206406e-05, "loss": 0.6885, "step": 819 }, { "epoch": 0.029214243725172347, "grad_norm": 1.6378636609628885, "learning_rate": 1.9454329774614473e-05, "loss": 0.594, "step": 820 }, { "epoch": 0.02924987085166646, "grad_norm": 1.7924449353283691, "learning_rate": 1.947805456702254e-05, "loss": 0.5648, "step": 821 }, { "epoch": 0.02928549797816057, "grad_norm": 2.6017081998638036, "learning_rate": 1.9501779359430607e-05, "loss": 0.6475, "step": 822 }, { "epoch": 0.029321125104654684, "grad_norm": 1.5608702528318776, "learning_rate": 1.952550415183867e-05, "loss": 0.5399, "step": 823 }, { "epoch": 0.029356752231148798, "grad_norm": 1.8496105919437595, "learning_rate": 1.9549228944246738e-05, "loss": 0.6092, "step": 824 }, { "epoch": 0.029392379357642908, "grad_norm": 1.7420304600405503, "learning_rate": 1.9572953736654805e-05, "loss": 0.5629, "step": 825 }, { "epoch": 0.029428006484137022, "grad_norm": 1.6158469247994873, "learning_rate": 1.9596678529062872e-05, "loss": 0.5785, "step": 826 }, { "epoch": 0.029463633610631135, "grad_norm": 1.6895107229904367, "learning_rate": 1.962040332147094e-05, "loss": 0.5881, "step": 827 }, { "epoch": 0.029499260737125246, "grad_norm": 1.6062449720773944, "learning_rate": 1.9644128113879003e-05, "loss": 0.5385, "step": 828 }, { "epoch": 0.02953488786361936, "grad_norm": 2.339611308819745, "learning_rate": 1.9667852906287073e-05, "loss": 0.6966, "step": 829 }, { "epoch": 0.029570514990113473, "grad_norm": 2.2398062036829813, "learning_rate": 1.969157769869514e-05, "loss": 0.6147, "step": 830 }, { "epoch": 0.029606142116607587, "grad_norm": 2.750897960856025, "learning_rate": 1.9715302491103204e-05, "loss": 0.6312, "step": 831 }, { "epoch": 0.029641769243101697, "grad_norm": 1.74706038547466, "learning_rate": 1.973902728351127e-05, "loss": 0.6111, "step": 832 }, { "epoch": 0.02967739636959581, "grad_norm": 1.8175166516752284, "learning_rate": 1.9762752075919338e-05, "loss": 0.5495, "step": 833 }, { "epoch": 0.029713023496089924, "grad_norm": 2.4047633537865467, "learning_rate": 1.9786476868327405e-05, "loss": 0.6182, "step": 834 }, { "epoch": 0.029748650622584034, "grad_norm": 1.6069119816612984, "learning_rate": 1.9810201660735472e-05, "loss": 0.5293, "step": 835 }, { "epoch": 0.029784277749078148, "grad_norm": 1.7081872046416544, "learning_rate": 1.9833926453143536e-05, "loss": 0.615, "step": 836 }, { "epoch": 0.029819904875572262, "grad_norm": 2.317147937656157, "learning_rate": 1.9857651245551603e-05, "loss": 0.6359, "step": 837 }, { "epoch": 0.029855532002066372, "grad_norm": 1.5577587739036025, "learning_rate": 1.988137603795967e-05, "loss": 0.5138, "step": 838 }, { "epoch": 0.029891159128560486, "grad_norm": 1.7394470020357622, "learning_rate": 1.9905100830367737e-05, "loss": 0.5839, "step": 839 }, { "epoch": 0.0299267862550546, "grad_norm": 1.6721190745270555, "learning_rate": 1.9928825622775804e-05, "loss": 0.5145, "step": 840 }, { "epoch": 0.02996241338154871, "grad_norm": 1.706144220499785, "learning_rate": 1.9952550415183868e-05, "loss": 0.5934, "step": 841 }, { "epoch": 0.029998040508042823, "grad_norm": 2.6709819563625787, "learning_rate": 1.9976275207591935e-05, "loss": 0.6765, "step": 842 }, { "epoch": 0.030033667634536937, "grad_norm": 2.697152142406989, "learning_rate": 2e-05, "loss": 0.6245, "step": 843 }, { "epoch": 0.03006929476103105, "grad_norm": 1.7481491175339778, "learning_rate": 1.9999999933421508e-05, "loss": 0.5439, "step": 844 }, { "epoch": 0.03010492188752516, "grad_norm": 1.7624338098891652, "learning_rate": 1.9999999733686023e-05, "loss": 0.5944, "step": 845 }, { "epoch": 0.030140549014019274, "grad_norm": 1.4293576084832218, "learning_rate": 1.9999999400793553e-05, "loss": 0.5604, "step": 846 }, { "epoch": 0.030176176140513388, "grad_norm": 2.2412880028896263, "learning_rate": 1.9999998934744103e-05, "loss": 0.6631, "step": 847 }, { "epoch": 0.0302118032670075, "grad_norm": 2.4723362634150345, "learning_rate": 1.9999998335537675e-05, "loss": 0.6774, "step": 848 }, { "epoch": 0.030247430393501612, "grad_norm": 2.5652219897202406, "learning_rate": 1.9999997603174283e-05, "loss": 0.7034, "step": 849 }, { "epoch": 0.030283057519995726, "grad_norm": 2.131920259203048, "learning_rate": 1.999999673765393e-05, "loss": 0.694, "step": 850 }, { "epoch": 0.030318684646489836, "grad_norm": 2.0370544882101993, "learning_rate": 1.9999995738976634e-05, "loss": 0.595, "step": 851 }, { "epoch": 0.03035431177298395, "grad_norm": 1.7767671954939268, "learning_rate": 1.99999946071424e-05, "loss": 0.5724, "step": 852 }, { "epoch": 0.030389938899478063, "grad_norm": 1.6012222724108875, "learning_rate": 1.9999993342151254e-05, "loss": 0.5691, "step": 853 }, { "epoch": 0.030425566025972177, "grad_norm": 2.078234577451606, "learning_rate": 1.9999991944003204e-05, "loss": 0.567, "step": 854 }, { "epoch": 0.030461193152466287, "grad_norm": 2.0227199606228643, "learning_rate": 1.9999990412698275e-05, "loss": 0.6144, "step": 855 }, { "epoch": 0.0304968202789604, "grad_norm": 2.177993861994139, "learning_rate": 1.999998874823648e-05, "loss": 0.6093, "step": 856 }, { "epoch": 0.030532447405454514, "grad_norm": 3.36900593368103, "learning_rate": 1.9999986950617843e-05, "loss": 0.6145, "step": 857 }, { "epoch": 0.030568074531948625, "grad_norm": 1.833284194390868, "learning_rate": 1.9999985019842398e-05, "loss": 0.585, "step": 858 }, { "epoch": 0.03060370165844274, "grad_norm": 1.5993839096056033, "learning_rate": 1.9999982955910158e-05, "loss": 0.5463, "step": 859 }, { "epoch": 0.030639328784936852, "grad_norm": 1.9459910200409756, "learning_rate": 1.9999980758821154e-05, "loss": 0.5588, "step": 860 }, { "epoch": 0.030674955911430962, "grad_norm": 1.891627495790606, "learning_rate": 1.9999978428575423e-05, "loss": 0.5883, "step": 861 }, { "epoch": 0.030710583037925076, "grad_norm": 4.553420611886144, "learning_rate": 1.9999975965172983e-05, "loss": 0.7608, "step": 862 }, { "epoch": 0.03074621016441919, "grad_norm": 2.0748523216584993, "learning_rate": 1.9999973368613876e-05, "loss": 0.5345, "step": 863 }, { "epoch": 0.0307818372909133, "grad_norm": 2.081398690206639, "learning_rate": 1.9999970638898134e-05, "loss": 0.7387, "step": 864 }, { "epoch": 0.030817464417407413, "grad_norm": 1.740649382004126, "learning_rate": 1.9999967776025794e-05, "loss": 0.5765, "step": 865 }, { "epoch": 0.030853091543901527, "grad_norm": 1.8665267644253412, "learning_rate": 1.9999964779996893e-05, "loss": 0.5544, "step": 866 }, { "epoch": 0.03088871867039564, "grad_norm": 2.9097648327720105, "learning_rate": 1.999996165081147e-05, "loss": 0.7124, "step": 867 }, { "epoch": 0.03092434579688975, "grad_norm": 1.971623674665617, "learning_rate": 1.999995838846957e-05, "loss": 0.5995, "step": 868 }, { "epoch": 0.030959972923383865, "grad_norm": 2.1805422739414264, "learning_rate": 1.9999954992971234e-05, "loss": 0.5493, "step": 869 }, { "epoch": 0.03099560004987798, "grad_norm": 1.938232401160258, "learning_rate": 1.9999951464316507e-05, "loss": 0.5746, "step": 870 }, { "epoch": 0.03103122717637209, "grad_norm": 2.277992026746974, "learning_rate": 1.9999947802505438e-05, "loss": 0.6247, "step": 871 }, { "epoch": 0.031066854302866202, "grad_norm": 2.7345585396349343, "learning_rate": 1.9999944007538073e-05, "loss": 0.5777, "step": 872 }, { "epoch": 0.031102481429360316, "grad_norm": 2.7979156417705875, "learning_rate": 1.9999940079414462e-05, "loss": 0.688, "step": 873 }, { "epoch": 0.031138108555854426, "grad_norm": 2.0933990283558663, "learning_rate": 1.9999936018134664e-05, "loss": 0.577, "step": 874 }, { "epoch": 0.03117373568234854, "grad_norm": 1.9327427667935781, "learning_rate": 1.9999931823698727e-05, "loss": 0.6833, "step": 875 }, { "epoch": 0.031209362808842653, "grad_norm": 2.1161213523035833, "learning_rate": 1.9999927496106708e-05, "loss": 0.5181, "step": 876 }, { "epoch": 0.031244989935336764, "grad_norm": 3.3868880225090576, "learning_rate": 1.9999923035358663e-05, "loss": 0.7471, "step": 877 }, { "epoch": 0.03128061706183088, "grad_norm": 1.8786028207918637, "learning_rate": 1.9999918441454657e-05, "loss": 0.5606, "step": 878 }, { "epoch": 0.03131624418832499, "grad_norm": 1.6448430033434767, "learning_rate": 1.9999913714394744e-05, "loss": 0.5557, "step": 879 }, { "epoch": 0.031351871314819105, "grad_norm": 2.4601278473421866, "learning_rate": 1.999990885417899e-05, "loss": 0.6275, "step": 880 }, { "epoch": 0.031387498441313215, "grad_norm": 2.5989780182712674, "learning_rate": 1.999990386080746e-05, "loss": 0.559, "step": 881 }, { "epoch": 0.031423125567807325, "grad_norm": 1.8455863468326297, "learning_rate": 1.9999898734280227e-05, "loss": 0.6023, "step": 882 }, { "epoch": 0.03145875269430144, "grad_norm": 2.226723736296335, "learning_rate": 1.9999893474597347e-05, "loss": 0.6217, "step": 883 }, { "epoch": 0.03149437982079555, "grad_norm": 3.6628820414513474, "learning_rate": 1.9999888081758895e-05, "loss": 0.6481, "step": 884 }, { "epoch": 0.03153000694728967, "grad_norm": 1.706525054057737, "learning_rate": 1.9999882555764946e-05, "loss": 0.5393, "step": 885 }, { "epoch": 0.03156563407378378, "grad_norm": 2.511939951277383, "learning_rate": 1.999987689661557e-05, "loss": 0.6646, "step": 886 }, { "epoch": 0.03160126120027789, "grad_norm": 1.7000505287958059, "learning_rate": 1.9999871104310846e-05, "loss": 0.5339, "step": 887 }, { "epoch": 0.03163688832677201, "grad_norm": 1.6620499519957157, "learning_rate": 1.9999865178850847e-05, "loss": 0.5685, "step": 888 }, { "epoch": 0.03167251545326612, "grad_norm": 2.5322664335524103, "learning_rate": 1.9999859120235655e-05, "loss": 0.6457, "step": 889 }, { "epoch": 0.03170814257976023, "grad_norm": 2.592864429543557, "learning_rate": 1.999985292846535e-05, "loss": 0.6449, "step": 890 }, { "epoch": 0.031743769706254345, "grad_norm": 1.9044359398533044, "learning_rate": 1.9999846603540008e-05, "loss": 0.5907, "step": 891 }, { "epoch": 0.031779396832748455, "grad_norm": 1.8338482511092136, "learning_rate": 1.9999840145459725e-05, "loss": 0.5975, "step": 892 }, { "epoch": 0.031815023959242565, "grad_norm": 1.754746976676261, "learning_rate": 1.9999833554224575e-05, "loss": 0.5458, "step": 893 }, { "epoch": 0.03185065108573668, "grad_norm": 2.6812575431627965, "learning_rate": 1.9999826829834657e-05, "loss": 0.685, "step": 894 }, { "epoch": 0.03188627821223079, "grad_norm": 2.5791095737175405, "learning_rate": 1.9999819972290055e-05, "loss": 0.5847, "step": 895 }, { "epoch": 0.0319219053387249, "grad_norm": 3.503399033615648, "learning_rate": 1.999981298159086e-05, "loss": 0.7229, "step": 896 }, { "epoch": 0.03195753246521902, "grad_norm": 2.026223714431898, "learning_rate": 1.999980585773717e-05, "loss": 0.6507, "step": 897 }, { "epoch": 0.03199315959171313, "grad_norm": 2.700508980855242, "learning_rate": 1.9999798600729067e-05, "loss": 0.667, "step": 898 }, { "epoch": 0.03202878671820724, "grad_norm": 1.7458459713515624, "learning_rate": 1.9999791210566658e-05, "loss": 0.5911, "step": 899 }, { "epoch": 0.03206441384470136, "grad_norm": 1.557102499104207, "learning_rate": 1.9999783687250046e-05, "loss": 0.5906, "step": 900 }, { "epoch": 0.03210004097119547, "grad_norm": 2.9214484161363723, "learning_rate": 1.9999776030779316e-05, "loss": 0.6179, "step": 901 }, { "epoch": 0.03213566809768958, "grad_norm": 1.4648215605332324, "learning_rate": 1.9999768241154583e-05, "loss": 0.5668, "step": 902 }, { "epoch": 0.032171295224183695, "grad_norm": 2.718872658035536, "learning_rate": 1.999976031837595e-05, "loss": 0.5685, "step": 903 }, { "epoch": 0.032206922350677805, "grad_norm": 1.707421456872485, "learning_rate": 1.9999752262443516e-05, "loss": 0.5831, "step": 904 }, { "epoch": 0.032242549477171915, "grad_norm": 2.14736063192114, "learning_rate": 1.999974407335739e-05, "loss": 0.7497, "step": 905 }, { "epoch": 0.03227817660366603, "grad_norm": 1.680375085866841, "learning_rate": 1.9999735751117684e-05, "loss": 0.5675, "step": 906 }, { "epoch": 0.03231380373016014, "grad_norm": 1.4697439838852338, "learning_rate": 1.9999727295724506e-05, "loss": 0.5136, "step": 907 }, { "epoch": 0.03234943085665426, "grad_norm": 2.158651036268924, "learning_rate": 1.999971870717797e-05, "loss": 0.5641, "step": 908 }, { "epoch": 0.03238505798314837, "grad_norm": 2.4504069365042946, "learning_rate": 1.9999709985478188e-05, "loss": 0.7173, "step": 909 }, { "epoch": 0.03242068510964248, "grad_norm": 2.4371662320052865, "learning_rate": 1.999970113062528e-05, "loss": 0.6899, "step": 910 }, { "epoch": 0.0324563122361366, "grad_norm": 1.9529629439239213, "learning_rate": 1.999969214261936e-05, "loss": 0.5886, "step": 911 }, { "epoch": 0.03249193936263071, "grad_norm": 1.8334202974661726, "learning_rate": 1.9999683021460553e-05, "loss": 0.5811, "step": 912 }, { "epoch": 0.03252756648912482, "grad_norm": 1.7487122872825072, "learning_rate": 1.9999673767148976e-05, "loss": 0.619, "step": 913 }, { "epoch": 0.032563193615618935, "grad_norm": 1.758039832247882, "learning_rate": 1.9999664379684754e-05, "loss": 0.5792, "step": 914 }, { "epoch": 0.032598820742113045, "grad_norm": 2.5606040980720257, "learning_rate": 1.9999654859068013e-05, "loss": 0.6893, "step": 915 }, { "epoch": 0.032634447868607155, "grad_norm": 4.349829764111781, "learning_rate": 1.9999645205298876e-05, "loss": 0.7043, "step": 916 }, { "epoch": 0.03267007499510127, "grad_norm": 1.6811348946272877, "learning_rate": 1.9999635418377474e-05, "loss": 0.5571, "step": 917 }, { "epoch": 0.03270570212159538, "grad_norm": 1.8515831165123444, "learning_rate": 1.9999625498303936e-05, "loss": 0.5784, "step": 918 }, { "epoch": 0.03274132924808949, "grad_norm": 1.8797481086610457, "learning_rate": 1.9999615445078394e-05, "loss": 0.5967, "step": 919 }, { "epoch": 0.03277695637458361, "grad_norm": 1.7330608560316232, "learning_rate": 1.9999605258700983e-05, "loss": 0.5431, "step": 920 }, { "epoch": 0.03281258350107772, "grad_norm": 2.9445254984480567, "learning_rate": 1.9999594939171844e-05, "loss": 0.6912, "step": 921 }, { "epoch": 0.03284821062757183, "grad_norm": 1.5124027912245757, "learning_rate": 1.9999584486491107e-05, "loss": 0.523, "step": 922 }, { "epoch": 0.03288383775406595, "grad_norm": 1.5543554057014446, "learning_rate": 1.999957390065891e-05, "loss": 0.6075, "step": 923 }, { "epoch": 0.03291946488056006, "grad_norm": 1.5630629181517635, "learning_rate": 1.99995631816754e-05, "loss": 0.5339, "step": 924 }, { "epoch": 0.03295509200705417, "grad_norm": 1.6390602277142663, "learning_rate": 1.9999552329540715e-05, "loss": 0.5884, "step": 925 }, { "epoch": 0.032990719133548285, "grad_norm": 3.0711943349417052, "learning_rate": 1.9999541344255006e-05, "loss": 0.636, "step": 926 }, { "epoch": 0.033026346260042395, "grad_norm": 2.1966113715571742, "learning_rate": 1.999953022581841e-05, "loss": 0.6393, "step": 927 }, { "epoch": 0.033061973386536506, "grad_norm": 1.592512186151803, "learning_rate": 1.9999518974231082e-05, "loss": 0.5695, "step": 928 }, { "epoch": 0.03309760051303062, "grad_norm": 1.7596329770511343, "learning_rate": 1.999950758949317e-05, "loss": 0.5844, "step": 929 }, { "epoch": 0.03313322763952473, "grad_norm": 3.4568186586863128, "learning_rate": 1.999949607160482e-05, "loss": 0.6883, "step": 930 }, { "epoch": 0.03316885476601885, "grad_norm": 2.229515675996706, "learning_rate": 1.9999484420566197e-05, "loss": 0.7294, "step": 931 }, { "epoch": 0.03320448189251296, "grad_norm": 2.179637336603746, "learning_rate": 1.999947263637745e-05, "loss": 0.635, "step": 932 }, { "epoch": 0.03324010901900707, "grad_norm": 1.5317658242113705, "learning_rate": 1.999946071903873e-05, "loss": 0.5927, "step": 933 }, { "epoch": 0.03327573614550119, "grad_norm": 1.6265152261604694, "learning_rate": 1.9999448668550205e-05, "loss": 0.6034, "step": 934 }, { "epoch": 0.0333113632719953, "grad_norm": 1.5232446954406689, "learning_rate": 1.9999436484912033e-05, "loss": 0.5186, "step": 935 }, { "epoch": 0.03334699039848941, "grad_norm": 1.5924688569464462, "learning_rate": 1.999942416812437e-05, "loss": 0.5405, "step": 936 }, { "epoch": 0.033382617524983525, "grad_norm": 1.4408449469392195, "learning_rate": 1.999941171818739e-05, "loss": 0.5146, "step": 937 }, { "epoch": 0.033418244651477635, "grad_norm": 1.4485829552789433, "learning_rate": 1.9999399135101255e-05, "loss": 0.4912, "step": 938 }, { "epoch": 0.033453871777971746, "grad_norm": 1.3369744962734869, "learning_rate": 1.9999386418866126e-05, "loss": 0.5499, "step": 939 }, { "epoch": 0.03348949890446586, "grad_norm": 1.7200343686738866, "learning_rate": 1.9999373569482178e-05, "loss": 0.5726, "step": 940 }, { "epoch": 0.03352512603095997, "grad_norm": 1.6389250458240359, "learning_rate": 1.9999360586949584e-05, "loss": 0.5332, "step": 941 }, { "epoch": 0.03356075315745408, "grad_norm": 2.745846021511539, "learning_rate": 1.9999347471268517e-05, "loss": 0.6275, "step": 942 }, { "epoch": 0.0335963802839482, "grad_norm": 2.2394798470398856, "learning_rate": 1.9999334222439146e-05, "loss": 0.5679, "step": 943 }, { "epoch": 0.03363200741044231, "grad_norm": 1.6910081626805815, "learning_rate": 1.999932084046165e-05, "loss": 0.5638, "step": 944 }, { "epoch": 0.03366763453693642, "grad_norm": 1.6488516593893592, "learning_rate": 1.999930732533621e-05, "loss": 0.5574, "step": 945 }, { "epoch": 0.03370326166343054, "grad_norm": 1.61696962854524, "learning_rate": 1.9999293677063004e-05, "loss": 0.6192, "step": 946 }, { "epoch": 0.03373888878992465, "grad_norm": 1.4859175611405193, "learning_rate": 1.9999279895642212e-05, "loss": 0.5623, "step": 947 }, { "epoch": 0.03377451591641876, "grad_norm": 1.650031996149008, "learning_rate": 1.999926598107402e-05, "loss": 0.5904, "step": 948 }, { "epoch": 0.033810143042912875, "grad_norm": 3.5930085466805215, "learning_rate": 1.9999251933358614e-05, "loss": 0.6079, "step": 949 }, { "epoch": 0.033845770169406986, "grad_norm": 1.6102994984158565, "learning_rate": 1.999923775249618e-05, "loss": 0.6042, "step": 950 }, { "epoch": 0.033881397295901096, "grad_norm": 2.0651649296499004, "learning_rate": 1.99992234384869e-05, "loss": 0.6113, "step": 951 }, { "epoch": 0.03391702442239521, "grad_norm": 1.7539388315031905, "learning_rate": 1.9999208991330975e-05, "loss": 0.5857, "step": 952 }, { "epoch": 0.03395265154888932, "grad_norm": 1.4419972995057462, "learning_rate": 1.9999194411028596e-05, "loss": 0.5453, "step": 953 }, { "epoch": 0.03398827867538344, "grad_norm": 1.525975152248488, "learning_rate": 1.999917969757995e-05, "loss": 0.5744, "step": 954 }, { "epoch": 0.03402390580187755, "grad_norm": 1.586592530991568, "learning_rate": 1.9999164850985238e-05, "loss": 0.5826, "step": 955 }, { "epoch": 0.03405953292837166, "grad_norm": 2.3861202616695905, "learning_rate": 1.999914987124466e-05, "loss": 0.6543, "step": 956 }, { "epoch": 0.03409516005486578, "grad_norm": 1.6671645431635969, "learning_rate": 1.999913475835841e-05, "loss": 0.5512, "step": 957 }, { "epoch": 0.03413078718135989, "grad_norm": 1.6686646838042534, "learning_rate": 1.9999119512326694e-05, "loss": 0.5614, "step": 958 }, { "epoch": 0.034166414307854, "grad_norm": 3.281255882572006, "learning_rate": 1.999910413314971e-05, "loss": 0.6968, "step": 959 }, { "epoch": 0.034202041434348116, "grad_norm": 3.129107911673184, "learning_rate": 1.9999088620827666e-05, "loss": 0.6104, "step": 960 }, { "epoch": 0.034237668560842226, "grad_norm": 2.482726401853099, "learning_rate": 1.9999072975360766e-05, "loss": 0.686, "step": 961 }, { "epoch": 0.034273295687336336, "grad_norm": 1.9928043242900928, "learning_rate": 1.9999057196749225e-05, "loss": 0.5766, "step": 962 }, { "epoch": 0.03430892281383045, "grad_norm": 1.743157283924953, "learning_rate": 1.9999041284993246e-05, "loss": 0.5559, "step": 963 }, { "epoch": 0.03434454994032456, "grad_norm": 2.597569865290837, "learning_rate": 1.9999025240093045e-05, "loss": 0.6591, "step": 964 }, { "epoch": 0.034380177066818673, "grad_norm": 1.6054327559498422, "learning_rate": 1.999900906204883e-05, "loss": 0.5109, "step": 965 }, { "epoch": 0.03441580419331279, "grad_norm": 1.5380245161083508, "learning_rate": 1.9998992750860827e-05, "loss": 0.4916, "step": 966 }, { "epoch": 0.0344514313198069, "grad_norm": 2.181489149986937, "learning_rate": 1.999897630652924e-05, "loss": 0.7164, "step": 967 }, { "epoch": 0.03448705844630101, "grad_norm": 1.6091913256161268, "learning_rate": 1.9998959729054296e-05, "loss": 0.6003, "step": 968 }, { "epoch": 0.03452268557279513, "grad_norm": 2.5384708915803915, "learning_rate": 1.9998943018436218e-05, "loss": 0.7334, "step": 969 }, { "epoch": 0.03455831269928924, "grad_norm": 2.279578316352723, "learning_rate": 1.9998926174675223e-05, "loss": 0.7057, "step": 970 }, { "epoch": 0.03459393982578335, "grad_norm": 1.8814283631921205, "learning_rate": 1.9998909197771538e-05, "loss": 0.5903, "step": 971 }, { "epoch": 0.034629566952277466, "grad_norm": 3.582145406083926, "learning_rate": 1.9998892087725387e-05, "loss": 0.6489, "step": 972 }, { "epoch": 0.034665194078771576, "grad_norm": 2.628833819337947, "learning_rate": 1.9998874844536997e-05, "loss": 0.6684, "step": 973 }, { "epoch": 0.034700821205265686, "grad_norm": 1.5017417338970012, "learning_rate": 1.9998857468206604e-05, "loss": 0.543, "step": 974 }, { "epoch": 0.0347364483317598, "grad_norm": 1.6350920888432527, "learning_rate": 1.999883995873443e-05, "loss": 0.5594, "step": 975 }, { "epoch": 0.034772075458253913, "grad_norm": 1.906405570591804, "learning_rate": 1.9998822316120718e-05, "loss": 0.6377, "step": 976 }, { "epoch": 0.034807702584748024, "grad_norm": 1.644745485487029, "learning_rate": 1.9998804540365695e-05, "loss": 0.5629, "step": 977 }, { "epoch": 0.03484332971124214, "grad_norm": 1.6941103666072894, "learning_rate": 1.9998786631469602e-05, "loss": 0.5879, "step": 978 }, { "epoch": 0.03487895683773625, "grad_norm": 1.6363139221833647, "learning_rate": 1.999876858943268e-05, "loss": 0.5706, "step": 979 }, { "epoch": 0.03491458396423037, "grad_norm": 1.612384117593206, "learning_rate": 1.999875041425516e-05, "loss": 0.6099, "step": 980 }, { "epoch": 0.03495021109072448, "grad_norm": 1.4968064976227597, "learning_rate": 1.999873210593729e-05, "loss": 0.5654, "step": 981 }, { "epoch": 0.03498583821721859, "grad_norm": 1.696989023517478, "learning_rate": 1.9998713664479313e-05, "loss": 0.5766, "step": 982 }, { "epoch": 0.035021465343712706, "grad_norm": 3.1734216115048755, "learning_rate": 1.9998695089881477e-05, "loss": 0.7228, "step": 983 }, { "epoch": 0.035057092470206816, "grad_norm": 2.602380483736501, "learning_rate": 1.9998676382144027e-05, "loss": 0.7368, "step": 984 }, { "epoch": 0.035092719596700926, "grad_norm": 1.5599428469468355, "learning_rate": 1.999865754126721e-05, "loss": 0.5312, "step": 985 }, { "epoch": 0.03512834672319504, "grad_norm": 2.869802691654362, "learning_rate": 1.9998638567251283e-05, "loss": 0.7115, "step": 986 }, { "epoch": 0.035163973849689154, "grad_norm": 2.9538260665343685, "learning_rate": 1.999861946009649e-05, "loss": 0.7221, "step": 987 }, { "epoch": 0.035199600976183264, "grad_norm": 1.6982022872658018, "learning_rate": 1.9998600219803097e-05, "loss": 0.5799, "step": 988 }, { "epoch": 0.03523522810267738, "grad_norm": 2.44815081195676, "learning_rate": 1.999858084637135e-05, "loss": 0.7035, "step": 989 }, { "epoch": 0.03527085522917149, "grad_norm": 1.6071884578775162, "learning_rate": 1.999856133980151e-05, "loss": 0.5638, "step": 990 }, { "epoch": 0.0353064823556656, "grad_norm": 1.421134136470788, "learning_rate": 1.9998541700093836e-05, "loss": 0.5404, "step": 991 }, { "epoch": 0.03534210948215972, "grad_norm": 1.6163747416343375, "learning_rate": 1.999852192724859e-05, "loss": 0.6033, "step": 992 }, { "epoch": 0.03537773660865383, "grad_norm": 1.5393704947346425, "learning_rate": 1.999850202126604e-05, "loss": 0.5414, "step": 993 }, { "epoch": 0.03541336373514794, "grad_norm": 2.371880425229531, "learning_rate": 1.9998481982146445e-05, "loss": 0.5787, "step": 994 }, { "epoch": 0.035448990861642056, "grad_norm": 2.5993187994636187, "learning_rate": 1.9998461809890073e-05, "loss": 0.6074, "step": 995 }, { "epoch": 0.035484617988136166, "grad_norm": 1.6111744814246776, "learning_rate": 1.9998441504497194e-05, "loss": 0.5439, "step": 996 }, { "epoch": 0.035520245114630276, "grad_norm": 1.7424635647660167, "learning_rate": 1.999842106596808e-05, "loss": 0.6099, "step": 997 }, { "epoch": 0.035555872241124394, "grad_norm": 2.8464874901043404, "learning_rate": 1.9998400494302998e-05, "loss": 0.6985, "step": 998 }, { "epoch": 0.035591499367618504, "grad_norm": 2.825175225702065, "learning_rate": 1.9998379789502226e-05, "loss": 0.7612, "step": 999 }, { "epoch": 0.035627126494112614, "grad_norm": 2.0651826091127763, "learning_rate": 1.999835895156604e-05, "loss": 0.7472, "step": 1000 }, { "epoch": 0.03566275362060673, "grad_norm": 2.4807711396042897, "learning_rate": 1.9998337980494714e-05, "loss": 0.6974, "step": 1001 }, { "epoch": 0.03569838074710084, "grad_norm": 2.511969481548052, "learning_rate": 1.9998316876288532e-05, "loss": 0.6952, "step": 1002 }, { "epoch": 0.03573400787359496, "grad_norm": 2.7438672930004735, "learning_rate": 1.9998295638947768e-05, "loss": 0.7426, "step": 1003 }, { "epoch": 0.03576963500008907, "grad_norm": 2.120612659618016, "learning_rate": 1.9998274268472714e-05, "loss": 0.6626, "step": 1004 }, { "epoch": 0.03580526212658318, "grad_norm": 2.7184784323390825, "learning_rate": 1.9998252764863646e-05, "loss": 0.6919, "step": 1005 }, { "epoch": 0.035840889253077296, "grad_norm": 1.7288715323569526, "learning_rate": 1.9998231128120858e-05, "loss": 0.5768, "step": 1006 }, { "epoch": 0.035876516379571406, "grad_norm": 1.9500356887903973, "learning_rate": 1.9998209358244632e-05, "loss": 0.5455, "step": 1007 }, { "epoch": 0.035912143506065516, "grad_norm": 2.146044257632708, "learning_rate": 1.999818745523526e-05, "loss": 0.689, "step": 1008 }, { "epoch": 0.035947770632559634, "grad_norm": 2.9899852133439886, "learning_rate": 1.9998165419093032e-05, "loss": 0.7194, "step": 1009 }, { "epoch": 0.035983397759053744, "grad_norm": 2.3999914920513903, "learning_rate": 1.9998143249818245e-05, "loss": 0.618, "step": 1010 }, { "epoch": 0.036019024885547854, "grad_norm": 2.386887285164048, "learning_rate": 1.9998120947411195e-05, "loss": 0.672, "step": 1011 }, { "epoch": 0.03605465201204197, "grad_norm": 1.8925953421282735, "learning_rate": 1.9998098511872173e-05, "loss": 0.6447, "step": 1012 }, { "epoch": 0.03609027913853608, "grad_norm": 1.7905801176494347, "learning_rate": 1.999807594320148e-05, "loss": 0.4909, "step": 1013 }, { "epoch": 0.03612590626503019, "grad_norm": 2.7628254787952833, "learning_rate": 1.999805324139942e-05, "loss": 0.6885, "step": 1014 }, { "epoch": 0.03616153339152431, "grad_norm": 2.51987539870195, "learning_rate": 1.9998030406466293e-05, "loss": 0.5796, "step": 1015 }, { "epoch": 0.03619716051801842, "grad_norm": 1.481092048410637, "learning_rate": 1.99980074384024e-05, "loss": 0.5299, "step": 1016 }, { "epoch": 0.03623278764451253, "grad_norm": 2.206927055757981, "learning_rate": 1.9997984337208055e-05, "loss": 0.6449, "step": 1017 }, { "epoch": 0.036268414771006646, "grad_norm": 5.339294528245345, "learning_rate": 1.9997961102883553e-05, "loss": 0.681, "step": 1018 }, { "epoch": 0.036304041897500756, "grad_norm": 2.5147338478511085, "learning_rate": 1.999793773542922e-05, "loss": 0.5768, "step": 1019 }, { "epoch": 0.03633966902399487, "grad_norm": 1.895154683107786, "learning_rate": 1.999791423484535e-05, "loss": 0.6337, "step": 1020 }, { "epoch": 0.036375296150488984, "grad_norm": 2.9753382388539804, "learning_rate": 1.9997890601132266e-05, "loss": 0.673, "step": 1021 }, { "epoch": 0.036410923276983094, "grad_norm": 2.4788224970326027, "learning_rate": 1.999786683429028e-05, "loss": 0.7029, "step": 1022 }, { "epoch": 0.036446550403477204, "grad_norm": 2.5371594957937806, "learning_rate": 1.999784293431971e-05, "loss": 0.6487, "step": 1023 }, { "epoch": 0.03648217752997132, "grad_norm": 2.2012957534653648, "learning_rate": 1.9997818901220873e-05, "loss": 0.6531, "step": 1024 }, { "epoch": 0.03651780465646543, "grad_norm": 1.7303775400071733, "learning_rate": 1.999779473499409e-05, "loss": 0.5984, "step": 1025 }, { "epoch": 0.03655343178295955, "grad_norm": 2.425181005263833, "learning_rate": 1.999777043563968e-05, "loss": 0.6952, "step": 1026 }, { "epoch": 0.03658905890945366, "grad_norm": 2.482977268700618, "learning_rate": 1.9997746003157975e-05, "loss": 0.6724, "step": 1027 }, { "epoch": 0.03662468603594777, "grad_norm": 1.981302759853544, "learning_rate": 1.9997721437549286e-05, "loss": 0.5846, "step": 1028 }, { "epoch": 0.036660313162441886, "grad_norm": 2.0982841977717053, "learning_rate": 1.9997696738813954e-05, "loss": 0.6271, "step": 1029 }, { "epoch": 0.036695940288935996, "grad_norm": 2.0420942870657455, "learning_rate": 1.99976719069523e-05, "loss": 0.6199, "step": 1030 }, { "epoch": 0.03673156741543011, "grad_norm": 2.0452069051208364, "learning_rate": 1.9997646941964655e-05, "loss": 0.6819, "step": 1031 }, { "epoch": 0.036767194541924224, "grad_norm": 1.8727216605693324, "learning_rate": 1.9997621843851353e-05, "loss": 0.5435, "step": 1032 }, { "epoch": 0.036802821668418334, "grad_norm": 2.679705090174384, "learning_rate": 1.9997596612612732e-05, "loss": 0.7035, "step": 1033 }, { "epoch": 0.036838448794912444, "grad_norm": 2.4680901063561547, "learning_rate": 1.999757124824912e-05, "loss": 0.6322, "step": 1034 }, { "epoch": 0.03687407592140656, "grad_norm": 2.2897673907339007, "learning_rate": 1.999754575076086e-05, "loss": 0.6403, "step": 1035 }, { "epoch": 0.03690970304790067, "grad_norm": 4.981117079740189, "learning_rate": 1.9997520120148294e-05, "loss": 0.6759, "step": 1036 }, { "epoch": 0.03694533017439478, "grad_norm": 2.2743936311704065, "learning_rate": 1.9997494356411752e-05, "loss": 0.6409, "step": 1037 }, { "epoch": 0.0369809573008889, "grad_norm": 2.3043003456957463, "learning_rate": 1.999746845955159e-05, "loss": 0.7201, "step": 1038 }, { "epoch": 0.03701658442738301, "grad_norm": 1.7073260502983558, "learning_rate": 1.999744242956815e-05, "loss": 0.5347, "step": 1039 }, { "epoch": 0.03705221155387712, "grad_norm": 2.6247481820808005, "learning_rate": 1.9997416266461772e-05, "loss": 0.6204, "step": 1040 }, { "epoch": 0.037087838680371236, "grad_norm": 1.6035762081325415, "learning_rate": 1.999738997023281e-05, "loss": 0.5604, "step": 1041 }, { "epoch": 0.03712346580686535, "grad_norm": 3.081074334402596, "learning_rate": 1.999736354088161e-05, "loss": 0.6006, "step": 1042 }, { "epoch": 0.03715909293335946, "grad_norm": 2.4464322959796276, "learning_rate": 1.999733697840853e-05, "loss": 0.6403, "step": 1043 }, { "epoch": 0.037194720059853574, "grad_norm": 1.6468119480157444, "learning_rate": 1.9997310282813923e-05, "loss": 0.5225, "step": 1044 }, { "epoch": 0.037230347186347684, "grad_norm": 2.441310494466177, "learning_rate": 1.9997283454098137e-05, "loss": 0.6697, "step": 1045 }, { "epoch": 0.037265974312841794, "grad_norm": 2.7913279513256044, "learning_rate": 1.9997256492261538e-05, "loss": 0.6342, "step": 1046 }, { "epoch": 0.03730160143933591, "grad_norm": 1.8381561436187646, "learning_rate": 1.9997229397304475e-05, "loss": 0.5684, "step": 1047 }, { "epoch": 0.03733722856583002, "grad_norm": 2.666474997907829, "learning_rate": 1.999720216922732e-05, "loss": 0.6389, "step": 1048 }, { "epoch": 0.03737285569232414, "grad_norm": 1.777564680212256, "learning_rate": 1.9997174808030432e-05, "loss": 0.5955, "step": 1049 }, { "epoch": 0.03740848281881825, "grad_norm": 2.1073709824313487, "learning_rate": 1.999714731371417e-05, "loss": 0.6737, "step": 1050 }, { "epoch": 0.03744410994531236, "grad_norm": 1.5777729414941166, "learning_rate": 1.9997119686278905e-05, "loss": 0.5607, "step": 1051 }, { "epoch": 0.037479737071806477, "grad_norm": 1.5521545186691283, "learning_rate": 1.9997091925725006e-05, "loss": 0.5603, "step": 1052 }, { "epoch": 0.03751536419830059, "grad_norm": 1.5014564881193966, "learning_rate": 1.9997064032052838e-05, "loss": 0.5665, "step": 1053 }, { "epoch": 0.0375509913247947, "grad_norm": 1.4630328214399524, "learning_rate": 1.9997036005262775e-05, "loss": 0.5733, "step": 1054 }, { "epoch": 0.037586618451288814, "grad_norm": 2.189403317027976, "learning_rate": 1.999700784535519e-05, "loss": 0.7595, "step": 1055 }, { "epoch": 0.037622245577782924, "grad_norm": 1.5794066930204294, "learning_rate": 1.9996979552330457e-05, "loss": 0.5367, "step": 1056 }, { "epoch": 0.037657872704277034, "grad_norm": 2.4480413283240114, "learning_rate": 1.9996951126188957e-05, "loss": 0.6366, "step": 1057 }, { "epoch": 0.03769349983077115, "grad_norm": 2.753410430957504, "learning_rate": 1.9996922566931063e-05, "loss": 0.6073, "step": 1058 }, { "epoch": 0.03772912695726526, "grad_norm": 3.016207538740983, "learning_rate": 1.9996893874557157e-05, "loss": 0.7681, "step": 1059 }, { "epoch": 0.03776475408375937, "grad_norm": 1.592563202004805, "learning_rate": 1.9996865049067625e-05, "loss": 0.5564, "step": 1060 }, { "epoch": 0.03780038121025349, "grad_norm": 2.294951994597062, "learning_rate": 1.9996836090462844e-05, "loss": 0.7412, "step": 1061 }, { "epoch": 0.0378360083367476, "grad_norm": 1.729517129827346, "learning_rate": 1.9996806998743207e-05, "loss": 0.5777, "step": 1062 }, { "epoch": 0.03787163546324171, "grad_norm": 2.0733590287920265, "learning_rate": 1.9996777773909093e-05, "loss": 0.7427, "step": 1063 }, { "epoch": 0.03790726258973583, "grad_norm": 2.3742151231884328, "learning_rate": 1.9996748415960895e-05, "loss": 0.7429, "step": 1064 }, { "epoch": 0.03794288971622994, "grad_norm": 1.7270465550528198, "learning_rate": 1.999671892489901e-05, "loss": 0.556, "step": 1065 }, { "epoch": 0.03797851684272405, "grad_norm": 2.2700019315344777, "learning_rate": 1.9996689300723823e-05, "loss": 0.6044, "step": 1066 }, { "epoch": 0.038014143969218164, "grad_norm": 1.52697907390452, "learning_rate": 1.9996659543435728e-05, "loss": 0.5717, "step": 1067 }, { "epoch": 0.038049771095712275, "grad_norm": 2.2959188607268732, "learning_rate": 1.9996629653035128e-05, "loss": 0.7203, "step": 1068 }, { "epoch": 0.038085398222206385, "grad_norm": 2.455444038382252, "learning_rate": 1.9996599629522413e-05, "loss": 0.6753, "step": 1069 }, { "epoch": 0.0381210253487005, "grad_norm": 1.5821829123772022, "learning_rate": 1.999656947289799e-05, "loss": 0.5582, "step": 1070 }, { "epoch": 0.03815665247519461, "grad_norm": 2.2697523983305303, "learning_rate": 1.9996539183162253e-05, "loss": 0.6234, "step": 1071 }, { "epoch": 0.03819227960168873, "grad_norm": 1.8815621945759635, "learning_rate": 1.9996508760315614e-05, "loss": 0.5456, "step": 1072 }, { "epoch": 0.03822790672818284, "grad_norm": 2.515347750136821, "learning_rate": 1.9996478204358473e-05, "loss": 0.6888, "step": 1073 }, { "epoch": 0.03826353385467695, "grad_norm": 2.9062154310463586, "learning_rate": 1.9996447515291236e-05, "loss": 0.6543, "step": 1074 }, { "epoch": 0.03829916098117107, "grad_norm": 1.7841220482194906, "learning_rate": 1.9996416693114312e-05, "loss": 0.555, "step": 1075 }, { "epoch": 0.03833478810766518, "grad_norm": 2.351506206133227, "learning_rate": 1.9996385737828116e-05, "loss": 0.6462, "step": 1076 }, { "epoch": 0.03837041523415929, "grad_norm": 3.850384343634844, "learning_rate": 1.9996354649433054e-05, "loss": 0.6122, "step": 1077 }, { "epoch": 0.038406042360653404, "grad_norm": 1.7228385295200244, "learning_rate": 1.9996323427929546e-05, "loss": 0.5825, "step": 1078 }, { "epoch": 0.038441669487147515, "grad_norm": 1.5526655057218568, "learning_rate": 1.9996292073318e-05, "loss": 0.5557, "step": 1079 }, { "epoch": 0.038477296613641625, "grad_norm": 1.5367354000788924, "learning_rate": 1.9996260585598845e-05, "loss": 0.5187, "step": 1080 }, { "epoch": 0.03851292374013574, "grad_norm": 2.2235358412915365, "learning_rate": 1.9996228964772485e-05, "loss": 0.6817, "step": 1081 }, { "epoch": 0.03854855086662985, "grad_norm": 2.3010943765302145, "learning_rate": 1.9996197210839357e-05, "loss": 0.6726, "step": 1082 }, { "epoch": 0.03858417799312396, "grad_norm": 2.703031209686954, "learning_rate": 1.9996165323799873e-05, "loss": 0.6214, "step": 1083 }, { "epoch": 0.03861980511961808, "grad_norm": 2.226806026453244, "learning_rate": 1.9996133303654458e-05, "loss": 0.6828, "step": 1084 }, { "epoch": 0.03865543224611219, "grad_norm": 2.4613553402950745, "learning_rate": 1.9996101150403543e-05, "loss": 0.7429, "step": 1085 }, { "epoch": 0.0386910593726063, "grad_norm": 1.7547660114882913, "learning_rate": 1.9996068864047558e-05, "loss": 0.5813, "step": 1086 }, { "epoch": 0.03872668649910042, "grad_norm": 1.5219900120598218, "learning_rate": 1.9996036444586925e-05, "loss": 0.5823, "step": 1087 }, { "epoch": 0.03876231362559453, "grad_norm": 1.680672954616601, "learning_rate": 1.9996003892022084e-05, "loss": 0.6133, "step": 1088 }, { "epoch": 0.03879794075208864, "grad_norm": 1.59515479917737, "learning_rate": 1.999597120635346e-05, "loss": 0.6538, "step": 1089 }, { "epoch": 0.038833567878582755, "grad_norm": 1.4745816659424305, "learning_rate": 1.9995938387581494e-05, "loss": 0.5309, "step": 1090 }, { "epoch": 0.038869195005076865, "grad_norm": 2.4721508503607343, "learning_rate": 1.9995905435706623e-05, "loss": 0.7257, "step": 1091 }, { "epoch": 0.038904822131570975, "grad_norm": 2.482040601052791, "learning_rate": 1.9995872350729282e-05, "loss": 0.7065, "step": 1092 }, { "epoch": 0.03894044925806509, "grad_norm": 3.079174006393166, "learning_rate": 1.9995839132649916e-05, "loss": 0.6268, "step": 1093 }, { "epoch": 0.0389760763845592, "grad_norm": 1.4366404535989303, "learning_rate": 1.9995805781468965e-05, "loss": 0.5895, "step": 1094 }, { "epoch": 0.03901170351105332, "grad_norm": 2.9719960364278393, "learning_rate": 1.9995772297186872e-05, "loss": 0.7396, "step": 1095 }, { "epoch": 0.03904733063754743, "grad_norm": 2.204951862430209, "learning_rate": 1.9995738679804086e-05, "loss": 0.6572, "step": 1096 }, { "epoch": 0.03908295776404154, "grad_norm": 1.6097564704370468, "learning_rate": 1.9995704929321053e-05, "loss": 0.5203, "step": 1097 }, { "epoch": 0.03911858489053566, "grad_norm": 2.41028875261739, "learning_rate": 1.999567104573822e-05, "loss": 0.6987, "step": 1098 }, { "epoch": 0.03915421201702977, "grad_norm": 1.4070709250542561, "learning_rate": 1.9995637029056042e-05, "loss": 0.5192, "step": 1099 }, { "epoch": 0.03918983914352388, "grad_norm": 1.5589247995497644, "learning_rate": 1.999560287927497e-05, "loss": 0.5709, "step": 1100 }, { "epoch": 0.039225466270017995, "grad_norm": 2.2485108920324333, "learning_rate": 1.999556859639546e-05, "loss": 0.7094, "step": 1101 }, { "epoch": 0.039261093396512105, "grad_norm": 2.1048616105647193, "learning_rate": 1.9995534180417965e-05, "loss": 0.6345, "step": 1102 }, { "epoch": 0.039296720523006215, "grad_norm": 2.198556459041046, "learning_rate": 1.999549963134295e-05, "loss": 0.7045, "step": 1103 }, { "epoch": 0.03933234764950033, "grad_norm": 1.5958796729170244, "learning_rate": 1.9995464949170867e-05, "loss": 0.5839, "step": 1104 }, { "epoch": 0.03936797477599444, "grad_norm": 1.7583276913702444, "learning_rate": 1.9995430133902184e-05, "loss": 0.5235, "step": 1105 }, { "epoch": 0.03940360190248855, "grad_norm": 1.6317277584393888, "learning_rate": 1.9995395185537364e-05, "loss": 0.5445, "step": 1106 }, { "epoch": 0.03943922902898267, "grad_norm": 1.4128277169916075, "learning_rate": 1.999536010407687e-05, "loss": 0.5632, "step": 1107 }, { "epoch": 0.03947485615547678, "grad_norm": 2.901221006999806, "learning_rate": 1.9995324889521167e-05, "loss": 0.7014, "step": 1108 }, { "epoch": 0.03951048328197089, "grad_norm": 3.6017026111801345, "learning_rate": 1.999528954187073e-05, "loss": 0.638, "step": 1109 }, { "epoch": 0.03954611040846501, "grad_norm": 2.066170543333192, "learning_rate": 1.9995254061126023e-05, "loss": 0.6231, "step": 1110 }, { "epoch": 0.03958173753495912, "grad_norm": 2.4121957946515784, "learning_rate": 1.9995218447287522e-05, "loss": 0.6835, "step": 1111 }, { "epoch": 0.03961736466145323, "grad_norm": 1.7788318919187514, "learning_rate": 1.9995182700355703e-05, "loss": 0.6704, "step": 1112 }, { "epoch": 0.039652991787947345, "grad_norm": 1.8346466765895866, "learning_rate": 1.999514682033104e-05, "loss": 0.5732, "step": 1113 }, { "epoch": 0.039688618914441455, "grad_norm": 1.782140685243504, "learning_rate": 1.9995110807214015e-05, "loss": 0.5743, "step": 1114 }, { "epoch": 0.039724246040935565, "grad_norm": 2.479155951165672, "learning_rate": 1.9995074661005098e-05, "loss": 0.7234, "step": 1115 }, { "epoch": 0.03975987316742968, "grad_norm": 2.396322665378456, "learning_rate": 1.9995038381704777e-05, "loss": 0.5861, "step": 1116 }, { "epoch": 0.03979550029392379, "grad_norm": 2.1018238456310074, "learning_rate": 1.9995001969313534e-05, "loss": 0.5535, "step": 1117 }, { "epoch": 0.0398311274204179, "grad_norm": 3.0234137536039203, "learning_rate": 1.9994965423831853e-05, "loss": 0.6895, "step": 1118 }, { "epoch": 0.03986675454691202, "grad_norm": 1.6032956917426573, "learning_rate": 1.999492874526022e-05, "loss": 0.5729, "step": 1119 }, { "epoch": 0.03990238167340613, "grad_norm": 2.2315818348668737, "learning_rate": 1.9994891933599128e-05, "loss": 0.6607, "step": 1120 }, { "epoch": 0.03993800879990025, "grad_norm": 2.402303363150337, "learning_rate": 1.9994854988849063e-05, "loss": 0.7471, "step": 1121 }, { "epoch": 0.03997363592639436, "grad_norm": 1.749912803308359, "learning_rate": 1.9994817911010518e-05, "loss": 0.5369, "step": 1122 }, { "epoch": 0.04000926305288847, "grad_norm": 1.9855790969510583, "learning_rate": 1.9994780700083985e-05, "loss": 0.5922, "step": 1123 }, { "epoch": 0.040044890179382585, "grad_norm": 2.805542543573043, "learning_rate": 1.9994743356069962e-05, "loss": 0.6641, "step": 1124 }, { "epoch": 0.040080517305876695, "grad_norm": 1.6112482920660969, "learning_rate": 1.9994705878968946e-05, "loss": 0.5459, "step": 1125 }, { "epoch": 0.040116144432370805, "grad_norm": 1.47862979317857, "learning_rate": 1.9994668268781433e-05, "loss": 0.5929, "step": 1126 }, { "epoch": 0.04015177155886492, "grad_norm": 1.6059036727447245, "learning_rate": 1.999463052550793e-05, "loss": 0.6261, "step": 1127 }, { "epoch": 0.04018739868535903, "grad_norm": 1.5395749987091163, "learning_rate": 1.9994592649148935e-05, "loss": 0.6244, "step": 1128 }, { "epoch": 0.04022302581185314, "grad_norm": 2.368759602911135, "learning_rate": 1.999455463970495e-05, "loss": 0.5965, "step": 1129 }, { "epoch": 0.04025865293834726, "grad_norm": 1.5427904758138773, "learning_rate": 1.9994516497176485e-05, "loss": 0.604, "step": 1130 }, { "epoch": 0.04029428006484137, "grad_norm": 2.115779524147135, "learning_rate": 1.999447822156405e-05, "loss": 0.5992, "step": 1131 }, { "epoch": 0.04032990719133548, "grad_norm": 2.1557176010566406, "learning_rate": 1.999443981286815e-05, "loss": 0.7465, "step": 1132 }, { "epoch": 0.0403655343178296, "grad_norm": 2.51745491049062, "learning_rate": 1.9994401271089295e-05, "loss": 0.6271, "step": 1133 }, { "epoch": 0.04040116144432371, "grad_norm": 2.4974287701000137, "learning_rate": 1.9994362596228005e-05, "loss": 0.6698, "step": 1134 }, { "epoch": 0.04043678857081782, "grad_norm": 1.6868716762827978, "learning_rate": 1.999432378828479e-05, "loss": 0.5378, "step": 1135 }, { "epoch": 0.040472415697311935, "grad_norm": 1.7635069364806695, "learning_rate": 1.999428484726017e-05, "loss": 0.5427, "step": 1136 }, { "epoch": 0.040508042823806045, "grad_norm": 1.5240136155518391, "learning_rate": 1.999424577315466e-05, "loss": 0.5545, "step": 1137 }, { "epoch": 0.040543669950300155, "grad_norm": 2.705048085866608, "learning_rate": 1.9994206565968783e-05, "loss": 0.6964, "step": 1138 }, { "epoch": 0.04057929707679427, "grad_norm": 2.047829203298459, "learning_rate": 1.999416722570306e-05, "loss": 0.6542, "step": 1139 }, { "epoch": 0.04061492420328838, "grad_norm": 2.5010659200613206, "learning_rate": 1.9994127752358014e-05, "loss": 0.6558, "step": 1140 }, { "epoch": 0.04065055132978249, "grad_norm": 2.021493565800565, "learning_rate": 1.999408814593417e-05, "loss": 0.6129, "step": 1141 }, { "epoch": 0.04068617845627661, "grad_norm": 2.1669611133678934, "learning_rate": 1.999404840643206e-05, "loss": 0.6131, "step": 1142 }, { "epoch": 0.04072180558277072, "grad_norm": 1.4743168364500294, "learning_rate": 1.999400853385221e-05, "loss": 0.5138, "step": 1143 }, { "epoch": 0.04075743270926484, "grad_norm": 2.0569152816853173, "learning_rate": 1.999396852819515e-05, "loss": 0.7519, "step": 1144 }, { "epoch": 0.04079305983575895, "grad_norm": 1.838196408271598, "learning_rate": 1.9993928389461414e-05, "loss": 0.5793, "step": 1145 }, { "epoch": 0.04082868696225306, "grad_norm": 1.837240392661566, "learning_rate": 1.9993888117651537e-05, "loss": 0.658, "step": 1146 }, { "epoch": 0.040864314088747175, "grad_norm": 1.4957309879176828, "learning_rate": 1.9993847712766052e-05, "loss": 0.5601, "step": 1147 }, { "epoch": 0.040899941215241285, "grad_norm": 1.761579069966494, "learning_rate": 1.99938071748055e-05, "loss": 0.5429, "step": 1148 }, { "epoch": 0.040935568341735395, "grad_norm": 1.6608299167428253, "learning_rate": 1.999376650377042e-05, "loss": 0.5814, "step": 1149 }, { "epoch": 0.04097119546822951, "grad_norm": 1.635821741553358, "learning_rate": 1.9993725699661354e-05, "loss": 0.54, "step": 1150 }, { "epoch": 0.04100682259472362, "grad_norm": 1.5764662889128673, "learning_rate": 1.9993684762478846e-05, "loss": 0.614, "step": 1151 }, { "epoch": 0.04104244972121773, "grad_norm": 2.293820893485768, "learning_rate": 1.999364369222344e-05, "loss": 0.6509, "step": 1152 }, { "epoch": 0.04107807684771185, "grad_norm": 2.5519854441603704, "learning_rate": 1.9993602488895684e-05, "loss": 0.6518, "step": 1153 }, { "epoch": 0.04111370397420596, "grad_norm": 2.126697149147884, "learning_rate": 1.9993561152496123e-05, "loss": 0.685, "step": 1154 }, { "epoch": 0.04114933110070007, "grad_norm": 1.452952985114855, "learning_rate": 1.999351968302531e-05, "loss": 0.5496, "step": 1155 }, { "epoch": 0.04118495822719419, "grad_norm": 1.4216973877409527, "learning_rate": 1.99934780804838e-05, "loss": 0.5891, "step": 1156 }, { "epoch": 0.0412205853536883, "grad_norm": 2.228086631330292, "learning_rate": 1.9993436344872147e-05, "loss": 0.6983, "step": 1157 }, { "epoch": 0.04125621248018241, "grad_norm": 2.2298611423167354, "learning_rate": 1.99933944761909e-05, "loss": 0.6957, "step": 1158 }, { "epoch": 0.041291839606676525, "grad_norm": 1.4625901375710137, "learning_rate": 1.9993352474440622e-05, "loss": 0.6037, "step": 1159 }, { "epoch": 0.041327466733170636, "grad_norm": 1.4764079960768672, "learning_rate": 1.9993310339621866e-05, "loss": 0.547, "step": 1160 }, { "epoch": 0.041363093859664746, "grad_norm": 2.076671343466842, "learning_rate": 1.9993268071735206e-05, "loss": 0.6719, "step": 1161 }, { "epoch": 0.04139872098615886, "grad_norm": 2.202489033072544, "learning_rate": 1.9993225670781192e-05, "loss": 0.6531, "step": 1162 }, { "epoch": 0.04143434811265297, "grad_norm": 2.366359018035933, "learning_rate": 1.9993183136760392e-05, "loss": 0.6282, "step": 1163 }, { "epoch": 0.04146997523914708, "grad_norm": 1.8121359062812687, "learning_rate": 1.999314046967338e-05, "loss": 0.6037, "step": 1164 }, { "epoch": 0.0415056023656412, "grad_norm": 2.432596384318843, "learning_rate": 1.9993097669520712e-05, "loss": 0.6186, "step": 1165 }, { "epoch": 0.04154122949213531, "grad_norm": 4.7463864673841565, "learning_rate": 1.9993054736302967e-05, "loss": 0.6861, "step": 1166 }, { "epoch": 0.04157685661862943, "grad_norm": 1.6713944188941285, "learning_rate": 1.999301167002071e-05, "loss": 0.5578, "step": 1167 }, { "epoch": 0.04161248374512354, "grad_norm": 1.637126523743615, "learning_rate": 1.999296847067452e-05, "loss": 0.5194, "step": 1168 }, { "epoch": 0.04164811087161765, "grad_norm": 2.643862352790166, "learning_rate": 1.999292513826497e-05, "loss": 0.6626, "step": 1169 }, { "epoch": 0.041683737998111765, "grad_norm": 2.43208893699471, "learning_rate": 1.9992881672792637e-05, "loss": 0.6024, "step": 1170 }, { "epoch": 0.041719365124605876, "grad_norm": 1.5672485282668926, "learning_rate": 1.99928380742581e-05, "loss": 0.567, "step": 1171 }, { "epoch": 0.041754992251099986, "grad_norm": 2.8566536995431164, "learning_rate": 1.9992794342661937e-05, "loss": 0.6682, "step": 1172 }, { "epoch": 0.0417906193775941, "grad_norm": 1.8754296560872166, "learning_rate": 1.999275047800474e-05, "loss": 0.6407, "step": 1173 }, { "epoch": 0.04182624650408821, "grad_norm": 2.753835709524637, "learning_rate": 1.9992706480287077e-05, "loss": 0.6451, "step": 1174 }, { "epoch": 0.04186187363058232, "grad_norm": 2.1224913248395048, "learning_rate": 1.9992662349509547e-05, "loss": 0.6734, "step": 1175 }, { "epoch": 0.04189750075707644, "grad_norm": 2.254804641470024, "learning_rate": 1.9992618085672736e-05, "loss": 0.6017, "step": 1176 }, { "epoch": 0.04193312788357055, "grad_norm": 1.6258296645535768, "learning_rate": 1.9992573688777226e-05, "loss": 0.5205, "step": 1177 }, { "epoch": 0.04196875501006466, "grad_norm": 1.7389065908515091, "learning_rate": 1.9992529158823616e-05, "loss": 0.5526, "step": 1178 }, { "epoch": 0.04200438213655878, "grad_norm": 1.5329959525875074, "learning_rate": 1.9992484495812496e-05, "loss": 0.5417, "step": 1179 }, { "epoch": 0.04204000926305289, "grad_norm": 2.21734150499949, "learning_rate": 1.9992439699744457e-05, "loss": 0.6501, "step": 1180 }, { "epoch": 0.042075636389547, "grad_norm": 2.6996046195325194, "learning_rate": 1.99923947706201e-05, "loss": 0.6091, "step": 1181 }, { "epoch": 0.042111263516041116, "grad_norm": 1.6712288107879987, "learning_rate": 1.9992349708440026e-05, "loss": 0.5658, "step": 1182 }, { "epoch": 0.042146890642535226, "grad_norm": 2.3099176892585853, "learning_rate": 1.999230451320483e-05, "loss": 0.6873, "step": 1183 }, { "epoch": 0.042182517769029336, "grad_norm": 1.6340328122568142, "learning_rate": 1.9992259184915115e-05, "loss": 0.5525, "step": 1184 }, { "epoch": 0.04221814489552345, "grad_norm": 3.0545287127917278, "learning_rate": 1.9992213723571482e-05, "loss": 0.6261, "step": 1185 }, { "epoch": 0.04225377202201756, "grad_norm": 2.590990973150484, "learning_rate": 1.9992168129174544e-05, "loss": 0.6581, "step": 1186 }, { "epoch": 0.042289399148511674, "grad_norm": 1.970445639862125, "learning_rate": 1.99921224017249e-05, "loss": 0.5658, "step": 1187 }, { "epoch": 0.04232502627500579, "grad_norm": 1.3988151273247955, "learning_rate": 1.9992076541223163e-05, "loss": 0.4563, "step": 1188 }, { "epoch": 0.0423606534014999, "grad_norm": 2.9203172150189087, "learning_rate": 1.999203054766994e-05, "loss": 0.6205, "step": 1189 }, { "epoch": 0.04239628052799402, "grad_norm": 1.4698608329447647, "learning_rate": 1.9991984421065852e-05, "loss": 0.5756, "step": 1190 }, { "epoch": 0.04243190765448813, "grad_norm": 2.2995742659869007, "learning_rate": 1.9991938161411505e-05, "loss": 0.7089, "step": 1191 }, { "epoch": 0.04246753478098224, "grad_norm": 1.5535512702466365, "learning_rate": 1.9991891768707515e-05, "loss": 0.5431, "step": 1192 }, { "epoch": 0.042503161907476356, "grad_norm": 1.7731546666817986, "learning_rate": 1.9991845242954505e-05, "loss": 0.6272, "step": 1193 }, { "epoch": 0.042538789033970466, "grad_norm": 2.2062303770753884, "learning_rate": 1.999179858415309e-05, "loss": 0.6296, "step": 1194 }, { "epoch": 0.042574416160464576, "grad_norm": 1.5250872942681337, "learning_rate": 1.9991751792303892e-05, "loss": 0.5925, "step": 1195 }, { "epoch": 0.04261004328695869, "grad_norm": 3.225987681504548, "learning_rate": 1.9991704867407537e-05, "loss": 0.6014, "step": 1196 }, { "epoch": 0.0426456704134528, "grad_norm": 2.768436086745788, "learning_rate": 1.999165780946465e-05, "loss": 0.6507, "step": 1197 }, { "epoch": 0.042681297539946914, "grad_norm": 2.059900958179866, "learning_rate": 1.9991610618475852e-05, "loss": 0.6139, "step": 1198 }, { "epoch": 0.04271692466644103, "grad_norm": 1.3671573974340114, "learning_rate": 1.9991563294441777e-05, "loss": 0.5615, "step": 1199 }, { "epoch": 0.04275255179293514, "grad_norm": 1.7979307318009567, "learning_rate": 1.999151583736305e-05, "loss": 0.5741, "step": 1200 }, { "epoch": 0.04278817891942925, "grad_norm": 1.8856381398591249, "learning_rate": 1.9991468247240307e-05, "loss": 0.6781, "step": 1201 }, { "epoch": 0.04282380604592337, "grad_norm": 1.6722408873422845, "learning_rate": 1.9991420524074183e-05, "loss": 0.591, "step": 1202 }, { "epoch": 0.04285943317241748, "grad_norm": 2.793960565979352, "learning_rate": 1.999137266786531e-05, "loss": 0.6662, "step": 1203 }, { "epoch": 0.04289506029891159, "grad_norm": 1.5354541387801273, "learning_rate": 1.999132467861433e-05, "loss": 0.6144, "step": 1204 }, { "epoch": 0.042930687425405706, "grad_norm": 1.5167058498889257, "learning_rate": 1.999127655632187e-05, "loss": 0.5504, "step": 1205 }, { "epoch": 0.042966314551899816, "grad_norm": 1.7445703730040725, "learning_rate": 1.9991228300988586e-05, "loss": 0.7163, "step": 1206 }, { "epoch": 0.043001941678393926, "grad_norm": 2.5519234654315732, "learning_rate": 1.9991179912615112e-05, "loss": 0.6258, "step": 1207 }, { "epoch": 0.04303756880488804, "grad_norm": 1.471569763683072, "learning_rate": 1.9991131391202093e-05, "loss": 0.5472, "step": 1208 }, { "epoch": 0.043073195931382154, "grad_norm": 2.3525476058751114, "learning_rate": 1.9991082736750176e-05, "loss": 0.6586, "step": 1209 }, { "epoch": 0.043108823057876264, "grad_norm": 3.050557731182539, "learning_rate": 1.9991033949260008e-05, "loss": 0.6577, "step": 1210 }, { "epoch": 0.04314445018437038, "grad_norm": 1.8511827598451311, "learning_rate": 1.9990985028732243e-05, "loss": 0.5705, "step": 1211 }, { "epoch": 0.04318007731086449, "grad_norm": 1.4264373163904078, "learning_rate": 1.999093597516753e-05, "loss": 0.5513, "step": 1212 }, { "epoch": 0.04321570443735861, "grad_norm": 1.4748130861555875, "learning_rate": 1.9990886788566515e-05, "loss": 0.5052, "step": 1213 }, { "epoch": 0.04325133156385272, "grad_norm": 2.030011569425698, "learning_rate": 1.9990837468929864e-05, "loss": 0.7019, "step": 1214 }, { "epoch": 0.04328695869034683, "grad_norm": 2.7587688524010834, "learning_rate": 1.9990788016258227e-05, "loss": 0.7343, "step": 1215 }, { "epoch": 0.043322585816840946, "grad_norm": 2.6600431447468567, "learning_rate": 1.9990738430552267e-05, "loss": 0.6658, "step": 1216 }, { "epoch": 0.043358212943335056, "grad_norm": 1.7001900529844896, "learning_rate": 1.999068871181264e-05, "loss": 0.5991, "step": 1217 }, { "epoch": 0.043393840069829166, "grad_norm": 1.5796632035499258, "learning_rate": 1.9990638860040007e-05, "loss": 0.6371, "step": 1218 }, { "epoch": 0.04342946719632328, "grad_norm": 1.4661527933926184, "learning_rate": 1.999058887523504e-05, "loss": 0.5787, "step": 1219 }, { "epoch": 0.043465094322817394, "grad_norm": 1.51811675287017, "learning_rate": 1.9990538757398394e-05, "loss": 0.5411, "step": 1220 }, { "epoch": 0.043500721449311504, "grad_norm": 2.444955690491294, "learning_rate": 1.9990488506530746e-05, "loss": 0.7203, "step": 1221 }, { "epoch": 0.04353634857580562, "grad_norm": 2.1645365119715283, "learning_rate": 1.9990438122632757e-05, "loss": 0.6853, "step": 1222 }, { "epoch": 0.04357197570229973, "grad_norm": 2.47079158406179, "learning_rate": 1.9990387605705104e-05, "loss": 0.6612, "step": 1223 }, { "epoch": 0.04360760282879384, "grad_norm": 2.547010006252241, "learning_rate": 1.9990336955748456e-05, "loss": 0.6356, "step": 1224 }, { "epoch": 0.04364322995528796, "grad_norm": 1.5525583437970778, "learning_rate": 1.999028617276349e-05, "loss": 0.5379, "step": 1225 }, { "epoch": 0.04367885708178207, "grad_norm": 2.1555683274337736, "learning_rate": 1.9990235256750878e-05, "loss": 0.6589, "step": 1226 }, { "epoch": 0.04371448420827618, "grad_norm": 2.8609304454063684, "learning_rate": 1.99901842077113e-05, "loss": 0.6039, "step": 1227 }, { "epoch": 0.043750111334770296, "grad_norm": 1.5108387528566076, "learning_rate": 1.999013302564544e-05, "loss": 0.5865, "step": 1228 }, { "epoch": 0.043785738461264406, "grad_norm": 2.446569047400587, "learning_rate": 1.9990081710553976e-05, "loss": 0.6774, "step": 1229 }, { "epoch": 0.043821365587758516, "grad_norm": 2.5104118690298614, "learning_rate": 1.999003026243759e-05, "loss": 0.6445, "step": 1230 }, { "epoch": 0.043856992714252634, "grad_norm": 1.5136443620406854, "learning_rate": 1.9989978681296968e-05, "loss": 0.554, "step": 1231 }, { "epoch": 0.043892619840746744, "grad_norm": 1.5028864507039903, "learning_rate": 1.9989926967132798e-05, "loss": 0.5842, "step": 1232 }, { "epoch": 0.043928246967240854, "grad_norm": 1.4923898867688288, "learning_rate": 1.998987511994577e-05, "loss": 0.6366, "step": 1233 }, { "epoch": 0.04396387409373497, "grad_norm": 2.079722344812138, "learning_rate": 1.998982313973657e-05, "loss": 0.7103, "step": 1234 }, { "epoch": 0.04399950122022908, "grad_norm": 1.5241485269096102, "learning_rate": 1.998977102650589e-05, "loss": 0.5602, "step": 1235 }, { "epoch": 0.04403512834672319, "grad_norm": 2.359508403899693, "learning_rate": 1.998971878025443e-05, "loss": 0.696, "step": 1236 }, { "epoch": 0.04407075547321731, "grad_norm": 2.166774809872781, "learning_rate": 1.998966640098288e-05, "loss": 0.7129, "step": 1237 }, { "epoch": 0.04410638259971142, "grad_norm": 1.4832245644974342, "learning_rate": 1.998961388869194e-05, "loss": 0.4699, "step": 1238 }, { "epoch": 0.044142009726205536, "grad_norm": 1.669375186998595, "learning_rate": 1.9989561243382313e-05, "loss": 0.6243, "step": 1239 }, { "epoch": 0.044177636852699646, "grad_norm": 2.1638991772555562, "learning_rate": 1.998950846505469e-05, "loss": 0.7042, "step": 1240 }, { "epoch": 0.044213263979193757, "grad_norm": 2.2480499707266954, "learning_rate": 1.9989455553709784e-05, "loss": 0.6553, "step": 1241 }, { "epoch": 0.044248891105687874, "grad_norm": 2.6098705598458642, "learning_rate": 1.998940250934829e-05, "loss": 0.6042, "step": 1242 }, { "epoch": 0.044284518232181984, "grad_norm": 2.4279877573094315, "learning_rate": 1.998934933197092e-05, "loss": 0.723, "step": 1243 }, { "epoch": 0.044320145358676094, "grad_norm": 1.9177675660798867, "learning_rate": 1.9989296021578382e-05, "loss": 0.5716, "step": 1244 }, { "epoch": 0.04435577248517021, "grad_norm": 2.1053661648439785, "learning_rate": 1.998924257817139e-05, "loss": 0.559, "step": 1245 }, { "epoch": 0.04439139961166432, "grad_norm": 1.678644506000278, "learning_rate": 1.998918900175065e-05, "loss": 0.5641, "step": 1246 }, { "epoch": 0.04442702673815843, "grad_norm": 2.0405961598883096, "learning_rate": 1.998913529231687e-05, "loss": 0.6582, "step": 1247 }, { "epoch": 0.04446265386465255, "grad_norm": 2.322253751778209, "learning_rate": 1.998908144987078e-05, "loss": 0.6535, "step": 1248 }, { "epoch": 0.04449828099114666, "grad_norm": 2.376486739312001, "learning_rate": 1.9989027474413083e-05, "loss": 0.6723, "step": 1249 }, { "epoch": 0.04453390811764077, "grad_norm": 3.167981692575479, "learning_rate": 1.9988973365944505e-05, "loss": 0.8165, "step": 1250 }, { "epoch": 0.044569535244134886, "grad_norm": 1.8667620846910906, "learning_rate": 1.9988919124465765e-05, "loss": 0.5724, "step": 1251 }, { "epoch": 0.044605162370628997, "grad_norm": 2.425637799809798, "learning_rate": 1.9988864749977584e-05, "loss": 0.718, "step": 1252 }, { "epoch": 0.04464078949712311, "grad_norm": 1.5096994979068155, "learning_rate": 1.9988810242480686e-05, "loss": 0.5389, "step": 1253 }, { "epoch": 0.044676416623617224, "grad_norm": 4.238792402688405, "learning_rate": 1.99887556019758e-05, "loss": 0.6272, "step": 1254 }, { "epoch": 0.044712043750111334, "grad_norm": 2.2302818372257374, "learning_rate": 1.998870082846365e-05, "loss": 0.6005, "step": 1255 }, { "epoch": 0.044747670876605444, "grad_norm": 1.5376962865712582, "learning_rate": 1.9988645921944968e-05, "loss": 0.5559, "step": 1256 }, { "epoch": 0.04478329800309956, "grad_norm": 2.2152430275967214, "learning_rate": 1.9988590882420486e-05, "loss": 0.5745, "step": 1257 }, { "epoch": 0.04481892512959367, "grad_norm": 1.4992054419527112, "learning_rate": 1.998853570989093e-05, "loss": 0.4816, "step": 1258 }, { "epoch": 0.04485455225608778, "grad_norm": 1.6163581695558307, "learning_rate": 1.9988480404357044e-05, "loss": 0.5976, "step": 1259 }, { "epoch": 0.0448901793825819, "grad_norm": 2.8366283284011797, "learning_rate": 1.9988424965819556e-05, "loss": 0.6978, "step": 1260 }, { "epoch": 0.04492580650907601, "grad_norm": 2.032942447606372, "learning_rate": 1.998836939427921e-05, "loss": 0.5358, "step": 1261 }, { "epoch": 0.044961433635570126, "grad_norm": 1.545268168145807, "learning_rate": 1.9988313689736746e-05, "loss": 0.5723, "step": 1262 }, { "epoch": 0.04499706076206424, "grad_norm": 2.375816682275043, "learning_rate": 1.99882578521929e-05, "loss": 0.6842, "step": 1263 }, { "epoch": 0.04503268788855835, "grad_norm": 2.6160979195103558, "learning_rate": 1.998820188164842e-05, "loss": 0.618, "step": 1264 }, { "epoch": 0.045068315015052464, "grad_norm": 1.5839714678983816, "learning_rate": 1.9988145778104055e-05, "loss": 0.5154, "step": 1265 }, { "epoch": 0.045103942141546574, "grad_norm": 1.5632085568222833, "learning_rate": 1.9988089541560543e-05, "loss": 0.6119, "step": 1266 }, { "epoch": 0.045139569268040684, "grad_norm": 1.9670656578095123, "learning_rate": 1.998803317201864e-05, "loss": 0.5519, "step": 1267 }, { "epoch": 0.0451751963945348, "grad_norm": 1.8991109469307794, "learning_rate": 1.998797666947909e-05, "loss": 0.6941, "step": 1268 }, { "epoch": 0.04521082352102891, "grad_norm": 1.635215849050729, "learning_rate": 1.9987920033942653e-05, "loss": 0.5414, "step": 1269 }, { "epoch": 0.04524645064752302, "grad_norm": 1.5277748933257786, "learning_rate": 1.998786326541008e-05, "loss": 0.5698, "step": 1270 }, { "epoch": 0.04528207777401714, "grad_norm": 1.5154896586078261, "learning_rate": 1.9987806363882124e-05, "loss": 0.5463, "step": 1271 }, { "epoch": 0.04531770490051125, "grad_norm": 1.48624283121224, "learning_rate": 1.998774932935955e-05, "loss": 0.5606, "step": 1272 }, { "epoch": 0.04535333202700536, "grad_norm": 2.3272671361811166, "learning_rate": 1.9987692161843107e-05, "loss": 0.6646, "step": 1273 }, { "epoch": 0.04538895915349948, "grad_norm": 2.2683824978521048, "learning_rate": 1.998763486133356e-05, "loss": 0.6526, "step": 1274 }, { "epoch": 0.04542458627999359, "grad_norm": 4.97339063610075, "learning_rate": 1.998757742783168e-05, "loss": 0.6303, "step": 1275 }, { "epoch": 0.0454602134064877, "grad_norm": 1.8808546766356753, "learning_rate": 1.9987519861338226e-05, "loss": 0.6875, "step": 1276 }, { "epoch": 0.045495840532981814, "grad_norm": 2.0901802028354872, "learning_rate": 1.998746216185396e-05, "loss": 0.7266, "step": 1277 }, { "epoch": 0.045531467659475924, "grad_norm": 2.8381410635357622, "learning_rate": 1.998740432937966e-05, "loss": 0.6448, "step": 1278 }, { "epoch": 0.045567094785970035, "grad_norm": 2.018260518720068, "learning_rate": 1.9987346363916085e-05, "loss": 0.6415, "step": 1279 }, { "epoch": 0.04560272191246415, "grad_norm": 2.4671242512240505, "learning_rate": 1.9987288265464017e-05, "loss": 0.7174, "step": 1280 }, { "epoch": 0.04563834903895826, "grad_norm": 1.6054171688815877, "learning_rate": 1.9987230034024222e-05, "loss": 0.5368, "step": 1281 }, { "epoch": 0.04567397616545237, "grad_norm": 1.6701615208758749, "learning_rate": 1.9987171669597476e-05, "loss": 0.5833, "step": 1282 }, { "epoch": 0.04570960329194649, "grad_norm": 2.1718479144741223, "learning_rate": 1.9987113172184562e-05, "loss": 0.6547, "step": 1283 }, { "epoch": 0.0457452304184406, "grad_norm": 1.69866436925894, "learning_rate": 1.9987054541786256e-05, "loss": 0.574, "step": 1284 }, { "epoch": 0.04578085754493472, "grad_norm": 1.5626153704786492, "learning_rate": 1.9986995778403336e-05, "loss": 0.5366, "step": 1285 }, { "epoch": 0.04581648467142883, "grad_norm": 1.5088500659022095, "learning_rate": 1.998693688203659e-05, "loss": 0.5409, "step": 1286 }, { "epoch": 0.04585211179792294, "grad_norm": 1.4837350926201052, "learning_rate": 1.9986877852686794e-05, "loss": 0.5587, "step": 1287 }, { "epoch": 0.045887738924417054, "grad_norm": 1.4822180384007673, "learning_rate": 1.9986818690354745e-05, "loss": 0.5443, "step": 1288 }, { "epoch": 0.045923366050911164, "grad_norm": 3.226193508949785, "learning_rate": 1.998675939504122e-05, "loss": 0.63, "step": 1289 }, { "epoch": 0.045958993177405275, "grad_norm": 1.6325835721835684, "learning_rate": 1.9986699966747018e-05, "loss": 0.5777, "step": 1290 }, { "epoch": 0.04599462030389939, "grad_norm": 1.587460020066661, "learning_rate": 1.998664040547292e-05, "loss": 0.5922, "step": 1291 }, { "epoch": 0.0460302474303935, "grad_norm": 1.429793310576156, "learning_rate": 1.9986580711219726e-05, "loss": 0.5476, "step": 1292 }, { "epoch": 0.04606587455688761, "grad_norm": 1.5590661611266956, "learning_rate": 1.9986520883988233e-05, "loss": 0.5354, "step": 1293 }, { "epoch": 0.04610150168338173, "grad_norm": 1.8045427137576133, "learning_rate": 1.998646092377923e-05, "loss": 0.5655, "step": 1294 }, { "epoch": 0.04613712880987584, "grad_norm": 6.219977686778161, "learning_rate": 1.9986400830593524e-05, "loss": 0.6733, "step": 1295 }, { "epoch": 0.04617275593636995, "grad_norm": 4.024311730250713, "learning_rate": 1.9986340604431908e-05, "loss": 0.5726, "step": 1296 }, { "epoch": 0.04620838306286407, "grad_norm": 10.512380550071958, "learning_rate": 1.9986280245295186e-05, "loss": 0.6333, "step": 1297 }, { "epoch": 0.04624401018935818, "grad_norm": 3.75481531123436, "learning_rate": 1.9986219753184167e-05, "loss": 0.529, "step": 1298 }, { "epoch": 0.04627963731585229, "grad_norm": 10.035110685139147, "learning_rate": 1.998615912809965e-05, "loss": 0.6655, "step": 1299 }, { "epoch": 0.046315264442346404, "grad_norm": 26.523185600265617, "learning_rate": 1.9986098370042443e-05, "loss": 0.5873, "step": 1300 }, { "epoch": 0.046350891568840515, "grad_norm": 9.394770126786796, "learning_rate": 1.9986037479013357e-05, "loss": 0.642, "step": 1301 }, { "epoch": 0.046386518695334625, "grad_norm": 3.3375557441872754, "learning_rate": 1.9985976455013203e-05, "loss": 0.7703, "step": 1302 }, { "epoch": 0.04642214582182874, "grad_norm": 5.265919345061893, "learning_rate": 1.9985915298042792e-05, "loss": 0.6292, "step": 1303 }, { "epoch": 0.04645777294832285, "grad_norm": 5.3841429931844, "learning_rate": 1.998585400810294e-05, "loss": 0.7121, "step": 1304 }, { "epoch": 0.04649340007481696, "grad_norm": 5.022797551913598, "learning_rate": 1.998579258519446e-05, "loss": 0.7043, "step": 1305 }, { "epoch": 0.04652902720131108, "grad_norm": 6.154961612836546, "learning_rate": 1.998573102931817e-05, "loss": 0.5807, "step": 1306 }, { "epoch": 0.04656465432780519, "grad_norm": 3.595577910686948, "learning_rate": 1.9985669340474894e-05, "loss": 0.5524, "step": 1307 }, { "epoch": 0.04660028145429931, "grad_norm": 5.808364476491327, "learning_rate": 1.9985607518665456e-05, "loss": 0.7731, "step": 1308 }, { "epoch": 0.04663590858079342, "grad_norm": 3.372663256669157, "learning_rate": 1.998554556389067e-05, "loss": 0.6501, "step": 1309 }, { "epoch": 0.04667153570728753, "grad_norm": 4.489885003165054, "learning_rate": 1.9985483476151362e-05, "loss": 0.5156, "step": 1310 }, { "epoch": 0.046707162833781644, "grad_norm": 4.186644151480073, "learning_rate": 1.998542125544837e-05, "loss": 0.6148, "step": 1311 }, { "epoch": 0.046742789960275755, "grad_norm": 3.63115668432235, "learning_rate": 1.998535890178251e-05, "loss": 0.7574, "step": 1312 }, { "epoch": 0.046778417086769865, "grad_norm": 3.879102604625324, "learning_rate": 1.9985296415154614e-05, "loss": 0.7266, "step": 1313 }, { "epoch": 0.04681404421326398, "grad_norm": 3.9455120837219844, "learning_rate": 1.998523379556552e-05, "loss": 0.6869, "step": 1314 }, { "epoch": 0.04684967133975809, "grad_norm": 3.0713438461839817, "learning_rate": 1.998517104301606e-05, "loss": 0.622, "step": 1315 }, { "epoch": 0.0468852984662522, "grad_norm": 2.857967354286344, "learning_rate": 1.998510815750707e-05, "loss": 0.6923, "step": 1316 }, { "epoch": 0.04692092559274632, "grad_norm": 3.4617789842808047, "learning_rate": 1.9985045139039383e-05, "loss": 0.6269, "step": 1317 }, { "epoch": 0.04695655271924043, "grad_norm": 2.7502040983376235, "learning_rate": 1.998498198761384e-05, "loss": 0.5809, "step": 1318 }, { "epoch": 0.04699217984573454, "grad_norm": 2.7738443288030568, "learning_rate": 1.9984918703231285e-05, "loss": 0.6309, "step": 1319 }, { "epoch": 0.04702780697222866, "grad_norm": 4.55389828302441, "learning_rate": 1.9984855285892555e-05, "loss": 0.609, "step": 1320 }, { "epoch": 0.04706343409872277, "grad_norm": 3.1531891329844783, "learning_rate": 1.99847917355985e-05, "loss": 0.6551, "step": 1321 }, { "epoch": 0.04709906122521688, "grad_norm": 2.8765938611415183, "learning_rate": 1.9984728052349965e-05, "loss": 0.7269, "step": 1322 }, { "epoch": 0.047134688351710995, "grad_norm": 4.522378165587878, "learning_rate": 1.9984664236147795e-05, "loss": 0.6567, "step": 1323 }, { "epoch": 0.047170315478205105, "grad_norm": 4.605319936740106, "learning_rate": 1.998460028699284e-05, "loss": 0.7153, "step": 1324 }, { "epoch": 0.047205942604699215, "grad_norm": 4.041840392203654, "learning_rate": 1.9984536204885957e-05, "loss": 0.609, "step": 1325 }, { "epoch": 0.04724156973119333, "grad_norm": 14.01623065311527, "learning_rate": 1.9984471989827996e-05, "loss": 0.7277, "step": 1326 }, { "epoch": 0.04727719685768744, "grad_norm": 3.456739683466038, "learning_rate": 1.9984407641819812e-05, "loss": 0.6623, "step": 1327 }, { "epoch": 0.04731282398418155, "grad_norm": 21.943901568944963, "learning_rate": 1.998434316086226e-05, "loss": 0.6394, "step": 1328 }, { "epoch": 0.04734845111067567, "grad_norm": 3.7631379052747373, "learning_rate": 1.9984278546956202e-05, "loss": 0.6273, "step": 1329 }, { "epoch": 0.04738407823716978, "grad_norm": 3.106951802194448, "learning_rate": 1.9984213800102494e-05, "loss": 0.6312, "step": 1330 }, { "epoch": 0.0474197053636639, "grad_norm": 3.826971544894008, "learning_rate": 1.9984148920302e-05, "loss": 0.6085, "step": 1331 }, { "epoch": 0.04745533249015801, "grad_norm": 3.436788507523537, "learning_rate": 1.9984083907555587e-05, "loss": 0.5933, "step": 1332 }, { "epoch": 0.04749095961665212, "grad_norm": 2.8119449489135713, "learning_rate": 1.998401876186412e-05, "loss": 0.6413, "step": 1333 }, { "epoch": 0.047526586743146235, "grad_norm": 3.3975800164721854, "learning_rate": 1.998395348322846e-05, "loss": 0.5592, "step": 1334 }, { "epoch": 0.047562213869640345, "grad_norm": 5.697921328261222, "learning_rate": 1.998388807164948e-05, "loss": 0.6408, "step": 1335 }, { "epoch": 0.047597840996134455, "grad_norm": 3.6591516278779443, "learning_rate": 1.9983822527128058e-05, "loss": 0.7518, "step": 1336 }, { "epoch": 0.04763346812262857, "grad_norm": 3.2003634261105454, "learning_rate": 1.9983756849665058e-05, "loss": 0.6934, "step": 1337 }, { "epoch": 0.04766909524912268, "grad_norm": 3.9974587265581434, "learning_rate": 1.9983691039261358e-05, "loss": 0.635, "step": 1338 }, { "epoch": 0.04770472237561679, "grad_norm": 4.746866406504354, "learning_rate": 1.9983625095917833e-05, "loss": 0.6323, "step": 1339 }, { "epoch": 0.04774034950211091, "grad_norm": 6.821269196181517, "learning_rate": 1.9983559019635363e-05, "loss": 0.6608, "step": 1340 }, { "epoch": 0.04777597662860502, "grad_norm": 4.842637271045115, "learning_rate": 1.9983492810414823e-05, "loss": 0.7088, "step": 1341 }, { "epoch": 0.04781160375509913, "grad_norm": 6.095756651435987, "learning_rate": 1.99834264682571e-05, "loss": 0.6616, "step": 1342 }, { "epoch": 0.04784723088159325, "grad_norm": 4.675642607233053, "learning_rate": 1.998335999316308e-05, "loss": 0.6145, "step": 1343 }, { "epoch": 0.04788285800808736, "grad_norm": 7.436898539947782, "learning_rate": 1.9983293385133636e-05, "loss": 0.6564, "step": 1344 }, { "epoch": 0.04791848513458147, "grad_norm": 6.85703278038998, "learning_rate": 1.9983226644169668e-05, "loss": 0.6941, "step": 1345 }, { "epoch": 0.047954112261075585, "grad_norm": 9.955615759059539, "learning_rate": 1.9983159770272057e-05, "loss": 0.6925, "step": 1346 }, { "epoch": 0.047989739387569695, "grad_norm": 4.699782552740191, "learning_rate": 1.9983092763441698e-05, "loss": 0.6864, "step": 1347 }, { "epoch": 0.048025366514063805, "grad_norm": 13.816632422211752, "learning_rate": 1.9983025623679483e-05, "loss": 0.6363, "step": 1348 }, { "epoch": 0.04806099364055792, "grad_norm": 7.093782981748434, "learning_rate": 1.9982958350986296e-05, "loss": 0.6207, "step": 1349 }, { "epoch": 0.04809662076705203, "grad_norm": 4.509799374829411, "learning_rate": 1.9982890945363048e-05, "loss": 0.623, "step": 1350 }, { "epoch": 0.04813224789354614, "grad_norm": 8.866350495461973, "learning_rate": 1.9982823406810627e-05, "loss": 0.72, "step": 1351 }, { "epoch": 0.04816787502004026, "grad_norm": 17.934959553257894, "learning_rate": 1.9982755735329937e-05, "loss": 0.6599, "step": 1352 }, { "epoch": 0.04820350214653437, "grad_norm": 5.874548677564341, "learning_rate": 1.9982687930921873e-05, "loss": 0.6804, "step": 1353 }, { "epoch": 0.04823912927302849, "grad_norm": 3.6648190562197804, "learning_rate": 1.9982619993587343e-05, "loss": 0.6112, "step": 1354 }, { "epoch": 0.0482747563995226, "grad_norm": 4.319572159784523, "learning_rate": 1.9982551923327255e-05, "loss": 0.6055, "step": 1355 }, { "epoch": 0.04831038352601671, "grad_norm": 6.6884786793365105, "learning_rate": 1.9982483720142506e-05, "loss": 0.6346, "step": 1356 }, { "epoch": 0.048346010652510825, "grad_norm": 4.522891729006273, "learning_rate": 1.998241538403401e-05, "loss": 0.6483, "step": 1357 }, { "epoch": 0.048381637779004935, "grad_norm": 5.8186971031041965, "learning_rate": 1.998234691500267e-05, "loss": 0.6316, "step": 1358 }, { "epoch": 0.048417264905499045, "grad_norm": 3.5303873261798837, "learning_rate": 1.998227831304941e-05, "loss": 0.5865, "step": 1359 }, { "epoch": 0.04845289203199316, "grad_norm": 5.182813752870008, "learning_rate": 1.9982209578175136e-05, "loss": 0.7016, "step": 1360 }, { "epoch": 0.04848851915848727, "grad_norm": 3.0250401024246836, "learning_rate": 1.9982140710380765e-05, "loss": 0.6643, "step": 1361 }, { "epoch": 0.04852414628498138, "grad_norm": 4.718135262943726, "learning_rate": 1.9982071709667216e-05, "loss": 0.6122, "step": 1362 }, { "epoch": 0.0485597734114755, "grad_norm": 4.2038957498444, "learning_rate": 1.9982002576035403e-05, "loss": 0.5842, "step": 1363 }, { "epoch": 0.04859540053796961, "grad_norm": 4.987962906524147, "learning_rate": 1.9981933309486245e-05, "loss": 0.5765, "step": 1364 }, { "epoch": 0.04863102766446372, "grad_norm": 2.0308610699901717, "learning_rate": 1.998186391002067e-05, "loss": 0.5923, "step": 1365 }, { "epoch": 0.04866665479095784, "grad_norm": 2.303559231839257, "learning_rate": 1.9981794377639603e-05, "loss": 0.5847, "step": 1366 }, { "epoch": 0.04870228191745195, "grad_norm": 5.981221558925014, "learning_rate": 1.9981724712343963e-05, "loss": 0.668, "step": 1367 }, { "epoch": 0.04873790904394606, "grad_norm": 6.806232941962953, "learning_rate": 1.9981654914134684e-05, "loss": 0.7091, "step": 1368 }, { "epoch": 0.048773536170440175, "grad_norm": 7.1847761090253766, "learning_rate": 1.9981584983012694e-05, "loss": 0.7063, "step": 1369 }, { "epoch": 0.048809163296934285, "grad_norm": 3.460834442837515, "learning_rate": 1.998151491897892e-05, "loss": 0.6741, "step": 1370 }, { "epoch": 0.048844790423428396, "grad_norm": 4.168688983391285, "learning_rate": 1.99814447220343e-05, "loss": 0.6951, "step": 1371 }, { "epoch": 0.04888041754992251, "grad_norm": 3.1890385561600674, "learning_rate": 1.9981374392179765e-05, "loss": 0.5938, "step": 1372 }, { "epoch": 0.04891604467641662, "grad_norm": 3.9985054621223006, "learning_rate": 1.998130392941626e-05, "loss": 0.6285, "step": 1373 }, { "epoch": 0.04895167180291073, "grad_norm": 4.579493683949071, "learning_rate": 1.9981233333744708e-05, "loss": 0.7596, "step": 1374 }, { "epoch": 0.04898729892940485, "grad_norm": 3.6545494255516995, "learning_rate": 1.998116260516606e-05, "loss": 0.5903, "step": 1375 }, { "epoch": 0.04902292605589896, "grad_norm": 4.324933816510712, "learning_rate": 1.998109174368126e-05, "loss": 0.6301, "step": 1376 }, { "epoch": 0.04905855318239307, "grad_norm": 2.597247868946158, "learning_rate": 1.9981020749291246e-05, "loss": 0.559, "step": 1377 }, { "epoch": 0.04909418030888719, "grad_norm": 5.504418292591998, "learning_rate": 1.9980949621996962e-05, "loss": 0.6243, "step": 1378 }, { "epoch": 0.0491298074353813, "grad_norm": 16.345045612821913, "learning_rate": 1.9980878361799356e-05, "loss": 0.6197, "step": 1379 }, { "epoch": 0.049165434561875415, "grad_norm": 8.514018045109776, "learning_rate": 1.998080696869938e-05, "loss": 0.6003, "step": 1380 }, { "epoch": 0.049201061688369525, "grad_norm": 6.951778097278988, "learning_rate": 1.9980735442697984e-05, "loss": 0.6841, "step": 1381 }, { "epoch": 0.049236688814863636, "grad_norm": 3.235773394281371, "learning_rate": 1.998066378379612e-05, "loss": 0.6207, "step": 1382 }, { "epoch": 0.04927231594135775, "grad_norm": 5.511991261038838, "learning_rate": 1.998059199199474e-05, "loss": 0.7238, "step": 1383 }, { "epoch": 0.04930794306785186, "grad_norm": 2.6620083662093195, "learning_rate": 1.9980520067294803e-05, "loss": 0.6982, "step": 1384 }, { "epoch": 0.04934357019434597, "grad_norm": 3.4977681108175616, "learning_rate": 1.9980448009697264e-05, "loss": 0.5262, "step": 1385 }, { "epoch": 0.04937919732084009, "grad_norm": 2.39883690117706, "learning_rate": 1.9980375819203084e-05, "loss": 0.5855, "step": 1386 }, { "epoch": 0.0494148244473342, "grad_norm": 4.836519811058751, "learning_rate": 1.9980303495813227e-05, "loss": 0.6085, "step": 1387 }, { "epoch": 0.04945045157382831, "grad_norm": 3.7588745273696325, "learning_rate": 1.998023103952865e-05, "loss": 0.634, "step": 1388 }, { "epoch": 0.04948607870032243, "grad_norm": 2.4937291288925154, "learning_rate": 1.998015845035032e-05, "loss": 0.6703, "step": 1389 }, { "epoch": 0.04952170582681654, "grad_norm": 4.193522387418361, "learning_rate": 1.9980085728279207e-05, "loss": 0.615, "step": 1390 }, { "epoch": 0.04955733295331065, "grad_norm": 8.827834621072094, "learning_rate": 1.9980012873316276e-05, "loss": 0.5775, "step": 1391 }, { "epoch": 0.049592960079804765, "grad_norm": 4.859751524542217, "learning_rate": 1.9979939885462498e-05, "loss": 0.7005, "step": 1392 }, { "epoch": 0.049628587206298876, "grad_norm": 2.768233327244985, "learning_rate": 1.9979866764718846e-05, "loss": 0.6301, "step": 1393 }, { "epoch": 0.049664214332792986, "grad_norm": 8.576206159186304, "learning_rate": 1.997979351108629e-05, "loss": 0.7263, "step": 1394 }, { "epoch": 0.0496998414592871, "grad_norm": 3.8411580736173665, "learning_rate": 1.997972012456581e-05, "loss": 0.6406, "step": 1395 }, { "epoch": 0.04973546858578121, "grad_norm": 3.411146246348724, "learning_rate": 1.997964660515838e-05, "loss": 0.6082, "step": 1396 }, { "epoch": 0.04977109571227532, "grad_norm": 3.880226785051415, "learning_rate": 1.9979572952864982e-05, "loss": 0.5891, "step": 1397 }, { "epoch": 0.04980672283876944, "grad_norm": 2.976730812182904, "learning_rate": 1.9979499167686592e-05, "loss": 0.5686, "step": 1398 }, { "epoch": 0.04984234996526355, "grad_norm": 2.5332298568551783, "learning_rate": 1.9979425249624196e-05, "loss": 0.5082, "step": 1399 }, { "epoch": 0.04987797709175766, "grad_norm": 4.256022510130291, "learning_rate": 1.997935119867878e-05, "loss": 0.7204, "step": 1400 }, { "epoch": 0.04991360421825178, "grad_norm": 4.240253090971089, "learning_rate": 1.9979277014851328e-05, "loss": 0.6987, "step": 1401 }, { "epoch": 0.04994923134474589, "grad_norm": 10.272621744350483, "learning_rate": 1.9979202698142823e-05, "loss": 0.6423, "step": 1402 }, { "epoch": 0.049984858471240005, "grad_norm": 3.3224125918819287, "learning_rate": 1.9979128248554263e-05, "loss": 0.5753, "step": 1403 }, { "epoch": 0.050020485597734116, "grad_norm": 4.26243145154525, "learning_rate": 1.9979053666086633e-05, "loss": 0.6025, "step": 1404 }, { "epoch": 0.050056112724228226, "grad_norm": 4.244842709605612, "learning_rate": 1.9978978950740927e-05, "loss": 0.6743, "step": 1405 }, { "epoch": 0.05009173985072234, "grad_norm": 5.818128868842674, "learning_rate": 1.9978904102518143e-05, "loss": 0.6732, "step": 1406 }, { "epoch": 0.05012736697721645, "grad_norm": 8.774986636268954, "learning_rate": 1.997882912141928e-05, "loss": 0.5849, "step": 1407 }, { "epoch": 0.05016299410371056, "grad_norm": 5.429190515822007, "learning_rate": 1.9978754007445328e-05, "loss": 0.7551, "step": 1408 }, { "epoch": 0.05019862123020468, "grad_norm": 2.899686441295528, "learning_rate": 1.9978678760597293e-05, "loss": 0.6482, "step": 1409 }, { "epoch": 0.05023424835669879, "grad_norm": 7.700601809232407, "learning_rate": 1.9978603380876175e-05, "loss": 0.6631, "step": 1410 }, { "epoch": 0.0502698754831929, "grad_norm": 4.1111782044262, "learning_rate": 1.997852786828298e-05, "loss": 0.6333, "step": 1411 }, { "epoch": 0.05030550260968702, "grad_norm": 3.4992776765609532, "learning_rate": 1.9978452222818712e-05, "loss": 0.6647, "step": 1412 }, { "epoch": 0.05034112973618113, "grad_norm": 2.8762012586847603, "learning_rate": 1.9978376444484374e-05, "loss": 0.6115, "step": 1413 }, { "epoch": 0.05037675686267524, "grad_norm": 6.362177054468749, "learning_rate": 1.9978300533280984e-05, "loss": 0.5551, "step": 1414 }, { "epoch": 0.050412383989169356, "grad_norm": 5.833392159478765, "learning_rate": 1.9978224489209547e-05, "loss": 0.5618, "step": 1415 }, { "epoch": 0.050448011115663466, "grad_norm": 4.94209080833789, "learning_rate": 1.9978148312271072e-05, "loss": 0.5868, "step": 1416 }, { "epoch": 0.050483638242157576, "grad_norm": 7.316607190658738, "learning_rate": 1.9978072002466585e-05, "loss": 0.6256, "step": 1417 }, { "epoch": 0.05051926536865169, "grad_norm": 6.666273477031922, "learning_rate": 1.997799555979709e-05, "loss": 0.5715, "step": 1418 }, { "epoch": 0.0505548924951458, "grad_norm": 7.345448764223213, "learning_rate": 1.997791898426361e-05, "loss": 0.656, "step": 1419 }, { "epoch": 0.050590519621639914, "grad_norm": 3.8367357902734764, "learning_rate": 1.9977842275867167e-05, "loss": 0.6195, "step": 1420 }, { "epoch": 0.05062614674813403, "grad_norm": 11.078702862508525, "learning_rate": 1.9977765434608778e-05, "loss": 0.5948, "step": 1421 }, { "epoch": 0.05066177387462814, "grad_norm": 22.18432844336349, "learning_rate": 1.997768846048947e-05, "loss": 0.6271, "step": 1422 }, { "epoch": 0.05069740100112225, "grad_norm": 38.806768066732744, "learning_rate": 1.9977611353510268e-05, "loss": 0.6003, "step": 1423 }, { "epoch": 0.05073302812761637, "grad_norm": 13.021651604517496, "learning_rate": 1.9977534113672194e-05, "loss": 0.6778, "step": 1424 }, { "epoch": 0.05076865525411048, "grad_norm": 8.51730325238328, "learning_rate": 1.9977456740976277e-05, "loss": 0.7015, "step": 1425 }, { "epoch": 0.050804282380604596, "grad_norm": 28.02218402665788, "learning_rate": 1.9977379235423553e-05, "loss": 0.7257, "step": 1426 }, { "epoch": 0.050839909507098706, "grad_norm": 5.127814099298438, "learning_rate": 1.9977301597015047e-05, "loss": 0.6826, "step": 1427 }, { "epoch": 0.050875536633592816, "grad_norm": 4.183225892875683, "learning_rate": 1.9977223825751802e-05, "loss": 0.6999, "step": 1428 }, { "epoch": 0.05091116376008693, "grad_norm": 4.439026758652179, "learning_rate": 1.9977145921634848e-05, "loss": 0.6759, "step": 1429 }, { "epoch": 0.05094679088658104, "grad_norm": 1.6908569244683394, "learning_rate": 1.997706788466522e-05, "loss": 0.7151, "step": 1430 }, { "epoch": 0.050982418013075154, "grad_norm": 4.381440730040515, "learning_rate": 1.997698971484396e-05, "loss": 0.8012, "step": 1431 }, { "epoch": 0.05101804513956927, "grad_norm": 3.148437554084942, "learning_rate": 1.9976911412172106e-05, "loss": 0.8105, "step": 1432 }, { "epoch": 0.05105367226606338, "grad_norm": 2.7193345969538405, "learning_rate": 1.9976832976650705e-05, "loss": 0.7261, "step": 1433 }, { "epoch": 0.05108929939255749, "grad_norm": 2.590630675911264, "learning_rate": 1.99767544082808e-05, "loss": 0.6904, "step": 1434 }, { "epoch": 0.05112492651905161, "grad_norm": 2.7283344245962886, "learning_rate": 1.9976675707063436e-05, "loss": 0.756, "step": 1435 }, { "epoch": 0.05116055364554572, "grad_norm": 1.988125393248746, "learning_rate": 1.9976596872999664e-05, "loss": 0.5839, "step": 1436 }, { "epoch": 0.05119618077203983, "grad_norm": 2.7858736030003386, "learning_rate": 1.9976517906090528e-05, "loss": 0.7149, "step": 1437 }, { "epoch": 0.051231807898533946, "grad_norm": 2.7111999558630906, "learning_rate": 1.9976438806337085e-05, "loss": 0.6226, "step": 1438 }, { "epoch": 0.051267435025028056, "grad_norm": 2.8540812885403635, "learning_rate": 1.9976359573740388e-05, "loss": 0.6106, "step": 1439 }, { "epoch": 0.051303062151522166, "grad_norm": 1.8238320310121117, "learning_rate": 1.9976280208301488e-05, "loss": 0.6765, "step": 1440 }, { "epoch": 0.05133868927801628, "grad_norm": 3.0725070732503568, "learning_rate": 1.997620071002144e-05, "loss": 0.747, "step": 1441 }, { "epoch": 0.051374316404510394, "grad_norm": 1.9835451512056625, "learning_rate": 1.9976121078901313e-05, "loss": 0.6769, "step": 1442 }, { "epoch": 0.051409943531004504, "grad_norm": 2.8067894579894306, "learning_rate": 1.9976041314942156e-05, "loss": 0.7126, "step": 1443 }, { "epoch": 0.05144557065749862, "grad_norm": 2.641360247117484, "learning_rate": 1.997596141814504e-05, "loss": 0.6681, "step": 1444 }, { "epoch": 0.05148119778399273, "grad_norm": 2.116191229895602, "learning_rate": 1.9975881388511022e-05, "loss": 0.7359, "step": 1445 }, { "epoch": 0.05151682491048684, "grad_norm": 2.1096518887093545, "learning_rate": 1.9975801226041173e-05, "loss": 0.6924, "step": 1446 }, { "epoch": 0.05155245203698096, "grad_norm": 1.8798607185989396, "learning_rate": 1.9975720930736557e-05, "loss": 0.6153, "step": 1447 }, { "epoch": 0.05158807916347507, "grad_norm": 2.0388806717874965, "learning_rate": 1.9975640502598243e-05, "loss": 0.6893, "step": 1448 }, { "epoch": 0.051623706289969186, "grad_norm": 3.412492390942541, "learning_rate": 1.9975559941627303e-05, "loss": 0.6499, "step": 1449 }, { "epoch": 0.051659333416463296, "grad_norm": 2.60046585640548, "learning_rate": 1.9975479247824813e-05, "loss": 0.6603, "step": 1450 }, { "epoch": 0.051694960542957406, "grad_norm": 2.0333948696822097, "learning_rate": 1.9975398421191842e-05, "loss": 0.6452, "step": 1451 }, { "epoch": 0.051730587669451523, "grad_norm": 2.6519459543483164, "learning_rate": 1.9975317461729472e-05, "loss": 0.6381, "step": 1452 }, { "epoch": 0.051766214795945634, "grad_norm": 1.9388543538579022, "learning_rate": 1.9975236369438773e-05, "loss": 0.651, "step": 1453 }, { "epoch": 0.051801841922439744, "grad_norm": 1.9712234846010463, "learning_rate": 1.9975155144320833e-05, "loss": 0.6779, "step": 1454 }, { "epoch": 0.05183746904893386, "grad_norm": 1.9511417157824247, "learning_rate": 1.9975073786376725e-05, "loss": 0.7465, "step": 1455 }, { "epoch": 0.05187309617542797, "grad_norm": 2.1531213882799, "learning_rate": 1.9974992295607542e-05, "loss": 0.6823, "step": 1456 }, { "epoch": 0.05190872330192208, "grad_norm": 1.320080627296201, "learning_rate": 1.9974910672014366e-05, "loss": 0.5418, "step": 1457 }, { "epoch": 0.0519443504284162, "grad_norm": 2.10557964373689, "learning_rate": 1.9974828915598276e-05, "loss": 0.6772, "step": 1458 }, { "epoch": 0.05197997755491031, "grad_norm": 2.0652264863604777, "learning_rate": 1.9974747026360372e-05, "loss": 0.627, "step": 1459 }, { "epoch": 0.05201560468140442, "grad_norm": 2.0472362967717377, "learning_rate": 1.997466500430174e-05, "loss": 0.6379, "step": 1460 }, { "epoch": 0.052051231807898536, "grad_norm": 1.8741791550588363, "learning_rate": 1.9974582849423468e-05, "loss": 0.7182, "step": 1461 }, { "epoch": 0.052086858934392646, "grad_norm": 1.448722513017975, "learning_rate": 1.9974500561726657e-05, "loss": 0.6241, "step": 1462 }, { "epoch": 0.05212248606088676, "grad_norm": 1.4735892948844438, "learning_rate": 1.99744181412124e-05, "loss": 0.6302, "step": 1463 }, { "epoch": 0.052158113187380874, "grad_norm": 1.8633090322757657, "learning_rate": 1.9974335587881786e-05, "loss": 0.807, "step": 1464 }, { "epoch": 0.052193740313874984, "grad_norm": 2.2453824814167795, "learning_rate": 1.9974252901735926e-05, "loss": 0.7538, "step": 1465 }, { "epoch": 0.052229367440369094, "grad_norm": 1.3772771369756642, "learning_rate": 1.9974170082775923e-05, "loss": 0.6666, "step": 1466 }, { "epoch": 0.05226499456686321, "grad_norm": 2.0969532234242023, "learning_rate": 1.9974087131002865e-05, "loss": 0.6848, "step": 1467 }, { "epoch": 0.05230062169335732, "grad_norm": 1.8623013330781488, "learning_rate": 1.9974004046417873e-05, "loss": 0.6276, "step": 1468 }, { "epoch": 0.05233624881985143, "grad_norm": 2.065589924423805, "learning_rate": 1.997392082902204e-05, "loss": 0.6258, "step": 1469 }, { "epoch": 0.05237187594634555, "grad_norm": 1.489542975334706, "learning_rate": 1.9973837478816483e-05, "loss": 0.5807, "step": 1470 }, { "epoch": 0.05240750307283966, "grad_norm": 2.16384279269531, "learning_rate": 1.9973753995802307e-05, "loss": 0.672, "step": 1471 }, { "epoch": 0.052443130199333776, "grad_norm": 1.2698751769813392, "learning_rate": 1.9973670379980626e-05, "loss": 0.5592, "step": 1472 }, { "epoch": 0.052478757325827886, "grad_norm": 1.8272579736658903, "learning_rate": 1.997358663135255e-05, "loss": 0.7011, "step": 1473 }, { "epoch": 0.052514384452322, "grad_norm": 1.4082350401719692, "learning_rate": 1.99735027499192e-05, "loss": 0.5814, "step": 1474 }, { "epoch": 0.052550011578816114, "grad_norm": 2.100034577328927, "learning_rate": 1.997341873568169e-05, "loss": 0.7217, "step": 1475 }, { "epoch": 0.052585638705310224, "grad_norm": 1.883607061512249, "learning_rate": 1.997333458864114e-05, "loss": 0.6195, "step": 1476 }, { "epoch": 0.052621265831804334, "grad_norm": 1.6036182381718245, "learning_rate": 1.9973250308798665e-05, "loss": 0.614, "step": 1477 }, { "epoch": 0.05265689295829845, "grad_norm": 1.5696388287841283, "learning_rate": 1.9973165896155395e-05, "loss": 0.6562, "step": 1478 }, { "epoch": 0.05269252008479256, "grad_norm": 2.054681937109727, "learning_rate": 1.997308135071245e-05, "loss": 0.6845, "step": 1479 }, { "epoch": 0.05272814721128667, "grad_norm": 1.8197352108815603, "learning_rate": 1.997299667247095e-05, "loss": 0.6514, "step": 1480 }, { "epoch": 0.05276377433778079, "grad_norm": 1.7516312663054814, "learning_rate": 1.9972911861432033e-05, "loss": 0.6203, "step": 1481 }, { "epoch": 0.0527994014642749, "grad_norm": 1.3176426148562044, "learning_rate": 1.9972826917596827e-05, "loss": 0.6119, "step": 1482 }, { "epoch": 0.05283502859076901, "grad_norm": 3.466430568693501, "learning_rate": 1.9972741840966455e-05, "loss": 0.642, "step": 1483 }, { "epoch": 0.052870655717263126, "grad_norm": 1.4300310465772061, "learning_rate": 1.997265663154206e-05, "loss": 0.5837, "step": 1484 }, { "epoch": 0.05290628284375724, "grad_norm": 4.440874012444962, "learning_rate": 1.9972571289324767e-05, "loss": 0.6072, "step": 1485 }, { "epoch": 0.05294190997025135, "grad_norm": 5.100567938443328, "learning_rate": 1.9972485814315715e-05, "loss": 0.5501, "step": 1486 }, { "epoch": 0.052977537096745464, "grad_norm": 36.19448137659547, "learning_rate": 1.997240020651605e-05, "loss": 0.6559, "step": 1487 }, { "epoch": 0.053013164223239574, "grad_norm": 6.9886297349619255, "learning_rate": 1.9972314465926902e-05, "loss": 0.6242, "step": 1488 }, { "epoch": 0.053048791349733684, "grad_norm": 2.6749046319445258, "learning_rate": 1.9972228592549418e-05, "loss": 0.551, "step": 1489 }, { "epoch": 0.0530844184762278, "grad_norm": 2.4234910045757148, "learning_rate": 1.997214258638474e-05, "loss": 0.6823, "step": 1490 }, { "epoch": 0.05312004560272191, "grad_norm": 2.081425586799367, "learning_rate": 1.9972056447434012e-05, "loss": 0.6058, "step": 1491 }, { "epoch": 0.05315567272921602, "grad_norm": 1.8723090907497602, "learning_rate": 1.9971970175698387e-05, "loss": 0.6173, "step": 1492 }, { "epoch": 0.05319129985571014, "grad_norm": 2.131793104398788, "learning_rate": 1.9971883771179003e-05, "loss": 0.6567, "step": 1493 }, { "epoch": 0.05322692698220425, "grad_norm": 2.0267042943598734, "learning_rate": 1.997179723387702e-05, "loss": 0.7209, "step": 1494 }, { "epoch": 0.05326255410869836, "grad_norm": 1.6082856445406093, "learning_rate": 1.9971710563793584e-05, "loss": 0.5873, "step": 1495 }, { "epoch": 0.05329818123519248, "grad_norm": 1.7763320297056044, "learning_rate": 1.9971623760929856e-05, "loss": 0.6843, "step": 1496 }, { "epoch": 0.05333380836168659, "grad_norm": 2.150022826544989, "learning_rate": 1.9971536825286984e-05, "loss": 0.6535, "step": 1497 }, { "epoch": 0.053369435488180704, "grad_norm": 1.5095588116491543, "learning_rate": 1.9971449756866135e-05, "loss": 0.6364, "step": 1498 }, { "epoch": 0.053405062614674814, "grad_norm": 1.7026949628331052, "learning_rate": 1.997136255566846e-05, "loss": 0.6573, "step": 1499 }, { "epoch": 0.053440689741168924, "grad_norm": 1.7785116743229994, "learning_rate": 1.997127522169512e-05, "loss": 0.6768, "step": 1500 }, { "epoch": 0.05347631686766304, "grad_norm": 1.9441178224628954, "learning_rate": 1.9971187754947287e-05, "loss": 0.5902, "step": 1501 }, { "epoch": 0.05351194399415715, "grad_norm": 1.936553829120513, "learning_rate": 1.9971100155426117e-05, "loss": 0.6625, "step": 1502 }, { "epoch": 0.05354757112065126, "grad_norm": 1.3973558583191736, "learning_rate": 1.9971012423132776e-05, "loss": 0.642, "step": 1503 }, { "epoch": 0.05358319824714538, "grad_norm": 1.7933469711952674, "learning_rate": 1.9970924558068438e-05, "loss": 0.6251, "step": 1504 }, { "epoch": 0.05361882537363949, "grad_norm": 1.8286916958078865, "learning_rate": 1.9970836560234266e-05, "loss": 0.591, "step": 1505 }, { "epoch": 0.0536544525001336, "grad_norm": 2.2764608752897093, "learning_rate": 1.9970748429631445e-05, "loss": 0.742, "step": 1506 }, { "epoch": 0.05369007962662772, "grad_norm": 1.4310751903141756, "learning_rate": 1.997066016626113e-05, "loss": 0.5967, "step": 1507 }, { "epoch": 0.05372570675312183, "grad_norm": 1.4378053392988546, "learning_rate": 1.9970571770124513e-05, "loss": 0.5862, "step": 1508 }, { "epoch": 0.05376133387961594, "grad_norm": 2.4919879791486546, "learning_rate": 1.997048324122276e-05, "loss": 0.7083, "step": 1509 }, { "epoch": 0.053796961006110054, "grad_norm": 1.4386098972134451, "learning_rate": 1.9970394579557054e-05, "loss": 0.5763, "step": 1510 }, { "epoch": 0.053832588132604164, "grad_norm": 2.1332877299221815, "learning_rate": 1.9970305785128576e-05, "loss": 0.6957, "step": 1511 }, { "epoch": 0.053868215259098275, "grad_norm": 1.926216878399923, "learning_rate": 1.9970216857938508e-05, "loss": 0.6235, "step": 1512 }, { "epoch": 0.05390384238559239, "grad_norm": 1.3561031039849814, "learning_rate": 1.9970127797988033e-05, "loss": 0.5731, "step": 1513 }, { "epoch": 0.0539394695120865, "grad_norm": 1.9281889221408193, "learning_rate": 1.997003860527834e-05, "loss": 0.6406, "step": 1514 }, { "epoch": 0.05397509663858061, "grad_norm": 1.4735819766028642, "learning_rate": 1.996994927981061e-05, "loss": 0.5915, "step": 1515 }, { "epoch": 0.05401072376507473, "grad_norm": 1.4259576765937327, "learning_rate": 1.996985982158604e-05, "loss": 0.6376, "step": 1516 }, { "epoch": 0.05404635089156884, "grad_norm": 1.4271402861683862, "learning_rate": 1.996977023060582e-05, "loss": 0.6194, "step": 1517 }, { "epoch": 0.05408197801806295, "grad_norm": 1.2874878251433837, "learning_rate": 1.9969680506871138e-05, "loss": 0.6175, "step": 1518 }, { "epoch": 0.05411760514455707, "grad_norm": 1.2976562933956677, "learning_rate": 1.996959065038319e-05, "loss": 0.5204, "step": 1519 }, { "epoch": 0.05415323227105118, "grad_norm": 2.135245644494447, "learning_rate": 1.9969500661143174e-05, "loss": 0.657, "step": 1520 }, { "epoch": 0.054188859397545294, "grad_norm": 1.656971526435537, "learning_rate": 1.996941053915229e-05, "loss": 0.664, "step": 1521 }, { "epoch": 0.054224486524039404, "grad_norm": 2.4036175864284717, "learning_rate": 1.9969320284411736e-05, "loss": 0.7435, "step": 1522 }, { "epoch": 0.054260113650533515, "grad_norm": 2.180117851245941, "learning_rate": 1.9969229896922715e-05, "loss": 0.5333, "step": 1523 }, { "epoch": 0.05429574077702763, "grad_norm": 2.7459086657863967, "learning_rate": 1.996913937668643e-05, "loss": 0.7271, "step": 1524 }, { "epoch": 0.05433136790352174, "grad_norm": 2.385589764064676, "learning_rate": 1.996904872370409e-05, "loss": 0.6451, "step": 1525 }, { "epoch": 0.05436699503001585, "grad_norm": 1.3536301942607158, "learning_rate": 1.9968957937976893e-05, "loss": 0.6092, "step": 1526 }, { "epoch": 0.05440262215650997, "grad_norm": 1.815350328540799, "learning_rate": 1.996886701950605e-05, "loss": 0.6111, "step": 1527 }, { "epoch": 0.05443824928300408, "grad_norm": 1.3085161375926482, "learning_rate": 1.996877596829278e-05, "loss": 0.5828, "step": 1528 }, { "epoch": 0.05447387640949819, "grad_norm": 1.8686969028786473, "learning_rate": 1.9968684784338293e-05, "loss": 0.6413, "step": 1529 }, { "epoch": 0.05450950353599231, "grad_norm": 1.415896207906122, "learning_rate": 1.9968593467643795e-05, "loss": 0.5581, "step": 1530 }, { "epoch": 0.05454513066248642, "grad_norm": 1.3544173206236731, "learning_rate": 1.996850201821051e-05, "loss": 0.5657, "step": 1531 }, { "epoch": 0.05458075778898053, "grad_norm": 1.867711795789691, "learning_rate": 1.996841043603965e-05, "loss": 0.638, "step": 1532 }, { "epoch": 0.054616384915474644, "grad_norm": 2.153693091082929, "learning_rate": 1.9968318721132442e-05, "loss": 0.6809, "step": 1533 }, { "epoch": 0.054652012041968755, "grad_norm": 5.329501025838812, "learning_rate": 1.9968226873490103e-05, "loss": 0.7073, "step": 1534 }, { "epoch": 0.054687639168462865, "grad_norm": 2.984849704036723, "learning_rate": 1.996813489311385e-05, "loss": 0.6937, "step": 1535 }, { "epoch": 0.05472326629495698, "grad_norm": 1.4702009028358125, "learning_rate": 1.9968042780004917e-05, "loss": 0.6165, "step": 1536 }, { "epoch": 0.05475889342145109, "grad_norm": 1.9688245261603567, "learning_rate": 1.996795053416453e-05, "loss": 0.6017, "step": 1537 }, { "epoch": 0.0547945205479452, "grad_norm": 1.229955267069871, "learning_rate": 1.996785815559391e-05, "loss": 0.5709, "step": 1538 }, { "epoch": 0.05483014767443932, "grad_norm": 6.785748786092756, "learning_rate": 1.9967765644294293e-05, "loss": 0.7187, "step": 1539 }, { "epoch": 0.05486577480093343, "grad_norm": 2.3162525398044163, "learning_rate": 1.9967673000266908e-05, "loss": 0.6437, "step": 1540 }, { "epoch": 0.05490140192742754, "grad_norm": 1.6905876670548434, "learning_rate": 1.9967580223512994e-05, "loss": 0.586, "step": 1541 }, { "epoch": 0.05493702905392166, "grad_norm": 1.3786422356414887, "learning_rate": 1.9967487314033778e-05, "loss": 0.6178, "step": 1542 }, { "epoch": 0.05497265618041577, "grad_norm": 1.770946514093217, "learning_rate": 1.9967394271830506e-05, "loss": 0.5651, "step": 1543 }, { "epoch": 0.055008283306909884, "grad_norm": 2.712644666871361, "learning_rate": 1.996730109690441e-05, "loss": 0.592, "step": 1544 }, { "epoch": 0.055043910433403995, "grad_norm": 1.4923709148316888, "learning_rate": 1.996720778925673e-05, "loss": 0.5835, "step": 1545 }, { "epoch": 0.055079537559898105, "grad_norm": 1.6103640986606673, "learning_rate": 1.9967114348888716e-05, "loss": 0.5613, "step": 1546 }, { "epoch": 0.05511516468639222, "grad_norm": 3.3224282002381105, "learning_rate": 1.9967020775801605e-05, "loss": 0.6842, "step": 1547 }, { "epoch": 0.05515079181288633, "grad_norm": 7.681194214130297, "learning_rate": 1.9966927069996652e-05, "loss": 0.6658, "step": 1548 }, { "epoch": 0.05518641893938044, "grad_norm": 7.826163673843606, "learning_rate": 1.996683323147509e-05, "loss": 0.6945, "step": 1549 }, { "epoch": 0.05522204606587456, "grad_norm": 1.69920612549859, "learning_rate": 1.996673926023818e-05, "loss": 0.621, "step": 1550 }, { "epoch": 0.05525767319236867, "grad_norm": 2.334659572932871, "learning_rate": 1.9966645156287173e-05, "loss": 0.6526, "step": 1551 }, { "epoch": 0.05529330031886278, "grad_norm": 1.5115885592025156, "learning_rate": 1.996655091962332e-05, "loss": 0.498, "step": 1552 }, { "epoch": 0.0553289274453569, "grad_norm": 3.1951051006195237, "learning_rate": 1.9966456550247875e-05, "loss": 0.6515, "step": 1553 }, { "epoch": 0.05536455457185101, "grad_norm": 1.4056126971881084, "learning_rate": 1.9966362048162092e-05, "loss": 0.5737, "step": 1554 }, { "epoch": 0.05540018169834512, "grad_norm": 1.504988881576791, "learning_rate": 1.9966267413367232e-05, "loss": 0.5865, "step": 1555 }, { "epoch": 0.055435808824839235, "grad_norm": 1.5906565283089684, "learning_rate": 1.9966172645864556e-05, "loss": 0.5526, "step": 1556 }, { "epoch": 0.055471435951333345, "grad_norm": 1.8391443582914595, "learning_rate": 1.9966077745655327e-05, "loss": 0.5149, "step": 1557 }, { "epoch": 0.055507063077827455, "grad_norm": 2.4986388009768965, "learning_rate": 1.996598271274081e-05, "loss": 0.6902, "step": 1558 }, { "epoch": 0.05554269020432157, "grad_norm": 2.518378446726697, "learning_rate": 1.9965887547122263e-05, "loss": 0.6072, "step": 1559 }, { "epoch": 0.05557831733081568, "grad_norm": 1.6772695613145923, "learning_rate": 1.9965792248800955e-05, "loss": 0.522, "step": 1560 }, { "epoch": 0.05561394445730979, "grad_norm": 1.4946091386578206, "learning_rate": 1.996569681777816e-05, "loss": 0.5557, "step": 1561 }, { "epoch": 0.05564957158380391, "grad_norm": 2.298699667438971, "learning_rate": 1.996560125405515e-05, "loss": 0.6767, "step": 1562 }, { "epoch": 0.05568519871029802, "grad_norm": 1.5495464940171375, "learning_rate": 1.9965505557633188e-05, "loss": 0.6158, "step": 1563 }, { "epoch": 0.05572082583679213, "grad_norm": 1.2628165165186915, "learning_rate": 1.9965409728513558e-05, "loss": 0.6098, "step": 1564 }, { "epoch": 0.05575645296328625, "grad_norm": 2.448330544709413, "learning_rate": 1.996531376669753e-05, "loss": 0.6258, "step": 1565 }, { "epoch": 0.05579208008978036, "grad_norm": 1.4657732604292173, "learning_rate": 1.9965217672186387e-05, "loss": 0.621, "step": 1566 }, { "epoch": 0.055827707216274475, "grad_norm": 1.4753806856311462, "learning_rate": 1.99651214449814e-05, "loss": 0.5563, "step": 1567 }, { "epoch": 0.055863334342768585, "grad_norm": 1.470198969704587, "learning_rate": 1.996502508508386e-05, "loss": 0.587, "step": 1568 }, { "epoch": 0.055898961469262695, "grad_norm": 1.2690393327253386, "learning_rate": 1.9964928592495046e-05, "loss": 0.5121, "step": 1569 }, { "epoch": 0.05593458859575681, "grad_norm": 1.4024686216844053, "learning_rate": 1.996483196721624e-05, "loss": 0.5069, "step": 1570 }, { "epoch": 0.05597021572225092, "grad_norm": 2.0395274206432745, "learning_rate": 1.9964735209248734e-05, "loss": 0.639, "step": 1571 }, { "epoch": 0.05600584284874503, "grad_norm": 1.3809265536601338, "learning_rate": 1.996463831859381e-05, "loss": 0.5613, "step": 1572 }, { "epoch": 0.05604146997523915, "grad_norm": 1.3312680984159466, "learning_rate": 1.9964541295252765e-05, "loss": 0.537, "step": 1573 }, { "epoch": 0.05607709710173326, "grad_norm": 2.4625392106360624, "learning_rate": 1.996444413922689e-05, "loss": 0.63, "step": 1574 }, { "epoch": 0.05611272422822737, "grad_norm": 1.2969030126940153, "learning_rate": 1.9964346850517472e-05, "loss": 0.5816, "step": 1575 }, { "epoch": 0.05614835135472149, "grad_norm": 2.147009763782209, "learning_rate": 1.9964249429125812e-05, "loss": 0.6604, "step": 1576 }, { "epoch": 0.0561839784812156, "grad_norm": 1.5209631068238025, "learning_rate": 1.9964151875053207e-05, "loss": 0.5576, "step": 1577 }, { "epoch": 0.05621960560770971, "grad_norm": 1.3376550966780827, "learning_rate": 1.9964054188300955e-05, "loss": 0.5635, "step": 1578 }, { "epoch": 0.056255232734203825, "grad_norm": 1.9593425424135862, "learning_rate": 1.996395636887036e-05, "loss": 0.6734, "step": 1579 }, { "epoch": 0.056290859860697935, "grad_norm": 2.031445237865731, "learning_rate": 1.996385841676272e-05, "loss": 0.6572, "step": 1580 }, { "epoch": 0.056326486987192045, "grad_norm": 1.6884400034728828, "learning_rate": 1.9963760331979338e-05, "loss": 0.5581, "step": 1581 }, { "epoch": 0.05636211411368616, "grad_norm": 1.9271353759899426, "learning_rate": 1.9963662114521524e-05, "loss": 0.5836, "step": 1582 }, { "epoch": 0.05639774124018027, "grad_norm": 1.3883364478052742, "learning_rate": 1.9963563764390584e-05, "loss": 0.5683, "step": 1583 }, { "epoch": 0.05643336836667438, "grad_norm": 2.8680969823192655, "learning_rate": 1.996346528158783e-05, "loss": 0.6815, "step": 1584 }, { "epoch": 0.0564689954931685, "grad_norm": 3.532625030517319, "learning_rate": 1.9963366666114573e-05, "loss": 0.5645, "step": 1585 }, { "epoch": 0.05650462261966261, "grad_norm": 1.5227286164588945, "learning_rate": 1.9963267917972125e-05, "loss": 0.5634, "step": 1586 }, { "epoch": 0.05654024974615672, "grad_norm": 1.4505608706170077, "learning_rate": 1.99631690371618e-05, "loss": 0.5891, "step": 1587 }, { "epoch": 0.05657587687265084, "grad_norm": 1.492583521055065, "learning_rate": 1.996307002368491e-05, "loss": 0.5969, "step": 1588 }, { "epoch": 0.05661150399914495, "grad_norm": 1.4230190269872065, "learning_rate": 1.9962970877542786e-05, "loss": 0.5875, "step": 1589 }, { "epoch": 0.056647131125639065, "grad_norm": 2.0978031314461028, "learning_rate": 1.996287159873674e-05, "loss": 0.6294, "step": 1590 }, { "epoch": 0.056682758252133175, "grad_norm": 1.492094831152133, "learning_rate": 1.9962772187268093e-05, "loss": 0.5841, "step": 1591 }, { "epoch": 0.056718385378627285, "grad_norm": 2.0471365160597808, "learning_rate": 1.9962672643138173e-05, "loss": 0.6014, "step": 1592 }, { "epoch": 0.0567540125051214, "grad_norm": 2.2640897882927757, "learning_rate": 1.99625729663483e-05, "loss": 0.6935, "step": 1593 }, { "epoch": 0.05678963963161551, "grad_norm": 1.722786398683497, "learning_rate": 1.9962473156899803e-05, "loss": 0.5703, "step": 1594 }, { "epoch": 0.05682526675810962, "grad_norm": 1.9223452013156923, "learning_rate": 1.9962373214794016e-05, "loss": 0.6619, "step": 1595 }, { "epoch": 0.05686089388460374, "grad_norm": 3.051623930187719, "learning_rate": 1.996227314003227e-05, "loss": 0.7155, "step": 1596 }, { "epoch": 0.05689652101109785, "grad_norm": 1.4055616520997687, "learning_rate": 1.996217293261589e-05, "loss": 0.5088, "step": 1597 }, { "epoch": 0.05693214813759196, "grad_norm": 2.5284067417500657, "learning_rate": 1.9962072592546215e-05, "loss": 0.7207, "step": 1598 }, { "epoch": 0.05696777526408608, "grad_norm": 1.679423879346203, "learning_rate": 1.9961972119824577e-05, "loss": 0.5576, "step": 1599 }, { "epoch": 0.05700340239058019, "grad_norm": 2.2985307537765087, "learning_rate": 1.996187151445232e-05, "loss": 0.6061, "step": 1600 }, { "epoch": 0.0570390295170743, "grad_norm": 1.570732463988684, "learning_rate": 1.9961770776430783e-05, "loss": 0.6011, "step": 1601 }, { "epoch": 0.057074656643568415, "grad_norm": 1.604805291172509, "learning_rate": 1.9961669905761303e-05, "loss": 0.5672, "step": 1602 }, { "epoch": 0.057110283770062525, "grad_norm": 2.0853087766885143, "learning_rate": 1.9961568902445224e-05, "loss": 0.7066, "step": 1603 }, { "epoch": 0.057145910896556636, "grad_norm": 1.9014836655699772, "learning_rate": 1.9961467766483893e-05, "loss": 0.6114, "step": 1604 }, { "epoch": 0.05718153802305075, "grad_norm": 2.35869843205239, "learning_rate": 1.9961366497878657e-05, "loss": 0.6871, "step": 1605 }, { "epoch": 0.05721716514954486, "grad_norm": 3.1106589800309545, "learning_rate": 1.9961265096630863e-05, "loss": 0.6802, "step": 1606 }, { "epoch": 0.05725279227603897, "grad_norm": 2.85029798577412, "learning_rate": 1.996116356274186e-05, "loss": 0.6181, "step": 1607 }, { "epoch": 0.05728841940253309, "grad_norm": 1.6825752431810066, "learning_rate": 1.9961061896213006e-05, "loss": 0.6045, "step": 1608 }, { "epoch": 0.0573240465290272, "grad_norm": 1.5694778858251193, "learning_rate": 1.996096009704565e-05, "loss": 0.5139, "step": 1609 }, { "epoch": 0.05735967365552131, "grad_norm": 1.684795372309042, "learning_rate": 1.9960858165241145e-05, "loss": 0.5788, "step": 1610 }, { "epoch": 0.05739530078201543, "grad_norm": 2.5588853747244436, "learning_rate": 1.996075610080085e-05, "loss": 0.6738, "step": 1611 }, { "epoch": 0.05743092790850954, "grad_norm": 1.9869696750121835, "learning_rate": 1.996065390372613e-05, "loss": 0.6021, "step": 1612 }, { "epoch": 0.057466555035003655, "grad_norm": 1.5367337453407037, "learning_rate": 1.996055157401834e-05, "loss": 0.5427, "step": 1613 }, { "epoch": 0.057502182161497765, "grad_norm": 2.4423210559416195, "learning_rate": 1.9960449111678845e-05, "loss": 0.7025, "step": 1614 }, { "epoch": 0.057537809287991876, "grad_norm": 1.6145977291814075, "learning_rate": 1.9960346516709005e-05, "loss": 0.564, "step": 1615 }, { "epoch": 0.05757343641448599, "grad_norm": 1.9907150464271124, "learning_rate": 1.9960243789110188e-05, "loss": 0.5739, "step": 1616 }, { "epoch": 0.0576090635409801, "grad_norm": 2.224272691844369, "learning_rate": 1.9960140928883765e-05, "loss": 0.5225, "step": 1617 }, { "epoch": 0.05764469066747421, "grad_norm": 1.8065290451763538, "learning_rate": 1.9960037936031105e-05, "loss": 0.6051, "step": 1618 }, { "epoch": 0.05768031779396833, "grad_norm": 1.813846207876942, "learning_rate": 1.9959934810553577e-05, "loss": 0.5475, "step": 1619 }, { "epoch": 0.05771594492046244, "grad_norm": 1.965191323025254, "learning_rate": 1.9959831552452557e-05, "loss": 0.6193, "step": 1620 }, { "epoch": 0.05775157204695655, "grad_norm": 1.8720637135637894, "learning_rate": 1.9959728161729416e-05, "loss": 0.6194, "step": 1621 }, { "epoch": 0.05778719917345067, "grad_norm": 2.935908498675685, "learning_rate": 1.9959624638385534e-05, "loss": 0.6544, "step": 1622 }, { "epoch": 0.05782282629994478, "grad_norm": 1.8621216754368084, "learning_rate": 1.9959520982422293e-05, "loss": 0.5225, "step": 1623 }, { "epoch": 0.05785845342643889, "grad_norm": 2.0709130274242535, "learning_rate": 1.995941719384106e-05, "loss": 0.6211, "step": 1624 }, { "epoch": 0.057894080552933005, "grad_norm": 2.2366974019088994, "learning_rate": 1.9959313272643232e-05, "loss": 0.5867, "step": 1625 }, { "epoch": 0.057929707679427116, "grad_norm": 3.250628706525616, "learning_rate": 1.9959209218830184e-05, "loss": 0.6965, "step": 1626 }, { "epoch": 0.057965334805921226, "grad_norm": 2.509092707487553, "learning_rate": 1.9959105032403307e-05, "loss": 0.6601, "step": 1627 }, { "epoch": 0.05800096193241534, "grad_norm": 2.6344629835350073, "learning_rate": 1.9959000713363985e-05, "loss": 0.6271, "step": 1628 }, { "epoch": 0.05803658905890945, "grad_norm": 2.2936505349657677, "learning_rate": 1.9958896261713606e-05, "loss": 0.6558, "step": 1629 }, { "epoch": 0.05807221618540356, "grad_norm": 1.877389702882658, "learning_rate": 1.9958791677453565e-05, "loss": 0.6324, "step": 1630 }, { "epoch": 0.05810784331189768, "grad_norm": 1.5535786437026722, "learning_rate": 1.9958686960585247e-05, "loss": 0.5791, "step": 1631 }, { "epoch": 0.05814347043839179, "grad_norm": 1.693718065209726, "learning_rate": 1.9958582111110055e-05, "loss": 0.6176, "step": 1632 }, { "epoch": 0.0581790975648859, "grad_norm": 2.8258355898834906, "learning_rate": 1.9958477129029382e-05, "loss": 0.7246, "step": 1633 }, { "epoch": 0.05821472469138002, "grad_norm": 2.4032861585621084, "learning_rate": 1.9958372014344624e-05, "loss": 0.6419, "step": 1634 }, { "epoch": 0.05825035181787413, "grad_norm": 1.9030710667598218, "learning_rate": 1.9958266767057183e-05, "loss": 0.5149, "step": 1635 }, { "epoch": 0.05828597894436824, "grad_norm": 4.3035363518099246, "learning_rate": 1.995816138716846e-05, "loss": 0.6695, "step": 1636 }, { "epoch": 0.058321606070862356, "grad_norm": 2.618351507662809, "learning_rate": 1.9958055874679853e-05, "loss": 0.6115, "step": 1637 }, { "epoch": 0.058357233197356466, "grad_norm": 2.7825566494244085, "learning_rate": 1.9957950229592777e-05, "loss": 0.6569, "step": 1638 }, { "epoch": 0.05839286032385058, "grad_norm": 3.0047679170489663, "learning_rate": 1.9957844451908633e-05, "loss": 0.7034, "step": 1639 }, { "epoch": 0.05842848745034469, "grad_norm": 2.227593979210369, "learning_rate": 1.995773854162883e-05, "loss": 0.5638, "step": 1640 }, { "epoch": 0.058464114576838803, "grad_norm": 2.0498028036705827, "learning_rate": 1.9957632498754776e-05, "loss": 0.5514, "step": 1641 }, { "epoch": 0.05849974170333292, "grad_norm": 3.6780709159435436, "learning_rate": 1.9957526323287886e-05, "loss": 0.5459, "step": 1642 }, { "epoch": 0.05853536882982703, "grad_norm": 2.2384768265486756, "learning_rate": 1.9957420015229572e-05, "loss": 0.6132, "step": 1643 }, { "epoch": 0.05857099595632114, "grad_norm": 2.1431801967073243, "learning_rate": 1.9957313574581258e-05, "loss": 0.6821, "step": 1644 }, { "epoch": 0.05860662308281526, "grad_norm": 1.825868478935642, "learning_rate": 1.9957207001344345e-05, "loss": 0.5814, "step": 1645 }, { "epoch": 0.05864225020930937, "grad_norm": 1.642893598238094, "learning_rate": 1.9957100295520267e-05, "loss": 0.574, "step": 1646 }, { "epoch": 0.05867787733580348, "grad_norm": 1.713409065364237, "learning_rate": 1.9956993457110437e-05, "loss": 0.5578, "step": 1647 }, { "epoch": 0.058713504462297596, "grad_norm": 2.0006431356534358, "learning_rate": 1.9956886486116278e-05, "loss": 0.7337, "step": 1648 }, { "epoch": 0.058749131588791706, "grad_norm": 1.683108516861564, "learning_rate": 1.9956779382539216e-05, "loss": 0.5661, "step": 1649 }, { "epoch": 0.058784758715285816, "grad_norm": 2.587106212337289, "learning_rate": 1.995667214638068e-05, "loss": 0.5855, "step": 1650 }, { "epoch": 0.05882038584177993, "grad_norm": 2.3658703316270393, "learning_rate": 1.9956564777642093e-05, "loss": 0.6991, "step": 1651 }, { "epoch": 0.058856012968274043, "grad_norm": 2.1767656719847905, "learning_rate": 1.995645727632489e-05, "loss": 0.57, "step": 1652 }, { "epoch": 0.058891640094768154, "grad_norm": 1.7321357532445016, "learning_rate": 1.9956349642430494e-05, "loss": 0.5618, "step": 1653 }, { "epoch": 0.05892726722126227, "grad_norm": 2.227943839440102, "learning_rate": 1.9956241875960348e-05, "loss": 0.666, "step": 1654 }, { "epoch": 0.05896289434775638, "grad_norm": 2.0528922887583496, "learning_rate": 1.995613397691588e-05, "loss": 0.5765, "step": 1655 }, { "epoch": 0.05899852147425049, "grad_norm": 2.1058291855833167, "learning_rate": 1.995602594529853e-05, "loss": 0.6279, "step": 1656 }, { "epoch": 0.05903414860074461, "grad_norm": 1.9158279803282214, "learning_rate": 1.995591778110973e-05, "loss": 0.7015, "step": 1657 }, { "epoch": 0.05906977572723872, "grad_norm": 1.7757259545263744, "learning_rate": 1.9955809484350933e-05, "loss": 0.6111, "step": 1658 }, { "epoch": 0.05910540285373283, "grad_norm": 2.373612922692952, "learning_rate": 1.9955701055023574e-05, "loss": 0.5912, "step": 1659 }, { "epoch": 0.059141029980226946, "grad_norm": 2.3186824460649813, "learning_rate": 1.995559249312909e-05, "loss": 0.6, "step": 1660 }, { "epoch": 0.059176657106721056, "grad_norm": 2.1406973056869307, "learning_rate": 1.995548379866894e-05, "loss": 0.5482, "step": 1661 }, { "epoch": 0.05921228423321517, "grad_norm": 2.2350400402255137, "learning_rate": 1.9955374971644566e-05, "loss": 0.6807, "step": 1662 }, { "epoch": 0.059247911359709284, "grad_norm": 1.6492584493097073, "learning_rate": 1.995526601205741e-05, "loss": 0.5921, "step": 1663 }, { "epoch": 0.059283538486203394, "grad_norm": 2.8295199023993782, "learning_rate": 1.9955156919908932e-05, "loss": 0.7568, "step": 1664 }, { "epoch": 0.05931916561269751, "grad_norm": 1.6196952567687886, "learning_rate": 1.9955047695200584e-05, "loss": 0.5146, "step": 1665 }, { "epoch": 0.05935479273919162, "grad_norm": 1.8755535692015988, "learning_rate": 1.9954938337933815e-05, "loss": 0.583, "step": 1666 }, { "epoch": 0.05939041986568573, "grad_norm": 1.9270474670413507, "learning_rate": 1.9954828848110082e-05, "loss": 0.6508, "step": 1667 }, { "epoch": 0.05942604699217985, "grad_norm": 3.4450849751979917, "learning_rate": 1.9954719225730847e-05, "loss": 0.6343, "step": 1668 }, { "epoch": 0.05946167411867396, "grad_norm": 2.3255675334181394, "learning_rate": 1.9954609470797566e-05, "loss": 0.579, "step": 1669 }, { "epoch": 0.05949730124516807, "grad_norm": 2.351114627345311, "learning_rate": 1.9954499583311703e-05, "loss": 0.6601, "step": 1670 }, { "epoch": 0.059532928371662186, "grad_norm": 1.9471041669764226, "learning_rate": 1.9954389563274724e-05, "loss": 0.5603, "step": 1671 }, { "epoch": 0.059568555498156296, "grad_norm": 1.8218979685894738, "learning_rate": 1.9954279410688087e-05, "loss": 0.6041, "step": 1672 }, { "epoch": 0.059604182624650406, "grad_norm": 1.925393220002548, "learning_rate": 1.9954169125553262e-05, "loss": 0.6707, "step": 1673 }, { "epoch": 0.059639809751144524, "grad_norm": 2.4715173832263186, "learning_rate": 1.9954058707871717e-05, "loss": 0.5294, "step": 1674 }, { "epoch": 0.059675436877638634, "grad_norm": 4.680782767333273, "learning_rate": 1.9953948157644924e-05, "loss": 0.5535, "step": 1675 }, { "epoch": 0.059711064004132744, "grad_norm": 3.661267542668548, "learning_rate": 1.995383747487435e-05, "loss": 0.6182, "step": 1676 }, { "epoch": 0.05974669113062686, "grad_norm": 1.7769375874673128, "learning_rate": 1.995372665956148e-05, "loss": 0.5523, "step": 1677 }, { "epoch": 0.05978231825712097, "grad_norm": 1.9925371179624696, "learning_rate": 1.9953615711707775e-05, "loss": 0.5297, "step": 1678 }, { "epoch": 0.05981794538361508, "grad_norm": 3.377831097120912, "learning_rate": 1.9953504631314722e-05, "loss": 0.6104, "step": 1679 }, { "epoch": 0.0598535725101092, "grad_norm": 2.0325807486519873, "learning_rate": 1.9953393418383798e-05, "loss": 0.5705, "step": 1680 }, { "epoch": 0.05988919963660331, "grad_norm": 3.141144435867888, "learning_rate": 1.9953282072916487e-05, "loss": 0.6686, "step": 1681 }, { "epoch": 0.05992482676309742, "grad_norm": 2.4025322275324648, "learning_rate": 1.9953170594914267e-05, "loss": 0.6163, "step": 1682 }, { "epoch": 0.059960453889591536, "grad_norm": 1.9881229126062643, "learning_rate": 1.995305898437862e-05, "loss": 0.5459, "step": 1683 }, { "epoch": 0.059996081016085646, "grad_norm": 2.328671798415382, "learning_rate": 1.995294724131104e-05, "loss": 0.5664, "step": 1684 }, { "epoch": 0.060031708142579764, "grad_norm": 2.149936544375316, "learning_rate": 1.9952835365713006e-05, "loss": 0.5555, "step": 1685 }, { "epoch": 0.060067335269073874, "grad_norm": 2.2375846132063, "learning_rate": 1.9952723357586013e-05, "loss": 0.7102, "step": 1686 }, { "epoch": 0.060102962395567984, "grad_norm": 3.373112863573489, "learning_rate": 1.9952611216931558e-05, "loss": 0.7452, "step": 1687 }, { "epoch": 0.0601385895220621, "grad_norm": 3.1023742813161044, "learning_rate": 1.9952498943751123e-05, "loss": 0.4872, "step": 1688 }, { "epoch": 0.06017421664855621, "grad_norm": 3.092345657423049, "learning_rate": 1.9952386538046208e-05, "loss": 0.5851, "step": 1689 }, { "epoch": 0.06020984377505032, "grad_norm": 3.1160795543021957, "learning_rate": 1.9952273999818312e-05, "loss": 0.599, "step": 1690 }, { "epoch": 0.06024547090154444, "grad_norm": 3.863163566885546, "learning_rate": 1.9952161329068934e-05, "loss": 0.6683, "step": 1691 }, { "epoch": 0.06028109802803855, "grad_norm": 2.5380426877899738, "learning_rate": 1.9952048525799567e-05, "loss": 0.5969, "step": 1692 }, { "epoch": 0.06031672515453266, "grad_norm": 2.005198628724799, "learning_rate": 1.995193559001172e-05, "loss": 0.5737, "step": 1693 }, { "epoch": 0.060352352281026776, "grad_norm": 2.730690295206345, "learning_rate": 1.9951822521706896e-05, "loss": 0.5561, "step": 1694 }, { "epoch": 0.060387979407520886, "grad_norm": 4.886974047296996, "learning_rate": 1.9951709320886596e-05, "loss": 0.5582, "step": 1695 }, { "epoch": 0.060423606534015, "grad_norm": 2.8687869109951385, "learning_rate": 1.9951595987552332e-05, "loss": 0.5768, "step": 1696 }, { "epoch": 0.060459233660509114, "grad_norm": 3.7686143596928554, "learning_rate": 1.9951482521705608e-05, "loss": 0.5486, "step": 1697 }, { "epoch": 0.060494860787003224, "grad_norm": 9.15574708310544, "learning_rate": 1.9951368923347945e-05, "loss": 0.6465, "step": 1698 }, { "epoch": 0.060530487913497334, "grad_norm": 3.0883601179189366, "learning_rate": 1.9951255192480844e-05, "loss": 0.6098, "step": 1699 }, { "epoch": 0.06056611503999145, "grad_norm": 4.1115631042677485, "learning_rate": 1.995114132910583e-05, "loss": 0.5785, "step": 1700 }, { "epoch": 0.06060174216648556, "grad_norm": 2.370293472589169, "learning_rate": 1.995102733322441e-05, "loss": 0.5533, "step": 1701 }, { "epoch": 0.06063736929297967, "grad_norm": 5.205423095327722, "learning_rate": 1.9950913204838108e-05, "loss": 0.6321, "step": 1702 }, { "epoch": 0.06067299641947379, "grad_norm": 5.047974011299402, "learning_rate": 1.9950798943948438e-05, "loss": 0.6731, "step": 1703 }, { "epoch": 0.0607086235459679, "grad_norm": 2.5480175475248976, "learning_rate": 1.9950684550556926e-05, "loss": 0.5844, "step": 1704 }, { "epoch": 0.06074425067246201, "grad_norm": 3.606812006298628, "learning_rate": 1.9950570024665094e-05, "loss": 0.6418, "step": 1705 }, { "epoch": 0.060779877798956126, "grad_norm": 3.8142969967305564, "learning_rate": 1.995045536627447e-05, "loss": 0.6759, "step": 1706 }, { "epoch": 0.06081550492545024, "grad_norm": 3.1184204701019627, "learning_rate": 1.9950340575386575e-05, "loss": 0.5807, "step": 1707 }, { "epoch": 0.060851132051944354, "grad_norm": 2.6936680534102364, "learning_rate": 1.9950225652002943e-05, "loss": 0.5739, "step": 1708 }, { "epoch": 0.060886759178438464, "grad_norm": 3.4633526293072245, "learning_rate": 1.99501105961251e-05, "loss": 0.7662, "step": 1709 }, { "epoch": 0.060922386304932574, "grad_norm": 3.993378585046687, "learning_rate": 1.994999540775458e-05, "loss": 0.7424, "step": 1710 }, { "epoch": 0.06095801343142669, "grad_norm": 5.2149203049510415, "learning_rate": 1.9949880086892914e-05, "loss": 0.5701, "step": 1711 }, { "epoch": 0.0609936405579208, "grad_norm": 4.506244620244628, "learning_rate": 1.9949764633541645e-05, "loss": 0.7205, "step": 1712 }, { "epoch": 0.06102926768441491, "grad_norm": 3.4722417205127214, "learning_rate": 1.99496490477023e-05, "loss": 0.6433, "step": 1713 }, { "epoch": 0.06106489481090903, "grad_norm": 6.988541472488657, "learning_rate": 1.994953332937643e-05, "loss": 0.6422, "step": 1714 }, { "epoch": 0.06110052193740314, "grad_norm": 3.882817322689962, "learning_rate": 1.9949417478565564e-05, "loss": 0.6079, "step": 1715 }, { "epoch": 0.06113614906389725, "grad_norm": 2.7487078332311454, "learning_rate": 1.994930149527125e-05, "loss": 0.6526, "step": 1716 }, { "epoch": 0.061171776190391366, "grad_norm": 3.6140002553044215, "learning_rate": 1.9949185379495034e-05, "loss": 0.6787, "step": 1717 }, { "epoch": 0.06120740331688548, "grad_norm": 2.628958235598255, "learning_rate": 1.994906913123846e-05, "loss": 0.7225, "step": 1718 }, { "epoch": 0.06124303044337959, "grad_norm": 3.3006098810527864, "learning_rate": 1.9948952750503077e-05, "loss": 0.6534, "step": 1719 }, { "epoch": 0.061278657569873704, "grad_norm": 2.874257713645124, "learning_rate": 1.9948836237290435e-05, "loss": 0.5721, "step": 1720 }, { "epoch": 0.061314284696367814, "grad_norm": 1.7380541072591866, "learning_rate": 1.9948719591602084e-05, "loss": 0.5757, "step": 1721 }, { "epoch": 0.061349911822861924, "grad_norm": 2.4093090371005386, "learning_rate": 1.9948602813439576e-05, "loss": 0.693, "step": 1722 }, { "epoch": 0.06138553894935604, "grad_norm": 3.2020286043227406, "learning_rate": 1.9948485902804472e-05, "loss": 0.6524, "step": 1723 }, { "epoch": 0.06142116607585015, "grad_norm": 1.873187934940613, "learning_rate": 1.9948368859698322e-05, "loss": 0.6236, "step": 1724 }, { "epoch": 0.06145679320234426, "grad_norm": 2.226560423242938, "learning_rate": 1.9948251684122687e-05, "loss": 0.576, "step": 1725 }, { "epoch": 0.06149242032883838, "grad_norm": 3.8479552256455123, "learning_rate": 1.9948134376079128e-05, "loss": 0.6927, "step": 1726 }, { "epoch": 0.06152804745533249, "grad_norm": 2.450483644032608, "learning_rate": 1.9948016935569208e-05, "loss": 0.6552, "step": 1727 }, { "epoch": 0.0615636745818266, "grad_norm": 1.396932764315989, "learning_rate": 1.9947899362594486e-05, "loss": 0.5826, "step": 1728 }, { "epoch": 0.06159930170832072, "grad_norm": 2.3226258982257852, "learning_rate": 1.9947781657156533e-05, "loss": 0.5704, "step": 1729 }, { "epoch": 0.06163492883481483, "grad_norm": 2.4233032114505852, "learning_rate": 1.9947663819256915e-05, "loss": 0.6914, "step": 1730 }, { "epoch": 0.061670555961308944, "grad_norm": 2.1816987242526347, "learning_rate": 1.9947545848897197e-05, "loss": 0.6566, "step": 1731 }, { "epoch": 0.061706183087803054, "grad_norm": 1.5392069111490596, "learning_rate": 1.9947427746078955e-05, "loss": 0.5847, "step": 1732 }, { "epoch": 0.061741810214297164, "grad_norm": 2.1740178063888647, "learning_rate": 1.994730951080376e-05, "loss": 0.6609, "step": 1733 }, { "epoch": 0.06177743734079128, "grad_norm": 5.3470201897817615, "learning_rate": 1.9947191143073185e-05, "loss": 0.6772, "step": 1734 }, { "epoch": 0.06181306446728539, "grad_norm": 1.5775802012463571, "learning_rate": 1.994707264288881e-05, "loss": 0.536, "step": 1735 }, { "epoch": 0.0618486915937795, "grad_norm": 1.8052212657967333, "learning_rate": 1.9946954010252207e-05, "loss": 0.6474, "step": 1736 }, { "epoch": 0.06188431872027362, "grad_norm": 1.805599908705596, "learning_rate": 1.9946835245164958e-05, "loss": 0.5989, "step": 1737 }, { "epoch": 0.06191994584676773, "grad_norm": 2.3455635748145203, "learning_rate": 1.994671634762865e-05, "loss": 0.6767, "step": 1738 }, { "epoch": 0.06195557297326184, "grad_norm": 1.4259117053090706, "learning_rate": 1.994659731764486e-05, "loss": 0.5718, "step": 1739 }, { "epoch": 0.06199120009975596, "grad_norm": 2.9368084713928484, "learning_rate": 1.9946478155215172e-05, "loss": 0.6395, "step": 1740 }, { "epoch": 0.06202682722625007, "grad_norm": 1.5005686192160528, "learning_rate": 1.9946358860341174e-05, "loss": 0.5407, "step": 1741 }, { "epoch": 0.06206245435274418, "grad_norm": 2.4405472250276046, "learning_rate": 1.994623943302446e-05, "loss": 0.6407, "step": 1742 }, { "epoch": 0.062098081479238294, "grad_norm": 2.275455804652201, "learning_rate": 1.9946119873266615e-05, "loss": 0.6652, "step": 1743 }, { "epoch": 0.062133708605732405, "grad_norm": 1.3146265768744108, "learning_rate": 1.9946000181069232e-05, "loss": 0.5324, "step": 1744 }, { "epoch": 0.062169335732226515, "grad_norm": 1.4530678822260297, "learning_rate": 1.9945880356433904e-05, "loss": 0.5852, "step": 1745 }, { "epoch": 0.06220496285872063, "grad_norm": 2.180951110295418, "learning_rate": 1.9945760399362227e-05, "loss": 0.6424, "step": 1746 }, { "epoch": 0.06224058998521474, "grad_norm": 1.732237486595813, "learning_rate": 1.99456403098558e-05, "loss": 0.5627, "step": 1747 }, { "epoch": 0.06227621711170885, "grad_norm": 2.0591654016115934, "learning_rate": 1.9945520087916218e-05, "loss": 0.7542, "step": 1748 }, { "epoch": 0.06231184423820297, "grad_norm": 1.4419694278603572, "learning_rate": 1.9945399733545085e-05, "loss": 0.6015, "step": 1749 }, { "epoch": 0.06234747136469708, "grad_norm": 2.0054271882358554, "learning_rate": 1.9945279246744004e-05, "loss": 0.711, "step": 1750 }, { "epoch": 0.06238309849119119, "grad_norm": 1.5182280795951282, "learning_rate": 1.994515862751458e-05, "loss": 0.6042, "step": 1751 }, { "epoch": 0.06241872561768531, "grad_norm": 1.732464319048884, "learning_rate": 1.9945037875858414e-05, "loss": 0.6761, "step": 1752 }, { "epoch": 0.06245435274417942, "grad_norm": 1.270136339569818, "learning_rate": 1.9944916991777122e-05, "loss": 0.5243, "step": 1753 }, { "epoch": 0.06248997987067353, "grad_norm": 2.1922090822566016, "learning_rate": 1.9944795975272308e-05, "loss": 0.6604, "step": 1754 }, { "epoch": 0.06252560699716764, "grad_norm": 1.4177327939256676, "learning_rate": 1.994467482634558e-05, "loss": 0.6109, "step": 1755 }, { "epoch": 0.06256123412366175, "grad_norm": 1.2902251738282842, "learning_rate": 1.9944553544998563e-05, "loss": 0.5655, "step": 1756 }, { "epoch": 0.06259686125015587, "grad_norm": 1.8849418419430932, "learning_rate": 1.9944432131232857e-05, "loss": 0.6112, "step": 1757 }, { "epoch": 0.06263248837664998, "grad_norm": 1.2474174529059279, "learning_rate": 1.9944310585050094e-05, "loss": 0.5851, "step": 1758 }, { "epoch": 0.06266811550314409, "grad_norm": 1.4574928235312719, "learning_rate": 1.994418890645188e-05, "loss": 0.5653, "step": 1759 }, { "epoch": 0.06270374262963821, "grad_norm": 1.3086533253054282, "learning_rate": 1.9944067095439845e-05, "loss": 0.5345, "step": 1760 }, { "epoch": 0.06273936975613231, "grad_norm": 1.3848846575742706, "learning_rate": 1.99439451520156e-05, "loss": 0.5318, "step": 1761 }, { "epoch": 0.06277499688262643, "grad_norm": 1.3582107371969843, "learning_rate": 1.9943823076180778e-05, "loss": 0.5509, "step": 1762 }, { "epoch": 0.06281062400912055, "grad_norm": 1.2517518312218228, "learning_rate": 1.9943700867937e-05, "loss": 0.5264, "step": 1763 }, { "epoch": 0.06284625113561465, "grad_norm": 1.473493116106608, "learning_rate": 1.9943578527285897e-05, "loss": 0.5986, "step": 1764 }, { "epoch": 0.06288187826210877, "grad_norm": 1.3671674471893394, "learning_rate": 1.994345605422909e-05, "loss": 0.5242, "step": 1765 }, { "epoch": 0.06291750538860288, "grad_norm": 1.3453163960115517, "learning_rate": 1.9943333448768226e-05, "loss": 0.5754, "step": 1766 }, { "epoch": 0.062953132515097, "grad_norm": 2.229619155106596, "learning_rate": 1.994321071090492e-05, "loss": 0.7169, "step": 1767 }, { "epoch": 0.0629887596415911, "grad_norm": 2.039866806857465, "learning_rate": 1.9943087840640813e-05, "loss": 0.6287, "step": 1768 }, { "epoch": 0.06302438676808522, "grad_norm": 1.3470972699952872, "learning_rate": 1.9942964837977542e-05, "loss": 0.5272, "step": 1769 }, { "epoch": 0.06306001389457934, "grad_norm": 1.909414323768722, "learning_rate": 1.9942841702916746e-05, "loss": 0.6508, "step": 1770 }, { "epoch": 0.06309564102107344, "grad_norm": 2.546092099055475, "learning_rate": 1.9942718435460066e-05, "loss": 0.7246, "step": 1771 }, { "epoch": 0.06313126814756756, "grad_norm": 1.3501398890452603, "learning_rate": 1.9942595035609137e-05, "loss": 0.578, "step": 1772 }, { "epoch": 0.06316689527406168, "grad_norm": 1.4706594771033976, "learning_rate": 1.9942471503365604e-05, "loss": 0.6078, "step": 1773 }, { "epoch": 0.06320252240055578, "grad_norm": 1.3770711877571156, "learning_rate": 1.9942347838731115e-05, "loss": 0.5664, "step": 1774 }, { "epoch": 0.0632381495270499, "grad_norm": 1.3204481666040748, "learning_rate": 1.9942224041707315e-05, "loss": 0.5888, "step": 1775 }, { "epoch": 0.06327377665354401, "grad_norm": 1.7991671787909826, "learning_rate": 1.9942100112295857e-05, "loss": 0.6051, "step": 1776 }, { "epoch": 0.06330940378003812, "grad_norm": 1.3137470646104543, "learning_rate": 1.994197605049838e-05, "loss": 0.6468, "step": 1777 }, { "epoch": 0.06334503090653223, "grad_norm": 1.296400689020026, "learning_rate": 1.994185185631655e-05, "loss": 0.5395, "step": 1778 }, { "epoch": 0.06338065803302635, "grad_norm": 1.6629429090826702, "learning_rate": 1.994172752975201e-05, "loss": 0.6033, "step": 1779 }, { "epoch": 0.06341628515952046, "grad_norm": 1.305988770512122, "learning_rate": 1.9941603070806415e-05, "loss": 0.5558, "step": 1780 }, { "epoch": 0.06345191228601457, "grad_norm": 2.232132183156297, "learning_rate": 1.9941478479481433e-05, "loss": 0.6381, "step": 1781 }, { "epoch": 0.06348753941250869, "grad_norm": 1.9370878617240481, "learning_rate": 1.9941353755778713e-05, "loss": 0.6981, "step": 1782 }, { "epoch": 0.06352316653900279, "grad_norm": 2.7980755327025495, "learning_rate": 1.994122889969992e-05, "loss": 0.6988, "step": 1783 }, { "epoch": 0.06355879366549691, "grad_norm": 1.4435653061060982, "learning_rate": 1.9941103911246714e-05, "loss": 0.5959, "step": 1784 }, { "epoch": 0.06359442079199103, "grad_norm": 1.3979046347014432, "learning_rate": 1.9940978790420762e-05, "loss": 0.5144, "step": 1785 }, { "epoch": 0.06363004791848513, "grad_norm": 1.9912143229889028, "learning_rate": 1.994085353722373e-05, "loss": 0.6709, "step": 1786 }, { "epoch": 0.06366567504497925, "grad_norm": 1.2398684072656663, "learning_rate": 1.9940728151657285e-05, "loss": 0.6226, "step": 1787 }, { "epoch": 0.06370130217147336, "grad_norm": 1.334774169040027, "learning_rate": 1.9940602633723097e-05, "loss": 0.5082, "step": 1788 }, { "epoch": 0.06373692929796747, "grad_norm": 1.4180616761002265, "learning_rate": 1.9940476983422834e-05, "loss": 0.5509, "step": 1789 }, { "epoch": 0.06377255642446159, "grad_norm": 2.451923609021124, "learning_rate": 1.9940351200758175e-05, "loss": 0.6496, "step": 1790 }, { "epoch": 0.0638081835509557, "grad_norm": 1.251951163179395, "learning_rate": 1.994022528573079e-05, "loss": 0.5381, "step": 1791 }, { "epoch": 0.0638438106774498, "grad_norm": 1.4568244054834214, "learning_rate": 1.994009923834236e-05, "loss": 0.5589, "step": 1792 }, { "epoch": 0.06387943780394392, "grad_norm": 1.2938341432472942, "learning_rate": 1.993997305859456e-05, "loss": 0.5334, "step": 1793 }, { "epoch": 0.06391506493043804, "grad_norm": 1.8176378521187055, "learning_rate": 1.993984674648907e-05, "loss": 0.6375, "step": 1794 }, { "epoch": 0.06395069205693214, "grad_norm": 1.3000192778516446, "learning_rate": 1.993972030202757e-05, "loss": 0.5136, "step": 1795 }, { "epoch": 0.06398631918342626, "grad_norm": 1.4426183454043757, "learning_rate": 1.9939593725211753e-05, "loss": 0.6537, "step": 1796 }, { "epoch": 0.06402194630992038, "grad_norm": 1.8084837157209324, "learning_rate": 1.9939467016043293e-05, "loss": 0.6127, "step": 1797 }, { "epoch": 0.06405757343641448, "grad_norm": 1.9440339314213306, "learning_rate": 1.9939340174523885e-05, "loss": 0.667, "step": 1798 }, { "epoch": 0.0640932005629086, "grad_norm": 1.8285508271323654, "learning_rate": 1.9939213200655216e-05, "loss": 0.6477, "step": 1799 }, { "epoch": 0.06412882768940271, "grad_norm": 1.3079309985914198, "learning_rate": 1.9939086094438975e-05, "loss": 0.6124, "step": 1800 }, { "epoch": 0.06416445481589682, "grad_norm": 1.2528912586971752, "learning_rate": 1.9938958855876854e-05, "loss": 0.572, "step": 1801 }, { "epoch": 0.06420008194239094, "grad_norm": 1.9172133995377385, "learning_rate": 1.993883148497055e-05, "loss": 0.6492, "step": 1802 }, { "epoch": 0.06423570906888505, "grad_norm": 2.488470297801146, "learning_rate": 1.993870398172176e-05, "loss": 0.6337, "step": 1803 }, { "epoch": 0.06427133619537916, "grad_norm": 1.635595797195819, "learning_rate": 1.993857634613218e-05, "loss": 0.5549, "step": 1804 }, { "epoch": 0.06430696332187327, "grad_norm": 1.9793027851986238, "learning_rate": 1.9938448578203508e-05, "loss": 0.5483, "step": 1805 }, { "epoch": 0.06434259044836739, "grad_norm": 1.504431241806549, "learning_rate": 1.9938320677937442e-05, "loss": 0.5425, "step": 1806 }, { "epoch": 0.0643782175748615, "grad_norm": 2.1168510021944487, "learning_rate": 1.9938192645335696e-05, "loss": 0.6712, "step": 1807 }, { "epoch": 0.06441384470135561, "grad_norm": 1.908898578157628, "learning_rate": 1.9938064480399967e-05, "loss": 0.7015, "step": 1808 }, { "epoch": 0.06444947182784973, "grad_norm": 1.91795194934055, "learning_rate": 1.993793618313196e-05, "loss": 0.6608, "step": 1809 }, { "epoch": 0.06448509895434383, "grad_norm": 1.6774410883110233, "learning_rate": 1.993780775353339e-05, "loss": 0.6535, "step": 1810 }, { "epoch": 0.06452072608083795, "grad_norm": 1.3507253865463147, "learning_rate": 1.9937679191605964e-05, "loss": 0.5664, "step": 1811 }, { "epoch": 0.06455635320733207, "grad_norm": 1.2696500250229161, "learning_rate": 1.9937550497351394e-05, "loss": 0.5135, "step": 1812 }, { "epoch": 0.06459198033382617, "grad_norm": 1.3198450221387543, "learning_rate": 1.9937421670771392e-05, "loss": 0.531, "step": 1813 }, { "epoch": 0.06462760746032029, "grad_norm": 1.4835775280625538, "learning_rate": 1.9937292711867676e-05, "loss": 0.5559, "step": 1814 }, { "epoch": 0.0646632345868144, "grad_norm": 1.6206640063742241, "learning_rate": 1.9937163620641957e-05, "loss": 0.5616, "step": 1815 }, { "epoch": 0.06469886171330852, "grad_norm": 1.189124225823801, "learning_rate": 1.9937034397095963e-05, "loss": 0.4602, "step": 1816 }, { "epoch": 0.06473448883980262, "grad_norm": 2.2096006659579155, "learning_rate": 1.9936905041231408e-05, "loss": 0.7232, "step": 1817 }, { "epoch": 0.06477011596629674, "grad_norm": 1.3289105040440294, "learning_rate": 1.9936775553050017e-05, "loss": 0.562, "step": 1818 }, { "epoch": 0.06480574309279086, "grad_norm": 2.3874604893430402, "learning_rate": 1.9936645932553515e-05, "loss": 0.7306, "step": 1819 }, { "epoch": 0.06484137021928496, "grad_norm": 2.0402665762662444, "learning_rate": 1.993651617974363e-05, "loss": 0.6593, "step": 1820 }, { "epoch": 0.06487699734577908, "grad_norm": 1.9170708685684361, "learning_rate": 1.9936386294622084e-05, "loss": 0.6416, "step": 1821 }, { "epoch": 0.0649126244722732, "grad_norm": 1.3825737099176227, "learning_rate": 1.9936256277190608e-05, "loss": 0.535, "step": 1822 }, { "epoch": 0.0649482515987673, "grad_norm": 2.1146677951419193, "learning_rate": 1.9936126127450935e-05, "loss": 0.6618, "step": 1823 }, { "epoch": 0.06498387872526142, "grad_norm": 1.9009461138974417, "learning_rate": 1.99359958454048e-05, "loss": 0.673, "step": 1824 }, { "epoch": 0.06501950585175553, "grad_norm": 1.61144477753537, "learning_rate": 1.9935865431053936e-05, "loss": 0.5985, "step": 1825 }, { "epoch": 0.06505513297824964, "grad_norm": 1.9115234287483447, "learning_rate": 1.9935734884400076e-05, "loss": 0.5815, "step": 1826 }, { "epoch": 0.06509076010474375, "grad_norm": 1.4106950907206977, "learning_rate": 1.9935604205444962e-05, "loss": 0.5875, "step": 1827 }, { "epoch": 0.06512638723123787, "grad_norm": 1.9185541004910371, "learning_rate": 1.9935473394190335e-05, "loss": 0.6596, "step": 1828 }, { "epoch": 0.06516201435773197, "grad_norm": 1.4550674933495338, "learning_rate": 1.9935342450637934e-05, "loss": 0.5973, "step": 1829 }, { "epoch": 0.06519764148422609, "grad_norm": 2.1900054108792544, "learning_rate": 1.9935211374789506e-05, "loss": 0.6195, "step": 1830 }, { "epoch": 0.06523326861072021, "grad_norm": 1.8883770254593586, "learning_rate": 1.993508016664679e-05, "loss": 0.5669, "step": 1831 }, { "epoch": 0.06526889573721431, "grad_norm": 1.8010032396935631, "learning_rate": 1.993494882621154e-05, "loss": 0.5956, "step": 1832 }, { "epoch": 0.06530452286370843, "grad_norm": 2.4563020340343744, "learning_rate": 1.99348173534855e-05, "loss": 0.664, "step": 1833 }, { "epoch": 0.06534014999020255, "grad_norm": 1.4752010446183483, "learning_rate": 1.993468574847043e-05, "loss": 0.6159, "step": 1834 }, { "epoch": 0.06537577711669665, "grad_norm": 1.4455628264316076, "learning_rate": 1.993455401116807e-05, "loss": 0.5539, "step": 1835 }, { "epoch": 0.06541140424319077, "grad_norm": 2.1469207676127544, "learning_rate": 1.993442214158018e-05, "loss": 0.6911, "step": 1836 }, { "epoch": 0.06544703136968488, "grad_norm": 1.4224580457640275, "learning_rate": 1.9934290139708516e-05, "loss": 0.5854, "step": 1837 }, { "epoch": 0.06548265849617899, "grad_norm": 1.7045183615349515, "learning_rate": 1.9934158005554833e-05, "loss": 0.7062, "step": 1838 }, { "epoch": 0.0655182856226731, "grad_norm": 1.4060479101742374, "learning_rate": 1.9934025739120893e-05, "loss": 0.5443, "step": 1839 }, { "epoch": 0.06555391274916722, "grad_norm": 1.313730062975728, "learning_rate": 1.993389334040846e-05, "loss": 0.5805, "step": 1840 }, { "epoch": 0.06558953987566132, "grad_norm": 2.164911728604433, "learning_rate": 1.993376080941929e-05, "loss": 0.6882, "step": 1841 }, { "epoch": 0.06562516700215544, "grad_norm": 2.035668091332522, "learning_rate": 1.9933628146155154e-05, "loss": 0.6451, "step": 1842 }, { "epoch": 0.06566079412864956, "grad_norm": 1.3696896480004426, "learning_rate": 1.9933495350617814e-05, "loss": 0.537, "step": 1843 }, { "epoch": 0.06569642125514366, "grad_norm": 1.76886330791315, "learning_rate": 1.9933362422809043e-05, "loss": 0.5848, "step": 1844 }, { "epoch": 0.06573204838163778, "grad_norm": 1.9043179612822492, "learning_rate": 1.9933229362730608e-05, "loss": 0.6577, "step": 1845 }, { "epoch": 0.0657676755081319, "grad_norm": 1.3618476211224744, "learning_rate": 1.993309617038428e-05, "loss": 0.5627, "step": 1846 }, { "epoch": 0.065803302634626, "grad_norm": 1.9115006872303124, "learning_rate": 1.9932962845771834e-05, "loss": 0.7082, "step": 1847 }, { "epoch": 0.06583892976112012, "grad_norm": 1.4954904482195597, "learning_rate": 1.993282938889504e-05, "loss": 0.5739, "step": 1848 }, { "epoch": 0.06587455688761423, "grad_norm": 1.4220253391223086, "learning_rate": 1.9932695799755688e-05, "loss": 0.5494, "step": 1849 }, { "epoch": 0.06591018401410834, "grad_norm": 2.2844166633609406, "learning_rate": 1.9932562078355544e-05, "loss": 0.6947, "step": 1850 }, { "epoch": 0.06594581114060245, "grad_norm": 1.3263462165882405, "learning_rate": 1.9932428224696396e-05, "loss": 0.4727, "step": 1851 }, { "epoch": 0.06598143826709657, "grad_norm": 2.015654692418085, "learning_rate": 1.9932294238780024e-05, "loss": 0.6413, "step": 1852 }, { "epoch": 0.06601706539359067, "grad_norm": 1.3359087595770625, "learning_rate": 1.993216012060821e-05, "loss": 0.5247, "step": 1853 }, { "epoch": 0.06605269252008479, "grad_norm": 1.8945733183835636, "learning_rate": 1.9932025870182744e-05, "loss": 0.6412, "step": 1854 }, { "epoch": 0.06608831964657891, "grad_norm": 1.4614476171544493, "learning_rate": 1.993189148750541e-05, "loss": 0.5833, "step": 1855 }, { "epoch": 0.06612394677307301, "grad_norm": 1.4593921075137504, "learning_rate": 1.9931756972578e-05, "loss": 0.582, "step": 1856 }, { "epoch": 0.06615957389956713, "grad_norm": 1.878891354707556, "learning_rate": 1.9931622325402304e-05, "loss": 0.64, "step": 1857 }, { "epoch": 0.06619520102606125, "grad_norm": 1.4055785494409627, "learning_rate": 1.993148754598011e-05, "loss": 0.6112, "step": 1858 }, { "epoch": 0.06623082815255535, "grad_norm": 1.611635262980218, "learning_rate": 1.9931352634313222e-05, "loss": 0.7263, "step": 1859 }, { "epoch": 0.06626645527904947, "grad_norm": 2.032362352585886, "learning_rate": 1.9931217590403435e-05, "loss": 0.5862, "step": 1860 }, { "epoch": 0.06630208240554358, "grad_norm": 1.6342258191294383, "learning_rate": 1.993108241425254e-05, "loss": 0.6855, "step": 1861 }, { "epoch": 0.0663377095320377, "grad_norm": 4.0480863750432965, "learning_rate": 1.993094710586234e-05, "loss": 0.6631, "step": 1862 }, { "epoch": 0.0663733366585318, "grad_norm": 1.9864457518307832, "learning_rate": 1.993081166523464e-05, "loss": 0.5384, "step": 1863 }, { "epoch": 0.06640896378502592, "grad_norm": 1.6128465719109983, "learning_rate": 1.9930676092371243e-05, "loss": 0.6459, "step": 1864 }, { "epoch": 0.06644459091152004, "grad_norm": 1.842659910041634, "learning_rate": 1.993054038727395e-05, "loss": 0.5887, "step": 1865 }, { "epoch": 0.06648021803801414, "grad_norm": 1.401629327532761, "learning_rate": 1.9930404549944577e-05, "loss": 0.6016, "step": 1866 }, { "epoch": 0.06651584516450826, "grad_norm": 1.396201594817403, "learning_rate": 1.993026858038492e-05, "loss": 0.5398, "step": 1867 }, { "epoch": 0.06655147229100238, "grad_norm": 1.885586233782095, "learning_rate": 1.9930132478596795e-05, "loss": 0.6306, "step": 1868 }, { "epoch": 0.06658709941749648, "grad_norm": 1.2691943158424508, "learning_rate": 1.992999624458202e-05, "loss": 0.5268, "step": 1869 }, { "epoch": 0.0666227265439906, "grad_norm": 1.2615553031846964, "learning_rate": 1.99298598783424e-05, "loss": 0.5581, "step": 1870 }, { "epoch": 0.06665835367048471, "grad_norm": 1.2738651609909983, "learning_rate": 1.9929723379879758e-05, "loss": 0.5565, "step": 1871 }, { "epoch": 0.06669398079697882, "grad_norm": 1.372047041536614, "learning_rate": 1.9929586749195907e-05, "loss": 0.5442, "step": 1872 }, { "epoch": 0.06672960792347293, "grad_norm": 1.8543560564453172, "learning_rate": 1.9929449986292672e-05, "loss": 0.7289, "step": 1873 }, { "epoch": 0.06676523504996705, "grad_norm": 1.4490638103181295, "learning_rate": 1.9929313091171866e-05, "loss": 0.5392, "step": 1874 }, { "epoch": 0.06680086217646115, "grad_norm": 1.3462120209303103, "learning_rate": 1.9929176063835317e-05, "loss": 0.5519, "step": 1875 }, { "epoch": 0.06683648930295527, "grad_norm": 2.0182370506692524, "learning_rate": 1.992903890428485e-05, "loss": 0.6669, "step": 1876 }, { "epoch": 0.06687211642944939, "grad_norm": 1.3673173749827012, "learning_rate": 1.992890161252229e-05, "loss": 0.5421, "step": 1877 }, { "epoch": 0.06690774355594349, "grad_norm": 1.9445139788341776, "learning_rate": 1.9928764188549462e-05, "loss": 0.6511, "step": 1878 }, { "epoch": 0.06694337068243761, "grad_norm": 2.028603074954545, "learning_rate": 1.9928626632368204e-05, "loss": 0.67, "step": 1879 }, { "epoch": 0.06697899780893173, "grad_norm": 2.048624543007889, "learning_rate": 1.9928488943980338e-05, "loss": 0.6871, "step": 1880 }, { "epoch": 0.06701462493542583, "grad_norm": 1.3961366676921956, "learning_rate": 1.9928351123387703e-05, "loss": 0.5973, "step": 1881 }, { "epoch": 0.06705025206191995, "grad_norm": 1.7486628536102033, "learning_rate": 1.9928213170592133e-05, "loss": 0.7068, "step": 1882 }, { "epoch": 0.06708587918841406, "grad_norm": 1.6399184643737872, "learning_rate": 1.992807508559547e-05, "loss": 0.6008, "step": 1883 }, { "epoch": 0.06712150631490817, "grad_norm": 3.5403467818354573, "learning_rate": 1.992793686839954e-05, "loss": 0.4754, "step": 1884 }, { "epoch": 0.06715713344140228, "grad_norm": 1.527746626669502, "learning_rate": 1.99277985190062e-05, "loss": 0.5895, "step": 1885 }, { "epoch": 0.0671927605678964, "grad_norm": 1.3691576290010932, "learning_rate": 1.992766003741728e-05, "loss": 0.5894, "step": 1886 }, { "epoch": 0.0672283876943905, "grad_norm": 2.0160146053044152, "learning_rate": 1.9927521423634627e-05, "loss": 0.679, "step": 1887 }, { "epoch": 0.06726401482088462, "grad_norm": 2.006993749231654, "learning_rate": 1.992738267766009e-05, "loss": 0.7097, "step": 1888 }, { "epoch": 0.06729964194737874, "grad_norm": 1.333469634523961, "learning_rate": 1.992724379949551e-05, "loss": 0.5781, "step": 1889 }, { "epoch": 0.06733526907387284, "grad_norm": 1.8648070555896552, "learning_rate": 1.992710478914274e-05, "loss": 0.805, "step": 1890 }, { "epoch": 0.06737089620036696, "grad_norm": 1.7523907404384031, "learning_rate": 1.9926965646603632e-05, "loss": 0.5975, "step": 1891 }, { "epoch": 0.06740652332686108, "grad_norm": 1.627136190926465, "learning_rate": 1.9926826371880035e-05, "loss": 0.5508, "step": 1892 }, { "epoch": 0.06744215045335518, "grad_norm": 1.4440598907823812, "learning_rate": 1.9926686964973813e-05, "loss": 0.5746, "step": 1893 }, { "epoch": 0.0674777775798493, "grad_norm": 1.3150399637927155, "learning_rate": 1.9926547425886812e-05, "loss": 0.5833, "step": 1894 }, { "epoch": 0.06751340470634341, "grad_norm": 1.4956655766209512, "learning_rate": 1.9926407754620892e-05, "loss": 0.6098, "step": 1895 }, { "epoch": 0.06754903183283752, "grad_norm": 1.239723974715641, "learning_rate": 1.9926267951177917e-05, "loss": 0.5463, "step": 1896 }, { "epoch": 0.06758465895933163, "grad_norm": 1.3140938745112631, "learning_rate": 1.992612801555975e-05, "loss": 0.6033, "step": 1897 }, { "epoch": 0.06762028608582575, "grad_norm": 2.136323068387811, "learning_rate": 1.9925987947768244e-05, "loss": 0.6186, "step": 1898 }, { "epoch": 0.06765591321231985, "grad_norm": 1.7309586726755604, "learning_rate": 1.9925847747805274e-05, "loss": 0.5807, "step": 1899 }, { "epoch": 0.06769154033881397, "grad_norm": 1.9835888362330245, "learning_rate": 1.99257074156727e-05, "loss": 0.5591, "step": 1900 }, { "epoch": 0.06772716746530809, "grad_norm": 1.4604733337040239, "learning_rate": 1.99255669513724e-05, "loss": 0.5946, "step": 1901 }, { "epoch": 0.06776279459180219, "grad_norm": 2.291689014625213, "learning_rate": 1.992542635490623e-05, "loss": 0.6179, "step": 1902 }, { "epoch": 0.06779842171829631, "grad_norm": 1.6490023908431757, "learning_rate": 1.992528562627608e-05, "loss": 0.7057, "step": 1903 }, { "epoch": 0.06783404884479043, "grad_norm": 2.2133430358075756, "learning_rate": 1.9925144765483807e-05, "loss": 0.6015, "step": 1904 }, { "epoch": 0.06786967597128453, "grad_norm": 2.189434241750004, "learning_rate": 1.99250037725313e-05, "loss": 0.6623, "step": 1905 }, { "epoch": 0.06790530309777865, "grad_norm": 2.2662613533403078, "learning_rate": 1.992486264742043e-05, "loss": 0.6093, "step": 1906 }, { "epoch": 0.06794093022427276, "grad_norm": 1.8502644011501654, "learning_rate": 1.9924721390153072e-05, "loss": 0.6639, "step": 1907 }, { "epoch": 0.06797655735076688, "grad_norm": 1.3420466460606395, "learning_rate": 1.9924580000731116e-05, "loss": 0.5266, "step": 1908 }, { "epoch": 0.06801218447726098, "grad_norm": 1.3962049253454794, "learning_rate": 1.9924438479156437e-05, "loss": 0.592, "step": 1909 }, { "epoch": 0.0680478116037551, "grad_norm": 1.3184526945526005, "learning_rate": 1.9924296825430928e-05, "loss": 0.5353, "step": 1910 }, { "epoch": 0.06808343873024922, "grad_norm": 1.28727355608433, "learning_rate": 1.9924155039556464e-05, "loss": 0.5632, "step": 1911 }, { "epoch": 0.06811906585674332, "grad_norm": 2.7466925816469896, "learning_rate": 1.9924013121534943e-05, "loss": 0.6729, "step": 1912 }, { "epoch": 0.06815469298323744, "grad_norm": 1.3864847038235928, "learning_rate": 1.9923871071368253e-05, "loss": 0.5905, "step": 1913 }, { "epoch": 0.06819032010973156, "grad_norm": 2.1853541318590253, "learning_rate": 1.9923728889058277e-05, "loss": 0.6273, "step": 1914 }, { "epoch": 0.06822594723622566, "grad_norm": 2.203223843700309, "learning_rate": 1.992358657460692e-05, "loss": 0.5858, "step": 1915 }, { "epoch": 0.06826157436271978, "grad_norm": 1.593476649127412, "learning_rate": 1.992344412801607e-05, "loss": 0.6081, "step": 1916 }, { "epoch": 0.0682972014892139, "grad_norm": 1.9028920334902593, "learning_rate": 1.9923301549287624e-05, "loss": 0.5371, "step": 1917 }, { "epoch": 0.068332828615708, "grad_norm": 1.9309558113725478, "learning_rate": 1.9923158838423482e-05, "loss": 0.6301, "step": 1918 }, { "epoch": 0.06836845574220211, "grad_norm": 1.4249362786401767, "learning_rate": 1.9923015995425547e-05, "loss": 0.554, "step": 1919 }, { "epoch": 0.06840408286869623, "grad_norm": 1.3453421395052723, "learning_rate": 1.9922873020295714e-05, "loss": 0.5594, "step": 1920 }, { "epoch": 0.06843970999519033, "grad_norm": 2.379855643654357, "learning_rate": 1.9922729913035893e-05, "loss": 0.6603, "step": 1921 }, { "epoch": 0.06847533712168445, "grad_norm": 1.6182284745832995, "learning_rate": 1.992258667364799e-05, "loss": 0.6391, "step": 1922 }, { "epoch": 0.06851096424817857, "grad_norm": 1.1811084004086205, "learning_rate": 1.9922443302133906e-05, "loss": 0.4702, "step": 1923 }, { "epoch": 0.06854659137467267, "grad_norm": 1.3335706016038948, "learning_rate": 1.9922299798495555e-05, "loss": 0.6057, "step": 1924 }, { "epoch": 0.06858221850116679, "grad_norm": 1.211937066602994, "learning_rate": 1.9922156162734846e-05, "loss": 0.5695, "step": 1925 }, { "epoch": 0.0686178456276609, "grad_norm": 1.2496268781392583, "learning_rate": 1.9922012394853695e-05, "loss": 0.5761, "step": 1926 }, { "epoch": 0.06865347275415501, "grad_norm": 1.3740125724730574, "learning_rate": 1.9921868494854014e-05, "loss": 0.5634, "step": 1927 }, { "epoch": 0.06868909988064913, "grad_norm": 1.9586847278821256, "learning_rate": 1.992172446273772e-05, "loss": 0.6126, "step": 1928 }, { "epoch": 0.06872472700714324, "grad_norm": 1.6306790475304427, "learning_rate": 1.9921580298506728e-05, "loss": 0.5384, "step": 1929 }, { "epoch": 0.06876035413363735, "grad_norm": 1.8137661692421045, "learning_rate": 1.992143600216296e-05, "loss": 0.7187, "step": 1930 }, { "epoch": 0.06879598126013146, "grad_norm": 1.7900204102946098, "learning_rate": 1.9921291573708338e-05, "loss": 0.6745, "step": 1931 }, { "epoch": 0.06883160838662558, "grad_norm": 1.3288841993060199, "learning_rate": 1.9921147013144782e-05, "loss": 0.5522, "step": 1932 }, { "epoch": 0.06886723551311968, "grad_norm": 1.237652346624355, "learning_rate": 1.9921002320474218e-05, "loss": 0.5427, "step": 1933 }, { "epoch": 0.0689028626396138, "grad_norm": 1.0948176293633256, "learning_rate": 1.9920857495698575e-05, "loss": 0.5408, "step": 1934 }, { "epoch": 0.06893848976610792, "grad_norm": 1.9404360047549012, "learning_rate": 1.992071253881978e-05, "loss": 0.7123, "step": 1935 }, { "epoch": 0.06897411689260202, "grad_norm": 1.2998601739187807, "learning_rate": 1.9920567449839764e-05, "loss": 0.5515, "step": 1936 }, { "epoch": 0.06900974401909614, "grad_norm": 1.307482090490229, "learning_rate": 1.9920422228760458e-05, "loss": 0.4976, "step": 1937 }, { "epoch": 0.06904537114559026, "grad_norm": 1.3684816971760103, "learning_rate": 1.99202768755838e-05, "loss": 0.5198, "step": 1938 }, { "epoch": 0.06908099827208436, "grad_norm": 4.800998868849512, "learning_rate": 1.9920131390311716e-05, "loss": 0.7413, "step": 1939 }, { "epoch": 0.06911662539857848, "grad_norm": 1.377461360665615, "learning_rate": 1.991998577294615e-05, "loss": 0.5952, "step": 1940 }, { "epoch": 0.0691522525250726, "grad_norm": 1.3394799670541149, "learning_rate": 1.991984002348904e-05, "loss": 0.5422, "step": 1941 }, { "epoch": 0.0691878796515667, "grad_norm": 1.471514427035454, "learning_rate": 1.991969414194233e-05, "loss": 0.5894, "step": 1942 }, { "epoch": 0.06922350677806081, "grad_norm": 1.3104756720513837, "learning_rate": 1.9919548128307954e-05, "loss": 0.5409, "step": 1943 }, { "epoch": 0.06925913390455493, "grad_norm": 1.2989157062972048, "learning_rate": 1.9919401982587865e-05, "loss": 0.488, "step": 1944 }, { "epoch": 0.06929476103104903, "grad_norm": 1.3030953668393979, "learning_rate": 1.9919255704784004e-05, "loss": 0.5389, "step": 1945 }, { "epoch": 0.06933038815754315, "grad_norm": 1.9745133833021917, "learning_rate": 1.9919109294898317e-05, "loss": 0.6073, "step": 1946 }, { "epoch": 0.06936601528403727, "grad_norm": 2.1620908144300093, "learning_rate": 1.991896275293276e-05, "loss": 0.6381, "step": 1947 }, { "epoch": 0.06940164241053137, "grad_norm": 1.7448162586894869, "learning_rate": 1.991881607888928e-05, "loss": 0.6511, "step": 1948 }, { "epoch": 0.06943726953702549, "grad_norm": 2.2882899329639366, "learning_rate": 1.9918669272769832e-05, "loss": 0.7124, "step": 1949 }, { "epoch": 0.0694728966635196, "grad_norm": 2.1964817026096726, "learning_rate": 1.991852233457637e-05, "loss": 0.6458, "step": 1950 }, { "epoch": 0.06950852379001371, "grad_norm": 1.2674075206519653, "learning_rate": 1.991837526431085e-05, "loss": 0.4874, "step": 1951 }, { "epoch": 0.06954415091650783, "grad_norm": 1.3554045971124873, "learning_rate": 1.9918228061975234e-05, "loss": 0.516, "step": 1952 }, { "epoch": 0.06957977804300194, "grad_norm": 1.2716674418777765, "learning_rate": 1.9918080727571473e-05, "loss": 0.5396, "step": 1953 }, { "epoch": 0.06961540516949605, "grad_norm": 1.8725920484039893, "learning_rate": 1.991793326110154e-05, "loss": 0.6065, "step": 1954 }, { "epoch": 0.06965103229599016, "grad_norm": 2.062257472037283, "learning_rate": 1.991778566256739e-05, "loss": 0.6925, "step": 1955 }, { "epoch": 0.06968665942248428, "grad_norm": 2.080171316713332, "learning_rate": 1.9917637931970994e-05, "loss": 0.6, "step": 1956 }, { "epoch": 0.0697222865489784, "grad_norm": 1.9581943683697052, "learning_rate": 1.9917490069314315e-05, "loss": 0.59, "step": 1957 }, { "epoch": 0.0697579136754725, "grad_norm": 1.4471930844896337, "learning_rate": 1.9917342074599325e-05, "loss": 0.571, "step": 1958 }, { "epoch": 0.06979354080196662, "grad_norm": 1.3959339373391584, "learning_rate": 1.991719394782799e-05, "loss": 0.5463, "step": 1959 }, { "epoch": 0.06982916792846074, "grad_norm": 2.417429162311425, "learning_rate": 1.9917045689002294e-05, "loss": 0.6393, "step": 1960 }, { "epoch": 0.06986479505495484, "grad_norm": 2.0084935371678414, "learning_rate": 1.9916897298124197e-05, "loss": 0.6989, "step": 1961 }, { "epoch": 0.06990042218144896, "grad_norm": 1.3457369669187198, "learning_rate": 1.991674877519568e-05, "loss": 0.5595, "step": 1962 }, { "epoch": 0.06993604930794307, "grad_norm": 1.7434892880088098, "learning_rate": 1.9916600120218724e-05, "loss": 0.639, "step": 1963 }, { "epoch": 0.06997167643443718, "grad_norm": 1.6833207479251087, "learning_rate": 1.9916451333195305e-05, "loss": 0.5553, "step": 1964 }, { "epoch": 0.0700073035609313, "grad_norm": 2.0092853822747494, "learning_rate": 1.9916302414127408e-05, "loss": 0.7085, "step": 1965 }, { "epoch": 0.07004293068742541, "grad_norm": 1.7332414546268766, "learning_rate": 1.9916153363017012e-05, "loss": 0.6415, "step": 1966 }, { "epoch": 0.07007855781391951, "grad_norm": 1.7593982303285316, "learning_rate": 1.99160041798661e-05, "loss": 0.5994, "step": 1967 }, { "epoch": 0.07011418494041363, "grad_norm": 1.6442476194934965, "learning_rate": 1.9915854864676665e-05, "loss": 0.6078, "step": 1968 }, { "epoch": 0.07014981206690775, "grad_norm": 1.3814565578865832, "learning_rate": 1.991570541745069e-05, "loss": 0.5157, "step": 1969 }, { "epoch": 0.07018543919340185, "grad_norm": 1.3779824500802542, "learning_rate": 1.9915555838190167e-05, "loss": 0.5642, "step": 1970 }, { "epoch": 0.07022106631989597, "grad_norm": 1.2533170567454617, "learning_rate": 1.991540612689709e-05, "loss": 0.5803, "step": 1971 }, { "epoch": 0.07025669344639009, "grad_norm": 4.5978797339877255, "learning_rate": 1.9915256283573446e-05, "loss": 0.6311, "step": 1972 }, { "epoch": 0.07029232057288419, "grad_norm": 1.7091348019838883, "learning_rate": 1.9915106308221234e-05, "loss": 0.6205, "step": 1973 }, { "epoch": 0.07032794769937831, "grad_norm": 1.7792705157692497, "learning_rate": 1.9914956200842453e-05, "loss": 0.6086, "step": 1974 }, { "epoch": 0.07036357482587242, "grad_norm": 1.3949441028330958, "learning_rate": 1.99148059614391e-05, "loss": 0.5607, "step": 1975 }, { "epoch": 0.07039920195236653, "grad_norm": 1.6769522815442752, "learning_rate": 1.9914655590013177e-05, "loss": 0.6172, "step": 1976 }, { "epoch": 0.07043482907886064, "grad_norm": 1.649689637611104, "learning_rate": 1.9914505086566683e-05, "loss": 0.6064, "step": 1977 }, { "epoch": 0.07047045620535476, "grad_norm": 1.385110547018733, "learning_rate": 1.9914354451101627e-05, "loss": 0.5425, "step": 1978 }, { "epoch": 0.07050608333184887, "grad_norm": 1.8542976998957674, "learning_rate": 1.991420368362001e-05, "loss": 0.5759, "step": 1979 }, { "epoch": 0.07054171045834298, "grad_norm": 1.64878067329774, "learning_rate": 1.991405278412384e-05, "loss": 0.6123, "step": 1980 }, { "epoch": 0.0705773375848371, "grad_norm": 1.9664295150303375, "learning_rate": 1.991390175261513e-05, "loss": 0.6581, "step": 1981 }, { "epoch": 0.0706129647113312, "grad_norm": 1.5562989686125437, "learning_rate": 1.9913750589095886e-05, "loss": 0.6149, "step": 1982 }, { "epoch": 0.07064859183782532, "grad_norm": 1.723862853489455, "learning_rate": 1.9913599293568124e-05, "loss": 0.5878, "step": 1983 }, { "epoch": 0.07068421896431944, "grad_norm": 1.6876498660899149, "learning_rate": 1.9913447866033858e-05, "loss": 0.6899, "step": 1984 }, { "epoch": 0.07071984609081354, "grad_norm": 2.4469414805171823, "learning_rate": 1.9913296306495108e-05, "loss": 0.6559, "step": 1985 }, { "epoch": 0.07075547321730766, "grad_norm": 1.2092555185581002, "learning_rate": 1.9913144614953883e-05, "loss": 0.5194, "step": 1986 }, { "epoch": 0.07079110034380177, "grad_norm": 1.2834613003855573, "learning_rate": 1.9912992791412212e-05, "loss": 0.585, "step": 1987 }, { "epoch": 0.07082672747029588, "grad_norm": 2.400366716675593, "learning_rate": 1.9912840835872112e-05, "loss": 0.6288, "step": 1988 }, { "epoch": 0.07086235459679, "grad_norm": 1.2880029765034455, "learning_rate": 1.991268874833561e-05, "loss": 0.5459, "step": 1989 }, { "epoch": 0.07089798172328411, "grad_norm": 1.32281518027182, "learning_rate": 1.991253652880473e-05, "loss": 0.6084, "step": 1990 }, { "epoch": 0.07093360884977822, "grad_norm": 2.195604897751773, "learning_rate": 1.9912384177281494e-05, "loss": 0.634, "step": 1991 }, { "epoch": 0.07096923597627233, "grad_norm": 2.5415233944376396, "learning_rate": 1.9912231693767933e-05, "loss": 0.6948, "step": 1992 }, { "epoch": 0.07100486310276645, "grad_norm": 1.2693855968698844, "learning_rate": 1.9912079078266086e-05, "loss": 0.493, "step": 1993 }, { "epoch": 0.07104049022926055, "grad_norm": 2.0249778951467583, "learning_rate": 1.991192633077797e-05, "loss": 0.6312, "step": 1994 }, { "epoch": 0.07107611735575467, "grad_norm": 2.1896882436498606, "learning_rate": 1.991177345130563e-05, "loss": 0.6444, "step": 1995 }, { "epoch": 0.07111174448224879, "grad_norm": 1.4815328470305749, "learning_rate": 1.99116204398511e-05, "loss": 0.6248, "step": 1996 }, { "epoch": 0.07114737160874289, "grad_norm": 2.3860306719244684, "learning_rate": 1.9911467296416412e-05, "loss": 0.6179, "step": 1997 }, { "epoch": 0.07118299873523701, "grad_norm": 1.9193119650221748, "learning_rate": 1.9911314021003614e-05, "loss": 0.7059, "step": 1998 }, { "epoch": 0.07121862586173112, "grad_norm": 1.1605844953913285, "learning_rate": 1.991116061361474e-05, "loss": 0.5116, "step": 1999 }, { "epoch": 0.07125425298822523, "grad_norm": 1.7015368896901846, "learning_rate": 1.9911007074251832e-05, "loss": 0.5655, "step": 2000 }, { "epoch": 0.07128988011471935, "grad_norm": 2.4340139851862723, "learning_rate": 1.991085340291694e-05, "loss": 0.6274, "step": 2001 }, { "epoch": 0.07132550724121346, "grad_norm": 2.1816612676006044, "learning_rate": 1.9910699599612112e-05, "loss": 0.6637, "step": 2002 }, { "epoch": 0.07136113436770758, "grad_norm": 1.5186318941842658, "learning_rate": 1.9910545664339387e-05, "loss": 0.4827, "step": 2003 }, { "epoch": 0.07139676149420168, "grad_norm": 1.378832886475098, "learning_rate": 1.991039159710082e-05, "loss": 0.5536, "step": 2004 }, { "epoch": 0.0714323886206958, "grad_norm": 1.6552643798771933, "learning_rate": 1.9910237397898463e-05, "loss": 0.5591, "step": 2005 }, { "epoch": 0.07146801574718992, "grad_norm": 1.4097718143845641, "learning_rate": 1.9910083066734366e-05, "loss": 0.5623, "step": 2006 }, { "epoch": 0.07150364287368402, "grad_norm": 1.4776200403387305, "learning_rate": 1.990992860361059e-05, "loss": 0.5241, "step": 2007 }, { "epoch": 0.07153927000017814, "grad_norm": 2.0330058953668577, "learning_rate": 1.9909774008529184e-05, "loss": 0.649, "step": 2008 }, { "epoch": 0.07157489712667225, "grad_norm": 1.370722563227471, "learning_rate": 1.990961928149221e-05, "loss": 0.5876, "step": 2009 }, { "epoch": 0.07161052425316636, "grad_norm": 1.71430352644329, "learning_rate": 1.9909464422501733e-05, "loss": 0.6757, "step": 2010 }, { "epoch": 0.07164615137966047, "grad_norm": 2.013692947699419, "learning_rate": 1.9909309431559808e-05, "loss": 0.5616, "step": 2011 }, { "epoch": 0.07168177850615459, "grad_norm": 1.794397162117778, "learning_rate": 1.9909154308668503e-05, "loss": 0.5714, "step": 2012 }, { "epoch": 0.0717174056326487, "grad_norm": 2.3040469547929834, "learning_rate": 1.990899905382988e-05, "loss": 0.6161, "step": 2013 }, { "epoch": 0.07175303275914281, "grad_norm": 1.8809028620667174, "learning_rate": 1.9908843667046014e-05, "loss": 0.6721, "step": 2014 }, { "epoch": 0.07178865988563693, "grad_norm": 1.3069287387920288, "learning_rate": 1.9908688148318964e-05, "loss": 0.5362, "step": 2015 }, { "epoch": 0.07182428701213103, "grad_norm": 1.4968456370042693, "learning_rate": 1.9908532497650808e-05, "loss": 0.5646, "step": 2016 }, { "epoch": 0.07185991413862515, "grad_norm": 1.4186840456817282, "learning_rate": 1.9908376715043616e-05, "loss": 0.5499, "step": 2017 }, { "epoch": 0.07189554126511927, "grad_norm": 1.4361803927513688, "learning_rate": 1.990822080049946e-05, "loss": 0.5819, "step": 2018 }, { "epoch": 0.07193116839161337, "grad_norm": 1.1721257081309406, "learning_rate": 1.990806475402042e-05, "loss": 0.5467, "step": 2019 }, { "epoch": 0.07196679551810749, "grad_norm": 2.601009963933779, "learning_rate": 1.9907908575608573e-05, "loss": 0.601, "step": 2020 }, { "epoch": 0.0720024226446016, "grad_norm": 1.4010728662557799, "learning_rate": 1.9907752265265997e-05, "loss": 0.537, "step": 2021 }, { "epoch": 0.07203804977109571, "grad_norm": 1.631732436428987, "learning_rate": 1.9907595822994774e-05, "loss": 0.6558, "step": 2022 }, { "epoch": 0.07207367689758983, "grad_norm": 1.3322140765300727, "learning_rate": 1.990743924879699e-05, "loss": 0.5421, "step": 2023 }, { "epoch": 0.07210930402408394, "grad_norm": 1.442610684610097, "learning_rate": 1.9907282542674726e-05, "loss": 0.5873, "step": 2024 }, { "epoch": 0.07214493115057805, "grad_norm": 1.8338919508573772, "learning_rate": 1.990712570463007e-05, "loss": 0.7228, "step": 2025 }, { "epoch": 0.07218055827707216, "grad_norm": 2.182759061316889, "learning_rate": 1.990696873466511e-05, "loss": 0.6876, "step": 2026 }, { "epoch": 0.07221618540356628, "grad_norm": 1.247433242323788, "learning_rate": 1.990681163278194e-05, "loss": 0.5742, "step": 2027 }, { "epoch": 0.07225181253006038, "grad_norm": 1.9939667605452036, "learning_rate": 1.9906654398982647e-05, "loss": 0.5583, "step": 2028 }, { "epoch": 0.0722874396565545, "grad_norm": 2.065270008107282, "learning_rate": 1.9906497033269325e-05, "loss": 0.6871, "step": 2029 }, { "epoch": 0.07232306678304862, "grad_norm": 1.2798559665986171, "learning_rate": 1.990633953564407e-05, "loss": 0.5572, "step": 2030 }, { "epoch": 0.07235869390954272, "grad_norm": 1.8565251230585196, "learning_rate": 1.9906181906108983e-05, "loss": 0.6703, "step": 2031 }, { "epoch": 0.07239432103603684, "grad_norm": 1.3307234709627178, "learning_rate": 1.9906024144666158e-05, "loss": 0.5036, "step": 2032 }, { "epoch": 0.07242994816253095, "grad_norm": 1.7023818945989673, "learning_rate": 1.9905866251317697e-05, "loss": 0.6641, "step": 2033 }, { "epoch": 0.07246557528902506, "grad_norm": 1.957233684505031, "learning_rate": 1.9905708226065704e-05, "loss": 0.6668, "step": 2034 }, { "epoch": 0.07250120241551918, "grad_norm": 1.1763567770191545, "learning_rate": 1.9905550068912284e-05, "loss": 0.5091, "step": 2035 }, { "epoch": 0.07253682954201329, "grad_norm": 1.267005772378866, "learning_rate": 1.990539177985954e-05, "loss": 0.5523, "step": 2036 }, { "epoch": 0.0725724566685074, "grad_norm": 1.8749449633244086, "learning_rate": 1.990523335890958e-05, "loss": 0.6382, "step": 2037 }, { "epoch": 0.07260808379500151, "grad_norm": 1.897134679534456, "learning_rate": 1.9905074806064514e-05, "loss": 0.6748, "step": 2038 }, { "epoch": 0.07264371092149563, "grad_norm": 1.4113787138998086, "learning_rate": 1.9904916121326455e-05, "loss": 0.54, "step": 2039 }, { "epoch": 0.07267933804798973, "grad_norm": 1.939100857466455, "learning_rate": 1.9904757304697516e-05, "loss": 0.6285, "step": 2040 }, { "epoch": 0.07271496517448385, "grad_norm": 1.840559921434337, "learning_rate": 1.9904598356179808e-05, "loss": 0.6504, "step": 2041 }, { "epoch": 0.07275059230097797, "grad_norm": 1.2483609792479866, "learning_rate": 1.990443927577545e-05, "loss": 0.511, "step": 2042 }, { "epoch": 0.07278621942747207, "grad_norm": 1.982834741241715, "learning_rate": 1.9904280063486563e-05, "loss": 0.6122, "step": 2043 }, { "epoch": 0.07282184655396619, "grad_norm": 1.2486513314751113, "learning_rate": 1.9904120719315262e-05, "loss": 0.5603, "step": 2044 }, { "epoch": 0.0728574736804603, "grad_norm": 2.271411177217239, "learning_rate": 1.9903961243263673e-05, "loss": 0.6413, "step": 2045 }, { "epoch": 0.07289310080695441, "grad_norm": 1.7790024371676147, "learning_rate": 1.9903801635333915e-05, "loss": 0.5529, "step": 2046 }, { "epoch": 0.07292872793344853, "grad_norm": 1.1754397650272412, "learning_rate": 1.9903641895528117e-05, "loss": 0.5245, "step": 2047 }, { "epoch": 0.07296435505994264, "grad_norm": 1.669595875157554, "learning_rate": 1.9903482023848407e-05, "loss": 0.6534, "step": 2048 }, { "epoch": 0.07299998218643676, "grad_norm": 1.456354176817866, "learning_rate": 1.9903322020296908e-05, "loss": 0.5438, "step": 2049 }, { "epoch": 0.07303560931293086, "grad_norm": 2.0446668615028414, "learning_rate": 1.9903161884875756e-05, "loss": 0.7252, "step": 2050 }, { "epoch": 0.07307123643942498, "grad_norm": 1.5880439668469777, "learning_rate": 1.9903001617587082e-05, "loss": 0.6275, "step": 2051 }, { "epoch": 0.0731068635659191, "grad_norm": 1.4505313272414548, "learning_rate": 1.990284121843302e-05, "loss": 0.5673, "step": 2052 }, { "epoch": 0.0731424906924132, "grad_norm": 1.2521762347032455, "learning_rate": 1.9902680687415704e-05, "loss": 0.5241, "step": 2053 }, { "epoch": 0.07317811781890732, "grad_norm": 1.30851463683316, "learning_rate": 1.9902520024537277e-05, "loss": 0.5237, "step": 2054 }, { "epoch": 0.07321374494540143, "grad_norm": 1.1414404477353033, "learning_rate": 1.990235922979987e-05, "loss": 0.4626, "step": 2055 }, { "epoch": 0.07324937207189554, "grad_norm": 1.8868515319385004, "learning_rate": 1.990219830320563e-05, "loss": 0.6087, "step": 2056 }, { "epoch": 0.07328499919838966, "grad_norm": 1.3231235561725172, "learning_rate": 1.9902037244756697e-05, "loss": 0.5293, "step": 2057 }, { "epoch": 0.07332062632488377, "grad_norm": 1.4189830201614306, "learning_rate": 1.9901876054455217e-05, "loss": 0.5185, "step": 2058 }, { "epoch": 0.07335625345137788, "grad_norm": 1.7464910665669762, "learning_rate": 1.990171473230334e-05, "loss": 0.7106, "step": 2059 }, { "epoch": 0.07339188057787199, "grad_norm": 1.3875627931306733, "learning_rate": 1.9901553278303206e-05, "loss": 0.5502, "step": 2060 }, { "epoch": 0.07342750770436611, "grad_norm": 1.3262975802028096, "learning_rate": 1.9901391692456974e-05, "loss": 0.4717, "step": 2061 }, { "epoch": 0.07346313483086021, "grad_norm": 1.7914444786594692, "learning_rate": 1.9901229974766787e-05, "loss": 0.66, "step": 2062 }, { "epoch": 0.07349876195735433, "grad_norm": 1.8615803611459576, "learning_rate": 1.9901068125234805e-05, "loss": 0.6054, "step": 2063 }, { "epoch": 0.07353438908384845, "grad_norm": 2.382820469813497, "learning_rate": 1.990090614386318e-05, "loss": 0.6753, "step": 2064 }, { "epoch": 0.07357001621034255, "grad_norm": 1.4328576968755875, "learning_rate": 1.9900744030654066e-05, "loss": 0.614, "step": 2065 }, { "epoch": 0.07360564333683667, "grad_norm": 2.4438261680863826, "learning_rate": 1.9900581785609627e-05, "loss": 0.6451, "step": 2066 }, { "epoch": 0.07364127046333079, "grad_norm": 1.17954971946949, "learning_rate": 1.9900419408732024e-05, "loss": 0.5377, "step": 2067 }, { "epoch": 0.07367689758982489, "grad_norm": 1.358696885254295, "learning_rate": 1.9900256900023413e-05, "loss": 0.5619, "step": 2068 }, { "epoch": 0.073712524716319, "grad_norm": 6.656851267038066, "learning_rate": 1.990009425948596e-05, "loss": 0.6681, "step": 2069 }, { "epoch": 0.07374815184281312, "grad_norm": 1.4421506262009058, "learning_rate": 1.9899931487121838e-05, "loss": 0.5854, "step": 2070 }, { "epoch": 0.07378377896930723, "grad_norm": 1.2038760817051695, "learning_rate": 1.9899768582933205e-05, "loss": 0.591, "step": 2071 }, { "epoch": 0.07381940609580134, "grad_norm": 2.0483069306267114, "learning_rate": 1.9899605546922237e-05, "loss": 0.6817, "step": 2072 }, { "epoch": 0.07385503322229546, "grad_norm": 2.780756053005157, "learning_rate": 1.98994423790911e-05, "loss": 0.6752, "step": 2073 }, { "epoch": 0.07389066034878956, "grad_norm": 1.3175089480693147, "learning_rate": 1.9899279079441967e-05, "loss": 0.5631, "step": 2074 }, { "epoch": 0.07392628747528368, "grad_norm": 1.2354891800777654, "learning_rate": 1.9899115647977015e-05, "loss": 0.5794, "step": 2075 }, { "epoch": 0.0739619146017778, "grad_norm": 1.765788018971071, "learning_rate": 1.9898952084698418e-05, "loss": 0.6876, "step": 2076 }, { "epoch": 0.0739975417282719, "grad_norm": 1.3124753331451424, "learning_rate": 1.9898788389608356e-05, "loss": 0.552, "step": 2077 }, { "epoch": 0.07403316885476602, "grad_norm": 1.2990612269804596, "learning_rate": 1.9898624562709007e-05, "loss": 0.5679, "step": 2078 }, { "epoch": 0.07406879598126014, "grad_norm": 2.094576361679494, "learning_rate": 1.9898460604002555e-05, "loss": 0.6508, "step": 2079 }, { "epoch": 0.07410442310775424, "grad_norm": 1.2014865100268926, "learning_rate": 1.9898296513491178e-05, "loss": 0.5448, "step": 2080 }, { "epoch": 0.07414005023424836, "grad_norm": 2.0007868251027467, "learning_rate": 1.989813229117707e-05, "loss": 0.6572, "step": 2081 }, { "epoch": 0.07417567736074247, "grad_norm": 1.8830181828956063, "learning_rate": 1.9897967937062407e-05, "loss": 0.6251, "step": 2082 }, { "epoch": 0.07421130448723658, "grad_norm": 2.111623385711302, "learning_rate": 1.9897803451149385e-05, "loss": 0.6706, "step": 2083 }, { "epoch": 0.0742469316137307, "grad_norm": 1.9375872650216621, "learning_rate": 1.989763883344019e-05, "loss": 0.6497, "step": 2084 }, { "epoch": 0.07428255874022481, "grad_norm": 1.5435004421917204, "learning_rate": 1.9897474083937017e-05, "loss": 0.5417, "step": 2085 }, { "epoch": 0.07431818586671891, "grad_norm": 1.8589958934455286, "learning_rate": 1.989730920264206e-05, "loss": 0.6858, "step": 2086 }, { "epoch": 0.07435381299321303, "grad_norm": 1.412374850276621, "learning_rate": 1.9897144189557513e-05, "loss": 0.5382, "step": 2087 }, { "epoch": 0.07438944011970715, "grad_norm": 1.8751252017461133, "learning_rate": 1.989697904468557e-05, "loss": 0.6647, "step": 2088 }, { "epoch": 0.07442506724620125, "grad_norm": 2.717864945345698, "learning_rate": 1.9896813768028437e-05, "loss": 0.6294, "step": 2089 }, { "epoch": 0.07446069437269537, "grad_norm": 1.807602898507065, "learning_rate": 1.9896648359588312e-05, "loss": 0.6393, "step": 2090 }, { "epoch": 0.07449632149918949, "grad_norm": 1.8298084091198121, "learning_rate": 1.9896482819367393e-05, "loss": 0.5959, "step": 2091 }, { "epoch": 0.07453194862568359, "grad_norm": 1.8738851581314118, "learning_rate": 1.989631714736789e-05, "loss": 0.7019, "step": 2092 }, { "epoch": 0.0745675757521777, "grad_norm": 1.9596131584272691, "learning_rate": 1.9896151343592007e-05, "loss": 0.6776, "step": 2093 }, { "epoch": 0.07460320287867182, "grad_norm": 2.858467085955056, "learning_rate": 1.9895985408041954e-05, "loss": 0.6625, "step": 2094 }, { "epoch": 0.07463883000516593, "grad_norm": 1.9355834210213925, "learning_rate": 1.9895819340719934e-05, "loss": 0.5964, "step": 2095 }, { "epoch": 0.07467445713166004, "grad_norm": 2.041037494573311, "learning_rate": 1.9895653141628166e-05, "loss": 0.7086, "step": 2096 }, { "epoch": 0.07471008425815416, "grad_norm": 2.581051197454729, "learning_rate": 1.989548681076886e-05, "loss": 0.6635, "step": 2097 }, { "epoch": 0.07474571138464828, "grad_norm": 1.1955237122929239, "learning_rate": 1.9895320348144226e-05, "loss": 0.5055, "step": 2098 }, { "epoch": 0.07478133851114238, "grad_norm": 2.00320272989668, "learning_rate": 1.989515375375649e-05, "loss": 0.6503, "step": 2099 }, { "epoch": 0.0748169656376365, "grad_norm": 1.3482269317566835, "learning_rate": 1.9894987027607863e-05, "loss": 0.5331, "step": 2100 }, { "epoch": 0.07485259276413062, "grad_norm": 1.3657301876135268, "learning_rate": 1.9894820169700567e-05, "loss": 0.6493, "step": 2101 }, { "epoch": 0.07488821989062472, "grad_norm": 2.0652510779313293, "learning_rate": 1.989465318003683e-05, "loss": 0.6556, "step": 2102 }, { "epoch": 0.07492384701711884, "grad_norm": 1.6734748835765516, "learning_rate": 1.9894486058618863e-05, "loss": 0.6681, "step": 2103 }, { "epoch": 0.07495947414361295, "grad_norm": 1.2079536830196203, "learning_rate": 1.98943188054489e-05, "loss": 0.515, "step": 2104 }, { "epoch": 0.07499510127010706, "grad_norm": 1.4321598637651318, "learning_rate": 1.9894151420529167e-05, "loss": 0.5345, "step": 2105 }, { "epoch": 0.07503072839660117, "grad_norm": 1.562122805550031, "learning_rate": 1.98939839038619e-05, "loss": 0.5319, "step": 2106 }, { "epoch": 0.07506635552309529, "grad_norm": 1.4321221450127626, "learning_rate": 1.989381625544931e-05, "loss": 0.546, "step": 2107 }, { "epoch": 0.0751019826495894, "grad_norm": 1.5720604888885752, "learning_rate": 1.9893648475293646e-05, "loss": 0.5936, "step": 2108 }, { "epoch": 0.07513760977608351, "grad_norm": 1.3486875140754104, "learning_rate": 1.9893480563397135e-05, "loss": 0.5279, "step": 2109 }, { "epoch": 0.07517323690257763, "grad_norm": 1.217526836250924, "learning_rate": 1.989331251976202e-05, "loss": 0.5531, "step": 2110 }, { "epoch": 0.07520886402907173, "grad_norm": 1.970724570918526, "learning_rate": 1.9893144344390533e-05, "loss": 0.6571, "step": 2111 }, { "epoch": 0.07524449115556585, "grad_norm": 2.0017430709142476, "learning_rate": 1.989297603728491e-05, "loss": 0.6756, "step": 2112 }, { "epoch": 0.07528011828205997, "grad_norm": 1.5104091886470483, "learning_rate": 1.9892807598447398e-05, "loss": 0.5803, "step": 2113 }, { "epoch": 0.07531574540855407, "grad_norm": 2.8782999130504243, "learning_rate": 1.989263902788024e-05, "loss": 0.6039, "step": 2114 }, { "epoch": 0.07535137253504819, "grad_norm": 1.3741767433886203, "learning_rate": 1.9892470325585676e-05, "loss": 0.5606, "step": 2115 }, { "epoch": 0.0753869996615423, "grad_norm": 1.2556102615473916, "learning_rate": 1.9892301491565957e-05, "loss": 0.5087, "step": 2116 }, { "epoch": 0.0754226267880364, "grad_norm": 1.3744430474608993, "learning_rate": 1.989213252582333e-05, "loss": 0.5592, "step": 2117 }, { "epoch": 0.07545825391453052, "grad_norm": 1.6745655822318246, "learning_rate": 1.9891963428360043e-05, "loss": 0.6165, "step": 2118 }, { "epoch": 0.07549388104102464, "grad_norm": 1.6226821193601737, "learning_rate": 1.989179419917835e-05, "loss": 0.5142, "step": 2119 }, { "epoch": 0.07552950816751874, "grad_norm": 1.241233933315229, "learning_rate": 1.98916248382805e-05, "loss": 0.539, "step": 2120 }, { "epoch": 0.07556513529401286, "grad_norm": 1.4125978276954574, "learning_rate": 1.9891455345668756e-05, "loss": 0.5639, "step": 2121 }, { "epoch": 0.07560076242050698, "grad_norm": 1.2840137682412782, "learning_rate": 1.9891285721345366e-05, "loss": 0.5363, "step": 2122 }, { "epoch": 0.07563638954700108, "grad_norm": 1.3049684382479947, "learning_rate": 1.98911159653126e-05, "loss": 0.5343, "step": 2123 }, { "epoch": 0.0756720166734952, "grad_norm": 2.832534022608328, "learning_rate": 1.9890946077572702e-05, "loss": 0.6312, "step": 2124 }, { "epoch": 0.07570764379998932, "grad_norm": 1.3076903808172333, "learning_rate": 1.9890776058127948e-05, "loss": 0.5718, "step": 2125 }, { "epoch": 0.07574327092648342, "grad_norm": 1.9495774472261835, "learning_rate": 1.98906059069806e-05, "loss": 0.7058, "step": 2126 }, { "epoch": 0.07577889805297754, "grad_norm": 2.0848149527539053, "learning_rate": 1.9890435624132918e-05, "loss": 0.5961, "step": 2127 }, { "epoch": 0.07581452517947165, "grad_norm": 1.751309693298081, "learning_rate": 1.9890265209587175e-05, "loss": 0.6146, "step": 2128 }, { "epoch": 0.07585015230596576, "grad_norm": 1.9063004016643412, "learning_rate": 1.9890094663345635e-05, "loss": 0.6819, "step": 2129 }, { "epoch": 0.07588577943245987, "grad_norm": 1.3931240884928349, "learning_rate": 1.9889923985410576e-05, "loss": 0.5147, "step": 2130 }, { "epoch": 0.07592140655895399, "grad_norm": 1.8796449421015893, "learning_rate": 1.988975317578426e-05, "loss": 0.7198, "step": 2131 }, { "epoch": 0.0759570336854481, "grad_norm": 1.47493566751899, "learning_rate": 1.9889582234468973e-05, "loss": 0.5933, "step": 2132 }, { "epoch": 0.07599266081194221, "grad_norm": 1.321595063624483, "learning_rate": 1.9889411161466987e-05, "loss": 0.5376, "step": 2133 }, { "epoch": 0.07602828793843633, "grad_norm": 1.5166135467606263, "learning_rate": 1.9889239956780576e-05, "loss": 0.52, "step": 2134 }, { "epoch": 0.07606391506493043, "grad_norm": 1.701165758319299, "learning_rate": 1.9889068620412025e-05, "loss": 0.6272, "step": 2135 }, { "epoch": 0.07609954219142455, "grad_norm": 1.395121399836917, "learning_rate": 1.9888897152363613e-05, "loss": 0.651, "step": 2136 }, { "epoch": 0.07613516931791867, "grad_norm": 1.7562001947813353, "learning_rate": 1.988872555263762e-05, "loss": 0.6256, "step": 2137 }, { "epoch": 0.07617079644441277, "grad_norm": 1.3777688562572967, "learning_rate": 1.9888553821236338e-05, "loss": 0.5599, "step": 2138 }, { "epoch": 0.07620642357090689, "grad_norm": 1.6369668788315703, "learning_rate": 1.988838195816205e-05, "loss": 0.6152, "step": 2139 }, { "epoch": 0.076242050697401, "grad_norm": 1.2955047967658382, "learning_rate": 1.9888209963417045e-05, "loss": 0.5775, "step": 2140 }, { "epoch": 0.07627767782389511, "grad_norm": 1.426470723068985, "learning_rate": 1.988803783700361e-05, "loss": 0.6007, "step": 2141 }, { "epoch": 0.07631330495038922, "grad_norm": 1.8019227471644355, "learning_rate": 1.9887865578924042e-05, "loss": 0.6241, "step": 2142 }, { "epoch": 0.07634893207688334, "grad_norm": 2.0687213358213232, "learning_rate": 1.988769318918063e-05, "loss": 0.6619, "step": 2143 }, { "epoch": 0.07638455920337746, "grad_norm": 3.0717139345037796, "learning_rate": 1.9887520667775677e-05, "loss": 0.6081, "step": 2144 }, { "epoch": 0.07642018632987156, "grad_norm": 1.2780987351785795, "learning_rate": 1.9887348014711475e-05, "loss": 0.5494, "step": 2145 }, { "epoch": 0.07645581345636568, "grad_norm": 1.3727193238784308, "learning_rate": 1.988717522999032e-05, "loss": 0.5621, "step": 2146 }, { "epoch": 0.0764914405828598, "grad_norm": 1.2290704361052696, "learning_rate": 1.9887002313614518e-05, "loss": 0.552, "step": 2147 }, { "epoch": 0.0765270677093539, "grad_norm": 1.816750423246476, "learning_rate": 1.9886829265586368e-05, "loss": 0.6045, "step": 2148 }, { "epoch": 0.07656269483584802, "grad_norm": 2.1531516535503474, "learning_rate": 1.9886656085908176e-05, "loss": 0.6739, "step": 2149 }, { "epoch": 0.07659832196234213, "grad_norm": 1.8612982493243597, "learning_rate": 1.988648277458225e-05, "loss": 0.6408, "step": 2150 }, { "epoch": 0.07663394908883624, "grad_norm": 1.4747203967956666, "learning_rate": 1.9886309331610893e-05, "loss": 0.6093, "step": 2151 }, { "epoch": 0.07666957621533035, "grad_norm": 1.5554609260201386, "learning_rate": 1.988613575699642e-05, "loss": 0.5656, "step": 2152 }, { "epoch": 0.07670520334182447, "grad_norm": 1.7432341782624006, "learning_rate": 1.9885962050741138e-05, "loss": 0.6272, "step": 2153 }, { "epoch": 0.07674083046831857, "grad_norm": 1.1605802078930285, "learning_rate": 1.9885788212847358e-05, "loss": 0.5239, "step": 2154 }, { "epoch": 0.07677645759481269, "grad_norm": 1.3237547568660266, "learning_rate": 1.98856142433174e-05, "loss": 0.5379, "step": 2155 }, { "epoch": 0.07681208472130681, "grad_norm": 1.3150036818585304, "learning_rate": 1.988544014215358e-05, "loss": 0.5925, "step": 2156 }, { "epoch": 0.07684771184780091, "grad_norm": 2.141657869391644, "learning_rate": 1.9885265909358216e-05, "loss": 0.6147, "step": 2157 }, { "epoch": 0.07688333897429503, "grad_norm": 1.2674116515860525, "learning_rate": 1.9885091544933628e-05, "loss": 0.5553, "step": 2158 }, { "epoch": 0.07691896610078915, "grad_norm": 1.9017978571746854, "learning_rate": 1.9884917048882132e-05, "loss": 0.6432, "step": 2159 }, { "epoch": 0.07695459322728325, "grad_norm": 2.122205324371925, "learning_rate": 1.988474242120606e-05, "loss": 0.6702, "step": 2160 }, { "epoch": 0.07699022035377737, "grad_norm": 2.1432952645794634, "learning_rate": 1.9884567661907733e-05, "loss": 0.6257, "step": 2161 }, { "epoch": 0.07702584748027148, "grad_norm": 1.5363778073382741, "learning_rate": 1.988439277098948e-05, "loss": 0.572, "step": 2162 }, { "epoch": 0.07706147460676559, "grad_norm": 1.5031896321959688, "learning_rate": 1.9884217748453625e-05, "loss": 0.5462, "step": 2163 }, { "epoch": 0.0770971017332597, "grad_norm": 1.8997895290074058, "learning_rate": 1.9884042594302503e-05, "loss": 0.655, "step": 2164 }, { "epoch": 0.07713272885975382, "grad_norm": 2.529787887044338, "learning_rate": 1.988386730853845e-05, "loss": 0.6206, "step": 2165 }, { "epoch": 0.07716835598624792, "grad_norm": 1.2943419495417317, "learning_rate": 1.988369189116379e-05, "loss": 0.5388, "step": 2166 }, { "epoch": 0.07720398311274204, "grad_norm": 2.5324013856206182, "learning_rate": 1.9883516342180867e-05, "loss": 0.6003, "step": 2167 }, { "epoch": 0.07723961023923616, "grad_norm": 1.8433645782964345, "learning_rate": 1.9883340661592013e-05, "loss": 0.6705, "step": 2168 }, { "epoch": 0.07727523736573026, "grad_norm": 1.6614962009932048, "learning_rate": 1.9883164849399572e-05, "loss": 0.5881, "step": 2169 }, { "epoch": 0.07731086449222438, "grad_norm": 1.8876673050213826, "learning_rate": 1.9882988905605885e-05, "loss": 0.6221, "step": 2170 }, { "epoch": 0.0773464916187185, "grad_norm": 1.3401962868316029, "learning_rate": 1.988281283021329e-05, "loss": 0.572, "step": 2171 }, { "epoch": 0.0773821187452126, "grad_norm": 1.3555000485089057, "learning_rate": 1.9882636623224135e-05, "loss": 0.5413, "step": 2172 }, { "epoch": 0.07741774587170672, "grad_norm": 1.2268845669354684, "learning_rate": 1.9882460284640762e-05, "loss": 0.5107, "step": 2173 }, { "epoch": 0.07745337299820083, "grad_norm": 2.2474553000356536, "learning_rate": 1.988228381446553e-05, "loss": 0.6367, "step": 2174 }, { "epoch": 0.07748900012469494, "grad_norm": 1.359485321784934, "learning_rate": 1.9882107212700778e-05, "loss": 0.5382, "step": 2175 }, { "epoch": 0.07752462725118905, "grad_norm": 1.2479265467111302, "learning_rate": 1.9881930479348863e-05, "loss": 0.5276, "step": 2176 }, { "epoch": 0.07756025437768317, "grad_norm": 1.1929899087973537, "learning_rate": 1.9881753614412135e-05, "loss": 0.5084, "step": 2177 }, { "epoch": 0.07759588150417727, "grad_norm": 1.269122313350085, "learning_rate": 1.988157661789295e-05, "loss": 0.5038, "step": 2178 }, { "epoch": 0.07763150863067139, "grad_norm": 1.91739151262742, "learning_rate": 1.9881399489793667e-05, "loss": 0.6626, "step": 2179 }, { "epoch": 0.07766713575716551, "grad_norm": 1.2325336713070163, "learning_rate": 1.9881222230116645e-05, "loss": 0.5488, "step": 2180 }, { "epoch": 0.07770276288365961, "grad_norm": 1.3562139988909803, "learning_rate": 1.9881044838864238e-05, "loss": 0.5333, "step": 2181 }, { "epoch": 0.07773839001015373, "grad_norm": 2.116655444921019, "learning_rate": 1.9880867316038815e-05, "loss": 0.5728, "step": 2182 }, { "epoch": 0.07777401713664785, "grad_norm": 1.360191190143699, "learning_rate": 1.988068966164274e-05, "loss": 0.5482, "step": 2183 }, { "epoch": 0.07780964426314195, "grad_norm": 2.8555899433857275, "learning_rate": 1.988051187567837e-05, "loss": 0.6852, "step": 2184 }, { "epoch": 0.07784527138963607, "grad_norm": 2.3119297248802493, "learning_rate": 1.9880333958148085e-05, "loss": 0.5687, "step": 2185 }, { "epoch": 0.07788089851613018, "grad_norm": 1.3227275236305567, "learning_rate": 1.988015590905424e-05, "loss": 0.5552, "step": 2186 }, { "epoch": 0.07791652564262429, "grad_norm": 1.513881687588521, "learning_rate": 1.9879977728399217e-05, "loss": 0.5747, "step": 2187 }, { "epoch": 0.0779521527691184, "grad_norm": 1.4926532323047579, "learning_rate": 1.9879799416185386e-05, "loss": 0.5743, "step": 2188 }, { "epoch": 0.07798777989561252, "grad_norm": 1.6882061529071433, "learning_rate": 1.9879620972415118e-05, "loss": 0.6196, "step": 2189 }, { "epoch": 0.07802340702210664, "grad_norm": 2.2039997364865873, "learning_rate": 1.9879442397090797e-05, "loss": 0.6179, "step": 2190 }, { "epoch": 0.07805903414860074, "grad_norm": 1.2590888138117198, "learning_rate": 1.987926369021479e-05, "loss": 0.5467, "step": 2191 }, { "epoch": 0.07809466127509486, "grad_norm": 1.396142550355384, "learning_rate": 1.9879084851789478e-05, "loss": 0.5545, "step": 2192 }, { "epoch": 0.07813028840158898, "grad_norm": 1.3067575141878092, "learning_rate": 1.9878905881817254e-05, "loss": 0.5429, "step": 2193 }, { "epoch": 0.07816591552808308, "grad_norm": 1.4699912058304216, "learning_rate": 1.987872678030049e-05, "loss": 0.5828, "step": 2194 }, { "epoch": 0.0782015426545772, "grad_norm": 1.3654012200181669, "learning_rate": 1.987854754724157e-05, "loss": 0.5906, "step": 2195 }, { "epoch": 0.07823716978107131, "grad_norm": 1.9945897017560692, "learning_rate": 1.987836818264289e-05, "loss": 0.5413, "step": 2196 }, { "epoch": 0.07827279690756542, "grad_norm": 1.1922395029746309, "learning_rate": 1.9878188686506834e-05, "loss": 0.5189, "step": 2197 }, { "epoch": 0.07830842403405953, "grad_norm": 1.364435720396074, "learning_rate": 1.987800905883579e-05, "loss": 0.5693, "step": 2198 }, { "epoch": 0.07834405116055365, "grad_norm": 1.3927964542630442, "learning_rate": 1.9877829299632146e-05, "loss": 0.5467, "step": 2199 }, { "epoch": 0.07837967828704775, "grad_norm": 2.3089332786674484, "learning_rate": 1.9877649408898306e-05, "loss": 0.6005, "step": 2200 }, { "epoch": 0.07841530541354187, "grad_norm": 1.4203851256695785, "learning_rate": 1.9877469386636658e-05, "loss": 0.522, "step": 2201 }, { "epoch": 0.07845093254003599, "grad_norm": 2.0006419406232006, "learning_rate": 1.9877289232849605e-05, "loss": 0.7224, "step": 2202 }, { "epoch": 0.07848655966653009, "grad_norm": 1.7769201198043387, "learning_rate": 1.9877108947539537e-05, "loss": 0.6808, "step": 2203 }, { "epoch": 0.07852218679302421, "grad_norm": 2.0636308186401027, "learning_rate": 1.9876928530708862e-05, "loss": 0.6389, "step": 2204 }, { "epoch": 0.07855781391951833, "grad_norm": 2.041112284986588, "learning_rate": 1.987674798235998e-05, "loss": 0.6393, "step": 2205 }, { "epoch": 0.07859344104601243, "grad_norm": 1.3021809415045051, "learning_rate": 1.9876567302495295e-05, "loss": 0.5617, "step": 2206 }, { "epoch": 0.07862906817250655, "grad_norm": 1.808352329079325, "learning_rate": 1.987638649111721e-05, "loss": 0.5499, "step": 2207 }, { "epoch": 0.07866469529900066, "grad_norm": 1.3001404840667847, "learning_rate": 1.987620554822814e-05, "loss": 0.5391, "step": 2208 }, { "epoch": 0.07870032242549477, "grad_norm": 1.2570655878892447, "learning_rate": 1.9876024473830488e-05, "loss": 0.5985, "step": 2209 }, { "epoch": 0.07873594955198888, "grad_norm": 1.3406610625456974, "learning_rate": 1.987584326792667e-05, "loss": 0.5188, "step": 2210 }, { "epoch": 0.078771576678483, "grad_norm": 2.517419885485161, "learning_rate": 1.9875661930519093e-05, "loss": 0.5139, "step": 2211 }, { "epoch": 0.0788072038049771, "grad_norm": 1.1240479880431062, "learning_rate": 1.9875480461610176e-05, "loss": 0.538, "step": 2212 }, { "epoch": 0.07884283093147122, "grad_norm": 1.877124380824623, "learning_rate": 1.9875298861202332e-05, "loss": 0.6509, "step": 2213 }, { "epoch": 0.07887845805796534, "grad_norm": 1.16607603808456, "learning_rate": 1.9875117129297986e-05, "loss": 0.5911, "step": 2214 }, { "epoch": 0.07891408518445944, "grad_norm": 1.9950881138659982, "learning_rate": 1.987493526589955e-05, "loss": 0.6856, "step": 2215 }, { "epoch": 0.07894971231095356, "grad_norm": 1.8392289534580786, "learning_rate": 1.9874753271009445e-05, "loss": 0.7097, "step": 2216 }, { "epoch": 0.07898533943744768, "grad_norm": 1.610200306758983, "learning_rate": 1.9874571144630105e-05, "loss": 0.5497, "step": 2217 }, { "epoch": 0.07902096656394178, "grad_norm": 1.3798695654180473, "learning_rate": 1.9874388886763944e-05, "loss": 0.5462, "step": 2218 }, { "epoch": 0.0790565936904359, "grad_norm": 1.3462251667692986, "learning_rate": 1.9874206497413397e-05, "loss": 0.5416, "step": 2219 }, { "epoch": 0.07909222081693001, "grad_norm": 1.3781423399326789, "learning_rate": 1.9874023976580886e-05, "loss": 0.5169, "step": 2220 }, { "epoch": 0.07912784794342412, "grad_norm": 1.3558097967756886, "learning_rate": 1.9873841324268842e-05, "loss": 0.5427, "step": 2221 }, { "epoch": 0.07916347506991823, "grad_norm": 3.2407797971603167, "learning_rate": 1.9873658540479703e-05, "loss": 0.6134, "step": 2222 }, { "epoch": 0.07919910219641235, "grad_norm": 1.4956680458557001, "learning_rate": 1.98734756252159e-05, "loss": 0.5332, "step": 2223 }, { "epoch": 0.07923472932290646, "grad_norm": 1.4569526936135742, "learning_rate": 1.9873292578479867e-05, "loss": 0.599, "step": 2224 }, { "epoch": 0.07927035644940057, "grad_norm": 1.2907386900638647, "learning_rate": 1.987310940027404e-05, "loss": 0.5069, "step": 2225 }, { "epoch": 0.07930598357589469, "grad_norm": 2.1398699990290146, "learning_rate": 1.9872926090600862e-05, "loss": 0.6772, "step": 2226 }, { "epoch": 0.07934161070238879, "grad_norm": 1.3700571431773259, "learning_rate": 1.9872742649462772e-05, "loss": 0.5532, "step": 2227 }, { "epoch": 0.07937723782888291, "grad_norm": 1.7357744297692834, "learning_rate": 1.9872559076862214e-05, "loss": 0.7155, "step": 2228 }, { "epoch": 0.07941286495537703, "grad_norm": 3.6362003633250675, "learning_rate": 1.9872375372801627e-05, "loss": 0.61, "step": 2229 }, { "epoch": 0.07944849208187113, "grad_norm": 1.3481678323959958, "learning_rate": 1.9872191537283466e-05, "loss": 0.5072, "step": 2230 }, { "epoch": 0.07948411920836525, "grad_norm": 2.0071245169501672, "learning_rate": 1.9872007570310175e-05, "loss": 0.6251, "step": 2231 }, { "epoch": 0.07951974633485936, "grad_norm": 1.9009567921377455, "learning_rate": 1.98718234718842e-05, "loss": 0.614, "step": 2232 }, { "epoch": 0.07955537346135347, "grad_norm": 2.161838723944382, "learning_rate": 1.9871639242007997e-05, "loss": 0.6509, "step": 2233 }, { "epoch": 0.07959100058784759, "grad_norm": 1.9787184315474233, "learning_rate": 1.9871454880684016e-05, "loss": 0.6216, "step": 2234 }, { "epoch": 0.0796266277143417, "grad_norm": 1.3579115295557873, "learning_rate": 1.987127038791472e-05, "loss": 0.597, "step": 2235 }, { "epoch": 0.0796622548408358, "grad_norm": 2.1916719870928243, "learning_rate": 1.9871085763702553e-05, "loss": 0.5833, "step": 2236 }, { "epoch": 0.07969788196732992, "grad_norm": 2.4856211878066636, "learning_rate": 1.987090100804998e-05, "loss": 0.6828, "step": 2237 }, { "epoch": 0.07973350909382404, "grad_norm": 1.2952165826330824, "learning_rate": 1.9870716120959462e-05, "loss": 0.5649, "step": 2238 }, { "epoch": 0.07976913622031816, "grad_norm": 1.5609533792438701, "learning_rate": 1.987053110243346e-05, "loss": 0.6149, "step": 2239 }, { "epoch": 0.07980476334681226, "grad_norm": 2.5248193434330397, "learning_rate": 1.9870345952474436e-05, "loss": 0.6685, "step": 2240 }, { "epoch": 0.07984039047330638, "grad_norm": 1.9374694389358147, "learning_rate": 1.9870160671084857e-05, "loss": 0.6824, "step": 2241 }, { "epoch": 0.0798760175998005, "grad_norm": 1.9493988714875279, "learning_rate": 1.9869975258267192e-05, "loss": 0.6803, "step": 2242 }, { "epoch": 0.0799116447262946, "grad_norm": 1.2308741757131094, "learning_rate": 1.9869789714023906e-05, "loss": 0.4997, "step": 2243 }, { "epoch": 0.07994727185278871, "grad_norm": 1.3314456856018908, "learning_rate": 1.986960403835747e-05, "loss": 0.5253, "step": 2244 }, { "epoch": 0.07998289897928283, "grad_norm": 2.2286116344937232, "learning_rate": 1.9869418231270358e-05, "loss": 0.6256, "step": 2245 }, { "epoch": 0.08001852610577694, "grad_norm": 2.2708655564292886, "learning_rate": 1.9869232292765046e-05, "loss": 0.5426, "step": 2246 }, { "epoch": 0.08005415323227105, "grad_norm": 1.3463034650466292, "learning_rate": 1.986904622284401e-05, "loss": 0.5807, "step": 2247 }, { "epoch": 0.08008978035876517, "grad_norm": 1.8380493780521028, "learning_rate": 1.986886002150972e-05, "loss": 0.6782, "step": 2248 }, { "epoch": 0.08012540748525927, "grad_norm": 2.1600856892609404, "learning_rate": 1.9868673688764663e-05, "loss": 0.6418, "step": 2249 }, { "epoch": 0.08016103461175339, "grad_norm": 1.386832077036047, "learning_rate": 1.986848722461132e-05, "loss": 0.5647, "step": 2250 }, { "epoch": 0.08019666173824751, "grad_norm": 1.9803188308871713, "learning_rate": 1.986830062905217e-05, "loss": 0.6591, "step": 2251 }, { "epoch": 0.08023228886474161, "grad_norm": 2.441777503949099, "learning_rate": 1.9868113902089702e-05, "loss": 0.6653, "step": 2252 }, { "epoch": 0.08026791599123573, "grad_norm": 1.4055800807727459, "learning_rate": 1.9867927043726393e-05, "loss": 0.6055, "step": 2253 }, { "epoch": 0.08030354311772984, "grad_norm": 1.7213551806242027, "learning_rate": 1.9867740053964746e-05, "loss": 0.6251, "step": 2254 }, { "epoch": 0.08033917024422395, "grad_norm": 1.3836811425178597, "learning_rate": 1.986755293280724e-05, "loss": 0.5552, "step": 2255 }, { "epoch": 0.08037479737071807, "grad_norm": 1.99082561651056, "learning_rate": 1.986736568025637e-05, "loss": 0.5627, "step": 2256 }, { "epoch": 0.08041042449721218, "grad_norm": 1.3484837960150475, "learning_rate": 1.9867178296314627e-05, "loss": 0.5689, "step": 2257 }, { "epoch": 0.08044605162370629, "grad_norm": 1.2174577525870307, "learning_rate": 1.986699078098451e-05, "loss": 0.4687, "step": 2258 }, { "epoch": 0.0804816787502004, "grad_norm": 1.4505851244212344, "learning_rate": 1.9866803134268514e-05, "loss": 0.6071, "step": 2259 }, { "epoch": 0.08051730587669452, "grad_norm": 2.048584626540625, "learning_rate": 1.986661535616914e-05, "loss": 0.5979, "step": 2260 }, { "epoch": 0.08055293300318862, "grad_norm": 1.3164648324481818, "learning_rate": 1.9866427446688883e-05, "loss": 0.539, "step": 2261 }, { "epoch": 0.08058856012968274, "grad_norm": 1.6333196675828416, "learning_rate": 1.986623940583025e-05, "loss": 0.6115, "step": 2262 }, { "epoch": 0.08062418725617686, "grad_norm": 1.3274185163016508, "learning_rate": 1.986605123359574e-05, "loss": 0.5672, "step": 2263 }, { "epoch": 0.08065981438267096, "grad_norm": 1.2445617347936344, "learning_rate": 1.9865862929987866e-05, "loss": 0.5906, "step": 2264 }, { "epoch": 0.08069544150916508, "grad_norm": 1.368005712201304, "learning_rate": 1.986567449500913e-05, "loss": 0.5575, "step": 2265 }, { "epoch": 0.0807310686356592, "grad_norm": 2.2610922656373185, "learning_rate": 1.986548592866204e-05, "loss": 0.6588, "step": 2266 }, { "epoch": 0.0807666957621533, "grad_norm": 1.9941490495277028, "learning_rate": 1.9865297230949113e-05, "loss": 0.624, "step": 2267 }, { "epoch": 0.08080232288864742, "grad_norm": 1.9470685397062732, "learning_rate": 1.9865108401872856e-05, "loss": 0.6998, "step": 2268 }, { "epoch": 0.08083795001514153, "grad_norm": 1.7698709391574345, "learning_rate": 1.986491944143579e-05, "loss": 0.6401, "step": 2269 }, { "epoch": 0.08087357714163564, "grad_norm": 2.0026606833112544, "learning_rate": 1.986473034964042e-05, "loss": 0.566, "step": 2270 }, { "epoch": 0.08090920426812975, "grad_norm": 1.2795852227465447, "learning_rate": 1.9864541126489275e-05, "loss": 0.5623, "step": 2271 }, { "epoch": 0.08094483139462387, "grad_norm": 1.852001063383119, "learning_rate": 1.9864351771984864e-05, "loss": 0.7113, "step": 2272 }, { "epoch": 0.08098045852111797, "grad_norm": 3.5736299752327576, "learning_rate": 1.986416228612972e-05, "loss": 0.6574, "step": 2273 }, { "epoch": 0.08101608564761209, "grad_norm": 1.375300096242031, "learning_rate": 1.986397266892636e-05, "loss": 0.6107, "step": 2274 }, { "epoch": 0.08105171277410621, "grad_norm": 1.336673482901467, "learning_rate": 1.9863782920377305e-05, "loss": 0.562, "step": 2275 }, { "epoch": 0.08108733990060031, "grad_norm": 1.2752102880576923, "learning_rate": 1.986359304048509e-05, "loss": 0.5539, "step": 2276 }, { "epoch": 0.08112296702709443, "grad_norm": 2.013321187175899, "learning_rate": 1.986340302925224e-05, "loss": 0.6522, "step": 2277 }, { "epoch": 0.08115859415358855, "grad_norm": 1.1497373236976096, "learning_rate": 1.986321288668128e-05, "loss": 0.4702, "step": 2278 }, { "epoch": 0.08119422128008265, "grad_norm": 1.2779128948481238, "learning_rate": 1.986302261277475e-05, "loss": 0.5022, "step": 2279 }, { "epoch": 0.08122984840657677, "grad_norm": 3.016120066962818, "learning_rate": 1.986283220753518e-05, "loss": 0.6472, "step": 2280 }, { "epoch": 0.08126547553307088, "grad_norm": 1.3336602882632, "learning_rate": 1.9862641670965105e-05, "loss": 0.5761, "step": 2281 }, { "epoch": 0.08130110265956499, "grad_norm": 2.1548998599823497, "learning_rate": 1.9862451003067063e-05, "loss": 0.6426, "step": 2282 }, { "epoch": 0.0813367297860591, "grad_norm": 1.3728009248596724, "learning_rate": 1.986226020384359e-05, "loss": 0.6156, "step": 2283 }, { "epoch": 0.08137235691255322, "grad_norm": 1.2860586178037894, "learning_rate": 1.9862069273297233e-05, "loss": 0.5289, "step": 2284 }, { "epoch": 0.08140798403904734, "grad_norm": 2.081675631281151, "learning_rate": 1.986187821143053e-05, "loss": 0.688, "step": 2285 }, { "epoch": 0.08144361116554144, "grad_norm": 1.319596277251789, "learning_rate": 1.9861687018246022e-05, "loss": 0.5382, "step": 2286 }, { "epoch": 0.08147923829203556, "grad_norm": 1.7891248801664652, "learning_rate": 1.986149569374626e-05, "loss": 0.5464, "step": 2287 }, { "epoch": 0.08151486541852968, "grad_norm": 2.6955969545490763, "learning_rate": 1.986130423793379e-05, "loss": 0.6528, "step": 2288 }, { "epoch": 0.08155049254502378, "grad_norm": 1.2450911821641648, "learning_rate": 1.9861112650811166e-05, "loss": 0.518, "step": 2289 }, { "epoch": 0.0815861196715179, "grad_norm": 1.8275141221524658, "learning_rate": 1.986092093238093e-05, "loss": 0.5798, "step": 2290 }, { "epoch": 0.08162174679801201, "grad_norm": 1.4284859683927462, "learning_rate": 1.986072908264564e-05, "loss": 0.5341, "step": 2291 }, { "epoch": 0.08165737392450612, "grad_norm": 1.5317004617138514, "learning_rate": 1.9860537101607848e-05, "loss": 0.6357, "step": 2292 }, { "epoch": 0.08169300105100023, "grad_norm": 1.3229434355909007, "learning_rate": 1.9860344989270115e-05, "loss": 0.6262, "step": 2293 }, { "epoch": 0.08172862817749435, "grad_norm": 1.309575342383908, "learning_rate": 1.9860152745634994e-05, "loss": 0.5988, "step": 2294 }, { "epoch": 0.08176425530398845, "grad_norm": 1.586210523042395, "learning_rate": 1.985996037070505e-05, "loss": 0.6791, "step": 2295 }, { "epoch": 0.08179988243048257, "grad_norm": 2.0967042956670827, "learning_rate": 1.985976786448284e-05, "loss": 0.6211, "step": 2296 }, { "epoch": 0.08183550955697669, "grad_norm": 1.7318792261923266, "learning_rate": 1.9859575226970932e-05, "loss": 0.659, "step": 2297 }, { "epoch": 0.08187113668347079, "grad_norm": 1.9377418620862978, "learning_rate": 1.9859382458171887e-05, "loss": 0.6347, "step": 2298 }, { "epoch": 0.08190676380996491, "grad_norm": 1.3389799817967734, "learning_rate": 1.9859189558088275e-05, "loss": 0.5385, "step": 2299 }, { "epoch": 0.08194239093645903, "grad_norm": 2.027405401067189, "learning_rate": 1.985899652672266e-05, "loss": 0.6662, "step": 2300 }, { "epoch": 0.08197801806295313, "grad_norm": 1.949700911873416, "learning_rate": 1.9858803364077616e-05, "loss": 0.6438, "step": 2301 }, { "epoch": 0.08201364518944725, "grad_norm": 1.3106771496702765, "learning_rate": 1.9858610070155715e-05, "loss": 0.4989, "step": 2302 }, { "epoch": 0.08204927231594136, "grad_norm": 1.765055388882481, "learning_rate": 1.985841664495953e-05, "loss": 0.6468, "step": 2303 }, { "epoch": 0.08208489944243547, "grad_norm": 2.2639638018107875, "learning_rate": 1.9858223088491635e-05, "loss": 0.6801, "step": 2304 }, { "epoch": 0.08212052656892958, "grad_norm": 1.858591742382216, "learning_rate": 1.9858029400754613e-05, "loss": 0.5679, "step": 2305 }, { "epoch": 0.0821561536954237, "grad_norm": 1.270952833289561, "learning_rate": 1.9857835581751038e-05, "loss": 0.5346, "step": 2306 }, { "epoch": 0.0821917808219178, "grad_norm": 1.6870709467027332, "learning_rate": 1.985764163148349e-05, "loss": 0.6256, "step": 2307 }, { "epoch": 0.08222740794841192, "grad_norm": 1.3789709248260706, "learning_rate": 1.985744754995455e-05, "loss": 0.5581, "step": 2308 }, { "epoch": 0.08226303507490604, "grad_norm": 1.4031307638320556, "learning_rate": 1.9857253337166813e-05, "loss": 0.5303, "step": 2309 }, { "epoch": 0.08229866220140014, "grad_norm": 2.180476802562191, "learning_rate": 1.9857058993122856e-05, "loss": 0.6117, "step": 2310 }, { "epoch": 0.08233428932789426, "grad_norm": 2.2097509823196253, "learning_rate": 1.985686451782527e-05, "loss": 0.6097, "step": 2311 }, { "epoch": 0.08236991645438838, "grad_norm": 1.918411888867366, "learning_rate": 1.9856669911276637e-05, "loss": 0.7065, "step": 2312 }, { "epoch": 0.08240554358088248, "grad_norm": 1.8662351553277052, "learning_rate": 1.9856475173479565e-05, "loss": 0.6338, "step": 2313 }, { "epoch": 0.0824411707073766, "grad_norm": 1.526077001849758, "learning_rate": 1.9856280304436628e-05, "loss": 0.6195, "step": 2314 }, { "epoch": 0.08247679783387071, "grad_norm": 1.2705939445664933, "learning_rate": 1.9856085304150433e-05, "loss": 0.5446, "step": 2315 }, { "epoch": 0.08251242496036482, "grad_norm": 2.2441031994483414, "learning_rate": 1.9855890172623573e-05, "loss": 0.6102, "step": 2316 }, { "epoch": 0.08254805208685893, "grad_norm": 1.3555067278657973, "learning_rate": 1.9855694909858645e-05, "loss": 0.5432, "step": 2317 }, { "epoch": 0.08258367921335305, "grad_norm": 1.964271022851751, "learning_rate": 1.985549951585825e-05, "loss": 0.6522, "step": 2318 }, { "epoch": 0.08261930633984715, "grad_norm": 1.2631586240720953, "learning_rate": 1.9855303990624996e-05, "loss": 0.5022, "step": 2319 }, { "epoch": 0.08265493346634127, "grad_norm": 1.3832457562220795, "learning_rate": 1.9855108334161478e-05, "loss": 0.5245, "step": 2320 }, { "epoch": 0.08269056059283539, "grad_norm": 3.6558627457051793, "learning_rate": 1.9854912546470304e-05, "loss": 0.7152, "step": 2321 }, { "epoch": 0.08272618771932949, "grad_norm": 1.2887832948220719, "learning_rate": 1.985471662755408e-05, "loss": 0.5461, "step": 2322 }, { "epoch": 0.08276181484582361, "grad_norm": 1.3201925625697306, "learning_rate": 1.9854520577415414e-05, "loss": 0.6228, "step": 2323 }, { "epoch": 0.08279744197231773, "grad_norm": 1.4587302155726591, "learning_rate": 1.9854324396056922e-05, "loss": 0.5236, "step": 2324 }, { "epoch": 0.08283306909881183, "grad_norm": 2.589160465445009, "learning_rate": 1.9854128083481216e-05, "loss": 0.6788, "step": 2325 }, { "epoch": 0.08286869622530595, "grad_norm": 1.2358281686970964, "learning_rate": 1.98539316396909e-05, "loss": 0.5142, "step": 2326 }, { "epoch": 0.08290432335180006, "grad_norm": 1.416058404399762, "learning_rate": 1.9853735064688603e-05, "loss": 0.5709, "step": 2327 }, { "epoch": 0.08293995047829417, "grad_norm": 1.1979194913775486, "learning_rate": 1.9853538358476933e-05, "loss": 0.4925, "step": 2328 }, { "epoch": 0.08297557760478828, "grad_norm": 1.225175944973957, "learning_rate": 1.985334152105851e-05, "loss": 0.5434, "step": 2329 }, { "epoch": 0.0830112047312824, "grad_norm": 1.2480810642761169, "learning_rate": 1.985314455243596e-05, "loss": 0.5372, "step": 2330 }, { "epoch": 0.0830468318577765, "grad_norm": 1.4159305301871228, "learning_rate": 1.9852947452611906e-05, "loss": 0.5395, "step": 2331 }, { "epoch": 0.08308245898427062, "grad_norm": 1.226316196762255, "learning_rate": 1.9852750221588966e-05, "loss": 0.5331, "step": 2332 }, { "epoch": 0.08311808611076474, "grad_norm": 1.2871806230115754, "learning_rate": 1.9852552859369774e-05, "loss": 0.575, "step": 2333 }, { "epoch": 0.08315371323725886, "grad_norm": 1.4022565988421742, "learning_rate": 1.985235536595695e-05, "loss": 0.6024, "step": 2334 }, { "epoch": 0.08318934036375296, "grad_norm": 1.2176992798340254, "learning_rate": 1.985215774135313e-05, "loss": 0.5833, "step": 2335 }, { "epoch": 0.08322496749024708, "grad_norm": 1.3445232136060115, "learning_rate": 1.9851959985560944e-05, "loss": 0.5338, "step": 2336 }, { "epoch": 0.0832605946167412, "grad_norm": 2.048692246764697, "learning_rate": 1.9851762098583025e-05, "loss": 0.6204, "step": 2337 }, { "epoch": 0.0832962217432353, "grad_norm": 2.213408155966516, "learning_rate": 1.9851564080422007e-05, "loss": 0.615, "step": 2338 }, { "epoch": 0.08333184886972941, "grad_norm": 2.234422149532979, "learning_rate": 1.9851365931080525e-05, "loss": 0.6375, "step": 2339 }, { "epoch": 0.08336747599622353, "grad_norm": 1.4613124719633575, "learning_rate": 1.9851167650561223e-05, "loss": 0.5471, "step": 2340 }, { "epoch": 0.08340310312271763, "grad_norm": 2.430178860268733, "learning_rate": 1.985096923886674e-05, "loss": 0.7057, "step": 2341 }, { "epoch": 0.08343873024921175, "grad_norm": 1.131432083240623, "learning_rate": 1.985077069599971e-05, "loss": 0.519, "step": 2342 }, { "epoch": 0.08347435737570587, "grad_norm": 1.2755720277176805, "learning_rate": 1.9850572021962788e-05, "loss": 0.5429, "step": 2343 }, { "epoch": 0.08350998450219997, "grad_norm": 1.2109507876454928, "learning_rate": 1.9850373216758613e-05, "loss": 0.4763, "step": 2344 }, { "epoch": 0.08354561162869409, "grad_norm": 1.816878177587193, "learning_rate": 1.985017428038983e-05, "loss": 0.6412, "step": 2345 }, { "epoch": 0.0835812387551882, "grad_norm": 1.829284718201371, "learning_rate": 1.9849975212859095e-05, "loss": 0.6437, "step": 2346 }, { "epoch": 0.08361686588168231, "grad_norm": 2.0804393277385245, "learning_rate": 1.9849776014169054e-05, "loss": 0.6259, "step": 2347 }, { "epoch": 0.08365249300817643, "grad_norm": 1.3344810149788309, "learning_rate": 1.9849576684322365e-05, "loss": 0.5347, "step": 2348 }, { "epoch": 0.08368812013467054, "grad_norm": 1.3227446636437095, "learning_rate": 1.9849377223321674e-05, "loss": 0.5576, "step": 2349 }, { "epoch": 0.08372374726116465, "grad_norm": 1.8009575171452818, "learning_rate": 1.9849177631169643e-05, "loss": 0.6535, "step": 2350 }, { "epoch": 0.08375937438765876, "grad_norm": 2.2607838292295424, "learning_rate": 1.9848977907868927e-05, "loss": 0.6173, "step": 2351 }, { "epoch": 0.08379500151415288, "grad_norm": 1.908528580172033, "learning_rate": 1.9848778053422184e-05, "loss": 0.6774, "step": 2352 }, { "epoch": 0.08383062864064698, "grad_norm": 1.3028855316463566, "learning_rate": 1.984857806783208e-05, "loss": 0.5003, "step": 2353 }, { "epoch": 0.0838662557671411, "grad_norm": 1.9213041082128863, "learning_rate": 1.9848377951101274e-05, "loss": 0.6052, "step": 2354 }, { "epoch": 0.08390188289363522, "grad_norm": 2.476300910358312, "learning_rate": 1.9848177703232433e-05, "loss": 0.7436, "step": 2355 }, { "epoch": 0.08393751002012932, "grad_norm": 1.3101697031958814, "learning_rate": 1.984797732422822e-05, "loss": 0.5678, "step": 2356 }, { "epoch": 0.08397313714662344, "grad_norm": 1.3216378573885599, "learning_rate": 1.984777681409131e-05, "loss": 0.5434, "step": 2357 }, { "epoch": 0.08400876427311756, "grad_norm": 1.2528041463117001, "learning_rate": 1.9847576172824365e-05, "loss": 0.5325, "step": 2358 }, { "epoch": 0.08404439139961166, "grad_norm": 1.228235201080195, "learning_rate": 1.984737540043006e-05, "loss": 0.5356, "step": 2359 }, { "epoch": 0.08408001852610578, "grad_norm": 1.7709213325498587, "learning_rate": 1.984717449691107e-05, "loss": 0.6305, "step": 2360 }, { "epoch": 0.0841156456525999, "grad_norm": 1.2625209756475912, "learning_rate": 1.984697346227007e-05, "loss": 0.5264, "step": 2361 }, { "epoch": 0.084151272779094, "grad_norm": 1.270331563389559, "learning_rate": 1.9846772296509736e-05, "loss": 0.5226, "step": 2362 }, { "epoch": 0.08418689990558811, "grad_norm": 1.4008916336174793, "learning_rate": 1.984657099963274e-05, "loss": 0.5543, "step": 2363 }, { "epoch": 0.08422252703208223, "grad_norm": 2.2978001191552315, "learning_rate": 1.9846369571641776e-05, "loss": 0.6016, "step": 2364 }, { "epoch": 0.08425815415857633, "grad_norm": 1.3187695658137228, "learning_rate": 1.984616801253952e-05, "loss": 0.5089, "step": 2365 }, { "epoch": 0.08429378128507045, "grad_norm": 2.191811163528017, "learning_rate": 1.9845966322328647e-05, "loss": 0.6496, "step": 2366 }, { "epoch": 0.08432940841156457, "grad_norm": 1.9576215254164226, "learning_rate": 1.9845764501011854e-05, "loss": 0.5611, "step": 2367 }, { "epoch": 0.08436503553805867, "grad_norm": 1.3934208484428519, "learning_rate": 1.9845562548591826e-05, "loss": 0.5121, "step": 2368 }, { "epoch": 0.08440066266455279, "grad_norm": 1.852394363657411, "learning_rate": 1.984536046507125e-05, "loss": 0.6453, "step": 2369 }, { "epoch": 0.0844362897910469, "grad_norm": 1.4377595577442486, "learning_rate": 1.9845158250452822e-05, "loss": 0.5571, "step": 2370 }, { "epoch": 0.08447191691754101, "grad_norm": 2.3300872596075966, "learning_rate": 1.9844955904739226e-05, "loss": 0.6655, "step": 2371 }, { "epoch": 0.08450754404403513, "grad_norm": 1.2134534398273271, "learning_rate": 1.9844753427933163e-05, "loss": 0.4897, "step": 2372 }, { "epoch": 0.08454317117052924, "grad_norm": 1.3057183309003777, "learning_rate": 1.9844550820037326e-05, "loss": 0.5507, "step": 2373 }, { "epoch": 0.08457879829702335, "grad_norm": 1.1081174507337674, "learning_rate": 1.9844348081054413e-05, "loss": 0.4713, "step": 2374 }, { "epoch": 0.08461442542351746, "grad_norm": 2.15120218206985, "learning_rate": 1.9844145210987123e-05, "loss": 0.7205, "step": 2375 }, { "epoch": 0.08465005255001158, "grad_norm": 1.3791707081565718, "learning_rate": 1.9843942209838165e-05, "loss": 0.5357, "step": 2376 }, { "epoch": 0.08468567967650568, "grad_norm": 1.3560218049821722, "learning_rate": 1.9843739077610232e-05, "loss": 0.5404, "step": 2377 }, { "epoch": 0.0847213068029998, "grad_norm": 1.1949790364570083, "learning_rate": 1.9843535814306033e-05, "loss": 0.531, "step": 2378 }, { "epoch": 0.08475693392949392, "grad_norm": 1.7169560040535958, "learning_rate": 1.9843332419928277e-05, "loss": 0.682, "step": 2379 }, { "epoch": 0.08479256105598804, "grad_norm": 1.6251155253452854, "learning_rate": 1.9843128894479667e-05, "loss": 0.6689, "step": 2380 }, { "epoch": 0.08482818818248214, "grad_norm": 1.9256856685792856, "learning_rate": 1.9842925237962913e-05, "loss": 0.6153, "step": 2381 }, { "epoch": 0.08486381530897626, "grad_norm": 2.0033857640290487, "learning_rate": 1.9842721450380737e-05, "loss": 0.6714, "step": 2382 }, { "epoch": 0.08489944243547037, "grad_norm": 1.5066619634341878, "learning_rate": 1.9842517531735837e-05, "loss": 0.5668, "step": 2383 }, { "epoch": 0.08493506956196448, "grad_norm": 1.7117136162707665, "learning_rate": 1.9842313482030943e-05, "loss": 0.6173, "step": 2384 }, { "epoch": 0.0849706966884586, "grad_norm": 1.2508931348521923, "learning_rate": 1.984210930126876e-05, "loss": 0.5239, "step": 2385 }, { "epoch": 0.08500632381495271, "grad_norm": 2.9277122959384214, "learning_rate": 1.9841904989452018e-05, "loss": 0.6969, "step": 2386 }, { "epoch": 0.08504195094144681, "grad_norm": 2.0998009658520824, "learning_rate": 1.9841700546583432e-05, "loss": 0.6204, "step": 2387 }, { "epoch": 0.08507757806794093, "grad_norm": 1.3451024073446027, "learning_rate": 1.9841495972665717e-05, "loss": 0.5805, "step": 2388 }, { "epoch": 0.08511320519443505, "grad_norm": 1.2820096927906328, "learning_rate": 1.9841291267701613e-05, "loss": 0.5009, "step": 2389 }, { "epoch": 0.08514883232092915, "grad_norm": 1.8989026794089126, "learning_rate": 1.9841086431693832e-05, "loss": 0.6449, "step": 2390 }, { "epoch": 0.08518445944742327, "grad_norm": 1.2106443528767048, "learning_rate": 1.9840881464645107e-05, "loss": 0.5188, "step": 2391 }, { "epoch": 0.08522008657391739, "grad_norm": 1.9667904255151651, "learning_rate": 1.9840676366558172e-05, "loss": 0.7235, "step": 2392 }, { "epoch": 0.08525571370041149, "grad_norm": 1.8276357491516388, "learning_rate": 1.9840471137435746e-05, "loss": 0.6315, "step": 2393 }, { "epoch": 0.0852913408269056, "grad_norm": 2.0700360166922267, "learning_rate": 1.984026577728057e-05, "loss": 0.5884, "step": 2394 }, { "epoch": 0.08532696795339972, "grad_norm": 1.7542376498617909, "learning_rate": 1.984006028609538e-05, "loss": 0.6838, "step": 2395 }, { "epoch": 0.08536259507989383, "grad_norm": 1.8669858289202381, "learning_rate": 1.983985466388291e-05, "loss": 0.5931, "step": 2396 }, { "epoch": 0.08539822220638794, "grad_norm": 1.4110462199381673, "learning_rate": 1.9839648910645898e-05, "loss": 0.5913, "step": 2397 }, { "epoch": 0.08543384933288206, "grad_norm": 1.3309255524770707, "learning_rate": 1.9839443026387078e-05, "loss": 0.5792, "step": 2398 }, { "epoch": 0.08546947645937616, "grad_norm": 1.2803721650807929, "learning_rate": 1.98392370111092e-05, "loss": 0.5629, "step": 2399 }, { "epoch": 0.08550510358587028, "grad_norm": 1.9549849346349462, "learning_rate": 1.9839030864815004e-05, "loss": 0.6222, "step": 2400 }, { "epoch": 0.0855407307123644, "grad_norm": 1.2274284083122828, "learning_rate": 1.9838824587507236e-05, "loss": 0.5391, "step": 2401 }, { "epoch": 0.0855763578388585, "grad_norm": 1.1805399322554986, "learning_rate": 1.983861817918864e-05, "loss": 0.561, "step": 2402 }, { "epoch": 0.08561198496535262, "grad_norm": 1.3822067521797865, "learning_rate": 1.9838411639861966e-05, "loss": 0.5739, "step": 2403 }, { "epoch": 0.08564761209184674, "grad_norm": 2.583288781294228, "learning_rate": 1.9838204969529965e-05, "loss": 0.5181, "step": 2404 }, { "epoch": 0.08568323921834084, "grad_norm": 2.106631061887723, "learning_rate": 1.983799816819539e-05, "loss": 0.6432, "step": 2405 }, { "epoch": 0.08571886634483496, "grad_norm": 2.161676413285736, "learning_rate": 1.983779123586099e-05, "loss": 0.5722, "step": 2406 }, { "epoch": 0.08575449347132907, "grad_norm": 2.159459264969435, "learning_rate": 1.9837584172529526e-05, "loss": 0.5966, "step": 2407 }, { "epoch": 0.08579012059782318, "grad_norm": 1.3571937753531995, "learning_rate": 1.983737697820375e-05, "loss": 0.518, "step": 2408 }, { "epoch": 0.0858257477243173, "grad_norm": 1.403754233244493, "learning_rate": 1.983716965288643e-05, "loss": 0.5702, "step": 2409 }, { "epoch": 0.08586137485081141, "grad_norm": 1.3176305828448758, "learning_rate": 1.983696219658031e-05, "loss": 0.5942, "step": 2410 }, { "epoch": 0.08589700197730551, "grad_norm": 1.6149924792662178, "learning_rate": 1.983675460928817e-05, "loss": 0.6039, "step": 2411 }, { "epoch": 0.08593262910379963, "grad_norm": 2.173760021669223, "learning_rate": 1.9836546891012766e-05, "loss": 0.6121, "step": 2412 }, { "epoch": 0.08596825623029375, "grad_norm": 1.333702173340208, "learning_rate": 1.9836339041756866e-05, "loss": 0.5221, "step": 2413 }, { "epoch": 0.08600388335678785, "grad_norm": 1.289280003412436, "learning_rate": 1.983613106152323e-05, "loss": 0.4894, "step": 2414 }, { "epoch": 0.08603951048328197, "grad_norm": 1.6407223799045527, "learning_rate": 1.983592295031464e-05, "loss": 0.5483, "step": 2415 }, { "epoch": 0.08607513760977609, "grad_norm": 1.3447487155567934, "learning_rate": 1.983571470813386e-05, "loss": 0.5611, "step": 2416 }, { "epoch": 0.08611076473627019, "grad_norm": 1.3531057423454984, "learning_rate": 1.9835506334983665e-05, "loss": 0.5796, "step": 2417 }, { "epoch": 0.08614639186276431, "grad_norm": 1.3478515741633528, "learning_rate": 1.9835297830866827e-05, "loss": 0.5408, "step": 2418 }, { "epoch": 0.08618201898925842, "grad_norm": 1.4780798817575607, "learning_rate": 1.9835089195786122e-05, "loss": 0.5126, "step": 2419 }, { "epoch": 0.08621764611575253, "grad_norm": 2.0880857365908136, "learning_rate": 1.983488042974433e-05, "loss": 0.6474, "step": 2420 }, { "epoch": 0.08625327324224664, "grad_norm": 1.9170842482494184, "learning_rate": 1.9834671532744238e-05, "loss": 0.6846, "step": 2421 }, { "epoch": 0.08628890036874076, "grad_norm": 2.114997069934102, "learning_rate": 1.983446250478861e-05, "loss": 0.647, "step": 2422 }, { "epoch": 0.08632452749523487, "grad_norm": 2.649412919949181, "learning_rate": 1.9834253345880248e-05, "loss": 0.6417, "step": 2423 }, { "epoch": 0.08636015462172898, "grad_norm": 1.6729159922786447, "learning_rate": 1.9834044056021926e-05, "loss": 0.5427, "step": 2424 }, { "epoch": 0.0863957817482231, "grad_norm": 1.2655761996623498, "learning_rate": 1.9833834635216434e-05, "loss": 0.5396, "step": 2425 }, { "epoch": 0.08643140887471722, "grad_norm": 1.2630594484001427, "learning_rate": 1.983362508346656e-05, "loss": 0.5648, "step": 2426 }, { "epoch": 0.08646703600121132, "grad_norm": 1.4302735254009635, "learning_rate": 1.9833415400775092e-05, "loss": 0.5711, "step": 2427 }, { "epoch": 0.08650266312770544, "grad_norm": 2.266252124655492, "learning_rate": 1.983320558714483e-05, "loss": 0.643, "step": 2428 }, { "epoch": 0.08653829025419955, "grad_norm": 1.9240754488765275, "learning_rate": 1.983299564257856e-05, "loss": 0.6839, "step": 2429 }, { "epoch": 0.08657391738069366, "grad_norm": 2.0209252917680796, "learning_rate": 1.983278556707908e-05, "loss": 0.6124, "step": 2430 }, { "epoch": 0.08660954450718777, "grad_norm": 1.373980654271818, "learning_rate": 1.983257536064919e-05, "loss": 0.5402, "step": 2431 }, { "epoch": 0.08664517163368189, "grad_norm": 2.0696487481469807, "learning_rate": 1.9832365023291686e-05, "loss": 0.6244, "step": 2432 }, { "epoch": 0.086680798760176, "grad_norm": 1.54699224015199, "learning_rate": 1.9832154555009367e-05, "loss": 0.5658, "step": 2433 }, { "epoch": 0.08671642588667011, "grad_norm": 1.37458709772133, "learning_rate": 1.983194395580504e-05, "loss": 0.5225, "step": 2434 }, { "epoch": 0.08675205301316423, "grad_norm": 1.1984828015510856, "learning_rate": 1.9831733225681506e-05, "loss": 0.5514, "step": 2435 }, { "epoch": 0.08678768013965833, "grad_norm": 1.2086015330946847, "learning_rate": 1.983152236464157e-05, "loss": 0.4671, "step": 2436 }, { "epoch": 0.08682330726615245, "grad_norm": 1.9228158233022408, "learning_rate": 1.9831311372688044e-05, "loss": 0.6089, "step": 2437 }, { "epoch": 0.08685893439264657, "grad_norm": 1.2082273655106388, "learning_rate": 1.9831100249823732e-05, "loss": 0.4731, "step": 2438 }, { "epoch": 0.08689456151914067, "grad_norm": 2.862523117677808, "learning_rate": 1.9830888996051454e-05, "loss": 0.5977, "step": 2439 }, { "epoch": 0.08693018864563479, "grad_norm": 2.136971483961077, "learning_rate": 1.9830677611374012e-05, "loss": 0.673, "step": 2440 }, { "epoch": 0.0869658157721289, "grad_norm": 2.6854373533271048, "learning_rate": 1.9830466095794235e-05, "loss": 0.6273, "step": 2441 }, { "epoch": 0.08700144289862301, "grad_norm": 1.258005027764413, "learning_rate": 1.9830254449314922e-05, "loss": 0.5599, "step": 2442 }, { "epoch": 0.08703707002511712, "grad_norm": 1.4010673383544094, "learning_rate": 1.9830042671938904e-05, "loss": 0.5451, "step": 2443 }, { "epoch": 0.08707269715161124, "grad_norm": 1.2956488666961565, "learning_rate": 1.9829830763668998e-05, "loss": 0.549, "step": 2444 }, { "epoch": 0.08710832427810535, "grad_norm": 1.9679394634194576, "learning_rate": 1.9829618724508025e-05, "loss": 0.6737, "step": 2445 }, { "epoch": 0.08714395140459946, "grad_norm": 1.4555137171942358, "learning_rate": 1.9829406554458806e-05, "loss": 0.546, "step": 2446 }, { "epoch": 0.08717957853109358, "grad_norm": 2.3272442484240545, "learning_rate": 1.9829194253524168e-05, "loss": 0.7784, "step": 2447 }, { "epoch": 0.08721520565758768, "grad_norm": 2.0808538940713577, "learning_rate": 1.9828981821706937e-05, "loss": 0.6332, "step": 2448 }, { "epoch": 0.0872508327840818, "grad_norm": 1.194811369429885, "learning_rate": 1.9828769259009947e-05, "loss": 0.5573, "step": 2449 }, { "epoch": 0.08728645991057592, "grad_norm": 1.300627321181369, "learning_rate": 1.9828556565436026e-05, "loss": 0.5271, "step": 2450 }, { "epoch": 0.08732208703707002, "grad_norm": 1.4073495212793552, "learning_rate": 1.9828343740988002e-05, "loss": 0.5955, "step": 2451 }, { "epoch": 0.08735771416356414, "grad_norm": 1.259469852384721, "learning_rate": 1.982813078566871e-05, "loss": 0.5281, "step": 2452 }, { "epoch": 0.08739334129005825, "grad_norm": 1.1573425752243898, "learning_rate": 1.9827917699480984e-05, "loss": 0.4762, "step": 2453 }, { "epoch": 0.08742896841655236, "grad_norm": 1.8696626990891432, "learning_rate": 1.982770448242767e-05, "loss": 0.602, "step": 2454 }, { "epoch": 0.08746459554304647, "grad_norm": 1.1779533797770192, "learning_rate": 1.98274911345116e-05, "loss": 0.4954, "step": 2455 }, { "epoch": 0.08750022266954059, "grad_norm": 2.6154926538145786, "learning_rate": 1.9827277655735617e-05, "loss": 0.71, "step": 2456 }, { "epoch": 0.0875358497960347, "grad_norm": 1.2319842154625567, "learning_rate": 1.9827064046102564e-05, "loss": 0.5097, "step": 2457 }, { "epoch": 0.08757147692252881, "grad_norm": 1.831911012722921, "learning_rate": 1.9826850305615283e-05, "loss": 0.7306, "step": 2458 }, { "epoch": 0.08760710404902293, "grad_norm": 2.1490304298827736, "learning_rate": 1.9826636434276624e-05, "loss": 0.7287, "step": 2459 }, { "epoch": 0.08764273117551703, "grad_norm": 1.1349386092805653, "learning_rate": 1.982642243208943e-05, "loss": 0.5538, "step": 2460 }, { "epoch": 0.08767835830201115, "grad_norm": 1.1546770679540812, "learning_rate": 1.9826208299056554e-05, "loss": 0.5834, "step": 2461 }, { "epoch": 0.08771398542850527, "grad_norm": 1.3371977771379826, "learning_rate": 1.9825994035180846e-05, "loss": 0.5092, "step": 2462 }, { "epoch": 0.08774961255499937, "grad_norm": 2.0360891650495256, "learning_rate": 1.9825779640465157e-05, "loss": 0.6007, "step": 2463 }, { "epoch": 0.08778523968149349, "grad_norm": 1.8764344735626346, "learning_rate": 1.982556511491235e-05, "loss": 0.6364, "step": 2464 }, { "epoch": 0.0878208668079876, "grad_norm": 1.893103222568026, "learning_rate": 1.9825350458525272e-05, "loss": 0.5766, "step": 2465 }, { "epoch": 0.08785649393448171, "grad_norm": 1.6796605791541606, "learning_rate": 1.9825135671306784e-05, "loss": 0.6217, "step": 2466 }, { "epoch": 0.08789212106097583, "grad_norm": 1.7414820062192082, "learning_rate": 1.982492075325975e-05, "loss": 0.5721, "step": 2467 }, { "epoch": 0.08792774818746994, "grad_norm": 2.2109605704619937, "learning_rate": 1.982470570438703e-05, "loss": 0.6278, "step": 2468 }, { "epoch": 0.08796337531396405, "grad_norm": 1.2429237671645643, "learning_rate": 1.9824490524691482e-05, "loss": 0.5019, "step": 2469 }, { "epoch": 0.08799900244045816, "grad_norm": 2.377943629031714, "learning_rate": 1.9824275214175978e-05, "loss": 0.5937, "step": 2470 }, { "epoch": 0.08803462956695228, "grad_norm": 1.3397485937153144, "learning_rate": 1.982405977284338e-05, "loss": 0.5335, "step": 2471 }, { "epoch": 0.08807025669344638, "grad_norm": 2.3385519287795837, "learning_rate": 1.9823844200696564e-05, "loss": 0.6777, "step": 2472 }, { "epoch": 0.0881058838199405, "grad_norm": 1.6465412891587652, "learning_rate": 1.9823628497738397e-05, "loss": 0.5798, "step": 2473 }, { "epoch": 0.08814151094643462, "grad_norm": 1.2735291924470868, "learning_rate": 1.9823412663971746e-05, "loss": 0.5666, "step": 2474 }, { "epoch": 0.08817713807292873, "grad_norm": 2.5910005854802773, "learning_rate": 1.9823196699399488e-05, "loss": 0.6647, "step": 2475 }, { "epoch": 0.08821276519942284, "grad_norm": 1.6812297500572462, "learning_rate": 1.9822980604024506e-05, "loss": 0.6567, "step": 2476 }, { "epoch": 0.08824839232591696, "grad_norm": 1.3484828991940623, "learning_rate": 1.982276437784967e-05, "loss": 0.5118, "step": 2477 }, { "epoch": 0.08828401945241107, "grad_norm": 1.3390177818606928, "learning_rate": 1.9822548020877858e-05, "loss": 0.5469, "step": 2478 }, { "epoch": 0.08831964657890518, "grad_norm": 1.764279110200369, "learning_rate": 1.9822331533111952e-05, "loss": 0.6464, "step": 2479 }, { "epoch": 0.08835527370539929, "grad_norm": 1.5901788164591655, "learning_rate": 1.982211491455484e-05, "loss": 0.5749, "step": 2480 }, { "epoch": 0.08839090083189341, "grad_norm": 1.9329956164135478, "learning_rate": 1.98218981652094e-05, "loss": 0.6044, "step": 2481 }, { "epoch": 0.08842652795838751, "grad_norm": 1.7090248370822536, "learning_rate": 1.9821681285078522e-05, "loss": 0.6166, "step": 2482 }, { "epoch": 0.08846215508488163, "grad_norm": 1.8448921494747408, "learning_rate": 1.9821464274165095e-05, "loss": 0.6573, "step": 2483 }, { "epoch": 0.08849778221137575, "grad_norm": 1.7968257200730382, "learning_rate": 1.9821247132472004e-05, "loss": 0.6062, "step": 2484 }, { "epoch": 0.08853340933786985, "grad_norm": 1.423862358196875, "learning_rate": 1.9821029860002145e-05, "loss": 0.5091, "step": 2485 }, { "epoch": 0.08856903646436397, "grad_norm": 1.939862450149805, "learning_rate": 1.98208124567584e-05, "loss": 0.6405, "step": 2486 }, { "epoch": 0.08860466359085808, "grad_norm": 1.3355332345117563, "learning_rate": 1.9820594922743684e-05, "loss": 0.5373, "step": 2487 }, { "epoch": 0.08864029071735219, "grad_norm": 1.3005719229560164, "learning_rate": 1.982037725796088e-05, "loss": 0.5017, "step": 2488 }, { "epoch": 0.0886759178438463, "grad_norm": 1.9018818724038278, "learning_rate": 1.9820159462412885e-05, "loss": 0.681, "step": 2489 }, { "epoch": 0.08871154497034042, "grad_norm": 2.055072153758149, "learning_rate": 1.9819941536102604e-05, "loss": 0.6358, "step": 2490 }, { "epoch": 0.08874717209683453, "grad_norm": 1.252541512739663, "learning_rate": 1.981972347903294e-05, "loss": 0.5366, "step": 2491 }, { "epoch": 0.08878279922332864, "grad_norm": 1.338985549909833, "learning_rate": 1.9819505291206792e-05, "loss": 0.5974, "step": 2492 }, { "epoch": 0.08881842634982276, "grad_norm": 1.4624879385717278, "learning_rate": 1.9819286972627066e-05, "loss": 0.5676, "step": 2493 }, { "epoch": 0.08885405347631686, "grad_norm": 2.0099233462328945, "learning_rate": 1.9819068523296677e-05, "loss": 0.7163, "step": 2494 }, { "epoch": 0.08888968060281098, "grad_norm": 2.7370096395531403, "learning_rate": 1.981884994321852e-05, "loss": 0.5554, "step": 2495 }, { "epoch": 0.0889253077293051, "grad_norm": 3.0099139494277303, "learning_rate": 1.981863123239552e-05, "loss": 0.7316, "step": 2496 }, { "epoch": 0.0889609348557992, "grad_norm": 1.5905409443114973, "learning_rate": 1.9818412390830576e-05, "loss": 0.5711, "step": 2497 }, { "epoch": 0.08899656198229332, "grad_norm": 1.4512721260397017, "learning_rate": 1.9818193418526613e-05, "loss": 0.5672, "step": 2498 }, { "epoch": 0.08903218910878744, "grad_norm": 1.432857546322144, "learning_rate": 1.981797431548654e-05, "loss": 0.5735, "step": 2499 }, { "epoch": 0.08906781623528154, "grad_norm": 1.9581936311770893, "learning_rate": 1.9817755081713277e-05, "loss": 0.5736, "step": 2500 }, { "epoch": 0.08910344336177566, "grad_norm": 2.022023432763352, "learning_rate": 1.9817535717209743e-05, "loss": 0.6018, "step": 2501 }, { "epoch": 0.08913907048826977, "grad_norm": 1.3092715161130037, "learning_rate": 1.9817316221978863e-05, "loss": 0.4619, "step": 2502 }, { "epoch": 0.08917469761476388, "grad_norm": 1.453422763164204, "learning_rate": 1.981709659602355e-05, "loss": 0.561, "step": 2503 }, { "epoch": 0.08921032474125799, "grad_norm": 1.2958023457227053, "learning_rate": 1.9816876839346734e-05, "loss": 0.5414, "step": 2504 }, { "epoch": 0.08924595186775211, "grad_norm": 1.4283695750267476, "learning_rate": 1.9816656951951343e-05, "loss": 0.5689, "step": 2505 }, { "epoch": 0.08928157899424621, "grad_norm": 2.614689146798712, "learning_rate": 1.9816436933840303e-05, "loss": 0.6277, "step": 2506 }, { "epoch": 0.08931720612074033, "grad_norm": 1.2982261370045076, "learning_rate": 1.9816216785016545e-05, "loss": 0.5305, "step": 2507 }, { "epoch": 0.08935283324723445, "grad_norm": 1.766071949830363, "learning_rate": 1.9815996505483e-05, "loss": 0.6188, "step": 2508 }, { "epoch": 0.08938846037372855, "grad_norm": 1.9836717310493526, "learning_rate": 1.98157760952426e-05, "loss": 0.5899, "step": 2509 }, { "epoch": 0.08942408750022267, "grad_norm": 2.315425490306042, "learning_rate": 1.9815555554298277e-05, "loss": 0.6575, "step": 2510 }, { "epoch": 0.08945971462671679, "grad_norm": 1.8998095740948557, "learning_rate": 1.9815334882652976e-05, "loss": 0.5638, "step": 2511 }, { "epoch": 0.08949534175321089, "grad_norm": 3.142632168798949, "learning_rate": 1.9815114080309624e-05, "loss": 0.6089, "step": 2512 }, { "epoch": 0.089530968879705, "grad_norm": 1.3624700444179962, "learning_rate": 1.9814893147271172e-05, "loss": 0.5416, "step": 2513 }, { "epoch": 0.08956659600619912, "grad_norm": 1.4255479122725052, "learning_rate": 1.9814672083540558e-05, "loss": 0.5882, "step": 2514 }, { "epoch": 0.08960222313269323, "grad_norm": 1.7903156071132271, "learning_rate": 1.9814450889120725e-05, "loss": 0.6076, "step": 2515 }, { "epoch": 0.08963785025918734, "grad_norm": 1.5797005160980226, "learning_rate": 1.9814229564014618e-05, "loss": 0.6134, "step": 2516 }, { "epoch": 0.08967347738568146, "grad_norm": 2.063676353797064, "learning_rate": 1.9814008108225185e-05, "loss": 0.5949, "step": 2517 }, { "epoch": 0.08970910451217556, "grad_norm": 1.212163817711236, "learning_rate": 1.9813786521755373e-05, "loss": 0.5291, "step": 2518 }, { "epoch": 0.08974473163866968, "grad_norm": 1.2769118349069224, "learning_rate": 1.9813564804608138e-05, "loss": 0.5392, "step": 2519 }, { "epoch": 0.0897803587651638, "grad_norm": 2.185505330501619, "learning_rate": 1.9813342956786422e-05, "loss": 0.6092, "step": 2520 }, { "epoch": 0.08981598589165792, "grad_norm": 2.393730216755607, "learning_rate": 1.981312097829319e-05, "loss": 0.6672, "step": 2521 }, { "epoch": 0.08985161301815202, "grad_norm": 1.2764286579919242, "learning_rate": 1.9812898869131392e-05, "loss": 0.5389, "step": 2522 }, { "epoch": 0.08988724014464614, "grad_norm": 1.8290509740481922, "learning_rate": 1.9812676629303987e-05, "loss": 0.676, "step": 2523 }, { "epoch": 0.08992286727114025, "grad_norm": 1.1807580234501591, "learning_rate": 1.9812454258813934e-05, "loss": 0.5025, "step": 2524 }, { "epoch": 0.08995849439763436, "grad_norm": 1.92506712615317, "learning_rate": 1.9812231757664193e-05, "loss": 0.6251, "step": 2525 }, { "epoch": 0.08999412152412847, "grad_norm": 1.58424470265551, "learning_rate": 1.981200912585773e-05, "loss": 0.6063, "step": 2526 }, { "epoch": 0.09002974865062259, "grad_norm": 1.736665793834005, "learning_rate": 1.9811786363397508e-05, "loss": 0.6393, "step": 2527 }, { "epoch": 0.0900653757771167, "grad_norm": 1.3962244590482689, "learning_rate": 1.981156347028649e-05, "loss": 0.5127, "step": 2528 }, { "epoch": 0.09010100290361081, "grad_norm": 1.319556722880328, "learning_rate": 1.9811340446527646e-05, "loss": 0.5336, "step": 2529 }, { "epoch": 0.09013663003010493, "grad_norm": 1.391962215129335, "learning_rate": 1.9811117292123948e-05, "loss": 0.5624, "step": 2530 }, { "epoch": 0.09017225715659903, "grad_norm": 1.8002796840433637, "learning_rate": 1.9810894007078364e-05, "loss": 0.5815, "step": 2531 }, { "epoch": 0.09020788428309315, "grad_norm": 2.37054936788273, "learning_rate": 1.981067059139387e-05, "loss": 0.6243, "step": 2532 }, { "epoch": 0.09024351140958727, "grad_norm": 2.188348054275896, "learning_rate": 1.981044704507344e-05, "loss": 0.6644, "step": 2533 }, { "epoch": 0.09027913853608137, "grad_norm": 1.3929312786316548, "learning_rate": 1.981022336812005e-05, "loss": 0.5805, "step": 2534 }, { "epoch": 0.09031476566257549, "grad_norm": 1.3052004136476758, "learning_rate": 1.9809999560536683e-05, "loss": 0.5624, "step": 2535 }, { "epoch": 0.0903503927890696, "grad_norm": 1.464308880471503, "learning_rate": 1.9809775622326307e-05, "loss": 0.5421, "step": 2536 }, { "epoch": 0.0903860199155637, "grad_norm": 1.9681956873899797, "learning_rate": 1.9809551553491918e-05, "loss": 0.5451, "step": 2537 }, { "epoch": 0.09042164704205782, "grad_norm": 1.795523567753794, "learning_rate": 1.9809327354036493e-05, "loss": 0.7226, "step": 2538 }, { "epoch": 0.09045727416855194, "grad_norm": 1.3230487401600959, "learning_rate": 1.9809103023963017e-05, "loss": 0.5684, "step": 2539 }, { "epoch": 0.09049290129504604, "grad_norm": 1.4529249099928336, "learning_rate": 1.9808878563274476e-05, "loss": 0.5539, "step": 2540 }, { "epoch": 0.09052852842154016, "grad_norm": 1.3261213293201406, "learning_rate": 1.9808653971973865e-05, "loss": 0.5991, "step": 2541 }, { "epoch": 0.09056415554803428, "grad_norm": 1.3470495783117016, "learning_rate": 1.9808429250064164e-05, "loss": 0.5634, "step": 2542 }, { "epoch": 0.09059978267452838, "grad_norm": 1.7622492402162129, "learning_rate": 1.9808204397548377e-05, "loss": 0.717, "step": 2543 }, { "epoch": 0.0906354098010225, "grad_norm": 1.24043484720489, "learning_rate": 1.9807979414429494e-05, "loss": 0.5215, "step": 2544 }, { "epoch": 0.09067103692751662, "grad_norm": 1.296899058434088, "learning_rate": 1.9807754300710507e-05, "loss": 0.5801, "step": 2545 }, { "epoch": 0.09070666405401072, "grad_norm": 1.2229822074707137, "learning_rate": 1.9807529056394416e-05, "loss": 0.5698, "step": 2546 }, { "epoch": 0.09074229118050484, "grad_norm": 1.2585522023416795, "learning_rate": 1.9807303681484224e-05, "loss": 0.6029, "step": 2547 }, { "epoch": 0.09077791830699895, "grad_norm": 1.2825168013099628, "learning_rate": 1.9807078175982925e-05, "loss": 0.5167, "step": 2548 }, { "epoch": 0.09081354543349306, "grad_norm": 1.6549771767912074, "learning_rate": 1.9806852539893526e-05, "loss": 0.5984, "step": 2549 }, { "epoch": 0.09084917255998717, "grad_norm": 1.2584412709206612, "learning_rate": 1.9806626773219032e-05, "loss": 0.5414, "step": 2550 }, { "epoch": 0.09088479968648129, "grad_norm": 1.2103112360142434, "learning_rate": 1.9806400875962447e-05, "loss": 0.5748, "step": 2551 }, { "epoch": 0.0909204268129754, "grad_norm": 1.925037883171635, "learning_rate": 1.980617484812678e-05, "loss": 0.6982, "step": 2552 }, { "epoch": 0.09095605393946951, "grad_norm": 2.15366723145489, "learning_rate": 1.9805948689715043e-05, "loss": 0.633, "step": 2553 }, { "epoch": 0.09099168106596363, "grad_norm": 1.8831547757366391, "learning_rate": 1.9805722400730243e-05, "loss": 0.6577, "step": 2554 }, { "epoch": 0.09102730819245773, "grad_norm": 1.832755961318674, "learning_rate": 1.98054959811754e-05, "loss": 0.613, "step": 2555 }, { "epoch": 0.09106293531895185, "grad_norm": 1.7812604772482123, "learning_rate": 1.980526943105352e-05, "loss": 0.6508, "step": 2556 }, { "epoch": 0.09109856244544597, "grad_norm": 1.3061317130687515, "learning_rate": 1.9805042750367622e-05, "loss": 0.5038, "step": 2557 }, { "epoch": 0.09113418957194007, "grad_norm": 2.083024150117773, "learning_rate": 1.980481593912073e-05, "loss": 0.6731, "step": 2558 }, { "epoch": 0.09116981669843419, "grad_norm": 1.2352646635010514, "learning_rate": 1.9804588997315858e-05, "loss": 0.5208, "step": 2559 }, { "epoch": 0.0912054438249283, "grad_norm": 1.2731257018930817, "learning_rate": 1.9804361924956034e-05, "loss": 0.5395, "step": 2560 }, { "epoch": 0.0912410709514224, "grad_norm": 1.3295341263198632, "learning_rate": 1.9804134722044276e-05, "loss": 0.5294, "step": 2561 }, { "epoch": 0.09127669807791652, "grad_norm": 2.151710476503235, "learning_rate": 1.9803907388583613e-05, "loss": 0.6829, "step": 2562 }, { "epoch": 0.09131232520441064, "grad_norm": 1.7556923410101821, "learning_rate": 1.9803679924577066e-05, "loss": 0.5846, "step": 2563 }, { "epoch": 0.09134795233090474, "grad_norm": 1.6572323664276194, "learning_rate": 1.980345233002767e-05, "loss": 0.6816, "step": 2564 }, { "epoch": 0.09138357945739886, "grad_norm": 1.847507303880543, "learning_rate": 1.980322460493846e-05, "loss": 0.5899, "step": 2565 }, { "epoch": 0.09141920658389298, "grad_norm": 2.8149488715187116, "learning_rate": 1.9802996749312456e-05, "loss": 0.672, "step": 2566 }, { "epoch": 0.0914548337103871, "grad_norm": 1.357957866405597, "learning_rate": 1.9802768763152703e-05, "loss": 0.4961, "step": 2567 }, { "epoch": 0.0914904608368812, "grad_norm": 2.047627268772909, "learning_rate": 1.980254064646223e-05, "loss": 0.5974, "step": 2568 }, { "epoch": 0.09152608796337532, "grad_norm": 2.690437063095208, "learning_rate": 1.9802312399244075e-05, "loss": 0.5305, "step": 2569 }, { "epoch": 0.09156171508986943, "grad_norm": 1.2896821895199713, "learning_rate": 1.9802084021501282e-05, "loss": 0.5795, "step": 2570 }, { "epoch": 0.09159734221636354, "grad_norm": 1.7153725901764765, "learning_rate": 1.9801855513236888e-05, "loss": 0.6398, "step": 2571 }, { "epoch": 0.09163296934285765, "grad_norm": 2.3842584730736816, "learning_rate": 1.9801626874453933e-05, "loss": 0.7395, "step": 2572 }, { "epoch": 0.09166859646935177, "grad_norm": 1.4101525730266726, "learning_rate": 1.9801398105155472e-05, "loss": 0.5478, "step": 2573 }, { "epoch": 0.09170422359584587, "grad_norm": 1.8114176571028153, "learning_rate": 1.980116920534454e-05, "loss": 0.573, "step": 2574 }, { "epoch": 0.09173985072233999, "grad_norm": 1.249724709701995, "learning_rate": 1.980094017502419e-05, "loss": 0.5363, "step": 2575 }, { "epoch": 0.09177547784883411, "grad_norm": 1.2356231346089181, "learning_rate": 1.980071101419747e-05, "loss": 0.5663, "step": 2576 }, { "epoch": 0.09181110497532821, "grad_norm": 1.9119455140161223, "learning_rate": 1.9800481722867434e-05, "loss": 0.527, "step": 2577 }, { "epoch": 0.09184673210182233, "grad_norm": 2.0028323234887644, "learning_rate": 1.9800252301037135e-05, "loss": 0.6275, "step": 2578 }, { "epoch": 0.09188235922831645, "grad_norm": 1.2310614321949667, "learning_rate": 1.9800022748709624e-05, "loss": 0.5439, "step": 2579 }, { "epoch": 0.09191798635481055, "grad_norm": 1.6870439231846355, "learning_rate": 1.9799793065887966e-05, "loss": 0.5888, "step": 2580 }, { "epoch": 0.09195361348130467, "grad_norm": 1.4107879353313542, "learning_rate": 1.9799563252575208e-05, "loss": 0.5188, "step": 2581 }, { "epoch": 0.09198924060779878, "grad_norm": 1.5163803233267117, "learning_rate": 1.9799333308774415e-05, "loss": 0.5574, "step": 2582 }, { "epoch": 0.09202486773429289, "grad_norm": 1.5376323016143512, "learning_rate": 1.9799103234488654e-05, "loss": 0.5956, "step": 2583 }, { "epoch": 0.092060494860787, "grad_norm": 2.2594852457127312, "learning_rate": 1.979887302972098e-05, "loss": 0.7601, "step": 2584 }, { "epoch": 0.09209612198728112, "grad_norm": 1.525397207250845, "learning_rate": 1.9798642694474465e-05, "loss": 0.5808, "step": 2585 }, { "epoch": 0.09213174911377522, "grad_norm": 1.3767699078278974, "learning_rate": 1.9798412228752174e-05, "loss": 0.5315, "step": 2586 }, { "epoch": 0.09216737624026934, "grad_norm": 1.4471054485401307, "learning_rate": 1.9798181632557176e-05, "loss": 0.5205, "step": 2587 }, { "epoch": 0.09220300336676346, "grad_norm": 2.4152349468270984, "learning_rate": 1.9797950905892537e-05, "loss": 0.6257, "step": 2588 }, { "epoch": 0.09223863049325756, "grad_norm": 1.919397809581342, "learning_rate": 1.979772004876134e-05, "loss": 0.5662, "step": 2589 }, { "epoch": 0.09227425761975168, "grad_norm": 1.483104637692988, "learning_rate": 1.979748906116665e-05, "loss": 0.5771, "step": 2590 }, { "epoch": 0.0923098847462458, "grad_norm": 1.260211605548964, "learning_rate": 1.9797257943111542e-05, "loss": 0.4629, "step": 2591 }, { "epoch": 0.0923455118727399, "grad_norm": 1.8147354339087394, "learning_rate": 1.9797026694599097e-05, "loss": 0.633, "step": 2592 }, { "epoch": 0.09238113899923402, "grad_norm": 1.413476452528301, "learning_rate": 1.9796795315632394e-05, "loss": 0.5359, "step": 2593 }, { "epoch": 0.09241676612572813, "grad_norm": 1.7081289860998254, "learning_rate": 1.9796563806214516e-05, "loss": 0.634, "step": 2594 }, { "epoch": 0.09245239325222224, "grad_norm": 2.8747758655028743, "learning_rate": 1.9796332166348543e-05, "loss": 0.6794, "step": 2595 }, { "epoch": 0.09248802037871635, "grad_norm": 3.419633311392305, "learning_rate": 1.979610039603756e-05, "loss": 0.5994, "step": 2596 }, { "epoch": 0.09252364750521047, "grad_norm": 1.5357321248127538, "learning_rate": 1.9795868495284654e-05, "loss": 0.5311, "step": 2597 }, { "epoch": 0.09255927463170457, "grad_norm": 1.6802929858765814, "learning_rate": 1.979563646409291e-05, "loss": 0.644, "step": 2598 }, { "epoch": 0.09259490175819869, "grad_norm": 1.7722037829793016, "learning_rate": 1.979540430246542e-05, "loss": 0.6945, "step": 2599 }, { "epoch": 0.09263052888469281, "grad_norm": 1.462473552595656, "learning_rate": 1.9795172010405274e-05, "loss": 0.5446, "step": 2600 }, { "epoch": 0.09266615601118691, "grad_norm": 2.359859851723722, "learning_rate": 1.979493958791557e-05, "loss": 0.6739, "step": 2601 }, { "epoch": 0.09270178313768103, "grad_norm": 1.2981919078047064, "learning_rate": 1.9794707034999396e-05, "loss": 0.5922, "step": 2602 }, { "epoch": 0.09273741026417515, "grad_norm": 1.303670238429109, "learning_rate": 1.9794474351659854e-05, "loss": 0.5449, "step": 2603 }, { "epoch": 0.09277303739066925, "grad_norm": 2.429357925051399, "learning_rate": 1.979424153790004e-05, "loss": 0.6596, "step": 2604 }, { "epoch": 0.09280866451716337, "grad_norm": 1.762071001061402, "learning_rate": 1.979400859372305e-05, "loss": 0.7073, "step": 2605 }, { "epoch": 0.09284429164365748, "grad_norm": 1.3055719445397207, "learning_rate": 1.9793775519131994e-05, "loss": 0.5, "step": 2606 }, { "epoch": 0.09287991877015159, "grad_norm": 1.3987744447108204, "learning_rate": 1.9793542314129972e-05, "loss": 0.5186, "step": 2607 }, { "epoch": 0.0929155458966457, "grad_norm": 1.1790807084865107, "learning_rate": 1.9793308978720086e-05, "loss": 0.538, "step": 2608 }, { "epoch": 0.09295117302313982, "grad_norm": 1.2812750389998437, "learning_rate": 1.979307551290545e-05, "loss": 0.5546, "step": 2609 }, { "epoch": 0.09298680014963392, "grad_norm": 1.7173140578667017, "learning_rate": 1.9792841916689163e-05, "loss": 0.6506, "step": 2610 }, { "epoch": 0.09302242727612804, "grad_norm": 2.3463842084601714, "learning_rate": 1.9792608190074345e-05, "loss": 0.5514, "step": 2611 }, { "epoch": 0.09305805440262216, "grad_norm": 2.144976241731739, "learning_rate": 1.9792374333064105e-05, "loss": 0.6703, "step": 2612 }, { "epoch": 0.09309368152911626, "grad_norm": 1.3046526878969487, "learning_rate": 1.979214034566155e-05, "loss": 0.5579, "step": 2613 }, { "epoch": 0.09312930865561038, "grad_norm": 1.1765421842216333, "learning_rate": 1.979190622786981e-05, "loss": 0.497, "step": 2614 }, { "epoch": 0.0931649357821045, "grad_norm": 1.6082806101542917, "learning_rate": 1.9791671979691988e-05, "loss": 0.5917, "step": 2615 }, { "epoch": 0.09320056290859861, "grad_norm": 1.2633983931347432, "learning_rate": 1.9791437601131215e-05, "loss": 0.5656, "step": 2616 }, { "epoch": 0.09323619003509272, "grad_norm": 1.8561803084894746, "learning_rate": 1.9791203092190605e-05, "loss": 0.6352, "step": 2617 }, { "epoch": 0.09327181716158683, "grad_norm": 1.2335641736214829, "learning_rate": 1.979096845287328e-05, "loss": 0.5597, "step": 2618 }, { "epoch": 0.09330744428808095, "grad_norm": 2.5411472486046645, "learning_rate": 1.9790733683182366e-05, "loss": 0.7032, "step": 2619 }, { "epoch": 0.09334307141457505, "grad_norm": 1.3849708348555716, "learning_rate": 1.9790498783120992e-05, "loss": 0.524, "step": 2620 }, { "epoch": 0.09337869854106917, "grad_norm": 1.7667470321064662, "learning_rate": 1.9790263752692286e-05, "loss": 0.6675, "step": 2621 }, { "epoch": 0.09341432566756329, "grad_norm": 2.120667841711892, "learning_rate": 1.979002859189937e-05, "loss": 0.5338, "step": 2622 }, { "epoch": 0.09344995279405739, "grad_norm": 1.2516770287390582, "learning_rate": 1.9789793300745383e-05, "loss": 0.4871, "step": 2623 }, { "epoch": 0.09348557992055151, "grad_norm": 1.1907270915019335, "learning_rate": 1.9789557879233454e-05, "loss": 0.5215, "step": 2624 }, { "epoch": 0.09352120704704563, "grad_norm": 3.965105836959301, "learning_rate": 1.9789322327366722e-05, "loss": 0.6941, "step": 2625 }, { "epoch": 0.09355683417353973, "grad_norm": 1.3491783297141273, "learning_rate": 1.9789086645148318e-05, "loss": 0.5657, "step": 2626 }, { "epoch": 0.09359246130003385, "grad_norm": 1.1841790826096361, "learning_rate": 1.978885083258139e-05, "loss": 0.5531, "step": 2627 }, { "epoch": 0.09362808842652796, "grad_norm": 1.2725493603604225, "learning_rate": 1.9788614889669063e-05, "loss": 0.5202, "step": 2628 }, { "epoch": 0.09366371555302207, "grad_norm": 1.8801740628112318, "learning_rate": 1.978837881641449e-05, "loss": 0.5945, "step": 2629 }, { "epoch": 0.09369934267951618, "grad_norm": 1.328130452513613, "learning_rate": 1.9788142612820808e-05, "loss": 0.5194, "step": 2630 }, { "epoch": 0.0937349698060103, "grad_norm": 1.149428989271541, "learning_rate": 1.9787906278891172e-05, "loss": 0.5221, "step": 2631 }, { "epoch": 0.0937705969325044, "grad_norm": 1.3263220959022595, "learning_rate": 1.9787669814628718e-05, "loss": 0.5812, "step": 2632 }, { "epoch": 0.09380622405899852, "grad_norm": 1.2147062083878506, "learning_rate": 1.97874332200366e-05, "loss": 0.5902, "step": 2633 }, { "epoch": 0.09384185118549264, "grad_norm": 1.2364566654165734, "learning_rate": 1.978719649511797e-05, "loss": 0.4937, "step": 2634 }, { "epoch": 0.09387747831198674, "grad_norm": 1.343265870713445, "learning_rate": 1.9786959639875976e-05, "loss": 0.5294, "step": 2635 }, { "epoch": 0.09391310543848086, "grad_norm": 1.188115332250848, "learning_rate": 1.9786722654313773e-05, "loss": 0.498, "step": 2636 }, { "epoch": 0.09394873256497498, "grad_norm": 1.2406266190468602, "learning_rate": 1.9786485538434518e-05, "loss": 0.466, "step": 2637 }, { "epoch": 0.09398435969146908, "grad_norm": 2.165524012978332, "learning_rate": 1.9786248292241368e-05, "loss": 0.5994, "step": 2638 }, { "epoch": 0.0940199868179632, "grad_norm": 2.4178605419421313, "learning_rate": 1.978601091573748e-05, "loss": 0.6101, "step": 2639 }, { "epoch": 0.09405561394445731, "grad_norm": 1.3316616968498032, "learning_rate": 1.978577340892602e-05, "loss": 0.496, "step": 2640 }, { "epoch": 0.09409124107095142, "grad_norm": 1.8659211469071373, "learning_rate": 1.9785535771810145e-05, "loss": 0.6261, "step": 2641 }, { "epoch": 0.09412686819744553, "grad_norm": 2.082256069403341, "learning_rate": 1.978529800439302e-05, "loss": 0.5738, "step": 2642 }, { "epoch": 0.09416249532393965, "grad_norm": 1.589551322332579, "learning_rate": 1.9785060106677818e-05, "loss": 0.5828, "step": 2643 }, { "epoch": 0.09419812245043375, "grad_norm": 2.336517121205693, "learning_rate": 1.9784822078667698e-05, "loss": 0.6051, "step": 2644 }, { "epoch": 0.09423374957692787, "grad_norm": 1.5424915984428316, "learning_rate": 1.978458392036583e-05, "loss": 0.5329, "step": 2645 }, { "epoch": 0.09426937670342199, "grad_norm": 1.6713624084001673, "learning_rate": 1.9784345631775393e-05, "loss": 0.5368, "step": 2646 }, { "epoch": 0.09430500382991609, "grad_norm": 1.8605009200868252, "learning_rate": 1.978410721289955e-05, "loss": 0.614, "step": 2647 }, { "epoch": 0.09434063095641021, "grad_norm": 1.758864566033263, "learning_rate": 1.9783868663741487e-05, "loss": 0.6095, "step": 2648 }, { "epoch": 0.09437625808290433, "grad_norm": 1.7021931559222454, "learning_rate": 1.9783629984304368e-05, "loss": 0.5289, "step": 2649 }, { "epoch": 0.09441188520939843, "grad_norm": 1.4851702688811774, "learning_rate": 1.9783391174591383e-05, "loss": 0.4848, "step": 2650 }, { "epoch": 0.09444751233589255, "grad_norm": 1.7007839021830726, "learning_rate": 1.9783152234605703e-05, "loss": 0.5177, "step": 2651 }, { "epoch": 0.09448313946238666, "grad_norm": 2.538098536115964, "learning_rate": 1.9782913164350513e-05, "loss": 0.6434, "step": 2652 }, { "epoch": 0.09451876658888077, "grad_norm": 2.0858385240260593, "learning_rate": 1.9782673963828998e-05, "loss": 0.6332, "step": 2653 }, { "epoch": 0.09455439371537488, "grad_norm": 1.6404769666095242, "learning_rate": 1.978243463304434e-05, "loss": 0.5183, "step": 2654 }, { "epoch": 0.094590020841869, "grad_norm": 1.8173868637171646, "learning_rate": 1.9782195171999728e-05, "loss": 0.5497, "step": 2655 }, { "epoch": 0.0946256479683631, "grad_norm": 2.9025911770050765, "learning_rate": 1.978195558069835e-05, "loss": 0.6651, "step": 2656 }, { "epoch": 0.09466127509485722, "grad_norm": 1.6162760834190373, "learning_rate": 1.9781715859143396e-05, "loss": 0.61, "step": 2657 }, { "epoch": 0.09469690222135134, "grad_norm": 2.263767288930247, "learning_rate": 1.9781476007338058e-05, "loss": 0.6891, "step": 2658 }, { "epoch": 0.09473252934784544, "grad_norm": 2.0347732098088485, "learning_rate": 1.978123602528553e-05, "loss": 0.7088, "step": 2659 }, { "epoch": 0.09476815647433956, "grad_norm": 2.0107953831974053, "learning_rate": 1.978099591298901e-05, "loss": 0.7035, "step": 2660 }, { "epoch": 0.09480378360083368, "grad_norm": 2.244123408077117, "learning_rate": 1.978075567045169e-05, "loss": 0.5708, "step": 2661 }, { "epoch": 0.0948394107273278, "grad_norm": 2.633811350846421, "learning_rate": 1.9780515297676772e-05, "loss": 0.6535, "step": 2662 }, { "epoch": 0.0948750378538219, "grad_norm": 1.5479568212964976, "learning_rate": 1.978027479466746e-05, "loss": 0.5539, "step": 2663 }, { "epoch": 0.09491066498031601, "grad_norm": 1.947009170419744, "learning_rate": 1.978003416142695e-05, "loss": 0.5614, "step": 2664 }, { "epoch": 0.09494629210681013, "grad_norm": 2.1140987722139886, "learning_rate": 1.9779793397958448e-05, "loss": 0.6264, "step": 2665 }, { "epoch": 0.09498191923330424, "grad_norm": 1.376075755492759, "learning_rate": 1.9779552504265163e-05, "loss": 0.5703, "step": 2666 }, { "epoch": 0.09501754635979835, "grad_norm": 1.3360256246329534, "learning_rate": 1.9779311480350303e-05, "loss": 0.554, "step": 2667 }, { "epoch": 0.09505317348629247, "grad_norm": 2.4431793028099333, "learning_rate": 1.9779070326217074e-05, "loss": 0.5596, "step": 2668 }, { "epoch": 0.09508880061278657, "grad_norm": 2.629808573726883, "learning_rate": 1.9778829041868687e-05, "loss": 0.6246, "step": 2669 }, { "epoch": 0.09512442773928069, "grad_norm": 1.370566078397002, "learning_rate": 1.9778587627308357e-05, "loss": 0.5004, "step": 2670 }, { "epoch": 0.0951600548657748, "grad_norm": 1.7538123337505442, "learning_rate": 1.9778346082539298e-05, "loss": 0.6625, "step": 2671 }, { "epoch": 0.09519568199226891, "grad_norm": 1.9979649962728883, "learning_rate": 1.9778104407564726e-05, "loss": 0.6619, "step": 2672 }, { "epoch": 0.09523130911876303, "grad_norm": 1.4674394386152851, "learning_rate": 1.9777862602387862e-05, "loss": 0.5434, "step": 2673 }, { "epoch": 0.09526693624525714, "grad_norm": 1.9805922108646403, "learning_rate": 1.977762066701192e-05, "loss": 0.6713, "step": 2674 }, { "epoch": 0.09530256337175125, "grad_norm": 1.3545585164622242, "learning_rate": 1.9777378601440126e-05, "loss": 0.6214, "step": 2675 }, { "epoch": 0.09533819049824536, "grad_norm": 1.354793138198836, "learning_rate": 1.97771364056757e-05, "loss": 0.5702, "step": 2676 }, { "epoch": 0.09537381762473948, "grad_norm": 2.0245881746878647, "learning_rate": 1.977689407972187e-05, "loss": 0.6205, "step": 2677 }, { "epoch": 0.09540944475123359, "grad_norm": 1.3037407848863654, "learning_rate": 1.977665162358186e-05, "loss": 0.4942, "step": 2678 }, { "epoch": 0.0954450718777277, "grad_norm": 1.36562306034194, "learning_rate": 1.9776409037258903e-05, "loss": 0.571, "step": 2679 }, { "epoch": 0.09548069900422182, "grad_norm": 1.4009447610508237, "learning_rate": 1.9776166320756228e-05, "loss": 0.5728, "step": 2680 }, { "epoch": 0.09551632613071592, "grad_norm": 1.3427946654337417, "learning_rate": 1.9775923474077064e-05, "loss": 0.5611, "step": 2681 }, { "epoch": 0.09555195325721004, "grad_norm": 1.687738970039556, "learning_rate": 1.9775680497224643e-05, "loss": 0.6257, "step": 2682 }, { "epoch": 0.09558758038370416, "grad_norm": 1.8860306781622496, "learning_rate": 1.9775437390202205e-05, "loss": 0.6898, "step": 2683 }, { "epoch": 0.09562320751019826, "grad_norm": 1.2640414678684098, "learning_rate": 1.9775194153012985e-05, "loss": 0.5356, "step": 2684 }, { "epoch": 0.09565883463669238, "grad_norm": 2.2972879293423207, "learning_rate": 1.9774950785660226e-05, "loss": 0.6406, "step": 2685 }, { "epoch": 0.0956944617631865, "grad_norm": 1.3637620626608984, "learning_rate": 1.9774707288147163e-05, "loss": 0.5617, "step": 2686 }, { "epoch": 0.0957300888896806, "grad_norm": 2.1381752449994154, "learning_rate": 1.9774463660477038e-05, "loss": 0.6246, "step": 2687 }, { "epoch": 0.09576571601617472, "grad_norm": 1.8640460469529996, "learning_rate": 1.97742199026531e-05, "loss": 0.6822, "step": 2688 }, { "epoch": 0.09580134314266883, "grad_norm": 1.5750630543237785, "learning_rate": 1.977397601467859e-05, "loss": 0.5524, "step": 2689 }, { "epoch": 0.09583697026916294, "grad_norm": 1.8187893410101756, "learning_rate": 1.9773731996556758e-05, "loss": 0.6492, "step": 2690 }, { "epoch": 0.09587259739565705, "grad_norm": 2.8613832345592938, "learning_rate": 1.9773487848290856e-05, "loss": 0.5777, "step": 2691 }, { "epoch": 0.09590822452215117, "grad_norm": 1.390920375873562, "learning_rate": 1.977324356988413e-05, "loss": 0.5269, "step": 2692 }, { "epoch": 0.09594385164864527, "grad_norm": 1.2543165465909905, "learning_rate": 1.9772999161339833e-05, "loss": 0.5462, "step": 2693 }, { "epoch": 0.09597947877513939, "grad_norm": 1.204651221975373, "learning_rate": 1.9772754622661223e-05, "loss": 0.5779, "step": 2694 }, { "epoch": 0.09601510590163351, "grad_norm": 2.110720826849181, "learning_rate": 1.9772509953851556e-05, "loss": 0.629, "step": 2695 }, { "epoch": 0.09605073302812761, "grad_norm": 1.8395540303605329, "learning_rate": 1.9772265154914087e-05, "loss": 0.6776, "step": 2696 }, { "epoch": 0.09608636015462173, "grad_norm": 2.223109161266895, "learning_rate": 1.9772020225852077e-05, "loss": 0.5944, "step": 2697 }, { "epoch": 0.09612198728111584, "grad_norm": 2.137943071898215, "learning_rate": 1.9771775166668787e-05, "loss": 0.6694, "step": 2698 }, { "epoch": 0.09615761440760995, "grad_norm": 1.3304368165224039, "learning_rate": 1.977152997736748e-05, "loss": 0.5033, "step": 2699 }, { "epoch": 0.09619324153410407, "grad_norm": 1.4223193769947928, "learning_rate": 1.977128465795142e-05, "loss": 0.545, "step": 2700 }, { "epoch": 0.09622886866059818, "grad_norm": 2.157341962905005, "learning_rate": 1.9771039208423882e-05, "loss": 0.61, "step": 2701 }, { "epoch": 0.09626449578709229, "grad_norm": 1.2987574887124904, "learning_rate": 1.9770793628788123e-05, "loss": 0.4984, "step": 2702 }, { "epoch": 0.0963001229135864, "grad_norm": 1.3580563090685709, "learning_rate": 1.9770547919047416e-05, "loss": 0.5984, "step": 2703 }, { "epoch": 0.09633575004008052, "grad_norm": 1.3005194628191432, "learning_rate": 1.9770302079205036e-05, "loss": 0.5449, "step": 2704 }, { "epoch": 0.09637137716657462, "grad_norm": 1.9644468264627954, "learning_rate": 1.9770056109264254e-05, "loss": 0.627, "step": 2705 }, { "epoch": 0.09640700429306874, "grad_norm": 1.3800157097287054, "learning_rate": 1.976981000922835e-05, "loss": 0.5723, "step": 2706 }, { "epoch": 0.09644263141956286, "grad_norm": 1.3076925183396273, "learning_rate": 1.976956377910059e-05, "loss": 0.5399, "step": 2707 }, { "epoch": 0.09647825854605697, "grad_norm": 1.3893936818919308, "learning_rate": 1.9769317418884264e-05, "loss": 0.5473, "step": 2708 }, { "epoch": 0.09651388567255108, "grad_norm": 1.615696901926642, "learning_rate": 1.9769070928582648e-05, "loss": 0.6153, "step": 2709 }, { "epoch": 0.0965495127990452, "grad_norm": 1.3674349266779806, "learning_rate": 1.9768824308199025e-05, "loss": 0.4894, "step": 2710 }, { "epoch": 0.09658513992553931, "grad_norm": 1.4753715240690628, "learning_rate": 1.9768577557736677e-05, "loss": 0.5485, "step": 2711 }, { "epoch": 0.09662076705203342, "grad_norm": 1.2036995697429558, "learning_rate": 1.976833067719889e-05, "loss": 0.5187, "step": 2712 }, { "epoch": 0.09665639417852753, "grad_norm": 1.8813782768086402, "learning_rate": 1.9768083666588954e-05, "loss": 0.6965, "step": 2713 }, { "epoch": 0.09669202130502165, "grad_norm": 1.4157512912858092, "learning_rate": 1.9767836525910156e-05, "loss": 0.5906, "step": 2714 }, { "epoch": 0.09672764843151575, "grad_norm": 1.3735338051434256, "learning_rate": 1.9767589255165785e-05, "loss": 0.581, "step": 2715 }, { "epoch": 0.09676327555800987, "grad_norm": 1.597345509150699, "learning_rate": 1.976734185435914e-05, "loss": 0.6457, "step": 2716 }, { "epoch": 0.09679890268450399, "grad_norm": 1.320147275263468, "learning_rate": 1.976709432349351e-05, "loss": 0.5142, "step": 2717 }, { "epoch": 0.09683452981099809, "grad_norm": 1.3589549981353553, "learning_rate": 1.976684666257219e-05, "loss": 0.5119, "step": 2718 }, { "epoch": 0.09687015693749221, "grad_norm": 1.2697604391870407, "learning_rate": 1.9766598871598484e-05, "loss": 0.5478, "step": 2719 }, { "epoch": 0.09690578406398632, "grad_norm": 1.2024645733333927, "learning_rate": 1.9766350950575685e-05, "loss": 0.5516, "step": 2720 }, { "epoch": 0.09694141119048043, "grad_norm": 1.3979545810621647, "learning_rate": 1.9766102899507098e-05, "loss": 0.6242, "step": 2721 }, { "epoch": 0.09697703831697455, "grad_norm": 1.9035959028972804, "learning_rate": 1.9765854718396026e-05, "loss": 0.6007, "step": 2722 }, { "epoch": 0.09701266544346866, "grad_norm": 2.1028458420241, "learning_rate": 1.9765606407245766e-05, "loss": 0.5909, "step": 2723 }, { "epoch": 0.09704829256996277, "grad_norm": 1.8389749622925065, "learning_rate": 1.9765357966059638e-05, "loss": 0.6249, "step": 2724 }, { "epoch": 0.09708391969645688, "grad_norm": 2.163463605493435, "learning_rate": 1.976510939484094e-05, "loss": 0.5902, "step": 2725 }, { "epoch": 0.097119546822951, "grad_norm": 1.3052685819741703, "learning_rate": 1.9764860693592988e-05, "loss": 0.544, "step": 2726 }, { "epoch": 0.0971551739494451, "grad_norm": 1.1823203431043323, "learning_rate": 1.9764611862319087e-05, "loss": 0.5061, "step": 2727 }, { "epoch": 0.09719080107593922, "grad_norm": 1.8753037764592269, "learning_rate": 1.9764362901022556e-05, "loss": 0.6011, "step": 2728 }, { "epoch": 0.09722642820243334, "grad_norm": 1.7818513414878974, "learning_rate": 1.9764113809706707e-05, "loss": 0.6681, "step": 2729 }, { "epoch": 0.09726205532892744, "grad_norm": 2.446222181566472, "learning_rate": 1.976386458837486e-05, "loss": 0.6229, "step": 2730 }, { "epoch": 0.09729768245542156, "grad_norm": 1.2944986726685823, "learning_rate": 1.9763615237030328e-05, "loss": 0.5814, "step": 2731 }, { "epoch": 0.09733330958191568, "grad_norm": 1.8680148568486077, "learning_rate": 1.9763365755676438e-05, "loss": 0.6223, "step": 2732 }, { "epoch": 0.09736893670840978, "grad_norm": 1.6187461858550398, "learning_rate": 1.9763116144316506e-05, "loss": 0.5585, "step": 2733 }, { "epoch": 0.0974045638349039, "grad_norm": 1.402460375308644, "learning_rate": 1.976286640295386e-05, "loss": 0.5749, "step": 2734 }, { "epoch": 0.09744019096139801, "grad_norm": 1.2153818712259195, "learning_rate": 1.9762616531591826e-05, "loss": 0.5617, "step": 2735 }, { "epoch": 0.09747581808789212, "grad_norm": 1.8507324749936758, "learning_rate": 1.9762366530233724e-05, "loss": 0.6802, "step": 2736 }, { "epoch": 0.09751144521438623, "grad_norm": 1.254280822889748, "learning_rate": 1.9762116398882894e-05, "loss": 0.5587, "step": 2737 }, { "epoch": 0.09754707234088035, "grad_norm": 1.845220913840414, "learning_rate": 1.976186613754266e-05, "loss": 0.6212, "step": 2738 }, { "epoch": 0.09758269946737445, "grad_norm": 1.1602898815841227, "learning_rate": 1.976161574621635e-05, "loss": 0.5636, "step": 2739 }, { "epoch": 0.09761832659386857, "grad_norm": 1.7679192244638486, "learning_rate": 1.9761365224907306e-05, "loss": 0.5697, "step": 2740 }, { "epoch": 0.09765395372036269, "grad_norm": 1.7334224072167808, "learning_rate": 1.976111457361886e-05, "loss": 0.6605, "step": 2741 }, { "epoch": 0.09768958084685679, "grad_norm": 1.6662141025641473, "learning_rate": 1.9760863792354356e-05, "loss": 0.6952, "step": 2742 }, { "epoch": 0.09772520797335091, "grad_norm": 1.6563376267770085, "learning_rate": 1.9760612881117124e-05, "loss": 0.7236, "step": 2743 }, { "epoch": 0.09776083509984503, "grad_norm": 1.1924675992738503, "learning_rate": 1.9760361839910513e-05, "loss": 0.5553, "step": 2744 }, { "epoch": 0.09779646222633913, "grad_norm": 1.613252075473056, "learning_rate": 1.976011066873786e-05, "loss": 0.5433, "step": 2745 }, { "epoch": 0.09783208935283325, "grad_norm": 1.3604537397904342, "learning_rate": 1.975985936760251e-05, "loss": 0.5524, "step": 2746 }, { "epoch": 0.09786771647932736, "grad_norm": 1.7487702210185918, "learning_rate": 1.975960793650781e-05, "loss": 0.6323, "step": 2747 }, { "epoch": 0.09790334360582147, "grad_norm": 1.4623104045635702, "learning_rate": 1.9759356375457114e-05, "loss": 0.5403, "step": 2748 }, { "epoch": 0.09793897073231558, "grad_norm": 1.3167216845746872, "learning_rate": 1.975910468445376e-05, "loss": 0.5537, "step": 2749 }, { "epoch": 0.0979745978588097, "grad_norm": 1.7336002337578593, "learning_rate": 1.9758852863501108e-05, "loss": 0.5709, "step": 2750 }, { "epoch": 0.0980102249853038, "grad_norm": 1.78117331829862, "learning_rate": 1.975860091260251e-05, "loss": 0.6683, "step": 2751 }, { "epoch": 0.09804585211179792, "grad_norm": 1.1500815517132956, "learning_rate": 1.975834883176132e-05, "loss": 0.494, "step": 2752 }, { "epoch": 0.09808147923829204, "grad_norm": 4.355247581392041, "learning_rate": 1.9758096620980896e-05, "loss": 0.586, "step": 2753 }, { "epoch": 0.09811710636478614, "grad_norm": 1.7037900875891294, "learning_rate": 1.9757844280264594e-05, "loss": 0.6114, "step": 2754 }, { "epoch": 0.09815273349128026, "grad_norm": 1.9413955223716137, "learning_rate": 1.9757591809615772e-05, "loss": 0.609, "step": 2755 }, { "epoch": 0.09818836061777438, "grad_norm": 1.9609191236445203, "learning_rate": 1.97573392090378e-05, "loss": 0.584, "step": 2756 }, { "epoch": 0.09822398774426849, "grad_norm": 2.044385577544781, "learning_rate": 1.975708647853403e-05, "loss": 0.7161, "step": 2757 }, { "epoch": 0.0982596148707626, "grad_norm": 1.2771829517659714, "learning_rate": 1.975683361810784e-05, "loss": 0.5172, "step": 2758 }, { "epoch": 0.09829524199725671, "grad_norm": 1.2717186575179515, "learning_rate": 1.9756580627762587e-05, "loss": 0.5701, "step": 2759 }, { "epoch": 0.09833086912375083, "grad_norm": 1.1469305258451599, "learning_rate": 1.9756327507501647e-05, "loss": 0.4723, "step": 2760 }, { "epoch": 0.09836649625024493, "grad_norm": 1.3791584275911668, "learning_rate": 1.9756074257328387e-05, "loss": 0.5666, "step": 2761 }, { "epoch": 0.09840212337673905, "grad_norm": 1.3593785091211232, "learning_rate": 1.9755820877246175e-05, "loss": 0.565, "step": 2762 }, { "epoch": 0.09843775050323317, "grad_norm": 1.8690382395088125, "learning_rate": 1.975556736725839e-05, "loss": 0.6034, "step": 2763 }, { "epoch": 0.09847337762972727, "grad_norm": 2.1377853733052907, "learning_rate": 1.9755313727368408e-05, "loss": 0.6359, "step": 2764 }, { "epoch": 0.09850900475622139, "grad_norm": 1.909860390646999, "learning_rate": 1.9755059957579605e-05, "loss": 0.5681, "step": 2765 }, { "epoch": 0.0985446318827155, "grad_norm": 1.331253668197131, "learning_rate": 1.975480605789536e-05, "loss": 0.528, "step": 2766 }, { "epoch": 0.09858025900920961, "grad_norm": 1.3330755304217203, "learning_rate": 1.9754552028319057e-05, "loss": 0.5574, "step": 2767 }, { "epoch": 0.09861588613570373, "grad_norm": 1.4347288892858943, "learning_rate": 1.9754297868854075e-05, "loss": 0.5452, "step": 2768 }, { "epoch": 0.09865151326219784, "grad_norm": 1.7001463299872623, "learning_rate": 1.9754043579503793e-05, "loss": 0.6449, "step": 2769 }, { "epoch": 0.09868714038869195, "grad_norm": 1.421644919245047, "learning_rate": 1.975378916027161e-05, "loss": 0.5417, "step": 2770 }, { "epoch": 0.09872276751518606, "grad_norm": 1.2112639415315594, "learning_rate": 1.97535346111609e-05, "loss": 0.4866, "step": 2771 }, { "epoch": 0.09875839464168018, "grad_norm": 1.3223550156326114, "learning_rate": 1.9753279932175068e-05, "loss": 0.5325, "step": 2772 }, { "epoch": 0.09879402176817428, "grad_norm": 1.464802061181763, "learning_rate": 1.975302512331749e-05, "loss": 0.5583, "step": 2773 }, { "epoch": 0.0988296488946684, "grad_norm": 1.9519296705649996, "learning_rate": 1.975277018459157e-05, "loss": 0.6621, "step": 2774 }, { "epoch": 0.09886527602116252, "grad_norm": 1.2054842705463722, "learning_rate": 1.9752515116000693e-05, "loss": 0.5863, "step": 2775 }, { "epoch": 0.09890090314765662, "grad_norm": 1.2429928774277788, "learning_rate": 1.9752259917548266e-05, "loss": 0.5198, "step": 2776 }, { "epoch": 0.09893653027415074, "grad_norm": 1.3298986781243125, "learning_rate": 1.9752004589237678e-05, "loss": 0.566, "step": 2777 }, { "epoch": 0.09897215740064486, "grad_norm": 1.3164816056692386, "learning_rate": 1.9751749131072335e-05, "loss": 0.5338, "step": 2778 }, { "epoch": 0.09900778452713896, "grad_norm": 1.2441003329275406, "learning_rate": 1.9751493543055634e-05, "loss": 0.5461, "step": 2779 }, { "epoch": 0.09904341165363308, "grad_norm": 1.2925238497993672, "learning_rate": 1.9751237825190982e-05, "loss": 0.5447, "step": 2780 }, { "epoch": 0.0990790387801272, "grad_norm": 1.2699318199721723, "learning_rate": 1.9750981977481783e-05, "loss": 0.5422, "step": 2781 }, { "epoch": 0.0991146659066213, "grad_norm": 1.1951221458906416, "learning_rate": 1.9750725999931443e-05, "loss": 0.5668, "step": 2782 }, { "epoch": 0.09915029303311541, "grad_norm": 1.6915722266083182, "learning_rate": 1.9750469892543372e-05, "loss": 0.6649, "step": 2783 }, { "epoch": 0.09918592015960953, "grad_norm": 1.8370764663963108, "learning_rate": 1.9750213655320977e-05, "loss": 0.6197, "step": 2784 }, { "epoch": 0.09922154728610363, "grad_norm": 1.227665928264945, "learning_rate": 1.974995728826767e-05, "loss": 0.5433, "step": 2785 }, { "epoch": 0.09925717441259775, "grad_norm": 1.822633854136513, "learning_rate": 1.9749700791386872e-05, "loss": 0.7404, "step": 2786 }, { "epoch": 0.09929280153909187, "grad_norm": 1.1690415764464848, "learning_rate": 1.974944416468199e-05, "loss": 0.5219, "step": 2787 }, { "epoch": 0.09932842866558597, "grad_norm": 1.260398230044204, "learning_rate": 1.9749187408156448e-05, "loss": 0.5583, "step": 2788 }, { "epoch": 0.09936405579208009, "grad_norm": 1.676531820582552, "learning_rate": 1.974893052181366e-05, "loss": 0.6643, "step": 2789 }, { "epoch": 0.0993996829185742, "grad_norm": 1.2356567839394597, "learning_rate": 1.9748673505657045e-05, "loss": 0.4969, "step": 2790 }, { "epoch": 0.09943531004506831, "grad_norm": 1.3339452289960558, "learning_rate": 1.974841635969003e-05, "loss": 0.4804, "step": 2791 }, { "epoch": 0.09947093717156243, "grad_norm": 1.2777323854783984, "learning_rate": 1.9748159083916038e-05, "loss": 0.565, "step": 2792 }, { "epoch": 0.09950656429805654, "grad_norm": 1.2424688916739908, "learning_rate": 1.9747901678338496e-05, "loss": 0.5535, "step": 2793 }, { "epoch": 0.09954219142455065, "grad_norm": 2.0693919955477993, "learning_rate": 1.974764414296083e-05, "loss": 0.6652, "step": 2794 }, { "epoch": 0.09957781855104476, "grad_norm": 1.2625875479976447, "learning_rate": 1.9747386477786466e-05, "loss": 0.5443, "step": 2795 }, { "epoch": 0.09961344567753888, "grad_norm": 3.2916443989556186, "learning_rate": 1.974712868281884e-05, "loss": 0.6814, "step": 2796 }, { "epoch": 0.09964907280403298, "grad_norm": 2.0393545529234567, "learning_rate": 1.974687075806138e-05, "loss": 0.5709, "step": 2797 }, { "epoch": 0.0996846999305271, "grad_norm": 1.4392037402376505, "learning_rate": 1.9746612703517524e-05, "loss": 0.5203, "step": 2798 }, { "epoch": 0.09972032705702122, "grad_norm": 1.3624757207281413, "learning_rate": 1.974635451919071e-05, "loss": 0.5433, "step": 2799 }, { "epoch": 0.09975595418351532, "grad_norm": 1.631722124380877, "learning_rate": 1.9746096205084372e-05, "loss": 0.6129, "step": 2800 }, { "epoch": 0.09979158131000944, "grad_norm": 1.8713092615287827, "learning_rate": 1.9745837761201948e-05, "loss": 0.6678, "step": 2801 }, { "epoch": 0.09982720843650356, "grad_norm": 2.292457453063141, "learning_rate": 1.974557918754689e-05, "loss": 0.6648, "step": 2802 }, { "epoch": 0.09986283556299767, "grad_norm": 2.85760088330765, "learning_rate": 1.9745320484122626e-05, "loss": 0.701, "step": 2803 }, { "epoch": 0.09989846268949178, "grad_norm": 1.992829827612355, "learning_rate": 1.974506165093261e-05, "loss": 0.6742, "step": 2804 }, { "epoch": 0.0999340898159859, "grad_norm": 1.427194488466464, "learning_rate": 1.9744802687980285e-05, "loss": 0.5652, "step": 2805 }, { "epoch": 0.09996971694248001, "grad_norm": 1.698579305737714, "learning_rate": 1.974454359526911e-05, "loss": 0.5759, "step": 2806 }, { "epoch": 0.10000534406897411, "grad_norm": 1.376223076585275, "learning_rate": 1.9744284372802517e-05, "loss": 0.5167, "step": 2807 }, { "epoch": 0.10004097119546823, "grad_norm": 1.292672730990222, "learning_rate": 1.974402502058397e-05, "loss": 0.5269, "step": 2808 }, { "epoch": 0.10007659832196235, "grad_norm": 2.55371898461386, "learning_rate": 1.9743765538616918e-05, "loss": 0.6071, "step": 2809 }, { "epoch": 0.10011222544845645, "grad_norm": 1.678001037656363, "learning_rate": 1.9743505926904815e-05, "loss": 0.6179, "step": 2810 }, { "epoch": 0.10014785257495057, "grad_norm": 1.8994012492673091, "learning_rate": 1.9743246185451125e-05, "loss": 0.5396, "step": 2811 }, { "epoch": 0.10018347970144469, "grad_norm": 1.8746212800818614, "learning_rate": 1.97429863142593e-05, "loss": 0.6555, "step": 2812 }, { "epoch": 0.10021910682793879, "grad_norm": 1.372886054165347, "learning_rate": 1.97427263133328e-05, "loss": 0.5499, "step": 2813 }, { "epoch": 0.1002547339544329, "grad_norm": 1.6445903114988951, "learning_rate": 1.9742466182675093e-05, "loss": 0.5087, "step": 2814 }, { "epoch": 0.10029036108092702, "grad_norm": 1.6230576965272854, "learning_rate": 1.974220592228964e-05, "loss": 0.5763, "step": 2815 }, { "epoch": 0.10032598820742113, "grad_norm": 3.4326091973756307, "learning_rate": 1.97419455321799e-05, "loss": 0.6567, "step": 2816 }, { "epoch": 0.10036161533391524, "grad_norm": 3.3589895804222594, "learning_rate": 1.9741685012349347e-05, "loss": 0.5815, "step": 2817 }, { "epoch": 0.10039724246040936, "grad_norm": 2.4670990328413236, "learning_rate": 1.9741424362801453e-05, "loss": 0.5835, "step": 2818 }, { "epoch": 0.10043286958690346, "grad_norm": 2.335419932536396, "learning_rate": 1.9741163583539685e-05, "loss": 0.5609, "step": 2819 }, { "epoch": 0.10046849671339758, "grad_norm": 2.4635627730637015, "learning_rate": 1.974090267456751e-05, "loss": 0.587, "step": 2820 }, { "epoch": 0.1005041238398917, "grad_norm": 2.9031765671334173, "learning_rate": 1.974064163588841e-05, "loss": 0.5003, "step": 2821 }, { "epoch": 0.1005397509663858, "grad_norm": 2.095724775264993, "learning_rate": 1.9740380467505863e-05, "loss": 0.4902, "step": 2822 }, { "epoch": 0.10057537809287992, "grad_norm": 3.098002889074237, "learning_rate": 1.9740119169423337e-05, "loss": 0.5655, "step": 2823 }, { "epoch": 0.10061100521937404, "grad_norm": 2.841480483507556, "learning_rate": 1.9739857741644318e-05, "loss": 0.5845, "step": 2824 }, { "epoch": 0.10064663234586814, "grad_norm": 3.277243571378979, "learning_rate": 1.9739596184172284e-05, "loss": 0.5698, "step": 2825 }, { "epoch": 0.10068225947236226, "grad_norm": 1.98215214276474, "learning_rate": 1.973933449701072e-05, "loss": 0.5313, "step": 2826 }, { "epoch": 0.10071788659885637, "grad_norm": 2.1193413943901134, "learning_rate": 1.973907268016311e-05, "loss": 0.5042, "step": 2827 }, { "epoch": 0.10075351372535048, "grad_norm": 2.1210748028255417, "learning_rate": 1.9738810733632937e-05, "loss": 0.5734, "step": 2828 }, { "epoch": 0.1007891408518446, "grad_norm": 1.6393302260468419, "learning_rate": 1.9738548657423696e-05, "loss": 0.5328, "step": 2829 }, { "epoch": 0.10082476797833871, "grad_norm": 2.1180687856130276, "learning_rate": 1.9738286451538874e-05, "loss": 0.6807, "step": 2830 }, { "epoch": 0.10086039510483281, "grad_norm": 2.87012827993452, "learning_rate": 1.9738024115981957e-05, "loss": 0.6757, "step": 2831 }, { "epoch": 0.10089602223132693, "grad_norm": 2.7557764096366193, "learning_rate": 1.9737761650756446e-05, "loss": 0.5441, "step": 2832 }, { "epoch": 0.10093164935782105, "grad_norm": 2.331970639259649, "learning_rate": 1.973749905586583e-05, "loss": 0.4812, "step": 2833 }, { "epoch": 0.10096727648431515, "grad_norm": 2.5159506163255227, "learning_rate": 1.973723633131361e-05, "loss": 0.5032, "step": 2834 }, { "epoch": 0.10100290361080927, "grad_norm": 4.13151596999373, "learning_rate": 1.973697347710328e-05, "loss": 0.6584, "step": 2835 }, { "epoch": 0.10103853073730339, "grad_norm": 2.2621678687073428, "learning_rate": 1.9736710493238345e-05, "loss": 0.5248, "step": 2836 }, { "epoch": 0.10107415786379749, "grad_norm": 2.563180284939806, "learning_rate": 1.97364473797223e-05, "loss": 0.5655, "step": 2837 }, { "epoch": 0.1011097849902916, "grad_norm": 2.310259505631523, "learning_rate": 1.9736184136558657e-05, "loss": 0.5495, "step": 2838 }, { "epoch": 0.10114541211678572, "grad_norm": 1.8576213129928125, "learning_rate": 1.9735920763750916e-05, "loss": 0.4741, "step": 2839 }, { "epoch": 0.10118103924327983, "grad_norm": 1.7424476134116471, "learning_rate": 1.9735657261302587e-05, "loss": 0.5684, "step": 2840 }, { "epoch": 0.10121666636977394, "grad_norm": 2.213970622588308, "learning_rate": 1.973539362921717e-05, "loss": 0.6789, "step": 2841 }, { "epoch": 0.10125229349626806, "grad_norm": 1.8405012947115085, "learning_rate": 1.9735129867498188e-05, "loss": 0.5199, "step": 2842 }, { "epoch": 0.10128792062276216, "grad_norm": 1.9380015484606832, "learning_rate": 1.9734865976149145e-05, "loss": 0.523, "step": 2843 }, { "epoch": 0.10132354774925628, "grad_norm": 2.1754125044993002, "learning_rate": 1.973460195517356e-05, "loss": 0.5362, "step": 2844 }, { "epoch": 0.1013591748757504, "grad_norm": 3.1225277431201284, "learning_rate": 1.9734337804574943e-05, "loss": 0.6853, "step": 2845 }, { "epoch": 0.1013948020022445, "grad_norm": 2.9943864223791037, "learning_rate": 1.9734073524356817e-05, "loss": 0.6068, "step": 2846 }, { "epoch": 0.10143042912873862, "grad_norm": 1.8743539197503059, "learning_rate": 1.9733809114522695e-05, "loss": 0.6413, "step": 2847 }, { "epoch": 0.10146605625523274, "grad_norm": 1.6554719148431234, "learning_rate": 1.9733544575076103e-05, "loss": 0.6053, "step": 2848 }, { "epoch": 0.10150168338172684, "grad_norm": 2.108893818397921, "learning_rate": 1.9733279906020558e-05, "loss": 0.6777, "step": 2849 }, { "epoch": 0.10153731050822096, "grad_norm": 1.4666213138996675, "learning_rate": 1.9733015107359593e-05, "loss": 0.4663, "step": 2850 }, { "epoch": 0.10157293763471507, "grad_norm": 1.9108278753137646, "learning_rate": 1.9732750179096726e-05, "loss": 0.614, "step": 2851 }, { "epoch": 0.10160856476120919, "grad_norm": 1.5045131542273045, "learning_rate": 1.973248512123549e-05, "loss": 0.575, "step": 2852 }, { "epoch": 0.1016441918877033, "grad_norm": 1.3458803852961838, "learning_rate": 1.973221993377941e-05, "loss": 0.5461, "step": 2853 }, { "epoch": 0.10167981901419741, "grad_norm": 1.9952161872374405, "learning_rate": 1.9731954616732022e-05, "loss": 0.6966, "step": 2854 }, { "epoch": 0.10171544614069153, "grad_norm": 3.0749907879727427, "learning_rate": 1.9731689170096853e-05, "loss": 0.737, "step": 2855 }, { "epoch": 0.10175107326718563, "grad_norm": 2.1415858041505227, "learning_rate": 1.973142359387744e-05, "loss": 0.6359, "step": 2856 }, { "epoch": 0.10178670039367975, "grad_norm": 1.6533234359936766, "learning_rate": 1.9731157888077322e-05, "loss": 0.5491, "step": 2857 }, { "epoch": 0.10182232752017387, "grad_norm": 1.3400582758289628, "learning_rate": 1.9730892052700034e-05, "loss": 0.5363, "step": 2858 }, { "epoch": 0.10185795464666797, "grad_norm": 1.6142676072250184, "learning_rate": 1.973062608774912e-05, "loss": 0.6747, "step": 2859 }, { "epoch": 0.10189358177316209, "grad_norm": 1.8474264402375953, "learning_rate": 1.9730359993228114e-05, "loss": 0.6786, "step": 2860 }, { "epoch": 0.1019292088996562, "grad_norm": 1.7475099969037373, "learning_rate": 1.9730093769140567e-05, "loss": 0.6442, "step": 2861 }, { "epoch": 0.10196483602615031, "grad_norm": 1.376298362391628, "learning_rate": 1.9729827415490016e-05, "loss": 0.5313, "step": 2862 }, { "epoch": 0.10200046315264442, "grad_norm": 1.1481944330242895, "learning_rate": 1.9729560932280016e-05, "loss": 0.47, "step": 2863 }, { "epoch": 0.10203609027913854, "grad_norm": 2.0199836395540527, "learning_rate": 1.972929431951411e-05, "loss": 0.6161, "step": 2864 }, { "epoch": 0.10207171740563264, "grad_norm": 1.3839365570861264, "learning_rate": 1.972902757719585e-05, "loss": 0.5635, "step": 2865 }, { "epoch": 0.10210734453212676, "grad_norm": 2.209997145332939, "learning_rate": 1.972876070532879e-05, "loss": 0.6245, "step": 2866 }, { "epoch": 0.10214297165862088, "grad_norm": 1.671148185531875, "learning_rate": 1.972849370391648e-05, "loss": 0.7106, "step": 2867 }, { "epoch": 0.10217859878511498, "grad_norm": 2.3226216431192106, "learning_rate": 1.9728226572962474e-05, "loss": 0.6333, "step": 2868 }, { "epoch": 0.1022142259116091, "grad_norm": 1.3068414841523563, "learning_rate": 1.9727959312470335e-05, "loss": 0.532, "step": 2869 }, { "epoch": 0.10224985303810322, "grad_norm": 1.6799815733375643, "learning_rate": 1.9727691922443616e-05, "loss": 0.6352, "step": 2870 }, { "epoch": 0.10228548016459732, "grad_norm": 1.3316014242588727, "learning_rate": 1.972742440288588e-05, "loss": 0.5305, "step": 2871 }, { "epoch": 0.10232110729109144, "grad_norm": 1.9074387907184507, "learning_rate": 1.972715675380069e-05, "loss": 0.5546, "step": 2872 }, { "epoch": 0.10235673441758555, "grad_norm": 1.2831948725902527, "learning_rate": 1.9726888975191607e-05, "loss": 0.5483, "step": 2873 }, { "epoch": 0.10239236154407966, "grad_norm": 1.3717820124411562, "learning_rate": 1.9726621067062203e-05, "loss": 0.5427, "step": 2874 }, { "epoch": 0.10242798867057377, "grad_norm": 1.916330432085623, "learning_rate": 1.9726353029416037e-05, "loss": 0.6536, "step": 2875 }, { "epoch": 0.10246361579706789, "grad_norm": 2.036071065572201, "learning_rate": 1.972608486225668e-05, "loss": 0.7412, "step": 2876 }, { "epoch": 0.102499242923562, "grad_norm": 1.3582021055157467, "learning_rate": 1.972581656558771e-05, "loss": 0.5541, "step": 2877 }, { "epoch": 0.10253487005005611, "grad_norm": 1.7858447372865898, "learning_rate": 1.9725548139412693e-05, "loss": 0.6987, "step": 2878 }, { "epoch": 0.10257049717655023, "grad_norm": 1.31957794962313, "learning_rate": 1.9725279583735203e-05, "loss": 0.5104, "step": 2879 }, { "epoch": 0.10260612430304433, "grad_norm": 1.3185642878720352, "learning_rate": 1.972501089855882e-05, "loss": 0.5571, "step": 2880 }, { "epoch": 0.10264175142953845, "grad_norm": 1.7101740783855877, "learning_rate": 1.9724742083887117e-05, "loss": 0.6683, "step": 2881 }, { "epoch": 0.10267737855603257, "grad_norm": 1.7047288898702744, "learning_rate": 1.9724473139723682e-05, "loss": 0.589, "step": 2882 }, { "epoch": 0.10271300568252667, "grad_norm": 1.7641736459215074, "learning_rate": 1.9724204066072084e-05, "loss": 0.5913, "step": 2883 }, { "epoch": 0.10274863280902079, "grad_norm": 1.1978113176542986, "learning_rate": 1.9723934862935912e-05, "loss": 0.5435, "step": 2884 }, { "epoch": 0.1027842599355149, "grad_norm": 2.099652453562648, "learning_rate": 1.9723665530318753e-05, "loss": 0.6117, "step": 2885 }, { "epoch": 0.10281988706200901, "grad_norm": 1.8913416456279462, "learning_rate": 1.972339606822419e-05, "loss": 0.6633, "step": 2886 }, { "epoch": 0.10285551418850312, "grad_norm": 1.8671913367536435, "learning_rate": 1.9723126476655813e-05, "loss": 0.6064, "step": 2887 }, { "epoch": 0.10289114131499724, "grad_norm": 1.244850985005167, "learning_rate": 1.972285675561721e-05, "loss": 0.5164, "step": 2888 }, { "epoch": 0.10292676844149135, "grad_norm": 1.2606249984564244, "learning_rate": 1.9722586905111975e-05, "loss": 0.5161, "step": 2889 }, { "epoch": 0.10296239556798546, "grad_norm": 1.1907174427025768, "learning_rate": 1.97223169251437e-05, "loss": 0.5325, "step": 2890 }, { "epoch": 0.10299802269447958, "grad_norm": 1.939134895449908, "learning_rate": 1.9722046815715975e-05, "loss": 0.6891, "step": 2891 }, { "epoch": 0.10303364982097368, "grad_norm": 1.36714570779059, "learning_rate": 1.9721776576832408e-05, "loss": 0.5675, "step": 2892 }, { "epoch": 0.1030692769474678, "grad_norm": 1.6140723904381036, "learning_rate": 1.9721506208496585e-05, "loss": 0.6246, "step": 2893 }, { "epoch": 0.10310490407396192, "grad_norm": 1.757739439158781, "learning_rate": 1.9721235710712118e-05, "loss": 0.6254, "step": 2894 }, { "epoch": 0.10314053120045602, "grad_norm": 1.315382256364352, "learning_rate": 1.9720965083482596e-05, "loss": 0.5664, "step": 2895 }, { "epoch": 0.10317615832695014, "grad_norm": 1.3312753869013423, "learning_rate": 1.9720694326811633e-05, "loss": 0.5285, "step": 2896 }, { "epoch": 0.10321178545344425, "grad_norm": 2.451840714891548, "learning_rate": 1.972042344070283e-05, "loss": 0.6667, "step": 2897 }, { "epoch": 0.10324741257993837, "grad_norm": 1.4366560918633569, "learning_rate": 1.9720152425159792e-05, "loss": 0.5722, "step": 2898 }, { "epoch": 0.10328303970643248, "grad_norm": 1.121329187808989, "learning_rate": 1.9719881280186138e-05, "loss": 0.5034, "step": 2899 }, { "epoch": 0.10331866683292659, "grad_norm": 1.3828096015034652, "learning_rate": 1.9719610005785466e-05, "loss": 0.5543, "step": 2900 }, { "epoch": 0.10335429395942071, "grad_norm": 2.5584680633081196, "learning_rate": 1.9719338601961393e-05, "loss": 0.6397, "step": 2901 }, { "epoch": 0.10338992108591481, "grad_norm": 1.3230871136544764, "learning_rate": 1.9719067068717532e-05, "loss": 0.5223, "step": 2902 }, { "epoch": 0.10342554821240893, "grad_norm": 2.1024490126175475, "learning_rate": 1.97187954060575e-05, "loss": 0.645, "step": 2903 }, { "epoch": 0.10346117533890305, "grad_norm": 1.7084695015311695, "learning_rate": 1.9718523613984917e-05, "loss": 0.5964, "step": 2904 }, { "epoch": 0.10349680246539715, "grad_norm": 1.399056755295789, "learning_rate": 1.9718251692503397e-05, "loss": 0.5564, "step": 2905 }, { "epoch": 0.10353242959189127, "grad_norm": 2.127071090274679, "learning_rate": 1.9717979641616564e-05, "loss": 0.6529, "step": 2906 }, { "epoch": 0.10356805671838538, "grad_norm": 1.6013950739913878, "learning_rate": 1.971770746132804e-05, "loss": 0.622, "step": 2907 }, { "epoch": 0.10360368384487949, "grad_norm": 1.2577310053969009, "learning_rate": 1.971743515164145e-05, "loss": 0.4844, "step": 2908 }, { "epoch": 0.1036393109713736, "grad_norm": 1.6807626723265037, "learning_rate": 1.9717162712560417e-05, "loss": 0.5138, "step": 2909 }, { "epoch": 0.10367493809786772, "grad_norm": 1.8886798278388464, "learning_rate": 1.9716890144088568e-05, "loss": 0.5853, "step": 2910 }, { "epoch": 0.10371056522436183, "grad_norm": 1.2685023764459777, "learning_rate": 1.9716617446229537e-05, "loss": 0.5908, "step": 2911 }, { "epoch": 0.10374619235085594, "grad_norm": 2.8691861460638006, "learning_rate": 1.9716344618986955e-05, "loss": 0.6018, "step": 2912 }, { "epoch": 0.10378181947735006, "grad_norm": 1.566084638055672, "learning_rate": 1.9716071662364454e-05, "loss": 0.6037, "step": 2913 }, { "epoch": 0.10381744660384416, "grad_norm": 1.4181640543737022, "learning_rate": 1.9715798576365663e-05, "loss": 0.5291, "step": 2914 }, { "epoch": 0.10385307373033828, "grad_norm": 2.75021563914783, "learning_rate": 1.9715525360994227e-05, "loss": 0.5617, "step": 2915 }, { "epoch": 0.1038887008568324, "grad_norm": 2.0544378795336145, "learning_rate": 1.9715252016253776e-05, "loss": 0.7395, "step": 2916 }, { "epoch": 0.1039243279833265, "grad_norm": 1.420826754434303, "learning_rate": 1.9714978542147958e-05, "loss": 0.6102, "step": 2917 }, { "epoch": 0.10395995510982062, "grad_norm": 2.1788889847724238, "learning_rate": 1.9714704938680408e-05, "loss": 0.6907, "step": 2918 }, { "epoch": 0.10399558223631473, "grad_norm": 1.8643725366191473, "learning_rate": 1.9714431205854773e-05, "loss": 0.62, "step": 2919 }, { "epoch": 0.10403120936280884, "grad_norm": 1.8912907900527125, "learning_rate": 1.9714157343674695e-05, "loss": 0.6038, "step": 2920 }, { "epoch": 0.10406683648930296, "grad_norm": 2.031213951616843, "learning_rate": 1.9713883352143825e-05, "loss": 0.6347, "step": 2921 }, { "epoch": 0.10410246361579707, "grad_norm": 1.3518712628443166, "learning_rate": 1.9713609231265807e-05, "loss": 0.5789, "step": 2922 }, { "epoch": 0.10413809074229118, "grad_norm": 1.2748902967335527, "learning_rate": 1.9713334981044293e-05, "loss": 0.5352, "step": 2923 }, { "epoch": 0.10417371786878529, "grad_norm": 2.074649160620621, "learning_rate": 1.9713060601482934e-05, "loss": 0.7066, "step": 2924 }, { "epoch": 0.10420934499527941, "grad_norm": 1.1348545531546188, "learning_rate": 1.9712786092585384e-05, "loss": 0.5149, "step": 2925 }, { "epoch": 0.10424497212177351, "grad_norm": 2.0666439767571596, "learning_rate": 1.9712511454355298e-05, "loss": 0.6342, "step": 2926 }, { "epoch": 0.10428059924826763, "grad_norm": 1.1770580324033346, "learning_rate": 1.9712236686796332e-05, "loss": 0.5208, "step": 2927 }, { "epoch": 0.10431622637476175, "grad_norm": 1.340138372664536, "learning_rate": 1.971196178991215e-05, "loss": 0.5813, "step": 2928 }, { "epoch": 0.10435185350125585, "grad_norm": 1.6847417136145635, "learning_rate": 1.9711686763706404e-05, "loss": 0.5541, "step": 2929 }, { "epoch": 0.10438748062774997, "grad_norm": 1.2919749152022066, "learning_rate": 1.9711411608182764e-05, "loss": 0.5241, "step": 2930 }, { "epoch": 0.10442310775424408, "grad_norm": 1.1861291885281366, "learning_rate": 1.971113632334489e-05, "loss": 0.5161, "step": 2931 }, { "epoch": 0.10445873488073819, "grad_norm": 2.1865331200444427, "learning_rate": 1.9710860909196446e-05, "loss": 0.6384, "step": 2932 }, { "epoch": 0.1044943620072323, "grad_norm": 1.3053181810466334, "learning_rate": 1.9710585365741105e-05, "loss": 0.5299, "step": 2933 }, { "epoch": 0.10452998913372642, "grad_norm": 2.35335303706536, "learning_rate": 1.971030969298253e-05, "loss": 0.5193, "step": 2934 }, { "epoch": 0.10456561626022053, "grad_norm": 1.2301325170365212, "learning_rate": 1.9710033890924395e-05, "loss": 0.5437, "step": 2935 }, { "epoch": 0.10460124338671464, "grad_norm": 3.337801491873638, "learning_rate": 1.9709757959570373e-05, "loss": 0.6187, "step": 2936 }, { "epoch": 0.10463687051320876, "grad_norm": 1.1501475467884292, "learning_rate": 1.9709481898924133e-05, "loss": 0.5544, "step": 2937 }, { "epoch": 0.10467249763970286, "grad_norm": 1.244910729120703, "learning_rate": 1.970920570898936e-05, "loss": 0.5162, "step": 2938 }, { "epoch": 0.10470812476619698, "grad_norm": 1.1832877303019567, "learning_rate": 1.9708929389769725e-05, "loss": 0.5211, "step": 2939 }, { "epoch": 0.1047437518926911, "grad_norm": 1.1341927504708278, "learning_rate": 1.970865294126891e-05, "loss": 0.545, "step": 2940 }, { "epoch": 0.1047793790191852, "grad_norm": 1.1738604609000678, "learning_rate": 1.9708376363490593e-05, "loss": 0.5651, "step": 2941 }, { "epoch": 0.10481500614567932, "grad_norm": 1.7050350820554387, "learning_rate": 1.970809965643846e-05, "loss": 0.665, "step": 2942 }, { "epoch": 0.10485063327217344, "grad_norm": 1.2394150204532977, "learning_rate": 1.9707822820116193e-05, "loss": 0.5751, "step": 2943 }, { "epoch": 0.10488626039866755, "grad_norm": 2.327738206393993, "learning_rate": 1.970754585452748e-05, "loss": 0.6535, "step": 2944 }, { "epoch": 0.10492188752516166, "grad_norm": 1.9642121570053872, "learning_rate": 1.9707268759676013e-05, "loss": 0.6346, "step": 2945 }, { "epoch": 0.10495751465165577, "grad_norm": 1.175561723710873, "learning_rate": 1.9706991535565473e-05, "loss": 0.5232, "step": 2946 }, { "epoch": 0.10499314177814989, "grad_norm": 1.8796525246497657, "learning_rate": 1.970671418219956e-05, "loss": 0.6322, "step": 2947 }, { "epoch": 0.105028768904644, "grad_norm": 2.562696977876172, "learning_rate": 1.970643669958196e-05, "loss": 0.6422, "step": 2948 }, { "epoch": 0.10506439603113811, "grad_norm": 1.1991700058318446, "learning_rate": 1.970615908771637e-05, "loss": 0.5456, "step": 2949 }, { "epoch": 0.10510002315763223, "grad_norm": 1.7851569125403621, "learning_rate": 1.9705881346606487e-05, "loss": 0.6934, "step": 2950 }, { "epoch": 0.10513565028412633, "grad_norm": 1.5922326630366466, "learning_rate": 1.9705603476256012e-05, "loss": 0.606, "step": 2951 }, { "epoch": 0.10517127741062045, "grad_norm": 2.1416491730683753, "learning_rate": 1.9705325476668646e-05, "loss": 0.6377, "step": 2952 }, { "epoch": 0.10520690453711457, "grad_norm": 1.7446001362011931, "learning_rate": 1.9705047347848083e-05, "loss": 0.6861, "step": 2953 }, { "epoch": 0.10524253166360867, "grad_norm": 1.678609728375129, "learning_rate": 1.9704769089798032e-05, "loss": 0.5777, "step": 2954 }, { "epoch": 0.10527815879010279, "grad_norm": 2.328681861992256, "learning_rate": 1.9704490702522198e-05, "loss": 0.658, "step": 2955 }, { "epoch": 0.1053137859165969, "grad_norm": 1.2575390776489364, "learning_rate": 1.9704212186024288e-05, "loss": 0.5693, "step": 2956 }, { "epoch": 0.105349413043091, "grad_norm": 2.4724812818488076, "learning_rate": 1.970393354030801e-05, "loss": 0.6473, "step": 2957 }, { "epoch": 0.10538504016958512, "grad_norm": 1.9312755789334428, "learning_rate": 1.970365476537707e-05, "loss": 0.6659, "step": 2958 }, { "epoch": 0.10542066729607924, "grad_norm": 1.8503942141278662, "learning_rate": 1.970337586123519e-05, "loss": 0.6167, "step": 2959 }, { "epoch": 0.10545629442257334, "grad_norm": 1.8577318123674549, "learning_rate": 1.9703096827886076e-05, "loss": 0.6392, "step": 2960 }, { "epoch": 0.10549192154906746, "grad_norm": 1.5036162826144965, "learning_rate": 1.9702817665333443e-05, "loss": 0.5226, "step": 2961 }, { "epoch": 0.10552754867556158, "grad_norm": 1.893817478987917, "learning_rate": 1.9702538373581012e-05, "loss": 0.6424, "step": 2962 }, { "epoch": 0.10556317580205568, "grad_norm": 2.0368109851316922, "learning_rate": 1.9702258952632505e-05, "loss": 0.6444, "step": 2963 }, { "epoch": 0.1055988029285498, "grad_norm": 1.8374672014750733, "learning_rate": 1.9701979402491637e-05, "loss": 0.6729, "step": 2964 }, { "epoch": 0.10563443005504392, "grad_norm": 2.005170051660627, "learning_rate": 1.970169972316213e-05, "loss": 0.5454, "step": 2965 }, { "epoch": 0.10567005718153802, "grad_norm": 2.03347301779943, "learning_rate": 1.970141991464771e-05, "loss": 0.6784, "step": 2966 }, { "epoch": 0.10570568430803214, "grad_norm": 1.730549263055093, "learning_rate": 1.9701139976952104e-05, "loss": 0.5575, "step": 2967 }, { "epoch": 0.10574131143452625, "grad_norm": 1.2219493372810133, "learning_rate": 1.9700859910079037e-05, "loss": 0.5128, "step": 2968 }, { "epoch": 0.10577693856102036, "grad_norm": 1.2466463315924885, "learning_rate": 1.9700579714032242e-05, "loss": 0.5261, "step": 2969 }, { "epoch": 0.10581256568751447, "grad_norm": 1.7946215315658895, "learning_rate": 1.9700299388815447e-05, "loss": 0.5393, "step": 2970 }, { "epoch": 0.10584819281400859, "grad_norm": 1.7059505543954545, "learning_rate": 1.9700018934432385e-05, "loss": 0.6031, "step": 2971 }, { "epoch": 0.1058838199405027, "grad_norm": 1.3087185845964693, "learning_rate": 1.969973835088679e-05, "loss": 0.6006, "step": 2972 }, { "epoch": 0.10591944706699681, "grad_norm": 1.8427051711604512, "learning_rate": 1.9699457638182404e-05, "loss": 0.5914, "step": 2973 }, { "epoch": 0.10595507419349093, "grad_norm": 2.2963020167152406, "learning_rate": 1.9699176796322955e-05, "loss": 0.6309, "step": 2974 }, { "epoch": 0.10599070131998503, "grad_norm": 1.734704683082126, "learning_rate": 1.969889582531219e-05, "loss": 0.6987, "step": 2975 }, { "epoch": 0.10602632844647915, "grad_norm": 2.217362754492299, "learning_rate": 1.9698614725153847e-05, "loss": 0.7481, "step": 2976 }, { "epoch": 0.10606195557297327, "grad_norm": 1.570096114351853, "learning_rate": 1.9698333495851672e-05, "loss": 0.6197, "step": 2977 }, { "epoch": 0.10609758269946737, "grad_norm": 1.6042913793233045, "learning_rate": 1.9698052137409404e-05, "loss": 0.5501, "step": 2978 }, { "epoch": 0.10613320982596149, "grad_norm": 1.7927699619004023, "learning_rate": 1.9697770649830796e-05, "loss": 0.5346, "step": 2979 }, { "epoch": 0.1061688369524556, "grad_norm": 1.3921880340446304, "learning_rate": 1.9697489033119594e-05, "loss": 0.5926, "step": 2980 }, { "epoch": 0.1062044640789497, "grad_norm": 1.9304931937939183, "learning_rate": 1.9697207287279543e-05, "loss": 0.5916, "step": 2981 }, { "epoch": 0.10624009120544382, "grad_norm": 2.0374673487750643, "learning_rate": 1.9696925412314405e-05, "loss": 0.6924, "step": 2982 }, { "epoch": 0.10627571833193794, "grad_norm": 2.0826009404544683, "learning_rate": 1.9696643408227923e-05, "loss": 0.7088, "step": 2983 }, { "epoch": 0.10631134545843204, "grad_norm": 2.020454880028597, "learning_rate": 1.9696361275023856e-05, "loss": 0.6586, "step": 2984 }, { "epoch": 0.10634697258492616, "grad_norm": 1.2622131886080414, "learning_rate": 1.9696079012705964e-05, "loss": 0.5376, "step": 2985 }, { "epoch": 0.10638259971142028, "grad_norm": 1.2224804239926041, "learning_rate": 1.9695796621278004e-05, "loss": 0.5607, "step": 2986 }, { "epoch": 0.10641822683791438, "grad_norm": 1.341954506762387, "learning_rate": 1.9695514100743734e-05, "loss": 0.5895, "step": 2987 }, { "epoch": 0.1064538539644085, "grad_norm": 1.8232925319435205, "learning_rate": 1.9695231451106914e-05, "loss": 0.5098, "step": 2988 }, { "epoch": 0.10648948109090262, "grad_norm": 1.2371631676600994, "learning_rate": 1.9694948672371313e-05, "loss": 0.5208, "step": 2989 }, { "epoch": 0.10652510821739672, "grad_norm": 1.1758982045067463, "learning_rate": 1.9694665764540694e-05, "loss": 0.4458, "step": 2990 }, { "epoch": 0.10656073534389084, "grad_norm": 1.3108494101675234, "learning_rate": 1.9694382727618825e-05, "loss": 0.5279, "step": 2991 }, { "epoch": 0.10659636247038495, "grad_norm": 2.0884463844006476, "learning_rate": 1.9694099561609474e-05, "loss": 0.6383, "step": 2992 }, { "epoch": 0.10663198959687907, "grad_norm": 1.3783558667832359, "learning_rate": 1.9693816266516408e-05, "loss": 0.5605, "step": 2993 }, { "epoch": 0.10666761672337317, "grad_norm": 2.082741028571308, "learning_rate": 1.9693532842343405e-05, "loss": 0.6715, "step": 2994 }, { "epoch": 0.10670324384986729, "grad_norm": 1.3136315638306508, "learning_rate": 1.969324928909424e-05, "loss": 0.5089, "step": 2995 }, { "epoch": 0.10673887097636141, "grad_norm": 1.80761112619167, "learning_rate": 1.9692965606772682e-05, "loss": 0.5737, "step": 2996 }, { "epoch": 0.10677449810285551, "grad_norm": 2.435709136921201, "learning_rate": 1.969268179538251e-05, "loss": 0.6284, "step": 2997 }, { "epoch": 0.10681012522934963, "grad_norm": 1.7857648630890743, "learning_rate": 1.969239785492751e-05, "loss": 0.575, "step": 2998 }, { "epoch": 0.10684575235584375, "grad_norm": 1.9920385251318489, "learning_rate": 1.9692113785411456e-05, "loss": 0.6692, "step": 2999 }, { "epoch": 0.10688137948233785, "grad_norm": 1.728252767339061, "learning_rate": 1.9691829586838134e-05, "loss": 0.5663, "step": 3000 }, { "epoch": 0.10691700660883197, "grad_norm": 1.2600487981434452, "learning_rate": 1.9691545259211322e-05, "loss": 0.5682, "step": 3001 }, { "epoch": 0.10695263373532608, "grad_norm": 1.7979081296792876, "learning_rate": 1.9691260802534816e-05, "loss": 0.6714, "step": 3002 }, { "epoch": 0.10698826086182019, "grad_norm": 2.1533878215827444, "learning_rate": 1.9690976216812397e-05, "loss": 0.5917, "step": 3003 }, { "epoch": 0.1070238879883143, "grad_norm": 2.2301482631223237, "learning_rate": 1.9690691502047852e-05, "loss": 0.7018, "step": 3004 }, { "epoch": 0.10705951511480842, "grad_norm": 1.308559262828316, "learning_rate": 1.9690406658244984e-05, "loss": 0.5262, "step": 3005 }, { "epoch": 0.10709514224130252, "grad_norm": 1.5496924238063527, "learning_rate": 1.9690121685407572e-05, "loss": 0.556, "step": 3006 }, { "epoch": 0.10713076936779664, "grad_norm": 1.68863472610624, "learning_rate": 1.968983658353942e-05, "loss": 0.5791, "step": 3007 }, { "epoch": 0.10716639649429076, "grad_norm": 1.2728308585742287, "learning_rate": 1.968955135264432e-05, "loss": 0.5379, "step": 3008 }, { "epoch": 0.10720202362078486, "grad_norm": 1.2911075442466518, "learning_rate": 1.9689265992726072e-05, "loss": 0.5376, "step": 3009 }, { "epoch": 0.10723765074727898, "grad_norm": 1.14944817704984, "learning_rate": 1.9688980503788474e-05, "loss": 0.5418, "step": 3010 }, { "epoch": 0.1072732778737731, "grad_norm": 1.1911853061333326, "learning_rate": 1.968869488583533e-05, "loss": 0.5578, "step": 3011 }, { "epoch": 0.1073089050002672, "grad_norm": 1.2123042558700456, "learning_rate": 1.9688409138870443e-05, "loss": 0.486, "step": 3012 }, { "epoch": 0.10734453212676132, "grad_norm": 1.184859255862531, "learning_rate": 1.968812326289761e-05, "loss": 0.5666, "step": 3013 }, { "epoch": 0.10738015925325543, "grad_norm": 1.2916195089051485, "learning_rate": 1.968783725792065e-05, "loss": 0.5842, "step": 3014 }, { "epoch": 0.10741578637974954, "grad_norm": 1.5552214698664815, "learning_rate": 1.968755112394336e-05, "loss": 0.5905, "step": 3015 }, { "epoch": 0.10745141350624365, "grad_norm": 1.8111694546501842, "learning_rate": 1.968726486096956e-05, "loss": 0.6099, "step": 3016 }, { "epoch": 0.10748704063273777, "grad_norm": 1.2737970792737632, "learning_rate": 1.9686978469003057e-05, "loss": 0.5457, "step": 3017 }, { "epoch": 0.10752266775923187, "grad_norm": 2.0451385856793647, "learning_rate": 1.9686691948047665e-05, "loss": 0.6731, "step": 3018 }, { "epoch": 0.10755829488572599, "grad_norm": 1.4263118957174916, "learning_rate": 1.96864052981072e-05, "loss": 0.5382, "step": 3019 }, { "epoch": 0.10759392201222011, "grad_norm": 2.193006769492594, "learning_rate": 1.9686118519185474e-05, "loss": 0.8129, "step": 3020 }, { "epoch": 0.10762954913871421, "grad_norm": 1.8422359340003616, "learning_rate": 1.9685831611286312e-05, "loss": 0.5697, "step": 3021 }, { "epoch": 0.10766517626520833, "grad_norm": 1.303970801557344, "learning_rate": 1.9685544574413533e-05, "loss": 0.5617, "step": 3022 }, { "epoch": 0.10770080339170245, "grad_norm": 1.273251128857507, "learning_rate": 1.968525740857096e-05, "loss": 0.5111, "step": 3023 }, { "epoch": 0.10773643051819655, "grad_norm": 1.2249403599575543, "learning_rate": 1.968497011376241e-05, "loss": 0.5424, "step": 3024 }, { "epoch": 0.10777205764469067, "grad_norm": 1.1983924162264774, "learning_rate": 1.9684682689991716e-05, "loss": 0.5198, "step": 3025 }, { "epoch": 0.10780768477118478, "grad_norm": 1.2837713008062968, "learning_rate": 1.9684395137262703e-05, "loss": 0.5401, "step": 3026 }, { "epoch": 0.10784331189767889, "grad_norm": 2.1866579979322913, "learning_rate": 1.96841074555792e-05, "loss": 0.5854, "step": 3027 }, { "epoch": 0.107878939024173, "grad_norm": 2.057554659406607, "learning_rate": 1.9683819644945036e-05, "loss": 0.5974, "step": 3028 }, { "epoch": 0.10791456615066712, "grad_norm": 2.136373409583247, "learning_rate": 1.9683531705364048e-05, "loss": 0.6045, "step": 3029 }, { "epoch": 0.10795019327716122, "grad_norm": 1.485688544507517, "learning_rate": 1.9683243636840064e-05, "loss": 0.6409, "step": 3030 }, { "epoch": 0.10798582040365534, "grad_norm": 1.146945980459061, "learning_rate": 1.9682955439376923e-05, "loss": 0.4905, "step": 3031 }, { "epoch": 0.10802144753014946, "grad_norm": 2.239236695711473, "learning_rate": 1.9682667112978464e-05, "loss": 0.5002, "step": 3032 }, { "epoch": 0.10805707465664356, "grad_norm": 1.7627770043548536, "learning_rate": 1.9682378657648523e-05, "loss": 0.6767, "step": 3033 }, { "epoch": 0.10809270178313768, "grad_norm": 2.3229112744887175, "learning_rate": 1.968209007339094e-05, "loss": 0.602, "step": 3034 }, { "epoch": 0.1081283289096318, "grad_norm": 1.6983588234228413, "learning_rate": 1.9681801360209568e-05, "loss": 0.6266, "step": 3035 }, { "epoch": 0.1081639560361259, "grad_norm": 1.3203443532860242, "learning_rate": 1.968151251810824e-05, "loss": 0.531, "step": 3036 }, { "epoch": 0.10819958316262002, "grad_norm": 1.8878955238238608, "learning_rate": 1.96812235470908e-05, "loss": 0.6572, "step": 3037 }, { "epoch": 0.10823521028911413, "grad_norm": 1.7706176175876338, "learning_rate": 1.968093444716111e-05, "loss": 0.5663, "step": 3038 }, { "epoch": 0.10827083741560825, "grad_norm": 1.44738934290393, "learning_rate": 1.9680645218323005e-05, "loss": 0.5153, "step": 3039 }, { "epoch": 0.10830646454210235, "grad_norm": 2.135067442133885, "learning_rate": 1.968035586058035e-05, "loss": 0.5315, "step": 3040 }, { "epoch": 0.10834209166859647, "grad_norm": 1.1830759869473635, "learning_rate": 1.9680066373936984e-05, "loss": 0.5113, "step": 3041 }, { "epoch": 0.10837771879509059, "grad_norm": 1.1262303818801624, "learning_rate": 1.9679776758396772e-05, "loss": 0.5281, "step": 3042 }, { "epoch": 0.10841334592158469, "grad_norm": 1.777829182521016, "learning_rate": 1.9679487013963566e-05, "loss": 0.6484, "step": 3043 }, { "epoch": 0.10844897304807881, "grad_norm": 1.173961619472958, "learning_rate": 1.9679197140641223e-05, "loss": 0.5442, "step": 3044 }, { "epoch": 0.10848460017457293, "grad_norm": 1.9872721768361878, "learning_rate": 1.9678907138433605e-05, "loss": 0.7243, "step": 3045 }, { "epoch": 0.10852022730106703, "grad_norm": 1.76417083407615, "learning_rate": 1.9678617007344573e-05, "loss": 0.6343, "step": 3046 }, { "epoch": 0.10855585442756115, "grad_norm": 1.6004758361756921, "learning_rate": 1.967832674737799e-05, "loss": 0.6567, "step": 3047 }, { "epoch": 0.10859148155405526, "grad_norm": 1.9503335994960092, "learning_rate": 1.9678036358537726e-05, "loss": 0.6184, "step": 3048 }, { "epoch": 0.10862710868054937, "grad_norm": 1.8083778320409258, "learning_rate": 1.967774584082764e-05, "loss": 0.6313, "step": 3049 }, { "epoch": 0.10866273580704348, "grad_norm": 1.9173529647367173, "learning_rate": 1.9677455194251603e-05, "loss": 0.6424, "step": 3050 }, { "epoch": 0.1086983629335376, "grad_norm": 2.022437202722988, "learning_rate": 1.967716441881349e-05, "loss": 0.6736, "step": 3051 }, { "epoch": 0.1087339900600317, "grad_norm": 1.9555759459989523, "learning_rate": 1.9676873514517164e-05, "loss": 0.6543, "step": 3052 }, { "epoch": 0.10876961718652582, "grad_norm": 1.962880402721094, "learning_rate": 1.9676582481366508e-05, "loss": 0.5849, "step": 3053 }, { "epoch": 0.10880524431301994, "grad_norm": 2.2660794197767924, "learning_rate": 1.967629131936539e-05, "loss": 0.5575, "step": 3054 }, { "epoch": 0.10884087143951404, "grad_norm": 2.2149872223888263, "learning_rate": 1.9676000028517688e-05, "loss": 0.6243, "step": 3055 }, { "epoch": 0.10887649856600816, "grad_norm": 1.2358248371693183, "learning_rate": 1.9675708608827286e-05, "loss": 0.5948, "step": 3056 }, { "epoch": 0.10891212569250228, "grad_norm": 1.1797793434210708, "learning_rate": 1.9675417060298058e-05, "loss": 0.5095, "step": 3057 }, { "epoch": 0.10894775281899638, "grad_norm": 2.053887210721778, "learning_rate": 1.9675125382933894e-05, "loss": 0.6175, "step": 3058 }, { "epoch": 0.1089833799454905, "grad_norm": 1.158227144982549, "learning_rate": 1.967483357673867e-05, "loss": 0.5276, "step": 3059 }, { "epoch": 0.10901900707198461, "grad_norm": 1.3053310872462507, "learning_rate": 1.9674541641716272e-05, "loss": 0.5641, "step": 3060 }, { "epoch": 0.10905463419847872, "grad_norm": 1.3823190513901642, "learning_rate": 1.9674249577870593e-05, "loss": 0.552, "step": 3061 }, { "epoch": 0.10909026132497283, "grad_norm": 1.5486239723094886, "learning_rate": 1.9673957385205516e-05, "loss": 0.6155, "step": 3062 }, { "epoch": 0.10912588845146695, "grad_norm": 1.2068571532912113, "learning_rate": 1.9673665063724937e-05, "loss": 0.5799, "step": 3063 }, { "epoch": 0.10916151557796105, "grad_norm": 1.2008574810551198, "learning_rate": 1.967337261343275e-05, "loss": 0.5397, "step": 3064 }, { "epoch": 0.10919714270445517, "grad_norm": 2.255130738102184, "learning_rate": 1.967308003433284e-05, "loss": 0.7288, "step": 3065 }, { "epoch": 0.10923276983094929, "grad_norm": 1.9375567555453779, "learning_rate": 1.9672787326429107e-05, "loss": 0.6219, "step": 3066 }, { "epoch": 0.10926839695744339, "grad_norm": 1.9936189280332428, "learning_rate": 1.9672494489725453e-05, "loss": 0.5642, "step": 3067 }, { "epoch": 0.10930402408393751, "grad_norm": 1.2386999674525119, "learning_rate": 1.9672201524225774e-05, "loss": 0.5375, "step": 3068 }, { "epoch": 0.10933965121043163, "grad_norm": 1.238652019647616, "learning_rate": 1.967190842993397e-05, "loss": 0.5435, "step": 3069 }, { "epoch": 0.10937527833692573, "grad_norm": 1.3256227069688706, "learning_rate": 1.9671615206853947e-05, "loss": 0.6014, "step": 3070 }, { "epoch": 0.10941090546341985, "grad_norm": 1.8282947062725725, "learning_rate": 1.9671321854989605e-05, "loss": 0.707, "step": 3071 }, { "epoch": 0.10944653258991396, "grad_norm": 1.941609266993558, "learning_rate": 1.9671028374344858e-05, "loss": 0.6236, "step": 3072 }, { "epoch": 0.10948215971640807, "grad_norm": 1.4529152418482867, "learning_rate": 1.9670734764923604e-05, "loss": 0.5976, "step": 3073 }, { "epoch": 0.10951778684290218, "grad_norm": 1.255944744248335, "learning_rate": 1.9670441026729757e-05, "loss": 0.4782, "step": 3074 }, { "epoch": 0.1095534139693963, "grad_norm": 2.2678998369243533, "learning_rate": 1.9670147159767228e-05, "loss": 0.6009, "step": 3075 }, { "epoch": 0.1095890410958904, "grad_norm": 1.2232087723157126, "learning_rate": 1.9669853164039935e-05, "loss": 0.5178, "step": 3076 }, { "epoch": 0.10962466822238452, "grad_norm": 1.8545856822875697, "learning_rate": 1.9669559039551783e-05, "loss": 0.5875, "step": 3077 }, { "epoch": 0.10966029534887864, "grad_norm": 1.328911598346269, "learning_rate": 1.9669264786306695e-05, "loss": 0.5449, "step": 3078 }, { "epoch": 0.10969592247537274, "grad_norm": 1.7275324908910723, "learning_rate": 1.966897040430859e-05, "loss": 0.5757, "step": 3079 }, { "epoch": 0.10973154960186686, "grad_norm": 1.245969133278096, "learning_rate": 1.9668675893561384e-05, "loss": 0.5845, "step": 3080 }, { "epoch": 0.10976717672836098, "grad_norm": 1.1793243879635986, "learning_rate": 1.9668381254069e-05, "loss": 0.5263, "step": 3081 }, { "epoch": 0.10980280385485508, "grad_norm": 2.166541076600148, "learning_rate": 1.9668086485835368e-05, "loss": 0.6579, "step": 3082 }, { "epoch": 0.1098384309813492, "grad_norm": 1.2270290677926274, "learning_rate": 1.96677915888644e-05, "loss": 0.5624, "step": 3083 }, { "epoch": 0.10987405810784331, "grad_norm": 1.176433812937299, "learning_rate": 1.966749656316003e-05, "loss": 0.4693, "step": 3084 }, { "epoch": 0.10990968523433743, "grad_norm": 1.162766793950152, "learning_rate": 1.9667201408726187e-05, "loss": 0.501, "step": 3085 }, { "epoch": 0.10994531236083153, "grad_norm": 1.7849688227856242, "learning_rate": 1.9666906125566803e-05, "loss": 0.56, "step": 3086 }, { "epoch": 0.10998093948732565, "grad_norm": 1.1517397601879016, "learning_rate": 1.9666610713685804e-05, "loss": 0.539, "step": 3087 }, { "epoch": 0.11001656661381977, "grad_norm": 1.2790073334810967, "learning_rate": 1.966631517308713e-05, "loss": 0.5649, "step": 3088 }, { "epoch": 0.11005219374031387, "grad_norm": 2.0894068474579286, "learning_rate": 1.966601950377471e-05, "loss": 0.644, "step": 3089 }, { "epoch": 0.11008782086680799, "grad_norm": 1.6970583541776565, "learning_rate": 1.9665723705752483e-05, "loss": 0.6115, "step": 3090 }, { "epoch": 0.1101234479933021, "grad_norm": 2.00608547994571, "learning_rate": 1.9665427779024394e-05, "loss": 0.6583, "step": 3091 }, { "epoch": 0.11015907511979621, "grad_norm": 1.6923015673795263, "learning_rate": 1.9665131723594376e-05, "loss": 0.6585, "step": 3092 }, { "epoch": 0.11019470224629033, "grad_norm": 1.254304464324591, "learning_rate": 1.966483553946637e-05, "loss": 0.4954, "step": 3093 }, { "epoch": 0.11023032937278444, "grad_norm": 1.322584500576241, "learning_rate": 1.9664539226644327e-05, "loss": 0.5469, "step": 3094 }, { "epoch": 0.11026595649927855, "grad_norm": 1.3458595572596415, "learning_rate": 1.9664242785132188e-05, "loss": 0.6232, "step": 3095 }, { "epoch": 0.11030158362577266, "grad_norm": 1.2151106014817785, "learning_rate": 1.9663946214933905e-05, "loss": 0.5002, "step": 3096 }, { "epoch": 0.11033721075226678, "grad_norm": 1.5146386883524585, "learning_rate": 1.9663649516053417e-05, "loss": 0.5654, "step": 3097 }, { "epoch": 0.11037283787876088, "grad_norm": 1.4003520437506922, "learning_rate": 1.9663352688494686e-05, "loss": 0.5204, "step": 3098 }, { "epoch": 0.110408465005255, "grad_norm": 1.1788942206962398, "learning_rate": 1.966305573226166e-05, "loss": 0.516, "step": 3099 }, { "epoch": 0.11044409213174912, "grad_norm": 1.8849662468441755, "learning_rate": 1.966275864735829e-05, "loss": 0.6351, "step": 3100 }, { "epoch": 0.11047971925824322, "grad_norm": 1.7748088751658753, "learning_rate": 1.9662461433788537e-05, "loss": 0.5388, "step": 3101 }, { "epoch": 0.11051534638473734, "grad_norm": 1.7741818375566003, "learning_rate": 1.9662164091556355e-05, "loss": 0.577, "step": 3102 }, { "epoch": 0.11055097351123146, "grad_norm": 1.623572203340103, "learning_rate": 1.9661866620665705e-05, "loss": 0.624, "step": 3103 }, { "epoch": 0.11058660063772556, "grad_norm": 1.5129639320933017, "learning_rate": 1.9661569021120546e-05, "loss": 0.5324, "step": 3104 }, { "epoch": 0.11062222776421968, "grad_norm": 1.6739684829791144, "learning_rate": 1.9661271292924845e-05, "loss": 0.6125, "step": 3105 }, { "epoch": 0.1106578548907138, "grad_norm": 1.975341901329705, "learning_rate": 1.9660973436082563e-05, "loss": 0.6106, "step": 3106 }, { "epoch": 0.1106934820172079, "grad_norm": 1.8667686924292624, "learning_rate": 1.9660675450597667e-05, "loss": 0.6482, "step": 3107 }, { "epoch": 0.11072910914370201, "grad_norm": 1.3690321977177506, "learning_rate": 1.9660377336474127e-05, "loss": 0.5039, "step": 3108 }, { "epoch": 0.11076473627019613, "grad_norm": 1.2991666049761759, "learning_rate": 1.9660079093715906e-05, "loss": 0.4969, "step": 3109 }, { "epoch": 0.11080036339669024, "grad_norm": 1.4300704630014993, "learning_rate": 1.9659780722326983e-05, "loss": 0.5756, "step": 3110 }, { "epoch": 0.11083599052318435, "grad_norm": 2.000948094809803, "learning_rate": 1.9659482222311327e-05, "loss": 0.6781, "step": 3111 }, { "epoch": 0.11087161764967847, "grad_norm": 1.5122489575206295, "learning_rate": 1.9659183593672914e-05, "loss": 0.5703, "step": 3112 }, { "epoch": 0.11090724477617257, "grad_norm": 1.209290467363776, "learning_rate": 1.965888483641572e-05, "loss": 0.5288, "step": 3113 }, { "epoch": 0.11094287190266669, "grad_norm": 1.8522625578394083, "learning_rate": 1.9658585950543725e-05, "loss": 0.6384, "step": 3114 }, { "epoch": 0.11097849902916081, "grad_norm": 1.7475631774538447, "learning_rate": 1.9658286936060907e-05, "loss": 0.6286, "step": 3115 }, { "epoch": 0.11101412615565491, "grad_norm": 1.9107546390894845, "learning_rate": 1.9657987792971244e-05, "loss": 0.5923, "step": 3116 }, { "epoch": 0.11104975328214903, "grad_norm": 1.210896767623191, "learning_rate": 1.9657688521278726e-05, "loss": 0.4912, "step": 3117 }, { "epoch": 0.11108538040864314, "grad_norm": 1.1917447439374258, "learning_rate": 1.9657389120987332e-05, "loss": 0.4905, "step": 3118 }, { "epoch": 0.11112100753513725, "grad_norm": 1.2523532442861784, "learning_rate": 1.9657089592101057e-05, "loss": 0.5465, "step": 3119 }, { "epoch": 0.11115663466163136, "grad_norm": 2.311272537098403, "learning_rate": 1.965678993462388e-05, "loss": 0.6602, "step": 3120 }, { "epoch": 0.11119226178812548, "grad_norm": 1.3215982735743186, "learning_rate": 1.9656490148559797e-05, "loss": 0.5952, "step": 3121 }, { "epoch": 0.11122788891461959, "grad_norm": 1.9303740946637893, "learning_rate": 1.9656190233912796e-05, "loss": 0.6428, "step": 3122 }, { "epoch": 0.1112635160411137, "grad_norm": 1.2723249809946038, "learning_rate": 1.9655890190686878e-05, "loss": 0.5562, "step": 3123 }, { "epoch": 0.11129914316760782, "grad_norm": 1.2887847738099731, "learning_rate": 1.9655590018886027e-05, "loss": 0.5497, "step": 3124 }, { "epoch": 0.11133477029410192, "grad_norm": 1.280443688555393, "learning_rate": 1.965528971851425e-05, "loss": 0.5387, "step": 3125 }, { "epoch": 0.11137039742059604, "grad_norm": 1.4710555472303999, "learning_rate": 1.965498928957554e-05, "loss": 0.6101, "step": 3126 }, { "epoch": 0.11140602454709016, "grad_norm": 1.7777941485667532, "learning_rate": 1.96546887320739e-05, "loss": 0.6245, "step": 3127 }, { "epoch": 0.11144165167358426, "grad_norm": 1.213863760091427, "learning_rate": 1.965438804601333e-05, "loss": 0.5667, "step": 3128 }, { "epoch": 0.11147727880007838, "grad_norm": 1.2865572365238942, "learning_rate": 1.9654087231397834e-05, "loss": 0.541, "step": 3129 }, { "epoch": 0.1115129059265725, "grad_norm": 1.1670171249700692, "learning_rate": 1.9653786288231424e-05, "loss": 0.5091, "step": 3130 }, { "epoch": 0.1115485330530666, "grad_norm": 1.331025132132891, "learning_rate": 1.9653485216518094e-05, "loss": 0.6239, "step": 3131 }, { "epoch": 0.11158416017956072, "grad_norm": 1.8429252371247369, "learning_rate": 1.9653184016261867e-05, "loss": 0.6323, "step": 3132 }, { "epoch": 0.11161978730605483, "grad_norm": 1.8315837405879032, "learning_rate": 1.9652882687466746e-05, "loss": 0.5758, "step": 3133 }, { "epoch": 0.11165541443254895, "grad_norm": 1.3145531666892614, "learning_rate": 1.9652581230136748e-05, "loss": 0.6015, "step": 3134 }, { "epoch": 0.11169104155904305, "grad_norm": 1.331870411245678, "learning_rate": 1.965227964427588e-05, "loss": 0.5597, "step": 3135 }, { "epoch": 0.11172666868553717, "grad_norm": 1.2527645197443422, "learning_rate": 1.9651977929888163e-05, "loss": 0.505, "step": 3136 }, { "epoch": 0.11176229581203129, "grad_norm": 1.275291766820519, "learning_rate": 1.9651676086977613e-05, "loss": 0.5472, "step": 3137 }, { "epoch": 0.11179792293852539, "grad_norm": 1.9231324558508516, "learning_rate": 1.9651374115548255e-05, "loss": 0.7004, "step": 3138 }, { "epoch": 0.11183355006501951, "grad_norm": 4.652426033040294, "learning_rate": 1.9651072015604098e-05, "loss": 0.6444, "step": 3139 }, { "epoch": 0.11186917719151362, "grad_norm": 2.113515534900049, "learning_rate": 1.9650769787149175e-05, "loss": 0.7666, "step": 3140 }, { "epoch": 0.11190480431800773, "grad_norm": 2.470679717147385, "learning_rate": 1.9650467430187503e-05, "loss": 0.6771, "step": 3141 }, { "epoch": 0.11194043144450185, "grad_norm": 1.693847168171765, "learning_rate": 1.9650164944723116e-05, "loss": 0.6099, "step": 3142 }, { "epoch": 0.11197605857099596, "grad_norm": 1.683250971437446, "learning_rate": 1.9649862330760036e-05, "loss": 0.5175, "step": 3143 }, { "epoch": 0.11201168569749007, "grad_norm": 1.6018424239363052, "learning_rate": 1.9649559588302295e-05, "loss": 0.6608, "step": 3144 }, { "epoch": 0.11204731282398418, "grad_norm": 1.5205071109247619, "learning_rate": 1.9649256717353926e-05, "loss": 0.6801, "step": 3145 }, { "epoch": 0.1120829399504783, "grad_norm": 1.6566835479932445, "learning_rate": 1.9648953717918955e-05, "loss": 0.5407, "step": 3146 }, { "epoch": 0.1121185670769724, "grad_norm": 1.1776963517216474, "learning_rate": 1.964865059000142e-05, "loss": 0.5045, "step": 3147 }, { "epoch": 0.11215419420346652, "grad_norm": 1.2989064520364608, "learning_rate": 1.964834733360536e-05, "loss": 0.4894, "step": 3148 }, { "epoch": 0.11218982132996064, "grad_norm": 1.2521379927993137, "learning_rate": 1.9648043948734814e-05, "loss": 0.5478, "step": 3149 }, { "epoch": 0.11222544845645474, "grad_norm": 1.7939201575299537, "learning_rate": 1.9647740435393818e-05, "loss": 0.5832, "step": 3150 }, { "epoch": 0.11226107558294886, "grad_norm": 1.3590545301871588, "learning_rate": 1.9647436793586415e-05, "loss": 0.5737, "step": 3151 }, { "epoch": 0.11229670270944297, "grad_norm": 1.1829242751335942, "learning_rate": 1.9647133023316646e-05, "loss": 0.5704, "step": 3152 }, { "epoch": 0.11233232983593708, "grad_norm": 1.7361251972579594, "learning_rate": 1.964682912458856e-05, "loss": 0.5579, "step": 3153 }, { "epoch": 0.1123679569624312, "grad_norm": 1.2461955025387925, "learning_rate": 1.9646525097406198e-05, "loss": 0.492, "step": 3154 }, { "epoch": 0.11240358408892531, "grad_norm": 1.5257351151813856, "learning_rate": 1.9646220941773616e-05, "loss": 0.5821, "step": 3155 }, { "epoch": 0.11243921121541942, "grad_norm": 1.4509140619133767, "learning_rate": 1.9645916657694858e-05, "loss": 0.5919, "step": 3156 }, { "epoch": 0.11247483834191353, "grad_norm": 1.6072267646680982, "learning_rate": 1.9645612245173978e-05, "loss": 0.625, "step": 3157 }, { "epoch": 0.11251046546840765, "grad_norm": 1.2775677586347702, "learning_rate": 1.964530770421503e-05, "loss": 0.5259, "step": 3158 }, { "epoch": 0.11254609259490175, "grad_norm": 2.129431108318402, "learning_rate": 1.9645003034822064e-05, "loss": 0.6284, "step": 3159 }, { "epoch": 0.11258171972139587, "grad_norm": 1.2362558572451467, "learning_rate": 1.9644698236999144e-05, "loss": 0.5374, "step": 3160 }, { "epoch": 0.11261734684788999, "grad_norm": 2.638217649305535, "learning_rate": 1.964439331075033e-05, "loss": 0.577, "step": 3161 }, { "epoch": 0.11265297397438409, "grad_norm": 1.7144683929201008, "learning_rate": 1.964408825607967e-05, "loss": 0.6463, "step": 3162 }, { "epoch": 0.11268860110087821, "grad_norm": 2.56452106181446, "learning_rate": 1.964378307299124e-05, "loss": 0.6266, "step": 3163 }, { "epoch": 0.11272422822737233, "grad_norm": 1.6179311146417188, "learning_rate": 1.9643477761489097e-05, "loss": 0.5794, "step": 3164 }, { "epoch": 0.11275985535386643, "grad_norm": 1.231243896865687, "learning_rate": 1.9643172321577304e-05, "loss": 0.5619, "step": 3165 }, { "epoch": 0.11279548248036055, "grad_norm": 1.3291360364971527, "learning_rate": 1.9642866753259934e-05, "loss": 0.518, "step": 3166 }, { "epoch": 0.11283110960685466, "grad_norm": 1.1848278364226899, "learning_rate": 1.9642561056541052e-05, "loss": 0.4521, "step": 3167 }, { "epoch": 0.11286673673334877, "grad_norm": 1.1998785674502275, "learning_rate": 1.964225523142473e-05, "loss": 0.4962, "step": 3168 }, { "epoch": 0.11290236385984288, "grad_norm": 1.2042096297367324, "learning_rate": 1.9641949277915044e-05, "loss": 0.4757, "step": 3169 }, { "epoch": 0.112937990986337, "grad_norm": 3.4719644210613665, "learning_rate": 1.9641643196016058e-05, "loss": 0.7361, "step": 3170 }, { "epoch": 0.1129736181128311, "grad_norm": 1.3114798708515438, "learning_rate": 1.9641336985731854e-05, "loss": 0.5505, "step": 3171 }, { "epoch": 0.11300924523932522, "grad_norm": 1.145757390618955, "learning_rate": 1.9641030647066516e-05, "loss": 0.5097, "step": 3172 }, { "epoch": 0.11304487236581934, "grad_norm": 1.9195479044443706, "learning_rate": 1.964072418002411e-05, "loss": 0.6565, "step": 3173 }, { "epoch": 0.11308049949231344, "grad_norm": 1.7121411968213907, "learning_rate": 1.9640417584608726e-05, "loss": 0.6607, "step": 3174 }, { "epoch": 0.11311612661880756, "grad_norm": 1.7965117094279068, "learning_rate": 1.9640110860824442e-05, "loss": 0.6128, "step": 3175 }, { "epoch": 0.11315175374530168, "grad_norm": 1.308215700080099, "learning_rate": 1.9639804008675344e-05, "loss": 0.571, "step": 3176 }, { "epoch": 0.11318738087179578, "grad_norm": 1.7153389544749051, "learning_rate": 1.9639497028165518e-05, "loss": 0.6519, "step": 3177 }, { "epoch": 0.1132230079982899, "grad_norm": 1.2443504084762647, "learning_rate": 1.963918991929905e-05, "loss": 0.5565, "step": 3178 }, { "epoch": 0.11325863512478401, "grad_norm": 1.5628373812037564, "learning_rate": 1.9638882682080033e-05, "loss": 0.6545, "step": 3179 }, { "epoch": 0.11329426225127813, "grad_norm": 1.2713508445587822, "learning_rate": 1.9638575316512557e-05, "loss": 0.5398, "step": 3180 }, { "epoch": 0.11332988937777223, "grad_norm": 1.485668473066828, "learning_rate": 1.963826782260071e-05, "loss": 0.5421, "step": 3181 }, { "epoch": 0.11336551650426635, "grad_norm": 2.0961010761988206, "learning_rate": 1.9637960200348592e-05, "loss": 0.6454, "step": 3182 }, { "epoch": 0.11340114363076047, "grad_norm": 2.180849205582606, "learning_rate": 1.9637652449760297e-05, "loss": 0.6865, "step": 3183 }, { "epoch": 0.11343677075725457, "grad_norm": 1.7348561429194373, "learning_rate": 1.9637344570839924e-05, "loss": 0.6884, "step": 3184 }, { "epoch": 0.11347239788374869, "grad_norm": 1.2324933252196717, "learning_rate": 1.9637036563591573e-05, "loss": 0.4976, "step": 3185 }, { "epoch": 0.1135080250102428, "grad_norm": 1.339831177506797, "learning_rate": 1.963672842801934e-05, "loss": 0.5286, "step": 3186 }, { "epoch": 0.11354365213673691, "grad_norm": 1.3089715818221463, "learning_rate": 1.9636420164127337e-05, "loss": 0.5245, "step": 3187 }, { "epoch": 0.11357927926323103, "grad_norm": 2.3124373114263266, "learning_rate": 1.9636111771919665e-05, "loss": 0.6563, "step": 3188 }, { "epoch": 0.11361490638972514, "grad_norm": 1.9699234621591029, "learning_rate": 1.9635803251400427e-05, "loss": 0.6201, "step": 3189 }, { "epoch": 0.11365053351621925, "grad_norm": 1.1567492926730172, "learning_rate": 1.9635494602573732e-05, "loss": 0.4935, "step": 3190 }, { "epoch": 0.11368616064271336, "grad_norm": 1.2521110349048907, "learning_rate": 1.9635185825443694e-05, "loss": 0.5595, "step": 3191 }, { "epoch": 0.11372178776920748, "grad_norm": 1.6111135137253243, "learning_rate": 1.963487692001442e-05, "loss": 0.6358, "step": 3192 }, { "epoch": 0.11375741489570158, "grad_norm": 1.211541860279606, "learning_rate": 1.9634567886290027e-05, "loss": 0.5736, "step": 3193 }, { "epoch": 0.1137930420221957, "grad_norm": 1.2079227025884918, "learning_rate": 1.963425872427463e-05, "loss": 0.507, "step": 3194 }, { "epoch": 0.11382866914868982, "grad_norm": 1.623007528327064, "learning_rate": 1.963394943397234e-05, "loss": 0.5908, "step": 3195 }, { "epoch": 0.11386429627518392, "grad_norm": 1.6391788138676655, "learning_rate": 1.9633640015387282e-05, "loss": 0.6399, "step": 3196 }, { "epoch": 0.11389992340167804, "grad_norm": 1.7215741960010815, "learning_rate": 1.9633330468523572e-05, "loss": 0.6628, "step": 3197 }, { "epoch": 0.11393555052817216, "grad_norm": 1.6803261439179895, "learning_rate": 1.9633020793385338e-05, "loss": 0.5923, "step": 3198 }, { "epoch": 0.11397117765466626, "grad_norm": 1.811592244598926, "learning_rate": 1.9632710989976696e-05, "loss": 0.647, "step": 3199 }, { "epoch": 0.11400680478116038, "grad_norm": 1.3284639478676252, "learning_rate": 1.9632401058301774e-05, "loss": 0.516, "step": 3200 }, { "epoch": 0.11404243190765449, "grad_norm": 2.0955190646041952, "learning_rate": 1.9632090998364698e-05, "loss": 0.5758, "step": 3201 }, { "epoch": 0.1140780590341486, "grad_norm": 11.416259250843753, "learning_rate": 1.9631780810169602e-05, "loss": 0.6259, "step": 3202 }, { "epoch": 0.11411368616064271, "grad_norm": 1.4123131968409928, "learning_rate": 1.963147049372061e-05, "loss": 0.5261, "step": 3203 }, { "epoch": 0.11414931328713683, "grad_norm": 1.762349821349352, "learning_rate": 1.9631160049021852e-05, "loss": 0.6563, "step": 3204 }, { "epoch": 0.11418494041363093, "grad_norm": 1.9521716034924923, "learning_rate": 1.9630849476077473e-05, "loss": 0.5644, "step": 3205 }, { "epoch": 0.11422056754012505, "grad_norm": 1.8673211835692998, "learning_rate": 1.96305387748916e-05, "loss": 0.542, "step": 3206 }, { "epoch": 0.11425619466661917, "grad_norm": 1.896406343575535, "learning_rate": 1.9630227945468368e-05, "loss": 0.5613, "step": 3207 }, { "epoch": 0.11429182179311327, "grad_norm": 1.7335773640238785, "learning_rate": 1.9629916987811924e-05, "loss": 0.6356, "step": 3208 }, { "epoch": 0.11432744891960739, "grad_norm": 1.6721131046164126, "learning_rate": 1.9629605901926404e-05, "loss": 0.6415, "step": 3209 }, { "epoch": 0.1143630760461015, "grad_norm": 1.2637392866221324, "learning_rate": 1.962929468781595e-05, "loss": 0.5471, "step": 3210 }, { "epoch": 0.11439870317259561, "grad_norm": 1.215840264437351, "learning_rate": 1.962898334548471e-05, "loss": 0.5339, "step": 3211 }, { "epoch": 0.11443433029908973, "grad_norm": 1.797910902572876, "learning_rate": 1.9628671874936823e-05, "loss": 0.623, "step": 3212 }, { "epoch": 0.11446995742558384, "grad_norm": 1.149295433323441, "learning_rate": 1.9628360276176438e-05, "loss": 0.4965, "step": 3213 }, { "epoch": 0.11450558455207795, "grad_norm": 2.025598251462267, "learning_rate": 1.9628048549207707e-05, "loss": 0.6132, "step": 3214 }, { "epoch": 0.11454121167857206, "grad_norm": 1.3288785091916058, "learning_rate": 1.9627736694034782e-05, "loss": 0.4829, "step": 3215 }, { "epoch": 0.11457683880506618, "grad_norm": 2.034362138777092, "learning_rate": 1.9627424710661812e-05, "loss": 0.6419, "step": 3216 }, { "epoch": 0.11461246593156028, "grad_norm": 2.0687243756957807, "learning_rate": 1.9627112599092956e-05, "loss": 0.6182, "step": 3217 }, { "epoch": 0.1146480930580544, "grad_norm": 2.2413979822151653, "learning_rate": 1.9626800359332362e-05, "loss": 0.5456, "step": 3218 }, { "epoch": 0.11468372018454852, "grad_norm": 1.478202892402654, "learning_rate": 1.9626487991384194e-05, "loss": 0.5599, "step": 3219 }, { "epoch": 0.11471934731104262, "grad_norm": 1.964270274815463, "learning_rate": 1.9626175495252612e-05, "loss": 0.7581, "step": 3220 }, { "epoch": 0.11475497443753674, "grad_norm": 2.374610157807133, "learning_rate": 1.9625862870941774e-05, "loss": 0.5761, "step": 3221 }, { "epoch": 0.11479060156403086, "grad_norm": 1.9796528315120612, "learning_rate": 1.9625550118455844e-05, "loss": 0.6096, "step": 3222 }, { "epoch": 0.11482622869052496, "grad_norm": 1.3284981346161857, "learning_rate": 1.9625237237798985e-05, "loss": 0.4884, "step": 3223 }, { "epoch": 0.11486185581701908, "grad_norm": 2.1762794281161364, "learning_rate": 1.9624924228975365e-05, "loss": 0.7206, "step": 3224 }, { "epoch": 0.1148974829435132, "grad_norm": 1.4615010470559697, "learning_rate": 1.9624611091989152e-05, "loss": 0.5923, "step": 3225 }, { "epoch": 0.11493311007000731, "grad_norm": 1.3715407355706846, "learning_rate": 1.9624297826844513e-05, "loss": 0.529, "step": 3226 }, { "epoch": 0.11496873719650141, "grad_norm": 1.17902052295513, "learning_rate": 1.9623984433545625e-05, "loss": 0.4881, "step": 3227 }, { "epoch": 0.11500436432299553, "grad_norm": 1.227708345416681, "learning_rate": 1.9623670912096656e-05, "loss": 0.5205, "step": 3228 }, { "epoch": 0.11503999144948965, "grad_norm": 1.3505176505091092, "learning_rate": 1.9623357262501783e-05, "loss": 0.5583, "step": 3229 }, { "epoch": 0.11507561857598375, "grad_norm": 2.0363023950263663, "learning_rate": 1.962304348476518e-05, "loss": 0.6319, "step": 3230 }, { "epoch": 0.11511124570247787, "grad_norm": 1.7155089355207518, "learning_rate": 1.9622729578891028e-05, "loss": 0.651, "step": 3231 }, { "epoch": 0.11514687282897199, "grad_norm": 1.7853559981161071, "learning_rate": 1.9622415544883506e-05, "loss": 0.6461, "step": 3232 }, { "epoch": 0.11518249995546609, "grad_norm": 1.2508407476774337, "learning_rate": 1.9622101382746795e-05, "loss": 0.5475, "step": 3233 }, { "epoch": 0.1152181270819602, "grad_norm": 1.2298213910551719, "learning_rate": 1.9621787092485078e-05, "loss": 0.5645, "step": 3234 }, { "epoch": 0.11525375420845432, "grad_norm": 1.2364450191363174, "learning_rate": 1.962147267410254e-05, "loss": 0.5278, "step": 3235 }, { "epoch": 0.11528938133494843, "grad_norm": 3.108183504175527, "learning_rate": 1.9621158127603372e-05, "loss": 0.6607, "step": 3236 }, { "epoch": 0.11532500846144254, "grad_norm": 1.2968894094754864, "learning_rate": 1.9620843452991754e-05, "loss": 0.5209, "step": 3237 }, { "epoch": 0.11536063558793666, "grad_norm": 1.3314976940628631, "learning_rate": 1.9620528650271882e-05, "loss": 0.4986, "step": 3238 }, { "epoch": 0.11539626271443076, "grad_norm": 1.7937206256020295, "learning_rate": 1.962021371944795e-05, "loss": 0.52, "step": 3239 }, { "epoch": 0.11543188984092488, "grad_norm": 1.3746791150984816, "learning_rate": 1.9619898660524145e-05, "loss": 0.5716, "step": 3240 }, { "epoch": 0.115467516967419, "grad_norm": 1.3608579532066627, "learning_rate": 1.961958347350467e-05, "loss": 0.5525, "step": 3241 }, { "epoch": 0.1155031440939131, "grad_norm": 1.2362362569770289, "learning_rate": 1.961926815839371e-05, "loss": 0.5394, "step": 3242 }, { "epoch": 0.11553877122040722, "grad_norm": 1.8907589257153787, "learning_rate": 1.9618952715195476e-05, "loss": 0.5931, "step": 3243 }, { "epoch": 0.11557439834690134, "grad_norm": 1.3012723402152722, "learning_rate": 1.961863714391416e-05, "loss": 0.5371, "step": 3244 }, { "epoch": 0.11561002547339544, "grad_norm": 1.2799664725672737, "learning_rate": 1.9618321444553972e-05, "loss": 0.5297, "step": 3245 }, { "epoch": 0.11564565259988956, "grad_norm": 1.689163921602312, "learning_rate": 1.9618005617119106e-05, "loss": 0.6381, "step": 3246 }, { "epoch": 0.11568127972638367, "grad_norm": 1.9706480944346243, "learning_rate": 1.961768966161378e-05, "loss": 0.6974, "step": 3247 }, { "epoch": 0.11571690685287778, "grad_norm": 1.2164511707250296, "learning_rate": 1.9617373578042187e-05, "loss": 0.4933, "step": 3248 }, { "epoch": 0.1157525339793719, "grad_norm": 1.2783709622212445, "learning_rate": 1.9617057366408545e-05, "loss": 0.5536, "step": 3249 }, { "epoch": 0.11578816110586601, "grad_norm": 1.194602839751836, "learning_rate": 1.9616741026717063e-05, "loss": 0.575, "step": 3250 }, { "epoch": 0.11582378823236011, "grad_norm": 1.9446774201597603, "learning_rate": 1.961642455897195e-05, "loss": 0.6835, "step": 3251 }, { "epoch": 0.11585941535885423, "grad_norm": 1.1917276737430769, "learning_rate": 1.9616107963177423e-05, "loss": 0.5184, "step": 3252 }, { "epoch": 0.11589504248534835, "grad_norm": 1.8494737775167456, "learning_rate": 1.96157912393377e-05, "loss": 0.639, "step": 3253 }, { "epoch": 0.11593066961184245, "grad_norm": 1.488895690189651, "learning_rate": 1.9615474387456995e-05, "loss": 0.5468, "step": 3254 }, { "epoch": 0.11596629673833657, "grad_norm": 1.2444905859068602, "learning_rate": 1.9615157407539525e-05, "loss": 0.5505, "step": 3255 }, { "epoch": 0.11600192386483069, "grad_norm": 2.040601524661916, "learning_rate": 1.9614840299589516e-05, "loss": 0.6223, "step": 3256 }, { "epoch": 0.11603755099132479, "grad_norm": 1.8703802836392287, "learning_rate": 1.9614523063611185e-05, "loss": 0.603, "step": 3257 }, { "epoch": 0.1160731781178189, "grad_norm": 1.1336406692468823, "learning_rate": 1.9614205699608763e-05, "loss": 0.5714, "step": 3258 }, { "epoch": 0.11610880524431302, "grad_norm": 1.206364496980157, "learning_rate": 1.961388820758647e-05, "loss": 0.569, "step": 3259 }, { "epoch": 0.11614443237080713, "grad_norm": 2.384619924357649, "learning_rate": 1.9613570587548537e-05, "loss": 0.6435, "step": 3260 }, { "epoch": 0.11618005949730124, "grad_norm": 1.6163072391515574, "learning_rate": 1.9613252839499192e-05, "loss": 0.6346, "step": 3261 }, { "epoch": 0.11621568662379536, "grad_norm": 1.805901309475101, "learning_rate": 1.9612934963442663e-05, "loss": 0.6513, "step": 3262 }, { "epoch": 0.11625131375028946, "grad_norm": 1.1844219049623623, "learning_rate": 1.961261695938319e-05, "loss": 0.534, "step": 3263 }, { "epoch": 0.11628694087678358, "grad_norm": 1.5289559076531392, "learning_rate": 1.9612298827325e-05, "loss": 0.5558, "step": 3264 }, { "epoch": 0.1163225680032777, "grad_norm": 1.7311102302229027, "learning_rate": 1.9611980567272337e-05, "loss": 0.6078, "step": 3265 }, { "epoch": 0.1163581951297718, "grad_norm": 2.0923045942780933, "learning_rate": 1.961166217922943e-05, "loss": 0.5603, "step": 3266 }, { "epoch": 0.11639382225626592, "grad_norm": 1.1885937474168256, "learning_rate": 1.9611343663200527e-05, "loss": 0.6039, "step": 3267 }, { "epoch": 0.11642944938276004, "grad_norm": 1.6775553071650555, "learning_rate": 1.961102501918986e-05, "loss": 0.7639, "step": 3268 }, { "epoch": 0.11646507650925414, "grad_norm": 1.3133127329486631, "learning_rate": 1.961070624720168e-05, "loss": 0.5428, "step": 3269 }, { "epoch": 0.11650070363574826, "grad_norm": 1.1406146935909292, "learning_rate": 1.961038734724023e-05, "loss": 0.5336, "step": 3270 }, { "epoch": 0.11653633076224237, "grad_norm": 1.835864903947633, "learning_rate": 1.9610068319309757e-05, "loss": 0.6677, "step": 3271 }, { "epoch": 0.11657195788873648, "grad_norm": 1.4095722697882878, "learning_rate": 1.9609749163414503e-05, "loss": 0.5358, "step": 3272 }, { "epoch": 0.1166075850152306, "grad_norm": 1.179617678469165, "learning_rate": 1.9609429879558726e-05, "loss": 0.4979, "step": 3273 }, { "epoch": 0.11664321214172471, "grad_norm": 1.4367919182283126, "learning_rate": 1.960911046774667e-05, "loss": 0.6556, "step": 3274 }, { "epoch": 0.11667883926821883, "grad_norm": 1.3234861370183997, "learning_rate": 1.9608790927982592e-05, "loss": 0.5801, "step": 3275 }, { "epoch": 0.11671446639471293, "grad_norm": 1.7359745472294432, "learning_rate": 1.960847126027075e-05, "loss": 0.5968, "step": 3276 }, { "epoch": 0.11675009352120705, "grad_norm": 1.4973485697141542, "learning_rate": 1.9608151464615392e-05, "loss": 0.5076, "step": 3277 }, { "epoch": 0.11678572064770117, "grad_norm": 1.5207933157626357, "learning_rate": 1.9607831541020784e-05, "loss": 0.5397, "step": 3278 }, { "epoch": 0.11682134777419527, "grad_norm": 1.9747355087740435, "learning_rate": 1.9607511489491187e-05, "loss": 0.6664, "step": 3279 }, { "epoch": 0.11685697490068939, "grad_norm": 1.3263741080542084, "learning_rate": 1.9607191310030857e-05, "loss": 0.56, "step": 3280 }, { "epoch": 0.1168926020271835, "grad_norm": 1.8168655891592076, "learning_rate": 1.960687100264406e-05, "loss": 0.6514, "step": 3281 }, { "epoch": 0.11692822915367761, "grad_norm": 1.4436262603132344, "learning_rate": 1.9606550567335058e-05, "loss": 0.6075, "step": 3282 }, { "epoch": 0.11696385628017172, "grad_norm": 1.2557937099689969, "learning_rate": 1.960623000410812e-05, "loss": 0.5312, "step": 3283 }, { "epoch": 0.11699948340666584, "grad_norm": 1.292729338500945, "learning_rate": 1.9605909312967516e-05, "loss": 0.575, "step": 3284 }, { "epoch": 0.11703511053315994, "grad_norm": 1.2798312648802737, "learning_rate": 1.960558849391752e-05, "loss": 0.5689, "step": 3285 }, { "epoch": 0.11707073765965406, "grad_norm": 1.2484659007223702, "learning_rate": 1.9605267546962395e-05, "loss": 0.5509, "step": 3286 }, { "epoch": 0.11710636478614818, "grad_norm": 1.1879387194265585, "learning_rate": 1.960494647210642e-05, "loss": 0.5554, "step": 3287 }, { "epoch": 0.11714199191264228, "grad_norm": 2.0682106215233156, "learning_rate": 1.960462526935387e-05, "loss": 0.5811, "step": 3288 }, { "epoch": 0.1171776190391364, "grad_norm": 1.9669320344019776, "learning_rate": 1.960430393870902e-05, "loss": 0.5996, "step": 3289 }, { "epoch": 0.11721324616563052, "grad_norm": 1.3819179046798475, "learning_rate": 1.9603982480176152e-05, "loss": 0.5088, "step": 3290 }, { "epoch": 0.11724887329212462, "grad_norm": 1.4667572368747521, "learning_rate": 1.960366089375954e-05, "loss": 0.5099, "step": 3291 }, { "epoch": 0.11728450041861874, "grad_norm": 2.006646193903483, "learning_rate": 1.9603339179463475e-05, "loss": 0.5934, "step": 3292 }, { "epoch": 0.11732012754511285, "grad_norm": 1.2983286154206803, "learning_rate": 1.9603017337292235e-05, "loss": 0.5335, "step": 3293 }, { "epoch": 0.11735575467160696, "grad_norm": 1.2125333763209032, "learning_rate": 1.960269536725011e-05, "loss": 0.5592, "step": 3294 }, { "epoch": 0.11739138179810107, "grad_norm": 1.6933137245591894, "learning_rate": 1.960237326934138e-05, "loss": 0.6586, "step": 3295 }, { "epoch": 0.11742700892459519, "grad_norm": 1.783197977997993, "learning_rate": 1.960205104357034e-05, "loss": 0.4844, "step": 3296 }, { "epoch": 0.1174626360510893, "grad_norm": 1.3153804310561474, "learning_rate": 1.960172868994128e-05, "loss": 0.5549, "step": 3297 }, { "epoch": 0.11749826317758341, "grad_norm": 1.4173024304093822, "learning_rate": 1.960140620845849e-05, "loss": 0.5002, "step": 3298 }, { "epoch": 0.11753389030407753, "grad_norm": 1.8254696034498787, "learning_rate": 1.9601083599126266e-05, "loss": 0.6713, "step": 3299 }, { "epoch": 0.11756951743057163, "grad_norm": 1.7395487281743818, "learning_rate": 1.96007608619489e-05, "loss": 0.6313, "step": 3300 }, { "epoch": 0.11760514455706575, "grad_norm": 1.5881181939434268, "learning_rate": 1.9600437996930695e-05, "loss": 0.5101, "step": 3301 }, { "epoch": 0.11764077168355987, "grad_norm": 2.0643565561686157, "learning_rate": 1.960011500407595e-05, "loss": 0.5734, "step": 3302 }, { "epoch": 0.11767639881005397, "grad_norm": 2.6553765726513734, "learning_rate": 1.959979188338896e-05, "loss": 0.6394, "step": 3303 }, { "epoch": 0.11771202593654809, "grad_norm": 2.0753311123148412, "learning_rate": 1.9599468634874033e-05, "loss": 0.5604, "step": 3304 }, { "epoch": 0.1177476530630422, "grad_norm": 2.6782348453685243, "learning_rate": 1.9599145258535472e-05, "loss": 0.5302, "step": 3305 }, { "epoch": 0.11778328018953631, "grad_norm": 2.489614832037904, "learning_rate": 1.9598821754377582e-05, "loss": 0.6339, "step": 3306 }, { "epoch": 0.11781890731603042, "grad_norm": 2.3843758004489923, "learning_rate": 1.9598498122404674e-05, "loss": 0.5365, "step": 3307 }, { "epoch": 0.11785453444252454, "grad_norm": 2.2943910477376814, "learning_rate": 1.959817436262105e-05, "loss": 0.5252, "step": 3308 }, { "epoch": 0.11789016156901864, "grad_norm": 3.832088839643593, "learning_rate": 1.9597850475031026e-05, "loss": 0.5436, "step": 3309 }, { "epoch": 0.11792578869551276, "grad_norm": 2.2101057605524885, "learning_rate": 1.9597526459638913e-05, "loss": 0.6256, "step": 3310 }, { "epoch": 0.11796141582200688, "grad_norm": 7.030104577394876, "learning_rate": 1.959720231644903e-05, "loss": 0.6501, "step": 3311 }, { "epoch": 0.11799704294850098, "grad_norm": 1.9895671518788607, "learning_rate": 1.959687804546569e-05, "loss": 0.6494, "step": 3312 }, { "epoch": 0.1180326700749951, "grad_norm": 2.706712642214425, "learning_rate": 1.9596553646693212e-05, "loss": 0.647, "step": 3313 }, { "epoch": 0.11806829720148922, "grad_norm": 6.413253800314126, "learning_rate": 1.9596229120135916e-05, "loss": 0.697, "step": 3314 }, { "epoch": 0.11810392432798332, "grad_norm": 2.482065878294624, "learning_rate": 1.9595904465798116e-05, "loss": 0.5156, "step": 3315 }, { "epoch": 0.11813955145447744, "grad_norm": 2.846407317553088, "learning_rate": 1.9595579683684143e-05, "loss": 0.5333, "step": 3316 }, { "epoch": 0.11817517858097155, "grad_norm": 1.80477448672958, "learning_rate": 1.9595254773798322e-05, "loss": 0.6761, "step": 3317 }, { "epoch": 0.11821080570746566, "grad_norm": 1.9391239198472063, "learning_rate": 1.9594929736144978e-05, "loss": 0.5377, "step": 3318 }, { "epoch": 0.11824643283395977, "grad_norm": 1.6095484487159388, "learning_rate": 1.9594604570728433e-05, "loss": 0.546, "step": 3319 }, { "epoch": 0.11828205996045389, "grad_norm": 2.1465303593507046, "learning_rate": 1.9594279277553024e-05, "loss": 0.7036, "step": 3320 }, { "epoch": 0.11831768708694801, "grad_norm": 1.7523870685130603, "learning_rate": 1.9593953856623082e-05, "loss": 0.6115, "step": 3321 }, { "epoch": 0.11835331421344211, "grad_norm": 1.4065783217505166, "learning_rate": 1.959362830794294e-05, "loss": 0.5654, "step": 3322 }, { "epoch": 0.11838894133993623, "grad_norm": 1.3387938195018942, "learning_rate": 1.9593302631516926e-05, "loss": 0.5252, "step": 3323 }, { "epoch": 0.11842456846643035, "grad_norm": 1.9212884972999213, "learning_rate": 1.9592976827349387e-05, "loss": 0.6017, "step": 3324 }, { "epoch": 0.11846019559292445, "grad_norm": 1.1977077239974223, "learning_rate": 1.9592650895444654e-05, "loss": 0.4821, "step": 3325 }, { "epoch": 0.11849582271941857, "grad_norm": 1.6071817904637393, "learning_rate": 1.959232483580707e-05, "loss": 0.6368, "step": 3326 }, { "epoch": 0.11853144984591268, "grad_norm": 1.599016681746712, "learning_rate": 1.9591998648440976e-05, "loss": 0.5016, "step": 3327 }, { "epoch": 0.11856707697240679, "grad_norm": 1.5529355600554209, "learning_rate": 1.959167233335072e-05, "loss": 0.5051, "step": 3328 }, { "epoch": 0.1186027040989009, "grad_norm": 2.9657195913401404, "learning_rate": 1.9591345890540635e-05, "loss": 0.6215, "step": 3329 }, { "epoch": 0.11863833122539502, "grad_norm": 2.229591306612113, "learning_rate": 1.959101932001508e-05, "loss": 0.5394, "step": 3330 }, { "epoch": 0.11867395835188913, "grad_norm": 1.3274567855928603, "learning_rate": 1.95906926217784e-05, "loss": 0.5349, "step": 3331 }, { "epoch": 0.11870958547838324, "grad_norm": 1.2402906459769658, "learning_rate": 1.959036579583494e-05, "loss": 0.5283, "step": 3332 }, { "epoch": 0.11874521260487736, "grad_norm": 1.8370360865347561, "learning_rate": 1.959003884218906e-05, "loss": 0.6166, "step": 3333 }, { "epoch": 0.11878083973137146, "grad_norm": 2.165028855616492, "learning_rate": 1.958971176084511e-05, "loss": 0.6397, "step": 3334 }, { "epoch": 0.11881646685786558, "grad_norm": 1.3071188423391427, "learning_rate": 1.9589384551807442e-05, "loss": 0.5436, "step": 3335 }, { "epoch": 0.1188520939843597, "grad_norm": 1.2033160706573343, "learning_rate": 1.958905721508042e-05, "loss": 0.551, "step": 3336 }, { "epoch": 0.1188877211108538, "grad_norm": 1.2274144198197225, "learning_rate": 1.9588729750668394e-05, "loss": 0.5393, "step": 3337 }, { "epoch": 0.11892334823734792, "grad_norm": 1.9447127811092764, "learning_rate": 1.958840215857573e-05, "loss": 0.5441, "step": 3338 }, { "epoch": 0.11895897536384203, "grad_norm": 1.258729508333436, "learning_rate": 1.9588074438806792e-05, "loss": 0.5901, "step": 3339 }, { "epoch": 0.11899460249033614, "grad_norm": 1.309574127630161, "learning_rate": 1.958774659136594e-05, "loss": 0.5219, "step": 3340 }, { "epoch": 0.11903022961683025, "grad_norm": 1.3032909832364652, "learning_rate": 1.9587418616257544e-05, "loss": 0.5876, "step": 3341 }, { "epoch": 0.11906585674332437, "grad_norm": 1.3232353131981796, "learning_rate": 1.9587090513485966e-05, "loss": 0.5823, "step": 3342 }, { "epoch": 0.11910148386981848, "grad_norm": 1.8898315181141965, "learning_rate": 1.9586762283055575e-05, "loss": 0.7341, "step": 3343 }, { "epoch": 0.11913711099631259, "grad_norm": 1.2247342843163125, "learning_rate": 1.9586433924970742e-05, "loss": 0.5487, "step": 3344 }, { "epoch": 0.11917273812280671, "grad_norm": 2.597639871430925, "learning_rate": 1.9586105439235842e-05, "loss": 0.6569, "step": 3345 }, { "epoch": 0.11920836524930081, "grad_norm": 1.1973011430099596, "learning_rate": 1.9585776825855252e-05, "loss": 0.5513, "step": 3346 }, { "epoch": 0.11924399237579493, "grad_norm": 1.1884125558436356, "learning_rate": 1.958544808483334e-05, "loss": 0.5023, "step": 3347 }, { "epoch": 0.11927961950228905, "grad_norm": 1.201190203266547, "learning_rate": 1.9585119216174486e-05, "loss": 0.5002, "step": 3348 }, { "epoch": 0.11931524662878315, "grad_norm": 1.2407583745561708, "learning_rate": 1.958479021988307e-05, "loss": 0.5685, "step": 3349 }, { "epoch": 0.11935087375527727, "grad_norm": 1.2744536309427463, "learning_rate": 1.9584461095963475e-05, "loss": 0.5172, "step": 3350 }, { "epoch": 0.11938650088177138, "grad_norm": 2.315293995731453, "learning_rate": 1.9584131844420084e-05, "loss": 0.6016, "step": 3351 }, { "epoch": 0.11942212800826549, "grad_norm": 2.4285756263091827, "learning_rate": 1.958380246525727e-05, "loss": 0.6093, "step": 3352 }, { "epoch": 0.1194577551347596, "grad_norm": 1.8399136605926376, "learning_rate": 1.9583472958479434e-05, "loss": 0.7141, "step": 3353 }, { "epoch": 0.11949338226125372, "grad_norm": 1.3960378574470829, "learning_rate": 1.9583143324090957e-05, "loss": 0.5058, "step": 3354 }, { "epoch": 0.11952900938774783, "grad_norm": 1.4893754891925894, "learning_rate": 1.9582813562096228e-05, "loss": 0.4924, "step": 3355 }, { "epoch": 0.11956463651424194, "grad_norm": 2.598273147318514, "learning_rate": 1.9582483672499638e-05, "loss": 0.6751, "step": 3356 }, { "epoch": 0.11960026364073606, "grad_norm": 1.1944988912642691, "learning_rate": 1.9582153655305578e-05, "loss": 0.539, "step": 3357 }, { "epoch": 0.11963589076723016, "grad_norm": 1.2898001081779193, "learning_rate": 1.9581823510518445e-05, "loss": 0.5373, "step": 3358 }, { "epoch": 0.11967151789372428, "grad_norm": 2.3535922730600554, "learning_rate": 1.958149323814264e-05, "loss": 0.6262, "step": 3359 }, { "epoch": 0.1197071450202184, "grad_norm": 1.7664653597850044, "learning_rate": 1.9581162838182548e-05, "loss": 0.5556, "step": 3360 }, { "epoch": 0.1197427721467125, "grad_norm": 1.4329905997157693, "learning_rate": 1.958083231064258e-05, "loss": 0.5193, "step": 3361 }, { "epoch": 0.11977839927320662, "grad_norm": 2.9903807345980633, "learning_rate": 1.9580501655527132e-05, "loss": 0.626, "step": 3362 }, { "epoch": 0.11981402639970073, "grad_norm": 1.8788473540483035, "learning_rate": 1.958017087284061e-05, "loss": 0.6428, "step": 3363 }, { "epoch": 0.11984965352619484, "grad_norm": 1.964951079115647, "learning_rate": 1.9579839962587413e-05, "loss": 0.6653, "step": 3364 }, { "epoch": 0.11988528065268896, "grad_norm": 1.2083906963549058, "learning_rate": 1.9579508924771954e-05, "loss": 0.5534, "step": 3365 }, { "epoch": 0.11992090777918307, "grad_norm": 1.3155341490786514, "learning_rate": 1.9579177759398638e-05, "loss": 0.4474, "step": 3366 }, { "epoch": 0.11995653490567718, "grad_norm": 1.7156643530695965, "learning_rate": 1.9578846466471875e-05, "loss": 0.6098, "step": 3367 }, { "epoch": 0.11999216203217129, "grad_norm": 1.3139247482650225, "learning_rate": 1.9578515045996074e-05, "loss": 0.5086, "step": 3368 }, { "epoch": 0.12002778915866541, "grad_norm": 2.8048525588945137, "learning_rate": 1.9578183497975652e-05, "loss": 0.6737, "step": 3369 }, { "epoch": 0.12006341628515953, "grad_norm": 1.5159259926933843, "learning_rate": 1.9577851822415023e-05, "loss": 0.5254, "step": 3370 }, { "epoch": 0.12009904341165363, "grad_norm": 1.2715445390396654, "learning_rate": 1.95775200193186e-05, "loss": 0.546, "step": 3371 }, { "epoch": 0.12013467053814775, "grad_norm": 1.8228113205893195, "learning_rate": 1.9577188088690804e-05, "loss": 0.6302, "step": 3372 }, { "epoch": 0.12017029766464186, "grad_norm": 1.112997166338344, "learning_rate": 1.9576856030536055e-05, "loss": 0.4888, "step": 3373 }, { "epoch": 0.12020592479113597, "grad_norm": 2.37385549417634, "learning_rate": 1.9576523844858774e-05, "loss": 0.7032, "step": 3374 }, { "epoch": 0.12024155191763009, "grad_norm": 1.584402301129494, "learning_rate": 1.9576191531663383e-05, "loss": 0.5169, "step": 3375 }, { "epoch": 0.1202771790441242, "grad_norm": 1.835625473726074, "learning_rate": 1.957585909095431e-05, "loss": 0.6996, "step": 3376 }, { "epoch": 0.1203128061706183, "grad_norm": 2.290386954626032, "learning_rate": 1.9575526522735982e-05, "loss": 0.6181, "step": 3377 }, { "epoch": 0.12034843329711242, "grad_norm": 1.7403424078107126, "learning_rate": 1.957519382701282e-05, "loss": 0.5648, "step": 3378 }, { "epoch": 0.12038406042360654, "grad_norm": 1.2718224229161692, "learning_rate": 1.9574861003789264e-05, "loss": 0.5472, "step": 3379 }, { "epoch": 0.12041968755010064, "grad_norm": 1.982427954203825, "learning_rate": 1.957452805306974e-05, "loss": 0.6295, "step": 3380 }, { "epoch": 0.12045531467659476, "grad_norm": 3.2228704067395575, "learning_rate": 1.9574194974858686e-05, "loss": 0.6814, "step": 3381 }, { "epoch": 0.12049094180308888, "grad_norm": 1.2185875027750517, "learning_rate": 1.957386176916053e-05, "loss": 0.5092, "step": 3382 }, { "epoch": 0.12052656892958298, "grad_norm": 1.7836100371190724, "learning_rate": 1.9573528435979715e-05, "loss": 0.6407, "step": 3383 }, { "epoch": 0.1205621960560771, "grad_norm": 1.280845705786239, "learning_rate": 1.9573194975320672e-05, "loss": 0.5472, "step": 3384 }, { "epoch": 0.12059782318257121, "grad_norm": 1.785509967292959, "learning_rate": 1.9572861387187854e-05, "loss": 0.6241, "step": 3385 }, { "epoch": 0.12063345030906532, "grad_norm": 1.255980950118524, "learning_rate": 1.957252767158569e-05, "loss": 0.5493, "step": 3386 }, { "epoch": 0.12066907743555944, "grad_norm": 1.3110979783366303, "learning_rate": 1.9572193828518633e-05, "loss": 0.5771, "step": 3387 }, { "epoch": 0.12070470456205355, "grad_norm": 1.929574905039301, "learning_rate": 1.9571859857991124e-05, "loss": 0.6549, "step": 3388 }, { "epoch": 0.12074033168854766, "grad_norm": 2.001178973018562, "learning_rate": 1.957152576000761e-05, "loss": 0.6728, "step": 3389 }, { "epoch": 0.12077595881504177, "grad_norm": 1.2880736445166574, "learning_rate": 1.957119153457254e-05, "loss": 0.5378, "step": 3390 }, { "epoch": 0.12081158594153589, "grad_norm": 1.184252811981976, "learning_rate": 1.9570857181690365e-05, "loss": 0.453, "step": 3391 }, { "epoch": 0.12084721306803, "grad_norm": 1.9869520610493872, "learning_rate": 1.957052270136554e-05, "loss": 0.7131, "step": 3392 }, { "epoch": 0.12088284019452411, "grad_norm": 1.8570960594291521, "learning_rate": 1.9570188093602512e-05, "loss": 0.668, "step": 3393 }, { "epoch": 0.12091846732101823, "grad_norm": 1.5188061640478403, "learning_rate": 1.956985335840574e-05, "loss": 0.6346, "step": 3394 }, { "epoch": 0.12095409444751233, "grad_norm": 1.897522832772226, "learning_rate": 1.9569518495779682e-05, "loss": 0.5926, "step": 3395 }, { "epoch": 0.12098972157400645, "grad_norm": 1.3556137583876244, "learning_rate": 1.9569183505728796e-05, "loss": 0.5866, "step": 3396 }, { "epoch": 0.12102534870050057, "grad_norm": 1.788018757974229, "learning_rate": 1.9568848388257547e-05, "loss": 0.7381, "step": 3397 }, { "epoch": 0.12106097582699467, "grad_norm": 1.6498682214920237, "learning_rate": 1.9568513143370392e-05, "loss": 0.6083, "step": 3398 }, { "epoch": 0.12109660295348879, "grad_norm": 1.8378793471701118, "learning_rate": 1.9568177771071795e-05, "loss": 0.6192, "step": 3399 }, { "epoch": 0.1211322300799829, "grad_norm": 1.3412825788671907, "learning_rate": 1.9567842271366223e-05, "loss": 0.5454, "step": 3400 }, { "epoch": 0.121167857206477, "grad_norm": 1.297194348981078, "learning_rate": 1.9567506644258144e-05, "loss": 0.534, "step": 3401 }, { "epoch": 0.12120348433297112, "grad_norm": 1.1961786165545256, "learning_rate": 1.9567170889752027e-05, "loss": 0.5378, "step": 3402 }, { "epoch": 0.12123911145946524, "grad_norm": 1.0927936912577367, "learning_rate": 1.9566835007852344e-05, "loss": 0.5149, "step": 3403 }, { "epoch": 0.12127473858595934, "grad_norm": 1.7338012116669905, "learning_rate": 1.9566498998563565e-05, "loss": 0.6077, "step": 3404 }, { "epoch": 0.12131036571245346, "grad_norm": 1.33968585540158, "learning_rate": 1.9566162861890165e-05, "loss": 0.5454, "step": 3405 }, { "epoch": 0.12134599283894758, "grad_norm": 1.2189546329339462, "learning_rate": 1.9565826597836623e-05, "loss": 0.5229, "step": 3406 }, { "epoch": 0.12138161996544168, "grad_norm": 1.1960280470760785, "learning_rate": 1.9565490206407408e-05, "loss": 0.5281, "step": 3407 }, { "epoch": 0.1214172470919358, "grad_norm": 1.3150260905128193, "learning_rate": 1.9565153687607006e-05, "loss": 0.5294, "step": 3408 }, { "epoch": 0.12145287421842992, "grad_norm": 2.047646720776246, "learning_rate": 1.95648170414399e-05, "loss": 0.5357, "step": 3409 }, { "epoch": 0.12148850134492402, "grad_norm": 1.5533792440573129, "learning_rate": 1.956448026791057e-05, "loss": 0.6153, "step": 3410 }, { "epoch": 0.12152412847141814, "grad_norm": 1.8506063020187271, "learning_rate": 1.9564143367023498e-05, "loss": 0.6768, "step": 3411 }, { "epoch": 0.12155975559791225, "grad_norm": 1.1202757836241906, "learning_rate": 1.9563806338783174e-05, "loss": 0.5724, "step": 3412 }, { "epoch": 0.12159538272440636, "grad_norm": 1.2133150075833488, "learning_rate": 1.9563469183194082e-05, "loss": 0.5758, "step": 3413 }, { "epoch": 0.12163100985090047, "grad_norm": 1.9327416188571978, "learning_rate": 1.9563131900260715e-05, "loss": 0.6358, "step": 3414 }, { "epoch": 0.12166663697739459, "grad_norm": 1.1686781910832538, "learning_rate": 1.9562794489987562e-05, "loss": 0.5306, "step": 3415 }, { "epoch": 0.12170226410388871, "grad_norm": 1.1392755479538044, "learning_rate": 1.9562456952379117e-05, "loss": 0.4982, "step": 3416 }, { "epoch": 0.12173789123038281, "grad_norm": 1.2177163454714492, "learning_rate": 1.9562119287439874e-05, "loss": 0.5363, "step": 3417 }, { "epoch": 0.12177351835687693, "grad_norm": 1.1566932032770463, "learning_rate": 1.9561781495174327e-05, "loss": 0.5068, "step": 3418 }, { "epoch": 0.12180914548337105, "grad_norm": 1.275177213995762, "learning_rate": 1.9561443575586975e-05, "loss": 0.5534, "step": 3419 }, { "epoch": 0.12184477260986515, "grad_norm": 1.2985769296631393, "learning_rate": 1.956110552868232e-05, "loss": 0.5414, "step": 3420 }, { "epoch": 0.12188039973635927, "grad_norm": 2.5143979635568527, "learning_rate": 1.9560767354464864e-05, "loss": 0.593, "step": 3421 }, { "epoch": 0.12191602686285338, "grad_norm": 1.1219181383434653, "learning_rate": 1.9560429052939108e-05, "loss": 0.4993, "step": 3422 }, { "epoch": 0.12195165398934749, "grad_norm": 1.675645722987929, "learning_rate": 1.9560090624109554e-05, "loss": 0.6914, "step": 3423 }, { "epoch": 0.1219872811158416, "grad_norm": 2.280347658416553, "learning_rate": 1.9559752067980717e-05, "loss": 0.7566, "step": 3424 }, { "epoch": 0.12202290824233572, "grad_norm": 1.8724963884135206, "learning_rate": 1.9559413384557095e-05, "loss": 0.6908, "step": 3425 }, { "epoch": 0.12205853536882982, "grad_norm": 1.255767197779252, "learning_rate": 1.95590745738432e-05, "loss": 0.5524, "step": 3426 }, { "epoch": 0.12209416249532394, "grad_norm": 1.207920268962752, "learning_rate": 1.955873563584355e-05, "loss": 0.5301, "step": 3427 }, { "epoch": 0.12212978962181806, "grad_norm": 1.429280501184093, "learning_rate": 1.955839657056265e-05, "loss": 0.5265, "step": 3428 }, { "epoch": 0.12216541674831216, "grad_norm": 1.6532800613296108, "learning_rate": 1.9558057378005025e-05, "loss": 0.6516, "step": 3429 }, { "epoch": 0.12220104387480628, "grad_norm": 2.5227308956019545, "learning_rate": 1.955771805817518e-05, "loss": 0.6129, "step": 3430 }, { "epoch": 0.1222366710013004, "grad_norm": 1.2354155749976097, "learning_rate": 1.955737861107764e-05, "loss": 0.5215, "step": 3431 }, { "epoch": 0.1222722981277945, "grad_norm": 1.191047957845585, "learning_rate": 1.9557039036716926e-05, "loss": 0.4571, "step": 3432 }, { "epoch": 0.12230792525428862, "grad_norm": 1.2303076278728549, "learning_rate": 1.9556699335097553e-05, "loss": 0.5354, "step": 3433 }, { "epoch": 0.12234355238078273, "grad_norm": 1.7387953840794876, "learning_rate": 1.955635950622405e-05, "loss": 0.5685, "step": 3434 }, { "epoch": 0.12237917950727684, "grad_norm": 1.1969729329670202, "learning_rate": 1.955601955010094e-05, "loss": 0.5284, "step": 3435 }, { "epoch": 0.12241480663377095, "grad_norm": 1.9250407612642426, "learning_rate": 1.9555679466732754e-05, "loss": 0.6508, "step": 3436 }, { "epoch": 0.12245043376026507, "grad_norm": 1.1494654664510073, "learning_rate": 1.9555339256124012e-05, "loss": 0.5496, "step": 3437 }, { "epoch": 0.12248606088675917, "grad_norm": 1.8105020337834354, "learning_rate": 1.9554998918279253e-05, "loss": 0.5978, "step": 3438 }, { "epoch": 0.12252168801325329, "grad_norm": 1.9659845515496663, "learning_rate": 1.9554658453203003e-05, "loss": 0.7239, "step": 3439 }, { "epoch": 0.12255731513974741, "grad_norm": 1.5100112542886934, "learning_rate": 1.9554317860899798e-05, "loss": 0.6172, "step": 3440 }, { "epoch": 0.12259294226624151, "grad_norm": 1.7913505967896994, "learning_rate": 1.9553977141374173e-05, "loss": 0.5602, "step": 3441 }, { "epoch": 0.12262856939273563, "grad_norm": 1.8811498504201536, "learning_rate": 1.9553636294630663e-05, "loss": 0.69, "step": 3442 }, { "epoch": 0.12266419651922975, "grad_norm": 1.2577486631263124, "learning_rate": 1.955329532067381e-05, "loss": 0.5259, "step": 3443 }, { "epoch": 0.12269982364572385, "grad_norm": 1.4041333300364902, "learning_rate": 1.955295421950815e-05, "loss": 0.5666, "step": 3444 }, { "epoch": 0.12273545077221797, "grad_norm": 1.5804653421675725, "learning_rate": 1.9552612991138228e-05, "loss": 0.5437, "step": 3445 }, { "epoch": 0.12277107789871208, "grad_norm": 1.1980557528679698, "learning_rate": 1.955227163556859e-05, "loss": 0.4842, "step": 3446 }, { "epoch": 0.12280670502520619, "grad_norm": 1.7424883788882481, "learning_rate": 1.9551930152803775e-05, "loss": 0.6142, "step": 3447 }, { "epoch": 0.1228423321517003, "grad_norm": 1.8106532948487881, "learning_rate": 1.9551588542848337e-05, "loss": 0.5728, "step": 3448 }, { "epoch": 0.12287795927819442, "grad_norm": 1.254858232226272, "learning_rate": 1.955124680570682e-05, "loss": 0.5166, "step": 3449 }, { "epoch": 0.12291358640468852, "grad_norm": 2.2920950327381364, "learning_rate": 1.955090494138377e-05, "loss": 0.6025, "step": 3450 }, { "epoch": 0.12294921353118264, "grad_norm": 1.321263818090712, "learning_rate": 1.9550562949883755e-05, "loss": 0.568, "step": 3451 }, { "epoch": 0.12298484065767676, "grad_norm": 1.3914720469410489, "learning_rate": 1.9550220831211313e-05, "loss": 0.5403, "step": 3452 }, { "epoch": 0.12302046778417086, "grad_norm": 2.321217779642699, "learning_rate": 1.9549878585371006e-05, "loss": 0.6064, "step": 3453 }, { "epoch": 0.12305609491066498, "grad_norm": 2.15534768451286, "learning_rate": 1.9549536212367394e-05, "loss": 0.6793, "step": 3454 }, { "epoch": 0.1230917220371591, "grad_norm": 1.1942460244155804, "learning_rate": 1.9549193712205028e-05, "loss": 0.554, "step": 3455 }, { "epoch": 0.1231273491636532, "grad_norm": 1.1479569606056097, "learning_rate": 1.9548851084888478e-05, "loss": 0.4994, "step": 3456 }, { "epoch": 0.12316297629014732, "grad_norm": 1.8523102542182293, "learning_rate": 1.9548508330422297e-05, "loss": 0.627, "step": 3457 }, { "epoch": 0.12319860341664143, "grad_norm": 1.2336486435207417, "learning_rate": 1.9548165448811054e-05, "loss": 0.5047, "step": 3458 }, { "epoch": 0.12323423054313554, "grad_norm": 1.915513493869833, "learning_rate": 1.9547822440059318e-05, "loss": 0.658, "step": 3459 }, { "epoch": 0.12326985766962965, "grad_norm": 1.5005254761971292, "learning_rate": 1.954747930417165e-05, "loss": 0.6361, "step": 3460 }, { "epoch": 0.12330548479612377, "grad_norm": 1.9027926540162332, "learning_rate": 1.954713604115262e-05, "loss": 0.5706, "step": 3461 }, { "epoch": 0.12334111192261789, "grad_norm": 2.235612520241174, "learning_rate": 1.9546792651006805e-05, "loss": 0.6737, "step": 3462 }, { "epoch": 0.12337673904911199, "grad_norm": 1.324707370565127, "learning_rate": 1.954644913373877e-05, "loss": 0.5747, "step": 3463 }, { "epoch": 0.12341236617560611, "grad_norm": 1.2412960221831462, "learning_rate": 1.954610548935309e-05, "loss": 0.5323, "step": 3464 }, { "epoch": 0.12344799330210023, "grad_norm": 1.574392370555455, "learning_rate": 1.9545761717854346e-05, "loss": 0.646, "step": 3465 }, { "epoch": 0.12348362042859433, "grad_norm": 1.4154365716117059, "learning_rate": 1.9545417819247113e-05, "loss": 0.5881, "step": 3466 }, { "epoch": 0.12351924755508845, "grad_norm": 2.32658997222044, "learning_rate": 1.9545073793535972e-05, "loss": 0.5798, "step": 3467 }, { "epoch": 0.12355487468158256, "grad_norm": 1.2417001624220507, "learning_rate": 1.95447296407255e-05, "loss": 0.5968, "step": 3468 }, { "epoch": 0.12359050180807667, "grad_norm": 1.6460230187693765, "learning_rate": 1.954438536082028e-05, "loss": 0.6621, "step": 3469 }, { "epoch": 0.12362612893457078, "grad_norm": 1.3948687590332565, "learning_rate": 1.95440409538249e-05, "loss": 0.6189, "step": 3470 }, { "epoch": 0.1236617560610649, "grad_norm": 2.0722615746452644, "learning_rate": 1.954369641974394e-05, "loss": 0.5988, "step": 3471 }, { "epoch": 0.123697383187559, "grad_norm": 1.1829157247018394, "learning_rate": 1.9543351758581995e-05, "loss": 0.5321, "step": 3472 }, { "epoch": 0.12373301031405312, "grad_norm": 1.8355680559092808, "learning_rate": 1.954300697034365e-05, "loss": 0.6703, "step": 3473 }, { "epoch": 0.12376863744054724, "grad_norm": 1.2073394641639483, "learning_rate": 1.9542662055033497e-05, "loss": 0.5452, "step": 3474 }, { "epoch": 0.12380426456704134, "grad_norm": 2.0495681259270357, "learning_rate": 1.954231701265613e-05, "loss": 0.6397, "step": 3475 }, { "epoch": 0.12383989169353546, "grad_norm": 1.6950338592069365, "learning_rate": 1.954197184321614e-05, "loss": 0.6212, "step": 3476 }, { "epoch": 0.12387551882002958, "grad_norm": 1.8972665039183247, "learning_rate": 1.9541626546718128e-05, "loss": 0.6176, "step": 3477 }, { "epoch": 0.12391114594652368, "grad_norm": 1.850179427106119, "learning_rate": 1.9541281123166685e-05, "loss": 0.5681, "step": 3478 }, { "epoch": 0.1239467730730178, "grad_norm": 1.7102240303895453, "learning_rate": 1.954093557256642e-05, "loss": 0.6905, "step": 3479 }, { "epoch": 0.12398240019951191, "grad_norm": 1.3402897659681698, "learning_rate": 1.9540589894921927e-05, "loss": 0.5699, "step": 3480 }, { "epoch": 0.12401802732600602, "grad_norm": 2.007914091797934, "learning_rate": 1.9540244090237808e-05, "loss": 0.6893, "step": 3481 }, { "epoch": 0.12405365445250013, "grad_norm": 1.344497602331119, "learning_rate": 1.9539898158518676e-05, "loss": 0.544, "step": 3482 }, { "epoch": 0.12408928157899425, "grad_norm": 1.975899859921477, "learning_rate": 1.9539552099769128e-05, "loss": 0.5754, "step": 3483 }, { "epoch": 0.12412490870548835, "grad_norm": 1.270023685705445, "learning_rate": 1.9539205913993776e-05, "loss": 0.5218, "step": 3484 }, { "epoch": 0.12416053583198247, "grad_norm": 1.273883443246729, "learning_rate": 1.9538859601197235e-05, "loss": 0.5279, "step": 3485 }, { "epoch": 0.12419616295847659, "grad_norm": 1.3003857205948453, "learning_rate": 1.9538513161384105e-05, "loss": 0.5011, "step": 3486 }, { "epoch": 0.12423179008497069, "grad_norm": 1.2948317415117618, "learning_rate": 1.9538166594559008e-05, "loss": 0.5356, "step": 3487 }, { "epoch": 0.12426741721146481, "grad_norm": 1.221447292258254, "learning_rate": 1.953781990072656e-05, "loss": 0.4922, "step": 3488 }, { "epoch": 0.12430304433795893, "grad_norm": 1.5663354438526786, "learning_rate": 1.9537473079891365e-05, "loss": 0.6231, "step": 3489 }, { "epoch": 0.12433867146445303, "grad_norm": 1.3626425136494462, "learning_rate": 1.9537126132058055e-05, "loss": 0.5976, "step": 3490 }, { "epoch": 0.12437429859094715, "grad_norm": 1.2488603427874307, "learning_rate": 1.9536779057231243e-05, "loss": 0.5247, "step": 3491 }, { "epoch": 0.12440992571744126, "grad_norm": 1.2181895193433203, "learning_rate": 1.953643185541555e-05, "loss": 0.5229, "step": 3492 }, { "epoch": 0.12444555284393537, "grad_norm": 1.2177545326072172, "learning_rate": 1.953608452661561e-05, "loss": 0.5014, "step": 3493 }, { "epoch": 0.12448117997042948, "grad_norm": 1.4751434996278863, "learning_rate": 1.9535737070836028e-05, "loss": 0.598, "step": 3494 }, { "epoch": 0.1245168070969236, "grad_norm": 1.7222885506940173, "learning_rate": 1.9535389488081447e-05, "loss": 0.6873, "step": 3495 }, { "epoch": 0.1245524342234177, "grad_norm": 2.1209074758412547, "learning_rate": 1.9535041778356492e-05, "loss": 0.673, "step": 3496 }, { "epoch": 0.12458806134991182, "grad_norm": 1.6490365168928551, "learning_rate": 1.9534693941665788e-05, "loss": 0.6169, "step": 3497 }, { "epoch": 0.12462368847640594, "grad_norm": 1.3178977237476837, "learning_rate": 1.9534345978013972e-05, "loss": 0.5422, "step": 3498 }, { "epoch": 0.12465931560290004, "grad_norm": 2.033207308778205, "learning_rate": 1.9533997887405673e-05, "loss": 0.6724, "step": 3499 }, { "epoch": 0.12469494272939416, "grad_norm": 1.3340858946285228, "learning_rate": 1.9533649669845533e-05, "loss": 0.5636, "step": 3500 }, { "epoch": 0.12473056985588828, "grad_norm": 1.302600960445934, "learning_rate": 1.9533301325338176e-05, "loss": 0.5668, "step": 3501 }, { "epoch": 0.12476619698238238, "grad_norm": 1.9407159607127773, "learning_rate": 1.9532952853888255e-05, "loss": 0.6127, "step": 3502 }, { "epoch": 0.1248018241088765, "grad_norm": 1.2673133338851081, "learning_rate": 1.95326042555004e-05, "loss": 0.5776, "step": 3503 }, { "epoch": 0.12483745123537061, "grad_norm": 1.2242692047395987, "learning_rate": 1.953225553017926e-05, "loss": 0.5513, "step": 3504 }, { "epoch": 0.12487307836186472, "grad_norm": 1.2046066313450379, "learning_rate": 1.9531906677929472e-05, "loss": 0.5019, "step": 3505 }, { "epoch": 0.12490870548835883, "grad_norm": 1.1666285717129337, "learning_rate": 1.9531557698755686e-05, "loss": 0.5378, "step": 3506 }, { "epoch": 0.12494433261485295, "grad_norm": 1.190448426875096, "learning_rate": 1.9531208592662545e-05, "loss": 0.6096, "step": 3507 }, { "epoch": 0.12497995974134705, "grad_norm": 1.359069086936094, "learning_rate": 1.95308593596547e-05, "loss": 0.5519, "step": 3508 }, { "epoch": 0.12501558686784117, "grad_norm": 1.6742715184919954, "learning_rate": 1.9530509999736803e-05, "loss": 0.6055, "step": 3509 }, { "epoch": 0.12505121399433528, "grad_norm": 1.827835015117334, "learning_rate": 1.9530160512913502e-05, "loss": 0.6307, "step": 3510 }, { "epoch": 0.1250868411208294, "grad_norm": 1.3398389167254152, "learning_rate": 1.9529810899189453e-05, "loss": 0.5785, "step": 3511 }, { "epoch": 0.1251224682473235, "grad_norm": 1.2392205722604581, "learning_rate": 1.952946115856931e-05, "loss": 0.5547, "step": 3512 }, { "epoch": 0.1251580953738176, "grad_norm": 2.0944014598243257, "learning_rate": 1.9529111291057735e-05, "loss": 0.6556, "step": 3513 }, { "epoch": 0.12519372250031174, "grad_norm": 1.0909057469994639, "learning_rate": 1.952876129665938e-05, "loss": 0.4524, "step": 3514 }, { "epoch": 0.12522934962680585, "grad_norm": 1.1863945971410312, "learning_rate": 1.9528411175378905e-05, "loss": 0.457, "step": 3515 }, { "epoch": 0.12526497675329995, "grad_norm": 1.671467440399828, "learning_rate": 1.952806092722098e-05, "loss": 0.7347, "step": 3516 }, { "epoch": 0.12530060387979408, "grad_norm": 1.251632888784763, "learning_rate": 1.9527710552190262e-05, "loss": 0.5084, "step": 3517 }, { "epoch": 0.12533623100628818, "grad_norm": 1.2264978286923518, "learning_rate": 1.952736005029142e-05, "loss": 0.5831, "step": 3518 }, { "epoch": 0.1253718581327823, "grad_norm": 1.2283057417197116, "learning_rate": 1.9527009421529117e-05, "loss": 0.5204, "step": 3519 }, { "epoch": 0.12540748525927642, "grad_norm": 1.7866422659959071, "learning_rate": 1.952665866590803e-05, "loss": 0.5659, "step": 3520 }, { "epoch": 0.12544311238577052, "grad_norm": 1.7039055763947706, "learning_rate": 1.9526307783432817e-05, "loss": 0.6743, "step": 3521 }, { "epoch": 0.12547873951226463, "grad_norm": 1.6619342310093173, "learning_rate": 1.9525956774108163e-05, "loss": 0.6375, "step": 3522 }, { "epoch": 0.12551436663875876, "grad_norm": 1.64764091660658, "learning_rate": 1.9525605637938733e-05, "loss": 0.4978, "step": 3523 }, { "epoch": 0.12554999376525286, "grad_norm": 1.0838053011492852, "learning_rate": 1.9525254374929205e-05, "loss": 0.4911, "step": 3524 }, { "epoch": 0.12558562089174696, "grad_norm": 1.397669629975662, "learning_rate": 1.952490298508426e-05, "loss": 0.5316, "step": 3525 }, { "epoch": 0.1256212480182411, "grad_norm": 1.208362669875761, "learning_rate": 1.952455146840857e-05, "loss": 0.5162, "step": 3526 }, { "epoch": 0.1256568751447352, "grad_norm": 1.1635734102707291, "learning_rate": 1.9524199824906826e-05, "loss": 0.5127, "step": 3527 }, { "epoch": 0.1256925022712293, "grad_norm": 2.099213184660945, "learning_rate": 1.9523848054583703e-05, "loss": 0.5959, "step": 3528 }, { "epoch": 0.12572812939772343, "grad_norm": 1.8775697722622935, "learning_rate": 1.9523496157443885e-05, "loss": 0.5852, "step": 3529 }, { "epoch": 0.12576375652421753, "grad_norm": 1.2177344045458476, "learning_rate": 1.9523144133492057e-05, "loss": 0.5223, "step": 3530 }, { "epoch": 0.12579938365071164, "grad_norm": 2.2147183264396366, "learning_rate": 1.9522791982732912e-05, "loss": 0.6537, "step": 3531 }, { "epoch": 0.12583501077720577, "grad_norm": 1.6742866302451793, "learning_rate": 1.9522439705171132e-05, "loss": 0.6649, "step": 3532 }, { "epoch": 0.12587063790369987, "grad_norm": 1.7780451609731076, "learning_rate": 1.9522087300811417e-05, "loss": 0.6676, "step": 3533 }, { "epoch": 0.125906265030194, "grad_norm": 1.9240360033574053, "learning_rate": 1.952173476965845e-05, "loss": 0.6029, "step": 3534 }, { "epoch": 0.1259418921566881, "grad_norm": 1.8759630048286138, "learning_rate": 1.952138211171693e-05, "loss": 0.6954, "step": 3535 }, { "epoch": 0.1259775192831822, "grad_norm": 1.2737484275775834, "learning_rate": 1.952102932699155e-05, "loss": 0.5497, "step": 3536 }, { "epoch": 0.12601314640967634, "grad_norm": 1.215285782726115, "learning_rate": 1.9520676415487013e-05, "loss": 0.5123, "step": 3537 }, { "epoch": 0.12604877353617044, "grad_norm": 1.8958540072754904, "learning_rate": 1.9520323377208017e-05, "loss": 0.7135, "step": 3538 }, { "epoch": 0.12608440066266455, "grad_norm": 2.1606182333002386, "learning_rate": 1.9519970212159257e-05, "loss": 0.7041, "step": 3539 }, { "epoch": 0.12612002778915868, "grad_norm": 1.2218403375349287, "learning_rate": 1.951961692034544e-05, "loss": 0.5374, "step": 3540 }, { "epoch": 0.12615565491565278, "grad_norm": 1.177214096866539, "learning_rate": 1.951926350177127e-05, "loss": 0.498, "step": 3541 }, { "epoch": 0.12619128204214689, "grad_norm": 2.0637949465278163, "learning_rate": 1.9518909956441454e-05, "loss": 0.6596, "step": 3542 }, { "epoch": 0.12622690916864102, "grad_norm": 1.456494398500466, "learning_rate": 1.9518556284360696e-05, "loss": 0.5391, "step": 3543 }, { "epoch": 0.12626253629513512, "grad_norm": 1.754248297131124, "learning_rate": 1.951820248553371e-05, "loss": 0.6067, "step": 3544 }, { "epoch": 0.12629816342162922, "grad_norm": 1.7126840673946582, "learning_rate": 1.9517848559965207e-05, "loss": 0.6102, "step": 3545 }, { "epoch": 0.12633379054812335, "grad_norm": 1.5069357358543676, "learning_rate": 1.9517494507659894e-05, "loss": 0.5794, "step": 3546 }, { "epoch": 0.12636941767461746, "grad_norm": 1.2409647683383023, "learning_rate": 1.951714032862249e-05, "loss": 0.4972, "step": 3547 }, { "epoch": 0.12640504480111156, "grad_norm": 1.464686791912883, "learning_rate": 1.9516786022857714e-05, "loss": 0.5753, "step": 3548 }, { "epoch": 0.1264406719276057, "grad_norm": 2.1204203093144134, "learning_rate": 1.951643159037028e-05, "loss": 0.6877, "step": 3549 }, { "epoch": 0.1264762990540998, "grad_norm": 1.2682522170139467, "learning_rate": 1.9516077031164905e-05, "loss": 0.5336, "step": 3550 }, { "epoch": 0.1265119261805939, "grad_norm": 3.6512568837953894, "learning_rate": 1.951572234524631e-05, "loss": 0.6866, "step": 3551 }, { "epoch": 0.12654755330708803, "grad_norm": 1.59228646088918, "learning_rate": 1.9515367532619226e-05, "loss": 0.5053, "step": 3552 }, { "epoch": 0.12658318043358213, "grad_norm": 1.271961180493113, "learning_rate": 1.9515012593288373e-05, "loss": 0.5798, "step": 3553 }, { "epoch": 0.12661880756007624, "grad_norm": 1.8727336012655542, "learning_rate": 1.9514657527258474e-05, "loss": 0.6387, "step": 3554 }, { "epoch": 0.12665443468657037, "grad_norm": 1.680504030279802, "learning_rate": 1.951430233453426e-05, "loss": 0.6994, "step": 3555 }, { "epoch": 0.12669006181306447, "grad_norm": 1.9919019378036884, "learning_rate": 1.951394701512046e-05, "loss": 0.54, "step": 3556 }, { "epoch": 0.12672568893955857, "grad_norm": 1.297707219695959, "learning_rate": 1.9513591569021807e-05, "loss": 0.5342, "step": 3557 }, { "epoch": 0.1267613160660527, "grad_norm": 1.6660252814080263, "learning_rate": 1.9513235996243033e-05, "loss": 0.6466, "step": 3558 }, { "epoch": 0.1267969431925468, "grad_norm": 1.2918170146713575, "learning_rate": 1.951288029678887e-05, "loss": 0.5429, "step": 3559 }, { "epoch": 0.1268325703190409, "grad_norm": 1.307302208585382, "learning_rate": 1.9512524470664058e-05, "loss": 0.5991, "step": 3560 }, { "epoch": 0.12686819744553504, "grad_norm": 1.9406021272886642, "learning_rate": 1.9512168517873333e-05, "loss": 0.6787, "step": 3561 }, { "epoch": 0.12690382457202914, "grad_norm": 1.21695612847769, "learning_rate": 1.951181243842144e-05, "loss": 0.5363, "step": 3562 }, { "epoch": 0.12693945169852325, "grad_norm": 1.7895415617096782, "learning_rate": 1.951145623231311e-05, "loss": 0.6024, "step": 3563 }, { "epoch": 0.12697507882501738, "grad_norm": 1.898472946499965, "learning_rate": 1.9511099899553097e-05, "loss": 0.6069, "step": 3564 }, { "epoch": 0.12701070595151148, "grad_norm": 1.1882249741417867, "learning_rate": 1.9510743440146136e-05, "loss": 0.5154, "step": 3565 }, { "epoch": 0.12704633307800559, "grad_norm": 1.1952215107528519, "learning_rate": 1.9510386854096982e-05, "loss": 0.5603, "step": 3566 }, { "epoch": 0.12708196020449972, "grad_norm": 2.031809439129384, "learning_rate": 1.951003014141038e-05, "loss": 0.6268, "step": 3567 }, { "epoch": 0.12711758733099382, "grad_norm": 1.2793277016268374, "learning_rate": 1.9509673302091075e-05, "loss": 0.5754, "step": 3568 }, { "epoch": 0.12715321445748792, "grad_norm": 1.1833695610702446, "learning_rate": 1.950931633614383e-05, "loss": 0.5671, "step": 3569 }, { "epoch": 0.12718884158398205, "grad_norm": 1.2105920200930844, "learning_rate": 1.9508959243573387e-05, "loss": 0.5298, "step": 3570 }, { "epoch": 0.12722446871047616, "grad_norm": 2.327851966930602, "learning_rate": 1.9508602024384507e-05, "loss": 0.6168, "step": 3571 }, { "epoch": 0.12726009583697026, "grad_norm": 1.1751748874482169, "learning_rate": 1.950824467858194e-05, "loss": 0.5155, "step": 3572 }, { "epoch": 0.1272957229634644, "grad_norm": 2.3002001820959177, "learning_rate": 1.9507887206170456e-05, "loss": 0.6809, "step": 3573 }, { "epoch": 0.1273313500899585, "grad_norm": 1.5706428866359394, "learning_rate": 1.9507529607154807e-05, "loss": 0.6212, "step": 3574 }, { "epoch": 0.1273669772164526, "grad_norm": 1.305855913779787, "learning_rate": 1.9507171881539754e-05, "loss": 0.5145, "step": 3575 }, { "epoch": 0.12740260434294673, "grad_norm": 2.0071678921981264, "learning_rate": 1.950681402933006e-05, "loss": 0.6758, "step": 3576 }, { "epoch": 0.12743823146944083, "grad_norm": 2.6046785548074576, "learning_rate": 1.95064560505305e-05, "loss": 0.6107, "step": 3577 }, { "epoch": 0.12747385859593494, "grad_norm": 1.4315506033554801, "learning_rate": 1.9506097945145825e-05, "loss": 0.5685, "step": 3578 }, { "epoch": 0.12750948572242907, "grad_norm": 1.1539696591751678, "learning_rate": 1.9505739713180815e-05, "loss": 0.5273, "step": 3579 }, { "epoch": 0.12754511284892317, "grad_norm": 1.6214390557833382, "learning_rate": 1.9505381354640236e-05, "loss": 0.4722, "step": 3580 }, { "epoch": 0.12758073997541727, "grad_norm": 1.3198091136611543, "learning_rate": 1.9505022869528862e-05, "loss": 0.5717, "step": 3581 }, { "epoch": 0.1276163671019114, "grad_norm": 3.5170062799410275, "learning_rate": 1.9504664257851464e-05, "loss": 0.6014, "step": 3582 }, { "epoch": 0.1276519942284055, "grad_norm": 1.2015893981241823, "learning_rate": 1.9504305519612815e-05, "loss": 0.5399, "step": 3583 }, { "epoch": 0.1276876213548996, "grad_norm": 1.2427118449104184, "learning_rate": 1.95039466548177e-05, "loss": 0.6448, "step": 3584 }, { "epoch": 0.12772324848139374, "grad_norm": 1.2882984699825701, "learning_rate": 1.9503587663470887e-05, "loss": 0.5056, "step": 3585 }, { "epoch": 0.12775887560788785, "grad_norm": 1.1995050323694074, "learning_rate": 1.9503228545577167e-05, "loss": 0.5445, "step": 3586 }, { "epoch": 0.12779450273438195, "grad_norm": 1.860287331126816, "learning_rate": 1.9502869301141315e-05, "loss": 0.6087, "step": 3587 }, { "epoch": 0.12783012986087608, "grad_norm": 1.2102621924838632, "learning_rate": 1.9502509930168113e-05, "loss": 0.5857, "step": 3588 }, { "epoch": 0.12786575698737018, "grad_norm": 1.655302709916832, "learning_rate": 1.9502150432662354e-05, "loss": 0.5863, "step": 3589 }, { "epoch": 0.12790138411386429, "grad_norm": 1.216270986823595, "learning_rate": 1.9501790808628815e-05, "loss": 0.5255, "step": 3590 }, { "epoch": 0.12793701124035842, "grad_norm": 1.187048678723801, "learning_rate": 1.9501431058072295e-05, "loss": 0.5053, "step": 3591 }, { "epoch": 0.12797263836685252, "grad_norm": 1.330432687277322, "learning_rate": 1.9501071180997577e-05, "loss": 0.5404, "step": 3592 }, { "epoch": 0.12800826549334662, "grad_norm": 1.2226330979730875, "learning_rate": 1.9500711177409456e-05, "loss": 0.5072, "step": 3593 }, { "epoch": 0.12804389261984075, "grad_norm": 2.2145111129183475, "learning_rate": 1.9500351047312726e-05, "loss": 0.6041, "step": 3594 }, { "epoch": 0.12807951974633486, "grad_norm": 1.4974052624402496, "learning_rate": 1.9499990790712177e-05, "loss": 0.5497, "step": 3595 }, { "epoch": 0.12811514687282896, "grad_norm": 1.6463666180545986, "learning_rate": 1.9499630407612614e-05, "loss": 0.6194, "step": 3596 }, { "epoch": 0.1281507739993231, "grad_norm": 1.947923834347585, "learning_rate": 1.9499269898018832e-05, "loss": 0.6032, "step": 3597 }, { "epoch": 0.1281864011258172, "grad_norm": 1.252005718773025, "learning_rate": 1.9498909261935634e-05, "loss": 0.548, "step": 3598 }, { "epoch": 0.1282220282523113, "grad_norm": 1.72141956288492, "learning_rate": 1.9498548499367818e-05, "loss": 0.5765, "step": 3599 }, { "epoch": 0.12825765537880543, "grad_norm": 2.342745883111322, "learning_rate": 1.9498187610320188e-05, "loss": 0.7002, "step": 3600 }, { "epoch": 0.12829328250529953, "grad_norm": 2.142480062960569, "learning_rate": 1.9497826594797553e-05, "loss": 0.6767, "step": 3601 }, { "epoch": 0.12832890963179364, "grad_norm": 1.8351942990108858, "learning_rate": 1.9497465452804717e-05, "loss": 0.5911, "step": 3602 }, { "epoch": 0.12836453675828777, "grad_norm": 1.8084197503591162, "learning_rate": 1.949710418434649e-05, "loss": 0.6381, "step": 3603 }, { "epoch": 0.12840016388478187, "grad_norm": 1.603437666857697, "learning_rate": 1.9496742789427684e-05, "loss": 0.611, "step": 3604 }, { "epoch": 0.12843579101127597, "grad_norm": 1.2562377159201246, "learning_rate": 1.949638126805311e-05, "loss": 0.5383, "step": 3605 }, { "epoch": 0.1284714181377701, "grad_norm": 1.280729628638976, "learning_rate": 1.9496019620227583e-05, "loss": 0.5522, "step": 3606 }, { "epoch": 0.1285070452642642, "grad_norm": 2.081466335049656, "learning_rate": 1.9495657845955914e-05, "loss": 0.5812, "step": 3607 }, { "epoch": 0.1285426723907583, "grad_norm": 1.1014759775753646, "learning_rate": 1.9495295945242927e-05, "loss": 0.5098, "step": 3608 }, { "epoch": 0.12857829951725244, "grad_norm": 2.0355280197116095, "learning_rate": 1.9494933918093437e-05, "loss": 0.5731, "step": 3609 }, { "epoch": 0.12861392664374655, "grad_norm": 1.273782218878308, "learning_rate": 1.9494571764512265e-05, "loss": 0.5951, "step": 3610 }, { "epoch": 0.12864955377024065, "grad_norm": 1.2090996455040337, "learning_rate": 1.9494209484504233e-05, "loss": 0.5228, "step": 3611 }, { "epoch": 0.12868518089673478, "grad_norm": 1.6631561786685118, "learning_rate": 1.9493847078074166e-05, "loss": 0.6794, "step": 3612 }, { "epoch": 0.12872080802322888, "grad_norm": 1.2689032727152034, "learning_rate": 1.9493484545226887e-05, "loss": 0.544, "step": 3613 }, { "epoch": 0.128756435149723, "grad_norm": 1.1644012015580358, "learning_rate": 1.949312188596723e-05, "loss": 0.5881, "step": 3614 }, { "epoch": 0.12879206227621712, "grad_norm": 1.470047192595255, "learning_rate": 1.949275910030002e-05, "loss": 0.5499, "step": 3615 }, { "epoch": 0.12882768940271122, "grad_norm": 1.2644484967451026, "learning_rate": 1.9492396188230083e-05, "loss": 0.5202, "step": 3616 }, { "epoch": 0.12886331652920532, "grad_norm": 1.7483843143153663, "learning_rate": 1.9492033149762258e-05, "loss": 0.6846, "step": 3617 }, { "epoch": 0.12889894365569946, "grad_norm": 1.1966103323423147, "learning_rate": 1.9491669984901377e-05, "loss": 0.5257, "step": 3618 }, { "epoch": 0.12893457078219356, "grad_norm": 1.7816049197775612, "learning_rate": 1.9491306693652276e-05, "loss": 0.5759, "step": 3619 }, { "epoch": 0.12897019790868766, "grad_norm": 1.212989491717934, "learning_rate": 1.949094327601979e-05, "loss": 0.5237, "step": 3620 }, { "epoch": 0.1290058250351818, "grad_norm": 1.7652881111494438, "learning_rate": 1.9490579732008765e-05, "loss": 0.5542, "step": 3621 }, { "epoch": 0.1290414521616759, "grad_norm": 2.5570596975202338, "learning_rate": 1.9490216061624038e-05, "loss": 0.5515, "step": 3622 }, { "epoch": 0.12907707928817, "grad_norm": 1.1199136857901864, "learning_rate": 1.948985226487045e-05, "loss": 0.507, "step": 3623 }, { "epoch": 0.12911270641466413, "grad_norm": 1.0907358369215867, "learning_rate": 1.948948834175284e-05, "loss": 0.5166, "step": 3624 }, { "epoch": 0.12914833354115823, "grad_norm": 1.1937939194438452, "learning_rate": 1.9489124292276066e-05, "loss": 0.4757, "step": 3625 }, { "epoch": 0.12918396066765234, "grad_norm": 1.5349571242237738, "learning_rate": 1.9488760116444966e-05, "loss": 0.6664, "step": 3626 }, { "epoch": 0.12921958779414647, "grad_norm": 1.5422408228167823, "learning_rate": 1.9488395814264394e-05, "loss": 0.6154, "step": 3627 }, { "epoch": 0.12925521492064057, "grad_norm": 2.3891106138371767, "learning_rate": 1.9488031385739203e-05, "loss": 0.6234, "step": 3628 }, { "epoch": 0.1292908420471347, "grad_norm": 1.7681086002956727, "learning_rate": 1.948766683087424e-05, "loss": 0.6717, "step": 3629 }, { "epoch": 0.1293264691736288, "grad_norm": 1.9989366074990356, "learning_rate": 1.9487302149674362e-05, "loss": 0.6376, "step": 3630 }, { "epoch": 0.1293620963001229, "grad_norm": 1.6551640855268508, "learning_rate": 1.9486937342144423e-05, "loss": 0.6235, "step": 3631 }, { "epoch": 0.12939772342661704, "grad_norm": 2.4909244011774616, "learning_rate": 1.9486572408289285e-05, "loss": 0.5283, "step": 3632 }, { "epoch": 0.12943335055311114, "grad_norm": 1.7826935471243, "learning_rate": 1.9486207348113803e-05, "loss": 0.6699, "step": 3633 }, { "epoch": 0.12946897767960525, "grad_norm": 1.2889610174471169, "learning_rate": 1.9485842161622835e-05, "loss": 0.5145, "step": 3634 }, { "epoch": 0.12950460480609938, "grad_norm": 2.1176560454386855, "learning_rate": 1.9485476848821257e-05, "loss": 0.6489, "step": 3635 }, { "epoch": 0.12954023193259348, "grad_norm": 1.198840183784045, "learning_rate": 1.948511140971392e-05, "loss": 0.4595, "step": 3636 }, { "epoch": 0.12957585905908758, "grad_norm": 1.4144280565089442, "learning_rate": 1.9484745844305694e-05, "loss": 0.5436, "step": 3637 }, { "epoch": 0.12961148618558171, "grad_norm": 1.071115267387803, "learning_rate": 1.948438015260145e-05, "loss": 0.483, "step": 3638 }, { "epoch": 0.12964711331207582, "grad_norm": 1.4698505795397816, "learning_rate": 1.9484014334606058e-05, "loss": 0.6396, "step": 3639 }, { "epoch": 0.12968274043856992, "grad_norm": 1.2298515663317018, "learning_rate": 1.9483648390324382e-05, "loss": 0.5182, "step": 3640 }, { "epoch": 0.12971836756506405, "grad_norm": 1.1366421312775188, "learning_rate": 1.9483282319761298e-05, "loss": 0.4961, "step": 3641 }, { "epoch": 0.12975399469155816, "grad_norm": 1.9488302031264741, "learning_rate": 1.9482916122921685e-05, "loss": 0.6586, "step": 3642 }, { "epoch": 0.12978962181805226, "grad_norm": 2.0290469269751856, "learning_rate": 1.9482549799810412e-05, "loss": 0.5789, "step": 3643 }, { "epoch": 0.1298252489445464, "grad_norm": 1.238849270241606, "learning_rate": 1.948218335043236e-05, "loss": 0.5211, "step": 3644 }, { "epoch": 0.1298608760710405, "grad_norm": 1.2364795160383457, "learning_rate": 1.9481816774792415e-05, "loss": 0.5105, "step": 3645 }, { "epoch": 0.1298965031975346, "grad_norm": 2.4691777447525656, "learning_rate": 1.948145007289545e-05, "loss": 0.6368, "step": 3646 }, { "epoch": 0.12993213032402873, "grad_norm": 1.563810691029489, "learning_rate": 1.9481083244746347e-05, "loss": 0.5431, "step": 3647 }, { "epoch": 0.12996775745052283, "grad_norm": 1.2883116947809665, "learning_rate": 1.9480716290349998e-05, "loss": 0.5484, "step": 3648 }, { "epoch": 0.13000338457701693, "grad_norm": 1.2310118729270296, "learning_rate": 1.948034920971128e-05, "loss": 0.518, "step": 3649 }, { "epoch": 0.13003901170351106, "grad_norm": 1.2100718280406133, "learning_rate": 1.947998200283509e-05, "loss": 0.5517, "step": 3650 }, { "epoch": 0.13007463883000517, "grad_norm": 1.2949864910508706, "learning_rate": 1.9479614669726313e-05, "loss": 0.5527, "step": 3651 }, { "epoch": 0.13011026595649927, "grad_norm": 1.7879977303534487, "learning_rate": 1.9479247210389838e-05, "loss": 0.6026, "step": 3652 }, { "epoch": 0.1301458930829934, "grad_norm": 1.5402228083793326, "learning_rate": 1.9478879624830564e-05, "loss": 0.6505, "step": 3653 }, { "epoch": 0.1301815202094875, "grad_norm": 1.395791923345807, "learning_rate": 1.947851191305338e-05, "loss": 0.5502, "step": 3654 }, { "epoch": 0.1302171473359816, "grad_norm": 2.3992576056076333, "learning_rate": 1.947814407506319e-05, "loss": 0.641, "step": 3655 }, { "epoch": 0.13025277446247574, "grad_norm": 1.649029914016385, "learning_rate": 1.947777611086488e-05, "loss": 0.5564, "step": 3656 }, { "epoch": 0.13028840158896984, "grad_norm": 1.1355231339904808, "learning_rate": 1.9477408020463357e-05, "loss": 0.4835, "step": 3657 }, { "epoch": 0.13032402871546395, "grad_norm": 1.5680380335519708, "learning_rate": 1.9477039803863524e-05, "loss": 0.5678, "step": 3658 }, { "epoch": 0.13035965584195808, "grad_norm": 3.4616752755759834, "learning_rate": 1.9476671461070286e-05, "loss": 0.6178, "step": 3659 }, { "epoch": 0.13039528296845218, "grad_norm": 1.3376984421549458, "learning_rate": 1.9476302992088536e-05, "loss": 0.5471, "step": 3660 }, { "epoch": 0.13043091009494628, "grad_norm": 2.1620509985174072, "learning_rate": 1.9475934396923194e-05, "loss": 0.619, "step": 3661 }, { "epoch": 0.13046653722144042, "grad_norm": 2.087760763280712, "learning_rate": 1.947556567557916e-05, "loss": 0.6529, "step": 3662 }, { "epoch": 0.13050216434793452, "grad_norm": 2.2278530135833714, "learning_rate": 1.9475196828061346e-05, "loss": 0.6134, "step": 3663 }, { "epoch": 0.13053779147442862, "grad_norm": 1.7876553691524022, "learning_rate": 1.9474827854374662e-05, "loss": 0.6493, "step": 3664 }, { "epoch": 0.13057341860092275, "grad_norm": 1.2435353673594878, "learning_rate": 1.9474458754524023e-05, "loss": 0.5079, "step": 3665 }, { "epoch": 0.13060904572741686, "grad_norm": 2.107432778630245, "learning_rate": 1.9474089528514346e-05, "loss": 0.6686, "step": 3666 }, { "epoch": 0.13064467285391096, "grad_norm": 2.031981064776749, "learning_rate": 1.9473720176350546e-05, "loss": 0.6003, "step": 3667 }, { "epoch": 0.1306802999804051, "grad_norm": 1.2732787135980002, "learning_rate": 1.9473350698037537e-05, "loss": 0.5278, "step": 3668 }, { "epoch": 0.1307159271068992, "grad_norm": 2.214887771306917, "learning_rate": 1.947298109358024e-05, "loss": 0.5921, "step": 3669 }, { "epoch": 0.1307515542333933, "grad_norm": 1.4296688732745153, "learning_rate": 1.9472611362983583e-05, "loss": 0.5304, "step": 3670 }, { "epoch": 0.13078718135988743, "grad_norm": 2.6081531293537643, "learning_rate": 1.9472241506252482e-05, "loss": 0.5731, "step": 3671 }, { "epoch": 0.13082280848638153, "grad_norm": 1.2027360931040783, "learning_rate": 1.9471871523391866e-05, "loss": 0.489, "step": 3672 }, { "epoch": 0.13085843561287563, "grad_norm": 1.1359145708271146, "learning_rate": 1.9471501414406662e-05, "loss": 0.496, "step": 3673 }, { "epoch": 0.13089406273936977, "grad_norm": 1.8704769186276586, "learning_rate": 1.9471131179301793e-05, "loss": 0.6689, "step": 3674 }, { "epoch": 0.13092968986586387, "grad_norm": 1.2595208524404515, "learning_rate": 1.9470760818082192e-05, "loss": 0.5337, "step": 3675 }, { "epoch": 0.13096531699235797, "grad_norm": 2.045642975068311, "learning_rate": 1.947039033075279e-05, "loss": 0.636, "step": 3676 }, { "epoch": 0.1310009441188521, "grad_norm": 1.6674785976359745, "learning_rate": 1.9470019717318525e-05, "loss": 0.6009, "step": 3677 }, { "epoch": 0.1310365712453462, "grad_norm": 1.584942712455826, "learning_rate": 1.946964897778433e-05, "loss": 0.652, "step": 3678 }, { "epoch": 0.1310721983718403, "grad_norm": 1.2108257161890896, "learning_rate": 1.9469278112155138e-05, "loss": 0.5296, "step": 3679 }, { "epoch": 0.13110782549833444, "grad_norm": 1.7606726605905432, "learning_rate": 1.946890712043589e-05, "loss": 0.6946, "step": 3680 }, { "epoch": 0.13114345262482854, "grad_norm": 1.2655419511781525, "learning_rate": 1.946853600263152e-05, "loss": 0.5398, "step": 3681 }, { "epoch": 0.13117907975132265, "grad_norm": 1.9035260733828143, "learning_rate": 1.946816475874698e-05, "loss": 0.7096, "step": 3682 }, { "epoch": 0.13121470687781678, "grad_norm": 1.283318772123704, "learning_rate": 1.946779338878721e-05, "loss": 0.5531, "step": 3683 }, { "epoch": 0.13125033400431088, "grad_norm": 1.955290222416975, "learning_rate": 1.946742189275715e-05, "loss": 0.5838, "step": 3684 }, { "epoch": 0.13128596113080498, "grad_norm": 1.2868300780558612, "learning_rate": 1.946705027066175e-05, "loss": 0.5334, "step": 3685 }, { "epoch": 0.13132158825729912, "grad_norm": 1.8509539748192314, "learning_rate": 1.9466678522505963e-05, "loss": 0.5131, "step": 3686 }, { "epoch": 0.13135721538379322, "grad_norm": 1.2661263283635544, "learning_rate": 1.946630664829473e-05, "loss": 0.5222, "step": 3687 }, { "epoch": 0.13139284251028732, "grad_norm": 1.9554699572764027, "learning_rate": 1.9465934648033008e-05, "loss": 0.6145, "step": 3688 }, { "epoch": 0.13142846963678145, "grad_norm": 2.16403646332597, "learning_rate": 1.946556252172575e-05, "loss": 0.6354, "step": 3689 }, { "epoch": 0.13146409676327556, "grad_norm": 1.805082817761927, "learning_rate": 1.946519026937791e-05, "loss": 0.5754, "step": 3690 }, { "epoch": 0.13149972388976966, "grad_norm": 1.3467553611992662, "learning_rate": 1.946481789099445e-05, "loss": 0.5052, "step": 3691 }, { "epoch": 0.1315353510162638, "grad_norm": 1.9811739281467728, "learning_rate": 1.946444538658032e-05, "loss": 0.6472, "step": 3692 }, { "epoch": 0.1315709781427579, "grad_norm": 2.204700588247676, "learning_rate": 1.9464072756140487e-05, "loss": 0.6423, "step": 3693 }, { "epoch": 0.131606605269252, "grad_norm": 2.837733481942953, "learning_rate": 1.9463699999679908e-05, "loss": 0.6022, "step": 3694 }, { "epoch": 0.13164223239574613, "grad_norm": 1.3233135297797143, "learning_rate": 1.9463327117203554e-05, "loss": 0.5985, "step": 3695 }, { "epoch": 0.13167785952224023, "grad_norm": 1.2695719339696074, "learning_rate": 1.9462954108716378e-05, "loss": 0.518, "step": 3696 }, { "epoch": 0.13171348664873433, "grad_norm": 1.1948526209809789, "learning_rate": 1.9462580974223355e-05, "loss": 0.526, "step": 3697 }, { "epoch": 0.13174911377522847, "grad_norm": 1.3273477396041733, "learning_rate": 1.9462207713729454e-05, "loss": 0.4691, "step": 3698 }, { "epoch": 0.13178474090172257, "grad_norm": 1.613633499799466, "learning_rate": 1.9461834327239646e-05, "loss": 0.6162, "step": 3699 }, { "epoch": 0.13182036802821667, "grad_norm": 1.8757360415994595, "learning_rate": 1.94614608147589e-05, "loss": 0.6742, "step": 3700 }, { "epoch": 0.1318559951547108, "grad_norm": 2.4934630562109117, "learning_rate": 1.9461087176292185e-05, "loss": 0.5809, "step": 3701 }, { "epoch": 0.1318916222812049, "grad_norm": 1.2487806742461747, "learning_rate": 1.946071341184449e-05, "loss": 0.4963, "step": 3702 }, { "epoch": 0.131927249407699, "grad_norm": 1.180465993815016, "learning_rate": 1.946033952142077e-05, "loss": 0.5371, "step": 3703 }, { "epoch": 0.13196287653419314, "grad_norm": 1.7724697283425894, "learning_rate": 1.9459965505026028e-05, "loss": 0.5891, "step": 3704 }, { "epoch": 0.13199850366068724, "grad_norm": 1.1972227260537742, "learning_rate": 1.945959136266523e-05, "loss": 0.5144, "step": 3705 }, { "epoch": 0.13203413078718135, "grad_norm": 1.5467016017782627, "learning_rate": 1.945921709434336e-05, "loss": 0.617, "step": 3706 }, { "epoch": 0.13206975791367548, "grad_norm": 2.1698436210883196, "learning_rate": 1.9458842700065408e-05, "loss": 0.6043, "step": 3707 }, { "epoch": 0.13210538504016958, "grad_norm": 1.1822372601027733, "learning_rate": 1.9458468179836346e-05, "loss": 0.5681, "step": 3708 }, { "epoch": 0.13214101216666368, "grad_norm": 1.136420042877147, "learning_rate": 1.9458093533661176e-05, "loss": 0.4996, "step": 3709 }, { "epoch": 0.13217663929315782, "grad_norm": 1.3128221690951956, "learning_rate": 1.9457718761544877e-05, "loss": 0.547, "step": 3710 }, { "epoch": 0.13221226641965192, "grad_norm": 1.2020461043844979, "learning_rate": 1.9457343863492443e-05, "loss": 0.5572, "step": 3711 }, { "epoch": 0.13224789354614602, "grad_norm": 1.2186514327372053, "learning_rate": 1.945696883950886e-05, "loss": 0.5311, "step": 3712 }, { "epoch": 0.13228352067264015, "grad_norm": 1.9873404960704089, "learning_rate": 1.9456593689599135e-05, "loss": 0.6072, "step": 3713 }, { "epoch": 0.13231914779913426, "grad_norm": 1.2424992525551144, "learning_rate": 1.945621841376825e-05, "loss": 0.4914, "step": 3714 }, { "epoch": 0.13235477492562836, "grad_norm": 1.769254323542695, "learning_rate": 1.945584301202121e-05, "loss": 0.5822, "step": 3715 }, { "epoch": 0.1323904020521225, "grad_norm": 1.1727983215955142, "learning_rate": 1.9455467484363008e-05, "loss": 0.4901, "step": 3716 }, { "epoch": 0.1324260291786166, "grad_norm": 1.7423589174094263, "learning_rate": 1.945509183079865e-05, "loss": 0.6545, "step": 3717 }, { "epoch": 0.1324616563051107, "grad_norm": 1.2752244335399119, "learning_rate": 1.9454716051333137e-05, "loss": 0.5221, "step": 3718 }, { "epoch": 0.13249728343160483, "grad_norm": 1.884603558797947, "learning_rate": 1.945434014597147e-05, "loss": 0.5476, "step": 3719 }, { "epoch": 0.13253291055809893, "grad_norm": 1.2324703435202211, "learning_rate": 1.9453964114718655e-05, "loss": 0.5128, "step": 3720 }, { "epoch": 0.13256853768459306, "grad_norm": 1.2940473995453068, "learning_rate": 1.94535879575797e-05, "loss": 0.6136, "step": 3721 }, { "epoch": 0.13260416481108717, "grad_norm": 1.1823724634567068, "learning_rate": 1.9453211674559613e-05, "loss": 0.5021, "step": 3722 }, { "epoch": 0.13263979193758127, "grad_norm": 1.2455394523780337, "learning_rate": 1.9452835265663404e-05, "loss": 0.5409, "step": 3723 }, { "epoch": 0.1326754190640754, "grad_norm": 1.1780938173040312, "learning_rate": 1.945245873089609e-05, "loss": 0.4723, "step": 3724 }, { "epoch": 0.1327110461905695, "grad_norm": 1.6969344676424616, "learning_rate": 1.9452082070262678e-05, "loss": 0.605, "step": 3725 }, { "epoch": 0.1327466733170636, "grad_norm": 1.239413310496525, "learning_rate": 1.9451705283768186e-05, "loss": 0.5499, "step": 3726 }, { "epoch": 0.13278230044355774, "grad_norm": 1.832717446631869, "learning_rate": 1.9451328371417636e-05, "loss": 0.5999, "step": 3727 }, { "epoch": 0.13281792757005184, "grad_norm": 1.4010197137849234, "learning_rate": 1.945095133321604e-05, "loss": 0.5073, "step": 3728 }, { "epoch": 0.13285355469654594, "grad_norm": 1.7251149950293128, "learning_rate": 1.945057416916842e-05, "loss": 0.5875, "step": 3729 }, { "epoch": 0.13288918182304008, "grad_norm": 1.8452142350232126, "learning_rate": 1.94501968792798e-05, "loss": 0.5311, "step": 3730 }, { "epoch": 0.13292480894953418, "grad_norm": 1.2398826080467678, "learning_rate": 1.9449819463555206e-05, "loss": 0.5363, "step": 3731 }, { "epoch": 0.13296043607602828, "grad_norm": 1.2603336282393471, "learning_rate": 1.944944192199966e-05, "loss": 0.5802, "step": 3732 }, { "epoch": 0.1329960632025224, "grad_norm": 1.5674279387174408, "learning_rate": 1.944906425461819e-05, "loss": 0.6866, "step": 3733 }, { "epoch": 0.13303169032901652, "grad_norm": 1.7783886415814036, "learning_rate": 1.9448686461415822e-05, "loss": 0.6418, "step": 3734 }, { "epoch": 0.13306731745551062, "grad_norm": 1.25887919449839, "learning_rate": 1.9448308542397592e-05, "loss": 0.5041, "step": 3735 }, { "epoch": 0.13310294458200475, "grad_norm": 1.143507986838614, "learning_rate": 1.9447930497568528e-05, "loss": 0.5067, "step": 3736 }, { "epoch": 0.13313857170849885, "grad_norm": 1.98509535362838, "learning_rate": 1.9447552326933666e-05, "loss": 0.69, "step": 3737 }, { "epoch": 0.13317419883499296, "grad_norm": 1.1860741590178285, "learning_rate": 1.9447174030498043e-05, "loss": 0.5133, "step": 3738 }, { "epoch": 0.1332098259614871, "grad_norm": 1.831292275719804, "learning_rate": 1.9446795608266693e-05, "loss": 0.5918, "step": 3739 }, { "epoch": 0.1332454530879812, "grad_norm": 1.2441466168026454, "learning_rate": 1.9446417060244656e-05, "loss": 0.5255, "step": 3740 }, { "epoch": 0.1332810802144753, "grad_norm": 1.746973444157492, "learning_rate": 1.9446038386436977e-05, "loss": 0.5928, "step": 3741 }, { "epoch": 0.13331670734096943, "grad_norm": 1.1571888870108, "learning_rate": 1.9445659586848692e-05, "loss": 0.5343, "step": 3742 }, { "epoch": 0.13335233446746353, "grad_norm": 1.381521011262847, "learning_rate": 1.9445280661484848e-05, "loss": 0.6042, "step": 3743 }, { "epoch": 0.13338796159395763, "grad_norm": 2.410024759301169, "learning_rate": 1.944490161035049e-05, "loss": 0.675, "step": 3744 }, { "epoch": 0.13342358872045176, "grad_norm": 1.2632088935601051, "learning_rate": 1.9444522433450662e-05, "loss": 0.5524, "step": 3745 }, { "epoch": 0.13345921584694587, "grad_norm": 1.884116224431166, "learning_rate": 1.9444143130790422e-05, "loss": 0.6797, "step": 3746 }, { "epoch": 0.13349484297343997, "grad_norm": 1.170756198701548, "learning_rate": 1.944376370237481e-05, "loss": 0.52, "step": 3747 }, { "epoch": 0.1335304700999341, "grad_norm": 1.482947571091444, "learning_rate": 1.944338414820889e-05, "loss": 0.6909, "step": 3748 }, { "epoch": 0.1335660972264282, "grad_norm": 2.4074249083202455, "learning_rate": 1.9443004468297704e-05, "loss": 0.612, "step": 3749 }, { "epoch": 0.1336017243529223, "grad_norm": 1.852898747104689, "learning_rate": 1.9442624662646317e-05, "loss": 0.6577, "step": 3750 }, { "epoch": 0.13363735147941644, "grad_norm": 1.8436622219286598, "learning_rate": 1.9442244731259778e-05, "loss": 0.5522, "step": 3751 }, { "epoch": 0.13367297860591054, "grad_norm": 1.090292181748128, "learning_rate": 1.9441864674143158e-05, "loss": 0.5124, "step": 3752 }, { "epoch": 0.13370860573240465, "grad_norm": 2.0500407550645527, "learning_rate": 1.9441484491301505e-05, "loss": 0.6295, "step": 3753 }, { "epoch": 0.13374423285889878, "grad_norm": 1.1645971740219077, "learning_rate": 1.9441104182739884e-05, "loss": 0.5472, "step": 3754 }, { "epoch": 0.13377985998539288, "grad_norm": 1.08494863501594, "learning_rate": 1.9440723748463365e-05, "loss": 0.4623, "step": 3755 }, { "epoch": 0.13381548711188698, "grad_norm": 1.3852529585210882, "learning_rate": 1.944034318847701e-05, "loss": 0.689, "step": 3756 }, { "epoch": 0.1338511142383811, "grad_norm": 1.8758675570835794, "learning_rate": 1.943996250278589e-05, "loss": 0.6479, "step": 3757 }, { "epoch": 0.13388674136487522, "grad_norm": 3.628930032357648, "learning_rate": 1.943958169139507e-05, "loss": 0.6066, "step": 3758 }, { "epoch": 0.13392236849136932, "grad_norm": 1.2470485231820858, "learning_rate": 1.943920075430962e-05, "loss": 0.5428, "step": 3759 }, { "epoch": 0.13395799561786345, "grad_norm": 1.2016231441131062, "learning_rate": 1.9438819691534615e-05, "loss": 0.4583, "step": 3760 }, { "epoch": 0.13399362274435755, "grad_norm": 1.480661033051766, "learning_rate": 1.943843850307513e-05, "loss": 0.6332, "step": 3761 }, { "epoch": 0.13402924987085166, "grad_norm": 1.2375104743692034, "learning_rate": 1.9438057188936237e-05, "loss": 0.535, "step": 3762 }, { "epoch": 0.1340648769973458, "grad_norm": 1.2853070852350639, "learning_rate": 1.9437675749123017e-05, "loss": 0.5123, "step": 3763 }, { "epoch": 0.1341005041238399, "grad_norm": 1.6735158200349156, "learning_rate": 1.943729418364055e-05, "loss": 0.6771, "step": 3764 }, { "epoch": 0.134136131250334, "grad_norm": 1.9572843634579362, "learning_rate": 1.943691249249391e-05, "loss": 0.6475, "step": 3765 }, { "epoch": 0.13417175837682813, "grad_norm": 1.299888447324538, "learning_rate": 1.943653067568819e-05, "loss": 0.5512, "step": 3766 }, { "epoch": 0.13420738550332223, "grad_norm": 1.286755029101419, "learning_rate": 1.9436148733228463e-05, "loss": 0.5317, "step": 3767 }, { "epoch": 0.13424301262981633, "grad_norm": 1.5117785143729254, "learning_rate": 1.9435766665119823e-05, "loss": 0.5735, "step": 3768 }, { "epoch": 0.13427863975631046, "grad_norm": 1.616164216081863, "learning_rate": 1.943538447136735e-05, "loss": 0.5688, "step": 3769 }, { "epoch": 0.13431426688280457, "grad_norm": 1.6872075896567418, "learning_rate": 1.9435002151976144e-05, "loss": 0.6636, "step": 3770 }, { "epoch": 0.13434989400929867, "grad_norm": 1.8689300211031636, "learning_rate": 1.9434619706951286e-05, "loss": 0.626, "step": 3771 }, { "epoch": 0.1343855211357928, "grad_norm": 1.6959998534826608, "learning_rate": 1.9434237136297876e-05, "loss": 0.6632, "step": 3772 }, { "epoch": 0.1344211482622869, "grad_norm": 1.7809929810318632, "learning_rate": 1.9433854440021e-05, "loss": 0.6559, "step": 3773 }, { "epoch": 0.134456775388781, "grad_norm": 2.2189928176922584, "learning_rate": 1.943347161812576e-05, "loss": 0.6364, "step": 3774 }, { "epoch": 0.13449240251527514, "grad_norm": 3.1379731513934837, "learning_rate": 1.9433088670617254e-05, "loss": 0.6909, "step": 3775 }, { "epoch": 0.13452802964176924, "grad_norm": 2.110265130532374, "learning_rate": 1.9432705597500573e-05, "loss": 0.5954, "step": 3776 }, { "epoch": 0.13456365676826335, "grad_norm": 1.299217555075943, "learning_rate": 1.943232239878083e-05, "loss": 0.5382, "step": 3777 }, { "epoch": 0.13459928389475748, "grad_norm": 1.678560862080956, "learning_rate": 1.943193907446312e-05, "loss": 0.6675, "step": 3778 }, { "epoch": 0.13463491102125158, "grad_norm": 1.323022119294691, "learning_rate": 1.9431555624552546e-05, "loss": 0.5201, "step": 3779 }, { "epoch": 0.13467053814774568, "grad_norm": 1.2046837228536675, "learning_rate": 1.943117204905422e-05, "loss": 0.5173, "step": 3780 }, { "epoch": 0.13470616527423981, "grad_norm": 1.3077655185198078, "learning_rate": 1.9430788347973245e-05, "loss": 0.5127, "step": 3781 }, { "epoch": 0.13474179240073392, "grad_norm": 1.1162139833314624, "learning_rate": 1.943040452131473e-05, "loss": 0.4721, "step": 3782 }, { "epoch": 0.13477741952722802, "grad_norm": 1.2993779838060435, "learning_rate": 1.943002056908379e-05, "loss": 0.5704, "step": 3783 }, { "epoch": 0.13481304665372215, "grad_norm": 1.3237711514160033, "learning_rate": 1.942963649128553e-05, "loss": 0.543, "step": 3784 }, { "epoch": 0.13484867378021625, "grad_norm": 2.3049348218396895, "learning_rate": 1.942925228792507e-05, "loss": 0.6061, "step": 3785 }, { "epoch": 0.13488430090671036, "grad_norm": 1.2261688439811473, "learning_rate": 1.942886795900753e-05, "loss": 0.5052, "step": 3786 }, { "epoch": 0.1349199280332045, "grad_norm": 1.5783779489982843, "learning_rate": 1.9428483504538023e-05, "loss": 0.7311, "step": 3787 }, { "epoch": 0.1349555551596986, "grad_norm": 1.2636960631176928, "learning_rate": 1.9428098924521665e-05, "loss": 0.543, "step": 3788 }, { "epoch": 0.1349911822861927, "grad_norm": 1.2767858745417173, "learning_rate": 1.942771421896358e-05, "loss": 0.5474, "step": 3789 }, { "epoch": 0.13502680941268683, "grad_norm": 2.8110823628911445, "learning_rate": 1.942732938786889e-05, "loss": 0.6175, "step": 3790 }, { "epoch": 0.13506243653918093, "grad_norm": 1.6100144131758118, "learning_rate": 1.942694443124272e-05, "loss": 0.5714, "step": 3791 }, { "epoch": 0.13509806366567503, "grad_norm": 1.7014850889949098, "learning_rate": 1.9426559349090198e-05, "loss": 0.6489, "step": 3792 }, { "epoch": 0.13513369079216916, "grad_norm": 2.0779852567034647, "learning_rate": 1.942617414141645e-05, "loss": 0.5846, "step": 3793 }, { "epoch": 0.13516931791866327, "grad_norm": 1.2354591265307422, "learning_rate": 1.9425788808226604e-05, "loss": 0.4931, "step": 3794 }, { "epoch": 0.13520494504515737, "grad_norm": 1.508604926987472, "learning_rate": 1.9425403349525788e-05, "loss": 0.6286, "step": 3795 }, { "epoch": 0.1352405721716515, "grad_norm": 1.185669079239241, "learning_rate": 1.942501776531914e-05, "loss": 0.5095, "step": 3796 }, { "epoch": 0.1352761992981456, "grad_norm": 1.2502212587716646, "learning_rate": 1.9424632055611797e-05, "loss": 0.5353, "step": 3797 }, { "epoch": 0.1353118264246397, "grad_norm": 1.2120317622352315, "learning_rate": 1.9424246220408888e-05, "loss": 0.5252, "step": 3798 }, { "epoch": 0.13534745355113384, "grad_norm": 1.3012358828707735, "learning_rate": 1.942386025971555e-05, "loss": 0.511, "step": 3799 }, { "epoch": 0.13538308067762794, "grad_norm": 1.9492791327992114, "learning_rate": 1.9423474173536932e-05, "loss": 0.6291, "step": 3800 }, { "epoch": 0.13541870780412205, "grad_norm": 1.436486228161076, "learning_rate": 1.9423087961878165e-05, "loss": 0.5137, "step": 3801 }, { "epoch": 0.13545433493061618, "grad_norm": 1.9384594048820991, "learning_rate": 1.9422701624744396e-05, "loss": 0.6688, "step": 3802 }, { "epoch": 0.13548996205711028, "grad_norm": 1.2705793004724648, "learning_rate": 1.942231516214077e-05, "loss": 0.5528, "step": 3803 }, { "epoch": 0.13552558918360438, "grad_norm": 1.1464765635021552, "learning_rate": 1.942192857407243e-05, "loss": 0.5165, "step": 3804 }, { "epoch": 0.13556121631009851, "grad_norm": 1.654242202376475, "learning_rate": 1.9421541860544525e-05, "loss": 0.6414, "step": 3805 }, { "epoch": 0.13559684343659262, "grad_norm": 1.9018848293879809, "learning_rate": 1.9421155021562206e-05, "loss": 0.5903, "step": 3806 }, { "epoch": 0.13563247056308672, "grad_norm": 1.2662176502936617, "learning_rate": 1.9420768057130622e-05, "loss": 0.556, "step": 3807 }, { "epoch": 0.13566809768958085, "grad_norm": 7.499047719164222, "learning_rate": 1.9420380967254932e-05, "loss": 0.5515, "step": 3808 }, { "epoch": 0.13570372481607496, "grad_norm": 1.1878332663270277, "learning_rate": 1.941999375194028e-05, "loss": 0.5221, "step": 3809 }, { "epoch": 0.13573935194256906, "grad_norm": 1.113672012576542, "learning_rate": 1.941960641119183e-05, "loss": 0.4839, "step": 3810 }, { "epoch": 0.1357749790690632, "grad_norm": 1.7919796565271429, "learning_rate": 1.941921894501473e-05, "loss": 0.5974, "step": 3811 }, { "epoch": 0.1358106061955573, "grad_norm": 1.4179765208685116, "learning_rate": 1.941883135341415e-05, "loss": 0.5847, "step": 3812 }, { "epoch": 0.1358462333220514, "grad_norm": 1.5467554959813328, "learning_rate": 1.941844363639525e-05, "loss": 0.6048, "step": 3813 }, { "epoch": 0.13588186044854553, "grad_norm": 1.9422873298167758, "learning_rate": 1.9418055793963185e-05, "loss": 0.6435, "step": 3814 }, { "epoch": 0.13591748757503963, "grad_norm": 1.9107347222120354, "learning_rate": 1.9417667826123128e-05, "loss": 0.5649, "step": 3815 }, { "epoch": 0.13595311470153376, "grad_norm": 1.1512362004037444, "learning_rate": 1.941727973288024e-05, "loss": 0.518, "step": 3816 }, { "epoch": 0.13598874182802786, "grad_norm": 1.2413774239395965, "learning_rate": 1.9416891514239688e-05, "loss": 0.5441, "step": 3817 }, { "epoch": 0.13602436895452197, "grad_norm": 1.7681307481370805, "learning_rate": 1.9416503170206647e-05, "loss": 0.5593, "step": 3818 }, { "epoch": 0.1360599960810161, "grad_norm": 1.1670770279752745, "learning_rate": 1.9416114700786285e-05, "loss": 0.5246, "step": 3819 }, { "epoch": 0.1360956232075102, "grad_norm": 2.3841129991244485, "learning_rate": 1.9415726105983773e-05, "loss": 0.6372, "step": 3820 }, { "epoch": 0.1361312503340043, "grad_norm": 1.2275569480036648, "learning_rate": 1.9415337385804285e-05, "loss": 0.5565, "step": 3821 }, { "epoch": 0.13616687746049844, "grad_norm": 1.4115209390678474, "learning_rate": 1.9414948540253e-05, "loss": 0.5526, "step": 3822 }, { "epoch": 0.13620250458699254, "grad_norm": 1.257021870480003, "learning_rate": 1.9414559569335093e-05, "loss": 0.5441, "step": 3823 }, { "epoch": 0.13623813171348664, "grad_norm": 2.0818723488606667, "learning_rate": 1.9414170473055747e-05, "loss": 0.6799, "step": 3824 }, { "epoch": 0.13627375883998077, "grad_norm": 1.407004654744572, "learning_rate": 1.941378125142014e-05, "loss": 0.5896, "step": 3825 }, { "epoch": 0.13630938596647488, "grad_norm": 2.135573555651192, "learning_rate": 1.9413391904433458e-05, "loss": 0.6302, "step": 3826 }, { "epoch": 0.13634501309296898, "grad_norm": 1.8119416121025111, "learning_rate": 1.9413002432100882e-05, "loss": 0.6198, "step": 3827 }, { "epoch": 0.1363806402194631, "grad_norm": 1.2603888291483791, "learning_rate": 1.9412612834427597e-05, "loss": 0.5647, "step": 3828 }, { "epoch": 0.13641626734595722, "grad_norm": 1.3182530594835937, "learning_rate": 1.9412223111418795e-05, "loss": 0.5633, "step": 3829 }, { "epoch": 0.13645189447245132, "grad_norm": 1.62154202928481, "learning_rate": 1.9411833263079665e-05, "loss": 0.6606, "step": 3830 }, { "epoch": 0.13648752159894545, "grad_norm": 2.025272999292529, "learning_rate": 1.9411443289415396e-05, "loss": 0.7643, "step": 3831 }, { "epoch": 0.13652314872543955, "grad_norm": 1.971910868067791, "learning_rate": 1.9411053190431176e-05, "loss": 0.6009, "step": 3832 }, { "epoch": 0.13655877585193366, "grad_norm": 2.08196280006863, "learning_rate": 1.9410662966132213e-05, "loss": 0.7726, "step": 3833 }, { "epoch": 0.1365944029784278, "grad_norm": 1.2092911136081919, "learning_rate": 1.9410272616523688e-05, "loss": 0.5533, "step": 3834 }, { "epoch": 0.1366300301049219, "grad_norm": 3.413705535839601, "learning_rate": 1.9409882141610813e-05, "loss": 0.5855, "step": 3835 }, { "epoch": 0.136665657231416, "grad_norm": 1.7632434673776745, "learning_rate": 1.9409491541398774e-05, "loss": 0.5437, "step": 3836 }, { "epoch": 0.13670128435791012, "grad_norm": 1.8272594339093615, "learning_rate": 1.940910081589278e-05, "loss": 0.5979, "step": 3837 }, { "epoch": 0.13673691148440423, "grad_norm": 2.7708545803348255, "learning_rate": 1.9408709965098035e-05, "loss": 0.6175, "step": 3838 }, { "epoch": 0.13677253861089833, "grad_norm": 1.2685263596634397, "learning_rate": 1.9408318989019736e-05, "loss": 0.5568, "step": 3839 }, { "epoch": 0.13680816573739246, "grad_norm": 1.691484178959519, "learning_rate": 1.9407927887663097e-05, "loss": 0.5641, "step": 3840 }, { "epoch": 0.13684379286388657, "grad_norm": 2.2347579304905256, "learning_rate": 1.9407536661033323e-05, "loss": 0.5677, "step": 3841 }, { "epoch": 0.13687941999038067, "grad_norm": 1.194067269910947, "learning_rate": 1.9407145309135617e-05, "loss": 0.4616, "step": 3842 }, { "epoch": 0.1369150471168748, "grad_norm": 1.3137523256001542, "learning_rate": 1.9406753831975202e-05, "loss": 0.5429, "step": 3843 }, { "epoch": 0.1369506742433689, "grad_norm": 1.2216894830207607, "learning_rate": 1.9406362229557283e-05, "loss": 0.5123, "step": 3844 }, { "epoch": 0.136986301369863, "grad_norm": 4.260591969977043, "learning_rate": 1.9405970501887074e-05, "loss": 0.6984, "step": 3845 }, { "epoch": 0.13702192849635714, "grad_norm": 1.1402267492188904, "learning_rate": 1.9405578648969796e-05, "loss": 0.5143, "step": 3846 }, { "epoch": 0.13705755562285124, "grad_norm": 1.9242406288474831, "learning_rate": 1.9405186670810667e-05, "loss": 0.6309, "step": 3847 }, { "epoch": 0.13709318274934534, "grad_norm": 1.7399713255624811, "learning_rate": 1.9404794567414897e-05, "loss": 0.6017, "step": 3848 }, { "epoch": 0.13712880987583947, "grad_norm": 1.2748293713636991, "learning_rate": 1.940440233878772e-05, "loss": 0.5155, "step": 3849 }, { "epoch": 0.13716443700233358, "grad_norm": 1.329280397274314, "learning_rate": 1.9404009984934347e-05, "loss": 0.5638, "step": 3850 }, { "epoch": 0.13720006412882768, "grad_norm": 1.510630323259796, "learning_rate": 1.940361750586001e-05, "loss": 0.637, "step": 3851 }, { "epoch": 0.1372356912553218, "grad_norm": 1.9729992675280754, "learning_rate": 1.9403224901569936e-05, "loss": 0.616, "step": 3852 }, { "epoch": 0.13727131838181592, "grad_norm": 1.197069875547572, "learning_rate": 1.9402832172069347e-05, "loss": 0.521, "step": 3853 }, { "epoch": 0.13730694550831002, "grad_norm": 1.1917669973053882, "learning_rate": 1.9402439317363475e-05, "loss": 0.4968, "step": 3854 }, { "epoch": 0.13734257263480415, "grad_norm": 2.0202037590185387, "learning_rate": 1.940204633745755e-05, "loss": 0.6156, "step": 3855 }, { "epoch": 0.13737819976129825, "grad_norm": 1.27128244406371, "learning_rate": 1.9401653232356813e-05, "loss": 0.5368, "step": 3856 }, { "epoch": 0.13741382688779236, "grad_norm": 1.5348427890398062, "learning_rate": 1.940126000206649e-05, "loss": 0.5534, "step": 3857 }, { "epoch": 0.1374494540142865, "grad_norm": 1.7293341136242562, "learning_rate": 1.9400866646591816e-05, "loss": 0.6004, "step": 3858 }, { "epoch": 0.1374850811407806, "grad_norm": 1.7575317897026022, "learning_rate": 1.940047316593803e-05, "loss": 0.6434, "step": 3859 }, { "epoch": 0.1375207082672747, "grad_norm": 1.584862871823689, "learning_rate": 1.9400079560110376e-05, "loss": 0.6108, "step": 3860 }, { "epoch": 0.13755633539376882, "grad_norm": 1.864250737049499, "learning_rate": 1.9399685829114093e-05, "loss": 0.6586, "step": 3861 }, { "epoch": 0.13759196252026293, "grad_norm": 1.1919078272854988, "learning_rate": 1.9399291972954425e-05, "loss": 0.521, "step": 3862 }, { "epoch": 0.13762758964675703, "grad_norm": 1.9315289679204857, "learning_rate": 1.9398897991636613e-05, "loss": 0.6362, "step": 3863 }, { "epoch": 0.13766321677325116, "grad_norm": 1.2454652455808566, "learning_rate": 1.9398503885165902e-05, "loss": 0.5332, "step": 3864 }, { "epoch": 0.13769884389974527, "grad_norm": 1.4972046956140663, "learning_rate": 1.9398109653547547e-05, "loss": 0.6155, "step": 3865 }, { "epoch": 0.13773447102623937, "grad_norm": 2.310947212888839, "learning_rate": 1.939771529678679e-05, "loss": 0.6815, "step": 3866 }, { "epoch": 0.1377700981527335, "grad_norm": 2.2177886665521527, "learning_rate": 1.9397320814888883e-05, "loss": 0.6522, "step": 3867 }, { "epoch": 0.1378057252792276, "grad_norm": 1.7936848920423085, "learning_rate": 1.9396926207859085e-05, "loss": 0.6218, "step": 3868 }, { "epoch": 0.1378413524057217, "grad_norm": 1.1758947373124846, "learning_rate": 1.9396531475702646e-05, "loss": 0.5313, "step": 3869 }, { "epoch": 0.13787697953221584, "grad_norm": 1.1653035210100502, "learning_rate": 1.9396136618424824e-05, "loss": 0.5137, "step": 3870 }, { "epoch": 0.13791260665870994, "grad_norm": 1.727231365996934, "learning_rate": 1.939574163603087e-05, "loss": 0.6253, "step": 3871 }, { "epoch": 0.13794823378520404, "grad_norm": 1.773309498372922, "learning_rate": 1.9395346528526055e-05, "loss": 0.6213, "step": 3872 }, { "epoch": 0.13798386091169818, "grad_norm": 1.1853630932320849, "learning_rate": 1.9394951295915628e-05, "loss": 0.5423, "step": 3873 }, { "epoch": 0.13801948803819228, "grad_norm": 1.7266544209307577, "learning_rate": 1.9394555938204866e-05, "loss": 0.7097, "step": 3874 }, { "epoch": 0.13805511516468638, "grad_norm": 2.4823634640663625, "learning_rate": 1.939416045539902e-05, "loss": 0.6256, "step": 3875 }, { "epoch": 0.1380907422911805, "grad_norm": 1.1164513598561485, "learning_rate": 1.939376484750336e-05, "loss": 0.5406, "step": 3876 }, { "epoch": 0.13812636941767462, "grad_norm": 2.0495007386080544, "learning_rate": 1.939336911452316e-05, "loss": 0.6239, "step": 3877 }, { "epoch": 0.13816199654416872, "grad_norm": 2.3464142637549723, "learning_rate": 1.9392973256463682e-05, "loss": 0.6243, "step": 3878 }, { "epoch": 0.13819762367066285, "grad_norm": 1.739390321622836, "learning_rate": 1.93925772733302e-05, "loss": 0.647, "step": 3879 }, { "epoch": 0.13823325079715695, "grad_norm": 1.5076010158949236, "learning_rate": 1.9392181165127984e-05, "loss": 0.5769, "step": 3880 }, { "epoch": 0.13826887792365106, "grad_norm": 2.096876599524458, "learning_rate": 1.9391784931862313e-05, "loss": 0.6859, "step": 3881 }, { "epoch": 0.1383045050501452, "grad_norm": 1.2074559285606674, "learning_rate": 1.9391388573538463e-05, "loss": 0.566, "step": 3882 }, { "epoch": 0.1383401321766393, "grad_norm": 1.753159864918723, "learning_rate": 1.9390992090161706e-05, "loss": 0.5689, "step": 3883 }, { "epoch": 0.1383757593031334, "grad_norm": 1.357494153491143, "learning_rate": 1.9390595481737327e-05, "loss": 0.5678, "step": 3884 }, { "epoch": 0.13841138642962753, "grad_norm": 1.2623101589974455, "learning_rate": 1.9390198748270607e-05, "loss": 0.5131, "step": 3885 }, { "epoch": 0.13844701355612163, "grad_norm": 2.1571875402576786, "learning_rate": 1.9389801889766828e-05, "loss": 0.6456, "step": 3886 }, { "epoch": 0.13848264068261573, "grad_norm": 1.5698615370037439, "learning_rate": 1.9389404906231272e-05, "loss": 0.5143, "step": 3887 }, { "epoch": 0.13851826780910986, "grad_norm": 1.358884416145516, "learning_rate": 1.9389007797669225e-05, "loss": 0.5318, "step": 3888 }, { "epoch": 0.13855389493560397, "grad_norm": 1.8269580217433259, "learning_rate": 1.9388610564085982e-05, "loss": 0.5641, "step": 3889 }, { "epoch": 0.13858952206209807, "grad_norm": 1.171143967250792, "learning_rate": 1.9388213205486823e-05, "loss": 0.5228, "step": 3890 }, { "epoch": 0.1386251491885922, "grad_norm": 1.3426091586674695, "learning_rate": 1.9387815721877045e-05, "loss": 0.478, "step": 3891 }, { "epoch": 0.1386607763150863, "grad_norm": 1.9753800011160376, "learning_rate": 1.9387418113261938e-05, "loss": 0.7061, "step": 3892 }, { "epoch": 0.1386964034415804, "grad_norm": 1.2315418288816438, "learning_rate": 1.9387020379646798e-05, "loss": 0.6024, "step": 3893 }, { "epoch": 0.13873203056807454, "grad_norm": 1.2132205201590027, "learning_rate": 1.9386622521036917e-05, "loss": 0.5383, "step": 3894 }, { "epoch": 0.13876765769456864, "grad_norm": 1.639072947936617, "learning_rate": 1.93862245374376e-05, "loss": 0.6102, "step": 3895 }, { "epoch": 0.13880328482106274, "grad_norm": 1.7568315156574663, "learning_rate": 1.938582642885414e-05, "loss": 0.6814, "step": 3896 }, { "epoch": 0.13883891194755688, "grad_norm": 1.0890223313398337, "learning_rate": 1.9385428195291848e-05, "loss": 0.4632, "step": 3897 }, { "epoch": 0.13887453907405098, "grad_norm": 1.8789155347148445, "learning_rate": 1.9385029836756013e-05, "loss": 0.7481, "step": 3898 }, { "epoch": 0.13891016620054508, "grad_norm": 1.2546364144491735, "learning_rate": 1.938463135325195e-05, "loss": 0.502, "step": 3899 }, { "epoch": 0.1389457933270392, "grad_norm": 2.121378119995204, "learning_rate": 1.9384232744784955e-05, "loss": 0.454, "step": 3900 }, { "epoch": 0.13898142045353332, "grad_norm": 1.126261604127757, "learning_rate": 1.9383834011360347e-05, "loss": 0.4652, "step": 3901 }, { "epoch": 0.13901704758002742, "grad_norm": 1.3327112805297516, "learning_rate": 1.9383435152983428e-05, "loss": 0.545, "step": 3902 }, { "epoch": 0.13905267470652155, "grad_norm": 1.2513962502511282, "learning_rate": 1.9383036169659513e-05, "loss": 0.5709, "step": 3903 }, { "epoch": 0.13908830183301565, "grad_norm": 1.9502429760567752, "learning_rate": 1.938263706139391e-05, "loss": 0.5843, "step": 3904 }, { "epoch": 0.13912392895950976, "grad_norm": 1.219738307133733, "learning_rate": 1.938223782819194e-05, "loss": 0.5054, "step": 3905 }, { "epoch": 0.1391595560860039, "grad_norm": 1.1418370619753166, "learning_rate": 1.9381838470058913e-05, "loss": 0.4932, "step": 3906 }, { "epoch": 0.139195183212498, "grad_norm": 4.466235199213758, "learning_rate": 1.938143898700015e-05, "loss": 0.5835, "step": 3907 }, { "epoch": 0.1392308103389921, "grad_norm": 1.1456232754753117, "learning_rate": 1.938103937902097e-05, "loss": 0.5608, "step": 3908 }, { "epoch": 0.13926643746548623, "grad_norm": 1.2711003785370798, "learning_rate": 1.9380639646126692e-05, "loss": 0.5127, "step": 3909 }, { "epoch": 0.13930206459198033, "grad_norm": 1.364579129431999, "learning_rate": 1.9380239788322642e-05, "loss": 0.5125, "step": 3910 }, { "epoch": 0.13933769171847446, "grad_norm": 1.128409882600541, "learning_rate": 1.9379839805614144e-05, "loss": 0.482, "step": 3911 }, { "epoch": 0.13937331884496856, "grad_norm": 1.8796131957433848, "learning_rate": 1.9379439698006522e-05, "loss": 0.6228, "step": 3912 }, { "epoch": 0.13940894597146267, "grad_norm": 1.3499428357386234, "learning_rate": 1.9379039465505105e-05, "loss": 0.5415, "step": 3913 }, { "epoch": 0.1394445730979568, "grad_norm": 1.1175704324646765, "learning_rate": 1.937863910811522e-05, "loss": 0.5248, "step": 3914 }, { "epoch": 0.1394802002244509, "grad_norm": 1.1489625950705682, "learning_rate": 1.93782386258422e-05, "loss": 0.5543, "step": 3915 }, { "epoch": 0.139515827350945, "grad_norm": 1.2030704440416307, "learning_rate": 1.937783801869138e-05, "loss": 0.5692, "step": 3916 }, { "epoch": 0.13955145447743914, "grad_norm": 2.522053531219731, "learning_rate": 1.937743728666809e-05, "loss": 0.5564, "step": 3917 }, { "epoch": 0.13958708160393324, "grad_norm": 1.2109255099633345, "learning_rate": 1.9377036429777673e-05, "loss": 0.5189, "step": 3918 }, { "epoch": 0.13962270873042734, "grad_norm": 1.6864732331683618, "learning_rate": 1.9376635448025456e-05, "loss": 0.6457, "step": 3919 }, { "epoch": 0.13965833585692147, "grad_norm": 1.2595455818467127, "learning_rate": 1.9376234341416784e-05, "loss": 0.507, "step": 3920 }, { "epoch": 0.13969396298341558, "grad_norm": 1.06649272075772, "learning_rate": 1.9375833109957e-05, "loss": 0.4811, "step": 3921 }, { "epoch": 0.13972959010990968, "grad_norm": 1.2138350768209618, "learning_rate": 1.9375431753651447e-05, "loss": 0.5408, "step": 3922 }, { "epoch": 0.1397652172364038, "grad_norm": 1.822261141119437, "learning_rate": 1.9375030272505463e-05, "loss": 0.6231, "step": 3923 }, { "epoch": 0.1398008443628979, "grad_norm": 1.2324358613458193, "learning_rate": 1.93746286665244e-05, "loss": 0.5506, "step": 3924 }, { "epoch": 0.13983647148939202, "grad_norm": 1.260391619273978, "learning_rate": 1.9374226935713603e-05, "loss": 0.5003, "step": 3925 }, { "epoch": 0.13987209861588615, "grad_norm": 1.6379678603275054, "learning_rate": 1.9373825080078426e-05, "loss": 0.6016, "step": 3926 }, { "epoch": 0.13990772574238025, "grad_norm": 1.4247312324735224, "learning_rate": 1.9373423099624213e-05, "loss": 0.6122, "step": 3927 }, { "epoch": 0.13994335286887435, "grad_norm": 1.8063508338995593, "learning_rate": 1.937302099435632e-05, "loss": 0.649, "step": 3928 }, { "epoch": 0.13997897999536849, "grad_norm": 1.6644994091454794, "learning_rate": 1.9372618764280102e-05, "loss": 0.5693, "step": 3929 }, { "epoch": 0.1400146071218626, "grad_norm": 1.5850654720716022, "learning_rate": 1.937221640940091e-05, "loss": 0.6199, "step": 3930 }, { "epoch": 0.1400502342483567, "grad_norm": 1.2161327533247923, "learning_rate": 1.937181392972411e-05, "loss": 0.521, "step": 3931 }, { "epoch": 0.14008586137485082, "grad_norm": 1.1708239967633072, "learning_rate": 1.9371411325255057e-05, "loss": 0.5011, "step": 3932 }, { "epoch": 0.14012148850134493, "grad_norm": 1.6971346840702275, "learning_rate": 1.9371008595999114e-05, "loss": 0.6495, "step": 3933 }, { "epoch": 0.14015711562783903, "grad_norm": 1.1935131889167214, "learning_rate": 1.9370605741961638e-05, "loss": 0.5648, "step": 3934 }, { "epoch": 0.14019274275433316, "grad_norm": 1.2575667486340802, "learning_rate": 1.9370202763147997e-05, "loss": 0.5128, "step": 3935 }, { "epoch": 0.14022836988082726, "grad_norm": 1.180087514362665, "learning_rate": 1.9369799659563556e-05, "loss": 0.4976, "step": 3936 }, { "epoch": 0.14026399700732137, "grad_norm": 1.7217411813865735, "learning_rate": 1.9369396431213686e-05, "loss": 0.6541, "step": 3937 }, { "epoch": 0.1402996241338155, "grad_norm": 1.3107967947756975, "learning_rate": 1.936899307810375e-05, "loss": 0.5714, "step": 3938 }, { "epoch": 0.1403352512603096, "grad_norm": 1.2579034788117784, "learning_rate": 1.9368589600239127e-05, "loss": 0.5443, "step": 3939 }, { "epoch": 0.1403708783868037, "grad_norm": 1.2290179688185348, "learning_rate": 1.9368185997625184e-05, "loss": 0.5929, "step": 3940 }, { "epoch": 0.14040650551329784, "grad_norm": 1.2050360237793376, "learning_rate": 1.9367782270267294e-05, "loss": 0.5663, "step": 3941 }, { "epoch": 0.14044213263979194, "grad_norm": 1.6693346279880987, "learning_rate": 1.936737841817084e-05, "loss": 0.6986, "step": 3942 }, { "epoch": 0.14047775976628604, "grad_norm": 1.6734757211935727, "learning_rate": 1.936697444134119e-05, "loss": 0.7195, "step": 3943 }, { "epoch": 0.14051338689278017, "grad_norm": 1.539829448282742, "learning_rate": 1.936657033978373e-05, "loss": 0.5567, "step": 3944 }, { "epoch": 0.14054901401927428, "grad_norm": 2.6654514124156807, "learning_rate": 1.9366166113503843e-05, "loss": 0.632, "step": 3945 }, { "epoch": 0.14058464114576838, "grad_norm": 1.1249824902089147, "learning_rate": 1.9365761762506903e-05, "loss": 0.5418, "step": 3946 }, { "epoch": 0.1406202682722625, "grad_norm": 1.2230979167568634, "learning_rate": 1.93653572867983e-05, "loss": 0.5363, "step": 3947 }, { "epoch": 0.14065589539875661, "grad_norm": 1.7366342636636385, "learning_rate": 1.936495268638342e-05, "loss": 0.6357, "step": 3948 }, { "epoch": 0.14069152252525072, "grad_norm": 1.2837682228753031, "learning_rate": 1.9364547961267645e-05, "loss": 0.5881, "step": 3949 }, { "epoch": 0.14072714965174485, "grad_norm": 2.1361805194379437, "learning_rate": 1.9364143111456374e-05, "loss": 0.595, "step": 3950 }, { "epoch": 0.14076277677823895, "grad_norm": 1.6803529668835016, "learning_rate": 1.9363738136954988e-05, "loss": 0.6119, "step": 3951 }, { "epoch": 0.14079840390473305, "grad_norm": 1.1945764200056246, "learning_rate": 1.9363333037768886e-05, "loss": 0.544, "step": 3952 }, { "epoch": 0.14083403103122719, "grad_norm": 1.3640215745158, "learning_rate": 1.9362927813903457e-05, "loss": 0.575, "step": 3953 }, { "epoch": 0.1408696581577213, "grad_norm": 2.226651717387772, "learning_rate": 1.9362522465364106e-05, "loss": 0.6248, "step": 3954 }, { "epoch": 0.1409052852842154, "grad_norm": 1.8479322765647255, "learning_rate": 1.9362116992156223e-05, "loss": 0.6777, "step": 3955 }, { "epoch": 0.14094091241070952, "grad_norm": 1.2008850429957056, "learning_rate": 1.9361711394285202e-05, "loss": 0.5214, "step": 3956 }, { "epoch": 0.14097653953720363, "grad_norm": 1.3105232279171617, "learning_rate": 1.9361305671756458e-05, "loss": 0.4806, "step": 3957 }, { "epoch": 0.14101216666369773, "grad_norm": 1.1987963694183432, "learning_rate": 1.936089982457538e-05, "loss": 0.5352, "step": 3958 }, { "epoch": 0.14104779379019186, "grad_norm": 1.2556231277440923, "learning_rate": 1.9360493852747383e-05, "loss": 0.5297, "step": 3959 }, { "epoch": 0.14108342091668596, "grad_norm": 1.714105908249424, "learning_rate": 1.9360087756277862e-05, "loss": 0.5318, "step": 3960 }, { "epoch": 0.14111904804318007, "grad_norm": 1.6374519724813261, "learning_rate": 1.9359681535172233e-05, "loss": 0.6264, "step": 3961 }, { "epoch": 0.1411546751696742, "grad_norm": 1.1856480627036385, "learning_rate": 1.9359275189435904e-05, "loss": 0.5297, "step": 3962 }, { "epoch": 0.1411903022961683, "grad_norm": 1.8410999631746363, "learning_rate": 1.9358868719074283e-05, "loss": 0.6693, "step": 3963 }, { "epoch": 0.1412259294226624, "grad_norm": 1.2423124938864556, "learning_rate": 1.9358462124092784e-05, "loss": 0.495, "step": 3964 }, { "epoch": 0.14126155654915654, "grad_norm": 1.0844702002073652, "learning_rate": 1.935805540449682e-05, "loss": 0.5202, "step": 3965 }, { "epoch": 0.14129718367565064, "grad_norm": 1.7582728174604485, "learning_rate": 1.9357648560291807e-05, "loss": 0.623, "step": 3966 }, { "epoch": 0.14133281080214474, "grad_norm": 1.27849261856099, "learning_rate": 1.9357241591483165e-05, "loss": 0.5981, "step": 3967 }, { "epoch": 0.14136843792863887, "grad_norm": 2.400606211862733, "learning_rate": 1.935683449807631e-05, "loss": 0.6398, "step": 3968 }, { "epoch": 0.14140406505513298, "grad_norm": 1.5306323205912706, "learning_rate": 1.935642728007666e-05, "loss": 0.6058, "step": 3969 }, { "epoch": 0.14143969218162708, "grad_norm": 1.2627658152650776, "learning_rate": 1.9356019937489647e-05, "loss": 0.524, "step": 3970 }, { "epoch": 0.1414753193081212, "grad_norm": 1.421471441847195, "learning_rate": 1.9355612470320683e-05, "loss": 0.6532, "step": 3971 }, { "epoch": 0.14151094643461531, "grad_norm": 1.292428021168911, "learning_rate": 1.9355204878575202e-05, "loss": 0.5511, "step": 3972 }, { "epoch": 0.14154657356110942, "grad_norm": 2.1208918807694404, "learning_rate": 1.9354797162258632e-05, "loss": 0.6545, "step": 3973 }, { "epoch": 0.14158220068760355, "grad_norm": 1.251463004738204, "learning_rate": 1.9354389321376398e-05, "loss": 0.5291, "step": 3974 }, { "epoch": 0.14161782781409765, "grad_norm": 1.2185314634716864, "learning_rate": 1.935398135593393e-05, "loss": 0.5332, "step": 3975 }, { "epoch": 0.14165345494059176, "grad_norm": 1.881205343983438, "learning_rate": 1.9353573265936663e-05, "loss": 0.6864, "step": 3976 }, { "epoch": 0.1416890820670859, "grad_norm": 1.861501884376752, "learning_rate": 1.935316505139003e-05, "loss": 0.5846, "step": 3977 }, { "epoch": 0.14172470919358, "grad_norm": 2.2853751944938545, "learning_rate": 1.9352756712299467e-05, "loss": 0.6142, "step": 3978 }, { "epoch": 0.1417603363200741, "grad_norm": 1.2198003530960928, "learning_rate": 1.935234824867041e-05, "loss": 0.5493, "step": 3979 }, { "epoch": 0.14179596344656822, "grad_norm": 1.1798790719239483, "learning_rate": 1.93519396605083e-05, "loss": 0.5684, "step": 3980 }, { "epoch": 0.14183159057306233, "grad_norm": 1.8659341042671154, "learning_rate": 1.9351530947818577e-05, "loss": 0.6056, "step": 3981 }, { "epoch": 0.14186721769955643, "grad_norm": 1.1738409054820158, "learning_rate": 1.9351122110606682e-05, "loss": 0.4752, "step": 3982 }, { "epoch": 0.14190284482605056, "grad_norm": 1.073735062101992, "learning_rate": 1.9350713148878065e-05, "loss": 0.4736, "step": 3983 }, { "epoch": 0.14193847195254466, "grad_norm": 1.7655697521939866, "learning_rate": 1.9350304062638162e-05, "loss": 0.5829, "step": 3984 }, { "epoch": 0.14197409907903877, "grad_norm": 1.7503563578098345, "learning_rate": 1.9349894851892426e-05, "loss": 0.5981, "step": 3985 }, { "epoch": 0.1420097262055329, "grad_norm": 1.5896138645074525, "learning_rate": 1.9349485516646303e-05, "loss": 0.5996, "step": 3986 }, { "epoch": 0.142045353332027, "grad_norm": 1.219914520270306, "learning_rate": 1.934907605690525e-05, "loss": 0.5526, "step": 3987 }, { "epoch": 0.1420809804585211, "grad_norm": 1.7997878262371114, "learning_rate": 1.9348666472674712e-05, "loss": 0.6591, "step": 3988 }, { "epoch": 0.14211660758501524, "grad_norm": 1.7467591597093917, "learning_rate": 1.9348256763960146e-05, "loss": 0.5105, "step": 3989 }, { "epoch": 0.14215223471150934, "grad_norm": 1.3363616973097818, "learning_rate": 1.9347846930767006e-05, "loss": 0.5014, "step": 3990 }, { "epoch": 0.14218786183800344, "grad_norm": 1.2094909987544977, "learning_rate": 1.9347436973100753e-05, "loss": 0.5677, "step": 3991 }, { "epoch": 0.14222348896449757, "grad_norm": 1.2490882863382637, "learning_rate": 1.934702689096684e-05, "loss": 0.5252, "step": 3992 }, { "epoch": 0.14225911609099168, "grad_norm": 2.0575928040891913, "learning_rate": 1.934661668437073e-05, "loss": 0.6935, "step": 3993 }, { "epoch": 0.14229474321748578, "grad_norm": 1.2538768909702003, "learning_rate": 1.934620635331789e-05, "loss": 0.531, "step": 3994 }, { "epoch": 0.1423303703439799, "grad_norm": 1.6139141820430642, "learning_rate": 1.934579589781378e-05, "loss": 0.6127, "step": 3995 }, { "epoch": 0.14236599747047401, "grad_norm": 1.7136954789254595, "learning_rate": 1.934538531786386e-05, "loss": 0.5588, "step": 3996 }, { "epoch": 0.14240162459696812, "grad_norm": 1.7614134684747773, "learning_rate": 1.9344974613473608e-05, "loss": 0.6971, "step": 3997 }, { "epoch": 0.14243725172346225, "grad_norm": 1.679814326703056, "learning_rate": 1.9344563784648486e-05, "loss": 0.6058, "step": 3998 }, { "epoch": 0.14247287884995635, "grad_norm": 1.208667102230208, "learning_rate": 1.9344152831393963e-05, "loss": 0.4968, "step": 3999 }, { "epoch": 0.14250850597645046, "grad_norm": 2.063133956316682, "learning_rate": 1.934374175371552e-05, "loss": 0.718, "step": 4000 }, { "epoch": 0.1425441331029446, "grad_norm": 1.6839842515938672, "learning_rate": 1.9343330551618616e-05, "loss": 0.6071, "step": 4001 }, { "epoch": 0.1425797602294387, "grad_norm": 1.2092515960139074, "learning_rate": 1.934291922510874e-05, "loss": 0.5251, "step": 4002 }, { "epoch": 0.1426153873559328, "grad_norm": 2.219168276611177, "learning_rate": 1.9342507774191362e-05, "loss": 0.5863, "step": 4003 }, { "epoch": 0.14265101448242692, "grad_norm": 1.8447351619694383, "learning_rate": 1.9342096198871963e-05, "loss": 0.6143, "step": 4004 }, { "epoch": 0.14268664160892103, "grad_norm": 1.9113286156061045, "learning_rate": 1.934168449915603e-05, "loss": 0.6318, "step": 4005 }, { "epoch": 0.14272226873541516, "grad_norm": 1.9661443783562547, "learning_rate": 1.934127267504903e-05, "loss": 0.5838, "step": 4006 }, { "epoch": 0.14275789586190926, "grad_norm": 1.1869574857679592, "learning_rate": 1.934086072655646e-05, "loss": 0.4638, "step": 4007 }, { "epoch": 0.14279352298840337, "grad_norm": 1.8076825540893664, "learning_rate": 1.93404486536838e-05, "loss": 0.6741, "step": 4008 }, { "epoch": 0.1428291501148975, "grad_norm": 1.2201819118047637, "learning_rate": 1.9340036456436534e-05, "loss": 0.5052, "step": 4009 }, { "epoch": 0.1428647772413916, "grad_norm": 1.8502688851249862, "learning_rate": 1.933962413482016e-05, "loss": 0.5683, "step": 4010 }, { "epoch": 0.1429004043678857, "grad_norm": 1.623473360596649, "learning_rate": 1.933921168884016e-05, "loss": 0.5904, "step": 4011 }, { "epoch": 0.14293603149437983, "grad_norm": 1.3217680774104053, "learning_rate": 1.9338799118502026e-05, "loss": 0.5178, "step": 4012 }, { "epoch": 0.14297165862087394, "grad_norm": 1.8081332624242146, "learning_rate": 1.9338386423811254e-05, "loss": 0.5341, "step": 4013 }, { "epoch": 0.14300728574736804, "grad_norm": 1.3618230909145308, "learning_rate": 1.9337973604773345e-05, "loss": 0.5518, "step": 4014 }, { "epoch": 0.14304291287386217, "grad_norm": 1.313660288733632, "learning_rate": 1.9337560661393788e-05, "loss": 0.5301, "step": 4015 }, { "epoch": 0.14307854000035627, "grad_norm": 1.8032706495633222, "learning_rate": 1.9337147593678085e-05, "loss": 0.6381, "step": 4016 }, { "epoch": 0.14311416712685038, "grad_norm": 2.050841580762259, "learning_rate": 1.9336734401631735e-05, "loss": 0.6546, "step": 4017 }, { "epoch": 0.1431497942533445, "grad_norm": 1.2327146496783445, "learning_rate": 1.9336321085260236e-05, "loss": 0.5755, "step": 4018 }, { "epoch": 0.1431854213798386, "grad_norm": 1.2233735406578563, "learning_rate": 1.93359076445691e-05, "loss": 0.5221, "step": 4019 }, { "epoch": 0.14322104850633272, "grad_norm": 1.8357690680756613, "learning_rate": 1.933549407956383e-05, "loss": 0.6401, "step": 4020 }, { "epoch": 0.14325667563282685, "grad_norm": 1.6332685084782155, "learning_rate": 1.9335080390249928e-05, "loss": 0.5821, "step": 4021 }, { "epoch": 0.14329230275932095, "grad_norm": 1.8233795681124838, "learning_rate": 1.933466657663291e-05, "loss": 0.6476, "step": 4022 }, { "epoch": 0.14332792988581505, "grad_norm": 1.637454605633924, "learning_rate": 1.933425263871828e-05, "loss": 0.6831, "step": 4023 }, { "epoch": 0.14336355701230918, "grad_norm": 1.5800219665481803, "learning_rate": 1.933383857651155e-05, "loss": 0.5836, "step": 4024 }, { "epoch": 0.1433991841388033, "grad_norm": 2.4376246485632955, "learning_rate": 1.9333424390018237e-05, "loss": 0.6519, "step": 4025 }, { "epoch": 0.1434348112652974, "grad_norm": 1.834542223177959, "learning_rate": 1.9333010079243856e-05, "loss": 0.5745, "step": 4026 }, { "epoch": 0.14347043839179152, "grad_norm": 2.047350524530137, "learning_rate": 1.9332595644193918e-05, "loss": 0.6033, "step": 4027 }, { "epoch": 0.14350606551828562, "grad_norm": 1.8977182317385526, "learning_rate": 1.9332181084873947e-05, "loss": 0.6295, "step": 4028 }, { "epoch": 0.14354169264477973, "grad_norm": 1.909889442260063, "learning_rate": 1.9331766401289466e-05, "loss": 0.6483, "step": 4029 }, { "epoch": 0.14357731977127386, "grad_norm": 1.2039113386542204, "learning_rate": 1.933135159344599e-05, "loss": 0.4898, "step": 4030 }, { "epoch": 0.14361294689776796, "grad_norm": 1.485131053185686, "learning_rate": 1.9330936661349044e-05, "loss": 0.5247, "step": 4031 }, { "epoch": 0.14364857402426207, "grad_norm": 3.006004394210739, "learning_rate": 1.9330521605004157e-05, "loss": 0.5813, "step": 4032 }, { "epoch": 0.1436842011507562, "grad_norm": 1.1926661223903514, "learning_rate": 1.9330106424416852e-05, "loss": 0.5318, "step": 4033 }, { "epoch": 0.1437198282772503, "grad_norm": 1.32053767574374, "learning_rate": 1.932969111959266e-05, "loss": 0.5215, "step": 4034 }, { "epoch": 0.1437554554037444, "grad_norm": 1.6724102782619183, "learning_rate": 1.9329275690537108e-05, "loss": 0.5646, "step": 4035 }, { "epoch": 0.14379108253023853, "grad_norm": 1.3324813719892599, "learning_rate": 1.932886013725573e-05, "loss": 0.546, "step": 4036 }, { "epoch": 0.14382670965673264, "grad_norm": 1.9197099241401383, "learning_rate": 1.9328444459754057e-05, "loss": 0.5696, "step": 4037 }, { "epoch": 0.14386233678322674, "grad_norm": 2.125604599015873, "learning_rate": 1.932802865803763e-05, "loss": 0.6716, "step": 4038 }, { "epoch": 0.14389796390972087, "grad_norm": 1.3452090592053776, "learning_rate": 1.9327612732111976e-05, "loss": 0.5586, "step": 4039 }, { "epoch": 0.14393359103621498, "grad_norm": 1.8277881936384124, "learning_rate": 1.9327196681982644e-05, "loss": 0.6931, "step": 4040 }, { "epoch": 0.14396921816270908, "grad_norm": 1.3561182513553198, "learning_rate": 1.9326780507655165e-05, "loss": 0.4993, "step": 4041 }, { "epoch": 0.1440048452892032, "grad_norm": 1.7312193563153575, "learning_rate": 1.9326364209135088e-05, "loss": 0.6813, "step": 4042 }, { "epoch": 0.1440404724156973, "grad_norm": 1.2720202387394226, "learning_rate": 1.9325947786427953e-05, "loss": 0.5704, "step": 4043 }, { "epoch": 0.14407609954219142, "grad_norm": 1.1180611804229157, "learning_rate": 1.9325531239539303e-05, "loss": 0.5133, "step": 4044 }, { "epoch": 0.14411172666868555, "grad_norm": 2.0918591364361268, "learning_rate": 1.9325114568474684e-05, "loss": 0.7008, "step": 4045 }, { "epoch": 0.14414735379517965, "grad_norm": 1.8657576763074948, "learning_rate": 1.932469777323965e-05, "loss": 0.5588, "step": 4046 }, { "epoch": 0.14418298092167375, "grad_norm": 1.7358517254039485, "learning_rate": 1.9324280853839744e-05, "loss": 0.6987, "step": 4047 }, { "epoch": 0.14421860804816788, "grad_norm": 1.2311997132049928, "learning_rate": 1.9323863810280525e-05, "loss": 0.5045, "step": 4048 }, { "epoch": 0.144254235174662, "grad_norm": 1.097873001017162, "learning_rate": 1.9323446642567543e-05, "loss": 0.5198, "step": 4049 }, { "epoch": 0.1442898623011561, "grad_norm": 1.8474074589429939, "learning_rate": 1.932302935070635e-05, "loss": 0.6318, "step": 4050 }, { "epoch": 0.14432548942765022, "grad_norm": 1.45982171799627, "learning_rate": 1.93226119347025e-05, "loss": 0.6822, "step": 4051 }, { "epoch": 0.14436111655414433, "grad_norm": 1.9933987217957634, "learning_rate": 1.9322194394561563e-05, "loss": 0.6211, "step": 4052 }, { "epoch": 0.14439674368063843, "grad_norm": 1.301702850812268, "learning_rate": 1.932177673028909e-05, "loss": 0.5179, "step": 4053 }, { "epoch": 0.14443237080713256, "grad_norm": 1.963068489903661, "learning_rate": 1.9321358941890645e-05, "loss": 0.6315, "step": 4054 }, { "epoch": 0.14446799793362666, "grad_norm": 1.4896093134149873, "learning_rate": 1.932094102937179e-05, "loss": 0.5387, "step": 4055 }, { "epoch": 0.14450362506012077, "grad_norm": 1.0466555436789489, "learning_rate": 1.932052299273809e-05, "loss": 0.5178, "step": 4056 }, { "epoch": 0.1445392521866149, "grad_norm": 1.7146318623230306, "learning_rate": 1.9320104831995113e-05, "loss": 0.6685, "step": 4057 }, { "epoch": 0.144574879313109, "grad_norm": 1.24755349088785, "learning_rate": 1.9319686547148422e-05, "loss": 0.5216, "step": 4058 }, { "epoch": 0.1446105064396031, "grad_norm": 2.08836102085519, "learning_rate": 1.9319268138203595e-05, "loss": 0.5712, "step": 4059 }, { "epoch": 0.14464613356609723, "grad_norm": 1.3676464246696558, "learning_rate": 1.9318849605166195e-05, "loss": 0.5214, "step": 4060 }, { "epoch": 0.14468176069259134, "grad_norm": 1.2193346921036254, "learning_rate": 1.9318430948041802e-05, "loss": 0.5329, "step": 4061 }, { "epoch": 0.14471738781908544, "grad_norm": 1.2774078358010685, "learning_rate": 1.9318012166835987e-05, "loss": 0.5682, "step": 4062 }, { "epoch": 0.14475301494557957, "grad_norm": 1.6091106395909243, "learning_rate": 1.9317593261554325e-05, "loss": 0.5485, "step": 4063 }, { "epoch": 0.14478864207207368, "grad_norm": 1.6862055036259855, "learning_rate": 1.9317174232202398e-05, "loss": 0.5901, "step": 4064 }, { "epoch": 0.14482426919856778, "grad_norm": 1.6661384144296527, "learning_rate": 1.931675507878578e-05, "loss": 0.5949, "step": 4065 }, { "epoch": 0.1448598963250619, "grad_norm": 1.2132570395117535, "learning_rate": 1.9316335801310064e-05, "loss": 0.5326, "step": 4066 }, { "epoch": 0.144895523451556, "grad_norm": 1.7864652818562623, "learning_rate": 1.931591639978082e-05, "loss": 0.6819, "step": 4067 }, { "epoch": 0.14493115057805012, "grad_norm": 1.2368540536415422, "learning_rate": 1.9315496874203637e-05, "loss": 0.498, "step": 4068 }, { "epoch": 0.14496677770454425, "grad_norm": 1.9562990975191807, "learning_rate": 1.9315077224584106e-05, "loss": 0.6447, "step": 4069 }, { "epoch": 0.14500240483103835, "grad_norm": 1.3097711825668115, "learning_rate": 1.9314657450927807e-05, "loss": 0.5671, "step": 4070 }, { "epoch": 0.14503803195753245, "grad_norm": 1.7526108311571555, "learning_rate": 1.9314237553240335e-05, "loss": 0.6078, "step": 4071 }, { "epoch": 0.14507365908402658, "grad_norm": 1.7902572636224867, "learning_rate": 1.931381753152728e-05, "loss": 0.6412, "step": 4072 }, { "epoch": 0.1451092862105207, "grad_norm": 1.6744397365390649, "learning_rate": 1.9313397385794236e-05, "loss": 0.6724, "step": 4073 }, { "epoch": 0.1451449133370148, "grad_norm": 6.060502298804669, "learning_rate": 1.9312977116046793e-05, "loss": 0.6878, "step": 4074 }, { "epoch": 0.14518054046350892, "grad_norm": 1.8458290921227587, "learning_rate": 1.9312556722290553e-05, "loss": 0.6349, "step": 4075 }, { "epoch": 0.14521616759000303, "grad_norm": 1.709279666777108, "learning_rate": 1.931213620453111e-05, "loss": 0.5597, "step": 4076 }, { "epoch": 0.14525179471649713, "grad_norm": 1.2021980535141552, "learning_rate": 1.931171556277406e-05, "loss": 0.4889, "step": 4077 }, { "epoch": 0.14528742184299126, "grad_norm": 1.8612385529605104, "learning_rate": 1.9311294797025016e-05, "loss": 0.5842, "step": 4078 }, { "epoch": 0.14532304896948536, "grad_norm": 1.1982103143520613, "learning_rate": 1.931087390728957e-05, "loss": 0.5014, "step": 4079 }, { "epoch": 0.14535867609597947, "grad_norm": 2.014757004782357, "learning_rate": 1.931045289357333e-05, "loss": 0.5632, "step": 4080 }, { "epoch": 0.1453943032224736, "grad_norm": 1.7766676381706898, "learning_rate": 1.9310031755881903e-05, "loss": 0.6379, "step": 4081 }, { "epoch": 0.1454299303489677, "grad_norm": 1.8104393106788175, "learning_rate": 1.9309610494220895e-05, "loss": 0.5423, "step": 4082 }, { "epoch": 0.1454655574754618, "grad_norm": 1.1762953459225802, "learning_rate": 1.930918910859592e-05, "loss": 0.4861, "step": 4083 }, { "epoch": 0.14550118460195594, "grad_norm": 1.2905655597455208, "learning_rate": 1.930876759901258e-05, "loss": 0.5334, "step": 4084 }, { "epoch": 0.14553681172845004, "grad_norm": 1.2729498825133259, "learning_rate": 1.9308345965476498e-05, "loss": 0.5595, "step": 4085 }, { "epoch": 0.14557243885494414, "grad_norm": 1.3661911367917756, "learning_rate": 1.9307924207993278e-05, "loss": 0.5315, "step": 4086 }, { "epoch": 0.14560806598143827, "grad_norm": 1.2270025494670562, "learning_rate": 1.9307502326568545e-05, "loss": 0.5319, "step": 4087 }, { "epoch": 0.14564369310793238, "grad_norm": 2.2023239004952666, "learning_rate": 1.9307080321207913e-05, "loss": 0.562, "step": 4088 }, { "epoch": 0.14567932023442648, "grad_norm": 1.0896443301634127, "learning_rate": 1.9306658191917e-05, "loss": 0.4907, "step": 4089 }, { "epoch": 0.1457149473609206, "grad_norm": 2.123171969264505, "learning_rate": 1.930623593870143e-05, "loss": 0.6644, "step": 4090 }, { "epoch": 0.1457505744874147, "grad_norm": 1.341037302625986, "learning_rate": 1.930581356156682e-05, "loss": 0.507, "step": 4091 }, { "epoch": 0.14578620161390882, "grad_norm": 1.7108504501537207, "learning_rate": 1.93053910605188e-05, "loss": 0.606, "step": 4092 }, { "epoch": 0.14582182874040295, "grad_norm": 1.1995701700112875, "learning_rate": 1.9304968435562995e-05, "loss": 0.5129, "step": 4093 }, { "epoch": 0.14585745586689705, "grad_norm": 1.5472225710053147, "learning_rate": 1.930454568670503e-05, "loss": 0.625, "step": 4094 }, { "epoch": 0.14589308299339115, "grad_norm": 1.1075933694002902, "learning_rate": 1.9304122813950538e-05, "loss": 0.5061, "step": 4095 }, { "epoch": 0.14592871011988529, "grad_norm": 1.2640042415706427, "learning_rate": 1.9303699817305144e-05, "loss": 0.5613, "step": 4096 }, { "epoch": 0.1459643372463794, "grad_norm": 1.2463926019635785, "learning_rate": 1.9303276696774486e-05, "loss": 0.5202, "step": 4097 }, { "epoch": 0.14599996437287352, "grad_norm": 1.4171778917982851, "learning_rate": 1.9302853452364196e-05, "loss": 0.5631, "step": 4098 }, { "epoch": 0.14603559149936762, "grad_norm": 1.0952836091298153, "learning_rate": 1.930243008407991e-05, "loss": 0.5267, "step": 4099 }, { "epoch": 0.14607121862586173, "grad_norm": 1.1053773046742095, "learning_rate": 1.9302006591927266e-05, "loss": 0.5191, "step": 4100 }, { "epoch": 0.14610684575235586, "grad_norm": 1.5350863988532615, "learning_rate": 1.9301582975911903e-05, "loss": 0.6281, "step": 4101 }, { "epoch": 0.14614247287884996, "grad_norm": 1.1830864505923568, "learning_rate": 1.9301159236039458e-05, "loss": 0.485, "step": 4102 }, { "epoch": 0.14617810000534406, "grad_norm": 1.6726652166251414, "learning_rate": 1.930073537231558e-05, "loss": 0.4564, "step": 4103 }, { "epoch": 0.1462137271318382, "grad_norm": 1.3432305165184868, "learning_rate": 1.9300311384745908e-05, "loss": 0.5489, "step": 4104 }, { "epoch": 0.1462493542583323, "grad_norm": 1.2473797622732896, "learning_rate": 1.9299887273336088e-05, "loss": 0.5099, "step": 4105 }, { "epoch": 0.1462849813848264, "grad_norm": 1.622393225204376, "learning_rate": 1.9299463038091772e-05, "loss": 0.5335, "step": 4106 }, { "epoch": 0.14632060851132053, "grad_norm": 2.0095163968661494, "learning_rate": 1.9299038679018606e-05, "loss": 0.5638, "step": 4107 }, { "epoch": 0.14635623563781464, "grad_norm": 1.519160428019722, "learning_rate": 1.929861419612224e-05, "loss": 0.562, "step": 4108 }, { "epoch": 0.14639186276430874, "grad_norm": 1.231112531251128, "learning_rate": 1.929818958940832e-05, "loss": 0.4788, "step": 4109 }, { "epoch": 0.14642748989080287, "grad_norm": 1.223366461370316, "learning_rate": 1.9297764858882516e-05, "loss": 0.5489, "step": 4110 }, { "epoch": 0.14646311701729697, "grad_norm": 1.2931703833564558, "learning_rate": 1.929734000455047e-05, "loss": 0.5802, "step": 4111 }, { "epoch": 0.14649874414379108, "grad_norm": 1.139973732334664, "learning_rate": 1.9296915026417846e-05, "loss": 0.5141, "step": 4112 }, { "epoch": 0.1465343712702852, "grad_norm": 1.1941357055461161, "learning_rate": 1.9296489924490296e-05, "loss": 0.5126, "step": 4113 }, { "epoch": 0.1465699983967793, "grad_norm": 1.7170384948820465, "learning_rate": 1.9296064698773485e-05, "loss": 0.685, "step": 4114 }, { "epoch": 0.14660562552327341, "grad_norm": 1.9906720473245356, "learning_rate": 1.9295639349273076e-05, "loss": 0.6599, "step": 4115 }, { "epoch": 0.14664125264976755, "grad_norm": 1.241896773371021, "learning_rate": 1.9295213875994735e-05, "loss": 0.5498, "step": 4116 }, { "epoch": 0.14667687977626165, "grad_norm": 1.7334768015394584, "learning_rate": 1.929478827894412e-05, "loss": 0.58, "step": 4117 }, { "epoch": 0.14671250690275575, "grad_norm": 1.1193517230218435, "learning_rate": 1.9294362558126904e-05, "loss": 0.5407, "step": 4118 }, { "epoch": 0.14674813402924988, "grad_norm": 2.2236470745341594, "learning_rate": 1.9293936713548754e-05, "loss": 0.6092, "step": 4119 }, { "epoch": 0.14678376115574399, "grad_norm": 1.6228177371552075, "learning_rate": 1.929351074521534e-05, "loss": 0.6463, "step": 4120 }, { "epoch": 0.1468193882822381, "grad_norm": 1.2298102779488858, "learning_rate": 1.9293084653132336e-05, "loss": 0.5233, "step": 4121 }, { "epoch": 0.14685501540873222, "grad_norm": 1.201757371087502, "learning_rate": 1.9292658437305415e-05, "loss": 0.5023, "step": 4122 }, { "epoch": 0.14689064253522632, "grad_norm": 1.1927345462426149, "learning_rate": 1.929223209774025e-05, "loss": 0.4938, "step": 4123 }, { "epoch": 0.14692626966172043, "grad_norm": 1.6655289845150447, "learning_rate": 1.929180563444252e-05, "loss": 0.6346, "step": 4124 }, { "epoch": 0.14696189678821456, "grad_norm": 1.562165530231512, "learning_rate": 1.9291379047417903e-05, "loss": 0.5926, "step": 4125 }, { "epoch": 0.14699752391470866, "grad_norm": 1.8960639399984185, "learning_rate": 1.929095233667208e-05, "loss": 0.6205, "step": 4126 }, { "epoch": 0.14703315104120276, "grad_norm": 1.6247488421776917, "learning_rate": 1.9290525502210733e-05, "loss": 0.6094, "step": 4127 }, { "epoch": 0.1470687781676969, "grad_norm": 1.2019129304724663, "learning_rate": 1.9290098544039546e-05, "loss": 0.538, "step": 4128 }, { "epoch": 0.147104405294191, "grad_norm": 1.4668701560751791, "learning_rate": 1.9289671462164202e-05, "loss": 0.5906, "step": 4129 }, { "epoch": 0.1471400324206851, "grad_norm": 1.584521242823008, "learning_rate": 1.928924425659039e-05, "loss": 0.6255, "step": 4130 }, { "epoch": 0.14717565954717923, "grad_norm": 2.0118670410890838, "learning_rate": 1.9288816927323798e-05, "loss": 0.6087, "step": 4131 }, { "epoch": 0.14721128667367334, "grad_norm": 2.0470448439766074, "learning_rate": 1.9288389474370116e-05, "loss": 0.6986, "step": 4132 }, { "epoch": 0.14724691380016744, "grad_norm": 1.6170198045770456, "learning_rate": 1.928796189773504e-05, "loss": 0.6262, "step": 4133 }, { "epoch": 0.14728254092666157, "grad_norm": 2.0781105936312088, "learning_rate": 1.9287534197424255e-05, "loss": 0.5622, "step": 4134 }, { "epoch": 0.14731816805315567, "grad_norm": 2.0220680784925227, "learning_rate": 1.928710637344346e-05, "loss": 0.6573, "step": 4135 }, { "epoch": 0.14735379517964978, "grad_norm": 1.2468504262139632, "learning_rate": 1.9286678425798353e-05, "loss": 0.5171, "step": 4136 }, { "epoch": 0.1473894223061439, "grad_norm": 1.6376301870353285, "learning_rate": 1.9286250354494634e-05, "loss": 0.676, "step": 4137 }, { "epoch": 0.147425049432638, "grad_norm": 1.543885650937171, "learning_rate": 1.9285822159537996e-05, "loss": 0.6096, "step": 4138 }, { "epoch": 0.14746067655913211, "grad_norm": 2.0705334502060135, "learning_rate": 1.9285393840934152e-05, "loss": 0.5965, "step": 4139 }, { "epoch": 0.14749630368562625, "grad_norm": 1.8145972403513029, "learning_rate": 1.9284965398688793e-05, "loss": 0.5381, "step": 4140 }, { "epoch": 0.14753193081212035, "grad_norm": 1.8801860183726045, "learning_rate": 1.9284536832807635e-05, "loss": 0.6114, "step": 4141 }, { "epoch": 0.14756755793861445, "grad_norm": 1.6897117269081705, "learning_rate": 1.9284108143296378e-05, "loss": 0.5932, "step": 4142 }, { "epoch": 0.14760318506510858, "grad_norm": 1.5962423901040488, "learning_rate": 1.9283679330160726e-05, "loss": 0.6292, "step": 4143 }, { "epoch": 0.1476388121916027, "grad_norm": 1.2083566372575636, "learning_rate": 1.92832503934064e-05, "loss": 0.5012, "step": 4144 }, { "epoch": 0.1476744393180968, "grad_norm": 1.7790462948476369, "learning_rate": 1.9282821333039106e-05, "loss": 0.6128, "step": 4145 }, { "epoch": 0.14771006644459092, "grad_norm": 1.600869207789703, "learning_rate": 1.928239214906456e-05, "loss": 0.5373, "step": 4146 }, { "epoch": 0.14774569357108502, "grad_norm": 1.3723489429748517, "learning_rate": 1.928196284148847e-05, "loss": 0.4998, "step": 4147 }, { "epoch": 0.14778132069757913, "grad_norm": 1.1802999209182217, "learning_rate": 1.9281533410316558e-05, "loss": 0.5073, "step": 4148 }, { "epoch": 0.14781694782407326, "grad_norm": 1.1129707253137617, "learning_rate": 1.9281103855554545e-05, "loss": 0.5657, "step": 4149 }, { "epoch": 0.14785257495056736, "grad_norm": 1.8329946721533634, "learning_rate": 1.9280674177208145e-05, "loss": 0.597, "step": 4150 }, { "epoch": 0.14788820207706146, "grad_norm": 1.815954122296644, "learning_rate": 1.9280244375283083e-05, "loss": 0.5568, "step": 4151 }, { "epoch": 0.1479238292035556, "grad_norm": 1.3140553493279303, "learning_rate": 1.9279814449785076e-05, "loss": 0.5771, "step": 4152 }, { "epoch": 0.1479594563300497, "grad_norm": 1.6054608010146008, "learning_rate": 1.927938440071986e-05, "loss": 0.5397, "step": 4153 }, { "epoch": 0.1479950834565438, "grad_norm": 1.2828082901192792, "learning_rate": 1.927895422809315e-05, "loss": 0.5268, "step": 4154 }, { "epoch": 0.14803071058303793, "grad_norm": 1.1956908269640565, "learning_rate": 1.9278523931910678e-05, "loss": 0.5295, "step": 4155 }, { "epoch": 0.14806633770953204, "grad_norm": 1.6393028190021852, "learning_rate": 1.9278093512178175e-05, "loss": 0.5998, "step": 4156 }, { "epoch": 0.14810196483602614, "grad_norm": 1.6709694896219045, "learning_rate": 1.9277662968901377e-05, "loss": 0.6153, "step": 4157 }, { "epoch": 0.14813759196252027, "grad_norm": 1.2282536971112499, "learning_rate": 1.9277232302086006e-05, "loss": 0.5242, "step": 4158 }, { "epoch": 0.14817321908901437, "grad_norm": 1.1424148991093568, "learning_rate": 1.9276801511737805e-05, "loss": 0.583, "step": 4159 }, { "epoch": 0.14820884621550848, "grad_norm": 1.1126131618272923, "learning_rate": 1.9276370597862507e-05, "loss": 0.5306, "step": 4160 }, { "epoch": 0.1482444733420026, "grad_norm": 1.776847371711823, "learning_rate": 1.927593956046585e-05, "loss": 0.7043, "step": 4161 }, { "epoch": 0.1482801004684967, "grad_norm": 1.1413544430342215, "learning_rate": 1.9275508399553577e-05, "loss": 0.5015, "step": 4162 }, { "epoch": 0.14831572759499081, "grad_norm": 1.2677128507854, "learning_rate": 1.9275077115131424e-05, "loss": 0.4941, "step": 4163 }, { "epoch": 0.14835135472148495, "grad_norm": 1.1215984571892135, "learning_rate": 1.9274645707205138e-05, "loss": 0.549, "step": 4164 }, { "epoch": 0.14838698184797905, "grad_norm": 1.3033462783030956, "learning_rate": 1.927421417578046e-05, "loss": 0.5624, "step": 4165 }, { "epoch": 0.14842260897447315, "grad_norm": 1.3366822618360341, "learning_rate": 1.9273782520863137e-05, "loss": 0.5442, "step": 4166 }, { "epoch": 0.14845823610096728, "grad_norm": 1.0902483261119142, "learning_rate": 1.927335074245892e-05, "loss": 0.5359, "step": 4167 }, { "epoch": 0.1484938632274614, "grad_norm": 1.9982665145290834, "learning_rate": 1.9272918840573558e-05, "loss": 0.5945, "step": 4168 }, { "epoch": 0.1485294903539555, "grad_norm": 1.3223815227747295, "learning_rate": 1.9272486815212797e-05, "loss": 0.5554, "step": 4169 }, { "epoch": 0.14856511748044962, "grad_norm": 1.1698015314338785, "learning_rate": 1.9272054666382397e-05, "loss": 0.5392, "step": 4170 }, { "epoch": 0.14860074460694372, "grad_norm": 1.2183239900930005, "learning_rate": 1.9271622394088103e-05, "loss": 0.5386, "step": 4171 }, { "epoch": 0.14863637173343783, "grad_norm": 1.91963619508235, "learning_rate": 1.9271189998335677e-05, "loss": 0.6497, "step": 4172 }, { "epoch": 0.14867199885993196, "grad_norm": 1.822923126107137, "learning_rate": 1.927075747913088e-05, "loss": 0.5947, "step": 4173 }, { "epoch": 0.14870762598642606, "grad_norm": 1.1254529293769642, "learning_rate": 1.9270324836479465e-05, "loss": 0.5085, "step": 4174 }, { "epoch": 0.14874325311292017, "grad_norm": 1.2383753505342512, "learning_rate": 1.9269892070387196e-05, "loss": 0.5285, "step": 4175 }, { "epoch": 0.1487788802394143, "grad_norm": 1.1755408326087375, "learning_rate": 1.9269459180859834e-05, "loss": 0.4724, "step": 4176 }, { "epoch": 0.1488145073659084, "grad_norm": 1.127155112080088, "learning_rate": 1.9269026167903142e-05, "loss": 0.495, "step": 4177 }, { "epoch": 0.1488501344924025, "grad_norm": 1.237242075576531, "learning_rate": 1.926859303152289e-05, "loss": 0.4757, "step": 4178 }, { "epoch": 0.14888576161889663, "grad_norm": 1.5898602785919775, "learning_rate": 1.9268159771724845e-05, "loss": 0.6328, "step": 4179 }, { "epoch": 0.14892138874539074, "grad_norm": 1.4687749965418149, "learning_rate": 1.9267726388514775e-05, "loss": 0.5398, "step": 4180 }, { "epoch": 0.14895701587188484, "grad_norm": 1.1090348173593596, "learning_rate": 1.926729288189845e-05, "loss": 0.4802, "step": 4181 }, { "epoch": 0.14899264299837897, "grad_norm": 1.589629623539173, "learning_rate": 1.9266859251881637e-05, "loss": 0.6033, "step": 4182 }, { "epoch": 0.14902827012487307, "grad_norm": 2.0370821931246, "learning_rate": 1.9266425498470122e-05, "loss": 0.6312, "step": 4183 }, { "epoch": 0.14906389725136718, "grad_norm": 1.5994378120882384, "learning_rate": 1.926599162166967e-05, "loss": 0.5962, "step": 4184 }, { "epoch": 0.1490995243778613, "grad_norm": 1.3264453745085143, "learning_rate": 1.9265557621486067e-05, "loss": 0.5204, "step": 4185 }, { "epoch": 0.1491351515043554, "grad_norm": 1.2500803433341745, "learning_rate": 1.9265123497925088e-05, "loss": 0.4989, "step": 4186 }, { "epoch": 0.14917077863084952, "grad_norm": 1.1889346088575574, "learning_rate": 1.9264689250992514e-05, "loss": 0.5176, "step": 4187 }, { "epoch": 0.14920640575734365, "grad_norm": 1.7709203527264148, "learning_rate": 1.9264254880694123e-05, "loss": 0.6161, "step": 4188 }, { "epoch": 0.14924203288383775, "grad_norm": 1.2030376626177886, "learning_rate": 1.9263820387035705e-05, "loss": 0.5461, "step": 4189 }, { "epoch": 0.14927766001033185, "grad_norm": 1.3354081510861968, "learning_rate": 1.9263385770023048e-05, "loss": 0.5702, "step": 4190 }, { "epoch": 0.14931328713682598, "grad_norm": 1.8434217343883288, "learning_rate": 1.9262951029661926e-05, "loss": 0.6909, "step": 4191 }, { "epoch": 0.1493489142633201, "grad_norm": 1.7495035877824041, "learning_rate": 1.9262516165958143e-05, "loss": 0.6507, "step": 4192 }, { "epoch": 0.14938454138981422, "grad_norm": 1.5977842152269803, "learning_rate": 1.9262081178917483e-05, "loss": 0.6312, "step": 4193 }, { "epoch": 0.14942016851630832, "grad_norm": 2.294517695295049, "learning_rate": 1.9261646068545736e-05, "loss": 0.6979, "step": 4194 }, { "epoch": 0.14945579564280242, "grad_norm": 1.3603326109360305, "learning_rate": 1.92612108348487e-05, "loss": 0.5284, "step": 4195 }, { "epoch": 0.14949142276929656, "grad_norm": 1.1421917787869098, "learning_rate": 1.9260775477832165e-05, "loss": 0.4933, "step": 4196 }, { "epoch": 0.14952704989579066, "grad_norm": 1.9495359746128165, "learning_rate": 1.9260339997501936e-05, "loss": 0.6015, "step": 4197 }, { "epoch": 0.14956267702228476, "grad_norm": 1.1653311555094468, "learning_rate": 1.9259904393863804e-05, "loss": 0.5302, "step": 4198 }, { "epoch": 0.1495983041487789, "grad_norm": 1.1549220346903666, "learning_rate": 1.9259468666923575e-05, "loss": 0.5099, "step": 4199 }, { "epoch": 0.149633931275273, "grad_norm": 1.2713588738143868, "learning_rate": 1.9259032816687047e-05, "loss": 0.5456, "step": 4200 }, { "epoch": 0.1496695584017671, "grad_norm": 1.2862427292821197, "learning_rate": 1.9258596843160026e-05, "loss": 0.5215, "step": 4201 }, { "epoch": 0.14970518552826123, "grad_norm": 1.6454199077200944, "learning_rate": 1.925816074634832e-05, "loss": 0.6401, "step": 4202 }, { "epoch": 0.14974081265475533, "grad_norm": 1.626875754084191, "learning_rate": 1.9257724526257727e-05, "loss": 0.6709, "step": 4203 }, { "epoch": 0.14977643978124944, "grad_norm": 1.776505009591413, "learning_rate": 1.9257288182894064e-05, "loss": 0.5876, "step": 4204 }, { "epoch": 0.14981206690774357, "grad_norm": 1.7744555559559279, "learning_rate": 1.925685171626314e-05, "loss": 0.6536, "step": 4205 }, { "epoch": 0.14984769403423767, "grad_norm": 1.4952796457191244, "learning_rate": 1.925641512637076e-05, "loss": 0.6063, "step": 4206 }, { "epoch": 0.14988332116073178, "grad_norm": 1.337569858939797, "learning_rate": 1.925597841322275e-05, "loss": 0.6043, "step": 4207 }, { "epoch": 0.1499189482872259, "grad_norm": 1.2961648833995723, "learning_rate": 1.925554157682491e-05, "loss": 0.5529, "step": 4208 }, { "epoch": 0.14995457541372, "grad_norm": 1.7406881449039149, "learning_rate": 1.9255104617183068e-05, "loss": 0.6437, "step": 4209 }, { "epoch": 0.1499902025402141, "grad_norm": 1.1594760162776356, "learning_rate": 1.9254667534303044e-05, "loss": 0.5676, "step": 4210 }, { "epoch": 0.15002582966670824, "grad_norm": 1.8148573274908533, "learning_rate": 1.925423032819065e-05, "loss": 0.5933, "step": 4211 }, { "epoch": 0.15006145679320235, "grad_norm": 1.91738955403441, "learning_rate": 1.9253792998851707e-05, "loss": 0.6131, "step": 4212 }, { "epoch": 0.15009708391969645, "grad_norm": 3.05704877296013, "learning_rate": 1.9253355546292044e-05, "loss": 0.7508, "step": 4213 }, { "epoch": 0.15013271104619058, "grad_norm": 1.8084537369763949, "learning_rate": 1.9252917970517486e-05, "loss": 0.6072, "step": 4214 }, { "epoch": 0.15016833817268468, "grad_norm": 1.6753620375539076, "learning_rate": 1.9252480271533858e-05, "loss": 0.711, "step": 4215 }, { "epoch": 0.1502039652991788, "grad_norm": 1.2365352615949723, "learning_rate": 1.9252042449346987e-05, "loss": 0.5456, "step": 4216 }, { "epoch": 0.15023959242567292, "grad_norm": 1.2855092209293975, "learning_rate": 1.92516045039627e-05, "loss": 0.5593, "step": 4217 }, { "epoch": 0.15027521955216702, "grad_norm": 1.6180754061092895, "learning_rate": 1.9251166435386837e-05, "loss": 0.6, "step": 4218 }, { "epoch": 0.15031084667866113, "grad_norm": 1.1457120176018327, "learning_rate": 1.925072824362523e-05, "loss": 0.5181, "step": 4219 }, { "epoch": 0.15034647380515526, "grad_norm": 1.129778672240643, "learning_rate": 1.9250289928683706e-05, "loss": 0.4651, "step": 4220 }, { "epoch": 0.15038210093164936, "grad_norm": 1.2682051625966313, "learning_rate": 1.9249851490568106e-05, "loss": 0.5436, "step": 4221 }, { "epoch": 0.15041772805814346, "grad_norm": 1.7487889428273498, "learning_rate": 1.924941292928427e-05, "loss": 0.6435, "step": 4222 }, { "epoch": 0.1504533551846376, "grad_norm": 2.0492580775302036, "learning_rate": 1.9248974244838035e-05, "loss": 0.6914, "step": 4223 }, { "epoch": 0.1504889823111317, "grad_norm": 1.1943661589656716, "learning_rate": 1.9248535437235242e-05, "loss": 0.5223, "step": 4224 }, { "epoch": 0.1505246094376258, "grad_norm": 1.4620295955188305, "learning_rate": 1.9248096506481737e-05, "loss": 0.5949, "step": 4225 }, { "epoch": 0.15056023656411993, "grad_norm": 2.1827560697178394, "learning_rate": 1.9247657452583362e-05, "loss": 0.6543, "step": 4226 }, { "epoch": 0.15059586369061403, "grad_norm": 1.9051892475679295, "learning_rate": 1.9247218275545964e-05, "loss": 0.5767, "step": 4227 }, { "epoch": 0.15063149081710814, "grad_norm": 1.5207389055196552, "learning_rate": 1.9246778975375393e-05, "loss": 0.5688, "step": 4228 }, { "epoch": 0.15066711794360227, "grad_norm": 1.6916681763846917, "learning_rate": 1.9246339552077497e-05, "loss": 0.6112, "step": 4229 }, { "epoch": 0.15070274507009637, "grad_norm": 1.8230112606598117, "learning_rate": 1.9245900005658124e-05, "loss": 0.6436, "step": 4230 }, { "epoch": 0.15073837219659048, "grad_norm": 1.7240601797401753, "learning_rate": 1.9245460336123136e-05, "loss": 0.6861, "step": 4231 }, { "epoch": 0.1507739993230846, "grad_norm": 1.9068922009720355, "learning_rate": 1.9245020543478373e-05, "loss": 0.6459, "step": 4232 }, { "epoch": 0.1508096264495787, "grad_norm": 1.4935970022567602, "learning_rate": 1.9244580627729704e-05, "loss": 0.6492, "step": 4233 }, { "epoch": 0.1508452535760728, "grad_norm": 1.6374262663568226, "learning_rate": 1.9244140588882984e-05, "loss": 0.6536, "step": 4234 }, { "epoch": 0.15088088070256694, "grad_norm": 1.296143818819166, "learning_rate": 1.924370042694407e-05, "loss": 0.5123, "step": 4235 }, { "epoch": 0.15091650782906105, "grad_norm": 1.5290895167907634, "learning_rate": 1.9243260141918822e-05, "loss": 0.5453, "step": 4236 }, { "epoch": 0.15095213495555515, "grad_norm": 1.2395993126319518, "learning_rate": 1.9242819733813106e-05, "loss": 0.5387, "step": 4237 }, { "epoch": 0.15098776208204928, "grad_norm": 2.7597712869168647, "learning_rate": 1.9242379202632785e-05, "loss": 0.6609, "step": 4238 }, { "epoch": 0.15102338920854338, "grad_norm": 1.1710853016760019, "learning_rate": 1.9241938548383723e-05, "loss": 0.514, "step": 4239 }, { "epoch": 0.1510590163350375, "grad_norm": 1.3174791865943114, "learning_rate": 1.9241497771071792e-05, "loss": 0.5427, "step": 4240 }, { "epoch": 0.15109464346153162, "grad_norm": 1.2598187673614902, "learning_rate": 1.9241056870702857e-05, "loss": 0.5658, "step": 4241 }, { "epoch": 0.15113027058802572, "grad_norm": 1.3300652729807403, "learning_rate": 1.924061584728279e-05, "loss": 0.5147, "step": 4242 }, { "epoch": 0.15116589771451983, "grad_norm": 2.40832339628021, "learning_rate": 1.9240174700817464e-05, "loss": 0.6739, "step": 4243 }, { "epoch": 0.15120152484101396, "grad_norm": 1.3367698716756289, "learning_rate": 1.9239733431312756e-05, "loss": 0.4712, "step": 4244 }, { "epoch": 0.15123715196750806, "grad_norm": 1.7463015869915612, "learning_rate": 1.9239292038774538e-05, "loss": 0.6434, "step": 4245 }, { "epoch": 0.15127277909400216, "grad_norm": 2.612062641615783, "learning_rate": 1.9238850523208687e-05, "loss": 0.6585, "step": 4246 }, { "epoch": 0.1513084062204963, "grad_norm": 1.3023845945083659, "learning_rate": 1.923840888462109e-05, "loss": 0.5455, "step": 4247 }, { "epoch": 0.1513440333469904, "grad_norm": 1.2095873630940233, "learning_rate": 1.923796712301761e-05, "loss": 0.5676, "step": 4248 }, { "epoch": 0.1513796604734845, "grad_norm": 1.230385921602282, "learning_rate": 1.923752523840415e-05, "loss": 0.5567, "step": 4249 }, { "epoch": 0.15141528759997863, "grad_norm": 1.8093111856432134, "learning_rate": 1.923708323078658e-05, "loss": 0.6227, "step": 4250 }, { "epoch": 0.15145091472647274, "grad_norm": 1.340462838652078, "learning_rate": 1.9236641100170798e-05, "loss": 0.5972, "step": 4251 }, { "epoch": 0.15148654185296684, "grad_norm": 2.011110453453408, "learning_rate": 1.9236198846562676e-05, "loss": 0.5901, "step": 4252 }, { "epoch": 0.15152216897946097, "grad_norm": 1.3202702996456166, "learning_rate": 1.9235756469968112e-05, "loss": 0.5508, "step": 4253 }, { "epoch": 0.15155779610595507, "grad_norm": 1.2325475924535483, "learning_rate": 1.9235313970392997e-05, "loss": 0.5122, "step": 4254 }, { "epoch": 0.15159342323244918, "grad_norm": 1.6536147365298561, "learning_rate": 1.923487134784322e-05, "loss": 0.6512, "step": 4255 }, { "epoch": 0.1516290503589433, "grad_norm": 1.1747861456836857, "learning_rate": 1.9234428602324678e-05, "loss": 0.5137, "step": 4256 }, { "epoch": 0.1516646774854374, "grad_norm": 1.8439615281493702, "learning_rate": 1.9233985733843262e-05, "loss": 0.6644, "step": 4257 }, { "epoch": 0.1517003046119315, "grad_norm": 1.3682113046212843, "learning_rate": 1.9233542742404872e-05, "loss": 0.5963, "step": 4258 }, { "epoch": 0.15173593173842564, "grad_norm": 1.3362655004371176, "learning_rate": 1.923309962801541e-05, "loss": 0.5144, "step": 4259 }, { "epoch": 0.15177155886491975, "grad_norm": 2.1381657818630293, "learning_rate": 1.923265639068077e-05, "loss": 0.6032, "step": 4260 }, { "epoch": 0.15180718599141385, "grad_norm": 1.3343473288353243, "learning_rate": 1.9232213030406857e-05, "loss": 0.5229, "step": 4261 }, { "epoch": 0.15184281311790798, "grad_norm": 1.3686117092105046, "learning_rate": 1.9231769547199577e-05, "loss": 0.5421, "step": 4262 }, { "epoch": 0.15187844024440209, "grad_norm": 1.2825636263588216, "learning_rate": 1.923132594106483e-05, "loss": 0.5278, "step": 4263 }, { "epoch": 0.1519140673708962, "grad_norm": 1.6040177451912176, "learning_rate": 1.9230882212008528e-05, "loss": 0.5766, "step": 4264 }, { "epoch": 0.15194969449739032, "grad_norm": 1.5909904152186543, "learning_rate": 1.9230438360036577e-05, "loss": 0.7027, "step": 4265 }, { "epoch": 0.15198532162388442, "grad_norm": 1.3043984314941732, "learning_rate": 1.9229994385154884e-05, "loss": 0.5462, "step": 4266 }, { "epoch": 0.15202094875037853, "grad_norm": 1.6213199550149047, "learning_rate": 1.922955028736937e-05, "loss": 0.5986, "step": 4267 }, { "epoch": 0.15205657587687266, "grad_norm": 1.1833466166842779, "learning_rate": 1.922910606668594e-05, "loss": 0.5029, "step": 4268 }, { "epoch": 0.15209220300336676, "grad_norm": 1.7673635163966144, "learning_rate": 1.922866172311051e-05, "loss": 0.5603, "step": 4269 }, { "epoch": 0.15212783012986086, "grad_norm": 1.658694245558688, "learning_rate": 1.9228217256649e-05, "loss": 0.6803, "step": 4270 }, { "epoch": 0.152163457256355, "grad_norm": 1.3051672579291411, "learning_rate": 1.922777266730733e-05, "loss": 0.5188, "step": 4271 }, { "epoch": 0.1521990843828491, "grad_norm": 1.2704739173691424, "learning_rate": 1.9227327955091412e-05, "loss": 0.5598, "step": 4272 }, { "epoch": 0.1522347115093432, "grad_norm": 1.359896550213657, "learning_rate": 1.9226883120007178e-05, "loss": 0.5818, "step": 4273 }, { "epoch": 0.15227033863583733, "grad_norm": 1.8800813648226835, "learning_rate": 1.9226438162060544e-05, "loss": 0.5709, "step": 4274 }, { "epoch": 0.15230596576233144, "grad_norm": 1.3534488096327424, "learning_rate": 1.9225993081257435e-05, "loss": 0.4816, "step": 4275 }, { "epoch": 0.15234159288882554, "grad_norm": 2.108384749841118, "learning_rate": 1.9225547877603782e-05, "loss": 0.5345, "step": 4276 }, { "epoch": 0.15237722001531967, "grad_norm": 1.7396409847133454, "learning_rate": 1.922510255110551e-05, "loss": 0.573, "step": 4277 }, { "epoch": 0.15241284714181377, "grad_norm": 1.312969380848164, "learning_rate": 1.922465710176855e-05, "loss": 0.508, "step": 4278 }, { "epoch": 0.15244847426830788, "grad_norm": 2.6375148532724135, "learning_rate": 1.922421152959883e-05, "loss": 0.6937, "step": 4279 }, { "epoch": 0.152484101394802, "grad_norm": 1.9325914613648099, "learning_rate": 1.922376583460229e-05, "loss": 0.5508, "step": 4280 }, { "epoch": 0.1525197285212961, "grad_norm": 1.791524809592948, "learning_rate": 1.922332001678486e-05, "loss": 0.6161, "step": 4281 }, { "epoch": 0.15255535564779021, "grad_norm": 1.7374168555716027, "learning_rate": 1.9222874076152478e-05, "loss": 0.6319, "step": 4282 }, { "epoch": 0.15259098277428434, "grad_norm": 1.8068815547336026, "learning_rate": 1.9222428012711082e-05, "loss": 0.6217, "step": 4283 }, { "epoch": 0.15262660990077845, "grad_norm": 2.7922306129578236, "learning_rate": 1.9221981826466606e-05, "loss": 0.588, "step": 4284 }, { "epoch": 0.15266223702727255, "grad_norm": 1.2657322916108034, "learning_rate": 1.9221535517425002e-05, "loss": 0.5305, "step": 4285 }, { "epoch": 0.15269786415376668, "grad_norm": 1.1580864744759294, "learning_rate": 1.9221089085592203e-05, "loss": 0.4952, "step": 4286 }, { "epoch": 0.15273349128026079, "grad_norm": 1.6761685603330863, "learning_rate": 1.922064253097416e-05, "loss": 0.638, "step": 4287 }, { "epoch": 0.15276911840675492, "grad_norm": 1.88303929642101, "learning_rate": 1.9220195853576815e-05, "loss": 0.6844, "step": 4288 }, { "epoch": 0.15280474553324902, "grad_norm": 1.711784134572154, "learning_rate": 1.9219749053406118e-05, "loss": 0.6637, "step": 4289 }, { "epoch": 0.15284037265974312, "grad_norm": 1.2351560311366176, "learning_rate": 1.9219302130468017e-05, "loss": 0.5422, "step": 4290 }, { "epoch": 0.15287599978623725, "grad_norm": 1.204242025039708, "learning_rate": 1.9218855084768466e-05, "loss": 0.5541, "step": 4291 }, { "epoch": 0.15291162691273136, "grad_norm": 1.0636102621889603, "learning_rate": 1.9218407916313417e-05, "loss": 0.4909, "step": 4292 }, { "epoch": 0.15294725403922546, "grad_norm": 1.691739300598767, "learning_rate": 1.921796062510882e-05, "loss": 0.6257, "step": 4293 }, { "epoch": 0.1529828811657196, "grad_norm": 1.290522919701081, "learning_rate": 1.9217513211160637e-05, "loss": 0.546, "step": 4294 }, { "epoch": 0.1530185082922137, "grad_norm": 1.1505014932532867, "learning_rate": 1.9217065674474823e-05, "loss": 0.4698, "step": 4295 }, { "epoch": 0.1530541354187078, "grad_norm": 1.6629469963938293, "learning_rate": 1.9216618015057336e-05, "loss": 0.6844, "step": 4296 }, { "epoch": 0.15308976254520193, "grad_norm": 1.5590534902169184, "learning_rate": 1.921617023291414e-05, "loss": 0.5822, "step": 4297 }, { "epoch": 0.15312538967169603, "grad_norm": 1.4738322405863449, "learning_rate": 1.9215722328051195e-05, "loss": 0.6705, "step": 4298 }, { "epoch": 0.15316101679819014, "grad_norm": 1.351613472018911, "learning_rate": 1.9215274300474463e-05, "loss": 0.608, "step": 4299 }, { "epoch": 0.15319664392468427, "grad_norm": 1.201383700181399, "learning_rate": 1.9214826150189915e-05, "loss": 0.5316, "step": 4300 }, { "epoch": 0.15323227105117837, "grad_norm": 1.0974556409477436, "learning_rate": 1.921437787720352e-05, "loss": 0.5787, "step": 4301 }, { "epoch": 0.15326789817767247, "grad_norm": 1.6771421366100383, "learning_rate": 1.9213929481521237e-05, "loss": 0.7074, "step": 4302 }, { "epoch": 0.1533035253041666, "grad_norm": 1.2296983641631076, "learning_rate": 1.9213480963149042e-05, "loss": 0.551, "step": 4303 }, { "epoch": 0.1533391524306607, "grad_norm": 1.112352704163259, "learning_rate": 1.9213032322092913e-05, "loss": 0.5433, "step": 4304 }, { "epoch": 0.1533747795571548, "grad_norm": 1.7120864240927538, "learning_rate": 1.9212583558358816e-05, "loss": 0.6412, "step": 4305 }, { "epoch": 0.15341040668364894, "grad_norm": 2.7104619493573185, "learning_rate": 1.9212134671952733e-05, "loss": 0.6136, "step": 4306 }, { "epoch": 0.15344603381014305, "grad_norm": 1.2196355553863105, "learning_rate": 1.9211685662880636e-05, "loss": 0.5651, "step": 4307 }, { "epoch": 0.15348166093663715, "grad_norm": 1.183761713893787, "learning_rate": 1.92112365311485e-05, "loss": 0.5375, "step": 4308 }, { "epoch": 0.15351728806313128, "grad_norm": 1.2605790011074955, "learning_rate": 1.921078727676232e-05, "loss": 0.4959, "step": 4309 }, { "epoch": 0.15355291518962538, "grad_norm": 1.6531704862420122, "learning_rate": 1.9210337899728066e-05, "loss": 0.5939, "step": 4310 }, { "epoch": 0.1535885423161195, "grad_norm": 1.2629685871169378, "learning_rate": 1.9209888400051727e-05, "loss": 0.5445, "step": 4311 }, { "epoch": 0.15362416944261362, "grad_norm": 1.357712072786056, "learning_rate": 1.9209438777739283e-05, "loss": 0.6063, "step": 4312 }, { "epoch": 0.15365979656910772, "grad_norm": 1.2765007730708533, "learning_rate": 1.9208989032796728e-05, "loss": 0.5089, "step": 4313 }, { "epoch": 0.15369542369560182, "grad_norm": 1.2707372695666004, "learning_rate": 1.9208539165230044e-05, "loss": 0.5713, "step": 4314 }, { "epoch": 0.15373105082209595, "grad_norm": 1.807921886716642, "learning_rate": 1.9208089175045227e-05, "loss": 0.6584, "step": 4315 }, { "epoch": 0.15376667794859006, "grad_norm": 1.9450783928538717, "learning_rate": 1.9207639062248267e-05, "loss": 0.6524, "step": 4316 }, { "epoch": 0.15380230507508416, "grad_norm": 2.19662041674976, "learning_rate": 1.9207188826845156e-05, "loss": 0.6594, "step": 4317 }, { "epoch": 0.1538379322015783, "grad_norm": 1.8383199182764884, "learning_rate": 1.920673846884189e-05, "loss": 0.6782, "step": 4318 }, { "epoch": 0.1538735593280724, "grad_norm": 1.9442162560006797, "learning_rate": 1.9206287988244467e-05, "loss": 0.7194, "step": 4319 }, { "epoch": 0.1539091864545665, "grad_norm": 2.1632011584978765, "learning_rate": 1.9205837385058884e-05, "loss": 0.6006, "step": 4320 }, { "epoch": 0.15394481358106063, "grad_norm": 1.5371761698920512, "learning_rate": 1.9205386659291143e-05, "loss": 0.5671, "step": 4321 }, { "epoch": 0.15398044070755473, "grad_norm": 3.2564663589506604, "learning_rate": 1.9204935810947245e-05, "loss": 0.5718, "step": 4322 }, { "epoch": 0.15401606783404884, "grad_norm": 1.300158496726686, "learning_rate": 1.9204484840033187e-05, "loss": 0.5274, "step": 4323 }, { "epoch": 0.15405169496054297, "grad_norm": 1.2500604370678878, "learning_rate": 1.9204033746554987e-05, "loss": 0.5829, "step": 4324 }, { "epoch": 0.15408732208703707, "grad_norm": 1.8170050678072562, "learning_rate": 1.9203582530518644e-05, "loss": 0.5938, "step": 4325 }, { "epoch": 0.15412294921353117, "grad_norm": 1.6373570685944547, "learning_rate": 1.9203131191930162e-05, "loss": 0.5809, "step": 4326 }, { "epoch": 0.1541585763400253, "grad_norm": 1.1699570183180656, "learning_rate": 1.920267973079556e-05, "loss": 0.5105, "step": 4327 }, { "epoch": 0.1541942034665194, "grad_norm": 1.1224352374997082, "learning_rate": 1.920222814712084e-05, "loss": 0.5038, "step": 4328 }, { "epoch": 0.1542298305930135, "grad_norm": 1.6843734640742218, "learning_rate": 1.9201776440912024e-05, "loss": 0.57, "step": 4329 }, { "epoch": 0.15426545771950764, "grad_norm": 1.4820839518075366, "learning_rate": 1.9201324612175123e-05, "loss": 0.5415, "step": 4330 }, { "epoch": 0.15430108484600175, "grad_norm": 1.1643391867066868, "learning_rate": 1.920087266091615e-05, "loss": 0.477, "step": 4331 }, { "epoch": 0.15433671197249585, "grad_norm": 1.70178675782592, "learning_rate": 1.9200420587141133e-05, "loss": 0.6606, "step": 4332 }, { "epoch": 0.15437233909898998, "grad_norm": 1.7579073336032107, "learning_rate": 1.919996839085608e-05, "loss": 0.6864, "step": 4333 }, { "epoch": 0.15440796622548408, "grad_norm": 1.626831380350647, "learning_rate": 1.919951607206702e-05, "loss": 0.7173, "step": 4334 }, { "epoch": 0.1544435933519782, "grad_norm": 2.317971280031096, "learning_rate": 1.9199063630779973e-05, "loss": 0.6425, "step": 4335 }, { "epoch": 0.15447922047847232, "grad_norm": 1.3769792677939179, "learning_rate": 1.919861106700096e-05, "loss": 0.5541, "step": 4336 }, { "epoch": 0.15451484760496642, "grad_norm": 1.3733694447686706, "learning_rate": 1.9198158380736015e-05, "loss": 0.6556, "step": 4337 }, { "epoch": 0.15455047473146052, "grad_norm": 1.4483875496572058, "learning_rate": 1.9197705571991162e-05, "loss": 0.6124, "step": 4338 }, { "epoch": 0.15458610185795466, "grad_norm": 1.1367450026656423, "learning_rate": 1.919725264077243e-05, "loss": 0.464, "step": 4339 }, { "epoch": 0.15462172898444876, "grad_norm": 1.148868922487371, "learning_rate": 1.919679958708585e-05, "loss": 0.5415, "step": 4340 }, { "epoch": 0.15465735611094286, "grad_norm": 1.9454182316873396, "learning_rate": 1.9196346410937455e-05, "loss": 0.702, "step": 4341 }, { "epoch": 0.154692983237437, "grad_norm": 1.1835571562077507, "learning_rate": 1.9195893112333284e-05, "loss": 0.4767, "step": 4342 }, { "epoch": 0.1547286103639311, "grad_norm": 1.7484983312043307, "learning_rate": 1.9195439691279363e-05, "loss": 0.6782, "step": 4343 }, { "epoch": 0.1547642374904252, "grad_norm": 1.0806866844870402, "learning_rate": 1.9194986147781737e-05, "loss": 0.5324, "step": 4344 }, { "epoch": 0.15479986461691933, "grad_norm": 1.1868785512339561, "learning_rate": 1.9194532481846447e-05, "loss": 0.4958, "step": 4345 }, { "epoch": 0.15483549174341343, "grad_norm": 2.215396235734674, "learning_rate": 1.9194078693479524e-05, "loss": 0.7249, "step": 4346 }, { "epoch": 0.15487111886990754, "grad_norm": 1.9740638458501703, "learning_rate": 1.9193624782687023e-05, "loss": 0.6081, "step": 4347 }, { "epoch": 0.15490674599640167, "grad_norm": 1.4399957104597139, "learning_rate": 1.919317074947498e-05, "loss": 0.6533, "step": 4348 }, { "epoch": 0.15494237312289577, "grad_norm": 1.652085875217601, "learning_rate": 1.9192716593849443e-05, "loss": 0.6404, "step": 4349 }, { "epoch": 0.15497800024938987, "grad_norm": 1.5001220522517604, "learning_rate": 1.9192262315816457e-05, "loss": 0.5844, "step": 4350 }, { "epoch": 0.155013627375884, "grad_norm": 1.2888327924233294, "learning_rate": 1.9191807915382076e-05, "loss": 0.5025, "step": 4351 }, { "epoch": 0.1550492545023781, "grad_norm": 1.122990498802485, "learning_rate": 1.9191353392552346e-05, "loss": 0.5495, "step": 4352 }, { "epoch": 0.1550848816288722, "grad_norm": 1.9285523142326688, "learning_rate": 1.919089874733332e-05, "loss": 0.6757, "step": 4353 }, { "epoch": 0.15512050875536634, "grad_norm": 1.553747171728656, "learning_rate": 1.9190443979731056e-05, "loss": 0.6469, "step": 4354 }, { "epoch": 0.15515613588186045, "grad_norm": 1.1064154248185445, "learning_rate": 1.9189989089751605e-05, "loss": 0.526, "step": 4355 }, { "epoch": 0.15519176300835455, "grad_norm": 1.953395952193166, "learning_rate": 1.918953407740103e-05, "loss": 0.5886, "step": 4356 }, { "epoch": 0.15522739013484868, "grad_norm": 1.6288456317728515, "learning_rate": 1.9189078942685378e-05, "loss": 0.6188, "step": 4357 }, { "epoch": 0.15526301726134278, "grad_norm": 1.2158433528114585, "learning_rate": 1.9188623685610725e-05, "loss": 0.5291, "step": 4358 }, { "epoch": 0.1552986443878369, "grad_norm": 1.136354149548232, "learning_rate": 1.918816830618312e-05, "loss": 0.5523, "step": 4359 }, { "epoch": 0.15533427151433102, "grad_norm": 2.2893767435993992, "learning_rate": 1.9187712804408632e-05, "loss": 0.6199, "step": 4360 }, { "epoch": 0.15536989864082512, "grad_norm": 1.1455447512885455, "learning_rate": 1.918725718029333e-05, "loss": 0.5047, "step": 4361 }, { "epoch": 0.15540552576731922, "grad_norm": 2.0516370462863454, "learning_rate": 1.9186801433843272e-05, "loss": 0.603, "step": 4362 }, { "epoch": 0.15544115289381336, "grad_norm": 1.3372146673933005, "learning_rate": 1.918634556506454e-05, "loss": 0.5075, "step": 4363 }, { "epoch": 0.15547678002030746, "grad_norm": 1.8401012302235316, "learning_rate": 1.9185889573963185e-05, "loss": 0.6424, "step": 4364 }, { "epoch": 0.15551240714680156, "grad_norm": 1.3884095332336974, "learning_rate": 1.9185433460545296e-05, "loss": 0.556, "step": 4365 }, { "epoch": 0.1555480342732957, "grad_norm": 1.0866721360434866, "learning_rate": 1.918497722481694e-05, "loss": 0.5275, "step": 4366 }, { "epoch": 0.1555836613997898, "grad_norm": 1.0884863851818054, "learning_rate": 1.918452086678419e-05, "loss": 0.5187, "step": 4367 }, { "epoch": 0.1556192885262839, "grad_norm": 1.165521625026756, "learning_rate": 1.9184064386453127e-05, "loss": 0.5023, "step": 4368 }, { "epoch": 0.15565491565277803, "grad_norm": 1.7064854828976177, "learning_rate": 1.918360778382983e-05, "loss": 0.6437, "step": 4369 }, { "epoch": 0.15569054277927213, "grad_norm": 1.1368167480182503, "learning_rate": 1.9183151058920373e-05, "loss": 0.4726, "step": 4370 }, { "epoch": 0.15572616990576624, "grad_norm": 1.786286679992919, "learning_rate": 1.9182694211730842e-05, "loss": 0.6559, "step": 4371 }, { "epoch": 0.15576179703226037, "grad_norm": 1.2211377186536116, "learning_rate": 1.918223724226732e-05, "loss": 0.5158, "step": 4372 }, { "epoch": 0.15579742415875447, "grad_norm": 1.2029990827244144, "learning_rate": 1.9181780150535886e-05, "loss": 0.5276, "step": 4373 }, { "epoch": 0.15583305128524857, "grad_norm": 1.0651767671462964, "learning_rate": 1.9181322936542638e-05, "loss": 0.4937, "step": 4374 }, { "epoch": 0.1558686784117427, "grad_norm": 2.0432225523006515, "learning_rate": 1.9180865600293655e-05, "loss": 0.6654, "step": 4375 }, { "epoch": 0.1559043055382368, "grad_norm": 1.7912243216891819, "learning_rate": 1.9180408141795032e-05, "loss": 0.7022, "step": 4376 }, { "epoch": 0.1559399326647309, "grad_norm": 1.1808530911361397, "learning_rate": 1.9179950561052855e-05, "loss": 0.5647, "step": 4377 }, { "epoch": 0.15597555979122504, "grad_norm": 1.1002218769830088, "learning_rate": 1.917949285807322e-05, "loss": 0.5292, "step": 4378 }, { "epoch": 0.15601118691771915, "grad_norm": 1.9645306259354391, "learning_rate": 1.917903503286222e-05, "loss": 0.6064, "step": 4379 }, { "epoch": 0.15604681404421328, "grad_norm": 1.6080037312092943, "learning_rate": 1.9178577085425956e-05, "loss": 0.536, "step": 4380 }, { "epoch": 0.15608244117070738, "grad_norm": 1.272628391507608, "learning_rate": 1.917811901577052e-05, "loss": 0.5475, "step": 4381 }, { "epoch": 0.15611806829720148, "grad_norm": 2.4951975322237803, "learning_rate": 1.9177660823902017e-05, "loss": 0.6891, "step": 4382 }, { "epoch": 0.15615369542369562, "grad_norm": 1.8524035507049559, "learning_rate": 1.9177202509826545e-05, "loss": 0.6616, "step": 4383 }, { "epoch": 0.15618932255018972, "grad_norm": 1.1386095865589057, "learning_rate": 1.9176744073550204e-05, "loss": 0.504, "step": 4384 }, { "epoch": 0.15622494967668382, "grad_norm": 2.4592232064367674, "learning_rate": 1.9176285515079102e-05, "loss": 0.6599, "step": 4385 }, { "epoch": 0.15626057680317795, "grad_norm": 1.1180461446023544, "learning_rate": 1.9175826834419347e-05, "loss": 0.5462, "step": 4386 }, { "epoch": 0.15629620392967206, "grad_norm": 1.2481756590164557, "learning_rate": 1.9175368031577045e-05, "loss": 0.5039, "step": 4387 }, { "epoch": 0.15633183105616616, "grad_norm": 1.4806080681508171, "learning_rate": 1.9174909106558303e-05, "loss": 0.5084, "step": 4388 }, { "epoch": 0.1563674581826603, "grad_norm": 1.1848422710867643, "learning_rate": 1.917445005936923e-05, "loss": 0.5115, "step": 4389 }, { "epoch": 0.1564030853091544, "grad_norm": 1.5589759783356782, "learning_rate": 1.9173990890015948e-05, "loss": 0.6357, "step": 4390 }, { "epoch": 0.1564387124356485, "grad_norm": 1.1901617172333263, "learning_rate": 1.917353159850456e-05, "loss": 0.5104, "step": 4391 }, { "epoch": 0.15647433956214263, "grad_norm": 1.104598643558038, "learning_rate": 1.917307218484119e-05, "loss": 0.5021, "step": 4392 }, { "epoch": 0.15650996668863673, "grad_norm": 1.6281336478967103, "learning_rate": 1.917261264903195e-05, "loss": 0.7223, "step": 4393 }, { "epoch": 0.15654559381513083, "grad_norm": 1.1787844574987625, "learning_rate": 1.9172152991082963e-05, "loss": 0.526, "step": 4394 }, { "epoch": 0.15658122094162497, "grad_norm": 2.114589052892501, "learning_rate": 1.9171693211000348e-05, "loss": 0.6519, "step": 4395 }, { "epoch": 0.15661684806811907, "grad_norm": 1.9368735468718121, "learning_rate": 1.9171233308790225e-05, "loss": 0.622, "step": 4396 }, { "epoch": 0.15665247519461317, "grad_norm": 1.649226727972772, "learning_rate": 1.9170773284458725e-05, "loss": 0.5866, "step": 4397 }, { "epoch": 0.1566881023211073, "grad_norm": 1.281472796397149, "learning_rate": 1.9170313138011964e-05, "loss": 0.5556, "step": 4398 }, { "epoch": 0.1567237294476014, "grad_norm": 1.7121840269362287, "learning_rate": 1.9169852869456077e-05, "loss": 0.6639, "step": 4399 }, { "epoch": 0.1567593565740955, "grad_norm": 1.643049149158988, "learning_rate": 1.916939247879719e-05, "loss": 0.6205, "step": 4400 }, { "epoch": 0.15679498370058964, "grad_norm": 1.142304381793565, "learning_rate": 1.916893196604143e-05, "loss": 0.5557, "step": 4401 }, { "epoch": 0.15683061082708374, "grad_norm": 1.2871242214327527, "learning_rate": 1.9168471331194933e-05, "loss": 0.5017, "step": 4402 }, { "epoch": 0.15686623795357785, "grad_norm": 1.1668517671686471, "learning_rate": 1.9168010574263833e-05, "loss": 0.5337, "step": 4403 }, { "epoch": 0.15690186508007198, "grad_norm": 1.1849839384191825, "learning_rate": 1.9167549695254265e-05, "loss": 0.5481, "step": 4404 }, { "epoch": 0.15693749220656608, "grad_norm": 1.4997125478324522, "learning_rate": 1.916708869417236e-05, "loss": 0.5896, "step": 4405 }, { "epoch": 0.15697311933306018, "grad_norm": 1.8254683648190142, "learning_rate": 1.9166627571024267e-05, "loss": 0.6737, "step": 4406 }, { "epoch": 0.15700874645955432, "grad_norm": 1.4713014390518198, "learning_rate": 1.916616632581612e-05, "loss": 0.5935, "step": 4407 }, { "epoch": 0.15704437358604842, "grad_norm": 1.196831735997074, "learning_rate": 1.916570495855406e-05, "loss": 0.4875, "step": 4408 }, { "epoch": 0.15708000071254252, "grad_norm": 1.9827158551593382, "learning_rate": 1.9165243469244233e-05, "loss": 0.5951, "step": 4409 }, { "epoch": 0.15711562783903665, "grad_norm": 2.0749340202254363, "learning_rate": 1.916478185789278e-05, "loss": 0.6669, "step": 4410 }, { "epoch": 0.15715125496553076, "grad_norm": 1.346363805360437, "learning_rate": 1.9164320124505854e-05, "loss": 0.5234, "step": 4411 }, { "epoch": 0.15718688209202486, "grad_norm": 1.2493093482256732, "learning_rate": 1.91638582690896e-05, "loss": 0.5415, "step": 4412 }, { "epoch": 0.157222509218519, "grad_norm": 1.7617696797032931, "learning_rate": 1.916339629165017e-05, "loss": 0.5565, "step": 4413 }, { "epoch": 0.1572581363450131, "grad_norm": 1.91245717727317, "learning_rate": 1.916293419219371e-05, "loss": 0.7302, "step": 4414 }, { "epoch": 0.1572937634715072, "grad_norm": 1.5891408509936609, "learning_rate": 1.9162471970726377e-05, "loss": 0.6301, "step": 4415 }, { "epoch": 0.15732939059800133, "grad_norm": 1.2570756121756845, "learning_rate": 1.9162009627254327e-05, "loss": 0.5204, "step": 4416 }, { "epoch": 0.15736501772449543, "grad_norm": 4.786366117431003, "learning_rate": 1.9161547161783714e-05, "loss": 0.7181, "step": 4417 }, { "epoch": 0.15740064485098954, "grad_norm": 1.2593216560237073, "learning_rate": 1.9161084574320696e-05, "loss": 0.5495, "step": 4418 }, { "epoch": 0.15743627197748367, "grad_norm": 2.065349318791352, "learning_rate": 1.916062186487144e-05, "loss": 0.6361, "step": 4419 }, { "epoch": 0.15747189910397777, "grad_norm": 1.2759288821464987, "learning_rate": 1.9160159033442092e-05, "loss": 0.495, "step": 4420 }, { "epoch": 0.15750752623047187, "grad_norm": 1.1258942805261931, "learning_rate": 1.915969608003883e-05, "loss": 0.4696, "step": 4421 }, { "epoch": 0.157543153356966, "grad_norm": 1.1376612519239084, "learning_rate": 1.9159233004667807e-05, "loss": 0.5549, "step": 4422 }, { "epoch": 0.1575787804834601, "grad_norm": 1.9902745114967364, "learning_rate": 1.91587698073352e-05, "loss": 0.6535, "step": 4423 }, { "epoch": 0.1576144076099542, "grad_norm": 1.9974034029875543, "learning_rate": 1.915830648804717e-05, "loss": 0.6691, "step": 4424 }, { "epoch": 0.15765003473644834, "grad_norm": 1.3164048929638457, "learning_rate": 1.9157843046809886e-05, "loss": 0.5946, "step": 4425 }, { "epoch": 0.15768566186294244, "grad_norm": 1.8017636949139175, "learning_rate": 1.9157379483629524e-05, "loss": 0.6385, "step": 4426 }, { "epoch": 0.15772128898943655, "grad_norm": 3.4147365769326976, "learning_rate": 1.915691579851225e-05, "loss": 0.645, "step": 4427 }, { "epoch": 0.15775691611593068, "grad_norm": 1.584052131522244, "learning_rate": 1.9156451991464243e-05, "loss": 0.6804, "step": 4428 }, { "epoch": 0.15779254324242478, "grad_norm": 1.7659932574731465, "learning_rate": 1.915598806249168e-05, "loss": 0.6275, "step": 4429 }, { "epoch": 0.15782817036891889, "grad_norm": 2.246544586632664, "learning_rate": 1.915552401160073e-05, "loss": 0.6973, "step": 4430 }, { "epoch": 0.15786379749541302, "grad_norm": 1.3826756276823193, "learning_rate": 1.9155059838797588e-05, "loss": 0.5625, "step": 4431 }, { "epoch": 0.15789942462190712, "grad_norm": 1.4779705557910858, "learning_rate": 1.915459554408842e-05, "loss": 0.5713, "step": 4432 }, { "epoch": 0.15793505174840122, "grad_norm": 1.6239029176121764, "learning_rate": 1.9154131127479416e-05, "loss": 0.7649, "step": 4433 }, { "epoch": 0.15797067887489535, "grad_norm": 1.2109544101629452, "learning_rate": 1.9153666588976753e-05, "loss": 0.4746, "step": 4434 }, { "epoch": 0.15800630600138946, "grad_norm": 1.238528211491108, "learning_rate": 1.9153201928586625e-05, "loss": 0.5753, "step": 4435 }, { "epoch": 0.15804193312788356, "grad_norm": 1.2242402594314448, "learning_rate": 1.9152737146315213e-05, "loss": 0.553, "step": 4436 }, { "epoch": 0.1580775602543777, "grad_norm": 1.1660127163686904, "learning_rate": 1.915227224216871e-05, "loss": 0.5812, "step": 4437 }, { "epoch": 0.1581131873808718, "grad_norm": 1.1444961655617925, "learning_rate": 1.9151807216153305e-05, "loss": 0.5141, "step": 4438 }, { "epoch": 0.1581488145073659, "grad_norm": 1.181552696209814, "learning_rate": 1.9151342068275188e-05, "loss": 0.5003, "step": 4439 }, { "epoch": 0.15818444163386003, "grad_norm": 1.5088274283095338, "learning_rate": 1.915087679854056e-05, "loss": 0.6036, "step": 4440 }, { "epoch": 0.15822006876035413, "grad_norm": 1.8579010008106094, "learning_rate": 1.9150411406955606e-05, "loss": 0.7144, "step": 4441 }, { "epoch": 0.15825569588684824, "grad_norm": 1.5124620700693827, "learning_rate": 1.9149945893526532e-05, "loss": 0.5265, "step": 4442 }, { "epoch": 0.15829132301334237, "grad_norm": 1.2631522846738719, "learning_rate": 1.9149480258259535e-05, "loss": 0.4897, "step": 4443 }, { "epoch": 0.15832695013983647, "grad_norm": 1.3097948344171013, "learning_rate": 1.914901450116081e-05, "loss": 0.5176, "step": 4444 }, { "epoch": 0.15836257726633057, "grad_norm": 1.2108426414459075, "learning_rate": 1.914854862223656e-05, "loss": 0.5204, "step": 4445 }, { "epoch": 0.1583982043928247, "grad_norm": 1.7373239901431783, "learning_rate": 1.9148082621492994e-05, "loss": 0.6065, "step": 4446 }, { "epoch": 0.1584338315193188, "grad_norm": 2.1032903555558944, "learning_rate": 1.914761649893631e-05, "loss": 0.5999, "step": 4447 }, { "epoch": 0.1584694586458129, "grad_norm": 1.5372735441927765, "learning_rate": 1.9147150254572725e-05, "loss": 0.6001, "step": 4448 }, { "epoch": 0.15850508577230704, "grad_norm": 1.0791708819165822, "learning_rate": 1.9146683888408434e-05, "loss": 0.4683, "step": 4449 }, { "epoch": 0.15854071289880114, "grad_norm": 1.1540917522880099, "learning_rate": 1.914621740044966e-05, "loss": 0.5359, "step": 4450 }, { "epoch": 0.15857634002529525, "grad_norm": 1.1914144592981688, "learning_rate": 1.9145750790702606e-05, "loss": 0.5219, "step": 4451 }, { "epoch": 0.15861196715178938, "grad_norm": 1.7724285056847708, "learning_rate": 1.9145284059173485e-05, "loss": 0.6578, "step": 4452 }, { "epoch": 0.15864759427828348, "grad_norm": 1.1591429602618515, "learning_rate": 1.9144817205868515e-05, "loss": 0.5589, "step": 4453 }, { "epoch": 0.15868322140477759, "grad_norm": 1.2597816331400953, "learning_rate": 1.9144350230793916e-05, "loss": 0.5598, "step": 4454 }, { "epoch": 0.15871884853127172, "grad_norm": 2.634828743530438, "learning_rate": 1.91438831339559e-05, "loss": 0.595, "step": 4455 }, { "epoch": 0.15875447565776582, "grad_norm": 1.9072930629183096, "learning_rate": 1.914341591536069e-05, "loss": 0.6018, "step": 4456 }, { "epoch": 0.15879010278425992, "grad_norm": 1.8865190813179453, "learning_rate": 1.9142948575014504e-05, "loss": 0.6752, "step": 4457 }, { "epoch": 0.15882572991075405, "grad_norm": 1.261168897114374, "learning_rate": 1.914248111292357e-05, "loss": 0.505, "step": 4458 }, { "epoch": 0.15886135703724816, "grad_norm": 1.4957592500837076, "learning_rate": 1.9142013529094105e-05, "loss": 0.7059, "step": 4459 }, { "epoch": 0.15889698416374226, "grad_norm": 1.7229327971965174, "learning_rate": 1.9141545823532346e-05, "loss": 0.6344, "step": 4460 }, { "epoch": 0.1589326112902364, "grad_norm": 1.868414543369186, "learning_rate": 1.9141077996244512e-05, "loss": 0.6769, "step": 4461 }, { "epoch": 0.1589682384167305, "grad_norm": 1.4091951264240783, "learning_rate": 1.9140610047236834e-05, "loss": 0.4714, "step": 4462 }, { "epoch": 0.1590038655432246, "grad_norm": 1.8290504686301212, "learning_rate": 1.9140141976515545e-05, "loss": 0.6033, "step": 4463 }, { "epoch": 0.15903949266971873, "grad_norm": 2.7886748357562965, "learning_rate": 1.9139673784086875e-05, "loss": 0.614, "step": 4464 }, { "epoch": 0.15907511979621283, "grad_norm": 1.6420826144654561, "learning_rate": 1.9139205469957066e-05, "loss": 0.5949, "step": 4465 }, { "epoch": 0.15911074692270694, "grad_norm": 1.806621362379718, "learning_rate": 1.9138737034132343e-05, "loss": 0.6366, "step": 4466 }, { "epoch": 0.15914637404920107, "grad_norm": 2.2900577889757248, "learning_rate": 1.9138268476618953e-05, "loss": 0.6413, "step": 4467 }, { "epoch": 0.15918200117569517, "grad_norm": 1.2362505292543249, "learning_rate": 1.9137799797423128e-05, "loss": 0.5655, "step": 4468 }, { "epoch": 0.15921762830218927, "grad_norm": 1.712276841201928, "learning_rate": 1.913733099655111e-05, "loss": 0.6272, "step": 4469 }, { "epoch": 0.1592532554286834, "grad_norm": 1.2246079122231384, "learning_rate": 1.9136862074009147e-05, "loss": 0.5667, "step": 4470 }, { "epoch": 0.1592888825551775, "grad_norm": 2.235982877720603, "learning_rate": 1.913639302980348e-05, "loss": 0.7553, "step": 4471 }, { "epoch": 0.1593245096816716, "grad_norm": 2.1425336907292225, "learning_rate": 1.9135923863940352e-05, "loss": 0.6438, "step": 4472 }, { "epoch": 0.15936013680816574, "grad_norm": 2.24658181604337, "learning_rate": 1.913545457642601e-05, "loss": 0.6718, "step": 4473 }, { "epoch": 0.15939576393465985, "grad_norm": 1.3511754819716215, "learning_rate": 1.9134985167266708e-05, "loss": 0.5009, "step": 4474 }, { "epoch": 0.15943139106115398, "grad_norm": 1.84311934187191, "learning_rate": 1.913451563646869e-05, "loss": 0.6607, "step": 4475 }, { "epoch": 0.15946701818764808, "grad_norm": 1.1768423847455793, "learning_rate": 1.9134045984038216e-05, "loss": 0.5472, "step": 4476 }, { "epoch": 0.15950264531414218, "grad_norm": 2.143829316064273, "learning_rate": 1.9133576209981535e-05, "loss": 0.5618, "step": 4477 }, { "epoch": 0.15953827244063631, "grad_norm": 1.8396912263337766, "learning_rate": 1.91331063143049e-05, "loss": 0.5641, "step": 4478 }, { "epoch": 0.15957389956713042, "grad_norm": 2.625018021210089, "learning_rate": 1.9132636297014572e-05, "loss": 0.5738, "step": 4479 }, { "epoch": 0.15960952669362452, "grad_norm": 2.003546414189109, "learning_rate": 1.9132166158116806e-05, "loss": 0.6971, "step": 4480 }, { "epoch": 0.15964515382011865, "grad_norm": 1.9440119586157278, "learning_rate": 1.913169589761787e-05, "loss": 0.679, "step": 4481 }, { "epoch": 0.15968078094661275, "grad_norm": 2.3955794399139525, "learning_rate": 1.9131225515524017e-05, "loss": 0.641, "step": 4482 }, { "epoch": 0.15971640807310686, "grad_norm": 1.1833011981863313, "learning_rate": 1.9130755011841514e-05, "loss": 0.5766, "step": 4483 }, { "epoch": 0.159752035199601, "grad_norm": 2.6368520897299703, "learning_rate": 1.9130284386576627e-05, "loss": 0.6081, "step": 4484 }, { "epoch": 0.1597876623260951, "grad_norm": 2.60247534864183, "learning_rate": 1.912981363973562e-05, "loss": 0.5825, "step": 4485 }, { "epoch": 0.1598232894525892, "grad_norm": 1.2220121237454922, "learning_rate": 1.9129342771324765e-05, "loss": 0.5535, "step": 4486 }, { "epoch": 0.15985891657908333, "grad_norm": 1.8674989447923844, "learning_rate": 1.912887178135033e-05, "loss": 0.6992, "step": 4487 }, { "epoch": 0.15989454370557743, "grad_norm": 7.335698525325075, "learning_rate": 1.9128400669818586e-05, "loss": 0.5808, "step": 4488 }, { "epoch": 0.15993017083207153, "grad_norm": 1.199805865217878, "learning_rate": 1.9127929436735805e-05, "loss": 0.546, "step": 4489 }, { "epoch": 0.15996579795856566, "grad_norm": 1.7262335862095104, "learning_rate": 1.9127458082108265e-05, "loss": 0.7129, "step": 4490 }, { "epoch": 0.16000142508505977, "grad_norm": 1.6556079922142481, "learning_rate": 1.912698660594224e-05, "loss": 0.5986, "step": 4491 }, { "epoch": 0.16003705221155387, "grad_norm": 1.4028717260508803, "learning_rate": 1.912651500824401e-05, "loss": 0.6107, "step": 4492 }, { "epoch": 0.160072679338048, "grad_norm": 1.1537165152349005, "learning_rate": 1.9126043289019853e-05, "loss": 0.4797, "step": 4493 }, { "epoch": 0.1601083064645421, "grad_norm": 1.7307581533602914, "learning_rate": 1.912557144827605e-05, "loss": 0.6176, "step": 4494 }, { "epoch": 0.1601439335910362, "grad_norm": 1.0810388726482452, "learning_rate": 1.9125099486018887e-05, "loss": 0.4929, "step": 4495 }, { "epoch": 0.16017956071753034, "grad_norm": 2.005188288714615, "learning_rate": 1.9124627402254646e-05, "loss": 0.5595, "step": 4496 }, { "epoch": 0.16021518784402444, "grad_norm": 1.2737048621614269, "learning_rate": 1.912415519698961e-05, "loss": 0.5926, "step": 4497 }, { "epoch": 0.16025081497051855, "grad_norm": 1.1686251291574292, "learning_rate": 1.9123682870230075e-05, "loss": 0.4714, "step": 4498 }, { "epoch": 0.16028644209701268, "grad_norm": 1.1808212935487572, "learning_rate": 1.912321042198232e-05, "loss": 0.5189, "step": 4499 }, { "epoch": 0.16032206922350678, "grad_norm": 1.165938983464681, "learning_rate": 1.912273785225265e-05, "loss": 0.5088, "step": 4500 }, { "epoch": 0.16035769635000088, "grad_norm": 1.693128646365381, "learning_rate": 1.9122265161047342e-05, "loss": 0.5873, "step": 4501 }, { "epoch": 0.16039332347649501, "grad_norm": 2.226492938229141, "learning_rate": 1.91217923483727e-05, "loss": 0.6687, "step": 4502 }, { "epoch": 0.16042895060298912, "grad_norm": 1.2305692505435997, "learning_rate": 1.9121319414235017e-05, "loss": 0.5083, "step": 4503 }, { "epoch": 0.16046457772948322, "grad_norm": 1.330424884097968, "learning_rate": 1.9120846358640588e-05, "loss": 0.558, "step": 4504 }, { "epoch": 0.16050020485597735, "grad_norm": 1.2475605897437199, "learning_rate": 1.9120373181595716e-05, "loss": 0.5184, "step": 4505 }, { "epoch": 0.16053583198247146, "grad_norm": 1.6548203494746105, "learning_rate": 1.9119899883106702e-05, "loss": 0.6137, "step": 4506 }, { "epoch": 0.16057145910896556, "grad_norm": 1.146067882885797, "learning_rate": 1.9119426463179845e-05, "loss": 0.5228, "step": 4507 }, { "epoch": 0.1606070862354597, "grad_norm": 1.1398321015880666, "learning_rate": 1.9118952921821453e-05, "loss": 0.513, "step": 4508 }, { "epoch": 0.1606427133619538, "grad_norm": 2.0517584955297106, "learning_rate": 1.9118479259037826e-05, "loss": 0.5686, "step": 4509 }, { "epoch": 0.1606783404884479, "grad_norm": 1.7176173121338667, "learning_rate": 1.9118005474835275e-05, "loss": 0.6337, "step": 4510 }, { "epoch": 0.16071396761494203, "grad_norm": 1.2039793827532677, "learning_rate": 1.9117531569220106e-05, "loss": 0.5424, "step": 4511 }, { "epoch": 0.16074959474143613, "grad_norm": 1.206510052038331, "learning_rate": 1.9117057542198637e-05, "loss": 0.5175, "step": 4512 }, { "epoch": 0.16078522186793023, "grad_norm": 1.8409232574914804, "learning_rate": 1.911658339377717e-05, "loss": 0.5867, "step": 4513 }, { "epoch": 0.16082084899442436, "grad_norm": 1.927316866243077, "learning_rate": 1.9116109123962024e-05, "loss": 0.6773, "step": 4514 }, { "epoch": 0.16085647612091847, "grad_norm": 1.586215171832855, "learning_rate": 1.9115634732759512e-05, "loss": 0.5528, "step": 4515 }, { "epoch": 0.16089210324741257, "grad_norm": 1.2178407059578158, "learning_rate": 1.9115160220175952e-05, "loss": 0.5843, "step": 4516 }, { "epoch": 0.1609277303739067, "grad_norm": 1.2487443488931989, "learning_rate": 1.9114685586217666e-05, "loss": 0.5612, "step": 4517 }, { "epoch": 0.1609633575004008, "grad_norm": 1.3309184394691964, "learning_rate": 1.911421083089097e-05, "loss": 0.5014, "step": 4518 }, { "epoch": 0.1609989846268949, "grad_norm": 2.017082671687741, "learning_rate": 1.9113735954202184e-05, "loss": 0.6447, "step": 4519 }, { "epoch": 0.16103461175338904, "grad_norm": 4.8666755497582015, "learning_rate": 1.9113260956157634e-05, "loss": 0.5874, "step": 4520 }, { "epoch": 0.16107023887988314, "grad_norm": 1.8594346748628885, "learning_rate": 1.9112785836763645e-05, "loss": 0.5906, "step": 4521 }, { "epoch": 0.16110586600637725, "grad_norm": 2.0049985709709963, "learning_rate": 1.9112310596026545e-05, "loss": 0.6633, "step": 4522 }, { "epoch": 0.16114149313287138, "grad_norm": 1.1450586676961778, "learning_rate": 1.911183523395266e-05, "loss": 0.511, "step": 4523 }, { "epoch": 0.16117712025936548, "grad_norm": 1.1864685808786446, "learning_rate": 1.9111359750548314e-05, "loss": 0.5588, "step": 4524 }, { "epoch": 0.16121274738585958, "grad_norm": 1.1596017221839818, "learning_rate": 1.9110884145819852e-05, "loss": 0.5052, "step": 4525 }, { "epoch": 0.16124837451235371, "grad_norm": 1.1819620830481707, "learning_rate": 1.9110408419773595e-05, "loss": 0.5357, "step": 4526 }, { "epoch": 0.16128400163884782, "grad_norm": 1.7016293300595229, "learning_rate": 1.9109932572415886e-05, "loss": 0.6635, "step": 4527 }, { "epoch": 0.16131962876534192, "grad_norm": 1.1389039880717373, "learning_rate": 1.910945660375305e-05, "loss": 0.5427, "step": 4528 }, { "epoch": 0.16135525589183605, "grad_norm": 1.1923202438359772, "learning_rate": 1.910898051379144e-05, "loss": 0.5755, "step": 4529 }, { "epoch": 0.16139088301833016, "grad_norm": 1.2146222410262715, "learning_rate": 1.9108504302537386e-05, "loss": 0.5078, "step": 4530 }, { "epoch": 0.16142651014482426, "grad_norm": 1.3224195908790104, "learning_rate": 1.910802796999723e-05, "loss": 0.5717, "step": 4531 }, { "epoch": 0.1614621372713184, "grad_norm": 1.1251128323355106, "learning_rate": 1.9107551516177313e-05, "loss": 0.4861, "step": 4532 }, { "epoch": 0.1614977643978125, "grad_norm": 1.830603123833558, "learning_rate": 1.9107074941083987e-05, "loss": 0.5816, "step": 4533 }, { "epoch": 0.1615333915243066, "grad_norm": 1.3501557539622582, "learning_rate": 1.910659824472359e-05, "loss": 0.5972, "step": 4534 }, { "epoch": 0.16156901865080073, "grad_norm": 1.40376790058434, "learning_rate": 1.910612142710247e-05, "loss": 0.6565, "step": 4535 }, { "epoch": 0.16160464577729483, "grad_norm": 1.0982360192391527, "learning_rate": 1.9105644488226978e-05, "loss": 0.4867, "step": 4536 }, { "epoch": 0.16164027290378893, "grad_norm": 1.2170184329236255, "learning_rate": 1.910516742810347e-05, "loss": 0.5216, "step": 4537 }, { "epoch": 0.16167590003028307, "grad_norm": 1.3842933266682795, "learning_rate": 1.9104690246738288e-05, "loss": 0.5492, "step": 4538 }, { "epoch": 0.16171152715677717, "grad_norm": 1.3836991100858007, "learning_rate": 1.9104212944137796e-05, "loss": 0.5496, "step": 4539 }, { "epoch": 0.16174715428327127, "grad_norm": 1.221764066925297, "learning_rate": 1.9103735520308345e-05, "loss": 0.5481, "step": 4540 }, { "epoch": 0.1617827814097654, "grad_norm": 1.928510836319283, "learning_rate": 1.910325797525629e-05, "loss": 0.6952, "step": 4541 }, { "epoch": 0.1618184085362595, "grad_norm": 1.7625915372192271, "learning_rate": 1.910278030898799e-05, "loss": 0.6185, "step": 4542 }, { "epoch": 0.1618540356627536, "grad_norm": 1.2961191353953494, "learning_rate": 1.9102302521509814e-05, "loss": 0.5688, "step": 4543 }, { "epoch": 0.16188966278924774, "grad_norm": 1.7634395087950316, "learning_rate": 1.9101824612828113e-05, "loss": 0.625, "step": 4544 }, { "epoch": 0.16192528991574184, "grad_norm": 1.3716526000998814, "learning_rate": 1.910134658294926e-05, "loss": 0.5717, "step": 4545 }, { "epoch": 0.16196091704223595, "grad_norm": 1.5226318122169324, "learning_rate": 1.910086843187961e-05, "loss": 0.6172, "step": 4546 }, { "epoch": 0.16199654416873008, "grad_norm": 1.2626791573490659, "learning_rate": 1.910039015962554e-05, "loss": 0.5192, "step": 4547 }, { "epoch": 0.16203217129522418, "grad_norm": 1.4552108376191755, "learning_rate": 1.9099911766193414e-05, "loss": 0.5186, "step": 4548 }, { "epoch": 0.16206779842171828, "grad_norm": 1.8578306824926514, "learning_rate": 1.90994332515896e-05, "loss": 0.708, "step": 4549 }, { "epoch": 0.16210342554821242, "grad_norm": 1.2457718340229313, "learning_rate": 1.9098954615820477e-05, "loss": 0.5197, "step": 4550 }, { "epoch": 0.16213905267470652, "grad_norm": 1.9132795541456384, "learning_rate": 1.909847585889241e-05, "loss": 0.6312, "step": 4551 }, { "epoch": 0.16217467980120062, "grad_norm": 1.218424693690435, "learning_rate": 1.9097996980811774e-05, "loss": 0.4841, "step": 4552 }, { "epoch": 0.16221030692769475, "grad_norm": 1.3300261274658038, "learning_rate": 1.9097517981584955e-05, "loss": 0.5198, "step": 4553 }, { "epoch": 0.16224593405418886, "grad_norm": 1.73204041553829, "learning_rate": 1.909703886121832e-05, "loss": 0.5649, "step": 4554 }, { "epoch": 0.16228156118068296, "grad_norm": 1.8816709149229742, "learning_rate": 1.9096559619718256e-05, "loss": 0.613, "step": 4555 }, { "epoch": 0.1623171883071771, "grad_norm": 1.1381107209103989, "learning_rate": 1.909608025709115e-05, "loss": 0.4284, "step": 4556 }, { "epoch": 0.1623528154336712, "grad_norm": 1.4628369567181922, "learning_rate": 1.909560077334337e-05, "loss": 0.5272, "step": 4557 }, { "epoch": 0.1623884425601653, "grad_norm": 1.6781167215820911, "learning_rate": 1.9095121168481308e-05, "loss": 0.6113, "step": 4558 }, { "epoch": 0.16242406968665943, "grad_norm": 1.362568106784794, "learning_rate": 1.909464144251135e-05, "loss": 0.5626, "step": 4559 }, { "epoch": 0.16245969681315353, "grad_norm": 2.0355169288685078, "learning_rate": 1.9094161595439887e-05, "loss": 0.5993, "step": 4560 }, { "epoch": 0.16249532393964763, "grad_norm": 1.7581207684873075, "learning_rate": 1.9093681627273306e-05, "loss": 0.5516, "step": 4561 }, { "epoch": 0.16253095106614177, "grad_norm": 3.0858613243645303, "learning_rate": 1.9093201538018e-05, "loss": 0.6368, "step": 4562 }, { "epoch": 0.16256657819263587, "grad_norm": 1.8588582032684946, "learning_rate": 1.9092721327680356e-05, "loss": 0.6129, "step": 4563 }, { "epoch": 0.16260220531912997, "grad_norm": 2.1206585459957794, "learning_rate": 1.9092240996266772e-05, "loss": 0.6209, "step": 4564 }, { "epoch": 0.1626378324456241, "grad_norm": 1.425619832106798, "learning_rate": 1.9091760543783646e-05, "loss": 0.5006, "step": 4565 }, { "epoch": 0.1626734595721182, "grad_norm": 6.947952420634789, "learning_rate": 1.9091279970237373e-05, "loss": 0.5371, "step": 4566 }, { "epoch": 0.1627090866986123, "grad_norm": 1.181879360172853, "learning_rate": 1.9090799275634354e-05, "loss": 0.5047, "step": 4567 }, { "epoch": 0.16274471382510644, "grad_norm": 1.713004004666241, "learning_rate": 1.9090318459980986e-05, "loss": 0.6845, "step": 4568 }, { "epoch": 0.16278034095160054, "grad_norm": 1.8911888110071753, "learning_rate": 1.9089837523283673e-05, "loss": 0.5237, "step": 4569 }, { "epoch": 0.16281596807809467, "grad_norm": 1.8590160949849426, "learning_rate": 1.9089356465548824e-05, "loss": 0.6568, "step": 4570 }, { "epoch": 0.16285159520458878, "grad_norm": 4.458715426292254, "learning_rate": 1.9088875286782836e-05, "loss": 0.5575, "step": 4571 }, { "epoch": 0.16288722233108288, "grad_norm": 2.0726133136648683, "learning_rate": 1.9088393986992124e-05, "loss": 0.5597, "step": 4572 }, { "epoch": 0.162922849457577, "grad_norm": 1.6163479965941163, "learning_rate": 1.9087912566183094e-05, "loss": 0.5965, "step": 4573 }, { "epoch": 0.16295847658407112, "grad_norm": 1.6119166622892172, "learning_rate": 1.9087431024362156e-05, "loss": 0.5363, "step": 4574 }, { "epoch": 0.16299410371056522, "grad_norm": 1.301115490262957, "learning_rate": 1.908694936153572e-05, "loss": 0.4841, "step": 4575 }, { "epoch": 0.16302973083705935, "grad_norm": 1.5072614434719944, "learning_rate": 1.9086467577710203e-05, "loss": 0.5449, "step": 4576 }, { "epoch": 0.16306535796355345, "grad_norm": 1.8734527248087045, "learning_rate": 1.908598567289202e-05, "loss": 0.6492, "step": 4577 }, { "epoch": 0.16310098509004756, "grad_norm": 1.3989154416996443, "learning_rate": 1.9085503647087588e-05, "loss": 0.4879, "step": 4578 }, { "epoch": 0.1631366122165417, "grad_norm": 1.6800481932252713, "learning_rate": 1.9085021500303322e-05, "loss": 0.5379, "step": 4579 }, { "epoch": 0.1631722393430358, "grad_norm": 1.827226359082184, "learning_rate": 1.9084539232545645e-05, "loss": 0.5879, "step": 4580 }, { "epoch": 0.1632078664695299, "grad_norm": 2.3629293656082413, "learning_rate": 1.908405684382098e-05, "loss": 0.6031, "step": 4581 }, { "epoch": 0.16324349359602403, "grad_norm": 2.07082762831314, "learning_rate": 1.9083574334135745e-05, "loss": 0.6238, "step": 4582 }, { "epoch": 0.16327912072251813, "grad_norm": 1.4534266412339267, "learning_rate": 1.9083091703496373e-05, "loss": 0.5232, "step": 4583 }, { "epoch": 0.16331474784901223, "grad_norm": 1.289729596013263, "learning_rate": 1.9082608951909285e-05, "loss": 0.55, "step": 4584 }, { "epoch": 0.16335037497550636, "grad_norm": 1.9495408714323554, "learning_rate": 1.9082126079380908e-05, "loss": 0.5189, "step": 4585 }, { "epoch": 0.16338600210200047, "grad_norm": 1.914477544146337, "learning_rate": 1.908164308591768e-05, "loss": 0.6304, "step": 4586 }, { "epoch": 0.16342162922849457, "grad_norm": 1.5642278852712888, "learning_rate": 1.9081159971526022e-05, "loss": 0.5985, "step": 4587 }, { "epoch": 0.1634572563549887, "grad_norm": 1.5995365653017393, "learning_rate": 1.908067673621237e-05, "loss": 0.5564, "step": 4588 }, { "epoch": 0.1634928834814828, "grad_norm": 1.5276335290959178, "learning_rate": 1.9080193379983166e-05, "loss": 0.6757, "step": 4589 }, { "epoch": 0.1635285106079769, "grad_norm": 2.3490677077801747, "learning_rate": 1.9079709902844837e-05, "loss": 0.617, "step": 4590 }, { "epoch": 0.16356413773447104, "grad_norm": 1.6370676118667908, "learning_rate": 1.9079226304803823e-05, "loss": 0.4793, "step": 4591 }, { "epoch": 0.16359976486096514, "grad_norm": 1.939425581611866, "learning_rate": 1.9078742585866566e-05, "loss": 0.515, "step": 4592 }, { "epoch": 0.16363539198745924, "grad_norm": 1.7149726051710916, "learning_rate": 1.907825874603951e-05, "loss": 0.5049, "step": 4593 }, { "epoch": 0.16367101911395338, "grad_norm": 1.5799154285594088, "learning_rate": 1.907777478532909e-05, "loss": 0.5879, "step": 4594 }, { "epoch": 0.16370664624044748, "grad_norm": 1.7307235745895018, "learning_rate": 1.9077290703741753e-05, "loss": 0.518, "step": 4595 }, { "epoch": 0.16374227336694158, "grad_norm": 1.8758176501326052, "learning_rate": 1.9076806501283944e-05, "loss": 0.6736, "step": 4596 }, { "epoch": 0.1637779004934357, "grad_norm": 2.3670466509431414, "learning_rate": 1.9076322177962116e-05, "loss": 0.5536, "step": 4597 }, { "epoch": 0.16381352761992982, "grad_norm": 1.5520010705136253, "learning_rate": 1.907583773378271e-05, "loss": 0.582, "step": 4598 }, { "epoch": 0.16384915474642392, "grad_norm": 1.615216780689787, "learning_rate": 1.9075353168752184e-05, "loss": 0.4784, "step": 4599 }, { "epoch": 0.16388478187291805, "grad_norm": 1.4233629527613434, "learning_rate": 1.9074868482876988e-05, "loss": 0.4959, "step": 4600 }, { "epoch": 0.16392040899941215, "grad_norm": 1.4381703178359933, "learning_rate": 1.9074383676163574e-05, "loss": 0.5245, "step": 4601 }, { "epoch": 0.16395603612590626, "grad_norm": 1.522816038427834, "learning_rate": 1.9073898748618396e-05, "loss": 0.5287, "step": 4602 }, { "epoch": 0.1639916632524004, "grad_norm": 1.5498038267724967, "learning_rate": 1.907341370024792e-05, "loss": 0.516, "step": 4603 }, { "epoch": 0.1640272903788945, "grad_norm": 2.4873827814388774, "learning_rate": 1.9072928531058594e-05, "loss": 0.6322, "step": 4604 }, { "epoch": 0.1640629175053886, "grad_norm": 1.5736180296669622, "learning_rate": 1.9072443241056884e-05, "loss": 0.6385, "step": 4605 }, { "epoch": 0.16409854463188273, "grad_norm": 1.8132033072414167, "learning_rate": 1.9071957830249252e-05, "loss": 0.5908, "step": 4606 }, { "epoch": 0.16413417175837683, "grad_norm": 1.5675822541268059, "learning_rate": 1.907147229864216e-05, "loss": 0.5259, "step": 4607 }, { "epoch": 0.16416979888487093, "grad_norm": 1.5061327069572719, "learning_rate": 1.9070986646242072e-05, "loss": 0.5512, "step": 4608 }, { "epoch": 0.16420542601136506, "grad_norm": 2.5480581306900154, "learning_rate": 1.907050087305546e-05, "loss": 0.6042, "step": 4609 }, { "epoch": 0.16424105313785917, "grad_norm": 1.6104161068923752, "learning_rate": 1.9070014979088786e-05, "loss": 0.5698, "step": 4610 }, { "epoch": 0.16427668026435327, "grad_norm": 2.0401728181812566, "learning_rate": 1.9069528964348522e-05, "loss": 0.6462, "step": 4611 }, { "epoch": 0.1643123073908474, "grad_norm": 1.7861134295576908, "learning_rate": 1.9069042828841144e-05, "loss": 0.5966, "step": 4612 }, { "epoch": 0.1643479345173415, "grad_norm": 1.544741944644765, "learning_rate": 1.906855657257312e-05, "loss": 0.5724, "step": 4613 }, { "epoch": 0.1643835616438356, "grad_norm": 1.899127187027242, "learning_rate": 1.9068070195550928e-05, "loss": 0.6273, "step": 4614 }, { "epoch": 0.16441918877032974, "grad_norm": 1.795114333651609, "learning_rate": 1.9067583697781043e-05, "loss": 0.6112, "step": 4615 }, { "epoch": 0.16445481589682384, "grad_norm": 2.054935910878483, "learning_rate": 1.9067097079269942e-05, "loss": 0.6379, "step": 4616 }, { "epoch": 0.16449044302331794, "grad_norm": 1.5375438000567563, "learning_rate": 1.9066610340024106e-05, "loss": 0.5662, "step": 4617 }, { "epoch": 0.16452607014981208, "grad_norm": 1.8665452361612627, "learning_rate": 1.9066123480050017e-05, "loss": 0.5762, "step": 4618 }, { "epoch": 0.16456169727630618, "grad_norm": 1.9614714676167166, "learning_rate": 1.9065636499354157e-05, "loss": 0.6266, "step": 4619 }, { "epoch": 0.16459732440280028, "grad_norm": 1.935056133267607, "learning_rate": 1.906514939794301e-05, "loss": 0.5152, "step": 4620 }, { "epoch": 0.1646329515292944, "grad_norm": 1.7125846473033195, "learning_rate": 1.906466217582306e-05, "loss": 0.627, "step": 4621 }, { "epoch": 0.16466857865578852, "grad_norm": 1.5550423227040822, "learning_rate": 1.90641748330008e-05, "loss": 0.5967, "step": 4622 }, { "epoch": 0.16470420578228262, "grad_norm": 1.3866665462191765, "learning_rate": 1.906368736948272e-05, "loss": 0.5768, "step": 4623 }, { "epoch": 0.16473983290877675, "grad_norm": 1.8262832717295543, "learning_rate": 1.9063199785275302e-05, "loss": 0.6918, "step": 4624 }, { "epoch": 0.16477546003527085, "grad_norm": 2.080139332222163, "learning_rate": 1.9062712080385045e-05, "loss": 0.7203, "step": 4625 }, { "epoch": 0.16481108716176496, "grad_norm": 1.8420015448625788, "learning_rate": 1.9062224254818445e-05, "loss": 0.6275, "step": 4626 }, { "epoch": 0.1648467142882591, "grad_norm": 1.331065784998801, "learning_rate": 1.9061736308581996e-05, "loss": 0.526, "step": 4627 }, { "epoch": 0.1648823414147532, "grad_norm": 1.4109570254146697, "learning_rate": 1.9061248241682187e-05, "loss": 0.6274, "step": 4628 }, { "epoch": 0.1649179685412473, "grad_norm": 1.535870575108902, "learning_rate": 1.906076005412553e-05, "loss": 0.5985, "step": 4629 }, { "epoch": 0.16495359566774143, "grad_norm": 1.2000361974739209, "learning_rate": 1.906027174591852e-05, "loss": 0.5236, "step": 4630 }, { "epoch": 0.16498922279423553, "grad_norm": 1.3976847466393552, "learning_rate": 1.9059783317067658e-05, "loss": 0.5503, "step": 4631 }, { "epoch": 0.16502484992072963, "grad_norm": 1.9129148064804744, "learning_rate": 1.9059294767579447e-05, "loss": 0.7026, "step": 4632 }, { "epoch": 0.16506047704722376, "grad_norm": 1.3136451054605005, "learning_rate": 1.9058806097460395e-05, "loss": 0.5466, "step": 4633 }, { "epoch": 0.16509610417371787, "grad_norm": 1.7840688185080147, "learning_rate": 1.905831730671701e-05, "loss": 0.6667, "step": 4634 }, { "epoch": 0.16513173130021197, "grad_norm": 1.5256786381907959, "learning_rate": 1.90578283953558e-05, "loss": 0.7187, "step": 4635 }, { "epoch": 0.1651673584267061, "grad_norm": 1.9916137056082261, "learning_rate": 1.9057339363383268e-05, "loss": 0.6274, "step": 4636 }, { "epoch": 0.1652029855532002, "grad_norm": 2.920651250549839, "learning_rate": 1.9056850210805935e-05, "loss": 0.6332, "step": 4637 }, { "epoch": 0.1652386126796943, "grad_norm": 1.999328841398814, "learning_rate": 1.905636093763031e-05, "loss": 0.5837, "step": 4638 }, { "epoch": 0.16527423980618844, "grad_norm": 2.323592519399766, "learning_rate": 1.905587154386291e-05, "loss": 0.6095, "step": 4639 }, { "epoch": 0.16530986693268254, "grad_norm": 1.2929536543546007, "learning_rate": 1.905538202951025e-05, "loss": 0.5361, "step": 4640 }, { "epoch": 0.16534549405917665, "grad_norm": 1.3325514328473098, "learning_rate": 1.905489239457885e-05, "loss": 0.5818, "step": 4641 }, { "epoch": 0.16538112118567078, "grad_norm": 1.7577799671129755, "learning_rate": 1.9054402639075226e-05, "loss": 0.6475, "step": 4642 }, { "epoch": 0.16541674831216488, "grad_norm": 1.2196865053178219, "learning_rate": 1.9053912763005907e-05, "loss": 0.4769, "step": 4643 }, { "epoch": 0.16545237543865898, "grad_norm": 1.83698459913823, "learning_rate": 1.905342276637741e-05, "loss": 0.5929, "step": 4644 }, { "epoch": 0.16548800256515311, "grad_norm": 1.7089140157978187, "learning_rate": 1.905293264919626e-05, "loss": 0.6546, "step": 4645 }, { "epoch": 0.16552362969164722, "grad_norm": 1.9695511809659338, "learning_rate": 1.9052442411468984e-05, "loss": 0.6299, "step": 4646 }, { "epoch": 0.16555925681814132, "grad_norm": 1.8245843523594232, "learning_rate": 1.9051952053202107e-05, "loss": 0.5607, "step": 4647 }, { "epoch": 0.16559488394463545, "grad_norm": 2.4510178073664686, "learning_rate": 1.9051461574402165e-05, "loss": 0.6345, "step": 4648 }, { "epoch": 0.16563051107112955, "grad_norm": 1.954384047979618, "learning_rate": 1.9050970975075685e-05, "loss": 0.5777, "step": 4649 }, { "epoch": 0.16566613819762366, "grad_norm": 1.1724475911571692, "learning_rate": 1.90504802552292e-05, "loss": 0.5152, "step": 4650 }, { "epoch": 0.1657017653241178, "grad_norm": 1.3052255887800397, "learning_rate": 1.9049989414869245e-05, "loss": 0.5631, "step": 4651 }, { "epoch": 0.1657373924506119, "grad_norm": 1.3294266602431766, "learning_rate": 1.9049498454002354e-05, "loss": 0.4965, "step": 4652 }, { "epoch": 0.165773019577106, "grad_norm": 2.0283666452234934, "learning_rate": 1.9049007372635068e-05, "loss": 0.6283, "step": 4653 }, { "epoch": 0.16580864670360013, "grad_norm": 1.8784069708216704, "learning_rate": 1.9048516170773924e-05, "loss": 0.6717, "step": 4654 }, { "epoch": 0.16584427383009423, "grad_norm": 1.2248007565529981, "learning_rate": 1.904802484842546e-05, "loss": 0.4837, "step": 4655 }, { "epoch": 0.16587990095658833, "grad_norm": 1.3403664572289853, "learning_rate": 1.9047533405596225e-05, "loss": 0.4766, "step": 4656 }, { "epoch": 0.16591552808308246, "grad_norm": 1.5242170000725588, "learning_rate": 1.9047041842292752e-05, "loss": 0.6149, "step": 4657 }, { "epoch": 0.16595115520957657, "grad_norm": 1.254034491159179, "learning_rate": 1.90465501585216e-05, "loss": 0.5616, "step": 4658 }, { "epoch": 0.16598678233607067, "grad_norm": 1.8966138725796458, "learning_rate": 1.9046058354289307e-05, "loss": 0.6107, "step": 4659 }, { "epoch": 0.1660224094625648, "grad_norm": 1.4149093114873545, "learning_rate": 1.9045566429602425e-05, "loss": 0.5241, "step": 4660 }, { "epoch": 0.1660580365890589, "grad_norm": 1.3812666077754436, "learning_rate": 1.90450743844675e-05, "loss": 0.5073, "step": 4661 }, { "epoch": 0.166093663715553, "grad_norm": 1.1971084330541195, "learning_rate": 1.904458221889109e-05, "loss": 0.5272, "step": 4662 }, { "epoch": 0.16612929084204714, "grad_norm": 1.245991590260896, "learning_rate": 1.904408993287975e-05, "loss": 0.5187, "step": 4663 }, { "epoch": 0.16616491796854124, "grad_norm": 1.8558226955700083, "learning_rate": 1.9043597526440027e-05, "loss": 0.6693, "step": 4664 }, { "epoch": 0.16620054509503537, "grad_norm": 1.273747365252878, "learning_rate": 1.904310499957848e-05, "loss": 0.5447, "step": 4665 }, { "epoch": 0.16623617222152948, "grad_norm": 1.8765711577382462, "learning_rate": 1.904261235230167e-05, "loss": 0.6172, "step": 4666 }, { "epoch": 0.16627179934802358, "grad_norm": 1.2750287558816418, "learning_rate": 1.9042119584616157e-05, "loss": 0.564, "step": 4667 }, { "epoch": 0.1663074264745177, "grad_norm": 1.1667875379136365, "learning_rate": 1.9041626696528503e-05, "loss": 0.5013, "step": 4668 }, { "epoch": 0.16634305360101181, "grad_norm": 1.7884227819309595, "learning_rate": 1.904113368804527e-05, "loss": 0.644, "step": 4669 }, { "epoch": 0.16637868072750592, "grad_norm": 1.7267176737773915, "learning_rate": 1.904064055917302e-05, "loss": 0.69, "step": 4670 }, { "epoch": 0.16641430785400005, "grad_norm": 2.0401615854144306, "learning_rate": 1.9040147309918326e-05, "loss": 0.6375, "step": 4671 }, { "epoch": 0.16644993498049415, "grad_norm": 1.2177770613110974, "learning_rate": 1.903965394028775e-05, "loss": 0.5265, "step": 4672 }, { "epoch": 0.16648556210698826, "grad_norm": 1.638393296533366, "learning_rate": 1.903916045028786e-05, "loss": 0.6112, "step": 4673 }, { "epoch": 0.1665211892334824, "grad_norm": 1.3172928943185058, "learning_rate": 1.9038666839925234e-05, "loss": 0.5163, "step": 4674 }, { "epoch": 0.1665568163599765, "grad_norm": 1.1323661312785847, "learning_rate": 1.9038173109206443e-05, "loss": 0.474, "step": 4675 }, { "epoch": 0.1665924434864706, "grad_norm": 1.5912176541246514, "learning_rate": 1.9037679258138056e-05, "loss": 0.5961, "step": 4676 }, { "epoch": 0.16662807061296472, "grad_norm": 1.2840668843658014, "learning_rate": 1.9037185286726653e-05, "loss": 0.4753, "step": 4677 }, { "epoch": 0.16666369773945883, "grad_norm": 1.2340841739587245, "learning_rate": 1.9036691194978815e-05, "loss": 0.4986, "step": 4678 }, { "epoch": 0.16669932486595293, "grad_norm": 1.3019010910627542, "learning_rate": 1.9036196982901113e-05, "loss": 0.56, "step": 4679 }, { "epoch": 0.16673495199244706, "grad_norm": 2.892160353382638, "learning_rate": 1.9035702650500135e-05, "loss": 0.6544, "step": 4680 }, { "epoch": 0.16677057911894116, "grad_norm": 1.180025927573859, "learning_rate": 1.9035208197782456e-05, "loss": 0.5232, "step": 4681 }, { "epoch": 0.16680620624543527, "grad_norm": 1.3301643180265168, "learning_rate": 1.903471362475467e-05, "loss": 0.545, "step": 4682 }, { "epoch": 0.1668418333719294, "grad_norm": 1.7752595310302075, "learning_rate": 1.9034218931423356e-05, "loss": 0.6108, "step": 4683 }, { "epoch": 0.1668774604984235, "grad_norm": 1.3250666513597398, "learning_rate": 1.9033724117795103e-05, "loss": 0.5668, "step": 4684 }, { "epoch": 0.1669130876249176, "grad_norm": 1.22613874020595, "learning_rate": 1.90332291838765e-05, "loss": 0.5744, "step": 4685 }, { "epoch": 0.16694871475141174, "grad_norm": 1.9616653585975232, "learning_rate": 1.9032734129674132e-05, "loss": 0.6912, "step": 4686 }, { "epoch": 0.16698434187790584, "grad_norm": 2.252608317286909, "learning_rate": 1.90322389551946e-05, "loss": 0.6455, "step": 4687 }, { "epoch": 0.16701996900439994, "grad_norm": 2.1595634013664236, "learning_rate": 1.903174366044449e-05, "loss": 0.6961, "step": 4688 }, { "epoch": 0.16705559613089407, "grad_norm": 1.1608946444340782, "learning_rate": 1.90312482454304e-05, "loss": 0.5201, "step": 4689 }, { "epoch": 0.16709122325738818, "grad_norm": 1.315326785715424, "learning_rate": 1.9030752710158934e-05, "loss": 0.5837, "step": 4690 }, { "epoch": 0.16712685038388228, "grad_norm": 1.1118477978012742, "learning_rate": 1.9030257054636677e-05, "loss": 0.538, "step": 4691 }, { "epoch": 0.1671624775103764, "grad_norm": 1.549904537820907, "learning_rate": 1.9029761278870235e-05, "loss": 0.6159, "step": 4692 }, { "epoch": 0.16719810463687051, "grad_norm": 1.1580249049823335, "learning_rate": 1.9029265382866216e-05, "loss": 0.5628, "step": 4693 }, { "epoch": 0.16723373176336462, "grad_norm": 1.7481251511382494, "learning_rate": 1.902876936663121e-05, "loss": 0.6985, "step": 4694 }, { "epoch": 0.16726935888985875, "grad_norm": 1.1531421023151929, "learning_rate": 1.9028273230171833e-05, "loss": 0.4779, "step": 4695 }, { "epoch": 0.16730498601635285, "grad_norm": 1.4908837480118438, "learning_rate": 1.902777697349469e-05, "loss": 0.5428, "step": 4696 }, { "epoch": 0.16734061314284696, "grad_norm": 1.8884202090837132, "learning_rate": 1.902728059660638e-05, "loss": 0.6924, "step": 4697 }, { "epoch": 0.1673762402693411, "grad_norm": 1.8511585951421623, "learning_rate": 1.9026784099513525e-05, "loss": 0.6289, "step": 4698 }, { "epoch": 0.1674118673958352, "grad_norm": 1.2100445761711298, "learning_rate": 1.9026287482222727e-05, "loss": 0.5303, "step": 4699 }, { "epoch": 0.1674474945223293, "grad_norm": 1.2409934424762366, "learning_rate": 1.9025790744740606e-05, "loss": 0.4795, "step": 4700 }, { "epoch": 0.16748312164882342, "grad_norm": 1.7516689753175272, "learning_rate": 1.902529388707377e-05, "loss": 0.6114, "step": 4701 }, { "epoch": 0.16751874877531753, "grad_norm": 2.2280997663705278, "learning_rate": 1.9024796909228835e-05, "loss": 0.7141, "step": 4702 }, { "epoch": 0.16755437590181163, "grad_norm": 1.3199377147600493, "learning_rate": 1.9024299811212426e-05, "loss": 0.5091, "step": 4703 }, { "epoch": 0.16759000302830576, "grad_norm": 1.2188740808338279, "learning_rate": 1.9023802593031156e-05, "loss": 0.5172, "step": 4704 }, { "epoch": 0.16762563015479987, "grad_norm": 1.5987417579646443, "learning_rate": 1.9023305254691645e-05, "loss": 0.6886, "step": 4705 }, { "epoch": 0.16766125728129397, "grad_norm": 2.0854486869980278, "learning_rate": 1.902280779620052e-05, "loss": 0.6231, "step": 4706 }, { "epoch": 0.1676968844077881, "grad_norm": 1.323951307539664, "learning_rate": 1.90223102175644e-05, "loss": 0.534, "step": 4707 }, { "epoch": 0.1677325115342822, "grad_norm": 2.009161637194269, "learning_rate": 1.9021812518789917e-05, "loss": 0.5797, "step": 4708 }, { "epoch": 0.1677681386607763, "grad_norm": 1.726351637033224, "learning_rate": 1.9021314699883692e-05, "loss": 0.687, "step": 4709 }, { "epoch": 0.16780376578727044, "grad_norm": 1.3076657706507846, "learning_rate": 1.902081676085236e-05, "loss": 0.4871, "step": 4710 }, { "epoch": 0.16783939291376454, "grad_norm": 1.2158980495025857, "learning_rate": 1.902031870170254e-05, "loss": 0.5311, "step": 4711 }, { "epoch": 0.16787502004025864, "grad_norm": 1.407723787414255, "learning_rate": 1.9019820522440877e-05, "loss": 0.5381, "step": 4712 }, { "epoch": 0.16791064716675277, "grad_norm": 2.0843747014606606, "learning_rate": 1.9019322223073997e-05, "loss": 0.619, "step": 4713 }, { "epoch": 0.16794627429324688, "grad_norm": 1.851013847479164, "learning_rate": 1.901882380360854e-05, "loss": 0.6869, "step": 4714 }, { "epoch": 0.16798190141974098, "grad_norm": 2.1760859531265657, "learning_rate": 1.901832526405114e-05, "loss": 0.7191, "step": 4715 }, { "epoch": 0.1680175285462351, "grad_norm": 1.2847062218748775, "learning_rate": 1.9017826604408434e-05, "loss": 0.4366, "step": 4716 }, { "epoch": 0.16805315567272922, "grad_norm": 1.201915364193209, "learning_rate": 1.9017327824687065e-05, "loss": 0.5254, "step": 4717 }, { "epoch": 0.16808878279922332, "grad_norm": 1.2845729738431797, "learning_rate": 1.901682892489367e-05, "loss": 0.5328, "step": 4718 }, { "epoch": 0.16812440992571745, "grad_norm": 1.2331839961050841, "learning_rate": 1.90163299050349e-05, "loss": 0.5539, "step": 4719 }, { "epoch": 0.16816003705221155, "grad_norm": 1.3334600880177945, "learning_rate": 1.9015830765117394e-05, "loss": 0.5223, "step": 4720 }, { "epoch": 0.16819566417870566, "grad_norm": 1.6854275538643662, "learning_rate": 1.90153315051478e-05, "loss": 0.6466, "step": 4721 }, { "epoch": 0.1682312913051998, "grad_norm": 1.910894729813054, "learning_rate": 1.9014832125132764e-05, "loss": 0.5721, "step": 4722 }, { "epoch": 0.1682669184316939, "grad_norm": 1.1380889509249663, "learning_rate": 1.9014332625078937e-05, "loss": 0.4642, "step": 4723 }, { "epoch": 0.168302545558188, "grad_norm": 1.3552317307489608, "learning_rate": 1.9013833004992972e-05, "loss": 0.5652, "step": 4724 }, { "epoch": 0.16833817268468212, "grad_norm": 1.0987236214851852, "learning_rate": 1.901333326488152e-05, "loss": 0.4987, "step": 4725 }, { "epoch": 0.16837379981117623, "grad_norm": 1.7961341149558345, "learning_rate": 1.9012833404751237e-05, "loss": 0.5869, "step": 4726 }, { "epoch": 0.16840942693767033, "grad_norm": 1.7501147894441627, "learning_rate": 1.9012333424608775e-05, "loss": 0.6053, "step": 4727 }, { "epoch": 0.16844505406416446, "grad_norm": 2.213298053736709, "learning_rate": 1.9011833324460792e-05, "loss": 0.5421, "step": 4728 }, { "epoch": 0.16848068119065857, "grad_norm": 1.3620085150775203, "learning_rate": 1.9011333104313956e-05, "loss": 0.4856, "step": 4729 }, { "epoch": 0.16851630831715267, "grad_norm": 1.9430904861574565, "learning_rate": 1.9010832764174916e-05, "loss": 0.5873, "step": 4730 }, { "epoch": 0.1685519354436468, "grad_norm": 1.30381196246348, "learning_rate": 1.9010332304050344e-05, "loss": 0.4766, "step": 4731 }, { "epoch": 0.1685875625701409, "grad_norm": 1.4743956821927031, "learning_rate": 1.9009831723946894e-05, "loss": 0.5811, "step": 4732 }, { "epoch": 0.168623189696635, "grad_norm": 1.403633074239565, "learning_rate": 1.900933102387124e-05, "loss": 0.5739, "step": 4733 }, { "epoch": 0.16865881682312914, "grad_norm": 1.2181062282005084, "learning_rate": 1.9008830203830048e-05, "loss": 0.5837, "step": 4734 }, { "epoch": 0.16869444394962324, "grad_norm": 2.3053312668730763, "learning_rate": 1.9008329263829982e-05, "loss": 0.6643, "step": 4735 }, { "epoch": 0.16873007107611734, "grad_norm": 1.6488155332310326, "learning_rate": 1.9007828203877718e-05, "loss": 0.6481, "step": 4736 }, { "epoch": 0.16876569820261147, "grad_norm": 1.8048375500253382, "learning_rate": 1.9007327023979924e-05, "loss": 0.6355, "step": 4737 }, { "epoch": 0.16880132532910558, "grad_norm": 1.4770949815165786, "learning_rate": 1.9006825724143274e-05, "loss": 0.6497, "step": 4738 }, { "epoch": 0.16883695245559968, "grad_norm": 1.438701721425926, "learning_rate": 1.9006324304374442e-05, "loss": 0.4878, "step": 4739 }, { "epoch": 0.1688725795820938, "grad_norm": 1.2291198271017914, "learning_rate": 1.9005822764680112e-05, "loss": 0.5028, "step": 4740 }, { "epoch": 0.16890820670858792, "grad_norm": 1.8160688752516765, "learning_rate": 1.9005321105066956e-05, "loss": 0.5892, "step": 4741 }, { "epoch": 0.16894383383508202, "grad_norm": 1.2307029835046415, "learning_rate": 1.9004819325541653e-05, "loss": 0.5064, "step": 4742 }, { "epoch": 0.16897946096157615, "grad_norm": 1.6422769156764485, "learning_rate": 1.9004317426110888e-05, "loss": 0.6555, "step": 4743 }, { "epoch": 0.16901508808807025, "grad_norm": 1.6592681960385138, "learning_rate": 1.9003815406781342e-05, "loss": 0.5995, "step": 4744 }, { "epoch": 0.16905071521456436, "grad_norm": 1.157627128781241, "learning_rate": 1.9003313267559704e-05, "loss": 0.5031, "step": 4745 }, { "epoch": 0.1690863423410585, "grad_norm": 1.245634931438037, "learning_rate": 1.9002811008452654e-05, "loss": 0.5339, "step": 4746 }, { "epoch": 0.1691219694675526, "grad_norm": 1.8113079037165447, "learning_rate": 1.900230862946688e-05, "loss": 0.6128, "step": 4747 }, { "epoch": 0.1691575965940467, "grad_norm": 1.3552663280945167, "learning_rate": 1.900180613060908e-05, "loss": 0.5443, "step": 4748 }, { "epoch": 0.16919322372054083, "grad_norm": 3.273522288477894, "learning_rate": 1.900130351188594e-05, "loss": 0.665, "step": 4749 }, { "epoch": 0.16922885084703493, "grad_norm": 2.0142318235523686, "learning_rate": 1.9000800773304146e-05, "loss": 0.6146, "step": 4750 }, { "epoch": 0.16926447797352903, "grad_norm": 1.1884787255794937, "learning_rate": 1.90002979148704e-05, "loss": 0.4948, "step": 4751 }, { "epoch": 0.16930010510002316, "grad_norm": 1.889786246377487, "learning_rate": 1.8999794936591402e-05, "loss": 0.6594, "step": 4752 }, { "epoch": 0.16933573222651727, "grad_norm": 1.1798683740005558, "learning_rate": 1.899929183847384e-05, "loss": 0.5481, "step": 4753 }, { "epoch": 0.16937135935301137, "grad_norm": 1.298634675922988, "learning_rate": 1.8998788620524413e-05, "loss": 0.5428, "step": 4754 }, { "epoch": 0.1694069864795055, "grad_norm": 1.2677288325611484, "learning_rate": 1.8998285282749833e-05, "loss": 0.4962, "step": 4755 }, { "epoch": 0.1694426136059996, "grad_norm": 1.221526523379081, "learning_rate": 1.899778182515679e-05, "loss": 0.5878, "step": 4756 }, { "epoch": 0.16947824073249373, "grad_norm": 1.8675599341641562, "learning_rate": 1.8997278247751992e-05, "loss": 0.7125, "step": 4757 }, { "epoch": 0.16951386785898784, "grad_norm": 1.1064841541225863, "learning_rate": 1.899677455054215e-05, "loss": 0.5006, "step": 4758 }, { "epoch": 0.16954949498548194, "grad_norm": 1.5871015887085327, "learning_rate": 1.8996270733533965e-05, "loss": 0.6789, "step": 4759 }, { "epoch": 0.16958512211197607, "grad_norm": 1.291374239943088, "learning_rate": 1.8995766796734148e-05, "loss": 0.5315, "step": 4760 }, { "epoch": 0.16962074923847018, "grad_norm": 1.18152528149219, "learning_rate": 1.8995262740149407e-05, "loss": 0.5905, "step": 4761 }, { "epoch": 0.16965637636496428, "grad_norm": 1.094410510874954, "learning_rate": 1.8994758563786456e-05, "loss": 0.4992, "step": 4762 }, { "epoch": 0.1696920034914584, "grad_norm": 1.1792516536883884, "learning_rate": 1.8994254267652006e-05, "loss": 0.5046, "step": 4763 }, { "epoch": 0.1697276306179525, "grad_norm": 1.7541413126819438, "learning_rate": 1.8993749851752776e-05, "loss": 0.6349, "step": 4764 }, { "epoch": 0.16976325774444662, "grad_norm": 1.975095713211329, "learning_rate": 1.899324531609548e-05, "loss": 0.7101, "step": 4765 }, { "epoch": 0.16979888487094075, "grad_norm": 1.5539160230463234, "learning_rate": 1.899274066068684e-05, "loss": 0.6477, "step": 4766 }, { "epoch": 0.16983451199743485, "grad_norm": 1.3064162465438656, "learning_rate": 1.8992235885533566e-05, "loss": 0.4873, "step": 4767 }, { "epoch": 0.16987013912392895, "grad_norm": 1.7380815364638433, "learning_rate": 1.8991730990642388e-05, "loss": 0.5515, "step": 4768 }, { "epoch": 0.16990576625042308, "grad_norm": 1.8765317421470409, "learning_rate": 1.8991225976020026e-05, "loss": 0.5665, "step": 4769 }, { "epoch": 0.1699413933769172, "grad_norm": 1.781782696048035, "learning_rate": 1.8990720841673206e-05, "loss": 0.6451, "step": 4770 }, { "epoch": 0.1699770205034113, "grad_norm": 1.6584165138421374, "learning_rate": 1.8990215587608657e-05, "loss": 0.5544, "step": 4771 }, { "epoch": 0.17001264762990542, "grad_norm": 1.6044121277987882, "learning_rate": 1.8989710213833104e-05, "loss": 0.6333, "step": 4772 }, { "epoch": 0.17004827475639953, "grad_norm": 1.6978043554759585, "learning_rate": 1.8989204720353273e-05, "loss": 0.5503, "step": 4773 }, { "epoch": 0.17008390188289363, "grad_norm": 1.2170251919410287, "learning_rate": 1.89886991071759e-05, "loss": 0.5544, "step": 4774 }, { "epoch": 0.17011952900938776, "grad_norm": 1.330168206576543, "learning_rate": 1.8988193374307715e-05, "loss": 0.545, "step": 4775 }, { "epoch": 0.17015515613588186, "grad_norm": 1.758334713351735, "learning_rate": 1.8987687521755452e-05, "loss": 0.6197, "step": 4776 }, { "epoch": 0.17019078326237597, "grad_norm": 1.911680426399574, "learning_rate": 1.898718154952585e-05, "loss": 0.5991, "step": 4777 }, { "epoch": 0.1702264103888701, "grad_norm": 1.4068744269586901, "learning_rate": 1.898667545762564e-05, "loss": 0.5386, "step": 4778 }, { "epoch": 0.1702620375153642, "grad_norm": 2.83327309168853, "learning_rate": 1.8986169246061567e-05, "loss": 0.6696, "step": 4779 }, { "epoch": 0.1702976646418583, "grad_norm": 1.2156045611205508, "learning_rate": 1.898566291484037e-05, "loss": 0.4976, "step": 4780 }, { "epoch": 0.17033329176835244, "grad_norm": 3.739730260871289, "learning_rate": 1.898515646396879e-05, "loss": 0.6578, "step": 4781 }, { "epoch": 0.17036891889484654, "grad_norm": 2.238904206602122, "learning_rate": 1.8984649893453573e-05, "loss": 0.5161, "step": 4782 }, { "epoch": 0.17040454602134064, "grad_norm": 1.8984510353605557, "learning_rate": 1.898414320330146e-05, "loss": 0.6716, "step": 4783 }, { "epoch": 0.17044017314783477, "grad_norm": 1.6103654828007967, "learning_rate": 1.8983636393519202e-05, "loss": 0.6324, "step": 4784 }, { "epoch": 0.17047580027432888, "grad_norm": 1.3333601124373742, "learning_rate": 1.898312946411355e-05, "loss": 0.4935, "step": 4785 }, { "epoch": 0.17051142740082298, "grad_norm": 1.3435261694946523, "learning_rate": 1.8982622415091244e-05, "loss": 0.5168, "step": 4786 }, { "epoch": 0.1705470545273171, "grad_norm": 1.8153218432465525, "learning_rate": 1.8982115246459046e-05, "loss": 0.6159, "step": 4787 }, { "epoch": 0.1705826816538112, "grad_norm": 1.2178624915693934, "learning_rate": 1.8981607958223705e-05, "loss": 0.5498, "step": 4788 }, { "epoch": 0.17061830878030532, "grad_norm": 1.2847292556932588, "learning_rate": 1.8981100550391974e-05, "loss": 0.5217, "step": 4789 }, { "epoch": 0.17065393590679945, "grad_norm": 1.1662341250207748, "learning_rate": 1.8980593022970615e-05, "loss": 0.5642, "step": 4790 }, { "epoch": 0.17068956303329355, "grad_norm": 1.922606719608584, "learning_rate": 1.8980085375966382e-05, "loss": 0.6465, "step": 4791 }, { "epoch": 0.17072519015978765, "grad_norm": 1.177143066231072, "learning_rate": 1.8979577609386033e-05, "loss": 0.5418, "step": 4792 }, { "epoch": 0.17076081728628179, "grad_norm": 1.2206402519095596, "learning_rate": 1.8979069723236335e-05, "loss": 0.5191, "step": 4793 }, { "epoch": 0.1707964444127759, "grad_norm": 1.2849286609804533, "learning_rate": 1.8978561717524044e-05, "loss": 0.4936, "step": 4794 }, { "epoch": 0.17083207153927, "grad_norm": 1.1195622537319603, "learning_rate": 1.8978053592255935e-05, "loss": 0.5379, "step": 4795 }, { "epoch": 0.17086769866576412, "grad_norm": 1.308713309182047, "learning_rate": 1.8977545347438758e-05, "loss": 0.5253, "step": 4796 }, { "epoch": 0.17090332579225823, "grad_norm": 1.2974918688207793, "learning_rate": 1.8977036983079296e-05, "loss": 0.4928, "step": 4797 }, { "epoch": 0.17093895291875233, "grad_norm": 1.9173194279469663, "learning_rate": 1.897652849918431e-05, "loss": 0.6486, "step": 4798 }, { "epoch": 0.17097458004524646, "grad_norm": 1.25664338437147, "learning_rate": 1.8976019895760572e-05, "loss": 0.5369, "step": 4799 }, { "epoch": 0.17101020717174056, "grad_norm": 1.3532387736879294, "learning_rate": 1.897551117281486e-05, "loss": 0.5526, "step": 4800 }, { "epoch": 0.17104583429823467, "grad_norm": 1.4496027438285368, "learning_rate": 1.8975002330353938e-05, "loss": 0.584, "step": 4801 }, { "epoch": 0.1710814614247288, "grad_norm": 1.0698435320585598, "learning_rate": 1.8974493368384587e-05, "loss": 0.4832, "step": 4802 }, { "epoch": 0.1711170885512229, "grad_norm": 1.7489952188937568, "learning_rate": 1.8973984286913584e-05, "loss": 0.5776, "step": 4803 }, { "epoch": 0.171152715677717, "grad_norm": 1.8172946888386168, "learning_rate": 1.8973475085947714e-05, "loss": 0.6468, "step": 4804 }, { "epoch": 0.17118834280421114, "grad_norm": 2.2033006742665577, "learning_rate": 1.897296576549374e-05, "loss": 0.7242, "step": 4805 }, { "epoch": 0.17122396993070524, "grad_norm": 1.3236511868082903, "learning_rate": 1.8972456325558463e-05, "loss": 0.5637, "step": 4806 }, { "epoch": 0.17125959705719934, "grad_norm": 1.329540139390089, "learning_rate": 1.897194676614866e-05, "loss": 0.5839, "step": 4807 }, { "epoch": 0.17129522418369347, "grad_norm": 1.8188323451483082, "learning_rate": 1.8971437087271113e-05, "loss": 0.517, "step": 4808 }, { "epoch": 0.17133085131018758, "grad_norm": 1.1820669282324712, "learning_rate": 1.8970927288932607e-05, "loss": 0.5533, "step": 4809 }, { "epoch": 0.17136647843668168, "grad_norm": 1.6070683628784324, "learning_rate": 1.8970417371139934e-05, "loss": 0.6269, "step": 4810 }, { "epoch": 0.1714021055631758, "grad_norm": 1.9514835608095258, "learning_rate": 1.8969907333899885e-05, "loss": 0.6756, "step": 4811 }, { "epoch": 0.1714377326896699, "grad_norm": 1.2414677069123492, "learning_rate": 1.8969397177219252e-05, "loss": 0.5018, "step": 4812 }, { "epoch": 0.17147335981616402, "grad_norm": 1.163343656739594, "learning_rate": 1.896888690110483e-05, "loss": 0.5555, "step": 4813 }, { "epoch": 0.17150898694265815, "grad_norm": 2.281393687964241, "learning_rate": 1.8968376505563402e-05, "loss": 0.6785, "step": 4814 }, { "epoch": 0.17154461406915225, "grad_norm": 2.045456916147516, "learning_rate": 1.896786599060178e-05, "loss": 0.6476, "step": 4815 }, { "epoch": 0.17158024119564635, "grad_norm": 1.2937247530415241, "learning_rate": 1.8967355356226747e-05, "loss": 0.4941, "step": 4816 }, { "epoch": 0.17161586832214049, "grad_norm": 1.2048355247665659, "learning_rate": 1.8966844602445112e-05, "loss": 0.4964, "step": 4817 }, { "epoch": 0.1716514954486346, "grad_norm": 1.6265479788341752, "learning_rate": 1.8966333729263674e-05, "loss": 0.6112, "step": 4818 }, { "epoch": 0.1716871225751287, "grad_norm": 1.692051884657252, "learning_rate": 1.8965822736689238e-05, "loss": 0.5959, "step": 4819 }, { "epoch": 0.17172274970162282, "grad_norm": 1.3274049928652514, "learning_rate": 1.89653116247286e-05, "loss": 0.5249, "step": 4820 }, { "epoch": 0.17175837682811693, "grad_norm": 1.2050675903716146, "learning_rate": 1.8964800393388577e-05, "loss": 0.5195, "step": 4821 }, { "epoch": 0.17179400395461103, "grad_norm": 1.268403728953013, "learning_rate": 1.8964289042675967e-05, "loss": 0.5858, "step": 4822 }, { "epoch": 0.17182963108110516, "grad_norm": 1.3583903665486377, "learning_rate": 1.8963777572597583e-05, "loss": 0.5479, "step": 4823 }, { "epoch": 0.17186525820759926, "grad_norm": 1.6592938498762948, "learning_rate": 1.896326598316024e-05, "loss": 0.5929, "step": 4824 }, { "epoch": 0.17190088533409337, "grad_norm": 1.7096407547788548, "learning_rate": 1.896275427437074e-05, "loss": 0.619, "step": 4825 }, { "epoch": 0.1719365124605875, "grad_norm": 2.2707028388737216, "learning_rate": 1.8962242446235904e-05, "loss": 0.6199, "step": 4826 }, { "epoch": 0.1719721395870816, "grad_norm": 1.309943169578844, "learning_rate": 1.8961730498762544e-05, "loss": 0.4993, "step": 4827 }, { "epoch": 0.1720077667135757, "grad_norm": 1.9571602846360983, "learning_rate": 1.896121843195748e-05, "loss": 0.6011, "step": 4828 }, { "epoch": 0.17204339384006984, "grad_norm": 1.3130732288815592, "learning_rate": 1.896070624582753e-05, "loss": 0.5605, "step": 4829 }, { "epoch": 0.17207902096656394, "grad_norm": 1.281952223496561, "learning_rate": 1.896019394037951e-05, "loss": 0.5015, "step": 4830 }, { "epoch": 0.17211464809305804, "grad_norm": 1.1429661569821228, "learning_rate": 1.895968151562025e-05, "loss": 0.5293, "step": 4831 }, { "epoch": 0.17215027521955217, "grad_norm": 1.6134668454533312, "learning_rate": 1.8959168971556564e-05, "loss": 0.5314, "step": 4832 }, { "epoch": 0.17218590234604628, "grad_norm": 2.228900188339349, "learning_rate": 1.8958656308195284e-05, "loss": 0.6811, "step": 4833 }, { "epoch": 0.17222152947254038, "grad_norm": 1.9281746851114612, "learning_rate": 1.895814352554323e-05, "loss": 0.6794, "step": 4834 }, { "epoch": 0.1722571565990345, "grad_norm": 2.28967311334992, "learning_rate": 1.8957630623607236e-05, "loss": 0.6826, "step": 4835 }, { "epoch": 0.17229278372552861, "grad_norm": 1.3428229724182386, "learning_rate": 1.895711760239413e-05, "loss": 0.5498, "step": 4836 }, { "epoch": 0.17232841085202272, "grad_norm": 1.5511201897093019, "learning_rate": 1.895660446191074e-05, "loss": 0.5972, "step": 4837 }, { "epoch": 0.17236403797851685, "grad_norm": 1.6326761516409372, "learning_rate": 1.8956091202163903e-05, "loss": 0.5922, "step": 4838 }, { "epoch": 0.17239966510501095, "grad_norm": 1.8981035349093045, "learning_rate": 1.895557782316045e-05, "loss": 0.6733, "step": 4839 }, { "epoch": 0.17243529223150506, "grad_norm": 1.283263532671482, "learning_rate": 1.895506432490722e-05, "loss": 0.506, "step": 4840 }, { "epoch": 0.1724709193579992, "grad_norm": 1.1801021496214572, "learning_rate": 1.895455070741105e-05, "loss": 0.4973, "step": 4841 }, { "epoch": 0.1725065464844933, "grad_norm": 1.7625861690527411, "learning_rate": 1.895403697067878e-05, "loss": 0.6648, "step": 4842 }, { "epoch": 0.1725421736109874, "grad_norm": 1.5454110437616342, "learning_rate": 1.8953523114717247e-05, "loss": 0.6205, "step": 4843 }, { "epoch": 0.17257780073748152, "grad_norm": 2.2548929588324547, "learning_rate": 1.8953009139533297e-05, "loss": 0.6548, "step": 4844 }, { "epoch": 0.17261342786397563, "grad_norm": 1.2347682356560452, "learning_rate": 1.895249504513377e-05, "loss": 0.5472, "step": 4845 }, { "epoch": 0.17264905499046973, "grad_norm": 1.249309931946745, "learning_rate": 1.8951980831525515e-05, "loss": 0.4925, "step": 4846 }, { "epoch": 0.17268468211696386, "grad_norm": 1.7807930441690667, "learning_rate": 1.8951466498715378e-05, "loss": 0.6999, "step": 4847 }, { "epoch": 0.17272030924345796, "grad_norm": 1.7268103157108736, "learning_rate": 1.895095204671021e-05, "loss": 0.6695, "step": 4848 }, { "epoch": 0.17275593636995207, "grad_norm": 1.078882105697701, "learning_rate": 1.8950437475516855e-05, "loss": 0.506, "step": 4849 }, { "epoch": 0.1727915634964462, "grad_norm": 1.3172737521249454, "learning_rate": 1.8949922785142174e-05, "loss": 0.5829, "step": 4850 }, { "epoch": 0.1728271906229403, "grad_norm": 1.2052508714726933, "learning_rate": 1.894940797559301e-05, "loss": 0.5288, "step": 4851 }, { "epoch": 0.17286281774943443, "grad_norm": 1.9741225887106844, "learning_rate": 1.8948893046876228e-05, "loss": 0.6703, "step": 4852 }, { "epoch": 0.17289844487592854, "grad_norm": 2.0719824052088027, "learning_rate": 1.8948377998998678e-05, "loss": 0.6892, "step": 4853 }, { "epoch": 0.17293407200242264, "grad_norm": 2.4635680151138644, "learning_rate": 1.894786283196722e-05, "loss": 0.5558, "step": 4854 }, { "epoch": 0.17296969912891677, "grad_norm": 1.4002094203103255, "learning_rate": 1.8947347545788718e-05, "loss": 0.5244, "step": 4855 }, { "epoch": 0.17300532625541087, "grad_norm": 1.5316213712417808, "learning_rate": 1.8946832140470027e-05, "loss": 0.6025, "step": 4856 }, { "epoch": 0.17304095338190498, "grad_norm": 1.6261950105372653, "learning_rate": 1.894631661601801e-05, "loss": 0.6653, "step": 4857 }, { "epoch": 0.1730765805083991, "grad_norm": 1.5868321939839432, "learning_rate": 1.894580097243954e-05, "loss": 0.621, "step": 4858 }, { "epoch": 0.1731122076348932, "grad_norm": 1.687287782235256, "learning_rate": 1.8945285209741474e-05, "loss": 0.5405, "step": 4859 }, { "epoch": 0.17314783476138731, "grad_norm": 2.5586923422117005, "learning_rate": 1.894476932793068e-05, "loss": 0.7049, "step": 4860 }, { "epoch": 0.17318346188788145, "grad_norm": 1.1752180131828112, "learning_rate": 1.8944253327014036e-05, "loss": 0.508, "step": 4861 }, { "epoch": 0.17321908901437555, "grad_norm": 1.614237292066755, "learning_rate": 1.8943737206998404e-05, "loss": 0.5766, "step": 4862 }, { "epoch": 0.17325471614086965, "grad_norm": 2.1187011377592886, "learning_rate": 1.8943220967890657e-05, "loss": 0.5877, "step": 4863 }, { "epoch": 0.17329034326736378, "grad_norm": 2.3275342740769087, "learning_rate": 1.8942704609697678e-05, "loss": 0.5433, "step": 4864 }, { "epoch": 0.1733259703938579, "grad_norm": 1.2521785016376226, "learning_rate": 1.8942188132426334e-05, "loss": 0.5384, "step": 4865 }, { "epoch": 0.173361597520352, "grad_norm": 3.1592144551340824, "learning_rate": 1.8941671536083506e-05, "loss": 0.6147, "step": 4866 }, { "epoch": 0.17339722464684612, "grad_norm": 1.6837617214429952, "learning_rate": 1.894115482067607e-05, "loss": 0.6648, "step": 4867 }, { "epoch": 0.17343285177334022, "grad_norm": 1.8639196334403603, "learning_rate": 1.8940637986210907e-05, "loss": 0.5568, "step": 4868 }, { "epoch": 0.17346847889983433, "grad_norm": 1.3231909709385528, "learning_rate": 1.8940121032694902e-05, "loss": 0.4814, "step": 4869 }, { "epoch": 0.17350410602632846, "grad_norm": 1.4398714957268965, "learning_rate": 1.8939603960134935e-05, "loss": 0.5287, "step": 4870 }, { "epoch": 0.17353973315282256, "grad_norm": 1.1647641730033202, "learning_rate": 1.8939086768537894e-05, "loss": 0.5681, "step": 4871 }, { "epoch": 0.17357536027931667, "grad_norm": 1.9177952079674088, "learning_rate": 1.8938569457910664e-05, "loss": 0.6467, "step": 4872 }, { "epoch": 0.1736109874058108, "grad_norm": 2.4418215747969767, "learning_rate": 1.8938052028260136e-05, "loss": 0.6133, "step": 4873 }, { "epoch": 0.1736466145323049, "grad_norm": 1.2096821585126052, "learning_rate": 1.8937534479593195e-05, "loss": 0.5392, "step": 4874 }, { "epoch": 0.173682241658799, "grad_norm": 2.061084938052713, "learning_rate": 1.893701681191674e-05, "loss": 0.6043, "step": 4875 }, { "epoch": 0.17371786878529313, "grad_norm": 1.2175154230031537, "learning_rate": 1.8936499025237656e-05, "loss": 0.5319, "step": 4876 }, { "epoch": 0.17375349591178724, "grad_norm": 1.2048566124911155, "learning_rate": 1.8935981119562842e-05, "loss": 0.5142, "step": 4877 }, { "epoch": 0.17378912303828134, "grad_norm": 1.2840958593367107, "learning_rate": 1.8935463094899194e-05, "loss": 0.5487, "step": 4878 }, { "epoch": 0.17382475016477547, "grad_norm": 1.1947847763684696, "learning_rate": 1.8934944951253607e-05, "loss": 0.5514, "step": 4879 }, { "epoch": 0.17386037729126957, "grad_norm": 1.295656443491576, "learning_rate": 1.8934426688632986e-05, "loss": 0.5631, "step": 4880 }, { "epoch": 0.17389600441776368, "grad_norm": 1.9843912562888686, "learning_rate": 1.8933908307044223e-05, "loss": 0.6641, "step": 4881 }, { "epoch": 0.1739316315442578, "grad_norm": 1.6908924997759238, "learning_rate": 1.8933389806494232e-05, "loss": 0.6544, "step": 4882 }, { "epoch": 0.1739672586707519, "grad_norm": 1.9153895002720964, "learning_rate": 1.8932871186989913e-05, "loss": 0.6268, "step": 4883 }, { "epoch": 0.17400288579724602, "grad_norm": 1.2286682185140643, "learning_rate": 1.8932352448538166e-05, "loss": 0.5432, "step": 4884 }, { "epoch": 0.17403851292374015, "grad_norm": 1.4301655783432967, "learning_rate": 1.8931833591145904e-05, "loss": 0.5065, "step": 4885 }, { "epoch": 0.17407414005023425, "grad_norm": 1.4814808085859772, "learning_rate": 1.8931314614820038e-05, "loss": 0.5766, "step": 4886 }, { "epoch": 0.17410976717672835, "grad_norm": 2.5401809696611153, "learning_rate": 1.893079551956747e-05, "loss": 0.628, "step": 4887 }, { "epoch": 0.17414539430322248, "grad_norm": 1.410531015363289, "learning_rate": 1.8930276305395122e-05, "loss": 0.5842, "step": 4888 }, { "epoch": 0.1741810214297166, "grad_norm": 2.1419367707907364, "learning_rate": 1.8929756972309902e-05, "loss": 0.6001, "step": 4889 }, { "epoch": 0.1742166485562107, "grad_norm": 1.132860753614844, "learning_rate": 1.8929237520318725e-05, "loss": 0.5308, "step": 4890 }, { "epoch": 0.17425227568270482, "grad_norm": 1.5932150748227487, "learning_rate": 1.8928717949428508e-05, "loss": 0.5537, "step": 4891 }, { "epoch": 0.17428790280919892, "grad_norm": 1.5944847493862855, "learning_rate": 1.8928198259646177e-05, "loss": 0.5657, "step": 4892 }, { "epoch": 0.17432352993569303, "grad_norm": 1.2669948240335904, "learning_rate": 1.892767845097864e-05, "loss": 0.5374, "step": 4893 }, { "epoch": 0.17435915706218716, "grad_norm": 1.5173618055202867, "learning_rate": 1.892715852343282e-05, "loss": 0.6792, "step": 4894 }, { "epoch": 0.17439478418868126, "grad_norm": 1.9500229698218006, "learning_rate": 1.8926638477015654e-05, "loss": 0.658, "step": 4895 }, { "epoch": 0.17443041131517537, "grad_norm": 1.1763461196911502, "learning_rate": 1.8926118311734055e-05, "loss": 0.5197, "step": 4896 }, { "epoch": 0.1744660384416695, "grad_norm": 2.330069930257368, "learning_rate": 1.892559802759495e-05, "loss": 0.6307, "step": 4897 }, { "epoch": 0.1745016655681636, "grad_norm": 1.3017706821147568, "learning_rate": 1.8925077624605266e-05, "loss": 0.5173, "step": 4898 }, { "epoch": 0.1745372926946577, "grad_norm": 1.1854118552506254, "learning_rate": 1.892455710277194e-05, "loss": 0.5029, "step": 4899 }, { "epoch": 0.17457291982115183, "grad_norm": 1.0821928856581524, "learning_rate": 1.8924036462101896e-05, "loss": 0.4803, "step": 4900 }, { "epoch": 0.17460854694764594, "grad_norm": 1.9549706852005484, "learning_rate": 1.892351570260207e-05, "loss": 0.6469, "step": 4901 }, { "epoch": 0.17464417407414004, "grad_norm": 1.1514605023918512, "learning_rate": 1.8922994824279394e-05, "loss": 0.4917, "step": 4902 }, { "epoch": 0.17467980120063417, "grad_norm": 1.2554417543050589, "learning_rate": 1.8922473827140805e-05, "loss": 0.5478, "step": 4903 }, { "epoch": 0.17471542832712827, "grad_norm": 1.2923137677963565, "learning_rate": 1.892195271119324e-05, "loss": 0.5621, "step": 4904 }, { "epoch": 0.17475105545362238, "grad_norm": 2.059485268247408, "learning_rate": 1.8921431476443642e-05, "loss": 0.6259, "step": 4905 }, { "epoch": 0.1747866825801165, "grad_norm": 1.299127314093873, "learning_rate": 1.892091012289895e-05, "loss": 0.5265, "step": 4906 }, { "epoch": 0.1748223097066106, "grad_norm": 1.0922046980683997, "learning_rate": 1.8920388650566094e-05, "loss": 0.4816, "step": 4907 }, { "epoch": 0.17485793683310472, "grad_norm": 4.949921032439596, "learning_rate": 1.8919867059452038e-05, "loss": 0.6248, "step": 4908 }, { "epoch": 0.17489356395959885, "grad_norm": 1.867359893022514, "learning_rate": 1.8919345349563714e-05, "loss": 0.571, "step": 4909 }, { "epoch": 0.17492919108609295, "grad_norm": 1.7335195617760402, "learning_rate": 1.8918823520908073e-05, "loss": 0.6482, "step": 4910 }, { "epoch": 0.17496481821258705, "grad_norm": 1.8564925688223581, "learning_rate": 1.891830157349206e-05, "loss": 0.6131, "step": 4911 }, { "epoch": 0.17500044533908118, "grad_norm": 1.6054750275200813, "learning_rate": 1.891777950732263e-05, "loss": 0.5804, "step": 4912 }, { "epoch": 0.1750360724655753, "grad_norm": 1.8118531076868123, "learning_rate": 1.8917257322406735e-05, "loss": 0.6493, "step": 4913 }, { "epoch": 0.1750716995920694, "grad_norm": 1.6908069195923765, "learning_rate": 1.891673501875132e-05, "loss": 0.6475, "step": 4914 }, { "epoch": 0.17510732671856352, "grad_norm": 2.1099717274497305, "learning_rate": 1.8916212596363355e-05, "loss": 0.6092, "step": 4915 }, { "epoch": 0.17514295384505763, "grad_norm": 2.291797776689377, "learning_rate": 1.891569005524978e-05, "loss": 0.6449, "step": 4916 }, { "epoch": 0.17517858097155173, "grad_norm": 1.620675849931295, "learning_rate": 1.8915167395417563e-05, "loss": 0.6161, "step": 4917 }, { "epoch": 0.17521420809804586, "grad_norm": 2.0444988141018428, "learning_rate": 1.891464461687366e-05, "loss": 0.6137, "step": 4918 }, { "epoch": 0.17524983522453996, "grad_norm": 1.8747842050029149, "learning_rate": 1.8914121719625033e-05, "loss": 0.5213, "step": 4919 }, { "epoch": 0.17528546235103407, "grad_norm": 2.2898512223859497, "learning_rate": 1.8913598703678648e-05, "loss": 0.6492, "step": 4920 }, { "epoch": 0.1753210894775282, "grad_norm": 1.7565591643893594, "learning_rate": 1.891307556904146e-05, "loss": 0.5981, "step": 4921 }, { "epoch": 0.1753567166040223, "grad_norm": 1.5792362522478196, "learning_rate": 1.8912552315720446e-05, "loss": 0.4995, "step": 4922 }, { "epoch": 0.1753923437305164, "grad_norm": 2.852476887662085, "learning_rate": 1.8912028943722562e-05, "loss": 0.5721, "step": 4923 }, { "epoch": 0.17542797085701053, "grad_norm": 1.3898080959347607, "learning_rate": 1.891150545305479e-05, "loss": 0.5905, "step": 4924 }, { "epoch": 0.17546359798350464, "grad_norm": 1.221826664207122, "learning_rate": 1.891098184372409e-05, "loss": 0.5429, "step": 4925 }, { "epoch": 0.17549922510999874, "grad_norm": 1.1522349249238792, "learning_rate": 1.891045811573744e-05, "loss": 0.5183, "step": 4926 }, { "epoch": 0.17553485223649287, "grad_norm": 1.766545094906626, "learning_rate": 1.890993426910181e-05, "loss": 0.6547, "step": 4927 }, { "epoch": 0.17557047936298698, "grad_norm": 2.3441780863941175, "learning_rate": 1.8909410303824182e-05, "loss": 0.6777, "step": 4928 }, { "epoch": 0.17560610648948108, "grad_norm": 1.6012345334668545, "learning_rate": 1.8908886219911528e-05, "loss": 0.5076, "step": 4929 }, { "epoch": 0.1756417336159752, "grad_norm": 1.2523167346399673, "learning_rate": 1.8908362017370826e-05, "loss": 0.5268, "step": 4930 }, { "epoch": 0.1756773607424693, "grad_norm": 1.796369361593007, "learning_rate": 1.8907837696209055e-05, "loss": 0.6265, "step": 4931 }, { "epoch": 0.17571298786896342, "grad_norm": 1.2900509688283825, "learning_rate": 1.89073132564332e-05, "loss": 0.5557, "step": 4932 }, { "epoch": 0.17574861499545755, "grad_norm": 2.3090156826521904, "learning_rate": 1.8906788698050244e-05, "loss": 0.6199, "step": 4933 }, { "epoch": 0.17578424212195165, "grad_norm": 2.0354975630933216, "learning_rate": 1.890626402106717e-05, "loss": 0.5646, "step": 4934 }, { "epoch": 0.17581986924844575, "grad_norm": 1.2817057558779819, "learning_rate": 1.8905739225490966e-05, "loss": 0.5158, "step": 4935 }, { "epoch": 0.17585549637493988, "grad_norm": 2.170976496987667, "learning_rate": 1.890521431132862e-05, "loss": 0.6642, "step": 4936 }, { "epoch": 0.175891123501434, "grad_norm": 3.0049003141838893, "learning_rate": 1.890468927858712e-05, "loss": 0.6241, "step": 4937 }, { "epoch": 0.1759267506279281, "grad_norm": 1.1183158432189766, "learning_rate": 1.890416412727346e-05, "loss": 0.5201, "step": 4938 }, { "epoch": 0.17596237775442222, "grad_norm": 2.00642800903345, "learning_rate": 1.8903638857394627e-05, "loss": 0.6616, "step": 4939 }, { "epoch": 0.17599800488091633, "grad_norm": 1.5109404215866016, "learning_rate": 1.8903113468957624e-05, "loss": 0.6072, "step": 4940 }, { "epoch": 0.17603363200741043, "grad_norm": 1.2631705573165228, "learning_rate": 1.8902587961969442e-05, "loss": 0.567, "step": 4941 }, { "epoch": 0.17606925913390456, "grad_norm": 1.6937726295105033, "learning_rate": 1.8902062336437078e-05, "loss": 0.5586, "step": 4942 }, { "epoch": 0.17610488626039866, "grad_norm": 1.1261995069115232, "learning_rate": 1.890153659236753e-05, "loss": 0.5111, "step": 4943 }, { "epoch": 0.17614051338689277, "grad_norm": 1.2446366410799568, "learning_rate": 1.89010107297678e-05, "loss": 0.4871, "step": 4944 }, { "epoch": 0.1761761405133869, "grad_norm": 1.0443112906240106, "learning_rate": 1.8900484748644893e-05, "loss": 0.5491, "step": 4945 }, { "epoch": 0.176211767639881, "grad_norm": 1.8520423272218123, "learning_rate": 1.889995864900581e-05, "loss": 0.6172, "step": 4946 }, { "epoch": 0.17624739476637513, "grad_norm": 1.7092399666411995, "learning_rate": 1.8899432430857554e-05, "loss": 0.545, "step": 4947 }, { "epoch": 0.17628302189286923, "grad_norm": 2.091383447912544, "learning_rate": 1.889890609420714e-05, "loss": 0.6314, "step": 4948 }, { "epoch": 0.17631864901936334, "grad_norm": 1.7040580387926367, "learning_rate": 1.889837963906157e-05, "loss": 0.6057, "step": 4949 }, { "epoch": 0.17635427614585747, "grad_norm": 1.688056164764789, "learning_rate": 1.8897853065427853e-05, "loss": 0.6149, "step": 4950 }, { "epoch": 0.17638990327235157, "grad_norm": 1.846993705911217, "learning_rate": 1.8897326373313004e-05, "loss": 0.5995, "step": 4951 }, { "epoch": 0.17642553039884568, "grad_norm": 1.2008145464205031, "learning_rate": 1.8896799562724037e-05, "loss": 0.5107, "step": 4952 }, { "epoch": 0.1764611575253398, "grad_norm": 2.148289168834477, "learning_rate": 1.889627263366796e-05, "loss": 0.7945, "step": 4953 }, { "epoch": 0.1764967846518339, "grad_norm": 1.1111245712022169, "learning_rate": 1.8895745586151803e-05, "loss": 0.4874, "step": 4954 }, { "epoch": 0.176532411778328, "grad_norm": 2.1084277040351855, "learning_rate": 1.889521842018257e-05, "loss": 0.6707, "step": 4955 }, { "epoch": 0.17656803890482214, "grad_norm": 1.8169213894403253, "learning_rate": 1.889469113576729e-05, "loss": 0.5931, "step": 4956 }, { "epoch": 0.17660366603131625, "grad_norm": 1.6514329104933738, "learning_rate": 1.889416373291298e-05, "loss": 0.6168, "step": 4957 }, { "epoch": 0.17663929315781035, "grad_norm": 1.2734368221092707, "learning_rate": 1.889363621162666e-05, "loss": 0.517, "step": 4958 }, { "epoch": 0.17667492028430448, "grad_norm": 2.014686133507579, "learning_rate": 1.8893108571915357e-05, "loss": 0.6141, "step": 4959 }, { "epoch": 0.17671054741079859, "grad_norm": 1.2193861437923457, "learning_rate": 1.8892580813786102e-05, "loss": 0.5021, "step": 4960 }, { "epoch": 0.1767461745372927, "grad_norm": 1.598053898288802, "learning_rate": 1.8892052937245914e-05, "loss": 0.6624, "step": 4961 }, { "epoch": 0.17678180166378682, "grad_norm": 1.6772258047315156, "learning_rate": 1.889152494230183e-05, "loss": 0.5914, "step": 4962 }, { "epoch": 0.17681742879028092, "grad_norm": 1.4800956034442059, "learning_rate": 1.8890996828960874e-05, "loss": 0.5095, "step": 4963 }, { "epoch": 0.17685305591677503, "grad_norm": 1.4969693954177903, "learning_rate": 1.889046859723008e-05, "loss": 0.5332, "step": 4964 }, { "epoch": 0.17688868304326916, "grad_norm": 1.1940084146548282, "learning_rate": 1.888994024711648e-05, "loss": 0.505, "step": 4965 }, { "epoch": 0.17692431016976326, "grad_norm": 2.472473285716985, "learning_rate": 1.888941177862712e-05, "loss": 0.6808, "step": 4966 }, { "epoch": 0.17695993729625736, "grad_norm": 1.9073885487906455, "learning_rate": 1.8888883191769027e-05, "loss": 0.569, "step": 4967 }, { "epoch": 0.1769955644227515, "grad_norm": 1.6838540166740277, "learning_rate": 1.8888354486549238e-05, "loss": 0.6678, "step": 4968 }, { "epoch": 0.1770311915492456, "grad_norm": 1.2143958512112627, "learning_rate": 1.8887825662974796e-05, "loss": 0.5607, "step": 4969 }, { "epoch": 0.1770668186757397, "grad_norm": 1.6415017313474896, "learning_rate": 1.8887296721052745e-05, "loss": 0.6501, "step": 4970 }, { "epoch": 0.17710244580223383, "grad_norm": 1.380061977363261, "learning_rate": 1.8886767660790128e-05, "loss": 0.5251, "step": 4971 }, { "epoch": 0.17713807292872794, "grad_norm": 1.150417017425866, "learning_rate": 1.8886238482193983e-05, "loss": 0.5257, "step": 4972 }, { "epoch": 0.17717370005522204, "grad_norm": 1.1290756233080703, "learning_rate": 1.8885709185271363e-05, "loss": 0.5365, "step": 4973 }, { "epoch": 0.17720932718171617, "grad_norm": 1.807452065345885, "learning_rate": 1.8885179770029317e-05, "loss": 0.6682, "step": 4974 }, { "epoch": 0.17724495430821027, "grad_norm": 2.098606142737782, "learning_rate": 1.888465023647489e-05, "loss": 0.6209, "step": 4975 }, { "epoch": 0.17728058143470438, "grad_norm": 1.1709082655384955, "learning_rate": 1.8884120584615136e-05, "loss": 0.4843, "step": 4976 }, { "epoch": 0.1773162085611985, "grad_norm": 1.6546056729839398, "learning_rate": 1.8883590814457106e-05, "loss": 0.5357, "step": 4977 }, { "epoch": 0.1773518356876926, "grad_norm": 1.231436363858393, "learning_rate": 1.8883060926007856e-05, "loss": 0.4972, "step": 4978 }, { "epoch": 0.1773874628141867, "grad_norm": 1.1987794537529712, "learning_rate": 1.8882530919274442e-05, "loss": 0.5328, "step": 4979 }, { "epoch": 0.17742308994068084, "grad_norm": 1.2976481490524157, "learning_rate": 1.8882000794263916e-05, "loss": 0.5506, "step": 4980 }, { "epoch": 0.17745871706717495, "grad_norm": 1.222975244667231, "learning_rate": 1.8881470550983345e-05, "loss": 0.5182, "step": 4981 }, { "epoch": 0.17749434419366905, "grad_norm": 1.2619777764282412, "learning_rate": 1.8880940189439786e-05, "loss": 0.5725, "step": 4982 }, { "epoch": 0.17752997132016318, "grad_norm": 2.8401560577987213, "learning_rate": 1.88804097096403e-05, "loss": 0.5628, "step": 4983 }, { "epoch": 0.17756559844665729, "grad_norm": 1.1499052672896188, "learning_rate": 1.8879879111591954e-05, "loss": 0.5116, "step": 4984 }, { "epoch": 0.1776012255731514, "grad_norm": 1.6641427062626117, "learning_rate": 1.8879348395301805e-05, "loss": 0.5975, "step": 4985 }, { "epoch": 0.17763685269964552, "grad_norm": 2.210609146458586, "learning_rate": 1.8878817560776932e-05, "loss": 0.6138, "step": 4986 }, { "epoch": 0.17767247982613962, "grad_norm": 1.7361327102960815, "learning_rate": 1.8878286608024394e-05, "loss": 0.6149, "step": 4987 }, { "epoch": 0.17770810695263373, "grad_norm": 1.1321365650700241, "learning_rate": 1.8877755537051263e-05, "loss": 0.5005, "step": 4988 }, { "epoch": 0.17774373407912786, "grad_norm": 1.43706532757652, "learning_rate": 1.8877224347864612e-05, "loss": 0.5194, "step": 4989 }, { "epoch": 0.17777936120562196, "grad_norm": 1.434293757745752, "learning_rate": 1.8876693040471518e-05, "loss": 0.5832, "step": 4990 }, { "epoch": 0.17781498833211606, "grad_norm": 2.191322508589129, "learning_rate": 1.887616161487905e-05, "loss": 0.6527, "step": 4991 }, { "epoch": 0.1778506154586102, "grad_norm": 2.446100160770864, "learning_rate": 1.8875630071094285e-05, "loss": 0.6681, "step": 4992 }, { "epoch": 0.1778862425851043, "grad_norm": 1.2214616941217686, "learning_rate": 1.8875098409124302e-05, "loss": 0.4698, "step": 4993 }, { "epoch": 0.1779218697115984, "grad_norm": 1.8593716188755673, "learning_rate": 1.887456662897618e-05, "loss": 0.6345, "step": 4994 }, { "epoch": 0.17795749683809253, "grad_norm": 1.7380977543758662, "learning_rate": 1.8874034730657002e-05, "loss": 0.5757, "step": 4995 }, { "epoch": 0.17799312396458664, "grad_norm": 1.2392558912492657, "learning_rate": 1.8873502714173848e-05, "loss": 0.5533, "step": 4996 }, { "epoch": 0.17802875109108074, "grad_norm": 2.0155065171508184, "learning_rate": 1.8872970579533805e-05, "loss": 0.6079, "step": 4997 }, { "epoch": 0.17806437821757487, "grad_norm": 1.237365225650685, "learning_rate": 1.8872438326743955e-05, "loss": 0.5681, "step": 4998 }, { "epoch": 0.17810000534406897, "grad_norm": 2.005809919026549, "learning_rate": 1.8871905955811387e-05, "loss": 0.6022, "step": 4999 }, { "epoch": 0.17813563247056308, "grad_norm": 1.2231204359263759, "learning_rate": 1.887137346674319e-05, "loss": 0.491, "step": 5000 }, { "epoch": 0.1781712595970572, "grad_norm": 1.6413192283866127, "learning_rate": 1.8870840859546455e-05, "loss": 0.6605, "step": 5001 }, { "epoch": 0.1782068867235513, "grad_norm": 1.6729787433493555, "learning_rate": 1.8870308134228274e-05, "loss": 0.609, "step": 5002 }, { "epoch": 0.17824251385004541, "grad_norm": 1.898617169666074, "learning_rate": 1.886977529079574e-05, "loss": 0.6003, "step": 5003 }, { "epoch": 0.17827814097653955, "grad_norm": 1.3020578276824741, "learning_rate": 1.886924232925595e-05, "loss": 0.5545, "step": 5004 }, { "epoch": 0.17831376810303365, "grad_norm": 2.095763007002203, "learning_rate": 1.8868709249615994e-05, "loss": 0.5828, "step": 5005 }, { "epoch": 0.17834939522952775, "grad_norm": 1.1551767667798813, "learning_rate": 1.886817605188298e-05, "loss": 0.5034, "step": 5006 }, { "epoch": 0.17838502235602188, "grad_norm": 1.7192043233198173, "learning_rate": 1.8867642736064e-05, "loss": 0.6266, "step": 5007 }, { "epoch": 0.17842064948251599, "grad_norm": 1.6585465129358554, "learning_rate": 1.8867109302166165e-05, "loss": 0.551, "step": 5008 }, { "epoch": 0.1784562766090101, "grad_norm": 2.123055578268898, "learning_rate": 1.8866575750196568e-05, "loss": 0.5959, "step": 5009 }, { "epoch": 0.17849190373550422, "grad_norm": 1.8278108595990512, "learning_rate": 1.8866042080162317e-05, "loss": 0.5743, "step": 5010 }, { "epoch": 0.17852753086199832, "grad_norm": 1.234929374676052, "learning_rate": 1.886550829207052e-05, "loss": 0.5404, "step": 5011 }, { "epoch": 0.17856315798849243, "grad_norm": 1.67843064117365, "learning_rate": 1.8864974385928284e-05, "loss": 0.646, "step": 5012 }, { "epoch": 0.17859878511498656, "grad_norm": 1.246235563545352, "learning_rate": 1.8864440361742714e-05, "loss": 0.5773, "step": 5013 }, { "epoch": 0.17863441224148066, "grad_norm": 1.1718499894795837, "learning_rate": 1.886390621952093e-05, "loss": 0.5512, "step": 5014 }, { "epoch": 0.17867003936797476, "grad_norm": 1.8268843404442834, "learning_rate": 1.8863371959270034e-05, "loss": 0.6395, "step": 5015 }, { "epoch": 0.1787056664944689, "grad_norm": 1.7549331420861676, "learning_rate": 1.8862837580997147e-05, "loss": 0.6949, "step": 5016 }, { "epoch": 0.178741293620963, "grad_norm": 1.289437378529345, "learning_rate": 1.8862303084709385e-05, "loss": 0.5178, "step": 5017 }, { "epoch": 0.1787769207474571, "grad_norm": 2.073584334004015, "learning_rate": 1.886176847041386e-05, "loss": 0.6426, "step": 5018 }, { "epoch": 0.17881254787395123, "grad_norm": 2.490954829136024, "learning_rate": 1.8861233738117695e-05, "loss": 0.6555, "step": 5019 }, { "epoch": 0.17884817500044534, "grad_norm": 1.6918973277000502, "learning_rate": 1.8860698887828014e-05, "loss": 0.677, "step": 5020 }, { "epoch": 0.17888380212693944, "grad_norm": 1.6278400350215523, "learning_rate": 1.8860163919551925e-05, "loss": 0.6861, "step": 5021 }, { "epoch": 0.17891942925343357, "grad_norm": 1.7676774761165193, "learning_rate": 1.8859628833296568e-05, "loss": 0.765, "step": 5022 }, { "epoch": 0.17895505637992767, "grad_norm": 1.715376068375444, "learning_rate": 1.8859093629069057e-05, "loss": 0.6861, "step": 5023 }, { "epoch": 0.17899068350642178, "grad_norm": 1.8345966784037846, "learning_rate": 1.8858558306876524e-05, "loss": 0.6543, "step": 5024 }, { "epoch": 0.1790263106329159, "grad_norm": 1.773869502777536, "learning_rate": 1.8858022866726092e-05, "loss": 0.5972, "step": 5025 }, { "epoch": 0.17906193775941, "grad_norm": 1.5866687660300425, "learning_rate": 1.88574873086249e-05, "loss": 0.6994, "step": 5026 }, { "epoch": 0.17909756488590411, "grad_norm": 1.140677627647634, "learning_rate": 1.885695163258007e-05, "loss": 0.4912, "step": 5027 }, { "epoch": 0.17913319201239825, "grad_norm": 1.7722926780798287, "learning_rate": 1.8856415838598738e-05, "loss": 0.6097, "step": 5028 }, { "epoch": 0.17916881913889235, "grad_norm": 1.2342032434210775, "learning_rate": 1.885587992668804e-05, "loss": 0.5599, "step": 5029 }, { "epoch": 0.17920444626538645, "grad_norm": 1.8195683048361018, "learning_rate": 1.8855343896855107e-05, "loss": 0.6551, "step": 5030 }, { "epoch": 0.17924007339188058, "grad_norm": 1.4798436505848889, "learning_rate": 1.8854807749107086e-05, "loss": 0.6125, "step": 5031 }, { "epoch": 0.1792757005183747, "grad_norm": 1.2452851491304682, "learning_rate": 1.8854271483451106e-05, "loss": 0.521, "step": 5032 }, { "epoch": 0.1793113276448688, "grad_norm": 1.088209638033856, "learning_rate": 1.8853735099894317e-05, "loss": 0.4645, "step": 5033 }, { "epoch": 0.17934695477136292, "grad_norm": 1.3155529710207863, "learning_rate": 1.885319859844385e-05, "loss": 0.5779, "step": 5034 }, { "epoch": 0.17938258189785702, "grad_norm": 2.0654102727508192, "learning_rate": 1.8852661979106864e-05, "loss": 0.6305, "step": 5035 }, { "epoch": 0.17941820902435113, "grad_norm": 1.317916177114593, "learning_rate": 1.8852125241890492e-05, "loss": 0.6126, "step": 5036 }, { "epoch": 0.17945383615084526, "grad_norm": 1.4932994501506929, "learning_rate": 1.8851588386801886e-05, "loss": 0.4988, "step": 5037 }, { "epoch": 0.17948946327733936, "grad_norm": 1.9129764494424928, "learning_rate": 1.885105141384819e-05, "loss": 0.616, "step": 5038 }, { "epoch": 0.17952509040383346, "grad_norm": 1.5485371304139361, "learning_rate": 1.8850514323036564e-05, "loss": 0.6565, "step": 5039 }, { "epoch": 0.1795607175303276, "grad_norm": 1.4790517200810192, "learning_rate": 1.884997711437415e-05, "loss": 0.6164, "step": 5040 }, { "epoch": 0.1795963446568217, "grad_norm": 1.1349441940903202, "learning_rate": 1.8849439787868105e-05, "loss": 0.5214, "step": 5041 }, { "epoch": 0.17963197178331583, "grad_norm": 2.0132052530939917, "learning_rate": 1.8848902343525583e-05, "loss": 0.6934, "step": 5042 }, { "epoch": 0.17966759890980993, "grad_norm": 1.2199019033765428, "learning_rate": 1.8848364781353744e-05, "loss": 0.5513, "step": 5043 }, { "epoch": 0.17970322603630404, "grad_norm": 1.115765554604521, "learning_rate": 1.884782710135974e-05, "loss": 0.5028, "step": 5044 }, { "epoch": 0.17973885316279817, "grad_norm": 1.0447372860981612, "learning_rate": 1.8847289303550738e-05, "loss": 0.4532, "step": 5045 }, { "epoch": 0.17977448028929227, "grad_norm": 1.4986812354243848, "learning_rate": 1.8846751387933893e-05, "loss": 0.6058, "step": 5046 }, { "epoch": 0.17981010741578637, "grad_norm": 1.6166206800811278, "learning_rate": 1.8846213354516368e-05, "loss": 0.6004, "step": 5047 }, { "epoch": 0.1798457345422805, "grad_norm": 1.2541740989298926, "learning_rate": 1.8845675203305332e-05, "loss": 0.5658, "step": 5048 }, { "epoch": 0.1798813616687746, "grad_norm": 1.2667725839560315, "learning_rate": 1.8845136934307944e-05, "loss": 0.5227, "step": 5049 }, { "epoch": 0.1799169887952687, "grad_norm": 1.6276780935575619, "learning_rate": 1.884459854753138e-05, "loss": 0.6822, "step": 5050 }, { "epoch": 0.17995261592176284, "grad_norm": 1.2616219663072543, "learning_rate": 1.8844060042982804e-05, "loss": 0.5424, "step": 5051 }, { "epoch": 0.17998824304825695, "grad_norm": 1.3701431807743583, "learning_rate": 1.8843521420669383e-05, "loss": 0.541, "step": 5052 }, { "epoch": 0.18002387017475105, "grad_norm": 1.2710877540122736, "learning_rate": 1.8842982680598295e-05, "loss": 0.4873, "step": 5053 }, { "epoch": 0.18005949730124518, "grad_norm": 1.3195632565651232, "learning_rate": 1.8842443822776714e-05, "loss": 0.5361, "step": 5054 }, { "epoch": 0.18009512442773928, "grad_norm": 1.3002091539397655, "learning_rate": 1.8841904847211808e-05, "loss": 0.5517, "step": 5055 }, { "epoch": 0.1801307515542334, "grad_norm": 1.3275084302476043, "learning_rate": 1.8841365753910765e-05, "loss": 0.5418, "step": 5056 }, { "epoch": 0.18016637868072752, "grad_norm": 2.0277445890794024, "learning_rate": 1.884082654288075e-05, "loss": 0.5574, "step": 5057 }, { "epoch": 0.18020200580722162, "grad_norm": 2.6495516679429545, "learning_rate": 1.8840287214128958e-05, "loss": 0.6348, "step": 5058 }, { "epoch": 0.18023763293371572, "grad_norm": 1.5965105000503992, "learning_rate": 1.883974776766256e-05, "loss": 0.5717, "step": 5059 }, { "epoch": 0.18027326006020986, "grad_norm": 1.2600831199283526, "learning_rate": 1.8839208203488738e-05, "loss": 0.5001, "step": 5060 }, { "epoch": 0.18030888718670396, "grad_norm": 1.2454537243658337, "learning_rate": 1.8838668521614685e-05, "loss": 0.5111, "step": 5061 }, { "epoch": 0.18034451431319806, "grad_norm": 1.320488149364646, "learning_rate": 1.8838128722047583e-05, "loss": 0.5233, "step": 5062 }, { "epoch": 0.1803801414396922, "grad_norm": 1.11115821394094, "learning_rate": 1.8837588804794614e-05, "loss": 0.5414, "step": 5063 }, { "epoch": 0.1804157685661863, "grad_norm": 1.4828120418776594, "learning_rate": 1.8837048769862978e-05, "loss": 0.648, "step": 5064 }, { "epoch": 0.1804513956926804, "grad_norm": 1.5582566051891764, "learning_rate": 1.883650861725986e-05, "loss": 0.648, "step": 5065 }, { "epoch": 0.18048702281917453, "grad_norm": 1.2584693431922291, "learning_rate": 1.8835968346992454e-05, "loss": 0.5136, "step": 5066 }, { "epoch": 0.18052264994566863, "grad_norm": 1.6879725458680832, "learning_rate": 1.8835427959067952e-05, "loss": 0.5796, "step": 5067 }, { "epoch": 0.18055827707216274, "grad_norm": 1.5412945221956498, "learning_rate": 1.883488745349355e-05, "loss": 0.6449, "step": 5068 }, { "epoch": 0.18059390419865687, "grad_norm": 1.3222951800632095, "learning_rate": 1.883434683027645e-05, "loss": 0.5482, "step": 5069 }, { "epoch": 0.18062953132515097, "grad_norm": 1.3463181775734545, "learning_rate": 1.8833806089423845e-05, "loss": 0.5642, "step": 5070 }, { "epoch": 0.18066515845164507, "grad_norm": 1.534708274084972, "learning_rate": 1.8833265230942936e-05, "loss": 0.5629, "step": 5071 }, { "epoch": 0.1807007855781392, "grad_norm": 2.188194524892252, "learning_rate": 1.8832724254840927e-05, "loss": 0.6473, "step": 5072 }, { "epoch": 0.1807364127046333, "grad_norm": 1.7125020280712355, "learning_rate": 1.8832183161125026e-05, "loss": 0.5761, "step": 5073 }, { "epoch": 0.1807720398311274, "grad_norm": 1.7771574308568687, "learning_rate": 1.8831641949802425e-05, "loss": 0.5234, "step": 5074 }, { "epoch": 0.18080766695762154, "grad_norm": 1.2006511795345625, "learning_rate": 1.883110062088034e-05, "loss": 0.5267, "step": 5075 }, { "epoch": 0.18084329408411565, "grad_norm": 1.2411267492029272, "learning_rate": 1.8830559174365984e-05, "loss": 0.5229, "step": 5076 }, { "epoch": 0.18087892121060975, "grad_norm": 1.1797946097228216, "learning_rate": 1.8830017610266555e-05, "loss": 0.5418, "step": 5077 }, { "epoch": 0.18091454833710388, "grad_norm": 1.114149918745374, "learning_rate": 1.8829475928589272e-05, "loss": 0.5143, "step": 5078 }, { "epoch": 0.18095017546359798, "grad_norm": 1.261894759172897, "learning_rate": 1.8828934129341343e-05, "loss": 0.572, "step": 5079 }, { "epoch": 0.1809858025900921, "grad_norm": 1.0558477088762521, "learning_rate": 1.8828392212529988e-05, "loss": 0.5031, "step": 5080 }, { "epoch": 0.18102142971658622, "grad_norm": 1.154961238194245, "learning_rate": 1.8827850178162416e-05, "loss": 0.5515, "step": 5081 }, { "epoch": 0.18105705684308032, "grad_norm": 1.5878177395801576, "learning_rate": 1.882730802624585e-05, "loss": 0.6252, "step": 5082 }, { "epoch": 0.18109268396957443, "grad_norm": 1.1099575917688556, "learning_rate": 1.882676575678751e-05, "loss": 0.5109, "step": 5083 }, { "epoch": 0.18112831109606856, "grad_norm": 1.9315077633182387, "learning_rate": 1.8826223369794614e-05, "loss": 0.5851, "step": 5084 }, { "epoch": 0.18116393822256266, "grad_norm": 1.0134539649872336, "learning_rate": 1.8825680865274385e-05, "loss": 0.5192, "step": 5085 }, { "epoch": 0.18119956534905676, "grad_norm": 2.058829452169305, "learning_rate": 1.8825138243234047e-05, "loss": 0.672, "step": 5086 }, { "epoch": 0.1812351924755509, "grad_norm": 1.6341926785302188, "learning_rate": 1.8824595503680823e-05, "loss": 0.5641, "step": 5087 }, { "epoch": 0.181270819602045, "grad_norm": 1.6361704769401821, "learning_rate": 1.882405264662194e-05, "loss": 0.6937, "step": 5088 }, { "epoch": 0.1813064467285391, "grad_norm": 1.5770386296807142, "learning_rate": 1.882350967206463e-05, "loss": 0.6002, "step": 5089 }, { "epoch": 0.18134207385503323, "grad_norm": 1.7304940253795762, "learning_rate": 1.8822966580016124e-05, "loss": 0.5232, "step": 5090 }, { "epoch": 0.18137770098152733, "grad_norm": 1.2457548664106215, "learning_rate": 1.8822423370483647e-05, "loss": 0.5237, "step": 5091 }, { "epoch": 0.18141332810802144, "grad_norm": 1.2000946365898753, "learning_rate": 1.8821880043474443e-05, "loss": 0.463, "step": 5092 }, { "epoch": 0.18144895523451557, "grad_norm": 1.2010565812415275, "learning_rate": 1.882133659899573e-05, "loss": 0.5204, "step": 5093 }, { "epoch": 0.18148458236100967, "grad_norm": 1.2164021656965678, "learning_rate": 1.882079303705476e-05, "loss": 0.5132, "step": 5094 }, { "epoch": 0.18152020948750378, "grad_norm": 1.5531703305702325, "learning_rate": 1.8820249357658768e-05, "loss": 0.5851, "step": 5095 }, { "epoch": 0.1815558366139979, "grad_norm": 1.0600855800902234, "learning_rate": 1.8819705560814987e-05, "loss": 0.455, "step": 5096 }, { "epoch": 0.181591463740492, "grad_norm": 2.0742445679254815, "learning_rate": 1.8819161646530665e-05, "loss": 0.6139, "step": 5097 }, { "epoch": 0.1816270908669861, "grad_norm": 1.7093663884749506, "learning_rate": 1.8818617614813037e-05, "loss": 0.5849, "step": 5098 }, { "epoch": 0.18166271799348024, "grad_norm": 1.167272146550467, "learning_rate": 1.8818073465669354e-05, "loss": 0.5141, "step": 5099 }, { "epoch": 0.18169834511997435, "grad_norm": 1.718691307734114, "learning_rate": 1.8817529199106858e-05, "loss": 0.6784, "step": 5100 }, { "epoch": 0.18173397224646845, "grad_norm": 1.4484438070424936, "learning_rate": 1.88169848151328e-05, "loss": 0.6376, "step": 5101 }, { "epoch": 0.18176959937296258, "grad_norm": 1.2184371336346336, "learning_rate": 1.8816440313754425e-05, "loss": 0.5451, "step": 5102 }, { "epoch": 0.18180522649945668, "grad_norm": 1.8553686934943676, "learning_rate": 1.8815895694978987e-05, "loss": 0.5755, "step": 5103 }, { "epoch": 0.1818408536259508, "grad_norm": 1.6119809877377265, "learning_rate": 1.8815350958813733e-05, "loss": 0.6857, "step": 5104 }, { "epoch": 0.18187648075244492, "grad_norm": 1.2431068037025936, "learning_rate": 1.881480610526592e-05, "loss": 0.514, "step": 5105 }, { "epoch": 0.18191210787893902, "grad_norm": 2.239460754745298, "learning_rate": 1.8814261134342804e-05, "loss": 0.6295, "step": 5106 }, { "epoch": 0.18194773500543313, "grad_norm": 1.7755570389183355, "learning_rate": 1.881371604605164e-05, "loss": 0.5662, "step": 5107 }, { "epoch": 0.18198336213192726, "grad_norm": 1.3348804942149568, "learning_rate": 1.8813170840399686e-05, "loss": 0.5485, "step": 5108 }, { "epoch": 0.18201898925842136, "grad_norm": 1.1655102015028544, "learning_rate": 1.88126255173942e-05, "loss": 0.5123, "step": 5109 }, { "epoch": 0.18205461638491546, "grad_norm": 1.0884129271087941, "learning_rate": 1.881208007704245e-05, "loss": 0.4869, "step": 5110 }, { "epoch": 0.1820902435114096, "grad_norm": 1.1123872521500242, "learning_rate": 1.881153451935169e-05, "loss": 0.5407, "step": 5111 }, { "epoch": 0.1821258706379037, "grad_norm": 1.1515382530545493, "learning_rate": 1.881098884432919e-05, "loss": 0.5272, "step": 5112 }, { "epoch": 0.1821614977643978, "grad_norm": 2.1166415388836906, "learning_rate": 1.8810443051982215e-05, "loss": 0.6094, "step": 5113 }, { "epoch": 0.18219712489089193, "grad_norm": 1.1599461218368692, "learning_rate": 1.8809897142318038e-05, "loss": 0.5103, "step": 5114 }, { "epoch": 0.18223275201738603, "grad_norm": 1.115538098507793, "learning_rate": 1.8809351115343918e-05, "loss": 0.541, "step": 5115 }, { "epoch": 0.18226837914388014, "grad_norm": 1.7333847567212848, "learning_rate": 1.8808804971067132e-05, "loss": 0.6294, "step": 5116 }, { "epoch": 0.18230400627037427, "grad_norm": 1.1380500312109751, "learning_rate": 1.8808258709494945e-05, "loss": 0.5108, "step": 5117 }, { "epoch": 0.18233963339686837, "grad_norm": 1.17647595903063, "learning_rate": 1.8807712330634645e-05, "loss": 0.4894, "step": 5118 }, { "epoch": 0.18237526052336248, "grad_norm": 1.7710563824153585, "learning_rate": 1.8807165834493495e-05, "loss": 0.6225, "step": 5119 }, { "epoch": 0.1824108876498566, "grad_norm": 1.2382125323597235, "learning_rate": 1.8806619221078776e-05, "loss": 0.5322, "step": 5120 }, { "epoch": 0.1824465147763507, "grad_norm": 1.2799266941687923, "learning_rate": 1.8806072490397768e-05, "loss": 0.5303, "step": 5121 }, { "epoch": 0.1824821419028448, "grad_norm": 1.8017779121464597, "learning_rate": 1.880552564245775e-05, "loss": 0.7184, "step": 5122 }, { "epoch": 0.18251776902933894, "grad_norm": 1.698013602372227, "learning_rate": 1.8804978677266003e-05, "loss": 0.5505, "step": 5123 }, { "epoch": 0.18255339615583305, "grad_norm": 1.7212801982449004, "learning_rate": 1.880443159482981e-05, "loss": 0.6074, "step": 5124 }, { "epoch": 0.18258902328232715, "grad_norm": 1.1616779845890235, "learning_rate": 1.8803884395156456e-05, "loss": 0.5222, "step": 5125 }, { "epoch": 0.18262465040882128, "grad_norm": 1.1109413289113677, "learning_rate": 1.8803337078253228e-05, "loss": 0.4912, "step": 5126 }, { "epoch": 0.18266027753531539, "grad_norm": 1.2328740615227083, "learning_rate": 1.8802789644127413e-05, "loss": 0.5803, "step": 5127 }, { "epoch": 0.1826959046618095, "grad_norm": 1.1273094468325715, "learning_rate": 1.8802242092786302e-05, "loss": 0.4811, "step": 5128 }, { "epoch": 0.18273153178830362, "grad_norm": 1.80272563540858, "learning_rate": 1.8801694424237184e-05, "loss": 0.6937, "step": 5129 }, { "epoch": 0.18276715891479772, "grad_norm": 1.7446962602082212, "learning_rate": 1.8801146638487355e-05, "loss": 0.5881, "step": 5130 }, { "epoch": 0.18280278604129183, "grad_norm": 1.2449233155769814, "learning_rate": 1.8800598735544105e-05, "loss": 0.4853, "step": 5131 }, { "epoch": 0.18283841316778596, "grad_norm": 1.7805006422779945, "learning_rate": 1.8800050715414733e-05, "loss": 0.651, "step": 5132 }, { "epoch": 0.18287404029428006, "grad_norm": 1.1901220156482084, "learning_rate": 1.8799502578106533e-05, "loss": 0.5471, "step": 5133 }, { "epoch": 0.1829096674207742, "grad_norm": 1.192952303790719, "learning_rate": 1.8798954323626808e-05, "loss": 0.5036, "step": 5134 }, { "epoch": 0.1829452945472683, "grad_norm": 1.9879298626637096, "learning_rate": 1.8798405951982853e-05, "loss": 0.6092, "step": 5135 }, { "epoch": 0.1829809216737624, "grad_norm": 1.2481619380626996, "learning_rate": 1.8797857463181977e-05, "loss": 0.5119, "step": 5136 }, { "epoch": 0.18301654880025653, "grad_norm": 1.1705829269301815, "learning_rate": 1.8797308857231477e-05, "loss": 0.5132, "step": 5137 }, { "epoch": 0.18305217592675063, "grad_norm": 1.2200391841499894, "learning_rate": 1.879676013413866e-05, "loss": 0.4986, "step": 5138 }, { "epoch": 0.18308780305324474, "grad_norm": 1.749773817414072, "learning_rate": 1.8796211293910836e-05, "loss": 0.5863, "step": 5139 }, { "epoch": 0.18312343017973887, "grad_norm": 1.1398474455584582, "learning_rate": 1.8795662336555307e-05, "loss": 0.4716, "step": 5140 }, { "epoch": 0.18315905730623297, "grad_norm": 1.3083043908061993, "learning_rate": 1.8795113262079388e-05, "loss": 0.5376, "step": 5141 }, { "epoch": 0.18319468443272707, "grad_norm": 1.2022718649441142, "learning_rate": 1.879456407049039e-05, "loss": 0.5471, "step": 5142 }, { "epoch": 0.1832303115592212, "grad_norm": 1.180827203670101, "learning_rate": 1.8794014761795624e-05, "loss": 0.518, "step": 5143 }, { "epoch": 0.1832659386857153, "grad_norm": 2.0015363027190842, "learning_rate": 1.87934653360024e-05, "loss": 0.5789, "step": 5144 }, { "epoch": 0.1833015658122094, "grad_norm": 2.7893952386969865, "learning_rate": 1.8792915793118042e-05, "loss": 0.6478, "step": 5145 }, { "epoch": 0.18333719293870354, "grad_norm": 1.3812566642651947, "learning_rate": 1.8792366133149866e-05, "loss": 0.5958, "step": 5146 }, { "epoch": 0.18337282006519764, "grad_norm": 1.2547918020675175, "learning_rate": 1.8791816356105186e-05, "loss": 0.5554, "step": 5147 }, { "epoch": 0.18340844719169175, "grad_norm": 1.208521959388233, "learning_rate": 1.879126646199133e-05, "loss": 0.5119, "step": 5148 }, { "epoch": 0.18344407431818588, "grad_norm": 1.9892385431389763, "learning_rate": 1.8790716450815614e-05, "loss": 0.7093, "step": 5149 }, { "epoch": 0.18347970144467998, "grad_norm": 1.2733456498768798, "learning_rate": 1.8790166322585365e-05, "loss": 0.5474, "step": 5150 }, { "epoch": 0.18351532857117409, "grad_norm": 1.092613913634157, "learning_rate": 1.8789616077307905e-05, "loss": 0.5135, "step": 5151 }, { "epoch": 0.18355095569766822, "grad_norm": 1.1703511293447448, "learning_rate": 1.8789065714990567e-05, "loss": 0.4753, "step": 5152 }, { "epoch": 0.18358658282416232, "grad_norm": 1.2649312950825917, "learning_rate": 1.8788515235640675e-05, "loss": 0.5321, "step": 5153 }, { "epoch": 0.18362220995065642, "grad_norm": 1.4733655953692557, "learning_rate": 1.878796463926556e-05, "loss": 0.5896, "step": 5154 }, { "epoch": 0.18365783707715055, "grad_norm": 1.3062905881301399, "learning_rate": 1.878741392587255e-05, "loss": 0.5374, "step": 5155 }, { "epoch": 0.18369346420364466, "grad_norm": 1.2461185561297834, "learning_rate": 1.8786863095468985e-05, "loss": 0.5348, "step": 5156 }, { "epoch": 0.18372909133013876, "grad_norm": 1.2618980930033772, "learning_rate": 1.8786312148062196e-05, "loss": 0.5084, "step": 5157 }, { "epoch": 0.1837647184566329, "grad_norm": 1.6926543823336326, "learning_rate": 1.878576108365952e-05, "loss": 0.6633, "step": 5158 }, { "epoch": 0.183800345583127, "grad_norm": 1.3379939166462669, "learning_rate": 1.878520990226829e-05, "loss": 0.5454, "step": 5159 }, { "epoch": 0.1838359727096211, "grad_norm": 1.2060686279145303, "learning_rate": 1.8784658603895855e-05, "loss": 0.5183, "step": 5160 }, { "epoch": 0.18387159983611523, "grad_norm": 1.150057351609181, "learning_rate": 1.878410718854955e-05, "loss": 0.5192, "step": 5161 }, { "epoch": 0.18390722696260933, "grad_norm": 1.572032489848806, "learning_rate": 1.8783555656236714e-05, "loss": 0.6352, "step": 5162 }, { "epoch": 0.18394285408910344, "grad_norm": 1.2297337041999337, "learning_rate": 1.87830040069647e-05, "loss": 0.5019, "step": 5163 }, { "epoch": 0.18397848121559757, "grad_norm": 1.2174868432886985, "learning_rate": 1.8782452240740845e-05, "loss": 0.5516, "step": 5164 }, { "epoch": 0.18401410834209167, "grad_norm": 1.6803019744519285, "learning_rate": 1.87819003575725e-05, "loss": 0.6022, "step": 5165 }, { "epoch": 0.18404973546858577, "grad_norm": 1.215173917616972, "learning_rate": 1.8781348357467017e-05, "loss": 0.582, "step": 5166 }, { "epoch": 0.1840853625950799, "grad_norm": 2.16322715098405, "learning_rate": 1.878079624043174e-05, "loss": 0.5653, "step": 5167 }, { "epoch": 0.184120989721574, "grad_norm": 1.8577319696903818, "learning_rate": 1.878024400647402e-05, "loss": 0.6367, "step": 5168 }, { "epoch": 0.1841566168480681, "grad_norm": 1.2145711487448947, "learning_rate": 1.877969165560122e-05, "loss": 0.5097, "step": 5169 }, { "epoch": 0.18419224397456224, "grad_norm": 2.500741391714967, "learning_rate": 1.8779139187820687e-05, "loss": 0.5684, "step": 5170 }, { "epoch": 0.18422787110105635, "grad_norm": 1.1937422457300524, "learning_rate": 1.877858660313978e-05, "loss": 0.5047, "step": 5171 }, { "epoch": 0.18426349822755045, "grad_norm": 2.0217557274065494, "learning_rate": 1.8778033901565856e-05, "loss": 0.551, "step": 5172 }, { "epoch": 0.18429912535404458, "grad_norm": 1.2718844047491762, "learning_rate": 1.8777481083106274e-05, "loss": 0.5158, "step": 5173 }, { "epoch": 0.18433475248053868, "grad_norm": 1.907258635242494, "learning_rate": 1.8776928147768397e-05, "loss": 0.5902, "step": 5174 }, { "epoch": 0.18437037960703279, "grad_norm": 2.1099447505086797, "learning_rate": 1.8776375095559588e-05, "loss": 0.5307, "step": 5175 }, { "epoch": 0.18440600673352692, "grad_norm": 1.8761566697166563, "learning_rate": 1.877582192648721e-05, "loss": 0.631, "step": 5176 }, { "epoch": 0.18444163386002102, "grad_norm": 1.5820782846125883, "learning_rate": 1.877526864055863e-05, "loss": 0.6507, "step": 5177 }, { "epoch": 0.18447726098651512, "grad_norm": 2.412956950965338, "learning_rate": 1.8774715237781213e-05, "loss": 0.7244, "step": 5178 }, { "epoch": 0.18451288811300925, "grad_norm": 1.7708276914281433, "learning_rate": 1.877416171816233e-05, "loss": 0.5915, "step": 5179 }, { "epoch": 0.18454851523950336, "grad_norm": 1.393620367544013, "learning_rate": 1.8773608081709353e-05, "loss": 0.55, "step": 5180 }, { "epoch": 0.18458414236599746, "grad_norm": 2.100313158861609, "learning_rate": 1.8773054328429645e-05, "loss": 0.6285, "step": 5181 }, { "epoch": 0.1846197694924916, "grad_norm": 1.7190840926598865, "learning_rate": 1.8772500458330595e-05, "loss": 0.4848, "step": 5182 }, { "epoch": 0.1846553966189857, "grad_norm": 2.792070602581105, "learning_rate": 1.8771946471419563e-05, "loss": 0.5129, "step": 5183 }, { "epoch": 0.1846910237454798, "grad_norm": 3.2439448870551106, "learning_rate": 1.8771392367703937e-05, "loss": 0.5773, "step": 5184 }, { "epoch": 0.18472665087197393, "grad_norm": 1.6242433031015748, "learning_rate": 1.877083814719109e-05, "loss": 0.5048, "step": 5185 }, { "epoch": 0.18476227799846803, "grad_norm": 1.3349706988855612, "learning_rate": 1.87702838098884e-05, "loss": 0.4956, "step": 5186 }, { "epoch": 0.18479790512496214, "grad_norm": 2.4789146167807745, "learning_rate": 1.8769729355803254e-05, "loss": 0.5047, "step": 5187 }, { "epoch": 0.18483353225145627, "grad_norm": 2.0579437400586436, "learning_rate": 1.8769174784943032e-05, "loss": 0.5276, "step": 5188 }, { "epoch": 0.18486915937795037, "grad_norm": 1.8318367170625147, "learning_rate": 1.8768620097315117e-05, "loss": 0.5457, "step": 5189 }, { "epoch": 0.18490478650444447, "grad_norm": 2.140294716066824, "learning_rate": 1.8768065292926895e-05, "loss": 0.5907, "step": 5190 }, { "epoch": 0.1849404136309386, "grad_norm": 1.2163309123157, "learning_rate": 1.876751037178576e-05, "loss": 0.5549, "step": 5191 }, { "epoch": 0.1849760407574327, "grad_norm": 1.404022672941758, "learning_rate": 1.8766955333899092e-05, "loss": 0.5511, "step": 5192 }, { "epoch": 0.1850116678839268, "grad_norm": 1.8223095826564744, "learning_rate": 1.8766400179274287e-05, "loss": 0.5891, "step": 5193 }, { "epoch": 0.18504729501042094, "grad_norm": 1.3695410345890227, "learning_rate": 1.876584490791874e-05, "loss": 0.4933, "step": 5194 }, { "epoch": 0.18508292213691505, "grad_norm": 1.7288090527835565, "learning_rate": 1.8765289519839832e-05, "loss": 0.619, "step": 5195 }, { "epoch": 0.18511854926340915, "grad_norm": 2.0856515738750168, "learning_rate": 1.8764734015044975e-05, "loss": 0.5957, "step": 5196 }, { "epoch": 0.18515417638990328, "grad_norm": 1.189632146338779, "learning_rate": 1.8764178393541555e-05, "loss": 0.5398, "step": 5197 }, { "epoch": 0.18518980351639738, "grad_norm": 1.1282455120823118, "learning_rate": 1.876362265533698e-05, "loss": 0.5022, "step": 5198 }, { "epoch": 0.1852254306428915, "grad_norm": 1.2250781762575627, "learning_rate": 1.8763066800438638e-05, "loss": 0.5354, "step": 5199 }, { "epoch": 0.18526105776938562, "grad_norm": 1.2969790325477777, "learning_rate": 1.8762510828853936e-05, "loss": 0.5115, "step": 5200 }, { "epoch": 0.18529668489587972, "grad_norm": 1.221720920854935, "learning_rate": 1.8761954740590283e-05, "loss": 0.555, "step": 5201 }, { "epoch": 0.18533231202237382, "grad_norm": 1.129021184179637, "learning_rate": 1.876139853565507e-05, "loss": 0.5305, "step": 5202 }, { "epoch": 0.18536793914886796, "grad_norm": 1.1240718594672596, "learning_rate": 1.876084221405572e-05, "loss": 0.5248, "step": 5203 }, { "epoch": 0.18540356627536206, "grad_norm": 1.1129143034920204, "learning_rate": 1.876028577579963e-05, "loss": 0.4859, "step": 5204 }, { "epoch": 0.18543919340185616, "grad_norm": 1.2931378912737228, "learning_rate": 1.875972922089421e-05, "loss": 0.5648, "step": 5205 }, { "epoch": 0.1854748205283503, "grad_norm": 1.6860559908529253, "learning_rate": 1.8759172549346872e-05, "loss": 0.696, "step": 5206 }, { "epoch": 0.1855104476548444, "grad_norm": 1.7727441583213093, "learning_rate": 1.8758615761165034e-05, "loss": 0.5826, "step": 5207 }, { "epoch": 0.1855460747813385, "grad_norm": 2.350402690541009, "learning_rate": 1.87580588563561e-05, "loss": 0.5982, "step": 5208 }, { "epoch": 0.18558170190783263, "grad_norm": 1.4954791674978143, "learning_rate": 1.8757501834927495e-05, "loss": 0.6285, "step": 5209 }, { "epoch": 0.18561732903432673, "grad_norm": 1.6263458605011394, "learning_rate": 1.875694469688663e-05, "loss": 0.6493, "step": 5210 }, { "epoch": 0.18565295616082084, "grad_norm": 1.1378680580291438, "learning_rate": 1.8756387442240923e-05, "loss": 0.5102, "step": 5211 }, { "epoch": 0.18568858328731497, "grad_norm": 1.1510357643995448, "learning_rate": 1.87558300709978e-05, "loss": 0.4766, "step": 5212 }, { "epoch": 0.18572421041380907, "grad_norm": 1.1402660450786541, "learning_rate": 1.8755272583164678e-05, "loss": 0.4944, "step": 5213 }, { "epoch": 0.18575983754030317, "grad_norm": 1.2487262806021622, "learning_rate": 1.8754714978748983e-05, "loss": 0.5356, "step": 5214 }, { "epoch": 0.1857954646667973, "grad_norm": 1.1857425942269122, "learning_rate": 1.875415725775814e-05, "loss": 0.521, "step": 5215 }, { "epoch": 0.1858310917932914, "grad_norm": 1.209310591767474, "learning_rate": 1.8753599420199574e-05, "loss": 0.5213, "step": 5216 }, { "epoch": 0.1858667189197855, "grad_norm": 1.510356375325163, "learning_rate": 1.8753041466080713e-05, "loss": 0.5366, "step": 5217 }, { "epoch": 0.18590234604627964, "grad_norm": 1.072577157391761, "learning_rate": 1.8752483395408987e-05, "loss": 0.5361, "step": 5218 }, { "epoch": 0.18593797317277375, "grad_norm": 1.2813638150348257, "learning_rate": 1.8751925208191822e-05, "loss": 0.5374, "step": 5219 }, { "epoch": 0.18597360029926785, "grad_norm": 1.221332038313165, "learning_rate": 1.8751366904436663e-05, "loss": 0.5527, "step": 5220 }, { "epoch": 0.18600922742576198, "grad_norm": 1.1973370395833134, "learning_rate": 1.8750808484150936e-05, "loss": 0.5419, "step": 5221 }, { "epoch": 0.18604485455225608, "grad_norm": 1.7966366003431178, "learning_rate": 1.8750249947342075e-05, "loss": 0.6838, "step": 5222 }, { "epoch": 0.1860804816787502, "grad_norm": 2.2898588841787224, "learning_rate": 1.874969129401752e-05, "loss": 0.6496, "step": 5223 }, { "epoch": 0.18611610880524432, "grad_norm": 1.8799889516150243, "learning_rate": 1.874913252418471e-05, "loss": 0.6508, "step": 5224 }, { "epoch": 0.18615173593173842, "grad_norm": 1.248239051394982, "learning_rate": 1.8748573637851085e-05, "loss": 0.5374, "step": 5225 }, { "epoch": 0.18618736305823252, "grad_norm": 1.531816222082417, "learning_rate": 1.8748014635024086e-05, "loss": 0.5908, "step": 5226 }, { "epoch": 0.18622299018472666, "grad_norm": 1.2507686723584257, "learning_rate": 1.8747455515711158e-05, "loss": 0.5601, "step": 5227 }, { "epoch": 0.18625861731122076, "grad_norm": 1.2695837746657637, "learning_rate": 1.8746896279919745e-05, "loss": 0.5127, "step": 5228 }, { "epoch": 0.1862942444377149, "grad_norm": 1.7704971434894567, "learning_rate": 1.87463369276573e-05, "loss": 0.5418, "step": 5229 }, { "epoch": 0.186329871564209, "grad_norm": 1.2879541794897047, "learning_rate": 1.874577745893126e-05, "loss": 0.5331, "step": 5230 }, { "epoch": 0.1863654986907031, "grad_norm": 1.1388083205348951, "learning_rate": 1.8745217873749083e-05, "loss": 0.5239, "step": 5231 }, { "epoch": 0.18640112581719723, "grad_norm": 1.1262696397288725, "learning_rate": 1.8744658172118215e-05, "loss": 0.5216, "step": 5232 }, { "epoch": 0.18643675294369133, "grad_norm": 1.504356278149893, "learning_rate": 1.874409835404611e-05, "loss": 0.551, "step": 5233 }, { "epoch": 0.18647238007018543, "grad_norm": 2.2228814819971654, "learning_rate": 1.8743538419540228e-05, "loss": 0.6517, "step": 5234 }, { "epoch": 0.18650800719667956, "grad_norm": 1.3174255932255639, "learning_rate": 1.874297836860802e-05, "loss": 0.5719, "step": 5235 }, { "epoch": 0.18654363432317367, "grad_norm": 1.2610557610958342, "learning_rate": 1.8742418201256944e-05, "loss": 0.5552, "step": 5236 }, { "epoch": 0.18657926144966777, "grad_norm": 1.8972118652370424, "learning_rate": 1.8741857917494454e-05, "loss": 0.6173, "step": 5237 }, { "epoch": 0.1866148885761619, "grad_norm": 1.8104947514741654, "learning_rate": 1.874129751732802e-05, "loss": 0.6256, "step": 5238 }, { "epoch": 0.186650515702656, "grad_norm": 2.161087731738996, "learning_rate": 1.87407370007651e-05, "loss": 0.647, "step": 5239 }, { "epoch": 0.1866861428291501, "grad_norm": 1.5388842690297102, "learning_rate": 1.8740176367813156e-05, "loss": 0.4852, "step": 5240 }, { "epoch": 0.18672176995564424, "grad_norm": 2.0994032122848076, "learning_rate": 1.8739615618479657e-05, "loss": 0.6074, "step": 5241 }, { "epoch": 0.18675739708213834, "grad_norm": 1.973365384941273, "learning_rate": 1.8739054752772064e-05, "loss": 0.5556, "step": 5242 }, { "epoch": 0.18679302420863245, "grad_norm": 1.5331382830061384, "learning_rate": 1.873849377069785e-05, "loss": 0.63, "step": 5243 }, { "epoch": 0.18682865133512658, "grad_norm": 1.5394405844343682, "learning_rate": 1.8737932672264487e-05, "loss": 0.543, "step": 5244 }, { "epoch": 0.18686427846162068, "grad_norm": 2.0104767260257574, "learning_rate": 1.8737371457479437e-05, "loss": 0.6122, "step": 5245 }, { "epoch": 0.18689990558811478, "grad_norm": 1.9685206036766045, "learning_rate": 1.8736810126350184e-05, "loss": 0.5362, "step": 5246 }, { "epoch": 0.18693553271460892, "grad_norm": 2.465308020782682, "learning_rate": 1.8736248678884195e-05, "loss": 0.6587, "step": 5247 }, { "epoch": 0.18697115984110302, "grad_norm": 1.1473566477599915, "learning_rate": 1.8735687115088947e-05, "loss": 0.4899, "step": 5248 }, { "epoch": 0.18700678696759712, "grad_norm": 1.785160783152978, "learning_rate": 1.873512543497192e-05, "loss": 0.5782, "step": 5249 }, { "epoch": 0.18704241409409125, "grad_norm": 1.0812827639071882, "learning_rate": 1.8734563638540596e-05, "loss": 0.5154, "step": 5250 }, { "epoch": 0.18707804122058536, "grad_norm": 1.5818946743272286, "learning_rate": 1.873400172580245e-05, "loss": 0.6908, "step": 5251 }, { "epoch": 0.18711366834707946, "grad_norm": 1.1754367262408696, "learning_rate": 1.8733439696764968e-05, "loss": 0.5704, "step": 5252 }, { "epoch": 0.1871492954735736, "grad_norm": 1.129136174506715, "learning_rate": 1.873287755143563e-05, "loss": 0.4893, "step": 5253 }, { "epoch": 0.1871849226000677, "grad_norm": 1.6586329046251413, "learning_rate": 1.873231528982192e-05, "loss": 0.6123, "step": 5254 }, { "epoch": 0.1872205497265618, "grad_norm": 1.588533276056353, "learning_rate": 1.8731752911931332e-05, "loss": 0.5817, "step": 5255 }, { "epoch": 0.18725617685305593, "grad_norm": 1.2203427893925696, "learning_rate": 1.873119041777135e-05, "loss": 0.5049, "step": 5256 }, { "epoch": 0.18729180397955003, "grad_norm": 1.6544693624315676, "learning_rate": 1.8730627807349468e-05, "loss": 0.6976, "step": 5257 }, { "epoch": 0.18732743110604413, "grad_norm": 1.735394275002518, "learning_rate": 1.8730065080673173e-05, "loss": 0.6158, "step": 5258 }, { "epoch": 0.18736305823253827, "grad_norm": 1.0569114496222822, "learning_rate": 1.872950223774996e-05, "loss": 0.4849, "step": 5259 }, { "epoch": 0.18739868535903237, "grad_norm": 2.0563151521823824, "learning_rate": 1.872893927858732e-05, "loss": 0.5625, "step": 5260 }, { "epoch": 0.18743431248552647, "grad_norm": 1.842578008995813, "learning_rate": 1.8728376203192755e-05, "loss": 0.6232, "step": 5261 }, { "epoch": 0.1874699396120206, "grad_norm": 1.0666660386418783, "learning_rate": 1.872781301157376e-05, "loss": 0.4569, "step": 5262 }, { "epoch": 0.1875055667385147, "grad_norm": 1.5888400982589352, "learning_rate": 1.8727249703737832e-05, "loss": 0.6857, "step": 5263 }, { "epoch": 0.1875411938650088, "grad_norm": 1.9658353818372547, "learning_rate": 1.8726686279692475e-05, "loss": 0.6218, "step": 5264 }, { "epoch": 0.18757682099150294, "grad_norm": 1.2494095584140872, "learning_rate": 1.872612273944519e-05, "loss": 0.5734, "step": 5265 }, { "epoch": 0.18761244811799704, "grad_norm": 1.996515108124984, "learning_rate": 1.8725559083003483e-05, "loss": 0.6344, "step": 5266 }, { "epoch": 0.18764807524449115, "grad_norm": 1.1803803010548444, "learning_rate": 1.872499531037486e-05, "loss": 0.5324, "step": 5267 }, { "epoch": 0.18768370237098528, "grad_norm": 1.1111051701253016, "learning_rate": 1.8724431421566822e-05, "loss": 0.5311, "step": 5268 }, { "epoch": 0.18771932949747938, "grad_norm": 2.108240980730888, "learning_rate": 1.8723867416586885e-05, "loss": 0.6972, "step": 5269 }, { "epoch": 0.18775495662397348, "grad_norm": 1.5934451577654112, "learning_rate": 1.8723303295442553e-05, "loss": 0.55, "step": 5270 }, { "epoch": 0.18779058375046762, "grad_norm": 1.2107374078807163, "learning_rate": 1.8722739058141344e-05, "loss": 0.4718, "step": 5271 }, { "epoch": 0.18782621087696172, "grad_norm": 1.301624331485224, "learning_rate": 1.8722174704690762e-05, "loss": 0.531, "step": 5272 }, { "epoch": 0.18786183800345582, "grad_norm": 1.2191828569848813, "learning_rate": 1.8721610235098332e-05, "loss": 0.5539, "step": 5273 }, { "epoch": 0.18789746512994995, "grad_norm": 1.4130569430204307, "learning_rate": 1.8721045649371567e-05, "loss": 0.4987, "step": 5274 }, { "epoch": 0.18793309225644406, "grad_norm": 1.5864526092496924, "learning_rate": 1.8720480947517977e-05, "loss": 0.617, "step": 5275 }, { "epoch": 0.18796871938293816, "grad_norm": 1.6186543804157956, "learning_rate": 1.8719916129545094e-05, "loss": 0.5442, "step": 5276 }, { "epoch": 0.1880043465094323, "grad_norm": 1.8667370059710875, "learning_rate": 1.871935119546043e-05, "loss": 0.5746, "step": 5277 }, { "epoch": 0.1880399736359264, "grad_norm": 1.3520100284516003, "learning_rate": 1.871878614527151e-05, "loss": 0.5485, "step": 5278 }, { "epoch": 0.1880756007624205, "grad_norm": 1.811425851101614, "learning_rate": 1.8718220978985862e-05, "loss": 0.5576, "step": 5279 }, { "epoch": 0.18811122788891463, "grad_norm": 1.2275480992725707, "learning_rate": 1.8717655696611004e-05, "loss": 0.536, "step": 5280 }, { "epoch": 0.18814685501540873, "grad_norm": 1.2464433395323056, "learning_rate": 1.871709029815447e-05, "loss": 0.5599, "step": 5281 }, { "epoch": 0.18818248214190283, "grad_norm": 1.1168239547439176, "learning_rate": 1.8716524783623785e-05, "loss": 0.5193, "step": 5282 }, { "epoch": 0.18821810926839697, "grad_norm": 3.756539715311806, "learning_rate": 1.8715959153026476e-05, "loss": 0.6058, "step": 5283 }, { "epoch": 0.18825373639489107, "grad_norm": 1.6637632422333364, "learning_rate": 1.8715393406370085e-05, "loss": 0.5602, "step": 5284 }, { "epoch": 0.18828936352138517, "grad_norm": 1.8375990136300466, "learning_rate": 1.8714827543662134e-05, "loss": 0.5559, "step": 5285 }, { "epoch": 0.1883249906478793, "grad_norm": 1.0463573998424958, "learning_rate": 1.8714261564910167e-05, "loss": 0.5008, "step": 5286 }, { "epoch": 0.1883606177743734, "grad_norm": 2.1882067721101053, "learning_rate": 1.8713695470121714e-05, "loss": 0.7296, "step": 5287 }, { "epoch": 0.1883962449008675, "grad_norm": 1.1591896107707107, "learning_rate": 1.8713129259304315e-05, "loss": 0.5395, "step": 5288 }, { "epoch": 0.18843187202736164, "grad_norm": 1.1629071799703299, "learning_rate": 1.871256293246551e-05, "loss": 0.4898, "step": 5289 }, { "epoch": 0.18846749915385574, "grad_norm": 1.576194292717252, "learning_rate": 1.8711996489612842e-05, "loss": 0.64, "step": 5290 }, { "epoch": 0.18850312628034985, "grad_norm": 1.5445208296616908, "learning_rate": 1.871142993075385e-05, "loss": 0.6145, "step": 5291 }, { "epoch": 0.18853875340684398, "grad_norm": 1.20316240256921, "learning_rate": 1.871086325589608e-05, "loss": 0.5217, "step": 5292 }, { "epoch": 0.18857438053333808, "grad_norm": 1.1037220119891542, "learning_rate": 1.8710296465047077e-05, "loss": 0.469, "step": 5293 }, { "epoch": 0.18861000765983219, "grad_norm": 1.7099968181742649, "learning_rate": 1.8709729558214388e-05, "loss": 0.5411, "step": 5294 }, { "epoch": 0.18864563478632632, "grad_norm": 1.1088999954563028, "learning_rate": 1.870916253540556e-05, "loss": 0.5023, "step": 5295 }, { "epoch": 0.18868126191282042, "grad_norm": 1.7690015017168337, "learning_rate": 1.870859539662815e-05, "loss": 0.5554, "step": 5296 }, { "epoch": 0.18871688903931452, "grad_norm": 1.1214794471114438, "learning_rate": 1.87080281418897e-05, "loss": 0.5671, "step": 5297 }, { "epoch": 0.18875251616580865, "grad_norm": 1.1638772108093334, "learning_rate": 1.8707460771197773e-05, "loss": 0.522, "step": 5298 }, { "epoch": 0.18878814329230276, "grad_norm": 2.0433194373583676, "learning_rate": 1.870689328455992e-05, "loss": 0.6333, "step": 5299 }, { "epoch": 0.18882377041879686, "grad_norm": 1.8528119427017806, "learning_rate": 1.8706325681983694e-05, "loss": 0.615, "step": 5300 }, { "epoch": 0.188859397545291, "grad_norm": 1.2153427946206108, "learning_rate": 1.870575796347666e-05, "loss": 0.489, "step": 5301 }, { "epoch": 0.1888950246717851, "grad_norm": 2.03250424259923, "learning_rate": 1.870519012904637e-05, "loss": 0.5448, "step": 5302 }, { "epoch": 0.1889306517982792, "grad_norm": 1.230942134566765, "learning_rate": 1.870462217870039e-05, "loss": 0.5345, "step": 5303 }, { "epoch": 0.18896627892477333, "grad_norm": 1.2203045240527162, "learning_rate": 1.870405411244628e-05, "loss": 0.5599, "step": 5304 }, { "epoch": 0.18900190605126743, "grad_norm": 1.9084833110151447, "learning_rate": 1.870348593029161e-05, "loss": 0.5754, "step": 5305 }, { "epoch": 0.18903753317776154, "grad_norm": 1.210804758531274, "learning_rate": 1.870291763224394e-05, "loss": 0.5323, "step": 5306 }, { "epoch": 0.18907316030425567, "grad_norm": 1.2925023693726636, "learning_rate": 1.870234921831084e-05, "loss": 0.5201, "step": 5307 }, { "epoch": 0.18910878743074977, "grad_norm": 1.8393350208343144, "learning_rate": 1.8701780688499872e-05, "loss": 0.6428, "step": 5308 }, { "epoch": 0.18914441455724387, "grad_norm": 1.7872114640945955, "learning_rate": 1.8701212042818616e-05, "loss": 0.608, "step": 5309 }, { "epoch": 0.189180041683738, "grad_norm": 1.4144797911146123, "learning_rate": 1.8700643281274644e-05, "loss": 0.559, "step": 5310 }, { "epoch": 0.1892156688102321, "grad_norm": 1.2126103183748185, "learning_rate": 1.8700074403875517e-05, "loss": 0.56, "step": 5311 }, { "epoch": 0.1892512959367262, "grad_norm": 1.1153295845707127, "learning_rate": 1.8699505410628823e-05, "loss": 0.4883, "step": 5312 }, { "epoch": 0.18928692306322034, "grad_norm": 2.2134909549117148, "learning_rate": 1.8698936301542133e-05, "loss": 0.6447, "step": 5313 }, { "epoch": 0.18932255018971444, "grad_norm": 1.7663610728790882, "learning_rate": 1.8698367076623024e-05, "loss": 0.7073, "step": 5314 }, { "epoch": 0.18935817731620855, "grad_norm": 1.1323183339792737, "learning_rate": 1.869779773587908e-05, "loss": 0.5283, "step": 5315 }, { "epoch": 0.18939380444270268, "grad_norm": 1.1335892410910235, "learning_rate": 1.869722827931788e-05, "loss": 0.4706, "step": 5316 }, { "epoch": 0.18942943156919678, "grad_norm": 1.735451932043234, "learning_rate": 1.8696658706947005e-05, "loss": 0.6315, "step": 5317 }, { "epoch": 0.18946505869569089, "grad_norm": 1.240513464435361, "learning_rate": 1.8696089018774043e-05, "loss": 0.568, "step": 5318 }, { "epoch": 0.18950068582218502, "grad_norm": 1.9447820594018892, "learning_rate": 1.8695519214806576e-05, "loss": 0.5699, "step": 5319 }, { "epoch": 0.18953631294867912, "grad_norm": 1.1665248639599402, "learning_rate": 1.869494929505219e-05, "loss": 0.5552, "step": 5320 }, { "epoch": 0.18957194007517322, "grad_norm": 1.241587817467892, "learning_rate": 1.8694379259518482e-05, "loss": 0.5564, "step": 5321 }, { "epoch": 0.18960756720166735, "grad_norm": 1.145465089199888, "learning_rate": 1.8693809108213035e-05, "loss": 0.5346, "step": 5322 }, { "epoch": 0.18964319432816146, "grad_norm": 1.104749645544654, "learning_rate": 1.8693238841143442e-05, "loss": 0.4937, "step": 5323 }, { "epoch": 0.1896788214546556, "grad_norm": 2.303256672654762, "learning_rate": 1.8692668458317297e-05, "loss": 0.5715, "step": 5324 }, { "epoch": 0.1897144485811497, "grad_norm": 1.2101847453863377, "learning_rate": 1.8692097959742196e-05, "loss": 0.4776, "step": 5325 }, { "epoch": 0.1897500757076438, "grad_norm": 2.148732172738108, "learning_rate": 1.8691527345425737e-05, "loss": 0.633, "step": 5326 }, { "epoch": 0.18978570283413793, "grad_norm": 2.36586570793448, "learning_rate": 1.869095661537552e-05, "loss": 0.6694, "step": 5327 }, { "epoch": 0.18982132996063203, "grad_norm": 1.9138913142587939, "learning_rate": 1.8690385769599135e-05, "loss": 0.6355, "step": 5328 }, { "epoch": 0.18985695708712613, "grad_norm": 1.7395350127689782, "learning_rate": 1.868981480810419e-05, "loss": 0.6531, "step": 5329 }, { "epoch": 0.18989258421362026, "grad_norm": 1.1964320833955242, "learning_rate": 1.8689243730898288e-05, "loss": 0.5185, "step": 5330 }, { "epoch": 0.18992821134011437, "grad_norm": 1.1242044230622172, "learning_rate": 1.8688672537989032e-05, "loss": 0.5264, "step": 5331 }, { "epoch": 0.18996383846660847, "grad_norm": 1.1740311480106043, "learning_rate": 1.8688101229384028e-05, "loss": 0.5358, "step": 5332 }, { "epoch": 0.1899994655931026, "grad_norm": 1.2274091434699712, "learning_rate": 1.8687529805090884e-05, "loss": 0.5022, "step": 5333 }, { "epoch": 0.1900350927195967, "grad_norm": 1.0987302485983121, "learning_rate": 1.8686958265117207e-05, "loss": 0.5007, "step": 5334 }, { "epoch": 0.1900707198460908, "grad_norm": 1.5758266644624797, "learning_rate": 1.868638660947061e-05, "loss": 0.5641, "step": 5335 }, { "epoch": 0.19010634697258494, "grad_norm": 2.1128158586848445, "learning_rate": 1.8685814838158703e-05, "loss": 0.6408, "step": 5336 }, { "epoch": 0.19014197409907904, "grad_norm": 1.228069885534676, "learning_rate": 1.8685242951189106e-05, "loss": 0.5152, "step": 5337 }, { "epoch": 0.19017760122557315, "grad_norm": 2.2669043822611585, "learning_rate": 1.8684670948569425e-05, "loss": 0.6053, "step": 5338 }, { "epoch": 0.19021322835206728, "grad_norm": 2.309164129350627, "learning_rate": 1.8684098830307276e-05, "loss": 0.5922, "step": 5339 }, { "epoch": 0.19024885547856138, "grad_norm": 1.1263190567838508, "learning_rate": 1.8683526596410284e-05, "loss": 0.4844, "step": 5340 }, { "epoch": 0.19028448260505548, "grad_norm": 1.9145628255021383, "learning_rate": 1.8682954246886067e-05, "loss": 0.6331, "step": 5341 }, { "epoch": 0.1903201097315496, "grad_norm": 1.1602714052651444, "learning_rate": 1.8682381781742246e-05, "loss": 0.5188, "step": 5342 }, { "epoch": 0.19035573685804372, "grad_norm": 1.643613703793897, "learning_rate": 1.868180920098644e-05, "loss": 0.6013, "step": 5343 }, { "epoch": 0.19039136398453782, "grad_norm": 2.1439363327416583, "learning_rate": 1.8681236504626277e-05, "loss": 0.7001, "step": 5344 }, { "epoch": 0.19042699111103195, "grad_norm": 1.339789176224169, "learning_rate": 1.8680663692669383e-05, "loss": 0.6245, "step": 5345 }, { "epoch": 0.19046261823752605, "grad_norm": 1.1862407876629342, "learning_rate": 1.8680090765123388e-05, "loss": 0.4547, "step": 5346 }, { "epoch": 0.19049824536402016, "grad_norm": 1.870663380603342, "learning_rate": 1.867951772199591e-05, "loss": 0.6169, "step": 5347 }, { "epoch": 0.1905338724905143, "grad_norm": 1.2628827091453694, "learning_rate": 1.867894456329459e-05, "loss": 0.5483, "step": 5348 }, { "epoch": 0.1905694996170084, "grad_norm": 1.6374978508566405, "learning_rate": 1.867837128902706e-05, "loss": 0.5913, "step": 5349 }, { "epoch": 0.1906051267435025, "grad_norm": 1.2140620362229895, "learning_rate": 1.8677797899200945e-05, "loss": 0.5438, "step": 5350 }, { "epoch": 0.19064075386999663, "grad_norm": 1.6729473519307319, "learning_rate": 1.867722439382389e-05, "loss": 0.5385, "step": 5351 }, { "epoch": 0.19067638099649073, "grad_norm": 1.5373607068096933, "learning_rate": 1.8676650772903527e-05, "loss": 0.5524, "step": 5352 }, { "epoch": 0.19071200812298483, "grad_norm": 1.4652139920170493, "learning_rate": 1.867607703644749e-05, "loss": 0.6411, "step": 5353 }, { "epoch": 0.19074763524947896, "grad_norm": 1.645077953500124, "learning_rate": 1.867550318446343e-05, "loss": 0.554, "step": 5354 }, { "epoch": 0.19078326237597307, "grad_norm": 1.7120756380307645, "learning_rate": 1.8674929216958977e-05, "loss": 0.7262, "step": 5355 }, { "epoch": 0.19081888950246717, "grad_norm": 1.1528930459255513, "learning_rate": 1.867435513394178e-05, "loss": 0.5382, "step": 5356 }, { "epoch": 0.1908545166289613, "grad_norm": 1.1112090292637526, "learning_rate": 1.867378093541948e-05, "loss": 0.5084, "step": 5357 }, { "epoch": 0.1908901437554554, "grad_norm": 1.9664384793273921, "learning_rate": 1.8673206621399724e-05, "loss": 0.5587, "step": 5358 }, { "epoch": 0.1909257708819495, "grad_norm": 1.3026320374166862, "learning_rate": 1.867263219189016e-05, "loss": 0.6155, "step": 5359 }, { "epoch": 0.19096139800844364, "grad_norm": 1.0077659327949968, "learning_rate": 1.8672057646898435e-05, "loss": 0.5445, "step": 5360 }, { "epoch": 0.19099702513493774, "grad_norm": 1.55976642561345, "learning_rate": 1.8671482986432202e-05, "loss": 0.5318, "step": 5361 }, { "epoch": 0.19103265226143185, "grad_norm": 2.029496137328628, "learning_rate": 1.8670908210499112e-05, "loss": 0.7253, "step": 5362 }, { "epoch": 0.19106827938792598, "grad_norm": 1.518555419974101, "learning_rate": 1.867033331910682e-05, "loss": 0.5989, "step": 5363 }, { "epoch": 0.19110390651442008, "grad_norm": 1.1627834567894086, "learning_rate": 1.866975831226298e-05, "loss": 0.5248, "step": 5364 }, { "epoch": 0.19113953364091418, "grad_norm": 1.5632151033481168, "learning_rate": 1.8669183189975244e-05, "loss": 0.5535, "step": 5365 }, { "epoch": 0.19117516076740831, "grad_norm": 1.145269134665934, "learning_rate": 1.8668607952251277e-05, "loss": 0.5526, "step": 5366 }, { "epoch": 0.19121078789390242, "grad_norm": 1.4093441982849517, "learning_rate": 1.866803259909874e-05, "loss": 0.5767, "step": 5367 }, { "epoch": 0.19124641502039652, "grad_norm": 1.6744183852273424, "learning_rate": 1.8667457130525286e-05, "loss": 0.5032, "step": 5368 }, { "epoch": 0.19128204214689065, "grad_norm": 1.1633307160012127, "learning_rate": 1.866688154653858e-05, "loss": 0.4853, "step": 5369 }, { "epoch": 0.19131766927338476, "grad_norm": 1.2127121424713871, "learning_rate": 1.8666305847146292e-05, "loss": 0.5944, "step": 5370 }, { "epoch": 0.19135329639987886, "grad_norm": 1.6687270584856975, "learning_rate": 1.8665730032356085e-05, "loss": 0.5311, "step": 5371 }, { "epoch": 0.191388923526373, "grad_norm": 1.03315737524397, "learning_rate": 1.8665154102175626e-05, "loss": 0.5006, "step": 5372 }, { "epoch": 0.1914245506528671, "grad_norm": 1.8039715270982986, "learning_rate": 1.866457805661258e-05, "loss": 0.606, "step": 5373 }, { "epoch": 0.1914601777793612, "grad_norm": 1.2010813789911114, "learning_rate": 1.8664001895674623e-05, "loss": 0.5177, "step": 5374 }, { "epoch": 0.19149580490585533, "grad_norm": 1.7059218648088574, "learning_rate": 1.8663425619369424e-05, "loss": 0.6662, "step": 5375 }, { "epoch": 0.19153143203234943, "grad_norm": 1.104011577228881, "learning_rate": 1.866284922770466e-05, "loss": 0.4795, "step": 5376 }, { "epoch": 0.19156705915884353, "grad_norm": 1.5677549793819905, "learning_rate": 1.8662272720688e-05, "loss": 0.652, "step": 5377 }, { "epoch": 0.19160268628533766, "grad_norm": 1.1122570159153797, "learning_rate": 1.8661696098327125e-05, "loss": 0.5445, "step": 5378 }, { "epoch": 0.19163831341183177, "grad_norm": 1.1994523669386778, "learning_rate": 1.8661119360629714e-05, "loss": 0.5401, "step": 5379 }, { "epoch": 0.19167394053832587, "grad_norm": 2.080574618043182, "learning_rate": 1.8660542507603444e-05, "loss": 0.616, "step": 5380 }, { "epoch": 0.19170956766482, "grad_norm": 1.5887526149508873, "learning_rate": 1.8659965539255996e-05, "loss": 0.7121, "step": 5381 }, { "epoch": 0.1917451947913141, "grad_norm": 1.1433227096489569, "learning_rate": 1.8659388455595055e-05, "loss": 0.5185, "step": 5382 }, { "epoch": 0.1917808219178082, "grad_norm": 1.0794287915580227, "learning_rate": 1.8658811256628308e-05, "loss": 0.5256, "step": 5383 }, { "epoch": 0.19181644904430234, "grad_norm": 2.0068539933864424, "learning_rate": 1.865823394236343e-05, "loss": 0.5816, "step": 5384 }, { "epoch": 0.19185207617079644, "grad_norm": 1.0980654057540202, "learning_rate": 1.865765651280812e-05, "loss": 0.5, "step": 5385 }, { "epoch": 0.19188770329729055, "grad_norm": 1.101202310026406, "learning_rate": 1.8657078967970063e-05, "loss": 0.5213, "step": 5386 }, { "epoch": 0.19192333042378468, "grad_norm": 1.2436632329074804, "learning_rate": 1.8656501307856943e-05, "loss": 0.543, "step": 5387 }, { "epoch": 0.19195895755027878, "grad_norm": 1.4132128089816565, "learning_rate": 1.8655923532476463e-05, "loss": 0.493, "step": 5388 }, { "epoch": 0.19199458467677288, "grad_norm": 2.0764681326121717, "learning_rate": 1.865534564183631e-05, "loss": 0.5069, "step": 5389 }, { "epoch": 0.19203021180326701, "grad_norm": 1.577886190063808, "learning_rate": 1.8654767635944178e-05, "loss": 0.6383, "step": 5390 }, { "epoch": 0.19206583892976112, "grad_norm": 1.2354165542027886, "learning_rate": 1.8654189514807765e-05, "loss": 0.5472, "step": 5391 }, { "epoch": 0.19210146605625522, "grad_norm": 1.2235677236436902, "learning_rate": 1.8653611278434774e-05, "loss": 0.5838, "step": 5392 }, { "epoch": 0.19213709318274935, "grad_norm": 1.1611778436650297, "learning_rate": 1.8653032926832897e-05, "loss": 0.5531, "step": 5393 }, { "epoch": 0.19217272030924346, "grad_norm": 1.1300993201944436, "learning_rate": 1.865245446000984e-05, "loss": 0.5148, "step": 5394 }, { "epoch": 0.19220834743573756, "grad_norm": 1.1142301401443067, "learning_rate": 1.86518758779733e-05, "loss": 0.4828, "step": 5395 }, { "epoch": 0.1922439745622317, "grad_norm": 1.7322630891646087, "learning_rate": 1.865129718073099e-05, "loss": 0.6495, "step": 5396 }, { "epoch": 0.1922796016887258, "grad_norm": 1.2321672365863294, "learning_rate": 1.8650718368290613e-05, "loss": 0.5317, "step": 5397 }, { "epoch": 0.1923152288152199, "grad_norm": 1.5481421194249583, "learning_rate": 1.865013944065987e-05, "loss": 0.6569, "step": 5398 }, { "epoch": 0.19235085594171403, "grad_norm": 1.8619874035487574, "learning_rate": 1.8649560397846475e-05, "loss": 0.6122, "step": 5399 }, { "epoch": 0.19238648306820813, "grad_norm": 1.2072351225953113, "learning_rate": 1.864898123985814e-05, "loss": 0.4957, "step": 5400 }, { "epoch": 0.19242211019470223, "grad_norm": 1.3742135889291915, "learning_rate": 1.864840196670257e-05, "loss": 0.5363, "step": 5401 }, { "epoch": 0.19245773732119636, "grad_norm": 2.1050560401630607, "learning_rate": 1.8647822578387488e-05, "loss": 0.499, "step": 5402 }, { "epoch": 0.19249336444769047, "grad_norm": 1.2357186912758606, "learning_rate": 1.8647243074920604e-05, "loss": 0.52, "step": 5403 }, { "epoch": 0.19252899157418457, "grad_norm": 1.0561156172079091, "learning_rate": 1.864666345630963e-05, "loss": 0.5058, "step": 5404 }, { "epoch": 0.1925646187006787, "grad_norm": 1.1928470213220312, "learning_rate": 1.864608372256229e-05, "loss": 0.5451, "step": 5405 }, { "epoch": 0.1926002458271728, "grad_norm": 1.7352763670437508, "learning_rate": 1.86455038736863e-05, "loss": 0.5461, "step": 5406 }, { "epoch": 0.1926358729536669, "grad_norm": 1.7329268686299215, "learning_rate": 1.864492390968939e-05, "loss": 0.6558, "step": 5407 }, { "epoch": 0.19267150008016104, "grad_norm": 2.0436258038763606, "learning_rate": 1.864434383057927e-05, "loss": 0.6643, "step": 5408 }, { "epoch": 0.19270712720665514, "grad_norm": 1.2054025216360245, "learning_rate": 1.864376363636367e-05, "loss": 0.5564, "step": 5409 }, { "epoch": 0.19274275433314925, "grad_norm": 1.5229658711677028, "learning_rate": 1.8643183327050315e-05, "loss": 0.5676, "step": 5410 }, { "epoch": 0.19277838145964338, "grad_norm": 1.6530186023576523, "learning_rate": 1.8642602902646937e-05, "loss": 0.6432, "step": 5411 }, { "epoch": 0.19281400858613748, "grad_norm": 1.640196694867216, "learning_rate": 1.8642022363161255e-05, "loss": 0.6334, "step": 5412 }, { "epoch": 0.19284963571263158, "grad_norm": 1.166496318960891, "learning_rate": 1.864144170860101e-05, "loss": 0.5412, "step": 5413 }, { "epoch": 0.19288526283912572, "grad_norm": 1.4059516531183036, "learning_rate": 1.8640860938973925e-05, "loss": 0.5365, "step": 5414 }, { "epoch": 0.19292088996561982, "grad_norm": 1.6372534950572626, "learning_rate": 1.8640280054287738e-05, "loss": 0.6154, "step": 5415 }, { "epoch": 0.19295651709211395, "grad_norm": 1.3048566079814468, "learning_rate": 1.8639699054550187e-05, "loss": 0.4997, "step": 5416 }, { "epoch": 0.19299214421860805, "grad_norm": 1.6170630059025155, "learning_rate": 1.8639117939768997e-05, "loss": 0.6341, "step": 5417 }, { "epoch": 0.19302777134510216, "grad_norm": 2.114572155783274, "learning_rate": 1.8638536709951916e-05, "loss": 0.6929, "step": 5418 }, { "epoch": 0.1930633984715963, "grad_norm": 1.8323213747294593, "learning_rate": 1.8637955365106685e-05, "loss": 0.6052, "step": 5419 }, { "epoch": 0.1930990255980904, "grad_norm": 1.137283359023904, "learning_rate": 1.8637373905241036e-05, "loss": 0.5297, "step": 5420 }, { "epoch": 0.1931346527245845, "grad_norm": 1.5895793444320323, "learning_rate": 1.8636792330362716e-05, "loss": 0.6193, "step": 5421 }, { "epoch": 0.19317027985107862, "grad_norm": 2.113251730985031, "learning_rate": 1.8636210640479474e-05, "loss": 0.6758, "step": 5422 }, { "epoch": 0.19320590697757273, "grad_norm": 1.0640885507764666, "learning_rate": 1.8635628835599048e-05, "loss": 0.471, "step": 5423 }, { "epoch": 0.19324153410406683, "grad_norm": 1.1398530573279597, "learning_rate": 1.8635046915729186e-05, "loss": 0.5079, "step": 5424 }, { "epoch": 0.19327716123056096, "grad_norm": 1.1937410847831418, "learning_rate": 1.8634464880877644e-05, "loss": 0.4908, "step": 5425 }, { "epoch": 0.19331278835705507, "grad_norm": 1.5192277264426617, "learning_rate": 1.8633882731052163e-05, "loss": 0.5428, "step": 5426 }, { "epoch": 0.19334841548354917, "grad_norm": 2.099530238240331, "learning_rate": 1.8633300466260497e-05, "loss": 0.5472, "step": 5427 }, { "epoch": 0.1933840426100433, "grad_norm": 2.2012447495263805, "learning_rate": 1.8632718086510404e-05, "loss": 0.657, "step": 5428 }, { "epoch": 0.1934196697365374, "grad_norm": 1.2193767259748458, "learning_rate": 1.8632135591809635e-05, "loss": 0.5428, "step": 5429 }, { "epoch": 0.1934552968630315, "grad_norm": 1.1206270578421096, "learning_rate": 1.8631552982165946e-05, "loss": 0.4589, "step": 5430 }, { "epoch": 0.19349092398952564, "grad_norm": 1.3397413380529617, "learning_rate": 1.8630970257587096e-05, "loss": 0.5188, "step": 5431 }, { "epoch": 0.19352655111601974, "grad_norm": 2.3368508075657775, "learning_rate": 1.8630387418080848e-05, "loss": 0.6525, "step": 5432 }, { "epoch": 0.19356217824251384, "grad_norm": 2.213712122127887, "learning_rate": 1.8629804463654956e-05, "loss": 0.6073, "step": 5433 }, { "epoch": 0.19359780536900797, "grad_norm": 1.262396384632406, "learning_rate": 1.8629221394317186e-05, "loss": 0.5595, "step": 5434 }, { "epoch": 0.19363343249550208, "grad_norm": 1.5292762055220703, "learning_rate": 1.8628638210075305e-05, "loss": 0.5688, "step": 5435 }, { "epoch": 0.19366905962199618, "grad_norm": 1.3162776479218985, "learning_rate": 1.862805491093707e-05, "loss": 0.5871, "step": 5436 }, { "epoch": 0.1937046867484903, "grad_norm": 1.9385183306008924, "learning_rate": 1.8627471496910257e-05, "loss": 0.7126, "step": 5437 }, { "epoch": 0.19374031387498442, "grad_norm": 1.7540786734557963, "learning_rate": 1.862688796800263e-05, "loss": 0.5983, "step": 5438 }, { "epoch": 0.19377594100147852, "grad_norm": 1.1408764200224535, "learning_rate": 1.862630432422196e-05, "loss": 0.4987, "step": 5439 }, { "epoch": 0.19381156812797265, "grad_norm": 1.0962226825764425, "learning_rate": 1.8625720565576016e-05, "loss": 0.4639, "step": 5440 }, { "epoch": 0.19384719525446675, "grad_norm": 1.936578903205479, "learning_rate": 1.8625136692072577e-05, "loss": 0.6574, "step": 5441 }, { "epoch": 0.19388282238096086, "grad_norm": 1.0544213058005159, "learning_rate": 1.862455270371941e-05, "loss": 0.4961, "step": 5442 }, { "epoch": 0.193918449507455, "grad_norm": 1.1754831420305882, "learning_rate": 1.8623968600524302e-05, "loss": 0.5002, "step": 5443 }, { "epoch": 0.1939540766339491, "grad_norm": 1.163867566168433, "learning_rate": 1.862338438249502e-05, "loss": 0.5103, "step": 5444 }, { "epoch": 0.1939897037604432, "grad_norm": 1.1661996822004126, "learning_rate": 1.8622800049639347e-05, "loss": 0.469, "step": 5445 }, { "epoch": 0.19402533088693732, "grad_norm": 1.5780451201094465, "learning_rate": 1.8622215601965066e-05, "loss": 0.6605, "step": 5446 }, { "epoch": 0.19406095801343143, "grad_norm": 1.903976590747028, "learning_rate": 1.8621631039479954e-05, "loss": 0.68, "step": 5447 }, { "epoch": 0.19409658513992553, "grad_norm": 1.0947333527290428, "learning_rate": 1.8621046362191806e-05, "loss": 0.4773, "step": 5448 }, { "epoch": 0.19413221226641966, "grad_norm": 1.6346134326580624, "learning_rate": 1.8620461570108396e-05, "loss": 0.6367, "step": 5449 }, { "epoch": 0.19416783939291377, "grad_norm": 1.1150686793255502, "learning_rate": 1.8619876663237513e-05, "loss": 0.5361, "step": 5450 }, { "epoch": 0.19420346651940787, "grad_norm": 1.157904728332684, "learning_rate": 1.8619291641586947e-05, "loss": 0.5151, "step": 5451 }, { "epoch": 0.194239093645902, "grad_norm": 2.0286123002286796, "learning_rate": 1.861870650516449e-05, "loss": 0.5889, "step": 5452 }, { "epoch": 0.1942747207723961, "grad_norm": 1.88008605239593, "learning_rate": 1.861812125397793e-05, "loss": 0.6666, "step": 5453 }, { "epoch": 0.1943103478988902, "grad_norm": 1.218011339502635, "learning_rate": 1.8617535888035066e-05, "loss": 0.5317, "step": 5454 }, { "epoch": 0.19434597502538434, "grad_norm": 1.6048528251975822, "learning_rate": 1.8616950407343685e-05, "loss": 0.6071, "step": 5455 }, { "epoch": 0.19438160215187844, "grad_norm": 1.9983765645268519, "learning_rate": 1.861636481191159e-05, "loss": 0.6621, "step": 5456 }, { "epoch": 0.19441722927837254, "grad_norm": 1.053877992411941, "learning_rate": 1.861577910174657e-05, "loss": 0.5358, "step": 5457 }, { "epoch": 0.19445285640486668, "grad_norm": 1.13068416359856, "learning_rate": 1.8615193276856434e-05, "loss": 0.5012, "step": 5458 }, { "epoch": 0.19448848353136078, "grad_norm": 1.0693703929104226, "learning_rate": 1.8614607337248975e-05, "loss": 0.4853, "step": 5459 }, { "epoch": 0.19452411065785488, "grad_norm": 1.7569595394459758, "learning_rate": 1.8614021282932e-05, "loss": 0.6759, "step": 5460 }, { "epoch": 0.194559737784349, "grad_norm": 1.9075012453279592, "learning_rate": 1.861343511391331e-05, "loss": 0.6363, "step": 5461 }, { "epoch": 0.19459536491084312, "grad_norm": 1.0775110906667502, "learning_rate": 1.861284883020071e-05, "loss": 0.4612, "step": 5462 }, { "epoch": 0.19463099203733722, "grad_norm": 2.0280690059942263, "learning_rate": 1.861226243180201e-05, "loss": 0.6559, "step": 5463 }, { "epoch": 0.19466661916383135, "grad_norm": 1.4848262498867693, "learning_rate": 1.861167591872501e-05, "loss": 0.6001, "step": 5464 }, { "epoch": 0.19470224629032545, "grad_norm": 1.5155905987969924, "learning_rate": 1.8611089290977534e-05, "loss": 0.5271, "step": 5465 }, { "epoch": 0.19473787341681956, "grad_norm": 1.6579123821365167, "learning_rate": 1.861050254856738e-05, "loss": 0.5704, "step": 5466 }, { "epoch": 0.1947735005433137, "grad_norm": 1.5591534000206289, "learning_rate": 1.8609915691502367e-05, "loss": 0.6383, "step": 5467 }, { "epoch": 0.1948091276698078, "grad_norm": 1.1878340034596715, "learning_rate": 1.860932871979031e-05, "loss": 0.5208, "step": 5468 }, { "epoch": 0.1948447547963019, "grad_norm": 10.096692490345117, "learning_rate": 1.8608741633439023e-05, "loss": 0.5996, "step": 5469 }, { "epoch": 0.19488038192279603, "grad_norm": 1.1141176869964904, "learning_rate": 1.8608154432456324e-05, "loss": 0.4661, "step": 5470 }, { "epoch": 0.19491600904929013, "grad_norm": 1.6579476742755346, "learning_rate": 1.860756711685003e-05, "loss": 0.6634, "step": 5471 }, { "epoch": 0.19495163617578423, "grad_norm": 1.2802801891445263, "learning_rate": 1.8606979686627966e-05, "loss": 0.4946, "step": 5472 }, { "epoch": 0.19498726330227836, "grad_norm": 1.1705431876549206, "learning_rate": 1.860639214179795e-05, "loss": 0.5016, "step": 5473 }, { "epoch": 0.19502289042877247, "grad_norm": 1.7072063433375275, "learning_rate": 1.860580448236781e-05, "loss": 0.5971, "step": 5474 }, { "epoch": 0.19505851755526657, "grad_norm": 1.7012556969060224, "learning_rate": 1.8605216708345363e-05, "loss": 0.6645, "step": 5475 }, { "epoch": 0.1950941446817607, "grad_norm": 2.1060409469508063, "learning_rate": 1.8604628819738447e-05, "loss": 0.6396, "step": 5476 }, { "epoch": 0.1951297718082548, "grad_norm": 1.4681076573781333, "learning_rate": 1.8604040816554882e-05, "loss": 0.5443, "step": 5477 }, { "epoch": 0.1951653989347489, "grad_norm": 1.5780566746114348, "learning_rate": 1.8603452698802498e-05, "loss": 0.6002, "step": 5478 }, { "epoch": 0.19520102606124304, "grad_norm": 1.1648624366748548, "learning_rate": 1.860286446648913e-05, "loss": 0.4776, "step": 5479 }, { "epoch": 0.19523665318773714, "grad_norm": 1.4989758508198763, "learning_rate": 1.8602276119622605e-05, "loss": 0.557, "step": 5480 }, { "epoch": 0.19527228031423124, "grad_norm": 1.8529460807499682, "learning_rate": 1.8601687658210763e-05, "loss": 0.6227, "step": 5481 }, { "epoch": 0.19530790744072538, "grad_norm": 1.0827250938188493, "learning_rate": 1.860109908226144e-05, "loss": 0.5134, "step": 5482 }, { "epoch": 0.19534353456721948, "grad_norm": 1.2029269840409207, "learning_rate": 1.8600510391782465e-05, "loss": 0.5131, "step": 5483 }, { "epoch": 0.19537916169371358, "grad_norm": 1.545771397449588, "learning_rate": 1.859992158678169e-05, "loss": 0.6078, "step": 5484 }, { "epoch": 0.1954147888202077, "grad_norm": 1.8492576674156411, "learning_rate": 1.8599332667266942e-05, "loss": 0.6687, "step": 5485 }, { "epoch": 0.19545041594670182, "grad_norm": 1.2481398228625231, "learning_rate": 1.8598743633246073e-05, "loss": 0.5277, "step": 5486 }, { "epoch": 0.19548604307319592, "grad_norm": 1.1451861984856115, "learning_rate": 1.859815448472692e-05, "loss": 0.5866, "step": 5487 }, { "epoch": 0.19552167019969005, "grad_norm": 1.1600081716582216, "learning_rate": 1.8597565221717332e-05, "loss": 0.4818, "step": 5488 }, { "epoch": 0.19555729732618415, "grad_norm": 1.4986379525310545, "learning_rate": 1.859697584422515e-05, "loss": 0.6032, "step": 5489 }, { "epoch": 0.19559292445267826, "grad_norm": 1.0637067472684836, "learning_rate": 1.859638635225823e-05, "loss": 0.4976, "step": 5490 }, { "epoch": 0.1956285515791724, "grad_norm": 1.650688131509857, "learning_rate": 1.8595796745824417e-05, "loss": 0.6569, "step": 5491 }, { "epoch": 0.1956641787056665, "grad_norm": 1.2120891708090062, "learning_rate": 1.8595207024931558e-05, "loss": 0.5303, "step": 5492 }, { "epoch": 0.1956998058321606, "grad_norm": 1.0701665816474262, "learning_rate": 1.8594617189587515e-05, "loss": 0.4579, "step": 5493 }, { "epoch": 0.19573543295865473, "grad_norm": 1.7742789794843508, "learning_rate": 1.8594027239800132e-05, "loss": 0.5636, "step": 5494 }, { "epoch": 0.19577106008514883, "grad_norm": 1.4728256203857644, "learning_rate": 1.8593437175577274e-05, "loss": 0.6186, "step": 5495 }, { "epoch": 0.19580668721164293, "grad_norm": 1.603109712582169, "learning_rate": 1.8592846996926793e-05, "loss": 0.5764, "step": 5496 }, { "epoch": 0.19584231433813706, "grad_norm": 1.4935340225907658, "learning_rate": 1.8592256703856547e-05, "loss": 0.7261, "step": 5497 }, { "epoch": 0.19587794146463117, "grad_norm": 1.2441873438609816, "learning_rate": 1.8591666296374393e-05, "loss": 0.5128, "step": 5498 }, { "epoch": 0.19591356859112527, "grad_norm": 1.1020520259196929, "learning_rate": 1.8591075774488205e-05, "loss": 0.5192, "step": 5499 }, { "epoch": 0.1959491957176194, "grad_norm": 2.104982995057128, "learning_rate": 1.8590485138205833e-05, "loss": 0.5629, "step": 5500 }, { "epoch": 0.1959848228441135, "grad_norm": 1.7364686849400068, "learning_rate": 1.8589894387535148e-05, "loss": 0.6108, "step": 5501 }, { "epoch": 0.1960204499706076, "grad_norm": 1.1582123073621367, "learning_rate": 1.8589303522484017e-05, "loss": 0.5793, "step": 5502 }, { "epoch": 0.19605607709710174, "grad_norm": 1.0944760678150318, "learning_rate": 1.8588712543060304e-05, "loss": 0.5529, "step": 5503 }, { "epoch": 0.19609170422359584, "grad_norm": 1.056289286804187, "learning_rate": 1.8588121449271882e-05, "loss": 0.5077, "step": 5504 }, { "epoch": 0.19612733135008995, "grad_norm": 1.120709584521122, "learning_rate": 1.8587530241126617e-05, "loss": 0.5481, "step": 5505 }, { "epoch": 0.19616295847658408, "grad_norm": 1.669647300698209, "learning_rate": 1.8586938918632385e-05, "loss": 0.604, "step": 5506 }, { "epoch": 0.19619858560307818, "grad_norm": 1.9091628456419765, "learning_rate": 1.8586347481797064e-05, "loss": 0.5434, "step": 5507 }, { "epoch": 0.19623421272957228, "grad_norm": 1.680060811501011, "learning_rate": 1.858575593062852e-05, "loss": 0.5531, "step": 5508 }, { "epoch": 0.1962698398560664, "grad_norm": 1.558362496802887, "learning_rate": 1.8585164265134636e-05, "loss": 0.5885, "step": 5509 }, { "epoch": 0.19630546698256052, "grad_norm": 1.1553809713680023, "learning_rate": 1.858457248532329e-05, "loss": 0.5194, "step": 5510 }, { "epoch": 0.19634109410905465, "grad_norm": 1.2405194944664222, "learning_rate": 1.8583980591202362e-05, "loss": 0.517, "step": 5511 }, { "epoch": 0.19637672123554875, "grad_norm": 1.084569596497389, "learning_rate": 1.858338858277973e-05, "loss": 0.5147, "step": 5512 }, { "epoch": 0.19641234836204285, "grad_norm": 2.0254734075426506, "learning_rate": 1.8582796460063282e-05, "loss": 0.589, "step": 5513 }, { "epoch": 0.19644797548853699, "grad_norm": 1.677096906042616, "learning_rate": 1.85822042230609e-05, "loss": 0.5813, "step": 5514 }, { "epoch": 0.1964836026150311, "grad_norm": 1.0967653590890676, "learning_rate": 1.858161187178047e-05, "loss": 0.5068, "step": 5515 }, { "epoch": 0.1965192297415252, "grad_norm": 1.1610935153834077, "learning_rate": 1.8581019406229875e-05, "loss": 0.5495, "step": 5516 }, { "epoch": 0.19655485686801932, "grad_norm": 1.2697800619639108, "learning_rate": 1.8580426826417016e-05, "loss": 0.5433, "step": 5517 }, { "epoch": 0.19659048399451343, "grad_norm": 1.853887809406493, "learning_rate": 1.8579834132349773e-05, "loss": 0.603, "step": 5518 }, { "epoch": 0.19662611112100753, "grad_norm": 1.9234621483946952, "learning_rate": 1.857924132403604e-05, "loss": 0.6886, "step": 5519 }, { "epoch": 0.19666173824750166, "grad_norm": 3.061132340304384, "learning_rate": 1.8578648401483713e-05, "loss": 0.5857, "step": 5520 }, { "epoch": 0.19669736537399576, "grad_norm": 1.2355073386097668, "learning_rate": 1.857805536470069e-05, "loss": 0.4673, "step": 5521 }, { "epoch": 0.19673299250048987, "grad_norm": 1.660801665584707, "learning_rate": 1.857746221369486e-05, "loss": 0.594, "step": 5522 }, { "epoch": 0.196768619626984, "grad_norm": 1.189998343427919, "learning_rate": 1.857686894847413e-05, "loss": 0.5505, "step": 5523 }, { "epoch": 0.1968042467534781, "grad_norm": 1.1407176355227528, "learning_rate": 1.857627556904639e-05, "loss": 0.4982, "step": 5524 }, { "epoch": 0.1968398738799722, "grad_norm": 1.1643007411246342, "learning_rate": 1.857568207541955e-05, "loss": 0.5574, "step": 5525 }, { "epoch": 0.19687550100646634, "grad_norm": 1.9042048518062118, "learning_rate": 1.8575088467601504e-05, "loss": 0.633, "step": 5526 }, { "epoch": 0.19691112813296044, "grad_norm": 1.2816165571296954, "learning_rate": 1.8574494745600166e-05, "loss": 0.5343, "step": 5527 }, { "epoch": 0.19694675525945454, "grad_norm": 2.1720851251294206, "learning_rate": 1.8573900909423438e-05, "loss": 0.5901, "step": 5528 }, { "epoch": 0.19698238238594867, "grad_norm": 1.1162839336487715, "learning_rate": 1.857330695907922e-05, "loss": 0.4933, "step": 5529 }, { "epoch": 0.19701800951244278, "grad_norm": 1.3261919134695002, "learning_rate": 1.8572712894575435e-05, "loss": 0.5629, "step": 5530 }, { "epoch": 0.19705363663893688, "grad_norm": 1.7342311147171512, "learning_rate": 1.8572118715919983e-05, "loss": 0.7278, "step": 5531 }, { "epoch": 0.197089263765431, "grad_norm": 1.1989855552575064, "learning_rate": 1.857152442312078e-05, "loss": 0.5368, "step": 5532 }, { "epoch": 0.19712489089192511, "grad_norm": 1.1744175164953758, "learning_rate": 1.8570930016185737e-05, "loss": 0.5185, "step": 5533 }, { "epoch": 0.19716051801841922, "grad_norm": 1.5327467438906464, "learning_rate": 1.8570335495122772e-05, "loss": 0.625, "step": 5534 }, { "epoch": 0.19719614514491335, "grad_norm": 1.5592224286279464, "learning_rate": 1.8569740859939797e-05, "loss": 0.5639, "step": 5535 }, { "epoch": 0.19723177227140745, "grad_norm": 1.1967675015360582, "learning_rate": 1.856914611064473e-05, "loss": 0.4676, "step": 5536 }, { "epoch": 0.19726739939790155, "grad_norm": 1.2114152945619174, "learning_rate": 1.85685512472455e-05, "loss": 0.529, "step": 5537 }, { "epoch": 0.19730302652439569, "grad_norm": 4.2964382125616565, "learning_rate": 1.8567956269750014e-05, "loss": 0.5932, "step": 5538 }, { "epoch": 0.1973386536508898, "grad_norm": 1.82233284828192, "learning_rate": 1.8567361178166206e-05, "loss": 0.5998, "step": 5539 }, { "epoch": 0.1973742807773839, "grad_norm": 1.0806004002246914, "learning_rate": 1.8566765972501994e-05, "loss": 0.5605, "step": 5540 }, { "epoch": 0.19740990790387802, "grad_norm": 1.2463257448243272, "learning_rate": 1.8566170652765304e-05, "loss": 0.5344, "step": 5541 }, { "epoch": 0.19744553503037213, "grad_norm": 1.6903743348181888, "learning_rate": 1.8565575218964067e-05, "loss": 0.5453, "step": 5542 }, { "epoch": 0.19748116215686623, "grad_norm": 1.131766774146485, "learning_rate": 1.8564979671106212e-05, "loss": 0.4989, "step": 5543 }, { "epoch": 0.19751678928336036, "grad_norm": 1.1319334982148626, "learning_rate": 1.856438400919966e-05, "loss": 0.5648, "step": 5544 }, { "epoch": 0.19755241640985446, "grad_norm": 1.4678269379549231, "learning_rate": 1.8563788233252352e-05, "loss": 0.6975, "step": 5545 }, { "epoch": 0.19758804353634857, "grad_norm": 1.1572548078989464, "learning_rate": 1.8563192343272215e-05, "loss": 0.4768, "step": 5546 }, { "epoch": 0.1976236706628427, "grad_norm": 1.3876182283144902, "learning_rate": 1.856259633926719e-05, "loss": 0.5779, "step": 5547 }, { "epoch": 0.1976592977893368, "grad_norm": 2.1147749917811947, "learning_rate": 1.856200022124521e-05, "loss": 0.6432, "step": 5548 }, { "epoch": 0.1976949249158309, "grad_norm": 1.1183158280878813, "learning_rate": 1.8561403989214207e-05, "loss": 0.4713, "step": 5549 }, { "epoch": 0.19773055204232504, "grad_norm": 1.7422756316748709, "learning_rate": 1.856080764318213e-05, "loss": 0.6383, "step": 5550 }, { "epoch": 0.19776617916881914, "grad_norm": 1.6017248590114155, "learning_rate": 1.8560211183156918e-05, "loss": 0.6627, "step": 5551 }, { "epoch": 0.19780180629531324, "grad_norm": 1.2362988368840075, "learning_rate": 1.8559614609146508e-05, "loss": 0.541, "step": 5552 }, { "epoch": 0.19783743342180737, "grad_norm": 1.1573421712095238, "learning_rate": 1.855901792115885e-05, "loss": 0.5237, "step": 5553 }, { "epoch": 0.19787306054830148, "grad_norm": 1.9256646471561707, "learning_rate": 1.8558421119201882e-05, "loss": 0.6779, "step": 5554 }, { "epoch": 0.19790868767479558, "grad_norm": 1.9611812380461646, "learning_rate": 1.8557824203283553e-05, "loss": 0.5982, "step": 5555 }, { "epoch": 0.1979443148012897, "grad_norm": 1.861146130462093, "learning_rate": 1.8557227173411818e-05, "loss": 0.6187, "step": 5556 }, { "epoch": 0.19797994192778381, "grad_norm": 1.174168139346631, "learning_rate": 1.855663002959462e-05, "loss": 0.4878, "step": 5557 }, { "epoch": 0.19801556905427792, "grad_norm": 1.8984821478710645, "learning_rate": 1.8556032771839912e-05, "loss": 0.5782, "step": 5558 }, { "epoch": 0.19805119618077205, "grad_norm": 1.0948307566452278, "learning_rate": 1.8555435400155648e-05, "loss": 0.4914, "step": 5559 }, { "epoch": 0.19808682330726615, "grad_norm": 1.107023666841615, "learning_rate": 1.8554837914549783e-05, "loss": 0.5221, "step": 5560 }, { "epoch": 0.19812245043376026, "grad_norm": 1.1641252542145148, "learning_rate": 1.8554240315030268e-05, "loss": 0.5678, "step": 5561 }, { "epoch": 0.1981580775602544, "grad_norm": 1.6334076105272308, "learning_rate": 1.855364260160507e-05, "loss": 0.7836, "step": 5562 }, { "epoch": 0.1981937046867485, "grad_norm": 2.113834635720528, "learning_rate": 1.8553044774282135e-05, "loss": 0.5873, "step": 5563 }, { "epoch": 0.1982293318132426, "grad_norm": 1.1949350242181493, "learning_rate": 1.8552446833069434e-05, "loss": 0.5015, "step": 5564 }, { "epoch": 0.19826495893973672, "grad_norm": 2.2808573131594185, "learning_rate": 1.855184877797493e-05, "loss": 0.5881, "step": 5565 }, { "epoch": 0.19830058606623083, "grad_norm": 1.176172703254451, "learning_rate": 1.855125060900658e-05, "loss": 0.5587, "step": 5566 }, { "epoch": 0.19833621319272493, "grad_norm": 1.0281542856411605, "learning_rate": 1.8550652326172344e-05, "loss": 0.4737, "step": 5567 }, { "epoch": 0.19837184031921906, "grad_norm": 1.0895192226473596, "learning_rate": 1.8550053929480202e-05, "loss": 0.5129, "step": 5568 }, { "epoch": 0.19840746744571316, "grad_norm": 2.834235496994921, "learning_rate": 1.8549455418938114e-05, "loss": 0.6124, "step": 5569 }, { "epoch": 0.19844309457220727, "grad_norm": 1.6401706123628874, "learning_rate": 1.8548856794554053e-05, "loss": 0.5651, "step": 5570 }, { "epoch": 0.1984787216987014, "grad_norm": 1.1839500312599696, "learning_rate": 1.8548258056335987e-05, "loss": 0.5631, "step": 5571 }, { "epoch": 0.1985143488251955, "grad_norm": 1.7333346011657798, "learning_rate": 1.854765920429189e-05, "loss": 0.6043, "step": 5572 }, { "epoch": 0.1985499759516896, "grad_norm": 1.1239089406223604, "learning_rate": 1.8547060238429737e-05, "loss": 0.531, "step": 5573 }, { "epoch": 0.19858560307818374, "grad_norm": 1.6785936346526729, "learning_rate": 1.8546461158757504e-05, "loss": 0.6118, "step": 5574 }, { "epoch": 0.19862123020467784, "grad_norm": 1.6461821338963987, "learning_rate": 1.8545861965283166e-05, "loss": 0.6771, "step": 5575 }, { "epoch": 0.19865685733117194, "grad_norm": 1.6548022201957189, "learning_rate": 1.8545262658014702e-05, "loss": 0.5968, "step": 5576 }, { "epoch": 0.19869248445766607, "grad_norm": 1.7382992884641306, "learning_rate": 1.8544663236960094e-05, "loss": 0.646, "step": 5577 }, { "epoch": 0.19872811158416018, "grad_norm": 1.9208199260306322, "learning_rate": 1.854406370212732e-05, "loss": 0.7017, "step": 5578 }, { "epoch": 0.19876373871065428, "grad_norm": 1.2429235012168875, "learning_rate": 1.8543464053524365e-05, "loss": 0.5337, "step": 5579 }, { "epoch": 0.1987993658371484, "grad_norm": 1.982254406498503, "learning_rate": 1.854286429115922e-05, "loss": 0.5981, "step": 5580 }, { "epoch": 0.19883499296364252, "grad_norm": 1.0375187555620615, "learning_rate": 1.8542264415039864e-05, "loss": 0.5003, "step": 5581 }, { "epoch": 0.19887062009013662, "grad_norm": 1.1364613319995314, "learning_rate": 1.8541664425174287e-05, "loss": 0.5035, "step": 5582 }, { "epoch": 0.19890624721663075, "grad_norm": 1.0863442740628737, "learning_rate": 1.8541064321570477e-05, "loss": 0.5113, "step": 5583 }, { "epoch": 0.19894187434312485, "grad_norm": 1.1438627665668708, "learning_rate": 1.8540464104236428e-05, "loss": 0.5041, "step": 5584 }, { "epoch": 0.19897750146961896, "grad_norm": 1.244232798720904, "learning_rate": 1.8539863773180125e-05, "loss": 0.5378, "step": 5585 }, { "epoch": 0.1990131285961131, "grad_norm": 1.2152022738147796, "learning_rate": 1.853926332840957e-05, "loss": 0.5163, "step": 5586 }, { "epoch": 0.1990487557226072, "grad_norm": 1.085449934886861, "learning_rate": 1.8538662769932754e-05, "loss": 0.4995, "step": 5587 }, { "epoch": 0.1990843828491013, "grad_norm": 1.1334093182100704, "learning_rate": 1.8538062097757676e-05, "loss": 0.5632, "step": 5588 }, { "epoch": 0.19912000997559542, "grad_norm": 1.211907201927371, "learning_rate": 1.8537461311892334e-05, "loss": 0.5278, "step": 5589 }, { "epoch": 0.19915563710208953, "grad_norm": 1.2099344851838818, "learning_rate": 1.8536860412344725e-05, "loss": 0.5386, "step": 5590 }, { "epoch": 0.19919126422858363, "grad_norm": 1.1526358031424755, "learning_rate": 1.8536259399122855e-05, "loss": 0.5259, "step": 5591 }, { "epoch": 0.19922689135507776, "grad_norm": 1.8305562128773358, "learning_rate": 1.8535658272234724e-05, "loss": 0.5289, "step": 5592 }, { "epoch": 0.19926251848157187, "grad_norm": 1.1499334040356723, "learning_rate": 1.8535057031688337e-05, "loss": 0.533, "step": 5593 }, { "epoch": 0.19929814560806597, "grad_norm": 1.6324118299878434, "learning_rate": 1.8534455677491697e-05, "loss": 0.6489, "step": 5594 }, { "epoch": 0.1993337727345601, "grad_norm": 1.8516007673302874, "learning_rate": 1.853385420965282e-05, "loss": 0.6633, "step": 5595 }, { "epoch": 0.1993693998610542, "grad_norm": 1.1820313334521682, "learning_rate": 1.8533252628179705e-05, "loss": 0.4862, "step": 5596 }, { "epoch": 0.1994050269875483, "grad_norm": 1.7404984755781114, "learning_rate": 1.853265093308037e-05, "loss": 0.68, "step": 5597 }, { "epoch": 0.19944065411404244, "grad_norm": 2.0487942022113517, "learning_rate": 1.853204912436282e-05, "loss": 0.6512, "step": 5598 }, { "epoch": 0.19947628124053654, "grad_norm": 1.6471016978647757, "learning_rate": 1.8531447202035077e-05, "loss": 0.6418, "step": 5599 }, { "epoch": 0.19951190836703064, "grad_norm": 1.7050737658387414, "learning_rate": 1.853084516610515e-05, "loss": 0.6084, "step": 5600 }, { "epoch": 0.19954753549352477, "grad_norm": 1.798899407095265, "learning_rate": 1.8530243016581055e-05, "loss": 0.6315, "step": 5601 }, { "epoch": 0.19958316262001888, "grad_norm": 2.2871141189262425, "learning_rate": 1.8529640753470812e-05, "loss": 0.6044, "step": 5602 }, { "epoch": 0.19961878974651298, "grad_norm": 1.8980721228697894, "learning_rate": 1.8529038376782443e-05, "loss": 0.6046, "step": 5603 }, { "epoch": 0.1996544168730071, "grad_norm": 1.2977920334051964, "learning_rate": 1.8528435886523964e-05, "loss": 0.5332, "step": 5604 }, { "epoch": 0.19969004399950122, "grad_norm": 1.9323579432215856, "learning_rate": 1.85278332827034e-05, "loss": 0.6425, "step": 5605 }, { "epoch": 0.19972567112599535, "grad_norm": 1.6198888997769068, "learning_rate": 1.852723056532878e-05, "loss": 0.5596, "step": 5606 }, { "epoch": 0.19976129825248945, "grad_norm": 1.0849610684566433, "learning_rate": 1.852662773440812e-05, "loss": 0.5179, "step": 5607 }, { "epoch": 0.19979692537898355, "grad_norm": 1.1050307162952484, "learning_rate": 1.8526024789949453e-05, "loss": 0.5258, "step": 5608 }, { "epoch": 0.19983255250547768, "grad_norm": 1.1731924166871603, "learning_rate": 1.852542173196081e-05, "loss": 0.5269, "step": 5609 }, { "epoch": 0.1998681796319718, "grad_norm": 2.63576146077179, "learning_rate": 1.8524818560450214e-05, "loss": 0.5652, "step": 5610 }, { "epoch": 0.1999038067584659, "grad_norm": 2.0024978108921236, "learning_rate": 1.8524215275425698e-05, "loss": 0.6755, "step": 5611 }, { "epoch": 0.19993943388496002, "grad_norm": 1.1544441235036864, "learning_rate": 1.85236118768953e-05, "loss": 0.5248, "step": 5612 }, { "epoch": 0.19997506101145412, "grad_norm": 1.7485336377917284, "learning_rate": 1.8523008364867056e-05, "loss": 0.6573, "step": 5613 }, { "epoch": 0.20001068813794823, "grad_norm": 1.3955898750075428, "learning_rate": 1.8522404739348993e-05, "loss": 0.5801, "step": 5614 }, { "epoch": 0.20004631526444236, "grad_norm": 1.2255581996653804, "learning_rate": 1.8521801000349156e-05, "loss": 0.4938, "step": 5615 }, { "epoch": 0.20008194239093646, "grad_norm": 1.5162544746096285, "learning_rate": 1.852119714787558e-05, "loss": 0.5485, "step": 5616 }, { "epoch": 0.20011756951743057, "grad_norm": 1.5522722957907482, "learning_rate": 1.8520593181936312e-05, "loss": 0.5756, "step": 5617 }, { "epoch": 0.2001531966439247, "grad_norm": 1.4483933740469623, "learning_rate": 1.851998910253939e-05, "loss": 0.6842, "step": 5618 }, { "epoch": 0.2001888237704188, "grad_norm": 1.8271768815965006, "learning_rate": 1.8519384909692854e-05, "loss": 0.6356, "step": 5619 }, { "epoch": 0.2002244508969129, "grad_norm": 1.6644769184088466, "learning_rate": 1.851878060340476e-05, "loss": 0.626, "step": 5620 }, { "epoch": 0.20026007802340703, "grad_norm": 1.9186518972351805, "learning_rate": 1.851817618368314e-05, "loss": 0.5816, "step": 5621 }, { "epoch": 0.20029570514990114, "grad_norm": 1.2692249820334667, "learning_rate": 1.8517571650536056e-05, "loss": 0.633, "step": 5622 }, { "epoch": 0.20033133227639524, "grad_norm": 1.6540706643898289, "learning_rate": 1.851696700397155e-05, "loss": 0.6016, "step": 5623 }, { "epoch": 0.20036695940288937, "grad_norm": 1.670718591557882, "learning_rate": 1.8516362243997677e-05, "loss": 0.6283, "step": 5624 }, { "epoch": 0.20040258652938348, "grad_norm": 1.659105504586382, "learning_rate": 1.8515757370622487e-05, "loss": 0.6382, "step": 5625 }, { "epoch": 0.20043821365587758, "grad_norm": 1.667166045161789, "learning_rate": 1.8515152383854034e-05, "loss": 0.5822, "step": 5626 }, { "epoch": 0.2004738407823717, "grad_norm": 1.258410739512587, "learning_rate": 1.8514547283700376e-05, "loss": 0.5554, "step": 5627 }, { "epoch": 0.2005094679088658, "grad_norm": 1.5619813677058851, "learning_rate": 1.8513942070169572e-05, "loss": 0.6033, "step": 5628 }, { "epoch": 0.20054509503535992, "grad_norm": 1.2257883001678649, "learning_rate": 1.8513336743269676e-05, "loss": 0.5729, "step": 5629 }, { "epoch": 0.20058072216185405, "grad_norm": 1.2538741918121399, "learning_rate": 1.8512731303008752e-05, "loss": 0.5154, "step": 5630 }, { "epoch": 0.20061634928834815, "grad_norm": 1.8609879089030483, "learning_rate": 1.8512125749394857e-05, "loss": 0.5614, "step": 5631 }, { "epoch": 0.20065197641484225, "grad_norm": 1.1702124881428742, "learning_rate": 1.8511520082436064e-05, "loss": 0.477, "step": 5632 }, { "epoch": 0.20068760354133638, "grad_norm": 1.7933725394444575, "learning_rate": 1.8510914302140428e-05, "loss": 0.5395, "step": 5633 }, { "epoch": 0.2007232306678305, "grad_norm": 2.0733632484457147, "learning_rate": 1.851030840851602e-05, "loss": 0.6668, "step": 5634 }, { "epoch": 0.2007588577943246, "grad_norm": 2.4368347294996546, "learning_rate": 1.8509702401570906e-05, "loss": 0.6032, "step": 5635 }, { "epoch": 0.20079448492081872, "grad_norm": 1.1559649811291308, "learning_rate": 1.8509096281313158e-05, "loss": 0.5084, "step": 5636 }, { "epoch": 0.20083011204731283, "grad_norm": 1.2719219079221673, "learning_rate": 1.8508490047750845e-05, "loss": 0.51, "step": 5637 }, { "epoch": 0.20086573917380693, "grad_norm": 1.7092225521756361, "learning_rate": 1.8507883700892042e-05, "loss": 0.6582, "step": 5638 }, { "epoch": 0.20090136630030106, "grad_norm": 0.9763169290908351, "learning_rate": 1.8507277240744818e-05, "loss": 0.4848, "step": 5639 }, { "epoch": 0.20093699342679516, "grad_norm": 1.6002457937133214, "learning_rate": 1.8506670667317253e-05, "loss": 0.6695, "step": 5640 }, { "epoch": 0.20097262055328927, "grad_norm": 1.8672845944153802, "learning_rate": 1.8506063980617423e-05, "loss": 0.5704, "step": 5641 }, { "epoch": 0.2010082476797834, "grad_norm": 1.8095567348375798, "learning_rate": 1.8505457180653405e-05, "loss": 0.6218, "step": 5642 }, { "epoch": 0.2010438748062775, "grad_norm": 1.2724811003846412, "learning_rate": 1.8504850267433278e-05, "loss": 0.5235, "step": 5643 }, { "epoch": 0.2010795019327716, "grad_norm": 1.1659118193160343, "learning_rate": 1.8504243240965128e-05, "loss": 0.5294, "step": 5644 }, { "epoch": 0.20111512905926573, "grad_norm": 1.5005916692058914, "learning_rate": 1.850363610125703e-05, "loss": 0.5995, "step": 5645 }, { "epoch": 0.20115075618575984, "grad_norm": 1.6358737463774422, "learning_rate": 1.850302884831708e-05, "loss": 0.6033, "step": 5646 }, { "epoch": 0.20118638331225394, "grad_norm": 1.1090029857928003, "learning_rate": 1.8502421482153354e-05, "loss": 0.498, "step": 5647 }, { "epoch": 0.20122201043874807, "grad_norm": 1.4784615198411541, "learning_rate": 1.8501814002773946e-05, "loss": 0.6138, "step": 5648 }, { "epoch": 0.20125763756524218, "grad_norm": 1.0460184080287682, "learning_rate": 1.8501206410186938e-05, "loss": 0.4761, "step": 5649 }, { "epoch": 0.20129326469173628, "grad_norm": 2.1561756316679843, "learning_rate": 1.8500598704400427e-05, "loss": 0.6164, "step": 5650 }, { "epoch": 0.2013288918182304, "grad_norm": 1.1874127209146061, "learning_rate": 1.84999908854225e-05, "loss": 0.577, "step": 5651 }, { "epoch": 0.2013645189447245, "grad_norm": 1.0628171253832948, "learning_rate": 1.849938295326126e-05, "loss": 0.4668, "step": 5652 }, { "epoch": 0.20140014607121862, "grad_norm": 1.1765414051119993, "learning_rate": 1.849877490792479e-05, "loss": 0.5334, "step": 5653 }, { "epoch": 0.20143577319771275, "grad_norm": 1.1614765735216666, "learning_rate": 1.849816674942119e-05, "loss": 0.5315, "step": 5654 }, { "epoch": 0.20147140032420685, "grad_norm": 1.5683303012266818, "learning_rate": 1.8497558477758565e-05, "loss": 0.5031, "step": 5655 }, { "epoch": 0.20150702745070095, "grad_norm": 1.177635051602635, "learning_rate": 1.8496950092945006e-05, "loss": 0.5413, "step": 5656 }, { "epoch": 0.20154265457719509, "grad_norm": 1.2933544943938604, "learning_rate": 1.8496341594988617e-05, "loss": 0.5031, "step": 5657 }, { "epoch": 0.2015782817036892, "grad_norm": 1.3486568912440422, "learning_rate": 1.8495732983897504e-05, "loss": 0.5251, "step": 5658 }, { "epoch": 0.2016139088301833, "grad_norm": 1.5464780576786787, "learning_rate": 1.8495124259679767e-05, "loss": 0.5605, "step": 5659 }, { "epoch": 0.20164953595667742, "grad_norm": 1.093852775440165, "learning_rate": 1.849451542234351e-05, "loss": 0.506, "step": 5660 }, { "epoch": 0.20168516308317153, "grad_norm": 1.1356203220150785, "learning_rate": 1.8493906471896846e-05, "loss": 0.52, "step": 5661 }, { "epoch": 0.20172079020966563, "grad_norm": 2.0577274522881797, "learning_rate": 1.8493297408347882e-05, "loss": 0.6366, "step": 5662 }, { "epoch": 0.20175641733615976, "grad_norm": 1.2300809278779918, "learning_rate": 1.849268823170472e-05, "loss": 0.5509, "step": 5663 }, { "epoch": 0.20179204446265386, "grad_norm": 1.1928690514224776, "learning_rate": 1.8492078941975486e-05, "loss": 0.5085, "step": 5664 }, { "epoch": 0.20182767158914797, "grad_norm": 1.087040532233632, "learning_rate": 1.849146953916828e-05, "loss": 0.5094, "step": 5665 }, { "epoch": 0.2018632987156421, "grad_norm": 1.7085684716346992, "learning_rate": 1.8490860023291228e-05, "loss": 0.6007, "step": 5666 }, { "epoch": 0.2018989258421362, "grad_norm": 1.8727396928770585, "learning_rate": 1.8490250394352433e-05, "loss": 0.6676, "step": 5667 }, { "epoch": 0.2019345529686303, "grad_norm": 1.8645150239105925, "learning_rate": 1.8489640652360025e-05, "loss": 0.4315, "step": 5668 }, { "epoch": 0.20197018009512444, "grad_norm": 1.2899240397324139, "learning_rate": 1.8489030797322112e-05, "loss": 0.5498, "step": 5669 }, { "epoch": 0.20200580722161854, "grad_norm": 1.2597966686354958, "learning_rate": 1.8488420829246826e-05, "loss": 0.5228, "step": 5670 }, { "epoch": 0.20204143434811264, "grad_norm": 1.2344684361901075, "learning_rate": 1.848781074814228e-05, "loss": 0.5259, "step": 5671 }, { "epoch": 0.20207706147460677, "grad_norm": 2.4535702132838844, "learning_rate": 1.8487200554016602e-05, "loss": 0.6236, "step": 5672 }, { "epoch": 0.20211268860110088, "grad_norm": 1.1572342501420962, "learning_rate": 1.8486590246877916e-05, "loss": 0.5226, "step": 5673 }, { "epoch": 0.20214831572759498, "grad_norm": 1.1601516115331503, "learning_rate": 1.848597982673435e-05, "loss": 0.5476, "step": 5674 }, { "epoch": 0.2021839428540891, "grad_norm": 1.525655348220368, "learning_rate": 1.8485369293594033e-05, "loss": 0.6596, "step": 5675 }, { "epoch": 0.2022195699805832, "grad_norm": 1.7530189472487236, "learning_rate": 1.848475864746509e-05, "loss": 0.5499, "step": 5676 }, { "epoch": 0.20225519710707732, "grad_norm": 1.5241311506571213, "learning_rate": 1.8484147888355656e-05, "loss": 0.5648, "step": 5677 }, { "epoch": 0.20229082423357145, "grad_norm": 1.1778986931934283, "learning_rate": 1.8483537016273866e-05, "loss": 0.5481, "step": 5678 }, { "epoch": 0.20232645136006555, "grad_norm": 1.564079989152251, "learning_rate": 1.8482926031227845e-05, "loss": 0.5698, "step": 5679 }, { "epoch": 0.20236207848655965, "grad_norm": 1.7678726884657503, "learning_rate": 1.848231493322574e-05, "loss": 0.6121, "step": 5680 }, { "epoch": 0.20239770561305379, "grad_norm": 1.3576267042840342, "learning_rate": 1.848170372227568e-05, "loss": 0.5345, "step": 5681 }, { "epoch": 0.2024333327395479, "grad_norm": 1.6703550982324051, "learning_rate": 1.8481092398385805e-05, "loss": 0.5653, "step": 5682 }, { "epoch": 0.202468959866042, "grad_norm": 1.3066148500305867, "learning_rate": 1.848048096156426e-05, "loss": 0.5846, "step": 5683 }, { "epoch": 0.20250458699253612, "grad_norm": 2.299272326148446, "learning_rate": 1.8479869411819183e-05, "loss": 0.624, "step": 5684 }, { "epoch": 0.20254021411903023, "grad_norm": 2.048639196127395, "learning_rate": 1.8479257749158717e-05, "loss": 0.6394, "step": 5685 }, { "epoch": 0.20257584124552433, "grad_norm": 1.760971645042986, "learning_rate": 1.847864597359101e-05, "loss": 0.597, "step": 5686 }, { "epoch": 0.20261146837201846, "grad_norm": 1.8769421755002385, "learning_rate": 1.8478034085124204e-05, "loss": 0.5409, "step": 5687 }, { "epoch": 0.20264709549851256, "grad_norm": 1.8493664674434498, "learning_rate": 1.8477422083766446e-05, "loss": 0.6863, "step": 5688 }, { "epoch": 0.20268272262500667, "grad_norm": 1.2500133398547748, "learning_rate": 1.8476809969525892e-05, "loss": 0.5141, "step": 5689 }, { "epoch": 0.2027183497515008, "grad_norm": 1.6558746197022263, "learning_rate": 1.8476197742410683e-05, "loss": 0.6061, "step": 5690 }, { "epoch": 0.2027539768779949, "grad_norm": 1.8683679409130578, "learning_rate": 1.847558540242898e-05, "loss": 0.4994, "step": 5691 }, { "epoch": 0.202789604004489, "grad_norm": 1.325666413104182, "learning_rate": 1.847497294958893e-05, "loss": 0.5277, "step": 5692 }, { "epoch": 0.20282523113098314, "grad_norm": 1.1476983904576423, "learning_rate": 1.8474360383898694e-05, "loss": 0.5278, "step": 5693 }, { "epoch": 0.20286085825747724, "grad_norm": 1.2187377700522009, "learning_rate": 1.8473747705366427e-05, "loss": 0.4782, "step": 5694 }, { "epoch": 0.20289648538397134, "grad_norm": 1.179892818741776, "learning_rate": 1.8473134914000284e-05, "loss": 0.53, "step": 5695 }, { "epoch": 0.20293211251046547, "grad_norm": 1.6959170955255827, "learning_rate": 1.847252200980843e-05, "loss": 0.5906, "step": 5696 }, { "epoch": 0.20296773963695958, "grad_norm": 1.1404630795556854, "learning_rate": 1.8471908992799024e-05, "loss": 0.4971, "step": 5697 }, { "epoch": 0.20300336676345368, "grad_norm": 1.3575575200679355, "learning_rate": 1.8471295862980226e-05, "loss": 0.5776, "step": 5698 }, { "epoch": 0.2030389938899478, "grad_norm": 1.2305021162259246, "learning_rate": 1.8470682620360202e-05, "loss": 0.4934, "step": 5699 }, { "epoch": 0.20307462101644191, "grad_norm": 1.7767091794140597, "learning_rate": 1.847006926494712e-05, "loss": 0.6878, "step": 5700 }, { "epoch": 0.20311024814293605, "grad_norm": 1.2575914560495223, "learning_rate": 1.8469455796749148e-05, "loss": 0.5517, "step": 5701 }, { "epoch": 0.20314587526943015, "grad_norm": 1.5919772372749121, "learning_rate": 1.8468842215774447e-05, "loss": 0.6317, "step": 5702 }, { "epoch": 0.20318150239592425, "grad_norm": 1.284751339136499, "learning_rate": 1.8468228522031197e-05, "loss": 0.4841, "step": 5703 }, { "epoch": 0.20321712952241838, "grad_norm": 1.5571535276041157, "learning_rate": 1.8467614715527563e-05, "loss": 0.5675, "step": 5704 }, { "epoch": 0.20325275664891249, "grad_norm": 1.166233141098886, "learning_rate": 1.8467000796271717e-05, "loss": 0.4818, "step": 5705 }, { "epoch": 0.2032883837754066, "grad_norm": 1.8172001606837267, "learning_rate": 1.8466386764271843e-05, "loss": 0.5391, "step": 5706 }, { "epoch": 0.20332401090190072, "grad_norm": 1.2557687500202142, "learning_rate": 1.846577261953611e-05, "loss": 0.5794, "step": 5707 }, { "epoch": 0.20335963802839482, "grad_norm": 1.1065701256850264, "learning_rate": 1.8465158362072696e-05, "loss": 0.5232, "step": 5708 }, { "epoch": 0.20339526515488893, "grad_norm": 1.1553639144119607, "learning_rate": 1.8464543991889784e-05, "loss": 0.5402, "step": 5709 }, { "epoch": 0.20343089228138306, "grad_norm": 1.7344806131798387, "learning_rate": 1.846392950899555e-05, "loss": 0.6608, "step": 5710 }, { "epoch": 0.20346651940787716, "grad_norm": 1.2525896073721186, "learning_rate": 1.846331491339818e-05, "loss": 0.5522, "step": 5711 }, { "epoch": 0.20350214653437126, "grad_norm": 1.2433015732012462, "learning_rate": 1.8462700205105855e-05, "loss": 0.5115, "step": 5712 }, { "epoch": 0.2035377736608654, "grad_norm": 1.1479895802078872, "learning_rate": 1.846208538412676e-05, "loss": 0.4449, "step": 5713 }, { "epoch": 0.2035734007873595, "grad_norm": 1.5850353430414639, "learning_rate": 1.8461470450469086e-05, "loss": 0.619, "step": 5714 }, { "epoch": 0.2036090279138536, "grad_norm": 1.1361790142076618, "learning_rate": 1.8460855404141024e-05, "loss": 0.5837, "step": 5715 }, { "epoch": 0.20364465504034773, "grad_norm": 1.1503406593559324, "learning_rate": 1.846024024515075e-05, "loss": 0.4901, "step": 5716 }, { "epoch": 0.20368028216684184, "grad_norm": 1.6470439266701504, "learning_rate": 1.8459624973506467e-05, "loss": 0.6755, "step": 5717 }, { "epoch": 0.20371590929333594, "grad_norm": 1.137032729164237, "learning_rate": 1.8459009589216364e-05, "loss": 0.5538, "step": 5718 }, { "epoch": 0.20375153641983007, "grad_norm": 2.1415046925906336, "learning_rate": 1.845839409228864e-05, "loss": 0.6477, "step": 5719 }, { "epoch": 0.20378716354632417, "grad_norm": 1.26239911056258, "learning_rate": 1.8457778482731484e-05, "loss": 0.5164, "step": 5720 }, { "epoch": 0.20382279067281828, "grad_norm": 1.4143803628757021, "learning_rate": 1.8457162760553092e-05, "loss": 0.5865, "step": 5721 }, { "epoch": 0.2038584177993124, "grad_norm": 1.266075977836505, "learning_rate": 1.845654692576167e-05, "loss": 0.5178, "step": 5722 }, { "epoch": 0.2038940449258065, "grad_norm": 1.1339609543375209, "learning_rate": 1.8455930978365415e-05, "loss": 0.5295, "step": 5723 }, { "epoch": 0.20392967205230061, "grad_norm": 1.2077020483028285, "learning_rate": 1.845531491837253e-05, "loss": 0.554, "step": 5724 }, { "epoch": 0.20396529917879475, "grad_norm": 1.8053309432281115, "learning_rate": 1.845469874579121e-05, "loss": 0.621, "step": 5725 }, { "epoch": 0.20400092630528885, "grad_norm": 1.141903383837622, "learning_rate": 1.8454082460629677e-05, "loss": 0.5389, "step": 5726 }, { "epoch": 0.20403655343178295, "grad_norm": 1.6467543069738497, "learning_rate": 1.8453466062896122e-05, "loss": 0.6068, "step": 5727 }, { "epoch": 0.20407218055827708, "grad_norm": 1.24198090600118, "learning_rate": 1.8452849552598757e-05, "loss": 0.5633, "step": 5728 }, { "epoch": 0.2041078076847712, "grad_norm": 1.7021239716959842, "learning_rate": 1.8452232929745793e-05, "loss": 0.5436, "step": 5729 }, { "epoch": 0.2041434348112653, "grad_norm": 1.6225325291925585, "learning_rate": 1.845161619434544e-05, "loss": 0.5864, "step": 5730 }, { "epoch": 0.20417906193775942, "grad_norm": 1.2004233315258368, "learning_rate": 1.845099934640591e-05, "loss": 0.4847, "step": 5731 }, { "epoch": 0.20421468906425352, "grad_norm": 1.09913862565342, "learning_rate": 1.8450382385935414e-05, "loss": 0.4731, "step": 5732 }, { "epoch": 0.20425031619074763, "grad_norm": 1.503139156839305, "learning_rate": 1.8449765312942172e-05, "loss": 0.5923, "step": 5733 }, { "epoch": 0.20428594331724176, "grad_norm": 2.5775767297496865, "learning_rate": 1.84491481274344e-05, "loss": 0.5635, "step": 5734 }, { "epoch": 0.20432157044373586, "grad_norm": 1.3138369848764506, "learning_rate": 1.8448530829420313e-05, "loss": 0.5168, "step": 5735 }, { "epoch": 0.20435719757022996, "grad_norm": 1.118367206467647, "learning_rate": 1.8447913418908136e-05, "loss": 0.4946, "step": 5736 }, { "epoch": 0.2043928246967241, "grad_norm": 1.3793304035878955, "learning_rate": 1.8447295895906086e-05, "loss": 0.567, "step": 5737 }, { "epoch": 0.2044284518232182, "grad_norm": 1.7676679602967402, "learning_rate": 1.8446678260422388e-05, "loss": 0.5375, "step": 5738 }, { "epoch": 0.2044640789497123, "grad_norm": 1.3742179420216516, "learning_rate": 1.844606051246526e-05, "loss": 0.5181, "step": 5739 }, { "epoch": 0.20449970607620643, "grad_norm": 1.224562400694924, "learning_rate": 1.8445442652042937e-05, "loss": 0.5265, "step": 5740 }, { "epoch": 0.20453533320270054, "grad_norm": 1.804197237213769, "learning_rate": 1.844482467916364e-05, "loss": 0.6377, "step": 5741 }, { "epoch": 0.20457096032919464, "grad_norm": 1.7982601359989947, "learning_rate": 1.84442065938356e-05, "loss": 0.6231, "step": 5742 }, { "epoch": 0.20460658745568877, "grad_norm": 1.8606937596258322, "learning_rate": 1.844358839606705e-05, "loss": 0.5828, "step": 5743 }, { "epoch": 0.20464221458218287, "grad_norm": 1.6602161481448925, "learning_rate": 1.8442970085866214e-05, "loss": 0.5724, "step": 5744 }, { "epoch": 0.20467784170867698, "grad_norm": 1.9941736908794696, "learning_rate": 1.8442351663241337e-05, "loss": 0.6757, "step": 5745 }, { "epoch": 0.2047134688351711, "grad_norm": 1.852079145683076, "learning_rate": 1.844173312820064e-05, "loss": 0.7099, "step": 5746 }, { "epoch": 0.2047490959616652, "grad_norm": 1.1802967181726656, "learning_rate": 1.844111448075237e-05, "loss": 0.4952, "step": 5747 }, { "epoch": 0.20478472308815932, "grad_norm": 1.2399163099081043, "learning_rate": 1.8440495720904758e-05, "loss": 0.4871, "step": 5748 }, { "epoch": 0.20482035021465345, "grad_norm": 1.1035915627678277, "learning_rate": 1.8439876848666048e-05, "loss": 0.5423, "step": 5749 }, { "epoch": 0.20485597734114755, "grad_norm": 2.3511853401019294, "learning_rate": 1.8439257864044476e-05, "loss": 0.6356, "step": 5750 }, { "epoch": 0.20489160446764165, "grad_norm": 1.1293990841927486, "learning_rate": 1.8438638767048287e-05, "loss": 0.5005, "step": 5751 }, { "epoch": 0.20492723159413578, "grad_norm": 2.3932753887124907, "learning_rate": 1.8438019557685727e-05, "loss": 0.6591, "step": 5752 }, { "epoch": 0.2049628587206299, "grad_norm": 1.2271902216039485, "learning_rate": 1.843740023596504e-05, "loss": 0.493, "step": 5753 }, { "epoch": 0.204998485847124, "grad_norm": 2.06360478402138, "learning_rate": 1.8436780801894468e-05, "loss": 0.6378, "step": 5754 }, { "epoch": 0.20503411297361812, "grad_norm": 1.7417461539877372, "learning_rate": 1.8436161255482265e-05, "loss": 0.6076, "step": 5755 }, { "epoch": 0.20506974010011222, "grad_norm": 1.2837305510477488, "learning_rate": 1.8435541596736675e-05, "loss": 0.5519, "step": 5756 }, { "epoch": 0.20510536722660633, "grad_norm": 1.9651430687534932, "learning_rate": 1.8434921825665958e-05, "loss": 0.6511, "step": 5757 }, { "epoch": 0.20514099435310046, "grad_norm": 1.8262550309923564, "learning_rate": 1.8434301942278357e-05, "loss": 0.6252, "step": 5758 }, { "epoch": 0.20517662147959456, "grad_norm": 1.2592146860002864, "learning_rate": 1.843368194658213e-05, "loss": 0.5983, "step": 5759 }, { "epoch": 0.20521224860608867, "grad_norm": 1.8338316188824735, "learning_rate": 1.8433061838585537e-05, "loss": 0.5684, "step": 5760 }, { "epoch": 0.2052478757325828, "grad_norm": 1.233193635325145, "learning_rate": 1.8432441618296828e-05, "loss": 0.5696, "step": 5761 }, { "epoch": 0.2052835028590769, "grad_norm": 1.2204916021475658, "learning_rate": 1.8431821285724263e-05, "loss": 0.5461, "step": 5762 }, { "epoch": 0.205319129985571, "grad_norm": 1.4005571690686858, "learning_rate": 1.843120084087611e-05, "loss": 0.6199, "step": 5763 }, { "epoch": 0.20535475711206513, "grad_norm": 1.6629656158427593, "learning_rate": 1.8430580283760623e-05, "loss": 0.6514, "step": 5764 }, { "epoch": 0.20539038423855924, "grad_norm": 1.270920672292916, "learning_rate": 1.8429959614386064e-05, "loss": 0.5629, "step": 5765 }, { "epoch": 0.20542601136505334, "grad_norm": 1.7558590304240176, "learning_rate": 1.84293388327607e-05, "loss": 0.5193, "step": 5766 }, { "epoch": 0.20546163849154747, "grad_norm": 1.2504140383138536, "learning_rate": 1.84287179388928e-05, "loss": 0.5379, "step": 5767 }, { "epoch": 0.20549726561804157, "grad_norm": 2.015153888021099, "learning_rate": 1.842809693279063e-05, "loss": 0.6156, "step": 5768 }, { "epoch": 0.20553289274453568, "grad_norm": 1.0760186395235833, "learning_rate": 1.8427475814462458e-05, "loss": 0.5168, "step": 5769 }, { "epoch": 0.2055685198710298, "grad_norm": 1.915518168987928, "learning_rate": 1.8426854583916554e-05, "loss": 0.63, "step": 5770 }, { "epoch": 0.2056041469975239, "grad_norm": 1.2620886233942705, "learning_rate": 1.8426233241161193e-05, "loss": 0.4872, "step": 5771 }, { "epoch": 0.20563977412401802, "grad_norm": 1.9671804796577348, "learning_rate": 1.8425611786204647e-05, "loss": 0.6706, "step": 5772 }, { "epoch": 0.20567540125051215, "grad_norm": 2.068250705887194, "learning_rate": 1.842499021905519e-05, "loss": 0.6063, "step": 5773 }, { "epoch": 0.20571102837700625, "grad_norm": 1.7117310108631965, "learning_rate": 1.8424368539721098e-05, "loss": 0.637, "step": 5774 }, { "epoch": 0.20574665550350035, "grad_norm": 2.049356071547932, "learning_rate": 1.842374674821065e-05, "loss": 0.6706, "step": 5775 }, { "epoch": 0.20578228262999448, "grad_norm": 1.1342808405963785, "learning_rate": 1.842312484453213e-05, "loss": 0.5203, "step": 5776 }, { "epoch": 0.2058179097564886, "grad_norm": 2.526909813103969, "learning_rate": 1.8422502828693814e-05, "loss": 0.5355, "step": 5777 }, { "epoch": 0.2058535368829827, "grad_norm": 1.1782648853884194, "learning_rate": 1.8421880700703983e-05, "loss": 0.5181, "step": 5778 }, { "epoch": 0.20588916400947682, "grad_norm": 1.8513826331093548, "learning_rate": 1.8421258460570927e-05, "loss": 0.6123, "step": 5779 }, { "epoch": 0.20592479113597092, "grad_norm": 1.219757918672933, "learning_rate": 1.8420636108302926e-05, "loss": 0.521, "step": 5780 }, { "epoch": 0.20596041826246503, "grad_norm": 1.1245161910666888, "learning_rate": 1.8420013643908272e-05, "loss": 0.4952, "step": 5781 }, { "epoch": 0.20599604538895916, "grad_norm": 1.557279848202358, "learning_rate": 1.8419391067395248e-05, "loss": 0.6057, "step": 5782 }, { "epoch": 0.20603167251545326, "grad_norm": 1.2502092874044366, "learning_rate": 1.841876837877215e-05, "loss": 0.5273, "step": 5783 }, { "epoch": 0.20606729964194737, "grad_norm": 1.4871873175268064, "learning_rate": 1.841814557804727e-05, "loss": 0.6297, "step": 5784 }, { "epoch": 0.2061029267684415, "grad_norm": 1.2268669884629746, "learning_rate": 1.8417522665228894e-05, "loss": 0.5188, "step": 5785 }, { "epoch": 0.2061385538949356, "grad_norm": 1.2385501038698017, "learning_rate": 1.8416899640325318e-05, "loss": 0.5032, "step": 5786 }, { "epoch": 0.2061741810214297, "grad_norm": 1.0663850243841786, "learning_rate": 1.8416276503344843e-05, "loss": 0.5079, "step": 5787 }, { "epoch": 0.20620980814792383, "grad_norm": 1.0541128405216924, "learning_rate": 1.841565325429576e-05, "loss": 0.47, "step": 5788 }, { "epoch": 0.20624543527441794, "grad_norm": 1.1312564855544345, "learning_rate": 1.8415029893186375e-05, "loss": 0.4888, "step": 5789 }, { "epoch": 0.20628106240091204, "grad_norm": 1.2270247327440122, "learning_rate": 1.8414406420024984e-05, "loss": 0.5442, "step": 5790 }, { "epoch": 0.20631668952740617, "grad_norm": 1.6727314040254522, "learning_rate": 1.8413782834819895e-05, "loss": 0.6302, "step": 5791 }, { "epoch": 0.20635231665390028, "grad_norm": 1.0512114261321617, "learning_rate": 1.84131591375794e-05, "loss": 0.4808, "step": 5792 }, { "epoch": 0.2063879437803944, "grad_norm": 1.0983839764652148, "learning_rate": 1.8412535328311813e-05, "loss": 0.5568, "step": 5793 }, { "epoch": 0.2064235709068885, "grad_norm": 1.7856433510547818, "learning_rate": 1.841191140702544e-05, "loss": 0.595, "step": 5794 }, { "epoch": 0.2064591980333826, "grad_norm": 1.438650125859569, "learning_rate": 1.8411287373728584e-05, "loss": 0.5003, "step": 5795 }, { "epoch": 0.20649482515987674, "grad_norm": 1.308110979031433, "learning_rate": 1.8410663228429558e-05, "loss": 0.5276, "step": 5796 }, { "epoch": 0.20653045228637085, "grad_norm": 1.1901359521036123, "learning_rate": 1.8410038971136673e-05, "loss": 0.5503, "step": 5797 }, { "epoch": 0.20656607941286495, "grad_norm": 1.5126216422622412, "learning_rate": 1.8409414601858243e-05, "loss": 0.6962, "step": 5798 }, { "epoch": 0.20660170653935908, "grad_norm": 2.1261145490629407, "learning_rate": 1.840879012060258e-05, "loss": 0.637, "step": 5799 }, { "epoch": 0.20663733366585318, "grad_norm": 3.2099002689151055, "learning_rate": 1.8408165527377995e-05, "loss": 0.5698, "step": 5800 }, { "epoch": 0.2066729607923473, "grad_norm": 1.5366402428665937, "learning_rate": 1.840754082219281e-05, "loss": 0.7138, "step": 5801 }, { "epoch": 0.20670858791884142, "grad_norm": 1.2353775810198846, "learning_rate": 1.8406916005055345e-05, "loss": 0.5478, "step": 5802 }, { "epoch": 0.20674421504533552, "grad_norm": 2.1187536340826822, "learning_rate": 1.8406291075973914e-05, "loss": 0.6606, "step": 5803 }, { "epoch": 0.20677984217182963, "grad_norm": 1.1675604200197853, "learning_rate": 1.8405666034956842e-05, "loss": 0.5506, "step": 5804 }, { "epoch": 0.20681546929832376, "grad_norm": 1.13503683836689, "learning_rate": 1.8405040882012454e-05, "loss": 0.5562, "step": 5805 }, { "epoch": 0.20685109642481786, "grad_norm": 1.094996872707747, "learning_rate": 1.8404415617149073e-05, "loss": 0.5374, "step": 5806 }, { "epoch": 0.20688672355131196, "grad_norm": 1.1082755980654921, "learning_rate": 1.8403790240375022e-05, "loss": 0.4631, "step": 5807 }, { "epoch": 0.2069223506778061, "grad_norm": 1.2010348560172097, "learning_rate": 1.840316475169863e-05, "loss": 0.5105, "step": 5808 }, { "epoch": 0.2069579778043002, "grad_norm": 1.1109296367531827, "learning_rate": 1.8402539151128224e-05, "loss": 0.5782, "step": 5809 }, { "epoch": 0.2069936049307943, "grad_norm": 1.8769988800503736, "learning_rate": 1.840191343867214e-05, "loss": 0.5747, "step": 5810 }, { "epoch": 0.20702923205728843, "grad_norm": 1.2408578203141947, "learning_rate": 1.84012876143387e-05, "loss": 0.5208, "step": 5811 }, { "epoch": 0.20706485918378253, "grad_norm": 1.9245276874858157, "learning_rate": 1.840066167813625e-05, "loss": 0.5901, "step": 5812 }, { "epoch": 0.20710048631027664, "grad_norm": 1.1855012989039317, "learning_rate": 1.8400035630073114e-05, "loss": 0.492, "step": 5813 }, { "epoch": 0.20713611343677077, "grad_norm": 1.0978852474478276, "learning_rate": 1.839940947015763e-05, "loss": 0.5532, "step": 5814 }, { "epoch": 0.20717174056326487, "grad_norm": 1.1267483193601653, "learning_rate": 1.8398783198398146e-05, "loss": 0.5795, "step": 5815 }, { "epoch": 0.20720736768975898, "grad_norm": 1.0477704174949498, "learning_rate": 1.8398156814802987e-05, "loss": 0.4915, "step": 5816 }, { "epoch": 0.2072429948162531, "grad_norm": 1.1133371333868043, "learning_rate": 1.83975303193805e-05, "loss": 0.5137, "step": 5817 }, { "epoch": 0.2072786219427472, "grad_norm": 1.0136625886480775, "learning_rate": 1.839690371213903e-05, "loss": 0.4712, "step": 5818 }, { "epoch": 0.2073142490692413, "grad_norm": 2.851367002248305, "learning_rate": 1.8396276993086916e-05, "loss": 0.5681, "step": 5819 }, { "epoch": 0.20734987619573544, "grad_norm": 1.029174840811311, "learning_rate": 1.839565016223251e-05, "loss": 0.5246, "step": 5820 }, { "epoch": 0.20738550332222955, "grad_norm": 1.671469047210659, "learning_rate": 1.8395023219584146e-05, "loss": 0.6263, "step": 5821 }, { "epoch": 0.20742113044872365, "grad_norm": 1.1831392135478378, "learning_rate": 1.8394396165150185e-05, "loss": 0.5133, "step": 5822 }, { "epoch": 0.20745675757521778, "grad_norm": 1.5728386402702765, "learning_rate": 1.839376899893897e-05, "loss": 0.6589, "step": 5823 }, { "epoch": 0.20749238470171188, "grad_norm": 1.131074651197177, "learning_rate": 1.8393141720958856e-05, "loss": 0.5386, "step": 5824 }, { "epoch": 0.207528011828206, "grad_norm": 1.1455821249376457, "learning_rate": 1.8392514331218196e-05, "loss": 0.482, "step": 5825 }, { "epoch": 0.20756363895470012, "grad_norm": 1.138663708422406, "learning_rate": 1.8391886829725334e-05, "loss": 0.4838, "step": 5826 }, { "epoch": 0.20759926608119422, "grad_norm": 1.1291862855019166, "learning_rate": 1.8391259216488637e-05, "loss": 0.4888, "step": 5827 }, { "epoch": 0.20763489320768833, "grad_norm": 1.647013298620329, "learning_rate": 1.839063149151646e-05, "loss": 0.591, "step": 5828 }, { "epoch": 0.20767052033418246, "grad_norm": 1.077035640344425, "learning_rate": 1.839000365481716e-05, "loss": 0.5134, "step": 5829 }, { "epoch": 0.20770614746067656, "grad_norm": 1.22244765644603, "learning_rate": 1.8389375706399097e-05, "loss": 0.5244, "step": 5830 }, { "epoch": 0.20774177458717066, "grad_norm": 1.300240529993693, "learning_rate": 1.838874764627063e-05, "loss": 0.5003, "step": 5831 }, { "epoch": 0.2077774017136648, "grad_norm": 1.1486922811951799, "learning_rate": 1.8388119474440125e-05, "loss": 0.4761, "step": 5832 }, { "epoch": 0.2078130288401589, "grad_norm": 1.177469310157926, "learning_rate": 1.838749119091595e-05, "loss": 0.4524, "step": 5833 }, { "epoch": 0.207848655966653, "grad_norm": 1.0882933331200173, "learning_rate": 1.8386862795706462e-05, "loss": 0.5274, "step": 5834 }, { "epoch": 0.20788428309314713, "grad_norm": 1.1886678737175729, "learning_rate": 1.8386234288820034e-05, "loss": 0.5628, "step": 5835 }, { "epoch": 0.20791991021964124, "grad_norm": 1.73613901642778, "learning_rate": 1.8385605670265038e-05, "loss": 0.6914, "step": 5836 }, { "epoch": 0.20795553734613534, "grad_norm": 1.7533545800350436, "learning_rate": 1.8384976940049842e-05, "loss": 0.6479, "step": 5837 }, { "epoch": 0.20799116447262947, "grad_norm": 1.0912408882783156, "learning_rate": 1.8384348098182815e-05, "loss": 0.5239, "step": 5838 }, { "epoch": 0.20802679159912357, "grad_norm": 1.6105584689563426, "learning_rate": 1.8383719144672333e-05, "loss": 0.6407, "step": 5839 }, { "epoch": 0.20806241872561768, "grad_norm": 1.7205252618143811, "learning_rate": 1.8383090079526774e-05, "loss": 0.6378, "step": 5840 }, { "epoch": 0.2080980458521118, "grad_norm": 1.1468582063869335, "learning_rate": 1.8382460902754508e-05, "loss": 0.4958, "step": 5841 }, { "epoch": 0.2081336729786059, "grad_norm": 1.196490507247224, "learning_rate": 1.8381831614363915e-05, "loss": 0.567, "step": 5842 }, { "epoch": 0.2081693001051, "grad_norm": 1.106230704250586, "learning_rate": 1.838120221436338e-05, "loss": 0.525, "step": 5843 }, { "epoch": 0.20820492723159414, "grad_norm": 1.1708970068955298, "learning_rate": 1.8380572702761276e-05, "loss": 0.5293, "step": 5844 }, { "epoch": 0.20824055435808825, "grad_norm": 1.115746616703397, "learning_rate": 1.8379943079565987e-05, "loss": 0.5128, "step": 5845 }, { "epoch": 0.20827618148458235, "grad_norm": 1.6534439163963937, "learning_rate": 1.83793133447859e-05, "loss": 0.6953, "step": 5846 }, { "epoch": 0.20831180861107648, "grad_norm": 1.090508755994455, "learning_rate": 1.83786834984294e-05, "loss": 0.4687, "step": 5847 }, { "epoch": 0.20834743573757059, "grad_norm": 1.0431073352777969, "learning_rate": 1.8378053540504874e-05, "loss": 0.4559, "step": 5848 }, { "epoch": 0.2083830628640647, "grad_norm": 2.1484235554449396, "learning_rate": 1.8377423471020708e-05, "loss": 0.6866, "step": 5849 }, { "epoch": 0.20841868999055882, "grad_norm": 1.6812073947802808, "learning_rate": 1.8376793289985295e-05, "loss": 0.5639, "step": 5850 }, { "epoch": 0.20845431711705292, "grad_norm": 1.4209997033292556, "learning_rate": 1.837616299740702e-05, "loss": 0.5426, "step": 5851 }, { "epoch": 0.20848994424354703, "grad_norm": 1.3911287877508105, "learning_rate": 1.8375532593294282e-05, "loss": 0.5303, "step": 5852 }, { "epoch": 0.20852557137004116, "grad_norm": 1.0018676618410052, "learning_rate": 1.8374902077655474e-05, "loss": 0.455, "step": 5853 }, { "epoch": 0.20856119849653526, "grad_norm": 2.5332196613214846, "learning_rate": 1.837427145049899e-05, "loss": 0.6614, "step": 5854 }, { "epoch": 0.20859682562302936, "grad_norm": 1.77545032112898, "learning_rate": 1.837364071183323e-05, "loss": 0.5667, "step": 5855 }, { "epoch": 0.2086324527495235, "grad_norm": 1.3167665878721218, "learning_rate": 1.837300986166659e-05, "loss": 0.5332, "step": 5856 }, { "epoch": 0.2086680798760176, "grad_norm": 1.0951069352214664, "learning_rate": 1.8372378900007476e-05, "loss": 0.5073, "step": 5857 }, { "epoch": 0.2087037070025117, "grad_norm": 2.1946418300163324, "learning_rate": 1.837174782686428e-05, "loss": 0.6529, "step": 5858 }, { "epoch": 0.20873933412900583, "grad_norm": 1.498255463670146, "learning_rate": 1.837111664224541e-05, "loss": 0.6932, "step": 5859 }, { "epoch": 0.20877496125549994, "grad_norm": 1.6908885994016398, "learning_rate": 1.837048534615927e-05, "loss": 0.5972, "step": 5860 }, { "epoch": 0.20881058838199404, "grad_norm": 1.8228541884366531, "learning_rate": 1.836985393861427e-05, "loss": 0.5871, "step": 5861 }, { "epoch": 0.20884621550848817, "grad_norm": 1.9117046964305333, "learning_rate": 1.836922241961881e-05, "loss": 0.5471, "step": 5862 }, { "epoch": 0.20888184263498227, "grad_norm": 1.266428627415598, "learning_rate": 1.836859078918131e-05, "loss": 0.4951, "step": 5863 }, { "epoch": 0.20891746976147638, "grad_norm": 1.8276709890129164, "learning_rate": 1.8367959047310174e-05, "loss": 0.584, "step": 5864 }, { "epoch": 0.2089530968879705, "grad_norm": 2.0450218118435193, "learning_rate": 1.8367327194013808e-05, "loss": 0.5483, "step": 5865 }, { "epoch": 0.2089887240144646, "grad_norm": 1.4551795653825652, "learning_rate": 1.836669522930064e-05, "loss": 0.6127, "step": 5866 }, { "epoch": 0.20902435114095871, "grad_norm": 1.4903341091087186, "learning_rate": 1.8366063153179074e-05, "loss": 0.6121, "step": 5867 }, { "epoch": 0.20905997826745285, "grad_norm": 1.2474115971794852, "learning_rate": 1.8365430965657527e-05, "loss": 0.5053, "step": 5868 }, { "epoch": 0.20909560539394695, "grad_norm": 2.1594493347040267, "learning_rate": 1.836479866674442e-05, "loss": 0.5786, "step": 5869 }, { "epoch": 0.20913123252044105, "grad_norm": 2.2189793118119474, "learning_rate": 1.8364166256448173e-05, "loss": 0.5936, "step": 5870 }, { "epoch": 0.20916685964693518, "grad_norm": 1.1631972074135628, "learning_rate": 1.8363533734777207e-05, "loss": 0.4918, "step": 5871 }, { "epoch": 0.20920248677342929, "grad_norm": 1.1646367545060323, "learning_rate": 1.8362901101739942e-05, "loss": 0.5169, "step": 5872 }, { "epoch": 0.2092381138999234, "grad_norm": 1.655460273225535, "learning_rate": 1.8362268357344804e-05, "loss": 0.623, "step": 5873 }, { "epoch": 0.20927374102641752, "grad_norm": 1.4112536019437285, "learning_rate": 1.8361635501600222e-05, "loss": 0.5797, "step": 5874 }, { "epoch": 0.20930936815291162, "grad_norm": 1.9517063375203698, "learning_rate": 1.8361002534514613e-05, "loss": 0.6384, "step": 5875 }, { "epoch": 0.20934499527940573, "grad_norm": 1.1330652819533753, "learning_rate": 1.8360369456096413e-05, "loss": 0.4736, "step": 5876 }, { "epoch": 0.20938062240589986, "grad_norm": 1.069902655733983, "learning_rate": 1.8359736266354052e-05, "loss": 0.5341, "step": 5877 }, { "epoch": 0.20941624953239396, "grad_norm": 1.4971182388907156, "learning_rate": 1.8359102965295956e-05, "loss": 0.6086, "step": 5878 }, { "epoch": 0.20945187665888806, "grad_norm": 0.9790997583059591, "learning_rate": 1.8358469552930564e-05, "loss": 0.4766, "step": 5879 }, { "epoch": 0.2094875037853822, "grad_norm": 1.1970732458953057, "learning_rate": 1.8357836029266307e-05, "loss": 0.4996, "step": 5880 }, { "epoch": 0.2095231309118763, "grad_norm": 1.632789412304168, "learning_rate": 1.8357202394311624e-05, "loss": 0.6327, "step": 5881 }, { "epoch": 0.2095587580383704, "grad_norm": 1.8912964721419137, "learning_rate": 1.8356568648074946e-05, "loss": 0.6784, "step": 5882 }, { "epoch": 0.20959438516486453, "grad_norm": 1.5216547477158815, "learning_rate": 1.8355934790564718e-05, "loss": 0.561, "step": 5883 }, { "epoch": 0.20963001229135864, "grad_norm": 1.1800302917453638, "learning_rate": 1.8355300821789373e-05, "loss": 0.5688, "step": 5884 }, { "epoch": 0.20966563941785274, "grad_norm": 1.1047679411284406, "learning_rate": 1.8354666741757366e-05, "loss": 0.5076, "step": 5885 }, { "epoch": 0.20970126654434687, "grad_norm": 1.189901686240485, "learning_rate": 1.8354032550477126e-05, "loss": 0.5353, "step": 5886 }, { "epoch": 0.20973689367084097, "grad_norm": 1.9654284178831518, "learning_rate": 1.8353398247957105e-05, "loss": 0.5869, "step": 5887 }, { "epoch": 0.2097725207973351, "grad_norm": 1.2280086780631747, "learning_rate": 1.8352763834205748e-05, "loss": 0.5781, "step": 5888 }, { "epoch": 0.2098081479238292, "grad_norm": 1.178323138712247, "learning_rate": 1.83521293092315e-05, "loss": 0.5335, "step": 5889 }, { "epoch": 0.2098437750503233, "grad_norm": 1.0504612719323911, "learning_rate": 1.8351494673042814e-05, "loss": 0.477, "step": 5890 }, { "epoch": 0.20987940217681744, "grad_norm": 1.1273874118089147, "learning_rate": 1.835085992564814e-05, "loss": 0.5162, "step": 5891 }, { "epoch": 0.20991502930331155, "grad_norm": 1.157945635478641, "learning_rate": 1.8350225067055927e-05, "loss": 0.5365, "step": 5892 }, { "epoch": 0.20995065642980565, "grad_norm": 1.1611112029477666, "learning_rate": 1.8349590097274633e-05, "loss": 0.4736, "step": 5893 }, { "epoch": 0.20998628355629978, "grad_norm": 2.931765503844823, "learning_rate": 1.834895501631271e-05, "loss": 0.6368, "step": 5894 }, { "epoch": 0.21002191068279388, "grad_norm": 1.6520390808581167, "learning_rate": 1.8348319824178616e-05, "loss": 0.6699, "step": 5895 }, { "epoch": 0.210057537809288, "grad_norm": 1.1178744934203177, "learning_rate": 1.8347684520880807e-05, "loss": 0.5058, "step": 5896 }, { "epoch": 0.21009316493578212, "grad_norm": 1.1749196856442368, "learning_rate": 1.8347049106427745e-05, "loss": 0.5064, "step": 5897 }, { "epoch": 0.21012879206227622, "grad_norm": 1.7494801086380072, "learning_rate": 1.834641358082789e-05, "loss": 0.5374, "step": 5898 }, { "epoch": 0.21016441918877032, "grad_norm": 1.9241722305817839, "learning_rate": 1.8345777944089703e-05, "loss": 0.6899, "step": 5899 }, { "epoch": 0.21020004631526445, "grad_norm": 1.9478743507779948, "learning_rate": 1.834514219622165e-05, "loss": 0.6665, "step": 5900 }, { "epoch": 0.21023567344175856, "grad_norm": 2.318072924004887, "learning_rate": 1.83445063372322e-05, "loss": 0.691, "step": 5901 }, { "epoch": 0.21027130056825266, "grad_norm": 1.178804585398632, "learning_rate": 1.834387036712981e-05, "loss": 0.5288, "step": 5902 }, { "epoch": 0.2103069276947468, "grad_norm": 1.7387072395555119, "learning_rate": 1.8343234285922955e-05, "loss": 0.5996, "step": 5903 }, { "epoch": 0.2103425548212409, "grad_norm": 1.0885263727474963, "learning_rate": 1.8342598093620103e-05, "loss": 0.4807, "step": 5904 }, { "epoch": 0.210378181947735, "grad_norm": 1.273568563845266, "learning_rate": 1.834196179022973e-05, "loss": 0.5619, "step": 5905 }, { "epoch": 0.21041380907422913, "grad_norm": 1.1265017623116542, "learning_rate": 1.83413253757603e-05, "loss": 0.4755, "step": 5906 }, { "epoch": 0.21044943620072323, "grad_norm": 1.9914381273456847, "learning_rate": 1.8340688850220298e-05, "loss": 0.6534, "step": 5907 }, { "epoch": 0.21048506332721734, "grad_norm": 1.1120920175602882, "learning_rate": 1.8340052213618192e-05, "loss": 0.4877, "step": 5908 }, { "epoch": 0.21052069045371147, "grad_norm": 1.848769727520478, "learning_rate": 1.833941546596246e-05, "loss": 0.6899, "step": 5909 }, { "epoch": 0.21055631758020557, "grad_norm": 1.6230075549704948, "learning_rate": 1.8338778607261584e-05, "loss": 0.606, "step": 5910 }, { "epoch": 0.21059194470669967, "grad_norm": 1.7627539114623565, "learning_rate": 1.8338141637524042e-05, "loss": 0.6307, "step": 5911 }, { "epoch": 0.2106275718331938, "grad_norm": 1.86677403848144, "learning_rate": 1.8337504556758313e-05, "loss": 0.6242, "step": 5912 }, { "epoch": 0.2106631989596879, "grad_norm": 1.404074098832729, "learning_rate": 1.8336867364972887e-05, "loss": 0.5493, "step": 5913 }, { "epoch": 0.210698826086182, "grad_norm": 1.297037172549234, "learning_rate": 1.8336230062176245e-05, "loss": 0.5428, "step": 5914 }, { "epoch": 0.21073445321267614, "grad_norm": 1.1910713177769292, "learning_rate": 1.833559264837687e-05, "loss": 0.5252, "step": 5915 }, { "epoch": 0.21077008033917025, "grad_norm": 1.2138033253659277, "learning_rate": 1.8334955123583252e-05, "loss": 0.5109, "step": 5916 }, { "epoch": 0.21080570746566435, "grad_norm": 1.145722286664596, "learning_rate": 1.8334317487803884e-05, "loss": 0.5226, "step": 5917 }, { "epoch": 0.21084133459215848, "grad_norm": 3.2579628335170003, "learning_rate": 1.8333679741047255e-05, "loss": 0.5915, "step": 5918 }, { "epoch": 0.21087696171865258, "grad_norm": 1.3971802496312604, "learning_rate": 1.8333041883321852e-05, "loss": 0.5056, "step": 5919 }, { "epoch": 0.2109125888451467, "grad_norm": 2.246496996283527, "learning_rate": 1.833240391463617e-05, "loss": 0.5767, "step": 5920 }, { "epoch": 0.21094821597164082, "grad_norm": 2.1519254142884088, "learning_rate": 1.833176583499871e-05, "loss": 0.6461, "step": 5921 }, { "epoch": 0.21098384309813492, "grad_norm": 1.5839920696519538, "learning_rate": 1.8331127644417965e-05, "loss": 0.5969, "step": 5922 }, { "epoch": 0.21101947022462902, "grad_norm": 1.7655875806156833, "learning_rate": 1.833048934290243e-05, "loss": 0.7156, "step": 5923 }, { "epoch": 0.21105509735112316, "grad_norm": 2.357405328329115, "learning_rate": 1.8329850930460606e-05, "loss": 0.6059, "step": 5924 }, { "epoch": 0.21109072447761726, "grad_norm": 1.2843367808933863, "learning_rate": 1.8329212407100996e-05, "loss": 0.5388, "step": 5925 }, { "epoch": 0.21112635160411136, "grad_norm": 2.4276875255402364, "learning_rate": 1.83285737728321e-05, "loss": 0.6088, "step": 5926 }, { "epoch": 0.2111619787306055, "grad_norm": 1.1348666551178022, "learning_rate": 1.8327935027662426e-05, "loss": 0.5262, "step": 5927 }, { "epoch": 0.2111976058570996, "grad_norm": 1.040087203735987, "learning_rate": 1.832729617160047e-05, "loss": 0.5342, "step": 5928 }, { "epoch": 0.2112332329835937, "grad_norm": 1.864214269468874, "learning_rate": 1.8326657204654752e-05, "loss": 0.6095, "step": 5929 }, { "epoch": 0.21126886011008783, "grad_norm": 1.900302710347012, "learning_rate": 1.832601812683377e-05, "loss": 0.574, "step": 5930 }, { "epoch": 0.21130448723658193, "grad_norm": 1.182346155739218, "learning_rate": 1.8325378938146037e-05, "loss": 0.5082, "step": 5931 }, { "epoch": 0.21134011436307604, "grad_norm": 1.5458645624216432, "learning_rate": 1.8324739638600064e-05, "loss": 0.606, "step": 5932 }, { "epoch": 0.21137574148957017, "grad_norm": 1.2480660841616535, "learning_rate": 1.8324100228204367e-05, "loss": 0.5048, "step": 5933 }, { "epoch": 0.21141136861606427, "grad_norm": 1.9020953831738956, "learning_rate": 1.8323460706967453e-05, "loss": 0.6639, "step": 5934 }, { "epoch": 0.21144699574255837, "grad_norm": 1.061682806069973, "learning_rate": 1.8322821074897846e-05, "loss": 0.496, "step": 5935 }, { "epoch": 0.2114826228690525, "grad_norm": 1.769522533128799, "learning_rate": 1.8322181332004057e-05, "loss": 0.6078, "step": 5936 }, { "epoch": 0.2115182499955466, "grad_norm": 1.7385202517525171, "learning_rate": 1.8321541478294605e-05, "loss": 0.6306, "step": 5937 }, { "epoch": 0.2115538771220407, "grad_norm": 1.2127431550413041, "learning_rate": 1.8320901513778014e-05, "loss": 0.5095, "step": 5938 }, { "epoch": 0.21158950424853484, "grad_norm": 1.8448787581446286, "learning_rate": 1.8320261438462802e-05, "loss": 0.6339, "step": 5939 }, { "epoch": 0.21162513137502895, "grad_norm": 1.1574259870512722, "learning_rate": 1.8319621252357492e-05, "loss": 0.5083, "step": 5940 }, { "epoch": 0.21166075850152305, "grad_norm": 1.7504290336223882, "learning_rate": 1.8318980955470614e-05, "loss": 0.5192, "step": 5941 }, { "epoch": 0.21169638562801718, "grad_norm": 1.3335631343187784, "learning_rate": 1.8318340547810688e-05, "loss": 0.4864, "step": 5942 }, { "epoch": 0.21173201275451128, "grad_norm": 1.1168469333547195, "learning_rate": 1.8317700029386245e-05, "loss": 0.5178, "step": 5943 }, { "epoch": 0.2117676398810054, "grad_norm": 1.8463104726031203, "learning_rate": 1.831705940020581e-05, "loss": 0.5803, "step": 5944 }, { "epoch": 0.21180326700749952, "grad_norm": 2.9312582192758, "learning_rate": 1.8316418660277916e-05, "loss": 0.6325, "step": 5945 }, { "epoch": 0.21183889413399362, "grad_norm": 1.0903849119080655, "learning_rate": 1.8315777809611098e-05, "loss": 0.4813, "step": 5946 }, { "epoch": 0.21187452126048772, "grad_norm": 1.200688028309669, "learning_rate": 1.8315136848213883e-05, "loss": 0.5141, "step": 5947 }, { "epoch": 0.21191014838698186, "grad_norm": 1.1808307662444388, "learning_rate": 1.831449577609481e-05, "loss": 0.5814, "step": 5948 }, { "epoch": 0.21194577551347596, "grad_norm": 1.9535071625593814, "learning_rate": 1.8313854593262413e-05, "loss": 0.5901, "step": 5949 }, { "epoch": 0.21198140263997006, "grad_norm": 1.8081301283709021, "learning_rate": 1.8313213299725232e-05, "loss": 0.5951, "step": 5950 }, { "epoch": 0.2120170297664642, "grad_norm": 1.8218823963271584, "learning_rate": 1.8312571895491803e-05, "loss": 0.7016, "step": 5951 }, { "epoch": 0.2120526568929583, "grad_norm": 1.3790453495006896, "learning_rate": 1.8311930380570672e-05, "loss": 0.541, "step": 5952 }, { "epoch": 0.2120882840194524, "grad_norm": 1.1793032240382113, "learning_rate": 1.831128875497038e-05, "loss": 0.4773, "step": 5953 }, { "epoch": 0.21212391114594653, "grad_norm": 1.5791317861146552, "learning_rate": 1.8310647018699464e-05, "loss": 0.6063, "step": 5954 }, { "epoch": 0.21215953827244063, "grad_norm": 1.1739435353799115, "learning_rate": 1.8310005171766477e-05, "loss": 0.5532, "step": 5955 }, { "epoch": 0.21219516539893474, "grad_norm": 1.2387405498457484, "learning_rate": 1.8309363214179965e-05, "loss": 0.5101, "step": 5956 }, { "epoch": 0.21223079252542887, "grad_norm": 1.156062150063324, "learning_rate": 1.830872114594847e-05, "loss": 0.49, "step": 5957 }, { "epoch": 0.21226641965192297, "grad_norm": 1.2223587367333448, "learning_rate": 1.8308078967080547e-05, "loss": 0.5259, "step": 5958 }, { "epoch": 0.21230204677841708, "grad_norm": 1.5203536248021297, "learning_rate": 1.8307436677584743e-05, "loss": 0.5322, "step": 5959 }, { "epoch": 0.2123376739049112, "grad_norm": 1.9310016082770893, "learning_rate": 1.830679427746962e-05, "loss": 0.6518, "step": 5960 }, { "epoch": 0.2123733010314053, "grad_norm": 1.2293816313228758, "learning_rate": 1.830615176674372e-05, "loss": 0.5119, "step": 5961 }, { "epoch": 0.2124089281578994, "grad_norm": 1.7640551482374796, "learning_rate": 1.8305509145415602e-05, "loss": 0.5367, "step": 5962 }, { "epoch": 0.21244455528439354, "grad_norm": 1.2375667184322459, "learning_rate": 1.8304866413493826e-05, "loss": 0.553, "step": 5963 }, { "epoch": 0.21248018241088765, "grad_norm": 1.2336649498197687, "learning_rate": 1.8304223570986952e-05, "loss": 0.5021, "step": 5964 }, { "epoch": 0.21251580953738175, "grad_norm": 1.1445992058328422, "learning_rate": 1.8303580617903532e-05, "loss": 0.5386, "step": 5965 }, { "epoch": 0.21255143666387588, "grad_norm": 1.247149917305136, "learning_rate": 1.8302937554252138e-05, "loss": 0.5302, "step": 5966 }, { "epoch": 0.21258706379036998, "grad_norm": 1.05086791731699, "learning_rate": 1.8302294380041325e-05, "loss": 0.4971, "step": 5967 }, { "epoch": 0.2126226909168641, "grad_norm": 1.1726238603417836, "learning_rate": 1.8301651095279657e-05, "loss": 0.5382, "step": 5968 }, { "epoch": 0.21265831804335822, "grad_norm": 1.1892932809094254, "learning_rate": 1.8301007699975704e-05, "loss": 0.4712, "step": 5969 }, { "epoch": 0.21269394516985232, "grad_norm": 1.6945915197537402, "learning_rate": 1.830036419413803e-05, "loss": 0.6901, "step": 5970 }, { "epoch": 0.21272957229634643, "grad_norm": 1.610250796393166, "learning_rate": 1.8299720577775205e-05, "loss": 0.6254, "step": 5971 }, { "epoch": 0.21276519942284056, "grad_norm": 1.1783138582602248, "learning_rate": 1.82990768508958e-05, "loss": 0.4976, "step": 5972 }, { "epoch": 0.21280082654933466, "grad_norm": 1.226204054214162, "learning_rate": 1.8298433013508384e-05, "loss": 0.517, "step": 5973 }, { "epoch": 0.21283645367582876, "grad_norm": 1.3052328732421292, "learning_rate": 1.8297789065621536e-05, "loss": 0.4813, "step": 5974 }, { "epoch": 0.2128720808023229, "grad_norm": 1.5175133047804072, "learning_rate": 1.8297145007243827e-05, "loss": 0.5406, "step": 5975 }, { "epoch": 0.212907707928817, "grad_norm": 1.5060045274259828, "learning_rate": 1.829650083838383e-05, "loss": 0.6001, "step": 5976 }, { "epoch": 0.2129433350553111, "grad_norm": 1.5565905330026026, "learning_rate": 1.8295856559050126e-05, "loss": 0.6237, "step": 5977 }, { "epoch": 0.21297896218180523, "grad_norm": 1.5589696761877834, "learning_rate": 1.8295212169251295e-05, "loss": 0.6957, "step": 5978 }, { "epoch": 0.21301458930829933, "grad_norm": 1.0557636849519854, "learning_rate": 1.8294567668995914e-05, "loss": 0.5368, "step": 5979 }, { "epoch": 0.21305021643479344, "grad_norm": 1.1881978009017606, "learning_rate": 1.829392305829257e-05, "loss": 0.5804, "step": 5980 }, { "epoch": 0.21308584356128757, "grad_norm": 1.0647338987130972, "learning_rate": 1.829327833714984e-05, "loss": 0.4623, "step": 5981 }, { "epoch": 0.21312147068778167, "grad_norm": 1.1453989119769592, "learning_rate": 1.8292633505576313e-05, "loss": 0.5136, "step": 5982 }, { "epoch": 0.2131570978142758, "grad_norm": 1.0966202340895943, "learning_rate": 1.8291988563580576e-05, "loss": 0.533, "step": 5983 }, { "epoch": 0.2131927249407699, "grad_norm": 1.214897628103525, "learning_rate": 1.8291343511171217e-05, "loss": 0.5396, "step": 5984 }, { "epoch": 0.213228352067264, "grad_norm": 1.1230869387923406, "learning_rate": 1.829069834835682e-05, "loss": 0.4712, "step": 5985 }, { "epoch": 0.21326397919375814, "grad_norm": 1.1794099638425246, "learning_rate": 1.829005307514598e-05, "loss": 0.4983, "step": 5986 }, { "epoch": 0.21329960632025224, "grad_norm": 1.8975318308289995, "learning_rate": 1.828940769154729e-05, "loss": 0.5501, "step": 5987 }, { "epoch": 0.21333523344674635, "grad_norm": 1.4824991677442672, "learning_rate": 1.8288762197569343e-05, "loss": 0.6183, "step": 5988 }, { "epoch": 0.21337086057324048, "grad_norm": 1.9147922964881794, "learning_rate": 1.8288116593220737e-05, "loss": 0.5417, "step": 5989 }, { "epoch": 0.21340648769973458, "grad_norm": 1.2084436651765127, "learning_rate": 1.8287470878510062e-05, "loss": 0.4922, "step": 5990 }, { "epoch": 0.21344211482622868, "grad_norm": 1.556473707658132, "learning_rate": 1.8286825053445916e-05, "loss": 0.6901, "step": 5991 }, { "epoch": 0.21347774195272282, "grad_norm": 1.6810694620717672, "learning_rate": 1.8286179118036908e-05, "loss": 0.6881, "step": 5992 }, { "epoch": 0.21351336907921692, "grad_norm": 2.6909323636831757, "learning_rate": 1.828553307229163e-05, "loss": 0.6302, "step": 5993 }, { "epoch": 0.21354899620571102, "grad_norm": 1.1543917911062982, "learning_rate": 1.828488691621869e-05, "loss": 0.473, "step": 5994 }, { "epoch": 0.21358462333220515, "grad_norm": 1.7189297383700497, "learning_rate": 1.828424064982669e-05, "loss": 0.5744, "step": 5995 }, { "epoch": 0.21362025045869926, "grad_norm": 1.2215463090750636, "learning_rate": 1.828359427312423e-05, "loss": 0.5355, "step": 5996 }, { "epoch": 0.21365587758519336, "grad_norm": 1.0790801224733475, "learning_rate": 1.828294778611993e-05, "loss": 0.5116, "step": 5997 }, { "epoch": 0.2136915047116875, "grad_norm": 1.2390123106183664, "learning_rate": 1.828230118882239e-05, "loss": 0.5239, "step": 5998 }, { "epoch": 0.2137271318381816, "grad_norm": 1.4243480148512846, "learning_rate": 1.828165448124022e-05, "loss": 0.633, "step": 5999 }, { "epoch": 0.2137627589646757, "grad_norm": 1.18150923419814, "learning_rate": 1.8281007663382027e-05, "loss": 0.5451, "step": 6000 }, { "epoch": 0.21379838609116983, "grad_norm": 1.0906268152398413, "learning_rate": 1.8280360735256432e-05, "loss": 0.5619, "step": 6001 }, { "epoch": 0.21383401321766393, "grad_norm": 1.9326193917629906, "learning_rate": 1.8279713696872047e-05, "loss": 0.6347, "step": 6002 }, { "epoch": 0.21386964034415804, "grad_norm": 1.2251243274365422, "learning_rate": 1.8279066548237487e-05, "loss": 0.5101, "step": 6003 }, { "epoch": 0.21390526747065217, "grad_norm": 1.2094097697378619, "learning_rate": 1.827841928936137e-05, "loss": 0.5479, "step": 6004 }, { "epoch": 0.21394089459714627, "grad_norm": 1.1948006849794623, "learning_rate": 1.8277771920252312e-05, "loss": 0.5354, "step": 6005 }, { "epoch": 0.21397652172364037, "grad_norm": 1.5121769808201966, "learning_rate": 1.8277124440918935e-05, "loss": 0.5122, "step": 6006 }, { "epoch": 0.2140121488501345, "grad_norm": 1.196601976088452, "learning_rate": 1.8276476851369862e-05, "loss": 0.5252, "step": 6007 }, { "epoch": 0.2140477759766286, "grad_norm": 1.7432072084896248, "learning_rate": 1.8275829151613712e-05, "loss": 0.636, "step": 6008 }, { "epoch": 0.2140834031031227, "grad_norm": 1.379641620305511, "learning_rate": 1.8275181341659117e-05, "loss": 0.5345, "step": 6009 }, { "epoch": 0.21411903022961684, "grad_norm": 2.5091976705930175, "learning_rate": 1.8274533421514697e-05, "loss": 0.6231, "step": 6010 }, { "epoch": 0.21415465735611094, "grad_norm": 1.9592643759440693, "learning_rate": 1.827388539118908e-05, "loss": 0.6412, "step": 6011 }, { "epoch": 0.21419028448260505, "grad_norm": 1.9821290923094204, "learning_rate": 1.8273237250690894e-05, "loss": 0.6903, "step": 6012 }, { "epoch": 0.21422591160909918, "grad_norm": 1.4091556665516112, "learning_rate": 1.8272589000028774e-05, "loss": 0.4647, "step": 6013 }, { "epoch": 0.21426153873559328, "grad_norm": 1.334279782011494, "learning_rate": 1.8271940639211347e-05, "loss": 0.4717, "step": 6014 }, { "epoch": 0.21429716586208739, "grad_norm": 1.3625949153193608, "learning_rate": 1.8271292168247253e-05, "loss": 0.5228, "step": 6015 }, { "epoch": 0.21433279298858152, "grad_norm": 1.9526867072609777, "learning_rate": 1.8270643587145118e-05, "loss": 0.5649, "step": 6016 }, { "epoch": 0.21436842011507562, "grad_norm": 2.0255198184580996, "learning_rate": 1.8269994895913585e-05, "loss": 0.6178, "step": 6017 }, { "epoch": 0.21440404724156972, "grad_norm": 1.6518588014655853, "learning_rate": 1.826934609456129e-05, "loss": 0.6108, "step": 6018 }, { "epoch": 0.21443967436806385, "grad_norm": 1.5552807052960715, "learning_rate": 1.826869718309687e-05, "loss": 0.527, "step": 6019 }, { "epoch": 0.21447530149455796, "grad_norm": 1.9896632191872083, "learning_rate": 1.8268048161528966e-05, "loss": 0.5324, "step": 6020 }, { "epoch": 0.21451092862105206, "grad_norm": 3.809055631881374, "learning_rate": 1.8267399029866227e-05, "loss": 0.6225, "step": 6021 }, { "epoch": 0.2145465557475462, "grad_norm": 1.5882152941930217, "learning_rate": 1.826674978811729e-05, "loss": 0.5506, "step": 6022 }, { "epoch": 0.2145821828740403, "grad_norm": 2.856556326996892, "learning_rate": 1.8266100436290796e-05, "loss": 0.4763, "step": 6023 }, { "epoch": 0.2146178100005344, "grad_norm": 3.2305300990630115, "learning_rate": 1.8265450974395403e-05, "loss": 0.6283, "step": 6024 }, { "epoch": 0.21465343712702853, "grad_norm": 2.036587574335706, "learning_rate": 1.826480140243975e-05, "loss": 0.4951, "step": 6025 }, { "epoch": 0.21468906425352263, "grad_norm": 1.6758962749681647, "learning_rate": 1.826415172043249e-05, "loss": 0.633, "step": 6026 }, { "epoch": 0.21472469138001674, "grad_norm": 1.9210528054382838, "learning_rate": 1.8263501928382274e-05, "loss": 0.6265, "step": 6027 }, { "epoch": 0.21476031850651087, "grad_norm": 2.014428184502331, "learning_rate": 1.8262852026297754e-05, "loss": 0.516, "step": 6028 }, { "epoch": 0.21479594563300497, "grad_norm": 1.8191882220036941, "learning_rate": 1.8262202014187583e-05, "loss": 0.5176, "step": 6029 }, { "epoch": 0.21483157275949907, "grad_norm": 1.636253405754218, "learning_rate": 1.826155189206042e-05, "loss": 0.6108, "step": 6030 }, { "epoch": 0.2148671998859932, "grad_norm": 2.3136486540001244, "learning_rate": 1.8260901659924917e-05, "loss": 0.6306, "step": 6031 }, { "epoch": 0.2149028270124873, "grad_norm": 1.938585395804237, "learning_rate": 1.8260251317789738e-05, "loss": 0.5353, "step": 6032 }, { "epoch": 0.2149384541389814, "grad_norm": 1.748002705661573, "learning_rate": 1.8259600865663532e-05, "loss": 0.5814, "step": 6033 }, { "epoch": 0.21497408126547554, "grad_norm": 1.6561650349472576, "learning_rate": 1.825895030355497e-05, "loss": 0.6569, "step": 6034 }, { "epoch": 0.21500970839196965, "grad_norm": 1.6525796131338333, "learning_rate": 1.8258299631472717e-05, "loss": 0.5013, "step": 6035 }, { "epoch": 0.21504533551846375, "grad_norm": 2.45117642197216, "learning_rate": 1.8257648849425427e-05, "loss": 0.5388, "step": 6036 }, { "epoch": 0.21508096264495788, "grad_norm": 1.1661039956074173, "learning_rate": 1.8256997957421774e-05, "loss": 0.4925, "step": 6037 }, { "epoch": 0.21511658977145198, "grad_norm": 1.3649474298057063, "learning_rate": 1.825634695547042e-05, "loss": 0.5356, "step": 6038 }, { "epoch": 0.21515221689794609, "grad_norm": 1.35234907550291, "learning_rate": 1.8255695843580036e-05, "loss": 0.4857, "step": 6039 }, { "epoch": 0.21518784402444022, "grad_norm": 1.4442828602349913, "learning_rate": 1.8255044621759288e-05, "loss": 0.4886, "step": 6040 }, { "epoch": 0.21522347115093432, "grad_norm": 1.7375247055769598, "learning_rate": 1.8254393290016854e-05, "loss": 0.6298, "step": 6041 }, { "epoch": 0.21525909827742842, "grad_norm": 1.2488576303336119, "learning_rate": 1.8253741848361403e-05, "loss": 0.5315, "step": 6042 }, { "epoch": 0.21529472540392255, "grad_norm": 1.5316665687021205, "learning_rate": 1.8253090296801614e-05, "loss": 0.5702, "step": 6043 }, { "epoch": 0.21533035253041666, "grad_norm": 1.9116645936041132, "learning_rate": 1.8252438635346155e-05, "loss": 0.7352, "step": 6044 }, { "epoch": 0.21536597965691076, "grad_norm": 1.3270488162312408, "learning_rate": 1.8251786864003712e-05, "loss": 0.5633, "step": 6045 }, { "epoch": 0.2154016067834049, "grad_norm": 1.63386358565864, "learning_rate": 1.8251134982782952e-05, "loss": 0.5925, "step": 6046 }, { "epoch": 0.215437233909899, "grad_norm": 2.4367662300964077, "learning_rate": 1.825048299169257e-05, "loss": 0.6344, "step": 6047 }, { "epoch": 0.2154728610363931, "grad_norm": 1.2434981932101687, "learning_rate": 1.8249830890741234e-05, "loss": 0.5186, "step": 6048 }, { "epoch": 0.21550848816288723, "grad_norm": 1.2130095569357826, "learning_rate": 1.8249178679937637e-05, "loss": 0.5722, "step": 6049 }, { "epoch": 0.21554411528938133, "grad_norm": 1.6984010144329116, "learning_rate": 1.824852635929046e-05, "loss": 0.503, "step": 6050 }, { "epoch": 0.21557974241587544, "grad_norm": 1.2407932268621453, "learning_rate": 1.824787392880839e-05, "loss": 0.5655, "step": 6051 }, { "epoch": 0.21561536954236957, "grad_norm": 1.4391383668168156, "learning_rate": 1.8247221388500106e-05, "loss": 0.5069, "step": 6052 }, { "epoch": 0.21565099666886367, "grad_norm": 1.2992655491823055, "learning_rate": 1.8246568738374316e-05, "loss": 0.5124, "step": 6053 }, { "epoch": 0.21568662379535777, "grad_norm": 1.263223211958871, "learning_rate": 1.824591597843969e-05, "loss": 0.5145, "step": 6054 }, { "epoch": 0.2157222509218519, "grad_norm": 2.459264207952915, "learning_rate": 1.8245263108704937e-05, "loss": 0.6438, "step": 6055 }, { "epoch": 0.215757878048346, "grad_norm": 2.251452933010736, "learning_rate": 1.8244610129178737e-05, "loss": 0.6426, "step": 6056 }, { "epoch": 0.2157935051748401, "grad_norm": 1.4160562759004218, "learning_rate": 1.824395703986979e-05, "loss": 0.5784, "step": 6057 }, { "epoch": 0.21582913230133424, "grad_norm": 1.6832128943133686, "learning_rate": 1.8243303840786798e-05, "loss": 0.6469, "step": 6058 }, { "epoch": 0.21586475942782835, "grad_norm": 1.0816289089273172, "learning_rate": 1.824265053193845e-05, "loss": 0.4955, "step": 6059 }, { "epoch": 0.21590038655432245, "grad_norm": 1.2501640294616343, "learning_rate": 1.824199711333345e-05, "loss": 0.5309, "step": 6060 }, { "epoch": 0.21593601368081658, "grad_norm": 2.145357906221119, "learning_rate": 1.8241343584980495e-05, "loss": 0.5498, "step": 6061 }, { "epoch": 0.21597164080731068, "grad_norm": 1.8909876474224365, "learning_rate": 1.8240689946888296e-05, "loss": 0.66, "step": 6062 }, { "epoch": 0.2160072679338048, "grad_norm": 1.4603439917764323, "learning_rate": 1.8240036199065546e-05, "loss": 0.5645, "step": 6063 }, { "epoch": 0.21604289506029892, "grad_norm": 1.1241376198976418, "learning_rate": 1.823938234152095e-05, "loss": 0.5492, "step": 6064 }, { "epoch": 0.21607852218679302, "grad_norm": 1.1389866729899631, "learning_rate": 1.8238728374263228e-05, "loss": 0.5614, "step": 6065 }, { "epoch": 0.21611414931328712, "grad_norm": 1.3049942770944862, "learning_rate": 1.8238074297301076e-05, "loss": 0.5176, "step": 6066 }, { "epoch": 0.21614977643978125, "grad_norm": 1.2718987456827564, "learning_rate": 1.8237420110643205e-05, "loss": 0.4993, "step": 6067 }, { "epoch": 0.21618540356627536, "grad_norm": 1.7364198074769412, "learning_rate": 1.8236765814298328e-05, "loss": 0.6167, "step": 6068 }, { "epoch": 0.21622103069276946, "grad_norm": 1.1455171115461653, "learning_rate": 1.823611140827516e-05, "loss": 0.5336, "step": 6069 }, { "epoch": 0.2162566578192636, "grad_norm": 1.1388439989395631, "learning_rate": 1.823545689258241e-05, "loss": 0.4859, "step": 6070 }, { "epoch": 0.2162922849457577, "grad_norm": 1.2443935598192184, "learning_rate": 1.8234802267228794e-05, "loss": 0.5021, "step": 6071 }, { "epoch": 0.2163279120722518, "grad_norm": 1.0214882321126157, "learning_rate": 1.823414753222303e-05, "loss": 0.4841, "step": 6072 }, { "epoch": 0.21636353919874593, "grad_norm": 1.4598256483485383, "learning_rate": 1.823349268757384e-05, "loss": 0.5053, "step": 6073 }, { "epoch": 0.21639916632524003, "grad_norm": 1.5576880206189, "learning_rate": 1.8232837733289935e-05, "loss": 0.4672, "step": 6074 }, { "epoch": 0.21643479345173414, "grad_norm": 1.6302994164553881, "learning_rate": 1.8232182669380044e-05, "loss": 0.633, "step": 6075 }, { "epoch": 0.21647042057822827, "grad_norm": 3.6084707085523187, "learning_rate": 1.8231527495852884e-05, "loss": 0.6906, "step": 6076 }, { "epoch": 0.21650604770472237, "grad_norm": 1.115219050234889, "learning_rate": 1.8230872212717185e-05, "loss": 0.5053, "step": 6077 }, { "epoch": 0.2165416748312165, "grad_norm": 1.0760923595785754, "learning_rate": 1.8230216819981667e-05, "loss": 0.4745, "step": 6078 }, { "epoch": 0.2165773019577106, "grad_norm": 2.045361938772524, "learning_rate": 1.822956131765506e-05, "loss": 0.5794, "step": 6079 }, { "epoch": 0.2166129290842047, "grad_norm": 1.2211152932468468, "learning_rate": 1.8228905705746094e-05, "loss": 0.5152, "step": 6080 }, { "epoch": 0.21664855621069884, "grad_norm": 1.8463794008782342, "learning_rate": 1.8228249984263496e-05, "loss": 0.5759, "step": 6081 }, { "epoch": 0.21668418333719294, "grad_norm": 1.179036689672868, "learning_rate": 1.8227594153216e-05, "loss": 0.5141, "step": 6082 }, { "epoch": 0.21671981046368705, "grad_norm": 2.3509408563086645, "learning_rate": 1.8226938212612335e-05, "loss": 0.6527, "step": 6083 }, { "epoch": 0.21675543759018118, "grad_norm": 1.2156326051614086, "learning_rate": 1.8226282162461237e-05, "loss": 0.4884, "step": 6084 }, { "epoch": 0.21679106471667528, "grad_norm": 1.5327020117549106, "learning_rate": 1.8225626002771443e-05, "loss": 0.5876, "step": 6085 }, { "epoch": 0.21682669184316938, "grad_norm": 1.2072127238327102, "learning_rate": 1.822496973355169e-05, "loss": 0.56, "step": 6086 }, { "epoch": 0.21686231896966351, "grad_norm": 2.217869880790822, "learning_rate": 1.8224313354810714e-05, "loss": 0.5175, "step": 6087 }, { "epoch": 0.21689794609615762, "grad_norm": 1.322670645515027, "learning_rate": 1.822365686655726e-05, "loss": 0.5215, "step": 6088 }, { "epoch": 0.21693357322265172, "grad_norm": 1.227779035345082, "learning_rate": 1.8223000268800063e-05, "loss": 0.5275, "step": 6089 }, { "epoch": 0.21696920034914585, "grad_norm": 1.1346102367931357, "learning_rate": 1.8222343561547876e-05, "loss": 0.4833, "step": 6090 }, { "epoch": 0.21700482747563996, "grad_norm": 2.367625066651954, "learning_rate": 1.8221686744809434e-05, "loss": 0.6212, "step": 6091 }, { "epoch": 0.21704045460213406, "grad_norm": 1.6845633934404063, "learning_rate": 1.8221029818593484e-05, "loss": 0.6465, "step": 6092 }, { "epoch": 0.2170760817286282, "grad_norm": 1.1190877794414231, "learning_rate": 1.8220372782908778e-05, "loss": 0.4406, "step": 6093 }, { "epoch": 0.2171117088551223, "grad_norm": 2.7369696335109843, "learning_rate": 1.821971563776406e-05, "loss": 0.5357, "step": 6094 }, { "epoch": 0.2171473359816164, "grad_norm": 1.280047704930896, "learning_rate": 1.8219058383168088e-05, "loss": 0.5512, "step": 6095 }, { "epoch": 0.21718296310811053, "grad_norm": 1.1985390939121852, "learning_rate": 1.8218401019129607e-05, "loss": 0.5333, "step": 6096 }, { "epoch": 0.21721859023460463, "grad_norm": 1.0734525744176302, "learning_rate": 1.821774354565737e-05, "loss": 0.4473, "step": 6097 }, { "epoch": 0.21725421736109873, "grad_norm": 1.8976459580942764, "learning_rate": 1.8217085962760136e-05, "loss": 0.5663, "step": 6098 }, { "epoch": 0.21728984448759286, "grad_norm": 1.411319006488104, "learning_rate": 1.8216428270446657e-05, "loss": 0.5731, "step": 6099 }, { "epoch": 0.21732547161408697, "grad_norm": 1.6149303742169756, "learning_rate": 1.8215770468725696e-05, "loss": 0.4989, "step": 6100 }, { "epoch": 0.21736109874058107, "grad_norm": 1.039741927346796, "learning_rate": 1.8215112557606005e-05, "loss": 0.4582, "step": 6101 }, { "epoch": 0.2173967258670752, "grad_norm": 1.135048287879687, "learning_rate": 1.8214454537096352e-05, "loss": 0.5248, "step": 6102 }, { "epoch": 0.2174323529935693, "grad_norm": 1.5517142803431878, "learning_rate": 1.821379640720549e-05, "loss": 0.4976, "step": 6103 }, { "epoch": 0.2174679801200634, "grad_norm": 1.6702202483321522, "learning_rate": 1.821313816794219e-05, "loss": 0.6146, "step": 6104 }, { "epoch": 0.21750360724655754, "grad_norm": 1.0557985962213132, "learning_rate": 1.821247981931522e-05, "loss": 0.4401, "step": 6105 }, { "epoch": 0.21753923437305164, "grad_norm": 1.0870222245096677, "learning_rate": 1.8211821361333333e-05, "loss": 0.5413, "step": 6106 }, { "epoch": 0.21757486149954575, "grad_norm": 1.184701345266014, "learning_rate": 1.821116279400531e-05, "loss": 0.5221, "step": 6107 }, { "epoch": 0.21761048862603988, "grad_norm": 2.0533652414889767, "learning_rate": 1.8210504117339917e-05, "loss": 0.6486, "step": 6108 }, { "epoch": 0.21764611575253398, "grad_norm": 1.9410251445701485, "learning_rate": 1.8209845331345915e-05, "loss": 0.6239, "step": 6109 }, { "epoch": 0.21768174287902808, "grad_norm": 1.6600582327626388, "learning_rate": 1.820918643603209e-05, "loss": 0.5948, "step": 6110 }, { "epoch": 0.21771737000552221, "grad_norm": 1.1919623872787124, "learning_rate": 1.8208527431407206e-05, "loss": 0.5007, "step": 6111 }, { "epoch": 0.21775299713201632, "grad_norm": 1.1260469320543265, "learning_rate": 1.8207868317480046e-05, "loss": 0.4552, "step": 6112 }, { "epoch": 0.21778862425851042, "grad_norm": 1.3445132409223362, "learning_rate": 1.820720909425938e-05, "loss": 0.4984, "step": 6113 }, { "epoch": 0.21782425138500455, "grad_norm": 1.535423320218332, "learning_rate": 1.8206549761753992e-05, "loss": 0.5822, "step": 6114 }, { "epoch": 0.21785987851149866, "grad_norm": 1.5809488457877852, "learning_rate": 1.8205890319972653e-05, "loss": 0.6016, "step": 6115 }, { "epoch": 0.21789550563799276, "grad_norm": 1.5294612577796054, "learning_rate": 1.8205230768924148e-05, "loss": 0.5773, "step": 6116 }, { "epoch": 0.2179311327644869, "grad_norm": 1.934314955974301, "learning_rate": 1.8204571108617263e-05, "loss": 0.6491, "step": 6117 }, { "epoch": 0.217966759890981, "grad_norm": 2.186120712043856, "learning_rate": 1.8203911339060784e-05, "loss": 0.5834, "step": 6118 }, { "epoch": 0.2180023870174751, "grad_norm": 1.159889554997067, "learning_rate": 1.8203251460263485e-05, "loss": 0.489, "step": 6119 }, { "epoch": 0.21803801414396923, "grad_norm": 2.474317165850678, "learning_rate": 1.820259147223416e-05, "loss": 0.5624, "step": 6120 }, { "epoch": 0.21807364127046333, "grad_norm": 1.476157015250329, "learning_rate": 1.8201931374981595e-05, "loss": 0.5184, "step": 6121 }, { "epoch": 0.21810926839695743, "grad_norm": 2.3466613979065234, "learning_rate": 1.820127116851458e-05, "loss": 0.6486, "step": 6122 }, { "epoch": 0.21814489552345157, "grad_norm": 1.588015904830306, "learning_rate": 1.8200610852841913e-05, "loss": 0.5798, "step": 6123 }, { "epoch": 0.21818052264994567, "grad_norm": 1.1737185230605391, "learning_rate": 1.8199950427972374e-05, "loss": 0.5202, "step": 6124 }, { "epoch": 0.21821614977643977, "grad_norm": 2.0456498059329933, "learning_rate": 1.8199289893914768e-05, "loss": 0.6148, "step": 6125 }, { "epoch": 0.2182517769029339, "grad_norm": 1.0970177119634148, "learning_rate": 1.8198629250677884e-05, "loss": 0.4695, "step": 6126 }, { "epoch": 0.218287404029428, "grad_norm": 1.7079770058908965, "learning_rate": 1.8197968498270523e-05, "loss": 0.5767, "step": 6127 }, { "epoch": 0.2183230311559221, "grad_norm": 1.3425464155044176, "learning_rate": 1.819730763670148e-05, "loss": 0.5122, "step": 6128 }, { "epoch": 0.21835865828241624, "grad_norm": 1.4430199202596512, "learning_rate": 1.8196646665979557e-05, "loss": 0.5312, "step": 6129 }, { "epoch": 0.21839428540891034, "grad_norm": 1.2019281084387086, "learning_rate": 1.8195985586113557e-05, "loss": 0.508, "step": 6130 }, { "epoch": 0.21842991253540445, "grad_norm": 2.580643044124776, "learning_rate": 1.819532439711228e-05, "loss": 0.5998, "step": 6131 }, { "epoch": 0.21846553966189858, "grad_norm": 2.742042290535209, "learning_rate": 1.8194663098984524e-05, "loss": 0.5888, "step": 6132 }, { "epoch": 0.21850116678839268, "grad_norm": 2.038927577023964, "learning_rate": 1.8194001691739104e-05, "loss": 0.6124, "step": 6133 }, { "epoch": 0.21853679391488678, "grad_norm": 1.519998715788992, "learning_rate": 1.8193340175384826e-05, "loss": 0.6055, "step": 6134 }, { "epoch": 0.21857242104138092, "grad_norm": 1.829071511265327, "learning_rate": 1.8192678549930496e-05, "loss": 0.629, "step": 6135 }, { "epoch": 0.21860804816787502, "grad_norm": 1.2398944027454544, "learning_rate": 1.8192016815384925e-05, "loss": 0.4939, "step": 6136 }, { "epoch": 0.21864367529436912, "grad_norm": 1.4475970774480844, "learning_rate": 1.8191354971756923e-05, "loss": 0.4995, "step": 6137 }, { "epoch": 0.21867930242086325, "grad_norm": 1.2389490949334305, "learning_rate": 1.8190693019055307e-05, "loss": 0.4872, "step": 6138 }, { "epoch": 0.21871492954735736, "grad_norm": 1.3447283369545118, "learning_rate": 1.8190030957288884e-05, "loss": 0.47, "step": 6139 }, { "epoch": 0.21875055667385146, "grad_norm": 1.5484496565943953, "learning_rate": 1.8189368786466478e-05, "loss": 0.6398, "step": 6140 }, { "epoch": 0.2187861838003456, "grad_norm": 2.4701542830110363, "learning_rate": 1.81887065065969e-05, "loss": 0.6485, "step": 6141 }, { "epoch": 0.2188218109268397, "grad_norm": 1.770093442641906, "learning_rate": 1.8188044117688973e-05, "loss": 0.6026, "step": 6142 }, { "epoch": 0.2188574380533338, "grad_norm": 1.8101905722978606, "learning_rate": 1.8187381619751516e-05, "loss": 0.6263, "step": 6143 }, { "epoch": 0.21889306517982793, "grad_norm": 1.293785190344975, "learning_rate": 1.818671901279335e-05, "loss": 0.5115, "step": 6144 }, { "epoch": 0.21892869230632203, "grad_norm": 1.4510388985580316, "learning_rate": 1.8186056296823297e-05, "loss": 0.555, "step": 6145 }, { "epoch": 0.21896431943281613, "grad_norm": 1.7200733788240288, "learning_rate": 1.8185393471850184e-05, "loss": 0.7023, "step": 6146 }, { "epoch": 0.21899994655931027, "grad_norm": 2.644330034730304, "learning_rate": 1.8184730537882833e-05, "loss": 0.6495, "step": 6147 }, { "epoch": 0.21903557368580437, "grad_norm": 1.2183182297800503, "learning_rate": 1.8184067494930076e-05, "loss": 0.4995, "step": 6148 }, { "epoch": 0.21907120081229847, "grad_norm": 1.099441317971324, "learning_rate": 1.818340434300074e-05, "loss": 0.5546, "step": 6149 }, { "epoch": 0.2191068279387926, "grad_norm": 1.6936885516899671, "learning_rate": 1.8182741082103656e-05, "loss": 0.6072, "step": 6150 }, { "epoch": 0.2191424550652867, "grad_norm": 1.1755439544712893, "learning_rate": 1.8182077712247654e-05, "loss": 0.5392, "step": 6151 }, { "epoch": 0.2191780821917808, "grad_norm": 1.145829531819458, "learning_rate": 1.818141423344157e-05, "loss": 0.5077, "step": 6152 }, { "epoch": 0.21921370931827494, "grad_norm": 1.1848496825728014, "learning_rate": 1.8180750645694236e-05, "loss": 0.4981, "step": 6153 }, { "epoch": 0.21924933644476904, "grad_norm": 1.0370600356247024, "learning_rate": 1.8180086949014488e-05, "loss": 0.5172, "step": 6154 }, { "epoch": 0.21928496357126315, "grad_norm": 1.12296531501939, "learning_rate": 1.8179423143411166e-05, "loss": 0.5433, "step": 6155 }, { "epoch": 0.21932059069775728, "grad_norm": 1.1098349503154126, "learning_rate": 1.8178759228893108e-05, "loss": 0.4574, "step": 6156 }, { "epoch": 0.21935621782425138, "grad_norm": 1.770886657908591, "learning_rate": 1.8178095205469154e-05, "loss": 0.6649, "step": 6157 }, { "epoch": 0.21939184495074548, "grad_norm": 1.882437850531944, "learning_rate": 1.8177431073148144e-05, "loss": 0.5123, "step": 6158 }, { "epoch": 0.21942747207723962, "grad_norm": 1.4890422810007473, "learning_rate": 1.8176766831938927e-05, "loss": 0.6181, "step": 6159 }, { "epoch": 0.21946309920373372, "grad_norm": 3.343695024010314, "learning_rate": 1.817610248185034e-05, "loss": 0.6541, "step": 6160 }, { "epoch": 0.21949872633022782, "grad_norm": 1.2017953388711218, "learning_rate": 1.8175438022891238e-05, "loss": 0.5156, "step": 6161 }, { "epoch": 0.21953435345672195, "grad_norm": 1.253431717720117, "learning_rate": 1.8174773455070464e-05, "loss": 0.5585, "step": 6162 }, { "epoch": 0.21956998058321606, "grad_norm": 1.0725861367729115, "learning_rate": 1.8174108778396866e-05, "loss": 0.4543, "step": 6163 }, { "epoch": 0.21960560770971016, "grad_norm": 1.8801300355356738, "learning_rate": 1.8173443992879297e-05, "loss": 0.5756, "step": 6164 }, { "epoch": 0.2196412348362043, "grad_norm": 1.0567562315271473, "learning_rate": 1.8172779098526607e-05, "loss": 0.501, "step": 6165 }, { "epoch": 0.2196768619626984, "grad_norm": 1.1663159526292988, "learning_rate": 1.8172114095347653e-05, "loss": 0.4904, "step": 6166 }, { "epoch": 0.2197124890891925, "grad_norm": 1.2274931287739501, "learning_rate": 1.8171448983351284e-05, "loss": 0.4824, "step": 6167 }, { "epoch": 0.21974811621568663, "grad_norm": 1.1847891221062599, "learning_rate": 1.8170783762546363e-05, "loss": 0.4989, "step": 6168 }, { "epoch": 0.21978374334218073, "grad_norm": 1.657785649579391, "learning_rate": 1.8170118432941748e-05, "loss": 0.5534, "step": 6169 }, { "epoch": 0.21981937046867486, "grad_norm": 1.6293533656494794, "learning_rate": 1.816945299454629e-05, "loss": 0.7097, "step": 6170 }, { "epoch": 0.21985499759516897, "grad_norm": 2.10327101526509, "learning_rate": 1.816878744736886e-05, "loss": 0.5661, "step": 6171 }, { "epoch": 0.21989062472166307, "grad_norm": 1.091120193831298, "learning_rate": 1.8168121791418313e-05, "loss": 0.5147, "step": 6172 }, { "epoch": 0.2199262518481572, "grad_norm": 1.794545313058794, "learning_rate": 1.8167456026703515e-05, "loss": 0.6538, "step": 6173 }, { "epoch": 0.2199618789746513, "grad_norm": 1.7104737157434835, "learning_rate": 1.8166790153233335e-05, "loss": 0.6375, "step": 6174 }, { "epoch": 0.2199975061011454, "grad_norm": 1.555142548101414, "learning_rate": 1.8166124171016634e-05, "loss": 0.617, "step": 6175 }, { "epoch": 0.22003313322763954, "grad_norm": 1.1318018076898946, "learning_rate": 1.816545808006228e-05, "loss": 0.457, "step": 6176 }, { "epoch": 0.22006876035413364, "grad_norm": 2.4895252200394458, "learning_rate": 1.8164791880379145e-05, "loss": 0.642, "step": 6177 }, { "epoch": 0.22010438748062774, "grad_norm": 2.109079052988908, "learning_rate": 1.81641255719761e-05, "loss": 0.6795, "step": 6178 }, { "epoch": 0.22014001460712188, "grad_norm": 2.0079890301630217, "learning_rate": 1.8163459154862015e-05, "loss": 0.5997, "step": 6179 }, { "epoch": 0.22017564173361598, "grad_norm": 1.4020687982283377, "learning_rate": 1.8162792629045765e-05, "loss": 0.5253, "step": 6180 }, { "epoch": 0.22021126886011008, "grad_norm": 1.2710326440209514, "learning_rate": 1.816212599453623e-05, "loss": 0.4877, "step": 6181 }, { "epoch": 0.2202468959866042, "grad_norm": 1.7935060935080664, "learning_rate": 1.816145925134228e-05, "loss": 0.6481, "step": 6182 }, { "epoch": 0.22028252311309832, "grad_norm": 1.7010085861172077, "learning_rate": 1.816079239947279e-05, "loss": 0.6893, "step": 6183 }, { "epoch": 0.22031815023959242, "grad_norm": 1.7136524686212051, "learning_rate": 1.8160125438936652e-05, "loss": 0.5788, "step": 6184 }, { "epoch": 0.22035377736608655, "grad_norm": 1.1665094353177645, "learning_rate": 1.8159458369742742e-05, "loss": 0.4924, "step": 6185 }, { "epoch": 0.22038940449258065, "grad_norm": 1.5740450446489285, "learning_rate": 1.8158791191899934e-05, "loss": 0.6104, "step": 6186 }, { "epoch": 0.22042503161907476, "grad_norm": 1.0374234858026867, "learning_rate": 1.8158123905417124e-05, "loss": 0.5025, "step": 6187 }, { "epoch": 0.2204606587455689, "grad_norm": 1.1923680912295003, "learning_rate": 1.8157456510303188e-05, "loss": 0.5666, "step": 6188 }, { "epoch": 0.220496285872063, "grad_norm": 1.0982504826147765, "learning_rate": 1.8156789006567018e-05, "loss": 0.4453, "step": 6189 }, { "epoch": 0.2205319129985571, "grad_norm": 1.7088147712562722, "learning_rate": 1.8156121394217504e-05, "loss": 0.5258, "step": 6190 }, { "epoch": 0.22056754012505123, "grad_norm": 1.2405486178419491, "learning_rate": 1.815545367326353e-05, "loss": 0.4855, "step": 6191 }, { "epoch": 0.22060316725154533, "grad_norm": 1.9240594891835354, "learning_rate": 1.815478584371399e-05, "loss": 0.7308, "step": 6192 }, { "epoch": 0.22063879437803943, "grad_norm": 2.006928729200545, "learning_rate": 1.8154117905577777e-05, "loss": 0.7266, "step": 6193 }, { "epoch": 0.22067442150453356, "grad_norm": 1.7320607798478127, "learning_rate": 1.8153449858863784e-05, "loss": 0.5771, "step": 6194 }, { "epoch": 0.22071004863102767, "grad_norm": 1.83680881866595, "learning_rate": 1.815278170358091e-05, "loss": 0.6088, "step": 6195 }, { "epoch": 0.22074567575752177, "grad_norm": 1.6121404109223172, "learning_rate": 1.815211343973805e-05, "loss": 0.557, "step": 6196 }, { "epoch": 0.2207813028840159, "grad_norm": 1.8510873526358558, "learning_rate": 1.8151445067344097e-05, "loss": 0.5857, "step": 6197 }, { "epoch": 0.22081693001051, "grad_norm": 1.3549868038119801, "learning_rate": 1.8150776586407957e-05, "loss": 0.6011, "step": 6198 }, { "epoch": 0.2208525571370041, "grad_norm": 1.9673399659364599, "learning_rate": 1.815010799693853e-05, "loss": 0.5847, "step": 6199 }, { "epoch": 0.22088818426349824, "grad_norm": 2.391154437041903, "learning_rate": 1.8149439298944716e-05, "loss": 0.6115, "step": 6200 }, { "epoch": 0.22092381138999234, "grad_norm": 1.7587685818852483, "learning_rate": 1.8148770492435427e-05, "loss": 0.6992, "step": 6201 }, { "epoch": 0.22095943851648644, "grad_norm": 1.1852487228157707, "learning_rate": 1.8148101577419557e-05, "loss": 0.4913, "step": 6202 }, { "epoch": 0.22099506564298058, "grad_norm": 1.3226857237459897, "learning_rate": 1.8147432553906024e-05, "loss": 0.5173, "step": 6203 }, { "epoch": 0.22103069276947468, "grad_norm": 1.5786562026235653, "learning_rate": 1.814676342190373e-05, "loss": 0.5943, "step": 6204 }, { "epoch": 0.22106631989596878, "grad_norm": 1.0793674325114182, "learning_rate": 1.814609418142159e-05, "loss": 0.5125, "step": 6205 }, { "epoch": 0.2211019470224629, "grad_norm": 1.8330784039882815, "learning_rate": 1.8145424832468507e-05, "loss": 0.5905, "step": 6206 }, { "epoch": 0.22113757414895702, "grad_norm": 1.26353728970028, "learning_rate": 1.8144755375053402e-05, "loss": 0.547, "step": 6207 }, { "epoch": 0.22117320127545112, "grad_norm": 2.4352172017101914, "learning_rate": 1.8144085809185184e-05, "loss": 0.5982, "step": 6208 }, { "epoch": 0.22120882840194525, "grad_norm": 1.1866974507001649, "learning_rate": 1.8143416134872775e-05, "loss": 0.5146, "step": 6209 }, { "epoch": 0.22124445552843935, "grad_norm": 1.6850597197300297, "learning_rate": 1.814274635212509e-05, "loss": 0.7327, "step": 6210 }, { "epoch": 0.22128008265493346, "grad_norm": 1.1213266616575548, "learning_rate": 1.814207646095104e-05, "loss": 0.516, "step": 6211 }, { "epoch": 0.2213157097814276, "grad_norm": 1.5763715720884424, "learning_rate": 1.8141406461359553e-05, "loss": 0.6434, "step": 6212 }, { "epoch": 0.2213513369079217, "grad_norm": 1.6323237613372434, "learning_rate": 1.8140736353359546e-05, "loss": 0.5432, "step": 6213 }, { "epoch": 0.2213869640344158, "grad_norm": 1.5283468752609244, "learning_rate": 1.8140066136959945e-05, "loss": 0.6158, "step": 6214 }, { "epoch": 0.22142259116090993, "grad_norm": 2.2313882223210473, "learning_rate": 1.813939581216968e-05, "loss": 0.6902, "step": 6215 }, { "epoch": 0.22145821828740403, "grad_norm": 1.2736069052070425, "learning_rate": 1.8138725378997667e-05, "loss": 0.5279, "step": 6216 }, { "epoch": 0.22149384541389813, "grad_norm": 1.1837217625466778, "learning_rate": 1.8138054837452834e-05, "loss": 0.5149, "step": 6217 }, { "epoch": 0.22152947254039226, "grad_norm": 1.3708568404625432, "learning_rate": 1.8137384187544116e-05, "loss": 0.5934, "step": 6218 }, { "epoch": 0.22156509966688637, "grad_norm": 1.3962883458942974, "learning_rate": 1.8136713429280435e-05, "loss": 0.4924, "step": 6219 }, { "epoch": 0.22160072679338047, "grad_norm": 1.5533012146155383, "learning_rate": 1.813604256267073e-05, "loss": 0.4906, "step": 6220 }, { "epoch": 0.2216363539198746, "grad_norm": 1.7328357220917727, "learning_rate": 1.8135371587723937e-05, "loss": 0.6245, "step": 6221 }, { "epoch": 0.2216719810463687, "grad_norm": 1.9197906829873088, "learning_rate": 1.8134700504448982e-05, "loss": 0.6712, "step": 6222 }, { "epoch": 0.2217076081728628, "grad_norm": 1.6496456821395609, "learning_rate": 1.8134029312854803e-05, "loss": 0.5315, "step": 6223 }, { "epoch": 0.22174323529935694, "grad_norm": 1.2453706985443922, "learning_rate": 1.8133358012950338e-05, "loss": 0.4963, "step": 6224 }, { "epoch": 0.22177886242585104, "grad_norm": 1.2250467267241525, "learning_rate": 1.8132686604744524e-05, "loss": 0.592, "step": 6225 }, { "epoch": 0.22181448955234515, "grad_norm": 1.23429861041282, "learning_rate": 1.8132015088246308e-05, "loss": 0.5082, "step": 6226 }, { "epoch": 0.22185011667883928, "grad_norm": 1.7015215176254368, "learning_rate": 1.8131343463464622e-05, "loss": 0.6091, "step": 6227 }, { "epoch": 0.22188574380533338, "grad_norm": 1.5280277592394087, "learning_rate": 1.813067173040842e-05, "loss": 0.5258, "step": 6228 }, { "epoch": 0.22192137093182748, "grad_norm": 1.4167670473103806, "learning_rate": 1.8129999889086637e-05, "loss": 0.6673, "step": 6229 }, { "epoch": 0.22195699805832161, "grad_norm": 2.2238912636118746, "learning_rate": 1.8129327939508227e-05, "loss": 0.5308, "step": 6230 }, { "epoch": 0.22199262518481572, "grad_norm": 2.759347285071287, "learning_rate": 1.812865588168213e-05, "loss": 0.5979, "step": 6231 }, { "epoch": 0.22202825231130982, "grad_norm": 1.2060753865863836, "learning_rate": 1.81279837156173e-05, "loss": 0.5488, "step": 6232 }, { "epoch": 0.22206387943780395, "grad_norm": 1.438038559242545, "learning_rate": 1.812731144132268e-05, "loss": 0.4969, "step": 6233 }, { "epoch": 0.22209950656429805, "grad_norm": 1.2558029441522336, "learning_rate": 1.8126639058807235e-05, "loss": 0.5222, "step": 6234 }, { "epoch": 0.22213513369079216, "grad_norm": 1.5883388244855994, "learning_rate": 1.8125966568079908e-05, "loss": 0.5711, "step": 6235 }, { "epoch": 0.2221707608172863, "grad_norm": 1.1733395186471915, "learning_rate": 1.8125293969149655e-05, "loss": 0.5069, "step": 6236 }, { "epoch": 0.2222063879437804, "grad_norm": 1.2194432429151, "learning_rate": 1.8124621262025434e-05, "loss": 0.5674, "step": 6237 }, { "epoch": 0.2222420150702745, "grad_norm": 1.931110766060727, "learning_rate": 1.8123948446716204e-05, "loss": 0.5438, "step": 6238 }, { "epoch": 0.22227764219676863, "grad_norm": 1.554503032123883, "learning_rate": 1.812327552323092e-05, "loss": 0.5604, "step": 6239 }, { "epoch": 0.22231326932326273, "grad_norm": 1.2170885267387659, "learning_rate": 1.8122602491578546e-05, "loss": 0.5208, "step": 6240 }, { "epoch": 0.22234889644975683, "grad_norm": 1.0773206576963406, "learning_rate": 1.8121929351768042e-05, "loss": 0.5345, "step": 6241 }, { "epoch": 0.22238452357625096, "grad_norm": 1.824576325686119, "learning_rate": 1.8121256103808372e-05, "loss": 0.5882, "step": 6242 }, { "epoch": 0.22242015070274507, "grad_norm": 1.2493289244570491, "learning_rate": 1.8120582747708503e-05, "loss": 0.4665, "step": 6243 }, { "epoch": 0.22245577782923917, "grad_norm": 1.9934260924758662, "learning_rate": 1.8119909283477394e-05, "loss": 0.5414, "step": 6244 }, { "epoch": 0.2224914049557333, "grad_norm": 1.7177311814702851, "learning_rate": 1.811923571112402e-05, "loss": 0.6053, "step": 6245 }, { "epoch": 0.2225270320822274, "grad_norm": 1.1467024876203333, "learning_rate": 1.8118562030657347e-05, "loss": 0.4991, "step": 6246 }, { "epoch": 0.2225626592087215, "grad_norm": 1.5713886323354131, "learning_rate": 1.811788824208635e-05, "loss": 0.5478, "step": 6247 }, { "epoch": 0.22259828633521564, "grad_norm": 1.7762120680543352, "learning_rate": 1.811721434541999e-05, "loss": 0.6416, "step": 6248 }, { "epoch": 0.22263391346170974, "grad_norm": 1.4042565161175722, "learning_rate": 1.811654034066725e-05, "loss": 0.5258, "step": 6249 }, { "epoch": 0.22266954058820385, "grad_norm": 1.2939101532816693, "learning_rate": 1.81158662278371e-05, "loss": 0.5045, "step": 6250 }, { "epoch": 0.22270516771469798, "grad_norm": 1.1880916212392794, "learning_rate": 1.8115192006938525e-05, "loss": 0.5311, "step": 6251 }, { "epoch": 0.22274079484119208, "grad_norm": 1.7457882762030104, "learning_rate": 1.8114517677980493e-05, "loss": 0.6117, "step": 6252 }, { "epoch": 0.22277642196768618, "grad_norm": 1.8812424746016376, "learning_rate": 1.8113843240971983e-05, "loss": 0.562, "step": 6253 }, { "epoch": 0.22281204909418031, "grad_norm": 1.8855509209495338, "learning_rate": 1.8113168695921985e-05, "loss": 0.678, "step": 6254 }, { "epoch": 0.22284767622067442, "grad_norm": 1.2177002057973367, "learning_rate": 1.811249404283947e-05, "loss": 0.4349, "step": 6255 }, { "epoch": 0.22288330334716852, "grad_norm": 1.3294054916929336, "learning_rate": 1.811181928173343e-05, "loss": 0.5238, "step": 6256 }, { "epoch": 0.22291893047366265, "grad_norm": 1.7943950459702214, "learning_rate": 1.8111144412612846e-05, "loss": 0.648, "step": 6257 }, { "epoch": 0.22295455760015676, "grad_norm": 1.665690037790313, "learning_rate": 1.8110469435486703e-05, "loss": 0.6591, "step": 6258 }, { "epoch": 0.22299018472665086, "grad_norm": 1.4353244006117833, "learning_rate": 1.8109794350363992e-05, "loss": 0.6008, "step": 6259 }, { "epoch": 0.223025811853145, "grad_norm": 1.773887060422404, "learning_rate": 1.8109119157253697e-05, "loss": 0.555, "step": 6260 }, { "epoch": 0.2230614389796391, "grad_norm": 2.2370937199384695, "learning_rate": 1.8108443856164818e-05, "loss": 0.6292, "step": 6261 }, { "epoch": 0.2230970661061332, "grad_norm": 1.1683113329798247, "learning_rate": 1.810776844710634e-05, "loss": 0.4942, "step": 6262 }, { "epoch": 0.22313269323262733, "grad_norm": 1.702610533405462, "learning_rate": 1.8107092930087258e-05, "loss": 0.6703, "step": 6263 }, { "epoch": 0.22316832035912143, "grad_norm": 1.2750410914702959, "learning_rate": 1.8106417305116565e-05, "loss": 0.5641, "step": 6264 }, { "epoch": 0.22320394748561556, "grad_norm": 1.2499676251499057, "learning_rate": 1.8105741572203262e-05, "loss": 0.5018, "step": 6265 }, { "epoch": 0.22323957461210966, "grad_norm": 1.1418605953076215, "learning_rate": 1.8105065731356343e-05, "loss": 0.4746, "step": 6266 }, { "epoch": 0.22327520173860377, "grad_norm": 1.8306507468211692, "learning_rate": 1.810438978258481e-05, "loss": 0.5017, "step": 6267 }, { "epoch": 0.2233108288650979, "grad_norm": 1.8350165878755027, "learning_rate": 1.8103713725897664e-05, "loss": 0.6085, "step": 6268 }, { "epoch": 0.223346455991592, "grad_norm": 1.379271185005254, "learning_rate": 1.8103037561303903e-05, "loss": 0.5431, "step": 6269 }, { "epoch": 0.2233820831180861, "grad_norm": 1.241528694348748, "learning_rate": 1.810236128881253e-05, "loss": 0.5378, "step": 6270 }, { "epoch": 0.22341771024458024, "grad_norm": 1.1353168830316875, "learning_rate": 1.810168490843256e-05, "loss": 0.5256, "step": 6271 }, { "epoch": 0.22345333737107434, "grad_norm": 1.6110607508823895, "learning_rate": 1.8101008420172984e-05, "loss": 0.6163, "step": 6272 }, { "epoch": 0.22348896449756844, "grad_norm": 2.0324554963919272, "learning_rate": 1.8100331824042828e-05, "loss": 0.5952, "step": 6273 }, { "epoch": 0.22352459162406257, "grad_norm": 1.4424436103788503, "learning_rate": 1.8099655120051087e-05, "loss": 0.582, "step": 6274 }, { "epoch": 0.22356021875055668, "grad_norm": 1.3060325873768792, "learning_rate": 1.8098978308206775e-05, "loss": 0.5594, "step": 6275 }, { "epoch": 0.22359584587705078, "grad_norm": 1.1307582880393054, "learning_rate": 1.809830138851891e-05, "loss": 0.504, "step": 6276 }, { "epoch": 0.2236314730035449, "grad_norm": 1.4911261544508694, "learning_rate": 1.80976243609965e-05, "loss": 0.641, "step": 6277 }, { "epoch": 0.22366710013003901, "grad_norm": 1.5214630605824067, "learning_rate": 1.8096947225648566e-05, "loss": 0.5273, "step": 6278 }, { "epoch": 0.22370272725653312, "grad_norm": 1.104791932925854, "learning_rate": 1.8096269982484114e-05, "loss": 0.5079, "step": 6279 }, { "epoch": 0.22373835438302725, "grad_norm": 1.6367292623539345, "learning_rate": 1.809559263151217e-05, "loss": 0.6579, "step": 6280 }, { "epoch": 0.22377398150952135, "grad_norm": 1.207235062126258, "learning_rate": 1.809491517274175e-05, "loss": 0.5399, "step": 6281 }, { "epoch": 0.22380960863601546, "grad_norm": 1.740009082887775, "learning_rate": 1.8094237606181878e-05, "loss": 0.6228, "step": 6282 }, { "epoch": 0.2238452357625096, "grad_norm": 2.6338681961295256, "learning_rate": 1.809355993184158e-05, "loss": 0.5728, "step": 6283 }, { "epoch": 0.2238808628890037, "grad_norm": 1.272516347067455, "learning_rate": 1.8092882149729867e-05, "loss": 0.4678, "step": 6284 }, { "epoch": 0.2239164900154978, "grad_norm": 1.074574679796866, "learning_rate": 1.8092204259855773e-05, "loss": 0.5334, "step": 6285 }, { "epoch": 0.22395211714199192, "grad_norm": 1.2743333860741677, "learning_rate": 1.8091526262228328e-05, "loss": 0.5203, "step": 6286 }, { "epoch": 0.22398774426848603, "grad_norm": 2.660063155788374, "learning_rate": 1.809084815685655e-05, "loss": 0.6396, "step": 6287 }, { "epoch": 0.22402337139498013, "grad_norm": 1.1923776524920662, "learning_rate": 1.8090169943749477e-05, "loss": 0.5557, "step": 6288 }, { "epoch": 0.22405899852147426, "grad_norm": 1.1143628390853817, "learning_rate": 1.8089491622916133e-05, "loss": 0.4894, "step": 6289 }, { "epoch": 0.22409462564796837, "grad_norm": 1.1334817061357985, "learning_rate": 1.8088813194365555e-05, "loss": 0.5355, "step": 6290 }, { "epoch": 0.22413025277446247, "grad_norm": 2.0146256479249445, "learning_rate": 1.808813465810678e-05, "loss": 0.5866, "step": 6291 }, { "epoch": 0.2241658799009566, "grad_norm": 1.7825263663614823, "learning_rate": 1.808745601414883e-05, "loss": 0.6783, "step": 6292 }, { "epoch": 0.2242015070274507, "grad_norm": 2.64066312522784, "learning_rate": 1.808677726250076e-05, "loss": 0.5218, "step": 6293 }, { "epoch": 0.2242371341539448, "grad_norm": 1.5856565114899817, "learning_rate": 1.8086098403171593e-05, "loss": 0.5473, "step": 6294 }, { "epoch": 0.22427276128043894, "grad_norm": 1.5882981835751198, "learning_rate": 1.8085419436170373e-05, "loss": 0.6021, "step": 6295 }, { "epoch": 0.22430838840693304, "grad_norm": 2.113463840554715, "learning_rate": 1.8084740361506143e-05, "loss": 0.5614, "step": 6296 }, { "epoch": 0.22434401553342714, "grad_norm": 1.228160367841859, "learning_rate": 1.8084061179187945e-05, "loss": 0.5277, "step": 6297 }, { "epoch": 0.22437964265992127, "grad_norm": 2.266581377694421, "learning_rate": 1.8083381889224822e-05, "loss": 0.5983, "step": 6298 }, { "epoch": 0.22441526978641538, "grad_norm": 1.1962490047123573, "learning_rate": 1.808270249162582e-05, "loss": 0.515, "step": 6299 }, { "epoch": 0.22445089691290948, "grad_norm": 4.461464107139968, "learning_rate": 1.808202298639998e-05, "loss": 0.6184, "step": 6300 }, { "epoch": 0.2244865240394036, "grad_norm": 1.5002847419438192, "learning_rate": 1.8081343373556363e-05, "loss": 0.5897, "step": 6301 }, { "epoch": 0.22452215116589772, "grad_norm": 1.1197658809446103, "learning_rate": 1.8080663653104005e-05, "loss": 0.4976, "step": 6302 }, { "epoch": 0.22455777829239182, "grad_norm": 1.1223100510663793, "learning_rate": 1.8079983825051964e-05, "loss": 0.5091, "step": 6303 }, { "epoch": 0.22459340541888595, "grad_norm": 1.0871204894631687, "learning_rate": 1.807930388940929e-05, "loss": 0.5165, "step": 6304 }, { "epoch": 0.22462903254538005, "grad_norm": 1.5686482190441697, "learning_rate": 1.807862384618504e-05, "loss": 0.6425, "step": 6305 }, { "epoch": 0.22466465967187416, "grad_norm": 1.1459093595081613, "learning_rate": 1.8077943695388262e-05, "loss": 0.55, "step": 6306 }, { "epoch": 0.2247002867983683, "grad_norm": 1.8945041888020853, "learning_rate": 1.807726343702802e-05, "loss": 0.6594, "step": 6307 }, { "epoch": 0.2247359139248624, "grad_norm": 1.352443914877825, "learning_rate": 1.8076583071113372e-05, "loss": 0.5916, "step": 6308 }, { "epoch": 0.2247715410513565, "grad_norm": 1.3434565049245153, "learning_rate": 1.8075902597653372e-05, "loss": 0.5257, "step": 6309 }, { "epoch": 0.22480716817785062, "grad_norm": 1.7232869676151235, "learning_rate": 1.807522201665709e-05, "loss": 0.5535, "step": 6310 }, { "epoch": 0.22484279530434473, "grad_norm": 1.1506577935949254, "learning_rate": 1.8074541328133574e-05, "loss": 0.53, "step": 6311 }, { "epoch": 0.22487842243083883, "grad_norm": 2.00038457554906, "learning_rate": 1.80738605320919e-05, "loss": 0.6832, "step": 6312 }, { "epoch": 0.22491404955733296, "grad_norm": 1.5250032596983514, "learning_rate": 1.807317962854113e-05, "loss": 0.6075, "step": 6313 }, { "epoch": 0.22494967668382707, "grad_norm": 1.1925167902155203, "learning_rate": 1.8072498617490334e-05, "loss": 0.4872, "step": 6314 }, { "epoch": 0.22498530381032117, "grad_norm": 2.1967361365215723, "learning_rate": 1.807181749894857e-05, "loss": 0.5885, "step": 6315 }, { "epoch": 0.2250209309368153, "grad_norm": 1.3170284927438605, "learning_rate": 1.807113627292492e-05, "loss": 0.5163, "step": 6316 }, { "epoch": 0.2250565580633094, "grad_norm": 1.5028780839312534, "learning_rate": 1.8070454939428446e-05, "loss": 0.5775, "step": 6317 }, { "epoch": 0.2250921851898035, "grad_norm": 1.1808202403598536, "learning_rate": 1.8069773498468224e-05, "loss": 0.5621, "step": 6318 }, { "epoch": 0.22512781231629764, "grad_norm": 1.8273285957859962, "learning_rate": 1.8069091950053324e-05, "loss": 0.6573, "step": 6319 }, { "epoch": 0.22516343944279174, "grad_norm": 1.1549955896571282, "learning_rate": 1.806841029419283e-05, "loss": 0.4554, "step": 6320 }, { "epoch": 0.22519906656928584, "grad_norm": 1.269803220433351, "learning_rate": 1.8067728530895812e-05, "loss": 0.5439, "step": 6321 }, { "epoch": 0.22523469369577998, "grad_norm": 1.7230855687446103, "learning_rate": 1.806704666017135e-05, "loss": 0.5846, "step": 6322 }, { "epoch": 0.22527032082227408, "grad_norm": 1.142386193909555, "learning_rate": 1.8066364682028522e-05, "loss": 0.5072, "step": 6323 }, { "epoch": 0.22530594794876818, "grad_norm": 1.3800753260381018, "learning_rate": 1.8065682596476413e-05, "loss": 0.6329, "step": 6324 }, { "epoch": 0.2253415750752623, "grad_norm": 1.1932399596572527, "learning_rate": 1.80650004035241e-05, "loss": 0.4636, "step": 6325 }, { "epoch": 0.22537720220175642, "grad_norm": 2.2094385908635794, "learning_rate": 1.806431810318067e-05, "loss": 0.634, "step": 6326 }, { "epoch": 0.22541282932825052, "grad_norm": 2.3158980373232785, "learning_rate": 1.806363569545521e-05, "loss": 0.547, "step": 6327 }, { "epoch": 0.22544845645474465, "grad_norm": 1.5588160713994248, "learning_rate": 1.80629531803568e-05, "loss": 0.5354, "step": 6328 }, { "epoch": 0.22548408358123875, "grad_norm": 1.8850572827242897, "learning_rate": 1.8062270557894533e-05, "loss": 0.6627, "step": 6329 }, { "epoch": 0.22551971070773286, "grad_norm": 1.2091375904479167, "learning_rate": 1.8061587828077503e-05, "loss": 0.5353, "step": 6330 }, { "epoch": 0.225555337834227, "grad_norm": 1.1933412613520455, "learning_rate": 1.806090499091479e-05, "loss": 0.5502, "step": 6331 }, { "epoch": 0.2255909649607211, "grad_norm": 1.1289100018120528, "learning_rate": 1.8060222046415502e-05, "loss": 0.5449, "step": 6332 }, { "epoch": 0.2256265920872152, "grad_norm": 1.7307923477472322, "learning_rate": 1.8059538994588715e-05, "loss": 0.5816, "step": 6333 }, { "epoch": 0.22566221921370933, "grad_norm": 1.5353357705466808, "learning_rate": 1.805885583544354e-05, "loss": 0.5787, "step": 6334 }, { "epoch": 0.22569784634020343, "grad_norm": 1.5498092372077732, "learning_rate": 1.8058172568989062e-05, "loss": 0.5659, "step": 6335 }, { "epoch": 0.22573347346669753, "grad_norm": 1.6602220366228524, "learning_rate": 1.8057489195234384e-05, "loss": 0.6133, "step": 6336 }, { "epoch": 0.22576910059319166, "grad_norm": 1.372460487879214, "learning_rate": 1.8056805714188606e-05, "loss": 0.5537, "step": 6337 }, { "epoch": 0.22580472771968577, "grad_norm": 1.5267808124127071, "learning_rate": 1.8056122125860827e-05, "loss": 0.6317, "step": 6338 }, { "epoch": 0.22584035484617987, "grad_norm": 1.7815922253143623, "learning_rate": 1.8055438430260154e-05, "loss": 0.5988, "step": 6339 }, { "epoch": 0.225875981972674, "grad_norm": 1.253330069326722, "learning_rate": 1.8054754627395687e-05, "loss": 0.5133, "step": 6340 }, { "epoch": 0.2259116090991681, "grad_norm": 2.53011641944033, "learning_rate": 1.805407071727653e-05, "loss": 0.5997, "step": 6341 }, { "epoch": 0.2259472362256622, "grad_norm": 1.831682766081432, "learning_rate": 1.8053386699911795e-05, "loss": 0.6021, "step": 6342 }, { "epoch": 0.22598286335215634, "grad_norm": 1.757016366217332, "learning_rate": 1.8052702575310588e-05, "loss": 0.6625, "step": 6343 }, { "epoch": 0.22601849047865044, "grad_norm": 1.1346638132192979, "learning_rate": 1.8052018343482016e-05, "loss": 0.5221, "step": 6344 }, { "epoch": 0.22605411760514454, "grad_norm": 1.0391657547920585, "learning_rate": 1.805133400443519e-05, "loss": 0.4682, "step": 6345 }, { "epoch": 0.22608974473163868, "grad_norm": 1.1655615854095356, "learning_rate": 1.805064955817923e-05, "loss": 0.491, "step": 6346 }, { "epoch": 0.22612537185813278, "grad_norm": 1.3104069659831872, "learning_rate": 1.8049965004723238e-05, "loss": 0.516, "step": 6347 }, { "epoch": 0.22616099898462688, "grad_norm": 1.0555806069747546, "learning_rate": 1.8049280344076335e-05, "loss": 0.4695, "step": 6348 }, { "epoch": 0.226196626111121, "grad_norm": 1.0493390803992955, "learning_rate": 1.8048595576247642e-05, "loss": 0.538, "step": 6349 }, { "epoch": 0.22623225323761512, "grad_norm": 1.1612417423436017, "learning_rate": 1.804791070124627e-05, "loss": 0.5343, "step": 6350 }, { "epoch": 0.22626788036410922, "grad_norm": 1.7735327196377215, "learning_rate": 1.8047225719081344e-05, "loss": 0.6473, "step": 6351 }, { "epoch": 0.22630350749060335, "grad_norm": 1.2188563903066196, "learning_rate": 1.8046540629761983e-05, "loss": 0.5164, "step": 6352 }, { "epoch": 0.22633913461709745, "grad_norm": 2.034635238035914, "learning_rate": 1.804585543329731e-05, "loss": 0.5763, "step": 6353 }, { "epoch": 0.22637476174359156, "grad_norm": 1.6796937269539076, "learning_rate": 1.8045170129696444e-05, "loss": 0.6652, "step": 6354 }, { "epoch": 0.2264103888700857, "grad_norm": 1.5115895090334859, "learning_rate": 1.8044484718968517e-05, "loss": 0.6214, "step": 6355 }, { "epoch": 0.2264460159965798, "grad_norm": 1.0665804740980287, "learning_rate": 1.8043799201122656e-05, "loss": 0.558, "step": 6356 }, { "epoch": 0.2264816431230739, "grad_norm": 1.2050144968234888, "learning_rate": 1.8043113576167985e-05, "loss": 0.5014, "step": 6357 }, { "epoch": 0.22651727024956803, "grad_norm": 1.1804702687558701, "learning_rate": 1.804242784411363e-05, "loss": 0.5225, "step": 6358 }, { "epoch": 0.22655289737606213, "grad_norm": 1.2027403500478253, "learning_rate": 1.804174200496873e-05, "loss": 0.5427, "step": 6359 }, { "epoch": 0.22658852450255626, "grad_norm": 1.5767528705542995, "learning_rate": 1.8041056058742418e-05, "loss": 0.6362, "step": 6360 }, { "epoch": 0.22662415162905036, "grad_norm": 1.5271020759016563, "learning_rate": 1.8040370005443822e-05, "loss": 0.6151, "step": 6361 }, { "epoch": 0.22665977875554447, "grad_norm": 1.1687103803308188, "learning_rate": 1.8039683845082077e-05, "loss": 0.5425, "step": 6362 }, { "epoch": 0.2266954058820386, "grad_norm": 1.8306184303667423, "learning_rate": 1.8038997577666326e-05, "loss": 0.6795, "step": 6363 }, { "epoch": 0.2267310330085327, "grad_norm": 1.143872065801663, "learning_rate": 1.8038311203205698e-05, "loss": 0.5072, "step": 6364 }, { "epoch": 0.2267666601350268, "grad_norm": 1.5559007531211848, "learning_rate": 1.8037624721709344e-05, "loss": 0.5179, "step": 6365 }, { "epoch": 0.22680228726152094, "grad_norm": 1.5871349301892526, "learning_rate": 1.8036938133186397e-05, "loss": 0.6349, "step": 6366 }, { "epoch": 0.22683791438801504, "grad_norm": 1.0299394833707929, "learning_rate": 1.8036251437645998e-05, "loss": 0.4778, "step": 6367 }, { "epoch": 0.22687354151450914, "grad_norm": 1.7391154549466499, "learning_rate": 1.8035564635097296e-05, "loss": 0.5917, "step": 6368 }, { "epoch": 0.22690916864100327, "grad_norm": 1.7187226337489303, "learning_rate": 1.8034877725549435e-05, "loss": 0.6609, "step": 6369 }, { "epoch": 0.22694479576749738, "grad_norm": 1.13217078916357, "learning_rate": 1.803419070901156e-05, "loss": 0.5123, "step": 6370 }, { "epoch": 0.22698042289399148, "grad_norm": 1.7734081166002873, "learning_rate": 1.8033503585492823e-05, "loss": 0.5918, "step": 6371 }, { "epoch": 0.2270160500204856, "grad_norm": 1.4572486911725144, "learning_rate": 1.803281635500237e-05, "loss": 0.5865, "step": 6372 }, { "epoch": 0.2270516771469797, "grad_norm": 1.5786839079641075, "learning_rate": 1.8032129017549353e-05, "loss": 0.6003, "step": 6373 }, { "epoch": 0.22708730427347382, "grad_norm": 1.304963154805963, "learning_rate": 1.803144157314292e-05, "loss": 0.5444, "step": 6374 }, { "epoch": 0.22712293139996795, "grad_norm": 1.2439943156973432, "learning_rate": 1.803075402179223e-05, "loss": 0.5628, "step": 6375 }, { "epoch": 0.22715855852646205, "grad_norm": 1.6911811649176414, "learning_rate": 1.8030066363506438e-05, "loss": 0.6674, "step": 6376 }, { "epoch": 0.22719418565295615, "grad_norm": 1.5319695435622345, "learning_rate": 1.80293785982947e-05, "loss": 0.5746, "step": 6377 }, { "epoch": 0.22722981277945029, "grad_norm": 1.1196080935598902, "learning_rate": 1.8028690726166172e-05, "loss": 0.4885, "step": 6378 }, { "epoch": 0.2272654399059444, "grad_norm": 1.162275411278357, "learning_rate": 1.802800274713002e-05, "loss": 0.5463, "step": 6379 }, { "epoch": 0.2273010670324385, "grad_norm": 1.2085438989703465, "learning_rate": 1.8027314661195397e-05, "loss": 0.5313, "step": 6380 }, { "epoch": 0.22733669415893262, "grad_norm": 1.29066673266196, "learning_rate": 1.8026626468371468e-05, "loss": 0.5114, "step": 6381 }, { "epoch": 0.22737232128542673, "grad_norm": 1.1976418492121272, "learning_rate": 1.8025938168667398e-05, "loss": 0.5376, "step": 6382 }, { "epoch": 0.22740794841192083, "grad_norm": 1.1490047865210264, "learning_rate": 1.802524976209235e-05, "loss": 0.5055, "step": 6383 }, { "epoch": 0.22744357553841496, "grad_norm": 1.122583784980831, "learning_rate": 1.8024561248655494e-05, "loss": 0.5241, "step": 6384 }, { "epoch": 0.22747920266490906, "grad_norm": 2.0131769228599716, "learning_rate": 1.8023872628365997e-05, "loss": 0.6172, "step": 6385 }, { "epoch": 0.22751482979140317, "grad_norm": 1.1103173299157194, "learning_rate": 1.8023183901233024e-05, "loss": 0.4994, "step": 6386 }, { "epoch": 0.2275504569178973, "grad_norm": 1.7641276393524128, "learning_rate": 1.802249506726575e-05, "loss": 0.6699, "step": 6387 }, { "epoch": 0.2275860840443914, "grad_norm": 1.0958265798275122, "learning_rate": 1.8021806126473353e-05, "loss": 0.512, "step": 6388 }, { "epoch": 0.2276217111708855, "grad_norm": 1.0974327196436526, "learning_rate": 1.8021117078864994e-05, "loss": 0.4862, "step": 6389 }, { "epoch": 0.22765733829737964, "grad_norm": 1.5561279449676362, "learning_rate": 1.802042792444986e-05, "loss": 0.5968, "step": 6390 }, { "epoch": 0.22769296542387374, "grad_norm": 1.6092117603864917, "learning_rate": 1.8019738663237115e-05, "loss": 0.5135, "step": 6391 }, { "epoch": 0.22772859255036784, "grad_norm": 1.1770960024712727, "learning_rate": 1.801904929523595e-05, "loss": 0.4987, "step": 6392 }, { "epoch": 0.22776421967686197, "grad_norm": 1.183018662038673, "learning_rate": 1.8018359820455535e-05, "loss": 0.602, "step": 6393 }, { "epoch": 0.22779984680335608, "grad_norm": 1.194557504893146, "learning_rate": 1.8017670238905057e-05, "loss": 0.5206, "step": 6394 }, { "epoch": 0.22783547392985018, "grad_norm": 1.226030888167879, "learning_rate": 1.8016980550593695e-05, "loss": 0.4959, "step": 6395 }, { "epoch": 0.2278711010563443, "grad_norm": 2.2294510410069446, "learning_rate": 1.8016290755530636e-05, "loss": 0.6118, "step": 6396 }, { "epoch": 0.22790672818283841, "grad_norm": 1.4055453802685283, "learning_rate": 1.801560085372506e-05, "loss": 0.5649, "step": 6397 }, { "epoch": 0.22794235530933252, "grad_norm": 1.168747848503025, "learning_rate": 1.8014910845186154e-05, "loss": 0.4744, "step": 6398 }, { "epoch": 0.22797798243582665, "grad_norm": 1.1735482664763657, "learning_rate": 1.801422072992311e-05, "loss": 0.5068, "step": 6399 }, { "epoch": 0.22801360956232075, "grad_norm": 1.5418731766503486, "learning_rate": 1.8013530507945115e-05, "loss": 0.6314, "step": 6400 }, { "epoch": 0.22804923668881485, "grad_norm": 1.3118575893712785, "learning_rate": 1.801284017926136e-05, "loss": 0.5311, "step": 6401 }, { "epoch": 0.22808486381530899, "grad_norm": 1.7310039844410108, "learning_rate": 1.801214974388104e-05, "loss": 0.6108, "step": 6402 }, { "epoch": 0.2281204909418031, "grad_norm": 1.236681423039565, "learning_rate": 1.8011459201813345e-05, "loss": 0.4989, "step": 6403 }, { "epoch": 0.2281561180682972, "grad_norm": 1.7259709041856943, "learning_rate": 1.801076855306747e-05, "loss": 0.5501, "step": 6404 }, { "epoch": 0.22819174519479132, "grad_norm": 1.0748480956331736, "learning_rate": 1.8010077797652613e-05, "loss": 0.5002, "step": 6405 }, { "epoch": 0.22822737232128543, "grad_norm": 1.5875582250584865, "learning_rate": 1.800938693557797e-05, "loss": 0.5718, "step": 6406 }, { "epoch": 0.22826299944777953, "grad_norm": 1.129091616732911, "learning_rate": 1.8008695966852743e-05, "loss": 0.5214, "step": 6407 }, { "epoch": 0.22829862657427366, "grad_norm": 1.8969011071412432, "learning_rate": 1.800800489148613e-05, "loss": 0.6244, "step": 6408 }, { "epoch": 0.22833425370076776, "grad_norm": 1.1164298284356846, "learning_rate": 1.8007313709487334e-05, "loss": 0.4567, "step": 6409 }, { "epoch": 0.22836988082726187, "grad_norm": 1.1051253560373644, "learning_rate": 1.8006622420865564e-05, "loss": 0.5035, "step": 6410 }, { "epoch": 0.228405507953756, "grad_norm": 1.1770413585057238, "learning_rate": 1.8005931025630016e-05, "loss": 0.5446, "step": 6411 }, { "epoch": 0.2284411350802501, "grad_norm": 1.5040761775572387, "learning_rate": 1.8005239523789903e-05, "loss": 0.5647, "step": 6412 }, { "epoch": 0.2284767622067442, "grad_norm": 1.3425930965301571, "learning_rate": 1.800454791535443e-05, "loss": 0.5153, "step": 6413 }, { "epoch": 0.22851238933323834, "grad_norm": 1.080092911610815, "learning_rate": 1.8003856200332804e-05, "loss": 0.4941, "step": 6414 }, { "epoch": 0.22854801645973244, "grad_norm": 1.5171300281642428, "learning_rate": 1.800316437873424e-05, "loss": 0.582, "step": 6415 }, { "epoch": 0.22858364358622654, "grad_norm": 1.1577901586953279, "learning_rate": 1.800247245056795e-05, "loss": 0.4624, "step": 6416 }, { "epoch": 0.22861927071272067, "grad_norm": 1.4857594431216674, "learning_rate": 1.800178041584314e-05, "loss": 0.6091, "step": 6417 }, { "epoch": 0.22865489783921478, "grad_norm": 1.1806798120034392, "learning_rate": 1.800108827456904e-05, "loss": 0.4704, "step": 6418 }, { "epoch": 0.22869052496570888, "grad_norm": 1.1000919477923325, "learning_rate": 1.800039602675485e-05, "loss": 0.4894, "step": 6419 }, { "epoch": 0.228726152092203, "grad_norm": 1.2697656181403336, "learning_rate": 1.7999703672409802e-05, "loss": 0.5492, "step": 6420 }, { "epoch": 0.22876177921869711, "grad_norm": 1.6270399875895578, "learning_rate": 1.79990112115431e-05, "loss": 0.6793, "step": 6421 }, { "epoch": 0.22879740634519122, "grad_norm": 1.371664654129956, "learning_rate": 1.799831864416398e-05, "loss": 0.5602, "step": 6422 }, { "epoch": 0.22883303347168535, "grad_norm": 1.6376282518893224, "learning_rate": 1.7997625970281652e-05, "loss": 0.5703, "step": 6423 }, { "epoch": 0.22886866059817945, "grad_norm": 1.901399972378867, "learning_rate": 1.7996933189905347e-05, "loss": 0.6111, "step": 6424 }, { "epoch": 0.22890428772467356, "grad_norm": 1.6791107152328142, "learning_rate": 1.799624030304428e-05, "loss": 0.5713, "step": 6425 }, { "epoch": 0.2289399148511677, "grad_norm": 1.192761959609426, "learning_rate": 1.799554730970769e-05, "loss": 0.5299, "step": 6426 }, { "epoch": 0.2289755419776618, "grad_norm": 1.430602979990646, "learning_rate": 1.79948542099048e-05, "loss": 0.6132, "step": 6427 }, { "epoch": 0.2290111691041559, "grad_norm": 1.2084678661823667, "learning_rate": 1.7994161003644838e-05, "loss": 0.5347, "step": 6428 }, { "epoch": 0.22904679623065002, "grad_norm": 1.577973016222441, "learning_rate": 1.7993467690937033e-05, "loss": 0.6176, "step": 6429 }, { "epoch": 0.22908242335714413, "grad_norm": 1.2947035618693203, "learning_rate": 1.799277427179062e-05, "loss": 0.5409, "step": 6430 }, { "epoch": 0.22911805048363823, "grad_norm": 1.1971708907816478, "learning_rate": 1.799208074621483e-05, "loss": 0.5228, "step": 6431 }, { "epoch": 0.22915367761013236, "grad_norm": 1.1367228495745039, "learning_rate": 1.79913871142189e-05, "loss": 0.5411, "step": 6432 }, { "epoch": 0.22918930473662646, "grad_norm": 1.1640077881174602, "learning_rate": 1.799069337581206e-05, "loss": 0.5159, "step": 6433 }, { "epoch": 0.22922493186312057, "grad_norm": 1.4818110792185013, "learning_rate": 1.7989999531003558e-05, "loss": 0.5548, "step": 6434 }, { "epoch": 0.2292605589896147, "grad_norm": 1.1310923150537375, "learning_rate": 1.7989305579802625e-05, "loss": 0.5255, "step": 6435 }, { "epoch": 0.2292961861161088, "grad_norm": 1.6198470922928057, "learning_rate": 1.7988611522218505e-05, "loss": 0.5305, "step": 6436 }, { "epoch": 0.2293318132426029, "grad_norm": 1.2051259444616664, "learning_rate": 1.798791735826044e-05, "loss": 0.5625, "step": 6437 }, { "epoch": 0.22936744036909704, "grad_norm": 1.8406165751413752, "learning_rate": 1.7987223087937672e-05, "loss": 0.6313, "step": 6438 }, { "epoch": 0.22940306749559114, "grad_norm": 1.5857108747422268, "learning_rate": 1.7986528711259442e-05, "loss": 0.6323, "step": 6439 }, { "epoch": 0.22943869462208524, "grad_norm": 1.091960005559875, "learning_rate": 1.7985834228235002e-05, "loss": 0.4609, "step": 6440 }, { "epoch": 0.22947432174857937, "grad_norm": 2.07512315375369, "learning_rate": 1.7985139638873603e-05, "loss": 0.6374, "step": 6441 }, { "epoch": 0.22950994887507348, "grad_norm": 1.6857528594370685, "learning_rate": 1.798444494318448e-05, "loss": 0.5358, "step": 6442 }, { "epoch": 0.22954557600156758, "grad_norm": 1.6239172469718701, "learning_rate": 1.7983750141176898e-05, "loss": 0.5911, "step": 6443 }, { "epoch": 0.2295812031280617, "grad_norm": 1.8453806360652258, "learning_rate": 1.79830552328601e-05, "loss": 0.6291, "step": 6444 }, { "epoch": 0.22961683025455581, "grad_norm": 1.1299887619392044, "learning_rate": 1.798236021824334e-05, "loss": 0.5163, "step": 6445 }, { "epoch": 0.22965245738104992, "grad_norm": 1.590139251462227, "learning_rate": 1.7981665097335875e-05, "loss": 0.5714, "step": 6446 }, { "epoch": 0.22968808450754405, "grad_norm": 2.271824784409793, "learning_rate": 1.7980969870146963e-05, "loss": 0.6637, "step": 6447 }, { "epoch": 0.22972371163403815, "grad_norm": 1.1058807808587146, "learning_rate": 1.798027453668586e-05, "loss": 0.5237, "step": 6448 }, { "epoch": 0.22975933876053226, "grad_norm": 1.9703668125326668, "learning_rate": 1.797957909696182e-05, "loss": 0.5571, "step": 6449 }, { "epoch": 0.2297949658870264, "grad_norm": 1.1858172035939745, "learning_rate": 1.797888355098411e-05, "loss": 0.5465, "step": 6450 }, { "epoch": 0.2298305930135205, "grad_norm": 1.0512094633769413, "learning_rate": 1.7978187898761982e-05, "loss": 0.5183, "step": 6451 }, { "epoch": 0.22986622014001462, "grad_norm": 2.0022430347335582, "learning_rate": 1.7977492140304713e-05, "loss": 0.6704, "step": 6452 }, { "epoch": 0.22990184726650872, "grad_norm": 0.9906246493216054, "learning_rate": 1.7976796275621556e-05, "loss": 0.5002, "step": 6453 }, { "epoch": 0.22993747439300283, "grad_norm": 1.6269104337297182, "learning_rate": 1.7976100304721786e-05, "loss": 0.6634, "step": 6454 }, { "epoch": 0.22997310151949696, "grad_norm": 1.2181922804376142, "learning_rate": 1.7975404227614658e-05, "loss": 0.4924, "step": 6455 }, { "epoch": 0.23000872864599106, "grad_norm": 1.105895760377535, "learning_rate": 1.7974708044309453e-05, "loss": 0.5054, "step": 6456 }, { "epoch": 0.23004435577248517, "grad_norm": 1.3302667324631843, "learning_rate": 1.7974011754815435e-05, "loss": 0.5295, "step": 6457 }, { "epoch": 0.2300799828989793, "grad_norm": 1.6527287752399697, "learning_rate": 1.7973315359141875e-05, "loss": 0.552, "step": 6458 }, { "epoch": 0.2301156100254734, "grad_norm": 1.5652834496040147, "learning_rate": 1.7972618857298055e-05, "loss": 0.559, "step": 6459 }, { "epoch": 0.2301512371519675, "grad_norm": 2.114202004983222, "learning_rate": 1.7971922249293238e-05, "loss": 0.6436, "step": 6460 }, { "epoch": 0.23018686427846163, "grad_norm": 1.6060541750689155, "learning_rate": 1.7971225535136703e-05, "loss": 0.6442, "step": 6461 }, { "epoch": 0.23022249140495574, "grad_norm": 1.2772275038475196, "learning_rate": 1.7970528714837728e-05, "loss": 0.4641, "step": 6462 }, { "epoch": 0.23025811853144984, "grad_norm": 1.2858791511300423, "learning_rate": 1.7969831788405592e-05, "loss": 0.4707, "step": 6463 }, { "epoch": 0.23029374565794397, "grad_norm": 1.864524343127792, "learning_rate": 1.796913475584958e-05, "loss": 0.6773, "step": 6464 }, { "epoch": 0.23032937278443807, "grad_norm": 1.6170210174791948, "learning_rate": 1.7968437617178966e-05, "loss": 0.6068, "step": 6465 }, { "epoch": 0.23036499991093218, "grad_norm": 4.983146336148104, "learning_rate": 1.7967740372403036e-05, "loss": 0.6087, "step": 6466 }, { "epoch": 0.2304006270374263, "grad_norm": 2.145442749321616, "learning_rate": 1.7967043021531073e-05, "loss": 0.5925, "step": 6467 }, { "epoch": 0.2304362541639204, "grad_norm": 1.7599875174947626, "learning_rate": 1.796634556457236e-05, "loss": 0.6168, "step": 6468 }, { "epoch": 0.23047188129041452, "grad_norm": 1.6557197926759437, "learning_rate": 1.7965648001536193e-05, "loss": 0.5407, "step": 6469 }, { "epoch": 0.23050750841690865, "grad_norm": 1.3527077130598872, "learning_rate": 1.7964950332431854e-05, "loss": 0.5036, "step": 6470 }, { "epoch": 0.23054313554340275, "grad_norm": 1.3506723136381957, "learning_rate": 1.7964252557268633e-05, "loss": 0.5705, "step": 6471 }, { "epoch": 0.23057876266989685, "grad_norm": 3.671324151348635, "learning_rate": 1.7963554676055823e-05, "loss": 0.6281, "step": 6472 }, { "epoch": 0.23061438979639098, "grad_norm": 1.4581989993537032, "learning_rate": 1.7962856688802717e-05, "loss": 0.5229, "step": 6473 }, { "epoch": 0.2306500169228851, "grad_norm": 1.4245038140265749, "learning_rate": 1.7962158595518607e-05, "loss": 0.5367, "step": 6474 }, { "epoch": 0.2306856440493792, "grad_norm": 1.8537109919111405, "learning_rate": 1.796146039621279e-05, "loss": 0.5113, "step": 6475 }, { "epoch": 0.23072127117587332, "grad_norm": 1.580831770370579, "learning_rate": 1.7960762090894565e-05, "loss": 0.6281, "step": 6476 }, { "epoch": 0.23075689830236742, "grad_norm": 1.283148228779879, "learning_rate": 1.796006367957323e-05, "loss": 0.52, "step": 6477 }, { "epoch": 0.23079252542886153, "grad_norm": 1.1951955868751976, "learning_rate": 1.795936516225808e-05, "loss": 0.4768, "step": 6478 }, { "epoch": 0.23082815255535566, "grad_norm": 1.9698879077140206, "learning_rate": 1.795866653895842e-05, "loss": 0.6257, "step": 6479 }, { "epoch": 0.23086377968184976, "grad_norm": 5.20800907159323, "learning_rate": 1.795796780968355e-05, "loss": 0.5901, "step": 6480 }, { "epoch": 0.23089940680834387, "grad_norm": 3.3777380552457483, "learning_rate": 1.795726897444278e-05, "loss": 0.5899, "step": 6481 }, { "epoch": 0.230935033934838, "grad_norm": 1.1998980432499615, "learning_rate": 1.795657003324541e-05, "loss": 0.521, "step": 6482 }, { "epoch": 0.2309706610613321, "grad_norm": 1.3943573305473418, "learning_rate": 1.795587098610075e-05, "loss": 0.5156, "step": 6483 }, { "epoch": 0.2310062881878262, "grad_norm": 1.429056717070296, "learning_rate": 1.7955171833018107e-05, "loss": 0.4636, "step": 6484 }, { "epoch": 0.23104191531432033, "grad_norm": 1.7465596257637124, "learning_rate": 1.795447257400679e-05, "loss": 0.5636, "step": 6485 }, { "epoch": 0.23107754244081444, "grad_norm": 1.392320448854406, "learning_rate": 1.795377320907611e-05, "loss": 0.4746, "step": 6486 }, { "epoch": 0.23111316956730854, "grad_norm": 1.33713237119586, "learning_rate": 1.7953073738235382e-05, "loss": 0.5241, "step": 6487 }, { "epoch": 0.23114879669380267, "grad_norm": 2.237926214772548, "learning_rate": 1.795237416149392e-05, "loss": 0.6022, "step": 6488 }, { "epoch": 0.23118442382029677, "grad_norm": 2.455269450280065, "learning_rate": 1.7951674478861036e-05, "loss": 0.6286, "step": 6489 }, { "epoch": 0.23122005094679088, "grad_norm": 2.6429482638221833, "learning_rate": 1.795097469034605e-05, "loss": 0.5335, "step": 6490 }, { "epoch": 0.231255678073285, "grad_norm": 2.3686415108028553, "learning_rate": 1.7950274795958272e-05, "loss": 0.6503, "step": 6491 }, { "epoch": 0.2312913051997791, "grad_norm": 4.867630797489745, "learning_rate": 1.7949574795707034e-05, "loss": 0.5436, "step": 6492 }, { "epoch": 0.23132693232627322, "grad_norm": 1.5011182191045493, "learning_rate": 1.794887468960165e-05, "loss": 0.6156, "step": 6493 }, { "epoch": 0.23136255945276735, "grad_norm": 2.0878071062236385, "learning_rate": 1.7948174477651445e-05, "loss": 0.5392, "step": 6494 }, { "epoch": 0.23139818657926145, "grad_norm": 1.7778740743072508, "learning_rate": 1.794747415986574e-05, "loss": 0.7081, "step": 6495 }, { "epoch": 0.23143381370575555, "grad_norm": 1.8609267044315567, "learning_rate": 1.794677373625386e-05, "loss": 0.6681, "step": 6496 }, { "epoch": 0.23146944083224968, "grad_norm": 2.4485107072702372, "learning_rate": 1.7946073206825134e-05, "loss": 0.5578, "step": 6497 }, { "epoch": 0.2315050679587438, "grad_norm": 1.6523931809599488, "learning_rate": 1.794537257158889e-05, "loss": 0.4773, "step": 6498 }, { "epoch": 0.2315406950852379, "grad_norm": 1.3527474765182854, "learning_rate": 1.7944671830554455e-05, "loss": 0.5391, "step": 6499 }, { "epoch": 0.23157632221173202, "grad_norm": 1.4190182105369382, "learning_rate": 1.7943970983731164e-05, "loss": 0.4963, "step": 6500 }, { "epoch": 0.23161194933822613, "grad_norm": 1.301260670987213, "learning_rate": 1.7943270031128345e-05, "loss": 0.5797, "step": 6501 }, { "epoch": 0.23164757646472023, "grad_norm": 1.6907974764155698, "learning_rate": 1.7942568972755336e-05, "loss": 0.6232, "step": 6502 }, { "epoch": 0.23168320359121436, "grad_norm": 1.5811816742254055, "learning_rate": 1.7941867808621468e-05, "loss": 0.6687, "step": 6503 }, { "epoch": 0.23171883071770846, "grad_norm": 1.12013141018202, "learning_rate": 1.7941166538736082e-05, "loss": 0.5395, "step": 6504 }, { "epoch": 0.23175445784420257, "grad_norm": 1.5847587851914082, "learning_rate": 1.794046516310851e-05, "loss": 0.5872, "step": 6505 }, { "epoch": 0.2317900849706967, "grad_norm": 1.2685208723437886, "learning_rate": 1.7939763681748094e-05, "loss": 0.5233, "step": 6506 }, { "epoch": 0.2318257120971908, "grad_norm": 1.1915026255271408, "learning_rate": 1.7939062094664174e-05, "loss": 0.5051, "step": 6507 }, { "epoch": 0.2318613392236849, "grad_norm": 1.2471885140982866, "learning_rate": 1.7938360401866096e-05, "loss": 0.4885, "step": 6508 }, { "epoch": 0.23189696635017903, "grad_norm": 1.1475172557827493, "learning_rate": 1.7937658603363196e-05, "loss": 0.5433, "step": 6509 }, { "epoch": 0.23193259347667314, "grad_norm": 1.5230125144255535, "learning_rate": 1.7936956699164828e-05, "loss": 0.539, "step": 6510 }, { "epoch": 0.23196822060316724, "grad_norm": 1.1660263977821694, "learning_rate": 1.7936254689280336e-05, "loss": 0.5095, "step": 6511 }, { "epoch": 0.23200384772966137, "grad_norm": 1.1430849279441773, "learning_rate": 1.793555257371906e-05, "loss": 0.5075, "step": 6512 }, { "epoch": 0.23203947485615548, "grad_norm": 1.9200317151695903, "learning_rate": 1.793485035249036e-05, "loss": 0.5109, "step": 6513 }, { "epoch": 0.23207510198264958, "grad_norm": 1.7366911635766298, "learning_rate": 1.7934148025603576e-05, "loss": 0.5545, "step": 6514 }, { "epoch": 0.2321107291091437, "grad_norm": 1.6040538891828613, "learning_rate": 1.793344559306807e-05, "loss": 0.5323, "step": 6515 }, { "epoch": 0.2321463562356378, "grad_norm": 1.7342421510842254, "learning_rate": 1.793274305489319e-05, "loss": 0.5974, "step": 6516 }, { "epoch": 0.23218198336213192, "grad_norm": 1.313388088878975, "learning_rate": 1.793204041108829e-05, "loss": 0.4915, "step": 6517 }, { "epoch": 0.23221761048862605, "grad_norm": 1.2030400338091367, "learning_rate": 1.7931337661662728e-05, "loss": 0.4703, "step": 6518 }, { "epoch": 0.23225323761512015, "grad_norm": 1.142836408555085, "learning_rate": 1.793063480662586e-05, "loss": 0.5023, "step": 6519 }, { "epoch": 0.23228886474161425, "grad_norm": 2.740435872326697, "learning_rate": 1.792993184598705e-05, "loss": 0.632, "step": 6520 }, { "epoch": 0.23232449186810838, "grad_norm": 1.9208691955871178, "learning_rate": 1.7929228779755652e-05, "loss": 0.6143, "step": 6521 }, { "epoch": 0.2323601189946025, "grad_norm": 1.9433449231509181, "learning_rate": 1.7928525607941033e-05, "loss": 0.6042, "step": 6522 }, { "epoch": 0.2323957461210966, "grad_norm": 1.1220234915741485, "learning_rate": 1.7927822330552553e-05, "loss": 0.4682, "step": 6523 }, { "epoch": 0.23243137324759072, "grad_norm": 1.3983740188635285, "learning_rate": 1.7927118947599576e-05, "loss": 0.5028, "step": 6524 }, { "epoch": 0.23246700037408483, "grad_norm": 1.9060322928831666, "learning_rate": 1.792641545909147e-05, "loss": 0.679, "step": 6525 }, { "epoch": 0.23250262750057893, "grad_norm": 1.2228163587549814, "learning_rate": 1.7925711865037602e-05, "loss": 0.4776, "step": 6526 }, { "epoch": 0.23253825462707306, "grad_norm": 1.518588215023902, "learning_rate": 1.792500816544734e-05, "loss": 0.7125, "step": 6527 }, { "epoch": 0.23257388175356716, "grad_norm": 1.6529181587724786, "learning_rate": 1.7924304360330057e-05, "loss": 0.6441, "step": 6528 }, { "epoch": 0.23260950888006127, "grad_norm": 2.232430643521548, "learning_rate": 1.7923600449695124e-05, "loss": 0.7345, "step": 6529 }, { "epoch": 0.2326451360065554, "grad_norm": 1.284815786047033, "learning_rate": 1.792289643355191e-05, "loss": 0.5179, "step": 6530 }, { "epoch": 0.2326807631330495, "grad_norm": 1.8828758595204922, "learning_rate": 1.792219231190979e-05, "loss": 0.5983, "step": 6531 }, { "epoch": 0.2327163902595436, "grad_norm": 1.2605180258854012, "learning_rate": 1.7921488084778147e-05, "loss": 0.5239, "step": 6532 }, { "epoch": 0.23275201738603774, "grad_norm": 1.1524084086210962, "learning_rate": 1.7920783752166348e-05, "loss": 0.5089, "step": 6533 }, { "epoch": 0.23278764451253184, "grad_norm": 1.513356275135751, "learning_rate": 1.792007931408378e-05, "loss": 0.535, "step": 6534 }, { "epoch": 0.23282327163902594, "grad_norm": 1.1572383436775162, "learning_rate": 1.791937477053982e-05, "loss": 0.5383, "step": 6535 }, { "epoch": 0.23285889876552007, "grad_norm": 1.6889363289803614, "learning_rate": 1.7918670121543854e-05, "loss": 0.6629, "step": 6536 }, { "epoch": 0.23289452589201418, "grad_norm": 1.126405388120553, "learning_rate": 1.7917965367105253e-05, "loss": 0.5345, "step": 6537 }, { "epoch": 0.23293015301850828, "grad_norm": 1.110355301332314, "learning_rate": 1.7917260507233415e-05, "loss": 0.5344, "step": 6538 }, { "epoch": 0.2329657801450024, "grad_norm": 1.4732200214945426, "learning_rate": 1.7916555541937715e-05, "loss": 0.5009, "step": 6539 }, { "epoch": 0.2330014072714965, "grad_norm": 1.9407716603772938, "learning_rate": 1.7915850471227548e-05, "loss": 0.6108, "step": 6540 }, { "epoch": 0.23303703439799062, "grad_norm": 1.21289484376644, "learning_rate": 1.7915145295112297e-05, "loss": 0.493, "step": 6541 }, { "epoch": 0.23307266152448475, "grad_norm": 1.5829814888525917, "learning_rate": 1.7914440013601355e-05, "loss": 0.5904, "step": 6542 }, { "epoch": 0.23310828865097885, "grad_norm": 1.385878237703932, "learning_rate": 1.791373462670411e-05, "loss": 0.5731, "step": 6543 }, { "epoch": 0.23314391577747295, "grad_norm": 1.7381656088600748, "learning_rate": 1.791302913442996e-05, "loss": 0.6276, "step": 6544 }, { "epoch": 0.23317954290396709, "grad_norm": 1.6133813059451996, "learning_rate": 1.7912323536788297e-05, "loss": 0.6211, "step": 6545 }, { "epoch": 0.2332151700304612, "grad_norm": 1.5002178856012767, "learning_rate": 1.7911617833788513e-05, "loss": 0.6213, "step": 6546 }, { "epoch": 0.23325079715695532, "grad_norm": 1.274552605674186, "learning_rate": 1.791091202544001e-05, "loss": 0.52, "step": 6547 }, { "epoch": 0.23328642428344942, "grad_norm": 1.9933923471327675, "learning_rate": 1.791020611175218e-05, "loss": 0.5844, "step": 6548 }, { "epoch": 0.23332205140994353, "grad_norm": 1.3116733172769317, "learning_rate": 1.790950009273443e-05, "loss": 0.5188, "step": 6549 }, { "epoch": 0.23335767853643766, "grad_norm": 2.1669590235353495, "learning_rate": 1.7908793968396158e-05, "loss": 0.5532, "step": 6550 }, { "epoch": 0.23339330566293176, "grad_norm": 2.8776796974755503, "learning_rate": 1.7908087738746764e-05, "loss": 0.5472, "step": 6551 }, { "epoch": 0.23342893278942586, "grad_norm": 1.2019525410725742, "learning_rate": 1.7907381403795655e-05, "loss": 0.5164, "step": 6552 }, { "epoch": 0.23346455991592, "grad_norm": 1.3418120412803858, "learning_rate": 1.7906674963552237e-05, "loss": 0.4821, "step": 6553 }, { "epoch": 0.2335001870424141, "grad_norm": 2.043875577454795, "learning_rate": 1.7905968418025915e-05, "loss": 0.6166, "step": 6554 }, { "epoch": 0.2335358141689082, "grad_norm": 1.7235192463168854, "learning_rate": 1.79052617672261e-05, "loss": 0.5688, "step": 6555 }, { "epoch": 0.23357144129540233, "grad_norm": 3.170836200163713, "learning_rate": 1.7904555011162195e-05, "loss": 0.6412, "step": 6556 }, { "epoch": 0.23360706842189644, "grad_norm": 1.3647157695861005, "learning_rate": 1.7903848149843616e-05, "loss": 0.5812, "step": 6557 }, { "epoch": 0.23364269554839054, "grad_norm": 1.7355934976198826, "learning_rate": 1.7903141183279776e-05, "loss": 0.596, "step": 6558 }, { "epoch": 0.23367832267488467, "grad_norm": 1.9340517162214277, "learning_rate": 1.7902434111480086e-05, "loss": 0.593, "step": 6559 }, { "epoch": 0.23371394980137877, "grad_norm": 1.8683799051378045, "learning_rate": 1.7901726934453964e-05, "loss": 0.6405, "step": 6560 }, { "epoch": 0.23374957692787288, "grad_norm": 1.6749790906243605, "learning_rate": 1.790101965221082e-05, "loss": 0.5462, "step": 6561 }, { "epoch": 0.233785204054367, "grad_norm": 1.4038935992631327, "learning_rate": 1.790031226476008e-05, "loss": 0.6013, "step": 6562 }, { "epoch": 0.2338208311808611, "grad_norm": 2.004098989038553, "learning_rate": 1.7899604772111163e-05, "loss": 0.7118, "step": 6563 }, { "epoch": 0.23385645830735521, "grad_norm": 1.383923332153407, "learning_rate": 1.7898897174273486e-05, "loss": 0.6039, "step": 6564 }, { "epoch": 0.23389208543384934, "grad_norm": 1.2356738911762708, "learning_rate": 1.7898189471256468e-05, "loss": 0.5648, "step": 6565 }, { "epoch": 0.23392771256034345, "grad_norm": 1.3505877352102365, "learning_rate": 1.789748166306954e-05, "loss": 0.4992, "step": 6566 }, { "epoch": 0.23396333968683755, "grad_norm": 1.3644052325381046, "learning_rate": 1.7896773749722123e-05, "loss": 0.5171, "step": 6567 }, { "epoch": 0.23399896681333168, "grad_norm": 1.8539425989255407, "learning_rate": 1.7896065731223645e-05, "loss": 0.6359, "step": 6568 }, { "epoch": 0.23403459393982579, "grad_norm": 1.226296043392905, "learning_rate": 1.7895357607583534e-05, "loss": 0.5608, "step": 6569 }, { "epoch": 0.2340702210663199, "grad_norm": 1.3346693971131627, "learning_rate": 1.7894649378811214e-05, "loss": 0.5182, "step": 6570 }, { "epoch": 0.23410584819281402, "grad_norm": 1.404496250509687, "learning_rate": 1.789394104491612e-05, "loss": 0.4557, "step": 6571 }, { "epoch": 0.23414147531930812, "grad_norm": 1.5447824911227326, "learning_rate": 1.7893232605907686e-05, "loss": 0.554, "step": 6572 }, { "epoch": 0.23417710244580223, "grad_norm": 1.5664600550165635, "learning_rate": 1.789252406179534e-05, "loss": 0.6164, "step": 6573 }, { "epoch": 0.23421272957229636, "grad_norm": 1.2064241697974067, "learning_rate": 1.7891815412588524e-05, "loss": 0.4755, "step": 6574 }, { "epoch": 0.23424835669879046, "grad_norm": 1.2568065016012355, "learning_rate": 1.7891106658296665e-05, "loss": 0.5221, "step": 6575 }, { "epoch": 0.23428398382528456, "grad_norm": 1.2645454686648112, "learning_rate": 1.7890397798929207e-05, "loss": 0.4625, "step": 6576 }, { "epoch": 0.2343196109517787, "grad_norm": 1.9068341272342868, "learning_rate": 1.7889688834495585e-05, "loss": 0.628, "step": 6577 }, { "epoch": 0.2343552380782728, "grad_norm": 1.2780593176690367, "learning_rate": 1.7888979765005247e-05, "loss": 0.5451, "step": 6578 }, { "epoch": 0.2343908652047669, "grad_norm": 1.2403059019955949, "learning_rate": 1.7888270590467627e-05, "loss": 0.5248, "step": 6579 }, { "epoch": 0.23442649233126103, "grad_norm": 1.3682592696844142, "learning_rate": 1.788756131089217e-05, "loss": 0.4982, "step": 6580 }, { "epoch": 0.23446211945775514, "grad_norm": 1.2071589850574937, "learning_rate": 1.7886851926288318e-05, "loss": 0.4773, "step": 6581 }, { "epoch": 0.23449774658424924, "grad_norm": 1.5948645598691378, "learning_rate": 1.7886142436665525e-05, "loss": 0.5656, "step": 6582 }, { "epoch": 0.23453337371074337, "grad_norm": 1.1412654548308412, "learning_rate": 1.788543284203323e-05, "loss": 0.5304, "step": 6583 }, { "epoch": 0.23456900083723747, "grad_norm": 1.235819394606969, "learning_rate": 1.7884723142400887e-05, "loss": 0.5321, "step": 6584 }, { "epoch": 0.23460462796373158, "grad_norm": 1.234106573559808, "learning_rate": 1.7884013337777944e-05, "loss": 0.5339, "step": 6585 }, { "epoch": 0.2346402550902257, "grad_norm": 1.306746053068761, "learning_rate": 1.7883303428173854e-05, "loss": 0.4512, "step": 6586 }, { "epoch": 0.2346758822167198, "grad_norm": 1.183777769129212, "learning_rate": 1.7882593413598065e-05, "loss": 0.4667, "step": 6587 }, { "epoch": 0.23471150934321391, "grad_norm": 1.2914075988073233, "learning_rate": 1.788188329406004e-05, "loss": 0.5195, "step": 6588 }, { "epoch": 0.23474713646970805, "grad_norm": 1.5404933729212094, "learning_rate": 1.7881173069569225e-05, "loss": 0.681, "step": 6589 }, { "epoch": 0.23478276359620215, "grad_norm": 1.6489955653460031, "learning_rate": 1.7880462740135084e-05, "loss": 0.5666, "step": 6590 }, { "epoch": 0.23481839072269625, "grad_norm": 1.432544597605759, "learning_rate": 1.7879752305767073e-05, "loss": 0.5098, "step": 6591 }, { "epoch": 0.23485401784919038, "grad_norm": 1.4780006880876813, "learning_rate": 1.7879041766474654e-05, "loss": 0.5684, "step": 6592 }, { "epoch": 0.2348896449756845, "grad_norm": 1.3281733675446399, "learning_rate": 1.7878331122267284e-05, "loss": 0.476, "step": 6593 }, { "epoch": 0.2349252721021786, "grad_norm": 2.0764058567121015, "learning_rate": 1.7877620373154433e-05, "loss": 0.5944, "step": 6594 }, { "epoch": 0.23496089922867272, "grad_norm": 1.149523709440821, "learning_rate": 1.7876909519145558e-05, "loss": 0.493, "step": 6595 }, { "epoch": 0.23499652635516682, "grad_norm": 1.101526200739301, "learning_rate": 1.7876198560250127e-05, "loss": 0.5085, "step": 6596 }, { "epoch": 0.23503215348166093, "grad_norm": 1.4519878840688054, "learning_rate": 1.7875487496477604e-05, "loss": 0.5907, "step": 6597 }, { "epoch": 0.23506778060815506, "grad_norm": 1.0422418792982622, "learning_rate": 1.7874776327837463e-05, "loss": 0.4375, "step": 6598 }, { "epoch": 0.23510340773464916, "grad_norm": 2.376698663558486, "learning_rate": 1.7874065054339173e-05, "loss": 0.5973, "step": 6599 }, { "epoch": 0.23513903486114326, "grad_norm": 1.661786316814539, "learning_rate": 1.7873353675992203e-05, "loss": 0.5783, "step": 6600 }, { "epoch": 0.2351746619876374, "grad_norm": 1.4979339964270673, "learning_rate": 1.787264219280602e-05, "loss": 0.5861, "step": 6601 }, { "epoch": 0.2352102891141315, "grad_norm": 1.1976811152859455, "learning_rate": 1.787193060479011e-05, "loss": 0.5423, "step": 6602 }, { "epoch": 0.2352459162406256, "grad_norm": 1.8415687702027135, "learning_rate": 1.7871218911953942e-05, "loss": 0.5725, "step": 6603 }, { "epoch": 0.23528154336711973, "grad_norm": 2.1760321768553745, "learning_rate": 1.7870507114306988e-05, "loss": 0.561, "step": 6604 }, { "epoch": 0.23531717049361384, "grad_norm": 1.337835479487816, "learning_rate": 1.7869795211858735e-05, "loss": 0.5271, "step": 6605 }, { "epoch": 0.23535279762010794, "grad_norm": 1.1986809501727562, "learning_rate": 1.7869083204618655e-05, "loss": 0.5142, "step": 6606 }, { "epoch": 0.23538842474660207, "grad_norm": 1.5200557743238523, "learning_rate": 1.7868371092596233e-05, "loss": 0.6121, "step": 6607 }, { "epoch": 0.23542405187309617, "grad_norm": 1.2497950801947648, "learning_rate": 1.786765887580095e-05, "loss": 0.5294, "step": 6608 }, { "epoch": 0.23545967899959028, "grad_norm": 1.2757421280846404, "learning_rate": 1.786694655424229e-05, "loss": 0.5036, "step": 6609 }, { "epoch": 0.2354953061260844, "grad_norm": 1.5029697593148397, "learning_rate": 1.786623412792974e-05, "loss": 0.6315, "step": 6610 }, { "epoch": 0.2355309332525785, "grad_norm": 1.9290270257016435, "learning_rate": 1.7865521596872782e-05, "loss": 0.5634, "step": 6611 }, { "epoch": 0.23556656037907261, "grad_norm": 1.1314213814394998, "learning_rate": 1.7864808961080906e-05, "loss": 0.5367, "step": 6612 }, { "epoch": 0.23560218750556675, "grad_norm": 5.212340503657331, "learning_rate": 1.78640962205636e-05, "loss": 0.5815, "step": 6613 }, { "epoch": 0.23563781463206085, "grad_norm": 1.1124804918943165, "learning_rate": 1.786338337533036e-05, "loss": 0.4972, "step": 6614 }, { "epoch": 0.23567344175855495, "grad_norm": 1.8367698096437555, "learning_rate": 1.7862670425390672e-05, "loss": 0.611, "step": 6615 }, { "epoch": 0.23570906888504908, "grad_norm": 1.2246550488034769, "learning_rate": 1.7861957370754034e-05, "loss": 0.5561, "step": 6616 }, { "epoch": 0.2357446960115432, "grad_norm": 1.1334436554211542, "learning_rate": 1.7861244211429935e-05, "loss": 0.512, "step": 6617 }, { "epoch": 0.2357803231380373, "grad_norm": 2.410975418234829, "learning_rate": 1.7860530947427878e-05, "loss": 0.6617, "step": 6618 }, { "epoch": 0.23581595026453142, "grad_norm": 2.002737432214344, "learning_rate": 1.7859817578757353e-05, "loss": 0.6531, "step": 6619 }, { "epoch": 0.23585157739102552, "grad_norm": 2.1218740501807836, "learning_rate": 1.7859104105427867e-05, "loss": 0.6688, "step": 6620 }, { "epoch": 0.23588720451751963, "grad_norm": 1.87939120873316, "learning_rate": 1.7858390527448915e-05, "loss": 0.6399, "step": 6621 }, { "epoch": 0.23592283164401376, "grad_norm": 1.4135174556260692, "learning_rate": 1.7857676844830002e-05, "loss": 0.618, "step": 6622 }, { "epoch": 0.23595845877050786, "grad_norm": 1.8499760865889339, "learning_rate": 1.7856963057580626e-05, "loss": 0.5417, "step": 6623 }, { "epoch": 0.23599408589700197, "grad_norm": 1.142815963018559, "learning_rate": 1.7856249165710298e-05, "loss": 0.4725, "step": 6624 }, { "epoch": 0.2360297130234961, "grad_norm": 1.8621688298786696, "learning_rate": 1.785553516922852e-05, "loss": 0.6317, "step": 6625 }, { "epoch": 0.2360653401499902, "grad_norm": 2.1895644112363732, "learning_rate": 1.78548210681448e-05, "loss": 0.619, "step": 6626 }, { "epoch": 0.2361009672764843, "grad_norm": 2.3842881180411806, "learning_rate": 1.785410686246865e-05, "loss": 0.6214, "step": 6627 }, { "epoch": 0.23613659440297843, "grad_norm": 1.1198998277855992, "learning_rate": 1.785339255220958e-05, "loss": 0.4746, "step": 6628 }, { "epoch": 0.23617222152947254, "grad_norm": 1.592695459941807, "learning_rate": 1.785267813737709e-05, "loss": 0.5253, "step": 6629 }, { "epoch": 0.23620784865596664, "grad_norm": 1.5837812152236919, "learning_rate": 1.7851963617980707e-05, "loss": 0.6267, "step": 6630 }, { "epoch": 0.23624347578246077, "grad_norm": 1.7650594598859113, "learning_rate": 1.7851248994029943e-05, "loss": 0.6021, "step": 6631 }, { "epoch": 0.23627910290895487, "grad_norm": 2.120396539759443, "learning_rate": 1.785053426553431e-05, "loss": 0.63, "step": 6632 }, { "epoch": 0.23631473003544898, "grad_norm": 1.5154968411802954, "learning_rate": 1.7849819432503322e-05, "loss": 0.5925, "step": 6633 }, { "epoch": 0.2363503571619431, "grad_norm": 1.1232440822825955, "learning_rate": 1.7849104494946504e-05, "loss": 0.4958, "step": 6634 }, { "epoch": 0.2363859842884372, "grad_norm": 1.6177109543112165, "learning_rate": 1.7848389452873372e-05, "loss": 0.653, "step": 6635 }, { "epoch": 0.23642161141493132, "grad_norm": 1.2107848131269887, "learning_rate": 1.7847674306293454e-05, "loss": 0.4962, "step": 6636 }, { "epoch": 0.23645723854142545, "grad_norm": 1.1683192302067, "learning_rate": 1.784695905521626e-05, "loss": 0.5239, "step": 6637 }, { "epoch": 0.23649286566791955, "grad_norm": 2.07562155650387, "learning_rate": 1.7846243699651325e-05, "loss": 0.5974, "step": 6638 }, { "epoch": 0.23652849279441365, "grad_norm": 1.1919561428260608, "learning_rate": 1.7845528239608172e-05, "loss": 0.5415, "step": 6639 }, { "epoch": 0.23656411992090778, "grad_norm": 1.8272524336622666, "learning_rate": 1.7844812675096323e-05, "loss": 0.5997, "step": 6640 }, { "epoch": 0.2365997470474019, "grad_norm": 1.4020138556829944, "learning_rate": 1.7844097006125315e-05, "loss": 0.6171, "step": 6641 }, { "epoch": 0.23663537417389602, "grad_norm": 1.3568343561043354, "learning_rate": 1.7843381232704668e-05, "loss": 0.5306, "step": 6642 }, { "epoch": 0.23667100130039012, "grad_norm": 1.594883787306674, "learning_rate": 1.7842665354843923e-05, "loss": 0.6175, "step": 6643 }, { "epoch": 0.23670662842688422, "grad_norm": 3.243253347300732, "learning_rate": 1.7841949372552602e-05, "loss": 0.548, "step": 6644 }, { "epoch": 0.23674225555337836, "grad_norm": 1.3768994470538507, "learning_rate": 1.7841233285840248e-05, "loss": 0.5496, "step": 6645 }, { "epoch": 0.23677788267987246, "grad_norm": 2.4224730357709, "learning_rate": 1.7840517094716386e-05, "loss": 0.6108, "step": 6646 }, { "epoch": 0.23681350980636656, "grad_norm": 1.1422643018461813, "learning_rate": 1.7839800799190566e-05, "loss": 0.511, "step": 6647 }, { "epoch": 0.2368491369328607, "grad_norm": 1.0114389147252179, "learning_rate": 1.7839084399272317e-05, "loss": 0.4689, "step": 6648 }, { "epoch": 0.2368847640593548, "grad_norm": 1.3717392939482012, "learning_rate": 1.7838367894971175e-05, "loss": 0.5706, "step": 6649 }, { "epoch": 0.2369203911858489, "grad_norm": 1.9973042062673882, "learning_rate": 1.7837651286296692e-05, "loss": 0.6536, "step": 6650 }, { "epoch": 0.23695601831234303, "grad_norm": 1.171975410690839, "learning_rate": 1.78369345732584e-05, "loss": 0.5216, "step": 6651 }, { "epoch": 0.23699164543883713, "grad_norm": 1.9148681419371274, "learning_rate": 1.7836217755865846e-05, "loss": 0.614, "step": 6652 }, { "epoch": 0.23702727256533124, "grad_norm": 1.1065668072034907, "learning_rate": 1.783550083412858e-05, "loss": 0.4797, "step": 6653 }, { "epoch": 0.23706289969182537, "grad_norm": 1.4749653532513385, "learning_rate": 1.783478380805614e-05, "loss": 0.5764, "step": 6654 }, { "epoch": 0.23709852681831947, "grad_norm": 1.2501910668290723, "learning_rate": 1.7834066677658077e-05, "loss": 0.5105, "step": 6655 }, { "epoch": 0.23713415394481357, "grad_norm": 1.145385370495983, "learning_rate": 1.783334944294394e-05, "loss": 0.5278, "step": 6656 }, { "epoch": 0.2371697810713077, "grad_norm": 1.134970212460702, "learning_rate": 1.7832632103923286e-05, "loss": 0.4979, "step": 6657 }, { "epoch": 0.2372054081978018, "grad_norm": 1.0647289770210642, "learning_rate": 1.7831914660605654e-05, "loss": 0.4914, "step": 6658 }, { "epoch": 0.2372410353242959, "grad_norm": 1.380642325277606, "learning_rate": 1.7831197113000607e-05, "loss": 0.6532, "step": 6659 }, { "epoch": 0.23727666245079004, "grad_norm": 2.0863153320378722, "learning_rate": 1.7830479461117697e-05, "loss": 0.5633, "step": 6660 }, { "epoch": 0.23731228957728415, "grad_norm": 1.8764060662760533, "learning_rate": 1.7829761704966477e-05, "loss": 0.5958, "step": 6661 }, { "epoch": 0.23734791670377825, "grad_norm": 1.32258427554934, "learning_rate": 1.782904384455651e-05, "loss": 0.5371, "step": 6662 }, { "epoch": 0.23738354383027238, "grad_norm": 2.413391654611304, "learning_rate": 1.7828325879897353e-05, "loss": 0.7205, "step": 6663 }, { "epoch": 0.23741917095676648, "grad_norm": 1.196979299698252, "learning_rate": 1.782760781099856e-05, "loss": 0.5029, "step": 6664 }, { "epoch": 0.2374547980832606, "grad_norm": 1.710204842748091, "learning_rate": 1.7826889637869705e-05, "loss": 0.5856, "step": 6665 }, { "epoch": 0.23749042520975472, "grad_norm": 2.278598759978109, "learning_rate": 1.782617136052034e-05, "loss": 0.6061, "step": 6666 }, { "epoch": 0.23752605233624882, "grad_norm": 2.7820470114779043, "learning_rate": 1.7825452978960032e-05, "loss": 0.5667, "step": 6667 }, { "epoch": 0.23756167946274293, "grad_norm": 1.823506171781934, "learning_rate": 1.782473449319835e-05, "loss": 0.6055, "step": 6668 }, { "epoch": 0.23759730658923706, "grad_norm": 1.746459624683545, "learning_rate": 1.7824015903244857e-05, "loss": 0.6925, "step": 6669 }, { "epoch": 0.23763293371573116, "grad_norm": 1.1139305425336619, "learning_rate": 1.782329720910913e-05, "loss": 0.5211, "step": 6670 }, { "epoch": 0.23766856084222526, "grad_norm": 1.5185957716206102, "learning_rate": 1.7822578410800724e-05, "loss": 0.6005, "step": 6671 }, { "epoch": 0.2377041879687194, "grad_norm": 2.867129662457285, "learning_rate": 1.7821859508329223e-05, "loss": 0.6975, "step": 6672 }, { "epoch": 0.2377398150952135, "grad_norm": 1.1988371356175045, "learning_rate": 1.7821140501704195e-05, "loss": 0.5143, "step": 6673 }, { "epoch": 0.2377754422217076, "grad_norm": 1.5381712405870898, "learning_rate": 1.7820421390935213e-05, "loss": 0.564, "step": 6674 }, { "epoch": 0.23781106934820173, "grad_norm": 1.7360757431382765, "learning_rate": 1.7819702176031858e-05, "loss": 0.6039, "step": 6675 }, { "epoch": 0.23784669647469583, "grad_norm": 1.1342187338903946, "learning_rate": 1.7818982857003698e-05, "loss": 0.5426, "step": 6676 }, { "epoch": 0.23788232360118994, "grad_norm": 1.284086788905315, "learning_rate": 1.781826343386032e-05, "loss": 0.5148, "step": 6677 }, { "epoch": 0.23791795072768407, "grad_norm": 1.1282657991916332, "learning_rate": 1.7817543906611298e-05, "loss": 0.5016, "step": 6678 }, { "epoch": 0.23795357785417817, "grad_norm": 2.01288553369685, "learning_rate": 1.7816824275266213e-05, "loss": 0.627, "step": 6679 }, { "epoch": 0.23798920498067228, "grad_norm": 1.1519500039522166, "learning_rate": 1.7816104539834654e-05, "loss": 0.4812, "step": 6680 }, { "epoch": 0.2380248321071664, "grad_norm": 1.7659933904105007, "learning_rate": 1.7815384700326195e-05, "loss": 0.4914, "step": 6681 }, { "epoch": 0.2380604592336605, "grad_norm": 1.5516611088899492, "learning_rate": 1.781466475675043e-05, "loss": 0.6237, "step": 6682 }, { "epoch": 0.2380960863601546, "grad_norm": 2.981498917023366, "learning_rate": 1.7813944709116938e-05, "loss": 0.5566, "step": 6683 }, { "epoch": 0.23813171348664874, "grad_norm": 2.5906198487671683, "learning_rate": 1.7813224557435313e-05, "loss": 0.58, "step": 6684 }, { "epoch": 0.23816734061314285, "grad_norm": 1.7423122055866607, "learning_rate": 1.7812504301715142e-05, "loss": 0.5617, "step": 6685 }, { "epoch": 0.23820296773963695, "grad_norm": 1.2392612729096186, "learning_rate": 1.7811783941966013e-05, "loss": 0.5142, "step": 6686 }, { "epoch": 0.23823859486613108, "grad_norm": 1.7539140390244063, "learning_rate": 1.781106347819752e-05, "loss": 0.6166, "step": 6687 }, { "epoch": 0.23827422199262518, "grad_norm": 2.0267821316518235, "learning_rate": 1.781034291041926e-05, "loss": 0.6946, "step": 6688 }, { "epoch": 0.2383098491191193, "grad_norm": 1.8286036320686962, "learning_rate": 1.7809622238640826e-05, "loss": 0.6467, "step": 6689 }, { "epoch": 0.23834547624561342, "grad_norm": 1.5376047863097975, "learning_rate": 1.7808901462871814e-05, "loss": 0.5758, "step": 6690 }, { "epoch": 0.23838110337210752, "grad_norm": 1.6319016573444185, "learning_rate": 1.780818058312182e-05, "loss": 0.6542, "step": 6691 }, { "epoch": 0.23841673049860163, "grad_norm": 1.912949883537687, "learning_rate": 1.7807459599400442e-05, "loss": 0.5694, "step": 6692 }, { "epoch": 0.23845235762509576, "grad_norm": 1.1564442294008552, "learning_rate": 1.780673851171728e-05, "loss": 0.5558, "step": 6693 }, { "epoch": 0.23848798475158986, "grad_norm": 1.384552553412936, "learning_rate": 1.780601732008194e-05, "loss": 0.5025, "step": 6694 }, { "epoch": 0.23852361187808396, "grad_norm": 2.069567257409648, "learning_rate": 1.7805296024504026e-05, "loss": 0.6331, "step": 6695 }, { "epoch": 0.2385592390045781, "grad_norm": 1.2966762189613914, "learning_rate": 1.780457462499314e-05, "loss": 0.5131, "step": 6696 }, { "epoch": 0.2385948661310722, "grad_norm": 1.3002970201328832, "learning_rate": 1.7803853121558886e-05, "loss": 0.5499, "step": 6697 }, { "epoch": 0.2386304932575663, "grad_norm": 1.9311476065485966, "learning_rate": 1.7803131514210874e-05, "loss": 0.533, "step": 6698 }, { "epoch": 0.23866612038406043, "grad_norm": 1.3478844486356576, "learning_rate": 1.780240980295871e-05, "loss": 0.5296, "step": 6699 }, { "epoch": 0.23870174751055453, "grad_norm": 1.3486333000174777, "learning_rate": 1.780168798781201e-05, "loss": 0.4952, "step": 6700 }, { "epoch": 0.23873737463704864, "grad_norm": 1.4617687444054985, "learning_rate": 1.7800966068780378e-05, "loss": 0.5854, "step": 6701 }, { "epoch": 0.23877300176354277, "grad_norm": 1.1077438803432293, "learning_rate": 1.780024404587343e-05, "loss": 0.4764, "step": 6702 }, { "epoch": 0.23880862889003687, "grad_norm": 1.2152640375942119, "learning_rate": 1.779952191910078e-05, "loss": 0.5378, "step": 6703 }, { "epoch": 0.23884425601653098, "grad_norm": 1.3057639734213151, "learning_rate": 1.7798799688472046e-05, "loss": 0.5254, "step": 6704 }, { "epoch": 0.2388798831430251, "grad_norm": 1.23423283351918, "learning_rate": 1.7798077353996843e-05, "loss": 0.5548, "step": 6705 }, { "epoch": 0.2389155102695192, "grad_norm": 1.1636337541264383, "learning_rate": 1.779735491568479e-05, "loss": 0.5184, "step": 6706 }, { "epoch": 0.2389511373960133, "grad_norm": 2.061460518097042, "learning_rate": 1.7796632373545507e-05, "loss": 0.6121, "step": 6707 }, { "epoch": 0.23898676452250744, "grad_norm": 1.8955691002244686, "learning_rate": 1.7795909727588608e-05, "loss": 0.6676, "step": 6708 }, { "epoch": 0.23902239164900155, "grad_norm": 1.185105585693456, "learning_rate": 1.779518697782373e-05, "loss": 0.4972, "step": 6709 }, { "epoch": 0.23905801877549565, "grad_norm": 1.188356391413639, "learning_rate": 1.7794464124260484e-05, "loss": 0.4623, "step": 6710 }, { "epoch": 0.23909364590198978, "grad_norm": 1.2579740809102649, "learning_rate": 1.7793741166908504e-05, "loss": 0.516, "step": 6711 }, { "epoch": 0.23912927302848389, "grad_norm": 1.8462486670481193, "learning_rate": 1.779301810577741e-05, "loss": 0.6768, "step": 6712 }, { "epoch": 0.239164900154978, "grad_norm": 1.2083313663257178, "learning_rate": 1.7792294940876835e-05, "loss": 0.5375, "step": 6713 }, { "epoch": 0.23920052728147212, "grad_norm": 2.184736999755791, "learning_rate": 1.7791571672216404e-05, "loss": 0.58, "step": 6714 }, { "epoch": 0.23923615440796622, "grad_norm": 1.1122925094590965, "learning_rate": 1.779084829980575e-05, "loss": 0.4628, "step": 6715 }, { "epoch": 0.23927178153446033, "grad_norm": 2.3204309499178577, "learning_rate": 1.7790124823654506e-05, "loss": 0.5523, "step": 6716 }, { "epoch": 0.23930740866095446, "grad_norm": 2.6301498589474646, "learning_rate": 1.7789401243772307e-05, "loss": 0.7009, "step": 6717 }, { "epoch": 0.23934303578744856, "grad_norm": 2.0887628834604146, "learning_rate": 1.7788677560168782e-05, "loss": 0.5976, "step": 6718 }, { "epoch": 0.23937866291394266, "grad_norm": 1.744762900690796, "learning_rate": 1.7787953772853574e-05, "loss": 0.5595, "step": 6719 }, { "epoch": 0.2394142900404368, "grad_norm": 1.950075494739126, "learning_rate": 1.778722988183632e-05, "loss": 0.6192, "step": 6720 }, { "epoch": 0.2394499171669309, "grad_norm": 2.6580411118390903, "learning_rate": 1.7786505887126652e-05, "loss": 0.6411, "step": 6721 }, { "epoch": 0.239485544293425, "grad_norm": 1.5750590757808842, "learning_rate": 1.778578178873422e-05, "loss": 0.6679, "step": 6722 }, { "epoch": 0.23952117141991913, "grad_norm": 1.245261693187419, "learning_rate": 1.778505758666866e-05, "loss": 0.5061, "step": 6723 }, { "epoch": 0.23955679854641324, "grad_norm": 1.3383830929220473, "learning_rate": 1.7784333280939618e-05, "loss": 0.5494, "step": 6724 }, { "epoch": 0.23959242567290734, "grad_norm": 1.9361763726333785, "learning_rate": 1.7783608871556735e-05, "loss": 0.6523, "step": 6725 }, { "epoch": 0.23962805279940147, "grad_norm": 1.0490683895551216, "learning_rate": 1.778288435852966e-05, "loss": 0.4787, "step": 6726 }, { "epoch": 0.23966367992589557, "grad_norm": 1.2764864793675372, "learning_rate": 1.7782159741868042e-05, "loss": 0.5094, "step": 6727 }, { "epoch": 0.23969930705238968, "grad_norm": 1.492522672325354, "learning_rate": 1.7781435021581527e-05, "loss": 0.5645, "step": 6728 }, { "epoch": 0.2397349341788838, "grad_norm": 1.2202382422796663, "learning_rate": 1.7780710197679765e-05, "loss": 0.5028, "step": 6729 }, { "epoch": 0.2397705613053779, "grad_norm": 1.8779208580966469, "learning_rate": 1.777998527017241e-05, "loss": 0.5472, "step": 6730 }, { "epoch": 0.239806188431872, "grad_norm": 1.2837013871196454, "learning_rate": 1.777926023906911e-05, "loss": 0.5146, "step": 6731 }, { "epoch": 0.23984181555836614, "grad_norm": 1.1929363787888174, "learning_rate": 1.7778535104379523e-05, "loss": 0.5057, "step": 6732 }, { "epoch": 0.23987744268486025, "grad_norm": 1.1219648584288702, "learning_rate": 1.777780986611331e-05, "loss": 0.4539, "step": 6733 }, { "epoch": 0.23991306981135435, "grad_norm": 1.183339387287958, "learning_rate": 1.777708452428012e-05, "loss": 0.4874, "step": 6734 }, { "epoch": 0.23994869693784848, "grad_norm": 1.2004557102866018, "learning_rate": 1.7776359078889605e-05, "loss": 0.4693, "step": 6735 }, { "epoch": 0.23998432406434259, "grad_norm": 1.1408525111374652, "learning_rate": 1.7775633529951442e-05, "loss": 0.5143, "step": 6736 }, { "epoch": 0.24001995119083672, "grad_norm": 2.0523900752290603, "learning_rate": 1.777490787747528e-05, "loss": 0.5763, "step": 6737 }, { "epoch": 0.24005557831733082, "grad_norm": 1.1873809052804865, "learning_rate": 1.777418212147079e-05, "loss": 0.5108, "step": 6738 }, { "epoch": 0.24009120544382492, "grad_norm": 1.1430262085680094, "learning_rate": 1.7773456261947627e-05, "loss": 0.535, "step": 6739 }, { "epoch": 0.24012683257031905, "grad_norm": 1.1660382648452028, "learning_rate": 1.7772730298915463e-05, "loss": 0.4873, "step": 6740 }, { "epoch": 0.24016245969681316, "grad_norm": 1.732810526201022, "learning_rate": 1.7772004232383963e-05, "loss": 0.5873, "step": 6741 }, { "epoch": 0.24019808682330726, "grad_norm": 1.0959441981135913, "learning_rate": 1.7771278062362787e-05, "loss": 0.4921, "step": 6742 }, { "epoch": 0.2402337139498014, "grad_norm": 1.76987281065881, "learning_rate": 1.777055178886162e-05, "loss": 0.6627, "step": 6743 }, { "epoch": 0.2402693410762955, "grad_norm": 1.7621292578569718, "learning_rate": 1.7769825411890122e-05, "loss": 0.5493, "step": 6744 }, { "epoch": 0.2403049682027896, "grad_norm": 1.56180463889938, "learning_rate": 1.7769098931457968e-05, "loss": 0.5961, "step": 6745 }, { "epoch": 0.24034059532928373, "grad_norm": 1.25388255897559, "learning_rate": 1.7768372347574828e-05, "loss": 0.4868, "step": 6746 }, { "epoch": 0.24037622245577783, "grad_norm": 1.3141618158271708, "learning_rate": 1.7767645660250387e-05, "loss": 0.5403, "step": 6747 }, { "epoch": 0.24041184958227194, "grad_norm": 1.2575573425504925, "learning_rate": 1.776691886949431e-05, "loss": 0.5492, "step": 6748 }, { "epoch": 0.24044747670876607, "grad_norm": 2.564792165970358, "learning_rate": 1.7766191975316282e-05, "loss": 0.6195, "step": 6749 }, { "epoch": 0.24048310383526017, "grad_norm": 1.703203682261874, "learning_rate": 1.7765464977725977e-05, "loss": 0.5384, "step": 6750 }, { "epoch": 0.24051873096175427, "grad_norm": 1.871966012592625, "learning_rate": 1.776473787673308e-05, "loss": 0.6194, "step": 6751 }, { "epoch": 0.2405543580882484, "grad_norm": 1.2274852811607742, "learning_rate": 1.776401067234727e-05, "loss": 0.5173, "step": 6752 }, { "epoch": 0.2405899852147425, "grad_norm": 1.063697231518187, "learning_rate": 1.7763283364578234e-05, "loss": 0.4593, "step": 6753 }, { "epoch": 0.2406256123412366, "grad_norm": 1.2720191188333898, "learning_rate": 1.7762555953435653e-05, "loss": 0.5391, "step": 6754 }, { "epoch": 0.24066123946773074, "grad_norm": 1.1583001145036786, "learning_rate": 1.776182843892921e-05, "loss": 0.5069, "step": 6755 }, { "epoch": 0.24069686659422485, "grad_norm": 1.9270635899281177, "learning_rate": 1.7761100821068597e-05, "loss": 0.7394, "step": 6756 }, { "epoch": 0.24073249372071895, "grad_norm": 1.9230209388521968, "learning_rate": 1.7760373099863504e-05, "loss": 0.6158, "step": 6757 }, { "epoch": 0.24076812084721308, "grad_norm": 1.8426064681355965, "learning_rate": 1.775964527532362e-05, "loss": 0.5743, "step": 6758 }, { "epoch": 0.24080374797370718, "grad_norm": 2.1188597961050726, "learning_rate": 1.7758917347458634e-05, "loss": 0.6914, "step": 6759 }, { "epoch": 0.2408393751002013, "grad_norm": 1.6211658819226806, "learning_rate": 1.775818931627824e-05, "loss": 0.6723, "step": 6760 }, { "epoch": 0.24087500222669542, "grad_norm": 1.3038563239370198, "learning_rate": 1.775746118179213e-05, "loss": 0.4578, "step": 6761 }, { "epoch": 0.24091062935318952, "grad_norm": 1.1501636119654908, "learning_rate": 1.7756732944010007e-05, "loss": 0.4459, "step": 6762 }, { "epoch": 0.24094625647968362, "grad_norm": 1.2792510007808173, "learning_rate": 1.775600460294156e-05, "loss": 0.5084, "step": 6763 }, { "epoch": 0.24098188360617775, "grad_norm": 1.124275533955239, "learning_rate": 1.775527615859649e-05, "loss": 0.4596, "step": 6764 }, { "epoch": 0.24101751073267186, "grad_norm": 1.2680770793410263, "learning_rate": 1.7754547610984502e-05, "loss": 0.4745, "step": 6765 }, { "epoch": 0.24105313785916596, "grad_norm": 2.2045709887286296, "learning_rate": 1.775381896011529e-05, "loss": 0.5778, "step": 6766 }, { "epoch": 0.2410887649856601, "grad_norm": 2.1988452641289515, "learning_rate": 1.775309020599856e-05, "loss": 0.5767, "step": 6767 }, { "epoch": 0.2411243921121542, "grad_norm": 2.4904356570155826, "learning_rate": 1.7752361348644012e-05, "loss": 0.5674, "step": 6768 }, { "epoch": 0.2411600192386483, "grad_norm": 1.0960895074524515, "learning_rate": 1.7751632388061354e-05, "loss": 0.4168, "step": 6769 }, { "epoch": 0.24119564636514243, "grad_norm": 1.5399248952367868, "learning_rate": 1.7750903324260294e-05, "loss": 0.6091, "step": 6770 }, { "epoch": 0.24123127349163653, "grad_norm": 1.7261728506078686, "learning_rate": 1.775017415725054e-05, "loss": 0.7004, "step": 6771 }, { "epoch": 0.24126690061813064, "grad_norm": 1.2926224297319198, "learning_rate": 1.7749444887041797e-05, "loss": 0.4515, "step": 6772 }, { "epoch": 0.24130252774462477, "grad_norm": 2.0211230394664352, "learning_rate": 1.774871551364378e-05, "loss": 0.6752, "step": 6773 }, { "epoch": 0.24133815487111887, "grad_norm": 1.7993633300138039, "learning_rate": 1.77479860370662e-05, "loss": 0.6505, "step": 6774 }, { "epoch": 0.24137378199761297, "grad_norm": 1.230945643297053, "learning_rate": 1.7747256457318772e-05, "loss": 0.5811, "step": 6775 }, { "epoch": 0.2414094091241071, "grad_norm": 2.6326599190489874, "learning_rate": 1.7746526774411207e-05, "loss": 0.5822, "step": 6776 }, { "epoch": 0.2414450362506012, "grad_norm": 1.8005499836233745, "learning_rate": 1.7745796988353226e-05, "loss": 0.5716, "step": 6777 }, { "epoch": 0.2414806633770953, "grad_norm": 1.1120106478846195, "learning_rate": 1.774506709915454e-05, "loss": 0.4757, "step": 6778 }, { "epoch": 0.24151629050358944, "grad_norm": 2.3658984335917874, "learning_rate": 1.7744337106824878e-05, "loss": 0.6929, "step": 6779 }, { "epoch": 0.24155191763008355, "grad_norm": 1.212522661393305, "learning_rate": 1.774360701137395e-05, "loss": 0.4903, "step": 6780 }, { "epoch": 0.24158754475657765, "grad_norm": 1.70988299794468, "learning_rate": 1.7742876812811485e-05, "loss": 0.6101, "step": 6781 }, { "epoch": 0.24162317188307178, "grad_norm": 1.194415809897193, "learning_rate": 1.7742146511147202e-05, "loss": 0.5492, "step": 6782 }, { "epoch": 0.24165879900956588, "grad_norm": 1.1391981814869827, "learning_rate": 1.7741416106390828e-05, "loss": 0.4722, "step": 6783 }, { "epoch": 0.24169442613606, "grad_norm": 1.2705340003241756, "learning_rate": 1.774068559855208e-05, "loss": 0.5316, "step": 6784 }, { "epoch": 0.24173005326255412, "grad_norm": 2.1714445687972126, "learning_rate": 1.7739954987640705e-05, "loss": 0.6126, "step": 6785 }, { "epoch": 0.24176568038904822, "grad_norm": 1.5706275505062228, "learning_rate": 1.773922427366641e-05, "loss": 0.5728, "step": 6786 }, { "epoch": 0.24180130751554232, "grad_norm": 1.2742156267308882, "learning_rate": 1.773849345663894e-05, "loss": 0.5384, "step": 6787 }, { "epoch": 0.24183693464203646, "grad_norm": 1.8327461508865077, "learning_rate": 1.7737762536568017e-05, "loss": 0.6628, "step": 6788 }, { "epoch": 0.24187256176853056, "grad_norm": 1.0837293562927302, "learning_rate": 1.773703151346338e-05, "loss": 0.5451, "step": 6789 }, { "epoch": 0.24190818889502466, "grad_norm": 1.145028018004542, "learning_rate": 1.7736300387334758e-05, "loss": 0.4895, "step": 6790 }, { "epoch": 0.2419438160215188, "grad_norm": 1.8140194967413854, "learning_rate": 1.7735569158191892e-05, "loss": 0.6293, "step": 6791 }, { "epoch": 0.2419794431480129, "grad_norm": 1.0529892785938775, "learning_rate": 1.7734837826044514e-05, "loss": 0.4563, "step": 6792 }, { "epoch": 0.242015070274507, "grad_norm": 2.0200677045813276, "learning_rate": 1.7734106390902364e-05, "loss": 0.6755, "step": 6793 }, { "epoch": 0.24205069740100113, "grad_norm": 1.7167692305233446, "learning_rate": 1.7733374852775184e-05, "loss": 0.5471, "step": 6794 }, { "epoch": 0.24208632452749523, "grad_norm": 1.1102760479948592, "learning_rate": 1.7732643211672708e-05, "loss": 0.4852, "step": 6795 }, { "epoch": 0.24212195165398934, "grad_norm": 1.0465041489813953, "learning_rate": 1.7731911467604686e-05, "loss": 0.5331, "step": 6796 }, { "epoch": 0.24215757878048347, "grad_norm": 1.6644655311423293, "learning_rate": 1.7731179620580857e-05, "loss": 0.6119, "step": 6797 }, { "epoch": 0.24219320590697757, "grad_norm": 1.17386045391902, "learning_rate": 1.773044767061097e-05, "loss": 0.5528, "step": 6798 }, { "epoch": 0.24222883303347167, "grad_norm": 1.0854202159537756, "learning_rate": 1.772971561770477e-05, "loss": 0.47, "step": 6799 }, { "epoch": 0.2422644601599658, "grad_norm": 1.0631157697423328, "learning_rate": 1.7728983461872002e-05, "loss": 0.5177, "step": 6800 }, { "epoch": 0.2423000872864599, "grad_norm": 1.53321062426669, "learning_rate": 1.7728251203122414e-05, "loss": 0.6241, "step": 6801 }, { "epoch": 0.242335714412954, "grad_norm": 1.8902566278448567, "learning_rate": 1.772751884146576e-05, "loss": 0.6533, "step": 6802 }, { "epoch": 0.24237134153944814, "grad_norm": 1.1740752879937038, "learning_rate": 1.7726786376911795e-05, "loss": 0.4834, "step": 6803 }, { "epoch": 0.24240696866594225, "grad_norm": 1.5223883405634318, "learning_rate": 1.7726053809470266e-05, "loss": 0.6263, "step": 6804 }, { "epoch": 0.24244259579243635, "grad_norm": 4.129449409042315, "learning_rate": 1.772532113915093e-05, "loss": 0.6343, "step": 6805 }, { "epoch": 0.24247822291893048, "grad_norm": 1.16564705198728, "learning_rate": 1.7724588365963546e-05, "loss": 0.5248, "step": 6806 }, { "epoch": 0.24251385004542458, "grad_norm": 1.4344926932303472, "learning_rate": 1.7723855489917868e-05, "loss": 0.455, "step": 6807 }, { "epoch": 0.2425494771719187, "grad_norm": 1.164403530253363, "learning_rate": 1.7723122511023654e-05, "loss": 0.576, "step": 6808 }, { "epoch": 0.24258510429841282, "grad_norm": 1.1577493646588328, "learning_rate": 1.7722389429290668e-05, "loss": 0.5134, "step": 6809 }, { "epoch": 0.24262073142490692, "grad_norm": 1.9261344837425318, "learning_rate": 1.7721656244728665e-05, "loss": 0.6176, "step": 6810 }, { "epoch": 0.24265635855140102, "grad_norm": 1.2007070343697532, "learning_rate": 1.7720922957347417e-05, "loss": 0.4925, "step": 6811 }, { "epoch": 0.24269198567789516, "grad_norm": 1.6802538853347315, "learning_rate": 1.772018956715668e-05, "loss": 0.5796, "step": 6812 }, { "epoch": 0.24272761280438926, "grad_norm": 1.6681642872759594, "learning_rate": 1.771945607416622e-05, "loss": 0.7039, "step": 6813 }, { "epoch": 0.24276323993088336, "grad_norm": 1.7614443641714213, "learning_rate": 1.7718722478385815e-05, "loss": 0.6403, "step": 6814 }, { "epoch": 0.2427988670573775, "grad_norm": 1.2022604549023312, "learning_rate": 1.771798877982522e-05, "loss": 0.5204, "step": 6815 }, { "epoch": 0.2428344941838716, "grad_norm": 1.7496184049795, "learning_rate": 1.771725497849421e-05, "loss": 0.5814, "step": 6816 }, { "epoch": 0.2428701213103657, "grad_norm": 1.2794553996388247, "learning_rate": 1.7716521074402554e-05, "loss": 0.5201, "step": 6817 }, { "epoch": 0.24290574843685983, "grad_norm": 1.2721610786199478, "learning_rate": 1.771578706756003e-05, "loss": 0.5308, "step": 6818 }, { "epoch": 0.24294137556335393, "grad_norm": 1.1919725737308844, "learning_rate": 1.7715052957976406e-05, "loss": 0.5352, "step": 6819 }, { "epoch": 0.24297700268984804, "grad_norm": 1.1953616471420656, "learning_rate": 1.771431874566146e-05, "loss": 0.4555, "step": 6820 }, { "epoch": 0.24301262981634217, "grad_norm": 1.3093587078938886, "learning_rate": 1.771358443062497e-05, "loss": 0.4549, "step": 6821 }, { "epoch": 0.24304825694283627, "grad_norm": 1.1555011381841551, "learning_rate": 1.7712850012876708e-05, "loss": 0.5027, "step": 6822 }, { "epoch": 0.24308388406933037, "grad_norm": 1.1840121236969028, "learning_rate": 1.771211549242646e-05, "loss": 0.4997, "step": 6823 }, { "epoch": 0.2431195111958245, "grad_norm": 1.297322115349516, "learning_rate": 1.7711380869284e-05, "loss": 0.5278, "step": 6824 }, { "epoch": 0.2431551383223186, "grad_norm": 1.5878144113672314, "learning_rate": 1.771064614345912e-05, "loss": 0.6433, "step": 6825 }, { "epoch": 0.2431907654488127, "grad_norm": 1.8430477569926562, "learning_rate": 1.7709911314961593e-05, "loss": 0.6269, "step": 6826 }, { "epoch": 0.24322639257530684, "grad_norm": 1.1565674931912142, "learning_rate": 1.770917638380121e-05, "loss": 0.5035, "step": 6827 }, { "epoch": 0.24326201970180095, "grad_norm": 1.8603901607691686, "learning_rate": 1.7708441349987753e-05, "loss": 0.614, "step": 6828 }, { "epoch": 0.24329764682829508, "grad_norm": 1.6166389569190793, "learning_rate": 1.7707706213531012e-05, "loss": 0.6385, "step": 6829 }, { "epoch": 0.24333327395478918, "grad_norm": 1.2365263987662158, "learning_rate": 1.7706970974440776e-05, "loss": 0.5575, "step": 6830 }, { "epoch": 0.24336890108128328, "grad_norm": 1.6977411250476173, "learning_rate": 1.7706235632726834e-05, "loss": 0.5883, "step": 6831 }, { "epoch": 0.24340452820777742, "grad_norm": 1.2978789016391505, "learning_rate": 1.7705500188398977e-05, "loss": 0.5385, "step": 6832 }, { "epoch": 0.24344015533427152, "grad_norm": 1.839750275672186, "learning_rate": 1.7704764641467004e-05, "loss": 0.5977, "step": 6833 }, { "epoch": 0.24347578246076562, "grad_norm": 1.109071261577315, "learning_rate": 1.77040289919407e-05, "loss": 0.4448, "step": 6834 }, { "epoch": 0.24351140958725975, "grad_norm": 1.3285596420910755, "learning_rate": 1.7703293239829865e-05, "loss": 0.5362, "step": 6835 }, { "epoch": 0.24354703671375386, "grad_norm": 1.6869531804230238, "learning_rate": 1.77025573851443e-05, "loss": 0.5815, "step": 6836 }, { "epoch": 0.24358266384024796, "grad_norm": 1.5279235800312303, "learning_rate": 1.77018214278938e-05, "loss": 0.5928, "step": 6837 }, { "epoch": 0.2436182909667421, "grad_norm": 1.9991272977177106, "learning_rate": 1.7701085368088157e-05, "loss": 0.6254, "step": 6838 }, { "epoch": 0.2436539180932362, "grad_norm": 1.221743254803457, "learning_rate": 1.7700349205737186e-05, "loss": 0.4842, "step": 6839 }, { "epoch": 0.2436895452197303, "grad_norm": 1.5178261837112663, "learning_rate": 1.769961294085068e-05, "loss": 0.6387, "step": 6840 }, { "epoch": 0.24372517234622443, "grad_norm": 1.0591641721483234, "learning_rate": 1.7698876573438448e-05, "loss": 0.5135, "step": 6841 }, { "epoch": 0.24376079947271853, "grad_norm": 2.032060552313224, "learning_rate": 1.7698140103510294e-05, "loss": 0.651, "step": 6842 }, { "epoch": 0.24379642659921263, "grad_norm": 1.4087661802888827, "learning_rate": 1.769740353107602e-05, "loss": 0.5367, "step": 6843 }, { "epoch": 0.24383205372570677, "grad_norm": 1.9996272963260617, "learning_rate": 1.7696666856145443e-05, "loss": 0.6233, "step": 6844 }, { "epoch": 0.24386768085220087, "grad_norm": 2.1727233175368856, "learning_rate": 1.769593007872836e-05, "loss": 0.6697, "step": 6845 }, { "epoch": 0.24390330797869497, "grad_norm": 1.1869976321924298, "learning_rate": 1.7695193198834598e-05, "loss": 0.472, "step": 6846 }, { "epoch": 0.2439389351051891, "grad_norm": 1.4921114364452082, "learning_rate": 1.7694456216473955e-05, "loss": 0.7137, "step": 6847 }, { "epoch": 0.2439745622316832, "grad_norm": 1.9350983816939042, "learning_rate": 1.7693719131656247e-05, "loss": 0.5913, "step": 6848 }, { "epoch": 0.2440101893581773, "grad_norm": 1.2195528589903313, "learning_rate": 1.7692981944391293e-05, "loss": 0.4978, "step": 6849 }, { "epoch": 0.24404581648467144, "grad_norm": 1.4477977159955184, "learning_rate": 1.769224465468891e-05, "loss": 0.5712, "step": 6850 }, { "epoch": 0.24408144361116554, "grad_norm": 1.1875703723902664, "learning_rate": 1.769150726255891e-05, "loss": 0.5175, "step": 6851 }, { "epoch": 0.24411707073765965, "grad_norm": 1.669492407873707, "learning_rate": 1.7690769768011118e-05, "loss": 0.5804, "step": 6852 }, { "epoch": 0.24415269786415378, "grad_norm": 1.1993118946953838, "learning_rate": 1.769003217105535e-05, "loss": 0.5467, "step": 6853 }, { "epoch": 0.24418832499064788, "grad_norm": 1.1736925266749871, "learning_rate": 1.7689294471701424e-05, "loss": 0.4876, "step": 6854 }, { "epoch": 0.24422395211714198, "grad_norm": 1.114843894084216, "learning_rate": 1.7688556669959173e-05, "loss": 0.4493, "step": 6855 }, { "epoch": 0.24425957924363612, "grad_norm": 1.8723072519377095, "learning_rate": 1.7687818765838415e-05, "loss": 0.5708, "step": 6856 }, { "epoch": 0.24429520637013022, "grad_norm": 1.5421091145095038, "learning_rate": 1.7687080759348975e-05, "loss": 0.5307, "step": 6857 }, { "epoch": 0.24433083349662432, "grad_norm": 2.073640029117121, "learning_rate": 1.7686342650500682e-05, "loss": 0.5963, "step": 6858 }, { "epoch": 0.24436646062311845, "grad_norm": 1.4966346387217724, "learning_rate": 1.7685604439303365e-05, "loss": 0.5761, "step": 6859 }, { "epoch": 0.24440208774961256, "grad_norm": 1.5417801855655842, "learning_rate": 1.7684866125766853e-05, "loss": 0.6157, "step": 6860 }, { "epoch": 0.24443771487610666, "grad_norm": 1.1335552148433714, "learning_rate": 1.7684127709900976e-05, "loss": 0.5607, "step": 6861 }, { "epoch": 0.2444733420026008, "grad_norm": 1.1652559489165246, "learning_rate": 1.7683389191715568e-05, "loss": 0.4888, "step": 6862 }, { "epoch": 0.2445089691290949, "grad_norm": 1.2157850996460446, "learning_rate": 1.7682650571220462e-05, "loss": 0.5565, "step": 6863 }, { "epoch": 0.244544596255589, "grad_norm": 1.613158977885857, "learning_rate": 1.7681911848425498e-05, "loss": 0.5625, "step": 6864 }, { "epoch": 0.24458022338208313, "grad_norm": 1.7396487516659624, "learning_rate": 1.76811730233405e-05, "loss": 0.6335, "step": 6865 }, { "epoch": 0.24461585050857723, "grad_norm": 1.8587114085931973, "learning_rate": 1.7680434095975322e-05, "loss": 0.6309, "step": 6866 }, { "epoch": 0.24465147763507133, "grad_norm": 1.1639164510383522, "learning_rate": 1.7679695066339792e-05, "loss": 0.4852, "step": 6867 }, { "epoch": 0.24468710476156547, "grad_norm": 1.1610435547251423, "learning_rate": 1.7678955934443757e-05, "loss": 0.5011, "step": 6868 }, { "epoch": 0.24472273188805957, "grad_norm": 1.5847432257377443, "learning_rate": 1.7678216700297056e-05, "loss": 0.6018, "step": 6869 }, { "epoch": 0.24475835901455367, "grad_norm": 1.1318121261258252, "learning_rate": 1.7677477363909534e-05, "loss": 0.5056, "step": 6870 }, { "epoch": 0.2447939861410478, "grad_norm": 2.0721218207022307, "learning_rate": 1.7676737925291027e-05, "loss": 0.536, "step": 6871 }, { "epoch": 0.2448296132675419, "grad_norm": 1.7074472621165728, "learning_rate": 1.7675998384451398e-05, "loss": 0.6509, "step": 6872 }, { "epoch": 0.244865240394036, "grad_norm": 1.3509258041004877, "learning_rate": 1.767525874140048e-05, "loss": 0.5661, "step": 6873 }, { "epoch": 0.24490086752053014, "grad_norm": 1.1635101993043149, "learning_rate": 1.7674518996148128e-05, "loss": 0.5364, "step": 6874 }, { "epoch": 0.24493649464702424, "grad_norm": 1.8211117703629212, "learning_rate": 1.7673779148704192e-05, "loss": 0.6039, "step": 6875 }, { "epoch": 0.24497212177351835, "grad_norm": 1.7850375559734872, "learning_rate": 1.7673039199078525e-05, "loss": 0.6315, "step": 6876 }, { "epoch": 0.24500774890001248, "grad_norm": 1.7309732797236763, "learning_rate": 1.7672299147280975e-05, "loss": 0.5876, "step": 6877 }, { "epoch": 0.24504337602650658, "grad_norm": 1.2800532446789212, "learning_rate": 1.7671558993321403e-05, "loss": 0.5249, "step": 6878 }, { "epoch": 0.24507900315300069, "grad_norm": 1.1649630130599486, "learning_rate": 1.7670818737209657e-05, "loss": 0.5276, "step": 6879 }, { "epoch": 0.24511463027949482, "grad_norm": 2.7783482410463076, "learning_rate": 1.76700783789556e-05, "loss": 0.5925, "step": 6880 }, { "epoch": 0.24515025740598892, "grad_norm": 1.226151369964102, "learning_rate": 1.7669337918569086e-05, "loss": 0.5169, "step": 6881 }, { "epoch": 0.24518588453248302, "grad_norm": 1.5437950862112182, "learning_rate": 1.7668597356059977e-05, "loss": 0.5935, "step": 6882 }, { "epoch": 0.24522151165897715, "grad_norm": 1.1951757322733838, "learning_rate": 1.766785669143814e-05, "loss": 0.5255, "step": 6883 }, { "epoch": 0.24525713878547126, "grad_norm": 1.1873883700480319, "learning_rate": 1.7667115924713427e-05, "loss": 0.4891, "step": 6884 }, { "epoch": 0.24529276591196536, "grad_norm": 1.9105985689578473, "learning_rate": 1.7666375055895707e-05, "loss": 0.6408, "step": 6885 }, { "epoch": 0.2453283930384595, "grad_norm": 1.132076944092759, "learning_rate": 1.7665634084994842e-05, "loss": 0.4073, "step": 6886 }, { "epoch": 0.2453640201649536, "grad_norm": 1.1948466508634934, "learning_rate": 1.7664893012020704e-05, "loss": 0.5569, "step": 6887 }, { "epoch": 0.2453996472914477, "grad_norm": 2.266786827872321, "learning_rate": 1.7664151836983157e-05, "loss": 0.5326, "step": 6888 }, { "epoch": 0.24543527441794183, "grad_norm": 1.6363130956581746, "learning_rate": 1.7663410559892073e-05, "loss": 0.621, "step": 6889 }, { "epoch": 0.24547090154443593, "grad_norm": 1.1017865864278045, "learning_rate": 1.766266918075732e-05, "loss": 0.5206, "step": 6890 }, { "epoch": 0.24550652867093004, "grad_norm": 1.123395522900814, "learning_rate": 1.7661927699588772e-05, "loss": 0.4846, "step": 6891 }, { "epoch": 0.24554215579742417, "grad_norm": 1.5972889064075704, "learning_rate": 1.7661186116396304e-05, "loss": 0.5466, "step": 6892 }, { "epoch": 0.24557778292391827, "grad_norm": 1.3437220225408202, "learning_rate": 1.766044443118978e-05, "loss": 0.5012, "step": 6893 }, { "epoch": 0.24561341005041237, "grad_norm": 1.0954365594218924, "learning_rate": 1.765970264397909e-05, "loss": 0.5054, "step": 6894 }, { "epoch": 0.2456490371769065, "grad_norm": 1.5799722481029126, "learning_rate": 1.7658960754774104e-05, "loss": 0.6549, "step": 6895 }, { "epoch": 0.2456846643034006, "grad_norm": 1.107934125446644, "learning_rate": 1.7658218763584704e-05, "loss": 0.5127, "step": 6896 }, { "epoch": 0.2457202914298947, "grad_norm": 1.7095474936297672, "learning_rate": 1.7657476670420766e-05, "loss": 0.6316, "step": 6897 }, { "epoch": 0.24575591855638884, "grad_norm": 1.9314194620487475, "learning_rate": 1.7656734475292175e-05, "loss": 0.5801, "step": 6898 }, { "epoch": 0.24579154568288294, "grad_norm": 1.0760730271580858, "learning_rate": 1.7655992178208813e-05, "loss": 0.5225, "step": 6899 }, { "epoch": 0.24582717280937705, "grad_norm": 1.5470684963024306, "learning_rate": 1.765524977918056e-05, "loss": 0.5649, "step": 6900 }, { "epoch": 0.24586279993587118, "grad_norm": 4.047154786174866, "learning_rate": 1.765450727821731e-05, "loss": 0.5624, "step": 6901 }, { "epoch": 0.24589842706236528, "grad_norm": 1.805118371629957, "learning_rate": 1.7653764675328945e-05, "loss": 0.5456, "step": 6902 }, { "epoch": 0.24593405418885939, "grad_norm": 1.065705334964575, "learning_rate": 1.7653021970525356e-05, "loss": 0.4547, "step": 6903 }, { "epoch": 0.24596968131535352, "grad_norm": 1.3873376172427583, "learning_rate": 1.7652279163816422e-05, "loss": 0.5862, "step": 6904 }, { "epoch": 0.24600530844184762, "grad_norm": 1.688390287857876, "learning_rate": 1.765153625521205e-05, "loss": 0.5614, "step": 6905 }, { "epoch": 0.24604093556834172, "grad_norm": 1.7860339940285017, "learning_rate": 1.765079324472212e-05, "loss": 0.6816, "step": 6906 }, { "epoch": 0.24607656269483585, "grad_norm": 2.9306656682131154, "learning_rate": 1.7650050132356537e-05, "loss": 0.6199, "step": 6907 }, { "epoch": 0.24611218982132996, "grad_norm": 1.7977094167823815, "learning_rate": 1.7649306918125182e-05, "loss": 0.6285, "step": 6908 }, { "epoch": 0.24614781694782406, "grad_norm": 3.641038056568871, "learning_rate": 1.7648563602037962e-05, "loss": 0.5855, "step": 6909 }, { "epoch": 0.2461834440743182, "grad_norm": 1.682444761957967, "learning_rate": 1.764782018410477e-05, "loss": 0.5946, "step": 6910 }, { "epoch": 0.2462190712008123, "grad_norm": 1.2029467247454182, "learning_rate": 1.764707666433551e-05, "loss": 0.4678, "step": 6911 }, { "epoch": 0.2462546983273064, "grad_norm": 2.1794326066135574, "learning_rate": 1.7646333042740077e-05, "loss": 0.5758, "step": 6912 }, { "epoch": 0.24629032545380053, "grad_norm": 1.1473898065883361, "learning_rate": 1.7645589319328374e-05, "loss": 0.5509, "step": 6913 }, { "epoch": 0.24632595258029463, "grad_norm": 2.199691484220849, "learning_rate": 1.764484549411031e-05, "loss": 0.5893, "step": 6914 }, { "epoch": 0.24636157970678874, "grad_norm": 1.7697583694825596, "learning_rate": 1.764410156709578e-05, "loss": 0.5583, "step": 6915 }, { "epoch": 0.24639720683328287, "grad_norm": 1.589242052513008, "learning_rate": 1.7643357538294697e-05, "loss": 0.6344, "step": 6916 }, { "epoch": 0.24643283395977697, "grad_norm": 1.2172414753644512, "learning_rate": 1.7642613407716963e-05, "loss": 0.5436, "step": 6917 }, { "epoch": 0.24646846108627107, "grad_norm": 1.2247481975984045, "learning_rate": 1.7641869175372493e-05, "loss": 0.532, "step": 6918 }, { "epoch": 0.2465040882127652, "grad_norm": 1.6958736168362387, "learning_rate": 1.7641124841271194e-05, "loss": 0.5592, "step": 6919 }, { "epoch": 0.2465397153392593, "grad_norm": 1.3056275020604344, "learning_rate": 1.7640380405422976e-05, "loss": 0.5049, "step": 6920 }, { "epoch": 0.2465753424657534, "grad_norm": 1.1706738050520578, "learning_rate": 1.763963586783775e-05, "loss": 0.5214, "step": 6921 }, { "epoch": 0.24661096959224754, "grad_norm": 1.4383694438554326, "learning_rate": 1.7638891228525434e-05, "loss": 0.5182, "step": 6922 }, { "epoch": 0.24664659671874165, "grad_norm": 1.1723573219605232, "learning_rate": 1.763814648749594e-05, "loss": 0.4984, "step": 6923 }, { "epoch": 0.24668222384523578, "grad_norm": 1.9397375129797454, "learning_rate": 1.763740164475919e-05, "loss": 0.6979, "step": 6924 }, { "epoch": 0.24671785097172988, "grad_norm": 1.1734252719314473, "learning_rate": 1.7636656700325102e-05, "loss": 0.4819, "step": 6925 }, { "epoch": 0.24675347809822398, "grad_norm": 1.2354076508908516, "learning_rate": 1.7635911654203588e-05, "loss": 0.5065, "step": 6926 }, { "epoch": 0.2467891052247181, "grad_norm": 1.7326802201878497, "learning_rate": 1.7635166506404574e-05, "loss": 0.599, "step": 6927 }, { "epoch": 0.24682473235121222, "grad_norm": 1.4931043640541932, "learning_rate": 1.763442125693798e-05, "loss": 0.5713, "step": 6928 }, { "epoch": 0.24686035947770632, "grad_norm": 1.270936929468331, "learning_rate": 1.7633675905813732e-05, "loss": 0.4978, "step": 6929 }, { "epoch": 0.24689598660420045, "grad_norm": 1.8377435967896731, "learning_rate": 1.7632930453041752e-05, "loss": 0.5834, "step": 6930 }, { "epoch": 0.24693161373069455, "grad_norm": 1.2101672845022045, "learning_rate": 1.763218489863197e-05, "loss": 0.541, "step": 6931 }, { "epoch": 0.24696724085718866, "grad_norm": 1.522652620012505, "learning_rate": 1.7631439242594314e-05, "loss": 0.5078, "step": 6932 }, { "epoch": 0.2470028679836828, "grad_norm": 1.9104003535981224, "learning_rate": 1.7630693484938707e-05, "loss": 0.5203, "step": 6933 }, { "epoch": 0.2470384951101769, "grad_norm": 1.670705494445263, "learning_rate": 1.762994762567508e-05, "loss": 0.4848, "step": 6934 }, { "epoch": 0.247074122236671, "grad_norm": 1.308623552199222, "learning_rate": 1.762920166481337e-05, "loss": 0.5496, "step": 6935 }, { "epoch": 0.24710974936316513, "grad_norm": 1.2792395457236307, "learning_rate": 1.762845560236351e-05, "loss": 0.5604, "step": 6936 }, { "epoch": 0.24714537648965923, "grad_norm": 1.7638194018116458, "learning_rate": 1.762770943833543e-05, "loss": 0.7252, "step": 6937 }, { "epoch": 0.24718100361615333, "grad_norm": 3.931759033218765, "learning_rate": 1.7626963172739066e-05, "loss": 0.5647, "step": 6938 }, { "epoch": 0.24721663074264746, "grad_norm": 1.4901914030847117, "learning_rate": 1.7626216805584357e-05, "loss": 0.611, "step": 6939 }, { "epoch": 0.24725225786914157, "grad_norm": 2.170525194450953, "learning_rate": 1.762547033688124e-05, "loss": 0.6583, "step": 6940 }, { "epoch": 0.24728788499563567, "grad_norm": 1.0970894918714091, "learning_rate": 1.7624723766639658e-05, "loss": 0.4913, "step": 6941 }, { "epoch": 0.2473235121221298, "grad_norm": 1.8704182548261035, "learning_rate": 1.7623977094869547e-05, "loss": 0.5871, "step": 6942 }, { "epoch": 0.2473591392486239, "grad_norm": 2.8885917211134333, "learning_rate": 1.7623230321580856e-05, "loss": 0.6565, "step": 6943 }, { "epoch": 0.247394766375118, "grad_norm": 1.1347045526054111, "learning_rate": 1.7622483446783523e-05, "loss": 0.5017, "step": 6944 }, { "epoch": 0.24743039350161214, "grad_norm": 1.2998487758887523, "learning_rate": 1.7621736470487493e-05, "loss": 0.5153, "step": 6945 }, { "epoch": 0.24746602062810624, "grad_norm": 1.283300921194799, "learning_rate": 1.7620989392702715e-05, "loss": 0.4975, "step": 6946 }, { "epoch": 0.24750164775460035, "grad_norm": 1.293072937675037, "learning_rate": 1.762024221343914e-05, "loss": 0.4878, "step": 6947 }, { "epoch": 0.24753727488109448, "grad_norm": 1.2122197005225095, "learning_rate": 1.761949493270671e-05, "loss": 0.5257, "step": 6948 }, { "epoch": 0.24757290200758858, "grad_norm": 1.7858058506554428, "learning_rate": 1.761874755051538e-05, "loss": 0.6478, "step": 6949 }, { "epoch": 0.24760852913408268, "grad_norm": 1.5413913883911927, "learning_rate": 1.7618000066875104e-05, "loss": 0.4993, "step": 6950 }, { "epoch": 0.24764415626057681, "grad_norm": 1.5629442131218971, "learning_rate": 1.761725248179583e-05, "loss": 0.6074, "step": 6951 }, { "epoch": 0.24767978338707092, "grad_norm": 2.311521783731735, "learning_rate": 1.7616504795287516e-05, "loss": 0.578, "step": 6952 }, { "epoch": 0.24771541051356502, "grad_norm": 1.7741883071203237, "learning_rate": 1.7615757007360114e-05, "loss": 0.5818, "step": 6953 }, { "epoch": 0.24775103764005915, "grad_norm": 1.7748917379775342, "learning_rate": 1.7615009118023587e-05, "loss": 0.564, "step": 6954 }, { "epoch": 0.24778666476655326, "grad_norm": 1.7403156372506632, "learning_rate": 1.761426112728789e-05, "loss": 0.6099, "step": 6955 }, { "epoch": 0.24782229189304736, "grad_norm": 1.3034550750285323, "learning_rate": 1.7613513035162983e-05, "loss": 0.504, "step": 6956 }, { "epoch": 0.2478579190195415, "grad_norm": 1.843952451704499, "learning_rate": 1.761276484165883e-05, "loss": 0.6097, "step": 6957 }, { "epoch": 0.2478935461460356, "grad_norm": 1.3557818094264722, "learning_rate": 1.7612016546785396e-05, "loss": 0.5406, "step": 6958 }, { "epoch": 0.2479291732725297, "grad_norm": 1.1737301092465535, "learning_rate": 1.7611268150552637e-05, "loss": 0.4924, "step": 6959 }, { "epoch": 0.24796480039902383, "grad_norm": 1.158858827799794, "learning_rate": 1.7610519652970525e-05, "loss": 0.5108, "step": 6960 }, { "epoch": 0.24800042752551793, "grad_norm": 1.903192881307904, "learning_rate": 1.760977105404902e-05, "loss": 0.6161, "step": 6961 }, { "epoch": 0.24803605465201203, "grad_norm": 2.0121100078107945, "learning_rate": 1.7609022353798097e-05, "loss": 0.6042, "step": 6962 }, { "epoch": 0.24807168177850616, "grad_norm": 1.5700049259674258, "learning_rate": 1.7608273552227723e-05, "loss": 0.5488, "step": 6963 }, { "epoch": 0.24810730890500027, "grad_norm": 1.2975673638707648, "learning_rate": 1.7607524649347867e-05, "loss": 0.5065, "step": 6964 }, { "epoch": 0.24814293603149437, "grad_norm": 1.0795884840827743, "learning_rate": 1.7606775645168506e-05, "loss": 0.4715, "step": 6965 }, { "epoch": 0.2481785631579885, "grad_norm": 1.7611587500134425, "learning_rate": 1.760602653969961e-05, "loss": 0.5612, "step": 6966 }, { "epoch": 0.2482141902844826, "grad_norm": 2.0449156350571704, "learning_rate": 1.7605277332951155e-05, "loss": 0.7285, "step": 6967 }, { "epoch": 0.2482498174109767, "grad_norm": 1.120087363816585, "learning_rate": 1.7604528024933113e-05, "loss": 0.5193, "step": 6968 }, { "epoch": 0.24828544453747084, "grad_norm": 1.2376273298841831, "learning_rate": 1.7603778615655468e-05, "loss": 0.5114, "step": 6969 }, { "epoch": 0.24832107166396494, "grad_norm": 3.1980892512619783, "learning_rate": 1.7603029105128193e-05, "loss": 0.6443, "step": 6970 }, { "epoch": 0.24835669879045905, "grad_norm": 1.173846953253906, "learning_rate": 1.7602279493361277e-05, "loss": 0.5598, "step": 6971 }, { "epoch": 0.24839232591695318, "grad_norm": 1.3455670400275526, "learning_rate": 1.7601529780364692e-05, "loss": 0.5336, "step": 6972 }, { "epoch": 0.24842795304344728, "grad_norm": 2.0337949307007177, "learning_rate": 1.760077996614842e-05, "loss": 0.6338, "step": 6973 }, { "epoch": 0.24846358016994138, "grad_norm": 2.6019034448983667, "learning_rate": 1.760003005072246e-05, "loss": 0.5729, "step": 6974 }, { "epoch": 0.24849920729643551, "grad_norm": 2.4274127698642434, "learning_rate": 1.759928003409678e-05, "loss": 0.6261, "step": 6975 }, { "epoch": 0.24853483442292962, "grad_norm": 1.5662569706564275, "learning_rate": 1.759852991628138e-05, "loss": 0.5892, "step": 6976 }, { "epoch": 0.24857046154942372, "grad_norm": 1.3176871212074177, "learning_rate": 1.759777969728624e-05, "loss": 0.5648, "step": 6977 }, { "epoch": 0.24860608867591785, "grad_norm": 1.8478096264682575, "learning_rate": 1.7597029377121356e-05, "loss": 0.6263, "step": 6978 }, { "epoch": 0.24864171580241196, "grad_norm": 2.118509009671848, "learning_rate": 1.759627895579671e-05, "loss": 0.601, "step": 6979 }, { "epoch": 0.24867734292890606, "grad_norm": 2.4905919739709033, "learning_rate": 1.7595528433322308e-05, "loss": 0.628, "step": 6980 }, { "epoch": 0.2487129700554002, "grad_norm": 4.217771570285711, "learning_rate": 1.759477780970813e-05, "loss": 0.6707, "step": 6981 }, { "epoch": 0.2487485971818943, "grad_norm": 1.3940462745403344, "learning_rate": 1.759402708496418e-05, "loss": 0.518, "step": 6982 }, { "epoch": 0.2487842243083884, "grad_norm": 1.9686545524564, "learning_rate": 1.7593276259100447e-05, "loss": 0.5424, "step": 6983 }, { "epoch": 0.24881985143488253, "grad_norm": 1.634217305352532, "learning_rate": 1.7592525332126938e-05, "loss": 0.5417, "step": 6984 }, { "epoch": 0.24885547856137663, "grad_norm": 2.226430213446079, "learning_rate": 1.7591774304053646e-05, "loss": 0.5984, "step": 6985 }, { "epoch": 0.24889110568787073, "grad_norm": 1.1224559225668123, "learning_rate": 1.759102317489057e-05, "loss": 0.4961, "step": 6986 }, { "epoch": 0.24892673281436486, "grad_norm": 1.421148587217465, "learning_rate": 1.7590271944647718e-05, "loss": 0.5297, "step": 6987 }, { "epoch": 0.24896235994085897, "grad_norm": 1.0756038507210792, "learning_rate": 1.758952061333509e-05, "loss": 0.4876, "step": 6988 }, { "epoch": 0.24899798706735307, "grad_norm": 1.2557098361259256, "learning_rate": 1.758876918096269e-05, "loss": 0.5246, "step": 6989 }, { "epoch": 0.2490336141938472, "grad_norm": 1.87486356142742, "learning_rate": 1.758801764754052e-05, "loss": 0.5614, "step": 6990 }, { "epoch": 0.2490692413203413, "grad_norm": 1.8398012626550386, "learning_rate": 1.7587266013078596e-05, "loss": 0.4724, "step": 6991 }, { "epoch": 0.2491048684468354, "grad_norm": 1.1599401597163068, "learning_rate": 1.758651427758692e-05, "loss": 0.4511, "step": 6992 }, { "epoch": 0.24914049557332954, "grad_norm": 1.060712256995105, "learning_rate": 1.7585762441075504e-05, "loss": 0.5423, "step": 6993 }, { "epoch": 0.24917612269982364, "grad_norm": 1.6132429878130174, "learning_rate": 1.758501050355436e-05, "loss": 0.5146, "step": 6994 }, { "epoch": 0.24921174982631775, "grad_norm": 1.4690586064384443, "learning_rate": 1.75842584650335e-05, "loss": 0.5211, "step": 6995 }, { "epoch": 0.24924737695281188, "grad_norm": 1.5229109734978032, "learning_rate": 1.7583506325522936e-05, "loss": 0.5869, "step": 6996 }, { "epoch": 0.24928300407930598, "grad_norm": 1.1662524947404853, "learning_rate": 1.7582754085032683e-05, "loss": 0.5542, "step": 6997 }, { "epoch": 0.24931863120580008, "grad_norm": 2.6569789348479405, "learning_rate": 1.7582001743572763e-05, "loss": 0.5871, "step": 6998 }, { "epoch": 0.24935425833229422, "grad_norm": 1.981882151036804, "learning_rate": 1.7581249301153187e-05, "loss": 0.6051, "step": 6999 }, { "epoch": 0.24938988545878832, "grad_norm": 2.0868508085918624, "learning_rate": 1.758049675778398e-05, "loss": 0.6822, "step": 7000 }, { "epoch": 0.24942551258528242, "grad_norm": 2.8626715614397975, "learning_rate": 1.757974411347516e-05, "loss": 0.6728, "step": 7001 }, { "epoch": 0.24946113971177655, "grad_norm": 1.9349909741088152, "learning_rate": 1.757899136823675e-05, "loss": 0.6427, "step": 7002 }, { "epoch": 0.24949676683827066, "grad_norm": 1.5994331243304238, "learning_rate": 1.757823852207877e-05, "loss": 0.6767, "step": 7003 }, { "epoch": 0.24953239396476476, "grad_norm": 1.7423551767685421, "learning_rate": 1.757748557501125e-05, "loss": 0.6439, "step": 7004 }, { "epoch": 0.2495680210912589, "grad_norm": 1.8870715789278438, "learning_rate": 1.757673252704421e-05, "loss": 0.6041, "step": 7005 }, { "epoch": 0.249603648217753, "grad_norm": 1.2964681684686246, "learning_rate": 1.7575979378187683e-05, "loss": 0.5427, "step": 7006 }, { "epoch": 0.2496392753442471, "grad_norm": 1.6364275792524703, "learning_rate": 1.757522612845169e-05, "loss": 0.5427, "step": 7007 }, { "epoch": 0.24967490247074123, "grad_norm": 1.3097719909350143, "learning_rate": 1.7574472777846276e-05, "loss": 0.4964, "step": 7008 }, { "epoch": 0.24971052959723533, "grad_norm": 2.189209725316778, "learning_rate": 1.7573719326381455e-05, "loss": 0.5771, "step": 7009 }, { "epoch": 0.24974615672372943, "grad_norm": 1.296094980130625, "learning_rate": 1.757296577406727e-05, "loss": 0.5018, "step": 7010 }, { "epoch": 0.24978178385022357, "grad_norm": 1.1717620734896808, "learning_rate": 1.757221212091375e-05, "loss": 0.4864, "step": 7011 }, { "epoch": 0.24981741097671767, "grad_norm": 1.5833471730750324, "learning_rate": 1.7571458366930937e-05, "loss": 0.5454, "step": 7012 }, { "epoch": 0.24985303810321177, "grad_norm": 2.234165966653855, "learning_rate": 1.7570704512128865e-05, "loss": 0.6545, "step": 7013 }, { "epoch": 0.2498886652297059, "grad_norm": 1.4785984563216776, "learning_rate": 1.7569950556517566e-05, "loss": 0.6428, "step": 7014 }, { "epoch": 0.2499242923562, "grad_norm": 1.1974468212789942, "learning_rate": 1.7569196500107087e-05, "loss": 0.5155, "step": 7015 }, { "epoch": 0.2499599194826941, "grad_norm": 1.8427242180037067, "learning_rate": 1.7568442342907463e-05, "loss": 0.6553, "step": 7016 }, { "epoch": 0.24999554660918824, "grad_norm": 1.763041942564035, "learning_rate": 1.7567688084928743e-05, "loss": 0.6755, "step": 7017 }, { "epoch": 0.25003117373568234, "grad_norm": 1.739083621168562, "learning_rate": 1.7566933726180963e-05, "loss": 0.5846, "step": 7018 }, { "epoch": 0.2500668008621765, "grad_norm": 1.9641236121951495, "learning_rate": 1.7566179266674175e-05, "loss": 0.5307, "step": 7019 }, { "epoch": 0.25010242798867055, "grad_norm": 1.3223777436803419, "learning_rate": 1.7565424706418422e-05, "loss": 0.5029, "step": 7020 }, { "epoch": 0.2501380551151647, "grad_norm": 1.971874778897127, "learning_rate": 1.7564670045423747e-05, "loss": 0.6603, "step": 7021 }, { "epoch": 0.2501736822416588, "grad_norm": 1.7421982343543405, "learning_rate": 1.7563915283700205e-05, "loss": 0.5904, "step": 7022 }, { "epoch": 0.2502093093681529, "grad_norm": 1.0945047361152989, "learning_rate": 1.7563160421257845e-05, "loss": 0.522, "step": 7023 }, { "epoch": 0.250244936494647, "grad_norm": 1.7530029670138234, "learning_rate": 1.7562405458106718e-05, "loss": 0.5205, "step": 7024 }, { "epoch": 0.25028056362114115, "grad_norm": 1.3781355075032218, "learning_rate": 1.7561650394256874e-05, "loss": 0.5083, "step": 7025 }, { "epoch": 0.2503161907476352, "grad_norm": 1.644700578257319, "learning_rate": 1.7560895229718374e-05, "loss": 0.6004, "step": 7026 }, { "epoch": 0.25035181787412936, "grad_norm": 1.1620221650312759, "learning_rate": 1.7560139964501263e-05, "loss": 0.544, "step": 7027 }, { "epoch": 0.2503874450006235, "grad_norm": 2.0087496616203917, "learning_rate": 1.755938459861561e-05, "loss": 0.6575, "step": 7028 }, { "epoch": 0.25042307212711756, "grad_norm": 1.5567103389698467, "learning_rate": 1.7558629132071465e-05, "loss": 0.5644, "step": 7029 }, { "epoch": 0.2504586992536117, "grad_norm": 1.6151266388457812, "learning_rate": 1.755787356487889e-05, "loss": 0.6254, "step": 7030 }, { "epoch": 0.2504943263801058, "grad_norm": 1.7625881673043213, "learning_rate": 1.7557117897047944e-05, "loss": 0.6155, "step": 7031 }, { "epoch": 0.2505299535065999, "grad_norm": 1.647518466206359, "learning_rate": 1.7556362128588695e-05, "loss": 0.5233, "step": 7032 }, { "epoch": 0.25056558063309403, "grad_norm": 1.9259525520504324, "learning_rate": 1.75556062595112e-05, "loss": 0.5003, "step": 7033 }, { "epoch": 0.25060120775958816, "grad_norm": 1.6874159710929275, "learning_rate": 1.7554850289825527e-05, "loss": 0.5579, "step": 7034 }, { "epoch": 0.25063683488608224, "grad_norm": 1.1143430583311311, "learning_rate": 1.7554094219541743e-05, "loss": 0.5508, "step": 7035 }, { "epoch": 0.25067246201257637, "grad_norm": 1.8609526097298603, "learning_rate": 1.755333804866991e-05, "loss": 0.5677, "step": 7036 }, { "epoch": 0.2507080891390705, "grad_norm": 1.1031120324813923, "learning_rate": 1.755258177722011e-05, "loss": 0.4513, "step": 7037 }, { "epoch": 0.2507437162655646, "grad_norm": 1.161349975875971, "learning_rate": 1.7551825405202397e-05, "loss": 0.4956, "step": 7038 }, { "epoch": 0.2507793433920587, "grad_norm": 1.2633408450432453, "learning_rate": 1.7551068932626854e-05, "loss": 0.5403, "step": 7039 }, { "epoch": 0.25081497051855284, "grad_norm": 1.6518244843790464, "learning_rate": 1.7550312359503547e-05, "loss": 0.6322, "step": 7040 }, { "epoch": 0.2508505976450469, "grad_norm": 1.616575594588499, "learning_rate": 1.7549555685842556e-05, "loss": 0.6244, "step": 7041 }, { "epoch": 0.25088622477154104, "grad_norm": 1.0920484731609765, "learning_rate": 1.7548798911653956e-05, "loss": 0.526, "step": 7042 }, { "epoch": 0.2509218518980352, "grad_norm": 1.092946299388673, "learning_rate": 1.754804203694782e-05, "loss": 0.4475, "step": 7043 }, { "epoch": 0.25095747902452925, "grad_norm": 1.9392090574172185, "learning_rate": 1.7547285061734227e-05, "loss": 0.5849, "step": 7044 }, { "epoch": 0.2509931061510234, "grad_norm": 2.062908529513215, "learning_rate": 1.754652798602326e-05, "loss": 0.6206, "step": 7045 }, { "epoch": 0.2510287332775175, "grad_norm": 1.7190784367819139, "learning_rate": 1.7545770809824997e-05, "loss": 0.6243, "step": 7046 }, { "epoch": 0.2510643604040116, "grad_norm": 1.136841185623304, "learning_rate": 1.7545013533149525e-05, "loss": 0.4819, "step": 7047 }, { "epoch": 0.2510999875305057, "grad_norm": 1.0586361039392247, "learning_rate": 1.754425615600692e-05, "loss": 0.5589, "step": 7048 }, { "epoch": 0.25113561465699985, "grad_norm": 1.0723759327709845, "learning_rate": 1.7543498678407275e-05, "loss": 0.5448, "step": 7049 }, { "epoch": 0.2511712417834939, "grad_norm": 1.030298995981059, "learning_rate": 1.754274110036067e-05, "loss": 0.4614, "step": 7050 }, { "epoch": 0.25120686890998806, "grad_norm": 2.5551854229325706, "learning_rate": 1.7541983421877196e-05, "loss": 0.6092, "step": 7051 }, { "epoch": 0.2512424960364822, "grad_norm": 1.2058802249453917, "learning_rate": 1.754122564296694e-05, "loss": 0.4853, "step": 7052 }, { "epoch": 0.25127812316297626, "grad_norm": 1.1621382442588892, "learning_rate": 1.7540467763639994e-05, "loss": 0.5203, "step": 7053 }, { "epoch": 0.2513137502894704, "grad_norm": 1.8851707027416098, "learning_rate": 1.753970978390645e-05, "loss": 0.5366, "step": 7054 }, { "epoch": 0.2513493774159645, "grad_norm": 1.0968048010402724, "learning_rate": 1.75389517037764e-05, "loss": 0.5028, "step": 7055 }, { "epoch": 0.2513850045424586, "grad_norm": 1.1117206220697167, "learning_rate": 1.7538193523259938e-05, "loss": 0.5019, "step": 7056 }, { "epoch": 0.25142063166895273, "grad_norm": 1.6666841497638332, "learning_rate": 1.753743524236716e-05, "loss": 0.5681, "step": 7057 }, { "epoch": 0.25145625879544686, "grad_norm": 1.0569692753140736, "learning_rate": 1.7536676861108167e-05, "loss": 0.4746, "step": 7058 }, { "epoch": 0.25149188592194094, "grad_norm": 2.1665838568826654, "learning_rate": 1.753591837949305e-05, "loss": 0.6815, "step": 7059 }, { "epoch": 0.25152751304843507, "grad_norm": 1.5355378117345277, "learning_rate": 1.7535159797531914e-05, "loss": 0.4882, "step": 7060 }, { "epoch": 0.2515631401749292, "grad_norm": 1.088634784294451, "learning_rate": 1.7534401115234857e-05, "loss": 0.5603, "step": 7061 }, { "epoch": 0.2515987673014233, "grad_norm": 1.8374465937088926, "learning_rate": 1.7533642332611986e-05, "loss": 0.5822, "step": 7062 }, { "epoch": 0.2516343944279174, "grad_norm": 2.8951364283971657, "learning_rate": 1.7532883449673396e-05, "loss": 0.6374, "step": 7063 }, { "epoch": 0.25167002155441154, "grad_norm": 1.1838370088519135, "learning_rate": 1.7532124466429202e-05, "loss": 0.4924, "step": 7064 }, { "epoch": 0.25170564868090567, "grad_norm": 2.022835434097388, "learning_rate": 1.7531365382889508e-05, "loss": 0.6129, "step": 7065 }, { "epoch": 0.25174127580739974, "grad_norm": 1.0880673525528244, "learning_rate": 1.7530606199064416e-05, "loss": 0.5108, "step": 7066 }, { "epoch": 0.2517769029338939, "grad_norm": 1.037994552308981, "learning_rate": 1.752984691496404e-05, "loss": 0.4889, "step": 7067 }, { "epoch": 0.251812530060388, "grad_norm": 1.7898736505798023, "learning_rate": 1.752908753059849e-05, "loss": 0.478, "step": 7068 }, { "epoch": 0.2518481571868821, "grad_norm": 1.1221651241860824, "learning_rate": 1.7528328045977876e-05, "loss": 0.4397, "step": 7069 }, { "epoch": 0.2518837843133762, "grad_norm": 2.320997559480588, "learning_rate": 1.7527568461112313e-05, "loss": 0.6829, "step": 7070 }, { "epoch": 0.25191941143987034, "grad_norm": 1.0498131854268324, "learning_rate": 1.7526808776011913e-05, "loss": 0.502, "step": 7071 }, { "epoch": 0.2519550385663644, "grad_norm": 1.5809362408953012, "learning_rate": 1.7526048990686793e-05, "loss": 0.549, "step": 7072 }, { "epoch": 0.25199066569285855, "grad_norm": 2.031628179155845, "learning_rate": 1.752528910514707e-05, "loss": 0.559, "step": 7073 }, { "epoch": 0.2520262928193527, "grad_norm": 1.1276839548151922, "learning_rate": 1.7524529119402867e-05, "loss": 0.5101, "step": 7074 }, { "epoch": 0.25206191994584676, "grad_norm": 1.216168968733998, "learning_rate": 1.7523769033464298e-05, "loss": 0.5137, "step": 7075 }, { "epoch": 0.2520975470723409, "grad_norm": 1.6406602828397696, "learning_rate": 1.7523008847341484e-05, "loss": 0.6146, "step": 7076 }, { "epoch": 0.252133174198835, "grad_norm": 1.093847132951071, "learning_rate": 1.752224856104455e-05, "loss": 0.489, "step": 7077 }, { "epoch": 0.2521688013253291, "grad_norm": 0.9961579083001498, "learning_rate": 1.7521488174583616e-05, "loss": 0.4702, "step": 7078 }, { "epoch": 0.2522044284518232, "grad_norm": 1.656555598373267, "learning_rate": 1.752072768796881e-05, "loss": 0.6071, "step": 7079 }, { "epoch": 0.25224005557831736, "grad_norm": 1.164549146966372, "learning_rate": 1.7519967101210264e-05, "loss": 0.4946, "step": 7080 }, { "epoch": 0.25227568270481143, "grad_norm": 1.1662427766422465, "learning_rate": 1.7519206414318092e-05, "loss": 0.5389, "step": 7081 }, { "epoch": 0.25231130983130556, "grad_norm": 2.0459788129461915, "learning_rate": 1.7518445627302438e-05, "loss": 0.5979, "step": 7082 }, { "epoch": 0.2523469369577997, "grad_norm": 1.5236992033888046, "learning_rate": 1.7517684740173418e-05, "loss": 0.5044, "step": 7083 }, { "epoch": 0.25238256408429377, "grad_norm": 1.137621514401241, "learning_rate": 1.7516923752941174e-05, "loss": 0.4725, "step": 7084 }, { "epoch": 0.2524181912107879, "grad_norm": 1.596753891561118, "learning_rate": 1.751616266561584e-05, "loss": 0.5989, "step": 7085 }, { "epoch": 0.25245381833728203, "grad_norm": 1.511856329468634, "learning_rate": 1.7515401478207542e-05, "loss": 0.618, "step": 7086 }, { "epoch": 0.2524894454637761, "grad_norm": 1.5660748736532064, "learning_rate": 1.7514640190726425e-05, "loss": 0.624, "step": 7087 }, { "epoch": 0.25252507259027024, "grad_norm": 1.580067884380838, "learning_rate": 1.751387880318262e-05, "loss": 0.6296, "step": 7088 }, { "epoch": 0.25256069971676437, "grad_norm": 1.9481875455870312, "learning_rate": 1.7513117315586265e-05, "loss": 0.6166, "step": 7089 }, { "epoch": 0.25259632684325845, "grad_norm": 1.1114283808832506, "learning_rate": 1.75123557279475e-05, "loss": 0.5008, "step": 7090 }, { "epoch": 0.2526319539697526, "grad_norm": 1.1010334941275057, "learning_rate": 1.7511594040276468e-05, "loss": 0.48, "step": 7091 }, { "epoch": 0.2526675810962467, "grad_norm": 1.6939954583635508, "learning_rate": 1.7510832252583317e-05, "loss": 0.6619, "step": 7092 }, { "epoch": 0.2527032082227408, "grad_norm": 1.0581068721245632, "learning_rate": 1.7510070364878177e-05, "loss": 0.5324, "step": 7093 }, { "epoch": 0.2527388353492349, "grad_norm": 1.6472915427128443, "learning_rate": 1.7509308377171205e-05, "loss": 0.5531, "step": 7094 }, { "epoch": 0.25277446247572904, "grad_norm": 1.618275663757686, "learning_rate": 1.7508546289472543e-05, "loss": 0.6446, "step": 7095 }, { "epoch": 0.2528100896022231, "grad_norm": 1.6952035391410478, "learning_rate": 1.7507784101792342e-05, "loss": 0.6444, "step": 7096 }, { "epoch": 0.25284571672871725, "grad_norm": 1.0795660546949994, "learning_rate": 1.7507021814140745e-05, "loss": 0.4826, "step": 7097 }, { "epoch": 0.2528813438552114, "grad_norm": 1.2168622333895338, "learning_rate": 1.7506259426527903e-05, "loss": 0.4755, "step": 7098 }, { "epoch": 0.25291697098170546, "grad_norm": 1.1023709804264956, "learning_rate": 1.7505496938963974e-05, "loss": 0.5048, "step": 7099 }, { "epoch": 0.2529525981081996, "grad_norm": 1.1784183467661244, "learning_rate": 1.7504734351459105e-05, "loss": 0.5265, "step": 7100 }, { "epoch": 0.2529882252346937, "grad_norm": 1.8360997997631636, "learning_rate": 1.7503971664023454e-05, "loss": 0.5557, "step": 7101 }, { "epoch": 0.2530238523611878, "grad_norm": 1.0880597490333388, "learning_rate": 1.7503208876667175e-05, "loss": 0.4869, "step": 7102 }, { "epoch": 0.2530594794876819, "grad_norm": 0.9966694523776614, "learning_rate": 1.7502445989400426e-05, "loss": 0.4606, "step": 7103 }, { "epoch": 0.25309510661417606, "grad_norm": 1.1570882300150493, "learning_rate": 1.750168300223336e-05, "loss": 0.523, "step": 7104 }, { "epoch": 0.25313073374067013, "grad_norm": 1.062653572640552, "learning_rate": 1.7500919915176147e-05, "loss": 0.4609, "step": 7105 }, { "epoch": 0.25316636086716426, "grad_norm": 1.0384310824456944, "learning_rate": 1.750015672823894e-05, "loss": 0.467, "step": 7106 }, { "epoch": 0.2532019879936584, "grad_norm": 1.6352234449142007, "learning_rate": 1.7499393441431903e-05, "loss": 0.672, "step": 7107 }, { "epoch": 0.25323761512015247, "grad_norm": 1.4036201738377243, "learning_rate": 1.7498630054765204e-05, "loss": 0.6082, "step": 7108 }, { "epoch": 0.2532732422466466, "grad_norm": 1.994691709318112, "learning_rate": 1.7497866568249e-05, "loss": 0.5237, "step": 7109 }, { "epoch": 0.25330886937314073, "grad_norm": 1.6220693764837273, "learning_rate": 1.749710298189347e-05, "loss": 0.5285, "step": 7110 }, { "epoch": 0.2533444964996348, "grad_norm": 1.1332289993865292, "learning_rate": 1.7496339295708767e-05, "loss": 0.5009, "step": 7111 }, { "epoch": 0.25338012362612894, "grad_norm": 1.2332880528321286, "learning_rate": 1.749557550970507e-05, "loss": 0.5327, "step": 7112 }, { "epoch": 0.25341575075262307, "grad_norm": 1.167804332102143, "learning_rate": 1.7494811623892543e-05, "loss": 0.4894, "step": 7113 }, { "epoch": 0.25345137787911715, "grad_norm": 1.7227954331594773, "learning_rate": 1.7494047638281363e-05, "loss": 0.5808, "step": 7114 }, { "epoch": 0.2534870050056113, "grad_norm": 2.1785726561416547, "learning_rate": 1.7493283552881696e-05, "loss": 0.6294, "step": 7115 }, { "epoch": 0.2535226321321054, "grad_norm": 1.147429194271815, "learning_rate": 1.7492519367703725e-05, "loss": 0.4897, "step": 7116 }, { "epoch": 0.2535582592585995, "grad_norm": 1.4523068717932544, "learning_rate": 1.749175508275762e-05, "loss": 0.585, "step": 7117 }, { "epoch": 0.2535938863850936, "grad_norm": 1.2131712966421615, "learning_rate": 1.7490990698053562e-05, "loss": 0.5361, "step": 7118 }, { "epoch": 0.25362951351158775, "grad_norm": 1.546245415117508, "learning_rate": 1.7490226213601725e-05, "loss": 0.5768, "step": 7119 }, { "epoch": 0.2536651406380818, "grad_norm": 1.275528000633684, "learning_rate": 1.7489461629412293e-05, "loss": 0.5172, "step": 7120 }, { "epoch": 0.25370076776457595, "grad_norm": 1.1806625183408335, "learning_rate": 1.748869694549544e-05, "loss": 0.5501, "step": 7121 }, { "epoch": 0.2537363948910701, "grad_norm": 3.0251339273826945, "learning_rate": 1.748793216186136e-05, "loss": 0.5733, "step": 7122 }, { "epoch": 0.25377202201756416, "grad_norm": 1.8722966446617837, "learning_rate": 1.7487167278520222e-05, "loss": 0.5969, "step": 7123 }, { "epoch": 0.2538076491440583, "grad_norm": 1.0838043945188396, "learning_rate": 1.7486402295482224e-05, "loss": 0.5155, "step": 7124 }, { "epoch": 0.2538432762705524, "grad_norm": 1.642030306032132, "learning_rate": 1.748563721275754e-05, "loss": 0.588, "step": 7125 }, { "epoch": 0.2538789033970465, "grad_norm": 1.0994410799688514, "learning_rate": 1.7484872030356373e-05, "loss": 0.5442, "step": 7126 }, { "epoch": 0.2539145305235406, "grad_norm": 1.1340319588609709, "learning_rate": 1.74841067482889e-05, "loss": 0.574, "step": 7127 }, { "epoch": 0.25395015765003476, "grad_norm": 1.2600566170819163, "learning_rate": 1.748334136656531e-05, "loss": 0.5369, "step": 7128 }, { "epoch": 0.25398578477652883, "grad_norm": 1.7170312577486984, "learning_rate": 1.7482575885195806e-05, "loss": 0.6196, "step": 7129 }, { "epoch": 0.25402141190302296, "grad_norm": 1.775172460723537, "learning_rate": 1.7481810304190568e-05, "loss": 0.5843, "step": 7130 }, { "epoch": 0.2540570390295171, "grad_norm": 1.2040525327365474, "learning_rate": 1.74810446235598e-05, "loss": 0.5626, "step": 7131 }, { "epoch": 0.25409266615601117, "grad_norm": 1.1346521548673647, "learning_rate": 1.7480278843313692e-05, "loss": 0.4882, "step": 7132 }, { "epoch": 0.2541282932825053, "grad_norm": 1.0737651250046203, "learning_rate": 1.7479512963462448e-05, "loss": 0.5454, "step": 7133 }, { "epoch": 0.25416392040899943, "grad_norm": 4.089797605065023, "learning_rate": 1.7478746984016256e-05, "loss": 0.6851, "step": 7134 }, { "epoch": 0.2541995475354935, "grad_norm": 1.6077056119657849, "learning_rate": 1.747798090498532e-05, "loss": 0.6131, "step": 7135 }, { "epoch": 0.25423517466198764, "grad_norm": 1.047376264644933, "learning_rate": 1.7477214726379844e-05, "loss": 0.5043, "step": 7136 }, { "epoch": 0.25427080178848177, "grad_norm": 1.1173760216588176, "learning_rate": 1.7476448448210027e-05, "loss": 0.5369, "step": 7137 }, { "epoch": 0.25430642891497585, "grad_norm": 1.8475189939207786, "learning_rate": 1.747568207048607e-05, "loss": 0.6487, "step": 7138 }, { "epoch": 0.25434205604147, "grad_norm": 1.3497490557005902, "learning_rate": 1.7474915593218185e-05, "loss": 0.5668, "step": 7139 }, { "epoch": 0.2543776831679641, "grad_norm": 2.4425528562244594, "learning_rate": 1.7474149016416576e-05, "loss": 0.6244, "step": 7140 }, { "epoch": 0.2544133102944582, "grad_norm": 1.5826871434980754, "learning_rate": 1.747338234009144e-05, "loss": 0.6723, "step": 7141 }, { "epoch": 0.2544489374209523, "grad_norm": 1.109046389775086, "learning_rate": 1.7472615564253004e-05, "loss": 0.4976, "step": 7142 }, { "epoch": 0.25448456454744645, "grad_norm": 1.3097977740733704, "learning_rate": 1.7471848688911465e-05, "loss": 0.4914, "step": 7143 }, { "epoch": 0.2545201916739405, "grad_norm": 1.171718349503691, "learning_rate": 1.747108171407704e-05, "loss": 0.5146, "step": 7144 }, { "epoch": 0.25455581880043465, "grad_norm": 2.3076806639027363, "learning_rate": 1.7470314639759936e-05, "loss": 0.5465, "step": 7145 }, { "epoch": 0.2545914459269288, "grad_norm": 1.2958456035207038, "learning_rate": 1.7469547465970373e-05, "loss": 0.5776, "step": 7146 }, { "epoch": 0.25462707305342286, "grad_norm": 1.1281546427990645, "learning_rate": 1.7468780192718566e-05, "loss": 0.4422, "step": 7147 }, { "epoch": 0.254662700179917, "grad_norm": 1.9774985828252392, "learning_rate": 1.746801282001473e-05, "loss": 0.611, "step": 7148 }, { "epoch": 0.2546983273064111, "grad_norm": 1.2288926164930634, "learning_rate": 1.7467245347869083e-05, "loss": 0.5448, "step": 7149 }, { "epoch": 0.2547339544329052, "grad_norm": 1.1060397975478482, "learning_rate": 1.7466477776291845e-05, "loss": 0.5225, "step": 7150 }, { "epoch": 0.2547695815593993, "grad_norm": 1.189420246989699, "learning_rate": 1.7465710105293238e-05, "loss": 0.5155, "step": 7151 }, { "epoch": 0.25480520868589346, "grad_norm": 2.517199629891233, "learning_rate": 1.746494233488348e-05, "loss": 0.5553, "step": 7152 }, { "epoch": 0.25484083581238753, "grad_norm": 1.3994802309553636, "learning_rate": 1.7464174465072798e-05, "loss": 0.5498, "step": 7153 }, { "epoch": 0.25487646293888166, "grad_norm": 1.7012909254559365, "learning_rate": 1.7463406495871416e-05, "loss": 0.581, "step": 7154 }, { "epoch": 0.2549120900653758, "grad_norm": 1.3017811117999165, "learning_rate": 1.746263842728956e-05, "loss": 0.5155, "step": 7155 }, { "epoch": 0.25494771719186987, "grad_norm": 1.9455639190751683, "learning_rate": 1.7461870259337457e-05, "loss": 0.5907, "step": 7156 }, { "epoch": 0.254983344318364, "grad_norm": 1.398475087773712, "learning_rate": 1.7461101992025334e-05, "loss": 0.56, "step": 7157 }, { "epoch": 0.25501897144485813, "grad_norm": 1.1799596367814404, "learning_rate": 1.746033362536343e-05, "loss": 0.5093, "step": 7158 }, { "epoch": 0.2550545985713522, "grad_norm": 1.2008991207866284, "learning_rate": 1.7459565159361963e-05, "loss": 0.5119, "step": 7159 }, { "epoch": 0.25509022569784634, "grad_norm": 1.173427253827198, "learning_rate": 1.745879659403117e-05, "loss": 0.4923, "step": 7160 }, { "epoch": 0.25512585282434047, "grad_norm": 1.7087863619244612, "learning_rate": 1.7458027929381292e-05, "loss": 0.6059, "step": 7161 }, { "epoch": 0.25516147995083455, "grad_norm": 1.1220858000420608, "learning_rate": 1.7457259165422557e-05, "loss": 0.5229, "step": 7162 }, { "epoch": 0.2551971070773287, "grad_norm": 1.9325756858314935, "learning_rate": 1.7456490302165208e-05, "loss": 0.6125, "step": 7163 }, { "epoch": 0.2552327342038228, "grad_norm": 1.2817654559168048, "learning_rate": 1.7455721339619477e-05, "loss": 0.5517, "step": 7164 }, { "epoch": 0.2552683613303169, "grad_norm": 1.0851697187633846, "learning_rate": 1.7454952277795602e-05, "loss": 0.5092, "step": 7165 }, { "epoch": 0.255303988456811, "grad_norm": 1.767105617695797, "learning_rate": 1.745418311670383e-05, "loss": 0.5959, "step": 7166 }, { "epoch": 0.25533961558330515, "grad_norm": 1.1766230152986497, "learning_rate": 1.7453413856354396e-05, "loss": 0.521, "step": 7167 }, { "epoch": 0.2553752427097992, "grad_norm": 1.1798503283626969, "learning_rate": 1.745264449675755e-05, "loss": 0.495, "step": 7168 }, { "epoch": 0.25541086983629335, "grad_norm": 1.8471391419114311, "learning_rate": 1.745187503792353e-05, "loss": 0.521, "step": 7169 }, { "epoch": 0.2554464969627875, "grad_norm": 1.9321998572142651, "learning_rate": 1.745110547986259e-05, "loss": 0.6777, "step": 7170 }, { "epoch": 0.25548212408928156, "grad_norm": 1.3784792309074567, "learning_rate": 1.7450335822584968e-05, "loss": 0.5483, "step": 7171 }, { "epoch": 0.2555177512157757, "grad_norm": 1.9687874759652324, "learning_rate": 1.744956606610092e-05, "loss": 0.5965, "step": 7172 }, { "epoch": 0.2555533783422698, "grad_norm": 2.007410318701521, "learning_rate": 1.744879621042069e-05, "loss": 0.6162, "step": 7173 }, { "epoch": 0.2555890054687639, "grad_norm": 1.8804471054846799, "learning_rate": 1.7448026255554537e-05, "loss": 0.5482, "step": 7174 }, { "epoch": 0.25562463259525803, "grad_norm": 1.6877044995166264, "learning_rate": 1.7447256201512707e-05, "loss": 0.5502, "step": 7175 }, { "epoch": 0.25566025972175216, "grad_norm": 1.0174239992303515, "learning_rate": 1.7446486048305454e-05, "loss": 0.5042, "step": 7176 }, { "epoch": 0.25569588684824623, "grad_norm": 1.6330984760941392, "learning_rate": 1.7445715795943033e-05, "loss": 0.6129, "step": 7177 }, { "epoch": 0.25573151397474037, "grad_norm": 1.146938420088567, "learning_rate": 1.7444945444435703e-05, "loss": 0.4981, "step": 7178 }, { "epoch": 0.2557671411012345, "grad_norm": 1.5798215044120147, "learning_rate": 1.744417499379372e-05, "loss": 0.5621, "step": 7179 }, { "epoch": 0.25580276822772857, "grad_norm": 1.2017350050471651, "learning_rate": 1.7443404444027347e-05, "loss": 0.5195, "step": 7180 }, { "epoch": 0.2558383953542227, "grad_norm": 1.0089033731752128, "learning_rate": 1.7442633795146838e-05, "loss": 0.5084, "step": 7181 }, { "epoch": 0.25587402248071683, "grad_norm": 1.252805149253771, "learning_rate": 1.744186304716246e-05, "loss": 0.5636, "step": 7182 }, { "epoch": 0.2559096496072109, "grad_norm": 1.100639385432991, "learning_rate": 1.7441092200084472e-05, "loss": 0.5198, "step": 7183 }, { "epoch": 0.25594527673370504, "grad_norm": 1.786949839708985, "learning_rate": 1.744032125392314e-05, "loss": 0.6155, "step": 7184 }, { "epoch": 0.25598090386019917, "grad_norm": 2.3405161691939655, "learning_rate": 1.7439550208688737e-05, "loss": 0.6107, "step": 7185 }, { "epoch": 0.25601653098669325, "grad_norm": 1.3123131794247864, "learning_rate": 1.7438779064391515e-05, "loss": 0.4749, "step": 7186 }, { "epoch": 0.2560521581131874, "grad_norm": 2.2626241756943193, "learning_rate": 1.7438007821041752e-05, "loss": 0.764, "step": 7187 }, { "epoch": 0.2560877852396815, "grad_norm": 1.0123661040284213, "learning_rate": 1.7437236478649718e-05, "loss": 0.4864, "step": 7188 }, { "epoch": 0.2561234123661756, "grad_norm": 1.2321186798685009, "learning_rate": 1.743646503722568e-05, "loss": 0.5002, "step": 7189 }, { "epoch": 0.2561590394926697, "grad_norm": 2.160199834170902, "learning_rate": 1.743569349677991e-05, "loss": 0.619, "step": 7190 }, { "epoch": 0.25619466661916385, "grad_norm": 1.6262742085965578, "learning_rate": 1.7434921857322692e-05, "loss": 0.5309, "step": 7191 }, { "epoch": 0.2562302937456579, "grad_norm": 1.1037506303192546, "learning_rate": 1.7434150118864286e-05, "loss": 0.5005, "step": 7192 }, { "epoch": 0.25626592087215205, "grad_norm": 1.1841675765423738, "learning_rate": 1.7433378281414975e-05, "loss": 0.4821, "step": 7193 }, { "epoch": 0.2563015479986462, "grad_norm": 1.6510537690554363, "learning_rate": 1.743260634498504e-05, "loss": 0.5774, "step": 7194 }, { "epoch": 0.25633717512514026, "grad_norm": 1.1277907202502993, "learning_rate": 1.7431834309584753e-05, "loss": 0.5436, "step": 7195 }, { "epoch": 0.2563728022516344, "grad_norm": 1.6339188252392873, "learning_rate": 1.7431062175224404e-05, "loss": 0.6446, "step": 7196 }, { "epoch": 0.2564084293781285, "grad_norm": 2.065192633572886, "learning_rate": 1.743028994191426e-05, "loss": 0.6798, "step": 7197 }, { "epoch": 0.2564440565046226, "grad_norm": 1.3704478038657886, "learning_rate": 1.7429517609664618e-05, "loss": 0.5277, "step": 7198 }, { "epoch": 0.25647968363111673, "grad_norm": 1.1505181440471048, "learning_rate": 1.7428745178485754e-05, "loss": 0.5366, "step": 7199 }, { "epoch": 0.25651531075761086, "grad_norm": 1.6151576683381415, "learning_rate": 1.7427972648387953e-05, "loss": 0.5995, "step": 7200 }, { "epoch": 0.25655093788410493, "grad_norm": 1.7099210753732161, "learning_rate": 1.7427200019381504e-05, "loss": 0.6091, "step": 7201 }, { "epoch": 0.25658656501059907, "grad_norm": 1.1308251382252288, "learning_rate": 1.74264272914767e-05, "loss": 0.5161, "step": 7202 }, { "epoch": 0.2566221921370932, "grad_norm": 1.5428180404495133, "learning_rate": 1.7425654464683822e-05, "loss": 0.6563, "step": 7203 }, { "epoch": 0.2566578192635873, "grad_norm": 1.1784920534095176, "learning_rate": 1.742488153901317e-05, "loss": 0.5215, "step": 7204 }, { "epoch": 0.2566934463900814, "grad_norm": 1.0960767064244208, "learning_rate": 1.7424108514475025e-05, "loss": 0.5229, "step": 7205 }, { "epoch": 0.25672907351657553, "grad_norm": 1.8616395041671085, "learning_rate": 1.742333539107969e-05, "loss": 0.6272, "step": 7206 }, { "epoch": 0.2567647006430696, "grad_norm": 1.3811054646662135, "learning_rate": 1.7422562168837452e-05, "loss": 0.4943, "step": 7207 }, { "epoch": 0.25680032776956374, "grad_norm": 1.629575454370856, "learning_rate": 1.742178884775861e-05, "loss": 0.5973, "step": 7208 }, { "epoch": 0.25683595489605787, "grad_norm": 1.6177957902879432, "learning_rate": 1.7421015427853468e-05, "loss": 0.5898, "step": 7209 }, { "epoch": 0.25687158202255195, "grad_norm": 1.2370084134987804, "learning_rate": 1.7420241909132312e-05, "loss": 0.5208, "step": 7210 }, { "epoch": 0.2569072091490461, "grad_norm": 1.1406626384635583, "learning_rate": 1.741946829160545e-05, "loss": 0.5328, "step": 7211 }, { "epoch": 0.2569428362755402, "grad_norm": 1.500180647109347, "learning_rate": 1.7418694575283185e-05, "loss": 0.6012, "step": 7212 }, { "epoch": 0.2569784634020343, "grad_norm": 1.1052649894093256, "learning_rate": 1.7417920760175817e-05, "loss": 0.4918, "step": 7213 }, { "epoch": 0.2570140905285284, "grad_norm": 1.5081428010270599, "learning_rate": 1.7417146846293646e-05, "loss": 0.6222, "step": 7214 }, { "epoch": 0.25704971765502255, "grad_norm": 1.2033098415799597, "learning_rate": 1.741637283364698e-05, "loss": 0.5325, "step": 7215 }, { "epoch": 0.2570853447815166, "grad_norm": 1.6344638098507325, "learning_rate": 1.7415598722246127e-05, "loss": 0.5777, "step": 7216 }, { "epoch": 0.25712097190801075, "grad_norm": 1.7133424053844255, "learning_rate": 1.7414824512101395e-05, "loss": 0.5867, "step": 7217 }, { "epoch": 0.2571565990345049, "grad_norm": 1.9815924736949633, "learning_rate": 1.7414050203223092e-05, "loss": 0.6302, "step": 7218 }, { "epoch": 0.25719222616099896, "grad_norm": 1.0781801697123155, "learning_rate": 1.7413275795621527e-05, "loss": 0.475, "step": 7219 }, { "epoch": 0.2572278532874931, "grad_norm": 1.614495878631688, "learning_rate": 1.7412501289307013e-05, "loss": 0.5982, "step": 7220 }, { "epoch": 0.2572634804139872, "grad_norm": 1.7848602335898063, "learning_rate": 1.7411726684289863e-05, "loss": 0.5853, "step": 7221 }, { "epoch": 0.2572991075404813, "grad_norm": 1.0921254619758745, "learning_rate": 1.7410951980580396e-05, "loss": 0.4444, "step": 7222 }, { "epoch": 0.25733473466697543, "grad_norm": 1.2518995903967711, "learning_rate": 1.7410177178188917e-05, "loss": 0.4962, "step": 7223 }, { "epoch": 0.25737036179346956, "grad_norm": 1.1425535935617743, "learning_rate": 1.7409402277125753e-05, "loss": 0.5411, "step": 7224 }, { "epoch": 0.25740598891996364, "grad_norm": 1.6286242031007805, "learning_rate": 1.740862727740122e-05, "loss": 0.6277, "step": 7225 }, { "epoch": 0.25744161604645777, "grad_norm": 2.2023179999440554, "learning_rate": 1.7407852179025636e-05, "loss": 0.5653, "step": 7226 }, { "epoch": 0.2574772431729519, "grad_norm": 1.9920486104059345, "learning_rate": 1.740707698200932e-05, "loss": 0.5567, "step": 7227 }, { "epoch": 0.257512870299446, "grad_norm": 1.2206488240227882, "learning_rate": 1.74063016863626e-05, "loss": 0.4742, "step": 7228 }, { "epoch": 0.2575484974259401, "grad_norm": 1.219450868896617, "learning_rate": 1.7405526292095794e-05, "loss": 0.5715, "step": 7229 }, { "epoch": 0.25758412455243423, "grad_norm": 2.242147937102166, "learning_rate": 1.7404750799219228e-05, "loss": 0.625, "step": 7230 }, { "epoch": 0.2576197516789283, "grad_norm": 1.6770270084701542, "learning_rate": 1.7403975207743233e-05, "loss": 0.6258, "step": 7231 }, { "epoch": 0.25765537880542244, "grad_norm": 1.6106519879268617, "learning_rate": 1.740319951767813e-05, "loss": 0.4934, "step": 7232 }, { "epoch": 0.2576910059319166, "grad_norm": 1.240546983676612, "learning_rate": 1.7402423729034252e-05, "loss": 0.54, "step": 7233 }, { "epoch": 0.25772663305841065, "grad_norm": 1.701975193659599, "learning_rate": 1.7401647841821928e-05, "loss": 0.5976, "step": 7234 }, { "epoch": 0.2577622601849048, "grad_norm": 1.1092630002052035, "learning_rate": 1.7400871856051492e-05, "loss": 0.5283, "step": 7235 }, { "epoch": 0.2577978873113989, "grad_norm": 1.0713398335615545, "learning_rate": 1.7400095771733273e-05, "loss": 0.4801, "step": 7236 }, { "epoch": 0.257833514437893, "grad_norm": 1.755278130379285, "learning_rate": 1.7399319588877603e-05, "loss": 0.5458, "step": 7237 }, { "epoch": 0.2578691415643871, "grad_norm": 1.7215594942883317, "learning_rate": 1.7398543307494824e-05, "loss": 0.5788, "step": 7238 }, { "epoch": 0.25790476869088125, "grad_norm": 1.7628013161743414, "learning_rate": 1.739776692759527e-05, "loss": 0.6251, "step": 7239 }, { "epoch": 0.2579403958173753, "grad_norm": 5.562235026548027, "learning_rate": 1.7396990449189275e-05, "loss": 0.6758, "step": 7240 }, { "epoch": 0.25797602294386945, "grad_norm": 1.58857202440849, "learning_rate": 1.7396213872287187e-05, "loss": 0.5893, "step": 7241 }, { "epoch": 0.2580116500703636, "grad_norm": 1.1308003927302845, "learning_rate": 1.7395437196899338e-05, "loss": 0.519, "step": 7242 }, { "epoch": 0.25804727719685766, "grad_norm": 1.1309035307635955, "learning_rate": 1.7394660423036076e-05, "loss": 0.4743, "step": 7243 }, { "epoch": 0.2580829043233518, "grad_norm": 1.619067582051819, "learning_rate": 1.739388355070774e-05, "loss": 0.5851, "step": 7244 }, { "epoch": 0.2581185314498459, "grad_norm": 1.0536803108240136, "learning_rate": 1.7393106579924676e-05, "loss": 0.5016, "step": 7245 }, { "epoch": 0.25815415857634, "grad_norm": 1.7649631833523916, "learning_rate": 1.7392329510697233e-05, "loss": 0.5805, "step": 7246 }, { "epoch": 0.25818978570283413, "grad_norm": 2.085859449396754, "learning_rate": 1.7391552343035753e-05, "loss": 0.5695, "step": 7247 }, { "epoch": 0.25822541282932826, "grad_norm": 1.3899118391223313, "learning_rate": 1.739077507695059e-05, "loss": 0.5061, "step": 7248 }, { "epoch": 0.25826103995582234, "grad_norm": 1.0339366767493905, "learning_rate": 1.738999771245209e-05, "loss": 0.4474, "step": 7249 }, { "epoch": 0.25829666708231647, "grad_norm": 1.7265108693673983, "learning_rate": 1.7389220249550605e-05, "loss": 0.5318, "step": 7250 }, { "epoch": 0.2583322942088106, "grad_norm": 1.491974926567358, "learning_rate": 1.7388442688256485e-05, "loss": 0.671, "step": 7251 }, { "epoch": 0.2583679213353047, "grad_norm": 1.1273144414403586, "learning_rate": 1.7387665028580088e-05, "loss": 0.4714, "step": 7252 }, { "epoch": 0.2584035484617988, "grad_norm": 1.6869140631949258, "learning_rate": 1.7386887270531768e-05, "loss": 0.5367, "step": 7253 }, { "epoch": 0.25843917558829294, "grad_norm": 1.2989658675656732, "learning_rate": 1.7386109414121883e-05, "loss": 0.5575, "step": 7254 }, { "epoch": 0.25847480271478707, "grad_norm": 1.4280518522863856, "learning_rate": 1.7385331459360786e-05, "loss": 0.5817, "step": 7255 }, { "epoch": 0.25851042984128114, "grad_norm": 1.1000054597647502, "learning_rate": 1.7384553406258842e-05, "loss": 0.4465, "step": 7256 }, { "epoch": 0.2585460569677753, "grad_norm": 1.6328732069404261, "learning_rate": 1.7383775254826402e-05, "loss": 0.5437, "step": 7257 }, { "epoch": 0.2585816840942694, "grad_norm": 1.085014028465867, "learning_rate": 1.7382997005073837e-05, "loss": 0.4858, "step": 7258 }, { "epoch": 0.2586173112207635, "grad_norm": 1.870546667879151, "learning_rate": 1.7382218657011507e-05, "loss": 0.6334, "step": 7259 }, { "epoch": 0.2586529383472576, "grad_norm": 1.4911777731330786, "learning_rate": 1.7381440210649774e-05, "loss": 0.5815, "step": 7260 }, { "epoch": 0.25868856547375174, "grad_norm": 1.6108121970341527, "learning_rate": 1.7380661665999006e-05, "loss": 0.5951, "step": 7261 }, { "epoch": 0.2587241926002458, "grad_norm": 3.8312092892148217, "learning_rate": 1.737988302306957e-05, "loss": 0.5502, "step": 7262 }, { "epoch": 0.25875981972673995, "grad_norm": 1.1923809580548994, "learning_rate": 1.737910428187183e-05, "loss": 0.5714, "step": 7263 }, { "epoch": 0.2587954468532341, "grad_norm": 1.8851218815985253, "learning_rate": 1.7378325442416165e-05, "loss": 0.6291, "step": 7264 }, { "epoch": 0.25883107397972815, "grad_norm": 1.505690809109452, "learning_rate": 1.7377546504712933e-05, "loss": 0.5789, "step": 7265 }, { "epoch": 0.2588667011062223, "grad_norm": 3.2145142948436685, "learning_rate": 1.7376767468772518e-05, "loss": 0.624, "step": 7266 }, { "epoch": 0.2589023282327164, "grad_norm": 1.097354498706909, "learning_rate": 1.7375988334605283e-05, "loss": 0.4631, "step": 7267 }, { "epoch": 0.2589379553592105, "grad_norm": 2.112073868288962, "learning_rate": 1.7375209102221614e-05, "loss": 0.5225, "step": 7268 }, { "epoch": 0.2589735824857046, "grad_norm": 1.1229307242724729, "learning_rate": 1.7374429771631877e-05, "loss": 0.544, "step": 7269 }, { "epoch": 0.25900920961219875, "grad_norm": 1.176242054056881, "learning_rate": 1.737365034284645e-05, "loss": 0.5044, "step": 7270 }, { "epoch": 0.25904483673869283, "grad_norm": 1.1464128793625736, "learning_rate": 1.7372870815875723e-05, "loss": 0.5246, "step": 7271 }, { "epoch": 0.25908046386518696, "grad_norm": 1.7700977015382495, "learning_rate": 1.7372091190730062e-05, "loss": 0.5321, "step": 7272 }, { "epoch": 0.2591160909916811, "grad_norm": 1.0635105152038362, "learning_rate": 1.7371311467419856e-05, "loss": 0.4623, "step": 7273 }, { "epoch": 0.25915171811817517, "grad_norm": 1.024188199748261, "learning_rate": 1.7370531645955485e-05, "loss": 0.4853, "step": 7274 }, { "epoch": 0.2591873452446693, "grad_norm": 1.1149521183154636, "learning_rate": 1.736975172634733e-05, "loss": 0.5284, "step": 7275 }, { "epoch": 0.25922297237116343, "grad_norm": 2.1348466265094506, "learning_rate": 1.7368971708605785e-05, "loss": 0.6393, "step": 7276 }, { "epoch": 0.2592585994976575, "grad_norm": 1.6742342649053301, "learning_rate": 1.736819159274123e-05, "loss": 0.626, "step": 7277 }, { "epoch": 0.25929422662415164, "grad_norm": 1.767503301385463, "learning_rate": 1.736741137876405e-05, "loss": 0.6142, "step": 7278 }, { "epoch": 0.25932985375064577, "grad_norm": 1.074185291142221, "learning_rate": 1.736663106668464e-05, "loss": 0.4707, "step": 7279 }, { "epoch": 0.25936548087713984, "grad_norm": 1.11975620598259, "learning_rate": 1.736585065651339e-05, "loss": 0.5093, "step": 7280 }, { "epoch": 0.259401108003634, "grad_norm": 2.289214423498096, "learning_rate": 1.736507014826069e-05, "loss": 0.6049, "step": 7281 }, { "epoch": 0.2594367351301281, "grad_norm": 1.8258456063484796, "learning_rate": 1.736428954193693e-05, "loss": 0.6091, "step": 7282 }, { "epoch": 0.2594723622566222, "grad_norm": 2.1529466519843123, "learning_rate": 1.7363508837552513e-05, "loss": 0.65, "step": 7283 }, { "epoch": 0.2595079893831163, "grad_norm": 1.8123021511057653, "learning_rate": 1.7362728035117826e-05, "loss": 0.6075, "step": 7284 }, { "epoch": 0.25954361650961044, "grad_norm": 1.4171380765246524, "learning_rate": 1.7361947134643265e-05, "loss": 0.5023, "step": 7285 }, { "epoch": 0.2595792436361045, "grad_norm": 1.7262893763191618, "learning_rate": 1.7361166136139238e-05, "loss": 0.6558, "step": 7286 }, { "epoch": 0.25961487076259865, "grad_norm": 1.8023506974912276, "learning_rate": 1.7360385039616137e-05, "loss": 0.6591, "step": 7287 }, { "epoch": 0.2596504978890928, "grad_norm": 1.168874045122674, "learning_rate": 1.7359603845084365e-05, "loss": 0.5436, "step": 7288 }, { "epoch": 0.25968612501558686, "grad_norm": 1.804596223625835, "learning_rate": 1.735882255255432e-05, "loss": 0.6188, "step": 7289 }, { "epoch": 0.259721752142081, "grad_norm": 2.5242388904448556, "learning_rate": 1.7358041162036414e-05, "loss": 0.6124, "step": 7290 }, { "epoch": 0.2597573792685751, "grad_norm": 1.7265925933101876, "learning_rate": 1.7357259673541045e-05, "loss": 0.5848, "step": 7291 }, { "epoch": 0.2597930063950692, "grad_norm": 1.0544898020887585, "learning_rate": 1.7356478087078625e-05, "loss": 0.4863, "step": 7292 }, { "epoch": 0.2598286335215633, "grad_norm": 1.0534402058396182, "learning_rate": 1.735569640265955e-05, "loss": 0.4585, "step": 7293 }, { "epoch": 0.25986426064805745, "grad_norm": 1.1194814785741498, "learning_rate": 1.735491462029424e-05, "loss": 0.5144, "step": 7294 }, { "epoch": 0.25989988777455153, "grad_norm": 1.1062183972638449, "learning_rate": 1.73541327399931e-05, "loss": 0.4792, "step": 7295 }, { "epoch": 0.25993551490104566, "grad_norm": 2.2156624540640957, "learning_rate": 1.7353350761766544e-05, "loss": 0.5815, "step": 7296 }, { "epoch": 0.2599711420275398, "grad_norm": 1.7371224572759045, "learning_rate": 1.735256868562498e-05, "loss": 0.601, "step": 7297 }, { "epoch": 0.26000676915403387, "grad_norm": 1.1192079273009827, "learning_rate": 1.7351786511578827e-05, "loss": 0.4968, "step": 7298 }, { "epoch": 0.260042396280528, "grad_norm": 1.0922375663354071, "learning_rate": 1.73510042396385e-05, "loss": 0.5344, "step": 7299 }, { "epoch": 0.26007802340702213, "grad_norm": 1.6199639219948676, "learning_rate": 1.7350221869814413e-05, "loss": 0.7348, "step": 7300 }, { "epoch": 0.2601136505335162, "grad_norm": 1.7124293431719901, "learning_rate": 1.7349439402116983e-05, "loss": 0.6649, "step": 7301 }, { "epoch": 0.26014927766001034, "grad_norm": 1.4231358152539015, "learning_rate": 1.734865683655663e-05, "loss": 0.6889, "step": 7302 }, { "epoch": 0.26018490478650447, "grad_norm": 1.6162462592718663, "learning_rate": 1.7347874173143773e-05, "loss": 0.615, "step": 7303 }, { "epoch": 0.26022053191299854, "grad_norm": 1.2030390743582204, "learning_rate": 1.734709141188884e-05, "loss": 0.4997, "step": 7304 }, { "epoch": 0.2602561590394927, "grad_norm": 1.160893349963275, "learning_rate": 1.7346308552802245e-05, "loss": 0.5067, "step": 7305 }, { "epoch": 0.2602917861659868, "grad_norm": 1.8253430522649619, "learning_rate": 1.7345525595894418e-05, "loss": 0.6729, "step": 7306 }, { "epoch": 0.2603274132924809, "grad_norm": 1.4543355842511354, "learning_rate": 1.7344742541175782e-05, "loss": 0.5711, "step": 7307 }, { "epoch": 0.260363040418975, "grad_norm": 1.6419333412393375, "learning_rate": 1.734395938865677e-05, "loss": 0.5591, "step": 7308 }, { "epoch": 0.26039866754546914, "grad_norm": 2.19198631379811, "learning_rate": 1.7343176138347803e-05, "loss": 0.6146, "step": 7309 }, { "epoch": 0.2604342946719632, "grad_norm": 1.7257710474920989, "learning_rate": 1.7342392790259314e-05, "loss": 0.6511, "step": 7310 }, { "epoch": 0.26046992179845735, "grad_norm": 1.026808922415933, "learning_rate": 1.734160934440173e-05, "loss": 0.4951, "step": 7311 }, { "epoch": 0.2605055489249515, "grad_norm": 1.257889364462615, "learning_rate": 1.7340825800785494e-05, "loss": 0.5423, "step": 7312 }, { "epoch": 0.26054117605144556, "grad_norm": 1.1735003429300366, "learning_rate": 1.734004215942102e-05, "loss": 0.5488, "step": 7313 }, { "epoch": 0.2605768031779397, "grad_norm": 3.392720409722902, "learning_rate": 1.7339258420318765e-05, "loss": 0.6913, "step": 7314 }, { "epoch": 0.2606124303044338, "grad_norm": 1.446293498311628, "learning_rate": 1.7338474583489148e-05, "loss": 0.5499, "step": 7315 }, { "epoch": 0.2606480574309279, "grad_norm": 1.125817158961674, "learning_rate": 1.7337690648942615e-05, "loss": 0.4807, "step": 7316 }, { "epoch": 0.260683684557422, "grad_norm": 1.687596744320451, "learning_rate": 1.7336906616689598e-05, "loss": 0.5996, "step": 7317 }, { "epoch": 0.26071931168391616, "grad_norm": 1.2728994964142928, "learning_rate": 1.7336122486740546e-05, "loss": 0.5315, "step": 7318 }, { "epoch": 0.26075493881041023, "grad_norm": 1.1577449821460397, "learning_rate": 1.7335338259105895e-05, "loss": 0.5031, "step": 7319 }, { "epoch": 0.26079056593690436, "grad_norm": 1.1422738585964822, "learning_rate": 1.7334553933796087e-05, "loss": 0.4818, "step": 7320 }, { "epoch": 0.2608261930633985, "grad_norm": 1.0871653983975111, "learning_rate": 1.7333769510821567e-05, "loss": 0.4764, "step": 7321 }, { "epoch": 0.26086182018989257, "grad_norm": 1.9851911514624154, "learning_rate": 1.7332984990192782e-05, "loss": 0.6385, "step": 7322 }, { "epoch": 0.2608974473163867, "grad_norm": 1.1973968870872431, "learning_rate": 1.7332200371920173e-05, "loss": 0.5289, "step": 7323 }, { "epoch": 0.26093307444288083, "grad_norm": 1.5124793236388767, "learning_rate": 1.7331415656014196e-05, "loss": 0.5488, "step": 7324 }, { "epoch": 0.2609687015693749, "grad_norm": 2.192133680426303, "learning_rate": 1.7330630842485287e-05, "loss": 0.6831, "step": 7325 }, { "epoch": 0.26100432869586904, "grad_norm": 1.186432701681829, "learning_rate": 1.732984593134391e-05, "loss": 0.4548, "step": 7326 }, { "epoch": 0.26103995582236317, "grad_norm": 1.1725786283799717, "learning_rate": 1.732906092260051e-05, "loss": 0.5073, "step": 7327 }, { "epoch": 0.26107558294885724, "grad_norm": 1.7457871492009127, "learning_rate": 1.7328275816265542e-05, "loss": 0.6325, "step": 7328 }, { "epoch": 0.2611112100753514, "grad_norm": 1.851658582745822, "learning_rate": 1.732749061234946e-05, "loss": 0.4961, "step": 7329 }, { "epoch": 0.2611468372018455, "grad_norm": 1.423458966106117, "learning_rate": 1.7326705310862716e-05, "loss": 0.6057, "step": 7330 }, { "epoch": 0.2611824643283396, "grad_norm": 1.0919891216458752, "learning_rate": 1.7325919911815774e-05, "loss": 0.4851, "step": 7331 }, { "epoch": 0.2612180914548337, "grad_norm": 1.0888234379728432, "learning_rate": 1.7325134415219085e-05, "loss": 0.4974, "step": 7332 }, { "epoch": 0.26125371858132784, "grad_norm": 1.6015177004328462, "learning_rate": 1.732434882108311e-05, "loss": 0.663, "step": 7333 }, { "epoch": 0.2612893457078219, "grad_norm": 1.231516773503468, "learning_rate": 1.7323563129418317e-05, "loss": 0.4915, "step": 7334 }, { "epoch": 0.26132497283431605, "grad_norm": 2.1701974494314564, "learning_rate": 1.7322777340235154e-05, "loss": 0.6247, "step": 7335 }, { "epoch": 0.2613605999608102, "grad_norm": 2.168501022156476, "learning_rate": 1.7321991453544096e-05, "loss": 0.5587, "step": 7336 }, { "epoch": 0.26139622708730426, "grad_norm": 1.6667659642210346, "learning_rate": 1.7321205469355603e-05, "loss": 0.7214, "step": 7337 }, { "epoch": 0.2614318542137984, "grad_norm": 2.1705158941116176, "learning_rate": 1.7320419387680142e-05, "loss": 0.5968, "step": 7338 }, { "epoch": 0.2614674813402925, "grad_norm": 1.4256130684767254, "learning_rate": 1.7319633208528185e-05, "loss": 0.5431, "step": 7339 }, { "epoch": 0.2615031084667866, "grad_norm": 1.2142903655665964, "learning_rate": 1.7318846931910192e-05, "loss": 0.5231, "step": 7340 }, { "epoch": 0.2615387355932807, "grad_norm": 1.2051219830778692, "learning_rate": 1.7318060557836634e-05, "loss": 0.5629, "step": 7341 }, { "epoch": 0.26157436271977486, "grad_norm": 1.2016498677290495, "learning_rate": 1.731727408631799e-05, "loss": 0.5648, "step": 7342 }, { "epoch": 0.26160998984626893, "grad_norm": 1.1416362004127871, "learning_rate": 1.7316487517364723e-05, "loss": 0.5149, "step": 7343 }, { "epoch": 0.26164561697276306, "grad_norm": 1.1512348335330616, "learning_rate": 1.731570085098731e-05, "loss": 0.4538, "step": 7344 }, { "epoch": 0.2616812440992572, "grad_norm": 1.5279238630816618, "learning_rate": 1.7314914087196232e-05, "loss": 0.62, "step": 7345 }, { "epoch": 0.26171687122575127, "grad_norm": 2.1620700677572287, "learning_rate": 1.7314127226001954e-05, "loss": 0.6553, "step": 7346 }, { "epoch": 0.2617524983522454, "grad_norm": 1.5093087012735313, "learning_rate": 1.7313340267414964e-05, "loss": 0.5625, "step": 7347 }, { "epoch": 0.26178812547873953, "grad_norm": 1.0759384373340297, "learning_rate": 1.7312553211445738e-05, "loss": 0.536, "step": 7348 }, { "epoch": 0.2618237526052336, "grad_norm": 1.7131939074018843, "learning_rate": 1.7311766058104752e-05, "loss": 0.5821, "step": 7349 }, { "epoch": 0.26185937973172774, "grad_norm": 1.7233483193487629, "learning_rate": 1.731097880740249e-05, "loss": 0.6308, "step": 7350 }, { "epoch": 0.26189500685822187, "grad_norm": 2.1568711396665257, "learning_rate": 1.731019145934943e-05, "loss": 0.643, "step": 7351 }, { "epoch": 0.26193063398471594, "grad_norm": 1.5907939806761715, "learning_rate": 1.730940401395607e-05, "loss": 0.6455, "step": 7352 }, { "epoch": 0.2619662611112101, "grad_norm": 1.7107374911965831, "learning_rate": 1.7308616471232885e-05, "loss": 0.6341, "step": 7353 }, { "epoch": 0.2620018882377042, "grad_norm": 1.1156210847794072, "learning_rate": 1.7307828831190362e-05, "loss": 0.4953, "step": 7354 }, { "epoch": 0.2620375153641983, "grad_norm": 1.1020802227728737, "learning_rate": 1.7307041093838986e-05, "loss": 0.484, "step": 7355 }, { "epoch": 0.2620731424906924, "grad_norm": 0.9787805620799814, "learning_rate": 1.7306253259189255e-05, "loss": 0.5068, "step": 7356 }, { "epoch": 0.26210876961718654, "grad_norm": 1.6132563574937109, "learning_rate": 1.7305465327251657e-05, "loss": 0.5656, "step": 7357 }, { "epoch": 0.2621443967436806, "grad_norm": 2.0598325138613474, "learning_rate": 1.730467729803668e-05, "loss": 0.6269, "step": 7358 }, { "epoch": 0.26218002387017475, "grad_norm": 1.8503556773047682, "learning_rate": 1.7303889171554816e-05, "loss": 0.5889, "step": 7359 }, { "epoch": 0.2622156509966689, "grad_norm": 1.8853921507067535, "learning_rate": 1.7303100947816567e-05, "loss": 0.5183, "step": 7360 }, { "epoch": 0.26225127812316296, "grad_norm": 1.957281704529437, "learning_rate": 1.7302312626832423e-05, "loss": 0.6578, "step": 7361 }, { "epoch": 0.2622869052496571, "grad_norm": 1.10346031497902, "learning_rate": 1.730152420861288e-05, "loss": 0.4744, "step": 7362 }, { "epoch": 0.2623225323761512, "grad_norm": 1.5018726509400568, "learning_rate": 1.730073569316844e-05, "loss": 0.629, "step": 7363 }, { "epoch": 0.2623581595026453, "grad_norm": 1.5342726629204415, "learning_rate": 1.7299947080509603e-05, "loss": 0.625, "step": 7364 }, { "epoch": 0.2623937866291394, "grad_norm": 1.1216348866988912, "learning_rate": 1.7299158370646866e-05, "loss": 0.4718, "step": 7365 }, { "epoch": 0.26242941375563356, "grad_norm": 1.0558275129670027, "learning_rate": 1.7298369563590734e-05, "loss": 0.5045, "step": 7366 }, { "epoch": 0.26246504088212763, "grad_norm": 1.1080327740784732, "learning_rate": 1.729758065935171e-05, "loss": 0.5051, "step": 7367 }, { "epoch": 0.26250066800862176, "grad_norm": 1.0807720770064928, "learning_rate": 1.72967916579403e-05, "loss": 0.5447, "step": 7368 }, { "epoch": 0.2625362951351159, "grad_norm": 1.1272682086604582, "learning_rate": 1.729600255936701e-05, "loss": 0.5379, "step": 7369 }, { "epoch": 0.26257192226160997, "grad_norm": 0.9952543968941041, "learning_rate": 1.7295213363642346e-05, "loss": 0.5023, "step": 7370 }, { "epoch": 0.2626075493881041, "grad_norm": 1.5643134139678179, "learning_rate": 1.7294424070776813e-05, "loss": 0.644, "step": 7371 }, { "epoch": 0.26264317651459823, "grad_norm": 1.1223462769535824, "learning_rate": 1.729363468078093e-05, "loss": 0.497, "step": 7372 }, { "epoch": 0.2626788036410923, "grad_norm": 1.465805895062004, "learning_rate": 1.72928451936652e-05, "loss": 0.571, "step": 7373 }, { "epoch": 0.26271443076758644, "grad_norm": 1.7590583288831014, "learning_rate": 1.7292055609440138e-05, "loss": 0.6123, "step": 7374 }, { "epoch": 0.26275005789408057, "grad_norm": 1.187846405139529, "learning_rate": 1.729126592811626e-05, "loss": 0.5083, "step": 7375 }, { "epoch": 0.26278568502057464, "grad_norm": 1.1714562832170436, "learning_rate": 1.7290476149704082e-05, "loss": 0.4921, "step": 7376 }, { "epoch": 0.2628213121470688, "grad_norm": 1.5259152906128148, "learning_rate": 1.7289686274214116e-05, "loss": 0.6312, "step": 7377 }, { "epoch": 0.2628569392735629, "grad_norm": 1.1582157895290515, "learning_rate": 1.7288896301656885e-05, "loss": 0.5341, "step": 7378 }, { "epoch": 0.262892566400057, "grad_norm": 2.142160252446372, "learning_rate": 1.72881062320429e-05, "loss": 0.6515, "step": 7379 }, { "epoch": 0.2629281935265511, "grad_norm": 1.2067149020012555, "learning_rate": 1.728731606538269e-05, "loss": 0.5161, "step": 7380 }, { "epoch": 0.26296382065304524, "grad_norm": 1.4019520657443354, "learning_rate": 1.7286525801686774e-05, "loss": 0.6534, "step": 7381 }, { "epoch": 0.2629994477795393, "grad_norm": 1.1172900796148524, "learning_rate": 1.7285735440965673e-05, "loss": 0.5187, "step": 7382 }, { "epoch": 0.26303507490603345, "grad_norm": 1.1976168429558824, "learning_rate": 1.728494498322991e-05, "loss": 0.5481, "step": 7383 }, { "epoch": 0.2630707020325276, "grad_norm": 1.576766741168938, "learning_rate": 1.7284154428490014e-05, "loss": 0.635, "step": 7384 }, { "epoch": 0.26310632915902166, "grad_norm": 1.1587921028417605, "learning_rate": 1.728336377675651e-05, "loss": 0.5165, "step": 7385 }, { "epoch": 0.2631419562855158, "grad_norm": 2.148411737212118, "learning_rate": 1.7282573028039927e-05, "loss": 0.6253, "step": 7386 }, { "epoch": 0.2631775834120099, "grad_norm": 1.1216926801760865, "learning_rate": 1.7281782182350796e-05, "loss": 0.4724, "step": 7387 }, { "epoch": 0.263213210538504, "grad_norm": 1.4072793029564015, "learning_rate": 1.7280991239699643e-05, "loss": 0.5973, "step": 7388 }, { "epoch": 0.2632488376649981, "grad_norm": 1.4051745332924575, "learning_rate": 1.7280200200097003e-05, "loss": 0.5668, "step": 7389 }, { "epoch": 0.26328446479149226, "grad_norm": 1.842019085832085, "learning_rate": 1.727940906355341e-05, "loss": 0.6791, "step": 7390 }, { "epoch": 0.26332009191798633, "grad_norm": 1.2583245623655057, "learning_rate": 1.7278617830079394e-05, "loss": 0.5171, "step": 7391 }, { "epoch": 0.26335571904448046, "grad_norm": 1.5771931900876222, "learning_rate": 1.72778264996855e-05, "loss": 0.6171, "step": 7392 }, { "epoch": 0.2633913461709746, "grad_norm": 1.3275628429230422, "learning_rate": 1.727703507238225e-05, "loss": 0.6643, "step": 7393 }, { "epoch": 0.26342697329746867, "grad_norm": 1.5105245547203048, "learning_rate": 1.72762435481802e-05, "loss": 0.5607, "step": 7394 }, { "epoch": 0.2634626004239628, "grad_norm": 1.2718531716650634, "learning_rate": 1.727545192708988e-05, "loss": 0.5518, "step": 7395 }, { "epoch": 0.26349822755045693, "grad_norm": 2.2155763884722974, "learning_rate": 1.727466020912183e-05, "loss": 0.6381, "step": 7396 }, { "epoch": 0.263533854676951, "grad_norm": 1.5064901637139487, "learning_rate": 1.7273868394286596e-05, "loss": 0.6209, "step": 7397 }, { "epoch": 0.26356948180344514, "grad_norm": 1.7099599754196184, "learning_rate": 1.727307648259472e-05, "loss": 0.5825, "step": 7398 }, { "epoch": 0.26360510892993927, "grad_norm": 1.1724390092185824, "learning_rate": 1.7272284474056748e-05, "loss": 0.5422, "step": 7399 }, { "epoch": 0.26364073605643334, "grad_norm": 1.703327918573311, "learning_rate": 1.7271492368683224e-05, "loss": 0.6382, "step": 7400 }, { "epoch": 0.2636763631829275, "grad_norm": 1.0500123010903883, "learning_rate": 1.72707001664847e-05, "loss": 0.5033, "step": 7401 }, { "epoch": 0.2637119903094216, "grad_norm": 1.9892896711164336, "learning_rate": 1.7269907867471715e-05, "loss": 0.575, "step": 7402 }, { "epoch": 0.2637476174359157, "grad_norm": 1.3088747476691232, "learning_rate": 1.7269115471654828e-05, "loss": 0.5375, "step": 7403 }, { "epoch": 0.2637832445624098, "grad_norm": 1.7805876182652707, "learning_rate": 1.726832297904459e-05, "loss": 0.5993, "step": 7404 }, { "epoch": 0.26381887168890394, "grad_norm": 1.190634501313769, "learning_rate": 1.7267530389651546e-05, "loss": 0.4793, "step": 7405 }, { "epoch": 0.263854498815398, "grad_norm": 1.173798628577957, "learning_rate": 1.7266737703486263e-05, "loss": 0.5098, "step": 7406 }, { "epoch": 0.26389012594189215, "grad_norm": 1.5243246062872229, "learning_rate": 1.726594492055928e-05, "loss": 0.6063, "step": 7407 }, { "epoch": 0.2639257530683863, "grad_norm": 1.7776342222481092, "learning_rate": 1.726515204088117e-05, "loss": 0.5684, "step": 7408 }, { "epoch": 0.26396138019488036, "grad_norm": 1.2678331124873508, "learning_rate": 1.7264359064462478e-05, "loss": 0.5299, "step": 7409 }, { "epoch": 0.2639970073213745, "grad_norm": 1.9038948386043166, "learning_rate": 1.7263565991313767e-05, "loss": 0.6317, "step": 7410 }, { "epoch": 0.2640326344478686, "grad_norm": 1.6972117906644861, "learning_rate": 1.72627728214456e-05, "loss": 0.6367, "step": 7411 }, { "epoch": 0.2640682615743627, "grad_norm": 1.1864484402823903, "learning_rate": 1.7261979554868533e-05, "loss": 0.4935, "step": 7412 }, { "epoch": 0.2641038887008568, "grad_norm": 2.0219639096975914, "learning_rate": 1.7261186191593135e-05, "loss": 0.5496, "step": 7413 }, { "epoch": 0.26413951582735096, "grad_norm": 1.7029562830756555, "learning_rate": 1.726039273162997e-05, "loss": 0.5903, "step": 7414 }, { "epoch": 0.26417514295384503, "grad_norm": 1.2417590328821757, "learning_rate": 1.7259599174989598e-05, "loss": 0.5051, "step": 7415 }, { "epoch": 0.26421077008033916, "grad_norm": 2.3471489871152826, "learning_rate": 1.725880552168259e-05, "loss": 0.5771, "step": 7416 }, { "epoch": 0.2642463972068333, "grad_norm": 1.5291669140413575, "learning_rate": 1.725801177171951e-05, "loss": 0.649, "step": 7417 }, { "epoch": 0.26428202433332737, "grad_norm": 1.4792240133225882, "learning_rate": 1.7257217925110934e-05, "loss": 0.5875, "step": 7418 }, { "epoch": 0.2643176514598215, "grad_norm": 1.7467999675688353, "learning_rate": 1.7256423981867425e-05, "loss": 0.6648, "step": 7419 }, { "epoch": 0.26435327858631563, "grad_norm": 1.5843896760042566, "learning_rate": 1.7255629941999565e-05, "loss": 0.5958, "step": 7420 }, { "epoch": 0.2643889057128097, "grad_norm": 1.9065964766971673, "learning_rate": 1.7254835805517913e-05, "loss": 0.6133, "step": 7421 }, { "epoch": 0.26442453283930384, "grad_norm": 1.2268296029446504, "learning_rate": 1.7254041572433058e-05, "loss": 0.5195, "step": 7422 }, { "epoch": 0.26446015996579797, "grad_norm": 1.5883889347193376, "learning_rate": 1.7253247242755565e-05, "loss": 0.6676, "step": 7423 }, { "epoch": 0.26449578709229205, "grad_norm": 1.1200706476943267, "learning_rate": 1.7252452816496017e-05, "loss": 0.4948, "step": 7424 }, { "epoch": 0.2645314142187862, "grad_norm": 2.1203635821516147, "learning_rate": 1.725165829366499e-05, "loss": 0.6443, "step": 7425 }, { "epoch": 0.2645670413452803, "grad_norm": 1.5546476886195608, "learning_rate": 1.7250863674273063e-05, "loss": 0.5532, "step": 7426 }, { "epoch": 0.2646026684717744, "grad_norm": 1.2006988623578478, "learning_rate": 1.7250068958330822e-05, "loss": 0.494, "step": 7427 }, { "epoch": 0.2646382955982685, "grad_norm": 1.1989522298864557, "learning_rate": 1.724927414584884e-05, "loss": 0.5572, "step": 7428 }, { "epoch": 0.26467392272476264, "grad_norm": 1.0945931479210074, "learning_rate": 1.724847923683771e-05, "loss": 0.5043, "step": 7429 }, { "epoch": 0.2647095498512567, "grad_norm": 1.1001070669390043, "learning_rate": 1.7247684231308012e-05, "loss": 0.4872, "step": 7430 }, { "epoch": 0.26474517697775085, "grad_norm": 1.07242778602301, "learning_rate": 1.7246889129270332e-05, "loss": 0.5234, "step": 7431 }, { "epoch": 0.264780804104245, "grad_norm": 1.7717098328061298, "learning_rate": 1.724609393073526e-05, "loss": 0.6499, "step": 7432 }, { "epoch": 0.26481643123073906, "grad_norm": 1.6962909481117532, "learning_rate": 1.724529863571338e-05, "loss": 0.6633, "step": 7433 }, { "epoch": 0.2648520583572332, "grad_norm": 1.1688844743676285, "learning_rate": 1.7244503244215285e-05, "loss": 0.509, "step": 7434 }, { "epoch": 0.2648876854837273, "grad_norm": 2.168310568742621, "learning_rate": 1.7243707756251566e-05, "loss": 0.6599, "step": 7435 }, { "epoch": 0.2649233126102214, "grad_norm": 1.1851449383742407, "learning_rate": 1.7242912171832814e-05, "loss": 0.4724, "step": 7436 }, { "epoch": 0.2649589397367155, "grad_norm": 1.083739664047341, "learning_rate": 1.7242116490969623e-05, "loss": 0.4484, "step": 7437 }, { "epoch": 0.26499456686320966, "grad_norm": 1.9136100049486167, "learning_rate": 1.7241320713672592e-05, "loss": 0.7276, "step": 7438 }, { "epoch": 0.26503019398970373, "grad_norm": 2.0231964083757195, "learning_rate": 1.7240524839952315e-05, "loss": 0.6341, "step": 7439 }, { "epoch": 0.26506582111619786, "grad_norm": 1.486061381026872, "learning_rate": 1.7239728869819384e-05, "loss": 0.5412, "step": 7440 }, { "epoch": 0.265101448242692, "grad_norm": 1.1448054063881334, "learning_rate": 1.723893280328441e-05, "loss": 0.4576, "step": 7441 }, { "epoch": 0.2651370753691861, "grad_norm": 1.4702299997521038, "learning_rate": 1.7238136640357977e-05, "loss": 0.5213, "step": 7442 }, { "epoch": 0.2651727024956802, "grad_norm": 1.9299957953353413, "learning_rate": 1.72373403810507e-05, "loss": 0.5701, "step": 7443 }, { "epoch": 0.26520832962217433, "grad_norm": 1.0212472296361865, "learning_rate": 1.723654402537318e-05, "loss": 0.5102, "step": 7444 }, { "epoch": 0.26524395674866846, "grad_norm": 2.1132751585998997, "learning_rate": 1.7235747573336016e-05, "loss": 0.6954, "step": 7445 }, { "epoch": 0.26527958387516254, "grad_norm": 2.0253338414465523, "learning_rate": 1.723495102494982e-05, "loss": 0.6989, "step": 7446 }, { "epoch": 0.26531521100165667, "grad_norm": 1.8285732587283126, "learning_rate": 1.7234154380225185e-05, "loss": 0.6676, "step": 7447 }, { "epoch": 0.2653508381281508, "grad_norm": 1.1340587881008406, "learning_rate": 1.7233357639172735e-05, "loss": 0.5118, "step": 7448 }, { "epoch": 0.2653864652546449, "grad_norm": 1.1262409868196501, "learning_rate": 1.7232560801803067e-05, "loss": 0.5023, "step": 7449 }, { "epoch": 0.265422092381139, "grad_norm": 1.5776978574188398, "learning_rate": 1.7231763868126804e-05, "loss": 0.6152, "step": 7450 }, { "epoch": 0.26545771950763314, "grad_norm": 1.5657858881001323, "learning_rate": 1.7230966838154543e-05, "loss": 0.6042, "step": 7451 }, { "epoch": 0.2654933466341272, "grad_norm": 2.0964767917048928, "learning_rate": 1.723016971189691e-05, "loss": 0.5537, "step": 7452 }, { "epoch": 0.26552897376062135, "grad_norm": 1.992359417106728, "learning_rate": 1.722937248936451e-05, "loss": 0.6045, "step": 7453 }, { "epoch": 0.2655646008871155, "grad_norm": 1.1696933434988852, "learning_rate": 1.722857517056796e-05, "loss": 0.4792, "step": 7454 }, { "epoch": 0.26560022801360955, "grad_norm": 1.435570623398564, "learning_rate": 1.7227777755517886e-05, "loss": 0.6452, "step": 7455 }, { "epoch": 0.2656358551401037, "grad_norm": 1.1603361258031015, "learning_rate": 1.7226980244224896e-05, "loss": 0.5077, "step": 7456 }, { "epoch": 0.2656714822665978, "grad_norm": 1.0273673544445066, "learning_rate": 1.722618263669961e-05, "loss": 0.5207, "step": 7457 }, { "epoch": 0.2657071093930919, "grad_norm": 1.108235049055524, "learning_rate": 1.7225384932952655e-05, "loss": 0.4955, "step": 7458 }, { "epoch": 0.265742736519586, "grad_norm": 1.1497698600733448, "learning_rate": 1.7224587132994648e-05, "loss": 0.5265, "step": 7459 }, { "epoch": 0.26577836364608015, "grad_norm": 1.1203076630222752, "learning_rate": 1.7223789236836212e-05, "loss": 0.5417, "step": 7460 }, { "epoch": 0.2658139907725742, "grad_norm": 1.1708605675921633, "learning_rate": 1.7222991244487976e-05, "loss": 0.5379, "step": 7461 }, { "epoch": 0.26584961789906836, "grad_norm": 1.6716673726592028, "learning_rate": 1.7222193155960565e-05, "loss": 0.6587, "step": 7462 }, { "epoch": 0.2658852450255625, "grad_norm": 1.272635265835712, "learning_rate": 1.72213949712646e-05, "loss": 0.5238, "step": 7463 }, { "epoch": 0.26592087215205656, "grad_norm": 1.6505153812986695, "learning_rate": 1.7220596690410712e-05, "loss": 0.5521, "step": 7464 }, { "epoch": 0.2659564992785507, "grad_norm": 1.550255478570259, "learning_rate": 1.7219798313409538e-05, "loss": 0.5275, "step": 7465 }, { "epoch": 0.2659921264050448, "grad_norm": 1.4493561132400725, "learning_rate": 1.72189998402717e-05, "loss": 0.6133, "step": 7466 }, { "epoch": 0.2660277535315389, "grad_norm": 2.0988419829784997, "learning_rate": 1.7218201271007832e-05, "loss": 0.6435, "step": 7467 }, { "epoch": 0.26606338065803303, "grad_norm": 1.1587997643425318, "learning_rate": 1.7217402605628572e-05, "loss": 0.5357, "step": 7468 }, { "epoch": 0.26609900778452716, "grad_norm": 1.616151668192974, "learning_rate": 1.721660384414455e-05, "loss": 0.5896, "step": 7469 }, { "epoch": 0.26613463491102124, "grad_norm": 1.5102741932581607, "learning_rate": 1.7215804986566405e-05, "loss": 0.5572, "step": 7470 }, { "epoch": 0.26617026203751537, "grad_norm": 1.62906411437661, "learning_rate": 1.721500603290477e-05, "loss": 0.6206, "step": 7471 }, { "epoch": 0.2662058891640095, "grad_norm": 1.5696082606587434, "learning_rate": 1.7214206983170287e-05, "loss": 0.5841, "step": 7472 }, { "epoch": 0.2662415162905036, "grad_norm": 1.6120778159982507, "learning_rate": 1.7213407837373596e-05, "loss": 0.5844, "step": 7473 }, { "epoch": 0.2662771434169977, "grad_norm": 1.1194134993403708, "learning_rate": 1.7212608595525342e-05, "loss": 0.5156, "step": 7474 }, { "epoch": 0.26631277054349184, "grad_norm": 1.642966098782836, "learning_rate": 1.7211809257636158e-05, "loss": 0.6873, "step": 7475 }, { "epoch": 0.2663483976699859, "grad_norm": 1.646316513291391, "learning_rate": 1.7211009823716695e-05, "loss": 0.4955, "step": 7476 }, { "epoch": 0.26638402479648005, "grad_norm": 1.2467164929108314, "learning_rate": 1.7210210293777597e-05, "loss": 0.5214, "step": 7477 }, { "epoch": 0.2664196519229742, "grad_norm": 1.655070787431359, "learning_rate": 1.7209410667829506e-05, "loss": 0.6282, "step": 7478 }, { "epoch": 0.26645527904946825, "grad_norm": 1.0253040842823233, "learning_rate": 1.720861094588308e-05, "loss": 0.4521, "step": 7479 }, { "epoch": 0.2664909061759624, "grad_norm": 1.1626744535309539, "learning_rate": 1.7207811127948953e-05, "loss": 0.5108, "step": 7480 }, { "epoch": 0.2665265333024565, "grad_norm": 1.8271950568602309, "learning_rate": 1.7207011214037785e-05, "loss": 0.6192, "step": 7481 }, { "epoch": 0.2665621604289506, "grad_norm": 1.456430340947668, "learning_rate": 1.7206211204160227e-05, "loss": 0.6033, "step": 7482 }, { "epoch": 0.2665977875554447, "grad_norm": 1.3905980140447733, "learning_rate": 1.720541109832693e-05, "loss": 0.5906, "step": 7483 }, { "epoch": 0.26663341468193885, "grad_norm": 1.6950360146090098, "learning_rate": 1.7204610896548547e-05, "loss": 0.6152, "step": 7484 }, { "epoch": 0.2666690418084329, "grad_norm": 1.2131764548539647, "learning_rate": 1.720381059883573e-05, "loss": 0.5267, "step": 7485 }, { "epoch": 0.26670466893492706, "grad_norm": 1.1352555922978163, "learning_rate": 1.7203010205199144e-05, "loss": 0.5309, "step": 7486 }, { "epoch": 0.2667402960614212, "grad_norm": 1.1456933603070256, "learning_rate": 1.7202209715649444e-05, "loss": 0.5258, "step": 7487 }, { "epoch": 0.26677592318791526, "grad_norm": 1.70634485805759, "learning_rate": 1.7201409130197288e-05, "loss": 0.608, "step": 7488 }, { "epoch": 0.2668115503144094, "grad_norm": 1.1781200497121707, "learning_rate": 1.720060844885333e-05, "loss": 0.5237, "step": 7489 }, { "epoch": 0.2668471774409035, "grad_norm": 1.0787153380351389, "learning_rate": 1.7199807671628243e-05, "loss": 0.5428, "step": 7490 }, { "epoch": 0.2668828045673976, "grad_norm": 1.1515505184173123, "learning_rate": 1.7199006798532683e-05, "loss": 0.4749, "step": 7491 }, { "epoch": 0.26691843169389173, "grad_norm": 1.231032800479276, "learning_rate": 1.7198205829577314e-05, "loss": 0.4779, "step": 7492 }, { "epoch": 0.26695405882038586, "grad_norm": 1.0358179363013256, "learning_rate": 1.7197404764772807e-05, "loss": 0.4951, "step": 7493 }, { "epoch": 0.26698968594687994, "grad_norm": 1.1216274695569166, "learning_rate": 1.7196603604129823e-05, "loss": 0.5044, "step": 7494 }, { "epoch": 0.26702531307337407, "grad_norm": 1.288895538221294, "learning_rate": 1.719580234765903e-05, "loss": 0.5108, "step": 7495 }, { "epoch": 0.2670609401998682, "grad_norm": 1.2497377877167073, "learning_rate": 1.7195000995371104e-05, "loss": 0.5151, "step": 7496 }, { "epoch": 0.2670965673263623, "grad_norm": 1.1727860993974704, "learning_rate": 1.7194199547276708e-05, "loss": 0.4814, "step": 7497 }, { "epoch": 0.2671321944528564, "grad_norm": 2.1011854398462764, "learning_rate": 1.7193398003386514e-05, "loss": 0.5794, "step": 7498 }, { "epoch": 0.26716782157935054, "grad_norm": 1.865744891561887, "learning_rate": 1.71925963637112e-05, "loss": 0.6112, "step": 7499 }, { "epoch": 0.2672034487058446, "grad_norm": 1.136105220070886, "learning_rate": 1.7191794628261437e-05, "loss": 0.4809, "step": 7500 }, { "epoch": 0.26723907583233875, "grad_norm": 1.7455648568385502, "learning_rate": 1.71909927970479e-05, "loss": 0.5844, "step": 7501 }, { "epoch": 0.2672747029588329, "grad_norm": 1.8962222200303003, "learning_rate": 1.7190190870081272e-05, "loss": 0.6037, "step": 7502 }, { "epoch": 0.26731033008532695, "grad_norm": 1.4256856876444026, "learning_rate": 1.7189388847372227e-05, "loss": 0.622, "step": 7503 }, { "epoch": 0.2673459572118211, "grad_norm": 1.7328956033070495, "learning_rate": 1.718858672893144e-05, "loss": 0.5467, "step": 7504 }, { "epoch": 0.2673815843383152, "grad_norm": 1.414112421337126, "learning_rate": 1.7187784514769598e-05, "loss": 0.6197, "step": 7505 }, { "epoch": 0.2674172114648093, "grad_norm": 1.2591528923558386, "learning_rate": 1.7186982204897386e-05, "loss": 0.4805, "step": 7506 }, { "epoch": 0.2674528385913034, "grad_norm": 2.160433052043115, "learning_rate": 1.7186179799325477e-05, "loss": 0.5423, "step": 7507 }, { "epoch": 0.26748846571779755, "grad_norm": 1.2403428570745243, "learning_rate": 1.7185377298064562e-05, "loss": 0.5095, "step": 7508 }, { "epoch": 0.2675240928442916, "grad_norm": 1.9475707798295299, "learning_rate": 1.718457470112533e-05, "loss": 0.5931, "step": 7509 }, { "epoch": 0.26755971997078576, "grad_norm": 1.7075904743652848, "learning_rate": 1.7183772008518458e-05, "loss": 0.6716, "step": 7510 }, { "epoch": 0.2675953470972799, "grad_norm": 1.2294939177712403, "learning_rate": 1.718296922025464e-05, "loss": 0.5288, "step": 7511 }, { "epoch": 0.26763097422377397, "grad_norm": 2.000037949542221, "learning_rate": 1.7182166336344572e-05, "loss": 0.6202, "step": 7512 }, { "epoch": 0.2676666013502681, "grad_norm": 1.484994445295197, "learning_rate": 1.7181363356798938e-05, "loss": 0.5804, "step": 7513 }, { "epoch": 0.2677022284767622, "grad_norm": 1.991443181560533, "learning_rate": 1.718056028162843e-05, "loss": 0.6143, "step": 7514 }, { "epoch": 0.2677378556032563, "grad_norm": 1.5326774942941201, "learning_rate": 1.717975711084374e-05, "loss": 0.5613, "step": 7515 }, { "epoch": 0.26777348272975043, "grad_norm": 1.5750970107204176, "learning_rate": 1.7178953844455572e-05, "loss": 0.517, "step": 7516 }, { "epoch": 0.26780910985624456, "grad_norm": 1.3581390248027747, "learning_rate": 1.7178150482474613e-05, "loss": 0.5045, "step": 7517 }, { "epoch": 0.26784473698273864, "grad_norm": 1.301291739110441, "learning_rate": 1.7177347024911562e-05, "loss": 0.5116, "step": 7518 }, { "epoch": 0.26788036410923277, "grad_norm": 1.4932457332156817, "learning_rate": 1.717654347177712e-05, "loss": 0.5661, "step": 7519 }, { "epoch": 0.2679159912357269, "grad_norm": 1.1884591160059186, "learning_rate": 1.7175739823081985e-05, "loss": 0.5261, "step": 7520 }, { "epoch": 0.267951618362221, "grad_norm": 1.5145612616195383, "learning_rate": 1.717493607883686e-05, "loss": 0.5406, "step": 7521 }, { "epoch": 0.2679872454887151, "grad_norm": 1.6163930804828586, "learning_rate": 1.7174132239052443e-05, "loss": 0.5605, "step": 7522 }, { "epoch": 0.26802287261520924, "grad_norm": 1.2336390474354808, "learning_rate": 1.7173328303739442e-05, "loss": 0.5084, "step": 7523 }, { "epoch": 0.2680584997417033, "grad_norm": 1.5070963135803819, "learning_rate": 1.7172524272908562e-05, "loss": 0.4611, "step": 7524 }, { "epoch": 0.26809412686819745, "grad_norm": 2.1520798864186292, "learning_rate": 1.717172014657051e-05, "loss": 0.6493, "step": 7525 }, { "epoch": 0.2681297539946916, "grad_norm": 1.135479422694062, "learning_rate": 1.717091592473599e-05, "loss": 0.4727, "step": 7526 }, { "epoch": 0.26816538112118565, "grad_norm": 1.639511591245324, "learning_rate": 1.7170111607415706e-05, "loss": 0.6074, "step": 7527 }, { "epoch": 0.2682010082476798, "grad_norm": 1.2657507122217406, "learning_rate": 1.7169307194620383e-05, "loss": 0.5324, "step": 7528 }, { "epoch": 0.2682366353741739, "grad_norm": 1.1128062062051556, "learning_rate": 1.716850268636072e-05, "loss": 0.4942, "step": 7529 }, { "epoch": 0.268272262500668, "grad_norm": 1.5043356797296508, "learning_rate": 1.7167698082647433e-05, "loss": 0.5471, "step": 7530 }, { "epoch": 0.2683078896271621, "grad_norm": 1.765226043845455, "learning_rate": 1.7166893383491237e-05, "loss": 0.6186, "step": 7531 }, { "epoch": 0.26834351675365625, "grad_norm": 1.2104712909020587, "learning_rate": 1.7166088588902843e-05, "loss": 0.5042, "step": 7532 }, { "epoch": 0.26837914388015033, "grad_norm": 1.522415221722007, "learning_rate": 1.7165283698892977e-05, "loss": 0.5481, "step": 7533 }, { "epoch": 0.26841477100664446, "grad_norm": 1.7186477473544015, "learning_rate": 1.7164478713472346e-05, "loss": 0.6104, "step": 7534 }, { "epoch": 0.2684503981331386, "grad_norm": 1.1707301957292693, "learning_rate": 1.7163673632651676e-05, "loss": 0.4965, "step": 7535 }, { "epoch": 0.26848602525963267, "grad_norm": 1.5950684499016923, "learning_rate": 1.7162868456441678e-05, "loss": 0.5047, "step": 7536 }, { "epoch": 0.2685216523861268, "grad_norm": 2.317931043053829, "learning_rate": 1.7162063184853087e-05, "loss": 0.6013, "step": 7537 }, { "epoch": 0.26855727951262093, "grad_norm": 1.201364890950901, "learning_rate": 1.7161257817896613e-05, "loss": 0.4995, "step": 7538 }, { "epoch": 0.268592906639115, "grad_norm": 1.651255240931113, "learning_rate": 1.7160452355582993e-05, "loss": 0.476, "step": 7539 }, { "epoch": 0.26862853376560913, "grad_norm": 1.097968131615993, "learning_rate": 1.715964679792294e-05, "loss": 0.5237, "step": 7540 }, { "epoch": 0.26866416089210327, "grad_norm": 1.6923977805835406, "learning_rate": 1.7158841144927186e-05, "loss": 0.5563, "step": 7541 }, { "epoch": 0.26869978801859734, "grad_norm": 1.1050845599201096, "learning_rate": 1.7158035396606457e-05, "loss": 0.5088, "step": 7542 }, { "epoch": 0.26873541514509147, "grad_norm": 1.5322667620720714, "learning_rate": 1.715722955297149e-05, "loss": 0.629, "step": 7543 }, { "epoch": 0.2687710422715856, "grad_norm": 1.0432569231493818, "learning_rate": 1.7156423614033e-05, "loss": 0.47, "step": 7544 }, { "epoch": 0.2688066693980797, "grad_norm": 1.5709057842357446, "learning_rate": 1.7155617579801735e-05, "loss": 0.5649, "step": 7545 }, { "epoch": 0.2688422965245738, "grad_norm": 1.1298803520056393, "learning_rate": 1.7154811450288414e-05, "loss": 0.5276, "step": 7546 }, { "epoch": 0.26887792365106794, "grad_norm": 1.0879597957400693, "learning_rate": 1.7154005225503785e-05, "loss": 0.5006, "step": 7547 }, { "epoch": 0.268913550777562, "grad_norm": 2.149633524765073, "learning_rate": 1.715319890545857e-05, "loss": 0.6451, "step": 7548 }, { "epoch": 0.26894917790405615, "grad_norm": 1.9159222549462809, "learning_rate": 1.7152392490163516e-05, "loss": 0.6759, "step": 7549 }, { "epoch": 0.2689848050305503, "grad_norm": 2.1020662075485492, "learning_rate": 1.7151585979629354e-05, "loss": 0.5537, "step": 7550 }, { "epoch": 0.26902043215704435, "grad_norm": 1.2939162719931971, "learning_rate": 1.7150779373866825e-05, "loss": 0.5124, "step": 7551 }, { "epoch": 0.2690560592835385, "grad_norm": 1.8043239498627572, "learning_rate": 1.7149972672886673e-05, "loss": 0.6486, "step": 7552 }, { "epoch": 0.2690916864100326, "grad_norm": 1.4017651949225414, "learning_rate": 1.7149165876699635e-05, "loss": 0.6096, "step": 7553 }, { "epoch": 0.2691273135365267, "grad_norm": 1.1850851084555276, "learning_rate": 1.714835898531646e-05, "loss": 0.4889, "step": 7554 }, { "epoch": 0.2691629406630208, "grad_norm": 1.8630259363659882, "learning_rate": 1.7147551998747886e-05, "loss": 0.6051, "step": 7555 }, { "epoch": 0.26919856778951495, "grad_norm": 1.5769889055649, "learning_rate": 1.7146744917004662e-05, "loss": 0.5926, "step": 7556 }, { "epoch": 0.26923419491600903, "grad_norm": 1.5788233753330674, "learning_rate": 1.7145937740097537e-05, "loss": 0.6467, "step": 7557 }, { "epoch": 0.26926982204250316, "grad_norm": 1.6502854545769334, "learning_rate": 1.7145130468037252e-05, "loss": 0.5629, "step": 7558 }, { "epoch": 0.2693054491689973, "grad_norm": 1.298001754093719, "learning_rate": 1.714432310083456e-05, "loss": 0.5133, "step": 7559 }, { "epoch": 0.26934107629549137, "grad_norm": 1.2415700138550196, "learning_rate": 1.7143515638500216e-05, "loss": 0.5604, "step": 7560 }, { "epoch": 0.2693767034219855, "grad_norm": 1.5756226255807892, "learning_rate": 1.714270808104497e-05, "loss": 0.6323, "step": 7561 }, { "epoch": 0.26941233054847963, "grad_norm": 1.1617733868396627, "learning_rate": 1.7141900428479568e-05, "loss": 0.5324, "step": 7562 }, { "epoch": 0.2694479576749737, "grad_norm": 1.1800958787305207, "learning_rate": 1.7141092680814772e-05, "loss": 0.5058, "step": 7563 }, { "epoch": 0.26948358480146783, "grad_norm": 1.0642567836745607, "learning_rate": 1.7140284838061336e-05, "loss": 0.5063, "step": 7564 }, { "epoch": 0.26951921192796197, "grad_norm": 1.837536402213224, "learning_rate": 1.7139476900230013e-05, "loss": 0.6032, "step": 7565 }, { "epoch": 0.26955483905445604, "grad_norm": 1.1634589128390997, "learning_rate": 1.713866886733157e-05, "loss": 0.5029, "step": 7566 }, { "epoch": 0.2695904661809502, "grad_norm": 1.1668198212079872, "learning_rate": 1.713786073937676e-05, "loss": 0.5133, "step": 7567 }, { "epoch": 0.2696260933074443, "grad_norm": 1.7686755530341343, "learning_rate": 1.7137052516376344e-05, "loss": 0.5546, "step": 7568 }, { "epoch": 0.2696617204339384, "grad_norm": 1.1517294548483687, "learning_rate": 1.7136244198341087e-05, "loss": 0.517, "step": 7569 }, { "epoch": 0.2696973475604325, "grad_norm": 1.0945643958277813, "learning_rate": 1.713543578528175e-05, "loss": 0.5174, "step": 7570 }, { "epoch": 0.26973297468692664, "grad_norm": 1.935101934608251, "learning_rate": 1.7134627277209097e-05, "loss": 0.5894, "step": 7571 }, { "epoch": 0.2697686018134207, "grad_norm": 1.0890024428877878, "learning_rate": 1.7133818674133893e-05, "loss": 0.4928, "step": 7572 }, { "epoch": 0.26980422893991485, "grad_norm": 1.4072538612656629, "learning_rate": 1.7133009976066912e-05, "loss": 0.5156, "step": 7573 }, { "epoch": 0.269839856066409, "grad_norm": 1.0298395291910918, "learning_rate": 1.7132201183018915e-05, "loss": 0.4764, "step": 7574 }, { "epoch": 0.26987548319290305, "grad_norm": 1.957434413847198, "learning_rate": 1.7131392295000676e-05, "loss": 0.682, "step": 7575 }, { "epoch": 0.2699111103193972, "grad_norm": 1.1779964101462248, "learning_rate": 1.7130583312022962e-05, "loss": 0.5026, "step": 7576 }, { "epoch": 0.2699467374458913, "grad_norm": 1.7637398782996747, "learning_rate": 1.7129774234096544e-05, "loss": 0.5279, "step": 7577 }, { "epoch": 0.2699823645723854, "grad_norm": 1.1813343534138265, "learning_rate": 1.7128965061232204e-05, "loss": 0.5187, "step": 7578 }, { "epoch": 0.2700179916988795, "grad_norm": 1.0507539003584732, "learning_rate": 1.7128155793440712e-05, "loss": 0.4714, "step": 7579 }, { "epoch": 0.27005361882537365, "grad_norm": 1.2737086227703363, "learning_rate": 1.712734643073284e-05, "loss": 0.4953, "step": 7580 }, { "epoch": 0.27008924595186773, "grad_norm": 1.39254819366292, "learning_rate": 1.7126536973119366e-05, "loss": 0.4894, "step": 7581 }, { "epoch": 0.27012487307836186, "grad_norm": 1.523241961675064, "learning_rate": 1.7125727420611074e-05, "loss": 0.5532, "step": 7582 }, { "epoch": 0.270160500204856, "grad_norm": 1.777359800907656, "learning_rate": 1.7124917773218745e-05, "loss": 0.5879, "step": 7583 }, { "epoch": 0.27019612733135007, "grad_norm": 1.5186021822183466, "learning_rate": 1.712410803095315e-05, "loss": 0.659, "step": 7584 }, { "epoch": 0.2702317544578442, "grad_norm": 2.3523832655712362, "learning_rate": 1.7123298193825083e-05, "loss": 0.5836, "step": 7585 }, { "epoch": 0.27026738158433833, "grad_norm": 1.7763801880694483, "learning_rate": 1.7122488261845316e-05, "loss": 0.6504, "step": 7586 }, { "epoch": 0.2703030087108324, "grad_norm": 1.7642284477121095, "learning_rate": 1.7121678235024642e-05, "loss": 0.537, "step": 7587 }, { "epoch": 0.27033863583732654, "grad_norm": 1.134115800715194, "learning_rate": 1.7120868113373842e-05, "loss": 0.4418, "step": 7588 }, { "epoch": 0.27037426296382067, "grad_norm": 1.5133330450981621, "learning_rate": 1.712005789690371e-05, "loss": 0.5271, "step": 7589 }, { "epoch": 0.27040989009031474, "grad_norm": 1.4097319505858525, "learning_rate": 1.711924758562503e-05, "loss": 0.4753, "step": 7590 }, { "epoch": 0.2704455172168089, "grad_norm": 1.5761068205562832, "learning_rate": 1.711843717954859e-05, "loss": 0.5466, "step": 7591 }, { "epoch": 0.270481144343303, "grad_norm": 1.7957047415142458, "learning_rate": 1.711762667868518e-05, "loss": 0.6057, "step": 7592 }, { "epoch": 0.2705167714697971, "grad_norm": 1.160420314977297, "learning_rate": 1.7116816083045603e-05, "loss": 0.5032, "step": 7593 }, { "epoch": 0.2705523985962912, "grad_norm": 1.6228941396416496, "learning_rate": 1.7116005392640646e-05, "loss": 0.6391, "step": 7594 }, { "epoch": 0.27058802572278534, "grad_norm": 1.6826577964983989, "learning_rate": 1.7115194607481095e-05, "loss": 0.6269, "step": 7595 }, { "epoch": 0.2706236528492794, "grad_norm": 1.45636738339437, "learning_rate": 1.711438372757776e-05, "loss": 0.7139, "step": 7596 }, { "epoch": 0.27065927997577355, "grad_norm": 1.444492080307371, "learning_rate": 1.7113572752941436e-05, "loss": 0.5798, "step": 7597 }, { "epoch": 0.2706949071022677, "grad_norm": 1.6728722186057987, "learning_rate": 1.7112761683582914e-05, "loss": 0.5704, "step": 7598 }, { "epoch": 0.27073053422876175, "grad_norm": 1.2112136439459569, "learning_rate": 1.7111950519513e-05, "loss": 0.5218, "step": 7599 }, { "epoch": 0.2707661613552559, "grad_norm": 1.341409110567694, "learning_rate": 1.7111139260742494e-05, "loss": 0.6134, "step": 7600 }, { "epoch": 0.27080178848175, "grad_norm": 1.081020966672101, "learning_rate": 1.71103279072822e-05, "loss": 0.4913, "step": 7601 }, { "epoch": 0.2708374156082441, "grad_norm": 1.0842577474360577, "learning_rate": 1.7109516459142916e-05, "loss": 0.5204, "step": 7602 }, { "epoch": 0.2708730427347382, "grad_norm": 1.1697287061037114, "learning_rate": 1.7108704916335454e-05, "loss": 0.4998, "step": 7603 }, { "epoch": 0.27090866986123235, "grad_norm": 1.56522991453845, "learning_rate": 1.7107893278870616e-05, "loss": 0.519, "step": 7604 }, { "epoch": 0.27094429698772643, "grad_norm": 1.107637740035335, "learning_rate": 1.710708154675921e-05, "loss": 0.4718, "step": 7605 }, { "epoch": 0.27097992411422056, "grad_norm": 1.090897756618861, "learning_rate": 1.710626972001205e-05, "loss": 0.495, "step": 7606 }, { "epoch": 0.2710155512407147, "grad_norm": 1.6600772186833783, "learning_rate": 1.710545779863994e-05, "loss": 0.6892, "step": 7607 }, { "epoch": 0.27105117836720877, "grad_norm": 1.0350872350585996, "learning_rate": 1.710464578265369e-05, "loss": 0.5009, "step": 7608 }, { "epoch": 0.2710868054937029, "grad_norm": 1.2257017692046595, "learning_rate": 1.7103833672064117e-05, "loss": 0.4826, "step": 7609 }, { "epoch": 0.27112243262019703, "grad_norm": 1.0536435685670729, "learning_rate": 1.7103021466882036e-05, "loss": 0.4642, "step": 7610 }, { "epoch": 0.2711580597466911, "grad_norm": 1.0872274653009644, "learning_rate": 1.7102209167118257e-05, "loss": 0.5411, "step": 7611 }, { "epoch": 0.27119368687318524, "grad_norm": 1.6039316529273073, "learning_rate": 1.7101396772783595e-05, "loss": 0.5057, "step": 7612 }, { "epoch": 0.27122931399967937, "grad_norm": 1.1078577897834962, "learning_rate": 1.7100584283888876e-05, "loss": 0.4869, "step": 7613 }, { "epoch": 0.27126494112617344, "grad_norm": 1.485712098275326, "learning_rate": 1.7099771700444916e-05, "loss": 0.5154, "step": 7614 }, { "epoch": 0.2713005682526676, "grad_norm": 1.0422721246772428, "learning_rate": 1.709895902246253e-05, "loss": 0.4739, "step": 7615 }, { "epoch": 0.2713361953791617, "grad_norm": 1.0396715096381208, "learning_rate": 1.7098146249952545e-05, "loss": 0.4834, "step": 7616 }, { "epoch": 0.2713718225056558, "grad_norm": 1.1375172808475078, "learning_rate": 1.709733338292578e-05, "loss": 0.5228, "step": 7617 }, { "epoch": 0.2714074496321499, "grad_norm": 2.2305387816114703, "learning_rate": 1.709652042139306e-05, "loss": 0.5413, "step": 7618 }, { "epoch": 0.27144307675864404, "grad_norm": 1.1734579560256553, "learning_rate": 1.709570736536521e-05, "loss": 0.5049, "step": 7619 }, { "epoch": 0.2714787038851381, "grad_norm": 1.7451456152298501, "learning_rate": 1.709489421485306e-05, "loss": 0.6274, "step": 7620 }, { "epoch": 0.27151433101163225, "grad_norm": 1.1523270081302, "learning_rate": 1.709408096986743e-05, "loss": 0.4517, "step": 7621 }, { "epoch": 0.2715499581381264, "grad_norm": 1.0644234039801495, "learning_rate": 1.7093267630419156e-05, "loss": 0.5023, "step": 7622 }, { "epoch": 0.27158558526462045, "grad_norm": 1.186965355164172, "learning_rate": 1.7092454196519064e-05, "loss": 0.4388, "step": 7623 }, { "epoch": 0.2716212123911146, "grad_norm": 1.2591710700995822, "learning_rate": 1.709164066817799e-05, "loss": 0.5219, "step": 7624 }, { "epoch": 0.2716568395176087, "grad_norm": 1.7594730324306693, "learning_rate": 1.7090827045406763e-05, "loss": 0.6187, "step": 7625 }, { "epoch": 0.2716924666441028, "grad_norm": 1.2042080850139967, "learning_rate": 1.7090013328216214e-05, "loss": 0.4846, "step": 7626 }, { "epoch": 0.2717280937705969, "grad_norm": 1.085417874907701, "learning_rate": 1.7089199516617188e-05, "loss": 0.5496, "step": 7627 }, { "epoch": 0.27176372089709105, "grad_norm": 1.677870437729397, "learning_rate": 1.7088385610620508e-05, "loss": 0.6307, "step": 7628 }, { "epoch": 0.27179934802358513, "grad_norm": 1.6029710114406028, "learning_rate": 1.7087571610237026e-05, "loss": 0.5722, "step": 7629 }, { "epoch": 0.27183497515007926, "grad_norm": 1.6390682451259504, "learning_rate": 1.708675751547757e-05, "loss": 0.5414, "step": 7630 }, { "epoch": 0.2718706022765734, "grad_norm": 1.172001021624197, "learning_rate": 1.7085943326352988e-05, "loss": 0.5229, "step": 7631 }, { "epoch": 0.2719062294030675, "grad_norm": 1.8047836163980544, "learning_rate": 1.708512904287411e-05, "loss": 0.7003, "step": 7632 }, { "epoch": 0.2719418565295616, "grad_norm": 1.043627215176847, "learning_rate": 1.7084314665051795e-05, "loss": 0.4665, "step": 7633 }, { "epoch": 0.27197748365605573, "grad_norm": 2.0033209305894495, "learning_rate": 1.7083500192896876e-05, "loss": 0.6023, "step": 7634 }, { "epoch": 0.27201311078254986, "grad_norm": 1.1591684086394005, "learning_rate": 1.70826856264202e-05, "loss": 0.5233, "step": 7635 }, { "epoch": 0.27204873790904394, "grad_norm": 1.2559989577007775, "learning_rate": 1.7081870965632618e-05, "loss": 0.4764, "step": 7636 }, { "epoch": 0.27208436503553807, "grad_norm": 1.2133295274229323, "learning_rate": 1.7081056210544968e-05, "loss": 0.5332, "step": 7637 }, { "epoch": 0.2721199921620322, "grad_norm": 1.4477092461190517, "learning_rate": 1.7080241361168108e-05, "loss": 0.6215, "step": 7638 }, { "epoch": 0.2721556192885263, "grad_norm": 1.0972764194284286, "learning_rate": 1.7079426417512886e-05, "loss": 0.52, "step": 7639 }, { "epoch": 0.2721912464150204, "grad_norm": 1.062628673904472, "learning_rate": 1.7078611379590154e-05, "loss": 0.4691, "step": 7640 }, { "epoch": 0.27222687354151454, "grad_norm": 1.5191660538835468, "learning_rate": 1.707779624741076e-05, "loss": 0.6277, "step": 7641 }, { "epoch": 0.2722625006680086, "grad_norm": 1.5633072310514278, "learning_rate": 1.7076981020985567e-05, "loss": 0.5865, "step": 7642 }, { "epoch": 0.27229812779450274, "grad_norm": 1.94495758239894, "learning_rate": 1.707616570032542e-05, "loss": 0.6028, "step": 7643 }, { "epoch": 0.2723337549209969, "grad_norm": 1.327996555630996, "learning_rate": 1.7075350285441185e-05, "loss": 0.5093, "step": 7644 }, { "epoch": 0.27236938204749095, "grad_norm": 1.7277825634326134, "learning_rate": 1.7074534776343715e-05, "loss": 0.5607, "step": 7645 }, { "epoch": 0.2724050091739851, "grad_norm": 1.1666291968361526, "learning_rate": 1.7073719173043865e-05, "loss": 0.5129, "step": 7646 }, { "epoch": 0.2724406363004792, "grad_norm": 1.6606488720220909, "learning_rate": 1.7072903475552505e-05, "loss": 0.5485, "step": 7647 }, { "epoch": 0.2724762634269733, "grad_norm": 2.1037490304659667, "learning_rate": 1.707208768388049e-05, "loss": 0.5848, "step": 7648 }, { "epoch": 0.2725118905534674, "grad_norm": 1.3052392347724704, "learning_rate": 1.7071271798038686e-05, "loss": 0.481, "step": 7649 }, { "epoch": 0.27254751767996155, "grad_norm": 1.6374980716780236, "learning_rate": 1.7070455818037955e-05, "loss": 0.5661, "step": 7650 }, { "epoch": 0.2725831448064556, "grad_norm": 1.1860267305640089, "learning_rate": 1.706963974388916e-05, "loss": 0.545, "step": 7651 }, { "epoch": 0.27261877193294975, "grad_norm": 1.743851148779767, "learning_rate": 1.7068823575603172e-05, "loss": 0.5523, "step": 7652 }, { "epoch": 0.2726543990594439, "grad_norm": 1.0661852858794993, "learning_rate": 1.7068007313190858e-05, "loss": 0.4595, "step": 7653 }, { "epoch": 0.27269002618593796, "grad_norm": 1.198774666597267, "learning_rate": 1.7067190956663084e-05, "loss": 0.5217, "step": 7654 }, { "epoch": 0.2727256533124321, "grad_norm": 1.1262394886527929, "learning_rate": 1.7066374506030726e-05, "loss": 0.484, "step": 7655 }, { "epoch": 0.2727612804389262, "grad_norm": 1.535073117000507, "learning_rate": 1.706555796130465e-05, "loss": 0.6399, "step": 7656 }, { "epoch": 0.2727969075654203, "grad_norm": 1.5829917514107261, "learning_rate": 1.706474132249573e-05, "loss": 0.5994, "step": 7657 }, { "epoch": 0.27283253469191443, "grad_norm": 1.109819695790863, "learning_rate": 1.7063924589614847e-05, "loss": 0.5471, "step": 7658 }, { "epoch": 0.27286816181840856, "grad_norm": 1.5048877530137401, "learning_rate": 1.7063107762672865e-05, "loss": 0.5174, "step": 7659 }, { "epoch": 0.27290378894490264, "grad_norm": 1.7210964373783053, "learning_rate": 1.7062290841680668e-05, "loss": 0.5557, "step": 7660 }, { "epoch": 0.27293941607139677, "grad_norm": 1.2041180057663314, "learning_rate": 1.7061473826649135e-05, "loss": 0.4838, "step": 7661 }, { "epoch": 0.2729750431978909, "grad_norm": 1.3570302971496824, "learning_rate": 1.7060656717589142e-05, "loss": 0.5357, "step": 7662 }, { "epoch": 0.273010670324385, "grad_norm": 1.1144414121148432, "learning_rate": 1.7059839514511565e-05, "loss": 0.5005, "step": 7663 }, { "epoch": 0.2730462974508791, "grad_norm": 1.1966647354215711, "learning_rate": 1.7059022217427293e-05, "loss": 0.4824, "step": 7664 }, { "epoch": 0.27308192457737324, "grad_norm": 1.6713668870688594, "learning_rate": 1.705820482634721e-05, "loss": 0.5872, "step": 7665 }, { "epoch": 0.2731175517038673, "grad_norm": 1.2719547404906455, "learning_rate": 1.705738734128219e-05, "loss": 0.5083, "step": 7666 }, { "epoch": 0.27315317883036144, "grad_norm": 1.0437942662582271, "learning_rate": 1.7056569762243127e-05, "loss": 0.52, "step": 7667 }, { "epoch": 0.2731888059568556, "grad_norm": 1.0859028895997347, "learning_rate": 1.7055752089240907e-05, "loss": 0.4854, "step": 7668 }, { "epoch": 0.27322443308334965, "grad_norm": 1.1577934074756673, "learning_rate": 1.7054934322286415e-05, "loss": 0.4561, "step": 7669 }, { "epoch": 0.2732600602098438, "grad_norm": 1.24492559668044, "learning_rate": 1.705411646139054e-05, "loss": 0.501, "step": 7670 }, { "epoch": 0.2732956873363379, "grad_norm": 1.8998433335534515, "learning_rate": 1.7053298506564174e-05, "loss": 0.5345, "step": 7671 }, { "epoch": 0.273331314462832, "grad_norm": 2.012721056572939, "learning_rate": 1.7052480457818213e-05, "loss": 0.6583, "step": 7672 }, { "epoch": 0.2733669415893261, "grad_norm": 1.131716950401912, "learning_rate": 1.705166231516354e-05, "loss": 0.5166, "step": 7673 }, { "epoch": 0.27340256871582025, "grad_norm": 2.540234870462666, "learning_rate": 1.7050844078611058e-05, "loss": 0.5479, "step": 7674 }, { "epoch": 0.2734381958423143, "grad_norm": 1.0423538967055572, "learning_rate": 1.7050025748171655e-05, "loss": 0.5062, "step": 7675 }, { "epoch": 0.27347382296880846, "grad_norm": 1.0013690801139046, "learning_rate": 1.7049207323856232e-05, "loss": 0.5232, "step": 7676 }, { "epoch": 0.2735094500953026, "grad_norm": 1.54944408184595, "learning_rate": 1.704838880567569e-05, "loss": 0.6644, "step": 7677 }, { "epoch": 0.27354507722179666, "grad_norm": 2.111064573722793, "learning_rate": 1.7047570193640924e-05, "loss": 0.5279, "step": 7678 }, { "epoch": 0.2735807043482908, "grad_norm": 1.7309197400793253, "learning_rate": 1.7046751487762835e-05, "loss": 0.6433, "step": 7679 }, { "epoch": 0.2736163314747849, "grad_norm": 1.0481006746849948, "learning_rate": 1.704593268805232e-05, "loss": 0.5384, "step": 7680 }, { "epoch": 0.273651958601279, "grad_norm": 1.7695330847544641, "learning_rate": 1.704511379452029e-05, "loss": 0.6196, "step": 7681 }, { "epoch": 0.27368758572777313, "grad_norm": 1.5359926709654765, "learning_rate": 1.7044294807177647e-05, "loss": 0.4685, "step": 7682 }, { "epoch": 0.27372321285426726, "grad_norm": 1.1413705129922447, "learning_rate": 1.704347572603529e-05, "loss": 0.493, "step": 7683 }, { "epoch": 0.27375883998076134, "grad_norm": 1.1738130313176522, "learning_rate": 1.7042656551104137e-05, "loss": 0.5046, "step": 7684 }, { "epoch": 0.27379446710725547, "grad_norm": 1.55787531780573, "learning_rate": 1.7041837282395084e-05, "loss": 0.6205, "step": 7685 }, { "epoch": 0.2738300942337496, "grad_norm": 1.5042839981153295, "learning_rate": 1.7041017919919046e-05, "loss": 0.6807, "step": 7686 }, { "epoch": 0.2738657213602437, "grad_norm": 1.4015687150649858, "learning_rate": 1.7040198463686938e-05, "loss": 0.5486, "step": 7687 }, { "epoch": 0.2739013484867378, "grad_norm": 1.8136421132580023, "learning_rate": 1.7039378913709664e-05, "loss": 0.6045, "step": 7688 }, { "epoch": 0.27393697561323194, "grad_norm": 1.1008621630813546, "learning_rate": 1.703855926999814e-05, "loss": 0.4979, "step": 7689 }, { "epoch": 0.273972602739726, "grad_norm": 2.208325123132032, "learning_rate": 1.7037739532563276e-05, "loss": 0.5684, "step": 7690 }, { "epoch": 0.27400822986622014, "grad_norm": 1.1343648312790653, "learning_rate": 1.7036919701415994e-05, "loss": 0.4989, "step": 7691 }, { "epoch": 0.2740438569927143, "grad_norm": 1.653448341257992, "learning_rate": 1.7036099776567207e-05, "loss": 0.5708, "step": 7692 }, { "epoch": 0.27407948411920835, "grad_norm": 1.4423275131633566, "learning_rate": 1.7035279758027835e-05, "loss": 0.6219, "step": 7693 }, { "epoch": 0.2741151112457025, "grad_norm": 1.4250773201562863, "learning_rate": 1.7034459645808794e-05, "loss": 0.5841, "step": 7694 }, { "epoch": 0.2741507383721966, "grad_norm": 1.112034989647926, "learning_rate": 1.7033639439921007e-05, "loss": 0.5363, "step": 7695 }, { "epoch": 0.2741863654986907, "grad_norm": 1.6451390198091542, "learning_rate": 1.7032819140375395e-05, "loss": 0.6688, "step": 7696 }, { "epoch": 0.2742219926251848, "grad_norm": 2.4886396601317466, "learning_rate": 1.7031998747182877e-05, "loss": 0.5595, "step": 7697 }, { "epoch": 0.27425761975167895, "grad_norm": 1.7547299550486606, "learning_rate": 1.7031178260354388e-05, "loss": 0.6671, "step": 7698 }, { "epoch": 0.274293246878173, "grad_norm": 5.294770653026021, "learning_rate": 1.703035767990084e-05, "loss": 0.5086, "step": 7699 }, { "epoch": 0.27432887400466716, "grad_norm": 1.3530776716554196, "learning_rate": 1.702953700583317e-05, "loss": 0.5583, "step": 7700 }, { "epoch": 0.2743645011311613, "grad_norm": 1.0333651027572375, "learning_rate": 1.7028716238162296e-05, "loss": 0.459, "step": 7701 }, { "epoch": 0.27440012825765536, "grad_norm": 1.0846956588869932, "learning_rate": 1.7027895376899154e-05, "loss": 0.5226, "step": 7702 }, { "epoch": 0.2744357553841495, "grad_norm": 1.248875002503767, "learning_rate": 1.702707442205467e-05, "loss": 0.5298, "step": 7703 }, { "epoch": 0.2744713825106436, "grad_norm": 1.2588388080534032, "learning_rate": 1.7026253373639784e-05, "loss": 0.5088, "step": 7704 }, { "epoch": 0.2745070096371377, "grad_norm": 1.8906728619585889, "learning_rate": 1.702543223166542e-05, "loss": 0.6315, "step": 7705 }, { "epoch": 0.27454263676363183, "grad_norm": 1.0855651550287846, "learning_rate": 1.702461099614252e-05, "loss": 0.4848, "step": 7706 }, { "epoch": 0.27457826389012596, "grad_norm": 1.2180983034306319, "learning_rate": 1.7023789667082012e-05, "loss": 0.5095, "step": 7707 }, { "epoch": 0.27461389101662004, "grad_norm": 1.770973590095933, "learning_rate": 1.7022968244494832e-05, "loss": 0.6126, "step": 7708 }, { "epoch": 0.27464951814311417, "grad_norm": 1.0435804099020436, "learning_rate": 1.7022146728391926e-05, "loss": 0.4799, "step": 7709 }, { "epoch": 0.2746851452696083, "grad_norm": 1.177394521402667, "learning_rate": 1.7021325118784225e-05, "loss": 0.492, "step": 7710 }, { "epoch": 0.2747207723961024, "grad_norm": 1.5738550462619412, "learning_rate": 1.7020503415682675e-05, "loss": 0.6889, "step": 7711 }, { "epoch": 0.2747563995225965, "grad_norm": 1.9484749566294144, "learning_rate": 1.701968161909821e-05, "loss": 0.6162, "step": 7712 }, { "epoch": 0.27479202664909064, "grad_norm": 2.044683398430355, "learning_rate": 1.7018859729041785e-05, "loss": 0.6636, "step": 7713 }, { "epoch": 0.2748276537755847, "grad_norm": 1.2107488063819156, "learning_rate": 1.701803774552433e-05, "loss": 0.5618, "step": 7714 }, { "epoch": 0.27486328090207884, "grad_norm": 1.2541512515475588, "learning_rate": 1.70172156685568e-05, "loss": 0.4878, "step": 7715 }, { "epoch": 0.274898908028573, "grad_norm": 1.8138653072281359, "learning_rate": 1.7016393498150135e-05, "loss": 0.5698, "step": 7716 }, { "epoch": 0.27493453515506705, "grad_norm": 1.325952602590499, "learning_rate": 1.7015571234315294e-05, "loss": 0.5035, "step": 7717 }, { "epoch": 0.2749701622815612, "grad_norm": 1.300509303218089, "learning_rate": 1.7014748877063212e-05, "loss": 0.5454, "step": 7718 }, { "epoch": 0.2750057894080553, "grad_norm": 1.1230630899932923, "learning_rate": 1.7013926426404854e-05, "loss": 0.4855, "step": 7719 }, { "epoch": 0.2750414165345494, "grad_norm": 1.1288267022962022, "learning_rate": 1.701310388235115e-05, "loss": 0.4542, "step": 7720 }, { "epoch": 0.2750770436610435, "grad_norm": 1.088664040878978, "learning_rate": 1.7012281244913076e-05, "loss": 0.4969, "step": 7721 }, { "epoch": 0.27511267078753765, "grad_norm": 1.0824382469434366, "learning_rate": 1.7011458514101575e-05, "loss": 0.4482, "step": 7722 }, { "epoch": 0.2751482979140317, "grad_norm": 1.1105491171219317, "learning_rate": 1.7010635689927602e-05, "loss": 0.4669, "step": 7723 }, { "epoch": 0.27518392504052586, "grad_norm": 1.4669614761763816, "learning_rate": 1.7009812772402115e-05, "loss": 0.6121, "step": 7724 }, { "epoch": 0.27521955216702, "grad_norm": 1.067250507104659, "learning_rate": 1.7008989761536067e-05, "loss": 0.492, "step": 7725 }, { "epoch": 0.27525517929351406, "grad_norm": 1.1985423347932262, "learning_rate": 1.7008166657340425e-05, "loss": 0.4771, "step": 7726 }, { "epoch": 0.2752908064200082, "grad_norm": 1.576259200726395, "learning_rate": 1.7007343459826147e-05, "loss": 0.5574, "step": 7727 }, { "epoch": 0.2753264335465023, "grad_norm": 1.1364781416714467, "learning_rate": 1.700652016900419e-05, "loss": 0.4903, "step": 7728 }, { "epoch": 0.2753620606729964, "grad_norm": 1.7051388005087817, "learning_rate": 1.7005696784885518e-05, "loss": 0.5647, "step": 7729 }, { "epoch": 0.27539768779949053, "grad_norm": 1.1397171133523991, "learning_rate": 1.70048733074811e-05, "loss": 0.532, "step": 7730 }, { "epoch": 0.27543331492598466, "grad_norm": 1.134158844051295, "learning_rate": 1.7004049736801898e-05, "loss": 0.5454, "step": 7731 }, { "epoch": 0.27546894205247874, "grad_norm": 1.1776160575225085, "learning_rate": 1.700322607285888e-05, "loss": 0.5416, "step": 7732 }, { "epoch": 0.27550456917897287, "grad_norm": 1.4465967976073622, "learning_rate": 1.700240231566301e-05, "loss": 0.6891, "step": 7733 }, { "epoch": 0.275540196305467, "grad_norm": 0.9622560639432599, "learning_rate": 1.700157846522526e-05, "loss": 0.4357, "step": 7734 }, { "epoch": 0.2755758234319611, "grad_norm": 1.279799195351912, "learning_rate": 1.7000754521556596e-05, "loss": 0.524, "step": 7735 }, { "epoch": 0.2756114505584552, "grad_norm": 1.7502972503908218, "learning_rate": 1.6999930484667992e-05, "loss": 0.6, "step": 7736 }, { "epoch": 0.27564707768494934, "grad_norm": 1.690683289525277, "learning_rate": 1.6999106354570422e-05, "loss": 0.6423, "step": 7737 }, { "epoch": 0.2756827048114434, "grad_norm": 1.04850685176387, "learning_rate": 1.699828213127486e-05, "loss": 0.4684, "step": 7738 }, { "epoch": 0.27571833193793754, "grad_norm": 2.052590354830315, "learning_rate": 1.6997457814792278e-05, "loss": 0.5991, "step": 7739 }, { "epoch": 0.2757539590644317, "grad_norm": 1.223536735828306, "learning_rate": 1.6996633405133656e-05, "loss": 0.6054, "step": 7740 }, { "epoch": 0.27578958619092575, "grad_norm": 1.446809234842663, "learning_rate": 1.6995808902309968e-05, "loss": 0.6729, "step": 7741 }, { "epoch": 0.2758252133174199, "grad_norm": 2.115974888561898, "learning_rate": 1.6994984306332195e-05, "loss": 0.523, "step": 7742 }, { "epoch": 0.275860840443914, "grad_norm": 1.9057490015471443, "learning_rate": 1.6994159617211318e-05, "loss": 0.5687, "step": 7743 }, { "epoch": 0.2758964675704081, "grad_norm": 1.0481990541464337, "learning_rate": 1.6993334834958318e-05, "loss": 0.4663, "step": 7744 }, { "epoch": 0.2759320946969022, "grad_norm": 1.4751464552777074, "learning_rate": 1.6992509959584176e-05, "loss": 0.5714, "step": 7745 }, { "epoch": 0.27596772182339635, "grad_norm": 2.983882439299629, "learning_rate": 1.6991684991099873e-05, "loss": 0.5455, "step": 7746 }, { "epoch": 0.2760033489498904, "grad_norm": 2.608928166912141, "learning_rate": 1.6990859929516404e-05, "loss": 0.616, "step": 7747 }, { "epoch": 0.27603897607638456, "grad_norm": 1.8360316396450027, "learning_rate": 1.6990034774844745e-05, "loss": 0.5353, "step": 7748 }, { "epoch": 0.2760746032028787, "grad_norm": 1.1072638803924604, "learning_rate": 1.6989209527095885e-05, "loss": 0.5087, "step": 7749 }, { "epoch": 0.27611023032937276, "grad_norm": 1.154003159278275, "learning_rate": 1.698838418628082e-05, "loss": 0.5297, "step": 7750 }, { "epoch": 0.2761458574558669, "grad_norm": 1.1348758135009904, "learning_rate": 1.698755875241053e-05, "loss": 0.5108, "step": 7751 }, { "epoch": 0.276181484582361, "grad_norm": 20.945611536391286, "learning_rate": 1.698673322549601e-05, "loss": 0.4858, "step": 7752 }, { "epoch": 0.2762171117088551, "grad_norm": 1.1487091480819018, "learning_rate": 1.698590760554826e-05, "loss": 0.5221, "step": 7753 }, { "epoch": 0.27625273883534923, "grad_norm": 2.095785941834821, "learning_rate": 1.6985081892578264e-05, "loss": 0.4959, "step": 7754 }, { "epoch": 0.27628836596184336, "grad_norm": 1.129476316524489, "learning_rate": 1.6984256086597022e-05, "loss": 0.5009, "step": 7755 }, { "epoch": 0.27632399308833744, "grad_norm": 1.005287745358592, "learning_rate": 1.6983430187615527e-05, "loss": 0.4942, "step": 7756 }, { "epoch": 0.27635962021483157, "grad_norm": 1.1081013620726756, "learning_rate": 1.698260419564478e-05, "loss": 0.5318, "step": 7757 }, { "epoch": 0.2763952473413257, "grad_norm": 1.094816486145594, "learning_rate": 1.6981778110695775e-05, "loss": 0.4722, "step": 7758 }, { "epoch": 0.2764308744678198, "grad_norm": 1.062077793307631, "learning_rate": 1.6980951932779514e-05, "loss": 0.5328, "step": 7759 }, { "epoch": 0.2764665015943139, "grad_norm": 1.186471070064305, "learning_rate": 1.6980125661907e-05, "loss": 0.4692, "step": 7760 }, { "epoch": 0.27650212872080804, "grad_norm": 1.06023726109529, "learning_rate": 1.6979299298089237e-05, "loss": 0.5134, "step": 7761 }, { "epoch": 0.2765377558473021, "grad_norm": 1.074758016936866, "learning_rate": 1.697847284133722e-05, "loss": 0.4827, "step": 7762 }, { "epoch": 0.27657338297379624, "grad_norm": 1.9046999984128494, "learning_rate": 1.697764629166196e-05, "loss": 0.5911, "step": 7763 }, { "epoch": 0.2766090101002904, "grad_norm": 1.1352386189575583, "learning_rate": 1.6976819649074465e-05, "loss": 0.4996, "step": 7764 }, { "epoch": 0.27664463722678445, "grad_norm": 2.185909578499956, "learning_rate": 1.697599291358574e-05, "loss": 0.6341, "step": 7765 }, { "epoch": 0.2766802643532786, "grad_norm": 1.2288937315316304, "learning_rate": 1.697516608520679e-05, "loss": 0.5259, "step": 7766 }, { "epoch": 0.2767158914797727, "grad_norm": 2.760844427591542, "learning_rate": 1.6974339163948636e-05, "loss": 0.5199, "step": 7767 }, { "epoch": 0.2767515186062668, "grad_norm": 1.2025414559710919, "learning_rate": 1.6973512149822272e-05, "loss": 0.5008, "step": 7768 }, { "epoch": 0.2767871457327609, "grad_norm": 1.0667717752892003, "learning_rate": 1.6972685042838725e-05, "loss": 0.4335, "step": 7769 }, { "epoch": 0.27682277285925505, "grad_norm": 1.4593067535757978, "learning_rate": 1.6971857843009e-05, "loss": 0.5888, "step": 7770 }, { "epoch": 0.2768583999857491, "grad_norm": 1.20596601103069, "learning_rate": 1.6971030550344122e-05, "loss": 0.5551, "step": 7771 }, { "epoch": 0.27689402711224326, "grad_norm": 1.3402137782456511, "learning_rate": 1.6970203164855092e-05, "loss": 0.5892, "step": 7772 }, { "epoch": 0.2769296542387374, "grad_norm": 1.5356903492706082, "learning_rate": 1.696937568655294e-05, "loss": 0.6921, "step": 7773 }, { "epoch": 0.27696528136523146, "grad_norm": 1.1604779069970592, "learning_rate": 1.696854811544867e-05, "loss": 0.513, "step": 7774 }, { "epoch": 0.2770009084917256, "grad_norm": 1.4395520162522784, "learning_rate": 1.696772045155332e-05, "loss": 0.6079, "step": 7775 }, { "epoch": 0.2770365356182197, "grad_norm": 1.0391779175853069, "learning_rate": 1.69668926948779e-05, "loss": 0.4498, "step": 7776 }, { "epoch": 0.2770721627447138, "grad_norm": 1.8095030186782997, "learning_rate": 1.6966064845433436e-05, "loss": 0.6824, "step": 7777 }, { "epoch": 0.27710778987120793, "grad_norm": 1.226327838156696, "learning_rate": 1.6965236903230947e-05, "loss": 0.5709, "step": 7778 }, { "epoch": 0.27714341699770206, "grad_norm": 1.1566722261398295, "learning_rate": 1.696440886828146e-05, "loss": 0.527, "step": 7779 }, { "epoch": 0.27717904412419614, "grad_norm": 1.1676841793973751, "learning_rate": 1.6963580740596003e-05, "loss": 0.5065, "step": 7780 }, { "epoch": 0.27721467125069027, "grad_norm": 1.163684838732349, "learning_rate": 1.6962752520185598e-05, "loss": 0.4628, "step": 7781 }, { "epoch": 0.2772502983771844, "grad_norm": 1.050614099174664, "learning_rate": 1.6961924207061283e-05, "loss": 0.5042, "step": 7782 }, { "epoch": 0.2772859255036785, "grad_norm": 1.2401424518049453, "learning_rate": 1.6961095801234075e-05, "loss": 0.5096, "step": 7783 }, { "epoch": 0.2773215526301726, "grad_norm": 1.1467336687960008, "learning_rate": 1.6960267302715014e-05, "loss": 0.5351, "step": 7784 }, { "epoch": 0.27735717975666674, "grad_norm": 1.5462532583880284, "learning_rate": 1.6959438711515126e-05, "loss": 0.6058, "step": 7785 }, { "epoch": 0.2773928068831608, "grad_norm": 1.0435251313748615, "learning_rate": 1.695861002764545e-05, "loss": 0.4992, "step": 7786 }, { "epoch": 0.27742843400965495, "grad_norm": 1.4636049390940025, "learning_rate": 1.6957781251117012e-05, "loss": 0.552, "step": 7787 }, { "epoch": 0.2774640611361491, "grad_norm": 1.0940405390786707, "learning_rate": 1.695695238194086e-05, "loss": 0.5118, "step": 7788 }, { "epoch": 0.27749968826264315, "grad_norm": 1.2882370265132836, "learning_rate": 1.6956123420128025e-05, "loss": 0.5988, "step": 7789 }, { "epoch": 0.2775353153891373, "grad_norm": 1.1665817185841239, "learning_rate": 1.695529436568954e-05, "loss": 0.5575, "step": 7790 }, { "epoch": 0.2775709425156314, "grad_norm": 1.5180185703598525, "learning_rate": 1.6954465218636452e-05, "loss": 0.5317, "step": 7791 }, { "epoch": 0.2776065696421255, "grad_norm": 1.7826923537471733, "learning_rate": 1.6953635978979798e-05, "loss": 0.5806, "step": 7792 }, { "epoch": 0.2776421967686196, "grad_norm": 1.175387551090058, "learning_rate": 1.695280664673062e-05, "loss": 0.5344, "step": 7793 }, { "epoch": 0.27767782389511375, "grad_norm": 2.029323888553775, "learning_rate": 1.6951977221899962e-05, "loss": 0.6055, "step": 7794 }, { "epoch": 0.2777134510216078, "grad_norm": 1.1381819255578027, "learning_rate": 1.695114770449887e-05, "loss": 0.4889, "step": 7795 }, { "epoch": 0.27774907814810196, "grad_norm": 1.108288913932921, "learning_rate": 1.6950318094538386e-05, "loss": 0.4477, "step": 7796 }, { "epoch": 0.2777847052745961, "grad_norm": 1.3079679479225277, "learning_rate": 1.6949488392029557e-05, "loss": 0.4748, "step": 7797 }, { "epoch": 0.27782033240109016, "grad_norm": 1.638125402292466, "learning_rate": 1.6948658596983435e-05, "loss": 0.6589, "step": 7798 }, { "epoch": 0.2778559595275843, "grad_norm": 1.170389782561419, "learning_rate": 1.6947828709411066e-05, "loss": 0.5188, "step": 7799 }, { "epoch": 0.2778915866540784, "grad_norm": 1.2113707162087117, "learning_rate": 1.69469987293235e-05, "loss": 0.5011, "step": 7800 }, { "epoch": 0.2779272137805725, "grad_norm": 1.4955023458803718, "learning_rate": 1.6946168656731793e-05, "loss": 0.5764, "step": 7801 }, { "epoch": 0.27796284090706663, "grad_norm": 1.1693967809955146, "learning_rate": 1.6945338491646995e-05, "loss": 0.5192, "step": 7802 }, { "epoch": 0.27799846803356076, "grad_norm": 1.573606204808036, "learning_rate": 1.694450823408016e-05, "loss": 0.5447, "step": 7803 }, { "epoch": 0.27803409516005484, "grad_norm": 1.067545548437786, "learning_rate": 1.6943677884042343e-05, "loss": 0.4661, "step": 7804 }, { "epoch": 0.27806972228654897, "grad_norm": 1.9530448620608813, "learning_rate": 1.69428474415446e-05, "loss": 0.5874, "step": 7805 }, { "epoch": 0.2781053494130431, "grad_norm": 1.106724413935052, "learning_rate": 1.6942016906597997e-05, "loss": 0.4712, "step": 7806 }, { "epoch": 0.2781409765395372, "grad_norm": 1.6887729790070674, "learning_rate": 1.694118627921358e-05, "loss": 0.6195, "step": 7807 }, { "epoch": 0.2781766036660313, "grad_norm": 1.1636584615982313, "learning_rate": 1.694035555940242e-05, "loss": 0.5231, "step": 7808 }, { "epoch": 0.27821223079252544, "grad_norm": 1.4963352229869462, "learning_rate": 1.6939524747175574e-05, "loss": 0.4924, "step": 7809 }, { "epoch": 0.2782478579190195, "grad_norm": 2.0353336105741344, "learning_rate": 1.6938693842544105e-05, "loss": 0.6519, "step": 7810 }, { "epoch": 0.27828348504551365, "grad_norm": 1.1305161428970003, "learning_rate": 1.6937862845519078e-05, "loss": 0.5414, "step": 7811 }, { "epoch": 0.2783191121720078, "grad_norm": 1.0867970505934064, "learning_rate": 1.693703175611156e-05, "loss": 0.5373, "step": 7812 }, { "epoch": 0.27835473929850185, "grad_norm": 1.3966721667508641, "learning_rate": 1.693620057433261e-05, "loss": 0.5122, "step": 7813 }, { "epoch": 0.278390366424996, "grad_norm": 1.1404731450037877, "learning_rate": 1.6935369300193304e-05, "loss": 0.4978, "step": 7814 }, { "epoch": 0.2784259935514901, "grad_norm": 1.076595124940253, "learning_rate": 1.693453793370471e-05, "loss": 0.5001, "step": 7815 }, { "epoch": 0.2784616206779842, "grad_norm": 1.8531291820562201, "learning_rate": 1.6933706474877895e-05, "loss": 0.6104, "step": 7816 }, { "epoch": 0.2784972478044783, "grad_norm": 1.4249712962917278, "learning_rate": 1.693287492372393e-05, "loss": 0.5874, "step": 7817 }, { "epoch": 0.27853287493097245, "grad_norm": 1.0502904293567237, "learning_rate": 1.6932043280253892e-05, "loss": 0.522, "step": 7818 }, { "epoch": 0.2785685020574666, "grad_norm": 1.1880108327296102, "learning_rate": 1.6931211544478855e-05, "loss": 0.4736, "step": 7819 }, { "epoch": 0.27860412918396066, "grad_norm": 1.2780146618978319, "learning_rate": 1.6930379716409885e-05, "loss": 0.5542, "step": 7820 }, { "epoch": 0.2786397563104548, "grad_norm": 1.4213874573296272, "learning_rate": 1.692954779605807e-05, "loss": 0.5625, "step": 7821 }, { "epoch": 0.2786753834369489, "grad_norm": 1.253634205284154, "learning_rate": 1.6928715783434482e-05, "loss": 0.5429, "step": 7822 }, { "epoch": 0.278711010563443, "grad_norm": 1.0661846628878604, "learning_rate": 1.6927883678550197e-05, "loss": 0.4842, "step": 7823 }, { "epoch": 0.2787466376899371, "grad_norm": 2.0455107711283946, "learning_rate": 1.6927051481416304e-05, "loss": 0.5546, "step": 7824 }, { "epoch": 0.27878226481643126, "grad_norm": 1.7201178270391335, "learning_rate": 1.6926219192043873e-05, "loss": 0.6142, "step": 7825 }, { "epoch": 0.27881789194292533, "grad_norm": 1.4803334348046746, "learning_rate": 1.6925386810443994e-05, "loss": 0.516, "step": 7826 }, { "epoch": 0.27885351906941946, "grad_norm": 1.1355056360686333, "learning_rate": 1.6924554336627754e-05, "loss": 0.5462, "step": 7827 }, { "epoch": 0.2788891461959136, "grad_norm": 1.8718992562090537, "learning_rate": 1.692372177060623e-05, "loss": 0.5566, "step": 7828 }, { "epoch": 0.27892477332240767, "grad_norm": 1.6295375532843237, "learning_rate": 1.692288911239051e-05, "loss": 0.6487, "step": 7829 }, { "epoch": 0.2789604004489018, "grad_norm": 1.0522336190082913, "learning_rate": 1.6922056361991685e-05, "loss": 0.4629, "step": 7830 }, { "epoch": 0.27899602757539593, "grad_norm": 1.3422629545861455, "learning_rate": 1.692122351942084e-05, "loss": 0.5007, "step": 7831 }, { "epoch": 0.27903165470189, "grad_norm": 1.1709290667588674, "learning_rate": 1.6920390584689063e-05, "loss": 0.4507, "step": 7832 }, { "epoch": 0.27906728182838414, "grad_norm": 1.5856568324089395, "learning_rate": 1.6919557557807454e-05, "loss": 0.6422, "step": 7833 }, { "epoch": 0.27910290895487827, "grad_norm": 1.4358670238187068, "learning_rate": 1.6918724438787095e-05, "loss": 0.5423, "step": 7834 }, { "epoch": 0.27913853608137235, "grad_norm": 1.1454986039589834, "learning_rate": 1.6917891227639086e-05, "loss": 0.4767, "step": 7835 }, { "epoch": 0.2791741632078665, "grad_norm": 1.0553183302670492, "learning_rate": 1.691705792437452e-05, "loss": 0.5203, "step": 7836 }, { "epoch": 0.2792097903343606, "grad_norm": 1.1106023531660734, "learning_rate": 1.691622452900449e-05, "loss": 0.4941, "step": 7837 }, { "epoch": 0.2792454174608547, "grad_norm": 1.102502729835337, "learning_rate": 1.69153910415401e-05, "loss": 0.5187, "step": 7838 }, { "epoch": 0.2792810445873488, "grad_norm": 1.1695574045286723, "learning_rate": 1.6914557461992445e-05, "loss": 0.4972, "step": 7839 }, { "epoch": 0.27931667171384295, "grad_norm": 1.139495644163874, "learning_rate": 1.6913723790372626e-05, "loss": 0.4917, "step": 7840 }, { "epoch": 0.279352298840337, "grad_norm": 1.055879999271983, "learning_rate": 1.691289002669174e-05, "loss": 0.4787, "step": 7841 }, { "epoch": 0.27938792596683115, "grad_norm": 1.1429827307835214, "learning_rate": 1.691205617096089e-05, "loss": 0.5726, "step": 7842 }, { "epoch": 0.2794235530933253, "grad_norm": 1.969518114600432, "learning_rate": 1.691122222319118e-05, "loss": 0.6565, "step": 7843 }, { "epoch": 0.27945918021981936, "grad_norm": 1.383844302948442, "learning_rate": 1.691038818339372e-05, "loss": 0.5912, "step": 7844 }, { "epoch": 0.2794948073463135, "grad_norm": 1.6028515659982536, "learning_rate": 1.690955405157961e-05, "loss": 0.6254, "step": 7845 }, { "epoch": 0.2795304344728076, "grad_norm": 1.2140693167839278, "learning_rate": 1.6908719827759954e-05, "loss": 0.4978, "step": 7846 }, { "epoch": 0.2795660615993017, "grad_norm": 1.153222161925711, "learning_rate": 1.690788551194587e-05, "loss": 0.5014, "step": 7847 }, { "epoch": 0.2796016887257958, "grad_norm": 2.00998039246402, "learning_rate": 1.6907051104148456e-05, "loss": 0.5925, "step": 7848 }, { "epoch": 0.27963731585228996, "grad_norm": 1.1102566185876481, "learning_rate": 1.6906216604378836e-05, "loss": 0.5034, "step": 7849 }, { "epoch": 0.27967294297878403, "grad_norm": 1.480048646720293, "learning_rate": 1.690538201264811e-05, "loss": 0.5791, "step": 7850 }, { "epoch": 0.27970857010527816, "grad_norm": 2.0182402618126667, "learning_rate": 1.69045473289674e-05, "loss": 0.5822, "step": 7851 }, { "epoch": 0.2797441972317723, "grad_norm": 1.551911973301401, "learning_rate": 1.6903712553347812e-05, "loss": 0.6438, "step": 7852 }, { "epoch": 0.27977982435826637, "grad_norm": 1.923482737954199, "learning_rate": 1.6902877685800467e-05, "loss": 0.5475, "step": 7853 }, { "epoch": 0.2798154514847605, "grad_norm": 1.098408182138369, "learning_rate": 1.690204272633648e-05, "loss": 0.5136, "step": 7854 }, { "epoch": 0.27985107861125463, "grad_norm": 1.4155701369576363, "learning_rate": 1.6901207674966972e-05, "loss": 0.5904, "step": 7855 }, { "epoch": 0.2798867057377487, "grad_norm": 1.1048643880913271, "learning_rate": 1.690037253170306e-05, "loss": 0.5123, "step": 7856 }, { "epoch": 0.27992233286424284, "grad_norm": 1.2114914412706592, "learning_rate": 1.6899537296555865e-05, "loss": 0.5449, "step": 7857 }, { "epoch": 0.27995795999073697, "grad_norm": 1.7612200464461472, "learning_rate": 1.6898701969536508e-05, "loss": 0.5453, "step": 7858 }, { "epoch": 0.27999358711723105, "grad_norm": 1.0959391916660075, "learning_rate": 1.689786655065611e-05, "loss": 0.4753, "step": 7859 }, { "epoch": 0.2800292142437252, "grad_norm": 1.259948403697737, "learning_rate": 1.68970310399258e-05, "loss": 0.5249, "step": 7860 }, { "epoch": 0.2800648413702193, "grad_norm": 1.1165446357865387, "learning_rate": 1.68961954373567e-05, "loss": 0.4874, "step": 7861 }, { "epoch": 0.2801004684967134, "grad_norm": 1.1492255081400011, "learning_rate": 1.689535974295994e-05, "loss": 0.4907, "step": 7862 }, { "epoch": 0.2801360956232075, "grad_norm": 1.1128727610537967, "learning_rate": 1.689452395674664e-05, "loss": 0.515, "step": 7863 }, { "epoch": 0.28017172274970165, "grad_norm": 1.0976585987198308, "learning_rate": 1.689368807872794e-05, "loss": 0.512, "step": 7864 }, { "epoch": 0.2802073498761957, "grad_norm": 2.6042872996891906, "learning_rate": 1.6892852108914963e-05, "loss": 0.6469, "step": 7865 }, { "epoch": 0.28024297700268985, "grad_norm": 1.1222817859818432, "learning_rate": 1.689201604731884e-05, "loss": 0.4145, "step": 7866 }, { "epoch": 0.280278604129184, "grad_norm": 1.6220958490016106, "learning_rate": 1.689117989395071e-05, "loss": 0.6692, "step": 7867 }, { "epoch": 0.28031423125567806, "grad_norm": 1.255722069404579, "learning_rate": 1.68903436488217e-05, "loss": 0.4613, "step": 7868 }, { "epoch": 0.2803498583821722, "grad_norm": 1.7716371931928239, "learning_rate": 1.6889507311942946e-05, "loss": 0.6388, "step": 7869 }, { "epoch": 0.2803854855086663, "grad_norm": 1.1781159781575952, "learning_rate": 1.6888670883325593e-05, "loss": 0.5071, "step": 7870 }, { "epoch": 0.2804211126351604, "grad_norm": 1.7777234991245137, "learning_rate": 1.6887834362980768e-05, "loss": 0.5162, "step": 7871 }, { "epoch": 0.2804567397616545, "grad_norm": 2.0772530688427495, "learning_rate": 1.688699775091962e-05, "loss": 0.6048, "step": 7872 }, { "epoch": 0.28049236688814866, "grad_norm": 1.096658248280147, "learning_rate": 1.6886161047153278e-05, "loss": 0.4729, "step": 7873 }, { "epoch": 0.28052799401464273, "grad_norm": 1.1751264333626033, "learning_rate": 1.6885324251692887e-05, "loss": 0.4977, "step": 7874 }, { "epoch": 0.28056362114113687, "grad_norm": 1.1791139477420156, "learning_rate": 1.6884487364549596e-05, "loss": 0.4879, "step": 7875 }, { "epoch": 0.280599248267631, "grad_norm": 1.0372653344711718, "learning_rate": 1.688365038573454e-05, "loss": 0.4589, "step": 7876 }, { "epoch": 0.28063487539412507, "grad_norm": 2.218912119592175, "learning_rate": 1.688281331525887e-05, "loss": 0.5216, "step": 7877 }, { "epoch": 0.2806705025206192, "grad_norm": 1.6083725499545447, "learning_rate": 1.6881976153133724e-05, "loss": 0.5869, "step": 7878 }, { "epoch": 0.28070612964711333, "grad_norm": 1.2658623275630123, "learning_rate": 1.6881138899370263e-05, "loss": 0.5745, "step": 7879 }, { "epoch": 0.2807417567736074, "grad_norm": 1.6800306898963175, "learning_rate": 1.6880301553979627e-05, "loss": 0.544, "step": 7880 }, { "epoch": 0.28077738390010154, "grad_norm": 1.9190958987398767, "learning_rate": 1.687946411697296e-05, "loss": 0.5836, "step": 7881 }, { "epoch": 0.28081301102659567, "grad_norm": 1.03409938777769, "learning_rate": 1.6878626588361427e-05, "loss": 0.5061, "step": 7882 }, { "epoch": 0.28084863815308975, "grad_norm": 1.849060467466531, "learning_rate": 1.6877788968156172e-05, "loss": 0.6109, "step": 7883 }, { "epoch": 0.2808842652795839, "grad_norm": 1.6531366974958333, "learning_rate": 1.6876951256368346e-05, "loss": 0.5543, "step": 7884 }, { "epoch": 0.280919892406078, "grad_norm": 1.1209807064119115, "learning_rate": 1.6876113453009112e-05, "loss": 0.4783, "step": 7885 }, { "epoch": 0.2809555195325721, "grad_norm": 1.2033128626850134, "learning_rate": 1.6875275558089618e-05, "loss": 0.538, "step": 7886 }, { "epoch": 0.2809911466590662, "grad_norm": 1.5515288010984125, "learning_rate": 1.6874437571621025e-05, "loss": 0.561, "step": 7887 }, { "epoch": 0.28102677378556035, "grad_norm": 1.4251750330436324, "learning_rate": 1.6873599493614492e-05, "loss": 0.6218, "step": 7888 }, { "epoch": 0.2810624009120544, "grad_norm": 1.1067321082980346, "learning_rate": 1.6872761324081174e-05, "loss": 0.4961, "step": 7889 }, { "epoch": 0.28109802803854855, "grad_norm": 1.0810181597589115, "learning_rate": 1.6871923063032238e-05, "loss": 0.5198, "step": 7890 }, { "epoch": 0.2811336551650427, "grad_norm": 3.2359683315708403, "learning_rate": 1.6871084710478844e-05, "loss": 0.6515, "step": 7891 }, { "epoch": 0.28116928229153676, "grad_norm": 1.1035984202289357, "learning_rate": 1.6870246266432153e-05, "loss": 0.5045, "step": 7892 }, { "epoch": 0.2812049094180309, "grad_norm": 1.0525387281790144, "learning_rate": 1.686940773090333e-05, "loss": 0.5215, "step": 7893 }, { "epoch": 0.281240536544525, "grad_norm": 1.1874015192306469, "learning_rate": 1.6868569103903544e-05, "loss": 0.5293, "step": 7894 }, { "epoch": 0.2812761636710191, "grad_norm": 1.5617241408548765, "learning_rate": 1.686773038544396e-05, "loss": 0.6104, "step": 7895 }, { "epoch": 0.28131179079751323, "grad_norm": 1.1212406007251736, "learning_rate": 1.686689157553574e-05, "loss": 0.5129, "step": 7896 }, { "epoch": 0.28134741792400736, "grad_norm": 1.1316203525559299, "learning_rate": 1.6866052674190067e-05, "loss": 0.511, "step": 7897 }, { "epoch": 0.28138304505050143, "grad_norm": 1.4390897179210176, "learning_rate": 1.68652136814181e-05, "loss": 0.6564, "step": 7898 }, { "epoch": 0.28141867217699557, "grad_norm": 1.2444744031496755, "learning_rate": 1.686437459723101e-05, "loss": 0.5231, "step": 7899 }, { "epoch": 0.2814542993034897, "grad_norm": 1.6392224955120105, "learning_rate": 1.6863535421639982e-05, "loss": 0.6609, "step": 7900 }, { "epoch": 0.28148992642998377, "grad_norm": 1.240188451572169, "learning_rate": 1.6862696154656178e-05, "loss": 0.5368, "step": 7901 }, { "epoch": 0.2815255535564779, "grad_norm": 1.6179182056222365, "learning_rate": 1.686185679629078e-05, "loss": 0.5755, "step": 7902 }, { "epoch": 0.28156118068297203, "grad_norm": 1.1109212010678873, "learning_rate": 1.6861017346554958e-05, "loss": 0.4742, "step": 7903 }, { "epoch": 0.2815968078094661, "grad_norm": 1.6897156918908673, "learning_rate": 1.68601778054599e-05, "loss": 0.5649, "step": 7904 }, { "epoch": 0.28163243493596024, "grad_norm": 1.156807647000625, "learning_rate": 1.685933817301678e-05, "loss": 0.5187, "step": 7905 }, { "epoch": 0.28166806206245437, "grad_norm": 1.4874404961417702, "learning_rate": 1.6858498449236774e-05, "loss": 0.5514, "step": 7906 }, { "epoch": 0.28170368918894845, "grad_norm": 1.6952956198215634, "learning_rate": 1.6857658634131068e-05, "loss": 0.611, "step": 7907 }, { "epoch": 0.2817393163154426, "grad_norm": 1.1394656694451981, "learning_rate": 1.6856818727710847e-05, "loss": 0.5517, "step": 7908 }, { "epoch": 0.2817749434419367, "grad_norm": 1.1990463308650001, "learning_rate": 1.6855978729987287e-05, "loss": 0.4809, "step": 7909 }, { "epoch": 0.2818105705684308, "grad_norm": 1.0955312214135637, "learning_rate": 1.685513864097158e-05, "loss": 0.4543, "step": 7910 }, { "epoch": 0.2818461976949249, "grad_norm": 1.482737511349584, "learning_rate": 1.6854298460674915e-05, "loss": 0.6011, "step": 7911 }, { "epoch": 0.28188182482141905, "grad_norm": 1.09810495364995, "learning_rate": 1.685345818910847e-05, "loss": 0.4737, "step": 7912 }, { "epoch": 0.2819174519479131, "grad_norm": 1.836947502615461, "learning_rate": 1.685261782628344e-05, "loss": 0.5951, "step": 7913 }, { "epoch": 0.28195307907440725, "grad_norm": 1.0895642610146696, "learning_rate": 1.6851777372211014e-05, "loss": 0.4546, "step": 7914 }, { "epoch": 0.2819887062009014, "grad_norm": 1.6929353485608163, "learning_rate": 1.6850936826902386e-05, "loss": 0.6317, "step": 7915 }, { "epoch": 0.28202433332739546, "grad_norm": 1.2792371824990474, "learning_rate": 1.685009619036874e-05, "loss": 0.5194, "step": 7916 }, { "epoch": 0.2820599604538896, "grad_norm": 1.0758188976644971, "learning_rate": 1.684925546262128e-05, "loss": 0.5275, "step": 7917 }, { "epoch": 0.2820955875803837, "grad_norm": 1.8568727119367459, "learning_rate": 1.6848414643671197e-05, "loss": 0.5881, "step": 7918 }, { "epoch": 0.2821312147068778, "grad_norm": 1.1336339286370285, "learning_rate": 1.684757373352968e-05, "loss": 0.4554, "step": 7919 }, { "epoch": 0.28216684183337193, "grad_norm": 1.6218801682835835, "learning_rate": 1.6846732732207938e-05, "loss": 0.634, "step": 7920 }, { "epoch": 0.28220246895986606, "grad_norm": 1.1733830275006745, "learning_rate": 1.6845891639717164e-05, "loss": 0.5018, "step": 7921 }, { "epoch": 0.28223809608636014, "grad_norm": 1.5134154353925215, "learning_rate": 1.6845050456068555e-05, "loss": 0.6074, "step": 7922 }, { "epoch": 0.28227372321285427, "grad_norm": 1.250589767504875, "learning_rate": 1.684420918127332e-05, "loss": 0.559, "step": 7923 }, { "epoch": 0.2823093503393484, "grad_norm": 1.9211247443063515, "learning_rate": 1.6843367815342648e-05, "loss": 0.6448, "step": 7924 }, { "epoch": 0.2823449774658425, "grad_norm": 1.1859685585992583, "learning_rate": 1.6842526358287755e-05, "loss": 0.5135, "step": 7925 }, { "epoch": 0.2823806045923366, "grad_norm": 1.2550164758284734, "learning_rate": 1.684168481011984e-05, "loss": 0.5087, "step": 7926 }, { "epoch": 0.28241623171883073, "grad_norm": 1.2623444710148375, "learning_rate": 1.684084317085011e-05, "loss": 0.5181, "step": 7927 }, { "epoch": 0.2824518588453248, "grad_norm": 1.546311967558438, "learning_rate": 1.684000144048977e-05, "loss": 0.5997, "step": 7928 }, { "epoch": 0.28248748597181894, "grad_norm": 1.7957120209947128, "learning_rate": 1.6839159619050036e-05, "loss": 0.6489, "step": 7929 }, { "epoch": 0.2825231130983131, "grad_norm": 1.648747110211265, "learning_rate": 1.6838317706542105e-05, "loss": 0.5238, "step": 7930 }, { "epoch": 0.28255874022480715, "grad_norm": 1.1910115266355104, "learning_rate": 1.6837475702977198e-05, "loss": 0.5205, "step": 7931 }, { "epoch": 0.2825943673513013, "grad_norm": 1.644985139720509, "learning_rate": 1.6836633608366518e-05, "loss": 0.5653, "step": 7932 }, { "epoch": 0.2826299944777954, "grad_norm": 1.8743476108430417, "learning_rate": 1.6835791422721287e-05, "loss": 0.592, "step": 7933 }, { "epoch": 0.2826656216042895, "grad_norm": 1.4800798331845533, "learning_rate": 1.6834949146052717e-05, "loss": 0.5873, "step": 7934 }, { "epoch": 0.2827012487307836, "grad_norm": 1.120655783443205, "learning_rate": 1.6834106778372018e-05, "loss": 0.4597, "step": 7935 }, { "epoch": 0.28273687585727775, "grad_norm": 1.6204727848363345, "learning_rate": 1.683326431969041e-05, "loss": 0.5781, "step": 7936 }, { "epoch": 0.2827725029837718, "grad_norm": 1.6706638171080075, "learning_rate": 1.6832421770019115e-05, "loss": 0.617, "step": 7937 }, { "epoch": 0.28280813011026595, "grad_norm": 2.0896332575122694, "learning_rate": 1.6831579129369347e-05, "loss": 0.5742, "step": 7938 }, { "epoch": 0.2828437572367601, "grad_norm": 1.6446981875475777, "learning_rate": 1.6830736397752326e-05, "loss": 0.6079, "step": 7939 }, { "epoch": 0.28287938436325416, "grad_norm": 1.6279572638481072, "learning_rate": 1.682989357517928e-05, "loss": 0.6123, "step": 7940 }, { "epoch": 0.2829150114897483, "grad_norm": 1.833752522689794, "learning_rate": 1.6829050661661424e-05, "loss": 0.6327, "step": 7941 }, { "epoch": 0.2829506386162424, "grad_norm": 2.1368064964897084, "learning_rate": 1.6828207657209982e-05, "loss": 0.6636, "step": 7942 }, { "epoch": 0.2829862657427365, "grad_norm": 1.1693424592527704, "learning_rate": 1.6827364561836187e-05, "loss": 0.4849, "step": 7943 }, { "epoch": 0.28302189286923063, "grad_norm": 1.234729544256134, "learning_rate": 1.682652137555126e-05, "loss": 0.5265, "step": 7944 }, { "epoch": 0.28305751999572476, "grad_norm": 1.540499794355009, "learning_rate": 1.6825678098366428e-05, "loss": 0.6156, "step": 7945 }, { "epoch": 0.28309314712221884, "grad_norm": 2.1499807819971757, "learning_rate": 1.6824834730292923e-05, "loss": 0.6519, "step": 7946 }, { "epoch": 0.28312877424871297, "grad_norm": 1.5086241285665054, "learning_rate": 1.6823991271341972e-05, "loss": 0.6266, "step": 7947 }, { "epoch": 0.2831644013752071, "grad_norm": 1.3126238275914397, "learning_rate": 1.682314772152481e-05, "loss": 0.5301, "step": 7948 }, { "epoch": 0.2832000285017012, "grad_norm": 1.0869204382842388, "learning_rate": 1.6822304080852663e-05, "loss": 0.4574, "step": 7949 }, { "epoch": 0.2832356556281953, "grad_norm": 1.6579498017129903, "learning_rate": 1.6821460349336773e-05, "loss": 0.578, "step": 7950 }, { "epoch": 0.28327128275468944, "grad_norm": 1.5629056629601596, "learning_rate": 1.682061652698837e-05, "loss": 0.5784, "step": 7951 }, { "epoch": 0.2833069098811835, "grad_norm": 1.174544969079865, "learning_rate": 1.6819772613818686e-05, "loss": 0.5255, "step": 7952 }, { "epoch": 0.28334253700767764, "grad_norm": 1.2009744917054153, "learning_rate": 1.6818928609838967e-05, "loss": 0.5458, "step": 7953 }, { "epoch": 0.2833781641341718, "grad_norm": 1.6487901279286994, "learning_rate": 1.6818084515060444e-05, "loss": 0.6783, "step": 7954 }, { "epoch": 0.28341379126066585, "grad_norm": 2.081944019229534, "learning_rate": 1.6817240329494365e-05, "loss": 0.5602, "step": 7955 }, { "epoch": 0.28344941838716, "grad_norm": 1.1544857751420832, "learning_rate": 1.681639605315196e-05, "loss": 0.5065, "step": 7956 }, { "epoch": 0.2834850455136541, "grad_norm": 2.4170146996163218, "learning_rate": 1.681555168604448e-05, "loss": 0.6974, "step": 7957 }, { "epoch": 0.2835206726401482, "grad_norm": 1.9137774936771546, "learning_rate": 1.681470722818317e-05, "loss": 0.5321, "step": 7958 }, { "epoch": 0.2835562997666423, "grad_norm": 1.8240891752821469, "learning_rate": 1.6813862679579265e-05, "loss": 0.6219, "step": 7959 }, { "epoch": 0.28359192689313645, "grad_norm": 1.2353359193015674, "learning_rate": 1.681301804024402e-05, "loss": 0.5118, "step": 7960 }, { "epoch": 0.2836275540196305, "grad_norm": 2.012952762705544, "learning_rate": 1.6812173310188674e-05, "loss": 0.6564, "step": 7961 }, { "epoch": 0.28366318114612465, "grad_norm": 1.8071665419897118, "learning_rate": 1.681132848942448e-05, "loss": 0.5929, "step": 7962 }, { "epoch": 0.2836988082726188, "grad_norm": 1.5604255954947104, "learning_rate": 1.6810483577962684e-05, "loss": 0.4834, "step": 7963 }, { "epoch": 0.28373443539911286, "grad_norm": 1.2524466626704902, "learning_rate": 1.6809638575814542e-05, "loss": 0.5965, "step": 7964 }, { "epoch": 0.283770062525607, "grad_norm": 1.2777558395682937, "learning_rate": 1.68087934829913e-05, "loss": 0.5207, "step": 7965 }, { "epoch": 0.2838056896521011, "grad_norm": 2.63513565728008, "learning_rate": 1.6807948299504216e-05, "loss": 0.6235, "step": 7966 }, { "epoch": 0.2838413167785952, "grad_norm": 1.0985387259271524, "learning_rate": 1.6807103025364542e-05, "loss": 0.5079, "step": 7967 }, { "epoch": 0.28387694390508933, "grad_norm": 1.2727817412358733, "learning_rate": 1.6806257660583534e-05, "loss": 0.459, "step": 7968 }, { "epoch": 0.28391257103158346, "grad_norm": 1.215173455605007, "learning_rate": 1.6805412205172443e-05, "loss": 0.4835, "step": 7969 }, { "epoch": 0.28394819815807754, "grad_norm": 1.1756495164207963, "learning_rate": 1.6804566659142536e-05, "loss": 0.5201, "step": 7970 }, { "epoch": 0.28398382528457167, "grad_norm": 1.0501499553361022, "learning_rate": 1.680372102250507e-05, "loss": 0.485, "step": 7971 }, { "epoch": 0.2840194524110658, "grad_norm": 1.820871543881506, "learning_rate": 1.6802875295271297e-05, "loss": 0.4893, "step": 7972 }, { "epoch": 0.2840550795375599, "grad_norm": 1.3049364025084866, "learning_rate": 1.680202947745249e-05, "loss": 0.5161, "step": 7973 }, { "epoch": 0.284090706664054, "grad_norm": 1.2145137944993067, "learning_rate": 1.6801183569059903e-05, "loss": 0.5121, "step": 7974 }, { "epoch": 0.28412633379054814, "grad_norm": 1.0870231281018938, "learning_rate": 1.6800337570104806e-05, "loss": 0.5281, "step": 7975 }, { "epoch": 0.2841619609170422, "grad_norm": 1.6572480476157725, "learning_rate": 1.679949148059846e-05, "loss": 0.6187, "step": 7976 }, { "epoch": 0.28419758804353634, "grad_norm": 1.594395686664621, "learning_rate": 1.679864530055213e-05, "loss": 0.6457, "step": 7977 }, { "epoch": 0.2842332151700305, "grad_norm": 2.0374486397091607, "learning_rate": 1.6797799029977086e-05, "loss": 0.6228, "step": 7978 }, { "epoch": 0.28426884229652455, "grad_norm": 3.016005106862865, "learning_rate": 1.6796952668884602e-05, "loss": 0.5489, "step": 7979 }, { "epoch": 0.2843044694230187, "grad_norm": 1.3790484602832187, "learning_rate": 1.6796106217285933e-05, "loss": 0.5593, "step": 7980 }, { "epoch": 0.2843400965495128, "grad_norm": 1.6665968207815662, "learning_rate": 1.6795259675192367e-05, "loss": 0.5988, "step": 7981 }, { "epoch": 0.2843757236760069, "grad_norm": 1.538459875865541, "learning_rate": 1.6794413042615168e-05, "loss": 0.6302, "step": 7982 }, { "epoch": 0.284411350802501, "grad_norm": 1.8145800940233723, "learning_rate": 1.6793566319565607e-05, "loss": 0.5393, "step": 7983 }, { "epoch": 0.28444697792899515, "grad_norm": 1.959614746265584, "learning_rate": 1.6792719506054967e-05, "loss": 0.608, "step": 7984 }, { "epoch": 0.2844826050554892, "grad_norm": 1.38694350205181, "learning_rate": 1.6791872602094513e-05, "loss": 0.4964, "step": 7985 }, { "epoch": 0.28451823218198335, "grad_norm": 1.4325947598875668, "learning_rate": 1.6791025607695532e-05, "loss": 0.5542, "step": 7986 }, { "epoch": 0.2845538593084775, "grad_norm": 1.7235097732775309, "learning_rate": 1.67901785228693e-05, "loss": 0.5804, "step": 7987 }, { "epoch": 0.28458948643497156, "grad_norm": 1.1826647271956956, "learning_rate": 1.678933134762709e-05, "loss": 0.527, "step": 7988 }, { "epoch": 0.2846251135614657, "grad_norm": 1.5496759584701334, "learning_rate": 1.678848408198019e-05, "loss": 0.5022, "step": 7989 }, { "epoch": 0.2846607406879598, "grad_norm": 1.2765642345130002, "learning_rate": 1.678763672593988e-05, "loss": 0.4329, "step": 7990 }, { "epoch": 0.2846963678144539, "grad_norm": 1.1257991206841953, "learning_rate": 1.678678927951744e-05, "loss": 0.4914, "step": 7991 }, { "epoch": 0.28473199494094803, "grad_norm": 1.21832935187292, "learning_rate": 1.678594174272416e-05, "loss": 0.4924, "step": 7992 }, { "epoch": 0.28476762206744216, "grad_norm": 1.6398658195171576, "learning_rate": 1.6785094115571323e-05, "loss": 0.5441, "step": 7993 }, { "epoch": 0.28480324919393624, "grad_norm": 2.544708903142105, "learning_rate": 1.6784246398070215e-05, "loss": 0.5777, "step": 7994 }, { "epoch": 0.28483887632043037, "grad_norm": 1.4268410833501592, "learning_rate": 1.678339859023212e-05, "loss": 0.5932, "step": 7995 }, { "epoch": 0.2848745034469245, "grad_norm": 1.9348177912218432, "learning_rate": 1.6782550692068338e-05, "loss": 0.6062, "step": 7996 }, { "epoch": 0.2849101305734186, "grad_norm": 1.5444106987801887, "learning_rate": 1.678170270359015e-05, "loss": 0.6304, "step": 7997 }, { "epoch": 0.2849457576999127, "grad_norm": 1.0628902496111334, "learning_rate": 1.678085462480885e-05, "loss": 0.5012, "step": 7998 }, { "epoch": 0.28498138482640684, "grad_norm": 2.0305064489593594, "learning_rate": 1.6780006455735735e-05, "loss": 0.6191, "step": 7999 }, { "epoch": 0.2850170119529009, "grad_norm": 1.8548906749221512, "learning_rate": 1.677915819638209e-05, "loss": 0.5762, "step": 8000 }, { "epoch": 0.28505263907939504, "grad_norm": 1.3602942850409243, "learning_rate": 1.6778309846759216e-05, "loss": 0.6446, "step": 8001 }, { "epoch": 0.2850882662058892, "grad_norm": 1.6218285717750163, "learning_rate": 1.6777461406878412e-05, "loss": 0.5733, "step": 8002 }, { "epoch": 0.28512389333238325, "grad_norm": 1.1080107034340991, "learning_rate": 1.6776612876750967e-05, "loss": 0.5409, "step": 8003 }, { "epoch": 0.2851595204588774, "grad_norm": 1.1932071486946156, "learning_rate": 1.6775764256388187e-05, "loss": 0.4808, "step": 8004 }, { "epoch": 0.2851951475853715, "grad_norm": 1.1341867417675664, "learning_rate": 1.6774915545801372e-05, "loss": 0.5496, "step": 8005 }, { "epoch": 0.2852307747118656, "grad_norm": 1.4037733430749828, "learning_rate": 1.6774066745001818e-05, "loss": 0.5194, "step": 8006 }, { "epoch": 0.2852664018383597, "grad_norm": 1.7793685023803572, "learning_rate": 1.6773217854000834e-05, "loss": 0.599, "step": 8007 }, { "epoch": 0.28530202896485385, "grad_norm": 1.3467766335971834, "learning_rate": 1.6772368872809716e-05, "loss": 0.572, "step": 8008 }, { "epoch": 0.285337656091348, "grad_norm": 1.017164640865475, "learning_rate": 1.6771519801439775e-05, "loss": 0.4859, "step": 8009 }, { "epoch": 0.28537328321784206, "grad_norm": 1.6950280203491723, "learning_rate": 1.6770670639902316e-05, "loss": 0.6026, "step": 8010 }, { "epoch": 0.2854089103443362, "grad_norm": 1.6893338650083567, "learning_rate": 1.6769821388208645e-05, "loss": 0.6039, "step": 8011 }, { "epoch": 0.2854445374708303, "grad_norm": 1.2119203718041076, "learning_rate": 1.6768972046370068e-05, "loss": 0.5332, "step": 8012 }, { "epoch": 0.2854801645973244, "grad_norm": 1.0968606100346103, "learning_rate": 1.67681226143979e-05, "loss": 0.4581, "step": 8013 }, { "epoch": 0.2855157917238185, "grad_norm": 1.614082513135156, "learning_rate": 1.6767273092303448e-05, "loss": 0.6495, "step": 8014 }, { "epoch": 0.28555141885031265, "grad_norm": 1.0873393669679976, "learning_rate": 1.6766423480098023e-05, "loss": 0.5406, "step": 8015 }, { "epoch": 0.28558704597680673, "grad_norm": 1.8799584502300628, "learning_rate": 1.676557377779294e-05, "loss": 0.6513, "step": 8016 }, { "epoch": 0.28562267310330086, "grad_norm": 1.018116148124435, "learning_rate": 1.6764723985399515e-05, "loss": 0.4588, "step": 8017 }, { "epoch": 0.285658300229795, "grad_norm": 1.059454035589224, "learning_rate": 1.6763874102929058e-05, "loss": 0.4695, "step": 8018 }, { "epoch": 0.28569392735628907, "grad_norm": 1.3204511559104912, "learning_rate": 1.6763024130392893e-05, "loss": 0.6463, "step": 8019 }, { "epoch": 0.2857295544827832, "grad_norm": 1.6586074816490473, "learning_rate": 1.6762174067802334e-05, "loss": 0.6302, "step": 8020 }, { "epoch": 0.28576518160927733, "grad_norm": 1.0526121411364866, "learning_rate": 1.6761323915168702e-05, "loss": 0.4872, "step": 8021 }, { "epoch": 0.2858008087357714, "grad_norm": 1.9676635012665447, "learning_rate": 1.6760473672503316e-05, "loss": 0.6611, "step": 8022 }, { "epoch": 0.28583643586226554, "grad_norm": 1.1976436048005525, "learning_rate": 1.6759623339817497e-05, "loss": 0.4703, "step": 8023 }, { "epoch": 0.28587206298875967, "grad_norm": 1.1130067202078302, "learning_rate": 1.675877291712257e-05, "loss": 0.4824, "step": 8024 }, { "epoch": 0.28590769011525374, "grad_norm": 1.080377602742229, "learning_rate": 1.6757922404429854e-05, "loss": 0.4777, "step": 8025 }, { "epoch": 0.2859433172417479, "grad_norm": 1.3477566585093441, "learning_rate": 1.675707180175068e-05, "loss": 0.4893, "step": 8026 }, { "epoch": 0.285978944368242, "grad_norm": 1.6907649899476562, "learning_rate": 1.6756221109096375e-05, "loss": 0.4568, "step": 8027 }, { "epoch": 0.2860145714947361, "grad_norm": 1.3381736095353411, "learning_rate": 1.675537032647826e-05, "loss": 0.525, "step": 8028 }, { "epoch": 0.2860501986212302, "grad_norm": 1.7869557241527392, "learning_rate": 1.6754519453907666e-05, "loss": 0.5564, "step": 8029 }, { "epoch": 0.28608582574772434, "grad_norm": 1.4356000482095548, "learning_rate": 1.6753668491395923e-05, "loss": 0.6058, "step": 8030 }, { "epoch": 0.2861214528742184, "grad_norm": 2.0589951277831546, "learning_rate": 1.675281743895437e-05, "loss": 0.5479, "step": 8031 }, { "epoch": 0.28615708000071255, "grad_norm": 1.0744410778880071, "learning_rate": 1.6751966296594328e-05, "loss": 0.5242, "step": 8032 }, { "epoch": 0.2861927071272067, "grad_norm": 1.0491842582187119, "learning_rate": 1.675111506432714e-05, "loss": 0.4815, "step": 8033 }, { "epoch": 0.28622833425370076, "grad_norm": 1.1449435471627865, "learning_rate": 1.675026374216413e-05, "loss": 0.4883, "step": 8034 }, { "epoch": 0.2862639613801949, "grad_norm": 1.0340221629599071, "learning_rate": 1.6749412330116643e-05, "loss": 0.4901, "step": 8035 }, { "epoch": 0.286299588506689, "grad_norm": 1.2137603227272795, "learning_rate": 1.6748560828196013e-05, "loss": 0.5051, "step": 8036 }, { "epoch": 0.2863352156331831, "grad_norm": 1.1399940576125906, "learning_rate": 1.674770923641358e-05, "loss": 0.5215, "step": 8037 }, { "epoch": 0.2863708427596772, "grad_norm": 1.74934039410919, "learning_rate": 1.674685755478068e-05, "loss": 0.6173, "step": 8038 }, { "epoch": 0.28640646988617136, "grad_norm": 1.6960604953602092, "learning_rate": 1.674600578330866e-05, "loss": 0.5713, "step": 8039 }, { "epoch": 0.28644209701266543, "grad_norm": 1.5860032090080767, "learning_rate": 1.6745153922008853e-05, "loss": 0.6445, "step": 8040 }, { "epoch": 0.28647772413915956, "grad_norm": 1.7443314632993008, "learning_rate": 1.6744301970892608e-05, "loss": 0.5984, "step": 8041 }, { "epoch": 0.2865133512656537, "grad_norm": 1.5654525065742815, "learning_rate": 1.674344992997127e-05, "loss": 0.5689, "step": 8042 }, { "epoch": 0.28654897839214777, "grad_norm": 1.1957035150512019, "learning_rate": 1.6742597799256182e-05, "loss": 0.4966, "step": 8043 }, { "epoch": 0.2865846055186419, "grad_norm": 1.7474584945454195, "learning_rate": 1.674174557875869e-05, "loss": 0.5577, "step": 8044 }, { "epoch": 0.28662023264513603, "grad_norm": 1.5567542751858494, "learning_rate": 1.6740893268490147e-05, "loss": 0.6127, "step": 8045 }, { "epoch": 0.2866558597716301, "grad_norm": 2.019403150841141, "learning_rate": 1.6740040868461897e-05, "loss": 0.5818, "step": 8046 }, { "epoch": 0.28669148689812424, "grad_norm": 1.6088902686284572, "learning_rate": 1.673918837868529e-05, "loss": 0.558, "step": 8047 }, { "epoch": 0.28672711402461837, "grad_norm": 1.058662521664243, "learning_rate": 1.673833579917168e-05, "loss": 0.4843, "step": 8048 }, { "epoch": 0.28676274115111244, "grad_norm": 1.8709627326956724, "learning_rate": 1.673748312993242e-05, "loss": 0.568, "step": 8049 }, { "epoch": 0.2867983682776066, "grad_norm": 1.7262354369393056, "learning_rate": 1.6736630370978862e-05, "loss": 0.5794, "step": 8050 }, { "epoch": 0.2868339954041007, "grad_norm": 1.9485192502259996, "learning_rate": 1.6735777522322362e-05, "loss": 0.5835, "step": 8051 }, { "epoch": 0.2868696225305948, "grad_norm": 1.7815934600669245, "learning_rate": 1.6734924583974275e-05, "loss": 0.6585, "step": 8052 }, { "epoch": 0.2869052496570889, "grad_norm": 1.145024932919087, "learning_rate": 1.6734071555945965e-05, "loss": 0.4799, "step": 8053 }, { "epoch": 0.28694087678358304, "grad_norm": 2.747316858979222, "learning_rate": 1.6733218438248782e-05, "loss": 0.6137, "step": 8054 }, { "epoch": 0.2869765039100771, "grad_norm": 1.3485515874766651, "learning_rate": 1.673236523089409e-05, "loss": 0.4867, "step": 8055 }, { "epoch": 0.28701213103657125, "grad_norm": 1.7418343743798574, "learning_rate": 1.6731511933893246e-05, "loss": 0.645, "step": 8056 }, { "epoch": 0.2870477581630654, "grad_norm": 1.5441272051761963, "learning_rate": 1.6730658547257617e-05, "loss": 0.5989, "step": 8057 }, { "epoch": 0.28708338528955946, "grad_norm": 1.4816800878888383, "learning_rate": 1.6729805070998567e-05, "loss": 0.5492, "step": 8058 }, { "epoch": 0.2871190124160536, "grad_norm": 1.8874312742211143, "learning_rate": 1.672895150512746e-05, "loss": 0.6178, "step": 8059 }, { "epoch": 0.2871546395425477, "grad_norm": 1.7173037472726833, "learning_rate": 1.6728097849655654e-05, "loss": 0.7274, "step": 8060 }, { "epoch": 0.2871902666690418, "grad_norm": 1.081458168456495, "learning_rate": 1.672724410459453e-05, "loss": 0.4872, "step": 8061 }, { "epoch": 0.2872258937955359, "grad_norm": 1.214090308888834, "learning_rate": 1.6726390269955444e-05, "loss": 0.5097, "step": 8062 }, { "epoch": 0.28726152092203006, "grad_norm": 1.1891498918062013, "learning_rate": 1.672553634574977e-05, "loss": 0.5305, "step": 8063 }, { "epoch": 0.28729714804852413, "grad_norm": 1.0786136107028856, "learning_rate": 1.6724682331988882e-05, "loss": 0.5125, "step": 8064 }, { "epoch": 0.28733277517501826, "grad_norm": 0.9972378375140255, "learning_rate": 1.6723828228684144e-05, "loss": 0.4987, "step": 8065 }, { "epoch": 0.2873684023015124, "grad_norm": 1.1704554488044858, "learning_rate": 1.6722974035846937e-05, "loss": 0.5007, "step": 8066 }, { "epoch": 0.28740402942800647, "grad_norm": 1.6627305022816783, "learning_rate": 1.672211975348863e-05, "loss": 0.6236, "step": 8067 }, { "epoch": 0.2874396565545006, "grad_norm": 1.5185749522840783, "learning_rate": 1.67212653816206e-05, "loss": 0.6324, "step": 8068 }, { "epoch": 0.28747528368099473, "grad_norm": 1.6062832415823862, "learning_rate": 1.6720410920254227e-05, "loss": 0.5419, "step": 8069 }, { "epoch": 0.2875109108074888, "grad_norm": 1.4649232153026288, "learning_rate": 1.671955636940088e-05, "loss": 0.6002, "step": 8070 }, { "epoch": 0.28754653793398294, "grad_norm": 1.1441960819280919, "learning_rate": 1.6718701729071946e-05, "loss": 0.4883, "step": 8071 }, { "epoch": 0.28758216506047707, "grad_norm": 1.1374526106971505, "learning_rate": 1.6717846999278805e-05, "loss": 0.5272, "step": 8072 }, { "epoch": 0.28761779218697114, "grad_norm": 1.1384499175908305, "learning_rate": 1.6716992180032832e-05, "loss": 0.473, "step": 8073 }, { "epoch": 0.2876534193134653, "grad_norm": 1.6071758710523212, "learning_rate": 1.671613727134541e-05, "loss": 0.6407, "step": 8074 }, { "epoch": 0.2876890464399594, "grad_norm": 1.5811470983286664, "learning_rate": 1.671528227322793e-05, "loss": 0.5944, "step": 8075 }, { "epoch": 0.2877246735664535, "grad_norm": 1.1888088915409678, "learning_rate": 1.6714427185691775e-05, "loss": 0.515, "step": 8076 }, { "epoch": 0.2877603006929476, "grad_norm": 1.2329977774918262, "learning_rate": 1.6713572008748328e-05, "loss": 0.5656, "step": 8077 }, { "epoch": 0.28779592781944174, "grad_norm": 1.9901757064724697, "learning_rate": 1.6712716742408973e-05, "loss": 0.5451, "step": 8078 }, { "epoch": 0.2878315549459358, "grad_norm": 1.0780954127437063, "learning_rate": 1.6711861386685108e-05, "loss": 0.4478, "step": 8079 }, { "epoch": 0.28786718207242995, "grad_norm": 1.1638656569752541, "learning_rate": 1.6711005941588113e-05, "loss": 0.4812, "step": 8080 }, { "epoch": 0.2879028091989241, "grad_norm": 1.2606915990822183, "learning_rate": 1.6710150407129386e-05, "loss": 0.5603, "step": 8081 }, { "epoch": 0.28793843632541816, "grad_norm": 1.6131470655224374, "learning_rate": 1.6709294783320314e-05, "loss": 0.5347, "step": 8082 }, { "epoch": 0.2879740634519123, "grad_norm": 1.2456963675251351, "learning_rate": 1.6708439070172296e-05, "loss": 0.5051, "step": 8083 }, { "epoch": 0.2880096905784064, "grad_norm": 1.0631064373298926, "learning_rate": 1.6707583267696722e-05, "loss": 0.4888, "step": 8084 }, { "epoch": 0.2880453177049005, "grad_norm": 1.3694383879552858, "learning_rate": 1.6706727375904983e-05, "loss": 0.5517, "step": 8085 }, { "epoch": 0.2880809448313946, "grad_norm": 1.1465722459039132, "learning_rate": 1.6705871394808487e-05, "loss": 0.4814, "step": 8086 }, { "epoch": 0.28811657195788876, "grad_norm": 1.5192984975996633, "learning_rate": 1.6705015324418624e-05, "loss": 0.5389, "step": 8087 }, { "epoch": 0.28815219908438283, "grad_norm": 1.5826708712488415, "learning_rate": 1.6704159164746797e-05, "loss": 0.5922, "step": 8088 }, { "epoch": 0.28818782621087696, "grad_norm": 1.5992058714691704, "learning_rate": 1.67033029158044e-05, "loss": 0.5475, "step": 8089 }, { "epoch": 0.2882234533373711, "grad_norm": 1.6805804176734078, "learning_rate": 1.6702446577602846e-05, "loss": 0.7037, "step": 8090 }, { "epoch": 0.28825908046386517, "grad_norm": 1.1891411586649088, "learning_rate": 1.670159015015353e-05, "loss": 0.4922, "step": 8091 }, { "epoch": 0.2882947075903593, "grad_norm": 1.2381825517907548, "learning_rate": 1.6700733633467855e-05, "loss": 0.5314, "step": 8092 }, { "epoch": 0.28833033471685343, "grad_norm": 1.1445848401407854, "learning_rate": 1.6699877027557227e-05, "loss": 0.4721, "step": 8093 }, { "epoch": 0.2883659618433475, "grad_norm": 1.4555670538220409, "learning_rate": 1.6699020332433054e-05, "loss": 0.6317, "step": 8094 }, { "epoch": 0.28840158896984164, "grad_norm": 1.0653650037147235, "learning_rate": 1.6698163548106744e-05, "loss": 0.5082, "step": 8095 }, { "epoch": 0.28843721609633577, "grad_norm": 1.5837566339208369, "learning_rate": 1.6697306674589704e-05, "loss": 0.5818, "step": 8096 }, { "epoch": 0.28847284322282984, "grad_norm": 1.1994340004490593, "learning_rate": 1.6696449711893346e-05, "loss": 0.4989, "step": 8097 }, { "epoch": 0.288508470349324, "grad_norm": 1.1455911237675667, "learning_rate": 1.6695592660029077e-05, "loss": 0.5481, "step": 8098 }, { "epoch": 0.2885440974758181, "grad_norm": 1.7338099930835016, "learning_rate": 1.6694735519008312e-05, "loss": 0.5344, "step": 8099 }, { "epoch": 0.2885797246023122, "grad_norm": 1.1204585516990921, "learning_rate": 1.6693878288842466e-05, "loss": 0.4772, "step": 8100 }, { "epoch": 0.2886153517288063, "grad_norm": 1.4812348727486977, "learning_rate": 1.6693020969542953e-05, "loss": 0.5723, "step": 8101 }, { "epoch": 0.28865097885530044, "grad_norm": 1.6181369154984608, "learning_rate": 1.6692163561121185e-05, "loss": 0.5037, "step": 8102 }, { "epoch": 0.2886866059817945, "grad_norm": 1.9244441690657168, "learning_rate": 1.6691306063588583e-05, "loss": 0.6664, "step": 8103 }, { "epoch": 0.28872223310828865, "grad_norm": 1.3932547391472268, "learning_rate": 1.6690448476956565e-05, "loss": 0.4728, "step": 8104 }, { "epoch": 0.2887578602347828, "grad_norm": 2.508436084196789, "learning_rate": 1.6689590801236545e-05, "loss": 0.5548, "step": 8105 }, { "epoch": 0.28879348736127686, "grad_norm": 1.3427372011308587, "learning_rate": 1.6688733036439952e-05, "loss": 0.6329, "step": 8106 }, { "epoch": 0.288829114487771, "grad_norm": 1.8318301007749036, "learning_rate": 1.66878751825782e-05, "loss": 0.5951, "step": 8107 }, { "epoch": 0.2888647416142651, "grad_norm": 1.1026521375551874, "learning_rate": 1.668701723966272e-05, "loss": 0.5256, "step": 8108 }, { "epoch": 0.2889003687407592, "grad_norm": 2.0196245328759157, "learning_rate": 1.6686159207704927e-05, "loss": 0.6316, "step": 8109 }, { "epoch": 0.2889359958672533, "grad_norm": 1.8352299429058614, "learning_rate": 1.6685301086716252e-05, "loss": 0.5413, "step": 8110 }, { "epoch": 0.28897162299374746, "grad_norm": 1.1580777748741866, "learning_rate": 1.6684442876708124e-05, "loss": 0.5146, "step": 8111 }, { "epoch": 0.28900725012024153, "grad_norm": 1.1130534285831377, "learning_rate": 1.668358457769196e-05, "loss": 0.5345, "step": 8112 }, { "epoch": 0.28904287724673566, "grad_norm": 1.1387342403042031, "learning_rate": 1.66827261896792e-05, "loss": 0.4837, "step": 8113 }, { "epoch": 0.2890785043732298, "grad_norm": 1.4758552246010201, "learning_rate": 1.6681867712681274e-05, "loss": 0.5836, "step": 8114 }, { "epoch": 0.28911413149972387, "grad_norm": 1.3607291384053526, "learning_rate": 1.66810091467096e-05, "loss": 0.5571, "step": 8115 }, { "epoch": 0.289149758626218, "grad_norm": 1.4450818210114071, "learning_rate": 1.6680150491775626e-05, "loss": 0.5747, "step": 8116 }, { "epoch": 0.28918538575271213, "grad_norm": 1.1034554787295439, "learning_rate": 1.6679291747890777e-05, "loss": 0.4924, "step": 8117 }, { "epoch": 0.2892210128792062, "grad_norm": 1.6915641114093427, "learning_rate": 1.6678432915066488e-05, "loss": 0.5294, "step": 8118 }, { "epoch": 0.28925664000570034, "grad_norm": 1.0899793987987432, "learning_rate": 1.66775739933142e-05, "loss": 0.5199, "step": 8119 }, { "epoch": 0.28929226713219447, "grad_norm": 1.0624395039804466, "learning_rate": 1.6676714982645345e-05, "loss": 0.4361, "step": 8120 }, { "epoch": 0.28932789425868854, "grad_norm": 1.4386324760026274, "learning_rate": 1.6675855883071364e-05, "loss": 0.5356, "step": 8121 }, { "epoch": 0.2893635213851827, "grad_norm": 2.050653941791776, "learning_rate": 1.6674996694603694e-05, "loss": 0.6453, "step": 8122 }, { "epoch": 0.2893991485116768, "grad_norm": 1.288451753794984, "learning_rate": 1.667413741725378e-05, "loss": 0.486, "step": 8123 }, { "epoch": 0.2894347756381709, "grad_norm": 1.1853495393368776, "learning_rate": 1.6673278051033062e-05, "loss": 0.5407, "step": 8124 }, { "epoch": 0.289470402764665, "grad_norm": 1.8179887884792127, "learning_rate": 1.667241859595298e-05, "loss": 0.6493, "step": 8125 }, { "epoch": 0.28950602989115914, "grad_norm": 1.9312765816150343, "learning_rate": 1.6671559052024983e-05, "loss": 0.5653, "step": 8126 }, { "epoch": 0.2895416570176532, "grad_norm": 1.4771824892040868, "learning_rate": 1.667069941926051e-05, "loss": 0.5691, "step": 8127 }, { "epoch": 0.28957728414414735, "grad_norm": 1.1568872735591444, "learning_rate": 1.6669839697671014e-05, "loss": 0.4929, "step": 8128 }, { "epoch": 0.2896129112706415, "grad_norm": 1.194437892858589, "learning_rate": 1.666897988726794e-05, "loss": 0.5253, "step": 8129 }, { "epoch": 0.28964853839713556, "grad_norm": 1.6302757071009684, "learning_rate": 1.666811998806274e-05, "loss": 0.648, "step": 8130 }, { "epoch": 0.2896841655236297, "grad_norm": 1.1626709847104837, "learning_rate": 1.666726000006686e-05, "loss": 0.4624, "step": 8131 }, { "epoch": 0.2897197926501238, "grad_norm": 1.7450518439075973, "learning_rate": 1.6666399923291748e-05, "loss": 0.5757, "step": 8132 }, { "epoch": 0.2897554197766179, "grad_norm": 1.150127301765532, "learning_rate": 1.6665539757748866e-05, "loss": 0.5174, "step": 8133 }, { "epoch": 0.289791046903112, "grad_norm": 1.0952305041224049, "learning_rate": 1.6664679503449668e-05, "loss": 0.5063, "step": 8134 }, { "epoch": 0.28982667402960616, "grad_norm": 2.1230778684375657, "learning_rate": 1.6663819160405597e-05, "loss": 0.5269, "step": 8135 }, { "epoch": 0.28986230115610023, "grad_norm": 1.6950220953782327, "learning_rate": 1.666295872862812e-05, "loss": 0.6458, "step": 8136 }, { "epoch": 0.28989792828259436, "grad_norm": 1.0726235603496006, "learning_rate": 1.666209820812869e-05, "loss": 0.511, "step": 8137 }, { "epoch": 0.2899335554090885, "grad_norm": 1.2462063766054035, "learning_rate": 1.6661237598918766e-05, "loss": 0.543, "step": 8138 }, { "epoch": 0.28996918253558257, "grad_norm": 1.6521233958040267, "learning_rate": 1.66603769010098e-05, "loss": 0.5741, "step": 8139 }, { "epoch": 0.2900048096620767, "grad_norm": 1.6512311240739022, "learning_rate": 1.6659516114413273e-05, "loss": 0.6653, "step": 8140 }, { "epoch": 0.29004043678857083, "grad_norm": 1.1081519191904465, "learning_rate": 1.6658655239140627e-05, "loss": 0.4727, "step": 8141 }, { "epoch": 0.2900760639150649, "grad_norm": 1.0865743910912489, "learning_rate": 1.665779427520333e-05, "loss": 0.4553, "step": 8142 }, { "epoch": 0.29011169104155904, "grad_norm": 1.042772599755676, "learning_rate": 1.6656933222612856e-05, "loss": 0.4747, "step": 8143 }, { "epoch": 0.29014731816805317, "grad_norm": 1.1767193628680161, "learning_rate": 1.6656072081380657e-05, "loss": 0.5199, "step": 8144 }, { "epoch": 0.29018294529454725, "grad_norm": 1.4156204806887596, "learning_rate": 1.6655210851518207e-05, "loss": 0.6304, "step": 8145 }, { "epoch": 0.2902185724210414, "grad_norm": 1.1234088619958185, "learning_rate": 1.6654349533036975e-05, "loss": 0.5209, "step": 8146 }, { "epoch": 0.2902541995475355, "grad_norm": 1.1206235936178999, "learning_rate": 1.6653488125948425e-05, "loss": 0.4685, "step": 8147 }, { "epoch": 0.2902898266740296, "grad_norm": 1.6098018192188006, "learning_rate": 1.665262663026403e-05, "loss": 0.5493, "step": 8148 }, { "epoch": 0.2903254538005237, "grad_norm": 1.3055758310738765, "learning_rate": 1.6651765045995264e-05, "loss": 0.5966, "step": 8149 }, { "epoch": 0.29036108092701784, "grad_norm": 1.3733186905006014, "learning_rate": 1.66509033731536e-05, "loss": 0.6198, "step": 8150 }, { "epoch": 0.2903967080535119, "grad_norm": 1.1833104354325372, "learning_rate": 1.6650041611750505e-05, "loss": 0.5095, "step": 8151 }, { "epoch": 0.29043233518000605, "grad_norm": 1.6891787268189866, "learning_rate": 1.6649179761797457e-05, "loss": 0.6271, "step": 8152 }, { "epoch": 0.2904679623065002, "grad_norm": 1.371747899665676, "learning_rate": 1.6648317823305933e-05, "loss": 0.6455, "step": 8153 }, { "epoch": 0.29050358943299426, "grad_norm": 1.19070916550566, "learning_rate": 1.664745579628741e-05, "loss": 0.4595, "step": 8154 }, { "epoch": 0.2905392165594884, "grad_norm": 1.821115468185473, "learning_rate": 1.6646593680753368e-05, "loss": 0.6509, "step": 8155 }, { "epoch": 0.2905748436859825, "grad_norm": 1.4467619012926238, "learning_rate": 1.6645731476715286e-05, "loss": 0.5599, "step": 8156 }, { "epoch": 0.2906104708124766, "grad_norm": 1.2017534893828112, "learning_rate": 1.6644869184184644e-05, "loss": 0.5113, "step": 8157 }, { "epoch": 0.2906460979389707, "grad_norm": 1.2884619213432382, "learning_rate": 1.6644006803172926e-05, "loss": 0.5473, "step": 8158 }, { "epoch": 0.29068172506546486, "grad_norm": 1.0988684156357282, "learning_rate": 1.664314433369161e-05, "loss": 0.4951, "step": 8159 }, { "epoch": 0.29071735219195893, "grad_norm": 1.4558603747476546, "learning_rate": 1.6642281775752187e-05, "loss": 0.6156, "step": 8160 }, { "epoch": 0.29075297931845306, "grad_norm": 1.694329810560556, "learning_rate": 1.6641419129366137e-05, "loss": 0.662, "step": 8161 }, { "epoch": 0.2907886064449472, "grad_norm": 1.0524052376627056, "learning_rate": 1.6640556394544953e-05, "loss": 0.4921, "step": 8162 }, { "epoch": 0.29082423357144127, "grad_norm": 1.347240101242719, "learning_rate": 1.6639693571300115e-05, "loss": 0.6414, "step": 8163 }, { "epoch": 0.2908598606979354, "grad_norm": 1.1471936595424694, "learning_rate": 1.6638830659643116e-05, "loss": 0.4542, "step": 8164 }, { "epoch": 0.29089548782442953, "grad_norm": 1.8069808861267316, "learning_rate": 1.663796765958545e-05, "loss": 0.7073, "step": 8165 }, { "epoch": 0.2909311149509236, "grad_norm": 1.7822677642921094, "learning_rate": 1.6637104571138603e-05, "loss": 0.5442, "step": 8166 }, { "epoch": 0.29096674207741774, "grad_norm": 1.5087483216474031, "learning_rate": 1.663624139431407e-05, "loss": 0.6204, "step": 8167 }, { "epoch": 0.29100236920391187, "grad_norm": 1.0595889410970136, "learning_rate": 1.6635378129123342e-05, "loss": 0.4602, "step": 8168 }, { "epoch": 0.29103799633040595, "grad_norm": 2.1395270888807167, "learning_rate": 1.663451477557792e-05, "loss": 0.6288, "step": 8169 }, { "epoch": 0.2910736234569001, "grad_norm": 1.5855619383266197, "learning_rate": 1.6633651333689296e-05, "loss": 0.6009, "step": 8170 }, { "epoch": 0.2911092505833942, "grad_norm": 1.111833020414936, "learning_rate": 1.6632787803468968e-05, "loss": 0.4817, "step": 8171 }, { "epoch": 0.2911448777098883, "grad_norm": 1.1535920007946494, "learning_rate": 1.6631924184928432e-05, "loss": 0.5237, "step": 8172 }, { "epoch": 0.2911805048363824, "grad_norm": 1.6364776041011586, "learning_rate": 1.663106047807919e-05, "loss": 0.6558, "step": 8173 }, { "epoch": 0.29121613196287655, "grad_norm": 1.0958580191982268, "learning_rate": 1.6630196682932746e-05, "loss": 0.5217, "step": 8174 }, { "epoch": 0.2912517590893706, "grad_norm": 1.4003823142429752, "learning_rate": 1.6629332799500593e-05, "loss": 0.5911, "step": 8175 }, { "epoch": 0.29128738621586475, "grad_norm": 1.7654970849920901, "learning_rate": 1.6628468827794245e-05, "loss": 0.5772, "step": 8176 }, { "epoch": 0.2913230133423589, "grad_norm": 2.163608482733814, "learning_rate": 1.6627604767825203e-05, "loss": 0.6508, "step": 8177 }, { "epoch": 0.29135864046885296, "grad_norm": 2.2692052127937754, "learning_rate": 1.6626740619604967e-05, "loss": 0.5686, "step": 8178 }, { "epoch": 0.2913942675953471, "grad_norm": 1.2210637655393897, "learning_rate": 1.662587638314505e-05, "loss": 0.4896, "step": 8179 }, { "epoch": 0.2914298947218412, "grad_norm": 1.1861042211361263, "learning_rate": 1.6625012058456956e-05, "loss": 0.5055, "step": 8180 }, { "epoch": 0.2914655218483353, "grad_norm": 1.8958874102186083, "learning_rate": 1.66241476455522e-05, "loss": 0.5586, "step": 8181 }, { "epoch": 0.2915011489748294, "grad_norm": 1.0696689984610899, "learning_rate": 1.6623283144442283e-05, "loss": 0.5059, "step": 8182 }, { "epoch": 0.29153677610132356, "grad_norm": 1.4246737047232516, "learning_rate": 1.6622418555138727e-05, "loss": 0.636, "step": 8183 }, { "epoch": 0.29157240322781763, "grad_norm": 1.3635908853290666, "learning_rate": 1.6621553877653034e-05, "loss": 0.526, "step": 8184 }, { "epoch": 0.29160803035431176, "grad_norm": 1.1405749279194475, "learning_rate": 1.6620689111996724e-05, "loss": 0.4909, "step": 8185 }, { "epoch": 0.2916436574808059, "grad_norm": 1.6923654880260044, "learning_rate": 1.6619824258181313e-05, "loss": 0.5725, "step": 8186 }, { "epoch": 0.29167928460729997, "grad_norm": 1.1440032017091235, "learning_rate": 1.6618959316218313e-05, "loss": 0.5102, "step": 8187 }, { "epoch": 0.2917149117337941, "grad_norm": 1.0940356927472572, "learning_rate": 1.6618094286119245e-05, "loss": 0.47, "step": 8188 }, { "epoch": 0.29175053886028823, "grad_norm": 1.0472227581804023, "learning_rate": 1.661722916789562e-05, "loss": 0.4646, "step": 8189 }, { "epoch": 0.2917861659867823, "grad_norm": 1.7480916887089764, "learning_rate": 1.6616363961558972e-05, "loss": 0.6782, "step": 8190 }, { "epoch": 0.29182179311327644, "grad_norm": 1.45986942217875, "learning_rate": 1.661549866712081e-05, "loss": 0.5982, "step": 8191 }, { "epoch": 0.29185742023977057, "grad_norm": 1.1988083380411954, "learning_rate": 1.661463328459266e-05, "loss": 0.5042, "step": 8192 }, { "epoch": 0.29189304736626465, "grad_norm": 1.3728933295886088, "learning_rate": 1.6613767813986045e-05, "loss": 0.5741, "step": 8193 }, { "epoch": 0.2919286744927588, "grad_norm": 1.1473604316361508, "learning_rate": 1.6612902255312485e-05, "loss": 0.4814, "step": 8194 }, { "epoch": 0.2919643016192529, "grad_norm": 1.4980610191960082, "learning_rate": 1.6612036608583514e-05, "loss": 0.6161, "step": 8195 }, { "epoch": 0.29199992874574704, "grad_norm": 2.008154133540382, "learning_rate": 1.6611170873810653e-05, "loss": 0.5856, "step": 8196 }, { "epoch": 0.2920355558722411, "grad_norm": 1.529481871180861, "learning_rate": 1.6610305051005427e-05, "loss": 0.5573, "step": 8197 }, { "epoch": 0.29207118299873525, "grad_norm": 1.603295560663468, "learning_rate": 1.660943914017937e-05, "loss": 0.5619, "step": 8198 }, { "epoch": 0.2921068101252294, "grad_norm": 1.13083021735348, "learning_rate": 1.6608573141344014e-05, "loss": 0.4686, "step": 8199 }, { "epoch": 0.29214243725172345, "grad_norm": 1.2438194966995966, "learning_rate": 1.6607707054510886e-05, "loss": 0.4714, "step": 8200 }, { "epoch": 0.2921780643782176, "grad_norm": 2.1921403541628237, "learning_rate": 1.660684087969152e-05, "loss": 0.6022, "step": 8201 }, { "epoch": 0.2922136915047117, "grad_norm": 1.0592488849761506, "learning_rate": 1.660597461689745e-05, "loss": 0.4361, "step": 8202 }, { "epoch": 0.2922493186312058, "grad_norm": 1.324288469538767, "learning_rate": 1.6605108266140208e-05, "loss": 0.5129, "step": 8203 }, { "epoch": 0.2922849457576999, "grad_norm": 1.4630635074810174, "learning_rate": 1.660424182743134e-05, "loss": 0.6581, "step": 8204 }, { "epoch": 0.29232057288419405, "grad_norm": 1.0236985400754839, "learning_rate": 1.6603375300782368e-05, "loss": 0.4528, "step": 8205 }, { "epoch": 0.2923562000106881, "grad_norm": 1.2305255135729403, "learning_rate": 1.660250868620484e-05, "loss": 0.5709, "step": 8206 }, { "epoch": 0.29239182713718226, "grad_norm": 1.0492485238936904, "learning_rate": 1.6601641983710298e-05, "loss": 0.5127, "step": 8207 }, { "epoch": 0.2924274542636764, "grad_norm": 1.068348284302056, "learning_rate": 1.6600775193310276e-05, "loss": 0.4471, "step": 8208 }, { "epoch": 0.29246308139017047, "grad_norm": 1.2079655708948958, "learning_rate": 1.659990831501632e-05, "loss": 0.5384, "step": 8209 }, { "epoch": 0.2924987085166646, "grad_norm": 1.3047373600443046, "learning_rate": 1.659904134883997e-05, "loss": 0.4955, "step": 8210 }, { "epoch": 0.2925343356431587, "grad_norm": 1.0776565750163247, "learning_rate": 1.6598174294792774e-05, "loss": 0.4806, "step": 8211 }, { "epoch": 0.2925699627696528, "grad_norm": 1.2208161886106497, "learning_rate": 1.6597307152886273e-05, "loss": 0.5299, "step": 8212 }, { "epoch": 0.29260558989614693, "grad_norm": 1.151159229460581, "learning_rate": 1.6596439923132016e-05, "loss": 0.4977, "step": 8213 }, { "epoch": 0.29264121702264106, "grad_norm": 1.1097937412091674, "learning_rate": 1.659557260554155e-05, "loss": 0.5129, "step": 8214 }, { "epoch": 0.29267684414913514, "grad_norm": 1.1486682078789587, "learning_rate": 1.659470520012643e-05, "loss": 0.4584, "step": 8215 }, { "epoch": 0.29271247127562927, "grad_norm": 1.0129042019801564, "learning_rate": 1.65938377068982e-05, "loss": 0.4843, "step": 8216 }, { "epoch": 0.2927480984021234, "grad_norm": 1.020641131236499, "learning_rate": 1.6592970125868405e-05, "loss": 0.5115, "step": 8217 }, { "epoch": 0.2927837255286175, "grad_norm": 1.5366713201460074, "learning_rate": 1.6592102457048613e-05, "loss": 0.5326, "step": 8218 }, { "epoch": 0.2928193526551116, "grad_norm": 1.1359023238879458, "learning_rate": 1.6591234700450366e-05, "loss": 0.5012, "step": 8219 }, { "epoch": 0.29285497978160574, "grad_norm": 1.7005404773012394, "learning_rate": 1.659036685608522e-05, "loss": 0.5974, "step": 8220 }, { "epoch": 0.2928906069080998, "grad_norm": 1.3495422857022896, "learning_rate": 1.6589498923964734e-05, "loss": 0.4986, "step": 8221 }, { "epoch": 0.29292623403459395, "grad_norm": 1.2405308959092338, "learning_rate": 1.6588630904100465e-05, "loss": 0.5565, "step": 8222 }, { "epoch": 0.2929618611610881, "grad_norm": 1.127478684261643, "learning_rate": 1.658776279650397e-05, "loss": 0.5342, "step": 8223 }, { "epoch": 0.29299748828758215, "grad_norm": 1.8814719598053256, "learning_rate": 1.6586894601186804e-05, "loss": 0.631, "step": 8224 }, { "epoch": 0.2930331154140763, "grad_norm": 1.150571616457621, "learning_rate": 1.6586026318160538e-05, "loss": 0.5177, "step": 8225 }, { "epoch": 0.2930687425405704, "grad_norm": 1.6991897688982003, "learning_rate": 1.6585157947436727e-05, "loss": 0.6103, "step": 8226 }, { "epoch": 0.2931043696670645, "grad_norm": 1.0805295512218185, "learning_rate": 1.6584289489026937e-05, "loss": 0.4964, "step": 8227 }, { "epoch": 0.2931399967935586, "grad_norm": 1.4594848430836989, "learning_rate": 1.6583420942942728e-05, "loss": 0.6119, "step": 8228 }, { "epoch": 0.29317562392005275, "grad_norm": 1.9458336116078214, "learning_rate": 1.6582552309195667e-05, "loss": 0.6244, "step": 8229 }, { "epoch": 0.29321125104654683, "grad_norm": 1.8811306627168802, "learning_rate": 1.6581683587797322e-05, "loss": 0.6424, "step": 8230 }, { "epoch": 0.29324687817304096, "grad_norm": 1.0423920559829336, "learning_rate": 1.6580814778759263e-05, "loss": 0.4412, "step": 8231 }, { "epoch": 0.2932825052995351, "grad_norm": 1.0998517357043704, "learning_rate": 1.657994588209305e-05, "loss": 0.4783, "step": 8232 }, { "epoch": 0.29331813242602917, "grad_norm": 0.9496543847777841, "learning_rate": 1.657907689781026e-05, "loss": 0.4747, "step": 8233 }, { "epoch": 0.2933537595525233, "grad_norm": 1.7091972146044254, "learning_rate": 1.6578207825922466e-05, "loss": 0.6417, "step": 8234 }, { "epoch": 0.2933893866790174, "grad_norm": 1.0720946671164775, "learning_rate": 1.6577338666441232e-05, "loss": 0.4725, "step": 8235 }, { "epoch": 0.2934250138055115, "grad_norm": 1.197833308258231, "learning_rate": 1.657646941937814e-05, "loss": 0.4972, "step": 8236 }, { "epoch": 0.29346064093200563, "grad_norm": 1.6758517469170797, "learning_rate": 1.657560008474476e-05, "loss": 0.6018, "step": 8237 }, { "epoch": 0.29349626805849977, "grad_norm": 1.035983533793348, "learning_rate": 1.657473066255267e-05, "loss": 0.4578, "step": 8238 }, { "epoch": 0.29353189518499384, "grad_norm": 1.6334752620509445, "learning_rate": 1.6573861152813445e-05, "loss": 0.6528, "step": 8239 }, { "epoch": 0.29356752231148797, "grad_norm": 1.5929027561962381, "learning_rate": 1.657299155553866e-05, "loss": 0.5278, "step": 8240 }, { "epoch": 0.2936031494379821, "grad_norm": 1.5773934277735702, "learning_rate": 1.6572121870739903e-05, "loss": 0.6126, "step": 8241 }, { "epoch": 0.2936387765644762, "grad_norm": 1.7324819572940648, "learning_rate": 1.657125209842875e-05, "loss": 0.6036, "step": 8242 }, { "epoch": 0.2936744036909703, "grad_norm": 1.0339143328504283, "learning_rate": 1.6570382238616778e-05, "loss": 0.4645, "step": 8243 }, { "epoch": 0.29371003081746444, "grad_norm": 1.1853854376960409, "learning_rate": 1.6569512291315577e-05, "loss": 0.5205, "step": 8244 }, { "epoch": 0.2937456579439585, "grad_norm": 1.5016431767751046, "learning_rate": 1.656864225653673e-05, "loss": 0.6472, "step": 8245 }, { "epoch": 0.29378128507045265, "grad_norm": 1.1410237514730306, "learning_rate": 1.6567772134291816e-05, "loss": 0.5471, "step": 8246 }, { "epoch": 0.2938169121969468, "grad_norm": 1.4695503905952727, "learning_rate": 1.6566901924592425e-05, "loss": 0.5499, "step": 8247 }, { "epoch": 0.29385253932344085, "grad_norm": 1.160364932183187, "learning_rate": 1.6566031627450147e-05, "loss": 0.5348, "step": 8248 }, { "epoch": 0.293888166449935, "grad_norm": 1.6452654238219493, "learning_rate": 1.656516124287657e-05, "loss": 0.4917, "step": 8249 }, { "epoch": 0.2939237935764291, "grad_norm": 1.9447342474272808, "learning_rate": 1.656429077088328e-05, "loss": 0.6057, "step": 8250 }, { "epoch": 0.2939594207029232, "grad_norm": 1.4178097460706285, "learning_rate": 1.6563420211481873e-05, "loss": 0.5613, "step": 8251 }, { "epoch": 0.2939950478294173, "grad_norm": 1.4972925408046769, "learning_rate": 1.6562549564683937e-05, "loss": 0.61, "step": 8252 }, { "epoch": 0.29403067495591145, "grad_norm": 1.2211284403796852, "learning_rate": 1.6561678830501064e-05, "loss": 0.5281, "step": 8253 }, { "epoch": 0.29406630208240553, "grad_norm": 1.1038120869962524, "learning_rate": 1.6560808008944852e-05, "loss": 0.4577, "step": 8254 }, { "epoch": 0.29410192920889966, "grad_norm": 1.2145840636686975, "learning_rate": 1.65599371000269e-05, "loss": 0.5148, "step": 8255 }, { "epoch": 0.2941375563353938, "grad_norm": 1.2889326271829316, "learning_rate": 1.6559066103758797e-05, "loss": 0.5005, "step": 8256 }, { "epoch": 0.29417318346188787, "grad_norm": 1.2009265993032896, "learning_rate": 1.655819502015214e-05, "loss": 0.4935, "step": 8257 }, { "epoch": 0.294208810588382, "grad_norm": 1.560680000554306, "learning_rate": 1.655732384921854e-05, "loss": 0.6527, "step": 8258 }, { "epoch": 0.29424443771487613, "grad_norm": 1.1289289107567, "learning_rate": 1.6556452590969583e-05, "loss": 0.5135, "step": 8259 }, { "epoch": 0.2942800648413702, "grad_norm": 1.0683319314726774, "learning_rate": 1.6555581245416882e-05, "loss": 0.4799, "step": 8260 }, { "epoch": 0.29431569196786433, "grad_norm": 1.0662196634804524, "learning_rate": 1.6554709812572032e-05, "loss": 0.452, "step": 8261 }, { "epoch": 0.29435131909435847, "grad_norm": 1.2189370410292328, "learning_rate": 1.655383829244664e-05, "loss": 0.5378, "step": 8262 }, { "epoch": 0.29438694622085254, "grad_norm": 1.1224137095264957, "learning_rate": 1.6552966685052312e-05, "loss": 0.5437, "step": 8263 }, { "epoch": 0.29442257334734667, "grad_norm": 1.4971832995032637, "learning_rate": 1.655209499040065e-05, "loss": 0.6225, "step": 8264 }, { "epoch": 0.2944582004738408, "grad_norm": 1.6971389133133155, "learning_rate": 1.6551223208503263e-05, "loss": 0.6032, "step": 8265 }, { "epoch": 0.2944938276003349, "grad_norm": 1.6043544057734649, "learning_rate": 1.6550351339371763e-05, "loss": 0.6092, "step": 8266 }, { "epoch": 0.294529454726829, "grad_norm": 1.2209569230166644, "learning_rate": 1.6549479383017755e-05, "loss": 0.5375, "step": 8267 }, { "epoch": 0.29456508185332314, "grad_norm": 2.046894310570276, "learning_rate": 1.6548607339452853e-05, "loss": 0.5467, "step": 8268 }, { "epoch": 0.2946007089798172, "grad_norm": 1.1429328572648945, "learning_rate": 1.6547735208688664e-05, "loss": 0.5008, "step": 8269 }, { "epoch": 0.29463633610631135, "grad_norm": 1.049797998701693, "learning_rate": 1.6546862990736805e-05, "loss": 0.5276, "step": 8270 }, { "epoch": 0.2946719632328055, "grad_norm": 1.8782105335235697, "learning_rate": 1.654599068560889e-05, "loss": 0.5596, "step": 8271 }, { "epoch": 0.29470759035929955, "grad_norm": 2.4667598326243865, "learning_rate": 1.6545118293316533e-05, "loss": 0.5603, "step": 8272 }, { "epoch": 0.2947432174857937, "grad_norm": 1.6352798702557088, "learning_rate": 1.6544245813871353e-05, "loss": 0.5248, "step": 8273 }, { "epoch": 0.2947788446122878, "grad_norm": 1.1821716994464393, "learning_rate": 1.6543373247284965e-05, "loss": 0.5414, "step": 8274 }, { "epoch": 0.2948144717387819, "grad_norm": 1.670097391120147, "learning_rate": 1.654250059356899e-05, "loss": 0.6055, "step": 8275 }, { "epoch": 0.294850098865276, "grad_norm": 1.5089856820590972, "learning_rate": 1.654162785273504e-05, "loss": 0.5795, "step": 8276 }, { "epoch": 0.29488572599177015, "grad_norm": 1.2873119272059683, "learning_rate": 1.6540755024794748e-05, "loss": 0.593, "step": 8277 }, { "epoch": 0.29492135311826423, "grad_norm": 1.1670017517518407, "learning_rate": 1.6539882109759734e-05, "loss": 0.4662, "step": 8278 }, { "epoch": 0.29495698024475836, "grad_norm": 0.9410920178891121, "learning_rate": 1.653900910764161e-05, "loss": 0.5002, "step": 8279 }, { "epoch": 0.2949926073712525, "grad_norm": 1.1450507437665116, "learning_rate": 1.653813601845202e-05, "loss": 0.5074, "step": 8280 }, { "epoch": 0.29502823449774657, "grad_norm": 1.014118108923681, "learning_rate": 1.653726284220257e-05, "loss": 0.4904, "step": 8281 }, { "epoch": 0.2950638616242407, "grad_norm": 1.543482527353327, "learning_rate": 1.6536389578904898e-05, "loss": 0.6528, "step": 8282 }, { "epoch": 0.29509948875073483, "grad_norm": 1.1298483057083246, "learning_rate": 1.6535516228570633e-05, "loss": 0.4655, "step": 8283 }, { "epoch": 0.2951351158772289, "grad_norm": 1.1603822347935897, "learning_rate": 1.6534642791211403e-05, "loss": 0.4956, "step": 8284 }, { "epoch": 0.29517074300372304, "grad_norm": 1.0912506515973408, "learning_rate": 1.653376926683883e-05, "loss": 0.4968, "step": 8285 }, { "epoch": 0.29520637013021717, "grad_norm": 1.0378242911975668, "learning_rate": 1.6532895655464556e-05, "loss": 0.4958, "step": 8286 }, { "epoch": 0.29524199725671124, "grad_norm": 1.100868677794826, "learning_rate": 1.653202195710021e-05, "loss": 0.552, "step": 8287 }, { "epoch": 0.2952776243832054, "grad_norm": 1.1104689687933482, "learning_rate": 1.6531148171757427e-05, "loss": 0.4613, "step": 8288 }, { "epoch": 0.2953132515096995, "grad_norm": 1.9346604370696885, "learning_rate": 1.6530274299447843e-05, "loss": 0.5323, "step": 8289 }, { "epoch": 0.2953488786361936, "grad_norm": 1.0869622201882128, "learning_rate": 1.6529400340183087e-05, "loss": 0.5073, "step": 8290 }, { "epoch": 0.2953845057626877, "grad_norm": 2.096356803124613, "learning_rate": 1.6528526293974806e-05, "loss": 0.5529, "step": 8291 }, { "epoch": 0.29542013288918184, "grad_norm": 1.4121372805911852, "learning_rate": 1.6527652160834633e-05, "loss": 0.4792, "step": 8292 }, { "epoch": 0.2954557600156759, "grad_norm": 1.0492556824156822, "learning_rate": 1.6526777940774205e-05, "loss": 0.4848, "step": 8293 }, { "epoch": 0.29549138714217005, "grad_norm": 1.3898163251026807, "learning_rate": 1.6525903633805175e-05, "loss": 0.5885, "step": 8294 }, { "epoch": 0.2955270142686642, "grad_norm": 1.3278789141776193, "learning_rate": 1.6525029239939166e-05, "loss": 0.6616, "step": 8295 }, { "epoch": 0.29556264139515825, "grad_norm": 1.0351505674031234, "learning_rate": 1.652415475918784e-05, "loss": 0.4723, "step": 8296 }, { "epoch": 0.2955982685216524, "grad_norm": 1.1108935908975066, "learning_rate": 1.652328019156283e-05, "loss": 0.4487, "step": 8297 }, { "epoch": 0.2956338956481465, "grad_norm": 1.5767212746586907, "learning_rate": 1.652240553707579e-05, "loss": 0.6004, "step": 8298 }, { "epoch": 0.2956695227746406, "grad_norm": 1.1393850941110075, "learning_rate": 1.6521530795738356e-05, "loss": 0.501, "step": 8299 }, { "epoch": 0.2957051499011347, "grad_norm": 1.6662055716212274, "learning_rate": 1.652065596756218e-05, "loss": 0.5368, "step": 8300 }, { "epoch": 0.29574077702762885, "grad_norm": 1.2196548783963574, "learning_rate": 1.6519781052558918e-05, "loss": 0.4943, "step": 8301 }, { "epoch": 0.29577640415412293, "grad_norm": 1.1065758072214795, "learning_rate": 1.651890605074021e-05, "loss": 0.4931, "step": 8302 }, { "epoch": 0.29581203128061706, "grad_norm": 1.1532349976169562, "learning_rate": 1.6518030962117713e-05, "loss": 0.4693, "step": 8303 }, { "epoch": 0.2958476584071112, "grad_norm": 1.5900998435318918, "learning_rate": 1.651715578670308e-05, "loss": 0.6204, "step": 8304 }, { "epoch": 0.29588328553360527, "grad_norm": 1.3851224513281128, "learning_rate": 1.6516280524507956e-05, "loss": 0.5835, "step": 8305 }, { "epoch": 0.2959189126600994, "grad_norm": 1.6449589654508674, "learning_rate": 1.6515405175544005e-05, "loss": 0.6393, "step": 8306 }, { "epoch": 0.29595453978659353, "grad_norm": 1.2231971164039543, "learning_rate": 1.651452973982288e-05, "loss": 0.5218, "step": 8307 }, { "epoch": 0.2959901669130876, "grad_norm": 1.8244040208618433, "learning_rate": 1.6513654217356242e-05, "loss": 0.5985, "step": 8308 }, { "epoch": 0.29602579403958174, "grad_norm": 1.6801920061362572, "learning_rate": 1.6512778608155743e-05, "loss": 0.5645, "step": 8309 }, { "epoch": 0.29606142116607587, "grad_norm": 1.5564155106972604, "learning_rate": 1.6511902912233043e-05, "loss": 0.6358, "step": 8310 }, { "epoch": 0.29609704829256994, "grad_norm": 1.0709562957291991, "learning_rate": 1.6511027129599807e-05, "loss": 0.5248, "step": 8311 }, { "epoch": 0.2961326754190641, "grad_norm": 1.20113244869456, "learning_rate": 1.6510151260267694e-05, "loss": 0.5108, "step": 8312 }, { "epoch": 0.2961683025455582, "grad_norm": 1.0537808516527378, "learning_rate": 1.6509275304248366e-05, "loss": 0.4905, "step": 8313 }, { "epoch": 0.2962039296720523, "grad_norm": 1.4151870149162136, "learning_rate": 1.6508399261553484e-05, "loss": 0.4685, "step": 8314 }, { "epoch": 0.2962395567985464, "grad_norm": 1.222524574282405, "learning_rate": 1.650752313219472e-05, "loss": 0.5003, "step": 8315 }, { "epoch": 0.29627518392504054, "grad_norm": 1.230096375457036, "learning_rate": 1.6506646916183732e-05, "loss": 0.5673, "step": 8316 }, { "epoch": 0.2963108110515346, "grad_norm": 1.1152699604804537, "learning_rate": 1.6505770613532196e-05, "loss": 0.5062, "step": 8317 }, { "epoch": 0.29634643817802875, "grad_norm": 1.704199241886803, "learning_rate": 1.650489422425178e-05, "loss": 0.5998, "step": 8318 }, { "epoch": 0.2963820653045229, "grad_norm": 1.5327405634362306, "learning_rate": 1.650401774835415e-05, "loss": 0.6281, "step": 8319 }, { "epoch": 0.29641769243101695, "grad_norm": 1.4701798606990175, "learning_rate": 1.650314118585098e-05, "loss": 0.5664, "step": 8320 }, { "epoch": 0.2964533195575111, "grad_norm": 1.168689266095638, "learning_rate": 1.650226453675393e-05, "loss": 0.4965, "step": 8321 }, { "epoch": 0.2964889466840052, "grad_norm": 1.0631991785937311, "learning_rate": 1.650138780107469e-05, "loss": 0.5048, "step": 8322 }, { "epoch": 0.2965245738104993, "grad_norm": 1.8320451668474091, "learning_rate": 1.6500510978824928e-05, "loss": 0.7053, "step": 8323 }, { "epoch": 0.2965602009369934, "grad_norm": 1.2138109226974654, "learning_rate": 1.6499634070016316e-05, "loss": 0.4958, "step": 8324 }, { "epoch": 0.29659582806348755, "grad_norm": 1.3861957476307447, "learning_rate": 1.649875707466053e-05, "loss": 0.6042, "step": 8325 }, { "epoch": 0.29663145518998163, "grad_norm": 1.3990086335933019, "learning_rate": 1.6497879992769258e-05, "loss": 0.5575, "step": 8326 }, { "epoch": 0.29666708231647576, "grad_norm": 1.0376629525788106, "learning_rate": 1.6497002824354172e-05, "loss": 0.5172, "step": 8327 }, { "epoch": 0.2967027094429699, "grad_norm": 1.194598170334176, "learning_rate": 1.6496125569426946e-05, "loss": 0.5184, "step": 8328 }, { "epoch": 0.29673833656946397, "grad_norm": 1.5263959548745774, "learning_rate": 1.6495248227999272e-05, "loss": 0.5602, "step": 8329 }, { "epoch": 0.2967739636959581, "grad_norm": 1.4595444517858966, "learning_rate": 1.6494370800082826e-05, "loss": 0.6374, "step": 8330 }, { "epoch": 0.29680959082245223, "grad_norm": 1.0920977834713212, "learning_rate": 1.6493493285689293e-05, "loss": 0.5015, "step": 8331 }, { "epoch": 0.2968452179489463, "grad_norm": 1.034690344929526, "learning_rate": 1.6492615684830362e-05, "loss": 0.4759, "step": 8332 }, { "epoch": 0.29688084507544044, "grad_norm": 1.9398845772529625, "learning_rate": 1.6491737997517713e-05, "loss": 0.5786, "step": 8333 }, { "epoch": 0.29691647220193457, "grad_norm": 1.13728806435231, "learning_rate": 1.6490860223763037e-05, "loss": 0.496, "step": 8334 }, { "epoch": 0.29695209932842864, "grad_norm": 1.2066354524714524, "learning_rate": 1.6489982363578017e-05, "loss": 0.5405, "step": 8335 }, { "epoch": 0.2969877264549228, "grad_norm": 1.1436640746295113, "learning_rate": 1.6489104416974346e-05, "loss": 0.5054, "step": 8336 }, { "epoch": 0.2970233535814169, "grad_norm": 1.6988931093943833, "learning_rate": 1.6488226383963715e-05, "loss": 0.6785, "step": 8337 }, { "epoch": 0.297058980707911, "grad_norm": 2.748859084444953, "learning_rate": 1.6487348264557817e-05, "loss": 0.6374, "step": 8338 }, { "epoch": 0.2970946078344051, "grad_norm": 1.6210342467243046, "learning_rate": 1.648647005876834e-05, "loss": 0.5629, "step": 8339 }, { "epoch": 0.29713023496089924, "grad_norm": 1.6334420022629734, "learning_rate": 1.6485591766606978e-05, "loss": 0.6224, "step": 8340 }, { "epoch": 0.2971658620873933, "grad_norm": 1.1576358542530532, "learning_rate": 1.648471338808543e-05, "loss": 0.4694, "step": 8341 }, { "epoch": 0.29720148921388745, "grad_norm": 1.942418914475392, "learning_rate": 1.6483834923215395e-05, "loss": 0.607, "step": 8342 }, { "epoch": 0.2972371163403816, "grad_norm": 1.6569458904733851, "learning_rate": 1.648295637200856e-05, "loss": 0.6104, "step": 8343 }, { "epoch": 0.29727274346687566, "grad_norm": 2.0057990853594507, "learning_rate": 1.6482077734476634e-05, "loss": 0.5814, "step": 8344 }, { "epoch": 0.2973083705933698, "grad_norm": 1.1004778437066793, "learning_rate": 1.6481199010631312e-05, "loss": 0.4981, "step": 8345 }, { "epoch": 0.2973439977198639, "grad_norm": 2.039885665196904, "learning_rate": 1.6480320200484293e-05, "loss": 0.5771, "step": 8346 }, { "epoch": 0.297379624846358, "grad_norm": 1.0241519959116645, "learning_rate": 1.647944130404728e-05, "loss": 0.4403, "step": 8347 }, { "epoch": 0.2974152519728521, "grad_norm": 1.158615227884711, "learning_rate": 1.6478562321331982e-05, "loss": 0.4666, "step": 8348 }, { "epoch": 0.29745087909934625, "grad_norm": 1.0429891449789799, "learning_rate": 1.6477683252350095e-05, "loss": 0.4622, "step": 8349 }, { "epoch": 0.29748650622584033, "grad_norm": 1.022019561697406, "learning_rate": 1.647680409711333e-05, "loss": 0.5349, "step": 8350 }, { "epoch": 0.29752213335233446, "grad_norm": 1.0821772270798953, "learning_rate": 1.6475924855633386e-05, "loss": 0.4779, "step": 8351 }, { "epoch": 0.2975577604788286, "grad_norm": 1.0402626393645313, "learning_rate": 1.647504552792198e-05, "loss": 0.5101, "step": 8352 }, { "epoch": 0.29759338760532267, "grad_norm": 1.0473146949398304, "learning_rate": 1.6474166113990817e-05, "loss": 0.499, "step": 8353 }, { "epoch": 0.2976290147318168, "grad_norm": 1.0597133837046466, "learning_rate": 1.6473286613851606e-05, "loss": 0.5119, "step": 8354 }, { "epoch": 0.29766464185831093, "grad_norm": 1.1501681954225715, "learning_rate": 1.647240702751606e-05, "loss": 0.5201, "step": 8355 }, { "epoch": 0.297700268984805, "grad_norm": 1.9587837489900521, "learning_rate": 1.647152735499589e-05, "loss": 0.5844, "step": 8356 }, { "epoch": 0.29773589611129914, "grad_norm": 1.7458964855038401, "learning_rate": 1.6470647596302812e-05, "loss": 0.6068, "step": 8357 }, { "epoch": 0.29777152323779327, "grad_norm": 1.6665880017268484, "learning_rate": 1.6469767751448538e-05, "loss": 0.6593, "step": 8358 }, { "epoch": 0.29780715036428734, "grad_norm": 1.588091678308389, "learning_rate": 1.6468887820444785e-05, "loss": 0.5417, "step": 8359 }, { "epoch": 0.2978427774907815, "grad_norm": 1.6523037257030002, "learning_rate": 1.6468007803303267e-05, "loss": 0.5603, "step": 8360 }, { "epoch": 0.2978784046172756, "grad_norm": 1.1167577073148602, "learning_rate": 1.6467127700035703e-05, "loss": 0.4846, "step": 8361 }, { "epoch": 0.2979140317437697, "grad_norm": 1.1375077351679597, "learning_rate": 1.6466247510653813e-05, "loss": 0.4688, "step": 8362 }, { "epoch": 0.2979496588702638, "grad_norm": 1.2361659520733619, "learning_rate": 1.646536723516932e-05, "loss": 0.4915, "step": 8363 }, { "epoch": 0.29798528599675794, "grad_norm": 1.7381489263058385, "learning_rate": 1.6464486873593942e-05, "loss": 0.5917, "step": 8364 }, { "epoch": 0.298020913123252, "grad_norm": 1.0938820565663254, "learning_rate": 1.6463606425939404e-05, "loss": 0.487, "step": 8365 }, { "epoch": 0.29805654024974615, "grad_norm": 1.3157192369796098, "learning_rate": 1.646272589221743e-05, "loss": 0.4939, "step": 8366 }, { "epoch": 0.2980921673762403, "grad_norm": 1.2386566546534148, "learning_rate": 1.6461845272439743e-05, "loss": 0.5053, "step": 8367 }, { "epoch": 0.29812779450273436, "grad_norm": 1.5628284856436965, "learning_rate": 1.646096456661807e-05, "loss": 0.4952, "step": 8368 }, { "epoch": 0.2981634216292285, "grad_norm": 1.0381395249199548, "learning_rate": 1.6460083774764135e-05, "loss": 0.5043, "step": 8369 }, { "epoch": 0.2981990487557226, "grad_norm": 1.428861956332711, "learning_rate": 1.645920289688967e-05, "loss": 0.4994, "step": 8370 }, { "epoch": 0.2982346758822167, "grad_norm": 1.6742471672604922, "learning_rate": 1.6458321933006407e-05, "loss": 0.5518, "step": 8371 }, { "epoch": 0.2982703030087108, "grad_norm": 1.179078271592345, "learning_rate": 1.6457440883126074e-05, "loss": 0.5002, "step": 8372 }, { "epoch": 0.29830593013520496, "grad_norm": 1.1156258893124196, "learning_rate": 1.6456559747260396e-05, "loss": 0.503, "step": 8373 }, { "epoch": 0.29834155726169903, "grad_norm": 1.1554645020284198, "learning_rate": 1.6455678525421116e-05, "loss": 0.4938, "step": 8374 }, { "epoch": 0.29837718438819316, "grad_norm": 1.510870890694163, "learning_rate": 1.645479721761996e-05, "loss": 0.6279, "step": 8375 }, { "epoch": 0.2984128115146873, "grad_norm": 1.6548657366013382, "learning_rate": 1.6453915823868673e-05, "loss": 0.548, "step": 8376 }, { "epoch": 0.29844843864118137, "grad_norm": 1.102975511198209, "learning_rate": 1.6453034344178983e-05, "loss": 0.5669, "step": 8377 }, { "epoch": 0.2984840657676755, "grad_norm": 1.6083160549572773, "learning_rate": 1.6452152778562633e-05, "loss": 0.5742, "step": 8378 }, { "epoch": 0.29851969289416963, "grad_norm": 1.8607369911713019, "learning_rate": 1.6451271127031357e-05, "loss": 0.626, "step": 8379 }, { "epoch": 0.2985553200206637, "grad_norm": 1.338832951382474, "learning_rate": 1.6450389389596898e-05, "loss": 0.5242, "step": 8380 }, { "epoch": 0.29859094714715784, "grad_norm": 1.3316101443929946, "learning_rate": 1.6449507566270993e-05, "loss": 0.6157, "step": 8381 }, { "epoch": 0.29862657427365197, "grad_norm": 1.2028654415694384, "learning_rate": 1.644862565706539e-05, "loss": 0.5229, "step": 8382 }, { "epoch": 0.2986622014001461, "grad_norm": 1.6969377577186406, "learning_rate": 1.6447743661991826e-05, "loss": 0.6136, "step": 8383 }, { "epoch": 0.2986978285266402, "grad_norm": 1.569310284770317, "learning_rate": 1.6446861581062047e-05, "loss": 0.6026, "step": 8384 }, { "epoch": 0.2987334556531343, "grad_norm": 1.6952557050339518, "learning_rate": 1.6445979414287806e-05, "loss": 0.6197, "step": 8385 }, { "epoch": 0.29876908277962844, "grad_norm": 1.0967615416261203, "learning_rate": 1.6445097161680837e-05, "loss": 0.5123, "step": 8386 }, { "epoch": 0.2988047099061225, "grad_norm": 1.0569196904868048, "learning_rate": 1.64442148232529e-05, "loss": 0.529, "step": 8387 }, { "epoch": 0.29884033703261664, "grad_norm": 1.0508853656422208, "learning_rate": 1.6443332399015734e-05, "loss": 0.4924, "step": 8388 }, { "epoch": 0.2988759641591108, "grad_norm": 1.1247760997457434, "learning_rate": 1.6442449888981096e-05, "loss": 0.5015, "step": 8389 }, { "epoch": 0.29891159128560485, "grad_norm": 1.3610359093465232, "learning_rate": 1.6441567293160737e-05, "loss": 0.5758, "step": 8390 }, { "epoch": 0.298947218412099, "grad_norm": 1.0304641373454242, "learning_rate": 1.6440684611566403e-05, "loss": 0.4793, "step": 8391 }, { "epoch": 0.2989828455385931, "grad_norm": 1.5551176028497073, "learning_rate": 1.643980184420985e-05, "loss": 0.5663, "step": 8392 }, { "epoch": 0.2990184726650872, "grad_norm": 1.4773359065096139, "learning_rate": 1.6438918991102842e-05, "loss": 0.6195, "step": 8393 }, { "epoch": 0.2990540997915813, "grad_norm": 1.1489287822450542, "learning_rate": 1.643803605225712e-05, "loss": 0.5077, "step": 8394 }, { "epoch": 0.29908972691807545, "grad_norm": 1.1087260512728208, "learning_rate": 1.6437153027684452e-05, "loss": 0.528, "step": 8395 }, { "epoch": 0.2991253540445695, "grad_norm": 1.8305597058277545, "learning_rate": 1.6436269917396593e-05, "loss": 0.5859, "step": 8396 }, { "epoch": 0.29916098117106366, "grad_norm": 1.6904314951547779, "learning_rate": 1.6435386721405297e-05, "loss": 0.5498, "step": 8397 }, { "epoch": 0.2991966082975578, "grad_norm": 1.099936197580993, "learning_rate": 1.643450343972233e-05, "loss": 0.4574, "step": 8398 }, { "epoch": 0.29923223542405186, "grad_norm": 1.0372311681416175, "learning_rate": 1.6433620072359452e-05, "loss": 0.4429, "step": 8399 }, { "epoch": 0.299267862550546, "grad_norm": 1.236121731556468, "learning_rate": 1.6432736619328425e-05, "loss": 0.4929, "step": 8400 }, { "epoch": 0.2993034896770401, "grad_norm": 1.8213311887862182, "learning_rate": 1.643185308064102e-05, "loss": 0.6829, "step": 8401 }, { "epoch": 0.2993391168035342, "grad_norm": 1.5507039851861006, "learning_rate": 1.643096945630899e-05, "loss": 0.5615, "step": 8402 }, { "epoch": 0.29937474393002833, "grad_norm": 1.2586494550757559, "learning_rate": 1.6430085746344107e-05, "loss": 0.5545, "step": 8403 }, { "epoch": 0.29941037105652246, "grad_norm": 2.0367297236851876, "learning_rate": 1.642920195075814e-05, "loss": 0.6223, "step": 8404 }, { "epoch": 0.29944599818301654, "grad_norm": 1.0833493707794257, "learning_rate": 1.6428318069562854e-05, "loss": 0.4565, "step": 8405 }, { "epoch": 0.29948162530951067, "grad_norm": 1.180855719231827, "learning_rate": 1.642743410277002e-05, "loss": 0.5349, "step": 8406 }, { "epoch": 0.2995172524360048, "grad_norm": 1.137848847208681, "learning_rate": 1.642655005039141e-05, "loss": 0.511, "step": 8407 }, { "epoch": 0.2995528795624989, "grad_norm": 1.5091537847649632, "learning_rate": 1.6425665912438794e-05, "loss": 0.6205, "step": 8408 }, { "epoch": 0.299588506688993, "grad_norm": 1.1435249979486055, "learning_rate": 1.642478168892394e-05, "loss": 0.4764, "step": 8409 }, { "epoch": 0.29962413381548714, "grad_norm": 2.758017834709528, "learning_rate": 1.6423897379858633e-05, "loss": 0.5999, "step": 8410 }, { "epoch": 0.2996597609419812, "grad_norm": 1.2147867430280097, "learning_rate": 1.6423012985254638e-05, "loss": 0.4969, "step": 8411 }, { "epoch": 0.29969538806847534, "grad_norm": 2.912327107526319, "learning_rate": 1.6422128505123737e-05, "loss": 0.5877, "step": 8412 }, { "epoch": 0.2997310151949695, "grad_norm": 2.2265601199475387, "learning_rate": 1.642124393947771e-05, "loss": 0.574, "step": 8413 }, { "epoch": 0.29976664232146355, "grad_norm": 1.1919084488005438, "learning_rate": 1.6420359288328325e-05, "loss": 0.4805, "step": 8414 }, { "epoch": 0.2998022694479577, "grad_norm": 1.1615279602954225, "learning_rate": 1.6419474551687375e-05, "loss": 0.4711, "step": 8415 }, { "epoch": 0.2998378965744518, "grad_norm": 1.7891294155543878, "learning_rate": 1.641858972956663e-05, "loss": 0.6661, "step": 8416 }, { "epoch": 0.2998735237009459, "grad_norm": 1.3860711755189452, "learning_rate": 1.641770482197788e-05, "loss": 0.481, "step": 8417 }, { "epoch": 0.29990915082744, "grad_norm": 1.14022352631132, "learning_rate": 1.64168198289329e-05, "loss": 0.5011, "step": 8418 }, { "epoch": 0.29994477795393415, "grad_norm": 1.253298156497022, "learning_rate": 1.6415934750443483e-05, "loss": 0.5076, "step": 8419 }, { "epoch": 0.2999804050804282, "grad_norm": 1.067306382720132, "learning_rate": 1.6415049586521407e-05, "loss": 0.437, "step": 8420 }, { "epoch": 0.30001603220692236, "grad_norm": 1.9287458478030803, "learning_rate": 1.6414164337178467e-05, "loss": 0.6447, "step": 8421 }, { "epoch": 0.3000516593334165, "grad_norm": 1.1597152457404185, "learning_rate": 1.641327900242644e-05, "loss": 0.5409, "step": 8422 }, { "epoch": 0.30008728645991056, "grad_norm": 1.686883361859506, "learning_rate": 1.6412393582277122e-05, "loss": 0.5865, "step": 8423 }, { "epoch": 0.3001229135864047, "grad_norm": 1.141862918720281, "learning_rate": 1.6411508076742305e-05, "loss": 0.478, "step": 8424 }, { "epoch": 0.3001585407128988, "grad_norm": 1.594947510043607, "learning_rate": 1.6410622485833774e-05, "loss": 0.5817, "step": 8425 }, { "epoch": 0.3001941678393929, "grad_norm": 1.7779800017116585, "learning_rate": 1.6409736809563325e-05, "loss": 0.6664, "step": 8426 }, { "epoch": 0.30022979496588703, "grad_norm": 2.016614424196554, "learning_rate": 1.640885104794275e-05, "loss": 0.5642, "step": 8427 }, { "epoch": 0.30026542209238116, "grad_norm": 1.0925012654936208, "learning_rate": 1.6407965200983843e-05, "loss": 0.4934, "step": 8428 }, { "epoch": 0.30030104921887524, "grad_norm": 1.1492118457298803, "learning_rate": 1.64070792686984e-05, "loss": 0.5034, "step": 8429 }, { "epoch": 0.30033667634536937, "grad_norm": 1.6546354641242933, "learning_rate": 1.6406193251098218e-05, "loss": 0.6831, "step": 8430 }, { "epoch": 0.3003723034718635, "grad_norm": 2.250507876799874, "learning_rate": 1.6405307148195095e-05, "loss": 0.5699, "step": 8431 }, { "epoch": 0.3004079305983576, "grad_norm": 1.1330470856273311, "learning_rate": 1.640442096000083e-05, "loss": 0.5284, "step": 8432 }, { "epoch": 0.3004435577248517, "grad_norm": 1.1283429376609153, "learning_rate": 1.6403534686527223e-05, "loss": 0.5794, "step": 8433 }, { "epoch": 0.30047918485134584, "grad_norm": 1.3680307375589829, "learning_rate": 1.6402648327786076e-05, "loss": 0.461, "step": 8434 }, { "epoch": 0.3005148119778399, "grad_norm": 2.080986489863369, "learning_rate": 1.6401761883789196e-05, "loss": 0.5628, "step": 8435 }, { "epoch": 0.30055043910433404, "grad_norm": 1.5264752466728062, "learning_rate": 1.6400875354548377e-05, "loss": 0.5654, "step": 8436 }, { "epoch": 0.3005860662308282, "grad_norm": 1.5559786336526216, "learning_rate": 1.6399988740075433e-05, "loss": 0.6217, "step": 8437 }, { "epoch": 0.30062169335732225, "grad_norm": 1.155407734380888, "learning_rate": 1.6399102040382163e-05, "loss": 0.4909, "step": 8438 }, { "epoch": 0.3006573204838164, "grad_norm": 1.7086860617344186, "learning_rate": 1.6398215255480376e-05, "loss": 0.5811, "step": 8439 }, { "epoch": 0.3006929476103105, "grad_norm": 1.2701110958628914, "learning_rate": 1.639732838538188e-05, "loss": 0.5251, "step": 8440 }, { "epoch": 0.3007285747368046, "grad_norm": 1.5518363702460565, "learning_rate": 1.6396441430098485e-05, "loss": 0.6381, "step": 8441 }, { "epoch": 0.3007642018632987, "grad_norm": 1.3419907882795972, "learning_rate": 1.6395554389642007e-05, "loss": 0.5575, "step": 8442 }, { "epoch": 0.30079982898979285, "grad_norm": 1.1094833014149281, "learning_rate": 1.6394667264024246e-05, "loss": 0.5012, "step": 8443 }, { "epoch": 0.3008354561162869, "grad_norm": 1.1127224846221333, "learning_rate": 1.639378005325702e-05, "loss": 0.4455, "step": 8444 }, { "epoch": 0.30087108324278106, "grad_norm": 1.1211802338659498, "learning_rate": 1.639289275735215e-05, "loss": 0.4866, "step": 8445 }, { "epoch": 0.3009067103692752, "grad_norm": 2.2670570690603635, "learning_rate": 1.639200537632144e-05, "loss": 0.5933, "step": 8446 }, { "epoch": 0.30094233749576926, "grad_norm": 1.0952412156393025, "learning_rate": 1.6391117910176712e-05, "loss": 0.4751, "step": 8447 }, { "epoch": 0.3009779646222634, "grad_norm": 1.537209723933058, "learning_rate": 1.639023035892978e-05, "loss": 0.5987, "step": 8448 }, { "epoch": 0.3010135917487575, "grad_norm": 1.1585601371404877, "learning_rate": 1.6389342722592468e-05, "loss": 0.4798, "step": 8449 }, { "epoch": 0.3010492188752516, "grad_norm": 2.671833916826744, "learning_rate": 1.6388455001176587e-05, "loss": 0.649, "step": 8450 }, { "epoch": 0.30108484600174573, "grad_norm": 1.5395998373453552, "learning_rate": 1.6387567194693967e-05, "loss": 0.5893, "step": 8451 }, { "epoch": 0.30112047312823986, "grad_norm": 1.7002121702488502, "learning_rate": 1.6386679303156422e-05, "loss": 0.4662, "step": 8452 }, { "epoch": 0.30115610025473394, "grad_norm": 1.1852938607189845, "learning_rate": 1.638579132657578e-05, "loss": 0.5125, "step": 8453 }, { "epoch": 0.30119172738122807, "grad_norm": 1.1571422365975361, "learning_rate": 1.6384903264963864e-05, "loss": 0.4957, "step": 8454 }, { "epoch": 0.3012273545077222, "grad_norm": 1.6030486608485788, "learning_rate": 1.6384015118332494e-05, "loss": 0.5753, "step": 8455 }, { "epoch": 0.3012629816342163, "grad_norm": 1.9427192357363368, "learning_rate": 1.6383126886693507e-05, "loss": 0.7084, "step": 8456 }, { "epoch": 0.3012986087607104, "grad_norm": 1.6822231290468228, "learning_rate": 1.6382238570058718e-05, "loss": 0.5863, "step": 8457 }, { "epoch": 0.30133423588720454, "grad_norm": 1.0657555266981158, "learning_rate": 1.638135016843996e-05, "loss": 0.4334, "step": 8458 }, { "epoch": 0.3013698630136986, "grad_norm": 1.803856937233099, "learning_rate": 1.638046168184907e-05, "loss": 0.6774, "step": 8459 }, { "epoch": 0.30140549014019274, "grad_norm": 1.4862039725405876, "learning_rate": 1.637957311029787e-05, "loss": 0.5035, "step": 8460 }, { "epoch": 0.3014411172666869, "grad_norm": 1.1416988255365568, "learning_rate": 1.63786844537982e-05, "loss": 0.4813, "step": 8461 }, { "epoch": 0.30147674439318095, "grad_norm": 1.1430366959620508, "learning_rate": 1.6377795712361882e-05, "loss": 0.4791, "step": 8462 }, { "epoch": 0.3015123715196751, "grad_norm": 0.975132938370361, "learning_rate": 1.6376906886000758e-05, "loss": 0.4398, "step": 8463 }, { "epoch": 0.3015479986461692, "grad_norm": 1.1504063878509716, "learning_rate": 1.637601797472666e-05, "loss": 0.4627, "step": 8464 }, { "epoch": 0.3015836257726633, "grad_norm": 1.5966193768894552, "learning_rate": 1.637512897855143e-05, "loss": 0.596, "step": 8465 }, { "epoch": 0.3016192528991574, "grad_norm": 1.1420055549728625, "learning_rate": 1.63742398974869e-05, "loss": 0.5028, "step": 8466 }, { "epoch": 0.30165488002565155, "grad_norm": 1.1088052433160271, "learning_rate": 1.637335073154491e-05, "loss": 0.4616, "step": 8467 }, { "epoch": 0.3016905071521456, "grad_norm": 1.1256951931440096, "learning_rate": 1.63724614807373e-05, "loss": 0.516, "step": 8468 }, { "epoch": 0.30172613427863976, "grad_norm": 1.0702162031700493, "learning_rate": 1.637157214507591e-05, "loss": 0.5177, "step": 8469 }, { "epoch": 0.3017617614051339, "grad_norm": 1.1229063055256647, "learning_rate": 1.6370682724572583e-05, "loss": 0.4572, "step": 8470 }, { "epoch": 0.30179738853162796, "grad_norm": 1.7263963949863792, "learning_rate": 1.636979321923917e-05, "loss": 0.6785, "step": 8471 }, { "epoch": 0.3018330156581221, "grad_norm": 1.1022364196475474, "learning_rate": 1.63689036290875e-05, "loss": 0.5493, "step": 8472 }, { "epoch": 0.3018686427846162, "grad_norm": 1.563104860243994, "learning_rate": 1.636801395412943e-05, "loss": 0.5362, "step": 8473 }, { "epoch": 0.3019042699111103, "grad_norm": 1.5334062884634843, "learning_rate": 1.6367124194376802e-05, "loss": 0.5964, "step": 8474 }, { "epoch": 0.30193989703760443, "grad_norm": 1.1465682844164, "learning_rate": 1.6366234349841465e-05, "loss": 0.5483, "step": 8475 }, { "epoch": 0.30197552416409856, "grad_norm": 1.2970110158733783, "learning_rate": 1.636534442053527e-05, "loss": 0.5871, "step": 8476 }, { "epoch": 0.30201115129059264, "grad_norm": 1.5372309679395684, "learning_rate": 1.6364454406470064e-05, "loss": 0.5751, "step": 8477 }, { "epoch": 0.30204677841708677, "grad_norm": 1.5475171456996024, "learning_rate": 1.6363564307657696e-05, "loss": 0.5952, "step": 8478 }, { "epoch": 0.3020824055435809, "grad_norm": 1.1092756951575173, "learning_rate": 1.6362674124110026e-05, "loss": 0.4873, "step": 8479 }, { "epoch": 0.302118032670075, "grad_norm": 1.7540078509624335, "learning_rate": 1.6361783855838898e-05, "loss": 0.6002, "step": 8480 }, { "epoch": 0.3021536597965691, "grad_norm": 1.8213674537319047, "learning_rate": 1.6360893502856175e-05, "loss": 0.6413, "step": 8481 }, { "epoch": 0.30218928692306324, "grad_norm": 1.7359605860351435, "learning_rate": 1.6360003065173707e-05, "loss": 0.5425, "step": 8482 }, { "epoch": 0.3022249140495573, "grad_norm": 1.6117695996982082, "learning_rate": 1.6359112542803355e-05, "loss": 0.5952, "step": 8483 }, { "epoch": 0.30226054117605144, "grad_norm": 1.471595575904782, "learning_rate": 1.6358221935756975e-05, "loss": 0.605, "step": 8484 }, { "epoch": 0.3022961683025456, "grad_norm": 1.2243794883601362, "learning_rate": 1.6357331244046423e-05, "loss": 0.5077, "step": 8485 }, { "epoch": 0.30233179542903965, "grad_norm": 1.503067112940474, "learning_rate": 1.6356440467683564e-05, "loss": 0.6279, "step": 8486 }, { "epoch": 0.3023674225555338, "grad_norm": 1.4580500039124473, "learning_rate": 1.6355549606680257e-05, "loss": 0.594, "step": 8487 }, { "epoch": 0.3024030496820279, "grad_norm": 1.1160750251076128, "learning_rate": 1.6354658661048364e-05, "loss": 0.5132, "step": 8488 }, { "epoch": 0.302438676808522, "grad_norm": 1.120666686463186, "learning_rate": 1.635376763079975e-05, "loss": 0.4519, "step": 8489 }, { "epoch": 0.3024743039350161, "grad_norm": 1.1164944257908063, "learning_rate": 1.6352876515946276e-05, "loss": 0.4517, "step": 8490 }, { "epoch": 0.30250993106151025, "grad_norm": 1.4556778465211333, "learning_rate": 1.6351985316499815e-05, "loss": 0.6158, "step": 8491 }, { "epoch": 0.3025455581880043, "grad_norm": 1.9163478646560284, "learning_rate": 1.6351094032472228e-05, "loss": 0.5487, "step": 8492 }, { "epoch": 0.30258118531449846, "grad_norm": 1.0063932231114638, "learning_rate": 1.6350202663875385e-05, "loss": 0.5259, "step": 8493 }, { "epoch": 0.3026168124409926, "grad_norm": 1.2000615893908448, "learning_rate": 1.6349311210721156e-05, "loss": 0.4915, "step": 8494 }, { "epoch": 0.30265243956748666, "grad_norm": 1.1895515524402054, "learning_rate": 1.634841967302141e-05, "loss": 0.4918, "step": 8495 }, { "epoch": 0.3026880666939808, "grad_norm": 1.8423266882437683, "learning_rate": 1.6347528050788016e-05, "loss": 0.6173, "step": 8496 }, { "epoch": 0.3027236938204749, "grad_norm": 1.178118708057123, "learning_rate": 1.6346636344032852e-05, "loss": 0.5209, "step": 8497 }, { "epoch": 0.302759320946969, "grad_norm": 1.662519998440057, "learning_rate": 1.6345744552767788e-05, "loss": 0.6546, "step": 8498 }, { "epoch": 0.30279494807346313, "grad_norm": 1.156037148579245, "learning_rate": 1.6344852677004702e-05, "loss": 0.5279, "step": 8499 }, { "epoch": 0.30283057519995726, "grad_norm": 1.6443062516186326, "learning_rate": 1.6343960716755467e-05, "loss": 0.6251, "step": 8500 }, { "epoch": 0.30286620232645134, "grad_norm": 1.136858130094602, "learning_rate": 1.6343068672031958e-05, "loss": 0.4698, "step": 8501 }, { "epoch": 0.30290182945294547, "grad_norm": 1.2670380535138526, "learning_rate": 1.6342176542846057e-05, "loss": 0.6175, "step": 8502 }, { "epoch": 0.3029374565794396, "grad_norm": 1.2725083274176465, "learning_rate": 1.6341284329209647e-05, "loss": 0.5931, "step": 8503 }, { "epoch": 0.3029730837059337, "grad_norm": 1.0752118212094557, "learning_rate": 1.63403920311346e-05, "loss": 0.4672, "step": 8504 }, { "epoch": 0.3030087108324278, "grad_norm": 1.9009900877804782, "learning_rate": 1.63394996486328e-05, "loss": 0.4788, "step": 8505 }, { "epoch": 0.30304433795892194, "grad_norm": 1.12192216354068, "learning_rate": 1.6338607181716132e-05, "loss": 0.5376, "step": 8506 }, { "epoch": 0.303079965085416, "grad_norm": 1.0875308214998125, "learning_rate": 1.633771463039648e-05, "loss": 0.4838, "step": 8507 }, { "epoch": 0.30311559221191015, "grad_norm": 1.6244763487055478, "learning_rate": 1.6336821994685728e-05, "loss": 0.5761, "step": 8508 }, { "epoch": 0.3031512193384043, "grad_norm": 0.988284328148686, "learning_rate": 1.6335929274595762e-05, "loss": 0.5291, "step": 8509 }, { "epoch": 0.30318684646489835, "grad_norm": 1.1393240660488526, "learning_rate": 1.633503647013847e-05, "loss": 0.4859, "step": 8510 }, { "epoch": 0.3032224735913925, "grad_norm": 1.505548633893093, "learning_rate": 1.6334143581325736e-05, "loss": 0.5729, "step": 8511 }, { "epoch": 0.3032581007178866, "grad_norm": 1.4571293734793798, "learning_rate": 1.6333250608169454e-05, "loss": 0.5743, "step": 8512 }, { "epoch": 0.3032937278443807, "grad_norm": 1.0148527412431334, "learning_rate": 1.6332357550681515e-05, "loss": 0.4898, "step": 8513 }, { "epoch": 0.3033293549708748, "grad_norm": 1.1184254138537977, "learning_rate": 1.633146440887381e-05, "loss": 0.531, "step": 8514 }, { "epoch": 0.30336498209736895, "grad_norm": 1.1919716559525548, "learning_rate": 1.633057118275823e-05, "loss": 0.5277, "step": 8515 }, { "epoch": 0.303400609223863, "grad_norm": 1.4119485329917336, "learning_rate": 1.6329677872346674e-05, "loss": 0.6268, "step": 8516 }, { "epoch": 0.30343623635035716, "grad_norm": 1.1600758411849899, "learning_rate": 1.632878447765103e-05, "loss": 0.4748, "step": 8517 }, { "epoch": 0.3034718634768513, "grad_norm": 1.0462081618285395, "learning_rate": 1.632789099868319e-05, "loss": 0.4742, "step": 8518 }, { "epoch": 0.30350749060334536, "grad_norm": 1.518292887122122, "learning_rate": 1.6326997435455067e-05, "loss": 0.6058, "step": 8519 }, { "epoch": 0.3035431177298395, "grad_norm": 1.200946264705845, "learning_rate": 1.6326103787978546e-05, "loss": 0.4802, "step": 8520 }, { "epoch": 0.3035787448563336, "grad_norm": 1.4629686381211926, "learning_rate": 1.6325210056265537e-05, "loss": 0.5838, "step": 8521 }, { "epoch": 0.3036143719828277, "grad_norm": 1.0828722938618027, "learning_rate": 1.6324316240327932e-05, "loss": 0.4393, "step": 8522 }, { "epoch": 0.30364999910932183, "grad_norm": 1.2968311724358679, "learning_rate": 1.6323422340177634e-05, "loss": 0.5985, "step": 8523 }, { "epoch": 0.30368562623581596, "grad_norm": 1.2950265598381787, "learning_rate": 1.6322528355826545e-05, "loss": 0.5806, "step": 8524 }, { "epoch": 0.30372125336231004, "grad_norm": 1.1929366901922345, "learning_rate": 1.632163428728658e-05, "loss": 0.5059, "step": 8525 }, { "epoch": 0.30375688048880417, "grad_norm": 1.245748070372707, "learning_rate": 1.632074013456963e-05, "loss": 0.54, "step": 8526 }, { "epoch": 0.3037925076152983, "grad_norm": 1.0668460219160452, "learning_rate": 1.631984589768761e-05, "loss": 0.4632, "step": 8527 }, { "epoch": 0.3038281347417924, "grad_norm": 1.0785399976119852, "learning_rate": 1.6318951576652424e-05, "loss": 0.485, "step": 8528 }, { "epoch": 0.3038637618682865, "grad_norm": 1.5615055194564982, "learning_rate": 1.6318057171475977e-05, "loss": 0.6679, "step": 8529 }, { "epoch": 0.30389938899478064, "grad_norm": 1.675228065174595, "learning_rate": 1.6317162682170188e-05, "loss": 0.6327, "step": 8530 }, { "epoch": 0.3039350161212747, "grad_norm": 1.730464400150634, "learning_rate": 1.6316268108746963e-05, "loss": 0.517, "step": 8531 }, { "epoch": 0.30397064324776885, "grad_norm": 1.5941019596753996, "learning_rate": 1.631537345121821e-05, "loss": 0.5286, "step": 8532 }, { "epoch": 0.304006270374263, "grad_norm": 1.8690656717635543, "learning_rate": 1.6314478709595845e-05, "loss": 0.6231, "step": 8533 }, { "epoch": 0.30404189750075705, "grad_norm": 1.5819456774080947, "learning_rate": 1.6313583883891785e-05, "loss": 0.6204, "step": 8534 }, { "epoch": 0.3040775246272512, "grad_norm": 1.1061509827350455, "learning_rate": 1.6312688974117943e-05, "loss": 0.552, "step": 8535 }, { "epoch": 0.3041131517537453, "grad_norm": 1.1731902699796612, "learning_rate": 1.6311793980286232e-05, "loss": 0.4786, "step": 8536 }, { "epoch": 0.3041487788802394, "grad_norm": 0.9790719951564603, "learning_rate": 1.6310898902408574e-05, "loss": 0.4924, "step": 8537 }, { "epoch": 0.3041844060067335, "grad_norm": 1.1973377999105408, "learning_rate": 1.6310003740496887e-05, "loss": 0.4907, "step": 8538 }, { "epoch": 0.30422003313322765, "grad_norm": 1.1356547981391623, "learning_rate": 1.6309108494563088e-05, "loss": 0.5379, "step": 8539 }, { "epoch": 0.3042556602597217, "grad_norm": 1.0986918994128962, "learning_rate": 1.63082131646191e-05, "loss": 0.5129, "step": 8540 }, { "epoch": 0.30429128738621586, "grad_norm": 1.4809315282894038, "learning_rate": 1.6307317750676846e-05, "loss": 0.6193, "step": 8541 }, { "epoch": 0.30432691451271, "grad_norm": 0.9881921402623222, "learning_rate": 1.6306422252748245e-05, "loss": 0.4661, "step": 8542 }, { "epoch": 0.30436254163920407, "grad_norm": 2.34064239316424, "learning_rate": 1.6305526670845225e-05, "loss": 0.6333, "step": 8543 }, { "epoch": 0.3043981687656982, "grad_norm": 1.4109409507453994, "learning_rate": 1.630463100497971e-05, "loss": 0.5447, "step": 8544 }, { "epoch": 0.3044337958921923, "grad_norm": 1.1609499102254461, "learning_rate": 1.630373525516363e-05, "loss": 0.5277, "step": 8545 }, { "epoch": 0.3044694230186864, "grad_norm": 1.816805273194472, "learning_rate": 1.6302839421408906e-05, "loss": 0.5451, "step": 8546 }, { "epoch": 0.30450505014518053, "grad_norm": 1.1940884443841664, "learning_rate": 1.630194350372747e-05, "loss": 0.4972, "step": 8547 }, { "epoch": 0.30454067727167466, "grad_norm": 2.2400309488115906, "learning_rate": 1.630104750213125e-05, "loss": 0.6163, "step": 8548 }, { "epoch": 0.30457630439816874, "grad_norm": 1.1585621804463841, "learning_rate": 1.630015141663218e-05, "loss": 0.472, "step": 8549 }, { "epoch": 0.30461193152466287, "grad_norm": 1.9708414288017144, "learning_rate": 1.629925524724219e-05, "loss": 0.6126, "step": 8550 }, { "epoch": 0.304647558651157, "grad_norm": 1.8324283795576617, "learning_rate": 1.6298358993973212e-05, "loss": 0.5986, "step": 8551 }, { "epoch": 0.3046831857776511, "grad_norm": 1.540349632674321, "learning_rate": 1.629746265683718e-05, "loss": 0.5611, "step": 8552 }, { "epoch": 0.3047188129041452, "grad_norm": 2.213926087961077, "learning_rate": 1.6296566235846037e-05, "loss": 0.5283, "step": 8553 }, { "epoch": 0.30475444003063934, "grad_norm": 2.1890764943024474, "learning_rate": 1.629566973101171e-05, "loss": 0.6231, "step": 8554 }, { "epoch": 0.3047900671571334, "grad_norm": 1.4610514025234596, "learning_rate": 1.629477314234614e-05, "loss": 0.5114, "step": 8555 }, { "epoch": 0.30482569428362755, "grad_norm": 1.7601100296097987, "learning_rate": 1.6293876469861266e-05, "loss": 0.5994, "step": 8556 }, { "epoch": 0.3048613214101217, "grad_norm": 1.1502313635772086, "learning_rate": 1.629297971356903e-05, "loss": 0.4403, "step": 8557 }, { "epoch": 0.30489694853661575, "grad_norm": 1.6099847786198607, "learning_rate": 1.6292082873481368e-05, "loss": 0.5787, "step": 8558 }, { "epoch": 0.3049325756631099, "grad_norm": 1.6831247616108176, "learning_rate": 1.6291185949610223e-05, "loss": 0.6313, "step": 8559 }, { "epoch": 0.304968202789604, "grad_norm": 1.5653697170925156, "learning_rate": 1.6290288941967546e-05, "loss": 0.587, "step": 8560 }, { "epoch": 0.3050038299160981, "grad_norm": 1.7848699015764455, "learning_rate": 1.628939185056527e-05, "loss": 0.6063, "step": 8561 }, { "epoch": 0.3050394570425922, "grad_norm": 1.0863303873601788, "learning_rate": 1.6288494675415346e-05, "loss": 0.5125, "step": 8562 }, { "epoch": 0.30507508416908635, "grad_norm": 1.5606952526972027, "learning_rate": 1.628759741652972e-05, "loss": 0.527, "step": 8563 }, { "epoch": 0.30511071129558043, "grad_norm": 1.1455471400365151, "learning_rate": 1.6286700073920343e-05, "loss": 0.5238, "step": 8564 }, { "epoch": 0.30514633842207456, "grad_norm": 1.1400114220096744, "learning_rate": 1.6285802647599156e-05, "loss": 0.4708, "step": 8565 }, { "epoch": 0.3051819655485687, "grad_norm": 1.4682990742253677, "learning_rate": 1.6284905137578114e-05, "loss": 0.5367, "step": 8566 }, { "epoch": 0.30521759267506277, "grad_norm": 1.117150978775427, "learning_rate": 1.628400754386917e-05, "loss": 0.5, "step": 8567 }, { "epoch": 0.3052532198015569, "grad_norm": 1.910204774924543, "learning_rate": 1.628310986648427e-05, "loss": 0.6233, "step": 8568 }, { "epoch": 0.305288846928051, "grad_norm": 1.8694035167523297, "learning_rate": 1.6282212105435372e-05, "loss": 0.5551, "step": 8569 }, { "epoch": 0.3053244740545451, "grad_norm": 1.1941778344900993, "learning_rate": 1.628131426073443e-05, "loss": 0.5043, "step": 8570 }, { "epoch": 0.30536010118103923, "grad_norm": 1.6542466710252763, "learning_rate": 1.6280416332393396e-05, "loss": 0.5802, "step": 8571 }, { "epoch": 0.30539572830753337, "grad_norm": 1.337089976541987, "learning_rate": 1.627951832042423e-05, "loss": 0.5707, "step": 8572 }, { "epoch": 0.3054313554340275, "grad_norm": 1.253988230306802, "learning_rate": 1.6278620224838887e-05, "loss": 0.4767, "step": 8573 }, { "epoch": 0.30546698256052157, "grad_norm": 1.5577334785222903, "learning_rate": 1.6277722045649328e-05, "loss": 0.6601, "step": 8574 }, { "epoch": 0.3055026096870157, "grad_norm": 1.451469552718479, "learning_rate": 1.6276823782867514e-05, "loss": 0.5073, "step": 8575 }, { "epoch": 0.30553823681350983, "grad_norm": 2.5709623868733034, "learning_rate": 1.62759254365054e-05, "loss": 0.4758, "step": 8576 }, { "epoch": 0.3055738639400039, "grad_norm": 1.1901101072976314, "learning_rate": 1.627502700657496e-05, "loss": 0.5274, "step": 8577 }, { "epoch": 0.30560949106649804, "grad_norm": 1.2071867784048818, "learning_rate": 1.6274128493088142e-05, "loss": 0.466, "step": 8578 }, { "epoch": 0.30564511819299217, "grad_norm": 1.6419679507444114, "learning_rate": 1.6273229896056922e-05, "loss": 0.4645, "step": 8579 }, { "epoch": 0.30568074531948625, "grad_norm": 1.621257407679621, "learning_rate": 1.6272331215493257e-05, "loss": 0.6308, "step": 8580 }, { "epoch": 0.3057163724459804, "grad_norm": 1.1340768826378966, "learning_rate": 1.627143245140912e-05, "loss": 0.4748, "step": 8581 }, { "epoch": 0.3057519995724745, "grad_norm": 1.2464122120634316, "learning_rate": 1.6270533603816476e-05, "loss": 0.5254, "step": 8582 }, { "epoch": 0.3057876266989686, "grad_norm": 1.1185006317618644, "learning_rate": 1.6269634672727296e-05, "loss": 0.4843, "step": 8583 }, { "epoch": 0.3058232538254627, "grad_norm": 1.4942069770697766, "learning_rate": 1.6268735658153546e-05, "loss": 0.5407, "step": 8584 }, { "epoch": 0.30585888095195685, "grad_norm": 1.0947088764362631, "learning_rate": 1.62678365601072e-05, "loss": 0.4872, "step": 8585 }, { "epoch": 0.3058945080784509, "grad_norm": 1.7193482561169713, "learning_rate": 1.626693737860023e-05, "loss": 0.6231, "step": 8586 }, { "epoch": 0.30593013520494505, "grad_norm": 1.0591240118389245, "learning_rate": 1.6266038113644605e-05, "loss": 0.5479, "step": 8587 }, { "epoch": 0.3059657623314392, "grad_norm": 1.1410156019889868, "learning_rate": 1.6265138765252308e-05, "loss": 0.4971, "step": 8588 }, { "epoch": 0.30600138945793326, "grad_norm": 1.3832000651362866, "learning_rate": 1.6264239333435305e-05, "loss": 0.5055, "step": 8589 }, { "epoch": 0.3060370165844274, "grad_norm": 1.434576848819748, "learning_rate": 1.626333981820558e-05, "loss": 0.6008, "step": 8590 }, { "epoch": 0.3060726437109215, "grad_norm": 1.1392505026345816, "learning_rate": 1.6262440219575106e-05, "loss": 0.4764, "step": 8591 }, { "epoch": 0.3061082708374156, "grad_norm": 1.074397994213154, "learning_rate": 1.6261540537555863e-05, "loss": 0.5015, "step": 8592 }, { "epoch": 0.30614389796390973, "grad_norm": 1.1935272529021914, "learning_rate": 1.626064077215983e-05, "loss": 0.5369, "step": 8593 }, { "epoch": 0.30617952509040386, "grad_norm": 1.6620657655982187, "learning_rate": 1.6259740923398994e-05, "loss": 0.5312, "step": 8594 }, { "epoch": 0.30621515221689793, "grad_norm": 1.6609601755199372, "learning_rate": 1.625884099128533e-05, "loss": 0.5835, "step": 8595 }, { "epoch": 0.30625077934339207, "grad_norm": 1.6514707653554912, "learning_rate": 1.625794097583082e-05, "loss": 0.5964, "step": 8596 }, { "epoch": 0.3062864064698862, "grad_norm": 1.2660892551141285, "learning_rate": 1.6257040877047454e-05, "loss": 0.4855, "step": 8597 }, { "epoch": 0.30632203359638027, "grad_norm": 1.2034631245462717, "learning_rate": 1.6256140694947217e-05, "loss": 0.5131, "step": 8598 }, { "epoch": 0.3063576607228744, "grad_norm": 1.1920016019475088, "learning_rate": 1.6255240429542092e-05, "loss": 0.5709, "step": 8599 }, { "epoch": 0.30639328784936853, "grad_norm": 1.2328282350664908, "learning_rate": 1.6254340080844068e-05, "loss": 0.5128, "step": 8600 }, { "epoch": 0.3064289149758626, "grad_norm": 1.2105995713389142, "learning_rate": 1.6253439648865137e-05, "loss": 0.5162, "step": 8601 }, { "epoch": 0.30646454210235674, "grad_norm": 1.0273839539820964, "learning_rate": 1.6252539133617283e-05, "loss": 0.5107, "step": 8602 }, { "epoch": 0.30650016922885087, "grad_norm": 1.1163455910102698, "learning_rate": 1.6251638535112505e-05, "loss": 0.5051, "step": 8603 }, { "epoch": 0.30653579635534495, "grad_norm": 1.0406245728881698, "learning_rate": 1.6250737853362785e-05, "loss": 0.4875, "step": 8604 }, { "epoch": 0.3065714234818391, "grad_norm": 2.269326860566241, "learning_rate": 1.6249837088380122e-05, "loss": 0.575, "step": 8605 }, { "epoch": 0.3066070506083332, "grad_norm": 4.528603217766211, "learning_rate": 1.6248936240176512e-05, "loss": 0.4739, "step": 8606 }, { "epoch": 0.3066426777348273, "grad_norm": 1.1878115370647138, "learning_rate": 1.6248035308763947e-05, "loss": 0.5196, "step": 8607 }, { "epoch": 0.3066783048613214, "grad_norm": 1.16878886740069, "learning_rate": 1.6247134294154426e-05, "loss": 0.5025, "step": 8608 }, { "epoch": 0.30671393198781555, "grad_norm": 1.822409040581901, "learning_rate": 1.6246233196359944e-05, "loss": 0.6008, "step": 8609 }, { "epoch": 0.3067495591143096, "grad_norm": 1.1916469236549452, "learning_rate": 1.6245332015392502e-05, "loss": 0.4624, "step": 8610 }, { "epoch": 0.30678518624080375, "grad_norm": 3.2104623136343573, "learning_rate": 1.62444307512641e-05, "loss": 0.5417, "step": 8611 }, { "epoch": 0.3068208133672979, "grad_norm": 1.4628454668506663, "learning_rate": 1.6243529403986737e-05, "loss": 0.5322, "step": 8612 }, { "epoch": 0.30685644049379196, "grad_norm": 1.4410886709568171, "learning_rate": 1.6242627973572415e-05, "loss": 0.6022, "step": 8613 }, { "epoch": 0.3068920676202861, "grad_norm": 1.177516403852618, "learning_rate": 1.6241726460033136e-05, "loss": 0.5263, "step": 8614 }, { "epoch": 0.3069276947467802, "grad_norm": 2.3480863682473587, "learning_rate": 1.624082486338091e-05, "loss": 0.5566, "step": 8615 }, { "epoch": 0.3069633218732743, "grad_norm": 1.1830745626660404, "learning_rate": 1.623992318362774e-05, "loss": 0.492, "step": 8616 }, { "epoch": 0.30699894899976843, "grad_norm": 1.499388002807871, "learning_rate": 1.623902142078563e-05, "loss": 0.5505, "step": 8617 }, { "epoch": 0.30703457612626256, "grad_norm": 1.2572456754113213, "learning_rate": 1.623811957486659e-05, "loss": 0.4958, "step": 8618 }, { "epoch": 0.30707020325275663, "grad_norm": 1.2018877687134124, "learning_rate": 1.6237217645882623e-05, "loss": 0.5138, "step": 8619 }, { "epoch": 0.30710583037925077, "grad_norm": 1.2101568304059516, "learning_rate": 1.623631563384575e-05, "loss": 0.4425, "step": 8620 }, { "epoch": 0.3071414575057449, "grad_norm": 1.1420016480608102, "learning_rate": 1.6235413538767974e-05, "loss": 0.5366, "step": 8621 }, { "epoch": 0.307177084632239, "grad_norm": 1.9583445648802502, "learning_rate": 1.6234511360661305e-05, "loss": 0.614, "step": 8622 }, { "epoch": 0.3072127117587331, "grad_norm": 1.8203842010472537, "learning_rate": 1.623360909953776e-05, "loss": 0.5622, "step": 8623 }, { "epoch": 0.30724833888522723, "grad_norm": 1.30324578574817, "learning_rate": 1.6232706755409354e-05, "loss": 0.5388, "step": 8624 }, { "epoch": 0.3072839660117213, "grad_norm": 1.59349468350643, "learning_rate": 1.6231804328288104e-05, "loss": 0.6771, "step": 8625 }, { "epoch": 0.30731959313821544, "grad_norm": 1.4961853158953775, "learning_rate": 1.623090181818602e-05, "loss": 0.5135, "step": 8626 }, { "epoch": 0.30735522026470957, "grad_norm": 1.1883249124743998, "learning_rate": 1.6229999225115122e-05, "loss": 0.5129, "step": 8627 }, { "epoch": 0.30739084739120365, "grad_norm": 1.1508941376032014, "learning_rate": 1.6229096549087434e-05, "loss": 0.4815, "step": 8628 }, { "epoch": 0.3074264745176978, "grad_norm": 3.090281920109412, "learning_rate": 1.6228193790114968e-05, "loss": 0.5781, "step": 8629 }, { "epoch": 0.3074621016441919, "grad_norm": 1.7531232191019097, "learning_rate": 1.6227290948209748e-05, "loss": 0.6279, "step": 8630 }, { "epoch": 0.307497728770686, "grad_norm": 1.5772877440039494, "learning_rate": 1.62263880233838e-05, "loss": 0.5753, "step": 8631 }, { "epoch": 0.3075333558971801, "grad_norm": 2.569853988885915, "learning_rate": 1.6225485015649142e-05, "loss": 0.6295, "step": 8632 }, { "epoch": 0.30756898302367425, "grad_norm": 1.6113767380567656, "learning_rate": 1.62245819250178e-05, "loss": 0.5719, "step": 8633 }, { "epoch": 0.3076046101501683, "grad_norm": 2.0803152096135507, "learning_rate": 1.6223678751501798e-05, "loss": 0.4895, "step": 8634 }, { "epoch": 0.30764023727666245, "grad_norm": 1.33835752985127, "learning_rate": 1.6222775495113165e-05, "loss": 0.5465, "step": 8635 }, { "epoch": 0.3076758644031566, "grad_norm": 1.3752191323696978, "learning_rate": 1.6221872155863923e-05, "loss": 0.5265, "step": 8636 }, { "epoch": 0.30771149152965066, "grad_norm": 1.2023114487590236, "learning_rate": 1.6220968733766108e-05, "loss": 0.4273, "step": 8637 }, { "epoch": 0.3077471186561448, "grad_norm": 1.4477163151205106, "learning_rate": 1.6220065228831746e-05, "loss": 0.5482, "step": 8638 }, { "epoch": 0.3077827457826389, "grad_norm": 1.4858795539138545, "learning_rate": 1.6219161641072863e-05, "loss": 0.5806, "step": 8639 }, { "epoch": 0.307818372909133, "grad_norm": 1.0140871771255877, "learning_rate": 1.62182579705015e-05, "loss": 0.4564, "step": 8640 }, { "epoch": 0.30785400003562713, "grad_norm": 1.3920803061326112, "learning_rate": 1.621735421712969e-05, "loss": 0.5687, "step": 8641 }, { "epoch": 0.30788962716212126, "grad_norm": 1.0651981840036573, "learning_rate": 1.6216450380969456e-05, "loss": 0.5268, "step": 8642 }, { "epoch": 0.30792525428861534, "grad_norm": 1.0870107942383032, "learning_rate": 1.621554646203284e-05, "loss": 0.4445, "step": 8643 }, { "epoch": 0.30796088141510947, "grad_norm": 1.3744752346692894, "learning_rate": 1.6214642460331884e-05, "loss": 0.4838, "step": 8644 }, { "epoch": 0.3079965085416036, "grad_norm": 1.0594354218212243, "learning_rate": 1.6213738375878617e-05, "loss": 0.4711, "step": 8645 }, { "epoch": 0.3080321356680977, "grad_norm": 1.127569761868897, "learning_rate": 1.621283420868508e-05, "loss": 0.4604, "step": 8646 }, { "epoch": 0.3080677627945918, "grad_norm": 1.6921609018624728, "learning_rate": 1.6211929958763314e-05, "loss": 0.6122, "step": 8647 }, { "epoch": 0.30810338992108594, "grad_norm": 1.1723371269818146, "learning_rate": 1.6211025626125358e-05, "loss": 0.5695, "step": 8648 }, { "epoch": 0.30813901704758, "grad_norm": 2.2964807936846734, "learning_rate": 1.621012121078325e-05, "loss": 0.6189, "step": 8649 }, { "epoch": 0.30817464417407414, "grad_norm": 1.1390492863066926, "learning_rate": 1.6209216712749044e-05, "loss": 0.4831, "step": 8650 }, { "epoch": 0.3082102713005683, "grad_norm": 1.063330211455823, "learning_rate": 1.6208312132034776e-05, "loss": 0.4599, "step": 8651 }, { "epoch": 0.30824589842706235, "grad_norm": 1.0876025329329924, "learning_rate": 1.620740746865249e-05, "loss": 0.4559, "step": 8652 }, { "epoch": 0.3082815255535565, "grad_norm": 1.3754816331923736, "learning_rate": 1.620650272261424e-05, "loss": 0.5334, "step": 8653 }, { "epoch": 0.3083171526800506, "grad_norm": 1.0920939882917713, "learning_rate": 1.6205597893932065e-05, "loss": 0.4737, "step": 8654 }, { "epoch": 0.3083527798065447, "grad_norm": 2.211985680654155, "learning_rate": 1.6204692982618015e-05, "loss": 0.6459, "step": 8655 }, { "epoch": 0.3083884069330388, "grad_norm": 1.7310476137862154, "learning_rate": 1.6203787988684144e-05, "loss": 0.55, "step": 8656 }, { "epoch": 0.30842403405953295, "grad_norm": 1.1539341478059453, "learning_rate": 1.6202882912142497e-05, "loss": 0.5214, "step": 8657 }, { "epoch": 0.308459661186027, "grad_norm": 1.089209860502633, "learning_rate": 1.620197775300513e-05, "loss": 0.5018, "step": 8658 }, { "epoch": 0.30849528831252115, "grad_norm": 1.21284860322311, "learning_rate": 1.6201072511284096e-05, "loss": 0.4878, "step": 8659 }, { "epoch": 0.3085309154390153, "grad_norm": 1.885643827755521, "learning_rate": 1.6200167186991445e-05, "loss": 0.6913, "step": 8660 }, { "epoch": 0.30856654256550936, "grad_norm": 2.1427321821487615, "learning_rate": 1.6199261780139238e-05, "loss": 0.5812, "step": 8661 }, { "epoch": 0.3086021696920035, "grad_norm": 1.3124241508575925, "learning_rate": 1.619835629073952e-05, "loss": 0.4587, "step": 8662 }, { "epoch": 0.3086377968184976, "grad_norm": 1.8189424042463906, "learning_rate": 1.6197450718804358e-05, "loss": 0.6039, "step": 8663 }, { "epoch": 0.3086734239449917, "grad_norm": 1.5467063764799145, "learning_rate": 1.6196545064345813e-05, "loss": 0.6682, "step": 8664 }, { "epoch": 0.30870905107148583, "grad_norm": 1.9608662252800126, "learning_rate": 1.6195639327375935e-05, "loss": 0.5132, "step": 8665 }, { "epoch": 0.30874467819797996, "grad_norm": 1.180757680788658, "learning_rate": 1.619473350790679e-05, "loss": 0.4941, "step": 8666 }, { "epoch": 0.30878030532447404, "grad_norm": 1.5548935497065337, "learning_rate": 1.6193827605950438e-05, "loss": 0.5912, "step": 8667 }, { "epoch": 0.30881593245096817, "grad_norm": 1.227859602335175, "learning_rate": 1.6192921621518946e-05, "loss": 0.4909, "step": 8668 }, { "epoch": 0.3088515595774623, "grad_norm": 1.3864308065980748, "learning_rate": 1.6192015554624368e-05, "loss": 0.561, "step": 8669 }, { "epoch": 0.3088871867039564, "grad_norm": 1.8966592243704516, "learning_rate": 1.6191109405278776e-05, "loss": 0.5666, "step": 8670 }, { "epoch": 0.3089228138304505, "grad_norm": 0.9783860951818721, "learning_rate": 1.6190203173494236e-05, "loss": 0.4837, "step": 8671 }, { "epoch": 0.30895844095694464, "grad_norm": 1.6400876956056736, "learning_rate": 1.6189296859282816e-05, "loss": 0.639, "step": 8672 }, { "epoch": 0.3089940680834387, "grad_norm": 1.622303592339775, "learning_rate": 1.618839046265658e-05, "loss": 0.6165, "step": 8673 }, { "epoch": 0.30902969520993284, "grad_norm": 1.2463221797422945, "learning_rate": 1.6187483983627602e-05, "loss": 0.4769, "step": 8674 }, { "epoch": 0.309065322336427, "grad_norm": 1.7387176537727937, "learning_rate": 1.6186577422207946e-05, "loss": 0.5347, "step": 8675 }, { "epoch": 0.30910094946292105, "grad_norm": 1.8159391781950684, "learning_rate": 1.618567077840969e-05, "loss": 0.5628, "step": 8676 }, { "epoch": 0.3091365765894152, "grad_norm": 1.6227293632397288, "learning_rate": 1.6184764052244904e-05, "loss": 0.5985, "step": 8677 }, { "epoch": 0.3091722037159093, "grad_norm": 1.2306179487194038, "learning_rate": 1.6183857243725662e-05, "loss": 0.4827, "step": 8678 }, { "epoch": 0.3092078308424034, "grad_norm": 1.2134049484589138, "learning_rate": 1.618295035286404e-05, "loss": 0.4844, "step": 8679 }, { "epoch": 0.3092434579688975, "grad_norm": 2.1315507160316867, "learning_rate": 1.618204337967211e-05, "loss": 0.5802, "step": 8680 }, { "epoch": 0.30927908509539165, "grad_norm": 1.158142048576853, "learning_rate": 1.6181136324161954e-05, "loss": 0.5406, "step": 8681 }, { "epoch": 0.3093147122218857, "grad_norm": 1.1577557152767735, "learning_rate": 1.6180229186345643e-05, "loss": 0.4972, "step": 8682 }, { "epoch": 0.30935033934837985, "grad_norm": 1.5995359545734829, "learning_rate": 1.617932196623526e-05, "loss": 0.6356, "step": 8683 }, { "epoch": 0.309385966474874, "grad_norm": 1.7686973548277378, "learning_rate": 1.617841466384289e-05, "loss": 0.6528, "step": 8684 }, { "epoch": 0.30942159360136806, "grad_norm": 1.7382822251908345, "learning_rate": 1.617750727918061e-05, "loss": 0.5212, "step": 8685 }, { "epoch": 0.3094572207278622, "grad_norm": 1.690218460984882, "learning_rate": 1.6176599812260504e-05, "loss": 0.5954, "step": 8686 }, { "epoch": 0.3094928478543563, "grad_norm": 1.1879832631394285, "learning_rate": 1.617569226309465e-05, "loss": 0.5063, "step": 8687 }, { "epoch": 0.3095284749808504, "grad_norm": 1.2622305596910397, "learning_rate": 1.6174784631695138e-05, "loss": 0.5003, "step": 8688 }, { "epoch": 0.30956410210734453, "grad_norm": 1.1658532495471514, "learning_rate": 1.6173876918074057e-05, "loss": 0.5211, "step": 8689 }, { "epoch": 0.30959972923383866, "grad_norm": 1.529027306156567, "learning_rate": 1.6172969122243488e-05, "loss": 0.5634, "step": 8690 }, { "epoch": 0.30963535636033274, "grad_norm": 1.130562355976245, "learning_rate": 1.6172061244215517e-05, "loss": 0.4422, "step": 8691 }, { "epoch": 0.30967098348682687, "grad_norm": 2.2193935256775266, "learning_rate": 1.617115328400224e-05, "loss": 0.582, "step": 8692 }, { "epoch": 0.309706610613321, "grad_norm": 1.9459781213930485, "learning_rate": 1.617024524161574e-05, "loss": 0.5759, "step": 8693 }, { "epoch": 0.3097422377398151, "grad_norm": 1.606992966296374, "learning_rate": 1.6169337117068114e-05, "loss": 0.6777, "step": 8694 }, { "epoch": 0.3097778648663092, "grad_norm": 1.3022226190654649, "learning_rate": 1.6168428910371455e-05, "loss": 0.4778, "step": 8695 }, { "epoch": 0.30981349199280334, "grad_norm": 1.2344845684696049, "learning_rate": 1.6167520621537847e-05, "loss": 0.4826, "step": 8696 }, { "epoch": 0.3098491191192974, "grad_norm": 1.8741785376888025, "learning_rate": 1.6166612250579398e-05, "loss": 0.5252, "step": 8697 }, { "epoch": 0.30988474624579154, "grad_norm": 1.790816451260561, "learning_rate": 1.6165703797508193e-05, "loss": 0.4545, "step": 8698 }, { "epoch": 0.3099203733722857, "grad_norm": 1.16228473422331, "learning_rate": 1.6164795262336334e-05, "loss": 0.449, "step": 8699 }, { "epoch": 0.30995600049877975, "grad_norm": 1.7606248798447528, "learning_rate": 1.6163886645075916e-05, "loss": 0.6473, "step": 8700 }, { "epoch": 0.3099916276252739, "grad_norm": 1.6482074872205124, "learning_rate": 1.6162977945739036e-05, "loss": 0.6149, "step": 8701 }, { "epoch": 0.310027254751768, "grad_norm": 1.187130514521613, "learning_rate": 1.6162069164337803e-05, "loss": 0.504, "step": 8702 }, { "epoch": 0.3100628818782621, "grad_norm": 1.399367115348653, "learning_rate": 1.6161160300884308e-05, "loss": 0.4912, "step": 8703 }, { "epoch": 0.3100985090047562, "grad_norm": 1.2360272969785124, "learning_rate": 1.6160251355390658e-05, "loss": 0.4975, "step": 8704 }, { "epoch": 0.31013413613125035, "grad_norm": 1.2412870781002594, "learning_rate": 1.615934232786896e-05, "loss": 0.4844, "step": 8705 }, { "epoch": 0.3101697632577444, "grad_norm": 1.1510880258756127, "learning_rate": 1.6158433218331305e-05, "loss": 0.5149, "step": 8706 }, { "epoch": 0.31020539038423856, "grad_norm": 1.6541908692135263, "learning_rate": 1.6157524026789814e-05, "loss": 0.6261, "step": 8707 }, { "epoch": 0.3102410175107327, "grad_norm": 1.8008553519249864, "learning_rate": 1.6156614753256583e-05, "loss": 0.6207, "step": 8708 }, { "epoch": 0.31027664463722676, "grad_norm": 1.225562671865531, "learning_rate": 1.6155705397743726e-05, "loss": 0.4719, "step": 8709 }, { "epoch": 0.3103122717637209, "grad_norm": 1.294607920838987, "learning_rate": 1.615479596026335e-05, "loss": 0.5211, "step": 8710 }, { "epoch": 0.310347898890215, "grad_norm": 1.6897457919322776, "learning_rate": 1.615388644082756e-05, "loss": 0.5464, "step": 8711 }, { "epoch": 0.3103835260167091, "grad_norm": 1.5398586021479321, "learning_rate": 1.6152976839448472e-05, "loss": 0.5623, "step": 8712 }, { "epoch": 0.31041915314320323, "grad_norm": 1.2704672627986178, "learning_rate": 1.6152067156138198e-05, "loss": 0.4403, "step": 8713 }, { "epoch": 0.31045478026969736, "grad_norm": 1.3681716888164794, "learning_rate": 1.615115739090885e-05, "loss": 0.5096, "step": 8714 }, { "epoch": 0.31049040739619144, "grad_norm": 1.46138716673416, "learning_rate": 1.615024754377254e-05, "loss": 0.4413, "step": 8715 }, { "epoch": 0.31052603452268557, "grad_norm": 1.7707196560867802, "learning_rate": 1.6149337614741388e-05, "loss": 0.5964, "step": 8716 }, { "epoch": 0.3105616616491797, "grad_norm": 1.390306366364986, "learning_rate": 1.6148427603827504e-05, "loss": 0.4763, "step": 8717 }, { "epoch": 0.3105972887756738, "grad_norm": 1.3248297243467135, "learning_rate": 1.614751751104301e-05, "loss": 0.4786, "step": 8718 }, { "epoch": 0.3106329159021679, "grad_norm": 1.734322645058835, "learning_rate": 1.6146607336400024e-05, "loss": 0.5376, "step": 8719 }, { "epoch": 0.31066854302866204, "grad_norm": 1.493557885049933, "learning_rate": 1.6145697079910666e-05, "loss": 0.6382, "step": 8720 }, { "epoch": 0.3107041701551561, "grad_norm": 1.22711555063104, "learning_rate": 1.6144786741587056e-05, "loss": 0.4475, "step": 8721 }, { "epoch": 0.31073979728165024, "grad_norm": 1.76129437950309, "learning_rate": 1.6143876321441313e-05, "loss": 0.5312, "step": 8722 }, { "epoch": 0.3107754244081444, "grad_norm": 1.5951519984303097, "learning_rate": 1.6142965819485564e-05, "loss": 0.5486, "step": 8723 }, { "epoch": 0.31081105153463845, "grad_norm": 1.784190332009495, "learning_rate": 1.614205523573193e-05, "loss": 0.6231, "step": 8724 }, { "epoch": 0.3108466786611326, "grad_norm": 2.4265215844222356, "learning_rate": 1.6141144570192538e-05, "loss": 0.5514, "step": 8725 }, { "epoch": 0.3108823057876267, "grad_norm": 1.6399548240850867, "learning_rate": 1.614023382287952e-05, "loss": 0.6201, "step": 8726 }, { "epoch": 0.3109179329141208, "grad_norm": 1.4104166719847808, "learning_rate": 1.613932299380499e-05, "loss": 0.5229, "step": 8727 }, { "epoch": 0.3109535600406149, "grad_norm": 2.0180310102676637, "learning_rate": 1.6138412082981084e-05, "loss": 0.6174, "step": 8728 }, { "epoch": 0.31098918716710905, "grad_norm": 1.6959191749277776, "learning_rate": 1.6137501090419932e-05, "loss": 0.5007, "step": 8729 }, { "epoch": 0.3110248142936031, "grad_norm": 1.8916904204162655, "learning_rate": 1.6136590016133663e-05, "loss": 0.5394, "step": 8730 }, { "epoch": 0.31106044142009726, "grad_norm": 1.5124966786968381, "learning_rate": 1.6135678860134408e-05, "loss": 0.5701, "step": 8731 }, { "epoch": 0.3110960685465914, "grad_norm": 1.6387074214203161, "learning_rate": 1.61347676224343e-05, "loss": 0.601, "step": 8732 }, { "epoch": 0.31113169567308546, "grad_norm": 1.9371863298347276, "learning_rate": 1.6133856303045476e-05, "loss": 0.5214, "step": 8733 }, { "epoch": 0.3111673227995796, "grad_norm": 1.2367198466152847, "learning_rate": 1.6132944901980062e-05, "loss": 0.4866, "step": 8734 }, { "epoch": 0.3112029499260737, "grad_norm": 2.7705202052444062, "learning_rate": 1.6132033419250203e-05, "loss": 0.4971, "step": 8735 }, { "epoch": 0.3112385770525678, "grad_norm": 1.2733264540109432, "learning_rate": 1.6131121854868035e-05, "loss": 0.447, "step": 8736 }, { "epoch": 0.31127420417906193, "grad_norm": 1.3446976136704774, "learning_rate": 1.6130210208845695e-05, "loss": 0.5343, "step": 8737 }, { "epoch": 0.31130983130555606, "grad_norm": 1.860182864858488, "learning_rate": 1.612929848119532e-05, "loss": 0.7135, "step": 8738 }, { "epoch": 0.31134545843205014, "grad_norm": 1.6954740936178958, "learning_rate": 1.612838667192905e-05, "loss": 0.483, "step": 8739 }, { "epoch": 0.31138108555854427, "grad_norm": 1.2899785546639102, "learning_rate": 1.6127474781059028e-05, "loss": 0.4509, "step": 8740 }, { "epoch": 0.3114167126850384, "grad_norm": 1.1927811486508966, "learning_rate": 1.61265628085974e-05, "loss": 0.4594, "step": 8741 }, { "epoch": 0.3114523398115325, "grad_norm": 1.6951250159414786, "learning_rate": 1.61256507545563e-05, "loss": 0.4784, "step": 8742 }, { "epoch": 0.3114879669380266, "grad_norm": 1.184603252246146, "learning_rate": 1.612473861894788e-05, "loss": 0.5262, "step": 8743 }, { "epoch": 0.31152359406452074, "grad_norm": 1.3210020505423838, "learning_rate": 1.612382640178429e-05, "loss": 0.5418, "step": 8744 }, { "epoch": 0.3115592211910148, "grad_norm": 1.1372636207435654, "learning_rate": 1.6122914103077665e-05, "loss": 0.485, "step": 8745 }, { "epoch": 0.31159484831750894, "grad_norm": 1.4102512643533602, "learning_rate": 1.6122001722840162e-05, "loss": 0.5045, "step": 8746 }, { "epoch": 0.3116304754440031, "grad_norm": 1.4713318229461365, "learning_rate": 1.6121089261083927e-05, "loss": 0.5726, "step": 8747 }, { "epoch": 0.31166610257049715, "grad_norm": 1.159504933637949, "learning_rate": 1.6120176717821107e-05, "loss": 0.4476, "step": 8748 }, { "epoch": 0.3117017296969913, "grad_norm": 1.115323553371223, "learning_rate": 1.6119264093063864e-05, "loss": 0.5272, "step": 8749 }, { "epoch": 0.3117373568234854, "grad_norm": 1.6594924295853497, "learning_rate": 1.6118351386824335e-05, "loss": 0.4633, "step": 8750 }, { "epoch": 0.3117729839499795, "grad_norm": 1.5958520036016102, "learning_rate": 1.611743859911468e-05, "loss": 0.5407, "step": 8751 }, { "epoch": 0.3118086110764736, "grad_norm": 1.6366477416960659, "learning_rate": 1.611652572994706e-05, "loss": 0.5584, "step": 8752 }, { "epoch": 0.31184423820296775, "grad_norm": 1.4558449803666984, "learning_rate": 1.6115612779333624e-05, "loss": 0.5237, "step": 8753 }, { "epoch": 0.3118798653294618, "grad_norm": 1.1903616809033957, "learning_rate": 1.6114699747286528e-05, "loss": 0.521, "step": 8754 }, { "epoch": 0.31191549245595596, "grad_norm": 2.053601223511052, "learning_rate": 1.6113786633817925e-05, "loss": 0.5928, "step": 8755 }, { "epoch": 0.3119511195824501, "grad_norm": 1.6045243490082173, "learning_rate": 1.6112873438939987e-05, "loss": 0.5691, "step": 8756 }, { "epoch": 0.31198674670894416, "grad_norm": 2.164713152998222, "learning_rate": 1.6111960162664863e-05, "loss": 0.5156, "step": 8757 }, { "epoch": 0.3120223738354383, "grad_norm": 1.2577597678458896, "learning_rate": 1.611104680500472e-05, "loss": 0.5998, "step": 8758 }, { "epoch": 0.3120580009619324, "grad_norm": 1.057776418529825, "learning_rate": 1.6110133365971715e-05, "loss": 0.4842, "step": 8759 }, { "epoch": 0.31209362808842656, "grad_norm": 0.9918395771517844, "learning_rate": 1.6109219845578016e-05, "loss": 0.5031, "step": 8760 }, { "epoch": 0.31212925521492063, "grad_norm": 1.1539589700859838, "learning_rate": 1.6108306243835784e-05, "loss": 0.5006, "step": 8761 }, { "epoch": 0.31216488234141476, "grad_norm": 2.4758902801754856, "learning_rate": 1.6107392560757182e-05, "loss": 0.6051, "step": 8762 }, { "epoch": 0.3122005094679089, "grad_norm": 1.1223604836118368, "learning_rate": 1.6106478796354382e-05, "loss": 0.4766, "step": 8763 }, { "epoch": 0.31223613659440297, "grad_norm": 1.5732815404755238, "learning_rate": 1.610556495063955e-05, "loss": 0.6334, "step": 8764 }, { "epoch": 0.3122717637208971, "grad_norm": 1.6732111141678216, "learning_rate": 1.610465102362485e-05, "loss": 0.6059, "step": 8765 }, { "epoch": 0.31230739084739123, "grad_norm": 1.6336197405073918, "learning_rate": 1.6103737015322458e-05, "loss": 0.5837, "step": 8766 }, { "epoch": 0.3123430179738853, "grad_norm": 1.544841338463815, "learning_rate": 1.610282292574454e-05, "loss": 0.6649, "step": 8767 }, { "epoch": 0.31237864510037944, "grad_norm": 1.6326445072119908, "learning_rate": 1.610190875490327e-05, "loss": 0.5816, "step": 8768 }, { "epoch": 0.31241427222687357, "grad_norm": 1.0733706725457088, "learning_rate": 1.6100994502810813e-05, "loss": 0.4519, "step": 8769 }, { "epoch": 0.31244989935336764, "grad_norm": 1.161484372548981, "learning_rate": 1.6100080169479357e-05, "loss": 0.4405, "step": 8770 }, { "epoch": 0.3124855264798618, "grad_norm": 1.7703737978616485, "learning_rate": 1.609916575492107e-05, "loss": 0.5414, "step": 8771 }, { "epoch": 0.3125211536063559, "grad_norm": 1.5046324863180873, "learning_rate": 1.6098251259148125e-05, "loss": 0.5824, "step": 8772 }, { "epoch": 0.31255678073285, "grad_norm": 1.0768955674062888, "learning_rate": 1.6097336682172703e-05, "loss": 0.5021, "step": 8773 }, { "epoch": 0.3125924078593441, "grad_norm": 1.8279838202863439, "learning_rate": 1.609642202400698e-05, "loss": 0.6329, "step": 8774 }, { "epoch": 0.31262803498583824, "grad_norm": 1.353589832202095, "learning_rate": 1.609550728466314e-05, "loss": 0.579, "step": 8775 }, { "epoch": 0.3126636621123323, "grad_norm": 1.1796994345959162, "learning_rate": 1.609459246415336e-05, "loss": 0.4799, "step": 8776 }, { "epoch": 0.31269928923882645, "grad_norm": 1.5167743428892728, "learning_rate": 1.609367756248982e-05, "loss": 0.5517, "step": 8777 }, { "epoch": 0.3127349163653206, "grad_norm": 1.612047211706241, "learning_rate": 1.6092762579684703e-05, "loss": 0.6655, "step": 8778 }, { "epoch": 0.31277054349181466, "grad_norm": 2.143560720861461, "learning_rate": 1.6091847515750192e-05, "loss": 0.6904, "step": 8779 }, { "epoch": 0.3128061706183088, "grad_norm": 1.0667602129915394, "learning_rate": 1.6090932370698477e-05, "loss": 0.4789, "step": 8780 }, { "epoch": 0.3128417977448029, "grad_norm": 1.4556924229298531, "learning_rate": 1.6090017144541737e-05, "loss": 0.5551, "step": 8781 }, { "epoch": 0.312877424871297, "grad_norm": 2.0373166601389365, "learning_rate": 1.6089101837292168e-05, "loss": 0.5414, "step": 8782 }, { "epoch": 0.3129130519977911, "grad_norm": 1.1258996073674188, "learning_rate": 1.6088186448961948e-05, "loss": 0.4847, "step": 8783 }, { "epoch": 0.31294867912428526, "grad_norm": 1.1419505063839748, "learning_rate": 1.608727097956327e-05, "loss": 0.4857, "step": 8784 }, { "epoch": 0.31298430625077933, "grad_norm": 1.0541084004082104, "learning_rate": 1.608635542910832e-05, "loss": 0.4406, "step": 8785 }, { "epoch": 0.31301993337727346, "grad_norm": 1.113273370411686, "learning_rate": 1.60854397976093e-05, "loss": 0.503, "step": 8786 }, { "epoch": 0.3130555605037676, "grad_norm": 1.2008395385101938, "learning_rate": 1.6084524085078392e-05, "loss": 0.4962, "step": 8787 }, { "epoch": 0.31309118763026167, "grad_norm": 1.0579920690992952, "learning_rate": 1.6083608291527793e-05, "loss": 0.4693, "step": 8788 }, { "epoch": 0.3131268147567558, "grad_norm": 1.5484306809700976, "learning_rate": 1.6082692416969698e-05, "loss": 0.5542, "step": 8789 }, { "epoch": 0.31316244188324993, "grad_norm": 1.045137878503815, "learning_rate": 1.6081776461416306e-05, "loss": 0.4759, "step": 8790 }, { "epoch": 0.313198069009744, "grad_norm": 1.1958138496486883, "learning_rate": 1.6080860424879805e-05, "loss": 0.5025, "step": 8791 }, { "epoch": 0.31323369613623814, "grad_norm": 1.6313402246245374, "learning_rate": 1.60799443073724e-05, "loss": 0.5513, "step": 8792 }, { "epoch": 0.31326932326273227, "grad_norm": 1.1264910447174636, "learning_rate": 1.607902810890628e-05, "loss": 0.5107, "step": 8793 }, { "epoch": 0.31330495038922634, "grad_norm": 1.1153389660246094, "learning_rate": 1.6078111829493657e-05, "loss": 0.4833, "step": 8794 }, { "epoch": 0.3133405775157205, "grad_norm": 1.0273600461685992, "learning_rate": 1.607719546914673e-05, "loss": 0.4903, "step": 8795 }, { "epoch": 0.3133762046422146, "grad_norm": 2.081304919453432, "learning_rate": 1.6076279027877694e-05, "loss": 0.6008, "step": 8796 }, { "epoch": 0.3134118317687087, "grad_norm": 1.429026474328582, "learning_rate": 1.6075362505698756e-05, "loss": 0.5903, "step": 8797 }, { "epoch": 0.3134474588952028, "grad_norm": 1.1096123015059798, "learning_rate": 1.6074445902622117e-05, "loss": 0.5169, "step": 8798 }, { "epoch": 0.31348308602169694, "grad_norm": 1.0185052211077592, "learning_rate": 1.607352921865999e-05, "loss": 0.4959, "step": 8799 }, { "epoch": 0.313518713148191, "grad_norm": 1.5216484400775985, "learning_rate": 1.6072612453824575e-05, "loss": 0.5752, "step": 8800 }, { "epoch": 0.31355434027468515, "grad_norm": 1.1002109420116104, "learning_rate": 1.6071695608128077e-05, "loss": 0.4643, "step": 8801 }, { "epoch": 0.3135899674011793, "grad_norm": 1.1508439279263887, "learning_rate": 1.607077868158271e-05, "loss": 0.4708, "step": 8802 }, { "epoch": 0.31362559452767336, "grad_norm": 1.6791514334824256, "learning_rate": 1.6069861674200682e-05, "loss": 0.5835, "step": 8803 }, { "epoch": 0.3136612216541675, "grad_norm": 1.717102141367581, "learning_rate": 1.6068944585994203e-05, "loss": 0.6509, "step": 8804 }, { "epoch": 0.3136968487806616, "grad_norm": 1.2801847123513628, "learning_rate": 1.606802741697549e-05, "loss": 0.4782, "step": 8805 }, { "epoch": 0.3137324759071557, "grad_norm": 1.498994088760058, "learning_rate": 1.6067110167156743e-05, "loss": 0.7102, "step": 8806 }, { "epoch": 0.3137681030336498, "grad_norm": 2.446154521475103, "learning_rate": 1.6066192836550186e-05, "loss": 0.5457, "step": 8807 }, { "epoch": 0.31380373016014396, "grad_norm": 1.7232556203093412, "learning_rate": 1.6065275425168034e-05, "loss": 0.55, "step": 8808 }, { "epoch": 0.31383935728663803, "grad_norm": 1.4135673485384024, "learning_rate": 1.6064357933022496e-05, "loss": 0.5144, "step": 8809 }, { "epoch": 0.31387498441313216, "grad_norm": 1.9428368386349282, "learning_rate": 1.6063440360125797e-05, "loss": 0.6206, "step": 8810 }, { "epoch": 0.3139106115396263, "grad_norm": 2.3787864613637155, "learning_rate": 1.6062522706490148e-05, "loss": 0.5606, "step": 8811 }, { "epoch": 0.31394623866612037, "grad_norm": 1.216674680339361, "learning_rate": 1.6061604972127774e-05, "loss": 0.5039, "step": 8812 }, { "epoch": 0.3139818657926145, "grad_norm": 1.1263256824302175, "learning_rate": 1.6060687157050894e-05, "loss": 0.4407, "step": 8813 }, { "epoch": 0.31401749291910863, "grad_norm": 1.0836496159538225, "learning_rate": 1.6059769261271727e-05, "loss": 0.5158, "step": 8814 }, { "epoch": 0.3140531200456027, "grad_norm": 1.223829765546032, "learning_rate": 1.6058851284802498e-05, "loss": 0.5593, "step": 8815 }, { "epoch": 0.31408874717209684, "grad_norm": 1.1009444510116033, "learning_rate": 1.6057933227655427e-05, "loss": 0.4674, "step": 8816 }, { "epoch": 0.31412437429859097, "grad_norm": 1.7651187991934985, "learning_rate": 1.605701508984274e-05, "loss": 0.5837, "step": 8817 }, { "epoch": 0.31416000142508504, "grad_norm": 1.0911217826609756, "learning_rate": 1.6056096871376667e-05, "loss": 0.4773, "step": 8818 }, { "epoch": 0.3141956285515792, "grad_norm": 1.9836795331418662, "learning_rate": 1.6055178572269433e-05, "loss": 0.5145, "step": 8819 }, { "epoch": 0.3142312556780733, "grad_norm": 1.0699872131530965, "learning_rate": 1.605426019253326e-05, "loss": 0.5001, "step": 8820 }, { "epoch": 0.3142668828045674, "grad_norm": 1.0751795202128573, "learning_rate": 1.6053341732180382e-05, "loss": 0.4724, "step": 8821 }, { "epoch": 0.3143025099310615, "grad_norm": 2.061517853750352, "learning_rate": 1.6052423191223026e-05, "loss": 0.6193, "step": 8822 }, { "epoch": 0.31433813705755564, "grad_norm": 1.1295593836958377, "learning_rate": 1.6051504569673426e-05, "loss": 0.4992, "step": 8823 }, { "epoch": 0.3143737641840497, "grad_norm": 2.343723257055822, "learning_rate": 1.6050585867543814e-05, "loss": 0.5699, "step": 8824 }, { "epoch": 0.31440939131054385, "grad_norm": 1.084585479709645, "learning_rate": 1.604966708484642e-05, "loss": 0.4693, "step": 8825 }, { "epoch": 0.314445018437038, "grad_norm": 1.2172522759596773, "learning_rate": 1.604874822159348e-05, "loss": 0.4917, "step": 8826 }, { "epoch": 0.31448064556353206, "grad_norm": 2.0244548843335712, "learning_rate": 1.604782927779723e-05, "loss": 0.6082, "step": 8827 }, { "epoch": 0.3145162726900262, "grad_norm": 1.131832224909272, "learning_rate": 1.6046910253469908e-05, "loss": 0.5029, "step": 8828 }, { "epoch": 0.3145518998165203, "grad_norm": 1.6438028700125946, "learning_rate": 1.6045991148623752e-05, "loss": 0.5665, "step": 8829 }, { "epoch": 0.3145875269430144, "grad_norm": 1.113253575802499, "learning_rate": 1.604507196327099e-05, "loss": 0.4921, "step": 8830 }, { "epoch": 0.3146231540695085, "grad_norm": 1.2254404445367657, "learning_rate": 1.6044152697423876e-05, "loss": 0.4977, "step": 8831 }, { "epoch": 0.31465878119600266, "grad_norm": 1.9930344505885105, "learning_rate": 1.604323335109464e-05, "loss": 0.5355, "step": 8832 }, { "epoch": 0.31469440832249673, "grad_norm": 1.5975598773600213, "learning_rate": 1.6042313924295535e-05, "loss": 0.6203, "step": 8833 }, { "epoch": 0.31473003544899086, "grad_norm": 1.1929901083728904, "learning_rate": 1.6041394417038788e-05, "loss": 0.4859, "step": 8834 }, { "epoch": 0.314765662575485, "grad_norm": 1.1514362338464113, "learning_rate": 1.604047482933666e-05, "loss": 0.4976, "step": 8835 }, { "epoch": 0.31480128970197907, "grad_norm": 3.1022748563124605, "learning_rate": 1.603955516120138e-05, "loss": 0.5988, "step": 8836 }, { "epoch": 0.3148369168284732, "grad_norm": 1.138319946978169, "learning_rate": 1.6038635412645204e-05, "loss": 0.5383, "step": 8837 }, { "epoch": 0.31487254395496733, "grad_norm": 1.0203288020407384, "learning_rate": 1.6037715583680377e-05, "loss": 0.4357, "step": 8838 }, { "epoch": 0.3149081710814614, "grad_norm": 1.180297312106556, "learning_rate": 1.603679567431915e-05, "loss": 0.5093, "step": 8839 }, { "epoch": 0.31494379820795554, "grad_norm": 1.8163132970247942, "learning_rate": 1.6035875684573764e-05, "loss": 0.5122, "step": 8840 }, { "epoch": 0.31497942533444967, "grad_norm": 1.2190804400933006, "learning_rate": 1.603495561445648e-05, "loss": 0.5221, "step": 8841 }, { "epoch": 0.31501505246094375, "grad_norm": 1.132030523672967, "learning_rate": 1.603403546397954e-05, "loss": 0.5148, "step": 8842 }, { "epoch": 0.3150506795874379, "grad_norm": 1.2987117503342938, "learning_rate": 1.6033115233155202e-05, "loss": 0.5327, "step": 8843 }, { "epoch": 0.315086306713932, "grad_norm": 1.475749215757613, "learning_rate": 1.6032194921995718e-05, "loss": 0.4582, "step": 8844 }, { "epoch": 0.3151219338404261, "grad_norm": 1.3084919179818055, "learning_rate": 1.6031274530513342e-05, "loss": 0.4985, "step": 8845 }, { "epoch": 0.3151575609669202, "grad_norm": 0.981078438115194, "learning_rate": 1.6030354058720328e-05, "loss": 0.4475, "step": 8846 }, { "epoch": 0.31519318809341434, "grad_norm": 1.1537295414709072, "learning_rate": 1.6029433506628932e-05, "loss": 0.4799, "step": 8847 }, { "epoch": 0.3152288152199084, "grad_norm": 1.0336989810978956, "learning_rate": 1.6028512874251423e-05, "loss": 0.455, "step": 8848 }, { "epoch": 0.31526444234640255, "grad_norm": 1.133964567621442, "learning_rate": 1.6027592161600046e-05, "loss": 0.5005, "step": 8849 }, { "epoch": 0.3153000694728967, "grad_norm": 2.358139636633606, "learning_rate": 1.6026671368687066e-05, "loss": 0.537, "step": 8850 }, { "epoch": 0.31533569659939076, "grad_norm": 1.155881206186914, "learning_rate": 1.6025750495524745e-05, "loss": 0.5326, "step": 8851 }, { "epoch": 0.3153713237258849, "grad_norm": 1.2489781416610917, "learning_rate": 1.6024829542125347e-05, "loss": 0.4929, "step": 8852 }, { "epoch": 0.315406950852379, "grad_norm": 1.491461710095168, "learning_rate": 1.602390850850113e-05, "loss": 0.562, "step": 8853 }, { "epoch": 0.3154425779788731, "grad_norm": 1.382059306302741, "learning_rate": 1.602298739466436e-05, "loss": 0.5204, "step": 8854 }, { "epoch": 0.3154782051053672, "grad_norm": 1.1656202059594143, "learning_rate": 1.6022066200627306e-05, "loss": 0.5218, "step": 8855 }, { "epoch": 0.31551383223186136, "grad_norm": 1.9267600854166547, "learning_rate": 1.602114492640223e-05, "loss": 0.6615, "step": 8856 }, { "epoch": 0.31554945935835543, "grad_norm": 1.248626930804629, "learning_rate": 1.6020223572001403e-05, "loss": 0.5234, "step": 8857 }, { "epoch": 0.31558508648484956, "grad_norm": 1.6191334850787351, "learning_rate": 1.6019302137437086e-05, "loss": 0.6055, "step": 8858 }, { "epoch": 0.3156207136113437, "grad_norm": 1.0631681863323517, "learning_rate": 1.6018380622721554e-05, "loss": 0.527, "step": 8859 }, { "epoch": 0.31565634073783777, "grad_norm": 1.6419205767319007, "learning_rate": 1.601745902786708e-05, "loss": 0.599, "step": 8860 }, { "epoch": 0.3156919678643319, "grad_norm": 1.0592044377940195, "learning_rate": 1.6016537352885936e-05, "loss": 0.4871, "step": 8861 }, { "epoch": 0.31572759499082603, "grad_norm": 1.7634894098797633, "learning_rate": 1.601561559779039e-05, "loss": 0.6021, "step": 8862 }, { "epoch": 0.3157632221173201, "grad_norm": 1.21838973616266, "learning_rate": 1.6014693762592716e-05, "loss": 0.5212, "step": 8863 }, { "epoch": 0.31579884924381424, "grad_norm": 1.1973669467873014, "learning_rate": 1.6013771847305187e-05, "loss": 0.556, "step": 8864 }, { "epoch": 0.31583447637030837, "grad_norm": 1.0864970917435552, "learning_rate": 1.6012849851940088e-05, "loss": 0.4913, "step": 8865 }, { "epoch": 0.31587010349680245, "grad_norm": 1.6599184545432581, "learning_rate": 1.601192777650969e-05, "loss": 0.5801, "step": 8866 }, { "epoch": 0.3159057306232966, "grad_norm": 1.0174573199704444, "learning_rate": 1.601100562102627e-05, "loss": 0.4471, "step": 8867 }, { "epoch": 0.3159413577497907, "grad_norm": 3.0097233962109757, "learning_rate": 1.601008338550211e-05, "loss": 0.5987, "step": 8868 }, { "epoch": 0.3159769848762848, "grad_norm": 2.1622929257930976, "learning_rate": 1.6009161069949488e-05, "loss": 0.6294, "step": 8869 }, { "epoch": 0.3160126120027789, "grad_norm": 1.6000642813230719, "learning_rate": 1.6008238674380688e-05, "loss": 0.5946, "step": 8870 }, { "epoch": 0.31604823912927305, "grad_norm": 1.941818248115084, "learning_rate": 1.6007316198807986e-05, "loss": 0.5982, "step": 8871 }, { "epoch": 0.3160838662557671, "grad_norm": 1.908024679595133, "learning_rate": 1.6006393643243675e-05, "loss": 0.6044, "step": 8872 }, { "epoch": 0.31611949338226125, "grad_norm": 1.1740006870060529, "learning_rate": 1.6005471007700033e-05, "loss": 0.5148, "step": 8873 }, { "epoch": 0.3161551205087554, "grad_norm": 1.8871014999546059, "learning_rate": 1.6004548292189345e-05, "loss": 0.6078, "step": 8874 }, { "epoch": 0.31619074763524946, "grad_norm": 1.0661898567799386, "learning_rate": 1.60036254967239e-05, "loss": 0.4818, "step": 8875 }, { "epoch": 0.3162263747617436, "grad_norm": 1.6289406903880066, "learning_rate": 1.6002702621315988e-05, "loss": 0.6146, "step": 8876 }, { "epoch": 0.3162620018882377, "grad_norm": 1.3626975229120526, "learning_rate": 1.6001779665977892e-05, "loss": 0.5312, "step": 8877 }, { "epoch": 0.3162976290147318, "grad_norm": 1.832900282965291, "learning_rate": 1.6000856630721906e-05, "loss": 0.5356, "step": 8878 }, { "epoch": 0.3163332561412259, "grad_norm": 1.1472286468080035, "learning_rate": 1.599993351556032e-05, "loss": 0.4758, "step": 8879 }, { "epoch": 0.31636888326772006, "grad_norm": 1.093288690225771, "learning_rate": 1.5999010320505424e-05, "loss": 0.5113, "step": 8880 }, { "epoch": 0.31640451039421413, "grad_norm": 1.2463529711431747, "learning_rate": 1.5998087045569515e-05, "loss": 0.4859, "step": 8881 }, { "epoch": 0.31644013752070826, "grad_norm": 1.206565040876919, "learning_rate": 1.599716369076488e-05, "loss": 0.54, "step": 8882 }, { "epoch": 0.3164757646472024, "grad_norm": 1.6975011274099372, "learning_rate": 1.5996240256103824e-05, "loss": 0.5343, "step": 8883 }, { "epoch": 0.31651139177369647, "grad_norm": 1.6319619535007053, "learning_rate": 1.5995316741598636e-05, "loss": 0.5849, "step": 8884 }, { "epoch": 0.3165470189001906, "grad_norm": 1.6166692446684001, "learning_rate": 1.5994393147261616e-05, "loss": 0.5076, "step": 8885 }, { "epoch": 0.31658264602668473, "grad_norm": 1.3623078001853086, "learning_rate": 1.599346947310506e-05, "loss": 0.5884, "step": 8886 }, { "epoch": 0.3166182731531788, "grad_norm": 1.2250103444977702, "learning_rate": 1.5992545719141268e-05, "loss": 0.5036, "step": 8887 }, { "epoch": 0.31665390027967294, "grad_norm": 1.1962513781104631, "learning_rate": 1.5991621885382542e-05, "loss": 0.489, "step": 8888 }, { "epoch": 0.31668952740616707, "grad_norm": 1.1935138217119803, "learning_rate": 1.5990697971841183e-05, "loss": 0.5663, "step": 8889 }, { "epoch": 0.31672515453266115, "grad_norm": 1.6583262843950366, "learning_rate": 1.5989773978529496e-05, "loss": 0.6083, "step": 8890 }, { "epoch": 0.3167607816591553, "grad_norm": 2.136401665655915, "learning_rate": 1.5988849905459775e-05, "loss": 0.5347, "step": 8891 }, { "epoch": 0.3167964087856494, "grad_norm": 1.8381232007771906, "learning_rate": 1.5987925752644334e-05, "loss": 0.6634, "step": 8892 }, { "epoch": 0.3168320359121435, "grad_norm": 1.04753899995664, "learning_rate": 1.598700152009548e-05, "loss": 0.4729, "step": 8893 }, { "epoch": 0.3168676630386376, "grad_norm": 1.9650702855995938, "learning_rate": 1.5986077207825513e-05, "loss": 0.6039, "step": 8894 }, { "epoch": 0.31690329016513175, "grad_norm": 1.2621693647017709, "learning_rate": 1.5985152815846745e-05, "loss": 0.5278, "step": 8895 }, { "epoch": 0.3169389172916258, "grad_norm": 1.1377498879833547, "learning_rate": 1.5984228344171484e-05, "loss": 0.4955, "step": 8896 }, { "epoch": 0.31697454441811995, "grad_norm": 1.2824370125962665, "learning_rate": 1.598330379281204e-05, "loss": 0.5286, "step": 8897 }, { "epoch": 0.3170101715446141, "grad_norm": 1.0621056207261947, "learning_rate": 1.5982379161780722e-05, "loss": 0.4775, "step": 8898 }, { "epoch": 0.31704579867110816, "grad_norm": 1.9371093382729148, "learning_rate": 1.5981454451089846e-05, "loss": 0.6053, "step": 8899 }, { "epoch": 0.3170814257976023, "grad_norm": 1.2202284142063842, "learning_rate": 1.5980529660751723e-05, "loss": 0.5107, "step": 8900 }, { "epoch": 0.3171170529240964, "grad_norm": 1.1651828327442775, "learning_rate": 1.597960479077867e-05, "loss": 0.4787, "step": 8901 }, { "epoch": 0.3171526800505905, "grad_norm": 1.086551022136999, "learning_rate": 1.5978679841182995e-05, "loss": 0.5043, "step": 8902 }, { "epoch": 0.3171883071770846, "grad_norm": 1.6637190942044644, "learning_rate": 1.597775481197702e-05, "loss": 0.6041, "step": 8903 }, { "epoch": 0.31722393430357876, "grad_norm": 1.7202214160183218, "learning_rate": 1.5976829703173063e-05, "loss": 0.5353, "step": 8904 }, { "epoch": 0.31725956143007283, "grad_norm": 1.1681473971536542, "learning_rate": 1.5975904514783442e-05, "loss": 0.4684, "step": 8905 }, { "epoch": 0.31729518855656696, "grad_norm": 1.17412097864642, "learning_rate": 1.5974979246820472e-05, "loss": 0.5434, "step": 8906 }, { "epoch": 0.3173308156830611, "grad_norm": 1.0938447186701183, "learning_rate": 1.597405389929648e-05, "loss": 0.453, "step": 8907 }, { "epoch": 0.31736644280955517, "grad_norm": 1.4443488502036868, "learning_rate": 1.5973128472223783e-05, "loss": 0.5604, "step": 8908 }, { "epoch": 0.3174020699360493, "grad_norm": 1.6256036628587391, "learning_rate": 1.597220296561471e-05, "loss": 0.5384, "step": 8909 }, { "epoch": 0.31743769706254343, "grad_norm": 1.4919567910150073, "learning_rate": 1.5971277379481573e-05, "loss": 0.6268, "step": 8910 }, { "epoch": 0.3174733241890375, "grad_norm": 1.6358244958786994, "learning_rate": 1.597035171383671e-05, "loss": 0.6707, "step": 8911 }, { "epoch": 0.31750895131553164, "grad_norm": 1.2683765773075693, "learning_rate": 1.5969425968692437e-05, "loss": 0.5064, "step": 8912 }, { "epoch": 0.31754457844202577, "grad_norm": 1.1381151124756264, "learning_rate": 1.5968500144061085e-05, "loss": 0.5029, "step": 8913 }, { "epoch": 0.31758020556851985, "grad_norm": 1.224544731164024, "learning_rate": 1.5967574239954983e-05, "loss": 0.494, "step": 8914 }, { "epoch": 0.317615832695014, "grad_norm": 1.1822223084876848, "learning_rate": 1.596664825638646e-05, "loss": 0.4989, "step": 8915 }, { "epoch": 0.3176514598215081, "grad_norm": 1.302839153824734, "learning_rate": 1.596572219336784e-05, "loss": 0.5234, "step": 8916 }, { "epoch": 0.3176870869480022, "grad_norm": 1.6205362197848956, "learning_rate": 1.5964796050911467e-05, "loss": 0.543, "step": 8917 }, { "epoch": 0.3177227140744963, "grad_norm": 1.0620956273499746, "learning_rate": 1.596386982902966e-05, "loss": 0.488, "step": 8918 }, { "epoch": 0.31775834120099045, "grad_norm": 1.0426707745298531, "learning_rate": 1.5962943527734757e-05, "loss": 0.4431, "step": 8919 }, { "epoch": 0.3177939683274845, "grad_norm": 1.1546950726722467, "learning_rate": 1.5962017147039096e-05, "loss": 0.4909, "step": 8920 }, { "epoch": 0.31782959545397865, "grad_norm": 1.7977668348737401, "learning_rate": 1.596109068695501e-05, "loss": 0.6501, "step": 8921 }, { "epoch": 0.3178652225804728, "grad_norm": 1.7956995860047966, "learning_rate": 1.5960164147494835e-05, "loss": 0.5535, "step": 8922 }, { "epoch": 0.31790084970696686, "grad_norm": 1.0521221426787257, "learning_rate": 1.5959237528670907e-05, "loss": 0.4899, "step": 8923 }, { "epoch": 0.317936476833461, "grad_norm": 1.402660583905485, "learning_rate": 1.5958310830495567e-05, "loss": 0.6042, "step": 8924 }, { "epoch": 0.3179721039599551, "grad_norm": 1.8866789729398852, "learning_rate": 1.5957384052981152e-05, "loss": 0.6074, "step": 8925 }, { "epoch": 0.3180077310864492, "grad_norm": 1.0996977384830375, "learning_rate": 1.5956457196140008e-05, "loss": 0.4653, "step": 8926 }, { "epoch": 0.31804335821294333, "grad_norm": 1.2800104715846072, "learning_rate": 1.595553025998447e-05, "loss": 0.5443, "step": 8927 }, { "epoch": 0.31807898533943746, "grad_norm": 1.355623924939554, "learning_rate": 1.595460324452688e-05, "loss": 0.5824, "step": 8928 }, { "epoch": 0.31811461246593153, "grad_norm": 1.1283037690677715, "learning_rate": 1.595367614977959e-05, "loss": 0.5002, "step": 8929 }, { "epoch": 0.31815023959242567, "grad_norm": 1.1421522517049667, "learning_rate": 1.5952748975754944e-05, "loss": 0.5339, "step": 8930 }, { "epoch": 0.3181858667189198, "grad_norm": 1.9629502251263147, "learning_rate": 1.595182172246528e-05, "loss": 0.5363, "step": 8931 }, { "epoch": 0.31822149384541387, "grad_norm": 1.064937254874696, "learning_rate": 1.5950894389922954e-05, "loss": 0.5326, "step": 8932 }, { "epoch": 0.318257120971908, "grad_norm": 1.1751365062540373, "learning_rate": 1.5949966978140303e-05, "loss": 0.4952, "step": 8933 }, { "epoch": 0.31829274809840213, "grad_norm": 1.5450343880124775, "learning_rate": 1.5949039487129687e-05, "loss": 0.6215, "step": 8934 }, { "epoch": 0.3183283752248962, "grad_norm": 1.4426579232982082, "learning_rate": 1.5948111916903456e-05, "loss": 0.6239, "step": 8935 }, { "epoch": 0.31836400235139034, "grad_norm": 1.0069252410977254, "learning_rate": 1.594718426747395e-05, "loss": 0.4771, "step": 8936 }, { "epoch": 0.31839962947788447, "grad_norm": 1.0890370993299896, "learning_rate": 1.594625653885353e-05, "loss": 0.4742, "step": 8937 }, { "epoch": 0.31843525660437855, "grad_norm": 1.714182982796359, "learning_rate": 1.594532873105455e-05, "loss": 0.572, "step": 8938 }, { "epoch": 0.3184708837308727, "grad_norm": 3.23318034358671, "learning_rate": 1.5944400844089365e-05, "loss": 0.5823, "step": 8939 }, { "epoch": 0.3185065108573668, "grad_norm": 4.611071609310565, "learning_rate": 1.5943472877970325e-05, "loss": 0.6152, "step": 8940 }, { "epoch": 0.3185421379838609, "grad_norm": 1.0659563339532037, "learning_rate": 1.594254483270979e-05, "loss": 0.4605, "step": 8941 }, { "epoch": 0.318577765110355, "grad_norm": 1.0705660987360053, "learning_rate": 1.5941616708320115e-05, "loss": 0.4751, "step": 8942 }, { "epoch": 0.31861339223684915, "grad_norm": 1.1019121504210831, "learning_rate": 1.5940688504813664e-05, "loss": 0.4787, "step": 8943 }, { "epoch": 0.3186490193633432, "grad_norm": 1.5739920680720625, "learning_rate": 1.593976022220279e-05, "loss": 0.6353, "step": 8944 }, { "epoch": 0.31868464648983735, "grad_norm": 1.0483546807175248, "learning_rate": 1.593883186049986e-05, "loss": 0.4751, "step": 8945 }, { "epoch": 0.3187202736163315, "grad_norm": 1.5339528397594564, "learning_rate": 1.593790341971723e-05, "loss": 0.5413, "step": 8946 }, { "epoch": 0.31875590074282556, "grad_norm": 1.6696684472807894, "learning_rate": 1.5936974899867267e-05, "loss": 0.5801, "step": 8947 }, { "epoch": 0.3187915278693197, "grad_norm": 1.2188535831329013, "learning_rate": 1.5936046300962336e-05, "loss": 0.4564, "step": 8948 }, { "epoch": 0.3188271549958138, "grad_norm": 1.137111822829522, "learning_rate": 1.5935117623014795e-05, "loss": 0.5536, "step": 8949 }, { "epoch": 0.31886278212230795, "grad_norm": 1.0228525866784248, "learning_rate": 1.5934188866037017e-05, "loss": 0.483, "step": 8950 }, { "epoch": 0.31889840924880203, "grad_norm": 1.6163918757996505, "learning_rate": 1.5933260030041364e-05, "loss": 0.6096, "step": 8951 }, { "epoch": 0.31893403637529616, "grad_norm": 1.266532832411547, "learning_rate": 1.5932331115040208e-05, "loss": 0.5148, "step": 8952 }, { "epoch": 0.3189696635017903, "grad_norm": 1.3729306243938608, "learning_rate": 1.5931402121045918e-05, "loss": 0.6023, "step": 8953 }, { "epoch": 0.31900529062828437, "grad_norm": 1.9282528438686193, "learning_rate": 1.5930473048070863e-05, "loss": 0.6086, "step": 8954 }, { "epoch": 0.3190409177547785, "grad_norm": 1.234231507848216, "learning_rate": 1.5929543896127412e-05, "loss": 0.4968, "step": 8955 }, { "epoch": 0.31907654488127263, "grad_norm": 1.690548244825015, "learning_rate": 1.592861466522794e-05, "loss": 0.6751, "step": 8956 }, { "epoch": 0.3191121720077667, "grad_norm": 1.0357336413806666, "learning_rate": 1.5927685355384822e-05, "loss": 0.4432, "step": 8957 }, { "epoch": 0.31914779913426083, "grad_norm": 1.7820817869658763, "learning_rate": 1.5926755966610426e-05, "loss": 0.6389, "step": 8958 }, { "epoch": 0.31918342626075497, "grad_norm": 1.4481516252686024, "learning_rate": 1.5925826498917138e-05, "loss": 0.531, "step": 8959 }, { "epoch": 0.31921905338724904, "grad_norm": 1.2233360919317813, "learning_rate": 1.5924896952317323e-05, "loss": 0.4973, "step": 8960 }, { "epoch": 0.31925468051374317, "grad_norm": 1.7676094383150105, "learning_rate": 1.592396732682337e-05, "loss": 0.6013, "step": 8961 }, { "epoch": 0.3192903076402373, "grad_norm": 4.928981925083446, "learning_rate": 1.5923037622447644e-05, "loss": 0.694, "step": 8962 }, { "epoch": 0.3193259347667314, "grad_norm": 1.1651231985408184, "learning_rate": 1.5922107839202535e-05, "loss": 0.54, "step": 8963 }, { "epoch": 0.3193615618932255, "grad_norm": 1.1271431095241953, "learning_rate": 1.5921177977100423e-05, "loss": 0.4907, "step": 8964 }, { "epoch": 0.31939718901971964, "grad_norm": 1.1611885948412894, "learning_rate": 1.5920248036153685e-05, "loss": 0.5289, "step": 8965 }, { "epoch": 0.3194328161462137, "grad_norm": 1.0441676444227714, "learning_rate": 1.5919318016374707e-05, "loss": 0.4901, "step": 8966 }, { "epoch": 0.31946844327270785, "grad_norm": 1.6475540847620136, "learning_rate": 1.5918387917775873e-05, "loss": 0.5426, "step": 8967 }, { "epoch": 0.319504070399202, "grad_norm": 2.480097197953562, "learning_rate": 1.5917457740369565e-05, "loss": 0.5829, "step": 8968 }, { "epoch": 0.31953969752569605, "grad_norm": 1.464660820754432, "learning_rate": 1.5916527484168173e-05, "loss": 0.5835, "step": 8969 }, { "epoch": 0.3195753246521902, "grad_norm": 1.330180113459892, "learning_rate": 1.5915597149184083e-05, "loss": 0.5379, "step": 8970 }, { "epoch": 0.3196109517786843, "grad_norm": 1.659784725099722, "learning_rate": 1.591466673542968e-05, "loss": 0.5932, "step": 8971 }, { "epoch": 0.3196465789051784, "grad_norm": 1.1185780610908354, "learning_rate": 1.591373624291736e-05, "loss": 0.5309, "step": 8972 }, { "epoch": 0.3196822060316725, "grad_norm": 1.6763603015728725, "learning_rate": 1.59128056716595e-05, "loss": 0.5435, "step": 8973 }, { "epoch": 0.31971783315816665, "grad_norm": 1.1143338205743265, "learning_rate": 1.5911875021668506e-05, "loss": 0.4874, "step": 8974 }, { "epoch": 0.31975346028466073, "grad_norm": 1.1636841908379687, "learning_rate": 1.5910944292956764e-05, "loss": 0.5376, "step": 8975 }, { "epoch": 0.31978908741115486, "grad_norm": 1.5681919487910114, "learning_rate": 1.5910013485536664e-05, "loss": 0.5545, "step": 8976 }, { "epoch": 0.319824714537649, "grad_norm": 1.4984944481722775, "learning_rate": 1.5909082599420606e-05, "loss": 0.575, "step": 8977 }, { "epoch": 0.31986034166414307, "grad_norm": 1.2602339328652579, "learning_rate": 1.5908151634620985e-05, "loss": 0.5177, "step": 8978 }, { "epoch": 0.3198959687906372, "grad_norm": 1.6246843696892252, "learning_rate": 1.5907220591150186e-05, "loss": 0.6183, "step": 8979 }, { "epoch": 0.31993159591713133, "grad_norm": 1.085089940939789, "learning_rate": 1.5906289469020626e-05, "loss": 0.4496, "step": 8980 }, { "epoch": 0.3199672230436254, "grad_norm": 2.25671990616355, "learning_rate": 1.590535826824469e-05, "loss": 0.5549, "step": 8981 }, { "epoch": 0.32000285017011953, "grad_norm": 1.0469285366997552, "learning_rate": 1.590442698883478e-05, "loss": 0.4724, "step": 8982 }, { "epoch": 0.32003847729661367, "grad_norm": 2.0027414872557143, "learning_rate": 1.5903495630803302e-05, "loss": 0.6229, "step": 8983 }, { "epoch": 0.32007410442310774, "grad_norm": 1.5130288352632346, "learning_rate": 1.590256419416265e-05, "loss": 0.6806, "step": 8984 }, { "epoch": 0.3201097315496019, "grad_norm": 1.227627601374459, "learning_rate": 1.590163267892523e-05, "loss": 0.5351, "step": 8985 }, { "epoch": 0.320145358676096, "grad_norm": 1.1277971384434833, "learning_rate": 1.5900701085103444e-05, "loss": 0.4908, "step": 8986 }, { "epoch": 0.3201809858025901, "grad_norm": 1.1311464233713298, "learning_rate": 1.5899769412709704e-05, "loss": 0.5367, "step": 8987 }, { "epoch": 0.3202166129290842, "grad_norm": 1.0894568123971193, "learning_rate": 1.5898837661756405e-05, "loss": 0.4677, "step": 8988 }, { "epoch": 0.32025224005557834, "grad_norm": 1.5342451648857227, "learning_rate": 1.5897905832255962e-05, "loss": 0.5839, "step": 8989 }, { "epoch": 0.3202878671820724, "grad_norm": 1.2737994366039518, "learning_rate": 1.589697392422078e-05, "loss": 0.5138, "step": 8990 }, { "epoch": 0.32032349430856655, "grad_norm": 1.1839247916697577, "learning_rate": 1.589604193766327e-05, "loss": 0.5093, "step": 8991 }, { "epoch": 0.3203591214350607, "grad_norm": 0.9757656685423637, "learning_rate": 1.5895109872595837e-05, "loss": 0.5026, "step": 8992 }, { "epoch": 0.32039474856155475, "grad_norm": 1.1515921836034848, "learning_rate": 1.58941777290309e-05, "loss": 0.4651, "step": 8993 }, { "epoch": 0.3204303756880489, "grad_norm": 2.093118476938984, "learning_rate": 1.5893245506980866e-05, "loss": 0.5814, "step": 8994 }, { "epoch": 0.320466002814543, "grad_norm": 1.1893740918563378, "learning_rate": 1.5892313206458148e-05, "loss": 0.4922, "step": 8995 }, { "epoch": 0.3205016299410371, "grad_norm": 1.0670714724481063, "learning_rate": 1.589138082747516e-05, "loss": 0.5228, "step": 8996 }, { "epoch": 0.3205372570675312, "grad_norm": 1.60912809123008, "learning_rate": 1.5890448370044318e-05, "loss": 0.581, "step": 8997 }, { "epoch": 0.32057288419402535, "grad_norm": 1.4936433150527162, "learning_rate": 1.588951583417804e-05, "loss": 0.5942, "step": 8998 }, { "epoch": 0.32060851132051943, "grad_norm": 1.0970113026267978, "learning_rate": 1.5888583219888743e-05, "loss": 0.4946, "step": 8999 }, { "epoch": 0.32064413844701356, "grad_norm": 1.3361044006629663, "learning_rate": 1.5887650527188844e-05, "loss": 0.5246, "step": 9000 }, { "epoch": 0.3206797655735077, "grad_norm": 1.7147897362198996, "learning_rate": 1.5886717756090765e-05, "loss": 0.5008, "step": 9001 }, { "epoch": 0.32071539270000177, "grad_norm": 1.6005360781166427, "learning_rate": 1.588578490660692e-05, "loss": 0.6012, "step": 9002 }, { "epoch": 0.3207510198264959, "grad_norm": 1.2485699457182822, "learning_rate": 1.588485197874974e-05, "loss": 0.5351, "step": 9003 }, { "epoch": 0.32078664695299003, "grad_norm": 1.7121065712095827, "learning_rate": 1.5883918972531637e-05, "loss": 0.5756, "step": 9004 }, { "epoch": 0.3208222740794841, "grad_norm": 1.4769027847247702, "learning_rate": 1.5882985887965045e-05, "loss": 0.5811, "step": 9005 }, { "epoch": 0.32085790120597824, "grad_norm": 1.2554623447816549, "learning_rate": 1.5882052725062382e-05, "loss": 0.5539, "step": 9006 }, { "epoch": 0.32089352833247237, "grad_norm": 1.0611742187684619, "learning_rate": 1.5881119483836073e-05, "loss": 0.4504, "step": 9007 }, { "epoch": 0.32092915545896644, "grad_norm": 1.1707690282142915, "learning_rate": 1.5880186164298553e-05, "loss": 0.4656, "step": 9008 }, { "epoch": 0.3209647825854606, "grad_norm": 1.0504243892378617, "learning_rate": 1.5879252766462243e-05, "loss": 0.4861, "step": 9009 }, { "epoch": 0.3210004097119547, "grad_norm": 2.026355988481531, "learning_rate": 1.587831929033957e-05, "loss": 0.6281, "step": 9010 }, { "epoch": 0.3210360368384488, "grad_norm": 1.1830352840150569, "learning_rate": 1.5877385735942967e-05, "loss": 0.4527, "step": 9011 }, { "epoch": 0.3210716639649429, "grad_norm": 1.1358873342121427, "learning_rate": 1.5876452103284868e-05, "loss": 0.4857, "step": 9012 }, { "epoch": 0.32110729109143704, "grad_norm": 1.9219597548944327, "learning_rate": 1.58755183923777e-05, "loss": 0.6471, "step": 9013 }, { "epoch": 0.3211429182179311, "grad_norm": 1.9036221983734736, "learning_rate": 1.58745846032339e-05, "loss": 0.6584, "step": 9014 }, { "epoch": 0.32117854534442525, "grad_norm": 1.1947138248575537, "learning_rate": 1.5873650735865894e-05, "loss": 0.492, "step": 9015 }, { "epoch": 0.3212141724709194, "grad_norm": 1.0405235520905867, "learning_rate": 1.587271679028613e-05, "loss": 0.4539, "step": 9016 }, { "epoch": 0.32124979959741345, "grad_norm": 1.5102463146933924, "learning_rate": 1.5871782766507033e-05, "loss": 0.5168, "step": 9017 }, { "epoch": 0.3212854267239076, "grad_norm": 1.656051792251063, "learning_rate": 1.5870848664541046e-05, "loss": 0.6352, "step": 9018 }, { "epoch": 0.3213210538504017, "grad_norm": 1.7600392747453508, "learning_rate": 1.5869914484400606e-05, "loss": 0.612, "step": 9019 }, { "epoch": 0.3213566809768958, "grad_norm": 1.0343419860555894, "learning_rate": 1.586898022609815e-05, "loss": 0.4785, "step": 9020 }, { "epoch": 0.3213923081033899, "grad_norm": 3.001698161271786, "learning_rate": 1.586804588964612e-05, "loss": 0.499, "step": 9021 }, { "epoch": 0.32142793522988405, "grad_norm": 1.0495290229621963, "learning_rate": 1.5867111475056957e-05, "loss": 0.4978, "step": 9022 }, { "epoch": 0.32146356235637813, "grad_norm": 1.8417818213730655, "learning_rate": 1.5866176982343108e-05, "loss": 0.5561, "step": 9023 }, { "epoch": 0.32149918948287226, "grad_norm": 1.2404116617435876, "learning_rate": 1.586524241151701e-05, "loss": 0.4928, "step": 9024 }, { "epoch": 0.3215348166093664, "grad_norm": 1.29335913788287, "learning_rate": 1.586430776259111e-05, "loss": 0.4708, "step": 9025 }, { "epoch": 0.32157044373586047, "grad_norm": 1.0600867611315563, "learning_rate": 1.586337303557785e-05, "loss": 0.4835, "step": 9026 }, { "epoch": 0.3216060708623546, "grad_norm": 1.6073825925971685, "learning_rate": 1.5862438230489682e-05, "loss": 0.5926, "step": 9027 }, { "epoch": 0.32164169798884873, "grad_norm": 1.0543533948727135, "learning_rate": 1.5861503347339052e-05, "loss": 0.5187, "step": 9028 }, { "epoch": 0.3216773251153428, "grad_norm": 1.3811681385639702, "learning_rate": 1.5860568386138408e-05, "loss": 0.5667, "step": 9029 }, { "epoch": 0.32171295224183694, "grad_norm": 2.916735051840316, "learning_rate": 1.5859633346900202e-05, "loss": 0.4403, "step": 9030 }, { "epoch": 0.32174857936833107, "grad_norm": 1.6478763321230554, "learning_rate": 1.585869822963688e-05, "loss": 0.5531, "step": 9031 }, { "epoch": 0.32178420649482514, "grad_norm": 2.903448394930918, "learning_rate": 1.5857763034360892e-05, "loss": 0.5307, "step": 9032 }, { "epoch": 0.3218198336213193, "grad_norm": 1.4588057707559905, "learning_rate": 1.5856827761084698e-05, "loss": 0.6101, "step": 9033 }, { "epoch": 0.3218554607478134, "grad_norm": 1.057618442544038, "learning_rate": 1.585589240982075e-05, "loss": 0.5076, "step": 9034 }, { "epoch": 0.3218910878743075, "grad_norm": 1.8113889573953779, "learning_rate": 1.58549569805815e-05, "loss": 0.6756, "step": 9035 }, { "epoch": 0.3219267150008016, "grad_norm": 1.5987532644798053, "learning_rate": 1.5854021473379408e-05, "loss": 0.5802, "step": 9036 }, { "epoch": 0.32196234212729574, "grad_norm": 1.0795606653217977, "learning_rate": 1.5853085888226926e-05, "loss": 0.4838, "step": 9037 }, { "epoch": 0.3219979692537898, "grad_norm": 1.6605202979447884, "learning_rate": 1.585215022513652e-05, "loss": 0.6735, "step": 9038 }, { "epoch": 0.32203359638028395, "grad_norm": 1.1253487077913142, "learning_rate": 1.5851214484120637e-05, "loss": 0.5264, "step": 9039 }, { "epoch": 0.3220692235067781, "grad_norm": 1.8053580935045201, "learning_rate": 1.585027866519174e-05, "loss": 0.6337, "step": 9040 }, { "epoch": 0.32210485063327216, "grad_norm": 1.6192554257600917, "learning_rate": 1.58493427683623e-05, "loss": 0.5595, "step": 9041 }, { "epoch": 0.3221404777597663, "grad_norm": 1.0605367909456522, "learning_rate": 1.584840679364477e-05, "loss": 0.5188, "step": 9042 }, { "epoch": 0.3221761048862604, "grad_norm": 2.8264559581415276, "learning_rate": 1.5847470741051617e-05, "loss": 0.637, "step": 9043 }, { "epoch": 0.3222117320127545, "grad_norm": 1.2668421331346749, "learning_rate": 1.5846534610595306e-05, "loss": 0.5008, "step": 9044 }, { "epoch": 0.3222473591392486, "grad_norm": 1.4643980773449452, "learning_rate": 1.5845598402288297e-05, "loss": 0.5384, "step": 9045 }, { "epoch": 0.32228298626574275, "grad_norm": 1.315312877510731, "learning_rate": 1.5844662116143057e-05, "loss": 0.4891, "step": 9046 }, { "epoch": 0.32231861339223683, "grad_norm": 1.8067282964589253, "learning_rate": 1.584372575217206e-05, "loss": 0.5925, "step": 9047 }, { "epoch": 0.32235424051873096, "grad_norm": 1.092911522940285, "learning_rate": 1.584278931038777e-05, "loss": 0.5171, "step": 9048 }, { "epoch": 0.3223898676452251, "grad_norm": 1.1990459501098762, "learning_rate": 1.5841852790802655e-05, "loss": 0.5043, "step": 9049 }, { "epoch": 0.32242549477171917, "grad_norm": 1.1148816442736393, "learning_rate": 1.584091619342919e-05, "loss": 0.5065, "step": 9050 }, { "epoch": 0.3224611218982133, "grad_norm": 1.1768724598520768, "learning_rate": 1.5839979518279836e-05, "loss": 0.5165, "step": 9051 }, { "epoch": 0.32249674902470743, "grad_norm": 1.7738768952898198, "learning_rate": 1.583904276536708e-05, "loss": 0.5636, "step": 9052 }, { "epoch": 0.3225323761512015, "grad_norm": 2.444455176786311, "learning_rate": 1.5838105934703385e-05, "loss": 0.6641, "step": 9053 }, { "epoch": 0.32256800327769564, "grad_norm": 1.210870925666303, "learning_rate": 1.583716902630123e-05, "loss": 0.483, "step": 9054 }, { "epoch": 0.32260363040418977, "grad_norm": 2.5860936344706804, "learning_rate": 1.583623204017309e-05, "loss": 0.6083, "step": 9055 }, { "epoch": 0.32263925753068384, "grad_norm": 1.4921597015893175, "learning_rate": 1.583529497633144e-05, "loss": 0.6286, "step": 9056 }, { "epoch": 0.322674884657178, "grad_norm": 1.7621740589570447, "learning_rate": 1.583435783478876e-05, "loss": 0.6021, "step": 9057 }, { "epoch": 0.3227105117836721, "grad_norm": 1.4398343922410874, "learning_rate": 1.5833420615557523e-05, "loss": 0.629, "step": 9058 }, { "epoch": 0.3227461389101662, "grad_norm": 1.6621844979899945, "learning_rate": 1.583248331865022e-05, "loss": 0.6144, "step": 9059 }, { "epoch": 0.3227817660366603, "grad_norm": 1.6178480188980149, "learning_rate": 1.583154594407932e-05, "loss": 0.6375, "step": 9060 }, { "epoch": 0.32281739316315444, "grad_norm": 1.1004716684445277, "learning_rate": 1.583060849185731e-05, "loss": 0.4762, "step": 9061 }, { "epoch": 0.3228530202896485, "grad_norm": 1.1764898153662893, "learning_rate": 1.5829670961996673e-05, "loss": 0.481, "step": 9062 }, { "epoch": 0.32288864741614265, "grad_norm": 1.1901534882736717, "learning_rate": 1.5828733354509895e-05, "loss": 0.4941, "step": 9063 }, { "epoch": 0.3229242745426368, "grad_norm": 1.2527641729178247, "learning_rate": 1.5827795669409458e-05, "loss": 0.5023, "step": 9064 }, { "epoch": 0.32295990166913086, "grad_norm": 1.9138308392845254, "learning_rate": 1.5826857906707846e-05, "loss": 0.5804, "step": 9065 }, { "epoch": 0.322995528795625, "grad_norm": 1.0917063338932864, "learning_rate": 1.582592006641755e-05, "loss": 0.4673, "step": 9066 }, { "epoch": 0.3230311559221191, "grad_norm": 1.1367929538458394, "learning_rate": 1.5824982148551055e-05, "loss": 0.5023, "step": 9067 }, { "epoch": 0.3230667830486132, "grad_norm": 1.4462480299063358, "learning_rate": 1.5824044153120853e-05, "loss": 0.5172, "step": 9068 }, { "epoch": 0.3231024101751073, "grad_norm": 2.370089330505728, "learning_rate": 1.582310608013943e-05, "loss": 0.554, "step": 9069 }, { "epoch": 0.32313803730160146, "grad_norm": 1.9108793906818031, "learning_rate": 1.5822167929619282e-05, "loss": 0.6804, "step": 9070 }, { "epoch": 0.32317366442809553, "grad_norm": 1.226390307588214, "learning_rate": 1.5821229701572897e-05, "loss": 0.476, "step": 9071 }, { "epoch": 0.32320929155458966, "grad_norm": 1.2883523062325104, "learning_rate": 1.582029139601277e-05, "loss": 0.5358, "step": 9072 }, { "epoch": 0.3232449186810838, "grad_norm": 2.1039996374762304, "learning_rate": 1.5819353012951392e-05, "loss": 0.5569, "step": 9073 }, { "epoch": 0.32328054580757787, "grad_norm": 1.0785987673606316, "learning_rate": 1.5818414552401264e-05, "loss": 0.4782, "step": 9074 }, { "epoch": 0.323316172934072, "grad_norm": 1.8020735651120046, "learning_rate": 1.581747601437488e-05, "loss": 0.6234, "step": 9075 }, { "epoch": 0.32335180006056613, "grad_norm": 1.692537287815624, "learning_rate": 1.5816537398884737e-05, "loss": 0.6094, "step": 9076 }, { "epoch": 0.3233874271870602, "grad_norm": 1.1474734930940307, "learning_rate": 1.5815598705943333e-05, "loss": 0.5627, "step": 9077 }, { "epoch": 0.32342305431355434, "grad_norm": 2.3051425649098776, "learning_rate": 1.5814659935563165e-05, "loss": 0.6585, "step": 9078 }, { "epoch": 0.32345868144004847, "grad_norm": 1.568007072217871, "learning_rate": 1.5813721087756735e-05, "loss": 0.6105, "step": 9079 }, { "epoch": 0.32349430856654254, "grad_norm": 1.1276580376328489, "learning_rate": 1.581278216253655e-05, "loss": 0.4965, "step": 9080 }, { "epoch": 0.3235299356930367, "grad_norm": 1.8929319049644384, "learning_rate": 1.5811843159915105e-05, "loss": 0.6434, "step": 9081 }, { "epoch": 0.3235655628195308, "grad_norm": 1.6491278146643518, "learning_rate": 1.5810904079904906e-05, "loss": 0.4986, "step": 9082 }, { "epoch": 0.3236011899460249, "grad_norm": 1.5447573592688033, "learning_rate": 1.5809964922518458e-05, "loss": 0.5428, "step": 9083 }, { "epoch": 0.323636817072519, "grad_norm": 2.5083714789488534, "learning_rate": 1.5809025687768265e-05, "loss": 0.587, "step": 9084 }, { "epoch": 0.32367244419901314, "grad_norm": 1.0792850657658555, "learning_rate": 1.5808086375666834e-05, "loss": 0.4954, "step": 9085 }, { "epoch": 0.3237080713255072, "grad_norm": 1.228164709743279, "learning_rate": 1.5807146986226677e-05, "loss": 0.4882, "step": 9086 }, { "epoch": 0.32374369845200135, "grad_norm": 1.2351710875710618, "learning_rate": 1.5806207519460296e-05, "loss": 0.539, "step": 9087 }, { "epoch": 0.3237793255784955, "grad_norm": 1.4834791251866264, "learning_rate": 1.5805267975380206e-05, "loss": 0.5197, "step": 9088 }, { "epoch": 0.32381495270498956, "grad_norm": 1.6456547645849078, "learning_rate": 1.580432835399891e-05, "loss": 0.6039, "step": 9089 }, { "epoch": 0.3238505798314837, "grad_norm": 1.5992068604919758, "learning_rate": 1.580338865532893e-05, "loss": 0.5882, "step": 9090 }, { "epoch": 0.3238862069579778, "grad_norm": 1.3896268063732817, "learning_rate": 1.5802448879382773e-05, "loss": 0.5941, "step": 9091 }, { "epoch": 0.3239218340844719, "grad_norm": 1.4431420718354437, "learning_rate": 1.580150902617295e-05, "loss": 0.5951, "step": 9092 }, { "epoch": 0.323957461210966, "grad_norm": 1.1141523848847261, "learning_rate": 1.5800569095711983e-05, "loss": 0.4845, "step": 9093 }, { "epoch": 0.32399308833746016, "grad_norm": 1.1364681592831214, "learning_rate": 1.5799629088012384e-05, "loss": 0.4815, "step": 9094 }, { "epoch": 0.32402871546395423, "grad_norm": 1.4674561838187516, "learning_rate": 1.5798689003086668e-05, "loss": 0.5944, "step": 9095 }, { "epoch": 0.32406434259044836, "grad_norm": 1.6338996891766198, "learning_rate": 1.579774884094736e-05, "loss": 0.5623, "step": 9096 }, { "epoch": 0.3240999697169425, "grad_norm": 1.3125992738095988, "learning_rate": 1.5796808601606966e-05, "loss": 0.5442, "step": 9097 }, { "epoch": 0.32413559684343657, "grad_norm": 1.2704690771369878, "learning_rate": 1.579586828507802e-05, "loss": 0.5363, "step": 9098 }, { "epoch": 0.3241712239699307, "grad_norm": 1.8350076537262803, "learning_rate": 1.5794927891373032e-05, "loss": 0.5751, "step": 9099 }, { "epoch": 0.32420685109642483, "grad_norm": 1.5019417741521668, "learning_rate": 1.5793987420504532e-05, "loss": 0.5717, "step": 9100 }, { "epoch": 0.3242424782229189, "grad_norm": 1.8320628787968083, "learning_rate": 1.5793046872485036e-05, "loss": 0.5588, "step": 9101 }, { "epoch": 0.32427810534941304, "grad_norm": 1.36443709202687, "learning_rate": 1.579210624732708e-05, "loss": 0.5092, "step": 9102 }, { "epoch": 0.32431373247590717, "grad_norm": 1.6471369394221123, "learning_rate": 1.5791165545043174e-05, "loss": 0.5981, "step": 9103 }, { "epoch": 0.32434935960240124, "grad_norm": 1.3425548839138597, "learning_rate": 1.579022476564585e-05, "loss": 0.5672, "step": 9104 }, { "epoch": 0.3243849867288954, "grad_norm": 1.841597288659343, "learning_rate": 1.578928390914764e-05, "loss": 0.5598, "step": 9105 }, { "epoch": 0.3244206138553895, "grad_norm": 1.0947459571813867, "learning_rate": 1.5788342975561067e-05, "loss": 0.4449, "step": 9106 }, { "epoch": 0.3244562409818836, "grad_norm": 2.3807899795133003, "learning_rate": 1.578740196489866e-05, "loss": 0.5897, "step": 9107 }, { "epoch": 0.3244918681083777, "grad_norm": 1.1058307854957545, "learning_rate": 1.5786460877172954e-05, "loss": 0.4835, "step": 9108 }, { "epoch": 0.32452749523487184, "grad_norm": 1.3190183535939173, "learning_rate": 1.5785519712396473e-05, "loss": 0.5081, "step": 9109 }, { "epoch": 0.3245631223613659, "grad_norm": 1.4592254383969931, "learning_rate": 1.5784578470581755e-05, "loss": 0.628, "step": 9110 }, { "epoch": 0.32459874948786005, "grad_norm": 1.244955746393256, "learning_rate": 1.5783637151741335e-05, "loss": 0.5237, "step": 9111 }, { "epoch": 0.3246343766143542, "grad_norm": 1.1900925319756874, "learning_rate": 1.5782695755887737e-05, "loss": 0.536, "step": 9112 }, { "epoch": 0.32467000374084826, "grad_norm": 1.0930736732870028, "learning_rate": 1.5781754283033504e-05, "loss": 0.4521, "step": 9113 }, { "epoch": 0.3247056308673424, "grad_norm": 1.086435873980275, "learning_rate": 1.5780812733191176e-05, "loss": 0.5111, "step": 9114 }, { "epoch": 0.3247412579938365, "grad_norm": 1.1514730030702303, "learning_rate": 1.5779871106373283e-05, "loss": 0.5194, "step": 9115 }, { "epoch": 0.3247768851203306, "grad_norm": 1.7734927927397708, "learning_rate": 1.577892940259237e-05, "loss": 0.5638, "step": 9116 }, { "epoch": 0.3248125122468247, "grad_norm": 1.3287268890130597, "learning_rate": 1.577798762186097e-05, "loss": 0.4309, "step": 9117 }, { "epoch": 0.32484813937331886, "grad_norm": 1.162207628523857, "learning_rate": 1.5777045764191625e-05, "loss": 0.4749, "step": 9118 }, { "epoch": 0.32488376649981293, "grad_norm": 1.16300378431212, "learning_rate": 1.577610382959688e-05, "loss": 0.5084, "step": 9119 }, { "epoch": 0.32491939362630706, "grad_norm": 1.515582435407444, "learning_rate": 1.5775161818089273e-05, "loss": 0.625, "step": 9120 }, { "epoch": 0.3249550207528012, "grad_norm": 3.830756761038001, "learning_rate": 1.577421972968135e-05, "loss": 0.5809, "step": 9121 }, { "epoch": 0.32499064787929527, "grad_norm": 1.068632527035362, "learning_rate": 1.577327756438566e-05, "loss": 0.483, "step": 9122 }, { "epoch": 0.3250262750057894, "grad_norm": 1.8916757835345215, "learning_rate": 1.577233532221474e-05, "loss": 0.611, "step": 9123 }, { "epoch": 0.32506190213228353, "grad_norm": 1.1312686734488526, "learning_rate": 1.5771393003181138e-05, "loss": 0.4955, "step": 9124 }, { "epoch": 0.3250975292587776, "grad_norm": 1.2536814408725427, "learning_rate": 1.5770450607297406e-05, "loss": 0.5323, "step": 9125 }, { "epoch": 0.32513315638527174, "grad_norm": 1.1009045636020969, "learning_rate": 1.5769508134576094e-05, "loss": 0.4817, "step": 9126 }, { "epoch": 0.32516878351176587, "grad_norm": 1.433258437322303, "learning_rate": 1.5768565585029747e-05, "loss": 0.5635, "step": 9127 }, { "epoch": 0.32520441063825994, "grad_norm": 1.7908668207495393, "learning_rate": 1.5767622958670915e-05, "loss": 0.6303, "step": 9128 }, { "epoch": 0.3252400377647541, "grad_norm": 1.3135529085006592, "learning_rate": 1.5766680255512155e-05, "loss": 0.4597, "step": 9129 }, { "epoch": 0.3252756648912482, "grad_norm": 1.6461648276916068, "learning_rate": 1.5765737475566013e-05, "loss": 0.5857, "step": 9130 }, { "epoch": 0.3253112920177423, "grad_norm": 1.0522029200784002, "learning_rate": 1.5764794618845055e-05, "loss": 0.4755, "step": 9131 }, { "epoch": 0.3253469191442364, "grad_norm": 1.9996949536436703, "learning_rate": 1.5763851685361818e-05, "loss": 0.5729, "step": 9132 }, { "epoch": 0.32538254627073054, "grad_norm": 1.0420093235579417, "learning_rate": 1.576290867512887e-05, "loss": 0.4821, "step": 9133 }, { "epoch": 0.3254181733972246, "grad_norm": 1.4093081117922561, "learning_rate": 1.5761965588158767e-05, "loss": 0.5017, "step": 9134 }, { "epoch": 0.32545380052371875, "grad_norm": 1.0586308394580466, "learning_rate": 1.5761022424464063e-05, "loss": 0.4498, "step": 9135 }, { "epoch": 0.3254894276502129, "grad_norm": 1.0676230413463752, "learning_rate": 1.5760079184057316e-05, "loss": 0.4761, "step": 9136 }, { "epoch": 0.325525054776707, "grad_norm": 1.2388736402061282, "learning_rate": 1.5759135866951095e-05, "loss": 0.4807, "step": 9137 }, { "epoch": 0.3255606819032011, "grad_norm": 1.8254852298763333, "learning_rate": 1.575819247315795e-05, "loss": 0.5452, "step": 9138 }, { "epoch": 0.3255963090296952, "grad_norm": 2.148320755250653, "learning_rate": 1.575724900269045e-05, "loss": 0.5473, "step": 9139 }, { "epoch": 0.32563193615618935, "grad_norm": 1.7124785794475437, "learning_rate": 1.5756305455561154e-05, "loss": 0.6556, "step": 9140 }, { "epoch": 0.3256675632826834, "grad_norm": 1.6536153562396811, "learning_rate": 1.5755361831782626e-05, "loss": 0.561, "step": 9141 }, { "epoch": 0.32570319040917756, "grad_norm": 1.324832189647295, "learning_rate": 1.5754418131367436e-05, "loss": 0.5608, "step": 9142 }, { "epoch": 0.3257388175356717, "grad_norm": 1.2967808317467657, "learning_rate": 1.5753474354328142e-05, "loss": 0.5142, "step": 9143 }, { "epoch": 0.32577444466216576, "grad_norm": 1.36582835194547, "learning_rate": 1.575253050067732e-05, "loss": 0.5769, "step": 9144 }, { "epoch": 0.3258100717886599, "grad_norm": 1.7112832995900187, "learning_rate": 1.575158657042753e-05, "loss": 0.5867, "step": 9145 }, { "epoch": 0.325845698915154, "grad_norm": 1.4100734124936118, "learning_rate": 1.575064256359135e-05, "loss": 0.4935, "step": 9146 }, { "epoch": 0.3258813260416481, "grad_norm": 1.4983998177666276, "learning_rate": 1.5749698480181335e-05, "loss": 0.4948, "step": 9147 }, { "epoch": 0.32591695316814223, "grad_norm": 1.6588920653195758, "learning_rate": 1.5748754320210074e-05, "loss": 0.529, "step": 9148 }, { "epoch": 0.32595258029463636, "grad_norm": 1.6446709567277533, "learning_rate": 1.574781008369013e-05, "loss": 0.5801, "step": 9149 }, { "epoch": 0.32598820742113044, "grad_norm": 1.2304051258033575, "learning_rate": 1.5746865770634073e-05, "loss": 0.4712, "step": 9150 }, { "epoch": 0.32602383454762457, "grad_norm": 1.1570742019654476, "learning_rate": 1.5745921381054487e-05, "loss": 0.4794, "step": 9151 }, { "epoch": 0.3260594616741187, "grad_norm": 2.607056572828962, "learning_rate": 1.574497691496394e-05, "loss": 0.5385, "step": 9152 }, { "epoch": 0.3260950888006128, "grad_norm": 1.5049322111160814, "learning_rate": 1.5744032372375007e-05, "loss": 0.5265, "step": 9153 }, { "epoch": 0.3261307159271069, "grad_norm": 1.6189052548217995, "learning_rate": 1.5743087753300274e-05, "loss": 0.6054, "step": 9154 }, { "epoch": 0.32616634305360104, "grad_norm": 1.972183502124887, "learning_rate": 1.574214305775231e-05, "loss": 0.6076, "step": 9155 }, { "epoch": 0.3262019701800951, "grad_norm": 1.2832435271060314, "learning_rate": 1.5741198285743694e-05, "loss": 0.518, "step": 9156 }, { "epoch": 0.32623759730658924, "grad_norm": 1.1255355026879938, "learning_rate": 1.5740253437287014e-05, "loss": 0.4477, "step": 9157 }, { "epoch": 0.3262732244330834, "grad_norm": 1.147158350676268, "learning_rate": 1.5739308512394847e-05, "loss": 0.4395, "step": 9158 }, { "epoch": 0.32630885155957745, "grad_norm": 1.5878999945318495, "learning_rate": 1.5738363511079776e-05, "loss": 0.5946, "step": 9159 }, { "epoch": 0.3263444786860716, "grad_norm": 1.1429009215378843, "learning_rate": 1.573741843335438e-05, "loss": 0.5229, "step": 9160 }, { "epoch": 0.3263801058125657, "grad_norm": 1.6351068878750457, "learning_rate": 1.5736473279231255e-05, "loss": 0.4745, "step": 9161 }, { "epoch": 0.3264157329390598, "grad_norm": 2.1624941962545887, "learning_rate": 1.5735528048722974e-05, "loss": 0.652, "step": 9162 }, { "epoch": 0.3264513600655539, "grad_norm": 2.052747364234001, "learning_rate": 1.573458274184213e-05, "loss": 0.5963, "step": 9163 }, { "epoch": 0.32648698719204805, "grad_norm": 1.185370220769878, "learning_rate": 1.5733637358601304e-05, "loss": 0.4802, "step": 9164 }, { "epoch": 0.3265226143185421, "grad_norm": 1.1200515924685555, "learning_rate": 1.5732691899013094e-05, "loss": 0.4547, "step": 9165 }, { "epoch": 0.32655824144503626, "grad_norm": 2.0559942402347713, "learning_rate": 1.573174636309008e-05, "loss": 0.5551, "step": 9166 }, { "epoch": 0.3265938685715304, "grad_norm": 1.6439015112892796, "learning_rate": 1.573080075084486e-05, "loss": 0.5956, "step": 9167 }, { "epoch": 0.32662949569802446, "grad_norm": 1.616673239674628, "learning_rate": 1.5729855062290024e-05, "loss": 0.4927, "step": 9168 }, { "epoch": 0.3266651228245186, "grad_norm": 1.9480374627185268, "learning_rate": 1.572890929743816e-05, "loss": 0.5821, "step": 9169 }, { "epoch": 0.3267007499510127, "grad_norm": 1.6873940490768784, "learning_rate": 1.572796345630186e-05, "loss": 0.6267, "step": 9170 }, { "epoch": 0.3267363770775068, "grad_norm": 1.7646562332637223, "learning_rate": 1.5727017538893732e-05, "loss": 0.6279, "step": 9171 }, { "epoch": 0.32677200420400093, "grad_norm": 1.2277673152683473, "learning_rate": 1.572607154522636e-05, "loss": 0.4777, "step": 9172 }, { "epoch": 0.32680763133049506, "grad_norm": 1.833038573607411, "learning_rate": 1.572512547531234e-05, "loss": 0.5974, "step": 9173 }, { "epoch": 0.32684325845698914, "grad_norm": 1.711778656930264, "learning_rate": 1.5724179329164274e-05, "loss": 0.613, "step": 9174 }, { "epoch": 0.32687888558348327, "grad_norm": 1.6369337803284842, "learning_rate": 1.572323310679476e-05, "loss": 0.5382, "step": 9175 }, { "epoch": 0.3269145127099774, "grad_norm": 1.3450139228902709, "learning_rate": 1.5722286808216397e-05, "loss": 0.4778, "step": 9176 }, { "epoch": 0.3269501398364715, "grad_norm": 1.47332668516566, "learning_rate": 1.5721340433441786e-05, "loss": 0.587, "step": 9177 }, { "epoch": 0.3269857669629656, "grad_norm": 1.57669315904562, "learning_rate": 1.5720393982483528e-05, "loss": 0.5095, "step": 9178 }, { "epoch": 0.32702139408945974, "grad_norm": 1.9490471658794541, "learning_rate": 1.5719447455354225e-05, "loss": 0.5876, "step": 9179 }, { "epoch": 0.3270570212159538, "grad_norm": 1.4323471916912598, "learning_rate": 1.571850085206648e-05, "loss": 0.523, "step": 9180 }, { "epoch": 0.32709264834244794, "grad_norm": 1.7137638418078263, "learning_rate": 1.57175541726329e-05, "loss": 0.5605, "step": 9181 }, { "epoch": 0.3271282754689421, "grad_norm": 1.696596357325441, "learning_rate": 1.571660741706609e-05, "loss": 0.4936, "step": 9182 }, { "epoch": 0.32716390259543615, "grad_norm": 1.6750310610878978, "learning_rate": 1.571566058537866e-05, "loss": 0.4727, "step": 9183 }, { "epoch": 0.3271995297219303, "grad_norm": 1.7898271441607896, "learning_rate": 1.571471367758321e-05, "loss": 0.6175, "step": 9184 }, { "epoch": 0.3272351568484244, "grad_norm": 1.5270667643367406, "learning_rate": 1.5713766693692356e-05, "loss": 0.4909, "step": 9185 }, { "epoch": 0.3272707839749185, "grad_norm": 1.9064521696515357, "learning_rate": 1.5712819633718702e-05, "loss": 0.4688, "step": 9186 }, { "epoch": 0.3273064111014126, "grad_norm": 1.4994634915222518, "learning_rate": 1.5711872497674863e-05, "loss": 0.4541, "step": 9187 }, { "epoch": 0.32734203822790675, "grad_norm": 1.420661514202353, "learning_rate": 1.571092528557345e-05, "loss": 0.6046, "step": 9188 }, { "epoch": 0.3273776653544008, "grad_norm": 1.8888597173446982, "learning_rate": 1.5709977997427077e-05, "loss": 0.568, "step": 9189 }, { "epoch": 0.32741329248089496, "grad_norm": 1.5442953980499181, "learning_rate": 1.570903063324835e-05, "loss": 0.4918, "step": 9190 }, { "epoch": 0.3274489196073891, "grad_norm": 1.865316339300964, "learning_rate": 1.570808319304989e-05, "loss": 0.5435, "step": 9191 }, { "epoch": 0.32748454673388316, "grad_norm": 1.948475698078808, "learning_rate": 1.570713567684432e-05, "loss": 0.482, "step": 9192 }, { "epoch": 0.3275201738603773, "grad_norm": 1.642717825372638, "learning_rate": 1.5706188084644242e-05, "loss": 0.5092, "step": 9193 }, { "epoch": 0.3275558009868714, "grad_norm": 1.5334964563387052, "learning_rate": 1.5705240416462288e-05, "loss": 0.5205, "step": 9194 }, { "epoch": 0.3275914281133655, "grad_norm": 2.03988775904685, "learning_rate": 1.5704292672311063e-05, "loss": 0.6395, "step": 9195 }, { "epoch": 0.32762705523985963, "grad_norm": 2.2510624348739094, "learning_rate": 1.57033448522032e-05, "loss": 0.531, "step": 9196 }, { "epoch": 0.32766268236635376, "grad_norm": 1.7760073637499487, "learning_rate": 1.570239695615131e-05, "loss": 0.4932, "step": 9197 }, { "epoch": 0.32769830949284784, "grad_norm": 2.327362106815762, "learning_rate": 1.5701448984168018e-05, "loss": 0.624, "step": 9198 }, { "epoch": 0.32773393661934197, "grad_norm": 1.6044522336506981, "learning_rate": 1.5700500936265954e-05, "loss": 0.5006, "step": 9199 }, { "epoch": 0.3277695637458361, "grad_norm": 1.3863525356894677, "learning_rate": 1.5699552812457726e-05, "loss": 0.4957, "step": 9200 }, { "epoch": 0.3278051908723302, "grad_norm": 1.7110477029134221, "learning_rate": 1.5698604612755977e-05, "loss": 0.5214, "step": 9201 }, { "epoch": 0.3278408179988243, "grad_norm": 1.5302434412904464, "learning_rate": 1.569765633717332e-05, "loss": 0.5903, "step": 9202 }, { "epoch": 0.32787644512531844, "grad_norm": 1.425781153585228, "learning_rate": 1.569670798572239e-05, "loss": 0.4661, "step": 9203 }, { "epoch": 0.3279120722518125, "grad_norm": 1.480036995310168, "learning_rate": 1.569575955841581e-05, "loss": 0.4468, "step": 9204 }, { "epoch": 0.32794769937830665, "grad_norm": 1.5803228929799737, "learning_rate": 1.5694811055266207e-05, "loss": 0.49, "step": 9205 }, { "epoch": 0.3279833265048008, "grad_norm": 1.5293596821999482, "learning_rate": 1.5693862476286218e-05, "loss": 0.4669, "step": 9206 }, { "epoch": 0.32801895363129485, "grad_norm": 1.7852269366952767, "learning_rate": 1.569291382148847e-05, "loss": 0.5581, "step": 9207 }, { "epoch": 0.328054580757789, "grad_norm": 1.8269046817551147, "learning_rate": 1.5691965090885594e-05, "loss": 0.6468, "step": 9208 }, { "epoch": 0.3280902078842831, "grad_norm": 1.2147612113676631, "learning_rate": 1.5691016284490225e-05, "loss": 0.4663, "step": 9209 }, { "epoch": 0.3281258350107772, "grad_norm": 1.8784684714027293, "learning_rate": 1.5690067402314996e-05, "loss": 0.6311, "step": 9210 }, { "epoch": 0.3281614621372713, "grad_norm": 1.5449361108042476, "learning_rate": 1.5689118444372544e-05, "loss": 0.6535, "step": 9211 }, { "epoch": 0.32819708926376545, "grad_norm": 1.470473101021472, "learning_rate": 1.5688169410675503e-05, "loss": 0.6323, "step": 9212 }, { "epoch": 0.3282327163902595, "grad_norm": 1.5019613856026448, "learning_rate": 1.568722030123651e-05, "loss": 0.6208, "step": 9213 }, { "epoch": 0.32826834351675366, "grad_norm": 1.2445172959476711, "learning_rate": 1.5686271116068203e-05, "loss": 0.4647, "step": 9214 }, { "epoch": 0.3283039706432478, "grad_norm": 1.4989069497900167, "learning_rate": 1.568532185518322e-05, "loss": 0.618, "step": 9215 }, { "epoch": 0.32833959776974186, "grad_norm": 1.721416428382604, "learning_rate": 1.5684372518594203e-05, "loss": 0.4565, "step": 9216 }, { "epoch": 0.328375224896236, "grad_norm": 1.8840842648483171, "learning_rate": 1.5683423106313798e-05, "loss": 0.5767, "step": 9217 }, { "epoch": 0.3284108520227301, "grad_norm": 2.7258122968712635, "learning_rate": 1.5682473618354638e-05, "loss": 0.5193, "step": 9218 }, { "epoch": 0.3284464791492242, "grad_norm": 1.847214101113089, "learning_rate": 1.568152405472937e-05, "loss": 0.6403, "step": 9219 }, { "epoch": 0.32848210627571833, "grad_norm": 1.1712281618521578, "learning_rate": 1.568057441545064e-05, "loss": 0.4495, "step": 9220 }, { "epoch": 0.32851773340221246, "grad_norm": 1.3635109760266377, "learning_rate": 1.5679624700531088e-05, "loss": 0.5675, "step": 9221 }, { "epoch": 0.32855336052870654, "grad_norm": 1.17529608690334, "learning_rate": 1.5678674909983364e-05, "loss": 0.4839, "step": 9222 }, { "epoch": 0.32858898765520067, "grad_norm": 1.2966290443554458, "learning_rate": 1.5677725043820117e-05, "loss": 0.4676, "step": 9223 }, { "epoch": 0.3286246147816948, "grad_norm": 1.3043648655954494, "learning_rate": 1.5676775102053993e-05, "loss": 0.5, "step": 9224 }, { "epoch": 0.3286602419081889, "grad_norm": 1.342106159152518, "learning_rate": 1.5675825084697635e-05, "loss": 0.5013, "step": 9225 }, { "epoch": 0.328695869034683, "grad_norm": 0.9913385780751666, "learning_rate": 1.5674874991763705e-05, "loss": 0.4522, "step": 9226 }, { "epoch": 0.32873149616117714, "grad_norm": 1.4314605695622709, "learning_rate": 1.5673924823264845e-05, "loss": 0.6129, "step": 9227 }, { "epoch": 0.3287671232876712, "grad_norm": 1.2139360465201654, "learning_rate": 1.5672974579213705e-05, "loss": 0.4997, "step": 9228 }, { "epoch": 0.32880275041416535, "grad_norm": 2.1180809967447907, "learning_rate": 1.5672024259622955e-05, "loss": 0.5114, "step": 9229 }, { "epoch": 0.3288383775406595, "grad_norm": 1.6518314080438887, "learning_rate": 1.5671073864505227e-05, "loss": 0.5152, "step": 9230 }, { "epoch": 0.32887400466715355, "grad_norm": 2.3995211777251733, "learning_rate": 1.5670123393873194e-05, "loss": 0.5561, "step": 9231 }, { "epoch": 0.3289096317936477, "grad_norm": 1.8276370036754859, "learning_rate": 1.56691728477395e-05, "loss": 0.5881, "step": 9232 }, { "epoch": 0.3289452589201418, "grad_norm": 1.4942383569741267, "learning_rate": 1.5668222226116808e-05, "loss": 0.5574, "step": 9233 }, { "epoch": 0.3289808860466359, "grad_norm": 1.2529255624351556, "learning_rate": 1.5667271529017777e-05, "loss": 0.4405, "step": 9234 }, { "epoch": 0.32901651317313, "grad_norm": 1.7587665288848993, "learning_rate": 1.5666320756455063e-05, "loss": 0.6141, "step": 9235 }, { "epoch": 0.32905214029962415, "grad_norm": 1.2776639165559462, "learning_rate": 1.5665369908441328e-05, "loss": 0.4861, "step": 9236 }, { "epoch": 0.3290877674261182, "grad_norm": 1.226487538459183, "learning_rate": 1.5664418984989235e-05, "loss": 0.5146, "step": 9237 }, { "epoch": 0.32912339455261236, "grad_norm": 1.7346897225589017, "learning_rate": 1.566346798611144e-05, "loss": 0.5447, "step": 9238 }, { "epoch": 0.3291590216791065, "grad_norm": 1.1916419884785958, "learning_rate": 1.5662516911820612e-05, "loss": 0.4888, "step": 9239 }, { "epoch": 0.32919464880560056, "grad_norm": 1.6555974598190777, "learning_rate": 1.5661565762129414e-05, "loss": 0.6853, "step": 9240 }, { "epoch": 0.3292302759320947, "grad_norm": 1.5442376679231082, "learning_rate": 1.566061453705051e-05, "loss": 0.6243, "step": 9241 }, { "epoch": 0.3292659030585888, "grad_norm": 1.171305245167591, "learning_rate": 1.5659663236596564e-05, "loss": 0.4679, "step": 9242 }, { "epoch": 0.3293015301850829, "grad_norm": 1.900367601560187, "learning_rate": 1.565871186078025e-05, "loss": 0.6208, "step": 9243 }, { "epoch": 0.32933715731157703, "grad_norm": 1.4743680085073854, "learning_rate": 1.5657760409614232e-05, "loss": 0.5961, "step": 9244 }, { "epoch": 0.32937278443807116, "grad_norm": 1.135599865824112, "learning_rate": 1.565680888311118e-05, "loss": 0.5086, "step": 9245 }, { "epoch": 0.32940841156456524, "grad_norm": 1.0872059625432522, "learning_rate": 1.5655857281283764e-05, "loss": 0.4815, "step": 9246 }, { "epoch": 0.32944403869105937, "grad_norm": 1.603382180265704, "learning_rate": 1.5654905604144656e-05, "loss": 0.5456, "step": 9247 }, { "epoch": 0.3294796658175535, "grad_norm": 1.2135552534204606, "learning_rate": 1.5653953851706524e-05, "loss": 0.5335, "step": 9248 }, { "epoch": 0.3295152929440476, "grad_norm": 1.1871265885547604, "learning_rate": 1.5653002023982047e-05, "loss": 0.4156, "step": 9249 }, { "epoch": 0.3295509200705417, "grad_norm": 1.520276342716806, "learning_rate": 1.5652050120983896e-05, "loss": 0.5706, "step": 9250 }, { "epoch": 0.32958654719703584, "grad_norm": 1.160207046973232, "learning_rate": 1.5651098142724744e-05, "loss": 0.5193, "step": 9251 }, { "epoch": 0.3296221743235299, "grad_norm": 1.4771113912602567, "learning_rate": 1.5650146089217276e-05, "loss": 0.4988, "step": 9252 }, { "epoch": 0.32965780145002405, "grad_norm": 1.1180543386897723, "learning_rate": 1.564919396047416e-05, "loss": 0.5322, "step": 9253 }, { "epoch": 0.3296934285765182, "grad_norm": 1.0099926173881275, "learning_rate": 1.564824175650808e-05, "loss": 0.4921, "step": 9254 }, { "epoch": 0.32972905570301225, "grad_norm": 1.6380770357762597, "learning_rate": 1.564728947733171e-05, "loss": 0.7093, "step": 9255 }, { "epoch": 0.3297646828295064, "grad_norm": 1.069201028441039, "learning_rate": 1.5646337122957736e-05, "loss": 0.5156, "step": 9256 }, { "epoch": 0.3298003099560005, "grad_norm": 1.3777801259459659, "learning_rate": 1.5645384693398837e-05, "loss": 0.568, "step": 9257 }, { "epoch": 0.3298359370824946, "grad_norm": 1.160386763039796, "learning_rate": 1.5644432188667695e-05, "loss": 0.4762, "step": 9258 }, { "epoch": 0.3298715642089887, "grad_norm": 1.6321611205094442, "learning_rate": 1.5643479608776992e-05, "loss": 0.6055, "step": 9259 }, { "epoch": 0.32990719133548285, "grad_norm": 1.3046510592639744, "learning_rate": 1.5642526953739413e-05, "loss": 0.5689, "step": 9260 }, { "epoch": 0.32994281846197693, "grad_norm": 1.6207200888767295, "learning_rate": 1.5641574223567643e-05, "loss": 0.7026, "step": 9261 }, { "epoch": 0.32997844558847106, "grad_norm": 1.1509622620401747, "learning_rate": 1.564062141827437e-05, "loss": 0.5053, "step": 9262 }, { "epoch": 0.3300140727149652, "grad_norm": 1.166120221728341, "learning_rate": 1.563966853787228e-05, "loss": 0.4805, "step": 9263 }, { "epoch": 0.33004969984145927, "grad_norm": 1.7956650484077359, "learning_rate": 1.5638715582374065e-05, "loss": 0.5358, "step": 9264 }, { "epoch": 0.3300853269679534, "grad_norm": 1.1968873572185867, "learning_rate": 1.5637762551792404e-05, "loss": 0.4657, "step": 9265 }, { "epoch": 0.3301209540944475, "grad_norm": 1.804143985913827, "learning_rate": 1.563680944614e-05, "loss": 0.579, "step": 9266 }, { "epoch": 0.3301565812209416, "grad_norm": 1.1410019834158227, "learning_rate": 1.5635856265429538e-05, "loss": 0.4825, "step": 9267 }, { "epoch": 0.33019220834743573, "grad_norm": 1.6998721641976773, "learning_rate": 1.5634903009673706e-05, "loss": 0.5545, "step": 9268 }, { "epoch": 0.33022783547392986, "grad_norm": 1.154514710384808, "learning_rate": 1.5633949678885208e-05, "loss": 0.4596, "step": 9269 }, { "epoch": 0.33026346260042394, "grad_norm": 1.5163327464074605, "learning_rate": 1.5632996273076727e-05, "loss": 0.4558, "step": 9270 }, { "epoch": 0.33029908972691807, "grad_norm": 1.0885092515038508, "learning_rate": 1.5632042792260965e-05, "loss": 0.5136, "step": 9271 }, { "epoch": 0.3303347168534122, "grad_norm": 1.101763934129069, "learning_rate": 1.5631089236450616e-05, "loss": 0.4721, "step": 9272 }, { "epoch": 0.3303703439799063, "grad_norm": 1.6678514911307687, "learning_rate": 1.563013560565838e-05, "loss": 0.5848, "step": 9273 }, { "epoch": 0.3304059711064004, "grad_norm": 1.4282743672121148, "learning_rate": 1.5629181899896952e-05, "loss": 0.4707, "step": 9274 }, { "epoch": 0.33044159823289454, "grad_norm": 1.9343450486169844, "learning_rate": 1.562822811917903e-05, "loss": 0.6302, "step": 9275 }, { "epoch": 0.3304772253593886, "grad_norm": 2.2003575330830043, "learning_rate": 1.562727426351732e-05, "loss": 0.5982, "step": 9276 }, { "epoch": 0.33051285248588275, "grad_norm": 1.2936732940518443, "learning_rate": 1.562632033292452e-05, "loss": 0.5164, "step": 9277 }, { "epoch": 0.3305484796123769, "grad_norm": 1.146931632475021, "learning_rate": 1.562536632741333e-05, "loss": 0.4644, "step": 9278 }, { "epoch": 0.33058410673887095, "grad_norm": 2.070458068661015, "learning_rate": 1.562441224699646e-05, "loss": 0.6914, "step": 9279 }, { "epoch": 0.3306197338653651, "grad_norm": 1.0239666674906522, "learning_rate": 1.5623458091686606e-05, "loss": 0.4871, "step": 9280 }, { "epoch": 0.3306553609918592, "grad_norm": 1.1205108573758447, "learning_rate": 1.5622503861496475e-05, "loss": 0.5185, "step": 9281 }, { "epoch": 0.3306909881183533, "grad_norm": 1.627919200176909, "learning_rate": 1.562154955643878e-05, "loss": 0.6151, "step": 9282 }, { "epoch": 0.3307266152448474, "grad_norm": 1.9392455989991055, "learning_rate": 1.562059517652622e-05, "loss": 0.6693, "step": 9283 }, { "epoch": 0.33076224237134155, "grad_norm": 1.8565751903429109, "learning_rate": 1.5619640721771507e-05, "loss": 0.5424, "step": 9284 }, { "epoch": 0.33079786949783563, "grad_norm": 1.4612211176837009, "learning_rate": 1.5618686192187353e-05, "loss": 0.6132, "step": 9285 }, { "epoch": 0.33083349662432976, "grad_norm": 1.0820848245760935, "learning_rate": 1.561773158778646e-05, "loss": 0.4788, "step": 9286 }, { "epoch": 0.3308691237508239, "grad_norm": 1.1084422398808005, "learning_rate": 1.5616776908581546e-05, "loss": 0.4977, "step": 9287 }, { "epoch": 0.33090475087731797, "grad_norm": 1.1149011141728764, "learning_rate": 1.5615822154585326e-05, "loss": 0.5033, "step": 9288 }, { "epoch": 0.3309403780038121, "grad_norm": 0.9858046625644722, "learning_rate": 1.5614867325810503e-05, "loss": 0.4547, "step": 9289 }, { "epoch": 0.33097600513030623, "grad_norm": 1.357941657601103, "learning_rate": 1.56139124222698e-05, "loss": 0.6055, "step": 9290 }, { "epoch": 0.3310116322568003, "grad_norm": 1.1193652167751345, "learning_rate": 1.5612957443975928e-05, "loss": 0.4883, "step": 9291 }, { "epoch": 0.33104725938329443, "grad_norm": 1.196051856941624, "learning_rate": 1.5612002390941604e-05, "loss": 0.5392, "step": 9292 }, { "epoch": 0.33108288650978857, "grad_norm": 1.0589466759108395, "learning_rate": 1.561104726317955e-05, "loss": 0.4539, "step": 9293 }, { "epoch": 0.33111851363628264, "grad_norm": 1.4927796245661331, "learning_rate": 1.5610092060702477e-05, "loss": 0.5891, "step": 9294 }, { "epoch": 0.33115414076277677, "grad_norm": 2.114229027975405, "learning_rate": 1.5609136783523106e-05, "loss": 0.5118, "step": 9295 }, { "epoch": 0.3311897678892709, "grad_norm": 1.1840855097341643, "learning_rate": 1.560818143165416e-05, "loss": 0.5282, "step": 9296 }, { "epoch": 0.331225395015765, "grad_norm": 1.6651647648454575, "learning_rate": 1.560722600510835e-05, "loss": 0.6303, "step": 9297 }, { "epoch": 0.3312610221422591, "grad_norm": 1.3283892221663338, "learning_rate": 1.5606270503898414e-05, "loss": 0.4931, "step": 9298 }, { "epoch": 0.33129664926875324, "grad_norm": 1.1037031383321758, "learning_rate": 1.560531492803707e-05, "loss": 0.4685, "step": 9299 }, { "epoch": 0.3313322763952473, "grad_norm": 1.6685778686032933, "learning_rate": 1.5604359277537037e-05, "loss": 0.6165, "step": 9300 }, { "epoch": 0.33136790352174145, "grad_norm": 2.8922417042825286, "learning_rate": 1.560340355241104e-05, "loss": 0.619, "step": 9301 }, { "epoch": 0.3314035306482356, "grad_norm": 1.1462300976735385, "learning_rate": 1.5602447752671813e-05, "loss": 0.5012, "step": 9302 }, { "epoch": 0.33143915777472965, "grad_norm": 0.9984662323632082, "learning_rate": 1.5601491878332077e-05, "loss": 0.475, "step": 9303 }, { "epoch": 0.3314747849012238, "grad_norm": 2.4835752946991203, "learning_rate": 1.5600535929404563e-05, "loss": 0.6138, "step": 9304 }, { "epoch": 0.3315104120277179, "grad_norm": 1.513729329177019, "learning_rate": 1.5599579905901996e-05, "loss": 0.5922, "step": 9305 }, { "epoch": 0.331546039154212, "grad_norm": 1.666663132262346, "learning_rate": 1.5598623807837107e-05, "loss": 0.5962, "step": 9306 }, { "epoch": 0.3315816662807061, "grad_norm": 1.063607184951979, "learning_rate": 1.559766763522263e-05, "loss": 0.5175, "step": 9307 }, { "epoch": 0.33161729340720025, "grad_norm": 1.5249894954461531, "learning_rate": 1.55967113880713e-05, "loss": 0.5441, "step": 9308 }, { "epoch": 0.33165292053369433, "grad_norm": 1.681788713889081, "learning_rate": 1.559575506639584e-05, "loss": 0.6359, "step": 9309 }, { "epoch": 0.33168854766018846, "grad_norm": 1.4201586651267766, "learning_rate": 1.559479867020899e-05, "loss": 0.4786, "step": 9310 }, { "epoch": 0.3317241747866826, "grad_norm": 1.702974680064025, "learning_rate": 1.559384219952349e-05, "loss": 0.5902, "step": 9311 }, { "epoch": 0.33175980191317667, "grad_norm": 1.997540141200501, "learning_rate": 1.559288565435207e-05, "loss": 0.5611, "step": 9312 }, { "epoch": 0.3317954290396708, "grad_norm": 1.9037471129093875, "learning_rate": 1.5591929034707468e-05, "loss": 0.5852, "step": 9313 }, { "epoch": 0.33183105616616493, "grad_norm": 1.1177081965271372, "learning_rate": 1.5590972340602425e-05, "loss": 0.5142, "step": 9314 }, { "epoch": 0.331866683292659, "grad_norm": 1.7168352466960097, "learning_rate": 1.5590015572049673e-05, "loss": 0.6549, "step": 9315 }, { "epoch": 0.33190231041915313, "grad_norm": 1.2856639904084615, "learning_rate": 1.5589058729061957e-05, "loss": 0.5199, "step": 9316 }, { "epoch": 0.33193793754564727, "grad_norm": 1.248188730234071, "learning_rate": 1.558810181165202e-05, "loss": 0.4849, "step": 9317 }, { "epoch": 0.33197356467214134, "grad_norm": 1.220630047895177, "learning_rate": 1.55871448198326e-05, "loss": 0.4634, "step": 9318 }, { "epoch": 0.3320091917986355, "grad_norm": 1.1817475058777356, "learning_rate": 1.5586187753616443e-05, "loss": 0.536, "step": 9319 }, { "epoch": 0.3320448189251296, "grad_norm": 1.5283849664553844, "learning_rate": 1.558523061301629e-05, "loss": 0.5781, "step": 9320 }, { "epoch": 0.3320804460516237, "grad_norm": 1.1727072425480047, "learning_rate": 1.5584273398044887e-05, "loss": 0.4927, "step": 9321 }, { "epoch": 0.3321160731781178, "grad_norm": 2.417916531795476, "learning_rate": 1.558331610871498e-05, "loss": 0.5951, "step": 9322 }, { "epoch": 0.33215170030461194, "grad_norm": 1.231291655549245, "learning_rate": 1.5582358745039318e-05, "loss": 0.4892, "step": 9323 }, { "epoch": 0.332187327431106, "grad_norm": 1.1373307812237003, "learning_rate": 1.558140130703065e-05, "loss": 0.502, "step": 9324 }, { "epoch": 0.33222295455760015, "grad_norm": 1.2590229852903942, "learning_rate": 1.5580443794701715e-05, "loss": 0.5182, "step": 9325 }, { "epoch": 0.3322585816840943, "grad_norm": 1.4370972969082674, "learning_rate": 1.557948620806528e-05, "loss": 0.5513, "step": 9326 }, { "epoch": 0.3322942088105884, "grad_norm": 1.2690049881615204, "learning_rate": 1.5578528547134078e-05, "loss": 0.5005, "step": 9327 }, { "epoch": 0.3323298359370825, "grad_norm": 1.5421902878512843, "learning_rate": 1.5577570811920873e-05, "loss": 0.5641, "step": 9328 }, { "epoch": 0.3323654630635766, "grad_norm": 2.0451254267558805, "learning_rate": 1.5576613002438413e-05, "loss": 0.5917, "step": 9329 }, { "epoch": 0.33240109019007075, "grad_norm": 1.223156728637788, "learning_rate": 1.5575655118699454e-05, "loss": 0.5045, "step": 9330 }, { "epoch": 0.3324367173165648, "grad_norm": 1.242881762736674, "learning_rate": 1.557469716071675e-05, "loss": 0.4617, "step": 9331 }, { "epoch": 0.33247234444305895, "grad_norm": 1.4173674763676687, "learning_rate": 1.5573739128503053e-05, "loss": 0.6435, "step": 9332 }, { "epoch": 0.3325079715695531, "grad_norm": 1.3056854498960848, "learning_rate": 1.557278102207113e-05, "loss": 0.4568, "step": 9333 }, { "epoch": 0.33254359869604716, "grad_norm": 1.7116628581196673, "learning_rate": 1.557182284143373e-05, "loss": 0.5404, "step": 9334 }, { "epoch": 0.3325792258225413, "grad_norm": 1.4376515107772434, "learning_rate": 1.5570864586603613e-05, "loss": 0.6584, "step": 9335 }, { "epoch": 0.3326148529490354, "grad_norm": 1.7947106985415124, "learning_rate": 1.5569906257593542e-05, "loss": 0.6474, "step": 9336 }, { "epoch": 0.3326504800755295, "grad_norm": 1.66177576406408, "learning_rate": 1.5568947854416277e-05, "loss": 0.5753, "step": 9337 }, { "epoch": 0.33268610720202363, "grad_norm": 1.6007153444691862, "learning_rate": 1.5567989377084574e-05, "loss": 0.5584, "step": 9338 }, { "epoch": 0.33272173432851776, "grad_norm": 1.3411868060607062, "learning_rate": 1.5567030825611203e-05, "loss": 0.5747, "step": 9339 }, { "epoch": 0.33275736145501184, "grad_norm": 1.21443826024811, "learning_rate": 1.5566072200008928e-05, "loss": 0.5387, "step": 9340 }, { "epoch": 0.33279298858150597, "grad_norm": 1.4061648684785373, "learning_rate": 1.5565113500290513e-05, "loss": 0.5585, "step": 9341 }, { "epoch": 0.3328286157080001, "grad_norm": 1.7899240975196358, "learning_rate": 1.5564154726468717e-05, "loss": 0.5889, "step": 9342 }, { "epoch": 0.3328642428344942, "grad_norm": 1.1187939774271336, "learning_rate": 1.5563195878556312e-05, "loss": 0.5259, "step": 9343 }, { "epoch": 0.3328998699609883, "grad_norm": 1.7002172689468669, "learning_rate": 1.5562236956566064e-05, "loss": 0.5798, "step": 9344 }, { "epoch": 0.33293549708748243, "grad_norm": 1.099992229633278, "learning_rate": 1.5561277960510747e-05, "loss": 0.5547, "step": 9345 }, { "epoch": 0.3329711242139765, "grad_norm": 1.3913734982263832, "learning_rate": 1.5560318890403125e-05, "loss": 0.5413, "step": 9346 }, { "epoch": 0.33300675134047064, "grad_norm": 1.2252897338902702, "learning_rate": 1.5559359746255972e-05, "loss": 0.446, "step": 9347 }, { "epoch": 0.3330423784669648, "grad_norm": 1.2900147682740517, "learning_rate": 1.5558400528082057e-05, "loss": 0.5729, "step": 9348 }, { "epoch": 0.33307800559345885, "grad_norm": 1.7305553722842033, "learning_rate": 1.5557441235894156e-05, "loss": 0.58, "step": 9349 }, { "epoch": 0.333113632719953, "grad_norm": 1.4141130329037341, "learning_rate": 1.5556481869705038e-05, "loss": 0.5365, "step": 9350 }, { "epoch": 0.3331492598464471, "grad_norm": 1.7128626603687385, "learning_rate": 1.5555522429527484e-05, "loss": 0.5514, "step": 9351 }, { "epoch": 0.3331848869729412, "grad_norm": 1.3418584652287162, "learning_rate": 1.555456291537426e-05, "loss": 0.5166, "step": 9352 }, { "epoch": 0.3332205140994353, "grad_norm": 1.7217587743449259, "learning_rate": 1.5553603327258153e-05, "loss": 0.5703, "step": 9353 }, { "epoch": 0.33325614122592945, "grad_norm": 1.0629412405316851, "learning_rate": 1.5552643665191933e-05, "loss": 0.499, "step": 9354 }, { "epoch": 0.3332917683524235, "grad_norm": 1.142029991253139, "learning_rate": 1.5551683929188383e-05, "loss": 0.4811, "step": 9355 }, { "epoch": 0.33332739547891765, "grad_norm": 1.348172644948658, "learning_rate": 1.555072411926028e-05, "loss": 0.5324, "step": 9356 }, { "epoch": 0.3333630226054118, "grad_norm": 1.444663540271221, "learning_rate": 1.5549764235420407e-05, "loss": 0.6162, "step": 9357 }, { "epoch": 0.33339864973190586, "grad_norm": 1.878342447562084, "learning_rate": 1.554880427768154e-05, "loss": 0.609, "step": 9358 }, { "epoch": 0.3334342768584, "grad_norm": 1.8668691361510532, "learning_rate": 1.5547844246056472e-05, "loss": 0.5825, "step": 9359 }, { "epoch": 0.3334699039848941, "grad_norm": 1.4861885006259816, "learning_rate": 1.5546884140557977e-05, "loss": 0.5532, "step": 9360 }, { "epoch": 0.3335055311113882, "grad_norm": 1.1223342462425503, "learning_rate": 1.5545923961198842e-05, "loss": 0.4849, "step": 9361 }, { "epoch": 0.33354115823788233, "grad_norm": 1.4441098396367038, "learning_rate": 1.5544963707991855e-05, "loss": 0.63, "step": 9362 }, { "epoch": 0.33357678536437646, "grad_norm": 1.621080390218648, "learning_rate": 1.55440033809498e-05, "loss": 0.625, "step": 9363 }, { "epoch": 0.33361241249087054, "grad_norm": 1.66908771282261, "learning_rate": 1.5543042980085467e-05, "loss": 0.5739, "step": 9364 }, { "epoch": 0.33364803961736467, "grad_norm": 1.2106529378876445, "learning_rate": 1.5542082505411636e-05, "loss": 0.4576, "step": 9365 }, { "epoch": 0.3336836667438588, "grad_norm": 1.9547679850127744, "learning_rate": 1.5541121956941107e-05, "loss": 0.5733, "step": 9366 }, { "epoch": 0.3337192938703529, "grad_norm": 1.205696432221651, "learning_rate": 1.5540161334686665e-05, "loss": 0.4817, "step": 9367 }, { "epoch": 0.333754920996847, "grad_norm": 1.5940848242722436, "learning_rate": 1.5539200638661106e-05, "loss": 0.6322, "step": 9368 }, { "epoch": 0.33379054812334114, "grad_norm": 1.4495691607380676, "learning_rate": 1.5538239868877215e-05, "loss": 0.5401, "step": 9369 }, { "epoch": 0.3338261752498352, "grad_norm": 1.2215864382070767, "learning_rate": 1.553727902534779e-05, "loss": 0.4936, "step": 9370 }, { "epoch": 0.33386180237632934, "grad_norm": 2.0824136011847623, "learning_rate": 1.5536318108085624e-05, "loss": 0.6087, "step": 9371 }, { "epoch": 0.3338974295028235, "grad_norm": 1.144118562612704, "learning_rate": 1.5535357117103514e-05, "loss": 0.4917, "step": 9372 }, { "epoch": 0.33393305662931755, "grad_norm": 1.6603337392134165, "learning_rate": 1.5534396052414257e-05, "loss": 0.6191, "step": 9373 }, { "epoch": 0.3339686837558117, "grad_norm": 1.458635587187348, "learning_rate": 1.5533434914030644e-05, "loss": 0.6053, "step": 9374 }, { "epoch": 0.3340043108823058, "grad_norm": 1.3788798953055663, "learning_rate": 1.5532473701965477e-05, "loss": 0.5724, "step": 9375 }, { "epoch": 0.3340399380087999, "grad_norm": 1.2358619507128044, "learning_rate": 1.553151241623156e-05, "loss": 0.486, "step": 9376 }, { "epoch": 0.334075565135294, "grad_norm": 1.4188327004472128, "learning_rate": 1.5530551056841687e-05, "loss": 0.6064, "step": 9377 }, { "epoch": 0.33411119226178815, "grad_norm": 1.16204724892493, "learning_rate": 1.5529589623808657e-05, "loss": 0.5264, "step": 9378 }, { "epoch": 0.3341468193882822, "grad_norm": 1.4631528178815298, "learning_rate": 1.5528628117145282e-05, "loss": 0.6493, "step": 9379 }, { "epoch": 0.33418244651477635, "grad_norm": 1.3222708040537383, "learning_rate": 1.5527666536864357e-05, "loss": 0.525, "step": 9380 }, { "epoch": 0.3342180736412705, "grad_norm": 1.583983941041683, "learning_rate": 1.552670488297869e-05, "loss": 0.534, "step": 9381 }, { "epoch": 0.33425370076776456, "grad_norm": 1.0969501868055076, "learning_rate": 1.5525743155501084e-05, "loss": 0.4992, "step": 9382 }, { "epoch": 0.3342893278942587, "grad_norm": 1.2234327264296498, "learning_rate": 1.5524781354444344e-05, "loss": 0.5807, "step": 9383 }, { "epoch": 0.3343249550207528, "grad_norm": 1.1326901768814504, "learning_rate": 1.552381947982128e-05, "loss": 0.5149, "step": 9384 }, { "epoch": 0.3343605821472469, "grad_norm": 1.5354658175088611, "learning_rate": 1.5522857531644697e-05, "loss": 0.5489, "step": 9385 }, { "epoch": 0.33439620927374103, "grad_norm": 1.0864879032001713, "learning_rate": 1.5521895509927408e-05, "loss": 0.4711, "step": 9386 }, { "epoch": 0.33443183640023516, "grad_norm": 1.6044232423543001, "learning_rate": 1.552093341468222e-05, "loss": 0.5333, "step": 9387 }, { "epoch": 0.33446746352672924, "grad_norm": 1.9032789477648702, "learning_rate": 1.5519971245921946e-05, "loss": 0.6119, "step": 9388 }, { "epoch": 0.33450309065322337, "grad_norm": 1.134142613508444, "learning_rate": 1.5519009003659392e-05, "loss": 0.5331, "step": 9389 }, { "epoch": 0.3345387177797175, "grad_norm": 1.4558451107326194, "learning_rate": 1.5518046687907378e-05, "loss": 0.5375, "step": 9390 }, { "epoch": 0.3345743449062116, "grad_norm": 2.0399099907330416, "learning_rate": 1.551708429867872e-05, "loss": 0.5824, "step": 9391 }, { "epoch": 0.3346099720327057, "grad_norm": 1.636513985045146, "learning_rate": 1.551612183598622e-05, "loss": 0.5773, "step": 9392 }, { "epoch": 0.33464559915919984, "grad_norm": 1.3653509095483933, "learning_rate": 1.551515929984271e-05, "loss": 0.4985, "step": 9393 }, { "epoch": 0.3346812262856939, "grad_norm": 1.1118123005939975, "learning_rate": 1.5514196690260993e-05, "loss": 0.5125, "step": 9394 }, { "epoch": 0.33471685341218804, "grad_norm": 1.1108737261151096, "learning_rate": 1.5513234007253894e-05, "loss": 0.5182, "step": 9395 }, { "epoch": 0.3347524805386822, "grad_norm": 1.1570724171586901, "learning_rate": 1.5512271250834233e-05, "loss": 0.46, "step": 9396 }, { "epoch": 0.33478810766517625, "grad_norm": 1.9194004223511598, "learning_rate": 1.5511308421014826e-05, "loss": 0.673, "step": 9397 }, { "epoch": 0.3348237347916704, "grad_norm": 1.0364990132081835, "learning_rate": 1.55103455178085e-05, "loss": 0.4738, "step": 9398 }, { "epoch": 0.3348593619181645, "grad_norm": 1.4041572456309046, "learning_rate": 1.5509382541228065e-05, "loss": 0.5854, "step": 9399 }, { "epoch": 0.3348949890446586, "grad_norm": 1.6005955210640561, "learning_rate": 1.5508419491286354e-05, "loss": 0.6179, "step": 9400 }, { "epoch": 0.3349306161711527, "grad_norm": 1.7692918980592434, "learning_rate": 1.5507456367996184e-05, "loss": 0.6134, "step": 9401 }, { "epoch": 0.33496624329764685, "grad_norm": 1.0741957599028014, "learning_rate": 1.550649317137039e-05, "loss": 0.4951, "step": 9402 }, { "epoch": 0.3350018704241409, "grad_norm": 1.6539296338071137, "learning_rate": 1.5505529901421782e-05, "loss": 0.6407, "step": 9403 }, { "epoch": 0.33503749755063505, "grad_norm": 1.7267177723487694, "learning_rate": 1.5504566558163204e-05, "loss": 0.5877, "step": 9404 }, { "epoch": 0.3350731246771292, "grad_norm": 2.168364675099356, "learning_rate": 1.550360314160747e-05, "loss": 0.5588, "step": 9405 }, { "epoch": 0.33510875180362326, "grad_norm": 1.3003895037734663, "learning_rate": 1.5502639651767414e-05, "loss": 0.4583, "step": 9406 }, { "epoch": 0.3351443789301174, "grad_norm": 2.007145425429247, "learning_rate": 1.5501676088655864e-05, "loss": 0.5682, "step": 9407 }, { "epoch": 0.3351800060566115, "grad_norm": 1.2919545346356236, "learning_rate": 1.550071245228565e-05, "loss": 0.5469, "step": 9408 }, { "epoch": 0.3352156331831056, "grad_norm": 2.2921089455881147, "learning_rate": 1.549974874266961e-05, "loss": 0.6104, "step": 9409 }, { "epoch": 0.33525126030959973, "grad_norm": 1.203484850976736, "learning_rate": 1.549878495982057e-05, "loss": 0.5183, "step": 9410 }, { "epoch": 0.33528688743609386, "grad_norm": 1.031589950692463, "learning_rate": 1.549782110375136e-05, "loss": 0.4748, "step": 9411 }, { "epoch": 0.33532251456258794, "grad_norm": 1.4555186419715904, "learning_rate": 1.5496857174474824e-05, "loss": 0.5508, "step": 9412 }, { "epoch": 0.33535814168908207, "grad_norm": 1.0854463849695553, "learning_rate": 1.549589317200379e-05, "loss": 0.4589, "step": 9413 }, { "epoch": 0.3353937688155762, "grad_norm": 1.4946514136283717, "learning_rate": 1.5494929096351098e-05, "loss": 0.6438, "step": 9414 }, { "epoch": 0.3354293959420703, "grad_norm": 1.2475969458456146, "learning_rate": 1.5493964947529584e-05, "loss": 0.5122, "step": 9415 }, { "epoch": 0.3354650230685644, "grad_norm": 1.6198318626121861, "learning_rate": 1.5493000725552087e-05, "loss": 0.4766, "step": 9416 }, { "epoch": 0.33550065019505854, "grad_norm": 1.125934110399418, "learning_rate": 1.5492036430431446e-05, "loss": 0.4876, "step": 9417 }, { "epoch": 0.3355362773215526, "grad_norm": 2.1017977713262272, "learning_rate": 1.5491072062180504e-05, "loss": 0.611, "step": 9418 }, { "epoch": 0.33557190444804674, "grad_norm": 1.0934075975740563, "learning_rate": 1.5490107620812096e-05, "loss": 0.4687, "step": 9419 }, { "epoch": 0.3356075315745409, "grad_norm": 1.3604593036590347, "learning_rate": 1.548914310633907e-05, "loss": 0.5105, "step": 9420 }, { "epoch": 0.33564315870103495, "grad_norm": 2.147332294804271, "learning_rate": 1.5488178518774265e-05, "loss": 0.6263, "step": 9421 }, { "epoch": 0.3356787858275291, "grad_norm": 1.408459184809846, "learning_rate": 1.5487213858130528e-05, "loss": 0.5759, "step": 9422 }, { "epoch": 0.3357144129540232, "grad_norm": 2.046694191680857, "learning_rate": 1.5486249124420702e-05, "loss": 0.6445, "step": 9423 }, { "epoch": 0.3357500400805173, "grad_norm": 2.078604326886664, "learning_rate": 1.5485284317657635e-05, "loss": 0.579, "step": 9424 }, { "epoch": 0.3357856672070114, "grad_norm": 2.019686259344559, "learning_rate": 1.548431943785417e-05, "loss": 0.5567, "step": 9425 }, { "epoch": 0.33582129433350555, "grad_norm": 1.3938449241454673, "learning_rate": 1.5483354485023164e-05, "loss": 0.5389, "step": 9426 }, { "epoch": 0.3358569214599996, "grad_norm": 1.3067736071739782, "learning_rate": 1.5482389459177454e-05, "loss": 0.5862, "step": 9427 }, { "epoch": 0.33589254858649376, "grad_norm": 2.6404400390109366, "learning_rate": 1.5481424360329906e-05, "loss": 0.5289, "step": 9428 }, { "epoch": 0.3359281757129879, "grad_norm": 1.415904960680851, "learning_rate": 1.5480459188493352e-05, "loss": 0.5182, "step": 9429 }, { "epoch": 0.33596380283948196, "grad_norm": 1.0837081910483322, "learning_rate": 1.5479493943680657e-05, "loss": 0.4647, "step": 9430 }, { "epoch": 0.3359994299659761, "grad_norm": 1.123238976482765, "learning_rate": 1.547852862590467e-05, "loss": 0.4997, "step": 9431 }, { "epoch": 0.3360350570924702, "grad_norm": 1.1216810322475315, "learning_rate": 1.547756323517825e-05, "loss": 0.4493, "step": 9432 }, { "epoch": 0.3360706842189643, "grad_norm": 2.293045510161855, "learning_rate": 1.5476597771514243e-05, "loss": 0.5396, "step": 9433 }, { "epoch": 0.33610631134545843, "grad_norm": 0.9547840679051115, "learning_rate": 1.5475632234925505e-05, "loss": 0.4324, "step": 9434 }, { "epoch": 0.33614193847195256, "grad_norm": 1.7099076407673017, "learning_rate": 1.54746666254249e-05, "loss": 0.5457, "step": 9435 }, { "epoch": 0.33617756559844664, "grad_norm": 1.0299813060604714, "learning_rate": 1.547370094302528e-05, "loss": 0.5042, "step": 9436 }, { "epoch": 0.33621319272494077, "grad_norm": 1.1829312921806727, "learning_rate": 1.5472735187739507e-05, "loss": 0.4789, "step": 9437 }, { "epoch": 0.3362488198514349, "grad_norm": 2.621819427728769, "learning_rate": 1.547176935958044e-05, "loss": 0.5839, "step": 9438 }, { "epoch": 0.336284446977929, "grad_norm": 1.637695374713526, "learning_rate": 1.5470803458560943e-05, "loss": 0.5448, "step": 9439 }, { "epoch": 0.3363200741044231, "grad_norm": 1.1779243060768603, "learning_rate": 1.546983748469387e-05, "loss": 0.4689, "step": 9440 }, { "epoch": 0.33635570123091724, "grad_norm": 1.5888537381551435, "learning_rate": 1.546887143799209e-05, "loss": 0.634, "step": 9441 }, { "epoch": 0.3363913283574113, "grad_norm": 1.026409043940151, "learning_rate": 1.5467905318468462e-05, "loss": 0.4303, "step": 9442 }, { "epoch": 0.33642695548390544, "grad_norm": 1.4220372079425545, "learning_rate": 1.5466939126135856e-05, "loss": 0.6332, "step": 9443 }, { "epoch": 0.3364625826103996, "grad_norm": 1.4829404429232016, "learning_rate": 1.546597286100713e-05, "loss": 0.6503, "step": 9444 }, { "epoch": 0.33649820973689365, "grad_norm": 1.0633022712876958, "learning_rate": 1.546500652309516e-05, "loss": 0.4464, "step": 9445 }, { "epoch": 0.3365338368633878, "grad_norm": 1.323685847199081, "learning_rate": 1.5464040112412804e-05, "loss": 0.5792, "step": 9446 }, { "epoch": 0.3365694639898819, "grad_norm": 1.1732870957589288, "learning_rate": 1.5463073628972935e-05, "loss": 0.494, "step": 9447 }, { "epoch": 0.336605091116376, "grad_norm": 2.348942004096211, "learning_rate": 1.5462107072788422e-05, "loss": 0.5906, "step": 9448 }, { "epoch": 0.3366407182428701, "grad_norm": 1.150442180282612, "learning_rate": 1.546114044387214e-05, "loss": 0.4815, "step": 9449 }, { "epoch": 0.33667634536936425, "grad_norm": 1.61952430072783, "learning_rate": 1.546017374223695e-05, "loss": 0.5507, "step": 9450 }, { "epoch": 0.3367119724958583, "grad_norm": 1.4491983128771038, "learning_rate": 1.5459206967895735e-05, "loss": 0.6402, "step": 9451 }, { "epoch": 0.33674759962235246, "grad_norm": 1.05235166657599, "learning_rate": 1.5458240120861363e-05, "loss": 0.542, "step": 9452 }, { "epoch": 0.3367832267488466, "grad_norm": 1.6684355262619281, "learning_rate": 1.5457273201146708e-05, "loss": 0.6369, "step": 9453 }, { "epoch": 0.33681885387534066, "grad_norm": 1.8033346955932357, "learning_rate": 1.5456306208764646e-05, "loss": 0.5792, "step": 9454 }, { "epoch": 0.3368544810018348, "grad_norm": 1.2453219586020459, "learning_rate": 1.545533914372805e-05, "loss": 0.5225, "step": 9455 }, { "epoch": 0.3368901081283289, "grad_norm": 1.441148667277281, "learning_rate": 1.5454372006049803e-05, "loss": 0.6039, "step": 9456 }, { "epoch": 0.336925735254823, "grad_norm": 1.511653905678121, "learning_rate": 1.5453404795742783e-05, "loss": 0.4822, "step": 9457 }, { "epoch": 0.33696136238131713, "grad_norm": 2.092051931203792, "learning_rate": 1.545243751281986e-05, "loss": 0.6081, "step": 9458 }, { "epoch": 0.33699698950781126, "grad_norm": 1.4595302399026857, "learning_rate": 1.545147015729392e-05, "loss": 0.5867, "step": 9459 }, { "epoch": 0.33703261663430534, "grad_norm": 1.0488074884919147, "learning_rate": 1.5450502729177853e-05, "loss": 0.4719, "step": 9460 }, { "epoch": 0.33706824376079947, "grad_norm": 1.1290260743147211, "learning_rate": 1.5449535228484526e-05, "loss": 0.4908, "step": 9461 }, { "epoch": 0.3371038708872936, "grad_norm": 1.2172200325663471, "learning_rate": 1.544856765522683e-05, "loss": 0.4871, "step": 9462 }, { "epoch": 0.3371394980137877, "grad_norm": 1.1237355635514046, "learning_rate": 1.544760000941765e-05, "loss": 0.473, "step": 9463 }, { "epoch": 0.3371751251402818, "grad_norm": 1.4189537510694048, "learning_rate": 1.544663229106986e-05, "loss": 0.4394, "step": 9464 }, { "epoch": 0.33721075226677594, "grad_norm": 1.069157408097137, "learning_rate": 1.5445664500196366e-05, "loss": 0.4368, "step": 9465 }, { "epoch": 0.33724637939327, "grad_norm": 1.0495249783306055, "learning_rate": 1.5444696636810036e-05, "loss": 0.4519, "step": 9466 }, { "epoch": 0.33728200651976414, "grad_norm": 1.571040361476126, "learning_rate": 1.5443728700923764e-05, "loss": 0.552, "step": 9467 }, { "epoch": 0.3373176336462583, "grad_norm": 1.4772146678201052, "learning_rate": 1.5442760692550443e-05, "loss": 0.5388, "step": 9468 }, { "epoch": 0.33735326077275235, "grad_norm": 1.1123652222975848, "learning_rate": 1.5441792611702956e-05, "loss": 0.4584, "step": 9469 }, { "epoch": 0.3373888878992465, "grad_norm": 1.2450557138220082, "learning_rate": 1.5440824458394196e-05, "loss": 0.4694, "step": 9470 }, { "epoch": 0.3374245150257406, "grad_norm": 1.1354301008462724, "learning_rate": 1.5439856232637058e-05, "loss": 0.4465, "step": 9471 }, { "epoch": 0.3374601421522347, "grad_norm": 1.2936632453509742, "learning_rate": 1.543888793444443e-05, "loss": 0.4867, "step": 9472 }, { "epoch": 0.3374957692787288, "grad_norm": 1.6218871740553324, "learning_rate": 1.543791956382921e-05, "loss": 0.571, "step": 9473 }, { "epoch": 0.33753139640522295, "grad_norm": 1.0584371555432663, "learning_rate": 1.543695112080429e-05, "loss": 0.5163, "step": 9474 }, { "epoch": 0.337567023531717, "grad_norm": 1.5605726982979593, "learning_rate": 1.5435982605382565e-05, "loss": 0.5779, "step": 9475 }, { "epoch": 0.33760265065821116, "grad_norm": 1.2866427486831644, "learning_rate": 1.5435014017576932e-05, "loss": 0.5672, "step": 9476 }, { "epoch": 0.3376382777847053, "grad_norm": 1.2068554486567893, "learning_rate": 1.543404535740029e-05, "loss": 0.4629, "step": 9477 }, { "epoch": 0.33767390491119936, "grad_norm": 1.3525612065648387, "learning_rate": 1.5433076624865533e-05, "loss": 0.4979, "step": 9478 }, { "epoch": 0.3377095320376935, "grad_norm": 1.2392248152081313, "learning_rate": 1.5432107819985566e-05, "loss": 0.5167, "step": 9479 }, { "epoch": 0.3377451591641876, "grad_norm": 2.719610260715072, "learning_rate": 1.5431138942773284e-05, "loss": 0.6133, "step": 9480 }, { "epoch": 0.3377807862906817, "grad_norm": 1.1988215256143697, "learning_rate": 1.543016999324159e-05, "loss": 0.5041, "step": 9481 }, { "epoch": 0.33781641341717583, "grad_norm": 1.5256659387473441, "learning_rate": 1.542920097140339e-05, "loss": 0.6138, "step": 9482 }, { "epoch": 0.33785204054366996, "grad_norm": 1.6944250249772255, "learning_rate": 1.5428231877271584e-05, "loss": 0.68, "step": 9483 }, { "epoch": 0.33788766767016404, "grad_norm": 1.1155126996647196, "learning_rate": 1.5427262710859072e-05, "loss": 0.5191, "step": 9484 }, { "epoch": 0.33792329479665817, "grad_norm": 1.1666839992193678, "learning_rate": 1.542629347217877e-05, "loss": 0.562, "step": 9485 }, { "epoch": 0.3379589219231523, "grad_norm": 1.7026537687356722, "learning_rate": 1.5425324161243572e-05, "loss": 0.7069, "step": 9486 }, { "epoch": 0.3379945490496464, "grad_norm": 1.1579206385016145, "learning_rate": 1.5424354778066395e-05, "loss": 0.4697, "step": 9487 }, { "epoch": 0.3380301761761405, "grad_norm": 1.1799474542501565, "learning_rate": 1.542338532266014e-05, "loss": 0.4679, "step": 9488 }, { "epoch": 0.33806580330263464, "grad_norm": 1.113171680668124, "learning_rate": 1.5422415795037722e-05, "loss": 0.4754, "step": 9489 }, { "epoch": 0.3381014304291287, "grad_norm": 1.387881182224579, "learning_rate": 1.5421446195212045e-05, "loss": 0.6362, "step": 9490 }, { "epoch": 0.33813705755562284, "grad_norm": 1.4264360832039646, "learning_rate": 1.5420476523196023e-05, "loss": 0.5954, "step": 9491 }, { "epoch": 0.338172684682117, "grad_norm": 1.3535735736614227, "learning_rate": 1.5419506779002568e-05, "loss": 0.4925, "step": 9492 }, { "epoch": 0.33820831180861105, "grad_norm": 1.1251285598673724, "learning_rate": 1.5418536962644593e-05, "loss": 0.5094, "step": 9493 }, { "epoch": 0.3382439389351052, "grad_norm": 1.5009884602987504, "learning_rate": 1.5417567074135006e-05, "loss": 0.6479, "step": 9494 }, { "epoch": 0.3382795660615993, "grad_norm": 1.877356423110229, "learning_rate": 1.5416597113486734e-05, "loss": 0.5579, "step": 9495 }, { "epoch": 0.3383151931880934, "grad_norm": 1.4508257537670952, "learning_rate": 1.541562708071268e-05, "loss": 0.4938, "step": 9496 }, { "epoch": 0.3383508203145875, "grad_norm": 1.020910593733844, "learning_rate": 1.5414656975825767e-05, "loss": 0.4694, "step": 9497 }, { "epoch": 0.33838644744108165, "grad_norm": 1.4313944131153349, "learning_rate": 1.5413686798838913e-05, "loss": 0.5507, "step": 9498 }, { "epoch": 0.3384220745675757, "grad_norm": 1.3804895441155345, "learning_rate": 1.541271654976504e-05, "loss": 0.5309, "step": 9499 }, { "epoch": 0.33845770169406986, "grad_norm": 1.6673098507406752, "learning_rate": 1.5411746228617058e-05, "loss": 0.6295, "step": 9500 }, { "epoch": 0.338493328820564, "grad_norm": 1.575213474289571, "learning_rate": 1.541077583540789e-05, "loss": 0.6346, "step": 9501 }, { "epoch": 0.33852895594705806, "grad_norm": 1.3021033471140948, "learning_rate": 1.5409805370150464e-05, "loss": 0.4572, "step": 9502 }, { "epoch": 0.3385645830735522, "grad_norm": 1.4911738784560153, "learning_rate": 1.5408834832857695e-05, "loss": 0.6045, "step": 9503 }, { "epoch": 0.3386002102000463, "grad_norm": 2.09320516595281, "learning_rate": 1.5407864223542512e-05, "loss": 0.5663, "step": 9504 }, { "epoch": 0.3386358373265404, "grad_norm": 1.1128333517893614, "learning_rate": 1.5406893542217835e-05, "loss": 0.4597, "step": 9505 }, { "epoch": 0.33867146445303453, "grad_norm": 2.085074336875724, "learning_rate": 1.540592278889659e-05, "loss": 0.5868, "step": 9506 }, { "epoch": 0.33870709157952866, "grad_norm": 1.7522419865436618, "learning_rate": 1.540495196359171e-05, "loss": 0.5888, "step": 9507 }, { "epoch": 0.33874271870602274, "grad_norm": 1.8919236350981072, "learning_rate": 1.5403981066316115e-05, "loss": 0.6781, "step": 9508 }, { "epoch": 0.33877834583251687, "grad_norm": 0.9678377730687863, "learning_rate": 1.5403010097082734e-05, "loss": 0.3808, "step": 9509 }, { "epoch": 0.338813972959011, "grad_norm": 1.1990757141355544, "learning_rate": 1.5402039055904496e-05, "loss": 0.488, "step": 9510 }, { "epoch": 0.3388496000855051, "grad_norm": 1.1258595250402266, "learning_rate": 1.5401067942794332e-05, "loss": 0.5192, "step": 9511 }, { "epoch": 0.3388852272119992, "grad_norm": 1.1084544186300653, "learning_rate": 1.5400096757765176e-05, "loss": 0.5006, "step": 9512 }, { "epoch": 0.33892085433849334, "grad_norm": 1.1614965119713356, "learning_rate": 1.5399125500829956e-05, "loss": 0.5372, "step": 9513 }, { "epoch": 0.33895648146498747, "grad_norm": 1.2617206184755363, "learning_rate": 1.5398154172001606e-05, "loss": 0.5319, "step": 9514 }, { "epoch": 0.33899210859148154, "grad_norm": 1.5287910529246729, "learning_rate": 1.539718277129306e-05, "loss": 0.548, "step": 9515 }, { "epoch": 0.3390277357179757, "grad_norm": 1.05894836726992, "learning_rate": 1.5396211298717248e-05, "loss": 0.4637, "step": 9516 }, { "epoch": 0.3390633628444698, "grad_norm": 1.0776525940623127, "learning_rate": 1.539523975428712e-05, "loss": 0.4812, "step": 9517 }, { "epoch": 0.3390989899709639, "grad_norm": 1.2619327025985516, "learning_rate": 1.5394268138015598e-05, "loss": 0.4544, "step": 9518 }, { "epoch": 0.339134617097458, "grad_norm": 1.1664645758887542, "learning_rate": 1.5393296449915627e-05, "loss": 0.4729, "step": 9519 }, { "epoch": 0.33917024422395214, "grad_norm": 1.8361446959184353, "learning_rate": 1.539232469000014e-05, "loss": 0.5603, "step": 9520 }, { "epoch": 0.3392058713504462, "grad_norm": 1.1035997678610339, "learning_rate": 1.539135285828209e-05, "loss": 0.4636, "step": 9521 }, { "epoch": 0.33924149847694035, "grad_norm": 1.1467567467257347, "learning_rate": 1.53903809547744e-05, "loss": 0.4837, "step": 9522 }, { "epoch": 0.3392771256034345, "grad_norm": 1.4893772762710915, "learning_rate": 1.5389408979490022e-05, "loss": 0.5805, "step": 9523 }, { "epoch": 0.33931275272992856, "grad_norm": 1.271998277506711, "learning_rate": 1.53884369324419e-05, "loss": 0.4513, "step": 9524 }, { "epoch": 0.3393483798564227, "grad_norm": 1.1253947400998052, "learning_rate": 1.538746481364297e-05, "loss": 0.4677, "step": 9525 }, { "epoch": 0.3393840069829168, "grad_norm": 1.5861260502316685, "learning_rate": 1.538649262310618e-05, "loss": 0.6147, "step": 9526 }, { "epoch": 0.3394196341094109, "grad_norm": 1.2160805351423165, "learning_rate": 1.538552036084448e-05, "loss": 0.4993, "step": 9527 }, { "epoch": 0.339455261235905, "grad_norm": 1.0572529353180917, "learning_rate": 1.538454802687081e-05, "loss": 0.5064, "step": 9528 }, { "epoch": 0.33949088836239916, "grad_norm": 1.0403221512251204, "learning_rate": 1.538357562119812e-05, "loss": 0.4386, "step": 9529 }, { "epoch": 0.33952651548889323, "grad_norm": 1.1361956335031191, "learning_rate": 1.5382603143839353e-05, "loss": 0.5264, "step": 9530 }, { "epoch": 0.33956214261538736, "grad_norm": 1.1458084753181434, "learning_rate": 1.5381630594807465e-05, "loss": 0.5231, "step": 9531 }, { "epoch": 0.3395977697418815, "grad_norm": 1.0817214539406794, "learning_rate": 1.5380657974115407e-05, "loss": 0.4575, "step": 9532 }, { "epoch": 0.33963339686837557, "grad_norm": 1.0623001185281085, "learning_rate": 1.5379685281776125e-05, "loss": 0.47, "step": 9533 }, { "epoch": 0.3396690239948697, "grad_norm": 1.6194283609089575, "learning_rate": 1.5378712517802576e-05, "loss": 0.6847, "step": 9534 }, { "epoch": 0.33970465112136383, "grad_norm": 1.0925696244547967, "learning_rate": 1.537773968220771e-05, "loss": 0.5205, "step": 9535 }, { "epoch": 0.3397402782478579, "grad_norm": 1.1422564163163007, "learning_rate": 1.5376766775004476e-05, "loss": 0.5002, "step": 9536 }, { "epoch": 0.33977590537435204, "grad_norm": 1.1159217143493827, "learning_rate": 1.537579379620584e-05, "loss": 0.4671, "step": 9537 }, { "epoch": 0.33981153250084617, "grad_norm": 1.0752325079546818, "learning_rate": 1.5374820745824753e-05, "loss": 0.4798, "step": 9538 }, { "epoch": 0.33984715962734025, "grad_norm": 2.135488051021569, "learning_rate": 1.537384762387417e-05, "loss": 0.5851, "step": 9539 }, { "epoch": 0.3398827867538344, "grad_norm": 1.5031382273287908, "learning_rate": 1.5372874430367044e-05, "loss": 0.5784, "step": 9540 }, { "epoch": 0.3399184138803285, "grad_norm": 1.449924117236052, "learning_rate": 1.5371901165316342e-05, "loss": 0.6335, "step": 9541 }, { "epoch": 0.3399540410068226, "grad_norm": 2.0984841558642477, "learning_rate": 1.5370927828735026e-05, "loss": 0.5541, "step": 9542 }, { "epoch": 0.3399896681333167, "grad_norm": 0.9790222868734949, "learning_rate": 1.5369954420636048e-05, "loss": 0.5, "step": 9543 }, { "epoch": 0.34002529525981084, "grad_norm": 1.8303453250238577, "learning_rate": 1.536898094103237e-05, "loss": 0.6047, "step": 9544 }, { "epoch": 0.3400609223863049, "grad_norm": 1.6813234373111083, "learning_rate": 1.5368007389936965e-05, "loss": 0.6125, "step": 9545 }, { "epoch": 0.34009654951279905, "grad_norm": 1.6965332281691397, "learning_rate": 1.5367033767362787e-05, "loss": 0.6376, "step": 9546 }, { "epoch": 0.3401321766392932, "grad_norm": 2.141968807540382, "learning_rate": 1.5366060073322803e-05, "loss": 0.6297, "step": 9547 }, { "epoch": 0.34016780376578726, "grad_norm": 1.5056947991276968, "learning_rate": 1.536508630782998e-05, "loss": 0.4999, "step": 9548 }, { "epoch": 0.3402034308922814, "grad_norm": 1.0130057846667915, "learning_rate": 1.536411247089728e-05, "loss": 0.4502, "step": 9549 }, { "epoch": 0.3402390580187755, "grad_norm": 1.87260325781699, "learning_rate": 1.536313856253768e-05, "loss": 0.5716, "step": 9550 }, { "epoch": 0.3402746851452696, "grad_norm": 1.2067180139093883, "learning_rate": 1.536216458276413e-05, "loss": 0.4447, "step": 9551 }, { "epoch": 0.3403103122717637, "grad_norm": 1.3995735905270057, "learning_rate": 1.5361190531589615e-05, "loss": 0.5053, "step": 9552 }, { "epoch": 0.34034593939825786, "grad_norm": 1.192933405101763, "learning_rate": 1.5360216409027104e-05, "loss": 0.4789, "step": 9553 }, { "epoch": 0.34038156652475193, "grad_norm": 1.5628471067970295, "learning_rate": 1.5359242215089565e-05, "loss": 0.4498, "step": 9554 }, { "epoch": 0.34041719365124606, "grad_norm": 1.0701824637015744, "learning_rate": 1.5358267949789968e-05, "loss": 0.4326, "step": 9555 }, { "epoch": 0.3404528207777402, "grad_norm": 1.1612894331097865, "learning_rate": 1.535729361314129e-05, "loss": 0.4743, "step": 9556 }, { "epoch": 0.34048844790423427, "grad_norm": 1.3483656035509008, "learning_rate": 1.53563192051565e-05, "loss": 0.6017, "step": 9557 }, { "epoch": 0.3405240750307284, "grad_norm": 1.3522335377420276, "learning_rate": 1.535534472584858e-05, "loss": 0.569, "step": 9558 }, { "epoch": 0.34055970215722253, "grad_norm": 0.9951750863435115, "learning_rate": 1.5354370175230498e-05, "loss": 0.4405, "step": 9559 }, { "epoch": 0.3405953292837166, "grad_norm": 1.056518162176964, "learning_rate": 1.5353395553315234e-05, "loss": 0.4982, "step": 9560 }, { "epoch": 0.34063095641021074, "grad_norm": 1.2085907207880424, "learning_rate": 1.5352420860115768e-05, "loss": 0.5018, "step": 9561 }, { "epoch": 0.34066658353670487, "grad_norm": 1.4686596387338833, "learning_rate": 1.5351446095645076e-05, "loss": 0.5993, "step": 9562 }, { "epoch": 0.34070221066319895, "grad_norm": 1.1650802122447335, "learning_rate": 1.535047125991614e-05, "loss": 0.531, "step": 9563 }, { "epoch": 0.3407378377896931, "grad_norm": 1.464389388042511, "learning_rate": 1.534949635294194e-05, "loss": 0.5206, "step": 9564 }, { "epoch": 0.3407734649161872, "grad_norm": 1.1030077852004216, "learning_rate": 1.5348521374735458e-05, "loss": 0.5038, "step": 9565 }, { "epoch": 0.3408090920426813, "grad_norm": 1.4702878242073725, "learning_rate": 1.5347546325309672e-05, "loss": 0.6324, "step": 9566 }, { "epoch": 0.3408447191691754, "grad_norm": 2.391807102848081, "learning_rate": 1.534657120467757e-05, "loss": 0.5621, "step": 9567 }, { "epoch": 0.34088034629566955, "grad_norm": 1.5252055258386668, "learning_rate": 1.534559601285214e-05, "loss": 0.5913, "step": 9568 }, { "epoch": 0.3409159734221636, "grad_norm": 1.1826476012202505, "learning_rate": 1.534462074984636e-05, "loss": 0.5247, "step": 9569 }, { "epoch": 0.34095160054865775, "grad_norm": 1.915439162053247, "learning_rate": 1.5343645415673216e-05, "loss": 0.6105, "step": 9570 }, { "epoch": 0.3409872276751519, "grad_norm": 1.1108085805755812, "learning_rate": 1.53426700103457e-05, "loss": 0.4361, "step": 9571 }, { "epoch": 0.34102285480164596, "grad_norm": 1.656404829844978, "learning_rate": 1.5341694533876802e-05, "loss": 0.5732, "step": 9572 }, { "epoch": 0.3410584819281401, "grad_norm": 1.0928870628718983, "learning_rate": 1.5340718986279505e-05, "loss": 0.4854, "step": 9573 }, { "epoch": 0.3410941090546342, "grad_norm": 1.2682414616531619, "learning_rate": 1.5339743367566796e-05, "loss": 0.5264, "step": 9574 }, { "epoch": 0.3411297361811283, "grad_norm": 1.1467721461244926, "learning_rate": 1.5338767677751677e-05, "loss": 0.5003, "step": 9575 }, { "epoch": 0.3411653633076224, "grad_norm": 1.7675419049350232, "learning_rate": 1.5337791916847137e-05, "loss": 0.6074, "step": 9576 }, { "epoch": 0.34120099043411656, "grad_norm": 1.6078764527715108, "learning_rate": 1.5336816084866165e-05, "loss": 0.5664, "step": 9577 }, { "epoch": 0.34123661756061063, "grad_norm": 1.114471983367791, "learning_rate": 1.5335840181821757e-05, "loss": 0.4932, "step": 9578 }, { "epoch": 0.34127224468710476, "grad_norm": 1.1944247308668392, "learning_rate": 1.5334864207726908e-05, "loss": 0.5101, "step": 9579 }, { "epoch": 0.3413078718135989, "grad_norm": 1.8757437441203821, "learning_rate": 1.533388816259461e-05, "loss": 0.6349, "step": 9580 }, { "epoch": 0.34134349894009297, "grad_norm": 1.8223179970447787, "learning_rate": 1.5332912046437866e-05, "loss": 0.7033, "step": 9581 }, { "epoch": 0.3413791260665871, "grad_norm": 1.5712556620776592, "learning_rate": 1.533193585926967e-05, "loss": 0.6564, "step": 9582 }, { "epoch": 0.34141475319308123, "grad_norm": 1.1229705669333536, "learning_rate": 1.533095960110302e-05, "loss": 0.413, "step": 9583 }, { "epoch": 0.3414503803195753, "grad_norm": 1.1110733064118354, "learning_rate": 1.5329983271950916e-05, "loss": 0.5161, "step": 9584 }, { "epoch": 0.34148600744606944, "grad_norm": 1.0575344079407938, "learning_rate": 1.5329006871826365e-05, "loss": 0.4334, "step": 9585 }, { "epoch": 0.34152163457256357, "grad_norm": 1.6152671393801297, "learning_rate": 1.5328030400742355e-05, "loss": 0.5449, "step": 9586 }, { "epoch": 0.34155726169905765, "grad_norm": 1.1121773867145752, "learning_rate": 1.53270538587119e-05, "loss": 0.4701, "step": 9587 }, { "epoch": 0.3415928888255518, "grad_norm": 1.1880669072910894, "learning_rate": 1.5326077245747998e-05, "loss": 0.5085, "step": 9588 }, { "epoch": 0.3416285159520459, "grad_norm": 1.0411432197217863, "learning_rate": 1.5325100561863656e-05, "loss": 0.4854, "step": 9589 }, { "epoch": 0.34166414307854, "grad_norm": 1.42405855081221, "learning_rate": 1.532412380707188e-05, "loss": 0.625, "step": 9590 }, { "epoch": 0.3416997702050341, "grad_norm": 1.0438656925348149, "learning_rate": 1.5323146981385676e-05, "loss": 0.4671, "step": 9591 }, { "epoch": 0.34173539733152825, "grad_norm": 1.0037641765420466, "learning_rate": 1.532217008481804e-05, "loss": 0.478, "step": 9592 }, { "epoch": 0.3417710244580223, "grad_norm": 0.9866885586726208, "learning_rate": 1.5321193117381998e-05, "loss": 0.5107, "step": 9593 }, { "epoch": 0.34180665158451645, "grad_norm": 1.4991609926346248, "learning_rate": 1.5320216079090546e-05, "loss": 0.6352, "step": 9594 }, { "epoch": 0.3418422787110106, "grad_norm": 1.700173367050247, "learning_rate": 1.5319238969956697e-05, "loss": 0.6076, "step": 9595 }, { "epoch": 0.34187790583750466, "grad_norm": 1.107189455968059, "learning_rate": 1.5318261789993466e-05, "loss": 0.4844, "step": 9596 }, { "epoch": 0.3419135329639988, "grad_norm": 2.6473572662332323, "learning_rate": 1.5317284539213862e-05, "loss": 0.5714, "step": 9597 }, { "epoch": 0.3419491600904929, "grad_norm": 1.3801876227924705, "learning_rate": 1.53163072176309e-05, "loss": 0.6137, "step": 9598 }, { "epoch": 0.341984787216987, "grad_norm": 1.562654928939778, "learning_rate": 1.531532982525759e-05, "loss": 0.4778, "step": 9599 }, { "epoch": 0.3420204143434811, "grad_norm": 1.0917422789238813, "learning_rate": 1.5314352362106943e-05, "loss": 0.4927, "step": 9600 }, { "epoch": 0.34205604146997526, "grad_norm": 1.4927594886303004, "learning_rate": 1.5313374828191987e-05, "loss": 0.5858, "step": 9601 }, { "epoch": 0.34209166859646933, "grad_norm": 1.394699950950735, "learning_rate": 1.5312397223525728e-05, "loss": 0.5823, "step": 9602 }, { "epoch": 0.34212729572296346, "grad_norm": 1.822899613648307, "learning_rate": 1.5311419548121186e-05, "loss": 0.583, "step": 9603 }, { "epoch": 0.3421629228494576, "grad_norm": 1.471172376722361, "learning_rate": 1.531044180199138e-05, "loss": 0.5389, "step": 9604 }, { "epoch": 0.34219854997595167, "grad_norm": 1.5466598532365243, "learning_rate": 1.5309463985149332e-05, "loss": 0.5796, "step": 9605 }, { "epoch": 0.3422341771024458, "grad_norm": 1.1199100037470342, "learning_rate": 1.530848609760806e-05, "loss": 0.4804, "step": 9606 }, { "epoch": 0.34226980422893993, "grad_norm": 2.201890277785868, "learning_rate": 1.5307508139380583e-05, "loss": 0.5595, "step": 9607 }, { "epoch": 0.342305431355434, "grad_norm": 1.3133549402996174, "learning_rate": 1.530653011047993e-05, "loss": 0.5083, "step": 9608 }, { "epoch": 0.34234105848192814, "grad_norm": 1.095711641061403, "learning_rate": 1.5305552010919115e-05, "loss": 0.4852, "step": 9609 }, { "epoch": 0.34237668560842227, "grad_norm": 1.1517390755970354, "learning_rate": 1.5304573840711166e-05, "loss": 0.5118, "step": 9610 }, { "epoch": 0.34241231273491635, "grad_norm": 1.139510647942341, "learning_rate": 1.530359559986911e-05, "loss": 0.4536, "step": 9611 }, { "epoch": 0.3424479398614105, "grad_norm": 1.131431410422578, "learning_rate": 1.5302617288405972e-05, "loss": 0.478, "step": 9612 }, { "epoch": 0.3424835669879046, "grad_norm": 1.6031996796693866, "learning_rate": 1.530163890633478e-05, "loss": 0.6025, "step": 9613 }, { "epoch": 0.3425191941143987, "grad_norm": 1.4029845815749968, "learning_rate": 1.5300660453668557e-05, "loss": 0.5965, "step": 9614 }, { "epoch": 0.3425548212408928, "grad_norm": 1.0973960141010548, "learning_rate": 1.529968193042034e-05, "loss": 0.4859, "step": 9615 }, { "epoch": 0.34259044836738695, "grad_norm": 1.0221309314105398, "learning_rate": 1.529870333660315e-05, "loss": 0.4523, "step": 9616 }, { "epoch": 0.342626075493881, "grad_norm": 1.1930270590607004, "learning_rate": 1.5297724672230023e-05, "loss": 0.4517, "step": 9617 }, { "epoch": 0.34266170262037515, "grad_norm": 1.1317166199983897, "learning_rate": 1.529674593731399e-05, "loss": 0.5254, "step": 9618 }, { "epoch": 0.3426973297468693, "grad_norm": 1.0750977622327222, "learning_rate": 1.529576713186808e-05, "loss": 0.4447, "step": 9619 }, { "epoch": 0.34273295687336336, "grad_norm": 1.1287624169786983, "learning_rate": 1.5294788255905328e-05, "loss": 0.4679, "step": 9620 }, { "epoch": 0.3427685839998575, "grad_norm": 0.9552401633392746, "learning_rate": 1.529380930943877e-05, "loss": 0.4379, "step": 9621 }, { "epoch": 0.3428042111263516, "grad_norm": 1.0989770071016582, "learning_rate": 1.5292830292481445e-05, "loss": 0.4873, "step": 9622 }, { "epoch": 0.3428398382528457, "grad_norm": 1.0259523742551662, "learning_rate": 1.5291851205046383e-05, "loss": 0.4935, "step": 9623 }, { "epoch": 0.3428754653793398, "grad_norm": 1.9218245251169472, "learning_rate": 1.5290872047146623e-05, "loss": 0.6457, "step": 9624 }, { "epoch": 0.34291109250583396, "grad_norm": 1.2357756329150578, "learning_rate": 1.5289892818795205e-05, "loss": 0.566, "step": 9625 }, { "epoch": 0.34294671963232803, "grad_norm": 1.5683381719469518, "learning_rate": 1.5288913520005162e-05, "loss": 0.6269, "step": 9626 }, { "epoch": 0.34298234675882217, "grad_norm": 1.461447443898742, "learning_rate": 1.5287934150789543e-05, "loss": 0.5874, "step": 9627 }, { "epoch": 0.3430179738853163, "grad_norm": 1.1309444629555685, "learning_rate": 1.5286954711161387e-05, "loss": 0.4918, "step": 9628 }, { "epoch": 0.34305360101181037, "grad_norm": 1.23557288790402, "learning_rate": 1.528597520113373e-05, "loss": 0.4602, "step": 9629 }, { "epoch": 0.3430892281383045, "grad_norm": 1.4438403836432037, "learning_rate": 1.5284995620719618e-05, "loss": 0.6254, "step": 9630 }, { "epoch": 0.34312485526479863, "grad_norm": 1.4279837605898196, "learning_rate": 1.5284015969932097e-05, "loss": 0.5964, "step": 9631 }, { "epoch": 0.3431604823912927, "grad_norm": 1.5654515112807816, "learning_rate": 1.528303624878421e-05, "loss": 0.66, "step": 9632 }, { "epoch": 0.34319610951778684, "grad_norm": 1.1235563673676352, "learning_rate": 1.5282056457289002e-05, "loss": 0.5284, "step": 9633 }, { "epoch": 0.34323173664428097, "grad_norm": 2.268259357442333, "learning_rate": 1.528107659545952e-05, "loss": 0.6762, "step": 9634 }, { "epoch": 0.34326736377077505, "grad_norm": 1.0876417029399226, "learning_rate": 1.5280096663308813e-05, "loss": 0.4645, "step": 9635 }, { "epoch": 0.3433029908972692, "grad_norm": 1.124984011083996, "learning_rate": 1.5279116660849926e-05, "loss": 0.4833, "step": 9636 }, { "epoch": 0.3433386180237633, "grad_norm": 1.1316554302414883, "learning_rate": 1.5278136588095914e-05, "loss": 0.4657, "step": 9637 }, { "epoch": 0.3433742451502574, "grad_norm": 1.5627804919238053, "learning_rate": 1.527715644505982e-05, "loss": 0.5952, "step": 9638 }, { "epoch": 0.3434098722767515, "grad_norm": 1.1452965106634299, "learning_rate": 1.5276176231754705e-05, "loss": 0.4955, "step": 9639 }, { "epoch": 0.34344549940324565, "grad_norm": 1.2423649168365751, "learning_rate": 1.5275195948193613e-05, "loss": 0.4479, "step": 9640 }, { "epoch": 0.3434811265297397, "grad_norm": 1.4819330672631197, "learning_rate": 1.52742155943896e-05, "loss": 0.4537, "step": 9641 }, { "epoch": 0.34351675365623385, "grad_norm": 1.1606280652926544, "learning_rate": 1.5273235170355717e-05, "loss": 0.4812, "step": 9642 }, { "epoch": 0.343552380782728, "grad_norm": 1.1905866206371716, "learning_rate": 1.5272254676105026e-05, "loss": 0.5293, "step": 9643 }, { "epoch": 0.34358800790922206, "grad_norm": 1.085333362482474, "learning_rate": 1.5271274111650577e-05, "loss": 0.5567, "step": 9644 }, { "epoch": 0.3436236350357162, "grad_norm": 1.0892222406112784, "learning_rate": 1.5270293477005428e-05, "loss": 0.519, "step": 9645 }, { "epoch": 0.3436592621622103, "grad_norm": 1.5240612976033414, "learning_rate": 1.5269312772182637e-05, "loss": 0.5854, "step": 9646 }, { "epoch": 0.3436948892887044, "grad_norm": 2.048391720272699, "learning_rate": 1.526833199719527e-05, "loss": 0.6001, "step": 9647 }, { "epoch": 0.34373051641519853, "grad_norm": 1.163273495378544, "learning_rate": 1.5267351152056375e-05, "loss": 0.4935, "step": 9648 }, { "epoch": 0.34376614354169266, "grad_norm": 1.5388273805782153, "learning_rate": 1.5266370236779015e-05, "loss": 0.5683, "step": 9649 }, { "epoch": 0.34380177066818673, "grad_norm": 1.3651619299925164, "learning_rate": 1.526538925137626e-05, "loss": 0.4645, "step": 9650 }, { "epoch": 0.34383739779468087, "grad_norm": 1.2834369319947876, "learning_rate": 1.5264408195861164e-05, "loss": 0.4681, "step": 9651 }, { "epoch": 0.343873024921175, "grad_norm": 1.250964167885225, "learning_rate": 1.526342707024679e-05, "loss": 0.4937, "step": 9652 }, { "epoch": 0.3439086520476691, "grad_norm": 1.861735098461, "learning_rate": 1.526244587454621e-05, "loss": 0.5916, "step": 9653 }, { "epoch": 0.3439442791741632, "grad_norm": 1.1699709512981817, "learning_rate": 1.5261464608772487e-05, "loss": 0.5066, "step": 9654 }, { "epoch": 0.34397990630065733, "grad_norm": 2.0005597988978057, "learning_rate": 1.526048327293868e-05, "loss": 0.6166, "step": 9655 }, { "epoch": 0.3440155334271514, "grad_norm": 1.7608218994268545, "learning_rate": 1.5259501867057864e-05, "loss": 0.6161, "step": 9656 }, { "epoch": 0.34405116055364554, "grad_norm": 2.8774569011254716, "learning_rate": 1.5258520391143106e-05, "loss": 0.5202, "step": 9657 }, { "epoch": 0.34408678768013967, "grad_norm": 1.0915069679170057, "learning_rate": 1.525753884520747e-05, "loss": 0.4899, "step": 9658 }, { "epoch": 0.34412241480663375, "grad_norm": 1.33229780555359, "learning_rate": 1.525655722926403e-05, "loss": 0.559, "step": 9659 }, { "epoch": 0.3441580419331279, "grad_norm": 1.4073527501942715, "learning_rate": 1.5255575543325859e-05, "loss": 0.6003, "step": 9660 }, { "epoch": 0.344193669059622, "grad_norm": 1.3523324832880563, "learning_rate": 1.5254593787406026e-05, "loss": 0.4832, "step": 9661 }, { "epoch": 0.3442292961861161, "grad_norm": 1.0994200267282566, "learning_rate": 1.5253611961517603e-05, "loss": 0.5165, "step": 9662 }, { "epoch": 0.3442649233126102, "grad_norm": 1.9034974395248472, "learning_rate": 1.5252630065673662e-05, "loss": 0.657, "step": 9663 }, { "epoch": 0.34430055043910435, "grad_norm": 1.4963490260582806, "learning_rate": 1.5251648099887285e-05, "loss": 0.6706, "step": 9664 }, { "epoch": 0.3443361775655984, "grad_norm": 1.1465561768709251, "learning_rate": 1.525066606417154e-05, "loss": 0.4477, "step": 9665 }, { "epoch": 0.34437180469209255, "grad_norm": 1.1632678268303571, "learning_rate": 1.5249683958539504e-05, "loss": 0.4922, "step": 9666 }, { "epoch": 0.3444074318185867, "grad_norm": 1.1874559234951094, "learning_rate": 1.5248701783004261e-05, "loss": 0.4973, "step": 9667 }, { "epoch": 0.34444305894508076, "grad_norm": 1.2060024934962432, "learning_rate": 1.5247719537578882e-05, "loss": 0.5232, "step": 9668 }, { "epoch": 0.3444786860715749, "grad_norm": 1.5276870551681552, "learning_rate": 1.5246737222276454e-05, "loss": 0.5901, "step": 9669 }, { "epoch": 0.344514313198069, "grad_norm": 1.1896300453015154, "learning_rate": 1.5245754837110048e-05, "loss": 0.547, "step": 9670 }, { "epoch": 0.3445499403245631, "grad_norm": 1.0683024198670976, "learning_rate": 1.5244772382092751e-05, "loss": 0.4923, "step": 9671 }, { "epoch": 0.34458556745105723, "grad_norm": 1.1211608705483627, "learning_rate": 1.5243789857237646e-05, "loss": 0.4881, "step": 9672 }, { "epoch": 0.34462119457755136, "grad_norm": 1.0458769336852811, "learning_rate": 1.5242807262557815e-05, "loss": 0.4289, "step": 9673 }, { "epoch": 0.34465682170404544, "grad_norm": 1.0872386868902728, "learning_rate": 1.5241824598066335e-05, "loss": 0.5444, "step": 9674 }, { "epoch": 0.34469244883053957, "grad_norm": 1.1118700556752947, "learning_rate": 1.52408418637763e-05, "loss": 0.4556, "step": 9675 }, { "epoch": 0.3447280759570337, "grad_norm": 1.686999930003975, "learning_rate": 1.5239859059700794e-05, "loss": 0.5601, "step": 9676 }, { "epoch": 0.3447637030835278, "grad_norm": 1.5615462231948338, "learning_rate": 1.52388761858529e-05, "loss": 0.5562, "step": 9677 }, { "epoch": 0.3447993302100219, "grad_norm": 1.1231759833598032, "learning_rate": 1.5237893242245707e-05, "loss": 0.4965, "step": 9678 }, { "epoch": 0.34483495733651603, "grad_norm": 2.5199251862567715, "learning_rate": 1.5236910228892306e-05, "loss": 0.5782, "step": 9679 }, { "epoch": 0.3448705844630101, "grad_norm": 1.4849027263785874, "learning_rate": 1.5235927145805785e-05, "loss": 0.6567, "step": 9680 }, { "epoch": 0.34490621158950424, "grad_norm": 1.7715952928104877, "learning_rate": 1.5234943992999235e-05, "loss": 0.5516, "step": 9681 }, { "epoch": 0.3449418387159984, "grad_norm": 1.269094394555293, "learning_rate": 1.5233960770485742e-05, "loss": 0.4511, "step": 9682 }, { "epoch": 0.34497746584249245, "grad_norm": 1.5359509967820622, "learning_rate": 1.523297747827841e-05, "loss": 0.6488, "step": 9683 }, { "epoch": 0.3450130929689866, "grad_norm": 1.1622052016151254, "learning_rate": 1.5231994116390322e-05, "loss": 0.5514, "step": 9684 }, { "epoch": 0.3450487200954807, "grad_norm": 1.1193375746906786, "learning_rate": 1.5231010684834574e-05, "loss": 0.5056, "step": 9685 }, { "epoch": 0.3450843472219748, "grad_norm": 1.556807958471239, "learning_rate": 1.5230027183624265e-05, "loss": 0.4993, "step": 9686 }, { "epoch": 0.3451199743484689, "grad_norm": 1.3948650725367682, "learning_rate": 1.5229043612772488e-05, "loss": 0.484, "step": 9687 }, { "epoch": 0.34515560147496305, "grad_norm": 1.577037330026236, "learning_rate": 1.5228059972292339e-05, "loss": 0.5827, "step": 9688 }, { "epoch": 0.3451912286014571, "grad_norm": 1.2438314305263143, "learning_rate": 1.5227076262196917e-05, "loss": 0.4697, "step": 9689 }, { "epoch": 0.34522685572795125, "grad_norm": 1.1968510521283708, "learning_rate": 1.5226092482499321e-05, "loss": 0.4978, "step": 9690 }, { "epoch": 0.3452624828544454, "grad_norm": 1.4179394730306172, "learning_rate": 1.5225108633212652e-05, "loss": 0.5697, "step": 9691 }, { "epoch": 0.34529810998093946, "grad_norm": 1.8014977434934627, "learning_rate": 1.5224124714350011e-05, "loss": 0.5661, "step": 9692 }, { "epoch": 0.3453337371074336, "grad_norm": 2.0920002665447814, "learning_rate": 1.5223140725924494e-05, "loss": 0.6088, "step": 9693 }, { "epoch": 0.3453693642339277, "grad_norm": 1.409628723254422, "learning_rate": 1.5222156667949212e-05, "loss": 0.5904, "step": 9694 }, { "epoch": 0.3454049913604218, "grad_norm": 1.206413446816192, "learning_rate": 1.522117254043726e-05, "loss": 0.4805, "step": 9695 }, { "epoch": 0.34544061848691593, "grad_norm": 1.184793166226743, "learning_rate": 1.5220188343401749e-05, "loss": 0.4838, "step": 9696 }, { "epoch": 0.34547624561341006, "grad_norm": 1.1023135883807256, "learning_rate": 1.5219204076855781e-05, "loss": 0.4775, "step": 9697 }, { "epoch": 0.34551187273990414, "grad_norm": 1.3086892494641487, "learning_rate": 1.521821974081246e-05, "loss": 0.5196, "step": 9698 }, { "epoch": 0.34554749986639827, "grad_norm": 1.560841119692777, "learning_rate": 1.5217235335284902e-05, "loss": 0.5485, "step": 9699 }, { "epoch": 0.3455831269928924, "grad_norm": 1.142097601501428, "learning_rate": 1.5216250860286201e-05, "loss": 0.4861, "step": 9700 }, { "epoch": 0.3456187541193865, "grad_norm": 2.0291031386448872, "learning_rate": 1.5215266315829477e-05, "loss": 0.573, "step": 9701 }, { "epoch": 0.3456543812458806, "grad_norm": 1.378291941238473, "learning_rate": 1.5214281701927837e-05, "loss": 0.5427, "step": 9702 }, { "epoch": 0.34569000837237474, "grad_norm": 1.6753127353431625, "learning_rate": 1.5213297018594393e-05, "loss": 0.6258, "step": 9703 }, { "epoch": 0.34572563549886887, "grad_norm": 1.5296175289936853, "learning_rate": 1.5212312265842256e-05, "loss": 0.5664, "step": 9704 }, { "epoch": 0.34576126262536294, "grad_norm": 2.716078575987966, "learning_rate": 1.5211327443684533e-05, "loss": 0.5653, "step": 9705 }, { "epoch": 0.3457968897518571, "grad_norm": 1.7142940960561184, "learning_rate": 1.5210342552134348e-05, "loss": 0.636, "step": 9706 }, { "epoch": 0.3458325168783512, "grad_norm": 1.070360129382918, "learning_rate": 1.520935759120481e-05, "loss": 0.4647, "step": 9707 }, { "epoch": 0.3458681440048453, "grad_norm": 2.003153422528574, "learning_rate": 1.5208372560909031e-05, "loss": 0.5238, "step": 9708 }, { "epoch": 0.3459037711313394, "grad_norm": 2.3062527110460307, "learning_rate": 1.5207387461260132e-05, "loss": 0.6056, "step": 9709 }, { "epoch": 0.34593939825783354, "grad_norm": 1.1907540323211283, "learning_rate": 1.520640229227123e-05, "loss": 0.4676, "step": 9710 }, { "epoch": 0.3459750253843276, "grad_norm": 2.340038809326718, "learning_rate": 1.520541705395544e-05, "loss": 0.4891, "step": 9711 }, { "epoch": 0.34601065251082175, "grad_norm": 1.6903082651797876, "learning_rate": 1.5204431746325888e-05, "loss": 0.586, "step": 9712 }, { "epoch": 0.3460462796373159, "grad_norm": 1.1106601014397912, "learning_rate": 1.5203446369395685e-05, "loss": 0.4702, "step": 9713 }, { "epoch": 0.34608190676380995, "grad_norm": 1.1132484006555285, "learning_rate": 1.520246092317796e-05, "loss": 0.4402, "step": 9714 }, { "epoch": 0.3461175338903041, "grad_norm": 1.175663081607254, "learning_rate": 1.5201475407685831e-05, "loss": 0.4931, "step": 9715 }, { "epoch": 0.3461531610167982, "grad_norm": 1.5883717368293777, "learning_rate": 1.520048982293242e-05, "loss": 0.5574, "step": 9716 }, { "epoch": 0.3461887881432923, "grad_norm": 1.6571148666984956, "learning_rate": 1.5199504168930852e-05, "loss": 0.5801, "step": 9717 }, { "epoch": 0.3462244152697864, "grad_norm": 1.4376339858711873, "learning_rate": 1.5198518445694255e-05, "loss": 0.5324, "step": 9718 }, { "epoch": 0.34626004239628055, "grad_norm": 1.18924647589192, "learning_rate": 1.5197532653235747e-05, "loss": 0.4638, "step": 9719 }, { "epoch": 0.34629566952277463, "grad_norm": 1.1145219801510056, "learning_rate": 1.5196546791568461e-05, "loss": 0.5038, "step": 9720 }, { "epoch": 0.34633129664926876, "grad_norm": 1.8238524364637752, "learning_rate": 1.5195560860705522e-05, "loss": 0.5647, "step": 9721 }, { "epoch": 0.3463669237757629, "grad_norm": 1.5580586960092195, "learning_rate": 1.5194574860660058e-05, "loss": 0.5464, "step": 9722 }, { "epoch": 0.34640255090225697, "grad_norm": 1.9859621695734058, "learning_rate": 1.5193588791445199e-05, "loss": 0.6351, "step": 9723 }, { "epoch": 0.3464381780287511, "grad_norm": 1.4245070876137076, "learning_rate": 1.5192602653074077e-05, "loss": 0.6523, "step": 9724 }, { "epoch": 0.34647380515524523, "grad_norm": 1.5396538586854078, "learning_rate": 1.519161644555982e-05, "loss": 0.5483, "step": 9725 }, { "epoch": 0.3465094322817393, "grad_norm": 1.2020747689508922, "learning_rate": 1.5190630168915563e-05, "loss": 0.5467, "step": 9726 }, { "epoch": 0.34654505940823344, "grad_norm": 1.1649829013125341, "learning_rate": 1.5189643823154435e-05, "loss": 0.4823, "step": 9727 }, { "epoch": 0.34658068653472757, "grad_norm": 1.434726245674618, "learning_rate": 1.5188657408289574e-05, "loss": 0.586, "step": 9728 }, { "epoch": 0.34661631366122164, "grad_norm": 1.2709045071992418, "learning_rate": 1.518767092433411e-05, "loss": 0.5469, "step": 9729 }, { "epoch": 0.3466519407877158, "grad_norm": 1.135802283438799, "learning_rate": 1.5186684371301183e-05, "loss": 0.5314, "step": 9730 }, { "epoch": 0.3466875679142099, "grad_norm": 1.7807813868570732, "learning_rate": 1.5185697749203928e-05, "loss": 0.5747, "step": 9731 }, { "epoch": 0.346723195040704, "grad_norm": 2.3292919707830526, "learning_rate": 1.5184711058055483e-05, "loss": 0.5046, "step": 9732 }, { "epoch": 0.3467588221671981, "grad_norm": 1.1850681186398517, "learning_rate": 1.518372429786899e-05, "loss": 0.5119, "step": 9733 }, { "epoch": 0.34679444929369224, "grad_norm": 1.5495316279849063, "learning_rate": 1.518273746865758e-05, "loss": 0.5539, "step": 9734 }, { "epoch": 0.3468300764201863, "grad_norm": 1.5336517144376633, "learning_rate": 1.5181750570434402e-05, "loss": 0.6606, "step": 9735 }, { "epoch": 0.34686570354668045, "grad_norm": 1.2764223064053404, "learning_rate": 1.518076360321259e-05, "loss": 0.4882, "step": 9736 }, { "epoch": 0.3469013306731746, "grad_norm": 1.6583524311247746, "learning_rate": 1.5179776567005289e-05, "loss": 0.5982, "step": 9737 }, { "epoch": 0.34693695779966865, "grad_norm": 1.9436741714397037, "learning_rate": 1.5178789461825645e-05, "loss": 0.5248, "step": 9738 }, { "epoch": 0.3469725849261628, "grad_norm": 1.797860654828899, "learning_rate": 1.5177802287686798e-05, "loss": 0.6063, "step": 9739 }, { "epoch": 0.3470082120526569, "grad_norm": 1.4613690923368285, "learning_rate": 1.5176815044601897e-05, "loss": 0.6208, "step": 9740 }, { "epoch": 0.347043839179151, "grad_norm": 1.830874571418283, "learning_rate": 1.5175827732584082e-05, "loss": 0.5636, "step": 9741 }, { "epoch": 0.3470794663056451, "grad_norm": 1.7300050411847718, "learning_rate": 1.5174840351646504e-05, "loss": 0.6238, "step": 9742 }, { "epoch": 0.34711509343213925, "grad_norm": 1.1418022992869548, "learning_rate": 1.517385290180231e-05, "loss": 0.5071, "step": 9743 }, { "epoch": 0.34715072055863333, "grad_norm": 1.7572555143850193, "learning_rate": 1.5172865383064653e-05, "loss": 0.6941, "step": 9744 }, { "epoch": 0.34718634768512746, "grad_norm": 2.9143506037247735, "learning_rate": 1.517187779544667e-05, "loss": 0.589, "step": 9745 }, { "epoch": 0.3472219748116216, "grad_norm": 2.1785537664726173, "learning_rate": 1.5170890138961523e-05, "loss": 0.6197, "step": 9746 }, { "epoch": 0.34725760193811567, "grad_norm": 1.5865203459443924, "learning_rate": 1.5169902413622359e-05, "loss": 0.5652, "step": 9747 }, { "epoch": 0.3472932290646098, "grad_norm": 1.3381766950383314, "learning_rate": 1.516891461944233e-05, "loss": 0.5288, "step": 9748 }, { "epoch": 0.34732885619110393, "grad_norm": 1.1821647166611318, "learning_rate": 1.516792675643459e-05, "loss": 0.4834, "step": 9749 }, { "epoch": 0.347364483317598, "grad_norm": 1.1747456911384067, "learning_rate": 1.5166938824612296e-05, "loss": 0.453, "step": 9750 }, { "epoch": 0.34740011044409214, "grad_norm": 1.949683795624022, "learning_rate": 1.5165950823988598e-05, "loss": 0.5732, "step": 9751 }, { "epoch": 0.34743573757058627, "grad_norm": 1.5820136020498714, "learning_rate": 1.5164962754576654e-05, "loss": 0.5079, "step": 9752 }, { "epoch": 0.34747136469708034, "grad_norm": 1.179705731815534, "learning_rate": 1.5163974616389621e-05, "loss": 0.4967, "step": 9753 }, { "epoch": 0.3475069918235745, "grad_norm": 2.0084164050597617, "learning_rate": 1.5162986409440657e-05, "loss": 0.5941, "step": 9754 }, { "epoch": 0.3475426189500686, "grad_norm": 1.2227774725409737, "learning_rate": 1.5161998133742919e-05, "loss": 0.5503, "step": 9755 }, { "epoch": 0.3475782460765627, "grad_norm": 1.1945616620443313, "learning_rate": 1.5161009789309569e-05, "loss": 0.4245, "step": 9756 }, { "epoch": 0.3476138732030568, "grad_norm": 2.0813634985813394, "learning_rate": 1.5160021376153765e-05, "loss": 0.5445, "step": 9757 }, { "epoch": 0.34764950032955094, "grad_norm": 1.2556523799744248, "learning_rate": 1.5159032894288672e-05, "loss": 0.4889, "step": 9758 }, { "epoch": 0.347685127456045, "grad_norm": 1.7333532567151368, "learning_rate": 1.5158044343727447e-05, "loss": 0.5967, "step": 9759 }, { "epoch": 0.34772075458253915, "grad_norm": 1.3737299865482777, "learning_rate": 1.5157055724483258e-05, "loss": 0.5316, "step": 9760 }, { "epoch": 0.3477563817090333, "grad_norm": 2.093338024223977, "learning_rate": 1.5156067036569267e-05, "loss": 0.6025, "step": 9761 }, { "epoch": 0.34779200883552736, "grad_norm": 1.4215468767877222, "learning_rate": 1.5155078279998643e-05, "loss": 0.484, "step": 9762 }, { "epoch": 0.3478276359620215, "grad_norm": 1.3281711501934026, "learning_rate": 1.5154089454784547e-05, "loss": 0.4661, "step": 9763 }, { "epoch": 0.3478632630885156, "grad_norm": 1.4730192821445265, "learning_rate": 1.5153100560940143e-05, "loss": 0.5993, "step": 9764 }, { "epoch": 0.3478988902150097, "grad_norm": 1.303184092584823, "learning_rate": 1.5152111598478607e-05, "loss": 0.5152, "step": 9765 }, { "epoch": 0.3479345173415038, "grad_norm": 1.50361002907326, "learning_rate": 1.5151122567413104e-05, "loss": 0.5089, "step": 9766 }, { "epoch": 0.34797014446799795, "grad_norm": 1.9636689863575296, "learning_rate": 1.51501334677568e-05, "loss": 0.6319, "step": 9767 }, { "epoch": 0.34800577159449203, "grad_norm": 1.6127918694553485, "learning_rate": 1.5149144299522874e-05, "loss": 0.6303, "step": 9768 }, { "epoch": 0.34804139872098616, "grad_norm": 2.056624560461287, "learning_rate": 1.514815506272449e-05, "loss": 0.6149, "step": 9769 }, { "epoch": 0.3480770258474803, "grad_norm": 1.4857366867578847, "learning_rate": 1.5147165757374822e-05, "loss": 0.5003, "step": 9770 }, { "epoch": 0.34811265297397437, "grad_norm": 1.3660430183098962, "learning_rate": 1.5146176383487045e-05, "loss": 0.4649, "step": 9771 }, { "epoch": 0.3481482801004685, "grad_norm": 1.9398294712167214, "learning_rate": 1.5145186941074335e-05, "loss": 0.6004, "step": 9772 }, { "epoch": 0.34818390722696263, "grad_norm": 1.7859602902267004, "learning_rate": 1.5144197430149863e-05, "loss": 0.5595, "step": 9773 }, { "epoch": 0.3482195343534567, "grad_norm": 1.4764255087799798, "learning_rate": 1.5143207850726809e-05, "loss": 0.5922, "step": 9774 }, { "epoch": 0.34825516147995084, "grad_norm": 1.436679186488773, "learning_rate": 1.5142218202818342e-05, "loss": 0.5393, "step": 9775 }, { "epoch": 0.34829078860644497, "grad_norm": 1.750022000155295, "learning_rate": 1.5141228486437649e-05, "loss": 0.4653, "step": 9776 }, { "epoch": 0.34832641573293904, "grad_norm": 1.3071247876767398, "learning_rate": 1.5140238701597905e-05, "loss": 0.4861, "step": 9777 }, { "epoch": 0.3483620428594332, "grad_norm": 1.207161584237288, "learning_rate": 1.513924884831229e-05, "loss": 0.5296, "step": 9778 }, { "epoch": 0.3483976699859273, "grad_norm": 1.2501415296366194, "learning_rate": 1.5138258926593983e-05, "loss": 0.5685, "step": 9779 }, { "epoch": 0.3484332971124214, "grad_norm": 1.188375641094788, "learning_rate": 1.513726893645617e-05, "loss": 0.4777, "step": 9780 }, { "epoch": 0.3484689242389155, "grad_norm": 1.2977993840787425, "learning_rate": 1.5136278877912029e-05, "loss": 0.4617, "step": 9781 }, { "epoch": 0.34850455136540964, "grad_norm": 1.184981383866009, "learning_rate": 1.5135288750974742e-05, "loss": 0.4816, "step": 9782 }, { "epoch": 0.3485401784919037, "grad_norm": 1.106442730747005, "learning_rate": 1.51342985556575e-05, "loss": 0.5124, "step": 9783 }, { "epoch": 0.34857580561839785, "grad_norm": 1.020195194999237, "learning_rate": 1.5133308291973482e-05, "loss": 0.5031, "step": 9784 }, { "epoch": 0.348611432744892, "grad_norm": 1.5246451016252416, "learning_rate": 1.5132317959935874e-05, "loss": 0.5554, "step": 9785 }, { "epoch": 0.34864705987138606, "grad_norm": 1.5702213383175125, "learning_rate": 1.5131327559557867e-05, "loss": 0.5958, "step": 9786 }, { "epoch": 0.3486826869978802, "grad_norm": 1.3655463059558566, "learning_rate": 1.513033709085265e-05, "loss": 0.5493, "step": 9787 }, { "epoch": 0.3487183141243743, "grad_norm": 1.0706772198592083, "learning_rate": 1.5129346553833406e-05, "loss": 0.5026, "step": 9788 }, { "epoch": 0.3487539412508684, "grad_norm": 1.1901462737065374, "learning_rate": 1.5128355948513326e-05, "loss": 0.5191, "step": 9789 }, { "epoch": 0.3487895683773625, "grad_norm": 1.161157199234404, "learning_rate": 1.5127365274905603e-05, "loss": 0.4442, "step": 9790 }, { "epoch": 0.34882519550385666, "grad_norm": 1.0219276378931792, "learning_rate": 1.512637453302343e-05, "loss": 0.4103, "step": 9791 }, { "epoch": 0.34886082263035073, "grad_norm": 1.442582749432889, "learning_rate": 1.5125383722879994e-05, "loss": 0.6926, "step": 9792 }, { "epoch": 0.34889644975684486, "grad_norm": 1.2390750939393977, "learning_rate": 1.5124392844488491e-05, "loss": 0.5472, "step": 9793 }, { "epoch": 0.348932076883339, "grad_norm": 1.042127992236373, "learning_rate": 1.5123401897862118e-05, "loss": 0.463, "step": 9794 }, { "epoch": 0.34896770400983307, "grad_norm": 1.177062616167476, "learning_rate": 1.5122410883014067e-05, "loss": 0.5452, "step": 9795 }, { "epoch": 0.3490033311363272, "grad_norm": 1.0980315239312717, "learning_rate": 1.5121419799957534e-05, "loss": 0.4798, "step": 9796 }, { "epoch": 0.34903895826282133, "grad_norm": 1.0882925275466304, "learning_rate": 1.5120428648705716e-05, "loss": 0.4945, "step": 9797 }, { "epoch": 0.3490745853893154, "grad_norm": 1.216692237220578, "learning_rate": 1.5119437429271813e-05, "loss": 0.4971, "step": 9798 }, { "epoch": 0.34911021251580954, "grad_norm": 1.2016681474086723, "learning_rate": 1.5118446141669023e-05, "loss": 0.5303, "step": 9799 }, { "epoch": 0.34914583964230367, "grad_norm": 1.7114832580734045, "learning_rate": 1.5117454785910543e-05, "loss": 0.6148, "step": 9800 }, { "epoch": 0.34918146676879774, "grad_norm": 1.10835181908264, "learning_rate": 1.5116463362009576e-05, "loss": 0.5176, "step": 9801 }, { "epoch": 0.3492170938952919, "grad_norm": 1.5677586464764182, "learning_rate": 1.5115471869979326e-05, "loss": 0.5195, "step": 9802 }, { "epoch": 0.349252721021786, "grad_norm": 1.6215375089291264, "learning_rate": 1.5114480309832991e-05, "loss": 0.6387, "step": 9803 }, { "epoch": 0.3492883481482801, "grad_norm": 1.6950130720253078, "learning_rate": 1.5113488681583773e-05, "loss": 0.5501, "step": 9804 }, { "epoch": 0.3493239752747742, "grad_norm": 1.837013366921646, "learning_rate": 1.511249698524488e-05, "loss": 0.6304, "step": 9805 }, { "epoch": 0.34935960240126834, "grad_norm": 1.798181805901916, "learning_rate": 1.5111505220829519e-05, "loss": 0.544, "step": 9806 }, { "epoch": 0.3493952295277624, "grad_norm": 1.6131804690771359, "learning_rate": 1.5110513388350892e-05, "loss": 0.5646, "step": 9807 }, { "epoch": 0.34943085665425655, "grad_norm": 1.4032424845708293, "learning_rate": 1.5109521487822208e-05, "loss": 0.5843, "step": 9808 }, { "epoch": 0.3494664837807507, "grad_norm": 1.0420549755995905, "learning_rate": 1.5108529519256672e-05, "loss": 0.5125, "step": 9809 }, { "epoch": 0.34950211090724476, "grad_norm": 1.588550532579913, "learning_rate": 1.5107537482667497e-05, "loss": 0.6408, "step": 9810 }, { "epoch": 0.3495377380337389, "grad_norm": 1.0907994448338734, "learning_rate": 1.5106545378067887e-05, "loss": 0.4123, "step": 9811 }, { "epoch": 0.349573365160233, "grad_norm": 1.3995310243043604, "learning_rate": 1.510555320547106e-05, "loss": 0.5259, "step": 9812 }, { "epoch": 0.3496089922867271, "grad_norm": 1.5114725751295643, "learning_rate": 1.5104560964890223e-05, "loss": 0.5183, "step": 9813 }, { "epoch": 0.3496446194132212, "grad_norm": 1.813322930854027, "learning_rate": 1.5103568656338588e-05, "loss": 0.6334, "step": 9814 }, { "epoch": 0.34968024653971536, "grad_norm": 0.9709775196814903, "learning_rate": 1.5102576279829368e-05, "loss": 0.4652, "step": 9815 }, { "epoch": 0.34971587366620943, "grad_norm": 1.7371110273439174, "learning_rate": 1.510158383537578e-05, "loss": 0.5978, "step": 9816 }, { "epoch": 0.34975150079270356, "grad_norm": 1.57565075819017, "learning_rate": 1.5100591322991033e-05, "loss": 0.6099, "step": 9817 }, { "epoch": 0.3497871279191977, "grad_norm": 1.0447497601324474, "learning_rate": 1.509959874268835e-05, "loss": 0.508, "step": 9818 }, { "epoch": 0.34982275504569177, "grad_norm": 1.1787486207672986, "learning_rate": 1.509860609448095e-05, "loss": 0.5487, "step": 9819 }, { "epoch": 0.3498583821721859, "grad_norm": 1.202044907442796, "learning_rate": 1.5097613378382042e-05, "loss": 0.5283, "step": 9820 }, { "epoch": 0.34989400929868003, "grad_norm": 1.0133172655012606, "learning_rate": 1.5096620594404847e-05, "loss": 0.4483, "step": 9821 }, { "epoch": 0.3499296364251741, "grad_norm": 1.292211792387634, "learning_rate": 1.509562774256259e-05, "loss": 0.6095, "step": 9822 }, { "epoch": 0.34996526355166824, "grad_norm": 1.279850193872127, "learning_rate": 1.5094634822868486e-05, "loss": 0.5376, "step": 9823 }, { "epoch": 0.35000089067816237, "grad_norm": 1.1215126721969617, "learning_rate": 1.5093641835335762e-05, "loss": 0.5052, "step": 9824 }, { "epoch": 0.35003651780465644, "grad_norm": 1.1145550628452592, "learning_rate": 1.5092648779977635e-05, "loss": 0.5284, "step": 9825 }, { "epoch": 0.3500721449311506, "grad_norm": 1.3885704532583052, "learning_rate": 1.509165565680733e-05, "loss": 0.532, "step": 9826 }, { "epoch": 0.3501077720576447, "grad_norm": 1.064373706920718, "learning_rate": 1.5090662465838067e-05, "loss": 0.5028, "step": 9827 }, { "epoch": 0.3501433991841388, "grad_norm": 1.1038872476133226, "learning_rate": 1.5089669207083075e-05, "loss": 0.4831, "step": 9828 }, { "epoch": 0.3501790263106329, "grad_norm": 1.1083206146616515, "learning_rate": 1.5088675880555586e-05, "loss": 0.4686, "step": 9829 }, { "epoch": 0.35021465343712704, "grad_norm": 1.758965113138786, "learning_rate": 1.5087682486268818e-05, "loss": 0.5781, "step": 9830 }, { "epoch": 0.3502502805636211, "grad_norm": 1.0812068885295167, "learning_rate": 1.5086689024236e-05, "loss": 0.4695, "step": 9831 }, { "epoch": 0.35028590769011525, "grad_norm": 2.6424292161517218, "learning_rate": 1.5085695494470365e-05, "loss": 0.6251, "step": 9832 }, { "epoch": 0.3503215348166094, "grad_norm": 1.5773381861198266, "learning_rate": 1.5084701896985139e-05, "loss": 0.6283, "step": 9833 }, { "epoch": 0.35035716194310346, "grad_norm": 1.0585432602378004, "learning_rate": 1.5083708231793552e-05, "loss": 0.4953, "step": 9834 }, { "epoch": 0.3503927890695976, "grad_norm": 1.0986811975330384, "learning_rate": 1.5082714498908839e-05, "loss": 0.4595, "step": 9835 }, { "epoch": 0.3504284161960917, "grad_norm": 1.0695719633769663, "learning_rate": 1.508172069834423e-05, "loss": 0.4377, "step": 9836 }, { "epoch": 0.3504640433225858, "grad_norm": 2.0078896886457023, "learning_rate": 1.5080726830112955e-05, "loss": 0.7165, "step": 9837 }, { "epoch": 0.3504996704490799, "grad_norm": 1.1449801627151392, "learning_rate": 1.5079732894228252e-05, "loss": 0.5092, "step": 9838 }, { "epoch": 0.35053529757557406, "grad_norm": 1.3694528662206327, "learning_rate": 1.5078738890703355e-05, "loss": 0.5569, "step": 9839 }, { "epoch": 0.35057092470206813, "grad_norm": 1.3828428257085583, "learning_rate": 1.5077744819551499e-05, "loss": 0.5975, "step": 9840 }, { "epoch": 0.35060655182856226, "grad_norm": 1.578705529451974, "learning_rate": 1.5076750680785924e-05, "loss": 0.653, "step": 9841 }, { "epoch": 0.3506421789550564, "grad_norm": 1.2204052244386905, "learning_rate": 1.5075756474419862e-05, "loss": 0.4963, "step": 9842 }, { "epoch": 0.35067780608155047, "grad_norm": 1.1820287910070881, "learning_rate": 1.5074762200466557e-05, "loss": 0.5104, "step": 9843 }, { "epoch": 0.3507134332080446, "grad_norm": 1.5216672898952261, "learning_rate": 1.5073767858939248e-05, "loss": 0.5229, "step": 9844 }, { "epoch": 0.35074906033453873, "grad_norm": 1.0939960336257755, "learning_rate": 1.5072773449851172e-05, "loss": 0.5027, "step": 9845 }, { "epoch": 0.3507846874610328, "grad_norm": 2.0293094188608536, "learning_rate": 1.5071778973215573e-05, "loss": 0.5414, "step": 9846 }, { "epoch": 0.35082031458752694, "grad_norm": 1.8505825483834077, "learning_rate": 1.5070784429045689e-05, "loss": 0.578, "step": 9847 }, { "epoch": 0.35085594171402107, "grad_norm": 1.0477728664162131, "learning_rate": 1.5069789817354767e-05, "loss": 0.5055, "step": 9848 }, { "epoch": 0.35089156884051514, "grad_norm": 1.0727422441154515, "learning_rate": 1.506879513815605e-05, "loss": 0.492, "step": 9849 }, { "epoch": 0.3509271959670093, "grad_norm": 0.9599764305185426, "learning_rate": 1.5067800391462783e-05, "loss": 0.4382, "step": 9850 }, { "epoch": 0.3509628230935034, "grad_norm": 0.988768674633416, "learning_rate": 1.5066805577288212e-05, "loss": 0.4747, "step": 9851 }, { "epoch": 0.3509984502199975, "grad_norm": 1.4854523689875603, "learning_rate": 1.5065810695645584e-05, "loss": 0.6108, "step": 9852 }, { "epoch": 0.3510340773464916, "grad_norm": 1.1180657838897983, "learning_rate": 1.5064815746548144e-05, "loss": 0.4616, "step": 9853 }, { "epoch": 0.35106970447298574, "grad_norm": 1.9238018617938926, "learning_rate": 1.5063820730009145e-05, "loss": 0.5661, "step": 9854 }, { "epoch": 0.3511053315994798, "grad_norm": 1.0599777796745713, "learning_rate": 1.5062825646041832e-05, "loss": 0.4558, "step": 9855 }, { "epoch": 0.35114095872597395, "grad_norm": 1.0808632461547545, "learning_rate": 1.5061830494659458e-05, "loss": 0.4893, "step": 9856 }, { "epoch": 0.3511765858524681, "grad_norm": 1.7093953436793916, "learning_rate": 1.5060835275875269e-05, "loss": 0.617, "step": 9857 }, { "epoch": 0.35121221297896216, "grad_norm": 1.1422500054490339, "learning_rate": 1.5059839989702524e-05, "loss": 0.4903, "step": 9858 }, { "epoch": 0.3512478401054563, "grad_norm": 1.0185001306922872, "learning_rate": 1.5058844636154475e-05, "loss": 0.4783, "step": 9859 }, { "epoch": 0.3512834672319504, "grad_norm": 1.2593056164578067, "learning_rate": 1.5057849215244371e-05, "loss": 0.583, "step": 9860 }, { "epoch": 0.3513190943584445, "grad_norm": 1.1344461289718342, "learning_rate": 1.505685372698547e-05, "loss": 0.5254, "step": 9861 }, { "epoch": 0.3513547214849386, "grad_norm": 1.1814407093865638, "learning_rate": 1.5055858171391026e-05, "loss": 0.496, "step": 9862 }, { "epoch": 0.35139034861143276, "grad_norm": 1.4823978831068214, "learning_rate": 1.5054862548474298e-05, "loss": 0.5561, "step": 9863 }, { "epoch": 0.35142597573792683, "grad_norm": 2.31864396453384, "learning_rate": 1.5053866858248543e-05, "loss": 0.5901, "step": 9864 }, { "epoch": 0.35146160286442096, "grad_norm": 1.6126424275126199, "learning_rate": 1.5052871100727014e-05, "loss": 0.623, "step": 9865 }, { "epoch": 0.3514972299909151, "grad_norm": 1.1382324383320772, "learning_rate": 1.5051875275922979e-05, "loss": 0.4588, "step": 9866 }, { "epoch": 0.35153285711740917, "grad_norm": 1.0247714612713823, "learning_rate": 1.5050879383849693e-05, "loss": 0.4428, "step": 9867 }, { "epoch": 0.3515684842439033, "grad_norm": 1.1259109754185892, "learning_rate": 1.5049883424520416e-05, "loss": 0.5135, "step": 9868 }, { "epoch": 0.35160411137039743, "grad_norm": 1.8177877441502979, "learning_rate": 1.5048887397948413e-05, "loss": 0.5508, "step": 9869 }, { "epoch": 0.3516397384968915, "grad_norm": 1.5326826432749656, "learning_rate": 1.5047891304146945e-05, "loss": 0.532, "step": 9870 }, { "epoch": 0.35167536562338564, "grad_norm": 1.5399656010158551, "learning_rate": 1.5046895143129277e-05, "loss": 0.5572, "step": 9871 }, { "epoch": 0.35171099274987977, "grad_norm": 1.3726371219973332, "learning_rate": 1.504589891490867e-05, "loss": 0.574, "step": 9872 }, { "epoch": 0.35174661987637384, "grad_norm": 1.787077383368832, "learning_rate": 1.5044902619498394e-05, "loss": 0.6057, "step": 9873 }, { "epoch": 0.351782247002868, "grad_norm": 1.6280067878221929, "learning_rate": 1.5043906256911711e-05, "loss": 0.575, "step": 9874 }, { "epoch": 0.3518178741293621, "grad_norm": 1.6258939241101855, "learning_rate": 1.5042909827161894e-05, "loss": 0.626, "step": 9875 }, { "epoch": 0.3518535012558562, "grad_norm": 1.4775373343618856, "learning_rate": 1.5041913330262205e-05, "loss": 0.5552, "step": 9876 }, { "epoch": 0.3518891283823503, "grad_norm": 1.3864052520995747, "learning_rate": 1.5040916766225917e-05, "loss": 0.494, "step": 9877 }, { "epoch": 0.35192475550884444, "grad_norm": 1.9870391803009655, "learning_rate": 1.50399201350663e-05, "loss": 0.5342, "step": 9878 }, { "epoch": 0.3519603826353385, "grad_norm": 1.0907376247045926, "learning_rate": 1.5038923436796623e-05, "loss": 0.4767, "step": 9879 }, { "epoch": 0.35199600976183265, "grad_norm": 1.1889532881128329, "learning_rate": 1.503792667143016e-05, "loss": 0.5292, "step": 9880 }, { "epoch": 0.3520316368883268, "grad_norm": 1.865524645882663, "learning_rate": 1.5036929838980179e-05, "loss": 0.5819, "step": 9881 }, { "epoch": 0.35206726401482086, "grad_norm": 1.5972110574131793, "learning_rate": 1.5035932939459961e-05, "loss": 0.6389, "step": 9882 }, { "epoch": 0.352102891141315, "grad_norm": 1.0486305860646565, "learning_rate": 1.503493597288277e-05, "loss": 0.4912, "step": 9883 }, { "epoch": 0.3521385182678091, "grad_norm": 1.2728289022769648, "learning_rate": 1.5033938939261892e-05, "loss": 0.4763, "step": 9884 }, { "epoch": 0.3521741453943032, "grad_norm": 1.9580209144959904, "learning_rate": 1.5032941838610597e-05, "loss": 0.5746, "step": 9885 }, { "epoch": 0.3522097725207973, "grad_norm": 1.540898304760951, "learning_rate": 1.503194467094216e-05, "loss": 0.5721, "step": 9886 }, { "epoch": 0.35224539964729146, "grad_norm": 1.2693768843255153, "learning_rate": 1.5030947436269866e-05, "loss": 0.5232, "step": 9887 }, { "epoch": 0.35228102677378553, "grad_norm": 1.1898499012602206, "learning_rate": 1.5029950134606991e-05, "loss": 0.394, "step": 9888 }, { "epoch": 0.35231665390027966, "grad_norm": 2.0804689840360373, "learning_rate": 1.5028952765966815e-05, "loss": 0.6009, "step": 9889 }, { "epoch": 0.3523522810267738, "grad_norm": 1.1276768033520002, "learning_rate": 1.5027955330362619e-05, "loss": 0.4854, "step": 9890 }, { "epoch": 0.3523879081532679, "grad_norm": 1.1869498127474698, "learning_rate": 1.502695782780768e-05, "loss": 0.4731, "step": 9891 }, { "epoch": 0.352423535279762, "grad_norm": 1.1158438292714854, "learning_rate": 1.5025960258315285e-05, "loss": 0.4427, "step": 9892 }, { "epoch": 0.35245916240625613, "grad_norm": 2.507024723893013, "learning_rate": 1.5024962621898715e-05, "loss": 0.5321, "step": 9893 }, { "epoch": 0.35249478953275026, "grad_norm": 1.1003184744037982, "learning_rate": 1.5023964918571256e-05, "loss": 0.4992, "step": 9894 }, { "epoch": 0.35253041665924434, "grad_norm": 1.0717638262551845, "learning_rate": 1.5022967148346194e-05, "loss": 0.4753, "step": 9895 }, { "epoch": 0.35256604378573847, "grad_norm": 1.435896802560654, "learning_rate": 1.5021969311236815e-05, "loss": 0.5804, "step": 9896 }, { "epoch": 0.3526016709122326, "grad_norm": 1.0637135933493431, "learning_rate": 1.50209714072564e-05, "loss": 0.4582, "step": 9897 }, { "epoch": 0.3526372980387267, "grad_norm": 1.0495833648577946, "learning_rate": 1.5019973436418242e-05, "loss": 0.5122, "step": 9898 }, { "epoch": 0.3526729251652208, "grad_norm": 1.1596237781850154, "learning_rate": 1.5018975398735632e-05, "loss": 0.4479, "step": 9899 }, { "epoch": 0.35270855229171494, "grad_norm": 1.152266756034457, "learning_rate": 1.5017977294221854e-05, "loss": 0.4453, "step": 9900 }, { "epoch": 0.352744179418209, "grad_norm": 1.065182503941097, "learning_rate": 1.5016979122890203e-05, "loss": 0.5416, "step": 9901 }, { "epoch": 0.35277980654470315, "grad_norm": 1.0900460296540075, "learning_rate": 1.5015980884753967e-05, "loss": 0.4654, "step": 9902 }, { "epoch": 0.3528154336711973, "grad_norm": 1.3955062298129204, "learning_rate": 1.501498257982644e-05, "loss": 0.4849, "step": 9903 }, { "epoch": 0.35285106079769135, "grad_norm": 1.444942217475473, "learning_rate": 1.5013984208120915e-05, "loss": 0.5629, "step": 9904 }, { "epoch": 0.3528866879241855, "grad_norm": 1.521864987992069, "learning_rate": 1.5012985769650686e-05, "loss": 0.5577, "step": 9905 }, { "epoch": 0.3529223150506796, "grad_norm": 1.6807314875938524, "learning_rate": 1.5011987264429047e-05, "loss": 0.6541, "step": 9906 }, { "epoch": 0.3529579421771737, "grad_norm": 1.1061667106686925, "learning_rate": 1.5010988692469295e-05, "loss": 0.5019, "step": 9907 }, { "epoch": 0.3529935693036678, "grad_norm": 1.5102982651593861, "learning_rate": 1.5009990053784725e-05, "loss": 0.5568, "step": 9908 }, { "epoch": 0.35302919643016195, "grad_norm": 1.7156098798178838, "learning_rate": 1.5008991348388632e-05, "loss": 0.5792, "step": 9909 }, { "epoch": 0.353064823556656, "grad_norm": 1.3841424063864287, "learning_rate": 1.5007992576294323e-05, "loss": 0.6505, "step": 9910 }, { "epoch": 0.35310045068315016, "grad_norm": 1.1625989767439706, "learning_rate": 1.5006993737515092e-05, "loss": 0.4861, "step": 9911 }, { "epoch": 0.3531360778096443, "grad_norm": 1.2765262287464367, "learning_rate": 1.500599483206424e-05, "loss": 0.6497, "step": 9912 }, { "epoch": 0.35317170493613836, "grad_norm": 1.2563548778742983, "learning_rate": 1.5004995859955064e-05, "loss": 0.4748, "step": 9913 }, { "epoch": 0.3532073320626325, "grad_norm": 1.0930521870143581, "learning_rate": 1.5003996821200875e-05, "loss": 0.5344, "step": 9914 }, { "epoch": 0.3532429591891266, "grad_norm": 1.1966018965434297, "learning_rate": 1.500299771581497e-05, "loss": 0.477, "step": 9915 }, { "epoch": 0.3532785863156207, "grad_norm": 1.506160836232402, "learning_rate": 1.5001998543810654e-05, "loss": 0.5414, "step": 9916 }, { "epoch": 0.35331421344211483, "grad_norm": 1.4910032943821083, "learning_rate": 1.5000999305201227e-05, "loss": 0.5803, "step": 9917 }, { "epoch": 0.35334984056860896, "grad_norm": 1.3891571005126857, "learning_rate": 1.5000000000000002e-05, "loss": 0.4884, "step": 9918 }, { "epoch": 0.35338546769510304, "grad_norm": 1.1878837953352148, "learning_rate": 1.4999000628220281e-05, "loss": 0.4901, "step": 9919 }, { "epoch": 0.35342109482159717, "grad_norm": 1.200469104245857, "learning_rate": 1.4998001189875373e-05, "loss": 0.5164, "step": 9920 }, { "epoch": 0.3534567219480913, "grad_norm": 1.169525871120747, "learning_rate": 1.4997001684978585e-05, "loss": 0.4779, "step": 9921 }, { "epoch": 0.3534923490745854, "grad_norm": 1.2558683143913845, "learning_rate": 1.4996002113543227e-05, "loss": 0.5123, "step": 9922 }, { "epoch": 0.3535279762010795, "grad_norm": 1.101240894136664, "learning_rate": 1.499500247558261e-05, "loss": 0.5062, "step": 9923 }, { "epoch": 0.35356360332757364, "grad_norm": 1.0524369210400741, "learning_rate": 1.4994002771110043e-05, "loss": 0.5268, "step": 9924 }, { "epoch": 0.3535992304540677, "grad_norm": 1.0754258866533204, "learning_rate": 1.4993003000138839e-05, "loss": 0.4887, "step": 9925 }, { "epoch": 0.35363485758056185, "grad_norm": 1.1924167555229541, "learning_rate": 1.499200316268231e-05, "loss": 0.515, "step": 9926 }, { "epoch": 0.353670484707056, "grad_norm": 1.120818215758726, "learning_rate": 1.4991003258753772e-05, "loss": 0.5261, "step": 9927 }, { "epoch": 0.35370611183355005, "grad_norm": 1.3943383798320945, "learning_rate": 1.4990003288366533e-05, "loss": 0.5717, "step": 9928 }, { "epoch": 0.3537417389600442, "grad_norm": 1.7450231009097188, "learning_rate": 1.4989003251533913e-05, "loss": 0.5382, "step": 9929 }, { "epoch": 0.3537773660865383, "grad_norm": 1.5488842562420269, "learning_rate": 1.498800314826923e-05, "loss": 0.5771, "step": 9930 }, { "epoch": 0.3538129932130324, "grad_norm": 1.7729369590405144, "learning_rate": 1.4987002978585796e-05, "loss": 0.6466, "step": 9931 }, { "epoch": 0.3538486203395265, "grad_norm": 1.1878884697912078, "learning_rate": 1.4986002742496932e-05, "loss": 0.4617, "step": 9932 }, { "epoch": 0.35388424746602065, "grad_norm": 1.3947725375800835, "learning_rate": 1.4985002440015959e-05, "loss": 0.6083, "step": 9933 }, { "epoch": 0.3539198745925147, "grad_norm": 1.4033968844269238, "learning_rate": 1.4984002071156192e-05, "loss": 0.5859, "step": 9934 }, { "epoch": 0.35395550171900886, "grad_norm": 1.3133772934012966, "learning_rate": 1.4983001635930954e-05, "loss": 0.5203, "step": 9935 }, { "epoch": 0.353991128845503, "grad_norm": 1.2858518039828957, "learning_rate": 1.4982001134353568e-05, "loss": 0.4534, "step": 9936 }, { "epoch": 0.35402675597199706, "grad_norm": 1.1499443258785997, "learning_rate": 1.4981000566437354e-05, "loss": 0.5276, "step": 9937 }, { "epoch": 0.3540623830984912, "grad_norm": 1.057091475905003, "learning_rate": 1.4979999932195635e-05, "loss": 0.4814, "step": 9938 }, { "epoch": 0.3540980102249853, "grad_norm": 1.3032291393751074, "learning_rate": 1.4978999231641736e-05, "loss": 0.465, "step": 9939 }, { "epoch": 0.3541336373514794, "grad_norm": 1.1946183497778127, "learning_rate": 1.4977998464788983e-05, "loss": 0.4793, "step": 9940 }, { "epoch": 0.35416926447797353, "grad_norm": 1.6328012564350585, "learning_rate": 1.4976997631650701e-05, "loss": 0.4924, "step": 9941 }, { "epoch": 0.35420489160446766, "grad_norm": 1.6976937168347928, "learning_rate": 1.4975996732240214e-05, "loss": 0.6198, "step": 9942 }, { "epoch": 0.35424051873096174, "grad_norm": 1.1757760722014425, "learning_rate": 1.4974995766570856e-05, "loss": 0.4444, "step": 9943 }, { "epoch": 0.35427614585745587, "grad_norm": 1.487131604847805, "learning_rate": 1.497399473465595e-05, "loss": 0.5539, "step": 9944 }, { "epoch": 0.35431177298395, "grad_norm": 1.1606534608918926, "learning_rate": 1.497299363650883e-05, "loss": 0.4757, "step": 9945 }, { "epoch": 0.3543474001104441, "grad_norm": 1.1448548285503595, "learning_rate": 1.4971992472142822e-05, "loss": 0.4611, "step": 9946 }, { "epoch": 0.3543830272369382, "grad_norm": 1.9742154779065486, "learning_rate": 1.497099124157126e-05, "loss": 0.6045, "step": 9947 }, { "epoch": 0.35441865436343234, "grad_norm": 1.6994065164283878, "learning_rate": 1.4969989944807472e-05, "loss": 0.5993, "step": 9948 }, { "epoch": 0.3544542814899264, "grad_norm": 1.6610473267725152, "learning_rate": 1.4968988581864798e-05, "loss": 0.6125, "step": 9949 }, { "epoch": 0.35448990861642055, "grad_norm": 1.9679118780915597, "learning_rate": 1.4967987152756567e-05, "loss": 0.5368, "step": 9950 }, { "epoch": 0.3545255357429147, "grad_norm": 1.0486754189342604, "learning_rate": 1.4966985657496113e-05, "loss": 0.4268, "step": 9951 }, { "epoch": 0.35456116286940875, "grad_norm": 1.06130975936013, "learning_rate": 1.4965984096096775e-05, "loss": 0.5444, "step": 9952 }, { "epoch": 0.3545967899959029, "grad_norm": 1.4887749463944355, "learning_rate": 1.496498246857189e-05, "loss": 0.6243, "step": 9953 }, { "epoch": 0.354632417122397, "grad_norm": 1.3691338465292773, "learning_rate": 1.4963980774934788e-05, "loss": 0.5793, "step": 9954 }, { "epoch": 0.3546680442488911, "grad_norm": 1.2810680717600755, "learning_rate": 1.4962979015198813e-05, "loss": 0.5393, "step": 9955 }, { "epoch": 0.3547036713753852, "grad_norm": 1.553244217459576, "learning_rate": 1.4961977189377306e-05, "loss": 0.5819, "step": 9956 }, { "epoch": 0.35473929850187935, "grad_norm": 1.4405341946888235, "learning_rate": 1.4960975297483605e-05, "loss": 0.5485, "step": 9957 }, { "epoch": 0.3547749256283734, "grad_norm": 1.6282586484075896, "learning_rate": 1.4959973339531046e-05, "loss": 0.5811, "step": 9958 }, { "epoch": 0.35481055275486756, "grad_norm": 1.9440922595989352, "learning_rate": 1.4958971315532979e-05, "loss": 0.6104, "step": 9959 }, { "epoch": 0.3548461798813617, "grad_norm": 1.2018459317674992, "learning_rate": 1.4957969225502742e-05, "loss": 0.4945, "step": 9960 }, { "epoch": 0.35488180700785577, "grad_norm": 1.4251701929284242, "learning_rate": 1.4956967069453678e-05, "loss": 0.5397, "step": 9961 }, { "epoch": 0.3549174341343499, "grad_norm": 1.1366567353871657, "learning_rate": 1.4955964847399135e-05, "loss": 0.4618, "step": 9962 }, { "epoch": 0.354953061260844, "grad_norm": 1.7737134066237463, "learning_rate": 1.4954962559352457e-05, "loss": 0.5846, "step": 9963 }, { "epoch": 0.3549886883873381, "grad_norm": 1.0826622042607417, "learning_rate": 1.4953960205326987e-05, "loss": 0.5089, "step": 9964 }, { "epoch": 0.35502431551383223, "grad_norm": 1.7659673816581938, "learning_rate": 1.4952957785336073e-05, "loss": 0.5524, "step": 9965 }, { "epoch": 0.35505994264032636, "grad_norm": 1.960179471857212, "learning_rate": 1.4951955299393066e-05, "loss": 0.6767, "step": 9966 }, { "epoch": 0.35509556976682044, "grad_norm": 1.5325465265211113, "learning_rate": 1.4950952747511311e-05, "loss": 0.5261, "step": 9967 }, { "epoch": 0.35513119689331457, "grad_norm": 1.2495806486578553, "learning_rate": 1.4949950129704162e-05, "loss": 0.4685, "step": 9968 }, { "epoch": 0.3551668240198087, "grad_norm": 2.490046519959997, "learning_rate": 1.4948947445984967e-05, "loss": 0.5683, "step": 9969 }, { "epoch": 0.3552024511463028, "grad_norm": 1.1020403060945745, "learning_rate": 1.4947944696367079e-05, "loss": 0.4738, "step": 9970 }, { "epoch": 0.3552380782727969, "grad_norm": 1.6187641721912338, "learning_rate": 1.4946941880863846e-05, "loss": 0.5701, "step": 9971 }, { "epoch": 0.35527370539929104, "grad_norm": 1.21316949657411, "learning_rate": 1.4945938999488628e-05, "loss": 0.5442, "step": 9972 }, { "epoch": 0.3553093325257851, "grad_norm": 1.1738284719413397, "learning_rate": 1.494493605225477e-05, "loss": 0.4742, "step": 9973 }, { "epoch": 0.35534495965227925, "grad_norm": 1.1437671316268805, "learning_rate": 1.4943933039175636e-05, "loss": 0.4585, "step": 9974 }, { "epoch": 0.3553805867787734, "grad_norm": 2.2227668506319276, "learning_rate": 1.4942929960264576e-05, "loss": 0.5469, "step": 9975 }, { "epoch": 0.35541621390526745, "grad_norm": 1.0760438484612584, "learning_rate": 1.494192681553495e-05, "loss": 0.4735, "step": 9976 }, { "epoch": 0.3554518410317616, "grad_norm": 1.1165590870944717, "learning_rate": 1.4940923605000112e-05, "loss": 0.4816, "step": 9977 }, { "epoch": 0.3554874681582557, "grad_norm": 2.037204913090425, "learning_rate": 1.4939920328673422e-05, "loss": 0.6012, "step": 9978 }, { "epoch": 0.3555230952847498, "grad_norm": 1.1020576361036727, "learning_rate": 1.4938916986568246e-05, "loss": 0.5179, "step": 9979 }, { "epoch": 0.3555587224112439, "grad_norm": 1.6709760613170161, "learning_rate": 1.4937913578697933e-05, "loss": 0.5016, "step": 9980 }, { "epoch": 0.35559434953773805, "grad_norm": 1.5019512660505963, "learning_rate": 1.493691010507585e-05, "loss": 0.5277, "step": 9981 }, { "epoch": 0.35562997666423213, "grad_norm": 2.307843203504597, "learning_rate": 1.493590656571536e-05, "loss": 0.6107, "step": 9982 }, { "epoch": 0.35566560379072626, "grad_norm": 1.3537938116147832, "learning_rate": 1.4934902960629825e-05, "loss": 0.5635, "step": 9983 }, { "epoch": 0.3557012309172204, "grad_norm": 1.5149884664638253, "learning_rate": 1.4933899289832603e-05, "loss": 0.5651, "step": 9984 }, { "epoch": 0.35573685804371447, "grad_norm": 1.104252506900909, "learning_rate": 1.4932895553337068e-05, "loss": 0.5607, "step": 9985 }, { "epoch": 0.3557724851702086, "grad_norm": 1.0917013408908065, "learning_rate": 1.493189175115658e-05, "loss": 0.5073, "step": 9986 }, { "epoch": 0.3558081122967027, "grad_norm": 1.111967531244325, "learning_rate": 1.4930887883304503e-05, "loss": 0.4879, "step": 9987 }, { "epoch": 0.3558437394231968, "grad_norm": 1.1338685965863433, "learning_rate": 1.4929883949794213e-05, "loss": 0.4901, "step": 9988 }, { "epoch": 0.35587936654969093, "grad_norm": 1.0814198958655739, "learning_rate": 1.4928879950639068e-05, "loss": 0.4805, "step": 9989 }, { "epoch": 0.35591499367618507, "grad_norm": 1.9186239736275605, "learning_rate": 1.4927875885852442e-05, "loss": 0.6057, "step": 9990 }, { "epoch": 0.35595062080267914, "grad_norm": 1.575876282169966, "learning_rate": 1.4926871755447705e-05, "loss": 0.5996, "step": 9991 }, { "epoch": 0.35598624792917327, "grad_norm": 1.500618606621345, "learning_rate": 1.4925867559438225e-05, "loss": 0.5863, "step": 9992 }, { "epoch": 0.3560218750556674, "grad_norm": 1.375063918306472, "learning_rate": 1.4924863297837378e-05, "loss": 0.5131, "step": 9993 }, { "epoch": 0.3560575021821615, "grad_norm": 1.4118584472996982, "learning_rate": 1.4923858970658534e-05, "loss": 0.5733, "step": 9994 }, { "epoch": 0.3560931293086556, "grad_norm": 2.310573291013925, "learning_rate": 1.4922854577915064e-05, "loss": 0.6022, "step": 9995 }, { "epoch": 0.35612875643514974, "grad_norm": 1.2879458195014586, "learning_rate": 1.4921850119620348e-05, "loss": 0.5193, "step": 9996 }, { "epoch": 0.3561643835616438, "grad_norm": 1.3827973593036265, "learning_rate": 1.4920845595787755e-05, "loss": 0.5167, "step": 9997 }, { "epoch": 0.35620001068813795, "grad_norm": 1.6176001434268261, "learning_rate": 1.4919841006430663e-05, "loss": 0.5861, "step": 9998 }, { "epoch": 0.3562356378146321, "grad_norm": 1.0848103357186871, "learning_rate": 1.491883635156245e-05, "loss": 0.441, "step": 9999 }, { "epoch": 0.35627126494112615, "grad_norm": 1.0887031386045876, "learning_rate": 1.4917831631196494e-05, "loss": 0.4335, "step": 10000 }, { "epoch": 0.3563068920676203, "grad_norm": 1.1190643458013958, "learning_rate": 1.4916826845346168e-05, "loss": 0.4892, "step": 10001 }, { "epoch": 0.3563425191941144, "grad_norm": 1.2097893015784378, "learning_rate": 1.4915821994024859e-05, "loss": 0.5239, "step": 10002 }, { "epoch": 0.3563781463206085, "grad_norm": 1.3385872524441322, "learning_rate": 1.4914817077245945e-05, "loss": 0.5989, "step": 10003 }, { "epoch": 0.3564137734471026, "grad_norm": 1.043565689299263, "learning_rate": 1.4913812095022806e-05, "loss": 0.4998, "step": 10004 }, { "epoch": 0.35644940057359675, "grad_norm": 1.3949305360579922, "learning_rate": 1.4912807047368825e-05, "loss": 0.5397, "step": 10005 }, { "epoch": 0.35648502770009083, "grad_norm": 1.1455242319342402, "learning_rate": 1.4911801934297381e-05, "loss": 0.5644, "step": 10006 }, { "epoch": 0.35652065482658496, "grad_norm": 1.7329028645627098, "learning_rate": 1.4910796755821863e-05, "loss": 0.5787, "step": 10007 }, { "epoch": 0.3565562819530791, "grad_norm": 2.035121898961449, "learning_rate": 1.4909791511955656e-05, "loss": 0.641, "step": 10008 }, { "epoch": 0.35659190907957317, "grad_norm": 1.0195079726110752, "learning_rate": 1.4908786202712142e-05, "loss": 0.4824, "step": 10009 }, { "epoch": 0.3566275362060673, "grad_norm": 1.0870096657696067, "learning_rate": 1.4907780828104707e-05, "loss": 0.502, "step": 10010 }, { "epoch": 0.35666316333256143, "grad_norm": 1.7185090504175704, "learning_rate": 1.4906775388146742e-05, "loss": 0.6378, "step": 10011 }, { "epoch": 0.3566987904590555, "grad_norm": 1.0836022327580437, "learning_rate": 1.4905769882851633e-05, "loss": 0.4911, "step": 10012 }, { "epoch": 0.35673441758554963, "grad_norm": 1.0699302915952957, "learning_rate": 1.4904764312232766e-05, "loss": 0.4626, "step": 10013 }, { "epoch": 0.35677004471204377, "grad_norm": 1.5121491000992462, "learning_rate": 1.4903758676303538e-05, "loss": 0.555, "step": 10014 }, { "epoch": 0.35680567183853784, "grad_norm": 1.1309211988325216, "learning_rate": 1.4902752975077333e-05, "loss": 0.5335, "step": 10015 }, { "epoch": 0.35684129896503197, "grad_norm": 1.4789400565441653, "learning_rate": 1.4901747208567545e-05, "loss": 0.6061, "step": 10016 }, { "epoch": 0.3568769260915261, "grad_norm": 1.7649330432426114, "learning_rate": 1.4900741376787568e-05, "loss": 0.5775, "step": 10017 }, { "epoch": 0.3569125532180202, "grad_norm": 1.5908609122391644, "learning_rate": 1.4899735479750794e-05, "loss": 0.6498, "step": 10018 }, { "epoch": 0.3569481803445143, "grad_norm": 1.0654381814539773, "learning_rate": 1.4898729517470618e-05, "loss": 0.4361, "step": 10019 }, { "epoch": 0.35698380747100844, "grad_norm": 1.291585289131978, "learning_rate": 1.4897723489960436e-05, "loss": 0.4725, "step": 10020 }, { "epoch": 0.3570194345975025, "grad_norm": 1.0952034949123732, "learning_rate": 1.489671739723364e-05, "loss": 0.4831, "step": 10021 }, { "epoch": 0.35705506172399665, "grad_norm": 1.4250445868931032, "learning_rate": 1.489571123930363e-05, "loss": 0.5495, "step": 10022 }, { "epoch": 0.3570906888504908, "grad_norm": 1.1391959640067573, "learning_rate": 1.4894705016183803e-05, "loss": 0.4701, "step": 10023 }, { "epoch": 0.35712631597698485, "grad_norm": 1.9988305236163224, "learning_rate": 1.4893698727887555e-05, "loss": 0.5872, "step": 10024 }, { "epoch": 0.357161943103479, "grad_norm": 1.1874119289704645, "learning_rate": 1.489269237442829e-05, "loss": 0.4937, "step": 10025 }, { "epoch": 0.3571975702299731, "grad_norm": 1.0902873379048155, "learning_rate": 1.4891685955819405e-05, "loss": 0.4598, "step": 10026 }, { "epoch": 0.3572331973564672, "grad_norm": 1.2206727333231022, "learning_rate": 1.4890679472074308e-05, "loss": 0.5198, "step": 10027 }, { "epoch": 0.3572688244829613, "grad_norm": 1.2875150764153636, "learning_rate": 1.488967292320639e-05, "loss": 0.4709, "step": 10028 }, { "epoch": 0.35730445160945545, "grad_norm": 1.08816765129295, "learning_rate": 1.4888666309229058e-05, "loss": 0.4582, "step": 10029 }, { "epoch": 0.35734007873594953, "grad_norm": 1.335212964978044, "learning_rate": 1.4887659630155722e-05, "loss": 0.5501, "step": 10030 }, { "epoch": 0.35737570586244366, "grad_norm": 1.3622403636435214, "learning_rate": 1.488665288599978e-05, "loss": 0.572, "step": 10031 }, { "epoch": 0.3574113329889378, "grad_norm": 1.8077582289557375, "learning_rate": 1.4885646076774639e-05, "loss": 0.5541, "step": 10032 }, { "epoch": 0.35744696011543187, "grad_norm": 1.1092855218606135, "learning_rate": 1.4884639202493705e-05, "loss": 0.5561, "step": 10033 }, { "epoch": 0.357482587241926, "grad_norm": 1.4827800354828728, "learning_rate": 1.4883632263170387e-05, "loss": 0.5737, "step": 10034 }, { "epoch": 0.35751821436842013, "grad_norm": 1.4493633833736548, "learning_rate": 1.488262525881809e-05, "loss": 0.615, "step": 10035 }, { "epoch": 0.3575538414949142, "grad_norm": 1.9242083304003543, "learning_rate": 1.4881618189450224e-05, "loss": 0.5344, "step": 10036 }, { "epoch": 0.35758946862140834, "grad_norm": 1.6349359560528476, "learning_rate": 1.4880611055080202e-05, "loss": 0.5857, "step": 10037 }, { "epoch": 0.35762509574790247, "grad_norm": 1.3407251798286348, "learning_rate": 1.4879603855721434e-05, "loss": 0.4919, "step": 10038 }, { "epoch": 0.35766072287439654, "grad_norm": 3.0698504160969255, "learning_rate": 1.4878596591387329e-05, "loss": 0.6658, "step": 10039 }, { "epoch": 0.3576963500008907, "grad_norm": 1.4387940842031857, "learning_rate": 1.4877589262091299e-05, "loss": 0.613, "step": 10040 }, { "epoch": 0.3577319771273848, "grad_norm": 1.2464265702108945, "learning_rate": 1.487658186784676e-05, "loss": 0.5117, "step": 10041 }, { "epoch": 0.3577676042538789, "grad_norm": 1.21626509122213, "learning_rate": 1.4875574408667125e-05, "loss": 0.5297, "step": 10042 }, { "epoch": 0.357803231380373, "grad_norm": 2.217419004880849, "learning_rate": 1.4874566884565806e-05, "loss": 0.576, "step": 10043 }, { "epoch": 0.35783885850686714, "grad_norm": 1.106122849799309, "learning_rate": 1.4873559295556225e-05, "loss": 0.4714, "step": 10044 }, { "epoch": 0.3578744856333612, "grad_norm": 1.3137163771959315, "learning_rate": 1.4872551641651796e-05, "loss": 0.5065, "step": 10045 }, { "epoch": 0.35791011275985535, "grad_norm": 1.1642424095084942, "learning_rate": 1.4871543922865936e-05, "loss": 0.4449, "step": 10046 }, { "epoch": 0.3579457398863495, "grad_norm": 1.4903905064574043, "learning_rate": 1.4870536139212062e-05, "loss": 0.6142, "step": 10047 }, { "epoch": 0.35798136701284355, "grad_norm": 1.0939902821746543, "learning_rate": 1.4869528290703592e-05, "loss": 0.4577, "step": 10048 }, { "epoch": 0.3580169941393377, "grad_norm": 2.169364055457726, "learning_rate": 1.4868520377353954e-05, "loss": 0.5735, "step": 10049 }, { "epoch": 0.3580526212658318, "grad_norm": 1.6976193187482926, "learning_rate": 1.4867512399176564e-05, "loss": 0.5816, "step": 10050 }, { "epoch": 0.3580882483923259, "grad_norm": 1.0441447120346508, "learning_rate": 1.4866504356184844e-05, "loss": 0.4681, "step": 10051 }, { "epoch": 0.35812387551882, "grad_norm": 1.0483247771675364, "learning_rate": 1.4865496248392215e-05, "loss": 0.4489, "step": 10052 }, { "epoch": 0.35815950264531415, "grad_norm": 1.2061653917209265, "learning_rate": 1.4864488075812105e-05, "loss": 0.515, "step": 10053 }, { "epoch": 0.35819512977180823, "grad_norm": 1.603299760489524, "learning_rate": 1.4863479838457937e-05, "loss": 0.6257, "step": 10054 }, { "epoch": 0.35823075689830236, "grad_norm": 1.6848222773841652, "learning_rate": 1.486247153634313e-05, "loss": 0.6437, "step": 10055 }, { "epoch": 0.3582663840247965, "grad_norm": 1.1251783563451276, "learning_rate": 1.486146316948112e-05, "loss": 0.4787, "step": 10056 }, { "epoch": 0.35830201115129057, "grad_norm": 1.6901681173022065, "learning_rate": 1.4860454737885332e-05, "loss": 0.5585, "step": 10057 }, { "epoch": 0.3583376382777847, "grad_norm": 1.8761348319481552, "learning_rate": 1.4859446241569188e-05, "loss": 0.4914, "step": 10058 }, { "epoch": 0.35837326540427883, "grad_norm": 1.1189023524069481, "learning_rate": 1.485843768054612e-05, "loss": 0.5018, "step": 10059 }, { "epoch": 0.3584088925307729, "grad_norm": 1.5108768129625805, "learning_rate": 1.4857429054829561e-05, "loss": 0.5719, "step": 10060 }, { "epoch": 0.35844451965726704, "grad_norm": 1.4648903902102255, "learning_rate": 1.4856420364432943e-05, "loss": 0.6465, "step": 10061 }, { "epoch": 0.35848014678376117, "grad_norm": 1.768406075942834, "learning_rate": 1.4855411609369685e-05, "loss": 0.7159, "step": 10062 }, { "epoch": 0.35851577391025524, "grad_norm": 1.6122647128740584, "learning_rate": 1.4854402789653233e-05, "loss": 0.548, "step": 10063 }, { "epoch": 0.3585514010367494, "grad_norm": 1.1411849677289965, "learning_rate": 1.4853393905297013e-05, "loss": 0.4843, "step": 10064 }, { "epoch": 0.3585870281632435, "grad_norm": 1.1623854411304837, "learning_rate": 1.4852384956314463e-05, "loss": 0.5167, "step": 10065 }, { "epoch": 0.3586226552897376, "grad_norm": 2.079676627639434, "learning_rate": 1.4851375942719013e-05, "loss": 0.5944, "step": 10066 }, { "epoch": 0.3586582824162317, "grad_norm": 1.1822216590357983, "learning_rate": 1.4850366864524104e-05, "loss": 0.5137, "step": 10067 }, { "epoch": 0.35869390954272584, "grad_norm": 1.602603090653708, "learning_rate": 1.4849357721743169e-05, "loss": 0.4997, "step": 10068 }, { "epoch": 0.3587295366692199, "grad_norm": 1.2333679880686188, "learning_rate": 1.4848348514389646e-05, "loss": 0.6124, "step": 10069 }, { "epoch": 0.35876516379571405, "grad_norm": 1.1323209381936898, "learning_rate": 1.484733924247697e-05, "loss": 0.4709, "step": 10070 }, { "epoch": 0.3588007909222082, "grad_norm": 1.2137860522769437, "learning_rate": 1.4846329906018588e-05, "loss": 0.549, "step": 10071 }, { "epoch": 0.35883641804870225, "grad_norm": 1.2547644263411786, "learning_rate": 1.4845320505027938e-05, "loss": 0.5014, "step": 10072 }, { "epoch": 0.3588720451751964, "grad_norm": 1.0359861647936934, "learning_rate": 1.4844311039518457e-05, "loss": 0.4224, "step": 10073 }, { "epoch": 0.3589076723016905, "grad_norm": 1.0392451080345428, "learning_rate": 1.4843301509503584e-05, "loss": 0.468, "step": 10074 }, { "epoch": 0.3589432994281846, "grad_norm": 0.9782934937322155, "learning_rate": 1.4842291914996771e-05, "loss": 0.4411, "step": 10075 }, { "epoch": 0.3589789265546787, "grad_norm": 1.2979125664601243, "learning_rate": 1.4841282256011458e-05, "loss": 0.5648, "step": 10076 }, { "epoch": 0.35901455368117285, "grad_norm": 1.0332725742947773, "learning_rate": 1.4840272532561084e-05, "loss": 0.5305, "step": 10077 }, { "epoch": 0.35905018080766693, "grad_norm": 1.421653757973845, "learning_rate": 1.48392627446591e-05, "loss": 0.6434, "step": 10078 }, { "epoch": 0.35908580793416106, "grad_norm": 1.0558421127662814, "learning_rate": 1.4838252892318949e-05, "loss": 0.5069, "step": 10079 }, { "epoch": 0.3591214350606552, "grad_norm": 1.5023179069346857, "learning_rate": 1.4837242975554082e-05, "loss": 0.5587, "step": 10080 }, { "epoch": 0.3591570621871493, "grad_norm": 1.4434553475330405, "learning_rate": 1.483623299437794e-05, "loss": 0.6248, "step": 10081 }, { "epoch": 0.3591926893136434, "grad_norm": 1.14051888441277, "learning_rate": 1.4835222948803976e-05, "loss": 0.4936, "step": 10082 }, { "epoch": 0.35922831644013753, "grad_norm": 2.059843387256519, "learning_rate": 1.4834212838845639e-05, "loss": 0.5508, "step": 10083 }, { "epoch": 0.35926394356663166, "grad_norm": 1.1105165907798185, "learning_rate": 1.4833202664516381e-05, "loss": 0.4588, "step": 10084 }, { "epoch": 0.35929957069312574, "grad_norm": 1.6068806717896569, "learning_rate": 1.483219242582965e-05, "loss": 0.6345, "step": 10085 }, { "epoch": 0.35933519781961987, "grad_norm": 0.9892530717752723, "learning_rate": 1.48311821227989e-05, "loss": 0.468, "step": 10086 }, { "epoch": 0.359370824946114, "grad_norm": 1.601290409890584, "learning_rate": 1.483017175543758e-05, "loss": 0.5673, "step": 10087 }, { "epoch": 0.3594064520726081, "grad_norm": 0.995932593142435, "learning_rate": 1.4829161323759152e-05, "loss": 0.4579, "step": 10088 }, { "epoch": 0.3594420791991022, "grad_norm": 0.9977492600021005, "learning_rate": 1.482815082777706e-05, "loss": 0.4866, "step": 10089 }, { "epoch": 0.35947770632559634, "grad_norm": 1.4721447941226735, "learning_rate": 1.4827140267504772e-05, "loss": 0.5782, "step": 10090 }, { "epoch": 0.3595133334520904, "grad_norm": 1.1669624479630807, "learning_rate": 1.4826129642955734e-05, "loss": 0.471, "step": 10091 }, { "epoch": 0.35954896057858454, "grad_norm": 1.7046078525146398, "learning_rate": 1.4825118954143405e-05, "loss": 0.5951, "step": 10092 }, { "epoch": 0.3595845877050787, "grad_norm": 1.0995274144785503, "learning_rate": 1.4824108201081247e-05, "loss": 0.5067, "step": 10093 }, { "epoch": 0.35962021483157275, "grad_norm": 1.8671514131389102, "learning_rate": 1.4823097383782714e-05, "loss": 0.573, "step": 10094 }, { "epoch": 0.3596558419580669, "grad_norm": 1.226058896915771, "learning_rate": 1.4822086502261268e-05, "loss": 0.5293, "step": 10095 }, { "epoch": 0.359691469084561, "grad_norm": 1.5963424722076096, "learning_rate": 1.4821075556530371e-05, "loss": 0.4912, "step": 10096 }, { "epoch": 0.3597270962110551, "grad_norm": 1.0748040411447608, "learning_rate": 1.4820064546603484e-05, "loss": 0.465, "step": 10097 }, { "epoch": 0.3597627233375492, "grad_norm": 1.1881522124624784, "learning_rate": 1.481905347249407e-05, "loss": 0.4867, "step": 10098 }, { "epoch": 0.35979835046404335, "grad_norm": 1.5712107583521584, "learning_rate": 1.4818042334215585e-05, "loss": 0.5654, "step": 10099 }, { "epoch": 0.3598339775905374, "grad_norm": 2.018675962304227, "learning_rate": 1.4817031131781505e-05, "loss": 0.6192, "step": 10100 }, { "epoch": 0.35986960471703155, "grad_norm": 1.05640384294912, "learning_rate": 1.4816019865205284e-05, "loss": 0.4496, "step": 10101 }, { "epoch": 0.3599052318435257, "grad_norm": 2.1307241484417325, "learning_rate": 1.4815008534500397e-05, "loss": 0.4914, "step": 10102 }, { "epoch": 0.35994085897001976, "grad_norm": 1.1248849921726098, "learning_rate": 1.4813997139680303e-05, "loss": 0.5035, "step": 10103 }, { "epoch": 0.3599764860965139, "grad_norm": 1.4195331193920437, "learning_rate": 1.4812985680758471e-05, "loss": 0.6493, "step": 10104 }, { "epoch": 0.360012113223008, "grad_norm": 1.6234262386269407, "learning_rate": 1.4811974157748372e-05, "loss": 0.4993, "step": 10105 }, { "epoch": 0.3600477403495021, "grad_norm": 1.0882753345264886, "learning_rate": 1.4810962570663475e-05, "loss": 0.5026, "step": 10106 }, { "epoch": 0.36008336747599623, "grad_norm": 1.43748655780123, "learning_rate": 1.4809950919517245e-05, "loss": 0.5733, "step": 10107 }, { "epoch": 0.36011899460249036, "grad_norm": 1.4763471792256193, "learning_rate": 1.4808939204323158e-05, "loss": 0.5594, "step": 10108 }, { "epoch": 0.36015462172898444, "grad_norm": 1.689513829719036, "learning_rate": 1.4807927425094687e-05, "loss": 0.585, "step": 10109 }, { "epoch": 0.36019024885547857, "grad_norm": 1.5463508886587445, "learning_rate": 1.48069155818453e-05, "loss": 0.5786, "step": 10110 }, { "epoch": 0.3602258759819727, "grad_norm": 1.2786810479022077, "learning_rate": 1.4805903674588474e-05, "loss": 0.5629, "step": 10111 }, { "epoch": 0.3602615031084668, "grad_norm": 1.0865778829098314, "learning_rate": 1.4804891703337679e-05, "loss": 0.4992, "step": 10112 }, { "epoch": 0.3602971302349609, "grad_norm": 1.4208777313287584, "learning_rate": 1.4803879668106393e-05, "loss": 0.6802, "step": 10113 }, { "epoch": 0.36033275736145504, "grad_norm": 1.0738227138235479, "learning_rate": 1.480286756890809e-05, "loss": 0.5223, "step": 10114 }, { "epoch": 0.3603683844879491, "grad_norm": 1.0557473801325712, "learning_rate": 1.4801855405756251e-05, "loss": 0.4687, "step": 10115 }, { "epoch": 0.36040401161444324, "grad_norm": 1.2428024779132332, "learning_rate": 1.480084317866435e-05, "loss": 0.4865, "step": 10116 }, { "epoch": 0.3604396387409374, "grad_norm": 0.9925968199060065, "learning_rate": 1.4799830887645867e-05, "loss": 0.4844, "step": 10117 }, { "epoch": 0.36047526586743145, "grad_norm": 1.1251629353628987, "learning_rate": 1.4798818532714278e-05, "loss": 0.5642, "step": 10118 }, { "epoch": 0.3605108929939256, "grad_norm": 1.4834379133541085, "learning_rate": 1.4797806113883069e-05, "loss": 0.6025, "step": 10119 }, { "epoch": 0.3605465201204197, "grad_norm": 1.6325017851027792, "learning_rate": 1.4796793631165718e-05, "loss": 0.4788, "step": 10120 }, { "epoch": 0.3605821472469138, "grad_norm": 1.8790279750293162, "learning_rate": 1.4795781084575707e-05, "loss": 0.7177, "step": 10121 }, { "epoch": 0.3606177743734079, "grad_norm": 1.622695551116485, "learning_rate": 1.4794768474126518e-05, "loss": 0.5483, "step": 10122 }, { "epoch": 0.36065340149990205, "grad_norm": 1.1756543972889242, "learning_rate": 1.4793755799831637e-05, "loss": 0.5591, "step": 10123 }, { "epoch": 0.3606890286263961, "grad_norm": 2.1718726645351034, "learning_rate": 1.4792743061704547e-05, "loss": 0.5998, "step": 10124 }, { "epoch": 0.36072465575289026, "grad_norm": 1.4193938997500126, "learning_rate": 1.479173025975873e-05, "loss": 0.5551, "step": 10125 }, { "epoch": 0.3607602828793844, "grad_norm": 1.7820769899169493, "learning_rate": 1.4790717394007676e-05, "loss": 0.611, "step": 10126 }, { "epoch": 0.36079591000587846, "grad_norm": 1.0346815403105727, "learning_rate": 1.4789704464464873e-05, "loss": 0.5228, "step": 10127 }, { "epoch": 0.3608315371323726, "grad_norm": 1.1655240754100527, "learning_rate": 1.4788691471143808e-05, "loss": 0.4995, "step": 10128 }, { "epoch": 0.3608671642588667, "grad_norm": 1.0856660846938582, "learning_rate": 1.4787678414057969e-05, "loss": 0.4887, "step": 10129 }, { "epoch": 0.3609027913853608, "grad_norm": 1.409185404537703, "learning_rate": 1.478666529322084e-05, "loss": 0.5913, "step": 10130 }, { "epoch": 0.36093841851185493, "grad_norm": 1.754905156285458, "learning_rate": 1.4785652108645925e-05, "loss": 0.6275, "step": 10131 }, { "epoch": 0.36097404563834906, "grad_norm": 2.1871314041931433, "learning_rate": 1.4784638860346702e-05, "loss": 0.5089, "step": 10132 }, { "epoch": 0.36100967276484314, "grad_norm": 1.0620798129794882, "learning_rate": 1.4783625548336669e-05, "loss": 0.4634, "step": 10133 }, { "epoch": 0.36104529989133727, "grad_norm": 1.3402083212797335, "learning_rate": 1.478261217262932e-05, "loss": 0.5584, "step": 10134 }, { "epoch": 0.3610809270178314, "grad_norm": 0.9911629250155987, "learning_rate": 1.4781598733238147e-05, "loss": 0.4729, "step": 10135 }, { "epoch": 0.3611165541443255, "grad_norm": 1.668118808747907, "learning_rate": 1.4780585230176646e-05, "loss": 0.5563, "step": 10136 }, { "epoch": 0.3611521812708196, "grad_norm": 1.0534211150473078, "learning_rate": 1.4779571663458308e-05, "loss": 0.5016, "step": 10137 }, { "epoch": 0.36118780839731374, "grad_norm": 1.1746920618213785, "learning_rate": 1.4778558033096633e-05, "loss": 0.4356, "step": 10138 }, { "epoch": 0.3612234355238078, "grad_norm": 1.6695401476598224, "learning_rate": 1.4777544339105121e-05, "loss": 0.5881, "step": 10139 }, { "epoch": 0.36125906265030194, "grad_norm": 1.5659490583884172, "learning_rate": 1.4776530581497261e-05, "loss": 0.5637, "step": 10140 }, { "epoch": 0.3612946897767961, "grad_norm": 1.001963854075711, "learning_rate": 1.4775516760286561e-05, "loss": 0.4623, "step": 10141 }, { "epoch": 0.36133031690329015, "grad_norm": 1.5222381404714977, "learning_rate": 1.4774502875486518e-05, "loss": 0.5602, "step": 10142 }, { "epoch": 0.3613659440297843, "grad_norm": 1.0989154377777484, "learning_rate": 1.4773488927110633e-05, "loss": 0.4736, "step": 10143 }, { "epoch": 0.3614015711562784, "grad_norm": 1.094861258461703, "learning_rate": 1.4772474915172403e-05, "loss": 0.4228, "step": 10144 }, { "epoch": 0.3614371982827725, "grad_norm": 1.1485498324562042, "learning_rate": 1.4771460839685336e-05, "loss": 0.4892, "step": 10145 }, { "epoch": 0.3614728254092666, "grad_norm": 1.7297851908518165, "learning_rate": 1.4770446700662934e-05, "loss": 0.5854, "step": 10146 }, { "epoch": 0.36150845253576075, "grad_norm": 1.3271665426030697, "learning_rate": 1.47694324981187e-05, "loss": 0.5962, "step": 10147 }, { "epoch": 0.3615440796622548, "grad_norm": 1.1589202617529508, "learning_rate": 1.4768418232066135e-05, "loss": 0.4561, "step": 10148 }, { "epoch": 0.36157970678874896, "grad_norm": 1.151721402111669, "learning_rate": 1.476740390251875e-05, "loss": 0.5169, "step": 10149 }, { "epoch": 0.3616153339152431, "grad_norm": 1.5268708896783045, "learning_rate": 1.4766389509490051e-05, "loss": 0.5482, "step": 10150 }, { "epoch": 0.36165096104173716, "grad_norm": 1.420025895564528, "learning_rate": 1.4765375052993543e-05, "loss": 0.5681, "step": 10151 }, { "epoch": 0.3616865881682313, "grad_norm": 1.9217436716269023, "learning_rate": 1.4764360533042734e-05, "loss": 0.6618, "step": 10152 }, { "epoch": 0.3617222152947254, "grad_norm": 1.0873191850375221, "learning_rate": 1.4763345949651137e-05, "loss": 0.4498, "step": 10153 }, { "epoch": 0.3617578424212195, "grad_norm": 1.2475813327968885, "learning_rate": 1.476233130283226e-05, "loss": 0.4983, "step": 10154 }, { "epoch": 0.36179346954771363, "grad_norm": 1.1177798448763632, "learning_rate": 1.4761316592599611e-05, "loss": 0.4682, "step": 10155 }, { "epoch": 0.36182909667420776, "grad_norm": 1.150959544317082, "learning_rate": 1.4760301818966705e-05, "loss": 0.5146, "step": 10156 }, { "epoch": 0.36186472380070184, "grad_norm": 2.419211554153544, "learning_rate": 1.4759286981947052e-05, "loss": 0.5635, "step": 10157 }, { "epoch": 0.36190035092719597, "grad_norm": 1.03844971194706, "learning_rate": 1.4758272081554168e-05, "loss": 0.473, "step": 10158 }, { "epoch": 0.3619359780536901, "grad_norm": 3.0363566704766805, "learning_rate": 1.475725711780156e-05, "loss": 0.5689, "step": 10159 }, { "epoch": 0.3619716051801842, "grad_norm": 1.3232250684799631, "learning_rate": 1.4756242090702756e-05, "loss": 0.5644, "step": 10160 }, { "epoch": 0.3620072323066783, "grad_norm": 1.2757641056356859, "learning_rate": 1.475522700027126e-05, "loss": 0.4758, "step": 10161 }, { "epoch": 0.36204285943317244, "grad_norm": 0.9884981968303617, "learning_rate": 1.4754211846520594e-05, "loss": 0.4841, "step": 10162 }, { "epoch": 0.3620784865596665, "grad_norm": 2.6988873972054193, "learning_rate": 1.475319662946427e-05, "loss": 0.5852, "step": 10163 }, { "epoch": 0.36211411368616064, "grad_norm": 1.337070543694609, "learning_rate": 1.4752181349115816e-05, "loss": 0.5622, "step": 10164 }, { "epoch": 0.3621497408126548, "grad_norm": 1.100605801752073, "learning_rate": 1.4751166005488742e-05, "loss": 0.4993, "step": 10165 }, { "epoch": 0.36218536793914885, "grad_norm": 2.0068612814869815, "learning_rate": 1.4750150598596576e-05, "loss": 0.6315, "step": 10166 }, { "epoch": 0.362220995065643, "grad_norm": 1.8596048158085678, "learning_rate": 1.474913512845283e-05, "loss": 0.6149, "step": 10167 }, { "epoch": 0.3622566221921371, "grad_norm": 1.8993434600669437, "learning_rate": 1.4748119595071033e-05, "loss": 0.5933, "step": 10168 }, { "epoch": 0.3622922493186312, "grad_norm": 1.5895168369096724, "learning_rate": 1.4747103998464704e-05, "loss": 0.5419, "step": 10169 }, { "epoch": 0.3623278764451253, "grad_norm": 1.283625266986905, "learning_rate": 1.4746088338647367e-05, "loss": 0.5216, "step": 10170 }, { "epoch": 0.36236350357161945, "grad_norm": 2.96555033295453, "learning_rate": 1.4745072615632547e-05, "loss": 0.6252, "step": 10171 }, { "epoch": 0.3623991306981135, "grad_norm": 1.4689161811382738, "learning_rate": 1.474405682943377e-05, "loss": 0.461, "step": 10172 }, { "epoch": 0.36243475782460766, "grad_norm": 1.1706188601940932, "learning_rate": 1.4743040980064561e-05, "loss": 0.4881, "step": 10173 }, { "epoch": 0.3624703849511018, "grad_norm": 1.7223253499411888, "learning_rate": 1.4742025067538439e-05, "loss": 0.5471, "step": 10174 }, { "epoch": 0.36250601207759586, "grad_norm": 1.144134884613007, "learning_rate": 1.4741009091868946e-05, "loss": 0.5357, "step": 10175 }, { "epoch": 0.36254163920409, "grad_norm": 0.9996911679701302, "learning_rate": 1.4739993053069601e-05, "loss": 0.4406, "step": 10176 }, { "epoch": 0.3625772663305841, "grad_norm": 1.2938309534384334, "learning_rate": 1.4738976951153935e-05, "loss": 0.5776, "step": 10177 }, { "epoch": 0.3626128934570782, "grad_norm": 1.8095050422818941, "learning_rate": 1.4737960786135479e-05, "loss": 0.6038, "step": 10178 }, { "epoch": 0.36264852058357233, "grad_norm": 1.5527120853827125, "learning_rate": 1.4736944558027764e-05, "loss": 0.5444, "step": 10179 }, { "epoch": 0.36268414771006646, "grad_norm": 1.010217452754868, "learning_rate": 1.4735928266844322e-05, "loss": 0.4239, "step": 10180 }, { "epoch": 0.36271977483656054, "grad_norm": 2.410505102788847, "learning_rate": 1.4734911912598681e-05, "loss": 0.601, "step": 10181 }, { "epoch": 0.36275540196305467, "grad_norm": 1.6601088137623885, "learning_rate": 1.4733895495304382e-05, "loss": 0.616, "step": 10182 }, { "epoch": 0.3627910290895488, "grad_norm": 3.255854972160299, "learning_rate": 1.4732879014974956e-05, "loss": 0.6006, "step": 10183 }, { "epoch": 0.3628266562160429, "grad_norm": 1.765907933910316, "learning_rate": 1.4731862471623938e-05, "loss": 0.5168, "step": 10184 }, { "epoch": 0.362862283342537, "grad_norm": 1.412163979393068, "learning_rate": 1.473084586526486e-05, "loss": 0.5934, "step": 10185 }, { "epoch": 0.36289791046903114, "grad_norm": 1.6212024039875879, "learning_rate": 1.4729829195911265e-05, "loss": 0.5771, "step": 10186 }, { "epoch": 0.3629335375955252, "grad_norm": 1.8211943009366347, "learning_rate": 1.472881246357669e-05, "loss": 0.5869, "step": 10187 }, { "epoch": 0.36296916472201934, "grad_norm": 1.1823529243894069, "learning_rate": 1.472779566827467e-05, "loss": 0.4423, "step": 10188 }, { "epoch": 0.3630047918485135, "grad_norm": 1.1655984524666583, "learning_rate": 1.4726778810018745e-05, "loss": 0.4385, "step": 10189 }, { "epoch": 0.36304041897500755, "grad_norm": 2.1861788385164087, "learning_rate": 1.4725761888822458e-05, "loss": 0.5898, "step": 10190 }, { "epoch": 0.3630760461015017, "grad_norm": 1.2088348279560412, "learning_rate": 1.4724744904699349e-05, "loss": 0.4583, "step": 10191 }, { "epoch": 0.3631116732279958, "grad_norm": 1.2433943051596974, "learning_rate": 1.472372785766296e-05, "loss": 0.5443, "step": 10192 }, { "epoch": 0.3631473003544899, "grad_norm": 1.1980225541505565, "learning_rate": 1.472271074772683e-05, "loss": 0.4723, "step": 10193 }, { "epoch": 0.363182927480984, "grad_norm": 1.2671457615968005, "learning_rate": 1.4721693574904505e-05, "loss": 0.5217, "step": 10194 }, { "epoch": 0.36321855460747815, "grad_norm": 1.2544213006506484, "learning_rate": 1.4720676339209532e-05, "loss": 0.5004, "step": 10195 }, { "epoch": 0.3632541817339722, "grad_norm": 1.7715811086520388, "learning_rate": 1.4719659040655452e-05, "loss": 0.6337, "step": 10196 }, { "epoch": 0.36328980886046636, "grad_norm": 0.9547910010975285, "learning_rate": 1.4718641679255814e-05, "loss": 0.4745, "step": 10197 }, { "epoch": 0.3633254359869605, "grad_norm": 1.6931537189713937, "learning_rate": 1.4717624255024165e-05, "loss": 0.5417, "step": 10198 }, { "epoch": 0.36336106311345456, "grad_norm": 1.0357057321072662, "learning_rate": 1.471660676797405e-05, "loss": 0.443, "step": 10199 }, { "epoch": 0.3633966902399487, "grad_norm": 1.5914879487985, "learning_rate": 1.4715589218119019e-05, "loss": 0.5702, "step": 10200 }, { "epoch": 0.3634323173664428, "grad_norm": 1.9869243502929994, "learning_rate": 1.4714571605472622e-05, "loss": 0.551, "step": 10201 }, { "epoch": 0.3634679444929369, "grad_norm": 1.4294998104228867, "learning_rate": 1.4713553930048413e-05, "loss": 0.4545, "step": 10202 }, { "epoch": 0.36350357161943103, "grad_norm": 1.6909048255978778, "learning_rate": 1.4712536191859934e-05, "loss": 0.547, "step": 10203 }, { "epoch": 0.36353919874592516, "grad_norm": 1.284864607896767, "learning_rate": 1.4711518390920742e-05, "loss": 0.4914, "step": 10204 }, { "epoch": 0.36357482587241924, "grad_norm": 1.3667541892651256, "learning_rate": 1.4710500527244392e-05, "loss": 0.55, "step": 10205 }, { "epoch": 0.36361045299891337, "grad_norm": 1.6848139518519865, "learning_rate": 1.4709482600844433e-05, "loss": 0.4542, "step": 10206 }, { "epoch": 0.3636460801254075, "grad_norm": 1.3355807893398202, "learning_rate": 1.4708464611734421e-05, "loss": 0.5455, "step": 10207 }, { "epoch": 0.3636817072519016, "grad_norm": 1.8360628300401947, "learning_rate": 1.4707446559927914e-05, "loss": 0.5832, "step": 10208 }, { "epoch": 0.3637173343783957, "grad_norm": 1.960743260656092, "learning_rate": 1.4706428445438462e-05, "loss": 0.5464, "step": 10209 }, { "epoch": 0.36375296150488984, "grad_norm": 1.4279377236369033, "learning_rate": 1.470541026827963e-05, "loss": 0.4504, "step": 10210 }, { "epoch": 0.3637885886313839, "grad_norm": 1.1149306162539083, "learning_rate": 1.4704392028464968e-05, "loss": 0.4669, "step": 10211 }, { "epoch": 0.36382421575787804, "grad_norm": 1.1693406848618237, "learning_rate": 1.470337372600804e-05, "loss": 0.4505, "step": 10212 }, { "epoch": 0.3638598428843722, "grad_norm": 2.043710722728934, "learning_rate": 1.4702355360922403e-05, "loss": 0.555, "step": 10213 }, { "epoch": 0.36389547001086625, "grad_norm": 1.5630953367689555, "learning_rate": 1.4701336933221617e-05, "loss": 0.5447, "step": 10214 }, { "epoch": 0.3639310971373604, "grad_norm": 1.6471904519111813, "learning_rate": 1.4700318442919244e-05, "loss": 0.6354, "step": 10215 }, { "epoch": 0.3639667242638545, "grad_norm": 1.2006277674291594, "learning_rate": 1.4699299890028847e-05, "loss": 0.4581, "step": 10216 }, { "epoch": 0.3640023513903486, "grad_norm": 1.5213797118849328, "learning_rate": 1.4698281274563986e-05, "loss": 0.6022, "step": 10217 }, { "epoch": 0.3640379785168427, "grad_norm": 1.676442688342197, "learning_rate": 1.4697262596538227e-05, "loss": 0.5336, "step": 10218 }, { "epoch": 0.36407360564333685, "grad_norm": 1.3270242890489206, "learning_rate": 1.4696243855965131e-05, "loss": 0.5254, "step": 10219 }, { "epoch": 0.3641092327698309, "grad_norm": 1.3297911492455894, "learning_rate": 1.4695225052858264e-05, "loss": 0.5301, "step": 10220 }, { "epoch": 0.36414485989632506, "grad_norm": 1.08625875278871, "learning_rate": 1.4694206187231197e-05, "loss": 0.4772, "step": 10221 }, { "epoch": 0.3641804870228192, "grad_norm": 1.2157970067490091, "learning_rate": 1.4693187259097495e-05, "loss": 0.5094, "step": 10222 }, { "epoch": 0.36421611414931326, "grad_norm": 1.1247485759014872, "learning_rate": 1.4692168268470722e-05, "loss": 0.4336, "step": 10223 }, { "epoch": 0.3642517412758074, "grad_norm": 1.3268726715879178, "learning_rate": 1.4691149215364449e-05, "loss": 0.4377, "step": 10224 }, { "epoch": 0.3642873684023015, "grad_norm": 1.4533813576271033, "learning_rate": 1.4690130099792245e-05, "loss": 0.5259, "step": 10225 }, { "epoch": 0.3643229955287956, "grad_norm": 1.452898079257466, "learning_rate": 1.468911092176768e-05, "loss": 0.6064, "step": 10226 }, { "epoch": 0.36435862265528973, "grad_norm": 1.7526819142092338, "learning_rate": 1.4688091681304327e-05, "loss": 0.5166, "step": 10227 }, { "epoch": 0.36439424978178386, "grad_norm": 1.558478748432791, "learning_rate": 1.4687072378415756e-05, "loss": 0.4537, "step": 10228 }, { "epoch": 0.36442987690827794, "grad_norm": 1.960647579186236, "learning_rate": 1.4686053013115542e-05, "loss": 0.6536, "step": 10229 }, { "epoch": 0.36446550403477207, "grad_norm": 1.5914086757342705, "learning_rate": 1.4685033585417255e-05, "loss": 0.5361, "step": 10230 }, { "epoch": 0.3645011311612662, "grad_norm": 1.196446801194838, "learning_rate": 1.4684014095334472e-05, "loss": 0.494, "step": 10231 }, { "epoch": 0.3645367582877603, "grad_norm": 1.2860702018536447, "learning_rate": 1.4682994542880763e-05, "loss": 0.4735, "step": 10232 }, { "epoch": 0.3645723854142544, "grad_norm": 1.8074772315392285, "learning_rate": 1.4681974928069714e-05, "loss": 0.6196, "step": 10233 }, { "epoch": 0.36460801254074854, "grad_norm": 1.1711781423037755, "learning_rate": 1.4680955250914892e-05, "loss": 0.455, "step": 10234 }, { "epoch": 0.3646436396672426, "grad_norm": 1.4004486178128785, "learning_rate": 1.4679935511429883e-05, "loss": 0.5438, "step": 10235 }, { "epoch": 0.36467926679373674, "grad_norm": 1.1813900698180506, "learning_rate": 1.467891570962826e-05, "loss": 0.5271, "step": 10236 }, { "epoch": 0.3647148939202309, "grad_norm": 1.6088108180553555, "learning_rate": 1.4677895845523606e-05, "loss": 0.5933, "step": 10237 }, { "epoch": 0.36475052104672495, "grad_norm": 1.1335508203413085, "learning_rate": 1.4676875919129497e-05, "loss": 0.4192, "step": 10238 }, { "epoch": 0.3647861481732191, "grad_norm": 1.8339355303615514, "learning_rate": 1.4675855930459516e-05, "loss": 0.5742, "step": 10239 }, { "epoch": 0.3648217752997132, "grad_norm": 1.5460632132590435, "learning_rate": 1.4674835879527245e-05, "loss": 0.5881, "step": 10240 }, { "epoch": 0.3648574024262073, "grad_norm": 1.1736305540322258, "learning_rate": 1.4673815766346267e-05, "loss": 0.5217, "step": 10241 }, { "epoch": 0.3648930295527014, "grad_norm": 1.4695431033533661, "learning_rate": 1.4672795590930165e-05, "loss": 0.5426, "step": 10242 }, { "epoch": 0.36492865667919555, "grad_norm": 1.5240021304241054, "learning_rate": 1.4671775353292525e-05, "loss": 0.6365, "step": 10243 }, { "epoch": 0.3649642838056896, "grad_norm": 1.5170270205444536, "learning_rate": 1.4670755053446927e-05, "loss": 0.616, "step": 10244 }, { "epoch": 0.36499991093218376, "grad_norm": 1.6879606179709727, "learning_rate": 1.4669734691406965e-05, "loss": 0.5585, "step": 10245 }, { "epoch": 0.3650355380586779, "grad_norm": 3.0527757264813724, "learning_rate": 1.4668714267186219e-05, "loss": 0.6812, "step": 10246 }, { "epoch": 0.36507116518517196, "grad_norm": 1.218819858795048, "learning_rate": 1.4667693780798282e-05, "loss": 0.5007, "step": 10247 }, { "epoch": 0.3651067923116661, "grad_norm": 1.3413187961861257, "learning_rate": 1.4666673232256738e-05, "loss": 0.4939, "step": 10248 }, { "epoch": 0.3651424194381602, "grad_norm": 1.3088125520115403, "learning_rate": 1.4665652621575177e-05, "loss": 0.4902, "step": 10249 }, { "epoch": 0.3651780465646543, "grad_norm": 1.791483022357779, "learning_rate": 1.4664631948767195e-05, "loss": 0.5588, "step": 10250 }, { "epoch": 0.36521367369114843, "grad_norm": 1.7539667032009958, "learning_rate": 1.4663611213846374e-05, "loss": 0.5693, "step": 10251 }, { "epoch": 0.36524930081764256, "grad_norm": 1.4040568872115584, "learning_rate": 1.466259041682631e-05, "loss": 0.5477, "step": 10252 }, { "epoch": 0.36528492794413664, "grad_norm": 1.8384126579155518, "learning_rate": 1.4661569557720596e-05, "loss": 0.5303, "step": 10253 }, { "epoch": 0.36532055507063077, "grad_norm": 1.3057620265144998, "learning_rate": 1.4660548636542828e-05, "loss": 0.4279, "step": 10254 }, { "epoch": 0.3653561821971249, "grad_norm": 1.6043866400369025, "learning_rate": 1.4659527653306593e-05, "loss": 0.6529, "step": 10255 }, { "epoch": 0.365391809323619, "grad_norm": 1.6642053580751044, "learning_rate": 1.4658506608025493e-05, "loss": 0.4702, "step": 10256 }, { "epoch": 0.3654274364501131, "grad_norm": 1.4953385958374634, "learning_rate": 1.465748550071312e-05, "loss": 0.4317, "step": 10257 }, { "epoch": 0.36546306357660724, "grad_norm": 1.8379287365511978, "learning_rate": 1.4656464331383073e-05, "loss": 0.4627, "step": 10258 }, { "epoch": 0.3654986907031013, "grad_norm": 2.15508214129944, "learning_rate": 1.4655443100048951e-05, "loss": 0.4994, "step": 10259 }, { "epoch": 0.36553431782959545, "grad_norm": 1.4906147928586326, "learning_rate": 1.4654421806724346e-05, "loss": 0.5117, "step": 10260 }, { "epoch": 0.3655699449560896, "grad_norm": 1.7277969963681188, "learning_rate": 1.4653400451422863e-05, "loss": 0.4848, "step": 10261 }, { "epoch": 0.36560557208258365, "grad_norm": 2.329796373301259, "learning_rate": 1.4652379034158104e-05, "loss": 0.5175, "step": 10262 }, { "epoch": 0.3656411992090778, "grad_norm": 2.929675487611234, "learning_rate": 1.4651357554943662e-05, "loss": 0.4775, "step": 10263 }, { "epoch": 0.3656768263355719, "grad_norm": 1.6684797210319864, "learning_rate": 1.4650336013793145e-05, "loss": 0.5741, "step": 10264 }, { "epoch": 0.365712453462066, "grad_norm": 2.6726581171295964, "learning_rate": 1.4649314410720153e-05, "loss": 0.5031, "step": 10265 }, { "epoch": 0.3657480805885601, "grad_norm": 2.49387522567204, "learning_rate": 1.4648292745738291e-05, "loss": 0.5287, "step": 10266 }, { "epoch": 0.36578370771505425, "grad_norm": 2.967201654753114, "learning_rate": 1.464727101886116e-05, "loss": 0.5138, "step": 10267 }, { "epoch": 0.3658193348415484, "grad_norm": 2.577626108032218, "learning_rate": 1.4646249230102364e-05, "loss": 0.5072, "step": 10268 }, { "epoch": 0.36585496196804246, "grad_norm": 2.8165025528784833, "learning_rate": 1.4645227379475518e-05, "loss": 0.5879, "step": 10269 }, { "epoch": 0.3658905890945366, "grad_norm": 2.1384173516487945, "learning_rate": 1.464420546699422e-05, "loss": 0.5781, "step": 10270 }, { "epoch": 0.3659262162210307, "grad_norm": 2.5717156781502277, "learning_rate": 1.4643183492672081e-05, "loss": 0.4862, "step": 10271 }, { "epoch": 0.3659618433475248, "grad_norm": 1.4393365816085268, "learning_rate": 1.4642161456522708e-05, "loss": 0.5037, "step": 10272 }, { "epoch": 0.3659974704740189, "grad_norm": 1.9095018773049461, "learning_rate": 1.4641139358559714e-05, "loss": 0.6116, "step": 10273 }, { "epoch": 0.36603309760051306, "grad_norm": 2.099231151150184, "learning_rate": 1.4640117198796702e-05, "loss": 0.5207, "step": 10274 }, { "epoch": 0.36606872472700713, "grad_norm": 1.6703419673722455, "learning_rate": 1.4639094977247284e-05, "loss": 0.5954, "step": 10275 }, { "epoch": 0.36610435185350126, "grad_norm": 4.205377225968543, "learning_rate": 1.4638072693925076e-05, "loss": 0.4553, "step": 10276 }, { "epoch": 0.3661399789799954, "grad_norm": 2.7694608272389396, "learning_rate": 1.4637050348843691e-05, "loss": 0.4839, "step": 10277 }, { "epoch": 0.36617560610648947, "grad_norm": 2.932517947228459, "learning_rate": 1.4636027942016736e-05, "loss": 0.544, "step": 10278 }, { "epoch": 0.3662112332329836, "grad_norm": 1.7246948971237825, "learning_rate": 1.4635005473457828e-05, "loss": 0.5378, "step": 10279 }, { "epoch": 0.36624686035947773, "grad_norm": 1.821459750665228, "learning_rate": 1.4633982943180585e-05, "loss": 0.6088, "step": 10280 }, { "epoch": 0.3662824874859718, "grad_norm": 3.0985781482617987, "learning_rate": 1.463296035119862e-05, "loss": 0.5591, "step": 10281 }, { "epoch": 0.36631811461246594, "grad_norm": 2.7669941066494514, "learning_rate": 1.4631937697525546e-05, "loss": 0.4405, "step": 10282 }, { "epoch": 0.36635374173896007, "grad_norm": 2.883861753901301, "learning_rate": 1.4630914982174988e-05, "loss": 0.493, "step": 10283 }, { "epoch": 0.36638936886545415, "grad_norm": 7.182640915668635, "learning_rate": 1.462989220516056e-05, "loss": 0.4676, "step": 10284 }, { "epoch": 0.3664249959919483, "grad_norm": 2.0268963668474327, "learning_rate": 1.4628869366495881e-05, "loss": 0.6779, "step": 10285 }, { "epoch": 0.3664606231184424, "grad_norm": 2.9615329377831547, "learning_rate": 1.4627846466194569e-05, "loss": 0.6492, "step": 10286 }, { "epoch": 0.3664962502449365, "grad_norm": 2.6742508417942483, "learning_rate": 1.4626823504270248e-05, "loss": 0.5205, "step": 10287 }, { "epoch": 0.3665318773714306, "grad_norm": 3.0340993184841394, "learning_rate": 1.4625800480736539e-05, "loss": 0.4935, "step": 10288 }, { "epoch": 0.36656750449792475, "grad_norm": 2.0139086211002675, "learning_rate": 1.4624777395607061e-05, "loss": 0.5647, "step": 10289 }, { "epoch": 0.3666031316244188, "grad_norm": 1.9413294966069334, "learning_rate": 1.4623754248895439e-05, "loss": 0.5646, "step": 10290 }, { "epoch": 0.36663875875091295, "grad_norm": 1.4785748577557762, "learning_rate": 1.46227310406153e-05, "loss": 0.5134, "step": 10291 }, { "epoch": 0.3666743858774071, "grad_norm": 1.6668418169345163, "learning_rate": 1.4621707770780267e-05, "loss": 0.5819, "step": 10292 }, { "epoch": 0.36671001300390116, "grad_norm": 1.4101671007334584, "learning_rate": 1.4620684439403962e-05, "loss": 0.5304, "step": 10293 }, { "epoch": 0.3667456401303953, "grad_norm": 1.702729198135723, "learning_rate": 1.4619661046500015e-05, "loss": 0.5076, "step": 10294 }, { "epoch": 0.3667812672568894, "grad_norm": 1.399977874140156, "learning_rate": 1.4618637592082053e-05, "loss": 0.4966, "step": 10295 }, { "epoch": 0.3668168943833835, "grad_norm": 1.1720342491005642, "learning_rate": 1.4617614076163702e-05, "loss": 0.4727, "step": 10296 }, { "epoch": 0.3668525215098776, "grad_norm": 1.1830424652799, "learning_rate": 1.4616590498758591e-05, "loss": 0.4756, "step": 10297 }, { "epoch": 0.36688814863637176, "grad_norm": 1.5356056803218505, "learning_rate": 1.4615566859880353e-05, "loss": 0.5763, "step": 10298 }, { "epoch": 0.36692377576286583, "grad_norm": 1.7443371580659697, "learning_rate": 1.4614543159542618e-05, "loss": 0.5266, "step": 10299 }, { "epoch": 0.36695940288935996, "grad_norm": 1.2504112424603433, "learning_rate": 1.4613519397759013e-05, "loss": 0.4732, "step": 10300 }, { "epoch": 0.3669950300158541, "grad_norm": 1.2781498126762338, "learning_rate": 1.4612495574543169e-05, "loss": 0.4815, "step": 10301 }, { "epoch": 0.36703065714234817, "grad_norm": 1.3118670746309193, "learning_rate": 1.4611471689908725e-05, "loss": 0.5486, "step": 10302 }, { "epoch": 0.3670662842688423, "grad_norm": 1.1883192308039408, "learning_rate": 1.4610447743869313e-05, "loss": 0.5239, "step": 10303 }, { "epoch": 0.36710191139533643, "grad_norm": 1.6511325284816307, "learning_rate": 1.4609423736438568e-05, "loss": 0.5101, "step": 10304 }, { "epoch": 0.3671375385218305, "grad_norm": 1.565298157498943, "learning_rate": 1.4608399667630122e-05, "loss": 0.5546, "step": 10305 }, { "epoch": 0.36717316564832464, "grad_norm": 1.7970050948570955, "learning_rate": 1.4607375537457616e-05, "loss": 0.583, "step": 10306 }, { "epoch": 0.36720879277481877, "grad_norm": 2.53296213433459, "learning_rate": 1.4606351345934684e-05, "loss": 0.4422, "step": 10307 }, { "epoch": 0.36724441990131285, "grad_norm": 1.8480469686087049, "learning_rate": 1.4605327093074961e-05, "loss": 0.5436, "step": 10308 }, { "epoch": 0.367280047027807, "grad_norm": 1.4881800939739933, "learning_rate": 1.4604302778892091e-05, "loss": 0.4908, "step": 10309 }, { "epoch": 0.3673156741543011, "grad_norm": 1.914279542984461, "learning_rate": 1.4603278403399712e-05, "loss": 0.6554, "step": 10310 }, { "epoch": 0.3673513012807952, "grad_norm": 1.216561459544384, "learning_rate": 1.4602253966611462e-05, "loss": 0.4925, "step": 10311 }, { "epoch": 0.3673869284072893, "grad_norm": 1.6722439944703462, "learning_rate": 1.4601229468540985e-05, "loss": 0.5815, "step": 10312 }, { "epoch": 0.36742255553378345, "grad_norm": 1.4925980262311165, "learning_rate": 1.4600204909201919e-05, "loss": 0.515, "step": 10313 }, { "epoch": 0.3674581826602775, "grad_norm": 1.7994374273877005, "learning_rate": 1.4599180288607911e-05, "loss": 0.6564, "step": 10314 }, { "epoch": 0.36749380978677165, "grad_norm": 1.4079263188279358, "learning_rate": 1.4598155606772605e-05, "loss": 0.5156, "step": 10315 }, { "epoch": 0.3675294369132658, "grad_norm": 1.889558450443965, "learning_rate": 1.4597130863709638e-05, "loss": 0.5957, "step": 10316 }, { "epoch": 0.36756506403975986, "grad_norm": 1.8098657569476024, "learning_rate": 1.4596106059432664e-05, "loss": 0.5182, "step": 10317 }, { "epoch": 0.367600691166254, "grad_norm": 1.0805473174068858, "learning_rate": 1.4595081193955325e-05, "loss": 0.5076, "step": 10318 }, { "epoch": 0.3676363182927481, "grad_norm": 1.2446611141457253, "learning_rate": 1.4594056267291268e-05, "loss": 0.5182, "step": 10319 }, { "epoch": 0.3676719454192422, "grad_norm": 1.5711952679764298, "learning_rate": 1.4593031279454142e-05, "loss": 0.5578, "step": 10320 }, { "epoch": 0.3677075725457363, "grad_norm": 1.1605748785318633, "learning_rate": 1.4592006230457592e-05, "loss": 0.537, "step": 10321 }, { "epoch": 0.36774319967223046, "grad_norm": 1.0948920953737222, "learning_rate": 1.4590981120315271e-05, "loss": 0.4524, "step": 10322 }, { "epoch": 0.36777882679872453, "grad_norm": 2.0140082635244947, "learning_rate": 1.4589955949040823e-05, "loss": 0.5245, "step": 10323 }, { "epoch": 0.36781445392521867, "grad_norm": 1.1135906386294292, "learning_rate": 1.458893071664791e-05, "loss": 0.5092, "step": 10324 }, { "epoch": 0.3678500810517128, "grad_norm": 1.2019184278116855, "learning_rate": 1.4587905423150173e-05, "loss": 0.5036, "step": 10325 }, { "epoch": 0.36788570817820687, "grad_norm": 1.452139568516623, "learning_rate": 1.4586880068561269e-05, "loss": 0.5696, "step": 10326 }, { "epoch": 0.367921335304701, "grad_norm": 0.9473766026341733, "learning_rate": 1.458585465289485e-05, "loss": 0.462, "step": 10327 }, { "epoch": 0.36795696243119513, "grad_norm": 1.4535670549072517, "learning_rate": 1.4584829176164575e-05, "loss": 0.5187, "step": 10328 }, { "epoch": 0.3679925895576892, "grad_norm": 1.321895829342862, "learning_rate": 1.4583803638384093e-05, "loss": 0.5001, "step": 10329 }, { "epoch": 0.36802821668418334, "grad_norm": 1.8701920731172843, "learning_rate": 1.458277803956706e-05, "loss": 0.5002, "step": 10330 }, { "epoch": 0.36806384381067747, "grad_norm": 1.82135557254393, "learning_rate": 1.4581752379727134e-05, "loss": 0.6448, "step": 10331 }, { "epoch": 0.36809947093717155, "grad_norm": 1.5375574526021107, "learning_rate": 1.4580726658877975e-05, "loss": 0.5195, "step": 10332 }, { "epoch": 0.3681350980636657, "grad_norm": 1.3806242849822086, "learning_rate": 1.4579700877033238e-05, "loss": 0.5706, "step": 10333 }, { "epoch": 0.3681707251901598, "grad_norm": 1.4053321557416518, "learning_rate": 1.4578675034206582e-05, "loss": 0.5128, "step": 10334 }, { "epoch": 0.3682063523166539, "grad_norm": 1.1832656767317258, "learning_rate": 1.4577649130411669e-05, "loss": 0.5068, "step": 10335 }, { "epoch": 0.368241979443148, "grad_norm": 2.287907156772185, "learning_rate": 1.4576623165662157e-05, "loss": 0.5862, "step": 10336 }, { "epoch": 0.36827760656964215, "grad_norm": 1.9506618419205952, "learning_rate": 1.4575597139971708e-05, "loss": 0.6218, "step": 10337 }, { "epoch": 0.3683132336961362, "grad_norm": 3.363646823991494, "learning_rate": 1.4574571053353987e-05, "loss": 0.613, "step": 10338 }, { "epoch": 0.36834886082263035, "grad_norm": 1.0913942899144202, "learning_rate": 1.4573544905822652e-05, "loss": 0.4641, "step": 10339 }, { "epoch": 0.3683844879491245, "grad_norm": 1.4957859774633557, "learning_rate": 1.4572518697391374e-05, "loss": 0.6273, "step": 10340 }, { "epoch": 0.36842011507561856, "grad_norm": 1.2037946160438247, "learning_rate": 1.457149242807381e-05, "loss": 0.452, "step": 10341 }, { "epoch": 0.3684557422021127, "grad_norm": 2.084934981666745, "learning_rate": 1.4570466097883631e-05, "loss": 0.6507, "step": 10342 }, { "epoch": 0.3684913693286068, "grad_norm": 1.0874931006914705, "learning_rate": 1.45694397068345e-05, "loss": 0.4761, "step": 10343 }, { "epoch": 0.3685269964551009, "grad_norm": 1.5807176219055237, "learning_rate": 1.4568413254940086e-05, "loss": 0.6041, "step": 10344 }, { "epoch": 0.36856262358159503, "grad_norm": 1.1076887479638013, "learning_rate": 1.4567386742214055e-05, "loss": 0.4913, "step": 10345 }, { "epoch": 0.36859825070808916, "grad_norm": 1.6208561194683928, "learning_rate": 1.456636016867008e-05, "loss": 0.5788, "step": 10346 }, { "epoch": 0.36863387783458323, "grad_norm": 1.1800189616688155, "learning_rate": 1.4565333534321827e-05, "loss": 0.4954, "step": 10347 }, { "epoch": 0.36866950496107737, "grad_norm": 1.177875350941021, "learning_rate": 1.4564306839182965e-05, "loss": 0.5114, "step": 10348 }, { "epoch": 0.3687051320875715, "grad_norm": 1.4218587174949497, "learning_rate": 1.4563280083267171e-05, "loss": 0.5265, "step": 10349 }, { "epoch": 0.36874075921406557, "grad_norm": 1.3639521840959066, "learning_rate": 1.4562253266588109e-05, "loss": 0.5648, "step": 10350 }, { "epoch": 0.3687763863405597, "grad_norm": 1.3559440011799713, "learning_rate": 1.4561226389159457e-05, "loss": 0.6221, "step": 10351 }, { "epoch": 0.36881201346705383, "grad_norm": 1.2927305289121214, "learning_rate": 1.456019945099489e-05, "loss": 0.602, "step": 10352 }, { "epoch": 0.3688476405935479, "grad_norm": 1.422993453602145, "learning_rate": 1.4559172452108077e-05, "loss": 0.5882, "step": 10353 }, { "epoch": 0.36888326772004204, "grad_norm": 1.114896425132517, "learning_rate": 1.4558145392512696e-05, "loss": 0.5129, "step": 10354 }, { "epoch": 0.36891889484653617, "grad_norm": 1.2963685242337326, "learning_rate": 1.4557118272222426e-05, "loss": 0.5577, "step": 10355 }, { "epoch": 0.36895452197303025, "grad_norm": 1.3919929677467104, "learning_rate": 1.4556091091250939e-05, "loss": 0.607, "step": 10356 }, { "epoch": 0.3689901490995244, "grad_norm": 1.6301442924525136, "learning_rate": 1.4555063849611915e-05, "loss": 0.4946, "step": 10357 }, { "epoch": 0.3690257762260185, "grad_norm": 1.067848483988703, "learning_rate": 1.4554036547319033e-05, "loss": 0.5018, "step": 10358 }, { "epoch": 0.3690614033525126, "grad_norm": 1.1556292665331085, "learning_rate": 1.4553009184385966e-05, "loss": 0.4942, "step": 10359 }, { "epoch": 0.3690970304790067, "grad_norm": 0.9911330760748109, "learning_rate": 1.4551981760826407e-05, "loss": 0.4816, "step": 10360 }, { "epoch": 0.36913265760550085, "grad_norm": 1.2264995083123493, "learning_rate": 1.4550954276654023e-05, "loss": 0.5088, "step": 10361 }, { "epoch": 0.3691682847319949, "grad_norm": 1.0448358902773713, "learning_rate": 1.4549926731882505e-05, "loss": 0.4338, "step": 10362 }, { "epoch": 0.36920391185848905, "grad_norm": 1.5012022932785252, "learning_rate": 1.4548899126525532e-05, "loss": 0.5682, "step": 10363 }, { "epoch": 0.3692395389849832, "grad_norm": 1.4571344538108446, "learning_rate": 1.4547871460596787e-05, "loss": 0.5309, "step": 10364 }, { "epoch": 0.36927516611147726, "grad_norm": 1.1666061398363483, "learning_rate": 1.4546843734109957e-05, "loss": 0.4883, "step": 10365 }, { "epoch": 0.3693107932379714, "grad_norm": 1.1833595268606587, "learning_rate": 1.4545815947078722e-05, "loss": 0.5077, "step": 10366 }, { "epoch": 0.3693464203644655, "grad_norm": 1.127643165681648, "learning_rate": 1.4544788099516773e-05, "loss": 0.4955, "step": 10367 }, { "epoch": 0.3693820474909596, "grad_norm": 1.2429981282099318, "learning_rate": 1.454376019143779e-05, "loss": 0.5564, "step": 10368 }, { "epoch": 0.36941767461745373, "grad_norm": 1.4993209499436522, "learning_rate": 1.454273222285547e-05, "loss": 0.6431, "step": 10369 }, { "epoch": 0.36945330174394786, "grad_norm": 1.8037557722768154, "learning_rate": 1.454170419378349e-05, "loss": 0.5634, "step": 10370 }, { "epoch": 0.36948892887044194, "grad_norm": 1.0431694480369003, "learning_rate": 1.4540676104235545e-05, "loss": 0.4782, "step": 10371 }, { "epoch": 0.36952455599693607, "grad_norm": 1.1054853193882235, "learning_rate": 1.4539647954225325e-05, "loss": 0.4999, "step": 10372 }, { "epoch": 0.3695601831234302, "grad_norm": 1.6637998059935812, "learning_rate": 1.4538619743766518e-05, "loss": 0.6374, "step": 10373 }, { "epoch": 0.3695958102499243, "grad_norm": 1.4984109731455268, "learning_rate": 1.4537591472872819e-05, "loss": 0.4925, "step": 10374 }, { "epoch": 0.3696314373764184, "grad_norm": 1.1506943609818057, "learning_rate": 1.4536563141557918e-05, "loss": 0.5224, "step": 10375 }, { "epoch": 0.36966706450291253, "grad_norm": 1.5324481344456935, "learning_rate": 1.4535534749835505e-05, "loss": 0.549, "step": 10376 }, { "epoch": 0.3697026916294066, "grad_norm": 1.6921052959223986, "learning_rate": 1.453450629771928e-05, "loss": 0.5279, "step": 10377 }, { "epoch": 0.36973831875590074, "grad_norm": 1.7290076005634742, "learning_rate": 1.4533477785222933e-05, "loss": 0.6766, "step": 10378 }, { "epoch": 0.36977394588239487, "grad_norm": 1.0661991540999285, "learning_rate": 1.4532449212360158e-05, "loss": 0.4583, "step": 10379 }, { "epoch": 0.36980957300888895, "grad_norm": 1.8993744237163939, "learning_rate": 1.4531420579144657e-05, "loss": 0.5657, "step": 10380 }, { "epoch": 0.3698452001353831, "grad_norm": 1.4938645089888223, "learning_rate": 1.4530391885590125e-05, "loss": 0.5819, "step": 10381 }, { "epoch": 0.3698808272618772, "grad_norm": 1.3869656888078363, "learning_rate": 1.4529363131710254e-05, "loss": 0.6166, "step": 10382 }, { "epoch": 0.3699164543883713, "grad_norm": 1.1294697303002508, "learning_rate": 1.452833431751875e-05, "loss": 0.4942, "step": 10383 }, { "epoch": 0.3699520815148654, "grad_norm": 1.7366713713437747, "learning_rate": 1.4527305443029307e-05, "loss": 0.6351, "step": 10384 }, { "epoch": 0.36998770864135955, "grad_norm": 1.1065537901955083, "learning_rate": 1.452627650825563e-05, "loss": 0.4847, "step": 10385 }, { "epoch": 0.3700233357678536, "grad_norm": 1.1832251423221778, "learning_rate": 1.4525247513211418e-05, "loss": 0.5274, "step": 10386 }, { "epoch": 0.37005896289434775, "grad_norm": 1.1721484795103032, "learning_rate": 1.452421845791037e-05, "loss": 0.4911, "step": 10387 }, { "epoch": 0.3700945900208419, "grad_norm": 1.2923294046785032, "learning_rate": 1.4523189342366192e-05, "loss": 0.4997, "step": 10388 }, { "epoch": 0.37013021714733596, "grad_norm": 1.0355947570806796, "learning_rate": 1.4522160166592589e-05, "loss": 0.4939, "step": 10389 }, { "epoch": 0.3701658442738301, "grad_norm": 1.485673470281522, "learning_rate": 1.4521130930603259e-05, "loss": 0.5993, "step": 10390 }, { "epoch": 0.3702014714003242, "grad_norm": 1.6928268874041883, "learning_rate": 1.4520101634411912e-05, "loss": 0.6131, "step": 10391 }, { "epoch": 0.3702370985268183, "grad_norm": 1.057546862718344, "learning_rate": 1.4519072278032254e-05, "loss": 0.4934, "step": 10392 }, { "epoch": 0.37027272565331243, "grad_norm": 2.5388347872009702, "learning_rate": 1.4518042861477988e-05, "loss": 0.5198, "step": 10393 }, { "epoch": 0.37030835277980656, "grad_norm": 1.3471478538879949, "learning_rate": 1.4517013384762821e-05, "loss": 0.5726, "step": 10394 }, { "epoch": 0.37034397990630064, "grad_norm": 1.015155656054382, "learning_rate": 1.4515983847900466e-05, "loss": 0.4396, "step": 10395 }, { "epoch": 0.37037960703279477, "grad_norm": 1.0456343129383572, "learning_rate": 1.4514954250904632e-05, "loss": 0.511, "step": 10396 }, { "epoch": 0.3704152341592889, "grad_norm": 1.7947103117574716, "learning_rate": 1.4513924593789025e-05, "loss": 0.6153, "step": 10397 }, { "epoch": 0.370450861285783, "grad_norm": 1.0563912400661306, "learning_rate": 1.4512894876567355e-05, "loss": 0.4568, "step": 10398 }, { "epoch": 0.3704864884122771, "grad_norm": 1.1185763592382747, "learning_rate": 1.4511865099253335e-05, "loss": 0.4608, "step": 10399 }, { "epoch": 0.37052211553877124, "grad_norm": 1.7220512805665273, "learning_rate": 1.451083526186068e-05, "loss": 0.6727, "step": 10400 }, { "epoch": 0.3705577426652653, "grad_norm": 2.4723965830604095, "learning_rate": 1.45098053644031e-05, "loss": 0.5595, "step": 10401 }, { "epoch": 0.37059336979175944, "grad_norm": 1.0248168364157384, "learning_rate": 1.4508775406894308e-05, "loss": 0.5089, "step": 10402 }, { "epoch": 0.3706289969182536, "grad_norm": 1.1220490978578836, "learning_rate": 1.4507745389348023e-05, "loss": 0.4967, "step": 10403 }, { "epoch": 0.37066462404474765, "grad_norm": 1.5787282198493031, "learning_rate": 1.4506715311777956e-05, "loss": 0.6283, "step": 10404 }, { "epoch": 0.3707002511712418, "grad_norm": 1.159688365763086, "learning_rate": 1.4505685174197823e-05, "loss": 0.4263, "step": 10405 }, { "epoch": 0.3707358782977359, "grad_norm": 1.1016218763796501, "learning_rate": 1.4504654976621342e-05, "loss": 0.516, "step": 10406 }, { "epoch": 0.37077150542423, "grad_norm": 1.1776602061594443, "learning_rate": 1.4503624719062234e-05, "loss": 0.4992, "step": 10407 }, { "epoch": 0.3708071325507241, "grad_norm": 1.2435598048116292, "learning_rate": 1.4502594401534216e-05, "loss": 0.5122, "step": 10408 }, { "epoch": 0.37084275967721825, "grad_norm": 1.5803422188911151, "learning_rate": 1.4501564024051003e-05, "loss": 0.5228, "step": 10409 }, { "epoch": 0.3708783868037123, "grad_norm": 1.5795723849020005, "learning_rate": 1.4500533586626322e-05, "loss": 0.6049, "step": 10410 }, { "epoch": 0.37091401393020645, "grad_norm": 1.1794126853968159, "learning_rate": 1.449950308927389e-05, "loss": 0.5132, "step": 10411 }, { "epoch": 0.3709496410567006, "grad_norm": 1.5861986918589503, "learning_rate": 1.4498472532007429e-05, "loss": 0.6739, "step": 10412 }, { "epoch": 0.37098526818319466, "grad_norm": 1.2052943802946148, "learning_rate": 1.449744191484066e-05, "loss": 0.4866, "step": 10413 }, { "epoch": 0.3710208953096888, "grad_norm": 1.0688591677647838, "learning_rate": 1.449641123778731e-05, "loss": 0.4806, "step": 10414 }, { "epoch": 0.3710565224361829, "grad_norm": 2.2692932779970416, "learning_rate": 1.44953805008611e-05, "loss": 0.5403, "step": 10415 }, { "epoch": 0.371092149562677, "grad_norm": 1.742046890740364, "learning_rate": 1.4494349704075758e-05, "loss": 0.5597, "step": 10416 }, { "epoch": 0.37112777668917113, "grad_norm": 1.4872590478349603, "learning_rate": 1.4493318847445007e-05, "loss": 0.5606, "step": 10417 }, { "epoch": 0.37116340381566526, "grad_norm": 1.2252403764696977, "learning_rate": 1.4492287930982575e-05, "loss": 0.5152, "step": 10418 }, { "epoch": 0.37119903094215934, "grad_norm": 1.26508037818445, "learning_rate": 1.4491256954702193e-05, "loss": 0.4693, "step": 10419 }, { "epoch": 0.37123465806865347, "grad_norm": 3.079728510845186, "learning_rate": 1.4490225918617583e-05, "loss": 0.5538, "step": 10420 }, { "epoch": 0.3712702851951476, "grad_norm": 1.169504662014861, "learning_rate": 1.4489194822742474e-05, "loss": 0.4895, "step": 10421 }, { "epoch": 0.3713059123216417, "grad_norm": 3.639604254160936, "learning_rate": 1.4488163667090603e-05, "loss": 0.569, "step": 10422 }, { "epoch": 0.3713415394481358, "grad_norm": 1.1902429331332254, "learning_rate": 1.4487132451675692e-05, "loss": 0.4566, "step": 10423 }, { "epoch": 0.37137716657462994, "grad_norm": 1.0860992306753987, "learning_rate": 1.4486101176511476e-05, "loss": 0.5053, "step": 10424 }, { "epoch": 0.371412793701124, "grad_norm": 1.73617610078565, "learning_rate": 1.448506984161169e-05, "loss": 0.4782, "step": 10425 }, { "epoch": 0.37144842082761814, "grad_norm": 1.5449491409575424, "learning_rate": 1.4484038446990063e-05, "loss": 0.5601, "step": 10426 }, { "epoch": 0.3714840479541123, "grad_norm": 1.7026109439812454, "learning_rate": 1.4483006992660328e-05, "loss": 0.4827, "step": 10427 }, { "epoch": 0.37151967508060635, "grad_norm": 1.112025947079699, "learning_rate": 1.448197547863622e-05, "loss": 0.4731, "step": 10428 }, { "epoch": 0.3715553022071005, "grad_norm": 1.9883500138040235, "learning_rate": 1.4480943904931481e-05, "loss": 0.692, "step": 10429 }, { "epoch": 0.3715909293335946, "grad_norm": 1.2752436927433504, "learning_rate": 1.447991227155984e-05, "loss": 0.4793, "step": 10430 }, { "epoch": 0.3716265564600887, "grad_norm": 1.071092003534608, "learning_rate": 1.4478880578535038e-05, "loss": 0.4408, "step": 10431 }, { "epoch": 0.3716621835865828, "grad_norm": 1.7368642419210718, "learning_rate": 1.4477848825870807e-05, "loss": 0.6116, "step": 10432 }, { "epoch": 0.37169781071307695, "grad_norm": 1.1011183800006263, "learning_rate": 1.4476817013580889e-05, "loss": 0.4722, "step": 10433 }, { "epoch": 0.371733437839571, "grad_norm": 1.5690264891896701, "learning_rate": 1.4475785141679027e-05, "loss": 0.5582, "step": 10434 }, { "epoch": 0.37176906496606515, "grad_norm": 1.1585864936212815, "learning_rate": 1.4474753210178953e-05, "loss": 0.4791, "step": 10435 }, { "epoch": 0.3718046920925593, "grad_norm": 1.638979190837036, "learning_rate": 1.4473721219094414e-05, "loss": 0.5917, "step": 10436 }, { "epoch": 0.37184031921905336, "grad_norm": 1.3814098988069843, "learning_rate": 1.4472689168439152e-05, "loss": 0.5496, "step": 10437 }, { "epoch": 0.3718759463455475, "grad_norm": 1.9532828684858734, "learning_rate": 1.4471657058226906e-05, "loss": 0.5808, "step": 10438 }, { "epoch": 0.3719115734720416, "grad_norm": 1.6884172476314558, "learning_rate": 1.447062488847142e-05, "loss": 0.5361, "step": 10439 }, { "epoch": 0.3719472005985357, "grad_norm": 1.7542864757607795, "learning_rate": 1.4469592659186438e-05, "loss": 0.561, "step": 10440 }, { "epoch": 0.37198282772502983, "grad_norm": 1.1525340808637188, "learning_rate": 1.4468560370385706e-05, "loss": 0.5191, "step": 10441 }, { "epoch": 0.37201845485152396, "grad_norm": 1.897102956571461, "learning_rate": 1.4467528022082972e-05, "loss": 0.6016, "step": 10442 }, { "epoch": 0.37205408197801804, "grad_norm": 1.6841232378767912, "learning_rate": 1.4466495614291977e-05, "loss": 0.5561, "step": 10443 }, { "epoch": 0.37208970910451217, "grad_norm": 1.1979030348809852, "learning_rate": 1.4465463147026472e-05, "loss": 0.4373, "step": 10444 }, { "epoch": 0.3721253362310063, "grad_norm": 1.7518158180168106, "learning_rate": 1.4464430620300205e-05, "loss": 0.536, "step": 10445 }, { "epoch": 0.3721609633575004, "grad_norm": 1.7640602329244477, "learning_rate": 1.446339803412692e-05, "loss": 0.5602, "step": 10446 }, { "epoch": 0.3721965904839945, "grad_norm": 1.2727014410473179, "learning_rate": 1.4462365388520376e-05, "loss": 0.4964, "step": 10447 }, { "epoch": 0.37223221761048864, "grad_norm": 1.6752413855317858, "learning_rate": 1.4461332683494317e-05, "loss": 0.5304, "step": 10448 }, { "epoch": 0.3722678447369827, "grad_norm": 1.6848431504030525, "learning_rate": 1.4460299919062495e-05, "loss": 0.511, "step": 10449 }, { "epoch": 0.37230347186347684, "grad_norm": 1.2258861895863522, "learning_rate": 1.4459267095238659e-05, "loss": 0.4958, "step": 10450 }, { "epoch": 0.372339098989971, "grad_norm": 1.1312046381018748, "learning_rate": 1.4458234212036563e-05, "loss": 0.4881, "step": 10451 }, { "epoch": 0.37237472611646505, "grad_norm": 1.5365435022307914, "learning_rate": 1.4457201269469967e-05, "loss": 0.5871, "step": 10452 }, { "epoch": 0.3724103532429592, "grad_norm": 1.3728856051555693, "learning_rate": 1.4456168267552619e-05, "loss": 0.5137, "step": 10453 }, { "epoch": 0.3724459803694533, "grad_norm": 1.5660070429511186, "learning_rate": 1.4455135206298275e-05, "loss": 0.5064, "step": 10454 }, { "epoch": 0.3724816074959474, "grad_norm": 1.0415863218510972, "learning_rate": 1.4454102085720693e-05, "loss": 0.4853, "step": 10455 }, { "epoch": 0.3725172346224415, "grad_norm": 1.6696920879163595, "learning_rate": 1.4453068905833629e-05, "loss": 0.7062, "step": 10456 }, { "epoch": 0.37255286174893565, "grad_norm": 0.9923964480126902, "learning_rate": 1.4452035666650841e-05, "loss": 0.4767, "step": 10457 }, { "epoch": 0.3725884888754298, "grad_norm": 1.6507935586637916, "learning_rate": 1.4451002368186084e-05, "loss": 0.6132, "step": 10458 }, { "epoch": 0.37262411600192386, "grad_norm": 1.151245130478754, "learning_rate": 1.444996901045312e-05, "loss": 0.476, "step": 10459 }, { "epoch": 0.372659743128418, "grad_norm": 1.4645870532436382, "learning_rate": 1.4448935593465709e-05, "loss": 0.6278, "step": 10460 }, { "epoch": 0.3726953702549121, "grad_norm": 1.1142579061628584, "learning_rate": 1.4447902117237608e-05, "loss": 0.4711, "step": 10461 }, { "epoch": 0.3727309973814062, "grad_norm": 1.551985410082704, "learning_rate": 1.4446868581782586e-05, "loss": 0.4692, "step": 10462 }, { "epoch": 0.3727666245079003, "grad_norm": 0.9768235647375426, "learning_rate": 1.4445834987114395e-05, "loss": 0.4091, "step": 10463 }, { "epoch": 0.37280225163439445, "grad_norm": 1.7125520101392717, "learning_rate": 1.4444801333246809e-05, "loss": 0.5716, "step": 10464 }, { "epoch": 0.37283787876088853, "grad_norm": 1.671357547891567, "learning_rate": 1.4443767620193582e-05, "loss": 0.5863, "step": 10465 }, { "epoch": 0.37287350588738266, "grad_norm": 1.0895695162498058, "learning_rate": 1.4442733847968488e-05, "loss": 0.4535, "step": 10466 }, { "epoch": 0.3729091330138768, "grad_norm": 2.050912916800308, "learning_rate": 1.4441700016585287e-05, "loss": 0.593, "step": 10467 }, { "epoch": 0.37294476014037087, "grad_norm": 1.7624801112956472, "learning_rate": 1.4440666126057743e-05, "loss": 0.5991, "step": 10468 }, { "epoch": 0.372980387266865, "grad_norm": 1.2476896316363577, "learning_rate": 1.4439632176399628e-05, "loss": 0.5006, "step": 10469 }, { "epoch": 0.37301601439335913, "grad_norm": 1.1977370233539757, "learning_rate": 1.4438598167624704e-05, "loss": 0.4856, "step": 10470 }, { "epoch": 0.3730516415198532, "grad_norm": 1.2655442422044134, "learning_rate": 1.4437564099746746e-05, "loss": 0.4991, "step": 10471 }, { "epoch": 0.37308726864634734, "grad_norm": 1.172648402899086, "learning_rate": 1.4436529972779519e-05, "loss": 0.541, "step": 10472 }, { "epoch": 0.37312289577284147, "grad_norm": 2.0649994654292807, "learning_rate": 1.4435495786736796e-05, "loss": 0.5862, "step": 10473 }, { "epoch": 0.37315852289933554, "grad_norm": 1.6205011091872021, "learning_rate": 1.4434461541632344e-05, "loss": 0.5802, "step": 10474 }, { "epoch": 0.3731941500258297, "grad_norm": 1.4796270933878812, "learning_rate": 1.4433427237479938e-05, "loss": 0.5172, "step": 10475 }, { "epoch": 0.3732297771523238, "grad_norm": 1.6762383682114543, "learning_rate": 1.443239287429335e-05, "loss": 0.5028, "step": 10476 }, { "epoch": 0.3732654042788179, "grad_norm": 2.122001269005764, "learning_rate": 1.4431358452086353e-05, "loss": 0.585, "step": 10477 }, { "epoch": 0.373301031405312, "grad_norm": 1.3677492546906755, "learning_rate": 1.4430323970872721e-05, "loss": 0.4662, "step": 10478 }, { "epoch": 0.37333665853180614, "grad_norm": 1.4612430977855229, "learning_rate": 1.4429289430666229e-05, "loss": 0.5694, "step": 10479 }, { "epoch": 0.3733722856583002, "grad_norm": 1.164184918513269, "learning_rate": 1.4428254831480651e-05, "loss": 0.4712, "step": 10480 }, { "epoch": 0.37340791278479435, "grad_norm": 1.3142032374613106, "learning_rate": 1.4427220173329766e-05, "loss": 0.5459, "step": 10481 }, { "epoch": 0.3734435399112885, "grad_norm": 1.5258372035789198, "learning_rate": 1.4426185456227348e-05, "loss": 0.5985, "step": 10482 }, { "epoch": 0.37347916703778256, "grad_norm": 1.1468428006575773, "learning_rate": 1.4425150680187181e-05, "loss": 0.4764, "step": 10483 }, { "epoch": 0.3735147941642767, "grad_norm": 1.6673688396455377, "learning_rate": 1.4424115845223035e-05, "loss": 0.6871, "step": 10484 }, { "epoch": 0.3735504212907708, "grad_norm": 3.3126484604117223, "learning_rate": 1.4423080951348696e-05, "loss": 0.7046, "step": 10485 }, { "epoch": 0.3735860484172649, "grad_norm": 1.7477800169471354, "learning_rate": 1.4422045998577943e-05, "loss": 0.4866, "step": 10486 }, { "epoch": 0.373621675543759, "grad_norm": 1.7565548036132281, "learning_rate": 1.4421010986924557e-05, "loss": 0.4912, "step": 10487 }, { "epoch": 0.37365730267025316, "grad_norm": 1.340663719847498, "learning_rate": 1.4419975916402317e-05, "loss": 0.5707, "step": 10488 }, { "epoch": 0.37369292979674723, "grad_norm": 1.4984783901430099, "learning_rate": 1.4418940787025011e-05, "loss": 0.6571, "step": 10489 }, { "epoch": 0.37372855692324136, "grad_norm": 1.9948420722313558, "learning_rate": 1.4417905598806422e-05, "loss": 0.5715, "step": 10490 }, { "epoch": 0.3737641840497355, "grad_norm": 1.1437020473509367, "learning_rate": 1.4416870351760325e-05, "loss": 0.4673, "step": 10491 }, { "epoch": 0.37379981117622957, "grad_norm": 1.2592214118573488, "learning_rate": 1.4415835045900516e-05, "loss": 0.5106, "step": 10492 }, { "epoch": 0.3738354383027237, "grad_norm": 1.0255515955807322, "learning_rate": 1.4414799681240781e-05, "loss": 0.4624, "step": 10493 }, { "epoch": 0.37387106542921783, "grad_norm": 1.2399239975137915, "learning_rate": 1.4413764257794898e-05, "loss": 0.4859, "step": 10494 }, { "epoch": 0.3739066925557119, "grad_norm": 1.2659238287109917, "learning_rate": 1.4412728775576658e-05, "loss": 0.4771, "step": 10495 }, { "epoch": 0.37394231968220604, "grad_norm": 1.1800866045520941, "learning_rate": 1.4411693234599853e-05, "loss": 0.5149, "step": 10496 }, { "epoch": 0.37397794680870017, "grad_norm": 1.0434344319701006, "learning_rate": 1.4410657634878267e-05, "loss": 0.4539, "step": 10497 }, { "epoch": 0.37401357393519424, "grad_norm": 1.4951927609028717, "learning_rate": 1.4409621976425691e-05, "loss": 0.5932, "step": 10498 }, { "epoch": 0.3740492010616884, "grad_norm": 1.1249977598280259, "learning_rate": 1.4408586259255917e-05, "loss": 0.432, "step": 10499 }, { "epoch": 0.3740848281881825, "grad_norm": 1.2754876233630317, "learning_rate": 1.4407550483382737e-05, "loss": 0.5281, "step": 10500 }, { "epoch": 0.3741204553146766, "grad_norm": 3.021442279781738, "learning_rate": 1.4406514648819943e-05, "loss": 0.6287, "step": 10501 }, { "epoch": 0.3741560824411707, "grad_norm": 1.1337687652633686, "learning_rate": 1.4405478755581323e-05, "loss": 0.4627, "step": 10502 }, { "epoch": 0.37419170956766484, "grad_norm": 1.0909631204198904, "learning_rate": 1.4404442803680673e-05, "loss": 0.4456, "step": 10503 }, { "epoch": 0.3742273366941589, "grad_norm": 1.480605787709433, "learning_rate": 1.4403406793131792e-05, "loss": 0.5569, "step": 10504 }, { "epoch": 0.37426296382065305, "grad_norm": 1.9882449932876074, "learning_rate": 1.4402370723948471e-05, "loss": 0.5945, "step": 10505 }, { "epoch": 0.3742985909471472, "grad_norm": 1.2164103970136626, "learning_rate": 1.4401334596144506e-05, "loss": 0.4903, "step": 10506 }, { "epoch": 0.37433421807364126, "grad_norm": 1.4257397698101681, "learning_rate": 1.4400298409733695e-05, "loss": 0.6789, "step": 10507 }, { "epoch": 0.3743698452001354, "grad_norm": 1.143989882856147, "learning_rate": 1.4399262164729835e-05, "loss": 0.4473, "step": 10508 }, { "epoch": 0.3744054723266295, "grad_norm": 1.2431450362470056, "learning_rate": 1.4398225861146724e-05, "loss": 0.5317, "step": 10509 }, { "epoch": 0.3744410994531236, "grad_norm": 2.003710218316471, "learning_rate": 1.4397189498998161e-05, "loss": 0.6311, "step": 10510 }, { "epoch": 0.3744767265796177, "grad_norm": 1.145669598276039, "learning_rate": 1.4396153078297948e-05, "loss": 0.463, "step": 10511 }, { "epoch": 0.37451235370611186, "grad_norm": 1.1746962781802608, "learning_rate": 1.4395116599059882e-05, "loss": 0.5262, "step": 10512 }, { "epoch": 0.37454798083260593, "grad_norm": 1.7369470724780378, "learning_rate": 1.439408006129777e-05, "loss": 0.6139, "step": 10513 }, { "epoch": 0.37458360795910006, "grad_norm": 1.0910965547717768, "learning_rate": 1.4393043465025405e-05, "loss": 0.5251, "step": 10514 }, { "epoch": 0.3746192350855942, "grad_norm": 1.3092274638870127, "learning_rate": 1.43920068102566e-05, "loss": 0.4767, "step": 10515 }, { "epoch": 0.37465486221208827, "grad_norm": 1.0305524436390385, "learning_rate": 1.4390970097005155e-05, "loss": 0.4905, "step": 10516 }, { "epoch": 0.3746904893385824, "grad_norm": 2.052233356529317, "learning_rate": 1.4389933325284871e-05, "loss": 0.4915, "step": 10517 }, { "epoch": 0.37472611646507653, "grad_norm": 1.0661552129800858, "learning_rate": 1.438889649510956e-05, "loss": 0.461, "step": 10518 }, { "epoch": 0.3747617435915706, "grad_norm": 1.1415674959631623, "learning_rate": 1.4387859606493022e-05, "loss": 0.5108, "step": 10519 }, { "epoch": 0.37479737071806474, "grad_norm": 1.1539233688926294, "learning_rate": 1.4386822659449067e-05, "loss": 0.5051, "step": 10520 }, { "epoch": 0.37483299784455887, "grad_norm": 1.712615782508649, "learning_rate": 1.43857856539915e-05, "loss": 0.6366, "step": 10521 }, { "epoch": 0.37486862497105294, "grad_norm": 1.2059458487379549, "learning_rate": 1.4384748590134134e-05, "loss": 0.5064, "step": 10522 }, { "epoch": 0.3749042520975471, "grad_norm": 1.4481559701974245, "learning_rate": 1.4383711467890776e-05, "loss": 0.5331, "step": 10523 }, { "epoch": 0.3749398792240412, "grad_norm": 1.6412006427453492, "learning_rate": 1.4382674287275235e-05, "loss": 0.5575, "step": 10524 }, { "epoch": 0.3749755063505353, "grad_norm": 1.154723677959244, "learning_rate": 1.4381637048301323e-05, "loss": 0.4729, "step": 10525 }, { "epoch": 0.3750111334770294, "grad_norm": 1.3907812645542759, "learning_rate": 1.4380599750982848e-05, "loss": 0.5399, "step": 10526 }, { "epoch": 0.37504676060352354, "grad_norm": 1.5306917044573711, "learning_rate": 1.437956239533363e-05, "loss": 0.587, "step": 10527 }, { "epoch": 0.3750823877300176, "grad_norm": 1.1204904944487433, "learning_rate": 1.4378524981367474e-05, "loss": 0.501, "step": 10528 }, { "epoch": 0.37511801485651175, "grad_norm": 1.1716768360033833, "learning_rate": 1.4377487509098203e-05, "loss": 0.524, "step": 10529 }, { "epoch": 0.3751536419830059, "grad_norm": 1.4148491401734296, "learning_rate": 1.4376449978539621e-05, "loss": 0.5316, "step": 10530 }, { "epoch": 0.37518926910949996, "grad_norm": 1.1171035442865478, "learning_rate": 1.437541238970555e-05, "loss": 0.4402, "step": 10531 }, { "epoch": 0.3752248962359941, "grad_norm": 1.5920089333322014, "learning_rate": 1.4374374742609805e-05, "loss": 0.5292, "step": 10532 }, { "epoch": 0.3752605233624882, "grad_norm": 1.85223419987506, "learning_rate": 1.43733370372662e-05, "loss": 0.6184, "step": 10533 }, { "epoch": 0.3752961504889823, "grad_norm": 1.167911455426749, "learning_rate": 1.437229927368856e-05, "loss": 0.4919, "step": 10534 }, { "epoch": 0.3753317776154764, "grad_norm": 1.616599574164797, "learning_rate": 1.4371261451890697e-05, "loss": 0.5194, "step": 10535 }, { "epoch": 0.37536740474197056, "grad_norm": 1.0722919145727976, "learning_rate": 1.4370223571886432e-05, "loss": 0.4558, "step": 10536 }, { "epoch": 0.37540303186846463, "grad_norm": 1.1746655367377368, "learning_rate": 1.4369185633689587e-05, "loss": 0.4922, "step": 10537 }, { "epoch": 0.37543865899495876, "grad_norm": 1.9878840251065155, "learning_rate": 1.436814763731398e-05, "loss": 0.5525, "step": 10538 }, { "epoch": 0.3754742861214529, "grad_norm": 1.0757198129950427, "learning_rate": 1.4367109582773435e-05, "loss": 0.4508, "step": 10539 }, { "epoch": 0.37550991324794697, "grad_norm": 1.4951477732257965, "learning_rate": 1.4366071470081773e-05, "loss": 0.6315, "step": 10540 }, { "epoch": 0.3755455403744411, "grad_norm": 1.0298161563920842, "learning_rate": 1.436503329925282e-05, "loss": 0.5041, "step": 10541 }, { "epoch": 0.37558116750093523, "grad_norm": 2.375709886080901, "learning_rate": 1.4363995070300395e-05, "loss": 0.5374, "step": 10542 }, { "epoch": 0.3756167946274293, "grad_norm": 1.0717797425126447, "learning_rate": 1.4362956783238325e-05, "loss": 0.4475, "step": 10543 }, { "epoch": 0.37565242175392344, "grad_norm": 1.4162084586393227, "learning_rate": 1.4361918438080436e-05, "loss": 0.5801, "step": 10544 }, { "epoch": 0.37568804888041757, "grad_norm": 1.1733103529797977, "learning_rate": 1.4360880034840554e-05, "loss": 0.501, "step": 10545 }, { "epoch": 0.37572367600691164, "grad_norm": 1.1839156079358468, "learning_rate": 1.4359841573532509e-05, "loss": 0.5181, "step": 10546 }, { "epoch": 0.3757593031334058, "grad_norm": 2.2934366786394174, "learning_rate": 1.4358803054170122e-05, "loss": 0.6594, "step": 10547 }, { "epoch": 0.3757949302598999, "grad_norm": 1.2272801034684406, "learning_rate": 1.435776447676723e-05, "loss": 0.5104, "step": 10548 }, { "epoch": 0.375830557386394, "grad_norm": 1.0765142436175157, "learning_rate": 1.4356725841337655e-05, "loss": 0.4607, "step": 10549 }, { "epoch": 0.3758661845128881, "grad_norm": 1.0982963695338985, "learning_rate": 1.4355687147895232e-05, "loss": 0.5289, "step": 10550 }, { "epoch": 0.37590181163938224, "grad_norm": 1.1441347209786619, "learning_rate": 1.4354648396453789e-05, "loss": 0.4765, "step": 10551 }, { "epoch": 0.3759374387658763, "grad_norm": 1.1473743285683127, "learning_rate": 1.435360958702716e-05, "loss": 0.4855, "step": 10552 }, { "epoch": 0.37597306589237045, "grad_norm": 1.2897351453956285, "learning_rate": 1.4352570719629177e-05, "loss": 0.5467, "step": 10553 }, { "epoch": 0.3760086930188646, "grad_norm": 1.4369949999873506, "learning_rate": 1.435153179427367e-05, "loss": 0.5905, "step": 10554 }, { "epoch": 0.37604432014535866, "grad_norm": 1.1521191250521117, "learning_rate": 1.4350492810974474e-05, "loss": 0.4622, "step": 10555 }, { "epoch": 0.3760799472718528, "grad_norm": 1.156898197226079, "learning_rate": 1.4349453769745431e-05, "loss": 0.476, "step": 10556 }, { "epoch": 0.3761155743983469, "grad_norm": 1.1243440289234652, "learning_rate": 1.434841467060037e-05, "loss": 0.4922, "step": 10557 }, { "epoch": 0.376151201524841, "grad_norm": 1.0883375216263202, "learning_rate": 1.4347375513553126e-05, "loss": 0.4282, "step": 10558 }, { "epoch": 0.3761868286513351, "grad_norm": 1.6781906269313809, "learning_rate": 1.4346336298617537e-05, "loss": 0.6454, "step": 10559 }, { "epoch": 0.37622245577782926, "grad_norm": 1.3203788342218057, "learning_rate": 1.4345297025807447e-05, "loss": 0.49, "step": 10560 }, { "epoch": 0.37625808290432333, "grad_norm": 1.6669564442493676, "learning_rate": 1.4344257695136687e-05, "loss": 0.559, "step": 10561 }, { "epoch": 0.37629371003081746, "grad_norm": 1.8730007320862707, "learning_rate": 1.4343218306619098e-05, "loss": 0.5623, "step": 10562 }, { "epoch": 0.3763293371573116, "grad_norm": 1.6650372170578678, "learning_rate": 1.4342178860268523e-05, "loss": 0.6532, "step": 10563 }, { "epoch": 0.37636496428380567, "grad_norm": 1.7192503044843164, "learning_rate": 1.4341139356098803e-05, "loss": 0.6173, "step": 10564 }, { "epoch": 0.3764005914102998, "grad_norm": 1.5521051894521285, "learning_rate": 1.4340099794123779e-05, "loss": 0.5688, "step": 10565 }, { "epoch": 0.37643621853679393, "grad_norm": 1.7892017465722712, "learning_rate": 1.4339060174357289e-05, "loss": 0.553, "step": 10566 }, { "epoch": 0.376471845663288, "grad_norm": 1.2439851358095075, "learning_rate": 1.433802049681318e-05, "loss": 0.5, "step": 10567 }, { "epoch": 0.37650747278978214, "grad_norm": 1.4440626857150876, "learning_rate": 1.4336980761505297e-05, "loss": 0.5356, "step": 10568 }, { "epoch": 0.37654309991627627, "grad_norm": 1.846878107381894, "learning_rate": 1.4335940968447486e-05, "loss": 0.6049, "step": 10569 }, { "epoch": 0.37657872704277034, "grad_norm": 2.0469453208964996, "learning_rate": 1.4334901117653588e-05, "loss": 0.5873, "step": 10570 }, { "epoch": 0.3766143541692645, "grad_norm": 1.5449581345842536, "learning_rate": 1.4333861209137453e-05, "loss": 0.6167, "step": 10571 }, { "epoch": 0.3766499812957586, "grad_norm": 1.5727330139120985, "learning_rate": 1.4332821242912927e-05, "loss": 0.4605, "step": 10572 }, { "epoch": 0.3766856084222527, "grad_norm": 1.1759290158347064, "learning_rate": 1.4331781218993855e-05, "loss": 0.4872, "step": 10573 }, { "epoch": 0.3767212355487468, "grad_norm": 1.87363093554923, "learning_rate": 1.4330741137394091e-05, "loss": 0.5433, "step": 10574 }, { "epoch": 0.37675686267524094, "grad_norm": 1.2069762119962286, "learning_rate": 1.432970099812748e-05, "loss": 0.4287, "step": 10575 }, { "epoch": 0.376792489801735, "grad_norm": 1.5975840989238708, "learning_rate": 1.4328660801207876e-05, "loss": 0.5934, "step": 10576 }, { "epoch": 0.37682811692822915, "grad_norm": 1.5589895177475244, "learning_rate": 1.4327620546649126e-05, "loss": 0.5669, "step": 10577 }, { "epoch": 0.3768637440547233, "grad_norm": 1.1207521315139097, "learning_rate": 1.4326580234465084e-05, "loss": 0.452, "step": 10578 }, { "epoch": 0.37689937118121736, "grad_norm": 1.9133448386540555, "learning_rate": 1.4325539864669602e-05, "loss": 0.5479, "step": 10579 }, { "epoch": 0.3769349983077115, "grad_norm": 1.6380905656583447, "learning_rate": 1.4324499437276536e-05, "loss": 0.5089, "step": 10580 }, { "epoch": 0.3769706254342056, "grad_norm": 1.2304615231727993, "learning_rate": 1.4323458952299731e-05, "loss": 0.4811, "step": 10581 }, { "epoch": 0.3770062525606997, "grad_norm": 1.4564072407503068, "learning_rate": 1.4322418409753053e-05, "loss": 0.5005, "step": 10582 }, { "epoch": 0.3770418796871938, "grad_norm": 1.3345343067791269, "learning_rate": 1.4321377809650355e-05, "loss": 0.5022, "step": 10583 }, { "epoch": 0.37707750681368796, "grad_norm": 2.5374566445934637, "learning_rate": 1.4320337152005487e-05, "loss": 0.495, "step": 10584 }, { "epoch": 0.37711313394018203, "grad_norm": 2.413322835408531, "learning_rate": 1.431929643683231e-05, "loss": 0.5707, "step": 10585 }, { "epoch": 0.37714876106667616, "grad_norm": 1.431878833744725, "learning_rate": 1.4318255664144682e-05, "loss": 0.4763, "step": 10586 }, { "epoch": 0.3771843881931703, "grad_norm": 1.7207657714318196, "learning_rate": 1.4317214833956465e-05, "loss": 0.4972, "step": 10587 }, { "epoch": 0.37722001531966437, "grad_norm": 1.338490837925332, "learning_rate": 1.431617394628151e-05, "loss": 0.5245, "step": 10588 }, { "epoch": 0.3772556424461585, "grad_norm": 1.2134646017554476, "learning_rate": 1.4315133001133686e-05, "loss": 0.5266, "step": 10589 }, { "epoch": 0.37729126957265263, "grad_norm": 1.2926517868413636, "learning_rate": 1.4314091998526846e-05, "loss": 0.4359, "step": 10590 }, { "epoch": 0.3773268966991467, "grad_norm": 2.2524364654174343, "learning_rate": 1.431305093847486e-05, "loss": 0.5105, "step": 10591 }, { "epoch": 0.37736252382564084, "grad_norm": 1.548734790972212, "learning_rate": 1.4312009820991584e-05, "loss": 0.5781, "step": 10592 }, { "epoch": 0.37739815095213497, "grad_norm": 1.4190385942615928, "learning_rate": 1.4310968646090884e-05, "loss": 0.4863, "step": 10593 }, { "epoch": 0.37743377807862905, "grad_norm": 1.9314823249037727, "learning_rate": 1.4309927413786624e-05, "loss": 0.6209, "step": 10594 }, { "epoch": 0.3774694052051232, "grad_norm": 1.711904314121831, "learning_rate": 1.4308886124092668e-05, "loss": 0.5777, "step": 10595 }, { "epoch": 0.3775050323316173, "grad_norm": 1.62298737848262, "learning_rate": 1.4307844777022876e-05, "loss": 0.515, "step": 10596 }, { "epoch": 0.3775406594581114, "grad_norm": 2.0521961367613017, "learning_rate": 1.4306803372591126e-05, "loss": 0.6808, "step": 10597 }, { "epoch": 0.3775762865846055, "grad_norm": 1.1568392843243307, "learning_rate": 1.4305761910811278e-05, "loss": 0.4799, "step": 10598 }, { "epoch": 0.37761191371109964, "grad_norm": 1.354288925873252, "learning_rate": 1.43047203916972e-05, "loss": 0.5165, "step": 10599 }, { "epoch": 0.3776475408375937, "grad_norm": 1.0592270022413897, "learning_rate": 1.430367881526276e-05, "loss": 0.4301, "step": 10600 }, { "epoch": 0.37768316796408785, "grad_norm": 1.4920470714803806, "learning_rate": 1.4302637181521828e-05, "loss": 0.5991, "step": 10601 }, { "epoch": 0.377718795090582, "grad_norm": 1.1657811043891162, "learning_rate": 1.4301595490488275e-05, "loss": 0.4422, "step": 10602 }, { "epoch": 0.37775442221707606, "grad_norm": 1.164463606124045, "learning_rate": 1.4300553742175974e-05, "loss": 0.494, "step": 10603 }, { "epoch": 0.3777900493435702, "grad_norm": 1.3461897794380913, "learning_rate": 1.4299511936598791e-05, "loss": 0.4978, "step": 10604 }, { "epoch": 0.3778256764700643, "grad_norm": 1.1403421655868016, "learning_rate": 1.4298470073770605e-05, "loss": 0.4904, "step": 10605 }, { "epoch": 0.3778613035965584, "grad_norm": 1.6416538988308051, "learning_rate": 1.4297428153705282e-05, "loss": 0.6284, "step": 10606 }, { "epoch": 0.3778969307230525, "grad_norm": 1.143288202279748, "learning_rate": 1.4296386176416701e-05, "loss": 0.3824, "step": 10607 }, { "epoch": 0.37793255784954666, "grad_norm": 1.481575544881668, "learning_rate": 1.4295344141918734e-05, "loss": 0.6255, "step": 10608 }, { "epoch": 0.37796818497604073, "grad_norm": 1.1724636151430143, "learning_rate": 1.429430205022526e-05, "loss": 0.4491, "step": 10609 }, { "epoch": 0.37800381210253486, "grad_norm": 1.1817787046835795, "learning_rate": 1.429325990135015e-05, "loss": 0.4722, "step": 10610 }, { "epoch": 0.378039439229029, "grad_norm": 1.7940694582408312, "learning_rate": 1.4292217695307285e-05, "loss": 0.4871, "step": 10611 }, { "epoch": 0.37807506635552307, "grad_norm": 1.4122835711314847, "learning_rate": 1.429117543211054e-05, "loss": 0.6269, "step": 10612 }, { "epoch": 0.3781106934820172, "grad_norm": 1.856189109781434, "learning_rate": 1.4290133111773794e-05, "loss": 0.5374, "step": 10613 }, { "epoch": 0.37814632060851133, "grad_norm": 1.2105201433112684, "learning_rate": 1.4289090734310929e-05, "loss": 0.45, "step": 10614 }, { "epoch": 0.3781819477350054, "grad_norm": 1.6218891966945819, "learning_rate": 1.4288048299735821e-05, "loss": 0.5247, "step": 10615 }, { "epoch": 0.37821757486149954, "grad_norm": 1.1842040889060905, "learning_rate": 1.4287005808062355e-05, "loss": 0.4914, "step": 10616 }, { "epoch": 0.37825320198799367, "grad_norm": 1.863419242285418, "learning_rate": 1.4285963259304413e-05, "loss": 0.6388, "step": 10617 }, { "epoch": 0.37828882911448775, "grad_norm": 1.2027725364866355, "learning_rate": 1.4284920653475867e-05, "loss": 0.513, "step": 10618 }, { "epoch": 0.3783244562409819, "grad_norm": 1.1842846809297942, "learning_rate": 1.4283877990590613e-05, "loss": 0.463, "step": 10619 }, { "epoch": 0.378360083367476, "grad_norm": 1.5267816700999826, "learning_rate": 1.4282835270662528e-05, "loss": 0.5644, "step": 10620 }, { "epoch": 0.3783957104939701, "grad_norm": 1.6873852840953214, "learning_rate": 1.4281792493705497e-05, "loss": 0.5837, "step": 10621 }, { "epoch": 0.3784313376204642, "grad_norm": 1.2169329018525963, "learning_rate": 1.4280749659733406e-05, "loss": 0.5067, "step": 10622 }, { "epoch": 0.37846696474695835, "grad_norm": 1.1605102976255648, "learning_rate": 1.4279706768760142e-05, "loss": 0.4944, "step": 10623 }, { "epoch": 0.3785025918734524, "grad_norm": 1.5044242845999076, "learning_rate": 1.4278663820799591e-05, "loss": 0.5608, "step": 10624 }, { "epoch": 0.37853821899994655, "grad_norm": 1.4145843364212896, "learning_rate": 1.427762081586564e-05, "loss": 0.5595, "step": 10625 }, { "epoch": 0.3785738461264407, "grad_norm": 1.1151798079108397, "learning_rate": 1.4276577753972179e-05, "loss": 0.453, "step": 10626 }, { "epoch": 0.37860947325293476, "grad_norm": 1.070599987332411, "learning_rate": 1.4275534635133097e-05, "loss": 0.4634, "step": 10627 }, { "epoch": 0.3786451003794289, "grad_norm": 1.929946447532694, "learning_rate": 1.427449145936228e-05, "loss": 0.5305, "step": 10628 }, { "epoch": 0.378680727505923, "grad_norm": 1.2154121688660198, "learning_rate": 1.4273448226673622e-05, "loss": 0.4795, "step": 10629 }, { "epoch": 0.3787163546324171, "grad_norm": 1.1317726368570886, "learning_rate": 1.4272404937081017e-05, "loss": 0.4645, "step": 10630 }, { "epoch": 0.3787519817589112, "grad_norm": 1.4684158233677072, "learning_rate": 1.4271361590598351e-05, "loss": 0.5764, "step": 10631 }, { "epoch": 0.37878760888540536, "grad_norm": 2.7798241353986963, "learning_rate": 1.4270318187239522e-05, "loss": 0.628, "step": 10632 }, { "epoch": 0.37882323601189943, "grad_norm": 1.677990269404431, "learning_rate": 1.4269274727018419e-05, "loss": 0.5848, "step": 10633 }, { "epoch": 0.37885886313839356, "grad_norm": 1.571269309455779, "learning_rate": 1.4268231209948942e-05, "loss": 0.616, "step": 10634 }, { "epoch": 0.3788944902648877, "grad_norm": 1.1602651313177608, "learning_rate": 1.426718763604498e-05, "loss": 0.4585, "step": 10635 }, { "epoch": 0.37893011739138177, "grad_norm": 1.8425016384087263, "learning_rate": 1.4266144005320435e-05, "loss": 0.6129, "step": 10636 }, { "epoch": 0.3789657445178759, "grad_norm": 1.0892493736385656, "learning_rate": 1.4265100317789198e-05, "loss": 0.4999, "step": 10637 }, { "epoch": 0.37900137164437003, "grad_norm": 1.5193959081951867, "learning_rate": 1.426405657346517e-05, "loss": 0.5404, "step": 10638 }, { "epoch": 0.3790369987708641, "grad_norm": 1.069609860110364, "learning_rate": 1.426301277236225e-05, "loss": 0.4511, "step": 10639 }, { "epoch": 0.37907262589735824, "grad_norm": 1.4642492095990778, "learning_rate": 1.4261968914494335e-05, "loss": 0.5421, "step": 10640 }, { "epoch": 0.37910825302385237, "grad_norm": 1.2091253543424365, "learning_rate": 1.4260924999875323e-05, "loss": 0.4668, "step": 10641 }, { "epoch": 0.37914388015034645, "grad_norm": 1.929982670880839, "learning_rate": 1.4259881028519118e-05, "loss": 0.5882, "step": 10642 }, { "epoch": 0.3791795072768406, "grad_norm": 1.119717214693846, "learning_rate": 1.4258837000439618e-05, "loss": 0.4819, "step": 10643 }, { "epoch": 0.3792151344033347, "grad_norm": 1.1383310747403321, "learning_rate": 1.4257792915650728e-05, "loss": 0.5439, "step": 10644 }, { "epoch": 0.37925076152982884, "grad_norm": 1.188766035108719, "learning_rate": 1.4256748774166348e-05, "loss": 0.5037, "step": 10645 }, { "epoch": 0.3792863886563229, "grad_norm": 2.209334262734086, "learning_rate": 1.4255704576000387e-05, "loss": 0.5974, "step": 10646 }, { "epoch": 0.37932201578281705, "grad_norm": 1.6239864108339688, "learning_rate": 1.425466032116674e-05, "loss": 0.4855, "step": 10647 }, { "epoch": 0.3793576429093112, "grad_norm": 1.2313283482713424, "learning_rate": 1.4253616009679317e-05, "loss": 0.4631, "step": 10648 }, { "epoch": 0.37939327003580525, "grad_norm": 1.1451034142480092, "learning_rate": 1.4252571641552026e-05, "loss": 0.4884, "step": 10649 }, { "epoch": 0.3794288971622994, "grad_norm": 1.6909390433833524, "learning_rate": 1.425152721679877e-05, "loss": 0.5668, "step": 10650 }, { "epoch": 0.3794645242887935, "grad_norm": 1.4965483699156374, "learning_rate": 1.4250482735433459e-05, "loss": 0.5467, "step": 10651 }, { "epoch": 0.3795001514152876, "grad_norm": 1.0944199293876198, "learning_rate": 1.4249438197469997e-05, "loss": 0.4531, "step": 10652 }, { "epoch": 0.3795357785417817, "grad_norm": 1.4939692346428537, "learning_rate": 1.4248393602922299e-05, "loss": 0.5556, "step": 10653 }, { "epoch": 0.37957140566827585, "grad_norm": 1.22654889128677, "learning_rate": 1.424734895180427e-05, "loss": 0.4861, "step": 10654 }, { "epoch": 0.3796070327947699, "grad_norm": 2.337040960563444, "learning_rate": 1.4246304244129819e-05, "loss": 0.556, "step": 10655 }, { "epoch": 0.37964265992126406, "grad_norm": 1.6419777023777282, "learning_rate": 1.4245259479912859e-05, "loss": 0.4431, "step": 10656 }, { "epoch": 0.3796782870477582, "grad_norm": 1.3663165871416438, "learning_rate": 1.4244214659167304e-05, "loss": 0.4802, "step": 10657 }, { "epoch": 0.37971391417425226, "grad_norm": 1.3916553628412425, "learning_rate": 1.4243169781907062e-05, "loss": 0.6242, "step": 10658 }, { "epoch": 0.3797495413007464, "grad_norm": 1.6032265301295994, "learning_rate": 1.4242124848146048e-05, "loss": 0.5954, "step": 10659 }, { "epoch": 0.3797851684272405, "grad_norm": 1.5087524754590085, "learning_rate": 1.4241079857898177e-05, "loss": 0.6005, "step": 10660 }, { "epoch": 0.3798207955537346, "grad_norm": 1.041505367232782, "learning_rate": 1.4240034811177364e-05, "loss": 0.4522, "step": 10661 }, { "epoch": 0.37985642268022873, "grad_norm": 1.6455361664048662, "learning_rate": 1.4238989707997526e-05, "loss": 0.5417, "step": 10662 }, { "epoch": 0.37989204980672286, "grad_norm": 1.805758122165093, "learning_rate": 1.4237944548372572e-05, "loss": 0.5082, "step": 10663 }, { "epoch": 0.37992767693321694, "grad_norm": 1.3185784599985, "learning_rate": 1.4236899332316427e-05, "loss": 0.4672, "step": 10664 }, { "epoch": 0.37996330405971107, "grad_norm": 1.6881962138219229, "learning_rate": 1.4235854059843008e-05, "loss": 0.5713, "step": 10665 }, { "epoch": 0.3799989311862052, "grad_norm": 1.7478902364895292, "learning_rate": 1.4234808730966231e-05, "loss": 0.5478, "step": 10666 }, { "epoch": 0.3800345583126993, "grad_norm": 1.2567573703786377, "learning_rate": 1.4233763345700014e-05, "loss": 0.5182, "step": 10667 }, { "epoch": 0.3800701854391934, "grad_norm": 1.1104703383908732, "learning_rate": 1.423271790405828e-05, "loss": 0.5139, "step": 10668 }, { "epoch": 0.38010581256568754, "grad_norm": 1.4066490810857843, "learning_rate": 1.4231672406054948e-05, "loss": 0.5777, "step": 10669 }, { "epoch": 0.3801414396921816, "grad_norm": 1.6644671061599616, "learning_rate": 1.423062685170394e-05, "loss": 0.4811, "step": 10670 }, { "epoch": 0.38017706681867575, "grad_norm": 1.606357084692954, "learning_rate": 1.4229581241019176e-05, "loss": 0.5078, "step": 10671 }, { "epoch": 0.3802126939451699, "grad_norm": 1.5263943651692462, "learning_rate": 1.4228535574014586e-05, "loss": 0.5517, "step": 10672 }, { "epoch": 0.38024832107166395, "grad_norm": 1.1516851728366406, "learning_rate": 1.4227489850704088e-05, "loss": 0.4808, "step": 10673 }, { "epoch": 0.3802839481981581, "grad_norm": 1.24484739665458, "learning_rate": 1.4226444071101606e-05, "loss": 0.5104, "step": 10674 }, { "epoch": 0.3803195753246522, "grad_norm": 1.0969372714108914, "learning_rate": 1.4225398235221069e-05, "loss": 0.4797, "step": 10675 }, { "epoch": 0.3803552024511463, "grad_norm": 1.5285900838087711, "learning_rate": 1.42243523430764e-05, "loss": 0.6244, "step": 10676 }, { "epoch": 0.3803908295776404, "grad_norm": 1.4416320709947428, "learning_rate": 1.4223306394681529e-05, "loss": 0.6109, "step": 10677 }, { "epoch": 0.38042645670413455, "grad_norm": 1.042795208041572, "learning_rate": 1.4222260390050378e-05, "loss": 0.5118, "step": 10678 }, { "epoch": 0.38046208383062863, "grad_norm": 1.4735338891633865, "learning_rate": 1.4221214329196882e-05, "loss": 0.5828, "step": 10679 }, { "epoch": 0.38049771095712276, "grad_norm": 1.1337129221923778, "learning_rate": 1.4220168212134966e-05, "loss": 0.535, "step": 10680 }, { "epoch": 0.3805333380836169, "grad_norm": 1.1413667152501628, "learning_rate": 1.4219122038878557e-05, "loss": 0.538, "step": 10681 }, { "epoch": 0.38056896521011097, "grad_norm": 1.1548848075753364, "learning_rate": 1.4218075809441591e-05, "loss": 0.5294, "step": 10682 }, { "epoch": 0.3806045923366051, "grad_norm": 0.9746762294027804, "learning_rate": 1.4217029523837998e-05, "loss": 0.4315, "step": 10683 }, { "epoch": 0.3806402194630992, "grad_norm": 1.174557359299404, "learning_rate": 1.421598318208171e-05, "loss": 0.5083, "step": 10684 }, { "epoch": 0.3806758465895933, "grad_norm": 1.1163302458789606, "learning_rate": 1.4214936784186657e-05, "loss": 0.4632, "step": 10685 }, { "epoch": 0.38071147371608743, "grad_norm": 2.227842400941518, "learning_rate": 1.4213890330166777e-05, "loss": 0.5386, "step": 10686 }, { "epoch": 0.38074710084258157, "grad_norm": 1.1426505879384135, "learning_rate": 1.4212843820036005e-05, "loss": 0.5122, "step": 10687 }, { "epoch": 0.38078272796907564, "grad_norm": 2.129475106321834, "learning_rate": 1.4211797253808268e-05, "loss": 0.5882, "step": 10688 }, { "epoch": 0.38081835509556977, "grad_norm": 1.9622613248348173, "learning_rate": 1.4210750631497505e-05, "loss": 0.609, "step": 10689 }, { "epoch": 0.3808539822220639, "grad_norm": 1.4995258092604542, "learning_rate": 1.420970395311766e-05, "loss": 0.4567, "step": 10690 }, { "epoch": 0.380889609348558, "grad_norm": 1.485402073776033, "learning_rate": 1.4208657218682662e-05, "loss": 0.624, "step": 10691 }, { "epoch": 0.3809252364750521, "grad_norm": 1.652374428372749, "learning_rate": 1.420761042820645e-05, "loss": 0.5995, "step": 10692 }, { "epoch": 0.38096086360154624, "grad_norm": 1.0631412321043907, "learning_rate": 1.4206563581702966e-05, "loss": 0.5106, "step": 10693 }, { "epoch": 0.3809964907280403, "grad_norm": 1.2956127241442839, "learning_rate": 1.4205516679186144e-05, "loss": 0.4753, "step": 10694 }, { "epoch": 0.38103211785453445, "grad_norm": 1.6122475456255787, "learning_rate": 1.4204469720669934e-05, "loss": 0.4881, "step": 10695 }, { "epoch": 0.3810677449810286, "grad_norm": 1.3087282718937754, "learning_rate": 1.4203422706168267e-05, "loss": 0.5785, "step": 10696 }, { "epoch": 0.38110337210752265, "grad_norm": 2.110011111780736, "learning_rate": 1.4202375635695088e-05, "loss": 0.5873, "step": 10697 }, { "epoch": 0.3811389992340168, "grad_norm": 1.1843869005508831, "learning_rate": 1.420132850926434e-05, "loss": 0.48, "step": 10698 }, { "epoch": 0.3811746263605109, "grad_norm": 1.5508184592299303, "learning_rate": 1.420028132688997e-05, "loss": 0.6038, "step": 10699 }, { "epoch": 0.381210253487005, "grad_norm": 1.5267603540061945, "learning_rate": 1.4199234088585915e-05, "loss": 0.6287, "step": 10700 }, { "epoch": 0.3812458806134991, "grad_norm": 1.5238070486600186, "learning_rate": 1.4198186794366125e-05, "loss": 0.4961, "step": 10701 }, { "epoch": 0.38128150773999325, "grad_norm": 1.1232321895006474, "learning_rate": 1.4197139444244544e-05, "loss": 0.4714, "step": 10702 }, { "epoch": 0.38131713486648733, "grad_norm": 2.097542860731027, "learning_rate": 1.4196092038235116e-05, "loss": 0.6029, "step": 10703 }, { "epoch": 0.38135276199298146, "grad_norm": 1.8672351970780234, "learning_rate": 1.419504457635179e-05, "loss": 0.5554, "step": 10704 }, { "epoch": 0.3813883891194756, "grad_norm": 2.79318692032847, "learning_rate": 1.419399705860851e-05, "loss": 0.545, "step": 10705 }, { "epoch": 0.38142401624596967, "grad_norm": 1.701209912575533, "learning_rate": 1.4192949485019231e-05, "loss": 0.601, "step": 10706 }, { "epoch": 0.3814596433724638, "grad_norm": 1.6294522340774076, "learning_rate": 1.4191901855597902e-05, "loss": 0.5576, "step": 10707 }, { "epoch": 0.38149527049895793, "grad_norm": 1.1389525060033114, "learning_rate": 1.4190854170358464e-05, "loss": 0.5462, "step": 10708 }, { "epoch": 0.381530897625452, "grad_norm": 1.4341698744017968, "learning_rate": 1.4189806429314876e-05, "loss": 0.5058, "step": 10709 }, { "epoch": 0.38156652475194613, "grad_norm": 1.38112076205566, "learning_rate": 1.418875863248109e-05, "loss": 0.5183, "step": 10710 }, { "epoch": 0.38160215187844027, "grad_norm": 1.3276926486921454, "learning_rate": 1.418771077987105e-05, "loss": 0.5054, "step": 10711 }, { "epoch": 0.38163777900493434, "grad_norm": 1.8356541145591683, "learning_rate": 1.4186662871498718e-05, "loss": 0.5642, "step": 10712 }, { "epoch": 0.38167340613142847, "grad_norm": 1.1237827169805679, "learning_rate": 1.4185614907378043e-05, "loss": 0.5021, "step": 10713 }, { "epoch": 0.3817090332579226, "grad_norm": 1.5106362487815364, "learning_rate": 1.4184566887522977e-05, "loss": 0.5536, "step": 10714 }, { "epoch": 0.3817446603844167, "grad_norm": 1.1701588754846248, "learning_rate": 1.4183518811947479e-05, "loss": 0.4263, "step": 10715 }, { "epoch": 0.3817802875109108, "grad_norm": 3.3212338344212267, "learning_rate": 1.4182470680665505e-05, "loss": 0.599, "step": 10716 }, { "epoch": 0.38181591463740494, "grad_norm": 1.4616649219895759, "learning_rate": 1.4181422493691008e-05, "loss": 0.6958, "step": 10717 }, { "epoch": 0.381851541763899, "grad_norm": 1.6263734484451895, "learning_rate": 1.4180374251037951e-05, "loss": 0.5399, "step": 10718 }, { "epoch": 0.38188716889039315, "grad_norm": 1.3070365055592148, "learning_rate": 1.4179325952720287e-05, "loss": 0.5053, "step": 10719 }, { "epoch": 0.3819227960168873, "grad_norm": 1.6496243429158268, "learning_rate": 1.417827759875198e-05, "loss": 0.6264, "step": 10720 }, { "epoch": 0.38195842314338135, "grad_norm": 1.1472670576202928, "learning_rate": 1.4177229189146983e-05, "loss": 0.5182, "step": 10721 }, { "epoch": 0.3819940502698755, "grad_norm": 1.6253567912637448, "learning_rate": 1.4176180723919262e-05, "loss": 0.5978, "step": 10722 }, { "epoch": 0.3820296773963696, "grad_norm": 1.1124655535683914, "learning_rate": 1.4175132203082773e-05, "loss": 0.5188, "step": 10723 }, { "epoch": 0.3820653045228637, "grad_norm": 1.1585967591366906, "learning_rate": 1.4174083626651482e-05, "loss": 0.4742, "step": 10724 }, { "epoch": 0.3821009316493578, "grad_norm": 1.0896696287604848, "learning_rate": 1.4173034994639349e-05, "loss": 0.4784, "step": 10725 }, { "epoch": 0.38213655877585195, "grad_norm": 2.8073805847437265, "learning_rate": 1.4171986307060339e-05, "loss": 0.6291, "step": 10726 }, { "epoch": 0.38217218590234603, "grad_norm": 1.0614230669072406, "learning_rate": 1.4170937563928417e-05, "loss": 0.4656, "step": 10727 }, { "epoch": 0.38220781302884016, "grad_norm": 1.4272764474437807, "learning_rate": 1.4169888765257542e-05, "loss": 0.5616, "step": 10728 }, { "epoch": 0.3822434401553343, "grad_norm": 1.5214165780044102, "learning_rate": 1.4168839911061686e-05, "loss": 0.611, "step": 10729 }, { "epoch": 0.38227906728182837, "grad_norm": 1.541500020736556, "learning_rate": 1.4167791001354811e-05, "loss": 0.5728, "step": 10730 }, { "epoch": 0.3823146944083225, "grad_norm": 1.4814152803663785, "learning_rate": 1.416674203615089e-05, "loss": 0.5106, "step": 10731 }, { "epoch": 0.38235032153481663, "grad_norm": 1.472398293100433, "learning_rate": 1.4165693015463884e-05, "loss": 0.6505, "step": 10732 }, { "epoch": 0.3823859486613107, "grad_norm": 1.7587416485448875, "learning_rate": 1.4164643939307765e-05, "loss": 0.5942, "step": 10733 }, { "epoch": 0.38242157578780483, "grad_norm": 1.5892405048721638, "learning_rate": 1.4163594807696496e-05, "loss": 0.5897, "step": 10734 }, { "epoch": 0.38245720291429897, "grad_norm": 1.1215669254054579, "learning_rate": 1.4162545620644058e-05, "loss": 0.475, "step": 10735 }, { "epoch": 0.38249283004079304, "grad_norm": 1.2699919286338521, "learning_rate": 1.4161496378164414e-05, "loss": 0.5887, "step": 10736 }, { "epoch": 0.3825284571672872, "grad_norm": 1.0029644260284056, "learning_rate": 1.4160447080271533e-05, "loss": 0.4399, "step": 10737 }, { "epoch": 0.3825640842937813, "grad_norm": 1.0546408253632469, "learning_rate": 1.4159397726979395e-05, "loss": 0.4771, "step": 10738 }, { "epoch": 0.3825997114202754, "grad_norm": 0.989106950222655, "learning_rate": 1.415834831830197e-05, "loss": 0.5008, "step": 10739 }, { "epoch": 0.3826353385467695, "grad_norm": 0.8940219589736406, "learning_rate": 1.4157298854253228e-05, "loss": 0.4678, "step": 10740 }, { "epoch": 0.38267096567326364, "grad_norm": 1.135603182639365, "learning_rate": 1.4156249334847148e-05, "loss": 0.456, "step": 10741 }, { "epoch": 0.3827065927997577, "grad_norm": 1.5755110999387185, "learning_rate": 1.41551997600977e-05, "loss": 0.6102, "step": 10742 }, { "epoch": 0.38274221992625185, "grad_norm": 1.7197581418395937, "learning_rate": 1.4154150130018867e-05, "loss": 0.6342, "step": 10743 }, { "epoch": 0.382777847052746, "grad_norm": 1.7289600429211718, "learning_rate": 1.415310044462462e-05, "loss": 0.5726, "step": 10744 }, { "epoch": 0.38281347417924005, "grad_norm": 1.092186768673279, "learning_rate": 1.4152050703928934e-05, "loss": 0.5196, "step": 10745 }, { "epoch": 0.3828491013057342, "grad_norm": 2.661191869921736, "learning_rate": 1.4151000907945796e-05, "loss": 0.5779, "step": 10746 }, { "epoch": 0.3828847284322283, "grad_norm": 1.079855592469211, "learning_rate": 1.4149951056689177e-05, "loss": 0.4794, "step": 10747 }, { "epoch": 0.3829203555587224, "grad_norm": 1.1326690944388036, "learning_rate": 1.414890115017306e-05, "loss": 0.5002, "step": 10748 }, { "epoch": 0.3829559826852165, "grad_norm": 1.4736403357270695, "learning_rate": 1.4147851188411423e-05, "loss": 0.6181, "step": 10749 }, { "epoch": 0.38299160981171065, "grad_norm": 1.01746354524426, "learning_rate": 1.414680117141825e-05, "loss": 0.4704, "step": 10750 }, { "epoch": 0.38302723693820473, "grad_norm": 1.0083733305373617, "learning_rate": 1.4145751099207518e-05, "loss": 0.4881, "step": 10751 }, { "epoch": 0.38306286406469886, "grad_norm": 1.0237482672049936, "learning_rate": 1.4144700971793215e-05, "loss": 0.4964, "step": 10752 }, { "epoch": 0.383098491191193, "grad_norm": 0.9688743579019174, "learning_rate": 1.4143650789189321e-05, "loss": 0.4564, "step": 10753 }, { "epoch": 0.38313411831768707, "grad_norm": 1.202023237464304, "learning_rate": 1.414260055140982e-05, "loss": 0.4834, "step": 10754 }, { "epoch": 0.3831697454441812, "grad_norm": 1.300252545874459, "learning_rate": 1.4141550258468697e-05, "loss": 0.6512, "step": 10755 }, { "epoch": 0.38320537257067533, "grad_norm": 1.0575536545825028, "learning_rate": 1.4140499910379936e-05, "loss": 0.5173, "step": 10756 }, { "epoch": 0.3832409996971694, "grad_norm": 0.9991045049652886, "learning_rate": 1.4139449507157528e-05, "loss": 0.473, "step": 10757 }, { "epoch": 0.38327662682366354, "grad_norm": 1.776468709768588, "learning_rate": 1.4138399048815456e-05, "loss": 0.5361, "step": 10758 }, { "epoch": 0.38331225395015767, "grad_norm": 4.338501545257209, "learning_rate": 1.413734853536771e-05, "loss": 0.5751, "step": 10759 }, { "epoch": 0.38334788107665174, "grad_norm": 1.5966283347142844, "learning_rate": 1.4136297966828272e-05, "loss": 0.561, "step": 10760 }, { "epoch": 0.3833835082031459, "grad_norm": 1.572672282048895, "learning_rate": 1.4135247343211138e-05, "loss": 0.624, "step": 10761 }, { "epoch": 0.38341913532964, "grad_norm": 1.2134872474065843, "learning_rate": 1.4134196664530295e-05, "loss": 0.6291, "step": 10762 }, { "epoch": 0.3834547624561341, "grad_norm": 1.6785360746282876, "learning_rate": 1.4133145930799732e-05, "loss": 0.5836, "step": 10763 }, { "epoch": 0.3834903895826282, "grad_norm": 1.0014326631980723, "learning_rate": 1.4132095142033442e-05, "loss": 0.4336, "step": 10764 }, { "epoch": 0.38352601670912234, "grad_norm": 1.9982017914038046, "learning_rate": 1.413104429824542e-05, "loss": 0.5494, "step": 10765 }, { "epoch": 0.3835616438356164, "grad_norm": 1.0349862558504148, "learning_rate": 1.4129993399449653e-05, "loss": 0.4956, "step": 10766 }, { "epoch": 0.38359727096211055, "grad_norm": 1.5364457725901755, "learning_rate": 1.4128942445660139e-05, "loss": 0.5781, "step": 10767 }, { "epoch": 0.3836328980886047, "grad_norm": 1.3844585920400927, "learning_rate": 1.4127891436890867e-05, "loss": 0.6287, "step": 10768 }, { "epoch": 0.38366852521509875, "grad_norm": 1.0807783693231983, "learning_rate": 1.4126840373155841e-05, "loss": 0.468, "step": 10769 }, { "epoch": 0.3837041523415929, "grad_norm": 2.3751206607152713, "learning_rate": 1.412578925446905e-05, "loss": 0.5914, "step": 10770 }, { "epoch": 0.383739779468087, "grad_norm": 1.2271438284148177, "learning_rate": 1.4124738080844486e-05, "loss": 0.5144, "step": 10771 }, { "epoch": 0.3837754065945811, "grad_norm": 0.9785949547464985, "learning_rate": 1.4123686852296154e-05, "loss": 0.4806, "step": 10772 }, { "epoch": 0.3838110337210752, "grad_norm": 1.035726524232271, "learning_rate": 1.4122635568838052e-05, "loss": 0.4725, "step": 10773 }, { "epoch": 0.38384666084756935, "grad_norm": 0.9793000768943271, "learning_rate": 1.4121584230484172e-05, "loss": 0.4482, "step": 10774 }, { "epoch": 0.38388228797406343, "grad_norm": 1.034563661586646, "learning_rate": 1.4120532837248518e-05, "loss": 0.4745, "step": 10775 }, { "epoch": 0.38391791510055756, "grad_norm": 1.7702333687931462, "learning_rate": 1.411948138914509e-05, "loss": 0.6137, "step": 10776 }, { "epoch": 0.3839535422270517, "grad_norm": 0.966247066037663, "learning_rate": 1.411842988618789e-05, "loss": 0.5251, "step": 10777 }, { "epoch": 0.38398916935354577, "grad_norm": 1.0039758287770766, "learning_rate": 1.4117378328390916e-05, "loss": 0.3991, "step": 10778 }, { "epoch": 0.3840247964800399, "grad_norm": 1.0178497950480188, "learning_rate": 1.4116326715768169e-05, "loss": 0.4287, "step": 10779 }, { "epoch": 0.38406042360653403, "grad_norm": 1.3235677201311864, "learning_rate": 1.4115275048333658e-05, "loss": 0.6228, "step": 10780 }, { "epoch": 0.3840960507330281, "grad_norm": 0.9769160470149657, "learning_rate": 1.4114223326101384e-05, "loss": 0.4742, "step": 10781 }, { "epoch": 0.38413167785952224, "grad_norm": 1.016703488262748, "learning_rate": 1.4113171549085347e-05, "loss": 0.4752, "step": 10782 }, { "epoch": 0.38416730498601637, "grad_norm": 1.7844454537172116, "learning_rate": 1.411211971729956e-05, "loss": 0.619, "step": 10783 }, { "epoch": 0.38420293211251044, "grad_norm": 1.0664228241496825, "learning_rate": 1.4111067830758023e-05, "loss": 0.5035, "step": 10784 }, { "epoch": 0.3842385592390046, "grad_norm": 1.1566915999903493, "learning_rate": 1.4110015889474744e-05, "loss": 0.4608, "step": 10785 }, { "epoch": 0.3842741863654987, "grad_norm": 1.7553533308877718, "learning_rate": 1.4108963893463728e-05, "loss": 0.6011, "step": 10786 }, { "epoch": 0.3843098134919928, "grad_norm": 1.528702713137169, "learning_rate": 1.4107911842738988e-05, "loss": 0.5238, "step": 10787 }, { "epoch": 0.3843454406184869, "grad_norm": 3.349536888302005, "learning_rate": 1.4106859737314532e-05, "loss": 0.6348, "step": 10788 }, { "epoch": 0.38438106774498104, "grad_norm": 1.1474717204731235, "learning_rate": 1.4105807577204367e-05, "loss": 0.5235, "step": 10789 }, { "epoch": 0.3844166948714751, "grad_norm": 2.136901314305843, "learning_rate": 1.4104755362422502e-05, "loss": 0.5863, "step": 10790 }, { "epoch": 0.38445232199796925, "grad_norm": 1.5009769647436262, "learning_rate": 1.4103703092982954e-05, "loss": 0.5801, "step": 10791 }, { "epoch": 0.3844879491244634, "grad_norm": 1.5551318330681094, "learning_rate": 1.4102650768899731e-05, "loss": 0.5861, "step": 10792 }, { "epoch": 0.38452357625095746, "grad_norm": 1.1317021817871926, "learning_rate": 1.4101598390186842e-05, "loss": 0.4396, "step": 10793 }, { "epoch": 0.3845592033774516, "grad_norm": 0.974779596112312, "learning_rate": 1.4100545956858307e-05, "loss": 0.4905, "step": 10794 }, { "epoch": 0.3845948305039457, "grad_norm": 1.158471470833325, "learning_rate": 1.4099493468928136e-05, "loss": 0.4598, "step": 10795 }, { "epoch": 0.3846304576304398, "grad_norm": 1.025096844909471, "learning_rate": 1.4098440926410344e-05, "loss": 0.4759, "step": 10796 }, { "epoch": 0.3846660847569339, "grad_norm": 1.6419843173795714, "learning_rate": 1.4097388329318945e-05, "loss": 0.6014, "step": 10797 }, { "epoch": 0.38470171188342805, "grad_norm": 0.9942707300630685, "learning_rate": 1.4096335677667954e-05, "loss": 0.4839, "step": 10798 }, { "epoch": 0.38473733900992213, "grad_norm": 1.077320916479135, "learning_rate": 1.4095282971471396e-05, "loss": 0.545, "step": 10799 }, { "epoch": 0.38477296613641626, "grad_norm": 1.0505518550135375, "learning_rate": 1.409423021074328e-05, "loss": 0.4284, "step": 10800 }, { "epoch": 0.3848085932629104, "grad_norm": 1.7392611923650139, "learning_rate": 1.4093177395497626e-05, "loss": 0.5717, "step": 10801 }, { "epoch": 0.38484422038940447, "grad_norm": 2.4050305291061043, "learning_rate": 1.4092124525748457e-05, "loss": 0.5812, "step": 10802 }, { "epoch": 0.3848798475158986, "grad_norm": 1.398160119178697, "learning_rate": 1.4091071601509789e-05, "loss": 0.5069, "step": 10803 }, { "epoch": 0.38491547464239273, "grad_norm": 1.8814353554456045, "learning_rate": 1.4090018622795643e-05, "loss": 0.5895, "step": 10804 }, { "epoch": 0.3849511017688868, "grad_norm": 1.4735560184975276, "learning_rate": 1.4088965589620039e-05, "loss": 0.5608, "step": 10805 }, { "epoch": 0.38498672889538094, "grad_norm": 1.790668363432952, "learning_rate": 1.4087912501996999e-05, "loss": 0.5777, "step": 10806 }, { "epoch": 0.38502235602187507, "grad_norm": 1.1734701815068511, "learning_rate": 1.408685935994055e-05, "loss": 0.5537, "step": 10807 }, { "epoch": 0.38505798314836914, "grad_norm": 1.052128002628126, "learning_rate": 1.4085806163464709e-05, "loss": 0.4434, "step": 10808 }, { "epoch": 0.3850936102748633, "grad_norm": 1.8703439404130602, "learning_rate": 1.4084752912583504e-05, "loss": 0.5976, "step": 10809 }, { "epoch": 0.3851292374013574, "grad_norm": 1.4086496645389273, "learning_rate": 1.4083699607310959e-05, "loss": 0.5976, "step": 10810 }, { "epoch": 0.3851648645278515, "grad_norm": 1.0341696427869949, "learning_rate": 1.4082646247661101e-05, "loss": 0.4765, "step": 10811 }, { "epoch": 0.3852004916543456, "grad_norm": 1.3234928111925435, "learning_rate": 1.4081592833647952e-05, "loss": 0.545, "step": 10812 }, { "epoch": 0.38523611878083974, "grad_norm": 1.5939695383526782, "learning_rate": 1.4080539365285544e-05, "loss": 0.5489, "step": 10813 }, { "epoch": 0.3852717459073338, "grad_norm": 0.9700507846887629, "learning_rate": 1.4079485842587902e-05, "loss": 0.4223, "step": 10814 }, { "epoch": 0.38530737303382795, "grad_norm": 1.1397560205695665, "learning_rate": 1.4078432265569056e-05, "loss": 0.515, "step": 10815 }, { "epoch": 0.3853430001603221, "grad_norm": 1.099907657717389, "learning_rate": 1.407737863424303e-05, "loss": 0.4734, "step": 10816 }, { "epoch": 0.38537862728681616, "grad_norm": 1.0349620794134033, "learning_rate": 1.4076324948623858e-05, "loss": 0.4985, "step": 10817 }, { "epoch": 0.3854142544133103, "grad_norm": 1.4777852169529848, "learning_rate": 1.4075271208725572e-05, "loss": 0.6072, "step": 10818 }, { "epoch": 0.3854498815398044, "grad_norm": 0.9726483851191521, "learning_rate": 1.4074217414562198e-05, "loss": 0.4546, "step": 10819 }, { "epoch": 0.3854855086662985, "grad_norm": 1.0168514347825734, "learning_rate": 1.4073163566147775e-05, "loss": 0.4462, "step": 10820 }, { "epoch": 0.3855211357927926, "grad_norm": 1.7038169951950308, "learning_rate": 1.4072109663496332e-05, "loss": 0.6009, "step": 10821 }, { "epoch": 0.38555676291928676, "grad_norm": 1.4619285893735772, "learning_rate": 1.4071055706621902e-05, "loss": 0.6612, "step": 10822 }, { "epoch": 0.38559239004578083, "grad_norm": 1.127164203927629, "learning_rate": 1.407000169553852e-05, "loss": 0.5372, "step": 10823 }, { "epoch": 0.38562801717227496, "grad_norm": 1.1171539953586023, "learning_rate": 1.406894763026022e-05, "loss": 0.5517, "step": 10824 }, { "epoch": 0.3856636442987691, "grad_norm": 1.1413021876816976, "learning_rate": 1.4067893510801039e-05, "loss": 0.4584, "step": 10825 }, { "epoch": 0.38569927142526317, "grad_norm": 1.420773077883836, "learning_rate": 1.4066839337175013e-05, "loss": 0.6019, "step": 10826 }, { "epoch": 0.3857348985517573, "grad_norm": 1.0048993432551234, "learning_rate": 1.4065785109396174e-05, "loss": 0.5116, "step": 10827 }, { "epoch": 0.38577052567825143, "grad_norm": 2.2778710967570763, "learning_rate": 1.4064730827478571e-05, "loss": 0.5678, "step": 10828 }, { "epoch": 0.3858061528047455, "grad_norm": 1.015801308400877, "learning_rate": 1.4063676491436234e-05, "loss": 0.4348, "step": 10829 }, { "epoch": 0.38584177993123964, "grad_norm": 1.7908869596749717, "learning_rate": 1.4062622101283206e-05, "loss": 0.5421, "step": 10830 }, { "epoch": 0.38587740705773377, "grad_norm": 0.9932929631974173, "learning_rate": 1.4061567657033519e-05, "loss": 0.422, "step": 10831 }, { "epoch": 0.3859130341842279, "grad_norm": 1.823196502718564, "learning_rate": 1.4060513158701224e-05, "loss": 0.6011, "step": 10832 }, { "epoch": 0.385948661310722, "grad_norm": 1.0885283853999386, "learning_rate": 1.4059458606300358e-05, "loss": 0.4869, "step": 10833 }, { "epoch": 0.3859842884372161, "grad_norm": 1.0758225909593944, "learning_rate": 1.4058403999844963e-05, "loss": 0.4597, "step": 10834 }, { "epoch": 0.38601991556371024, "grad_norm": 1.441900571064915, "learning_rate": 1.405734933934908e-05, "loss": 0.5577, "step": 10835 }, { "epoch": 0.3860555426902043, "grad_norm": 1.7518052858737185, "learning_rate": 1.4056294624826755e-05, "loss": 0.6482, "step": 10836 }, { "epoch": 0.38609116981669844, "grad_norm": 1.360881599739903, "learning_rate": 1.4055239856292036e-05, "loss": 0.5494, "step": 10837 }, { "epoch": 0.3861267969431926, "grad_norm": 1.5653921634107117, "learning_rate": 1.405418503375896e-05, "loss": 0.4984, "step": 10838 }, { "epoch": 0.38616242406968665, "grad_norm": 1.0459498867451282, "learning_rate": 1.4053130157241578e-05, "loss": 0.4644, "step": 10839 }, { "epoch": 0.3861980511961808, "grad_norm": 1.1617811545180186, "learning_rate": 1.4052075226753936e-05, "loss": 0.5153, "step": 10840 }, { "epoch": 0.3862336783226749, "grad_norm": 1.7373696674074472, "learning_rate": 1.4051020242310081e-05, "loss": 0.5612, "step": 10841 }, { "epoch": 0.386269305449169, "grad_norm": 1.6872712212326124, "learning_rate": 1.4049965203924055e-05, "loss": 0.5527, "step": 10842 }, { "epoch": 0.3863049325756631, "grad_norm": 1.5137084048870406, "learning_rate": 1.4048910111609914e-05, "loss": 0.5738, "step": 10843 }, { "epoch": 0.38634055970215725, "grad_norm": 1.597255221230937, "learning_rate": 1.4047854965381706e-05, "loss": 0.6382, "step": 10844 }, { "epoch": 0.3863761868286513, "grad_norm": 1.8300105566268303, "learning_rate": 1.4046799765253481e-05, "loss": 0.6546, "step": 10845 }, { "epoch": 0.38641181395514546, "grad_norm": 1.1245275311431633, "learning_rate": 1.4045744511239284e-05, "loss": 0.5048, "step": 10846 }, { "epoch": 0.3864474410816396, "grad_norm": 0.9891650163275882, "learning_rate": 1.4044689203353174e-05, "loss": 0.4827, "step": 10847 }, { "epoch": 0.38648306820813366, "grad_norm": 1.0585123530347142, "learning_rate": 1.4043633841609201e-05, "loss": 0.4822, "step": 10848 }, { "epoch": 0.3865186953346278, "grad_norm": 1.4191083175884778, "learning_rate": 1.4042578426021418e-05, "loss": 0.5334, "step": 10849 }, { "epoch": 0.3865543224611219, "grad_norm": 0.9954136475390404, "learning_rate": 1.4041522956603872e-05, "loss": 0.5268, "step": 10850 }, { "epoch": 0.386589949587616, "grad_norm": 1.0342969560651647, "learning_rate": 1.4040467433370628e-05, "loss": 0.4457, "step": 10851 }, { "epoch": 0.38662557671411013, "grad_norm": 1.0943964440760523, "learning_rate": 1.4039411856335736e-05, "loss": 0.5275, "step": 10852 }, { "epoch": 0.38666120384060426, "grad_norm": 1.5525770586763683, "learning_rate": 1.403835622551325e-05, "loss": 0.6486, "step": 10853 }, { "epoch": 0.38669683096709834, "grad_norm": 1.1019191490746771, "learning_rate": 1.403730054091723e-05, "loss": 0.4607, "step": 10854 }, { "epoch": 0.38673245809359247, "grad_norm": 1.6143766864716105, "learning_rate": 1.4036244802561727e-05, "loss": 0.6027, "step": 10855 }, { "epoch": 0.3867680852200866, "grad_norm": 2.2781195066452473, "learning_rate": 1.4035189010460808e-05, "loss": 0.4794, "step": 10856 }, { "epoch": 0.3868037123465807, "grad_norm": 1.4745425071956164, "learning_rate": 1.4034133164628523e-05, "loss": 0.5839, "step": 10857 }, { "epoch": 0.3868393394730748, "grad_norm": 1.1539046705460152, "learning_rate": 1.4033077265078939e-05, "loss": 0.4951, "step": 10858 }, { "epoch": 0.38687496659956894, "grad_norm": 1.0887788391540167, "learning_rate": 1.4032021311826112e-05, "loss": 0.4872, "step": 10859 }, { "epoch": 0.386910593726063, "grad_norm": 1.0255644064189222, "learning_rate": 1.4030965304884101e-05, "loss": 0.4549, "step": 10860 }, { "epoch": 0.38694622085255714, "grad_norm": 3.482659672990918, "learning_rate": 1.402990924426697e-05, "loss": 0.5463, "step": 10861 }, { "epoch": 0.3869818479790513, "grad_norm": 1.0162168236247284, "learning_rate": 1.402885312998878e-05, "loss": 0.4754, "step": 10862 }, { "epoch": 0.38701747510554535, "grad_norm": 1.1957310350669605, "learning_rate": 1.4027796962063598e-05, "loss": 0.4888, "step": 10863 }, { "epoch": 0.3870531022320395, "grad_norm": 1.0837048504940916, "learning_rate": 1.402674074050548e-05, "loss": 0.4994, "step": 10864 }, { "epoch": 0.3870887293585336, "grad_norm": 1.1277216189156904, "learning_rate": 1.4025684465328495e-05, "loss": 0.4959, "step": 10865 }, { "epoch": 0.3871243564850277, "grad_norm": 1.105350502704001, "learning_rate": 1.4024628136546711e-05, "loss": 0.456, "step": 10866 }, { "epoch": 0.3871599836115218, "grad_norm": 1.6389631431001574, "learning_rate": 1.4023571754174189e-05, "loss": 0.6359, "step": 10867 }, { "epoch": 0.38719561073801595, "grad_norm": 1.2014890132179017, "learning_rate": 1.4022515318224992e-05, "loss": 0.4714, "step": 10868 }, { "epoch": 0.38723123786451, "grad_norm": 1.080030796862046, "learning_rate": 1.4021458828713196e-05, "loss": 0.5105, "step": 10869 }, { "epoch": 0.38726686499100416, "grad_norm": 1.4364570224712967, "learning_rate": 1.4020402285652865e-05, "loss": 0.6137, "step": 10870 }, { "epoch": 0.3873024921174983, "grad_norm": 1.625549429895492, "learning_rate": 1.4019345689058068e-05, "loss": 0.4667, "step": 10871 }, { "epoch": 0.38733811924399236, "grad_norm": 0.9548311964614721, "learning_rate": 1.4018289038942871e-05, "loss": 0.436, "step": 10872 }, { "epoch": 0.3873737463704865, "grad_norm": 1.8397244171634346, "learning_rate": 1.4017232335321351e-05, "loss": 0.5882, "step": 10873 }, { "epoch": 0.3874093734969806, "grad_norm": 1.5130048927230444, "learning_rate": 1.4016175578207573e-05, "loss": 0.5769, "step": 10874 }, { "epoch": 0.3874450006234747, "grad_norm": 1.169295673881647, "learning_rate": 1.4015118767615607e-05, "loss": 0.5009, "step": 10875 }, { "epoch": 0.38748062774996883, "grad_norm": 1.8164742659890414, "learning_rate": 1.4014061903559532e-05, "loss": 0.7636, "step": 10876 }, { "epoch": 0.38751625487646296, "grad_norm": 1.417802774476472, "learning_rate": 1.4013004986053416e-05, "loss": 0.5403, "step": 10877 }, { "epoch": 0.38755188200295704, "grad_norm": 1.4909715842273996, "learning_rate": 1.4011948015111334e-05, "loss": 0.5487, "step": 10878 }, { "epoch": 0.38758750912945117, "grad_norm": 1.1929717874840324, "learning_rate": 1.401089099074736e-05, "loss": 0.4914, "step": 10879 }, { "epoch": 0.3876231362559453, "grad_norm": 1.0808969115512477, "learning_rate": 1.4009833912975566e-05, "loss": 0.4035, "step": 10880 }, { "epoch": 0.3876587633824394, "grad_norm": 1.4838633841379325, "learning_rate": 1.4008776781810035e-05, "loss": 0.6072, "step": 10881 }, { "epoch": 0.3876943905089335, "grad_norm": 1.1648966228943962, "learning_rate": 1.4007719597264837e-05, "loss": 0.4621, "step": 10882 }, { "epoch": 0.38773001763542764, "grad_norm": 1.2671766625841818, "learning_rate": 1.400666235935405e-05, "loss": 0.5413, "step": 10883 }, { "epoch": 0.3877656447619217, "grad_norm": 1.330035953552838, "learning_rate": 1.4005605068091756e-05, "loss": 0.4952, "step": 10884 }, { "epoch": 0.38780127188841584, "grad_norm": 1.5065807571609937, "learning_rate": 1.400454772349203e-05, "loss": 0.5462, "step": 10885 }, { "epoch": 0.38783689901491, "grad_norm": 1.119955923278799, "learning_rate": 1.4003490325568953e-05, "loss": 0.4888, "step": 10886 }, { "epoch": 0.38787252614140405, "grad_norm": 1.2197843424066712, "learning_rate": 1.40024328743366e-05, "loss": 0.494, "step": 10887 }, { "epoch": 0.3879081532678982, "grad_norm": 1.4978367929017449, "learning_rate": 1.400137536980906e-05, "loss": 0.5849, "step": 10888 }, { "epoch": 0.3879437803943923, "grad_norm": 1.2298368769634336, "learning_rate": 1.400031781200041e-05, "loss": 0.4971, "step": 10889 }, { "epoch": 0.3879794075208864, "grad_norm": 1.1922765963900834, "learning_rate": 1.3999260200924728e-05, "loss": 0.4858, "step": 10890 }, { "epoch": 0.3880150346473805, "grad_norm": 1.4293923161069138, "learning_rate": 1.3998202536596104e-05, "loss": 0.647, "step": 10891 }, { "epoch": 0.38805066177387465, "grad_norm": 1.1614620772436903, "learning_rate": 1.3997144819028619e-05, "loss": 0.474, "step": 10892 }, { "epoch": 0.3880862889003687, "grad_norm": 1.6474226214307868, "learning_rate": 1.3996087048236357e-05, "loss": 0.4995, "step": 10893 }, { "epoch": 0.38812191602686286, "grad_norm": 2.714178341868617, "learning_rate": 1.39950292242334e-05, "loss": 0.5576, "step": 10894 }, { "epoch": 0.388157543153357, "grad_norm": 1.8998831666120504, "learning_rate": 1.3993971347033841e-05, "loss": 0.5904, "step": 10895 }, { "epoch": 0.38819317027985106, "grad_norm": 1.5860897198958113, "learning_rate": 1.3992913416651761e-05, "loss": 0.5812, "step": 10896 }, { "epoch": 0.3882287974063452, "grad_norm": 1.363415323778243, "learning_rate": 1.3991855433101247e-05, "loss": 0.5525, "step": 10897 }, { "epoch": 0.3882644245328393, "grad_norm": 1.1122158461872707, "learning_rate": 1.3990797396396385e-05, "loss": 0.5309, "step": 10898 }, { "epoch": 0.3883000516593334, "grad_norm": 1.2116154275041044, "learning_rate": 1.398973930655127e-05, "loss": 0.5063, "step": 10899 }, { "epoch": 0.38833567878582753, "grad_norm": 1.7844259167578451, "learning_rate": 1.3988681163579987e-05, "loss": 0.629, "step": 10900 }, { "epoch": 0.38837130591232166, "grad_norm": 1.0445300410818203, "learning_rate": 1.3987622967496623e-05, "loss": 0.4832, "step": 10901 }, { "epoch": 0.38840693303881574, "grad_norm": 1.2134785170257822, "learning_rate": 1.3986564718315277e-05, "loss": 0.5839, "step": 10902 }, { "epoch": 0.38844256016530987, "grad_norm": 1.6934763964120045, "learning_rate": 1.3985506416050032e-05, "loss": 0.5639, "step": 10903 }, { "epoch": 0.388478187291804, "grad_norm": 1.7197031158952767, "learning_rate": 1.3984448060714985e-05, "loss": 0.5934, "step": 10904 }, { "epoch": 0.3885138144182981, "grad_norm": 1.3859019304382467, "learning_rate": 1.3983389652324229e-05, "loss": 0.5184, "step": 10905 }, { "epoch": 0.3885494415447922, "grad_norm": 0.9555122027183943, "learning_rate": 1.3982331190891851e-05, "loss": 0.4423, "step": 10906 }, { "epoch": 0.38858506867128634, "grad_norm": 1.0464555195193572, "learning_rate": 1.3981272676431951e-05, "loss": 0.4986, "step": 10907 }, { "epoch": 0.3886206957977804, "grad_norm": 1.1301443102951017, "learning_rate": 1.3980214108958626e-05, "loss": 0.4209, "step": 10908 }, { "epoch": 0.38865632292427454, "grad_norm": 1.0511058423964896, "learning_rate": 1.3979155488485963e-05, "loss": 0.5012, "step": 10909 }, { "epoch": 0.3886919500507687, "grad_norm": 1.8459625773034232, "learning_rate": 1.3978096815028066e-05, "loss": 0.5786, "step": 10910 }, { "epoch": 0.38872757717726275, "grad_norm": 1.1259762734805017, "learning_rate": 1.397703808859903e-05, "loss": 0.487, "step": 10911 }, { "epoch": 0.3887632043037569, "grad_norm": 1.0175843141238805, "learning_rate": 1.3975979309212952e-05, "loss": 0.4872, "step": 10912 }, { "epoch": 0.388798831430251, "grad_norm": 1.9397974408053538, "learning_rate": 1.3974920476883927e-05, "loss": 0.5692, "step": 10913 }, { "epoch": 0.3888344585567451, "grad_norm": 2.124669302552029, "learning_rate": 1.3973861591626061e-05, "loss": 0.5913, "step": 10914 }, { "epoch": 0.3888700856832392, "grad_norm": 1.4495606673811061, "learning_rate": 1.397280265345345e-05, "loss": 0.5894, "step": 10915 }, { "epoch": 0.38890571280973335, "grad_norm": 1.0964183992439096, "learning_rate": 1.3971743662380195e-05, "loss": 0.5112, "step": 10916 }, { "epoch": 0.3889413399362274, "grad_norm": 1.623957775576316, "learning_rate": 1.3970684618420396e-05, "loss": 0.6393, "step": 10917 }, { "epoch": 0.38897696706272156, "grad_norm": 1.024303720925738, "learning_rate": 1.3969625521588157e-05, "loss": 0.4873, "step": 10918 }, { "epoch": 0.3890125941892157, "grad_norm": 1.565953708426769, "learning_rate": 1.396856637189758e-05, "loss": 0.6846, "step": 10919 }, { "epoch": 0.38904822131570976, "grad_norm": 1.1367867467645658, "learning_rate": 1.3967507169362766e-05, "loss": 0.4709, "step": 10920 }, { "epoch": 0.3890838484422039, "grad_norm": 1.4297442599773673, "learning_rate": 1.3966447913997823e-05, "loss": 0.5939, "step": 10921 }, { "epoch": 0.389119475568698, "grad_norm": 1.5486142388279276, "learning_rate": 1.3965388605816852e-05, "loss": 0.5456, "step": 10922 }, { "epoch": 0.3891551026951921, "grad_norm": 1.4338461433231913, "learning_rate": 1.396432924483396e-05, "loss": 0.6173, "step": 10923 }, { "epoch": 0.38919072982168623, "grad_norm": 2.179020548637709, "learning_rate": 1.3963269831063255e-05, "loss": 0.5745, "step": 10924 }, { "epoch": 0.38922635694818036, "grad_norm": 1.2313591430373219, "learning_rate": 1.3962210364518837e-05, "loss": 0.479, "step": 10925 }, { "epoch": 0.38926198407467444, "grad_norm": 1.2284217245713616, "learning_rate": 1.3961150845214824e-05, "loss": 0.4831, "step": 10926 }, { "epoch": 0.38929761120116857, "grad_norm": 1.3640631287480094, "learning_rate": 1.396009127316532e-05, "loss": 0.6543, "step": 10927 }, { "epoch": 0.3893332383276627, "grad_norm": 1.945910426637012, "learning_rate": 1.3959031648384427e-05, "loss": 0.555, "step": 10928 }, { "epoch": 0.3893688654541568, "grad_norm": 1.5536040410750453, "learning_rate": 1.3957971970886264e-05, "loss": 0.5393, "step": 10929 }, { "epoch": 0.3894044925806509, "grad_norm": 1.5159277162184421, "learning_rate": 1.3956912240684938e-05, "loss": 0.6175, "step": 10930 }, { "epoch": 0.38944011970714504, "grad_norm": 1.0769551746543435, "learning_rate": 1.395585245779456e-05, "loss": 0.4687, "step": 10931 }, { "epoch": 0.3894757468336391, "grad_norm": 2.0522952641162173, "learning_rate": 1.3954792622229238e-05, "loss": 0.4846, "step": 10932 }, { "epoch": 0.38951137396013324, "grad_norm": 1.398787458907168, "learning_rate": 1.3953732734003094e-05, "loss": 0.569, "step": 10933 }, { "epoch": 0.3895470010866274, "grad_norm": 1.214670255379293, "learning_rate": 1.3952672793130231e-05, "loss": 0.4826, "step": 10934 }, { "epoch": 0.38958262821312145, "grad_norm": 1.8726930519395073, "learning_rate": 1.3951612799624767e-05, "loss": 0.5291, "step": 10935 }, { "epoch": 0.3896182553396156, "grad_norm": 1.4879933650332047, "learning_rate": 1.3950552753500817e-05, "loss": 0.6087, "step": 10936 }, { "epoch": 0.3896538824661097, "grad_norm": 1.0744573219740048, "learning_rate": 1.3949492654772496e-05, "loss": 0.4855, "step": 10937 }, { "epoch": 0.3896895095926038, "grad_norm": 1.0086684732637181, "learning_rate": 1.394843250345392e-05, "loss": 0.4599, "step": 10938 }, { "epoch": 0.3897251367190979, "grad_norm": 1.0324155944700864, "learning_rate": 1.3947372299559204e-05, "loss": 0.4514, "step": 10939 }, { "epoch": 0.38976076384559205, "grad_norm": 1.0202857067178859, "learning_rate": 1.3946312043102468e-05, "loss": 0.5271, "step": 10940 }, { "epoch": 0.3897963909720861, "grad_norm": 1.0223642647082958, "learning_rate": 1.3945251734097829e-05, "loss": 0.4901, "step": 10941 }, { "epoch": 0.38983201809858026, "grad_norm": 1.472675853732635, "learning_rate": 1.3944191372559408e-05, "loss": 0.5319, "step": 10942 }, { "epoch": 0.3898676452250744, "grad_norm": 1.1470252452189988, "learning_rate": 1.3943130958501318e-05, "loss": 0.4562, "step": 10943 }, { "epoch": 0.38990327235156846, "grad_norm": 1.1320735496625534, "learning_rate": 1.3942070491937685e-05, "loss": 0.5206, "step": 10944 }, { "epoch": 0.3899388994780626, "grad_norm": 1.0224269913313688, "learning_rate": 1.3941009972882628e-05, "loss": 0.4796, "step": 10945 }, { "epoch": 0.3899745266045567, "grad_norm": 1.4093313738046203, "learning_rate": 1.3939949401350268e-05, "loss": 0.5832, "step": 10946 }, { "epoch": 0.3900101537310508, "grad_norm": 1.3532509036431626, "learning_rate": 1.3938888777354728e-05, "loss": 0.6101, "step": 10947 }, { "epoch": 0.39004578085754493, "grad_norm": 1.0945257043279395, "learning_rate": 1.3937828100910129e-05, "loss": 0.484, "step": 10948 }, { "epoch": 0.39008140798403906, "grad_norm": 1.2501340227161424, "learning_rate": 1.3936767372030602e-05, "loss": 0.6361, "step": 10949 }, { "epoch": 0.39011703511053314, "grad_norm": 1.0316802322833456, "learning_rate": 1.393570659073026e-05, "loss": 0.5033, "step": 10950 }, { "epoch": 0.39015266223702727, "grad_norm": 1.7542816592296473, "learning_rate": 1.3934645757023235e-05, "loss": 0.582, "step": 10951 }, { "epoch": 0.3901882893635214, "grad_norm": 1.042085349805989, "learning_rate": 1.3933584870923655e-05, "loss": 0.489, "step": 10952 }, { "epoch": 0.3902239164900155, "grad_norm": 1.4870185698100744, "learning_rate": 1.3932523932445641e-05, "loss": 0.5463, "step": 10953 }, { "epoch": 0.3902595436165096, "grad_norm": 1.0611714422861513, "learning_rate": 1.3931462941603322e-05, "loss": 0.4993, "step": 10954 }, { "epoch": 0.39029517074300374, "grad_norm": 0.9635244111092673, "learning_rate": 1.3930401898410826e-05, "loss": 0.4428, "step": 10955 }, { "epoch": 0.3903307978694978, "grad_norm": 1.5901198179806253, "learning_rate": 1.3929340802882283e-05, "loss": 0.549, "step": 10956 }, { "epoch": 0.39036642499599195, "grad_norm": 1.0974077781961555, "learning_rate": 1.3928279655031818e-05, "loss": 0.4635, "step": 10957 }, { "epoch": 0.3904020521224861, "grad_norm": 1.615179254691359, "learning_rate": 1.3927218454873566e-05, "loss": 0.6038, "step": 10958 }, { "epoch": 0.39043767924898015, "grad_norm": 1.5167988718275025, "learning_rate": 1.3926157202421653e-05, "loss": 0.5559, "step": 10959 }, { "epoch": 0.3904733063754743, "grad_norm": 0.9994121741588484, "learning_rate": 1.3925095897690213e-05, "loss": 0.5304, "step": 10960 }, { "epoch": 0.3905089335019684, "grad_norm": 1.5308171346263435, "learning_rate": 1.392403454069338e-05, "loss": 0.6124, "step": 10961 }, { "epoch": 0.3905445606284625, "grad_norm": 1.0839774918145113, "learning_rate": 1.3922973131445281e-05, "loss": 0.4621, "step": 10962 }, { "epoch": 0.3905801877549566, "grad_norm": 1.0485384468213665, "learning_rate": 1.3921911669960055e-05, "loss": 0.5177, "step": 10963 }, { "epoch": 0.39061581488145075, "grad_norm": 1.8178365804688443, "learning_rate": 1.3920850156251835e-05, "loss": 0.56, "step": 10964 }, { "epoch": 0.3906514420079448, "grad_norm": 1.6742135144408956, "learning_rate": 1.3919788590334751e-05, "loss": 0.6102, "step": 10965 }, { "epoch": 0.39068706913443896, "grad_norm": 1.0284054259906392, "learning_rate": 1.3918726972222947e-05, "loss": 0.4757, "step": 10966 }, { "epoch": 0.3907226962609331, "grad_norm": 1.6211265610966334, "learning_rate": 1.3917665301930552e-05, "loss": 0.5423, "step": 10967 }, { "epoch": 0.39075832338742716, "grad_norm": 1.224909201686524, "learning_rate": 1.3916603579471705e-05, "loss": 0.4927, "step": 10968 }, { "epoch": 0.3907939505139213, "grad_norm": 1.1933034272638592, "learning_rate": 1.3915541804860542e-05, "loss": 0.5488, "step": 10969 }, { "epoch": 0.3908295776404154, "grad_norm": 1.3963200231752686, "learning_rate": 1.3914479978111206e-05, "loss": 0.5391, "step": 10970 }, { "epoch": 0.3908652047669095, "grad_norm": 1.6689392398740988, "learning_rate": 1.391341809923783e-05, "loss": 0.4799, "step": 10971 }, { "epoch": 0.39090083189340363, "grad_norm": 1.6930543511420872, "learning_rate": 1.3912356168254559e-05, "loss": 0.5764, "step": 10972 }, { "epoch": 0.39093645901989776, "grad_norm": 1.6480669001793704, "learning_rate": 1.391129418517553e-05, "loss": 0.5839, "step": 10973 }, { "epoch": 0.39097208614639184, "grad_norm": 1.5581988364901664, "learning_rate": 1.3910232150014887e-05, "loss": 0.5971, "step": 10974 }, { "epoch": 0.39100771327288597, "grad_norm": 1.0122386739145899, "learning_rate": 1.3909170062786768e-05, "loss": 0.4904, "step": 10975 }, { "epoch": 0.3910433403993801, "grad_norm": 1.106908531206637, "learning_rate": 1.3908107923505317e-05, "loss": 0.4991, "step": 10976 }, { "epoch": 0.3910789675258742, "grad_norm": 1.627176115483191, "learning_rate": 1.3907045732184679e-05, "loss": 0.5431, "step": 10977 }, { "epoch": 0.3911145946523683, "grad_norm": 1.077494068272338, "learning_rate": 1.3905983488838996e-05, "loss": 0.5224, "step": 10978 }, { "epoch": 0.39115022177886244, "grad_norm": 1.4486645801522853, "learning_rate": 1.3904921193482412e-05, "loss": 0.5838, "step": 10979 }, { "epoch": 0.3911858489053565, "grad_norm": 1.416766819218722, "learning_rate": 1.3903858846129073e-05, "loss": 0.5081, "step": 10980 }, { "epoch": 0.39122147603185065, "grad_norm": 1.6807263584365773, "learning_rate": 1.3902796446793125e-05, "loss": 0.5492, "step": 10981 }, { "epoch": 0.3912571031583448, "grad_norm": 1.5178106237406732, "learning_rate": 1.3901733995488713e-05, "loss": 0.5622, "step": 10982 }, { "epoch": 0.39129273028483885, "grad_norm": 1.5499458407482403, "learning_rate": 1.3900671492229986e-05, "loss": 0.6258, "step": 10983 }, { "epoch": 0.391328357411333, "grad_norm": 1.1046552039883137, "learning_rate": 1.3899608937031092e-05, "loss": 0.5089, "step": 10984 }, { "epoch": 0.3913639845378271, "grad_norm": 1.1420980697613619, "learning_rate": 1.3898546329906182e-05, "loss": 0.4929, "step": 10985 }, { "epoch": 0.3913996116643212, "grad_norm": 1.0490476824968553, "learning_rate": 1.38974836708694e-05, "loss": 0.4847, "step": 10986 }, { "epoch": 0.3914352387908153, "grad_norm": 1.049771321140003, "learning_rate": 1.38964209599349e-05, "loss": 0.4665, "step": 10987 }, { "epoch": 0.39147086591730945, "grad_norm": 0.9864965890328373, "learning_rate": 1.3895358197116831e-05, "loss": 0.4371, "step": 10988 }, { "epoch": 0.3915064930438035, "grad_norm": 0.9933797091992046, "learning_rate": 1.3894295382429346e-05, "loss": 0.5131, "step": 10989 }, { "epoch": 0.39154212017029766, "grad_norm": 1.3462107397991405, "learning_rate": 1.3893232515886594e-05, "loss": 0.6077, "step": 10990 }, { "epoch": 0.3915777472967918, "grad_norm": 1.5627667322925975, "learning_rate": 1.3892169597502729e-05, "loss": 0.57, "step": 10991 }, { "epoch": 0.39161337442328586, "grad_norm": 1.1315971103332971, "learning_rate": 1.3891106627291907e-05, "loss": 0.4386, "step": 10992 }, { "epoch": 0.39164900154978, "grad_norm": 1.6117136714394755, "learning_rate": 1.3890043605268284e-05, "loss": 0.4865, "step": 10993 }, { "epoch": 0.3916846286762741, "grad_norm": 1.3532506368229347, "learning_rate": 1.3888980531446005e-05, "loss": 0.4339, "step": 10994 }, { "epoch": 0.3917202558027682, "grad_norm": 1.7291840584386173, "learning_rate": 1.3887917405839236e-05, "loss": 0.6214, "step": 10995 }, { "epoch": 0.39175588292926233, "grad_norm": 1.1117997907792214, "learning_rate": 1.3886854228462128e-05, "loss": 0.461, "step": 10996 }, { "epoch": 0.39179151005575646, "grad_norm": 2.28931708315788, "learning_rate": 1.3885790999328839e-05, "loss": 0.547, "step": 10997 }, { "epoch": 0.39182713718225054, "grad_norm": 1.4898771870929044, "learning_rate": 1.3884727718453527e-05, "loss": 0.6426, "step": 10998 }, { "epoch": 0.39186276430874467, "grad_norm": 1.1541407470899057, "learning_rate": 1.388366438585035e-05, "loss": 0.4517, "step": 10999 }, { "epoch": 0.3918983914352388, "grad_norm": 1.7619536538757403, "learning_rate": 1.388260100153347e-05, "loss": 0.5353, "step": 11000 }, { "epoch": 0.3919340185617329, "grad_norm": 0.9946808421414167, "learning_rate": 1.388153756551704e-05, "loss": 0.4048, "step": 11001 }, { "epoch": 0.391969645688227, "grad_norm": 1.5392367784140917, "learning_rate": 1.3880474077815227e-05, "loss": 0.5245, "step": 11002 }, { "epoch": 0.39200527281472114, "grad_norm": 1.5347542201396092, "learning_rate": 1.3879410538442185e-05, "loss": 0.5421, "step": 11003 }, { "epoch": 0.3920408999412152, "grad_norm": 1.4120001860078715, "learning_rate": 1.3878346947412087e-05, "loss": 0.5886, "step": 11004 }, { "epoch": 0.39207652706770935, "grad_norm": 1.6542143450698261, "learning_rate": 1.3877283304739085e-05, "loss": 0.6128, "step": 11005 }, { "epoch": 0.3921121541942035, "grad_norm": 1.5482327615314988, "learning_rate": 1.3876219610437344e-05, "loss": 0.5176, "step": 11006 }, { "epoch": 0.39214778132069755, "grad_norm": 1.597841897101248, "learning_rate": 1.3875155864521031e-05, "loss": 0.6759, "step": 11007 }, { "epoch": 0.3921834084471917, "grad_norm": 1.0940179664080365, "learning_rate": 1.387409206700431e-05, "loss": 0.4911, "step": 11008 }, { "epoch": 0.3922190355736858, "grad_norm": 1.5949663506516247, "learning_rate": 1.3873028217901346e-05, "loss": 0.5182, "step": 11009 }, { "epoch": 0.3922546627001799, "grad_norm": 1.0321266931896504, "learning_rate": 1.38719643172263e-05, "loss": 0.4276, "step": 11010 }, { "epoch": 0.392290289826674, "grad_norm": 1.1481955861797744, "learning_rate": 1.3870900364993347e-05, "loss": 0.4901, "step": 11011 }, { "epoch": 0.39232591695316815, "grad_norm": 1.027273289413468, "learning_rate": 1.3869836361216649e-05, "loss": 0.4703, "step": 11012 }, { "epoch": 0.39236154407966223, "grad_norm": 1.7530087618368972, "learning_rate": 1.3868772305910376e-05, "loss": 0.6216, "step": 11013 }, { "epoch": 0.39239717120615636, "grad_norm": 2.024935052300906, "learning_rate": 1.3867708199088692e-05, "loss": 0.5828, "step": 11014 }, { "epoch": 0.3924327983326505, "grad_norm": 1.1140247444603577, "learning_rate": 1.3866644040765774e-05, "loss": 0.5167, "step": 11015 }, { "epoch": 0.39246842545914457, "grad_norm": 1.0926494893760041, "learning_rate": 1.3865579830955787e-05, "loss": 0.4419, "step": 11016 }, { "epoch": 0.3925040525856387, "grad_norm": 1.1767366231116736, "learning_rate": 1.3864515569672898e-05, "loss": 0.4946, "step": 11017 }, { "epoch": 0.3925396797121328, "grad_norm": 1.055777188822479, "learning_rate": 1.3863451256931286e-05, "loss": 0.4629, "step": 11018 }, { "epoch": 0.3925753068386269, "grad_norm": 1.167115770000899, "learning_rate": 1.3862386892745123e-05, "loss": 0.5291, "step": 11019 }, { "epoch": 0.39261093396512103, "grad_norm": 1.7944038438281693, "learning_rate": 1.3861322477128578e-05, "loss": 0.5537, "step": 11020 }, { "epoch": 0.39264656109161516, "grad_norm": 1.5706663035723032, "learning_rate": 1.3860258010095821e-05, "loss": 0.4916, "step": 11021 }, { "epoch": 0.3926821882181093, "grad_norm": 1.5585520785457427, "learning_rate": 1.3859193491661037e-05, "loss": 0.6173, "step": 11022 }, { "epoch": 0.39271781534460337, "grad_norm": 0.9358312904823791, "learning_rate": 1.385812892183839e-05, "loss": 0.4504, "step": 11023 }, { "epoch": 0.3927534424710975, "grad_norm": 1.074377375831993, "learning_rate": 1.385706430064206e-05, "loss": 0.4306, "step": 11024 }, { "epoch": 0.39278906959759163, "grad_norm": 1.0788119170451635, "learning_rate": 1.3855999628086223e-05, "loss": 0.5065, "step": 11025 }, { "epoch": 0.3928246967240857, "grad_norm": 1.5818887297167572, "learning_rate": 1.3854934904185059e-05, "loss": 0.4811, "step": 11026 }, { "epoch": 0.39286032385057984, "grad_norm": 1.5642871462984531, "learning_rate": 1.385387012895274e-05, "loss": 0.6578, "step": 11027 }, { "epoch": 0.39289595097707397, "grad_norm": 1.511994782465601, "learning_rate": 1.3852805302403445e-05, "loss": 0.496, "step": 11028 }, { "epoch": 0.39293157810356805, "grad_norm": 1.1560443356103485, "learning_rate": 1.3851740424551352e-05, "loss": 0.4821, "step": 11029 }, { "epoch": 0.3929672052300622, "grad_norm": 1.4492599337269425, "learning_rate": 1.3850675495410647e-05, "loss": 0.6323, "step": 11030 }, { "epoch": 0.3930028323565563, "grad_norm": 1.0647163249325675, "learning_rate": 1.3849610514995508e-05, "loss": 0.4686, "step": 11031 }, { "epoch": 0.3930384594830504, "grad_norm": 2.202911980810665, "learning_rate": 1.3848545483320111e-05, "loss": 0.5933, "step": 11032 }, { "epoch": 0.3930740866095445, "grad_norm": 1.0727292764089784, "learning_rate": 1.3847480400398643e-05, "loss": 0.4851, "step": 11033 }, { "epoch": 0.39310971373603865, "grad_norm": 1.0528248634087525, "learning_rate": 1.3846415266245284e-05, "loss": 0.4905, "step": 11034 }, { "epoch": 0.3931453408625327, "grad_norm": 1.091319019377624, "learning_rate": 1.3845350080874217e-05, "loss": 0.4931, "step": 11035 }, { "epoch": 0.39318096798902685, "grad_norm": 1.0345197759677087, "learning_rate": 1.3844284844299625e-05, "loss": 0.4788, "step": 11036 }, { "epoch": 0.393216595115521, "grad_norm": 0.95802771911398, "learning_rate": 1.3843219556535693e-05, "loss": 0.4299, "step": 11037 }, { "epoch": 0.39325222224201506, "grad_norm": 1.1003556714953848, "learning_rate": 1.3842154217596608e-05, "loss": 0.4909, "step": 11038 }, { "epoch": 0.3932878493685092, "grad_norm": 1.0877368270116616, "learning_rate": 1.3841088827496553e-05, "loss": 0.4643, "step": 11039 }, { "epoch": 0.3933234764950033, "grad_norm": 1.088946687346267, "learning_rate": 1.3840023386249715e-05, "loss": 0.4662, "step": 11040 }, { "epoch": 0.3933591036214974, "grad_norm": 1.23888557546435, "learning_rate": 1.3838957893870281e-05, "loss": 0.4791, "step": 11041 }, { "epoch": 0.39339473074799153, "grad_norm": 1.7611239255832758, "learning_rate": 1.3837892350372444e-05, "loss": 0.5997, "step": 11042 }, { "epoch": 0.39343035787448566, "grad_norm": 1.0611848472285683, "learning_rate": 1.3836826755770386e-05, "loss": 0.4167, "step": 11043 }, { "epoch": 0.39346598500097973, "grad_norm": 1.793541777318834, "learning_rate": 1.3835761110078293e-05, "loss": 0.6072, "step": 11044 }, { "epoch": 0.39350161212747387, "grad_norm": 2.4939110989458535, "learning_rate": 1.3834695413310363e-05, "loss": 0.5421, "step": 11045 }, { "epoch": 0.393537239253968, "grad_norm": 1.2145398404217953, "learning_rate": 1.3833629665480785e-05, "loss": 0.5203, "step": 11046 }, { "epoch": 0.39357286638046207, "grad_norm": 1.0806329242975221, "learning_rate": 1.3832563866603748e-05, "loss": 0.4763, "step": 11047 }, { "epoch": 0.3936084935069562, "grad_norm": 1.4411743401313186, "learning_rate": 1.3831498016693443e-05, "loss": 0.5306, "step": 11048 }, { "epoch": 0.39364412063345033, "grad_norm": 2.1315198451711312, "learning_rate": 1.3830432115764066e-05, "loss": 0.5641, "step": 11049 }, { "epoch": 0.3936797477599444, "grad_norm": 1.5378429057915362, "learning_rate": 1.3829366163829808e-05, "loss": 0.6036, "step": 11050 }, { "epoch": 0.39371537488643854, "grad_norm": 1.2149420841519267, "learning_rate": 1.3828300160904856e-05, "loss": 0.4189, "step": 11051 }, { "epoch": 0.39375100201293267, "grad_norm": 1.2874765226070832, "learning_rate": 1.3827234107003417e-05, "loss": 0.5217, "step": 11052 }, { "epoch": 0.39378662913942675, "grad_norm": 1.0808786790604472, "learning_rate": 1.3826168002139682e-05, "loss": 0.4679, "step": 11053 }, { "epoch": 0.3938222562659209, "grad_norm": 1.1351872908710032, "learning_rate": 1.3825101846327847e-05, "loss": 0.4883, "step": 11054 }, { "epoch": 0.393857883392415, "grad_norm": 1.3251728312620954, "learning_rate": 1.3824035639582101e-05, "loss": 0.5247, "step": 11055 }, { "epoch": 0.3938935105189091, "grad_norm": 1.2741479970169483, "learning_rate": 1.382296938191665e-05, "loss": 0.5441, "step": 11056 }, { "epoch": 0.3939291376454032, "grad_norm": 1.0710122041265935, "learning_rate": 1.3821903073345695e-05, "loss": 0.469, "step": 11057 }, { "epoch": 0.39396476477189735, "grad_norm": 1.4699687258575462, "learning_rate": 1.3820836713883424e-05, "loss": 0.6482, "step": 11058 }, { "epoch": 0.3940003918983914, "grad_norm": 1.1390261974729967, "learning_rate": 1.3819770303544042e-05, "loss": 0.4859, "step": 11059 }, { "epoch": 0.39403601902488555, "grad_norm": 1.1277998647248086, "learning_rate": 1.381870384234175e-05, "loss": 0.4893, "step": 11060 }, { "epoch": 0.3940716461513797, "grad_norm": 1.4524759335272355, "learning_rate": 1.3817637330290748e-05, "loss": 0.6379, "step": 11061 }, { "epoch": 0.39410727327787376, "grad_norm": 1.9017848058690026, "learning_rate": 1.3816570767405234e-05, "loss": 0.6024, "step": 11062 }, { "epoch": 0.3941429004043679, "grad_norm": 1.203287247035358, "learning_rate": 1.381550415369941e-05, "loss": 0.4744, "step": 11063 }, { "epoch": 0.394178527530862, "grad_norm": 1.4111682109142731, "learning_rate": 1.3814437489187486e-05, "loss": 0.5012, "step": 11064 }, { "epoch": 0.3942141546573561, "grad_norm": 1.0239252321560817, "learning_rate": 1.3813370773883662e-05, "loss": 0.4645, "step": 11065 }, { "epoch": 0.39424978178385023, "grad_norm": 1.4870981207159215, "learning_rate": 1.3812304007802135e-05, "loss": 0.4675, "step": 11066 }, { "epoch": 0.39428540891034436, "grad_norm": 1.027209251386839, "learning_rate": 1.381123719095712e-05, "loss": 0.4903, "step": 11067 }, { "epoch": 0.39432103603683843, "grad_norm": 1.0619597721098586, "learning_rate": 1.3810170323362817e-05, "loss": 0.4652, "step": 11068 }, { "epoch": 0.39435666316333257, "grad_norm": 1.1127066339887497, "learning_rate": 1.3809103405033435e-05, "loss": 0.5369, "step": 11069 }, { "epoch": 0.3943922902898267, "grad_norm": 0.9895920061555973, "learning_rate": 1.3808036435983175e-05, "loss": 0.4528, "step": 11070 }, { "epoch": 0.3944279174163208, "grad_norm": 1.7215816582128145, "learning_rate": 1.380696941622625e-05, "loss": 0.5439, "step": 11071 }, { "epoch": 0.3944635445428149, "grad_norm": 1.3028677575805139, "learning_rate": 1.3805902345776867e-05, "loss": 0.5017, "step": 11072 }, { "epoch": 0.39449917166930903, "grad_norm": 1.0590466636897422, "learning_rate": 1.380483522464923e-05, "loss": 0.5146, "step": 11073 }, { "epoch": 0.3945347987958031, "grad_norm": 1.218339908281081, "learning_rate": 1.3803768052857559e-05, "loss": 0.483, "step": 11074 }, { "epoch": 0.39457042592229724, "grad_norm": 1.06609415646971, "learning_rate": 1.3802700830416052e-05, "loss": 0.4849, "step": 11075 }, { "epoch": 0.39460605304879137, "grad_norm": 1.6485082201622705, "learning_rate": 1.380163355733893e-05, "loss": 0.6529, "step": 11076 }, { "epoch": 0.39464168017528545, "grad_norm": 1.1868765695916499, "learning_rate": 1.3800566233640396e-05, "loss": 0.482, "step": 11077 }, { "epoch": 0.3946773073017796, "grad_norm": 1.4026040278082514, "learning_rate": 1.3799498859334668e-05, "loss": 0.5204, "step": 11078 }, { "epoch": 0.3947129344282737, "grad_norm": 1.0570734023333381, "learning_rate": 1.379843143443596e-05, "loss": 0.4609, "step": 11079 }, { "epoch": 0.3947485615547678, "grad_norm": 1.1178315542135013, "learning_rate": 1.3797363958958478e-05, "loss": 0.4916, "step": 11080 }, { "epoch": 0.3947841886812619, "grad_norm": 1.4515556791987543, "learning_rate": 1.379629643291644e-05, "loss": 0.5954, "step": 11081 }, { "epoch": 0.39481981580775605, "grad_norm": 1.1092374549143647, "learning_rate": 1.3795228856324065e-05, "loss": 0.4662, "step": 11082 }, { "epoch": 0.3948554429342501, "grad_norm": 1.5955121879199115, "learning_rate": 1.3794161229195566e-05, "loss": 0.5294, "step": 11083 }, { "epoch": 0.39489107006074425, "grad_norm": 1.0766376834173201, "learning_rate": 1.3793093551545154e-05, "loss": 0.492, "step": 11084 }, { "epoch": 0.3949266971872384, "grad_norm": 1.1441049330609396, "learning_rate": 1.3792025823387055e-05, "loss": 0.4463, "step": 11085 }, { "epoch": 0.39496232431373246, "grad_norm": 1.5798415446235645, "learning_rate": 1.3790958044735477e-05, "loss": 0.6403, "step": 11086 }, { "epoch": 0.3949979514402266, "grad_norm": 1.8593267169725904, "learning_rate": 1.3789890215604645e-05, "loss": 0.5998, "step": 11087 }, { "epoch": 0.3950335785667207, "grad_norm": 2.307310111263223, "learning_rate": 1.3788822336008776e-05, "loss": 0.5956, "step": 11088 }, { "epoch": 0.3950692056932148, "grad_norm": 1.3281280965491715, "learning_rate": 1.3787754405962088e-05, "loss": 0.6051, "step": 11089 }, { "epoch": 0.39510483281970893, "grad_norm": 1.108484043268292, "learning_rate": 1.3786686425478806e-05, "loss": 0.4618, "step": 11090 }, { "epoch": 0.39514045994620306, "grad_norm": 1.3578552477852543, "learning_rate": 1.3785618394573144e-05, "loss": 0.6293, "step": 11091 }, { "epoch": 0.39517608707269714, "grad_norm": 1.5121392044502617, "learning_rate": 1.3784550313259328e-05, "loss": 0.618, "step": 11092 }, { "epoch": 0.39521171419919127, "grad_norm": 1.242160895075851, "learning_rate": 1.378348218155158e-05, "loss": 0.5129, "step": 11093 }, { "epoch": 0.3952473413256854, "grad_norm": 1.1043025331471488, "learning_rate": 1.3782413999464122e-05, "loss": 0.4636, "step": 11094 }, { "epoch": 0.3952829684521795, "grad_norm": 1.820229912953833, "learning_rate": 1.3781345767011178e-05, "loss": 0.5516, "step": 11095 }, { "epoch": 0.3953185955786736, "grad_norm": 1.1001230465139311, "learning_rate": 1.3780277484206972e-05, "loss": 0.487, "step": 11096 }, { "epoch": 0.39535422270516773, "grad_norm": 1.5247177250017894, "learning_rate": 1.3779209151065729e-05, "loss": 0.5466, "step": 11097 }, { "epoch": 0.3953898498316618, "grad_norm": 1.846818489483682, "learning_rate": 1.3778140767601673e-05, "loss": 0.5013, "step": 11098 }, { "epoch": 0.39542547695815594, "grad_norm": 1.0483607560500685, "learning_rate": 1.3777072333829033e-05, "loss": 0.4983, "step": 11099 }, { "epoch": 0.3954611040846501, "grad_norm": 1.176527858727339, "learning_rate": 1.3776003849762036e-05, "loss": 0.4621, "step": 11100 }, { "epoch": 0.39549673121114415, "grad_norm": 1.7454059658235113, "learning_rate": 1.3774935315414908e-05, "loss": 0.5667, "step": 11101 }, { "epoch": 0.3955323583376383, "grad_norm": 1.6389394318882569, "learning_rate": 1.3773866730801877e-05, "loss": 0.6361, "step": 11102 }, { "epoch": 0.3955679854641324, "grad_norm": 1.0179610449814067, "learning_rate": 1.3772798095937172e-05, "loss": 0.4771, "step": 11103 }, { "epoch": 0.3956036125906265, "grad_norm": 1.1910292702793748, "learning_rate": 1.3771729410835027e-05, "loss": 0.4857, "step": 11104 }, { "epoch": 0.3956392397171206, "grad_norm": 1.1368890489806833, "learning_rate": 1.3770660675509666e-05, "loss": 0.4915, "step": 11105 }, { "epoch": 0.39567486684361475, "grad_norm": 1.5377039897166058, "learning_rate": 1.3769591889975322e-05, "loss": 0.5913, "step": 11106 }, { "epoch": 0.3957104939701088, "grad_norm": 1.1223447987522053, "learning_rate": 1.3768523054246228e-05, "loss": 0.4213, "step": 11107 }, { "epoch": 0.39574612109660295, "grad_norm": 1.6285445663034024, "learning_rate": 1.3767454168336616e-05, "loss": 0.6074, "step": 11108 }, { "epoch": 0.3957817482230971, "grad_norm": 1.1665323323328194, "learning_rate": 1.3766385232260717e-05, "loss": 0.4898, "step": 11109 }, { "epoch": 0.39581737534959116, "grad_norm": 1.0596179193930833, "learning_rate": 1.3765316246032767e-05, "loss": 0.4476, "step": 11110 }, { "epoch": 0.3958530024760853, "grad_norm": 1.6270116158992587, "learning_rate": 1.3764247209666999e-05, "loss": 0.5726, "step": 11111 }, { "epoch": 0.3958886296025794, "grad_norm": 1.5146991369927592, "learning_rate": 1.376317812317765e-05, "loss": 0.5395, "step": 11112 }, { "epoch": 0.3959242567290735, "grad_norm": 1.0710072811489872, "learning_rate": 1.3762108986578952e-05, "loss": 0.4527, "step": 11113 }, { "epoch": 0.39595988385556763, "grad_norm": 1.0907205013989456, "learning_rate": 1.3761039799885143e-05, "loss": 0.5214, "step": 11114 }, { "epoch": 0.39599551098206176, "grad_norm": 1.140189363414081, "learning_rate": 1.3759970563110461e-05, "loss": 0.5368, "step": 11115 }, { "epoch": 0.39603113810855584, "grad_norm": 1.5290487825296288, "learning_rate": 1.3758901276269146e-05, "loss": 0.5163, "step": 11116 }, { "epoch": 0.39606676523504997, "grad_norm": 1.7619039536645311, "learning_rate": 1.375783193937543e-05, "loss": 0.6158, "step": 11117 }, { "epoch": 0.3961023923615441, "grad_norm": 1.040455681067296, "learning_rate": 1.3756762552443555e-05, "loss": 0.4642, "step": 11118 }, { "epoch": 0.3961380194880382, "grad_norm": 1.615243927940571, "learning_rate": 1.375569311548776e-05, "loss": 0.4729, "step": 11119 }, { "epoch": 0.3961736466145323, "grad_norm": 1.0780849475819527, "learning_rate": 1.375462362852229e-05, "loss": 0.4831, "step": 11120 }, { "epoch": 0.39620927374102644, "grad_norm": 1.0848115316187468, "learning_rate": 1.3753554091561379e-05, "loss": 0.4642, "step": 11121 }, { "epoch": 0.3962449008675205, "grad_norm": 1.7123489864455255, "learning_rate": 1.375248450461927e-05, "loss": 0.6153, "step": 11122 }, { "epoch": 0.39628052799401464, "grad_norm": 1.2159592835957085, "learning_rate": 1.375141486771021e-05, "loss": 0.4551, "step": 11123 }, { "epoch": 0.3963161551205088, "grad_norm": 1.6509736176285883, "learning_rate": 1.3750345180848438e-05, "loss": 0.6051, "step": 11124 }, { "epoch": 0.39635178224700285, "grad_norm": 0.9750490628318832, "learning_rate": 1.37492754440482e-05, "loss": 0.4822, "step": 11125 }, { "epoch": 0.396387409373497, "grad_norm": 1.4889504155427304, "learning_rate": 1.3748205657323737e-05, "loss": 0.6417, "step": 11126 }, { "epoch": 0.3964230364999911, "grad_norm": 1.0851713030269539, "learning_rate": 1.3747135820689294e-05, "loss": 0.4532, "step": 11127 }, { "epoch": 0.3964586636264852, "grad_norm": 1.4812687206464052, "learning_rate": 1.3746065934159123e-05, "loss": 0.6302, "step": 11128 }, { "epoch": 0.3964942907529793, "grad_norm": 1.5701179064273163, "learning_rate": 1.3744995997747462e-05, "loss": 0.5636, "step": 11129 }, { "epoch": 0.39652991787947345, "grad_norm": 1.0582942274544767, "learning_rate": 1.3743926011468565e-05, "loss": 0.4853, "step": 11130 }, { "epoch": 0.3965655450059675, "grad_norm": 1.4046942042574044, "learning_rate": 1.3742855975336676e-05, "loss": 0.4378, "step": 11131 }, { "epoch": 0.39660117213246165, "grad_norm": 1.1607135847918015, "learning_rate": 1.374178588936604e-05, "loss": 0.5021, "step": 11132 }, { "epoch": 0.3966367992589558, "grad_norm": 1.266781694624202, "learning_rate": 1.374071575357091e-05, "loss": 0.5926, "step": 11133 }, { "epoch": 0.39667242638544986, "grad_norm": 1.0264048910834387, "learning_rate": 1.3739645567965538e-05, "loss": 0.409, "step": 11134 }, { "epoch": 0.396708053511944, "grad_norm": 1.0864150350919801, "learning_rate": 1.373857533256417e-05, "loss": 0.4632, "step": 11135 }, { "epoch": 0.3967436806384381, "grad_norm": 1.5062813830799318, "learning_rate": 1.3737505047381058e-05, "loss": 0.4838, "step": 11136 }, { "epoch": 0.3967793077649322, "grad_norm": 1.019125024706328, "learning_rate": 1.3736434712430455e-05, "loss": 0.4512, "step": 11137 }, { "epoch": 0.39681493489142633, "grad_norm": 1.3666322370049564, "learning_rate": 1.3735364327726613e-05, "loss": 0.5468, "step": 11138 }, { "epoch": 0.39685056201792046, "grad_norm": 1.3226829767233053, "learning_rate": 1.3734293893283783e-05, "loss": 0.5829, "step": 11139 }, { "epoch": 0.39688618914441454, "grad_norm": 1.5932559224603962, "learning_rate": 1.3733223409116216e-05, "loss": 0.5664, "step": 11140 }, { "epoch": 0.39692181627090867, "grad_norm": 1.9644791508824988, "learning_rate": 1.3732152875238176e-05, "loss": 0.5752, "step": 11141 }, { "epoch": 0.3969574433974028, "grad_norm": 1.3275706873883781, "learning_rate": 1.373108229166391e-05, "loss": 0.5538, "step": 11142 }, { "epoch": 0.3969930705238969, "grad_norm": 1.1067678243504377, "learning_rate": 1.3730011658407676e-05, "loss": 0.4748, "step": 11143 }, { "epoch": 0.397028697650391, "grad_norm": 1.1497809115510902, "learning_rate": 1.3728940975483726e-05, "loss": 0.4886, "step": 11144 }, { "epoch": 0.39706432477688514, "grad_norm": 1.8512464032044782, "learning_rate": 1.372787024290632e-05, "loss": 0.5234, "step": 11145 }, { "epoch": 0.3970999519033792, "grad_norm": 1.036229138556769, "learning_rate": 1.3726799460689722e-05, "loss": 0.47, "step": 11146 }, { "epoch": 0.39713557902987334, "grad_norm": 1.2713988007008048, "learning_rate": 1.3725728628848181e-05, "loss": 0.498, "step": 11147 }, { "epoch": 0.3971712061563675, "grad_norm": 1.1838635887994857, "learning_rate": 1.3724657747395957e-05, "loss": 0.4901, "step": 11148 }, { "epoch": 0.39720683328286155, "grad_norm": 2.5125829520643888, "learning_rate": 1.3723586816347316e-05, "loss": 0.5839, "step": 11149 }, { "epoch": 0.3972424604093557, "grad_norm": 0.9509033400661627, "learning_rate": 1.3722515835716515e-05, "loss": 0.4264, "step": 11150 }, { "epoch": 0.3972780875358498, "grad_norm": 1.2570967109807731, "learning_rate": 1.372144480551781e-05, "loss": 0.5295, "step": 11151 }, { "epoch": 0.3973137146623439, "grad_norm": 1.4993521152464693, "learning_rate": 1.3720373725765466e-05, "loss": 0.5912, "step": 11152 }, { "epoch": 0.397349341788838, "grad_norm": 1.6173168127695148, "learning_rate": 1.3719302596473747e-05, "loss": 0.5578, "step": 11153 }, { "epoch": 0.39738496891533215, "grad_norm": 2.711183853323105, "learning_rate": 1.3718231417656913e-05, "loss": 0.6474, "step": 11154 }, { "epoch": 0.3974205960418262, "grad_norm": 1.2853999772502551, "learning_rate": 1.371716018932923e-05, "loss": 0.484, "step": 11155 }, { "epoch": 0.39745622316832036, "grad_norm": 1.6411366036330042, "learning_rate": 1.371608891150496e-05, "loss": 0.5989, "step": 11156 }, { "epoch": 0.3974918502948145, "grad_norm": 1.2639007082711804, "learning_rate": 1.371501758419837e-05, "loss": 0.5065, "step": 11157 }, { "epoch": 0.39752747742130856, "grad_norm": 1.934663010961873, "learning_rate": 1.3713946207423724e-05, "loss": 0.5777, "step": 11158 }, { "epoch": 0.3975631045478027, "grad_norm": 2.4949148413269318, "learning_rate": 1.3712874781195286e-05, "loss": 0.6438, "step": 11159 }, { "epoch": 0.3975987316742968, "grad_norm": 1.636368887232024, "learning_rate": 1.3711803305527326e-05, "loss": 0.5396, "step": 11160 }, { "epoch": 0.3976343588007909, "grad_norm": 1.505885805122918, "learning_rate": 1.3710731780434115e-05, "loss": 0.5698, "step": 11161 }, { "epoch": 0.39766998592728503, "grad_norm": 1.1557919681833808, "learning_rate": 1.3709660205929914e-05, "loss": 0.5333, "step": 11162 }, { "epoch": 0.39770561305377916, "grad_norm": 1.1850807767002414, "learning_rate": 1.3708588582028993e-05, "loss": 0.5153, "step": 11163 }, { "epoch": 0.39774124018027324, "grad_norm": 1.0397300542885841, "learning_rate": 1.3707516908745624e-05, "loss": 0.4731, "step": 11164 }, { "epoch": 0.39777686730676737, "grad_norm": 1.1365295260623833, "learning_rate": 1.3706445186094075e-05, "loss": 0.4614, "step": 11165 }, { "epoch": 0.3978124944332615, "grad_norm": 1.0153308692685936, "learning_rate": 1.3705373414088618e-05, "loss": 0.502, "step": 11166 }, { "epoch": 0.3978481215597556, "grad_norm": 1.128891785443056, "learning_rate": 1.3704301592743522e-05, "loss": 0.4887, "step": 11167 }, { "epoch": 0.3978837486862497, "grad_norm": 1.0803011385477517, "learning_rate": 1.3703229722073065e-05, "loss": 0.4918, "step": 11168 }, { "epoch": 0.39791937581274384, "grad_norm": 1.1981681328638027, "learning_rate": 1.3702157802091515e-05, "loss": 0.4957, "step": 11169 }, { "epoch": 0.3979550029392379, "grad_norm": 1.0168935314165308, "learning_rate": 1.3701085832813145e-05, "loss": 0.4741, "step": 11170 }, { "epoch": 0.39799063006573204, "grad_norm": 1.1672132345355015, "learning_rate": 1.3700013814252232e-05, "loss": 0.5067, "step": 11171 }, { "epoch": 0.3980262571922262, "grad_norm": 1.165280133140789, "learning_rate": 1.3698941746423047e-05, "loss": 0.5856, "step": 11172 }, { "epoch": 0.39806188431872025, "grad_norm": 1.8760558091720771, "learning_rate": 1.3697869629339869e-05, "loss": 0.5529, "step": 11173 }, { "epoch": 0.3980975114452144, "grad_norm": 1.1046324122872226, "learning_rate": 1.3696797463016972e-05, "loss": 0.4593, "step": 11174 }, { "epoch": 0.3981331385717085, "grad_norm": 1.0982899154571968, "learning_rate": 1.3695725247468633e-05, "loss": 0.4369, "step": 11175 }, { "epoch": 0.3981687656982026, "grad_norm": 1.061247142082898, "learning_rate": 1.3694652982709126e-05, "loss": 0.4552, "step": 11176 }, { "epoch": 0.3982043928246967, "grad_norm": 1.2479390692887042, "learning_rate": 1.3693580668752735e-05, "loss": 0.4838, "step": 11177 }, { "epoch": 0.39824001995119085, "grad_norm": 2.210368909337688, "learning_rate": 1.3692508305613733e-05, "loss": 0.5908, "step": 11178 }, { "epoch": 0.3982756470776849, "grad_norm": 2.136144945396737, "learning_rate": 1.3691435893306406e-05, "loss": 0.5886, "step": 11179 }, { "epoch": 0.39831127420417906, "grad_norm": 3.071440516070461, "learning_rate": 1.3690363431845029e-05, "loss": 0.6149, "step": 11180 }, { "epoch": 0.3983469013306732, "grad_norm": 2.230255766012076, "learning_rate": 1.3689290921243883e-05, "loss": 0.599, "step": 11181 }, { "epoch": 0.39838252845716726, "grad_norm": 1.1071927966254937, "learning_rate": 1.3688218361517246e-05, "loss": 0.4562, "step": 11182 }, { "epoch": 0.3984181555836614, "grad_norm": 1.2001541248053778, "learning_rate": 1.3687145752679409e-05, "loss": 0.4941, "step": 11183 }, { "epoch": 0.3984537827101555, "grad_norm": 2.0441359159221637, "learning_rate": 1.3686073094744647e-05, "loss": 0.6408, "step": 11184 }, { "epoch": 0.3984894098366496, "grad_norm": 1.1239842405299594, "learning_rate": 1.3685000387727245e-05, "loss": 0.501, "step": 11185 }, { "epoch": 0.39852503696314373, "grad_norm": 0.985538686751551, "learning_rate": 1.3683927631641486e-05, "loss": 0.4409, "step": 11186 }, { "epoch": 0.39856066408963786, "grad_norm": 1.6646298832677398, "learning_rate": 1.368285482650166e-05, "loss": 0.6523, "step": 11187 }, { "epoch": 0.39859629121613194, "grad_norm": 1.3622979781788418, "learning_rate": 1.3681781972322046e-05, "loss": 0.6076, "step": 11188 }, { "epoch": 0.39863191834262607, "grad_norm": 1.1001097252650245, "learning_rate": 1.3680709069116929e-05, "loss": 0.4742, "step": 11189 }, { "epoch": 0.3986675454691202, "grad_norm": 1.440054621065943, "learning_rate": 1.3679636116900598e-05, "loss": 0.5715, "step": 11190 }, { "epoch": 0.3987031725956143, "grad_norm": 1.0136344219605495, "learning_rate": 1.3678563115687342e-05, "loss": 0.5033, "step": 11191 }, { "epoch": 0.3987387997221084, "grad_norm": 1.050939493183513, "learning_rate": 1.3677490065491447e-05, "loss": 0.4736, "step": 11192 }, { "epoch": 0.39877442684860254, "grad_norm": 1.080710235429021, "learning_rate": 1.3676416966327201e-05, "loss": 0.4937, "step": 11193 }, { "epoch": 0.3988100539750966, "grad_norm": 1.1229485283739191, "learning_rate": 1.3675343818208894e-05, "loss": 0.4567, "step": 11194 }, { "epoch": 0.39884568110159074, "grad_norm": 1.0902959031062764, "learning_rate": 1.3674270621150817e-05, "loss": 0.4421, "step": 11195 }, { "epoch": 0.3988813082280849, "grad_norm": 1.4633957821945458, "learning_rate": 1.3673197375167255e-05, "loss": 0.6144, "step": 11196 }, { "epoch": 0.39891693535457895, "grad_norm": 3.0383837631349015, "learning_rate": 1.3672124080272504e-05, "loss": 0.6081, "step": 11197 }, { "epoch": 0.3989525624810731, "grad_norm": 3.720561958429047, "learning_rate": 1.3671050736480855e-05, "loss": 0.6144, "step": 11198 }, { "epoch": 0.3989881896075672, "grad_norm": 1.077587845864493, "learning_rate": 1.3669977343806599e-05, "loss": 0.4679, "step": 11199 }, { "epoch": 0.3990238167340613, "grad_norm": 1.2566477565410532, "learning_rate": 1.3668903902264027e-05, "loss": 0.486, "step": 11200 }, { "epoch": 0.3990594438605554, "grad_norm": 1.1042795938101975, "learning_rate": 1.3667830411867438e-05, "loss": 0.4502, "step": 11201 }, { "epoch": 0.39909507098704955, "grad_norm": 1.7149866566633103, "learning_rate": 1.3666756872631122e-05, "loss": 0.5875, "step": 11202 }, { "epoch": 0.3991306981135436, "grad_norm": 1.9633642329991916, "learning_rate": 1.3665683284569377e-05, "loss": 0.6367, "step": 11203 }, { "epoch": 0.39916632524003776, "grad_norm": 1.120484780821386, "learning_rate": 1.3664609647696497e-05, "loss": 0.467, "step": 11204 }, { "epoch": 0.3992019523665319, "grad_norm": 1.0936793508631395, "learning_rate": 1.3663535962026777e-05, "loss": 0.4662, "step": 11205 }, { "epoch": 0.39923757949302596, "grad_norm": 1.0967411823749196, "learning_rate": 1.3662462227574515e-05, "loss": 0.4685, "step": 11206 }, { "epoch": 0.3992732066195201, "grad_norm": 1.624022526911653, "learning_rate": 1.3661388444354009e-05, "loss": 0.5345, "step": 11207 }, { "epoch": 0.3993088337460142, "grad_norm": 0.9426128026554419, "learning_rate": 1.3660314612379554e-05, "loss": 0.4442, "step": 11208 }, { "epoch": 0.39934446087250836, "grad_norm": 1.0348829089125025, "learning_rate": 1.3659240731665455e-05, "loss": 0.4932, "step": 11209 }, { "epoch": 0.39938008799900243, "grad_norm": 1.805028751483189, "learning_rate": 1.3658166802226007e-05, "loss": 0.5545, "step": 11210 }, { "epoch": 0.39941571512549656, "grad_norm": 2.017424782487192, "learning_rate": 1.365709282407551e-05, "loss": 0.622, "step": 11211 }, { "epoch": 0.3994513422519907, "grad_norm": 1.1086209317005482, "learning_rate": 1.3656018797228269e-05, "loss": 0.4936, "step": 11212 }, { "epoch": 0.39948696937848477, "grad_norm": 1.384018706460199, "learning_rate": 1.365494472169858e-05, "loss": 0.618, "step": 11213 }, { "epoch": 0.3995225965049789, "grad_norm": 1.5300706295256892, "learning_rate": 1.3653870597500747e-05, "loss": 0.5695, "step": 11214 }, { "epoch": 0.39955822363147303, "grad_norm": 1.7114523268300461, "learning_rate": 1.3652796424649073e-05, "loss": 0.5269, "step": 11215 }, { "epoch": 0.3995938507579671, "grad_norm": 1.6570304824356668, "learning_rate": 1.3651722203157863e-05, "loss": 0.5933, "step": 11216 }, { "epoch": 0.39962947788446124, "grad_norm": 1.5238777047589325, "learning_rate": 1.365064793304142e-05, "loss": 0.6077, "step": 11217 }, { "epoch": 0.39966510501095537, "grad_norm": 1.5986612331832168, "learning_rate": 1.3649573614314045e-05, "loss": 0.5587, "step": 11218 }, { "epoch": 0.39970073213744944, "grad_norm": 1.1993246727193763, "learning_rate": 1.3648499246990048e-05, "loss": 0.4693, "step": 11219 }, { "epoch": 0.3997363592639436, "grad_norm": 1.0261405959206285, "learning_rate": 1.3647424831083733e-05, "loss": 0.4674, "step": 11220 }, { "epoch": 0.3997719863904377, "grad_norm": 1.0954069489531113, "learning_rate": 1.3646350366609409e-05, "loss": 0.4857, "step": 11221 }, { "epoch": 0.3998076135169318, "grad_norm": 1.1827689081002857, "learning_rate": 1.3645275853581379e-05, "loss": 0.493, "step": 11222 }, { "epoch": 0.3998432406434259, "grad_norm": 1.0767955482494882, "learning_rate": 1.3644201292013955e-05, "loss": 0.4627, "step": 11223 }, { "epoch": 0.39987886776992004, "grad_norm": 1.73250738750109, "learning_rate": 1.3643126681921441e-05, "loss": 0.5247, "step": 11224 }, { "epoch": 0.3999144948964141, "grad_norm": 1.4856997717250826, "learning_rate": 1.3642052023318151e-05, "loss": 0.616, "step": 11225 }, { "epoch": 0.39995012202290825, "grad_norm": 1.4588866431986642, "learning_rate": 1.3640977316218391e-05, "loss": 0.5625, "step": 11226 }, { "epoch": 0.3999857491494024, "grad_norm": 2.022329191929598, "learning_rate": 1.3639902560636473e-05, "loss": 0.6113, "step": 11227 }, { "epoch": 0.40002137627589646, "grad_norm": 1.6534669345914965, "learning_rate": 1.363882775658671e-05, "loss": 0.6155, "step": 11228 }, { "epoch": 0.4000570034023906, "grad_norm": 1.1407748741003167, "learning_rate": 1.3637752904083412e-05, "loss": 0.5062, "step": 11229 }, { "epoch": 0.4000926305288847, "grad_norm": 1.5076658992769778, "learning_rate": 1.3636678003140889e-05, "loss": 0.5793, "step": 11230 }, { "epoch": 0.4001282576553788, "grad_norm": 1.042903302117646, "learning_rate": 1.363560305377346e-05, "loss": 0.4352, "step": 11231 }, { "epoch": 0.4001638847818729, "grad_norm": 1.089622337857362, "learning_rate": 1.3634528055995433e-05, "loss": 0.471, "step": 11232 }, { "epoch": 0.40019951190836706, "grad_norm": 1.1015969869980642, "learning_rate": 1.3633453009821127e-05, "loss": 0.4955, "step": 11233 }, { "epoch": 0.40023513903486113, "grad_norm": 1.25656509274557, "learning_rate": 1.3632377915264852e-05, "loss": 0.4338, "step": 11234 }, { "epoch": 0.40027076616135526, "grad_norm": 1.5390896283330897, "learning_rate": 1.3631302772340926e-05, "loss": 0.5851, "step": 11235 }, { "epoch": 0.4003063932878494, "grad_norm": 1.0275297033079442, "learning_rate": 1.3630227581063666e-05, "loss": 0.4714, "step": 11236 }, { "epoch": 0.40034202041434347, "grad_norm": 1.412615828141271, "learning_rate": 1.3629152341447389e-05, "loss": 0.5461, "step": 11237 }, { "epoch": 0.4003776475408376, "grad_norm": 1.736835766558886, "learning_rate": 1.362807705350641e-05, "loss": 0.5706, "step": 11238 }, { "epoch": 0.40041327466733173, "grad_norm": 1.01602876623092, "learning_rate": 1.3627001717255053e-05, "loss": 0.4765, "step": 11239 }, { "epoch": 0.4004489017938258, "grad_norm": 1.0066198910812854, "learning_rate": 1.362592633270763e-05, "loss": 0.4545, "step": 11240 }, { "epoch": 0.40048452892031994, "grad_norm": 1.2157160219951677, "learning_rate": 1.3624850899878463e-05, "loss": 0.5228, "step": 11241 }, { "epoch": 0.40052015604681407, "grad_norm": 1.5938857844733754, "learning_rate": 1.3623775418781874e-05, "loss": 0.6293, "step": 11242 }, { "epoch": 0.40055578317330814, "grad_norm": 1.5107364950699183, "learning_rate": 1.3622699889432184e-05, "loss": 0.5771, "step": 11243 }, { "epoch": 0.4005914102998023, "grad_norm": 1.6989205805490752, "learning_rate": 1.362162431184371e-05, "loss": 0.5934, "step": 11244 }, { "epoch": 0.4006270374262964, "grad_norm": 1.0779084822666019, "learning_rate": 1.3620548686030779e-05, "loss": 0.4485, "step": 11245 }, { "epoch": 0.4006626645527905, "grad_norm": 1.9577867014303858, "learning_rate": 1.3619473012007712e-05, "loss": 0.6117, "step": 11246 }, { "epoch": 0.4006982916792846, "grad_norm": 1.3051651144082197, "learning_rate": 1.3618397289788831e-05, "loss": 0.4678, "step": 11247 }, { "epoch": 0.40073391880577874, "grad_norm": 1.3575045631010094, "learning_rate": 1.361732151938846e-05, "loss": 0.538, "step": 11248 }, { "epoch": 0.4007695459322728, "grad_norm": 1.1256318406694854, "learning_rate": 1.3616245700820922e-05, "loss": 0.4852, "step": 11249 }, { "epoch": 0.40080517305876695, "grad_norm": 1.185790895243896, "learning_rate": 1.361516983410055e-05, "loss": 0.4965, "step": 11250 }, { "epoch": 0.4008408001852611, "grad_norm": 1.686802979929441, "learning_rate": 1.3614093919241664e-05, "loss": 0.576, "step": 11251 }, { "epoch": 0.40087642731175516, "grad_norm": 1.5789926005752974, "learning_rate": 1.3613017956258591e-05, "loss": 0.6032, "step": 11252 }, { "epoch": 0.4009120544382493, "grad_norm": 1.6317826664908777, "learning_rate": 1.3611941945165656e-05, "loss": 0.5481, "step": 11253 }, { "epoch": 0.4009476815647434, "grad_norm": 1.5619442044122538, "learning_rate": 1.3610865885977194e-05, "loss": 0.5689, "step": 11254 }, { "epoch": 0.4009833086912375, "grad_norm": 1.2995537087906137, "learning_rate": 1.3609789778707527e-05, "loss": 0.605, "step": 11255 }, { "epoch": 0.4010189358177316, "grad_norm": 1.0377694346475532, "learning_rate": 1.3608713623370983e-05, "loss": 0.4258, "step": 11256 }, { "epoch": 0.40105456294422576, "grad_norm": 1.073967971461542, "learning_rate": 1.3607637419981898e-05, "loss": 0.4734, "step": 11257 }, { "epoch": 0.40109019007071983, "grad_norm": 1.0796624161264514, "learning_rate": 1.3606561168554597e-05, "loss": 0.4673, "step": 11258 }, { "epoch": 0.40112581719721396, "grad_norm": 1.013589458604862, "learning_rate": 1.3605484869103416e-05, "loss": 0.4883, "step": 11259 }, { "epoch": 0.4011614443237081, "grad_norm": 1.090019241588249, "learning_rate": 1.360440852164268e-05, "loss": 0.4179, "step": 11260 }, { "epoch": 0.40119707145020217, "grad_norm": 1.0734220190256785, "learning_rate": 1.3603332126186728e-05, "loss": 0.5053, "step": 11261 }, { "epoch": 0.4012326985766963, "grad_norm": 1.0555363336298536, "learning_rate": 1.360225568274989e-05, "loss": 0.4145, "step": 11262 }, { "epoch": 0.40126832570319043, "grad_norm": 1.0147876420524662, "learning_rate": 1.3601179191346502e-05, "loss": 0.4806, "step": 11263 }, { "epoch": 0.4013039528296845, "grad_norm": 1.1854137605509112, "learning_rate": 1.360010265199089e-05, "loss": 0.4734, "step": 11264 }, { "epoch": 0.40133957995617864, "grad_norm": 1.0973619981554117, "learning_rate": 1.3599026064697399e-05, "loss": 0.461, "step": 11265 }, { "epoch": 0.40137520708267277, "grad_norm": 1.1547832166825966, "learning_rate": 1.3597949429480361e-05, "loss": 0.4955, "step": 11266 }, { "epoch": 0.40141083420916684, "grad_norm": 0.9922525431687237, "learning_rate": 1.3596872746354108e-05, "loss": 0.4284, "step": 11267 }, { "epoch": 0.401446461335661, "grad_norm": 1.1175470891846881, "learning_rate": 1.3595796015332986e-05, "loss": 0.4958, "step": 11268 }, { "epoch": 0.4014820884621551, "grad_norm": 1.8365732332436078, "learning_rate": 1.3594719236431321e-05, "loss": 0.5995, "step": 11269 }, { "epoch": 0.4015177155886492, "grad_norm": 1.3875672165304263, "learning_rate": 1.3593642409663461e-05, "loss": 0.5878, "step": 11270 }, { "epoch": 0.4015533427151433, "grad_norm": 1.4417418987455348, "learning_rate": 1.3592565535043738e-05, "loss": 0.5817, "step": 11271 }, { "epoch": 0.40158896984163744, "grad_norm": 1.1734559334120487, "learning_rate": 1.3591488612586492e-05, "loss": 0.4433, "step": 11272 }, { "epoch": 0.4016245969681315, "grad_norm": 1.1499361094441112, "learning_rate": 1.3590411642306068e-05, "loss": 0.4928, "step": 11273 }, { "epoch": 0.40166022409462565, "grad_norm": 1.0315394190192742, "learning_rate": 1.3589334624216804e-05, "loss": 0.3808, "step": 11274 }, { "epoch": 0.4016958512211198, "grad_norm": 1.8083299479189, "learning_rate": 1.3588257558333039e-05, "loss": 0.6074, "step": 11275 }, { "epoch": 0.40173147834761386, "grad_norm": 1.7092113777907163, "learning_rate": 1.3587180444669117e-05, "loss": 0.5411, "step": 11276 }, { "epoch": 0.401767105474108, "grad_norm": 1.4024369890713673, "learning_rate": 1.358610328323938e-05, "loss": 0.5796, "step": 11277 }, { "epoch": 0.4018027326006021, "grad_norm": 1.1210446668233913, "learning_rate": 1.358502607405817e-05, "loss": 0.4381, "step": 11278 }, { "epoch": 0.4018383597270962, "grad_norm": 1.4541685205682284, "learning_rate": 1.3583948817139833e-05, "loss": 0.509, "step": 11279 }, { "epoch": 0.4018739868535903, "grad_norm": 1.4381676918016297, "learning_rate": 1.3582871512498714e-05, "loss": 0.6368, "step": 11280 }, { "epoch": 0.40190961398008446, "grad_norm": 1.1186106746510782, "learning_rate": 1.3581794160149158e-05, "loss": 0.5154, "step": 11281 }, { "epoch": 0.40194524110657853, "grad_norm": 1.1464616909368344, "learning_rate": 1.3580716760105504e-05, "loss": 0.4638, "step": 11282 }, { "epoch": 0.40198086823307266, "grad_norm": 1.2905778295662407, "learning_rate": 1.3579639312382105e-05, "loss": 0.6119, "step": 11283 }, { "epoch": 0.4020164953595668, "grad_norm": 1.3255525724720485, "learning_rate": 1.3578561816993307e-05, "loss": 0.5236, "step": 11284 }, { "epoch": 0.40205212248606087, "grad_norm": 1.0891035203850779, "learning_rate": 1.3577484273953458e-05, "loss": 0.4906, "step": 11285 }, { "epoch": 0.402087749612555, "grad_norm": 1.0607132438155593, "learning_rate": 1.3576406683276905e-05, "loss": 0.4377, "step": 11286 }, { "epoch": 0.40212337673904913, "grad_norm": 1.4189753981970785, "learning_rate": 1.3575329044977998e-05, "loss": 0.5163, "step": 11287 }, { "epoch": 0.4021590038655432, "grad_norm": 1.6495672151832357, "learning_rate": 1.3574251359071087e-05, "loss": 0.5106, "step": 11288 }, { "epoch": 0.40219463099203734, "grad_norm": 1.2028728889153224, "learning_rate": 1.357317362557052e-05, "loss": 0.4768, "step": 11289 }, { "epoch": 0.40223025811853147, "grad_norm": 1.671235683812313, "learning_rate": 1.3572095844490645e-05, "loss": 0.6447, "step": 11290 }, { "epoch": 0.40226588524502555, "grad_norm": 1.021665299518345, "learning_rate": 1.357101801584582e-05, "loss": 0.49, "step": 11291 }, { "epoch": 0.4023015123715197, "grad_norm": 1.6291882640090456, "learning_rate": 1.3569940139650395e-05, "loss": 0.5153, "step": 11292 }, { "epoch": 0.4023371394980138, "grad_norm": 1.4204947062506645, "learning_rate": 1.356886221591872e-05, "loss": 0.6174, "step": 11293 }, { "epoch": 0.4023727666245079, "grad_norm": 1.089880358473386, "learning_rate": 1.356778424466515e-05, "loss": 0.5182, "step": 11294 }, { "epoch": 0.402408393751002, "grad_norm": 1.1603173341166253, "learning_rate": 1.3566706225904041e-05, "loss": 0.4429, "step": 11295 }, { "epoch": 0.40244402087749614, "grad_norm": 1.4747367326099083, "learning_rate": 1.3565628159649745e-05, "loss": 0.5743, "step": 11296 }, { "epoch": 0.4024796480039902, "grad_norm": 1.3084709302600293, "learning_rate": 1.3564550045916616e-05, "loss": 0.5419, "step": 11297 }, { "epoch": 0.40251527513048435, "grad_norm": 1.085941203015688, "learning_rate": 1.3563471884719012e-05, "loss": 0.424, "step": 11298 }, { "epoch": 0.4025509022569785, "grad_norm": 0.9909627385357085, "learning_rate": 1.3562393676071293e-05, "loss": 0.5204, "step": 11299 }, { "epoch": 0.40258652938347256, "grad_norm": 1.0037096560784156, "learning_rate": 1.356131541998781e-05, "loss": 0.4858, "step": 11300 }, { "epoch": 0.4026221565099667, "grad_norm": 1.1038754375759265, "learning_rate": 1.356023711648292e-05, "loss": 0.5297, "step": 11301 }, { "epoch": 0.4026577836364608, "grad_norm": 1.175453297891583, "learning_rate": 1.3559158765570987e-05, "loss": 0.4768, "step": 11302 }, { "epoch": 0.4026934107629549, "grad_norm": 1.5297909547190087, "learning_rate": 1.355808036726637e-05, "loss": 0.6442, "step": 11303 }, { "epoch": 0.402729037889449, "grad_norm": 1.7945037994347168, "learning_rate": 1.3557001921583419e-05, "loss": 0.5756, "step": 11304 }, { "epoch": 0.40276466501594316, "grad_norm": 0.9835247504397054, "learning_rate": 1.3555923428536506e-05, "loss": 0.438, "step": 11305 }, { "epoch": 0.40280029214243723, "grad_norm": 1.0952118804476738, "learning_rate": 1.3554844888139986e-05, "loss": 0.5215, "step": 11306 }, { "epoch": 0.40283591926893136, "grad_norm": 1.0718935963863578, "learning_rate": 1.3553766300408225e-05, "loss": 0.4974, "step": 11307 }, { "epoch": 0.4028715463954255, "grad_norm": 1.5384997408068999, "learning_rate": 1.3552687665355577e-05, "loss": 0.5167, "step": 11308 }, { "epoch": 0.40290717352191957, "grad_norm": 1.4745160043661996, "learning_rate": 1.3551608982996412e-05, "loss": 0.5704, "step": 11309 }, { "epoch": 0.4029428006484137, "grad_norm": 0.9453169564809092, "learning_rate": 1.3550530253345088e-05, "loss": 0.4472, "step": 11310 }, { "epoch": 0.40297842777490783, "grad_norm": 1.3559268554574853, "learning_rate": 1.3549451476415976e-05, "loss": 0.609, "step": 11311 }, { "epoch": 0.4030140549014019, "grad_norm": 1.0936127638407904, "learning_rate": 1.3548372652223434e-05, "loss": 0.5138, "step": 11312 }, { "epoch": 0.40304968202789604, "grad_norm": 1.309506413640109, "learning_rate": 1.3547293780781832e-05, "loss": 0.5995, "step": 11313 }, { "epoch": 0.40308530915439017, "grad_norm": 1.820688558453031, "learning_rate": 1.3546214862105533e-05, "loss": 0.5829, "step": 11314 }, { "epoch": 0.40312093628088425, "grad_norm": 1.3111647989263655, "learning_rate": 1.3545135896208904e-05, "loss": 0.5713, "step": 11315 }, { "epoch": 0.4031565634073784, "grad_norm": 1.0664150001642732, "learning_rate": 1.3544056883106313e-05, "loss": 0.4703, "step": 11316 }, { "epoch": 0.4031921905338725, "grad_norm": 1.471019284985559, "learning_rate": 1.3542977822812126e-05, "loss": 0.6308, "step": 11317 }, { "epoch": 0.4032278176603666, "grad_norm": 1.1936363863814008, "learning_rate": 1.3541898715340716e-05, "loss": 0.4819, "step": 11318 }, { "epoch": 0.4032634447868607, "grad_norm": 1.0430463983805351, "learning_rate": 1.3540819560706446e-05, "loss": 0.4541, "step": 11319 }, { "epoch": 0.40329907191335485, "grad_norm": 1.2015580646830815, "learning_rate": 1.353974035892369e-05, "loss": 0.4952, "step": 11320 }, { "epoch": 0.4033346990398489, "grad_norm": 1.5326375040868199, "learning_rate": 1.3538661110006816e-05, "loss": 0.5537, "step": 11321 }, { "epoch": 0.40337032616634305, "grad_norm": 1.1225350043340594, "learning_rate": 1.3537581813970197e-05, "loss": 0.4779, "step": 11322 }, { "epoch": 0.4034059532928372, "grad_norm": 1.0680440270139644, "learning_rate": 1.3536502470828204e-05, "loss": 0.518, "step": 11323 }, { "epoch": 0.40344158041933126, "grad_norm": 1.4644381065768206, "learning_rate": 1.3535423080595209e-05, "loss": 0.492, "step": 11324 }, { "epoch": 0.4034772075458254, "grad_norm": 1.0718653635324895, "learning_rate": 1.3534343643285583e-05, "loss": 0.4849, "step": 11325 }, { "epoch": 0.4035128346723195, "grad_norm": 1.8592653176503893, "learning_rate": 1.3533264158913704e-05, "loss": 0.5963, "step": 11326 }, { "epoch": 0.4035484617988136, "grad_norm": 1.0863008081605456, "learning_rate": 1.3532184627493939e-05, "loss": 0.4628, "step": 11327 }, { "epoch": 0.4035840889253077, "grad_norm": 1.1212371140134514, "learning_rate": 1.3531105049040667e-05, "loss": 0.4807, "step": 11328 }, { "epoch": 0.40361971605180186, "grad_norm": 1.0726862934047567, "learning_rate": 1.3530025423568266e-05, "loss": 0.4748, "step": 11329 }, { "epoch": 0.40365534317829593, "grad_norm": 0.9988192671096309, "learning_rate": 1.3528945751091108e-05, "loss": 0.4457, "step": 11330 }, { "epoch": 0.40369097030479006, "grad_norm": 1.0880448785037402, "learning_rate": 1.3527866031623569e-05, "loss": 0.4517, "step": 11331 }, { "epoch": 0.4037265974312842, "grad_norm": 1.179417086379894, "learning_rate": 1.352678626518003e-05, "loss": 0.474, "step": 11332 }, { "epoch": 0.40376222455777827, "grad_norm": 1.034959861924822, "learning_rate": 1.3525706451774866e-05, "loss": 0.4818, "step": 11333 }, { "epoch": 0.4037978516842724, "grad_norm": 2.129851999303282, "learning_rate": 1.3524626591422458e-05, "loss": 0.515, "step": 11334 }, { "epoch": 0.40383347881076653, "grad_norm": 1.64065340469845, "learning_rate": 1.3523546684137183e-05, "loss": 0.5915, "step": 11335 }, { "epoch": 0.4038691059372606, "grad_norm": 1.6141052109447864, "learning_rate": 1.352246672993342e-05, "loss": 0.5824, "step": 11336 }, { "epoch": 0.40390473306375474, "grad_norm": 1.5851478561738879, "learning_rate": 1.3521386728825551e-05, "loss": 0.555, "step": 11337 }, { "epoch": 0.40394036019024887, "grad_norm": 1.611481069787669, "learning_rate": 1.3520306680827954e-05, "loss": 0.4987, "step": 11338 }, { "epoch": 0.40397598731674295, "grad_norm": 1.669008346370248, "learning_rate": 1.3519226585955016e-05, "loss": 0.5729, "step": 11339 }, { "epoch": 0.4040116144432371, "grad_norm": 1.0929044741073173, "learning_rate": 1.3518146444221114e-05, "loss": 0.4868, "step": 11340 }, { "epoch": 0.4040472415697312, "grad_norm": 1.668101699751385, "learning_rate": 1.3517066255640635e-05, "loss": 0.48, "step": 11341 }, { "epoch": 0.4040828686962253, "grad_norm": 1.3943945061284884, "learning_rate": 1.351598602022796e-05, "loss": 0.6199, "step": 11342 }, { "epoch": 0.4041184958227194, "grad_norm": 1.0660353786372454, "learning_rate": 1.3514905737997474e-05, "loss": 0.4416, "step": 11343 }, { "epoch": 0.40415412294921355, "grad_norm": 1.1858026730401203, "learning_rate": 1.3513825408963561e-05, "loss": 0.505, "step": 11344 }, { "epoch": 0.4041897500757076, "grad_norm": 1.0863467628825925, "learning_rate": 1.3512745033140608e-05, "loss": 0.5078, "step": 11345 }, { "epoch": 0.40422537720220175, "grad_norm": 1.7824450514591388, "learning_rate": 1.3511664610542996e-05, "loss": 0.5836, "step": 11346 }, { "epoch": 0.4042610043286959, "grad_norm": 1.7198573277419933, "learning_rate": 1.3510584141185117e-05, "loss": 0.5824, "step": 11347 }, { "epoch": 0.40429663145518996, "grad_norm": 1.1499920534503625, "learning_rate": 1.350950362508136e-05, "loss": 0.5199, "step": 11348 }, { "epoch": 0.4043322585816841, "grad_norm": 1.0946692282291302, "learning_rate": 1.3508423062246104e-05, "loss": 0.4601, "step": 11349 }, { "epoch": 0.4043678857081782, "grad_norm": 1.068123976079763, "learning_rate": 1.3507342452693743e-05, "loss": 0.4806, "step": 11350 }, { "epoch": 0.4044035128346723, "grad_norm": 1.503833324686848, "learning_rate": 1.3506261796438666e-05, "loss": 0.5973, "step": 11351 }, { "epoch": 0.4044391399611664, "grad_norm": 1.0684068362039523, "learning_rate": 1.3505181093495265e-05, "loss": 0.4694, "step": 11352 }, { "epoch": 0.40447476708766056, "grad_norm": 1.5782602553788465, "learning_rate": 1.3504100343877928e-05, "loss": 0.607, "step": 11353 }, { "epoch": 0.40451039421415463, "grad_norm": 1.1191609637725675, "learning_rate": 1.3503019547601044e-05, "loss": 0.4553, "step": 11354 }, { "epoch": 0.40454602134064876, "grad_norm": 1.850469711365767, "learning_rate": 1.3501938704679005e-05, "loss": 0.6096, "step": 11355 }, { "epoch": 0.4045816484671429, "grad_norm": 1.6645584287505655, "learning_rate": 1.350085781512621e-05, "loss": 0.5369, "step": 11356 }, { "epoch": 0.40461727559363697, "grad_norm": 1.01850411540217, "learning_rate": 1.3499776878957038e-05, "loss": 0.468, "step": 11357 }, { "epoch": 0.4046529027201311, "grad_norm": 1.204953317218008, "learning_rate": 1.3498695896185896e-05, "loss": 0.5105, "step": 11358 }, { "epoch": 0.40468852984662523, "grad_norm": 1.368470713629764, "learning_rate": 1.3497614866827173e-05, "loss": 0.5834, "step": 11359 }, { "epoch": 0.4047241569731193, "grad_norm": 1.0316880583081114, "learning_rate": 1.3496533790895262e-05, "loss": 0.4802, "step": 11360 }, { "epoch": 0.40475978409961344, "grad_norm": 1.1222898091090947, "learning_rate": 1.3495452668404558e-05, "loss": 0.5128, "step": 11361 }, { "epoch": 0.40479541122610757, "grad_norm": 1.111166510592714, "learning_rate": 1.3494371499369461e-05, "loss": 0.425, "step": 11362 }, { "epoch": 0.40483103835260165, "grad_norm": 1.517334368232487, "learning_rate": 1.3493290283804363e-05, "loss": 0.5637, "step": 11363 }, { "epoch": 0.4048666654790958, "grad_norm": 1.429519229411656, "learning_rate": 1.3492209021723665e-05, "loss": 0.5529, "step": 11364 }, { "epoch": 0.4049022926055899, "grad_norm": 1.413679898316668, "learning_rate": 1.349112771314176e-05, "loss": 0.477, "step": 11365 }, { "epoch": 0.404937919732084, "grad_norm": 1.3969290254876905, "learning_rate": 1.3490046358073055e-05, "loss": 0.5463, "step": 11366 }, { "epoch": 0.4049735468585781, "grad_norm": 1.2763095267456679, "learning_rate": 1.348896495653194e-05, "loss": 0.4691, "step": 11367 }, { "epoch": 0.40500917398507225, "grad_norm": 1.1114719632563206, "learning_rate": 1.3487883508532815e-05, "loss": 0.5016, "step": 11368 }, { "epoch": 0.4050448011115663, "grad_norm": 1.5609040491235735, "learning_rate": 1.3486802014090088e-05, "loss": 0.6658, "step": 11369 }, { "epoch": 0.40508042823806045, "grad_norm": 1.6606840124587106, "learning_rate": 1.3485720473218153e-05, "loss": 0.5017, "step": 11370 }, { "epoch": 0.4051160553645546, "grad_norm": 1.6340983863582192, "learning_rate": 1.3484638885931416e-05, "loss": 0.646, "step": 11371 }, { "epoch": 0.40515168249104866, "grad_norm": 1.662136896982208, "learning_rate": 1.3483557252244272e-05, "loss": 0.5752, "step": 11372 }, { "epoch": 0.4051873096175428, "grad_norm": 1.482516137013508, "learning_rate": 1.3482475572171132e-05, "loss": 0.6102, "step": 11373 }, { "epoch": 0.4052229367440369, "grad_norm": 1.7639137618260923, "learning_rate": 1.3481393845726396e-05, "loss": 0.5038, "step": 11374 }, { "epoch": 0.405258563870531, "grad_norm": 1.3509041722043078, "learning_rate": 1.3480312072924465e-05, "loss": 0.5422, "step": 11375 }, { "epoch": 0.4052941909970251, "grad_norm": 1.6509051172772826, "learning_rate": 1.3479230253779745e-05, "loss": 0.6243, "step": 11376 }, { "epoch": 0.40532981812351926, "grad_norm": 1.4555771656515866, "learning_rate": 1.3478148388306644e-05, "loss": 0.6495, "step": 11377 }, { "epoch": 0.40536544525001333, "grad_norm": 1.6045192636896604, "learning_rate": 1.347706647651957e-05, "loss": 0.5165, "step": 11378 }, { "epoch": 0.40540107237650747, "grad_norm": 1.1000565516802188, "learning_rate": 1.3475984518432918e-05, "loss": 0.4988, "step": 11379 }, { "epoch": 0.4054366995030016, "grad_norm": 1.2147217753294932, "learning_rate": 1.3474902514061108e-05, "loss": 0.5045, "step": 11380 }, { "epoch": 0.40547232662949567, "grad_norm": 1.3400087823509679, "learning_rate": 1.3473820463418542e-05, "loss": 0.5434, "step": 11381 }, { "epoch": 0.4055079537559898, "grad_norm": 1.1324460121379918, "learning_rate": 1.347273836651963e-05, "loss": 0.512, "step": 11382 }, { "epoch": 0.40554358088248393, "grad_norm": 1.0321242354438227, "learning_rate": 1.3471656223378773e-05, "loss": 0.4596, "step": 11383 }, { "epoch": 0.405579208008978, "grad_norm": 1.7886031383890493, "learning_rate": 1.3470574034010394e-05, "loss": 0.5747, "step": 11384 }, { "epoch": 0.40561483513547214, "grad_norm": 1.023412341001757, "learning_rate": 1.3469491798428891e-05, "loss": 0.4967, "step": 11385 }, { "epoch": 0.40565046226196627, "grad_norm": 1.1046584747906716, "learning_rate": 1.3468409516648679e-05, "loss": 0.4446, "step": 11386 }, { "epoch": 0.40568608938846035, "grad_norm": 1.2917554690934245, "learning_rate": 1.3467327188684171e-05, "loss": 0.5375, "step": 11387 }, { "epoch": 0.4057217165149545, "grad_norm": 1.0703733602419956, "learning_rate": 1.3466244814549781e-05, "loss": 0.4718, "step": 11388 }, { "epoch": 0.4057573436414486, "grad_norm": 1.1950694983963264, "learning_rate": 1.3465162394259919e-05, "loss": 0.4752, "step": 11389 }, { "epoch": 0.4057929707679427, "grad_norm": 1.057048219119886, "learning_rate": 1.3464079927828995e-05, "loss": 0.4948, "step": 11390 }, { "epoch": 0.4058285978944368, "grad_norm": 1.2636110764452608, "learning_rate": 1.3462997415271424e-05, "loss": 0.5512, "step": 11391 }, { "epoch": 0.40586422502093095, "grad_norm": 1.2862919054117352, "learning_rate": 1.3461914856601625e-05, "loss": 0.472, "step": 11392 }, { "epoch": 0.405899852147425, "grad_norm": 1.0905374693022507, "learning_rate": 1.346083225183401e-05, "loss": 0.4588, "step": 11393 }, { "epoch": 0.40593547927391915, "grad_norm": 2.3667900554189494, "learning_rate": 1.3459749600982991e-05, "loss": 0.5766, "step": 11394 }, { "epoch": 0.4059711064004133, "grad_norm": 1.0331339730925988, "learning_rate": 1.3458666904062992e-05, "loss": 0.4979, "step": 11395 }, { "epoch": 0.40600673352690736, "grad_norm": 1.5098051108186847, "learning_rate": 1.3457584161088424e-05, "loss": 0.5904, "step": 11396 }, { "epoch": 0.4060423606534015, "grad_norm": 1.0414691316190015, "learning_rate": 1.3456501372073706e-05, "loss": 0.4179, "step": 11397 }, { "epoch": 0.4060779877798956, "grad_norm": 1.5425380132219675, "learning_rate": 1.3455418537033253e-05, "loss": 0.5358, "step": 11398 }, { "epoch": 0.40611361490638975, "grad_norm": 1.1479073547812353, "learning_rate": 1.3454335655981491e-05, "loss": 0.4819, "step": 11399 }, { "epoch": 0.40614924203288383, "grad_norm": 1.4588231746967237, "learning_rate": 1.3453252728932837e-05, "loss": 0.6258, "step": 11400 }, { "epoch": 0.40618486915937796, "grad_norm": 1.922556153779632, "learning_rate": 1.3452169755901707e-05, "loss": 0.5573, "step": 11401 }, { "epoch": 0.4062204962858721, "grad_norm": 1.6884783673950523, "learning_rate": 1.3451086736902524e-05, "loss": 0.5566, "step": 11402 }, { "epoch": 0.40625612341236617, "grad_norm": 1.5461254293571316, "learning_rate": 1.3450003671949707e-05, "loss": 0.5908, "step": 11403 }, { "epoch": 0.4062917505388603, "grad_norm": 1.1711790915442788, "learning_rate": 1.3448920561057683e-05, "loss": 0.4497, "step": 11404 }, { "epoch": 0.40632737766535443, "grad_norm": 1.2783168149829942, "learning_rate": 1.3447837404240868e-05, "loss": 0.5075, "step": 11405 }, { "epoch": 0.4063630047918485, "grad_norm": 1.2828486947067335, "learning_rate": 1.3446754201513689e-05, "loss": 0.487, "step": 11406 }, { "epoch": 0.40639863191834263, "grad_norm": 2.7740564795486393, "learning_rate": 1.3445670952890573e-05, "loss": 0.538, "step": 11407 }, { "epoch": 0.40643425904483677, "grad_norm": 1.1695503444997708, "learning_rate": 1.3444587658385934e-05, "loss": 0.4789, "step": 11408 }, { "epoch": 0.40646988617133084, "grad_norm": 1.1534462370165968, "learning_rate": 1.3443504318014204e-05, "loss": 0.4727, "step": 11409 }, { "epoch": 0.40650551329782497, "grad_norm": 1.5625927865061733, "learning_rate": 1.3442420931789808e-05, "loss": 0.5209, "step": 11410 }, { "epoch": 0.4065411404243191, "grad_norm": 1.0438418397967333, "learning_rate": 1.3441337499727171e-05, "loss": 0.4965, "step": 11411 }, { "epoch": 0.4065767675508132, "grad_norm": 1.1184138644470492, "learning_rate": 1.3440254021840722e-05, "loss": 0.5022, "step": 11412 }, { "epoch": 0.4066123946773073, "grad_norm": 1.8308581694446968, "learning_rate": 1.3439170498144882e-05, "loss": 0.5798, "step": 11413 }, { "epoch": 0.40664802180380144, "grad_norm": 1.6426838412107134, "learning_rate": 1.3438086928654087e-05, "loss": 0.5304, "step": 11414 }, { "epoch": 0.4066836489302955, "grad_norm": 1.8216196694347568, "learning_rate": 1.343700331338276e-05, "loss": 0.6316, "step": 11415 }, { "epoch": 0.40671927605678965, "grad_norm": 1.1645886088354205, "learning_rate": 1.3435919652345331e-05, "loss": 0.4895, "step": 11416 }, { "epoch": 0.4067549031832838, "grad_norm": 1.6422758330786367, "learning_rate": 1.3434835945556228e-05, "loss": 0.625, "step": 11417 }, { "epoch": 0.40679053030977785, "grad_norm": 1.2121446061790766, "learning_rate": 1.3433752193029888e-05, "loss": 0.4165, "step": 11418 }, { "epoch": 0.406826157436272, "grad_norm": 1.973947465900001, "learning_rate": 1.3432668394780735e-05, "loss": 0.5985, "step": 11419 }, { "epoch": 0.4068617845627661, "grad_norm": 1.2425873551836342, "learning_rate": 1.3431584550823203e-05, "loss": 0.4828, "step": 11420 }, { "epoch": 0.4068974116892602, "grad_norm": 1.2593599480772444, "learning_rate": 1.3430500661171721e-05, "loss": 0.4409, "step": 11421 }, { "epoch": 0.4069330388157543, "grad_norm": 1.1196606127738489, "learning_rate": 1.3429416725840731e-05, "loss": 0.484, "step": 11422 }, { "epoch": 0.40696866594224845, "grad_norm": 1.3131438133847055, "learning_rate": 1.3428332744844659e-05, "loss": 0.4982, "step": 11423 }, { "epoch": 0.40700429306874253, "grad_norm": 1.03324127166532, "learning_rate": 1.3427248718197935e-05, "loss": 0.4233, "step": 11424 }, { "epoch": 0.40703992019523666, "grad_norm": 1.0074287166519982, "learning_rate": 1.3426164645915005e-05, "loss": 0.4629, "step": 11425 }, { "epoch": 0.4070755473217308, "grad_norm": 1.1006998052077506, "learning_rate": 1.3425080528010297e-05, "loss": 0.4789, "step": 11426 }, { "epoch": 0.40711117444822487, "grad_norm": 1.1385225000345696, "learning_rate": 1.3423996364498248e-05, "loss": 0.4537, "step": 11427 }, { "epoch": 0.407146801574719, "grad_norm": 2.2243279070153124, "learning_rate": 1.3422912155393291e-05, "loss": 0.5445, "step": 11428 }, { "epoch": 0.40718242870121313, "grad_norm": 1.2275556653678081, "learning_rate": 1.3421827900709867e-05, "loss": 0.4778, "step": 11429 }, { "epoch": 0.4072180558277072, "grad_norm": 1.1709322401581812, "learning_rate": 1.3420743600462416e-05, "loss": 0.4608, "step": 11430 }, { "epoch": 0.40725368295420133, "grad_norm": 1.5299120709321634, "learning_rate": 1.341965925466537e-05, "loss": 0.4939, "step": 11431 }, { "epoch": 0.40728931008069547, "grad_norm": 2.187717482286686, "learning_rate": 1.3418574863333169e-05, "loss": 0.5997, "step": 11432 }, { "epoch": 0.40732493720718954, "grad_norm": 1.386296370219677, "learning_rate": 1.3417490426480258e-05, "loss": 0.448, "step": 11433 }, { "epoch": 0.4073605643336837, "grad_norm": 1.085617383686303, "learning_rate": 1.3416405944121074e-05, "loss": 0.4246, "step": 11434 }, { "epoch": 0.4073961914601778, "grad_norm": 1.3418278574955838, "learning_rate": 1.3415321416270055e-05, "loss": 0.5297, "step": 11435 }, { "epoch": 0.4074318185866719, "grad_norm": 1.5516358449155967, "learning_rate": 1.3414236842941644e-05, "loss": 0.6261, "step": 11436 }, { "epoch": 0.407467445713166, "grad_norm": 1.1181414930639215, "learning_rate": 1.3413152224150282e-05, "loss": 0.4786, "step": 11437 }, { "epoch": 0.40750307283966014, "grad_norm": 1.0394873025409666, "learning_rate": 1.3412067559910416e-05, "loss": 0.4273, "step": 11438 }, { "epoch": 0.4075386999661542, "grad_norm": 1.0290310376875156, "learning_rate": 1.341098285023648e-05, "loss": 0.4316, "step": 11439 }, { "epoch": 0.40757432709264835, "grad_norm": 1.7693919971793322, "learning_rate": 1.3409898095142928e-05, "loss": 0.5766, "step": 11440 }, { "epoch": 0.4076099542191425, "grad_norm": 1.2768683266783825, "learning_rate": 1.3408813294644199e-05, "loss": 0.5265, "step": 11441 }, { "epoch": 0.40764558134563655, "grad_norm": 1.2132128074466793, "learning_rate": 1.3407728448754737e-05, "loss": 0.445, "step": 11442 }, { "epoch": 0.4076812084721307, "grad_norm": 1.2658691183916846, "learning_rate": 1.340664355748899e-05, "loss": 0.568, "step": 11443 }, { "epoch": 0.4077168355986248, "grad_norm": 1.2127640351894364, "learning_rate": 1.34055586208614e-05, "loss": 0.4398, "step": 11444 }, { "epoch": 0.4077524627251189, "grad_norm": 1.0779260416794, "learning_rate": 1.340447363888642e-05, "loss": 0.5006, "step": 11445 }, { "epoch": 0.407788089851613, "grad_norm": 1.1482885407210541, "learning_rate": 1.3403388611578491e-05, "loss": 0.4693, "step": 11446 }, { "epoch": 0.40782371697810715, "grad_norm": 1.1304886760264228, "learning_rate": 1.3402303538952065e-05, "loss": 0.4668, "step": 11447 }, { "epoch": 0.40785934410460123, "grad_norm": 1.181137968514526, "learning_rate": 1.3401218421021591e-05, "loss": 0.4956, "step": 11448 }, { "epoch": 0.40789497123109536, "grad_norm": 1.0490167579391714, "learning_rate": 1.3400133257801517e-05, "loss": 0.4966, "step": 11449 }, { "epoch": 0.4079305983575895, "grad_norm": 1.1126062275114996, "learning_rate": 1.3399048049306289e-05, "loss": 0.4539, "step": 11450 }, { "epoch": 0.40796622548408357, "grad_norm": 1.0777537118062988, "learning_rate": 1.3397962795550363e-05, "loss": 0.4861, "step": 11451 }, { "epoch": 0.4080018526105777, "grad_norm": 1.8083796012772348, "learning_rate": 1.3396877496548185e-05, "loss": 0.5531, "step": 11452 }, { "epoch": 0.40803747973707183, "grad_norm": 1.7445844133824253, "learning_rate": 1.3395792152314211e-05, "loss": 0.5489, "step": 11453 }, { "epoch": 0.4080731068635659, "grad_norm": 1.682456118743206, "learning_rate": 1.3394706762862887e-05, "loss": 0.5672, "step": 11454 }, { "epoch": 0.40810873399006004, "grad_norm": 3.2773780329673174, "learning_rate": 1.3393621328208673e-05, "loss": 0.6376, "step": 11455 }, { "epoch": 0.40814436111655417, "grad_norm": 1.09745472514701, "learning_rate": 1.339253584836602e-05, "loss": 0.4949, "step": 11456 }, { "epoch": 0.40817998824304824, "grad_norm": 1.5615466031254095, "learning_rate": 1.339145032334938e-05, "loss": 0.5704, "step": 11457 }, { "epoch": 0.4082156153695424, "grad_norm": 1.114365461370348, "learning_rate": 1.3390364753173206e-05, "loss": 0.4613, "step": 11458 }, { "epoch": 0.4082512424960365, "grad_norm": 1.1133649195348172, "learning_rate": 1.3389279137851961e-05, "loss": 0.4512, "step": 11459 }, { "epoch": 0.4082868696225306, "grad_norm": 1.413637246118942, "learning_rate": 1.3388193477400091e-05, "loss": 0.5821, "step": 11460 }, { "epoch": 0.4083224967490247, "grad_norm": 1.637567076904639, "learning_rate": 1.3387107771832058e-05, "loss": 0.5734, "step": 11461 }, { "epoch": 0.40835812387551884, "grad_norm": 1.230171053100222, "learning_rate": 1.3386022021162317e-05, "loss": 0.4777, "step": 11462 }, { "epoch": 0.4083937510020129, "grad_norm": 1.4749014611606166, "learning_rate": 1.3384936225405326e-05, "loss": 0.4877, "step": 11463 }, { "epoch": 0.40842937812850705, "grad_norm": 1.039352981811045, "learning_rate": 1.3383850384575546e-05, "loss": 0.4363, "step": 11464 }, { "epoch": 0.4084650052550012, "grad_norm": 1.3772648115399755, "learning_rate": 1.3382764498687429e-05, "loss": 0.5171, "step": 11465 }, { "epoch": 0.40850063238149525, "grad_norm": 1.4238737902575649, "learning_rate": 1.3381678567755438e-05, "loss": 0.6556, "step": 11466 }, { "epoch": 0.4085362595079894, "grad_norm": 1.1031947198607772, "learning_rate": 1.3380592591794034e-05, "loss": 0.4834, "step": 11467 }, { "epoch": 0.4085718866344835, "grad_norm": 1.6952643977967528, "learning_rate": 1.337950657081768e-05, "loss": 0.5514, "step": 11468 }, { "epoch": 0.4086075137609776, "grad_norm": 1.4922193660231446, "learning_rate": 1.337842050484083e-05, "loss": 0.5647, "step": 11469 }, { "epoch": 0.4086431408874717, "grad_norm": 1.028153443552013, "learning_rate": 1.3377334393877952e-05, "loss": 0.4407, "step": 11470 }, { "epoch": 0.40867876801396585, "grad_norm": 1.00991951029601, "learning_rate": 1.3376248237943505e-05, "loss": 0.4384, "step": 11471 }, { "epoch": 0.40871439514045993, "grad_norm": 1.2610424520374883, "learning_rate": 1.3375162037051955e-05, "loss": 0.4479, "step": 11472 }, { "epoch": 0.40875002226695406, "grad_norm": 1.4050727319149607, "learning_rate": 1.3374075791217758e-05, "loss": 0.5461, "step": 11473 }, { "epoch": 0.4087856493934482, "grad_norm": 1.643038182920227, "learning_rate": 1.3372989500455389e-05, "loss": 0.5486, "step": 11474 }, { "epoch": 0.40882127651994227, "grad_norm": 1.107768881200868, "learning_rate": 1.3371903164779306e-05, "loss": 0.4734, "step": 11475 }, { "epoch": 0.4088569036464364, "grad_norm": 1.0442673424773887, "learning_rate": 1.3370816784203973e-05, "loss": 0.4963, "step": 11476 }, { "epoch": 0.40889253077293053, "grad_norm": 1.126761172773765, "learning_rate": 1.336973035874386e-05, "loss": 0.4856, "step": 11477 }, { "epoch": 0.4089281578994246, "grad_norm": 1.0992208176687672, "learning_rate": 1.3368643888413431e-05, "loss": 0.4955, "step": 11478 }, { "epoch": 0.40896378502591874, "grad_norm": 1.073342710094227, "learning_rate": 1.3367557373227156e-05, "loss": 0.4731, "step": 11479 }, { "epoch": 0.40899941215241287, "grad_norm": 1.1615578151127877, "learning_rate": 1.33664708131995e-05, "loss": 0.496, "step": 11480 }, { "epoch": 0.40903503927890694, "grad_norm": 1.4753802006376775, "learning_rate": 1.3365384208344932e-05, "loss": 0.5973, "step": 11481 }, { "epoch": 0.4090706664054011, "grad_norm": 1.1393005297066137, "learning_rate": 1.3364297558677921e-05, "loss": 0.5071, "step": 11482 }, { "epoch": 0.4091062935318952, "grad_norm": 1.1066226752587893, "learning_rate": 1.3363210864212937e-05, "loss": 0.4902, "step": 11483 }, { "epoch": 0.4091419206583893, "grad_norm": 1.8774209831941988, "learning_rate": 1.3362124124964449e-05, "loss": 0.5817, "step": 11484 }, { "epoch": 0.4091775477848834, "grad_norm": 1.0201296662050656, "learning_rate": 1.3361037340946927e-05, "loss": 0.4346, "step": 11485 }, { "epoch": 0.40921317491137754, "grad_norm": 1.8607682986575103, "learning_rate": 1.3359950512174847e-05, "loss": 0.5572, "step": 11486 }, { "epoch": 0.4092488020378716, "grad_norm": 1.0779594542203381, "learning_rate": 1.3358863638662673e-05, "loss": 0.4436, "step": 11487 }, { "epoch": 0.40928442916436575, "grad_norm": 1.5812874802248262, "learning_rate": 1.3357776720424887e-05, "loss": 0.6141, "step": 11488 }, { "epoch": 0.4093200562908599, "grad_norm": 1.0925025135531687, "learning_rate": 1.3356689757475954e-05, "loss": 0.4617, "step": 11489 }, { "epoch": 0.40935568341735395, "grad_norm": 1.1018899292742321, "learning_rate": 1.335560274983035e-05, "loss": 0.4504, "step": 11490 }, { "epoch": 0.4093913105438481, "grad_norm": 1.0449503728767233, "learning_rate": 1.3354515697502552e-05, "loss": 0.4545, "step": 11491 }, { "epoch": 0.4094269376703422, "grad_norm": 1.0592367560001872, "learning_rate": 1.335342860050703e-05, "loss": 0.4726, "step": 11492 }, { "epoch": 0.4094625647968363, "grad_norm": 0.9938936806132955, "learning_rate": 1.3352341458858264e-05, "loss": 0.4748, "step": 11493 }, { "epoch": 0.4094981919233304, "grad_norm": 1.0693212744353857, "learning_rate": 1.3351254272570731e-05, "loss": 0.4571, "step": 11494 }, { "epoch": 0.40953381904982455, "grad_norm": 1.1278303090999355, "learning_rate": 1.3350167041658901e-05, "loss": 0.4512, "step": 11495 }, { "epoch": 0.40956944617631863, "grad_norm": 1.170041731330786, "learning_rate": 1.3349079766137259e-05, "loss": 0.5048, "step": 11496 }, { "epoch": 0.40960507330281276, "grad_norm": 1.649178768271447, "learning_rate": 1.3347992446020275e-05, "loss": 0.5982, "step": 11497 }, { "epoch": 0.4096407004293069, "grad_norm": 1.0495116802087154, "learning_rate": 1.3346905081322435e-05, "loss": 0.4724, "step": 11498 }, { "epoch": 0.40967632755580097, "grad_norm": 1.631030113301401, "learning_rate": 1.3345817672058212e-05, "loss": 0.5259, "step": 11499 }, { "epoch": 0.4097119546822951, "grad_norm": 1.039969061727427, "learning_rate": 1.3344730218242088e-05, "loss": 0.438, "step": 11500 }, { "epoch": 0.40974758180878923, "grad_norm": 1.0766435805051409, "learning_rate": 1.3343642719888545e-05, "loss": 0.4879, "step": 11501 }, { "epoch": 0.4097832089352833, "grad_norm": 1.1381133749364478, "learning_rate": 1.334255517701206e-05, "loss": 0.5201, "step": 11502 }, { "epoch": 0.40981883606177744, "grad_norm": 1.2861369553584516, "learning_rate": 1.3341467589627118e-05, "loss": 0.5619, "step": 11503 }, { "epoch": 0.40985446318827157, "grad_norm": 1.2708577899509472, "learning_rate": 1.33403799577482e-05, "loss": 0.4668, "step": 11504 }, { "epoch": 0.40989009031476564, "grad_norm": 1.329260906839646, "learning_rate": 1.3339292281389789e-05, "loss": 0.4741, "step": 11505 }, { "epoch": 0.4099257174412598, "grad_norm": 1.4388195457225814, "learning_rate": 1.3338204560566364e-05, "loss": 0.6452, "step": 11506 }, { "epoch": 0.4099613445677539, "grad_norm": 1.7590033220812924, "learning_rate": 1.3337116795292413e-05, "loss": 0.5946, "step": 11507 }, { "epoch": 0.409996971694248, "grad_norm": 1.1494873549025293, "learning_rate": 1.333602898558242e-05, "loss": 0.4712, "step": 11508 }, { "epoch": 0.4100325988207421, "grad_norm": 1.025522584402984, "learning_rate": 1.333494113145087e-05, "loss": 0.4446, "step": 11509 }, { "epoch": 0.41006822594723624, "grad_norm": 1.1021626033416552, "learning_rate": 1.3333853232912245e-05, "loss": 0.4482, "step": 11510 }, { "epoch": 0.4101038530737303, "grad_norm": 1.346328485226591, "learning_rate": 1.3332765289981038e-05, "loss": 0.4613, "step": 11511 }, { "epoch": 0.41013948020022445, "grad_norm": 1.9448068307639648, "learning_rate": 1.333167730267173e-05, "loss": 0.5801, "step": 11512 }, { "epoch": 0.4101751073267186, "grad_norm": 1.4662550479322791, "learning_rate": 1.3330589270998809e-05, "loss": 0.5214, "step": 11513 }, { "epoch": 0.41021073445321266, "grad_norm": 1.3626228110808227, "learning_rate": 1.3329501194976764e-05, "loss": 0.5761, "step": 11514 }, { "epoch": 0.4102463615797068, "grad_norm": 1.2663987897372948, "learning_rate": 1.3328413074620086e-05, "loss": 0.5192, "step": 11515 }, { "epoch": 0.4102819887062009, "grad_norm": 1.0978705308205967, "learning_rate": 1.3327324909943257e-05, "loss": 0.4487, "step": 11516 }, { "epoch": 0.410317615832695, "grad_norm": 1.1117357405046264, "learning_rate": 1.3326236700960777e-05, "loss": 0.4699, "step": 11517 }, { "epoch": 0.4103532429591891, "grad_norm": 1.7295831405257147, "learning_rate": 1.3325148447687126e-05, "loss": 0.6095, "step": 11518 }, { "epoch": 0.41038887008568325, "grad_norm": 1.4570420510556392, "learning_rate": 1.3324060150136799e-05, "loss": 0.57, "step": 11519 }, { "epoch": 0.41042449721217733, "grad_norm": 1.084225100657189, "learning_rate": 1.3322971808324292e-05, "loss": 0.4451, "step": 11520 }, { "epoch": 0.41046012433867146, "grad_norm": 1.0988031458388732, "learning_rate": 1.3321883422264088e-05, "loss": 0.4945, "step": 11521 }, { "epoch": 0.4104957514651656, "grad_norm": 1.3297098551499258, "learning_rate": 1.3320794991970688e-05, "loss": 0.5601, "step": 11522 }, { "epoch": 0.41053137859165967, "grad_norm": 1.0910805329392583, "learning_rate": 1.3319706517458579e-05, "loss": 0.5114, "step": 11523 }, { "epoch": 0.4105670057181538, "grad_norm": 1.0577887836805802, "learning_rate": 1.331861799874226e-05, "loss": 0.4485, "step": 11524 }, { "epoch": 0.41060263284464793, "grad_norm": 1.07523999592189, "learning_rate": 1.3317529435836219e-05, "loss": 0.4879, "step": 11525 }, { "epoch": 0.410638259971142, "grad_norm": 1.1609243523904877, "learning_rate": 1.3316440828754957e-05, "loss": 0.5044, "step": 11526 }, { "epoch": 0.41067388709763614, "grad_norm": 1.1752510332561648, "learning_rate": 1.3315352177512967e-05, "loss": 0.5037, "step": 11527 }, { "epoch": 0.41070951422413027, "grad_norm": 1.1600544947298392, "learning_rate": 1.3314263482124748e-05, "loss": 0.4944, "step": 11528 }, { "epoch": 0.41074514135062434, "grad_norm": 1.1068020156494178, "learning_rate": 1.3313174742604788e-05, "loss": 0.4869, "step": 11529 }, { "epoch": 0.4107807684771185, "grad_norm": 1.073900310630389, "learning_rate": 1.3312085958967595e-05, "loss": 0.4705, "step": 11530 }, { "epoch": 0.4108163956036126, "grad_norm": 1.078763356883108, "learning_rate": 1.3310997131227664e-05, "loss": 0.499, "step": 11531 }, { "epoch": 0.4108520227301067, "grad_norm": 1.384946056100233, "learning_rate": 1.330990825939949e-05, "loss": 0.5665, "step": 11532 }, { "epoch": 0.4108876498566008, "grad_norm": 1.1556633244733547, "learning_rate": 1.3308819343497572e-05, "loss": 0.4845, "step": 11533 }, { "epoch": 0.41092327698309494, "grad_norm": 1.2739113207385833, "learning_rate": 1.3307730383536415e-05, "loss": 0.5641, "step": 11534 }, { "epoch": 0.410958904109589, "grad_norm": 1.2728345176373532, "learning_rate": 1.3306641379530515e-05, "loss": 0.6368, "step": 11535 }, { "epoch": 0.41099453123608315, "grad_norm": 1.002330832602996, "learning_rate": 1.3305552331494371e-05, "loss": 0.4326, "step": 11536 }, { "epoch": 0.4110301583625773, "grad_norm": 1.0955070909241544, "learning_rate": 1.3304463239442489e-05, "loss": 0.4449, "step": 11537 }, { "epoch": 0.41106578548907136, "grad_norm": 1.2877650772426814, "learning_rate": 1.3303374103389371e-05, "loss": 0.5391, "step": 11538 }, { "epoch": 0.4111014126155655, "grad_norm": 1.666717871380907, "learning_rate": 1.3302284923349517e-05, "loss": 0.5838, "step": 11539 }, { "epoch": 0.4111370397420596, "grad_norm": 1.0882665539929248, "learning_rate": 1.330119569933743e-05, "loss": 0.4663, "step": 11540 }, { "epoch": 0.4111726668685537, "grad_norm": 1.0748985535167692, "learning_rate": 1.3300106431367617e-05, "loss": 0.4587, "step": 11541 }, { "epoch": 0.4112082939950478, "grad_norm": 1.135094691324662, "learning_rate": 1.3299017119454579e-05, "loss": 0.4774, "step": 11542 }, { "epoch": 0.41124392112154196, "grad_norm": 2.2751008911897888, "learning_rate": 1.3297927763612822e-05, "loss": 0.6381, "step": 11543 }, { "epoch": 0.41127954824803603, "grad_norm": 1.405406379416694, "learning_rate": 1.3296838363856852e-05, "loss": 0.5109, "step": 11544 }, { "epoch": 0.41131517537453016, "grad_norm": 1.5117837486561636, "learning_rate": 1.3295748920201175e-05, "loss": 0.5606, "step": 11545 }, { "epoch": 0.4113508025010243, "grad_norm": 1.2947118312187784, "learning_rate": 1.3294659432660297e-05, "loss": 0.5203, "step": 11546 }, { "epoch": 0.41138642962751837, "grad_norm": 1.2162327779091087, "learning_rate": 1.3293569901248727e-05, "loss": 0.4611, "step": 11547 }, { "epoch": 0.4114220567540125, "grad_norm": 1.8799441778716557, "learning_rate": 1.3292480325980967e-05, "loss": 0.5264, "step": 11548 }, { "epoch": 0.41145768388050663, "grad_norm": 1.0357153740330751, "learning_rate": 1.3291390706871536e-05, "loss": 0.4979, "step": 11549 }, { "epoch": 0.4114933110070007, "grad_norm": 1.7299915345321717, "learning_rate": 1.3290301043934935e-05, "loss": 0.49, "step": 11550 }, { "epoch": 0.41152893813349484, "grad_norm": 1.4023868052054802, "learning_rate": 1.3289211337185674e-05, "loss": 0.5558, "step": 11551 }, { "epoch": 0.41156456525998897, "grad_norm": 1.8358612946417927, "learning_rate": 1.3288121586638268e-05, "loss": 0.61, "step": 11552 }, { "epoch": 0.41160019238648304, "grad_norm": 1.728694333333, "learning_rate": 1.3287031792307226e-05, "loss": 0.611, "step": 11553 }, { "epoch": 0.4116358195129772, "grad_norm": 1.480279013261061, "learning_rate": 1.3285941954207055e-05, "loss": 0.5702, "step": 11554 }, { "epoch": 0.4116714466394713, "grad_norm": 1.2595551847059892, "learning_rate": 1.328485207235227e-05, "loss": 0.4841, "step": 11555 }, { "epoch": 0.4117070737659654, "grad_norm": 1.377054346379981, "learning_rate": 1.3283762146757387e-05, "loss": 0.5516, "step": 11556 }, { "epoch": 0.4117427008924595, "grad_norm": 1.4656940860420555, "learning_rate": 1.3282672177436913e-05, "loss": 0.5852, "step": 11557 }, { "epoch": 0.41177832801895364, "grad_norm": 1.1255540686900183, "learning_rate": 1.3281582164405364e-05, "loss": 0.4975, "step": 11558 }, { "epoch": 0.4118139551454477, "grad_norm": 1.696042965011256, "learning_rate": 1.3280492107677255e-05, "loss": 0.5818, "step": 11559 }, { "epoch": 0.41184958227194185, "grad_norm": 1.7988666985027881, "learning_rate": 1.3279402007267101e-05, "loss": 0.471, "step": 11560 }, { "epoch": 0.411885209398436, "grad_norm": 1.1052231706075422, "learning_rate": 1.327831186318942e-05, "loss": 0.4706, "step": 11561 }, { "epoch": 0.41192083652493006, "grad_norm": 1.0225641081642665, "learning_rate": 1.327722167545872e-05, "loss": 0.5042, "step": 11562 }, { "epoch": 0.4119564636514242, "grad_norm": 1.116195269697669, "learning_rate": 1.3276131444089526e-05, "loss": 0.4804, "step": 11563 }, { "epoch": 0.4119920907779183, "grad_norm": 1.1369416899928833, "learning_rate": 1.3275041169096351e-05, "loss": 0.454, "step": 11564 }, { "epoch": 0.4120277179044124, "grad_norm": 1.1060864778032091, "learning_rate": 1.3273950850493717e-05, "loss": 0.4792, "step": 11565 }, { "epoch": 0.4120633450309065, "grad_norm": 1.5607646902652303, "learning_rate": 1.3272860488296134e-05, "loss": 0.5195, "step": 11566 }, { "epoch": 0.41209897215740066, "grad_norm": 1.6730679688399124, "learning_rate": 1.327177008251813e-05, "loss": 0.5407, "step": 11567 }, { "epoch": 0.41213459928389473, "grad_norm": 3.1859460007588103, "learning_rate": 1.3270679633174219e-05, "loss": 0.6361, "step": 11568 }, { "epoch": 0.41217022641038886, "grad_norm": 1.093128204983767, "learning_rate": 1.326958914027892e-05, "loss": 0.4679, "step": 11569 }, { "epoch": 0.412205853536883, "grad_norm": 1.65219081438902, "learning_rate": 1.326849860384676e-05, "loss": 0.5678, "step": 11570 }, { "epoch": 0.41224148066337707, "grad_norm": 1.6787024208900772, "learning_rate": 1.3267408023892255e-05, "loss": 0.5189, "step": 11571 }, { "epoch": 0.4122771077898712, "grad_norm": 1.2465069841329761, "learning_rate": 1.3266317400429928e-05, "loss": 0.4939, "step": 11572 }, { "epoch": 0.41231273491636533, "grad_norm": 1.4755165115320015, "learning_rate": 1.3265226733474304e-05, "loss": 0.5393, "step": 11573 }, { "epoch": 0.4123483620428594, "grad_norm": 1.1559826127146642, "learning_rate": 1.32641360230399e-05, "loss": 0.4473, "step": 11574 }, { "epoch": 0.41238398916935354, "grad_norm": 1.1924737158953518, "learning_rate": 1.3263045269141248e-05, "loss": 0.5213, "step": 11575 }, { "epoch": 0.41241961629584767, "grad_norm": 1.9215330482480948, "learning_rate": 1.3261954471792864e-05, "loss": 0.5708, "step": 11576 }, { "epoch": 0.41245524342234174, "grad_norm": 1.040973390497971, "learning_rate": 1.3260863631009277e-05, "loss": 0.454, "step": 11577 }, { "epoch": 0.4124908705488359, "grad_norm": 1.1457323062228348, "learning_rate": 1.325977274680501e-05, "loss": 0.4589, "step": 11578 }, { "epoch": 0.41252649767533, "grad_norm": 1.0441845072082696, "learning_rate": 1.3258681819194595e-05, "loss": 0.4364, "step": 11579 }, { "epoch": 0.4125621248018241, "grad_norm": 1.1837651268627671, "learning_rate": 1.3257590848192552e-05, "loss": 0.5601, "step": 11580 }, { "epoch": 0.4125977519283182, "grad_norm": 1.3566864377672236, "learning_rate": 1.3256499833813407e-05, "loss": 0.5809, "step": 11581 }, { "epoch": 0.41263337905481234, "grad_norm": 2.053160891711095, "learning_rate": 1.3255408776071692e-05, "loss": 0.541, "step": 11582 }, { "epoch": 0.4126690061813064, "grad_norm": 2.2188062341465207, "learning_rate": 1.3254317674981935e-05, "loss": 0.621, "step": 11583 }, { "epoch": 0.41270463330780055, "grad_norm": 1.4800833054706035, "learning_rate": 1.3253226530558663e-05, "loss": 0.6083, "step": 11584 }, { "epoch": 0.4127402604342947, "grad_norm": 1.2431118598709496, "learning_rate": 1.3252135342816406e-05, "loss": 0.4852, "step": 11585 }, { "epoch": 0.4127758875607888, "grad_norm": 1.63976612768262, "learning_rate": 1.3251044111769693e-05, "loss": 0.5349, "step": 11586 }, { "epoch": 0.4128115146872829, "grad_norm": 1.1508190982463828, "learning_rate": 1.3249952837433058e-05, "loss": 0.5106, "step": 11587 }, { "epoch": 0.412847141813777, "grad_norm": 1.1581172127179973, "learning_rate": 1.3248861519821026e-05, "loss": 0.4555, "step": 11588 }, { "epoch": 0.41288276894027115, "grad_norm": 1.1334802603819902, "learning_rate": 1.3247770158948131e-05, "loss": 0.483, "step": 11589 }, { "epoch": 0.4129183960667652, "grad_norm": 1.0916680640503529, "learning_rate": 1.3246678754828913e-05, "loss": 0.4677, "step": 11590 }, { "epoch": 0.41295402319325936, "grad_norm": 1.2930538513992467, "learning_rate": 1.3245587307477891e-05, "loss": 0.4815, "step": 11591 }, { "epoch": 0.4129896503197535, "grad_norm": 1.625033615877515, "learning_rate": 1.3244495816909608e-05, "loss": 0.5551, "step": 11592 }, { "epoch": 0.41302527744624756, "grad_norm": 1.1081594525834857, "learning_rate": 1.3243404283138596e-05, "loss": 0.4354, "step": 11593 }, { "epoch": 0.4130609045727417, "grad_norm": 1.166667158129884, "learning_rate": 1.3242312706179388e-05, "loss": 0.4837, "step": 11594 }, { "epoch": 0.4130965316992358, "grad_norm": 1.9103738548608882, "learning_rate": 1.324122108604652e-05, "loss": 0.5706, "step": 11595 }, { "epoch": 0.4131321588257299, "grad_norm": 1.1987170467487607, "learning_rate": 1.3240129422754528e-05, "loss": 0.4552, "step": 11596 }, { "epoch": 0.41316778595222403, "grad_norm": 1.1566617279598812, "learning_rate": 1.3239037716317951e-05, "loss": 0.3982, "step": 11597 }, { "epoch": 0.41320341307871816, "grad_norm": 1.2659785349302062, "learning_rate": 1.323794596675132e-05, "loss": 0.479, "step": 11598 }, { "epoch": 0.41323904020521224, "grad_norm": 1.3093215394701743, "learning_rate": 1.3236854174069177e-05, "loss": 0.4972, "step": 11599 }, { "epoch": 0.41327466733170637, "grad_norm": 1.0767555974133836, "learning_rate": 1.3235762338286056e-05, "loss": 0.4498, "step": 11600 }, { "epoch": 0.4133102944582005, "grad_norm": 1.0521032529234817, "learning_rate": 1.3234670459416498e-05, "loss": 0.4532, "step": 11601 }, { "epoch": 0.4133459215846946, "grad_norm": 1.5446348761052047, "learning_rate": 1.3233578537475043e-05, "loss": 0.5003, "step": 11602 }, { "epoch": 0.4133815487111887, "grad_norm": 1.393027581987522, "learning_rate": 1.3232486572476229e-05, "loss": 0.5228, "step": 11603 }, { "epoch": 0.41341717583768284, "grad_norm": 1.0521468563957068, "learning_rate": 1.3231394564434596e-05, "loss": 0.4578, "step": 11604 }, { "epoch": 0.4134528029641769, "grad_norm": 1.3537080493977198, "learning_rate": 1.3230302513364689e-05, "loss": 0.4991, "step": 11605 }, { "epoch": 0.41348843009067104, "grad_norm": 1.8728742584062268, "learning_rate": 1.3229210419281042e-05, "loss": 0.6309, "step": 11606 }, { "epoch": 0.4135240572171652, "grad_norm": 1.1611992259560877, "learning_rate": 1.3228118282198202e-05, "loss": 0.4559, "step": 11607 }, { "epoch": 0.41355968434365925, "grad_norm": 1.1486424067648755, "learning_rate": 1.3227026102130713e-05, "loss": 0.4873, "step": 11608 }, { "epoch": 0.4135953114701534, "grad_norm": 1.2445299829641658, "learning_rate": 1.3225933879093116e-05, "loss": 0.4774, "step": 11609 }, { "epoch": 0.4136309385966475, "grad_norm": 1.4260439461471472, "learning_rate": 1.3224841613099955e-05, "loss": 0.6247, "step": 11610 }, { "epoch": 0.4136665657231416, "grad_norm": 1.4960598688599098, "learning_rate": 1.322374930416577e-05, "loss": 0.564, "step": 11611 }, { "epoch": 0.4137021928496357, "grad_norm": 1.1068254232736257, "learning_rate": 1.3222656952305113e-05, "loss": 0.4569, "step": 11612 }, { "epoch": 0.41373781997612985, "grad_norm": 1.06991091346801, "learning_rate": 1.3221564557532526e-05, "loss": 0.4571, "step": 11613 }, { "epoch": 0.4137734471026239, "grad_norm": 1.1165017032821147, "learning_rate": 1.3220472119862553e-05, "loss": 0.4883, "step": 11614 }, { "epoch": 0.41380907422911806, "grad_norm": 1.5455218883437525, "learning_rate": 1.3219379639309743e-05, "loss": 0.5756, "step": 11615 }, { "epoch": 0.4138447013556122, "grad_norm": 2.4609888854895963, "learning_rate": 1.3218287115888647e-05, "loss": 0.5966, "step": 11616 }, { "epoch": 0.41388032848210626, "grad_norm": 1.460715054816572, "learning_rate": 1.3217194549613804e-05, "loss": 0.5215, "step": 11617 }, { "epoch": 0.4139159556086004, "grad_norm": 1.417903898127091, "learning_rate": 1.3216101940499768e-05, "loss": 0.5122, "step": 11618 }, { "epoch": 0.4139515827350945, "grad_norm": 1.1337281539748938, "learning_rate": 1.3215009288561087e-05, "loss": 0.4831, "step": 11619 }, { "epoch": 0.4139872098615886, "grad_norm": 1.0238830840725466, "learning_rate": 1.3213916593812313e-05, "loss": 0.4134, "step": 11620 }, { "epoch": 0.41402283698808273, "grad_norm": 1.0496509620260726, "learning_rate": 1.3212823856267991e-05, "loss": 0.4827, "step": 11621 }, { "epoch": 0.41405846411457686, "grad_norm": 1.592996158374464, "learning_rate": 1.321173107594267e-05, "loss": 0.5355, "step": 11622 }, { "epoch": 0.41409409124107094, "grad_norm": 1.6081742264920145, "learning_rate": 1.321063825285091e-05, "loss": 0.5293, "step": 11623 }, { "epoch": 0.41412971836756507, "grad_norm": 1.0488996437672837, "learning_rate": 1.3209545387007257e-05, "loss": 0.4165, "step": 11624 }, { "epoch": 0.4141653454940592, "grad_norm": 1.3774700203874424, "learning_rate": 1.3208452478426261e-05, "loss": 0.6106, "step": 11625 }, { "epoch": 0.4142009726205533, "grad_norm": 1.1338479193646338, "learning_rate": 1.3207359527122479e-05, "loss": 0.5026, "step": 11626 }, { "epoch": 0.4142365997470474, "grad_norm": 1.1893644095213283, "learning_rate": 1.3206266533110466e-05, "loss": 0.4845, "step": 11627 }, { "epoch": 0.41427222687354154, "grad_norm": 1.1588646404369032, "learning_rate": 1.3205173496404768e-05, "loss": 0.4676, "step": 11628 }, { "epoch": 0.4143078540000356, "grad_norm": 1.9411198462819574, "learning_rate": 1.320408041701995e-05, "loss": 0.5854, "step": 11629 }, { "epoch": 0.41434348112652974, "grad_norm": 1.1454897097349912, "learning_rate": 1.3202987294970555e-05, "loss": 0.4835, "step": 11630 }, { "epoch": 0.4143791082530239, "grad_norm": 1.1165210334660942, "learning_rate": 1.3201894130271148e-05, "loss": 0.4704, "step": 11631 }, { "epoch": 0.41441473537951795, "grad_norm": 1.1095203130999591, "learning_rate": 1.3200800922936287e-05, "loss": 0.4919, "step": 11632 }, { "epoch": 0.4144503625060121, "grad_norm": 1.0789386970791757, "learning_rate": 1.319970767298052e-05, "loss": 0.4809, "step": 11633 }, { "epoch": 0.4144859896325062, "grad_norm": 1.083768382630114, "learning_rate": 1.3198614380418412e-05, "loss": 0.4634, "step": 11634 }, { "epoch": 0.4145216167590003, "grad_norm": 1.6096875470582233, "learning_rate": 1.3197521045264517e-05, "loss": 0.6136, "step": 11635 }, { "epoch": 0.4145572438854944, "grad_norm": 1.1253471676649642, "learning_rate": 1.3196427667533393e-05, "loss": 0.4637, "step": 11636 }, { "epoch": 0.41459287101198855, "grad_norm": 2.6322759186174927, "learning_rate": 1.31953342472396e-05, "loss": 0.5548, "step": 11637 }, { "epoch": 0.4146284981384826, "grad_norm": 1.4949577528751103, "learning_rate": 1.3194240784397701e-05, "loss": 0.5727, "step": 11638 }, { "epoch": 0.41466412526497676, "grad_norm": 1.056756528392793, "learning_rate": 1.319314727902225e-05, "loss": 0.4336, "step": 11639 }, { "epoch": 0.4146997523914709, "grad_norm": 1.1155597789965974, "learning_rate": 1.3192053731127812e-05, "loss": 0.4817, "step": 11640 }, { "epoch": 0.41473537951796496, "grad_norm": 1.0608639124379795, "learning_rate": 1.3190960140728946e-05, "loss": 0.463, "step": 11641 }, { "epoch": 0.4147710066444591, "grad_norm": 1.0731537172668864, "learning_rate": 1.3189866507840217e-05, "loss": 0.4282, "step": 11642 }, { "epoch": 0.4148066337709532, "grad_norm": 1.0651194118777394, "learning_rate": 1.318877283247619e-05, "loss": 0.4495, "step": 11643 }, { "epoch": 0.4148422608974473, "grad_norm": 1.3894104733866424, "learning_rate": 1.3187679114651415e-05, "loss": 0.4777, "step": 11644 }, { "epoch": 0.41487788802394143, "grad_norm": 1.1169859571494716, "learning_rate": 1.318658535438047e-05, "loss": 0.4187, "step": 11645 }, { "epoch": 0.41491351515043556, "grad_norm": 1.4712310577014436, "learning_rate": 1.3185491551677917e-05, "loss": 0.6463, "step": 11646 }, { "epoch": 0.41494914227692964, "grad_norm": 6.304457019815835, "learning_rate": 1.3184397706558314e-05, "loss": 0.5183, "step": 11647 }, { "epoch": 0.41498476940342377, "grad_norm": 1.4573053409514967, "learning_rate": 1.3183303819036226e-05, "loss": 0.6038, "step": 11648 }, { "epoch": 0.4150203965299179, "grad_norm": 2.4755580742958303, "learning_rate": 1.3182209889126228e-05, "loss": 0.5892, "step": 11649 }, { "epoch": 0.415056023656412, "grad_norm": 1.2640293930499593, "learning_rate": 1.318111591684288e-05, "loss": 0.4851, "step": 11650 }, { "epoch": 0.4150916507829061, "grad_norm": 1.2381472798319133, "learning_rate": 1.3180021902200749e-05, "loss": 0.5154, "step": 11651 }, { "epoch": 0.41512727790940024, "grad_norm": 1.6693808630592728, "learning_rate": 1.3178927845214402e-05, "loss": 0.5656, "step": 11652 }, { "epoch": 0.4151629050358943, "grad_norm": 1.2141490193623607, "learning_rate": 1.3177833745898409e-05, "loss": 0.4271, "step": 11653 }, { "epoch": 0.41519853216238845, "grad_norm": 1.1324020018755725, "learning_rate": 1.3176739604267344e-05, "loss": 0.4314, "step": 11654 }, { "epoch": 0.4152341592888826, "grad_norm": 1.5376446591588744, "learning_rate": 1.3175645420335768e-05, "loss": 0.5396, "step": 11655 }, { "epoch": 0.41526978641537665, "grad_norm": 1.1449623554219368, "learning_rate": 1.317455119411825e-05, "loss": 0.4574, "step": 11656 }, { "epoch": 0.4153054135418708, "grad_norm": 2.066453781300612, "learning_rate": 1.3173456925629368e-05, "loss": 0.4361, "step": 11657 }, { "epoch": 0.4153410406683649, "grad_norm": 2.0385336346509604, "learning_rate": 1.3172362614883687e-05, "loss": 0.5877, "step": 11658 }, { "epoch": 0.415376667794859, "grad_norm": 1.6208397255460714, "learning_rate": 1.317126826189578e-05, "loss": 0.6526, "step": 11659 }, { "epoch": 0.4154122949213531, "grad_norm": 1.5600297970979649, "learning_rate": 1.317017386668022e-05, "loss": 0.6393, "step": 11660 }, { "epoch": 0.41544792204784725, "grad_norm": 1.3747765086016284, "learning_rate": 1.3169079429251581e-05, "loss": 0.5828, "step": 11661 }, { "epoch": 0.4154835491743413, "grad_norm": 1.6260289002509314, "learning_rate": 1.3167984949624436e-05, "loss": 0.5053, "step": 11662 }, { "epoch": 0.41551917630083546, "grad_norm": 1.282162296627271, "learning_rate": 1.3166890427813348e-05, "loss": 0.5709, "step": 11663 }, { "epoch": 0.4155548034273296, "grad_norm": 1.1186777581817953, "learning_rate": 1.3165795863832909e-05, "loss": 0.4413, "step": 11664 }, { "epoch": 0.41559043055382366, "grad_norm": 0.9710025952204406, "learning_rate": 1.3164701257697683e-05, "loss": 0.4415, "step": 11665 }, { "epoch": 0.4156260576803178, "grad_norm": 1.9975583877504877, "learning_rate": 1.3163606609422249e-05, "loss": 0.533, "step": 11666 }, { "epoch": 0.4156616848068119, "grad_norm": 1.0981030140420744, "learning_rate": 1.316251191902118e-05, "loss": 0.4733, "step": 11667 }, { "epoch": 0.415697311933306, "grad_norm": 1.7456715548266222, "learning_rate": 1.3161417186509053e-05, "loss": 0.5544, "step": 11668 }, { "epoch": 0.41573293905980013, "grad_norm": 1.08831879522847, "learning_rate": 1.316032241190045e-05, "loss": 0.4341, "step": 11669 }, { "epoch": 0.41576856618629426, "grad_norm": 1.1109480586620055, "learning_rate": 1.3159227595209942e-05, "loss": 0.4969, "step": 11670 }, { "epoch": 0.41580419331278834, "grad_norm": 1.556580360736394, "learning_rate": 1.3158132736452111e-05, "loss": 0.6461, "step": 11671 }, { "epoch": 0.41583982043928247, "grad_norm": 2.019748414126232, "learning_rate": 1.3157037835641537e-05, "loss": 0.6069, "step": 11672 }, { "epoch": 0.4158754475657766, "grad_norm": 5.4740249882436345, "learning_rate": 1.3155942892792796e-05, "loss": 0.4814, "step": 11673 }, { "epoch": 0.4159110746922707, "grad_norm": 1.1471663810383361, "learning_rate": 1.3154847907920472e-05, "loss": 0.5412, "step": 11674 }, { "epoch": 0.4159467018187648, "grad_norm": 1.2120407625803977, "learning_rate": 1.315375288103914e-05, "loss": 0.4999, "step": 11675 }, { "epoch": 0.41598232894525894, "grad_norm": 1.4312466990552704, "learning_rate": 1.3152657812163385e-05, "loss": 0.6557, "step": 11676 }, { "epoch": 0.416017956071753, "grad_norm": 1.0455965303441719, "learning_rate": 1.3151562701307787e-05, "loss": 0.4453, "step": 11677 }, { "epoch": 0.41605358319824715, "grad_norm": 1.962884646656034, "learning_rate": 1.3150467548486929e-05, "loss": 0.6503, "step": 11678 }, { "epoch": 0.4160892103247413, "grad_norm": 1.5077226682800826, "learning_rate": 1.3149372353715395e-05, "loss": 0.5763, "step": 11679 }, { "epoch": 0.41612483745123535, "grad_norm": 1.7626299327698094, "learning_rate": 1.3148277117007765e-05, "loss": 0.6084, "step": 11680 }, { "epoch": 0.4161604645777295, "grad_norm": 1.1221097612853081, "learning_rate": 1.3147181838378628e-05, "loss": 0.4686, "step": 11681 }, { "epoch": 0.4161960917042236, "grad_norm": 1.5042020228577322, "learning_rate": 1.314608651784256e-05, "loss": 0.5271, "step": 11682 }, { "epoch": 0.4162317188307177, "grad_norm": 1.1625259057877204, "learning_rate": 1.3144991155414155e-05, "loss": 0.4589, "step": 11683 }, { "epoch": 0.4162673459572118, "grad_norm": 1.5698019503472704, "learning_rate": 1.3143895751107993e-05, "loss": 0.5352, "step": 11684 }, { "epoch": 0.41630297308370595, "grad_norm": 1.8739662154611023, "learning_rate": 1.3142800304938662e-05, "loss": 0.5664, "step": 11685 }, { "epoch": 0.4163386002102, "grad_norm": 0.9977540131730976, "learning_rate": 1.3141704816920746e-05, "loss": 0.4974, "step": 11686 }, { "epoch": 0.41637422733669416, "grad_norm": 1.3385040321721322, "learning_rate": 1.3140609287068838e-05, "loss": 0.5747, "step": 11687 }, { "epoch": 0.4164098544631883, "grad_norm": 1.1438740568932604, "learning_rate": 1.3139513715397521e-05, "loss": 0.5063, "step": 11688 }, { "epoch": 0.41644548158968236, "grad_norm": 1.437093273364869, "learning_rate": 1.3138418101921384e-05, "loss": 0.5752, "step": 11689 }, { "epoch": 0.4164811087161765, "grad_norm": 1.444562088495773, "learning_rate": 1.3137322446655017e-05, "loss": 0.5537, "step": 11690 }, { "epoch": 0.4165167358426706, "grad_norm": 1.0686685082556584, "learning_rate": 1.3136226749613009e-05, "loss": 0.4718, "step": 11691 }, { "epoch": 0.4165523629691647, "grad_norm": 1.6541494225160833, "learning_rate": 1.313513101080995e-05, "loss": 0.4522, "step": 11692 }, { "epoch": 0.41658799009565883, "grad_norm": 1.2753743646877633, "learning_rate": 1.3134035230260427e-05, "loss": 0.5043, "step": 11693 }, { "epoch": 0.41662361722215296, "grad_norm": 1.6538619998665478, "learning_rate": 1.3132939407979038e-05, "loss": 0.573, "step": 11694 }, { "epoch": 0.41665924434864704, "grad_norm": 1.1117075450788196, "learning_rate": 1.3131843543980369e-05, "loss": 0.4757, "step": 11695 }, { "epoch": 0.41669487147514117, "grad_norm": 1.4854174323746445, "learning_rate": 1.3130747638279016e-05, "loss": 0.5948, "step": 11696 }, { "epoch": 0.4167304986016353, "grad_norm": 1.604837055327169, "learning_rate": 1.3129651690889569e-05, "loss": 0.5251, "step": 11697 }, { "epoch": 0.4167661257281294, "grad_norm": 1.1772997886855845, "learning_rate": 1.312855570182662e-05, "loss": 0.4405, "step": 11698 }, { "epoch": 0.4168017528546235, "grad_norm": 1.0943614152111851, "learning_rate": 1.3127459671104767e-05, "loss": 0.397, "step": 11699 }, { "epoch": 0.41683737998111764, "grad_norm": 1.7859675991238493, "learning_rate": 1.3126363598738602e-05, "loss": 0.5606, "step": 11700 }, { "epoch": 0.4168730071076117, "grad_norm": 1.4532111932127285, "learning_rate": 1.3125267484742721e-05, "loss": 0.6065, "step": 11701 }, { "epoch": 0.41690863423410585, "grad_norm": 1.2315972870755634, "learning_rate": 1.312417132913172e-05, "loss": 0.5205, "step": 11702 }, { "epoch": 0.4169442613606, "grad_norm": 1.0792772014113092, "learning_rate": 1.3123075131920196e-05, "loss": 0.4442, "step": 11703 }, { "epoch": 0.41697988848709405, "grad_norm": 1.1525677757271473, "learning_rate": 1.312197889312274e-05, "loss": 0.4222, "step": 11704 }, { "epoch": 0.4170155156135882, "grad_norm": 1.098937707671529, "learning_rate": 1.3120882612753953e-05, "loss": 0.4716, "step": 11705 }, { "epoch": 0.4170511427400823, "grad_norm": 1.85553643157892, "learning_rate": 1.3119786290828437e-05, "loss": 0.5624, "step": 11706 }, { "epoch": 0.4170867698665764, "grad_norm": 1.3480525520643822, "learning_rate": 1.311868992736078e-05, "loss": 0.5644, "step": 11707 }, { "epoch": 0.4171223969930705, "grad_norm": 1.0942883502063787, "learning_rate": 1.3117593522365588e-05, "loss": 0.4775, "step": 11708 }, { "epoch": 0.41715802411956465, "grad_norm": 1.142418055614935, "learning_rate": 1.311649707585746e-05, "loss": 0.4543, "step": 11709 }, { "epoch": 0.4171936512460587, "grad_norm": 1.0272984614166603, "learning_rate": 1.3115400587850996e-05, "loss": 0.4442, "step": 11710 }, { "epoch": 0.41722927837255286, "grad_norm": 1.1011531596723396, "learning_rate": 1.3114304058360795e-05, "loss": 0.4327, "step": 11711 }, { "epoch": 0.417264905499047, "grad_norm": 1.166392610727228, "learning_rate": 1.3113207487401458e-05, "loss": 0.4759, "step": 11712 }, { "epoch": 0.41730053262554107, "grad_norm": 1.0473775855846141, "learning_rate": 1.311211087498759e-05, "loss": 0.4765, "step": 11713 }, { "epoch": 0.4173361597520352, "grad_norm": 1.0973952230773336, "learning_rate": 1.3111014221133789e-05, "loss": 0.4833, "step": 11714 }, { "epoch": 0.4173717868785293, "grad_norm": 1.0800317208519528, "learning_rate": 1.3109917525854657e-05, "loss": 0.4924, "step": 11715 }, { "epoch": 0.4174074140050234, "grad_norm": 1.0114245843609018, "learning_rate": 1.3108820789164802e-05, "loss": 0.4503, "step": 11716 }, { "epoch": 0.41744304113151753, "grad_norm": 1.3888166911342283, "learning_rate": 1.3107724011078826e-05, "loss": 0.5141, "step": 11717 }, { "epoch": 0.41747866825801166, "grad_norm": 0.992572245553194, "learning_rate": 1.3106627191611333e-05, "loss": 0.4292, "step": 11718 }, { "epoch": 0.41751429538450574, "grad_norm": 1.6640527791706359, "learning_rate": 1.3105530330776924e-05, "loss": 0.6291, "step": 11719 }, { "epoch": 0.41754992251099987, "grad_norm": 1.3516315216788948, "learning_rate": 1.3104433428590209e-05, "loss": 0.5874, "step": 11720 }, { "epoch": 0.417585549637494, "grad_norm": 1.078449562556243, "learning_rate": 1.3103336485065791e-05, "loss": 0.4551, "step": 11721 }, { "epoch": 0.4176211767639881, "grad_norm": 1.5578309578368255, "learning_rate": 1.3102239500218283e-05, "loss": 0.6227, "step": 11722 }, { "epoch": 0.4176568038904822, "grad_norm": 1.079311385363897, "learning_rate": 1.3101142474062284e-05, "loss": 0.4868, "step": 11723 }, { "epoch": 0.41769243101697634, "grad_norm": 1.1383095421325657, "learning_rate": 1.310004540661241e-05, "loss": 0.5399, "step": 11724 }, { "epoch": 0.4177280581434704, "grad_norm": 1.4042360970845513, "learning_rate": 1.3098948297883263e-05, "loss": 0.5705, "step": 11725 }, { "epoch": 0.41776368526996455, "grad_norm": 1.0107479875497378, "learning_rate": 1.3097851147889452e-05, "loss": 0.4595, "step": 11726 }, { "epoch": 0.4177993123964587, "grad_norm": 1.8641949835361937, "learning_rate": 1.3096753956645586e-05, "loss": 0.6151, "step": 11727 }, { "epoch": 0.41783493952295275, "grad_norm": 1.0861803823915877, "learning_rate": 1.3095656724166282e-05, "loss": 0.4317, "step": 11728 }, { "epoch": 0.4178705666494469, "grad_norm": 1.3584616920238242, "learning_rate": 1.3094559450466143e-05, "loss": 0.5501, "step": 11729 }, { "epoch": 0.417906193775941, "grad_norm": 1.0510485054696916, "learning_rate": 1.3093462135559778e-05, "loss": 0.4241, "step": 11730 }, { "epoch": 0.4179418209024351, "grad_norm": 1.0699449948626956, "learning_rate": 1.3092364779461805e-05, "loss": 0.4688, "step": 11731 }, { "epoch": 0.4179774480289292, "grad_norm": 1.0534514262409842, "learning_rate": 1.3091267382186832e-05, "loss": 0.486, "step": 11732 }, { "epoch": 0.41801307515542335, "grad_norm": 1.8499144947897308, "learning_rate": 1.3090169943749475e-05, "loss": 0.5604, "step": 11733 }, { "epoch": 0.41804870228191743, "grad_norm": 1.570659601853163, "learning_rate": 1.3089072464164345e-05, "loss": 0.5548, "step": 11734 }, { "epoch": 0.41808432940841156, "grad_norm": 1.3748881954168328, "learning_rate": 1.3087974943446055e-05, "loss": 0.5338, "step": 11735 }, { "epoch": 0.4181199565349057, "grad_norm": 1.663088353069458, "learning_rate": 1.3086877381609223e-05, "loss": 0.653, "step": 11736 }, { "epoch": 0.41815558366139977, "grad_norm": 1.8583481054234543, "learning_rate": 1.308577977866846e-05, "loss": 0.5708, "step": 11737 }, { "epoch": 0.4181912107878939, "grad_norm": 1.7897608221600434, "learning_rate": 1.308468213463838e-05, "loss": 0.5468, "step": 11738 }, { "epoch": 0.418226837914388, "grad_norm": 1.0304847082802018, "learning_rate": 1.3083584449533605e-05, "loss": 0.4499, "step": 11739 }, { "epoch": 0.4182624650408821, "grad_norm": 1.1389982643714198, "learning_rate": 1.3082486723368746e-05, "loss": 0.4892, "step": 11740 }, { "epoch": 0.41829809216737623, "grad_norm": 1.706355472636372, "learning_rate": 1.3081388956158421e-05, "loss": 0.5764, "step": 11741 }, { "epoch": 0.41833371929387037, "grad_norm": 1.1853327730738543, "learning_rate": 1.308029114791725e-05, "loss": 0.5604, "step": 11742 }, { "epoch": 0.41836934642036444, "grad_norm": 1.041833816717955, "learning_rate": 1.3079193298659849e-05, "loss": 0.44, "step": 11743 }, { "epoch": 0.41840497354685857, "grad_norm": 1.8211856862434939, "learning_rate": 1.3078095408400834e-05, "loss": 0.5519, "step": 11744 }, { "epoch": 0.4184406006733527, "grad_norm": 1.0853934542396169, "learning_rate": 1.3076997477154829e-05, "loss": 0.4125, "step": 11745 }, { "epoch": 0.4184762277998468, "grad_norm": 1.7021364219380113, "learning_rate": 1.3075899504936454e-05, "loss": 0.5584, "step": 11746 }, { "epoch": 0.4185118549263409, "grad_norm": 1.0729359505146487, "learning_rate": 1.3074801491760326e-05, "loss": 0.4625, "step": 11747 }, { "epoch": 0.41854748205283504, "grad_norm": 1.0937876563444815, "learning_rate": 1.3073703437641066e-05, "loss": 0.4838, "step": 11748 }, { "epoch": 0.4185831091793291, "grad_norm": 1.9689821297612269, "learning_rate": 1.3072605342593297e-05, "loss": 0.595, "step": 11749 }, { "epoch": 0.41861873630582325, "grad_norm": 1.1711249794170147, "learning_rate": 1.307150720663164e-05, "loss": 0.4543, "step": 11750 }, { "epoch": 0.4186543634323174, "grad_norm": 1.1113575699282296, "learning_rate": 1.3070409029770717e-05, "loss": 0.5281, "step": 11751 }, { "epoch": 0.41868999055881145, "grad_norm": 2.8775887586687223, "learning_rate": 1.3069310812025151e-05, "loss": 0.5225, "step": 11752 }, { "epoch": 0.4187256176853056, "grad_norm": 1.3981915025293097, "learning_rate": 1.306821255340957e-05, "loss": 0.5148, "step": 11753 }, { "epoch": 0.4187612448117997, "grad_norm": 1.497748880286993, "learning_rate": 1.3067114253938593e-05, "loss": 0.5498, "step": 11754 }, { "epoch": 0.4187968719382938, "grad_norm": 1.1162161498104033, "learning_rate": 1.3066015913626841e-05, "loss": 0.4523, "step": 11755 }, { "epoch": 0.4188324990647879, "grad_norm": 1.066153470038417, "learning_rate": 1.306491753248895e-05, "loss": 0.4854, "step": 11756 }, { "epoch": 0.41886812619128205, "grad_norm": 1.794879619793809, "learning_rate": 1.3063819110539537e-05, "loss": 0.5388, "step": 11757 }, { "epoch": 0.41890375331777613, "grad_norm": 1.1743353801850114, "learning_rate": 1.3062720647793231e-05, "loss": 0.4674, "step": 11758 }, { "epoch": 0.41893938044427026, "grad_norm": 1.0891726133546389, "learning_rate": 1.306162214426466e-05, "loss": 0.477, "step": 11759 }, { "epoch": 0.4189750075707644, "grad_norm": 1.1973310568119693, "learning_rate": 1.306052359996845e-05, "loss": 0.4253, "step": 11760 }, { "epoch": 0.41901063469725847, "grad_norm": 1.1320698186176277, "learning_rate": 1.3059425014919229e-05, "loss": 0.454, "step": 11761 }, { "epoch": 0.4190462618237526, "grad_norm": 1.9472263597018649, "learning_rate": 1.3058326389131626e-05, "loss": 0.5591, "step": 11762 }, { "epoch": 0.41908188895024673, "grad_norm": 1.0353463494309212, "learning_rate": 1.3057227722620272e-05, "loss": 0.4237, "step": 11763 }, { "epoch": 0.4191175160767408, "grad_norm": 1.5286768537074045, "learning_rate": 1.3056129015399787e-05, "loss": 0.5454, "step": 11764 }, { "epoch": 0.41915314320323493, "grad_norm": 1.0874138243671636, "learning_rate": 1.3055030267484814e-05, "loss": 0.4709, "step": 11765 }, { "epoch": 0.41918877032972907, "grad_norm": 1.1951688860917786, "learning_rate": 1.3053931478889976e-05, "loss": 0.3929, "step": 11766 }, { "epoch": 0.41922439745622314, "grad_norm": 1.1836469856256788, "learning_rate": 1.3052832649629905e-05, "loss": 0.4811, "step": 11767 }, { "epoch": 0.4192600245827173, "grad_norm": 1.6594499999729433, "learning_rate": 1.3051733779719233e-05, "loss": 0.5706, "step": 11768 }, { "epoch": 0.4192956517092114, "grad_norm": 1.0953692384102804, "learning_rate": 1.3050634869172594e-05, "loss": 0.4578, "step": 11769 }, { "epoch": 0.4193312788357055, "grad_norm": 1.0795095807907242, "learning_rate": 1.3049535918004618e-05, "loss": 0.4602, "step": 11770 }, { "epoch": 0.4193669059621996, "grad_norm": 1.155527593360017, "learning_rate": 1.3048436926229942e-05, "loss": 0.4455, "step": 11771 }, { "epoch": 0.41940253308869374, "grad_norm": 1.2738244331258497, "learning_rate": 1.3047337893863197e-05, "loss": 0.6295, "step": 11772 }, { "epoch": 0.4194381602151878, "grad_norm": 1.0623643074845799, "learning_rate": 1.3046238820919018e-05, "loss": 0.486, "step": 11773 }, { "epoch": 0.41947378734168195, "grad_norm": 1.226170836879084, "learning_rate": 1.3045139707412041e-05, "loss": 0.465, "step": 11774 }, { "epoch": 0.4195094144681761, "grad_norm": 1.1682080149122804, "learning_rate": 1.3044040553356896e-05, "loss": 0.498, "step": 11775 }, { "epoch": 0.4195450415946702, "grad_norm": 1.229342749201222, "learning_rate": 1.3042941358768227e-05, "loss": 0.5134, "step": 11776 }, { "epoch": 0.4195806687211643, "grad_norm": 1.2085440641538356, "learning_rate": 1.3041842123660667e-05, "loss": 0.4915, "step": 11777 }, { "epoch": 0.4196162958476584, "grad_norm": 1.21690510974166, "learning_rate": 1.304074284804885e-05, "loss": 0.4718, "step": 11778 }, { "epoch": 0.41965192297415255, "grad_norm": 1.6847719060132986, "learning_rate": 1.3039643531947417e-05, "loss": 0.5971, "step": 11779 }, { "epoch": 0.4196875501006466, "grad_norm": 2.399789278740891, "learning_rate": 1.3038544175371009e-05, "loss": 0.5416, "step": 11780 }, { "epoch": 0.41972317722714075, "grad_norm": 2.062520324953589, "learning_rate": 1.303744477833426e-05, "loss": 0.5815, "step": 11781 }, { "epoch": 0.4197588043536349, "grad_norm": 1.6020295350052371, "learning_rate": 1.3036345340851813e-05, "loss": 0.5554, "step": 11782 }, { "epoch": 0.41979443148012896, "grad_norm": 1.300900131289661, "learning_rate": 1.30352458629383e-05, "loss": 0.5063, "step": 11783 }, { "epoch": 0.4198300586066231, "grad_norm": 1.229654219616885, "learning_rate": 1.3034146344608372e-05, "loss": 0.4683, "step": 11784 }, { "epoch": 0.4198656857331172, "grad_norm": 1.479331897762778, "learning_rate": 1.3033046785876665e-05, "loss": 0.5049, "step": 11785 }, { "epoch": 0.4199013128596113, "grad_norm": 1.589712393537665, "learning_rate": 1.3031947186757814e-05, "loss": 0.5685, "step": 11786 }, { "epoch": 0.41993693998610543, "grad_norm": 2.2319945030183885, "learning_rate": 1.3030847547266472e-05, "loss": 0.661, "step": 11787 }, { "epoch": 0.41997256711259956, "grad_norm": 1.1519006215961258, "learning_rate": 1.3029747867417275e-05, "loss": 0.4961, "step": 11788 }, { "epoch": 0.42000819423909364, "grad_norm": 1.1648927088539422, "learning_rate": 1.3028648147224867e-05, "loss": 0.4956, "step": 11789 }, { "epoch": 0.42004382136558777, "grad_norm": 1.1655048167654687, "learning_rate": 1.302754838670389e-05, "loss": 0.4752, "step": 11790 }, { "epoch": 0.4200794484920819, "grad_norm": 1.2692451570437273, "learning_rate": 1.3026448585868994e-05, "loss": 0.4211, "step": 11791 }, { "epoch": 0.420115075618576, "grad_norm": 1.1032691171307345, "learning_rate": 1.302534874473482e-05, "loss": 0.4485, "step": 11792 }, { "epoch": 0.4201507027450701, "grad_norm": 1.0785157774367937, "learning_rate": 1.3024248863316012e-05, "loss": 0.5003, "step": 11793 }, { "epoch": 0.42018632987156423, "grad_norm": 1.0826688439062362, "learning_rate": 1.3023148941627215e-05, "loss": 0.5155, "step": 11794 }, { "epoch": 0.4202219569980583, "grad_norm": 1.5531340721855158, "learning_rate": 1.302204897968308e-05, "loss": 0.4434, "step": 11795 }, { "epoch": 0.42025758412455244, "grad_norm": 1.3335802673162402, "learning_rate": 1.3020948977498246e-05, "loss": 0.5148, "step": 11796 }, { "epoch": 0.4202932112510466, "grad_norm": 1.0329127916398648, "learning_rate": 1.3019848935087367e-05, "loss": 0.4524, "step": 11797 }, { "epoch": 0.42032883837754065, "grad_norm": 1.6403013702874272, "learning_rate": 1.3018748852465089e-05, "loss": 0.5431, "step": 11798 }, { "epoch": 0.4203644655040348, "grad_norm": 1.6866443162709512, "learning_rate": 1.3017648729646061e-05, "loss": 0.5875, "step": 11799 }, { "epoch": 0.4204000926305289, "grad_norm": 1.3689409168398423, "learning_rate": 1.3016548566644931e-05, "loss": 0.5134, "step": 11800 }, { "epoch": 0.420435719757023, "grad_norm": 1.255765609782534, "learning_rate": 1.3015448363476346e-05, "loss": 0.5062, "step": 11801 }, { "epoch": 0.4204713468835171, "grad_norm": 1.0955955758232072, "learning_rate": 1.3014348120154959e-05, "loss": 0.4337, "step": 11802 }, { "epoch": 0.42050697401001125, "grad_norm": 1.112843357495596, "learning_rate": 1.301324783669542e-05, "loss": 0.4585, "step": 11803 }, { "epoch": 0.4205426011365053, "grad_norm": 1.706730721938087, "learning_rate": 1.301214751311238e-05, "loss": 0.6128, "step": 11804 }, { "epoch": 0.42057822826299945, "grad_norm": 1.8109492511265917, "learning_rate": 1.3011047149420491e-05, "loss": 0.5669, "step": 11805 }, { "epoch": 0.4206138553894936, "grad_norm": 1.18439219679266, "learning_rate": 1.3009946745634406e-05, "loss": 0.5017, "step": 11806 }, { "epoch": 0.42064948251598766, "grad_norm": 1.2531256394720567, "learning_rate": 1.3008846301768773e-05, "loss": 0.4613, "step": 11807 }, { "epoch": 0.4206851096424818, "grad_norm": 1.5527230841250959, "learning_rate": 1.3007745817838251e-05, "loss": 0.5626, "step": 11808 }, { "epoch": 0.4207207367689759, "grad_norm": 1.1177990365649502, "learning_rate": 1.300664529385749e-05, "loss": 0.4504, "step": 11809 }, { "epoch": 0.42075636389547, "grad_norm": 1.2837726858831335, "learning_rate": 1.3005544729841146e-05, "loss": 0.4843, "step": 11810 }, { "epoch": 0.42079199102196413, "grad_norm": 1.2225090265633918, "learning_rate": 1.3004444125803875e-05, "loss": 0.4507, "step": 11811 }, { "epoch": 0.42082761814845826, "grad_norm": 0.9771107974690807, "learning_rate": 1.3003343481760327e-05, "loss": 0.428, "step": 11812 }, { "epoch": 0.42086324527495234, "grad_norm": 1.8511154802169478, "learning_rate": 1.300224279772516e-05, "loss": 0.4814, "step": 11813 }, { "epoch": 0.42089887240144647, "grad_norm": 1.1153006873589726, "learning_rate": 1.3001142073713035e-05, "loss": 0.452, "step": 11814 }, { "epoch": 0.4209344995279406, "grad_norm": 1.748404424755553, "learning_rate": 1.3000041309738607e-05, "loss": 0.6085, "step": 11815 }, { "epoch": 0.4209701266544347, "grad_norm": 1.7746083928176433, "learning_rate": 1.2998940505816529e-05, "loss": 0.5781, "step": 11816 }, { "epoch": 0.4210057537809288, "grad_norm": 1.1704729237419962, "learning_rate": 1.2997839661961464e-05, "loss": 0.4561, "step": 11817 }, { "epoch": 0.42104138090742294, "grad_norm": 2.43255453189925, "learning_rate": 1.2996738778188067e-05, "loss": 0.5783, "step": 11818 }, { "epoch": 0.421077008033917, "grad_norm": 1.640413742597179, "learning_rate": 1.2995637854511e-05, "loss": 0.5932, "step": 11819 }, { "epoch": 0.42111263516041114, "grad_norm": 1.29434043827005, "learning_rate": 1.299453689094492e-05, "loss": 0.4534, "step": 11820 }, { "epoch": 0.4211482622869053, "grad_norm": 1.234688948781679, "learning_rate": 1.2993435887504489e-05, "loss": 0.5057, "step": 11821 }, { "epoch": 0.42118388941339935, "grad_norm": 1.250940955025396, "learning_rate": 1.2992334844204368e-05, "loss": 0.4979, "step": 11822 }, { "epoch": 0.4212195165398935, "grad_norm": 1.2434532064312414, "learning_rate": 1.2991233761059214e-05, "loss": 0.5257, "step": 11823 }, { "epoch": 0.4212551436663876, "grad_norm": 1.313661360691085, "learning_rate": 1.2990132638083694e-05, "loss": 0.5025, "step": 11824 }, { "epoch": 0.4212907707928817, "grad_norm": 1.83252971844685, "learning_rate": 1.2989031475292464e-05, "loss": 0.5844, "step": 11825 }, { "epoch": 0.4213263979193758, "grad_norm": 1.1032484885925846, "learning_rate": 1.2987930272700196e-05, "loss": 0.4666, "step": 11826 }, { "epoch": 0.42136202504586995, "grad_norm": 1.412386701814896, "learning_rate": 1.2986829030321543e-05, "loss": 0.5489, "step": 11827 }, { "epoch": 0.421397652172364, "grad_norm": 1.4135302447613083, "learning_rate": 1.2985727748171177e-05, "loss": 0.5414, "step": 11828 }, { "epoch": 0.42143327929885815, "grad_norm": 1.470574717917401, "learning_rate": 1.298462642626376e-05, "loss": 0.5544, "step": 11829 }, { "epoch": 0.4214689064253523, "grad_norm": 1.7071763161786777, "learning_rate": 1.2983525064613955e-05, "loss": 0.5553, "step": 11830 }, { "epoch": 0.42150453355184636, "grad_norm": 1.0709556341537223, "learning_rate": 1.2982423663236428e-05, "loss": 0.4296, "step": 11831 }, { "epoch": 0.4215401606783405, "grad_norm": 1.2638950840988672, "learning_rate": 1.2981322222145846e-05, "loss": 0.5205, "step": 11832 }, { "epoch": 0.4215757878048346, "grad_norm": 1.9764133685857763, "learning_rate": 1.2980220741356874e-05, "loss": 0.5672, "step": 11833 }, { "epoch": 0.4216114149313287, "grad_norm": 1.1811784495808952, "learning_rate": 1.297911922088418e-05, "loss": 0.4671, "step": 11834 }, { "epoch": 0.42164704205782283, "grad_norm": 1.1852573015625014, "learning_rate": 1.2978017660742433e-05, "loss": 0.4601, "step": 11835 }, { "epoch": 0.42168266918431696, "grad_norm": 1.4433248519377675, "learning_rate": 1.2976916060946295e-05, "loss": 0.4299, "step": 11836 }, { "epoch": 0.42171829631081104, "grad_norm": 1.5283636767552595, "learning_rate": 1.2975814421510444e-05, "loss": 0.5342, "step": 11837 }, { "epoch": 0.42175392343730517, "grad_norm": 1.108027923419829, "learning_rate": 1.2974712742449543e-05, "loss": 0.4538, "step": 11838 }, { "epoch": 0.4217895505637993, "grad_norm": 1.7451535678503065, "learning_rate": 1.2973611023778259e-05, "loss": 0.5678, "step": 11839 }, { "epoch": 0.4218251776902934, "grad_norm": 1.1268408168743058, "learning_rate": 1.2972509265511269e-05, "loss": 0.4575, "step": 11840 }, { "epoch": 0.4218608048167875, "grad_norm": 1.1958784106087488, "learning_rate": 1.297140746766324e-05, "loss": 0.5183, "step": 11841 }, { "epoch": 0.42189643194328164, "grad_norm": 1.2413783192831578, "learning_rate": 1.2970305630248841e-05, "loss": 0.4747, "step": 11842 }, { "epoch": 0.4219320590697757, "grad_norm": 1.5627067899079339, "learning_rate": 1.296920375328275e-05, "loss": 0.6184, "step": 11843 }, { "epoch": 0.42196768619626984, "grad_norm": 2.2347952311168857, "learning_rate": 1.2968101836779634e-05, "loss": 0.6326, "step": 11844 }, { "epoch": 0.422003313322764, "grad_norm": 1.7525140595579527, "learning_rate": 1.296699988075417e-05, "loss": 0.6923, "step": 11845 }, { "epoch": 0.42203894044925805, "grad_norm": 1.2697948705671103, "learning_rate": 1.2965897885221024e-05, "loss": 0.4022, "step": 11846 }, { "epoch": 0.4220745675757522, "grad_norm": 1.2355904140975689, "learning_rate": 1.2964795850194878e-05, "loss": 0.4958, "step": 11847 }, { "epoch": 0.4221101947022463, "grad_norm": 1.2221749231036472, "learning_rate": 1.2963693775690403e-05, "loss": 0.4777, "step": 11848 }, { "epoch": 0.4221458218287404, "grad_norm": 1.8245599444033558, "learning_rate": 1.2962591661722273e-05, "loss": 0.5718, "step": 11849 }, { "epoch": 0.4221814489552345, "grad_norm": 1.0940939265674854, "learning_rate": 1.2961489508305162e-05, "loss": 0.4583, "step": 11850 }, { "epoch": 0.42221707608172865, "grad_norm": 1.6012293300493499, "learning_rate": 1.2960387315453753e-05, "loss": 0.6318, "step": 11851 }, { "epoch": 0.4222527032082227, "grad_norm": 1.3984010865105299, "learning_rate": 1.2959285083182715e-05, "loss": 0.5631, "step": 11852 }, { "epoch": 0.42228833033471685, "grad_norm": 1.222954987969594, "learning_rate": 1.2958182811506728e-05, "loss": 0.468, "step": 11853 }, { "epoch": 0.422323957461211, "grad_norm": 1.5206229522435817, "learning_rate": 1.2957080500440469e-05, "loss": 0.6223, "step": 11854 }, { "epoch": 0.42235958458770506, "grad_norm": 1.6424247930585323, "learning_rate": 1.2955978149998616e-05, "loss": 0.5321, "step": 11855 }, { "epoch": 0.4223952117141992, "grad_norm": 1.248352789657504, "learning_rate": 1.2954875760195851e-05, "loss": 0.4634, "step": 11856 }, { "epoch": 0.4224308388406933, "grad_norm": 1.1518963157106334, "learning_rate": 1.2953773331046845e-05, "loss": 0.5096, "step": 11857 }, { "epoch": 0.4224664659671874, "grad_norm": 1.2278287492035787, "learning_rate": 1.2952670862566285e-05, "loss": 0.5029, "step": 11858 }, { "epoch": 0.42250209309368153, "grad_norm": 1.556207334333397, "learning_rate": 1.2951568354768846e-05, "loss": 0.5681, "step": 11859 }, { "epoch": 0.42253772022017566, "grad_norm": 1.2615629310116754, "learning_rate": 1.2950465807669215e-05, "loss": 0.5032, "step": 11860 }, { "epoch": 0.42257334734666974, "grad_norm": 1.1709785751007928, "learning_rate": 1.2949363221282067e-05, "loss": 0.5012, "step": 11861 }, { "epoch": 0.42260897447316387, "grad_norm": 1.711070903956291, "learning_rate": 1.2948260595622088e-05, "loss": 0.4873, "step": 11862 }, { "epoch": 0.422644601599658, "grad_norm": 1.2636206159455148, "learning_rate": 1.2947157930703959e-05, "loss": 0.5093, "step": 11863 }, { "epoch": 0.4226802287261521, "grad_norm": 1.1329184137914292, "learning_rate": 1.2946055226542362e-05, "loss": 0.4826, "step": 11864 }, { "epoch": 0.4227158558526462, "grad_norm": 1.1143591877751338, "learning_rate": 1.2944952483151978e-05, "loss": 0.46, "step": 11865 }, { "epoch": 0.42275148297914034, "grad_norm": 1.1027180281624909, "learning_rate": 1.2943849700547495e-05, "loss": 0.4395, "step": 11866 }, { "epoch": 0.4227871101056344, "grad_norm": 2.191432166897596, "learning_rate": 1.2942746878743597e-05, "loss": 0.5361, "step": 11867 }, { "epoch": 0.42282273723212854, "grad_norm": 1.3517262540480073, "learning_rate": 1.2941644017754964e-05, "loss": 0.5939, "step": 11868 }, { "epoch": 0.4228583643586227, "grad_norm": 1.107384830382386, "learning_rate": 1.2940541117596287e-05, "loss": 0.4672, "step": 11869 }, { "epoch": 0.42289399148511675, "grad_norm": 1.6453382734264914, "learning_rate": 1.2939438178282254e-05, "loss": 0.5626, "step": 11870 }, { "epoch": 0.4229296186116109, "grad_norm": 1.505603714242384, "learning_rate": 1.2938335199827539e-05, "loss": 0.4958, "step": 11871 }, { "epoch": 0.422965245738105, "grad_norm": 1.7033425212699904, "learning_rate": 1.293723218224684e-05, "loss": 0.501, "step": 11872 }, { "epoch": 0.4230008728645991, "grad_norm": 1.549305383624147, "learning_rate": 1.2936129125554842e-05, "loss": 0.5541, "step": 11873 }, { "epoch": 0.4230364999910932, "grad_norm": 1.577692250013106, "learning_rate": 1.2935026029766234e-05, "loss": 0.6325, "step": 11874 }, { "epoch": 0.42307212711758735, "grad_norm": 1.6965398451346794, "learning_rate": 1.2933922894895703e-05, "loss": 0.6045, "step": 11875 }, { "epoch": 0.4231077542440814, "grad_norm": 1.3988732517956355, "learning_rate": 1.2932819720957934e-05, "loss": 0.5374, "step": 11876 }, { "epoch": 0.42314338137057556, "grad_norm": 2.0151016206935344, "learning_rate": 1.2931716507967624e-05, "loss": 0.5372, "step": 11877 }, { "epoch": 0.4231790084970697, "grad_norm": 1.2871888320055016, "learning_rate": 1.2930613255939458e-05, "loss": 0.4692, "step": 11878 }, { "epoch": 0.42321463562356376, "grad_norm": 1.4585800124200774, "learning_rate": 1.2929509964888129e-05, "loss": 0.5137, "step": 11879 }, { "epoch": 0.4232502627500579, "grad_norm": 1.3944438011863551, "learning_rate": 1.2928406634828327e-05, "loss": 0.5522, "step": 11880 }, { "epoch": 0.423285889876552, "grad_norm": 1.49812798469135, "learning_rate": 1.2927303265774745e-05, "loss": 0.5995, "step": 11881 }, { "epoch": 0.4233215170030461, "grad_norm": 1.7327328051909308, "learning_rate": 1.292619985774207e-05, "loss": 0.5045, "step": 11882 }, { "epoch": 0.42335714412954023, "grad_norm": 1.1344516884493474, "learning_rate": 1.2925096410745002e-05, "loss": 0.4525, "step": 11883 }, { "epoch": 0.42339277125603436, "grad_norm": 1.9077684465025186, "learning_rate": 1.2923992924798229e-05, "loss": 0.4075, "step": 11884 }, { "epoch": 0.42342839838252844, "grad_norm": 1.336038486296267, "learning_rate": 1.292288939991645e-05, "loss": 0.53, "step": 11885 }, { "epoch": 0.42346402550902257, "grad_norm": 2.6134101007493244, "learning_rate": 1.2921785836114353e-05, "loss": 0.5276, "step": 11886 }, { "epoch": 0.4234996526355167, "grad_norm": 1.1771241424896044, "learning_rate": 1.2920682233406634e-05, "loss": 0.4776, "step": 11887 }, { "epoch": 0.4235352797620108, "grad_norm": 1.0555131078177735, "learning_rate": 1.2919578591807992e-05, "loss": 0.4786, "step": 11888 }, { "epoch": 0.4235709068885049, "grad_norm": 1.3128304584713764, "learning_rate": 1.2918474911333121e-05, "loss": 0.5521, "step": 11889 }, { "epoch": 0.42360653401499904, "grad_norm": 1.133471080956027, "learning_rate": 1.2917371191996714e-05, "loss": 0.4896, "step": 11890 }, { "epoch": 0.4236421611414931, "grad_norm": 1.7815371188458458, "learning_rate": 1.2916267433813472e-05, "loss": 0.5408, "step": 11891 }, { "epoch": 0.42367778826798724, "grad_norm": 1.749736942910697, "learning_rate": 1.291516363679809e-05, "loss": 0.5538, "step": 11892 }, { "epoch": 0.4237134153944814, "grad_norm": 2.400483631059764, "learning_rate": 1.2914059800965268e-05, "loss": 0.4717, "step": 11893 }, { "epoch": 0.42374904252097545, "grad_norm": 1.6402089264365332, "learning_rate": 1.2912955926329702e-05, "loss": 0.527, "step": 11894 }, { "epoch": 0.4237846696474696, "grad_norm": 1.3969452884924534, "learning_rate": 1.291185201290609e-05, "loss": 0.5984, "step": 11895 }, { "epoch": 0.4238202967739637, "grad_norm": 1.0259715489326373, "learning_rate": 1.2910748060709138e-05, "loss": 0.456, "step": 11896 }, { "epoch": 0.4238559239004578, "grad_norm": 1.0848652281727544, "learning_rate": 1.2909644069753537e-05, "loss": 0.4581, "step": 11897 }, { "epoch": 0.4238915510269519, "grad_norm": 1.1627996391907105, "learning_rate": 1.2908540040053992e-05, "loss": 0.4732, "step": 11898 }, { "epoch": 0.42392717815344605, "grad_norm": 0.9772126801539783, "learning_rate": 1.2907435971625204e-05, "loss": 0.4034, "step": 11899 }, { "epoch": 0.4239628052799401, "grad_norm": 1.3921412156701967, "learning_rate": 1.2906331864481874e-05, "loss": 0.5212, "step": 11900 }, { "epoch": 0.42399843240643426, "grad_norm": 1.603352728564454, "learning_rate": 1.2905227718638704e-05, "loss": 0.604, "step": 11901 }, { "epoch": 0.4240340595329284, "grad_norm": 1.4518617459775176, "learning_rate": 1.2904123534110396e-05, "loss": 0.5777, "step": 11902 }, { "epoch": 0.42406968665942246, "grad_norm": 1.4338527876362581, "learning_rate": 1.2903019310911653e-05, "loss": 0.5575, "step": 11903 }, { "epoch": 0.4241053137859166, "grad_norm": 1.030532149325227, "learning_rate": 1.290191504905718e-05, "loss": 0.4098, "step": 11904 }, { "epoch": 0.4241409409124107, "grad_norm": 1.1626077860173922, "learning_rate": 1.2900810748561678e-05, "loss": 0.4509, "step": 11905 }, { "epoch": 0.4241765680389048, "grad_norm": 1.54883210961863, "learning_rate": 1.2899706409439854e-05, "loss": 0.5325, "step": 11906 }, { "epoch": 0.42421219516539893, "grad_norm": 1.0607621494471766, "learning_rate": 1.2898602031706413e-05, "loss": 0.4767, "step": 11907 }, { "epoch": 0.42424782229189306, "grad_norm": 1.1694885890093223, "learning_rate": 1.289749761537606e-05, "loss": 0.4772, "step": 11908 }, { "epoch": 0.42428344941838714, "grad_norm": 1.4961295850393788, "learning_rate": 1.2896393160463499e-05, "loss": 0.5958, "step": 11909 }, { "epoch": 0.42431907654488127, "grad_norm": 1.1184901467518094, "learning_rate": 1.289528866698344e-05, "loss": 0.475, "step": 11910 }, { "epoch": 0.4243547036713754, "grad_norm": 1.2993427279982706, "learning_rate": 1.289418413495059e-05, "loss": 0.5125, "step": 11911 }, { "epoch": 0.4243903307978695, "grad_norm": 1.5853687026898866, "learning_rate": 1.2893079564379657e-05, "loss": 0.6173, "step": 11912 }, { "epoch": 0.4244259579243636, "grad_norm": 1.2210645579156576, "learning_rate": 1.289197495528534e-05, "loss": 0.4858, "step": 11913 }, { "epoch": 0.42446158505085774, "grad_norm": 1.756224165642931, "learning_rate": 1.2890870307682362e-05, "loss": 0.5816, "step": 11914 }, { "epoch": 0.4244972121773518, "grad_norm": 1.1048128823369703, "learning_rate": 1.2889765621585424e-05, "loss": 0.4342, "step": 11915 }, { "epoch": 0.42453283930384594, "grad_norm": 1.0116542696776225, "learning_rate": 1.2888660897009233e-05, "loss": 0.4939, "step": 11916 }, { "epoch": 0.4245684664303401, "grad_norm": 1.727331937443087, "learning_rate": 1.2887556133968506e-05, "loss": 0.686, "step": 11917 }, { "epoch": 0.42460409355683415, "grad_norm": 1.9292179630400084, "learning_rate": 1.2886451332477952e-05, "loss": 0.5725, "step": 11918 }, { "epoch": 0.4246397206833283, "grad_norm": 1.5918541936678468, "learning_rate": 1.288534649255228e-05, "loss": 0.5641, "step": 11919 }, { "epoch": 0.4246753478098224, "grad_norm": 1.3552898020230626, "learning_rate": 1.2884241614206204e-05, "loss": 0.4967, "step": 11920 }, { "epoch": 0.4247109749363165, "grad_norm": 1.2274614613082242, "learning_rate": 1.2883136697454432e-05, "loss": 0.4785, "step": 11921 }, { "epoch": 0.4247466020628106, "grad_norm": 1.1464277506511136, "learning_rate": 1.288203174231168e-05, "loss": 0.4737, "step": 11922 }, { "epoch": 0.42478222918930475, "grad_norm": 1.0739899911587187, "learning_rate": 1.2880926748792663e-05, "loss": 0.4606, "step": 11923 }, { "epoch": 0.4248178563157988, "grad_norm": 1.5478795753052847, "learning_rate": 1.287982171691209e-05, "loss": 0.6533, "step": 11924 }, { "epoch": 0.42485348344229296, "grad_norm": 1.8230321975222625, "learning_rate": 1.2878716646684681e-05, "loss": 0.5542, "step": 11925 }, { "epoch": 0.4248891105687871, "grad_norm": 0.9893936710554234, "learning_rate": 1.2877611538125148e-05, "loss": 0.4891, "step": 11926 }, { "epoch": 0.42492473769528116, "grad_norm": 1.7213688298344696, "learning_rate": 1.2876506391248206e-05, "loss": 0.6153, "step": 11927 }, { "epoch": 0.4249603648217753, "grad_norm": 1.7116416715895069, "learning_rate": 1.2875401206068567e-05, "loss": 0.5579, "step": 11928 }, { "epoch": 0.4249959919482694, "grad_norm": 1.0089170031705799, "learning_rate": 1.2874295982600953e-05, "loss": 0.4308, "step": 11929 }, { "epoch": 0.4250316190747635, "grad_norm": 1.1590022888162563, "learning_rate": 1.287319072086008e-05, "loss": 0.4478, "step": 11930 }, { "epoch": 0.42506724620125763, "grad_norm": 1.5848272071644964, "learning_rate": 1.2872085420860665e-05, "loss": 0.5515, "step": 11931 }, { "epoch": 0.42510287332775176, "grad_norm": 1.0292768505245211, "learning_rate": 1.2870980082617423e-05, "loss": 0.45, "step": 11932 }, { "epoch": 0.42513850045424584, "grad_norm": 1.7030356736585288, "learning_rate": 1.2869874706145075e-05, "loss": 0.5521, "step": 11933 }, { "epoch": 0.42517412758073997, "grad_norm": 1.0100961667809798, "learning_rate": 1.2868769291458341e-05, "loss": 0.4632, "step": 11934 }, { "epoch": 0.4252097547072341, "grad_norm": 1.2027231383852037, "learning_rate": 1.2867663838571938e-05, "loss": 0.5042, "step": 11935 }, { "epoch": 0.4252453818337282, "grad_norm": 1.1201112038265892, "learning_rate": 1.2866558347500586e-05, "loss": 0.4846, "step": 11936 }, { "epoch": 0.4252810089602223, "grad_norm": 1.1018075685619089, "learning_rate": 1.2865452818259007e-05, "loss": 0.4962, "step": 11937 }, { "epoch": 0.42531663608671644, "grad_norm": 0.9617602226005395, "learning_rate": 1.286434725086192e-05, "loss": 0.4168, "step": 11938 }, { "epoch": 0.4253522632132105, "grad_norm": 2.415331978254427, "learning_rate": 1.2863241645324048e-05, "loss": 0.5416, "step": 11939 }, { "epoch": 0.42538789033970464, "grad_norm": 1.771659515633278, "learning_rate": 1.2862136001660109e-05, "loss": 0.6193, "step": 11940 }, { "epoch": 0.4254235174661988, "grad_norm": 1.052104897265131, "learning_rate": 1.2861030319884832e-05, "loss": 0.4396, "step": 11941 }, { "epoch": 0.42545914459269285, "grad_norm": 1.6467514978455513, "learning_rate": 1.2859924600012937e-05, "loss": 0.6017, "step": 11942 }, { "epoch": 0.425494771719187, "grad_norm": 1.248655125152788, "learning_rate": 1.2858818842059145e-05, "loss": 0.4076, "step": 11943 }, { "epoch": 0.4255303988456811, "grad_norm": 1.182884174163826, "learning_rate": 1.2857713046038183e-05, "loss": 0.459, "step": 11944 }, { "epoch": 0.4255660259721752, "grad_norm": 1.5987492186512873, "learning_rate": 1.2856607211964773e-05, "loss": 0.5474, "step": 11945 }, { "epoch": 0.4256016530986693, "grad_norm": 1.1473598764544706, "learning_rate": 1.2855501339853644e-05, "loss": 0.4901, "step": 11946 }, { "epoch": 0.42563728022516345, "grad_norm": 1.3640890811916349, "learning_rate": 1.2854395429719514e-05, "loss": 0.4595, "step": 11947 }, { "epoch": 0.4256729073516575, "grad_norm": 1.2115195533665721, "learning_rate": 1.2853289481577119e-05, "loss": 0.501, "step": 11948 }, { "epoch": 0.42570853447815166, "grad_norm": 1.1302824482106435, "learning_rate": 1.2852183495441179e-05, "loss": 0.4441, "step": 11949 }, { "epoch": 0.4257441616046458, "grad_norm": 1.1022226692816492, "learning_rate": 1.285107747132642e-05, "loss": 0.4561, "step": 11950 }, { "epoch": 0.42577978873113986, "grad_norm": 1.181024333746256, "learning_rate": 1.2849971409247575e-05, "loss": 0.4734, "step": 11951 }, { "epoch": 0.425815415857634, "grad_norm": 2.685106355817569, "learning_rate": 1.2848865309219365e-05, "loss": 0.6155, "step": 11952 }, { "epoch": 0.4258510429841281, "grad_norm": 1.3970705468645275, "learning_rate": 1.2847759171256523e-05, "loss": 0.5479, "step": 11953 }, { "epoch": 0.4258866701106222, "grad_norm": 1.1850856155496965, "learning_rate": 1.2846652995373777e-05, "loss": 0.4795, "step": 11954 }, { "epoch": 0.42592229723711633, "grad_norm": 1.0236495891713344, "learning_rate": 1.2845546781585857e-05, "loss": 0.4591, "step": 11955 }, { "epoch": 0.42595792436361046, "grad_norm": 1.1096107595854838, "learning_rate": 1.2844440529907496e-05, "loss": 0.5084, "step": 11956 }, { "epoch": 0.42599355149010454, "grad_norm": 1.137845678914626, "learning_rate": 1.284333424035342e-05, "loss": 0.4907, "step": 11957 }, { "epoch": 0.42602917861659867, "grad_norm": 1.0721427357077824, "learning_rate": 1.284222791293836e-05, "loss": 0.4213, "step": 11958 }, { "epoch": 0.4260648057430928, "grad_norm": 1.8399908965276022, "learning_rate": 1.2841121547677047e-05, "loss": 0.5513, "step": 11959 }, { "epoch": 0.4261004328695869, "grad_norm": 2.019326950046572, "learning_rate": 1.2840015144584216e-05, "loss": 0.6628, "step": 11960 }, { "epoch": 0.426136059996081, "grad_norm": 1.125120567156753, "learning_rate": 1.28389087036746e-05, "loss": 0.4582, "step": 11961 }, { "epoch": 0.42617168712257514, "grad_norm": 1.8527276691912804, "learning_rate": 1.2837802224962928e-05, "loss": 0.704, "step": 11962 }, { "epoch": 0.42620731424906927, "grad_norm": 3.41518303773123, "learning_rate": 1.2836695708463936e-05, "loss": 0.61, "step": 11963 }, { "epoch": 0.42624294137556334, "grad_norm": 1.3389509996892037, "learning_rate": 1.2835589154192358e-05, "loss": 0.5378, "step": 11964 }, { "epoch": 0.4262785685020575, "grad_norm": 1.857586554373005, "learning_rate": 1.2834482562162929e-05, "loss": 0.556, "step": 11965 }, { "epoch": 0.4263141956285516, "grad_norm": 3.778135641415072, "learning_rate": 1.2833375932390383e-05, "loss": 0.6228, "step": 11966 }, { "epoch": 0.4263498227550457, "grad_norm": 1.1246249432932878, "learning_rate": 1.2832269264889459e-05, "loss": 0.4669, "step": 11967 }, { "epoch": 0.4263854498815398, "grad_norm": 2.271685768816196, "learning_rate": 1.2831162559674888e-05, "loss": 0.5604, "step": 11968 }, { "epoch": 0.42642107700803394, "grad_norm": 1.583548985897178, "learning_rate": 1.2830055816761409e-05, "loss": 0.6354, "step": 11969 }, { "epoch": 0.426456704134528, "grad_norm": 1.1893526596553348, "learning_rate": 1.2828949036163757e-05, "loss": 0.5117, "step": 11970 }, { "epoch": 0.42649233126102215, "grad_norm": 1.5171491392935514, "learning_rate": 1.2827842217896674e-05, "loss": 0.5658, "step": 11971 }, { "epoch": 0.4265279583875163, "grad_norm": 1.0907554238704165, "learning_rate": 1.2826735361974891e-05, "loss": 0.4715, "step": 11972 }, { "epoch": 0.42656358551401036, "grad_norm": 1.1222710682334947, "learning_rate": 1.2825628468413156e-05, "loss": 0.4647, "step": 11973 }, { "epoch": 0.4265992126405045, "grad_norm": 1.4931534040898597, "learning_rate": 1.2824521537226203e-05, "loss": 0.569, "step": 11974 }, { "epoch": 0.4266348397669986, "grad_norm": 1.447613863066803, "learning_rate": 1.2823414568428767e-05, "loss": 0.5869, "step": 11975 }, { "epoch": 0.4266704668934927, "grad_norm": 1.2434612694069116, "learning_rate": 1.2822307562035593e-05, "loss": 0.4962, "step": 11976 }, { "epoch": 0.4267060940199868, "grad_norm": 1.0208501680502042, "learning_rate": 1.2821200518061422e-05, "loss": 0.4171, "step": 11977 }, { "epoch": 0.42674172114648096, "grad_norm": 1.082043800782622, "learning_rate": 1.2820093436520998e-05, "loss": 0.4637, "step": 11978 }, { "epoch": 0.42677734827297503, "grad_norm": 1.9299258616003767, "learning_rate": 1.2818986317429057e-05, "loss": 0.5143, "step": 11979 }, { "epoch": 0.42681297539946916, "grad_norm": 1.6103800962079553, "learning_rate": 1.2817879160800337e-05, "loss": 0.5468, "step": 11980 }, { "epoch": 0.4268486025259633, "grad_norm": 1.1079422415955784, "learning_rate": 1.2816771966649593e-05, "loss": 0.4921, "step": 11981 }, { "epoch": 0.42688422965245737, "grad_norm": 1.2663888390695792, "learning_rate": 1.2815664734991559e-05, "loss": 0.5234, "step": 11982 }, { "epoch": 0.4269198567789515, "grad_norm": 2.0604579471425324, "learning_rate": 1.2814557465840984e-05, "loss": 0.5563, "step": 11983 }, { "epoch": 0.42695548390544563, "grad_norm": 1.9452155955872028, "learning_rate": 1.2813450159212604e-05, "loss": 0.5841, "step": 11984 }, { "epoch": 0.4269911110319397, "grad_norm": 2.063413923561011, "learning_rate": 1.2812342815121173e-05, "loss": 0.5384, "step": 11985 }, { "epoch": 0.42702673815843384, "grad_norm": 2.2450451085454226, "learning_rate": 1.2811235433581428e-05, "loss": 0.5869, "step": 11986 }, { "epoch": 0.42706236528492797, "grad_norm": 1.0787776361495478, "learning_rate": 1.2810128014608121e-05, "loss": 0.4984, "step": 11987 }, { "epoch": 0.42709799241142204, "grad_norm": 1.1852519650534705, "learning_rate": 1.2809020558215992e-05, "loss": 0.5043, "step": 11988 }, { "epoch": 0.4271336195379162, "grad_norm": 1.5387183127157935, "learning_rate": 1.2807913064419797e-05, "loss": 0.5493, "step": 11989 }, { "epoch": 0.4271692466644103, "grad_norm": 1.1810471432114125, "learning_rate": 1.2806805533234272e-05, "loss": 0.4589, "step": 11990 }, { "epoch": 0.4272048737909044, "grad_norm": 1.3989863233101731, "learning_rate": 1.280569796467417e-05, "loss": 0.5051, "step": 11991 }, { "epoch": 0.4272405009173985, "grad_norm": 1.5931374472423425, "learning_rate": 1.280459035875424e-05, "loss": 0.6133, "step": 11992 }, { "epoch": 0.42727612804389264, "grad_norm": 1.7041814031479925, "learning_rate": 1.280348271548923e-05, "loss": 0.6742, "step": 11993 }, { "epoch": 0.4273117551703867, "grad_norm": 1.1938046225309802, "learning_rate": 1.2802375034893886e-05, "loss": 0.4524, "step": 11994 }, { "epoch": 0.42734738229688085, "grad_norm": 1.1438819878353936, "learning_rate": 1.280126731698296e-05, "loss": 0.4633, "step": 11995 }, { "epoch": 0.427383009423375, "grad_norm": 1.346693990706263, "learning_rate": 1.2800159561771201e-05, "loss": 0.4792, "step": 11996 }, { "epoch": 0.42741863654986906, "grad_norm": 1.1513758002860346, "learning_rate": 1.2799051769273363e-05, "loss": 0.4476, "step": 11997 }, { "epoch": 0.4274542636763632, "grad_norm": 1.3383927401974178, "learning_rate": 1.2797943939504192e-05, "loss": 0.4997, "step": 11998 }, { "epoch": 0.4274898908028573, "grad_norm": 1.1246919552922778, "learning_rate": 1.279683607247844e-05, "loss": 0.4629, "step": 11999 }, { "epoch": 0.4275255179293514, "grad_norm": 1.1939107945917633, "learning_rate": 1.2795728168210864e-05, "loss": 0.4951, "step": 12000 }, { "epoch": 0.4275611450558455, "grad_norm": 1.5572238050389944, "learning_rate": 1.2794620226716213e-05, "loss": 0.6491, "step": 12001 }, { "epoch": 0.42759677218233966, "grad_norm": 1.1063205625695514, "learning_rate": 1.2793512248009241e-05, "loss": 0.461, "step": 12002 }, { "epoch": 0.42763239930883373, "grad_norm": 1.7025550847350952, "learning_rate": 1.2792404232104699e-05, "loss": 0.5412, "step": 12003 }, { "epoch": 0.42766802643532786, "grad_norm": 1.527530401692466, "learning_rate": 1.2791296179017343e-05, "loss": 0.4981, "step": 12004 }, { "epoch": 0.427703653561822, "grad_norm": 1.2590566658473006, "learning_rate": 1.279018808876193e-05, "loss": 0.4636, "step": 12005 }, { "epoch": 0.42773928068831607, "grad_norm": 1.551454900925541, "learning_rate": 1.2789079961353208e-05, "loss": 0.5698, "step": 12006 }, { "epoch": 0.4277749078148102, "grad_norm": 1.5609063750143681, "learning_rate": 1.278797179680594e-05, "loss": 0.602, "step": 12007 }, { "epoch": 0.42781053494130433, "grad_norm": 1.523675842982343, "learning_rate": 1.278686359513488e-05, "loss": 0.5435, "step": 12008 }, { "epoch": 0.4278461620677984, "grad_norm": 1.0754423911320579, "learning_rate": 1.2785755356354778e-05, "loss": 0.4446, "step": 12009 }, { "epoch": 0.42788178919429254, "grad_norm": 1.1484185438568788, "learning_rate": 1.27846470804804e-05, "loss": 0.4811, "step": 12010 }, { "epoch": 0.42791741632078667, "grad_norm": 2.326915914107795, "learning_rate": 1.2783538767526497e-05, "loss": 0.6658, "step": 12011 }, { "epoch": 0.42795304344728075, "grad_norm": 1.527984617297398, "learning_rate": 1.2782430417507836e-05, "loss": 0.566, "step": 12012 }, { "epoch": 0.4279886705737749, "grad_norm": 1.448856999607575, "learning_rate": 1.2781322030439165e-05, "loss": 0.5195, "step": 12013 }, { "epoch": 0.428024297700269, "grad_norm": 1.307560391187308, "learning_rate": 1.2780213606335246e-05, "loss": 0.5266, "step": 12014 }, { "epoch": 0.4280599248267631, "grad_norm": 1.1079911296742722, "learning_rate": 1.2779105145210842e-05, "loss": 0.5127, "step": 12015 }, { "epoch": 0.4280955519532572, "grad_norm": 1.5102587870848347, "learning_rate": 1.2777996647080711e-05, "loss": 0.5596, "step": 12016 }, { "epoch": 0.42813117907975135, "grad_norm": 1.7670759198408887, "learning_rate": 1.2776888111959609e-05, "loss": 0.509, "step": 12017 }, { "epoch": 0.4281668062062454, "grad_norm": 1.616659682330635, "learning_rate": 1.2775779539862305e-05, "loss": 0.5134, "step": 12018 }, { "epoch": 0.42820243333273955, "grad_norm": 1.5261874716872426, "learning_rate": 1.2774670930803555e-05, "loss": 0.535, "step": 12019 }, { "epoch": 0.4282380604592337, "grad_norm": 1.2811273659837803, "learning_rate": 1.2773562284798122e-05, "loss": 0.3913, "step": 12020 }, { "epoch": 0.42827368758572776, "grad_norm": 1.0998311338454738, "learning_rate": 1.2772453601860768e-05, "loss": 0.4696, "step": 12021 }, { "epoch": 0.4283093147122219, "grad_norm": 1.4978793252543021, "learning_rate": 1.2771344882006256e-05, "loss": 0.5006, "step": 12022 }, { "epoch": 0.428344941838716, "grad_norm": 1.1424617661758534, "learning_rate": 1.2770236125249351e-05, "loss": 0.5059, "step": 12023 }, { "epoch": 0.4283805689652101, "grad_norm": 1.7055495993360197, "learning_rate": 1.2769127331604817e-05, "loss": 0.6507, "step": 12024 }, { "epoch": 0.4284161960917042, "grad_norm": 1.3915396796749189, "learning_rate": 1.2768018501087412e-05, "loss": 0.4633, "step": 12025 }, { "epoch": 0.42845182321819836, "grad_norm": 1.3224682299341948, "learning_rate": 1.276690963371191e-05, "loss": 0.4897, "step": 12026 }, { "epoch": 0.42848745034469243, "grad_norm": 1.1720582168938096, "learning_rate": 1.2765800729493073e-05, "loss": 0.4722, "step": 12027 }, { "epoch": 0.42852307747118656, "grad_norm": 1.670872225717355, "learning_rate": 1.2764691788445664e-05, "loss": 0.4836, "step": 12028 }, { "epoch": 0.4285587045976807, "grad_norm": 1.7502217763238737, "learning_rate": 1.2763582810584452e-05, "loss": 0.4596, "step": 12029 }, { "epoch": 0.42859433172417477, "grad_norm": 1.1551761437698171, "learning_rate": 1.2762473795924204e-05, "loss": 0.4758, "step": 12030 }, { "epoch": 0.4286299588506689, "grad_norm": 1.2117750264704998, "learning_rate": 1.2761364744479688e-05, "loss": 0.4839, "step": 12031 }, { "epoch": 0.42866558597716303, "grad_norm": 1.2376282161411218, "learning_rate": 1.2760255656265665e-05, "loss": 0.4387, "step": 12032 }, { "epoch": 0.4287012131036571, "grad_norm": 1.5470633296574965, "learning_rate": 1.275914653129691e-05, "loss": 0.6407, "step": 12033 }, { "epoch": 0.42873684023015124, "grad_norm": 1.8970225557157934, "learning_rate": 1.2758037369588195e-05, "loss": 0.5835, "step": 12034 }, { "epoch": 0.42877246735664537, "grad_norm": 1.7263840260756012, "learning_rate": 1.2756928171154284e-05, "loss": 0.4284, "step": 12035 }, { "epoch": 0.42880809448313945, "grad_norm": 1.933893558982576, "learning_rate": 1.2755818936009943e-05, "loss": 0.574, "step": 12036 }, { "epoch": 0.4288437216096336, "grad_norm": 1.098562688067087, "learning_rate": 1.2754709664169948e-05, "loss": 0.4803, "step": 12037 }, { "epoch": 0.4288793487361277, "grad_norm": 1.7028356426450828, "learning_rate": 1.2753600355649071e-05, "loss": 0.6005, "step": 12038 }, { "epoch": 0.4289149758626218, "grad_norm": 1.8635117679840523, "learning_rate": 1.275249101046208e-05, "loss": 0.545, "step": 12039 }, { "epoch": 0.4289506029891159, "grad_norm": 1.0005432084499353, "learning_rate": 1.2751381628623746e-05, "loss": 0.413, "step": 12040 }, { "epoch": 0.42898623011561005, "grad_norm": 1.0967085344886378, "learning_rate": 1.2750272210148844e-05, "loss": 0.4686, "step": 12041 }, { "epoch": 0.4290218572421041, "grad_norm": 1.2862165183493839, "learning_rate": 1.2749162755052145e-05, "loss": 0.6336, "step": 12042 }, { "epoch": 0.42905748436859825, "grad_norm": 1.2575249942602884, "learning_rate": 1.274805326334842e-05, "loss": 0.476, "step": 12043 }, { "epoch": 0.4290931114950924, "grad_norm": 1.169080269225528, "learning_rate": 1.2746943735052445e-05, "loss": 0.5089, "step": 12044 }, { "epoch": 0.42912873862158646, "grad_norm": 1.0914556630551717, "learning_rate": 1.2745834170178996e-05, "loss": 0.4618, "step": 12045 }, { "epoch": 0.4291643657480806, "grad_norm": 1.599989972484523, "learning_rate": 1.2744724568742848e-05, "loss": 0.5887, "step": 12046 }, { "epoch": 0.4291999928745747, "grad_norm": 1.0727534141376422, "learning_rate": 1.274361493075877e-05, "loss": 0.4727, "step": 12047 }, { "epoch": 0.4292356200010688, "grad_norm": 1.1960102211676558, "learning_rate": 1.2742505256241543e-05, "loss": 0.4691, "step": 12048 }, { "epoch": 0.4292712471275629, "grad_norm": 1.411677426743118, "learning_rate": 1.2741395545205944e-05, "loss": 0.669, "step": 12049 }, { "epoch": 0.42930687425405706, "grad_norm": 1.9074223742829548, "learning_rate": 1.2740285797666742e-05, "loss": 0.6095, "step": 12050 }, { "epoch": 0.42934250138055113, "grad_norm": 1.6222506763496525, "learning_rate": 1.2739176013638724e-05, "loss": 0.5757, "step": 12051 }, { "epoch": 0.42937812850704526, "grad_norm": 1.2424111432887206, "learning_rate": 1.2738066193136661e-05, "loss": 0.5132, "step": 12052 }, { "epoch": 0.4294137556335394, "grad_norm": 1.1431451057477604, "learning_rate": 1.2736956336175332e-05, "loss": 0.5109, "step": 12053 }, { "epoch": 0.42944938276003347, "grad_norm": 1.040343760304305, "learning_rate": 1.2735846442769516e-05, "loss": 0.4721, "step": 12054 }, { "epoch": 0.4294850098865276, "grad_norm": 1.1192360013052554, "learning_rate": 1.2734736512933992e-05, "loss": 0.4843, "step": 12055 }, { "epoch": 0.42952063701302173, "grad_norm": 1.0988397080462244, "learning_rate": 1.2733626546683542e-05, "loss": 0.4414, "step": 12056 }, { "epoch": 0.4295562641395158, "grad_norm": 1.168204194056277, "learning_rate": 1.2732516544032944e-05, "loss": 0.4569, "step": 12057 }, { "epoch": 0.42959189126600994, "grad_norm": 1.1947345700017986, "learning_rate": 1.2731406504996979e-05, "loss": 0.4888, "step": 12058 }, { "epoch": 0.42962751839250407, "grad_norm": 1.268836316508902, "learning_rate": 1.2730296429590423e-05, "loss": 0.5497, "step": 12059 }, { "epoch": 0.42966314551899815, "grad_norm": 1.0097668051890347, "learning_rate": 1.2729186317828066e-05, "loss": 0.454, "step": 12060 }, { "epoch": 0.4296987726454923, "grad_norm": 1.2508066252353758, "learning_rate": 1.2728076169724686e-05, "loss": 0.4961, "step": 12061 }, { "epoch": 0.4297343997719864, "grad_norm": 1.8616993759330243, "learning_rate": 1.2726965985295061e-05, "loss": 0.6489, "step": 12062 }, { "epoch": 0.4297700268984805, "grad_norm": 1.7279113687605758, "learning_rate": 1.2725855764553981e-05, "loss": 0.6266, "step": 12063 }, { "epoch": 0.4298056540249746, "grad_norm": 1.6936348210107335, "learning_rate": 1.2724745507516225e-05, "loss": 0.5633, "step": 12064 }, { "epoch": 0.42984128115146875, "grad_norm": 1.0826923831547974, "learning_rate": 1.2723635214196579e-05, "loss": 0.4525, "step": 12065 }, { "epoch": 0.4298769082779628, "grad_norm": 1.2957392054766512, "learning_rate": 1.2722524884609824e-05, "loss": 0.5203, "step": 12066 }, { "epoch": 0.42991253540445695, "grad_norm": 1.6538332521596535, "learning_rate": 1.2721414518770748e-05, "loss": 0.5939, "step": 12067 }, { "epoch": 0.4299481625309511, "grad_norm": 1.270683899935621, "learning_rate": 1.2720304116694137e-05, "loss": 0.466, "step": 12068 }, { "epoch": 0.42998378965744516, "grad_norm": 1.5296964444863723, "learning_rate": 1.2719193678394775e-05, "loss": 0.58, "step": 12069 }, { "epoch": 0.4300194167839393, "grad_norm": 1.4297593628964334, "learning_rate": 1.2718083203887445e-05, "loss": 0.6184, "step": 12070 }, { "epoch": 0.4300550439104334, "grad_norm": 1.3816064270415582, "learning_rate": 1.2716972693186941e-05, "loss": 0.4976, "step": 12071 }, { "epoch": 0.4300906710369275, "grad_norm": 1.7785007585026507, "learning_rate": 1.2715862146308049e-05, "loss": 0.6384, "step": 12072 }, { "epoch": 0.4301262981634216, "grad_norm": 1.2560191059552812, "learning_rate": 1.2714751563265551e-05, "loss": 0.4703, "step": 12073 }, { "epoch": 0.43016192528991576, "grad_norm": 1.08959847407862, "learning_rate": 1.271364094407424e-05, "loss": 0.4438, "step": 12074 }, { "epoch": 0.43019755241640983, "grad_norm": 1.9688545756134883, "learning_rate": 1.2712530288748901e-05, "loss": 0.6232, "step": 12075 }, { "epoch": 0.43023317954290397, "grad_norm": 0.9889208551252494, "learning_rate": 1.2711419597304326e-05, "loss": 0.4189, "step": 12076 }, { "epoch": 0.4302688066693981, "grad_norm": 1.8931479590546894, "learning_rate": 1.2710308869755303e-05, "loss": 0.5628, "step": 12077 }, { "epoch": 0.43030443379589217, "grad_norm": 1.3574313304642376, "learning_rate": 1.2709198106116624e-05, "loss": 0.5072, "step": 12078 }, { "epoch": 0.4303400609223863, "grad_norm": 1.1853980035042817, "learning_rate": 1.2708087306403078e-05, "loss": 0.4919, "step": 12079 }, { "epoch": 0.43037568804888043, "grad_norm": 1.6299073235302544, "learning_rate": 1.2706976470629457e-05, "loss": 0.6116, "step": 12080 }, { "epoch": 0.4304113151753745, "grad_norm": 1.1591822372278546, "learning_rate": 1.2705865598810551e-05, "loss": 0.4686, "step": 12081 }, { "epoch": 0.43044694230186864, "grad_norm": 1.58415778871782, "learning_rate": 1.2704754690961157e-05, "loss": 0.6357, "step": 12082 }, { "epoch": 0.43048256942836277, "grad_norm": 1.526872230777302, "learning_rate": 1.2703643747096064e-05, "loss": 0.5758, "step": 12083 }, { "epoch": 0.43051819655485685, "grad_norm": 1.1475397086903953, "learning_rate": 1.2702532767230062e-05, "loss": 0.4922, "step": 12084 }, { "epoch": 0.430553823681351, "grad_norm": 1.0658534601830283, "learning_rate": 1.2701421751377944e-05, "loss": 0.4323, "step": 12085 }, { "epoch": 0.4305894508078451, "grad_norm": 1.1855597131050806, "learning_rate": 1.2700310699554513e-05, "loss": 0.4956, "step": 12086 }, { "epoch": 0.4306250779343392, "grad_norm": 1.574259488142451, "learning_rate": 1.2699199611774558e-05, "loss": 0.6384, "step": 12087 }, { "epoch": 0.4306607050608333, "grad_norm": 1.2737768972038666, "learning_rate": 1.269808848805287e-05, "loss": 0.6127, "step": 12088 }, { "epoch": 0.43069633218732745, "grad_norm": 1.317668464732962, "learning_rate": 1.2696977328404248e-05, "loss": 0.5824, "step": 12089 }, { "epoch": 0.4307319593138215, "grad_norm": 1.5824323614959628, "learning_rate": 1.2695866132843486e-05, "loss": 0.5915, "step": 12090 }, { "epoch": 0.43076758644031565, "grad_norm": 1.2492636638696457, "learning_rate": 1.2694754901385386e-05, "loss": 0.5004, "step": 12091 }, { "epoch": 0.4308032135668098, "grad_norm": 1.1295645455416126, "learning_rate": 1.2693643634044738e-05, "loss": 0.5369, "step": 12092 }, { "epoch": 0.43083884069330386, "grad_norm": 1.4968785516226735, "learning_rate": 1.2692532330836346e-05, "loss": 0.5736, "step": 12093 }, { "epoch": 0.430874467819798, "grad_norm": 1.5904315728722687, "learning_rate": 1.2691420991775003e-05, "loss": 0.5903, "step": 12094 }, { "epoch": 0.4309100949462921, "grad_norm": 2.0883244519168347, "learning_rate": 1.2690309616875509e-05, "loss": 0.6062, "step": 12095 }, { "epoch": 0.4309457220727862, "grad_norm": 1.100718919535913, "learning_rate": 1.2689198206152657e-05, "loss": 0.4552, "step": 12096 }, { "epoch": 0.43098134919928033, "grad_norm": 1.303292022906445, "learning_rate": 1.2688086759621256e-05, "loss": 0.5163, "step": 12097 }, { "epoch": 0.43101697632577446, "grad_norm": 1.1446648434251456, "learning_rate": 1.26869752772961e-05, "loss": 0.4331, "step": 12098 }, { "epoch": 0.43105260345226853, "grad_norm": 1.166514620725194, "learning_rate": 1.268586375919199e-05, "loss": 0.488, "step": 12099 }, { "epoch": 0.43108823057876267, "grad_norm": 1.2975635416400775, "learning_rate": 1.2684752205323727e-05, "loss": 0.4919, "step": 12100 }, { "epoch": 0.4311238577052568, "grad_norm": 1.0641447894342086, "learning_rate": 1.2683640615706113e-05, "loss": 0.4944, "step": 12101 }, { "epoch": 0.43115948483175087, "grad_norm": 1.1417992749244896, "learning_rate": 1.2682528990353943e-05, "loss": 0.505, "step": 12102 }, { "epoch": 0.431195111958245, "grad_norm": 1.010450268447833, "learning_rate": 1.268141732928203e-05, "loss": 0.4495, "step": 12103 }, { "epoch": 0.43123073908473913, "grad_norm": 1.2124033540321837, "learning_rate": 1.2680305632505168e-05, "loss": 0.5149, "step": 12104 }, { "epoch": 0.4312663662112332, "grad_norm": 1.0045674521391594, "learning_rate": 1.2679193900038165e-05, "loss": 0.4561, "step": 12105 }, { "epoch": 0.43130199333772734, "grad_norm": 1.0874470725208256, "learning_rate": 1.2678082131895824e-05, "loss": 0.4317, "step": 12106 }, { "epoch": 0.43133762046422147, "grad_norm": 1.2575332419277532, "learning_rate": 1.2676970328092944e-05, "loss": 0.4747, "step": 12107 }, { "epoch": 0.43137324759071555, "grad_norm": 1.1049356082145143, "learning_rate": 1.2675858488644334e-05, "loss": 0.4378, "step": 12108 }, { "epoch": 0.4314088747172097, "grad_norm": 1.8682760941895695, "learning_rate": 1.2674746613564799e-05, "loss": 0.5521, "step": 12109 }, { "epoch": 0.4314445018437038, "grad_norm": 1.1300898481403028, "learning_rate": 1.2673634702869144e-05, "loss": 0.4603, "step": 12110 }, { "epoch": 0.4314801289701979, "grad_norm": 1.1420927016052091, "learning_rate": 1.2672522756572173e-05, "loss": 0.4475, "step": 12111 }, { "epoch": 0.431515756096692, "grad_norm": 1.3862473639954727, "learning_rate": 1.2671410774688693e-05, "loss": 0.5295, "step": 12112 }, { "epoch": 0.43155138322318615, "grad_norm": 1.4684955910146509, "learning_rate": 1.267029875723351e-05, "loss": 0.6792, "step": 12113 }, { "epoch": 0.4315870103496802, "grad_norm": 1.2695591673336184, "learning_rate": 1.2669186704221436e-05, "loss": 0.586, "step": 12114 }, { "epoch": 0.43162263747617435, "grad_norm": 1.6039309243108641, "learning_rate": 1.2668074615667272e-05, "loss": 0.5754, "step": 12115 }, { "epoch": 0.4316582646026685, "grad_norm": 1.382705439347799, "learning_rate": 1.2666962491585832e-05, "loss": 0.5374, "step": 12116 }, { "epoch": 0.43169389172916256, "grad_norm": 1.5544262979529682, "learning_rate": 1.2665850331991923e-05, "loss": 0.5663, "step": 12117 }, { "epoch": 0.4317295188556567, "grad_norm": 1.5588571648029246, "learning_rate": 1.266473813690035e-05, "loss": 0.5905, "step": 12118 }, { "epoch": 0.4317651459821508, "grad_norm": 1.4289216698464637, "learning_rate": 1.266362590632593e-05, "loss": 0.5755, "step": 12119 }, { "epoch": 0.4318007731086449, "grad_norm": 1.1792259409403711, "learning_rate": 1.2662513640283467e-05, "loss": 0.4857, "step": 12120 }, { "epoch": 0.43183640023513903, "grad_norm": 1.4656168033323151, "learning_rate": 1.2661401338787776e-05, "loss": 0.5812, "step": 12121 }, { "epoch": 0.43187202736163316, "grad_norm": 1.6697307066596463, "learning_rate": 1.2660289001853663e-05, "loss": 0.4952, "step": 12122 }, { "epoch": 0.43190765448812724, "grad_norm": 1.1251492415660234, "learning_rate": 1.2659176629495947e-05, "loss": 0.4839, "step": 12123 }, { "epoch": 0.43194328161462137, "grad_norm": 1.0934706263245226, "learning_rate": 1.265806422172943e-05, "loss": 0.5057, "step": 12124 }, { "epoch": 0.4319789087411155, "grad_norm": 1.0401926455803696, "learning_rate": 1.2656951778568933e-05, "loss": 0.3991, "step": 12125 }, { "epoch": 0.4320145358676096, "grad_norm": 1.5738746302341013, "learning_rate": 1.2655839300029265e-05, "loss": 0.5347, "step": 12126 }, { "epoch": 0.4320501629941037, "grad_norm": 1.5298594000643293, "learning_rate": 1.265472678612524e-05, "loss": 0.5825, "step": 12127 }, { "epoch": 0.43208579012059783, "grad_norm": 1.1477787462303846, "learning_rate": 1.2653614236871676e-05, "loss": 0.4613, "step": 12128 }, { "epoch": 0.4321214172470919, "grad_norm": 1.7291364359204544, "learning_rate": 1.2652501652283378e-05, "loss": 0.5827, "step": 12129 }, { "epoch": 0.43215704437358604, "grad_norm": 1.5885163461467988, "learning_rate": 1.2651389032375172e-05, "loss": 0.6472, "step": 12130 }, { "epoch": 0.43219267150008017, "grad_norm": 1.1338860813037823, "learning_rate": 1.2650276377161865e-05, "loss": 0.4372, "step": 12131 }, { "epoch": 0.43222829862657425, "grad_norm": 1.6116889241759143, "learning_rate": 1.2649163686658276e-05, "loss": 0.6209, "step": 12132 }, { "epoch": 0.4322639257530684, "grad_norm": 1.6480622039742197, "learning_rate": 1.2648050960879222e-05, "loss": 0.5377, "step": 12133 }, { "epoch": 0.4322995528795625, "grad_norm": 1.166890040662636, "learning_rate": 1.2646938199839516e-05, "loss": 0.4409, "step": 12134 }, { "epoch": 0.4323351800060566, "grad_norm": 1.030375551494806, "learning_rate": 1.2645825403553979e-05, "loss": 0.5108, "step": 12135 }, { "epoch": 0.4323708071325507, "grad_norm": 1.0841100229952356, "learning_rate": 1.2644712572037427e-05, "loss": 0.4246, "step": 12136 }, { "epoch": 0.43240643425904485, "grad_norm": 1.1941063717227978, "learning_rate": 1.2643599705304678e-05, "loss": 0.5474, "step": 12137 }, { "epoch": 0.4324420613855389, "grad_norm": 1.173388036635933, "learning_rate": 1.2642486803370553e-05, "loss": 0.4667, "step": 12138 }, { "epoch": 0.43247768851203305, "grad_norm": 1.078945413577371, "learning_rate": 1.2641373866249868e-05, "loss": 0.4849, "step": 12139 }, { "epoch": 0.4325133156385272, "grad_norm": 1.0778279198265053, "learning_rate": 1.2640260893957445e-05, "loss": 0.4295, "step": 12140 }, { "epoch": 0.43254894276502126, "grad_norm": 2.1851071205744064, "learning_rate": 1.2639147886508104e-05, "loss": 0.5189, "step": 12141 }, { "epoch": 0.4325845698915154, "grad_norm": 1.1053281484638557, "learning_rate": 1.263803484391666e-05, "loss": 0.4571, "step": 12142 }, { "epoch": 0.4326201970180095, "grad_norm": 1.1261563741613663, "learning_rate": 1.2636921766197942e-05, "loss": 0.4723, "step": 12143 }, { "epoch": 0.4326558241445036, "grad_norm": 1.649006264981596, "learning_rate": 1.2635808653366764e-05, "loss": 0.5903, "step": 12144 }, { "epoch": 0.43269145127099773, "grad_norm": 1.3188093733279838, "learning_rate": 1.2634695505437955e-05, "loss": 0.4888, "step": 12145 }, { "epoch": 0.43272707839749186, "grad_norm": 1.1994497095478516, "learning_rate": 1.2633582322426332e-05, "loss": 0.4746, "step": 12146 }, { "epoch": 0.43276270552398594, "grad_norm": 1.4467333482476843, "learning_rate": 1.2632469104346722e-05, "loss": 0.5856, "step": 12147 }, { "epoch": 0.43279833265048007, "grad_norm": 1.5625919895583409, "learning_rate": 1.263135585121394e-05, "loss": 0.5583, "step": 12148 }, { "epoch": 0.4328339597769742, "grad_norm": 1.415190141024666, "learning_rate": 1.263024256304282e-05, "loss": 0.5633, "step": 12149 }, { "epoch": 0.4328695869034683, "grad_norm": 1.2764946157738408, "learning_rate": 1.262912923984818e-05, "loss": 0.4565, "step": 12150 }, { "epoch": 0.4329052140299624, "grad_norm": 1.317502773283719, "learning_rate": 1.262801588164485e-05, "loss": 0.4994, "step": 12151 }, { "epoch": 0.43294084115645654, "grad_norm": 1.2603242644344221, "learning_rate": 1.262690248844765e-05, "loss": 0.4645, "step": 12152 }, { "epoch": 0.43297646828295067, "grad_norm": 1.1131077994633816, "learning_rate": 1.2625789060271405e-05, "loss": 0.494, "step": 12153 }, { "epoch": 0.43301209540944474, "grad_norm": 1.3226005514115498, "learning_rate": 1.2624675597130946e-05, "loss": 0.5591, "step": 12154 }, { "epoch": 0.4330477225359389, "grad_norm": 1.095351184098069, "learning_rate": 1.2623562099041096e-05, "loss": 0.4224, "step": 12155 }, { "epoch": 0.433083349662433, "grad_norm": 1.5441952901324625, "learning_rate": 1.2622448566016683e-05, "loss": 0.6074, "step": 12156 }, { "epoch": 0.4331189767889271, "grad_norm": 1.0415718840125292, "learning_rate": 1.2621334998072534e-05, "loss": 0.497, "step": 12157 }, { "epoch": 0.4331546039154212, "grad_norm": 1.3936478798256116, "learning_rate": 1.262022139522348e-05, "loss": 0.5538, "step": 12158 }, { "epoch": 0.43319023104191534, "grad_norm": 1.1166997098192606, "learning_rate": 1.2619107757484341e-05, "loss": 0.4918, "step": 12159 }, { "epoch": 0.4332258581684094, "grad_norm": 1.1438476927055818, "learning_rate": 1.2617994084869954e-05, "loss": 0.433, "step": 12160 }, { "epoch": 0.43326148529490355, "grad_norm": 1.1192111911698708, "learning_rate": 1.2616880377395147e-05, "loss": 0.5016, "step": 12161 }, { "epoch": 0.4332971124213977, "grad_norm": 1.5945370092008415, "learning_rate": 1.261576663507475e-05, "loss": 0.531, "step": 12162 }, { "epoch": 0.43333273954789175, "grad_norm": 1.1533766822729712, "learning_rate": 1.2614652857923589e-05, "loss": 0.4432, "step": 12163 }, { "epoch": 0.4333683666743859, "grad_norm": 1.5300316260582603, "learning_rate": 1.26135390459565e-05, "loss": 0.5499, "step": 12164 }, { "epoch": 0.43340399380088, "grad_norm": 1.0130662035702858, "learning_rate": 1.261242519918831e-05, "loss": 0.4843, "step": 12165 }, { "epoch": 0.4334396209273741, "grad_norm": 1.0470315004254185, "learning_rate": 1.2611311317633854e-05, "loss": 0.4491, "step": 12166 }, { "epoch": 0.4334752480538682, "grad_norm": 1.3782995760565178, "learning_rate": 1.2610197401307963e-05, "loss": 0.5293, "step": 12167 }, { "epoch": 0.43351087518036235, "grad_norm": 1.0646452261417976, "learning_rate": 1.2609083450225468e-05, "loss": 0.4738, "step": 12168 }, { "epoch": 0.43354650230685643, "grad_norm": 1.4233045868711571, "learning_rate": 1.2607969464401204e-05, "loss": 0.5657, "step": 12169 }, { "epoch": 0.43358212943335056, "grad_norm": 0.9918213360100567, "learning_rate": 1.260685544385e-05, "loss": 0.4257, "step": 12170 }, { "epoch": 0.4336177565598447, "grad_norm": 1.14541760884889, "learning_rate": 1.2605741388586696e-05, "loss": 0.4542, "step": 12171 }, { "epoch": 0.43365338368633877, "grad_norm": 1.048418254303679, "learning_rate": 1.2604627298626126e-05, "loss": 0.4525, "step": 12172 }, { "epoch": 0.4336890108128329, "grad_norm": 1.5162614116318576, "learning_rate": 1.2603513173983121e-05, "loss": 0.509, "step": 12173 }, { "epoch": 0.43372463793932703, "grad_norm": 1.0771410312019805, "learning_rate": 1.260239901467252e-05, "loss": 0.4395, "step": 12174 }, { "epoch": 0.4337602650658211, "grad_norm": 1.207105709905223, "learning_rate": 1.2601284820709156e-05, "loss": 0.447, "step": 12175 }, { "epoch": 0.43379589219231524, "grad_norm": 1.7206696319611463, "learning_rate": 1.2600170592107868e-05, "loss": 0.6433, "step": 12176 }, { "epoch": 0.43383151931880937, "grad_norm": 1.1508511953591771, "learning_rate": 1.259905632888349e-05, "loss": 0.4098, "step": 12177 }, { "epoch": 0.43386714644530344, "grad_norm": 1.1419631480875698, "learning_rate": 1.2597942031050859e-05, "loss": 0.4923, "step": 12178 }, { "epoch": 0.4339027735717976, "grad_norm": 1.0142974585854654, "learning_rate": 1.2596827698624816e-05, "loss": 0.4368, "step": 12179 }, { "epoch": 0.4339384006982917, "grad_norm": 1.7634767933501958, "learning_rate": 1.2595713331620196e-05, "loss": 0.531, "step": 12180 }, { "epoch": 0.4339740278247858, "grad_norm": 1.0313303159293572, "learning_rate": 1.2594598930051838e-05, "loss": 0.4804, "step": 12181 }, { "epoch": 0.4340096549512799, "grad_norm": 1.1338911506379223, "learning_rate": 1.2593484493934582e-05, "loss": 0.527, "step": 12182 }, { "epoch": 0.43404528207777404, "grad_norm": 1.0846038563638705, "learning_rate": 1.2592370023283268e-05, "loss": 0.5017, "step": 12183 }, { "epoch": 0.4340809092042681, "grad_norm": 1.0403330793186325, "learning_rate": 1.2591255518112737e-05, "loss": 0.4326, "step": 12184 }, { "epoch": 0.43411653633076225, "grad_norm": 1.4672160133081498, "learning_rate": 1.2590140978437826e-05, "loss": 0.5503, "step": 12185 }, { "epoch": 0.4341521634572564, "grad_norm": 1.7070418427491718, "learning_rate": 1.2589026404273376e-05, "loss": 0.5659, "step": 12186 }, { "epoch": 0.43418779058375045, "grad_norm": 1.9745505260803362, "learning_rate": 1.2587911795634231e-05, "loss": 0.5581, "step": 12187 }, { "epoch": 0.4342234177102446, "grad_norm": 1.3174293842110787, "learning_rate": 1.2586797152535233e-05, "loss": 0.5706, "step": 12188 }, { "epoch": 0.4342590448367387, "grad_norm": 1.3102783081785716, "learning_rate": 1.2585682474991221e-05, "loss": 0.5526, "step": 12189 }, { "epoch": 0.4342946719632328, "grad_norm": 1.4947606977709997, "learning_rate": 1.258456776301704e-05, "loss": 0.5435, "step": 12190 }, { "epoch": 0.4343302990897269, "grad_norm": 1.1451322817532854, "learning_rate": 1.2583453016627533e-05, "loss": 0.4539, "step": 12191 }, { "epoch": 0.43436592621622105, "grad_norm": 1.0225072619337747, "learning_rate": 1.2582338235837544e-05, "loss": 0.5199, "step": 12192 }, { "epoch": 0.43440155334271513, "grad_norm": 1.4561591525277335, "learning_rate": 1.2581223420661914e-05, "loss": 0.6073, "step": 12193 }, { "epoch": 0.43443718046920926, "grad_norm": 1.2117666882276397, "learning_rate": 1.258010857111549e-05, "loss": 0.5176, "step": 12194 }, { "epoch": 0.4344728075957034, "grad_norm": 1.0463806886573244, "learning_rate": 1.257899368721312e-05, "loss": 0.4437, "step": 12195 }, { "epoch": 0.43450843472219747, "grad_norm": 1.3975718933421597, "learning_rate": 1.2577878768969644e-05, "loss": 0.4485, "step": 12196 }, { "epoch": 0.4345440618486916, "grad_norm": 3.613373679646495, "learning_rate": 1.2576763816399913e-05, "loss": 0.5773, "step": 12197 }, { "epoch": 0.43457968897518573, "grad_norm": 1.5588107640384827, "learning_rate": 1.2575648829518769e-05, "loss": 0.499, "step": 12198 }, { "epoch": 0.4346153161016798, "grad_norm": 1.1020350573590074, "learning_rate": 1.257453380834106e-05, "loss": 0.5181, "step": 12199 }, { "epoch": 0.43465094322817394, "grad_norm": 1.0318051442356997, "learning_rate": 1.2573418752881635e-05, "loss": 0.4519, "step": 12200 }, { "epoch": 0.43468657035466807, "grad_norm": 1.1484441975140087, "learning_rate": 1.257230366315534e-05, "loss": 0.4989, "step": 12201 }, { "epoch": 0.43472219748116214, "grad_norm": 1.0728503871380282, "learning_rate": 1.2571188539177027e-05, "loss": 0.4539, "step": 12202 }, { "epoch": 0.4347578246076563, "grad_norm": 1.1784747329665806, "learning_rate": 1.2570073380961536e-05, "loss": 0.4685, "step": 12203 }, { "epoch": 0.4347934517341504, "grad_norm": 1.0965105186439859, "learning_rate": 1.2568958188523723e-05, "loss": 0.4985, "step": 12204 }, { "epoch": 0.4348290788606445, "grad_norm": 0.9619010695275239, "learning_rate": 1.2567842961878438e-05, "loss": 0.3829, "step": 12205 }, { "epoch": 0.4348647059871386, "grad_norm": 1.127481291950895, "learning_rate": 1.2566727701040527e-05, "loss": 0.4769, "step": 12206 }, { "epoch": 0.43490033311363274, "grad_norm": 1.1204087345837908, "learning_rate": 1.2565612406024845e-05, "loss": 0.5407, "step": 12207 }, { "epoch": 0.4349359602401268, "grad_norm": 1.535723305155318, "learning_rate": 1.2564497076846239e-05, "loss": 0.5275, "step": 12208 }, { "epoch": 0.43497158736662095, "grad_norm": 1.519246858034529, "learning_rate": 1.2563381713519562e-05, "loss": 0.5938, "step": 12209 }, { "epoch": 0.4350072144931151, "grad_norm": 1.0387837388118848, "learning_rate": 1.2562266316059665e-05, "loss": 0.4597, "step": 12210 }, { "epoch": 0.43504284161960916, "grad_norm": 1.7389859025432677, "learning_rate": 1.2561150884481403e-05, "loss": 0.5581, "step": 12211 }, { "epoch": 0.4350784687461033, "grad_norm": 1.5721291719123094, "learning_rate": 1.2560035418799626e-05, "loss": 0.5858, "step": 12212 }, { "epoch": 0.4351140958725974, "grad_norm": 1.564828364439785, "learning_rate": 1.2558919919029189e-05, "loss": 0.521, "step": 12213 }, { "epoch": 0.4351497229990915, "grad_norm": 1.189010827875331, "learning_rate": 1.2557804385184944e-05, "loss": 0.4379, "step": 12214 }, { "epoch": 0.4351853501255856, "grad_norm": 1.1219837583368848, "learning_rate": 1.2556688817281745e-05, "loss": 0.4719, "step": 12215 }, { "epoch": 0.43522097725207975, "grad_norm": 1.5813999425069707, "learning_rate": 1.2555573215334449e-05, "loss": 0.5888, "step": 12216 }, { "epoch": 0.43525660437857383, "grad_norm": 1.197974200555213, "learning_rate": 1.2554457579357906e-05, "loss": 0.5211, "step": 12217 }, { "epoch": 0.43529223150506796, "grad_norm": 1.1661131957713617, "learning_rate": 1.2553341909366979e-05, "loss": 0.4798, "step": 12218 }, { "epoch": 0.4353278586315621, "grad_norm": 1.2217017419556986, "learning_rate": 1.2552226205376517e-05, "loss": 0.4996, "step": 12219 }, { "epoch": 0.43536348575805617, "grad_norm": 1.0833467490123982, "learning_rate": 1.255111046740138e-05, "loss": 0.4911, "step": 12220 }, { "epoch": 0.4353991128845503, "grad_norm": 1.0940996387255362, "learning_rate": 1.2549994695456424e-05, "loss": 0.4823, "step": 12221 }, { "epoch": 0.43543474001104443, "grad_norm": 1.618157723073714, "learning_rate": 1.2548878889556508e-05, "loss": 0.6096, "step": 12222 }, { "epoch": 0.4354703671375385, "grad_norm": 1.0183019697156042, "learning_rate": 1.2547763049716486e-05, "loss": 0.4311, "step": 12223 }, { "epoch": 0.43550599426403264, "grad_norm": 1.4517377375032616, "learning_rate": 1.2546647175951218e-05, "loss": 0.5662, "step": 12224 }, { "epoch": 0.43554162139052677, "grad_norm": 1.5259867376691847, "learning_rate": 1.2545531268275565e-05, "loss": 0.624, "step": 12225 }, { "epoch": 0.43557724851702084, "grad_norm": 1.597296810728604, "learning_rate": 1.254441532670438e-05, "loss": 0.5408, "step": 12226 }, { "epoch": 0.435612875643515, "grad_norm": 1.4014332174110935, "learning_rate": 1.2543299351252532e-05, "loss": 0.6449, "step": 12227 }, { "epoch": 0.4356485027700091, "grad_norm": 1.2355692719034908, "learning_rate": 1.2542183341934873e-05, "loss": 0.5003, "step": 12228 }, { "epoch": 0.4356841298965032, "grad_norm": 1.4698852354189855, "learning_rate": 1.2541067298766263e-05, "loss": 0.5728, "step": 12229 }, { "epoch": 0.4357197570229973, "grad_norm": 1.511965138637855, "learning_rate": 1.2539951221761567e-05, "loss": 0.5018, "step": 12230 }, { "epoch": 0.43575538414949144, "grad_norm": 1.2100091992497373, "learning_rate": 1.2538835110935648e-05, "loss": 0.489, "step": 12231 }, { "epoch": 0.4357910112759855, "grad_norm": 1.4286374879173065, "learning_rate": 1.2537718966303363e-05, "loss": 0.5562, "step": 12232 }, { "epoch": 0.43582663840247965, "grad_norm": 1.1719506522300798, "learning_rate": 1.2536602787879576e-05, "loss": 0.5377, "step": 12233 }, { "epoch": 0.4358622655289738, "grad_norm": 1.57758929861515, "learning_rate": 1.253548657567915e-05, "loss": 0.5244, "step": 12234 }, { "epoch": 0.43589789265546786, "grad_norm": 1.581753527575404, "learning_rate": 1.2534370329716949e-05, "loss": 0.5761, "step": 12235 }, { "epoch": 0.435933519781962, "grad_norm": 1.5477817486091472, "learning_rate": 1.2533254050007837e-05, "loss": 0.5663, "step": 12236 }, { "epoch": 0.4359691469084561, "grad_norm": 1.0819151926784978, "learning_rate": 1.2532137736566674e-05, "loss": 0.4563, "step": 12237 }, { "epoch": 0.4360047740349502, "grad_norm": 1.4761563258542774, "learning_rate": 1.2531021389408329e-05, "loss": 0.617, "step": 12238 }, { "epoch": 0.4360404011614443, "grad_norm": 1.5749744726416353, "learning_rate": 1.2529905008547662e-05, "loss": 0.6324, "step": 12239 }, { "epoch": 0.43607602828793846, "grad_norm": 1.2843737295705273, "learning_rate": 1.2528788593999544e-05, "loss": 0.5279, "step": 12240 }, { "epoch": 0.43611165541443253, "grad_norm": 1.160439822195392, "learning_rate": 1.252767214577884e-05, "loss": 0.4874, "step": 12241 }, { "epoch": 0.43614728254092666, "grad_norm": 1.0270352795355646, "learning_rate": 1.252655566390041e-05, "loss": 0.43, "step": 12242 }, { "epoch": 0.4361829096674208, "grad_norm": 1.4469105313148318, "learning_rate": 1.2525439148379127e-05, "loss": 0.5497, "step": 12243 }, { "epoch": 0.43621853679391487, "grad_norm": 1.2941947577142883, "learning_rate": 1.2524322599229862e-05, "loss": 0.4192, "step": 12244 }, { "epoch": 0.436254163920409, "grad_norm": 1.2315058851818286, "learning_rate": 1.2523206016467468e-05, "loss": 0.4864, "step": 12245 }, { "epoch": 0.43628979104690313, "grad_norm": 1.7503165992621625, "learning_rate": 1.2522089400106829e-05, "loss": 0.6029, "step": 12246 }, { "epoch": 0.4363254181733972, "grad_norm": 1.0707842974091242, "learning_rate": 1.2520972750162804e-05, "loss": 0.4375, "step": 12247 }, { "epoch": 0.43636104529989134, "grad_norm": 1.5108676571873443, "learning_rate": 1.251985606665027e-05, "loss": 0.4911, "step": 12248 }, { "epoch": 0.43639667242638547, "grad_norm": 1.5696599917027056, "learning_rate": 1.2518739349584083e-05, "loss": 0.4795, "step": 12249 }, { "epoch": 0.43643229955287954, "grad_norm": 1.9543338954530045, "learning_rate": 1.2517622598979127e-05, "loss": 0.5475, "step": 12250 }, { "epoch": 0.4364679266793737, "grad_norm": 1.5282991323720578, "learning_rate": 1.2516505814850264e-05, "loss": 0.6065, "step": 12251 }, { "epoch": 0.4365035538058678, "grad_norm": 1.2284824150153286, "learning_rate": 1.2515388997212366e-05, "loss": 0.4801, "step": 12252 }, { "epoch": 0.4365391809323619, "grad_norm": 1.0972675315491751, "learning_rate": 1.2514272146080307e-05, "loss": 0.4783, "step": 12253 }, { "epoch": 0.436574808058856, "grad_norm": 1.0057947578311737, "learning_rate": 1.2513155261468957e-05, "loss": 0.4508, "step": 12254 }, { "epoch": 0.43661043518535014, "grad_norm": 1.4879493683825218, "learning_rate": 1.251203834339319e-05, "loss": 0.5335, "step": 12255 }, { "epoch": 0.4366460623118442, "grad_norm": 1.7683746580315112, "learning_rate": 1.251092139186787e-05, "loss": 0.6278, "step": 12256 }, { "epoch": 0.43668168943833835, "grad_norm": 1.040082798637245, "learning_rate": 1.2509804406907881e-05, "loss": 0.4337, "step": 12257 }, { "epoch": 0.4367173165648325, "grad_norm": 1.058446480902225, "learning_rate": 1.2508687388528094e-05, "loss": 0.4424, "step": 12258 }, { "epoch": 0.43675294369132656, "grad_norm": 1.1645744533337008, "learning_rate": 1.2507570336743383e-05, "loss": 0.4945, "step": 12259 }, { "epoch": 0.4367885708178207, "grad_norm": 0.9982976517574486, "learning_rate": 1.2506453251568614e-05, "loss": 0.4956, "step": 12260 }, { "epoch": 0.4368241979443148, "grad_norm": 1.5849709957987181, "learning_rate": 1.2505336133018671e-05, "loss": 0.6364, "step": 12261 }, { "epoch": 0.4368598250708089, "grad_norm": 1.8173043233582524, "learning_rate": 1.2504218981108424e-05, "loss": 0.5961, "step": 12262 }, { "epoch": 0.436895452197303, "grad_norm": 1.3555303461766532, "learning_rate": 1.2503101795852752e-05, "loss": 0.5749, "step": 12263 }, { "epoch": 0.43693107932379716, "grad_norm": 1.9274087862229272, "learning_rate": 1.2501984577266531e-05, "loss": 0.445, "step": 12264 }, { "epoch": 0.43696670645029123, "grad_norm": 1.0205657306655456, "learning_rate": 1.2500867325364636e-05, "loss": 0.4324, "step": 12265 }, { "epoch": 0.43700233357678536, "grad_norm": 1.5094990959205028, "learning_rate": 1.2499750040161947e-05, "loss": 0.5445, "step": 12266 }, { "epoch": 0.4370379607032795, "grad_norm": 1.1993200249246654, "learning_rate": 1.2498632721673337e-05, "loss": 0.4976, "step": 12267 }, { "epoch": 0.43707358782977357, "grad_norm": 1.4200144736380917, "learning_rate": 1.2497515369913684e-05, "loss": 0.5296, "step": 12268 }, { "epoch": 0.4371092149562677, "grad_norm": 1.23402223757775, "learning_rate": 1.249639798489787e-05, "loss": 0.491, "step": 12269 }, { "epoch": 0.43714484208276183, "grad_norm": 1.881368864693336, "learning_rate": 1.2495280566640773e-05, "loss": 0.579, "step": 12270 }, { "epoch": 0.4371804692092559, "grad_norm": 1.2238752785034874, "learning_rate": 1.249416311515727e-05, "loss": 0.4263, "step": 12271 }, { "epoch": 0.43721609633575004, "grad_norm": 1.233931851531844, "learning_rate": 1.2493045630462242e-05, "loss": 0.4973, "step": 12272 }, { "epoch": 0.43725172346224417, "grad_norm": 2.0738059629493795, "learning_rate": 1.2491928112570568e-05, "loss": 0.5706, "step": 12273 }, { "epoch": 0.43728735058873824, "grad_norm": 1.2239688718886375, "learning_rate": 1.249081056149713e-05, "loss": 0.4289, "step": 12274 }, { "epoch": 0.4373229777152324, "grad_norm": 1.8954002696926084, "learning_rate": 1.2489692977256807e-05, "loss": 0.5316, "step": 12275 }, { "epoch": 0.4373586048417265, "grad_norm": 1.1019292070504718, "learning_rate": 1.2488575359864483e-05, "loss": 0.4465, "step": 12276 }, { "epoch": 0.4373942319682206, "grad_norm": 1.2033093569966602, "learning_rate": 1.248745770933504e-05, "loss": 0.4768, "step": 12277 }, { "epoch": 0.4374298590947147, "grad_norm": 2.0601864038163074, "learning_rate": 1.2486340025683359e-05, "loss": 0.6092, "step": 12278 }, { "epoch": 0.43746548622120884, "grad_norm": 1.0541544911412928, "learning_rate": 1.248522230892432e-05, "loss": 0.4344, "step": 12279 }, { "epoch": 0.4375011133477029, "grad_norm": 1.1462360830159208, "learning_rate": 1.2484104559072812e-05, "loss": 0.4872, "step": 12280 }, { "epoch": 0.43753674047419705, "grad_norm": 1.467752334144931, "learning_rate": 1.2482986776143715e-05, "loss": 0.5504, "step": 12281 }, { "epoch": 0.4375723676006912, "grad_norm": 1.865449053487096, "learning_rate": 1.2481868960151911e-05, "loss": 0.5786, "step": 12282 }, { "epoch": 0.43760799472718526, "grad_norm": 1.492648089199511, "learning_rate": 1.248075111111229e-05, "loss": 0.5464, "step": 12283 }, { "epoch": 0.4376436218536794, "grad_norm": 1.4045508637179989, "learning_rate": 1.2479633229039734e-05, "loss": 0.4625, "step": 12284 }, { "epoch": 0.4376792489801735, "grad_norm": 1.6052520688470913, "learning_rate": 1.2478515313949127e-05, "loss": 0.6336, "step": 12285 }, { "epoch": 0.4377148761066676, "grad_norm": 1.123834813018822, "learning_rate": 1.2477397365855355e-05, "loss": 0.4567, "step": 12286 }, { "epoch": 0.4377505032331617, "grad_norm": 1.666734928652914, "learning_rate": 1.2476279384773307e-05, "loss": 0.5705, "step": 12287 }, { "epoch": 0.43778613035965586, "grad_norm": 1.4267059552255272, "learning_rate": 1.2475161370717868e-05, "loss": 0.6185, "step": 12288 }, { "epoch": 0.43782175748614993, "grad_norm": 1.4159257401497571, "learning_rate": 1.2474043323703926e-05, "loss": 0.6116, "step": 12289 }, { "epoch": 0.43785738461264406, "grad_norm": 0.93817527503412, "learning_rate": 1.2472925243746365e-05, "loss": 0.4262, "step": 12290 }, { "epoch": 0.4378930117391382, "grad_norm": 1.329075389789151, "learning_rate": 1.2471807130860079e-05, "loss": 0.5362, "step": 12291 }, { "epoch": 0.43792863886563227, "grad_norm": 1.1519264462435712, "learning_rate": 1.2470688985059954e-05, "loss": 0.4862, "step": 12292 }, { "epoch": 0.4379642659921264, "grad_norm": 1.155890181793907, "learning_rate": 1.2469570806360876e-05, "loss": 0.4846, "step": 12293 }, { "epoch": 0.43799989311862053, "grad_norm": 1.600849472766897, "learning_rate": 1.2468452594777738e-05, "loss": 0.5681, "step": 12294 }, { "epoch": 0.4380355202451146, "grad_norm": 1.0706894124709834, "learning_rate": 1.2467334350325429e-05, "loss": 0.4228, "step": 12295 }, { "epoch": 0.43807114737160874, "grad_norm": 1.1579312253055227, "learning_rate": 1.2466216073018838e-05, "loss": 0.5174, "step": 12296 }, { "epoch": 0.43810677449810287, "grad_norm": 1.4704750874832824, "learning_rate": 1.2465097762872853e-05, "loss": 0.5729, "step": 12297 }, { "epoch": 0.43814240162459694, "grad_norm": 1.1369740625986713, "learning_rate": 1.2463979419902371e-05, "loss": 0.4704, "step": 12298 }, { "epoch": 0.4381780287510911, "grad_norm": 1.1668226730146998, "learning_rate": 1.2462861044122279e-05, "loss": 0.5153, "step": 12299 }, { "epoch": 0.4382136558775852, "grad_norm": 1.0699166714252524, "learning_rate": 1.2461742635547473e-05, "loss": 0.5042, "step": 12300 }, { "epoch": 0.4382492830040793, "grad_norm": 1.1108170044982604, "learning_rate": 1.2460624194192842e-05, "loss": 0.4169, "step": 12301 }, { "epoch": 0.4382849101305734, "grad_norm": 1.9727301605098932, "learning_rate": 1.2459505720073281e-05, "loss": 0.6221, "step": 12302 }, { "epoch": 0.43832053725706754, "grad_norm": 1.1872157350052162, "learning_rate": 1.2458387213203682e-05, "loss": 0.5229, "step": 12303 }, { "epoch": 0.4383561643835616, "grad_norm": 1.131780325406587, "learning_rate": 1.2457268673598939e-05, "loss": 0.4604, "step": 12304 }, { "epoch": 0.43839179151005575, "grad_norm": 1.178695500168944, "learning_rate": 1.2456150101273945e-05, "loss": 0.4649, "step": 12305 }, { "epoch": 0.4384274186365499, "grad_norm": 2.18416062826405, "learning_rate": 1.2455031496243597e-05, "loss": 0.5731, "step": 12306 }, { "epoch": 0.43846304576304396, "grad_norm": 1.1340465495842977, "learning_rate": 1.245391285852279e-05, "loss": 0.4743, "step": 12307 }, { "epoch": 0.4384986728895381, "grad_norm": 1.2997157088192886, "learning_rate": 1.2452794188126413e-05, "loss": 0.6069, "step": 12308 }, { "epoch": 0.4385343000160322, "grad_norm": 1.1906036131160396, "learning_rate": 1.245167548506937e-05, "loss": 0.4813, "step": 12309 }, { "epoch": 0.4385699271425263, "grad_norm": 1.1026609287104467, "learning_rate": 1.2450556749366556e-05, "loss": 0.4719, "step": 12310 }, { "epoch": 0.4386055542690204, "grad_norm": 1.7391350680319952, "learning_rate": 1.2449437981032864e-05, "loss": 0.6168, "step": 12311 }, { "epoch": 0.43864118139551456, "grad_norm": 1.4253838896679152, "learning_rate": 1.2448319180083193e-05, "loss": 0.5409, "step": 12312 }, { "epoch": 0.43867680852200863, "grad_norm": 1.176307035947567, "learning_rate": 1.2447200346532443e-05, "loss": 0.4287, "step": 12313 }, { "epoch": 0.43871243564850276, "grad_norm": 1.4234827222792106, "learning_rate": 1.244608148039551e-05, "loss": 0.5714, "step": 12314 }, { "epoch": 0.4387480627749969, "grad_norm": 1.1086348445828316, "learning_rate": 1.244496258168729e-05, "loss": 0.4405, "step": 12315 }, { "epoch": 0.43878368990149097, "grad_norm": 1.2082179763558594, "learning_rate": 1.2443843650422684e-05, "loss": 0.4629, "step": 12316 }, { "epoch": 0.4388193170279851, "grad_norm": 1.0941436946938363, "learning_rate": 1.2442724686616594e-05, "loss": 0.4384, "step": 12317 }, { "epoch": 0.43885494415447923, "grad_norm": 1.102324350657692, "learning_rate": 1.2441605690283915e-05, "loss": 0.4623, "step": 12318 }, { "epoch": 0.4388905712809733, "grad_norm": 1.1189211076831864, "learning_rate": 1.244048666143955e-05, "loss": 0.4861, "step": 12319 }, { "epoch": 0.43892619840746744, "grad_norm": 1.0579315962673097, "learning_rate": 1.2439367600098402e-05, "loss": 0.4454, "step": 12320 }, { "epoch": 0.43896182553396157, "grad_norm": 1.1915216668381396, "learning_rate": 1.2438248506275366e-05, "loss": 0.4933, "step": 12321 }, { "epoch": 0.43899745266045564, "grad_norm": 3.9797524402950484, "learning_rate": 1.2437129379985349e-05, "loss": 0.6022, "step": 12322 }, { "epoch": 0.4390330797869498, "grad_norm": 1.671307123705258, "learning_rate": 1.243601022124325e-05, "loss": 0.5388, "step": 12323 }, { "epoch": 0.4390687069134439, "grad_norm": 1.1594563039425887, "learning_rate": 1.2434891030063971e-05, "loss": 0.4507, "step": 12324 }, { "epoch": 0.439104334039938, "grad_norm": 1.4958815349543582, "learning_rate": 1.243377180646242e-05, "loss": 0.5614, "step": 12325 }, { "epoch": 0.4391399611664321, "grad_norm": 1.007154693268954, "learning_rate": 1.2432652550453494e-05, "loss": 0.4542, "step": 12326 }, { "epoch": 0.43917558829292624, "grad_norm": 1.191827535663703, "learning_rate": 1.2431533262052099e-05, "loss": 0.425, "step": 12327 }, { "epoch": 0.4392112154194203, "grad_norm": 1.1157466555015563, "learning_rate": 1.2430413941273138e-05, "loss": 0.4699, "step": 12328 }, { "epoch": 0.43924684254591445, "grad_norm": 2.486951550173959, "learning_rate": 1.242929458813152e-05, "loss": 0.5779, "step": 12329 }, { "epoch": 0.4392824696724086, "grad_norm": 1.5456476383773217, "learning_rate": 1.2428175202642144e-05, "loss": 0.6025, "step": 12330 }, { "epoch": 0.43931809679890266, "grad_norm": 1.7330358264434356, "learning_rate": 1.2427055784819917e-05, "loss": 0.584, "step": 12331 }, { "epoch": 0.4393537239253968, "grad_norm": 1.1297172075997985, "learning_rate": 1.2425936334679745e-05, "loss": 0.4542, "step": 12332 }, { "epoch": 0.4393893510518909, "grad_norm": 1.0548672341883802, "learning_rate": 1.2424816852236539e-05, "loss": 0.4404, "step": 12333 }, { "epoch": 0.439424978178385, "grad_norm": 1.2136436527248444, "learning_rate": 1.24236973375052e-05, "loss": 0.4332, "step": 12334 }, { "epoch": 0.4394606053048791, "grad_norm": 1.2164346996647952, "learning_rate": 1.2422577790500634e-05, "loss": 0.4646, "step": 12335 }, { "epoch": 0.43949623243137326, "grad_norm": 1.692187640856341, "learning_rate": 1.2421458211237755e-05, "loss": 0.6237, "step": 12336 }, { "epoch": 0.43953185955786733, "grad_norm": 2.257484274228729, "learning_rate": 1.2420338599731467e-05, "loss": 0.5544, "step": 12337 }, { "epoch": 0.43956748668436146, "grad_norm": 1.0852703043376768, "learning_rate": 1.2419218955996677e-05, "loss": 0.4655, "step": 12338 }, { "epoch": 0.4396031138108556, "grad_norm": 1.6634964849237048, "learning_rate": 1.24180992800483e-05, "loss": 0.5301, "step": 12339 }, { "epoch": 0.4396387409373497, "grad_norm": 1.2507906390781525, "learning_rate": 1.2416979571901238e-05, "loss": 0.5125, "step": 12340 }, { "epoch": 0.4396743680638438, "grad_norm": 1.0728315725944502, "learning_rate": 1.2415859831570405e-05, "loss": 0.3955, "step": 12341 }, { "epoch": 0.43970999519033793, "grad_norm": 1.1766992755802097, "learning_rate": 1.2414740059070707e-05, "loss": 0.4651, "step": 12342 }, { "epoch": 0.43974562231683206, "grad_norm": 1.1300447756310583, "learning_rate": 1.2413620254417058e-05, "loss": 0.467, "step": 12343 }, { "epoch": 0.43978124944332614, "grad_norm": 1.1783073292336164, "learning_rate": 1.2412500417624367e-05, "loss": 0.4441, "step": 12344 }, { "epoch": 0.43981687656982027, "grad_norm": 1.5118238818558323, "learning_rate": 1.241138054870755e-05, "loss": 0.5562, "step": 12345 }, { "epoch": 0.4398525036963144, "grad_norm": 1.8175015028285462, "learning_rate": 1.2410260647681513e-05, "loss": 0.5399, "step": 12346 }, { "epoch": 0.4398881308228085, "grad_norm": 1.394067720636834, "learning_rate": 1.2409140714561172e-05, "loss": 0.5916, "step": 12347 }, { "epoch": 0.4399237579493026, "grad_norm": 1.3778953463471593, "learning_rate": 1.240802074936144e-05, "loss": 0.5875, "step": 12348 }, { "epoch": 0.43995938507579674, "grad_norm": 1.5128530014076103, "learning_rate": 1.2406900752097228e-05, "loss": 0.5864, "step": 12349 }, { "epoch": 0.4399950122022908, "grad_norm": 1.210805853904518, "learning_rate": 1.2405780722783447e-05, "loss": 0.4789, "step": 12350 }, { "epoch": 0.44003063932878494, "grad_norm": 1.0684714796642636, "learning_rate": 1.2404660661435017e-05, "loss": 0.4832, "step": 12351 }, { "epoch": 0.4400662664552791, "grad_norm": 1.1065297318182195, "learning_rate": 1.240354056806685e-05, "loss": 0.4447, "step": 12352 }, { "epoch": 0.44010189358177315, "grad_norm": 1.5917320163653468, "learning_rate": 1.2402420442693858e-05, "loss": 0.5288, "step": 12353 }, { "epoch": 0.4401375207082673, "grad_norm": 2.159526345920442, "learning_rate": 1.2401300285330962e-05, "loss": 0.5547, "step": 12354 }, { "epoch": 0.4401731478347614, "grad_norm": 1.6129660293127324, "learning_rate": 1.2400180095993072e-05, "loss": 0.5987, "step": 12355 }, { "epoch": 0.4402087749612555, "grad_norm": 1.739077416457283, "learning_rate": 1.2399059874695104e-05, "loss": 0.5424, "step": 12356 }, { "epoch": 0.4402444020877496, "grad_norm": 1.1523713585247712, "learning_rate": 1.239793962145198e-05, "loss": 0.4991, "step": 12357 }, { "epoch": 0.44028002921424375, "grad_norm": 1.572536190457961, "learning_rate": 1.2396819336278614e-05, "loss": 0.6006, "step": 12358 }, { "epoch": 0.4403156563407378, "grad_norm": 1.4449032552968628, "learning_rate": 1.2395699019189924e-05, "loss": 0.5309, "step": 12359 }, { "epoch": 0.44035128346723196, "grad_norm": 1.4368122694025505, "learning_rate": 1.2394578670200825e-05, "loss": 0.5133, "step": 12360 }, { "epoch": 0.4403869105937261, "grad_norm": 2.949642529685701, "learning_rate": 1.2393458289326237e-05, "loss": 0.5157, "step": 12361 }, { "epoch": 0.44042253772022016, "grad_norm": 1.4251650644907694, "learning_rate": 1.2392337876581081e-05, "loss": 0.5251, "step": 12362 }, { "epoch": 0.4404581648467143, "grad_norm": 1.2281320207410131, "learning_rate": 1.2391217431980273e-05, "loss": 0.5289, "step": 12363 }, { "epoch": 0.4404937919732084, "grad_norm": 1.3558968777003457, "learning_rate": 1.2390096955538733e-05, "loss": 0.5631, "step": 12364 }, { "epoch": 0.4405294190997025, "grad_norm": 1.6326490599359589, "learning_rate": 1.2388976447271384e-05, "loss": 0.5284, "step": 12365 }, { "epoch": 0.44056504622619663, "grad_norm": 1.16396014020109, "learning_rate": 1.2387855907193141e-05, "loss": 0.4171, "step": 12366 }, { "epoch": 0.44060067335269076, "grad_norm": 1.1692669764434627, "learning_rate": 1.2386735335318926e-05, "loss": 0.4544, "step": 12367 }, { "epoch": 0.44063630047918484, "grad_norm": 1.0998634333687864, "learning_rate": 1.2385614731663667e-05, "loss": 0.4745, "step": 12368 }, { "epoch": 0.44067192760567897, "grad_norm": 1.6710755020427464, "learning_rate": 1.2384494096242277e-05, "loss": 0.5059, "step": 12369 }, { "epoch": 0.4407075547321731, "grad_norm": 1.0956355466049545, "learning_rate": 1.2383373429069683e-05, "loss": 0.4786, "step": 12370 }, { "epoch": 0.4407431818586672, "grad_norm": 1.5797546010139942, "learning_rate": 1.2382252730160804e-05, "loss": 0.5397, "step": 12371 }, { "epoch": 0.4407788089851613, "grad_norm": 2.3847713591318587, "learning_rate": 1.2381131999530564e-05, "loss": 0.5495, "step": 12372 }, { "epoch": 0.44081443611165544, "grad_norm": 1.8814924390811754, "learning_rate": 1.238001123719389e-05, "loss": 0.6004, "step": 12373 }, { "epoch": 0.4408500632381495, "grad_norm": 1.273883442528087, "learning_rate": 1.2378890443165704e-05, "loss": 0.4638, "step": 12374 }, { "epoch": 0.44088569036464365, "grad_norm": 1.6830005944184696, "learning_rate": 1.2377769617460927e-05, "loss": 0.5899, "step": 12375 }, { "epoch": 0.4409213174911378, "grad_norm": 1.2038317258029736, "learning_rate": 1.2376648760094482e-05, "loss": 0.4468, "step": 12376 }, { "epoch": 0.44095694461763185, "grad_norm": 1.2627013998134307, "learning_rate": 1.2375527871081303e-05, "loss": 0.495, "step": 12377 }, { "epoch": 0.440992571744126, "grad_norm": 1.200319196438352, "learning_rate": 1.2374406950436308e-05, "loss": 0.5708, "step": 12378 }, { "epoch": 0.4410281988706201, "grad_norm": 1.8520565669971683, "learning_rate": 1.2373285998174424e-05, "loss": 0.5402, "step": 12379 }, { "epoch": 0.4410638259971142, "grad_norm": 1.129869851400731, "learning_rate": 1.2372165014310578e-05, "loss": 0.4687, "step": 12380 }, { "epoch": 0.4410994531236083, "grad_norm": 1.8069167990371204, "learning_rate": 1.2371043998859698e-05, "loss": 0.5609, "step": 12381 }, { "epoch": 0.44113508025010245, "grad_norm": 1.0955776101230335, "learning_rate": 1.236992295183671e-05, "loss": 0.4577, "step": 12382 }, { "epoch": 0.4411707073765965, "grad_norm": 1.454912199001718, "learning_rate": 1.236880187325654e-05, "loss": 0.5641, "step": 12383 }, { "epoch": 0.44120633450309066, "grad_norm": 1.9941603239649102, "learning_rate": 1.2367680763134117e-05, "loss": 0.6485, "step": 12384 }, { "epoch": 0.4412419616295848, "grad_norm": 1.2435664927491514, "learning_rate": 1.2366559621484373e-05, "loss": 0.4577, "step": 12385 }, { "epoch": 0.44127758875607886, "grad_norm": 1.2416982063582278, "learning_rate": 1.2365438448322234e-05, "loss": 0.4743, "step": 12386 }, { "epoch": 0.441313215882573, "grad_norm": 1.0551400988538806, "learning_rate": 1.2364317243662623e-05, "loss": 0.4076, "step": 12387 }, { "epoch": 0.4413488430090671, "grad_norm": 1.8005181591886823, "learning_rate": 1.236319600752048e-05, "loss": 0.5477, "step": 12388 }, { "epoch": 0.4413844701355612, "grad_norm": 1.175667141417619, "learning_rate": 1.236207473991073e-05, "loss": 0.4515, "step": 12389 }, { "epoch": 0.44142009726205533, "grad_norm": 1.0239900330983644, "learning_rate": 1.2360953440848303e-05, "loss": 0.4496, "step": 12390 }, { "epoch": 0.44145572438854946, "grad_norm": 1.1985528793809626, "learning_rate": 1.2359832110348126e-05, "loss": 0.4883, "step": 12391 }, { "epoch": 0.44149135151504354, "grad_norm": 1.090207912433675, "learning_rate": 1.2358710748425142e-05, "loss": 0.4765, "step": 12392 }, { "epoch": 0.44152697864153767, "grad_norm": 1.4689371031683305, "learning_rate": 1.2357589355094275e-05, "loss": 0.4386, "step": 12393 }, { "epoch": 0.4415626057680318, "grad_norm": 1.4727597894941207, "learning_rate": 1.2356467930370455e-05, "loss": 0.5581, "step": 12394 }, { "epoch": 0.4415982328945259, "grad_norm": 1.2617787842261967, "learning_rate": 1.2355346474268619e-05, "loss": 0.4653, "step": 12395 }, { "epoch": 0.44163386002102, "grad_norm": 1.1660305821344121, "learning_rate": 1.2354224986803697e-05, "loss": 0.4208, "step": 12396 }, { "epoch": 0.44166948714751414, "grad_norm": 1.1473101327135993, "learning_rate": 1.2353103467990627e-05, "loss": 0.4903, "step": 12397 }, { "epoch": 0.4417051142740082, "grad_norm": 1.2191081999515125, "learning_rate": 1.2351981917844335e-05, "loss": 0.5158, "step": 12398 }, { "epoch": 0.44174074140050235, "grad_norm": 0.9995386974043969, "learning_rate": 1.2350860336379765e-05, "loss": 0.4412, "step": 12399 }, { "epoch": 0.4417763685269965, "grad_norm": 1.6125067622612546, "learning_rate": 1.2349738723611845e-05, "loss": 0.6064, "step": 12400 }, { "epoch": 0.44181199565349055, "grad_norm": 1.8020874812575478, "learning_rate": 1.2348617079555507e-05, "loss": 0.6088, "step": 12401 }, { "epoch": 0.4418476227799847, "grad_norm": 1.1118805261222442, "learning_rate": 1.2347495404225693e-05, "loss": 0.4535, "step": 12402 }, { "epoch": 0.4418832499064788, "grad_norm": 1.180615772239357, "learning_rate": 1.2346373697637339e-05, "loss": 0.4495, "step": 12403 }, { "epoch": 0.4419188770329729, "grad_norm": 1.857476776421717, "learning_rate": 1.234525195980538e-05, "loss": 0.4891, "step": 12404 }, { "epoch": 0.441954504159467, "grad_norm": 1.1719134914355116, "learning_rate": 1.234413019074475e-05, "loss": 0.4877, "step": 12405 }, { "epoch": 0.44199013128596115, "grad_norm": 1.9642578812875207, "learning_rate": 1.2343008390470388e-05, "loss": 0.56, "step": 12406 }, { "epoch": 0.4420257584124552, "grad_norm": 1.129400101979305, "learning_rate": 1.2341886558997231e-05, "loss": 0.4364, "step": 12407 }, { "epoch": 0.44206138553894936, "grad_norm": 1.4432419249995592, "learning_rate": 1.234076469634022e-05, "loss": 0.5295, "step": 12408 }, { "epoch": 0.4420970126654435, "grad_norm": 1.5635300159214824, "learning_rate": 1.233964280251429e-05, "loss": 0.6142, "step": 12409 }, { "epoch": 0.44213263979193757, "grad_norm": 1.1151603817800442, "learning_rate": 1.2338520877534379e-05, "loss": 0.4345, "step": 12410 }, { "epoch": 0.4421682669184317, "grad_norm": 1.6633977433601645, "learning_rate": 1.2337398921415431e-05, "loss": 0.5865, "step": 12411 }, { "epoch": 0.4422038940449258, "grad_norm": 1.689461841561227, "learning_rate": 1.2336276934172382e-05, "loss": 0.5839, "step": 12412 }, { "epoch": 0.4422395211714199, "grad_norm": 1.1713646814890626, "learning_rate": 1.2335154915820167e-05, "loss": 0.4563, "step": 12413 }, { "epoch": 0.44227514829791403, "grad_norm": 0.9072254154800057, "learning_rate": 1.2334032866373737e-05, "loss": 0.4197, "step": 12414 }, { "epoch": 0.44231077542440816, "grad_norm": 1.0777665997962902, "learning_rate": 1.233291078584803e-05, "loss": 0.4788, "step": 12415 }, { "epoch": 0.44234640255090224, "grad_norm": 1.0317315712027768, "learning_rate": 1.2331788674257982e-05, "loss": 0.4555, "step": 12416 }, { "epoch": 0.44238202967739637, "grad_norm": 1.552678500899287, "learning_rate": 1.2330666531618538e-05, "loss": 0.5349, "step": 12417 }, { "epoch": 0.4424176568038905, "grad_norm": 1.052793403953955, "learning_rate": 1.2329544357944643e-05, "loss": 0.4789, "step": 12418 }, { "epoch": 0.4424532839303846, "grad_norm": 1.4221360001684933, "learning_rate": 1.2328422153251234e-05, "loss": 0.5645, "step": 12419 }, { "epoch": 0.4424889110568787, "grad_norm": 1.9095883268012717, "learning_rate": 1.2327299917553257e-05, "loss": 0.6424, "step": 12420 }, { "epoch": 0.44252453818337284, "grad_norm": 1.585544019973784, "learning_rate": 1.2326177650865656e-05, "loss": 0.5919, "step": 12421 }, { "epoch": 0.4425601653098669, "grad_norm": 0.9917523278620815, "learning_rate": 1.2325055353203373e-05, "loss": 0.4322, "step": 12422 }, { "epoch": 0.44259579243636105, "grad_norm": 1.3070242482624042, "learning_rate": 1.2323933024581355e-05, "loss": 0.4445, "step": 12423 }, { "epoch": 0.4426314195628552, "grad_norm": 1.804484752245513, "learning_rate": 1.232281066501454e-05, "loss": 0.5803, "step": 12424 }, { "epoch": 0.44266704668934925, "grad_norm": 1.713672043823455, "learning_rate": 1.2321688274517877e-05, "loss": 0.6436, "step": 12425 }, { "epoch": 0.4427026738158434, "grad_norm": 1.1480419830246935, "learning_rate": 1.2320565853106317e-05, "loss": 0.4308, "step": 12426 }, { "epoch": 0.4427383009423375, "grad_norm": 1.312883278341974, "learning_rate": 1.2319443400794796e-05, "loss": 0.5283, "step": 12427 }, { "epoch": 0.4427739280688316, "grad_norm": 1.658881490721841, "learning_rate": 1.2318320917598268e-05, "loss": 0.5323, "step": 12428 }, { "epoch": 0.4428095551953257, "grad_norm": 1.2233100942072141, "learning_rate": 1.2317198403531676e-05, "loss": 0.4968, "step": 12429 }, { "epoch": 0.44284518232181985, "grad_norm": 1.495130861716882, "learning_rate": 1.2316075858609969e-05, "loss": 0.5926, "step": 12430 }, { "epoch": 0.44288080944831393, "grad_norm": 1.015343509442576, "learning_rate": 1.231495328284809e-05, "loss": 0.4218, "step": 12431 }, { "epoch": 0.44291643657480806, "grad_norm": 1.4832686554847836, "learning_rate": 1.2313830676260989e-05, "loss": 0.472, "step": 12432 }, { "epoch": 0.4429520637013022, "grad_norm": 1.7411976402381424, "learning_rate": 1.2312708038863618e-05, "loss": 0.5385, "step": 12433 }, { "epoch": 0.44298769082779627, "grad_norm": 1.1638295142977804, "learning_rate": 1.2311585370670925e-05, "loss": 0.5194, "step": 12434 }, { "epoch": 0.4430233179542904, "grad_norm": 1.4877287361930789, "learning_rate": 1.2310462671697853e-05, "loss": 0.5925, "step": 12435 }, { "epoch": 0.4430589450807845, "grad_norm": 1.5314801676449876, "learning_rate": 1.2309339941959358e-05, "loss": 0.5501, "step": 12436 }, { "epoch": 0.4430945722072786, "grad_norm": 1.1526982498472886, "learning_rate": 1.2308217181470385e-05, "loss": 0.4512, "step": 12437 }, { "epoch": 0.44313019933377273, "grad_norm": 1.1427586802197505, "learning_rate": 1.230709439024589e-05, "loss": 0.4991, "step": 12438 }, { "epoch": 0.44316582646026687, "grad_norm": 1.3092683187234773, "learning_rate": 1.230597156830082e-05, "loss": 0.5504, "step": 12439 }, { "epoch": 0.44320145358676094, "grad_norm": 1.1061087318793417, "learning_rate": 1.2304848715650123e-05, "loss": 0.5011, "step": 12440 }, { "epoch": 0.44323708071325507, "grad_norm": 1.6469384291313587, "learning_rate": 1.2303725832308759e-05, "loss": 0.5485, "step": 12441 }, { "epoch": 0.4432727078397492, "grad_norm": 1.763289855409123, "learning_rate": 1.2302602918291674e-05, "loss": 0.5297, "step": 12442 }, { "epoch": 0.4433083349662433, "grad_norm": 1.10656711187388, "learning_rate": 1.2301479973613822e-05, "loss": 0.456, "step": 12443 }, { "epoch": 0.4433439620927374, "grad_norm": 1.0498954967209213, "learning_rate": 1.2300356998290155e-05, "loss": 0.4878, "step": 12444 }, { "epoch": 0.44337958921923154, "grad_norm": 1.4984588461806858, "learning_rate": 1.2299233992335628e-05, "loss": 0.5446, "step": 12445 }, { "epoch": 0.4434152163457256, "grad_norm": 1.0348422224788456, "learning_rate": 1.229811095576519e-05, "loss": 0.4315, "step": 12446 }, { "epoch": 0.44345084347221975, "grad_norm": 1.1770587936059573, "learning_rate": 1.2296987888593802e-05, "loss": 0.4975, "step": 12447 }, { "epoch": 0.4434864705987139, "grad_norm": 1.3815861292879779, "learning_rate": 1.2295864790836411e-05, "loss": 0.5308, "step": 12448 }, { "epoch": 0.44352209772520795, "grad_norm": 1.7571443681651497, "learning_rate": 1.2294741662507978e-05, "loss": 0.5839, "step": 12449 }, { "epoch": 0.4435577248517021, "grad_norm": 1.0402252348071483, "learning_rate": 1.2293618503623459e-05, "loss": 0.4576, "step": 12450 }, { "epoch": 0.4435933519781962, "grad_norm": 1.4911276411761591, "learning_rate": 1.2292495314197801e-05, "loss": 0.5159, "step": 12451 }, { "epoch": 0.4436289791046903, "grad_norm": 1.6890305020096739, "learning_rate": 1.2291372094245969e-05, "loss": 0.5058, "step": 12452 }, { "epoch": 0.4436646062311844, "grad_norm": 1.5570252663870336, "learning_rate": 1.2290248843782915e-05, "loss": 0.5167, "step": 12453 }, { "epoch": 0.44370023335767855, "grad_norm": 2.1049734641829843, "learning_rate": 1.2289125562823594e-05, "loss": 0.5615, "step": 12454 }, { "epoch": 0.44373586048417263, "grad_norm": 1.1667731557248957, "learning_rate": 1.2288002251382969e-05, "loss": 0.4842, "step": 12455 }, { "epoch": 0.44377148761066676, "grad_norm": 1.1678983986301927, "learning_rate": 1.2286878909475997e-05, "loss": 0.3704, "step": 12456 }, { "epoch": 0.4438071147371609, "grad_norm": 1.4035786539658477, "learning_rate": 1.2285755537117631e-05, "loss": 0.5867, "step": 12457 }, { "epoch": 0.44384274186365497, "grad_norm": 1.118919524632767, "learning_rate": 1.2284632134322832e-05, "loss": 0.4204, "step": 12458 }, { "epoch": 0.4438783689901491, "grad_norm": 1.2412829178862421, "learning_rate": 1.2283508701106559e-05, "loss": 0.4921, "step": 12459 }, { "epoch": 0.44391399611664323, "grad_norm": 1.693188800370455, "learning_rate": 1.2282385237483771e-05, "loss": 0.5512, "step": 12460 }, { "epoch": 0.4439496232431373, "grad_norm": 1.403409053477964, "learning_rate": 1.2281261743469433e-05, "loss": 0.4649, "step": 12461 }, { "epoch": 0.44398525036963143, "grad_norm": 1.1039371468467165, "learning_rate": 1.2280138219078494e-05, "loss": 0.4487, "step": 12462 }, { "epoch": 0.44402087749612557, "grad_norm": 1.733309149778503, "learning_rate": 1.2279014664325925e-05, "loss": 0.6747, "step": 12463 }, { "epoch": 0.44405650462261964, "grad_norm": 1.1782656577387958, "learning_rate": 1.2277891079226683e-05, "loss": 0.4839, "step": 12464 }, { "epoch": 0.44409213174911377, "grad_norm": 1.5444321463860327, "learning_rate": 1.2276767463795727e-05, "loss": 0.5932, "step": 12465 }, { "epoch": 0.4441277588756079, "grad_norm": 1.8161709073844974, "learning_rate": 1.2275643818048023e-05, "loss": 0.5043, "step": 12466 }, { "epoch": 0.444163386002102, "grad_norm": 1.497740595920771, "learning_rate": 1.227452014199853e-05, "loss": 0.5284, "step": 12467 }, { "epoch": 0.4441990131285961, "grad_norm": 1.3805860656814954, "learning_rate": 1.2273396435662212e-05, "loss": 0.4948, "step": 12468 }, { "epoch": 0.44423464025509024, "grad_norm": 1.7519600162133284, "learning_rate": 1.227227269905403e-05, "loss": 0.5475, "step": 12469 }, { "epoch": 0.4442702673815843, "grad_norm": 1.789259461042386, "learning_rate": 1.227114893218895e-05, "loss": 0.5634, "step": 12470 }, { "epoch": 0.44430589450807845, "grad_norm": 1.6573513459855214, "learning_rate": 1.227002513508193e-05, "loss": 0.5626, "step": 12471 }, { "epoch": 0.4443415216345726, "grad_norm": 1.7126173173286483, "learning_rate": 1.2268901307747947e-05, "loss": 0.6167, "step": 12472 }, { "epoch": 0.44437714876106665, "grad_norm": 3.346287013347141, "learning_rate": 1.226777745020195e-05, "loss": 0.598, "step": 12473 }, { "epoch": 0.4444127758875608, "grad_norm": 1.0296504225347345, "learning_rate": 1.2266653562458914e-05, "loss": 0.4542, "step": 12474 }, { "epoch": 0.4444484030140549, "grad_norm": 1.796227405167835, "learning_rate": 1.2265529644533804e-05, "loss": 0.5248, "step": 12475 }, { "epoch": 0.444484030140549, "grad_norm": 1.6992829872251662, "learning_rate": 1.226440569644158e-05, "loss": 0.5548, "step": 12476 }, { "epoch": 0.4445196572670431, "grad_norm": 1.4196055257533364, "learning_rate": 1.2263281718197213e-05, "loss": 0.4766, "step": 12477 }, { "epoch": 0.44455528439353725, "grad_norm": 1.128383166460526, "learning_rate": 1.2262157709815668e-05, "loss": 0.4891, "step": 12478 }, { "epoch": 0.44459091152003133, "grad_norm": 3.1214870030803126, "learning_rate": 1.226103367131191e-05, "loss": 0.5805, "step": 12479 }, { "epoch": 0.44462653864652546, "grad_norm": 1.5009418089426274, "learning_rate": 1.225990960270091e-05, "loss": 0.5052, "step": 12480 }, { "epoch": 0.4446621657730196, "grad_norm": 0.9577541505617909, "learning_rate": 1.2258785503997635e-05, "loss": 0.4774, "step": 12481 }, { "epoch": 0.44469779289951367, "grad_norm": 1.1609957115259142, "learning_rate": 1.2257661375217052e-05, "loss": 0.4802, "step": 12482 }, { "epoch": 0.4447334200260078, "grad_norm": 1.5185860852969988, "learning_rate": 1.2256537216374127e-05, "loss": 0.4712, "step": 12483 }, { "epoch": 0.44476904715250193, "grad_norm": 2.1649751638869574, "learning_rate": 1.225541302748383e-05, "loss": 0.6225, "step": 12484 }, { "epoch": 0.444804674278996, "grad_norm": 1.2538974745494946, "learning_rate": 1.2254288808561137e-05, "loss": 0.4199, "step": 12485 }, { "epoch": 0.44484030140549013, "grad_norm": 1.2090971932768488, "learning_rate": 1.2253164559621014e-05, "loss": 0.4693, "step": 12486 }, { "epoch": 0.44487592853198427, "grad_norm": 2.575155313195357, "learning_rate": 1.2252040280678426e-05, "loss": 0.5516, "step": 12487 }, { "epoch": 0.44491155565847834, "grad_norm": 1.5220579187295706, "learning_rate": 1.225091597174835e-05, "loss": 0.562, "step": 12488 }, { "epoch": 0.4449471827849725, "grad_norm": 1.781305143300197, "learning_rate": 1.2249791632845753e-05, "loss": 0.4895, "step": 12489 }, { "epoch": 0.4449828099114666, "grad_norm": 1.2108076745385827, "learning_rate": 1.2248667263985609e-05, "loss": 0.4824, "step": 12490 }, { "epoch": 0.4450184370379607, "grad_norm": 2.927579145115329, "learning_rate": 1.2247542865182887e-05, "loss": 0.6024, "step": 12491 }, { "epoch": 0.4450540641644548, "grad_norm": 1.5460846241798742, "learning_rate": 1.2246418436452563e-05, "loss": 0.5686, "step": 12492 }, { "epoch": 0.44508969129094894, "grad_norm": 1.0657488086113671, "learning_rate": 1.2245293977809607e-05, "loss": 0.4316, "step": 12493 }, { "epoch": 0.445125318417443, "grad_norm": 1.1799927395748144, "learning_rate": 1.2244169489268989e-05, "loss": 0.4811, "step": 12494 }, { "epoch": 0.44516094554393715, "grad_norm": 1.2577913124994398, "learning_rate": 1.2243044970845688e-05, "loss": 0.4464, "step": 12495 }, { "epoch": 0.4451965726704313, "grad_norm": 1.5625003837390816, "learning_rate": 1.2241920422554676e-05, "loss": 0.5796, "step": 12496 }, { "epoch": 0.44523219979692535, "grad_norm": 1.1899823525041904, "learning_rate": 1.2240795844410926e-05, "loss": 0.5032, "step": 12497 }, { "epoch": 0.4452678269234195, "grad_norm": 1.0523639743804172, "learning_rate": 1.2239671236429413e-05, "loss": 0.4757, "step": 12498 }, { "epoch": 0.4453034540499136, "grad_norm": 1.1159707242261938, "learning_rate": 1.2238546598625112e-05, "loss": 0.4838, "step": 12499 }, { "epoch": 0.4453390811764077, "grad_norm": 2.1538145074048085, "learning_rate": 1.2237421931013001e-05, "loss": 0.5445, "step": 12500 }, { "epoch": 0.4453747083029018, "grad_norm": 1.2836487516881538, "learning_rate": 1.223629723360805e-05, "loss": 0.4158, "step": 12501 }, { "epoch": 0.44541033542939595, "grad_norm": 1.4564780201865735, "learning_rate": 1.2235172506425239e-05, "loss": 0.5382, "step": 12502 }, { "epoch": 0.44544596255589003, "grad_norm": 1.1164995305866017, "learning_rate": 1.2234047749479543e-05, "loss": 0.4496, "step": 12503 }, { "epoch": 0.44548158968238416, "grad_norm": 1.0829321755781365, "learning_rate": 1.2232922962785942e-05, "loss": 0.4898, "step": 12504 }, { "epoch": 0.4455172168088783, "grad_norm": 1.3121618885265387, "learning_rate": 1.223179814635941e-05, "loss": 0.5739, "step": 12505 }, { "epoch": 0.44555284393537237, "grad_norm": 1.0633573359594897, "learning_rate": 1.2230673300214925e-05, "loss": 0.5026, "step": 12506 }, { "epoch": 0.4455884710618665, "grad_norm": 1.0757251696404018, "learning_rate": 1.2229548424367466e-05, "loss": 0.4978, "step": 12507 }, { "epoch": 0.44562409818836063, "grad_norm": 1.1465897728655194, "learning_rate": 1.2228423518832011e-05, "loss": 0.438, "step": 12508 }, { "epoch": 0.4456597253148547, "grad_norm": 1.1251322130848118, "learning_rate": 1.2227298583623541e-05, "loss": 0.4129, "step": 12509 }, { "epoch": 0.44569535244134884, "grad_norm": 1.4610762322219204, "learning_rate": 1.2226173618757032e-05, "loss": 0.6479, "step": 12510 }, { "epoch": 0.44573097956784297, "grad_norm": 1.4035272686503397, "learning_rate": 1.2225048624247467e-05, "loss": 0.576, "step": 12511 }, { "epoch": 0.44576660669433704, "grad_norm": 1.1844316140142195, "learning_rate": 1.2223923600109822e-05, "loss": 0.446, "step": 12512 }, { "epoch": 0.4458022338208312, "grad_norm": 1.0164423565875953, "learning_rate": 1.2222798546359082e-05, "loss": 0.4916, "step": 12513 }, { "epoch": 0.4458378609473253, "grad_norm": 1.1334264537583552, "learning_rate": 1.2221673463010225e-05, "loss": 0.4788, "step": 12514 }, { "epoch": 0.4458734880738194, "grad_norm": 1.9666368891617978, "learning_rate": 1.2220548350078232e-05, "loss": 0.5465, "step": 12515 }, { "epoch": 0.4459091152003135, "grad_norm": 1.1008796559158003, "learning_rate": 1.2219423207578087e-05, "loss": 0.4486, "step": 12516 }, { "epoch": 0.44594474232680764, "grad_norm": 1.1338428060918824, "learning_rate": 1.2218298035524768e-05, "loss": 0.4702, "step": 12517 }, { "epoch": 0.4459803694533017, "grad_norm": 1.6269134295612169, "learning_rate": 1.2217172833933261e-05, "loss": 0.4774, "step": 12518 }, { "epoch": 0.44601599657979585, "grad_norm": 1.2163773528515704, "learning_rate": 1.221604760281855e-05, "loss": 0.5211, "step": 12519 }, { "epoch": 0.44605162370629, "grad_norm": 2.1108658963674545, "learning_rate": 1.2214922342195616e-05, "loss": 0.6569, "step": 12520 }, { "epoch": 0.44608725083278405, "grad_norm": 1.6044379692944126, "learning_rate": 1.2213797052079441e-05, "loss": 0.5811, "step": 12521 }, { "epoch": 0.4461228779592782, "grad_norm": 1.082638552241891, "learning_rate": 1.2212671732485012e-05, "loss": 0.4895, "step": 12522 }, { "epoch": 0.4461585050857723, "grad_norm": 1.6756805271497257, "learning_rate": 1.2211546383427315e-05, "loss": 0.5492, "step": 12523 }, { "epoch": 0.4461941322122664, "grad_norm": 1.8492134640050726, "learning_rate": 1.2210421004921329e-05, "loss": 0.5084, "step": 12524 }, { "epoch": 0.4462297593387605, "grad_norm": 1.0865858664599035, "learning_rate": 1.2209295596982043e-05, "loss": 0.4229, "step": 12525 }, { "epoch": 0.44626538646525465, "grad_norm": 1.9095014300170667, "learning_rate": 1.2208170159624442e-05, "loss": 0.6242, "step": 12526 }, { "epoch": 0.44630101359174873, "grad_norm": 2.5833579311369452, "learning_rate": 1.2207044692863512e-05, "loss": 0.6166, "step": 12527 }, { "epoch": 0.44633664071824286, "grad_norm": 1.0569642185864085, "learning_rate": 1.2205919196714237e-05, "loss": 0.4412, "step": 12528 }, { "epoch": 0.446372267844737, "grad_norm": 2.0113421294569656, "learning_rate": 1.2204793671191607e-05, "loss": 0.5254, "step": 12529 }, { "epoch": 0.4464078949712311, "grad_norm": 1.948946577230213, "learning_rate": 1.220366811631061e-05, "loss": 0.5323, "step": 12530 }, { "epoch": 0.4464435220977252, "grad_norm": 1.3912314681727826, "learning_rate": 1.2202542532086232e-05, "loss": 0.455, "step": 12531 }, { "epoch": 0.44647914922421933, "grad_norm": 1.6643998695180706, "learning_rate": 1.2201416918533462e-05, "loss": 0.6383, "step": 12532 }, { "epoch": 0.44651477635071346, "grad_norm": 1.094463750014786, "learning_rate": 1.2200291275667282e-05, "loss": 0.4519, "step": 12533 }, { "epoch": 0.44655040347720754, "grad_norm": 1.071498621220272, "learning_rate": 1.219916560350269e-05, "loss": 0.4826, "step": 12534 }, { "epoch": 0.44658603060370167, "grad_norm": 1.3929565224595026, "learning_rate": 1.219803990205467e-05, "loss": 0.5665, "step": 12535 }, { "epoch": 0.4466216577301958, "grad_norm": 1.8720139587272688, "learning_rate": 1.2196914171338209e-05, "loss": 0.5785, "step": 12536 }, { "epoch": 0.4466572848566899, "grad_norm": 1.1663205616198449, "learning_rate": 1.2195788411368305e-05, "loss": 0.4334, "step": 12537 }, { "epoch": 0.446692911983184, "grad_norm": 1.0789196586539385, "learning_rate": 1.2194662622159943e-05, "loss": 0.4928, "step": 12538 }, { "epoch": 0.44672853910967814, "grad_norm": 1.1760504063799682, "learning_rate": 1.2193536803728113e-05, "loss": 0.4531, "step": 12539 }, { "epoch": 0.4467641662361722, "grad_norm": 1.503313068685348, "learning_rate": 1.2192410956087803e-05, "loss": 0.5273, "step": 12540 }, { "epoch": 0.44679979336266634, "grad_norm": 1.2687288513077175, "learning_rate": 1.2191285079254014e-05, "loss": 0.5441, "step": 12541 }, { "epoch": 0.4468354204891605, "grad_norm": 1.6779511426563367, "learning_rate": 1.2190159173241731e-05, "loss": 0.6453, "step": 12542 }, { "epoch": 0.44687104761565455, "grad_norm": 1.0243646192619584, "learning_rate": 1.218903323806595e-05, "loss": 0.4583, "step": 12543 }, { "epoch": 0.4469066747421487, "grad_norm": 1.6977698266721961, "learning_rate": 1.2187907273741656e-05, "loss": 0.622, "step": 12544 }, { "epoch": 0.4469423018686428, "grad_norm": 1.0102996328822529, "learning_rate": 1.2186781280283854e-05, "loss": 0.4519, "step": 12545 }, { "epoch": 0.4469779289951369, "grad_norm": 2.1785348492673533, "learning_rate": 1.2185655257707527e-05, "loss": 0.5679, "step": 12546 }, { "epoch": 0.447013556121631, "grad_norm": 1.5231234017462063, "learning_rate": 1.2184529206027674e-05, "loss": 0.5917, "step": 12547 }, { "epoch": 0.44704918324812515, "grad_norm": 1.5805359811609798, "learning_rate": 1.2183403125259287e-05, "loss": 0.5022, "step": 12548 }, { "epoch": 0.4470848103746192, "grad_norm": 1.1098892009209578, "learning_rate": 1.218227701541736e-05, "loss": 0.5458, "step": 12549 }, { "epoch": 0.44712043750111335, "grad_norm": 1.1986754775212636, "learning_rate": 1.2181150876516891e-05, "loss": 0.4495, "step": 12550 }, { "epoch": 0.4471560646276075, "grad_norm": 1.8252465234826638, "learning_rate": 1.2180024708572873e-05, "loss": 0.574, "step": 12551 }, { "epoch": 0.44719169175410156, "grad_norm": 1.7929534487832794, "learning_rate": 1.2178898511600301e-05, "loss": 0.6447, "step": 12552 }, { "epoch": 0.4472273188805957, "grad_norm": 1.2120344632370401, "learning_rate": 1.2177772285614176e-05, "loss": 0.5119, "step": 12553 }, { "epoch": 0.4472629460070898, "grad_norm": 1.2165261996974457, "learning_rate": 1.217664603062949e-05, "loss": 0.4517, "step": 12554 }, { "epoch": 0.4472985731335839, "grad_norm": 1.0590156119879062, "learning_rate": 1.2175519746661235e-05, "loss": 0.4359, "step": 12555 }, { "epoch": 0.44733420026007803, "grad_norm": 1.1377158834267334, "learning_rate": 1.2174393433724419e-05, "loss": 0.4691, "step": 12556 }, { "epoch": 0.44736982738657216, "grad_norm": 1.4780533451632083, "learning_rate": 1.2173267091834035e-05, "loss": 0.5275, "step": 12557 }, { "epoch": 0.44740545451306624, "grad_norm": 1.467891037669561, "learning_rate": 1.217214072100508e-05, "loss": 0.5818, "step": 12558 }, { "epoch": 0.44744108163956037, "grad_norm": 1.1575571970159888, "learning_rate": 1.2171014321252552e-05, "loss": 0.4941, "step": 12559 }, { "epoch": 0.4474767087660545, "grad_norm": 1.1562319847612195, "learning_rate": 1.2169887892591452e-05, "loss": 0.4854, "step": 12560 }, { "epoch": 0.4475123358925486, "grad_norm": 1.1377884404213217, "learning_rate": 1.2168761435036778e-05, "loss": 0.4614, "step": 12561 }, { "epoch": 0.4475479630190427, "grad_norm": 1.1889466596513512, "learning_rate": 1.2167634948603529e-05, "loss": 0.5069, "step": 12562 }, { "epoch": 0.44758359014553684, "grad_norm": 1.1310456665950113, "learning_rate": 1.2166508433306704e-05, "loss": 0.4409, "step": 12563 }, { "epoch": 0.4476192172720309, "grad_norm": 1.1121825388047795, "learning_rate": 1.2165381889161305e-05, "loss": 0.4587, "step": 12564 }, { "epoch": 0.44765484439852504, "grad_norm": 1.4857241503868461, "learning_rate": 1.2164255316182336e-05, "loss": 0.5258, "step": 12565 }, { "epoch": 0.4476904715250192, "grad_norm": 1.0607329266932748, "learning_rate": 1.2163128714384791e-05, "loss": 0.4243, "step": 12566 }, { "epoch": 0.44772609865151325, "grad_norm": 1.133912591783457, "learning_rate": 1.2162002083783676e-05, "loss": 0.5106, "step": 12567 }, { "epoch": 0.4477617257780074, "grad_norm": 1.1428548183231944, "learning_rate": 1.2160875424393997e-05, "loss": 0.4565, "step": 12568 }, { "epoch": 0.4477973529045015, "grad_norm": 1.1190420363245717, "learning_rate": 1.2159748736230747e-05, "loss": 0.4564, "step": 12569 }, { "epoch": 0.4478329800309956, "grad_norm": 1.304409035106989, "learning_rate": 1.2158622019308932e-05, "loss": 0.5781, "step": 12570 }, { "epoch": 0.4478686071574897, "grad_norm": 1.7858187286464924, "learning_rate": 1.2157495273643558e-05, "loss": 0.5454, "step": 12571 }, { "epoch": 0.44790423428398385, "grad_norm": 1.5366418456631865, "learning_rate": 1.2156368499249629e-05, "loss": 0.4704, "step": 12572 }, { "epoch": 0.4479398614104779, "grad_norm": 1.3967278377244619, "learning_rate": 1.2155241696142141e-05, "loss": 0.5083, "step": 12573 }, { "epoch": 0.44797548853697206, "grad_norm": 1.2597233043350609, "learning_rate": 1.2154114864336107e-05, "loss": 0.4431, "step": 12574 }, { "epoch": 0.4480111156634662, "grad_norm": 1.3045904968851945, "learning_rate": 1.2152988003846525e-05, "loss": 0.4379, "step": 12575 }, { "epoch": 0.44804674278996026, "grad_norm": 1.1484444065135908, "learning_rate": 1.2151861114688406e-05, "loss": 0.477, "step": 12576 }, { "epoch": 0.4480823699164544, "grad_norm": 1.2099023135014726, "learning_rate": 1.2150734196876747e-05, "loss": 0.5064, "step": 12577 }, { "epoch": 0.4481179970429485, "grad_norm": 1.1910648556344097, "learning_rate": 1.2149607250426565e-05, "loss": 0.4906, "step": 12578 }, { "epoch": 0.4481536241694426, "grad_norm": 1.181927186096594, "learning_rate": 1.2148480275352857e-05, "loss": 0.4332, "step": 12579 }, { "epoch": 0.44818925129593673, "grad_norm": 2.0606333592974466, "learning_rate": 1.2147353271670634e-05, "loss": 0.531, "step": 12580 }, { "epoch": 0.44822487842243086, "grad_norm": 1.3615958394676428, "learning_rate": 1.2146226239394898e-05, "loss": 0.5172, "step": 12581 }, { "epoch": 0.44826050554892494, "grad_norm": 1.50544943373391, "learning_rate": 1.2145099178540665e-05, "loss": 0.4967, "step": 12582 }, { "epoch": 0.44829613267541907, "grad_norm": 1.2933252203529075, "learning_rate": 1.2143972089122932e-05, "loss": 0.5177, "step": 12583 }, { "epoch": 0.4483317598019132, "grad_norm": 1.1640415935902135, "learning_rate": 1.2142844971156714e-05, "loss": 0.4932, "step": 12584 }, { "epoch": 0.4483673869284073, "grad_norm": 1.1966475696805237, "learning_rate": 1.2141717824657018e-05, "loss": 0.4931, "step": 12585 }, { "epoch": 0.4484030140549014, "grad_norm": 1.1561673364588096, "learning_rate": 1.214059064963885e-05, "loss": 0.4613, "step": 12586 }, { "epoch": 0.44843864118139554, "grad_norm": 1.1160254401664655, "learning_rate": 1.2139463446117224e-05, "loss": 0.4247, "step": 12587 }, { "epoch": 0.4484742683078896, "grad_norm": 1.4109164981386046, "learning_rate": 1.2138336214107148e-05, "loss": 0.5233, "step": 12588 }, { "epoch": 0.44850989543438374, "grad_norm": 1.4071340474964482, "learning_rate": 1.2137208953623626e-05, "loss": 0.4579, "step": 12589 }, { "epoch": 0.4485455225608779, "grad_norm": 1.818194944040471, "learning_rate": 1.2136081664681675e-05, "loss": 0.5948, "step": 12590 }, { "epoch": 0.44858114968737195, "grad_norm": 1.1441260479577375, "learning_rate": 1.2134954347296306e-05, "loss": 0.4475, "step": 12591 }, { "epoch": 0.4486167768138661, "grad_norm": 1.460064929076152, "learning_rate": 1.2133827001482524e-05, "loss": 0.4925, "step": 12592 }, { "epoch": 0.4486524039403602, "grad_norm": 1.4667110932035954, "learning_rate": 1.2132699627255347e-05, "loss": 0.5701, "step": 12593 }, { "epoch": 0.4486880310668543, "grad_norm": 1.9315898389666664, "learning_rate": 1.2131572224629783e-05, "loss": 0.5732, "step": 12594 }, { "epoch": 0.4487236581933484, "grad_norm": 1.3833540729463851, "learning_rate": 1.2130444793620844e-05, "loss": 0.6079, "step": 12595 }, { "epoch": 0.44875928531984255, "grad_norm": 1.6473797697683172, "learning_rate": 1.2129317334243542e-05, "loss": 0.4661, "step": 12596 }, { "epoch": 0.4487949124463366, "grad_norm": 1.1773280830178796, "learning_rate": 1.2128189846512895e-05, "loss": 0.4648, "step": 12597 }, { "epoch": 0.44883053957283076, "grad_norm": 1.136888510039731, "learning_rate": 1.2127062330443908e-05, "loss": 0.5285, "step": 12598 }, { "epoch": 0.4488661666993249, "grad_norm": 1.2053823325929125, "learning_rate": 1.2125934786051605e-05, "loss": 0.4854, "step": 12599 }, { "epoch": 0.44890179382581896, "grad_norm": 1.401828180574916, "learning_rate": 1.2124807213350991e-05, "loss": 0.5706, "step": 12600 }, { "epoch": 0.4489374209523131, "grad_norm": 1.4161042086687747, "learning_rate": 1.2123679612357085e-05, "loss": 0.4759, "step": 12601 }, { "epoch": 0.4489730480788072, "grad_norm": 1.4362221863646027, "learning_rate": 1.21225519830849e-05, "loss": 0.6064, "step": 12602 }, { "epoch": 0.4490086752053013, "grad_norm": 1.218378757223427, "learning_rate": 1.2121424325549449e-05, "loss": 0.3885, "step": 12603 }, { "epoch": 0.44904430233179543, "grad_norm": 1.8958892304715707, "learning_rate": 1.2120296639765754e-05, "loss": 0.6118, "step": 12604 }, { "epoch": 0.44907992945828956, "grad_norm": 1.8942356463064276, "learning_rate": 1.2119168925748827e-05, "loss": 0.5061, "step": 12605 }, { "epoch": 0.44911555658478364, "grad_norm": 1.0802572608550234, "learning_rate": 1.2118041183513686e-05, "loss": 0.4636, "step": 12606 }, { "epoch": 0.44915118371127777, "grad_norm": 1.0991789912808823, "learning_rate": 1.2116913413075342e-05, "loss": 0.5161, "step": 12607 }, { "epoch": 0.4491868108377719, "grad_norm": 1.6363358259828604, "learning_rate": 1.211578561444882e-05, "loss": 0.5976, "step": 12608 }, { "epoch": 0.449222437964266, "grad_norm": 1.723864632231557, "learning_rate": 1.2114657787649131e-05, "loss": 0.5625, "step": 12609 }, { "epoch": 0.4492580650907601, "grad_norm": 1.235457267850091, "learning_rate": 1.2113529932691297e-05, "loss": 0.4863, "step": 12610 }, { "epoch": 0.44929369221725424, "grad_norm": 1.0649083950439342, "learning_rate": 1.2112402049590333e-05, "loss": 0.4471, "step": 12611 }, { "epoch": 0.4493293193437483, "grad_norm": 1.5988371549493372, "learning_rate": 1.2111274138361262e-05, "loss": 0.4873, "step": 12612 }, { "epoch": 0.44936494647024244, "grad_norm": 1.2774441519831725, "learning_rate": 1.2110146199019099e-05, "loss": 0.4748, "step": 12613 }, { "epoch": 0.4494005735967366, "grad_norm": 1.8128293174638923, "learning_rate": 1.2109018231578863e-05, "loss": 0.598, "step": 12614 }, { "epoch": 0.44943620072323065, "grad_norm": 1.122822795080085, "learning_rate": 1.2107890236055578e-05, "loss": 0.4836, "step": 12615 }, { "epoch": 0.4494718278497248, "grad_norm": 1.0837762561889224, "learning_rate": 1.2106762212464257e-05, "loss": 0.4549, "step": 12616 }, { "epoch": 0.4495074549762189, "grad_norm": 1.095664917530053, "learning_rate": 1.210563416081993e-05, "loss": 0.5062, "step": 12617 }, { "epoch": 0.449543082102713, "grad_norm": 1.19019611777339, "learning_rate": 1.2104506081137608e-05, "loss": 0.5426, "step": 12618 }, { "epoch": 0.4495787092292071, "grad_norm": 1.5176890053261096, "learning_rate": 1.2103377973432317e-05, "loss": 0.5476, "step": 12619 }, { "epoch": 0.44961433635570125, "grad_norm": 1.2395042669722858, "learning_rate": 1.2102249837719083e-05, "loss": 0.4306, "step": 12620 }, { "epoch": 0.4496499634821953, "grad_norm": 1.8045478605140797, "learning_rate": 1.2101121674012916e-05, "loss": 0.5814, "step": 12621 }, { "epoch": 0.44968559060868946, "grad_norm": 1.1506107750787065, "learning_rate": 1.2099993482328846e-05, "loss": 0.4867, "step": 12622 }, { "epoch": 0.4497212177351836, "grad_norm": 1.148099512892814, "learning_rate": 1.20988652626819e-05, "loss": 0.4518, "step": 12623 }, { "epoch": 0.44975684486167766, "grad_norm": 1.091563294149784, "learning_rate": 1.2097737015087094e-05, "loss": 0.42, "step": 12624 }, { "epoch": 0.4497924719881718, "grad_norm": 1.2406328274271226, "learning_rate": 1.2096608739559454e-05, "loss": 0.5551, "step": 12625 }, { "epoch": 0.4498280991146659, "grad_norm": 1.2157626629518792, "learning_rate": 1.2095480436114e-05, "loss": 0.4495, "step": 12626 }, { "epoch": 0.44986372624116, "grad_norm": 2.9657657844226075, "learning_rate": 1.2094352104765762e-05, "loss": 0.5201, "step": 12627 }, { "epoch": 0.44989935336765413, "grad_norm": 1.6174084919175435, "learning_rate": 1.2093223745529763e-05, "loss": 0.5546, "step": 12628 }, { "epoch": 0.44993498049414826, "grad_norm": 1.636849290275066, "learning_rate": 1.2092095358421023e-05, "loss": 0.5807, "step": 12629 }, { "epoch": 0.44997060762064234, "grad_norm": 1.0829103993553753, "learning_rate": 1.2090966943454574e-05, "loss": 0.5046, "step": 12630 }, { "epoch": 0.45000623474713647, "grad_norm": 1.1654127707753057, "learning_rate": 1.2089838500645438e-05, "loss": 0.429, "step": 12631 }, { "epoch": 0.4500418618736306, "grad_norm": 1.5134928594179822, "learning_rate": 1.2088710030008642e-05, "loss": 0.5318, "step": 12632 }, { "epoch": 0.4500774890001247, "grad_norm": 1.4968837434574254, "learning_rate": 1.2087581531559208e-05, "loss": 0.5815, "step": 12633 }, { "epoch": 0.4501131161266188, "grad_norm": 1.6294754420869237, "learning_rate": 1.208645300531217e-05, "loss": 0.4807, "step": 12634 }, { "epoch": 0.45014874325311294, "grad_norm": 1.570328606324306, "learning_rate": 1.2085324451282554e-05, "loss": 0.5743, "step": 12635 }, { "epoch": 0.450184370379607, "grad_norm": 1.1339816725706362, "learning_rate": 1.2084195869485385e-05, "loss": 0.4804, "step": 12636 }, { "epoch": 0.45021999750610114, "grad_norm": 2.2183956272792065, "learning_rate": 1.2083067259935688e-05, "loss": 0.5508, "step": 12637 }, { "epoch": 0.4502556246325953, "grad_norm": 1.486240209652066, "learning_rate": 1.2081938622648494e-05, "loss": 0.4602, "step": 12638 }, { "epoch": 0.45029125175908935, "grad_norm": 1.0566070319770702, "learning_rate": 1.2080809957638837e-05, "loss": 0.4744, "step": 12639 }, { "epoch": 0.4503268788855835, "grad_norm": 2.097184591369582, "learning_rate": 1.2079681264921737e-05, "loss": 0.5468, "step": 12640 }, { "epoch": 0.4503625060120776, "grad_norm": 1.1312258711272893, "learning_rate": 1.2078552544512227e-05, "loss": 0.4505, "step": 12641 }, { "epoch": 0.4503981331385717, "grad_norm": 1.5299326481809368, "learning_rate": 1.2077423796425341e-05, "loss": 0.5839, "step": 12642 }, { "epoch": 0.4504337602650658, "grad_norm": 1.2639221432694177, "learning_rate": 1.2076295020676103e-05, "loss": 0.5446, "step": 12643 }, { "epoch": 0.45046938739155995, "grad_norm": 1.2783899291400913, "learning_rate": 1.2075166217279544e-05, "loss": 0.5173, "step": 12644 }, { "epoch": 0.450505014518054, "grad_norm": 1.5491314788549764, "learning_rate": 1.2074037386250696e-05, "loss": 0.5935, "step": 12645 }, { "epoch": 0.45054064164454816, "grad_norm": 1.0486707728656266, "learning_rate": 1.2072908527604595e-05, "loss": 0.4693, "step": 12646 }, { "epoch": 0.4505762687710423, "grad_norm": 1.4937973129350657, "learning_rate": 1.2071779641356264e-05, "loss": 0.6028, "step": 12647 }, { "epoch": 0.45061189589753636, "grad_norm": 1.1626360439879297, "learning_rate": 1.2070650727520738e-05, "loss": 0.5009, "step": 12648 }, { "epoch": 0.4506475230240305, "grad_norm": 1.5766199417891964, "learning_rate": 1.2069521786113055e-05, "loss": 0.5495, "step": 12649 }, { "epoch": 0.4506831501505246, "grad_norm": 1.5091864194199334, "learning_rate": 1.206839281714824e-05, "loss": 0.6289, "step": 12650 }, { "epoch": 0.4507187772770187, "grad_norm": 1.0129635675281345, "learning_rate": 1.2067263820641331e-05, "loss": 0.4898, "step": 12651 }, { "epoch": 0.45075440440351283, "grad_norm": 1.1595308630246628, "learning_rate": 1.2066134796607356e-05, "loss": 0.4521, "step": 12652 }, { "epoch": 0.45079003153000696, "grad_norm": 1.1888343044884153, "learning_rate": 1.2065005745061354e-05, "loss": 0.4977, "step": 12653 }, { "epoch": 0.45082565865650104, "grad_norm": 1.0822522506946008, "learning_rate": 1.2063876666018357e-05, "loss": 0.4519, "step": 12654 }, { "epoch": 0.45086128578299517, "grad_norm": 1.330189543073269, "learning_rate": 1.20627475594934e-05, "loss": 0.51, "step": 12655 }, { "epoch": 0.4508969129094893, "grad_norm": 1.5140200437559357, "learning_rate": 1.2061618425501513e-05, "loss": 0.504, "step": 12656 }, { "epoch": 0.4509325400359834, "grad_norm": 1.5004700597145708, "learning_rate": 1.2060489264057741e-05, "loss": 0.5519, "step": 12657 }, { "epoch": 0.4509681671624775, "grad_norm": 1.102515498973791, "learning_rate": 1.2059360075177114e-05, "loss": 0.4554, "step": 12658 }, { "epoch": 0.45100379428897164, "grad_norm": 1.3817973614636996, "learning_rate": 1.2058230858874667e-05, "loss": 0.4322, "step": 12659 }, { "epoch": 0.4510394214154657, "grad_norm": 1.827460611416283, "learning_rate": 1.2057101615165436e-05, "loss": 0.6575, "step": 12660 }, { "epoch": 0.45107504854195984, "grad_norm": 1.7443511142004016, "learning_rate": 1.205597234406446e-05, "loss": 0.5628, "step": 12661 }, { "epoch": 0.451110675668454, "grad_norm": 1.073275483293703, "learning_rate": 1.2054843045586779e-05, "loss": 0.4131, "step": 12662 }, { "epoch": 0.45114630279494805, "grad_norm": 1.1071414735664369, "learning_rate": 1.2053713719747423e-05, "loss": 0.5106, "step": 12663 }, { "epoch": 0.4511819299214422, "grad_norm": 1.1478648675432304, "learning_rate": 1.2052584366561433e-05, "loss": 0.4663, "step": 12664 }, { "epoch": 0.4512175570479363, "grad_norm": 1.1375702889475385, "learning_rate": 1.205145498604385e-05, "loss": 0.5072, "step": 12665 }, { "epoch": 0.4512531841744304, "grad_norm": 1.2961042699221421, "learning_rate": 1.2050325578209707e-05, "loss": 0.4726, "step": 12666 }, { "epoch": 0.4512888113009245, "grad_norm": 1.9522791364123402, "learning_rate": 1.2049196143074045e-05, "loss": 0.5131, "step": 12667 }, { "epoch": 0.45132443842741865, "grad_norm": 1.2108113938394125, "learning_rate": 1.2048066680651908e-05, "loss": 0.4923, "step": 12668 }, { "epoch": 0.4513600655539127, "grad_norm": 1.0930593370450619, "learning_rate": 1.2046937190958329e-05, "loss": 0.4641, "step": 12669 }, { "epoch": 0.45139569268040686, "grad_norm": 1.5207822012536218, "learning_rate": 1.2045807674008355e-05, "loss": 0.5379, "step": 12670 }, { "epoch": 0.451431319806901, "grad_norm": 1.2465798793694778, "learning_rate": 1.2044678129817016e-05, "loss": 0.4521, "step": 12671 }, { "epoch": 0.45146694693339506, "grad_norm": 1.6039594906408907, "learning_rate": 1.2043548558399361e-05, "loss": 0.5093, "step": 12672 }, { "epoch": 0.4515025740598892, "grad_norm": 1.0021472755672305, "learning_rate": 1.204241895977043e-05, "loss": 0.4587, "step": 12673 }, { "epoch": 0.4515382011863833, "grad_norm": 1.004399669159606, "learning_rate": 1.2041289333945258e-05, "loss": 0.4324, "step": 12674 }, { "epoch": 0.4515738283128774, "grad_norm": 1.3546339919168167, "learning_rate": 1.2040159680938897e-05, "loss": 0.568, "step": 12675 }, { "epoch": 0.45160945543937153, "grad_norm": 1.3459965517153276, "learning_rate": 1.2039030000766383e-05, "loss": 0.5576, "step": 12676 }, { "epoch": 0.45164508256586566, "grad_norm": 1.4851570120457407, "learning_rate": 1.2037900293442759e-05, "loss": 0.5268, "step": 12677 }, { "epoch": 0.45168070969235974, "grad_norm": 1.7672898862090156, "learning_rate": 1.2036770558983067e-05, "loss": 0.6098, "step": 12678 }, { "epoch": 0.45171633681885387, "grad_norm": 1.0250603819626025, "learning_rate": 1.2035640797402351e-05, "loss": 0.4682, "step": 12679 }, { "epoch": 0.451751963945348, "grad_norm": 1.615945286252559, "learning_rate": 1.2034511008715655e-05, "loss": 0.5775, "step": 12680 }, { "epoch": 0.4517875910718421, "grad_norm": 1.97571470168971, "learning_rate": 1.2033381192938026e-05, "loss": 0.5502, "step": 12681 }, { "epoch": 0.4518232181983362, "grad_norm": 1.5970019078458955, "learning_rate": 1.20322513500845e-05, "loss": 0.6402, "step": 12682 }, { "epoch": 0.45185884532483034, "grad_norm": 1.0913055195495405, "learning_rate": 1.2031121480170132e-05, "loss": 0.4999, "step": 12683 }, { "epoch": 0.4518944724513244, "grad_norm": 1.055784529590962, "learning_rate": 1.2029991583209959e-05, "loss": 0.4082, "step": 12684 }, { "epoch": 0.45193009957781854, "grad_norm": 1.0861380780593959, "learning_rate": 1.2028861659219028e-05, "loss": 0.4885, "step": 12685 }, { "epoch": 0.4519657267043127, "grad_norm": 1.7883934652962648, "learning_rate": 1.2027731708212389e-05, "loss": 0.5802, "step": 12686 }, { "epoch": 0.45200135383080675, "grad_norm": 1.0873849664748276, "learning_rate": 1.2026601730205085e-05, "loss": 0.4531, "step": 12687 }, { "epoch": 0.4520369809573009, "grad_norm": 1.5376308983405853, "learning_rate": 1.2025471725212162e-05, "loss": 0.4805, "step": 12688 }, { "epoch": 0.452072608083795, "grad_norm": 1.4357743306808746, "learning_rate": 1.2024341693248664e-05, "loss": 0.5162, "step": 12689 }, { "epoch": 0.4521082352102891, "grad_norm": 1.4759934210864314, "learning_rate": 1.2023211634329644e-05, "loss": 0.5556, "step": 12690 }, { "epoch": 0.4521438623367832, "grad_norm": 1.0424992856612252, "learning_rate": 1.2022081548470147e-05, "loss": 0.3963, "step": 12691 }, { "epoch": 0.45217948946327735, "grad_norm": 1.1036374886332454, "learning_rate": 1.202095143568522e-05, "loss": 0.4508, "step": 12692 }, { "epoch": 0.4522151165897714, "grad_norm": 1.0780049325599061, "learning_rate": 1.2019821295989913e-05, "loss": 0.4322, "step": 12693 }, { "epoch": 0.45225074371626556, "grad_norm": 1.488118631852623, "learning_rate": 1.2018691129399273e-05, "loss": 0.5829, "step": 12694 }, { "epoch": 0.4522863708427597, "grad_norm": 1.3960536629847586, "learning_rate": 1.201756093592835e-05, "loss": 0.5767, "step": 12695 }, { "epoch": 0.45232199796925376, "grad_norm": 0.9463395480952641, "learning_rate": 1.2016430715592193e-05, "loss": 0.3773, "step": 12696 }, { "epoch": 0.4523576250957479, "grad_norm": 1.0871920912420798, "learning_rate": 1.2015300468405852e-05, "loss": 0.4645, "step": 12697 }, { "epoch": 0.452393252222242, "grad_norm": 1.037505895651242, "learning_rate": 1.2014170194384375e-05, "loss": 0.5001, "step": 12698 }, { "epoch": 0.4524288793487361, "grad_norm": 1.0840937210295727, "learning_rate": 1.2013039893542817e-05, "loss": 0.4601, "step": 12699 }, { "epoch": 0.45246450647523023, "grad_norm": 1.0768332630286246, "learning_rate": 1.2011909565896223e-05, "loss": 0.421, "step": 12700 }, { "epoch": 0.45250013360172436, "grad_norm": 1.2416496048440615, "learning_rate": 1.2010779211459649e-05, "loss": 0.4777, "step": 12701 }, { "epoch": 0.45253576072821844, "grad_norm": 0.9842067717255849, "learning_rate": 1.2009648830248141e-05, "loss": 0.4994, "step": 12702 }, { "epoch": 0.45257138785471257, "grad_norm": 1.0338604610341993, "learning_rate": 1.2008518422276757e-05, "loss": 0.4461, "step": 12703 }, { "epoch": 0.4526070149812067, "grad_norm": 1.0149982999646021, "learning_rate": 1.2007387987560545e-05, "loss": 0.446, "step": 12704 }, { "epoch": 0.4526426421077008, "grad_norm": 1.115946303400081, "learning_rate": 1.200625752611456e-05, "loss": 0.4477, "step": 12705 }, { "epoch": 0.4526782692341949, "grad_norm": 1.0754556839854512, "learning_rate": 1.2005127037953854e-05, "loss": 0.4875, "step": 12706 }, { "epoch": 0.45271389636068904, "grad_norm": 1.3881052333435575, "learning_rate": 1.2003996523093482e-05, "loss": 0.6064, "step": 12707 }, { "epoch": 0.4527495234871831, "grad_norm": 1.0952160829231652, "learning_rate": 1.200286598154849e-05, "loss": 0.466, "step": 12708 }, { "epoch": 0.45278515061367725, "grad_norm": 1.3095574205525282, "learning_rate": 1.2001735413333941e-05, "loss": 0.5906, "step": 12709 }, { "epoch": 0.4528207777401714, "grad_norm": 1.93517659422266, "learning_rate": 1.2000604818464887e-05, "loss": 0.5891, "step": 12710 }, { "epoch": 0.45285640486666545, "grad_norm": 1.4250834280875446, "learning_rate": 1.1999474196956378e-05, "loss": 0.534, "step": 12711 }, { "epoch": 0.4528920319931596, "grad_norm": 1.3797540642482877, "learning_rate": 1.1998343548823474e-05, "loss": 0.5744, "step": 12712 }, { "epoch": 0.4529276591196537, "grad_norm": 1.081560928104801, "learning_rate": 1.199721287408123e-05, "loss": 0.463, "step": 12713 }, { "epoch": 0.4529632862461478, "grad_norm": 1.4187135725138176, "learning_rate": 1.1996082172744702e-05, "loss": 0.5622, "step": 12714 }, { "epoch": 0.4529989133726419, "grad_norm": 1.5978582094103329, "learning_rate": 1.1994951444828941e-05, "loss": 0.5267, "step": 12715 }, { "epoch": 0.45303454049913605, "grad_norm": 1.0776399177302178, "learning_rate": 1.1993820690349008e-05, "loss": 0.4618, "step": 12716 }, { "epoch": 0.4530701676256302, "grad_norm": 1.4770810245495285, "learning_rate": 1.1992689909319958e-05, "loss": 0.5426, "step": 12717 }, { "epoch": 0.45310579475212426, "grad_norm": 1.9796826629369033, "learning_rate": 1.1991559101756853e-05, "loss": 0.6227, "step": 12718 }, { "epoch": 0.4531414218786184, "grad_norm": 1.1055479594382558, "learning_rate": 1.1990428267674744e-05, "loss": 0.4999, "step": 12719 }, { "epoch": 0.4531770490051125, "grad_norm": 1.8943279459769065, "learning_rate": 1.1989297407088692e-05, "loss": 0.5617, "step": 12720 }, { "epoch": 0.4532126761316066, "grad_norm": 1.125342375052891, "learning_rate": 1.1988166520013755e-05, "loss": 0.4878, "step": 12721 }, { "epoch": 0.4532483032581007, "grad_norm": 1.063920662779232, "learning_rate": 1.1987035606464987e-05, "loss": 0.4635, "step": 12722 }, { "epoch": 0.45328393038459486, "grad_norm": 1.162523819319235, "learning_rate": 1.1985904666457455e-05, "loss": 0.4282, "step": 12723 }, { "epoch": 0.45331955751108893, "grad_norm": 2.122915876837053, "learning_rate": 1.1984773700006215e-05, "loss": 0.5322, "step": 12724 }, { "epoch": 0.45335518463758306, "grad_norm": 1.1056909270564899, "learning_rate": 1.1983642707126325e-05, "loss": 0.4598, "step": 12725 }, { "epoch": 0.4533908117640772, "grad_norm": 1.0701673012661235, "learning_rate": 1.1982511687832845e-05, "loss": 0.4202, "step": 12726 }, { "epoch": 0.45342643889057127, "grad_norm": 1.1394111553408162, "learning_rate": 1.1981380642140836e-05, "loss": 0.4722, "step": 12727 }, { "epoch": 0.4534620660170654, "grad_norm": 1.4767095612832757, "learning_rate": 1.1980249570065359e-05, "loss": 0.6598, "step": 12728 }, { "epoch": 0.45349769314355953, "grad_norm": 1.4027405592799196, "learning_rate": 1.1979118471621477e-05, "loss": 0.5781, "step": 12729 }, { "epoch": 0.4535333202700536, "grad_norm": 1.4773134715589507, "learning_rate": 1.1977987346824244e-05, "loss": 0.5227, "step": 12730 }, { "epoch": 0.45356894739654774, "grad_norm": 1.8896727950591505, "learning_rate": 1.1976856195688733e-05, "loss": 0.4764, "step": 12731 }, { "epoch": 0.45360457452304187, "grad_norm": 1.0984975818320477, "learning_rate": 1.1975725018229999e-05, "loss": 0.4456, "step": 12732 }, { "epoch": 0.45364020164953595, "grad_norm": 1.7296119707077409, "learning_rate": 1.1974593814463105e-05, "loss": 0.5617, "step": 12733 }, { "epoch": 0.4536758287760301, "grad_norm": 1.4580628389507362, "learning_rate": 1.197346258440311e-05, "loss": 0.5995, "step": 12734 }, { "epoch": 0.4537114559025242, "grad_norm": 1.82207978078945, "learning_rate": 1.1972331328065085e-05, "loss": 0.5842, "step": 12735 }, { "epoch": 0.4537470830290183, "grad_norm": 1.166627173078126, "learning_rate": 1.1971200045464088e-05, "loss": 0.4847, "step": 12736 }, { "epoch": 0.4537827101555124, "grad_norm": 1.742394227353753, "learning_rate": 1.1970068736615187e-05, "loss": 0.5978, "step": 12737 }, { "epoch": 0.45381833728200655, "grad_norm": 1.6001418282956106, "learning_rate": 1.1968937401533442e-05, "loss": 0.4915, "step": 12738 }, { "epoch": 0.4538539644085006, "grad_norm": 1.0738649073843465, "learning_rate": 1.196780604023392e-05, "loss": 0.4542, "step": 12739 }, { "epoch": 0.45388959153499475, "grad_norm": 1.3850744041773442, "learning_rate": 1.1966674652731687e-05, "loss": 0.4849, "step": 12740 }, { "epoch": 0.4539252186614889, "grad_norm": 2.013284124059489, "learning_rate": 1.1965543239041803e-05, "loss": 0.5869, "step": 12741 }, { "epoch": 0.45396084578798296, "grad_norm": 1.1973743387259037, "learning_rate": 1.1964411799179338e-05, "loss": 0.5102, "step": 12742 }, { "epoch": 0.4539964729144771, "grad_norm": 1.3840555972470856, "learning_rate": 1.1963280333159359e-05, "loss": 0.5449, "step": 12743 }, { "epoch": 0.4540321000409712, "grad_norm": 1.0913738624901796, "learning_rate": 1.196214884099693e-05, "loss": 0.4694, "step": 12744 }, { "epoch": 0.4540677271674653, "grad_norm": 1.6626956004357056, "learning_rate": 1.1961017322707114e-05, "loss": 0.5517, "step": 12745 }, { "epoch": 0.4541033542939594, "grad_norm": 1.1580440721028824, "learning_rate": 1.1959885778304985e-05, "loss": 0.5338, "step": 12746 }, { "epoch": 0.45413898142045356, "grad_norm": 1.0610859901243672, "learning_rate": 1.1958754207805606e-05, "loss": 0.427, "step": 12747 }, { "epoch": 0.45417460854694763, "grad_norm": 1.8543927969821619, "learning_rate": 1.1957622611224044e-05, "loss": 0.5309, "step": 12748 }, { "epoch": 0.45421023567344176, "grad_norm": 1.0775551411359436, "learning_rate": 1.1956490988575369e-05, "loss": 0.4785, "step": 12749 }, { "epoch": 0.4542458627999359, "grad_norm": 1.2903020665407858, "learning_rate": 1.195535933987465e-05, "loss": 0.5117, "step": 12750 }, { "epoch": 0.45428148992642997, "grad_norm": 1.1521336348329765, "learning_rate": 1.1954227665136956e-05, "loss": 0.4977, "step": 12751 }, { "epoch": 0.4543171170529241, "grad_norm": 1.3198146163069087, "learning_rate": 1.1953095964377353e-05, "loss": 0.4381, "step": 12752 }, { "epoch": 0.45435274417941823, "grad_norm": 1.0890610880664624, "learning_rate": 1.195196423761091e-05, "loss": 0.4475, "step": 12753 }, { "epoch": 0.4543883713059123, "grad_norm": 1.7837604875560398, "learning_rate": 1.1950832484852704e-05, "loss": 0.6181, "step": 12754 }, { "epoch": 0.45442399843240644, "grad_norm": 1.314358969048131, "learning_rate": 1.1949700706117797e-05, "loss": 0.5127, "step": 12755 }, { "epoch": 0.45445962555890057, "grad_norm": 2.063485252803922, "learning_rate": 1.1948568901421262e-05, "loss": 0.5483, "step": 12756 }, { "epoch": 0.45449525268539465, "grad_norm": 2.0208273068893994, "learning_rate": 1.1947437070778171e-05, "loss": 0.568, "step": 12757 }, { "epoch": 0.4545308798118888, "grad_norm": 1.1528697380153696, "learning_rate": 1.1946305214203595e-05, "loss": 0.463, "step": 12758 }, { "epoch": 0.4545665069383829, "grad_norm": 1.2335143655174272, "learning_rate": 1.1945173331712603e-05, "loss": 0.4703, "step": 12759 }, { "epoch": 0.454602134064877, "grad_norm": 1.4156234478957688, "learning_rate": 1.1944041423320267e-05, "loss": 0.5859, "step": 12760 }, { "epoch": 0.4546377611913711, "grad_norm": 1.0844944243033687, "learning_rate": 1.1942909489041663e-05, "loss": 0.4888, "step": 12761 }, { "epoch": 0.45467338831786525, "grad_norm": 1.0872797106005287, "learning_rate": 1.1941777528891862e-05, "loss": 0.4484, "step": 12762 }, { "epoch": 0.4547090154443593, "grad_norm": 1.1195450863379324, "learning_rate": 1.1940645542885935e-05, "loss": 0.4716, "step": 12763 }, { "epoch": 0.45474464257085345, "grad_norm": 1.438183146267767, "learning_rate": 1.1939513531038955e-05, "loss": 0.6103, "step": 12764 }, { "epoch": 0.4547802696973476, "grad_norm": 1.5408933379077845, "learning_rate": 1.1938381493366e-05, "loss": 0.5766, "step": 12765 }, { "epoch": 0.45481589682384166, "grad_norm": 1.6647687899363537, "learning_rate": 1.1937249429882139e-05, "loss": 0.508, "step": 12766 }, { "epoch": 0.4548515239503358, "grad_norm": 1.5416327708613673, "learning_rate": 1.1936117340602445e-05, "loss": 0.5324, "step": 12767 }, { "epoch": 0.4548871510768299, "grad_norm": 1.6418071977989266, "learning_rate": 1.1934985225541998e-05, "loss": 0.5973, "step": 12768 }, { "epoch": 0.454922778203324, "grad_norm": 1.036834289597959, "learning_rate": 1.1933853084715872e-05, "loss": 0.4586, "step": 12769 }, { "epoch": 0.4549584053298181, "grad_norm": 1.1535310761573316, "learning_rate": 1.1932720918139139e-05, "loss": 0.4813, "step": 12770 }, { "epoch": 0.45499403245631226, "grad_norm": 1.8481230225487209, "learning_rate": 1.1931588725826873e-05, "loss": 0.5555, "step": 12771 }, { "epoch": 0.45502965958280633, "grad_norm": 1.682965416283879, "learning_rate": 1.1930456507794153e-05, "loss": 0.5525, "step": 12772 }, { "epoch": 0.45506528670930046, "grad_norm": 1.8292959894269663, "learning_rate": 1.1929324264056059e-05, "loss": 0.6052, "step": 12773 }, { "epoch": 0.4551009138357946, "grad_norm": 1.774479739136282, "learning_rate": 1.1928191994627663e-05, "loss": 0.5864, "step": 12774 }, { "epoch": 0.45513654096228867, "grad_norm": 1.0714522776200304, "learning_rate": 1.1927059699524042e-05, "loss": 0.4898, "step": 12775 }, { "epoch": 0.4551721680887828, "grad_norm": 1.3491769822435515, "learning_rate": 1.1925927378760274e-05, "loss": 0.4629, "step": 12776 }, { "epoch": 0.45520779521527693, "grad_norm": 1.1999483687741885, "learning_rate": 1.192479503235144e-05, "loss": 0.4494, "step": 12777 }, { "epoch": 0.455243422341771, "grad_norm": 1.8419930146444685, "learning_rate": 1.1923662660312611e-05, "loss": 0.7004, "step": 12778 }, { "epoch": 0.45527904946826514, "grad_norm": 1.23003704412949, "learning_rate": 1.192253026265887e-05, "loss": 0.484, "step": 12779 }, { "epoch": 0.45531467659475927, "grad_norm": 1.0117818114556123, "learning_rate": 1.1921397839405296e-05, "loss": 0.4763, "step": 12780 }, { "epoch": 0.45535030372125335, "grad_norm": 1.922409102861084, "learning_rate": 1.1920265390566966e-05, "loss": 0.6056, "step": 12781 }, { "epoch": 0.4553859308477475, "grad_norm": 4.254748655321277, "learning_rate": 1.1919132916158957e-05, "loss": 0.5953, "step": 12782 }, { "epoch": 0.4554215579742416, "grad_norm": 1.4010492472397205, "learning_rate": 1.1918000416196354e-05, "loss": 0.5025, "step": 12783 }, { "epoch": 0.4554571851007357, "grad_norm": 1.1832646204080173, "learning_rate": 1.1916867890694237e-05, "loss": 0.4746, "step": 12784 }, { "epoch": 0.4554928122272298, "grad_norm": 1.425459352080474, "learning_rate": 1.1915735339667682e-05, "loss": 0.5414, "step": 12785 }, { "epoch": 0.45552843935372395, "grad_norm": 1.7179872057865413, "learning_rate": 1.1914602763131772e-05, "loss": 0.3803, "step": 12786 }, { "epoch": 0.455564066480218, "grad_norm": 1.5924097234560872, "learning_rate": 1.191347016110159e-05, "loss": 0.58, "step": 12787 }, { "epoch": 0.45559969360671215, "grad_norm": 1.2434488644698865, "learning_rate": 1.1912337533592216e-05, "loss": 0.4898, "step": 12788 }, { "epoch": 0.4556353207332063, "grad_norm": 1.7510217360537226, "learning_rate": 1.191120488061873e-05, "loss": 0.5746, "step": 12789 }, { "epoch": 0.45567094785970036, "grad_norm": 1.6932646450118176, "learning_rate": 1.1910072202196212e-05, "loss": 0.5228, "step": 12790 }, { "epoch": 0.4557065749861945, "grad_norm": 2.0844318680138065, "learning_rate": 1.1908939498339751e-05, "loss": 0.5631, "step": 12791 }, { "epoch": 0.4557422021126886, "grad_norm": 1.3801948342137418, "learning_rate": 1.1907806769064425e-05, "loss": 0.4911, "step": 12792 }, { "epoch": 0.4557778292391827, "grad_norm": 1.3227959360652586, "learning_rate": 1.1906674014385318e-05, "loss": 0.4291, "step": 12793 }, { "epoch": 0.45581345636567683, "grad_norm": 2.2479570276653225, "learning_rate": 1.1905541234317513e-05, "loss": 0.6854, "step": 12794 }, { "epoch": 0.45584908349217096, "grad_norm": 1.68561869286533, "learning_rate": 1.1904408428876098e-05, "loss": 0.5887, "step": 12795 }, { "epoch": 0.45588471061866503, "grad_norm": 1.0642839633509813, "learning_rate": 1.1903275598076152e-05, "loss": 0.4708, "step": 12796 }, { "epoch": 0.45592033774515917, "grad_norm": 1.3453426512789561, "learning_rate": 1.1902142741932761e-05, "loss": 0.4942, "step": 12797 }, { "epoch": 0.4559559648716533, "grad_norm": 1.542735475503603, "learning_rate": 1.1901009860461008e-05, "loss": 0.5581, "step": 12798 }, { "epoch": 0.45599159199814737, "grad_norm": 1.7969072880306294, "learning_rate": 1.1899876953675981e-05, "loss": 0.6216, "step": 12799 }, { "epoch": 0.4560272191246415, "grad_norm": 2.0724049575156034, "learning_rate": 1.1898744021592768e-05, "loss": 0.5104, "step": 12800 }, { "epoch": 0.45606284625113563, "grad_norm": 1.1123663774808927, "learning_rate": 1.1897611064226447e-05, "loss": 0.4468, "step": 12801 }, { "epoch": 0.4560984733776297, "grad_norm": 1.7379660469578986, "learning_rate": 1.189647808159211e-05, "loss": 0.5392, "step": 12802 }, { "epoch": 0.45613410050412384, "grad_norm": 1.1577147288273508, "learning_rate": 1.1895345073704842e-05, "loss": 0.4904, "step": 12803 }, { "epoch": 0.45616972763061797, "grad_norm": 1.0769670489700118, "learning_rate": 1.1894212040579727e-05, "loss": 0.4753, "step": 12804 }, { "epoch": 0.45620535475711205, "grad_norm": 1.044756411001822, "learning_rate": 1.1893078982231858e-05, "loss": 0.4558, "step": 12805 }, { "epoch": 0.4562409818836062, "grad_norm": 1.3563424964489725, "learning_rate": 1.1891945898676316e-05, "loss": 0.5324, "step": 12806 }, { "epoch": 0.4562766090101003, "grad_norm": 1.7696979274645723, "learning_rate": 1.1890812789928194e-05, "loss": 0.6372, "step": 12807 }, { "epoch": 0.4563122361365944, "grad_norm": 1.994727434584779, "learning_rate": 1.1889679656002579e-05, "loss": 0.5793, "step": 12808 }, { "epoch": 0.4563478632630885, "grad_norm": 1.1270551543018066, "learning_rate": 1.1888546496914557e-05, "loss": 0.4365, "step": 12809 }, { "epoch": 0.45638349038958265, "grad_norm": 1.3429121423813815, "learning_rate": 1.1887413312679219e-05, "loss": 0.4536, "step": 12810 }, { "epoch": 0.4564191175160767, "grad_norm": 1.089076406129969, "learning_rate": 1.1886280103311655e-05, "loss": 0.487, "step": 12811 }, { "epoch": 0.45645474464257085, "grad_norm": 2.087203655626736, "learning_rate": 1.1885146868826949e-05, "loss": 0.577, "step": 12812 }, { "epoch": 0.456490371769065, "grad_norm": 1.6850826838310884, "learning_rate": 1.18840136092402e-05, "loss": 0.575, "step": 12813 }, { "epoch": 0.45652599889555906, "grad_norm": 1.4373073154471194, "learning_rate": 1.1882880324566491e-05, "loss": 0.6334, "step": 12814 }, { "epoch": 0.4565616260220532, "grad_norm": 1.1130974380888703, "learning_rate": 1.1881747014820912e-05, "loss": 0.4718, "step": 12815 }, { "epoch": 0.4565972531485473, "grad_norm": 1.557473116042074, "learning_rate": 1.1880613680018558e-05, "loss": 0.5812, "step": 12816 }, { "epoch": 0.4566328802750414, "grad_norm": 1.4070095030626062, "learning_rate": 1.1879480320174516e-05, "loss": 0.5211, "step": 12817 }, { "epoch": 0.45666850740153553, "grad_norm": 1.2191507725881745, "learning_rate": 1.1878346935303884e-05, "loss": 0.4261, "step": 12818 }, { "epoch": 0.45670413452802966, "grad_norm": 1.3389921551789707, "learning_rate": 1.1877213525421746e-05, "loss": 0.4983, "step": 12819 }, { "epoch": 0.45673976165452373, "grad_norm": 1.7061367430104322, "learning_rate": 1.18760800905432e-05, "loss": 0.5422, "step": 12820 }, { "epoch": 0.45677538878101787, "grad_norm": 1.1368818290973965, "learning_rate": 1.1874946630683337e-05, "loss": 0.4584, "step": 12821 }, { "epoch": 0.456811015907512, "grad_norm": 1.8600104442196308, "learning_rate": 1.187381314585725e-05, "loss": 0.5557, "step": 12822 }, { "epoch": 0.4568466430340061, "grad_norm": 1.2996073388816651, "learning_rate": 1.1872679636080027e-05, "loss": 0.4958, "step": 12823 }, { "epoch": 0.4568822701605002, "grad_norm": 1.1755567414558663, "learning_rate": 1.187154610136677e-05, "loss": 0.4637, "step": 12824 }, { "epoch": 0.45691789728699433, "grad_norm": 1.1273890377331948, "learning_rate": 1.1870412541732566e-05, "loss": 0.424, "step": 12825 }, { "epoch": 0.4569535244134884, "grad_norm": 1.4137866524831038, "learning_rate": 1.1869278957192511e-05, "loss": 0.4943, "step": 12826 }, { "epoch": 0.45698915153998254, "grad_norm": 2.1982012277145317, "learning_rate": 1.1868145347761701e-05, "loss": 0.5139, "step": 12827 }, { "epoch": 0.45702477866647667, "grad_norm": 1.5638707695354703, "learning_rate": 1.186701171345523e-05, "loss": 0.494, "step": 12828 }, { "epoch": 0.45706040579297075, "grad_norm": 1.4030672141075324, "learning_rate": 1.186587805428819e-05, "loss": 0.5634, "step": 12829 }, { "epoch": 0.4570960329194649, "grad_norm": 1.3075556665805763, "learning_rate": 1.1864744370275685e-05, "loss": 0.5067, "step": 12830 }, { "epoch": 0.457131660045959, "grad_norm": 1.9085637237967423, "learning_rate": 1.1863610661432802e-05, "loss": 0.5956, "step": 12831 }, { "epoch": 0.4571672871724531, "grad_norm": 1.3923791193362, "learning_rate": 1.1862476927774641e-05, "loss": 0.458, "step": 12832 }, { "epoch": 0.4572029142989472, "grad_norm": 2.157155881454323, "learning_rate": 1.1861343169316301e-05, "loss": 0.577, "step": 12833 }, { "epoch": 0.45723854142544135, "grad_norm": 1.1423516240088862, "learning_rate": 1.1860209386072874e-05, "loss": 0.4645, "step": 12834 }, { "epoch": 0.4572741685519354, "grad_norm": 1.1198345591441698, "learning_rate": 1.1859075578059457e-05, "loss": 0.4824, "step": 12835 }, { "epoch": 0.45730979567842955, "grad_norm": 1.6818671988662934, "learning_rate": 1.185794174529115e-05, "loss": 0.6528, "step": 12836 }, { "epoch": 0.4573454228049237, "grad_norm": 1.4452737613836457, "learning_rate": 1.1856807887783052e-05, "loss": 0.5264, "step": 12837 }, { "epoch": 0.45738104993141776, "grad_norm": 2.6535728715821136, "learning_rate": 1.1855674005550256e-05, "loss": 0.6141, "step": 12838 }, { "epoch": 0.4574166770579119, "grad_norm": 1.3065028596710324, "learning_rate": 1.1854540098607865e-05, "loss": 0.6037, "step": 12839 }, { "epoch": 0.457452304184406, "grad_norm": 1.2527411413002771, "learning_rate": 1.1853406166970975e-05, "loss": 0.479, "step": 12840 }, { "epoch": 0.4574879313109001, "grad_norm": 1.0463698296095074, "learning_rate": 1.185227221065469e-05, "loss": 0.4615, "step": 12841 }, { "epoch": 0.45752355843739423, "grad_norm": 1.1851763377386406, "learning_rate": 1.1851138229674101e-05, "loss": 0.466, "step": 12842 }, { "epoch": 0.45755918556388836, "grad_norm": 1.0928813600413099, "learning_rate": 1.1850004224044315e-05, "loss": 0.4987, "step": 12843 }, { "epoch": 0.45759481269038244, "grad_norm": 1.1834287169804545, "learning_rate": 1.1848870193780432e-05, "loss": 0.4828, "step": 12844 }, { "epoch": 0.45763043981687657, "grad_norm": 1.6241959003474515, "learning_rate": 1.184773613889755e-05, "loss": 0.4238, "step": 12845 }, { "epoch": 0.4576660669433707, "grad_norm": 1.7584952131954483, "learning_rate": 1.1846602059410765e-05, "loss": 0.6314, "step": 12846 }, { "epoch": 0.4577016940698648, "grad_norm": 1.6051113338765994, "learning_rate": 1.1845467955335188e-05, "loss": 0.5204, "step": 12847 }, { "epoch": 0.4577373211963589, "grad_norm": 1.0543366879083613, "learning_rate": 1.1844333826685913e-05, "loss": 0.4939, "step": 12848 }, { "epoch": 0.45777294832285303, "grad_norm": 1.4688628437909943, "learning_rate": 1.1843199673478044e-05, "loss": 0.5639, "step": 12849 }, { "epoch": 0.4578085754493471, "grad_norm": 1.0176004244984167, "learning_rate": 1.1842065495726682e-05, "loss": 0.4159, "step": 12850 }, { "epoch": 0.45784420257584124, "grad_norm": 1.3279014213124856, "learning_rate": 1.1840931293446931e-05, "loss": 0.4947, "step": 12851 }, { "epoch": 0.4578798297023354, "grad_norm": 1.5843820930523944, "learning_rate": 1.1839797066653893e-05, "loss": 0.5717, "step": 12852 }, { "epoch": 0.45791545682882945, "grad_norm": 1.4700615516405036, "learning_rate": 1.1838662815362673e-05, "loss": 0.5972, "step": 12853 }, { "epoch": 0.4579510839553236, "grad_norm": 1.1340573237058829, "learning_rate": 1.183752853958837e-05, "loss": 0.4583, "step": 12854 }, { "epoch": 0.4579867110818177, "grad_norm": 1.191206682627451, "learning_rate": 1.1836394239346092e-05, "loss": 0.46, "step": 12855 }, { "epoch": 0.4580223382083118, "grad_norm": 0.965721028873838, "learning_rate": 1.183525991465094e-05, "loss": 0.4428, "step": 12856 }, { "epoch": 0.4580579653348059, "grad_norm": 1.4804425403880301, "learning_rate": 1.183412556551802e-05, "loss": 0.4703, "step": 12857 }, { "epoch": 0.45809359246130005, "grad_norm": 1.5471445824598973, "learning_rate": 1.1832991191962435e-05, "loss": 0.6063, "step": 12858 }, { "epoch": 0.4581292195877941, "grad_norm": 1.1618694203542443, "learning_rate": 1.1831856793999292e-05, "loss": 0.483, "step": 12859 }, { "epoch": 0.45816484671428825, "grad_norm": 1.6216957944069768, "learning_rate": 1.1830722371643697e-05, "loss": 0.585, "step": 12860 }, { "epoch": 0.4582004738407824, "grad_norm": 1.0749632733801109, "learning_rate": 1.1829587924910753e-05, "loss": 0.4069, "step": 12861 }, { "epoch": 0.45823610096727646, "grad_norm": 1.2644169797008142, "learning_rate": 1.1828453453815567e-05, "loss": 0.4532, "step": 12862 }, { "epoch": 0.4582717280937706, "grad_norm": 1.235667797001859, "learning_rate": 1.1827318958373241e-05, "loss": 0.4851, "step": 12863 }, { "epoch": 0.4583073552202647, "grad_norm": 1.476719577192222, "learning_rate": 1.1826184438598891e-05, "loss": 0.6267, "step": 12864 }, { "epoch": 0.4583429823467588, "grad_norm": 1.4964114853538897, "learning_rate": 1.1825049894507616e-05, "loss": 0.5467, "step": 12865 }, { "epoch": 0.45837860947325293, "grad_norm": 1.2231302356036733, "learning_rate": 1.1823915326114528e-05, "loss": 0.5081, "step": 12866 }, { "epoch": 0.45841423659974706, "grad_norm": 2.046854703612137, "learning_rate": 1.1822780733434733e-05, "loss": 0.6113, "step": 12867 }, { "epoch": 0.45844986372624114, "grad_norm": 1.0910241151413336, "learning_rate": 1.1821646116483337e-05, "loss": 0.4674, "step": 12868 }, { "epoch": 0.45848549085273527, "grad_norm": 1.2955604592918413, "learning_rate": 1.1820511475275448e-05, "loss": 0.6057, "step": 12869 }, { "epoch": 0.4585211179792294, "grad_norm": 1.1940001772929982, "learning_rate": 1.1819376809826181e-05, "loss": 0.4282, "step": 12870 }, { "epoch": 0.4585567451057235, "grad_norm": 1.9576742819731332, "learning_rate": 1.1818242120150637e-05, "loss": 0.6299, "step": 12871 }, { "epoch": 0.4585923722322176, "grad_norm": 0.9913456992535378, "learning_rate": 1.1817107406263929e-05, "loss": 0.4656, "step": 12872 }, { "epoch": 0.45862799935871174, "grad_norm": 1.3860380995940975, "learning_rate": 1.1815972668181166e-05, "loss": 0.485, "step": 12873 }, { "epoch": 0.4586636264852058, "grad_norm": 1.319715586505005, "learning_rate": 1.1814837905917458e-05, "loss": 0.5288, "step": 12874 }, { "epoch": 0.45869925361169994, "grad_norm": 1.5020031213294869, "learning_rate": 1.1813703119487912e-05, "loss": 0.585, "step": 12875 }, { "epoch": 0.4587348807381941, "grad_norm": 1.0682626054888462, "learning_rate": 1.181256830890764e-05, "loss": 0.4217, "step": 12876 }, { "epoch": 0.45877050786468815, "grad_norm": 1.0180511880298155, "learning_rate": 1.181143347419176e-05, "loss": 0.4857, "step": 12877 }, { "epoch": 0.4588061349911823, "grad_norm": 2.244597188042039, "learning_rate": 1.1810298615355376e-05, "loss": 0.5331, "step": 12878 }, { "epoch": 0.4588417621176764, "grad_norm": 1.0102095697883777, "learning_rate": 1.18091637324136e-05, "loss": 0.4846, "step": 12879 }, { "epoch": 0.4588773892441705, "grad_norm": 1.9254053255748098, "learning_rate": 1.1808028825381541e-05, "loss": 0.6008, "step": 12880 }, { "epoch": 0.4589130163706646, "grad_norm": 1.6405769293323709, "learning_rate": 1.1806893894274319e-05, "loss": 0.6392, "step": 12881 }, { "epoch": 0.45894864349715875, "grad_norm": 0.9976952044879922, "learning_rate": 1.180575893910704e-05, "loss": 0.4232, "step": 12882 }, { "epoch": 0.4589842706236528, "grad_norm": 1.151560555417298, "learning_rate": 1.1804623959894817e-05, "loss": 0.5124, "step": 12883 }, { "epoch": 0.45901989775014695, "grad_norm": 1.1779826621318605, "learning_rate": 1.1803488956652767e-05, "loss": 0.506, "step": 12884 }, { "epoch": 0.4590555248766411, "grad_norm": 1.5503955805460912, "learning_rate": 1.1802353929396e-05, "loss": 0.5588, "step": 12885 }, { "epoch": 0.45909115200313516, "grad_norm": 1.8580321371370105, "learning_rate": 1.1801218878139628e-05, "loss": 0.539, "step": 12886 }, { "epoch": 0.4591267791296293, "grad_norm": 1.503127979577434, "learning_rate": 1.1800083802898769e-05, "loss": 0.6226, "step": 12887 }, { "epoch": 0.4591624062561234, "grad_norm": 1.6023729605970995, "learning_rate": 1.179894870368854e-05, "loss": 0.5512, "step": 12888 }, { "epoch": 0.4591980333826175, "grad_norm": 1.1407721900893975, "learning_rate": 1.1797813580524049e-05, "loss": 0.3916, "step": 12889 }, { "epoch": 0.45923366050911163, "grad_norm": 1.261035845367105, "learning_rate": 1.1796678433420415e-05, "loss": 0.476, "step": 12890 }, { "epoch": 0.45926928763560576, "grad_norm": 1.0815548915275381, "learning_rate": 1.1795543262392748e-05, "loss": 0.477, "step": 12891 }, { "epoch": 0.45930491476209984, "grad_norm": 1.2207067019252158, "learning_rate": 1.1794408067456171e-05, "loss": 0.5096, "step": 12892 }, { "epoch": 0.45934054188859397, "grad_norm": 1.39040259659417, "learning_rate": 1.1793272848625796e-05, "loss": 0.5874, "step": 12893 }, { "epoch": 0.4593761690150881, "grad_norm": 1.6128486994719131, "learning_rate": 1.1792137605916736e-05, "loss": 0.575, "step": 12894 }, { "epoch": 0.4594117961415822, "grad_norm": 1.673541144914138, "learning_rate": 1.1791002339344116e-05, "loss": 0.5641, "step": 12895 }, { "epoch": 0.4594474232680763, "grad_norm": 1.5909184529771399, "learning_rate": 1.1789867048923048e-05, "loss": 0.5528, "step": 12896 }, { "epoch": 0.45948305039457044, "grad_norm": 1.5274772695301726, "learning_rate": 1.1788731734668646e-05, "loss": 0.606, "step": 12897 }, { "epoch": 0.4595186775210645, "grad_norm": 1.134571196206809, "learning_rate": 1.1787596396596028e-05, "loss": 0.512, "step": 12898 }, { "epoch": 0.45955430464755864, "grad_norm": 1.717669891863218, "learning_rate": 1.178646103472032e-05, "loss": 0.5892, "step": 12899 }, { "epoch": 0.4595899317740528, "grad_norm": 1.0820899964342485, "learning_rate": 1.1785325649056633e-05, "loss": 0.4766, "step": 12900 }, { "epoch": 0.45962555890054685, "grad_norm": 1.4088339459772652, "learning_rate": 1.1784190239620089e-05, "loss": 0.5219, "step": 12901 }, { "epoch": 0.459661186027041, "grad_norm": 1.435290954690027, "learning_rate": 1.17830548064258e-05, "loss": 0.5387, "step": 12902 }, { "epoch": 0.4596968131535351, "grad_norm": 1.1465114596605854, "learning_rate": 1.1781919349488894e-05, "loss": 0.4835, "step": 12903 }, { "epoch": 0.45973244028002924, "grad_norm": 1.2360350946488097, "learning_rate": 1.1780783868824485e-05, "loss": 0.529, "step": 12904 }, { "epoch": 0.4597680674065233, "grad_norm": 1.1109797689451726, "learning_rate": 1.1779648364447694e-05, "loss": 0.4598, "step": 12905 }, { "epoch": 0.45980369453301745, "grad_norm": 1.725536540783129, "learning_rate": 1.1778512836373643e-05, "loss": 0.5464, "step": 12906 }, { "epoch": 0.4598393216595116, "grad_norm": 1.704328268330854, "learning_rate": 1.1777377284617448e-05, "loss": 0.626, "step": 12907 }, { "epoch": 0.45987494878600566, "grad_norm": 1.581621595278005, "learning_rate": 1.1776241709194235e-05, "loss": 0.5907, "step": 12908 }, { "epoch": 0.4599105759124998, "grad_norm": 1.388966878084813, "learning_rate": 1.177510611011912e-05, "loss": 0.5463, "step": 12909 }, { "epoch": 0.4599462030389939, "grad_norm": 1.658139296756815, "learning_rate": 1.1773970487407224e-05, "loss": 0.5242, "step": 12910 }, { "epoch": 0.459981830165488, "grad_norm": 1.4744237265202194, "learning_rate": 1.1772834841073676e-05, "loss": 0.5467, "step": 12911 }, { "epoch": 0.4600174572919821, "grad_norm": 1.0457741847892708, "learning_rate": 1.177169917113359e-05, "loss": 0.4375, "step": 12912 }, { "epoch": 0.46005308441847625, "grad_norm": 1.963458888340084, "learning_rate": 1.1770563477602091e-05, "loss": 0.6025, "step": 12913 }, { "epoch": 0.46008871154497033, "grad_norm": 1.1942361572551063, "learning_rate": 1.1769427760494303e-05, "loss": 0.4654, "step": 12914 }, { "epoch": 0.46012433867146446, "grad_norm": 1.1699011241347672, "learning_rate": 1.1768292019825348e-05, "loss": 0.5145, "step": 12915 }, { "epoch": 0.4601599657979586, "grad_norm": 1.452307074548646, "learning_rate": 1.1767156255610348e-05, "loss": 0.4774, "step": 12916 }, { "epoch": 0.46019559292445267, "grad_norm": 1.0753095878281043, "learning_rate": 1.1766020467864427e-05, "loss": 0.4365, "step": 12917 }, { "epoch": 0.4602312200509468, "grad_norm": 1.1213803822729522, "learning_rate": 1.1764884656602711e-05, "loss": 0.4692, "step": 12918 }, { "epoch": 0.46026684717744093, "grad_norm": 0.9748964239010847, "learning_rate": 1.1763748821840319e-05, "loss": 0.441, "step": 12919 }, { "epoch": 0.460302474303935, "grad_norm": 1.29518453258835, "learning_rate": 1.176261296359238e-05, "loss": 0.4606, "step": 12920 }, { "epoch": 0.46033810143042914, "grad_norm": 1.1637715876624966, "learning_rate": 1.1761477081874016e-05, "loss": 0.4696, "step": 12921 }, { "epoch": 0.46037372855692327, "grad_norm": 1.1413448169552232, "learning_rate": 1.1760341176700357e-05, "loss": 0.5134, "step": 12922 }, { "epoch": 0.46040935568341734, "grad_norm": 1.8014539217162946, "learning_rate": 1.1759205248086522e-05, "loss": 0.5749, "step": 12923 }, { "epoch": 0.4604449828099115, "grad_norm": 1.0789953341529306, "learning_rate": 1.1758069296047641e-05, "loss": 0.4273, "step": 12924 }, { "epoch": 0.4604806099364056, "grad_norm": 1.7197187857105392, "learning_rate": 1.1756933320598838e-05, "loss": 0.6045, "step": 12925 }, { "epoch": 0.4605162370628997, "grad_norm": 1.0388045057140332, "learning_rate": 1.175579732175524e-05, "loss": 0.4453, "step": 12926 }, { "epoch": 0.4605518641893938, "grad_norm": 1.3911293703062195, "learning_rate": 1.1754661299531975e-05, "loss": 0.5918, "step": 12927 }, { "epoch": 0.46058749131588794, "grad_norm": 1.1675701872195852, "learning_rate": 1.1753525253944164e-05, "loss": 0.4921, "step": 12928 }, { "epoch": 0.460623118442382, "grad_norm": 1.4440034890843494, "learning_rate": 1.1752389185006944e-05, "loss": 0.5082, "step": 12929 }, { "epoch": 0.46065874556887615, "grad_norm": 1.4069434570141905, "learning_rate": 1.1751253092735432e-05, "loss": 0.4839, "step": 12930 }, { "epoch": 0.4606943726953703, "grad_norm": 1.1153240479002575, "learning_rate": 1.1750116977144763e-05, "loss": 0.4511, "step": 12931 }, { "epoch": 0.46072999982186436, "grad_norm": 1.0704587151961484, "learning_rate": 1.1748980838250065e-05, "loss": 0.4317, "step": 12932 }, { "epoch": 0.4607656269483585, "grad_norm": 1.4271748246215719, "learning_rate": 1.1747844676066462e-05, "loss": 0.6402, "step": 12933 }, { "epoch": 0.4608012540748526, "grad_norm": 1.5748051998267207, "learning_rate": 1.1746708490609087e-05, "loss": 0.48, "step": 12934 }, { "epoch": 0.4608368812013467, "grad_norm": 1.0699580945847116, "learning_rate": 1.1745572281893068e-05, "loss": 0.4091, "step": 12935 }, { "epoch": 0.4608725083278408, "grad_norm": 1.5778659537554327, "learning_rate": 1.1744436049933531e-05, "loss": 0.5945, "step": 12936 }, { "epoch": 0.46090813545433496, "grad_norm": 1.7071984052003537, "learning_rate": 1.1743299794745612e-05, "loss": 0.595, "step": 12937 }, { "epoch": 0.46094376258082903, "grad_norm": 1.1554985067424006, "learning_rate": 1.1742163516344439e-05, "loss": 0.4881, "step": 12938 }, { "epoch": 0.46097938970732316, "grad_norm": 1.3025877815798128, "learning_rate": 1.1741027214745138e-05, "loss": 0.5525, "step": 12939 }, { "epoch": 0.4610150168338173, "grad_norm": 1.1107810246952892, "learning_rate": 1.1739890889962844e-05, "loss": 0.4877, "step": 12940 }, { "epoch": 0.46105064396031137, "grad_norm": 1.1745263817785843, "learning_rate": 1.1738754542012688e-05, "loss": 0.4963, "step": 12941 }, { "epoch": 0.4610862710868055, "grad_norm": 1.3960649456548362, "learning_rate": 1.1737618170909797e-05, "loss": 0.5373, "step": 12942 }, { "epoch": 0.46112189821329963, "grad_norm": 1.7276400520234463, "learning_rate": 1.1736481776669307e-05, "loss": 0.6684, "step": 12943 }, { "epoch": 0.4611575253397937, "grad_norm": 1.6178243358353908, "learning_rate": 1.1735345359306346e-05, "loss": 0.4943, "step": 12944 }, { "epoch": 0.46119315246628784, "grad_norm": 1.123832717230848, "learning_rate": 1.173420891883605e-05, "loss": 0.4688, "step": 12945 }, { "epoch": 0.46122877959278197, "grad_norm": 2.0170271222238156, "learning_rate": 1.1733072455273553e-05, "loss": 0.5523, "step": 12946 }, { "epoch": 0.46126440671927604, "grad_norm": 1.5032221801092918, "learning_rate": 1.173193596863398e-05, "loss": 0.5892, "step": 12947 }, { "epoch": 0.4613000338457702, "grad_norm": 1.65936365112131, "learning_rate": 1.1730799458932473e-05, "loss": 0.597, "step": 12948 }, { "epoch": 0.4613356609722643, "grad_norm": 1.543334335773705, "learning_rate": 1.172966292618416e-05, "loss": 0.5785, "step": 12949 }, { "epoch": 0.4613712880987584, "grad_norm": 1.0598788255068046, "learning_rate": 1.1728526370404176e-05, "loss": 0.4443, "step": 12950 }, { "epoch": 0.4614069152252525, "grad_norm": 1.1665199638769195, "learning_rate": 1.1727389791607655e-05, "loss": 0.5288, "step": 12951 }, { "epoch": 0.46144254235174664, "grad_norm": 1.0015176461352313, "learning_rate": 1.1726253189809732e-05, "loss": 0.4194, "step": 12952 }, { "epoch": 0.4614781694782407, "grad_norm": 1.069738123958038, "learning_rate": 1.1725116565025544e-05, "loss": 0.4186, "step": 12953 }, { "epoch": 0.46151379660473485, "grad_norm": 1.0535151173279922, "learning_rate": 1.1723979917270218e-05, "loss": 0.4795, "step": 12954 }, { "epoch": 0.461549423731229, "grad_norm": 1.9428890758215625, "learning_rate": 1.1722843246558899e-05, "loss": 0.5123, "step": 12955 }, { "epoch": 0.46158505085772306, "grad_norm": 1.0269743043947253, "learning_rate": 1.1721706552906716e-05, "loss": 0.4719, "step": 12956 }, { "epoch": 0.4616206779842172, "grad_norm": 1.0177996508529552, "learning_rate": 1.1720569836328805e-05, "loss": 0.4497, "step": 12957 }, { "epoch": 0.4616563051107113, "grad_norm": 2.2554384639443756, "learning_rate": 1.1719433096840305e-05, "loss": 0.5924, "step": 12958 }, { "epoch": 0.4616919322372054, "grad_norm": 2.743154554946443, "learning_rate": 1.1718296334456353e-05, "loss": 0.5643, "step": 12959 }, { "epoch": 0.4617275593636995, "grad_norm": 1.0468157484276035, "learning_rate": 1.1717159549192085e-05, "loss": 0.4231, "step": 12960 }, { "epoch": 0.46176318649019366, "grad_norm": 1.2855141363662033, "learning_rate": 1.1716022741062637e-05, "loss": 0.5823, "step": 12961 }, { "epoch": 0.46179881361668773, "grad_norm": 2.3810004249871066, "learning_rate": 1.1714885910083143e-05, "loss": 0.5959, "step": 12962 }, { "epoch": 0.46183444074318186, "grad_norm": 1.3343914429400072, "learning_rate": 1.1713749056268748e-05, "loss": 0.509, "step": 12963 }, { "epoch": 0.461870067869676, "grad_norm": 1.572778815138531, "learning_rate": 1.1712612179634587e-05, "loss": 0.59, "step": 12964 }, { "epoch": 0.46190569499617007, "grad_norm": 1.2914816251280572, "learning_rate": 1.1711475280195795e-05, "loss": 0.4566, "step": 12965 }, { "epoch": 0.4619413221226642, "grad_norm": 1.0155362835167185, "learning_rate": 1.1710338357967514e-05, "loss": 0.4191, "step": 12966 }, { "epoch": 0.46197694924915833, "grad_norm": 1.6314879139819958, "learning_rate": 1.1709201412964883e-05, "loss": 0.5847, "step": 12967 }, { "epoch": 0.4620125763756524, "grad_norm": 1.0768943481775162, "learning_rate": 1.1708064445203042e-05, "loss": 0.48, "step": 12968 }, { "epoch": 0.46204820350214654, "grad_norm": 1.185624714022573, "learning_rate": 1.1706927454697125e-05, "loss": 0.4387, "step": 12969 }, { "epoch": 0.46208383062864067, "grad_norm": 1.4439451905454466, "learning_rate": 1.1705790441462281e-05, "loss": 0.5485, "step": 12970 }, { "epoch": 0.46211945775513474, "grad_norm": 1.0996536088572502, "learning_rate": 1.1704653405513642e-05, "loss": 0.4623, "step": 12971 }, { "epoch": 0.4621550848816289, "grad_norm": 1.2249009970620255, "learning_rate": 1.1703516346866353e-05, "loss": 0.4536, "step": 12972 }, { "epoch": 0.462190712008123, "grad_norm": 1.7969847971856314, "learning_rate": 1.1702379265535551e-05, "loss": 0.652, "step": 12973 }, { "epoch": 0.4622263391346171, "grad_norm": 1.0748055232274234, "learning_rate": 1.1701242161536382e-05, "loss": 0.4244, "step": 12974 }, { "epoch": 0.4622619662611112, "grad_norm": 1.098409209054133, "learning_rate": 1.1700105034883982e-05, "loss": 0.4475, "step": 12975 }, { "epoch": 0.46229759338760534, "grad_norm": 2.046707967260112, "learning_rate": 1.1698967885593494e-05, "loss": 0.5594, "step": 12976 }, { "epoch": 0.4623332205140994, "grad_norm": 1.1122637713197538, "learning_rate": 1.1697830713680065e-05, "loss": 0.4627, "step": 12977 }, { "epoch": 0.46236884764059355, "grad_norm": 1.0306272715386549, "learning_rate": 1.1696693519158833e-05, "loss": 0.4756, "step": 12978 }, { "epoch": 0.4624044747670877, "grad_norm": 2.6936116391128775, "learning_rate": 1.1695556302044937e-05, "loss": 0.5615, "step": 12979 }, { "epoch": 0.46244010189358176, "grad_norm": 1.1280036323057903, "learning_rate": 1.1694419062353525e-05, "loss": 0.4383, "step": 12980 }, { "epoch": 0.4624757290200759, "grad_norm": 1.6566777797399632, "learning_rate": 1.1693281800099741e-05, "loss": 0.5311, "step": 12981 }, { "epoch": 0.46251135614657, "grad_norm": 1.5203200434545812, "learning_rate": 1.1692144515298724e-05, "loss": 0.5575, "step": 12982 }, { "epoch": 0.4625469832730641, "grad_norm": 1.9499482523391949, "learning_rate": 1.1691007207965623e-05, "loss": 0.5988, "step": 12983 }, { "epoch": 0.4625826103995582, "grad_norm": 1.0335206771700969, "learning_rate": 1.1689869878115577e-05, "loss": 0.4703, "step": 12984 }, { "epoch": 0.46261823752605236, "grad_norm": 1.4674606568777777, "learning_rate": 1.1688732525763733e-05, "loss": 0.5252, "step": 12985 }, { "epoch": 0.46265386465254643, "grad_norm": 1.0229160673356206, "learning_rate": 1.1687595150925235e-05, "loss": 0.4125, "step": 12986 }, { "epoch": 0.46268949177904056, "grad_norm": 1.468068644745651, "learning_rate": 1.1686457753615228e-05, "loss": 0.6177, "step": 12987 }, { "epoch": 0.4627251189055347, "grad_norm": 1.153823942705121, "learning_rate": 1.1685320333848857e-05, "loss": 0.4723, "step": 12988 }, { "epoch": 0.46276074603202877, "grad_norm": 1.556636020936235, "learning_rate": 1.168418289164127e-05, "loss": 0.5677, "step": 12989 }, { "epoch": 0.4627963731585229, "grad_norm": 1.112314459248975, "learning_rate": 1.1683045427007604e-05, "loss": 0.4125, "step": 12990 }, { "epoch": 0.46283200028501703, "grad_norm": 1.7615607591798572, "learning_rate": 1.1681907939963017e-05, "loss": 0.5145, "step": 12991 }, { "epoch": 0.4628676274115111, "grad_norm": 1.1394503415479784, "learning_rate": 1.168077043052265e-05, "loss": 0.4285, "step": 12992 }, { "epoch": 0.46290325453800524, "grad_norm": 1.3974863870050582, "learning_rate": 1.1679632898701649e-05, "loss": 0.6516, "step": 12993 }, { "epoch": 0.46293888166449937, "grad_norm": 1.7060599479436163, "learning_rate": 1.1678495344515164e-05, "loss": 0.5914, "step": 12994 }, { "epoch": 0.46297450879099344, "grad_norm": 1.128050720243874, "learning_rate": 1.1677357767978337e-05, "loss": 0.4805, "step": 12995 }, { "epoch": 0.4630101359174876, "grad_norm": 1.0545633035653128, "learning_rate": 1.1676220169106321e-05, "loss": 0.4371, "step": 12996 }, { "epoch": 0.4630457630439817, "grad_norm": 1.0374937596629124, "learning_rate": 1.1675082547914263e-05, "loss": 0.4538, "step": 12997 }, { "epoch": 0.4630813901704758, "grad_norm": 1.102095473179193, "learning_rate": 1.1673944904417309e-05, "loss": 0.4176, "step": 12998 }, { "epoch": 0.4631170172969699, "grad_norm": 1.6638830358538548, "learning_rate": 1.1672807238630606e-05, "loss": 0.5856, "step": 12999 }, { "epoch": 0.46315264442346404, "grad_norm": 1.0933803373228388, "learning_rate": 1.1671669550569309e-05, "loss": 0.4489, "step": 13000 }, { "epoch": 0.4631882715499581, "grad_norm": 1.5256897840781014, "learning_rate": 1.1670531840248562e-05, "loss": 0.5293, "step": 13001 }, { "epoch": 0.46322389867645225, "grad_norm": 1.2715116729435785, "learning_rate": 1.1669394107683514e-05, "loss": 0.6383, "step": 13002 }, { "epoch": 0.4632595258029464, "grad_norm": 1.146566486030506, "learning_rate": 1.1668256352889317e-05, "loss": 0.4683, "step": 13003 }, { "epoch": 0.46329515292944046, "grad_norm": 1.0576103693832983, "learning_rate": 1.1667118575881122e-05, "loss": 0.4412, "step": 13004 }, { "epoch": 0.4633307800559346, "grad_norm": 1.887870125406451, "learning_rate": 1.166598077667408e-05, "loss": 0.6233, "step": 13005 }, { "epoch": 0.4633664071824287, "grad_norm": 1.102229407939846, "learning_rate": 1.1664842955283335e-05, "loss": 0.4189, "step": 13006 }, { "epoch": 0.4634020343089228, "grad_norm": 1.5239657700649394, "learning_rate": 1.1663705111724046e-05, "loss": 0.5683, "step": 13007 }, { "epoch": 0.4634376614354169, "grad_norm": 1.846947317807615, "learning_rate": 1.166256724601136e-05, "loss": 0.6555, "step": 13008 }, { "epoch": 0.46347328856191106, "grad_norm": 1.1554983471882552, "learning_rate": 1.166142935816043e-05, "loss": 0.4335, "step": 13009 }, { "epoch": 0.46350891568840513, "grad_norm": 1.0496747127487105, "learning_rate": 1.1660291448186404e-05, "loss": 0.437, "step": 13010 }, { "epoch": 0.46354454281489926, "grad_norm": 1.3676177468816735, "learning_rate": 1.165915351610444e-05, "loss": 0.5199, "step": 13011 }, { "epoch": 0.4635801699413934, "grad_norm": 1.511716569346073, "learning_rate": 1.1658015561929686e-05, "loss": 0.582, "step": 13012 }, { "epoch": 0.46361579706788747, "grad_norm": 1.3950427781333634, "learning_rate": 1.1656877585677293e-05, "loss": 0.5244, "step": 13013 }, { "epoch": 0.4636514241943816, "grad_norm": 1.1617474376193493, "learning_rate": 1.1655739587362418e-05, "loss": 0.4942, "step": 13014 }, { "epoch": 0.46368705132087573, "grad_norm": 1.4706434791932717, "learning_rate": 1.1654601567000214e-05, "loss": 0.5228, "step": 13015 }, { "epoch": 0.4637226784473698, "grad_norm": 1.058010718896029, "learning_rate": 1.1653463524605834e-05, "loss": 0.4655, "step": 13016 }, { "epoch": 0.46375830557386394, "grad_norm": 1.1163262723459464, "learning_rate": 1.1652325460194434e-05, "loss": 0.4858, "step": 13017 }, { "epoch": 0.46379393270035807, "grad_norm": 1.5563845239241236, "learning_rate": 1.1651187373781161e-05, "loss": 0.5802, "step": 13018 }, { "epoch": 0.46382955982685214, "grad_norm": 1.268047794541899, "learning_rate": 1.1650049265381178e-05, "loss": 0.5542, "step": 13019 }, { "epoch": 0.4638651869533463, "grad_norm": 1.3497944999071594, "learning_rate": 1.1648911135009635e-05, "loss": 0.5899, "step": 13020 }, { "epoch": 0.4639008140798404, "grad_norm": 1.4573855577921557, "learning_rate": 1.1647772982681685e-05, "loss": 0.5543, "step": 13021 }, { "epoch": 0.4639364412063345, "grad_norm": 1.17229321127601, "learning_rate": 1.1646634808412488e-05, "loss": 0.5293, "step": 13022 }, { "epoch": 0.4639720683328286, "grad_norm": 1.1003427968661512, "learning_rate": 1.1645496612217198e-05, "loss": 0.4374, "step": 13023 }, { "epoch": 0.46400769545932274, "grad_norm": 1.2358220397005886, "learning_rate": 1.1644358394110971e-05, "loss": 0.4914, "step": 13024 }, { "epoch": 0.4640433225858168, "grad_norm": 1.4135285403644575, "learning_rate": 1.1643220154108956e-05, "loss": 0.4769, "step": 13025 }, { "epoch": 0.46407894971231095, "grad_norm": 1.3881468346300487, "learning_rate": 1.1642081892226322e-05, "loss": 0.4624, "step": 13026 }, { "epoch": 0.4641145768388051, "grad_norm": 1.2665314366211329, "learning_rate": 1.1640943608478222e-05, "loss": 0.4476, "step": 13027 }, { "epoch": 0.46415020396529916, "grad_norm": 1.0395160845460667, "learning_rate": 1.1639805302879807e-05, "loss": 0.4799, "step": 13028 }, { "epoch": 0.4641858310917933, "grad_norm": 1.528845949440754, "learning_rate": 1.1638666975446238e-05, "loss": 0.568, "step": 13029 }, { "epoch": 0.4642214582182874, "grad_norm": 1.4667960971622722, "learning_rate": 1.1637528626192676e-05, "loss": 0.5278, "step": 13030 }, { "epoch": 0.4642570853447815, "grad_norm": 1.0803472845563542, "learning_rate": 1.1636390255134273e-05, "loss": 0.4063, "step": 13031 }, { "epoch": 0.4642927124712756, "grad_norm": 1.3657440114264374, "learning_rate": 1.1635251862286189e-05, "loss": 0.56, "step": 13032 }, { "epoch": 0.46432833959776976, "grad_norm": 1.1382926592342282, "learning_rate": 1.1634113447663587e-05, "loss": 0.4229, "step": 13033 }, { "epoch": 0.46436396672426383, "grad_norm": 1.1018817109876222, "learning_rate": 1.1632975011281622e-05, "loss": 0.4504, "step": 13034 }, { "epoch": 0.46439959385075796, "grad_norm": 1.0020038857442461, "learning_rate": 1.1631836553155454e-05, "loss": 0.4295, "step": 13035 }, { "epoch": 0.4644352209772521, "grad_norm": 1.5626576786827329, "learning_rate": 1.1630698073300237e-05, "loss": 0.6502, "step": 13036 }, { "epoch": 0.46447084810374617, "grad_norm": 1.0980527746933069, "learning_rate": 1.1629559571731136e-05, "loss": 0.48, "step": 13037 }, { "epoch": 0.4645064752302403, "grad_norm": 2.5122675528259553, "learning_rate": 1.1628421048463315e-05, "loss": 0.5533, "step": 13038 }, { "epoch": 0.46454210235673443, "grad_norm": 1.6007941096944327, "learning_rate": 1.1627282503511927e-05, "loss": 0.5608, "step": 13039 }, { "epoch": 0.4645777294832285, "grad_norm": 1.1166669245076506, "learning_rate": 1.1626143936892133e-05, "loss": 0.4216, "step": 13040 }, { "epoch": 0.46461335660972264, "grad_norm": 2.4380555381961706, "learning_rate": 1.1625005348619098e-05, "loss": 0.6298, "step": 13041 }, { "epoch": 0.46464898373621677, "grad_norm": 1.3366233599852455, "learning_rate": 1.1623866738707983e-05, "loss": 0.6193, "step": 13042 }, { "epoch": 0.46468461086271085, "grad_norm": 1.4417517930984118, "learning_rate": 1.1622728107173947e-05, "loss": 0.5464, "step": 13043 }, { "epoch": 0.464720237989205, "grad_norm": 1.1926730161725834, "learning_rate": 1.1621589454032147e-05, "loss": 0.4597, "step": 13044 }, { "epoch": 0.4647558651156991, "grad_norm": 2.072917189665526, "learning_rate": 1.1620450779297753e-05, "loss": 0.5514, "step": 13045 }, { "epoch": 0.4647914922421932, "grad_norm": 1.142847029274792, "learning_rate": 1.1619312082985927e-05, "loss": 0.483, "step": 13046 }, { "epoch": 0.4648271193686873, "grad_norm": 1.4321555889429558, "learning_rate": 1.1618173365111825e-05, "loss": 0.5331, "step": 13047 }, { "epoch": 0.46486274649518144, "grad_norm": 4.116354611310188, "learning_rate": 1.1617034625690613e-05, "loss": 0.5418, "step": 13048 }, { "epoch": 0.4648983736216755, "grad_norm": 1.9176706943493673, "learning_rate": 1.1615895864737459e-05, "loss": 0.5907, "step": 13049 }, { "epoch": 0.46493400074816965, "grad_norm": 1.258875119636866, "learning_rate": 1.1614757082267518e-05, "loss": 0.4914, "step": 13050 }, { "epoch": 0.4649696278746638, "grad_norm": 1.4514075722058524, "learning_rate": 1.1613618278295959e-05, "loss": 0.5707, "step": 13051 }, { "epoch": 0.46500525500115786, "grad_norm": 1.1715140541766482, "learning_rate": 1.1612479452837946e-05, "loss": 0.4421, "step": 13052 }, { "epoch": 0.465040882127652, "grad_norm": 1.1192676854422694, "learning_rate": 1.1611340605908643e-05, "loss": 0.5313, "step": 13053 }, { "epoch": 0.4650765092541461, "grad_norm": 1.1157484567267393, "learning_rate": 1.1610201737523212e-05, "loss": 0.4579, "step": 13054 }, { "epoch": 0.4651121363806402, "grad_norm": 1.1053469950300618, "learning_rate": 1.160906284769682e-05, "loss": 0.4447, "step": 13055 }, { "epoch": 0.4651477635071343, "grad_norm": 1.4527482575603199, "learning_rate": 1.160792393644463e-05, "loss": 0.4697, "step": 13056 }, { "epoch": 0.46518339063362846, "grad_norm": 2.0165940933677984, "learning_rate": 1.160678500378181e-05, "loss": 0.5491, "step": 13057 }, { "epoch": 0.46521901776012253, "grad_norm": 1.4155892799664196, "learning_rate": 1.1605646049723523e-05, "loss": 0.4758, "step": 13058 }, { "epoch": 0.46525464488661666, "grad_norm": 1.4019962226317007, "learning_rate": 1.1604507074284938e-05, "loss": 0.563, "step": 13059 }, { "epoch": 0.4652902720131108, "grad_norm": 1.2429934530210751, "learning_rate": 1.1603368077481218e-05, "loss": 0.4487, "step": 13060 }, { "epoch": 0.46532589913960487, "grad_norm": 1.1911866812201377, "learning_rate": 1.1602229059327534e-05, "loss": 0.4791, "step": 13061 }, { "epoch": 0.465361526266099, "grad_norm": 1.2168201458722718, "learning_rate": 1.1601090019839049e-05, "loss": 0.5269, "step": 13062 }, { "epoch": 0.46539715339259313, "grad_norm": 1.147201355723049, "learning_rate": 1.1599950959030932e-05, "loss": 0.4446, "step": 13063 }, { "epoch": 0.4654327805190872, "grad_norm": 1.26442377221287, "learning_rate": 1.159881187691835e-05, "loss": 0.4618, "step": 13064 }, { "epoch": 0.46546840764558134, "grad_norm": 1.9974479316389249, "learning_rate": 1.159767277351647e-05, "loss": 0.4726, "step": 13065 }, { "epoch": 0.46550403477207547, "grad_norm": 1.0814736815943813, "learning_rate": 1.1596533648840458e-05, "loss": 0.4511, "step": 13066 }, { "epoch": 0.46553966189856955, "grad_norm": 1.325326202040977, "learning_rate": 1.1595394502905488e-05, "loss": 0.5571, "step": 13067 }, { "epoch": 0.4655752890250637, "grad_norm": 1.584755582464911, "learning_rate": 1.1594255335726725e-05, "loss": 0.5687, "step": 13068 }, { "epoch": 0.4656109161515578, "grad_norm": 1.6868701563673218, "learning_rate": 1.1593116147319334e-05, "loss": 0.5792, "step": 13069 }, { "epoch": 0.4656465432780519, "grad_norm": 1.4655742978259036, "learning_rate": 1.1591976937698493e-05, "loss": 0.5214, "step": 13070 }, { "epoch": 0.465682170404546, "grad_norm": 1.8670652872141653, "learning_rate": 1.1590837706879361e-05, "loss": 0.5304, "step": 13071 }, { "epoch": 0.46571779753104015, "grad_norm": 1.1460721851962767, "learning_rate": 1.1589698454877116e-05, "loss": 0.4478, "step": 13072 }, { "epoch": 0.4657534246575342, "grad_norm": 1.1711951160897982, "learning_rate": 1.1588559181706926e-05, "loss": 0.4284, "step": 13073 }, { "epoch": 0.46578905178402835, "grad_norm": 1.6454821650894276, "learning_rate": 1.1587419887383958e-05, "loss": 0.6182, "step": 13074 }, { "epoch": 0.4658246789105225, "grad_norm": 1.1966890149084188, "learning_rate": 1.1586280571923385e-05, "loss": 0.4342, "step": 13075 }, { "epoch": 0.46586030603701656, "grad_norm": 1.0990991280391893, "learning_rate": 1.158514123534038e-05, "loss": 0.5044, "step": 13076 }, { "epoch": 0.4658959331635107, "grad_norm": 1.5717420992206952, "learning_rate": 1.1584001877650109e-05, "loss": 0.5207, "step": 13077 }, { "epoch": 0.4659315602900048, "grad_norm": 1.4770543069790583, "learning_rate": 1.1582862498867747e-05, "loss": 0.6253, "step": 13078 }, { "epoch": 0.4659671874164989, "grad_norm": 1.6587862598779086, "learning_rate": 1.1581723099008464e-05, "loss": 0.5818, "step": 13079 }, { "epoch": 0.466002814542993, "grad_norm": 1.192233895029396, "learning_rate": 1.158058367808743e-05, "loss": 0.4753, "step": 13080 }, { "epoch": 0.46603844166948716, "grad_norm": 3.4054001843645705, "learning_rate": 1.1579444236119823e-05, "loss": 0.5094, "step": 13081 }, { "epoch": 0.46607406879598123, "grad_norm": 1.096487905846978, "learning_rate": 1.1578304773120811e-05, "loss": 0.4459, "step": 13082 }, { "epoch": 0.46610969592247536, "grad_norm": 1.1857135255167472, "learning_rate": 1.1577165289105565e-05, "loss": 0.4796, "step": 13083 }, { "epoch": 0.4661453230489695, "grad_norm": 1.1237581024202015, "learning_rate": 1.1576025784089264e-05, "loss": 0.4737, "step": 13084 }, { "epoch": 0.46618095017546357, "grad_norm": 1.1567095831114271, "learning_rate": 1.1574886258087077e-05, "loss": 0.4382, "step": 13085 }, { "epoch": 0.4662165773019577, "grad_norm": 1.2958242146972303, "learning_rate": 1.157374671111418e-05, "loss": 0.4654, "step": 13086 }, { "epoch": 0.46625220442845183, "grad_norm": 1.1903428284893467, "learning_rate": 1.1572607143185745e-05, "loss": 0.4433, "step": 13087 }, { "epoch": 0.4662878315549459, "grad_norm": 1.189187181134927, "learning_rate": 1.1571467554316947e-05, "loss": 0.5143, "step": 13088 }, { "epoch": 0.46632345868144004, "grad_norm": 1.8536300972948947, "learning_rate": 1.1570327944522959e-05, "loss": 0.5629, "step": 13089 }, { "epoch": 0.46635908580793417, "grad_norm": 1.179342013643964, "learning_rate": 1.1569188313818958e-05, "loss": 0.4171, "step": 13090 }, { "epoch": 0.46639471293442825, "grad_norm": 1.3038764741505724, "learning_rate": 1.1568048662220117e-05, "loss": 0.4455, "step": 13091 }, { "epoch": 0.4664303400609224, "grad_norm": 1.4518888556233016, "learning_rate": 1.1566908989741612e-05, "loss": 0.5634, "step": 13092 }, { "epoch": 0.4664659671874165, "grad_norm": 1.162391402777392, "learning_rate": 1.156576929639862e-05, "loss": 0.4741, "step": 13093 }, { "epoch": 0.46650159431391064, "grad_norm": 1.5269021553768007, "learning_rate": 1.1564629582206313e-05, "loss": 0.5222, "step": 13094 }, { "epoch": 0.4665372214404047, "grad_norm": 1.169831854714522, "learning_rate": 1.1563489847179872e-05, "loss": 0.4569, "step": 13095 }, { "epoch": 0.46657284856689885, "grad_norm": 1.1543970761668692, "learning_rate": 1.1562350091334466e-05, "loss": 0.4689, "step": 13096 }, { "epoch": 0.466608475693393, "grad_norm": 1.200070450582371, "learning_rate": 1.1561210314685282e-05, "loss": 0.4449, "step": 13097 }, { "epoch": 0.46664410281988705, "grad_norm": 1.6010346172066867, "learning_rate": 1.156007051724749e-05, "loss": 0.5545, "step": 13098 }, { "epoch": 0.4666797299463812, "grad_norm": 1.5319746108672103, "learning_rate": 1.155893069903627e-05, "loss": 0.5607, "step": 13099 }, { "epoch": 0.4667153570728753, "grad_norm": 1.8089043134038985, "learning_rate": 1.1557790860066796e-05, "loss": 0.5675, "step": 13100 }, { "epoch": 0.4667509841993694, "grad_norm": 1.1128694210248302, "learning_rate": 1.1556651000354246e-05, "loss": 0.415, "step": 13101 }, { "epoch": 0.4667866113258635, "grad_norm": 1.741387226565179, "learning_rate": 1.1555511119913805e-05, "loss": 0.6237, "step": 13102 }, { "epoch": 0.46682223845235765, "grad_norm": 1.363600536959231, "learning_rate": 1.1554371218760641e-05, "loss": 0.4688, "step": 13103 }, { "epoch": 0.4668578655788517, "grad_norm": 1.2368266172271702, "learning_rate": 1.1553231296909942e-05, "loss": 0.4773, "step": 13104 }, { "epoch": 0.46689349270534586, "grad_norm": 1.739786397893273, "learning_rate": 1.155209135437688e-05, "loss": 0.5805, "step": 13105 }, { "epoch": 0.46692911983184, "grad_norm": 1.0947728007885174, "learning_rate": 1.1550951391176636e-05, "loss": 0.5162, "step": 13106 }, { "epoch": 0.46696474695833406, "grad_norm": 1.79946919755114, "learning_rate": 1.154981140732439e-05, "loss": 0.6361, "step": 13107 }, { "epoch": 0.4670003740848282, "grad_norm": 1.2336877712960541, "learning_rate": 1.1548671402835325e-05, "loss": 0.5646, "step": 13108 }, { "epoch": 0.4670360012113223, "grad_norm": 1.1654948059829715, "learning_rate": 1.1547531377724616e-05, "loss": 0.4797, "step": 13109 }, { "epoch": 0.4670716283378164, "grad_norm": 1.9413582915857677, "learning_rate": 1.1546391332007448e-05, "loss": 0.486, "step": 13110 }, { "epoch": 0.46710725546431053, "grad_norm": 1.4993877753947267, "learning_rate": 1.1545251265698993e-05, "loss": 0.5552, "step": 13111 }, { "epoch": 0.46714288259080466, "grad_norm": 1.221449518201707, "learning_rate": 1.154411117881444e-05, "loss": 0.5322, "step": 13112 }, { "epoch": 0.46717850971729874, "grad_norm": 1.403753417885277, "learning_rate": 1.1542971071368968e-05, "loss": 0.5462, "step": 13113 }, { "epoch": 0.46721413684379287, "grad_norm": 1.246969960738947, "learning_rate": 1.1541830943377758e-05, "loss": 0.4624, "step": 13114 }, { "epoch": 0.467249763970287, "grad_norm": 1.480290637261792, "learning_rate": 1.1540690794855991e-05, "loss": 0.5616, "step": 13115 }, { "epoch": 0.4672853910967811, "grad_norm": 1.3878875560190567, "learning_rate": 1.153955062581885e-05, "loss": 0.5485, "step": 13116 }, { "epoch": 0.4673210182232752, "grad_norm": 2.1325968568325226, "learning_rate": 1.1538410436281513e-05, "loss": 0.5737, "step": 13117 }, { "epoch": 0.46735664534976934, "grad_norm": 1.1670903214769346, "learning_rate": 1.153727022625917e-05, "loss": 0.4684, "step": 13118 }, { "epoch": 0.4673922724762634, "grad_norm": 1.2521287955028184, "learning_rate": 1.1536129995766995e-05, "loss": 0.6119, "step": 13119 }, { "epoch": 0.46742789960275755, "grad_norm": 1.1012564240385734, "learning_rate": 1.153498974482018e-05, "loss": 0.4826, "step": 13120 }, { "epoch": 0.4674635267292517, "grad_norm": 1.470741180904836, "learning_rate": 1.1533849473433901e-05, "loss": 0.6287, "step": 13121 }, { "epoch": 0.46749915385574575, "grad_norm": 1.3571180661068665, "learning_rate": 1.1532709181623342e-05, "loss": 0.6342, "step": 13122 }, { "epoch": 0.4675347809822399, "grad_norm": 1.5171759979443122, "learning_rate": 1.1531568869403693e-05, "loss": 0.5842, "step": 13123 }, { "epoch": 0.467570408108734, "grad_norm": 1.0074750332475877, "learning_rate": 1.1530428536790134e-05, "loss": 0.398, "step": 13124 }, { "epoch": 0.4676060352352281, "grad_norm": 1.1722424864670713, "learning_rate": 1.1529288183797849e-05, "loss": 0.4394, "step": 13125 }, { "epoch": 0.4676416623617222, "grad_norm": 1.751091657818038, "learning_rate": 1.1528147810442021e-05, "loss": 0.5963, "step": 13126 }, { "epoch": 0.46767728948821635, "grad_norm": 1.2295555273776548, "learning_rate": 1.1527007416737836e-05, "loss": 0.4197, "step": 13127 }, { "epoch": 0.46771291661471043, "grad_norm": 1.5771626788704858, "learning_rate": 1.1525867002700484e-05, "loss": 0.6117, "step": 13128 }, { "epoch": 0.46774854374120456, "grad_norm": 1.1375770612688982, "learning_rate": 1.1524726568345145e-05, "loss": 0.4763, "step": 13129 }, { "epoch": 0.4677841708676987, "grad_norm": 1.9934336786452453, "learning_rate": 1.1523586113687001e-05, "loss": 0.5731, "step": 13130 }, { "epoch": 0.46781979799419277, "grad_norm": 1.8058217805872974, "learning_rate": 1.1522445638741248e-05, "loss": 0.4953, "step": 13131 }, { "epoch": 0.4678554251206869, "grad_norm": 1.2509115075716843, "learning_rate": 1.1521305143523066e-05, "loss": 0.4854, "step": 13132 }, { "epoch": 0.467891052247181, "grad_norm": 1.9895445484825875, "learning_rate": 1.152016462804764e-05, "loss": 0.6144, "step": 13133 }, { "epoch": 0.4679266793736751, "grad_norm": 1.484517390428529, "learning_rate": 1.1519024092330164e-05, "loss": 0.439, "step": 13134 }, { "epoch": 0.46796230650016923, "grad_norm": 1.6745460965249122, "learning_rate": 1.1517883536385822e-05, "loss": 0.5691, "step": 13135 }, { "epoch": 0.46799793362666336, "grad_norm": 1.029817867479503, "learning_rate": 1.1516742960229796e-05, "loss": 0.4803, "step": 13136 }, { "epoch": 0.46803356075315744, "grad_norm": 1.0689959365671955, "learning_rate": 1.1515602363877276e-05, "loss": 0.4219, "step": 13137 }, { "epoch": 0.46806918787965157, "grad_norm": 1.4659962532672985, "learning_rate": 1.1514461747343452e-05, "loss": 0.5562, "step": 13138 }, { "epoch": 0.4681048150061457, "grad_norm": 1.4877965317721664, "learning_rate": 1.1513321110643514e-05, "loss": 0.4792, "step": 13139 }, { "epoch": 0.4681404421326398, "grad_norm": 1.1216990830731883, "learning_rate": 1.1512180453792645e-05, "loss": 0.4686, "step": 13140 }, { "epoch": 0.4681760692591339, "grad_norm": 1.1312768834557363, "learning_rate": 1.1511039776806037e-05, "loss": 0.4659, "step": 13141 }, { "epoch": 0.46821169638562804, "grad_norm": 1.6124130744077894, "learning_rate": 1.1509899079698879e-05, "loss": 0.544, "step": 13142 }, { "epoch": 0.4682473235121221, "grad_norm": 1.9419522566824614, "learning_rate": 1.1508758362486358e-05, "loss": 0.5664, "step": 13143 }, { "epoch": 0.46828295063861625, "grad_norm": 1.3965661103423563, "learning_rate": 1.1507617625183664e-05, "loss": 0.5331, "step": 13144 }, { "epoch": 0.4683185777651104, "grad_norm": 1.7294338617819744, "learning_rate": 1.1506476867805992e-05, "loss": 0.656, "step": 13145 }, { "epoch": 0.46835420489160445, "grad_norm": 1.4515221825249929, "learning_rate": 1.1505336090368524e-05, "loss": 0.543, "step": 13146 }, { "epoch": 0.4683898320180986, "grad_norm": 1.1306797264139339, "learning_rate": 1.1504195292886455e-05, "loss": 0.4743, "step": 13147 }, { "epoch": 0.4684254591445927, "grad_norm": 1.2578338773531004, "learning_rate": 1.1503054475374972e-05, "loss": 0.4773, "step": 13148 }, { "epoch": 0.4684610862710868, "grad_norm": 1.0666672863210203, "learning_rate": 1.1501913637849272e-05, "loss": 0.4275, "step": 13149 }, { "epoch": 0.4684967133975809, "grad_norm": 2.2164035701135574, "learning_rate": 1.150077278032454e-05, "loss": 0.5871, "step": 13150 }, { "epoch": 0.46853234052407505, "grad_norm": 1.1700905534265862, "learning_rate": 1.1499631902815968e-05, "loss": 0.5153, "step": 13151 }, { "epoch": 0.46856796765056913, "grad_norm": 1.4760715543363172, "learning_rate": 1.149849100533875e-05, "loss": 0.5802, "step": 13152 }, { "epoch": 0.46860359477706326, "grad_norm": 1.5933093735167931, "learning_rate": 1.1497350087908079e-05, "loss": 0.6001, "step": 13153 }, { "epoch": 0.4686392219035574, "grad_norm": 1.3848499240931833, "learning_rate": 1.1496209150539144e-05, "loss": 0.6061, "step": 13154 }, { "epoch": 0.46867484903005147, "grad_norm": 1.2000516103419447, "learning_rate": 1.1495068193247138e-05, "loss": 0.4465, "step": 13155 }, { "epoch": 0.4687104761565456, "grad_norm": 1.1939619095959955, "learning_rate": 1.1493927216047253e-05, "loss": 0.4912, "step": 13156 }, { "epoch": 0.46874610328303973, "grad_norm": 1.7189412577265137, "learning_rate": 1.1492786218954686e-05, "loss": 0.513, "step": 13157 }, { "epoch": 0.4687817304095338, "grad_norm": 1.0371247157310106, "learning_rate": 1.1491645201984625e-05, "loss": 0.4461, "step": 13158 }, { "epoch": 0.46881735753602793, "grad_norm": 1.0333949007782621, "learning_rate": 1.1490504165152264e-05, "loss": 0.4159, "step": 13159 }, { "epoch": 0.46885298466252207, "grad_norm": 1.6716902463085421, "learning_rate": 1.14893631084728e-05, "loss": 0.5148, "step": 13160 }, { "epoch": 0.46888861178901614, "grad_norm": 1.8129183930745114, "learning_rate": 1.1488222031961428e-05, "loss": 0.539, "step": 13161 }, { "epoch": 0.46892423891551027, "grad_norm": 1.4412124080380766, "learning_rate": 1.1487080935633335e-05, "loss": 0.614, "step": 13162 }, { "epoch": 0.4689598660420044, "grad_norm": 6.004947858781098, "learning_rate": 1.1485939819503717e-05, "loss": 0.5665, "step": 13163 }, { "epoch": 0.4689954931684985, "grad_norm": 1.4266936358815216, "learning_rate": 1.1484798683587777e-05, "loss": 0.6011, "step": 13164 }, { "epoch": 0.4690311202949926, "grad_norm": 1.8546217497432904, "learning_rate": 1.1483657527900705e-05, "loss": 0.4819, "step": 13165 }, { "epoch": 0.46906674742148674, "grad_norm": 1.747805106939253, "learning_rate": 1.1482516352457695e-05, "loss": 0.606, "step": 13166 }, { "epoch": 0.4691023745479808, "grad_norm": 2.045820991353524, "learning_rate": 1.1481375157273942e-05, "loss": 0.5611, "step": 13167 }, { "epoch": 0.46913800167447495, "grad_norm": 1.1162352719514543, "learning_rate": 1.1480233942364645e-05, "loss": 0.4268, "step": 13168 }, { "epoch": 0.4691736288009691, "grad_norm": 1.5364710753073318, "learning_rate": 1.1479092707745e-05, "loss": 0.5525, "step": 13169 }, { "epoch": 0.46920925592746315, "grad_norm": 1.3605579481282828, "learning_rate": 1.1477951453430198e-05, "loss": 0.5358, "step": 13170 }, { "epoch": 0.4692448830539573, "grad_norm": 1.5656809979662467, "learning_rate": 1.147681017943544e-05, "loss": 0.5973, "step": 13171 }, { "epoch": 0.4692805101804514, "grad_norm": 1.6452791530626036, "learning_rate": 1.1475668885775926e-05, "loss": 0.558, "step": 13172 }, { "epoch": 0.4693161373069455, "grad_norm": 1.2827407448173203, "learning_rate": 1.1474527572466847e-05, "loss": 0.6119, "step": 13173 }, { "epoch": 0.4693517644334396, "grad_norm": 1.0958310414035044, "learning_rate": 1.14733862395234e-05, "loss": 0.4335, "step": 13174 }, { "epoch": 0.46938739155993375, "grad_norm": 1.454797699157443, "learning_rate": 1.1472244886960788e-05, "loss": 0.6195, "step": 13175 }, { "epoch": 0.46942301868642783, "grad_norm": 1.1108166861355147, "learning_rate": 1.1471103514794207e-05, "loss": 0.4798, "step": 13176 }, { "epoch": 0.46945864581292196, "grad_norm": 1.4812722468869393, "learning_rate": 1.1469962123038854e-05, "loss": 0.5053, "step": 13177 }, { "epoch": 0.4694942729394161, "grad_norm": 1.8588098037701988, "learning_rate": 1.1468820711709928e-05, "loss": 0.5892, "step": 13178 }, { "epoch": 0.46952990006591017, "grad_norm": 2.1889467607129465, "learning_rate": 1.1467679280822626e-05, "loss": 0.6103, "step": 13179 }, { "epoch": 0.4695655271924043, "grad_norm": 1.2168888909931213, "learning_rate": 1.146653783039215e-05, "loss": 0.4299, "step": 13180 }, { "epoch": 0.46960115431889843, "grad_norm": 1.4966639438355063, "learning_rate": 1.1465396360433698e-05, "loss": 0.5638, "step": 13181 }, { "epoch": 0.4696367814453925, "grad_norm": 1.6467577686730626, "learning_rate": 1.1464254870962467e-05, "loss": 0.6011, "step": 13182 }, { "epoch": 0.46967240857188663, "grad_norm": 1.6267503784111013, "learning_rate": 1.1463113361993663e-05, "loss": 0.6233, "step": 13183 }, { "epoch": 0.46970803569838077, "grad_norm": 2.5719915141548686, "learning_rate": 1.1461971833542481e-05, "loss": 0.5512, "step": 13184 }, { "epoch": 0.46974366282487484, "grad_norm": 1.1205012236608032, "learning_rate": 1.1460830285624119e-05, "loss": 0.4272, "step": 13185 }, { "epoch": 0.469779289951369, "grad_norm": 1.2224569449521698, "learning_rate": 1.145968871825378e-05, "loss": 0.5067, "step": 13186 }, { "epoch": 0.4698149170778631, "grad_norm": 1.564646862688053, "learning_rate": 1.145854713144667e-05, "loss": 0.5034, "step": 13187 }, { "epoch": 0.4698505442043572, "grad_norm": 1.619693856352562, "learning_rate": 1.1457405525217982e-05, "loss": 0.5312, "step": 13188 }, { "epoch": 0.4698861713308513, "grad_norm": 1.0879223874312491, "learning_rate": 1.145626389958292e-05, "loss": 0.4506, "step": 13189 }, { "epoch": 0.46992179845734544, "grad_norm": 1.062088666189598, "learning_rate": 1.1455122254556689e-05, "loss": 0.4579, "step": 13190 }, { "epoch": 0.4699574255838395, "grad_norm": 1.118882116077544, "learning_rate": 1.145398059015449e-05, "loss": 0.4668, "step": 13191 }, { "epoch": 0.46999305271033365, "grad_norm": 1.1581122916614837, "learning_rate": 1.1452838906391519e-05, "loss": 0.453, "step": 13192 }, { "epoch": 0.4700286798368278, "grad_norm": 1.2265313305268977, "learning_rate": 1.1451697203282984e-05, "loss": 0.5073, "step": 13193 }, { "epoch": 0.47006430696332185, "grad_norm": 1.219003231067112, "learning_rate": 1.1450555480844087e-05, "loss": 0.4715, "step": 13194 }, { "epoch": 0.470099934089816, "grad_norm": 1.3351942789166815, "learning_rate": 1.144941373909003e-05, "loss": 0.5296, "step": 13195 }, { "epoch": 0.4701355612163101, "grad_norm": 1.1702843640124088, "learning_rate": 1.1448271978036014e-05, "loss": 0.3952, "step": 13196 }, { "epoch": 0.4701711883428042, "grad_norm": 1.4700359532244212, "learning_rate": 1.1447130197697246e-05, "loss": 0.4631, "step": 13197 }, { "epoch": 0.4702068154692983, "grad_norm": 1.8554772875506804, "learning_rate": 1.1445988398088925e-05, "loss": 0.4677, "step": 13198 }, { "epoch": 0.47024244259579245, "grad_norm": 2.0025951113453457, "learning_rate": 1.1444846579226261e-05, "loss": 0.4955, "step": 13199 }, { "epoch": 0.47027806972228653, "grad_norm": 2.119191404798325, "learning_rate": 1.1443704741124455e-05, "loss": 0.5414, "step": 13200 }, { "epoch": 0.47031369684878066, "grad_norm": 1.4976981114429946, "learning_rate": 1.144256288379871e-05, "loss": 0.5864, "step": 13201 }, { "epoch": 0.4703493239752748, "grad_norm": 1.096806828963835, "learning_rate": 1.1441421007264234e-05, "loss": 0.439, "step": 13202 }, { "epoch": 0.47038495110176887, "grad_norm": 1.334629300783704, "learning_rate": 1.1440279111536228e-05, "loss": 0.3994, "step": 13203 }, { "epoch": 0.470420578228263, "grad_norm": 1.1039340414860475, "learning_rate": 1.1439137196629896e-05, "loss": 0.4633, "step": 13204 }, { "epoch": 0.47045620535475713, "grad_norm": 1.5724708225699429, "learning_rate": 1.143799526256045e-05, "loss": 0.5693, "step": 13205 }, { "epoch": 0.4704918324812512, "grad_norm": 1.593191937259384, "learning_rate": 1.1436853309343092e-05, "loss": 0.5105, "step": 13206 }, { "epoch": 0.47052745960774534, "grad_norm": 1.3723566376172558, "learning_rate": 1.1435711336993027e-05, "loss": 0.5876, "step": 13207 }, { "epoch": 0.47056308673423947, "grad_norm": 1.055604263956173, "learning_rate": 1.1434569345525462e-05, "loss": 0.4653, "step": 13208 }, { "epoch": 0.47059871386073354, "grad_norm": 1.7886105150217788, "learning_rate": 1.1433427334955605e-05, "loss": 0.5762, "step": 13209 }, { "epoch": 0.4706343409872277, "grad_norm": 1.4745805479338727, "learning_rate": 1.1432285305298658e-05, "loss": 0.5207, "step": 13210 }, { "epoch": 0.4706699681137218, "grad_norm": 1.2066500188900053, "learning_rate": 1.1431143256569835e-05, "loss": 0.5015, "step": 13211 }, { "epoch": 0.4707055952402159, "grad_norm": 1.5337784373734298, "learning_rate": 1.1430001188784335e-05, "loss": 0.5169, "step": 13212 }, { "epoch": 0.47074122236671, "grad_norm": 1.0738637210727475, "learning_rate": 1.1428859101957372e-05, "loss": 0.4713, "step": 13213 }, { "epoch": 0.47077684949320414, "grad_norm": 1.1036260774354063, "learning_rate": 1.1427716996104152e-05, "loss": 0.4893, "step": 13214 }, { "epoch": 0.4708124766196982, "grad_norm": 1.4473848203127173, "learning_rate": 1.1426574871239879e-05, "loss": 0.5154, "step": 13215 }, { "epoch": 0.47084810374619235, "grad_norm": 2.0664066726426733, "learning_rate": 1.1425432727379769e-05, "loss": 0.5876, "step": 13216 }, { "epoch": 0.4708837308726865, "grad_norm": 1.6501957608960491, "learning_rate": 1.1424290564539022e-05, "loss": 0.5609, "step": 13217 }, { "epoch": 0.47091935799918055, "grad_norm": 1.0225312144173888, "learning_rate": 1.1423148382732854e-05, "loss": 0.4838, "step": 13218 }, { "epoch": 0.4709549851256747, "grad_norm": 1.0274245703695533, "learning_rate": 1.1422006181976468e-05, "loss": 0.4506, "step": 13219 }, { "epoch": 0.4709906122521688, "grad_norm": 1.251357090820544, "learning_rate": 1.1420863962285078e-05, "loss": 0.4829, "step": 13220 }, { "epoch": 0.4710262393786629, "grad_norm": 1.9199137601883804, "learning_rate": 1.1419721723673889e-05, "loss": 0.5557, "step": 13221 }, { "epoch": 0.471061866505157, "grad_norm": 2.080791904034286, "learning_rate": 1.1418579466158113e-05, "loss": 0.5304, "step": 13222 }, { "epoch": 0.47109749363165115, "grad_norm": 1.5900074389864118, "learning_rate": 1.1417437189752961e-05, "loss": 0.5564, "step": 13223 }, { "epoch": 0.47113312075814523, "grad_norm": 1.0150016573999594, "learning_rate": 1.1416294894473641e-05, "loss": 0.4603, "step": 13224 }, { "epoch": 0.47116874788463936, "grad_norm": 1.093957191207167, "learning_rate": 1.1415152580335367e-05, "loss": 0.4477, "step": 13225 }, { "epoch": 0.4712043750111335, "grad_norm": 1.6061164862772306, "learning_rate": 1.1414010247353345e-05, "loss": 0.5977, "step": 13226 }, { "epoch": 0.47124000213762757, "grad_norm": 3.2885872455341394, "learning_rate": 1.1412867895542787e-05, "loss": 0.5372, "step": 13227 }, { "epoch": 0.4712756292641217, "grad_norm": 1.6579338376974397, "learning_rate": 1.1411725524918909e-05, "loss": 0.5971, "step": 13228 }, { "epoch": 0.47131125639061583, "grad_norm": 1.5094362557098429, "learning_rate": 1.1410583135496917e-05, "loss": 0.534, "step": 13229 }, { "epoch": 0.4713468835171099, "grad_norm": 1.647582175828789, "learning_rate": 1.1409440727292025e-05, "loss": 0.5548, "step": 13230 }, { "epoch": 0.47138251064360404, "grad_norm": 1.5122447807683275, "learning_rate": 1.1408298300319445e-05, "loss": 0.5612, "step": 13231 }, { "epoch": 0.47141813777009817, "grad_norm": 1.5688152336679217, "learning_rate": 1.1407155854594387e-05, "loss": 0.5371, "step": 13232 }, { "epoch": 0.47145376489659224, "grad_norm": 1.1150858153093546, "learning_rate": 1.1406013390132067e-05, "loss": 0.5103, "step": 13233 }, { "epoch": 0.4714893920230864, "grad_norm": 1.133815071846072, "learning_rate": 1.1404870906947693e-05, "loss": 0.491, "step": 13234 }, { "epoch": 0.4715250191495805, "grad_norm": 1.6320858089413746, "learning_rate": 1.1403728405056484e-05, "loss": 0.5277, "step": 13235 }, { "epoch": 0.4715606462760746, "grad_norm": 1.3888406599258423, "learning_rate": 1.1402585884473649e-05, "loss": 0.5076, "step": 13236 }, { "epoch": 0.4715962734025687, "grad_norm": 1.2867117189922415, "learning_rate": 1.1401443345214405e-05, "loss": 0.496, "step": 13237 }, { "epoch": 0.47163190052906284, "grad_norm": 1.4893317811747513, "learning_rate": 1.1400300787293961e-05, "loss": 0.5865, "step": 13238 }, { "epoch": 0.4716675276555569, "grad_norm": 1.122176472056194, "learning_rate": 1.1399158210727533e-05, "loss": 0.4527, "step": 13239 }, { "epoch": 0.47170315478205105, "grad_norm": 1.8390264104520848, "learning_rate": 1.1398015615530338e-05, "loss": 0.6629, "step": 13240 }, { "epoch": 0.4717387819085452, "grad_norm": 1.6432795262680957, "learning_rate": 1.1396873001717583e-05, "loss": 0.5919, "step": 13241 }, { "epoch": 0.47177440903503925, "grad_norm": 1.4794420461131494, "learning_rate": 1.1395730369304492e-05, "loss": 0.6353, "step": 13242 }, { "epoch": 0.4718100361615334, "grad_norm": 1.16470790470658, "learning_rate": 1.1394587718306276e-05, "loss": 0.4537, "step": 13243 }, { "epoch": 0.4718456632880275, "grad_norm": 1.9183580974202725, "learning_rate": 1.1393445048738145e-05, "loss": 0.5303, "step": 13244 }, { "epoch": 0.4718812904145216, "grad_norm": 1.5129630127178815, "learning_rate": 1.1392302360615322e-05, "loss": 0.5611, "step": 13245 }, { "epoch": 0.4719169175410157, "grad_norm": 1.5639980327939564, "learning_rate": 1.1391159653953018e-05, "loss": 0.5181, "step": 13246 }, { "epoch": 0.47195254466750985, "grad_norm": 1.46072411766178, "learning_rate": 1.1390016928766456e-05, "loss": 0.5707, "step": 13247 }, { "epoch": 0.47198817179400393, "grad_norm": 1.076691187372251, "learning_rate": 1.1388874185070843e-05, "loss": 0.4692, "step": 13248 }, { "epoch": 0.47202379892049806, "grad_norm": 1.043024392867854, "learning_rate": 1.1387731422881401e-05, "loss": 0.4647, "step": 13249 }, { "epoch": 0.4720594260469922, "grad_norm": 1.922687375925964, "learning_rate": 1.1386588642213346e-05, "loss": 0.5093, "step": 13250 }, { "epoch": 0.47209505317348627, "grad_norm": 1.1002529356428923, "learning_rate": 1.1385445843081892e-05, "loss": 0.4668, "step": 13251 }, { "epoch": 0.4721306802999804, "grad_norm": 1.109592322614588, "learning_rate": 1.1384303025502257e-05, "loss": 0.4882, "step": 13252 }, { "epoch": 0.47216630742647453, "grad_norm": 1.4400201273528996, "learning_rate": 1.1383160189489663e-05, "loss": 0.6081, "step": 13253 }, { "epoch": 0.4722019345529686, "grad_norm": 1.0113745096759037, "learning_rate": 1.1382017335059324e-05, "loss": 0.4845, "step": 13254 }, { "epoch": 0.47223756167946274, "grad_norm": 1.2582883066569748, "learning_rate": 1.1380874462226457e-05, "loss": 0.5326, "step": 13255 }, { "epoch": 0.47227318880595687, "grad_norm": 1.0825816430583997, "learning_rate": 1.137973157100628e-05, "loss": 0.3995, "step": 13256 }, { "epoch": 0.47230881593245094, "grad_norm": 1.1609438023353653, "learning_rate": 1.1378588661414014e-05, "loss": 0.4792, "step": 13257 }, { "epoch": 0.4723444430589451, "grad_norm": 1.3839132898417554, "learning_rate": 1.1377445733464876e-05, "loss": 0.5294, "step": 13258 }, { "epoch": 0.4723800701854392, "grad_norm": 1.1450063577886382, "learning_rate": 1.1376302787174088e-05, "loss": 0.511, "step": 13259 }, { "epoch": 0.4724156973119333, "grad_norm": 1.9810209301617847, "learning_rate": 1.1375159822556863e-05, "loss": 0.6301, "step": 13260 }, { "epoch": 0.4724513244384274, "grad_norm": 1.627613522177309, "learning_rate": 1.1374016839628423e-05, "loss": 0.6016, "step": 13261 }, { "epoch": 0.47248695156492154, "grad_norm": 1.3710274889969505, "learning_rate": 1.1372873838403993e-05, "loss": 0.5833, "step": 13262 }, { "epoch": 0.4725225786914156, "grad_norm": 1.4984545058503826, "learning_rate": 1.1371730818898785e-05, "loss": 0.5694, "step": 13263 }, { "epoch": 0.47255820581790975, "grad_norm": 0.9909920608105128, "learning_rate": 1.1370587781128021e-05, "loss": 0.4471, "step": 13264 }, { "epoch": 0.4725938329444039, "grad_norm": 1.0597612307401558, "learning_rate": 1.1369444725106925e-05, "loss": 0.4671, "step": 13265 }, { "epoch": 0.47262946007089796, "grad_norm": 1.1237704060816038, "learning_rate": 1.1368301650850717e-05, "loss": 0.4624, "step": 13266 }, { "epoch": 0.4726650871973921, "grad_norm": 1.1364683624212768, "learning_rate": 1.1367158558374614e-05, "loss": 0.4213, "step": 13267 }, { "epoch": 0.4727007143238862, "grad_norm": 1.5155601779869894, "learning_rate": 1.1366015447693837e-05, "loss": 0.5563, "step": 13268 }, { "epoch": 0.4727363414503803, "grad_norm": 1.0487166784645714, "learning_rate": 1.1364872318823612e-05, "loss": 0.4626, "step": 13269 }, { "epoch": 0.4727719685768744, "grad_norm": 1.107051205465795, "learning_rate": 1.1363729171779161e-05, "loss": 0.4605, "step": 13270 }, { "epoch": 0.47280759570336856, "grad_norm": 1.0320158122858867, "learning_rate": 1.1362586006575698e-05, "loss": 0.469, "step": 13271 }, { "epoch": 0.47284322282986263, "grad_norm": 1.1945692552996738, "learning_rate": 1.1361442823228453e-05, "loss": 0.488, "step": 13272 }, { "epoch": 0.47287884995635676, "grad_norm": 1.8998143298963481, "learning_rate": 1.1360299621752644e-05, "loss": 0.5498, "step": 13273 }, { "epoch": 0.4729144770828509, "grad_norm": 1.0650108793265962, "learning_rate": 1.1359156402163496e-05, "loss": 0.4952, "step": 13274 }, { "epoch": 0.47295010420934497, "grad_norm": 2.9783374287876265, "learning_rate": 1.1358013164476227e-05, "loss": 0.5739, "step": 13275 }, { "epoch": 0.4729857313358391, "grad_norm": 1.1161050252751181, "learning_rate": 1.1356869908706068e-05, "loss": 0.4539, "step": 13276 }, { "epoch": 0.47302135846233323, "grad_norm": 1.089239822649628, "learning_rate": 1.1355726634868236e-05, "loss": 0.4651, "step": 13277 }, { "epoch": 0.4730569855888273, "grad_norm": 1.8698604213159389, "learning_rate": 1.1354583342977953e-05, "loss": 0.5904, "step": 13278 }, { "epoch": 0.47309261271532144, "grad_norm": 1.1516129812541864, "learning_rate": 1.1353440033050446e-05, "loss": 0.488, "step": 13279 }, { "epoch": 0.47312823984181557, "grad_norm": 1.6883908323983738, "learning_rate": 1.1352296705100944e-05, "loss": 0.5498, "step": 13280 }, { "epoch": 0.4731638669683097, "grad_norm": 1.4772531534979154, "learning_rate": 1.1351153359144664e-05, "loss": 0.6066, "step": 13281 }, { "epoch": 0.4731994940948038, "grad_norm": 0.9335049096557476, "learning_rate": 1.1350009995196834e-05, "loss": 0.4213, "step": 13282 }, { "epoch": 0.4732351212212979, "grad_norm": 1.0577553070489634, "learning_rate": 1.1348866613272675e-05, "loss": 0.4944, "step": 13283 }, { "epoch": 0.47327074834779204, "grad_norm": 1.6568788374077268, "learning_rate": 1.1347723213387416e-05, "loss": 0.5997, "step": 13284 }, { "epoch": 0.4733063754742861, "grad_norm": 1.8810807190456271, "learning_rate": 1.134657979555628e-05, "loss": 0.6575, "step": 13285 }, { "epoch": 0.47334200260078024, "grad_norm": 1.0331078348905813, "learning_rate": 1.134543635979449e-05, "loss": 0.4451, "step": 13286 }, { "epoch": 0.4733776297272744, "grad_norm": 1.6820940216711937, "learning_rate": 1.1344292906117279e-05, "loss": 0.5417, "step": 13287 }, { "epoch": 0.47341325685376845, "grad_norm": 1.589852342694422, "learning_rate": 1.1343149434539866e-05, "loss": 0.5705, "step": 13288 }, { "epoch": 0.4734488839802626, "grad_norm": 2.623231720796235, "learning_rate": 1.1342005945077481e-05, "loss": 0.5785, "step": 13289 }, { "epoch": 0.4734845111067567, "grad_norm": 2.57082467685339, "learning_rate": 1.1340862437745344e-05, "loss": 0.5788, "step": 13290 }, { "epoch": 0.4735201382332508, "grad_norm": 1.1116777293553857, "learning_rate": 1.1339718912558691e-05, "loss": 0.4352, "step": 13291 }, { "epoch": 0.4735557653597449, "grad_norm": 0.9322562904919296, "learning_rate": 1.1338575369532743e-05, "loss": 0.4123, "step": 13292 }, { "epoch": 0.47359139248623905, "grad_norm": 1.1157575648905091, "learning_rate": 1.133743180868273e-05, "loss": 0.4497, "step": 13293 }, { "epoch": 0.4736270196127331, "grad_norm": 1.6753384092238537, "learning_rate": 1.1336288230023875e-05, "loss": 0.5836, "step": 13294 }, { "epoch": 0.47366264673922726, "grad_norm": 2.0511565676868933, "learning_rate": 1.1335144633571412e-05, "loss": 0.5224, "step": 13295 }, { "epoch": 0.4736982738657214, "grad_norm": 1.357207290906466, "learning_rate": 1.1334001019340562e-05, "loss": 0.5597, "step": 13296 }, { "epoch": 0.47373390099221546, "grad_norm": 1.5572354317963557, "learning_rate": 1.1332857387346554e-05, "loss": 0.557, "step": 13297 }, { "epoch": 0.4737695281187096, "grad_norm": 1.1419456673660395, "learning_rate": 1.1331713737604621e-05, "loss": 0.4337, "step": 13298 }, { "epoch": 0.4738051552452037, "grad_norm": 1.0840839312459596, "learning_rate": 1.1330570070129989e-05, "loss": 0.453, "step": 13299 }, { "epoch": 0.4738407823716978, "grad_norm": 1.1059603904684283, "learning_rate": 1.1329426384937887e-05, "loss": 0.4477, "step": 13300 }, { "epoch": 0.47387640949819193, "grad_norm": 1.6022742224755429, "learning_rate": 1.132828268204354e-05, "loss": 0.5768, "step": 13301 }, { "epoch": 0.47391203662468606, "grad_norm": 2.1742282707992318, "learning_rate": 1.1327138961462183e-05, "loss": 0.5886, "step": 13302 }, { "epoch": 0.47394766375118014, "grad_norm": 1.2210974178402496, "learning_rate": 1.1325995223209042e-05, "loss": 0.5201, "step": 13303 }, { "epoch": 0.47398329087767427, "grad_norm": 1.8021556507543417, "learning_rate": 1.132485146729935e-05, "loss": 0.6373, "step": 13304 }, { "epoch": 0.4740189180041684, "grad_norm": 1.4096862851278227, "learning_rate": 1.1323707693748335e-05, "loss": 0.5786, "step": 13305 }, { "epoch": 0.4740545451306625, "grad_norm": 1.3501335252635924, "learning_rate": 1.1322563902571227e-05, "loss": 0.5408, "step": 13306 }, { "epoch": 0.4740901722571566, "grad_norm": 1.7210297447945377, "learning_rate": 1.1321420093783254e-05, "loss": 0.612, "step": 13307 }, { "epoch": 0.47412579938365074, "grad_norm": 1.3721344237702828, "learning_rate": 1.132027626739965e-05, "loss": 0.5502, "step": 13308 }, { "epoch": 0.4741614265101448, "grad_norm": 2.072658424547499, "learning_rate": 1.1319132423435644e-05, "loss": 0.5242, "step": 13309 }, { "epoch": 0.47419705363663894, "grad_norm": 1.24084313178247, "learning_rate": 1.1317988561906468e-05, "loss": 0.4246, "step": 13310 }, { "epoch": 0.4742326807631331, "grad_norm": 2.0148107503946333, "learning_rate": 1.1316844682827355e-05, "loss": 0.5791, "step": 13311 }, { "epoch": 0.47426830788962715, "grad_norm": 1.496295051589409, "learning_rate": 1.1315700786213531e-05, "loss": 0.5071, "step": 13312 }, { "epoch": 0.4743039350161213, "grad_norm": 1.1263437474721811, "learning_rate": 1.1314556872080231e-05, "loss": 0.4888, "step": 13313 }, { "epoch": 0.4743395621426154, "grad_norm": 1.2088075447122075, "learning_rate": 1.1313412940442689e-05, "loss": 0.6129, "step": 13314 }, { "epoch": 0.4743751892691095, "grad_norm": 1.5311442131348407, "learning_rate": 1.1312268991316137e-05, "loss": 0.5665, "step": 13315 }, { "epoch": 0.4744108163956036, "grad_norm": 1.0399663046117824, "learning_rate": 1.1311125024715806e-05, "loss": 0.422, "step": 13316 }, { "epoch": 0.47444644352209775, "grad_norm": 1.110799760215896, "learning_rate": 1.130998104065693e-05, "loss": 0.4512, "step": 13317 }, { "epoch": 0.4744820706485918, "grad_norm": 2.385596732036823, "learning_rate": 1.1308837039154739e-05, "loss": 0.5015, "step": 13318 }, { "epoch": 0.47451769777508596, "grad_norm": 0.9693561973666451, "learning_rate": 1.1307693020224468e-05, "loss": 0.4636, "step": 13319 }, { "epoch": 0.4745533249015801, "grad_norm": 1.4553965799791342, "learning_rate": 1.130654898388135e-05, "loss": 0.6202, "step": 13320 }, { "epoch": 0.47458895202807416, "grad_norm": 1.4306763827047717, "learning_rate": 1.130540493014062e-05, "loss": 0.5748, "step": 13321 }, { "epoch": 0.4746245791545683, "grad_norm": 1.327651516504015, "learning_rate": 1.1304260859017511e-05, "loss": 0.5102, "step": 13322 }, { "epoch": 0.4746602062810624, "grad_norm": 1.0893073637454596, "learning_rate": 1.1303116770527257e-05, "loss": 0.4396, "step": 13323 }, { "epoch": 0.4746958334075565, "grad_norm": 1.5938173451690965, "learning_rate": 1.130197266468509e-05, "loss": 0.575, "step": 13324 }, { "epoch": 0.47473146053405063, "grad_norm": 1.6518973896406377, "learning_rate": 1.1300828541506247e-05, "loss": 0.5594, "step": 13325 }, { "epoch": 0.47476708766054476, "grad_norm": 1.0594511995599631, "learning_rate": 1.1299684401005967e-05, "loss": 0.4397, "step": 13326 }, { "epoch": 0.47480271478703884, "grad_norm": 1.4850282172949463, "learning_rate": 1.1298540243199476e-05, "loss": 0.5167, "step": 13327 }, { "epoch": 0.47483834191353297, "grad_norm": 1.4675173684322076, "learning_rate": 1.1297396068102019e-05, "loss": 0.5783, "step": 13328 }, { "epoch": 0.4748739690400271, "grad_norm": 1.6864153255090526, "learning_rate": 1.1296251875728822e-05, "loss": 0.5699, "step": 13329 }, { "epoch": 0.4749095961665212, "grad_norm": 1.3195158870730899, "learning_rate": 1.1295107666095131e-05, "loss": 0.5675, "step": 13330 }, { "epoch": 0.4749452232930153, "grad_norm": 1.0905590046374334, "learning_rate": 1.129396343921617e-05, "loss": 0.466, "step": 13331 }, { "epoch": 0.47498085041950944, "grad_norm": 1.1039032241400035, "learning_rate": 1.1292819195107185e-05, "loss": 0.4777, "step": 13332 }, { "epoch": 0.4750164775460035, "grad_norm": 1.0955090395958198, "learning_rate": 1.1291674933783408e-05, "loss": 0.4433, "step": 13333 }, { "epoch": 0.47505210467249764, "grad_norm": 1.0874105420118225, "learning_rate": 1.1290530655260075e-05, "loss": 0.4786, "step": 13334 }, { "epoch": 0.4750877317989918, "grad_norm": 1.0901250574067982, "learning_rate": 1.1289386359552426e-05, "loss": 0.4788, "step": 13335 }, { "epoch": 0.47512335892548585, "grad_norm": 1.3664273451773854, "learning_rate": 1.1288242046675698e-05, "loss": 0.4721, "step": 13336 }, { "epoch": 0.47515898605198, "grad_norm": 0.9535349332668834, "learning_rate": 1.1287097716645121e-05, "loss": 0.4652, "step": 13337 }, { "epoch": 0.4751946131784741, "grad_norm": 1.478711968630321, "learning_rate": 1.128595336947594e-05, "loss": 0.5366, "step": 13338 }, { "epoch": 0.4752302403049682, "grad_norm": 1.0935421512470174, "learning_rate": 1.1284809005183395e-05, "loss": 0.4532, "step": 13339 }, { "epoch": 0.4752658674314623, "grad_norm": 0.9057612843454296, "learning_rate": 1.1283664623782718e-05, "loss": 0.3897, "step": 13340 }, { "epoch": 0.47530149455795645, "grad_norm": 1.6918847555691456, "learning_rate": 1.1282520225289152e-05, "loss": 0.5569, "step": 13341 }, { "epoch": 0.4753371216844505, "grad_norm": 1.154098058306073, "learning_rate": 1.1281375809717927e-05, "loss": 0.4909, "step": 13342 }, { "epoch": 0.47537274881094466, "grad_norm": 1.2809659836752216, "learning_rate": 1.128023137708429e-05, "loss": 0.5118, "step": 13343 }, { "epoch": 0.4754083759374388, "grad_norm": 1.62883746393896, "learning_rate": 1.127908692740348e-05, "loss": 0.5556, "step": 13344 }, { "epoch": 0.47544400306393286, "grad_norm": 1.0090563197581837, "learning_rate": 1.1277942460690731e-05, "loss": 0.4622, "step": 13345 }, { "epoch": 0.475479630190427, "grad_norm": 1.4858138105389165, "learning_rate": 1.1276797976961285e-05, "loss": 0.5289, "step": 13346 }, { "epoch": 0.4755152573169211, "grad_norm": 1.0252738508266532, "learning_rate": 1.127565347623038e-05, "loss": 0.4675, "step": 13347 }, { "epoch": 0.4755508844434152, "grad_norm": 1.0243986443156632, "learning_rate": 1.127450895851326e-05, "loss": 0.4348, "step": 13348 }, { "epoch": 0.47558651156990933, "grad_norm": 1.4728809617896885, "learning_rate": 1.1273364423825162e-05, "loss": 0.6146, "step": 13349 }, { "epoch": 0.47562213869640346, "grad_norm": 1.075376916670747, "learning_rate": 1.1272219872181322e-05, "loss": 0.4751, "step": 13350 }, { "epoch": 0.47565776582289754, "grad_norm": 1.521014754364793, "learning_rate": 1.1271075303596992e-05, "loss": 0.6202, "step": 13351 }, { "epoch": 0.47569339294939167, "grad_norm": 1.0208558121026647, "learning_rate": 1.1269930718087403e-05, "loss": 0.4651, "step": 13352 }, { "epoch": 0.4757290200758858, "grad_norm": 1.9914717483285214, "learning_rate": 1.1268786115667798e-05, "loss": 0.6243, "step": 13353 }, { "epoch": 0.4757646472023799, "grad_norm": 1.1306518674240056, "learning_rate": 1.126764149635342e-05, "loss": 0.4751, "step": 13354 }, { "epoch": 0.475800274328874, "grad_norm": 1.735020110746497, "learning_rate": 1.1266496860159512e-05, "loss": 0.512, "step": 13355 }, { "epoch": 0.47583590145536814, "grad_norm": 1.492082006116538, "learning_rate": 1.1265352207101312e-05, "loss": 0.5726, "step": 13356 }, { "epoch": 0.4758715285818622, "grad_norm": 1.4433879606734532, "learning_rate": 1.1264207537194059e-05, "loss": 0.5407, "step": 13357 }, { "epoch": 0.47590715570835634, "grad_norm": 1.4611264583066792, "learning_rate": 1.1263062850453002e-05, "loss": 0.5713, "step": 13358 }, { "epoch": 0.4759427828348505, "grad_norm": 1.5248515818897013, "learning_rate": 1.1261918146893381e-05, "loss": 0.6009, "step": 13359 }, { "epoch": 0.47597840996134455, "grad_norm": 2.446627384398892, "learning_rate": 1.1260773426530436e-05, "loss": 0.5394, "step": 13360 }, { "epoch": 0.4760140370878387, "grad_norm": 1.677705218224384, "learning_rate": 1.1259628689379412e-05, "loss": 0.5495, "step": 13361 }, { "epoch": 0.4760496642143328, "grad_norm": 1.851930397253873, "learning_rate": 1.1258483935455551e-05, "loss": 0.548, "step": 13362 }, { "epoch": 0.4760852913408269, "grad_norm": 1.0290279615259745, "learning_rate": 1.12573391647741e-05, "loss": 0.4215, "step": 13363 }, { "epoch": 0.476120918467321, "grad_norm": 1.3805473723409871, "learning_rate": 1.1256194377350296e-05, "loss": 0.5746, "step": 13364 }, { "epoch": 0.47615654559381515, "grad_norm": 1.7708720127445934, "learning_rate": 1.1255049573199386e-05, "loss": 0.5827, "step": 13365 }, { "epoch": 0.4761921727203092, "grad_norm": 1.188413989952923, "learning_rate": 1.1253904752336615e-05, "loss": 0.4316, "step": 13366 }, { "epoch": 0.47622779984680336, "grad_norm": 1.6648414853440552, "learning_rate": 1.1252759914777224e-05, "loss": 0.6031, "step": 13367 }, { "epoch": 0.4762634269732975, "grad_norm": 1.0990357448247041, "learning_rate": 1.125161506053646e-05, "loss": 0.437, "step": 13368 }, { "epoch": 0.47629905409979156, "grad_norm": 1.0552892723190428, "learning_rate": 1.1250470189629567e-05, "loss": 0.4989, "step": 13369 }, { "epoch": 0.4763346812262857, "grad_norm": 1.0312353104996888, "learning_rate": 1.1249325302071793e-05, "loss": 0.439, "step": 13370 }, { "epoch": 0.4763703083527798, "grad_norm": 1.1142673621723664, "learning_rate": 1.1248180397878372e-05, "loss": 0.477, "step": 13371 }, { "epoch": 0.4764059354792739, "grad_norm": 1.5059095643912295, "learning_rate": 1.124703547706456e-05, "loss": 0.6026, "step": 13372 }, { "epoch": 0.47644156260576803, "grad_norm": 1.232362918444876, "learning_rate": 1.1245890539645598e-05, "loss": 0.5252, "step": 13373 }, { "epoch": 0.47647718973226216, "grad_norm": 1.055926594100654, "learning_rate": 1.1244745585636736e-05, "loss": 0.4198, "step": 13374 }, { "epoch": 0.47651281685875624, "grad_norm": 1.4657885093979153, "learning_rate": 1.1243600615053217e-05, "loss": 0.5211, "step": 13375 }, { "epoch": 0.47654844398525037, "grad_norm": 1.1499620722802952, "learning_rate": 1.124245562791028e-05, "loss": 0.486, "step": 13376 }, { "epoch": 0.4765840711117445, "grad_norm": 1.1017752994320498, "learning_rate": 1.1241310624223183e-05, "loss": 0.4851, "step": 13377 }, { "epoch": 0.4766196982382386, "grad_norm": 1.0768221201707198, "learning_rate": 1.1240165604007167e-05, "loss": 0.5235, "step": 13378 }, { "epoch": 0.4766553253647327, "grad_norm": 1.371063138014959, "learning_rate": 1.1239020567277479e-05, "loss": 0.6328, "step": 13379 }, { "epoch": 0.47669095249122684, "grad_norm": 1.0048026952465654, "learning_rate": 1.1237875514049365e-05, "loss": 0.4731, "step": 13380 }, { "epoch": 0.4767265796177209, "grad_norm": 1.8479262572901056, "learning_rate": 1.1236730444338076e-05, "loss": 0.6359, "step": 13381 }, { "epoch": 0.47676220674421504, "grad_norm": 1.1122052417583574, "learning_rate": 1.1235585358158854e-05, "loss": 0.459, "step": 13382 }, { "epoch": 0.4767978338707092, "grad_norm": 1.055627737659598, "learning_rate": 1.1234440255526948e-05, "loss": 0.4787, "step": 13383 }, { "epoch": 0.47683346099720325, "grad_norm": 1.9153239011322787, "learning_rate": 1.1233295136457609e-05, "loss": 0.5621, "step": 13384 }, { "epoch": 0.4768690881236974, "grad_norm": 1.712437252476453, "learning_rate": 1.1232150000966084e-05, "loss": 0.5066, "step": 13385 }, { "epoch": 0.4769047152501915, "grad_norm": 1.6713697552002669, "learning_rate": 1.1231004849067621e-05, "loss": 0.5399, "step": 13386 }, { "epoch": 0.4769403423766856, "grad_norm": 1.3033761757002895, "learning_rate": 1.1229859680777465e-05, "loss": 0.5176, "step": 13387 }, { "epoch": 0.4769759695031797, "grad_norm": 1.6758075929845284, "learning_rate": 1.1228714496110871e-05, "loss": 0.5408, "step": 13388 }, { "epoch": 0.47701159662967385, "grad_norm": 1.8804148118408441, "learning_rate": 1.1227569295083083e-05, "loss": 0.544, "step": 13389 }, { "epoch": 0.4770472237561679, "grad_norm": 1.4579438913396845, "learning_rate": 1.1226424077709353e-05, "loss": 0.5216, "step": 13390 }, { "epoch": 0.47708285088266206, "grad_norm": 1.5247109093271438, "learning_rate": 1.1225278844004927e-05, "loss": 0.6175, "step": 13391 }, { "epoch": 0.4771184780091562, "grad_norm": 1.1779250747732093, "learning_rate": 1.1224133593985061e-05, "loss": 0.4494, "step": 13392 }, { "epoch": 0.47715410513565026, "grad_norm": 1.1698668915169543, "learning_rate": 1.1222988327664997e-05, "loss": 0.4237, "step": 13393 }, { "epoch": 0.4771897322621444, "grad_norm": 1.1169232625435683, "learning_rate": 1.1221843045059989e-05, "loss": 0.4326, "step": 13394 }, { "epoch": 0.4772253593886385, "grad_norm": 1.128105164718478, "learning_rate": 1.1220697746185285e-05, "loss": 0.4504, "step": 13395 }, { "epoch": 0.4772609865151326, "grad_norm": 1.6896651001215635, "learning_rate": 1.121955243105614e-05, "loss": 0.63, "step": 13396 }, { "epoch": 0.47729661364162673, "grad_norm": 1.2679948488341075, "learning_rate": 1.1218407099687802e-05, "loss": 0.5619, "step": 13397 }, { "epoch": 0.47733224076812086, "grad_norm": 1.13572529963176, "learning_rate": 1.1217261752095518e-05, "loss": 0.4689, "step": 13398 }, { "epoch": 0.47736786789461494, "grad_norm": 1.1068229833848917, "learning_rate": 1.121611638829455e-05, "loss": 0.4841, "step": 13399 }, { "epoch": 0.47740349502110907, "grad_norm": 1.1327978232116087, "learning_rate": 1.1214971008300138e-05, "loss": 0.4833, "step": 13400 }, { "epoch": 0.4774391221476032, "grad_norm": 1.133053582548719, "learning_rate": 1.1213825612127537e-05, "loss": 0.4688, "step": 13401 }, { "epoch": 0.4774747492740973, "grad_norm": 1.1999289565573252, "learning_rate": 1.1212680199792001e-05, "loss": 0.4866, "step": 13402 }, { "epoch": 0.4775103764005914, "grad_norm": 1.7845942574267446, "learning_rate": 1.121153477130878e-05, "loss": 0.6293, "step": 13403 }, { "epoch": 0.47754600352708554, "grad_norm": 1.1299264957657904, "learning_rate": 1.121038932669313e-05, "loss": 0.4461, "step": 13404 }, { "epoch": 0.4775816306535796, "grad_norm": 1.6500210309906453, "learning_rate": 1.1209243865960295e-05, "loss": 0.4996, "step": 13405 }, { "epoch": 0.47761725778007375, "grad_norm": 1.0661397571982867, "learning_rate": 1.1208098389125533e-05, "loss": 0.4931, "step": 13406 }, { "epoch": 0.4776528849065679, "grad_norm": 1.3850371413476148, "learning_rate": 1.12069528962041e-05, "loss": 0.5805, "step": 13407 }, { "epoch": 0.47768851203306195, "grad_norm": 1.2177231946723803, "learning_rate": 1.1205807387211245e-05, "loss": 0.4422, "step": 13408 }, { "epoch": 0.4777241391595561, "grad_norm": 1.1449287261424959, "learning_rate": 1.1204661862162219e-05, "loss": 0.4124, "step": 13409 }, { "epoch": 0.4777597662860502, "grad_norm": 1.6726677283586615, "learning_rate": 1.1203516321072281e-05, "loss": 0.5411, "step": 13410 }, { "epoch": 0.4777953934125443, "grad_norm": 1.2503885658601512, "learning_rate": 1.120237076395668e-05, "loss": 0.4452, "step": 13411 }, { "epoch": 0.4778310205390384, "grad_norm": 1.6812407599957169, "learning_rate": 1.1201225190830676e-05, "loss": 0.5397, "step": 13412 }, { "epoch": 0.47786664766553255, "grad_norm": 1.1411103644510583, "learning_rate": 1.1200079601709513e-05, "loss": 0.4716, "step": 13413 }, { "epoch": 0.4779022747920266, "grad_norm": 1.5951582624286011, "learning_rate": 1.1198933996608456e-05, "loss": 0.5534, "step": 13414 }, { "epoch": 0.47793790191852076, "grad_norm": 1.1086892771971946, "learning_rate": 1.1197788375542754e-05, "loss": 0.4732, "step": 13415 }, { "epoch": 0.4779735290450149, "grad_norm": 1.6445967282304799, "learning_rate": 1.119664273852766e-05, "loss": 0.5212, "step": 13416 }, { "epoch": 0.47800915617150896, "grad_norm": 3.7831639930548873, "learning_rate": 1.119549708557843e-05, "loss": 0.4041, "step": 13417 }, { "epoch": 0.4780447832980031, "grad_norm": 2.426544889135363, "learning_rate": 1.1194351416710325e-05, "loss": 0.4919, "step": 13418 }, { "epoch": 0.4780804104244972, "grad_norm": 1.1021308276145334, "learning_rate": 1.1193205731938597e-05, "loss": 0.4303, "step": 13419 }, { "epoch": 0.4781160375509913, "grad_norm": 0.9854165806141338, "learning_rate": 1.1192060031278497e-05, "loss": 0.4271, "step": 13420 }, { "epoch": 0.47815166467748543, "grad_norm": 1.6851221785580184, "learning_rate": 1.1190914314745286e-05, "loss": 0.4842, "step": 13421 }, { "epoch": 0.47818729180397956, "grad_norm": 1.7475025044208985, "learning_rate": 1.118976858235422e-05, "loss": 0.5776, "step": 13422 }, { "epoch": 0.47822291893047364, "grad_norm": 1.5553107564321778, "learning_rate": 1.1188622834120553e-05, "loss": 0.4946, "step": 13423 }, { "epoch": 0.47825854605696777, "grad_norm": 1.1147205080523859, "learning_rate": 1.118747707005954e-05, "loss": 0.3872, "step": 13424 }, { "epoch": 0.4782941731834619, "grad_norm": 1.2679409926816276, "learning_rate": 1.1186331290186444e-05, "loss": 0.5074, "step": 13425 }, { "epoch": 0.478329800309956, "grad_norm": 1.3751493129814536, "learning_rate": 1.1185185494516515e-05, "loss": 0.633, "step": 13426 }, { "epoch": 0.4783654274364501, "grad_norm": 1.53677446267802, "learning_rate": 1.1184039683065014e-05, "loss": 0.5548, "step": 13427 }, { "epoch": 0.47840105456294424, "grad_norm": 1.182853318284824, "learning_rate": 1.1182893855847193e-05, "loss": 0.4305, "step": 13428 }, { "epoch": 0.4784366816894383, "grad_norm": 1.2745191536143285, "learning_rate": 1.1181748012878316e-05, "loss": 0.4128, "step": 13429 }, { "epoch": 0.47847230881593245, "grad_norm": 1.63589230935961, "learning_rate": 1.1180602154173638e-05, "loss": 0.5178, "step": 13430 }, { "epoch": 0.4785079359424266, "grad_norm": 1.68277910421375, "learning_rate": 1.1179456279748418e-05, "loss": 0.5927, "step": 13431 }, { "epoch": 0.47854356306892065, "grad_norm": 1.132291129763576, "learning_rate": 1.1178310389617912e-05, "loss": 0.4212, "step": 13432 }, { "epoch": 0.4785791901954148, "grad_norm": 1.242591023274898, "learning_rate": 1.1177164483797381e-05, "loss": 0.4699, "step": 13433 }, { "epoch": 0.4786148173219089, "grad_norm": 1.796611804383555, "learning_rate": 1.1176018562302082e-05, "loss": 0.5123, "step": 13434 }, { "epoch": 0.478650444448403, "grad_norm": 1.2191830591417188, "learning_rate": 1.1174872625147273e-05, "loss": 0.5096, "step": 13435 }, { "epoch": 0.4786860715748971, "grad_norm": 1.5499157893361968, "learning_rate": 1.1173726672348215e-05, "loss": 0.6218, "step": 13436 }, { "epoch": 0.47872169870139125, "grad_norm": 1.1276307716372478, "learning_rate": 1.1172580703920165e-05, "loss": 0.4925, "step": 13437 }, { "epoch": 0.4787573258278853, "grad_norm": 1.76741581032598, "learning_rate": 1.1171434719878385e-05, "loss": 0.5625, "step": 13438 }, { "epoch": 0.47879295295437946, "grad_norm": 1.6657422978313263, "learning_rate": 1.117028872023813e-05, "loss": 0.5743, "step": 13439 }, { "epoch": 0.4788285800808736, "grad_norm": 1.2335884767386025, "learning_rate": 1.1169142705014661e-05, "loss": 0.4524, "step": 13440 }, { "epoch": 0.47886420720736766, "grad_norm": 1.319046106997052, "learning_rate": 1.1167996674223245e-05, "loss": 0.525, "step": 13441 }, { "epoch": 0.4788998343338618, "grad_norm": 2.448796829834894, "learning_rate": 1.1166850627879135e-05, "loss": 0.5606, "step": 13442 }, { "epoch": 0.4789354614603559, "grad_norm": 1.6823524676116823, "learning_rate": 1.1165704565997593e-05, "loss": 0.5932, "step": 13443 }, { "epoch": 0.47897108858685, "grad_norm": 1.0902016270860828, "learning_rate": 1.116455848859388e-05, "loss": 0.4653, "step": 13444 }, { "epoch": 0.47900671571334413, "grad_norm": 1.0082491723972997, "learning_rate": 1.1163412395683257e-05, "loss": 0.4552, "step": 13445 }, { "epoch": 0.47904234283983826, "grad_norm": 1.6778248564852754, "learning_rate": 1.1162266287280985e-05, "loss": 0.5578, "step": 13446 }, { "epoch": 0.47907796996633234, "grad_norm": 1.4835172369082559, "learning_rate": 1.1161120163402324e-05, "loss": 0.4391, "step": 13447 }, { "epoch": 0.47911359709282647, "grad_norm": 1.394835071336067, "learning_rate": 1.115997402406254e-05, "loss": 0.6189, "step": 13448 }, { "epoch": 0.4791492242193206, "grad_norm": 1.120345637417738, "learning_rate": 1.1158827869276887e-05, "loss": 0.4111, "step": 13449 }, { "epoch": 0.4791848513458147, "grad_norm": 1.4831026476107696, "learning_rate": 1.115768169906063e-05, "loss": 0.5712, "step": 13450 }, { "epoch": 0.4792204784723088, "grad_norm": 1.1875798391143575, "learning_rate": 1.1156535513429034e-05, "loss": 0.5065, "step": 13451 }, { "epoch": 0.47925610559880294, "grad_norm": 1.4373724754463553, "learning_rate": 1.115538931239736e-05, "loss": 0.5741, "step": 13452 }, { "epoch": 0.479291732725297, "grad_norm": 1.5034724842233762, "learning_rate": 1.1154243095980868e-05, "loss": 0.5136, "step": 13453 }, { "epoch": 0.47932735985179115, "grad_norm": 1.0868871643674987, "learning_rate": 1.1153096864194823e-05, "loss": 0.4324, "step": 13454 }, { "epoch": 0.4793629869782853, "grad_norm": 1.2652695112111478, "learning_rate": 1.1151950617054487e-05, "loss": 0.4638, "step": 13455 }, { "epoch": 0.47939861410477935, "grad_norm": 1.149655535850997, "learning_rate": 1.1150804354575126e-05, "loss": 0.4528, "step": 13456 }, { "epoch": 0.4794342412312735, "grad_norm": 1.4983893922030025, "learning_rate": 1.1149658076772001e-05, "loss": 0.5448, "step": 13457 }, { "epoch": 0.4794698683577676, "grad_norm": 1.5665527501090344, "learning_rate": 1.114851178366037e-05, "loss": 0.5155, "step": 13458 }, { "epoch": 0.4795054954842617, "grad_norm": 1.050855203078194, "learning_rate": 1.1147365475255505e-05, "loss": 0.377, "step": 13459 }, { "epoch": 0.4795411226107558, "grad_norm": 2.754183611920197, "learning_rate": 1.1146219151572669e-05, "loss": 0.62, "step": 13460 }, { "epoch": 0.47957674973724995, "grad_norm": 2.0191327899570637, "learning_rate": 1.1145072812627117e-05, "loss": 0.4944, "step": 13461 }, { "epoch": 0.479612376863744, "grad_norm": 1.654321692581118, "learning_rate": 1.1143926458434127e-05, "loss": 0.5682, "step": 13462 }, { "epoch": 0.47964800399023816, "grad_norm": 1.3386823697155543, "learning_rate": 1.1142780089008953e-05, "loss": 0.5087, "step": 13463 }, { "epoch": 0.4796836311167323, "grad_norm": 1.5396587169253562, "learning_rate": 1.1141633704366865e-05, "loss": 0.5506, "step": 13464 }, { "epoch": 0.47971925824322637, "grad_norm": 3.3346433012364027, "learning_rate": 1.1140487304523126e-05, "loss": 0.681, "step": 13465 }, { "epoch": 0.4797548853697205, "grad_norm": 1.594485817813225, "learning_rate": 1.1139340889493e-05, "loss": 0.5513, "step": 13466 }, { "epoch": 0.4797905124962146, "grad_norm": 1.0410845826010795, "learning_rate": 1.1138194459291755e-05, "loss": 0.457, "step": 13467 }, { "epoch": 0.4798261396227087, "grad_norm": 1.2653986326642026, "learning_rate": 1.1137048013934657e-05, "loss": 0.4571, "step": 13468 }, { "epoch": 0.47986176674920283, "grad_norm": 1.3817281290456032, "learning_rate": 1.1135901553436964e-05, "loss": 0.6341, "step": 13469 }, { "epoch": 0.47989739387569696, "grad_norm": 1.0287091210592498, "learning_rate": 1.1134755077813955e-05, "loss": 0.4186, "step": 13470 }, { "epoch": 0.4799330210021911, "grad_norm": 1.1872704270697128, "learning_rate": 1.1133608587080887e-05, "loss": 0.4545, "step": 13471 }, { "epoch": 0.47996864812868517, "grad_norm": 1.4395443476236134, "learning_rate": 1.1132462081253028e-05, "loss": 0.585, "step": 13472 }, { "epoch": 0.4800042752551793, "grad_norm": 1.2813026921924626, "learning_rate": 1.1131315560345641e-05, "loss": 0.5134, "step": 13473 }, { "epoch": 0.48003990238167343, "grad_norm": 1.3155131813847392, "learning_rate": 1.1130169024374002e-05, "loss": 0.4758, "step": 13474 }, { "epoch": 0.4800755295081675, "grad_norm": 1.4467047148689114, "learning_rate": 1.1129022473353368e-05, "loss": 0.4505, "step": 13475 }, { "epoch": 0.48011115663466164, "grad_norm": 1.849284910143948, "learning_rate": 1.1127875907299013e-05, "loss": 0.5502, "step": 13476 }, { "epoch": 0.48014678376115577, "grad_norm": 1.3442215626160345, "learning_rate": 1.11267293262262e-05, "loss": 0.4056, "step": 13477 }, { "epoch": 0.48018241088764985, "grad_norm": 1.4196565943511137, "learning_rate": 1.1125582730150201e-05, "loss": 0.5926, "step": 13478 }, { "epoch": 0.480218038014144, "grad_norm": 0.9986103008622721, "learning_rate": 1.1124436119086281e-05, "loss": 0.4254, "step": 13479 }, { "epoch": 0.4802536651406381, "grad_norm": 1.1255136212913326, "learning_rate": 1.1123289493049707e-05, "loss": 0.5062, "step": 13480 }, { "epoch": 0.4802892922671322, "grad_norm": 1.0390289528007708, "learning_rate": 1.112214285205575e-05, "loss": 0.5036, "step": 13481 }, { "epoch": 0.4803249193936263, "grad_norm": 1.5057652137304864, "learning_rate": 1.1120996196119676e-05, "loss": 0.4576, "step": 13482 }, { "epoch": 0.48036054652012045, "grad_norm": 1.380596757808118, "learning_rate": 1.1119849525256754e-05, "loss": 0.4617, "step": 13483 }, { "epoch": 0.4803961736466145, "grad_norm": 1.1010309910533782, "learning_rate": 1.1118702839482251e-05, "loss": 0.4487, "step": 13484 }, { "epoch": 0.48043180077310865, "grad_norm": 1.6075386845693067, "learning_rate": 1.111755613881144e-05, "loss": 0.5409, "step": 13485 }, { "epoch": 0.4804674278996028, "grad_norm": 1.6835351991208862, "learning_rate": 1.1116409423259586e-05, "loss": 0.6152, "step": 13486 }, { "epoch": 0.48050305502609686, "grad_norm": 1.0639642564549465, "learning_rate": 1.111526269284196e-05, "loss": 0.4512, "step": 13487 }, { "epoch": 0.480538682152591, "grad_norm": 1.6167791335352213, "learning_rate": 1.1114115947573834e-05, "loss": 0.5361, "step": 13488 }, { "epoch": 0.4805743092790851, "grad_norm": 1.0978489387362613, "learning_rate": 1.1112969187470475e-05, "loss": 0.4671, "step": 13489 }, { "epoch": 0.4806099364055792, "grad_norm": 1.191396060469837, "learning_rate": 1.1111822412547152e-05, "loss": 0.4626, "step": 13490 }, { "epoch": 0.4806455635320733, "grad_norm": 1.5839994406941573, "learning_rate": 1.1110675622819135e-05, "loss": 0.5694, "step": 13491 }, { "epoch": 0.48068119065856746, "grad_norm": 1.1047557167911923, "learning_rate": 1.11095288183017e-05, "loss": 0.4929, "step": 13492 }, { "epoch": 0.48071681778506153, "grad_norm": 1.404547009282796, "learning_rate": 1.1108381999010111e-05, "loss": 0.5464, "step": 13493 }, { "epoch": 0.48075244491155567, "grad_norm": 1.43118370021058, "learning_rate": 1.1107235164959643e-05, "loss": 0.6286, "step": 13494 }, { "epoch": 0.4807880720380498, "grad_norm": 1.2603908022118333, "learning_rate": 1.1106088316165562e-05, "loss": 0.4332, "step": 13495 }, { "epoch": 0.48082369916454387, "grad_norm": 1.813043465615765, "learning_rate": 1.1104941452643143e-05, "loss": 0.5708, "step": 13496 }, { "epoch": 0.480859326291038, "grad_norm": 1.0853120459880203, "learning_rate": 1.1103794574407659e-05, "loss": 0.4198, "step": 13497 }, { "epoch": 0.48089495341753213, "grad_norm": 1.050543344575744, "learning_rate": 1.1102647681474372e-05, "loss": 0.4622, "step": 13498 }, { "epoch": 0.4809305805440262, "grad_norm": 1.7249214802452444, "learning_rate": 1.1101500773858564e-05, "loss": 0.503, "step": 13499 }, { "epoch": 0.48096620767052034, "grad_norm": 1.0331006027441336, "learning_rate": 1.1100353851575505e-05, "loss": 0.4358, "step": 13500 }, { "epoch": 0.48100183479701447, "grad_norm": 1.0874241310880486, "learning_rate": 1.1099206914640464e-05, "loss": 0.4853, "step": 13501 }, { "epoch": 0.48103746192350855, "grad_norm": 1.1951414318644442, "learning_rate": 1.1098059963068715e-05, "loss": 0.452, "step": 13502 }, { "epoch": 0.4810730890500027, "grad_norm": 1.0228575140115503, "learning_rate": 1.1096912996875529e-05, "loss": 0.4641, "step": 13503 }, { "epoch": 0.4811087161764968, "grad_norm": 1.6495461630617163, "learning_rate": 1.1095766016076182e-05, "loss": 0.593, "step": 13504 }, { "epoch": 0.4811443433029909, "grad_norm": 1.5013829122106115, "learning_rate": 1.1094619020685945e-05, "loss": 0.5604, "step": 13505 }, { "epoch": 0.481179970429485, "grad_norm": 1.067187220823925, "learning_rate": 1.1093472010720086e-05, "loss": 0.4453, "step": 13506 }, { "epoch": 0.48121559755597915, "grad_norm": 1.7316236388798072, "learning_rate": 1.1092324986193885e-05, "loss": 0.4805, "step": 13507 }, { "epoch": 0.4812512246824732, "grad_norm": 1.1817501012008917, "learning_rate": 1.1091177947122616e-05, "loss": 0.4793, "step": 13508 }, { "epoch": 0.48128685180896735, "grad_norm": 1.449160830445925, "learning_rate": 1.1090030893521549e-05, "loss": 0.4979, "step": 13509 }, { "epoch": 0.4813224789354615, "grad_norm": 1.2531851325407906, "learning_rate": 1.1088883825405953e-05, "loss": 0.4732, "step": 13510 }, { "epoch": 0.48135810606195556, "grad_norm": 1.121455903009463, "learning_rate": 1.1087736742791114e-05, "loss": 0.4506, "step": 13511 }, { "epoch": 0.4813937331884497, "grad_norm": 1.3027805219688442, "learning_rate": 1.1086589645692299e-05, "loss": 0.6166, "step": 13512 }, { "epoch": 0.4814293603149438, "grad_norm": 1.408187630639915, "learning_rate": 1.1085442534124784e-05, "loss": 0.5056, "step": 13513 }, { "epoch": 0.4814649874414379, "grad_norm": 1.1136317396607367, "learning_rate": 1.108429540810384e-05, "loss": 0.5067, "step": 13514 }, { "epoch": 0.48150061456793203, "grad_norm": 1.1910276032603806, "learning_rate": 1.1083148267644748e-05, "loss": 0.4291, "step": 13515 }, { "epoch": 0.48153624169442616, "grad_norm": 1.7592786178938375, "learning_rate": 1.1082001112762778e-05, "loss": 0.6149, "step": 13516 }, { "epoch": 0.48157186882092023, "grad_norm": 1.5099676756562783, "learning_rate": 1.1080853943473207e-05, "loss": 0.564, "step": 13517 }, { "epoch": 0.48160749594741437, "grad_norm": 1.013796880264817, "learning_rate": 1.1079706759791311e-05, "loss": 0.4759, "step": 13518 }, { "epoch": 0.4816431230739085, "grad_norm": 1.0841297442886133, "learning_rate": 1.1078559561732364e-05, "loss": 0.4381, "step": 13519 }, { "epoch": 0.4816787502004026, "grad_norm": 1.1913122336507365, "learning_rate": 1.1077412349311645e-05, "loss": 0.4553, "step": 13520 }, { "epoch": 0.4817143773268967, "grad_norm": 1.7466930028025107, "learning_rate": 1.1076265122544422e-05, "loss": 0.5882, "step": 13521 }, { "epoch": 0.48175000445339083, "grad_norm": 1.671476947906406, "learning_rate": 1.107511788144598e-05, "loss": 0.5556, "step": 13522 }, { "epoch": 0.4817856315798849, "grad_norm": 1.2995242146613146, "learning_rate": 1.1073970626031594e-05, "loss": 0.5993, "step": 13523 }, { "epoch": 0.48182125870637904, "grad_norm": 1.157426383931428, "learning_rate": 1.1072823356316535e-05, "loss": 0.4667, "step": 13524 }, { "epoch": 0.48185688583287317, "grad_norm": 1.1716141950937613, "learning_rate": 1.1071676072316084e-05, "loss": 0.4685, "step": 13525 }, { "epoch": 0.48189251295936725, "grad_norm": 1.5436910736477523, "learning_rate": 1.107052877404552e-05, "loss": 0.5544, "step": 13526 }, { "epoch": 0.4819281400858614, "grad_norm": 1.1784791331916211, "learning_rate": 1.1069381461520115e-05, "loss": 0.4852, "step": 13527 }, { "epoch": 0.4819637672123555, "grad_norm": 1.5337248719721164, "learning_rate": 1.1068234134755149e-05, "loss": 0.5132, "step": 13528 }, { "epoch": 0.4819993943388496, "grad_norm": 1.6005876559285674, "learning_rate": 1.1067086793765896e-05, "loss": 0.5741, "step": 13529 }, { "epoch": 0.4820350214653437, "grad_norm": 1.1861299484003387, "learning_rate": 1.106593943856764e-05, "loss": 0.5053, "step": 13530 }, { "epoch": 0.48207064859183785, "grad_norm": 1.5280178977520458, "learning_rate": 1.1064792069175654e-05, "loss": 0.5517, "step": 13531 }, { "epoch": 0.4821062757183319, "grad_norm": 1.559186430409541, "learning_rate": 1.1063644685605216e-05, "loss": 0.5141, "step": 13532 }, { "epoch": 0.48214190284482605, "grad_norm": 1.0295170081444154, "learning_rate": 1.1062497287871606e-05, "loss": 0.3961, "step": 13533 }, { "epoch": 0.4821775299713202, "grad_norm": 1.2784848611753663, "learning_rate": 1.1061349875990101e-05, "loss": 0.4684, "step": 13534 }, { "epoch": 0.48221315709781426, "grad_norm": 3.14257556821547, "learning_rate": 1.1060202449975984e-05, "loss": 0.5576, "step": 13535 }, { "epoch": 0.4822487842243084, "grad_norm": 1.5640426326436965, "learning_rate": 1.1059055009844528e-05, "loss": 0.6089, "step": 13536 }, { "epoch": 0.4822844113508025, "grad_norm": 1.7612623148121092, "learning_rate": 1.1057907555611017e-05, "loss": 0.504, "step": 13537 }, { "epoch": 0.4823200384772966, "grad_norm": 2.111968895873947, "learning_rate": 1.1056760087290724e-05, "loss": 0.5247, "step": 13538 }, { "epoch": 0.48235566560379073, "grad_norm": 1.1088061183239044, "learning_rate": 1.1055612604898932e-05, "loss": 0.4174, "step": 13539 }, { "epoch": 0.48239129273028486, "grad_norm": 1.5240659682867417, "learning_rate": 1.1054465108450922e-05, "loss": 0.5802, "step": 13540 }, { "epoch": 0.48242691985677894, "grad_norm": 1.1451561136230053, "learning_rate": 1.1053317597961972e-05, "loss": 0.4399, "step": 13541 }, { "epoch": 0.48246254698327307, "grad_norm": 1.7471891554939685, "learning_rate": 1.1052170073447364e-05, "loss": 0.5207, "step": 13542 }, { "epoch": 0.4824981741097672, "grad_norm": 1.5973410458567192, "learning_rate": 1.1051022534922372e-05, "loss": 0.6709, "step": 13543 }, { "epoch": 0.4825338012362613, "grad_norm": 1.3531794364955907, "learning_rate": 1.104987498240228e-05, "loss": 0.6333, "step": 13544 }, { "epoch": 0.4825694283627554, "grad_norm": 1.0848568086694903, "learning_rate": 1.1048727415902371e-05, "loss": 0.4809, "step": 13545 }, { "epoch": 0.48260505548924953, "grad_norm": 1.2167270791937104, "learning_rate": 1.1047579835437925e-05, "loss": 0.4406, "step": 13546 }, { "epoch": 0.4826406826157436, "grad_norm": 1.1519022341158514, "learning_rate": 1.1046432241024219e-05, "loss": 0.4481, "step": 13547 }, { "epoch": 0.48267630974223774, "grad_norm": 1.5037433786424597, "learning_rate": 1.1045284632676535e-05, "loss": 0.5663, "step": 13548 }, { "epoch": 0.4827119368687319, "grad_norm": 1.27385520411484, "learning_rate": 1.1044137010410158e-05, "loss": 0.5368, "step": 13549 }, { "epoch": 0.48274756399522595, "grad_norm": 1.349537029486892, "learning_rate": 1.1042989374240367e-05, "loss": 0.4897, "step": 13550 }, { "epoch": 0.4827831911217201, "grad_norm": 2.2127934671607603, "learning_rate": 1.104184172418244e-05, "loss": 0.5728, "step": 13551 }, { "epoch": 0.4828188182482142, "grad_norm": 1.1009664857906125, "learning_rate": 1.1040694060251667e-05, "loss": 0.4433, "step": 13552 }, { "epoch": 0.4828544453747083, "grad_norm": 1.397835661173231, "learning_rate": 1.1039546382463323e-05, "loss": 0.4661, "step": 13553 }, { "epoch": 0.4828900725012024, "grad_norm": 1.2039373202371784, "learning_rate": 1.1038398690832695e-05, "loss": 0.4408, "step": 13554 }, { "epoch": 0.48292569962769655, "grad_norm": 1.4356924872646595, "learning_rate": 1.1037250985375056e-05, "loss": 0.5478, "step": 13555 }, { "epoch": 0.4829613267541906, "grad_norm": 1.1606440234550395, "learning_rate": 1.10361032661057e-05, "loss": 0.4389, "step": 13556 }, { "epoch": 0.48299695388068475, "grad_norm": 1.5269204800970908, "learning_rate": 1.1034955533039904e-05, "loss": 0.5091, "step": 13557 }, { "epoch": 0.4830325810071789, "grad_norm": 1.1065741927929902, "learning_rate": 1.1033807786192951e-05, "loss": 0.4514, "step": 13558 }, { "epoch": 0.48306820813367296, "grad_norm": 1.6072685515976999, "learning_rate": 1.1032660025580127e-05, "loss": 0.5484, "step": 13559 }, { "epoch": 0.4831038352601671, "grad_norm": 1.013978345613013, "learning_rate": 1.103151225121671e-05, "loss": 0.4123, "step": 13560 }, { "epoch": 0.4831394623866612, "grad_norm": 1.5955062923000105, "learning_rate": 1.1030364463117988e-05, "loss": 0.4772, "step": 13561 }, { "epoch": 0.4831750895131553, "grad_norm": 1.3424880680916977, "learning_rate": 1.1029216661299242e-05, "loss": 0.5517, "step": 13562 }, { "epoch": 0.48321071663964943, "grad_norm": 1.185940862097373, "learning_rate": 1.1028068845775758e-05, "loss": 0.4908, "step": 13563 }, { "epoch": 0.48324634376614356, "grad_norm": 1.1349891319844632, "learning_rate": 1.102692101656282e-05, "loss": 0.4607, "step": 13564 }, { "epoch": 0.48328197089263764, "grad_norm": 1.244235718609692, "learning_rate": 1.1025773173675708e-05, "loss": 0.5087, "step": 13565 }, { "epoch": 0.48331759801913177, "grad_norm": 1.6194570751070585, "learning_rate": 1.1024625317129711e-05, "loss": 0.5867, "step": 13566 }, { "epoch": 0.4833532251456259, "grad_norm": 1.4221369767170295, "learning_rate": 1.102347744694011e-05, "loss": 0.4819, "step": 13567 }, { "epoch": 0.48338885227212, "grad_norm": 1.4564454330376126, "learning_rate": 1.1022329563122191e-05, "loss": 0.5221, "step": 13568 }, { "epoch": 0.4834244793986141, "grad_norm": 1.4520530834199536, "learning_rate": 1.1021181665691244e-05, "loss": 0.5591, "step": 13569 }, { "epoch": 0.48346010652510824, "grad_norm": 1.077427873250066, "learning_rate": 1.1020033754662542e-05, "loss": 0.4599, "step": 13570 }, { "epoch": 0.4834957336516023, "grad_norm": 1.0643497635858816, "learning_rate": 1.1018885830051385e-05, "loss": 0.464, "step": 13571 }, { "epoch": 0.48353136077809644, "grad_norm": 1.22213763378323, "learning_rate": 1.1017737891873048e-05, "loss": 0.4565, "step": 13572 }, { "epoch": 0.4835669879045906, "grad_norm": 1.0730350674764026, "learning_rate": 1.1016589940142817e-05, "loss": 0.421, "step": 13573 }, { "epoch": 0.48360261503108465, "grad_norm": 1.0934113417268796, "learning_rate": 1.1015441974875984e-05, "loss": 0.4765, "step": 13574 }, { "epoch": 0.4836382421575788, "grad_norm": 1.5152545458465732, "learning_rate": 1.101429399608783e-05, "loss": 0.5716, "step": 13575 }, { "epoch": 0.4836738692840729, "grad_norm": 1.1646179761429944, "learning_rate": 1.1013146003793644e-05, "loss": 0.4462, "step": 13576 }, { "epoch": 0.483709496410567, "grad_norm": 1.100758601453759, "learning_rate": 1.1011997998008707e-05, "loss": 0.4409, "step": 13577 }, { "epoch": 0.4837451235370611, "grad_norm": 1.2797383809397491, "learning_rate": 1.1010849978748314e-05, "loss": 0.4621, "step": 13578 }, { "epoch": 0.48378075066355525, "grad_norm": 1.1427900444157773, "learning_rate": 1.1009701946027743e-05, "loss": 0.4361, "step": 13579 }, { "epoch": 0.4838163777900493, "grad_norm": 1.4970568870612573, "learning_rate": 1.1008553899862287e-05, "loss": 0.6515, "step": 13580 }, { "epoch": 0.48385200491654345, "grad_norm": 1.3717107745547277, "learning_rate": 1.1007405840267229e-05, "loss": 0.6178, "step": 13581 }, { "epoch": 0.4838876320430376, "grad_norm": 1.0622018664062343, "learning_rate": 1.100625776725786e-05, "loss": 0.4485, "step": 13582 }, { "epoch": 0.48392325916953166, "grad_norm": 1.8365158540021014, "learning_rate": 1.1005109680849466e-05, "loss": 0.4841, "step": 13583 }, { "epoch": 0.4839588862960258, "grad_norm": 1.0357617604030482, "learning_rate": 1.1003961581057334e-05, "loss": 0.3977, "step": 13584 }, { "epoch": 0.4839945134225199, "grad_norm": 1.1176376944921438, "learning_rate": 1.100281346789675e-05, "loss": 0.4378, "step": 13585 }, { "epoch": 0.484030140549014, "grad_norm": 1.7894063128387152, "learning_rate": 1.1001665341383005e-05, "loss": 0.52, "step": 13586 }, { "epoch": 0.48406576767550813, "grad_norm": 1.5944761561641234, "learning_rate": 1.1000517201531387e-05, "loss": 0.5739, "step": 13587 }, { "epoch": 0.48410139480200226, "grad_norm": 1.757740481765215, "learning_rate": 1.099936904835718e-05, "loss": 0.5739, "step": 13588 }, { "epoch": 0.48413702192849634, "grad_norm": 1.2901723232240148, "learning_rate": 1.0998220881875679e-05, "loss": 0.5787, "step": 13589 }, { "epoch": 0.48417264905499047, "grad_norm": 1.177983758108689, "learning_rate": 1.0997072702102164e-05, "loss": 0.4724, "step": 13590 }, { "epoch": 0.4842082761814846, "grad_norm": 1.4235058885135174, "learning_rate": 1.0995924509051934e-05, "loss": 0.583, "step": 13591 }, { "epoch": 0.4842439033079787, "grad_norm": 1.1737762583664444, "learning_rate": 1.0994776302740271e-05, "loss": 0.4884, "step": 13592 }, { "epoch": 0.4842795304344728, "grad_norm": 1.7719721626171763, "learning_rate": 1.0993628083182468e-05, "loss": 0.5417, "step": 13593 }, { "epoch": 0.48431515756096694, "grad_norm": 1.4452905321119853, "learning_rate": 1.099247985039381e-05, "loss": 0.5293, "step": 13594 }, { "epoch": 0.484350784687461, "grad_norm": 1.1862528254162479, "learning_rate": 1.0991331604389593e-05, "loss": 0.4865, "step": 13595 }, { "epoch": 0.48438641181395514, "grad_norm": 1.1380049180284897, "learning_rate": 1.09901833451851e-05, "loss": 0.4434, "step": 13596 }, { "epoch": 0.4844220389404493, "grad_norm": 1.446422566546567, "learning_rate": 1.0989035072795624e-05, "loss": 0.6276, "step": 13597 }, { "epoch": 0.48445766606694335, "grad_norm": 1.178516615337763, "learning_rate": 1.0987886787236455e-05, "loss": 0.4639, "step": 13598 }, { "epoch": 0.4844932931934375, "grad_norm": 1.5062307796463885, "learning_rate": 1.0986738488522881e-05, "loss": 0.6155, "step": 13599 }, { "epoch": 0.4845289203199316, "grad_norm": 1.1358604504455283, "learning_rate": 1.0985590176670198e-05, "loss": 0.4862, "step": 13600 }, { "epoch": 0.4845645474464257, "grad_norm": 1.0877690015470713, "learning_rate": 1.098444185169369e-05, "loss": 0.4613, "step": 13601 }, { "epoch": 0.4846001745729198, "grad_norm": 1.1317345344549294, "learning_rate": 1.098329351360865e-05, "loss": 0.4913, "step": 13602 }, { "epoch": 0.48463580169941395, "grad_norm": 1.1901857916777, "learning_rate": 1.0982145162430373e-05, "loss": 0.5019, "step": 13603 }, { "epoch": 0.484671428825908, "grad_norm": 1.1795287575497262, "learning_rate": 1.0980996798174143e-05, "loss": 0.5034, "step": 13604 }, { "epoch": 0.48470705595240215, "grad_norm": 1.1289804836426602, "learning_rate": 1.0979848420855256e-05, "loss": 0.4788, "step": 13605 }, { "epoch": 0.4847426830788963, "grad_norm": 1.1242041041180413, "learning_rate": 1.0978700030489007e-05, "loss": 0.4578, "step": 13606 }, { "epoch": 0.48477831020539036, "grad_norm": 1.0684253325579507, "learning_rate": 1.0977551627090677e-05, "loss": 0.4001, "step": 13607 }, { "epoch": 0.4848139373318845, "grad_norm": 1.1525349252784887, "learning_rate": 1.0976403210675566e-05, "loss": 0.4624, "step": 13608 }, { "epoch": 0.4848495644583786, "grad_norm": 1.1558621861193537, "learning_rate": 1.0975254781258966e-05, "loss": 0.4365, "step": 13609 }, { "epoch": 0.4848851915848727, "grad_norm": 1.1162539575235282, "learning_rate": 1.0974106338856166e-05, "loss": 0.3926, "step": 13610 }, { "epoch": 0.48492081871136683, "grad_norm": 1.9761239264188766, "learning_rate": 1.0972957883482456e-05, "loss": 0.5395, "step": 13611 }, { "epoch": 0.48495644583786096, "grad_norm": 1.6979905040122492, "learning_rate": 1.0971809415153135e-05, "loss": 0.5172, "step": 13612 }, { "epoch": 0.48499207296435504, "grad_norm": 1.6607724876619698, "learning_rate": 1.097066093388349e-05, "loss": 0.5208, "step": 13613 }, { "epoch": 0.48502770009084917, "grad_norm": 2.0752855265325416, "learning_rate": 1.0969512439688816e-05, "loss": 0.5547, "step": 13614 }, { "epoch": 0.4850633272173433, "grad_norm": 1.4238176443820956, "learning_rate": 1.0968363932584407e-05, "loss": 0.5864, "step": 13615 }, { "epoch": 0.4850989543438374, "grad_norm": 1.4753990248247448, "learning_rate": 1.0967215412585557e-05, "loss": 0.6551, "step": 13616 }, { "epoch": 0.4851345814703315, "grad_norm": 1.3651406723700412, "learning_rate": 1.0966066879707555e-05, "loss": 0.5802, "step": 13617 }, { "epoch": 0.48517020859682564, "grad_norm": 1.3160045014877424, "learning_rate": 1.0964918333965699e-05, "loss": 0.4391, "step": 13618 }, { "epoch": 0.4852058357233197, "grad_norm": 1.2583494697198128, "learning_rate": 1.0963769775375281e-05, "loss": 0.4493, "step": 13619 }, { "epoch": 0.48524146284981384, "grad_norm": 1.094293456271705, "learning_rate": 1.0962621203951595e-05, "loss": 0.431, "step": 13620 }, { "epoch": 0.485277089976308, "grad_norm": 1.1305365050260205, "learning_rate": 1.0961472619709936e-05, "loss": 0.4555, "step": 13621 }, { "epoch": 0.48531271710280205, "grad_norm": 1.3417740859189038, "learning_rate": 1.0960324022665593e-05, "loss": 0.5854, "step": 13622 }, { "epoch": 0.4853483442292962, "grad_norm": 1.3215894071089378, "learning_rate": 1.0959175412833869e-05, "loss": 0.5588, "step": 13623 }, { "epoch": 0.4853839713557903, "grad_norm": 1.256532438871925, "learning_rate": 1.0958026790230052e-05, "loss": 0.6107, "step": 13624 }, { "epoch": 0.4854195984822844, "grad_norm": 1.1001788236018326, "learning_rate": 1.0956878154869437e-05, "loss": 0.4866, "step": 13625 }, { "epoch": 0.4854552256087785, "grad_norm": 1.9529647223177924, "learning_rate": 1.095572950676732e-05, "loss": 0.528, "step": 13626 }, { "epoch": 0.48549085273527265, "grad_norm": 1.5088189898133226, "learning_rate": 1.0954580845939e-05, "loss": 0.5585, "step": 13627 }, { "epoch": 0.4855264798617667, "grad_norm": 1.0938091573523534, "learning_rate": 1.095343217239977e-05, "loss": 0.5443, "step": 13628 }, { "epoch": 0.48556210698826086, "grad_norm": 1.0526549155873477, "learning_rate": 1.0952283486164923e-05, "loss": 0.4694, "step": 13629 }, { "epoch": 0.485597734114755, "grad_norm": 1.6023834186057249, "learning_rate": 1.0951134787249754e-05, "loss": 0.5427, "step": 13630 }, { "epoch": 0.48563336124124906, "grad_norm": 1.1269894486087035, "learning_rate": 1.0949986075669558e-05, "loss": 0.4443, "step": 13631 }, { "epoch": 0.4856689883677432, "grad_norm": 1.3624574734513162, "learning_rate": 1.0948837351439639e-05, "loss": 0.4747, "step": 13632 }, { "epoch": 0.4857046154942373, "grad_norm": 1.2409077531278834, "learning_rate": 1.0947688614575283e-05, "loss": 0.4876, "step": 13633 }, { "epoch": 0.4857402426207314, "grad_norm": 1.125561708251596, "learning_rate": 1.0946539865091794e-05, "loss": 0.4576, "step": 13634 }, { "epoch": 0.48577586974722553, "grad_norm": 1.314893796138117, "learning_rate": 1.0945391103004464e-05, "loss": 0.5525, "step": 13635 }, { "epoch": 0.48581149687371966, "grad_norm": 1.1888498910560668, "learning_rate": 1.094424232832859e-05, "loss": 0.4548, "step": 13636 }, { "epoch": 0.48584712400021374, "grad_norm": 1.2367857074687083, "learning_rate": 1.0943093541079467e-05, "loss": 0.5, "step": 13637 }, { "epoch": 0.48588275112670787, "grad_norm": 1.568873252902338, "learning_rate": 1.09419447412724e-05, "loss": 0.5607, "step": 13638 }, { "epoch": 0.485918378253202, "grad_norm": 1.9071772366253414, "learning_rate": 1.0940795928922676e-05, "loss": 0.624, "step": 13639 }, { "epoch": 0.4859540053796961, "grad_norm": 1.4702666857130948, "learning_rate": 1.0939647104045598e-05, "loss": 0.6042, "step": 13640 }, { "epoch": 0.4859896325061902, "grad_norm": 1.0998501591024457, "learning_rate": 1.0938498266656463e-05, "loss": 0.496, "step": 13641 }, { "epoch": 0.48602525963268434, "grad_norm": 1.671722719749852, "learning_rate": 1.0937349416770566e-05, "loss": 0.5841, "step": 13642 }, { "epoch": 0.4860608867591784, "grad_norm": 1.8145599159796078, "learning_rate": 1.0936200554403209e-05, "loss": 0.5688, "step": 13643 }, { "epoch": 0.48609651388567254, "grad_norm": 1.073662926736263, "learning_rate": 1.0935051679569684e-05, "loss": 0.4806, "step": 13644 }, { "epoch": 0.4861321410121667, "grad_norm": 1.8212468722060526, "learning_rate": 1.0933902792285294e-05, "loss": 0.5969, "step": 13645 }, { "epoch": 0.48616776813866075, "grad_norm": 1.8156856663730372, "learning_rate": 1.0932753892565335e-05, "loss": 0.6222, "step": 13646 }, { "epoch": 0.4862033952651549, "grad_norm": 1.7631922517251064, "learning_rate": 1.0931604980425108e-05, "loss": 0.596, "step": 13647 }, { "epoch": 0.486239022391649, "grad_norm": 1.4806097829696914, "learning_rate": 1.0930456055879905e-05, "loss": 0.4897, "step": 13648 }, { "epoch": 0.4862746495181431, "grad_norm": 1.2631295714851036, "learning_rate": 1.0929307118945032e-05, "loss": 0.4494, "step": 13649 }, { "epoch": 0.4863102766446372, "grad_norm": 1.1951718717738975, "learning_rate": 1.0928158169635785e-05, "loss": 0.4165, "step": 13650 }, { "epoch": 0.48634590377113135, "grad_norm": 1.5025982279198846, "learning_rate": 1.0927009207967465e-05, "loss": 0.4839, "step": 13651 }, { "epoch": 0.4863815308976254, "grad_norm": 1.5079771520547214, "learning_rate": 1.0925860233955368e-05, "loss": 0.5979, "step": 13652 }, { "epoch": 0.48641715802411956, "grad_norm": 1.5283848454729307, "learning_rate": 1.0924711247614794e-05, "loss": 0.5126, "step": 13653 }, { "epoch": 0.4864527851506137, "grad_norm": 1.5750893761706803, "learning_rate": 1.0923562248961046e-05, "loss": 0.6087, "step": 13654 }, { "epoch": 0.48648841227710776, "grad_norm": 1.0856394395392837, "learning_rate": 1.0922413238009417e-05, "loss": 0.4944, "step": 13655 }, { "epoch": 0.4865240394036019, "grad_norm": 1.190952753524734, "learning_rate": 1.0921264214775214e-05, "loss": 0.5176, "step": 13656 }, { "epoch": 0.486559666530096, "grad_norm": 1.3723656656763645, "learning_rate": 1.0920115179273734e-05, "loss": 0.5596, "step": 13657 }, { "epoch": 0.48659529365659016, "grad_norm": 1.1125757570677435, "learning_rate": 1.0918966131520276e-05, "loss": 0.4443, "step": 13658 }, { "epoch": 0.48663092078308423, "grad_norm": 1.0438766939335935, "learning_rate": 1.091781707153014e-05, "loss": 0.4507, "step": 13659 }, { "epoch": 0.48666654790957836, "grad_norm": 1.0891742187265678, "learning_rate": 1.091666799931863e-05, "loss": 0.4387, "step": 13660 }, { "epoch": 0.4867021750360725, "grad_norm": 1.0317268789359257, "learning_rate": 1.0915518914901047e-05, "loss": 0.4478, "step": 13661 }, { "epoch": 0.48673780216256657, "grad_norm": 1.1431203859244001, "learning_rate": 1.091436981829269e-05, "loss": 0.4881, "step": 13662 }, { "epoch": 0.4867734292890607, "grad_norm": 1.3081687456176088, "learning_rate": 1.0913220709508856e-05, "loss": 0.4458, "step": 13663 }, { "epoch": 0.48680905641555483, "grad_norm": 1.462462764105194, "learning_rate": 1.0912071588564852e-05, "loss": 0.4821, "step": 13664 }, { "epoch": 0.4868446835420489, "grad_norm": 1.1536599125093767, "learning_rate": 1.0910922455475978e-05, "loss": 0.4505, "step": 13665 }, { "epoch": 0.48688031066854304, "grad_norm": 1.1159092142001272, "learning_rate": 1.0909773310257535e-05, "loss": 0.4261, "step": 13666 }, { "epoch": 0.48691593779503717, "grad_norm": 1.1584644357438783, "learning_rate": 1.090862415292482e-05, "loss": 0.4303, "step": 13667 }, { "epoch": 0.48695156492153124, "grad_norm": 1.4032434214844294, "learning_rate": 1.0907474983493144e-05, "loss": 0.5427, "step": 13668 }, { "epoch": 0.4869871920480254, "grad_norm": 1.451772122458668, "learning_rate": 1.0906325801977804e-05, "loss": 0.6479, "step": 13669 }, { "epoch": 0.4870228191745195, "grad_norm": 1.495614938335093, "learning_rate": 1.09051766083941e-05, "loss": 0.5005, "step": 13670 }, { "epoch": 0.4870584463010136, "grad_norm": 1.2821510574409691, "learning_rate": 1.0904027402757337e-05, "loss": 0.6008, "step": 13671 }, { "epoch": 0.4870940734275077, "grad_norm": 1.5048698984208813, "learning_rate": 1.090287818508282e-05, "loss": 0.5837, "step": 13672 }, { "epoch": 0.48712970055400184, "grad_norm": 3.033138233239803, "learning_rate": 1.0901728955385847e-05, "loss": 0.5226, "step": 13673 }, { "epoch": 0.4871653276804959, "grad_norm": 1.4795857969042632, "learning_rate": 1.0900579713681722e-05, "loss": 0.5924, "step": 13674 }, { "epoch": 0.48720095480699005, "grad_norm": 1.4217887181424462, "learning_rate": 1.089943045998575e-05, "loss": 0.5494, "step": 13675 }, { "epoch": 0.4872365819334842, "grad_norm": 1.209361968475364, "learning_rate": 1.0898281194313234e-05, "loss": 0.4203, "step": 13676 }, { "epoch": 0.48727220905997826, "grad_norm": 1.5293125419227314, "learning_rate": 1.0897131916679475e-05, "loss": 0.6372, "step": 13677 }, { "epoch": 0.4873078361864724, "grad_norm": 2.146575313475871, "learning_rate": 1.0895982627099775e-05, "loss": 0.5594, "step": 13678 }, { "epoch": 0.4873434633129665, "grad_norm": 2.6062529387877693, "learning_rate": 1.0894833325589443e-05, "loss": 0.5726, "step": 13679 }, { "epoch": 0.4873790904394606, "grad_norm": 1.4917107222997272, "learning_rate": 1.089368401216378e-05, "loss": 0.5354, "step": 13680 }, { "epoch": 0.4874147175659547, "grad_norm": 1.7060204829100645, "learning_rate": 1.0892534686838086e-05, "loss": 0.5138, "step": 13681 }, { "epoch": 0.48745034469244886, "grad_norm": 1.116748510510991, "learning_rate": 1.0891385349627671e-05, "loss": 0.4633, "step": 13682 }, { "epoch": 0.48748597181894293, "grad_norm": 1.92877235109591, "learning_rate": 1.0890236000547839e-05, "loss": 0.5168, "step": 13683 }, { "epoch": 0.48752159894543706, "grad_norm": 1.2516888878512578, "learning_rate": 1.0889086639613891e-05, "loss": 0.509, "step": 13684 }, { "epoch": 0.4875572260719312, "grad_norm": 2.082844410383842, "learning_rate": 1.0887937266841132e-05, "loss": 0.5694, "step": 13685 }, { "epoch": 0.48759285319842527, "grad_norm": 1.056090404839885, "learning_rate": 1.0886787882244867e-05, "loss": 0.4483, "step": 13686 }, { "epoch": 0.4876284803249194, "grad_norm": 1.653801958825078, "learning_rate": 1.0885638485840404e-05, "loss": 0.5708, "step": 13687 }, { "epoch": 0.48766410745141353, "grad_norm": 1.70937096188416, "learning_rate": 1.0884489077643045e-05, "loss": 0.4839, "step": 13688 }, { "epoch": 0.4876997345779076, "grad_norm": 1.3729248019813889, "learning_rate": 1.0883339657668095e-05, "loss": 0.5133, "step": 13689 }, { "epoch": 0.48773536170440174, "grad_norm": 1.0976512942722116, "learning_rate": 1.088219022593086e-05, "loss": 0.4561, "step": 13690 }, { "epoch": 0.48777098883089587, "grad_norm": 1.7644655771078666, "learning_rate": 1.0881040782446645e-05, "loss": 0.5615, "step": 13691 }, { "epoch": 0.48780661595738994, "grad_norm": 2.285916236148066, "learning_rate": 1.087989132723076e-05, "loss": 0.558, "step": 13692 }, { "epoch": 0.4878422430838841, "grad_norm": 1.6834701810102632, "learning_rate": 1.0878741860298503e-05, "loss": 0.5575, "step": 13693 }, { "epoch": 0.4878778702103782, "grad_norm": 1.0776852581058325, "learning_rate": 1.0877592381665183e-05, "loss": 0.4327, "step": 13694 }, { "epoch": 0.4879134973368723, "grad_norm": 1.4209927053811675, "learning_rate": 1.0876442891346109e-05, "loss": 0.5244, "step": 13695 }, { "epoch": 0.4879491244633664, "grad_norm": 1.091135290366855, "learning_rate": 1.0875293389356585e-05, "loss": 0.4771, "step": 13696 }, { "epoch": 0.48798475158986054, "grad_norm": 1.159666827486469, "learning_rate": 1.0874143875711918e-05, "loss": 0.4522, "step": 13697 }, { "epoch": 0.4880203787163546, "grad_norm": 1.2917061126645608, "learning_rate": 1.0872994350427414e-05, "loss": 0.6091, "step": 13698 }, { "epoch": 0.48805600584284875, "grad_norm": 1.1276966326045839, "learning_rate": 1.0871844813518381e-05, "loss": 0.4182, "step": 13699 }, { "epoch": 0.4880916329693429, "grad_norm": 1.0913980684301035, "learning_rate": 1.0870695265000121e-05, "loss": 0.4601, "step": 13700 }, { "epoch": 0.48812726009583696, "grad_norm": 1.641557424353893, "learning_rate": 1.0869545704887952e-05, "loss": 0.4905, "step": 13701 }, { "epoch": 0.4881628872223311, "grad_norm": 1.0034610308250005, "learning_rate": 1.086839613319717e-05, "loss": 0.4661, "step": 13702 }, { "epoch": 0.4881985143488252, "grad_norm": 1.550591334000108, "learning_rate": 1.0867246549943087e-05, "loss": 0.5512, "step": 13703 }, { "epoch": 0.4882341414753193, "grad_norm": 1.6141852843842295, "learning_rate": 1.0866096955141007e-05, "loss": 0.6022, "step": 13704 }, { "epoch": 0.4882697686018134, "grad_norm": 1.4651838176504237, "learning_rate": 1.0864947348806245e-05, "loss": 0.5433, "step": 13705 }, { "epoch": 0.48830539572830756, "grad_norm": 1.2095773793224784, "learning_rate": 1.0863797730954101e-05, "loss": 0.4664, "step": 13706 }, { "epoch": 0.48834102285480163, "grad_norm": 1.1433053976668117, "learning_rate": 1.086264810159989e-05, "loss": 0.4494, "step": 13707 }, { "epoch": 0.48837664998129576, "grad_norm": 1.6772018839481158, "learning_rate": 1.0861498460758915e-05, "loss": 0.6275, "step": 13708 }, { "epoch": 0.4884122771077899, "grad_norm": 1.0275010009095824, "learning_rate": 1.0860348808446486e-05, "loss": 0.4969, "step": 13709 }, { "epoch": 0.48844790423428397, "grad_norm": 2.2998485661644104, "learning_rate": 1.0859199144677911e-05, "loss": 0.6698, "step": 13710 }, { "epoch": 0.4884835313607781, "grad_norm": 1.47309134889522, "learning_rate": 1.0858049469468501e-05, "loss": 0.5889, "step": 13711 }, { "epoch": 0.48851915848727223, "grad_norm": 1.3828812809138695, "learning_rate": 1.0856899782833559e-05, "loss": 0.4953, "step": 13712 }, { "epoch": 0.4885547856137663, "grad_norm": 1.1734509343701571, "learning_rate": 1.08557500847884e-05, "loss": 0.4324, "step": 13713 }, { "epoch": 0.48859041274026044, "grad_norm": 1.7361882904927115, "learning_rate": 1.085460037534833e-05, "loss": 0.5244, "step": 13714 }, { "epoch": 0.48862603986675457, "grad_norm": 1.3329847355445748, "learning_rate": 1.0853450654528657e-05, "loss": 0.5907, "step": 13715 }, { "epoch": 0.48866166699324864, "grad_norm": 1.3968286015473204, "learning_rate": 1.0852300922344694e-05, "loss": 0.5781, "step": 13716 }, { "epoch": 0.4886972941197428, "grad_norm": 1.9303646613952172, "learning_rate": 1.0851151178811745e-05, "loss": 0.5516, "step": 13717 }, { "epoch": 0.4887329212462369, "grad_norm": 1.1314070127858487, "learning_rate": 1.0850001423945127e-05, "loss": 0.4696, "step": 13718 }, { "epoch": 0.488768548372731, "grad_norm": 2.144249832618864, "learning_rate": 1.0848851657760142e-05, "loss": 0.5134, "step": 13719 }, { "epoch": 0.4888041754992251, "grad_norm": 1.4134756356814542, "learning_rate": 1.0847701880272108e-05, "loss": 0.4828, "step": 13720 }, { "epoch": 0.48883980262571924, "grad_norm": 1.3890060494042766, "learning_rate": 1.0846552091496332e-05, "loss": 0.538, "step": 13721 }, { "epoch": 0.4888754297522133, "grad_norm": 1.5355769335073055, "learning_rate": 1.084540229144812e-05, "loss": 0.5376, "step": 13722 }, { "epoch": 0.48891105687870745, "grad_norm": 1.1080504434391518, "learning_rate": 1.0844252480142785e-05, "loss": 0.4477, "step": 13723 }, { "epoch": 0.4889466840052016, "grad_norm": 1.107402579073841, "learning_rate": 1.0843102657595641e-05, "loss": 0.4639, "step": 13724 }, { "epoch": 0.48898231113169566, "grad_norm": 1.1892930984096761, "learning_rate": 1.0841952823821996e-05, "loss": 0.4441, "step": 13725 }, { "epoch": 0.4890179382581898, "grad_norm": 1.4484547160792287, "learning_rate": 1.0840802978837156e-05, "loss": 0.5479, "step": 13726 }, { "epoch": 0.4890535653846839, "grad_norm": 1.055702890345305, "learning_rate": 1.083965312265644e-05, "loss": 0.4067, "step": 13727 }, { "epoch": 0.489089192511178, "grad_norm": 1.1283641505515594, "learning_rate": 1.0838503255295156e-05, "loss": 0.4775, "step": 13728 }, { "epoch": 0.4891248196376721, "grad_norm": 1.4259178240407493, "learning_rate": 1.0837353376768614e-05, "loss": 0.588, "step": 13729 }, { "epoch": 0.48916044676416626, "grad_norm": 1.1353121044138272, "learning_rate": 1.0836203487092127e-05, "loss": 0.4415, "step": 13730 }, { "epoch": 0.48919607389066033, "grad_norm": 1.443363126861631, "learning_rate": 1.0835053586281004e-05, "loss": 0.5222, "step": 13731 }, { "epoch": 0.48923170101715446, "grad_norm": 2.0415056124615116, "learning_rate": 1.0833903674350563e-05, "loss": 0.5494, "step": 13732 }, { "epoch": 0.4892673281436486, "grad_norm": 1.0667667399147673, "learning_rate": 1.0832753751316109e-05, "loss": 0.4509, "step": 13733 }, { "epoch": 0.48930295527014267, "grad_norm": 1.0300161444056015, "learning_rate": 1.0831603817192957e-05, "loss": 0.4336, "step": 13734 }, { "epoch": 0.4893385823966368, "grad_norm": 1.4025215317201696, "learning_rate": 1.083045387199642e-05, "loss": 0.6392, "step": 13735 }, { "epoch": 0.48937420952313093, "grad_norm": 1.2946185761742466, "learning_rate": 1.0829303915741808e-05, "loss": 0.5412, "step": 13736 }, { "epoch": 0.489409836649625, "grad_norm": 1.0745310317051195, "learning_rate": 1.0828153948444435e-05, "loss": 0.4361, "step": 13737 }, { "epoch": 0.48944546377611914, "grad_norm": 1.8107798685000218, "learning_rate": 1.0827003970119614e-05, "loss": 0.5119, "step": 13738 }, { "epoch": 0.48948109090261327, "grad_norm": 1.7852926258151098, "learning_rate": 1.0825853980782657e-05, "loss": 0.5467, "step": 13739 }, { "epoch": 0.48951671802910734, "grad_norm": 1.0927818914824392, "learning_rate": 1.0824703980448876e-05, "loss": 0.4456, "step": 13740 }, { "epoch": 0.4895523451556015, "grad_norm": 1.2128631866178032, "learning_rate": 1.0823553969133587e-05, "loss": 0.4623, "step": 13741 }, { "epoch": 0.4895879722820956, "grad_norm": 1.4395432975385707, "learning_rate": 1.0822403946852099e-05, "loss": 0.5489, "step": 13742 }, { "epoch": 0.4896235994085897, "grad_norm": 1.5848996577916525, "learning_rate": 1.0821253913619727e-05, "loss": 0.522, "step": 13743 }, { "epoch": 0.4896592265350838, "grad_norm": 1.0904770835592499, "learning_rate": 1.0820103869451788e-05, "loss": 0.43, "step": 13744 }, { "epoch": 0.48969485366157794, "grad_norm": 1.0493692697783337, "learning_rate": 1.081895381436359e-05, "loss": 0.4403, "step": 13745 }, { "epoch": 0.489730480788072, "grad_norm": 1.1250377630989654, "learning_rate": 1.0817803748370453e-05, "loss": 0.4438, "step": 13746 }, { "epoch": 0.48976610791456615, "grad_norm": 1.9401542163461762, "learning_rate": 1.0816653671487685e-05, "loss": 0.5897, "step": 13747 }, { "epoch": 0.4898017350410603, "grad_norm": 1.927802958069311, "learning_rate": 1.0815503583730604e-05, "loss": 0.5798, "step": 13748 }, { "epoch": 0.48983736216755436, "grad_norm": 1.9548751040711698, "learning_rate": 1.0814353485114521e-05, "loss": 0.6392, "step": 13749 }, { "epoch": 0.4898729892940485, "grad_norm": 1.035494879807253, "learning_rate": 1.0813203375654753e-05, "loss": 0.4126, "step": 13750 }, { "epoch": 0.4899086164205426, "grad_norm": 1.6506068062571535, "learning_rate": 1.0812053255366614e-05, "loss": 0.5867, "step": 13751 }, { "epoch": 0.4899442435470367, "grad_norm": 1.6811926958578631, "learning_rate": 1.0810903124265417e-05, "loss": 0.5423, "step": 13752 }, { "epoch": 0.4899798706735308, "grad_norm": 1.54930324587435, "learning_rate": 1.0809752982366479e-05, "loss": 0.6393, "step": 13753 }, { "epoch": 0.49001549780002496, "grad_norm": 1.1058435769529344, "learning_rate": 1.0808602829685113e-05, "loss": 0.4652, "step": 13754 }, { "epoch": 0.49005112492651903, "grad_norm": 1.4569962241175614, "learning_rate": 1.0807452666236638e-05, "loss": 0.6044, "step": 13755 }, { "epoch": 0.49008675205301316, "grad_norm": 1.8749882241229596, "learning_rate": 1.0806302492036363e-05, "loss": 0.5011, "step": 13756 }, { "epoch": 0.4901223791795073, "grad_norm": 1.1127974661727877, "learning_rate": 1.080515230709961e-05, "loss": 0.4579, "step": 13757 }, { "epoch": 0.49015800630600137, "grad_norm": 1.2457729037498348, "learning_rate": 1.080400211144169e-05, "loss": 0.5222, "step": 13758 }, { "epoch": 0.4901936334324955, "grad_norm": 1.067612179901686, "learning_rate": 1.080285190507792e-05, "loss": 0.5061, "step": 13759 }, { "epoch": 0.49022926055898963, "grad_norm": 1.1273341734817492, "learning_rate": 1.0801701688023614e-05, "loss": 0.4626, "step": 13760 }, { "epoch": 0.4902648876854837, "grad_norm": 1.1315185415113653, "learning_rate": 1.0800551460294092e-05, "loss": 0.4318, "step": 13761 }, { "epoch": 0.49030051481197784, "grad_norm": 1.4743173462517543, "learning_rate": 1.0799401221904668e-05, "loss": 0.5996, "step": 13762 }, { "epoch": 0.49033614193847197, "grad_norm": 1.1614469483860284, "learning_rate": 1.0798250972870653e-05, "loss": 0.4534, "step": 13763 }, { "epoch": 0.49037176906496605, "grad_norm": 1.103337631947006, "learning_rate": 1.0797100713207371e-05, "loss": 0.5015, "step": 13764 }, { "epoch": 0.4904073961914602, "grad_norm": 1.5372702786351762, "learning_rate": 1.0795950442930138e-05, "loss": 0.5246, "step": 13765 }, { "epoch": 0.4904430233179543, "grad_norm": 1.3084509083194538, "learning_rate": 1.079480016205427e-05, "loss": 0.479, "step": 13766 }, { "epoch": 0.4904786504444484, "grad_norm": 1.1760768939935433, "learning_rate": 1.079364987059508e-05, "loss": 0.4483, "step": 13767 }, { "epoch": 0.4905142775709425, "grad_norm": 1.3013085855308766, "learning_rate": 1.0792499568567885e-05, "loss": 0.5566, "step": 13768 }, { "epoch": 0.49054990469743665, "grad_norm": 1.0308123871606785, "learning_rate": 1.0791349255988009e-05, "loss": 0.4483, "step": 13769 }, { "epoch": 0.4905855318239307, "grad_norm": 1.1291712460791283, "learning_rate": 1.0790198932870761e-05, "loss": 0.4821, "step": 13770 }, { "epoch": 0.49062115895042485, "grad_norm": 1.3596983376901182, "learning_rate": 1.0789048599231462e-05, "loss": 0.4776, "step": 13771 }, { "epoch": 0.490656786076919, "grad_norm": 1.2756184392640362, "learning_rate": 1.0787898255085429e-05, "loss": 0.4465, "step": 13772 }, { "epoch": 0.49069241320341306, "grad_norm": 1.485115697171627, "learning_rate": 1.0786747900447983e-05, "loss": 0.5382, "step": 13773 }, { "epoch": 0.4907280403299072, "grad_norm": 1.343317428061729, "learning_rate": 1.0785597535334437e-05, "loss": 0.5752, "step": 13774 }, { "epoch": 0.4907636674564013, "grad_norm": 1.494257423385141, "learning_rate": 1.0784447159760107e-05, "loss": 0.6095, "step": 13775 }, { "epoch": 0.4907992945828954, "grad_norm": 1.1433430512452547, "learning_rate": 1.078329677374032e-05, "loss": 0.4532, "step": 13776 }, { "epoch": 0.4908349217093895, "grad_norm": 1.138998008370552, "learning_rate": 1.0782146377290388e-05, "loss": 0.4481, "step": 13777 }, { "epoch": 0.49087054883588366, "grad_norm": 1.6706830172435576, "learning_rate": 1.0780995970425633e-05, "loss": 0.5855, "step": 13778 }, { "epoch": 0.49090617596237773, "grad_norm": 1.1060345682873296, "learning_rate": 1.0779845553161364e-05, "loss": 0.4324, "step": 13779 }, { "epoch": 0.49094180308887186, "grad_norm": 1.615028160876531, "learning_rate": 1.077869512551291e-05, "loss": 0.638, "step": 13780 }, { "epoch": 0.490977430215366, "grad_norm": 1.278962847472533, "learning_rate": 1.0777544687495588e-05, "loss": 0.4603, "step": 13781 }, { "epoch": 0.49101305734186007, "grad_norm": 1.6324848229108964, "learning_rate": 1.077639423912471e-05, "loss": 0.5867, "step": 13782 }, { "epoch": 0.4910486844683542, "grad_norm": 1.1105367949754628, "learning_rate": 1.0775243780415606e-05, "loss": 0.5174, "step": 13783 }, { "epoch": 0.49108431159484833, "grad_norm": 1.173242670973088, "learning_rate": 1.0774093311383587e-05, "loss": 0.4432, "step": 13784 }, { "epoch": 0.4911199387213424, "grad_norm": 1.788726963771857, "learning_rate": 1.0772942832043974e-05, "loss": 0.4523, "step": 13785 }, { "epoch": 0.49115556584783654, "grad_norm": 1.1429716034343842, "learning_rate": 1.0771792342412084e-05, "loss": 0.4774, "step": 13786 }, { "epoch": 0.49119119297433067, "grad_norm": 1.658086307475478, "learning_rate": 1.0770641842503242e-05, "loss": 0.6047, "step": 13787 }, { "epoch": 0.49122682010082475, "grad_norm": 1.9272681446322202, "learning_rate": 1.0769491332332765e-05, "loss": 0.5055, "step": 13788 }, { "epoch": 0.4912624472273189, "grad_norm": 1.1356540687782573, "learning_rate": 1.0768340811915974e-05, "loss": 0.4688, "step": 13789 }, { "epoch": 0.491298074353813, "grad_norm": 1.1010011084792775, "learning_rate": 1.0767190281268187e-05, "loss": 0.4621, "step": 13790 }, { "epoch": 0.4913337014803071, "grad_norm": 1.1367555840341796, "learning_rate": 1.0766039740404726e-05, "loss": 0.5066, "step": 13791 }, { "epoch": 0.4913693286068012, "grad_norm": 1.3855145481748392, "learning_rate": 1.0764889189340911e-05, "loss": 0.5004, "step": 13792 }, { "epoch": 0.49140495573329535, "grad_norm": 1.5020858596677331, "learning_rate": 1.0763738628092061e-05, "loss": 0.5258, "step": 13793 }, { "epoch": 0.4914405828597894, "grad_norm": 1.1579797062970285, "learning_rate": 1.0762588056673497e-05, "loss": 0.5096, "step": 13794 }, { "epoch": 0.49147620998628355, "grad_norm": 1.2966639268996718, "learning_rate": 1.076143747510054e-05, "loss": 0.5378, "step": 13795 }, { "epoch": 0.4915118371127777, "grad_norm": 1.5242518268184309, "learning_rate": 1.0760286883388514e-05, "loss": 0.5106, "step": 13796 }, { "epoch": 0.49154746423927176, "grad_norm": 1.2038000658182602, "learning_rate": 1.075913628155273e-05, "loss": 0.501, "step": 13797 }, { "epoch": 0.4915830913657659, "grad_norm": 1.6574100098515423, "learning_rate": 1.075798566960852e-05, "loss": 0.618, "step": 13798 }, { "epoch": 0.49161871849226, "grad_norm": 1.0546790433038449, "learning_rate": 1.07568350475712e-05, "loss": 0.4057, "step": 13799 }, { "epoch": 0.4916543456187541, "grad_norm": 1.5699753884368521, "learning_rate": 1.0755684415456093e-05, "loss": 0.4747, "step": 13800 }, { "epoch": 0.4916899727452482, "grad_norm": 1.0504229981328685, "learning_rate": 1.075453377327852e-05, "loss": 0.4539, "step": 13801 }, { "epoch": 0.49172559987174236, "grad_norm": 1.1736304098154209, "learning_rate": 1.0753383121053801e-05, "loss": 0.4403, "step": 13802 }, { "epoch": 0.49176122699823643, "grad_norm": 1.333540790953531, "learning_rate": 1.0752232458797262e-05, "loss": 0.516, "step": 13803 }, { "epoch": 0.49179685412473056, "grad_norm": 1.6018832087600845, "learning_rate": 1.0751081786524223e-05, "loss": 0.5766, "step": 13804 }, { "epoch": 0.4918324812512247, "grad_norm": 1.5790977991684498, "learning_rate": 1.074993110425e-05, "loss": 0.5095, "step": 13805 }, { "epoch": 0.49186810837771877, "grad_norm": 1.8125995367705874, "learning_rate": 1.0748780411989924e-05, "loss": 0.4834, "step": 13806 }, { "epoch": 0.4919037355042129, "grad_norm": 1.1897169367688194, "learning_rate": 1.0747629709759313e-05, "loss": 0.4936, "step": 13807 }, { "epoch": 0.49193936263070703, "grad_norm": 1.4528825276988877, "learning_rate": 1.0746478997573486e-05, "loss": 0.5557, "step": 13808 }, { "epoch": 0.4919749897572011, "grad_norm": 1.0644362272803636, "learning_rate": 1.0745328275447774e-05, "loss": 0.4044, "step": 13809 }, { "epoch": 0.49201061688369524, "grad_norm": 1.0256185814441532, "learning_rate": 1.0744177543397492e-05, "loss": 0.4182, "step": 13810 }, { "epoch": 0.49204624401018937, "grad_norm": 1.1166046310513789, "learning_rate": 1.0743026801437968e-05, "loss": 0.4588, "step": 13811 }, { "epoch": 0.49208187113668345, "grad_norm": 1.2994103191619062, "learning_rate": 1.0741876049584523e-05, "loss": 0.4399, "step": 13812 }, { "epoch": 0.4921174982631776, "grad_norm": 3.0870895328261705, "learning_rate": 1.0740725287852479e-05, "loss": 0.6077, "step": 13813 }, { "epoch": 0.4921531253896717, "grad_norm": 1.5709962801016581, "learning_rate": 1.0739574516257159e-05, "loss": 0.6261, "step": 13814 }, { "epoch": 0.4921887525161658, "grad_norm": 1.454613959692219, "learning_rate": 1.073842373481389e-05, "loss": 0.5916, "step": 13815 }, { "epoch": 0.4922243796426599, "grad_norm": 2.2567691606613574, "learning_rate": 1.0737272943537989e-05, "loss": 0.5439, "step": 13816 }, { "epoch": 0.49226000676915405, "grad_norm": 1.259015932788506, "learning_rate": 1.0736122142444788e-05, "loss": 0.4479, "step": 13817 }, { "epoch": 0.4922956338956481, "grad_norm": 2.715739147193936, "learning_rate": 1.0734971331549604e-05, "loss": 0.4642, "step": 13818 }, { "epoch": 0.49233126102214225, "grad_norm": 1.2439324231983908, "learning_rate": 1.0733820510867763e-05, "loss": 0.5134, "step": 13819 }, { "epoch": 0.4923668881486364, "grad_norm": 1.1212560492623067, "learning_rate": 1.0732669680414587e-05, "loss": 0.448, "step": 13820 }, { "epoch": 0.49240251527513046, "grad_norm": 1.3740824521440311, "learning_rate": 1.0731518840205402e-05, "loss": 0.4462, "step": 13821 }, { "epoch": 0.4924381424016246, "grad_norm": 1.8054523706195522, "learning_rate": 1.0730367990255535e-05, "loss": 0.5403, "step": 13822 }, { "epoch": 0.4924737695281187, "grad_norm": 1.285477621555459, "learning_rate": 1.0729217130580309e-05, "loss": 0.4607, "step": 13823 }, { "epoch": 0.4925093966546128, "grad_norm": 1.2385993566891365, "learning_rate": 1.0728066261195042e-05, "loss": 0.4688, "step": 13824 }, { "epoch": 0.4925450237811069, "grad_norm": 1.4769668517314205, "learning_rate": 1.0726915382115068e-05, "loss": 0.496, "step": 13825 }, { "epoch": 0.49258065090760106, "grad_norm": 1.138744387950025, "learning_rate": 1.0725764493355709e-05, "loss": 0.4391, "step": 13826 }, { "epoch": 0.49261627803409513, "grad_norm": 1.489593674443834, "learning_rate": 1.0724613594932282e-05, "loss": 0.5815, "step": 13827 }, { "epoch": 0.49265190516058927, "grad_norm": 1.1257261133582932, "learning_rate": 1.0723462686860124e-05, "loss": 0.4522, "step": 13828 }, { "epoch": 0.4926875322870834, "grad_norm": 1.8433109472425837, "learning_rate": 1.0722311769154557e-05, "loss": 0.5467, "step": 13829 }, { "epoch": 0.49272315941357747, "grad_norm": 1.2250639445725617, "learning_rate": 1.07211608418309e-05, "loss": 0.4664, "step": 13830 }, { "epoch": 0.4927587865400716, "grad_norm": 1.3739926420383575, "learning_rate": 1.072000990490448e-05, "loss": 0.4918, "step": 13831 }, { "epoch": 0.49279441366656573, "grad_norm": 1.3791715558635256, "learning_rate": 1.0718858958390627e-05, "loss": 0.5274, "step": 13832 }, { "epoch": 0.4928300407930598, "grad_norm": 1.3769584976234042, "learning_rate": 1.0717708002304666e-05, "loss": 0.4498, "step": 13833 }, { "epoch": 0.49286566791955394, "grad_norm": 1.2277827393703784, "learning_rate": 1.071655703666192e-05, "loss": 0.4735, "step": 13834 }, { "epoch": 0.49290129504604807, "grad_norm": 1.8221171079351701, "learning_rate": 1.0715406061477716e-05, "loss": 0.5566, "step": 13835 }, { "epoch": 0.49293692217254215, "grad_norm": 1.8435404049994246, "learning_rate": 1.071425507676738e-05, "loss": 0.5052, "step": 13836 }, { "epoch": 0.4929725492990363, "grad_norm": 1.0589869730153734, "learning_rate": 1.0713104082546239e-05, "loss": 0.4913, "step": 13837 }, { "epoch": 0.4930081764255304, "grad_norm": 1.7599077637083613, "learning_rate": 1.0711953078829617e-05, "loss": 0.5244, "step": 13838 }, { "epoch": 0.4930438035520245, "grad_norm": 1.0970551679011882, "learning_rate": 1.0710802065632846e-05, "loss": 0.4612, "step": 13839 }, { "epoch": 0.4930794306785186, "grad_norm": 1.6825954305535014, "learning_rate": 1.0709651042971248e-05, "loss": 0.6019, "step": 13840 }, { "epoch": 0.49311505780501275, "grad_norm": 1.1757190421124726, "learning_rate": 1.070850001086015e-05, "loss": 0.4688, "step": 13841 }, { "epoch": 0.4931506849315068, "grad_norm": 1.3901511588594422, "learning_rate": 1.0707348969314877e-05, "loss": 0.5278, "step": 13842 }, { "epoch": 0.49318631205800095, "grad_norm": 1.0936277688182663, "learning_rate": 1.0706197918350758e-05, "loss": 0.4492, "step": 13843 }, { "epoch": 0.4932219391844951, "grad_norm": 1.1541764473759542, "learning_rate": 1.070504685798312e-05, "loss": 0.473, "step": 13844 }, { "epoch": 0.49325756631098916, "grad_norm": 1.3098930228747572, "learning_rate": 1.0703895788227292e-05, "loss": 0.5291, "step": 13845 }, { "epoch": 0.4932931934374833, "grad_norm": 1.1603384984328033, "learning_rate": 1.0702744709098598e-05, "loss": 0.4514, "step": 13846 }, { "epoch": 0.4933288205639774, "grad_norm": 1.4165243577341302, "learning_rate": 1.0701593620612369e-05, "loss": 0.5001, "step": 13847 }, { "epoch": 0.49336444769047155, "grad_norm": 1.3506374323212766, "learning_rate": 1.070044252278393e-05, "loss": 0.5153, "step": 13848 }, { "epoch": 0.49340007481696563, "grad_norm": 1.6967584930852257, "learning_rate": 1.069929141562861e-05, "loss": 0.5113, "step": 13849 }, { "epoch": 0.49343570194345976, "grad_norm": 1.4953900700228384, "learning_rate": 1.0698140299161735e-05, "loss": 0.5888, "step": 13850 }, { "epoch": 0.4934713290699539, "grad_norm": 1.2125613798045722, "learning_rate": 1.0696989173398633e-05, "loss": 0.4194, "step": 13851 }, { "epoch": 0.49350695619644797, "grad_norm": 1.2098654426562703, "learning_rate": 1.0695838038354635e-05, "loss": 0.4828, "step": 13852 }, { "epoch": 0.4935425833229421, "grad_norm": 1.3682882533539442, "learning_rate": 1.0694686894045066e-05, "loss": 0.4769, "step": 13853 }, { "epoch": 0.4935782104494362, "grad_norm": 1.156360844362486, "learning_rate": 1.0693535740485255e-05, "loss": 0.5068, "step": 13854 }, { "epoch": 0.4936138375759303, "grad_norm": 1.0564006456498107, "learning_rate": 1.0692384577690533e-05, "loss": 0.4681, "step": 13855 }, { "epoch": 0.49364946470242443, "grad_norm": 1.499998876320825, "learning_rate": 1.0691233405676222e-05, "loss": 0.6682, "step": 13856 }, { "epoch": 0.49368509182891857, "grad_norm": 1.325027791991425, "learning_rate": 1.0690082224457656e-05, "loss": 0.4639, "step": 13857 }, { "epoch": 0.49372071895541264, "grad_norm": 1.8731475841950178, "learning_rate": 1.0688931034050167e-05, "loss": 0.5972, "step": 13858 }, { "epoch": 0.49375634608190677, "grad_norm": 1.166273993736556, "learning_rate": 1.0687779834469079e-05, "loss": 0.4819, "step": 13859 }, { "epoch": 0.4937919732084009, "grad_norm": 1.496297247645197, "learning_rate": 1.068662862572972e-05, "loss": 0.5616, "step": 13860 }, { "epoch": 0.493827600334895, "grad_norm": 1.6274499567048957, "learning_rate": 1.068547740784742e-05, "loss": 0.6065, "step": 13861 }, { "epoch": 0.4938632274613891, "grad_norm": 1.5977426270040447, "learning_rate": 1.068432618083751e-05, "loss": 0.5739, "step": 13862 }, { "epoch": 0.49389885458788324, "grad_norm": 1.403446009134909, "learning_rate": 1.068317494471532e-05, "loss": 0.5409, "step": 13863 }, { "epoch": 0.4939344817143773, "grad_norm": 1.2713703540248433, "learning_rate": 1.0682023699496176e-05, "loss": 0.538, "step": 13864 }, { "epoch": 0.49397010884087145, "grad_norm": 1.072315995996637, "learning_rate": 1.0680872445195411e-05, "loss": 0.4676, "step": 13865 }, { "epoch": 0.4940057359673656, "grad_norm": 1.2580904515443665, "learning_rate": 1.0679721181828356e-05, "loss": 0.4687, "step": 13866 }, { "epoch": 0.49404136309385965, "grad_norm": 1.2057790832477315, "learning_rate": 1.0678569909410332e-05, "loss": 0.4399, "step": 13867 }, { "epoch": 0.4940769902203538, "grad_norm": 1.3430941485334265, "learning_rate": 1.067741862795668e-05, "loss": 0.4583, "step": 13868 }, { "epoch": 0.4941126173468479, "grad_norm": 1.2210934215648006, "learning_rate": 1.0676267337482723e-05, "loss": 0.4765, "step": 13869 }, { "epoch": 0.494148244473342, "grad_norm": 1.3109910281246537, "learning_rate": 1.0675116038003796e-05, "loss": 0.5429, "step": 13870 }, { "epoch": 0.4941838715998361, "grad_norm": 1.071473388691548, "learning_rate": 1.0673964729535227e-05, "loss": 0.3977, "step": 13871 }, { "epoch": 0.49421949872633025, "grad_norm": 1.4181527352856405, "learning_rate": 1.0672813412092345e-05, "loss": 0.6036, "step": 13872 }, { "epoch": 0.49425512585282433, "grad_norm": 1.5477914341923638, "learning_rate": 1.0671662085690484e-05, "loss": 0.5523, "step": 13873 }, { "epoch": 0.49429075297931846, "grad_norm": 1.5046941422571076, "learning_rate": 1.0670510750344971e-05, "loss": 0.5824, "step": 13874 }, { "epoch": 0.4943263801058126, "grad_norm": 1.6201848077369079, "learning_rate": 1.066935940607114e-05, "loss": 0.5447, "step": 13875 }, { "epoch": 0.49436200723230667, "grad_norm": 1.1220529694102612, "learning_rate": 1.0668208052884318e-05, "loss": 0.4166, "step": 13876 }, { "epoch": 0.4943976343588008, "grad_norm": 1.6274520910223302, "learning_rate": 1.0667056690799844e-05, "loss": 0.497, "step": 13877 }, { "epoch": 0.49443326148529493, "grad_norm": 1.1505634959744504, "learning_rate": 1.066590531983304e-05, "loss": 0.4883, "step": 13878 }, { "epoch": 0.494468888611789, "grad_norm": 1.0938742596580338, "learning_rate": 1.0664753939999241e-05, "loss": 0.4111, "step": 13879 }, { "epoch": 0.49450451573828313, "grad_norm": 1.4043806828262086, "learning_rate": 1.0663602551313779e-05, "loss": 0.5153, "step": 13880 }, { "epoch": 0.49454014286477727, "grad_norm": 1.793708835665706, "learning_rate": 1.0662451153791987e-05, "loss": 0.5798, "step": 13881 }, { "epoch": 0.49457576999127134, "grad_norm": 1.117349135748221, "learning_rate": 1.0661299747449194e-05, "loss": 0.4446, "step": 13882 }, { "epoch": 0.49461139711776547, "grad_norm": 1.5563653351382536, "learning_rate": 1.066014833230073e-05, "loss": 0.4968, "step": 13883 }, { "epoch": 0.4946470242442596, "grad_norm": 2.610756935026734, "learning_rate": 1.0658996908361934e-05, "loss": 0.5463, "step": 13884 }, { "epoch": 0.4946826513707537, "grad_norm": 1.2068431365466055, "learning_rate": 1.0657845475648132e-05, "loss": 0.4677, "step": 13885 }, { "epoch": 0.4947182784972478, "grad_norm": 1.171788157281458, "learning_rate": 1.0656694034174657e-05, "loss": 0.4718, "step": 13886 }, { "epoch": 0.49475390562374194, "grad_norm": 1.2583932782032858, "learning_rate": 1.0655542583956841e-05, "loss": 0.4713, "step": 13887 }, { "epoch": 0.494789532750236, "grad_norm": 1.1083621492632967, "learning_rate": 1.065439112501002e-05, "loss": 0.5118, "step": 13888 }, { "epoch": 0.49482515987673015, "grad_norm": 1.0848541967671306, "learning_rate": 1.0653239657349524e-05, "loss": 0.4487, "step": 13889 }, { "epoch": 0.4948607870032243, "grad_norm": 1.165915375307046, "learning_rate": 1.0652088180990681e-05, "loss": 0.4498, "step": 13890 }, { "epoch": 0.49489641412971835, "grad_norm": 1.5713660508020952, "learning_rate": 1.0650936695948827e-05, "loss": 0.6159, "step": 13891 }, { "epoch": 0.4949320412562125, "grad_norm": 1.0777416848526116, "learning_rate": 1.0649785202239302e-05, "loss": 0.5094, "step": 13892 }, { "epoch": 0.4949676683827066, "grad_norm": 1.2622782545746205, "learning_rate": 1.064863369987743e-05, "loss": 0.5111, "step": 13893 }, { "epoch": 0.4950032955092007, "grad_norm": 1.5385142991927743, "learning_rate": 1.0647482188878549e-05, "loss": 0.5821, "step": 13894 }, { "epoch": 0.4950389226356948, "grad_norm": 1.8519221707964417, "learning_rate": 1.0646330669257988e-05, "loss": 0.5877, "step": 13895 }, { "epoch": 0.49507454976218895, "grad_norm": 1.468231633102829, "learning_rate": 1.0645179141031084e-05, "loss": 0.5404, "step": 13896 }, { "epoch": 0.49511017688868303, "grad_norm": 1.1236515252342152, "learning_rate": 1.0644027604213168e-05, "loss": 0.4628, "step": 13897 }, { "epoch": 0.49514580401517716, "grad_norm": 1.5098743738154827, "learning_rate": 1.064287605881957e-05, "loss": 0.6016, "step": 13898 }, { "epoch": 0.4951814311416713, "grad_norm": 1.1304427720225894, "learning_rate": 1.0641724504865634e-05, "loss": 0.4414, "step": 13899 }, { "epoch": 0.49521705826816537, "grad_norm": 1.3438828707997539, "learning_rate": 1.0640572942366686e-05, "loss": 0.4684, "step": 13900 }, { "epoch": 0.4952526853946595, "grad_norm": 1.3244414743731847, "learning_rate": 1.0639421371338061e-05, "loss": 0.5663, "step": 13901 }, { "epoch": 0.49528831252115363, "grad_norm": 1.0656124932672448, "learning_rate": 1.063826979179509e-05, "loss": 0.4912, "step": 13902 }, { "epoch": 0.4953239396476477, "grad_norm": 1.654959713277276, "learning_rate": 1.0637118203753116e-05, "loss": 0.4909, "step": 13903 }, { "epoch": 0.49535956677414184, "grad_norm": 1.327803349186354, "learning_rate": 1.0635966607227466e-05, "loss": 0.4985, "step": 13904 }, { "epoch": 0.49539519390063597, "grad_norm": 1.7345990210002489, "learning_rate": 1.0634815002233478e-05, "loss": 0.5496, "step": 13905 }, { "epoch": 0.49543082102713004, "grad_norm": 1.3517675059130771, "learning_rate": 1.0633663388786481e-05, "loss": 0.4464, "step": 13906 }, { "epoch": 0.4954664481536242, "grad_norm": 1.018933255272388, "learning_rate": 1.0632511766901815e-05, "loss": 0.4255, "step": 13907 }, { "epoch": 0.4955020752801183, "grad_norm": 1.5185730685617345, "learning_rate": 1.0631360136594813e-05, "loss": 0.5777, "step": 13908 }, { "epoch": 0.4955377024066124, "grad_norm": 1.1288344782392277, "learning_rate": 1.0630208497880807e-05, "loss": 0.4762, "step": 13909 }, { "epoch": 0.4955733295331065, "grad_norm": 1.5050124698993452, "learning_rate": 1.0629056850775136e-05, "loss": 0.5205, "step": 13910 }, { "epoch": 0.49560895665960064, "grad_norm": 1.1881274893190268, "learning_rate": 1.0627905195293135e-05, "loss": 0.4965, "step": 13911 }, { "epoch": 0.4956445837860947, "grad_norm": 1.344614832020363, "learning_rate": 1.0626753531450138e-05, "loss": 0.5839, "step": 13912 }, { "epoch": 0.49568021091258885, "grad_norm": 1.0224952639582847, "learning_rate": 1.0625601859261477e-05, "loss": 0.4154, "step": 13913 }, { "epoch": 0.495715838039083, "grad_norm": 1.4559841450148714, "learning_rate": 1.062445017874249e-05, "loss": 0.5119, "step": 13914 }, { "epoch": 0.49575146516557705, "grad_norm": 1.7181216967887856, "learning_rate": 1.0623298489908512e-05, "loss": 0.5725, "step": 13915 }, { "epoch": 0.4957870922920712, "grad_norm": 1.8623121750123077, "learning_rate": 1.0622146792774882e-05, "loss": 0.5223, "step": 13916 }, { "epoch": 0.4958227194185653, "grad_norm": 1.056413834934124, "learning_rate": 1.0620995087356929e-05, "loss": 0.4423, "step": 13917 }, { "epoch": 0.4958583465450594, "grad_norm": 1.1758637342786453, "learning_rate": 1.0619843373669993e-05, "loss": 0.4637, "step": 13918 }, { "epoch": 0.4958939736715535, "grad_norm": 2.5469235669336663, "learning_rate": 1.0618691651729412e-05, "loss": 0.4755, "step": 13919 }, { "epoch": 0.49592960079804765, "grad_norm": 1.1287041801492277, "learning_rate": 1.0617539921550515e-05, "loss": 0.4573, "step": 13920 }, { "epoch": 0.49596522792454173, "grad_norm": 1.0267498863099978, "learning_rate": 1.0616388183148646e-05, "loss": 0.4324, "step": 13921 }, { "epoch": 0.49600085505103586, "grad_norm": 2.029924670960829, "learning_rate": 1.0615236436539137e-05, "loss": 0.5166, "step": 13922 }, { "epoch": 0.49603648217753, "grad_norm": 1.2891893442416806, "learning_rate": 1.0614084681737324e-05, "loss": 0.4148, "step": 13923 }, { "epoch": 0.49607210930402407, "grad_norm": 1.0756236913602857, "learning_rate": 1.0612932918758546e-05, "loss": 0.4266, "step": 13924 }, { "epoch": 0.4961077364305182, "grad_norm": 1.3508750795491578, "learning_rate": 1.0611781147618133e-05, "loss": 0.5379, "step": 13925 }, { "epoch": 0.49614336355701233, "grad_norm": 1.0422062340665421, "learning_rate": 1.061062936833143e-05, "loss": 0.4622, "step": 13926 }, { "epoch": 0.4961789906835064, "grad_norm": 1.3643065201409366, "learning_rate": 1.060947758091377e-05, "loss": 0.4353, "step": 13927 }, { "epoch": 0.49621461781000054, "grad_norm": 1.6340725811274477, "learning_rate": 1.0608325785380488e-05, "loss": 0.5989, "step": 13928 }, { "epoch": 0.49625024493649467, "grad_norm": 2.835775849403345, "learning_rate": 1.0607173981746926e-05, "loss": 0.482, "step": 13929 }, { "epoch": 0.49628587206298874, "grad_norm": 1.5034145210389995, "learning_rate": 1.0606022170028418e-05, "loss": 0.5977, "step": 13930 }, { "epoch": 0.4963214991894829, "grad_norm": 1.2467605625895282, "learning_rate": 1.0604870350240299e-05, "loss": 0.6271, "step": 13931 }, { "epoch": 0.496357126315977, "grad_norm": 1.4773536394217417, "learning_rate": 1.060371852239791e-05, "loss": 0.5919, "step": 13932 }, { "epoch": 0.4963927534424711, "grad_norm": 2.4878910337799316, "learning_rate": 1.0602566686516586e-05, "loss": 0.6635, "step": 13933 }, { "epoch": 0.4964283805689652, "grad_norm": 1.1725625641525854, "learning_rate": 1.0601414842611665e-05, "loss": 0.4452, "step": 13934 }, { "epoch": 0.49646400769545934, "grad_norm": 1.6993442916778752, "learning_rate": 1.0600262990698487e-05, "loss": 0.5456, "step": 13935 }, { "epoch": 0.4964996348219534, "grad_norm": 1.2190670026911896, "learning_rate": 1.0599111130792385e-05, "loss": 0.4347, "step": 13936 }, { "epoch": 0.49653526194844755, "grad_norm": 1.1649176129319834, "learning_rate": 1.0597959262908702e-05, "loss": 0.4564, "step": 13937 }, { "epoch": 0.4965708890749417, "grad_norm": 1.180182390440096, "learning_rate": 1.0596807387062772e-05, "loss": 0.4543, "step": 13938 }, { "epoch": 0.49660651620143575, "grad_norm": 1.3352244222077732, "learning_rate": 1.0595655503269933e-05, "loss": 0.5156, "step": 13939 }, { "epoch": 0.4966421433279299, "grad_norm": 1.2442083965122377, "learning_rate": 1.0594503611545528e-05, "loss": 0.4828, "step": 13940 }, { "epoch": 0.496677770454424, "grad_norm": 1.0702530199313458, "learning_rate": 1.0593351711904892e-05, "loss": 0.4954, "step": 13941 }, { "epoch": 0.4967133975809181, "grad_norm": 1.6348090802316273, "learning_rate": 1.0592199804363362e-05, "loss": 0.5979, "step": 13942 }, { "epoch": 0.4967490247074122, "grad_norm": 1.4290547360420978, "learning_rate": 1.0591047888936275e-05, "loss": 0.5225, "step": 13943 }, { "epoch": 0.49678465183390635, "grad_norm": 1.3029507613391107, "learning_rate": 1.0589895965638976e-05, "loss": 0.4764, "step": 13944 }, { "epoch": 0.49682027896040043, "grad_norm": 1.389273007687555, "learning_rate": 1.05887440344868e-05, "loss": 0.5716, "step": 13945 }, { "epoch": 0.49685590608689456, "grad_norm": 1.551157212941575, "learning_rate": 1.058759209549508e-05, "loss": 0.5436, "step": 13946 }, { "epoch": 0.4968915332133887, "grad_norm": 1.1261574633459035, "learning_rate": 1.0586440148679167e-05, "loss": 0.4847, "step": 13947 }, { "epoch": 0.49692716033988277, "grad_norm": 1.0996255330219586, "learning_rate": 1.058528819405439e-05, "loss": 0.4392, "step": 13948 }, { "epoch": 0.4969627874663769, "grad_norm": 1.6453942991518324, "learning_rate": 1.0584136231636094e-05, "loss": 0.5166, "step": 13949 }, { "epoch": 0.49699841459287103, "grad_norm": 1.1642369661585492, "learning_rate": 1.0582984261439616e-05, "loss": 0.4805, "step": 13950 }, { "epoch": 0.4970340417193651, "grad_norm": 1.4452067735834282, "learning_rate": 1.0581832283480291e-05, "loss": 0.4331, "step": 13951 }, { "epoch": 0.49706966884585924, "grad_norm": 1.8744993241802557, "learning_rate": 1.0580680297773467e-05, "loss": 0.6093, "step": 13952 }, { "epoch": 0.49710529597235337, "grad_norm": 1.2145376268932675, "learning_rate": 1.0579528304334477e-05, "loss": 0.4466, "step": 13953 }, { "epoch": 0.49714092309884744, "grad_norm": 1.46693171583307, "learning_rate": 1.0578376303178662e-05, "loss": 0.5928, "step": 13954 }, { "epoch": 0.4971765502253416, "grad_norm": 1.1327558847960997, "learning_rate": 1.0577224294321362e-05, "loss": 0.4487, "step": 13955 }, { "epoch": 0.4972121773518357, "grad_norm": 1.5299197175947876, "learning_rate": 1.057607227777792e-05, "loss": 0.5341, "step": 13956 }, { "epoch": 0.4972478044783298, "grad_norm": 1.6797437383461944, "learning_rate": 1.0574920253563672e-05, "loss": 0.5906, "step": 13957 }, { "epoch": 0.4972834316048239, "grad_norm": 1.2708468293860653, "learning_rate": 1.0573768221693956e-05, "loss": 0.4745, "step": 13958 }, { "epoch": 0.49731905873131804, "grad_norm": 1.1319891265192408, "learning_rate": 1.0572616182184119e-05, "loss": 0.4479, "step": 13959 }, { "epoch": 0.4973546858578121, "grad_norm": 1.5799125648021815, "learning_rate": 1.0571464135049493e-05, "loss": 0.6023, "step": 13960 }, { "epoch": 0.49739031298430625, "grad_norm": 1.327770428228308, "learning_rate": 1.0570312080305426e-05, "loss": 0.4726, "step": 13961 }, { "epoch": 0.4974259401108004, "grad_norm": 1.2164974939194353, "learning_rate": 1.0569160017967252e-05, "loss": 0.5555, "step": 13962 }, { "epoch": 0.49746156723729446, "grad_norm": 2.249664852694663, "learning_rate": 1.0568007948050319e-05, "loss": 0.5548, "step": 13963 }, { "epoch": 0.4974971943637886, "grad_norm": 1.0164245628045974, "learning_rate": 1.056685587056996e-05, "loss": 0.388, "step": 13964 }, { "epoch": 0.4975328214902827, "grad_norm": 1.2032251610267037, "learning_rate": 1.056570378554152e-05, "loss": 0.4673, "step": 13965 }, { "epoch": 0.4975684486167768, "grad_norm": 2.4899209675630725, "learning_rate": 1.0564551692980337e-05, "loss": 0.6084, "step": 13966 }, { "epoch": 0.4976040757432709, "grad_norm": 1.332304442509768, "learning_rate": 1.0563399592901755e-05, "loss": 0.4539, "step": 13967 }, { "epoch": 0.49763970286976505, "grad_norm": 2.3117975240058444, "learning_rate": 1.0562247485321116e-05, "loss": 0.5791, "step": 13968 }, { "epoch": 0.49767532999625913, "grad_norm": 1.2048553320505404, "learning_rate": 1.0561095370253754e-05, "loss": 0.5371, "step": 13969 }, { "epoch": 0.49771095712275326, "grad_norm": 1.1293753223899154, "learning_rate": 1.0559943247715019e-05, "loss": 0.4855, "step": 13970 }, { "epoch": 0.4977465842492474, "grad_norm": 1.3723255557738783, "learning_rate": 1.0558791117720244e-05, "loss": 0.4983, "step": 13971 }, { "epoch": 0.49778221137574147, "grad_norm": 1.1576430286846948, "learning_rate": 1.0557638980284778e-05, "loss": 0.3996, "step": 13972 }, { "epoch": 0.4978178385022356, "grad_norm": 1.5146960268254233, "learning_rate": 1.0556486835423957e-05, "loss": 0.6149, "step": 13973 }, { "epoch": 0.49785346562872973, "grad_norm": 1.190165871185029, "learning_rate": 1.0555334683153126e-05, "loss": 0.4634, "step": 13974 }, { "epoch": 0.4978890927552238, "grad_norm": 1.1509358822886213, "learning_rate": 1.0554182523487627e-05, "loss": 0.4502, "step": 13975 }, { "epoch": 0.49792471988171794, "grad_norm": 1.3155702988596163, "learning_rate": 1.0553030356442802e-05, "loss": 0.5037, "step": 13976 }, { "epoch": 0.49796034700821207, "grad_norm": 1.200013511267158, "learning_rate": 1.0551878182033985e-05, "loss": 0.399, "step": 13977 }, { "epoch": 0.49799597413470614, "grad_norm": 1.224907250739646, "learning_rate": 1.055072600027653e-05, "loss": 0.4408, "step": 13978 }, { "epoch": 0.4980316012612003, "grad_norm": 1.5755507066624688, "learning_rate": 1.0549573811185772e-05, "loss": 0.5712, "step": 13979 }, { "epoch": 0.4980672283876944, "grad_norm": 1.124916024110842, "learning_rate": 1.0548421614777051e-05, "loss": 0.4262, "step": 13980 }, { "epoch": 0.4981028555141885, "grad_norm": 1.1336077379487224, "learning_rate": 1.0547269411065718e-05, "loss": 0.4687, "step": 13981 }, { "epoch": 0.4981384826406826, "grad_norm": 1.0501171074994191, "learning_rate": 1.0546117200067107e-05, "loss": 0.4763, "step": 13982 }, { "epoch": 0.49817410976717674, "grad_norm": 1.015281232329822, "learning_rate": 1.0544964981796563e-05, "loss": 0.4248, "step": 13983 }, { "epoch": 0.4982097368936708, "grad_norm": 1.3257431427735968, "learning_rate": 1.054381275626943e-05, "loss": 0.5823, "step": 13984 }, { "epoch": 0.49824536402016495, "grad_norm": 1.6913914377731687, "learning_rate": 1.0542660523501051e-05, "loss": 0.4823, "step": 13985 }, { "epoch": 0.4982809911466591, "grad_norm": 1.6759421868594746, "learning_rate": 1.054150828350677e-05, "loss": 0.5082, "step": 13986 }, { "epoch": 0.49831661827315316, "grad_norm": 1.7143356704798247, "learning_rate": 1.0540356036301924e-05, "loss": 0.4797, "step": 13987 }, { "epoch": 0.4983522453996473, "grad_norm": 1.5991156149574839, "learning_rate": 1.053920378190186e-05, "loss": 0.6288, "step": 13988 }, { "epoch": 0.4983878725261414, "grad_norm": 1.5844035025526129, "learning_rate": 1.0538051520321924e-05, "loss": 0.5747, "step": 13989 }, { "epoch": 0.4984234996526355, "grad_norm": 1.5000925380367316, "learning_rate": 1.0536899251577455e-05, "loss": 0.5125, "step": 13990 }, { "epoch": 0.4984591267791296, "grad_norm": 1.4221778336967645, "learning_rate": 1.0535746975683793e-05, "loss": 0.6635, "step": 13991 }, { "epoch": 0.49849475390562376, "grad_norm": 1.7543233662098676, "learning_rate": 1.0534594692656291e-05, "loss": 0.5687, "step": 13992 }, { "epoch": 0.49853038103211783, "grad_norm": 1.505137429500144, "learning_rate": 1.0533442402510283e-05, "loss": 0.6214, "step": 13993 }, { "epoch": 0.49856600815861196, "grad_norm": 1.637195415240893, "learning_rate": 1.0532290105261118e-05, "loss": 0.5395, "step": 13994 }, { "epoch": 0.4986016352851061, "grad_norm": 1.6981375495292867, "learning_rate": 1.0531137800924136e-05, "loss": 0.5881, "step": 13995 }, { "epoch": 0.49863726241160017, "grad_norm": 1.6998445859546776, "learning_rate": 1.0529985489514687e-05, "loss": 0.5172, "step": 13996 }, { "epoch": 0.4986728895380943, "grad_norm": 1.0028455046316713, "learning_rate": 1.0528833171048109e-05, "loss": 0.4238, "step": 13997 }, { "epoch": 0.49870851666458843, "grad_norm": 1.055836247252831, "learning_rate": 1.052768084553975e-05, "loss": 0.5025, "step": 13998 }, { "epoch": 0.4987441437910825, "grad_norm": 1.292413330054927, "learning_rate": 1.0526528513004947e-05, "loss": 0.5964, "step": 13999 }, { "epoch": 0.49877977091757664, "grad_norm": 1.4350578323487397, "learning_rate": 1.0525376173459053e-05, "loss": 0.5238, "step": 14000 }, { "epoch": 0.49881539804407077, "grad_norm": 1.7978135462334213, "learning_rate": 1.0524223826917405e-05, "loss": 0.5546, "step": 14001 }, { "epoch": 0.49885102517056484, "grad_norm": 1.744457421093771, "learning_rate": 1.0523071473395352e-05, "loss": 0.5905, "step": 14002 }, { "epoch": 0.498886652297059, "grad_norm": 1.0166337636115774, "learning_rate": 1.0521919112908237e-05, "loss": 0.4602, "step": 14003 }, { "epoch": 0.4989222794235531, "grad_norm": 1.4616689374825462, "learning_rate": 1.0520766745471404e-05, "loss": 0.4993, "step": 14004 }, { "epoch": 0.4989579065500472, "grad_norm": 1.0555315264520795, "learning_rate": 1.0519614371100198e-05, "loss": 0.4649, "step": 14005 }, { "epoch": 0.4989935336765413, "grad_norm": 1.086420797038226, "learning_rate": 1.0518461989809962e-05, "loss": 0.4659, "step": 14006 }, { "epoch": 0.49902916080303544, "grad_norm": 1.6023281555179445, "learning_rate": 1.0517309601616042e-05, "loss": 0.5988, "step": 14007 }, { "epoch": 0.4990647879295295, "grad_norm": 1.0575221492814382, "learning_rate": 1.0516157206533783e-05, "loss": 0.4378, "step": 14008 }, { "epoch": 0.49910041505602365, "grad_norm": 1.6405032502344221, "learning_rate": 1.0515004804578532e-05, "loss": 0.5109, "step": 14009 }, { "epoch": 0.4991360421825178, "grad_norm": 1.3349728936178944, "learning_rate": 1.0513852395765631e-05, "loss": 0.5469, "step": 14010 }, { "epoch": 0.49917166930901186, "grad_norm": 2.127560910046293, "learning_rate": 1.0512699980110428e-05, "loss": 0.4887, "step": 14011 }, { "epoch": 0.499207296435506, "grad_norm": 1.1579123045940027, "learning_rate": 1.0511547557628265e-05, "loss": 0.4427, "step": 14012 }, { "epoch": 0.4992429235620001, "grad_norm": 1.0655682796051245, "learning_rate": 1.051039512833449e-05, "loss": 0.4684, "step": 14013 }, { "epoch": 0.4992785506884942, "grad_norm": 1.1520981039913945, "learning_rate": 1.0509242692244443e-05, "loss": 0.441, "step": 14014 }, { "epoch": 0.4993141778149883, "grad_norm": 1.148698048801644, "learning_rate": 1.0508090249373477e-05, "loss": 0.425, "step": 14015 }, { "epoch": 0.49934980494148246, "grad_norm": 1.5732956957003863, "learning_rate": 1.0506937799736935e-05, "loss": 0.5526, "step": 14016 }, { "epoch": 0.49938543206797653, "grad_norm": 1.007530378116374, "learning_rate": 1.050578534335016e-05, "loss": 0.4512, "step": 14017 }, { "epoch": 0.49942105919447066, "grad_norm": 2.0014323138998353, "learning_rate": 1.0504632880228497e-05, "loss": 0.4371, "step": 14018 }, { "epoch": 0.4994566863209648, "grad_norm": 1.7732108350967095, "learning_rate": 1.05034804103873e-05, "loss": 0.5884, "step": 14019 }, { "epoch": 0.49949231344745887, "grad_norm": 1.3935052534487176, "learning_rate": 1.0502327933841907e-05, "loss": 0.5396, "step": 14020 }, { "epoch": 0.499527940573953, "grad_norm": 1.0840162244077602, "learning_rate": 1.0501175450607664e-05, "loss": 0.4011, "step": 14021 }, { "epoch": 0.49956356770044713, "grad_norm": 1.7461095914474862, "learning_rate": 1.0500022960699923e-05, "loss": 0.5342, "step": 14022 }, { "epoch": 0.4995991948269412, "grad_norm": 1.1837198976633658, "learning_rate": 1.0498870464134026e-05, "loss": 0.476, "step": 14023 }, { "epoch": 0.49963482195343534, "grad_norm": 1.1195615831982437, "learning_rate": 1.0497717960925322e-05, "loss": 0.4223, "step": 14024 }, { "epoch": 0.49967044907992947, "grad_norm": 1.0426397096332174, "learning_rate": 1.049656545108915e-05, "loss": 0.4566, "step": 14025 }, { "epoch": 0.49970607620642354, "grad_norm": 1.226380049642346, "learning_rate": 1.0495412934640868e-05, "loss": 0.4777, "step": 14026 }, { "epoch": 0.4997417033329177, "grad_norm": 1.0991583347262466, "learning_rate": 1.0494260411595814e-05, "loss": 0.4564, "step": 14027 }, { "epoch": 0.4997773304594118, "grad_norm": 1.3783459377020584, "learning_rate": 1.0493107881969335e-05, "loss": 0.5768, "step": 14028 }, { "epoch": 0.4998129575859059, "grad_norm": 1.152583405325689, "learning_rate": 1.0491955345776782e-05, "loss": 0.467, "step": 14029 }, { "epoch": 0.4998485847124, "grad_norm": 1.1699324817440047, "learning_rate": 1.0490802803033497e-05, "loss": 0.4797, "step": 14030 }, { "epoch": 0.49988421183889414, "grad_norm": 1.8448281561170894, "learning_rate": 1.0489650253754836e-05, "loss": 0.5498, "step": 14031 }, { "epoch": 0.4999198389653882, "grad_norm": 1.2980814140947443, "learning_rate": 1.0488497697956134e-05, "loss": 0.4744, "step": 14032 }, { "epoch": 0.49995546609188235, "grad_norm": 1.3034956218016134, "learning_rate": 1.0487345135652745e-05, "loss": 0.5253, "step": 14033 }, { "epoch": 0.4999910932183765, "grad_norm": 1.2009245625775662, "learning_rate": 1.0486192566860016e-05, "loss": 0.5919, "step": 14034 }, { "epoch": 0.5000267203448706, "grad_norm": 1.0539776866482502, "learning_rate": 1.0485039991593293e-05, "loss": 0.4945, "step": 14035 }, { "epoch": 0.5000623474713647, "grad_norm": 1.1861509056039239, "learning_rate": 1.0483887409867921e-05, "loss": 0.5071, "step": 14036 }, { "epoch": 0.5000979745978588, "grad_norm": 1.0427180618176715, "learning_rate": 1.0482734821699254e-05, "loss": 0.4739, "step": 14037 }, { "epoch": 0.500133601724353, "grad_norm": 1.3471044145484172, "learning_rate": 1.0481582227102634e-05, "loss": 0.5134, "step": 14038 }, { "epoch": 0.500169228850847, "grad_norm": 1.1487944118263425, "learning_rate": 1.0480429626093408e-05, "loss": 0.4272, "step": 14039 }, { "epoch": 0.5002048559773411, "grad_norm": 1.1065675854588752, "learning_rate": 1.0479277018686926e-05, "loss": 0.456, "step": 14040 }, { "epoch": 0.5002404831038353, "grad_norm": 1.3958632544993952, "learning_rate": 1.0478124404898535e-05, "loss": 0.5135, "step": 14041 }, { "epoch": 0.5002761102303294, "grad_norm": 2.0562876001420842, "learning_rate": 1.0476971784743586e-05, "loss": 0.4969, "step": 14042 }, { "epoch": 0.5003117373568234, "grad_norm": 1.6194924387852714, "learning_rate": 1.0475819158237426e-05, "loss": 0.6195, "step": 14043 }, { "epoch": 0.5003473644833176, "grad_norm": 1.0918001333144958, "learning_rate": 1.0474666525395396e-05, "loss": 0.457, "step": 14044 }, { "epoch": 0.5003829916098117, "grad_norm": 1.3740984634974995, "learning_rate": 1.0473513886232854e-05, "loss": 0.5735, "step": 14045 }, { "epoch": 0.5004186187363058, "grad_norm": 1.251548790402261, "learning_rate": 1.0472361240765141e-05, "loss": 0.482, "step": 14046 }, { "epoch": 0.5004542458628, "grad_norm": 1.476063698987805, "learning_rate": 1.0471208589007607e-05, "loss": 0.4344, "step": 14047 }, { "epoch": 0.500489872989294, "grad_norm": 1.1549662475285867, "learning_rate": 1.0470055930975605e-05, "loss": 0.4955, "step": 14048 }, { "epoch": 0.5005255001157881, "grad_norm": 1.0661226127792918, "learning_rate": 1.0468903266684478e-05, "loss": 0.4745, "step": 14049 }, { "epoch": 0.5005611272422823, "grad_norm": 1.419181737006742, "learning_rate": 1.0467750596149576e-05, "loss": 0.5795, "step": 14050 }, { "epoch": 0.5005967543687764, "grad_norm": 1.026443721059554, "learning_rate": 1.0466597919386245e-05, "loss": 0.4056, "step": 14051 }, { "epoch": 0.5006323814952705, "grad_norm": 1.0656939915893056, "learning_rate": 1.046544523640984e-05, "loss": 0.4363, "step": 14052 }, { "epoch": 0.5006680086217646, "grad_norm": 1.7416040135116753, "learning_rate": 1.0464292547235706e-05, "loss": 0.4472, "step": 14053 }, { "epoch": 0.5007036357482587, "grad_norm": 1.5496376413896384, "learning_rate": 1.0463139851879192e-05, "loss": 0.577, "step": 14054 }, { "epoch": 0.5007392628747528, "grad_norm": 1.1772962076504268, "learning_rate": 1.0461987150355646e-05, "loss": 0.4729, "step": 14055 }, { "epoch": 0.500774890001247, "grad_norm": 1.3871628728342273, "learning_rate": 1.0460834442680419e-05, "loss": 0.5826, "step": 14056 }, { "epoch": 0.500810517127741, "grad_norm": 1.1162049192263916, "learning_rate": 1.045968172886886e-05, "loss": 0.4758, "step": 14057 }, { "epoch": 0.5008461442542351, "grad_norm": 1.8631711878538726, "learning_rate": 1.0458529008936317e-05, "loss": 0.4255, "step": 14058 }, { "epoch": 0.5008817713807293, "grad_norm": 1.396680917204875, "learning_rate": 1.0457376282898139e-05, "loss": 0.4389, "step": 14059 }, { "epoch": 0.5009173985072234, "grad_norm": 1.8165984358950085, "learning_rate": 1.0456223550769677e-05, "loss": 0.4938, "step": 14060 }, { "epoch": 0.5009530256337175, "grad_norm": 1.9100171393103698, "learning_rate": 1.0455070812566281e-05, "loss": 0.525, "step": 14061 }, { "epoch": 0.5009886527602117, "grad_norm": 1.0614491680892766, "learning_rate": 1.0453918068303298e-05, "loss": 0.457, "step": 14062 }, { "epoch": 0.5010242798867057, "grad_norm": 1.6245109573855434, "learning_rate": 1.0452765317996076e-05, "loss": 0.5426, "step": 14063 }, { "epoch": 0.5010599070131998, "grad_norm": 1.5421964127585763, "learning_rate": 1.045161256165997e-05, "loss": 0.5861, "step": 14064 }, { "epoch": 0.501095534139694, "grad_norm": 1.2342115734123482, "learning_rate": 1.0450459799310325e-05, "loss": 0.5566, "step": 14065 }, { "epoch": 0.5011311612661881, "grad_norm": 1.042516764529154, "learning_rate": 1.0449307030962494e-05, "loss": 0.4514, "step": 14066 }, { "epoch": 0.5011667883926821, "grad_norm": 1.3805987114327298, "learning_rate": 1.0448154256631828e-05, "loss": 0.611, "step": 14067 }, { "epoch": 0.5012024155191763, "grad_norm": 1.2131531769031716, "learning_rate": 1.0447001476333673e-05, "loss": 0.4845, "step": 14068 }, { "epoch": 0.5012380426456704, "grad_norm": 1.238917935806149, "learning_rate": 1.0445848690083382e-05, "loss": 0.4829, "step": 14069 }, { "epoch": 0.5012736697721645, "grad_norm": 2.050309110091824, "learning_rate": 1.0444695897896301e-05, "loss": 0.5649, "step": 14070 }, { "epoch": 0.5013092968986587, "grad_norm": 1.5102603770581615, "learning_rate": 1.0443543099787788e-05, "loss": 0.4277, "step": 14071 }, { "epoch": 0.5013449240251527, "grad_norm": 1.8227919195566393, "learning_rate": 1.0442390295773186e-05, "loss": 0.4829, "step": 14072 }, { "epoch": 0.5013805511516468, "grad_norm": 1.1405279566828246, "learning_rate": 1.0441237485867845e-05, "loss": 0.4771, "step": 14073 }, { "epoch": 0.501416178278141, "grad_norm": 2.2169162823669284, "learning_rate": 1.0440084670087122e-05, "loss": 0.5243, "step": 14074 }, { "epoch": 0.5014518054046351, "grad_norm": 1.3061102803651765, "learning_rate": 1.0438931848446363e-05, "loss": 0.6165, "step": 14075 }, { "epoch": 0.5014874325311292, "grad_norm": 1.4811060019384243, "learning_rate": 1.043777902096092e-05, "loss": 0.5765, "step": 14076 }, { "epoch": 0.5015230596576233, "grad_norm": 1.0885728780597852, "learning_rate": 1.0436626187646143e-05, "loss": 0.4726, "step": 14077 }, { "epoch": 0.5015586867841174, "grad_norm": 1.9315006573833828, "learning_rate": 1.0435473348517384e-05, "loss": 0.5459, "step": 14078 }, { "epoch": 0.5015943139106115, "grad_norm": 1.124734197341333, "learning_rate": 1.0434320503589994e-05, "loss": 0.483, "step": 14079 }, { "epoch": 0.5016299410371057, "grad_norm": 1.0756327699391275, "learning_rate": 1.0433167652879321e-05, "loss": 0.3846, "step": 14080 }, { "epoch": 0.5016655681635998, "grad_norm": 1.404051029354874, "learning_rate": 1.0432014796400719e-05, "loss": 0.4988, "step": 14081 }, { "epoch": 0.5017011952900938, "grad_norm": 1.7609445658887586, "learning_rate": 1.0430861934169536e-05, "loss": 0.542, "step": 14082 }, { "epoch": 0.501736822416588, "grad_norm": 1.6350464776990687, "learning_rate": 1.0429709066201128e-05, "loss": 0.5428, "step": 14083 }, { "epoch": 0.5017724495430821, "grad_norm": 1.3407860347076872, "learning_rate": 1.0428556192510841e-05, "loss": 0.5379, "step": 14084 }, { "epoch": 0.5018080766695762, "grad_norm": 1.368766973454077, "learning_rate": 1.0427403313114031e-05, "loss": 0.4481, "step": 14085 }, { "epoch": 0.5018437037960704, "grad_norm": 1.1268615452784578, "learning_rate": 1.0426250428026045e-05, "loss": 0.4435, "step": 14086 }, { "epoch": 0.5018793309225644, "grad_norm": 1.8417523943669432, "learning_rate": 1.0425097537262236e-05, "loss": 0.4769, "step": 14087 }, { "epoch": 0.5019149580490585, "grad_norm": 1.639943738481801, "learning_rate": 1.0423944640837959e-05, "loss": 0.5831, "step": 14088 }, { "epoch": 0.5019505851755527, "grad_norm": 1.6268538989328012, "learning_rate": 1.042279173876856e-05, "loss": 0.6321, "step": 14089 }, { "epoch": 0.5019862123020468, "grad_norm": 1.2159082265112477, "learning_rate": 1.0421638831069396e-05, "loss": 0.4355, "step": 14090 }, { "epoch": 0.5020218394285408, "grad_norm": 1.173212966040103, "learning_rate": 1.0420485917755816e-05, "loss": 0.4861, "step": 14091 }, { "epoch": 0.502057466555035, "grad_norm": 1.01532309041312, "learning_rate": 1.041933299884317e-05, "loss": 0.4454, "step": 14092 }, { "epoch": 0.5020930936815291, "grad_norm": 1.043908737495004, "learning_rate": 1.0418180074346816e-05, "loss": 0.4047, "step": 14093 }, { "epoch": 0.5021287208080232, "grad_norm": 1.0897927865327661, "learning_rate": 1.0417027144282099e-05, "loss": 0.4414, "step": 14094 }, { "epoch": 0.5021643479345174, "grad_norm": 1.4111038398806048, "learning_rate": 1.0415874208664376e-05, "loss": 0.5446, "step": 14095 }, { "epoch": 0.5021999750610114, "grad_norm": 1.489081309219843, "learning_rate": 1.0414721267508995e-05, "loss": 0.6182, "step": 14096 }, { "epoch": 0.5022356021875055, "grad_norm": 1.0956732695440865, "learning_rate": 1.041356832083131e-05, "loss": 0.4347, "step": 14097 }, { "epoch": 0.5022712293139997, "grad_norm": 1.6647118174779334, "learning_rate": 1.0412415368646674e-05, "loss": 0.4942, "step": 14098 }, { "epoch": 0.5023068564404938, "grad_norm": 1.4724356674557981, "learning_rate": 1.041126241097044e-05, "loss": 0.539, "step": 14099 }, { "epoch": 0.5023424835669879, "grad_norm": 1.222808623966282, "learning_rate": 1.0410109447817959e-05, "loss": 0.458, "step": 14100 }, { "epoch": 0.502378110693482, "grad_norm": 1.1589558119633003, "learning_rate": 1.0408956479204586e-05, "loss": 0.4811, "step": 14101 }, { "epoch": 0.5024137378199761, "grad_norm": 1.5848131509543408, "learning_rate": 1.0407803505145671e-05, "loss": 0.5441, "step": 14102 }, { "epoch": 0.5024493649464702, "grad_norm": 1.4990323265828496, "learning_rate": 1.0406650525656565e-05, "loss": 0.6149, "step": 14103 }, { "epoch": 0.5024849920729644, "grad_norm": 1.5888012751226592, "learning_rate": 1.0405497540752624e-05, "loss": 0.6186, "step": 14104 }, { "epoch": 0.5025206191994585, "grad_norm": 1.849633276063562, "learning_rate": 1.04043445504492e-05, "loss": 0.5738, "step": 14105 }, { "epoch": 0.5025562463259525, "grad_norm": 1.2003138353699727, "learning_rate": 1.040319155476165e-05, "loss": 0.4577, "step": 14106 }, { "epoch": 0.5025918734524467, "grad_norm": 1.106588696379341, "learning_rate": 1.0402038553705316e-05, "loss": 0.5387, "step": 14107 }, { "epoch": 0.5026275005789408, "grad_norm": 1.0820491277397308, "learning_rate": 1.0400885547295561e-05, "loss": 0.431, "step": 14108 }, { "epoch": 0.5026631277054349, "grad_norm": 1.5334032475822028, "learning_rate": 1.0399732535547735e-05, "loss": 0.5523, "step": 14109 }, { "epoch": 0.502698754831929, "grad_norm": 1.0757893642906846, "learning_rate": 1.0398579518477188e-05, "loss": 0.4541, "step": 14110 }, { "epoch": 0.5027343819584231, "grad_norm": 1.0307826519152992, "learning_rate": 1.0397426496099277e-05, "loss": 0.4382, "step": 14111 }, { "epoch": 0.5027700090849172, "grad_norm": 1.7711612498282463, "learning_rate": 1.0396273468429356e-05, "loss": 0.5671, "step": 14112 }, { "epoch": 0.5028056362114114, "grad_norm": 1.1892729208627895, "learning_rate": 1.0395120435482777e-05, "loss": 0.4294, "step": 14113 }, { "epoch": 0.5028412633379055, "grad_norm": 1.2849689359263161, "learning_rate": 1.0393967397274894e-05, "loss": 0.4209, "step": 14114 }, { "epoch": 0.5028768904643995, "grad_norm": 1.2514234809729663, "learning_rate": 1.0392814353821058e-05, "loss": 0.5129, "step": 14115 }, { "epoch": 0.5029125175908937, "grad_norm": 1.1587311241060747, "learning_rate": 1.0391661305136625e-05, "loss": 0.4444, "step": 14116 }, { "epoch": 0.5029481447173878, "grad_norm": 3.3071946350417676, "learning_rate": 1.0390508251236948e-05, "loss": 0.5707, "step": 14117 }, { "epoch": 0.5029837718438819, "grad_norm": 0.9731685821188752, "learning_rate": 1.0389355192137379e-05, "loss": 0.4315, "step": 14118 }, { "epoch": 0.5030193989703761, "grad_norm": 1.057659798238844, "learning_rate": 1.0388202127853274e-05, "loss": 0.4701, "step": 14119 }, { "epoch": 0.5030550260968701, "grad_norm": 1.0723282916280588, "learning_rate": 1.0387049058399989e-05, "loss": 0.4575, "step": 14120 }, { "epoch": 0.5030906532233642, "grad_norm": 2.059223014840328, "learning_rate": 1.038589598379287e-05, "loss": 0.6427, "step": 14121 }, { "epoch": 0.5031262803498584, "grad_norm": 1.4086961832604772, "learning_rate": 1.0384742904047281e-05, "loss": 0.5953, "step": 14122 }, { "epoch": 0.5031619074763525, "grad_norm": 1.2226055787704833, "learning_rate": 1.038358981917857e-05, "loss": 0.4071, "step": 14123 }, { "epoch": 0.5031975346028466, "grad_norm": 1.6999971728613363, "learning_rate": 1.0382436729202092e-05, "loss": 0.5083, "step": 14124 }, { "epoch": 0.5032331617293407, "grad_norm": 1.222944738587359, "learning_rate": 1.0381283634133203e-05, "loss": 0.4476, "step": 14125 }, { "epoch": 0.5032687888558348, "grad_norm": 1.1436894792790726, "learning_rate": 1.0380130533987254e-05, "loss": 0.4724, "step": 14126 }, { "epoch": 0.5033044159823289, "grad_norm": 1.5478223502417863, "learning_rate": 1.0378977428779601e-05, "loss": 0.5933, "step": 14127 }, { "epoch": 0.5033400431088231, "grad_norm": 2.0541449972523487, "learning_rate": 1.0377824318525603e-05, "loss": 0.5003, "step": 14128 }, { "epoch": 0.5033756702353172, "grad_norm": 1.4490293756655135, "learning_rate": 1.0376671203240605e-05, "loss": 0.4226, "step": 14129 }, { "epoch": 0.5034112973618113, "grad_norm": 1.9571654232410063, "learning_rate": 1.0375518082939969e-05, "loss": 0.552, "step": 14130 }, { "epoch": 0.5034469244883054, "grad_norm": 1.0522503934847387, "learning_rate": 1.0374364957639047e-05, "loss": 0.4173, "step": 14131 }, { "epoch": 0.5034825516147995, "grad_norm": 1.1209207487103368, "learning_rate": 1.0373211827353195e-05, "loss": 0.4535, "step": 14132 }, { "epoch": 0.5035181787412937, "grad_norm": 1.0889114960669184, "learning_rate": 1.0372058692097764e-05, "loss": 0.4428, "step": 14133 }, { "epoch": 0.5035538058677878, "grad_norm": 1.221103173986069, "learning_rate": 1.0370905551888111e-05, "loss": 0.4723, "step": 14134 }, { "epoch": 0.5035894329942818, "grad_norm": 1.1834227600311453, "learning_rate": 1.0369752406739594e-05, "loss": 0.4544, "step": 14135 }, { "epoch": 0.503625060120776, "grad_norm": 1.4780860736509316, "learning_rate": 1.0368599256667564e-05, "loss": 0.5843, "step": 14136 }, { "epoch": 0.5036606872472701, "grad_norm": 1.466316474630276, "learning_rate": 1.0367446101687377e-05, "loss": 0.662, "step": 14137 }, { "epoch": 0.5036963143737642, "grad_norm": 2.1920682731385424, "learning_rate": 1.0366292941814392e-05, "loss": 0.6208, "step": 14138 }, { "epoch": 0.5037319415002584, "grad_norm": 1.5863468398181744, "learning_rate": 1.0365139777063957e-05, "loss": 0.5538, "step": 14139 }, { "epoch": 0.5037675686267524, "grad_norm": 1.8939536963920198, "learning_rate": 1.0363986607451431e-05, "loss": 0.5728, "step": 14140 }, { "epoch": 0.5038031957532465, "grad_norm": 1.3931047925019535, "learning_rate": 1.0362833432992167e-05, "loss": 0.4895, "step": 14141 }, { "epoch": 0.5038388228797407, "grad_norm": 1.6199015332874134, "learning_rate": 1.0361680253701528e-05, "loss": 0.5499, "step": 14142 }, { "epoch": 0.5038744500062348, "grad_norm": 1.2055772261976596, "learning_rate": 1.0360527069594859e-05, "loss": 0.4057, "step": 14143 }, { "epoch": 0.5039100771327288, "grad_norm": 1.4188379161654212, "learning_rate": 1.035937388068752e-05, "loss": 0.4648, "step": 14144 }, { "epoch": 0.503945704259223, "grad_norm": 2.3048606796004067, "learning_rate": 1.0358220686994864e-05, "loss": 0.554, "step": 14145 }, { "epoch": 0.5039813313857171, "grad_norm": 1.4488779474389364, "learning_rate": 1.0357067488532255e-05, "loss": 0.541, "step": 14146 }, { "epoch": 0.5040169585122112, "grad_norm": 1.0497090833121985, "learning_rate": 1.035591428531504e-05, "loss": 0.4487, "step": 14147 }, { "epoch": 0.5040525856387054, "grad_norm": 1.1547935370346225, "learning_rate": 1.0354761077358577e-05, "loss": 0.4556, "step": 14148 }, { "epoch": 0.5040882127651994, "grad_norm": 1.9194029104291852, "learning_rate": 1.0353607864678225e-05, "loss": 0.5708, "step": 14149 }, { "epoch": 0.5041238398916935, "grad_norm": 1.6511040753965411, "learning_rate": 1.0352454647289336e-05, "loss": 0.6035, "step": 14150 }, { "epoch": 0.5041594670181877, "grad_norm": 1.237826755782823, "learning_rate": 1.0351301425207268e-05, "loss": 0.4352, "step": 14151 }, { "epoch": 0.5041950941446818, "grad_norm": 1.6660902806627296, "learning_rate": 1.0350148198447372e-05, "loss": 0.5525, "step": 14152 }, { "epoch": 0.5042307212711759, "grad_norm": 1.8800640782236266, "learning_rate": 1.0348994967025012e-05, "loss": 0.5288, "step": 14153 }, { "epoch": 0.50426634839767, "grad_norm": 1.9461133097826089, "learning_rate": 1.0347841730955539e-05, "loss": 0.5856, "step": 14154 }, { "epoch": 0.5043019755241641, "grad_norm": 1.079095176625551, "learning_rate": 1.0346688490254309e-05, "loss": 0.4284, "step": 14155 }, { "epoch": 0.5043376026506582, "grad_norm": 1.360548126545758, "learning_rate": 1.0345535244936677e-05, "loss": 0.5, "step": 14156 }, { "epoch": 0.5043732297771524, "grad_norm": 1.3398462414439127, "learning_rate": 1.0344381995018008e-05, "loss": 0.4822, "step": 14157 }, { "epoch": 0.5044088569036465, "grad_norm": 1.0801157536905852, "learning_rate": 1.034322874051365e-05, "loss": 0.4619, "step": 14158 }, { "epoch": 0.5044444840301405, "grad_norm": 1.1833306757655642, "learning_rate": 1.0342075481438959e-05, "loss": 0.4762, "step": 14159 }, { "epoch": 0.5044801111566347, "grad_norm": 2.349675936681391, "learning_rate": 1.0340922217809294e-05, "loss": 0.4993, "step": 14160 }, { "epoch": 0.5045157382831288, "grad_norm": 1.9275931269808342, "learning_rate": 1.0339768949640014e-05, "loss": 0.5079, "step": 14161 }, { "epoch": 0.5045513654096229, "grad_norm": 1.1833749648294287, "learning_rate": 1.0338615676946472e-05, "loss": 0.4342, "step": 14162 }, { "epoch": 0.504586992536117, "grad_norm": 1.604315303428782, "learning_rate": 1.0337462399744025e-05, "loss": 0.4993, "step": 14163 }, { "epoch": 0.5046226196626111, "grad_norm": 1.1777253162832946, "learning_rate": 1.0336309118048028e-05, "loss": 0.4514, "step": 14164 }, { "epoch": 0.5046582467891052, "grad_norm": 1.1074091228365985, "learning_rate": 1.0335155831873845e-05, "loss": 0.4635, "step": 14165 }, { "epoch": 0.5046938739155994, "grad_norm": 1.6091548242609437, "learning_rate": 1.0334002541236825e-05, "loss": 0.5704, "step": 14166 }, { "epoch": 0.5047295010420935, "grad_norm": 1.1945630777706002, "learning_rate": 1.0332849246152326e-05, "loss": 0.4739, "step": 14167 }, { "epoch": 0.5047651281685875, "grad_norm": 1.125761686543616, "learning_rate": 1.0331695946635708e-05, "loss": 0.4844, "step": 14168 }, { "epoch": 0.5048007552950817, "grad_norm": 1.2213324420709701, "learning_rate": 1.0330542642702327e-05, "loss": 0.4744, "step": 14169 }, { "epoch": 0.5048363824215758, "grad_norm": 1.1323032362907544, "learning_rate": 1.0329389334367543e-05, "loss": 0.4671, "step": 14170 }, { "epoch": 0.5048720095480699, "grad_norm": 1.579997047331368, "learning_rate": 1.0328236021646705e-05, "loss": 0.5798, "step": 14171 }, { "epoch": 0.5049076366745641, "grad_norm": 1.0879654394693932, "learning_rate": 1.0327082704555177e-05, "loss": 0.4459, "step": 14172 }, { "epoch": 0.5049432638010581, "grad_norm": 1.1285819936983978, "learning_rate": 1.0325929383108314e-05, "loss": 0.4781, "step": 14173 }, { "epoch": 0.5049788909275522, "grad_norm": 1.2310154763170877, "learning_rate": 1.0324776057321473e-05, "loss": 0.4741, "step": 14174 }, { "epoch": 0.5050145180540464, "grad_norm": 1.0551916376564756, "learning_rate": 1.0323622727210012e-05, "loss": 0.3945, "step": 14175 }, { "epoch": 0.5050501451805405, "grad_norm": 1.3210276235614018, "learning_rate": 1.032246939278929e-05, "loss": 0.5876, "step": 14176 }, { "epoch": 0.5050857723070346, "grad_norm": 1.5548532375877149, "learning_rate": 1.0321316054074662e-05, "loss": 0.4776, "step": 14177 }, { "epoch": 0.5051213994335287, "grad_norm": 1.182769537246931, "learning_rate": 1.0320162711081484e-05, "loss": 0.4468, "step": 14178 }, { "epoch": 0.5051570265600228, "grad_norm": 1.6335687082864168, "learning_rate": 1.0319009363825115e-05, "loss": 0.4436, "step": 14179 }, { "epoch": 0.5051926536865169, "grad_norm": 1.5642774457149873, "learning_rate": 1.0317856012320915e-05, "loss": 0.5354, "step": 14180 }, { "epoch": 0.5052282808130111, "grad_norm": 1.4413204857676785, "learning_rate": 1.0316702656584241e-05, "loss": 0.5544, "step": 14181 }, { "epoch": 0.5052639079395052, "grad_norm": 1.0819790391930257, "learning_rate": 1.0315549296630447e-05, "loss": 0.4859, "step": 14182 }, { "epoch": 0.5052995350659992, "grad_norm": 1.3023858227821368, "learning_rate": 1.0314395932474896e-05, "loss": 0.4529, "step": 14183 }, { "epoch": 0.5053351621924934, "grad_norm": 1.5218865186402581, "learning_rate": 1.0313242564132943e-05, "loss": 0.5849, "step": 14184 }, { "epoch": 0.5053707893189875, "grad_norm": 1.732014889122788, "learning_rate": 1.0312089191619948e-05, "loss": 0.4711, "step": 14185 }, { "epoch": 0.5054064164454816, "grad_norm": 1.0810741921899778, "learning_rate": 1.0310935814951263e-05, "loss": 0.4287, "step": 14186 }, { "epoch": 0.5054420435719758, "grad_norm": 1.209802938122017, "learning_rate": 1.0309782434142254e-05, "loss": 0.4874, "step": 14187 }, { "epoch": 0.5054776706984698, "grad_norm": 1.2170899112312912, "learning_rate": 1.0308629049208277e-05, "loss": 0.4681, "step": 14188 }, { "epoch": 0.5055132978249639, "grad_norm": 1.6532543617547, "learning_rate": 1.0307475660164683e-05, "loss": 0.5447, "step": 14189 }, { "epoch": 0.5055489249514581, "grad_norm": 1.1330541414287307, "learning_rate": 1.030632226702684e-05, "loss": 0.4279, "step": 14190 }, { "epoch": 0.5055845520779522, "grad_norm": 1.2064538002992649, "learning_rate": 1.0305168869810099e-05, "loss": 0.5037, "step": 14191 }, { "epoch": 0.5056201792044462, "grad_norm": 1.5890892162037407, "learning_rate": 1.0304015468529821e-05, "loss": 0.5624, "step": 14192 }, { "epoch": 0.5056558063309404, "grad_norm": 1.2462165167912662, "learning_rate": 1.0302862063201367e-05, "loss": 0.4864, "step": 14193 }, { "epoch": 0.5056914334574345, "grad_norm": 1.6211457177357693, "learning_rate": 1.0301708653840092e-05, "loss": 0.43, "step": 14194 }, { "epoch": 0.5057270605839286, "grad_norm": 1.1307054592215544, "learning_rate": 1.0300555240461356e-05, "loss": 0.4328, "step": 14195 }, { "epoch": 0.5057626877104228, "grad_norm": 1.8567467720612703, "learning_rate": 1.0299401823080517e-05, "loss": 0.6004, "step": 14196 }, { "epoch": 0.5057983148369168, "grad_norm": 3.130962806592722, "learning_rate": 1.0298248401712931e-05, "loss": 0.6089, "step": 14197 }, { "epoch": 0.5058339419634109, "grad_norm": 1.0837203637889812, "learning_rate": 1.0297094976373962e-05, "loss": 0.4274, "step": 14198 }, { "epoch": 0.5058695690899051, "grad_norm": 1.1580523055063432, "learning_rate": 1.0295941547078965e-05, "loss": 0.5036, "step": 14199 }, { "epoch": 0.5059051962163992, "grad_norm": 1.0785498785642906, "learning_rate": 1.0294788113843297e-05, "loss": 0.4415, "step": 14200 }, { "epoch": 0.5059408233428933, "grad_norm": 1.9558670950020884, "learning_rate": 1.0293634676682323e-05, "loss": 0.5943, "step": 14201 }, { "epoch": 0.5059764504693874, "grad_norm": 4.737795714424229, "learning_rate": 1.0292481235611394e-05, "loss": 0.5207, "step": 14202 }, { "epoch": 0.5060120775958815, "grad_norm": 1.5274119123798369, "learning_rate": 1.0291327790645876e-05, "loss": 0.4467, "step": 14203 }, { "epoch": 0.5060477047223756, "grad_norm": 1.6405921818961193, "learning_rate": 1.0290174341801122e-05, "loss": 0.5138, "step": 14204 }, { "epoch": 0.5060833318488698, "grad_norm": 1.1025518417488287, "learning_rate": 1.0289020889092496e-05, "loss": 0.4519, "step": 14205 }, { "epoch": 0.5061189589753639, "grad_norm": 1.3269931780195876, "learning_rate": 1.0287867432535357e-05, "loss": 0.5906, "step": 14206 }, { "epoch": 0.5061545861018579, "grad_norm": 1.9783830112914853, "learning_rate": 1.028671397214506e-05, "loss": 0.5599, "step": 14207 }, { "epoch": 0.5061902132283521, "grad_norm": 1.1981357727008157, "learning_rate": 1.0285560507936962e-05, "loss": 0.4221, "step": 14208 }, { "epoch": 0.5062258403548462, "grad_norm": 2.5627752469569955, "learning_rate": 1.0284407039926431e-05, "loss": 0.5737, "step": 14209 }, { "epoch": 0.5062614674813403, "grad_norm": 1.018097414840557, "learning_rate": 1.0283253568128821e-05, "loss": 0.4276, "step": 14210 }, { "epoch": 0.5062970946078345, "grad_norm": 1.1758327377407094, "learning_rate": 1.028210009255949e-05, "loss": 0.496, "step": 14211 }, { "epoch": 0.5063327217343285, "grad_norm": 1.5409404216956901, "learning_rate": 1.0280946613233803e-05, "loss": 0.5439, "step": 14212 }, { "epoch": 0.5063683488608226, "grad_norm": 1.2641024055339112, "learning_rate": 1.0279793130167112e-05, "loss": 0.4824, "step": 14213 }, { "epoch": 0.5064039759873168, "grad_norm": 1.3771090285095127, "learning_rate": 1.027863964337478e-05, "loss": 0.5533, "step": 14214 }, { "epoch": 0.5064396031138109, "grad_norm": 1.1781506450306214, "learning_rate": 1.0277486152872167e-05, "loss": 0.4425, "step": 14215 }, { "epoch": 0.5064752302403049, "grad_norm": 1.0455027852868053, "learning_rate": 1.027633265867463e-05, "loss": 0.4329, "step": 14216 }, { "epoch": 0.5065108573667991, "grad_norm": 1.5763974065566473, "learning_rate": 1.0275179160797533e-05, "loss": 0.4456, "step": 14217 }, { "epoch": 0.5065464844932932, "grad_norm": 1.9980741405869862, "learning_rate": 1.0274025659256233e-05, "loss": 0.5456, "step": 14218 }, { "epoch": 0.5065821116197873, "grad_norm": 1.1288042219865395, "learning_rate": 1.0272872154066089e-05, "loss": 0.4618, "step": 14219 }, { "epoch": 0.5066177387462815, "grad_norm": 1.6131610784328831, "learning_rate": 1.0271718645242462e-05, "loss": 0.6002, "step": 14220 }, { "epoch": 0.5066533658727755, "grad_norm": 1.291460820966202, "learning_rate": 1.0270565132800711e-05, "loss": 0.4841, "step": 14221 }, { "epoch": 0.5066889929992696, "grad_norm": 1.4738720419704374, "learning_rate": 1.0269411616756197e-05, "loss": 0.5543, "step": 14222 }, { "epoch": 0.5067246201257638, "grad_norm": 2.535507149170627, "learning_rate": 1.0268258097124277e-05, "loss": 0.5775, "step": 14223 }, { "epoch": 0.5067602472522579, "grad_norm": 1.007244586668632, "learning_rate": 1.0267104573920317e-05, "loss": 0.4331, "step": 14224 }, { "epoch": 0.506795874378752, "grad_norm": 1.2021580221395816, "learning_rate": 1.0265951047159668e-05, "loss": 0.4885, "step": 14225 }, { "epoch": 0.5068315015052461, "grad_norm": 2.0978628675013185, "learning_rate": 1.0264797516857696e-05, "loss": 0.5296, "step": 14226 }, { "epoch": 0.5068671286317402, "grad_norm": 1.2270307604986224, "learning_rate": 1.026364398302976e-05, "loss": 0.4493, "step": 14227 }, { "epoch": 0.5069027557582343, "grad_norm": 1.4867645765361204, "learning_rate": 1.0262490445691222e-05, "loss": 0.5916, "step": 14228 }, { "epoch": 0.5069383828847285, "grad_norm": 1.6242562475416171, "learning_rate": 1.0261336904857442e-05, "loss": 0.5168, "step": 14229 }, { "epoch": 0.5069740100112226, "grad_norm": 1.1538674206182071, "learning_rate": 1.0260183360543774e-05, "loss": 0.4754, "step": 14230 }, { "epoch": 0.5070096371377166, "grad_norm": 1.5993937539000838, "learning_rate": 1.0259029812765584e-05, "loss": 0.5123, "step": 14231 }, { "epoch": 0.5070452642642108, "grad_norm": 2.1934153468523405, "learning_rate": 1.0257876261538233e-05, "loss": 0.5436, "step": 14232 }, { "epoch": 0.5070808913907049, "grad_norm": 1.550260212740076, "learning_rate": 1.0256722706877078e-05, "loss": 0.591, "step": 14233 }, { "epoch": 0.507116518517199, "grad_norm": 1.8986222158751012, "learning_rate": 1.0255569148797479e-05, "loss": 0.5877, "step": 14234 }, { "epoch": 0.5071521456436932, "grad_norm": 1.1354185085002877, "learning_rate": 1.0254415587314801e-05, "loss": 0.5009, "step": 14235 }, { "epoch": 0.5071877727701872, "grad_norm": 1.7014719655737824, "learning_rate": 1.0253262022444401e-05, "loss": 0.6214, "step": 14236 }, { "epoch": 0.5072233998966813, "grad_norm": 1.1674390862717288, "learning_rate": 1.025210845420164e-05, "loss": 0.4901, "step": 14237 }, { "epoch": 0.5072590270231755, "grad_norm": 1.1523459363715043, "learning_rate": 1.0250954882601876e-05, "loss": 0.4793, "step": 14238 }, { "epoch": 0.5072946541496696, "grad_norm": 1.2463049109172983, "learning_rate": 1.0249801307660473e-05, "loss": 0.4686, "step": 14239 }, { "epoch": 0.5073302812761636, "grad_norm": 1.7006917473516334, "learning_rate": 1.0248647729392796e-05, "loss": 0.5419, "step": 14240 }, { "epoch": 0.5073659084026578, "grad_norm": 1.7810821977699334, "learning_rate": 1.0247494147814197e-05, "loss": 0.5329, "step": 14241 }, { "epoch": 0.5074015355291519, "grad_norm": 1.065383299873866, "learning_rate": 1.0246340562940038e-05, "loss": 0.4655, "step": 14242 }, { "epoch": 0.507437162655646, "grad_norm": 1.5562744192224134, "learning_rate": 1.0245186974785685e-05, "loss": 0.5898, "step": 14243 }, { "epoch": 0.5074727897821402, "grad_norm": 1.5655879179444825, "learning_rate": 1.0244033383366498e-05, "loss": 0.5536, "step": 14244 }, { "epoch": 0.5075084169086342, "grad_norm": 1.3993880007542425, "learning_rate": 1.0242879788697833e-05, "loss": 0.4821, "step": 14245 }, { "epoch": 0.5075440440351283, "grad_norm": 1.091472903655785, "learning_rate": 1.0241726190795056e-05, "loss": 0.434, "step": 14246 }, { "epoch": 0.5075796711616225, "grad_norm": 1.1040246069798558, "learning_rate": 1.0240572589673526e-05, "loss": 0.469, "step": 14247 }, { "epoch": 0.5076152982881166, "grad_norm": 1.402795313044176, "learning_rate": 1.02394189853486e-05, "loss": 0.5259, "step": 14248 }, { "epoch": 0.5076509254146107, "grad_norm": 2.316943955595632, "learning_rate": 1.0238265377835644e-05, "loss": 0.5295, "step": 14249 }, { "epoch": 0.5076865525411048, "grad_norm": 1.486798381018171, "learning_rate": 1.0237111767150018e-05, "loss": 0.5979, "step": 14250 }, { "epoch": 0.5077221796675989, "grad_norm": 1.416311764477864, "learning_rate": 1.0235958153307087e-05, "loss": 0.5108, "step": 14251 }, { "epoch": 0.507757806794093, "grad_norm": 1.0886318838762497, "learning_rate": 1.0234804536322206e-05, "loss": 0.4119, "step": 14252 }, { "epoch": 0.5077934339205872, "grad_norm": 1.2123497228406557, "learning_rate": 1.0233650916210736e-05, "loss": 0.4925, "step": 14253 }, { "epoch": 0.5078290610470813, "grad_norm": 1.6566171449548253, "learning_rate": 1.0232497292988043e-05, "loss": 0.515, "step": 14254 }, { "epoch": 0.5078646881735753, "grad_norm": 1.4074702987073995, "learning_rate": 1.0231343666669486e-05, "loss": 0.4845, "step": 14255 }, { "epoch": 0.5079003153000695, "grad_norm": 1.7376922005667164, "learning_rate": 1.0230190037270427e-05, "loss": 0.5479, "step": 14256 }, { "epoch": 0.5079359424265636, "grad_norm": 1.4285461715038774, "learning_rate": 1.0229036404806226e-05, "loss": 0.5141, "step": 14257 }, { "epoch": 0.5079715695530577, "grad_norm": 1.5773718546928677, "learning_rate": 1.0227882769292246e-05, "loss": 0.5674, "step": 14258 }, { "epoch": 0.5080071966795519, "grad_norm": 1.4731656874759358, "learning_rate": 1.0226729130743846e-05, "loss": 0.5321, "step": 14259 }, { "epoch": 0.5080428238060459, "grad_norm": 1.0744343197757418, "learning_rate": 1.0225575489176386e-05, "loss": 0.4244, "step": 14260 }, { "epoch": 0.50807845093254, "grad_norm": 1.9950468801052803, "learning_rate": 1.0224421844605236e-05, "loss": 0.5676, "step": 14261 }, { "epoch": 0.5081140780590342, "grad_norm": 1.1056679611567677, "learning_rate": 1.0223268197045752e-05, "loss": 0.4358, "step": 14262 }, { "epoch": 0.5081497051855283, "grad_norm": 1.4141441111979143, "learning_rate": 1.0222114546513296e-05, "loss": 0.5888, "step": 14263 }, { "epoch": 0.5081853323120223, "grad_norm": 1.1505615001642708, "learning_rate": 1.0220960893023226e-05, "loss": 0.4594, "step": 14264 }, { "epoch": 0.5082209594385165, "grad_norm": 1.5810540076411348, "learning_rate": 1.0219807236590911e-05, "loss": 0.436, "step": 14265 }, { "epoch": 0.5082565865650106, "grad_norm": 1.0992594721072375, "learning_rate": 1.0218653577231709e-05, "loss": 0.4234, "step": 14266 }, { "epoch": 0.5082922136915047, "grad_norm": 1.610355396528323, "learning_rate": 1.0217499914960977e-05, "loss": 0.5359, "step": 14267 }, { "epoch": 0.5083278408179989, "grad_norm": 1.0989537346461893, "learning_rate": 1.0216346249794087e-05, "loss": 0.456, "step": 14268 }, { "epoch": 0.5083634679444929, "grad_norm": 1.1421855916827193, "learning_rate": 1.0215192581746394e-05, "loss": 0.4325, "step": 14269 }, { "epoch": 0.508399095070987, "grad_norm": 1.5388609820071515, "learning_rate": 1.0214038910833261e-05, "loss": 0.5678, "step": 14270 }, { "epoch": 0.5084347221974812, "grad_norm": 1.053133153208934, "learning_rate": 1.0212885237070049e-05, "loss": 0.4804, "step": 14271 }, { "epoch": 0.5084703493239753, "grad_norm": 1.2352287379830338, "learning_rate": 1.0211731560472122e-05, "loss": 0.4851, "step": 14272 }, { "epoch": 0.5085059764504694, "grad_norm": 1.1237235485828572, "learning_rate": 1.0210577881054841e-05, "loss": 0.4858, "step": 14273 }, { "epoch": 0.5085416035769635, "grad_norm": 1.4071338349392486, "learning_rate": 1.0209424198833571e-05, "loss": 0.5157, "step": 14274 }, { "epoch": 0.5085772307034576, "grad_norm": 1.5415623464054844, "learning_rate": 1.0208270513823669e-05, "loss": 0.5182, "step": 14275 }, { "epoch": 0.5086128578299517, "grad_norm": 2.2448819678781797, "learning_rate": 1.0207116826040499e-05, "loss": 0.5002, "step": 14276 }, { "epoch": 0.5086484849564459, "grad_norm": 1.2837587724492976, "learning_rate": 1.0205963135499426e-05, "loss": 0.5609, "step": 14277 }, { "epoch": 0.50868411208294, "grad_norm": 1.547391982026642, "learning_rate": 1.020480944221581e-05, "loss": 0.5569, "step": 14278 }, { "epoch": 0.508719739209434, "grad_norm": 1.1943883189207116, "learning_rate": 1.020365574620501e-05, "loss": 0.4911, "step": 14279 }, { "epoch": 0.5087553663359282, "grad_norm": 1.6821905583947452, "learning_rate": 1.0202502047482395e-05, "loss": 0.5135, "step": 14280 }, { "epoch": 0.5087909934624223, "grad_norm": 1.0323612064428422, "learning_rate": 1.0201348346063322e-05, "loss": 0.4458, "step": 14281 }, { "epoch": 0.5088266205889164, "grad_norm": 1.07954937556266, "learning_rate": 1.0200194641963153e-05, "loss": 0.511, "step": 14282 }, { "epoch": 0.5088622477154106, "grad_norm": 1.513497730317829, "learning_rate": 1.0199040935197254e-05, "loss": 0.4514, "step": 14283 }, { "epoch": 0.5088978748419046, "grad_norm": 2.5109759594182886, "learning_rate": 1.0197887225780985e-05, "loss": 0.5777, "step": 14284 }, { "epoch": 0.5089335019683987, "grad_norm": 1.9872003973676373, "learning_rate": 1.0196733513729711e-05, "loss": 0.5944, "step": 14285 }, { "epoch": 0.5089691290948929, "grad_norm": 1.6518737434887094, "learning_rate": 1.019557979905879e-05, "loss": 0.5577, "step": 14286 }, { "epoch": 0.509004756221387, "grad_norm": 1.6193839041790905, "learning_rate": 1.019442608178359e-05, "loss": 0.6005, "step": 14287 }, { "epoch": 0.509040383347881, "grad_norm": 1.0444689361352468, "learning_rate": 1.0193272361919473e-05, "loss": 0.4179, "step": 14288 }, { "epoch": 0.5090760104743752, "grad_norm": 1.4345130620327056, "learning_rate": 1.0192118639481795e-05, "loss": 0.5443, "step": 14289 }, { "epoch": 0.5091116376008693, "grad_norm": 1.3441899941935083, "learning_rate": 1.0190964914485926e-05, "loss": 0.5582, "step": 14290 }, { "epoch": 0.5091472647273634, "grad_norm": 2.0464368026220443, "learning_rate": 1.0189811186947224e-05, "loss": 0.4873, "step": 14291 }, { "epoch": 0.5091828918538576, "grad_norm": 1.5685491463566295, "learning_rate": 1.0188657456881054e-05, "loss": 0.6103, "step": 14292 }, { "epoch": 0.5092185189803516, "grad_norm": 1.0460817017761312, "learning_rate": 1.0187503724302776e-05, "loss": 0.4839, "step": 14293 }, { "epoch": 0.5092541461068457, "grad_norm": 1.9188699286140227, "learning_rate": 1.0186349989227755e-05, "loss": 0.5803, "step": 14294 }, { "epoch": 0.5092897732333399, "grad_norm": 1.0998372417056224, "learning_rate": 1.0185196251671357e-05, "loss": 0.4393, "step": 14295 }, { "epoch": 0.509325400359834, "grad_norm": 4.263402104997236, "learning_rate": 1.0184042511648942e-05, "loss": 0.576, "step": 14296 }, { "epoch": 0.509361027486328, "grad_norm": 1.3638861902279171, "learning_rate": 1.018288876917587e-05, "loss": 0.611, "step": 14297 }, { "epoch": 0.5093966546128222, "grad_norm": 1.1448802867546368, "learning_rate": 1.0181735024267504e-05, "loss": 0.4763, "step": 14298 }, { "epoch": 0.5094322817393163, "grad_norm": 1.523379631503945, "learning_rate": 1.0180581276939213e-05, "loss": 0.5772, "step": 14299 }, { "epoch": 0.5094679088658104, "grad_norm": 1.6320291044671278, "learning_rate": 1.0179427527206354e-05, "loss": 0.5289, "step": 14300 }, { "epoch": 0.5095035359923046, "grad_norm": 1.080309313846061, "learning_rate": 1.0178273775084292e-05, "loss": 0.4535, "step": 14301 }, { "epoch": 0.5095391631187987, "grad_norm": 1.0960046876208176, "learning_rate": 1.0177120020588392e-05, "loss": 0.4696, "step": 14302 }, { "epoch": 0.5095747902452927, "grad_norm": 1.0199429513763583, "learning_rate": 1.0175966263734014e-05, "loss": 0.3953, "step": 14303 }, { "epoch": 0.5096104173717869, "grad_norm": 1.0333865931741224, "learning_rate": 1.0174812504536523e-05, "loss": 0.4616, "step": 14304 }, { "epoch": 0.509646044498281, "grad_norm": 1.5294057748573526, "learning_rate": 1.017365874301128e-05, "loss": 0.5768, "step": 14305 }, { "epoch": 0.5096816716247751, "grad_norm": 1.3232563750875495, "learning_rate": 1.0172504979173647e-05, "loss": 0.5697, "step": 14306 }, { "epoch": 0.5097172987512693, "grad_norm": 1.0915409341603601, "learning_rate": 1.0171351213038993e-05, "loss": 0.439, "step": 14307 }, { "epoch": 0.5097529258777633, "grad_norm": 1.5772647277087701, "learning_rate": 1.0170197444622678e-05, "loss": 0.5664, "step": 14308 }, { "epoch": 0.5097885530042574, "grad_norm": 1.0997230517104337, "learning_rate": 1.0169043673940062e-05, "loss": 0.4328, "step": 14309 }, { "epoch": 0.5098241801307516, "grad_norm": 1.4427325342829076, "learning_rate": 1.0167889901006516e-05, "loss": 0.4549, "step": 14310 }, { "epoch": 0.5098598072572457, "grad_norm": 2.3387926712909284, "learning_rate": 1.0166736125837395e-05, "loss": 0.5527, "step": 14311 }, { "epoch": 0.5098954343837397, "grad_norm": 1.3718770553985669, "learning_rate": 1.0165582348448066e-05, "loss": 0.5009, "step": 14312 }, { "epoch": 0.5099310615102339, "grad_norm": 1.1482085238712223, "learning_rate": 1.0164428568853894e-05, "loss": 0.4642, "step": 14313 }, { "epoch": 0.509966688636728, "grad_norm": 1.208233995700112, "learning_rate": 1.016327478707024e-05, "loss": 0.4527, "step": 14314 }, { "epoch": 0.5100023157632221, "grad_norm": 1.185187251096304, "learning_rate": 1.0162121003112471e-05, "loss": 0.5248, "step": 14315 }, { "epoch": 0.5100379428897163, "grad_norm": 1.4442519461954155, "learning_rate": 1.0160967216995942e-05, "loss": 0.5553, "step": 14316 }, { "epoch": 0.5100735700162103, "grad_norm": 1.1557506421539672, "learning_rate": 1.0159813428736027e-05, "loss": 0.4631, "step": 14317 }, { "epoch": 0.5101091971427044, "grad_norm": 1.0912710337707574, "learning_rate": 1.015865963834808e-05, "loss": 0.4343, "step": 14318 }, { "epoch": 0.5101448242691986, "grad_norm": 1.04740047582491, "learning_rate": 1.0157505845847473e-05, "loss": 0.4515, "step": 14319 }, { "epoch": 0.5101804513956927, "grad_norm": 2.464657190521218, "learning_rate": 1.0156352051249561e-05, "loss": 0.5481, "step": 14320 }, { "epoch": 0.5102160785221868, "grad_norm": 1.6001200187932691, "learning_rate": 1.0155198254569716e-05, "loss": 0.5157, "step": 14321 }, { "epoch": 0.5102517056486809, "grad_norm": 1.6434921772903128, "learning_rate": 1.0154044455823298e-05, "loss": 0.5299, "step": 14322 }, { "epoch": 0.510287332775175, "grad_norm": 1.5796156062000732, "learning_rate": 1.0152890655025669e-05, "loss": 0.5471, "step": 14323 }, { "epoch": 0.5103229599016691, "grad_norm": 1.1551635965944467, "learning_rate": 1.0151736852192193e-05, "loss": 0.4622, "step": 14324 }, { "epoch": 0.5103585870281633, "grad_norm": 1.4098967868011862, "learning_rate": 1.0150583047338236e-05, "loss": 0.566, "step": 14325 }, { "epoch": 0.5103942141546574, "grad_norm": 1.5065746776378754, "learning_rate": 1.0149429240479163e-05, "loss": 0.5164, "step": 14326 }, { "epoch": 0.5104298412811514, "grad_norm": 1.6861487174186744, "learning_rate": 1.014827543163033e-05, "loss": 0.4827, "step": 14327 }, { "epoch": 0.5104654684076456, "grad_norm": 1.1139925620511861, "learning_rate": 1.014712162080711e-05, "loss": 0.4369, "step": 14328 }, { "epoch": 0.5105010955341397, "grad_norm": 1.2759484554946052, "learning_rate": 1.014596780802486e-05, "loss": 0.4601, "step": 14329 }, { "epoch": 0.5105367226606338, "grad_norm": 1.032088696724107, "learning_rate": 1.014481399329895e-05, "loss": 0.4062, "step": 14330 }, { "epoch": 0.510572349787128, "grad_norm": 1.2525667370281708, "learning_rate": 1.0143660176644736e-05, "loss": 0.4327, "step": 14331 }, { "epoch": 0.510607976913622, "grad_norm": 1.5061151585052588, "learning_rate": 1.0142506358077588e-05, "loss": 0.529, "step": 14332 }, { "epoch": 0.5106436040401161, "grad_norm": 1.3116806530586176, "learning_rate": 1.0141352537612872e-05, "loss": 0.5533, "step": 14333 }, { "epoch": 0.5106792311666103, "grad_norm": 1.6007337630963803, "learning_rate": 1.0140198715265945e-05, "loss": 0.6197, "step": 14334 }, { "epoch": 0.5107148582931044, "grad_norm": 1.4831173224599201, "learning_rate": 1.0139044891052171e-05, "loss": 0.5871, "step": 14335 }, { "epoch": 0.5107504854195984, "grad_norm": 1.0650040145891306, "learning_rate": 1.0137891064986923e-05, "loss": 0.4608, "step": 14336 }, { "epoch": 0.5107861125460926, "grad_norm": 1.0607727191551872, "learning_rate": 1.0136737237085557e-05, "loss": 0.4684, "step": 14337 }, { "epoch": 0.5108217396725867, "grad_norm": 1.2916897988670535, "learning_rate": 1.0135583407363439e-05, "loss": 0.4708, "step": 14338 }, { "epoch": 0.5108573667990808, "grad_norm": 1.611790619652618, "learning_rate": 1.0134429575835932e-05, "loss": 0.4965, "step": 14339 }, { "epoch": 0.510892993925575, "grad_norm": 1.7151789383652414, "learning_rate": 1.0133275742518402e-05, "loss": 0.5591, "step": 14340 }, { "epoch": 0.510928621052069, "grad_norm": 1.3622459078424387, "learning_rate": 1.0132121907426212e-05, "loss": 0.5581, "step": 14341 }, { "epoch": 0.5109642481785631, "grad_norm": 1.116624077812536, "learning_rate": 1.0130968070574725e-05, "loss": 0.486, "step": 14342 }, { "epoch": 0.5109998753050573, "grad_norm": 1.5847568974793975, "learning_rate": 1.012981423197931e-05, "loss": 0.5308, "step": 14343 }, { "epoch": 0.5110355024315514, "grad_norm": 1.6524662347506793, "learning_rate": 1.0128660391655327e-05, "loss": 0.5978, "step": 14344 }, { "epoch": 0.5110711295580455, "grad_norm": 1.4145981366940323, "learning_rate": 1.012750654961814e-05, "loss": 0.5612, "step": 14345 }, { "epoch": 0.5111067566845396, "grad_norm": 1.3866282669843106, "learning_rate": 1.0126352705883113e-05, "loss": 0.5178, "step": 14346 }, { "epoch": 0.5111423838110337, "grad_norm": 1.3855604976203597, "learning_rate": 1.0125198860465615e-05, "loss": 0.6254, "step": 14347 }, { "epoch": 0.5111780109375278, "grad_norm": 1.5711521528926053, "learning_rate": 1.0124045013381003e-05, "loss": 0.5661, "step": 14348 }, { "epoch": 0.511213638064022, "grad_norm": 1.949950058384824, "learning_rate": 1.0122891164644646e-05, "loss": 0.5244, "step": 14349 }, { "epoch": 0.5112492651905161, "grad_norm": 1.2137445277608163, "learning_rate": 1.0121737314271908e-05, "loss": 0.5318, "step": 14350 }, { "epoch": 0.5112848923170101, "grad_norm": 1.2207272832939322, "learning_rate": 1.0120583462278153e-05, "loss": 0.478, "step": 14351 }, { "epoch": 0.5113205194435043, "grad_norm": 1.1214956911334102, "learning_rate": 1.0119429608678744e-05, "loss": 0.4835, "step": 14352 }, { "epoch": 0.5113561465699984, "grad_norm": 1.09178399349382, "learning_rate": 1.0118275753489047e-05, "loss": 0.4388, "step": 14353 }, { "epoch": 0.5113917736964925, "grad_norm": 1.0797380797943756, "learning_rate": 1.0117121896724422e-05, "loss": 0.458, "step": 14354 }, { "epoch": 0.5114274008229867, "grad_norm": 1.4470901369113989, "learning_rate": 1.011596803840024e-05, "loss": 0.5535, "step": 14355 }, { "epoch": 0.5114630279494807, "grad_norm": 1.401071092283604, "learning_rate": 1.0114814178531864e-05, "loss": 0.5303, "step": 14356 }, { "epoch": 0.5114986550759748, "grad_norm": 1.6736601447139674, "learning_rate": 1.0113660317134655e-05, "loss": 0.5971, "step": 14357 }, { "epoch": 0.511534282202469, "grad_norm": 1.0492669807994794, "learning_rate": 1.0112506454223981e-05, "loss": 0.4527, "step": 14358 }, { "epoch": 0.5115699093289631, "grad_norm": 1.0231225565789663, "learning_rate": 1.0111352589815204e-05, "loss": 0.4691, "step": 14359 }, { "epoch": 0.5116055364554571, "grad_norm": 1.191954159012405, "learning_rate": 1.011019872392369e-05, "loss": 0.4591, "step": 14360 }, { "epoch": 0.5116411635819513, "grad_norm": 1.4355917648271637, "learning_rate": 1.01090448565648e-05, "loss": 0.5819, "step": 14361 }, { "epoch": 0.5116767907084454, "grad_norm": 1.4333643921618253, "learning_rate": 1.0107890987753905e-05, "loss": 0.5266, "step": 14362 }, { "epoch": 0.5117124178349395, "grad_norm": 1.037830827085288, "learning_rate": 1.0106737117506367e-05, "loss": 0.4482, "step": 14363 }, { "epoch": 0.5117480449614337, "grad_norm": 1.3961019893462463, "learning_rate": 1.0105583245837547e-05, "loss": 0.4957, "step": 14364 }, { "epoch": 0.5117836720879277, "grad_norm": 1.0571251138863829, "learning_rate": 1.010442937276281e-05, "loss": 0.4051, "step": 14365 }, { "epoch": 0.5118192992144218, "grad_norm": 1.233786201460848, "learning_rate": 1.0103275498297526e-05, "loss": 0.4646, "step": 14366 }, { "epoch": 0.511854926340916, "grad_norm": 1.4741200860421808, "learning_rate": 1.0102121622457057e-05, "loss": 0.5382, "step": 14367 }, { "epoch": 0.5118905534674101, "grad_norm": 1.0808046764799633, "learning_rate": 1.0100967745256767e-05, "loss": 0.4216, "step": 14368 }, { "epoch": 0.5119261805939042, "grad_norm": 1.0726753635438497, "learning_rate": 1.009981386671202e-05, "loss": 0.4655, "step": 14369 }, { "epoch": 0.5119618077203983, "grad_norm": 1.0282823725339216, "learning_rate": 1.0098659986838185e-05, "loss": 0.4682, "step": 14370 }, { "epoch": 0.5119974348468924, "grad_norm": 1.144992289519677, "learning_rate": 1.0097506105650619e-05, "loss": 0.4285, "step": 14371 }, { "epoch": 0.5120330619733865, "grad_norm": 1.031111344825894, "learning_rate": 1.0096352223164692e-05, "loss": 0.4476, "step": 14372 }, { "epoch": 0.5120686890998807, "grad_norm": 0.9241170055928176, "learning_rate": 1.0095198339395769e-05, "loss": 0.4188, "step": 14373 }, { "epoch": 0.5121043162263748, "grad_norm": 1.4086035066107967, "learning_rate": 1.0094044454359214e-05, "loss": 0.4609, "step": 14374 }, { "epoch": 0.5121399433528688, "grad_norm": 1.529159225399968, "learning_rate": 1.0092890568070388e-05, "loss": 0.5422, "step": 14375 }, { "epoch": 0.512175570479363, "grad_norm": 1.6175330663016796, "learning_rate": 1.009173668054466e-05, "loss": 0.5268, "step": 14376 }, { "epoch": 0.5122111976058571, "grad_norm": 1.1374268393474323, "learning_rate": 1.0090582791797397e-05, "loss": 0.4354, "step": 14377 }, { "epoch": 0.5122468247323512, "grad_norm": 1.2790310031562402, "learning_rate": 1.0089428901843958e-05, "loss": 0.4208, "step": 14378 }, { "epoch": 0.5122824518588454, "grad_norm": 2.2157556986001623, "learning_rate": 1.0088275010699713e-05, "loss": 0.5541, "step": 14379 }, { "epoch": 0.5123180789853394, "grad_norm": 1.397484487800015, "learning_rate": 1.0087121118380021e-05, "loss": 0.5595, "step": 14380 }, { "epoch": 0.5123537061118335, "grad_norm": 1.1970939746769822, "learning_rate": 1.0085967224900256e-05, "loss": 0.5328, "step": 14381 }, { "epoch": 0.5123893332383277, "grad_norm": 1.1296960144882051, "learning_rate": 1.0084813330275774e-05, "loss": 0.4854, "step": 14382 }, { "epoch": 0.5124249603648218, "grad_norm": 1.1425907690227444, "learning_rate": 1.0083659434521941e-05, "loss": 0.4471, "step": 14383 }, { "epoch": 0.5124605874913158, "grad_norm": 1.121497144988561, "learning_rate": 1.0082505537654128e-05, "loss": 0.478, "step": 14384 }, { "epoch": 0.51249621461781, "grad_norm": 1.1384418227396533, "learning_rate": 1.0081351639687695e-05, "loss": 0.5203, "step": 14385 }, { "epoch": 0.5125318417443041, "grad_norm": 1.0472035046090122, "learning_rate": 1.0080197740638008e-05, "loss": 0.4043, "step": 14386 }, { "epoch": 0.5125674688707982, "grad_norm": 1.4003106276546973, "learning_rate": 1.0079043840520428e-05, "loss": 0.6335, "step": 14387 }, { "epoch": 0.5126030959972924, "grad_norm": 1.0317786966741727, "learning_rate": 1.007788993935033e-05, "loss": 0.4713, "step": 14388 }, { "epoch": 0.5126387231237864, "grad_norm": 1.0942578282359903, "learning_rate": 1.0076736037143069e-05, "loss": 0.4552, "step": 14389 }, { "epoch": 0.5126743502502805, "grad_norm": 1.094371860758878, "learning_rate": 1.0075582133914016e-05, "loss": 0.4595, "step": 14390 }, { "epoch": 0.5127099773767747, "grad_norm": 1.3533513557773211, "learning_rate": 1.0074428229678535e-05, "loss": 0.53, "step": 14391 }, { "epoch": 0.5127456045032688, "grad_norm": 1.5577426945523727, "learning_rate": 1.0073274324451989e-05, "loss": 0.6229, "step": 14392 }, { "epoch": 0.5127812316297629, "grad_norm": 1.4457519076820444, "learning_rate": 1.0072120418249744e-05, "loss": 0.4474, "step": 14393 }, { "epoch": 0.512816858756257, "grad_norm": 0.9809261993503187, "learning_rate": 1.0070966511087164e-05, "loss": 0.4147, "step": 14394 }, { "epoch": 0.5128524858827511, "grad_norm": 1.1027746669813068, "learning_rate": 1.0069812602979617e-05, "loss": 0.43, "step": 14395 }, { "epoch": 0.5128881130092452, "grad_norm": 1.2832167346238321, "learning_rate": 1.0068658693942466e-05, "loss": 0.5192, "step": 14396 }, { "epoch": 0.5129237401357394, "grad_norm": 1.365755336539242, "learning_rate": 1.0067504783991078e-05, "loss": 0.5346, "step": 14397 }, { "epoch": 0.5129593672622335, "grad_norm": 1.1180538317836899, "learning_rate": 1.0066350873140814e-05, "loss": 0.4756, "step": 14398 }, { "epoch": 0.5129949943887275, "grad_norm": 1.1086913620030907, "learning_rate": 1.006519696140704e-05, "loss": 0.4964, "step": 14399 }, { "epoch": 0.5130306215152217, "grad_norm": 1.6894812041732095, "learning_rate": 1.0064043048805129e-05, "loss": 0.6254, "step": 14400 }, { "epoch": 0.5130662486417158, "grad_norm": 1.684171871443131, "learning_rate": 1.0062889135350435e-05, "loss": 0.5849, "step": 14401 }, { "epoch": 0.5131018757682099, "grad_norm": 1.8625289421166418, "learning_rate": 1.0061735221058328e-05, "loss": 0.5738, "step": 14402 }, { "epoch": 0.5131375028947041, "grad_norm": 1.0669943662203747, "learning_rate": 1.0060581305944177e-05, "loss": 0.4591, "step": 14403 }, { "epoch": 0.5131731300211981, "grad_norm": 0.998480578109478, "learning_rate": 1.0059427390023342e-05, "loss": 0.4295, "step": 14404 }, { "epoch": 0.5132087571476922, "grad_norm": 1.4463112313993278, "learning_rate": 1.005827347331119e-05, "loss": 0.5988, "step": 14405 }, { "epoch": 0.5132443842741864, "grad_norm": 2.348217812538265, "learning_rate": 1.0057119555823085e-05, "loss": 0.5326, "step": 14406 }, { "epoch": 0.5132800114006805, "grad_norm": 1.6494153013138606, "learning_rate": 1.0055965637574394e-05, "loss": 0.5289, "step": 14407 }, { "epoch": 0.5133156385271745, "grad_norm": 1.002675953143971, "learning_rate": 1.0054811718580485e-05, "loss": 0.4016, "step": 14408 }, { "epoch": 0.5133512656536687, "grad_norm": 1.1618535259039326, "learning_rate": 1.0053657798856712e-05, "loss": 0.4342, "step": 14409 }, { "epoch": 0.5133868927801628, "grad_norm": 1.649730767488376, "learning_rate": 1.0052503878418451e-05, "loss": 0.614, "step": 14410 }, { "epoch": 0.5134225199066569, "grad_norm": 1.1758990494507249, "learning_rate": 1.0051349957281064e-05, "loss": 0.5062, "step": 14411 }, { "epoch": 0.5134581470331511, "grad_norm": 1.4602083978787506, "learning_rate": 1.0050196035459921e-05, "loss": 0.6131, "step": 14412 }, { "epoch": 0.5134937741596451, "grad_norm": 0.9992198445228416, "learning_rate": 1.0049042112970375e-05, "loss": 0.4731, "step": 14413 }, { "epoch": 0.5135294012861392, "grad_norm": 1.6422328014003222, "learning_rate": 1.0047888189827806e-05, "loss": 0.5375, "step": 14414 }, { "epoch": 0.5135650284126334, "grad_norm": 1.102039251730605, "learning_rate": 1.0046734266047568e-05, "loss": 0.489, "step": 14415 }, { "epoch": 0.5136006555391275, "grad_norm": 1.057218330983355, "learning_rate": 1.0045580341645033e-05, "loss": 0.4033, "step": 14416 }, { "epoch": 0.5136362826656216, "grad_norm": 1.0553703992636827, "learning_rate": 1.0044426416635562e-05, "loss": 0.3985, "step": 14417 }, { "epoch": 0.5136719097921157, "grad_norm": 1.152564868735465, "learning_rate": 1.0043272491034523e-05, "loss": 0.4419, "step": 14418 }, { "epoch": 0.5137075369186098, "grad_norm": 1.2622582074806648, "learning_rate": 1.0042118564857281e-05, "loss": 0.5531, "step": 14419 }, { "epoch": 0.5137431640451039, "grad_norm": 1.1280810171018634, "learning_rate": 1.0040964638119199e-05, "loss": 0.4626, "step": 14420 }, { "epoch": 0.5137787911715981, "grad_norm": 1.3325643356046653, "learning_rate": 1.0039810710835644e-05, "loss": 0.5537, "step": 14421 }, { "epoch": 0.5138144182980922, "grad_norm": 1.47841283606029, "learning_rate": 1.0038656783021985e-05, "loss": 0.5199, "step": 14422 }, { "epoch": 0.5138500454245862, "grad_norm": 1.234593750784954, "learning_rate": 1.0037502854693583e-05, "loss": 0.4595, "step": 14423 }, { "epoch": 0.5138856725510804, "grad_norm": 1.1406518578151568, "learning_rate": 1.0036348925865803e-05, "loss": 0.457, "step": 14424 }, { "epoch": 0.5139212996775745, "grad_norm": 1.1241404559725716, "learning_rate": 1.0035194996554012e-05, "loss": 0.4858, "step": 14425 }, { "epoch": 0.5139569268040686, "grad_norm": 1.6057231412499826, "learning_rate": 1.0034041066773577e-05, "loss": 0.6293, "step": 14426 }, { "epoch": 0.5139925539305628, "grad_norm": 1.2431729941859568, "learning_rate": 1.0032887136539859e-05, "loss": 0.4369, "step": 14427 }, { "epoch": 0.5140281810570568, "grad_norm": 1.5064949750454504, "learning_rate": 1.0031733205868223e-05, "loss": 0.5781, "step": 14428 }, { "epoch": 0.5140638081835509, "grad_norm": 2.786738096911854, "learning_rate": 1.0030579274774043e-05, "loss": 0.5669, "step": 14429 }, { "epoch": 0.5140994353100451, "grad_norm": 1.5712839971915187, "learning_rate": 1.0029425343272676e-05, "loss": 0.5348, "step": 14430 }, { "epoch": 0.5141350624365392, "grad_norm": 1.1157327945437228, "learning_rate": 1.0028271411379488e-05, "loss": 0.4625, "step": 14431 }, { "epoch": 0.5141706895630332, "grad_norm": 1.4448493881756388, "learning_rate": 1.002711747910985e-05, "loss": 0.5326, "step": 14432 }, { "epoch": 0.5142063166895274, "grad_norm": 2.987605832403213, "learning_rate": 1.002596354647912e-05, "loss": 0.5503, "step": 14433 }, { "epoch": 0.5142419438160215, "grad_norm": 1.6178315952061044, "learning_rate": 1.0024809613502671e-05, "loss": 0.5484, "step": 14434 }, { "epoch": 0.5142775709425156, "grad_norm": 1.1215394173489643, "learning_rate": 1.0023655680195865e-05, "loss": 0.4408, "step": 14435 }, { "epoch": 0.5143131980690098, "grad_norm": 1.380961105327906, "learning_rate": 1.0022501746574063e-05, "loss": 0.5172, "step": 14436 }, { "epoch": 0.5143488251955038, "grad_norm": 1.6885361852238454, "learning_rate": 1.0021347812652638e-05, "loss": 0.5774, "step": 14437 }, { "epoch": 0.5143844523219979, "grad_norm": 1.2513771249323609, "learning_rate": 1.0020193878446952e-05, "loss": 0.4693, "step": 14438 }, { "epoch": 0.5144200794484921, "grad_norm": 2.111002614577649, "learning_rate": 1.0019039943972367e-05, "loss": 0.4787, "step": 14439 }, { "epoch": 0.5144557065749862, "grad_norm": 1.2006869732093917, "learning_rate": 1.0017886009244253e-05, "loss": 0.4801, "step": 14440 }, { "epoch": 0.5144913337014803, "grad_norm": 1.5888209473175745, "learning_rate": 1.0016732074277977e-05, "loss": 0.5083, "step": 14441 }, { "epoch": 0.5145269608279744, "grad_norm": 1.1022361750658778, "learning_rate": 1.0015578139088901e-05, "loss": 0.4764, "step": 14442 }, { "epoch": 0.5145625879544685, "grad_norm": 1.0817544821734562, "learning_rate": 1.0014424203692388e-05, "loss": 0.4518, "step": 14443 }, { "epoch": 0.5145982150809626, "grad_norm": 1.1564765038456868, "learning_rate": 1.0013270268103808e-05, "loss": 0.4397, "step": 14444 }, { "epoch": 0.5146338422074568, "grad_norm": 1.3248153548160009, "learning_rate": 1.0012116332338526e-05, "loss": 0.4238, "step": 14445 }, { "epoch": 0.5146694693339509, "grad_norm": 1.136814646832656, "learning_rate": 1.0010962396411908e-05, "loss": 0.475, "step": 14446 }, { "epoch": 0.5147050964604449, "grad_norm": 1.1370361032944767, "learning_rate": 1.0009808460339314e-05, "loss": 0.425, "step": 14447 }, { "epoch": 0.5147407235869391, "grad_norm": 1.454812655503243, "learning_rate": 1.0008654524136116e-05, "loss": 0.5781, "step": 14448 }, { "epoch": 0.5147763507134332, "grad_norm": 1.4308379923347068, "learning_rate": 1.0007500587817677e-05, "loss": 0.5863, "step": 14449 }, { "epoch": 0.5148119778399273, "grad_norm": 1.068453166053246, "learning_rate": 1.0006346651399363e-05, "loss": 0.4104, "step": 14450 }, { "epoch": 0.5148476049664215, "grad_norm": 1.6920638119128517, "learning_rate": 1.000519271489654e-05, "loss": 0.518, "step": 14451 }, { "epoch": 0.5148832320929155, "grad_norm": 1.1190754719987825, "learning_rate": 1.000403877832457e-05, "loss": 0.4665, "step": 14452 }, { "epoch": 0.5149188592194096, "grad_norm": 2.401215388261764, "learning_rate": 1.0002884841698822e-05, "loss": 0.6195, "step": 14453 }, { "epoch": 0.5149544863459038, "grad_norm": 2.0031231685723823, "learning_rate": 1.0001730905034658e-05, "loss": 0.5749, "step": 14454 }, { "epoch": 0.5149901134723979, "grad_norm": 1.0529156863802374, "learning_rate": 1.0000576968347449e-05, "loss": 0.4376, "step": 14455 }, { "epoch": 0.515025740598892, "grad_norm": 1.178525648953779, "learning_rate": 9.999423031652556e-06, "loss": 0.3925, "step": 14456 }, { "epoch": 0.5150613677253861, "grad_norm": 1.804963810743437, "learning_rate": 9.998269094965345e-06, "loss": 0.6209, "step": 14457 }, { "epoch": 0.5150969948518802, "grad_norm": 2.4497891373524334, "learning_rate": 9.997115158301183e-06, "loss": 0.5234, "step": 14458 }, { "epoch": 0.5151326219783743, "grad_norm": 1.1106497835075892, "learning_rate": 9.995961221675433e-06, "loss": 0.4087, "step": 14459 }, { "epoch": 0.5151682491048685, "grad_norm": 1.1961465485483822, "learning_rate": 9.994807285103466e-06, "loss": 0.4508, "step": 14460 }, { "epoch": 0.5152038762313625, "grad_norm": 1.6254483857077673, "learning_rate": 9.99365334860064e-06, "loss": 0.5811, "step": 14461 }, { "epoch": 0.5152395033578566, "grad_norm": 1.1510408759579116, "learning_rate": 9.992499412182321e-06, "loss": 0.4514, "step": 14462 }, { "epoch": 0.5152751304843508, "grad_norm": 1.107962977957937, "learning_rate": 9.991345475863886e-06, "loss": 0.4736, "step": 14463 }, { "epoch": 0.5153107576108449, "grad_norm": 0.9607811799443592, "learning_rate": 9.990191539660686e-06, "loss": 0.4441, "step": 14464 }, { "epoch": 0.515346384737339, "grad_norm": 1.4778557420685328, "learning_rate": 9.989037603588096e-06, "loss": 0.514, "step": 14465 }, { "epoch": 0.5153820118638331, "grad_norm": 1.0446440802831032, "learning_rate": 9.987883667661474e-06, "loss": 0.4138, "step": 14466 }, { "epoch": 0.5154176389903272, "grad_norm": 1.7232436844735364, "learning_rate": 9.986729731896196e-06, "loss": 0.556, "step": 14467 }, { "epoch": 0.5154532661168213, "grad_norm": 1.1127549527513354, "learning_rate": 9.985575796307615e-06, "loss": 0.461, "step": 14468 }, { "epoch": 0.5154888932433155, "grad_norm": 1.2544448171739395, "learning_rate": 9.984421860911104e-06, "loss": 0.5009, "step": 14469 }, { "epoch": 0.5155245203698096, "grad_norm": 1.8594088733883953, "learning_rate": 9.983267925722026e-06, "loss": 0.6258, "step": 14470 }, { "epoch": 0.5155601474963036, "grad_norm": 1.0675428001839666, "learning_rate": 9.98211399075575e-06, "loss": 0.4043, "step": 14471 }, { "epoch": 0.5155957746227978, "grad_norm": 1.4480668958485032, "learning_rate": 9.980960056027637e-06, "loss": 0.4925, "step": 14472 }, { "epoch": 0.5156314017492919, "grad_norm": 1.1287331564601768, "learning_rate": 9.979806121553052e-06, "loss": 0.4614, "step": 14473 }, { "epoch": 0.515667028875786, "grad_norm": 0.9564362629774786, "learning_rate": 9.978652187347366e-06, "loss": 0.4521, "step": 14474 }, { "epoch": 0.5157026560022802, "grad_norm": 1.114839737865667, "learning_rate": 9.977498253425937e-06, "loss": 0.432, "step": 14475 }, { "epoch": 0.5157382831287742, "grad_norm": 1.0697741185374514, "learning_rate": 9.976344319804138e-06, "loss": 0.4417, "step": 14476 }, { "epoch": 0.5157739102552683, "grad_norm": 1.0892413961072567, "learning_rate": 9.975190386497329e-06, "loss": 0.4868, "step": 14477 }, { "epoch": 0.5158095373817625, "grad_norm": 1.5880891116055016, "learning_rate": 9.974036453520881e-06, "loss": 0.6051, "step": 14478 }, { "epoch": 0.5158451645082566, "grad_norm": 1.563347118477292, "learning_rate": 9.972882520890153e-06, "loss": 0.5512, "step": 14479 }, { "epoch": 0.5158807916347506, "grad_norm": 1.0434801249961472, "learning_rate": 9.971728588620513e-06, "loss": 0.4393, "step": 14480 }, { "epoch": 0.5159164187612448, "grad_norm": 1.5029487139722506, "learning_rate": 9.970574656727327e-06, "loss": 0.5954, "step": 14481 }, { "epoch": 0.5159520458877389, "grad_norm": 1.2347055006389986, "learning_rate": 9.969420725225962e-06, "loss": 0.4444, "step": 14482 }, { "epoch": 0.515987673014233, "grad_norm": 1.1547273873896178, "learning_rate": 9.968266794131778e-06, "loss": 0.4654, "step": 14483 }, { "epoch": 0.5160233001407272, "grad_norm": 1.1603834019047758, "learning_rate": 9.967112863460146e-06, "loss": 0.4261, "step": 14484 }, { "epoch": 0.5160589272672212, "grad_norm": 1.4201778257825952, "learning_rate": 9.965958933226428e-06, "loss": 0.5087, "step": 14485 }, { "epoch": 0.5160945543937153, "grad_norm": 1.1981859902849679, "learning_rate": 9.96480500344599e-06, "loss": 0.4451, "step": 14486 }, { "epoch": 0.5161301815202095, "grad_norm": 1.7191381154904306, "learning_rate": 9.9636510741342e-06, "loss": 0.5511, "step": 14487 }, { "epoch": 0.5161658086467036, "grad_norm": 1.5591046085077676, "learning_rate": 9.962497145306418e-06, "loss": 0.5318, "step": 14488 }, { "epoch": 0.5162014357731977, "grad_norm": 1.1469529326147414, "learning_rate": 9.961343216978018e-06, "loss": 0.4586, "step": 14489 }, { "epoch": 0.5162370628996918, "grad_norm": 1.0579342711167605, "learning_rate": 9.960189289164354e-06, "loss": 0.442, "step": 14490 }, { "epoch": 0.5162726900261859, "grad_norm": 1.053838177951104, "learning_rate": 9.959035361880803e-06, "loss": 0.4262, "step": 14491 }, { "epoch": 0.51630831715268, "grad_norm": 1.2197910715047178, "learning_rate": 9.95788143514272e-06, "loss": 0.5169, "step": 14492 }, { "epoch": 0.5163439442791742, "grad_norm": 1.0558469699801312, "learning_rate": 9.956727508965482e-06, "loss": 0.4027, "step": 14493 }, { "epoch": 0.5163795714056683, "grad_norm": 1.1278122243646105, "learning_rate": 9.955573583364441e-06, "loss": 0.4401, "step": 14494 }, { "epoch": 0.5164151985321623, "grad_norm": 1.175223632212147, "learning_rate": 9.954419658354972e-06, "loss": 0.4326, "step": 14495 }, { "epoch": 0.5164508256586565, "grad_norm": 1.053977298790366, "learning_rate": 9.953265733952434e-06, "loss": 0.441, "step": 14496 }, { "epoch": 0.5164864527851506, "grad_norm": 1.5651648133921072, "learning_rate": 9.952111810172195e-06, "loss": 0.5719, "step": 14497 }, { "epoch": 0.5165220799116447, "grad_norm": 1.3355880304883614, "learning_rate": 9.950957887029627e-06, "loss": 0.5651, "step": 14498 }, { "epoch": 0.5165577070381389, "grad_norm": 1.235309428576393, "learning_rate": 9.94980396454008e-06, "loss": 0.4749, "step": 14499 }, { "epoch": 0.5165933341646329, "grad_norm": 1.2912187223418288, "learning_rate": 9.948650042718938e-06, "loss": 0.5633, "step": 14500 }, { "epoch": 0.516628961291127, "grad_norm": 1.061370869223259, "learning_rate": 9.94749612158155e-06, "loss": 0.4412, "step": 14501 }, { "epoch": 0.5166645884176212, "grad_norm": 1.5484957243534754, "learning_rate": 9.946342201143291e-06, "loss": 0.5362, "step": 14502 }, { "epoch": 0.5167002155441153, "grad_norm": 1.0001161042726714, "learning_rate": 9.94518828141952e-06, "loss": 0.4707, "step": 14503 }, { "epoch": 0.5167358426706093, "grad_norm": 1.1266687809548412, "learning_rate": 9.94403436242561e-06, "loss": 0.447, "step": 14504 }, { "epoch": 0.5167714697971035, "grad_norm": 1.0610983110618748, "learning_rate": 9.942880444176918e-06, "loss": 0.433, "step": 14505 }, { "epoch": 0.5168070969235976, "grad_norm": 1.1021865705514624, "learning_rate": 9.941726526688815e-06, "loss": 0.4592, "step": 14506 }, { "epoch": 0.5168427240500918, "grad_norm": 1.1255334116035272, "learning_rate": 9.94057260997666e-06, "loss": 0.4788, "step": 14507 }, { "epoch": 0.5168783511765859, "grad_norm": 2.3524216256579624, "learning_rate": 9.939418694055823e-06, "loss": 0.5564, "step": 14508 }, { "epoch": 0.51691397830308, "grad_norm": 1.581941459802687, "learning_rate": 9.938264778941674e-06, "loss": 0.5327, "step": 14509 }, { "epoch": 0.5169496054295741, "grad_norm": 1.2092028193664914, "learning_rate": 9.937110864649568e-06, "loss": 0.422, "step": 14510 }, { "epoch": 0.5169852325560682, "grad_norm": 1.2975848183530458, "learning_rate": 9.935956951194876e-06, "loss": 0.5245, "step": 14511 }, { "epoch": 0.5170208596825623, "grad_norm": 2.709359175230693, "learning_rate": 9.934803038592959e-06, "loss": 0.5944, "step": 14512 }, { "epoch": 0.5170564868090565, "grad_norm": 1.845985073752394, "learning_rate": 9.93364912685919e-06, "loss": 0.5289, "step": 14513 }, { "epoch": 0.5170921139355505, "grad_norm": 1.1347917683023114, "learning_rate": 9.932495216008926e-06, "loss": 0.4272, "step": 14514 }, { "epoch": 0.5171277410620446, "grad_norm": 1.2457026359416306, "learning_rate": 9.931341306057539e-06, "loss": 0.4629, "step": 14515 }, { "epoch": 0.5171633681885388, "grad_norm": 1.3023360459860602, "learning_rate": 9.930187397020385e-06, "loss": 0.4776, "step": 14516 }, { "epoch": 0.5171989953150329, "grad_norm": 1.2006777939382227, "learning_rate": 9.92903348891284e-06, "loss": 0.4114, "step": 14517 }, { "epoch": 0.517234622441527, "grad_norm": 1.354106756976815, "learning_rate": 9.92787958175026e-06, "loss": 0.5781, "step": 14518 }, { "epoch": 0.5172702495680211, "grad_norm": 1.4448098957934699, "learning_rate": 9.926725675548013e-06, "loss": 0.5277, "step": 14519 }, { "epoch": 0.5173058766945152, "grad_norm": 1.2180002979705615, "learning_rate": 9.925571770321469e-06, "loss": 0.4459, "step": 14520 }, { "epoch": 0.5173415038210093, "grad_norm": 1.062751189277197, "learning_rate": 9.924417866085985e-06, "loss": 0.4434, "step": 14521 }, { "epoch": 0.5173771309475035, "grad_norm": 1.7816962289793767, "learning_rate": 9.923263962856933e-06, "loss": 0.5893, "step": 14522 }, { "epoch": 0.5174127580739976, "grad_norm": 1.7773848414558633, "learning_rate": 9.922110060649672e-06, "loss": 0.5669, "step": 14523 }, { "epoch": 0.5174483852004916, "grad_norm": 1.4385643070662448, "learning_rate": 9.920956159479575e-06, "loss": 0.5235, "step": 14524 }, { "epoch": 0.5174840123269858, "grad_norm": 1.3020310077369641, "learning_rate": 9.919802259361995e-06, "loss": 0.4849, "step": 14525 }, { "epoch": 0.5175196394534799, "grad_norm": 1.1988282128372645, "learning_rate": 9.918648360312312e-06, "loss": 0.5007, "step": 14526 }, { "epoch": 0.517555266579974, "grad_norm": 1.1871401411350728, "learning_rate": 9.917494462345877e-06, "loss": 0.4758, "step": 14527 }, { "epoch": 0.5175908937064682, "grad_norm": 1.122797654260871, "learning_rate": 9.916340565478062e-06, "loss": 0.4378, "step": 14528 }, { "epoch": 0.5176265208329622, "grad_norm": 1.5100548144290546, "learning_rate": 9.91518666972423e-06, "loss": 0.5662, "step": 14529 }, { "epoch": 0.5176621479594563, "grad_norm": 0.9764572116330053, "learning_rate": 9.91403277509975e-06, "loss": 0.4341, "step": 14530 }, { "epoch": 0.5176977750859505, "grad_norm": 1.4960823571058952, "learning_rate": 9.91287888161998e-06, "loss": 0.5603, "step": 14531 }, { "epoch": 0.5177334022124446, "grad_norm": 1.3529129385675291, "learning_rate": 9.91172498930029e-06, "loss": 0.5347, "step": 14532 }, { "epoch": 0.5177690293389386, "grad_norm": 1.314949983109559, "learning_rate": 9.910571098156045e-06, "loss": 0.4803, "step": 14533 }, { "epoch": 0.5178046564654328, "grad_norm": 1.3268970494683743, "learning_rate": 9.909417208202604e-06, "loss": 0.5024, "step": 14534 }, { "epoch": 0.5178402835919269, "grad_norm": 1.5201875050822264, "learning_rate": 9.908263319455344e-06, "loss": 0.4653, "step": 14535 }, { "epoch": 0.517875910718421, "grad_norm": 1.0854826409998226, "learning_rate": 9.907109431929615e-06, "loss": 0.4322, "step": 14536 }, { "epoch": 0.5179115378449152, "grad_norm": 0.9890324136457526, "learning_rate": 9.905955545640791e-06, "loss": 0.4386, "step": 14537 }, { "epoch": 0.5179471649714092, "grad_norm": 1.7882161585836063, "learning_rate": 9.904801660604234e-06, "loss": 0.6048, "step": 14538 }, { "epoch": 0.5179827920979033, "grad_norm": 1.3176245429854592, "learning_rate": 9.903647776835312e-06, "loss": 0.5145, "step": 14539 }, { "epoch": 0.5180184192243975, "grad_norm": 1.23493599448999, "learning_rate": 9.902493894349383e-06, "loss": 0.4127, "step": 14540 }, { "epoch": 0.5180540463508916, "grad_norm": 1.1540909275448852, "learning_rate": 9.901340013161822e-06, "loss": 0.49, "step": 14541 }, { "epoch": 0.5180896734773857, "grad_norm": 1.905425356929341, "learning_rate": 9.900186133287982e-06, "loss": 0.6175, "step": 14542 }, { "epoch": 0.5181253006038798, "grad_norm": 1.0926975992769157, "learning_rate": 9.899032254743235e-06, "loss": 0.4713, "step": 14543 }, { "epoch": 0.5181609277303739, "grad_norm": 1.1855713375259778, "learning_rate": 9.897878377542944e-06, "loss": 0.4615, "step": 14544 }, { "epoch": 0.518196554856868, "grad_norm": 1.5125953730431028, "learning_rate": 9.896724501702474e-06, "loss": 0.58, "step": 14545 }, { "epoch": 0.5182321819833622, "grad_norm": 2.0877461560797492, "learning_rate": 9.895570627237191e-06, "loss": 0.5949, "step": 14546 }, { "epoch": 0.5182678091098563, "grad_norm": 1.538699792374894, "learning_rate": 9.894416754162458e-06, "loss": 0.6145, "step": 14547 }, { "epoch": 0.5183034362363503, "grad_norm": 1.7690319244843056, "learning_rate": 9.893262882493638e-06, "loss": 0.5344, "step": 14548 }, { "epoch": 0.5183390633628445, "grad_norm": 2.187104173719387, "learning_rate": 9.892109012246097e-06, "loss": 0.5623, "step": 14549 }, { "epoch": 0.5183746904893386, "grad_norm": 1.0982119625210278, "learning_rate": 9.890955143435203e-06, "loss": 0.451, "step": 14550 }, { "epoch": 0.5184103176158327, "grad_norm": 1.0806311991402038, "learning_rate": 9.889801276076312e-06, "loss": 0.4866, "step": 14551 }, { "epoch": 0.5184459447423269, "grad_norm": 1.3026461251596848, "learning_rate": 9.888647410184801e-06, "loss": 0.5787, "step": 14552 }, { "epoch": 0.5184815718688209, "grad_norm": 1.27302951013181, "learning_rate": 9.887493545776022e-06, "loss": 0.6114, "step": 14553 }, { "epoch": 0.518517198995315, "grad_norm": 1.857516958046295, "learning_rate": 9.886339682865345e-06, "loss": 0.6469, "step": 14554 }, { "epoch": 0.5185528261218092, "grad_norm": 1.554470139484431, "learning_rate": 9.885185821468138e-06, "loss": 0.5892, "step": 14555 }, { "epoch": 0.5185884532483033, "grad_norm": 1.6625016657052563, "learning_rate": 9.88403196159976e-06, "loss": 0.5878, "step": 14556 }, { "epoch": 0.5186240803747973, "grad_norm": 1.5125780500530701, "learning_rate": 9.882878103275581e-06, "loss": 0.5408, "step": 14557 }, { "epoch": 0.5186597075012915, "grad_norm": 1.1837480475348199, "learning_rate": 9.881724246510958e-06, "loss": 0.4494, "step": 14558 }, { "epoch": 0.5186953346277856, "grad_norm": 1.1399825326051167, "learning_rate": 9.88057039132126e-06, "loss": 0.429, "step": 14559 }, { "epoch": 0.5187309617542797, "grad_norm": 1.2338197872308916, "learning_rate": 9.879416537721849e-06, "loss": 0.5028, "step": 14560 }, { "epoch": 0.5187665888807739, "grad_norm": 1.1919783646026056, "learning_rate": 9.878262685728097e-06, "loss": 0.4608, "step": 14561 }, { "epoch": 0.518802216007268, "grad_norm": 1.410752019424662, "learning_rate": 9.877108835355357e-06, "loss": 0.5611, "step": 14562 }, { "epoch": 0.518837843133762, "grad_norm": 1.1051703201210532, "learning_rate": 9.875954986619e-06, "loss": 0.5125, "step": 14563 }, { "epoch": 0.5188734702602562, "grad_norm": 1.0316344059654778, "learning_rate": 9.874801139534389e-06, "loss": 0.4551, "step": 14564 }, { "epoch": 0.5189090973867503, "grad_norm": 1.0996012066062555, "learning_rate": 9.873647294116888e-06, "loss": 0.4706, "step": 14565 }, { "epoch": 0.5189447245132444, "grad_norm": 1.804684371418538, "learning_rate": 9.872493450381863e-06, "loss": 0.5033, "step": 14566 }, { "epoch": 0.5189803516397385, "grad_norm": 1.0141643234213928, "learning_rate": 9.871339608344675e-06, "loss": 0.45, "step": 14567 }, { "epoch": 0.5190159787662326, "grad_norm": 1.9450373600503763, "learning_rate": 9.870185768020694e-06, "loss": 0.6027, "step": 14568 }, { "epoch": 0.5190516058927267, "grad_norm": 1.4508690578429246, "learning_rate": 9.869031929425275e-06, "loss": 0.5057, "step": 14569 }, { "epoch": 0.5190872330192209, "grad_norm": 1.6706001025117494, "learning_rate": 9.86787809257379e-06, "loss": 0.5902, "step": 14570 }, { "epoch": 0.519122860145715, "grad_norm": 1.8063672265120494, "learning_rate": 9.866724257481601e-06, "loss": 0.5139, "step": 14571 }, { "epoch": 0.519158487272209, "grad_norm": 1.2157455753528796, "learning_rate": 9.865570424164073e-06, "loss": 0.4925, "step": 14572 }, { "epoch": 0.5191941143987032, "grad_norm": 1.1195091826924692, "learning_rate": 9.864416592636565e-06, "loss": 0.4384, "step": 14573 }, { "epoch": 0.5192297415251973, "grad_norm": 1.5161544608437556, "learning_rate": 9.863262762914448e-06, "loss": 0.5602, "step": 14574 }, { "epoch": 0.5192653686516914, "grad_norm": 1.8923926485390068, "learning_rate": 9.862108935013079e-06, "loss": 0.45, "step": 14575 }, { "epoch": 0.5193009957781856, "grad_norm": 1.1378015733180962, "learning_rate": 9.860955108947832e-06, "loss": 0.5318, "step": 14576 }, { "epoch": 0.5193366229046796, "grad_norm": 0.9876196193806782, "learning_rate": 9.859801284734057e-06, "loss": 0.4145, "step": 14577 }, { "epoch": 0.5193722500311737, "grad_norm": 1.7379013083616355, "learning_rate": 9.85864746238713e-06, "loss": 0.547, "step": 14578 }, { "epoch": 0.5194078771576679, "grad_norm": 0.9892064183542751, "learning_rate": 9.857493641922414e-06, "loss": 0.4721, "step": 14579 }, { "epoch": 0.519443504284162, "grad_norm": 1.0869304310558647, "learning_rate": 9.856339823355265e-06, "loss": 0.4475, "step": 14580 }, { "epoch": 0.519479131410656, "grad_norm": 1.0378296075373343, "learning_rate": 9.855186006701054e-06, "loss": 0.404, "step": 14581 }, { "epoch": 0.5195147585371502, "grad_norm": 1.4257178188720265, "learning_rate": 9.85403219197514e-06, "loss": 0.6102, "step": 14582 }, { "epoch": 0.5195503856636443, "grad_norm": 1.8120326238902837, "learning_rate": 9.852878379192895e-06, "loss": 0.5342, "step": 14583 }, { "epoch": 0.5195860127901384, "grad_norm": 1.5412143058344114, "learning_rate": 9.851724568369673e-06, "loss": 0.5494, "step": 14584 }, { "epoch": 0.5196216399166326, "grad_norm": 2.2169924435688313, "learning_rate": 9.850570759520844e-06, "loss": 0.5855, "step": 14585 }, { "epoch": 0.5196572670431266, "grad_norm": 1.3514519647025764, "learning_rate": 9.849416952661765e-06, "loss": 0.5588, "step": 14586 }, { "epoch": 0.5196928941696207, "grad_norm": 1.0903879267441483, "learning_rate": 9.848263147807812e-06, "loss": 0.4076, "step": 14587 }, { "epoch": 0.5197285212961149, "grad_norm": 1.1109050407994323, "learning_rate": 9.847109344974333e-06, "loss": 0.4621, "step": 14588 }, { "epoch": 0.519764148422609, "grad_norm": 1.1357128957108193, "learning_rate": 9.845955544176704e-06, "loss": 0.5064, "step": 14589 }, { "epoch": 0.5197997755491031, "grad_norm": 1.5144923419931422, "learning_rate": 9.844801745430287e-06, "loss": 0.5877, "step": 14590 }, { "epoch": 0.5198354026755972, "grad_norm": 1.2844558221862457, "learning_rate": 9.843647948750439e-06, "loss": 0.4674, "step": 14591 }, { "epoch": 0.5198710298020913, "grad_norm": 1.5163028278958606, "learning_rate": 9.84249415415253e-06, "loss": 0.5368, "step": 14592 }, { "epoch": 0.5199066569285854, "grad_norm": 1.1010617268369656, "learning_rate": 9.841340361651921e-06, "loss": 0.4594, "step": 14593 }, { "epoch": 0.5199422840550796, "grad_norm": 1.7133887737439903, "learning_rate": 9.84018657126398e-06, "loss": 0.5908, "step": 14594 }, { "epoch": 0.5199779111815737, "grad_norm": 1.1150479198533416, "learning_rate": 9.83903278300406e-06, "loss": 0.4538, "step": 14595 }, { "epoch": 0.5200135383080677, "grad_norm": 1.2781553038869369, "learning_rate": 9.837878996887534e-06, "loss": 0.4149, "step": 14596 }, { "epoch": 0.5200491654345619, "grad_norm": 1.6737941201017463, "learning_rate": 9.836725212929762e-06, "loss": 0.5785, "step": 14597 }, { "epoch": 0.520084792561056, "grad_norm": 1.3090735465396863, "learning_rate": 9.83557143114611e-06, "loss": 0.4897, "step": 14598 }, { "epoch": 0.5201204196875501, "grad_norm": 1.4380244208058244, "learning_rate": 9.834417651551936e-06, "loss": 0.5903, "step": 14599 }, { "epoch": 0.5201560468140443, "grad_norm": 1.3066311041126368, "learning_rate": 9.833263874162604e-06, "loss": 0.5711, "step": 14600 }, { "epoch": 0.5201916739405383, "grad_norm": 1.6059463865024013, "learning_rate": 9.832110098993489e-06, "loss": 0.5064, "step": 14601 }, { "epoch": 0.5202273010670324, "grad_norm": 1.3144825691118103, "learning_rate": 9.830956326059938e-06, "loss": 0.5494, "step": 14602 }, { "epoch": 0.5202629281935266, "grad_norm": 1.5231522040976246, "learning_rate": 9.829802555377326e-06, "loss": 0.4814, "step": 14603 }, { "epoch": 0.5202985553200207, "grad_norm": 2.4096663349355367, "learning_rate": 9.828648786961009e-06, "loss": 0.5406, "step": 14604 }, { "epoch": 0.5203341824465147, "grad_norm": 1.0323399034327199, "learning_rate": 9.827495020826357e-06, "loss": 0.3971, "step": 14605 }, { "epoch": 0.5203698095730089, "grad_norm": 1.494880456500451, "learning_rate": 9.826341256988725e-06, "loss": 0.5895, "step": 14606 }, { "epoch": 0.520405436699503, "grad_norm": 1.5613347844123422, "learning_rate": 9.825187495463482e-06, "loss": 0.5734, "step": 14607 }, { "epoch": 0.5204410638259971, "grad_norm": 1.1313846146213073, "learning_rate": 9.82403373626599e-06, "loss": 0.4884, "step": 14608 }, { "epoch": 0.5204766909524913, "grad_norm": 1.1825069927883445, "learning_rate": 9.822879979411613e-06, "loss": 0.4402, "step": 14609 }, { "epoch": 0.5205123180789853, "grad_norm": 1.1062415497595184, "learning_rate": 9.82172622491571e-06, "loss": 0.4985, "step": 14610 }, { "epoch": 0.5205479452054794, "grad_norm": 1.4280435117070749, "learning_rate": 9.820572472793651e-06, "loss": 0.5166, "step": 14611 }, { "epoch": 0.5205835723319736, "grad_norm": 1.1173883571933434, "learning_rate": 9.81941872306079e-06, "loss": 0.5336, "step": 14612 }, { "epoch": 0.5206191994584677, "grad_norm": 1.0921730391564972, "learning_rate": 9.818264975732497e-06, "loss": 0.4485, "step": 14613 }, { "epoch": 0.5206548265849618, "grad_norm": 1.0608121212487718, "learning_rate": 9.817111230824133e-06, "loss": 0.4676, "step": 14614 }, { "epoch": 0.520690453711456, "grad_norm": 1.0870960267720522, "learning_rate": 9.81595748835106e-06, "loss": 0.436, "step": 14615 }, { "epoch": 0.52072608083795, "grad_norm": 1.5549198585873274, "learning_rate": 9.814803748328646e-06, "loss": 0.5399, "step": 14616 }, { "epoch": 0.5207617079644441, "grad_norm": 1.532856796875703, "learning_rate": 9.813650010772244e-06, "loss": 0.6147, "step": 14617 }, { "epoch": 0.5207973350909383, "grad_norm": 1.0350748587568759, "learning_rate": 9.812496275697226e-06, "loss": 0.4367, "step": 14618 }, { "epoch": 0.5208329622174324, "grad_norm": 1.5657915893336094, "learning_rate": 9.811342543118947e-06, "loss": 0.5429, "step": 14619 }, { "epoch": 0.5208685893439264, "grad_norm": 1.064125388714315, "learning_rate": 9.81018881305278e-06, "loss": 0.4303, "step": 14620 }, { "epoch": 0.5209042164704206, "grad_norm": 1.3270139550645632, "learning_rate": 9.809035085514078e-06, "loss": 0.4321, "step": 14621 }, { "epoch": 0.5209398435969147, "grad_norm": 0.9785765169567752, "learning_rate": 9.80788136051821e-06, "loss": 0.4009, "step": 14622 }, { "epoch": 0.5209754707234088, "grad_norm": 1.0652338333534885, "learning_rate": 9.80672763808053e-06, "loss": 0.4072, "step": 14623 }, { "epoch": 0.521011097849903, "grad_norm": 1.1340969921686261, "learning_rate": 9.80557391821641e-06, "loss": 0.4933, "step": 14624 }, { "epoch": 0.521046724976397, "grad_norm": 1.126426496357593, "learning_rate": 9.804420200941211e-06, "loss": 0.4718, "step": 14625 }, { "epoch": 0.5210823521028911, "grad_norm": 1.0934159638771306, "learning_rate": 9.803266486270289e-06, "loss": 0.4984, "step": 14626 }, { "epoch": 0.5211179792293853, "grad_norm": 1.6663511203739236, "learning_rate": 9.802112774219017e-06, "loss": 0.5595, "step": 14627 }, { "epoch": 0.5211536063558794, "grad_norm": 2.142075573872773, "learning_rate": 9.800959064802748e-06, "loss": 0.5306, "step": 14628 }, { "epoch": 0.5211892334823734, "grad_norm": 1.1388915291205204, "learning_rate": 9.79980535803685e-06, "loss": 0.4741, "step": 14629 }, { "epoch": 0.5212248606088676, "grad_norm": 1.1513063375574653, "learning_rate": 9.798651653936682e-06, "loss": 0.4423, "step": 14630 }, { "epoch": 0.5212604877353617, "grad_norm": 1.110097677414455, "learning_rate": 9.79749795251761e-06, "loss": 0.4061, "step": 14631 }, { "epoch": 0.5212961148618558, "grad_norm": 1.4342588762756638, "learning_rate": 9.796344253794993e-06, "loss": 0.565, "step": 14632 }, { "epoch": 0.52133174198835, "grad_norm": 1.1726268905078219, "learning_rate": 9.795190557784196e-06, "loss": 0.4059, "step": 14633 }, { "epoch": 0.521367369114844, "grad_norm": 1.4221877218975802, "learning_rate": 9.794036864500577e-06, "loss": 0.586, "step": 14634 }, { "epoch": 0.5214029962413381, "grad_norm": 1.471645756810762, "learning_rate": 9.7928831739595e-06, "loss": 0.5959, "step": 14635 }, { "epoch": 0.5214386233678323, "grad_norm": 1.0379230836265232, "learning_rate": 9.791729486176335e-06, "loss": 0.4052, "step": 14636 }, { "epoch": 0.5214742504943264, "grad_norm": 1.4406428765848829, "learning_rate": 9.790575801166432e-06, "loss": 0.5469, "step": 14637 }, { "epoch": 0.5215098776208205, "grad_norm": 1.1252585523740635, "learning_rate": 9.78942211894516e-06, "loss": 0.4846, "step": 14638 }, { "epoch": 0.5215455047473146, "grad_norm": 1.0926573354267237, "learning_rate": 9.788268439527878e-06, "loss": 0.409, "step": 14639 }, { "epoch": 0.5215811318738087, "grad_norm": 1.3586618753626838, "learning_rate": 9.787114762929953e-06, "loss": 0.613, "step": 14640 }, { "epoch": 0.5216167590003028, "grad_norm": 1.6160162198587342, "learning_rate": 9.78596108916674e-06, "loss": 0.4901, "step": 14641 }, { "epoch": 0.521652386126797, "grad_norm": 1.0494637268948497, "learning_rate": 9.78480741825361e-06, "loss": 0.4437, "step": 14642 }, { "epoch": 0.5216880132532911, "grad_norm": 1.1516333868900692, "learning_rate": 9.783653750205916e-06, "loss": 0.4095, "step": 14643 }, { "epoch": 0.5217236403797851, "grad_norm": 1.5791835699816603, "learning_rate": 9.782500085039025e-06, "loss": 0.6081, "step": 14644 }, { "epoch": 0.5217592675062793, "grad_norm": 1.4594344857284092, "learning_rate": 9.781346422768294e-06, "loss": 0.5951, "step": 14645 }, { "epoch": 0.5217948946327734, "grad_norm": 2.2147105040619373, "learning_rate": 9.780192763409089e-06, "loss": 0.5306, "step": 14646 }, { "epoch": 0.5218305217592675, "grad_norm": 2.0465969709487983, "learning_rate": 9.779039106976776e-06, "loss": 0.5881, "step": 14647 }, { "epoch": 0.5218661488857617, "grad_norm": 1.0870387755726652, "learning_rate": 9.777885453486706e-06, "loss": 0.4035, "step": 14648 }, { "epoch": 0.5219017760122557, "grad_norm": 1.1152457988087325, "learning_rate": 9.77673180295425e-06, "loss": 0.3927, "step": 14649 }, { "epoch": 0.5219374031387498, "grad_norm": 1.1975560960960823, "learning_rate": 9.775578155394764e-06, "loss": 0.4617, "step": 14650 }, { "epoch": 0.521973030265244, "grad_norm": 1.1517555745138854, "learning_rate": 9.774424510823615e-06, "loss": 0.5053, "step": 14651 }, { "epoch": 0.5220086573917381, "grad_norm": 1.7521673011552024, "learning_rate": 9.773270869256156e-06, "loss": 0.582, "step": 14652 }, { "epoch": 0.5220442845182321, "grad_norm": 1.90381644576267, "learning_rate": 9.772117230707759e-06, "loss": 0.5992, "step": 14653 }, { "epoch": 0.5220799116447263, "grad_norm": 1.0997715678607278, "learning_rate": 9.770963595193778e-06, "loss": 0.4419, "step": 14654 }, { "epoch": 0.5221155387712204, "grad_norm": 1.279846201756615, "learning_rate": 9.769809962729578e-06, "loss": 0.4751, "step": 14655 }, { "epoch": 0.5221511658977145, "grad_norm": 1.1364977763553967, "learning_rate": 9.768656333330516e-06, "loss": 0.4586, "step": 14656 }, { "epoch": 0.5221867930242087, "grad_norm": 1.2400267061574752, "learning_rate": 9.767502707011962e-06, "loss": 0.4406, "step": 14657 }, { "epoch": 0.5222224201507027, "grad_norm": 1.1180549506846982, "learning_rate": 9.766349083789266e-06, "loss": 0.4561, "step": 14658 }, { "epoch": 0.5222580472771968, "grad_norm": 1.29919197764252, "learning_rate": 9.765195463677796e-06, "loss": 0.4675, "step": 14659 }, { "epoch": 0.522293674403691, "grad_norm": 1.567050162132586, "learning_rate": 9.764041846692917e-06, "loss": 0.5447, "step": 14660 }, { "epoch": 0.5223293015301851, "grad_norm": 1.3198976032878977, "learning_rate": 9.76288823284998e-06, "loss": 0.4568, "step": 14661 }, { "epoch": 0.5223649286566792, "grad_norm": 1.0930668826376064, "learning_rate": 9.76173462216436e-06, "loss": 0.4723, "step": 14662 }, { "epoch": 0.5224005557831733, "grad_norm": 1.6048841100652587, "learning_rate": 9.760581014651404e-06, "loss": 0.6511, "step": 14663 }, { "epoch": 0.5224361829096674, "grad_norm": 1.0797900725467564, "learning_rate": 9.75942741032648e-06, "loss": 0.4708, "step": 14664 }, { "epoch": 0.5224718100361615, "grad_norm": 1.6597161669947518, "learning_rate": 9.758273809204948e-06, "loss": 0.5043, "step": 14665 }, { "epoch": 0.5225074371626557, "grad_norm": 1.1161307729411185, "learning_rate": 9.75712021130217e-06, "loss": 0.4179, "step": 14666 }, { "epoch": 0.5225430642891498, "grad_norm": 1.7550481997213667, "learning_rate": 9.755966616633504e-06, "loss": 0.5614, "step": 14667 }, { "epoch": 0.5225786914156438, "grad_norm": 1.1824252706647498, "learning_rate": 9.754813025214318e-06, "loss": 0.4788, "step": 14668 }, { "epoch": 0.522614318542138, "grad_norm": 1.1347832865272058, "learning_rate": 9.753659437059964e-06, "loss": 0.4527, "step": 14669 }, { "epoch": 0.5226499456686321, "grad_norm": 1.341110608578809, "learning_rate": 9.752505852185807e-06, "loss": 0.5398, "step": 14670 }, { "epoch": 0.5226855727951262, "grad_norm": 1.0198689309735625, "learning_rate": 9.751352270607208e-06, "loss": 0.4313, "step": 14671 }, { "epoch": 0.5227211999216204, "grad_norm": 1.2994120488820904, "learning_rate": 9.750198692339527e-06, "loss": 0.5298, "step": 14672 }, { "epoch": 0.5227568270481144, "grad_norm": 1.4672689473426916, "learning_rate": 9.749045117398129e-06, "loss": 0.6005, "step": 14673 }, { "epoch": 0.5227924541746085, "grad_norm": 1.0584676256945853, "learning_rate": 9.747891545798365e-06, "loss": 0.4292, "step": 14674 }, { "epoch": 0.5228280813011027, "grad_norm": 1.0424454965941878, "learning_rate": 9.746737977555604e-06, "loss": 0.4448, "step": 14675 }, { "epoch": 0.5228637084275968, "grad_norm": 1.8755148609967853, "learning_rate": 9.745584412685202e-06, "loss": 0.583, "step": 14676 }, { "epoch": 0.5228993355540908, "grad_norm": 1.5421219340264887, "learning_rate": 9.744430851202523e-06, "loss": 0.5939, "step": 14677 }, { "epoch": 0.522934962680585, "grad_norm": 1.5051188381534284, "learning_rate": 9.743277293122925e-06, "loss": 0.558, "step": 14678 }, { "epoch": 0.5229705898070791, "grad_norm": 1.2153650493019579, "learning_rate": 9.742123738461773e-06, "loss": 0.473, "step": 14679 }, { "epoch": 0.5230062169335732, "grad_norm": 1.0233172450079902, "learning_rate": 9.740970187234419e-06, "loss": 0.4619, "step": 14680 }, { "epoch": 0.5230418440600674, "grad_norm": 1.8752039694244864, "learning_rate": 9.739816639456227e-06, "loss": 0.5922, "step": 14681 }, { "epoch": 0.5230774711865614, "grad_norm": 1.213030350165897, "learning_rate": 9.738663095142562e-06, "loss": 0.5023, "step": 14682 }, { "epoch": 0.5231130983130555, "grad_norm": 2.741855042098423, "learning_rate": 9.737509554308778e-06, "loss": 0.4593, "step": 14683 }, { "epoch": 0.5231487254395497, "grad_norm": 1.1175267286951756, "learning_rate": 9.736356016970243e-06, "loss": 0.4746, "step": 14684 }, { "epoch": 0.5231843525660438, "grad_norm": 1.7121953499996572, "learning_rate": 9.735202483142306e-06, "loss": 0.6102, "step": 14685 }, { "epoch": 0.5232199796925379, "grad_norm": 1.5812908718702596, "learning_rate": 9.734048952840335e-06, "loss": 0.4989, "step": 14686 }, { "epoch": 0.523255606819032, "grad_norm": 1.083874138219982, "learning_rate": 9.732895426079688e-06, "loss": 0.435, "step": 14687 }, { "epoch": 0.5232912339455261, "grad_norm": 1.0662131669915702, "learning_rate": 9.731741902875728e-06, "loss": 0.4611, "step": 14688 }, { "epoch": 0.5233268610720202, "grad_norm": 1.5131852250937061, "learning_rate": 9.730588383243806e-06, "loss": 0.5526, "step": 14689 }, { "epoch": 0.5233624881985144, "grad_norm": 1.0539533383614805, "learning_rate": 9.729434867199294e-06, "loss": 0.5126, "step": 14690 }, { "epoch": 0.5233981153250085, "grad_norm": 1.4430155270844003, "learning_rate": 9.72828135475754e-06, "loss": 0.5602, "step": 14691 }, { "epoch": 0.5234337424515025, "grad_norm": 1.4257216484617714, "learning_rate": 9.727127845933915e-06, "loss": 0.5246, "step": 14692 }, { "epoch": 0.5234693695779967, "grad_norm": 1.3191180313872304, "learning_rate": 9.725974340743769e-06, "loss": 0.5192, "step": 14693 }, { "epoch": 0.5235049967044908, "grad_norm": 2.4990302840642307, "learning_rate": 9.724820839202468e-06, "loss": 0.5156, "step": 14694 }, { "epoch": 0.5235406238309849, "grad_norm": 1.156897167641611, "learning_rate": 9.723667341325373e-06, "loss": 0.4169, "step": 14695 }, { "epoch": 0.5235762509574791, "grad_norm": 1.568603199747232, "learning_rate": 9.722513847127836e-06, "loss": 0.5418, "step": 14696 }, { "epoch": 0.5236118780839731, "grad_norm": 1.2667243118080465, "learning_rate": 9.721360356625223e-06, "loss": 0.443, "step": 14697 }, { "epoch": 0.5236475052104672, "grad_norm": 1.1305336137487914, "learning_rate": 9.72020686983289e-06, "loss": 0.4485, "step": 14698 }, { "epoch": 0.5236831323369614, "grad_norm": 1.0796899821720052, "learning_rate": 9.719053386766204e-06, "loss": 0.4635, "step": 14699 }, { "epoch": 0.5237187594634555, "grad_norm": 1.0879742874926095, "learning_rate": 9.717899907440513e-06, "loss": 0.4548, "step": 14700 }, { "epoch": 0.5237543865899495, "grad_norm": 1.3254887508687796, "learning_rate": 9.716746431871184e-06, "loss": 0.5307, "step": 14701 }, { "epoch": 0.5237900137164437, "grad_norm": 2.7531252263553037, "learning_rate": 9.71559296007357e-06, "loss": 0.5406, "step": 14702 }, { "epoch": 0.5238256408429378, "grad_norm": 1.2637635306260733, "learning_rate": 9.71443949206304e-06, "loss": 0.555, "step": 14703 }, { "epoch": 0.5238612679694319, "grad_norm": 1.1821533989223787, "learning_rate": 9.713286027854943e-06, "loss": 0.452, "step": 14704 }, { "epoch": 0.5238968950959261, "grad_norm": 1.9278456828336699, "learning_rate": 9.712132567464645e-06, "loss": 0.5627, "step": 14705 }, { "epoch": 0.5239325222224201, "grad_norm": 1.380570115173822, "learning_rate": 9.710979110907506e-06, "loss": 0.5164, "step": 14706 }, { "epoch": 0.5239681493489142, "grad_norm": 1.765406136734872, "learning_rate": 9.709825658198878e-06, "loss": 0.5649, "step": 14707 }, { "epoch": 0.5240037764754084, "grad_norm": 1.2304241541502965, "learning_rate": 9.708672209354128e-06, "loss": 0.4404, "step": 14708 }, { "epoch": 0.5240394036019025, "grad_norm": 1.4720699219062727, "learning_rate": 9.707518764388607e-06, "loss": 0.5149, "step": 14709 }, { "epoch": 0.5240750307283966, "grad_norm": 2.3482288203533153, "learning_rate": 9.706365323317684e-06, "loss": 0.6045, "step": 14710 }, { "epoch": 0.5241106578548907, "grad_norm": 1.4050386833994575, "learning_rate": 9.705211886156706e-06, "loss": 0.5188, "step": 14711 }, { "epoch": 0.5241462849813848, "grad_norm": 2.0416612842932502, "learning_rate": 9.70405845292104e-06, "loss": 0.5722, "step": 14712 }, { "epoch": 0.5241819121078789, "grad_norm": 1.0327192802522462, "learning_rate": 9.702905023626041e-06, "loss": 0.4667, "step": 14713 }, { "epoch": 0.5242175392343731, "grad_norm": 1.0723052762669874, "learning_rate": 9.701751598287074e-06, "loss": 0.4704, "step": 14714 }, { "epoch": 0.5242531663608672, "grad_norm": 1.1401898517497697, "learning_rate": 9.700598176919485e-06, "loss": 0.4501, "step": 14715 }, { "epoch": 0.5242887934873612, "grad_norm": 1.1260140829326613, "learning_rate": 9.699444759538644e-06, "loss": 0.478, "step": 14716 }, { "epoch": 0.5243244206138554, "grad_norm": 1.5198418472474697, "learning_rate": 9.698291346159911e-06, "loss": 0.5404, "step": 14717 }, { "epoch": 0.5243600477403495, "grad_norm": 1.8621362360881881, "learning_rate": 9.697137936798635e-06, "loss": 0.5574, "step": 14718 }, { "epoch": 0.5243956748668436, "grad_norm": 1.417377876367988, "learning_rate": 9.69598453147018e-06, "loss": 0.4964, "step": 14719 }, { "epoch": 0.5244313019933378, "grad_norm": 1.4458557167148538, "learning_rate": 9.694831130189903e-06, "loss": 0.6659, "step": 14720 }, { "epoch": 0.5244669291198318, "grad_norm": 1.3620092153835053, "learning_rate": 9.693677732973166e-06, "loss": 0.4246, "step": 14721 }, { "epoch": 0.5245025562463259, "grad_norm": 1.1473573995650845, "learning_rate": 9.69252433983532e-06, "loss": 0.4732, "step": 14722 }, { "epoch": 0.5245381833728201, "grad_norm": 1.0368005053721183, "learning_rate": 9.691370950791728e-06, "loss": 0.4087, "step": 14723 }, { "epoch": 0.5245738104993142, "grad_norm": 1.7600642162454123, "learning_rate": 9.690217565857747e-06, "loss": 0.6232, "step": 14724 }, { "epoch": 0.5246094376258083, "grad_norm": 1.420710408841271, "learning_rate": 9.68906418504874e-06, "loss": 0.525, "step": 14725 }, { "epoch": 0.5246450647523024, "grad_norm": 1.1881337269224495, "learning_rate": 9.687910808380057e-06, "loss": 0.437, "step": 14726 }, { "epoch": 0.5246806918787965, "grad_norm": 1.5593941904113233, "learning_rate": 9.686757435867055e-06, "loss": 0.4415, "step": 14727 }, { "epoch": 0.5247163190052906, "grad_norm": 1.5858299929408683, "learning_rate": 9.685604067525106e-06, "loss": 0.5913, "step": 14728 }, { "epoch": 0.5247519461317848, "grad_norm": 1.0704816401015713, "learning_rate": 9.684450703369553e-06, "loss": 0.501, "step": 14729 }, { "epoch": 0.5247875732582788, "grad_norm": 1.4272441220369252, "learning_rate": 9.683297343415762e-06, "loss": 0.5161, "step": 14730 }, { "epoch": 0.5248232003847729, "grad_norm": 1.6799540383053548, "learning_rate": 9.682143987679086e-06, "loss": 0.5723, "step": 14731 }, { "epoch": 0.5248588275112671, "grad_norm": 1.6296473990764426, "learning_rate": 9.680990636174888e-06, "loss": 0.4947, "step": 14732 }, { "epoch": 0.5248944546377612, "grad_norm": 1.1625643764269324, "learning_rate": 9.67983728891852e-06, "loss": 0.4715, "step": 14733 }, { "epoch": 0.5249300817642553, "grad_norm": 2.0384700903374453, "learning_rate": 9.678683945925343e-06, "loss": 0.4999, "step": 14734 }, { "epoch": 0.5249657088907494, "grad_norm": 1.391624607803096, "learning_rate": 9.677530607210714e-06, "loss": 0.5638, "step": 14735 }, { "epoch": 0.5250013360172435, "grad_norm": 1.2222257974104813, "learning_rate": 9.676377272789991e-06, "loss": 0.4856, "step": 14736 }, { "epoch": 0.5250369631437376, "grad_norm": 1.1775799445314479, "learning_rate": 9.67522394267853e-06, "loss": 0.4328, "step": 14737 }, { "epoch": 0.5250725902702318, "grad_norm": 1.100845325813647, "learning_rate": 9.67407061689169e-06, "loss": 0.4377, "step": 14738 }, { "epoch": 0.5251082173967259, "grad_norm": 1.4352126214267371, "learning_rate": 9.672917295444826e-06, "loss": 0.4984, "step": 14739 }, { "epoch": 0.5251438445232199, "grad_norm": 2.1650935870517096, "learning_rate": 9.671763978353296e-06, "loss": 0.5952, "step": 14740 }, { "epoch": 0.5251794716497141, "grad_norm": 1.3459956976081884, "learning_rate": 9.67061066563246e-06, "loss": 0.6228, "step": 14741 }, { "epoch": 0.5252150987762082, "grad_norm": 1.9204436655363564, "learning_rate": 9.66945735729767e-06, "loss": 0.6653, "step": 14742 }, { "epoch": 0.5252507259027023, "grad_norm": 1.413678672300789, "learning_rate": 9.668304053364293e-06, "loss": 0.5895, "step": 14743 }, { "epoch": 0.5252863530291965, "grad_norm": 1.0514793477769004, "learning_rate": 9.667150753847676e-06, "loss": 0.407, "step": 14744 }, { "epoch": 0.5253219801556905, "grad_norm": 1.2008901160512477, "learning_rate": 9.665997458763179e-06, "loss": 0.4963, "step": 14745 }, { "epoch": 0.5253576072821846, "grad_norm": 1.2389832998403798, "learning_rate": 9.664844168126157e-06, "loss": 0.5026, "step": 14746 }, { "epoch": 0.5253932344086788, "grad_norm": 1.1082495455827033, "learning_rate": 9.663690881951975e-06, "loss": 0.4518, "step": 14747 }, { "epoch": 0.5254288615351729, "grad_norm": 1.0596159243418855, "learning_rate": 9.662537600255979e-06, "loss": 0.4468, "step": 14748 }, { "epoch": 0.525464488661667, "grad_norm": 1.0948844259043438, "learning_rate": 9.661384323053533e-06, "loss": 0.4425, "step": 14749 }, { "epoch": 0.5255001157881611, "grad_norm": 1.2725938378951769, "learning_rate": 9.660231050359989e-06, "loss": 0.4959, "step": 14750 }, { "epoch": 0.5255357429146552, "grad_norm": 1.0431802476829652, "learning_rate": 9.659077782190707e-06, "loss": 0.4476, "step": 14751 }, { "epoch": 0.5255713700411493, "grad_norm": 1.1907079837070738, "learning_rate": 9.657924518561046e-06, "loss": 0.4582, "step": 14752 }, { "epoch": 0.5256069971676435, "grad_norm": 1.1508447353709126, "learning_rate": 9.656771259486352e-06, "loss": 0.4461, "step": 14753 }, { "epoch": 0.5256426242941376, "grad_norm": 1.4787587804972806, "learning_rate": 9.655618004981996e-06, "loss": 0.5658, "step": 14754 }, { "epoch": 0.5256782514206316, "grad_norm": 2.0030929665232233, "learning_rate": 9.654464755063323e-06, "loss": 0.4579, "step": 14755 }, { "epoch": 0.5257138785471258, "grad_norm": 1.618597717057662, "learning_rate": 9.653311509745694e-06, "loss": 0.6435, "step": 14756 }, { "epoch": 0.5257495056736199, "grad_norm": 1.6007578687454898, "learning_rate": 9.652158269044463e-06, "loss": 0.5072, "step": 14757 }, { "epoch": 0.525785132800114, "grad_norm": 1.1228315093125054, "learning_rate": 9.651005032974994e-06, "loss": 0.4921, "step": 14758 }, { "epoch": 0.5258207599266082, "grad_norm": 1.4026497199192145, "learning_rate": 9.64985180155263e-06, "loss": 0.4762, "step": 14759 }, { "epoch": 0.5258563870531022, "grad_norm": 1.3438428492363963, "learning_rate": 9.648698574792737e-06, "loss": 0.4644, "step": 14760 }, { "epoch": 0.5258920141795963, "grad_norm": 1.0807903035754023, "learning_rate": 9.647545352710667e-06, "loss": 0.4373, "step": 14761 }, { "epoch": 0.5259276413060905, "grad_norm": 1.5707088131813405, "learning_rate": 9.646392135321776e-06, "loss": 0.5528, "step": 14762 }, { "epoch": 0.5259632684325846, "grad_norm": 1.1689186775188256, "learning_rate": 9.645238922641425e-06, "loss": 0.4687, "step": 14763 }, { "epoch": 0.5259988955590786, "grad_norm": 1.4726676319777063, "learning_rate": 9.64408571468496e-06, "loss": 0.5532, "step": 14764 }, { "epoch": 0.5260345226855728, "grad_norm": 1.6411643108679055, "learning_rate": 9.642932511467748e-06, "loss": 0.543, "step": 14765 }, { "epoch": 0.5260701498120669, "grad_norm": 1.911032038700837, "learning_rate": 9.641779313005136e-06, "loss": 0.5672, "step": 14766 }, { "epoch": 0.526105776938561, "grad_norm": 1.117579878811281, "learning_rate": 9.640626119312483e-06, "loss": 0.4208, "step": 14767 }, { "epoch": 0.5261414040650552, "grad_norm": 1.1000636080420994, "learning_rate": 9.639472930405143e-06, "loss": 0.4661, "step": 14768 }, { "epoch": 0.5261770311915492, "grad_norm": 1.8149158931172322, "learning_rate": 9.638319746298479e-06, "loss": 0.5392, "step": 14769 }, { "epoch": 0.5262126583180433, "grad_norm": 0.9766855396451788, "learning_rate": 9.637166567007834e-06, "loss": 0.4105, "step": 14770 }, { "epoch": 0.5262482854445375, "grad_norm": 1.0911938632559988, "learning_rate": 9.636013392548572e-06, "loss": 0.4967, "step": 14771 }, { "epoch": 0.5262839125710316, "grad_norm": 1.1806034505166119, "learning_rate": 9.634860222936047e-06, "loss": 0.4373, "step": 14772 }, { "epoch": 0.5263195396975257, "grad_norm": 1.4865850135513772, "learning_rate": 9.63370705818561e-06, "loss": 0.4855, "step": 14773 }, { "epoch": 0.5263551668240198, "grad_norm": 1.6257191782807583, "learning_rate": 9.632553898312625e-06, "loss": 0.489, "step": 14774 }, { "epoch": 0.5263907939505139, "grad_norm": 1.1227769858491687, "learning_rate": 9.631400743332437e-06, "loss": 0.425, "step": 14775 }, { "epoch": 0.526426421077008, "grad_norm": 1.825471058262992, "learning_rate": 9.630247593260409e-06, "loss": 0.5149, "step": 14776 }, { "epoch": 0.5264620482035022, "grad_norm": 1.0201953215718278, "learning_rate": 9.62909444811189e-06, "loss": 0.4481, "step": 14777 }, { "epoch": 0.5264976753299963, "grad_norm": 1.5164800321151954, "learning_rate": 9.627941307902238e-06, "loss": 0.5427, "step": 14778 }, { "epoch": 0.5265333024564903, "grad_norm": 1.7054311030693348, "learning_rate": 9.626788172646809e-06, "loss": 0.6405, "step": 14779 }, { "epoch": 0.5265689295829845, "grad_norm": 1.3409140666164896, "learning_rate": 9.625635042360958e-06, "loss": 0.5532, "step": 14780 }, { "epoch": 0.5266045567094786, "grad_norm": 1.068297329012161, "learning_rate": 9.624481917060035e-06, "loss": 0.454, "step": 14781 }, { "epoch": 0.5266401838359727, "grad_norm": 1.4917020098142113, "learning_rate": 9.623328796759399e-06, "loss": 0.5851, "step": 14782 }, { "epoch": 0.5266758109624669, "grad_norm": 1.2384038500226089, "learning_rate": 9.6221756814744e-06, "loss": 0.4816, "step": 14783 }, { "epoch": 0.5267114380889609, "grad_norm": 1.0742264245574082, "learning_rate": 9.621022571220404e-06, "loss": 0.4195, "step": 14784 }, { "epoch": 0.526747065215455, "grad_norm": 1.1586187359428448, "learning_rate": 9.61986946601275e-06, "loss": 0.458, "step": 14785 }, { "epoch": 0.5267826923419492, "grad_norm": 1.0791129072335826, "learning_rate": 9.6187163658668e-06, "loss": 0.4689, "step": 14786 }, { "epoch": 0.5268183194684433, "grad_norm": 1.0462193010222895, "learning_rate": 9.61756327079791e-06, "loss": 0.4729, "step": 14787 }, { "epoch": 0.5268539465949373, "grad_norm": 1.0994273484999433, "learning_rate": 9.616410180821431e-06, "loss": 0.4427, "step": 14788 }, { "epoch": 0.5268895737214315, "grad_norm": 1.1599075646790602, "learning_rate": 9.615257095952724e-06, "loss": 0.4667, "step": 14789 }, { "epoch": 0.5269252008479256, "grad_norm": 1.0857791379721933, "learning_rate": 9.61410401620713e-06, "loss": 0.4449, "step": 14790 }, { "epoch": 0.5269608279744197, "grad_norm": 1.1395751554953064, "learning_rate": 9.612950941600016e-06, "loss": 0.4405, "step": 14791 }, { "epoch": 0.5269964551009139, "grad_norm": 1.5266723762013317, "learning_rate": 9.611797872146729e-06, "loss": 0.6262, "step": 14792 }, { "epoch": 0.5270320822274079, "grad_norm": 1.0555466933948383, "learning_rate": 9.610644807862625e-06, "loss": 0.4617, "step": 14793 }, { "epoch": 0.527067709353902, "grad_norm": 1.6577968933899685, "learning_rate": 9.609491748763055e-06, "loss": 0.5736, "step": 14794 }, { "epoch": 0.5271033364803962, "grad_norm": 1.100873483100314, "learning_rate": 9.60833869486338e-06, "loss": 0.4288, "step": 14795 }, { "epoch": 0.5271389636068903, "grad_norm": 1.8100599108090334, "learning_rate": 9.607185646178946e-06, "loss": 0.5715, "step": 14796 }, { "epoch": 0.5271745907333844, "grad_norm": 1.4393294446696132, "learning_rate": 9.60603260272511e-06, "loss": 0.5173, "step": 14797 }, { "epoch": 0.5272102178598785, "grad_norm": 1.1024135697816204, "learning_rate": 9.604879564517225e-06, "loss": 0.4231, "step": 14798 }, { "epoch": 0.5272458449863726, "grad_norm": 1.6132956199371948, "learning_rate": 9.603726531570644e-06, "loss": 0.5502, "step": 14799 }, { "epoch": 0.5272814721128667, "grad_norm": 2.183095161148208, "learning_rate": 9.602573503900727e-06, "loss": 0.5484, "step": 14800 }, { "epoch": 0.5273170992393609, "grad_norm": 1.1350861347628145, "learning_rate": 9.601420481522813e-06, "loss": 0.4456, "step": 14801 }, { "epoch": 0.527352726365855, "grad_norm": 1.200589399841175, "learning_rate": 9.60026746445227e-06, "loss": 0.465, "step": 14802 }, { "epoch": 0.527388353492349, "grad_norm": 1.1808375855823614, "learning_rate": 9.599114452704442e-06, "loss": 0.4378, "step": 14803 }, { "epoch": 0.5274239806188432, "grad_norm": 1.330404681951265, "learning_rate": 9.597961446294687e-06, "loss": 0.5122, "step": 14804 }, { "epoch": 0.5274596077453373, "grad_norm": 1.387406905755402, "learning_rate": 9.596808445238354e-06, "loss": 0.4476, "step": 14805 }, { "epoch": 0.5274952348718314, "grad_norm": 1.5216533447543836, "learning_rate": 9.595655449550804e-06, "loss": 0.5668, "step": 14806 }, { "epoch": 0.5275308619983256, "grad_norm": 1.6279361944809176, "learning_rate": 9.594502459247378e-06, "loss": 0.6292, "step": 14807 }, { "epoch": 0.5275664891248196, "grad_norm": 1.7053974469569921, "learning_rate": 9.593349474343437e-06, "loss": 0.6096, "step": 14808 }, { "epoch": 0.5276021162513137, "grad_norm": 1.2198588734592601, "learning_rate": 9.592196494854332e-06, "loss": 0.4815, "step": 14809 }, { "epoch": 0.5276377433778079, "grad_norm": 1.1297179441830514, "learning_rate": 9.591043520795415e-06, "loss": 0.4346, "step": 14810 }, { "epoch": 0.527673370504302, "grad_norm": 1.9883365399150204, "learning_rate": 9.589890552182044e-06, "loss": 0.645, "step": 14811 }, { "epoch": 0.527708997630796, "grad_norm": 1.4612941787129552, "learning_rate": 9.588737589029562e-06, "loss": 0.596, "step": 14812 }, { "epoch": 0.5277446247572902, "grad_norm": 1.1565155654973358, "learning_rate": 9.58758463135333e-06, "loss": 0.4716, "step": 14813 }, { "epoch": 0.5277802518837843, "grad_norm": 1.4327526765591094, "learning_rate": 9.586431679168693e-06, "loss": 0.4712, "step": 14814 }, { "epoch": 0.5278158790102784, "grad_norm": 1.5869984276856544, "learning_rate": 9.58527873249101e-06, "loss": 0.5759, "step": 14815 }, { "epoch": 0.5278515061367726, "grad_norm": 1.0733981594990711, "learning_rate": 9.584125791335627e-06, "loss": 0.4589, "step": 14816 }, { "epoch": 0.5278871332632666, "grad_norm": 1.3260053606572908, "learning_rate": 9.582972855717906e-06, "loss": 0.5511, "step": 14817 }, { "epoch": 0.5279227603897607, "grad_norm": 1.4364784162836586, "learning_rate": 9.581819925653189e-06, "loss": 0.4972, "step": 14818 }, { "epoch": 0.5279583875162549, "grad_norm": 1.4806391995715265, "learning_rate": 9.580667001156833e-06, "loss": 0.556, "step": 14819 }, { "epoch": 0.527994014642749, "grad_norm": 1.2030598131254893, "learning_rate": 9.579514082244186e-06, "loss": 0.4603, "step": 14820 }, { "epoch": 0.528029641769243, "grad_norm": 1.5035843542229883, "learning_rate": 9.578361168930604e-06, "loss": 0.5526, "step": 14821 }, { "epoch": 0.5280652688957372, "grad_norm": 1.488315109105694, "learning_rate": 9.577208261231441e-06, "loss": 0.6119, "step": 14822 }, { "epoch": 0.5281008960222313, "grad_norm": 1.4757608710328471, "learning_rate": 9.576055359162043e-06, "loss": 0.495, "step": 14823 }, { "epoch": 0.5281365231487254, "grad_norm": 1.313202259974329, "learning_rate": 9.574902462737766e-06, "loss": 0.5198, "step": 14824 }, { "epoch": 0.5281721502752196, "grad_norm": 1.4869240863894981, "learning_rate": 9.573749571973956e-06, "loss": 0.6183, "step": 14825 }, { "epoch": 0.5282077774017137, "grad_norm": 1.1821056646593016, "learning_rate": 9.572596686885974e-06, "loss": 0.4537, "step": 14826 }, { "epoch": 0.5282434045282077, "grad_norm": 1.1517742129445503, "learning_rate": 9.571443807489162e-06, "loss": 0.4736, "step": 14827 }, { "epoch": 0.5282790316547019, "grad_norm": 1.2604813622644409, "learning_rate": 9.570290933798877e-06, "loss": 0.4434, "step": 14828 }, { "epoch": 0.528314658781196, "grad_norm": 1.3124478038003802, "learning_rate": 9.569138065830466e-06, "loss": 0.4877, "step": 14829 }, { "epoch": 0.5283502859076901, "grad_norm": 1.286625157830403, "learning_rate": 9.567985203599285e-06, "loss": 0.4999, "step": 14830 }, { "epoch": 0.5283859130341843, "grad_norm": 1.2020538578412066, "learning_rate": 9.566832347120682e-06, "loss": 0.4798, "step": 14831 }, { "epoch": 0.5284215401606783, "grad_norm": 3.2611777779495896, "learning_rate": 9.565679496410008e-06, "loss": 0.5606, "step": 14832 }, { "epoch": 0.5284571672871724, "grad_norm": 1.1901868406417295, "learning_rate": 9.564526651482619e-06, "loss": 0.4322, "step": 14833 }, { "epoch": 0.5284927944136666, "grad_norm": 1.199009770383033, "learning_rate": 9.563373812353858e-06, "loss": 0.4704, "step": 14834 }, { "epoch": 0.5285284215401607, "grad_norm": 1.5617941440901437, "learning_rate": 9.562220979039083e-06, "loss": 0.4859, "step": 14835 }, { "epoch": 0.5285640486666547, "grad_norm": 1.4395081963464826, "learning_rate": 9.561068151553637e-06, "loss": 0.5033, "step": 14836 }, { "epoch": 0.5285996757931489, "grad_norm": 1.1169909173523285, "learning_rate": 9.559915329912883e-06, "loss": 0.4583, "step": 14837 }, { "epoch": 0.528635302919643, "grad_norm": 1.0875937716723771, "learning_rate": 9.558762514132157e-06, "loss": 0.4412, "step": 14838 }, { "epoch": 0.5286709300461371, "grad_norm": 1.075106542283077, "learning_rate": 9.55760970422682e-06, "loss": 0.4558, "step": 14839 }, { "epoch": 0.5287065571726313, "grad_norm": 1.491290314147331, "learning_rate": 9.556456900212217e-06, "loss": 0.5899, "step": 14840 }, { "epoch": 0.5287421842991253, "grad_norm": 1.4897441382123697, "learning_rate": 9.5553041021037e-06, "loss": 0.5811, "step": 14841 }, { "epoch": 0.5287778114256194, "grad_norm": 1.2066947468661349, "learning_rate": 9.554151309916621e-06, "loss": 0.4785, "step": 14842 }, { "epoch": 0.5288134385521136, "grad_norm": 1.5827563911465519, "learning_rate": 9.552998523666327e-06, "loss": 0.5549, "step": 14843 }, { "epoch": 0.5288490656786077, "grad_norm": 1.1772733630074301, "learning_rate": 9.551845743368175e-06, "loss": 0.4342, "step": 14844 }, { "epoch": 0.5288846928051018, "grad_norm": 1.099230053007252, "learning_rate": 9.550692969037506e-06, "loss": 0.4063, "step": 14845 }, { "epoch": 0.5289203199315959, "grad_norm": 1.38739647036354, "learning_rate": 9.549540200689676e-06, "loss": 0.6192, "step": 14846 }, { "epoch": 0.52895594705809, "grad_norm": 1.6029714214195674, "learning_rate": 9.54838743834003e-06, "loss": 0.6374, "step": 14847 }, { "epoch": 0.5289915741845841, "grad_norm": 1.4135831974959192, "learning_rate": 9.547234682003928e-06, "loss": 0.5343, "step": 14848 }, { "epoch": 0.5290272013110783, "grad_norm": 0.9874575663233462, "learning_rate": 9.546081931696707e-06, "loss": 0.4337, "step": 14849 }, { "epoch": 0.5290628284375724, "grad_norm": 1.7431228366505382, "learning_rate": 9.544929187433724e-06, "loss": 0.5579, "step": 14850 }, { "epoch": 0.5290984555640664, "grad_norm": 1.109039781904249, "learning_rate": 9.543776449230325e-06, "loss": 0.4091, "step": 14851 }, { "epoch": 0.5291340826905606, "grad_norm": 2.1079711736918885, "learning_rate": 9.542623717101866e-06, "loss": 0.5914, "step": 14852 }, { "epoch": 0.5291697098170547, "grad_norm": 1.4735288661235968, "learning_rate": 9.541470991063685e-06, "loss": 0.5284, "step": 14853 }, { "epoch": 0.5292053369435488, "grad_norm": 1.1236640060176764, "learning_rate": 9.54031827113114e-06, "loss": 0.4526, "step": 14854 }, { "epoch": 0.529240964070043, "grad_norm": 1.6431282298447207, "learning_rate": 9.539165557319583e-06, "loss": 0.6548, "step": 14855 }, { "epoch": 0.529276591196537, "grad_norm": 1.176015565445936, "learning_rate": 9.538012849644355e-06, "loss": 0.4295, "step": 14856 }, { "epoch": 0.5293122183230311, "grad_norm": 0.9603448252263962, "learning_rate": 9.536860148120811e-06, "loss": 0.3981, "step": 14857 }, { "epoch": 0.5293478454495253, "grad_norm": 1.1319310437767771, "learning_rate": 9.535707452764296e-06, "loss": 0.4758, "step": 14858 }, { "epoch": 0.5293834725760194, "grad_norm": 1.3221596339454318, "learning_rate": 9.534554763590164e-06, "loss": 0.4203, "step": 14859 }, { "epoch": 0.5294190997025134, "grad_norm": 1.852588859506994, "learning_rate": 9.533402080613757e-06, "loss": 0.6121, "step": 14860 }, { "epoch": 0.5294547268290076, "grad_norm": 1.2279895058120682, "learning_rate": 9.53224940385043e-06, "loss": 0.432, "step": 14861 }, { "epoch": 0.5294903539555017, "grad_norm": 1.2030076769178304, "learning_rate": 9.531096733315527e-06, "loss": 0.4789, "step": 14862 }, { "epoch": 0.5295259810819958, "grad_norm": 2.2933827814885235, "learning_rate": 9.529944069024402e-06, "loss": 0.5517, "step": 14863 }, { "epoch": 0.52956160820849, "grad_norm": 1.6058456093132634, "learning_rate": 9.528791410992395e-06, "loss": 0.5389, "step": 14864 }, { "epoch": 0.529597235334984, "grad_norm": 1.1227113554963306, "learning_rate": 9.527638759234864e-06, "loss": 0.456, "step": 14865 }, { "epoch": 0.5296328624614781, "grad_norm": 1.383574936843789, "learning_rate": 9.526486113767151e-06, "loss": 0.5455, "step": 14866 }, { "epoch": 0.5296684895879723, "grad_norm": 1.1343323047733556, "learning_rate": 9.525333474604604e-06, "loss": 0.4727, "step": 14867 }, { "epoch": 0.5297041167144664, "grad_norm": 1.1408847672984672, "learning_rate": 9.524180841762577e-06, "loss": 0.4494, "step": 14868 }, { "epoch": 0.5297397438409605, "grad_norm": 1.1723406492786108, "learning_rate": 9.523028215256413e-06, "loss": 0.4786, "step": 14869 }, { "epoch": 0.5297753709674546, "grad_norm": 1.1291302837787778, "learning_rate": 9.521875595101467e-06, "loss": 0.4451, "step": 14870 }, { "epoch": 0.5298109980939487, "grad_norm": 1.087303047039004, "learning_rate": 9.520722981313076e-06, "loss": 0.4442, "step": 14871 }, { "epoch": 0.5298466252204428, "grad_norm": 1.0347415726062748, "learning_rate": 9.519570373906597e-06, "loss": 0.4244, "step": 14872 }, { "epoch": 0.529882252346937, "grad_norm": 1.5993374885581713, "learning_rate": 9.51841777289737e-06, "loss": 0.5436, "step": 14873 }, { "epoch": 0.529917879473431, "grad_norm": 1.1612480798923703, "learning_rate": 9.517265178300753e-06, "loss": 0.4952, "step": 14874 }, { "epoch": 0.5299535065999251, "grad_norm": 2.1035947287702887, "learning_rate": 9.516112590132082e-06, "loss": 0.5609, "step": 14875 }, { "epoch": 0.5299891337264193, "grad_norm": 1.0890563033932479, "learning_rate": 9.514960008406712e-06, "loss": 0.4138, "step": 14876 }, { "epoch": 0.5300247608529134, "grad_norm": 1.1558266990249313, "learning_rate": 9.513807433139987e-06, "loss": 0.4673, "step": 14877 }, { "epoch": 0.5300603879794075, "grad_norm": 1.6786510518660023, "learning_rate": 9.512654864347257e-06, "loss": 0.4695, "step": 14878 }, { "epoch": 0.5300960151059017, "grad_norm": 1.0211223282752062, "learning_rate": 9.511502302043867e-06, "loss": 0.4423, "step": 14879 }, { "epoch": 0.5301316422323957, "grad_norm": 1.5140762796951754, "learning_rate": 9.510349746245166e-06, "loss": 0.516, "step": 14880 }, { "epoch": 0.5301672693588898, "grad_norm": 1.5737865189798588, "learning_rate": 9.509197196966504e-06, "loss": 0.544, "step": 14881 }, { "epoch": 0.530202896485384, "grad_norm": 1.163663336542031, "learning_rate": 9.50804465422322e-06, "loss": 0.4538, "step": 14882 }, { "epoch": 0.5302385236118781, "grad_norm": 1.4673233159343972, "learning_rate": 9.506892118030668e-06, "loss": 0.5346, "step": 14883 }, { "epoch": 0.5302741507383723, "grad_norm": 1.164094658814896, "learning_rate": 9.50573958840419e-06, "loss": 0.4327, "step": 14884 }, { "epoch": 0.5303097778648663, "grad_norm": 1.1529064619949916, "learning_rate": 9.504587065359137e-06, "loss": 0.4831, "step": 14885 }, { "epoch": 0.5303454049913604, "grad_norm": 1.0733950798667067, "learning_rate": 9.50343454891085e-06, "loss": 0.418, "step": 14886 }, { "epoch": 0.5303810321178546, "grad_norm": 1.945912621492452, "learning_rate": 9.502282039074683e-06, "loss": 0.5168, "step": 14887 }, { "epoch": 0.5304166592443487, "grad_norm": 1.067412033269361, "learning_rate": 9.501129535865976e-06, "loss": 0.4316, "step": 14888 }, { "epoch": 0.5304522863708427, "grad_norm": 0.9730829103972147, "learning_rate": 9.499977039300078e-06, "loss": 0.4024, "step": 14889 }, { "epoch": 0.5304879134973369, "grad_norm": 1.1150135205279876, "learning_rate": 9.498824549392338e-06, "loss": 0.4597, "step": 14890 }, { "epoch": 0.530523540623831, "grad_norm": 1.5888912455923747, "learning_rate": 9.497672066158095e-06, "loss": 0.5554, "step": 14891 }, { "epoch": 0.5305591677503251, "grad_norm": 1.2422341024737078, "learning_rate": 9.496519589612704e-06, "loss": 0.4816, "step": 14892 }, { "epoch": 0.5305947948768193, "grad_norm": 1.4675488737574203, "learning_rate": 9.495367119771504e-06, "loss": 0.4573, "step": 14893 }, { "epoch": 0.5306304220033133, "grad_norm": 1.1068654952549561, "learning_rate": 9.494214656649844e-06, "loss": 0.4451, "step": 14894 }, { "epoch": 0.5306660491298074, "grad_norm": 1.1159068061025383, "learning_rate": 9.493062200263068e-06, "loss": 0.4772, "step": 14895 }, { "epoch": 0.5307016762563016, "grad_norm": 1.5732613026903648, "learning_rate": 9.491909750626526e-06, "loss": 0.5309, "step": 14896 }, { "epoch": 0.5307373033827957, "grad_norm": 1.6751534678406146, "learning_rate": 9.490757307755559e-06, "loss": 0.6148, "step": 14897 }, { "epoch": 0.5307729305092898, "grad_norm": 1.0118929118742728, "learning_rate": 9.489604871665515e-06, "loss": 0.4201, "step": 14898 }, { "epoch": 0.5308085576357839, "grad_norm": 0.9488935563839637, "learning_rate": 9.488452442371736e-06, "loss": 0.4496, "step": 14899 }, { "epoch": 0.530844184762278, "grad_norm": 1.4071034248037038, "learning_rate": 9.487300019889573e-06, "loss": 0.5367, "step": 14900 }, { "epoch": 0.5308798118887721, "grad_norm": 1.2776664647554026, "learning_rate": 9.48614760423437e-06, "loss": 0.57, "step": 14901 }, { "epoch": 0.5309154390152663, "grad_norm": 1.0496187021012366, "learning_rate": 9.48499519542147e-06, "loss": 0.4277, "step": 14902 }, { "epoch": 0.5309510661417604, "grad_norm": 3.2650450680140897, "learning_rate": 9.483842793466218e-06, "loss": 0.5342, "step": 14903 }, { "epoch": 0.5309866932682544, "grad_norm": 1.1930844573496557, "learning_rate": 9.482690398383958e-06, "loss": 0.4712, "step": 14904 }, { "epoch": 0.5310223203947486, "grad_norm": 1.1139293922450983, "learning_rate": 9.481538010190041e-06, "loss": 0.4572, "step": 14905 }, { "epoch": 0.5310579475212427, "grad_norm": 1.0445998029853567, "learning_rate": 9.480385628899804e-06, "loss": 0.409, "step": 14906 }, { "epoch": 0.5310935746477368, "grad_norm": 1.4199220715235934, "learning_rate": 9.4792332545286e-06, "loss": 0.5002, "step": 14907 }, { "epoch": 0.531129201774231, "grad_norm": 1.23047666238156, "learning_rate": 9.478080887091766e-06, "loss": 0.4627, "step": 14908 }, { "epoch": 0.531164828900725, "grad_norm": 1.0082694573812763, "learning_rate": 9.476928526604651e-06, "loss": 0.417, "step": 14909 }, { "epoch": 0.5312004560272191, "grad_norm": 1.6466885445291863, "learning_rate": 9.475776173082597e-06, "loss": 0.6148, "step": 14910 }, { "epoch": 0.5312360831537133, "grad_norm": 1.1846413337519903, "learning_rate": 9.474623826540952e-06, "loss": 0.4664, "step": 14911 }, { "epoch": 0.5312717102802074, "grad_norm": 1.1783135904315283, "learning_rate": 9.473471486995056e-06, "loss": 0.4662, "step": 14912 }, { "epoch": 0.5313073374067014, "grad_norm": 1.5170486870962316, "learning_rate": 9.472319154460254e-06, "loss": 0.5414, "step": 14913 }, { "epoch": 0.5313429645331956, "grad_norm": 1.3582686909563912, "learning_rate": 9.471166828951893e-06, "loss": 0.5921, "step": 14914 }, { "epoch": 0.5313785916596897, "grad_norm": 1.0710199138725605, "learning_rate": 9.470014510485313e-06, "loss": 0.4549, "step": 14915 }, { "epoch": 0.5314142187861838, "grad_norm": 1.1195274557394839, "learning_rate": 9.468862199075866e-06, "loss": 0.463, "step": 14916 }, { "epoch": 0.531449845912678, "grad_norm": 1.0606648782415797, "learning_rate": 9.467709894738882e-06, "loss": 0.4363, "step": 14917 }, { "epoch": 0.531485473039172, "grad_norm": 1.1426899747319443, "learning_rate": 9.46655759748972e-06, "loss": 0.4481, "step": 14918 }, { "epoch": 0.5315211001656661, "grad_norm": 1.860801936665113, "learning_rate": 9.465405307343714e-06, "loss": 0.5826, "step": 14919 }, { "epoch": 0.5315567272921603, "grad_norm": 1.3517247816487816, "learning_rate": 9.46425302431621e-06, "loss": 0.468, "step": 14920 }, { "epoch": 0.5315923544186544, "grad_norm": 1.238461247757559, "learning_rate": 9.46310074842255e-06, "loss": 0.4707, "step": 14921 }, { "epoch": 0.5316279815451485, "grad_norm": 1.5239282120681437, "learning_rate": 9.461948479678083e-06, "loss": 0.5546, "step": 14922 }, { "epoch": 0.5316636086716426, "grad_norm": 1.1781572217583616, "learning_rate": 9.460796218098143e-06, "loss": 0.4774, "step": 14923 }, { "epoch": 0.5316992357981367, "grad_norm": 1.3502132862830503, "learning_rate": 9.45964396369808e-06, "loss": 0.518, "step": 14924 }, { "epoch": 0.5317348629246308, "grad_norm": 0.9447188571645369, "learning_rate": 9.458491716493235e-06, "loss": 0.4249, "step": 14925 }, { "epoch": 0.531770490051125, "grad_norm": 1.2012755243574367, "learning_rate": 9.45733947649895e-06, "loss": 0.4922, "step": 14926 }, { "epoch": 0.531806117177619, "grad_norm": 1.0671242499954199, "learning_rate": 9.456187243730573e-06, "loss": 0.4468, "step": 14927 }, { "epoch": 0.5318417443041131, "grad_norm": 1.2033173373868242, "learning_rate": 9.455035018203439e-06, "loss": 0.4326, "step": 14928 }, { "epoch": 0.5318773714306073, "grad_norm": 1.1362545606644234, "learning_rate": 9.4538827999329e-06, "loss": 0.4277, "step": 14929 }, { "epoch": 0.5319129985571014, "grad_norm": 2.3574198714565, "learning_rate": 9.452730588934287e-06, "loss": 0.5339, "step": 14930 }, { "epoch": 0.5319486256835955, "grad_norm": 1.2176140500285482, "learning_rate": 9.451578385222952e-06, "loss": 0.4434, "step": 14931 }, { "epoch": 0.5319842528100897, "grad_norm": 1.0627955295323985, "learning_rate": 9.450426188814233e-06, "loss": 0.4077, "step": 14932 }, { "epoch": 0.5320198799365837, "grad_norm": 1.7010920954069229, "learning_rate": 9.449273999723476e-06, "loss": 0.6313, "step": 14933 }, { "epoch": 0.5320555070630778, "grad_norm": 1.2186782865705719, "learning_rate": 9.448121817966017e-06, "loss": 0.4935, "step": 14934 }, { "epoch": 0.532091134189572, "grad_norm": 1.3356137296340693, "learning_rate": 9.446969643557202e-06, "loss": 0.5223, "step": 14935 }, { "epoch": 0.5321267613160661, "grad_norm": 1.5299016458141328, "learning_rate": 9.445817476512376e-06, "loss": 0.5697, "step": 14936 }, { "epoch": 0.5321623884425601, "grad_norm": 1.1288805832316495, "learning_rate": 9.444665316846874e-06, "loss": 0.4211, "step": 14937 }, { "epoch": 0.5321980155690543, "grad_norm": 1.2447485670666323, "learning_rate": 9.443513164576046e-06, "loss": 0.4132, "step": 14938 }, { "epoch": 0.5322336426955484, "grad_norm": 1.4044212978005677, "learning_rate": 9.442361019715225e-06, "loss": 0.6145, "step": 14939 }, { "epoch": 0.5322692698220425, "grad_norm": 1.7410622067971977, "learning_rate": 9.441208882279759e-06, "loss": 0.539, "step": 14940 }, { "epoch": 0.5323048969485367, "grad_norm": 1.1809317248904225, "learning_rate": 9.440056752284985e-06, "loss": 0.4321, "step": 14941 }, { "epoch": 0.5323405240750307, "grad_norm": 1.4503027552880028, "learning_rate": 9.438904629746248e-06, "loss": 0.5678, "step": 14942 }, { "epoch": 0.5323761512015248, "grad_norm": 1.6436140184636634, "learning_rate": 9.437752514678888e-06, "loss": 0.5291, "step": 14943 }, { "epoch": 0.532411778328019, "grad_norm": 1.3643155413964638, "learning_rate": 9.436600407098248e-06, "loss": 0.5644, "step": 14944 }, { "epoch": 0.5324474054545131, "grad_norm": 1.360288428726423, "learning_rate": 9.435448307019666e-06, "loss": 0.4634, "step": 14945 }, { "epoch": 0.5324830325810072, "grad_norm": 1.2034786114364853, "learning_rate": 9.434296214458482e-06, "loss": 0.4781, "step": 14946 }, { "epoch": 0.5325186597075013, "grad_norm": 1.0751668611805691, "learning_rate": 9.433144129430043e-06, "loss": 0.4345, "step": 14947 }, { "epoch": 0.5325542868339954, "grad_norm": 1.580234817112209, "learning_rate": 9.431992051949683e-06, "loss": 0.5941, "step": 14948 }, { "epoch": 0.5325899139604895, "grad_norm": 2.926836035488197, "learning_rate": 9.43083998203275e-06, "loss": 0.567, "step": 14949 }, { "epoch": 0.5326255410869837, "grad_norm": 1.4220690560056917, "learning_rate": 9.429687919694576e-06, "loss": 0.6201, "step": 14950 }, { "epoch": 0.5326611682134778, "grad_norm": 1.138275968044848, "learning_rate": 9.42853586495051e-06, "loss": 0.4615, "step": 14951 }, { "epoch": 0.5326967953399718, "grad_norm": 1.5219769036454727, "learning_rate": 9.427383817815885e-06, "loss": 0.5616, "step": 14952 }, { "epoch": 0.532732422466466, "grad_norm": 1.1196182516326476, "learning_rate": 9.426231778306049e-06, "loss": 0.4072, "step": 14953 }, { "epoch": 0.5327680495929601, "grad_norm": 1.171623442849817, "learning_rate": 9.425079746436331e-06, "loss": 0.4748, "step": 14954 }, { "epoch": 0.5328036767194542, "grad_norm": 1.1063818728689496, "learning_rate": 9.423927722222085e-06, "loss": 0.4412, "step": 14955 }, { "epoch": 0.5328393038459484, "grad_norm": 1.2814360955401265, "learning_rate": 9.42277570567864e-06, "loss": 0.5622, "step": 14956 }, { "epoch": 0.5328749309724424, "grad_norm": 1.0774088510163211, "learning_rate": 9.421623696821342e-06, "loss": 0.464, "step": 14957 }, { "epoch": 0.5329105580989365, "grad_norm": 1.3382851614019875, "learning_rate": 9.420471695665527e-06, "loss": 0.6455, "step": 14958 }, { "epoch": 0.5329461852254307, "grad_norm": 1.3423265737043641, "learning_rate": 9.419319702226534e-06, "loss": 0.5821, "step": 14959 }, { "epoch": 0.5329818123519248, "grad_norm": 1.3119190113832133, "learning_rate": 9.418167716519712e-06, "loss": 0.4484, "step": 14960 }, { "epoch": 0.5330174394784188, "grad_norm": 1.0885755881407388, "learning_rate": 9.417015738560388e-06, "loss": 0.4248, "step": 14961 }, { "epoch": 0.533053066604913, "grad_norm": 1.3976461894053465, "learning_rate": 9.41586376836391e-06, "loss": 0.5837, "step": 14962 }, { "epoch": 0.5330886937314071, "grad_norm": 1.2256754389532432, "learning_rate": 9.414711805945612e-06, "loss": 0.4233, "step": 14963 }, { "epoch": 0.5331243208579012, "grad_norm": 1.7645912471546377, "learning_rate": 9.413559851320838e-06, "loss": 0.5121, "step": 14964 }, { "epoch": 0.5331599479843954, "grad_norm": 1.0018943461782104, "learning_rate": 9.412407904504921e-06, "loss": 0.4595, "step": 14965 }, { "epoch": 0.5331955751108894, "grad_norm": 3.5476620129172236, "learning_rate": 9.411255965513206e-06, "loss": 0.4615, "step": 14966 }, { "epoch": 0.5332312022373835, "grad_norm": 1.399230981187878, "learning_rate": 9.410104034361027e-06, "loss": 0.4963, "step": 14967 }, { "epoch": 0.5332668293638777, "grad_norm": 1.355202249090837, "learning_rate": 9.408952111063727e-06, "loss": 0.4765, "step": 14968 }, { "epoch": 0.5333024564903718, "grad_norm": 1.1065402065595173, "learning_rate": 9.407800195636642e-06, "loss": 0.4528, "step": 14969 }, { "epoch": 0.5333380836168659, "grad_norm": 1.2069798766796176, "learning_rate": 9.40664828809511e-06, "loss": 0.4666, "step": 14970 }, { "epoch": 0.53337371074336, "grad_norm": 1.2253905002930028, "learning_rate": 9.405496388454476e-06, "loss": 0.4412, "step": 14971 }, { "epoch": 0.5334093378698541, "grad_norm": 1.0386589545632219, "learning_rate": 9.404344496730067e-06, "loss": 0.4281, "step": 14972 }, { "epoch": 0.5334449649963482, "grad_norm": 1.1451856598336647, "learning_rate": 9.40319261293723e-06, "loss": 0.4516, "step": 14973 }, { "epoch": 0.5334805921228424, "grad_norm": 1.143952360965664, "learning_rate": 9.4020407370913e-06, "loss": 0.4552, "step": 14974 }, { "epoch": 0.5335162192493365, "grad_norm": 1.9406189568479666, "learning_rate": 9.40088886920762e-06, "loss": 0.6556, "step": 14975 }, { "epoch": 0.5335518463758305, "grad_norm": 1.3810018146707492, "learning_rate": 9.399737009301516e-06, "loss": 0.5478, "step": 14976 }, { "epoch": 0.5335874735023247, "grad_norm": 1.5863169077367532, "learning_rate": 9.398585157388338e-06, "loss": 0.5629, "step": 14977 }, { "epoch": 0.5336231006288188, "grad_norm": 1.1534406115642022, "learning_rate": 9.397433313483417e-06, "loss": 0.4336, "step": 14978 }, { "epoch": 0.5336587277553129, "grad_norm": 1.2593333017017385, "learning_rate": 9.396281477602095e-06, "loss": 0.4886, "step": 14979 }, { "epoch": 0.533694354881807, "grad_norm": 1.1700108197054149, "learning_rate": 9.395129649759703e-06, "loss": 0.4867, "step": 14980 }, { "epoch": 0.5337299820083011, "grad_norm": 1.1050075809055961, "learning_rate": 9.393977829971583e-06, "loss": 0.4857, "step": 14981 }, { "epoch": 0.5337656091347952, "grad_norm": 1.1632987172179725, "learning_rate": 9.392826018253078e-06, "loss": 0.4194, "step": 14982 }, { "epoch": 0.5338012362612894, "grad_norm": 1.0048534804373088, "learning_rate": 9.391674214619511e-06, "loss": 0.4094, "step": 14983 }, { "epoch": 0.5338368633877835, "grad_norm": 2.0767952727454646, "learning_rate": 9.390522419086232e-06, "loss": 0.5573, "step": 14984 }, { "epoch": 0.5338724905142775, "grad_norm": 1.107798457784455, "learning_rate": 9.389370631668571e-06, "loss": 0.419, "step": 14985 }, { "epoch": 0.5339081176407717, "grad_norm": 2.030125339695635, "learning_rate": 9.38821885238187e-06, "loss": 0.5825, "step": 14986 }, { "epoch": 0.5339437447672658, "grad_norm": 1.2034377014005049, "learning_rate": 9.38706708124146e-06, "loss": 0.4029, "step": 14987 }, { "epoch": 0.5339793718937599, "grad_norm": 1.390100199807069, "learning_rate": 9.38591531826268e-06, "loss": 0.4988, "step": 14988 }, { "epoch": 0.5340149990202541, "grad_norm": 1.3805036317717243, "learning_rate": 9.384763563460866e-06, "loss": 0.5838, "step": 14989 }, { "epoch": 0.5340506261467481, "grad_norm": 1.1551466306373288, "learning_rate": 9.383611816851359e-06, "loss": 0.428, "step": 14990 }, { "epoch": 0.5340862532732422, "grad_norm": 1.1611795120625028, "learning_rate": 9.382460078449487e-06, "loss": 0.4367, "step": 14991 }, { "epoch": 0.5341218803997364, "grad_norm": 1.150747467090193, "learning_rate": 9.381308348270593e-06, "loss": 0.432, "step": 14992 }, { "epoch": 0.5341575075262305, "grad_norm": 1.1282493691579922, "learning_rate": 9.380156626330009e-06, "loss": 0.4241, "step": 14993 }, { "epoch": 0.5341931346527246, "grad_norm": 1.4726759748254483, "learning_rate": 9.379004912643073e-06, "loss": 0.525, "step": 14994 }, { "epoch": 0.5342287617792187, "grad_norm": 1.469067078865431, "learning_rate": 9.377853207225122e-06, "loss": 0.4941, "step": 14995 }, { "epoch": 0.5342643889057128, "grad_norm": 1.000047770931299, "learning_rate": 9.376701510091488e-06, "loss": 0.4464, "step": 14996 }, { "epoch": 0.5343000160322069, "grad_norm": 1.132605308306152, "learning_rate": 9.375549821257515e-06, "loss": 0.4222, "step": 14997 }, { "epoch": 0.5343356431587011, "grad_norm": 1.5053833805321946, "learning_rate": 9.374398140738526e-06, "loss": 0.57, "step": 14998 }, { "epoch": 0.5343712702851952, "grad_norm": 1.4713250401464677, "learning_rate": 9.373246468549867e-06, "loss": 0.5824, "step": 14999 }, { "epoch": 0.5344068974116892, "grad_norm": 1.5119831576700118, "learning_rate": 9.372094804706867e-06, "loss": 0.57, "step": 15000 }, { "epoch": 0.5344425245381834, "grad_norm": 1.605249784131017, "learning_rate": 9.370943149224867e-06, "loss": 0.4902, "step": 15001 }, { "epoch": 0.5344781516646775, "grad_norm": 2.309165546104527, "learning_rate": 9.369791502119195e-06, "loss": 0.5354, "step": 15002 }, { "epoch": 0.5345137787911716, "grad_norm": 1.224422838935552, "learning_rate": 9.368639863405192e-06, "loss": 0.4614, "step": 15003 }, { "epoch": 0.5345494059176658, "grad_norm": 1.6661361458217405, "learning_rate": 9.367488233098187e-06, "loss": 0.5247, "step": 15004 }, { "epoch": 0.5345850330441598, "grad_norm": 1.4343874874357387, "learning_rate": 9.36633661121352e-06, "loss": 0.6139, "step": 15005 }, { "epoch": 0.5346206601706539, "grad_norm": 1.1566860326034447, "learning_rate": 9.365184997766525e-06, "loss": 0.4758, "step": 15006 }, { "epoch": 0.5346562872971481, "grad_norm": 1.2218336005333403, "learning_rate": 9.364033392772534e-06, "loss": 0.432, "step": 15007 }, { "epoch": 0.5346919144236422, "grad_norm": 1.1863066685954093, "learning_rate": 9.362881796246886e-06, "loss": 0.4646, "step": 15008 }, { "epoch": 0.5347275415501362, "grad_norm": 1.018202726626956, "learning_rate": 9.36173020820491e-06, "loss": 0.4283, "step": 15009 }, { "epoch": 0.5347631686766304, "grad_norm": 1.1095467164692525, "learning_rate": 9.360578628661942e-06, "loss": 0.4314, "step": 15010 }, { "epoch": 0.5347987958031245, "grad_norm": 1.098346776175019, "learning_rate": 9.359427057633317e-06, "loss": 0.4528, "step": 15011 }, { "epoch": 0.5348344229296186, "grad_norm": 1.3579754537249842, "learning_rate": 9.358275495134371e-06, "loss": 0.6436, "step": 15012 }, { "epoch": 0.5348700500561128, "grad_norm": 1.0804660068060896, "learning_rate": 9.35712394118043e-06, "loss": 0.4877, "step": 15013 }, { "epoch": 0.5349056771826068, "grad_norm": 1.364447365588641, "learning_rate": 9.355972395786837e-06, "loss": 0.4485, "step": 15014 }, { "epoch": 0.5349413043091009, "grad_norm": 1.07934736880136, "learning_rate": 9.35482085896892e-06, "loss": 0.4023, "step": 15015 }, { "epoch": 0.5349769314355951, "grad_norm": 1.0841025989623516, "learning_rate": 9.353669330742014e-06, "loss": 0.5018, "step": 15016 }, { "epoch": 0.5350125585620892, "grad_norm": 1.2984681911852571, "learning_rate": 9.352517811121453e-06, "loss": 0.4201, "step": 15017 }, { "epoch": 0.5350481856885833, "grad_norm": 1.3758167571798394, "learning_rate": 9.351366300122569e-06, "loss": 0.5449, "step": 15018 }, { "epoch": 0.5350838128150774, "grad_norm": 1.1302450232602506, "learning_rate": 9.350214797760701e-06, "loss": 0.4644, "step": 15019 }, { "epoch": 0.5351194399415715, "grad_norm": 2.180565506920723, "learning_rate": 9.34906330405117e-06, "loss": 0.4996, "step": 15020 }, { "epoch": 0.5351550670680656, "grad_norm": 1.4148812845893683, "learning_rate": 9.347911819009322e-06, "loss": 0.5138, "step": 15021 }, { "epoch": 0.5351906941945598, "grad_norm": 1.4023140342492229, "learning_rate": 9.346760342650481e-06, "loss": 0.5438, "step": 15022 }, { "epoch": 0.5352263213210539, "grad_norm": 1.2299551656023215, "learning_rate": 9.345608874989986e-06, "loss": 0.5432, "step": 15023 }, { "epoch": 0.5352619484475479, "grad_norm": 1.2542536155468103, "learning_rate": 9.34445741604316e-06, "loss": 0.5638, "step": 15024 }, { "epoch": 0.5352975755740421, "grad_norm": 1.0462372472183648, "learning_rate": 9.343305965825348e-06, "loss": 0.4582, "step": 15025 }, { "epoch": 0.5353332027005362, "grad_norm": 1.1996357424288069, "learning_rate": 9.342154524351871e-06, "loss": 0.4132, "step": 15026 }, { "epoch": 0.5353688298270303, "grad_norm": 1.1164640725711499, "learning_rate": 9.341003091638066e-06, "loss": 0.4598, "step": 15027 }, { "epoch": 0.5354044569535245, "grad_norm": 1.8942085447360788, "learning_rate": 9.339851667699272e-06, "loss": 0.5474, "step": 15028 }, { "epoch": 0.5354400840800185, "grad_norm": 1.4606248865751326, "learning_rate": 9.338700252550806e-06, "loss": 0.4847, "step": 15029 }, { "epoch": 0.5354757112065126, "grad_norm": 1.5728877928702065, "learning_rate": 9.337548846208016e-06, "loss": 0.621, "step": 15030 }, { "epoch": 0.5355113383330068, "grad_norm": 1.51140614523888, "learning_rate": 9.336397448686221e-06, "loss": 0.5203, "step": 15031 }, { "epoch": 0.5355469654595009, "grad_norm": 1.1084579158162002, "learning_rate": 9.33524606000076e-06, "loss": 0.4237, "step": 15032 }, { "epoch": 0.5355825925859949, "grad_norm": 1.1067865164829183, "learning_rate": 9.334094680166962e-06, "loss": 0.4335, "step": 15033 }, { "epoch": 0.5356182197124891, "grad_norm": 1.6181213931216347, "learning_rate": 9.332943309200163e-06, "loss": 0.6324, "step": 15034 }, { "epoch": 0.5356538468389832, "grad_norm": 1.0921798543933932, "learning_rate": 9.331791947115683e-06, "loss": 0.4673, "step": 15035 }, { "epoch": 0.5356894739654773, "grad_norm": 1.1997783848557326, "learning_rate": 9.330640593928863e-06, "loss": 0.4578, "step": 15036 }, { "epoch": 0.5357251010919715, "grad_norm": 1.0493661772984246, "learning_rate": 9.329489249655032e-06, "loss": 0.4557, "step": 15037 }, { "epoch": 0.5357607282184655, "grad_norm": 1.0833291788979411, "learning_rate": 9.328337914309521e-06, "loss": 0.4527, "step": 15038 }, { "epoch": 0.5357963553449596, "grad_norm": 1.4629789757294156, "learning_rate": 9.327186587907658e-06, "loss": 0.5066, "step": 15039 }, { "epoch": 0.5358319824714538, "grad_norm": 1.0495678272514528, "learning_rate": 9.326035270464777e-06, "loss": 0.4283, "step": 15040 }, { "epoch": 0.5358676095979479, "grad_norm": 1.6465451791738879, "learning_rate": 9.324883961996207e-06, "loss": 0.5119, "step": 15041 }, { "epoch": 0.535903236724442, "grad_norm": 1.0086663666466393, "learning_rate": 9.323732662517278e-06, "loss": 0.3837, "step": 15042 }, { "epoch": 0.5359388638509361, "grad_norm": 1.141589461565078, "learning_rate": 9.322581372043322e-06, "loss": 0.4813, "step": 15043 }, { "epoch": 0.5359744909774302, "grad_norm": 1.817978878647779, "learning_rate": 9.321430090589668e-06, "loss": 0.499, "step": 15044 }, { "epoch": 0.5360101181039243, "grad_norm": 1.342692202056374, "learning_rate": 9.32027881817165e-06, "loss": 0.6101, "step": 15045 }, { "epoch": 0.5360457452304185, "grad_norm": 1.1820165436991996, "learning_rate": 9.31912755480459e-06, "loss": 0.4853, "step": 15046 }, { "epoch": 0.5360813723569126, "grad_norm": 1.4343903196492276, "learning_rate": 9.317976300503827e-06, "loss": 0.6125, "step": 15047 }, { "epoch": 0.5361169994834066, "grad_norm": 1.1241862477145343, "learning_rate": 9.316825055284682e-06, "loss": 0.4476, "step": 15048 }, { "epoch": 0.5361526266099008, "grad_norm": 1.146493621863065, "learning_rate": 9.315673819162494e-06, "loss": 0.4304, "step": 15049 }, { "epoch": 0.5361882537363949, "grad_norm": 1.632780827482049, "learning_rate": 9.314522592152582e-06, "loss": 0.5669, "step": 15050 }, { "epoch": 0.536223880862889, "grad_norm": 1.1756526135598553, "learning_rate": 9.313371374270283e-06, "loss": 0.4652, "step": 15051 }, { "epoch": 0.5362595079893832, "grad_norm": 1.1597815137694143, "learning_rate": 9.312220165530926e-06, "loss": 0.4373, "step": 15052 }, { "epoch": 0.5362951351158772, "grad_norm": 1.2042059944730308, "learning_rate": 9.311068965949834e-06, "loss": 0.4655, "step": 15053 }, { "epoch": 0.5363307622423713, "grad_norm": 1.8140471531867048, "learning_rate": 9.309917775542345e-06, "loss": 0.5828, "step": 15054 }, { "epoch": 0.5363663893688655, "grad_norm": 1.0167628912623028, "learning_rate": 9.308766594323778e-06, "loss": 0.478, "step": 15055 }, { "epoch": 0.5364020164953596, "grad_norm": 1.155409424834724, "learning_rate": 9.307615422309474e-06, "loss": 0.453, "step": 15056 }, { "epoch": 0.5364376436218536, "grad_norm": 1.6305364324117952, "learning_rate": 9.306464259514747e-06, "loss": 0.6002, "step": 15057 }, { "epoch": 0.5364732707483478, "grad_norm": 1.1915082495858718, "learning_rate": 9.305313105954937e-06, "loss": 0.4327, "step": 15058 }, { "epoch": 0.5365088978748419, "grad_norm": 1.1059190793095781, "learning_rate": 9.304161961645366e-06, "loss": 0.4054, "step": 15059 }, { "epoch": 0.536544525001336, "grad_norm": 1.227614576606187, "learning_rate": 9.30301082660137e-06, "loss": 0.4596, "step": 15060 }, { "epoch": 0.5365801521278302, "grad_norm": 1.1630361884793865, "learning_rate": 9.301859700838268e-06, "loss": 0.4672, "step": 15061 }, { "epoch": 0.5366157792543242, "grad_norm": 1.180229972774781, "learning_rate": 9.300708584371392e-06, "loss": 0.4546, "step": 15062 }, { "epoch": 0.5366514063808183, "grad_norm": 1.4585972924687094, "learning_rate": 9.299557477216073e-06, "loss": 0.5775, "step": 15063 }, { "epoch": 0.5366870335073125, "grad_norm": 1.3862578833110706, "learning_rate": 9.298406379387631e-06, "loss": 0.5041, "step": 15064 }, { "epoch": 0.5367226606338066, "grad_norm": 1.212918502001348, "learning_rate": 9.297255290901404e-06, "loss": 0.4281, "step": 15065 }, { "epoch": 0.5367582877603007, "grad_norm": 1.7515360442585193, "learning_rate": 9.296104211772709e-06, "loss": 0.5481, "step": 15066 }, { "epoch": 0.5367939148867948, "grad_norm": 1.1668844559933296, "learning_rate": 9.294953142016883e-06, "loss": 0.4522, "step": 15067 }, { "epoch": 0.5368295420132889, "grad_norm": 1.4941351141334756, "learning_rate": 9.293802081649244e-06, "loss": 0.5003, "step": 15068 }, { "epoch": 0.536865169139783, "grad_norm": 1.2299808683664908, "learning_rate": 9.292651030685126e-06, "loss": 0.424, "step": 15069 }, { "epoch": 0.5369007962662772, "grad_norm": 1.1625930624730083, "learning_rate": 9.291499989139854e-06, "loss": 0.4137, "step": 15070 }, { "epoch": 0.5369364233927713, "grad_norm": 1.2231295201978158, "learning_rate": 9.290348957028757e-06, "loss": 0.4118, "step": 15071 }, { "epoch": 0.5369720505192653, "grad_norm": 1.4165627357734514, "learning_rate": 9.289197934367156e-06, "loss": 0.6283, "step": 15072 }, { "epoch": 0.5370076776457595, "grad_norm": 1.669383942707361, "learning_rate": 9.288046921170382e-06, "loss": 0.5935, "step": 15073 }, { "epoch": 0.5370433047722536, "grad_norm": 1.0708167568993983, "learning_rate": 9.286895917453763e-06, "loss": 0.4648, "step": 15074 }, { "epoch": 0.5370789318987477, "grad_norm": 1.4766456176422438, "learning_rate": 9.285744923232621e-06, "loss": 0.4765, "step": 15075 }, { "epoch": 0.5371145590252419, "grad_norm": 1.068573540786077, "learning_rate": 9.284593938522289e-06, "loss": 0.4241, "step": 15076 }, { "epoch": 0.5371501861517359, "grad_norm": 1.147296392129808, "learning_rate": 9.283442963338082e-06, "loss": 0.451, "step": 15077 }, { "epoch": 0.53718581327823, "grad_norm": 1.1026995808181526, "learning_rate": 9.28229199769534e-06, "loss": 0.4421, "step": 15078 }, { "epoch": 0.5372214404047242, "grad_norm": 1.333496609908158, "learning_rate": 9.281141041609374e-06, "loss": 0.5085, "step": 15079 }, { "epoch": 0.5372570675312183, "grad_norm": 1.4963964570802017, "learning_rate": 9.279990095095525e-06, "loss": 0.4715, "step": 15080 }, { "epoch": 0.5372926946577123, "grad_norm": 1.0260569554511279, "learning_rate": 9.278839158169103e-06, "loss": 0.4543, "step": 15081 }, { "epoch": 0.5373283217842065, "grad_norm": 1.5106470344232508, "learning_rate": 9.27768823084545e-06, "loss": 0.6232, "step": 15082 }, { "epoch": 0.5373639489107006, "grad_norm": 1.3088200001582366, "learning_rate": 9.276537313139878e-06, "loss": 0.584, "step": 15083 }, { "epoch": 0.5373995760371947, "grad_norm": 1.1543872833612774, "learning_rate": 9.27538640506772e-06, "loss": 0.4571, "step": 15084 }, { "epoch": 0.5374352031636889, "grad_norm": 1.4338163356709304, "learning_rate": 9.274235506644295e-06, "loss": 0.6153, "step": 15085 }, { "epoch": 0.5374708302901829, "grad_norm": 1.5870383094053881, "learning_rate": 9.273084617884932e-06, "loss": 0.5035, "step": 15086 }, { "epoch": 0.537506457416677, "grad_norm": 1.6971807505043228, "learning_rate": 9.27193373880496e-06, "loss": 0.5482, "step": 15087 }, { "epoch": 0.5375420845431712, "grad_norm": 1.8671535963765322, "learning_rate": 9.270782869419694e-06, "loss": 0.4239, "step": 15088 }, { "epoch": 0.5375777116696653, "grad_norm": 1.1840159532386094, "learning_rate": 9.269632009744466e-06, "loss": 0.4107, "step": 15089 }, { "epoch": 0.5376133387961594, "grad_norm": 1.081571987921791, "learning_rate": 9.268481159794598e-06, "loss": 0.4248, "step": 15090 }, { "epoch": 0.5376489659226535, "grad_norm": 1.0764673470609427, "learning_rate": 9.267330319585418e-06, "loss": 0.4542, "step": 15091 }, { "epoch": 0.5376845930491476, "grad_norm": 1.0486944127318263, "learning_rate": 9.26617948913224e-06, "loss": 0.4402, "step": 15092 }, { "epoch": 0.5377202201756417, "grad_norm": 1.0604186043998327, "learning_rate": 9.265028668450403e-06, "loss": 0.4614, "step": 15093 }, { "epoch": 0.5377558473021359, "grad_norm": 0.9724251249003569, "learning_rate": 9.263877857555216e-06, "loss": 0.4422, "step": 15094 }, { "epoch": 0.53779147442863, "grad_norm": 1.6967007106655227, "learning_rate": 9.262727056462013e-06, "loss": 0.518, "step": 15095 }, { "epoch": 0.537827101555124, "grad_norm": 1.1762606544123992, "learning_rate": 9.261576265186113e-06, "loss": 0.4011, "step": 15096 }, { "epoch": 0.5378627286816182, "grad_norm": 1.3231968739913904, "learning_rate": 9.26042548374284e-06, "loss": 0.4461, "step": 15097 }, { "epoch": 0.5378983558081123, "grad_norm": 1.1331317195015789, "learning_rate": 9.259274712147526e-06, "loss": 0.4803, "step": 15098 }, { "epoch": 0.5379339829346064, "grad_norm": 1.1086222901900342, "learning_rate": 9.25812395041548e-06, "loss": 0.4113, "step": 15099 }, { "epoch": 0.5379696100611006, "grad_norm": 1.9280373926396852, "learning_rate": 9.256973198562035e-06, "loss": 0.6678, "step": 15100 }, { "epoch": 0.5380052371875946, "grad_norm": 1.2612787940240173, "learning_rate": 9.255822456602508e-06, "loss": 0.502, "step": 15101 }, { "epoch": 0.5380408643140887, "grad_norm": 1.170753163656676, "learning_rate": 9.25467172455223e-06, "loss": 0.49, "step": 15102 }, { "epoch": 0.5380764914405829, "grad_norm": 1.659411058352249, "learning_rate": 9.253521002426515e-06, "loss": 0.546, "step": 15103 }, { "epoch": 0.538112118567077, "grad_norm": 1.225095523113843, "learning_rate": 9.252370290240692e-06, "loss": 0.4732, "step": 15104 }, { "epoch": 0.538147745693571, "grad_norm": 1.0428989585196429, "learning_rate": 9.251219588010078e-06, "loss": 0.4615, "step": 15105 }, { "epoch": 0.5381833728200652, "grad_norm": 1.1856297093028092, "learning_rate": 9.250068895750004e-06, "loss": 0.4607, "step": 15106 }, { "epoch": 0.5382189999465593, "grad_norm": 1.2096136621677873, "learning_rate": 9.24891821347578e-06, "loss": 0.4654, "step": 15107 }, { "epoch": 0.5382546270730534, "grad_norm": 1.0977622235211908, "learning_rate": 9.247767541202738e-06, "loss": 0.453, "step": 15108 }, { "epoch": 0.5382902541995476, "grad_norm": 1.2049028370957087, "learning_rate": 9.2466168789462e-06, "loss": 0.5019, "step": 15109 }, { "epoch": 0.5383258813260416, "grad_norm": 1.1906105187627802, "learning_rate": 9.24546622672148e-06, "loss": 0.4595, "step": 15110 }, { "epoch": 0.5383615084525357, "grad_norm": 1.0769942237096133, "learning_rate": 9.244315584543908e-06, "loss": 0.3959, "step": 15111 }, { "epoch": 0.5383971355790299, "grad_norm": 1.6688690755690512, "learning_rate": 9.243164952428801e-06, "loss": 0.6273, "step": 15112 }, { "epoch": 0.538432762705524, "grad_norm": 1.6772979976738378, "learning_rate": 9.242014330391483e-06, "loss": 0.6185, "step": 15113 }, { "epoch": 0.5384683898320181, "grad_norm": 1.1678222650408092, "learning_rate": 9.240863718447271e-06, "loss": 0.4835, "step": 15114 }, { "epoch": 0.5385040169585122, "grad_norm": 1.1137262604416927, "learning_rate": 9.239713116611493e-06, "loss": 0.4036, "step": 15115 }, { "epoch": 0.5385396440850063, "grad_norm": 1.6949109973550178, "learning_rate": 9.23856252489946e-06, "loss": 0.5273, "step": 15116 }, { "epoch": 0.5385752712115004, "grad_norm": 1.0646740624625584, "learning_rate": 9.237411943326508e-06, "loss": 0.3972, "step": 15117 }, { "epoch": 0.5386108983379946, "grad_norm": 1.6658142440586468, "learning_rate": 9.23626137190794e-06, "loss": 0.5054, "step": 15118 }, { "epoch": 0.5386465254644887, "grad_norm": 1.2174803133254561, "learning_rate": 9.235110810659094e-06, "loss": 0.4305, "step": 15119 }, { "epoch": 0.5386821525909827, "grad_norm": 1.146757994360789, "learning_rate": 9.233960259595277e-06, "loss": 0.4112, "step": 15120 }, { "epoch": 0.5387177797174769, "grad_norm": 1.140722396598504, "learning_rate": 9.232809718731815e-06, "loss": 0.4739, "step": 15121 }, { "epoch": 0.538753406843971, "grad_norm": 1.108584553979001, "learning_rate": 9.231659188084028e-06, "loss": 0.4125, "step": 15122 }, { "epoch": 0.5387890339704651, "grad_norm": 1.1741682415055366, "learning_rate": 9.230508667667235e-06, "loss": 0.4357, "step": 15123 }, { "epoch": 0.5388246610969593, "grad_norm": 1.1870338094679673, "learning_rate": 9.229358157496761e-06, "loss": 0.4176, "step": 15124 }, { "epoch": 0.5388602882234533, "grad_norm": 2.069671759660488, "learning_rate": 9.228207657587918e-06, "loss": 0.5883, "step": 15125 }, { "epoch": 0.5388959153499474, "grad_norm": 1.0601309016784133, "learning_rate": 9.227057167956031e-06, "loss": 0.4384, "step": 15126 }, { "epoch": 0.5389315424764416, "grad_norm": 1.50958197885616, "learning_rate": 9.225906688616418e-06, "loss": 0.5797, "step": 15127 }, { "epoch": 0.5389671696029357, "grad_norm": 1.6664899568789766, "learning_rate": 9.224756219584401e-06, "loss": 0.6449, "step": 15128 }, { "epoch": 0.5390027967294297, "grad_norm": 1.0642584129519972, "learning_rate": 9.223605760875291e-06, "loss": 0.4334, "step": 15129 }, { "epoch": 0.5390384238559239, "grad_norm": 1.4663318972110722, "learning_rate": 9.222455312504417e-06, "loss": 0.538, "step": 15130 }, { "epoch": 0.539074050982418, "grad_norm": 1.0375716088771234, "learning_rate": 9.221304874487093e-06, "loss": 0.4389, "step": 15131 }, { "epoch": 0.5391096781089121, "grad_norm": 2.6588239861672087, "learning_rate": 9.220154446838638e-06, "loss": 0.5999, "step": 15132 }, { "epoch": 0.5391453052354063, "grad_norm": 1.5041964514488522, "learning_rate": 9.219004029574372e-06, "loss": 0.6512, "step": 15133 }, { "epoch": 0.5391809323619003, "grad_norm": 2.3108983874308593, "learning_rate": 9.217853622709612e-06, "loss": 0.6275, "step": 15134 }, { "epoch": 0.5392165594883944, "grad_norm": 1.8634944457663838, "learning_rate": 9.216703226259682e-06, "loss": 0.4921, "step": 15135 }, { "epoch": 0.5392521866148886, "grad_norm": 1.473902904157257, "learning_rate": 9.215552840239893e-06, "loss": 0.4972, "step": 15136 }, { "epoch": 0.5392878137413827, "grad_norm": 1.0785220899209869, "learning_rate": 9.214402464665568e-06, "loss": 0.4437, "step": 15137 }, { "epoch": 0.5393234408678768, "grad_norm": 1.1496939679672136, "learning_rate": 9.213252099552019e-06, "loss": 0.4441, "step": 15138 }, { "epoch": 0.5393590679943709, "grad_norm": 1.4712401238557866, "learning_rate": 9.212101744914574e-06, "loss": 0.5531, "step": 15139 }, { "epoch": 0.539394695120865, "grad_norm": 1.8489201079152615, "learning_rate": 9.210951400768541e-06, "loss": 0.5358, "step": 15140 }, { "epoch": 0.5394303222473591, "grad_norm": 1.1191463932374766, "learning_rate": 9.209801067129244e-06, "loss": 0.4503, "step": 15141 }, { "epoch": 0.5394659493738533, "grad_norm": 1.814887651434417, "learning_rate": 9.208650744011996e-06, "loss": 0.4843, "step": 15142 }, { "epoch": 0.5395015765003474, "grad_norm": 2.897943747395967, "learning_rate": 9.207500431432115e-06, "loss": 0.5319, "step": 15143 }, { "epoch": 0.5395372036268414, "grad_norm": 2.053169449936989, "learning_rate": 9.206350129404924e-06, "loss": 0.6622, "step": 15144 }, { "epoch": 0.5395728307533356, "grad_norm": 1.11394664300011, "learning_rate": 9.20519983794573e-06, "loss": 0.4324, "step": 15145 }, { "epoch": 0.5396084578798297, "grad_norm": 1.1301446995558408, "learning_rate": 9.204049557069864e-06, "loss": 0.4916, "step": 15146 }, { "epoch": 0.5396440850063238, "grad_norm": 1.2359215202343816, "learning_rate": 9.202899286792629e-06, "loss": 0.4501, "step": 15147 }, { "epoch": 0.539679712132818, "grad_norm": 1.872682279977707, "learning_rate": 9.201749027129348e-06, "loss": 0.6026, "step": 15148 }, { "epoch": 0.539715339259312, "grad_norm": 1.224277163217243, "learning_rate": 9.200598778095335e-06, "loss": 0.5361, "step": 15149 }, { "epoch": 0.5397509663858061, "grad_norm": 1.230589108970424, "learning_rate": 9.199448539705913e-06, "loss": 0.4545, "step": 15150 }, { "epoch": 0.5397865935123003, "grad_norm": 1.5702248644258703, "learning_rate": 9.19829831197639e-06, "loss": 0.4823, "step": 15151 }, { "epoch": 0.5398222206387944, "grad_norm": 1.139827776857621, "learning_rate": 9.197148094922085e-06, "loss": 0.4683, "step": 15152 }, { "epoch": 0.5398578477652884, "grad_norm": 1.3291220185077701, "learning_rate": 9.195997888558312e-06, "loss": 0.559, "step": 15153 }, { "epoch": 0.5398934748917826, "grad_norm": 1.0510802025168366, "learning_rate": 9.194847692900392e-06, "loss": 0.4459, "step": 15154 }, { "epoch": 0.5399291020182767, "grad_norm": 1.2937580552587555, "learning_rate": 9.193697507963639e-06, "loss": 0.5376, "step": 15155 }, { "epoch": 0.5399647291447708, "grad_norm": 1.0474078299123144, "learning_rate": 9.192547333763362e-06, "loss": 0.4193, "step": 15156 }, { "epoch": 0.540000356271265, "grad_norm": 1.4653063534704924, "learning_rate": 9.191397170314888e-06, "loss": 0.6298, "step": 15157 }, { "epoch": 0.540035983397759, "grad_norm": 1.5874628425462773, "learning_rate": 9.190247017633523e-06, "loss": 0.5739, "step": 15158 }, { "epoch": 0.5400716105242531, "grad_norm": 1.0537388111839685, "learning_rate": 9.189096875734585e-06, "loss": 0.4207, "step": 15159 }, { "epoch": 0.5401072376507473, "grad_norm": 2.289507188200292, "learning_rate": 9.187946744633387e-06, "loss": 0.4792, "step": 15160 }, { "epoch": 0.5401428647772414, "grad_norm": 1.6407202973456367, "learning_rate": 9.186796624345252e-06, "loss": 0.5687, "step": 15161 }, { "epoch": 0.5401784919037355, "grad_norm": 1.422484484742402, "learning_rate": 9.185646514885482e-06, "loss": 0.5448, "step": 15162 }, { "epoch": 0.5402141190302296, "grad_norm": 1.657136196140288, "learning_rate": 9.184496416269401e-06, "loss": 0.5683, "step": 15163 }, { "epoch": 0.5402497461567237, "grad_norm": 1.1137538937199118, "learning_rate": 9.183346328512316e-06, "loss": 0.4412, "step": 15164 }, { "epoch": 0.5402853732832178, "grad_norm": 1.0949192312076492, "learning_rate": 9.182196251629552e-06, "loss": 0.4331, "step": 15165 }, { "epoch": 0.540321000409712, "grad_norm": 1.2224582186578559, "learning_rate": 9.181046185636411e-06, "loss": 0.465, "step": 15166 }, { "epoch": 0.5403566275362061, "grad_norm": 1.0865444083484423, "learning_rate": 9.179896130548215e-06, "loss": 0.4464, "step": 15167 }, { "epoch": 0.5403922546627001, "grad_norm": 1.3859053628860576, "learning_rate": 9.178746086380274e-06, "loss": 0.552, "step": 15168 }, { "epoch": 0.5404278817891943, "grad_norm": 1.1874572422904435, "learning_rate": 9.177596053147903e-06, "loss": 0.435, "step": 15169 }, { "epoch": 0.5404635089156884, "grad_norm": 1.4816150125698575, "learning_rate": 9.176446030866418e-06, "loss": 0.6223, "step": 15170 }, { "epoch": 0.5404991360421825, "grad_norm": 1.202194792839207, "learning_rate": 9.175296019551125e-06, "loss": 0.4569, "step": 15171 }, { "epoch": 0.5405347631686767, "grad_norm": 1.9926078417345123, "learning_rate": 9.174146019217348e-06, "loss": 0.5619, "step": 15172 }, { "epoch": 0.5405703902951707, "grad_norm": 1.1695205155122832, "learning_rate": 9.172996029880389e-06, "loss": 0.4256, "step": 15173 }, { "epoch": 0.5406060174216648, "grad_norm": 1.4828663059201566, "learning_rate": 9.171846051555568e-06, "loss": 0.5275, "step": 15174 }, { "epoch": 0.540641644548159, "grad_norm": 1.0802365515144632, "learning_rate": 9.170696084258193e-06, "loss": 0.438, "step": 15175 }, { "epoch": 0.5406772716746531, "grad_norm": 2.046126862389618, "learning_rate": 9.169546128003585e-06, "loss": 0.6291, "step": 15176 }, { "epoch": 0.5407128988011471, "grad_norm": 1.1474646793171186, "learning_rate": 9.168396182807046e-06, "loss": 0.4543, "step": 15177 }, { "epoch": 0.5407485259276413, "grad_norm": 1.6250238670036803, "learning_rate": 9.167246248683893e-06, "loss": 0.503, "step": 15178 }, { "epoch": 0.5407841530541354, "grad_norm": 2.2203328937956166, "learning_rate": 9.16609632564944e-06, "loss": 0.5504, "step": 15179 }, { "epoch": 0.5408197801806295, "grad_norm": 1.1622720429658118, "learning_rate": 9.164946413718996e-06, "loss": 0.3857, "step": 15180 }, { "epoch": 0.5408554073071237, "grad_norm": 1.4213119464571426, "learning_rate": 9.163796512907875e-06, "loss": 0.5475, "step": 15181 }, { "epoch": 0.5408910344336177, "grad_norm": 1.873806304116797, "learning_rate": 9.162646623231388e-06, "loss": 0.6352, "step": 15182 }, { "epoch": 0.5409266615601118, "grad_norm": 1.3398017493406893, "learning_rate": 9.161496744704848e-06, "loss": 0.4954, "step": 15183 }, { "epoch": 0.540962288686606, "grad_norm": 1.023422944036816, "learning_rate": 9.160346877343562e-06, "loss": 0.4659, "step": 15184 }, { "epoch": 0.5409979158131001, "grad_norm": 1.72593847191606, "learning_rate": 9.159197021162846e-06, "loss": 0.5131, "step": 15185 }, { "epoch": 0.5410335429395942, "grad_norm": 2.7860789189600563, "learning_rate": 9.15804717617801e-06, "loss": 0.5183, "step": 15186 }, { "epoch": 0.5410691700660883, "grad_norm": 1.099516193102403, "learning_rate": 9.156897342404365e-06, "loss": 0.46, "step": 15187 }, { "epoch": 0.5411047971925824, "grad_norm": 1.2370535127258386, "learning_rate": 9.155747519857218e-06, "loss": 0.5061, "step": 15188 }, { "epoch": 0.5411404243190765, "grad_norm": 1.1517705853586988, "learning_rate": 9.154597708551882e-06, "loss": 0.4418, "step": 15189 }, { "epoch": 0.5411760514455707, "grad_norm": 1.286453682188203, "learning_rate": 9.153447908503671e-06, "loss": 0.4543, "step": 15190 }, { "epoch": 0.5412116785720648, "grad_norm": 1.2799283698543893, "learning_rate": 9.152298119727892e-06, "loss": 0.5375, "step": 15191 }, { "epoch": 0.5412473056985588, "grad_norm": 1.1194559872797465, "learning_rate": 9.15114834223986e-06, "loss": 0.4253, "step": 15192 }, { "epoch": 0.541282932825053, "grad_norm": 1.1821214372535165, "learning_rate": 9.149998576054875e-06, "loss": 0.4603, "step": 15193 }, { "epoch": 0.5413185599515471, "grad_norm": 1.1452002707639846, "learning_rate": 9.148848821188257e-06, "loss": 0.4877, "step": 15194 }, { "epoch": 0.5413541870780412, "grad_norm": 1.2525784298050207, "learning_rate": 9.14769907765531e-06, "loss": 0.4965, "step": 15195 }, { "epoch": 0.5413898142045354, "grad_norm": 1.2112542932794892, "learning_rate": 9.146549345471348e-06, "loss": 0.4209, "step": 15196 }, { "epoch": 0.5414254413310294, "grad_norm": 1.6123261568035483, "learning_rate": 9.145399624651673e-06, "loss": 0.6381, "step": 15197 }, { "epoch": 0.5414610684575235, "grad_norm": 1.014991877095889, "learning_rate": 9.144249915211605e-06, "loss": 0.432, "step": 15198 }, { "epoch": 0.5414966955840177, "grad_norm": 1.7912867712868634, "learning_rate": 9.143100217166445e-06, "loss": 0.4561, "step": 15199 }, { "epoch": 0.5415323227105118, "grad_norm": 2.0071951040718528, "learning_rate": 9.141950530531502e-06, "loss": 0.5873, "step": 15200 }, { "epoch": 0.5415679498370058, "grad_norm": 2.0060537179907434, "learning_rate": 9.140800855322092e-06, "loss": 0.5122, "step": 15201 }, { "epoch": 0.5416035769635, "grad_norm": 1.1664436087593375, "learning_rate": 9.139651191553515e-06, "loss": 0.4654, "step": 15202 }, { "epoch": 0.5416392040899941, "grad_norm": 1.0524975790100228, "learning_rate": 9.13850153924109e-06, "loss": 0.391, "step": 15203 }, { "epoch": 0.5416748312164882, "grad_norm": 1.2672255240238068, "learning_rate": 9.137351898400113e-06, "loss": 0.4486, "step": 15204 }, { "epoch": 0.5417104583429824, "grad_norm": 1.3118570025364646, "learning_rate": 9.1362022690459e-06, "loss": 0.4077, "step": 15205 }, { "epoch": 0.5417460854694764, "grad_norm": 1.3698482585283387, "learning_rate": 9.135052651193757e-06, "loss": 0.5286, "step": 15206 }, { "epoch": 0.5417817125959705, "grad_norm": 1.5750641877020013, "learning_rate": 9.133903044858994e-06, "loss": 0.4041, "step": 15207 }, { "epoch": 0.5418173397224647, "grad_norm": 1.4929341545344843, "learning_rate": 9.132753450056916e-06, "loss": 0.5422, "step": 15208 }, { "epoch": 0.5418529668489588, "grad_norm": 1.2793540760468138, "learning_rate": 9.131603866802836e-06, "loss": 0.4107, "step": 15209 }, { "epoch": 0.5418885939754529, "grad_norm": 1.3784404481743968, "learning_rate": 9.130454295112051e-06, "loss": 0.5741, "step": 15210 }, { "epoch": 0.541924221101947, "grad_norm": 1.79217899771956, "learning_rate": 9.12930473499988e-06, "loss": 0.5599, "step": 15211 }, { "epoch": 0.5419598482284411, "grad_norm": 1.757609319062765, "learning_rate": 9.128155186481622e-06, "loss": 0.582, "step": 15212 }, { "epoch": 0.5419954753549352, "grad_norm": 1.2562823668761804, "learning_rate": 9.127005649572587e-06, "loss": 0.4155, "step": 15213 }, { "epoch": 0.5420311024814294, "grad_norm": 1.1745003524849085, "learning_rate": 9.125856124288087e-06, "loss": 0.4367, "step": 15214 }, { "epoch": 0.5420667296079235, "grad_norm": 1.1704499249988969, "learning_rate": 9.124706610643417e-06, "loss": 0.4558, "step": 15215 }, { "epoch": 0.5421023567344175, "grad_norm": 1.4789157104438693, "learning_rate": 9.123557108653894e-06, "loss": 0.549, "step": 15216 }, { "epoch": 0.5421379838609117, "grad_norm": 1.1003258440516377, "learning_rate": 9.122407618334819e-06, "loss": 0.4448, "step": 15217 }, { "epoch": 0.5421736109874058, "grad_norm": 1.9635260063719095, "learning_rate": 9.121258139701503e-06, "loss": 0.6012, "step": 15218 }, { "epoch": 0.5422092381138999, "grad_norm": 1.1923811075904145, "learning_rate": 9.120108672769243e-06, "loss": 0.5042, "step": 15219 }, { "epoch": 0.5422448652403941, "grad_norm": 1.5956427345794573, "learning_rate": 9.118959217553358e-06, "loss": 0.6278, "step": 15220 }, { "epoch": 0.5422804923668881, "grad_norm": 1.306590607694469, "learning_rate": 9.117809774069144e-06, "loss": 0.5381, "step": 15221 }, { "epoch": 0.5423161194933822, "grad_norm": 1.1052102134448418, "learning_rate": 9.116660342331909e-06, "loss": 0.4582, "step": 15222 }, { "epoch": 0.5423517466198764, "grad_norm": 1.1215833503937531, "learning_rate": 9.115510922356956e-06, "loss": 0.4243, "step": 15223 }, { "epoch": 0.5423873737463705, "grad_norm": 1.1751668156732327, "learning_rate": 9.114361514159596e-06, "loss": 0.4532, "step": 15224 }, { "epoch": 0.5424230008728645, "grad_norm": 1.5862203109957345, "learning_rate": 9.113212117755134e-06, "loss": 0.581, "step": 15225 }, { "epoch": 0.5424586279993587, "grad_norm": 1.3747568269205865, "learning_rate": 9.112062733158871e-06, "loss": 0.4645, "step": 15226 }, { "epoch": 0.5424942551258528, "grad_norm": 1.250227245761204, "learning_rate": 9.110913360386114e-06, "loss": 0.4961, "step": 15227 }, { "epoch": 0.5425298822523469, "grad_norm": 1.1688428220953093, "learning_rate": 9.109763999452163e-06, "loss": 0.4485, "step": 15228 }, { "epoch": 0.5425655093788411, "grad_norm": 1.6782073977312841, "learning_rate": 9.108614650372334e-06, "loss": 0.5334, "step": 15229 }, { "epoch": 0.5426011365053351, "grad_norm": 1.3617261508929217, "learning_rate": 9.107465313161916e-06, "loss": 0.5587, "step": 15230 }, { "epoch": 0.5426367636318292, "grad_norm": 1.2593414355066495, "learning_rate": 9.106315987836226e-06, "loss": 0.6012, "step": 15231 }, { "epoch": 0.5426723907583234, "grad_norm": 1.9773823194645137, "learning_rate": 9.105166674410559e-06, "loss": 0.4903, "step": 15232 }, { "epoch": 0.5427080178848175, "grad_norm": 1.1339129586455579, "learning_rate": 9.104017372900229e-06, "loss": 0.417, "step": 15233 }, { "epoch": 0.5427436450113116, "grad_norm": 1.120817252416487, "learning_rate": 9.102868083320528e-06, "loss": 0.4248, "step": 15234 }, { "epoch": 0.5427792721378057, "grad_norm": 1.036634018963386, "learning_rate": 9.101718805686767e-06, "loss": 0.4849, "step": 15235 }, { "epoch": 0.5428148992642998, "grad_norm": 0.9680877064251602, "learning_rate": 9.100569540014252e-06, "loss": 0.4069, "step": 15236 }, { "epoch": 0.5428505263907939, "grad_norm": 1.0204631401731248, "learning_rate": 9.09942028631828e-06, "loss": 0.4475, "step": 15237 }, { "epoch": 0.5428861535172881, "grad_norm": 1.3787666328454113, "learning_rate": 9.098271044614155e-06, "loss": 0.5794, "step": 15238 }, { "epoch": 0.5429217806437822, "grad_norm": 1.4686700517894637, "learning_rate": 9.097121814917182e-06, "loss": 0.616, "step": 15239 }, { "epoch": 0.5429574077702762, "grad_norm": 1.6208131127331655, "learning_rate": 9.095972597242666e-06, "loss": 0.5501, "step": 15240 }, { "epoch": 0.5429930348967704, "grad_norm": 1.5767138884763379, "learning_rate": 9.094823391605903e-06, "loss": 0.6184, "step": 15241 }, { "epoch": 0.5430286620232645, "grad_norm": 1.1642124242081253, "learning_rate": 9.093674198022201e-06, "loss": 0.4971, "step": 15242 }, { "epoch": 0.5430642891497586, "grad_norm": 1.2475275524636595, "learning_rate": 9.092525016506858e-06, "loss": 0.5221, "step": 15243 }, { "epoch": 0.5430999162762528, "grad_norm": 1.4710489184638071, "learning_rate": 9.091375847075183e-06, "loss": 0.4914, "step": 15244 }, { "epoch": 0.5431355434027468, "grad_norm": 1.1209573727041948, "learning_rate": 9.090226689742469e-06, "loss": 0.432, "step": 15245 }, { "epoch": 0.5431711705292409, "grad_norm": 1.2141911370672323, "learning_rate": 9.089077544524027e-06, "loss": 0.4211, "step": 15246 }, { "epoch": 0.5432067976557351, "grad_norm": 1.2361217691011825, "learning_rate": 9.087928411435151e-06, "loss": 0.4786, "step": 15247 }, { "epoch": 0.5432424247822292, "grad_norm": 1.750751751890715, "learning_rate": 9.086779290491145e-06, "loss": 0.5219, "step": 15248 }, { "epoch": 0.5432780519087232, "grad_norm": 1.283507524222195, "learning_rate": 9.085630181707314e-06, "loss": 0.5231, "step": 15249 }, { "epoch": 0.5433136790352174, "grad_norm": 1.5337820416227563, "learning_rate": 9.084481085098953e-06, "loss": 0.609, "step": 15250 }, { "epoch": 0.5433493061617115, "grad_norm": 1.0561685527629705, "learning_rate": 9.083332000681371e-06, "loss": 0.431, "step": 15251 }, { "epoch": 0.5433849332882056, "grad_norm": 1.4095595370139462, "learning_rate": 9.082182928469861e-06, "loss": 0.5166, "step": 15252 }, { "epoch": 0.5434205604146998, "grad_norm": 1.5959339017691674, "learning_rate": 9.081033868479727e-06, "loss": 0.4842, "step": 15253 }, { "epoch": 0.5434561875411938, "grad_norm": 1.120459758530045, "learning_rate": 9.07988482072627e-06, "loss": 0.455, "step": 15254 }, { "epoch": 0.5434918146676879, "grad_norm": 1.8007264514848238, "learning_rate": 9.078735785224791e-06, "loss": 0.592, "step": 15255 }, { "epoch": 0.5435274417941821, "grad_norm": 1.3466141332753305, "learning_rate": 9.077586761990586e-06, "loss": 0.5668, "step": 15256 }, { "epoch": 0.5435630689206762, "grad_norm": 1.1163912333300108, "learning_rate": 9.07643775103896e-06, "loss": 0.4589, "step": 15257 }, { "epoch": 0.5435986960471703, "grad_norm": 1.2234062466376845, "learning_rate": 9.075288752385208e-06, "loss": 0.4329, "step": 15258 }, { "epoch": 0.5436343231736644, "grad_norm": 1.1682388439801434, "learning_rate": 9.074139766044634e-06, "loss": 0.4458, "step": 15259 }, { "epoch": 0.5436699503001585, "grad_norm": 1.4910955028416113, "learning_rate": 9.072990792032537e-06, "loss": 0.5217, "step": 15260 }, { "epoch": 0.5437055774266527, "grad_norm": 1.2101371242820156, "learning_rate": 9.071841830364214e-06, "loss": 0.4472, "step": 15261 }, { "epoch": 0.5437412045531468, "grad_norm": 1.037071024834469, "learning_rate": 9.070692881054972e-06, "loss": 0.422, "step": 15262 }, { "epoch": 0.5437768316796409, "grad_norm": 1.2856855805039253, "learning_rate": 9.069543944120097e-06, "loss": 0.44, "step": 15263 }, { "epoch": 0.543812458806135, "grad_norm": 1.1184805279673986, "learning_rate": 9.068395019574897e-06, "loss": 0.4832, "step": 15264 }, { "epoch": 0.5438480859326291, "grad_norm": 1.0532213716436991, "learning_rate": 9.067246107434667e-06, "loss": 0.4195, "step": 15265 }, { "epoch": 0.5438837130591232, "grad_norm": 1.0969576848542082, "learning_rate": 9.06609720771471e-06, "loss": 0.4646, "step": 15266 }, { "epoch": 0.5439193401856174, "grad_norm": 1.1594353785008709, "learning_rate": 9.06494832043032e-06, "loss": 0.4489, "step": 15267 }, { "epoch": 0.5439549673121115, "grad_norm": 1.227411203581911, "learning_rate": 9.063799445596796e-06, "loss": 0.507, "step": 15268 }, { "epoch": 0.5439905944386055, "grad_norm": 1.8268899952605087, "learning_rate": 9.062650583229435e-06, "loss": 0.5186, "step": 15269 }, { "epoch": 0.5440262215650997, "grad_norm": 1.470673773254826, "learning_rate": 9.061501733343539e-06, "loss": 0.548, "step": 15270 }, { "epoch": 0.5440618486915938, "grad_norm": 1.1383494059177919, "learning_rate": 9.060352895954403e-06, "loss": 0.4749, "step": 15271 }, { "epoch": 0.5440974758180879, "grad_norm": 1.1034537603178183, "learning_rate": 9.059204071077324e-06, "loss": 0.4189, "step": 15272 }, { "epoch": 0.5441331029445821, "grad_norm": 1.0833417451420309, "learning_rate": 9.058055258727605e-06, "loss": 0.476, "step": 15273 }, { "epoch": 0.5441687300710761, "grad_norm": 1.6053336540717227, "learning_rate": 9.056906458920534e-06, "loss": 0.5729, "step": 15274 }, { "epoch": 0.5442043571975702, "grad_norm": 1.1943479594407358, "learning_rate": 9.055757671671414e-06, "loss": 0.4442, "step": 15275 }, { "epoch": 0.5442399843240644, "grad_norm": 1.0455038926815583, "learning_rate": 9.054608896995538e-06, "loss": 0.4248, "step": 15276 }, { "epoch": 0.5442756114505585, "grad_norm": 1.002656845357787, "learning_rate": 9.053460134908213e-06, "loss": 0.4215, "step": 15277 }, { "epoch": 0.5443112385770525, "grad_norm": 1.2478081293697945, "learning_rate": 9.052311385424719e-06, "loss": 0.5008, "step": 15278 }, { "epoch": 0.5443468657035467, "grad_norm": 1.2736745618563372, "learning_rate": 9.051162648560366e-06, "loss": 0.4865, "step": 15279 }, { "epoch": 0.5443824928300408, "grad_norm": 11.717710539103901, "learning_rate": 9.050013924330444e-06, "loss": 0.5825, "step": 15280 }, { "epoch": 0.5444181199565349, "grad_norm": 1.4421802274317195, "learning_rate": 9.048865212750248e-06, "loss": 0.5812, "step": 15281 }, { "epoch": 0.5444537470830291, "grad_norm": 1.2230206417854297, "learning_rate": 9.04771651383508e-06, "loss": 0.4411, "step": 15282 }, { "epoch": 0.5444893742095231, "grad_norm": 1.8920416211573448, "learning_rate": 9.046567827600231e-06, "loss": 0.4529, "step": 15283 }, { "epoch": 0.5445250013360172, "grad_norm": 1.219538999525739, "learning_rate": 9.045419154061002e-06, "loss": 0.3953, "step": 15284 }, { "epoch": 0.5445606284625114, "grad_norm": 1.7992346867301767, "learning_rate": 9.044270493232678e-06, "loss": 0.6031, "step": 15285 }, { "epoch": 0.5445962555890055, "grad_norm": 1.6828250066715777, "learning_rate": 9.043121845130564e-06, "loss": 0.5137, "step": 15286 }, { "epoch": 0.5446318827154996, "grad_norm": 1.017752365152544, "learning_rate": 9.04197320976995e-06, "loss": 0.4259, "step": 15287 }, { "epoch": 0.5446675098419937, "grad_norm": 1.2731925602585774, "learning_rate": 9.040824587166136e-06, "loss": 0.5174, "step": 15288 }, { "epoch": 0.5447031369684878, "grad_norm": 1.5014078135795013, "learning_rate": 9.039675977334408e-06, "loss": 0.4966, "step": 15289 }, { "epoch": 0.5447387640949819, "grad_norm": 1.2467050527705799, "learning_rate": 9.038527380290069e-06, "loss": 0.4433, "step": 15290 }, { "epoch": 0.5447743912214761, "grad_norm": 0.9969297466609264, "learning_rate": 9.037378796048407e-06, "loss": 0.3705, "step": 15291 }, { "epoch": 0.5448100183479702, "grad_norm": 1.45696609594808, "learning_rate": 9.036230224624724e-06, "loss": 0.5735, "step": 15292 }, { "epoch": 0.5448456454744642, "grad_norm": 1.386498372638596, "learning_rate": 9.035081666034304e-06, "loss": 0.5649, "step": 15293 }, { "epoch": 0.5448812726009584, "grad_norm": 3.342749849363809, "learning_rate": 9.033933120292445e-06, "loss": 0.5341, "step": 15294 }, { "epoch": 0.5449168997274525, "grad_norm": 1.1206850835028994, "learning_rate": 9.032784587414448e-06, "loss": 0.4309, "step": 15295 }, { "epoch": 0.5449525268539466, "grad_norm": 1.1416077316546087, "learning_rate": 9.031636067415595e-06, "loss": 0.4218, "step": 15296 }, { "epoch": 0.5449881539804408, "grad_norm": 1.5987932174702315, "learning_rate": 9.030487560311187e-06, "loss": 0.4907, "step": 15297 }, { "epoch": 0.5450237811069348, "grad_norm": 1.623618887313869, "learning_rate": 9.029339066116512e-06, "loss": 0.5829, "step": 15298 }, { "epoch": 0.5450594082334289, "grad_norm": 1.405214984414723, "learning_rate": 9.028190584846872e-06, "loss": 0.5053, "step": 15299 }, { "epoch": 0.5450950353599231, "grad_norm": 2.5305532032060007, "learning_rate": 9.027042116517547e-06, "loss": 0.6053, "step": 15300 }, { "epoch": 0.5451306624864172, "grad_norm": 1.2019699923666245, "learning_rate": 9.02589366114384e-06, "loss": 0.4907, "step": 15301 }, { "epoch": 0.5451662896129112, "grad_norm": 1.006702524873069, "learning_rate": 9.024745218741037e-06, "loss": 0.4385, "step": 15302 }, { "epoch": 0.5452019167394054, "grad_norm": 1.161730137505723, "learning_rate": 9.023596789324439e-06, "loss": 0.446, "step": 15303 }, { "epoch": 0.5452375438658995, "grad_norm": 1.1160149619611142, "learning_rate": 9.022448372909326e-06, "loss": 0.4473, "step": 15304 }, { "epoch": 0.5452731709923936, "grad_norm": 1.6755373651528565, "learning_rate": 9.021299969510998e-06, "loss": 0.4959, "step": 15305 }, { "epoch": 0.5453087981188878, "grad_norm": 1.3358878553797866, "learning_rate": 9.020151579144745e-06, "loss": 0.5125, "step": 15306 }, { "epoch": 0.5453444252453818, "grad_norm": 1.190656181804734, "learning_rate": 9.019003201825858e-06, "loss": 0.4379, "step": 15307 }, { "epoch": 0.5453800523718759, "grad_norm": 1.9121899066978885, "learning_rate": 9.017854837569629e-06, "loss": 0.5322, "step": 15308 }, { "epoch": 0.5454156794983701, "grad_norm": 1.074683874576039, "learning_rate": 9.016706486391349e-06, "loss": 0.4176, "step": 15309 }, { "epoch": 0.5454513066248642, "grad_norm": 1.3553098602129183, "learning_rate": 9.015558148306313e-06, "loss": 0.5682, "step": 15310 }, { "epoch": 0.5454869337513583, "grad_norm": 1.0998439755889358, "learning_rate": 9.014409823329807e-06, "loss": 0.4501, "step": 15311 }, { "epoch": 0.5455225608778524, "grad_norm": 1.6102636201664111, "learning_rate": 9.013261511477122e-06, "loss": 0.5782, "step": 15312 }, { "epoch": 0.5455581880043465, "grad_norm": 1.1754366348015874, "learning_rate": 9.012113212763547e-06, "loss": 0.4787, "step": 15313 }, { "epoch": 0.5455938151308406, "grad_norm": 2.2575935823656574, "learning_rate": 9.01096492720438e-06, "loss": 0.5236, "step": 15314 }, { "epoch": 0.5456294422573348, "grad_norm": 1.9225491546307607, "learning_rate": 9.009816654814903e-06, "loss": 0.5678, "step": 15315 }, { "epoch": 0.5456650693838289, "grad_norm": 1.187418248796943, "learning_rate": 9.008668395610409e-06, "loss": 0.403, "step": 15316 }, { "epoch": 0.5457006965103229, "grad_norm": 1.0516981266101233, "learning_rate": 9.007520149606191e-06, "loss": 0.433, "step": 15317 }, { "epoch": 0.5457363236368171, "grad_norm": 1.2650371691574485, "learning_rate": 9.006371916817533e-06, "loss": 0.4514, "step": 15318 }, { "epoch": 0.5457719507633112, "grad_norm": 1.3676368142469262, "learning_rate": 9.005223697259732e-06, "loss": 0.5674, "step": 15319 }, { "epoch": 0.5458075778898053, "grad_norm": 1.9385423757372144, "learning_rate": 9.004075490948066e-06, "loss": 0.6034, "step": 15320 }, { "epoch": 0.5458432050162995, "grad_norm": 1.219194307355038, "learning_rate": 9.002927297897837e-06, "loss": 0.4418, "step": 15321 }, { "epoch": 0.5458788321427935, "grad_norm": 1.0488104264861953, "learning_rate": 9.001779118124325e-06, "loss": 0.3977, "step": 15322 }, { "epoch": 0.5459144592692876, "grad_norm": 1.0726974327741192, "learning_rate": 9.000630951642823e-06, "loss": 0.4096, "step": 15323 }, { "epoch": 0.5459500863957818, "grad_norm": 1.362120385879558, "learning_rate": 8.999482798468616e-06, "loss": 0.4822, "step": 15324 }, { "epoch": 0.5459857135222759, "grad_norm": 1.5259452906088329, "learning_rate": 8.998334658617e-06, "loss": 0.5704, "step": 15325 }, { "epoch": 0.54602134064877, "grad_norm": 1.7889481136839513, "learning_rate": 8.997186532103252e-06, "loss": 0.5932, "step": 15326 }, { "epoch": 0.5460569677752641, "grad_norm": 1.0481022044222958, "learning_rate": 8.996038418942669e-06, "loss": 0.4457, "step": 15327 }, { "epoch": 0.5460925949017582, "grad_norm": 1.360685715179734, "learning_rate": 8.994890319150536e-06, "loss": 0.5244, "step": 15328 }, { "epoch": 0.5461282220282523, "grad_norm": 1.7618959362175641, "learning_rate": 8.99374223274214e-06, "loss": 0.5342, "step": 15329 }, { "epoch": 0.5461638491547465, "grad_norm": 1.2602431605926891, "learning_rate": 8.992594159732773e-06, "loss": 0.4507, "step": 15330 }, { "epoch": 0.5461994762812405, "grad_norm": 1.1726370555861882, "learning_rate": 8.991446100137715e-06, "loss": 0.4843, "step": 15331 }, { "epoch": 0.5462351034077346, "grad_norm": 1.4588089130435598, "learning_rate": 8.990298053972259e-06, "loss": 0.4792, "step": 15332 }, { "epoch": 0.5462707305342288, "grad_norm": 1.8832066935953495, "learning_rate": 8.98915002125169e-06, "loss": 0.5171, "step": 15333 }, { "epoch": 0.5463063576607229, "grad_norm": 1.8437371258418958, "learning_rate": 8.988002001991295e-06, "loss": 0.526, "step": 15334 }, { "epoch": 0.546341984787217, "grad_norm": 1.03938514653346, "learning_rate": 8.986853996206359e-06, "loss": 0.4092, "step": 15335 }, { "epoch": 0.5463776119137111, "grad_norm": 2.866663800276483, "learning_rate": 8.985706003912175e-06, "loss": 0.4887, "step": 15336 }, { "epoch": 0.5464132390402052, "grad_norm": 1.2871643062053568, "learning_rate": 8.98455802512402e-06, "loss": 0.4432, "step": 15337 }, { "epoch": 0.5464488661666993, "grad_norm": 1.1539246961823544, "learning_rate": 8.983410059857184e-06, "loss": 0.4175, "step": 15338 }, { "epoch": 0.5464844932931935, "grad_norm": 1.2903224503467474, "learning_rate": 8.982262108126956e-06, "loss": 0.462, "step": 15339 }, { "epoch": 0.5465201204196876, "grad_norm": 1.3092992534231556, "learning_rate": 8.981114169948617e-06, "loss": 0.4436, "step": 15340 }, { "epoch": 0.5465557475461816, "grad_norm": 1.4011730242066573, "learning_rate": 8.97996624533746e-06, "loss": 0.51, "step": 15341 }, { "epoch": 0.5465913746726758, "grad_norm": 1.3148766010272404, "learning_rate": 8.97881833430876e-06, "loss": 0.4916, "step": 15342 }, { "epoch": 0.5466270017991699, "grad_norm": 1.2291352340968031, "learning_rate": 8.977670436877812e-06, "loss": 0.5097, "step": 15343 }, { "epoch": 0.546662628925664, "grad_norm": 1.640656023361307, "learning_rate": 8.976522553059891e-06, "loss": 0.5877, "step": 15344 }, { "epoch": 0.5466982560521582, "grad_norm": 1.0671203947937302, "learning_rate": 8.975374682870294e-06, "loss": 0.4324, "step": 15345 }, { "epoch": 0.5467338831786522, "grad_norm": 2.053951585091987, "learning_rate": 8.974226826324293e-06, "loss": 0.5485, "step": 15346 }, { "epoch": 0.5467695103051463, "grad_norm": 1.5398584200980907, "learning_rate": 8.973078983437185e-06, "loss": 0.5412, "step": 15347 }, { "epoch": 0.5468051374316405, "grad_norm": 1.6317519378739658, "learning_rate": 8.971931154224245e-06, "loss": 0.4775, "step": 15348 }, { "epoch": 0.5468407645581346, "grad_norm": 1.1891019843106607, "learning_rate": 8.97078333870076e-06, "loss": 0.4362, "step": 15349 }, { "epoch": 0.5468763916846286, "grad_norm": 1.1932233746523302, "learning_rate": 8.969635536882014e-06, "loss": 0.4575, "step": 15350 }, { "epoch": 0.5469120188111228, "grad_norm": 1.1737474941777217, "learning_rate": 8.96848774878329e-06, "loss": 0.4379, "step": 15351 }, { "epoch": 0.5469476459376169, "grad_norm": 1.546636005316613, "learning_rate": 8.967339974419877e-06, "loss": 0.5539, "step": 15352 }, { "epoch": 0.546983273064111, "grad_norm": 1.2443394543407873, "learning_rate": 8.96619221380705e-06, "loss": 0.4478, "step": 15353 }, { "epoch": 0.5470189001906052, "grad_norm": 1.1694916351899884, "learning_rate": 8.965044466960099e-06, "loss": 0.4739, "step": 15354 }, { "epoch": 0.5470545273170992, "grad_norm": 1.067527601674866, "learning_rate": 8.963896733894302e-06, "loss": 0.497, "step": 15355 }, { "epoch": 0.5470901544435933, "grad_norm": 1.1851913197694006, "learning_rate": 8.962749014624948e-06, "loss": 0.5077, "step": 15356 }, { "epoch": 0.5471257815700875, "grad_norm": 1.0461922964795083, "learning_rate": 8.961601309167309e-06, "loss": 0.434, "step": 15357 }, { "epoch": 0.5471614086965816, "grad_norm": 1.5784299561004325, "learning_rate": 8.960453617536682e-06, "loss": 0.5428, "step": 15358 }, { "epoch": 0.5471970358230757, "grad_norm": 1.5736029082999543, "learning_rate": 8.959305939748336e-06, "loss": 0.562, "step": 15359 }, { "epoch": 0.5472326629495698, "grad_norm": 1.4373199035891704, "learning_rate": 8.958158275817562e-06, "loss": 0.5299, "step": 15360 }, { "epoch": 0.5472682900760639, "grad_norm": 1.1441926963152467, "learning_rate": 8.957010625759635e-06, "loss": 0.4895, "step": 15361 }, { "epoch": 0.547303917202558, "grad_norm": 1.1328587974887336, "learning_rate": 8.955862989589842e-06, "loss": 0.4258, "step": 15362 }, { "epoch": 0.5473395443290522, "grad_norm": 1.9554555330560799, "learning_rate": 8.954715367323468e-06, "loss": 0.5443, "step": 15363 }, { "epoch": 0.5473751714555463, "grad_norm": 1.8084208621139704, "learning_rate": 8.953567758975785e-06, "loss": 0.6013, "step": 15364 }, { "epoch": 0.5474107985820403, "grad_norm": 1.076468372501585, "learning_rate": 8.952420164562078e-06, "loss": 0.443, "step": 15365 }, { "epoch": 0.5474464257085345, "grad_norm": 1.2289643555968979, "learning_rate": 8.95127258409763e-06, "loss": 0.4785, "step": 15366 }, { "epoch": 0.5474820528350286, "grad_norm": 1.7927009440169204, "learning_rate": 8.950125017597724e-06, "loss": 0.5906, "step": 15367 }, { "epoch": 0.5475176799615227, "grad_norm": 1.0560484760095041, "learning_rate": 8.948977465077632e-06, "loss": 0.4754, "step": 15368 }, { "epoch": 0.5475533070880169, "grad_norm": 1.474516927991819, "learning_rate": 8.947829926552643e-06, "loss": 0.5271, "step": 15369 }, { "epoch": 0.5475889342145109, "grad_norm": 1.3767602816277171, "learning_rate": 8.94668240203803e-06, "loss": 0.499, "step": 15370 }, { "epoch": 0.547624561341005, "grad_norm": 1.3658187087662648, "learning_rate": 8.94553489154908e-06, "loss": 0.4728, "step": 15371 }, { "epoch": 0.5476601884674992, "grad_norm": 1.475510602010047, "learning_rate": 8.94438739510107e-06, "loss": 0.4688, "step": 15372 }, { "epoch": 0.5476958155939933, "grad_norm": 1.1237317396335929, "learning_rate": 8.943239912709281e-06, "loss": 0.4717, "step": 15373 }, { "epoch": 0.5477314427204873, "grad_norm": 1.0942230723627508, "learning_rate": 8.942092444388988e-06, "loss": 0.4183, "step": 15374 }, { "epoch": 0.5477670698469815, "grad_norm": 1.140773479455288, "learning_rate": 8.940944990155474e-06, "loss": 0.4456, "step": 15375 }, { "epoch": 0.5478026969734756, "grad_norm": 1.150885255541761, "learning_rate": 8.939797550024018e-06, "loss": 0.4273, "step": 15376 }, { "epoch": 0.5478383240999697, "grad_norm": 1.4489661733637533, "learning_rate": 8.938650124009899e-06, "loss": 0.6182, "step": 15377 }, { "epoch": 0.5478739512264639, "grad_norm": 1.6387598152787821, "learning_rate": 8.9375027121284e-06, "loss": 0.5439, "step": 15378 }, { "epoch": 0.547909578352958, "grad_norm": 1.4394902967097236, "learning_rate": 8.936355314394788e-06, "loss": 0.5408, "step": 15379 }, { "epoch": 0.547945205479452, "grad_norm": 1.0671208455131196, "learning_rate": 8.935207930824351e-06, "loss": 0.4456, "step": 15380 }, { "epoch": 0.5479808326059462, "grad_norm": 1.0079718476453756, "learning_rate": 8.934060561432364e-06, "loss": 0.3953, "step": 15381 }, { "epoch": 0.5480164597324403, "grad_norm": 1.394814493639194, "learning_rate": 8.93291320623411e-06, "loss": 0.6137, "step": 15382 }, { "epoch": 0.5480520868589344, "grad_norm": 1.453844898815475, "learning_rate": 8.931765865244854e-06, "loss": 0.4769, "step": 15383 }, { "epoch": 0.5480877139854285, "grad_norm": 1.0783733471962156, "learning_rate": 8.93061853847989e-06, "loss": 0.406, "step": 15384 }, { "epoch": 0.5481233411119226, "grad_norm": 1.103900919779051, "learning_rate": 8.929471225954484e-06, "loss": 0.4865, "step": 15385 }, { "epoch": 0.5481589682384167, "grad_norm": 1.1319174634997666, "learning_rate": 8.928323927683916e-06, "loss": 0.4704, "step": 15386 }, { "epoch": 0.5481945953649109, "grad_norm": 1.6010418022574908, "learning_rate": 8.927176643683467e-06, "loss": 0.5537, "step": 15387 }, { "epoch": 0.548230222491405, "grad_norm": 1.117654934962395, "learning_rate": 8.926029373968407e-06, "loss": 0.4429, "step": 15388 }, { "epoch": 0.548265849617899, "grad_norm": 1.1423555425835623, "learning_rate": 8.924882118554022e-06, "loss": 0.4336, "step": 15389 }, { "epoch": 0.5483014767443932, "grad_norm": 1.6530378037157352, "learning_rate": 8.92373487745558e-06, "loss": 0.6167, "step": 15390 }, { "epoch": 0.5483371038708873, "grad_norm": 1.0421478919588638, "learning_rate": 8.92258765068836e-06, "loss": 0.4534, "step": 15391 }, { "epoch": 0.5483727309973814, "grad_norm": 1.0664836128881936, "learning_rate": 8.921440438267637e-06, "loss": 0.4226, "step": 15392 }, { "epoch": 0.5484083581238756, "grad_norm": 1.3276097204214194, "learning_rate": 8.920293240208694e-06, "loss": 0.5075, "step": 15393 }, { "epoch": 0.5484439852503696, "grad_norm": 1.3551536054844815, "learning_rate": 8.919146056526796e-06, "loss": 0.5222, "step": 15394 }, { "epoch": 0.5484796123768637, "grad_norm": 1.6124624807289707, "learning_rate": 8.917998887237227e-06, "loss": 0.5588, "step": 15395 }, { "epoch": 0.5485152395033579, "grad_norm": 1.108482951282803, "learning_rate": 8.916851732355256e-06, "loss": 0.4255, "step": 15396 }, { "epoch": 0.548550866629852, "grad_norm": 1.6306268109963422, "learning_rate": 8.91570459189616e-06, "loss": 0.5072, "step": 15397 }, { "epoch": 0.548586493756346, "grad_norm": 1.1498873390632258, "learning_rate": 8.91455746587522e-06, "loss": 0.4481, "step": 15398 }, { "epoch": 0.5486221208828402, "grad_norm": 1.7929622578822006, "learning_rate": 8.913410354307701e-06, "loss": 0.5045, "step": 15399 }, { "epoch": 0.5486577480093343, "grad_norm": 1.3970391917592697, "learning_rate": 8.912263257208888e-06, "loss": 0.6235, "step": 15400 }, { "epoch": 0.5486933751358284, "grad_norm": 1.042785971784245, "learning_rate": 8.911116174594047e-06, "loss": 0.4183, "step": 15401 }, { "epoch": 0.5487290022623226, "grad_norm": 1.1294831186184702, "learning_rate": 8.909969106478456e-06, "loss": 0.4656, "step": 15402 }, { "epoch": 0.5487646293888166, "grad_norm": 1.5448615292686252, "learning_rate": 8.908822052877387e-06, "loss": 0.5407, "step": 15403 }, { "epoch": 0.5488002565153107, "grad_norm": 2.356183898999832, "learning_rate": 8.907675013806118e-06, "loss": 0.571, "step": 15404 }, { "epoch": 0.5488358836418049, "grad_norm": 1.5545647091887913, "learning_rate": 8.906527989279917e-06, "loss": 0.6401, "step": 15405 }, { "epoch": 0.548871510768299, "grad_norm": 1.2213392454870768, "learning_rate": 8.905380979314062e-06, "loss": 0.516, "step": 15406 }, { "epoch": 0.5489071378947931, "grad_norm": 1.5229045487731419, "learning_rate": 8.904233983923821e-06, "loss": 0.4622, "step": 15407 }, { "epoch": 0.5489427650212872, "grad_norm": 1.9698557836271045, "learning_rate": 8.903087003124471e-06, "loss": 0.5145, "step": 15408 }, { "epoch": 0.5489783921477813, "grad_norm": 1.2321491629903782, "learning_rate": 8.901940036931287e-06, "loss": 0.4408, "step": 15409 }, { "epoch": 0.5490140192742754, "grad_norm": 1.1382812566403318, "learning_rate": 8.900793085359536e-06, "loss": 0.4216, "step": 15410 }, { "epoch": 0.5490496464007696, "grad_norm": 1.0287347333721422, "learning_rate": 8.899646148424499e-06, "loss": 0.4462, "step": 15411 }, { "epoch": 0.5490852735272637, "grad_norm": 1.0969629100764342, "learning_rate": 8.898499226141438e-06, "loss": 0.4127, "step": 15412 }, { "epoch": 0.5491209006537577, "grad_norm": 1.5302290199964241, "learning_rate": 8.89735231852563e-06, "loss": 0.5766, "step": 15413 }, { "epoch": 0.5491565277802519, "grad_norm": 1.4048675341506238, "learning_rate": 8.896205425592346e-06, "loss": 0.4466, "step": 15414 }, { "epoch": 0.549192154906746, "grad_norm": 1.153445403416878, "learning_rate": 8.895058547356862e-06, "loss": 0.5135, "step": 15415 }, { "epoch": 0.5492277820332401, "grad_norm": 1.06149694963926, "learning_rate": 8.893911683834442e-06, "loss": 0.4626, "step": 15416 }, { "epoch": 0.5492634091597343, "grad_norm": 1.0464103336409953, "learning_rate": 8.892764835040362e-06, "loss": 0.4456, "step": 15417 }, { "epoch": 0.5492990362862283, "grad_norm": 1.112135974900418, "learning_rate": 8.89161800098989e-06, "loss": 0.3906, "step": 15418 }, { "epoch": 0.5493346634127224, "grad_norm": 1.633635827553135, "learning_rate": 8.890471181698304e-06, "loss": 0.5313, "step": 15419 }, { "epoch": 0.5493702905392166, "grad_norm": 1.094838549927255, "learning_rate": 8.889324377180866e-06, "loss": 0.44, "step": 15420 }, { "epoch": 0.5494059176657107, "grad_norm": 1.1543684471361852, "learning_rate": 8.88817758745285e-06, "loss": 0.4929, "step": 15421 }, { "epoch": 0.5494415447922048, "grad_norm": 2.013021995401313, "learning_rate": 8.88703081252953e-06, "loss": 0.5492, "step": 15422 }, { "epoch": 0.5494771719186989, "grad_norm": 1.6512681458505827, "learning_rate": 8.885884052426168e-06, "loss": 0.4739, "step": 15423 }, { "epoch": 0.549512799045193, "grad_norm": 1.785849565407893, "learning_rate": 8.88473730715804e-06, "loss": 0.5425, "step": 15424 }, { "epoch": 0.5495484261716871, "grad_norm": 1.1224620987157625, "learning_rate": 8.883590576740416e-06, "loss": 0.4237, "step": 15425 }, { "epoch": 0.5495840532981813, "grad_norm": 1.2766520902365652, "learning_rate": 8.882443861188566e-06, "loss": 0.5093, "step": 15426 }, { "epoch": 0.5496196804246753, "grad_norm": 1.0820596325848653, "learning_rate": 8.881297160517752e-06, "loss": 0.4438, "step": 15427 }, { "epoch": 0.5496553075511694, "grad_norm": 1.475381389335449, "learning_rate": 8.880150474743251e-06, "loss": 0.6425, "step": 15428 }, { "epoch": 0.5496909346776636, "grad_norm": 1.6718237866309957, "learning_rate": 8.879003803880328e-06, "loss": 0.5536, "step": 15429 }, { "epoch": 0.5497265618041577, "grad_norm": 1.1078757324851776, "learning_rate": 8.877857147944256e-06, "loss": 0.4764, "step": 15430 }, { "epoch": 0.5497621889306518, "grad_norm": 1.2998942528643125, "learning_rate": 8.876710506950296e-06, "loss": 0.5643, "step": 15431 }, { "epoch": 0.549797816057146, "grad_norm": 1.0030631315483094, "learning_rate": 8.875563880913719e-06, "loss": 0.4392, "step": 15432 }, { "epoch": 0.54983344318364, "grad_norm": 1.2799354362293875, "learning_rate": 8.8744172698498e-06, "loss": 0.5809, "step": 15433 }, { "epoch": 0.5498690703101341, "grad_norm": 1.4236286259040147, "learning_rate": 8.8732706737738e-06, "loss": 0.5572, "step": 15434 }, { "epoch": 0.5499046974366283, "grad_norm": 1.0653572308900696, "learning_rate": 8.872124092700988e-06, "loss": 0.4358, "step": 15435 }, { "epoch": 0.5499403245631224, "grad_norm": 1.1191381643209222, "learning_rate": 8.870977526646632e-06, "loss": 0.399, "step": 15436 }, { "epoch": 0.5499759516896164, "grad_norm": 1.5509085863984409, "learning_rate": 8.869830975626004e-06, "loss": 0.5108, "step": 15437 }, { "epoch": 0.5500115788161106, "grad_norm": 1.0439123655126292, "learning_rate": 8.86868443965436e-06, "loss": 0.4098, "step": 15438 }, { "epoch": 0.5500472059426047, "grad_norm": 1.9441400468221852, "learning_rate": 8.867537918746977e-06, "loss": 0.6615, "step": 15439 }, { "epoch": 0.5500828330690988, "grad_norm": 1.0999622789623793, "learning_rate": 8.866391412919116e-06, "loss": 0.4327, "step": 15440 }, { "epoch": 0.550118460195593, "grad_norm": 1.0898671840569283, "learning_rate": 8.86524492218605e-06, "loss": 0.4524, "step": 15441 }, { "epoch": 0.550154087322087, "grad_norm": 1.3635630456276169, "learning_rate": 8.864098446563037e-06, "loss": 0.5114, "step": 15442 }, { "epoch": 0.5501897144485811, "grad_norm": 1.121313940161429, "learning_rate": 8.862951986065346e-06, "loss": 0.4447, "step": 15443 }, { "epoch": 0.5502253415750753, "grad_norm": 1.043933900121125, "learning_rate": 8.861805540708246e-06, "loss": 0.425, "step": 15444 }, { "epoch": 0.5502609687015694, "grad_norm": 1.6334868458694183, "learning_rate": 8.860659110507e-06, "loss": 0.5882, "step": 15445 }, { "epoch": 0.5502965958280635, "grad_norm": 1.192738003567465, "learning_rate": 8.859512695476876e-06, "loss": 0.4312, "step": 15446 }, { "epoch": 0.5503322229545576, "grad_norm": 2.0481426055980623, "learning_rate": 8.858366295633137e-06, "loss": 0.4849, "step": 15447 }, { "epoch": 0.5503678500810517, "grad_norm": 1.0616737646457601, "learning_rate": 8.85721991099105e-06, "loss": 0.4087, "step": 15448 }, { "epoch": 0.5504034772075458, "grad_norm": 1.117422646722183, "learning_rate": 8.856073541565876e-06, "loss": 0.4893, "step": 15449 }, { "epoch": 0.55043910433404, "grad_norm": 1.0263387390929815, "learning_rate": 8.854927187372884e-06, "loss": 0.425, "step": 15450 }, { "epoch": 0.550474731460534, "grad_norm": 1.8960674607933545, "learning_rate": 8.853780848427335e-06, "loss": 0.5791, "step": 15451 }, { "epoch": 0.5505103585870281, "grad_norm": 1.1476013775468439, "learning_rate": 8.852634524744499e-06, "loss": 0.4963, "step": 15452 }, { "epoch": 0.5505459857135223, "grad_norm": 1.1274884640445988, "learning_rate": 8.851488216339632e-06, "loss": 0.4403, "step": 15453 }, { "epoch": 0.5505816128400164, "grad_norm": 1.0827665315993216, "learning_rate": 8.850341923228004e-06, "loss": 0.4132, "step": 15454 }, { "epoch": 0.5506172399665105, "grad_norm": 1.1935044099609273, "learning_rate": 8.849195645424877e-06, "loss": 0.4348, "step": 15455 }, { "epoch": 0.5506528670930046, "grad_norm": 1.1230414796028438, "learning_rate": 8.848049382945513e-06, "loss": 0.4561, "step": 15456 }, { "epoch": 0.5506884942194987, "grad_norm": 1.1899476550726853, "learning_rate": 8.84690313580518e-06, "loss": 0.4099, "step": 15457 }, { "epoch": 0.5507241213459928, "grad_norm": 1.1648575832144532, "learning_rate": 8.845756904019132e-06, "loss": 0.491, "step": 15458 }, { "epoch": 0.550759748472487, "grad_norm": 1.3836052393253855, "learning_rate": 8.844610687602643e-06, "loss": 0.5794, "step": 15459 }, { "epoch": 0.5507953755989811, "grad_norm": 1.9346350933221417, "learning_rate": 8.843464486570967e-06, "loss": 0.6461, "step": 15460 }, { "epoch": 0.5508310027254751, "grad_norm": 1.0176394230284944, "learning_rate": 8.842318300939372e-06, "loss": 0.4366, "step": 15461 }, { "epoch": 0.5508666298519693, "grad_norm": 1.5139430811724712, "learning_rate": 8.841172130723116e-06, "loss": 0.5984, "step": 15462 }, { "epoch": 0.5509022569784634, "grad_norm": 1.0420290122401288, "learning_rate": 8.840025975937467e-06, "loss": 0.4099, "step": 15463 }, { "epoch": 0.5509378841049575, "grad_norm": 1.1409911638686137, "learning_rate": 8.83887983659768e-06, "loss": 0.4279, "step": 15464 }, { "epoch": 0.5509735112314517, "grad_norm": 1.556741912779108, "learning_rate": 8.83773371271902e-06, "loss": 0.5121, "step": 15465 }, { "epoch": 0.5510091383579457, "grad_norm": 1.201283557246612, "learning_rate": 8.836587604316746e-06, "loss": 0.4829, "step": 15466 }, { "epoch": 0.5510447654844398, "grad_norm": 1.7813881128431424, "learning_rate": 8.83544151140612e-06, "loss": 0.5185, "step": 15467 }, { "epoch": 0.551080392610934, "grad_norm": 1.0416125532779448, "learning_rate": 8.83429543400241e-06, "loss": 0.4399, "step": 15468 }, { "epoch": 0.5511160197374281, "grad_norm": 1.022311070963223, "learning_rate": 8.833149372120867e-06, "loss": 0.4528, "step": 15469 }, { "epoch": 0.5511516468639222, "grad_norm": 1.377838122573322, "learning_rate": 8.832003325776758e-06, "loss": 0.4791, "step": 15470 }, { "epoch": 0.5511872739904163, "grad_norm": 1.5194668350069054, "learning_rate": 8.830857294985337e-06, "loss": 0.4828, "step": 15471 }, { "epoch": 0.5512229011169104, "grad_norm": 1.5748968894129707, "learning_rate": 8.829711279761873e-06, "loss": 0.538, "step": 15472 }, { "epoch": 0.5512585282434045, "grad_norm": 1.8646793173488247, "learning_rate": 8.828565280121619e-06, "loss": 0.6297, "step": 15473 }, { "epoch": 0.5512941553698987, "grad_norm": 1.1265271538783894, "learning_rate": 8.82741929607984e-06, "loss": 0.4422, "step": 15474 }, { "epoch": 0.5513297824963928, "grad_norm": 1.4344961823776705, "learning_rate": 8.826273327651788e-06, "loss": 0.5577, "step": 15475 }, { "epoch": 0.5513654096228868, "grad_norm": 1.5198138758836817, "learning_rate": 8.82512737485273e-06, "loss": 0.4844, "step": 15476 }, { "epoch": 0.551401036749381, "grad_norm": 1.4533464425412896, "learning_rate": 8.823981437697921e-06, "loss": 0.5604, "step": 15477 }, { "epoch": 0.5514366638758751, "grad_norm": 1.204737329879618, "learning_rate": 8.822835516202619e-06, "loss": 0.4581, "step": 15478 }, { "epoch": 0.5514722910023692, "grad_norm": 1.1608867613953058, "learning_rate": 8.821689610382091e-06, "loss": 0.435, "step": 15479 }, { "epoch": 0.5515079181288634, "grad_norm": 1.1177824579463285, "learning_rate": 8.820543720251584e-06, "loss": 0.4357, "step": 15480 }, { "epoch": 0.5515435452553574, "grad_norm": 1.1735977520821084, "learning_rate": 8.819397845826365e-06, "loss": 0.5039, "step": 15481 }, { "epoch": 0.5515791723818515, "grad_norm": 1.09335106157679, "learning_rate": 8.818251987121686e-06, "loss": 0.4423, "step": 15482 }, { "epoch": 0.5516147995083457, "grad_norm": 1.097730797592329, "learning_rate": 8.817106144152812e-06, "loss": 0.4516, "step": 15483 }, { "epoch": 0.5516504266348398, "grad_norm": 1.380649171364496, "learning_rate": 8.815960316934991e-06, "loss": 0.4473, "step": 15484 }, { "epoch": 0.5516860537613338, "grad_norm": 1.3788198965054272, "learning_rate": 8.814814505483492e-06, "loss": 0.5369, "step": 15485 }, { "epoch": 0.551721680887828, "grad_norm": 1.0381443881699453, "learning_rate": 8.813668709813561e-06, "loss": 0.4451, "step": 15486 }, { "epoch": 0.5517573080143221, "grad_norm": 1.9613661308909738, "learning_rate": 8.812522929940462e-06, "loss": 0.6054, "step": 15487 }, { "epoch": 0.5517929351408162, "grad_norm": 1.1381840031886272, "learning_rate": 8.81137716587945e-06, "loss": 0.4835, "step": 15488 }, { "epoch": 0.5518285622673104, "grad_norm": 1.0720510981365163, "learning_rate": 8.81023141764578e-06, "loss": 0.3957, "step": 15489 }, { "epoch": 0.5518641893938044, "grad_norm": 1.012282504667755, "learning_rate": 8.809085685254717e-06, "loss": 0.4331, "step": 15490 }, { "epoch": 0.5518998165202985, "grad_norm": 1.4022637548881256, "learning_rate": 8.807939968721504e-06, "loss": 0.5291, "step": 15491 }, { "epoch": 0.5519354436467927, "grad_norm": 1.0913491064371414, "learning_rate": 8.806794268061407e-06, "loss": 0.4761, "step": 15492 }, { "epoch": 0.5519710707732868, "grad_norm": 1.3747296533876692, "learning_rate": 8.805648583289675e-06, "loss": 0.5952, "step": 15493 }, { "epoch": 0.5520066978997809, "grad_norm": 1.0938500378987313, "learning_rate": 8.804502914421573e-06, "loss": 0.444, "step": 15494 }, { "epoch": 0.552042325026275, "grad_norm": 1.1168325109832133, "learning_rate": 8.803357261472344e-06, "loss": 0.406, "step": 15495 }, { "epoch": 0.5520779521527691, "grad_norm": 1.0968034805279503, "learning_rate": 8.802211624457251e-06, "loss": 0.4687, "step": 15496 }, { "epoch": 0.5521135792792632, "grad_norm": 1.303055557933984, "learning_rate": 8.801066003391547e-06, "loss": 0.5066, "step": 15497 }, { "epoch": 0.5521492064057574, "grad_norm": 1.1358206088183267, "learning_rate": 8.799920398290489e-06, "loss": 0.4667, "step": 15498 }, { "epoch": 0.5521848335322515, "grad_norm": 1.5520402036203167, "learning_rate": 8.798774809169328e-06, "loss": 0.5493, "step": 15499 }, { "epoch": 0.5522204606587455, "grad_norm": 1.3648589398137434, "learning_rate": 8.79762923604332e-06, "loss": 0.5369, "step": 15500 }, { "epoch": 0.5522560877852397, "grad_norm": 1.5106382838422916, "learning_rate": 8.796483678927724e-06, "loss": 0.5696, "step": 15501 }, { "epoch": 0.5522917149117338, "grad_norm": 1.0250268685137407, "learning_rate": 8.795338137837783e-06, "loss": 0.4135, "step": 15502 }, { "epoch": 0.5523273420382279, "grad_norm": 1.7337887858983978, "learning_rate": 8.794192612788758e-06, "loss": 0.6458, "step": 15503 }, { "epoch": 0.552362969164722, "grad_norm": 1.6997099596879224, "learning_rate": 8.793047103795902e-06, "loss": 0.5657, "step": 15504 }, { "epoch": 0.5523985962912161, "grad_norm": 1.1507314720949804, "learning_rate": 8.791901610874469e-06, "loss": 0.4686, "step": 15505 }, { "epoch": 0.5524342234177102, "grad_norm": 1.1900908562614985, "learning_rate": 8.790756134039708e-06, "loss": 0.4659, "step": 15506 }, { "epoch": 0.5524698505442044, "grad_norm": 1.4451692179969395, "learning_rate": 8.789610673306876e-06, "loss": 0.515, "step": 15507 }, { "epoch": 0.5525054776706985, "grad_norm": 1.3965099304789101, "learning_rate": 8.788465228691222e-06, "loss": 0.4762, "step": 15508 }, { "epoch": 0.5525411047971925, "grad_norm": 1.119922084088022, "learning_rate": 8.787319800208004e-06, "loss": 0.4529, "step": 15509 }, { "epoch": 0.5525767319236867, "grad_norm": 1.3793313124071218, "learning_rate": 8.786174387872464e-06, "loss": 0.5443, "step": 15510 }, { "epoch": 0.5526123590501808, "grad_norm": 1.5616573199236803, "learning_rate": 8.785028991699867e-06, "loss": 0.4494, "step": 15511 }, { "epoch": 0.5526479861766749, "grad_norm": 0.9684994419789151, "learning_rate": 8.783883611705455e-06, "loss": 0.46, "step": 15512 }, { "epoch": 0.5526836133031691, "grad_norm": 1.1415276118210527, "learning_rate": 8.78273824790448e-06, "loss": 0.4702, "step": 15513 }, { "epoch": 0.5527192404296631, "grad_norm": 1.0361716756303785, "learning_rate": 8.781592900312201e-06, "loss": 0.4749, "step": 15514 }, { "epoch": 0.5527548675561572, "grad_norm": 1.3785809774610938, "learning_rate": 8.780447568943862e-06, "loss": 0.5047, "step": 15515 }, { "epoch": 0.5527904946826514, "grad_norm": 1.1029844846925694, "learning_rate": 8.779302253814718e-06, "loss": 0.4731, "step": 15516 }, { "epoch": 0.5528261218091455, "grad_norm": 1.752778286209103, "learning_rate": 8.778156954940016e-06, "loss": 0.5123, "step": 15517 }, { "epoch": 0.5528617489356396, "grad_norm": 1.5290355883472597, "learning_rate": 8.777011672335008e-06, "loss": 0.5403, "step": 15518 }, { "epoch": 0.5528973760621337, "grad_norm": 1.0612834240651166, "learning_rate": 8.775866406014944e-06, "loss": 0.4689, "step": 15519 }, { "epoch": 0.5529330031886278, "grad_norm": 1.3422276771284933, "learning_rate": 8.774721155995078e-06, "loss": 0.5304, "step": 15520 }, { "epoch": 0.5529686303151219, "grad_norm": 1.045565300129624, "learning_rate": 8.773575922290649e-06, "loss": 0.4368, "step": 15521 }, { "epoch": 0.5530042574416161, "grad_norm": 1.4491302529638597, "learning_rate": 8.77243070491692e-06, "loss": 0.5717, "step": 15522 }, { "epoch": 0.5530398845681102, "grad_norm": 1.1453956780575187, "learning_rate": 8.771285503889132e-06, "loss": 0.4785, "step": 15523 }, { "epoch": 0.5530755116946042, "grad_norm": 1.6626134186735844, "learning_rate": 8.770140319222535e-06, "loss": 0.5538, "step": 15524 }, { "epoch": 0.5531111388210984, "grad_norm": 1.3222639804136125, "learning_rate": 8.768995150932382e-06, "loss": 0.5584, "step": 15525 }, { "epoch": 0.5531467659475925, "grad_norm": 2.8669020292507033, "learning_rate": 8.767849999033918e-06, "loss": 0.4882, "step": 15526 }, { "epoch": 0.5531823930740866, "grad_norm": 1.2428244333098162, "learning_rate": 8.766704863542395e-06, "loss": 0.4452, "step": 15527 }, { "epoch": 0.5532180202005808, "grad_norm": 1.4103796597324225, "learning_rate": 8.765559744473054e-06, "loss": 0.6222, "step": 15528 }, { "epoch": 0.5532536473270748, "grad_norm": 1.322229560008594, "learning_rate": 8.76441464184115e-06, "loss": 0.4575, "step": 15529 }, { "epoch": 0.5532892744535689, "grad_norm": 1.1261904350799257, "learning_rate": 8.763269555661928e-06, "loss": 0.4647, "step": 15530 }, { "epoch": 0.5533249015800631, "grad_norm": 1.674327095597346, "learning_rate": 8.76212448595064e-06, "loss": 0.5313, "step": 15531 }, { "epoch": 0.5533605287065572, "grad_norm": 1.02320648069831, "learning_rate": 8.760979432722525e-06, "loss": 0.4949, "step": 15532 }, { "epoch": 0.5533961558330512, "grad_norm": 0.9491389117621564, "learning_rate": 8.759834395992836e-06, "loss": 0.4046, "step": 15533 }, { "epoch": 0.5534317829595454, "grad_norm": 2.1588669089738954, "learning_rate": 8.758689375776819e-06, "loss": 0.5902, "step": 15534 }, { "epoch": 0.5534674100860395, "grad_norm": 1.0908903440115603, "learning_rate": 8.75754437208972e-06, "loss": 0.4197, "step": 15535 }, { "epoch": 0.5535030372125336, "grad_norm": 1.0557843254887822, "learning_rate": 8.756399384946788e-06, "loss": 0.4469, "step": 15536 }, { "epoch": 0.5535386643390278, "grad_norm": 1.045870363665628, "learning_rate": 8.755254414363264e-06, "loss": 0.4706, "step": 15537 }, { "epoch": 0.5535742914655218, "grad_norm": 1.417763117046183, "learning_rate": 8.754109460354403e-06, "loss": 0.5134, "step": 15538 }, { "epoch": 0.5536099185920159, "grad_norm": 1.487977814372698, "learning_rate": 8.752964522935442e-06, "loss": 0.5608, "step": 15539 }, { "epoch": 0.5536455457185101, "grad_norm": 1.124090050571964, "learning_rate": 8.751819602121631e-06, "loss": 0.4535, "step": 15540 }, { "epoch": 0.5536811728450042, "grad_norm": 1.9458534526046243, "learning_rate": 8.750674697928213e-06, "loss": 0.67, "step": 15541 }, { "epoch": 0.5537167999714983, "grad_norm": 1.4913193810144345, "learning_rate": 8.749529810370436e-06, "loss": 0.5409, "step": 15542 }, { "epoch": 0.5537524270979924, "grad_norm": 4.18432990732024, "learning_rate": 8.748384939463543e-06, "loss": 0.567, "step": 15543 }, { "epoch": 0.5537880542244865, "grad_norm": 1.2474577115574732, "learning_rate": 8.74724008522278e-06, "loss": 0.485, "step": 15544 }, { "epoch": 0.5538236813509806, "grad_norm": 1.1462860481815107, "learning_rate": 8.746095247663388e-06, "loss": 0.4508, "step": 15545 }, { "epoch": 0.5538593084774748, "grad_norm": 0.965972390174659, "learning_rate": 8.74495042680062e-06, "loss": 0.4518, "step": 15546 }, { "epoch": 0.5538949356039689, "grad_norm": 1.4803105548187632, "learning_rate": 8.743805622649708e-06, "loss": 0.629, "step": 15547 }, { "epoch": 0.5539305627304629, "grad_norm": 1.6178988934477176, "learning_rate": 8.742660835225902e-06, "loss": 0.5209, "step": 15548 }, { "epoch": 0.5539661898569571, "grad_norm": 1.2631267898225698, "learning_rate": 8.741516064544452e-06, "loss": 0.4707, "step": 15549 }, { "epoch": 0.5540018169834512, "grad_norm": 1.597378418183742, "learning_rate": 8.74037131062059e-06, "loss": 0.5495, "step": 15550 }, { "epoch": 0.5540374441099453, "grad_norm": 1.4705672617586705, "learning_rate": 8.739226573469568e-06, "loss": 0.5911, "step": 15551 }, { "epoch": 0.5540730712364395, "grad_norm": 2.4864849328548035, "learning_rate": 8.73808185310662e-06, "loss": 0.547, "step": 15552 }, { "epoch": 0.5541086983629335, "grad_norm": 1.3926900862774503, "learning_rate": 8.736937149547003e-06, "loss": 0.6114, "step": 15553 }, { "epoch": 0.5541443254894276, "grad_norm": 1.6679432395346554, "learning_rate": 8.735792462805943e-06, "loss": 0.5565, "step": 15554 }, { "epoch": 0.5541799526159218, "grad_norm": 1.57874936962684, "learning_rate": 8.734647792898693e-06, "loss": 0.5618, "step": 15555 }, { "epoch": 0.5542155797424159, "grad_norm": 1.2913106401804784, "learning_rate": 8.733503139840491e-06, "loss": 0.4447, "step": 15556 }, { "epoch": 0.5542512068689099, "grad_norm": 1.0513630876536204, "learning_rate": 8.732358503646585e-06, "loss": 0.4771, "step": 15557 }, { "epoch": 0.5542868339954041, "grad_norm": 1.1285422237492324, "learning_rate": 8.731213884332205e-06, "loss": 0.4242, "step": 15558 }, { "epoch": 0.5543224611218982, "grad_norm": 1.0658374227632081, "learning_rate": 8.730069281912597e-06, "loss": 0.4695, "step": 15559 }, { "epoch": 0.5543580882483923, "grad_norm": 1.104500345418406, "learning_rate": 8.728924696403011e-06, "loss": 0.4437, "step": 15560 }, { "epoch": 0.5543937153748865, "grad_norm": 1.5540831331330598, "learning_rate": 8.727780127818676e-06, "loss": 0.554, "step": 15561 }, { "epoch": 0.5544293425013805, "grad_norm": 1.6872910173383981, "learning_rate": 8.726635576174843e-06, "loss": 0.5119, "step": 15562 }, { "epoch": 0.5544649696278746, "grad_norm": 1.136634081208238, "learning_rate": 8.725491041486742e-06, "loss": 0.4385, "step": 15563 }, { "epoch": 0.5545005967543688, "grad_norm": 1.7420232100957724, "learning_rate": 8.724346523769624e-06, "loss": 0.5944, "step": 15564 }, { "epoch": 0.5545362238808629, "grad_norm": 2.202258876100647, "learning_rate": 8.723202023038719e-06, "loss": 0.5726, "step": 15565 }, { "epoch": 0.554571851007357, "grad_norm": 2.369803021226212, "learning_rate": 8.722057539309274e-06, "loss": 0.6635, "step": 15566 }, { "epoch": 0.5546074781338511, "grad_norm": 1.5547824386074591, "learning_rate": 8.720913072596524e-06, "loss": 0.5725, "step": 15567 }, { "epoch": 0.5546431052603452, "grad_norm": 1.163478341669269, "learning_rate": 8.719768622915714e-06, "loss": 0.4624, "step": 15568 }, { "epoch": 0.5546787323868393, "grad_norm": 1.3089460647034097, "learning_rate": 8.718624190282076e-06, "loss": 0.5756, "step": 15569 }, { "epoch": 0.5547143595133335, "grad_norm": 1.4018038135998674, "learning_rate": 8.717479774710853e-06, "loss": 0.5249, "step": 15570 }, { "epoch": 0.5547499866398276, "grad_norm": 0.9854415932144474, "learning_rate": 8.716335376217285e-06, "loss": 0.4456, "step": 15571 }, { "epoch": 0.5547856137663216, "grad_norm": 1.4512694177464462, "learning_rate": 8.715190994816607e-06, "loss": 0.6301, "step": 15572 }, { "epoch": 0.5548212408928158, "grad_norm": 1.4732822362064226, "learning_rate": 8.714046630524061e-06, "loss": 0.5314, "step": 15573 }, { "epoch": 0.5548568680193099, "grad_norm": 2.129089442188213, "learning_rate": 8.712902283354879e-06, "loss": 0.5608, "step": 15574 }, { "epoch": 0.554892495145804, "grad_norm": 1.4648632330989622, "learning_rate": 8.711757953324309e-06, "loss": 0.5218, "step": 15575 }, { "epoch": 0.5549281222722982, "grad_norm": 1.8424299506611639, "learning_rate": 8.710613640447578e-06, "loss": 0.4899, "step": 15576 }, { "epoch": 0.5549637493987922, "grad_norm": 1.4007619742209607, "learning_rate": 8.709469344739928e-06, "loss": 0.5165, "step": 15577 }, { "epoch": 0.5549993765252863, "grad_norm": 1.032544582521873, "learning_rate": 8.708325066216595e-06, "loss": 0.4826, "step": 15578 }, { "epoch": 0.5550350036517805, "grad_norm": 1.6164495476526164, "learning_rate": 8.707180804892822e-06, "loss": 0.5524, "step": 15579 }, { "epoch": 0.5550706307782746, "grad_norm": 1.1716129241121707, "learning_rate": 8.706036560783832e-06, "loss": 0.4549, "step": 15580 }, { "epoch": 0.5551062579047686, "grad_norm": 1.2739746009155255, "learning_rate": 8.704892333904874e-06, "loss": 0.4788, "step": 15581 }, { "epoch": 0.5551418850312628, "grad_norm": 1.1190907038812006, "learning_rate": 8.703748124271179e-06, "loss": 0.4509, "step": 15582 }, { "epoch": 0.5551775121577569, "grad_norm": 1.4472052813426137, "learning_rate": 8.702603931897983e-06, "loss": 0.4433, "step": 15583 }, { "epoch": 0.555213139284251, "grad_norm": 1.24971568877471, "learning_rate": 8.701459756800525e-06, "loss": 0.4626, "step": 15584 }, { "epoch": 0.5552487664107452, "grad_norm": 1.1119099036302915, "learning_rate": 8.700315598994034e-06, "loss": 0.4212, "step": 15585 }, { "epoch": 0.5552843935372392, "grad_norm": 1.1495557667615028, "learning_rate": 8.699171458493754e-06, "loss": 0.4619, "step": 15586 }, { "epoch": 0.5553200206637333, "grad_norm": 1.1214952539524723, "learning_rate": 8.698027335314912e-06, "loss": 0.4287, "step": 15587 }, { "epoch": 0.5553556477902275, "grad_norm": 1.1317427156909172, "learning_rate": 8.696883229472748e-06, "loss": 0.4664, "step": 15588 }, { "epoch": 0.5553912749167216, "grad_norm": 2.1248170430109843, "learning_rate": 8.69573914098249e-06, "loss": 0.5546, "step": 15589 }, { "epoch": 0.5554269020432157, "grad_norm": 1.8277664777610196, "learning_rate": 8.694595069859386e-06, "loss": 0.5984, "step": 15590 }, { "epoch": 0.5554625291697098, "grad_norm": 1.1719372721392582, "learning_rate": 8.693451016118654e-06, "loss": 0.4342, "step": 15591 }, { "epoch": 0.5554981562962039, "grad_norm": 1.3935793567937522, "learning_rate": 8.692306979775537e-06, "loss": 0.5322, "step": 15592 }, { "epoch": 0.555533783422698, "grad_norm": 1.3048757837648932, "learning_rate": 8.691162960845265e-06, "loss": 0.4471, "step": 15593 }, { "epoch": 0.5555694105491922, "grad_norm": 1.1652450845899167, "learning_rate": 8.690018959343071e-06, "loss": 0.4377, "step": 15594 }, { "epoch": 0.5556050376756863, "grad_norm": 1.1748285631957958, "learning_rate": 8.688874975284198e-06, "loss": 0.4767, "step": 15595 }, { "epoch": 0.5556406648021803, "grad_norm": 1.482889492266916, "learning_rate": 8.687731008683864e-06, "loss": 0.5044, "step": 15596 }, { "epoch": 0.5556762919286745, "grad_norm": 1.1480040889711325, "learning_rate": 8.686587059557313e-06, "loss": 0.4853, "step": 15597 }, { "epoch": 0.5557119190551686, "grad_norm": 1.1044741880434983, "learning_rate": 8.685443127919769e-06, "loss": 0.3943, "step": 15598 }, { "epoch": 0.5557475461816627, "grad_norm": 1.1408254275149339, "learning_rate": 8.684299213786472e-06, "loss": 0.4254, "step": 15599 }, { "epoch": 0.5557831733081569, "grad_norm": 1.1306060080439513, "learning_rate": 8.68315531717265e-06, "loss": 0.4797, "step": 15600 }, { "epoch": 0.5558188004346509, "grad_norm": 1.786546805732943, "learning_rate": 8.682011438093537e-06, "loss": 0.491, "step": 15601 }, { "epoch": 0.555854427561145, "grad_norm": 1.4444799291414128, "learning_rate": 8.68086757656436e-06, "loss": 0.507, "step": 15602 }, { "epoch": 0.5558900546876392, "grad_norm": 1.4914751774197317, "learning_rate": 8.679723732600355e-06, "loss": 0.5452, "step": 15603 }, { "epoch": 0.5559256818141333, "grad_norm": 1.050280699802506, "learning_rate": 8.678579906216748e-06, "loss": 0.3982, "step": 15604 }, { "epoch": 0.5559613089406273, "grad_norm": 1.0283954497878622, "learning_rate": 8.677436097428775e-06, "loss": 0.4254, "step": 15605 }, { "epoch": 0.5559969360671215, "grad_norm": 1.473399610375859, "learning_rate": 8.676292306251667e-06, "loss": 0.6078, "step": 15606 }, { "epoch": 0.5560325631936156, "grad_norm": 1.3104054960066966, "learning_rate": 8.675148532700651e-06, "loss": 0.472, "step": 15607 }, { "epoch": 0.5560681903201097, "grad_norm": 1.736891260249169, "learning_rate": 8.67400477679096e-06, "loss": 0.5468, "step": 15608 }, { "epoch": 0.5561038174466039, "grad_norm": 1.620268210405076, "learning_rate": 8.672861038537818e-06, "loss": 0.6284, "step": 15609 }, { "epoch": 0.5561394445730979, "grad_norm": 1.7529359069633554, "learning_rate": 8.671717317956462e-06, "loss": 0.4779, "step": 15610 }, { "epoch": 0.556175071699592, "grad_norm": 1.0180699958022552, "learning_rate": 8.670573615062115e-06, "loss": 0.417, "step": 15611 }, { "epoch": 0.5562106988260862, "grad_norm": 1.104864766579979, "learning_rate": 8.669429929870016e-06, "loss": 0.4674, "step": 15612 }, { "epoch": 0.5562463259525803, "grad_norm": 1.517931358473089, "learning_rate": 8.66828626239538e-06, "loss": 0.627, "step": 15613 }, { "epoch": 0.5562819530790744, "grad_norm": 1.0795067361871362, "learning_rate": 8.667142612653449e-06, "loss": 0.4331, "step": 15614 }, { "epoch": 0.5563175802055685, "grad_norm": 1.2075535456452329, "learning_rate": 8.665998980659443e-06, "loss": 0.4208, "step": 15615 }, { "epoch": 0.5563532073320626, "grad_norm": 1.662965035746125, "learning_rate": 8.66485536642859e-06, "loss": 0.5527, "step": 15616 }, { "epoch": 0.5563888344585567, "grad_norm": 1.4994906050256205, "learning_rate": 8.663711769976129e-06, "loss": 0.5749, "step": 15617 }, { "epoch": 0.5564244615850509, "grad_norm": 1.3976375035050246, "learning_rate": 8.662568191317273e-06, "loss": 0.4599, "step": 15618 }, { "epoch": 0.556460088711545, "grad_norm": 1.4965169656336568, "learning_rate": 8.661424630467258e-06, "loss": 0.5974, "step": 15619 }, { "epoch": 0.556495715838039, "grad_norm": 1.0890072690239627, "learning_rate": 8.66028108744131e-06, "loss": 0.4823, "step": 15620 }, { "epoch": 0.5565313429645332, "grad_norm": 1.443940217444721, "learning_rate": 8.659137562254659e-06, "loss": 0.5866, "step": 15621 }, { "epoch": 0.5565669700910273, "grad_norm": 1.119011637269381, "learning_rate": 8.657994054922522e-06, "loss": 0.4625, "step": 15622 }, { "epoch": 0.5566025972175214, "grad_norm": 1.5947320393540878, "learning_rate": 8.656850565460139e-06, "loss": 0.5563, "step": 15623 }, { "epoch": 0.5566382243440156, "grad_norm": 1.3693010245501964, "learning_rate": 8.655707093882723e-06, "loss": 0.5299, "step": 15624 }, { "epoch": 0.5566738514705096, "grad_norm": 1.3276903338324006, "learning_rate": 8.654563640205512e-06, "loss": 0.5558, "step": 15625 }, { "epoch": 0.5567094785970037, "grad_norm": 1.171081682544864, "learning_rate": 8.653420204443723e-06, "loss": 0.4219, "step": 15626 }, { "epoch": 0.5567451057234979, "grad_norm": 1.2564523070099114, "learning_rate": 8.652276786612584e-06, "loss": 0.5175, "step": 15627 }, { "epoch": 0.556780732849992, "grad_norm": 0.9754369064505293, "learning_rate": 8.651133386727328e-06, "loss": 0.4339, "step": 15628 }, { "epoch": 0.556816359976486, "grad_norm": 1.0493985919808495, "learning_rate": 8.649990004803168e-06, "loss": 0.4621, "step": 15629 }, { "epoch": 0.5568519871029802, "grad_norm": 1.6433348047016259, "learning_rate": 8.648846640855338e-06, "loss": 0.5711, "step": 15630 }, { "epoch": 0.5568876142294743, "grad_norm": 1.6586921328945388, "learning_rate": 8.647703294899057e-06, "loss": 0.5504, "step": 15631 }, { "epoch": 0.5569232413559684, "grad_norm": 1.8439675500416615, "learning_rate": 8.646559966949555e-06, "loss": 0.6006, "step": 15632 }, { "epoch": 0.5569588684824626, "grad_norm": 1.5333502842884732, "learning_rate": 8.64541665702205e-06, "loss": 0.5141, "step": 15633 }, { "epoch": 0.5569944956089566, "grad_norm": 1.196868007103904, "learning_rate": 8.64427336513177e-06, "loss": 0.4673, "step": 15634 }, { "epoch": 0.5570301227354507, "grad_norm": 1.7373276752445677, "learning_rate": 8.643130091293937e-06, "loss": 0.5196, "step": 15635 }, { "epoch": 0.5570657498619449, "grad_norm": 1.5173422135034036, "learning_rate": 8.641986835523775e-06, "loss": 0.6163, "step": 15636 }, { "epoch": 0.557101376988439, "grad_norm": 1.0634386694488918, "learning_rate": 8.640843597836508e-06, "loss": 0.4509, "step": 15637 }, { "epoch": 0.5571370041149332, "grad_norm": 1.5128182535832118, "learning_rate": 8.639700378247362e-06, "loss": 0.5102, "step": 15638 }, { "epoch": 0.5571726312414272, "grad_norm": 1.1690034132934073, "learning_rate": 8.63855717677155e-06, "loss": 0.4489, "step": 15639 }, { "epoch": 0.5572082583679213, "grad_norm": 1.7266879893707792, "learning_rate": 8.637413993424303e-06, "loss": 0.5375, "step": 15640 }, { "epoch": 0.5572438854944155, "grad_norm": 1.0249559756098154, "learning_rate": 8.636270828220844e-06, "loss": 0.4398, "step": 15641 }, { "epoch": 0.5572795126209096, "grad_norm": 1.2239898902283757, "learning_rate": 8.635127681176387e-06, "loss": 0.5763, "step": 15642 }, { "epoch": 0.5573151397474037, "grad_norm": 1.174262054954799, "learning_rate": 8.633984552306165e-06, "loss": 0.4066, "step": 15643 }, { "epoch": 0.5573507668738978, "grad_norm": 1.4434735309659816, "learning_rate": 8.63284144162539e-06, "loss": 0.5717, "step": 15644 }, { "epoch": 0.5573863940003919, "grad_norm": 1.5156167137466956, "learning_rate": 8.631698349149287e-06, "loss": 0.4496, "step": 15645 }, { "epoch": 0.557422021126886, "grad_norm": 1.0391541352267404, "learning_rate": 8.630555274893076e-06, "loss": 0.4808, "step": 15646 }, { "epoch": 0.5574576482533802, "grad_norm": 1.1577830809901266, "learning_rate": 8.629412218871982e-06, "loss": 0.3832, "step": 15647 }, { "epoch": 0.5574932753798743, "grad_norm": 1.3038280320590998, "learning_rate": 8.628269181101216e-06, "loss": 0.5125, "step": 15648 }, { "epoch": 0.5575289025063683, "grad_norm": 1.6978532870253786, "learning_rate": 8.627126161596012e-06, "loss": 0.5408, "step": 15649 }, { "epoch": 0.5575645296328625, "grad_norm": 1.7476581727645868, "learning_rate": 8.625983160371579e-06, "loss": 0.5703, "step": 15650 }, { "epoch": 0.5576001567593566, "grad_norm": 1.0452892552042576, "learning_rate": 8.624840177443138e-06, "loss": 0.4331, "step": 15651 }, { "epoch": 0.5576357838858507, "grad_norm": 1.0268574503944263, "learning_rate": 8.623697212825917e-06, "loss": 0.4528, "step": 15652 }, { "epoch": 0.5576714110123449, "grad_norm": 1.1375573829798977, "learning_rate": 8.622554266535124e-06, "loss": 0.475, "step": 15653 }, { "epoch": 0.5577070381388389, "grad_norm": 1.3603359780960205, "learning_rate": 8.62141133858599e-06, "loss": 0.5576, "step": 15654 }, { "epoch": 0.557742665265333, "grad_norm": 1.0894518390228365, "learning_rate": 8.620268428993723e-06, "loss": 0.4844, "step": 15655 }, { "epoch": 0.5577782923918272, "grad_norm": 1.8693197561649715, "learning_rate": 8.619125537773548e-06, "loss": 0.6125, "step": 15656 }, { "epoch": 0.5578139195183213, "grad_norm": 1.3222614141388915, "learning_rate": 8.61798266494068e-06, "loss": 0.5415, "step": 15657 }, { "epoch": 0.5578495466448153, "grad_norm": 1.1300723789110874, "learning_rate": 8.616839810510342e-06, "loss": 0.4497, "step": 15658 }, { "epoch": 0.5578851737713095, "grad_norm": 1.0237812970329592, "learning_rate": 8.615696974497745e-06, "loss": 0.4167, "step": 15659 }, { "epoch": 0.5579208008978036, "grad_norm": 1.5630885202036386, "learning_rate": 8.614554156918113e-06, "loss": 0.5924, "step": 15660 }, { "epoch": 0.5579564280242977, "grad_norm": 1.0013503174679421, "learning_rate": 8.613411357786659e-06, "loss": 0.4387, "step": 15661 }, { "epoch": 0.5579920551507919, "grad_norm": 1.3355149127886916, "learning_rate": 8.6122685771186e-06, "loss": 0.537, "step": 15662 }, { "epoch": 0.5580276822772859, "grad_norm": 1.0447038114749332, "learning_rate": 8.611125814929158e-06, "loss": 0.4185, "step": 15663 }, { "epoch": 0.55806330940378, "grad_norm": 1.1243125264893912, "learning_rate": 8.609983071233545e-06, "loss": 0.4468, "step": 15664 }, { "epoch": 0.5580989365302742, "grad_norm": 1.2218639676387129, "learning_rate": 8.608840346046983e-06, "loss": 0.4338, "step": 15665 }, { "epoch": 0.5581345636567683, "grad_norm": 1.0966922118377942, "learning_rate": 8.60769763938468e-06, "loss": 0.4187, "step": 15666 }, { "epoch": 0.5581701907832624, "grad_norm": 1.033166368822813, "learning_rate": 8.606554951261858e-06, "loss": 0.4783, "step": 15667 }, { "epoch": 0.5582058179097565, "grad_norm": 1.1609579938400683, "learning_rate": 8.605412281693727e-06, "loss": 0.439, "step": 15668 }, { "epoch": 0.5582414450362506, "grad_norm": 1.1539433107641943, "learning_rate": 8.604269630695513e-06, "loss": 0.4538, "step": 15669 }, { "epoch": 0.5582770721627447, "grad_norm": 2.1653948805361893, "learning_rate": 8.60312699828242e-06, "loss": 0.5318, "step": 15670 }, { "epoch": 0.5583126992892389, "grad_norm": 1.3133896241539877, "learning_rate": 8.601984384469669e-06, "loss": 0.5032, "step": 15671 }, { "epoch": 0.558348326415733, "grad_norm": 1.1954214367062381, "learning_rate": 8.600841789272469e-06, "loss": 0.5903, "step": 15672 }, { "epoch": 0.558383953542227, "grad_norm": 1.0136830485873356, "learning_rate": 8.599699212706045e-06, "loss": 0.4519, "step": 15673 }, { "epoch": 0.5584195806687212, "grad_norm": 2.5623066880508305, "learning_rate": 8.598556654785598e-06, "loss": 0.5245, "step": 15674 }, { "epoch": 0.5584552077952153, "grad_norm": 1.0605304436444418, "learning_rate": 8.59741411552635e-06, "loss": 0.4548, "step": 15675 }, { "epoch": 0.5584908349217094, "grad_norm": 1.4528272801886533, "learning_rate": 8.59627159494352e-06, "loss": 0.5373, "step": 15676 }, { "epoch": 0.5585264620482036, "grad_norm": 1.1676548593611922, "learning_rate": 8.595129093052308e-06, "loss": 0.4335, "step": 15677 }, { "epoch": 0.5585620891746976, "grad_norm": 1.1403244473844618, "learning_rate": 8.593986609867936e-06, "loss": 0.4091, "step": 15678 }, { "epoch": 0.5585977163011917, "grad_norm": 1.089413424700605, "learning_rate": 8.592844145405615e-06, "loss": 0.4398, "step": 15679 }, { "epoch": 0.5586333434276859, "grad_norm": 1.781866836429315, "learning_rate": 8.591701699680562e-06, "loss": 0.4416, "step": 15680 }, { "epoch": 0.55866897055418, "grad_norm": 1.4936381436575172, "learning_rate": 8.590559272707978e-06, "loss": 0.5442, "step": 15681 }, { "epoch": 0.558704597680674, "grad_norm": 1.1206355922836984, "learning_rate": 8.589416864503088e-06, "loss": 0.4671, "step": 15682 }, { "epoch": 0.5587402248071682, "grad_norm": 1.0363056192088151, "learning_rate": 8.588274475081094e-06, "loss": 0.4701, "step": 15683 }, { "epoch": 0.5587758519336623, "grad_norm": 1.3086970896094317, "learning_rate": 8.587132104457216e-06, "loss": 0.4866, "step": 15684 }, { "epoch": 0.5588114790601564, "grad_norm": 1.798364452846793, "learning_rate": 8.58598975264666e-06, "loss": 0.5565, "step": 15685 }, { "epoch": 0.5588471061866506, "grad_norm": 1.3474755729338515, "learning_rate": 8.584847419664635e-06, "loss": 0.5737, "step": 15686 }, { "epoch": 0.5588827333131446, "grad_norm": 1.2089632943379929, "learning_rate": 8.583705105526362e-06, "loss": 0.4537, "step": 15687 }, { "epoch": 0.5589183604396387, "grad_norm": 1.3392623018350545, "learning_rate": 8.58256281024704e-06, "loss": 0.6082, "step": 15688 }, { "epoch": 0.5589539875661329, "grad_norm": 1.5160512742312748, "learning_rate": 8.58142053384189e-06, "loss": 0.6136, "step": 15689 }, { "epoch": 0.558989614692627, "grad_norm": 1.104981549334218, "learning_rate": 8.580278276326113e-06, "loss": 0.4805, "step": 15690 }, { "epoch": 0.559025241819121, "grad_norm": 1.1758331760670018, "learning_rate": 8.579136037714929e-06, "loss": 0.4702, "step": 15691 }, { "epoch": 0.5590608689456152, "grad_norm": 1.520345394828534, "learning_rate": 8.577993818023535e-06, "loss": 0.5096, "step": 15692 }, { "epoch": 0.5590964960721093, "grad_norm": 1.5372149589362154, "learning_rate": 8.576851617267151e-06, "loss": 0.5748, "step": 15693 }, { "epoch": 0.5591321231986034, "grad_norm": 1.1210540638203281, "learning_rate": 8.57570943546098e-06, "loss": 0.4601, "step": 15694 }, { "epoch": 0.5591677503250976, "grad_norm": 1.0501918416935452, "learning_rate": 8.574567272620238e-06, "loss": 0.383, "step": 15695 }, { "epoch": 0.5592033774515917, "grad_norm": 1.1522596620807348, "learning_rate": 8.573425128760124e-06, "loss": 0.4309, "step": 15696 }, { "epoch": 0.5592390045780857, "grad_norm": 1.406073899439702, "learning_rate": 8.57228300389585e-06, "loss": 0.4661, "step": 15697 }, { "epoch": 0.5592746317045799, "grad_norm": 1.389060387905872, "learning_rate": 8.571140898042631e-06, "loss": 0.4564, "step": 15698 }, { "epoch": 0.559310258831074, "grad_norm": 1.6802030452741408, "learning_rate": 8.569998811215667e-06, "loss": 0.5692, "step": 15699 }, { "epoch": 0.5593458859575681, "grad_norm": 1.2830315927818814, "learning_rate": 8.568856743430169e-06, "loss": 0.4874, "step": 15700 }, { "epoch": 0.5593815130840623, "grad_norm": 0.9751130419768615, "learning_rate": 8.567714694701344e-06, "loss": 0.401, "step": 15701 }, { "epoch": 0.5594171402105563, "grad_norm": 1.3992923549800642, "learning_rate": 8.566572665044402e-06, "loss": 0.5918, "step": 15702 }, { "epoch": 0.5594527673370504, "grad_norm": 1.0072022850279019, "learning_rate": 8.565430654474542e-06, "loss": 0.4135, "step": 15703 }, { "epoch": 0.5594883944635446, "grad_norm": 1.3290051799751708, "learning_rate": 8.564288663006977e-06, "loss": 0.4842, "step": 15704 }, { "epoch": 0.5595240215900387, "grad_norm": 1.2595671281458725, "learning_rate": 8.563146690656911e-06, "loss": 0.5826, "step": 15705 }, { "epoch": 0.5595596487165327, "grad_norm": 1.4873205297903769, "learning_rate": 8.562004737439555e-06, "loss": 0.488, "step": 15706 }, { "epoch": 0.5595952758430269, "grad_norm": 1.173927527170835, "learning_rate": 8.560862803370107e-06, "loss": 0.4824, "step": 15707 }, { "epoch": 0.559630902969521, "grad_norm": 1.3972168808290693, "learning_rate": 8.559720888463777e-06, "loss": 0.5234, "step": 15708 }, { "epoch": 0.5596665300960151, "grad_norm": 1.399555422414512, "learning_rate": 8.558578992735771e-06, "loss": 0.501, "step": 15709 }, { "epoch": 0.5597021572225093, "grad_norm": 1.2080768248936742, "learning_rate": 8.557437116201292e-06, "loss": 0.5015, "step": 15710 }, { "epoch": 0.5597377843490033, "grad_norm": 1.022610457720316, "learning_rate": 8.556295258875548e-06, "loss": 0.4356, "step": 15711 }, { "epoch": 0.5597734114754974, "grad_norm": 1.0378215854301456, "learning_rate": 8.555153420773739e-06, "loss": 0.4254, "step": 15712 }, { "epoch": 0.5598090386019916, "grad_norm": 1.8478955125342351, "learning_rate": 8.554011601911077e-06, "loss": 0.5676, "step": 15713 }, { "epoch": 0.5598446657284857, "grad_norm": 1.033988076644154, "learning_rate": 8.552869802302756e-06, "loss": 0.4773, "step": 15714 }, { "epoch": 0.5598802928549798, "grad_norm": 1.101752181700125, "learning_rate": 8.55172802196399e-06, "loss": 0.4833, "step": 15715 }, { "epoch": 0.5599159199814739, "grad_norm": 1.6881091547177511, "learning_rate": 8.550586260909974e-06, "loss": 0.5339, "step": 15716 }, { "epoch": 0.559951547107968, "grad_norm": 1.1660463926428968, "learning_rate": 8.549444519155918e-06, "loss": 0.4694, "step": 15717 }, { "epoch": 0.5599871742344621, "grad_norm": 1.0085884058314725, "learning_rate": 8.548302796717018e-06, "loss": 0.3996, "step": 15718 }, { "epoch": 0.5600228013609563, "grad_norm": 1.5680833554903062, "learning_rate": 8.547161093608484e-06, "loss": 0.4654, "step": 15719 }, { "epoch": 0.5600584284874504, "grad_norm": 0.9706939041968644, "learning_rate": 8.546019409845514e-06, "loss": 0.4193, "step": 15720 }, { "epoch": 0.5600940556139444, "grad_norm": 0.9904168842741765, "learning_rate": 8.544877745443311e-06, "loss": 0.4745, "step": 15721 }, { "epoch": 0.5601296827404386, "grad_norm": 1.039486522006193, "learning_rate": 8.543736100417083e-06, "loss": 0.438, "step": 15722 }, { "epoch": 0.5601653098669327, "grad_norm": 2.808542198905083, "learning_rate": 8.542594474782018e-06, "loss": 0.5998, "step": 15723 }, { "epoch": 0.5602009369934268, "grad_norm": 1.331526251834364, "learning_rate": 8.541452868553335e-06, "loss": 0.5292, "step": 15724 }, { "epoch": 0.560236564119921, "grad_norm": 1.5488044399191305, "learning_rate": 8.540311281746222e-06, "loss": 0.5627, "step": 15725 }, { "epoch": 0.560272191246415, "grad_norm": 1.6585093082288467, "learning_rate": 8.539169714375885e-06, "loss": 0.5063, "step": 15726 }, { "epoch": 0.5603078183729091, "grad_norm": 1.3003335849595037, "learning_rate": 8.538028166457524e-06, "loss": 0.5242, "step": 15727 }, { "epoch": 0.5603434454994033, "grad_norm": 1.6288052257362995, "learning_rate": 8.536886638006344e-06, "loss": 0.5528, "step": 15728 }, { "epoch": 0.5603790726258974, "grad_norm": 1.1816052797157572, "learning_rate": 8.535745129037535e-06, "loss": 0.4357, "step": 15729 }, { "epoch": 0.5604146997523914, "grad_norm": 1.6365779136725864, "learning_rate": 8.534603639566307e-06, "loss": 0.5309, "step": 15730 }, { "epoch": 0.5604503268788856, "grad_norm": 1.6497200155707337, "learning_rate": 8.533462169607853e-06, "loss": 0.4871, "step": 15731 }, { "epoch": 0.5604859540053797, "grad_norm": 1.0396620093832885, "learning_rate": 8.532320719177375e-06, "loss": 0.4715, "step": 15732 }, { "epoch": 0.5605215811318738, "grad_norm": 1.1733253879177394, "learning_rate": 8.531179288290077e-06, "loss": 0.4899, "step": 15733 }, { "epoch": 0.560557208258368, "grad_norm": 1.2361474334484517, "learning_rate": 8.53003787696115e-06, "loss": 0.4892, "step": 15734 }, { "epoch": 0.560592835384862, "grad_norm": 1.0819520419955415, "learning_rate": 8.528896485205796e-06, "loss": 0.4508, "step": 15735 }, { "epoch": 0.5606284625113561, "grad_norm": 1.5293133682904112, "learning_rate": 8.527755113039213e-06, "loss": 0.6242, "step": 15736 }, { "epoch": 0.5606640896378503, "grad_norm": 1.5534191569611824, "learning_rate": 8.526613760476601e-06, "loss": 0.6034, "step": 15737 }, { "epoch": 0.5606997167643444, "grad_norm": 1.288408783993235, "learning_rate": 8.525472427533156e-06, "loss": 0.5379, "step": 15738 }, { "epoch": 0.5607353438908385, "grad_norm": 1.1961819684749788, "learning_rate": 8.524331114224079e-06, "loss": 0.4516, "step": 15739 }, { "epoch": 0.5607709710173326, "grad_norm": 1.0273485726310716, "learning_rate": 8.523189820564562e-06, "loss": 0.4521, "step": 15740 }, { "epoch": 0.5608065981438267, "grad_norm": 2.957859123468794, "learning_rate": 8.522048546569805e-06, "loss": 0.575, "step": 15741 }, { "epoch": 0.5608422252703208, "grad_norm": 1.4714091809866485, "learning_rate": 8.520907292255004e-06, "loss": 0.5599, "step": 15742 }, { "epoch": 0.560877852396815, "grad_norm": 1.3052887905847466, "learning_rate": 8.519766057635355e-06, "loss": 0.4598, "step": 15743 }, { "epoch": 0.560913479523309, "grad_norm": 1.043054712299427, "learning_rate": 8.518624842726062e-06, "loss": 0.4307, "step": 15744 }, { "epoch": 0.5609491066498031, "grad_norm": 1.3175597634378666, "learning_rate": 8.517483647542307e-06, "loss": 0.5672, "step": 15745 }, { "epoch": 0.5609847337762973, "grad_norm": 1.700085961935436, "learning_rate": 8.516342472099299e-06, "loss": 0.5553, "step": 15746 }, { "epoch": 0.5610203609027914, "grad_norm": 1.1250981180868949, "learning_rate": 8.515201316412222e-06, "loss": 0.412, "step": 15747 }, { "epoch": 0.5610559880292855, "grad_norm": 1.1099718543901076, "learning_rate": 8.514060180496285e-06, "loss": 0.3954, "step": 15748 }, { "epoch": 0.5610916151557797, "grad_norm": 1.3419583209672725, "learning_rate": 8.512919064366669e-06, "loss": 0.4931, "step": 15749 }, { "epoch": 0.5611272422822737, "grad_norm": 1.101293548550431, "learning_rate": 8.511777968038579e-06, "loss": 0.4524, "step": 15750 }, { "epoch": 0.5611628694087678, "grad_norm": 1.04328074798652, "learning_rate": 8.510636891527203e-06, "loss": 0.4115, "step": 15751 }, { "epoch": 0.561198496535262, "grad_norm": 1.5138999498480203, "learning_rate": 8.509495834847738e-06, "loss": 0.4688, "step": 15752 }, { "epoch": 0.5612341236617561, "grad_norm": 1.2516541535274743, "learning_rate": 8.508354798015378e-06, "loss": 0.5019, "step": 15753 }, { "epoch": 0.5612697507882501, "grad_norm": 1.077682755696289, "learning_rate": 8.507213781045316e-06, "loss": 0.4224, "step": 15754 }, { "epoch": 0.5613053779147443, "grad_norm": 1.7930035676994094, "learning_rate": 8.50607278395275e-06, "loss": 0.5397, "step": 15755 }, { "epoch": 0.5613410050412384, "grad_norm": 1.4298047697316414, "learning_rate": 8.504931806752865e-06, "loss": 0.5336, "step": 15756 }, { "epoch": 0.5613766321677325, "grad_norm": 1.1649234494462837, "learning_rate": 8.50379084946086e-06, "loss": 0.4392, "step": 15757 }, { "epoch": 0.5614122592942267, "grad_norm": 1.0484962559424504, "learning_rate": 8.502649912091921e-06, "loss": 0.4511, "step": 15758 }, { "epoch": 0.5614478864207207, "grad_norm": 1.4690135545042464, "learning_rate": 8.501508994661252e-06, "loss": 0.4808, "step": 15759 }, { "epoch": 0.5614835135472148, "grad_norm": 1.1020554967582843, "learning_rate": 8.500368097184034e-06, "loss": 0.411, "step": 15760 }, { "epoch": 0.561519140673709, "grad_norm": 1.4056992888768933, "learning_rate": 8.499227219675463e-06, "loss": 0.5312, "step": 15761 }, { "epoch": 0.5615547678002031, "grad_norm": 0.9693842151439677, "learning_rate": 8.49808636215073e-06, "loss": 0.4056, "step": 15762 }, { "epoch": 0.5615903949266972, "grad_norm": 1.4092352405542192, "learning_rate": 8.49694552462503e-06, "loss": 0.5483, "step": 15763 }, { "epoch": 0.5616260220531913, "grad_norm": 1.0124813644265955, "learning_rate": 8.495804707113547e-06, "loss": 0.386, "step": 15764 }, { "epoch": 0.5616616491796854, "grad_norm": 1.5087952625404248, "learning_rate": 8.49466390963148e-06, "loss": 0.5975, "step": 15765 }, { "epoch": 0.5616972763061795, "grad_norm": 1.3331699902895384, "learning_rate": 8.493523132194012e-06, "loss": 0.4606, "step": 15766 }, { "epoch": 0.5617329034326737, "grad_norm": 1.0220813303837233, "learning_rate": 8.492382374816334e-06, "loss": 0.4005, "step": 15767 }, { "epoch": 0.5617685305591678, "grad_norm": 1.1468333106285178, "learning_rate": 8.491241637513644e-06, "loss": 0.4494, "step": 15768 }, { "epoch": 0.5618041576856618, "grad_norm": 1.4049129441137045, "learning_rate": 8.490100920301121e-06, "loss": 0.4985, "step": 15769 }, { "epoch": 0.561839784812156, "grad_norm": 1.06216920453636, "learning_rate": 8.488960223193968e-06, "loss": 0.4348, "step": 15770 }, { "epoch": 0.5618754119386501, "grad_norm": 1.4308691728730438, "learning_rate": 8.487819546207357e-06, "loss": 0.5364, "step": 15771 }, { "epoch": 0.5619110390651442, "grad_norm": 1.5072661906756508, "learning_rate": 8.48667888935649e-06, "loss": 0.5149, "step": 15772 }, { "epoch": 0.5619466661916384, "grad_norm": 1.0868912220871525, "learning_rate": 8.48553825265655e-06, "loss": 0.4855, "step": 15773 }, { "epoch": 0.5619822933181324, "grad_norm": 1.5951258787970028, "learning_rate": 8.484397636122727e-06, "loss": 0.4744, "step": 15774 }, { "epoch": 0.5620179204446265, "grad_norm": 1.6401744763572388, "learning_rate": 8.483257039770207e-06, "loss": 0.5103, "step": 15775 }, { "epoch": 0.5620535475711207, "grad_norm": 1.4982386465090243, "learning_rate": 8.482116463614185e-06, "loss": 0.5674, "step": 15776 }, { "epoch": 0.5620891746976148, "grad_norm": 1.1128273250839709, "learning_rate": 8.480975907669838e-06, "loss": 0.4682, "step": 15777 }, { "epoch": 0.5621248018241088, "grad_norm": 1.1705792150908807, "learning_rate": 8.479835371952359e-06, "loss": 0.4688, "step": 15778 }, { "epoch": 0.562160428950603, "grad_norm": 1.4683217222297942, "learning_rate": 8.478694856476937e-06, "loss": 0.5819, "step": 15779 }, { "epoch": 0.5621960560770971, "grad_norm": 1.5462878194664846, "learning_rate": 8.477554361258753e-06, "loss": 0.5634, "step": 15780 }, { "epoch": 0.5622316832035912, "grad_norm": 1.7214179733186854, "learning_rate": 8.476413886313002e-06, "loss": 0.53, "step": 15781 }, { "epoch": 0.5622673103300854, "grad_norm": 1.5654887159916755, "learning_rate": 8.47527343165486e-06, "loss": 0.4327, "step": 15782 }, { "epoch": 0.5623029374565794, "grad_norm": 1.040866247836255, "learning_rate": 8.474132997299521e-06, "loss": 0.4074, "step": 15783 }, { "epoch": 0.5623385645830735, "grad_norm": 1.3108326992451862, "learning_rate": 8.472992583262166e-06, "loss": 0.5553, "step": 15784 }, { "epoch": 0.5623741917095677, "grad_norm": 1.0432493581009201, "learning_rate": 8.471852189557986e-06, "loss": 0.4533, "step": 15785 }, { "epoch": 0.5624098188360618, "grad_norm": 1.4064978507564707, "learning_rate": 8.470711816202154e-06, "loss": 0.5344, "step": 15786 }, { "epoch": 0.5624454459625559, "grad_norm": 1.8083963117928277, "learning_rate": 8.46957146320987e-06, "loss": 0.5512, "step": 15787 }, { "epoch": 0.56248107308905, "grad_norm": 1.6066215166347755, "learning_rate": 8.468431130596309e-06, "loss": 0.6469, "step": 15788 }, { "epoch": 0.5625167002155441, "grad_norm": 1.2295994361891667, "learning_rate": 8.467290818376658e-06, "loss": 0.4579, "step": 15789 }, { "epoch": 0.5625523273420382, "grad_norm": 1.0229194967046555, "learning_rate": 8.4661505265661e-06, "loss": 0.446, "step": 15790 }, { "epoch": 0.5625879544685324, "grad_norm": 1.2039763556740646, "learning_rate": 8.465010255179822e-06, "loss": 0.4202, "step": 15791 }, { "epoch": 0.5626235815950265, "grad_norm": 1.6389272765716132, "learning_rate": 8.463870004233008e-06, "loss": 0.5439, "step": 15792 }, { "epoch": 0.5626592087215205, "grad_norm": 1.439807367931052, "learning_rate": 8.462729773740833e-06, "loss": 0.5759, "step": 15793 }, { "epoch": 0.5626948358480147, "grad_norm": 1.2204528893263324, "learning_rate": 8.461589563718489e-06, "loss": 0.4002, "step": 15794 }, { "epoch": 0.5627304629745088, "grad_norm": 1.4982435422183533, "learning_rate": 8.460449374181152e-06, "loss": 0.4834, "step": 15795 }, { "epoch": 0.5627660901010029, "grad_norm": 1.853436671642391, "learning_rate": 8.459309205144014e-06, "loss": 0.5003, "step": 15796 }, { "epoch": 0.5628017172274971, "grad_norm": 1.2803930596335567, "learning_rate": 8.458169056622245e-06, "loss": 0.4611, "step": 15797 }, { "epoch": 0.5628373443539911, "grad_norm": 1.0260367477224914, "learning_rate": 8.457028928631035e-06, "loss": 0.4453, "step": 15798 }, { "epoch": 0.5628729714804852, "grad_norm": 1.312444671570739, "learning_rate": 8.455888821185561e-06, "loss": 0.4529, "step": 15799 }, { "epoch": 0.5629085986069794, "grad_norm": 1.0943249931067651, "learning_rate": 8.454748734301009e-06, "loss": 0.408, "step": 15800 }, { "epoch": 0.5629442257334735, "grad_norm": 1.3076188332297323, "learning_rate": 8.453608667992556e-06, "loss": 0.4661, "step": 15801 }, { "epoch": 0.5629798528599675, "grad_norm": 1.7788355327423877, "learning_rate": 8.452468622275384e-06, "loss": 0.5679, "step": 15802 }, { "epoch": 0.5630154799864617, "grad_norm": 1.5803334526205925, "learning_rate": 8.451328597164679e-06, "loss": 0.5706, "step": 15803 }, { "epoch": 0.5630511071129558, "grad_norm": 1.0774775909352803, "learning_rate": 8.45018859267561e-06, "loss": 0.4351, "step": 15804 }, { "epoch": 0.5630867342394499, "grad_norm": 1.3430584310587925, "learning_rate": 8.449048608823367e-06, "loss": 0.4905, "step": 15805 }, { "epoch": 0.5631223613659441, "grad_norm": 1.7108705200371326, "learning_rate": 8.447908645623121e-06, "loss": 0.5353, "step": 15806 }, { "epoch": 0.5631579884924381, "grad_norm": 1.4927080109010094, "learning_rate": 8.446768703090065e-06, "loss": 0.532, "step": 15807 }, { "epoch": 0.5631936156189322, "grad_norm": 1.4822580854769019, "learning_rate": 8.44562878123936e-06, "loss": 0.5184, "step": 15808 }, { "epoch": 0.5632292427454264, "grad_norm": 1.2042036052407723, "learning_rate": 8.444488880086201e-06, "loss": 0.4483, "step": 15809 }, { "epoch": 0.5632648698719205, "grad_norm": 1.493740135996604, "learning_rate": 8.443348999645756e-06, "loss": 0.548, "step": 15810 }, { "epoch": 0.5633004969984146, "grad_norm": 1.345859108406023, "learning_rate": 8.44220913993321e-06, "loss": 0.5104, "step": 15811 }, { "epoch": 0.5633361241249087, "grad_norm": 1.1657456152262846, "learning_rate": 8.441069300963733e-06, "loss": 0.4638, "step": 15812 }, { "epoch": 0.5633717512514028, "grad_norm": 1.7209225729779527, "learning_rate": 8.43992948275251e-06, "loss": 0.5073, "step": 15813 }, { "epoch": 0.5634073783778969, "grad_norm": 1.59493944876109, "learning_rate": 8.438789685314721e-06, "loss": 0.6067, "step": 15814 }, { "epoch": 0.5634430055043911, "grad_norm": 1.474693579299109, "learning_rate": 8.437649908665532e-06, "loss": 0.6022, "step": 15815 }, { "epoch": 0.5634786326308852, "grad_norm": 1.3545656269694393, "learning_rate": 8.436510152820133e-06, "loss": 0.5062, "step": 15816 }, { "epoch": 0.5635142597573792, "grad_norm": 1.926914734468406, "learning_rate": 8.435370417793687e-06, "loss": 0.5624, "step": 15817 }, { "epoch": 0.5635498868838734, "grad_norm": 1.6053288863341988, "learning_rate": 8.434230703601386e-06, "loss": 0.6071, "step": 15818 }, { "epoch": 0.5635855140103675, "grad_norm": 1.0896382396469302, "learning_rate": 8.43309101025839e-06, "loss": 0.424, "step": 15819 }, { "epoch": 0.5636211411368616, "grad_norm": 1.1331417261032217, "learning_rate": 8.431951337779886e-06, "loss": 0.461, "step": 15820 }, { "epoch": 0.5636567682633558, "grad_norm": 1.1262836331449875, "learning_rate": 8.430811686181044e-06, "loss": 0.4877, "step": 15821 }, { "epoch": 0.5636923953898498, "grad_norm": 1.0775309249536267, "learning_rate": 8.429672055477046e-06, "loss": 0.4688, "step": 15822 }, { "epoch": 0.5637280225163439, "grad_norm": 1.0178592866615463, "learning_rate": 8.428532445683058e-06, "loss": 0.3797, "step": 15823 }, { "epoch": 0.5637636496428381, "grad_norm": 1.3130236361799492, "learning_rate": 8.427392856814255e-06, "loss": 0.5589, "step": 15824 }, { "epoch": 0.5637992767693322, "grad_norm": 0.9530776102049564, "learning_rate": 8.426253288885822e-06, "loss": 0.4484, "step": 15825 }, { "epoch": 0.5638349038958262, "grad_norm": 1.094776463994165, "learning_rate": 8.425113741912922e-06, "loss": 0.4743, "step": 15826 }, { "epoch": 0.5638705310223204, "grad_norm": 1.0820367483990057, "learning_rate": 8.423974215910738e-06, "loss": 0.4256, "step": 15827 }, { "epoch": 0.5639061581488145, "grad_norm": 1.0416211906175716, "learning_rate": 8.422834710894434e-06, "loss": 0.4584, "step": 15828 }, { "epoch": 0.5639417852753086, "grad_norm": 1.0540526229805303, "learning_rate": 8.421695226879194e-06, "loss": 0.4096, "step": 15829 }, { "epoch": 0.5639774124018028, "grad_norm": 1.4453555131342977, "learning_rate": 8.42055576388018e-06, "loss": 0.5061, "step": 15830 }, { "epoch": 0.5640130395282968, "grad_norm": 1.6086904488426907, "learning_rate": 8.419416321912573e-06, "loss": 0.5841, "step": 15831 }, { "epoch": 0.5640486666547909, "grad_norm": 1.0732077203469472, "learning_rate": 8.41827690099154e-06, "loss": 0.4778, "step": 15832 }, { "epoch": 0.5640842937812851, "grad_norm": 1.1348954651280052, "learning_rate": 8.41713750113226e-06, "loss": 0.4594, "step": 15833 }, { "epoch": 0.5641199209077792, "grad_norm": 1.3497438162728015, "learning_rate": 8.415998122349896e-06, "loss": 0.5625, "step": 15834 }, { "epoch": 0.5641555480342733, "grad_norm": 1.1046753127411162, "learning_rate": 8.414858764659623e-06, "loss": 0.4844, "step": 15835 }, { "epoch": 0.5641911751607674, "grad_norm": 1.3037803016538059, "learning_rate": 8.413719428076617e-06, "loss": 0.4702, "step": 15836 }, { "epoch": 0.5642268022872615, "grad_norm": 1.151546883801224, "learning_rate": 8.412580112616044e-06, "loss": 0.42, "step": 15837 }, { "epoch": 0.5642624294137556, "grad_norm": 1.552664507891207, "learning_rate": 8.411440818293077e-06, "loss": 0.5581, "step": 15838 }, { "epoch": 0.5642980565402498, "grad_norm": 1.0953474122542435, "learning_rate": 8.410301545122884e-06, "loss": 0.4559, "step": 15839 }, { "epoch": 0.5643336836667439, "grad_norm": 1.2885887173676918, "learning_rate": 8.409162293120642e-06, "loss": 0.5499, "step": 15840 }, { "epoch": 0.5643693107932379, "grad_norm": 1.549930760393693, "learning_rate": 8.40802306230151e-06, "loss": 0.5655, "step": 15841 }, { "epoch": 0.5644049379197321, "grad_norm": 1.0102440260649088, "learning_rate": 8.406883852680669e-06, "loss": 0.465, "step": 15842 }, { "epoch": 0.5644405650462262, "grad_norm": 1.3199964479296746, "learning_rate": 8.405744664273278e-06, "loss": 0.4809, "step": 15843 }, { "epoch": 0.5644761921727203, "grad_norm": 1.0944611226243957, "learning_rate": 8.404605497094517e-06, "loss": 0.4256, "step": 15844 }, { "epoch": 0.5645118192992145, "grad_norm": 2.150558332406549, "learning_rate": 8.403466351159543e-06, "loss": 0.5623, "step": 15845 }, { "epoch": 0.5645474464257085, "grad_norm": 1.0837699299835166, "learning_rate": 8.402327226483535e-06, "loss": 0.4128, "step": 15846 }, { "epoch": 0.5645830735522026, "grad_norm": 1.4380247730830256, "learning_rate": 8.401188123081653e-06, "loss": 0.475, "step": 15847 }, { "epoch": 0.5646187006786968, "grad_norm": 1.3306924499545951, "learning_rate": 8.400049040969068e-06, "loss": 0.5127, "step": 15848 }, { "epoch": 0.5646543278051909, "grad_norm": 1.104611607684704, "learning_rate": 8.398909980160954e-06, "loss": 0.4665, "step": 15849 }, { "epoch": 0.564689954931685, "grad_norm": 1.458551093844544, "learning_rate": 8.397770940672466e-06, "loss": 0.5682, "step": 15850 }, { "epoch": 0.5647255820581791, "grad_norm": 1.0703116606521164, "learning_rate": 8.396631922518784e-06, "loss": 0.4393, "step": 15851 }, { "epoch": 0.5647612091846732, "grad_norm": 1.3823211355674307, "learning_rate": 8.395492925715064e-06, "loss": 0.4924, "step": 15852 }, { "epoch": 0.5647968363111673, "grad_norm": 1.2580671848268994, "learning_rate": 8.394353950276479e-06, "loss": 0.5112, "step": 15853 }, { "epoch": 0.5648324634376615, "grad_norm": 1.239624684718141, "learning_rate": 8.393214996218193e-06, "loss": 0.4957, "step": 15854 }, { "epoch": 0.5648680905641555, "grad_norm": 0.990382484093965, "learning_rate": 8.392076063555376e-06, "loss": 0.4186, "step": 15855 }, { "epoch": 0.5649037176906496, "grad_norm": 1.0143743889962407, "learning_rate": 8.390937152303184e-06, "loss": 0.4464, "step": 15856 }, { "epoch": 0.5649393448171438, "grad_norm": 1.6741498107562336, "learning_rate": 8.389798262476793e-06, "loss": 0.608, "step": 15857 }, { "epoch": 0.5649749719436379, "grad_norm": 1.456100487685786, "learning_rate": 8.388659394091362e-06, "loss": 0.5075, "step": 15858 }, { "epoch": 0.565010599070132, "grad_norm": 1.105390106050195, "learning_rate": 8.387520547162054e-06, "loss": 0.4869, "step": 15859 }, { "epoch": 0.5650462261966261, "grad_norm": 1.01755823764434, "learning_rate": 8.386381721704043e-06, "loss": 0.386, "step": 15860 }, { "epoch": 0.5650818533231202, "grad_norm": 1.3852116190857873, "learning_rate": 8.38524291773248e-06, "loss": 0.5664, "step": 15861 }, { "epoch": 0.5651174804496143, "grad_norm": 1.501972836935737, "learning_rate": 8.384104135262545e-06, "loss": 0.5517, "step": 15862 }, { "epoch": 0.5651531075761085, "grad_norm": 1.4647649386017636, "learning_rate": 8.382965374309387e-06, "loss": 0.5635, "step": 15863 }, { "epoch": 0.5651887347026026, "grad_norm": 1.2186157727649742, "learning_rate": 8.381826634888178e-06, "loss": 0.4377, "step": 15864 }, { "epoch": 0.5652243618290966, "grad_norm": 1.2062659997764242, "learning_rate": 8.380687917014076e-06, "loss": 0.4316, "step": 15865 }, { "epoch": 0.5652599889555908, "grad_norm": 1.3265925052259904, "learning_rate": 8.37954922070225e-06, "loss": 0.5972, "step": 15866 }, { "epoch": 0.5652956160820849, "grad_norm": 1.4380052772785925, "learning_rate": 8.378410545967856e-06, "loss": 0.5842, "step": 15867 }, { "epoch": 0.565331243208579, "grad_norm": 1.0357901643322809, "learning_rate": 8.37727189282606e-06, "loss": 0.474, "step": 15868 }, { "epoch": 0.5653668703350732, "grad_norm": 1.0183400640135223, "learning_rate": 8.37613326129202e-06, "loss": 0.4538, "step": 15869 }, { "epoch": 0.5654024974615672, "grad_norm": 1.2695959692651246, "learning_rate": 8.374994651380902e-06, "loss": 0.5175, "step": 15870 }, { "epoch": 0.5654381245880613, "grad_norm": 1.1075570459309318, "learning_rate": 8.373856063107869e-06, "loss": 0.4522, "step": 15871 }, { "epoch": 0.5654737517145555, "grad_norm": 1.7630239069602505, "learning_rate": 8.372717496488076e-06, "loss": 0.5763, "step": 15872 }, { "epoch": 0.5655093788410496, "grad_norm": 1.0040058318644751, "learning_rate": 8.371578951536689e-06, "loss": 0.4371, "step": 15873 }, { "epoch": 0.5655450059675436, "grad_norm": 1.0746138999655979, "learning_rate": 8.370440428268864e-06, "loss": 0.442, "step": 15874 }, { "epoch": 0.5655806330940378, "grad_norm": 1.086630794791651, "learning_rate": 8.369301926699766e-06, "loss": 0.4722, "step": 15875 }, { "epoch": 0.5656162602205319, "grad_norm": 1.1507379021164317, "learning_rate": 8.36816344684455e-06, "loss": 0.4548, "step": 15876 }, { "epoch": 0.565651887347026, "grad_norm": 1.1378089169990167, "learning_rate": 8.367024988718383e-06, "loss": 0.4265, "step": 15877 }, { "epoch": 0.5656875144735202, "grad_norm": 1.7641857086474833, "learning_rate": 8.365886552336415e-06, "loss": 0.5485, "step": 15878 }, { "epoch": 0.5657231416000142, "grad_norm": 1.1994013082587094, "learning_rate": 8.364748137713813e-06, "loss": 0.4748, "step": 15879 }, { "epoch": 0.5657587687265083, "grad_norm": 1.6594387720407864, "learning_rate": 8.363609744865729e-06, "loss": 0.5527, "step": 15880 }, { "epoch": 0.5657943958530025, "grad_norm": 1.5400696034348402, "learning_rate": 8.362471373807326e-06, "loss": 0.5522, "step": 15881 }, { "epoch": 0.5658300229794966, "grad_norm": 1.1642323200234432, "learning_rate": 8.361333024553765e-06, "loss": 0.4162, "step": 15882 }, { "epoch": 0.5658656501059907, "grad_norm": 0.9850366220847739, "learning_rate": 8.360194697120195e-06, "loss": 0.4261, "step": 15883 }, { "epoch": 0.5659012772324848, "grad_norm": 1.0647780362927883, "learning_rate": 8.359056391521783e-06, "loss": 0.4529, "step": 15884 }, { "epoch": 0.5659369043589789, "grad_norm": 2.26123375403101, "learning_rate": 8.357918107773678e-06, "loss": 0.5225, "step": 15885 }, { "epoch": 0.565972531485473, "grad_norm": 1.2881752173412588, "learning_rate": 8.356779845891047e-06, "loss": 0.4714, "step": 15886 }, { "epoch": 0.5660081586119672, "grad_norm": 1.2312901817446098, "learning_rate": 8.355641605889034e-06, "loss": 0.5232, "step": 15887 }, { "epoch": 0.5660437857384613, "grad_norm": 1.6563589076864909, "learning_rate": 8.354503387782807e-06, "loss": 0.625, "step": 15888 }, { "epoch": 0.5660794128649553, "grad_norm": 1.136992031347444, "learning_rate": 8.353365191587514e-06, "loss": 0.4101, "step": 15889 }, { "epoch": 0.5661150399914495, "grad_norm": 1.4318376952710465, "learning_rate": 8.352227017318319e-06, "loss": 0.551, "step": 15890 }, { "epoch": 0.5661506671179436, "grad_norm": 1.7255545328497703, "learning_rate": 8.351088864990369e-06, "loss": 0.5628, "step": 15891 }, { "epoch": 0.5661862942444377, "grad_norm": 1.151704354410493, "learning_rate": 8.349950734618829e-06, "loss": 0.416, "step": 15892 }, { "epoch": 0.5662219213709319, "grad_norm": 1.2199844890713778, "learning_rate": 8.34881262621884e-06, "loss": 0.4758, "step": 15893 }, { "epoch": 0.5662575484974259, "grad_norm": 1.1396890383265454, "learning_rate": 8.347674539805568e-06, "loss": 0.4306, "step": 15894 }, { "epoch": 0.56629317562392, "grad_norm": 1.3532977743827619, "learning_rate": 8.346536475394168e-06, "loss": 0.5579, "step": 15895 }, { "epoch": 0.5663288027504142, "grad_norm": 1.1128740003812134, "learning_rate": 8.345398432999784e-06, "loss": 0.451, "step": 15896 }, { "epoch": 0.5663644298769083, "grad_norm": 1.5708806442509775, "learning_rate": 8.344260412637585e-06, "loss": 0.5333, "step": 15897 }, { "epoch": 0.5664000570034023, "grad_norm": 1.392390715930117, "learning_rate": 8.34312241432271e-06, "loss": 0.4712, "step": 15898 }, { "epoch": 0.5664356841298965, "grad_norm": 1.1506503995870017, "learning_rate": 8.34198443807032e-06, "loss": 0.4025, "step": 15899 }, { "epoch": 0.5664713112563906, "grad_norm": 1.8043890221956944, "learning_rate": 8.340846483895565e-06, "loss": 0.5371, "step": 15900 }, { "epoch": 0.5665069383828847, "grad_norm": 1.146240490798922, "learning_rate": 8.339708551813599e-06, "loss": 0.4415, "step": 15901 }, { "epoch": 0.5665425655093789, "grad_norm": 1.4523329773553493, "learning_rate": 8.338570641839573e-06, "loss": 0.5462, "step": 15902 }, { "epoch": 0.566578192635873, "grad_norm": 1.7477605865951678, "learning_rate": 8.337432753988645e-06, "loss": 0.4512, "step": 15903 }, { "epoch": 0.566613819762367, "grad_norm": 1.0665071330722624, "learning_rate": 8.336294888275956e-06, "loss": 0.3934, "step": 15904 }, { "epoch": 0.5666494468888612, "grad_norm": 1.203417963560963, "learning_rate": 8.335157044716663e-06, "loss": 0.4422, "step": 15905 }, { "epoch": 0.5666850740153553, "grad_norm": 1.168452156828747, "learning_rate": 8.334019223325924e-06, "loss": 0.4331, "step": 15906 }, { "epoch": 0.5667207011418494, "grad_norm": 1.0016860768676095, "learning_rate": 8.332881424118878e-06, "loss": 0.4349, "step": 15907 }, { "epoch": 0.5667563282683435, "grad_norm": 0.985802818081875, "learning_rate": 8.331743647110686e-06, "loss": 0.4319, "step": 15908 }, { "epoch": 0.5667919553948376, "grad_norm": 1.5463458691443819, "learning_rate": 8.330605892316489e-06, "loss": 0.5279, "step": 15909 }, { "epoch": 0.5668275825213317, "grad_norm": 1.0612467364187388, "learning_rate": 8.329468159751443e-06, "loss": 0.3994, "step": 15910 }, { "epoch": 0.5668632096478259, "grad_norm": 2.2438923957791213, "learning_rate": 8.328330449430694e-06, "loss": 0.5502, "step": 15911 }, { "epoch": 0.56689883677432, "grad_norm": 1.980539806945383, "learning_rate": 8.327192761369399e-06, "loss": 0.5724, "step": 15912 }, { "epoch": 0.566934463900814, "grad_norm": 1.2696265665081417, "learning_rate": 8.326055095582694e-06, "loss": 0.4821, "step": 15913 }, { "epoch": 0.5669700910273082, "grad_norm": 1.628243571817192, "learning_rate": 8.324917452085743e-06, "loss": 0.5821, "step": 15914 }, { "epoch": 0.5670057181538023, "grad_norm": 1.9076801497305904, "learning_rate": 8.32377983089368e-06, "loss": 0.6084, "step": 15915 }, { "epoch": 0.5670413452802964, "grad_norm": 1.081206123696059, "learning_rate": 8.322642232021664e-06, "loss": 0.3944, "step": 15916 }, { "epoch": 0.5670769724067906, "grad_norm": 1.091393974391781, "learning_rate": 8.32150465548484e-06, "loss": 0.3789, "step": 15917 }, { "epoch": 0.5671125995332846, "grad_norm": 2.0565253644843478, "learning_rate": 8.320367101298351e-06, "loss": 0.5398, "step": 15918 }, { "epoch": 0.5671482266597787, "grad_norm": 1.0936518257006576, "learning_rate": 8.319229569477355e-06, "loss": 0.4521, "step": 15919 }, { "epoch": 0.5671838537862729, "grad_norm": 1.69663493899983, "learning_rate": 8.318092060036984e-06, "loss": 0.5332, "step": 15920 }, { "epoch": 0.567219480912767, "grad_norm": 1.3603225102145409, "learning_rate": 8.316954572992397e-06, "loss": 0.5498, "step": 15921 }, { "epoch": 0.567255108039261, "grad_norm": 1.3452221488005405, "learning_rate": 8.315817108358735e-06, "loss": 0.4811, "step": 15922 }, { "epoch": 0.5672907351657552, "grad_norm": 1.6338634670687389, "learning_rate": 8.314679666151148e-06, "loss": 0.4966, "step": 15923 }, { "epoch": 0.5673263622922493, "grad_norm": 1.3048282592616998, "learning_rate": 8.313542246384775e-06, "loss": 0.4405, "step": 15924 }, { "epoch": 0.5673619894187434, "grad_norm": 1.5988429550755574, "learning_rate": 8.312404849074769e-06, "loss": 0.5348, "step": 15925 }, { "epoch": 0.5673976165452376, "grad_norm": 1.2502685388441352, "learning_rate": 8.31126747423627e-06, "loss": 0.4796, "step": 15926 }, { "epoch": 0.5674332436717316, "grad_norm": 1.9189548229073201, "learning_rate": 8.310130121884425e-06, "loss": 0.5728, "step": 15927 }, { "epoch": 0.5674688707982257, "grad_norm": 1.7261456984649464, "learning_rate": 8.308992792034379e-06, "loss": 0.529, "step": 15928 }, { "epoch": 0.5675044979247199, "grad_norm": 1.316030465011456, "learning_rate": 8.307855484701274e-06, "loss": 0.5341, "step": 15929 }, { "epoch": 0.567540125051214, "grad_norm": 1.1384728808786435, "learning_rate": 8.306718199900262e-06, "loss": 0.4307, "step": 15930 }, { "epoch": 0.5675757521777081, "grad_norm": 2.072382923032563, "learning_rate": 8.305580937646476e-06, "loss": 0.5065, "step": 15931 }, { "epoch": 0.5676113793042022, "grad_norm": 1.0945678320788115, "learning_rate": 8.304443697955066e-06, "loss": 0.4242, "step": 15932 }, { "epoch": 0.5676470064306963, "grad_norm": 1.1344458112397853, "learning_rate": 8.30330648084117e-06, "loss": 0.3699, "step": 15933 }, { "epoch": 0.5676826335571904, "grad_norm": 1.2539800775971262, "learning_rate": 8.30216928631994e-06, "loss": 0.5584, "step": 15934 }, { "epoch": 0.5677182606836846, "grad_norm": 1.7076634948752385, "learning_rate": 8.301032114406507e-06, "loss": 0.4888, "step": 15935 }, { "epoch": 0.5677538878101787, "grad_norm": 1.56048238936528, "learning_rate": 8.299894965116022e-06, "loss": 0.5198, "step": 15936 }, { "epoch": 0.5677895149366727, "grad_norm": 1.2606448854164065, "learning_rate": 8.298757838463622e-06, "loss": 0.4606, "step": 15937 }, { "epoch": 0.5678251420631669, "grad_norm": 1.4722759523767357, "learning_rate": 8.297620734464454e-06, "loss": 0.4982, "step": 15938 }, { "epoch": 0.567860769189661, "grad_norm": 1.4404118852892205, "learning_rate": 8.296483653133649e-06, "loss": 0.6232, "step": 15939 }, { "epoch": 0.5678963963161551, "grad_norm": 1.1957248801683507, "learning_rate": 8.295346594486358e-06, "loss": 0.4257, "step": 15940 }, { "epoch": 0.5679320234426493, "grad_norm": 1.4876847876522314, "learning_rate": 8.294209558537722e-06, "loss": 0.6083, "step": 15941 }, { "epoch": 0.5679676505691433, "grad_norm": 1.318048917327896, "learning_rate": 8.293072545302875e-06, "loss": 0.6159, "step": 15942 }, { "epoch": 0.5680032776956374, "grad_norm": 1.0811229306227086, "learning_rate": 8.291935554796962e-06, "loss": 0.4005, "step": 15943 }, { "epoch": 0.5680389048221316, "grad_norm": 1.0453230065512942, "learning_rate": 8.290798587035118e-06, "loss": 0.3984, "step": 15944 }, { "epoch": 0.5680745319486257, "grad_norm": 1.1725699218903642, "learning_rate": 8.28966164203249e-06, "loss": 0.4417, "step": 15945 }, { "epoch": 0.5681101590751197, "grad_norm": 1.1131539218368638, "learning_rate": 8.288524719804209e-06, "loss": 0.4346, "step": 15946 }, { "epoch": 0.5681457862016139, "grad_norm": 1.1526487773894896, "learning_rate": 8.287387820365418e-06, "loss": 0.4115, "step": 15947 }, { "epoch": 0.568181413328108, "grad_norm": 1.0220595121967646, "learning_rate": 8.286250943731254e-06, "loss": 0.3993, "step": 15948 }, { "epoch": 0.5682170404546021, "grad_norm": 1.0757615869373076, "learning_rate": 8.285114089916862e-06, "loss": 0.4481, "step": 15949 }, { "epoch": 0.5682526675810963, "grad_norm": 1.1160720844280319, "learning_rate": 8.283977258937367e-06, "loss": 0.3873, "step": 15950 }, { "epoch": 0.5682882947075903, "grad_norm": 1.4158659482636204, "learning_rate": 8.282840450807917e-06, "loss": 0.532, "step": 15951 }, { "epoch": 0.5683239218340844, "grad_norm": 1.4127881786798813, "learning_rate": 8.28170366554365e-06, "loss": 0.5665, "step": 15952 }, { "epoch": 0.5683595489605786, "grad_norm": 1.5975323837634416, "learning_rate": 8.280566903159695e-06, "loss": 0.5939, "step": 15953 }, { "epoch": 0.5683951760870727, "grad_norm": 1.1170221712872066, "learning_rate": 8.279430163671197e-06, "loss": 0.4162, "step": 15954 }, { "epoch": 0.5684308032135668, "grad_norm": 1.9161514819865109, "learning_rate": 8.278293447093286e-06, "loss": 0.5778, "step": 15955 }, { "epoch": 0.568466430340061, "grad_norm": 1.1733230632896139, "learning_rate": 8.277156753441106e-06, "loss": 0.4722, "step": 15956 }, { "epoch": 0.568502057466555, "grad_norm": 0.8891916547952453, "learning_rate": 8.276020082729784e-06, "loss": 0.3913, "step": 15957 }, { "epoch": 0.5685376845930491, "grad_norm": 1.0537243491955366, "learning_rate": 8.27488343497446e-06, "loss": 0.4425, "step": 15958 }, { "epoch": 0.5685733117195433, "grad_norm": 1.6875412622966557, "learning_rate": 8.27374681019027e-06, "loss": 0.6159, "step": 15959 }, { "epoch": 0.5686089388460374, "grad_norm": 1.0109740148231754, "learning_rate": 8.272610208392348e-06, "loss": 0.4042, "step": 15960 }, { "epoch": 0.5686445659725314, "grad_norm": 1.1316757753892672, "learning_rate": 8.271473629595827e-06, "loss": 0.3932, "step": 15961 }, { "epoch": 0.5686801930990256, "grad_norm": 1.0921214442680238, "learning_rate": 8.270337073815839e-06, "loss": 0.4386, "step": 15962 }, { "epoch": 0.5687158202255197, "grad_norm": 1.2548042304809601, "learning_rate": 8.26920054106753e-06, "loss": 0.4549, "step": 15963 }, { "epoch": 0.5687514473520138, "grad_norm": 1.7686051911956993, "learning_rate": 8.268064031366021e-06, "loss": 0.5378, "step": 15964 }, { "epoch": 0.568787074478508, "grad_norm": 1.1767189529013136, "learning_rate": 8.26692754472645e-06, "loss": 0.4624, "step": 15965 }, { "epoch": 0.568822701605002, "grad_norm": 2.3087003786576075, "learning_rate": 8.265791081163949e-06, "loss": 0.508, "step": 15966 }, { "epoch": 0.5688583287314961, "grad_norm": 1.0734508437140273, "learning_rate": 8.264654640693658e-06, "loss": 0.4265, "step": 15967 }, { "epoch": 0.5688939558579903, "grad_norm": 1.2935279424345636, "learning_rate": 8.263518223330698e-06, "loss": 0.4513, "step": 15968 }, { "epoch": 0.5689295829844844, "grad_norm": 1.1051379044923104, "learning_rate": 8.262381829090208e-06, "loss": 0.442, "step": 15969 }, { "epoch": 0.5689652101109784, "grad_norm": 1.399608116366029, "learning_rate": 8.261245457987318e-06, "loss": 0.5019, "step": 15970 }, { "epoch": 0.5690008372374726, "grad_norm": 1.641105622548857, "learning_rate": 8.260109110037163e-06, "loss": 0.4679, "step": 15971 }, { "epoch": 0.5690364643639667, "grad_norm": 1.0689572937294471, "learning_rate": 8.258972785254867e-06, "loss": 0.4527, "step": 15972 }, { "epoch": 0.5690720914904608, "grad_norm": 1.0551432496938202, "learning_rate": 8.257836483655566e-06, "loss": 0.3896, "step": 15973 }, { "epoch": 0.569107718616955, "grad_norm": 1.5854996808335664, "learning_rate": 8.25670020525439e-06, "loss": 0.5105, "step": 15974 }, { "epoch": 0.569143345743449, "grad_norm": 1.0305312807236005, "learning_rate": 8.255563950066468e-06, "loss": 0.4368, "step": 15975 }, { "epoch": 0.5691789728699431, "grad_norm": 1.8147869735514375, "learning_rate": 8.254427718106934e-06, "loss": 0.5379, "step": 15976 }, { "epoch": 0.5692145999964373, "grad_norm": 1.1158230099133495, "learning_rate": 8.253291509390913e-06, "loss": 0.4442, "step": 15977 }, { "epoch": 0.5692502271229314, "grad_norm": 1.350840906017846, "learning_rate": 8.25215532393354e-06, "loss": 0.602, "step": 15978 }, { "epoch": 0.5692858542494255, "grad_norm": 1.7307955773269195, "learning_rate": 8.251019161749938e-06, "loss": 0.555, "step": 15979 }, { "epoch": 0.5693214813759196, "grad_norm": 1.1274383390846574, "learning_rate": 8.249883022855238e-06, "loss": 0.4058, "step": 15980 }, { "epoch": 0.5693571085024137, "grad_norm": 1.9102268803272657, "learning_rate": 8.24874690726457e-06, "loss": 0.4648, "step": 15981 }, { "epoch": 0.5693927356289078, "grad_norm": 1.5286375108441466, "learning_rate": 8.247610814993062e-06, "loss": 0.5298, "step": 15982 }, { "epoch": 0.569428362755402, "grad_norm": 1.0497178867338557, "learning_rate": 8.246474746055837e-06, "loss": 0.415, "step": 15983 }, { "epoch": 0.5694639898818961, "grad_norm": 1.1396300295103599, "learning_rate": 8.24533870046803e-06, "loss": 0.4627, "step": 15984 }, { "epoch": 0.5694996170083901, "grad_norm": 1.3583366502766043, "learning_rate": 8.244202678244763e-06, "loss": 0.5686, "step": 15985 }, { "epoch": 0.5695352441348843, "grad_norm": 1.1467191895634827, "learning_rate": 8.243066679401162e-06, "loss": 0.4389, "step": 15986 }, { "epoch": 0.5695708712613784, "grad_norm": 1.0108213899724858, "learning_rate": 8.241930703952362e-06, "loss": 0.4059, "step": 15987 }, { "epoch": 0.5696064983878725, "grad_norm": 1.3704398726243083, "learning_rate": 8.240794751913476e-06, "loss": 0.5221, "step": 15988 }, { "epoch": 0.5696421255143667, "grad_norm": 1.524443556888392, "learning_rate": 8.239658823299647e-06, "loss": 0.4843, "step": 15989 }, { "epoch": 0.5696777526408607, "grad_norm": 1.1886113771093707, "learning_rate": 8.238522918125984e-06, "loss": 0.4346, "step": 15990 }, { "epoch": 0.5697133797673548, "grad_norm": 1.575184370599946, "learning_rate": 8.237387036407622e-06, "loss": 0.5046, "step": 15991 }, { "epoch": 0.569749006893849, "grad_norm": 1.5057441444254016, "learning_rate": 8.236251178159683e-06, "loss": 0.5562, "step": 15992 }, { "epoch": 0.5697846340203431, "grad_norm": 1.0289457698974673, "learning_rate": 8.235115343397295e-06, "loss": 0.4595, "step": 15993 }, { "epoch": 0.5698202611468371, "grad_norm": 1.545792321653449, "learning_rate": 8.233979532135576e-06, "loss": 0.5775, "step": 15994 }, { "epoch": 0.5698558882733313, "grad_norm": 1.5007351545630097, "learning_rate": 8.232843744389655e-06, "loss": 0.5455, "step": 15995 }, { "epoch": 0.5698915153998254, "grad_norm": 1.0687272469600857, "learning_rate": 8.231707980174655e-06, "loss": 0.43, "step": 15996 }, { "epoch": 0.5699271425263195, "grad_norm": 1.4713020153331344, "learning_rate": 8.230572239505697e-06, "loss": 0.6272, "step": 15997 }, { "epoch": 0.5699627696528137, "grad_norm": 1.0343382544379107, "learning_rate": 8.229436522397912e-06, "loss": 0.4284, "step": 15998 }, { "epoch": 0.5699983967793077, "grad_norm": 1.0717880608992716, "learning_rate": 8.228300828866412e-06, "loss": 0.4479, "step": 15999 }, { "epoch": 0.5700340239058018, "grad_norm": 1.6665373598849582, "learning_rate": 8.227165158926328e-06, "loss": 0.5494, "step": 16000 }, { "epoch": 0.570069651032296, "grad_norm": 1.6610553157815224, "learning_rate": 8.226029512592777e-06, "loss": 0.5148, "step": 16001 }, { "epoch": 0.5701052781587901, "grad_norm": 1.1470531576469412, "learning_rate": 8.224893889880884e-06, "loss": 0.4348, "step": 16002 }, { "epoch": 0.5701409052852842, "grad_norm": 1.6242534239665736, "learning_rate": 8.223758290805768e-06, "loss": 0.5699, "step": 16003 }, { "epoch": 0.5701765324117783, "grad_norm": 1.4888942751842849, "learning_rate": 8.222622715382555e-06, "loss": 0.5212, "step": 16004 }, { "epoch": 0.5702121595382724, "grad_norm": 1.12300601111344, "learning_rate": 8.22148716362636e-06, "loss": 0.4762, "step": 16005 }, { "epoch": 0.5702477866647665, "grad_norm": 1.68213365145231, "learning_rate": 8.220351635552308e-06, "loss": 0.6024, "step": 16006 }, { "epoch": 0.5702834137912607, "grad_norm": 1.375974381451403, "learning_rate": 8.219216131175516e-06, "loss": 0.4445, "step": 16007 }, { "epoch": 0.5703190409177548, "grad_norm": 1.162184050830276, "learning_rate": 8.218080650511107e-06, "loss": 0.4419, "step": 16008 }, { "epoch": 0.5703546680442488, "grad_norm": 1.0256385279181435, "learning_rate": 8.216945193574204e-06, "loss": 0.4023, "step": 16009 }, { "epoch": 0.570390295170743, "grad_norm": 1.0473077513251698, "learning_rate": 8.215809760379915e-06, "loss": 0.4254, "step": 16010 }, { "epoch": 0.5704259222972371, "grad_norm": 1.630796848608716, "learning_rate": 8.21467435094337e-06, "loss": 0.5616, "step": 16011 }, { "epoch": 0.5704615494237312, "grad_norm": 1.1552936044041726, "learning_rate": 8.213538965279681e-06, "loss": 0.3795, "step": 16012 }, { "epoch": 0.5704971765502254, "grad_norm": 1.4708584787772327, "learning_rate": 8.212403603403974e-06, "loss": 0.5586, "step": 16013 }, { "epoch": 0.5705328036767194, "grad_norm": 1.0613564926602748, "learning_rate": 8.211268265331358e-06, "loss": 0.4334, "step": 16014 }, { "epoch": 0.5705684308032136, "grad_norm": 1.4701744544295006, "learning_rate": 8.210132951076959e-06, "loss": 0.5643, "step": 16015 }, { "epoch": 0.5706040579297077, "grad_norm": 1.7352747924127905, "learning_rate": 8.208997660655887e-06, "loss": 0.5683, "step": 16016 }, { "epoch": 0.5706396850562018, "grad_norm": 1.3412662189923445, "learning_rate": 8.207862394083265e-06, "loss": 0.5258, "step": 16017 }, { "epoch": 0.570675312182696, "grad_norm": 1.1838603069626983, "learning_rate": 8.206727151374208e-06, "loss": 0.4809, "step": 16018 }, { "epoch": 0.57071093930919, "grad_norm": 1.5017625705581032, "learning_rate": 8.205591932543836e-06, "loss": 0.6341, "step": 16019 }, { "epoch": 0.5707465664356841, "grad_norm": 1.0072852180332474, "learning_rate": 8.204456737607255e-06, "loss": 0.4164, "step": 16020 }, { "epoch": 0.5707821935621783, "grad_norm": 1.0957563778593327, "learning_rate": 8.203321566579589e-06, "loss": 0.4685, "step": 16021 }, { "epoch": 0.5708178206886724, "grad_norm": 1.1700055252325088, "learning_rate": 8.202186419475955e-06, "loss": 0.4416, "step": 16022 }, { "epoch": 0.5708534478151664, "grad_norm": 2.1745756512825967, "learning_rate": 8.201051296311462e-06, "loss": 0.4791, "step": 16023 }, { "epoch": 0.5708890749416606, "grad_norm": 1.4522952098734496, "learning_rate": 8.199916197101233e-06, "loss": 0.453, "step": 16024 }, { "epoch": 0.5709247020681547, "grad_norm": 1.4398617986318538, "learning_rate": 8.19878112186037e-06, "loss": 0.495, "step": 16025 }, { "epoch": 0.5709603291946488, "grad_norm": 1.8627524588398996, "learning_rate": 8.197646070604004e-06, "loss": 0.5526, "step": 16026 }, { "epoch": 0.570995956321143, "grad_norm": 1.2903566991439535, "learning_rate": 8.196511043347237e-06, "loss": 0.4695, "step": 16027 }, { "epoch": 0.571031583447637, "grad_norm": 1.6123649352995106, "learning_rate": 8.195376040105186e-06, "loss": 0.4918, "step": 16028 }, { "epoch": 0.5710672105741311, "grad_norm": 1.8184720432500106, "learning_rate": 8.194241060892962e-06, "loss": 0.5134, "step": 16029 }, { "epoch": 0.5711028377006253, "grad_norm": 1.1113458383469865, "learning_rate": 8.193106105725686e-06, "loss": 0.4509, "step": 16030 }, { "epoch": 0.5711384648271194, "grad_norm": 1.3343752273472251, "learning_rate": 8.19197117461846e-06, "loss": 0.5228, "step": 16031 }, { "epoch": 0.5711740919536135, "grad_norm": 1.2208218717746948, "learning_rate": 8.190836267586403e-06, "loss": 0.4552, "step": 16032 }, { "epoch": 0.5712097190801076, "grad_norm": 1.2185856276974367, "learning_rate": 8.189701384644627e-06, "loss": 0.5169, "step": 16033 }, { "epoch": 0.5712453462066017, "grad_norm": 1.113302202668578, "learning_rate": 8.18856652580824e-06, "loss": 0.4609, "step": 16034 }, { "epoch": 0.5712809733330958, "grad_norm": 2.6383295191802314, "learning_rate": 8.187431691092361e-06, "loss": 0.4837, "step": 16035 }, { "epoch": 0.57131660045959, "grad_norm": 0.9901194426990254, "learning_rate": 8.18629688051209e-06, "loss": 0.3487, "step": 16036 }, { "epoch": 0.5713522275860841, "grad_norm": 1.9684992885000947, "learning_rate": 8.185162094082547e-06, "loss": 0.517, "step": 16037 }, { "epoch": 0.5713878547125781, "grad_norm": 1.4015929677916483, "learning_rate": 8.184027331818837e-06, "loss": 0.4762, "step": 16038 }, { "epoch": 0.5714234818390723, "grad_norm": 1.285270572492695, "learning_rate": 8.182892593736073e-06, "loss": 0.4786, "step": 16039 }, { "epoch": 0.5714591089655664, "grad_norm": 1.1743383750700787, "learning_rate": 8.181757879849364e-06, "loss": 0.4295, "step": 16040 }, { "epoch": 0.5714947360920605, "grad_norm": 1.0945998977303868, "learning_rate": 8.180623190173824e-06, "loss": 0.4305, "step": 16041 }, { "epoch": 0.5715303632185547, "grad_norm": 1.1534856403325808, "learning_rate": 8.179488524724553e-06, "loss": 0.4635, "step": 16042 }, { "epoch": 0.5715659903450487, "grad_norm": 1.0687457243006944, "learning_rate": 8.178353883516665e-06, "loss": 0.4625, "step": 16043 }, { "epoch": 0.5716016174715428, "grad_norm": 1.5523698075389152, "learning_rate": 8.177219266565269e-06, "loss": 0.5378, "step": 16044 }, { "epoch": 0.571637244598037, "grad_norm": 1.1691128565516842, "learning_rate": 8.176084673885471e-06, "loss": 0.4451, "step": 16045 }, { "epoch": 0.5716728717245311, "grad_norm": 1.0979950597541435, "learning_rate": 8.174950105492386e-06, "loss": 0.4544, "step": 16046 }, { "epoch": 0.5717084988510251, "grad_norm": 1.2920765640918892, "learning_rate": 8.173815561401112e-06, "loss": 0.4153, "step": 16047 }, { "epoch": 0.5717441259775193, "grad_norm": 1.0657207915733509, "learning_rate": 8.17268104162676e-06, "loss": 0.4242, "step": 16048 }, { "epoch": 0.5717797531040134, "grad_norm": 1.7183451107407244, "learning_rate": 8.171546546184436e-06, "loss": 0.5954, "step": 16049 }, { "epoch": 0.5718153802305075, "grad_norm": 1.3569183991856129, "learning_rate": 8.170412075089254e-06, "loss": 0.5716, "step": 16050 }, { "epoch": 0.5718510073570017, "grad_norm": 1.1185796693906462, "learning_rate": 8.169277628356305e-06, "loss": 0.4692, "step": 16051 }, { "epoch": 0.5718866344834957, "grad_norm": 1.1032092466086327, "learning_rate": 8.168143206000712e-06, "loss": 0.4274, "step": 16052 }, { "epoch": 0.5719222616099898, "grad_norm": 1.129952145034171, "learning_rate": 8.167008808037568e-06, "loss": 0.4735, "step": 16053 }, { "epoch": 0.571957888736484, "grad_norm": 1.5738201529205214, "learning_rate": 8.165874434481982e-06, "loss": 0.5413, "step": 16054 }, { "epoch": 0.5719935158629781, "grad_norm": 1.0938472936165322, "learning_rate": 8.164740085349062e-06, "loss": 0.4055, "step": 16055 }, { "epoch": 0.5720291429894722, "grad_norm": 1.7579146746136778, "learning_rate": 8.16360576065391e-06, "loss": 0.4999, "step": 16056 }, { "epoch": 0.5720647701159663, "grad_norm": 1.0986469977338709, "learning_rate": 8.162471460411634e-06, "loss": 0.4632, "step": 16057 }, { "epoch": 0.5721003972424604, "grad_norm": 1.132257665543747, "learning_rate": 8.16133718463733e-06, "loss": 0.4377, "step": 16058 }, { "epoch": 0.5721360243689545, "grad_norm": 1.3176866963583942, "learning_rate": 8.16020293334611e-06, "loss": 0.4709, "step": 16059 }, { "epoch": 0.5721716514954487, "grad_norm": 1.618452459835543, "learning_rate": 8.15906870655307e-06, "loss": 0.4809, "step": 16060 }, { "epoch": 0.5722072786219428, "grad_norm": 1.1134208297486574, "learning_rate": 8.157934504273323e-06, "loss": 0.44, "step": 16061 }, { "epoch": 0.5722429057484368, "grad_norm": 1.0624769565725838, "learning_rate": 8.156800326521961e-06, "loss": 0.393, "step": 16062 }, { "epoch": 0.572278532874931, "grad_norm": 1.6132262694519122, "learning_rate": 8.155666173314092e-06, "loss": 0.5908, "step": 16063 }, { "epoch": 0.5723141600014251, "grad_norm": 1.1188041903108124, "learning_rate": 8.154532044664815e-06, "loss": 0.4462, "step": 16064 }, { "epoch": 0.5723497871279192, "grad_norm": 1.1391977867969967, "learning_rate": 8.153397940589238e-06, "loss": 0.428, "step": 16065 }, { "epoch": 0.5723854142544134, "grad_norm": 1.1477945477589435, "learning_rate": 8.152263861102455e-06, "loss": 0.4664, "step": 16066 }, { "epoch": 0.5724210413809074, "grad_norm": 1.1233415108174687, "learning_rate": 8.15112980621957e-06, "loss": 0.4221, "step": 16067 }, { "epoch": 0.5724566685074015, "grad_norm": 1.872235919684611, "learning_rate": 8.149995775955686e-06, "loss": 0.5534, "step": 16068 }, { "epoch": 0.5724922956338957, "grad_norm": 1.2862760090890593, "learning_rate": 8.148861770325899e-06, "loss": 0.4521, "step": 16069 }, { "epoch": 0.5725279227603898, "grad_norm": 1.464428764702581, "learning_rate": 8.147727789345313e-06, "loss": 0.5905, "step": 16070 }, { "epoch": 0.5725635498868838, "grad_norm": 1.1470681985262816, "learning_rate": 8.146593833029024e-06, "loss": 0.4107, "step": 16071 }, { "epoch": 0.572599177013378, "grad_norm": 2.211267619577454, "learning_rate": 8.145459901392138e-06, "loss": 0.5802, "step": 16072 }, { "epoch": 0.5726348041398721, "grad_norm": 1.118821963984399, "learning_rate": 8.144325994449746e-06, "loss": 0.4651, "step": 16073 }, { "epoch": 0.5726704312663662, "grad_norm": 0.9971410681493665, "learning_rate": 8.143192112216953e-06, "loss": 0.3878, "step": 16074 }, { "epoch": 0.5727060583928604, "grad_norm": 1.3904472456309709, "learning_rate": 8.142058254708853e-06, "loss": 0.5621, "step": 16075 }, { "epoch": 0.5727416855193544, "grad_norm": 1.1230733660432428, "learning_rate": 8.140924421940548e-06, "loss": 0.4317, "step": 16076 }, { "epoch": 0.5727773126458485, "grad_norm": 1.0203356035992381, "learning_rate": 8.13979061392713e-06, "loss": 0.4264, "step": 16077 }, { "epoch": 0.5728129397723427, "grad_norm": 1.2398854363177347, "learning_rate": 8.1386568306837e-06, "loss": 0.4268, "step": 16078 }, { "epoch": 0.5728485668988368, "grad_norm": 1.634751037779834, "learning_rate": 8.13752307222536e-06, "loss": 0.5766, "step": 16079 }, { "epoch": 0.5728841940253309, "grad_norm": 1.7379282694905742, "learning_rate": 8.136389338567198e-06, "loss": 0.4533, "step": 16080 }, { "epoch": 0.572919821151825, "grad_norm": 2.1128873594789415, "learning_rate": 8.135255629724317e-06, "loss": 0.4958, "step": 16081 }, { "epoch": 0.5729554482783191, "grad_norm": 1.2101432081834071, "learning_rate": 8.134121945711808e-06, "loss": 0.5689, "step": 16082 }, { "epoch": 0.5729910754048132, "grad_norm": 1.647428875655423, "learning_rate": 8.132988286544773e-06, "loss": 0.6101, "step": 16083 }, { "epoch": 0.5730267025313074, "grad_norm": 1.0114945020599517, "learning_rate": 8.131854652238302e-06, "loss": 0.417, "step": 16084 }, { "epoch": 0.5730623296578015, "grad_norm": 1.5984702747324564, "learning_rate": 8.130721042807494e-06, "loss": 0.5649, "step": 16085 }, { "epoch": 0.5730979567842955, "grad_norm": 1.098388767211458, "learning_rate": 8.129587458267439e-06, "loss": 0.4457, "step": 16086 }, { "epoch": 0.5731335839107897, "grad_norm": 1.1574140124569272, "learning_rate": 8.128453898633237e-06, "loss": 0.4458, "step": 16087 }, { "epoch": 0.5731692110372838, "grad_norm": 1.0484725095988259, "learning_rate": 8.127320363919976e-06, "loss": 0.4232, "step": 16088 }, { "epoch": 0.5732048381637779, "grad_norm": 1.5912388318225998, "learning_rate": 8.126186854142752e-06, "loss": 0.571, "step": 16089 }, { "epoch": 0.5732404652902721, "grad_norm": 1.7964289856104738, "learning_rate": 8.125053369316666e-06, "loss": 0.4915, "step": 16090 }, { "epoch": 0.5732760924167661, "grad_norm": 1.0333847658959046, "learning_rate": 8.123919909456801e-06, "loss": 0.4769, "step": 16091 }, { "epoch": 0.5733117195432602, "grad_norm": 1.830937733895122, "learning_rate": 8.122786474578255e-06, "loss": 0.5658, "step": 16092 }, { "epoch": 0.5733473466697544, "grad_norm": 1.0955402710450963, "learning_rate": 8.121653064696118e-06, "loss": 0.4521, "step": 16093 }, { "epoch": 0.5733829737962485, "grad_norm": 1.6826153549192202, "learning_rate": 8.120519679825487e-06, "loss": 0.5154, "step": 16094 }, { "epoch": 0.5734186009227425, "grad_norm": 1.1721854488161518, "learning_rate": 8.119386319981445e-06, "loss": 0.4818, "step": 16095 }, { "epoch": 0.5734542280492367, "grad_norm": 1.0310637604927841, "learning_rate": 8.118252985179091e-06, "loss": 0.4187, "step": 16096 }, { "epoch": 0.5734898551757308, "grad_norm": 1.7193545429515398, "learning_rate": 8.117119675433514e-06, "loss": 0.608, "step": 16097 }, { "epoch": 0.5735254823022249, "grad_norm": 1.8749151353198623, "learning_rate": 8.115986390759805e-06, "loss": 0.5781, "step": 16098 }, { "epoch": 0.5735611094287191, "grad_norm": 1.150405960478417, "learning_rate": 8.114853131173053e-06, "loss": 0.4695, "step": 16099 }, { "epoch": 0.5735967365552131, "grad_norm": 1.64964035795727, "learning_rate": 8.11371989668835e-06, "loss": 0.6049, "step": 16100 }, { "epoch": 0.5736323636817072, "grad_norm": 0.9572400884150143, "learning_rate": 8.112586687320783e-06, "loss": 0.4403, "step": 16101 }, { "epoch": 0.5736679908082014, "grad_norm": 1.8253534903037305, "learning_rate": 8.111453503085444e-06, "loss": 0.5503, "step": 16102 }, { "epoch": 0.5737036179346955, "grad_norm": 1.18520416910468, "learning_rate": 8.110320343997423e-06, "loss": 0.4059, "step": 16103 }, { "epoch": 0.5737392450611896, "grad_norm": 1.1558437211790091, "learning_rate": 8.109187210071806e-06, "loss": 0.4863, "step": 16104 }, { "epoch": 0.5737748721876837, "grad_norm": 1.634394904043667, "learning_rate": 8.108054101323686e-06, "loss": 0.5573, "step": 16105 }, { "epoch": 0.5738104993141778, "grad_norm": 1.770173151148472, "learning_rate": 8.106921017768145e-06, "loss": 0.5575, "step": 16106 }, { "epoch": 0.5738461264406719, "grad_norm": 1.058604310695927, "learning_rate": 8.105787959420274e-06, "loss": 0.4928, "step": 16107 }, { "epoch": 0.5738817535671661, "grad_norm": 1.6295180321097475, "learning_rate": 8.104654926295161e-06, "loss": 0.6167, "step": 16108 }, { "epoch": 0.5739173806936602, "grad_norm": 1.014046209152356, "learning_rate": 8.103521918407894e-06, "loss": 0.4568, "step": 16109 }, { "epoch": 0.5739530078201542, "grad_norm": 1.084546750036943, "learning_rate": 8.102388935773556e-06, "loss": 0.4323, "step": 16110 }, { "epoch": 0.5739886349466484, "grad_norm": 1.4556428766395217, "learning_rate": 8.101255978407237e-06, "loss": 0.5457, "step": 16111 }, { "epoch": 0.5740242620731425, "grad_norm": 1.089209660786971, "learning_rate": 8.10012304632402e-06, "loss": 0.48, "step": 16112 }, { "epoch": 0.5740598891996366, "grad_norm": 0.9967402348561264, "learning_rate": 8.098990139538992e-06, "loss": 0.4255, "step": 16113 }, { "epoch": 0.5740955163261308, "grad_norm": 1.0991807651941758, "learning_rate": 8.097857258067242e-06, "loss": 0.4444, "step": 16114 }, { "epoch": 0.5741311434526248, "grad_norm": 1.7507455990050929, "learning_rate": 8.09672440192385e-06, "loss": 0.588, "step": 16115 }, { "epoch": 0.5741667705791189, "grad_norm": 1.5445684770424501, "learning_rate": 8.095591571123906e-06, "loss": 0.4924, "step": 16116 }, { "epoch": 0.5742023977056131, "grad_norm": 1.2898009181495043, "learning_rate": 8.094458765682487e-06, "loss": 0.4881, "step": 16117 }, { "epoch": 0.5742380248321072, "grad_norm": 1.1072093467143875, "learning_rate": 8.093325985614686e-06, "loss": 0.3875, "step": 16118 }, { "epoch": 0.5742736519586013, "grad_norm": 1.0242739985845366, "learning_rate": 8.092193230935577e-06, "loss": 0.4345, "step": 16119 }, { "epoch": 0.5743092790850954, "grad_norm": 0.9920498480576289, "learning_rate": 8.091060501660254e-06, "loss": 0.401, "step": 16120 }, { "epoch": 0.5743449062115895, "grad_norm": 1.1284432047390076, "learning_rate": 8.089927797803791e-06, "loss": 0.413, "step": 16121 }, { "epoch": 0.5743805333380836, "grad_norm": 1.622541254208122, "learning_rate": 8.088795119381277e-06, "loss": 0.5292, "step": 16122 }, { "epoch": 0.5744161604645778, "grad_norm": 1.5024414285033645, "learning_rate": 8.087662466407788e-06, "loss": 0.5765, "step": 16123 }, { "epoch": 0.5744517875910718, "grad_norm": 1.8806192356265514, "learning_rate": 8.086529838898411e-06, "loss": 0.5871, "step": 16124 }, { "epoch": 0.5744874147175659, "grad_norm": 1.0997438259372607, "learning_rate": 8.08539723686823e-06, "loss": 0.4284, "step": 16125 }, { "epoch": 0.5745230418440601, "grad_norm": 1.6466101328470797, "learning_rate": 8.084264660332316e-06, "loss": 0.5812, "step": 16126 }, { "epoch": 0.5745586689705542, "grad_norm": 1.1500301048115573, "learning_rate": 8.083132109305766e-06, "loss": 0.4892, "step": 16127 }, { "epoch": 0.5745942960970483, "grad_norm": 1.6888448175500776, "learning_rate": 8.081999583803646e-06, "loss": 0.5746, "step": 16128 }, { "epoch": 0.5746299232235424, "grad_norm": 1.4625058812083795, "learning_rate": 8.080867083841045e-06, "loss": 0.5784, "step": 16129 }, { "epoch": 0.5746655503500365, "grad_norm": 1.4678030464627712, "learning_rate": 8.079734609433038e-06, "loss": 0.5742, "step": 16130 }, { "epoch": 0.5747011774765306, "grad_norm": 1.1954858879033101, "learning_rate": 8.07860216059471e-06, "loss": 0.4364, "step": 16131 }, { "epoch": 0.5747368046030248, "grad_norm": 1.528658297809885, "learning_rate": 8.077469737341134e-06, "loss": 0.6233, "step": 16132 }, { "epoch": 0.5747724317295189, "grad_norm": 1.7523539799026708, "learning_rate": 8.076337339687395e-06, "loss": 0.4989, "step": 16133 }, { "epoch": 0.5748080588560129, "grad_norm": 2.0280327671753917, "learning_rate": 8.075204967648565e-06, "loss": 0.5023, "step": 16134 }, { "epoch": 0.5748436859825071, "grad_norm": 1.6810762757510838, "learning_rate": 8.074072621239726e-06, "loss": 0.4994, "step": 16135 }, { "epoch": 0.5748793131090012, "grad_norm": 1.0538250786090742, "learning_rate": 8.072940300475963e-06, "loss": 0.4451, "step": 16136 }, { "epoch": 0.5749149402354953, "grad_norm": 1.5151787240308703, "learning_rate": 8.07180800537234e-06, "loss": 0.499, "step": 16137 }, { "epoch": 0.5749505673619895, "grad_norm": 1.1514670791270598, "learning_rate": 8.070675735943945e-06, "loss": 0.48, "step": 16138 }, { "epoch": 0.5749861944884835, "grad_norm": 1.9599227964385255, "learning_rate": 8.069543492205847e-06, "loss": 0.6087, "step": 16139 }, { "epoch": 0.5750218216149776, "grad_norm": 1.4455932929950543, "learning_rate": 8.06841127417313e-06, "loss": 0.5244, "step": 16140 }, { "epoch": 0.5750574487414718, "grad_norm": 1.1539947260110865, "learning_rate": 8.067279081860866e-06, "loss": 0.4837, "step": 16141 }, { "epoch": 0.5750930758679659, "grad_norm": 1.3897325624036865, "learning_rate": 8.066146915284134e-06, "loss": 0.5557, "step": 16142 }, { "epoch": 0.57512870299446, "grad_norm": 1.0194286040944431, "learning_rate": 8.065014774458004e-06, "loss": 0.4177, "step": 16143 }, { "epoch": 0.5751643301209541, "grad_norm": 3.7335613200808573, "learning_rate": 8.063882659397556e-06, "loss": 0.5516, "step": 16144 }, { "epoch": 0.5751999572474482, "grad_norm": 1.1729751073433499, "learning_rate": 8.062750570117865e-06, "loss": 0.4271, "step": 16145 }, { "epoch": 0.5752355843739423, "grad_norm": 1.140379008023266, "learning_rate": 8.061618506634006e-06, "loss": 0.4376, "step": 16146 }, { "epoch": 0.5752712115004365, "grad_norm": 1.1570464676558956, "learning_rate": 8.060486468961047e-06, "loss": 0.435, "step": 16147 }, { "epoch": 0.5753068386269306, "grad_norm": 1.2071654402369505, "learning_rate": 8.059354457114066e-06, "loss": 0.4562, "step": 16148 }, { "epoch": 0.5753424657534246, "grad_norm": 1.2252694217617748, "learning_rate": 8.05822247110814e-06, "loss": 0.4568, "step": 16149 }, { "epoch": 0.5753780928799188, "grad_norm": 1.0763806877464577, "learning_rate": 8.057090510958336e-06, "loss": 0.4337, "step": 16150 }, { "epoch": 0.5754137200064129, "grad_norm": 1.1248813799153, "learning_rate": 8.055958576679735e-06, "loss": 0.3866, "step": 16151 }, { "epoch": 0.575449347132907, "grad_norm": 1.1053122340603365, "learning_rate": 8.054826668287398e-06, "loss": 0.4646, "step": 16152 }, { "epoch": 0.5754849742594011, "grad_norm": 1.0300557036344522, "learning_rate": 8.05369478579641e-06, "loss": 0.4248, "step": 16153 }, { "epoch": 0.5755206013858952, "grad_norm": 1.4039702568399033, "learning_rate": 8.05256292922183e-06, "loss": 0.5714, "step": 16154 }, { "epoch": 0.5755562285123893, "grad_norm": 1.7174468863640642, "learning_rate": 8.051431098578742e-06, "loss": 0.5951, "step": 16155 }, { "epoch": 0.5755918556388835, "grad_norm": 1.1015528457477823, "learning_rate": 8.050299293882205e-06, "loss": 0.468, "step": 16156 }, { "epoch": 0.5756274827653776, "grad_norm": 1.494534444590072, "learning_rate": 8.0491675151473e-06, "loss": 0.457, "step": 16157 }, { "epoch": 0.5756631098918716, "grad_norm": 1.033045993085268, "learning_rate": 8.048035762389091e-06, "loss": 0.409, "step": 16158 }, { "epoch": 0.5756987370183658, "grad_norm": 1.1055254959430862, "learning_rate": 8.046904035622648e-06, "loss": 0.4136, "step": 16159 }, { "epoch": 0.5757343641448599, "grad_norm": 1.7653104983483343, "learning_rate": 8.045772334863046e-06, "loss": 0.4857, "step": 16160 }, { "epoch": 0.575769991271354, "grad_norm": 1.4587487166642632, "learning_rate": 8.044640660125351e-06, "loss": 0.603, "step": 16161 }, { "epoch": 0.5758056183978482, "grad_norm": 1.1440353812516453, "learning_rate": 8.043509011424634e-06, "loss": 0.4459, "step": 16162 }, { "epoch": 0.5758412455243422, "grad_norm": 1.2195194690418654, "learning_rate": 8.04237738877596e-06, "loss": 0.435, "step": 16163 }, { "epoch": 0.5758768726508363, "grad_norm": 1.8538182798287086, "learning_rate": 8.0412457921944e-06, "loss": 0.5533, "step": 16164 }, { "epoch": 0.5759124997773305, "grad_norm": 1.7554539381072616, "learning_rate": 8.040114221695019e-06, "loss": 0.5578, "step": 16165 }, { "epoch": 0.5759481269038246, "grad_norm": 1.2950519295100122, "learning_rate": 8.038982677292889e-06, "loss": 0.5284, "step": 16166 }, { "epoch": 0.5759837540303187, "grad_norm": 1.1316533539232705, "learning_rate": 8.037851159003074e-06, "loss": 0.4443, "step": 16167 }, { "epoch": 0.5760193811568128, "grad_norm": 1.0802260373928472, "learning_rate": 8.036719666840646e-06, "loss": 0.473, "step": 16168 }, { "epoch": 0.5760550082833069, "grad_norm": 1.173131189901881, "learning_rate": 8.035588200820665e-06, "loss": 0.4039, "step": 16169 }, { "epoch": 0.576090635409801, "grad_norm": 1.6977565116782034, "learning_rate": 8.034456760958198e-06, "loss": 0.5607, "step": 16170 }, { "epoch": 0.5761262625362952, "grad_norm": 1.1125820831680335, "learning_rate": 8.033325347268318e-06, "loss": 0.4409, "step": 16171 }, { "epoch": 0.5761618896627893, "grad_norm": 1.5300803173481992, "learning_rate": 8.03219395976608e-06, "loss": 0.4331, "step": 16172 }, { "epoch": 0.5761975167892833, "grad_norm": 1.1255966411094094, "learning_rate": 8.031062598466561e-06, "loss": 0.4699, "step": 16173 }, { "epoch": 0.5762331439157775, "grad_norm": 1.0598012868721824, "learning_rate": 8.029931263384816e-06, "loss": 0.4491, "step": 16174 }, { "epoch": 0.5762687710422716, "grad_norm": 1.3067110003710254, "learning_rate": 8.028799954535915e-06, "loss": 0.4785, "step": 16175 }, { "epoch": 0.5763043981687657, "grad_norm": 1.1268319733713554, "learning_rate": 8.027668671934917e-06, "loss": 0.4652, "step": 16176 }, { "epoch": 0.5763400252952599, "grad_norm": 1.8436378020503454, "learning_rate": 8.026537415596895e-06, "loss": 0.5832, "step": 16177 }, { "epoch": 0.5763756524217539, "grad_norm": 1.507916845301635, "learning_rate": 8.0254061855369e-06, "loss": 0.5374, "step": 16178 }, { "epoch": 0.576411279548248, "grad_norm": 1.0389087448915773, "learning_rate": 8.024274981770006e-06, "loss": 0.391, "step": 16179 }, { "epoch": 0.5764469066747422, "grad_norm": 1.7916636590774968, "learning_rate": 8.023143804311269e-06, "loss": 0.6353, "step": 16180 }, { "epoch": 0.5764825338012363, "grad_norm": 1.2370849705404237, "learning_rate": 8.022012653175754e-06, "loss": 0.5126, "step": 16181 }, { "epoch": 0.5765181609277303, "grad_norm": 1.604329628339756, "learning_rate": 8.020881528378525e-06, "loss": 0.5106, "step": 16182 }, { "epoch": 0.5765537880542245, "grad_norm": 1.6946817373177316, "learning_rate": 8.019750429934641e-06, "loss": 0.6002, "step": 16183 }, { "epoch": 0.5765894151807186, "grad_norm": 1.1530408532185354, "learning_rate": 8.018619357859167e-06, "loss": 0.4048, "step": 16184 }, { "epoch": 0.5766250423072127, "grad_norm": 1.395602494714261, "learning_rate": 8.017488312167158e-06, "loss": 0.5723, "step": 16185 }, { "epoch": 0.5766606694337069, "grad_norm": 1.6101981024594874, "learning_rate": 8.016357292873679e-06, "loss": 0.5869, "step": 16186 }, { "epoch": 0.5766962965602009, "grad_norm": 1.4104931634021876, "learning_rate": 8.015226299993786e-06, "loss": 0.6258, "step": 16187 }, { "epoch": 0.576731923686695, "grad_norm": 1.3906875280978748, "learning_rate": 8.014095333542548e-06, "loss": 0.5088, "step": 16188 }, { "epoch": 0.5767675508131892, "grad_norm": 1.305815914775919, "learning_rate": 8.012964393535015e-06, "loss": 0.4097, "step": 16189 }, { "epoch": 0.5768031779396833, "grad_norm": 1.1518173732449568, "learning_rate": 8.011833479986251e-06, "loss": 0.4695, "step": 16190 }, { "epoch": 0.5768388050661774, "grad_norm": 1.1264635116034998, "learning_rate": 8.010702592911311e-06, "loss": 0.4697, "step": 16191 }, { "epoch": 0.5768744321926715, "grad_norm": 1.8298048311644801, "learning_rate": 8.00957173232526e-06, "loss": 0.5959, "step": 16192 }, { "epoch": 0.5769100593191656, "grad_norm": 1.1528381036819642, "learning_rate": 8.008440898243149e-06, "loss": 0.4362, "step": 16193 }, { "epoch": 0.5769456864456597, "grad_norm": 1.7555139003649507, "learning_rate": 8.00731009068004e-06, "loss": 0.6446, "step": 16194 }, { "epoch": 0.5769813135721539, "grad_norm": 1.0712545264283204, "learning_rate": 8.006179309650996e-06, "loss": 0.4728, "step": 16195 }, { "epoch": 0.577016940698648, "grad_norm": 1.1496359239481702, "learning_rate": 8.005048555171062e-06, "loss": 0.4864, "step": 16196 }, { "epoch": 0.577052567825142, "grad_norm": 1.4509192163477875, "learning_rate": 8.003917827255303e-06, "loss": 0.584, "step": 16197 }, { "epoch": 0.5770881949516362, "grad_norm": 2.0162160331698065, "learning_rate": 8.002787125918771e-06, "loss": 0.5668, "step": 16198 }, { "epoch": 0.5771238220781303, "grad_norm": 1.7848563543408251, "learning_rate": 8.001656451176529e-06, "loss": 0.5279, "step": 16199 }, { "epoch": 0.5771594492046244, "grad_norm": 1.1398843067048496, "learning_rate": 8.000525803043623e-06, "loss": 0.4505, "step": 16200 }, { "epoch": 0.5771950763311186, "grad_norm": 1.3462916773613813, "learning_rate": 7.999395181535118e-06, "loss": 0.446, "step": 16201 }, { "epoch": 0.5772307034576126, "grad_norm": 1.143718379441114, "learning_rate": 7.998264586666062e-06, "loss": 0.4439, "step": 16202 }, { "epoch": 0.5772663305841067, "grad_norm": 1.3200436236397244, "learning_rate": 7.997134018451513e-06, "loss": 0.489, "step": 16203 }, { "epoch": 0.5773019577106009, "grad_norm": 1.2767977385569829, "learning_rate": 7.996003476906521e-06, "loss": 0.4812, "step": 16204 }, { "epoch": 0.577337584837095, "grad_norm": 1.746654580978113, "learning_rate": 7.994872962046146e-06, "loss": 0.4956, "step": 16205 }, { "epoch": 0.577373211963589, "grad_norm": 1.550219642876046, "learning_rate": 7.993742473885441e-06, "loss": 0.571, "step": 16206 }, { "epoch": 0.5774088390900832, "grad_norm": 1.4248835184913855, "learning_rate": 7.992612012439456e-06, "loss": 0.5062, "step": 16207 }, { "epoch": 0.5774444662165773, "grad_norm": 1.1732402318487958, "learning_rate": 7.991481577723244e-06, "loss": 0.453, "step": 16208 }, { "epoch": 0.5774800933430714, "grad_norm": 1.4074124059952247, "learning_rate": 7.990351169751859e-06, "loss": 0.4817, "step": 16209 }, { "epoch": 0.5775157204695656, "grad_norm": 1.1511139413998677, "learning_rate": 7.989220788540356e-06, "loss": 0.4177, "step": 16210 }, { "epoch": 0.5775513475960596, "grad_norm": 1.1416346882795905, "learning_rate": 7.988090434103779e-06, "loss": 0.4289, "step": 16211 }, { "epoch": 0.5775869747225537, "grad_norm": 1.1467405639487764, "learning_rate": 7.986960106457188e-06, "loss": 0.4612, "step": 16212 }, { "epoch": 0.5776226018490479, "grad_norm": 2.09760775980879, "learning_rate": 7.985829805615626e-06, "loss": 0.537, "step": 16213 }, { "epoch": 0.577658228975542, "grad_norm": 1.3700558837295886, "learning_rate": 7.984699531594153e-06, "loss": 0.5337, "step": 16214 }, { "epoch": 0.577693856102036, "grad_norm": 1.1501470562852445, "learning_rate": 7.983569284407808e-06, "loss": 0.4149, "step": 16215 }, { "epoch": 0.5777294832285302, "grad_norm": 1.4469987359549967, "learning_rate": 7.982439064071649e-06, "loss": 0.429, "step": 16216 }, { "epoch": 0.5777651103550243, "grad_norm": 1.0984510444055113, "learning_rate": 7.98130887060073e-06, "loss": 0.4521, "step": 16217 }, { "epoch": 0.5778007374815184, "grad_norm": 1.0701928704080754, "learning_rate": 7.980178704010089e-06, "loss": 0.4302, "step": 16218 }, { "epoch": 0.5778363646080126, "grad_norm": 1.1757535281057898, "learning_rate": 7.979048564314782e-06, "loss": 0.4408, "step": 16219 }, { "epoch": 0.5778719917345067, "grad_norm": 1.181218287987536, "learning_rate": 7.977918451529855e-06, "loss": 0.4402, "step": 16220 }, { "epoch": 0.5779076188610007, "grad_norm": 1.3530660086549167, "learning_rate": 7.976788365670361e-06, "loss": 0.4352, "step": 16221 }, { "epoch": 0.5779432459874949, "grad_norm": 1.7201084237161561, "learning_rate": 7.97565830675134e-06, "loss": 0.51, "step": 16222 }, { "epoch": 0.577978873113989, "grad_norm": 1.1867037682926196, "learning_rate": 7.974528274787843e-06, "loss": 0.4682, "step": 16223 }, { "epoch": 0.5780145002404831, "grad_norm": 1.2109350039654168, "learning_rate": 7.973398269794919e-06, "loss": 0.4474, "step": 16224 }, { "epoch": 0.5780501273669773, "grad_norm": 1.4472871742647, "learning_rate": 7.972268291787616e-06, "loss": 0.5627, "step": 16225 }, { "epoch": 0.5780857544934713, "grad_norm": 1.1972065618985734, "learning_rate": 7.971138340780975e-06, "loss": 0.3957, "step": 16226 }, { "epoch": 0.5781213816199654, "grad_norm": 2.763738907380315, "learning_rate": 7.970008416790046e-06, "loss": 0.5486, "step": 16227 }, { "epoch": 0.5781570087464596, "grad_norm": 1.329292416541038, "learning_rate": 7.968878519829872e-06, "loss": 0.4981, "step": 16228 }, { "epoch": 0.5781926358729537, "grad_norm": 2.8594221079933857, "learning_rate": 7.967748649915502e-06, "loss": 0.5469, "step": 16229 }, { "epoch": 0.5782282629994477, "grad_norm": 1.4431018673133804, "learning_rate": 7.96661880706198e-06, "loss": 0.5178, "step": 16230 }, { "epoch": 0.5782638901259419, "grad_norm": 1.403101402058078, "learning_rate": 7.965488991284346e-06, "loss": 0.4737, "step": 16231 }, { "epoch": 0.578299517252436, "grad_norm": 1.7493992978124422, "learning_rate": 7.964359202597654e-06, "loss": 0.4989, "step": 16232 }, { "epoch": 0.5783351443789301, "grad_norm": 1.4544230130814644, "learning_rate": 7.963229441016938e-06, "loss": 0.5758, "step": 16233 }, { "epoch": 0.5783707715054243, "grad_norm": 1.628055856739365, "learning_rate": 7.962099706557248e-06, "loss": 0.4708, "step": 16234 }, { "epoch": 0.5784063986319183, "grad_norm": 1.17599182761655, "learning_rate": 7.960969999233622e-06, "loss": 0.4272, "step": 16235 }, { "epoch": 0.5784420257584124, "grad_norm": 1.0808275176843678, "learning_rate": 7.959840319061108e-06, "loss": 0.4555, "step": 16236 }, { "epoch": 0.5784776528849066, "grad_norm": 1.6757839419343492, "learning_rate": 7.958710666054744e-06, "loss": 0.5457, "step": 16237 }, { "epoch": 0.5785132800114007, "grad_norm": 1.4626791017612277, "learning_rate": 7.957581040229577e-06, "loss": 0.5963, "step": 16238 }, { "epoch": 0.5785489071378948, "grad_norm": 1.217398812495717, "learning_rate": 7.956451441600642e-06, "loss": 0.4141, "step": 16239 }, { "epoch": 0.5785845342643889, "grad_norm": 1.50838640852277, "learning_rate": 7.955321870182985e-06, "loss": 0.5315, "step": 16240 }, { "epoch": 0.578620161390883, "grad_norm": 1.1933168099460312, "learning_rate": 7.95419232599165e-06, "loss": 0.451, "step": 16241 }, { "epoch": 0.5786557885173771, "grad_norm": 1.5567108126292433, "learning_rate": 7.953062809041671e-06, "loss": 0.5093, "step": 16242 }, { "epoch": 0.5786914156438713, "grad_norm": 1.5304431033368355, "learning_rate": 7.951933319348095e-06, "loss": 0.5561, "step": 16243 }, { "epoch": 0.5787270427703654, "grad_norm": 1.1306875588256473, "learning_rate": 7.950803856925955e-06, "loss": 0.4896, "step": 16244 }, { "epoch": 0.5787626698968594, "grad_norm": 1.5387365091840055, "learning_rate": 7.949674421790297e-06, "loss": 0.5712, "step": 16245 }, { "epoch": 0.5787982970233536, "grad_norm": 2.0305258981381233, "learning_rate": 7.948545013956152e-06, "loss": 0.6178, "step": 16246 }, { "epoch": 0.5788339241498477, "grad_norm": 1.6396946617568324, "learning_rate": 7.94741563343857e-06, "loss": 0.5983, "step": 16247 }, { "epoch": 0.5788695512763418, "grad_norm": 1.3374975596114342, "learning_rate": 7.94628628025258e-06, "loss": 0.4267, "step": 16248 }, { "epoch": 0.578905178402836, "grad_norm": 1.5927166549440763, "learning_rate": 7.945156954413226e-06, "loss": 0.6178, "step": 16249 }, { "epoch": 0.57894080552933, "grad_norm": 1.5526238395846919, "learning_rate": 7.94402765593554e-06, "loss": 0.4582, "step": 16250 }, { "epoch": 0.5789764326558241, "grad_norm": 1.4426580827005913, "learning_rate": 7.942898384834564e-06, "loss": 0.5392, "step": 16251 }, { "epoch": 0.5790120597823183, "grad_norm": 1.1583954123373237, "learning_rate": 7.941769141125338e-06, "loss": 0.4358, "step": 16252 }, { "epoch": 0.5790476869088124, "grad_norm": 1.7498734629242063, "learning_rate": 7.940639924822886e-06, "loss": 0.4689, "step": 16253 }, { "epoch": 0.5790833140353064, "grad_norm": 1.6536671379176286, "learning_rate": 7.939510735942262e-06, "loss": 0.4449, "step": 16254 }, { "epoch": 0.5791189411618006, "grad_norm": 1.7039652622283286, "learning_rate": 7.938381574498485e-06, "loss": 0.4999, "step": 16255 }, { "epoch": 0.5791545682882947, "grad_norm": 1.0309963071482346, "learning_rate": 7.937252440506604e-06, "loss": 0.4331, "step": 16256 }, { "epoch": 0.5791901954147888, "grad_norm": 1.1692215368030041, "learning_rate": 7.936123333981645e-06, "loss": 0.4062, "step": 16257 }, { "epoch": 0.579225822541283, "grad_norm": 1.3657970540372677, "learning_rate": 7.93499425493865e-06, "loss": 0.5935, "step": 16258 }, { "epoch": 0.579261449667777, "grad_norm": 1.679547320723263, "learning_rate": 7.933865203392646e-06, "loss": 0.5206, "step": 16259 }, { "epoch": 0.5792970767942711, "grad_norm": 1.1247960104133865, "learning_rate": 7.932736179358674e-06, "loss": 0.4544, "step": 16260 }, { "epoch": 0.5793327039207653, "grad_norm": 1.7071351205006524, "learning_rate": 7.931607182851762e-06, "loss": 0.5082, "step": 16261 }, { "epoch": 0.5793683310472594, "grad_norm": 1.6958033008353723, "learning_rate": 7.930478213886947e-06, "loss": 0.5012, "step": 16262 }, { "epoch": 0.5794039581737535, "grad_norm": 1.5278934659860501, "learning_rate": 7.929349272479264e-06, "loss": 0.5684, "step": 16263 }, { "epoch": 0.5794395853002476, "grad_norm": 1.1517720675522995, "learning_rate": 7.92822035864374e-06, "loss": 0.4321, "step": 16264 }, { "epoch": 0.5794752124267417, "grad_norm": 1.3745928511062437, "learning_rate": 7.927091472395409e-06, "loss": 0.5823, "step": 16265 }, { "epoch": 0.5795108395532358, "grad_norm": 1.2963837866923604, "learning_rate": 7.925962613749304e-06, "loss": 0.4301, "step": 16266 }, { "epoch": 0.57954646667973, "grad_norm": 1.2418714875529195, "learning_rate": 7.924833782720457e-06, "loss": 0.4732, "step": 16267 }, { "epoch": 0.579582093806224, "grad_norm": 1.6742828719041454, "learning_rate": 7.923704979323899e-06, "loss": 0.5564, "step": 16268 }, { "epoch": 0.5796177209327181, "grad_norm": 1.5742902794629987, "learning_rate": 7.922576203574664e-06, "loss": 0.6425, "step": 16269 }, { "epoch": 0.5796533480592123, "grad_norm": 1.5623128839908058, "learning_rate": 7.921447455487775e-06, "loss": 0.5914, "step": 16270 }, { "epoch": 0.5796889751857064, "grad_norm": 1.2467523566837708, "learning_rate": 7.920318735078267e-06, "loss": 0.4893, "step": 16271 }, { "epoch": 0.5797246023122005, "grad_norm": 1.3221636027884356, "learning_rate": 7.919190042361166e-06, "loss": 0.4675, "step": 16272 }, { "epoch": 0.5797602294386947, "grad_norm": 2.9782515285968714, "learning_rate": 7.918061377351509e-06, "loss": 0.5409, "step": 16273 }, { "epoch": 0.5797958565651887, "grad_norm": 1.6326973452227187, "learning_rate": 7.916932740064315e-06, "loss": 0.6, "step": 16274 }, { "epoch": 0.5798314836916828, "grad_norm": 1.393635161927534, "learning_rate": 7.91580413051462e-06, "loss": 0.4671, "step": 16275 }, { "epoch": 0.579867110818177, "grad_norm": 2.214744177535134, "learning_rate": 7.91467554871745e-06, "loss": 0.5486, "step": 16276 }, { "epoch": 0.5799027379446711, "grad_norm": 1.1598220654058256, "learning_rate": 7.91354699468783e-06, "loss": 0.436, "step": 16277 }, { "epoch": 0.5799383650711651, "grad_norm": 1.101321122246683, "learning_rate": 7.912418468440794e-06, "loss": 0.444, "step": 16278 }, { "epoch": 0.5799739921976593, "grad_norm": 1.4404091014872158, "learning_rate": 7.91128996999136e-06, "loss": 0.4731, "step": 16279 }, { "epoch": 0.5800096193241534, "grad_norm": 1.6801412236411473, "learning_rate": 7.910161499354567e-06, "loss": 0.5364, "step": 16280 }, { "epoch": 0.5800452464506475, "grad_norm": 1.196623320802532, "learning_rate": 7.90903305654543e-06, "loss": 0.5094, "step": 16281 }, { "epoch": 0.5800808735771417, "grad_norm": 1.2943955551818946, "learning_rate": 7.90790464157898e-06, "loss": 0.4685, "step": 16282 }, { "epoch": 0.5801165007036357, "grad_norm": 1.3979503377236868, "learning_rate": 7.90677625447024e-06, "loss": 0.5872, "step": 16283 }, { "epoch": 0.5801521278301298, "grad_norm": 1.5370086849531004, "learning_rate": 7.905647895234243e-06, "loss": 0.4837, "step": 16284 }, { "epoch": 0.580187754956624, "grad_norm": 3.2830013503726647, "learning_rate": 7.904519563886003e-06, "loss": 0.4733, "step": 16285 }, { "epoch": 0.5802233820831181, "grad_norm": 1.8805314974528766, "learning_rate": 7.903391260440549e-06, "loss": 0.5555, "step": 16286 }, { "epoch": 0.5802590092096122, "grad_norm": 1.0687493065967615, "learning_rate": 7.90226298491291e-06, "loss": 0.3899, "step": 16287 }, { "epoch": 0.5802946363361063, "grad_norm": 1.0200004329121461, "learning_rate": 7.901134737318102e-06, "loss": 0.4102, "step": 16288 }, { "epoch": 0.5803302634626004, "grad_norm": 1.45849952661314, "learning_rate": 7.900006517671155e-06, "loss": 0.4408, "step": 16289 }, { "epoch": 0.5803658905890945, "grad_norm": 1.6294121784214837, "learning_rate": 7.898878325987086e-06, "loss": 0.5483, "step": 16290 }, { "epoch": 0.5804015177155887, "grad_norm": 1.5111576000955271, "learning_rate": 7.897750162280924e-06, "loss": 0.5094, "step": 16291 }, { "epoch": 0.5804371448420828, "grad_norm": 1.2250915718228232, "learning_rate": 7.896622026567685e-06, "loss": 0.4431, "step": 16292 }, { "epoch": 0.5804727719685768, "grad_norm": 1.1664507349897086, "learning_rate": 7.895493918862395e-06, "loss": 0.4199, "step": 16293 }, { "epoch": 0.580508399095071, "grad_norm": 1.0107748714511786, "learning_rate": 7.894365839180074e-06, "loss": 0.4201, "step": 16294 }, { "epoch": 0.5805440262215651, "grad_norm": 1.5294709384486844, "learning_rate": 7.893237787535746e-06, "loss": 0.5246, "step": 16295 }, { "epoch": 0.5805796533480592, "grad_norm": 1.1220489834801288, "learning_rate": 7.892109763944426e-06, "loss": 0.4284, "step": 16296 }, { "epoch": 0.5806152804745534, "grad_norm": 1.6615384544157152, "learning_rate": 7.890981768421138e-06, "loss": 0.4929, "step": 16297 }, { "epoch": 0.5806509076010474, "grad_norm": 1.4013402805065507, "learning_rate": 7.889853800980905e-06, "loss": 0.5526, "step": 16298 }, { "epoch": 0.5806865347275415, "grad_norm": 1.139832001047894, "learning_rate": 7.888725861638739e-06, "loss": 0.4378, "step": 16299 }, { "epoch": 0.5807221618540357, "grad_norm": 1.2425853268490776, "learning_rate": 7.88759795040967e-06, "loss": 0.4237, "step": 16300 }, { "epoch": 0.5807577889805298, "grad_norm": 1.729243559090593, "learning_rate": 7.886470067308705e-06, "loss": 0.5809, "step": 16301 }, { "epoch": 0.5807934161070238, "grad_norm": 2.0762752280101573, "learning_rate": 7.885342212350872e-06, "loss": 0.6215, "step": 16302 }, { "epoch": 0.580829043233518, "grad_norm": 1.7501392770339133, "learning_rate": 7.884214385551183e-06, "loss": 0.5796, "step": 16303 }, { "epoch": 0.5808646703600121, "grad_norm": 1.1186329768597991, "learning_rate": 7.88308658692466e-06, "loss": 0.41, "step": 16304 }, { "epoch": 0.5809002974865062, "grad_norm": 1.1885499259740968, "learning_rate": 7.881958816486317e-06, "loss": 0.4059, "step": 16305 }, { "epoch": 0.5809359246130004, "grad_norm": 2.3981610412768273, "learning_rate": 7.880831074251176e-06, "loss": 0.4362, "step": 16306 }, { "epoch": 0.5809715517394944, "grad_norm": 1.1893071881438109, "learning_rate": 7.879703360234247e-06, "loss": 0.4498, "step": 16307 }, { "epoch": 0.5810071788659885, "grad_norm": 1.1797826635613413, "learning_rate": 7.878575674450551e-06, "loss": 0.4015, "step": 16308 }, { "epoch": 0.5810428059924827, "grad_norm": 1.1342730958293084, "learning_rate": 7.877448016915104e-06, "loss": 0.444, "step": 16309 }, { "epoch": 0.5810784331189768, "grad_norm": 1.4785856841584704, "learning_rate": 7.876320387642916e-06, "loss": 0.5466, "step": 16310 }, { "epoch": 0.5811140602454709, "grad_norm": 1.155455962133417, "learning_rate": 7.875192786649012e-06, "loss": 0.4169, "step": 16311 }, { "epoch": 0.581149687371965, "grad_norm": 1.629723472434318, "learning_rate": 7.874065213948399e-06, "loss": 0.5427, "step": 16312 }, { "epoch": 0.5811853144984591, "grad_norm": 1.3883160111290844, "learning_rate": 7.872937669556094e-06, "loss": 0.541, "step": 16313 }, { "epoch": 0.5812209416249532, "grad_norm": 1.875974201166467, "learning_rate": 7.871810153487109e-06, "loss": 0.5681, "step": 16314 }, { "epoch": 0.5812565687514474, "grad_norm": 2.1662766002147817, "learning_rate": 7.870682665756461e-06, "loss": 0.5436, "step": 16315 }, { "epoch": 0.5812921958779415, "grad_norm": 1.2049511426615402, "learning_rate": 7.869555206379158e-06, "loss": 0.5022, "step": 16316 }, { "epoch": 0.5813278230044355, "grad_norm": 1.6469419806935046, "learning_rate": 7.868427775370222e-06, "loss": 0.5172, "step": 16317 }, { "epoch": 0.5813634501309297, "grad_norm": 1.68676515916755, "learning_rate": 7.867300372744658e-06, "loss": 0.5626, "step": 16318 }, { "epoch": 0.5813990772574238, "grad_norm": 1.6897755220290245, "learning_rate": 7.86617299851748e-06, "loss": 0.552, "step": 16319 }, { "epoch": 0.5814347043839179, "grad_norm": 1.4287781340154895, "learning_rate": 7.865045652703697e-06, "loss": 0.5732, "step": 16320 }, { "epoch": 0.581470331510412, "grad_norm": 1.5967018510358308, "learning_rate": 7.863918335318325e-06, "loss": 0.5421, "step": 16321 }, { "epoch": 0.5815059586369061, "grad_norm": 1.2738003080071993, "learning_rate": 7.862791046376376e-06, "loss": 0.4784, "step": 16322 }, { "epoch": 0.5815415857634002, "grad_norm": 1.1481075026649485, "learning_rate": 7.861663785892857e-06, "loss": 0.4247, "step": 16323 }, { "epoch": 0.5815772128898944, "grad_norm": 1.7063091650786335, "learning_rate": 7.860536553882778e-06, "loss": 0.5633, "step": 16324 }, { "epoch": 0.5816128400163885, "grad_norm": 2.5721160457940884, "learning_rate": 7.85940935036115e-06, "loss": 0.5523, "step": 16325 }, { "epoch": 0.5816484671428825, "grad_norm": 1.1769505562610496, "learning_rate": 7.858282175342987e-06, "loss": 0.4253, "step": 16326 }, { "epoch": 0.5816840942693767, "grad_norm": 1.3695842321511715, "learning_rate": 7.857155028843288e-06, "loss": 0.5471, "step": 16327 }, { "epoch": 0.5817197213958708, "grad_norm": 1.741428508958476, "learning_rate": 7.856027910877073e-06, "loss": 0.5476, "step": 16328 }, { "epoch": 0.5817553485223649, "grad_norm": 1.5523466996047677, "learning_rate": 7.854900821459339e-06, "loss": 0.4585, "step": 16329 }, { "epoch": 0.5817909756488591, "grad_norm": 1.2221770833557657, "learning_rate": 7.853773760605104e-06, "loss": 0.4717, "step": 16330 }, { "epoch": 0.5818266027753531, "grad_norm": 1.0858178965500016, "learning_rate": 7.852646728329368e-06, "loss": 0.4511, "step": 16331 }, { "epoch": 0.5818622299018472, "grad_norm": 1.2542041518787574, "learning_rate": 7.851519724647143e-06, "loss": 0.4663, "step": 16332 }, { "epoch": 0.5818978570283414, "grad_norm": 1.079111859371867, "learning_rate": 7.85039274957344e-06, "loss": 0.4324, "step": 16333 }, { "epoch": 0.5819334841548355, "grad_norm": 1.250650650627756, "learning_rate": 7.849265803123253e-06, "loss": 0.4876, "step": 16334 }, { "epoch": 0.5819691112813296, "grad_norm": 2.8808576796027583, "learning_rate": 7.848138885311599e-06, "loss": 0.5819, "step": 16335 }, { "epoch": 0.5820047384078237, "grad_norm": 1.1010564772780622, "learning_rate": 7.847011996153477e-06, "loss": 0.4081, "step": 16336 }, { "epoch": 0.5820403655343178, "grad_norm": 1.2079771418069616, "learning_rate": 7.845885135663898e-06, "loss": 0.4731, "step": 16337 }, { "epoch": 0.5820759926608119, "grad_norm": 1.2139774060333883, "learning_rate": 7.844758303857862e-06, "loss": 0.4478, "step": 16338 }, { "epoch": 0.5821116197873061, "grad_norm": 1.7648624627758414, "learning_rate": 7.843631500750376e-06, "loss": 0.5577, "step": 16339 }, { "epoch": 0.5821472469138002, "grad_norm": 1.031355296581283, "learning_rate": 7.842504726356444e-06, "loss": 0.4113, "step": 16340 }, { "epoch": 0.5821828740402942, "grad_norm": 1.0414018243928473, "learning_rate": 7.841377980691071e-06, "loss": 0.3885, "step": 16341 }, { "epoch": 0.5822185011667884, "grad_norm": 1.4351411756467218, "learning_rate": 7.840251263769256e-06, "loss": 0.4889, "step": 16342 }, { "epoch": 0.5822541282932825, "grad_norm": 1.57910578252291, "learning_rate": 7.839124575606005e-06, "loss": 0.4697, "step": 16343 }, { "epoch": 0.5822897554197766, "grad_norm": 1.5267008760247158, "learning_rate": 7.837997916216325e-06, "loss": 0.5361, "step": 16344 }, { "epoch": 0.5823253825462708, "grad_norm": 1.518614466619737, "learning_rate": 7.83687128561521e-06, "loss": 0.5648, "step": 16345 }, { "epoch": 0.5823610096727648, "grad_norm": 1.1802466085347307, "learning_rate": 7.835744683817668e-06, "loss": 0.4471, "step": 16346 }, { "epoch": 0.5823966367992589, "grad_norm": 1.7050304649241432, "learning_rate": 7.834618110838695e-06, "loss": 0.4815, "step": 16347 }, { "epoch": 0.5824322639257531, "grad_norm": 1.835403032672992, "learning_rate": 7.833491566693301e-06, "loss": 0.6011, "step": 16348 }, { "epoch": 0.5824678910522472, "grad_norm": 1.4165646864542125, "learning_rate": 7.832365051396476e-06, "loss": 0.4473, "step": 16349 }, { "epoch": 0.5825035181787412, "grad_norm": 1.39189806768828, "learning_rate": 7.831238564963227e-06, "loss": 0.545, "step": 16350 }, { "epoch": 0.5825391453052354, "grad_norm": 1.093493018882831, "learning_rate": 7.830112107408553e-06, "loss": 0.4083, "step": 16351 }, { "epoch": 0.5825747724317295, "grad_norm": 2.004554965859603, "learning_rate": 7.828985678747453e-06, "loss": 0.5611, "step": 16352 }, { "epoch": 0.5826103995582236, "grad_norm": 1.392006214095406, "learning_rate": 7.827859278994924e-06, "loss": 0.5827, "step": 16353 }, { "epoch": 0.5826460266847178, "grad_norm": 1.215523356113798, "learning_rate": 7.826732908165966e-06, "loss": 0.4084, "step": 16354 }, { "epoch": 0.5826816538112118, "grad_norm": 1.060264232604761, "learning_rate": 7.825606566275583e-06, "loss": 0.4087, "step": 16355 }, { "epoch": 0.5827172809377059, "grad_norm": 1.1915616683174801, "learning_rate": 7.824480253338765e-06, "loss": 0.461, "step": 16356 }, { "epoch": 0.5827529080642001, "grad_norm": 1.6242120796628416, "learning_rate": 7.823353969370516e-06, "loss": 0.4549, "step": 16357 }, { "epoch": 0.5827885351906942, "grad_norm": 2.517758528584239, "learning_rate": 7.822227714385826e-06, "loss": 0.5438, "step": 16358 }, { "epoch": 0.5828241623171883, "grad_norm": 1.7541972458088404, "learning_rate": 7.821101488399702e-06, "loss": 0.5226, "step": 16359 }, { "epoch": 0.5828597894436824, "grad_norm": 1.1714574424456232, "learning_rate": 7.819975291427128e-06, "loss": 0.4158, "step": 16360 }, { "epoch": 0.5828954165701765, "grad_norm": 1.244672308664691, "learning_rate": 7.818849123483112e-06, "loss": 0.4416, "step": 16361 }, { "epoch": 0.5829310436966706, "grad_norm": 1.5452092088665692, "learning_rate": 7.817722984582641e-06, "loss": 0.5244, "step": 16362 }, { "epoch": 0.5829666708231648, "grad_norm": 1.232327961503978, "learning_rate": 7.816596874740718e-06, "loss": 0.4295, "step": 16363 }, { "epoch": 0.5830022979496589, "grad_norm": 1.7205379563112073, "learning_rate": 7.815470793972331e-06, "loss": 0.5597, "step": 16364 }, { "epoch": 0.5830379250761529, "grad_norm": 1.0268627168441067, "learning_rate": 7.814344742292476e-06, "loss": 0.4018, "step": 16365 }, { "epoch": 0.5830735522026471, "grad_norm": 1.759494292526676, "learning_rate": 7.81321871971615e-06, "loss": 0.6551, "step": 16366 }, { "epoch": 0.5831091793291412, "grad_norm": 1.0789490787653357, "learning_rate": 7.812092726258342e-06, "loss": 0.4286, "step": 16367 }, { "epoch": 0.5831448064556353, "grad_norm": 1.1939609373366082, "learning_rate": 7.810966761934053e-06, "loss": 0.4137, "step": 16368 }, { "epoch": 0.5831804335821295, "grad_norm": 1.115446070536927, "learning_rate": 7.809840826758269e-06, "loss": 0.4396, "step": 16369 }, { "epoch": 0.5832160607086235, "grad_norm": 1.1327968847376215, "learning_rate": 7.80871492074599e-06, "loss": 0.4208, "step": 16370 }, { "epoch": 0.5832516878351176, "grad_norm": 1.4640236732321124, "learning_rate": 7.807589043912197e-06, "loss": 0.5615, "step": 16371 }, { "epoch": 0.5832873149616118, "grad_norm": 1.1792234615346828, "learning_rate": 7.806463196271892e-06, "loss": 0.4239, "step": 16372 }, { "epoch": 0.5833229420881059, "grad_norm": 1.0473675919416099, "learning_rate": 7.80533737784006e-06, "loss": 0.441, "step": 16373 }, { "epoch": 0.5833585692145999, "grad_norm": 2.0426876150346684, "learning_rate": 7.8042115886317e-06, "loss": 0.592, "step": 16374 }, { "epoch": 0.5833941963410941, "grad_norm": 1.1134884486326733, "learning_rate": 7.803085828661792e-06, "loss": 0.4416, "step": 16375 }, { "epoch": 0.5834298234675882, "grad_norm": 1.2320508574647029, "learning_rate": 7.801960097945337e-06, "loss": 0.4031, "step": 16376 }, { "epoch": 0.5834654505940823, "grad_norm": 1.6715192884506653, "learning_rate": 7.800834396497313e-06, "loss": 0.5814, "step": 16377 }, { "epoch": 0.5835010777205765, "grad_norm": 1.3437327141424242, "learning_rate": 7.799708724332718e-06, "loss": 0.5224, "step": 16378 }, { "epoch": 0.5835367048470705, "grad_norm": 1.1786617420161658, "learning_rate": 7.798583081466542e-06, "loss": 0.4659, "step": 16379 }, { "epoch": 0.5835723319735646, "grad_norm": 1.1132887856948799, "learning_rate": 7.797457467913768e-06, "loss": 0.4545, "step": 16380 }, { "epoch": 0.5836079591000588, "grad_norm": 1.901259275737969, "learning_rate": 7.796331883689392e-06, "loss": 0.4709, "step": 16381 }, { "epoch": 0.5836435862265529, "grad_norm": 1.5106557604901711, "learning_rate": 7.795206328808394e-06, "loss": 0.5811, "step": 16382 }, { "epoch": 0.583679213353047, "grad_norm": 1.5997026117374118, "learning_rate": 7.794080803285765e-06, "loss": 0.5968, "step": 16383 }, { "epoch": 0.5837148404795411, "grad_norm": 1.9559424024537955, "learning_rate": 7.792955307136491e-06, "loss": 0.5682, "step": 16384 }, { "epoch": 0.5837504676060352, "grad_norm": 1.2535666088676067, "learning_rate": 7.791829840375563e-06, "loss": 0.4198, "step": 16385 }, { "epoch": 0.5837860947325293, "grad_norm": 1.6844483752701096, "learning_rate": 7.790704403017962e-06, "loss": 0.5789, "step": 16386 }, { "epoch": 0.5838217218590235, "grad_norm": 1.0573493799069889, "learning_rate": 7.789578995078674e-06, "loss": 0.4333, "step": 16387 }, { "epoch": 0.5838573489855176, "grad_norm": 1.5720021772211692, "learning_rate": 7.788453616572689e-06, "loss": 0.5933, "step": 16388 }, { "epoch": 0.5838929761120117, "grad_norm": 2.546429608921117, "learning_rate": 7.787328267514988e-06, "loss": 0.487, "step": 16389 }, { "epoch": 0.5839286032385058, "grad_norm": 1.4557160559400657, "learning_rate": 7.786202947920562e-06, "loss": 0.614, "step": 16390 }, { "epoch": 0.5839642303649999, "grad_norm": 1.1210291784958486, "learning_rate": 7.785077657804386e-06, "loss": 0.462, "step": 16391 }, { "epoch": 0.5839998574914941, "grad_norm": 1.6003288167910252, "learning_rate": 7.783952397181454e-06, "loss": 0.5277, "step": 16392 }, { "epoch": 0.5840354846179882, "grad_norm": 1.492710312849437, "learning_rate": 7.78282716606674e-06, "loss": 0.548, "step": 16393 }, { "epoch": 0.5840711117444822, "grad_norm": 1.3092647146768313, "learning_rate": 7.781701964475234e-06, "loss": 0.4399, "step": 16394 }, { "epoch": 0.5841067388709764, "grad_norm": 1.295020574678335, "learning_rate": 7.780576792421917e-06, "loss": 0.4806, "step": 16395 }, { "epoch": 0.5841423659974705, "grad_norm": 1.3996459722273076, "learning_rate": 7.779451649921774e-06, "loss": 0.5344, "step": 16396 }, { "epoch": 0.5841779931239646, "grad_norm": 1.2591707309473699, "learning_rate": 7.77832653698978e-06, "loss": 0.4778, "step": 16397 }, { "epoch": 0.5842136202504588, "grad_norm": 1.4123687348644522, "learning_rate": 7.777201453640923e-06, "loss": 0.5389, "step": 16398 }, { "epoch": 0.5842492473769528, "grad_norm": 2.2924707227369385, "learning_rate": 7.77607639989018e-06, "loss": 0.5724, "step": 16399 }, { "epoch": 0.5842848745034469, "grad_norm": 1.4186553484866502, "learning_rate": 7.77495137575254e-06, "loss": 0.6168, "step": 16400 }, { "epoch": 0.5843205016299411, "grad_norm": 1.349036469603359, "learning_rate": 7.773826381242973e-06, "loss": 0.5421, "step": 16401 }, { "epoch": 0.5843561287564352, "grad_norm": 1.4245330861736143, "learning_rate": 7.772701416376462e-06, "loss": 0.5188, "step": 16402 }, { "epoch": 0.5843917558829292, "grad_norm": 1.4827001649011715, "learning_rate": 7.771576481167992e-06, "loss": 0.4356, "step": 16403 }, { "epoch": 0.5844273830094234, "grad_norm": 1.0610067841430686, "learning_rate": 7.770451575632536e-06, "loss": 0.4523, "step": 16404 }, { "epoch": 0.5844630101359175, "grad_norm": 1.4322646501197172, "learning_rate": 7.769326699785078e-06, "loss": 0.5656, "step": 16405 }, { "epoch": 0.5844986372624116, "grad_norm": 1.2235942004132518, "learning_rate": 7.768201853640593e-06, "loss": 0.4122, "step": 16406 }, { "epoch": 0.5845342643889058, "grad_norm": 1.1039458342716773, "learning_rate": 7.767077037214063e-06, "loss": 0.4364, "step": 16407 }, { "epoch": 0.5845698915153998, "grad_norm": 1.5957458076016395, "learning_rate": 7.765952250520459e-06, "loss": 0.5464, "step": 16408 }, { "epoch": 0.5846055186418939, "grad_norm": 1.4639467936169794, "learning_rate": 7.764827493574765e-06, "loss": 0.5595, "step": 16409 }, { "epoch": 0.5846411457683881, "grad_norm": 1.0912253397171872, "learning_rate": 7.763702766391952e-06, "loss": 0.4764, "step": 16410 }, { "epoch": 0.5846767728948822, "grad_norm": 1.4641481250583168, "learning_rate": 7.762578068987006e-06, "loss": 0.6244, "step": 16411 }, { "epoch": 0.5847124000213763, "grad_norm": 1.5141265258028178, "learning_rate": 7.76145340137489e-06, "loss": 0.5585, "step": 16412 }, { "epoch": 0.5847480271478704, "grad_norm": 0.9899416106365266, "learning_rate": 7.760328763570589e-06, "loss": 0.4124, "step": 16413 }, { "epoch": 0.5847836542743645, "grad_norm": 1.1166407744005118, "learning_rate": 7.759204155589076e-06, "loss": 0.404, "step": 16414 }, { "epoch": 0.5848192814008586, "grad_norm": 1.4013609342372553, "learning_rate": 7.758079577445325e-06, "loss": 0.5326, "step": 16415 }, { "epoch": 0.5848549085273528, "grad_norm": 1.411323994991922, "learning_rate": 7.756955029154316e-06, "loss": 0.5152, "step": 16416 }, { "epoch": 0.5848905356538469, "grad_norm": 1.1226979781893054, "learning_rate": 7.755830510731011e-06, "loss": 0.4478, "step": 16417 }, { "epoch": 0.5849261627803409, "grad_norm": 1.0902801135948639, "learning_rate": 7.7547060221904e-06, "loss": 0.441, "step": 16418 }, { "epoch": 0.5849617899068351, "grad_norm": 1.3290893601297011, "learning_rate": 7.753581563547442e-06, "loss": 0.5694, "step": 16419 }, { "epoch": 0.5849974170333292, "grad_norm": 1.1962254241577892, "learning_rate": 7.752457134817116e-06, "loss": 0.4438, "step": 16420 }, { "epoch": 0.5850330441598233, "grad_norm": 1.139638241255215, "learning_rate": 7.751332736014394e-06, "loss": 0.4465, "step": 16421 }, { "epoch": 0.5850686712863175, "grad_norm": 1.092695868548279, "learning_rate": 7.750208367154253e-06, "loss": 0.4385, "step": 16422 }, { "epoch": 0.5851042984128115, "grad_norm": 1.152365755294523, "learning_rate": 7.749084028251654e-06, "loss": 0.4427, "step": 16423 }, { "epoch": 0.5851399255393056, "grad_norm": 1.0811007420282441, "learning_rate": 7.747959719321575e-06, "loss": 0.4395, "step": 16424 }, { "epoch": 0.5851755526657998, "grad_norm": 1.0262463929215349, "learning_rate": 7.74683544037899e-06, "loss": 0.3949, "step": 16425 }, { "epoch": 0.5852111797922939, "grad_norm": 1.4289825229384403, "learning_rate": 7.745711191438863e-06, "loss": 0.6071, "step": 16426 }, { "epoch": 0.5852468069187879, "grad_norm": 2.4154312404278038, "learning_rate": 7.744586972516171e-06, "loss": 0.5917, "step": 16427 }, { "epoch": 0.5852824340452821, "grad_norm": 1.222904804542383, "learning_rate": 7.743462783625878e-06, "loss": 0.4501, "step": 16428 }, { "epoch": 0.5853180611717762, "grad_norm": 1.1894535094105272, "learning_rate": 7.742338624782955e-06, "loss": 0.449, "step": 16429 }, { "epoch": 0.5853536882982703, "grad_norm": 1.8073814305635025, "learning_rate": 7.741214496002368e-06, "loss": 0.5193, "step": 16430 }, { "epoch": 0.5853893154247645, "grad_norm": 1.4036753837737828, "learning_rate": 7.740090397299093e-06, "loss": 0.5999, "step": 16431 }, { "epoch": 0.5854249425512585, "grad_norm": 1.2151139391823862, "learning_rate": 7.738966328688091e-06, "loss": 0.4609, "step": 16432 }, { "epoch": 0.5854605696777526, "grad_norm": 1.9130672280647227, "learning_rate": 7.737842290184337e-06, "loss": 0.5216, "step": 16433 }, { "epoch": 0.5854961968042468, "grad_norm": 1.4741304134313358, "learning_rate": 7.73671828180279e-06, "loss": 0.5624, "step": 16434 }, { "epoch": 0.5855318239307409, "grad_norm": 1.066995567897105, "learning_rate": 7.735594303558422e-06, "loss": 0.4242, "step": 16435 }, { "epoch": 0.585567451057235, "grad_norm": 0.9750061838865, "learning_rate": 7.734470355466199e-06, "loss": 0.4458, "step": 16436 }, { "epoch": 0.5856030781837291, "grad_norm": 2.0477150953273484, "learning_rate": 7.733346437541085e-06, "loss": 0.5794, "step": 16437 }, { "epoch": 0.5856387053102232, "grad_norm": 1.105607480843737, "learning_rate": 7.732222549798051e-06, "loss": 0.4203, "step": 16438 }, { "epoch": 0.5856743324367173, "grad_norm": 1.8217221164757107, "learning_rate": 7.731098692252058e-06, "loss": 0.5445, "step": 16439 }, { "epoch": 0.5857099595632115, "grad_norm": 1.5738092049323884, "learning_rate": 7.729974864918071e-06, "loss": 0.4418, "step": 16440 }, { "epoch": 0.5857455866897056, "grad_norm": 1.1739135344198952, "learning_rate": 7.728851067811055e-06, "loss": 0.4306, "step": 16441 }, { "epoch": 0.5857812138161996, "grad_norm": 1.856376045433733, "learning_rate": 7.727727300945976e-06, "loss": 0.5119, "step": 16442 }, { "epoch": 0.5858168409426938, "grad_norm": 1.3262487156646579, "learning_rate": 7.726603564337791e-06, "loss": 0.5022, "step": 16443 }, { "epoch": 0.5858524680691879, "grad_norm": 1.710130426177954, "learning_rate": 7.725479858001475e-06, "loss": 0.4814, "step": 16444 }, { "epoch": 0.585888095195682, "grad_norm": 1.283958217769525, "learning_rate": 7.72435618195198e-06, "loss": 0.4797, "step": 16445 }, { "epoch": 0.5859237223221762, "grad_norm": 1.0740303350364513, "learning_rate": 7.723232536204277e-06, "loss": 0.4337, "step": 16446 }, { "epoch": 0.5859593494486702, "grad_norm": 1.3611113831633934, "learning_rate": 7.72210892077332e-06, "loss": 0.5118, "step": 16447 }, { "epoch": 0.5859949765751643, "grad_norm": 1.6575470460325403, "learning_rate": 7.720985335674077e-06, "loss": 0.5299, "step": 16448 }, { "epoch": 0.5860306037016585, "grad_norm": 1.4513775680015648, "learning_rate": 7.719861780921508e-06, "loss": 0.5182, "step": 16449 }, { "epoch": 0.5860662308281526, "grad_norm": 2.41733433709651, "learning_rate": 7.71873825653057e-06, "loss": 0.5894, "step": 16450 }, { "epoch": 0.5861018579546466, "grad_norm": 1.3749147553070367, "learning_rate": 7.71761476251623e-06, "loss": 0.5107, "step": 16451 }, { "epoch": 0.5861374850811408, "grad_norm": 1.0059029436227012, "learning_rate": 7.716491298893443e-06, "loss": 0.4371, "step": 16452 }, { "epoch": 0.5861731122076349, "grad_norm": 1.2029937032553601, "learning_rate": 7.715367865677173e-06, "loss": 0.4234, "step": 16453 }, { "epoch": 0.586208739334129, "grad_norm": 1.7837874983142112, "learning_rate": 7.714244462882372e-06, "loss": 0.5131, "step": 16454 }, { "epoch": 0.5862443664606232, "grad_norm": 1.6314878405421784, "learning_rate": 7.71312109052401e-06, "loss": 0.5742, "step": 16455 }, { "epoch": 0.5862799935871172, "grad_norm": 1.4590448008443722, "learning_rate": 7.711997748617033e-06, "loss": 0.5286, "step": 16456 }, { "epoch": 0.5863156207136113, "grad_norm": 1.1878354238846198, "learning_rate": 7.71087443717641e-06, "loss": 0.4237, "step": 16457 }, { "epoch": 0.5863512478401055, "grad_norm": 1.6467005747568584, "learning_rate": 7.709751156217088e-06, "loss": 0.5351, "step": 16458 }, { "epoch": 0.5863868749665996, "grad_norm": 1.3831202303806878, "learning_rate": 7.708627905754033e-06, "loss": 0.4623, "step": 16459 }, { "epoch": 0.5864225020930937, "grad_norm": 1.1152978237829505, "learning_rate": 7.707504685802202e-06, "loss": 0.4448, "step": 16460 }, { "epoch": 0.5864581292195878, "grad_norm": 1.1595427896118211, "learning_rate": 7.706381496376545e-06, "loss": 0.4549, "step": 16461 }, { "epoch": 0.5864937563460819, "grad_norm": 1.1573060877380033, "learning_rate": 7.705258337492023e-06, "loss": 0.448, "step": 16462 }, { "epoch": 0.586529383472576, "grad_norm": 1.5549017832798453, "learning_rate": 7.704135209163589e-06, "loss": 0.635, "step": 16463 }, { "epoch": 0.5865650105990702, "grad_norm": 1.1874395070854407, "learning_rate": 7.703012111406203e-06, "loss": 0.4694, "step": 16464 }, { "epoch": 0.5866006377255643, "grad_norm": 1.5693273940957897, "learning_rate": 7.701889044234811e-06, "loss": 0.5419, "step": 16465 }, { "epoch": 0.5866362648520583, "grad_norm": 1.0982354065140716, "learning_rate": 7.700766007664377e-06, "loss": 0.47, "step": 16466 }, { "epoch": 0.5866718919785525, "grad_norm": 1.1338858371337879, "learning_rate": 7.699643001709847e-06, "loss": 0.4184, "step": 16467 }, { "epoch": 0.5867075191050466, "grad_norm": 1.2163673724067965, "learning_rate": 7.69852002638618e-06, "loss": 0.4472, "step": 16468 }, { "epoch": 0.5867431462315407, "grad_norm": 1.733818287169609, "learning_rate": 7.697397081708327e-06, "loss": 0.576, "step": 16469 }, { "epoch": 0.5867787733580349, "grad_norm": 1.5011887141210316, "learning_rate": 7.69627416769124e-06, "loss": 0.5359, "step": 16470 }, { "epoch": 0.5868144004845289, "grad_norm": 1.523799378092603, "learning_rate": 7.695151284349878e-06, "loss": 0.566, "step": 16471 }, { "epoch": 0.586850027611023, "grad_norm": 1.122130252339564, "learning_rate": 7.694028431699182e-06, "loss": 0.3952, "step": 16472 }, { "epoch": 0.5868856547375172, "grad_norm": 2.001164315130921, "learning_rate": 7.692905609754113e-06, "loss": 0.5053, "step": 16473 }, { "epoch": 0.5869212818640113, "grad_norm": 1.575235118196519, "learning_rate": 7.691782818529615e-06, "loss": 0.5395, "step": 16474 }, { "epoch": 0.5869569089905053, "grad_norm": 1.7882425055156395, "learning_rate": 7.690660058040647e-06, "loss": 0.5186, "step": 16475 }, { "epoch": 0.5869925361169995, "grad_norm": 1.9432963087295037, "learning_rate": 7.68953732830215e-06, "loss": 0.6798, "step": 16476 }, { "epoch": 0.5870281632434936, "grad_norm": 1.0716855520711797, "learning_rate": 7.68841462932908e-06, "loss": 0.4633, "step": 16477 }, { "epoch": 0.5870637903699877, "grad_norm": 1.284199953744966, "learning_rate": 7.687291961136383e-06, "loss": 0.4178, "step": 16478 }, { "epoch": 0.5870994174964819, "grad_norm": 1.6212072544055398, "learning_rate": 7.686169323739014e-06, "loss": 0.6032, "step": 16479 }, { "epoch": 0.5871350446229759, "grad_norm": 1.0632741436630802, "learning_rate": 7.685046717151913e-06, "loss": 0.4417, "step": 16480 }, { "epoch": 0.58717067174947, "grad_norm": 1.6618788066641057, "learning_rate": 7.683924141390033e-06, "loss": 0.5841, "step": 16481 }, { "epoch": 0.5872062988759642, "grad_norm": 1.3220670294522927, "learning_rate": 7.682801596468328e-06, "loss": 0.4775, "step": 16482 }, { "epoch": 0.5872419260024583, "grad_norm": 1.2459741928902903, "learning_rate": 7.681679082401734e-06, "loss": 0.4514, "step": 16483 }, { "epoch": 0.5872775531289524, "grad_norm": 1.1435985012193728, "learning_rate": 7.680556599205205e-06, "loss": 0.3944, "step": 16484 }, { "epoch": 0.5873131802554465, "grad_norm": 1.803142340428658, "learning_rate": 7.679434146893685e-06, "loss": 0.5695, "step": 16485 }, { "epoch": 0.5873488073819406, "grad_norm": 1.1708052412082852, "learning_rate": 7.678311725482126e-06, "loss": 0.4404, "step": 16486 }, { "epoch": 0.5873844345084347, "grad_norm": 1.542275410105487, "learning_rate": 7.677189334985464e-06, "loss": 0.5268, "step": 16487 }, { "epoch": 0.5874200616349289, "grad_norm": 1.204838117542179, "learning_rate": 7.676066975418652e-06, "loss": 0.4239, "step": 16488 }, { "epoch": 0.587455688761423, "grad_norm": 1.703118008015207, "learning_rate": 7.674944646796629e-06, "loss": 0.572, "step": 16489 }, { "epoch": 0.587491315887917, "grad_norm": 1.4315407861684437, "learning_rate": 7.67382234913435e-06, "loss": 0.5616, "step": 16490 }, { "epoch": 0.5875269430144112, "grad_norm": 1.0897672727438437, "learning_rate": 7.672700082446747e-06, "loss": 0.424, "step": 16491 }, { "epoch": 0.5875625701409053, "grad_norm": 1.5053849101756518, "learning_rate": 7.67157784674877e-06, "loss": 0.5433, "step": 16492 }, { "epoch": 0.5875981972673994, "grad_norm": 1.4084484913934374, "learning_rate": 7.67045564205536e-06, "loss": 0.5271, "step": 16493 }, { "epoch": 0.5876338243938936, "grad_norm": 1.4468493834251874, "learning_rate": 7.669333468381462e-06, "loss": 0.5408, "step": 16494 }, { "epoch": 0.5876694515203876, "grad_norm": 1.578766949239289, "learning_rate": 7.66821132574202e-06, "loss": 0.461, "step": 16495 }, { "epoch": 0.5877050786468817, "grad_norm": 1.2065259194197393, "learning_rate": 7.667089214151972e-06, "loss": 0.4814, "step": 16496 }, { "epoch": 0.5877407057733759, "grad_norm": 2.1039419044571366, "learning_rate": 7.665967133626265e-06, "loss": 0.5505, "step": 16497 }, { "epoch": 0.58777633289987, "grad_norm": 1.1382018802753615, "learning_rate": 7.664845084179833e-06, "loss": 0.4486, "step": 16498 }, { "epoch": 0.587811960026364, "grad_norm": 1.5424429218520241, "learning_rate": 7.663723065827623e-06, "loss": 0.5575, "step": 16499 }, { "epoch": 0.5878475871528582, "grad_norm": 1.6479675762659456, "learning_rate": 7.662601078584572e-06, "loss": 0.4877, "step": 16500 }, { "epoch": 0.5878832142793523, "grad_norm": 1.5380907656552527, "learning_rate": 7.661479122465625e-06, "loss": 0.5679, "step": 16501 }, { "epoch": 0.5879188414058464, "grad_norm": 1.1247006088682228, "learning_rate": 7.660357197485714e-06, "loss": 0.4402, "step": 16502 }, { "epoch": 0.5879544685323406, "grad_norm": 3.4913408011901894, "learning_rate": 7.659235303659784e-06, "loss": 0.5571, "step": 16503 }, { "epoch": 0.5879900956588346, "grad_norm": 1.5825073734458968, "learning_rate": 7.65811344100277e-06, "loss": 0.6072, "step": 16504 }, { "epoch": 0.5880257227853287, "grad_norm": 1.124825288563622, "learning_rate": 7.656991609529613e-06, "loss": 0.4773, "step": 16505 }, { "epoch": 0.5880613499118229, "grad_norm": 2.0954640178082866, "learning_rate": 7.655869809255252e-06, "loss": 0.505, "step": 16506 }, { "epoch": 0.588096977038317, "grad_norm": 1.2225171321117305, "learning_rate": 7.654748040194621e-06, "loss": 0.4431, "step": 16507 }, { "epoch": 0.5881326041648111, "grad_norm": 1.5383975084388166, "learning_rate": 7.653626302362665e-06, "loss": 0.5957, "step": 16508 }, { "epoch": 0.5881682312913052, "grad_norm": 1.0668968162847108, "learning_rate": 7.652504595774309e-06, "loss": 0.408, "step": 16509 }, { "epoch": 0.5882038584177993, "grad_norm": 1.438699987913461, "learning_rate": 7.651382920444495e-06, "loss": 0.5172, "step": 16510 }, { "epoch": 0.5882394855442934, "grad_norm": 1.3472373906942179, "learning_rate": 7.65026127638816e-06, "loss": 0.5039, "step": 16511 }, { "epoch": 0.5882751126707876, "grad_norm": 1.2606183799863595, "learning_rate": 7.64913966362024e-06, "loss": 0.4638, "step": 16512 }, { "epoch": 0.5883107397972817, "grad_norm": 1.5552870492788005, "learning_rate": 7.648018082155666e-06, "loss": 0.5135, "step": 16513 }, { "epoch": 0.5883463669237757, "grad_norm": 1.0722111988671357, "learning_rate": 7.646896532009378e-06, "loss": 0.4459, "step": 16514 }, { "epoch": 0.5883819940502699, "grad_norm": 1.1557191692099842, "learning_rate": 7.645775013196305e-06, "loss": 0.4397, "step": 16515 }, { "epoch": 0.588417621176764, "grad_norm": 1.6171961648008224, "learning_rate": 7.644653525731383e-06, "loss": 0.5409, "step": 16516 }, { "epoch": 0.5884532483032581, "grad_norm": 1.6327521827028781, "learning_rate": 7.643532069629549e-06, "loss": 0.5247, "step": 16517 }, { "epoch": 0.5884888754297523, "grad_norm": 1.162024852898801, "learning_rate": 7.642410644905726e-06, "loss": 0.4438, "step": 16518 }, { "epoch": 0.5885245025562463, "grad_norm": 1.4522901932697307, "learning_rate": 7.641289251574861e-06, "loss": 0.5372, "step": 16519 }, { "epoch": 0.5885601296827404, "grad_norm": 1.611887986115645, "learning_rate": 7.640167889651872e-06, "loss": 0.5541, "step": 16520 }, { "epoch": 0.5885957568092346, "grad_norm": 1.4255740227425246, "learning_rate": 7.639046559151702e-06, "loss": 0.5088, "step": 16521 }, { "epoch": 0.5886313839357287, "grad_norm": 1.1736750674364471, "learning_rate": 7.637925260089273e-06, "loss": 0.4254, "step": 16522 }, { "epoch": 0.5886670110622227, "grad_norm": 1.004669153346814, "learning_rate": 7.636803992479525e-06, "loss": 0.4126, "step": 16523 }, { "epoch": 0.5887026381887169, "grad_norm": 1.0604112968438861, "learning_rate": 7.635682756337378e-06, "loss": 0.4002, "step": 16524 }, { "epoch": 0.588738265315211, "grad_norm": 1.136592854267942, "learning_rate": 7.634561551677773e-06, "loss": 0.4476, "step": 16525 }, { "epoch": 0.5887738924417051, "grad_norm": 1.2786486980108895, "learning_rate": 7.633440378515629e-06, "loss": 0.4208, "step": 16526 }, { "epoch": 0.5888095195681993, "grad_norm": 1.1238289515707127, "learning_rate": 7.632319236865886e-06, "loss": 0.4367, "step": 16527 }, { "epoch": 0.5888451466946933, "grad_norm": 1.652938808451173, "learning_rate": 7.631198126743463e-06, "loss": 0.5971, "step": 16528 }, { "epoch": 0.5888807738211874, "grad_norm": 1.6185328584952872, "learning_rate": 7.630077048163292e-06, "loss": 0.6069, "step": 16529 }, { "epoch": 0.5889164009476816, "grad_norm": 1.8835556049803555, "learning_rate": 7.628956001140305e-06, "loss": 0.5827, "step": 16530 }, { "epoch": 0.5889520280741757, "grad_norm": 1.60072982207725, "learning_rate": 7.627834985689423e-06, "loss": 0.5547, "step": 16531 }, { "epoch": 0.5889876552006698, "grad_norm": 1.0844178918333867, "learning_rate": 7.626714001825579e-06, "loss": 0.4321, "step": 16532 }, { "epoch": 0.5890232823271639, "grad_norm": 2.2425715541872293, "learning_rate": 7.6255930495636945e-06, "loss": 0.5727, "step": 16533 }, { "epoch": 0.589058909453658, "grad_norm": 1.0482232085933885, "learning_rate": 7.624472128918702e-06, "loss": 0.4126, "step": 16534 }, { "epoch": 0.5890945365801521, "grad_norm": 1.5713955667765764, "learning_rate": 7.62335123990552e-06, "loss": 0.5104, "step": 16535 }, { "epoch": 0.5891301637066463, "grad_norm": 2.391326722871946, "learning_rate": 7.622230382539079e-06, "loss": 0.5017, "step": 16536 }, { "epoch": 0.5891657908331404, "grad_norm": 1.3852548260184692, "learning_rate": 7.6211095568343e-06, "loss": 0.5622, "step": 16537 }, { "epoch": 0.5892014179596344, "grad_norm": 1.3349255455570173, "learning_rate": 7.619988762806114e-06, "loss": 0.5675, "step": 16538 }, { "epoch": 0.5892370450861286, "grad_norm": 1.1925302585128195, "learning_rate": 7.618868000469438e-06, "loss": 0.4558, "step": 16539 }, { "epoch": 0.5892726722126227, "grad_norm": 1.3393250693718555, "learning_rate": 7.617747269839199e-06, "loss": 0.4816, "step": 16540 }, { "epoch": 0.5893082993391168, "grad_norm": 1.077493808447588, "learning_rate": 7.616626570930321e-06, "loss": 0.4381, "step": 16541 }, { "epoch": 0.589343926465611, "grad_norm": 1.1401686872247376, "learning_rate": 7.615505903757725e-06, "loss": 0.4546, "step": 16542 }, { "epoch": 0.589379553592105, "grad_norm": 1.2032401291574693, "learning_rate": 7.614385268336336e-06, "loss": 0.4895, "step": 16543 }, { "epoch": 0.5894151807185991, "grad_norm": 1.7525666904977593, "learning_rate": 7.613264664681072e-06, "loss": 0.556, "step": 16544 }, { "epoch": 0.5894508078450933, "grad_norm": 1.1484656894490168, "learning_rate": 7.612144092806864e-06, "loss": 0.4458, "step": 16545 }, { "epoch": 0.5894864349715874, "grad_norm": 1.5931638872174902, "learning_rate": 7.611023552728621e-06, "loss": 0.5176, "step": 16546 }, { "epoch": 0.5895220620980814, "grad_norm": 2.1854898419846895, "learning_rate": 7.60990304446127e-06, "loss": 0.5324, "step": 16547 }, { "epoch": 0.5895576892245756, "grad_norm": 1.1384622401455962, "learning_rate": 7.608782568019729e-06, "loss": 0.4858, "step": 16548 }, { "epoch": 0.5895933163510697, "grad_norm": 1.1428954808610783, "learning_rate": 7.607662123418924e-06, "loss": 0.4405, "step": 16549 }, { "epoch": 0.5896289434775638, "grad_norm": 1.2849050114744616, "learning_rate": 7.6065417106737646e-06, "loss": 0.5224, "step": 16550 }, { "epoch": 0.589664570604058, "grad_norm": 1.048137994032173, "learning_rate": 7.605421329799176e-06, "loss": 0.4672, "step": 16551 }, { "epoch": 0.589700197730552, "grad_norm": 1.625063491625267, "learning_rate": 7.60430098081008e-06, "loss": 0.6031, "step": 16552 }, { "epoch": 0.5897358248570461, "grad_norm": 1.2088369725680639, "learning_rate": 7.603180663721387e-06, "loss": 0.4213, "step": 16553 }, { "epoch": 0.5897714519835403, "grad_norm": 1.3981545603577432, "learning_rate": 7.602060378548023e-06, "loss": 0.5596, "step": 16554 }, { "epoch": 0.5898070791100344, "grad_norm": 1.5974415640826876, "learning_rate": 7.6009401253048956e-06, "loss": 0.4732, "step": 16555 }, { "epoch": 0.5898427062365285, "grad_norm": 1.1321784632573793, "learning_rate": 7.599819904006933e-06, "loss": 0.3917, "step": 16556 }, { "epoch": 0.5898783333630226, "grad_norm": 1.4749148452843124, "learning_rate": 7.598699714669042e-06, "loss": 0.5596, "step": 16557 }, { "epoch": 0.5899139604895167, "grad_norm": 1.3389092593525802, "learning_rate": 7.597579557306144e-06, "loss": 0.5071, "step": 16558 }, { "epoch": 0.5899495876160108, "grad_norm": 1.0621075483430393, "learning_rate": 7.5964594319331516e-06, "loss": 0.4175, "step": 16559 }, { "epoch": 0.589985214742505, "grad_norm": 1.095706270315975, "learning_rate": 7.595339338564986e-06, "loss": 0.4448, "step": 16560 }, { "epoch": 0.5900208418689991, "grad_norm": 1.1397109815981268, "learning_rate": 7.5942192772165546e-06, "loss": 0.4602, "step": 16561 }, { "epoch": 0.5900564689954931, "grad_norm": 1.3665910198930173, "learning_rate": 7.593099247902775e-06, "loss": 0.477, "step": 16562 }, { "epoch": 0.5900920961219873, "grad_norm": 1.7087875689897827, "learning_rate": 7.591979250638562e-06, "loss": 0.5634, "step": 16563 }, { "epoch": 0.5901277232484814, "grad_norm": 1.3341871130815337, "learning_rate": 7.5908592854388275e-06, "loss": 0.4484, "step": 16564 }, { "epoch": 0.5901633503749755, "grad_norm": 1.1140470222626528, "learning_rate": 7.58973935231849e-06, "loss": 0.443, "step": 16565 }, { "epoch": 0.5901989775014697, "grad_norm": 1.2690855672505335, "learning_rate": 7.5886194512924525e-06, "loss": 0.4894, "step": 16566 }, { "epoch": 0.5902346046279637, "grad_norm": 1.1157248540963178, "learning_rate": 7.587499582375635e-06, "loss": 0.4375, "step": 16567 }, { "epoch": 0.5902702317544578, "grad_norm": 1.1665368100541567, "learning_rate": 7.586379745582945e-06, "loss": 0.4272, "step": 16568 }, { "epoch": 0.590305858880952, "grad_norm": 1.5747697865597963, "learning_rate": 7.585259940929297e-06, "loss": 0.5661, "step": 16569 }, { "epoch": 0.5903414860074461, "grad_norm": 1.6499900562948506, "learning_rate": 7.584140168429599e-06, "loss": 0.5409, "step": 16570 }, { "epoch": 0.5903771131339401, "grad_norm": 1.2580606343111402, "learning_rate": 7.5830204280987675e-06, "loss": 0.5231, "step": 16571 }, { "epoch": 0.5904127402604343, "grad_norm": 1.1867097487192089, "learning_rate": 7.581900719951705e-06, "loss": 0.4553, "step": 16572 }, { "epoch": 0.5904483673869284, "grad_norm": 1.0905864190837125, "learning_rate": 7.580781044003324e-06, "loss": 0.4294, "step": 16573 }, { "epoch": 0.5904839945134225, "grad_norm": 2.1676434271695526, "learning_rate": 7.579661400268535e-06, "loss": 0.4708, "step": 16574 }, { "epoch": 0.5905196216399167, "grad_norm": 1.1480196584526519, "learning_rate": 7.578541788762245e-06, "loss": 0.4904, "step": 16575 }, { "epoch": 0.5905552487664107, "grad_norm": 1.4546231175860167, "learning_rate": 7.5774222094993675e-06, "loss": 0.5527, "step": 16576 }, { "epoch": 0.5905908758929048, "grad_norm": 2.063083957596686, "learning_rate": 7.576302662494803e-06, "loss": 0.4908, "step": 16577 }, { "epoch": 0.590626503019399, "grad_norm": 2.0031927006401973, "learning_rate": 7.575183147763465e-06, "loss": 0.5502, "step": 16578 }, { "epoch": 0.5906621301458931, "grad_norm": 1.1145383648064, "learning_rate": 7.574063665320256e-06, "loss": 0.4487, "step": 16579 }, { "epoch": 0.5906977572723872, "grad_norm": 1.6193086500715133, "learning_rate": 7.572944215180089e-06, "loss": 0.5574, "step": 16580 }, { "epoch": 0.5907333843988813, "grad_norm": 1.195379211043361, "learning_rate": 7.571824797357859e-06, "loss": 0.4413, "step": 16581 }, { "epoch": 0.5907690115253754, "grad_norm": 1.0896701027645384, "learning_rate": 7.570705411868486e-06, "loss": 0.46, "step": 16582 }, { "epoch": 0.5908046386518695, "grad_norm": 1.463873789985286, "learning_rate": 7.569586058726865e-06, "loss": 0.5389, "step": 16583 }, { "epoch": 0.5908402657783637, "grad_norm": 1.7192531184509767, "learning_rate": 7.568466737947905e-06, "loss": 0.5265, "step": 16584 }, { "epoch": 0.5908758929048578, "grad_norm": 1.1215934693883354, "learning_rate": 7.567347449546508e-06, "loss": 0.4695, "step": 16585 }, { "epoch": 0.5909115200313518, "grad_norm": 1.1106132427762978, "learning_rate": 7.566228193537581e-06, "loss": 0.4704, "step": 16586 }, { "epoch": 0.590947147157846, "grad_norm": 1.1898418243284075, "learning_rate": 7.56510896993603e-06, "loss": 0.4433, "step": 16587 }, { "epoch": 0.5909827742843401, "grad_norm": 1.1803648263404722, "learning_rate": 7.563989778756752e-06, "loss": 0.3995, "step": 16588 }, { "epoch": 0.5910184014108342, "grad_norm": 2.135049991031846, "learning_rate": 7.562870620014654e-06, "loss": 0.5674, "step": 16589 }, { "epoch": 0.5910540285373284, "grad_norm": 1.3997976862578452, "learning_rate": 7.561751493724635e-06, "loss": 0.5162, "step": 16590 }, { "epoch": 0.5910896556638224, "grad_norm": 1.6835318549931642, "learning_rate": 7.560632399901604e-06, "loss": 0.4755, "step": 16591 }, { "epoch": 0.5911252827903165, "grad_norm": 1.1586243687002078, "learning_rate": 7.559513338560452e-06, "loss": 0.452, "step": 16592 }, { "epoch": 0.5911609099168107, "grad_norm": 1.7437619411443994, "learning_rate": 7.558394309716088e-06, "loss": 0.5453, "step": 16593 }, { "epoch": 0.5911965370433048, "grad_norm": 1.5854111000994482, "learning_rate": 7.55727531338341e-06, "loss": 0.52, "step": 16594 }, { "epoch": 0.5912321641697988, "grad_norm": 1.5681765411499862, "learning_rate": 7.5561563495773196e-06, "loss": 0.563, "step": 16595 }, { "epoch": 0.591267791296293, "grad_norm": 1.0820804413735035, "learning_rate": 7.555037418312713e-06, "loss": 0.4396, "step": 16596 }, { "epoch": 0.5913034184227871, "grad_norm": 1.1668067310349874, "learning_rate": 7.553918519604493e-06, "loss": 0.408, "step": 16597 }, { "epoch": 0.5913390455492812, "grad_norm": 1.121076986229615, "learning_rate": 7.552799653467561e-06, "loss": 0.432, "step": 16598 }, { "epoch": 0.5913746726757754, "grad_norm": 1.0694171155118932, "learning_rate": 7.551680819916808e-06, "loss": 0.4497, "step": 16599 }, { "epoch": 0.5914102998022694, "grad_norm": 1.2518216659963939, "learning_rate": 7.550562018967138e-06, "loss": 0.4323, "step": 16600 }, { "epoch": 0.5914459269287635, "grad_norm": 1.2204327120288894, "learning_rate": 7.549443250633446e-06, "loss": 0.4806, "step": 16601 }, { "epoch": 0.5914815540552577, "grad_norm": 1.1692608728198477, "learning_rate": 7.5483245149306326e-06, "loss": 0.4582, "step": 16602 }, { "epoch": 0.5915171811817518, "grad_norm": 1.7042565221562973, "learning_rate": 7.5472058118735895e-06, "loss": 0.4842, "step": 16603 }, { "epoch": 0.5915528083082459, "grad_norm": 1.161402360199259, "learning_rate": 7.546087141477216e-06, "loss": 0.4487, "step": 16604 }, { "epoch": 0.59158843543474, "grad_norm": 1.1082741015269806, "learning_rate": 7.544968503756406e-06, "loss": 0.4594, "step": 16605 }, { "epoch": 0.5916240625612341, "grad_norm": 1.2879748108335607, "learning_rate": 7.54384989872606e-06, "loss": 0.4696, "step": 16606 }, { "epoch": 0.5916596896877282, "grad_norm": 1.3712021342934149, "learning_rate": 7.542731326401063e-06, "loss": 0.519, "step": 16607 }, { "epoch": 0.5916953168142224, "grad_norm": 1.5717140031405812, "learning_rate": 7.541612786796318e-06, "loss": 0.5753, "step": 16608 }, { "epoch": 0.5917309439407165, "grad_norm": 1.0942101292712767, "learning_rate": 7.540494279926722e-06, "loss": 0.4307, "step": 16609 }, { "epoch": 0.5917665710672105, "grad_norm": 1.796525404456373, "learning_rate": 7.5393758058071584e-06, "loss": 0.5624, "step": 16610 }, { "epoch": 0.5918021981937047, "grad_norm": 1.8571844499262626, "learning_rate": 7.538257364452529e-06, "loss": 0.5574, "step": 16611 }, { "epoch": 0.5918378253201988, "grad_norm": 1.3891138446693214, "learning_rate": 7.537138955877721e-06, "loss": 0.5478, "step": 16612 }, { "epoch": 0.5918734524466929, "grad_norm": 1.775587317425802, "learning_rate": 7.536020580097634e-06, "loss": 0.5261, "step": 16613 }, { "epoch": 0.5919090795731871, "grad_norm": 1.6787344001321873, "learning_rate": 7.53490223712715e-06, "loss": 0.4691, "step": 16614 }, { "epoch": 0.5919447066996811, "grad_norm": 1.8757425273387263, "learning_rate": 7.533783926981168e-06, "loss": 0.4836, "step": 16615 }, { "epoch": 0.5919803338261752, "grad_norm": 1.0692137638592567, "learning_rate": 7.5326656496745755e-06, "loss": 0.3947, "step": 16616 }, { "epoch": 0.5920159609526694, "grad_norm": 1.74342085806713, "learning_rate": 7.531547405222268e-06, "loss": 0.5671, "step": 16617 }, { "epoch": 0.5920515880791635, "grad_norm": 1.4919112036168924, "learning_rate": 7.530429193639128e-06, "loss": 0.4388, "step": 16618 }, { "epoch": 0.5920872152056575, "grad_norm": 1.2083406826996068, "learning_rate": 7.52931101494005e-06, "loss": 0.4213, "step": 16619 }, { "epoch": 0.5921228423321517, "grad_norm": 1.1412302076558944, "learning_rate": 7.528192869139923e-06, "loss": 0.4496, "step": 16620 }, { "epoch": 0.5921584694586458, "grad_norm": 1.2617539429691103, "learning_rate": 7.527074756253635e-06, "loss": 0.447, "step": 16621 }, { "epoch": 0.5921940965851399, "grad_norm": 1.5020281253548247, "learning_rate": 7.5259566762960765e-06, "loss": 0.6791, "step": 16622 }, { "epoch": 0.5922297237116341, "grad_norm": 1.6416477006400199, "learning_rate": 7.524838629282133e-06, "loss": 0.531, "step": 16623 }, { "epoch": 0.5922653508381281, "grad_norm": 1.38687282689606, "learning_rate": 7.523720615226697e-06, "loss": 0.575, "step": 16624 }, { "epoch": 0.5923009779646222, "grad_norm": 1.5183688582031303, "learning_rate": 7.522602634144648e-06, "loss": 0.4175, "step": 16625 }, { "epoch": 0.5923366050911164, "grad_norm": 1.2172863453068026, "learning_rate": 7.5214846860508775e-06, "loss": 0.4525, "step": 16626 }, { "epoch": 0.5923722322176105, "grad_norm": 1.7505434668722575, "learning_rate": 7.52036677096027e-06, "loss": 0.5827, "step": 16627 }, { "epoch": 0.5924078593441046, "grad_norm": 1.1827370374279116, "learning_rate": 7.519248888887715e-06, "loss": 0.422, "step": 16628 }, { "epoch": 0.5924434864705987, "grad_norm": 1.0996640461574259, "learning_rate": 7.5181310398480914e-06, "loss": 0.4136, "step": 16629 }, { "epoch": 0.5924791135970928, "grad_norm": 1.5710195914485423, "learning_rate": 7.51701322385629e-06, "loss": 0.6054, "step": 16630 }, { "epoch": 0.5925147407235869, "grad_norm": 1.796858287440396, "learning_rate": 7.51589544092719e-06, "loss": 0.5603, "step": 16631 }, { "epoch": 0.5925503678500811, "grad_norm": 1.040710101929262, "learning_rate": 7.51477769107568e-06, "loss": 0.4074, "step": 16632 }, { "epoch": 0.5925859949765752, "grad_norm": 1.1236943752540853, "learning_rate": 7.5136599743166425e-06, "loss": 0.4331, "step": 16633 }, { "epoch": 0.5926216221030692, "grad_norm": 1.6227329006673132, "learning_rate": 7.51254229066496e-06, "loss": 0.5761, "step": 16634 }, { "epoch": 0.5926572492295634, "grad_norm": 1.5311733236904248, "learning_rate": 7.511424640135519e-06, "loss": 0.5973, "step": 16635 }, { "epoch": 0.5926928763560575, "grad_norm": 1.290157690615044, "learning_rate": 7.510307022743194e-06, "loss": 0.4526, "step": 16636 }, { "epoch": 0.5927285034825516, "grad_norm": 1.275500569759729, "learning_rate": 7.509189438502873e-06, "loss": 0.4813, "step": 16637 }, { "epoch": 0.5927641306090458, "grad_norm": 1.2462395898678675, "learning_rate": 7.508071887429433e-06, "loss": 0.4608, "step": 16638 }, { "epoch": 0.5927997577355398, "grad_norm": 1.390001634506769, "learning_rate": 7.506954369537764e-06, "loss": 0.5077, "step": 16639 }, { "epoch": 0.5928353848620339, "grad_norm": 1.0631431343492919, "learning_rate": 7.505836884842734e-06, "loss": 0.4475, "step": 16640 }, { "epoch": 0.5928710119885281, "grad_norm": 1.143625640079249, "learning_rate": 7.504719433359232e-06, "loss": 0.4286, "step": 16641 }, { "epoch": 0.5929066391150222, "grad_norm": 1.4065968110331921, "learning_rate": 7.5036020151021325e-06, "loss": 0.4912, "step": 16642 }, { "epoch": 0.5929422662415162, "grad_norm": 1.2283697520421124, "learning_rate": 7.502484630086318e-06, "loss": 0.4271, "step": 16643 }, { "epoch": 0.5929778933680104, "grad_norm": 1.5780647555842988, "learning_rate": 7.501367278326667e-06, "loss": 0.5801, "step": 16644 }, { "epoch": 0.5930135204945045, "grad_norm": 1.5420429950205996, "learning_rate": 7.5002499598380555e-06, "loss": 0.5746, "step": 16645 }, { "epoch": 0.5930491476209986, "grad_norm": 1.5340517745620033, "learning_rate": 7.499132674635367e-06, "loss": 0.583, "step": 16646 }, { "epoch": 0.5930847747474928, "grad_norm": 1.1949990162778086, "learning_rate": 7.4980154227334715e-06, "loss": 0.4757, "step": 16647 }, { "epoch": 0.5931204018739868, "grad_norm": 1.1382277387524367, "learning_rate": 7.496898204147249e-06, "loss": 0.4222, "step": 16648 }, { "epoch": 0.5931560290004809, "grad_norm": 1.134347786005808, "learning_rate": 7.4957810188915766e-06, "loss": 0.4686, "step": 16649 }, { "epoch": 0.5931916561269751, "grad_norm": 1.1097966971374564, "learning_rate": 7.4946638669813355e-06, "loss": 0.4515, "step": 16650 }, { "epoch": 0.5932272832534692, "grad_norm": 1.1364880818097707, "learning_rate": 7.493546748431391e-06, "loss": 0.4511, "step": 16651 }, { "epoch": 0.5932629103799633, "grad_norm": 1.619588306757849, "learning_rate": 7.492429663256624e-06, "loss": 0.589, "step": 16652 }, { "epoch": 0.5932985375064574, "grad_norm": 1.610108283675257, "learning_rate": 7.491312611471908e-06, "loss": 0.5096, "step": 16653 }, { "epoch": 0.5933341646329515, "grad_norm": 1.859032681809898, "learning_rate": 7.490195593092122e-06, "loss": 0.6121, "step": 16654 }, { "epoch": 0.5933697917594456, "grad_norm": 1.0581193185587319, "learning_rate": 7.489078608132132e-06, "loss": 0.3953, "step": 16655 }, { "epoch": 0.5934054188859398, "grad_norm": 1.1049069909004314, "learning_rate": 7.487961656606813e-06, "loss": 0.4274, "step": 16656 }, { "epoch": 0.5934410460124339, "grad_norm": 1.294336176264963, "learning_rate": 7.486844738531047e-06, "loss": 0.4213, "step": 16657 }, { "epoch": 0.5934766731389279, "grad_norm": 1.529698787253689, "learning_rate": 7.485727853919695e-06, "loss": 0.5134, "step": 16658 }, { "epoch": 0.5935123002654221, "grad_norm": 1.0934942432120032, "learning_rate": 7.484611002787636e-06, "loss": 0.4309, "step": 16659 }, { "epoch": 0.5935479273919162, "grad_norm": 1.7081221212905389, "learning_rate": 7.483494185149738e-06, "loss": 0.5094, "step": 16660 }, { "epoch": 0.5935835545184103, "grad_norm": 1.077470941373799, "learning_rate": 7.48237740102088e-06, "loss": 0.483, "step": 16661 }, { "epoch": 0.5936191816449045, "grad_norm": 1.3973009489422903, "learning_rate": 7.4812606504159195e-06, "loss": 0.5543, "step": 16662 }, { "epoch": 0.5936548087713985, "grad_norm": 1.2065986536039655, "learning_rate": 7.480143933349737e-06, "loss": 0.4001, "step": 16663 }, { "epoch": 0.5936904358978926, "grad_norm": 1.2562080841279946, "learning_rate": 7.479027249837196e-06, "loss": 0.4842, "step": 16664 }, { "epoch": 0.5937260630243868, "grad_norm": 1.0863362473634957, "learning_rate": 7.477910599893176e-06, "loss": 0.4642, "step": 16665 }, { "epoch": 0.5937616901508809, "grad_norm": 1.4333022732898741, "learning_rate": 7.476793983532533e-06, "loss": 0.5392, "step": 16666 }, { "epoch": 0.593797317277375, "grad_norm": 1.2054643658781246, "learning_rate": 7.475677400770142e-06, "loss": 0.4754, "step": 16667 }, { "epoch": 0.5938329444038691, "grad_norm": 1.6362633968984523, "learning_rate": 7.474560851620873e-06, "loss": 0.4376, "step": 16668 }, { "epoch": 0.5938685715303632, "grad_norm": 1.1171840639835668, "learning_rate": 7.473444336099591e-06, "loss": 0.4147, "step": 16669 }, { "epoch": 0.5939041986568573, "grad_norm": 1.0839945981259203, "learning_rate": 7.472327854221164e-06, "loss": 0.403, "step": 16670 }, { "epoch": 0.5939398257833515, "grad_norm": 1.1623383263272546, "learning_rate": 7.471211406000457e-06, "loss": 0.434, "step": 16671 }, { "epoch": 0.5939754529098455, "grad_norm": 1.1927169882837827, "learning_rate": 7.470094991452341e-06, "loss": 0.4387, "step": 16672 }, { "epoch": 0.5940110800363396, "grad_norm": 0.9912044326500424, "learning_rate": 7.468978610591675e-06, "loss": 0.3996, "step": 16673 }, { "epoch": 0.5940467071628338, "grad_norm": 1.0777845137256168, "learning_rate": 7.4678622634333295e-06, "loss": 0.3874, "step": 16674 }, { "epoch": 0.5940823342893279, "grad_norm": 1.1521974748570285, "learning_rate": 7.466745949992167e-06, "loss": 0.3891, "step": 16675 }, { "epoch": 0.594117961415822, "grad_norm": 1.0251340164994844, "learning_rate": 7.465629670283055e-06, "loss": 0.4154, "step": 16676 }, { "epoch": 0.5941535885423161, "grad_norm": 1.303603876150643, "learning_rate": 7.464513424320852e-06, "loss": 0.4361, "step": 16677 }, { "epoch": 0.5941892156688102, "grad_norm": 1.837891958928727, "learning_rate": 7.463397212120426e-06, "loss": 0.5503, "step": 16678 }, { "epoch": 0.5942248427953043, "grad_norm": 1.081597920695067, "learning_rate": 7.462281033696639e-06, "loss": 0.4552, "step": 16679 }, { "epoch": 0.5942604699217985, "grad_norm": 1.0297375726238212, "learning_rate": 7.461164889064354e-06, "loss": 0.4146, "step": 16680 }, { "epoch": 0.5942960970482926, "grad_norm": 1.0814255531667867, "learning_rate": 7.460048778238436e-06, "loss": 0.4096, "step": 16681 }, { "epoch": 0.5943317241747866, "grad_norm": 0.9938518567671566, "learning_rate": 7.458932701233738e-06, "loss": 0.4325, "step": 16682 }, { "epoch": 0.5943673513012808, "grad_norm": 1.4680267639909699, "learning_rate": 7.4578166580651335e-06, "loss": 0.7084, "step": 16683 }, { "epoch": 0.5944029784277749, "grad_norm": 1.327904727744844, "learning_rate": 7.456700648747473e-06, "loss": 0.4975, "step": 16684 }, { "epoch": 0.594438605554269, "grad_norm": 1.105179450279963, "learning_rate": 7.455584673295621e-06, "loss": 0.4629, "step": 16685 }, { "epoch": 0.5944742326807632, "grad_norm": 1.1592438671409595, "learning_rate": 7.454468731724438e-06, "loss": 0.4498, "step": 16686 }, { "epoch": 0.5945098598072572, "grad_norm": 1.0906481216452681, "learning_rate": 7.453352824048786e-06, "loss": 0.4735, "step": 16687 }, { "epoch": 0.5945454869337513, "grad_norm": 2.1856740110950748, "learning_rate": 7.452236950283518e-06, "loss": 0.5456, "step": 16688 }, { "epoch": 0.5945811140602455, "grad_norm": 1.4648932115944708, "learning_rate": 7.451121110443496e-06, "loss": 0.5554, "step": 16689 }, { "epoch": 0.5946167411867396, "grad_norm": 1.1152961492585505, "learning_rate": 7.450005304543578e-06, "loss": 0.4267, "step": 16690 }, { "epoch": 0.5946523683132336, "grad_norm": 1.0730688119781409, "learning_rate": 7.448889532598621e-06, "loss": 0.4868, "step": 16691 }, { "epoch": 0.5946879954397278, "grad_norm": 1.3831531122791063, "learning_rate": 7.447773794623487e-06, "loss": 0.493, "step": 16692 }, { "epoch": 0.5947236225662219, "grad_norm": 1.6962294476400361, "learning_rate": 7.4466580906330254e-06, "loss": 0.6431, "step": 16693 }, { "epoch": 0.594759249692716, "grad_norm": 1.731081432176491, "learning_rate": 7.445542420642097e-06, "loss": 0.5479, "step": 16694 }, { "epoch": 0.5947948768192102, "grad_norm": 1.2644563507925417, "learning_rate": 7.444426784665555e-06, "loss": 0.5035, "step": 16695 }, { "epoch": 0.5948305039457042, "grad_norm": 1.3027032968098793, "learning_rate": 7.443311182718258e-06, "loss": 0.5192, "step": 16696 }, { "epoch": 0.5948661310721983, "grad_norm": 1.37943124974966, "learning_rate": 7.442195614815058e-06, "loss": 0.5298, "step": 16697 }, { "epoch": 0.5949017581986925, "grad_norm": 1.11424690229972, "learning_rate": 7.441080080970816e-06, "loss": 0.4124, "step": 16698 }, { "epoch": 0.5949373853251866, "grad_norm": 1.9050194038199382, "learning_rate": 7.439964581200376e-06, "loss": 0.5332, "step": 16699 }, { "epoch": 0.5949730124516807, "grad_norm": 1.9559723476997597, "learning_rate": 7.4388491155186e-06, "loss": 0.4701, "step": 16700 }, { "epoch": 0.5950086395781748, "grad_norm": 1.1166640650041544, "learning_rate": 7.437733683940336e-06, "loss": 0.4775, "step": 16701 }, { "epoch": 0.5950442667046689, "grad_norm": 1.0777380969270018, "learning_rate": 7.436618286480439e-06, "loss": 0.4267, "step": 16702 }, { "epoch": 0.595079893831163, "grad_norm": 1.5373465018857804, "learning_rate": 7.435502923153766e-06, "loss": 0.5353, "step": 16703 }, { "epoch": 0.5951155209576572, "grad_norm": 1.1231206200986037, "learning_rate": 7.434387593975158e-06, "loss": 0.4582, "step": 16704 }, { "epoch": 0.5951511480841513, "grad_norm": 1.1049306649079567, "learning_rate": 7.4332722989594754e-06, "loss": 0.4427, "step": 16705 }, { "epoch": 0.5951867752106453, "grad_norm": 1.2179576187128496, "learning_rate": 7.4321570381215645e-06, "loss": 0.4415, "step": 16706 }, { "epoch": 0.5952224023371395, "grad_norm": 1.4272262201562576, "learning_rate": 7.43104181147628e-06, "loss": 0.5464, "step": 16707 }, { "epoch": 0.5952580294636336, "grad_norm": 1.1878748659040874, "learning_rate": 7.429926619038465e-06, "loss": 0.4129, "step": 16708 }, { "epoch": 0.5952936565901277, "grad_norm": 1.4644326410770006, "learning_rate": 7.42881146082298e-06, "loss": 0.5331, "step": 16709 }, { "epoch": 0.5953292837166219, "grad_norm": 1.421216401685702, "learning_rate": 7.427696336844663e-06, "loss": 0.5601, "step": 16710 }, { "epoch": 0.5953649108431159, "grad_norm": 1.5082276746375685, "learning_rate": 7.426581247118368e-06, "loss": 0.5727, "step": 16711 }, { "epoch": 0.59540053796961, "grad_norm": 1.086469020782137, "learning_rate": 7.425466191658941e-06, "loss": 0.4052, "step": 16712 }, { "epoch": 0.5954361650961042, "grad_norm": 1.1349445376461744, "learning_rate": 7.424351170481232e-06, "loss": 0.4054, "step": 16713 }, { "epoch": 0.5954717922225983, "grad_norm": 1.8681195504977057, "learning_rate": 7.423236183600091e-06, "loss": 0.5676, "step": 16714 }, { "epoch": 0.5955074193490923, "grad_norm": 1.4402473067579382, "learning_rate": 7.422121231030357e-06, "loss": 0.458, "step": 16715 }, { "epoch": 0.5955430464755865, "grad_norm": 1.1655202745135564, "learning_rate": 7.421006312786882e-06, "loss": 0.4419, "step": 16716 }, { "epoch": 0.5955786736020806, "grad_norm": 1.0642081940492032, "learning_rate": 7.419891428884509e-06, "loss": 0.4493, "step": 16717 }, { "epoch": 0.5956143007285747, "grad_norm": 1.5554492783036586, "learning_rate": 7.41877657933809e-06, "loss": 0.5542, "step": 16718 }, { "epoch": 0.5956499278550689, "grad_norm": 1.1688289646158823, "learning_rate": 7.417661764162458e-06, "loss": 0.4533, "step": 16719 }, { "epoch": 0.595685554981563, "grad_norm": 1.445182398096377, "learning_rate": 7.416546983372472e-06, "loss": 0.538, "step": 16720 }, { "epoch": 0.595721182108057, "grad_norm": 1.581642974873334, "learning_rate": 7.415432236982962e-06, "loss": 0.507, "step": 16721 }, { "epoch": 0.5957568092345512, "grad_norm": 1.2303476896157515, "learning_rate": 7.414317525008783e-06, "loss": 0.4976, "step": 16722 }, { "epoch": 0.5957924363610453, "grad_norm": 1.236591500728637, "learning_rate": 7.413202847464769e-06, "loss": 0.4645, "step": 16723 }, { "epoch": 0.5958280634875394, "grad_norm": 1.7071589174307147, "learning_rate": 7.412088204365768e-06, "loss": 0.5676, "step": 16724 }, { "epoch": 0.5958636906140335, "grad_norm": 1.266245398680091, "learning_rate": 7.410973595726627e-06, "loss": 0.4551, "step": 16725 }, { "epoch": 0.5958993177405276, "grad_norm": 1.1426610122941803, "learning_rate": 7.409859021562177e-06, "loss": 0.4426, "step": 16726 }, { "epoch": 0.5959349448670217, "grad_norm": 1.7129748435245047, "learning_rate": 7.408744481887267e-06, "loss": 0.5678, "step": 16727 }, { "epoch": 0.5959705719935159, "grad_norm": 0.9733251521086889, "learning_rate": 7.4076299767167325e-06, "loss": 0.4137, "step": 16728 }, { "epoch": 0.59600619912001, "grad_norm": 1.4421458601746853, "learning_rate": 7.406515506065422e-06, "loss": 0.514, "step": 16729 }, { "epoch": 0.596041826246504, "grad_norm": 1.108164434087957, "learning_rate": 7.405401069948163e-06, "loss": 0.4292, "step": 16730 }, { "epoch": 0.5960774533729982, "grad_norm": 1.674822508505832, "learning_rate": 7.404286668379808e-06, "loss": 0.5507, "step": 16731 }, { "epoch": 0.5961130804994923, "grad_norm": 1.4564574336937444, "learning_rate": 7.403172301375188e-06, "loss": 0.5097, "step": 16732 }, { "epoch": 0.5961487076259864, "grad_norm": 1.3335496982380028, "learning_rate": 7.4020579689491434e-06, "loss": 0.4772, "step": 16733 }, { "epoch": 0.5961843347524806, "grad_norm": 1.375657572096951, "learning_rate": 7.400943671116513e-06, "loss": 0.4584, "step": 16734 }, { "epoch": 0.5962199618789746, "grad_norm": 1.4872972863714187, "learning_rate": 7.399829407892134e-06, "loss": 0.5717, "step": 16735 }, { "epoch": 0.5962555890054687, "grad_norm": 1.1296472106850852, "learning_rate": 7.3987151792908475e-06, "loss": 0.4029, "step": 16736 }, { "epoch": 0.5962912161319629, "grad_norm": 0.9737823689567163, "learning_rate": 7.397600985327482e-06, "loss": 0.4916, "step": 16737 }, { "epoch": 0.596326843258457, "grad_norm": 1.1079263847618221, "learning_rate": 7.39648682601688e-06, "loss": 0.4376, "step": 16738 }, { "epoch": 0.596362470384951, "grad_norm": 1.0830391894349374, "learning_rate": 7.395372701373876e-06, "loss": 0.4092, "step": 16739 }, { "epoch": 0.5963980975114452, "grad_norm": 1.6036904919039756, "learning_rate": 7.394258611413308e-06, "loss": 0.5472, "step": 16740 }, { "epoch": 0.5964337246379393, "grad_norm": 1.4472698988070036, "learning_rate": 7.393144556150003e-06, "loss": 0.4945, "step": 16741 }, { "epoch": 0.5964693517644334, "grad_norm": 1.121706199125315, "learning_rate": 7.392030535598802e-06, "loss": 0.465, "step": 16742 }, { "epoch": 0.5965049788909276, "grad_norm": 1.1475193702475834, "learning_rate": 7.390916549774536e-06, "loss": 0.418, "step": 16743 }, { "epoch": 0.5965406060174216, "grad_norm": 1.1194795313074, "learning_rate": 7.389802598692044e-06, "loss": 0.4431, "step": 16744 }, { "epoch": 0.5965762331439157, "grad_norm": 2.0622008823596962, "learning_rate": 7.388688682366147e-06, "loss": 0.4833, "step": 16745 }, { "epoch": 0.5966118602704099, "grad_norm": 1.1453382205129563, "learning_rate": 7.387574800811694e-06, "loss": 0.4298, "step": 16746 }, { "epoch": 0.596647487396904, "grad_norm": 1.1303160284388207, "learning_rate": 7.3864609540435036e-06, "loss": 0.4514, "step": 16747 }, { "epoch": 0.5966831145233981, "grad_norm": 1.147752837891865, "learning_rate": 7.385347142076412e-06, "loss": 0.4333, "step": 16748 }, { "epoch": 0.5967187416498922, "grad_norm": 1.2033340971357165, "learning_rate": 7.384233364925253e-06, "loss": 0.4819, "step": 16749 }, { "epoch": 0.5967543687763863, "grad_norm": 1.6856057191082547, "learning_rate": 7.383119622604853e-06, "loss": 0.4844, "step": 16750 }, { "epoch": 0.5967899959028804, "grad_norm": 1.3747920944212249, "learning_rate": 7.3820059151300485e-06, "loss": 0.5543, "step": 16751 }, { "epoch": 0.5968256230293746, "grad_norm": 1.6352683440015483, "learning_rate": 7.3808922425156605e-06, "loss": 0.521, "step": 16752 }, { "epoch": 0.5968612501558687, "grad_norm": 1.3562412756133821, "learning_rate": 7.379778604776527e-06, "loss": 0.5612, "step": 16753 }, { "epoch": 0.5968968772823627, "grad_norm": 1.208726565605891, "learning_rate": 7.378665001927468e-06, "loss": 0.3913, "step": 16754 }, { "epoch": 0.5969325044088569, "grad_norm": 1.2121480585301219, "learning_rate": 7.377551433983322e-06, "loss": 0.4011, "step": 16755 }, { "epoch": 0.596968131535351, "grad_norm": 1.3809856453683322, "learning_rate": 7.376437900958908e-06, "loss": 0.5763, "step": 16756 }, { "epoch": 0.5970037586618451, "grad_norm": 1.0920179838697424, "learning_rate": 7.375324402869057e-06, "loss": 0.3967, "step": 16757 }, { "epoch": 0.5970393857883393, "grad_norm": 1.119435915484783, "learning_rate": 7.374210939728596e-06, "loss": 0.4229, "step": 16758 }, { "epoch": 0.5970750129148333, "grad_norm": 1.460713677194426, "learning_rate": 7.373097511552352e-06, "loss": 0.5757, "step": 16759 }, { "epoch": 0.5971106400413274, "grad_norm": 1.0894896091846256, "learning_rate": 7.3719841183551525e-06, "loss": 0.4271, "step": 16760 }, { "epoch": 0.5971462671678216, "grad_norm": 1.3216803479455497, "learning_rate": 7.370870760151818e-06, "loss": 0.5077, "step": 16761 }, { "epoch": 0.5971818942943157, "grad_norm": 1.5913650131788548, "learning_rate": 7.369757436957183e-06, "loss": 0.425, "step": 16762 }, { "epoch": 0.5972175214208097, "grad_norm": 1.372375358390109, "learning_rate": 7.3686441487860615e-06, "loss": 0.5033, "step": 16763 }, { "epoch": 0.5972531485473039, "grad_norm": 1.143369532594859, "learning_rate": 7.367530895653283e-06, "loss": 0.4304, "step": 16764 }, { "epoch": 0.597288775673798, "grad_norm": 1.1272574061212264, "learning_rate": 7.36641767757367e-06, "loss": 0.4492, "step": 16765 }, { "epoch": 0.5973244028002922, "grad_norm": 1.5738916281003048, "learning_rate": 7.36530449456205e-06, "loss": 0.483, "step": 16766 }, { "epoch": 0.5973600299267863, "grad_norm": 2.1059449014596283, "learning_rate": 7.364191346633238e-06, "loss": 0.4858, "step": 16767 }, { "epoch": 0.5973956570532803, "grad_norm": 1.0248913696618036, "learning_rate": 7.363078233802063e-06, "loss": 0.4039, "step": 16768 }, { "epoch": 0.5974312841797745, "grad_norm": 1.665742690520666, "learning_rate": 7.3619651560833415e-06, "loss": 0.493, "step": 16769 }, { "epoch": 0.5974669113062686, "grad_norm": 1.7854219289904332, "learning_rate": 7.360852113491899e-06, "loss": 0.5524, "step": 16770 }, { "epoch": 0.5975025384327627, "grad_norm": 1.5806783699258309, "learning_rate": 7.359739106042556e-06, "loss": 0.5714, "step": 16771 }, { "epoch": 0.5975381655592569, "grad_norm": 1.085556540883487, "learning_rate": 7.358626133750131e-06, "loss": 0.3917, "step": 16772 }, { "epoch": 0.597573792685751, "grad_norm": 1.1639814689996446, "learning_rate": 7.35751319662945e-06, "loss": 0.4606, "step": 16773 }, { "epoch": 0.597609419812245, "grad_norm": 1.19535470838776, "learning_rate": 7.3564002946953226e-06, "loss": 0.3777, "step": 16774 }, { "epoch": 0.5976450469387392, "grad_norm": 1.1810777321435706, "learning_rate": 7.355287427962577e-06, "loss": 0.4921, "step": 16775 }, { "epoch": 0.5976806740652333, "grad_norm": 1.044149015030078, "learning_rate": 7.354174596446022e-06, "loss": 0.4599, "step": 16776 }, { "epoch": 0.5977163011917274, "grad_norm": 1.1880044379973438, "learning_rate": 7.3530618001604885e-06, "loss": 0.4157, "step": 16777 }, { "epoch": 0.5977519283182215, "grad_norm": 1.0289314899009854, "learning_rate": 7.351949039120782e-06, "loss": 0.4161, "step": 16778 }, { "epoch": 0.5977875554447156, "grad_norm": 1.2361543305562879, "learning_rate": 7.3508363133417286e-06, "loss": 0.3999, "step": 16779 }, { "epoch": 0.5978231825712097, "grad_norm": 1.1708760497908568, "learning_rate": 7.349723622838139e-06, "loss": 0.4722, "step": 16780 }, { "epoch": 0.5978588096977039, "grad_norm": 1.3257941171214505, "learning_rate": 7.34861096762483e-06, "loss": 0.4554, "step": 16781 }, { "epoch": 0.597894436824198, "grad_norm": 1.5268138027789633, "learning_rate": 7.347498347716624e-06, "loss": 0.5356, "step": 16782 }, { "epoch": 0.597930063950692, "grad_norm": 1.2328611115388561, "learning_rate": 7.346385763128325e-06, "loss": 0.4429, "step": 16783 }, { "epoch": 0.5979656910771862, "grad_norm": 2.0841972532046227, "learning_rate": 7.345273213874762e-06, "loss": 0.5771, "step": 16784 }, { "epoch": 0.5980013182036803, "grad_norm": 1.446016320207372, "learning_rate": 7.344160699970736e-06, "loss": 0.5794, "step": 16785 }, { "epoch": 0.5980369453301744, "grad_norm": 1.1117240586787935, "learning_rate": 7.34304822143107e-06, "loss": 0.4377, "step": 16786 }, { "epoch": 0.5980725724566686, "grad_norm": 1.6571573404081796, "learning_rate": 7.34193577827057e-06, "loss": 0.5112, "step": 16787 }, { "epoch": 0.5981081995831626, "grad_norm": 1.4559487199171983, "learning_rate": 7.34082337050406e-06, "loss": 0.4437, "step": 16788 }, { "epoch": 0.5981438267096567, "grad_norm": 1.5139737388132326, "learning_rate": 7.339710998146339e-06, "loss": 0.4933, "step": 16789 }, { "epoch": 0.5981794538361509, "grad_norm": 1.1550033355158278, "learning_rate": 7.338598661212229e-06, "loss": 0.435, "step": 16790 }, { "epoch": 0.598215080962645, "grad_norm": 1.1143903478951156, "learning_rate": 7.337486359716535e-06, "loss": 0.4568, "step": 16791 }, { "epoch": 0.598250708089139, "grad_norm": 1.197432908292269, "learning_rate": 7.3363740936740744e-06, "loss": 0.4091, "step": 16792 }, { "epoch": 0.5982863352156332, "grad_norm": 1.3947336103794243, "learning_rate": 7.335261863099652e-06, "loss": 0.515, "step": 16793 }, { "epoch": 0.5983219623421273, "grad_norm": 1.2354187134507422, "learning_rate": 7.334149668008078e-06, "loss": 0.4511, "step": 16794 }, { "epoch": 0.5983575894686214, "grad_norm": 1.1944687111204861, "learning_rate": 7.333037508414171e-06, "loss": 0.4805, "step": 16795 }, { "epoch": 0.5983932165951156, "grad_norm": 0.9469100317136045, "learning_rate": 7.331925384332729e-06, "loss": 0.3919, "step": 16796 }, { "epoch": 0.5984288437216096, "grad_norm": 1.1028530110763002, "learning_rate": 7.330813295778567e-06, "loss": 0.4899, "step": 16797 }, { "epoch": 0.5984644708481037, "grad_norm": 1.0608942486099828, "learning_rate": 7.329701242766491e-06, "loss": 0.43, "step": 16798 }, { "epoch": 0.5985000979745979, "grad_norm": 1.2120347424096303, "learning_rate": 7.3285892253113115e-06, "loss": 0.5277, "step": 16799 }, { "epoch": 0.598535725101092, "grad_norm": 1.0707007633995211, "learning_rate": 7.3274772434278315e-06, "loss": 0.4084, "step": 16800 }, { "epoch": 0.5985713522275861, "grad_norm": 1.62422235908383, "learning_rate": 7.326365297130861e-06, "loss": 0.5719, "step": 16801 }, { "epoch": 0.5986069793540802, "grad_norm": 1.5254874611725944, "learning_rate": 7.325253386435203e-06, "loss": 0.518, "step": 16802 }, { "epoch": 0.5986426064805743, "grad_norm": 1.1698693560637123, "learning_rate": 7.324141511355671e-06, "loss": 0.4258, "step": 16803 }, { "epoch": 0.5986782336070684, "grad_norm": 1.0864073477781566, "learning_rate": 7.32302967190706e-06, "loss": 0.432, "step": 16804 }, { "epoch": 0.5987138607335626, "grad_norm": 1.3363499954355906, "learning_rate": 7.32191786810418e-06, "loss": 0.4791, "step": 16805 }, { "epoch": 0.5987494878600567, "grad_norm": 1.0385478678674112, "learning_rate": 7.320806099961837e-06, "loss": 0.4121, "step": 16806 }, { "epoch": 0.5987851149865507, "grad_norm": 1.6871528727514848, "learning_rate": 7.319694367494833e-06, "loss": 0.5774, "step": 16807 }, { "epoch": 0.5988207421130449, "grad_norm": 1.7454281261700202, "learning_rate": 7.318582670717973e-06, "loss": 0.5252, "step": 16808 }, { "epoch": 0.598856369239539, "grad_norm": 1.4270635181948876, "learning_rate": 7.317471009646057e-06, "loss": 0.5002, "step": 16809 }, { "epoch": 0.5988919963660331, "grad_norm": 1.05675264565755, "learning_rate": 7.316359384293894e-06, "loss": 0.4069, "step": 16810 }, { "epoch": 0.5989276234925273, "grad_norm": 1.998952277857804, "learning_rate": 7.315247794676276e-06, "loss": 0.4752, "step": 16811 }, { "epoch": 0.5989632506190213, "grad_norm": 1.3953628732454806, "learning_rate": 7.314136240808014e-06, "loss": 0.5534, "step": 16812 }, { "epoch": 0.5989988777455154, "grad_norm": 1.1021606599726992, "learning_rate": 7.313024722703902e-06, "loss": 0.4157, "step": 16813 }, { "epoch": 0.5990345048720096, "grad_norm": 1.1508601877823865, "learning_rate": 7.311913240378748e-06, "loss": 0.4608, "step": 16814 }, { "epoch": 0.5990701319985037, "grad_norm": 1.2391182107419256, "learning_rate": 7.310801793847344e-06, "loss": 0.4289, "step": 16815 }, { "epoch": 0.5991057591249977, "grad_norm": 1.6781053949468758, "learning_rate": 7.309690383124494e-06, "loss": 0.5708, "step": 16816 }, { "epoch": 0.5991413862514919, "grad_norm": 1.5910026197954295, "learning_rate": 7.308579008225001e-06, "loss": 0.536, "step": 16817 }, { "epoch": 0.599177013377986, "grad_norm": 1.3141265222984526, "learning_rate": 7.307467669163655e-06, "loss": 0.5558, "step": 16818 }, { "epoch": 0.5992126405044801, "grad_norm": 1.3217041712001236, "learning_rate": 7.3063563659552635e-06, "loss": 0.5697, "step": 16819 }, { "epoch": 0.5992482676309743, "grad_norm": 1.4654317374843089, "learning_rate": 7.305245098614614e-06, "loss": 0.528, "step": 16820 }, { "epoch": 0.5992838947574683, "grad_norm": 1.137637305270565, "learning_rate": 7.304133867156516e-06, "loss": 0.3893, "step": 16821 }, { "epoch": 0.5993195218839624, "grad_norm": 1.6802951077403496, "learning_rate": 7.3030226715957555e-06, "loss": 0.5628, "step": 16822 }, { "epoch": 0.5993551490104566, "grad_norm": 1.1694757989768532, "learning_rate": 7.301911511947135e-06, "loss": 0.4431, "step": 16823 }, { "epoch": 0.5993907761369507, "grad_norm": 1.513228912624338, "learning_rate": 7.300800388225446e-06, "loss": 0.5411, "step": 16824 }, { "epoch": 0.5994264032634448, "grad_norm": 1.4675535692827968, "learning_rate": 7.299689300445491e-06, "loss": 0.6855, "step": 16825 }, { "epoch": 0.599462030389939, "grad_norm": 1.085943326246501, "learning_rate": 7.2985782486220565e-06, "loss": 0.4271, "step": 16826 }, { "epoch": 0.599497657516433, "grad_norm": 1.1826256961882702, "learning_rate": 7.297467232769943e-06, "loss": 0.3923, "step": 16827 }, { "epoch": 0.5995332846429271, "grad_norm": 1.3844537319493067, "learning_rate": 7.29635625290394e-06, "loss": 0.4747, "step": 16828 }, { "epoch": 0.5995689117694213, "grad_norm": 1.458680348426656, "learning_rate": 7.295245309038843e-06, "loss": 0.5172, "step": 16829 }, { "epoch": 0.5996045388959154, "grad_norm": 1.4891760657543733, "learning_rate": 7.29413440118945e-06, "loss": 0.5778, "step": 16830 }, { "epoch": 0.5996401660224094, "grad_norm": 1.1106196629324039, "learning_rate": 7.293023529370544e-06, "loss": 0.3924, "step": 16831 }, { "epoch": 0.5996757931489036, "grad_norm": 1.52701837053869, "learning_rate": 7.291912693596924e-06, "loss": 0.5399, "step": 16832 }, { "epoch": 0.5997114202753977, "grad_norm": 1.1218154335302704, "learning_rate": 7.290801893883377e-06, "loss": 0.432, "step": 16833 }, { "epoch": 0.5997470474018918, "grad_norm": 1.89410075938559, "learning_rate": 7.289691130244699e-06, "loss": 0.5228, "step": 16834 }, { "epoch": 0.599782674528386, "grad_norm": 1.4740191026918454, "learning_rate": 7.288580402695676e-06, "loss": 0.4895, "step": 16835 }, { "epoch": 0.59981830165488, "grad_norm": 1.4913528728978482, "learning_rate": 7.287469711251104e-06, "loss": 0.5227, "step": 16836 }, { "epoch": 0.5998539287813741, "grad_norm": 1.0499853249452475, "learning_rate": 7.286359055925764e-06, "loss": 0.4038, "step": 16837 }, { "epoch": 0.5998895559078683, "grad_norm": 1.564524216085717, "learning_rate": 7.285248436734453e-06, "loss": 0.5892, "step": 16838 }, { "epoch": 0.5999251830343624, "grad_norm": 1.4723817391148992, "learning_rate": 7.284137853691954e-06, "loss": 0.5627, "step": 16839 }, { "epoch": 0.5999608101608565, "grad_norm": 1.1782594384278529, "learning_rate": 7.283027306813058e-06, "loss": 0.4516, "step": 16840 }, { "epoch": 0.5999964372873506, "grad_norm": 1.2093179767729063, "learning_rate": 7.2819167961125555e-06, "loss": 0.4332, "step": 16841 }, { "epoch": 0.6000320644138447, "grad_norm": 1.5453713260428807, "learning_rate": 7.280806321605227e-06, "loss": 0.5452, "step": 16842 }, { "epoch": 0.6000676915403388, "grad_norm": 1.7155491357926693, "learning_rate": 7.2796958833058665e-06, "loss": 0.5875, "step": 16843 }, { "epoch": 0.600103318666833, "grad_norm": 1.532848458908017, "learning_rate": 7.278585481229253e-06, "loss": 0.546, "step": 16844 }, { "epoch": 0.600138945793327, "grad_norm": 1.6070324052090987, "learning_rate": 7.2774751153901805e-06, "loss": 0.5202, "step": 16845 }, { "epoch": 0.6001745729198211, "grad_norm": 1.483939690802336, "learning_rate": 7.276364785803425e-06, "loss": 0.5544, "step": 16846 }, { "epoch": 0.6002102000463153, "grad_norm": 1.6772629764346212, "learning_rate": 7.27525449248378e-06, "loss": 0.5386, "step": 16847 }, { "epoch": 0.6002458271728094, "grad_norm": 1.0454682224955452, "learning_rate": 7.274144235446024e-06, "loss": 0.4178, "step": 16848 }, { "epoch": 0.6002814542993035, "grad_norm": 1.6802296750856016, "learning_rate": 7.273034014704943e-06, "loss": 0.5688, "step": 16849 }, { "epoch": 0.6003170814257976, "grad_norm": 1.6975987206299878, "learning_rate": 7.271923830275318e-06, "loss": 0.5588, "step": 16850 }, { "epoch": 0.6003527085522917, "grad_norm": 1.0870329454919911, "learning_rate": 7.270813682171934e-06, "loss": 0.4786, "step": 16851 }, { "epoch": 0.6003883356787858, "grad_norm": 1.2925635419136807, "learning_rate": 7.269703570409579e-06, "loss": 0.5258, "step": 16852 }, { "epoch": 0.60042396280528, "grad_norm": 1.2082296627619622, "learning_rate": 7.268593495003025e-06, "loss": 0.3821, "step": 16853 }, { "epoch": 0.6004595899317741, "grad_norm": 1.2533120799067607, "learning_rate": 7.2674834559670585e-06, "loss": 0.4333, "step": 16854 }, { "epoch": 0.6004952170582681, "grad_norm": 1.071545257730547, "learning_rate": 7.266373453316458e-06, "loss": 0.4468, "step": 16855 }, { "epoch": 0.6005308441847623, "grad_norm": 1.3406869138961703, "learning_rate": 7.26526348706601e-06, "loss": 0.5659, "step": 16856 }, { "epoch": 0.6005664713112564, "grad_norm": 1.4781913408839662, "learning_rate": 7.264153557230487e-06, "loss": 0.5239, "step": 16857 }, { "epoch": 0.6006020984377505, "grad_norm": 1.136535283701988, "learning_rate": 7.2630436638246715e-06, "loss": 0.4267, "step": 16858 }, { "epoch": 0.6006377255642447, "grad_norm": 1.1799771543648954, "learning_rate": 7.2619338068633425e-06, "loss": 0.4573, "step": 16859 }, { "epoch": 0.6006733526907387, "grad_norm": 1.6676205992648763, "learning_rate": 7.26082398636128e-06, "loss": 0.5578, "step": 16860 }, { "epoch": 0.6007089798172328, "grad_norm": 1.5963089270023563, "learning_rate": 7.259714202333259e-06, "loss": 0.4878, "step": 16861 }, { "epoch": 0.600744606943727, "grad_norm": 1.647803652886752, "learning_rate": 7.258604454794058e-06, "loss": 0.584, "step": 16862 }, { "epoch": 0.6007802340702211, "grad_norm": 1.3791677374857148, "learning_rate": 7.25749474375846e-06, "loss": 0.5956, "step": 16863 }, { "epoch": 0.6008158611967152, "grad_norm": 1.0640219073313597, "learning_rate": 7.256385069241232e-06, "loss": 0.3788, "step": 16864 }, { "epoch": 0.6008514883232093, "grad_norm": 1.0711137448502388, "learning_rate": 7.255275431257156e-06, "loss": 0.4381, "step": 16865 }, { "epoch": 0.6008871154497034, "grad_norm": 1.3511247404851305, "learning_rate": 7.254165829821004e-06, "loss": 0.5561, "step": 16866 }, { "epoch": 0.6009227425761975, "grad_norm": 1.4271670990490608, "learning_rate": 7.253056264947558e-06, "loss": 0.5355, "step": 16867 }, { "epoch": 0.6009583697026917, "grad_norm": 1.0985554458676559, "learning_rate": 7.251946736651583e-06, "loss": 0.4237, "step": 16868 }, { "epoch": 0.6009939968291858, "grad_norm": 1.07530241309017, "learning_rate": 7.2508372449478605e-06, "loss": 0.402, "step": 16869 }, { "epoch": 0.6010296239556798, "grad_norm": 1.0956379902133588, "learning_rate": 7.249727789851159e-06, "loss": 0.435, "step": 16870 }, { "epoch": 0.601065251082174, "grad_norm": 1.0710456136442195, "learning_rate": 7.248618371376259e-06, "loss": 0.4044, "step": 16871 }, { "epoch": 0.6011008782086681, "grad_norm": 1.1554631849055528, "learning_rate": 7.247508989537922e-06, "loss": 0.4324, "step": 16872 }, { "epoch": 0.6011365053351622, "grad_norm": 1.2284028540021739, "learning_rate": 7.246399644350932e-06, "loss": 0.4676, "step": 16873 }, { "epoch": 0.6011721324616563, "grad_norm": 1.478913045662309, "learning_rate": 7.2452903358300534e-06, "loss": 0.5794, "step": 16874 }, { "epoch": 0.6012077595881504, "grad_norm": 1.070503915157871, "learning_rate": 7.2441810639900575e-06, "loss": 0.4038, "step": 16875 }, { "epoch": 0.6012433867146445, "grad_norm": 0.9607264283240937, "learning_rate": 7.24307182884572e-06, "loss": 0.4227, "step": 16876 }, { "epoch": 0.6012790138411387, "grad_norm": 4.206344496851456, "learning_rate": 7.241962630411805e-06, "loss": 0.5362, "step": 16877 }, { "epoch": 0.6013146409676328, "grad_norm": 1.5628505389226497, "learning_rate": 7.2408534687030905e-06, "loss": 0.4754, "step": 16878 }, { "epoch": 0.6013502680941268, "grad_norm": 1.116210249459325, "learning_rate": 7.239744343734336e-06, "loss": 0.4642, "step": 16879 }, { "epoch": 0.601385895220621, "grad_norm": 1.2900869737861962, "learning_rate": 7.238635255520318e-06, "loss": 0.5795, "step": 16880 }, { "epoch": 0.6014215223471151, "grad_norm": 1.300771735445882, "learning_rate": 7.237526204075798e-06, "loss": 0.4494, "step": 16881 }, { "epoch": 0.6014571494736092, "grad_norm": 1.192352915526327, "learning_rate": 7.236417189415552e-06, "loss": 0.4744, "step": 16882 }, { "epoch": 0.6014927766001034, "grad_norm": 1.554134700770604, "learning_rate": 7.2353082115543376e-06, "loss": 0.5049, "step": 16883 }, { "epoch": 0.6015284037265974, "grad_norm": 2.105730700108537, "learning_rate": 7.234199270506932e-06, "loss": 0.5989, "step": 16884 }, { "epoch": 0.6015640308530915, "grad_norm": 2.2058612508620405, "learning_rate": 7.233090366288092e-06, "loss": 0.5795, "step": 16885 }, { "epoch": 0.6015996579795857, "grad_norm": 1.588475562753593, "learning_rate": 7.231981498912588e-06, "loss": 0.5813, "step": 16886 }, { "epoch": 0.6016352851060798, "grad_norm": 1.362471076314794, "learning_rate": 7.230872668395188e-06, "loss": 0.4223, "step": 16887 }, { "epoch": 0.6016709122325739, "grad_norm": 1.1955960110289374, "learning_rate": 7.229763874750651e-06, "loss": 0.4786, "step": 16888 }, { "epoch": 0.601706539359068, "grad_norm": 1.2095529471478343, "learning_rate": 7.228655117993748e-06, "loss": 0.4868, "step": 16889 }, { "epoch": 0.6017421664855621, "grad_norm": 1.120042587851911, "learning_rate": 7.227546398139236e-06, "loss": 0.444, "step": 16890 }, { "epoch": 0.6017777936120562, "grad_norm": 1.0719214234632841, "learning_rate": 7.226437715201882e-06, "loss": 0.4287, "step": 16891 }, { "epoch": 0.6018134207385504, "grad_norm": 2.0534340480812254, "learning_rate": 7.2253290691964485e-06, "loss": 0.5327, "step": 16892 }, { "epoch": 0.6018490478650445, "grad_norm": 1.4052168231012205, "learning_rate": 7.224220460137701e-06, "loss": 0.5344, "step": 16893 }, { "epoch": 0.6018846749915385, "grad_norm": 1.2468019154400736, "learning_rate": 7.223111888040394e-06, "loss": 0.4496, "step": 16894 }, { "epoch": 0.6019203021180327, "grad_norm": 1.4518296836124382, "learning_rate": 7.222003352919296e-06, "loss": 0.5219, "step": 16895 }, { "epoch": 0.6019559292445268, "grad_norm": 1.1856596289847001, "learning_rate": 7.220894854789161e-06, "loss": 0.4359, "step": 16896 }, { "epoch": 0.6019915563710209, "grad_norm": 1.069815607072988, "learning_rate": 7.219786393664755e-06, "loss": 0.4346, "step": 16897 }, { "epoch": 0.602027183497515, "grad_norm": 1.1692799980087953, "learning_rate": 7.2186779695608375e-06, "loss": 0.4731, "step": 16898 }, { "epoch": 0.6020628106240091, "grad_norm": 1.1986471799566216, "learning_rate": 7.217569582492165e-06, "loss": 0.4536, "step": 16899 }, { "epoch": 0.6020984377505032, "grad_norm": 1.840443379730772, "learning_rate": 7.2164612324735035e-06, "loss": 0.5836, "step": 16900 }, { "epoch": 0.6021340648769974, "grad_norm": 1.8522613987060152, "learning_rate": 7.215352919519603e-06, "loss": 0.5367, "step": 16901 }, { "epoch": 0.6021696920034915, "grad_norm": 1.0798357445219218, "learning_rate": 7.214244643645224e-06, "loss": 0.4279, "step": 16902 }, { "epoch": 0.6022053191299855, "grad_norm": 1.613254774012974, "learning_rate": 7.213136404865124e-06, "loss": 0.5537, "step": 16903 }, { "epoch": 0.6022409462564797, "grad_norm": 1.319390766999322, "learning_rate": 7.212028203194066e-06, "loss": 0.4887, "step": 16904 }, { "epoch": 0.6022765733829738, "grad_norm": 1.139316829145212, "learning_rate": 7.210920038646795e-06, "loss": 0.4592, "step": 16905 }, { "epoch": 0.6023122005094679, "grad_norm": 1.4374449874807627, "learning_rate": 7.209811911238076e-06, "loss": 0.5304, "step": 16906 }, { "epoch": 0.6023478276359621, "grad_norm": 1.5615460314969372, "learning_rate": 7.208703820982659e-06, "loss": 0.6034, "step": 16907 }, { "epoch": 0.6023834547624561, "grad_norm": 1.8479426142439315, "learning_rate": 7.207595767895303e-06, "loss": 0.5118, "step": 16908 }, { "epoch": 0.6024190818889502, "grad_norm": 1.0876079061528434, "learning_rate": 7.206487751990763e-06, "loss": 0.4561, "step": 16909 }, { "epoch": 0.6024547090154444, "grad_norm": 1.6587098224604953, "learning_rate": 7.205379773283787e-06, "loss": 0.535, "step": 16910 }, { "epoch": 0.6024903361419385, "grad_norm": 1.1741123889851033, "learning_rate": 7.204271831789138e-06, "loss": 0.437, "step": 16911 }, { "epoch": 0.6025259632684326, "grad_norm": 1.1316450374309412, "learning_rate": 7.2031639275215606e-06, "loss": 0.4016, "step": 16912 }, { "epoch": 0.6025615903949267, "grad_norm": 2.720211059224792, "learning_rate": 7.202056060495811e-06, "loss": 0.4716, "step": 16913 }, { "epoch": 0.6025972175214208, "grad_norm": 1.0047171455483594, "learning_rate": 7.200948230726639e-06, "loss": 0.4115, "step": 16914 }, { "epoch": 0.6026328446479149, "grad_norm": 1.653653809334035, "learning_rate": 7.199840438228803e-06, "loss": 0.5337, "step": 16915 }, { "epoch": 0.6026684717744091, "grad_norm": 1.1881548848870758, "learning_rate": 7.1987326830170434e-06, "loss": 0.4964, "step": 16916 }, { "epoch": 0.6027040989009032, "grad_norm": 1.4444071362713091, "learning_rate": 7.197624965106118e-06, "loss": 0.523, "step": 16917 }, { "epoch": 0.6027397260273972, "grad_norm": 1.44440637707329, "learning_rate": 7.196517284510774e-06, "loss": 0.5243, "step": 16918 }, { "epoch": 0.6027753531538914, "grad_norm": 1.114479437638575, "learning_rate": 7.195409641245765e-06, "loss": 0.4551, "step": 16919 }, { "epoch": 0.6028109802803855, "grad_norm": 1.0919908361600323, "learning_rate": 7.194302035325833e-06, "loss": 0.429, "step": 16920 }, { "epoch": 0.6028466074068796, "grad_norm": 1.1134082701063301, "learning_rate": 7.193194466765727e-06, "loss": 0.4515, "step": 16921 }, { "epoch": 0.6028822345333738, "grad_norm": 1.1573808292084053, "learning_rate": 7.192086935580207e-06, "loss": 0.4386, "step": 16922 }, { "epoch": 0.6029178616598678, "grad_norm": 1.1879100529595308, "learning_rate": 7.1909794417840075e-06, "loss": 0.4496, "step": 16923 }, { "epoch": 0.6029534887863619, "grad_norm": 1.6180332476499788, "learning_rate": 7.1898719853918805e-06, "loss": 0.6682, "step": 16924 }, { "epoch": 0.6029891159128561, "grad_norm": 1.1410377859176897, "learning_rate": 7.1887645664185724e-06, "loss": 0.4213, "step": 16925 }, { "epoch": 0.6030247430393502, "grad_norm": 1.1561774194893424, "learning_rate": 7.187657184878833e-06, "loss": 0.4374, "step": 16926 }, { "epoch": 0.6030603701658442, "grad_norm": 1.468569029297455, "learning_rate": 7.186549840787398e-06, "loss": 0.5146, "step": 16927 }, { "epoch": 0.6030959972923384, "grad_norm": 1.115494809109237, "learning_rate": 7.185442534159021e-06, "loss": 0.4007, "step": 16928 }, { "epoch": 0.6031316244188325, "grad_norm": 1.4752492400273443, "learning_rate": 7.184335265008442e-06, "loss": 0.5105, "step": 16929 }, { "epoch": 0.6031672515453266, "grad_norm": 1.7107626295218863, "learning_rate": 7.183228033350411e-06, "loss": 0.5044, "step": 16930 }, { "epoch": 0.6032028786718208, "grad_norm": 1.2835831589980036, "learning_rate": 7.1821208391996644e-06, "loss": 0.4431, "step": 16931 }, { "epoch": 0.6032385057983148, "grad_norm": 1.1769189885492388, "learning_rate": 7.1810136825709474e-06, "loss": 0.4459, "step": 16932 }, { "epoch": 0.6032741329248089, "grad_norm": 1.2067159848039188, "learning_rate": 7.179906563479006e-06, "loss": 0.4375, "step": 16933 }, { "epoch": 0.6033097600513031, "grad_norm": 1.3467228489049312, "learning_rate": 7.1787994819385785e-06, "loss": 0.5825, "step": 16934 }, { "epoch": 0.6033453871777972, "grad_norm": 1.054031115988434, "learning_rate": 7.177692437964408e-06, "loss": 0.4157, "step": 16935 }, { "epoch": 0.6033810143042913, "grad_norm": 1.5561776346089822, "learning_rate": 7.176585431571235e-06, "loss": 0.577, "step": 16936 }, { "epoch": 0.6034166414307854, "grad_norm": 1.3063492441675468, "learning_rate": 7.175478462773804e-06, "loss": 0.5554, "step": 16937 }, { "epoch": 0.6034522685572795, "grad_norm": 1.1543032831376148, "learning_rate": 7.174371531586848e-06, "loss": 0.4207, "step": 16938 }, { "epoch": 0.6034878956837736, "grad_norm": 1.3587903924039304, "learning_rate": 7.1732646380251106e-06, "loss": 0.5448, "step": 16939 }, { "epoch": 0.6035235228102678, "grad_norm": 1.5060071374507822, "learning_rate": 7.172157782103328e-06, "loss": 0.5518, "step": 16940 }, { "epoch": 0.6035591499367619, "grad_norm": 1.2954115182745807, "learning_rate": 7.171050963836248e-06, "loss": 0.4551, "step": 16941 }, { "epoch": 0.6035947770632559, "grad_norm": 1.5052887066473786, "learning_rate": 7.169944183238595e-06, "loss": 0.5316, "step": 16942 }, { "epoch": 0.6036304041897501, "grad_norm": 1.6382575246558762, "learning_rate": 7.1688374403251134e-06, "loss": 0.4902, "step": 16943 }, { "epoch": 0.6036660313162442, "grad_norm": 1.1827503762773335, "learning_rate": 7.167730735110544e-06, "loss": 0.4492, "step": 16944 }, { "epoch": 0.6037016584427383, "grad_norm": 1.0638568310528296, "learning_rate": 7.166624067609616e-06, "loss": 0.392, "step": 16945 }, { "epoch": 0.6037372855692325, "grad_norm": 1.8079054368574154, "learning_rate": 7.165517437837074e-06, "loss": 0.5501, "step": 16946 }, { "epoch": 0.6037729126957265, "grad_norm": 1.595686257110739, "learning_rate": 7.1644108458076415e-06, "loss": 0.5365, "step": 16947 }, { "epoch": 0.6038085398222206, "grad_norm": 1.7190783209371574, "learning_rate": 7.163304291536067e-06, "loss": 0.453, "step": 16948 }, { "epoch": 0.6038441669487148, "grad_norm": 5.2769260257881445, "learning_rate": 7.1621977750370745e-06, "loss": 0.4391, "step": 16949 }, { "epoch": 0.6038797940752089, "grad_norm": 1.0630850919495851, "learning_rate": 7.161091296325404e-06, "loss": 0.3871, "step": 16950 }, { "epoch": 0.6039154212017029, "grad_norm": 1.0840736215712592, "learning_rate": 7.1599848554157855e-06, "loss": 0.3638, "step": 16951 }, { "epoch": 0.6039510483281971, "grad_norm": 1.1924380901418101, "learning_rate": 7.158878452322957e-06, "loss": 0.4559, "step": 16952 }, { "epoch": 0.6039866754546912, "grad_norm": 1.0676699354491848, "learning_rate": 7.157772087061645e-06, "loss": 0.4145, "step": 16953 }, { "epoch": 0.6040223025811853, "grad_norm": 1.073583397723984, "learning_rate": 7.156665759646586e-06, "loss": 0.4202, "step": 16954 }, { "epoch": 0.6040579297076795, "grad_norm": 1.2273086396806043, "learning_rate": 7.1555594700925065e-06, "loss": 0.4196, "step": 16955 }, { "epoch": 0.6040935568341735, "grad_norm": 1.5732361607416014, "learning_rate": 7.154453218414142e-06, "loss": 0.5177, "step": 16956 }, { "epoch": 0.6041291839606676, "grad_norm": 1.711432669314585, "learning_rate": 7.1533470046262255e-06, "loss": 0.4624, "step": 16957 }, { "epoch": 0.6041648110871618, "grad_norm": 1.0663198702591201, "learning_rate": 7.1522408287434774e-06, "loss": 0.4083, "step": 16958 }, { "epoch": 0.6042004382136559, "grad_norm": 1.597119558050342, "learning_rate": 7.1511346907806386e-06, "loss": 0.5629, "step": 16959 }, { "epoch": 0.60423606534015, "grad_norm": 1.1759198274156684, "learning_rate": 7.150028590752429e-06, "loss": 0.4307, "step": 16960 }, { "epoch": 0.6042716924666441, "grad_norm": 1.1490834958374736, "learning_rate": 7.148922528673583e-06, "loss": 0.4151, "step": 16961 }, { "epoch": 0.6043073195931382, "grad_norm": 1.0637376021492821, "learning_rate": 7.147816504558824e-06, "loss": 0.4193, "step": 16962 }, { "epoch": 0.6043429467196323, "grad_norm": 1.5908662176421573, "learning_rate": 7.146710518422885e-06, "loss": 0.5378, "step": 16963 }, { "epoch": 0.6043785738461265, "grad_norm": 1.057816151756253, "learning_rate": 7.145604570280487e-06, "loss": 0.416, "step": 16964 }, { "epoch": 0.6044142009726206, "grad_norm": 1.278633745357239, "learning_rate": 7.144498660146362e-06, "loss": 0.4396, "step": 16965 }, { "epoch": 0.6044498280991146, "grad_norm": 1.057994505258552, "learning_rate": 7.143392788035229e-06, "loss": 0.4155, "step": 16966 }, { "epoch": 0.6044854552256088, "grad_norm": 1.3015523250871122, "learning_rate": 7.142286953961819e-06, "loss": 0.4499, "step": 16967 }, { "epoch": 0.6045210823521029, "grad_norm": 1.1648891238206942, "learning_rate": 7.141181157940859e-06, "loss": 0.4551, "step": 16968 }, { "epoch": 0.604556709478597, "grad_norm": 1.3548870942353108, "learning_rate": 7.140075399987068e-06, "loss": 0.5212, "step": 16969 }, { "epoch": 0.6045923366050912, "grad_norm": 1.9598663803414826, "learning_rate": 7.138969680115171e-06, "loss": 0.548, "step": 16970 }, { "epoch": 0.6046279637315852, "grad_norm": 1.1695204725633377, "learning_rate": 7.137863998339892e-06, "loss": 0.4189, "step": 16971 }, { "epoch": 0.6046635908580793, "grad_norm": 1.1244808317751345, "learning_rate": 7.136758354675956e-06, "loss": 0.4737, "step": 16972 }, { "epoch": 0.6046992179845735, "grad_norm": 1.599898320058216, "learning_rate": 7.135652749138082e-06, "loss": 0.459, "step": 16973 }, { "epoch": 0.6047348451110676, "grad_norm": 1.2248338636500786, "learning_rate": 7.134547181740998e-06, "loss": 0.4633, "step": 16974 }, { "epoch": 0.6047704722375616, "grad_norm": 1.7678697954681408, "learning_rate": 7.1334416524994175e-06, "loss": 0.5483, "step": 16975 }, { "epoch": 0.6048060993640558, "grad_norm": 1.764335313734823, "learning_rate": 7.132336161428066e-06, "loss": 0.5384, "step": 16976 }, { "epoch": 0.6048417264905499, "grad_norm": 10.519209121800314, "learning_rate": 7.131230708541661e-06, "loss": 0.5377, "step": 16977 }, { "epoch": 0.604877353617044, "grad_norm": 1.1271779235153394, "learning_rate": 7.130125293854925e-06, "loss": 0.377, "step": 16978 }, { "epoch": 0.6049129807435382, "grad_norm": 1.7622665021541903, "learning_rate": 7.12901991738258e-06, "loss": 0.6424, "step": 16979 }, { "epoch": 0.6049486078700322, "grad_norm": 1.1296669501252965, "learning_rate": 7.127914579139338e-06, "loss": 0.4072, "step": 16980 }, { "epoch": 0.6049842349965263, "grad_norm": 1.1752030558579745, "learning_rate": 7.126809279139923e-06, "loss": 0.3997, "step": 16981 }, { "epoch": 0.6050198621230205, "grad_norm": 1.1223351311186978, "learning_rate": 7.125704017399048e-06, "loss": 0.4659, "step": 16982 }, { "epoch": 0.6050554892495146, "grad_norm": 1.2090211255545955, "learning_rate": 7.124598793931436e-06, "loss": 0.4315, "step": 16983 }, { "epoch": 0.6050911163760087, "grad_norm": 1.578991909898338, "learning_rate": 7.123493608751797e-06, "loss": 0.5756, "step": 16984 }, { "epoch": 0.6051267435025028, "grad_norm": 1.1542185023388074, "learning_rate": 7.1223884618748564e-06, "loss": 0.4363, "step": 16985 }, { "epoch": 0.6051623706289969, "grad_norm": 1.1368591227170772, "learning_rate": 7.121283353315321e-06, "loss": 0.4667, "step": 16986 }, { "epoch": 0.605197997755491, "grad_norm": 1.8885537280715523, "learning_rate": 7.12017828308791e-06, "loss": 0.5008, "step": 16987 }, { "epoch": 0.6052336248819852, "grad_norm": 1.1414092662765871, "learning_rate": 7.119073251207338e-06, "loss": 0.471, "step": 16988 }, { "epoch": 0.6052692520084793, "grad_norm": 1.4960601187458784, "learning_rate": 7.117968257688319e-06, "loss": 0.6, "step": 16989 }, { "epoch": 0.6053048791349733, "grad_norm": 1.0514191631503287, "learning_rate": 7.116863302545571e-06, "loss": 0.4339, "step": 16990 }, { "epoch": 0.6053405062614675, "grad_norm": 1.2330709126850297, "learning_rate": 7.115758385793799e-06, "loss": 0.4723, "step": 16991 }, { "epoch": 0.6053761333879616, "grad_norm": 1.2135065778650154, "learning_rate": 7.114653507447723e-06, "loss": 0.4613, "step": 16992 }, { "epoch": 0.6054117605144557, "grad_norm": 1.5130702407168581, "learning_rate": 7.11354866752205e-06, "loss": 0.5529, "step": 16993 }, { "epoch": 0.6054473876409499, "grad_norm": 1.558441860519205, "learning_rate": 7.112443866031497e-06, "loss": 0.4272, "step": 16994 }, { "epoch": 0.6054830147674439, "grad_norm": 1.6551997910684815, "learning_rate": 7.111339102990768e-06, "loss": 0.545, "step": 16995 }, { "epoch": 0.605518641893938, "grad_norm": 1.866357180480609, "learning_rate": 7.110234378414581e-06, "loss": 0.6008, "step": 16996 }, { "epoch": 0.6055542690204322, "grad_norm": 1.67346653474548, "learning_rate": 7.1091296923176416e-06, "loss": 0.4932, "step": 16997 }, { "epoch": 0.6055898961469263, "grad_norm": 1.3779216652604742, "learning_rate": 7.108025044714661e-06, "loss": 0.4496, "step": 16998 }, { "epoch": 0.6056255232734203, "grad_norm": 1.1290066946999309, "learning_rate": 7.106920435620348e-06, "loss": 0.4397, "step": 16999 }, { "epoch": 0.6056611503999145, "grad_norm": 1.5681796728329271, "learning_rate": 7.105815865049414e-06, "loss": 0.5578, "step": 17000 }, { "epoch": 0.6056967775264086, "grad_norm": 1.4547227673173542, "learning_rate": 7.104711333016563e-06, "loss": 0.5169, "step": 17001 }, { "epoch": 0.6057324046529027, "grad_norm": 1.5241409880350596, "learning_rate": 7.103606839536502e-06, "loss": 0.5255, "step": 17002 }, { "epoch": 0.6057680317793969, "grad_norm": 1.1034308653864293, "learning_rate": 7.1025023846239435e-06, "loss": 0.4514, "step": 17003 }, { "epoch": 0.6058036589058909, "grad_norm": 1.1275022478682177, "learning_rate": 7.101397968293588e-06, "loss": 0.4467, "step": 17004 }, { "epoch": 0.605839286032385, "grad_norm": 1.4910723600489497, "learning_rate": 7.10029359056015e-06, "loss": 0.4682, "step": 17005 }, { "epoch": 0.6058749131588792, "grad_norm": 1.1338395438496678, "learning_rate": 7.099189251438326e-06, "loss": 0.4234, "step": 17006 }, { "epoch": 0.6059105402853733, "grad_norm": 1.3715636309378902, "learning_rate": 7.098084950942825e-06, "loss": 0.5357, "step": 17007 }, { "epoch": 0.6059461674118674, "grad_norm": 2.0958258741280527, "learning_rate": 7.096980689088349e-06, "loss": 0.5981, "step": 17008 }, { "epoch": 0.6059817945383615, "grad_norm": 1.4860630694146284, "learning_rate": 7.09587646588961e-06, "loss": 0.4762, "step": 17009 }, { "epoch": 0.6060174216648556, "grad_norm": 1.1471967893985833, "learning_rate": 7.094772281361298e-06, "loss": 0.4323, "step": 17010 }, { "epoch": 0.6060530487913497, "grad_norm": 1.3851459180757362, "learning_rate": 7.0936681355181306e-06, "loss": 0.4951, "step": 17011 }, { "epoch": 0.6060886759178439, "grad_norm": 1.5105480600807444, "learning_rate": 7.0925640283747986e-06, "loss": 0.5973, "step": 17012 }, { "epoch": 0.606124303044338, "grad_norm": 1.4058292078404866, "learning_rate": 7.0914599599460095e-06, "loss": 0.5844, "step": 17013 }, { "epoch": 0.606159930170832, "grad_norm": 1.4932027536456156, "learning_rate": 7.090355930246466e-06, "loss": 0.4951, "step": 17014 }, { "epoch": 0.6061955572973262, "grad_norm": 1.322246290205188, "learning_rate": 7.089251939290865e-06, "loss": 0.466, "step": 17015 }, { "epoch": 0.6062311844238203, "grad_norm": 1.7342025881019651, "learning_rate": 7.088147987093912e-06, "loss": 0.5336, "step": 17016 }, { "epoch": 0.6062668115503144, "grad_norm": 1.4133818585926312, "learning_rate": 7.087044073670301e-06, "loss": 0.5065, "step": 17017 }, { "epoch": 0.6063024386768086, "grad_norm": 1.2030660845772572, "learning_rate": 7.085940199034735e-06, "loss": 0.447, "step": 17018 }, { "epoch": 0.6063380658033026, "grad_norm": 1.2185717537137952, "learning_rate": 7.084836363201911e-06, "loss": 0.4982, "step": 17019 }, { "epoch": 0.6063736929297967, "grad_norm": 1.613361491418873, "learning_rate": 7.083732566186533e-06, "loss": 0.4525, "step": 17020 }, { "epoch": 0.6064093200562909, "grad_norm": 1.220510780112677, "learning_rate": 7.082628808003289e-06, "loss": 0.4627, "step": 17021 }, { "epoch": 0.606444947182785, "grad_norm": 1.16290942183405, "learning_rate": 7.081525088666884e-06, "loss": 0.4567, "step": 17022 }, { "epoch": 0.606480574309279, "grad_norm": 1.4447418185550305, "learning_rate": 7.080421408192011e-06, "loss": 0.5044, "step": 17023 }, { "epoch": 0.6065162014357732, "grad_norm": 1.1844658455798838, "learning_rate": 7.079317766593366e-06, "loss": 0.4128, "step": 17024 }, { "epoch": 0.6065518285622673, "grad_norm": 1.0526159198445182, "learning_rate": 7.078214163885649e-06, "loss": 0.4518, "step": 17025 }, { "epoch": 0.6065874556887614, "grad_norm": 1.7575830597362856, "learning_rate": 7.077110600083552e-06, "loss": 0.4509, "step": 17026 }, { "epoch": 0.6066230828152556, "grad_norm": 1.6288819316035694, "learning_rate": 7.076007075201773e-06, "loss": 0.5224, "step": 17027 }, { "epoch": 0.6066587099417496, "grad_norm": 1.3105096555672666, "learning_rate": 7.074903589255e-06, "loss": 0.5376, "step": 17028 }, { "epoch": 0.6066943370682437, "grad_norm": 1.628726375761471, "learning_rate": 7.0738001422579315e-06, "loss": 0.5374, "step": 17029 }, { "epoch": 0.6067299641947379, "grad_norm": 1.2805519498457458, "learning_rate": 7.072696734225258e-06, "loss": 0.4081, "step": 17030 }, { "epoch": 0.606765591321232, "grad_norm": 1.2686056598794484, "learning_rate": 7.071593365171677e-06, "loss": 0.4648, "step": 17031 }, { "epoch": 0.606801218447726, "grad_norm": 1.4370461142218638, "learning_rate": 7.070490035111874e-06, "loss": 0.5771, "step": 17032 }, { "epoch": 0.6068368455742202, "grad_norm": 1.1282127948062517, "learning_rate": 7.069386744060546e-06, "loss": 0.4096, "step": 17033 }, { "epoch": 0.6068724727007143, "grad_norm": 1.1232514636397506, "learning_rate": 7.068283492032379e-06, "loss": 0.4238, "step": 17034 }, { "epoch": 0.6069080998272084, "grad_norm": 1.421103713726957, "learning_rate": 7.067180279042066e-06, "loss": 0.5726, "step": 17035 }, { "epoch": 0.6069437269537026, "grad_norm": 1.6181170335742587, "learning_rate": 7.0660771051043005e-06, "loss": 0.519, "step": 17036 }, { "epoch": 0.6069793540801967, "grad_norm": 1.9647973098014981, "learning_rate": 7.064973970233766e-06, "loss": 0.5059, "step": 17037 }, { "epoch": 0.6070149812066907, "grad_norm": 1.3588317016414375, "learning_rate": 7.06387087444516e-06, "loss": 0.4432, "step": 17038 }, { "epoch": 0.6070506083331849, "grad_norm": 1.072473517198523, "learning_rate": 7.062767817753161e-06, "loss": 0.3962, "step": 17039 }, { "epoch": 0.607086235459679, "grad_norm": 1.5657656095798616, "learning_rate": 7.061664800172463e-06, "loss": 0.5452, "step": 17040 }, { "epoch": 0.6071218625861731, "grad_norm": 1.865758014800226, "learning_rate": 7.06056182171775e-06, "loss": 0.5708, "step": 17041 }, { "epoch": 0.6071574897126673, "grad_norm": 1.2251527209845376, "learning_rate": 7.059458882403716e-06, "loss": 0.4088, "step": 17042 }, { "epoch": 0.6071931168391613, "grad_norm": 1.1456005705298196, "learning_rate": 7.058355982245038e-06, "loss": 0.4321, "step": 17043 }, { "epoch": 0.6072287439656554, "grad_norm": 1.3088925987877165, "learning_rate": 7.057253121256407e-06, "loss": 0.5115, "step": 17044 }, { "epoch": 0.6072643710921496, "grad_norm": 1.1380931755484853, "learning_rate": 7.056150299452507e-06, "loss": 0.4084, "step": 17045 }, { "epoch": 0.6072999982186437, "grad_norm": 2.6212607314718324, "learning_rate": 7.055047516848026e-06, "loss": 0.601, "step": 17046 }, { "epoch": 0.6073356253451377, "grad_norm": 1.5841209371191576, "learning_rate": 7.053944773457641e-06, "loss": 0.5316, "step": 17047 }, { "epoch": 0.6073712524716319, "grad_norm": 1.1844552796215733, "learning_rate": 7.052842069296042e-06, "loss": 0.414, "step": 17048 }, { "epoch": 0.607406879598126, "grad_norm": 1.689885004019248, "learning_rate": 7.051739404377914e-06, "loss": 0.5379, "step": 17049 }, { "epoch": 0.6074425067246201, "grad_norm": 1.9948925647915654, "learning_rate": 7.050636778717933e-06, "loss": 0.5027, "step": 17050 }, { "epoch": 0.6074781338511143, "grad_norm": 1.499247215785489, "learning_rate": 7.049534192330787e-06, "loss": 0.5135, "step": 17051 }, { "epoch": 0.6075137609776083, "grad_norm": 1.1818978021201312, "learning_rate": 7.048431645231153e-06, "loss": 0.4243, "step": 17052 }, { "epoch": 0.6075493881041024, "grad_norm": 1.6956750892758474, "learning_rate": 7.04732913743372e-06, "loss": 0.5653, "step": 17053 }, { "epoch": 0.6075850152305966, "grad_norm": 1.7867691323337385, "learning_rate": 7.046226668953158e-06, "loss": 0.5204, "step": 17054 }, { "epoch": 0.6076206423570907, "grad_norm": 1.1254104620196053, "learning_rate": 7.045124239804155e-06, "loss": 0.4589, "step": 17055 }, { "epoch": 0.6076562694835848, "grad_norm": 1.4721346222385365, "learning_rate": 7.044021850001386e-06, "loss": 0.5359, "step": 17056 }, { "epoch": 0.6076918966100789, "grad_norm": 1.430816966088108, "learning_rate": 7.042919499559538e-06, "loss": 0.483, "step": 17057 }, { "epoch": 0.607727523736573, "grad_norm": 2.136245694468592, "learning_rate": 7.0418171884932766e-06, "loss": 0.5229, "step": 17058 }, { "epoch": 0.6077631508630671, "grad_norm": 1.7427540144963298, "learning_rate": 7.0407149168172865e-06, "loss": 0.4897, "step": 17059 }, { "epoch": 0.6077987779895613, "grad_norm": 1.1483524502197202, "learning_rate": 7.039612684546251e-06, "loss": 0.4415, "step": 17060 }, { "epoch": 0.6078344051160554, "grad_norm": 1.537717501247493, "learning_rate": 7.0385104916948386e-06, "loss": 0.6014, "step": 17061 }, { "epoch": 0.6078700322425494, "grad_norm": 1.1795695302245464, "learning_rate": 7.03740833827773e-06, "loss": 0.455, "step": 17062 }, { "epoch": 0.6079056593690436, "grad_norm": 1.4240730941791004, "learning_rate": 7.0363062243095994e-06, "loss": 0.4809, "step": 17063 }, { "epoch": 0.6079412864955377, "grad_norm": 1.2632423296320152, "learning_rate": 7.035204149805126e-06, "loss": 0.4966, "step": 17064 }, { "epoch": 0.6079769136220318, "grad_norm": 1.3119193110618246, "learning_rate": 7.034102114778978e-06, "loss": 0.4883, "step": 17065 }, { "epoch": 0.608012540748526, "grad_norm": 1.1810081809405997, "learning_rate": 7.033000119245836e-06, "loss": 0.3919, "step": 17066 }, { "epoch": 0.60804816787502, "grad_norm": 1.8043089300690118, "learning_rate": 7.031898163220367e-06, "loss": 0.5425, "step": 17067 }, { "epoch": 0.6080837950015141, "grad_norm": 1.6515681550279013, "learning_rate": 7.0307962467172555e-06, "loss": 0.5781, "step": 17068 }, { "epoch": 0.6081194221280083, "grad_norm": 1.1663503603718393, "learning_rate": 7.029694369751161e-06, "loss": 0.4515, "step": 17069 }, { "epoch": 0.6081550492545024, "grad_norm": 1.2130008706050817, "learning_rate": 7.028592532336762e-06, "loss": 0.4315, "step": 17070 }, { "epoch": 0.6081906763809964, "grad_norm": 1.2320839688928906, "learning_rate": 7.0274907344887335e-06, "loss": 0.4393, "step": 17071 }, { "epoch": 0.6082263035074906, "grad_norm": 1.2329242429363918, "learning_rate": 7.026388976221742e-06, "loss": 0.4488, "step": 17072 }, { "epoch": 0.6082619306339847, "grad_norm": 1.648123452638743, "learning_rate": 7.0252872575504614e-06, "loss": 0.5879, "step": 17073 }, { "epoch": 0.6082975577604788, "grad_norm": 1.5220700275450378, "learning_rate": 7.0241855784895575e-06, "loss": 0.5738, "step": 17074 }, { "epoch": 0.608333184886973, "grad_norm": 1.2281548071153054, "learning_rate": 7.023083939053706e-06, "loss": 0.4629, "step": 17075 }, { "epoch": 0.608368812013467, "grad_norm": 1.6374630226857143, "learning_rate": 7.0219823392575694e-06, "loss": 0.6082, "step": 17076 }, { "epoch": 0.6084044391399611, "grad_norm": 1.4698704637691944, "learning_rate": 7.020880779115821e-06, "loss": 0.5806, "step": 17077 }, { "epoch": 0.6084400662664553, "grad_norm": 1.193255112578291, "learning_rate": 7.019779258643127e-06, "loss": 0.4402, "step": 17078 }, { "epoch": 0.6084756933929494, "grad_norm": 1.4524078695728893, "learning_rate": 7.018677777854158e-06, "loss": 0.5606, "step": 17079 }, { "epoch": 0.6085113205194435, "grad_norm": 1.4778962289963524, "learning_rate": 7.017576336763574e-06, "loss": 0.5151, "step": 17080 }, { "epoch": 0.6085469476459376, "grad_norm": 1.0954581299207347, "learning_rate": 7.016474935386049e-06, "loss": 0.4356, "step": 17081 }, { "epoch": 0.6085825747724317, "grad_norm": 1.3767771519225314, "learning_rate": 7.015373573736241e-06, "loss": 0.5475, "step": 17082 }, { "epoch": 0.6086182018989258, "grad_norm": 1.3925822623963786, "learning_rate": 7.014272251828822e-06, "loss": 0.608, "step": 17083 }, { "epoch": 0.60865382902542, "grad_norm": 1.284352988119911, "learning_rate": 7.013170969678457e-06, "loss": 0.4673, "step": 17084 }, { "epoch": 0.608689456151914, "grad_norm": 1.121110796444816, "learning_rate": 7.012069727299805e-06, "loss": 0.4595, "step": 17085 }, { "epoch": 0.6087250832784081, "grad_norm": 1.4281237637207482, "learning_rate": 7.010968524707538e-06, "loss": 0.4416, "step": 17086 }, { "epoch": 0.6087607104049023, "grad_norm": 1.226799249600125, "learning_rate": 7.009867361916309e-06, "loss": 0.4668, "step": 17087 }, { "epoch": 0.6087963375313964, "grad_norm": 1.316814309815421, "learning_rate": 7.00876623894079e-06, "loss": 0.4472, "step": 17088 }, { "epoch": 0.6088319646578905, "grad_norm": 1.040546138135555, "learning_rate": 7.007665155795636e-06, "loss": 0.4018, "step": 17089 }, { "epoch": 0.6088675917843847, "grad_norm": 1.1784202062951747, "learning_rate": 7.006564112495515e-06, "loss": 0.4531, "step": 17090 }, { "epoch": 0.6089032189108787, "grad_norm": 1.467810296098259, "learning_rate": 7.005463109055084e-06, "loss": 0.6147, "step": 17091 }, { "epoch": 0.6089388460373728, "grad_norm": 1.2351132889630128, "learning_rate": 7.004362145489004e-06, "loss": 0.4339, "step": 17092 }, { "epoch": 0.608974473163867, "grad_norm": 1.127044773293163, "learning_rate": 7.003261221811935e-06, "loss": 0.4227, "step": 17093 }, { "epoch": 0.6090101002903611, "grad_norm": 1.212029230237141, "learning_rate": 7.002160338038537e-06, "loss": 0.5053, "step": 17094 }, { "epoch": 0.6090457274168551, "grad_norm": 1.1905099247574542, "learning_rate": 7.0010594941834745e-06, "loss": 0.3959, "step": 17095 }, { "epoch": 0.6090813545433493, "grad_norm": 1.153752124577388, "learning_rate": 6.999958690261396e-06, "loss": 0.4038, "step": 17096 }, { "epoch": 0.6091169816698434, "grad_norm": 1.5568750449857989, "learning_rate": 6.998857926286966e-06, "loss": 0.5326, "step": 17097 }, { "epoch": 0.6091526087963375, "grad_norm": 1.1475719681140735, "learning_rate": 6.99775720227484e-06, "loss": 0.4469, "step": 17098 }, { "epoch": 0.6091882359228317, "grad_norm": 1.7704258725301854, "learning_rate": 6.996656518239676e-06, "loss": 0.5474, "step": 17099 }, { "epoch": 0.6092238630493257, "grad_norm": 1.2757272320986448, "learning_rate": 6.995555874196128e-06, "loss": 0.5101, "step": 17100 }, { "epoch": 0.6092594901758198, "grad_norm": 1.3076662545082287, "learning_rate": 6.994455270158858e-06, "loss": 0.4652, "step": 17101 }, { "epoch": 0.609295117302314, "grad_norm": 1.1942243907270822, "learning_rate": 6.993354706142512e-06, "loss": 0.4794, "step": 17102 }, { "epoch": 0.6093307444288081, "grad_norm": 1.1031380901704575, "learning_rate": 6.992254182161753e-06, "loss": 0.4163, "step": 17103 }, { "epoch": 0.6093663715553022, "grad_norm": 1.1730882376424328, "learning_rate": 6.991153698231228e-06, "loss": 0.4624, "step": 17104 }, { "epoch": 0.6094019986817963, "grad_norm": 1.4752157399191148, "learning_rate": 6.990053254365597e-06, "loss": 0.6018, "step": 17105 }, { "epoch": 0.6094376258082904, "grad_norm": 1.0867429108892812, "learning_rate": 6.988952850579512e-06, "loss": 0.3965, "step": 17106 }, { "epoch": 0.6094732529347845, "grad_norm": 1.1384027337793463, "learning_rate": 6.987852486887622e-06, "loss": 0.4766, "step": 17107 }, { "epoch": 0.6095088800612787, "grad_norm": 1.6035972134142256, "learning_rate": 6.986752163304582e-06, "loss": 0.5407, "step": 17108 }, { "epoch": 0.6095445071877728, "grad_norm": 1.0553608463069661, "learning_rate": 6.985651879845044e-06, "loss": 0.4358, "step": 17109 }, { "epoch": 0.6095801343142668, "grad_norm": 1.6190248534111702, "learning_rate": 6.98455163652366e-06, "loss": 0.5567, "step": 17110 }, { "epoch": 0.609615761440761, "grad_norm": 1.2715072100634914, "learning_rate": 6.983451433355072e-06, "loss": 0.4602, "step": 17111 }, { "epoch": 0.6096513885672551, "grad_norm": 1.2627671114034513, "learning_rate": 6.9823512703539445e-06, "loss": 0.4188, "step": 17112 }, { "epoch": 0.6096870156937492, "grad_norm": 1.935402836981625, "learning_rate": 6.9812511475349135e-06, "loss": 0.5697, "step": 17113 }, { "epoch": 0.6097226428202434, "grad_norm": 1.2202740424529035, "learning_rate": 6.980151064912636e-06, "loss": 0.4104, "step": 17114 }, { "epoch": 0.6097582699467374, "grad_norm": 1.6047645211342116, "learning_rate": 6.979051022501755e-06, "loss": 0.3575, "step": 17115 }, { "epoch": 0.6097938970732315, "grad_norm": 1.2738923701852916, "learning_rate": 6.977951020316922e-06, "loss": 0.4475, "step": 17116 }, { "epoch": 0.6098295241997257, "grad_norm": 1.195296922339401, "learning_rate": 6.976851058372789e-06, "loss": 0.4465, "step": 17117 }, { "epoch": 0.6098651513262198, "grad_norm": 1.0851955539367888, "learning_rate": 6.97575113668399e-06, "loss": 0.3868, "step": 17118 }, { "epoch": 0.6099007784527138, "grad_norm": 1.5437812215349513, "learning_rate": 6.974651255265183e-06, "loss": 0.5047, "step": 17119 }, { "epoch": 0.609936405579208, "grad_norm": 1.2156246571811526, "learning_rate": 6.973551414131005e-06, "loss": 0.4943, "step": 17120 }, { "epoch": 0.6099720327057021, "grad_norm": 1.1249666047079216, "learning_rate": 6.972451613296112e-06, "loss": 0.4451, "step": 17121 }, { "epoch": 0.6100076598321962, "grad_norm": 1.301483077864537, "learning_rate": 6.971351852775137e-06, "loss": 0.5346, "step": 17122 }, { "epoch": 0.6100432869586904, "grad_norm": 1.4502802402496946, "learning_rate": 6.970252132582729e-06, "loss": 0.5073, "step": 17123 }, { "epoch": 0.6100789140851844, "grad_norm": 1.1034336456117964, "learning_rate": 6.969152452733532e-06, "loss": 0.4155, "step": 17124 }, { "epoch": 0.6101145412116785, "grad_norm": 1.0625848936402331, "learning_rate": 6.968052813242189e-06, "loss": 0.415, "step": 17125 }, { "epoch": 0.6101501683381727, "grad_norm": 1.4675157924115942, "learning_rate": 6.96695321412334e-06, "loss": 0.5114, "step": 17126 }, { "epoch": 0.6101857954646668, "grad_norm": 1.184158820060239, "learning_rate": 6.965853655391633e-06, "loss": 0.4312, "step": 17127 }, { "epoch": 0.6102214225911609, "grad_norm": 1.1723352386560284, "learning_rate": 6.964754137061702e-06, "loss": 0.4248, "step": 17128 }, { "epoch": 0.610257049717655, "grad_norm": 1.6481468149246246, "learning_rate": 6.963654659148192e-06, "loss": 0.5146, "step": 17129 }, { "epoch": 0.6102926768441491, "grad_norm": 1.0493453151845862, "learning_rate": 6.962555221665742e-06, "loss": 0.4269, "step": 17130 }, { "epoch": 0.6103283039706432, "grad_norm": 1.0745741772448705, "learning_rate": 6.961455824628992e-06, "loss": 0.3997, "step": 17131 }, { "epoch": 0.6103639310971374, "grad_norm": 1.6577494846902294, "learning_rate": 6.960356468052585e-06, "loss": 0.5467, "step": 17132 }, { "epoch": 0.6103995582236315, "grad_norm": 1.5816881916858634, "learning_rate": 6.959257151951153e-06, "loss": 0.5922, "step": 17133 }, { "epoch": 0.6104351853501255, "grad_norm": 1.6024715389628799, "learning_rate": 6.958157876339337e-06, "loss": 0.5878, "step": 17134 }, { "epoch": 0.6104708124766197, "grad_norm": 1.5149586337267178, "learning_rate": 6.957058641231775e-06, "loss": 0.4514, "step": 17135 }, { "epoch": 0.6105064396031138, "grad_norm": 1.7415750265808407, "learning_rate": 6.955959446643107e-06, "loss": 0.513, "step": 17136 }, { "epoch": 0.6105420667296079, "grad_norm": 1.125529670192037, "learning_rate": 6.954860292587964e-06, "loss": 0.4673, "step": 17137 }, { "epoch": 0.610577693856102, "grad_norm": 1.5717868775903157, "learning_rate": 6.953761179080987e-06, "loss": 0.5618, "step": 17138 }, { "epoch": 0.6106133209825961, "grad_norm": 1.482303340775927, "learning_rate": 6.952662106136806e-06, "loss": 0.4209, "step": 17139 }, { "epoch": 0.6106489481090902, "grad_norm": 1.2560101767614547, "learning_rate": 6.95156307377006e-06, "loss": 0.421, "step": 17140 }, { "epoch": 0.6106845752355844, "grad_norm": 1.1148428395472394, "learning_rate": 6.9504640819953825e-06, "loss": 0.3924, "step": 17141 }, { "epoch": 0.6107202023620785, "grad_norm": 1.2935038519923003, "learning_rate": 6.949365130827407e-06, "loss": 0.3985, "step": 17142 }, { "epoch": 0.6107558294885727, "grad_norm": 2.633370755511627, "learning_rate": 6.94826622028077e-06, "loss": 0.5279, "step": 17143 }, { "epoch": 0.6107914566150667, "grad_norm": 1.593026727415013, "learning_rate": 6.947167350370098e-06, "loss": 0.5986, "step": 17144 }, { "epoch": 0.6108270837415608, "grad_norm": 1.2627097653430792, "learning_rate": 6.946068521110028e-06, "loss": 0.4219, "step": 17145 }, { "epoch": 0.610862710868055, "grad_norm": 1.439031285694843, "learning_rate": 6.944969732515188e-06, "loss": 0.4229, "step": 17146 }, { "epoch": 0.6108983379945491, "grad_norm": 2.129583546028388, "learning_rate": 6.943870984600216e-06, "loss": 0.5293, "step": 17147 }, { "epoch": 0.6109339651210431, "grad_norm": 1.676414697427719, "learning_rate": 6.9427722773797325e-06, "loss": 0.5472, "step": 17148 }, { "epoch": 0.6109695922475373, "grad_norm": 2.013392345912391, "learning_rate": 6.941673610868378e-06, "loss": 0.5426, "step": 17149 }, { "epoch": 0.6110052193740314, "grad_norm": 1.7194156315699212, "learning_rate": 6.9405749850807725e-06, "loss": 0.611, "step": 17150 }, { "epoch": 0.6110408465005255, "grad_norm": 1.2822751715810932, "learning_rate": 6.9394764000315504e-06, "loss": 0.5007, "step": 17151 }, { "epoch": 0.6110764736270197, "grad_norm": 1.8764040634885697, "learning_rate": 6.938377855735341e-06, "loss": 0.5091, "step": 17152 }, { "epoch": 0.6111121007535137, "grad_norm": 1.1106671064578728, "learning_rate": 6.9372793522067685e-06, "loss": 0.4086, "step": 17153 }, { "epoch": 0.6111477278800078, "grad_norm": 1.217645587269542, "learning_rate": 6.936180889460467e-06, "loss": 0.4671, "step": 17154 }, { "epoch": 0.611183355006502, "grad_norm": 1.5634961843237305, "learning_rate": 6.935082467511053e-06, "loss": 0.4406, "step": 17155 }, { "epoch": 0.6112189821329961, "grad_norm": 1.168300029935381, "learning_rate": 6.93398408637316e-06, "loss": 0.4671, "step": 17156 }, { "epoch": 0.6112546092594902, "grad_norm": 1.1721709854978652, "learning_rate": 6.932885746061411e-06, "loss": 0.4336, "step": 17157 }, { "epoch": 0.6112902363859843, "grad_norm": 1.4780581216530442, "learning_rate": 6.931787446590435e-06, "loss": 0.5757, "step": 17158 }, { "epoch": 0.6113258635124784, "grad_norm": 1.600842019049433, "learning_rate": 6.9306891879748505e-06, "loss": 0.5486, "step": 17159 }, { "epoch": 0.6113614906389725, "grad_norm": 1.5087598772175077, "learning_rate": 6.929590970229287e-06, "loss": 0.5281, "step": 17160 }, { "epoch": 0.6113971177654667, "grad_norm": 1.3678738527091834, "learning_rate": 6.928492793368365e-06, "loss": 0.534, "step": 17161 }, { "epoch": 0.6114327448919608, "grad_norm": 1.3523513784130663, "learning_rate": 6.927394657406705e-06, "loss": 0.4312, "step": 17162 }, { "epoch": 0.6114683720184548, "grad_norm": 1.3920160784810596, "learning_rate": 6.926296562358936e-06, "loss": 0.5878, "step": 17163 }, { "epoch": 0.611503999144949, "grad_norm": 1.5325534570469501, "learning_rate": 6.925198508239677e-06, "loss": 0.5482, "step": 17164 }, { "epoch": 0.6115396262714431, "grad_norm": 1.0659775847625392, "learning_rate": 6.9241004950635505e-06, "loss": 0.3902, "step": 17165 }, { "epoch": 0.6115752533979372, "grad_norm": 1.7896778485694984, "learning_rate": 6.923002522845172e-06, "loss": 0.529, "step": 17166 }, { "epoch": 0.6116108805244314, "grad_norm": 1.7962326049582813, "learning_rate": 6.921904591599169e-06, "loss": 0.5394, "step": 17167 }, { "epoch": 0.6116465076509254, "grad_norm": 1.5594775095666247, "learning_rate": 6.920806701340154e-06, "loss": 0.5888, "step": 17168 }, { "epoch": 0.6116821347774195, "grad_norm": 1.1987900253454375, "learning_rate": 6.919708852082756e-06, "loss": 0.3915, "step": 17169 }, { "epoch": 0.6117177619039137, "grad_norm": 1.212284324992072, "learning_rate": 6.918611043841582e-06, "loss": 0.4882, "step": 17170 }, { "epoch": 0.6117533890304078, "grad_norm": 1.645437522072354, "learning_rate": 6.917513276631258e-06, "loss": 0.5703, "step": 17171 }, { "epoch": 0.6117890161569018, "grad_norm": 1.2033385545886046, "learning_rate": 6.916415550466398e-06, "loss": 0.4582, "step": 17172 }, { "epoch": 0.611824643283396, "grad_norm": 1.20531408163443, "learning_rate": 6.915317865361623e-06, "loss": 0.5079, "step": 17173 }, { "epoch": 0.6118602704098901, "grad_norm": 1.1526621255205582, "learning_rate": 6.9142202213315425e-06, "loss": 0.463, "step": 17174 }, { "epoch": 0.6118958975363842, "grad_norm": 1.2514785095108474, "learning_rate": 6.913122618390777e-06, "loss": 0.4586, "step": 17175 }, { "epoch": 0.6119315246628784, "grad_norm": 1.7986148474869663, "learning_rate": 6.912025056553946e-06, "loss": 0.5578, "step": 17176 }, { "epoch": 0.6119671517893724, "grad_norm": 1.8077940959921364, "learning_rate": 6.910927535835656e-06, "loss": 0.5098, "step": 17177 }, { "epoch": 0.6120027789158665, "grad_norm": 1.694052534376222, "learning_rate": 6.909830056250527e-06, "loss": 0.576, "step": 17178 }, { "epoch": 0.6120384060423607, "grad_norm": 1.1510082169880262, "learning_rate": 6.908732617813168e-06, "loss": 0.4467, "step": 17179 }, { "epoch": 0.6120740331688548, "grad_norm": 1.6462524820421254, "learning_rate": 6.9076352205382005e-06, "loss": 0.5906, "step": 17180 }, { "epoch": 0.6121096602953489, "grad_norm": 1.5752025459447225, "learning_rate": 6.906537864440226e-06, "loss": 0.5593, "step": 17181 }, { "epoch": 0.612145287421843, "grad_norm": 1.662144067640281, "learning_rate": 6.905440549533864e-06, "loss": 0.5338, "step": 17182 }, { "epoch": 0.6121809145483371, "grad_norm": 1.1064540579170765, "learning_rate": 6.904343275833722e-06, "loss": 0.4037, "step": 17183 }, { "epoch": 0.6122165416748312, "grad_norm": 1.5305872024880898, "learning_rate": 6.903246043354418e-06, "loss": 0.5108, "step": 17184 }, { "epoch": 0.6122521688013254, "grad_norm": 1.1819509526982668, "learning_rate": 6.902148852110553e-06, "loss": 0.4123, "step": 17185 }, { "epoch": 0.6122877959278195, "grad_norm": 1.3419733139112051, "learning_rate": 6.901051702116739e-06, "loss": 0.5073, "step": 17186 }, { "epoch": 0.6123234230543135, "grad_norm": 1.6407014168572926, "learning_rate": 6.899954593387594e-06, "loss": 0.5772, "step": 17187 }, { "epoch": 0.6123590501808077, "grad_norm": 1.133494718182769, "learning_rate": 6.8988575259377156e-06, "loss": 0.4812, "step": 17188 }, { "epoch": 0.6123946773073018, "grad_norm": 1.6588966787578643, "learning_rate": 6.897760499781719e-06, "loss": 0.53, "step": 17189 }, { "epoch": 0.6124303044337959, "grad_norm": 1.266112215440408, "learning_rate": 6.896663514934208e-06, "loss": 0.4246, "step": 17190 }, { "epoch": 0.6124659315602901, "grad_norm": 1.1528642098100224, "learning_rate": 6.895566571409795e-06, "loss": 0.4344, "step": 17191 }, { "epoch": 0.6125015586867841, "grad_norm": 1.0388103460473612, "learning_rate": 6.89446966922308e-06, "loss": 0.4347, "step": 17192 }, { "epoch": 0.6125371858132782, "grad_norm": 1.148509505697197, "learning_rate": 6.893372808388674e-06, "loss": 0.4467, "step": 17193 }, { "epoch": 0.6125728129397724, "grad_norm": 1.1555804168981352, "learning_rate": 6.892275988921178e-06, "loss": 0.4367, "step": 17194 }, { "epoch": 0.6126084400662665, "grad_norm": 1.7578211848224246, "learning_rate": 6.891179210835202e-06, "loss": 0.4707, "step": 17195 }, { "epoch": 0.6126440671927605, "grad_norm": 1.1905348456845324, "learning_rate": 6.890082474145345e-06, "loss": 0.4287, "step": 17196 }, { "epoch": 0.6126796943192547, "grad_norm": 1.9024704004020576, "learning_rate": 6.888985778866214e-06, "loss": 0.5243, "step": 17197 }, { "epoch": 0.6127153214457488, "grad_norm": 1.2306138251849221, "learning_rate": 6.887889125012413e-06, "loss": 0.4745, "step": 17198 }, { "epoch": 0.6127509485722429, "grad_norm": 1.1779635415895895, "learning_rate": 6.886792512598542e-06, "loss": 0.4552, "step": 17199 }, { "epoch": 0.6127865756987371, "grad_norm": 1.1776507634633833, "learning_rate": 6.885695941639208e-06, "loss": 0.4619, "step": 17200 }, { "epoch": 0.6128222028252311, "grad_norm": 2.4040634503089673, "learning_rate": 6.884599412149004e-06, "loss": 0.4505, "step": 17201 }, { "epoch": 0.6128578299517252, "grad_norm": 1.8262732469391145, "learning_rate": 6.883502924142543e-06, "loss": 0.4819, "step": 17202 }, { "epoch": 0.6128934570782194, "grad_norm": 1.5349944195336294, "learning_rate": 6.882406477634413e-06, "loss": 0.5188, "step": 17203 }, { "epoch": 0.6129290842047135, "grad_norm": 1.251949379105768, "learning_rate": 6.881310072639224e-06, "loss": 0.4678, "step": 17204 }, { "epoch": 0.6129647113312076, "grad_norm": 1.6783362270619597, "learning_rate": 6.8802137091715685e-06, "loss": 0.5432, "step": 17205 }, { "epoch": 0.6130003384577017, "grad_norm": 1.1946719979748202, "learning_rate": 6.879117387246052e-06, "loss": 0.447, "step": 17206 }, { "epoch": 0.6130359655841958, "grad_norm": 2.2166487038823814, "learning_rate": 6.878021106877264e-06, "loss": 0.5074, "step": 17207 }, { "epoch": 0.6130715927106899, "grad_norm": 1.5836935684121747, "learning_rate": 6.876924868079811e-06, "loss": 0.5219, "step": 17208 }, { "epoch": 0.6131072198371841, "grad_norm": 2.2599788808628465, "learning_rate": 6.875828670868282e-06, "loss": 0.5686, "step": 17209 }, { "epoch": 0.6131428469636782, "grad_norm": 1.2430503542843832, "learning_rate": 6.8747325152572785e-06, "loss": 0.424, "step": 17210 }, { "epoch": 0.6131784740901722, "grad_norm": 1.2979572010932436, "learning_rate": 6.873636401261401e-06, "loss": 0.4671, "step": 17211 }, { "epoch": 0.6132141012166664, "grad_norm": 1.0969978617541822, "learning_rate": 6.872540328895233e-06, "loss": 0.4803, "step": 17212 }, { "epoch": 0.6132497283431605, "grad_norm": 1.804017398914156, "learning_rate": 6.871444298173382e-06, "loss": 0.5092, "step": 17213 }, { "epoch": 0.6132853554696546, "grad_norm": 1.5006564327001854, "learning_rate": 6.870348309110435e-06, "loss": 0.5331, "step": 17214 }, { "epoch": 0.6133209825961488, "grad_norm": 1.2995330349175442, "learning_rate": 6.869252361720988e-06, "loss": 0.4297, "step": 17215 }, { "epoch": 0.6133566097226428, "grad_norm": 1.2147552958060581, "learning_rate": 6.868156456019633e-06, "loss": 0.4003, "step": 17216 }, { "epoch": 0.6133922368491369, "grad_norm": 1.2484586443453631, "learning_rate": 6.867060592020967e-06, "loss": 0.4837, "step": 17217 }, { "epoch": 0.6134278639756311, "grad_norm": 1.5574997742206473, "learning_rate": 6.8659647697395745e-06, "loss": 0.4551, "step": 17218 }, { "epoch": 0.6134634911021252, "grad_norm": 1.5963139278944416, "learning_rate": 6.864868989190056e-06, "loss": 0.5352, "step": 17219 }, { "epoch": 0.6134991182286192, "grad_norm": 1.4099375481982195, "learning_rate": 6.863773250386994e-06, "loss": 0.4351, "step": 17220 }, { "epoch": 0.6135347453551134, "grad_norm": 1.2458375193352602, "learning_rate": 6.862677553344984e-06, "loss": 0.4264, "step": 17221 }, { "epoch": 0.6135703724816075, "grad_norm": 1.5509875104870525, "learning_rate": 6.86158189807862e-06, "loss": 0.5184, "step": 17222 }, { "epoch": 0.6136059996081016, "grad_norm": 1.3026947824934412, "learning_rate": 6.860486284602479e-06, "loss": 0.4362, "step": 17223 }, { "epoch": 0.6136416267345958, "grad_norm": 1.4252736126147463, "learning_rate": 6.859390712931165e-06, "loss": 0.5068, "step": 17224 }, { "epoch": 0.6136772538610898, "grad_norm": 1.296400525691255, "learning_rate": 6.858295183079254e-06, "loss": 0.5534, "step": 17225 }, { "epoch": 0.6137128809875839, "grad_norm": 1.0956214288295678, "learning_rate": 6.857199695061341e-06, "loss": 0.4132, "step": 17226 }, { "epoch": 0.6137485081140781, "grad_norm": 1.5449427168781467, "learning_rate": 6.856104248892008e-06, "loss": 0.5169, "step": 17227 }, { "epoch": 0.6137841352405722, "grad_norm": 1.2331937228705712, "learning_rate": 6.855008844585849e-06, "loss": 0.4365, "step": 17228 }, { "epoch": 0.6138197623670663, "grad_norm": 1.1787292499214537, "learning_rate": 6.8539134821574415e-06, "loss": 0.4155, "step": 17229 }, { "epoch": 0.6138553894935604, "grad_norm": 1.140532242134684, "learning_rate": 6.852818161621377e-06, "loss": 0.469, "step": 17230 }, { "epoch": 0.6138910166200545, "grad_norm": 1.366564586351526, "learning_rate": 6.851722882992237e-06, "loss": 0.5253, "step": 17231 }, { "epoch": 0.6139266437465486, "grad_norm": 1.230139264792598, "learning_rate": 6.850627646284606e-06, "loss": 0.4147, "step": 17232 }, { "epoch": 0.6139622708730428, "grad_norm": 1.9863199484228449, "learning_rate": 6.8495324515130744e-06, "loss": 0.5383, "step": 17233 }, { "epoch": 0.6139978979995369, "grad_norm": 1.7507419165249802, "learning_rate": 6.848437298692216e-06, "loss": 0.6182, "step": 17234 }, { "epoch": 0.6140335251260309, "grad_norm": 1.2025614497481538, "learning_rate": 6.847342187836619e-06, "loss": 0.4442, "step": 17235 }, { "epoch": 0.6140691522525251, "grad_norm": 1.2354462912955233, "learning_rate": 6.846247118960863e-06, "loss": 0.5811, "step": 17236 }, { "epoch": 0.6141047793790192, "grad_norm": 1.9252881975052158, "learning_rate": 6.8451520920795326e-06, "loss": 0.5496, "step": 17237 }, { "epoch": 0.6141404065055133, "grad_norm": 1.289141457083793, "learning_rate": 6.844057107207205e-06, "loss": 0.4254, "step": 17238 }, { "epoch": 0.6141760336320075, "grad_norm": 1.1065044190312279, "learning_rate": 6.842962164358467e-06, "loss": 0.3844, "step": 17239 }, { "epoch": 0.6142116607585015, "grad_norm": 2.4820811236060765, "learning_rate": 6.84186726354789e-06, "loss": 0.455, "step": 17240 }, { "epoch": 0.6142472878849956, "grad_norm": 2.118832800532317, "learning_rate": 6.8407724047900616e-06, "loss": 0.5096, "step": 17241 }, { "epoch": 0.6142829150114898, "grad_norm": 1.2497180995780748, "learning_rate": 6.839677588099554e-06, "loss": 0.4204, "step": 17242 }, { "epoch": 0.6143185421379839, "grad_norm": 2.0171405731293013, "learning_rate": 6.838582813490947e-06, "loss": 0.4689, "step": 17243 }, { "epoch": 0.614354169264478, "grad_norm": 1.206210038123571, "learning_rate": 6.837488080978825e-06, "loss": 0.4314, "step": 17244 }, { "epoch": 0.6143897963909721, "grad_norm": 1.1971911658225507, "learning_rate": 6.836393390577755e-06, "loss": 0.4774, "step": 17245 }, { "epoch": 0.6144254235174662, "grad_norm": 1.1146163865652647, "learning_rate": 6.83529874230232e-06, "loss": 0.4526, "step": 17246 }, { "epoch": 0.6144610506439603, "grad_norm": 1.1365964260616561, "learning_rate": 6.834204136167093e-06, "loss": 0.4235, "step": 17247 }, { "epoch": 0.6144966777704545, "grad_norm": 1.5232619566590524, "learning_rate": 6.833109572186653e-06, "loss": 0.529, "step": 17248 }, { "epoch": 0.6145323048969485, "grad_norm": 1.7334064809932168, "learning_rate": 6.832015050375569e-06, "loss": 0.4935, "step": 17249 }, { "epoch": 0.6145679320234426, "grad_norm": 1.9416165170668938, "learning_rate": 6.830920570748423e-06, "loss": 0.5421, "step": 17250 }, { "epoch": 0.6146035591499368, "grad_norm": 1.1407519876044205, "learning_rate": 6.829826133319781e-06, "loss": 0.4765, "step": 17251 }, { "epoch": 0.6146391862764309, "grad_norm": 1.8314264261167277, "learning_rate": 6.8287317381042216e-06, "loss": 0.6127, "step": 17252 }, { "epoch": 0.614674813402925, "grad_norm": 1.2099525225664391, "learning_rate": 6.827637385116315e-06, "loss": 0.4587, "step": 17253 }, { "epoch": 0.6147104405294191, "grad_norm": 1.6648667763570473, "learning_rate": 6.826543074370637e-06, "loss": 0.5983, "step": 17254 }, { "epoch": 0.6147460676559132, "grad_norm": 1.3614603176901676, "learning_rate": 6.825448805881753e-06, "loss": 0.5863, "step": 17255 }, { "epoch": 0.6147816947824073, "grad_norm": 1.5754951791348417, "learning_rate": 6.824354579664236e-06, "loss": 0.5345, "step": 17256 }, { "epoch": 0.6148173219089015, "grad_norm": 1.2118050366172874, "learning_rate": 6.823260395732661e-06, "loss": 0.4326, "step": 17257 }, { "epoch": 0.6148529490353956, "grad_norm": 1.5986854045008139, "learning_rate": 6.82216625410159e-06, "loss": 0.5455, "step": 17258 }, { "epoch": 0.6148885761618896, "grad_norm": 1.9898116319141868, "learning_rate": 6.821072154785602e-06, "loss": 0.5154, "step": 17259 }, { "epoch": 0.6149242032883838, "grad_norm": 1.2108360451510203, "learning_rate": 6.819978097799256e-06, "loss": 0.4512, "step": 17260 }, { "epoch": 0.6149598304148779, "grad_norm": 1.6387466417527217, "learning_rate": 6.818884083157125e-06, "loss": 0.5336, "step": 17261 }, { "epoch": 0.614995457541372, "grad_norm": 1.272060537738071, "learning_rate": 6.817790110873774e-06, "loss": 0.428, "step": 17262 }, { "epoch": 0.6150310846678662, "grad_norm": 1.8509094011340212, "learning_rate": 6.816696180963776e-06, "loss": 0.4785, "step": 17263 }, { "epoch": 0.6150667117943602, "grad_norm": 1.338452447951384, "learning_rate": 6.81560229344169e-06, "loss": 0.4793, "step": 17264 }, { "epoch": 0.6151023389208543, "grad_norm": 2.5733041717131626, "learning_rate": 6.81450844832209e-06, "loss": 0.5377, "step": 17265 }, { "epoch": 0.6151379660473485, "grad_norm": 1.1334256513009795, "learning_rate": 6.813414645619531e-06, "loss": 0.4265, "step": 17266 }, { "epoch": 0.6151735931738426, "grad_norm": 1.6142163546252652, "learning_rate": 6.812320885348585e-06, "loss": 0.5355, "step": 17267 }, { "epoch": 0.6152092203003366, "grad_norm": 1.3500395794261666, "learning_rate": 6.8112271675238154e-06, "loss": 0.4136, "step": 17268 }, { "epoch": 0.6152448474268308, "grad_norm": 1.614015702675713, "learning_rate": 6.810133492159783e-06, "loss": 0.5448, "step": 17269 }, { "epoch": 0.6152804745533249, "grad_norm": 1.7876622181179924, "learning_rate": 6.8090398592710575e-06, "loss": 0.5535, "step": 17270 }, { "epoch": 0.615316101679819, "grad_norm": 1.950037037095905, "learning_rate": 6.807946268872192e-06, "loss": 0.5584, "step": 17271 }, { "epoch": 0.6153517288063132, "grad_norm": 1.1347129521243702, "learning_rate": 6.806852720977754e-06, "loss": 0.4203, "step": 17272 }, { "epoch": 0.6153873559328072, "grad_norm": 1.5990841507954507, "learning_rate": 6.805759215602303e-06, "loss": 0.5736, "step": 17273 }, { "epoch": 0.6154229830593013, "grad_norm": 1.5466196471040146, "learning_rate": 6.804665752760406e-06, "loss": 0.5965, "step": 17274 }, { "epoch": 0.6154586101857955, "grad_norm": 1.4450684803603155, "learning_rate": 6.8035723324666105e-06, "loss": 0.5275, "step": 17275 }, { "epoch": 0.6154942373122896, "grad_norm": 2.191020596785154, "learning_rate": 6.80247895473549e-06, "loss": 0.5637, "step": 17276 }, { "epoch": 0.6155298644387837, "grad_norm": 1.1325864989451528, "learning_rate": 6.801385619581592e-06, "loss": 0.3927, "step": 17277 }, { "epoch": 0.6155654915652778, "grad_norm": 1.4842993579430743, "learning_rate": 6.800292327019481e-06, "loss": 0.5369, "step": 17278 }, { "epoch": 0.6156011186917719, "grad_norm": 1.769288222757495, "learning_rate": 6.799199077063716e-06, "loss": 0.5772, "step": 17279 }, { "epoch": 0.615636745818266, "grad_norm": 1.3215797409786698, "learning_rate": 6.798105869728851e-06, "loss": 0.4808, "step": 17280 }, { "epoch": 0.6156723729447602, "grad_norm": 1.1254502525767691, "learning_rate": 6.797012705029447e-06, "loss": 0.4555, "step": 17281 }, { "epoch": 0.6157080000712543, "grad_norm": 1.6451809032652376, "learning_rate": 6.795919582980055e-06, "loss": 0.5369, "step": 17282 }, { "epoch": 0.6157436271977483, "grad_norm": 1.3577207808739193, "learning_rate": 6.794826503595235e-06, "loss": 0.4912, "step": 17283 }, { "epoch": 0.6157792543242425, "grad_norm": 1.2907987771115172, "learning_rate": 6.793733466889539e-06, "loss": 0.4833, "step": 17284 }, { "epoch": 0.6158148814507366, "grad_norm": 1.6122906471508427, "learning_rate": 6.7926404728775255e-06, "loss": 0.5645, "step": 17285 }, { "epoch": 0.6158505085772307, "grad_norm": 1.7902975164170758, "learning_rate": 6.791547521573743e-06, "loss": 0.562, "step": 17286 }, { "epoch": 0.6158861357037249, "grad_norm": 1.1218093025520637, "learning_rate": 6.790454612992749e-06, "loss": 0.3976, "step": 17287 }, { "epoch": 0.6159217628302189, "grad_norm": 1.2551998718588966, "learning_rate": 6.789361747149092e-06, "loss": 0.4726, "step": 17288 }, { "epoch": 0.615957389956713, "grad_norm": 1.1582666765469267, "learning_rate": 6.78826892405733e-06, "loss": 0.4722, "step": 17289 }, { "epoch": 0.6159930170832072, "grad_norm": 1.3332723899464656, "learning_rate": 6.787176143732013e-06, "loss": 0.504, "step": 17290 }, { "epoch": 0.6160286442097013, "grad_norm": 1.6646830306027944, "learning_rate": 6.786083406187689e-06, "loss": 0.6741, "step": 17291 }, { "epoch": 0.6160642713361953, "grad_norm": 1.8147812760016884, "learning_rate": 6.784990711438916e-06, "loss": 0.507, "step": 17292 }, { "epoch": 0.6160998984626895, "grad_norm": 1.1095312529361607, "learning_rate": 6.783898059500233e-06, "loss": 0.4638, "step": 17293 }, { "epoch": 0.6161355255891836, "grad_norm": 1.6163323586365212, "learning_rate": 6.7828054503861985e-06, "loss": 0.5593, "step": 17294 }, { "epoch": 0.6161711527156777, "grad_norm": 1.5112442402643071, "learning_rate": 6.7817128841113556e-06, "loss": 0.4469, "step": 17295 }, { "epoch": 0.6162067798421719, "grad_norm": 1.5659524090635213, "learning_rate": 6.78062036069026e-06, "loss": 0.5249, "step": 17296 }, { "epoch": 0.616242406968666, "grad_norm": 1.9965068521374345, "learning_rate": 6.779527880137451e-06, "loss": 0.5126, "step": 17297 }, { "epoch": 0.61627803409516, "grad_norm": 1.8424642833668, "learning_rate": 6.778435442467479e-06, "loss": 0.5489, "step": 17298 }, { "epoch": 0.6163136612216542, "grad_norm": 1.5540268656193748, "learning_rate": 6.777343047694891e-06, "loss": 0.5199, "step": 17299 }, { "epoch": 0.6163492883481483, "grad_norm": 1.0499892165131137, "learning_rate": 6.776250695834234e-06, "loss": 0.3811, "step": 17300 }, { "epoch": 0.6163849154746424, "grad_norm": 1.1148784866625627, "learning_rate": 6.775158386900049e-06, "loss": 0.4349, "step": 17301 }, { "epoch": 0.6164205426011365, "grad_norm": 1.6558486564206065, "learning_rate": 6.774066120906886e-06, "loss": 0.5121, "step": 17302 }, { "epoch": 0.6164561697276306, "grad_norm": 1.4449977703348407, "learning_rate": 6.77297389786929e-06, "loss": 0.5304, "step": 17303 }, { "epoch": 0.6164917968541247, "grad_norm": 1.1867322574787857, "learning_rate": 6.771881717801798e-06, "loss": 0.4285, "step": 17304 }, { "epoch": 0.6165274239806189, "grad_norm": 1.054701556370979, "learning_rate": 6.77078958071896e-06, "loss": 0.4185, "step": 17305 }, { "epoch": 0.616563051107113, "grad_norm": 2.2505358119854084, "learning_rate": 6.769697486635314e-06, "loss": 0.5762, "step": 17306 }, { "epoch": 0.616598678233607, "grad_norm": 1.2420973605319827, "learning_rate": 6.768605435565407e-06, "loss": 0.4425, "step": 17307 }, { "epoch": 0.6166343053601012, "grad_norm": 1.5940487944539619, "learning_rate": 6.767513427523775e-06, "loss": 0.5383, "step": 17308 }, { "epoch": 0.6166699324865953, "grad_norm": 1.587814235576749, "learning_rate": 6.766421462524962e-06, "loss": 0.5247, "step": 17309 }, { "epoch": 0.6167055596130894, "grad_norm": 1.5352685846005056, "learning_rate": 6.765329540583505e-06, "loss": 0.491, "step": 17310 }, { "epoch": 0.6167411867395836, "grad_norm": 1.4109061638409777, "learning_rate": 6.76423766171395e-06, "loss": 0.6042, "step": 17311 }, { "epoch": 0.6167768138660776, "grad_norm": 1.2004875131251476, "learning_rate": 6.763145825930826e-06, "loss": 0.4674, "step": 17312 }, { "epoch": 0.6168124409925717, "grad_norm": 1.8951865766619456, "learning_rate": 6.762054033248681e-06, "loss": 0.5113, "step": 17313 }, { "epoch": 0.6168480681190659, "grad_norm": 1.397508613723625, "learning_rate": 6.760962283682054e-06, "loss": 0.6135, "step": 17314 }, { "epoch": 0.61688369524556, "grad_norm": 1.2419400055181666, "learning_rate": 6.7598705772454724e-06, "loss": 0.4328, "step": 17315 }, { "epoch": 0.616919322372054, "grad_norm": 1.2248128542560015, "learning_rate": 6.758778913953481e-06, "loss": 0.4159, "step": 17316 }, { "epoch": 0.6169549494985482, "grad_norm": 1.3407097051948582, "learning_rate": 6.7576872938206115e-06, "loss": 0.5399, "step": 17317 }, { "epoch": 0.6169905766250423, "grad_norm": 1.0974649741270002, "learning_rate": 6.756595716861407e-06, "loss": 0.3854, "step": 17318 }, { "epoch": 0.6170262037515364, "grad_norm": 1.5155580672716527, "learning_rate": 6.755504183090393e-06, "loss": 0.566, "step": 17319 }, { "epoch": 0.6170618308780306, "grad_norm": 1.5664367340682075, "learning_rate": 6.754412692522112e-06, "loss": 0.5316, "step": 17320 }, { "epoch": 0.6170974580045246, "grad_norm": 1.1383936081664285, "learning_rate": 6.7533212451710914e-06, "loss": 0.405, "step": 17321 }, { "epoch": 0.6171330851310187, "grad_norm": 1.0215233296669157, "learning_rate": 6.752229841051872e-06, "loss": 0.4023, "step": 17322 }, { "epoch": 0.6171687122575129, "grad_norm": 1.6697168545067427, "learning_rate": 6.751138480178978e-06, "loss": 0.4871, "step": 17323 }, { "epoch": 0.617204339384007, "grad_norm": 1.4930394379005874, "learning_rate": 6.750047162566944e-06, "loss": 0.5007, "step": 17324 }, { "epoch": 0.6172399665105011, "grad_norm": 1.5749718958064656, "learning_rate": 6.7489558882303084e-06, "loss": 0.4298, "step": 17325 }, { "epoch": 0.6172755936369952, "grad_norm": 1.3790783684928014, "learning_rate": 6.747864657183596e-06, "loss": 0.5227, "step": 17326 }, { "epoch": 0.6173112207634893, "grad_norm": 1.2550238362173325, "learning_rate": 6.746773469441339e-06, "loss": 0.4575, "step": 17327 }, { "epoch": 0.6173468478899834, "grad_norm": 1.1239520036555815, "learning_rate": 6.745682325018066e-06, "loss": 0.4414, "step": 17328 }, { "epoch": 0.6173824750164776, "grad_norm": 1.6168651962852334, "learning_rate": 6.744591223928311e-06, "loss": 0.5002, "step": 17329 }, { "epoch": 0.6174181021429717, "grad_norm": 1.4520909586373825, "learning_rate": 6.743500166186595e-06, "loss": 0.5587, "step": 17330 }, { "epoch": 0.6174537292694657, "grad_norm": 1.0992538847350015, "learning_rate": 6.742409151807453e-06, "loss": 0.4375, "step": 17331 }, { "epoch": 0.6174893563959599, "grad_norm": 1.2545356159434475, "learning_rate": 6.741318180805408e-06, "loss": 0.4264, "step": 17332 }, { "epoch": 0.617524983522454, "grad_norm": 1.410555420523545, "learning_rate": 6.740227253194993e-06, "loss": 0.5278, "step": 17333 }, { "epoch": 0.6175606106489481, "grad_norm": 1.0478664512042846, "learning_rate": 6.739136368990726e-06, "loss": 0.4372, "step": 17334 }, { "epoch": 0.6175962377754423, "grad_norm": 1.2540629768778624, "learning_rate": 6.738045528207137e-06, "loss": 0.5002, "step": 17335 }, { "epoch": 0.6176318649019363, "grad_norm": 1.936401134593854, "learning_rate": 6.736954730858756e-06, "loss": 0.5006, "step": 17336 }, { "epoch": 0.6176674920284304, "grad_norm": 1.0926029274113152, "learning_rate": 6.7358639769601e-06, "loss": 0.4017, "step": 17337 }, { "epoch": 0.6177031191549246, "grad_norm": 1.6624977336840678, "learning_rate": 6.7347732665257e-06, "loss": 0.5462, "step": 17338 }, { "epoch": 0.6177387462814187, "grad_norm": 1.5374615249060941, "learning_rate": 6.733682599570073e-06, "loss": 0.5501, "step": 17339 }, { "epoch": 0.6177743734079127, "grad_norm": 1.4293766820333704, "learning_rate": 6.732591976107749e-06, "loss": 0.5176, "step": 17340 }, { "epoch": 0.6178100005344069, "grad_norm": 1.2247982535472395, "learning_rate": 6.731501396153243e-06, "loss": 0.3963, "step": 17341 }, { "epoch": 0.617845627660901, "grad_norm": 1.1726972915348899, "learning_rate": 6.730410859721081e-06, "loss": 0.4835, "step": 17342 }, { "epoch": 0.6178812547873951, "grad_norm": 1.2352056903047273, "learning_rate": 6.729320366825785e-06, "loss": 0.4651, "step": 17343 }, { "epoch": 0.6179168819138893, "grad_norm": 1.467577158984033, "learning_rate": 6.7282299174818756e-06, "loss": 0.6634, "step": 17344 }, { "epoch": 0.6179525090403833, "grad_norm": 1.5758614850051436, "learning_rate": 6.7271395117038685e-06, "loss": 0.5313, "step": 17345 }, { "epoch": 0.6179881361668774, "grad_norm": 1.9198293248977996, "learning_rate": 6.7260491495062885e-06, "loss": 0.5551, "step": 17346 }, { "epoch": 0.6180237632933716, "grad_norm": 1.0781482726580265, "learning_rate": 6.724958830903649e-06, "loss": 0.4371, "step": 17347 }, { "epoch": 0.6180593904198657, "grad_norm": 1.4234582620578986, "learning_rate": 6.7238685559104745e-06, "loss": 0.5519, "step": 17348 }, { "epoch": 0.6180950175463598, "grad_norm": 2.094105899380655, "learning_rate": 6.722778324541283e-06, "loss": 0.4644, "step": 17349 }, { "epoch": 0.618130644672854, "grad_norm": 1.0575858141273915, "learning_rate": 6.721688136810582e-06, "loss": 0.3996, "step": 17350 }, { "epoch": 0.618166271799348, "grad_norm": 1.3648137346718914, "learning_rate": 6.720597992732901e-06, "loss": 0.5976, "step": 17351 }, { "epoch": 0.6182018989258421, "grad_norm": 1.0966278321743408, "learning_rate": 6.719507892322747e-06, "loss": 0.4043, "step": 17352 }, { "epoch": 0.6182375260523363, "grad_norm": 1.3799598467809793, "learning_rate": 6.7184178355946395e-06, "loss": 0.4424, "step": 17353 }, { "epoch": 0.6182731531788304, "grad_norm": 1.262274598476967, "learning_rate": 6.717327822563089e-06, "loss": 0.4311, "step": 17354 }, { "epoch": 0.6183087803053244, "grad_norm": 1.1446014912857416, "learning_rate": 6.71623785324262e-06, "loss": 0.4215, "step": 17355 }, { "epoch": 0.6183444074318186, "grad_norm": 1.2498750929338331, "learning_rate": 6.715147927647733e-06, "loss": 0.493, "step": 17356 }, { "epoch": 0.6183800345583127, "grad_norm": 2.5154212824445668, "learning_rate": 6.714058045792949e-06, "loss": 0.5464, "step": 17357 }, { "epoch": 0.6184156616848068, "grad_norm": 1.1875945671312578, "learning_rate": 6.712968207692778e-06, "loss": 0.4438, "step": 17358 }, { "epoch": 0.618451288811301, "grad_norm": 1.0546264261740133, "learning_rate": 6.711878413361732e-06, "loss": 0.4378, "step": 17359 }, { "epoch": 0.618486915937795, "grad_norm": 1.6482793891154608, "learning_rate": 6.710788662814328e-06, "loss": 0.5256, "step": 17360 }, { "epoch": 0.6185225430642891, "grad_norm": 1.080907425571272, "learning_rate": 6.709698956065067e-06, "loss": 0.444, "step": 17361 }, { "epoch": 0.6185581701907833, "grad_norm": 1.5804907838496682, "learning_rate": 6.708609293128467e-06, "loss": 0.5378, "step": 17362 }, { "epoch": 0.6185937973172774, "grad_norm": 1.1223709409688731, "learning_rate": 6.7075196740190326e-06, "loss": 0.4387, "step": 17363 }, { "epoch": 0.6186294244437714, "grad_norm": 1.1963538028185314, "learning_rate": 6.706430098751278e-06, "loss": 0.4266, "step": 17364 }, { "epoch": 0.6186650515702656, "grad_norm": 1.2065279046854414, "learning_rate": 6.7053405673397045e-06, "loss": 0.4751, "step": 17365 }, { "epoch": 0.6187006786967597, "grad_norm": 1.1252459536745885, "learning_rate": 6.7042510797988304e-06, "loss": 0.4265, "step": 17366 }, { "epoch": 0.6187363058232538, "grad_norm": 1.1697999516437183, "learning_rate": 6.703161636143152e-06, "loss": 0.4525, "step": 17367 }, { "epoch": 0.618771932949748, "grad_norm": 1.6144558265301128, "learning_rate": 6.702072236387182e-06, "loss": 0.5149, "step": 17368 }, { "epoch": 0.618807560076242, "grad_norm": 1.1319032072055728, "learning_rate": 6.700982880545424e-06, "loss": 0.4095, "step": 17369 }, { "epoch": 0.6188431872027361, "grad_norm": 1.0965238194652907, "learning_rate": 6.699893568632384e-06, "loss": 0.4022, "step": 17370 }, { "epoch": 0.6188788143292303, "grad_norm": 1.0846850754284107, "learning_rate": 6.698804300662574e-06, "loss": 0.3911, "step": 17371 }, { "epoch": 0.6189144414557244, "grad_norm": 1.1093301564831841, "learning_rate": 6.697715076650486e-06, "loss": 0.4391, "step": 17372 }, { "epoch": 0.6189500685822185, "grad_norm": 1.6338251325123667, "learning_rate": 6.6966258966106314e-06, "loss": 0.5685, "step": 17373 }, { "epoch": 0.6189856957087126, "grad_norm": 1.2142046852621846, "learning_rate": 6.69553676055751e-06, "loss": 0.4099, "step": 17374 }, { "epoch": 0.6190213228352067, "grad_norm": 1.2168117354577792, "learning_rate": 6.6944476685056324e-06, "loss": 0.4166, "step": 17375 }, { "epoch": 0.6190569499617008, "grad_norm": 1.9904247128038726, "learning_rate": 6.693358620469488e-06, "loss": 0.5375, "step": 17376 }, { "epoch": 0.619092577088195, "grad_norm": 1.2254094964189235, "learning_rate": 6.692269616463589e-06, "loss": 0.4622, "step": 17377 }, { "epoch": 0.6191282042146891, "grad_norm": 1.3759219674322172, "learning_rate": 6.691180656502429e-06, "loss": 0.5485, "step": 17378 }, { "epoch": 0.6191638313411831, "grad_norm": 1.0485992544085556, "learning_rate": 6.690091740600515e-06, "loss": 0.423, "step": 17379 }, { "epoch": 0.6191994584676773, "grad_norm": 1.1477862523472309, "learning_rate": 6.689002868772339e-06, "loss": 0.4203, "step": 17380 }, { "epoch": 0.6192350855941714, "grad_norm": 1.4206073299662578, "learning_rate": 6.687914041032409e-06, "loss": 0.6185, "step": 17381 }, { "epoch": 0.6192707127206655, "grad_norm": 1.2566108656924442, "learning_rate": 6.686825257395213e-06, "loss": 0.385, "step": 17382 }, { "epoch": 0.6193063398471597, "grad_norm": 1.6740592848486704, "learning_rate": 6.6857365178752565e-06, "loss": 0.588, "step": 17383 }, { "epoch": 0.6193419669736537, "grad_norm": 1.7612588144177417, "learning_rate": 6.684647822487036e-06, "loss": 0.4644, "step": 17384 }, { "epoch": 0.6193775941001478, "grad_norm": 1.1930604500403326, "learning_rate": 6.683559171245044e-06, "loss": 0.4648, "step": 17385 }, { "epoch": 0.619413221226642, "grad_norm": 1.199943328266267, "learning_rate": 6.682470564163785e-06, "loss": 0.4247, "step": 17386 }, { "epoch": 0.6194488483531361, "grad_norm": 1.2542280273033481, "learning_rate": 6.681382001257743e-06, "loss": 0.4411, "step": 17387 }, { "epoch": 0.6194844754796301, "grad_norm": 1.4906345363889986, "learning_rate": 6.680293482541424e-06, "loss": 0.5556, "step": 17388 }, { "epoch": 0.6195201026061243, "grad_norm": 1.156624559508642, "learning_rate": 6.679205008029315e-06, "loss": 0.4748, "step": 17389 }, { "epoch": 0.6195557297326184, "grad_norm": 1.3791978602099726, "learning_rate": 6.678116577735915e-06, "loss": 0.5604, "step": 17390 }, { "epoch": 0.6195913568591125, "grad_norm": 1.6162175041231672, "learning_rate": 6.677028191675712e-06, "loss": 0.4432, "step": 17391 }, { "epoch": 0.6196269839856067, "grad_norm": 1.4133230568906199, "learning_rate": 6.6759398498632044e-06, "loss": 0.5139, "step": 17392 }, { "epoch": 0.6196626111121007, "grad_norm": 1.3779082060826169, "learning_rate": 6.674851552312878e-06, "loss": 0.4987, "step": 17393 }, { "epoch": 0.6196982382385948, "grad_norm": 1.3512605423089352, "learning_rate": 6.6737632990392264e-06, "loss": 0.5131, "step": 17394 }, { "epoch": 0.619733865365089, "grad_norm": 1.5263643898180599, "learning_rate": 6.672675090056744e-06, "loss": 0.5303, "step": 17395 }, { "epoch": 0.6197694924915831, "grad_norm": 1.1503088260761796, "learning_rate": 6.671586925379916e-06, "loss": 0.4602, "step": 17396 }, { "epoch": 0.6198051196180772, "grad_norm": 1.3833281493020764, "learning_rate": 6.67049880502324e-06, "loss": 0.3884, "step": 17397 }, { "epoch": 0.6198407467445713, "grad_norm": 1.0526110115375331, "learning_rate": 6.669410729001193e-06, "loss": 0.4508, "step": 17398 }, { "epoch": 0.6198763738710654, "grad_norm": 1.6393408565424714, "learning_rate": 6.668322697328273e-06, "loss": 0.4474, "step": 17399 }, { "epoch": 0.6199120009975595, "grad_norm": 1.4461040187464258, "learning_rate": 6.667234710018964e-06, "loss": 0.5948, "step": 17400 }, { "epoch": 0.6199476281240537, "grad_norm": 1.5898612194556019, "learning_rate": 6.666146767087756e-06, "loss": 0.4774, "step": 17401 }, { "epoch": 0.6199832552505478, "grad_norm": 1.226492171667375, "learning_rate": 6.6650588685491325e-06, "loss": 0.4273, "step": 17402 }, { "epoch": 0.6200188823770418, "grad_norm": 1.233919399763212, "learning_rate": 6.663971014417585e-06, "loss": 0.4538, "step": 17403 }, { "epoch": 0.620054509503536, "grad_norm": 1.222569394342386, "learning_rate": 6.66288320470759e-06, "loss": 0.4482, "step": 17404 }, { "epoch": 0.6200901366300301, "grad_norm": 1.4152647381847914, "learning_rate": 6.661795439433638e-06, "loss": 0.5107, "step": 17405 }, { "epoch": 0.6201257637565242, "grad_norm": 1.1685645978333565, "learning_rate": 6.660707718610215e-06, "loss": 0.4208, "step": 17406 }, { "epoch": 0.6201613908830184, "grad_norm": 1.2105028276779728, "learning_rate": 6.659620042251802e-06, "loss": 0.4416, "step": 17407 }, { "epoch": 0.6201970180095124, "grad_norm": 1.1215260384289305, "learning_rate": 6.658532410372886e-06, "loss": 0.4109, "step": 17408 }, { "epoch": 0.6202326451360065, "grad_norm": 1.2906144908515853, "learning_rate": 6.657444822987942e-06, "loss": 0.4585, "step": 17409 }, { "epoch": 0.6202682722625007, "grad_norm": 1.6828458797410095, "learning_rate": 6.656357280111458e-06, "loss": 0.5883, "step": 17410 }, { "epoch": 0.6203038993889948, "grad_norm": 1.9536137432352283, "learning_rate": 6.655269781757914e-06, "loss": 0.6066, "step": 17411 }, { "epoch": 0.6203395265154888, "grad_norm": 1.1324498851886882, "learning_rate": 6.654182327941794e-06, "loss": 0.4313, "step": 17412 }, { "epoch": 0.620375153641983, "grad_norm": 1.4624400996449183, "learning_rate": 6.653094918677567e-06, "loss": 0.5446, "step": 17413 }, { "epoch": 0.6204107807684771, "grad_norm": 1.8279123920758436, "learning_rate": 6.652007553979729e-06, "loss": 0.4786, "step": 17414 }, { "epoch": 0.6204464078949712, "grad_norm": 1.6208836282998957, "learning_rate": 6.6509202338627456e-06, "loss": 0.5155, "step": 17415 }, { "epoch": 0.6204820350214654, "grad_norm": 1.3006922667446301, "learning_rate": 6.6498329583411e-06, "loss": 0.4518, "step": 17416 }, { "epoch": 0.6205176621479594, "grad_norm": 1.243557768548232, "learning_rate": 6.648745727429272e-06, "loss": 0.456, "step": 17417 }, { "epoch": 0.6205532892744535, "grad_norm": 1.3066335597344947, "learning_rate": 6.647658541141735e-06, "loss": 0.4439, "step": 17418 }, { "epoch": 0.6205889164009477, "grad_norm": 1.1667211478187929, "learning_rate": 6.646571399492972e-06, "loss": 0.4911, "step": 17419 }, { "epoch": 0.6206245435274418, "grad_norm": 1.638605385193325, "learning_rate": 6.645484302497452e-06, "loss": 0.5065, "step": 17420 }, { "epoch": 0.6206601706539359, "grad_norm": 1.664682411531577, "learning_rate": 6.644397250169652e-06, "loss": 0.5333, "step": 17421 }, { "epoch": 0.62069579778043, "grad_norm": 1.1032849729167988, "learning_rate": 6.643310242524048e-06, "loss": 0.4485, "step": 17422 }, { "epoch": 0.6207314249069241, "grad_norm": 2.35609406691689, "learning_rate": 6.642223279575118e-06, "loss": 0.5988, "step": 17423 }, { "epoch": 0.6207670520334182, "grad_norm": 1.8233717747938196, "learning_rate": 6.641136361337328e-06, "loss": 0.6123, "step": 17424 }, { "epoch": 0.6208026791599124, "grad_norm": 1.3444488224068285, "learning_rate": 6.640049487825158e-06, "loss": 0.5119, "step": 17425 }, { "epoch": 0.6208383062864065, "grad_norm": 1.1312608645108817, "learning_rate": 6.638962659053074e-06, "loss": 0.4102, "step": 17426 }, { "epoch": 0.6208739334129005, "grad_norm": 1.6258686971340248, "learning_rate": 6.637875875035555e-06, "loss": 0.5516, "step": 17427 }, { "epoch": 0.6209095605393947, "grad_norm": 1.450483257472878, "learning_rate": 6.636789135787065e-06, "loss": 0.5439, "step": 17428 }, { "epoch": 0.6209451876658888, "grad_norm": 1.8962839348601996, "learning_rate": 6.635702441322081e-06, "loss": 0.6131, "step": 17429 }, { "epoch": 0.6209808147923829, "grad_norm": 1.702243735316567, "learning_rate": 6.634615791655071e-06, "loss": 0.5454, "step": 17430 }, { "epoch": 0.6210164419188771, "grad_norm": 1.4894966031182784, "learning_rate": 6.633529186800503e-06, "loss": 0.5213, "step": 17431 }, { "epoch": 0.6210520690453711, "grad_norm": 2.0053927317833646, "learning_rate": 6.632442626772847e-06, "loss": 0.5983, "step": 17432 }, { "epoch": 0.6210876961718652, "grad_norm": 1.7452081967501094, "learning_rate": 6.63135611158657e-06, "loss": 0.5573, "step": 17433 }, { "epoch": 0.6211233232983594, "grad_norm": 1.3761220637422564, "learning_rate": 6.630269641256144e-06, "loss": 0.4899, "step": 17434 }, { "epoch": 0.6211589504248535, "grad_norm": 1.6098111557298609, "learning_rate": 6.62918321579603e-06, "loss": 0.4739, "step": 17435 }, { "epoch": 0.6211945775513475, "grad_norm": 1.4298996678549007, "learning_rate": 6.6280968352207e-06, "loss": 0.5559, "step": 17436 }, { "epoch": 0.6212302046778417, "grad_norm": 1.465534435851761, "learning_rate": 6.6270104995446146e-06, "loss": 0.4972, "step": 17437 }, { "epoch": 0.6212658318043358, "grad_norm": 1.3834627926369323, "learning_rate": 6.625924208782246e-06, "loss": 0.5781, "step": 17438 }, { "epoch": 0.6213014589308299, "grad_norm": 1.470255199828714, "learning_rate": 6.62483796294805e-06, "loss": 0.4367, "step": 17439 }, { "epoch": 0.6213370860573241, "grad_norm": 1.1213127164679346, "learning_rate": 6.623751762056496e-06, "loss": 0.3984, "step": 17440 }, { "epoch": 0.6213727131838181, "grad_norm": 1.1599465713981292, "learning_rate": 6.622665606122052e-06, "loss": 0.451, "step": 17441 }, { "epoch": 0.6214083403103122, "grad_norm": 1.7217282221553663, "learning_rate": 6.621579495159172e-06, "loss": 0.5344, "step": 17442 }, { "epoch": 0.6214439674368064, "grad_norm": 1.0899330203391464, "learning_rate": 6.620493429182324e-06, "loss": 0.4314, "step": 17443 }, { "epoch": 0.6214795945633005, "grad_norm": 1.6097505796426816, "learning_rate": 6.6194074082059665e-06, "loss": 0.4863, "step": 17444 }, { "epoch": 0.6215152216897946, "grad_norm": 1.684017132454411, "learning_rate": 6.618321432244565e-06, "loss": 0.5706, "step": 17445 }, { "epoch": 0.6215508488162887, "grad_norm": 1.43868627332283, "learning_rate": 6.617235501312575e-06, "loss": 0.5021, "step": 17446 }, { "epoch": 0.6215864759427828, "grad_norm": 1.2878273334466528, "learning_rate": 6.616149615424461e-06, "loss": 0.4856, "step": 17447 }, { "epoch": 0.6216221030692769, "grad_norm": 1.1087462154638317, "learning_rate": 6.615063774594677e-06, "loss": 0.4728, "step": 17448 }, { "epoch": 0.6216577301957711, "grad_norm": 1.305458444041544, "learning_rate": 6.613977978837689e-06, "loss": 0.5688, "step": 17449 }, { "epoch": 0.6216933573222652, "grad_norm": 1.5985855414170052, "learning_rate": 6.612892228167946e-06, "loss": 0.5324, "step": 17450 }, { "epoch": 0.6217289844487592, "grad_norm": 1.2902905533816431, "learning_rate": 6.61180652259991e-06, "loss": 0.486, "step": 17451 }, { "epoch": 0.6217646115752534, "grad_norm": 1.77514489749634, "learning_rate": 6.610720862148043e-06, "loss": 0.5968, "step": 17452 }, { "epoch": 0.6218002387017475, "grad_norm": 1.132375184007851, "learning_rate": 6.6096352468267935e-06, "loss": 0.4563, "step": 17453 }, { "epoch": 0.6218358658282416, "grad_norm": 1.5227387925267222, "learning_rate": 6.608549676650622e-06, "loss": 0.4755, "step": 17454 }, { "epoch": 0.6218714929547358, "grad_norm": 1.3542184800743367, "learning_rate": 6.607464151633981e-06, "loss": 0.4425, "step": 17455 }, { "epoch": 0.6219071200812298, "grad_norm": 1.5770334042079222, "learning_rate": 6.606378671791329e-06, "loss": 0.474, "step": 17456 }, { "epoch": 0.6219427472077239, "grad_norm": 1.7057533438796637, "learning_rate": 6.6052932371371135e-06, "loss": 0.5313, "step": 17457 }, { "epoch": 0.6219783743342181, "grad_norm": 1.2012165965768655, "learning_rate": 6.604207847685793e-06, "loss": 0.436, "step": 17458 }, { "epoch": 0.6220140014607122, "grad_norm": 1.1247098587803848, "learning_rate": 6.603122503451818e-06, "loss": 0.4299, "step": 17459 }, { "epoch": 0.6220496285872062, "grad_norm": 1.2344049861858275, "learning_rate": 6.602037204449643e-06, "loss": 0.4526, "step": 17460 }, { "epoch": 0.6220852557137004, "grad_norm": 1.2075605736674662, "learning_rate": 6.6009519506937144e-06, "loss": 0.4175, "step": 17461 }, { "epoch": 0.6221208828401945, "grad_norm": 1.0806112124651095, "learning_rate": 6.599866742198486e-06, "loss": 0.4619, "step": 17462 }, { "epoch": 0.6221565099666886, "grad_norm": 1.2428493206862434, "learning_rate": 6.59878157897841e-06, "loss": 0.4377, "step": 17463 }, { "epoch": 0.6221921370931828, "grad_norm": 1.3908985573426362, "learning_rate": 6.597696461047935e-06, "loss": 0.4519, "step": 17464 }, { "epoch": 0.6222277642196768, "grad_norm": 1.3706898521185842, "learning_rate": 6.59661138842151e-06, "loss": 0.5377, "step": 17465 }, { "epoch": 0.6222633913461709, "grad_norm": 1.2117849130899532, "learning_rate": 6.595526361113581e-06, "loss": 0.4316, "step": 17466 }, { "epoch": 0.6222990184726651, "grad_norm": 1.0974429064907012, "learning_rate": 6.594441379138603e-06, "loss": 0.4437, "step": 17467 }, { "epoch": 0.6223346455991592, "grad_norm": 1.1481056601868393, "learning_rate": 6.593356442511014e-06, "loss": 0.4337, "step": 17468 }, { "epoch": 0.6223702727256533, "grad_norm": 1.4530211329306235, "learning_rate": 6.592271551245266e-06, "loss": 0.5199, "step": 17469 }, { "epoch": 0.6224058998521474, "grad_norm": 1.7720332625591768, "learning_rate": 6.591186705355803e-06, "loss": 0.6274, "step": 17470 }, { "epoch": 0.6224415269786415, "grad_norm": 1.5927215683225824, "learning_rate": 6.590101904857076e-06, "loss": 0.5199, "step": 17471 }, { "epoch": 0.6224771541051356, "grad_norm": 1.5738548869593223, "learning_rate": 6.589017149763522e-06, "loss": 0.5554, "step": 17472 }, { "epoch": 0.6225127812316298, "grad_norm": 1.7773892772458293, "learning_rate": 6.58793244008959e-06, "loss": 0.6098, "step": 17473 }, { "epoch": 0.6225484083581239, "grad_norm": 1.8168512398710792, "learning_rate": 6.586847775849721e-06, "loss": 0.5252, "step": 17474 }, { "epoch": 0.6225840354846179, "grad_norm": 1.2423607892364306, "learning_rate": 6.585763157058358e-06, "loss": 0.4802, "step": 17475 }, { "epoch": 0.6226196626111121, "grad_norm": 1.4268605834891188, "learning_rate": 6.584678583729947e-06, "loss": 0.64, "step": 17476 }, { "epoch": 0.6226552897376062, "grad_norm": 1.1832065122641955, "learning_rate": 6.583594055878928e-06, "loss": 0.4222, "step": 17477 }, { "epoch": 0.6226909168641003, "grad_norm": 1.88659474672358, "learning_rate": 6.5825095735197445e-06, "loss": 0.5074, "step": 17478 }, { "epoch": 0.6227265439905945, "grad_norm": 1.9096850349628094, "learning_rate": 6.58142513666683e-06, "loss": 0.5649, "step": 17479 }, { "epoch": 0.6227621711170885, "grad_norm": 1.8194145997486413, "learning_rate": 6.580340745334633e-06, "loss": 0.636, "step": 17480 }, { "epoch": 0.6227977982435826, "grad_norm": 1.7293634769439898, "learning_rate": 6.579256399537587e-06, "loss": 0.5718, "step": 17481 }, { "epoch": 0.6228334253700768, "grad_norm": 1.1354134999262033, "learning_rate": 6.578172099290136e-06, "loss": 0.4376, "step": 17482 }, { "epoch": 0.6228690524965709, "grad_norm": 1.1690420594430613, "learning_rate": 6.577087844606712e-06, "loss": 0.4677, "step": 17483 }, { "epoch": 0.622904679623065, "grad_norm": 1.1409071981080685, "learning_rate": 6.576003635501758e-06, "loss": 0.4428, "step": 17484 }, { "epoch": 0.6229403067495591, "grad_norm": 1.6307432209401374, "learning_rate": 6.574919471989706e-06, "loss": 0.5621, "step": 17485 }, { "epoch": 0.6229759338760532, "grad_norm": 1.1160660044895943, "learning_rate": 6.573835354084995e-06, "loss": 0.3911, "step": 17486 }, { "epoch": 0.6230115610025473, "grad_norm": 1.7030393918572926, "learning_rate": 6.572751281802065e-06, "loss": 0.5541, "step": 17487 }, { "epoch": 0.6230471881290415, "grad_norm": 1.0776957352449466, "learning_rate": 6.571667255155343e-06, "loss": 0.4022, "step": 17488 }, { "epoch": 0.6230828152555355, "grad_norm": 1.0826877911405082, "learning_rate": 6.570583274159272e-06, "loss": 0.4355, "step": 17489 }, { "epoch": 0.6231184423820296, "grad_norm": 1.0778900624057202, "learning_rate": 6.569499338828278e-06, "loss": 0.4021, "step": 17490 }, { "epoch": 0.6231540695085238, "grad_norm": 1.1613870104350226, "learning_rate": 6.5684154491768005e-06, "loss": 0.4025, "step": 17491 }, { "epoch": 0.6231896966350179, "grad_norm": 1.1309905054955647, "learning_rate": 6.567331605219267e-06, "loss": 0.4388, "step": 17492 }, { "epoch": 0.623225323761512, "grad_norm": 1.4530433553128956, "learning_rate": 6.566247806970119e-06, "loss": 0.5475, "step": 17493 }, { "epoch": 0.6232609508880061, "grad_norm": 1.5806304948937389, "learning_rate": 6.565164054443775e-06, "loss": 0.5534, "step": 17494 }, { "epoch": 0.6232965780145002, "grad_norm": 1.6174077000911957, "learning_rate": 6.564080347654675e-06, "loss": 0.5075, "step": 17495 }, { "epoch": 0.6233322051409943, "grad_norm": 1.4515045973654372, "learning_rate": 6.5629966866172445e-06, "loss": 0.5763, "step": 17496 }, { "epoch": 0.6233678322674885, "grad_norm": 1.1283225881059908, "learning_rate": 6.561913071345916e-06, "loss": 0.4657, "step": 17497 }, { "epoch": 0.6234034593939826, "grad_norm": 1.33226064391537, "learning_rate": 6.560829501855121e-06, "loss": 0.5669, "step": 17498 }, { "epoch": 0.6234390865204766, "grad_norm": 1.5303056931856234, "learning_rate": 6.559745978159282e-06, "loss": 0.5088, "step": 17499 }, { "epoch": 0.6234747136469708, "grad_norm": 1.4574854581642267, "learning_rate": 6.558662500272832e-06, "loss": 0.5229, "step": 17500 }, { "epoch": 0.6235103407734649, "grad_norm": 1.3571699980129874, "learning_rate": 6.557579068210193e-06, "loss": 0.5755, "step": 17501 }, { "epoch": 0.623545967899959, "grad_norm": 1.5392467306897715, "learning_rate": 6.556495681985797e-06, "loss": 0.5452, "step": 17502 }, { "epoch": 0.6235815950264532, "grad_norm": 1.139974854186008, "learning_rate": 6.555412341614067e-06, "loss": 0.3997, "step": 17503 }, { "epoch": 0.6236172221529472, "grad_norm": 1.5502417062789209, "learning_rate": 6.554329047109433e-06, "loss": 0.4943, "step": 17504 }, { "epoch": 0.6236528492794413, "grad_norm": 1.1549475192108203, "learning_rate": 6.553245798486312e-06, "loss": 0.4123, "step": 17505 }, { "epoch": 0.6236884764059355, "grad_norm": 1.353196610135801, "learning_rate": 6.552162595759135e-06, "loss": 0.5433, "step": 17506 }, { "epoch": 0.6237241035324296, "grad_norm": 1.1531639254253236, "learning_rate": 6.551079438942319e-06, "loss": 0.3992, "step": 17507 }, { "epoch": 0.6237597306589237, "grad_norm": 1.1714006159867005, "learning_rate": 6.549996328050296e-06, "loss": 0.4396, "step": 17508 }, { "epoch": 0.6237953577854178, "grad_norm": 1.4132985421086852, "learning_rate": 6.54891326309748e-06, "loss": 0.5328, "step": 17509 }, { "epoch": 0.6238309849119119, "grad_norm": 1.1218209287547856, "learning_rate": 6.547830244098296e-06, "loss": 0.4422, "step": 17510 }, { "epoch": 0.623866612038406, "grad_norm": 1.1444389443061571, "learning_rate": 6.546747271067167e-06, "loss": 0.4468, "step": 17511 }, { "epoch": 0.6239022391649002, "grad_norm": 2.1049583377851175, "learning_rate": 6.545664344018509e-06, "loss": 0.6476, "step": 17512 }, { "epoch": 0.6239378662913942, "grad_norm": 1.5530784139754956, "learning_rate": 6.544581462966748e-06, "loss": 0.5443, "step": 17513 }, { "epoch": 0.6239734934178883, "grad_norm": 1.6550897523592707, "learning_rate": 6.543498627926296e-06, "loss": 0.605, "step": 17514 }, { "epoch": 0.6240091205443825, "grad_norm": 1.6281025329275498, "learning_rate": 6.54241583891158e-06, "loss": 0.7279, "step": 17515 }, { "epoch": 0.6240447476708766, "grad_norm": 1.281211336183875, "learning_rate": 6.541333095937011e-06, "loss": 0.4879, "step": 17516 }, { "epoch": 0.6240803747973707, "grad_norm": 1.1777254936452128, "learning_rate": 6.540250399017012e-06, "loss": 0.4561, "step": 17517 }, { "epoch": 0.6241160019238648, "grad_norm": 1.233385133181899, "learning_rate": 6.539167748165993e-06, "loss": 0.4473, "step": 17518 }, { "epoch": 0.6241516290503589, "grad_norm": 1.928129358319845, "learning_rate": 6.538085143398381e-06, "loss": 0.5455, "step": 17519 }, { "epoch": 0.6241872561768531, "grad_norm": 1.2052481078203505, "learning_rate": 6.537002584728578e-06, "loss": 0.448, "step": 17520 }, { "epoch": 0.6242228833033472, "grad_norm": 1.0939195744340173, "learning_rate": 6.535920072171008e-06, "loss": 0.448, "step": 17521 }, { "epoch": 0.6242585104298413, "grad_norm": 1.4634687348782187, "learning_rate": 6.534837605740085e-06, "loss": 0.5265, "step": 17522 }, { "epoch": 0.6242941375563354, "grad_norm": 1.1681376998330613, "learning_rate": 6.533755185450219e-06, "loss": 0.4213, "step": 17523 }, { "epoch": 0.6243297646828295, "grad_norm": 1.3228739349695073, "learning_rate": 6.53267281131583e-06, "loss": 0.5164, "step": 17524 }, { "epoch": 0.6243653918093236, "grad_norm": 1.0883067442460497, "learning_rate": 6.531590483351322e-06, "loss": 0.3871, "step": 17525 }, { "epoch": 0.6244010189358178, "grad_norm": 1.1006397375618697, "learning_rate": 6.530508201571113e-06, "loss": 0.4209, "step": 17526 }, { "epoch": 0.6244366460623119, "grad_norm": 1.0956007846273252, "learning_rate": 6.529425965989611e-06, "loss": 0.3958, "step": 17527 }, { "epoch": 0.6244722731888059, "grad_norm": 0.9736190441611915, "learning_rate": 6.528343776621229e-06, "loss": 0.3911, "step": 17528 }, { "epoch": 0.6245079003153001, "grad_norm": 1.2460577650539855, "learning_rate": 6.5272616334803755e-06, "loss": 0.5452, "step": 17529 }, { "epoch": 0.6245435274417942, "grad_norm": 1.4987233217167788, "learning_rate": 6.5261795365814625e-06, "loss": 0.5402, "step": 17530 }, { "epoch": 0.6245791545682883, "grad_norm": 1.1484371081512696, "learning_rate": 6.525097485938894e-06, "loss": 0.4638, "step": 17531 }, { "epoch": 0.6246147816947825, "grad_norm": 1.1252875899652117, "learning_rate": 6.524015481567081e-06, "loss": 0.4271, "step": 17532 }, { "epoch": 0.6246504088212765, "grad_norm": 1.8024752750438697, "learning_rate": 6.522933523480435e-06, "loss": 0.6118, "step": 17533 }, { "epoch": 0.6246860359477706, "grad_norm": 1.4112123775007257, "learning_rate": 6.521851611693355e-06, "loss": 0.5458, "step": 17534 }, { "epoch": 0.6247216630742648, "grad_norm": 1.3372212586919634, "learning_rate": 6.520769746220258e-06, "loss": 0.5241, "step": 17535 }, { "epoch": 0.6247572902007589, "grad_norm": 1.8183918732441307, "learning_rate": 6.519687927075539e-06, "loss": 0.5439, "step": 17536 }, { "epoch": 0.624792917327253, "grad_norm": 1.0978026559857879, "learning_rate": 6.51860615427361e-06, "loss": 0.4432, "step": 17537 }, { "epoch": 0.6248285444537471, "grad_norm": 1.9083731882384585, "learning_rate": 6.5175244278288705e-06, "loss": 0.5571, "step": 17538 }, { "epoch": 0.6248641715802412, "grad_norm": 1.419362126867698, "learning_rate": 6.516442747755731e-06, "loss": 0.5143, "step": 17539 }, { "epoch": 0.6248997987067353, "grad_norm": 1.1569560889505275, "learning_rate": 6.5153611140685865e-06, "loss": 0.4303, "step": 17540 }, { "epoch": 0.6249354258332295, "grad_norm": 1.562882280187316, "learning_rate": 6.5142795267818505e-06, "loss": 0.49, "step": 17541 }, { "epoch": 0.6249710529597235, "grad_norm": 1.3034328626545102, "learning_rate": 6.513197985909914e-06, "loss": 0.5326, "step": 17542 }, { "epoch": 0.6250066800862176, "grad_norm": 1.2343421518134325, "learning_rate": 6.512116491467184e-06, "loss": 0.4405, "step": 17543 }, { "epoch": 0.6250423072127118, "grad_norm": 1.2656903367504924, "learning_rate": 6.511035043468064e-06, "loss": 0.4751, "step": 17544 }, { "epoch": 0.6250779343392059, "grad_norm": 1.1377648019433513, "learning_rate": 6.509953641926947e-06, "loss": 0.4332, "step": 17545 }, { "epoch": 0.6251135614657, "grad_norm": 1.489953780057507, "learning_rate": 6.508872286858241e-06, "loss": 0.5233, "step": 17546 }, { "epoch": 0.6251491885921941, "grad_norm": 1.4605623484600172, "learning_rate": 6.507790978276337e-06, "loss": 0.5537, "step": 17547 }, { "epoch": 0.6251848157186882, "grad_norm": 1.5686582235240472, "learning_rate": 6.506709716195639e-06, "loss": 0.5738, "step": 17548 }, { "epoch": 0.6252204428451823, "grad_norm": 1.6407513097346926, "learning_rate": 6.505628500630541e-06, "loss": 0.5585, "step": 17549 }, { "epoch": 0.6252560699716765, "grad_norm": 1.5546826529679014, "learning_rate": 6.5045473315954455e-06, "loss": 0.5509, "step": 17550 }, { "epoch": 0.6252916970981706, "grad_norm": 1.3704976496189023, "learning_rate": 6.503466209104743e-06, "loss": 0.4461, "step": 17551 }, { "epoch": 0.6253273242246646, "grad_norm": 1.2182118744334773, "learning_rate": 6.502385133172833e-06, "loss": 0.4778, "step": 17552 }, { "epoch": 0.6253629513511588, "grad_norm": 1.0399757080051604, "learning_rate": 6.5013041038141055e-06, "loss": 0.4443, "step": 17553 }, { "epoch": 0.6253985784776529, "grad_norm": 1.178924506635006, "learning_rate": 6.5002231210429636e-06, "loss": 0.4212, "step": 17554 }, { "epoch": 0.625434205604147, "grad_norm": 3.6077047283006736, "learning_rate": 6.499142184873795e-06, "loss": 0.5114, "step": 17555 }, { "epoch": 0.6254698327306412, "grad_norm": 1.101692992344937, "learning_rate": 6.4980612953209946e-06, "loss": 0.4816, "step": 17556 }, { "epoch": 0.6255054598571352, "grad_norm": 1.104538146909884, "learning_rate": 6.49698045239896e-06, "loss": 0.4119, "step": 17557 }, { "epoch": 0.6255410869836293, "grad_norm": 1.2046654985654555, "learning_rate": 6.495899656122075e-06, "loss": 0.3861, "step": 17558 }, { "epoch": 0.6255767141101235, "grad_norm": 1.5576334716699733, "learning_rate": 6.494818906504737e-06, "loss": 0.6726, "step": 17559 }, { "epoch": 0.6256123412366176, "grad_norm": 1.2030263720405623, "learning_rate": 6.493738203561333e-06, "loss": 0.4108, "step": 17560 }, { "epoch": 0.6256479683631117, "grad_norm": 1.1632337011032783, "learning_rate": 6.49265754730626e-06, "loss": 0.4805, "step": 17561 }, { "epoch": 0.6256835954896058, "grad_norm": 1.079697343916844, "learning_rate": 6.4915769377538995e-06, "loss": 0.4238, "step": 17562 }, { "epoch": 0.6257192226160999, "grad_norm": 1.1551070545655588, "learning_rate": 6.490496374918647e-06, "loss": 0.4732, "step": 17563 }, { "epoch": 0.625754849742594, "grad_norm": 1.4439457113585714, "learning_rate": 6.4894158588148844e-06, "loss": 0.5564, "step": 17564 }, { "epoch": 0.6257904768690882, "grad_norm": 1.0704168429083474, "learning_rate": 6.488335389457007e-06, "loss": 0.4359, "step": 17565 }, { "epoch": 0.6258261039955823, "grad_norm": 1.128223557967159, "learning_rate": 6.4872549668593954e-06, "loss": 0.4276, "step": 17566 }, { "epoch": 0.6258617311220763, "grad_norm": 1.7315775168789809, "learning_rate": 6.486174591036439e-06, "loss": 0.5547, "step": 17567 }, { "epoch": 0.6258973582485705, "grad_norm": 1.175212442241604, "learning_rate": 6.485094262002529e-06, "loss": 0.4513, "step": 17568 }, { "epoch": 0.6259329853750646, "grad_norm": 1.0339495426049465, "learning_rate": 6.484013979772041e-06, "loss": 0.3742, "step": 17569 }, { "epoch": 0.6259686125015587, "grad_norm": 1.1606397086551032, "learning_rate": 6.482933744359366e-06, "loss": 0.4265, "step": 17570 }, { "epoch": 0.6260042396280528, "grad_norm": 1.2106926548778558, "learning_rate": 6.481853555778886e-06, "loss": 0.4279, "step": 17571 }, { "epoch": 0.6260398667545469, "grad_norm": 1.1781984088824933, "learning_rate": 6.480773414044988e-06, "loss": 0.4262, "step": 17572 }, { "epoch": 0.626075493881041, "grad_norm": 1.1405673600561175, "learning_rate": 6.479693319172047e-06, "loss": 0.4156, "step": 17573 }, { "epoch": 0.6261111210075352, "grad_norm": 1.0633836035689257, "learning_rate": 6.478613271174453e-06, "loss": 0.4238, "step": 17574 }, { "epoch": 0.6261467481340293, "grad_norm": 1.0632837297176343, "learning_rate": 6.477533270066583e-06, "loss": 0.4128, "step": 17575 }, { "epoch": 0.6261823752605233, "grad_norm": 1.010150347836343, "learning_rate": 6.476453315862823e-06, "loss": 0.4503, "step": 17576 }, { "epoch": 0.6262180023870175, "grad_norm": 1.1477591639396725, "learning_rate": 6.4753734085775435e-06, "loss": 0.4186, "step": 17577 }, { "epoch": 0.6262536295135116, "grad_norm": 1.045672515819364, "learning_rate": 6.474293548225132e-06, "loss": 0.3728, "step": 17578 }, { "epoch": 0.6262892566400057, "grad_norm": 1.9780904045899528, "learning_rate": 6.473213734819972e-06, "loss": 0.5904, "step": 17579 }, { "epoch": 0.6263248837664999, "grad_norm": 1.174109724429984, "learning_rate": 6.4721339683764306e-06, "loss": 0.3959, "step": 17580 }, { "epoch": 0.6263605108929939, "grad_norm": 1.0166620996011564, "learning_rate": 6.471054248908895e-06, "loss": 0.4339, "step": 17581 }, { "epoch": 0.626396138019488, "grad_norm": 1.4042603193015981, "learning_rate": 6.469974576431734e-06, "loss": 0.5125, "step": 17582 }, { "epoch": 0.6264317651459822, "grad_norm": 1.099309114523947, "learning_rate": 6.468894950959336e-06, "loss": 0.3985, "step": 17583 }, { "epoch": 0.6264673922724763, "grad_norm": 1.3069955433445517, "learning_rate": 6.467815372506064e-06, "loss": 0.5157, "step": 17584 }, { "epoch": 0.6265030193989704, "grad_norm": 1.2193762836057453, "learning_rate": 6.466735841086303e-06, "loss": 0.4068, "step": 17585 }, { "epoch": 0.6265386465254645, "grad_norm": 1.4790281562525658, "learning_rate": 6.465656356714419e-06, "loss": 0.5223, "step": 17586 }, { "epoch": 0.6265742736519586, "grad_norm": 1.3820970767277871, "learning_rate": 6.464576919404797e-06, "loss": 0.5029, "step": 17587 }, { "epoch": 0.6266099007784527, "grad_norm": 2.1643786729688617, "learning_rate": 6.463497529171799e-06, "loss": 0.5498, "step": 17588 }, { "epoch": 0.6266455279049469, "grad_norm": 1.553311708614215, "learning_rate": 6.462418186029802e-06, "loss": 0.5894, "step": 17589 }, { "epoch": 0.626681155031441, "grad_norm": 1.233684020992471, "learning_rate": 6.461338889993185e-06, "loss": 0.4489, "step": 17590 }, { "epoch": 0.626716782157935, "grad_norm": 1.4876646378393086, "learning_rate": 6.460259641076312e-06, "loss": 0.6181, "step": 17591 }, { "epoch": 0.6267524092844292, "grad_norm": 1.7475011927273951, "learning_rate": 6.459180439293555e-06, "loss": 0.5131, "step": 17592 }, { "epoch": 0.6267880364109233, "grad_norm": 1.5494220019574727, "learning_rate": 6.4581012846592865e-06, "loss": 0.5066, "step": 17593 }, { "epoch": 0.6268236635374174, "grad_norm": 1.4515665851302517, "learning_rate": 6.4570221771878775e-06, "loss": 0.4882, "step": 17594 }, { "epoch": 0.6268592906639116, "grad_norm": 1.1604339458121296, "learning_rate": 6.4559431168936905e-06, "loss": 0.3888, "step": 17595 }, { "epoch": 0.6268949177904056, "grad_norm": 1.2123346242959918, "learning_rate": 6.4548641037911e-06, "loss": 0.4003, "step": 17596 }, { "epoch": 0.6269305449168997, "grad_norm": 1.096812819845326, "learning_rate": 6.453785137894469e-06, "loss": 0.4669, "step": 17597 }, { "epoch": 0.6269661720433939, "grad_norm": 1.0991215998637476, "learning_rate": 6.4527062192181735e-06, "loss": 0.3835, "step": 17598 }, { "epoch": 0.627001799169888, "grad_norm": 1.3544219264149102, "learning_rate": 6.451627347776569e-06, "loss": 0.4608, "step": 17599 }, { "epoch": 0.627037426296382, "grad_norm": 1.196340894747804, "learning_rate": 6.450548523584028e-06, "loss": 0.3769, "step": 17600 }, { "epoch": 0.6270730534228762, "grad_norm": 1.871079738014273, "learning_rate": 6.449469746654913e-06, "loss": 0.5061, "step": 17601 }, { "epoch": 0.6271086805493703, "grad_norm": 1.2970532140741124, "learning_rate": 6.4483910170035905e-06, "loss": 0.4417, "step": 17602 }, { "epoch": 0.6271443076758644, "grad_norm": 1.1095623070482976, "learning_rate": 6.447312334644425e-06, "loss": 0.449, "step": 17603 }, { "epoch": 0.6271799348023586, "grad_norm": 1.0313817899953615, "learning_rate": 6.446233699591778e-06, "loss": 0.4384, "step": 17604 }, { "epoch": 0.6272155619288526, "grad_norm": 1.1513322361068268, "learning_rate": 6.445155111860016e-06, "loss": 0.4542, "step": 17605 }, { "epoch": 0.6272511890553467, "grad_norm": 1.3454862168334794, "learning_rate": 6.444076571463495e-06, "loss": 0.5979, "step": 17606 }, { "epoch": 0.6272868161818409, "grad_norm": 1.0728930581095741, "learning_rate": 6.442998078416583e-06, "loss": 0.3985, "step": 17607 }, { "epoch": 0.627322443308335, "grad_norm": 1.138379804816638, "learning_rate": 6.441919632733634e-06, "loss": 0.456, "step": 17608 }, { "epoch": 0.627358070434829, "grad_norm": 1.3592817991279893, "learning_rate": 6.440841234429018e-06, "loss": 0.4215, "step": 17609 }, { "epoch": 0.6273936975613232, "grad_norm": 1.0989794301444964, "learning_rate": 6.439762883517082e-06, "loss": 0.408, "step": 17610 }, { "epoch": 0.6274293246878173, "grad_norm": 1.2738614903662224, "learning_rate": 6.438684580012196e-06, "loss": 0.5975, "step": 17611 }, { "epoch": 0.6274649518143114, "grad_norm": 1.073146322685353, "learning_rate": 6.437606323928711e-06, "loss": 0.4325, "step": 17612 }, { "epoch": 0.6275005789408056, "grad_norm": 1.4199566573539746, "learning_rate": 6.436528115280988e-06, "loss": 0.5948, "step": 17613 }, { "epoch": 0.6275362060672997, "grad_norm": 1.0608504251785018, "learning_rate": 6.435449954083388e-06, "loss": 0.4191, "step": 17614 }, { "epoch": 0.6275718331937937, "grad_norm": 1.045976784506359, "learning_rate": 6.434371840350257e-06, "loss": 0.3807, "step": 17615 }, { "epoch": 0.6276074603202879, "grad_norm": 1.0898956474157824, "learning_rate": 6.433293774095963e-06, "loss": 0.4322, "step": 17616 }, { "epoch": 0.627643087446782, "grad_norm": 1.5682419892559638, "learning_rate": 6.432215755334852e-06, "loss": 0.4408, "step": 17617 }, { "epoch": 0.6276787145732761, "grad_norm": 1.6832118765054092, "learning_rate": 6.431137784081283e-06, "loss": 0.5419, "step": 17618 }, { "epoch": 0.6277143416997703, "grad_norm": 1.7968114928735237, "learning_rate": 6.430059860349607e-06, "loss": 0.4779, "step": 17619 }, { "epoch": 0.6277499688262643, "grad_norm": 1.2276847949848377, "learning_rate": 6.428981984154183e-06, "loss": 0.3872, "step": 17620 }, { "epoch": 0.6277855959527584, "grad_norm": 1.197752285347617, "learning_rate": 6.427904155509357e-06, "loss": 0.47, "step": 17621 }, { "epoch": 0.6278212230792526, "grad_norm": 1.7847973740529917, "learning_rate": 6.4268263744294866e-06, "loss": 0.5063, "step": 17622 }, { "epoch": 0.6278568502057467, "grad_norm": 1.964344282841041, "learning_rate": 6.425748640928917e-06, "loss": 0.5441, "step": 17623 }, { "epoch": 0.6278924773322407, "grad_norm": 1.1174534160229037, "learning_rate": 6.424670955022003e-06, "loss": 0.4334, "step": 17624 }, { "epoch": 0.6279281044587349, "grad_norm": 1.953563643440689, "learning_rate": 6.423593316723098e-06, "loss": 0.5041, "step": 17625 }, { "epoch": 0.627963731585229, "grad_norm": 1.1606504282702426, "learning_rate": 6.422515726046543e-06, "loss": 0.4934, "step": 17626 }, { "epoch": 0.6279993587117231, "grad_norm": 1.015904693321628, "learning_rate": 6.421438183006695e-06, "loss": 0.3925, "step": 17627 }, { "epoch": 0.6280349858382173, "grad_norm": 1.0366509033727176, "learning_rate": 6.420360687617897e-06, "loss": 0.4359, "step": 17628 }, { "epoch": 0.6280706129647113, "grad_norm": 1.1796491135086236, "learning_rate": 6.419283239894499e-06, "loss": 0.4333, "step": 17629 }, { "epoch": 0.6281062400912054, "grad_norm": 1.6562492473181003, "learning_rate": 6.418205839850847e-06, "loss": 0.5857, "step": 17630 }, { "epoch": 0.6281418672176996, "grad_norm": 1.146391503612031, "learning_rate": 6.417128487501291e-06, "loss": 0.3992, "step": 17631 }, { "epoch": 0.6281774943441937, "grad_norm": 1.5017022184219717, "learning_rate": 6.416051182860169e-06, "loss": 0.5544, "step": 17632 }, { "epoch": 0.6282131214706878, "grad_norm": 1.1388691030922475, "learning_rate": 6.4149739259418335e-06, "loss": 0.4706, "step": 17633 }, { "epoch": 0.6282487485971819, "grad_norm": 1.0139576685932457, "learning_rate": 6.413896716760624e-06, "loss": 0.3906, "step": 17634 }, { "epoch": 0.628284375723676, "grad_norm": 1.2942659129436966, "learning_rate": 6.412819555330889e-06, "loss": 0.4522, "step": 17635 }, { "epoch": 0.6283200028501701, "grad_norm": 2.0229794382600033, "learning_rate": 6.411742441666966e-06, "loss": 0.5391, "step": 17636 }, { "epoch": 0.6283556299766643, "grad_norm": 1.09645235689895, "learning_rate": 6.4106653757831994e-06, "loss": 0.4574, "step": 17637 }, { "epoch": 0.6283912571031584, "grad_norm": 1.4956486245418599, "learning_rate": 6.409588357693935e-06, "loss": 0.5005, "step": 17638 }, { "epoch": 0.6284268842296524, "grad_norm": 1.1238524382898198, "learning_rate": 6.408511387413508e-06, "loss": 0.4582, "step": 17639 }, { "epoch": 0.6284625113561466, "grad_norm": 1.2526433624358424, "learning_rate": 6.407434464956266e-06, "loss": 0.5273, "step": 17640 }, { "epoch": 0.6284981384826407, "grad_norm": 1.0234818192980772, "learning_rate": 6.406357590336543e-06, "loss": 0.4404, "step": 17641 }, { "epoch": 0.6285337656091348, "grad_norm": 1.0535524996794796, "learning_rate": 6.405280763568683e-06, "loss": 0.3938, "step": 17642 }, { "epoch": 0.628569392735629, "grad_norm": 1.0291651899764263, "learning_rate": 6.404203984667019e-06, "loss": 0.4284, "step": 17643 }, { "epoch": 0.628605019862123, "grad_norm": 2.22347449805676, "learning_rate": 6.403127253645893e-06, "loss": 0.4946, "step": 17644 }, { "epoch": 0.6286406469886171, "grad_norm": 1.5713213045489411, "learning_rate": 6.402050570519642e-06, "loss": 0.5216, "step": 17645 }, { "epoch": 0.6286762741151113, "grad_norm": 1.433007787343852, "learning_rate": 6.400973935302605e-06, "loss": 0.5003, "step": 17646 }, { "epoch": 0.6287119012416054, "grad_norm": 1.0726416535125998, "learning_rate": 6.399897348009112e-06, "loss": 0.3922, "step": 17647 }, { "epoch": 0.6287475283680994, "grad_norm": 1.089010136704305, "learning_rate": 6.398820808653503e-06, "loss": 0.4483, "step": 17648 }, { "epoch": 0.6287831554945936, "grad_norm": 1.393968799674238, "learning_rate": 6.397744317250112e-06, "loss": 0.5442, "step": 17649 }, { "epoch": 0.6288187826210877, "grad_norm": 1.5308766365936037, "learning_rate": 6.396667873813273e-06, "loss": 0.4854, "step": 17650 }, { "epoch": 0.6288544097475818, "grad_norm": 1.1912460757802597, "learning_rate": 6.395591478357323e-06, "loss": 0.4465, "step": 17651 }, { "epoch": 0.628890036874076, "grad_norm": 1.5491909339993852, "learning_rate": 6.394515130896586e-06, "loss": 0.5794, "step": 17652 }, { "epoch": 0.62892566400057, "grad_norm": 1.0093819443325498, "learning_rate": 6.393438831445407e-06, "loss": 0.3844, "step": 17653 }, { "epoch": 0.6289612911270641, "grad_norm": 1.6709679657940568, "learning_rate": 6.3923625800181055e-06, "loss": 0.605, "step": 17654 }, { "epoch": 0.6289969182535583, "grad_norm": 1.2955998327698224, "learning_rate": 6.39128637662902e-06, "loss": 0.6061, "step": 17655 }, { "epoch": 0.6290325453800524, "grad_norm": 1.57736494297832, "learning_rate": 6.390210221292476e-06, "loss": 0.5549, "step": 17656 }, { "epoch": 0.6290681725065465, "grad_norm": 1.0997225587945345, "learning_rate": 6.3891341140228125e-06, "loss": 0.3909, "step": 17657 }, { "epoch": 0.6291037996330406, "grad_norm": 1.1883457874627728, "learning_rate": 6.388058054834345e-06, "loss": 0.4647, "step": 17658 }, { "epoch": 0.6291394267595347, "grad_norm": 1.6005260093697495, "learning_rate": 6.386982043741412e-06, "loss": 0.5687, "step": 17659 }, { "epoch": 0.6291750538860288, "grad_norm": 1.7271181057282599, "learning_rate": 6.385906080758339e-06, "loss": 0.6117, "step": 17660 }, { "epoch": 0.629210681012523, "grad_norm": 1.0576556647271393, "learning_rate": 6.38483016589945e-06, "loss": 0.418, "step": 17661 }, { "epoch": 0.629246308139017, "grad_norm": 1.4854453520239403, "learning_rate": 6.383754299179079e-06, "loss": 0.4887, "step": 17662 }, { "epoch": 0.6292819352655111, "grad_norm": 1.332361667582487, "learning_rate": 6.382678480611544e-06, "loss": 0.6029, "step": 17663 }, { "epoch": 0.6293175623920053, "grad_norm": 1.3801550177775648, "learning_rate": 6.3816027102111745e-06, "loss": 0.5294, "step": 17664 }, { "epoch": 0.6293531895184994, "grad_norm": 1.3579593154241025, "learning_rate": 6.380526987992291e-06, "loss": 0.5145, "step": 17665 }, { "epoch": 0.6293888166449935, "grad_norm": 1.40647251636861, "learning_rate": 6.379451313969224e-06, "loss": 0.496, "step": 17666 }, { "epoch": 0.6294244437714877, "grad_norm": 1.0681007411930579, "learning_rate": 6.378375688156291e-06, "loss": 0.3807, "step": 17667 }, { "epoch": 0.6294600708979817, "grad_norm": 1.1133941828994443, "learning_rate": 6.377300110567821e-06, "loss": 0.4363, "step": 17668 }, { "epoch": 0.6294956980244758, "grad_norm": 1.3162083580695276, "learning_rate": 6.376224581218128e-06, "loss": 0.4248, "step": 17669 }, { "epoch": 0.62953132515097, "grad_norm": 1.5373244189552828, "learning_rate": 6.375149100121537e-06, "loss": 0.5535, "step": 17670 }, { "epoch": 0.6295669522774641, "grad_norm": 1.477069579417697, "learning_rate": 6.374073667292373e-06, "loss": 0.4755, "step": 17671 }, { "epoch": 0.6296025794039581, "grad_norm": 1.5902781450225825, "learning_rate": 6.372998282744949e-06, "loss": 0.4921, "step": 17672 }, { "epoch": 0.6296382065304523, "grad_norm": 1.1035507515497427, "learning_rate": 6.3719229464935915e-06, "loss": 0.3998, "step": 17673 }, { "epoch": 0.6296738336569464, "grad_norm": 1.030845930060635, "learning_rate": 6.3708476585526135e-06, "loss": 0.3884, "step": 17674 }, { "epoch": 0.6297094607834405, "grad_norm": 1.229263754734168, "learning_rate": 6.369772418936337e-06, "loss": 0.4457, "step": 17675 }, { "epoch": 0.6297450879099347, "grad_norm": 1.5965133761018473, "learning_rate": 6.368697227659075e-06, "loss": 0.4645, "step": 17676 }, { "epoch": 0.6297807150364287, "grad_norm": 1.0936500957565805, "learning_rate": 6.367622084735153e-06, "loss": 0.4313, "step": 17677 }, { "epoch": 0.6298163421629228, "grad_norm": 1.1144189892307312, "learning_rate": 6.366546990178876e-06, "loss": 0.4211, "step": 17678 }, { "epoch": 0.629851969289417, "grad_norm": 1.36101552894025, "learning_rate": 6.365471944004571e-06, "loss": 0.5426, "step": 17679 }, { "epoch": 0.6298875964159111, "grad_norm": 1.2982173537371524, "learning_rate": 6.364396946226543e-06, "loss": 0.4843, "step": 17680 }, { "epoch": 0.6299232235424052, "grad_norm": 1.6698717011214628, "learning_rate": 6.363321996859113e-06, "loss": 0.4415, "step": 17681 }, { "epoch": 0.6299588506688993, "grad_norm": 1.0040562496565584, "learning_rate": 6.3622470959165905e-06, "loss": 0.3968, "step": 17682 }, { "epoch": 0.6299944777953934, "grad_norm": 1.8155160524403882, "learning_rate": 6.361172243413291e-06, "loss": 0.5115, "step": 17683 }, { "epoch": 0.6300301049218875, "grad_norm": 1.1171773492235448, "learning_rate": 6.36009743936353e-06, "loss": 0.4115, "step": 17684 }, { "epoch": 0.6300657320483817, "grad_norm": 1.2126697033541307, "learning_rate": 6.359022683781611e-06, "loss": 0.5218, "step": 17685 }, { "epoch": 0.6301013591748758, "grad_norm": 1.0818017911657298, "learning_rate": 6.357947976681853e-06, "loss": 0.4397, "step": 17686 }, { "epoch": 0.6301369863013698, "grad_norm": 1.0885186211131637, "learning_rate": 6.35687331807856e-06, "loss": 0.4189, "step": 17687 }, { "epoch": 0.630172613427864, "grad_norm": 1.2392932751538184, "learning_rate": 6.355798707986052e-06, "loss": 0.5145, "step": 17688 }, { "epoch": 0.6302082405543581, "grad_norm": 1.0678638928682556, "learning_rate": 6.3547241464186245e-06, "loss": 0.412, "step": 17689 }, { "epoch": 0.6302438676808522, "grad_norm": 1.3686183508979826, "learning_rate": 6.353649633390596e-06, "loss": 0.5449, "step": 17690 }, { "epoch": 0.6302794948073464, "grad_norm": 1.7093777568180406, "learning_rate": 6.352575168916268e-06, "loss": 0.4302, "step": 17691 }, { "epoch": 0.6303151219338404, "grad_norm": 1.3275010463739725, "learning_rate": 6.351500753009956e-06, "loss": 0.5185, "step": 17692 }, { "epoch": 0.6303507490603345, "grad_norm": 1.0837203390178434, "learning_rate": 6.3504263856859575e-06, "loss": 0.4336, "step": 17693 }, { "epoch": 0.6303863761868287, "grad_norm": 1.0538143707684575, "learning_rate": 6.349352066958583e-06, "loss": 0.4082, "step": 17694 }, { "epoch": 0.6304220033133228, "grad_norm": 1.1328782483093567, "learning_rate": 6.3482777968421415e-06, "loss": 0.4439, "step": 17695 }, { "epoch": 0.6304576304398168, "grad_norm": 1.194924532673151, "learning_rate": 6.347203575350929e-06, "loss": 0.4726, "step": 17696 }, { "epoch": 0.630493257566311, "grad_norm": 1.1568457122261429, "learning_rate": 6.346129402499256e-06, "loss": 0.4054, "step": 17697 }, { "epoch": 0.6305288846928051, "grad_norm": 1.8015662466540303, "learning_rate": 6.345055278301422e-06, "loss": 0.5515, "step": 17698 }, { "epoch": 0.6305645118192992, "grad_norm": 1.1084486419912551, "learning_rate": 6.343981202771736e-06, "loss": 0.4354, "step": 17699 }, { "epoch": 0.6306001389457934, "grad_norm": 1.0639878600454895, "learning_rate": 6.342907175924492e-06, "loss": 0.4316, "step": 17700 }, { "epoch": 0.6306357660722874, "grad_norm": 1.5115834152355778, "learning_rate": 6.341833197773997e-06, "loss": 0.5269, "step": 17701 }, { "epoch": 0.6306713931987815, "grad_norm": 1.2900014480357984, "learning_rate": 6.340759268334546e-06, "loss": 0.4512, "step": 17702 }, { "epoch": 0.6307070203252757, "grad_norm": 1.5354095849993803, "learning_rate": 6.339685387620449e-06, "loss": 0.5203, "step": 17703 }, { "epoch": 0.6307426474517698, "grad_norm": 1.8971489533478991, "learning_rate": 6.338611555645993e-06, "loss": 0.564, "step": 17704 }, { "epoch": 0.6307782745782639, "grad_norm": 1.085409895462143, "learning_rate": 6.337537772425486e-06, "loss": 0.4078, "step": 17705 }, { "epoch": 0.630813901704758, "grad_norm": 1.29921638066297, "learning_rate": 6.3364640379732266e-06, "loss": 0.5985, "step": 17706 }, { "epoch": 0.6308495288312521, "grad_norm": 1.2102589034467925, "learning_rate": 6.3353903523035056e-06, "loss": 0.4745, "step": 17707 }, { "epoch": 0.6308851559577462, "grad_norm": 1.4481232542804305, "learning_rate": 6.334316715430625e-06, "loss": 0.5247, "step": 17708 }, { "epoch": 0.6309207830842404, "grad_norm": 1.4506053743449718, "learning_rate": 6.333243127368878e-06, "loss": 0.5541, "step": 17709 }, { "epoch": 0.6309564102107345, "grad_norm": 1.4978726193495082, "learning_rate": 6.332169588132565e-06, "loss": 0.4871, "step": 17710 }, { "epoch": 0.6309920373372285, "grad_norm": 1.4853517981146525, "learning_rate": 6.331096097735973e-06, "loss": 0.4872, "step": 17711 }, { "epoch": 0.6310276644637227, "grad_norm": 1.91082951784313, "learning_rate": 6.330022656193405e-06, "loss": 0.5524, "step": 17712 }, { "epoch": 0.6310632915902168, "grad_norm": 1.1078697142242413, "learning_rate": 6.3289492635191475e-06, "loss": 0.468, "step": 17713 }, { "epoch": 0.6310989187167109, "grad_norm": 1.0754941582805126, "learning_rate": 6.3278759197275e-06, "loss": 0.4274, "step": 17714 }, { "epoch": 0.631134545843205, "grad_norm": 1.2822625685723035, "learning_rate": 6.326802624832749e-06, "loss": 0.562, "step": 17715 }, { "epoch": 0.6311701729696991, "grad_norm": 1.1486396592571149, "learning_rate": 6.3257293788491835e-06, "loss": 0.4401, "step": 17716 }, { "epoch": 0.6312058000961932, "grad_norm": 1.1170097139885555, "learning_rate": 6.3246561817911076e-06, "loss": 0.4276, "step": 17717 }, { "epoch": 0.6312414272226874, "grad_norm": 1.8761583191967441, "learning_rate": 6.323583033672799e-06, "loss": 0.5496, "step": 17718 }, { "epoch": 0.6312770543491815, "grad_norm": 2.6813319457723708, "learning_rate": 6.322509934508554e-06, "loss": 0.5837, "step": 17719 }, { "epoch": 0.6313126814756755, "grad_norm": 1.342490640104238, "learning_rate": 6.321436884312657e-06, "loss": 0.4716, "step": 17720 }, { "epoch": 0.6313483086021697, "grad_norm": 1.1509792877473137, "learning_rate": 6.3203638830994044e-06, "loss": 0.409, "step": 17721 }, { "epoch": 0.6313839357286638, "grad_norm": 1.0864396393761164, "learning_rate": 6.319290930883073e-06, "loss": 0.4221, "step": 17722 }, { "epoch": 0.6314195628551579, "grad_norm": 1.0377597898540019, "learning_rate": 6.3182180276779594e-06, "loss": 0.3896, "step": 17723 }, { "epoch": 0.6314551899816521, "grad_norm": 1.8475070462569214, "learning_rate": 6.317145173498343e-06, "loss": 0.5179, "step": 17724 }, { "epoch": 0.6314908171081461, "grad_norm": 1.6833836456573836, "learning_rate": 6.316072368358517e-06, "loss": 0.5954, "step": 17725 }, { "epoch": 0.6315264442346402, "grad_norm": 1.6986439203184602, "learning_rate": 6.3149996122727585e-06, "loss": 0.5535, "step": 17726 }, { "epoch": 0.6315620713611344, "grad_norm": 1.3083172337854556, "learning_rate": 6.313926905255358e-06, "loss": 0.5027, "step": 17727 }, { "epoch": 0.6315976984876285, "grad_norm": 1.6201776205018679, "learning_rate": 6.312854247320594e-06, "loss": 0.596, "step": 17728 }, { "epoch": 0.6316333256141226, "grad_norm": 1.4099229127540895, "learning_rate": 6.311781638482754e-06, "loss": 0.4365, "step": 17729 }, { "epoch": 0.6316689527406167, "grad_norm": 1.3476429358122384, "learning_rate": 6.3107090787561206e-06, "loss": 0.5065, "step": 17730 }, { "epoch": 0.6317045798671108, "grad_norm": 1.5513101692757043, "learning_rate": 6.309636568154973e-06, "loss": 0.5697, "step": 17731 }, { "epoch": 0.6317402069936049, "grad_norm": 1.7840312003136267, "learning_rate": 6.308564106693597e-06, "loss": 0.6029, "step": 17732 }, { "epoch": 0.6317758341200991, "grad_norm": 1.143247182616068, "learning_rate": 6.307491694386267e-06, "loss": 0.4684, "step": 17733 }, { "epoch": 0.6318114612465932, "grad_norm": 1.0960033154395719, "learning_rate": 6.306419331247269e-06, "loss": 0.4371, "step": 17734 }, { "epoch": 0.6318470883730872, "grad_norm": 1.149511175503862, "learning_rate": 6.305347017290875e-06, "loss": 0.417, "step": 17735 }, { "epoch": 0.6318827154995814, "grad_norm": 1.2296436459935502, "learning_rate": 6.304274752531374e-06, "loss": 0.4312, "step": 17736 }, { "epoch": 0.6319183426260755, "grad_norm": 1.0836794669841405, "learning_rate": 6.303202536983033e-06, "loss": 0.452, "step": 17737 }, { "epoch": 0.6319539697525696, "grad_norm": 1.104288201305625, "learning_rate": 6.302130370660135e-06, "loss": 0.433, "step": 17738 }, { "epoch": 0.6319895968790638, "grad_norm": 1.6108609833207839, "learning_rate": 6.3010582535769546e-06, "loss": 0.5047, "step": 17739 }, { "epoch": 0.6320252240055578, "grad_norm": 1.186199782187594, "learning_rate": 6.299986185747769e-06, "loss": 0.4403, "step": 17740 }, { "epoch": 0.6320608511320519, "grad_norm": 1.8586913687699913, "learning_rate": 6.298914167186856e-06, "loss": 0.5154, "step": 17741 }, { "epoch": 0.6320964782585461, "grad_norm": 1.1994091911010916, "learning_rate": 6.297842197908486e-06, "loss": 0.4326, "step": 17742 }, { "epoch": 0.6321321053850402, "grad_norm": 1.407776933675106, "learning_rate": 6.296770277926938e-06, "loss": 0.4904, "step": 17743 }, { "epoch": 0.6321677325115342, "grad_norm": 1.0625790335170686, "learning_rate": 6.295698407256479e-06, "loss": 0.4148, "step": 17744 }, { "epoch": 0.6322033596380284, "grad_norm": 1.575175977724529, "learning_rate": 6.2946265859113855e-06, "loss": 0.5411, "step": 17745 }, { "epoch": 0.6322389867645225, "grad_norm": 1.1296463340151586, "learning_rate": 6.293554813905927e-06, "loss": 0.4272, "step": 17746 }, { "epoch": 0.6322746138910166, "grad_norm": 0.9848251696109449, "learning_rate": 6.292483091254382e-06, "loss": 0.4057, "step": 17747 }, { "epoch": 0.6323102410175108, "grad_norm": 2.092291169507392, "learning_rate": 6.291411417971012e-06, "loss": 0.5836, "step": 17748 }, { "epoch": 0.6323458681440048, "grad_norm": 1.0269499867584768, "learning_rate": 6.290339794070092e-06, "loss": 0.4277, "step": 17749 }, { "epoch": 0.6323814952704989, "grad_norm": 1.4645492645698146, "learning_rate": 6.289268219565889e-06, "loss": 0.5111, "step": 17750 }, { "epoch": 0.6324171223969931, "grad_norm": 1.558511435759378, "learning_rate": 6.288196694472673e-06, "loss": 0.5958, "step": 17751 }, { "epoch": 0.6324527495234872, "grad_norm": 1.1163364338185104, "learning_rate": 6.287125218804717e-06, "loss": 0.4355, "step": 17752 }, { "epoch": 0.6324883766499813, "grad_norm": 1.1218665321464707, "learning_rate": 6.286053792576276e-06, "loss": 0.4146, "step": 17753 }, { "epoch": 0.6325240037764754, "grad_norm": 1.2921733287404826, "learning_rate": 6.284982415801634e-06, "loss": 0.4316, "step": 17754 }, { "epoch": 0.6325596309029695, "grad_norm": 1.1496365289237995, "learning_rate": 6.283911088495041e-06, "loss": 0.5033, "step": 17755 }, { "epoch": 0.6325952580294636, "grad_norm": 1.252431433127368, "learning_rate": 6.2828398106707734e-06, "loss": 0.4714, "step": 17756 }, { "epoch": 0.6326308851559578, "grad_norm": 1.064227253284118, "learning_rate": 6.2817685823430885e-06, "loss": 0.3817, "step": 17757 }, { "epoch": 0.6326665122824519, "grad_norm": 1.27754424950843, "learning_rate": 6.280697403526258e-06, "loss": 0.4271, "step": 17758 }, { "epoch": 0.6327021394089459, "grad_norm": 1.1753233443874642, "learning_rate": 6.279626274234537e-06, "loss": 0.4314, "step": 17759 }, { "epoch": 0.6327377665354401, "grad_norm": 1.3511355066486903, "learning_rate": 6.278555194482194e-06, "loss": 0.4993, "step": 17760 }, { "epoch": 0.6327733936619342, "grad_norm": 1.4608381345912278, "learning_rate": 6.277484164283489e-06, "loss": 0.5895, "step": 17761 }, { "epoch": 0.6328090207884283, "grad_norm": 1.9365227132344078, "learning_rate": 6.2764131836526844e-06, "loss": 0.406, "step": 17762 }, { "epoch": 0.6328446479149225, "grad_norm": 1.4176656318860479, "learning_rate": 6.275342252604044e-06, "loss": 0.5316, "step": 17763 }, { "epoch": 0.6328802750414165, "grad_norm": 1.1720377319020183, "learning_rate": 6.274271371151821e-06, "loss": 0.4938, "step": 17764 }, { "epoch": 0.6329159021679106, "grad_norm": 1.0466580382434607, "learning_rate": 6.273200539310282e-06, "loss": 0.4625, "step": 17765 }, { "epoch": 0.6329515292944048, "grad_norm": 1.1627942209471345, "learning_rate": 6.272129757093679e-06, "loss": 0.4263, "step": 17766 }, { "epoch": 0.6329871564208989, "grad_norm": 1.8441094399025204, "learning_rate": 6.271059024516277e-06, "loss": 0.5414, "step": 17767 }, { "epoch": 0.6330227835473929, "grad_norm": 1.2058972168759057, "learning_rate": 6.269988341592328e-06, "loss": 0.4395, "step": 17768 }, { "epoch": 0.6330584106738871, "grad_norm": 1.6964980982245714, "learning_rate": 6.2689177083360956e-06, "loss": 0.5147, "step": 17769 }, { "epoch": 0.6330940378003812, "grad_norm": 1.7395713066294936, "learning_rate": 6.267847124761828e-06, "loss": 0.575, "step": 17770 }, { "epoch": 0.6331296649268753, "grad_norm": 1.0333877667565827, "learning_rate": 6.2667765908837856e-06, "loss": 0.4456, "step": 17771 }, { "epoch": 0.6331652920533695, "grad_norm": 1.115306104240754, "learning_rate": 6.2657061067162205e-06, "loss": 0.4587, "step": 17772 }, { "epoch": 0.6332009191798635, "grad_norm": 1.1426642761625323, "learning_rate": 6.264635672273393e-06, "loss": 0.3901, "step": 17773 }, { "epoch": 0.6332365463063576, "grad_norm": 1.4347124277636825, "learning_rate": 6.263565287569549e-06, "loss": 0.4982, "step": 17774 }, { "epoch": 0.6332721734328518, "grad_norm": 1.2457789074599381, "learning_rate": 6.262494952618944e-06, "loss": 0.467, "step": 17775 }, { "epoch": 0.6333078005593459, "grad_norm": 1.5268086986379021, "learning_rate": 6.261424667435833e-06, "loss": 0.5249, "step": 17776 }, { "epoch": 0.63334342768584, "grad_norm": 1.1546875742496352, "learning_rate": 6.260354432034463e-06, "loss": 0.435, "step": 17777 }, { "epoch": 0.6333790548123341, "grad_norm": 1.036581120885406, "learning_rate": 6.2592842464290935e-06, "loss": 0.3875, "step": 17778 }, { "epoch": 0.6334146819388282, "grad_norm": 1.2526522106329494, "learning_rate": 6.258214110633962e-06, "loss": 0.5367, "step": 17779 }, { "epoch": 0.6334503090653223, "grad_norm": 1.1591638487758646, "learning_rate": 6.257144024663331e-06, "loss": 0.4488, "step": 17780 }, { "epoch": 0.6334859361918165, "grad_norm": 1.5029778351648622, "learning_rate": 6.256073988531439e-06, "loss": 0.543, "step": 17781 }, { "epoch": 0.6335215633183106, "grad_norm": 1.239820458447706, "learning_rate": 6.2550040022525405e-06, "loss": 0.4318, "step": 17782 }, { "epoch": 0.6335571904448046, "grad_norm": 1.1980342823006234, "learning_rate": 6.25393406584088e-06, "loss": 0.3994, "step": 17783 }, { "epoch": 0.6335928175712988, "grad_norm": 1.6884230571362262, "learning_rate": 6.252864179310709e-06, "loss": 0.5931, "step": 17784 }, { "epoch": 0.6336284446977929, "grad_norm": 1.8477116913144227, "learning_rate": 6.251794342676267e-06, "loss": 0.5916, "step": 17785 }, { "epoch": 0.633664071824287, "grad_norm": 1.9734889640349709, "learning_rate": 6.250724555951804e-06, "loss": 0.5758, "step": 17786 }, { "epoch": 0.6336996989507812, "grad_norm": 1.1809006827126292, "learning_rate": 6.249654819151565e-06, "loss": 0.4825, "step": 17787 }, { "epoch": 0.6337353260772752, "grad_norm": 1.146681603756186, "learning_rate": 6.248585132289792e-06, "loss": 0.4069, "step": 17788 }, { "epoch": 0.6337709532037693, "grad_norm": 1.0919520714321387, "learning_rate": 6.2475154953807336e-06, "loss": 0.4301, "step": 17789 }, { "epoch": 0.6338065803302635, "grad_norm": 1.1367513728317264, "learning_rate": 6.246445908438625e-06, "loss": 0.426, "step": 17790 }, { "epoch": 0.6338422074567576, "grad_norm": 1.7658551803961273, "learning_rate": 6.245376371477716e-06, "loss": 0.526, "step": 17791 }, { "epoch": 0.6338778345832516, "grad_norm": 1.158332211537029, "learning_rate": 6.244306884512241e-06, "loss": 0.4002, "step": 17792 }, { "epoch": 0.6339134617097458, "grad_norm": 1.071391150852011, "learning_rate": 6.24323744755645e-06, "loss": 0.4051, "step": 17793 }, { "epoch": 0.6339490888362399, "grad_norm": 1.0907795689513942, "learning_rate": 6.242168060624572e-06, "loss": 0.3688, "step": 17794 }, { "epoch": 0.633984715962734, "grad_norm": 1.3583162649115221, "learning_rate": 6.241098723730861e-06, "loss": 0.5228, "step": 17795 }, { "epoch": 0.6340203430892282, "grad_norm": 1.6784768046240919, "learning_rate": 6.2400294368895405e-06, "loss": 0.5581, "step": 17796 }, { "epoch": 0.6340559702157222, "grad_norm": 1.3743761711127394, "learning_rate": 6.238960200114857e-06, "loss": 0.4699, "step": 17797 }, { "epoch": 0.6340915973422163, "grad_norm": 1.0473375161205694, "learning_rate": 6.23789101342105e-06, "loss": 0.403, "step": 17798 }, { "epoch": 0.6341272244687105, "grad_norm": 1.1411486691133375, "learning_rate": 6.236821876822352e-06, "loss": 0.4127, "step": 17799 }, { "epoch": 0.6341628515952046, "grad_norm": 1.4623991761115025, "learning_rate": 6.235752790333004e-06, "loss": 0.5108, "step": 17800 }, { "epoch": 0.6341984787216987, "grad_norm": 1.550518651749976, "learning_rate": 6.234683753967234e-06, "loss": 0.5875, "step": 17801 }, { "epoch": 0.6342341058481928, "grad_norm": 1.117935623242803, "learning_rate": 6.233614767739287e-06, "loss": 0.4187, "step": 17802 }, { "epoch": 0.6342697329746869, "grad_norm": 1.0803027898714996, "learning_rate": 6.232545831663386e-06, "loss": 0.4744, "step": 17803 }, { "epoch": 0.634305360101181, "grad_norm": 2.0124998128488563, "learning_rate": 6.231476945753776e-06, "loss": 0.5083, "step": 17804 }, { "epoch": 0.6343409872276752, "grad_norm": 1.1958985913409554, "learning_rate": 6.23040811002468e-06, "loss": 0.4327, "step": 17805 }, { "epoch": 0.6343766143541693, "grad_norm": 1.1747585612291844, "learning_rate": 6.229339324490339e-06, "loss": 0.437, "step": 17806 }, { "epoch": 0.6344122414806633, "grad_norm": 1.8819176814447531, "learning_rate": 6.228270589164977e-06, "loss": 0.5285, "step": 17807 }, { "epoch": 0.6344478686071575, "grad_norm": 1.4377159422959187, "learning_rate": 6.22720190406283e-06, "loss": 0.5991, "step": 17808 }, { "epoch": 0.6344834957336516, "grad_norm": 1.4716753004984644, "learning_rate": 6.226133269198124e-06, "loss": 0.5406, "step": 17809 }, { "epoch": 0.6345191228601457, "grad_norm": 1.6341083508272225, "learning_rate": 6.225064684585094e-06, "loss": 0.5343, "step": 17810 }, { "epoch": 0.6345547499866399, "grad_norm": 1.6069765712407222, "learning_rate": 6.223996150237968e-06, "loss": 0.5064, "step": 17811 }, { "epoch": 0.6345903771131339, "grad_norm": 1.1339534279707493, "learning_rate": 6.222927666170968e-06, "loss": 0.4247, "step": 17812 }, { "epoch": 0.634626004239628, "grad_norm": 1.7112426784824966, "learning_rate": 6.22185923239833e-06, "loss": 0.5835, "step": 17813 }, { "epoch": 0.6346616313661222, "grad_norm": 1.125314528056885, "learning_rate": 6.220790848934274e-06, "loss": 0.4057, "step": 17814 }, { "epoch": 0.6346972584926163, "grad_norm": 1.160032042882306, "learning_rate": 6.219722515793034e-06, "loss": 0.4031, "step": 17815 }, { "epoch": 0.6347328856191103, "grad_norm": 1.910647417418918, "learning_rate": 6.218654232988823e-06, "loss": 0.5891, "step": 17816 }, { "epoch": 0.6347685127456045, "grad_norm": 2.2168654042257043, "learning_rate": 6.217586000535883e-06, "loss": 0.4384, "step": 17817 }, { "epoch": 0.6348041398720986, "grad_norm": 1.289770809590114, "learning_rate": 6.216517818448423e-06, "loss": 0.4632, "step": 17818 }, { "epoch": 0.6348397669985927, "grad_norm": 1.1855877820145406, "learning_rate": 6.215449686740675e-06, "loss": 0.4835, "step": 17819 }, { "epoch": 0.6348753941250869, "grad_norm": 2.097656105821489, "learning_rate": 6.214381605426858e-06, "loss": 0.615, "step": 17820 }, { "epoch": 0.6349110212515809, "grad_norm": 1.7351355579256453, "learning_rate": 6.213313574521196e-06, "loss": 0.5163, "step": 17821 }, { "epoch": 0.634946648378075, "grad_norm": 1.1629597349904515, "learning_rate": 6.2122455940379145e-06, "loss": 0.4183, "step": 17822 }, { "epoch": 0.6349822755045692, "grad_norm": 1.1046166533400976, "learning_rate": 6.211177663991226e-06, "loss": 0.4635, "step": 17823 }, { "epoch": 0.6350179026310633, "grad_norm": 1.5593478413355744, "learning_rate": 6.210109784395358e-06, "loss": 0.5146, "step": 17824 }, { "epoch": 0.6350535297575574, "grad_norm": 2.1602097580195765, "learning_rate": 6.209041955264524e-06, "loss": 0.425, "step": 17825 }, { "epoch": 0.6350891568840515, "grad_norm": 1.1343986198504297, "learning_rate": 6.207974176612952e-06, "loss": 0.3881, "step": 17826 }, { "epoch": 0.6351247840105456, "grad_norm": 1.4886419827219537, "learning_rate": 6.206906448454848e-06, "loss": 0.5127, "step": 17827 }, { "epoch": 0.6351604111370397, "grad_norm": 1.2795519659768895, "learning_rate": 6.20583877080444e-06, "loss": 0.5386, "step": 17828 }, { "epoch": 0.6351960382635339, "grad_norm": 1.8565832286774826, "learning_rate": 6.204771143675936e-06, "loss": 0.5585, "step": 17829 }, { "epoch": 0.635231665390028, "grad_norm": 1.1379757437991243, "learning_rate": 6.203703567083561e-06, "loss": 0.4688, "step": 17830 }, { "epoch": 0.635267292516522, "grad_norm": 1.1576012897323456, "learning_rate": 6.202636041041524e-06, "loss": 0.3954, "step": 17831 }, { "epoch": 0.6353029196430162, "grad_norm": 1.263239075432088, "learning_rate": 6.201568565564042e-06, "loss": 0.4917, "step": 17832 }, { "epoch": 0.6353385467695103, "grad_norm": 1.2538670581971392, "learning_rate": 6.200501140665334e-06, "loss": 0.5616, "step": 17833 }, { "epoch": 0.6353741738960044, "grad_norm": 1.4029006463112594, "learning_rate": 6.199433766359605e-06, "loss": 0.5355, "step": 17834 }, { "epoch": 0.6354098010224986, "grad_norm": 1.4538860957719746, "learning_rate": 6.1983664426610735e-06, "loss": 0.5112, "step": 17835 }, { "epoch": 0.6354454281489926, "grad_norm": 1.7974474309847053, "learning_rate": 6.197299169583948e-06, "loss": 0.4739, "step": 17836 }, { "epoch": 0.6354810552754867, "grad_norm": 1.775133448053126, "learning_rate": 6.1962319471424455e-06, "loss": 0.5382, "step": 17837 }, { "epoch": 0.6355166824019809, "grad_norm": 1.6011940200406183, "learning_rate": 6.19516477535077e-06, "loss": 0.5315, "step": 17838 }, { "epoch": 0.635552309528475, "grad_norm": 1.697515721697654, "learning_rate": 6.194097654223137e-06, "loss": 0.5363, "step": 17839 }, { "epoch": 0.635587936654969, "grad_norm": 1.918637442700067, "learning_rate": 6.193030583773752e-06, "loss": 0.4954, "step": 17840 }, { "epoch": 0.6356235637814632, "grad_norm": 1.1708244031325874, "learning_rate": 6.191963564016829e-06, "loss": 0.4206, "step": 17841 }, { "epoch": 0.6356591909079573, "grad_norm": 1.0892083829248178, "learning_rate": 6.1908965949665675e-06, "loss": 0.4011, "step": 17842 }, { "epoch": 0.6356948180344514, "grad_norm": 1.411172657203749, "learning_rate": 6.189829676637182e-06, "loss": 0.5286, "step": 17843 }, { "epoch": 0.6357304451609456, "grad_norm": 1.338994117540324, "learning_rate": 6.188762809042882e-06, "loss": 0.4822, "step": 17844 }, { "epoch": 0.6357660722874396, "grad_norm": 1.4487050744656664, "learning_rate": 6.187695992197864e-06, "loss": 0.4593, "step": 17845 }, { "epoch": 0.6358016994139337, "grad_norm": 1.0587018200866183, "learning_rate": 6.186629226116342e-06, "loss": 0.419, "step": 17846 }, { "epoch": 0.6358373265404279, "grad_norm": 1.120506243653291, "learning_rate": 6.185562510812514e-06, "loss": 0.3891, "step": 17847 }, { "epoch": 0.635872953666922, "grad_norm": 1.275807377637196, "learning_rate": 6.184495846300591e-06, "loss": 0.4405, "step": 17848 }, { "epoch": 0.6359085807934161, "grad_norm": 1.2767255312536991, "learning_rate": 6.18342923259477e-06, "loss": 0.4622, "step": 17849 }, { "epoch": 0.6359442079199102, "grad_norm": 1.6456411042533854, "learning_rate": 6.182362669709257e-06, "loss": 0.5255, "step": 17850 }, { "epoch": 0.6359798350464043, "grad_norm": 1.069413155078838, "learning_rate": 6.181296157658252e-06, "loss": 0.4042, "step": 17851 }, { "epoch": 0.6360154621728984, "grad_norm": 1.5083071289173173, "learning_rate": 6.180229696455963e-06, "loss": 0.5691, "step": 17852 }, { "epoch": 0.6360510892993926, "grad_norm": 1.698018415401207, "learning_rate": 6.179163286116581e-06, "loss": 0.5992, "step": 17853 }, { "epoch": 0.6360867164258867, "grad_norm": 1.6796272572154693, "learning_rate": 6.178096926654311e-06, "loss": 0.5196, "step": 17854 }, { "epoch": 0.6361223435523807, "grad_norm": 1.1400373439052596, "learning_rate": 6.177030618083351e-06, "loss": 0.4249, "step": 17855 }, { "epoch": 0.6361579706788749, "grad_norm": 1.125794616179587, "learning_rate": 6.175964360417899e-06, "loss": 0.4187, "step": 17856 }, { "epoch": 0.636193597805369, "grad_norm": 1.0885912318304587, "learning_rate": 6.174898153672157e-06, "loss": 0.4038, "step": 17857 }, { "epoch": 0.6362292249318631, "grad_norm": 1.131577613686729, "learning_rate": 6.173831997860319e-06, "loss": 0.4268, "step": 17858 }, { "epoch": 0.6362648520583573, "grad_norm": 1.9249851884660787, "learning_rate": 6.172765892996585e-06, "loss": 0.4803, "step": 17859 }, { "epoch": 0.6363004791848513, "grad_norm": 1.2846796572961832, "learning_rate": 6.171699839095144e-06, "loss": 0.4184, "step": 17860 }, { "epoch": 0.6363361063113454, "grad_norm": 1.026588621307734, "learning_rate": 6.170633836170199e-06, "loss": 0.4107, "step": 17861 }, { "epoch": 0.6363717334378396, "grad_norm": 1.7069908341574136, "learning_rate": 6.169567884235938e-06, "loss": 0.4737, "step": 17862 }, { "epoch": 0.6364073605643337, "grad_norm": 1.6804965464356532, "learning_rate": 6.168501983306562e-06, "loss": 0.5687, "step": 17863 }, { "epoch": 0.6364429876908277, "grad_norm": 1.0650527239181882, "learning_rate": 6.1674361333962565e-06, "loss": 0.418, "step": 17864 }, { "epoch": 0.6364786148173219, "grad_norm": 1.4292809955260015, "learning_rate": 6.166370334519219e-06, "loss": 0.5749, "step": 17865 }, { "epoch": 0.636514241943816, "grad_norm": 1.3811515589923409, "learning_rate": 6.165304586689638e-06, "loss": 0.5204, "step": 17866 }, { "epoch": 0.6365498690703101, "grad_norm": 1.076392347534082, "learning_rate": 6.1642388899217075e-06, "loss": 0.4278, "step": 17867 }, { "epoch": 0.6365854961968043, "grad_norm": 1.6527152981789943, "learning_rate": 6.163173244229618e-06, "loss": 0.5068, "step": 17868 }, { "epoch": 0.6366211233232983, "grad_norm": 1.230152199279453, "learning_rate": 6.162107649627558e-06, "loss": 0.4499, "step": 17869 }, { "epoch": 0.6366567504497924, "grad_norm": 1.2987551274153204, "learning_rate": 6.16104210612972e-06, "loss": 0.5295, "step": 17870 }, { "epoch": 0.6366923775762866, "grad_norm": 1.5543113675005011, "learning_rate": 6.159976613750287e-06, "loss": 0.5193, "step": 17871 }, { "epoch": 0.6367280047027807, "grad_norm": 1.866648911360492, "learning_rate": 6.15891117250345e-06, "loss": 0.5722, "step": 17872 }, { "epoch": 0.6367636318292748, "grad_norm": 1.1236645674407795, "learning_rate": 6.1578457824033935e-06, "loss": 0.4561, "step": 17873 }, { "epoch": 0.6367992589557689, "grad_norm": 1.3749907167969784, "learning_rate": 6.156780443464311e-06, "loss": 0.4352, "step": 17874 }, { "epoch": 0.636834886082263, "grad_norm": 1.0616073962825208, "learning_rate": 6.155715155700378e-06, "loss": 0.4069, "step": 17875 }, { "epoch": 0.6368705132087571, "grad_norm": 1.3342756441784525, "learning_rate": 6.154649919125788e-06, "loss": 0.4679, "step": 17876 }, { "epoch": 0.6369061403352513, "grad_norm": 1.0533670093261664, "learning_rate": 6.1535847337547185e-06, "loss": 0.4138, "step": 17877 }, { "epoch": 0.6369417674617454, "grad_norm": 1.5320263458852494, "learning_rate": 6.152519599601359e-06, "loss": 0.5537, "step": 17878 }, { "epoch": 0.6369773945882394, "grad_norm": 1.1038512157046816, "learning_rate": 6.151454516679891e-06, "loss": 0.4437, "step": 17879 }, { "epoch": 0.6370130217147336, "grad_norm": 1.7746061415587457, "learning_rate": 6.150389485004493e-06, "loss": 0.4488, "step": 17880 }, { "epoch": 0.6370486488412277, "grad_norm": 1.1150070236003904, "learning_rate": 6.149324504589354e-06, "loss": 0.4223, "step": 17881 }, { "epoch": 0.6370842759677218, "grad_norm": 1.8689331776002391, "learning_rate": 6.148259575448648e-06, "loss": 0.6613, "step": 17882 }, { "epoch": 0.637119903094216, "grad_norm": 1.3565733218337155, "learning_rate": 6.147194697596559e-06, "loss": 0.4194, "step": 17883 }, { "epoch": 0.63715553022071, "grad_norm": 1.7751275878409252, "learning_rate": 6.1461298710472635e-06, "loss": 0.5497, "step": 17884 }, { "epoch": 0.6371911573472041, "grad_norm": 1.4906950937717134, "learning_rate": 6.145065095814947e-06, "loss": 0.4508, "step": 17885 }, { "epoch": 0.6372267844736983, "grad_norm": 1.2009023736429898, "learning_rate": 6.144000371913779e-06, "loss": 0.4321, "step": 17886 }, { "epoch": 0.6372624116001924, "grad_norm": 1.5798829910667063, "learning_rate": 6.1429356993579435e-06, "loss": 0.5484, "step": 17887 }, { "epoch": 0.6372980387266864, "grad_norm": 1.0772470252004427, "learning_rate": 6.141871078161612e-06, "loss": 0.4245, "step": 17888 }, { "epoch": 0.6373336658531806, "grad_norm": 2.0195431547117844, "learning_rate": 6.140806508338966e-06, "loss": 0.5887, "step": 17889 }, { "epoch": 0.6373692929796747, "grad_norm": 1.0933268010124213, "learning_rate": 6.139741989904181e-06, "loss": 0.4009, "step": 17890 }, { "epoch": 0.6374049201061688, "grad_norm": 1.4825901301208104, "learning_rate": 6.138677522871424e-06, "loss": 0.5979, "step": 17891 }, { "epoch": 0.637440547232663, "grad_norm": 1.241520621320623, "learning_rate": 6.13761310725488e-06, "loss": 0.4484, "step": 17892 }, { "epoch": 0.637476174359157, "grad_norm": 1.0614095237986978, "learning_rate": 6.136548743068713e-06, "loss": 0.3757, "step": 17893 }, { "epoch": 0.6375118014856511, "grad_norm": 1.5632852310942138, "learning_rate": 6.135484430327103e-06, "loss": 0.5749, "step": 17894 }, { "epoch": 0.6375474286121453, "grad_norm": 1.3094483540171997, "learning_rate": 6.1344201690442175e-06, "loss": 0.5353, "step": 17895 }, { "epoch": 0.6375830557386394, "grad_norm": 1.157635813466602, "learning_rate": 6.133355959234232e-06, "loss": 0.4191, "step": 17896 }, { "epoch": 0.6376186828651336, "grad_norm": 1.7346511533336684, "learning_rate": 6.1322918009113105e-06, "loss": 0.5416, "step": 17897 }, { "epoch": 0.6376543099916276, "grad_norm": 1.48382175629203, "learning_rate": 6.13122769408963e-06, "loss": 0.6195, "step": 17898 }, { "epoch": 0.6376899371181217, "grad_norm": 1.077966443352899, "learning_rate": 6.1301636387833545e-06, "loss": 0.4275, "step": 17899 }, { "epoch": 0.6377255642446159, "grad_norm": 1.1746138854980501, "learning_rate": 6.129099635006659e-06, "loss": 0.4001, "step": 17900 }, { "epoch": 0.63776119137111, "grad_norm": 1.140281275403277, "learning_rate": 6.128035682773703e-06, "loss": 0.4101, "step": 17901 }, { "epoch": 0.6377968184976041, "grad_norm": 2.382227258805452, "learning_rate": 6.1269717820986585e-06, "loss": 0.5607, "step": 17902 }, { "epoch": 0.6378324456240982, "grad_norm": 2.7743848454870754, "learning_rate": 6.125907932995694e-06, "loss": 0.5378, "step": 17903 }, { "epoch": 0.6378680727505923, "grad_norm": 1.2498537204980709, "learning_rate": 6.124844135478971e-06, "loss": 0.4288, "step": 17904 }, { "epoch": 0.6379036998770864, "grad_norm": 1.3540395417744255, "learning_rate": 6.123780389562659e-06, "loss": 0.4172, "step": 17905 }, { "epoch": 0.6379393270035806, "grad_norm": 1.5146649422226992, "learning_rate": 6.1227166952609174e-06, "loss": 0.5235, "step": 17906 }, { "epoch": 0.6379749541300747, "grad_norm": 1.0566640691050855, "learning_rate": 6.1216530525879185e-06, "loss": 0.4091, "step": 17907 }, { "epoch": 0.6380105812565687, "grad_norm": 1.3185672511237154, "learning_rate": 6.120589461557815e-06, "loss": 0.442, "step": 17908 }, { "epoch": 0.6380462083830629, "grad_norm": 1.521453221345835, "learning_rate": 6.119525922184777e-06, "loss": 0.5462, "step": 17909 }, { "epoch": 0.638081835509557, "grad_norm": 1.1695482106535982, "learning_rate": 6.118462434482961e-06, "loss": 0.4812, "step": 17910 }, { "epoch": 0.6381174626360511, "grad_norm": 1.2932763441503288, "learning_rate": 6.1173989984665365e-06, "loss": 0.4618, "step": 17911 }, { "epoch": 0.6381530897625453, "grad_norm": 1.1158137292509154, "learning_rate": 6.116335614149652e-06, "loss": 0.3854, "step": 17912 }, { "epoch": 0.6381887168890393, "grad_norm": 1.182648273650162, "learning_rate": 6.115272281546474e-06, "loss": 0.438, "step": 17913 }, { "epoch": 0.6382243440155334, "grad_norm": 1.2143762817886015, "learning_rate": 6.114209000671163e-06, "loss": 0.4484, "step": 17914 }, { "epoch": 0.6382599711420276, "grad_norm": 1.2550585967142636, "learning_rate": 6.113145771537874e-06, "loss": 0.5182, "step": 17915 }, { "epoch": 0.6382955982685217, "grad_norm": 2.0056853847335248, "learning_rate": 6.112082594160769e-06, "loss": 0.512, "step": 17916 }, { "epoch": 0.6383312253950157, "grad_norm": 1.785665974773071, "learning_rate": 6.111019468553995e-06, "loss": 0.5721, "step": 17917 }, { "epoch": 0.6383668525215099, "grad_norm": 1.7683101793354306, "learning_rate": 6.109956394731722e-06, "loss": 0.5375, "step": 17918 }, { "epoch": 0.638402479648004, "grad_norm": 1.1163597101194405, "learning_rate": 6.1088933727080935e-06, "loss": 0.4243, "step": 17919 }, { "epoch": 0.6384381067744981, "grad_norm": 1.3380433561840526, "learning_rate": 6.107830402497272e-06, "loss": 0.5275, "step": 17920 }, { "epoch": 0.6384737339009923, "grad_norm": 1.1118020121504766, "learning_rate": 6.1067674841134075e-06, "loss": 0.4221, "step": 17921 }, { "epoch": 0.6385093610274863, "grad_norm": 1.9967215989290448, "learning_rate": 6.1057046175706605e-06, "loss": 0.5289, "step": 17922 }, { "epoch": 0.6385449881539804, "grad_norm": 1.099539874986276, "learning_rate": 6.104641802883172e-06, "loss": 0.4436, "step": 17923 }, { "epoch": 0.6385806152804746, "grad_norm": 1.6407837579141868, "learning_rate": 6.103579040065102e-06, "loss": 0.516, "step": 17924 }, { "epoch": 0.6386162424069687, "grad_norm": 1.7600727383111285, "learning_rate": 6.102516329130601e-06, "loss": 0.4368, "step": 17925 }, { "epoch": 0.6386518695334628, "grad_norm": 1.2443767490111983, "learning_rate": 6.10145367009382e-06, "loss": 0.4082, "step": 17926 }, { "epoch": 0.6386874966599569, "grad_norm": 1.5372853412808642, "learning_rate": 6.10039106296891e-06, "loss": 0.5643, "step": 17927 }, { "epoch": 0.638723123786451, "grad_norm": 1.2730793197975554, "learning_rate": 6.099328507770015e-06, "loss": 0.4479, "step": 17928 }, { "epoch": 0.6387587509129451, "grad_norm": 1.3920193275374209, "learning_rate": 6.098266004511291e-06, "loss": 0.5133, "step": 17929 }, { "epoch": 0.6387943780394393, "grad_norm": 1.0977735263253936, "learning_rate": 6.097203553206879e-06, "loss": 0.4204, "step": 17930 }, { "epoch": 0.6388300051659334, "grad_norm": 1.202445895253403, "learning_rate": 6.09614115387093e-06, "loss": 0.387, "step": 17931 }, { "epoch": 0.6388656322924274, "grad_norm": 1.2337234966633135, "learning_rate": 6.09507880651759e-06, "loss": 0.4692, "step": 17932 }, { "epoch": 0.6389012594189216, "grad_norm": 1.2250670795401881, "learning_rate": 6.09401651116101e-06, "loss": 0.4881, "step": 17933 }, { "epoch": 0.6389368865454157, "grad_norm": 1.5709514137776752, "learning_rate": 6.0929542678153244e-06, "loss": 0.5326, "step": 17934 }, { "epoch": 0.6389725136719098, "grad_norm": 1.1313022235381212, "learning_rate": 6.0918920764946856e-06, "loss": 0.4224, "step": 17935 }, { "epoch": 0.639008140798404, "grad_norm": 1.0737470008189511, "learning_rate": 6.090829937213234e-06, "loss": 0.4217, "step": 17936 }, { "epoch": 0.639043767924898, "grad_norm": 1.0449596957762128, "learning_rate": 6.089767849985114e-06, "loss": 0.4, "step": 17937 }, { "epoch": 0.6390793950513921, "grad_norm": 1.0127600130939787, "learning_rate": 6.088705814824473e-06, "loss": 0.3919, "step": 17938 }, { "epoch": 0.6391150221778863, "grad_norm": 1.7542588198217335, "learning_rate": 6.0876438317454434e-06, "loss": 0.5291, "step": 17939 }, { "epoch": 0.6391506493043804, "grad_norm": 2.5077957478593422, "learning_rate": 6.0865819007621716e-06, "loss": 0.5863, "step": 17940 }, { "epoch": 0.6391862764308744, "grad_norm": 1.146905142921153, "learning_rate": 6.085520021888797e-06, "loss": 0.4331, "step": 17941 }, { "epoch": 0.6392219035573686, "grad_norm": 1.8748311464260172, "learning_rate": 6.084458195139462e-06, "loss": 0.4513, "step": 17942 }, { "epoch": 0.6392575306838627, "grad_norm": 1.2439191569805297, "learning_rate": 6.083396420528298e-06, "loss": 0.4573, "step": 17943 }, { "epoch": 0.6392931578103568, "grad_norm": 1.7451967497326744, "learning_rate": 6.082334698069454e-06, "loss": 0.4797, "step": 17944 }, { "epoch": 0.639328784936851, "grad_norm": 1.5589111335957924, "learning_rate": 6.081273027777057e-06, "loss": 0.5217, "step": 17945 }, { "epoch": 0.639364412063345, "grad_norm": 2.7573425519327173, "learning_rate": 6.080211409665251e-06, "loss": 0.512, "step": 17946 }, { "epoch": 0.6394000391898391, "grad_norm": 1.3548242617040784, "learning_rate": 6.079149843748168e-06, "loss": 0.434, "step": 17947 }, { "epoch": 0.6394356663163333, "grad_norm": 1.6379134061336746, "learning_rate": 6.078088330039945e-06, "loss": 0.4854, "step": 17948 }, { "epoch": 0.6394712934428274, "grad_norm": 1.4085379175185355, "learning_rate": 6.077026868554721e-06, "loss": 0.5037, "step": 17949 }, { "epoch": 0.6395069205693215, "grad_norm": 1.353341987913374, "learning_rate": 6.0759654593066235e-06, "loss": 0.5894, "step": 17950 }, { "epoch": 0.6395425476958156, "grad_norm": 1.2361797529823433, "learning_rate": 6.074904102309789e-06, "loss": 0.4548, "step": 17951 }, { "epoch": 0.6395781748223097, "grad_norm": 1.470220426244115, "learning_rate": 6.0738427975783485e-06, "loss": 0.5758, "step": 17952 }, { "epoch": 0.6396138019488038, "grad_norm": 1.7658407478088707, "learning_rate": 6.072781545126441e-06, "loss": 0.4898, "step": 17953 }, { "epoch": 0.639649429075298, "grad_norm": 2.640056609606709, "learning_rate": 6.071720344968186e-06, "loss": 0.4978, "step": 17954 }, { "epoch": 0.6396850562017921, "grad_norm": 1.663495216176424, "learning_rate": 6.0706591971177235e-06, "loss": 0.4874, "step": 17955 }, { "epoch": 0.6397206833282861, "grad_norm": 2.4347471871531616, "learning_rate": 6.069598101589177e-06, "loss": 0.4326, "step": 17956 }, { "epoch": 0.6397563104547803, "grad_norm": 1.0318998313012397, "learning_rate": 6.068537058396682e-06, "loss": 0.3975, "step": 17957 }, { "epoch": 0.6397919375812744, "grad_norm": 1.4537994169333703, "learning_rate": 6.067476067554361e-06, "loss": 0.5158, "step": 17958 }, { "epoch": 0.6398275647077685, "grad_norm": 1.9646578239853156, "learning_rate": 6.066415129076346e-06, "loss": 0.5091, "step": 17959 }, { "epoch": 0.6398631918342627, "grad_norm": 1.1262365468863305, "learning_rate": 6.0653542429767664e-06, "loss": 0.4139, "step": 17960 }, { "epoch": 0.6398988189607567, "grad_norm": 1.6591748725678181, "learning_rate": 6.064293409269742e-06, "loss": 0.5488, "step": 17961 }, { "epoch": 0.6399344460872508, "grad_norm": 1.9178308755161062, "learning_rate": 6.063232627969402e-06, "loss": 0.4928, "step": 17962 }, { "epoch": 0.639970073213745, "grad_norm": 1.9785772854211168, "learning_rate": 6.06217189908987e-06, "loss": 0.5558, "step": 17963 }, { "epoch": 0.6400057003402391, "grad_norm": 1.2061816026580867, "learning_rate": 6.061111222645276e-06, "loss": 0.4394, "step": 17964 }, { "epoch": 0.6400413274667331, "grad_norm": 1.8218904777635307, "learning_rate": 6.060050598649735e-06, "loss": 0.5005, "step": 17965 }, { "epoch": 0.6400769545932273, "grad_norm": 1.4228746676484891, "learning_rate": 6.058990027117376e-06, "loss": 0.5552, "step": 17966 }, { "epoch": 0.6401125817197214, "grad_norm": 1.5945395142420886, "learning_rate": 6.057929508062318e-06, "loss": 0.5567, "step": 17967 }, { "epoch": 0.6401482088462155, "grad_norm": 1.4326361014770423, "learning_rate": 6.0568690414986875e-06, "loss": 0.4785, "step": 17968 }, { "epoch": 0.6401838359727097, "grad_norm": 1.8390150046827385, "learning_rate": 6.055808627440596e-06, "loss": 0.5156, "step": 17969 }, { "epoch": 0.6402194630992037, "grad_norm": 1.2659109967567448, "learning_rate": 6.0547482659021704e-06, "loss": 0.3895, "step": 17970 }, { "epoch": 0.6402550902256978, "grad_norm": 1.5164665151388939, "learning_rate": 6.053687956897534e-06, "loss": 0.6291, "step": 17971 }, { "epoch": 0.640290717352192, "grad_norm": 1.6274700364102903, "learning_rate": 6.052627700440796e-06, "loss": 0.4941, "step": 17972 }, { "epoch": 0.6403263444786861, "grad_norm": 2.1328505359533394, "learning_rate": 6.051567496546083e-06, "loss": 0.5161, "step": 17973 }, { "epoch": 0.6403619716051802, "grad_norm": 1.268979148995726, "learning_rate": 6.050507345227505e-06, "loss": 0.4121, "step": 17974 }, { "epoch": 0.6403975987316743, "grad_norm": 1.191030660220948, "learning_rate": 6.049447246499187e-06, "loss": 0.4568, "step": 17975 }, { "epoch": 0.6404332258581684, "grad_norm": 1.072294100926575, "learning_rate": 6.048387200375236e-06, "loss": 0.4236, "step": 17976 }, { "epoch": 0.6404688529846625, "grad_norm": 1.2442854004383928, "learning_rate": 6.047327206869773e-06, "loss": 0.4526, "step": 17977 }, { "epoch": 0.6405044801111567, "grad_norm": 2.1011172127697417, "learning_rate": 6.0462672659969105e-06, "loss": 0.5861, "step": 17978 }, { "epoch": 0.6405401072376508, "grad_norm": 1.3607455836902331, "learning_rate": 6.045207377770765e-06, "loss": 0.4177, "step": 17979 }, { "epoch": 0.6405757343641448, "grad_norm": 1.4042566200027267, "learning_rate": 6.044147542205443e-06, "loss": 0.4454, "step": 17980 }, { "epoch": 0.640611361490639, "grad_norm": 1.5015392643556855, "learning_rate": 6.043087759315066e-06, "loss": 0.5125, "step": 17981 }, { "epoch": 0.6406469886171331, "grad_norm": 1.4984703870804748, "learning_rate": 6.042028029113738e-06, "loss": 0.4636, "step": 17982 }, { "epoch": 0.6406826157436272, "grad_norm": 1.211656688647576, "learning_rate": 6.040968351615572e-06, "loss": 0.4353, "step": 17983 }, { "epoch": 0.6407182428701214, "grad_norm": 1.8160678164901967, "learning_rate": 6.039908726834684e-06, "loss": 0.5015, "step": 17984 }, { "epoch": 0.6407538699966154, "grad_norm": 1.3493784608719503, "learning_rate": 6.038849154785176e-06, "loss": 0.4045, "step": 17985 }, { "epoch": 0.6407894971231095, "grad_norm": 1.0580212286975674, "learning_rate": 6.037789635481164e-06, "loss": 0.4399, "step": 17986 }, { "epoch": 0.6408251242496037, "grad_norm": 1.4363473136677642, "learning_rate": 6.0367301689367485e-06, "loss": 0.3982, "step": 17987 }, { "epoch": 0.6408607513760978, "grad_norm": 1.1560786503383536, "learning_rate": 6.0356707551660434e-06, "loss": 0.4338, "step": 17988 }, { "epoch": 0.6408963785025918, "grad_norm": 1.1368807760011719, "learning_rate": 6.034611394183151e-06, "loss": 0.4147, "step": 17989 }, { "epoch": 0.640932005629086, "grad_norm": 1.202919517980995, "learning_rate": 6.033552086002184e-06, "loss": 0.452, "step": 17990 }, { "epoch": 0.6409676327555801, "grad_norm": 1.2136191065718553, "learning_rate": 6.032492830637238e-06, "loss": 0.4249, "step": 17991 }, { "epoch": 0.6410032598820742, "grad_norm": 1.2191758249385096, "learning_rate": 6.031433628102426e-06, "loss": 0.4067, "step": 17992 }, { "epoch": 0.6410388870085684, "grad_norm": 2.1652054308001434, "learning_rate": 6.030374478411847e-06, "loss": 0.5935, "step": 17993 }, { "epoch": 0.6410745141350624, "grad_norm": 1.332732838336445, "learning_rate": 6.029315381579606e-06, "loss": 0.4935, "step": 17994 }, { "epoch": 0.6411101412615565, "grad_norm": 1.1873541429032706, "learning_rate": 6.028256337619807e-06, "loss": 0.4482, "step": 17995 }, { "epoch": 0.6411457683880507, "grad_norm": 1.211134946194001, "learning_rate": 6.0271973465465505e-06, "loss": 0.4234, "step": 17996 }, { "epoch": 0.6411813955145448, "grad_norm": 1.3849622142756206, "learning_rate": 6.026138408373942e-06, "loss": 0.5147, "step": 17997 }, { "epoch": 0.6412170226410389, "grad_norm": 1.1949041340968625, "learning_rate": 6.0250795231160745e-06, "loss": 0.4342, "step": 17998 }, { "epoch": 0.641252649767533, "grad_norm": 1.097791575618586, "learning_rate": 6.0240206907870526e-06, "loss": 0.4171, "step": 17999 }, { "epoch": 0.6412882768940271, "grad_norm": 1.7251456921407846, "learning_rate": 6.022961911400973e-06, "loss": 0.543, "step": 18000 }, { "epoch": 0.6413239040205212, "grad_norm": 1.6344083488200536, "learning_rate": 6.021903184971938e-06, "loss": 0.5398, "step": 18001 }, { "epoch": 0.6413595311470154, "grad_norm": 1.2586580184730374, "learning_rate": 6.02084451151404e-06, "loss": 0.397, "step": 18002 }, { "epoch": 0.6413951582735095, "grad_norm": 1.1281262703581547, "learning_rate": 6.019785891041381e-06, "loss": 0.4385, "step": 18003 }, { "epoch": 0.6414307854000035, "grad_norm": 1.3459782854317677, "learning_rate": 6.018727323568052e-06, "loss": 0.4731, "step": 18004 }, { "epoch": 0.6414664125264977, "grad_norm": 1.3713042585709403, "learning_rate": 6.017668809108151e-06, "loss": 0.5789, "step": 18005 }, { "epoch": 0.6415020396529918, "grad_norm": 1.299196750968949, "learning_rate": 6.016610347675775e-06, "loss": 0.4183, "step": 18006 }, { "epoch": 0.6415376667794859, "grad_norm": 1.4884670994114875, "learning_rate": 6.015551939285016e-06, "loss": 0.4508, "step": 18007 }, { "epoch": 0.6415732939059801, "grad_norm": 1.169234495817304, "learning_rate": 6.014493583949971e-06, "loss": 0.4067, "step": 18008 }, { "epoch": 0.6416089210324741, "grad_norm": 1.3513333015100353, "learning_rate": 6.013435281684726e-06, "loss": 0.4236, "step": 18009 }, { "epoch": 0.6416445481589682, "grad_norm": 1.195695180923578, "learning_rate": 6.012377032503378e-06, "loss": 0.4279, "step": 18010 }, { "epoch": 0.6416801752854624, "grad_norm": 1.541530980012524, "learning_rate": 6.0113188364200145e-06, "loss": 0.4642, "step": 18011 }, { "epoch": 0.6417158024119565, "grad_norm": 1.6494305679689552, "learning_rate": 6.010260693448735e-06, "loss": 0.5365, "step": 18012 }, { "epoch": 0.6417514295384505, "grad_norm": 1.3720931136792027, "learning_rate": 6.009202603603616e-06, "loss": 0.4765, "step": 18013 }, { "epoch": 0.6417870566649447, "grad_norm": 1.67144166331266, "learning_rate": 6.008144566898758e-06, "loss": 0.5699, "step": 18014 }, { "epoch": 0.6418226837914388, "grad_norm": 1.212248013427672, "learning_rate": 6.007086583348243e-06, "loss": 0.4434, "step": 18015 }, { "epoch": 0.6418583109179329, "grad_norm": 1.5082608042115966, "learning_rate": 6.0060286529661604e-06, "loss": 0.5228, "step": 18016 }, { "epoch": 0.6418939380444271, "grad_norm": 1.3965658861975245, "learning_rate": 6.004970775766601e-06, "loss": 0.4544, "step": 18017 }, { "epoch": 0.6419295651709211, "grad_norm": 1.2099647944332181, "learning_rate": 6.003912951763644e-06, "loss": 0.4418, "step": 18018 }, { "epoch": 0.6419651922974152, "grad_norm": 1.172104407334176, "learning_rate": 6.002855180971384e-06, "loss": 0.434, "step": 18019 }, { "epoch": 0.6420008194239094, "grad_norm": 1.068170410866359, "learning_rate": 6.001797463403898e-06, "loss": 0.3765, "step": 18020 }, { "epoch": 0.6420364465504035, "grad_norm": 1.1719876042883082, "learning_rate": 6.000739799075275e-06, "loss": 0.408, "step": 18021 }, { "epoch": 0.6420720736768976, "grad_norm": 1.1878301182050848, "learning_rate": 5.999682187999594e-06, "loss": 0.4071, "step": 18022 }, { "epoch": 0.6421077008033917, "grad_norm": 1.8331012989378437, "learning_rate": 5.998624630190946e-06, "loss": 0.5443, "step": 18023 }, { "epoch": 0.6421433279298858, "grad_norm": 1.4628968426124407, "learning_rate": 5.997567125663401e-06, "loss": 0.4853, "step": 18024 }, { "epoch": 0.6421789550563799, "grad_norm": 1.4782541053774139, "learning_rate": 5.996509674431053e-06, "loss": 0.4723, "step": 18025 }, { "epoch": 0.6422145821828741, "grad_norm": 1.4115760178857115, "learning_rate": 5.995452276507973e-06, "loss": 0.5056, "step": 18026 }, { "epoch": 0.6422502093093682, "grad_norm": 1.2440990316184912, "learning_rate": 5.99439493190825e-06, "loss": 0.4345, "step": 18027 }, { "epoch": 0.6422858364358622, "grad_norm": 1.5131187157428792, "learning_rate": 5.993337640645953e-06, "loss": 0.5524, "step": 18028 }, { "epoch": 0.6423214635623564, "grad_norm": 1.1385509936273879, "learning_rate": 5.992280402735167e-06, "loss": 0.4282, "step": 18029 }, { "epoch": 0.6423570906888505, "grad_norm": 1.119896133876606, "learning_rate": 5.991223218189969e-06, "loss": 0.4617, "step": 18030 }, { "epoch": 0.6423927178153446, "grad_norm": 1.4122765144998821, "learning_rate": 5.9901660870244345e-06, "loss": 0.4612, "step": 18031 }, { "epoch": 0.6424283449418388, "grad_norm": 1.232270806636033, "learning_rate": 5.989109009252644e-06, "loss": 0.4574, "step": 18032 }, { "epoch": 0.6424639720683328, "grad_norm": 1.1829342663431464, "learning_rate": 5.988051984888668e-06, "loss": 0.4375, "step": 18033 }, { "epoch": 0.6424995991948269, "grad_norm": 1.2265155315481882, "learning_rate": 5.986995013946588e-06, "loss": 0.4427, "step": 18034 }, { "epoch": 0.6425352263213211, "grad_norm": 1.1674941921680957, "learning_rate": 5.985938096440471e-06, "loss": 0.3975, "step": 18035 }, { "epoch": 0.6425708534478152, "grad_norm": 1.2841090618599433, "learning_rate": 5.984881232384394e-06, "loss": 0.4486, "step": 18036 }, { "epoch": 0.6426064805743092, "grad_norm": 1.2813093410779437, "learning_rate": 5.98382442179243e-06, "loss": 0.4259, "step": 18037 }, { "epoch": 0.6426421077008034, "grad_norm": 1.1451887614859249, "learning_rate": 5.982767664678654e-06, "loss": 0.4374, "step": 18038 }, { "epoch": 0.6426777348272975, "grad_norm": 2.2477420384040347, "learning_rate": 5.98171096105713e-06, "loss": 0.45, "step": 18039 }, { "epoch": 0.6427133619537916, "grad_norm": 1.4139355868812726, "learning_rate": 5.9806543109419335e-06, "loss": 0.5103, "step": 18040 }, { "epoch": 0.6427489890802858, "grad_norm": 1.510357596818781, "learning_rate": 5.979597714347136e-06, "loss": 0.5071, "step": 18041 }, { "epoch": 0.6427846162067798, "grad_norm": 1.1755831587781096, "learning_rate": 5.9785411712868045e-06, "loss": 0.4329, "step": 18042 }, { "epoch": 0.6428202433332739, "grad_norm": 1.701649261773657, "learning_rate": 5.977484681775011e-06, "loss": 0.5384, "step": 18043 }, { "epoch": 0.6428558704597681, "grad_norm": 1.436600016406463, "learning_rate": 5.976428245825814e-06, "loss": 0.5218, "step": 18044 }, { "epoch": 0.6428914975862622, "grad_norm": 1.4005140185392178, "learning_rate": 5.9753718634532944e-06, "loss": 0.5495, "step": 18045 }, { "epoch": 0.6429271247127563, "grad_norm": 1.5293849180301409, "learning_rate": 5.974315534671507e-06, "loss": 0.4421, "step": 18046 }, { "epoch": 0.6429627518392504, "grad_norm": 1.1032361870891632, "learning_rate": 5.973259259494524e-06, "loss": 0.4235, "step": 18047 }, { "epoch": 0.6429983789657445, "grad_norm": 1.6553495153871358, "learning_rate": 5.972203037936406e-06, "loss": 0.6046, "step": 18048 }, { "epoch": 0.6430340060922386, "grad_norm": 1.1468539974366592, "learning_rate": 5.971146870011224e-06, "loss": 0.4466, "step": 18049 }, { "epoch": 0.6430696332187328, "grad_norm": 1.6430762106031758, "learning_rate": 5.970090755733034e-06, "loss": 0.4788, "step": 18050 }, { "epoch": 0.6431052603452269, "grad_norm": 1.1120753607008023, "learning_rate": 5.9690346951159005e-06, "loss": 0.4375, "step": 18051 }, { "epoch": 0.6431408874717209, "grad_norm": 1.2406154128818883, "learning_rate": 5.9679786881738925e-06, "loss": 0.4178, "step": 18052 }, { "epoch": 0.6431765145982151, "grad_norm": 1.2359375430781698, "learning_rate": 5.966922734921062e-06, "loss": 0.391, "step": 18053 }, { "epoch": 0.6432121417247092, "grad_norm": 1.5405031477852837, "learning_rate": 5.965866835371479e-06, "loss": 0.6429, "step": 18054 }, { "epoch": 0.6432477688512033, "grad_norm": 1.3139482190847918, "learning_rate": 5.964810989539196e-06, "loss": 0.4353, "step": 18055 }, { "epoch": 0.6432833959776975, "grad_norm": 1.837391545659468, "learning_rate": 5.963755197438275e-06, "loss": 0.5244, "step": 18056 }, { "epoch": 0.6433190231041915, "grad_norm": 1.490040245261789, "learning_rate": 5.962699459082775e-06, "loss": 0.5746, "step": 18057 }, { "epoch": 0.6433546502306856, "grad_norm": 1.1912932037687445, "learning_rate": 5.961643774486754e-06, "loss": 0.4127, "step": 18058 }, { "epoch": 0.6433902773571798, "grad_norm": 1.1671939533062423, "learning_rate": 5.960588143664266e-06, "loss": 0.4238, "step": 18059 }, { "epoch": 0.6434259044836739, "grad_norm": 1.1680536661386807, "learning_rate": 5.959532566629376e-06, "loss": 0.4334, "step": 18060 }, { "epoch": 0.643461531610168, "grad_norm": 1.08354387968244, "learning_rate": 5.958477043396129e-06, "loss": 0.4189, "step": 18061 }, { "epoch": 0.6434971587366621, "grad_norm": 2.1982016288859922, "learning_rate": 5.9574215739785875e-06, "loss": 0.4809, "step": 18062 }, { "epoch": 0.6435327858631562, "grad_norm": 1.637781181350972, "learning_rate": 5.956366158390802e-06, "loss": 0.5529, "step": 18063 }, { "epoch": 0.6435684129896503, "grad_norm": 1.3655386880810663, "learning_rate": 5.955310796646826e-06, "loss": 0.5108, "step": 18064 }, { "epoch": 0.6436040401161445, "grad_norm": 1.3956100489029086, "learning_rate": 5.954255488760719e-06, "loss": 0.4933, "step": 18065 }, { "epoch": 0.6436396672426385, "grad_norm": 1.1246676164716394, "learning_rate": 5.953200234746524e-06, "loss": 0.436, "step": 18066 }, { "epoch": 0.6436752943691326, "grad_norm": 1.0990632222670769, "learning_rate": 5.9521450346182975e-06, "loss": 0.4199, "step": 18067 }, { "epoch": 0.6437109214956268, "grad_norm": 1.2174594549006186, "learning_rate": 5.951089888390087e-06, "loss": 0.3919, "step": 18068 }, { "epoch": 0.6437465486221209, "grad_norm": 1.1085454482594337, "learning_rate": 5.950034796075948e-06, "loss": 0.428, "step": 18069 }, { "epoch": 0.643782175748615, "grad_norm": 2.141244623376678, "learning_rate": 5.948979757689923e-06, "loss": 0.4494, "step": 18070 }, { "epoch": 0.6438178028751091, "grad_norm": 1.105737718227038, "learning_rate": 5.947924773246068e-06, "loss": 0.4399, "step": 18071 }, { "epoch": 0.6438534300016032, "grad_norm": 1.074967138434885, "learning_rate": 5.946869842758425e-06, "loss": 0.3838, "step": 18072 }, { "epoch": 0.6438890571280973, "grad_norm": 1.8184167116650198, "learning_rate": 5.945814966241043e-06, "loss": 0.5434, "step": 18073 }, { "epoch": 0.6439246842545915, "grad_norm": 1.2259809896506964, "learning_rate": 5.944760143707966e-06, "loss": 0.4621, "step": 18074 }, { "epoch": 0.6439603113810856, "grad_norm": 1.0917488100858734, "learning_rate": 5.943705375173243e-06, "loss": 0.4195, "step": 18075 }, { "epoch": 0.6439959385075796, "grad_norm": 1.0821188913799475, "learning_rate": 5.942650660650923e-06, "loss": 0.4199, "step": 18076 }, { "epoch": 0.6440315656340738, "grad_norm": 1.1959941288012768, "learning_rate": 5.941596000155041e-06, "loss": 0.3989, "step": 18077 }, { "epoch": 0.6440671927605679, "grad_norm": 1.5577472306087816, "learning_rate": 5.940541393699646e-06, "loss": 0.544, "step": 18078 }, { "epoch": 0.644102819887062, "grad_norm": 1.2879045861747214, "learning_rate": 5.939486841298779e-06, "loss": 0.5102, "step": 18079 }, { "epoch": 0.6441384470135562, "grad_norm": 1.621231055619626, "learning_rate": 5.9384323429664845e-06, "loss": 0.5394, "step": 18080 }, { "epoch": 0.6441740741400502, "grad_norm": 1.4789123682834595, "learning_rate": 5.9373778987167984e-06, "loss": 0.5782, "step": 18081 }, { "epoch": 0.6442097012665443, "grad_norm": 1.6367014094036725, "learning_rate": 5.93632350856377e-06, "loss": 0.576, "step": 18082 }, { "epoch": 0.6442453283930385, "grad_norm": 1.4673496252842089, "learning_rate": 5.935269172521432e-06, "loss": 0.4572, "step": 18083 }, { "epoch": 0.6442809555195326, "grad_norm": 1.3015057286381628, "learning_rate": 5.934214890603826e-06, "loss": 0.4383, "step": 18084 }, { "epoch": 0.6443165826460266, "grad_norm": 1.549099622797635, "learning_rate": 5.933160662824991e-06, "loss": 0.5886, "step": 18085 }, { "epoch": 0.6443522097725208, "grad_norm": 1.1158108409991134, "learning_rate": 5.932106489198962e-06, "loss": 0.4638, "step": 18086 }, { "epoch": 0.6443878368990149, "grad_norm": 1.3454302605373651, "learning_rate": 5.931052369739783e-06, "loss": 0.5316, "step": 18087 }, { "epoch": 0.644423464025509, "grad_norm": 1.4684039921661853, "learning_rate": 5.929998304461482e-06, "loss": 0.5246, "step": 18088 }, { "epoch": 0.6444590911520032, "grad_norm": 1.2390660498602553, "learning_rate": 5.928944293378101e-06, "loss": 0.4268, "step": 18089 }, { "epoch": 0.6444947182784972, "grad_norm": 1.1629445683642745, "learning_rate": 5.92789033650367e-06, "loss": 0.3929, "step": 18090 }, { "epoch": 0.6445303454049913, "grad_norm": 1.2536565766199999, "learning_rate": 5.926836433852228e-06, "loss": 0.4155, "step": 18091 }, { "epoch": 0.6445659725314855, "grad_norm": 1.2301739738538395, "learning_rate": 5.9257825854378024e-06, "loss": 0.4578, "step": 18092 }, { "epoch": 0.6446015996579796, "grad_norm": 1.17016002088814, "learning_rate": 5.924728791274432e-06, "loss": 0.5187, "step": 18093 }, { "epoch": 0.6446372267844737, "grad_norm": 1.1588599457268707, "learning_rate": 5.923675051376144e-06, "loss": 0.4531, "step": 18094 }, { "epoch": 0.6446728539109678, "grad_norm": 1.7025028011402037, "learning_rate": 5.922621365756974e-06, "loss": 0.5375, "step": 18095 }, { "epoch": 0.6447084810374619, "grad_norm": 1.5338200765237768, "learning_rate": 5.921567734430949e-06, "loss": 0.5513, "step": 18096 }, { "epoch": 0.644744108163956, "grad_norm": 1.2544412506021796, "learning_rate": 5.920514157412099e-06, "loss": 0.4618, "step": 18097 }, { "epoch": 0.6447797352904502, "grad_norm": 1.5048372045437528, "learning_rate": 5.919460634714459e-06, "loss": 0.5343, "step": 18098 }, { "epoch": 0.6448153624169443, "grad_norm": 1.681020930034511, "learning_rate": 5.91840716635205e-06, "loss": 0.5615, "step": 18099 }, { "epoch": 0.6448509895434383, "grad_norm": 1.769785129775597, "learning_rate": 5.917353752338902e-06, "loss": 0.5383, "step": 18100 }, { "epoch": 0.6448866166699325, "grad_norm": 1.0987007049124484, "learning_rate": 5.916300392689042e-06, "loss": 0.4286, "step": 18101 }, { "epoch": 0.6449222437964266, "grad_norm": 1.288062092101118, "learning_rate": 5.9152470874165004e-06, "loss": 0.4491, "step": 18102 }, { "epoch": 0.6449578709229207, "grad_norm": 1.3304531443296845, "learning_rate": 5.914193836535294e-06, "loss": 0.5492, "step": 18103 }, { "epoch": 0.6449934980494149, "grad_norm": 6.837073559416597, "learning_rate": 5.913140640059456e-06, "loss": 0.5649, "step": 18104 }, { "epoch": 0.6450291251759089, "grad_norm": 1.3799190971715531, "learning_rate": 5.912087498003003e-06, "loss": 0.4974, "step": 18105 }, { "epoch": 0.645064752302403, "grad_norm": 1.2335715550927808, "learning_rate": 5.911034410379968e-06, "loss": 0.425, "step": 18106 }, { "epoch": 0.6451003794288972, "grad_norm": 1.1806514518684874, "learning_rate": 5.909981377204361e-06, "loss": 0.4076, "step": 18107 }, { "epoch": 0.6451360065553913, "grad_norm": 1.308635404910487, "learning_rate": 5.908928398490217e-06, "loss": 0.438, "step": 18108 }, { "epoch": 0.6451716336818853, "grad_norm": 1.7243950056933812, "learning_rate": 5.907875474251546e-06, "loss": 0.5855, "step": 18109 }, { "epoch": 0.6452072608083795, "grad_norm": 1.148330157908428, "learning_rate": 5.906822604502375e-06, "loss": 0.3769, "step": 18110 }, { "epoch": 0.6452428879348736, "grad_norm": 1.979541957452564, "learning_rate": 5.905769789256722e-06, "loss": 0.5813, "step": 18111 }, { "epoch": 0.6452785150613677, "grad_norm": 1.414577563397028, "learning_rate": 5.904717028528606e-06, "loss": 0.422, "step": 18112 }, { "epoch": 0.6453141421878619, "grad_norm": 1.6403106177612286, "learning_rate": 5.903664322332048e-06, "loss": 0.5718, "step": 18113 }, { "epoch": 0.645349769314356, "grad_norm": 1.1980281531118238, "learning_rate": 5.902611670681058e-06, "loss": 0.4163, "step": 18114 }, { "epoch": 0.64538539644085, "grad_norm": 1.205321671392154, "learning_rate": 5.901559073589661e-06, "loss": 0.4196, "step": 18115 }, { "epoch": 0.6454210235673442, "grad_norm": 1.8867716356020152, "learning_rate": 5.900506531071867e-06, "loss": 0.5379, "step": 18116 }, { "epoch": 0.6454566506938383, "grad_norm": 1.1414022306368468, "learning_rate": 5.899454043141697e-06, "loss": 0.4411, "step": 18117 }, { "epoch": 0.6454922778203324, "grad_norm": 1.6924621218248308, "learning_rate": 5.8984016098131605e-06, "loss": 0.5028, "step": 18118 }, { "epoch": 0.6455279049468265, "grad_norm": 1.3280752732201455, "learning_rate": 5.897349231100273e-06, "loss": 0.3994, "step": 18119 }, { "epoch": 0.6455635320733206, "grad_norm": 1.444998348761537, "learning_rate": 5.896296907017049e-06, "loss": 0.4532, "step": 18120 }, { "epoch": 0.6455991591998147, "grad_norm": 1.19087949874981, "learning_rate": 5.895244637577498e-06, "loss": 0.4562, "step": 18121 }, { "epoch": 0.6456347863263089, "grad_norm": 1.1609943386127797, "learning_rate": 5.894192422795635e-06, "loss": 0.4857, "step": 18122 }, { "epoch": 0.645670413452803, "grad_norm": 1.4116416625714807, "learning_rate": 5.893140262685469e-06, "loss": 0.6138, "step": 18123 }, { "epoch": 0.645706040579297, "grad_norm": 1.4029547787860492, "learning_rate": 5.892088157261016e-06, "loss": 0.5751, "step": 18124 }, { "epoch": 0.6457416677057912, "grad_norm": 1.0527253052592072, "learning_rate": 5.891036106536274e-06, "loss": 0.407, "step": 18125 }, { "epoch": 0.6457772948322853, "grad_norm": 1.1740263243156603, "learning_rate": 5.889984110525262e-06, "loss": 0.4622, "step": 18126 }, { "epoch": 0.6458129219587794, "grad_norm": 1.1813241854759158, "learning_rate": 5.888932169241981e-06, "loss": 0.4271, "step": 18127 }, { "epoch": 0.6458485490852736, "grad_norm": 1.1523580864584293, "learning_rate": 5.8878802827004465e-06, "loss": 0.4682, "step": 18128 }, { "epoch": 0.6458841762117676, "grad_norm": 1.0362823150321439, "learning_rate": 5.8868284509146564e-06, "loss": 0.4065, "step": 18129 }, { "epoch": 0.6459198033382617, "grad_norm": 2.240957070450203, "learning_rate": 5.885776673898622e-06, "loss": 0.4529, "step": 18130 }, { "epoch": 0.6459554304647559, "grad_norm": 1.5588532951713547, "learning_rate": 5.884724951666345e-06, "loss": 0.5293, "step": 18131 }, { "epoch": 0.64599105759125, "grad_norm": 1.683669881483128, "learning_rate": 5.883673284231831e-06, "loss": 0.5083, "step": 18132 }, { "epoch": 0.646026684717744, "grad_norm": 1.30715101978035, "learning_rate": 5.882621671609087e-06, "loss": 0.4243, "step": 18133 }, { "epoch": 0.6460623118442382, "grad_norm": 1.0942284561327407, "learning_rate": 5.8815701138121115e-06, "loss": 0.4374, "step": 18134 }, { "epoch": 0.6460979389707323, "grad_norm": 1.2736188567068598, "learning_rate": 5.8805186108549114e-06, "loss": 0.4274, "step": 18135 }, { "epoch": 0.6461335660972264, "grad_norm": 1.169458160249946, "learning_rate": 5.879467162751482e-06, "loss": 0.4615, "step": 18136 }, { "epoch": 0.6461691932237206, "grad_norm": 1.5014686019080878, "learning_rate": 5.878415769515831e-06, "loss": 0.525, "step": 18137 }, { "epoch": 0.6462048203502146, "grad_norm": 1.6343378579186159, "learning_rate": 5.877364431161952e-06, "loss": 0.429, "step": 18138 }, { "epoch": 0.6462404474767087, "grad_norm": 1.4648861292163207, "learning_rate": 5.876313147703849e-06, "loss": 0.5633, "step": 18139 }, { "epoch": 0.6462760746032029, "grad_norm": 1.4530441462445058, "learning_rate": 5.875261919155517e-06, "loss": 0.5329, "step": 18140 }, { "epoch": 0.646311701729697, "grad_norm": 1.0519065654230584, "learning_rate": 5.8742107455309574e-06, "loss": 0.3909, "step": 18141 }, { "epoch": 0.6463473288561911, "grad_norm": 2.0102208258060377, "learning_rate": 5.873159626844163e-06, "loss": 0.5218, "step": 18142 }, { "epoch": 0.6463829559826852, "grad_norm": 1.2907506356100418, "learning_rate": 5.872108563109132e-06, "loss": 0.4317, "step": 18143 }, { "epoch": 0.6464185831091793, "grad_norm": 1.3507102341492425, "learning_rate": 5.871057554339865e-06, "loss": 0.4628, "step": 18144 }, { "epoch": 0.6464542102356734, "grad_norm": 1.6140071451802223, "learning_rate": 5.870006600550348e-06, "loss": 0.5118, "step": 18145 }, { "epoch": 0.6464898373621676, "grad_norm": 1.3944377324147448, "learning_rate": 5.868955701754584e-06, "loss": 0.472, "step": 18146 }, { "epoch": 0.6465254644886617, "grad_norm": 1.1909468310267985, "learning_rate": 5.86790485796656e-06, "loss": 0.4207, "step": 18147 }, { "epoch": 0.6465610916151557, "grad_norm": 1.4101565650238237, "learning_rate": 5.8668540692002714e-06, "loss": 0.4953, "step": 18148 }, { "epoch": 0.6465967187416499, "grad_norm": 1.5119610432698436, "learning_rate": 5.865803335469708e-06, "loss": 0.6699, "step": 18149 }, { "epoch": 0.646632345868144, "grad_norm": 1.1655539840145175, "learning_rate": 5.8647526567888676e-06, "loss": 0.4532, "step": 18150 }, { "epoch": 0.6466679729946381, "grad_norm": 1.5743214386722442, "learning_rate": 5.863702033171732e-06, "loss": 0.5724, "step": 18151 }, { "epoch": 0.6467036001211323, "grad_norm": 1.5254708868499693, "learning_rate": 5.8626514646322965e-06, "loss": 0.5664, "step": 18152 }, { "epoch": 0.6467392272476263, "grad_norm": 1.051112604051448, "learning_rate": 5.861600951184545e-06, "loss": 0.4194, "step": 18153 }, { "epoch": 0.6467748543741204, "grad_norm": 1.1099205565379366, "learning_rate": 5.860550492842476e-06, "loss": 0.383, "step": 18154 }, { "epoch": 0.6468104815006146, "grad_norm": 1.3757895424827657, "learning_rate": 5.859500089620065e-06, "loss": 0.5328, "step": 18155 }, { "epoch": 0.6468461086271087, "grad_norm": 1.3634927027289094, "learning_rate": 5.858449741531303e-06, "loss": 0.5181, "step": 18156 }, { "epoch": 0.6468817357536027, "grad_norm": 1.148443676557731, "learning_rate": 5.8573994485901835e-06, "loss": 0.4156, "step": 18157 }, { "epoch": 0.6469173628800969, "grad_norm": 1.4265108817129362, "learning_rate": 5.856349210810681e-06, "loss": 0.5884, "step": 18158 }, { "epoch": 0.646952990006591, "grad_norm": 1.1346861379151523, "learning_rate": 5.855299028206788e-06, "loss": 0.4287, "step": 18159 }, { "epoch": 0.6469886171330851, "grad_norm": 1.5902754918200825, "learning_rate": 5.854248900792483e-06, "loss": 0.5415, "step": 18160 }, { "epoch": 0.6470242442595793, "grad_norm": 1.8571481599408364, "learning_rate": 5.8531988285817555e-06, "loss": 0.5107, "step": 18161 }, { "epoch": 0.6470598713860733, "grad_norm": 1.58852733125174, "learning_rate": 5.852148811588581e-06, "loss": 0.4576, "step": 18162 }, { "epoch": 0.6470954985125674, "grad_norm": 1.6252157712424065, "learning_rate": 5.851098849826944e-06, "loss": 0.6119, "step": 18163 }, { "epoch": 0.6471311256390616, "grad_norm": 1.1547747060766111, "learning_rate": 5.8500489433108265e-06, "loss": 0.4559, "step": 18164 }, { "epoch": 0.6471667527655557, "grad_norm": 1.455685238606035, "learning_rate": 5.848999092054211e-06, "loss": 0.5406, "step": 18165 }, { "epoch": 0.6472023798920498, "grad_norm": 1.534750989472876, "learning_rate": 5.847949296071068e-06, "loss": 0.4897, "step": 18166 }, { "epoch": 0.647238007018544, "grad_norm": 1.3505546534966182, "learning_rate": 5.846899555375385e-06, "loss": 0.4795, "step": 18167 }, { "epoch": 0.647273634145038, "grad_norm": 1.7082664577953872, "learning_rate": 5.845849869981137e-06, "loss": 0.5497, "step": 18168 }, { "epoch": 0.6473092612715321, "grad_norm": 1.155001181415648, "learning_rate": 5.844800239902301e-06, "loss": 0.3518, "step": 18169 }, { "epoch": 0.6473448883980263, "grad_norm": 1.220232546110414, "learning_rate": 5.8437506651528555e-06, "loss": 0.4788, "step": 18170 }, { "epoch": 0.6473805155245204, "grad_norm": 1.2412803017038314, "learning_rate": 5.842701145746773e-06, "loss": 0.4241, "step": 18171 }, { "epoch": 0.6474161426510144, "grad_norm": 1.0511170349382664, "learning_rate": 5.8416516816980355e-06, "loss": 0.3863, "step": 18172 }, { "epoch": 0.6474517697775086, "grad_norm": 1.3851320323150347, "learning_rate": 5.840602273020607e-06, "loss": 0.5087, "step": 18173 }, { "epoch": 0.6474873969040027, "grad_norm": 1.5910794783710458, "learning_rate": 5.839552919728469e-06, "loss": 0.4937, "step": 18174 }, { "epoch": 0.6475230240304968, "grad_norm": 1.2301905808334526, "learning_rate": 5.838503621835592e-06, "loss": 0.4895, "step": 18175 }, { "epoch": 0.647558651156991, "grad_norm": 1.053349579633706, "learning_rate": 5.8374543793559455e-06, "loss": 0.4032, "step": 18176 }, { "epoch": 0.647594278283485, "grad_norm": 1.1643775127253917, "learning_rate": 5.836405192303504e-06, "loss": 0.4486, "step": 18177 }, { "epoch": 0.6476299054099791, "grad_norm": 1.1820356542171648, "learning_rate": 5.835356060692236e-06, "loss": 0.4334, "step": 18178 }, { "epoch": 0.6476655325364733, "grad_norm": 1.2459690846318496, "learning_rate": 5.83430698453612e-06, "loss": 0.4052, "step": 18179 }, { "epoch": 0.6477011596629674, "grad_norm": 1.1296399988687167, "learning_rate": 5.833257963849111e-06, "loss": 0.418, "step": 18180 }, { "epoch": 0.6477367867894614, "grad_norm": 1.2541149546944217, "learning_rate": 5.8322089986451906e-06, "loss": 0.4254, "step": 18181 }, { "epoch": 0.6477724139159556, "grad_norm": 1.530583038365505, "learning_rate": 5.831160088938315e-06, "loss": 0.5341, "step": 18182 }, { "epoch": 0.6478080410424497, "grad_norm": 1.160312065095979, "learning_rate": 5.830111234742459e-06, "loss": 0.4153, "step": 18183 }, { "epoch": 0.6478436681689438, "grad_norm": 1.8960524642430696, "learning_rate": 5.829062436071584e-06, "loss": 0.5537, "step": 18184 }, { "epoch": 0.647879295295438, "grad_norm": 1.0990551479791182, "learning_rate": 5.828013692939664e-06, "loss": 0.4237, "step": 18185 }, { "epoch": 0.647914922421932, "grad_norm": 1.0941658216655061, "learning_rate": 5.826965005360652e-06, "loss": 0.4839, "step": 18186 }, { "epoch": 0.6479505495484261, "grad_norm": 1.189854287767438, "learning_rate": 5.825916373348524e-06, "loss": 0.388, "step": 18187 }, { "epoch": 0.6479861766749203, "grad_norm": 1.1317874874305556, "learning_rate": 5.82486779691723e-06, "loss": 0.4379, "step": 18188 }, { "epoch": 0.6480218038014144, "grad_norm": 1.4842596595092088, "learning_rate": 5.823819276080741e-06, "loss": 0.5104, "step": 18189 }, { "epoch": 0.6480574309279085, "grad_norm": 1.4304881215837617, "learning_rate": 5.8227708108530225e-06, "loss": 0.585, "step": 18190 }, { "epoch": 0.6480930580544026, "grad_norm": 1.5123525210275939, "learning_rate": 5.821722401248025e-06, "loss": 0.5022, "step": 18191 }, { "epoch": 0.6481286851808967, "grad_norm": 1.5100956885293744, "learning_rate": 5.820674047279713e-06, "loss": 0.4842, "step": 18192 }, { "epoch": 0.6481643123073908, "grad_norm": 1.1436303984719958, "learning_rate": 5.819625748962049e-06, "loss": 0.4597, "step": 18193 }, { "epoch": 0.648199939433885, "grad_norm": 1.1766878420909186, "learning_rate": 5.818577506308993e-06, "loss": 0.4406, "step": 18194 }, { "epoch": 0.6482355665603791, "grad_norm": 1.41656258385111, "learning_rate": 5.8175293193344975e-06, "loss": 0.5206, "step": 18195 }, { "epoch": 0.6482711936868731, "grad_norm": 1.6144435573992253, "learning_rate": 5.8164811880525254e-06, "loss": 0.4849, "step": 18196 }, { "epoch": 0.6483068208133673, "grad_norm": 1.6335094883169377, "learning_rate": 5.815433112477027e-06, "loss": 0.5294, "step": 18197 }, { "epoch": 0.6483424479398614, "grad_norm": 1.2538233550852877, "learning_rate": 5.814385092621961e-06, "loss": 0.4156, "step": 18198 }, { "epoch": 0.6483780750663555, "grad_norm": 1.0708234736433264, "learning_rate": 5.813337128501284e-06, "loss": 0.4184, "step": 18199 }, { "epoch": 0.6484137021928497, "grad_norm": 1.4508601307286177, "learning_rate": 5.812289220128953e-06, "loss": 0.5318, "step": 18200 }, { "epoch": 0.6484493293193437, "grad_norm": 1.2681395891932699, "learning_rate": 5.811241367518914e-06, "loss": 0.4872, "step": 18201 }, { "epoch": 0.6484849564458378, "grad_norm": 1.6639249568718468, "learning_rate": 5.8101935706851235e-06, "loss": 0.5779, "step": 18202 }, { "epoch": 0.648520583572332, "grad_norm": 1.2849275042249038, "learning_rate": 5.809145829641538e-06, "loss": 0.4668, "step": 18203 }, { "epoch": 0.6485562106988261, "grad_norm": 2.0373683973212207, "learning_rate": 5.808098144402101e-06, "loss": 0.6108, "step": 18204 }, { "epoch": 0.6485918378253201, "grad_norm": 1.0597816610703341, "learning_rate": 5.807050514980772e-06, "loss": 0.4336, "step": 18205 }, { "epoch": 0.6486274649518143, "grad_norm": 1.2515810920570476, "learning_rate": 5.8060029413914905e-06, "loss": 0.4287, "step": 18206 }, { "epoch": 0.6486630920783084, "grad_norm": 1.1539200847701705, "learning_rate": 5.804955423648213e-06, "loss": 0.3983, "step": 18207 }, { "epoch": 0.6486987192048025, "grad_norm": 1.598987428375962, "learning_rate": 5.803907961764886e-06, "loss": 0.5722, "step": 18208 }, { "epoch": 0.6487343463312967, "grad_norm": 1.3602370009618876, "learning_rate": 5.80286055575546e-06, "loss": 0.4124, "step": 18209 }, { "epoch": 0.6487699734577907, "grad_norm": 1.2993822193318496, "learning_rate": 5.8018132056338776e-06, "loss": 0.4703, "step": 18210 }, { "epoch": 0.6488056005842848, "grad_norm": 2.6191027305416124, "learning_rate": 5.80076591141409e-06, "loss": 0.4642, "step": 18211 }, { "epoch": 0.648841227710779, "grad_norm": 1.4183048282166184, "learning_rate": 5.7997186731100355e-06, "loss": 0.4731, "step": 18212 }, { "epoch": 0.6488768548372731, "grad_norm": 1.227592515110191, "learning_rate": 5.7986714907356614e-06, "loss": 0.446, "step": 18213 }, { "epoch": 0.6489124819637672, "grad_norm": 1.124804372151042, "learning_rate": 5.7976243643049124e-06, "loss": 0.4617, "step": 18214 }, { "epoch": 0.6489481090902613, "grad_norm": 1.6928576131758066, "learning_rate": 5.796577293831734e-06, "loss": 0.5737, "step": 18215 }, { "epoch": 0.6489837362167554, "grad_norm": 1.6803477883751619, "learning_rate": 5.79553027933007e-06, "loss": 0.5974, "step": 18216 }, { "epoch": 0.6490193633432495, "grad_norm": 1.9210023448683249, "learning_rate": 5.794483320813856e-06, "loss": 0.5127, "step": 18217 }, { "epoch": 0.6490549904697437, "grad_norm": 1.5501492276539681, "learning_rate": 5.79343641829704e-06, "loss": 0.4595, "step": 18218 }, { "epoch": 0.6490906175962378, "grad_norm": 1.6208540674885161, "learning_rate": 5.792389571793553e-06, "loss": 0.5036, "step": 18219 }, { "epoch": 0.6491262447227318, "grad_norm": 1.3758237821818962, "learning_rate": 5.791342781317345e-06, "loss": 0.4412, "step": 18220 }, { "epoch": 0.649161871849226, "grad_norm": 1.2458142915342847, "learning_rate": 5.790296046882342e-06, "loss": 0.4316, "step": 18221 }, { "epoch": 0.6491974989757201, "grad_norm": 1.1022143860896967, "learning_rate": 5.789249368502496e-06, "loss": 0.403, "step": 18222 }, { "epoch": 0.6492331261022142, "grad_norm": 1.2897133874274231, "learning_rate": 5.788202746191735e-06, "loss": 0.4181, "step": 18223 }, { "epoch": 0.6492687532287084, "grad_norm": 1.2158276019443057, "learning_rate": 5.787156179963998e-06, "loss": 0.4718, "step": 18224 }, { "epoch": 0.6493043803552024, "grad_norm": 1.299397027266554, "learning_rate": 5.786109669833224e-06, "loss": 0.4918, "step": 18225 }, { "epoch": 0.6493400074816965, "grad_norm": 0.9931831326069843, "learning_rate": 5.785063215813342e-06, "loss": 0.357, "step": 18226 }, { "epoch": 0.6493756346081907, "grad_norm": 1.2059430663298505, "learning_rate": 5.784016817918294e-06, "loss": 0.425, "step": 18227 }, { "epoch": 0.6494112617346848, "grad_norm": 2.5570038598546185, "learning_rate": 5.782970476162004e-06, "loss": 0.5337, "step": 18228 }, { "epoch": 0.6494468888611789, "grad_norm": 1.2088786727390926, "learning_rate": 5.78192419055841e-06, "loss": 0.4451, "step": 18229 }, { "epoch": 0.649482515987673, "grad_norm": 1.1828787649482977, "learning_rate": 5.7808779611214425e-06, "loss": 0.4257, "step": 18230 }, { "epoch": 0.6495181431141671, "grad_norm": 1.3019949114397162, "learning_rate": 5.77983178786504e-06, "loss": 0.4381, "step": 18231 }, { "epoch": 0.6495537702406612, "grad_norm": 1.8471479079280266, "learning_rate": 5.778785670803122e-06, "loss": 0.5111, "step": 18232 }, { "epoch": 0.6495893973671554, "grad_norm": 1.1390610764659739, "learning_rate": 5.777739609949626e-06, "loss": 0.4315, "step": 18233 }, { "epoch": 0.6496250244936495, "grad_norm": 1.650737995002818, "learning_rate": 5.7766936053184764e-06, "loss": 0.575, "step": 18234 }, { "epoch": 0.6496606516201435, "grad_norm": 1.1875186134583737, "learning_rate": 5.775647656923602e-06, "loss": 0.4032, "step": 18235 }, { "epoch": 0.6496962787466377, "grad_norm": 1.0405396007465828, "learning_rate": 5.774601764778933e-06, "loss": 0.4229, "step": 18236 }, { "epoch": 0.6497319058731318, "grad_norm": 1.5971208858156931, "learning_rate": 5.773555928898393e-06, "loss": 0.5114, "step": 18237 }, { "epoch": 0.6497675329996259, "grad_norm": 1.6019239829752894, "learning_rate": 5.772510149295916e-06, "loss": 0.4872, "step": 18238 }, { "epoch": 0.64980316012612, "grad_norm": 1.1863133356666047, "learning_rate": 5.771464425985416e-06, "loss": 0.4069, "step": 18239 }, { "epoch": 0.6498387872526141, "grad_norm": 1.0885681463129406, "learning_rate": 5.770418758980827e-06, "loss": 0.4216, "step": 18240 }, { "epoch": 0.6498744143791082, "grad_norm": 1.8622897090870678, "learning_rate": 5.769373148296063e-06, "loss": 0.5259, "step": 18241 }, { "epoch": 0.6499100415056024, "grad_norm": 1.1085364453386197, "learning_rate": 5.7683275939450575e-06, "loss": 0.4334, "step": 18242 }, { "epoch": 0.6499456686320965, "grad_norm": 1.5815762490183565, "learning_rate": 5.767282095941725e-06, "loss": 0.4803, "step": 18243 }, { "epoch": 0.6499812957585905, "grad_norm": 1.012246720354825, "learning_rate": 5.7662366542999896e-06, "loss": 0.414, "step": 18244 }, { "epoch": 0.6500169228850847, "grad_norm": 1.604693642364885, "learning_rate": 5.765191269033772e-06, "loss": 0.5396, "step": 18245 }, { "epoch": 0.6500525500115788, "grad_norm": 1.5605955664895828, "learning_rate": 5.764145940156996e-06, "loss": 0.5596, "step": 18246 }, { "epoch": 0.6500881771380729, "grad_norm": 1.1174474770649867, "learning_rate": 5.763100667683574e-06, "loss": 0.399, "step": 18247 }, { "epoch": 0.6501238042645671, "grad_norm": 1.1088807327584458, "learning_rate": 5.762055451627428e-06, "loss": 0.4372, "step": 18248 }, { "epoch": 0.6501594313910611, "grad_norm": 1.6781202445588497, "learning_rate": 5.76101029200248e-06, "loss": 0.5366, "step": 18249 }, { "epoch": 0.6501950585175552, "grad_norm": 1.964960944647918, "learning_rate": 5.759965188822638e-06, "loss": 0.4925, "step": 18250 }, { "epoch": 0.6502306856440494, "grad_norm": 1.0673100137070304, "learning_rate": 5.758920142101825e-06, "loss": 0.4238, "step": 18251 }, { "epoch": 0.6502663127705435, "grad_norm": 1.1902687707437813, "learning_rate": 5.757875151853953e-06, "loss": 0.4209, "step": 18252 }, { "epoch": 0.6503019398970376, "grad_norm": 1.0028350802638986, "learning_rate": 5.756830218092942e-06, "loss": 0.4363, "step": 18253 }, { "epoch": 0.6503375670235317, "grad_norm": 1.4878297712980049, "learning_rate": 5.7557853408327e-06, "loss": 0.4958, "step": 18254 }, { "epoch": 0.6503731941500258, "grad_norm": 1.4215241687281661, "learning_rate": 5.754740520087145e-06, "loss": 0.488, "step": 18255 }, { "epoch": 0.6504088212765199, "grad_norm": 2.533571979126509, "learning_rate": 5.753695755870185e-06, "loss": 0.4983, "step": 18256 }, { "epoch": 0.6504444484030141, "grad_norm": 1.2535246651906802, "learning_rate": 5.752651048195738e-06, "loss": 0.5059, "step": 18257 }, { "epoch": 0.6504800755295082, "grad_norm": 1.1237521145207654, "learning_rate": 5.751606397077703e-06, "loss": 0.4854, "step": 18258 }, { "epoch": 0.6505157026560022, "grad_norm": 1.095289985875709, "learning_rate": 5.75056180253e-06, "loss": 0.4198, "step": 18259 }, { "epoch": 0.6505513297824964, "grad_norm": 1.5232494893792148, "learning_rate": 5.749517264566542e-06, "loss": 0.4644, "step": 18260 }, { "epoch": 0.6505869569089905, "grad_norm": 1.9559264807445087, "learning_rate": 5.748472783201229e-06, "loss": 0.5891, "step": 18261 }, { "epoch": 0.6506225840354846, "grad_norm": 1.4628809498403912, "learning_rate": 5.747428358447977e-06, "loss": 0.5434, "step": 18262 }, { "epoch": 0.6506582111619788, "grad_norm": 3.2619176536844194, "learning_rate": 5.7463839903206844e-06, "loss": 0.496, "step": 18263 }, { "epoch": 0.6506938382884728, "grad_norm": 1.1003788786449045, "learning_rate": 5.745339678833266e-06, "loss": 0.393, "step": 18264 }, { "epoch": 0.6507294654149669, "grad_norm": 1.1374536509290558, "learning_rate": 5.744295423999619e-06, "loss": 0.4181, "step": 18265 }, { "epoch": 0.6507650925414611, "grad_norm": 1.1655412986744316, "learning_rate": 5.743251225833653e-06, "loss": 0.3889, "step": 18266 }, { "epoch": 0.6508007196679552, "grad_norm": 1.0615784378403121, "learning_rate": 5.742207084349274e-06, "loss": 0.4562, "step": 18267 }, { "epoch": 0.6508363467944492, "grad_norm": 1.107748454679829, "learning_rate": 5.7411629995603855e-06, "loss": 0.4296, "step": 18268 }, { "epoch": 0.6508719739209434, "grad_norm": 1.7385032327508134, "learning_rate": 5.740118971480886e-06, "loss": 0.4992, "step": 18269 }, { "epoch": 0.6509076010474375, "grad_norm": 1.9244208744117244, "learning_rate": 5.739075000124679e-06, "loss": 0.4257, "step": 18270 }, { "epoch": 0.6509432281739316, "grad_norm": 1.0346002096244054, "learning_rate": 5.73803108550567e-06, "loss": 0.4151, "step": 18271 }, { "epoch": 0.6509788553004258, "grad_norm": 1.4514659771225373, "learning_rate": 5.736987227637752e-06, "loss": 0.6207, "step": 18272 }, { "epoch": 0.6510144824269198, "grad_norm": 1.1782266380832744, "learning_rate": 5.73594342653483e-06, "loss": 0.4639, "step": 18273 }, { "epoch": 0.651050109553414, "grad_norm": 3.2546850591624987, "learning_rate": 5.734899682210801e-06, "loss": 0.5534, "step": 18274 }, { "epoch": 0.6510857366799081, "grad_norm": 1.1696720780751135, "learning_rate": 5.733855994679569e-06, "loss": 0.439, "step": 18275 }, { "epoch": 0.6511213638064022, "grad_norm": 1.4349023331857964, "learning_rate": 5.732812363955021e-06, "loss": 0.5459, "step": 18276 }, { "epoch": 0.6511569909328964, "grad_norm": 1.130418139585636, "learning_rate": 5.731768790051062e-06, "loss": 0.4376, "step": 18277 }, { "epoch": 0.6511926180593904, "grad_norm": 1.6251210105370475, "learning_rate": 5.7307252729815835e-06, "loss": 0.5223, "step": 18278 }, { "epoch": 0.6512282451858845, "grad_norm": 2.4015638607636616, "learning_rate": 5.729681812760484e-06, "loss": 0.5823, "step": 18279 }, { "epoch": 0.6512638723123787, "grad_norm": 1.122878875448494, "learning_rate": 5.728638409401652e-06, "loss": 0.3882, "step": 18280 }, { "epoch": 0.6512994994388728, "grad_norm": 1.6087204076225026, "learning_rate": 5.727595062918987e-06, "loss": 0.4796, "step": 18281 }, { "epoch": 0.6513351265653669, "grad_norm": 1.4349395538530294, "learning_rate": 5.7265517733263785e-06, "loss": 0.5068, "step": 18282 }, { "epoch": 0.651370753691861, "grad_norm": 1.2090425948357906, "learning_rate": 5.72550854063772e-06, "loss": 0.4275, "step": 18283 }, { "epoch": 0.6514063808183551, "grad_norm": 1.5105375799138796, "learning_rate": 5.724465364866907e-06, "loss": 0.5357, "step": 18284 }, { "epoch": 0.6514420079448492, "grad_norm": 1.5692924380197102, "learning_rate": 5.723422246027822e-06, "loss": 0.5094, "step": 18285 }, { "epoch": 0.6514776350713434, "grad_norm": 1.1876513835316007, "learning_rate": 5.7223791841343635e-06, "loss": 0.3588, "step": 18286 }, { "epoch": 0.6515132621978375, "grad_norm": 1.5158782974441702, "learning_rate": 5.721336179200412e-06, "loss": 0.502, "step": 18287 }, { "epoch": 0.6515488893243315, "grad_norm": 1.3300735838220663, "learning_rate": 5.720293231239859e-06, "loss": 0.4258, "step": 18288 }, { "epoch": 0.6515845164508257, "grad_norm": 1.0471464618413193, "learning_rate": 5.719250340266595e-06, "loss": 0.3877, "step": 18289 }, { "epoch": 0.6516201435773198, "grad_norm": 1.086331566382387, "learning_rate": 5.718207506294506e-06, "loss": 0.4216, "step": 18290 }, { "epoch": 0.6516557707038139, "grad_norm": 1.6906864297128859, "learning_rate": 5.717164729337474e-06, "loss": 0.5364, "step": 18291 }, { "epoch": 0.651691397830308, "grad_norm": 1.802126399281343, "learning_rate": 5.716122009409393e-06, "loss": 0.614, "step": 18292 }, { "epoch": 0.6517270249568021, "grad_norm": 1.868125722355095, "learning_rate": 5.715079346524135e-06, "loss": 0.485, "step": 18293 }, { "epoch": 0.6517626520832962, "grad_norm": 1.359609237470628, "learning_rate": 5.7140367406955925e-06, "loss": 0.4597, "step": 18294 }, { "epoch": 0.6517982792097904, "grad_norm": 1.12625578258423, "learning_rate": 5.7129941919376485e-06, "loss": 0.4012, "step": 18295 }, { "epoch": 0.6518339063362845, "grad_norm": 1.591863558192426, "learning_rate": 5.711951700264176e-06, "loss": 0.4993, "step": 18296 }, { "epoch": 0.6518695334627785, "grad_norm": 1.5216590174351332, "learning_rate": 5.710909265689073e-06, "loss": 0.4464, "step": 18297 }, { "epoch": 0.6519051605892727, "grad_norm": 1.1757115033058714, "learning_rate": 5.709866888226206e-06, "loss": 0.4164, "step": 18298 }, { "epoch": 0.6519407877157668, "grad_norm": 1.331907728487787, "learning_rate": 5.7088245678894635e-06, "loss": 0.483, "step": 18299 }, { "epoch": 0.6519764148422609, "grad_norm": 1.1038369092112033, "learning_rate": 5.70778230469272e-06, "loss": 0.4133, "step": 18300 }, { "epoch": 0.6520120419687551, "grad_norm": 1.5489834711341637, "learning_rate": 5.706740098649857e-06, "loss": 0.4287, "step": 18301 }, { "epoch": 0.6520476690952491, "grad_norm": 1.2578826848215954, "learning_rate": 5.705697949774746e-06, "loss": 0.4387, "step": 18302 }, { "epoch": 0.6520832962217432, "grad_norm": 1.4168761474194411, "learning_rate": 5.704655858081268e-06, "loss": 0.5538, "step": 18303 }, { "epoch": 0.6521189233482374, "grad_norm": 1.1956066279432447, "learning_rate": 5.7036138235833005e-06, "loss": 0.4524, "step": 18304 }, { "epoch": 0.6521545504747315, "grad_norm": 1.1608215034265776, "learning_rate": 5.7025718462947175e-06, "loss": 0.3892, "step": 18305 }, { "epoch": 0.6521901776012256, "grad_norm": 1.3629806344330566, "learning_rate": 5.7015299262293984e-06, "loss": 0.5188, "step": 18306 }, { "epoch": 0.6522258047277197, "grad_norm": 1.5654559053535908, "learning_rate": 5.700488063401209e-06, "loss": 0.5496, "step": 18307 }, { "epoch": 0.6522614318542138, "grad_norm": 1.4212673530547721, "learning_rate": 5.6994462578240305e-06, "loss": 0.4782, "step": 18308 }, { "epoch": 0.6522970589807079, "grad_norm": 1.0809755875641758, "learning_rate": 5.698404509511726e-06, "loss": 0.4134, "step": 18309 }, { "epoch": 0.6523326861072021, "grad_norm": 1.1534476325371987, "learning_rate": 5.697362818478172e-06, "loss": 0.4113, "step": 18310 }, { "epoch": 0.6523683132336962, "grad_norm": 1.608987268691529, "learning_rate": 5.696321184737241e-06, "loss": 0.5369, "step": 18311 }, { "epoch": 0.6524039403601902, "grad_norm": 1.395049538780482, "learning_rate": 5.695279608302804e-06, "loss": 0.5383, "step": 18312 }, { "epoch": 0.6524395674866844, "grad_norm": 1.197608064922379, "learning_rate": 5.694238089188724e-06, "loss": 0.4031, "step": 18313 }, { "epoch": 0.6524751946131785, "grad_norm": 1.8717497817433928, "learning_rate": 5.693196627408879e-06, "loss": 0.632, "step": 18314 }, { "epoch": 0.6525108217396726, "grad_norm": 1.1440140270615724, "learning_rate": 5.692155222977125e-06, "loss": 0.4277, "step": 18315 }, { "epoch": 0.6525464488661668, "grad_norm": 2.0688734088221405, "learning_rate": 5.691113875907337e-06, "loss": 0.4626, "step": 18316 }, { "epoch": 0.6525820759926608, "grad_norm": 1.4125086031606808, "learning_rate": 5.690072586213382e-06, "loss": 0.4994, "step": 18317 }, { "epoch": 0.6526177031191549, "grad_norm": 1.1566103266630787, "learning_rate": 5.68903135390912e-06, "loss": 0.4396, "step": 18318 }, { "epoch": 0.6526533302456491, "grad_norm": 1.195165870431744, "learning_rate": 5.687990179008418e-06, "loss": 0.4348, "step": 18319 }, { "epoch": 0.6526889573721432, "grad_norm": 1.7588288100151808, "learning_rate": 5.68694906152514e-06, "loss": 0.541, "step": 18320 }, { "epoch": 0.6527245844986372, "grad_norm": 0.9964954194737606, "learning_rate": 5.685908001473155e-06, "loss": 0.3943, "step": 18321 }, { "epoch": 0.6527602116251314, "grad_norm": 1.0286043552442001, "learning_rate": 5.684866998866316e-06, "loss": 0.336, "step": 18322 }, { "epoch": 0.6527958387516255, "grad_norm": 1.7969816156150655, "learning_rate": 5.683826053718493e-06, "loss": 0.5754, "step": 18323 }, { "epoch": 0.6528314658781196, "grad_norm": 1.0838719084372113, "learning_rate": 5.682785166043541e-06, "loss": 0.4053, "step": 18324 }, { "epoch": 0.6528670930046138, "grad_norm": 1.2121906051901863, "learning_rate": 5.681744335855319e-06, "loss": 0.4257, "step": 18325 }, { "epoch": 0.6529027201311078, "grad_norm": 1.6030609265830267, "learning_rate": 5.680703563167691e-06, "loss": 0.593, "step": 18326 }, { "epoch": 0.6529383472576019, "grad_norm": 1.1570343815786523, "learning_rate": 5.679662847994518e-06, "loss": 0.4134, "step": 18327 }, { "epoch": 0.6529739743840961, "grad_norm": 1.4917527688387124, "learning_rate": 5.67862219034965e-06, "loss": 0.5303, "step": 18328 }, { "epoch": 0.6530096015105902, "grad_norm": 1.1450258065356005, "learning_rate": 5.6775815902469465e-06, "loss": 0.4248, "step": 18329 }, { "epoch": 0.6530452286370843, "grad_norm": 1.5149861208210786, "learning_rate": 5.67654104770027e-06, "loss": 0.5369, "step": 18330 }, { "epoch": 0.6530808557635784, "grad_norm": 1.4894691302486869, "learning_rate": 5.675500562723468e-06, "loss": 0.5095, "step": 18331 }, { "epoch": 0.6531164828900725, "grad_norm": 2.388185203151537, "learning_rate": 5.674460135330403e-06, "loss": 0.4895, "step": 18332 }, { "epoch": 0.6531521100165666, "grad_norm": 1.2345035425237914, "learning_rate": 5.673419765534915e-06, "loss": 0.4355, "step": 18333 }, { "epoch": 0.6531877371430608, "grad_norm": 1.0654911176634436, "learning_rate": 5.672379453350877e-06, "loss": 0.4196, "step": 18334 }, { "epoch": 0.6532233642695549, "grad_norm": 1.5385605553900943, "learning_rate": 5.671339198792126e-06, "loss": 0.4759, "step": 18335 }, { "epoch": 0.6532589913960489, "grad_norm": 1.3007060290512271, "learning_rate": 5.670299001872523e-06, "loss": 0.4592, "step": 18336 }, { "epoch": 0.6532946185225431, "grad_norm": 1.1245159061942558, "learning_rate": 5.669258862605913e-06, "loss": 0.4224, "step": 18337 }, { "epoch": 0.6533302456490372, "grad_norm": 1.8958791509750625, "learning_rate": 5.66821878100615e-06, "loss": 0.5702, "step": 18338 }, { "epoch": 0.6533658727755313, "grad_norm": 1.067143113620074, "learning_rate": 5.66717875708708e-06, "loss": 0.4566, "step": 18339 }, { "epoch": 0.6534014999020255, "grad_norm": 0.9706369312636752, "learning_rate": 5.66613879086255e-06, "loss": 0.4465, "step": 18340 }, { "epoch": 0.6534371270285195, "grad_norm": 2.139345208892775, "learning_rate": 5.665098882346414e-06, "loss": 0.4367, "step": 18341 }, { "epoch": 0.6534727541550136, "grad_norm": 1.2165579037600973, "learning_rate": 5.664059031552516e-06, "loss": 0.5023, "step": 18342 }, { "epoch": 0.6535083812815078, "grad_norm": 1.0570220218253477, "learning_rate": 5.663019238494705e-06, "loss": 0.4397, "step": 18343 }, { "epoch": 0.6535440084080019, "grad_norm": 1.2067684095455407, "learning_rate": 5.661979503186821e-06, "loss": 0.4308, "step": 18344 }, { "epoch": 0.6535796355344959, "grad_norm": 1.6816663916626493, "learning_rate": 5.660939825642716e-06, "loss": 0.5289, "step": 18345 }, { "epoch": 0.6536152626609901, "grad_norm": 1.6178703052192553, "learning_rate": 5.659900205876226e-06, "loss": 0.519, "step": 18346 }, { "epoch": 0.6536508897874842, "grad_norm": 1.0796968096187007, "learning_rate": 5.658860643901199e-06, "loss": 0.4, "step": 18347 }, { "epoch": 0.6536865169139783, "grad_norm": 1.2004745809313861, "learning_rate": 5.6578211397314765e-06, "loss": 0.3908, "step": 18348 }, { "epoch": 0.6537221440404725, "grad_norm": 1.1560526943298688, "learning_rate": 5.656781693380904e-06, "loss": 0.3933, "step": 18349 }, { "epoch": 0.6537577711669665, "grad_norm": 1.4205696374321852, "learning_rate": 5.6557423048633155e-06, "loss": 0.5602, "step": 18350 }, { "epoch": 0.6537933982934606, "grad_norm": 1.105214517130092, "learning_rate": 5.654702974192554e-06, "loss": 0.401, "step": 18351 }, { "epoch": 0.6538290254199548, "grad_norm": 1.6818621895574162, "learning_rate": 5.653663701382464e-06, "loss": 0.5614, "step": 18352 }, { "epoch": 0.6538646525464489, "grad_norm": 1.1281980928895834, "learning_rate": 5.652624486446876e-06, "loss": 0.4672, "step": 18353 }, { "epoch": 0.653900279672943, "grad_norm": 1.1299549069832864, "learning_rate": 5.651585329399636e-06, "loss": 0.4781, "step": 18354 }, { "epoch": 0.6539359067994371, "grad_norm": 1.1240023610540943, "learning_rate": 5.650546230254572e-06, "loss": 0.4729, "step": 18355 }, { "epoch": 0.6539715339259312, "grad_norm": 1.6119868743347667, "learning_rate": 5.649507189025525e-06, "loss": 0.5729, "step": 18356 }, { "epoch": 0.6540071610524253, "grad_norm": 1.8352849808477996, "learning_rate": 5.648468205726331e-06, "loss": 0.5061, "step": 18357 }, { "epoch": 0.6540427881789195, "grad_norm": 1.0115712630470701, "learning_rate": 5.647429280370828e-06, "loss": 0.4042, "step": 18358 }, { "epoch": 0.6540784153054136, "grad_norm": 1.0770057204527415, "learning_rate": 5.646390412972842e-06, "loss": 0.4459, "step": 18359 }, { "epoch": 0.6541140424319076, "grad_norm": 1.134178877550198, "learning_rate": 5.645351603546216e-06, "loss": 0.3978, "step": 18360 }, { "epoch": 0.6541496695584018, "grad_norm": 1.3102666677570387, "learning_rate": 5.644312852104772e-06, "loss": 0.5385, "step": 18361 }, { "epoch": 0.6541852966848959, "grad_norm": 1.0683388230250428, "learning_rate": 5.643274158662346e-06, "loss": 0.3817, "step": 18362 }, { "epoch": 0.65422092381139, "grad_norm": 1.14507108533903, "learning_rate": 5.642235523232772e-06, "loss": 0.3836, "step": 18363 }, { "epoch": 0.6542565509378842, "grad_norm": 1.4549447780526519, "learning_rate": 5.641196945829877e-06, "loss": 0.4857, "step": 18364 }, { "epoch": 0.6542921780643782, "grad_norm": 1.191266451688746, "learning_rate": 5.640158426467495e-06, "loss": 0.4244, "step": 18365 }, { "epoch": 0.6543278051908723, "grad_norm": 1.4731083480912122, "learning_rate": 5.639119965159447e-06, "loss": 0.561, "step": 18366 }, { "epoch": 0.6543634323173665, "grad_norm": 1.5580626722392688, "learning_rate": 5.638081561919568e-06, "loss": 0.5102, "step": 18367 }, { "epoch": 0.6543990594438606, "grad_norm": 1.2128156400037577, "learning_rate": 5.637043216761679e-06, "loss": 0.4496, "step": 18368 }, { "epoch": 0.6544346865703546, "grad_norm": 1.104852447718318, "learning_rate": 5.636004929699612e-06, "loss": 0.431, "step": 18369 }, { "epoch": 0.6544703136968488, "grad_norm": 1.5270716956862322, "learning_rate": 5.6349667007471866e-06, "loss": 0.5068, "step": 18370 }, { "epoch": 0.6545059408233429, "grad_norm": 1.7979037624932517, "learning_rate": 5.633928529918229e-06, "loss": 0.5049, "step": 18371 }, { "epoch": 0.654541567949837, "grad_norm": 2.0242631784336793, "learning_rate": 5.632890417226566e-06, "loss": 0.5097, "step": 18372 }, { "epoch": 0.6545771950763312, "grad_norm": 1.5202041605496805, "learning_rate": 5.631852362686023e-06, "loss": 0.4438, "step": 18373 }, { "epoch": 0.6546128222028252, "grad_norm": 1.460034645941949, "learning_rate": 5.6308143663104166e-06, "loss": 0.5394, "step": 18374 }, { "epoch": 0.6546484493293193, "grad_norm": 1.2419115317483325, "learning_rate": 5.629776428113569e-06, "loss": 0.4066, "step": 18375 }, { "epoch": 0.6546840764558135, "grad_norm": 1.51944219200712, "learning_rate": 5.628738548109307e-06, "loss": 0.4627, "step": 18376 }, { "epoch": 0.6547197035823076, "grad_norm": 1.2515793862702373, "learning_rate": 5.627700726311444e-06, "loss": 0.3964, "step": 18377 }, { "epoch": 0.6547553307088017, "grad_norm": 2.054885963250146, "learning_rate": 5.6266629627338e-06, "loss": 0.499, "step": 18378 }, { "epoch": 0.6547909578352958, "grad_norm": 1.700691203361343, "learning_rate": 5.625625257390197e-06, "loss": 0.5623, "step": 18379 }, { "epoch": 0.6548265849617899, "grad_norm": 1.382169782822952, "learning_rate": 5.624587610294454e-06, "loss": 0.4975, "step": 18380 }, { "epoch": 0.654862212088284, "grad_norm": 1.682866168078613, "learning_rate": 5.623550021460382e-06, "loss": 0.5892, "step": 18381 }, { "epoch": 0.6548978392147782, "grad_norm": 1.6888095143288677, "learning_rate": 5.622512490901804e-06, "loss": 0.493, "step": 18382 }, { "epoch": 0.6549334663412723, "grad_norm": 1.5822386823625447, "learning_rate": 5.621475018632527e-06, "loss": 0.5707, "step": 18383 }, { "epoch": 0.6549690934677663, "grad_norm": 1.1494159097550272, "learning_rate": 5.620437604666375e-06, "loss": 0.4507, "step": 18384 }, { "epoch": 0.6550047205942605, "grad_norm": 1.2272936105815515, "learning_rate": 5.619400249017151e-06, "loss": 0.383, "step": 18385 }, { "epoch": 0.6550403477207546, "grad_norm": 1.101447714307406, "learning_rate": 5.618362951698677e-06, "loss": 0.4554, "step": 18386 }, { "epoch": 0.6550759748472487, "grad_norm": 1.1305369555079083, "learning_rate": 5.6173257127247685e-06, "loss": 0.44, "step": 18387 }, { "epoch": 0.6551116019737429, "grad_norm": 1.243911990312469, "learning_rate": 5.616288532109225e-06, "loss": 0.4758, "step": 18388 }, { "epoch": 0.6551472291002369, "grad_norm": 1.4874749654545734, "learning_rate": 5.61525140986587e-06, "loss": 0.5133, "step": 18389 }, { "epoch": 0.655182856226731, "grad_norm": 1.4745537210766964, "learning_rate": 5.614214346008502e-06, "loss": 0.6122, "step": 18390 }, { "epoch": 0.6552184833532252, "grad_norm": 1.1225692002188201, "learning_rate": 5.613177340550939e-06, "loss": 0.4205, "step": 18391 }, { "epoch": 0.6552541104797193, "grad_norm": 1.5030217963325003, "learning_rate": 5.612140393506982e-06, "loss": 0.5208, "step": 18392 }, { "epoch": 0.6552897376062133, "grad_norm": 1.1872103433827235, "learning_rate": 5.611103504890444e-06, "loss": 0.4232, "step": 18393 }, { "epoch": 0.6553253647327075, "grad_norm": 1.7211107789471867, "learning_rate": 5.610066674715129e-06, "loss": 0.4786, "step": 18394 }, { "epoch": 0.6553609918592016, "grad_norm": 1.111272606589021, "learning_rate": 5.609029902994849e-06, "loss": 0.3912, "step": 18395 }, { "epoch": 0.6553966189856957, "grad_norm": 0.9997625900651718, "learning_rate": 5.607993189743401e-06, "loss": 0.3891, "step": 18396 }, { "epoch": 0.6554322461121899, "grad_norm": 1.032336947752363, "learning_rate": 5.606956534974594e-06, "loss": 0.4233, "step": 18397 }, { "epoch": 0.6554678732386839, "grad_norm": 1.1658007492683626, "learning_rate": 5.605919938702235e-06, "loss": 0.4544, "step": 18398 }, { "epoch": 0.655503500365178, "grad_norm": 1.5787002369180971, "learning_rate": 5.6048834009401196e-06, "loss": 0.544, "step": 18399 }, { "epoch": 0.6555391274916722, "grad_norm": 1.9460482239079306, "learning_rate": 5.603846921702054e-06, "loss": 0.527, "step": 18400 }, { "epoch": 0.6555747546181663, "grad_norm": 1.0911296739655718, "learning_rate": 5.602810501001839e-06, "loss": 0.3978, "step": 18401 }, { "epoch": 0.6556103817446604, "grad_norm": 1.5721232408553212, "learning_rate": 5.601774138853278e-06, "loss": 0.5132, "step": 18402 }, { "epoch": 0.6556460088711545, "grad_norm": 1.4039292621253212, "learning_rate": 5.600737835270167e-06, "loss": 0.5399, "step": 18403 }, { "epoch": 0.6556816359976486, "grad_norm": 1.1325930228324772, "learning_rate": 5.599701590266309e-06, "loss": 0.3671, "step": 18404 }, { "epoch": 0.6557172631241427, "grad_norm": 1.2390942562702458, "learning_rate": 5.598665403855497e-06, "loss": 0.4564, "step": 18405 }, { "epoch": 0.6557528902506369, "grad_norm": 1.3796136772054894, "learning_rate": 5.597629276051535e-06, "loss": 0.4869, "step": 18406 }, { "epoch": 0.655788517377131, "grad_norm": 2.1199161731528844, "learning_rate": 5.596593206868213e-06, "loss": 0.5447, "step": 18407 }, { "epoch": 0.655824144503625, "grad_norm": 1.6367086972925258, "learning_rate": 5.595557196319329e-06, "loss": 0.5302, "step": 18408 }, { "epoch": 0.6558597716301192, "grad_norm": 1.0948474871808629, "learning_rate": 5.59452124441868e-06, "loss": 0.4144, "step": 18409 }, { "epoch": 0.6558953987566133, "grad_norm": 1.0747771656678882, "learning_rate": 5.593485351180059e-06, "loss": 0.4521, "step": 18410 }, { "epoch": 0.6559310258831074, "grad_norm": 1.116777080376173, "learning_rate": 5.5924495166172664e-06, "loss": 0.4272, "step": 18411 }, { "epoch": 0.6559666530096016, "grad_norm": 1.484501714800016, "learning_rate": 5.5914137407440835e-06, "loss": 0.5025, "step": 18412 }, { "epoch": 0.6560022801360956, "grad_norm": 1.6785098086077523, "learning_rate": 5.590378023574312e-06, "loss": 0.4951, "step": 18413 }, { "epoch": 0.6560379072625897, "grad_norm": 1.1395542421560414, "learning_rate": 5.5893423651217375e-06, "loss": 0.4321, "step": 18414 }, { "epoch": 0.6560735343890839, "grad_norm": 1.601226462332457, "learning_rate": 5.58830676540015e-06, "loss": 0.5227, "step": 18415 }, { "epoch": 0.656109161515578, "grad_norm": 1.1752977452803963, "learning_rate": 5.587271224423343e-06, "loss": 0.4379, "step": 18416 }, { "epoch": 0.656144788642072, "grad_norm": 1.0882174991794729, "learning_rate": 5.586235742205106e-06, "loss": 0.388, "step": 18417 }, { "epoch": 0.6561804157685662, "grad_norm": 1.083120281664764, "learning_rate": 5.585200318759223e-06, "loss": 0.3927, "step": 18418 }, { "epoch": 0.6562160428950603, "grad_norm": 1.5059736759343012, "learning_rate": 5.584164954099487e-06, "loss": 0.5579, "step": 18419 }, { "epoch": 0.6562516700215544, "grad_norm": 1.2203520970034676, "learning_rate": 5.583129648239677e-06, "loss": 0.4436, "step": 18420 }, { "epoch": 0.6562872971480486, "grad_norm": 1.347539262616235, "learning_rate": 5.582094401193583e-06, "loss": 0.5333, "step": 18421 }, { "epoch": 0.6563229242745426, "grad_norm": 1.1341405005861065, "learning_rate": 5.58105921297499e-06, "loss": 0.4238, "step": 18422 }, { "epoch": 0.6563585514010367, "grad_norm": 1.4451268458858462, "learning_rate": 5.580024083597682e-06, "loss": 0.5646, "step": 18423 }, { "epoch": 0.6563941785275309, "grad_norm": 1.147987449194554, "learning_rate": 5.578989013075446e-06, "loss": 0.4127, "step": 18424 }, { "epoch": 0.656429805654025, "grad_norm": 2.1917883396595843, "learning_rate": 5.577954001422059e-06, "loss": 0.5129, "step": 18425 }, { "epoch": 0.656465432780519, "grad_norm": 1.163788234839853, "learning_rate": 5.576919048651308e-06, "loss": 0.4221, "step": 18426 }, { "epoch": 0.6565010599070132, "grad_norm": 1.3131783628808034, "learning_rate": 5.575884154776968e-06, "loss": 0.5014, "step": 18427 }, { "epoch": 0.6565366870335073, "grad_norm": 1.1752288839196905, "learning_rate": 5.5748493198128275e-06, "loss": 0.4474, "step": 18428 }, { "epoch": 0.6565723141600014, "grad_norm": 1.2357874230427115, "learning_rate": 5.573814543772656e-06, "loss": 0.431, "step": 18429 }, { "epoch": 0.6566079412864956, "grad_norm": 1.0976007720828176, "learning_rate": 5.572779826670238e-06, "loss": 0.4219, "step": 18430 }, { "epoch": 0.6566435684129897, "grad_norm": 1.1302247002500176, "learning_rate": 5.571745168519351e-06, "loss": 0.4274, "step": 18431 }, { "epoch": 0.6566791955394837, "grad_norm": 1.1800569924316724, "learning_rate": 5.570710569333773e-06, "loss": 0.4787, "step": 18432 }, { "epoch": 0.6567148226659779, "grad_norm": 1.1454653589758808, "learning_rate": 5.569676029127282e-06, "loss": 0.4166, "step": 18433 }, { "epoch": 0.656750449792472, "grad_norm": 1.5488672696794041, "learning_rate": 5.568641547913649e-06, "loss": 0.4529, "step": 18434 }, { "epoch": 0.6567860769189661, "grad_norm": 2.195132702321659, "learning_rate": 5.5676071257066535e-06, "loss": 0.5071, "step": 18435 }, { "epoch": 0.6568217040454603, "grad_norm": 1.3978578346636783, "learning_rate": 5.566572762520064e-06, "loss": 0.5607, "step": 18436 }, { "epoch": 0.6568573311719543, "grad_norm": 1.1195540003708329, "learning_rate": 5.565538458367658e-06, "loss": 0.4539, "step": 18437 }, { "epoch": 0.6568929582984484, "grad_norm": 1.1807926083267373, "learning_rate": 5.564504213263205e-06, "loss": 0.4509, "step": 18438 }, { "epoch": 0.6569285854249426, "grad_norm": 1.2770620010778742, "learning_rate": 5.5634700272204835e-06, "loss": 0.4349, "step": 18439 }, { "epoch": 0.6569642125514367, "grad_norm": 1.79038180970163, "learning_rate": 5.562435900253256e-06, "loss": 0.5597, "step": 18440 }, { "epoch": 0.6569998396779307, "grad_norm": 1.226283018188697, "learning_rate": 5.5614018323753e-06, "loss": 0.4639, "step": 18441 }, { "epoch": 0.6570354668044249, "grad_norm": 1.4360865681310784, "learning_rate": 5.560367823600377e-06, "loss": 0.5839, "step": 18442 }, { "epoch": 0.657071093930919, "grad_norm": 1.1465963685614429, "learning_rate": 5.559333873942259e-06, "loss": 0.4206, "step": 18443 }, { "epoch": 0.6571067210574131, "grad_norm": 1.1213935957123256, "learning_rate": 5.558299983414719e-06, "loss": 0.4233, "step": 18444 }, { "epoch": 0.6571423481839073, "grad_norm": 1.1057277691369127, "learning_rate": 5.557266152031515e-06, "loss": 0.441, "step": 18445 }, { "epoch": 0.6571779753104013, "grad_norm": 1.4509185800923938, "learning_rate": 5.556232379806417e-06, "loss": 0.5032, "step": 18446 }, { "epoch": 0.6572136024368954, "grad_norm": 1.118927838057726, "learning_rate": 5.555198666753192e-06, "loss": 0.4573, "step": 18447 }, { "epoch": 0.6572492295633896, "grad_norm": 1.2082245914731191, "learning_rate": 5.554165012885606e-06, "loss": 0.442, "step": 18448 }, { "epoch": 0.6572848566898837, "grad_norm": 1.0870322855654777, "learning_rate": 5.553131418217417e-06, "loss": 0.4609, "step": 18449 }, { "epoch": 0.6573204838163778, "grad_norm": 1.0702219871350815, "learning_rate": 5.552097882762395e-06, "loss": 0.4222, "step": 18450 }, { "epoch": 0.6573561109428719, "grad_norm": 1.0476805307307668, "learning_rate": 5.551064406534296e-06, "loss": 0.3778, "step": 18451 }, { "epoch": 0.657391738069366, "grad_norm": 1.6190390388598015, "learning_rate": 5.550030989546883e-06, "loss": 0.5922, "step": 18452 }, { "epoch": 0.6574273651958601, "grad_norm": 1.602603313688056, "learning_rate": 5.5489976318139175e-06, "loss": 0.5392, "step": 18453 }, { "epoch": 0.6574629923223543, "grad_norm": 1.409745365211103, "learning_rate": 5.547964333349164e-06, "loss": 0.504, "step": 18454 }, { "epoch": 0.6574986194488484, "grad_norm": 1.5948050824448943, "learning_rate": 5.546931094166372e-06, "loss": 0.5597, "step": 18455 }, { "epoch": 0.6575342465753424, "grad_norm": 1.3820599282376338, "learning_rate": 5.545897914279307e-06, "loss": 0.6214, "step": 18456 }, { "epoch": 0.6575698737018366, "grad_norm": 1.481208586940114, "learning_rate": 5.5448647937017275e-06, "loss": 0.4969, "step": 18457 }, { "epoch": 0.6576055008283307, "grad_norm": 1.6862384198802165, "learning_rate": 5.543831732447383e-06, "loss": 0.5645, "step": 18458 }, { "epoch": 0.6576411279548248, "grad_norm": 1.0462496432091162, "learning_rate": 5.542798730530038e-06, "loss": 0.4188, "step": 18459 }, { "epoch": 0.657676755081319, "grad_norm": 1.140364424134808, "learning_rate": 5.541765787963435e-06, "loss": 0.4513, "step": 18460 }, { "epoch": 0.657712382207813, "grad_norm": 1.4397340118857027, "learning_rate": 5.5407329047613455e-06, "loss": 0.5701, "step": 18461 }, { "epoch": 0.6577480093343071, "grad_norm": 1.4256560333320905, "learning_rate": 5.539700080937509e-06, "loss": 0.5154, "step": 18462 }, { "epoch": 0.6577836364608013, "grad_norm": 1.2580464873791783, "learning_rate": 5.5386673165056884e-06, "loss": 0.4313, "step": 18463 }, { "epoch": 0.6578192635872954, "grad_norm": 1.0832006656655733, "learning_rate": 5.537634611479626e-06, "loss": 0.4149, "step": 18464 }, { "epoch": 0.6578548907137894, "grad_norm": 1.0955870597045256, "learning_rate": 5.536601965873083e-06, "loss": 0.3884, "step": 18465 }, { "epoch": 0.6578905178402836, "grad_norm": 1.1766151843198973, "learning_rate": 5.535569379699799e-06, "loss": 0.43, "step": 18466 }, { "epoch": 0.6579261449667777, "grad_norm": 1.10910869170472, "learning_rate": 5.53453685297353e-06, "loss": 0.4079, "step": 18467 }, { "epoch": 0.6579617720932718, "grad_norm": 1.2500613884796807, "learning_rate": 5.533504385708024e-06, "loss": 0.4462, "step": 18468 }, { "epoch": 0.657997399219766, "grad_norm": 1.2309722615710845, "learning_rate": 5.532471977917029e-06, "loss": 0.5931, "step": 18469 }, { "epoch": 0.65803302634626, "grad_norm": 2.0842525382292094, "learning_rate": 5.531439629614296e-06, "loss": 0.5908, "step": 18470 }, { "epoch": 0.6580686534727541, "grad_norm": 1.1741083814519064, "learning_rate": 5.530407340813564e-06, "loss": 0.4436, "step": 18471 }, { "epoch": 0.6581042805992483, "grad_norm": 1.2445879384017446, "learning_rate": 5.529375111528586e-06, "loss": 0.4242, "step": 18472 }, { "epoch": 0.6581399077257424, "grad_norm": 1.4844676569390558, "learning_rate": 5.5283429417730996e-06, "loss": 0.5255, "step": 18473 }, { "epoch": 0.6581755348522365, "grad_norm": 1.3723338455176843, "learning_rate": 5.5273108315608514e-06, "loss": 0.4826, "step": 18474 }, { "epoch": 0.6582111619787306, "grad_norm": 1.0846272186779855, "learning_rate": 5.526278780905587e-06, "loss": 0.4296, "step": 18475 }, { "epoch": 0.6582467891052247, "grad_norm": 1.3634032682547783, "learning_rate": 5.5252467898210505e-06, "loss": 0.5028, "step": 18476 }, { "epoch": 0.6582824162317188, "grad_norm": 1.079654312033307, "learning_rate": 5.524214858320978e-06, "loss": 0.4044, "step": 18477 }, { "epoch": 0.658318043358213, "grad_norm": 1.0974633364930002, "learning_rate": 5.523182986419111e-06, "loss": 0.4497, "step": 18478 }, { "epoch": 0.658353670484707, "grad_norm": 1.4837450209649519, "learning_rate": 5.5221511741291975e-06, "loss": 0.499, "step": 18479 }, { "epoch": 0.6583892976112011, "grad_norm": 1.1458097805737928, "learning_rate": 5.5211194214649665e-06, "loss": 0.4308, "step": 18480 }, { "epoch": 0.6584249247376953, "grad_norm": 2.782996447834405, "learning_rate": 5.520087728440164e-06, "loss": 0.543, "step": 18481 }, { "epoch": 0.6584605518641894, "grad_norm": 1.6768654746850071, "learning_rate": 5.5190560950685215e-06, "loss": 0.5366, "step": 18482 }, { "epoch": 0.6584961789906835, "grad_norm": 1.1417452795551621, "learning_rate": 5.5180245213637785e-06, "loss": 0.3859, "step": 18483 }, { "epoch": 0.6585318061171777, "grad_norm": 1.1893789681495763, "learning_rate": 5.5169930073396725e-06, "loss": 0.4194, "step": 18484 }, { "epoch": 0.6585674332436717, "grad_norm": 1.0990935596709797, "learning_rate": 5.515961553009942e-06, "loss": 0.4129, "step": 18485 }, { "epoch": 0.6586030603701658, "grad_norm": 1.1578278991112494, "learning_rate": 5.514930158388313e-06, "loss": 0.3844, "step": 18486 }, { "epoch": 0.65863868749666, "grad_norm": 1.0920418776760594, "learning_rate": 5.513898823488528e-06, "loss": 0.4022, "step": 18487 }, { "epoch": 0.6586743146231541, "grad_norm": 1.2215232437937011, "learning_rate": 5.512867548324312e-06, "loss": 0.4337, "step": 18488 }, { "epoch": 0.6587099417496481, "grad_norm": 1.5314087983991727, "learning_rate": 5.511836332909401e-06, "loss": 0.5139, "step": 18489 }, { "epoch": 0.6587455688761423, "grad_norm": 1.0773706154540914, "learning_rate": 5.510805177257525e-06, "loss": 0.3791, "step": 18490 }, { "epoch": 0.6587811960026364, "grad_norm": 1.1732247985689324, "learning_rate": 5.509774081382417e-06, "loss": 0.4452, "step": 18491 }, { "epoch": 0.6588168231291305, "grad_norm": 1.433769294522668, "learning_rate": 5.5087430452978105e-06, "loss": 0.5089, "step": 18492 }, { "epoch": 0.6588524502556247, "grad_norm": 1.286986055623228, "learning_rate": 5.507712069017425e-06, "loss": 0.5155, "step": 18493 }, { "epoch": 0.6588880773821187, "grad_norm": 1.5164809304256235, "learning_rate": 5.506681152554996e-06, "loss": 0.5616, "step": 18494 }, { "epoch": 0.6589237045086128, "grad_norm": 1.1449708632936983, "learning_rate": 5.505650295924245e-06, "loss": 0.4475, "step": 18495 }, { "epoch": 0.658959331635107, "grad_norm": 1.5350127062428138, "learning_rate": 5.504619499138906e-06, "loss": 0.4847, "step": 18496 }, { "epoch": 0.6589949587616011, "grad_norm": 1.4260793740240065, "learning_rate": 5.5035887622126915e-06, "loss": 0.537, "step": 18497 }, { "epoch": 0.6590305858880952, "grad_norm": 1.1374789539534944, "learning_rate": 5.502558085159344e-06, "loss": 0.4474, "step": 18498 }, { "epoch": 0.6590662130145893, "grad_norm": 1.4165746288522154, "learning_rate": 5.501527467992576e-06, "loss": 0.5741, "step": 18499 }, { "epoch": 0.6591018401410834, "grad_norm": 1.996060815477523, "learning_rate": 5.5004969107261166e-06, "loss": 0.5571, "step": 18500 }, { "epoch": 0.6591374672675775, "grad_norm": 1.398990857882054, "learning_rate": 5.499466413373682e-06, "loss": 0.4076, "step": 18501 }, { "epoch": 0.6591730943940717, "grad_norm": 1.498843842091324, "learning_rate": 5.4984359759489966e-06, "loss": 0.5704, "step": 18502 }, { "epoch": 0.6592087215205658, "grad_norm": 1.1757269686682144, "learning_rate": 5.49740559846579e-06, "loss": 0.4425, "step": 18503 }, { "epoch": 0.6592443486470598, "grad_norm": 1.4190534766231782, "learning_rate": 5.496375280937768e-06, "loss": 0.5227, "step": 18504 }, { "epoch": 0.659279975773554, "grad_norm": 1.3354014386791273, "learning_rate": 5.495345023378658e-06, "loss": 0.6157, "step": 18505 }, { "epoch": 0.6593156029000481, "grad_norm": 1.5001724227358226, "learning_rate": 5.494314825802178e-06, "loss": 0.5371, "step": 18506 }, { "epoch": 0.6593512300265422, "grad_norm": 1.295183039679229, "learning_rate": 5.493284688222049e-06, "loss": 0.4197, "step": 18507 }, { "epoch": 0.6593868571530364, "grad_norm": 1.10743861289203, "learning_rate": 5.492254610651979e-06, "loss": 0.4205, "step": 18508 }, { "epoch": 0.6594224842795304, "grad_norm": 1.1656270593921298, "learning_rate": 5.491224593105695e-06, "loss": 0.4311, "step": 18509 }, { "epoch": 0.6594581114060245, "grad_norm": 1.2703747568903536, "learning_rate": 5.490194635596904e-06, "loss": 0.4076, "step": 18510 }, { "epoch": 0.6594937385325187, "grad_norm": 1.140049163551322, "learning_rate": 5.489164738139322e-06, "loss": 0.4533, "step": 18511 }, { "epoch": 0.6595293656590128, "grad_norm": 1.7457300354546736, "learning_rate": 5.488134900746665e-06, "loss": 0.4769, "step": 18512 }, { "epoch": 0.6595649927855068, "grad_norm": 1.3885030899910127, "learning_rate": 5.487105123432645e-06, "loss": 0.5004, "step": 18513 }, { "epoch": 0.659600619912001, "grad_norm": 1.1510644245786759, "learning_rate": 5.486075406210979e-06, "loss": 0.4041, "step": 18514 }, { "epoch": 0.6596362470384951, "grad_norm": 1.6767990780692354, "learning_rate": 5.48504574909537e-06, "loss": 0.4881, "step": 18515 }, { "epoch": 0.6596718741649892, "grad_norm": 1.4913454995779967, "learning_rate": 5.484016152099536e-06, "loss": 0.5044, "step": 18516 }, { "epoch": 0.6597075012914834, "grad_norm": 1.2320408148959903, "learning_rate": 5.48298661523718e-06, "loss": 0.4465, "step": 18517 }, { "epoch": 0.6597431284179774, "grad_norm": 2.111457296144642, "learning_rate": 5.481957138522018e-06, "loss": 0.4918, "step": 18518 }, { "epoch": 0.6597787555444715, "grad_norm": 1.505663157237907, "learning_rate": 5.480927721967752e-06, "loss": 0.5308, "step": 18519 }, { "epoch": 0.6598143826709657, "grad_norm": 1.4987434631950238, "learning_rate": 5.47989836558809e-06, "loss": 0.5256, "step": 18520 }, { "epoch": 0.6598500097974598, "grad_norm": 1.0803820959478927, "learning_rate": 5.4788690693967414e-06, "loss": 0.4103, "step": 18521 }, { "epoch": 0.6598856369239539, "grad_norm": 1.0295464567917483, "learning_rate": 5.477839833407416e-06, "loss": 0.4045, "step": 18522 }, { "epoch": 0.659921264050448, "grad_norm": 1.129692360891767, "learning_rate": 5.47681065763381e-06, "loss": 0.4178, "step": 18523 }, { "epoch": 0.6599568911769421, "grad_norm": 1.4361368849272473, "learning_rate": 5.475781542089631e-06, "loss": 0.3915, "step": 18524 }, { "epoch": 0.6599925183034362, "grad_norm": 1.6087702895211895, "learning_rate": 5.474752486788587e-06, "loss": 0.488, "step": 18525 }, { "epoch": 0.6600281454299304, "grad_norm": 1.1888880249506593, "learning_rate": 5.473723491744373e-06, "loss": 0.4546, "step": 18526 }, { "epoch": 0.6600637725564245, "grad_norm": 1.4442381470008367, "learning_rate": 5.472694556970693e-06, "loss": 0.5189, "step": 18527 }, { "epoch": 0.6600993996829185, "grad_norm": 1.4203769216614484, "learning_rate": 5.4716656824812505e-06, "loss": 0.4848, "step": 18528 }, { "epoch": 0.6601350268094127, "grad_norm": 1.159562249886375, "learning_rate": 5.470636868289749e-06, "loss": 0.4181, "step": 18529 }, { "epoch": 0.6601706539359068, "grad_norm": 1.6060638692893117, "learning_rate": 5.469608114409879e-06, "loss": 0.4838, "step": 18530 }, { "epoch": 0.6602062810624009, "grad_norm": 1.436771523492259, "learning_rate": 5.4685794208553465e-06, "loss": 0.5199, "step": 18531 }, { "epoch": 0.660241908188895, "grad_norm": 1.1528439845167375, "learning_rate": 5.467550787639844e-06, "loss": 0.3716, "step": 18532 }, { "epoch": 0.6602775353153891, "grad_norm": 1.0804728651489717, "learning_rate": 5.466522214777074e-06, "loss": 0.4074, "step": 18533 }, { "epoch": 0.6603131624418832, "grad_norm": 1.2974758587697512, "learning_rate": 5.465493702280725e-06, "loss": 0.5932, "step": 18534 }, { "epoch": 0.6603487895683774, "grad_norm": 1.2408662514721198, "learning_rate": 5.464465250164498e-06, "loss": 0.4326, "step": 18535 }, { "epoch": 0.6603844166948715, "grad_norm": 1.164439045317243, "learning_rate": 5.463436858442085e-06, "loss": 0.443, "step": 18536 }, { "epoch": 0.6604200438213655, "grad_norm": 1.3193457447181962, "learning_rate": 5.462408527127182e-06, "loss": 0.4294, "step": 18537 }, { "epoch": 0.6604556709478597, "grad_norm": 2.220701573089515, "learning_rate": 5.461380256233484e-06, "loss": 0.5632, "step": 18538 }, { "epoch": 0.6604912980743538, "grad_norm": 1.7293678771937557, "learning_rate": 5.4603520457746775e-06, "loss": 0.6175, "step": 18539 }, { "epoch": 0.6605269252008479, "grad_norm": 1.2227054607954677, "learning_rate": 5.459323895764459e-06, "loss": 0.4422, "step": 18540 }, { "epoch": 0.6605625523273421, "grad_norm": 2.4851768545755766, "learning_rate": 5.458295806216514e-06, "loss": 0.7141, "step": 18541 }, { "epoch": 0.6605981794538361, "grad_norm": 1.1671157053733305, "learning_rate": 5.457267777144535e-06, "loss": 0.4397, "step": 18542 }, { "epoch": 0.6606338065803302, "grad_norm": 1.4647559593035624, "learning_rate": 5.45623980856221e-06, "loss": 0.5159, "step": 18543 }, { "epoch": 0.6606694337068244, "grad_norm": 1.1039003292575131, "learning_rate": 5.455211900483232e-06, "loss": 0.4293, "step": 18544 }, { "epoch": 0.6607050608333185, "grad_norm": 1.0590763099054754, "learning_rate": 5.454184052921278e-06, "loss": 0.4072, "step": 18545 }, { "epoch": 0.6607406879598126, "grad_norm": 1.0189192978022836, "learning_rate": 5.453156265890048e-06, "loss": 0.3993, "step": 18546 }, { "epoch": 0.6607763150863067, "grad_norm": 1.3558395190528318, "learning_rate": 5.452128539403215e-06, "loss": 0.4478, "step": 18547 }, { "epoch": 0.6608119422128008, "grad_norm": 1.8533786271823172, "learning_rate": 5.451100873474468e-06, "loss": 0.5776, "step": 18548 }, { "epoch": 0.6608475693392949, "grad_norm": 1.162609459493668, "learning_rate": 5.450073268117495e-06, "loss": 0.4217, "step": 18549 }, { "epoch": 0.6608831964657891, "grad_norm": 1.3942752593397063, "learning_rate": 5.449045723345976e-06, "loss": 0.5962, "step": 18550 }, { "epoch": 0.6609188235922832, "grad_norm": 1.4170763028280107, "learning_rate": 5.448018239173597e-06, "loss": 0.4848, "step": 18551 }, { "epoch": 0.6609544507187772, "grad_norm": 1.7221602742300868, "learning_rate": 5.446990815614033e-06, "loss": 0.6517, "step": 18552 }, { "epoch": 0.6609900778452714, "grad_norm": 2.1668787964568743, "learning_rate": 5.445963452680974e-06, "loss": 0.5519, "step": 18553 }, { "epoch": 0.6610257049717655, "grad_norm": 1.4333047480703076, "learning_rate": 5.4449361503880895e-06, "loss": 0.4729, "step": 18554 }, { "epoch": 0.6610613320982596, "grad_norm": 1.5353243988091012, "learning_rate": 5.443908908749067e-06, "loss": 0.4938, "step": 18555 }, { "epoch": 0.6610969592247538, "grad_norm": 1.0522768007028132, "learning_rate": 5.442881727777579e-06, "loss": 0.4103, "step": 18556 }, { "epoch": 0.6611325863512478, "grad_norm": 1.3534890979693983, "learning_rate": 5.441854607487305e-06, "loss": 0.3935, "step": 18557 }, { "epoch": 0.6611682134777419, "grad_norm": 1.0106498176522927, "learning_rate": 5.4408275478919246e-06, "loss": 0.3515, "step": 18558 }, { "epoch": 0.6612038406042361, "grad_norm": 1.1127096399156295, "learning_rate": 5.439800549005112e-06, "loss": 0.4183, "step": 18559 }, { "epoch": 0.6612394677307302, "grad_norm": 1.302127400363151, "learning_rate": 5.438773610840546e-06, "loss": 0.4719, "step": 18560 }, { "epoch": 0.6612750948572242, "grad_norm": 1.4340697294689129, "learning_rate": 5.437746733411893e-06, "loss": 0.5434, "step": 18561 }, { "epoch": 0.6613107219837184, "grad_norm": 4.654949884211819, "learning_rate": 5.436719916732836e-06, "loss": 0.4971, "step": 18562 }, { "epoch": 0.6613463491102125, "grad_norm": 1.155440791629716, "learning_rate": 5.435693160817037e-06, "loss": 0.4151, "step": 18563 }, { "epoch": 0.6613819762367066, "grad_norm": 1.1192124767481222, "learning_rate": 5.434666465678176e-06, "loss": 0.4249, "step": 18564 }, { "epoch": 0.6614176033632008, "grad_norm": 1.197829404673679, "learning_rate": 5.43363983132992e-06, "loss": 0.4244, "step": 18565 }, { "epoch": 0.6614532304896948, "grad_norm": 1.2127375633507123, "learning_rate": 5.432613257785946e-06, "loss": 0.4499, "step": 18566 }, { "epoch": 0.6614888576161889, "grad_norm": 1.6435799781726204, "learning_rate": 5.431586745059916e-06, "loss": 0.4452, "step": 18567 }, { "epoch": 0.6615244847426831, "grad_norm": 1.3214247682443216, "learning_rate": 5.430560293165504e-06, "loss": 0.4297, "step": 18568 }, { "epoch": 0.6615601118691772, "grad_norm": 1.4819481746373622, "learning_rate": 5.429533902116373e-06, "loss": 0.4809, "step": 18569 }, { "epoch": 0.6615957389956713, "grad_norm": 1.6084218177809324, "learning_rate": 5.428507571926192e-06, "loss": 0.5677, "step": 18570 }, { "epoch": 0.6616313661221654, "grad_norm": 1.0696224192671293, "learning_rate": 5.427481302608632e-06, "loss": 0.4351, "step": 18571 }, { "epoch": 0.6616669932486595, "grad_norm": 1.0870473314294302, "learning_rate": 5.426455094177346e-06, "loss": 0.4238, "step": 18572 }, { "epoch": 0.6617026203751536, "grad_norm": 1.2042004481388926, "learning_rate": 5.425428946646016e-06, "loss": 0.412, "step": 18573 }, { "epoch": 0.6617382475016478, "grad_norm": 1.852460125683099, "learning_rate": 5.4244028600282925e-06, "loss": 0.5818, "step": 18574 }, { "epoch": 0.6617738746281419, "grad_norm": 1.5422155959301798, "learning_rate": 5.423376834337848e-06, "loss": 0.5565, "step": 18575 }, { "epoch": 0.6618095017546359, "grad_norm": 1.0839128919751488, "learning_rate": 5.422350869588334e-06, "loss": 0.3808, "step": 18576 }, { "epoch": 0.6618451288811301, "grad_norm": 1.075884982880801, "learning_rate": 5.421324965793423e-06, "loss": 0.4455, "step": 18577 }, { "epoch": 0.6618807560076242, "grad_norm": 1.126550222516111, "learning_rate": 5.420299122966766e-06, "loss": 0.4, "step": 18578 }, { "epoch": 0.6619163831341183, "grad_norm": 1.128486629009498, "learning_rate": 5.4192733411220275e-06, "loss": 0.4353, "step": 18579 }, { "epoch": 0.6619520102606125, "grad_norm": 1.6398120121160937, "learning_rate": 5.418247620272866e-06, "loss": 0.5767, "step": 18580 }, { "epoch": 0.6619876373871065, "grad_norm": 1.1859089877092306, "learning_rate": 5.4172219604329436e-06, "loss": 0.4495, "step": 18581 }, { "epoch": 0.6620232645136006, "grad_norm": 1.5073588649879306, "learning_rate": 5.4161963616159104e-06, "loss": 0.4498, "step": 18582 }, { "epoch": 0.6620588916400948, "grad_norm": 1.05417668410108, "learning_rate": 5.415170823835427e-06, "loss": 0.4143, "step": 18583 }, { "epoch": 0.6620945187665889, "grad_norm": 1.5450241539097356, "learning_rate": 5.414145347105152e-06, "loss": 0.512, "step": 18584 }, { "epoch": 0.6621301458930829, "grad_norm": 1.085440916789705, "learning_rate": 5.413119931438733e-06, "loss": 0.418, "step": 18585 }, { "epoch": 0.6621657730195771, "grad_norm": 1.073199847316896, "learning_rate": 5.412094576849829e-06, "loss": 0.4237, "step": 18586 }, { "epoch": 0.6622014001460712, "grad_norm": 1.4648701080534363, "learning_rate": 5.411069283352091e-06, "loss": 0.5322, "step": 18587 }, { "epoch": 0.6622370272725653, "grad_norm": 1.4278771665957455, "learning_rate": 5.410044050959178e-06, "loss": 0.5295, "step": 18588 }, { "epoch": 0.6622726543990595, "grad_norm": 1.3029057172577723, "learning_rate": 5.409018879684732e-06, "loss": 0.4359, "step": 18589 }, { "epoch": 0.6623082815255535, "grad_norm": 1.5226195096448953, "learning_rate": 5.407993769542412e-06, "loss": 0.4447, "step": 18590 }, { "epoch": 0.6623439086520476, "grad_norm": 1.4222981188661175, "learning_rate": 5.406968720545862e-06, "loss": 0.5315, "step": 18591 }, { "epoch": 0.6623795357785418, "grad_norm": 1.8599853695523558, "learning_rate": 5.405943732708738e-06, "loss": 0.5546, "step": 18592 }, { "epoch": 0.6624151629050359, "grad_norm": 1.155858783088429, "learning_rate": 5.404918806044679e-06, "loss": 0.4945, "step": 18593 }, { "epoch": 0.66245079003153, "grad_norm": 2.1046839432695985, "learning_rate": 5.403893940567338e-06, "loss": 0.5333, "step": 18594 }, { "epoch": 0.6624864171580241, "grad_norm": 1.5549645823829537, "learning_rate": 5.402869136290363e-06, "loss": 0.4653, "step": 18595 }, { "epoch": 0.6625220442845182, "grad_norm": 1.2131794559312568, "learning_rate": 5.401844393227398e-06, "loss": 0.462, "step": 18596 }, { "epoch": 0.6625576714110123, "grad_norm": 1.2635987120230254, "learning_rate": 5.400819711392092e-06, "loss": 0.4446, "step": 18597 }, { "epoch": 0.6625932985375065, "grad_norm": 1.1015152522473213, "learning_rate": 5.399795090798082e-06, "loss": 0.4372, "step": 18598 }, { "epoch": 0.6626289256640006, "grad_norm": 1.272718000463101, "learning_rate": 5.398770531459021e-06, "loss": 0.4062, "step": 18599 }, { "epoch": 0.6626645527904946, "grad_norm": 1.465971455585632, "learning_rate": 5.397746033388542e-06, "loss": 0.4933, "step": 18600 }, { "epoch": 0.6627001799169888, "grad_norm": 1.2565045059828654, "learning_rate": 5.396721596600292e-06, "loss": 0.3998, "step": 18601 }, { "epoch": 0.6627358070434829, "grad_norm": 1.1550003644910352, "learning_rate": 5.395697221107909e-06, "loss": 0.4104, "step": 18602 }, { "epoch": 0.662771434169977, "grad_norm": 1.7081852932209192, "learning_rate": 5.394672906925043e-06, "loss": 0.4731, "step": 18603 }, { "epoch": 0.6628070612964712, "grad_norm": 1.1638495215974711, "learning_rate": 5.39364865406532e-06, "loss": 0.4009, "step": 18604 }, { "epoch": 0.6628426884229652, "grad_norm": 1.2108422031244699, "learning_rate": 5.3926244625423855e-06, "loss": 0.4008, "step": 18605 }, { "epoch": 0.6628783155494593, "grad_norm": 1.961530359197747, "learning_rate": 5.39160033236988e-06, "loss": 0.5385, "step": 18606 }, { "epoch": 0.6629139426759535, "grad_norm": 1.6262934552901134, "learning_rate": 5.3905762635614334e-06, "loss": 0.5376, "step": 18607 }, { "epoch": 0.6629495698024476, "grad_norm": 1.0739589161405845, "learning_rate": 5.38955225613069e-06, "loss": 0.4405, "step": 18608 }, { "epoch": 0.6629851969289416, "grad_norm": 12.941572514664205, "learning_rate": 5.388528310091277e-06, "loss": 0.469, "step": 18609 }, { "epoch": 0.6630208240554358, "grad_norm": 1.1763173999118348, "learning_rate": 5.387504425456833e-06, "loss": 0.3973, "step": 18610 }, { "epoch": 0.6630564511819299, "grad_norm": 1.5296515679664053, "learning_rate": 5.386480602240991e-06, "loss": 0.4495, "step": 18611 }, { "epoch": 0.663092078308424, "grad_norm": 1.6999995855714243, "learning_rate": 5.385456840457388e-06, "loss": 0.5728, "step": 18612 }, { "epoch": 0.6631277054349182, "grad_norm": 1.1230504303259718, "learning_rate": 5.384433140119649e-06, "loss": 0.3701, "step": 18613 }, { "epoch": 0.6631633325614122, "grad_norm": 1.4194833452211144, "learning_rate": 5.3834095012414125e-06, "loss": 0.5569, "step": 18614 }, { "epoch": 0.6631989596879063, "grad_norm": 1.1131328627262875, "learning_rate": 5.382385923836302e-06, "loss": 0.4108, "step": 18615 }, { "epoch": 0.6632345868144005, "grad_norm": 1.2535676106687073, "learning_rate": 5.3813624079179495e-06, "loss": 0.4358, "step": 18616 }, { "epoch": 0.6632702139408946, "grad_norm": 1.7217885346907587, "learning_rate": 5.380338953499986e-06, "loss": 0.4867, "step": 18617 }, { "epoch": 0.6633058410673887, "grad_norm": 1.1400358967588016, "learning_rate": 5.379315560596038e-06, "loss": 0.4092, "step": 18618 }, { "epoch": 0.6633414681938828, "grad_norm": 1.118080722281817, "learning_rate": 5.378292229219736e-06, "loss": 0.4072, "step": 18619 }, { "epoch": 0.6633770953203769, "grad_norm": 1.0538583590788086, "learning_rate": 5.3772689593847e-06, "loss": 0.3703, "step": 18620 }, { "epoch": 0.663412722446871, "grad_norm": 1.1408647532224874, "learning_rate": 5.376245751104563e-06, "loss": 0.3847, "step": 18621 }, { "epoch": 0.6634483495733652, "grad_norm": 1.2527882339212484, "learning_rate": 5.375222604392942e-06, "loss": 0.3985, "step": 18622 }, { "epoch": 0.6634839766998593, "grad_norm": 1.5111782435593333, "learning_rate": 5.374199519263468e-06, "loss": 0.5747, "step": 18623 }, { "epoch": 0.6635196038263533, "grad_norm": 1.5626114429423672, "learning_rate": 5.373176495729752e-06, "loss": 0.631, "step": 18624 }, { "epoch": 0.6635552309528475, "grad_norm": 1.0949244370070148, "learning_rate": 5.372153533805433e-06, "loss": 0.4204, "step": 18625 }, { "epoch": 0.6635908580793416, "grad_norm": 1.200931673311803, "learning_rate": 5.3711306335041224e-06, "loss": 0.4086, "step": 18626 }, { "epoch": 0.6636264852058357, "grad_norm": 1.1885238578152546, "learning_rate": 5.370107794839445e-06, "loss": 0.4119, "step": 18627 }, { "epoch": 0.6636621123323299, "grad_norm": 1.1563313104928896, "learning_rate": 5.369085017825015e-06, "loss": 0.4531, "step": 18628 }, { "epoch": 0.6636977394588239, "grad_norm": 1.5188429193517916, "learning_rate": 5.368062302474455e-06, "loss": 0.556, "step": 18629 }, { "epoch": 0.663733366585318, "grad_norm": 1.455638444948066, "learning_rate": 5.367039648801386e-06, "loss": 0.4814, "step": 18630 }, { "epoch": 0.6637689937118122, "grad_norm": 1.218297940316087, "learning_rate": 5.366017056819419e-06, "loss": 0.4111, "step": 18631 }, { "epoch": 0.6638046208383063, "grad_norm": 1.3845897764192123, "learning_rate": 5.364994526542173e-06, "loss": 0.579, "step": 18632 }, { "epoch": 0.6638402479648003, "grad_norm": 2.4509485235383672, "learning_rate": 5.363972057983266e-06, "loss": 0.4636, "step": 18633 }, { "epoch": 0.6638758750912945, "grad_norm": 1.1138045613776746, "learning_rate": 5.3629496511563154e-06, "loss": 0.3888, "step": 18634 }, { "epoch": 0.6639115022177886, "grad_norm": 1.5719103945021646, "learning_rate": 5.361927306074925e-06, "loss": 0.4473, "step": 18635 }, { "epoch": 0.6639471293442827, "grad_norm": 1.330610026189833, "learning_rate": 5.3609050227527205e-06, "loss": 0.6152, "step": 18636 }, { "epoch": 0.6639827564707769, "grad_norm": 1.1575283919695982, "learning_rate": 5.359882801203305e-06, "loss": 0.4425, "step": 18637 }, { "epoch": 0.664018383597271, "grad_norm": 1.114732481274064, "learning_rate": 5.3588606414402914e-06, "loss": 0.4421, "step": 18638 }, { "epoch": 0.664054010723765, "grad_norm": 1.1916129886459657, "learning_rate": 5.3578385434772916e-06, "loss": 0.4771, "step": 18639 }, { "epoch": 0.6640896378502592, "grad_norm": 1.529644970918208, "learning_rate": 5.356816507327918e-06, "loss": 0.5697, "step": 18640 }, { "epoch": 0.6641252649767533, "grad_norm": 1.1145227093219687, "learning_rate": 5.355794533005782e-06, "loss": 0.4419, "step": 18641 }, { "epoch": 0.6641608921032474, "grad_norm": 1.478334352520724, "learning_rate": 5.354772620524482e-06, "loss": 0.5328, "step": 18642 }, { "epoch": 0.6641965192297415, "grad_norm": 1.588545239631025, "learning_rate": 5.353750769897636e-06, "loss": 0.5769, "step": 18643 }, { "epoch": 0.6642321463562356, "grad_norm": 1.9456387618677675, "learning_rate": 5.3527289811388435e-06, "loss": 0.4487, "step": 18644 }, { "epoch": 0.6642677734827297, "grad_norm": 0.9494857426157929, "learning_rate": 5.3517072542617164e-06, "loss": 0.3747, "step": 18645 }, { "epoch": 0.6643034006092239, "grad_norm": 1.421493326834453, "learning_rate": 5.350685589279852e-06, "loss": 0.5047, "step": 18646 }, { "epoch": 0.664339027735718, "grad_norm": 1.6243633956906114, "learning_rate": 5.349663986206859e-06, "loss": 0.4845, "step": 18647 }, { "epoch": 0.664374654862212, "grad_norm": 1.176037141490287, "learning_rate": 5.34864244505634e-06, "loss": 0.3877, "step": 18648 }, { "epoch": 0.6644102819887062, "grad_norm": 1.2065576403127949, "learning_rate": 5.3476209658419024e-06, "loss": 0.4298, "step": 18649 }, { "epoch": 0.6644459091152003, "grad_norm": 1.503258511795023, "learning_rate": 5.346599548577138e-06, "loss": 0.5942, "step": 18650 }, { "epoch": 0.6644815362416945, "grad_norm": 1.5373681619302142, "learning_rate": 5.345578193275654e-06, "loss": 0.4936, "step": 18651 }, { "epoch": 0.6645171633681886, "grad_norm": 1.1426372507079658, "learning_rate": 5.344556899951055e-06, "loss": 0.4081, "step": 18652 }, { "epoch": 0.6645527904946826, "grad_norm": 1.4522497871603004, "learning_rate": 5.343535668616928e-06, "loss": 0.4777, "step": 18653 }, { "epoch": 0.6645884176211768, "grad_norm": 1.3982095677319302, "learning_rate": 5.34251449928688e-06, "loss": 0.5405, "step": 18654 }, { "epoch": 0.6646240447476709, "grad_norm": 1.5688947025081885, "learning_rate": 5.341493391974507e-06, "loss": 0.5294, "step": 18655 }, { "epoch": 0.664659671874165, "grad_norm": 1.2702991492009166, "learning_rate": 5.340472346693409e-06, "loss": 0.4406, "step": 18656 }, { "epoch": 0.6646952990006592, "grad_norm": 1.1514761038493182, "learning_rate": 5.339451363457175e-06, "loss": 0.4118, "step": 18657 }, { "epoch": 0.6647309261271532, "grad_norm": 1.2260284411783564, "learning_rate": 5.338430442279408e-06, "loss": 0.3959, "step": 18658 }, { "epoch": 0.6647665532536473, "grad_norm": 1.4224057994597676, "learning_rate": 5.337409583173693e-06, "loss": 0.499, "step": 18659 }, { "epoch": 0.6648021803801415, "grad_norm": 1.4492135661782501, "learning_rate": 5.336388786153632e-06, "loss": 0.5576, "step": 18660 }, { "epoch": 0.6648378075066356, "grad_norm": 1.074136094890462, "learning_rate": 5.335368051232808e-06, "loss": 0.4041, "step": 18661 }, { "epoch": 0.6648734346331296, "grad_norm": 1.917331547484437, "learning_rate": 5.334347378424825e-06, "loss": 0.571, "step": 18662 }, { "epoch": 0.6649090617596238, "grad_norm": 1.4086290785273774, "learning_rate": 5.333326767743263e-06, "loss": 0.4996, "step": 18663 }, { "epoch": 0.6649446888861179, "grad_norm": 1.157574802307634, "learning_rate": 5.332306219201718e-06, "loss": 0.4394, "step": 18664 }, { "epoch": 0.664980316012612, "grad_norm": 1.6064260068950231, "learning_rate": 5.331285732813782e-06, "loss": 0.5006, "step": 18665 }, { "epoch": 0.6650159431391062, "grad_norm": 1.0926122918086971, "learning_rate": 5.330265308593037e-06, "loss": 0.4329, "step": 18666 }, { "epoch": 0.6650515702656002, "grad_norm": 1.5505961692794605, "learning_rate": 5.329244946553076e-06, "loss": 0.5641, "step": 18667 }, { "epoch": 0.6650871973920943, "grad_norm": 1.1511041517207987, "learning_rate": 5.328224646707479e-06, "loss": 0.4467, "step": 18668 }, { "epoch": 0.6651228245185885, "grad_norm": 1.0688205487570388, "learning_rate": 5.327204409069837e-06, "loss": 0.383, "step": 18669 }, { "epoch": 0.6651584516450826, "grad_norm": 1.315327069024425, "learning_rate": 5.326184233653735e-06, "loss": 0.486, "step": 18670 }, { "epoch": 0.6651940787715767, "grad_norm": 1.185158609355432, "learning_rate": 5.325164120472759e-06, "loss": 0.3996, "step": 18671 }, { "epoch": 0.6652297058980708, "grad_norm": 1.1877890800954156, "learning_rate": 5.324144069540487e-06, "loss": 0.3751, "step": 18672 }, { "epoch": 0.6652653330245649, "grad_norm": 2.002118482594399, "learning_rate": 5.32312408087051e-06, "loss": 0.5097, "step": 18673 }, { "epoch": 0.665300960151059, "grad_norm": 1.0294188288584118, "learning_rate": 5.3221041544764e-06, "loss": 0.436, "step": 18674 }, { "epoch": 0.6653365872775532, "grad_norm": 1.4915275104628083, "learning_rate": 5.321084290371742e-06, "loss": 0.5495, "step": 18675 }, { "epoch": 0.6653722144040473, "grad_norm": 1.7678544450559794, "learning_rate": 5.3200644885701184e-06, "loss": 0.5103, "step": 18676 }, { "epoch": 0.6654078415305413, "grad_norm": 1.492055943725467, "learning_rate": 5.319044749085105e-06, "loss": 0.5724, "step": 18677 }, { "epoch": 0.6654434686570355, "grad_norm": 1.146281299976021, "learning_rate": 5.318025071930288e-06, "loss": 0.4298, "step": 18678 }, { "epoch": 0.6654790957835296, "grad_norm": 1.2984627147265821, "learning_rate": 5.317005457119237e-06, "loss": 0.4665, "step": 18679 }, { "epoch": 0.6655147229100237, "grad_norm": 1.8790512577891423, "learning_rate": 5.315985904665533e-06, "loss": 0.5567, "step": 18680 }, { "epoch": 0.6655503500365179, "grad_norm": 1.4517147344221024, "learning_rate": 5.31496641458275e-06, "loss": 0.5445, "step": 18681 }, { "epoch": 0.6655859771630119, "grad_norm": 1.6232053398427726, "learning_rate": 5.313946986884464e-06, "loss": 0.568, "step": 18682 }, { "epoch": 0.665621604289506, "grad_norm": 1.40355224952708, "learning_rate": 5.3129276215842475e-06, "loss": 0.5384, "step": 18683 }, { "epoch": 0.6656572314160002, "grad_norm": 1.512401016839652, "learning_rate": 5.311908318695676e-06, "loss": 0.5552, "step": 18684 }, { "epoch": 0.6656928585424943, "grad_norm": 1.2176462755716253, "learning_rate": 5.310889078232321e-06, "loss": 0.4105, "step": 18685 }, { "epoch": 0.6657284856689883, "grad_norm": 1.5401838347942234, "learning_rate": 5.3098699002077554e-06, "loss": 0.5032, "step": 18686 }, { "epoch": 0.6657641127954825, "grad_norm": 2.092181859846191, "learning_rate": 5.308850784635555e-06, "loss": 0.552, "step": 18687 }, { "epoch": 0.6657997399219766, "grad_norm": 1.463255209750015, "learning_rate": 5.307831731529281e-06, "loss": 0.4628, "step": 18688 }, { "epoch": 0.6658353670484707, "grad_norm": 1.4485544750023143, "learning_rate": 5.306812740902511e-06, "loss": 0.4778, "step": 18689 }, { "epoch": 0.6658709941749649, "grad_norm": 1.1357976696535284, "learning_rate": 5.3057938127688045e-06, "loss": 0.398, "step": 18690 }, { "epoch": 0.665906621301459, "grad_norm": 1.3891351821814166, "learning_rate": 5.304774947141737e-06, "loss": 0.6058, "step": 18691 }, { "epoch": 0.665942248427953, "grad_norm": 1.4976167147982105, "learning_rate": 5.303756144034871e-06, "loss": 0.5144, "step": 18692 }, { "epoch": 0.6659778755544472, "grad_norm": 1.073098397740706, "learning_rate": 5.302737403461778e-06, "loss": 0.4496, "step": 18693 }, { "epoch": 0.6660135026809413, "grad_norm": 1.1327568095341107, "learning_rate": 5.301718725436018e-06, "loss": 0.4326, "step": 18694 }, { "epoch": 0.6660491298074354, "grad_norm": 1.6953083566584684, "learning_rate": 5.30070010997116e-06, "loss": 0.4743, "step": 18695 }, { "epoch": 0.6660847569339295, "grad_norm": 1.7386790865283912, "learning_rate": 5.299681557080759e-06, "loss": 0.5153, "step": 18696 }, { "epoch": 0.6661203840604236, "grad_norm": 1.0997690902195916, "learning_rate": 5.2986630667783846e-06, "loss": 0.4452, "step": 18697 }, { "epoch": 0.6661560111869177, "grad_norm": 1.1662332093090362, "learning_rate": 5.2976446390776015e-06, "loss": 0.4458, "step": 18698 }, { "epoch": 0.6661916383134119, "grad_norm": 1.3179414405143715, "learning_rate": 5.29662627399196e-06, "loss": 0.4376, "step": 18699 }, { "epoch": 0.666227265439906, "grad_norm": 1.1166722210681885, "learning_rate": 5.2956079715350326e-06, "loss": 0.4555, "step": 18700 }, { "epoch": 0.6662628925664, "grad_norm": 1.493001560323673, "learning_rate": 5.294589731720372e-06, "loss": 0.6117, "step": 18701 }, { "epoch": 0.6662985196928942, "grad_norm": 1.3337584532542874, "learning_rate": 5.293571554561541e-06, "loss": 0.5271, "step": 18702 }, { "epoch": 0.6663341468193883, "grad_norm": 1.1156321965326317, "learning_rate": 5.292553440072089e-06, "loss": 0.4325, "step": 18703 }, { "epoch": 0.6663697739458824, "grad_norm": 1.5070601443168827, "learning_rate": 5.291535388265583e-06, "loss": 0.4432, "step": 18704 }, { "epoch": 0.6664054010723766, "grad_norm": 1.4700859687802352, "learning_rate": 5.29051739915557e-06, "loss": 0.5127, "step": 18705 }, { "epoch": 0.6664410281988706, "grad_norm": 1.5285528274138795, "learning_rate": 5.289499472755611e-06, "loss": 0.5895, "step": 18706 }, { "epoch": 0.6664766553253647, "grad_norm": 1.511097089740816, "learning_rate": 5.288481609079259e-06, "loss": 0.4751, "step": 18707 }, { "epoch": 0.6665122824518589, "grad_norm": 1.3820856249474076, "learning_rate": 5.287463808140069e-06, "loss": 0.5116, "step": 18708 }, { "epoch": 0.666547909578353, "grad_norm": 1.6426650177380275, "learning_rate": 5.286446069951591e-06, "loss": 0.5842, "step": 18709 }, { "epoch": 0.666583536704847, "grad_norm": 1.5508773982710864, "learning_rate": 5.285428394527377e-06, "loss": 0.5596, "step": 18710 }, { "epoch": 0.6666191638313412, "grad_norm": 1.474122506640511, "learning_rate": 5.2844107818809835e-06, "loss": 0.5331, "step": 18711 }, { "epoch": 0.6666547909578353, "grad_norm": 1.059579205979958, "learning_rate": 5.2833932320259515e-06, "loss": 0.4082, "step": 18712 }, { "epoch": 0.6666904180843294, "grad_norm": 1.540974153416124, "learning_rate": 5.282375744975836e-06, "loss": 0.6258, "step": 18713 }, { "epoch": 0.6667260452108236, "grad_norm": 1.0995210037585486, "learning_rate": 5.281358320744186e-06, "loss": 0.4209, "step": 18714 }, { "epoch": 0.6667616723373176, "grad_norm": 1.224788689110625, "learning_rate": 5.28034095934455e-06, "loss": 0.4134, "step": 18715 }, { "epoch": 0.6667972994638117, "grad_norm": 1.6397388467439216, "learning_rate": 5.27932366079047e-06, "loss": 0.4709, "step": 18716 }, { "epoch": 0.6668329265903059, "grad_norm": 1.4918700138562726, "learning_rate": 5.2783064250954986e-06, "loss": 0.5782, "step": 18717 }, { "epoch": 0.6668685537168, "grad_norm": 1.3703402055831908, "learning_rate": 5.277289252273175e-06, "loss": 0.5071, "step": 18718 }, { "epoch": 0.6669041808432941, "grad_norm": 1.1114441286092511, "learning_rate": 5.276272142337048e-06, "loss": 0.4203, "step": 18719 }, { "epoch": 0.6669398079697882, "grad_norm": 1.8233987858790952, "learning_rate": 5.275255095300656e-06, "loss": 0.5682, "step": 18720 }, { "epoch": 0.6669754350962823, "grad_norm": 1.1499908020890242, "learning_rate": 5.274238111177545e-06, "loss": 0.4508, "step": 18721 }, { "epoch": 0.6670110622227764, "grad_norm": 1.75359821538901, "learning_rate": 5.273221189981257e-06, "loss": 0.499, "step": 18722 }, { "epoch": 0.6670466893492706, "grad_norm": 1.6038086465452914, "learning_rate": 5.272204331725331e-06, "loss": 0.4891, "step": 18723 }, { "epoch": 0.6670823164757647, "grad_norm": 1.720816199332754, "learning_rate": 5.271187536423314e-06, "loss": 0.6007, "step": 18724 }, { "epoch": 0.6671179436022587, "grad_norm": 1.2239959441850816, "learning_rate": 5.270170804088736e-06, "loss": 0.468, "step": 18725 }, { "epoch": 0.6671535707287529, "grad_norm": 0.9811033045278781, "learning_rate": 5.2691541347351434e-06, "loss": 0.3791, "step": 18726 }, { "epoch": 0.667189197855247, "grad_norm": 1.5858970542065929, "learning_rate": 5.2681375283760676e-06, "loss": 0.461, "step": 18727 }, { "epoch": 0.6672248249817411, "grad_norm": 1.220606258090619, "learning_rate": 5.267120985025047e-06, "loss": 0.4333, "step": 18728 }, { "epoch": 0.6672604521082353, "grad_norm": 1.479764445838071, "learning_rate": 5.266104504695617e-06, "loss": 0.5429, "step": 18729 }, { "epoch": 0.6672960792347293, "grad_norm": 1.1122299837693943, "learning_rate": 5.26508808740132e-06, "loss": 0.4129, "step": 18730 }, { "epoch": 0.6673317063612234, "grad_norm": 1.1281961439507229, "learning_rate": 5.264071733155681e-06, "loss": 0.4623, "step": 18731 }, { "epoch": 0.6673673334877176, "grad_norm": 1.200690410479779, "learning_rate": 5.263055441972237e-06, "loss": 0.4169, "step": 18732 }, { "epoch": 0.6674029606142117, "grad_norm": 1.1569834112814774, "learning_rate": 5.262039213864524e-06, "loss": 0.4136, "step": 18733 }, { "epoch": 0.6674385877407057, "grad_norm": 1.1955115852530391, "learning_rate": 5.261023048846066e-06, "loss": 0.4219, "step": 18734 }, { "epoch": 0.6674742148671999, "grad_norm": 1.6473336727711634, "learning_rate": 5.260006946930404e-06, "loss": 0.526, "step": 18735 }, { "epoch": 0.667509841993694, "grad_norm": 1.115260516503304, "learning_rate": 5.258990908131058e-06, "loss": 0.3862, "step": 18736 }, { "epoch": 0.6675454691201881, "grad_norm": 1.1294201760396716, "learning_rate": 5.2579749324615605e-06, "loss": 0.4029, "step": 18737 }, { "epoch": 0.6675810962466823, "grad_norm": 1.2733241608397585, "learning_rate": 5.256959019935444e-06, "loss": 0.4074, "step": 18738 }, { "epoch": 0.6676167233731763, "grad_norm": 1.4526235022111256, "learning_rate": 5.255943170566234e-06, "loss": 0.5404, "step": 18739 }, { "epoch": 0.6676523504996704, "grad_norm": 1.7879173981799223, "learning_rate": 5.2549273843674544e-06, "loss": 0.5005, "step": 18740 }, { "epoch": 0.6676879776261646, "grad_norm": 1.3902186586252292, "learning_rate": 5.253911661352637e-06, "loss": 0.4861, "step": 18741 }, { "epoch": 0.6677236047526587, "grad_norm": 1.5011917383463709, "learning_rate": 5.252896001535299e-06, "loss": 0.5569, "step": 18742 }, { "epoch": 0.6677592318791528, "grad_norm": 1.4972050045903456, "learning_rate": 5.251880404928971e-06, "loss": 0.5177, "step": 18743 }, { "epoch": 0.667794859005647, "grad_norm": 1.6307370907507777, "learning_rate": 5.250864871547171e-06, "loss": 0.5503, "step": 18744 }, { "epoch": 0.667830486132141, "grad_norm": 1.1875935154664539, "learning_rate": 5.249849401403427e-06, "loss": 0.4407, "step": 18745 }, { "epoch": 0.6678661132586351, "grad_norm": 1.2394405607947765, "learning_rate": 5.24883399451126e-06, "loss": 0.4153, "step": 18746 }, { "epoch": 0.6679017403851293, "grad_norm": 1.4507141766345253, "learning_rate": 5.247818650884187e-06, "loss": 0.527, "step": 18747 }, { "epoch": 0.6679373675116234, "grad_norm": 1.1495643928948505, "learning_rate": 5.2468033705357336e-06, "loss": 0.4621, "step": 18748 }, { "epoch": 0.6679729946381174, "grad_norm": 1.5004257739336204, "learning_rate": 5.245788153479412e-06, "loss": 0.5216, "step": 18749 }, { "epoch": 0.6680086217646116, "grad_norm": 1.5492693222331024, "learning_rate": 5.244772999728743e-06, "loss": 0.4955, "step": 18750 }, { "epoch": 0.6680442488911057, "grad_norm": 1.1906111803336112, "learning_rate": 5.243757909297247e-06, "loss": 0.4481, "step": 18751 }, { "epoch": 0.6680798760175998, "grad_norm": 1.3414899086262346, "learning_rate": 5.242742882198441e-06, "loss": 0.5152, "step": 18752 }, { "epoch": 0.668115503144094, "grad_norm": 1.8304811873465332, "learning_rate": 5.241727918445836e-06, "loss": 0.4245, "step": 18753 }, { "epoch": 0.668151130270588, "grad_norm": 1.139410311934037, "learning_rate": 5.240713018052953e-06, "loss": 0.4327, "step": 18754 }, { "epoch": 0.6681867573970821, "grad_norm": 1.0586918447439582, "learning_rate": 5.2396981810332995e-06, "loss": 0.3933, "step": 18755 }, { "epoch": 0.6682223845235763, "grad_norm": 1.2806400667468958, "learning_rate": 5.238683407400391e-06, "loss": 0.435, "step": 18756 }, { "epoch": 0.6682580116500704, "grad_norm": 1.1270321329694342, "learning_rate": 5.237668697167746e-06, "loss": 0.4373, "step": 18757 }, { "epoch": 0.6682936387765644, "grad_norm": 1.3119378827414288, "learning_rate": 5.236654050348866e-06, "loss": 0.4597, "step": 18758 }, { "epoch": 0.6683292659030586, "grad_norm": 1.5217052525581578, "learning_rate": 5.235639466957266e-06, "loss": 0.5581, "step": 18759 }, { "epoch": 0.6683648930295527, "grad_norm": 2.0916375411703627, "learning_rate": 5.234624947006458e-06, "loss": 0.5264, "step": 18760 }, { "epoch": 0.6684005201560468, "grad_norm": 1.5273941832591078, "learning_rate": 5.233610490509953e-06, "loss": 0.6088, "step": 18761 }, { "epoch": 0.668436147282541, "grad_norm": 1.2926833068275667, "learning_rate": 5.232596097481252e-06, "loss": 0.4008, "step": 18762 }, { "epoch": 0.668471774409035, "grad_norm": 1.413852102962159, "learning_rate": 5.2315817679338695e-06, "loss": 0.5173, "step": 18763 }, { "epoch": 0.6685074015355291, "grad_norm": 1.7083059970853094, "learning_rate": 5.230567501881305e-06, "loss": 0.5619, "step": 18764 }, { "epoch": 0.6685430286620233, "grad_norm": 1.6117087200472773, "learning_rate": 5.229553299337069e-06, "loss": 0.5315, "step": 18765 }, { "epoch": 0.6685786557885174, "grad_norm": 2.4992937212045088, "learning_rate": 5.228539160314665e-06, "loss": 0.5927, "step": 18766 }, { "epoch": 0.6686142829150115, "grad_norm": 1.3008998576617044, "learning_rate": 5.227525084827596e-06, "loss": 0.405, "step": 18767 }, { "epoch": 0.6686499100415056, "grad_norm": 1.4874214722848813, "learning_rate": 5.226511072889371e-06, "loss": 0.484, "step": 18768 }, { "epoch": 0.6686855371679997, "grad_norm": 1.1994174843078493, "learning_rate": 5.225497124513484e-06, "loss": 0.393, "step": 18769 }, { "epoch": 0.6687211642944938, "grad_norm": 1.4257648774849514, "learning_rate": 5.224483239713443e-06, "loss": 0.4692, "step": 18770 }, { "epoch": 0.668756791420988, "grad_norm": 1.4402705667285431, "learning_rate": 5.22346941850274e-06, "loss": 0.5271, "step": 18771 }, { "epoch": 0.6687924185474821, "grad_norm": 1.1648884172892011, "learning_rate": 5.222455660894887e-06, "loss": 0.4371, "step": 18772 }, { "epoch": 0.6688280456739761, "grad_norm": 1.10456489071922, "learning_rate": 5.221441966903371e-06, "loss": 0.3787, "step": 18773 }, { "epoch": 0.6688636728004703, "grad_norm": 1.76105586561224, "learning_rate": 5.220428336541695e-06, "loss": 0.5077, "step": 18774 }, { "epoch": 0.6688992999269644, "grad_norm": 1.8360313769244143, "learning_rate": 5.219414769823357e-06, "loss": 0.5434, "step": 18775 }, { "epoch": 0.6689349270534585, "grad_norm": 1.1114667389818804, "learning_rate": 5.218401266761857e-06, "loss": 0.4012, "step": 18776 }, { "epoch": 0.6689705541799527, "grad_norm": 1.3883658090824413, "learning_rate": 5.217387827370681e-06, "loss": 0.6105, "step": 18777 }, { "epoch": 0.6690061813064467, "grad_norm": 1.5004866617782913, "learning_rate": 5.2163744516633305e-06, "loss": 0.578, "step": 18778 }, { "epoch": 0.6690418084329408, "grad_norm": 2.124188849939254, "learning_rate": 5.215361139653301e-06, "loss": 0.534, "step": 18779 }, { "epoch": 0.669077435559435, "grad_norm": 1.3268441038006533, "learning_rate": 5.214347891354079e-06, "loss": 0.5492, "step": 18780 }, { "epoch": 0.6691130626859291, "grad_norm": 1.359033956177898, "learning_rate": 5.2133347067791585e-06, "loss": 0.5148, "step": 18781 }, { "epoch": 0.6691486898124231, "grad_norm": 2.0417052106640234, "learning_rate": 5.212321585942033e-06, "loss": 0.55, "step": 18782 }, { "epoch": 0.6691843169389173, "grad_norm": 1.2301761044596946, "learning_rate": 5.211308528856196e-06, "loss": 0.4319, "step": 18783 }, { "epoch": 0.6692199440654114, "grad_norm": 1.4512692535480591, "learning_rate": 5.210295535535128e-06, "loss": 0.5171, "step": 18784 }, { "epoch": 0.6692555711919055, "grad_norm": 1.4443066881745474, "learning_rate": 5.209282605992328e-06, "loss": 0.4764, "step": 18785 }, { "epoch": 0.6692911983183997, "grad_norm": 1.631796675717119, "learning_rate": 5.208269740241273e-06, "loss": 0.581, "step": 18786 }, { "epoch": 0.6693268254448937, "grad_norm": 1.3483558683397383, "learning_rate": 5.207256938295461e-06, "loss": 0.5105, "step": 18787 }, { "epoch": 0.6693624525713878, "grad_norm": 1.2228557258899526, "learning_rate": 5.206244200168364e-06, "loss": 0.4275, "step": 18788 }, { "epoch": 0.669398079697882, "grad_norm": 1.8291215761682729, "learning_rate": 5.2052315258734846e-06, "loss": 0.5417, "step": 18789 }, { "epoch": 0.6694337068243761, "grad_norm": 2.2625723446780497, "learning_rate": 5.204218915424295e-06, "loss": 0.5202, "step": 18790 }, { "epoch": 0.6694693339508702, "grad_norm": 1.8085328076864173, "learning_rate": 5.2032063688342815e-06, "loss": 0.4434, "step": 18791 }, { "epoch": 0.6695049610773643, "grad_norm": 1.1278478338335969, "learning_rate": 5.202193886116933e-06, "loss": 0.455, "step": 18792 }, { "epoch": 0.6695405882038584, "grad_norm": 1.126680785882958, "learning_rate": 5.201181467285723e-06, "loss": 0.4782, "step": 18793 }, { "epoch": 0.6695762153303525, "grad_norm": 1.524620194778165, "learning_rate": 5.200169112354139e-06, "loss": 0.5075, "step": 18794 }, { "epoch": 0.6696118424568467, "grad_norm": 1.0879647994201167, "learning_rate": 5.199156821335654e-06, "loss": 0.3843, "step": 18795 }, { "epoch": 0.6696474695833408, "grad_norm": 1.6534414908620654, "learning_rate": 5.1981445942437515e-06, "loss": 0.5442, "step": 18796 }, { "epoch": 0.6696830967098348, "grad_norm": 1.334884773908649, "learning_rate": 5.19713243109191e-06, "loss": 0.4308, "step": 18797 }, { "epoch": 0.669718723836329, "grad_norm": 1.2104291002556684, "learning_rate": 5.1961203318936116e-06, "loss": 0.4211, "step": 18798 }, { "epoch": 0.6697543509628231, "grad_norm": 1.2628485568915453, "learning_rate": 5.195108296662325e-06, "loss": 0.41, "step": 18799 }, { "epoch": 0.6697899780893172, "grad_norm": 1.304204116457039, "learning_rate": 5.194096325411533e-06, "loss": 0.4237, "step": 18800 }, { "epoch": 0.6698256052158114, "grad_norm": 1.151475580002683, "learning_rate": 5.193084418154703e-06, "loss": 0.4023, "step": 18801 }, { "epoch": 0.6698612323423054, "grad_norm": 1.1741129183498393, "learning_rate": 5.192072574905316e-06, "loss": 0.413, "step": 18802 }, { "epoch": 0.6698968594687995, "grad_norm": 1.4795949365652885, "learning_rate": 5.191060795676841e-06, "loss": 0.5278, "step": 18803 }, { "epoch": 0.6699324865952937, "grad_norm": 1.5839367018281132, "learning_rate": 5.190049080482754e-06, "loss": 0.5414, "step": 18804 }, { "epoch": 0.6699681137217878, "grad_norm": 1.3907810420571145, "learning_rate": 5.1890374293365295e-06, "loss": 0.4657, "step": 18805 }, { "epoch": 0.6700037408482818, "grad_norm": 1.5985704463551882, "learning_rate": 5.188025842251629e-06, "loss": 0.5729, "step": 18806 }, { "epoch": 0.670039367974776, "grad_norm": 1.1083178883877967, "learning_rate": 5.187014319241533e-06, "loss": 0.4178, "step": 18807 }, { "epoch": 0.6700749951012701, "grad_norm": 1.5226413408789166, "learning_rate": 5.186002860319701e-06, "loss": 0.5567, "step": 18808 }, { "epoch": 0.6701106222277642, "grad_norm": 1.3412593672145225, "learning_rate": 5.184991465499609e-06, "loss": 0.4467, "step": 18809 }, { "epoch": 0.6701462493542584, "grad_norm": 1.495429561439038, "learning_rate": 5.183980134794719e-06, "loss": 0.612, "step": 18810 }, { "epoch": 0.6701818764807524, "grad_norm": 2.3679989448633627, "learning_rate": 5.182968868218498e-06, "loss": 0.5084, "step": 18811 }, { "epoch": 0.6702175036072465, "grad_norm": 1.1384073163469963, "learning_rate": 5.181957665784414e-06, "loss": 0.4397, "step": 18812 }, { "epoch": 0.6702531307337407, "grad_norm": 1.1135677021649963, "learning_rate": 5.1809465275059325e-06, "loss": 0.3986, "step": 18813 }, { "epoch": 0.6702887578602348, "grad_norm": 1.5012192758994625, "learning_rate": 5.179935453396519e-06, "loss": 0.587, "step": 18814 }, { "epoch": 0.6703243849867289, "grad_norm": 1.1295523565140144, "learning_rate": 5.178924443469629e-06, "loss": 0.4106, "step": 18815 }, { "epoch": 0.670360012113223, "grad_norm": 1.354971612310844, "learning_rate": 5.177913497738736e-06, "loss": 0.5157, "step": 18816 }, { "epoch": 0.6703956392397171, "grad_norm": 1.109814116260457, "learning_rate": 5.17690261621729e-06, "loss": 0.4388, "step": 18817 }, { "epoch": 0.6704312663662112, "grad_norm": 1.2712224076820275, "learning_rate": 5.175891798918757e-06, "loss": 0.4666, "step": 18818 }, { "epoch": 0.6704668934927054, "grad_norm": 1.2240721903587424, "learning_rate": 5.174881045856597e-06, "loss": 0.4433, "step": 18819 }, { "epoch": 0.6705025206191995, "grad_norm": 1.3817212466309587, "learning_rate": 5.173870357044271e-06, "loss": 0.4543, "step": 18820 }, { "epoch": 0.6705381477456935, "grad_norm": 1.2076366847135653, "learning_rate": 5.1728597324952315e-06, "loss": 0.4417, "step": 18821 }, { "epoch": 0.6705737748721877, "grad_norm": 1.3341299142848746, "learning_rate": 5.171849172222943e-06, "loss": 0.4568, "step": 18822 }, { "epoch": 0.6706094019986818, "grad_norm": 1.212364210192817, "learning_rate": 5.170838676240853e-06, "loss": 0.4632, "step": 18823 }, { "epoch": 0.6706450291251759, "grad_norm": 1.5500588435122857, "learning_rate": 5.16982824456242e-06, "loss": 0.4584, "step": 18824 }, { "epoch": 0.6706806562516701, "grad_norm": 1.756856763868229, "learning_rate": 5.1688178772011025e-06, "loss": 0.5326, "step": 18825 }, { "epoch": 0.6707162833781641, "grad_norm": 2.0096601051919984, "learning_rate": 5.167807574170349e-06, "loss": 0.5285, "step": 18826 }, { "epoch": 0.6707519105046582, "grad_norm": 1.9722880433183765, "learning_rate": 5.166797335483622e-06, "loss": 0.5337, "step": 18827 }, { "epoch": 0.6707875376311524, "grad_norm": 1.5143477721817955, "learning_rate": 5.165787161154361e-06, "loss": 0.5602, "step": 18828 }, { "epoch": 0.6708231647576465, "grad_norm": 1.4312662821543047, "learning_rate": 5.164777051196027e-06, "loss": 0.4808, "step": 18829 }, { "epoch": 0.6708587918841405, "grad_norm": 1.582265374312741, "learning_rate": 5.163767005622063e-06, "loss": 0.4689, "step": 18830 }, { "epoch": 0.6708944190106347, "grad_norm": 1.4911889592806336, "learning_rate": 5.162757024445925e-06, "loss": 0.5524, "step": 18831 }, { "epoch": 0.6709300461371288, "grad_norm": 1.1965947067413152, "learning_rate": 5.161747107681054e-06, "loss": 0.4704, "step": 18832 }, { "epoch": 0.6709656732636229, "grad_norm": 1.1278325797027575, "learning_rate": 5.160737255340902e-06, "loss": 0.425, "step": 18833 }, { "epoch": 0.6710013003901171, "grad_norm": 1.1368302352846829, "learning_rate": 5.159727467438918e-06, "loss": 0.4257, "step": 18834 }, { "epoch": 0.6710369275166111, "grad_norm": 2.6048522621317205, "learning_rate": 5.1587177439885475e-06, "loss": 0.5578, "step": 18835 }, { "epoch": 0.6710725546431052, "grad_norm": 1.268677896609626, "learning_rate": 5.15770808500323e-06, "loss": 0.4687, "step": 18836 }, { "epoch": 0.6711081817695994, "grad_norm": 1.1213803434547092, "learning_rate": 5.156698490496414e-06, "loss": 0.425, "step": 18837 }, { "epoch": 0.6711438088960935, "grad_norm": 1.1437204218931396, "learning_rate": 5.155688960481549e-06, "loss": 0.4059, "step": 18838 }, { "epoch": 0.6711794360225876, "grad_norm": 1.3074038600802214, "learning_rate": 5.1546794949720655e-06, "loss": 0.4174, "step": 18839 }, { "epoch": 0.6712150631490817, "grad_norm": 1.5469356930590668, "learning_rate": 5.153670093981411e-06, "loss": 0.5777, "step": 18840 }, { "epoch": 0.6712506902755758, "grad_norm": 2.1522475903635385, "learning_rate": 5.152660757523028e-06, "loss": 0.5038, "step": 18841 }, { "epoch": 0.6712863174020699, "grad_norm": 1.7280660084255517, "learning_rate": 5.151651485610359e-06, "loss": 0.5004, "step": 18842 }, { "epoch": 0.6713219445285641, "grad_norm": 1.0567481393083016, "learning_rate": 5.1506422782568345e-06, "loss": 0.4138, "step": 18843 }, { "epoch": 0.6713575716550582, "grad_norm": 1.1598400883487547, "learning_rate": 5.1496331354759015e-06, "loss": 0.4177, "step": 18844 }, { "epoch": 0.6713931987815522, "grad_norm": 1.1643703464353374, "learning_rate": 5.148624057280988e-06, "loss": 0.4423, "step": 18845 }, { "epoch": 0.6714288259080464, "grad_norm": 1.2396647606290572, "learning_rate": 5.147615043685543e-06, "loss": 0.508, "step": 18846 }, { "epoch": 0.6714644530345405, "grad_norm": 1.657363662937001, "learning_rate": 5.14660609470299e-06, "loss": 0.5011, "step": 18847 }, { "epoch": 0.6715000801610346, "grad_norm": 1.1156962827349222, "learning_rate": 5.145597210346769e-06, "loss": 0.3884, "step": 18848 }, { "epoch": 0.6715357072875288, "grad_norm": 1.1046639853335471, "learning_rate": 5.144588390630315e-06, "loss": 0.4028, "step": 18849 }, { "epoch": 0.6715713344140228, "grad_norm": 1.7703498190488656, "learning_rate": 5.14357963556706e-06, "loss": 0.6075, "step": 18850 }, { "epoch": 0.6716069615405169, "grad_norm": 1.643400676457797, "learning_rate": 5.14257094517044e-06, "loss": 0.5132, "step": 18851 }, { "epoch": 0.6716425886670111, "grad_norm": 1.19962006794585, "learning_rate": 5.141562319453879e-06, "loss": 0.4075, "step": 18852 }, { "epoch": 0.6716782157935052, "grad_norm": 1.6790876589884742, "learning_rate": 5.140553758430816e-06, "loss": 0.5394, "step": 18853 }, { "epoch": 0.6717138429199992, "grad_norm": 1.1362531858892335, "learning_rate": 5.139545262114673e-06, "loss": 0.4195, "step": 18854 }, { "epoch": 0.6717494700464934, "grad_norm": 1.3974065406841991, "learning_rate": 5.138536830518881e-06, "loss": 0.4586, "step": 18855 }, { "epoch": 0.6717850971729875, "grad_norm": 1.5228961941398784, "learning_rate": 5.137528463656869e-06, "loss": 0.4951, "step": 18856 }, { "epoch": 0.6718207242994816, "grad_norm": 1.697813415803415, "learning_rate": 5.136520161542069e-06, "loss": 0.4278, "step": 18857 }, { "epoch": 0.6718563514259758, "grad_norm": 1.4006156320259944, "learning_rate": 5.135511924187897e-06, "loss": 0.4136, "step": 18858 }, { "epoch": 0.6718919785524698, "grad_norm": 1.2190121069653563, "learning_rate": 5.1345037516077846e-06, "loss": 0.4551, "step": 18859 }, { "epoch": 0.6719276056789639, "grad_norm": 1.2976327374372687, "learning_rate": 5.13349564381516e-06, "loss": 0.4117, "step": 18860 }, { "epoch": 0.6719632328054581, "grad_norm": 1.074222289003478, "learning_rate": 5.132487600823438e-06, "loss": 0.3738, "step": 18861 }, { "epoch": 0.6719988599319522, "grad_norm": 1.5395300670842067, "learning_rate": 5.13147962264605e-06, "loss": 0.4947, "step": 18862 }, { "epoch": 0.6720344870584463, "grad_norm": 1.0818531918346503, "learning_rate": 5.130471709296405e-06, "loss": 0.4046, "step": 18863 }, { "epoch": 0.6720701141849404, "grad_norm": 1.1598055072387936, "learning_rate": 5.129463860787942e-06, "loss": 0.4187, "step": 18864 }, { "epoch": 0.6721057413114345, "grad_norm": 1.0813050521993268, "learning_rate": 5.1284560771340674e-06, "loss": 0.4542, "step": 18865 }, { "epoch": 0.6721413684379286, "grad_norm": 1.3939871304971765, "learning_rate": 5.127448358348208e-06, "loss": 0.4901, "step": 18866 }, { "epoch": 0.6721769955644228, "grad_norm": 1.3737251466067684, "learning_rate": 5.126440704443777e-06, "loss": 0.4493, "step": 18867 }, { "epoch": 0.6722126226909169, "grad_norm": 1.519130792412556, "learning_rate": 5.125433115434198e-06, "loss": 0.5779, "step": 18868 }, { "epoch": 0.6722482498174109, "grad_norm": 1.1388446516896222, "learning_rate": 5.12442559133288e-06, "loss": 0.4102, "step": 18869 }, { "epoch": 0.6722838769439051, "grad_norm": 1.172941367085645, "learning_rate": 5.1234181321532436e-06, "loss": 0.4597, "step": 18870 }, { "epoch": 0.6723195040703992, "grad_norm": 1.9521100316484166, "learning_rate": 5.122410737908703e-06, "loss": 0.5395, "step": 18871 }, { "epoch": 0.6723551311968933, "grad_norm": 1.1400827392936355, "learning_rate": 5.121403408612672e-06, "loss": 0.4243, "step": 18872 }, { "epoch": 0.6723907583233875, "grad_norm": 1.6125260459838289, "learning_rate": 5.1203961442785685e-06, "loss": 0.4401, "step": 18873 }, { "epoch": 0.6724263854498815, "grad_norm": 1.362726554758801, "learning_rate": 5.119388944919798e-06, "loss": 0.5132, "step": 18874 }, { "epoch": 0.6724620125763756, "grad_norm": 2.0824416913971526, "learning_rate": 5.1183818105497785e-06, "loss": 0.5009, "step": 18875 }, { "epoch": 0.6724976397028698, "grad_norm": 1.1445940947544093, "learning_rate": 5.117374741181913e-06, "loss": 0.4204, "step": 18876 }, { "epoch": 0.6725332668293639, "grad_norm": 1.4616914439707291, "learning_rate": 5.116367736829616e-06, "loss": 0.4199, "step": 18877 }, { "epoch": 0.672568893955858, "grad_norm": 1.465736615224728, "learning_rate": 5.115360797506297e-06, "loss": 0.4789, "step": 18878 }, { "epoch": 0.6726045210823521, "grad_norm": 1.3874987715727378, "learning_rate": 5.114353923225366e-06, "loss": 0.3975, "step": 18879 }, { "epoch": 0.6726401482088462, "grad_norm": 1.3088181673243668, "learning_rate": 5.113347114000222e-06, "loss": 0.4674, "step": 18880 }, { "epoch": 0.6726757753353403, "grad_norm": 1.451107651158048, "learning_rate": 5.112340369844283e-06, "loss": 0.5658, "step": 18881 }, { "epoch": 0.6727114024618345, "grad_norm": 1.5678352821697685, "learning_rate": 5.111333690770943e-06, "loss": 0.5359, "step": 18882 }, { "epoch": 0.6727470295883285, "grad_norm": 1.8396143766076434, "learning_rate": 5.110327076793613e-06, "loss": 0.5797, "step": 18883 }, { "epoch": 0.6727826567148226, "grad_norm": 1.2963110193480651, "learning_rate": 5.1093205279256986e-06, "loss": 0.3931, "step": 18884 }, { "epoch": 0.6728182838413168, "grad_norm": 1.5766521188777884, "learning_rate": 5.1083140441805955e-06, "loss": 0.4354, "step": 18885 }, { "epoch": 0.6728539109678109, "grad_norm": 1.5939130077719803, "learning_rate": 5.107307625571711e-06, "loss": 0.5724, "step": 18886 }, { "epoch": 0.672889538094305, "grad_norm": 1.3067005680584536, "learning_rate": 5.1063012721124446e-06, "loss": 0.4097, "step": 18887 }, { "epoch": 0.6729251652207991, "grad_norm": 1.6896826578883564, "learning_rate": 5.105294983816203e-06, "loss": 0.4892, "step": 18888 }, { "epoch": 0.6729607923472932, "grad_norm": 1.5058161248367168, "learning_rate": 5.104288760696373e-06, "loss": 0.4968, "step": 18889 }, { "epoch": 0.6729964194737873, "grad_norm": 1.667604411893886, "learning_rate": 5.103282602766366e-06, "loss": 0.4283, "step": 18890 }, { "epoch": 0.6730320466002815, "grad_norm": 1.223309370999591, "learning_rate": 5.102276510039569e-06, "loss": 0.4654, "step": 18891 }, { "epoch": 0.6730676737267756, "grad_norm": 1.3296149839943305, "learning_rate": 5.101270482529384e-06, "loss": 0.5574, "step": 18892 }, { "epoch": 0.6731033008532696, "grad_norm": 1.5056489235101855, "learning_rate": 5.100264520249205e-06, "loss": 0.5513, "step": 18893 }, { "epoch": 0.6731389279797638, "grad_norm": 1.4170818774764082, "learning_rate": 5.099258623212431e-06, "loss": 0.5647, "step": 18894 }, { "epoch": 0.6731745551062579, "grad_norm": 1.552563769430714, "learning_rate": 5.098252791432456e-06, "loss": 0.4391, "step": 18895 }, { "epoch": 0.673210182232752, "grad_norm": 1.2449939528004514, "learning_rate": 5.097247024922669e-06, "loss": 0.4674, "step": 18896 }, { "epoch": 0.6732458093592462, "grad_norm": 1.6139465758566884, "learning_rate": 5.096241323696467e-06, "loss": 0.6487, "step": 18897 }, { "epoch": 0.6732814364857402, "grad_norm": 1.1942123677668552, "learning_rate": 5.095235687767236e-06, "loss": 0.4365, "step": 18898 }, { "epoch": 0.6733170636122343, "grad_norm": 1.1765399898749314, "learning_rate": 5.094230117148373e-06, "loss": 0.4227, "step": 18899 }, { "epoch": 0.6733526907387285, "grad_norm": 1.0716956654683618, "learning_rate": 5.093224611853262e-06, "loss": 0.3932, "step": 18900 }, { "epoch": 0.6733883178652226, "grad_norm": 1.162636221760218, "learning_rate": 5.092219171895296e-06, "loss": 0.3822, "step": 18901 }, { "epoch": 0.6734239449917166, "grad_norm": 1.0485161823296723, "learning_rate": 5.09121379728786e-06, "loss": 0.3674, "step": 18902 }, { "epoch": 0.6734595721182108, "grad_norm": 1.1933843840909661, "learning_rate": 5.090208488044348e-06, "loss": 0.4971, "step": 18903 }, { "epoch": 0.6734951992447049, "grad_norm": 1.0655937349623787, "learning_rate": 5.089203244178138e-06, "loss": 0.4205, "step": 18904 }, { "epoch": 0.673530826371199, "grad_norm": 1.796734541342346, "learning_rate": 5.08819806570262e-06, "loss": 0.5767, "step": 18905 }, { "epoch": 0.6735664534976932, "grad_norm": 1.261493994720186, "learning_rate": 5.087192952631181e-06, "loss": 0.4031, "step": 18906 }, { "epoch": 0.6736020806241872, "grad_norm": 1.0923686612085057, "learning_rate": 5.086187904977198e-06, "loss": 0.4067, "step": 18907 }, { "epoch": 0.6736377077506813, "grad_norm": 1.7078169805473005, "learning_rate": 5.085182922754057e-06, "loss": 0.5472, "step": 18908 }, { "epoch": 0.6736733348771755, "grad_norm": 1.309611962115729, "learning_rate": 5.084178005975141e-06, "loss": 0.4, "step": 18909 }, { "epoch": 0.6737089620036696, "grad_norm": 1.2405910851625768, "learning_rate": 5.0831731546538354e-06, "loss": 0.4448, "step": 18910 }, { "epoch": 0.6737445891301637, "grad_norm": 1.1884348404289413, "learning_rate": 5.0821683688035105e-06, "loss": 0.4317, "step": 18911 }, { "epoch": 0.6737802162566578, "grad_norm": 1.1548815465075257, "learning_rate": 5.081163648437556e-06, "loss": 0.4128, "step": 18912 }, { "epoch": 0.6738158433831519, "grad_norm": 1.5216886150175968, "learning_rate": 5.0801589935693415e-06, "loss": 0.5394, "step": 18913 }, { "epoch": 0.673851470509646, "grad_norm": 1.5338843512802969, "learning_rate": 5.0791544042122484e-06, "loss": 0.5002, "step": 18914 }, { "epoch": 0.6738870976361402, "grad_norm": 1.0932190197893403, "learning_rate": 5.078149880379655e-06, "loss": 0.4444, "step": 18915 }, { "epoch": 0.6739227247626343, "grad_norm": 1.2656794230399966, "learning_rate": 5.0771454220849395e-06, "loss": 0.5037, "step": 18916 }, { "epoch": 0.6739583518891283, "grad_norm": 1.1807016726485795, "learning_rate": 5.076141029341468e-06, "loss": 0.3989, "step": 18917 }, { "epoch": 0.6739939790156225, "grad_norm": 1.5677309659449354, "learning_rate": 5.075136702162622e-06, "loss": 0.5783, "step": 18918 }, { "epoch": 0.6740296061421166, "grad_norm": 1.604357037669029, "learning_rate": 5.074132440561778e-06, "loss": 0.5823, "step": 18919 }, { "epoch": 0.6740652332686107, "grad_norm": 1.1036975885815723, "learning_rate": 5.073128244552298e-06, "loss": 0.3814, "step": 18920 }, { "epoch": 0.6741008603951049, "grad_norm": 1.2730718417150175, "learning_rate": 5.072124114147563e-06, "loss": 0.4374, "step": 18921 }, { "epoch": 0.6741364875215989, "grad_norm": 1.5489134767062582, "learning_rate": 5.071120049360936e-06, "loss": 0.5324, "step": 18922 }, { "epoch": 0.674172114648093, "grad_norm": 2.0387276464684896, "learning_rate": 5.070116050205791e-06, "loss": 0.5456, "step": 18923 }, { "epoch": 0.6742077417745872, "grad_norm": 1.2834631944798105, "learning_rate": 5.069112116695496e-06, "loss": 0.49, "step": 18924 }, { "epoch": 0.6742433689010813, "grad_norm": 1.6949461338850313, "learning_rate": 5.068108248843424e-06, "loss": 0.5134, "step": 18925 }, { "epoch": 0.6742789960275754, "grad_norm": 1.6195652774744544, "learning_rate": 5.067104446662934e-06, "loss": 0.495, "step": 18926 }, { "epoch": 0.6743146231540695, "grad_norm": 1.0750748283692289, "learning_rate": 5.066100710167401e-06, "loss": 0.4335, "step": 18927 }, { "epoch": 0.6743502502805636, "grad_norm": 1.3368549542100696, "learning_rate": 5.0650970393701805e-06, "loss": 0.4487, "step": 18928 }, { "epoch": 0.6743858774070577, "grad_norm": 1.602866573627136, "learning_rate": 5.064093434284643e-06, "loss": 0.5567, "step": 18929 }, { "epoch": 0.6744215045335519, "grad_norm": 1.1034563677414193, "learning_rate": 5.06308989492415e-06, "loss": 0.4277, "step": 18930 }, { "epoch": 0.674457131660046, "grad_norm": 1.3282843541087794, "learning_rate": 5.062086421302066e-06, "loss": 0.4045, "step": 18931 }, { "epoch": 0.67449275878654, "grad_norm": 1.217322458072175, "learning_rate": 5.061083013431758e-06, "loss": 0.5072, "step": 18932 }, { "epoch": 0.6745283859130342, "grad_norm": 1.2551326987388691, "learning_rate": 5.060079671326577e-06, "loss": 0.4521, "step": 18933 }, { "epoch": 0.6745640130395283, "grad_norm": 1.211583998580605, "learning_rate": 5.059076394999891e-06, "loss": 0.3998, "step": 18934 }, { "epoch": 0.6745996401660224, "grad_norm": 1.096325218811434, "learning_rate": 5.058073184465054e-06, "loss": 0.3673, "step": 18935 }, { "epoch": 0.6746352672925165, "grad_norm": 1.7940767213670255, "learning_rate": 5.05707003973543e-06, "loss": 0.5532, "step": 18936 }, { "epoch": 0.6746708944190106, "grad_norm": 1.1472957761717588, "learning_rate": 5.05606696082437e-06, "loss": 0.475, "step": 18937 }, { "epoch": 0.6747065215455047, "grad_norm": 2.0907276729382644, "learning_rate": 5.055063947745234e-06, "loss": 0.5761, "step": 18938 }, { "epoch": 0.6747421486719989, "grad_norm": 1.3106787317844035, "learning_rate": 5.054061000511376e-06, "loss": 0.487, "step": 18939 }, { "epoch": 0.674777775798493, "grad_norm": 1.1056805521623747, "learning_rate": 5.053058119136154e-06, "loss": 0.4082, "step": 18940 }, { "epoch": 0.674813402924987, "grad_norm": 1.0954143858831333, "learning_rate": 5.0520553036329255e-06, "loss": 0.4186, "step": 18941 }, { "epoch": 0.6748490300514812, "grad_norm": 1.4018228615919295, "learning_rate": 5.051052554015034e-06, "loss": 0.5095, "step": 18942 }, { "epoch": 0.6748846571779753, "grad_norm": 1.1606689234514778, "learning_rate": 5.050049870295841e-06, "loss": 0.4383, "step": 18943 }, { "epoch": 0.6749202843044694, "grad_norm": 1.2823485734088438, "learning_rate": 5.04904725248869e-06, "loss": 0.4168, "step": 18944 }, { "epoch": 0.6749559114309636, "grad_norm": 2.0951760734100002, "learning_rate": 5.048044700606935e-06, "loss": 0.5296, "step": 18945 }, { "epoch": 0.6749915385574576, "grad_norm": 1.995164534979088, "learning_rate": 5.0470422146639275e-06, "loss": 0.4903, "step": 18946 }, { "epoch": 0.6750271656839517, "grad_norm": 2.4831393940630733, "learning_rate": 5.0460397946730174e-06, "loss": 0.5097, "step": 18947 }, { "epoch": 0.6750627928104459, "grad_norm": 1.1432988541234208, "learning_rate": 5.045037440647547e-06, "loss": 0.3871, "step": 18948 }, { "epoch": 0.67509841993694, "grad_norm": 1.4970725467195143, "learning_rate": 5.044035152600869e-06, "loss": 0.509, "step": 18949 }, { "epoch": 0.675134047063434, "grad_norm": 1.7022008134655804, "learning_rate": 5.043032930546323e-06, "loss": 0.4497, "step": 18950 }, { "epoch": 0.6751696741899282, "grad_norm": 1.2482108486823547, "learning_rate": 5.042030774497259e-06, "loss": 0.3815, "step": 18951 }, { "epoch": 0.6752053013164223, "grad_norm": 1.2039709292633374, "learning_rate": 5.041028684467022e-06, "loss": 0.4627, "step": 18952 }, { "epoch": 0.6752409284429164, "grad_norm": 1.0642847781692437, "learning_rate": 5.040026660468953e-06, "loss": 0.3848, "step": 18953 }, { "epoch": 0.6752765555694106, "grad_norm": 1.5699982011526183, "learning_rate": 5.0390247025164e-06, "loss": 0.4847, "step": 18954 }, { "epoch": 0.6753121826959047, "grad_norm": 1.2853826806684052, "learning_rate": 5.038022810622696e-06, "loss": 0.4723, "step": 18955 }, { "epoch": 0.6753478098223987, "grad_norm": 1.4156287244799453, "learning_rate": 5.037020984801191e-06, "loss": 0.509, "step": 18956 }, { "epoch": 0.6753834369488929, "grad_norm": 1.2191794632220638, "learning_rate": 5.036019225065215e-06, "loss": 0.4733, "step": 18957 }, { "epoch": 0.675419064075387, "grad_norm": 1.082657799847507, "learning_rate": 5.035017531428118e-06, "loss": 0.3736, "step": 18958 }, { "epoch": 0.6754546912018811, "grad_norm": 1.8148409086979465, "learning_rate": 5.034015903903229e-06, "loss": 0.5294, "step": 18959 }, { "epoch": 0.6754903183283752, "grad_norm": 1.7083515095868351, "learning_rate": 5.033014342503889e-06, "loss": 0.4676, "step": 18960 }, { "epoch": 0.6755259454548693, "grad_norm": 1.1514576195800241, "learning_rate": 5.032012847243435e-06, "loss": 0.3974, "step": 18961 }, { "epoch": 0.6755615725813634, "grad_norm": 1.0481125417732045, "learning_rate": 5.031011418135205e-06, "loss": 0.4204, "step": 18962 }, { "epoch": 0.6755971997078576, "grad_norm": 1.2473205209496887, "learning_rate": 5.030010055192529e-06, "loss": 0.4096, "step": 18963 }, { "epoch": 0.6756328268343517, "grad_norm": 1.607054886475509, "learning_rate": 5.029008758428742e-06, "loss": 0.4841, "step": 18964 }, { "epoch": 0.6756684539608457, "grad_norm": 1.0999431643916537, "learning_rate": 5.028007527857182e-06, "loss": 0.3986, "step": 18965 }, { "epoch": 0.6757040810873399, "grad_norm": 1.4736903969145208, "learning_rate": 5.027006363491174e-06, "loss": 0.451, "step": 18966 }, { "epoch": 0.675739708213834, "grad_norm": 1.117449115104823, "learning_rate": 5.02600526534405e-06, "loss": 0.4401, "step": 18967 }, { "epoch": 0.6757753353403281, "grad_norm": 1.0650018314506577, "learning_rate": 5.025004233429145e-06, "loss": 0.445, "step": 18968 }, { "epoch": 0.6758109624668223, "grad_norm": 1.2020846835060657, "learning_rate": 5.0240032677597875e-06, "loss": 0.3774, "step": 18969 }, { "epoch": 0.6758465895933163, "grad_norm": 1.368329557822835, "learning_rate": 5.023002368349301e-06, "loss": 0.4854, "step": 18970 }, { "epoch": 0.6758822167198104, "grad_norm": 1.152563494746273, "learning_rate": 5.022001535211022e-06, "loss": 0.3518, "step": 18971 }, { "epoch": 0.6759178438463046, "grad_norm": 1.2955235429241274, "learning_rate": 5.021000768358267e-06, "loss": 0.4061, "step": 18972 }, { "epoch": 0.6759534709727987, "grad_norm": 1.5714440795890676, "learning_rate": 5.020000067804372e-06, "loss": 0.4268, "step": 18973 }, { "epoch": 0.6759890980992928, "grad_norm": 1.5386161441222477, "learning_rate": 5.018999433562652e-06, "loss": 0.5185, "step": 18974 }, { "epoch": 0.6760247252257869, "grad_norm": 1.525307155385382, "learning_rate": 5.0179988656464366e-06, "loss": 0.4784, "step": 18975 }, { "epoch": 0.676060352352281, "grad_norm": 1.680585923358886, "learning_rate": 5.0169983640690476e-06, "loss": 0.5565, "step": 18976 }, { "epoch": 0.6760959794787751, "grad_norm": 1.212047252242431, "learning_rate": 5.0159979288438075e-06, "loss": 0.3911, "step": 18977 }, { "epoch": 0.6761316066052693, "grad_norm": 1.158453062559855, "learning_rate": 5.014997559984045e-06, "loss": 0.4192, "step": 18978 }, { "epoch": 0.6761672337317634, "grad_norm": 1.2368141227619924, "learning_rate": 5.013997257503069e-06, "loss": 0.4355, "step": 18979 }, { "epoch": 0.6762028608582574, "grad_norm": 1.231214173936706, "learning_rate": 5.012997021414208e-06, "loss": 0.4897, "step": 18980 }, { "epoch": 0.6762384879847516, "grad_norm": 1.0829479692670376, "learning_rate": 5.011996851730774e-06, "loss": 0.413, "step": 18981 }, { "epoch": 0.6762741151112457, "grad_norm": 1.5816023668818289, "learning_rate": 5.010996748466088e-06, "loss": 0.4946, "step": 18982 }, { "epoch": 0.6763097422377398, "grad_norm": 1.5777614986459065, "learning_rate": 5.009996711633469e-06, "loss": 0.5601, "step": 18983 }, { "epoch": 0.676345369364234, "grad_norm": 1.1214151022665868, "learning_rate": 5.008996741246234e-06, "loss": 0.4138, "step": 18984 }, { "epoch": 0.676380996490728, "grad_norm": 1.9018442773003907, "learning_rate": 5.007996837317692e-06, "loss": 0.5451, "step": 18985 }, { "epoch": 0.6764166236172221, "grad_norm": 1.362875755677011, "learning_rate": 5.006996999861163e-06, "loss": 0.5268, "step": 18986 }, { "epoch": 0.6764522507437163, "grad_norm": 1.8668744336058993, "learning_rate": 5.00599722888996e-06, "loss": 0.5367, "step": 18987 }, { "epoch": 0.6764878778702104, "grad_norm": 1.1076998245377085, "learning_rate": 5.004997524417392e-06, "loss": 0.4407, "step": 18988 }, { "epoch": 0.6765235049967044, "grad_norm": 1.1488675893933105, "learning_rate": 5.003997886456777e-06, "loss": 0.426, "step": 18989 }, { "epoch": 0.6765591321231986, "grad_norm": 1.1909524675388747, "learning_rate": 5.002998315021415e-06, "loss": 0.4392, "step": 18990 }, { "epoch": 0.6765947592496927, "grad_norm": 1.897841178601761, "learning_rate": 5.00199881012463e-06, "loss": 0.5711, "step": 18991 }, { "epoch": 0.6766303863761868, "grad_norm": 1.7643379596579853, "learning_rate": 5.00099937177972e-06, "loss": 0.5713, "step": 18992 }, { "epoch": 0.676666013502681, "grad_norm": 1.152955183286367, "learning_rate": 5.000000000000003e-06, "loss": 0.4295, "step": 18993 }, { "epoch": 0.676701640629175, "grad_norm": 1.400104779966289, "learning_rate": 4.999000694798776e-06, "loss": 0.532, "step": 18994 }, { "epoch": 0.6767372677556691, "grad_norm": 1.5613400589421214, "learning_rate": 4.998001456189353e-06, "loss": 0.611, "step": 18995 }, { "epoch": 0.6767728948821633, "grad_norm": 1.1906023768168508, "learning_rate": 4.997002284185034e-06, "loss": 0.4388, "step": 18996 }, { "epoch": 0.6768085220086574, "grad_norm": 1.1559593745692376, "learning_rate": 4.996003178799127e-06, "loss": 0.4684, "step": 18997 }, { "epoch": 0.6768441491351515, "grad_norm": 1.1216512351641852, "learning_rate": 4.995004140044935e-06, "loss": 0.5019, "step": 18998 }, { "epoch": 0.6768797762616456, "grad_norm": 1.2938014385657741, "learning_rate": 4.99400516793576e-06, "loss": 0.5245, "step": 18999 }, { "epoch": 0.6769154033881397, "grad_norm": 1.162844230960253, "learning_rate": 4.993006262484911e-06, "loss": 0.4538, "step": 19000 }, { "epoch": 0.6769510305146338, "grad_norm": 1.1795563130642104, "learning_rate": 4.992007423705677e-06, "loss": 0.4513, "step": 19001 }, { "epoch": 0.676986657641128, "grad_norm": 1.1486675345086148, "learning_rate": 4.99100865161137e-06, "loss": 0.4332, "step": 19002 }, { "epoch": 0.677022284767622, "grad_norm": 1.0374524713158586, "learning_rate": 4.990009946215279e-06, "loss": 0.3906, "step": 19003 }, { "epoch": 0.6770579118941161, "grad_norm": 2.2606027238701203, "learning_rate": 4.989011307530709e-06, "loss": 0.5786, "step": 19004 }, { "epoch": 0.6770935390206103, "grad_norm": 1.2857028826739083, "learning_rate": 4.988012735570955e-06, "loss": 0.5921, "step": 19005 }, { "epoch": 0.6771291661471044, "grad_norm": 1.0120839106160022, "learning_rate": 4.987014230349319e-06, "loss": 0.4255, "step": 19006 }, { "epoch": 0.6771647932735985, "grad_norm": 1.208041930085557, "learning_rate": 4.986015791879088e-06, "loss": 0.4599, "step": 19007 }, { "epoch": 0.6772004204000927, "grad_norm": 1.4490732133401647, "learning_rate": 4.9850174201735645e-06, "loss": 0.486, "step": 19008 }, { "epoch": 0.6772360475265867, "grad_norm": 1.0633199668938356, "learning_rate": 4.984019115246036e-06, "loss": 0.4678, "step": 19009 }, { "epoch": 0.6772716746530808, "grad_norm": 1.6042053467482111, "learning_rate": 4.983020877109799e-06, "loss": 0.49, "step": 19010 }, { "epoch": 0.677307301779575, "grad_norm": 1.4934148757742396, "learning_rate": 4.98202270577815e-06, "loss": 0.5287, "step": 19011 }, { "epoch": 0.6773429289060691, "grad_norm": 1.0826133634060977, "learning_rate": 4.981024601264372e-06, "loss": 0.3829, "step": 19012 }, { "epoch": 0.6773785560325631, "grad_norm": 1.4785548182417694, "learning_rate": 4.980026563581759e-06, "loss": 0.4479, "step": 19013 }, { "epoch": 0.6774141831590573, "grad_norm": 1.302930804774383, "learning_rate": 4.979028592743601e-06, "loss": 0.4353, "step": 19014 }, { "epoch": 0.6774498102855514, "grad_norm": 1.626409062100652, "learning_rate": 4.978030688763191e-06, "loss": 0.4971, "step": 19015 }, { "epoch": 0.6774854374120455, "grad_norm": 1.1640007057283324, "learning_rate": 4.977032851653808e-06, "loss": 0.431, "step": 19016 }, { "epoch": 0.6775210645385397, "grad_norm": 1.3940023043414898, "learning_rate": 4.9760350814287486e-06, "loss": 0.4702, "step": 19017 }, { "epoch": 0.6775566916650337, "grad_norm": 1.2876886426002954, "learning_rate": 4.975037378101289e-06, "loss": 0.4221, "step": 19018 }, { "epoch": 0.6775923187915278, "grad_norm": 1.3523592819917396, "learning_rate": 4.974039741684719e-06, "loss": 0.4587, "step": 19019 }, { "epoch": 0.677627945918022, "grad_norm": 1.5775314764071109, "learning_rate": 4.973042172192323e-06, "loss": 0.4952, "step": 19020 }, { "epoch": 0.6776635730445161, "grad_norm": 1.0881915520826275, "learning_rate": 4.972044669637384e-06, "loss": 0.4343, "step": 19021 }, { "epoch": 0.6776992001710102, "grad_norm": 1.1226678606282496, "learning_rate": 4.971047234033187e-06, "loss": 0.4759, "step": 19022 }, { "epoch": 0.6777348272975043, "grad_norm": 1.5324176572108497, "learning_rate": 4.970049865393009e-06, "loss": 0.5689, "step": 19023 }, { "epoch": 0.6777704544239984, "grad_norm": 1.1363427933000743, "learning_rate": 4.969052563730136e-06, "loss": 0.406, "step": 19024 }, { "epoch": 0.6778060815504925, "grad_norm": 1.2159949743040013, "learning_rate": 4.9680553290578405e-06, "loss": 0.4517, "step": 19025 }, { "epoch": 0.6778417086769867, "grad_norm": 1.6400166023750393, "learning_rate": 4.9670581613894094e-06, "loss": 0.5337, "step": 19026 }, { "epoch": 0.6778773358034808, "grad_norm": 1.1729318349743785, "learning_rate": 4.96606106073811e-06, "loss": 0.4916, "step": 19027 }, { "epoch": 0.6779129629299749, "grad_norm": 1.328395047542669, "learning_rate": 4.965064027117232e-06, "loss": 0.5123, "step": 19028 }, { "epoch": 0.677948590056469, "grad_norm": 1.155924612963533, "learning_rate": 4.964067060540043e-06, "loss": 0.4879, "step": 19029 }, { "epoch": 0.6779842171829631, "grad_norm": 1.140472358401462, "learning_rate": 4.963070161019825e-06, "loss": 0.4246, "step": 19030 }, { "epoch": 0.6780198443094573, "grad_norm": 1.0023113313692238, "learning_rate": 4.962073328569844e-06, "loss": 0.4045, "step": 19031 }, { "epoch": 0.6780554714359514, "grad_norm": 1.524635092031552, "learning_rate": 4.961076563203379e-06, "loss": 0.5658, "step": 19032 }, { "epoch": 0.6780910985624454, "grad_norm": 1.456667460225496, "learning_rate": 4.960079864933705e-06, "loss": 0.5363, "step": 19033 }, { "epoch": 0.6781267256889396, "grad_norm": 1.6106736014846768, "learning_rate": 4.959083233774085e-06, "loss": 0.5584, "step": 19034 }, { "epoch": 0.6781623528154337, "grad_norm": 1.2082318187711674, "learning_rate": 4.958086669737796e-06, "loss": 0.4257, "step": 19035 }, { "epoch": 0.6781979799419278, "grad_norm": 1.087478651736466, "learning_rate": 4.957090172838107e-06, "loss": 0.417, "step": 19036 }, { "epoch": 0.678233607068422, "grad_norm": 1.099131522088846, "learning_rate": 4.956093743088292e-06, "loss": 0.5006, "step": 19037 }, { "epoch": 0.678269234194916, "grad_norm": 1.2210496261515005, "learning_rate": 4.955097380501609e-06, "loss": 0.4401, "step": 19038 }, { "epoch": 0.6783048613214101, "grad_norm": 1.0950447173382973, "learning_rate": 4.954101085091335e-06, "loss": 0.4064, "step": 19039 }, { "epoch": 0.6783404884479043, "grad_norm": 1.6434607617383923, "learning_rate": 4.953104856870729e-06, "loss": 0.5812, "step": 19040 }, { "epoch": 0.6783761155743984, "grad_norm": 1.216370909091347, "learning_rate": 4.952108695853057e-06, "loss": 0.4612, "step": 19041 }, { "epoch": 0.6784117427008924, "grad_norm": 1.5982171117781956, "learning_rate": 4.951112602051587e-06, "loss": 0.5591, "step": 19042 }, { "epoch": 0.6784473698273866, "grad_norm": 1.1725168716052157, "learning_rate": 4.950116575479587e-06, "loss": 0.398, "step": 19043 }, { "epoch": 0.6784829969538807, "grad_norm": 1.2881701855836976, "learning_rate": 4.949120616150309e-06, "loss": 0.4274, "step": 19044 }, { "epoch": 0.6785186240803748, "grad_norm": 1.4238743445134558, "learning_rate": 4.948124724077021e-06, "loss": 0.5548, "step": 19045 }, { "epoch": 0.678554251206869, "grad_norm": 1.6047462561611703, "learning_rate": 4.947128899272988e-06, "loss": 0.4863, "step": 19046 }, { "epoch": 0.678589878333363, "grad_norm": 1.5816697022552477, "learning_rate": 4.94613314175146e-06, "loss": 0.5466, "step": 19047 }, { "epoch": 0.6786255054598571, "grad_norm": 1.0931322008818898, "learning_rate": 4.945137451525707e-06, "loss": 0.4464, "step": 19048 }, { "epoch": 0.6786611325863513, "grad_norm": 1.1606636658939578, "learning_rate": 4.944141828608977e-06, "loss": 0.3861, "step": 19049 }, { "epoch": 0.6786967597128454, "grad_norm": 1.0854632942216382, "learning_rate": 4.943146273014533e-06, "loss": 0.4221, "step": 19050 }, { "epoch": 0.6787323868393395, "grad_norm": 1.192415405544568, "learning_rate": 4.9421507847556305e-06, "loss": 0.398, "step": 19051 }, { "epoch": 0.6787680139658336, "grad_norm": 1.1226712732465436, "learning_rate": 4.94115536384553e-06, "loss": 0.4633, "step": 19052 }, { "epoch": 0.6788036410923277, "grad_norm": 1.4542357235550634, "learning_rate": 4.940160010297476e-06, "loss": 0.4835, "step": 19053 }, { "epoch": 0.6788392682188218, "grad_norm": 1.092327831191336, "learning_rate": 4.9391647241247345e-06, "loss": 0.4193, "step": 19054 }, { "epoch": 0.678874895345316, "grad_norm": 1.21071573320566, "learning_rate": 4.9381695053405464e-06, "loss": 0.4558, "step": 19055 }, { "epoch": 0.67891052247181, "grad_norm": 1.3970568279945037, "learning_rate": 4.937174353958171e-06, "loss": 0.4797, "step": 19056 }, { "epoch": 0.6789461495983041, "grad_norm": 1.2132973086777548, "learning_rate": 4.936179269990856e-06, "loss": 0.4273, "step": 19057 }, { "epoch": 0.6789817767247983, "grad_norm": 1.2421985835610172, "learning_rate": 4.935184253451855e-06, "loss": 0.4224, "step": 19058 }, { "epoch": 0.6790174038512924, "grad_norm": 1.129845883097935, "learning_rate": 4.934189304354418e-06, "loss": 0.3931, "step": 19059 }, { "epoch": 0.6790530309777865, "grad_norm": 1.1384923190985083, "learning_rate": 4.933194422711788e-06, "loss": 0.414, "step": 19060 }, { "epoch": 0.6790886581042807, "grad_norm": 1.0472818465733904, "learning_rate": 4.932199608537221e-06, "loss": 0.4142, "step": 19061 }, { "epoch": 0.6791242852307747, "grad_norm": 1.3646166110265063, "learning_rate": 4.931204861843952e-06, "loss": 0.4722, "step": 19062 }, { "epoch": 0.6791599123572688, "grad_norm": 1.1565943414467965, "learning_rate": 4.930210182645239e-06, "loss": 0.475, "step": 19063 }, { "epoch": 0.679195539483763, "grad_norm": 1.6538179546175544, "learning_rate": 4.929215570954315e-06, "loss": 0.4732, "step": 19064 }, { "epoch": 0.6792311666102571, "grad_norm": 1.0276830331641482, "learning_rate": 4.9282210267844325e-06, "loss": 0.4569, "step": 19065 }, { "epoch": 0.6792667937367511, "grad_norm": 1.2446812686173798, "learning_rate": 4.9272265501488304e-06, "loss": 0.4415, "step": 19066 }, { "epoch": 0.6793024208632453, "grad_norm": 1.1450077835760029, "learning_rate": 4.9262321410607526e-06, "loss": 0.4102, "step": 19067 }, { "epoch": 0.6793380479897394, "grad_norm": 1.1010452742736527, "learning_rate": 4.925237799533445e-06, "loss": 0.4698, "step": 19068 }, { "epoch": 0.6793736751162335, "grad_norm": 1.1399724545810261, "learning_rate": 4.9242435255801376e-06, "loss": 0.427, "step": 19069 }, { "epoch": 0.6794093022427277, "grad_norm": 1.1777275230381303, "learning_rate": 4.92324931921408e-06, "loss": 0.4179, "step": 19070 }, { "epoch": 0.6794449293692217, "grad_norm": 1.3246483586935867, "learning_rate": 4.922255180448504e-06, "loss": 0.5339, "step": 19071 }, { "epoch": 0.6794805564957158, "grad_norm": 1.3272324137750509, "learning_rate": 4.921261109296648e-06, "loss": 0.4591, "step": 19072 }, { "epoch": 0.67951618362221, "grad_norm": 1.505681855949542, "learning_rate": 4.92026710577175e-06, "loss": 0.5915, "step": 19073 }, { "epoch": 0.6795518107487041, "grad_norm": 1.1789505662234077, "learning_rate": 4.91927316988705e-06, "loss": 0.4208, "step": 19074 }, { "epoch": 0.6795874378751982, "grad_norm": 1.7292942076926663, "learning_rate": 4.918279301655775e-06, "loss": 0.5081, "step": 19075 }, { "epoch": 0.6796230650016923, "grad_norm": 1.204659122797218, "learning_rate": 4.917285501091167e-06, "loss": 0.4498, "step": 19076 }, { "epoch": 0.6796586921281864, "grad_norm": 1.4367319846884896, "learning_rate": 4.916291768206451e-06, "loss": 0.5155, "step": 19077 }, { "epoch": 0.6796943192546805, "grad_norm": 1.5383547639223851, "learning_rate": 4.9152981030148635e-06, "loss": 0.5471, "step": 19078 }, { "epoch": 0.6797299463811747, "grad_norm": 1.1960624806329112, "learning_rate": 4.9143045055296356e-06, "loss": 0.3963, "step": 19079 }, { "epoch": 0.6797655735076688, "grad_norm": 1.3031840922503755, "learning_rate": 4.913310975763999e-06, "loss": 0.4379, "step": 19080 }, { "epoch": 0.6798012006341628, "grad_norm": 1.4935775878191204, "learning_rate": 4.912317513731185e-06, "loss": 0.5802, "step": 19081 }, { "epoch": 0.679836827760657, "grad_norm": 1.161083736097341, "learning_rate": 4.911324119444416e-06, "loss": 0.4881, "step": 19082 }, { "epoch": 0.6798724548871511, "grad_norm": 1.3504873204746652, "learning_rate": 4.910330792916926e-06, "loss": 0.5046, "step": 19083 }, { "epoch": 0.6799080820136452, "grad_norm": 1.4690460714663385, "learning_rate": 4.909337534161936e-06, "loss": 0.5426, "step": 19084 }, { "epoch": 0.6799437091401394, "grad_norm": 1.0867321428733197, "learning_rate": 4.908344343192679e-06, "loss": 0.3875, "step": 19085 }, { "epoch": 0.6799793362666334, "grad_norm": 1.4154908745388937, "learning_rate": 4.907351220022371e-06, "loss": 0.4271, "step": 19086 }, { "epoch": 0.6800149633931275, "grad_norm": 2.398950753743511, "learning_rate": 4.906358164664242e-06, "loss": 0.5691, "step": 19087 }, { "epoch": 0.6800505905196217, "grad_norm": 1.5063208775132868, "learning_rate": 4.905365177131514e-06, "loss": 0.4346, "step": 19088 }, { "epoch": 0.6800862176461158, "grad_norm": 1.5560636714609726, "learning_rate": 4.904372257437413e-06, "loss": 0.5714, "step": 19089 }, { "epoch": 0.6801218447726098, "grad_norm": 1.611419769534404, "learning_rate": 4.903379405595154e-06, "loss": 0.5284, "step": 19090 }, { "epoch": 0.680157471899104, "grad_norm": 1.0879431774576522, "learning_rate": 4.902386621617961e-06, "loss": 0.3966, "step": 19091 }, { "epoch": 0.6801930990255981, "grad_norm": 1.283673975100312, "learning_rate": 4.9013939055190554e-06, "loss": 0.4635, "step": 19092 }, { "epoch": 0.6802287261520922, "grad_norm": 1.5040818957795923, "learning_rate": 4.90040125731165e-06, "loss": 0.5077, "step": 19093 }, { "epoch": 0.6802643532785864, "grad_norm": 1.1346263719500436, "learning_rate": 4.899408677008968e-06, "loss": 0.39, "step": 19094 }, { "epoch": 0.6802999804050804, "grad_norm": 1.4845407334857477, "learning_rate": 4.898416164624222e-06, "loss": 0.4535, "step": 19095 }, { "epoch": 0.6803356075315745, "grad_norm": 1.1707012968085586, "learning_rate": 4.897423720170637e-06, "loss": 0.444, "step": 19096 }, { "epoch": 0.6803712346580687, "grad_norm": 1.4095452895273501, "learning_rate": 4.896431343661416e-06, "loss": 0.5236, "step": 19097 }, { "epoch": 0.6804068617845628, "grad_norm": 1.197288920010799, "learning_rate": 4.895439035109784e-06, "loss": 0.3944, "step": 19098 }, { "epoch": 0.6804424889110569, "grad_norm": 1.2455437995234566, "learning_rate": 4.894446794528944e-06, "loss": 0.4247, "step": 19099 }, { "epoch": 0.680478116037551, "grad_norm": 1.144938070301269, "learning_rate": 4.893454621932118e-06, "loss": 0.416, "step": 19100 }, { "epoch": 0.6805137431640451, "grad_norm": 1.4944186837956546, "learning_rate": 4.892462517332509e-06, "loss": 0.5509, "step": 19101 }, { "epoch": 0.6805493702905392, "grad_norm": 1.158396093439052, "learning_rate": 4.8914704807433275e-06, "loss": 0.4188, "step": 19102 }, { "epoch": 0.6805849974170334, "grad_norm": 0.9822692779165526, "learning_rate": 4.890478512177796e-06, "loss": 0.3707, "step": 19103 }, { "epoch": 0.6806206245435275, "grad_norm": 1.1473614873104272, "learning_rate": 4.8894866116491095e-06, "loss": 0.425, "step": 19104 }, { "epoch": 0.6806562516700215, "grad_norm": 1.2386621207411743, "learning_rate": 4.888494779170484e-06, "loss": 0.47, "step": 19105 }, { "epoch": 0.6806918787965157, "grad_norm": 1.5955450340553727, "learning_rate": 4.887503014755121e-06, "loss": 0.4778, "step": 19106 }, { "epoch": 0.6807275059230098, "grad_norm": 1.7585709801949245, "learning_rate": 4.8865113184162315e-06, "loss": 0.5149, "step": 19107 }, { "epoch": 0.6807631330495039, "grad_norm": 1.689986730849042, "learning_rate": 4.885519690167015e-06, "loss": 0.6385, "step": 19108 }, { "epoch": 0.680798760175998, "grad_norm": 1.060089867843455, "learning_rate": 4.884528130020677e-06, "loss": 0.4428, "step": 19109 }, { "epoch": 0.6808343873024921, "grad_norm": 1.5637575854512942, "learning_rate": 4.8835366379904235e-06, "loss": 0.5173, "step": 19110 }, { "epoch": 0.6808700144289862, "grad_norm": 1.143470564114105, "learning_rate": 4.88254521408946e-06, "loss": 0.4077, "step": 19111 }, { "epoch": 0.6809056415554804, "grad_norm": 2.0443979044019898, "learning_rate": 4.88155385833098e-06, "loss": 0.5086, "step": 19112 }, { "epoch": 0.6809412686819745, "grad_norm": 1.1697713922087012, "learning_rate": 4.880562570728188e-06, "loss": 0.4527, "step": 19113 }, { "epoch": 0.6809768958084685, "grad_norm": 1.2873008913284052, "learning_rate": 4.879571351294287e-06, "loss": 0.4832, "step": 19114 }, { "epoch": 0.6810125229349627, "grad_norm": 1.7070936970821027, "learning_rate": 4.878580200042469e-06, "loss": 0.4262, "step": 19115 }, { "epoch": 0.6810481500614568, "grad_norm": 1.1973237699855879, "learning_rate": 4.877589116985935e-06, "loss": 0.4622, "step": 19116 }, { "epoch": 0.6810837771879509, "grad_norm": 1.6817142878880225, "learning_rate": 4.876598102137882e-06, "loss": 0.5728, "step": 19117 }, { "epoch": 0.6811194043144451, "grad_norm": 1.6566346057260863, "learning_rate": 4.875607155511511e-06, "loss": 0.568, "step": 19118 }, { "epoch": 0.6811550314409391, "grad_norm": 1.214867110296868, "learning_rate": 4.874616277120008e-06, "loss": 0.4835, "step": 19119 }, { "epoch": 0.6811906585674332, "grad_norm": 1.1978933891543357, "learning_rate": 4.873625466976575e-06, "loss": 0.3882, "step": 19120 }, { "epoch": 0.6812262856939274, "grad_norm": 1.7100190835563711, "learning_rate": 4.8726347250943995e-06, "loss": 0.5129, "step": 19121 }, { "epoch": 0.6812619128204215, "grad_norm": 1.1159762892725242, "learning_rate": 4.871644051486679e-06, "loss": 0.4318, "step": 19122 }, { "epoch": 0.6812975399469156, "grad_norm": 0.9517238057729068, "learning_rate": 4.8706534461666e-06, "loss": 0.3861, "step": 19123 }, { "epoch": 0.6813331670734097, "grad_norm": 1.1343665447544267, "learning_rate": 4.869662909147354e-06, "loss": 0.3766, "step": 19124 }, { "epoch": 0.6813687941999038, "grad_norm": 1.2744635923368723, "learning_rate": 4.868672440442134e-06, "loss": 0.444, "step": 19125 }, { "epoch": 0.6814044213263979, "grad_norm": 1.1365603299935385, "learning_rate": 4.8676820400641255e-06, "loss": 0.4198, "step": 19126 }, { "epoch": 0.6814400484528921, "grad_norm": 1.421918482363676, "learning_rate": 4.866691708026522e-06, "loss": 0.4974, "step": 19127 }, { "epoch": 0.6814756755793862, "grad_norm": 1.5799096809188187, "learning_rate": 4.865701444342503e-06, "loss": 0.5048, "step": 19128 }, { "epoch": 0.6815113027058802, "grad_norm": 1.4504356304879749, "learning_rate": 4.864711249025261e-06, "loss": 0.4831, "step": 19129 }, { "epoch": 0.6815469298323744, "grad_norm": 1.2094588593197004, "learning_rate": 4.863721122087975e-06, "loss": 0.446, "step": 19130 }, { "epoch": 0.6815825569588685, "grad_norm": 1.1175477627942958, "learning_rate": 4.862731063543833e-06, "loss": 0.4262, "step": 19131 }, { "epoch": 0.6816181840853626, "grad_norm": 1.1470348481692263, "learning_rate": 4.8617410734060166e-06, "loss": 0.4275, "step": 19132 }, { "epoch": 0.6816538112118568, "grad_norm": 1.6326221912271643, "learning_rate": 4.860751151687713e-06, "loss": 0.5149, "step": 19133 }, { "epoch": 0.6816894383383508, "grad_norm": 1.3331705105640734, "learning_rate": 4.859761298402097e-06, "loss": 0.4859, "step": 19134 }, { "epoch": 0.6817250654648449, "grad_norm": 1.3947396758547925, "learning_rate": 4.858771513562355e-06, "loss": 0.486, "step": 19135 }, { "epoch": 0.6817606925913391, "grad_norm": 1.4961633497044438, "learning_rate": 4.857781797181661e-06, "loss": 0.4811, "step": 19136 }, { "epoch": 0.6817963197178332, "grad_norm": 1.1447535480915438, "learning_rate": 4.856792149273195e-06, "loss": 0.432, "step": 19137 }, { "epoch": 0.6818319468443272, "grad_norm": 1.138150400373466, "learning_rate": 4.855802569850141e-06, "loss": 0.375, "step": 19138 }, { "epoch": 0.6818675739708214, "grad_norm": 1.2291959403874553, "learning_rate": 4.854813058925668e-06, "loss": 0.4187, "step": 19139 }, { "epoch": 0.6819032010973155, "grad_norm": 2.385688186612778, "learning_rate": 4.853823616512955e-06, "loss": 0.5412, "step": 19140 }, { "epoch": 0.6819388282238096, "grad_norm": 1.5513494648859207, "learning_rate": 4.852834242625177e-06, "loss": 0.4408, "step": 19141 }, { "epoch": 0.6819744553503038, "grad_norm": 1.2308102113695663, "learning_rate": 4.851844937275514e-06, "loss": 0.4655, "step": 19142 }, { "epoch": 0.6820100824767978, "grad_norm": 1.2095255014044666, "learning_rate": 4.85085570047713e-06, "loss": 0.4724, "step": 19143 }, { "epoch": 0.6820457096032919, "grad_norm": 1.151818996357801, "learning_rate": 4.8498665322432036e-06, "loss": 0.487, "step": 19144 }, { "epoch": 0.6820813367297861, "grad_norm": 1.0967316841811334, "learning_rate": 4.848877432586901e-06, "loss": 0.3889, "step": 19145 }, { "epoch": 0.6821169638562802, "grad_norm": 1.026713962885045, "learning_rate": 4.847888401521396e-06, "loss": 0.4005, "step": 19146 }, { "epoch": 0.6821525909827743, "grad_norm": 1.227188448815509, "learning_rate": 4.846899439059858e-06, "loss": 0.4393, "step": 19147 }, { "epoch": 0.6821882181092684, "grad_norm": 1.273990637097322, "learning_rate": 4.845910545215456e-06, "loss": 0.4743, "step": 19148 }, { "epoch": 0.6822238452357625, "grad_norm": 1.3746282299550647, "learning_rate": 4.844921720001361e-06, "loss": 0.552, "step": 19149 }, { "epoch": 0.6822594723622566, "grad_norm": 1.167511050474651, "learning_rate": 4.843932963430733e-06, "loss": 0.4558, "step": 19150 }, { "epoch": 0.6822950994887508, "grad_norm": 1.16374478676235, "learning_rate": 4.842944275516745e-06, "loss": 0.4051, "step": 19151 }, { "epoch": 0.6823307266152449, "grad_norm": 1.0717811853063284, "learning_rate": 4.8419556562725545e-06, "loss": 0.4452, "step": 19152 }, { "epoch": 0.6823663537417389, "grad_norm": 1.5786129519911658, "learning_rate": 4.8409671057113335e-06, "loss": 0.5554, "step": 19153 }, { "epoch": 0.6824019808682331, "grad_norm": 1.799264861519857, "learning_rate": 4.8399786238462345e-06, "loss": 0.4772, "step": 19154 }, { "epoch": 0.6824376079947272, "grad_norm": 1.133009028196215, "learning_rate": 4.838990210690434e-06, "loss": 0.425, "step": 19155 }, { "epoch": 0.6824732351212213, "grad_norm": 1.0643315009180605, "learning_rate": 4.838001866257083e-06, "loss": 0.4045, "step": 19156 }, { "epoch": 0.6825088622477155, "grad_norm": 1.7430353905608627, "learning_rate": 4.8370135905593475e-06, "loss": 0.5729, "step": 19157 }, { "epoch": 0.6825444893742095, "grad_norm": 1.190977953355337, "learning_rate": 4.836025383610382e-06, "loss": 0.4166, "step": 19158 }, { "epoch": 0.6825801165007036, "grad_norm": 1.1804273081180212, "learning_rate": 4.835037245423348e-06, "loss": 0.4408, "step": 19159 }, { "epoch": 0.6826157436271978, "grad_norm": 1.6786503374891084, "learning_rate": 4.8340491760114065e-06, "loss": 0.6153, "step": 19160 }, { "epoch": 0.6826513707536919, "grad_norm": 1.6193107256332515, "learning_rate": 4.833061175387708e-06, "loss": 0.4787, "step": 19161 }, { "epoch": 0.6826869978801859, "grad_norm": 1.4048548628921391, "learning_rate": 4.832073243565411e-06, "loss": 0.5276, "step": 19162 }, { "epoch": 0.6827226250066801, "grad_norm": 1.1608814479941854, "learning_rate": 4.83108538055767e-06, "loss": 0.3967, "step": 19163 }, { "epoch": 0.6827582521331742, "grad_norm": 1.6474578304845937, "learning_rate": 4.830097586377645e-06, "loss": 0.5381, "step": 19164 }, { "epoch": 0.6827938792596683, "grad_norm": 1.3745476853154432, "learning_rate": 4.82910986103848e-06, "loss": 0.4459, "step": 19165 }, { "epoch": 0.6828295063861625, "grad_norm": 1.6017782903639395, "learning_rate": 4.8281222045533345e-06, "loss": 0.5585, "step": 19166 }, { "epoch": 0.6828651335126565, "grad_norm": 1.2006800584119648, "learning_rate": 4.827134616935354e-06, "loss": 0.4376, "step": 19167 }, { "epoch": 0.6829007606391506, "grad_norm": 1.3048900398083976, "learning_rate": 4.826147098197691e-06, "loss": 0.5339, "step": 19168 }, { "epoch": 0.6829363877656448, "grad_norm": 1.5301790634310244, "learning_rate": 4.825159648353497e-06, "loss": 0.5205, "step": 19169 }, { "epoch": 0.6829720148921389, "grad_norm": 1.3916723721797892, "learning_rate": 4.8241722674159175e-06, "loss": 0.4857, "step": 19170 }, { "epoch": 0.683007642018633, "grad_norm": 1.146892005782853, "learning_rate": 4.823184955398107e-06, "loss": 0.4553, "step": 19171 }, { "epoch": 0.6830432691451271, "grad_norm": 1.1795422166381253, "learning_rate": 4.822197712313203e-06, "loss": 0.4182, "step": 19172 }, { "epoch": 0.6830788962716212, "grad_norm": 1.7347745666465568, "learning_rate": 4.821210538174359e-06, "loss": 0.6162, "step": 19173 }, { "epoch": 0.6831145233981153, "grad_norm": 1.133043702966262, "learning_rate": 4.820223432994713e-06, "loss": 0.422, "step": 19174 }, { "epoch": 0.6831501505246095, "grad_norm": 1.321621782315243, "learning_rate": 4.819236396787417e-06, "loss": 0.4727, "step": 19175 }, { "epoch": 0.6831857776511036, "grad_norm": 1.8847892914756446, "learning_rate": 4.818249429565605e-06, "loss": 0.5465, "step": 19176 }, { "epoch": 0.6832214047775976, "grad_norm": 1.7747766725065328, "learning_rate": 4.8172625313424225e-06, "loss": 0.533, "step": 19177 }, { "epoch": 0.6832570319040918, "grad_norm": 1.2646861071954438, "learning_rate": 4.816275702131013e-06, "loss": 0.4211, "step": 19178 }, { "epoch": 0.6832926590305859, "grad_norm": 1.549389070720424, "learning_rate": 4.8152889419445195e-06, "loss": 0.5458, "step": 19179 }, { "epoch": 0.68332828615708, "grad_norm": 1.0512899692477755, "learning_rate": 4.814302250796074e-06, "loss": 0.3895, "step": 19180 }, { "epoch": 0.6833639132835742, "grad_norm": 1.039640334333535, "learning_rate": 4.813315628698822e-06, "loss": 0.383, "step": 19181 }, { "epoch": 0.6833995404100682, "grad_norm": 1.1749776044924778, "learning_rate": 4.812329075665894e-06, "loss": 0.3923, "step": 19182 }, { "epoch": 0.6834351675365623, "grad_norm": 2.222072140876038, "learning_rate": 4.81134259171043e-06, "loss": 0.5564, "step": 19183 }, { "epoch": 0.6834707946630565, "grad_norm": 1.1259950988004175, "learning_rate": 4.8103561768455676e-06, "loss": 0.3991, "step": 19184 }, { "epoch": 0.6835064217895506, "grad_norm": 1.1633081617004073, "learning_rate": 4.809369831084439e-06, "loss": 0.4104, "step": 19185 }, { "epoch": 0.6835420489160446, "grad_norm": 1.5083932253089727, "learning_rate": 4.8083835544401825e-06, "loss": 0.4999, "step": 19186 }, { "epoch": 0.6835776760425388, "grad_norm": 1.6307804524048557, "learning_rate": 4.807397346925925e-06, "loss": 0.4314, "step": 19187 }, { "epoch": 0.6836133031690329, "grad_norm": 1.139839196604201, "learning_rate": 4.806411208554804e-06, "loss": 0.4066, "step": 19188 }, { "epoch": 0.683648930295527, "grad_norm": 1.613756504349828, "learning_rate": 4.805425139339944e-06, "loss": 0.5638, "step": 19189 }, { "epoch": 0.6836845574220212, "grad_norm": 1.3828209465633177, "learning_rate": 4.804439139294484e-06, "loss": 0.5034, "step": 19190 }, { "epoch": 0.6837201845485152, "grad_norm": 1.5412912482111283, "learning_rate": 4.80345320843154e-06, "loss": 0.5708, "step": 19191 }, { "epoch": 0.6837558116750093, "grad_norm": 1.7874054033422626, "learning_rate": 4.802467346764256e-06, "loss": 0.5392, "step": 19192 }, { "epoch": 0.6837914388015035, "grad_norm": 1.7854561450119277, "learning_rate": 4.801481554305748e-06, "loss": 0.5432, "step": 19193 }, { "epoch": 0.6838270659279976, "grad_norm": 1.5380541183840521, "learning_rate": 4.800495831069147e-06, "loss": 0.4924, "step": 19194 }, { "epoch": 0.6838626930544917, "grad_norm": 1.5884085326335644, "learning_rate": 4.799510177067583e-06, "loss": 0.5361, "step": 19195 }, { "epoch": 0.6838983201809858, "grad_norm": 1.21365342440976, "learning_rate": 4.798524592314171e-06, "loss": 0.4205, "step": 19196 }, { "epoch": 0.6839339473074799, "grad_norm": 1.1560939693235952, "learning_rate": 4.797539076822044e-06, "loss": 0.4026, "step": 19197 }, { "epoch": 0.683969574433974, "grad_norm": 1.4945994823580377, "learning_rate": 4.796553630604317e-06, "loss": 0.3742, "step": 19198 }, { "epoch": 0.6840052015604682, "grad_norm": 1.2102545156353766, "learning_rate": 4.795568253674115e-06, "loss": 0.4227, "step": 19199 }, { "epoch": 0.6840408286869623, "grad_norm": 1.1836924576749377, "learning_rate": 4.794582946044559e-06, "loss": 0.4252, "step": 19200 }, { "epoch": 0.6840764558134563, "grad_norm": 1.23732661328585, "learning_rate": 4.793597707728774e-06, "loss": 0.3571, "step": 19201 }, { "epoch": 0.6841120829399505, "grad_norm": 1.1252040744919583, "learning_rate": 4.792612538739871e-06, "loss": 0.4533, "step": 19202 }, { "epoch": 0.6841477100664446, "grad_norm": 1.152328537027218, "learning_rate": 4.791627439090975e-06, "loss": 0.4493, "step": 19203 }, { "epoch": 0.6841833371929387, "grad_norm": 1.2808705827831826, "learning_rate": 4.790642408795195e-06, "loss": 0.4255, "step": 19204 }, { "epoch": 0.6842189643194329, "grad_norm": 1.217038981038573, "learning_rate": 4.7896574478656545e-06, "loss": 0.4342, "step": 19205 }, { "epoch": 0.6842545914459269, "grad_norm": 1.5755227820611781, "learning_rate": 4.788672556315466e-06, "loss": 0.5452, "step": 19206 }, { "epoch": 0.684290218572421, "grad_norm": 1.5905752106844495, "learning_rate": 4.787687734157745e-06, "loss": 0.499, "step": 19207 }, { "epoch": 0.6843258456989152, "grad_norm": 1.4313437593616982, "learning_rate": 4.7867029814056085e-06, "loss": 0.6284, "step": 19208 }, { "epoch": 0.6843614728254093, "grad_norm": 2.0252158314777424, "learning_rate": 4.785718298072163e-06, "loss": 0.5273, "step": 19209 }, { "epoch": 0.6843970999519033, "grad_norm": 1.314573550934954, "learning_rate": 4.784733684170525e-06, "loss": 0.4113, "step": 19210 }, { "epoch": 0.6844327270783975, "grad_norm": 1.4601421153046361, "learning_rate": 4.783749139713801e-06, "loss": 0.4539, "step": 19211 }, { "epoch": 0.6844683542048916, "grad_norm": 1.3704339099293146, "learning_rate": 4.782764664715106e-06, "loss": 0.5212, "step": 19212 }, { "epoch": 0.6845039813313857, "grad_norm": 1.8509280936828294, "learning_rate": 4.781780259187543e-06, "loss": 0.4824, "step": 19213 }, { "epoch": 0.6845396084578799, "grad_norm": 1.474790532041609, "learning_rate": 4.780795923144222e-06, "loss": 0.5752, "step": 19214 }, { "epoch": 0.6845752355843739, "grad_norm": 1.7295687659239298, "learning_rate": 4.779811656598253e-06, "loss": 0.5955, "step": 19215 }, { "epoch": 0.684610862710868, "grad_norm": 1.5563305905904503, "learning_rate": 4.778827459562743e-06, "loss": 0.5696, "step": 19216 }, { "epoch": 0.6846464898373622, "grad_norm": 1.2221100953375836, "learning_rate": 4.777843332050791e-06, "loss": 0.4099, "step": 19217 }, { "epoch": 0.6846821169638563, "grad_norm": 2.6068108151514124, "learning_rate": 4.776859274075506e-06, "loss": 0.5549, "step": 19218 }, { "epoch": 0.6847177440903504, "grad_norm": 1.879348254733256, "learning_rate": 4.775875285649993e-06, "loss": 0.4301, "step": 19219 }, { "epoch": 0.6847533712168445, "grad_norm": 1.3689383671885806, "learning_rate": 4.774891366787349e-06, "loss": 0.3979, "step": 19220 }, { "epoch": 0.6847889983433386, "grad_norm": 1.4803261605120095, "learning_rate": 4.77390751750068e-06, "loss": 0.4675, "step": 19221 }, { "epoch": 0.6848246254698327, "grad_norm": 1.3614863490142215, "learning_rate": 4.7729237378030835e-06, "loss": 0.4442, "step": 19222 }, { "epoch": 0.6848602525963269, "grad_norm": 1.68199271612277, "learning_rate": 4.771940027707664e-06, "loss": 0.5773, "step": 19223 }, { "epoch": 0.684895879722821, "grad_norm": 1.140374620607511, "learning_rate": 4.770956387227516e-06, "loss": 0.4186, "step": 19224 }, { "epoch": 0.684931506849315, "grad_norm": 1.6354239199535532, "learning_rate": 4.76997281637574e-06, "loss": 0.5204, "step": 19225 }, { "epoch": 0.6849671339758092, "grad_norm": 1.2895726819836795, "learning_rate": 4.768989315165429e-06, "loss": 0.3861, "step": 19226 }, { "epoch": 0.6850027611023033, "grad_norm": 1.2253819714789018, "learning_rate": 4.768005883609684e-06, "loss": 0.431, "step": 19227 }, { "epoch": 0.6850383882287974, "grad_norm": 1.0954102852315521, "learning_rate": 4.767022521721595e-06, "loss": 0.3826, "step": 19228 }, { "epoch": 0.6850740153552916, "grad_norm": 1.3933616290110529, "learning_rate": 4.766039229514255e-06, "loss": 0.4667, "step": 19229 }, { "epoch": 0.6851096424817856, "grad_norm": 1.4886712364400583, "learning_rate": 4.765056007000768e-06, "loss": 0.5994, "step": 19230 }, { "epoch": 0.6851452696082797, "grad_norm": 1.2601883463753538, "learning_rate": 4.764072854194216e-06, "loss": 0.4422, "step": 19231 }, { "epoch": 0.6851808967347739, "grad_norm": 1.4317188448038656, "learning_rate": 4.763089771107697e-06, "loss": 0.5337, "step": 19232 }, { "epoch": 0.685216523861268, "grad_norm": 1.606505654179734, "learning_rate": 4.762106757754294e-06, "loss": 0.4195, "step": 19233 }, { "epoch": 0.685252150987762, "grad_norm": 1.3657496004651504, "learning_rate": 4.761123814147105e-06, "loss": 0.4727, "step": 19234 }, { "epoch": 0.6852877781142562, "grad_norm": 1.6539681517396572, "learning_rate": 4.76014094029921e-06, "loss": 0.4718, "step": 19235 }, { "epoch": 0.6853234052407503, "grad_norm": 1.2515031735501019, "learning_rate": 4.759158136223701e-06, "loss": 0.4388, "step": 19236 }, { "epoch": 0.6853590323672444, "grad_norm": 1.7022986376577205, "learning_rate": 4.758175401933665e-06, "loss": 0.4668, "step": 19237 }, { "epoch": 0.6853946594937386, "grad_norm": 2.048877309310361, "learning_rate": 4.757192737442191e-06, "loss": 0.5723, "step": 19238 }, { "epoch": 0.6854302866202326, "grad_norm": 1.6828853045872645, "learning_rate": 4.756210142762357e-06, "loss": 0.5249, "step": 19239 }, { "epoch": 0.6854659137467267, "grad_norm": 1.1920468994083833, "learning_rate": 4.755227617907249e-06, "loss": 0.4225, "step": 19240 }, { "epoch": 0.6855015408732209, "grad_norm": 1.4479609889477167, "learning_rate": 4.754245162889954e-06, "loss": 0.5018, "step": 19241 }, { "epoch": 0.685537167999715, "grad_norm": 1.186604309436307, "learning_rate": 4.75326277772355e-06, "loss": 0.3906, "step": 19242 }, { "epoch": 0.6855727951262091, "grad_norm": 1.5997215998341618, "learning_rate": 4.7522804624211175e-06, "loss": 0.436, "step": 19243 }, { "epoch": 0.6856084222527032, "grad_norm": 1.134377853752945, "learning_rate": 4.751298216995739e-06, "loss": 0.431, "step": 19244 }, { "epoch": 0.6856440493791973, "grad_norm": 1.2242341426706909, "learning_rate": 4.750316041460496e-06, "loss": 0.4225, "step": 19245 }, { "epoch": 0.6856796765056914, "grad_norm": 1.2020068949341944, "learning_rate": 4.749333935828463e-06, "loss": 0.4327, "step": 19246 }, { "epoch": 0.6857153036321856, "grad_norm": 1.3254176482137618, "learning_rate": 4.748351900112721e-06, "loss": 0.4298, "step": 19247 }, { "epoch": 0.6857509307586797, "grad_norm": 1.1834843500140908, "learning_rate": 4.74736993432634e-06, "loss": 0.3898, "step": 19248 }, { "epoch": 0.6857865578851737, "grad_norm": 1.6454045722188078, "learning_rate": 4.746388038482403e-06, "loss": 0.5048, "step": 19249 }, { "epoch": 0.6858221850116679, "grad_norm": 1.159364117137072, "learning_rate": 4.745406212593979e-06, "loss": 0.3974, "step": 19250 }, { "epoch": 0.685857812138162, "grad_norm": 1.3273488359928824, "learning_rate": 4.744424456674144e-06, "loss": 0.4261, "step": 19251 }, { "epoch": 0.6858934392646561, "grad_norm": 1.5386688598818639, "learning_rate": 4.74344277073597e-06, "loss": 0.528, "step": 19252 }, { "epoch": 0.6859290663911503, "grad_norm": 2.072175763227321, "learning_rate": 4.74246115479253e-06, "loss": 0.5024, "step": 19253 }, { "epoch": 0.6859646935176443, "grad_norm": 2.0268878874359797, "learning_rate": 4.741479608856898e-06, "loss": 0.5232, "step": 19254 }, { "epoch": 0.6860003206441384, "grad_norm": 1.1726359021809036, "learning_rate": 4.740498132942137e-06, "loss": 0.4234, "step": 19255 }, { "epoch": 0.6860359477706326, "grad_norm": 1.85162578520173, "learning_rate": 4.739516727061323e-06, "loss": 0.5054, "step": 19256 }, { "epoch": 0.6860715748971267, "grad_norm": 1.6533770700387391, "learning_rate": 4.738535391227517e-06, "loss": 0.5487, "step": 19257 }, { "epoch": 0.6861072020236207, "grad_norm": 1.4131305769362144, "learning_rate": 4.737554125453791e-06, "loss": 0.4142, "step": 19258 }, { "epoch": 0.6861428291501149, "grad_norm": 1.2318375483059492, "learning_rate": 4.736572929753208e-06, "loss": 0.4449, "step": 19259 }, { "epoch": 0.686178456276609, "grad_norm": 1.1490022092770633, "learning_rate": 4.73559180413884e-06, "loss": 0.4465, "step": 19260 }, { "epoch": 0.6862140834031031, "grad_norm": 1.8233042143033757, "learning_rate": 4.734610748623743e-06, "loss": 0.4953, "step": 19261 }, { "epoch": 0.6862497105295973, "grad_norm": 1.3380999838289855, "learning_rate": 4.733629763220988e-06, "loss": 0.4353, "step": 19262 }, { "epoch": 0.6862853376560913, "grad_norm": 1.636104887670962, "learning_rate": 4.7326488479436295e-06, "loss": 0.5033, "step": 19263 }, { "epoch": 0.6863209647825854, "grad_norm": 1.5906275207641531, "learning_rate": 4.7316680028047335e-06, "loss": 0.5281, "step": 19264 }, { "epoch": 0.6863565919090796, "grad_norm": 1.2997445925447642, "learning_rate": 4.730687227817365e-06, "loss": 0.3859, "step": 19265 }, { "epoch": 0.6863922190355737, "grad_norm": 1.2866698307947857, "learning_rate": 4.72970652299457e-06, "loss": 0.4293, "step": 19266 }, { "epoch": 0.6864278461620678, "grad_norm": 1.5445006472212597, "learning_rate": 4.728725888349425e-06, "loss": 0.4642, "step": 19267 }, { "epoch": 0.6864634732885619, "grad_norm": 1.4472806538176135, "learning_rate": 4.727745323894976e-06, "loss": 0.4305, "step": 19268 }, { "epoch": 0.686499100415056, "grad_norm": 1.3579236765079097, "learning_rate": 4.726764829644286e-06, "loss": 0.4034, "step": 19269 }, { "epoch": 0.6865347275415501, "grad_norm": 1.401107065226203, "learning_rate": 4.725784405610404e-06, "loss": 0.5644, "step": 19270 }, { "epoch": 0.6865703546680443, "grad_norm": 1.1920989575621583, "learning_rate": 4.724804051806393e-06, "loss": 0.4244, "step": 19271 }, { "epoch": 0.6866059817945384, "grad_norm": 1.454636361371869, "learning_rate": 4.723823768245299e-06, "loss": 0.4742, "step": 19272 }, { "epoch": 0.6866416089210324, "grad_norm": 1.490688024297593, "learning_rate": 4.7228435549401805e-06, "loss": 0.5276, "step": 19273 }, { "epoch": 0.6866772360475266, "grad_norm": 1.5510703514656645, "learning_rate": 4.721863411904087e-06, "loss": 0.485, "step": 19274 }, { "epoch": 0.6867128631740207, "grad_norm": 1.6165992249243064, "learning_rate": 4.720883339150073e-06, "loss": 0.6236, "step": 19275 }, { "epoch": 0.6867484903005148, "grad_norm": 1.1330237420788956, "learning_rate": 4.7199033366911905e-06, "loss": 0.4117, "step": 19276 }, { "epoch": 0.686784117427009, "grad_norm": 1.676040933246472, "learning_rate": 4.718923404540482e-06, "loss": 0.4559, "step": 19277 }, { "epoch": 0.686819744553503, "grad_norm": 1.4101890200671634, "learning_rate": 4.717943542711003e-06, "loss": 0.6012, "step": 19278 }, { "epoch": 0.6868553716799971, "grad_norm": 1.2752553756307379, "learning_rate": 4.716963751215794e-06, "loss": 0.4353, "step": 19279 }, { "epoch": 0.6868909988064913, "grad_norm": 1.5626742161498144, "learning_rate": 4.715984030067905e-06, "loss": 0.5086, "step": 19280 }, { "epoch": 0.6869266259329854, "grad_norm": 1.3598025297647398, "learning_rate": 4.715004379280382e-06, "loss": 0.4305, "step": 19281 }, { "epoch": 0.6869622530594794, "grad_norm": 1.3071372874734468, "learning_rate": 4.714024798866275e-06, "loss": 0.4456, "step": 19282 }, { "epoch": 0.6869978801859736, "grad_norm": 2.641322560793412, "learning_rate": 4.713045288838617e-06, "loss": 0.561, "step": 19283 }, { "epoch": 0.6870335073124677, "grad_norm": 1.2508209636012129, "learning_rate": 4.71206584921046e-06, "loss": 0.4687, "step": 19284 }, { "epoch": 0.6870691344389618, "grad_norm": 1.386316348144369, "learning_rate": 4.711086479994839e-06, "loss": 0.6023, "step": 19285 }, { "epoch": 0.687104761565456, "grad_norm": 1.7837262893446297, "learning_rate": 4.710107181204798e-06, "loss": 0.5037, "step": 19286 }, { "epoch": 0.68714038869195, "grad_norm": 1.6661275836849232, "learning_rate": 4.7091279528533816e-06, "loss": 0.5366, "step": 19287 }, { "epoch": 0.6871760158184441, "grad_norm": 1.3004211933848375, "learning_rate": 4.708148794953621e-06, "loss": 0.3949, "step": 19288 }, { "epoch": 0.6872116429449383, "grad_norm": 1.9154861381490846, "learning_rate": 4.707169707518558e-06, "loss": 0.4522, "step": 19289 }, { "epoch": 0.6872472700714324, "grad_norm": 1.3693565818941758, "learning_rate": 4.706190690561229e-06, "loss": 0.4663, "step": 19290 }, { "epoch": 0.6872828971979265, "grad_norm": 1.9693135475199584, "learning_rate": 4.705211744094674e-06, "loss": 0.5506, "step": 19291 }, { "epoch": 0.6873185243244206, "grad_norm": 1.5590871283699614, "learning_rate": 4.704232868131924e-06, "loss": 0.5116, "step": 19292 }, { "epoch": 0.6873541514509147, "grad_norm": 1.209757972549402, "learning_rate": 4.703254062686017e-06, "loss": 0.5731, "step": 19293 }, { "epoch": 0.6873897785774088, "grad_norm": 2.0047374294920126, "learning_rate": 4.7022753277699816e-06, "loss": 0.4543, "step": 19294 }, { "epoch": 0.687425405703903, "grad_norm": 1.2130436398416504, "learning_rate": 4.701296663396852e-06, "loss": 0.4221, "step": 19295 }, { "epoch": 0.6874610328303971, "grad_norm": 1.5903015728439682, "learning_rate": 4.700318069579663e-06, "loss": 0.4993, "step": 19296 }, { "epoch": 0.6874966599568911, "grad_norm": 1.2674670549244487, "learning_rate": 4.699339546331441e-06, "loss": 0.4441, "step": 19297 }, { "epoch": 0.6875322870833853, "grad_norm": 1.607220389875457, "learning_rate": 4.698361093665222e-06, "loss": 0.5694, "step": 19298 }, { "epoch": 0.6875679142098794, "grad_norm": 1.2833144954969835, "learning_rate": 4.697382711594027e-06, "loss": 0.4376, "step": 19299 }, { "epoch": 0.6876035413363735, "grad_norm": 1.6549984619347045, "learning_rate": 4.6964044001308915e-06, "loss": 0.5628, "step": 19300 }, { "epoch": 0.6876391684628677, "grad_norm": 1.463288870104069, "learning_rate": 4.695426159288835e-06, "loss": 0.4731, "step": 19301 }, { "epoch": 0.6876747955893617, "grad_norm": 1.5168416706954757, "learning_rate": 4.694447989080891e-06, "loss": 0.5212, "step": 19302 }, { "epoch": 0.6877104227158558, "grad_norm": 1.830150003487388, "learning_rate": 4.693469889520074e-06, "loss": 0.5374, "step": 19303 }, { "epoch": 0.68774604984235, "grad_norm": 1.2962455367260373, "learning_rate": 4.692491860619418e-06, "loss": 0.4136, "step": 19304 }, { "epoch": 0.6877816769688441, "grad_norm": 1.3761890132186148, "learning_rate": 4.691513902391941e-06, "loss": 0.5126, "step": 19305 }, { "epoch": 0.6878173040953381, "grad_norm": 1.4286658517306574, "learning_rate": 4.690536014850671e-06, "loss": 0.4604, "step": 19306 }, { "epoch": 0.6878529312218323, "grad_norm": 1.52862644207601, "learning_rate": 4.68955819800862e-06, "loss": 0.5149, "step": 19307 }, { "epoch": 0.6878885583483264, "grad_norm": 1.3231705476431745, "learning_rate": 4.688580451878818e-06, "loss": 0.4469, "step": 19308 }, { "epoch": 0.6879241854748205, "grad_norm": 1.7827831819396147, "learning_rate": 4.687602776474276e-06, "loss": 0.6236, "step": 19309 }, { "epoch": 0.6879598126013147, "grad_norm": 1.1283170278096815, "learning_rate": 4.686625171808016e-06, "loss": 0.4206, "step": 19310 }, { "epoch": 0.6879954397278087, "grad_norm": 1.9726920182480816, "learning_rate": 4.685647637893056e-06, "loss": 0.488, "step": 19311 }, { "epoch": 0.6880310668543028, "grad_norm": 1.8232150578158963, "learning_rate": 4.6846701747424115e-06, "loss": 0.4724, "step": 19312 }, { "epoch": 0.688066693980797, "grad_norm": 1.4180775631702975, "learning_rate": 4.683692782369104e-06, "loss": 0.51, "step": 19313 }, { "epoch": 0.6881023211072911, "grad_norm": 1.485154712302821, "learning_rate": 4.682715460786138e-06, "loss": 0.4827, "step": 19314 }, { "epoch": 0.6881379482337852, "grad_norm": 1.1689306663850405, "learning_rate": 4.681738210006536e-06, "loss": 0.4162, "step": 19315 }, { "epoch": 0.6881735753602793, "grad_norm": 1.239787088141253, "learning_rate": 4.680761030043303e-06, "loss": 0.4526, "step": 19316 }, { "epoch": 0.6882092024867734, "grad_norm": 1.618045283984827, "learning_rate": 4.67978392090946e-06, "loss": 0.4499, "step": 19317 }, { "epoch": 0.6882448296132675, "grad_norm": 1.5200840227082155, "learning_rate": 4.678806882618003e-06, "loss": 0.5955, "step": 19318 }, { "epoch": 0.6882804567397617, "grad_norm": 1.3753193263239907, "learning_rate": 4.677829915181961e-06, "loss": 0.4458, "step": 19319 }, { "epoch": 0.6883160838662558, "grad_norm": 1.2618403758449053, "learning_rate": 4.676853018614329e-06, "loss": 0.4756, "step": 19320 }, { "epoch": 0.6883517109927498, "grad_norm": 1.2503068994206201, "learning_rate": 4.675876192928121e-06, "loss": 0.3971, "step": 19321 }, { "epoch": 0.688387338119244, "grad_norm": 1.0799032393441508, "learning_rate": 4.674899438136346e-06, "loss": 0.4044, "step": 19322 }, { "epoch": 0.6884229652457381, "grad_norm": 1.2204189967259347, "learning_rate": 4.673922754252001e-06, "loss": 0.433, "step": 19323 }, { "epoch": 0.6884585923722322, "grad_norm": 1.4097355183063693, "learning_rate": 4.672946141288104e-06, "loss": 0.4299, "step": 19324 }, { "epoch": 0.6884942194987264, "grad_norm": 1.669859529768762, "learning_rate": 4.671969599257648e-06, "loss": 0.5873, "step": 19325 }, { "epoch": 0.6885298466252204, "grad_norm": 1.1829428602638863, "learning_rate": 4.6709931281736395e-06, "loss": 0.4276, "step": 19326 }, { "epoch": 0.6885654737517145, "grad_norm": 1.544815824822664, "learning_rate": 4.670016728049084e-06, "loss": 0.4488, "step": 19327 }, { "epoch": 0.6886011008782087, "grad_norm": 1.2686210957654687, "learning_rate": 4.669040398896983e-06, "loss": 0.3874, "step": 19328 }, { "epoch": 0.6886367280047028, "grad_norm": 1.5057641603239587, "learning_rate": 4.668064140730333e-06, "loss": 0.5757, "step": 19329 }, { "epoch": 0.6886723551311968, "grad_norm": 1.1262964551753354, "learning_rate": 4.66708795356214e-06, "loss": 0.3919, "step": 19330 }, { "epoch": 0.688707982257691, "grad_norm": 1.2669018893916697, "learning_rate": 4.666111837405392e-06, "loss": 0.4542, "step": 19331 }, { "epoch": 0.6887436093841851, "grad_norm": 1.4766656177176312, "learning_rate": 4.665135792273095e-06, "loss": 0.5514, "step": 19332 }, { "epoch": 0.6887792365106792, "grad_norm": 1.3676941151842796, "learning_rate": 4.664159818178244e-06, "loss": 0.4007, "step": 19333 }, { "epoch": 0.6888148636371734, "grad_norm": 1.1196377198935952, "learning_rate": 4.6631839151338345e-06, "loss": 0.4138, "step": 19334 }, { "epoch": 0.6888504907636674, "grad_norm": 1.4371518504390288, "learning_rate": 4.662208083152865e-06, "loss": 0.4452, "step": 19335 }, { "epoch": 0.6888861178901615, "grad_norm": 1.507842463653518, "learning_rate": 4.661232322248321e-06, "loss": 0.5473, "step": 19336 }, { "epoch": 0.6889217450166557, "grad_norm": 1.3061231172698466, "learning_rate": 4.660256632433205e-06, "loss": 0.4867, "step": 19337 }, { "epoch": 0.6889573721431498, "grad_norm": 1.1963157073809236, "learning_rate": 4.6592810137205e-06, "loss": 0.4092, "step": 19338 }, { "epoch": 0.6889929992696439, "grad_norm": 1.3706674088373973, "learning_rate": 4.658305466123205e-06, "loss": 0.5195, "step": 19339 }, { "epoch": 0.689028626396138, "grad_norm": 1.5339846913767488, "learning_rate": 4.657329989654303e-06, "loss": 0.5526, "step": 19340 }, { "epoch": 0.6890642535226321, "grad_norm": 1.2971971092080914, "learning_rate": 4.656354584326787e-06, "loss": 0.4275, "step": 19341 }, { "epoch": 0.6890998806491262, "grad_norm": 1.3469080159076217, "learning_rate": 4.6553792501536436e-06, "loss": 0.4358, "step": 19342 }, { "epoch": 0.6891355077756204, "grad_norm": 1.4540257325730845, "learning_rate": 4.654403987147865e-06, "loss": 0.5709, "step": 19343 }, { "epoch": 0.6891711349021145, "grad_norm": 1.753473206138269, "learning_rate": 4.65342879532243e-06, "loss": 0.4762, "step": 19344 }, { "epoch": 0.6892067620286085, "grad_norm": 1.8859203357794603, "learning_rate": 4.652453674690328e-06, "loss": 0.5506, "step": 19345 }, { "epoch": 0.6892423891551027, "grad_norm": 1.1989882296084986, "learning_rate": 4.651478625264547e-06, "loss": 0.464, "step": 19346 }, { "epoch": 0.6892780162815968, "grad_norm": 1.4436150023024434, "learning_rate": 4.650503647058062e-06, "loss": 0.5133, "step": 19347 }, { "epoch": 0.6893136434080909, "grad_norm": 1.3344629602903912, "learning_rate": 4.64952874008386e-06, "loss": 0.3911, "step": 19348 }, { "epoch": 0.6893492705345851, "grad_norm": 1.7019820936210739, "learning_rate": 4.648553904354923e-06, "loss": 0.5341, "step": 19349 }, { "epoch": 0.6893848976610791, "grad_norm": 1.257384832718042, "learning_rate": 4.6475791398842355e-06, "loss": 0.4175, "step": 19350 }, { "epoch": 0.6894205247875732, "grad_norm": 1.3191131813436727, "learning_rate": 4.6466044466847685e-06, "loss": 0.4917, "step": 19351 }, { "epoch": 0.6894561519140674, "grad_norm": 1.2486386990460125, "learning_rate": 4.645629824769509e-06, "loss": 0.4223, "step": 19352 }, { "epoch": 0.6894917790405615, "grad_norm": 1.7105753384760956, "learning_rate": 4.6446552741514266e-06, "loss": 0.4444, "step": 19353 }, { "epoch": 0.6895274061670555, "grad_norm": 1.7077656732987052, "learning_rate": 4.643680794843506e-06, "loss": 0.5133, "step": 19354 }, { "epoch": 0.6895630332935497, "grad_norm": 1.1719857838697347, "learning_rate": 4.642706386858713e-06, "loss": 0.3806, "step": 19355 }, { "epoch": 0.6895986604200438, "grad_norm": 1.1094469755455474, "learning_rate": 4.641732050210032e-06, "loss": 0.4214, "step": 19356 }, { "epoch": 0.6896342875465379, "grad_norm": 1.4168164896845674, "learning_rate": 4.640757784910438e-06, "loss": 0.4918, "step": 19357 }, { "epoch": 0.6896699146730321, "grad_norm": 1.2586830625140666, "learning_rate": 4.639783590972896e-06, "loss": 0.4347, "step": 19358 }, { "epoch": 0.6897055417995261, "grad_norm": 1.5476402570309364, "learning_rate": 4.638809468410385e-06, "loss": 0.5353, "step": 19359 }, { "epoch": 0.6897411689260202, "grad_norm": 1.4237199985386935, "learning_rate": 4.637835417235871e-06, "loss": 0.5508, "step": 19360 }, { "epoch": 0.6897767960525144, "grad_norm": 1.254804716864461, "learning_rate": 4.63686143746233e-06, "loss": 0.4269, "step": 19361 }, { "epoch": 0.6898124231790085, "grad_norm": 1.1049868730276449, "learning_rate": 4.635887529102723e-06, "loss": 0.4236, "step": 19362 }, { "epoch": 0.6898480503055026, "grad_norm": 1.5109425359044717, "learning_rate": 4.634913692170024e-06, "loss": 0.5881, "step": 19363 }, { "epoch": 0.6898836774319967, "grad_norm": 1.0802548027425456, "learning_rate": 4.633939926677197e-06, "loss": 0.3935, "step": 19364 }, { "epoch": 0.6899193045584908, "grad_norm": 1.096432828054446, "learning_rate": 4.632966232637216e-06, "loss": 0.3699, "step": 19365 }, { "epoch": 0.6899549316849849, "grad_norm": 1.45623742009012, "learning_rate": 4.631992610063036e-06, "loss": 0.5203, "step": 19366 }, { "epoch": 0.6899905588114791, "grad_norm": 1.3765235925107677, "learning_rate": 4.631019058967627e-06, "loss": 0.5921, "step": 19367 }, { "epoch": 0.6900261859379732, "grad_norm": 1.4004178312538043, "learning_rate": 4.6300455793639565e-06, "loss": 0.5503, "step": 19368 }, { "epoch": 0.6900618130644672, "grad_norm": 1.199090005943257, "learning_rate": 4.6290721712649776e-06, "loss": 0.4641, "step": 19369 }, { "epoch": 0.6900974401909614, "grad_norm": 1.1128926413025302, "learning_rate": 4.628098834683658e-06, "loss": 0.4868, "step": 19370 }, { "epoch": 0.6901330673174555, "grad_norm": 1.1463558122533166, "learning_rate": 4.6271255696329556e-06, "loss": 0.406, "step": 19371 }, { "epoch": 0.6901686944439496, "grad_norm": 1.2881625721418375, "learning_rate": 4.626152376125837e-06, "loss": 0.5538, "step": 19372 }, { "epoch": 0.6902043215704438, "grad_norm": 1.165356884671908, "learning_rate": 4.625179254175251e-06, "loss": 0.4364, "step": 19373 }, { "epoch": 0.6902399486969378, "grad_norm": 1.6552449685426385, "learning_rate": 4.624206203794164e-06, "loss": 0.5603, "step": 19374 }, { "epoch": 0.6902755758234319, "grad_norm": 1.2862308797886857, "learning_rate": 4.623233224995524e-06, "loss": 0.4065, "step": 19375 }, { "epoch": 0.6903112029499261, "grad_norm": 1.196581742481208, "learning_rate": 4.6222603177922965e-06, "loss": 0.3986, "step": 19376 }, { "epoch": 0.6903468300764202, "grad_norm": 1.245198913052675, "learning_rate": 4.6212874821974275e-06, "loss": 0.4228, "step": 19377 }, { "epoch": 0.6903824572029142, "grad_norm": 1.4791312893245874, "learning_rate": 4.620314718223876e-06, "loss": 0.6003, "step": 19378 }, { "epoch": 0.6904180843294084, "grad_norm": 1.1260682030533793, "learning_rate": 4.619342025884593e-06, "loss": 0.4317, "step": 19379 }, { "epoch": 0.6904537114559025, "grad_norm": 1.1882402357255342, "learning_rate": 4.6183694051925335e-06, "loss": 0.3819, "step": 19380 }, { "epoch": 0.6904893385823966, "grad_norm": 1.0765269862988005, "learning_rate": 4.61739685616065e-06, "loss": 0.3751, "step": 19381 }, { "epoch": 0.6905249657088908, "grad_norm": 1.6724629710824104, "learning_rate": 4.616424378801885e-06, "loss": 0.439, "step": 19382 }, { "epoch": 0.6905605928353848, "grad_norm": 1.2705762194131711, "learning_rate": 4.615451973129196e-06, "loss": 0.3956, "step": 19383 }, { "epoch": 0.6905962199618789, "grad_norm": 1.3682115850596552, "learning_rate": 4.6144796391555245e-06, "loss": 0.4199, "step": 19384 }, { "epoch": 0.6906318470883731, "grad_norm": 1.8520587149979941, "learning_rate": 4.613507376893821e-06, "loss": 0.5772, "step": 19385 }, { "epoch": 0.6906674742148672, "grad_norm": 1.3953533221210228, "learning_rate": 4.612535186357031e-06, "loss": 0.4554, "step": 19386 }, { "epoch": 0.6907031013413613, "grad_norm": 1.102540634294765, "learning_rate": 4.611563067558105e-06, "loss": 0.403, "step": 19387 }, { "epoch": 0.6907387284678554, "grad_norm": 1.5710145551095462, "learning_rate": 4.6105910205099795e-06, "loss": 0.5054, "step": 19388 }, { "epoch": 0.6907743555943495, "grad_norm": 1.7841696768294961, "learning_rate": 4.609619045225604e-06, "loss": 0.5824, "step": 19389 }, { "epoch": 0.6908099827208436, "grad_norm": 1.1475094737307543, "learning_rate": 4.6086471417179155e-06, "loss": 0.3849, "step": 19390 }, { "epoch": 0.6908456098473378, "grad_norm": 2.298006485447611, "learning_rate": 4.60767530999986e-06, "loss": 0.4789, "step": 19391 }, { "epoch": 0.6908812369738319, "grad_norm": 1.2509559343647494, "learning_rate": 4.606703550084378e-06, "loss": 0.4219, "step": 19392 }, { "epoch": 0.6909168641003259, "grad_norm": 1.2071560629496798, "learning_rate": 4.6057318619844014e-06, "loss": 0.4323, "step": 19393 }, { "epoch": 0.6909524912268201, "grad_norm": 1.5519320507113246, "learning_rate": 4.604760245712883e-06, "loss": 0.5377, "step": 19394 }, { "epoch": 0.6909881183533142, "grad_norm": 2.26318341208478, "learning_rate": 4.60378870128275e-06, "loss": 0.4302, "step": 19395 }, { "epoch": 0.6910237454798083, "grad_norm": 1.4653427138840256, "learning_rate": 4.602817228706945e-06, "loss": 0.4815, "step": 19396 }, { "epoch": 0.6910593726063025, "grad_norm": 1.2165403662274337, "learning_rate": 4.601845827998398e-06, "loss": 0.432, "step": 19397 }, { "epoch": 0.6910949997327965, "grad_norm": 1.5367435179370672, "learning_rate": 4.6008744991700505e-06, "loss": 0.565, "step": 19398 }, { "epoch": 0.6911306268592906, "grad_norm": 1.151332811919466, "learning_rate": 4.599903242234829e-06, "loss": 0.4091, "step": 19399 }, { "epoch": 0.6911662539857848, "grad_norm": 1.5719279087125986, "learning_rate": 4.59893205720567e-06, "loss": 0.5053, "step": 19400 }, { "epoch": 0.6912018811122789, "grad_norm": 1.1717110683891656, "learning_rate": 4.597960944095505e-06, "loss": 0.4056, "step": 19401 }, { "epoch": 0.691237508238773, "grad_norm": 1.5120963831440128, "learning_rate": 4.5969899029172674e-06, "loss": 0.6057, "step": 19402 }, { "epoch": 0.6912731353652671, "grad_norm": 1.3755193060420947, "learning_rate": 4.596018933683889e-06, "loss": 0.5082, "step": 19403 }, { "epoch": 0.6913087624917612, "grad_norm": 1.1579168066746464, "learning_rate": 4.595048036408291e-06, "loss": 0.3654, "step": 19404 }, { "epoch": 0.6913443896182554, "grad_norm": 1.150484095344685, "learning_rate": 4.59407721110341e-06, "loss": 0.378, "step": 19405 }, { "epoch": 0.6913800167447495, "grad_norm": 1.1993079362304566, "learning_rate": 4.593106457782167e-06, "loss": 0.4201, "step": 19406 }, { "epoch": 0.6914156438712435, "grad_norm": 1.3827973669555194, "learning_rate": 4.59213577645749e-06, "loss": 0.4334, "step": 19407 }, { "epoch": 0.6914512709977377, "grad_norm": 2.693384575086167, "learning_rate": 4.5911651671423055e-06, "loss": 0.5228, "step": 19408 }, { "epoch": 0.6914868981242318, "grad_norm": 1.2315780625597046, "learning_rate": 4.590194629849541e-06, "loss": 0.391, "step": 19409 }, { "epoch": 0.6915225252507259, "grad_norm": 2.1642754823496526, "learning_rate": 4.5892241645921125e-06, "loss": 0.5245, "step": 19410 }, { "epoch": 0.6915581523772201, "grad_norm": 1.0624487749966367, "learning_rate": 4.5882537713829486e-06, "loss": 0.4269, "step": 19411 }, { "epoch": 0.6915937795037141, "grad_norm": 1.306768444761063, "learning_rate": 4.587283450234965e-06, "loss": 0.447, "step": 19412 }, { "epoch": 0.6916294066302082, "grad_norm": 1.408376672716367, "learning_rate": 4.5863132011610866e-06, "loss": 0.3998, "step": 19413 }, { "epoch": 0.6916650337567024, "grad_norm": 1.123385976877988, "learning_rate": 4.585343024174235e-06, "loss": 0.4293, "step": 19414 }, { "epoch": 0.6917006608831965, "grad_norm": 1.0909147339559364, "learning_rate": 4.584372919287322e-06, "loss": 0.4198, "step": 19415 }, { "epoch": 0.6917362880096906, "grad_norm": 2.239643931262492, "learning_rate": 4.583402886513269e-06, "loss": 0.5404, "step": 19416 }, { "epoch": 0.6917719151361847, "grad_norm": 1.2549976618380019, "learning_rate": 4.582432925864993e-06, "loss": 0.3943, "step": 19417 }, { "epoch": 0.6918075422626788, "grad_norm": 1.6288627327367515, "learning_rate": 4.581463037355411e-06, "loss": 0.5061, "step": 19418 }, { "epoch": 0.6918431693891729, "grad_norm": 1.8665816720540398, "learning_rate": 4.580493220997434e-06, "loss": 0.5769, "step": 19419 }, { "epoch": 0.6918787965156671, "grad_norm": 1.4415890236019024, "learning_rate": 4.579523476803981e-06, "loss": 0.5285, "step": 19420 }, { "epoch": 0.6919144236421612, "grad_norm": 1.3773516557728698, "learning_rate": 4.5785538047879586e-06, "loss": 0.5492, "step": 19421 }, { "epoch": 0.6919500507686552, "grad_norm": 1.2115757967218723, "learning_rate": 4.577584204962281e-06, "loss": 0.4459, "step": 19422 }, { "epoch": 0.6919856778951494, "grad_norm": 2.0692662759983547, "learning_rate": 4.576614677339859e-06, "loss": 0.5459, "step": 19423 }, { "epoch": 0.6920213050216435, "grad_norm": 1.1598837328420113, "learning_rate": 4.575645221933604e-06, "loss": 0.3851, "step": 19424 }, { "epoch": 0.6920569321481376, "grad_norm": 1.6231469696332215, "learning_rate": 4.574675838756429e-06, "loss": 0.5976, "step": 19425 }, { "epoch": 0.6920925592746318, "grad_norm": 1.0932706248447537, "learning_rate": 4.573706527821232e-06, "loss": 0.405, "step": 19426 }, { "epoch": 0.6921281864011258, "grad_norm": 1.3789863512058593, "learning_rate": 4.57273728914093e-06, "loss": 0.4608, "step": 19427 }, { "epoch": 0.6921638135276199, "grad_norm": 1.200240252790791, "learning_rate": 4.571768122728421e-06, "loss": 0.4208, "step": 19428 }, { "epoch": 0.6921994406541141, "grad_norm": 1.3503886573891652, "learning_rate": 4.570799028596617e-06, "loss": 0.4599, "step": 19429 }, { "epoch": 0.6922350677806082, "grad_norm": 1.5518245350036242, "learning_rate": 4.56983000675841e-06, "loss": 0.5519, "step": 19430 }, { "epoch": 0.6922706949071022, "grad_norm": 1.194990045802927, "learning_rate": 4.56886105722672e-06, "loss": 0.3905, "step": 19431 }, { "epoch": 0.6923063220335964, "grad_norm": 1.1729717595521558, "learning_rate": 4.5678921800144385e-06, "loss": 0.3676, "step": 19432 }, { "epoch": 0.6923419491600905, "grad_norm": 1.3368333198258389, "learning_rate": 4.5669233751344725e-06, "loss": 0.4468, "step": 19433 }, { "epoch": 0.6923775762865846, "grad_norm": 1.8373549927163146, "learning_rate": 4.565954642599715e-06, "loss": 0.5255, "step": 19434 }, { "epoch": 0.6924132034130788, "grad_norm": 1.2394163988172953, "learning_rate": 4.564985982423073e-06, "loss": 0.4271, "step": 19435 }, { "epoch": 0.6924488305395728, "grad_norm": 1.2440184731155806, "learning_rate": 4.56401739461744e-06, "loss": 0.3691, "step": 19436 }, { "epoch": 0.6924844576660669, "grad_norm": 1.5755558863441144, "learning_rate": 4.563048879195713e-06, "loss": 0.4623, "step": 19437 }, { "epoch": 0.6925200847925611, "grad_norm": 1.7207356196910621, "learning_rate": 4.56208043617079e-06, "loss": 0.4596, "step": 19438 }, { "epoch": 0.6925557119190552, "grad_norm": 1.5720165088499845, "learning_rate": 4.5611120655555675e-06, "loss": 0.5113, "step": 19439 }, { "epoch": 0.6925913390455493, "grad_norm": 1.7770164421875838, "learning_rate": 4.560143767362943e-06, "loss": 0.5956, "step": 19440 }, { "epoch": 0.6926269661720434, "grad_norm": 1.1250634248130231, "learning_rate": 4.5591755416058035e-06, "loss": 0.4344, "step": 19441 }, { "epoch": 0.6926625932985375, "grad_norm": 1.5916197671414256, "learning_rate": 4.558207388297048e-06, "loss": 0.5103, "step": 19442 }, { "epoch": 0.6926982204250316, "grad_norm": 1.324312956797372, "learning_rate": 4.557239307449562e-06, "loss": 0.4012, "step": 19443 }, { "epoch": 0.6927338475515258, "grad_norm": 1.2128256070800163, "learning_rate": 4.556271299076237e-06, "loss": 0.44, "step": 19444 }, { "epoch": 0.6927694746780199, "grad_norm": 1.9462700437051637, "learning_rate": 4.555303363189965e-06, "loss": 0.5485, "step": 19445 }, { "epoch": 0.6928051018045139, "grad_norm": 1.5059543017801162, "learning_rate": 4.5543354998036385e-06, "loss": 0.5531, "step": 19446 }, { "epoch": 0.6928407289310081, "grad_norm": 1.7207999251297146, "learning_rate": 4.5533677089301375e-06, "loss": 0.5177, "step": 19447 }, { "epoch": 0.6928763560575022, "grad_norm": 1.3270650173332181, "learning_rate": 4.552399990582351e-06, "loss": 0.4482, "step": 19448 }, { "epoch": 0.6929119831839963, "grad_norm": 1.206527863711129, "learning_rate": 4.551432344773172e-06, "loss": 0.4105, "step": 19449 }, { "epoch": 0.6929476103104905, "grad_norm": 1.2662821344008892, "learning_rate": 4.550464771515475e-06, "loss": 0.4266, "step": 19450 }, { "epoch": 0.6929832374369845, "grad_norm": 1.5099383359820053, "learning_rate": 4.549497270822153e-06, "loss": 0.5993, "step": 19451 }, { "epoch": 0.6930188645634786, "grad_norm": 1.3150149957611543, "learning_rate": 4.54852984270608e-06, "loss": 0.5382, "step": 19452 }, { "epoch": 0.6930544916899728, "grad_norm": 1.9631139216118454, "learning_rate": 4.547562487180143e-06, "loss": 0.5437, "step": 19453 }, { "epoch": 0.6930901188164669, "grad_norm": 1.2873514990102182, "learning_rate": 4.546595204257222e-06, "loss": 0.4027, "step": 19454 }, { "epoch": 0.693125745942961, "grad_norm": 1.4332031393879396, "learning_rate": 4.545627993950201e-06, "loss": 0.4416, "step": 19455 }, { "epoch": 0.6931613730694551, "grad_norm": 1.597232656707073, "learning_rate": 4.544660856271952e-06, "loss": 0.5363, "step": 19456 }, { "epoch": 0.6931970001959492, "grad_norm": 1.2768479862974398, "learning_rate": 4.543693791235361e-06, "loss": 0.4273, "step": 19457 }, { "epoch": 0.6932326273224433, "grad_norm": 1.9793248306697162, "learning_rate": 4.542726798853297e-06, "loss": 0.5025, "step": 19458 }, { "epoch": 0.6932682544489375, "grad_norm": 1.6239875104231636, "learning_rate": 4.541759879138641e-06, "loss": 0.44, "step": 19459 }, { "epoch": 0.6933038815754315, "grad_norm": 1.723554965420389, "learning_rate": 4.540793032104266e-06, "loss": 0.6062, "step": 19460 }, { "epoch": 0.6933395087019256, "grad_norm": 1.3672104983664093, "learning_rate": 4.539826257763048e-06, "loss": 0.4028, "step": 19461 }, { "epoch": 0.6933751358284198, "grad_norm": 1.203072941191984, "learning_rate": 4.538859556127864e-06, "loss": 0.4205, "step": 19462 }, { "epoch": 0.6934107629549139, "grad_norm": 1.1657452713262375, "learning_rate": 4.537892927211577e-06, "loss": 0.393, "step": 19463 }, { "epoch": 0.693446390081408, "grad_norm": 1.2983619339501202, "learning_rate": 4.536926371027068e-06, "loss": 0.3819, "step": 19464 }, { "epoch": 0.6934820172079021, "grad_norm": 1.3247406930678267, "learning_rate": 4.535959887587199e-06, "loss": 0.4318, "step": 19465 }, { "epoch": 0.6935176443343962, "grad_norm": 1.6582166034849364, "learning_rate": 4.5349934769048474e-06, "loss": 0.4681, "step": 19466 }, { "epoch": 0.6935532714608903, "grad_norm": 1.5444260283727225, "learning_rate": 4.5340271389928735e-06, "loss": 0.5467, "step": 19467 }, { "epoch": 0.6935888985873845, "grad_norm": 2.1057245162957012, "learning_rate": 4.5330608738641495e-06, "loss": 0.5421, "step": 19468 }, { "epoch": 0.6936245257138786, "grad_norm": 1.6137795188945625, "learning_rate": 4.532094681531539e-06, "loss": 0.4921, "step": 19469 }, { "epoch": 0.6936601528403726, "grad_norm": 1.4471496276399276, "learning_rate": 4.5311285620079144e-06, "loss": 0.4897, "step": 19470 }, { "epoch": 0.6936957799668668, "grad_norm": 1.865290172436636, "learning_rate": 4.530162515306132e-06, "loss": 0.5612, "step": 19471 }, { "epoch": 0.6937314070933609, "grad_norm": 2.1695561535369294, "learning_rate": 4.529196541439059e-06, "loss": 0.5622, "step": 19472 }, { "epoch": 0.693767034219855, "grad_norm": 1.3539965197367856, "learning_rate": 4.528230640419562e-06, "loss": 0.4708, "step": 19473 }, { "epoch": 0.6938026613463492, "grad_norm": 1.2919557196384617, "learning_rate": 4.5272648122604945e-06, "loss": 0.4201, "step": 19474 }, { "epoch": 0.6938382884728432, "grad_norm": 1.3986539043363047, "learning_rate": 4.526299056974721e-06, "loss": 0.4001, "step": 19475 }, { "epoch": 0.6938739155993373, "grad_norm": 1.7104396503624806, "learning_rate": 4.525333374575103e-06, "loss": 0.5171, "step": 19476 }, { "epoch": 0.6939095427258315, "grad_norm": 1.5483157484666252, "learning_rate": 4.524367765074499e-06, "loss": 0.5646, "step": 19477 }, { "epoch": 0.6939451698523256, "grad_norm": 1.5436591718793151, "learning_rate": 4.523402228485763e-06, "loss": 0.5033, "step": 19478 }, { "epoch": 0.6939807969788196, "grad_norm": 2.174052752922234, "learning_rate": 4.522436764821758e-06, "loss": 0.4445, "step": 19479 }, { "epoch": 0.6940164241053138, "grad_norm": 1.6213004732369476, "learning_rate": 4.521471374095332e-06, "loss": 0.4316, "step": 19480 }, { "epoch": 0.6940520512318079, "grad_norm": 1.3066686409436943, "learning_rate": 4.520506056319347e-06, "loss": 0.4114, "step": 19481 }, { "epoch": 0.694087678358302, "grad_norm": 1.6985564647019544, "learning_rate": 4.519540811506647e-06, "loss": 0.4426, "step": 19482 }, { "epoch": 0.6941233054847962, "grad_norm": 1.495553983501162, "learning_rate": 4.518575639670097e-06, "loss": 0.4973, "step": 19483 }, { "epoch": 0.6941589326112902, "grad_norm": 1.3657183287601433, "learning_rate": 4.517610540822545e-06, "loss": 0.4487, "step": 19484 }, { "epoch": 0.6941945597377843, "grad_norm": 1.402974810078478, "learning_rate": 4.516645514976837e-06, "loss": 0.4285, "step": 19485 }, { "epoch": 0.6942301868642785, "grad_norm": 1.2908773350939349, "learning_rate": 4.515680562145832e-06, "loss": 0.3873, "step": 19486 }, { "epoch": 0.6942658139907726, "grad_norm": 1.6812298530603493, "learning_rate": 4.514715682342369e-06, "loss": 0.5508, "step": 19487 }, { "epoch": 0.6943014411172667, "grad_norm": 1.6238775175267721, "learning_rate": 4.513750875579303e-06, "loss": 0.4193, "step": 19488 }, { "epoch": 0.6943370682437608, "grad_norm": 1.549362632139132, "learning_rate": 4.5127861418694775e-06, "loss": 0.4859, "step": 19489 }, { "epoch": 0.6943726953702549, "grad_norm": 1.6383514575922733, "learning_rate": 4.5118214812257386e-06, "loss": 0.4178, "step": 19490 }, { "epoch": 0.694408322496749, "grad_norm": 1.3201913795031532, "learning_rate": 4.5108568936609336e-06, "loss": 0.462, "step": 19491 }, { "epoch": 0.6944439496232432, "grad_norm": 1.863951526895356, "learning_rate": 4.509892379187908e-06, "loss": 0.5224, "step": 19492 }, { "epoch": 0.6944795767497373, "grad_norm": 1.7156839186243595, "learning_rate": 4.5089279378194995e-06, "loss": 0.5998, "step": 19493 }, { "epoch": 0.6945152038762313, "grad_norm": 1.7035389181319571, "learning_rate": 4.507963569568554e-06, "loss": 0.5257, "step": 19494 }, { "epoch": 0.6945508310027255, "grad_norm": 1.3362144929838822, "learning_rate": 4.506999274447915e-06, "loss": 0.4499, "step": 19495 }, { "epoch": 0.6945864581292196, "grad_norm": 2.151246834676312, "learning_rate": 4.506035052470418e-06, "loss": 0.4833, "step": 19496 }, { "epoch": 0.6946220852557137, "grad_norm": 1.3427572632711355, "learning_rate": 4.505070903648903e-06, "loss": 0.4414, "step": 19497 }, { "epoch": 0.6946577123822079, "grad_norm": 1.698370762545723, "learning_rate": 4.50410682799621e-06, "loss": 0.5263, "step": 19498 }, { "epoch": 0.6946933395087019, "grad_norm": 1.2101881485061534, "learning_rate": 4.50314282552518e-06, "loss": 0.4694, "step": 19499 }, { "epoch": 0.694728966635196, "grad_norm": 1.2178923306803597, "learning_rate": 4.502178896248641e-06, "loss": 0.4419, "step": 19500 }, { "epoch": 0.6947645937616902, "grad_norm": 1.7340027894464383, "learning_rate": 4.501215040179438e-06, "loss": 0.5298, "step": 19501 }, { "epoch": 0.6948002208881843, "grad_norm": 1.5128754383648455, "learning_rate": 4.500251257330395e-06, "loss": 0.4649, "step": 19502 }, { "epoch": 0.6948358480146783, "grad_norm": 2.6408643169724835, "learning_rate": 4.499287547714354e-06, "loss": 0.5219, "step": 19503 }, { "epoch": 0.6948714751411725, "grad_norm": 1.2125686401704183, "learning_rate": 4.498323911344142e-06, "loss": 0.3916, "step": 19504 }, { "epoch": 0.6949071022676666, "grad_norm": 1.5843751100839922, "learning_rate": 4.497360348232587e-06, "loss": 0.5585, "step": 19505 }, { "epoch": 0.6949427293941607, "grad_norm": 1.1998044089788897, "learning_rate": 4.496396858392533e-06, "loss": 0.4296, "step": 19506 }, { "epoch": 0.6949783565206549, "grad_norm": 1.4705669160235277, "learning_rate": 4.495433441836798e-06, "loss": 0.4077, "step": 19507 }, { "epoch": 0.695013983647149, "grad_norm": 1.5002745816335488, "learning_rate": 4.494470098578218e-06, "loss": 0.4848, "step": 19508 }, { "epoch": 0.695049610773643, "grad_norm": 1.3329689145565184, "learning_rate": 4.493506828629613e-06, "loss": 0.4088, "step": 19509 }, { "epoch": 0.6950852379001372, "grad_norm": 1.4032908985116732, "learning_rate": 4.492543632003817e-06, "loss": 0.4875, "step": 19510 }, { "epoch": 0.6951208650266313, "grad_norm": 1.5174044108763027, "learning_rate": 4.4915805087136485e-06, "loss": 0.5409, "step": 19511 }, { "epoch": 0.6951564921531254, "grad_norm": 1.13141237960798, "learning_rate": 4.490617458771936e-06, "loss": 0.4385, "step": 19512 }, { "epoch": 0.6951921192796195, "grad_norm": 1.2248153110348043, "learning_rate": 4.489654482191503e-06, "loss": 0.4323, "step": 19513 }, { "epoch": 0.6952277464061136, "grad_norm": 1.5882706832428664, "learning_rate": 4.488691578985176e-06, "loss": 0.528, "step": 19514 }, { "epoch": 0.6952633735326077, "grad_norm": 2.0726111283356587, "learning_rate": 4.487728749165768e-06, "loss": 0.5295, "step": 19515 }, { "epoch": 0.6952990006591019, "grad_norm": 1.8916289025267228, "learning_rate": 4.486765992746108e-06, "loss": 0.5639, "step": 19516 }, { "epoch": 0.695334627785596, "grad_norm": 2.7992517574877, "learning_rate": 4.4858033097390095e-06, "loss": 0.3893, "step": 19517 }, { "epoch": 0.69537025491209, "grad_norm": 2.5162403592021008, "learning_rate": 4.4848407001572945e-06, "loss": 0.4876, "step": 19518 }, { "epoch": 0.6954058820385842, "grad_norm": 1.9995901362520336, "learning_rate": 4.48387816401378e-06, "loss": 0.4732, "step": 19519 }, { "epoch": 0.6954415091650783, "grad_norm": 1.8252342573393763, "learning_rate": 4.482915701321283e-06, "loss": 0.5661, "step": 19520 }, { "epoch": 0.6954771362915724, "grad_norm": 1.1611762543261603, "learning_rate": 4.4819533120926236e-06, "loss": 0.4157, "step": 19521 }, { "epoch": 0.6955127634180666, "grad_norm": 3.0602175186712195, "learning_rate": 4.48099099634061e-06, "loss": 0.5093, "step": 19522 }, { "epoch": 0.6955483905445606, "grad_norm": 1.352003883530561, "learning_rate": 4.480028754078061e-06, "loss": 0.4444, "step": 19523 }, { "epoch": 0.6955840176710547, "grad_norm": 1.3943255445982115, "learning_rate": 4.479066585317784e-06, "loss": 0.405, "step": 19524 }, { "epoch": 0.6956196447975489, "grad_norm": 1.837937476026814, "learning_rate": 4.478104490072598e-06, "loss": 0.5002, "step": 19525 }, { "epoch": 0.695655271924043, "grad_norm": 1.310833022729856, "learning_rate": 4.477142468355307e-06, "loss": 0.3943, "step": 19526 }, { "epoch": 0.695690899050537, "grad_norm": 1.7568859116480542, "learning_rate": 4.476180520178725e-06, "loss": 0.565, "step": 19527 }, { "epoch": 0.6957265261770312, "grad_norm": 1.6278356925959447, "learning_rate": 4.475218645555659e-06, "loss": 0.5346, "step": 19528 }, { "epoch": 0.6957621533035253, "grad_norm": 2.7565636579919337, "learning_rate": 4.474256844498918e-06, "loss": 0.6133, "step": 19529 }, { "epoch": 0.6957977804300194, "grad_norm": 2.047558376058399, "learning_rate": 4.473295117021314e-06, "loss": 0.5044, "step": 19530 }, { "epoch": 0.6958334075565136, "grad_norm": 1.324183777169559, "learning_rate": 4.472333463135644e-06, "loss": 0.3725, "step": 19531 }, { "epoch": 0.6958690346830076, "grad_norm": 1.5509098303694369, "learning_rate": 4.471371882854723e-06, "loss": 0.4721, "step": 19532 }, { "epoch": 0.6959046618095017, "grad_norm": 1.3480073033212445, "learning_rate": 4.470410376191344e-06, "loss": 0.4373, "step": 19533 }, { "epoch": 0.6959402889359959, "grad_norm": 1.4064352197853904, "learning_rate": 4.4694489431583165e-06, "loss": 0.4432, "step": 19534 }, { "epoch": 0.69597591606249, "grad_norm": 1.6114591665583864, "learning_rate": 4.468487583768442e-06, "loss": 0.4156, "step": 19535 }, { "epoch": 0.6960115431889841, "grad_norm": 1.627907321688586, "learning_rate": 4.467526298034525e-06, "loss": 0.5718, "step": 19536 }, { "epoch": 0.6960471703154782, "grad_norm": 1.3119854883190436, "learning_rate": 4.466565085969359e-06, "loss": 0.4308, "step": 19537 }, { "epoch": 0.6960827974419723, "grad_norm": 1.2775557099354435, "learning_rate": 4.4656039475857505e-06, "loss": 0.4115, "step": 19538 }, { "epoch": 0.6961184245684664, "grad_norm": 1.5924947171592223, "learning_rate": 4.464642882896489e-06, "loss": 0.488, "step": 19539 }, { "epoch": 0.6961540516949606, "grad_norm": 2.8130025518815023, "learning_rate": 4.463681891914378e-06, "loss": 0.6248, "step": 19540 }, { "epoch": 0.6961896788214547, "grad_norm": 1.5896837340345795, "learning_rate": 4.462720974652215e-06, "loss": 0.4972, "step": 19541 }, { "epoch": 0.6962253059479487, "grad_norm": 1.3259308941819117, "learning_rate": 4.461760131122789e-06, "loss": 0.4331, "step": 19542 }, { "epoch": 0.6962609330744429, "grad_norm": 2.2521159035039493, "learning_rate": 4.460799361338898e-06, "loss": 0.6253, "step": 19543 }, { "epoch": 0.696296560200937, "grad_norm": 1.365815740958066, "learning_rate": 4.459838665313335e-06, "loss": 0.4209, "step": 19544 }, { "epoch": 0.6963321873274311, "grad_norm": 1.9159040325922498, "learning_rate": 4.458878043058897e-06, "loss": 0.6357, "step": 19545 }, { "epoch": 0.6963678144539253, "grad_norm": 1.7608391977351516, "learning_rate": 4.4579174945883654e-06, "loss": 0.4991, "step": 19546 }, { "epoch": 0.6964034415804193, "grad_norm": 1.6641912297246133, "learning_rate": 4.4569570199145405e-06, "loss": 0.5263, "step": 19547 }, { "epoch": 0.6964390687069134, "grad_norm": 1.6609872653676532, "learning_rate": 4.455996619050204e-06, "loss": 0.4843, "step": 19548 }, { "epoch": 0.6964746958334076, "grad_norm": 1.197028384581701, "learning_rate": 4.455036292008148e-06, "loss": 0.4402, "step": 19549 }, { "epoch": 0.6965103229599017, "grad_norm": 1.286930445661568, "learning_rate": 4.4540760388011584e-06, "loss": 0.4371, "step": 19550 }, { "epoch": 0.6965459500863957, "grad_norm": 1.3353482264357155, "learning_rate": 4.453115859442023e-06, "loss": 0.431, "step": 19551 }, { "epoch": 0.6965815772128899, "grad_norm": 1.1918490090398488, "learning_rate": 4.452155753943531e-06, "loss": 0.4392, "step": 19552 }, { "epoch": 0.696617204339384, "grad_norm": 1.874099580791718, "learning_rate": 4.451195722318459e-06, "loss": 0.4555, "step": 19553 }, { "epoch": 0.6966528314658781, "grad_norm": 2.0902078019212293, "learning_rate": 4.450235764579598e-06, "loss": 0.4396, "step": 19554 }, { "epoch": 0.6966884585923723, "grad_norm": 1.6627565292328768, "learning_rate": 4.449275880739723e-06, "loss": 0.5041, "step": 19555 }, { "epoch": 0.6967240857188663, "grad_norm": 1.6781918289680544, "learning_rate": 4.448316070811622e-06, "loss": 0.4868, "step": 19556 }, { "epoch": 0.6967597128453604, "grad_norm": 1.5322276234063612, "learning_rate": 4.447356334808067e-06, "loss": 0.5249, "step": 19557 }, { "epoch": 0.6967953399718546, "grad_norm": 1.523153239276501, "learning_rate": 4.446396672741852e-06, "loss": 0.5377, "step": 19558 }, { "epoch": 0.6968309670983487, "grad_norm": 1.9450466025443922, "learning_rate": 4.4454370846257414e-06, "loss": 0.4885, "step": 19559 }, { "epoch": 0.6968665942248428, "grad_norm": 1.4893356201763808, "learning_rate": 4.4444775704725225e-06, "loss": 0.3764, "step": 19560 }, { "epoch": 0.696902221351337, "grad_norm": 1.3823061610415828, "learning_rate": 4.443518130294963e-06, "loss": 0.3954, "step": 19561 }, { "epoch": 0.696937848477831, "grad_norm": 1.535104597177133, "learning_rate": 4.44255876410585e-06, "loss": 0.4525, "step": 19562 }, { "epoch": 0.6969734756043251, "grad_norm": 2.131139383092936, "learning_rate": 4.441599471917946e-06, "loss": 0.5697, "step": 19563 }, { "epoch": 0.6970091027308193, "grad_norm": 2.0162065277353305, "learning_rate": 4.44064025374403e-06, "loss": 0.5826, "step": 19564 }, { "epoch": 0.6970447298573134, "grad_norm": 1.6483807297482451, "learning_rate": 4.439681109596875e-06, "loss": 0.4691, "step": 19565 }, { "epoch": 0.6970803569838074, "grad_norm": 3.7612346339675207, "learning_rate": 4.4387220394892525e-06, "loss": 0.5356, "step": 19566 }, { "epoch": 0.6971159841103016, "grad_norm": 1.7677879499925502, "learning_rate": 4.437763043433937e-06, "loss": 0.5336, "step": 19567 }, { "epoch": 0.6971516112367957, "grad_norm": 1.4861823078839516, "learning_rate": 4.43680412144369e-06, "loss": 0.4048, "step": 19568 }, { "epoch": 0.6971872383632898, "grad_norm": 1.5888537887396503, "learning_rate": 4.435845273531289e-06, "loss": 0.4574, "step": 19569 }, { "epoch": 0.697222865489784, "grad_norm": 1.6443740393018897, "learning_rate": 4.434886499709493e-06, "loss": 0.414, "step": 19570 }, { "epoch": 0.697258492616278, "grad_norm": 1.789824696347389, "learning_rate": 4.433927799991073e-06, "loss": 0.5336, "step": 19571 }, { "epoch": 0.6972941197427721, "grad_norm": 1.4238914306622719, "learning_rate": 4.432969174388795e-06, "loss": 0.4099, "step": 19572 }, { "epoch": 0.6973297468692663, "grad_norm": 8.511462703366073, "learning_rate": 4.432010622915428e-06, "loss": 0.4609, "step": 19573 }, { "epoch": 0.6973653739957604, "grad_norm": 1.5875694995400065, "learning_rate": 4.431052145583726e-06, "loss": 0.5303, "step": 19574 }, { "epoch": 0.6974010011222544, "grad_norm": 1.6679322867365325, "learning_rate": 4.4300937424064595e-06, "loss": 0.5749, "step": 19575 }, { "epoch": 0.6974366282487486, "grad_norm": 1.4941930921862026, "learning_rate": 4.4291354133963895e-06, "loss": 0.4112, "step": 19576 }, { "epoch": 0.6974722553752427, "grad_norm": 2.7881802495639243, "learning_rate": 4.428177158566273e-06, "loss": 0.561, "step": 19577 }, { "epoch": 0.6975078825017368, "grad_norm": 1.9041920235993781, "learning_rate": 4.427218977928875e-06, "loss": 0.3982, "step": 19578 }, { "epoch": 0.697543509628231, "grad_norm": 1.8342456100517983, "learning_rate": 4.426260871496948e-06, "loss": 0.4412, "step": 19579 }, { "epoch": 0.697579136754725, "grad_norm": 1.6414917208741233, "learning_rate": 4.425302839283253e-06, "loss": 0.4934, "step": 19580 }, { "epoch": 0.6976147638812191, "grad_norm": 1.6214593204366918, "learning_rate": 4.424344881300547e-06, "loss": 0.4125, "step": 19581 }, { "epoch": 0.6976503910077133, "grad_norm": 1.462251125322583, "learning_rate": 4.4233869975615905e-06, "loss": 0.3896, "step": 19582 }, { "epoch": 0.6976860181342074, "grad_norm": 1.3830807655354345, "learning_rate": 4.4224291880791305e-06, "loss": 0.4131, "step": 19583 }, { "epoch": 0.6977216452607015, "grad_norm": 1.6498243958820362, "learning_rate": 4.421471452865928e-06, "loss": 0.4618, "step": 19584 }, { "epoch": 0.6977572723871956, "grad_norm": 1.8285836519495695, "learning_rate": 4.420513791934727e-06, "loss": 0.4372, "step": 19585 }, { "epoch": 0.6977928995136897, "grad_norm": 1.6044595517811773, "learning_rate": 4.419556205298286e-06, "loss": 0.461, "step": 19586 }, { "epoch": 0.6978285266401838, "grad_norm": 3.991622144977934, "learning_rate": 4.418598692969355e-06, "loss": 0.6002, "step": 19587 }, { "epoch": 0.697864153766678, "grad_norm": 1.6074814820023764, "learning_rate": 4.4176412549606825e-06, "loss": 0.4027, "step": 19588 }, { "epoch": 0.6978997808931721, "grad_norm": 1.662465568187326, "learning_rate": 4.416683891285023e-06, "loss": 0.4228, "step": 19589 }, { "epoch": 0.6979354080196661, "grad_norm": 1.3922438270627842, "learning_rate": 4.415726601955116e-06, "loss": 0.3823, "step": 19590 }, { "epoch": 0.6979710351461603, "grad_norm": 2.0498797837414986, "learning_rate": 4.414769386983716e-06, "loss": 0.5377, "step": 19591 }, { "epoch": 0.6980066622726544, "grad_norm": 1.7155397160274428, "learning_rate": 4.413812246383562e-06, "loss": 0.4733, "step": 19592 }, { "epoch": 0.6980422893991485, "grad_norm": 2.1000679314411577, "learning_rate": 4.412855180167406e-06, "loss": 0.4564, "step": 19593 }, { "epoch": 0.6980779165256427, "grad_norm": 1.5497883542533515, "learning_rate": 4.4118981883479805e-06, "loss": 0.446, "step": 19594 }, { "epoch": 0.6981135436521367, "grad_norm": 1.35839485760668, "learning_rate": 4.410941270938045e-06, "loss": 0.4307, "step": 19595 }, { "epoch": 0.6981491707786308, "grad_norm": 1.6058485145353079, "learning_rate": 4.409984427950328e-06, "loss": 0.4706, "step": 19596 }, { "epoch": 0.698184797905125, "grad_norm": 1.6977192547499833, "learning_rate": 4.409027659397577e-06, "loss": 0.4157, "step": 19597 }, { "epoch": 0.6982204250316191, "grad_norm": 1.8117613905160235, "learning_rate": 4.408070965292534e-06, "loss": 0.5335, "step": 19598 }, { "epoch": 0.6982560521581131, "grad_norm": 2.223322473618212, "learning_rate": 4.407114345647929e-06, "loss": 0.4069, "step": 19599 }, { "epoch": 0.6982916792846073, "grad_norm": 1.8207191317856748, "learning_rate": 4.406157800476512e-06, "loss": 0.5048, "step": 19600 }, { "epoch": 0.6983273064111014, "grad_norm": 1.4211490965197269, "learning_rate": 4.405201329791008e-06, "loss": 0.4484, "step": 19601 }, { "epoch": 0.6983629335375955, "grad_norm": 1.5467776529116075, "learning_rate": 4.40424493360416e-06, "loss": 0.3977, "step": 19602 }, { "epoch": 0.6983985606640897, "grad_norm": 1.60530662874927, "learning_rate": 4.403288611928702e-06, "loss": 0.3938, "step": 19603 }, { "epoch": 0.6984341877905837, "grad_norm": 1.4766780367173182, "learning_rate": 4.402332364777371e-06, "loss": 0.5486, "step": 19604 }, { "epoch": 0.6984698149170778, "grad_norm": 1.411719701904136, "learning_rate": 4.401376192162894e-06, "loss": 0.4163, "step": 19605 }, { "epoch": 0.698505442043572, "grad_norm": 1.696125234957798, "learning_rate": 4.400420094098009e-06, "loss": 0.4617, "step": 19606 }, { "epoch": 0.6985410691700661, "grad_norm": 1.9161960031835512, "learning_rate": 4.399464070595442e-06, "loss": 0.4095, "step": 19607 }, { "epoch": 0.6985766962965602, "grad_norm": 1.361633952051636, "learning_rate": 4.398508121667925e-06, "loss": 0.4143, "step": 19608 }, { "epoch": 0.6986123234230543, "grad_norm": 1.8943543812494141, "learning_rate": 4.397552247328187e-06, "loss": 0.5396, "step": 19609 }, { "epoch": 0.6986479505495484, "grad_norm": 1.7038267512397998, "learning_rate": 4.396596447588956e-06, "loss": 0.5662, "step": 19610 }, { "epoch": 0.6986835776760425, "grad_norm": 1.9357495900945174, "learning_rate": 4.3956407224629645e-06, "loss": 0.5237, "step": 19611 }, { "epoch": 0.6987192048025367, "grad_norm": 2.382118102079703, "learning_rate": 4.394685071962931e-06, "loss": 0.5264, "step": 19612 }, { "epoch": 0.6987548319290308, "grad_norm": 1.5281002607518444, "learning_rate": 4.393729496101587e-06, "loss": 0.5232, "step": 19613 }, { "epoch": 0.6987904590555248, "grad_norm": 4.076047350751672, "learning_rate": 4.3927739948916496e-06, "loss": 0.5263, "step": 19614 }, { "epoch": 0.698826086182019, "grad_norm": 1.3836312907097252, "learning_rate": 4.391818568345848e-06, "loss": 0.4065, "step": 19615 }, { "epoch": 0.6988617133085131, "grad_norm": 1.6585996619379564, "learning_rate": 4.390863216476899e-06, "loss": 0.4353, "step": 19616 }, { "epoch": 0.6988973404350072, "grad_norm": 1.5415605152439826, "learning_rate": 4.389907939297528e-06, "loss": 0.5027, "step": 19617 }, { "epoch": 0.6989329675615014, "grad_norm": 1.373691022975807, "learning_rate": 4.388952736820453e-06, "loss": 0.4638, "step": 19618 }, { "epoch": 0.6989685946879954, "grad_norm": 1.3290400487728733, "learning_rate": 4.387997609058399e-06, "loss": 0.3834, "step": 19619 }, { "epoch": 0.6990042218144895, "grad_norm": 1.56421936244428, "learning_rate": 4.387042556024074e-06, "loss": 0.4349, "step": 19620 }, { "epoch": 0.6990398489409837, "grad_norm": 1.8350730997797025, "learning_rate": 4.386087577730201e-06, "loss": 0.4653, "step": 19621 }, { "epoch": 0.6990754760674778, "grad_norm": 1.8003023607644981, "learning_rate": 4.3851326741895e-06, "loss": 0.5261, "step": 19622 }, { "epoch": 0.6991111031939719, "grad_norm": 1.6936925256854185, "learning_rate": 4.384177845414679e-06, "loss": 0.4093, "step": 19623 }, { "epoch": 0.699146730320466, "grad_norm": 1.322916579071022, "learning_rate": 4.383223091418454e-06, "loss": 0.4172, "step": 19624 }, { "epoch": 0.6991823574469601, "grad_norm": 1.4438244660938033, "learning_rate": 4.382268412213539e-06, "loss": 0.4284, "step": 19625 }, { "epoch": 0.6992179845734542, "grad_norm": 1.4545151019613174, "learning_rate": 4.381313807812652e-06, "loss": 0.4276, "step": 19626 }, { "epoch": 0.6992536116999484, "grad_norm": 1.300732468987008, "learning_rate": 4.380359278228494e-06, "loss": 0.4087, "step": 19627 }, { "epoch": 0.6992892388264424, "grad_norm": 1.7746954596127253, "learning_rate": 4.3794048234737855e-06, "loss": 0.5471, "step": 19628 }, { "epoch": 0.6993248659529365, "grad_norm": 1.4562658470589902, "learning_rate": 4.378450443561224e-06, "loss": 0.4211, "step": 19629 }, { "epoch": 0.6993604930794307, "grad_norm": 1.5171178390615876, "learning_rate": 4.37749613850353e-06, "loss": 0.4764, "step": 19630 }, { "epoch": 0.6993961202059248, "grad_norm": 1.1125501122894568, "learning_rate": 4.3765419083134e-06, "loss": 0.3418, "step": 19631 }, { "epoch": 0.6994317473324189, "grad_norm": 1.7923956405624089, "learning_rate": 4.3755877530035415e-06, "loss": 0.4849, "step": 19632 }, { "epoch": 0.699467374458913, "grad_norm": 1.863427414871778, "learning_rate": 4.37463367258667e-06, "loss": 0.4301, "step": 19633 }, { "epoch": 0.6995030015854071, "grad_norm": 1.504118330751156, "learning_rate": 4.373679667075481e-06, "loss": 0.4108, "step": 19634 }, { "epoch": 0.6995386287119012, "grad_norm": 1.800733450974101, "learning_rate": 4.372725736482682e-06, "loss": 0.5635, "step": 19635 }, { "epoch": 0.6995742558383954, "grad_norm": 1.349182916262399, "learning_rate": 4.37177188082097e-06, "loss": 0.404, "step": 19636 }, { "epoch": 0.6996098829648895, "grad_norm": 1.7618663913149535, "learning_rate": 4.370818100103054e-06, "loss": 0.5715, "step": 19637 }, { "epoch": 0.6996455100913835, "grad_norm": 1.3133808781932943, "learning_rate": 4.369864394341624e-06, "loss": 0.4307, "step": 19638 }, { "epoch": 0.6996811372178777, "grad_norm": 1.5259000113143917, "learning_rate": 4.3689107635493855e-06, "loss": 0.4547, "step": 19639 }, { "epoch": 0.6997167643443718, "grad_norm": 1.532483711701101, "learning_rate": 4.367957207739036e-06, "loss": 0.636, "step": 19640 }, { "epoch": 0.6997523914708659, "grad_norm": 1.4366828163816416, "learning_rate": 4.367003726923277e-06, "loss": 0.4081, "step": 19641 }, { "epoch": 0.6997880185973601, "grad_norm": 1.2883142787898016, "learning_rate": 4.366050321114797e-06, "loss": 0.4207, "step": 19642 }, { "epoch": 0.6998236457238541, "grad_norm": 1.5570120655830764, "learning_rate": 4.365096990326298e-06, "loss": 0.5342, "step": 19643 }, { "epoch": 0.6998592728503482, "grad_norm": 1.357342029329213, "learning_rate": 4.364143734570467e-06, "loss": 0.4122, "step": 19644 }, { "epoch": 0.6998948999768424, "grad_norm": 1.3055009678785126, "learning_rate": 4.363190553860003e-06, "loss": 0.4375, "step": 19645 }, { "epoch": 0.6999305271033365, "grad_norm": 1.599789919786172, "learning_rate": 4.362237448207596e-06, "loss": 0.5956, "step": 19646 }, { "epoch": 0.6999661542298306, "grad_norm": 1.6965158575906016, "learning_rate": 4.361284417625938e-06, "loss": 0.5676, "step": 19647 }, { "epoch": 0.7000017813563247, "grad_norm": 1.4422472795405559, "learning_rate": 4.360331462127721e-06, "loss": 0.4116, "step": 19648 }, { "epoch": 0.7000374084828188, "grad_norm": 1.538573031260255, "learning_rate": 4.3593785817256305e-06, "loss": 0.5328, "step": 19649 }, { "epoch": 0.7000730356093129, "grad_norm": 1.4969586968614566, "learning_rate": 4.35842577643236e-06, "loss": 0.4273, "step": 19650 }, { "epoch": 0.7001086627358071, "grad_norm": 2.060001329740657, "learning_rate": 4.357473046260591e-06, "loss": 0.5573, "step": 19651 }, { "epoch": 0.7001442898623012, "grad_norm": 2.209718030727306, "learning_rate": 4.356520391223015e-06, "loss": 0.5914, "step": 19652 }, { "epoch": 0.7001799169887952, "grad_norm": 1.4831212970221055, "learning_rate": 4.355567811332311e-06, "loss": 0.4731, "step": 19653 }, { "epoch": 0.7002155441152894, "grad_norm": 1.4190456361258028, "learning_rate": 4.354615306601166e-06, "loss": 0.4021, "step": 19654 }, { "epoch": 0.7002511712417835, "grad_norm": 1.667355569673125, "learning_rate": 4.3536628770422655e-06, "loss": 0.4928, "step": 19655 }, { "epoch": 0.7002867983682776, "grad_norm": 1.2207400809484963, "learning_rate": 4.35271052266829e-06, "loss": 0.3957, "step": 19656 }, { "epoch": 0.7003224254947717, "grad_norm": 1.8269394470271014, "learning_rate": 4.3517582434919235e-06, "loss": 0.4452, "step": 19657 }, { "epoch": 0.7003580526212658, "grad_norm": 1.8148793215090915, "learning_rate": 4.350806039525841e-06, "loss": 0.5725, "step": 19658 }, { "epoch": 0.7003936797477599, "grad_norm": 1.5624368982231318, "learning_rate": 4.349853910782728e-06, "loss": 0.5332, "step": 19659 }, { "epoch": 0.7004293068742541, "grad_norm": 1.4630312977502786, "learning_rate": 4.348901857275257e-06, "loss": 0.4205, "step": 19660 }, { "epoch": 0.7004649340007482, "grad_norm": 1.6035381305236087, "learning_rate": 4.347949879016107e-06, "loss": 0.5249, "step": 19661 }, { "epoch": 0.7005005611272422, "grad_norm": 1.3476409591533223, "learning_rate": 4.346997976017955e-06, "loss": 0.4132, "step": 19662 }, { "epoch": 0.7005361882537364, "grad_norm": 1.4216382140300277, "learning_rate": 4.346046148293479e-06, "loss": 0.3966, "step": 19663 }, { "epoch": 0.7005718153802305, "grad_norm": 1.3790016216871124, "learning_rate": 4.345094395855347e-06, "loss": 0.4395, "step": 19664 }, { "epoch": 0.7006074425067246, "grad_norm": 1.4036072038984913, "learning_rate": 4.34414271871624e-06, "loss": 0.4308, "step": 19665 }, { "epoch": 0.7006430696332188, "grad_norm": 1.3315198888282476, "learning_rate": 4.343191116888822e-06, "loss": 0.4931, "step": 19666 }, { "epoch": 0.7006786967597128, "grad_norm": 1.2859945890670286, "learning_rate": 4.342239590385768e-06, "loss": 0.4008, "step": 19667 }, { "epoch": 0.7007143238862069, "grad_norm": 1.268374623559733, "learning_rate": 4.341288139219752e-06, "loss": 0.3717, "step": 19668 }, { "epoch": 0.7007499510127011, "grad_norm": 1.8127730640580986, "learning_rate": 4.340336763403437e-06, "loss": 0.4463, "step": 19669 }, { "epoch": 0.7007855781391952, "grad_norm": 1.4481980548674884, "learning_rate": 4.339385462949492e-06, "loss": 0.3531, "step": 19670 }, { "epoch": 0.7008212052656893, "grad_norm": 1.3709267312102011, "learning_rate": 4.338434237870587e-06, "loss": 0.4081, "step": 19671 }, { "epoch": 0.7008568323921834, "grad_norm": 1.408840606433997, "learning_rate": 4.337483088179391e-06, "loss": 0.4149, "step": 19672 }, { "epoch": 0.7008924595186775, "grad_norm": 1.7156264314166552, "learning_rate": 4.336532013888563e-06, "loss": 0.5559, "step": 19673 }, { "epoch": 0.7009280866451716, "grad_norm": 1.5300133205453763, "learning_rate": 4.335581015010772e-06, "loss": 0.4782, "step": 19674 }, { "epoch": 0.7009637137716658, "grad_norm": 1.4584303412271524, "learning_rate": 4.334630091558675e-06, "loss": 0.4557, "step": 19675 }, { "epoch": 0.7009993408981599, "grad_norm": 1.3883428032066145, "learning_rate": 4.3336792435449385e-06, "loss": 0.3894, "step": 19676 }, { "epoch": 0.7010349680246539, "grad_norm": 1.3200486351651841, "learning_rate": 4.332728470982224e-06, "loss": 0.4108, "step": 19677 }, { "epoch": 0.7010705951511481, "grad_norm": 1.4313477395609497, "learning_rate": 4.3317777738831914e-06, "loss": 0.3689, "step": 19678 }, { "epoch": 0.7011062222776422, "grad_norm": 1.4139058429349058, "learning_rate": 4.330827152260502e-06, "loss": 0.4489, "step": 19679 }, { "epoch": 0.7011418494041363, "grad_norm": 1.435690528410414, "learning_rate": 4.329876606126808e-06, "loss": 0.4756, "step": 19680 }, { "epoch": 0.7011774765306305, "grad_norm": 1.2057192269618209, "learning_rate": 4.3289261354947755e-06, "loss": 0.4022, "step": 19681 }, { "epoch": 0.7012131036571245, "grad_norm": 1.4965772226254341, "learning_rate": 4.3279757403770495e-06, "loss": 0.4266, "step": 19682 }, { "epoch": 0.7012487307836186, "grad_norm": 1.401051330852285, "learning_rate": 4.327025420786293e-06, "loss": 0.4487, "step": 19683 }, { "epoch": 0.7012843579101128, "grad_norm": 1.4552622171781322, "learning_rate": 4.326075176735157e-06, "loss": 0.5062, "step": 19684 }, { "epoch": 0.7013199850366069, "grad_norm": 1.3547077367643587, "learning_rate": 4.3251250082363e-06, "loss": 0.4294, "step": 19685 }, { "epoch": 0.7013556121631009, "grad_norm": 1.4199511032472751, "learning_rate": 4.324174915302366e-06, "loss": 0.6223, "step": 19686 }, { "epoch": 0.7013912392895951, "grad_norm": 1.146119975467121, "learning_rate": 4.323224897946013e-06, "loss": 0.4395, "step": 19687 }, { "epoch": 0.7014268664160892, "grad_norm": 3.3253850257302977, "learning_rate": 4.322274956179887e-06, "loss": 0.565, "step": 19688 }, { "epoch": 0.7014624935425833, "grad_norm": 1.7477460563261404, "learning_rate": 4.321325090016641e-06, "loss": 0.5337, "step": 19689 }, { "epoch": 0.7014981206690775, "grad_norm": 1.626113615792294, "learning_rate": 4.320375299468916e-06, "loss": 0.5307, "step": 19690 }, { "epoch": 0.7015337477955715, "grad_norm": 1.3032911004157135, "learning_rate": 4.3194255845493645e-06, "loss": 0.441, "step": 19691 }, { "epoch": 0.7015693749220656, "grad_norm": 1.4299169833603598, "learning_rate": 4.318475945270632e-06, "loss": 0.3923, "step": 19692 }, { "epoch": 0.7016050020485598, "grad_norm": 2.1465903350846878, "learning_rate": 4.3175263816453625e-06, "loss": 0.531, "step": 19693 }, { "epoch": 0.7016406291750539, "grad_norm": 1.4766410894721522, "learning_rate": 4.316576893686206e-06, "loss": 0.5274, "step": 19694 }, { "epoch": 0.701676256301548, "grad_norm": 1.3548734372640463, "learning_rate": 4.315627481405796e-06, "loss": 0.4086, "step": 19695 }, { "epoch": 0.7017118834280421, "grad_norm": 1.6068324957636808, "learning_rate": 4.314678144816783e-06, "loss": 0.569, "step": 19696 }, { "epoch": 0.7017475105545362, "grad_norm": 1.5317037588146396, "learning_rate": 4.313728883931801e-06, "loss": 0.4764, "step": 19697 }, { "epoch": 0.7017831376810303, "grad_norm": 1.7603716500660143, "learning_rate": 4.312779698763493e-06, "loss": 0.5439, "step": 19698 }, { "epoch": 0.7018187648075245, "grad_norm": 1.8476465250317866, "learning_rate": 4.311830589324499e-06, "loss": 0.5561, "step": 19699 }, { "epoch": 0.7018543919340186, "grad_norm": 1.54865285932954, "learning_rate": 4.31088155562746e-06, "loss": 0.6003, "step": 19700 }, { "epoch": 0.7018900190605126, "grad_norm": 1.5527085322641403, "learning_rate": 4.309932597685006e-06, "loss": 0.6101, "step": 19701 }, { "epoch": 0.7019256461870068, "grad_norm": 1.4515381545076027, "learning_rate": 4.308983715509776e-06, "loss": 0.4189, "step": 19702 }, { "epoch": 0.7019612733135009, "grad_norm": 1.3092528872997438, "learning_rate": 4.308034909114409e-06, "loss": 0.4524, "step": 19703 }, { "epoch": 0.701996900439995, "grad_norm": 1.5213859458278176, "learning_rate": 4.307086178511533e-06, "loss": 0.4625, "step": 19704 }, { "epoch": 0.7020325275664892, "grad_norm": 1.2577742504706926, "learning_rate": 4.306137523713787e-06, "loss": 0.4323, "step": 19705 }, { "epoch": 0.7020681546929832, "grad_norm": 1.9694659877227427, "learning_rate": 4.305188944733796e-06, "loss": 0.4956, "step": 19706 }, { "epoch": 0.7021037818194773, "grad_norm": 1.3931512414330696, "learning_rate": 4.304240441584194e-06, "loss": 0.3904, "step": 19707 }, { "epoch": 0.7021394089459715, "grad_norm": 1.3770540655768795, "learning_rate": 4.3032920142776125e-06, "loss": 0.4509, "step": 19708 }, { "epoch": 0.7021750360724656, "grad_norm": 1.5995198126986134, "learning_rate": 4.302343662826683e-06, "loss": 0.5488, "step": 19709 }, { "epoch": 0.7022106631989596, "grad_norm": 2.187269876309395, "learning_rate": 4.301395387244026e-06, "loss": 0.4903, "step": 19710 }, { "epoch": 0.7022462903254538, "grad_norm": 1.7939132199349388, "learning_rate": 4.300447187542276e-06, "loss": 0.5037, "step": 19711 }, { "epoch": 0.7022819174519479, "grad_norm": 1.435998725218386, "learning_rate": 4.299499063734053e-06, "loss": 0.407, "step": 19712 }, { "epoch": 0.702317544578442, "grad_norm": 1.3297072215648678, "learning_rate": 4.298551015831983e-06, "loss": 0.3903, "step": 19713 }, { "epoch": 0.7023531717049362, "grad_norm": 1.5687140277355458, "learning_rate": 4.297603043848692e-06, "loss": 0.4936, "step": 19714 }, { "epoch": 0.7023887988314302, "grad_norm": 1.642289807981016, "learning_rate": 4.296655147796802e-06, "loss": 0.4975, "step": 19715 }, { "epoch": 0.7024244259579243, "grad_norm": 1.3330020207473117, "learning_rate": 4.295707327688938e-06, "loss": 0.4706, "step": 19716 }, { "epoch": 0.7024600530844185, "grad_norm": 1.3445624166032861, "learning_rate": 4.2947595835377155e-06, "loss": 0.4485, "step": 19717 }, { "epoch": 0.7024956802109126, "grad_norm": 1.766014487240461, "learning_rate": 4.293811915355761e-06, "loss": 0.5048, "step": 19718 }, { "epoch": 0.7025313073374067, "grad_norm": 1.457970713509719, "learning_rate": 4.292864323155684e-06, "loss": 0.4366, "step": 19719 }, { "epoch": 0.7025669344639008, "grad_norm": 1.4014443338813096, "learning_rate": 4.291916806950112e-06, "loss": 0.4236, "step": 19720 }, { "epoch": 0.7026025615903949, "grad_norm": 1.3246178120262317, "learning_rate": 4.29096936675165e-06, "loss": 0.4367, "step": 19721 }, { "epoch": 0.702638188716889, "grad_norm": 1.752716671464558, "learning_rate": 4.290022002572929e-06, "loss": 0.5294, "step": 19722 }, { "epoch": 0.7026738158433832, "grad_norm": 1.6467720308680014, "learning_rate": 4.289074714426552e-06, "loss": 0.521, "step": 19723 }, { "epoch": 0.7027094429698773, "grad_norm": 1.3085124705584754, "learning_rate": 4.2881275023251375e-06, "loss": 0.5246, "step": 19724 }, { "epoch": 0.7027450700963713, "grad_norm": 1.4251880107441255, "learning_rate": 4.287180366281301e-06, "loss": 0.4586, "step": 19725 }, { "epoch": 0.7027806972228655, "grad_norm": 1.4536728202530929, "learning_rate": 4.286233306307647e-06, "loss": 0.4881, "step": 19726 }, { "epoch": 0.7028163243493596, "grad_norm": 1.576296495224819, "learning_rate": 4.285286322416794e-06, "loss": 0.4223, "step": 19727 }, { "epoch": 0.7028519514758537, "grad_norm": 1.2874486178057092, "learning_rate": 4.284339414621343e-06, "loss": 0.3926, "step": 19728 }, { "epoch": 0.7028875786023479, "grad_norm": 1.5123640155121967, "learning_rate": 4.28339258293391e-06, "loss": 0.4319, "step": 19729 }, { "epoch": 0.7029232057288419, "grad_norm": 1.8848039471530527, "learning_rate": 4.2824458273671e-06, "loss": 0.6157, "step": 19730 }, { "epoch": 0.702958832855336, "grad_norm": 1.7804854626298563, "learning_rate": 4.281499147933522e-06, "loss": 0.5191, "step": 19731 }, { "epoch": 0.7029944599818302, "grad_norm": 1.8896567675424139, "learning_rate": 4.280552544645779e-06, "loss": 0.5858, "step": 19732 }, { "epoch": 0.7030300871083243, "grad_norm": 1.5849710788627167, "learning_rate": 4.279606017516478e-06, "loss": 0.4872, "step": 19733 }, { "epoch": 0.7030657142348183, "grad_norm": 1.716456196664448, "learning_rate": 4.278659566558218e-06, "loss": 0.4672, "step": 19734 }, { "epoch": 0.7031013413613125, "grad_norm": 1.726205839130178, "learning_rate": 4.277713191783604e-06, "loss": 0.4757, "step": 19735 }, { "epoch": 0.7031369684878066, "grad_norm": 1.750736164105378, "learning_rate": 4.27676689320524e-06, "loss": 0.5307, "step": 19736 }, { "epoch": 0.7031725956143007, "grad_norm": 1.227940864407388, "learning_rate": 4.275820670835724e-06, "loss": 0.4157, "step": 19737 }, { "epoch": 0.7032082227407949, "grad_norm": 1.3518970225112585, "learning_rate": 4.274874524687661e-06, "loss": 0.4264, "step": 19738 }, { "epoch": 0.7032438498672889, "grad_norm": 1.2939639330530737, "learning_rate": 4.2739284547736426e-06, "loss": 0.4477, "step": 19739 }, { "epoch": 0.703279476993783, "grad_norm": 2.218917481576153, "learning_rate": 4.272982461106271e-06, "loss": 0.5807, "step": 19740 }, { "epoch": 0.7033151041202772, "grad_norm": 1.3595691920916402, "learning_rate": 4.272036543698138e-06, "loss": 0.4225, "step": 19741 }, { "epoch": 0.7033507312467713, "grad_norm": 1.6586571854141754, "learning_rate": 4.271090702561846e-06, "loss": 0.4307, "step": 19742 }, { "epoch": 0.7033863583732654, "grad_norm": 1.3967569210671402, "learning_rate": 4.270144937709981e-06, "loss": 0.5314, "step": 19743 }, { "epoch": 0.7034219854997595, "grad_norm": 1.6339158683275314, "learning_rate": 4.269199249155142e-06, "loss": 0.595, "step": 19744 }, { "epoch": 0.7034576126262536, "grad_norm": 1.2967368506420112, "learning_rate": 4.26825363690992e-06, "loss": 0.4207, "step": 19745 }, { "epoch": 0.7034932397527477, "grad_norm": 1.8107349796036771, "learning_rate": 4.267308100986911e-06, "loss": 0.5404, "step": 19746 }, { "epoch": 0.7035288668792419, "grad_norm": 1.3236148342710996, "learning_rate": 4.266362641398698e-06, "loss": 0.4376, "step": 19747 }, { "epoch": 0.703564494005736, "grad_norm": 1.3193049598378734, "learning_rate": 4.265417258157875e-06, "loss": 0.425, "step": 19748 }, { "epoch": 0.70360012113223, "grad_norm": 1.7819800601019828, "learning_rate": 4.2644719512770314e-06, "loss": 0.5088, "step": 19749 }, { "epoch": 0.7036357482587242, "grad_norm": 1.4366111066106186, "learning_rate": 4.26352672076875e-06, "loss": 0.5629, "step": 19750 }, { "epoch": 0.7036713753852183, "grad_norm": 1.318330984578542, "learning_rate": 4.2625815666456194e-06, "loss": 0.415, "step": 19751 }, { "epoch": 0.7037070025117124, "grad_norm": 2.4224313351111464, "learning_rate": 4.261636488920225e-06, "loss": 0.5827, "step": 19752 }, { "epoch": 0.7037426296382066, "grad_norm": 1.295787784556827, "learning_rate": 4.260691487605157e-06, "loss": 0.4113, "step": 19753 }, { "epoch": 0.7037782567647006, "grad_norm": 2.711271859082639, "learning_rate": 4.259746562712988e-06, "loss": 0.6001, "step": 19754 }, { "epoch": 0.7038138838911947, "grad_norm": 1.7125068378955361, "learning_rate": 4.258801714256309e-06, "loss": 0.561, "step": 19755 }, { "epoch": 0.7038495110176889, "grad_norm": 2.8254451908438933, "learning_rate": 4.2578569422476955e-06, "loss": 0.3693, "step": 19756 }, { "epoch": 0.703885138144183, "grad_norm": 1.2668143629159208, "learning_rate": 4.2569122466997335e-06, "loss": 0.4012, "step": 19757 }, { "epoch": 0.703920765270677, "grad_norm": 1.511116754438478, "learning_rate": 4.2559676276249926e-06, "loss": 0.3397, "step": 19758 }, { "epoch": 0.7039563923971712, "grad_norm": 1.3878610505614741, "learning_rate": 4.2550230850360604e-06, "loss": 0.4312, "step": 19759 }, { "epoch": 0.7039920195236653, "grad_norm": 1.9533443080236177, "learning_rate": 4.2540786189455156e-06, "loss": 0.6307, "step": 19760 }, { "epoch": 0.7040276466501594, "grad_norm": 1.6462426480672467, "learning_rate": 4.253134229365926e-06, "loss": 0.4894, "step": 19761 }, { "epoch": 0.7040632737766536, "grad_norm": 1.6290759386398777, "learning_rate": 4.252189916309874e-06, "loss": 0.6214, "step": 19762 }, { "epoch": 0.7040989009031476, "grad_norm": 1.991029405670219, "learning_rate": 4.251245679789928e-06, "loss": 0.5066, "step": 19763 }, { "epoch": 0.7041345280296417, "grad_norm": 1.2834283573632812, "learning_rate": 4.250301519818668e-06, "loss": 0.3911, "step": 19764 }, { "epoch": 0.7041701551561359, "grad_norm": 1.1986308009597935, "learning_rate": 4.2493574364086575e-06, "loss": 0.3887, "step": 19765 }, { "epoch": 0.70420578228263, "grad_norm": 1.4121054204178152, "learning_rate": 4.248413429572473e-06, "loss": 0.433, "step": 19766 }, { "epoch": 0.704241409409124, "grad_norm": 2.9137172543460164, "learning_rate": 4.247469499322683e-06, "loss": 0.4947, "step": 19767 }, { "epoch": 0.7042770365356182, "grad_norm": 1.3428099872987356, "learning_rate": 4.246525645671862e-06, "loss": 0.4938, "step": 19768 }, { "epoch": 0.7043126636621123, "grad_norm": 1.6636473769912197, "learning_rate": 4.245581868632569e-06, "loss": 0.4911, "step": 19769 }, { "epoch": 0.7043482907886064, "grad_norm": 1.3916653012574884, "learning_rate": 4.244638168217379e-06, "loss": 0.433, "step": 19770 }, { "epoch": 0.7043839179151006, "grad_norm": 1.2872618656488881, "learning_rate": 4.243694544438851e-06, "loss": 0.4144, "step": 19771 }, { "epoch": 0.7044195450415947, "grad_norm": 1.1890181507402202, "learning_rate": 4.2427509973095536e-06, "loss": 0.406, "step": 19772 }, { "epoch": 0.7044551721680887, "grad_norm": 1.451166076705715, "learning_rate": 4.241807526842051e-06, "loss": 0.3613, "step": 19773 }, { "epoch": 0.7044907992945829, "grad_norm": 1.8556073089279812, "learning_rate": 4.2408641330489066e-06, "loss": 0.509, "step": 19774 }, { "epoch": 0.704526426421077, "grad_norm": 1.571078301421029, "learning_rate": 4.239920815942684e-06, "loss": 0.447, "step": 19775 }, { "epoch": 0.7045620535475711, "grad_norm": 1.8405139411772695, "learning_rate": 4.238977575535939e-06, "loss": 0.5467, "step": 19776 }, { "epoch": 0.7045976806740653, "grad_norm": 1.4453247036156869, "learning_rate": 4.238034411841238e-06, "loss": 0.369, "step": 19777 }, { "epoch": 0.7046333078005593, "grad_norm": 1.5765518974451114, "learning_rate": 4.237091324871132e-06, "loss": 0.4612, "step": 19778 }, { "epoch": 0.7046689349270534, "grad_norm": 1.716017913730497, "learning_rate": 4.2361483146381865e-06, "loss": 0.5104, "step": 19779 }, { "epoch": 0.7047045620535476, "grad_norm": 1.2398774780614672, "learning_rate": 4.2352053811549525e-06, "loss": 0.4176, "step": 19780 }, { "epoch": 0.7047401891800417, "grad_norm": 1.6736662237466118, "learning_rate": 4.234262524433986e-06, "loss": 0.5659, "step": 19781 }, { "epoch": 0.7047758163065359, "grad_norm": 1.5138187335087425, "learning_rate": 4.233319744487846e-06, "loss": 0.5433, "step": 19782 }, { "epoch": 0.7048114434330299, "grad_norm": 1.6625441932482548, "learning_rate": 4.2323770413290845e-06, "loss": 0.4208, "step": 19783 }, { "epoch": 0.704847070559524, "grad_norm": 2.1105124541134663, "learning_rate": 4.2314344149702554e-06, "loss": 0.5535, "step": 19784 }, { "epoch": 0.7048826976860182, "grad_norm": 1.5238741817229864, "learning_rate": 4.230491865423908e-06, "loss": 0.4617, "step": 19785 }, { "epoch": 0.7049183248125123, "grad_norm": 1.7958893513395284, "learning_rate": 4.2295493927025955e-06, "loss": 0.4738, "step": 19786 }, { "epoch": 0.7049539519390063, "grad_norm": 2.089319305357697, "learning_rate": 4.228606996818864e-06, "loss": 0.5118, "step": 19787 }, { "epoch": 0.7049895790655005, "grad_norm": 1.41327677975299, "learning_rate": 4.227664677785264e-06, "loss": 0.3778, "step": 19788 }, { "epoch": 0.7050252061919946, "grad_norm": 1.3362224018618358, "learning_rate": 4.2267224356143436e-06, "loss": 0.4314, "step": 19789 }, { "epoch": 0.7050608333184887, "grad_norm": 1.3861328460383937, "learning_rate": 4.225780270318652e-06, "loss": 0.3913, "step": 19790 }, { "epoch": 0.7050964604449829, "grad_norm": 1.948653668228989, "learning_rate": 4.224838181910729e-06, "loss": 0.5302, "step": 19791 }, { "epoch": 0.7051320875714769, "grad_norm": 1.5412787320398154, "learning_rate": 4.223896170403126e-06, "loss": 0.5123, "step": 19792 }, { "epoch": 0.705167714697971, "grad_norm": 1.1908717355813152, "learning_rate": 4.222954235808379e-06, "loss": 0.4319, "step": 19793 }, { "epoch": 0.7052033418244652, "grad_norm": 1.8184886297342313, "learning_rate": 4.222012378139033e-06, "loss": 0.4457, "step": 19794 }, { "epoch": 0.7052389689509593, "grad_norm": 1.7033241001225967, "learning_rate": 4.221070597407636e-06, "loss": 0.4283, "step": 19795 }, { "epoch": 0.7052745960774534, "grad_norm": 1.3937266473979415, "learning_rate": 4.220128893626715e-06, "loss": 0.4205, "step": 19796 }, { "epoch": 0.7053102232039475, "grad_norm": 1.2312787171787662, "learning_rate": 4.219187266808826e-06, "loss": 0.4235, "step": 19797 }, { "epoch": 0.7053458503304416, "grad_norm": 1.3840279061004115, "learning_rate": 4.218245716966495e-06, "loss": 0.4327, "step": 19798 }, { "epoch": 0.7053814774569357, "grad_norm": 1.7841890751958647, "learning_rate": 4.217304244112267e-06, "loss": 0.5452, "step": 19799 }, { "epoch": 0.7054171045834299, "grad_norm": 1.4004484823472634, "learning_rate": 4.21636284825867e-06, "loss": 0.374, "step": 19800 }, { "epoch": 0.705452731709924, "grad_norm": 1.382623377255781, "learning_rate": 4.21542152941825e-06, "loss": 0.4446, "step": 19801 }, { "epoch": 0.705488358836418, "grad_norm": 2.1532674174526636, "learning_rate": 4.214480287603531e-06, "loss": 0.5053, "step": 19802 }, { "epoch": 0.7055239859629122, "grad_norm": 1.4983847956088723, "learning_rate": 4.21353912282705e-06, "loss": 0.3938, "step": 19803 }, { "epoch": 0.7055596130894063, "grad_norm": 1.9821461028634044, "learning_rate": 4.212598035101341e-06, "loss": 0.5168, "step": 19804 }, { "epoch": 0.7055952402159004, "grad_norm": 1.7866257701269843, "learning_rate": 4.211657024438934e-06, "loss": 0.5732, "step": 19805 }, { "epoch": 0.7056308673423946, "grad_norm": 1.2385402772988916, "learning_rate": 4.210716090852364e-06, "loss": 0.4464, "step": 19806 }, { "epoch": 0.7056664944688886, "grad_norm": 1.7223232893547349, "learning_rate": 4.209775234354151e-06, "loss": 0.4812, "step": 19807 }, { "epoch": 0.7057021215953827, "grad_norm": 1.5625827574544475, "learning_rate": 4.208834454956831e-06, "loss": 0.4955, "step": 19808 }, { "epoch": 0.7057377487218769, "grad_norm": 1.5024927627421805, "learning_rate": 4.2078937526729255e-06, "loss": 0.5408, "step": 19809 }, { "epoch": 0.705773375848371, "grad_norm": 1.2736048062284593, "learning_rate": 4.2069531275149634e-06, "loss": 0.447, "step": 19810 }, { "epoch": 0.705809002974865, "grad_norm": 2.1033996007534617, "learning_rate": 4.206012579495469e-06, "loss": 0.3866, "step": 19811 }, { "epoch": 0.7058446301013592, "grad_norm": 1.39316874985136, "learning_rate": 4.205072108626971e-06, "loss": 0.4416, "step": 19812 }, { "epoch": 0.7058802572278533, "grad_norm": 1.188731917478106, "learning_rate": 4.204131714921984e-06, "loss": 0.413, "step": 19813 }, { "epoch": 0.7059158843543474, "grad_norm": 1.2683649436398443, "learning_rate": 4.203191398393037e-06, "loss": 0.3874, "step": 19814 }, { "epoch": 0.7059515114808416, "grad_norm": 1.2913500368440576, "learning_rate": 4.2022511590526464e-06, "loss": 0.3955, "step": 19815 }, { "epoch": 0.7059871386073356, "grad_norm": 1.4118280029518133, "learning_rate": 4.201310996913337e-06, "loss": 0.4575, "step": 19816 }, { "epoch": 0.7060227657338297, "grad_norm": 1.3714180011277337, "learning_rate": 4.20037091198762e-06, "loss": 0.4481, "step": 19817 }, { "epoch": 0.7060583928603239, "grad_norm": 1.2918536864462593, "learning_rate": 4.19943090428802e-06, "loss": 0.3612, "step": 19818 }, { "epoch": 0.706094019986818, "grad_norm": 1.4342996303213609, "learning_rate": 4.1984909738270505e-06, "loss": 0.4133, "step": 19819 }, { "epoch": 0.706129647113312, "grad_norm": 1.9561540819506982, "learning_rate": 4.1975511206172295e-06, "loss": 0.445, "step": 19820 }, { "epoch": 0.7061652742398062, "grad_norm": 1.8001329161652426, "learning_rate": 4.196611344671073e-06, "loss": 0.5472, "step": 19821 }, { "epoch": 0.7062009013663003, "grad_norm": 1.3677997861072149, "learning_rate": 4.195671646001091e-06, "loss": 0.4714, "step": 19822 }, { "epoch": 0.7062365284927944, "grad_norm": 1.4430377941178643, "learning_rate": 4.194732024619801e-06, "loss": 0.4606, "step": 19823 }, { "epoch": 0.7062721556192886, "grad_norm": 1.433958785088551, "learning_rate": 4.193792480539708e-06, "loss": 0.4178, "step": 19824 }, { "epoch": 0.7063077827457827, "grad_norm": 1.447470471535626, "learning_rate": 4.192853013773327e-06, "loss": 0.3946, "step": 19825 }, { "epoch": 0.7063434098722767, "grad_norm": 1.5051278864626843, "learning_rate": 4.191913624333166e-06, "loss": 0.4746, "step": 19826 }, { "epoch": 0.7063790369987709, "grad_norm": 2.0089791727390183, "learning_rate": 4.190974312231739e-06, "loss": 0.5171, "step": 19827 }, { "epoch": 0.706414664125265, "grad_norm": 1.8360817095686057, "learning_rate": 4.190035077481546e-06, "loss": 0.46, "step": 19828 }, { "epoch": 0.7064502912517591, "grad_norm": 1.4224279505393151, "learning_rate": 4.189095920095096e-06, "loss": 0.471, "step": 19829 }, { "epoch": 0.7064859183782533, "grad_norm": 1.6533643118375951, "learning_rate": 4.188156840084899e-06, "loss": 0.4571, "step": 19830 }, { "epoch": 0.7065215455047473, "grad_norm": 1.6048088135222416, "learning_rate": 4.187217837463453e-06, "loss": 0.5145, "step": 19831 }, { "epoch": 0.7065571726312414, "grad_norm": 1.6788852729288528, "learning_rate": 4.186278912243268e-06, "loss": 0.5237, "step": 19832 }, { "epoch": 0.7065927997577356, "grad_norm": 1.642689742283475, "learning_rate": 4.1853400644368395e-06, "loss": 0.4821, "step": 19833 }, { "epoch": 0.7066284268842297, "grad_norm": 1.9305194183950358, "learning_rate": 4.1844012940566715e-06, "loss": 0.566, "step": 19834 }, { "epoch": 0.7066640540107237, "grad_norm": 1.5017381506582212, "learning_rate": 4.183462601115264e-06, "loss": 0.3879, "step": 19835 }, { "epoch": 0.7066996811372179, "grad_norm": 1.281331488037617, "learning_rate": 4.182523985625123e-06, "loss": 0.4823, "step": 19836 }, { "epoch": 0.706735308263712, "grad_norm": 1.3369573406149495, "learning_rate": 4.181585447598737e-06, "loss": 0.4238, "step": 19837 }, { "epoch": 0.7067709353902061, "grad_norm": 1.987765163077946, "learning_rate": 4.180646987048612e-06, "loss": 0.5671, "step": 19838 }, { "epoch": 0.7068065625167003, "grad_norm": 1.4732439481313115, "learning_rate": 4.1797086039872355e-06, "loss": 0.4068, "step": 19839 }, { "epoch": 0.7068421896431943, "grad_norm": 1.2877257440108099, "learning_rate": 4.178770298427107e-06, "loss": 0.4309, "step": 19840 }, { "epoch": 0.7068778167696884, "grad_norm": 1.3252953762448247, "learning_rate": 4.177832070380721e-06, "loss": 0.4515, "step": 19841 }, { "epoch": 0.7069134438961826, "grad_norm": 1.5204487794219421, "learning_rate": 4.1768939198605705e-06, "loss": 0.4273, "step": 19842 }, { "epoch": 0.7069490710226767, "grad_norm": 1.9267186742779676, "learning_rate": 4.17595584687915e-06, "loss": 0.4853, "step": 19843 }, { "epoch": 0.7069846981491708, "grad_norm": 1.3628023140302163, "learning_rate": 4.1750178514489456e-06, "loss": 0.441, "step": 19844 }, { "epoch": 0.7070203252756649, "grad_norm": 1.1969379731529937, "learning_rate": 4.174079933582454e-06, "loss": 0.4097, "step": 19845 }, { "epoch": 0.707055952402159, "grad_norm": 1.3232909463428426, "learning_rate": 4.173142093292156e-06, "loss": 0.4071, "step": 19846 }, { "epoch": 0.7070915795286531, "grad_norm": 1.854071472936443, "learning_rate": 4.1722043305905445e-06, "loss": 0.5081, "step": 19847 }, { "epoch": 0.7071272066551473, "grad_norm": 1.9143309713316863, "learning_rate": 4.171266645490105e-06, "loss": 0.5634, "step": 19848 }, { "epoch": 0.7071628337816414, "grad_norm": 1.386682631776783, "learning_rate": 4.170329038003328e-06, "loss": 0.4258, "step": 19849 }, { "epoch": 0.7071984609081354, "grad_norm": 1.1868006818995838, "learning_rate": 4.169391508142691e-06, "loss": 0.4141, "step": 19850 }, { "epoch": 0.7072340880346296, "grad_norm": 1.6082265050220168, "learning_rate": 4.168454055920681e-06, "loss": 0.4765, "step": 19851 }, { "epoch": 0.7072697151611237, "grad_norm": 1.3918734769612073, "learning_rate": 4.167516681349785e-06, "loss": 0.4283, "step": 19852 }, { "epoch": 0.7073053422876178, "grad_norm": 1.307440761745653, "learning_rate": 4.166579384442477e-06, "loss": 0.4403, "step": 19853 }, { "epoch": 0.707340969414112, "grad_norm": 1.3907104108023745, "learning_rate": 4.165642165211246e-06, "loss": 0.5307, "step": 19854 }, { "epoch": 0.707376596540606, "grad_norm": 1.348113359471402, "learning_rate": 4.164705023668564e-06, "loss": 0.3714, "step": 19855 }, { "epoch": 0.7074122236671001, "grad_norm": 1.6756581585530004, "learning_rate": 4.1637679598269136e-06, "loss": 0.5936, "step": 19856 }, { "epoch": 0.7074478507935943, "grad_norm": 1.552930337132645, "learning_rate": 4.162830973698771e-06, "loss": 0.5514, "step": 19857 }, { "epoch": 0.7074834779200884, "grad_norm": 1.436849529734789, "learning_rate": 4.161894065296618e-06, "loss": 0.4153, "step": 19858 }, { "epoch": 0.7075191050465824, "grad_norm": 1.609032824977728, "learning_rate": 4.160957234632923e-06, "loss": 0.4487, "step": 19859 }, { "epoch": 0.7075547321730766, "grad_norm": 1.5729960368031868, "learning_rate": 4.160020481720166e-06, "loss": 0.5345, "step": 19860 }, { "epoch": 0.7075903592995707, "grad_norm": 1.5834106562132486, "learning_rate": 4.159083806570816e-06, "loss": 0.5062, "step": 19861 }, { "epoch": 0.7076259864260648, "grad_norm": 1.619423845594983, "learning_rate": 4.158147209197347e-06, "loss": 0.588, "step": 19862 }, { "epoch": 0.707661613552559, "grad_norm": 1.6815388982485349, "learning_rate": 4.157210689612231e-06, "loss": 0.4229, "step": 19863 }, { "epoch": 0.707697240679053, "grad_norm": 1.6712608855679545, "learning_rate": 4.15627424782794e-06, "loss": 0.5003, "step": 19864 }, { "epoch": 0.7077328678055471, "grad_norm": 1.6901962171554321, "learning_rate": 4.155337883856943e-06, "loss": 0.4008, "step": 19865 }, { "epoch": 0.7077684949320413, "grad_norm": 1.6072953725539803, "learning_rate": 4.154401597711706e-06, "loss": 0.5101, "step": 19866 }, { "epoch": 0.7078041220585354, "grad_norm": 1.2464600726231314, "learning_rate": 4.1534653894046995e-06, "loss": 0.3878, "step": 19867 }, { "epoch": 0.7078397491850295, "grad_norm": 1.7558720385198725, "learning_rate": 4.152529258948384e-06, "loss": 0.507, "step": 19868 }, { "epoch": 0.7078753763115236, "grad_norm": 1.4477006093158058, "learning_rate": 4.151593206355233e-06, "loss": 0.5994, "step": 19869 }, { "epoch": 0.7079110034380177, "grad_norm": 1.749693758891754, "learning_rate": 4.1506572316377035e-06, "loss": 0.5418, "step": 19870 }, { "epoch": 0.7079466305645118, "grad_norm": 1.2701170075573507, "learning_rate": 4.14972133480826e-06, "loss": 0.4397, "step": 19871 }, { "epoch": 0.707982257691006, "grad_norm": 1.6491805929566536, "learning_rate": 4.148785515879367e-06, "loss": 0.4982, "step": 19872 }, { "epoch": 0.7080178848175, "grad_norm": 1.5261627105380755, "learning_rate": 4.147849774863488e-06, "loss": 0.4394, "step": 19873 }, { "epoch": 0.7080535119439941, "grad_norm": 1.4203119026045539, "learning_rate": 4.146914111773076e-06, "loss": 0.4161, "step": 19874 }, { "epoch": 0.7080891390704883, "grad_norm": 1.351891188363625, "learning_rate": 4.145978526620593e-06, "loss": 0.413, "step": 19875 }, { "epoch": 0.7081247661969824, "grad_norm": 1.6244698462658216, "learning_rate": 4.1450430194185025e-06, "loss": 0.5419, "step": 19876 }, { "epoch": 0.7081603933234765, "grad_norm": 1.6083682152937004, "learning_rate": 4.144107590179251e-06, "loss": 0.5054, "step": 19877 }, { "epoch": 0.7081960204499707, "grad_norm": 1.3059623198837134, "learning_rate": 4.143172238915302e-06, "loss": 0.385, "step": 19878 }, { "epoch": 0.7082316475764647, "grad_norm": 1.5916231626673394, "learning_rate": 4.142236965639107e-06, "loss": 0.4659, "step": 19879 }, { "epoch": 0.7082672747029588, "grad_norm": 2.462012753747669, "learning_rate": 4.1413017703631255e-06, "loss": 0.4669, "step": 19880 }, { "epoch": 0.708302901829453, "grad_norm": 1.4855832206697765, "learning_rate": 4.140366653099802e-06, "loss": 0.4068, "step": 19881 }, { "epoch": 0.7083385289559471, "grad_norm": 1.3463884265215076, "learning_rate": 4.139431613861595e-06, "loss": 0.4202, "step": 19882 }, { "epoch": 0.7083741560824411, "grad_norm": 1.4205269164419945, "learning_rate": 4.13849665266095e-06, "loss": 0.3795, "step": 19883 }, { "epoch": 0.7084097832089353, "grad_norm": 1.4475807954850637, "learning_rate": 4.137561769510322e-06, "loss": 0.3923, "step": 19884 }, { "epoch": 0.7084454103354294, "grad_norm": 1.3091598170116703, "learning_rate": 4.1366269644221505e-06, "loss": 0.3774, "step": 19885 }, { "epoch": 0.7084810374619235, "grad_norm": 1.280691881787627, "learning_rate": 4.135692237408891e-06, "loss": 0.4268, "step": 19886 }, { "epoch": 0.7085166645884177, "grad_norm": 1.3676952308151518, "learning_rate": 4.134757588482993e-06, "loss": 0.4245, "step": 19887 }, { "epoch": 0.7085522917149117, "grad_norm": 1.7042425942139876, "learning_rate": 4.133823017656894e-06, "loss": 0.4881, "step": 19888 }, { "epoch": 0.7085879188414058, "grad_norm": 2.1253597435282092, "learning_rate": 4.132888524943044e-06, "loss": 0.532, "step": 19889 }, { "epoch": 0.7086235459679, "grad_norm": 1.373145211637114, "learning_rate": 4.131954110353882e-06, "loss": 0.424, "step": 19890 }, { "epoch": 0.7086591730943941, "grad_norm": 1.2978243596183883, "learning_rate": 4.131019773901855e-06, "loss": 0.4411, "step": 19891 }, { "epoch": 0.7086948002208882, "grad_norm": 1.5847333309902445, "learning_rate": 4.130085515599399e-06, "loss": 0.4665, "step": 19892 }, { "epoch": 0.7087304273473823, "grad_norm": 1.790897784893759, "learning_rate": 4.1291513354589576e-06, "loss": 0.4358, "step": 19893 }, { "epoch": 0.7087660544738764, "grad_norm": 1.4319210198589005, "learning_rate": 4.128217233492968e-06, "loss": 0.3984, "step": 19894 }, { "epoch": 0.7088016816003705, "grad_norm": 1.32326952527652, "learning_rate": 4.127283209713875e-06, "loss": 0.429, "step": 19895 }, { "epoch": 0.7088373087268647, "grad_norm": 1.6417574383792368, "learning_rate": 4.126349264134106e-06, "loss": 0.5787, "step": 19896 }, { "epoch": 0.7088729358533588, "grad_norm": 2.3448780090446855, "learning_rate": 4.125415396766107e-06, "loss": 0.5597, "step": 19897 }, { "epoch": 0.7089085629798528, "grad_norm": 1.6516624240889493, "learning_rate": 4.124481607622304e-06, "loss": 0.4026, "step": 19898 }, { "epoch": 0.708944190106347, "grad_norm": 1.6280002246921528, "learning_rate": 4.123547896715135e-06, "loss": 0.468, "step": 19899 }, { "epoch": 0.7089798172328411, "grad_norm": 1.2373859874694897, "learning_rate": 4.1226142640570335e-06, "loss": 0.4439, "step": 19900 }, { "epoch": 0.7090154443593352, "grad_norm": 1.3779503962719666, "learning_rate": 4.121680709660431e-06, "loss": 0.4068, "step": 19901 }, { "epoch": 0.7090510714858294, "grad_norm": 1.6041960118580711, "learning_rate": 4.120747233537762e-06, "loss": 0.5204, "step": 19902 }, { "epoch": 0.7090866986123234, "grad_norm": 1.2897613623819857, "learning_rate": 4.119813835701449e-06, "loss": 0.4087, "step": 19903 }, { "epoch": 0.7091223257388175, "grad_norm": 1.255988511643403, "learning_rate": 4.1188805161639285e-06, "loss": 0.4167, "step": 19904 }, { "epoch": 0.7091579528653117, "grad_norm": 1.2849409014439919, "learning_rate": 4.117947274937622e-06, "loss": 0.4091, "step": 19905 }, { "epoch": 0.7091935799918058, "grad_norm": 1.3265388399341262, "learning_rate": 4.117014112034961e-06, "loss": 0.451, "step": 19906 }, { "epoch": 0.7092292071182998, "grad_norm": 2.1115896503517906, "learning_rate": 4.116081027468367e-06, "loss": 0.5357, "step": 19907 }, { "epoch": 0.709264834244794, "grad_norm": 1.27012080095152, "learning_rate": 4.115148021250266e-06, "loss": 0.4332, "step": 19908 }, { "epoch": 0.7093004613712881, "grad_norm": 1.3268970593110716, "learning_rate": 4.114215093393082e-06, "loss": 0.3868, "step": 19909 }, { "epoch": 0.7093360884977822, "grad_norm": 1.256048556829488, "learning_rate": 4.1132822439092375e-06, "loss": 0.413, "step": 19910 }, { "epoch": 0.7093717156242764, "grad_norm": 2.0294399344259895, "learning_rate": 4.112349472811159e-06, "loss": 0.533, "step": 19911 }, { "epoch": 0.7094073427507704, "grad_norm": 1.8126598826984446, "learning_rate": 4.111416780111259e-06, "loss": 0.537, "step": 19912 }, { "epoch": 0.7094429698772645, "grad_norm": 1.4445759717745112, "learning_rate": 4.110484165821963e-06, "loss": 0.4391, "step": 19913 }, { "epoch": 0.7094785970037587, "grad_norm": 1.4152060051018605, "learning_rate": 4.109551629955685e-06, "loss": 0.4269, "step": 19914 }, { "epoch": 0.7095142241302528, "grad_norm": 1.6134548756817744, "learning_rate": 4.108619172524842e-06, "loss": 0.4621, "step": 19915 }, { "epoch": 0.7095498512567469, "grad_norm": 1.3106458543415065, "learning_rate": 4.107686793541854e-06, "loss": 0.4298, "step": 19916 }, { "epoch": 0.709585478383241, "grad_norm": 1.6625187949384441, "learning_rate": 4.106754493019138e-06, "loss": 0.4154, "step": 19917 }, { "epoch": 0.7096211055097351, "grad_norm": 1.1742682105991882, "learning_rate": 4.105822270969102e-06, "loss": 0.384, "step": 19918 }, { "epoch": 0.7096567326362292, "grad_norm": 1.2044324501730448, "learning_rate": 4.104890127404165e-06, "loss": 0.3789, "step": 19919 }, { "epoch": 0.7096923597627234, "grad_norm": 1.2348582981088136, "learning_rate": 4.103958062336733e-06, "loss": 0.4233, "step": 19920 }, { "epoch": 0.7097279868892175, "grad_norm": 1.3997700247466411, "learning_rate": 4.103026075779221e-06, "loss": 0.4345, "step": 19921 }, { "epoch": 0.7097636140157115, "grad_norm": 1.4284823028044626, "learning_rate": 4.102094167744043e-06, "loss": 0.396, "step": 19922 }, { "epoch": 0.7097992411422057, "grad_norm": 1.7752436940123584, "learning_rate": 4.101162338243595e-06, "loss": 0.5485, "step": 19923 }, { "epoch": 0.7098348682686998, "grad_norm": 1.7153250558769688, "learning_rate": 4.1002305872903005e-06, "loss": 0.5402, "step": 19924 }, { "epoch": 0.7098704953951939, "grad_norm": 1.4838716719632759, "learning_rate": 4.099298914896556e-06, "loss": 0.5658, "step": 19925 }, { "epoch": 0.709906122521688, "grad_norm": 1.6034666882364754, "learning_rate": 4.098367321074775e-06, "loss": 0.5302, "step": 19926 }, { "epoch": 0.7099417496481821, "grad_norm": 1.757385186853698, "learning_rate": 4.097435805837354e-06, "loss": 0.5002, "step": 19927 }, { "epoch": 0.7099773767746762, "grad_norm": 1.335977439342748, "learning_rate": 4.0965043691967045e-06, "loss": 0.442, "step": 19928 }, { "epoch": 0.7100130039011704, "grad_norm": 1.3219896685524974, "learning_rate": 4.095573011165223e-06, "loss": 0.3801, "step": 19929 }, { "epoch": 0.7100486310276645, "grad_norm": 1.6994040063583613, "learning_rate": 4.0946417317553124e-06, "loss": 0.5825, "step": 19930 }, { "epoch": 0.7100842581541585, "grad_norm": 1.684577793203841, "learning_rate": 4.093710530979375e-06, "loss": 0.4834, "step": 19931 }, { "epoch": 0.7101198852806527, "grad_norm": 1.4485242962112121, "learning_rate": 4.092779408849811e-06, "loss": 0.372, "step": 19932 }, { "epoch": 0.7101555124071468, "grad_norm": 1.4077225962160007, "learning_rate": 4.0918483653790205e-06, "loss": 0.4512, "step": 19933 }, { "epoch": 0.7101911395336409, "grad_norm": 1.7320767337945908, "learning_rate": 4.0909174005793965e-06, "loss": 0.5338, "step": 19934 }, { "epoch": 0.7102267666601351, "grad_norm": 1.3109626476353746, "learning_rate": 4.0899865144633395e-06, "loss": 0.4263, "step": 19935 }, { "epoch": 0.7102623937866291, "grad_norm": 1.2931054256432888, "learning_rate": 4.08905570704324e-06, "loss": 0.4072, "step": 19936 }, { "epoch": 0.7102980209131232, "grad_norm": 1.4454021061629756, "learning_rate": 4.088124978331496e-06, "loss": 0.4493, "step": 19937 }, { "epoch": 0.7103336480396174, "grad_norm": 1.2810967059070049, "learning_rate": 4.087194328340499e-06, "loss": 0.4245, "step": 19938 }, { "epoch": 0.7103692751661115, "grad_norm": 1.6501691889100425, "learning_rate": 4.086263757082646e-06, "loss": 0.4397, "step": 19939 }, { "epoch": 0.7104049022926056, "grad_norm": 2.298130863506216, "learning_rate": 4.085333264570322e-06, "loss": 0.4747, "step": 19940 }, { "epoch": 0.7104405294190997, "grad_norm": 1.319644749766759, "learning_rate": 4.084402850815922e-06, "loss": 0.4541, "step": 19941 }, { "epoch": 0.7104761565455938, "grad_norm": 1.7406076519009732, "learning_rate": 4.083472515831829e-06, "loss": 0.5653, "step": 19942 }, { "epoch": 0.7105117836720879, "grad_norm": 1.712155155454284, "learning_rate": 4.08254225963044e-06, "loss": 0.4728, "step": 19943 }, { "epoch": 0.7105474107985821, "grad_norm": 1.4959870804868276, "learning_rate": 4.081612082224132e-06, "loss": 0.4823, "step": 19944 }, { "epoch": 0.7105830379250762, "grad_norm": 1.5380097879958476, "learning_rate": 4.080681983625297e-06, "loss": 0.4631, "step": 19945 }, { "epoch": 0.7106186650515702, "grad_norm": 1.5052307793534268, "learning_rate": 4.079751963846317e-06, "loss": 0.4633, "step": 19946 }, { "epoch": 0.7106542921780644, "grad_norm": 1.5377351849075633, "learning_rate": 4.078822022899579e-06, "loss": 0.5185, "step": 19947 }, { "epoch": 0.7106899193045585, "grad_norm": 1.8397208412852812, "learning_rate": 4.077892160797468e-06, "loss": 0.4978, "step": 19948 }, { "epoch": 0.7107255464310526, "grad_norm": 1.1958498497010162, "learning_rate": 4.076962377552357e-06, "loss": 0.3894, "step": 19949 }, { "epoch": 0.7107611735575468, "grad_norm": 1.4381130511511233, "learning_rate": 4.0760326731766376e-06, "loss": 0.4193, "step": 19950 }, { "epoch": 0.7107968006840408, "grad_norm": 1.1819707949124014, "learning_rate": 4.075103047682679e-06, "loss": 0.4146, "step": 19951 }, { "epoch": 0.7108324278105349, "grad_norm": 1.790041441801792, "learning_rate": 4.074173501082865e-06, "loss": 0.4628, "step": 19952 }, { "epoch": 0.7108680549370291, "grad_norm": 1.5284808459696568, "learning_rate": 4.073244033389572e-06, "loss": 0.5347, "step": 19953 }, { "epoch": 0.7109036820635232, "grad_norm": 1.7865777345797702, "learning_rate": 4.0723146446151815e-06, "loss": 0.449, "step": 19954 }, { "epoch": 0.7109393091900172, "grad_norm": 1.7444372112662596, "learning_rate": 4.071385334772061e-06, "loss": 0.5385, "step": 19955 }, { "epoch": 0.7109749363165114, "grad_norm": 1.4789748532647156, "learning_rate": 4.070456103872589e-06, "loss": 0.3965, "step": 19956 }, { "epoch": 0.7110105634430055, "grad_norm": 1.3490338737038043, "learning_rate": 4.0695269519291416e-06, "loss": 0.4699, "step": 19957 }, { "epoch": 0.7110461905694996, "grad_norm": 1.4405955060213327, "learning_rate": 4.068597878954085e-06, "loss": 0.4111, "step": 19958 }, { "epoch": 0.7110818176959938, "grad_norm": 1.3200562305125632, "learning_rate": 4.067668884959795e-06, "loss": 0.4057, "step": 19959 }, { "epoch": 0.7111174448224878, "grad_norm": 2.1113455128408103, "learning_rate": 4.0667399699586355e-06, "loss": 0.4503, "step": 19960 }, { "epoch": 0.7111530719489819, "grad_norm": 1.5898017741725916, "learning_rate": 4.065811133962987e-06, "loss": 0.4016, "step": 19961 }, { "epoch": 0.7111886990754761, "grad_norm": 1.6689276778146789, "learning_rate": 4.064882376985206e-06, "loss": 0.5164, "step": 19962 }, { "epoch": 0.7112243262019702, "grad_norm": 1.4649585231213225, "learning_rate": 4.06395369903767e-06, "loss": 0.5141, "step": 19963 }, { "epoch": 0.7112599533284643, "grad_norm": 1.7622796789759212, "learning_rate": 4.063025100132735e-06, "loss": 0.5308, "step": 19964 }, { "epoch": 0.7112955804549584, "grad_norm": 1.5863821567069545, "learning_rate": 4.062096580282775e-06, "loss": 0.4812, "step": 19965 }, { "epoch": 0.7113312075814525, "grad_norm": 1.4496598783737793, "learning_rate": 4.061168139500145e-06, "loss": 0.3985, "step": 19966 }, { "epoch": 0.7113668347079466, "grad_norm": 1.6499919296901893, "learning_rate": 4.0602397777972125e-06, "loss": 0.4266, "step": 19967 }, { "epoch": 0.7114024618344408, "grad_norm": 1.5670220404610338, "learning_rate": 4.059311495186338e-06, "loss": 0.5061, "step": 19968 }, { "epoch": 0.7114380889609349, "grad_norm": 1.8131130234426012, "learning_rate": 4.058383291679884e-06, "loss": 0.5528, "step": 19969 }, { "epoch": 0.7114737160874289, "grad_norm": 1.2629813819264863, "learning_rate": 4.057455167290213e-06, "loss": 0.3799, "step": 19970 }, { "epoch": 0.7115093432139231, "grad_norm": 1.6525637399779325, "learning_rate": 4.056527122029677e-06, "loss": 0.4529, "step": 19971 }, { "epoch": 0.7115449703404172, "grad_norm": 1.773885237115, "learning_rate": 4.0555991559106385e-06, "loss": 0.4917, "step": 19972 }, { "epoch": 0.7115805974669113, "grad_norm": 1.5568988124960244, "learning_rate": 4.05467126894545e-06, "loss": 0.4811, "step": 19973 }, { "epoch": 0.7116162245934055, "grad_norm": 1.6758622842554989, "learning_rate": 4.05374346114647e-06, "loss": 0.5196, "step": 19974 }, { "epoch": 0.7116518517198995, "grad_norm": 1.598644280397532, "learning_rate": 4.05281573252605e-06, "loss": 0.5373, "step": 19975 }, { "epoch": 0.7116874788463936, "grad_norm": 1.6384248992888333, "learning_rate": 4.05188808309655e-06, "loss": 0.4091, "step": 19976 }, { "epoch": 0.7117231059728878, "grad_norm": 1.5781835578429066, "learning_rate": 4.050960512870313e-06, "loss": 0.5214, "step": 19977 }, { "epoch": 0.7117587330993819, "grad_norm": 1.7002165462653018, "learning_rate": 4.0500330218596965e-06, "loss": 0.5502, "step": 19978 }, { "epoch": 0.7117943602258759, "grad_norm": 1.492366640445102, "learning_rate": 4.049105610077052e-06, "loss": 0.3573, "step": 19979 }, { "epoch": 0.7118299873523701, "grad_norm": 1.4612962009929504, "learning_rate": 4.048178277534722e-06, "loss": 0.534, "step": 19980 }, { "epoch": 0.7118656144788642, "grad_norm": 1.5875832833979457, "learning_rate": 4.0472510242450614e-06, "loss": 0.411, "step": 19981 }, { "epoch": 0.7119012416053583, "grad_norm": 1.764585637246327, "learning_rate": 4.046323850220412e-06, "loss": 0.5144, "step": 19982 }, { "epoch": 0.7119368687318525, "grad_norm": 1.6567233361842233, "learning_rate": 4.045396755473121e-06, "loss": 0.5647, "step": 19983 }, { "epoch": 0.7119724958583465, "grad_norm": 1.787906111194887, "learning_rate": 4.0444697400155345e-06, "loss": 0.441, "step": 19984 }, { "epoch": 0.7120081229848406, "grad_norm": 1.408777768860002, "learning_rate": 4.043542803859999e-06, "loss": 0.4366, "step": 19985 }, { "epoch": 0.7120437501113348, "grad_norm": 1.7126317104041484, "learning_rate": 4.042615947018851e-06, "loss": 0.5469, "step": 19986 }, { "epoch": 0.7120793772378289, "grad_norm": 1.4872355837608349, "learning_rate": 4.041689169504439e-06, "loss": 0.3786, "step": 19987 }, { "epoch": 0.712115004364323, "grad_norm": 1.3803521541370092, "learning_rate": 4.040762471329097e-06, "loss": 0.4205, "step": 19988 }, { "epoch": 0.7121506314908171, "grad_norm": 1.9593614168738773, "learning_rate": 4.039835852505169e-06, "loss": 0.5252, "step": 19989 }, { "epoch": 0.7121862586173112, "grad_norm": 1.7327169989730744, "learning_rate": 4.038909313044991e-06, "loss": 0.454, "step": 19990 }, { "epoch": 0.7122218857438053, "grad_norm": 1.5404991826050956, "learning_rate": 4.037982852960902e-06, "loss": 0.4162, "step": 19991 }, { "epoch": 0.7122575128702995, "grad_norm": 1.7462634274182613, "learning_rate": 4.037056472265244e-06, "loss": 0.4252, "step": 19992 }, { "epoch": 0.7122931399967936, "grad_norm": 1.5298726408332641, "learning_rate": 4.036130170970342e-06, "loss": 0.5016, "step": 19993 }, { "epoch": 0.7123287671232876, "grad_norm": 1.2174206662596563, "learning_rate": 4.035203949088538e-06, "loss": 0.4259, "step": 19994 }, { "epoch": 0.7123643942497818, "grad_norm": 1.859789216805346, "learning_rate": 4.03427780663216e-06, "loss": 0.5085, "step": 19995 }, { "epoch": 0.7124000213762759, "grad_norm": 2.4418704351584664, "learning_rate": 4.033351743613546e-06, "loss": 0.4808, "step": 19996 }, { "epoch": 0.71243564850277, "grad_norm": 1.7527932314197625, "learning_rate": 4.0324257600450215e-06, "loss": 0.3778, "step": 19997 }, { "epoch": 0.7124712756292642, "grad_norm": 1.6018330182671456, "learning_rate": 4.0314998559389185e-06, "loss": 0.4115, "step": 19998 }, { "epoch": 0.7125069027557582, "grad_norm": 1.5844467343939053, "learning_rate": 4.030574031307566e-06, "loss": 0.4262, "step": 19999 }, { "epoch": 0.7125425298822523, "grad_norm": 1.4025797625686083, "learning_rate": 4.0296482861632965e-06, "loss": 0.4273, "step": 20000 }, { "epoch": 0.7125781570087465, "grad_norm": 2.0850983300378427, "learning_rate": 4.0287226205184294e-06, "loss": 0.5624, "step": 20001 }, { "epoch": 0.7126137841352406, "grad_norm": 2.503438199429491, "learning_rate": 4.027797034385295e-06, "loss": 0.4202, "step": 20002 }, { "epoch": 0.7126494112617346, "grad_norm": 1.4437350052574236, "learning_rate": 4.02687152777622e-06, "loss": 0.363, "step": 20003 }, { "epoch": 0.7126850383882288, "grad_norm": 1.5247451481643877, "learning_rate": 4.0259461007035225e-06, "loss": 0.5942, "step": 20004 }, { "epoch": 0.7127206655147229, "grad_norm": 1.4686828272818377, "learning_rate": 4.0250207531795284e-06, "loss": 0.49, "step": 20005 }, { "epoch": 0.712756292641217, "grad_norm": 2.2387125622055826, "learning_rate": 4.024095485216559e-06, "loss": 0.408, "step": 20006 }, { "epoch": 0.7127919197677112, "grad_norm": 1.4152821275281655, "learning_rate": 4.023170296826939e-06, "loss": 0.4669, "step": 20007 }, { "epoch": 0.7128275468942052, "grad_norm": 1.6277209853144334, "learning_rate": 4.022245188022981e-06, "loss": 0.492, "step": 20008 }, { "epoch": 0.7128631740206993, "grad_norm": 1.3119692387561281, "learning_rate": 4.0213201588170094e-06, "loss": 0.4284, "step": 20009 }, { "epoch": 0.7128988011471935, "grad_norm": 1.4183023516612256, "learning_rate": 4.020395209221335e-06, "loss": 0.539, "step": 20010 }, { "epoch": 0.7129344282736876, "grad_norm": 2.48191945689394, "learning_rate": 4.019470339248278e-06, "loss": 0.5328, "step": 20011 }, { "epoch": 0.7129700554001817, "grad_norm": 1.7689727349886946, "learning_rate": 4.018545548910153e-06, "loss": 0.3779, "step": 20012 }, { "epoch": 0.7130056825266758, "grad_norm": 1.3807466825814658, "learning_rate": 4.017620838219276e-06, "loss": 0.4165, "step": 20013 }, { "epoch": 0.7130413096531699, "grad_norm": 1.5535986394424184, "learning_rate": 4.016696207187962e-06, "loss": 0.5343, "step": 20014 }, { "epoch": 0.713076936779664, "grad_norm": 1.2329364456782315, "learning_rate": 4.015771655828517e-06, "loss": 0.4126, "step": 20015 }, { "epoch": 0.7131125639061582, "grad_norm": 1.8512474832039663, "learning_rate": 4.014847184153258e-06, "loss": 0.4976, "step": 20016 }, { "epoch": 0.7131481910326523, "grad_norm": 1.65657503720933, "learning_rate": 4.013922792174488e-06, "loss": 0.4831, "step": 20017 }, { "epoch": 0.7131838181591463, "grad_norm": 1.3864507825595003, "learning_rate": 4.012998479904525e-06, "loss": 0.3992, "step": 20018 }, { "epoch": 0.7132194452856405, "grad_norm": 1.5998568199798766, "learning_rate": 4.012074247355667e-06, "loss": 0.4695, "step": 20019 }, { "epoch": 0.7132550724121346, "grad_norm": 1.3160088978239297, "learning_rate": 4.011150094540227e-06, "loss": 0.5299, "step": 20020 }, { "epoch": 0.7132906995386287, "grad_norm": 1.571949395471788, "learning_rate": 4.0102260214705094e-06, "loss": 0.3831, "step": 20021 }, { "epoch": 0.7133263266651229, "grad_norm": 1.7691544474928451, "learning_rate": 4.009302028158821e-06, "loss": 0.5254, "step": 20022 }, { "epoch": 0.7133619537916169, "grad_norm": 1.3512385653068155, "learning_rate": 4.00837811461746e-06, "loss": 0.4327, "step": 20023 }, { "epoch": 0.713397580918111, "grad_norm": 1.725377424776928, "learning_rate": 4.007454280858733e-06, "loss": 0.49, "step": 20024 }, { "epoch": 0.7134332080446052, "grad_norm": 1.421434857862973, "learning_rate": 4.006530526894945e-06, "loss": 0.4333, "step": 20025 }, { "epoch": 0.7134688351710993, "grad_norm": 1.2930432706460488, "learning_rate": 4.005606852738388e-06, "loss": 0.4418, "step": 20026 }, { "epoch": 0.7135044622975933, "grad_norm": 1.6844701524324317, "learning_rate": 4.0046832584013666e-06, "loss": 0.5447, "step": 20027 }, { "epoch": 0.7135400894240875, "grad_norm": 1.2807507619710843, "learning_rate": 4.003759743896176e-06, "loss": 0.3963, "step": 20028 }, { "epoch": 0.7135757165505816, "grad_norm": 1.7559816230639513, "learning_rate": 4.002836309235121e-06, "loss": 0.4782, "step": 20029 }, { "epoch": 0.7136113436770757, "grad_norm": 1.3967602246371902, "learning_rate": 4.001912954430487e-06, "loss": 0.4048, "step": 20030 }, { "epoch": 0.7136469708035699, "grad_norm": 1.7374561233320538, "learning_rate": 4.000989679494579e-06, "loss": 0.4607, "step": 20031 }, { "epoch": 0.7136825979300639, "grad_norm": 1.3239284637133064, "learning_rate": 4.000066484439683e-06, "loss": 0.4637, "step": 20032 }, { "epoch": 0.713718225056558, "grad_norm": 1.7604772075466582, "learning_rate": 3.999143369278099e-06, "loss": 0.5843, "step": 20033 }, { "epoch": 0.7137538521830522, "grad_norm": 1.3224996726467988, "learning_rate": 3.998220334022111e-06, "loss": 0.4134, "step": 20034 }, { "epoch": 0.7137894793095463, "grad_norm": 1.3485905567213028, "learning_rate": 3.997297378684015e-06, "loss": 0.4392, "step": 20035 }, { "epoch": 0.7138251064360404, "grad_norm": 2.264634762790556, "learning_rate": 3.9963745032761005e-06, "loss": 0.4988, "step": 20036 }, { "epoch": 0.7138607335625345, "grad_norm": 1.4442666958500257, "learning_rate": 3.995451707810655e-06, "loss": 0.4828, "step": 20037 }, { "epoch": 0.7138963606890286, "grad_norm": 1.4502368623565318, "learning_rate": 3.9945289922999705e-06, "loss": 0.4628, "step": 20038 }, { "epoch": 0.7139319878155227, "grad_norm": 1.3355801266867708, "learning_rate": 3.9936063567563275e-06, "loss": 0.4319, "step": 20039 }, { "epoch": 0.7139676149420169, "grad_norm": 1.7898242967475422, "learning_rate": 3.992683801192016e-06, "loss": 0.5668, "step": 20040 }, { "epoch": 0.714003242068511, "grad_norm": 1.2239098962619508, "learning_rate": 3.9917613256193164e-06, "loss": 0.3727, "step": 20041 }, { "epoch": 0.714038869195005, "grad_norm": 1.4471409458433464, "learning_rate": 3.990838930050513e-06, "loss": 0.4718, "step": 20042 }, { "epoch": 0.7140744963214992, "grad_norm": 1.454753433460766, "learning_rate": 3.989916614497891e-06, "loss": 0.5066, "step": 20043 }, { "epoch": 0.7141101234479933, "grad_norm": 2.29753500243131, "learning_rate": 3.988994378973733e-06, "loss": 0.513, "step": 20044 }, { "epoch": 0.7141457505744874, "grad_norm": 1.4256080240919364, "learning_rate": 3.988072223490312e-06, "loss": 0.3765, "step": 20045 }, { "epoch": 0.7141813777009816, "grad_norm": 1.3114429753995793, "learning_rate": 3.9871501480599154e-06, "loss": 0.4204, "step": 20046 }, { "epoch": 0.7142170048274756, "grad_norm": 1.3350461637834306, "learning_rate": 3.986228152694813e-06, "loss": 0.4412, "step": 20047 }, { "epoch": 0.7142526319539697, "grad_norm": 1.6171452232893835, "learning_rate": 3.9853062374072884e-06, "loss": 0.4402, "step": 20048 }, { "epoch": 0.7142882590804639, "grad_norm": 1.8597308701502357, "learning_rate": 3.984384402209613e-06, "loss": 0.4938, "step": 20049 }, { "epoch": 0.714323886206958, "grad_norm": 1.6757043072998472, "learning_rate": 3.983462647114065e-06, "loss": 0.4227, "step": 20050 }, { "epoch": 0.714359513333452, "grad_norm": 1.797446027701157, "learning_rate": 3.98254097213292e-06, "loss": 0.5949, "step": 20051 }, { "epoch": 0.7143951404599462, "grad_norm": 1.2436022591469154, "learning_rate": 3.981619377278445e-06, "loss": 0.4029, "step": 20052 }, { "epoch": 0.7144307675864403, "grad_norm": 1.77915733642752, "learning_rate": 3.980697862562918e-06, "loss": 0.5247, "step": 20053 }, { "epoch": 0.7144663947129344, "grad_norm": 1.670931576589851, "learning_rate": 3.979776427998603e-06, "loss": 0.4993, "step": 20054 }, { "epoch": 0.7145020218394286, "grad_norm": 1.3789825873607813, "learning_rate": 3.978855073597776e-06, "loss": 0.4054, "step": 20055 }, { "epoch": 0.7145376489659226, "grad_norm": 2.0092713739940153, "learning_rate": 3.977933799372699e-06, "loss": 0.6023, "step": 20056 }, { "epoch": 0.7145732760924167, "grad_norm": 1.6607836428683065, "learning_rate": 3.977012605335642e-06, "loss": 0.6386, "step": 20057 }, { "epoch": 0.7146089032189109, "grad_norm": 1.5400347643623902, "learning_rate": 3.9760914914988716e-06, "loss": 0.4093, "step": 20058 }, { "epoch": 0.714644530345405, "grad_norm": 1.3646243728697598, "learning_rate": 3.975170457874654e-06, "loss": 0.4161, "step": 20059 }, { "epoch": 0.7146801574718991, "grad_norm": 1.5119886995552354, "learning_rate": 3.974249504475256e-06, "loss": 0.5164, "step": 20060 }, { "epoch": 0.7147157845983932, "grad_norm": 1.9529233617039277, "learning_rate": 3.9733286313129346e-06, "loss": 0.5523, "step": 20061 }, { "epoch": 0.7147514117248873, "grad_norm": 1.6042481043667067, "learning_rate": 3.972407838399959e-06, "loss": 0.4566, "step": 20062 }, { "epoch": 0.7147870388513814, "grad_norm": 1.960572428880878, "learning_rate": 3.971487125748581e-06, "loss": 0.5683, "step": 20063 }, { "epoch": 0.7148226659778756, "grad_norm": 1.6017287527676822, "learning_rate": 3.9705664933710665e-06, "loss": 0.5397, "step": 20064 }, { "epoch": 0.7148582931043697, "grad_norm": 1.842306188334068, "learning_rate": 3.969645941279674e-06, "loss": 0.5434, "step": 20065 }, { "epoch": 0.7148939202308637, "grad_norm": 1.8399671525743366, "learning_rate": 3.9687254694866625e-06, "loss": 0.4345, "step": 20066 }, { "epoch": 0.7149295473573579, "grad_norm": 1.6125488662052418, "learning_rate": 3.9678050780042855e-06, "loss": 0.5261, "step": 20067 }, { "epoch": 0.714965174483852, "grad_norm": 1.4203041198524629, "learning_rate": 3.966884766844803e-06, "loss": 0.374, "step": 20068 }, { "epoch": 0.7150008016103461, "grad_norm": 1.6623576395010151, "learning_rate": 3.9659645360204625e-06, "loss": 0.5346, "step": 20069 }, { "epoch": 0.7150364287368403, "grad_norm": 1.9265653471735609, "learning_rate": 3.965044385543526e-06, "loss": 0.3559, "step": 20070 }, { "epoch": 0.7150720558633343, "grad_norm": 1.3236104212685214, "learning_rate": 3.964124315426237e-06, "loss": 0.4244, "step": 20071 }, { "epoch": 0.7151076829898284, "grad_norm": 1.5343530712634437, "learning_rate": 3.963204325680853e-06, "loss": 0.4117, "step": 20072 }, { "epoch": 0.7151433101163226, "grad_norm": 1.6575381798909397, "learning_rate": 3.962284416319623e-06, "loss": 0.4426, "step": 20073 }, { "epoch": 0.7151789372428167, "grad_norm": 1.854720397435537, "learning_rate": 3.9613645873547944e-06, "loss": 0.5581, "step": 20074 }, { "epoch": 0.7152145643693107, "grad_norm": 1.335824712236117, "learning_rate": 3.960444838798623e-06, "loss": 0.4034, "step": 20075 }, { "epoch": 0.7152501914958049, "grad_norm": 1.2835761280631128, "learning_rate": 3.959525170663344e-06, "loss": 0.343, "step": 20076 }, { "epoch": 0.715285818622299, "grad_norm": 1.6173775859891266, "learning_rate": 3.9586055829612145e-06, "loss": 0.4298, "step": 20077 }, { "epoch": 0.7153214457487931, "grad_norm": 1.1678178552053824, "learning_rate": 3.95768607570447e-06, "loss": 0.4287, "step": 20078 }, { "epoch": 0.7153570728752873, "grad_norm": 1.5222587573499897, "learning_rate": 3.956766648905359e-06, "loss": 0.41, "step": 20079 }, { "epoch": 0.7153927000017813, "grad_norm": 1.2525644993654639, "learning_rate": 3.955847302576125e-06, "loss": 0.3791, "step": 20080 }, { "epoch": 0.7154283271282754, "grad_norm": 1.2594482367029813, "learning_rate": 3.954928036729011e-06, "loss": 0.4264, "step": 20081 }, { "epoch": 0.7154639542547696, "grad_norm": 1.4197537034506738, "learning_rate": 3.954008851376252e-06, "loss": 0.4199, "step": 20082 }, { "epoch": 0.7154995813812637, "grad_norm": 1.242372051538043, "learning_rate": 3.953089746530091e-06, "loss": 0.3945, "step": 20083 }, { "epoch": 0.7155352085077578, "grad_norm": 1.2153877877017545, "learning_rate": 3.952170722202771e-06, "loss": 0.4108, "step": 20084 }, { "epoch": 0.715570835634252, "grad_norm": 1.9080076070267793, "learning_rate": 3.951251778406521e-06, "loss": 0.5355, "step": 20085 }, { "epoch": 0.715606462760746, "grad_norm": 1.697212400327505, "learning_rate": 3.950332915153585e-06, "loss": 0.5525, "step": 20086 }, { "epoch": 0.7156420898872401, "grad_norm": 1.7110837618445633, "learning_rate": 3.949414132456187e-06, "loss": 0.5082, "step": 20087 }, { "epoch": 0.7156777170137343, "grad_norm": 1.3145919803396418, "learning_rate": 3.948495430326576e-06, "loss": 0.4438, "step": 20088 }, { "epoch": 0.7157133441402284, "grad_norm": 2.087885445354135, "learning_rate": 3.947576808776976e-06, "loss": 0.5022, "step": 20089 }, { "epoch": 0.7157489712667224, "grad_norm": 1.3843330966665093, "learning_rate": 3.946658267819624e-06, "loss": 0.4087, "step": 20090 }, { "epoch": 0.7157845983932166, "grad_norm": 1.8865273937636784, "learning_rate": 3.945739807466744e-06, "loss": 0.5503, "step": 20091 }, { "epoch": 0.7158202255197107, "grad_norm": 1.231168119442237, "learning_rate": 3.944821427730574e-06, "loss": 0.4195, "step": 20092 }, { "epoch": 0.7158558526462048, "grad_norm": 1.9875423078116938, "learning_rate": 3.943903128623336e-06, "loss": 0.4461, "step": 20093 }, { "epoch": 0.715891479772699, "grad_norm": 1.8105926037983608, "learning_rate": 3.94298491015726e-06, "loss": 0.4987, "step": 20094 }, { "epoch": 0.715927106899193, "grad_norm": 1.4899015313581299, "learning_rate": 3.942066772344574e-06, "loss": 0.5151, "step": 20095 }, { "epoch": 0.7159627340256871, "grad_norm": 1.271500381664386, "learning_rate": 3.941148715197503e-06, "loss": 0.3961, "step": 20096 }, { "epoch": 0.7159983611521813, "grad_norm": 1.5102989161159417, "learning_rate": 3.940230738728276e-06, "loss": 0.488, "step": 20097 }, { "epoch": 0.7160339882786754, "grad_norm": 1.7871851773351202, "learning_rate": 3.9393128429491075e-06, "loss": 0.5791, "step": 20098 }, { "epoch": 0.7160696154051694, "grad_norm": 1.258147277619964, "learning_rate": 3.938395027872228e-06, "loss": 0.4062, "step": 20099 }, { "epoch": 0.7161052425316636, "grad_norm": 1.2656496718191972, "learning_rate": 3.937477293509853e-06, "loss": 0.3944, "step": 20100 }, { "epoch": 0.7161408696581577, "grad_norm": 1.269510130781158, "learning_rate": 3.936559639874205e-06, "loss": 0.4455, "step": 20101 }, { "epoch": 0.7161764967846518, "grad_norm": 1.5496299341872422, "learning_rate": 3.935642066977504e-06, "loss": 0.4498, "step": 20102 }, { "epoch": 0.716212123911146, "grad_norm": 1.2471272376778573, "learning_rate": 3.93472457483197e-06, "loss": 0.4166, "step": 20103 }, { "epoch": 0.71624775103764, "grad_norm": 1.1514393029620422, "learning_rate": 3.933807163449815e-06, "loss": 0.4025, "step": 20104 }, { "epoch": 0.7162833781641341, "grad_norm": 1.508320706394318, "learning_rate": 3.932889832843257e-06, "loss": 0.448, "step": 20105 }, { "epoch": 0.7163190052906283, "grad_norm": 1.5282209678732321, "learning_rate": 3.931972583024516e-06, "loss": 0.5079, "step": 20106 }, { "epoch": 0.7163546324171224, "grad_norm": 1.5421444685165635, "learning_rate": 3.931055414005798e-06, "loss": 0.4928, "step": 20107 }, { "epoch": 0.7163902595436165, "grad_norm": 1.1887011883528178, "learning_rate": 3.930138325799322e-06, "loss": 0.4115, "step": 20108 }, { "epoch": 0.7164258866701106, "grad_norm": 1.830832090045892, "learning_rate": 3.9292213184172925e-06, "loss": 0.5116, "step": 20109 }, { "epoch": 0.7164615137966047, "grad_norm": 1.5224619209472472, "learning_rate": 3.928304391871925e-06, "loss": 0.3939, "step": 20110 }, { "epoch": 0.7164971409230988, "grad_norm": 1.7449685645962953, "learning_rate": 3.927387546175428e-06, "loss": 0.4413, "step": 20111 }, { "epoch": 0.716532768049593, "grad_norm": 1.6234101173031443, "learning_rate": 3.9264707813400145e-06, "loss": 0.6115, "step": 20112 }, { "epoch": 0.7165683951760871, "grad_norm": 1.3437909810652764, "learning_rate": 3.925554097377884e-06, "loss": 0.4111, "step": 20113 }, { "epoch": 0.7166040223025811, "grad_norm": 1.6033658546418, "learning_rate": 3.92463749430125e-06, "loss": 0.5833, "step": 20114 }, { "epoch": 0.7166396494290753, "grad_norm": 1.268168654408032, "learning_rate": 3.923720972122311e-06, "loss": 0.4165, "step": 20115 }, { "epoch": 0.7166752765555694, "grad_norm": 1.7770482997964991, "learning_rate": 3.922804530853274e-06, "loss": 0.4569, "step": 20116 }, { "epoch": 0.7167109036820635, "grad_norm": 1.5861580341674815, "learning_rate": 3.921888170506342e-06, "loss": 0.4693, "step": 20117 }, { "epoch": 0.7167465308085577, "grad_norm": 1.3696084171086675, "learning_rate": 3.9209718910937174e-06, "loss": 0.483, "step": 20118 }, { "epoch": 0.7167821579350517, "grad_norm": 1.2088169223722558, "learning_rate": 3.9200556926276046e-06, "loss": 0.378, "step": 20119 }, { "epoch": 0.7168177850615458, "grad_norm": 1.481071447989786, "learning_rate": 3.919139575120197e-06, "loss": 0.4876, "step": 20120 }, { "epoch": 0.71685341218804, "grad_norm": 1.4198517106094697, "learning_rate": 3.918223538583699e-06, "loss": 0.4834, "step": 20121 }, { "epoch": 0.7168890393145341, "grad_norm": 1.5814631132422712, "learning_rate": 3.917307583030303e-06, "loss": 0.462, "step": 20122 }, { "epoch": 0.7169246664410281, "grad_norm": 2.042982327669102, "learning_rate": 3.91639170847221e-06, "loss": 0.5034, "step": 20123 }, { "epoch": 0.7169602935675223, "grad_norm": 3.2581553995877255, "learning_rate": 3.915475914921608e-06, "loss": 0.3943, "step": 20124 }, { "epoch": 0.7169959206940164, "grad_norm": 1.2323784152680413, "learning_rate": 3.914560202390703e-06, "loss": 0.4107, "step": 20125 }, { "epoch": 0.7170315478205105, "grad_norm": 2.9369271331280076, "learning_rate": 3.913644570891681e-06, "loss": 0.5206, "step": 20126 }, { "epoch": 0.7170671749470047, "grad_norm": 1.1567674170024334, "learning_rate": 3.912729020436737e-06, "loss": 0.3788, "step": 20127 }, { "epoch": 0.7171028020734987, "grad_norm": 1.674208458188481, "learning_rate": 3.911813551038057e-06, "loss": 0.6023, "step": 20128 }, { "epoch": 0.7171384291999928, "grad_norm": 1.5987256813953625, "learning_rate": 3.910898162707836e-06, "loss": 0.476, "step": 20129 }, { "epoch": 0.717174056326487, "grad_norm": 1.2468533426529742, "learning_rate": 3.909982855458265e-06, "loss": 0.4133, "step": 20130 }, { "epoch": 0.7172096834529811, "grad_norm": 1.4934942669559355, "learning_rate": 3.9090676293015265e-06, "loss": 0.5923, "step": 20131 }, { "epoch": 0.7172453105794752, "grad_norm": 1.424126855407617, "learning_rate": 3.908152484249809e-06, "loss": 0.4001, "step": 20132 }, { "epoch": 0.7172809377059693, "grad_norm": 1.6580361999226443, "learning_rate": 3.907237420315298e-06, "loss": 0.5488, "step": 20133 }, { "epoch": 0.7173165648324634, "grad_norm": 1.3814292767272678, "learning_rate": 3.906322437510185e-06, "loss": 0.4055, "step": 20134 }, { "epoch": 0.7173521919589575, "grad_norm": 1.3727090146734493, "learning_rate": 3.905407535846644e-06, "loss": 0.4563, "step": 20135 }, { "epoch": 0.7173878190854517, "grad_norm": 1.4764663679336472, "learning_rate": 3.904492715336864e-06, "loss": 0.4906, "step": 20136 }, { "epoch": 0.7174234462119458, "grad_norm": 1.4061248633987038, "learning_rate": 3.903577975993021e-06, "loss": 0.3929, "step": 20137 }, { "epoch": 0.7174590733384398, "grad_norm": 1.443546288752865, "learning_rate": 3.9026633178272985e-06, "loss": 0.4517, "step": 20138 }, { "epoch": 0.717494700464934, "grad_norm": 1.610887131474124, "learning_rate": 3.901748740851876e-06, "loss": 0.4024, "step": 20139 }, { "epoch": 0.7175303275914281, "grad_norm": 1.2637198719325016, "learning_rate": 3.90083424507893e-06, "loss": 0.4071, "step": 20140 }, { "epoch": 0.7175659547179222, "grad_norm": 1.229341856759852, "learning_rate": 3.899919830520644e-06, "loss": 0.4088, "step": 20141 }, { "epoch": 0.7176015818444164, "grad_norm": 1.5923292917460692, "learning_rate": 3.899005497189186e-06, "loss": 0.4946, "step": 20142 }, { "epoch": 0.7176372089709104, "grad_norm": 2.4036085280951247, "learning_rate": 3.8980912450967366e-06, "loss": 0.491, "step": 20143 }, { "epoch": 0.7176728360974045, "grad_norm": 1.5246990607239934, "learning_rate": 3.897177074255464e-06, "loss": 0.3976, "step": 20144 }, { "epoch": 0.7177084632238987, "grad_norm": 1.4764195083042406, "learning_rate": 3.896262984677548e-06, "loss": 0.3865, "step": 20145 }, { "epoch": 0.7177440903503928, "grad_norm": 1.7031335681886133, "learning_rate": 3.895348976375153e-06, "loss": 0.5116, "step": 20146 }, { "epoch": 0.7177797174768868, "grad_norm": 1.6871342426749114, "learning_rate": 3.894435049360453e-06, "loss": 0.474, "step": 20147 }, { "epoch": 0.717815344603381, "grad_norm": 1.5809880176170277, "learning_rate": 3.893521203645618e-06, "loss": 0.541, "step": 20148 }, { "epoch": 0.7178509717298751, "grad_norm": 1.3570571921247019, "learning_rate": 3.89260743924282e-06, "loss": 0.4492, "step": 20149 }, { "epoch": 0.7178865988563692, "grad_norm": 2.2044609958567603, "learning_rate": 3.8916937561642195e-06, "loss": 0.5447, "step": 20150 }, { "epoch": 0.7179222259828634, "grad_norm": 2.1788236992410477, "learning_rate": 3.890780154421986e-06, "loss": 0.4414, "step": 20151 }, { "epoch": 0.7179578531093574, "grad_norm": 1.7306523196801782, "learning_rate": 3.889866634028287e-06, "loss": 0.5713, "step": 20152 }, { "epoch": 0.7179934802358515, "grad_norm": 1.3665315720958457, "learning_rate": 3.888953194995283e-06, "loss": 0.4245, "step": 20153 }, { "epoch": 0.7180291073623457, "grad_norm": 1.3868022303203154, "learning_rate": 3.888039837335138e-06, "loss": 0.4706, "step": 20154 }, { "epoch": 0.7180647344888398, "grad_norm": 1.7017851015518415, "learning_rate": 3.887126561060014e-06, "loss": 0.5011, "step": 20155 }, { "epoch": 0.7181003616153339, "grad_norm": 1.6495382173336333, "learning_rate": 3.886213366182077e-06, "loss": 0.5646, "step": 20156 }, { "epoch": 0.718135988741828, "grad_norm": 1.4198267554339374, "learning_rate": 3.885300252713477e-06, "loss": 0.3838, "step": 20157 }, { "epoch": 0.7181716158683221, "grad_norm": 1.4991999510348408, "learning_rate": 3.884387220666383e-06, "loss": 0.5801, "step": 20158 }, { "epoch": 0.7182072429948163, "grad_norm": 1.4910347874706609, "learning_rate": 3.883474270052944e-06, "loss": 0.4348, "step": 20159 }, { "epoch": 0.7182428701213104, "grad_norm": 1.6830374744393786, "learning_rate": 3.882561400885324e-06, "loss": 0.5126, "step": 20160 }, { "epoch": 0.7182784972478045, "grad_norm": 1.4092496370810499, "learning_rate": 3.881648613175671e-06, "loss": 0.4598, "step": 20161 }, { "epoch": 0.7183141243742986, "grad_norm": 5.7020046706937295, "learning_rate": 3.880735906936143e-06, "loss": 0.4268, "step": 20162 }, { "epoch": 0.7183497515007927, "grad_norm": 1.4167367101157615, "learning_rate": 3.879823282178894e-06, "loss": 0.4565, "step": 20163 }, { "epoch": 0.7183853786272868, "grad_norm": 2.191968969334575, "learning_rate": 3.8789107389160755e-06, "loss": 0.4676, "step": 20164 }, { "epoch": 0.718421005753781, "grad_norm": 1.428264954490679, "learning_rate": 3.8779982771598414e-06, "loss": 0.3933, "step": 20165 }, { "epoch": 0.7184566328802751, "grad_norm": 1.6455505201149374, "learning_rate": 3.877085896922337e-06, "loss": 0.5181, "step": 20166 }, { "epoch": 0.7184922600067691, "grad_norm": 1.1694775595664386, "learning_rate": 3.876173598215717e-06, "loss": 0.4259, "step": 20167 }, { "epoch": 0.7185278871332633, "grad_norm": 1.7725379895355564, "learning_rate": 3.875261381052121e-06, "loss": 0.5618, "step": 20168 }, { "epoch": 0.7185635142597574, "grad_norm": 1.4740655241449612, "learning_rate": 3.874349245443702e-06, "loss": 0.5162, "step": 20169 }, { "epoch": 0.7185991413862515, "grad_norm": 1.6596982038732992, "learning_rate": 3.873437191402605e-06, "loss": 0.4855, "step": 20170 }, { "epoch": 0.7186347685127457, "grad_norm": 1.562286584224884, "learning_rate": 3.872525218940976e-06, "loss": 0.552, "step": 20171 }, { "epoch": 0.7186703956392397, "grad_norm": 1.1670445964172798, "learning_rate": 3.871613328070953e-06, "loss": 0.3849, "step": 20172 }, { "epoch": 0.7187060227657338, "grad_norm": 1.322533775284653, "learning_rate": 3.870701518804687e-06, "loss": 0.4165, "step": 20173 }, { "epoch": 0.718741649892228, "grad_norm": 1.6275288245255068, "learning_rate": 3.869789791154309e-06, "loss": 0.484, "step": 20174 }, { "epoch": 0.7187772770187221, "grad_norm": 1.3765521108722596, "learning_rate": 3.868878145131965e-06, "loss": 0.4257, "step": 20175 }, { "epoch": 0.7188129041452161, "grad_norm": 1.2523540264894784, "learning_rate": 3.867966580749796e-06, "loss": 0.4312, "step": 20176 }, { "epoch": 0.7188485312717103, "grad_norm": 1.5684110417075863, "learning_rate": 3.867055098019936e-06, "loss": 0.5504, "step": 20177 }, { "epoch": 0.7188841583982044, "grad_norm": 1.2895603290197517, "learning_rate": 3.8661436969545285e-06, "loss": 0.4625, "step": 20178 }, { "epoch": 0.7189197855246985, "grad_norm": 1.311829814840079, "learning_rate": 3.865232377565701e-06, "loss": 0.572, "step": 20179 }, { "epoch": 0.7189554126511927, "grad_norm": 2.5977728315582045, "learning_rate": 3.864321139865597e-06, "loss": 0.5096, "step": 20180 }, { "epoch": 0.7189910397776867, "grad_norm": 2.055241907980544, "learning_rate": 3.863409983866341e-06, "loss": 0.4839, "step": 20181 }, { "epoch": 0.7190266669041808, "grad_norm": 1.3562757598544524, "learning_rate": 3.862498909580074e-06, "loss": 0.4189, "step": 20182 }, { "epoch": 0.719062294030675, "grad_norm": 1.7132890417316884, "learning_rate": 3.86158791701892e-06, "loss": 0.4638, "step": 20183 }, { "epoch": 0.7190979211571691, "grad_norm": 1.6009028597662758, "learning_rate": 3.860677006195012e-06, "loss": 0.4862, "step": 20184 }, { "epoch": 0.7191335482836632, "grad_norm": 1.267795570815699, "learning_rate": 3.859766177120483e-06, "loss": 0.4256, "step": 20185 }, { "epoch": 0.7191691754101573, "grad_norm": 1.3247383333893044, "learning_rate": 3.858855429807459e-06, "loss": 0.4564, "step": 20186 }, { "epoch": 0.7192048025366514, "grad_norm": 1.2721281421330346, "learning_rate": 3.85794476426807e-06, "loss": 0.4599, "step": 20187 }, { "epoch": 0.7192404296631455, "grad_norm": 1.7537718131252193, "learning_rate": 3.857034180514437e-06, "loss": 0.5283, "step": 20188 }, { "epoch": 0.7192760567896397, "grad_norm": 1.868560756969666, "learning_rate": 3.856123678558691e-06, "loss": 0.5395, "step": 20189 }, { "epoch": 0.7193116839161338, "grad_norm": 1.1439436680391677, "learning_rate": 3.855213258412948e-06, "loss": 0.3838, "step": 20190 }, { "epoch": 0.7193473110426278, "grad_norm": 1.1935024656658373, "learning_rate": 3.854302920089336e-06, "loss": 0.4185, "step": 20191 }, { "epoch": 0.719382938169122, "grad_norm": 1.5080601702573955, "learning_rate": 3.853392663599977e-06, "loss": 0.5213, "step": 20192 }, { "epoch": 0.7194185652956161, "grad_norm": 1.2635347292939523, "learning_rate": 3.852482488956992e-06, "loss": 0.4527, "step": 20193 }, { "epoch": 0.7194541924221102, "grad_norm": 1.238764747072534, "learning_rate": 3.851572396172498e-06, "loss": 0.4314, "step": 20194 }, { "epoch": 0.7194898195486044, "grad_norm": 2.252018136043231, "learning_rate": 3.8506623852586174e-06, "loss": 0.4722, "step": 20195 }, { "epoch": 0.7195254466750984, "grad_norm": 1.3105582766071513, "learning_rate": 3.849752456227463e-06, "loss": 0.4388, "step": 20196 }, { "epoch": 0.7195610738015925, "grad_norm": 1.352509072529052, "learning_rate": 3.848842609091157e-06, "loss": 0.3991, "step": 20197 }, { "epoch": 0.7195967009280867, "grad_norm": 1.1924251224711473, "learning_rate": 3.847932843861807e-06, "loss": 0.3734, "step": 20198 }, { "epoch": 0.7196323280545808, "grad_norm": 1.341937765944349, "learning_rate": 3.847023160551528e-06, "loss": 0.4788, "step": 20199 }, { "epoch": 0.7196679551810748, "grad_norm": 1.1007585618548674, "learning_rate": 3.8461135591724416e-06, "loss": 0.4871, "step": 20200 }, { "epoch": 0.719703582307569, "grad_norm": 1.467980739932212, "learning_rate": 3.845204039736653e-06, "loss": 0.5144, "step": 20201 }, { "epoch": 0.7197392094340631, "grad_norm": 1.710229724875699, "learning_rate": 3.844294602256277e-06, "loss": 0.596, "step": 20202 }, { "epoch": 0.7197748365605572, "grad_norm": 1.7954982906871495, "learning_rate": 3.8433852467434175e-06, "loss": 0.4405, "step": 20203 }, { "epoch": 0.7198104636870514, "grad_norm": 2.7895291524601014, "learning_rate": 3.842475973210191e-06, "loss": 0.4892, "step": 20204 }, { "epoch": 0.7198460908135454, "grad_norm": 1.4407499874968144, "learning_rate": 3.841566781668696e-06, "loss": 0.5284, "step": 20205 }, { "epoch": 0.7198817179400395, "grad_norm": 1.503715850553087, "learning_rate": 3.8406576721310454e-06, "loss": 0.5105, "step": 20206 }, { "epoch": 0.7199173450665337, "grad_norm": 1.2128809438701253, "learning_rate": 3.839748644609342e-06, "loss": 0.4163, "step": 20207 }, { "epoch": 0.7199529721930278, "grad_norm": 1.3780950162332966, "learning_rate": 3.838839699115695e-06, "loss": 0.4363, "step": 20208 }, { "epoch": 0.7199885993195219, "grad_norm": 1.209831234686793, "learning_rate": 3.837930835662199e-06, "loss": 0.4394, "step": 20209 }, { "epoch": 0.720024226446016, "grad_norm": 1.9203533253773117, "learning_rate": 3.837022054260963e-06, "loss": 0.5519, "step": 20210 }, { "epoch": 0.7200598535725101, "grad_norm": 1.6317289322432842, "learning_rate": 3.836113354924088e-06, "loss": 0.5204, "step": 20211 }, { "epoch": 0.7200954806990042, "grad_norm": 1.3201137229198248, "learning_rate": 3.835204737663669e-06, "loss": 0.4952, "step": 20212 }, { "epoch": 0.7201311078254984, "grad_norm": 1.420086952901944, "learning_rate": 3.834296202491809e-06, "loss": 0.4349, "step": 20213 }, { "epoch": 0.7201667349519925, "grad_norm": 1.3525386983798875, "learning_rate": 3.833387749420603e-06, "loss": 0.4453, "step": 20214 }, { "epoch": 0.7202023620784865, "grad_norm": 1.258089987325049, "learning_rate": 3.832479378462153e-06, "loss": 0.4295, "step": 20215 }, { "epoch": 0.7202379892049807, "grad_norm": 1.3273866683086812, "learning_rate": 3.831571089628547e-06, "loss": 0.5386, "step": 20216 }, { "epoch": 0.7202736163314748, "grad_norm": 1.4916043280559421, "learning_rate": 3.830662882931888e-06, "loss": 0.5491, "step": 20217 }, { "epoch": 0.7203092434579689, "grad_norm": 1.5450491583711226, "learning_rate": 3.829754758384262e-06, "loss": 0.4908, "step": 20218 }, { "epoch": 0.7203448705844631, "grad_norm": 1.0515016525592054, "learning_rate": 3.828846715997766e-06, "loss": 0.4179, "step": 20219 }, { "epoch": 0.7203804977109571, "grad_norm": 1.2941523472262921, "learning_rate": 3.8279387557844875e-06, "loss": 0.5927, "step": 20220 }, { "epoch": 0.7204161248374512, "grad_norm": 1.605358353335159, "learning_rate": 3.8270308777565165e-06, "loss": 0.4812, "step": 20221 }, { "epoch": 0.7204517519639454, "grad_norm": 1.4097453579513304, "learning_rate": 3.826123081925946e-06, "loss": 0.4059, "step": 20222 }, { "epoch": 0.7204873790904395, "grad_norm": 1.0756223082354173, "learning_rate": 3.82521536830486e-06, "loss": 0.4172, "step": 20223 }, { "epoch": 0.7205230062169335, "grad_norm": 1.6325656525190742, "learning_rate": 3.824307736905351e-06, "loss": 0.5462, "step": 20224 }, { "epoch": 0.7205586333434277, "grad_norm": 1.4752160823366445, "learning_rate": 3.823400187739499e-06, "loss": 0.4579, "step": 20225 }, { "epoch": 0.7205942604699218, "grad_norm": 1.3668845225888475, "learning_rate": 3.822492720819393e-06, "loss": 0.4886, "step": 20226 }, { "epoch": 0.7206298875964159, "grad_norm": 1.245646677215857, "learning_rate": 3.821585336157112e-06, "loss": 0.4275, "step": 20227 }, { "epoch": 0.7206655147229101, "grad_norm": 1.928746590269587, "learning_rate": 3.820678033764739e-06, "loss": 0.4455, "step": 20228 }, { "epoch": 0.7207011418494041, "grad_norm": 1.2989868247678942, "learning_rate": 3.8197708136543585e-06, "loss": 0.4503, "step": 20229 }, { "epoch": 0.7207367689758982, "grad_norm": 1.634924378265178, "learning_rate": 3.818863675838052e-06, "loss": 0.5832, "step": 20230 }, { "epoch": 0.7207723961023924, "grad_norm": 1.589904246112174, "learning_rate": 3.817956620327893e-06, "loss": 0.5264, "step": 20231 }, { "epoch": 0.7208080232288865, "grad_norm": 1.2113906887066928, "learning_rate": 3.817049647135962e-06, "loss": 0.438, "step": 20232 }, { "epoch": 0.7208436503553806, "grad_norm": 1.227909140600386, "learning_rate": 3.8161427562743415e-06, "loss": 0.4471, "step": 20233 }, { "epoch": 0.7208792774818747, "grad_norm": 1.5455048955859765, "learning_rate": 3.815235947755097e-06, "loss": 0.5729, "step": 20234 }, { "epoch": 0.7209149046083688, "grad_norm": 1.4054352697170946, "learning_rate": 3.814329221590314e-06, "loss": 0.4564, "step": 20235 }, { "epoch": 0.7209505317348629, "grad_norm": 1.2764605022784226, "learning_rate": 3.8134225777920565e-06, "loss": 0.4553, "step": 20236 }, { "epoch": 0.7209861588613571, "grad_norm": 1.5877913073719188, "learning_rate": 3.812516016372402e-06, "loss": 0.4843, "step": 20237 }, { "epoch": 0.7210217859878512, "grad_norm": 1.2218429586521333, "learning_rate": 3.8116095373434204e-06, "loss": 0.3875, "step": 20238 }, { "epoch": 0.7210574131143452, "grad_norm": 1.6528219794110528, "learning_rate": 3.810703140717188e-06, "loss": 0.5571, "step": 20239 }, { "epoch": 0.7210930402408394, "grad_norm": 1.226674933025671, "learning_rate": 3.809796826505765e-06, "loss": 0.4217, "step": 20240 }, { "epoch": 0.7211286673673335, "grad_norm": 1.1833450653230921, "learning_rate": 3.8088905947212274e-06, "loss": 0.4759, "step": 20241 }, { "epoch": 0.7211642944938276, "grad_norm": 1.607229098505311, "learning_rate": 3.8079844453756367e-06, "loss": 0.5192, "step": 20242 }, { "epoch": 0.7211999216203218, "grad_norm": 1.1998761054732408, "learning_rate": 3.8070783784810593e-06, "loss": 0.4132, "step": 20243 }, { "epoch": 0.7212355487468158, "grad_norm": 1.1919660307316775, "learning_rate": 3.8061723940495634e-06, "loss": 0.4035, "step": 20244 }, { "epoch": 0.7212711758733099, "grad_norm": 1.310583022295312, "learning_rate": 3.8052664920932103e-06, "loss": 0.4143, "step": 20245 }, { "epoch": 0.7213068029998041, "grad_norm": 1.8384205957943789, "learning_rate": 3.8043606726240676e-06, "loss": 0.4778, "step": 20246 }, { "epoch": 0.7213424301262982, "grad_norm": 1.585048095155369, "learning_rate": 3.8034549356541896e-06, "loss": 0.5276, "step": 20247 }, { "epoch": 0.7213780572527922, "grad_norm": 1.6824402173937738, "learning_rate": 3.8025492811956434e-06, "loss": 0.5078, "step": 20248 }, { "epoch": 0.7214136843792864, "grad_norm": 1.2585679256036777, "learning_rate": 3.801643709260482e-06, "loss": 0.4356, "step": 20249 }, { "epoch": 0.7214493115057805, "grad_norm": 1.530266373943639, "learning_rate": 3.8007382198607703e-06, "loss": 0.5589, "step": 20250 }, { "epoch": 0.7214849386322746, "grad_norm": 1.1730989099243025, "learning_rate": 3.799832813008556e-06, "loss": 0.428, "step": 20251 }, { "epoch": 0.7215205657587688, "grad_norm": 1.133881903188226, "learning_rate": 3.798927488715908e-06, "loss": 0.4471, "step": 20252 }, { "epoch": 0.7215561928852628, "grad_norm": 1.4366847583891547, "learning_rate": 3.79802224699487e-06, "loss": 0.5843, "step": 20253 }, { "epoch": 0.7215918200117569, "grad_norm": 1.5533135226527934, "learning_rate": 3.797117087857506e-06, "loss": 0.5551, "step": 20254 }, { "epoch": 0.7216274471382511, "grad_norm": 1.397002580703342, "learning_rate": 3.7962120113158597e-06, "loss": 0.391, "step": 20255 }, { "epoch": 0.7216630742647452, "grad_norm": 1.7773035268719595, "learning_rate": 3.7953070173819863e-06, "loss": 0.5238, "step": 20256 }, { "epoch": 0.7216987013912393, "grad_norm": 1.0878611003113714, "learning_rate": 3.7944021060679404e-06, "loss": 0.4123, "step": 20257 }, { "epoch": 0.7217343285177334, "grad_norm": 1.200930655885017, "learning_rate": 3.7934972773857637e-06, "loss": 0.4527, "step": 20258 }, { "epoch": 0.7217699556442275, "grad_norm": 1.7348026679577881, "learning_rate": 3.7925925313475097e-06, "loss": 0.512, "step": 20259 }, { "epoch": 0.7218055827707216, "grad_norm": 2.0764072245205676, "learning_rate": 3.7916878679652247e-06, "loss": 0.4978, "step": 20260 }, { "epoch": 0.7218412098972158, "grad_norm": 1.6759595536781526, "learning_rate": 3.7907832872509587e-06, "loss": 0.5661, "step": 20261 }, { "epoch": 0.7218768370237099, "grad_norm": 2.072750247718664, "learning_rate": 3.7898787892167497e-06, "loss": 0.494, "step": 20262 }, { "epoch": 0.7219124641502039, "grad_norm": 1.1668097836470233, "learning_rate": 3.788974373874649e-06, "loss": 0.4231, "step": 20263 }, { "epoch": 0.7219480912766981, "grad_norm": 1.817728466659281, "learning_rate": 3.788070041236691e-06, "loss": 0.6395, "step": 20264 }, { "epoch": 0.7219837184031922, "grad_norm": 1.1888146735380771, "learning_rate": 3.787165791314923e-06, "loss": 0.4356, "step": 20265 }, { "epoch": 0.7220193455296863, "grad_norm": 1.2659886830654261, "learning_rate": 3.7862616241213855e-06, "loss": 0.4592, "step": 20266 }, { "epoch": 0.7220549726561805, "grad_norm": 1.193259134777282, "learning_rate": 3.7853575396681176e-06, "loss": 0.4031, "step": 20267 }, { "epoch": 0.7220905997826745, "grad_norm": 2.0816554125278612, "learning_rate": 3.784453537967161e-06, "loss": 0.4506, "step": 20268 }, { "epoch": 0.7221262269091686, "grad_norm": 1.6368666462124972, "learning_rate": 3.7835496190305465e-06, "loss": 0.5007, "step": 20269 }, { "epoch": 0.7221618540356628, "grad_norm": 1.2342789665494511, "learning_rate": 3.782645782870317e-06, "loss": 0.4342, "step": 20270 }, { "epoch": 0.7221974811621569, "grad_norm": 1.442633742671446, "learning_rate": 3.781742029498501e-06, "loss": 0.5024, "step": 20271 }, { "epoch": 0.722233108288651, "grad_norm": 1.9171934195905973, "learning_rate": 3.78083835892714e-06, "loss": 0.5384, "step": 20272 }, { "epoch": 0.7222687354151451, "grad_norm": 1.8103966736758832, "learning_rate": 3.7799347711682597e-06, "loss": 0.5388, "step": 20273 }, { "epoch": 0.7223043625416392, "grad_norm": 1.5853484007782508, "learning_rate": 3.7790312662338957e-06, "loss": 0.5428, "step": 20274 }, { "epoch": 0.7223399896681333, "grad_norm": 1.613975813664056, "learning_rate": 3.778127844136078e-06, "loss": 0.5125, "step": 20275 }, { "epoch": 0.7223756167946275, "grad_norm": 1.3993138688974334, "learning_rate": 3.7772245048868418e-06, "loss": 0.3957, "step": 20276 }, { "epoch": 0.7224112439211215, "grad_norm": 1.0522241882103593, "learning_rate": 3.776321248498206e-06, "loss": 0.4086, "step": 20277 }, { "epoch": 0.7224468710476156, "grad_norm": 1.746743073286117, "learning_rate": 3.7754180749822033e-06, "loss": 0.5159, "step": 20278 }, { "epoch": 0.7224824981741098, "grad_norm": 2.048242444872909, "learning_rate": 3.774514984350862e-06, "loss": 0.4992, "step": 20279 }, { "epoch": 0.7225181253006039, "grad_norm": 1.3042772861567056, "learning_rate": 3.773611976616203e-06, "loss": 0.5396, "step": 20280 }, { "epoch": 0.722553752427098, "grad_norm": 1.6633672839624856, "learning_rate": 3.772709051790252e-06, "loss": 0.6018, "step": 20281 }, { "epoch": 0.7225893795535921, "grad_norm": 1.2065029897482284, "learning_rate": 3.771806209885033e-06, "loss": 0.4752, "step": 20282 }, { "epoch": 0.7226250066800862, "grad_norm": 1.2375768738365014, "learning_rate": 3.7709034509125706e-06, "loss": 0.4106, "step": 20283 }, { "epoch": 0.7226606338065803, "grad_norm": 1.289432154475369, "learning_rate": 3.7700007748848788e-06, "loss": 0.4224, "step": 20284 }, { "epoch": 0.7226962609330745, "grad_norm": 1.866897694542447, "learning_rate": 3.769098181813986e-06, "loss": 0.5146, "step": 20285 }, { "epoch": 0.7227318880595686, "grad_norm": 1.7580890639166487, "learning_rate": 3.7681956717119005e-06, "loss": 0.5042, "step": 20286 }, { "epoch": 0.7227675151860626, "grad_norm": 1.137940089605225, "learning_rate": 3.767293244590651e-06, "loss": 0.4122, "step": 20287 }, { "epoch": 0.7228031423125568, "grad_norm": 1.4954720090311997, "learning_rate": 3.76639090046224e-06, "loss": 0.5052, "step": 20288 }, { "epoch": 0.7228387694390509, "grad_norm": 1.199988371988818, "learning_rate": 3.765488639338699e-06, "loss": 0.4406, "step": 20289 }, { "epoch": 0.722874396565545, "grad_norm": 1.3556499790347711, "learning_rate": 3.7645864612320305e-06, "loss": 0.361, "step": 20290 }, { "epoch": 0.7229100236920392, "grad_norm": 1.2705701527004925, "learning_rate": 3.763684366154251e-06, "loss": 0.3744, "step": 20291 }, { "epoch": 0.7229456508185332, "grad_norm": 1.2266199813389256, "learning_rate": 3.762782354117378e-06, "loss": 0.4351, "step": 20292 }, { "epoch": 0.7229812779450273, "grad_norm": 1.9762853673839584, "learning_rate": 3.7618804251334127e-06, "loss": 0.5846, "step": 20293 }, { "epoch": 0.7230169050715215, "grad_norm": 1.6277388315234866, "learning_rate": 3.760978579214374e-06, "loss": 0.5269, "step": 20294 }, { "epoch": 0.7230525321980156, "grad_norm": 1.2611317657635939, "learning_rate": 3.7600768163722625e-06, "loss": 0.4129, "step": 20295 }, { "epoch": 0.7230881593245096, "grad_norm": 1.949254014728667, "learning_rate": 3.75917513661909e-06, "loss": 0.5541, "step": 20296 }, { "epoch": 0.7231237864510038, "grad_norm": 1.4592002381020994, "learning_rate": 3.758273539966862e-06, "loss": 0.5917, "step": 20297 }, { "epoch": 0.7231594135774979, "grad_norm": 1.2836114475941351, "learning_rate": 3.7573720264275883e-06, "loss": 0.4079, "step": 20298 }, { "epoch": 0.723195040703992, "grad_norm": 1.4799261569576274, "learning_rate": 3.7564705960132662e-06, "loss": 0.4618, "step": 20299 }, { "epoch": 0.7232306678304862, "grad_norm": 1.7869166354803458, "learning_rate": 3.755569248735905e-06, "loss": 0.5012, "step": 20300 }, { "epoch": 0.7232662949569802, "grad_norm": 1.341586163270038, "learning_rate": 3.7546679846075006e-06, "loss": 0.5359, "step": 20301 }, { "epoch": 0.7233019220834743, "grad_norm": 1.408043328287254, "learning_rate": 3.753766803640058e-06, "loss": 0.526, "step": 20302 }, { "epoch": 0.7233375492099685, "grad_norm": 1.2297308536235116, "learning_rate": 3.7528657058455753e-06, "loss": 0.406, "step": 20303 }, { "epoch": 0.7233731763364626, "grad_norm": 1.2937276731471687, "learning_rate": 3.7519646912360517e-06, "loss": 0.4405, "step": 20304 }, { "epoch": 0.7234088034629567, "grad_norm": 1.2080398807143702, "learning_rate": 3.75106375982349e-06, "loss": 0.4122, "step": 20305 }, { "epoch": 0.7234444305894508, "grad_norm": 1.7624902097362027, "learning_rate": 3.7501629116198777e-06, "loss": 0.5542, "step": 20306 }, { "epoch": 0.7234800577159449, "grad_norm": 1.4227777166737978, "learning_rate": 3.749262146637219e-06, "loss": 0.4937, "step": 20307 }, { "epoch": 0.723515684842439, "grad_norm": 1.7493258403604766, "learning_rate": 3.7483614648875e-06, "loss": 0.4388, "step": 20308 }, { "epoch": 0.7235513119689332, "grad_norm": 1.1406120827106574, "learning_rate": 3.7474608663827215e-06, "loss": 0.3731, "step": 20309 }, { "epoch": 0.7235869390954273, "grad_norm": 1.5812341396781882, "learning_rate": 3.746560351134868e-06, "loss": 0.5178, "step": 20310 }, { "epoch": 0.7236225662219213, "grad_norm": 1.183768081274163, "learning_rate": 3.745659919155934e-06, "loss": 0.4278, "step": 20311 }, { "epoch": 0.7236581933484155, "grad_norm": 1.1761418085219641, "learning_rate": 3.74475957045791e-06, "loss": 0.4406, "step": 20312 }, { "epoch": 0.7236938204749096, "grad_norm": 1.2898429342578153, "learning_rate": 3.743859305052785e-06, "loss": 0.4172, "step": 20313 }, { "epoch": 0.7237294476014037, "grad_norm": 1.521628711893601, "learning_rate": 3.7429591229525484e-06, "loss": 0.4968, "step": 20314 }, { "epoch": 0.7237650747278979, "grad_norm": 1.8863521925811657, "learning_rate": 3.7420590241691823e-06, "loss": 0.5615, "step": 20315 }, { "epoch": 0.7238007018543919, "grad_norm": 1.4726214752737468, "learning_rate": 3.7411590087146767e-06, "loss": 0.5163, "step": 20316 }, { "epoch": 0.723836328980886, "grad_norm": 1.2026027694761032, "learning_rate": 3.7402590766010103e-06, "loss": 0.412, "step": 20317 }, { "epoch": 0.7238719561073802, "grad_norm": 1.4985560657205186, "learning_rate": 3.7393592278401704e-06, "loss": 0.4639, "step": 20318 }, { "epoch": 0.7239075832338743, "grad_norm": 1.178479469126715, "learning_rate": 3.7384594624441383e-06, "loss": 0.4432, "step": 20319 }, { "epoch": 0.7239432103603683, "grad_norm": 1.4774973412782508, "learning_rate": 3.7375597804248974e-06, "loss": 0.5329, "step": 20320 }, { "epoch": 0.7239788374868625, "grad_norm": 1.1550918226688545, "learning_rate": 3.7366601817944225e-06, "loss": 0.3824, "step": 20321 }, { "epoch": 0.7240144646133566, "grad_norm": 1.4082495855190693, "learning_rate": 3.7357606665646984e-06, "loss": 0.5006, "step": 20322 }, { "epoch": 0.7240500917398507, "grad_norm": 1.2906101674993538, "learning_rate": 3.7348612347476965e-06, "loss": 0.4513, "step": 20323 }, { "epoch": 0.7240857188663449, "grad_norm": 1.4900355070789142, "learning_rate": 3.7339618863553983e-06, "loss": 0.4848, "step": 20324 }, { "epoch": 0.724121345992839, "grad_norm": 1.2356770329626785, "learning_rate": 3.7330626213997755e-06, "loss": 0.4216, "step": 20325 }, { "epoch": 0.724156973119333, "grad_norm": 1.2056007356698393, "learning_rate": 3.7321634398928e-06, "loss": 0.4102, "step": 20326 }, { "epoch": 0.7241926002458272, "grad_norm": 1.4454723040391637, "learning_rate": 3.731264341846457e-06, "loss": 0.5448, "step": 20327 }, { "epoch": 0.7242282273723213, "grad_norm": 1.2052423721555974, "learning_rate": 3.7303653272727057e-06, "loss": 0.408, "step": 20328 }, { "epoch": 0.7242638544988154, "grad_norm": 1.5337721264232052, "learning_rate": 3.7294663961835263e-06, "loss": 0.5052, "step": 20329 }, { "epoch": 0.7242994816253095, "grad_norm": 1.3098403557294034, "learning_rate": 3.7285675485908823e-06, "loss": 0.5066, "step": 20330 }, { "epoch": 0.7243351087518036, "grad_norm": 1.4600669036247809, "learning_rate": 3.7276687845067474e-06, "loss": 0.4686, "step": 20331 }, { "epoch": 0.7243707358782977, "grad_norm": 1.302882562942107, "learning_rate": 3.7267701039430836e-06, "loss": 0.4799, "step": 20332 }, { "epoch": 0.7244063630047919, "grad_norm": 1.085017359405072, "learning_rate": 3.72587150691186e-06, "loss": 0.3918, "step": 20333 }, { "epoch": 0.724441990131286, "grad_norm": 2.0918171848013327, "learning_rate": 3.7249729934250433e-06, "loss": 0.5978, "step": 20334 }, { "epoch": 0.72447761725778, "grad_norm": 1.1347036471818892, "learning_rate": 3.7240745634946005e-06, "loss": 0.4083, "step": 20335 }, { "epoch": 0.7245132443842742, "grad_norm": 1.1109155964428505, "learning_rate": 3.723176217132488e-06, "loss": 0.4266, "step": 20336 }, { "epoch": 0.7245488715107683, "grad_norm": 1.165444940642834, "learning_rate": 3.7222779543506714e-06, "loss": 0.439, "step": 20337 }, { "epoch": 0.7245844986372624, "grad_norm": 1.2150815858953838, "learning_rate": 3.721379775161116e-06, "loss": 0.4107, "step": 20338 }, { "epoch": 0.7246201257637566, "grad_norm": 1.1899518499169845, "learning_rate": 3.7204816795757727e-06, "loss": 0.4075, "step": 20339 }, { "epoch": 0.7246557528902506, "grad_norm": 1.9078014062770006, "learning_rate": 3.7195836676066057e-06, "loss": 0.5369, "step": 20340 }, { "epoch": 0.7246913800167447, "grad_norm": 1.5658614403946427, "learning_rate": 3.7186857392655708e-06, "loss": 0.459, "step": 20341 }, { "epoch": 0.7247270071432389, "grad_norm": 1.23457625370665, "learning_rate": 3.7177878945646306e-06, "loss": 0.3619, "step": 20342 }, { "epoch": 0.724762634269733, "grad_norm": 1.4172184883731649, "learning_rate": 3.7168901335157313e-06, "loss": 0.5062, "step": 20343 }, { "epoch": 0.724798261396227, "grad_norm": 1.2047525400200154, "learning_rate": 3.7159924561308346e-06, "loss": 0.4481, "step": 20344 }, { "epoch": 0.7248338885227212, "grad_norm": 1.1916258902249588, "learning_rate": 3.7150948624218876e-06, "loss": 0.3807, "step": 20345 }, { "epoch": 0.7248695156492153, "grad_norm": 1.2311209738215978, "learning_rate": 3.714197352400849e-06, "loss": 0.4166, "step": 20346 }, { "epoch": 0.7249051427757094, "grad_norm": 1.0211668328254013, "learning_rate": 3.713299926079663e-06, "loss": 0.4605, "step": 20347 }, { "epoch": 0.7249407699022036, "grad_norm": 1.1196235276020934, "learning_rate": 3.7124025834702816e-06, "loss": 0.3959, "step": 20348 }, { "epoch": 0.7249763970286976, "grad_norm": 1.6085925010377309, "learning_rate": 3.7115053245846554e-06, "loss": 0.5105, "step": 20349 }, { "epoch": 0.7250120241551917, "grad_norm": 1.4658825742367254, "learning_rate": 3.710608149434731e-06, "loss": 0.5292, "step": 20350 }, { "epoch": 0.7250476512816859, "grad_norm": 1.221163760261404, "learning_rate": 3.709711058032459e-06, "loss": 0.4072, "step": 20351 }, { "epoch": 0.72508327840818, "grad_norm": 1.5203986894550743, "learning_rate": 3.708814050389776e-06, "loss": 0.4984, "step": 20352 }, { "epoch": 0.7251189055346741, "grad_norm": 1.6809257170335874, "learning_rate": 3.7079171265186363e-06, "loss": 0.4691, "step": 20353 }, { "epoch": 0.7251545326611682, "grad_norm": 1.1421603342410216, "learning_rate": 3.7070202864309733e-06, "loss": 0.4811, "step": 20354 }, { "epoch": 0.7251901597876623, "grad_norm": 1.5563305910759673, "learning_rate": 3.7061235301387342e-06, "loss": 0.5418, "step": 20355 }, { "epoch": 0.7252257869141564, "grad_norm": 1.567235138458052, "learning_rate": 3.70522685765386e-06, "loss": 0.5066, "step": 20356 }, { "epoch": 0.7252614140406506, "grad_norm": 1.2099525277135932, "learning_rate": 3.704330268988293e-06, "loss": 0.4048, "step": 20357 }, { "epoch": 0.7252970411671447, "grad_norm": 1.2376743818829767, "learning_rate": 3.7034337641539652e-06, "loss": 0.4289, "step": 20358 }, { "epoch": 0.7253326682936387, "grad_norm": 1.1238520183280662, "learning_rate": 3.7025373431628185e-06, "loss": 0.3961, "step": 20359 }, { "epoch": 0.7253682954201329, "grad_norm": 1.4320603319161325, "learning_rate": 3.7016410060267914e-06, "loss": 0.5894, "step": 20360 }, { "epoch": 0.725403922546627, "grad_norm": 2.1524941615009143, "learning_rate": 3.7007447527578134e-06, "loss": 0.4821, "step": 20361 }, { "epoch": 0.7254395496731211, "grad_norm": 1.4586610476125874, "learning_rate": 3.6998485833678256e-06, "loss": 0.5399, "step": 20362 }, { "epoch": 0.7254751767996153, "grad_norm": 1.676922992254682, "learning_rate": 3.698952497868751e-06, "loss": 0.5835, "step": 20363 }, { "epoch": 0.7255108039261093, "grad_norm": 1.4083397082663258, "learning_rate": 3.698056496272534e-06, "loss": 0.5644, "step": 20364 }, { "epoch": 0.7255464310526034, "grad_norm": 1.1538796268384792, "learning_rate": 3.6971605785910967e-06, "loss": 0.4345, "step": 20365 }, { "epoch": 0.7255820581790976, "grad_norm": 1.2757315997020264, "learning_rate": 3.6962647448363753e-06, "loss": 0.4124, "step": 20366 }, { "epoch": 0.7256176853055917, "grad_norm": 1.1798471138173923, "learning_rate": 3.6953689950202908e-06, "loss": 0.4281, "step": 20367 }, { "epoch": 0.7256533124320858, "grad_norm": 1.2435446840087092, "learning_rate": 3.6944733291547784e-06, "loss": 0.3721, "step": 20368 }, { "epoch": 0.7256889395585799, "grad_norm": 1.6772032532898482, "learning_rate": 3.6935777472517576e-06, "loss": 0.4801, "step": 20369 }, { "epoch": 0.725724566685074, "grad_norm": 1.4383971452956392, "learning_rate": 3.692682249323157e-06, "loss": 0.5045, "step": 20370 }, { "epoch": 0.7257601938115681, "grad_norm": 1.8589759678023798, "learning_rate": 3.6917868353809007e-06, "loss": 0.4789, "step": 20371 }, { "epoch": 0.7257958209380623, "grad_norm": 1.7577727587051106, "learning_rate": 3.690891505436912e-06, "loss": 0.5065, "step": 20372 }, { "epoch": 0.7258314480645564, "grad_norm": 1.2934115719078085, "learning_rate": 3.689996259503116e-06, "loss": 0.4684, "step": 20373 }, { "epoch": 0.7258670751910504, "grad_norm": 1.7020896712379083, "learning_rate": 3.6891010975914264e-06, "loss": 0.5468, "step": 20374 }, { "epoch": 0.7259027023175446, "grad_norm": 1.4944934502397462, "learning_rate": 3.6882060197137714e-06, "loss": 0.5167, "step": 20375 }, { "epoch": 0.7259383294440387, "grad_norm": 1.6842557397620606, "learning_rate": 3.687311025882061e-06, "loss": 0.5496, "step": 20376 }, { "epoch": 0.7259739565705328, "grad_norm": 1.1867299167521295, "learning_rate": 3.6864161161082156e-06, "loss": 0.4061, "step": 20377 }, { "epoch": 0.726009583697027, "grad_norm": 1.2204842156011877, "learning_rate": 3.6855212904041538e-06, "loss": 0.4332, "step": 20378 }, { "epoch": 0.726045210823521, "grad_norm": 1.3156587419352612, "learning_rate": 3.6846265487817924e-06, "loss": 0.5347, "step": 20379 }, { "epoch": 0.7260808379500151, "grad_norm": 1.4176655169174808, "learning_rate": 3.6837318912530394e-06, "loss": 0.4887, "step": 20380 }, { "epoch": 0.7261164650765093, "grad_norm": 1.20195664258588, "learning_rate": 3.6828373178298147e-06, "loss": 0.4643, "step": 20381 }, { "epoch": 0.7261520922030034, "grad_norm": 1.051674757203501, "learning_rate": 3.6819428285240235e-06, "loss": 0.4528, "step": 20382 }, { "epoch": 0.7261877193294974, "grad_norm": 1.569564071448115, "learning_rate": 3.6810484233475787e-06, "loss": 0.4951, "step": 20383 }, { "epoch": 0.7262233464559916, "grad_norm": 1.1783469128102977, "learning_rate": 3.6801541023123943e-06, "loss": 0.3443, "step": 20384 }, { "epoch": 0.7262589735824857, "grad_norm": 1.62504347527625, "learning_rate": 3.6792598654303724e-06, "loss": 0.4541, "step": 20385 }, { "epoch": 0.7262946007089798, "grad_norm": 1.4598992041558305, "learning_rate": 3.678365712713423e-06, "loss": 0.5638, "step": 20386 }, { "epoch": 0.726330227835474, "grad_norm": 1.269830676059442, "learning_rate": 3.6774716441734524e-06, "loss": 0.3705, "step": 20387 }, { "epoch": 0.726365854961968, "grad_norm": 1.2564282857457378, "learning_rate": 3.67657765982237e-06, "loss": 0.4039, "step": 20388 }, { "epoch": 0.7264014820884621, "grad_norm": 1.723989087709749, "learning_rate": 3.675683759672073e-06, "loss": 0.504, "step": 20389 }, { "epoch": 0.7264371092149563, "grad_norm": 1.512718787459996, "learning_rate": 3.674789943734469e-06, "loss": 0.4998, "step": 20390 }, { "epoch": 0.7264727363414504, "grad_norm": 1.253894726891081, "learning_rate": 3.673896212021456e-06, "loss": 0.4427, "step": 20391 }, { "epoch": 0.7265083634679445, "grad_norm": 1.5492067075627898, "learning_rate": 3.6730025645449363e-06, "loss": 0.5368, "step": 20392 }, { "epoch": 0.7265439905944386, "grad_norm": 3.0199359598583335, "learning_rate": 3.6721090013168093e-06, "loss": 0.427, "step": 20393 }, { "epoch": 0.7265796177209327, "grad_norm": 1.6230187452890132, "learning_rate": 3.6712155223489745e-06, "loss": 0.5271, "step": 20394 }, { "epoch": 0.7266152448474268, "grad_norm": 1.699014939488491, "learning_rate": 3.6703221276533306e-06, "loss": 0.5549, "step": 20395 }, { "epoch": 0.726650871973921, "grad_norm": 1.1750549375136377, "learning_rate": 3.66942881724177e-06, "loss": 0.4031, "step": 20396 }, { "epoch": 0.726686499100415, "grad_norm": 1.8908381322812489, "learning_rate": 3.668535591126192e-06, "loss": 0.508, "step": 20397 }, { "epoch": 0.7267221262269091, "grad_norm": 1.3608198882606224, "learning_rate": 3.6676424493184848e-06, "loss": 0.4448, "step": 20398 }, { "epoch": 0.7267577533534033, "grad_norm": 1.4579878468344176, "learning_rate": 3.6667493918305485e-06, "loss": 0.5222, "step": 20399 }, { "epoch": 0.7267933804798974, "grad_norm": 1.6822622128507592, "learning_rate": 3.665856418674266e-06, "loss": 0.5371, "step": 20400 }, { "epoch": 0.7268290076063915, "grad_norm": 1.4920456868493186, "learning_rate": 3.664963529861534e-06, "loss": 0.3487, "step": 20401 }, { "epoch": 0.7268646347328857, "grad_norm": 2.015373737961021, "learning_rate": 3.6640707254042395e-06, "loss": 0.4844, "step": 20402 }, { "epoch": 0.7269002618593797, "grad_norm": 1.15443307108851, "learning_rate": 3.663178005314275e-06, "loss": 0.399, "step": 20403 }, { "epoch": 0.7269358889858738, "grad_norm": 1.0254167505135048, "learning_rate": 3.6622853696035222e-06, "loss": 0.381, "step": 20404 }, { "epoch": 0.726971516112368, "grad_norm": 1.3003025125812326, "learning_rate": 3.661392818283869e-06, "loss": 0.3981, "step": 20405 }, { "epoch": 0.7270071432388621, "grad_norm": 1.2382243095220542, "learning_rate": 3.6605003513672045e-06, "loss": 0.3714, "step": 20406 }, { "epoch": 0.7270427703653561, "grad_norm": 1.5297349777306908, "learning_rate": 3.6596079688654053e-06, "loss": 0.4337, "step": 20407 }, { "epoch": 0.7270783974918503, "grad_norm": 1.295782147710801, "learning_rate": 3.6587156707903572e-06, "loss": 0.3881, "step": 20408 }, { "epoch": 0.7271140246183444, "grad_norm": 1.1364645241150815, "learning_rate": 3.657823457153943e-06, "loss": 0.3799, "step": 20409 }, { "epoch": 0.7271496517448385, "grad_norm": 1.4621959765467847, "learning_rate": 3.656931327968045e-06, "loss": 0.5391, "step": 20410 }, { "epoch": 0.7271852788713327, "grad_norm": 1.6952890190971945, "learning_rate": 3.656039283244537e-06, "loss": 0.3727, "step": 20411 }, { "epoch": 0.7272209059978267, "grad_norm": 1.7752153776403514, "learning_rate": 3.655147322995304e-06, "loss": 0.5566, "step": 20412 }, { "epoch": 0.7272565331243208, "grad_norm": 1.621363279505361, "learning_rate": 3.6542554472322147e-06, "loss": 0.5196, "step": 20413 }, { "epoch": 0.727292160250815, "grad_norm": 1.379525223740956, "learning_rate": 3.6533636559671537e-06, "loss": 0.4172, "step": 20414 }, { "epoch": 0.7273277873773091, "grad_norm": 1.1539379666046377, "learning_rate": 3.6524719492119844e-06, "loss": 0.4131, "step": 20415 }, { "epoch": 0.7273634145038032, "grad_norm": 1.2325630330518964, "learning_rate": 3.6515803269785944e-06, "loss": 0.3861, "step": 20416 }, { "epoch": 0.7273990416302973, "grad_norm": 1.9891311696492417, "learning_rate": 3.6506887892788466e-06, "loss": 0.5293, "step": 20417 }, { "epoch": 0.7274346687567914, "grad_norm": 1.3769228885589668, "learning_rate": 3.6497973361246153e-06, "loss": 0.4331, "step": 20418 }, { "epoch": 0.7274702958832855, "grad_norm": 1.2020637338886213, "learning_rate": 3.648905967527775e-06, "loss": 0.3987, "step": 20419 }, { "epoch": 0.7275059230097797, "grad_norm": 1.1818651922623418, "learning_rate": 3.6480146835001874e-06, "loss": 0.401, "step": 20420 }, { "epoch": 0.7275415501362738, "grad_norm": 1.1770348905707615, "learning_rate": 3.647123484053727e-06, "loss": 0.3783, "step": 20421 }, { "epoch": 0.7275771772627678, "grad_norm": 1.2717747084356192, "learning_rate": 3.646232369200254e-06, "loss": 0.4012, "step": 20422 }, { "epoch": 0.727612804389262, "grad_norm": 1.2270987737291346, "learning_rate": 3.645341338951639e-06, "loss": 0.3902, "step": 20423 }, { "epoch": 0.7276484315157561, "grad_norm": 1.9065372972369783, "learning_rate": 3.644450393319745e-06, "loss": 0.4741, "step": 20424 }, { "epoch": 0.7276840586422502, "grad_norm": 1.570033963992886, "learning_rate": 3.6435595323164395e-06, "loss": 0.4578, "step": 20425 }, { "epoch": 0.7277196857687444, "grad_norm": 2.550021456196559, "learning_rate": 3.642668755953579e-06, "loss": 0.5697, "step": 20426 }, { "epoch": 0.7277553128952384, "grad_norm": 1.4642465034579806, "learning_rate": 3.641778064243031e-06, "loss": 0.3987, "step": 20427 }, { "epoch": 0.7277909400217325, "grad_norm": 1.6445598998993445, "learning_rate": 3.640887457196648e-06, "loss": 0.4373, "step": 20428 }, { "epoch": 0.7278265671482267, "grad_norm": 1.758213366934624, "learning_rate": 3.639996934826294e-06, "loss": 0.4973, "step": 20429 }, { "epoch": 0.7278621942747208, "grad_norm": 1.3283071328511404, "learning_rate": 3.639106497143826e-06, "loss": 0.4056, "step": 20430 }, { "epoch": 0.7278978214012148, "grad_norm": 1.3445062254131845, "learning_rate": 3.638216144161101e-06, "loss": 0.4467, "step": 20431 }, { "epoch": 0.727933448527709, "grad_norm": 1.199467711420151, "learning_rate": 3.6373258758899778e-06, "loss": 0.4166, "step": 20432 }, { "epoch": 0.7279690756542031, "grad_norm": 2.547367966109974, "learning_rate": 3.636435692342305e-06, "loss": 0.4574, "step": 20433 }, { "epoch": 0.7280047027806972, "grad_norm": 1.475857678074658, "learning_rate": 3.635545593529941e-06, "loss": 0.4896, "step": 20434 }, { "epoch": 0.7280403299071914, "grad_norm": 1.2325063742150233, "learning_rate": 3.634655579464733e-06, "loss": 0.4121, "step": 20435 }, { "epoch": 0.7280759570336854, "grad_norm": 1.8204684338775876, "learning_rate": 3.6337656501585395e-06, "loss": 0.5623, "step": 20436 }, { "epoch": 0.7281115841601795, "grad_norm": 1.42042259827995, "learning_rate": 3.6328758056232017e-06, "loss": 0.4374, "step": 20437 }, { "epoch": 0.7281472112866737, "grad_norm": 1.6013727856550561, "learning_rate": 3.6319860458705735e-06, "loss": 0.5553, "step": 20438 }, { "epoch": 0.7281828384131678, "grad_norm": 1.3429189101542458, "learning_rate": 3.6310963709125014e-06, "loss": 0.4054, "step": 20439 }, { "epoch": 0.7282184655396619, "grad_norm": 1.4701044655055715, "learning_rate": 3.6302067807608334e-06, "loss": 0.4974, "step": 20440 }, { "epoch": 0.728254092666156, "grad_norm": 1.5933127393564621, "learning_rate": 3.6293172754274177e-06, "loss": 0.5621, "step": 20441 }, { "epoch": 0.7282897197926501, "grad_norm": 1.3866678770833394, "learning_rate": 3.6284278549240912e-06, "loss": 0.4485, "step": 20442 }, { "epoch": 0.7283253469191442, "grad_norm": 2.020018352614151, "learning_rate": 3.6275385192627056e-06, "loss": 0.4347, "step": 20443 }, { "epoch": 0.7283609740456384, "grad_norm": 1.425696657606341, "learning_rate": 3.626649268455095e-06, "loss": 0.5105, "step": 20444 }, { "epoch": 0.7283966011721325, "grad_norm": 1.5101290233590636, "learning_rate": 3.625760102513103e-06, "loss": 0.5342, "step": 20445 }, { "epoch": 0.7284322282986265, "grad_norm": 1.3950603500598346, "learning_rate": 3.6248710214485717e-06, "loss": 0.4447, "step": 20446 }, { "epoch": 0.7284678554251207, "grad_norm": 1.2796405195873608, "learning_rate": 3.6239820252733416e-06, "loss": 0.3875, "step": 20447 }, { "epoch": 0.7285034825516148, "grad_norm": 1.2794190076709875, "learning_rate": 3.6230931139992443e-06, "loss": 0.3832, "step": 20448 }, { "epoch": 0.7285391096781089, "grad_norm": 1.3332178705295132, "learning_rate": 3.6222042876381227e-06, "loss": 0.5023, "step": 20449 }, { "epoch": 0.728574736804603, "grad_norm": 1.452700235242227, "learning_rate": 3.621315546201806e-06, "loss": 0.4699, "step": 20450 }, { "epoch": 0.7286103639310971, "grad_norm": 1.480459299304111, "learning_rate": 3.6204268897021333e-06, "loss": 0.5692, "step": 20451 }, { "epoch": 0.7286459910575912, "grad_norm": 1.266824393709653, "learning_rate": 3.6195383181509304e-06, "loss": 0.416, "step": 20452 }, { "epoch": 0.7286816181840854, "grad_norm": 1.3526147241755935, "learning_rate": 3.618649831560037e-06, "loss": 0.4507, "step": 20453 }, { "epoch": 0.7287172453105795, "grad_norm": 1.2532638460265657, "learning_rate": 3.6177614299412845e-06, "loss": 0.416, "step": 20454 }, { "epoch": 0.7287528724370735, "grad_norm": 1.7559508974355749, "learning_rate": 3.616873113306496e-06, "loss": 0.4297, "step": 20455 }, { "epoch": 0.7287884995635677, "grad_norm": 1.2890958527498202, "learning_rate": 3.615984881667508e-06, "loss": 0.4089, "step": 20456 }, { "epoch": 0.7288241266900618, "grad_norm": 1.4695406990339903, "learning_rate": 3.61509673503614e-06, "loss": 0.5102, "step": 20457 }, { "epoch": 0.7288597538165559, "grad_norm": 1.4715821853543578, "learning_rate": 3.6142086734242242e-06, "loss": 0.4979, "step": 20458 }, { "epoch": 0.7288953809430501, "grad_norm": 1.9157532413007512, "learning_rate": 3.6133206968435795e-06, "loss": 0.6108, "step": 20459 }, { "epoch": 0.7289310080695441, "grad_norm": 1.5346650507993804, "learning_rate": 3.6124328053060354e-06, "loss": 0.6326, "step": 20460 }, { "epoch": 0.7289666351960382, "grad_norm": 1.6846381438750986, "learning_rate": 3.6115449988234118e-06, "loss": 0.4966, "step": 20461 }, { "epoch": 0.7290022623225324, "grad_norm": 1.8685778019407013, "learning_rate": 3.6106572774075354e-06, "loss": 0.5369, "step": 20462 }, { "epoch": 0.7290378894490265, "grad_norm": 1.4682680330863007, "learning_rate": 3.609769641070221e-06, "loss": 0.4033, "step": 20463 }, { "epoch": 0.7290735165755206, "grad_norm": 1.7364222649450514, "learning_rate": 3.608882089823289e-06, "loss": 0.5018, "step": 20464 }, { "epoch": 0.7291091437020147, "grad_norm": 1.0296906703265845, "learning_rate": 3.6079946236785633e-06, "loss": 0.3938, "step": 20465 }, { "epoch": 0.7291447708285088, "grad_norm": 1.1924779824765814, "learning_rate": 3.6071072426478527e-06, "loss": 0.4103, "step": 20466 }, { "epoch": 0.7291803979550029, "grad_norm": 2.49744611826962, "learning_rate": 3.6062199467429784e-06, "loss": 0.5457, "step": 20467 }, { "epoch": 0.7292160250814971, "grad_norm": 1.1566125127866647, "learning_rate": 3.605332735975754e-06, "loss": 0.3872, "step": 20468 }, { "epoch": 0.7292516522079912, "grad_norm": 1.5837142105603401, "learning_rate": 3.604445610357997e-06, "loss": 0.4893, "step": 20469 }, { "epoch": 0.7292872793344852, "grad_norm": 1.133687250953475, "learning_rate": 3.603558569901514e-06, "loss": 0.3895, "step": 20470 }, { "epoch": 0.7293229064609794, "grad_norm": 1.4142902769461507, "learning_rate": 3.6026716146181227e-06, "loss": 0.4688, "step": 20471 }, { "epoch": 0.7293585335874735, "grad_norm": 1.4782149667419868, "learning_rate": 3.6017847445196273e-06, "loss": 0.5287, "step": 20472 }, { "epoch": 0.7293941607139676, "grad_norm": 2.185152618413178, "learning_rate": 3.600897959617843e-06, "loss": 0.5344, "step": 20473 }, { "epoch": 0.7294297878404618, "grad_norm": 1.64492443059346, "learning_rate": 3.6000112599245717e-06, "loss": 0.5467, "step": 20474 }, { "epoch": 0.7294654149669558, "grad_norm": 1.377275195462272, "learning_rate": 3.599124645451624e-06, "loss": 0.5313, "step": 20475 }, { "epoch": 0.7295010420934499, "grad_norm": 1.5156253759571598, "learning_rate": 3.598238116210806e-06, "loss": 0.5042, "step": 20476 }, { "epoch": 0.7295366692199441, "grad_norm": 1.57814974207001, "learning_rate": 3.5973516722139212e-06, "loss": 0.4756, "step": 20477 }, { "epoch": 0.7295722963464382, "grad_norm": 1.8313022759184652, "learning_rate": 3.596465313472778e-06, "loss": 0.4832, "step": 20478 }, { "epoch": 0.7296079234729322, "grad_norm": 1.7613909064373725, "learning_rate": 3.5955790399991717e-06, "loss": 0.5697, "step": 20479 }, { "epoch": 0.7296435505994264, "grad_norm": 1.5545899894911321, "learning_rate": 3.5946928518049097e-06, "loss": 0.5736, "step": 20480 }, { "epoch": 0.7296791777259205, "grad_norm": 1.294188862904303, "learning_rate": 3.5938067489017857e-06, "loss": 0.4568, "step": 20481 }, { "epoch": 0.7297148048524146, "grad_norm": 1.095967939643683, "learning_rate": 3.5929207313016033e-06, "loss": 0.4001, "step": 20482 }, { "epoch": 0.7297504319789088, "grad_norm": 1.0904572032818098, "learning_rate": 3.59203479901616e-06, "loss": 0.385, "step": 20483 }, { "epoch": 0.7297860591054028, "grad_norm": 1.2811131532186861, "learning_rate": 3.591148952057255e-06, "loss": 0.4501, "step": 20484 }, { "epoch": 0.7298216862318969, "grad_norm": 1.6572081932235332, "learning_rate": 3.590263190436678e-06, "loss": 0.4993, "step": 20485 }, { "epoch": 0.7298573133583911, "grad_norm": 1.8036241169858696, "learning_rate": 3.5893775141662267e-06, "loss": 0.6047, "step": 20486 }, { "epoch": 0.7298929404848852, "grad_norm": 1.1848538670444404, "learning_rate": 3.5884919232576976e-06, "loss": 0.4447, "step": 20487 }, { "epoch": 0.7299285676113793, "grad_norm": 1.3530317975912705, "learning_rate": 3.5876064177228775e-06, "loss": 0.4491, "step": 20488 }, { "epoch": 0.7299641947378734, "grad_norm": 1.205124626491829, "learning_rate": 3.586720997573563e-06, "loss": 0.4523, "step": 20489 }, { "epoch": 0.7299998218643675, "grad_norm": 1.1492224630623955, "learning_rate": 3.585835662821534e-06, "loss": 0.4708, "step": 20490 }, { "epoch": 0.7300354489908616, "grad_norm": 1.16394232003664, "learning_rate": 3.5849504134785927e-06, "loss": 0.4329, "step": 20491 }, { "epoch": 0.7300710761173558, "grad_norm": 1.264206200664399, "learning_rate": 3.5840652495565178e-06, "loss": 0.4351, "step": 20492 }, { "epoch": 0.7301067032438499, "grad_norm": 2.0945008138921195, "learning_rate": 3.583180171067101e-06, "loss": 0.5379, "step": 20493 }, { "epoch": 0.7301423303703439, "grad_norm": 1.1356207389339137, "learning_rate": 3.582295178022123e-06, "loss": 0.4042, "step": 20494 }, { "epoch": 0.7301779574968381, "grad_norm": 1.2781586148955388, "learning_rate": 3.581410270433373e-06, "loss": 0.4218, "step": 20495 }, { "epoch": 0.7302135846233322, "grad_norm": 1.2294330395954234, "learning_rate": 3.5805254483126295e-06, "loss": 0.3759, "step": 20496 }, { "epoch": 0.7302492117498263, "grad_norm": 1.1661406951229967, "learning_rate": 3.579640711671676e-06, "loss": 0.3815, "step": 20497 }, { "epoch": 0.7302848388763205, "grad_norm": 1.1002853046422505, "learning_rate": 3.5787560605222926e-06, "loss": 0.3462, "step": 20498 }, { "epoch": 0.7303204660028145, "grad_norm": 1.9627574118729239, "learning_rate": 3.5778714948762617e-06, "loss": 0.5678, "step": 20499 }, { "epoch": 0.7303560931293086, "grad_norm": 1.0448303009699724, "learning_rate": 3.5769870147453643e-06, "loss": 0.3669, "step": 20500 }, { "epoch": 0.7303917202558028, "grad_norm": 1.279234593407615, "learning_rate": 3.5761026201413708e-06, "loss": 0.4532, "step": 20501 }, { "epoch": 0.7304273473822969, "grad_norm": 1.8356043025674411, "learning_rate": 3.575218311076063e-06, "loss": 0.5279, "step": 20502 }, { "epoch": 0.7304629745087909, "grad_norm": 1.1361493327205348, "learning_rate": 3.574334087561212e-06, "loss": 0.4021, "step": 20503 }, { "epoch": 0.7304986016352851, "grad_norm": 1.6576961232183207, "learning_rate": 3.573449949608594e-06, "loss": 0.5605, "step": 20504 }, { "epoch": 0.7305342287617792, "grad_norm": 1.1597730328574396, "learning_rate": 3.57256589722998e-06, "loss": 0.3903, "step": 20505 }, { "epoch": 0.7305698558882733, "grad_norm": 1.1290808442064593, "learning_rate": 3.5716819304371485e-06, "loss": 0.3867, "step": 20506 }, { "epoch": 0.7306054830147675, "grad_norm": 1.6386010071806096, "learning_rate": 3.5707980492418624e-06, "loss": 0.4573, "step": 20507 }, { "epoch": 0.7306411101412615, "grad_norm": 1.4265765497346041, "learning_rate": 3.569914253655896e-06, "loss": 0.41, "step": 20508 }, { "epoch": 0.7306767372677556, "grad_norm": 1.28685655772135, "learning_rate": 3.5690305436910133e-06, "loss": 0.4225, "step": 20509 }, { "epoch": 0.7307123643942498, "grad_norm": 1.4603837492981115, "learning_rate": 3.5681469193589834e-06, "loss": 0.4733, "step": 20510 }, { "epoch": 0.7307479915207439, "grad_norm": 1.588477066883248, "learning_rate": 3.5672633806715763e-06, "loss": 0.5172, "step": 20511 }, { "epoch": 0.730783618647238, "grad_norm": 1.2451076150229416, "learning_rate": 3.5663799276405496e-06, "loss": 0.4293, "step": 20512 }, { "epoch": 0.7308192457737321, "grad_norm": 1.4854152018258928, "learning_rate": 3.5654965602776724e-06, "loss": 0.5169, "step": 20513 }, { "epoch": 0.7308548729002262, "grad_norm": 1.4098827819360873, "learning_rate": 3.564613278594705e-06, "loss": 0.5267, "step": 20514 }, { "epoch": 0.7308905000267203, "grad_norm": 2.2586983705852983, "learning_rate": 3.5637300826034126e-06, "loss": 0.566, "step": 20515 }, { "epoch": 0.7309261271532145, "grad_norm": 1.1651330178926362, "learning_rate": 3.5628469723155513e-06, "loss": 0.4423, "step": 20516 }, { "epoch": 0.7309617542797086, "grad_norm": 1.344966154014329, "learning_rate": 3.5619639477428836e-06, "loss": 0.409, "step": 20517 }, { "epoch": 0.7309973814062026, "grad_norm": 1.2154161452635315, "learning_rate": 3.5610810088971627e-06, "loss": 0.4084, "step": 20518 }, { "epoch": 0.7310330085326968, "grad_norm": 1.553428859353747, "learning_rate": 3.5601981557901488e-06, "loss": 0.5087, "step": 20519 }, { "epoch": 0.7310686356591909, "grad_norm": 1.2297403933983886, "learning_rate": 3.5593153884335986e-06, "loss": 0.4495, "step": 20520 }, { "epoch": 0.731104262785685, "grad_norm": 1.5671592401290133, "learning_rate": 3.5584327068392643e-06, "loss": 0.5467, "step": 20521 }, { "epoch": 0.7311398899121792, "grad_norm": 1.1889868643796881, "learning_rate": 3.557550111018906e-06, "loss": 0.4298, "step": 20522 }, { "epoch": 0.7311755170386732, "grad_norm": 1.5240853297299501, "learning_rate": 3.556667600984266e-06, "loss": 0.5154, "step": 20523 }, { "epoch": 0.7312111441651673, "grad_norm": 1.0849506016139316, "learning_rate": 3.555785176747104e-06, "loss": 0.397, "step": 20524 }, { "epoch": 0.7312467712916615, "grad_norm": 1.3933767770358734, "learning_rate": 3.554902838319164e-06, "loss": 0.494, "step": 20525 }, { "epoch": 0.7312823984181556, "grad_norm": 1.2921261732114202, "learning_rate": 3.554020585712201e-06, "loss": 0.5267, "step": 20526 }, { "epoch": 0.7313180255446496, "grad_norm": 1.3834221743865798, "learning_rate": 3.5531384189379515e-06, "loss": 0.4875, "step": 20527 }, { "epoch": 0.7313536526711438, "grad_norm": 1.4515551374956042, "learning_rate": 3.5522563380081778e-06, "loss": 0.5779, "step": 20528 }, { "epoch": 0.7313892797976379, "grad_norm": 1.8938398454268848, "learning_rate": 3.5513743429346135e-06, "loss": 0.5952, "step": 20529 }, { "epoch": 0.731424906924132, "grad_norm": 1.5320255390280333, "learning_rate": 3.5504924337290113e-06, "loss": 0.4404, "step": 20530 }, { "epoch": 0.7314605340506262, "grad_norm": 1.8871298775024279, "learning_rate": 3.549610610403107e-06, "loss": 0.4962, "step": 20531 }, { "epoch": 0.7314961611771202, "grad_norm": 1.2064052301416226, "learning_rate": 3.5487288729686454e-06, "loss": 0.3977, "step": 20532 }, { "epoch": 0.7315317883036143, "grad_norm": 1.1190268618242147, "learning_rate": 3.5478472214373716e-06, "loss": 0.4253, "step": 20533 }, { "epoch": 0.7315674154301085, "grad_norm": 1.7136033318105832, "learning_rate": 3.5469656558210186e-06, "loss": 0.4831, "step": 20534 }, { "epoch": 0.7316030425566026, "grad_norm": 1.1925789550765353, "learning_rate": 3.5460841761313282e-06, "loss": 0.4213, "step": 20535 }, { "epoch": 0.7316386696830968, "grad_norm": 1.0804076427860119, "learning_rate": 3.5452027823800384e-06, "loss": 0.4235, "step": 20536 }, { "epoch": 0.7316742968095908, "grad_norm": 1.7219395468775403, "learning_rate": 3.544321474578888e-06, "loss": 0.5282, "step": 20537 }, { "epoch": 0.7317099239360849, "grad_norm": 1.347925645355576, "learning_rate": 3.543440252739606e-06, "loss": 0.5148, "step": 20538 }, { "epoch": 0.7317455510625791, "grad_norm": 1.1902010746993645, "learning_rate": 3.5425591168739336e-06, "loss": 0.4264, "step": 20539 }, { "epoch": 0.7317811781890732, "grad_norm": 1.8992130883526894, "learning_rate": 3.541678066993597e-06, "loss": 0.4935, "step": 20540 }, { "epoch": 0.7318168053155673, "grad_norm": 1.6451327084615557, "learning_rate": 3.5407971031103305e-06, "loss": 0.4469, "step": 20541 }, { "epoch": 0.7318524324420614, "grad_norm": 1.730722922561657, "learning_rate": 3.5399162252358653e-06, "loss": 0.5176, "step": 20542 }, { "epoch": 0.7318880595685555, "grad_norm": 1.3319215555596584, "learning_rate": 3.5390354333819344e-06, "loss": 0.5234, "step": 20543 }, { "epoch": 0.7319236866950496, "grad_norm": 1.5479002244770521, "learning_rate": 3.538154727560259e-06, "loss": 0.4939, "step": 20544 }, { "epoch": 0.7319593138215438, "grad_norm": 1.1788759345561575, "learning_rate": 3.53727410778257e-06, "loss": 0.3857, "step": 20545 }, { "epoch": 0.7319949409480379, "grad_norm": 1.5452981921810744, "learning_rate": 3.536393574060597e-06, "loss": 0.4801, "step": 20546 }, { "epoch": 0.7320305680745319, "grad_norm": 2.1440677302910416, "learning_rate": 3.5355131264060573e-06, "loss": 0.5999, "step": 20547 }, { "epoch": 0.7320661952010261, "grad_norm": 1.7188726941817174, "learning_rate": 3.5346327648306834e-06, "loss": 0.5905, "step": 20548 }, { "epoch": 0.7321018223275202, "grad_norm": 1.5248307460275414, "learning_rate": 3.5337524893461894e-06, "loss": 0.4762, "step": 20549 }, { "epoch": 0.7321374494540143, "grad_norm": 1.4469129925256734, "learning_rate": 3.5328722999642997e-06, "loss": 0.5202, "step": 20550 }, { "epoch": 0.7321730765805085, "grad_norm": 1.3216432176860498, "learning_rate": 3.531992196696736e-06, "loss": 0.4542, "step": 20551 }, { "epoch": 0.7322087037070025, "grad_norm": 1.7037660741067635, "learning_rate": 3.531112179555222e-06, "loss": 0.5911, "step": 20552 }, { "epoch": 0.7322443308334966, "grad_norm": 1.1440068044938028, "learning_rate": 3.530232248551466e-06, "loss": 0.4418, "step": 20553 }, { "epoch": 0.7322799579599908, "grad_norm": 1.169197197255815, "learning_rate": 3.5293524036971927e-06, "loss": 0.4154, "step": 20554 }, { "epoch": 0.7323155850864849, "grad_norm": 1.4974183099301734, "learning_rate": 3.528472645004112e-06, "loss": 0.4941, "step": 20555 }, { "epoch": 0.7323512122129789, "grad_norm": 1.8356447835138363, "learning_rate": 3.527592972483942e-06, "loss": 0.5095, "step": 20556 }, { "epoch": 0.7323868393394731, "grad_norm": 1.0743816993343391, "learning_rate": 3.5267133861483948e-06, "loss": 0.4238, "step": 20557 }, { "epoch": 0.7324224664659672, "grad_norm": 1.1168133444001076, "learning_rate": 3.5258338860091836e-06, "loss": 0.3829, "step": 20558 }, { "epoch": 0.7324580935924613, "grad_norm": 1.0871488459411376, "learning_rate": 3.5249544720780225e-06, "loss": 0.4317, "step": 20559 }, { "epoch": 0.7324937207189555, "grad_norm": 1.360301846302279, "learning_rate": 3.5240751443666157e-06, "loss": 0.424, "step": 20560 }, { "epoch": 0.7325293478454495, "grad_norm": 1.1914508193874485, "learning_rate": 3.523195902886677e-06, "loss": 0.4446, "step": 20561 }, { "epoch": 0.7325649749719436, "grad_norm": 1.2644768168976996, "learning_rate": 3.5223167476499098e-06, "loss": 0.3772, "step": 20562 }, { "epoch": 0.7326006020984378, "grad_norm": 1.6201012592561552, "learning_rate": 3.5214376786680248e-06, "loss": 0.5106, "step": 20563 }, { "epoch": 0.7326362292249319, "grad_norm": 1.1606002985436021, "learning_rate": 3.5205586959527228e-06, "loss": 0.3867, "step": 20564 }, { "epoch": 0.732671856351426, "grad_norm": 1.460582042308897, "learning_rate": 3.5196797995157105e-06, "loss": 0.5021, "step": 20565 }, { "epoch": 0.7327074834779201, "grad_norm": 1.9063769991160382, "learning_rate": 3.5188009893686916e-06, "loss": 0.4533, "step": 20566 }, { "epoch": 0.7327431106044142, "grad_norm": 1.325377646233857, "learning_rate": 3.517922265523367e-06, "loss": 0.4626, "step": 20567 }, { "epoch": 0.7327787377309083, "grad_norm": 1.3700887035849822, "learning_rate": 3.517043627991441e-06, "loss": 0.3814, "step": 20568 }, { "epoch": 0.7328143648574025, "grad_norm": 1.1507670922969915, "learning_rate": 3.5161650767846077e-06, "loss": 0.4425, "step": 20569 }, { "epoch": 0.7328499919838966, "grad_norm": 1.1693731315025626, "learning_rate": 3.5152866119145722e-06, "loss": 0.4517, "step": 20570 }, { "epoch": 0.7328856191103906, "grad_norm": 1.8042397096789096, "learning_rate": 3.514408233393024e-06, "loss": 0.4725, "step": 20571 }, { "epoch": 0.7329212462368848, "grad_norm": 1.6968452562627534, "learning_rate": 3.5135299412316635e-06, "loss": 0.476, "step": 20572 }, { "epoch": 0.7329568733633789, "grad_norm": 1.1813680675629352, "learning_rate": 3.5126517354421853e-06, "loss": 0.4627, "step": 20573 }, { "epoch": 0.732992500489873, "grad_norm": 1.4082235726160004, "learning_rate": 3.511773616036287e-06, "loss": 0.4544, "step": 20574 }, { "epoch": 0.7330281276163672, "grad_norm": 1.6665741459897718, "learning_rate": 3.5108955830256552e-06, "loss": 0.5335, "step": 20575 }, { "epoch": 0.7330637547428612, "grad_norm": 1.1942082494395827, "learning_rate": 3.5100176364219875e-06, "loss": 0.4294, "step": 20576 }, { "epoch": 0.7330993818693553, "grad_norm": 1.2699866454099755, "learning_rate": 3.509139776236967e-06, "loss": 0.411, "step": 20577 }, { "epoch": 0.7331350089958495, "grad_norm": 1.5077518486142003, "learning_rate": 3.508262002482289e-06, "loss": 0.5505, "step": 20578 }, { "epoch": 0.7331706361223436, "grad_norm": 1.15208854154395, "learning_rate": 3.5073843151696387e-06, "loss": 0.382, "step": 20579 }, { "epoch": 0.7332062632488376, "grad_norm": 1.1808640288231138, "learning_rate": 3.5065067143107046e-06, "loss": 0.4545, "step": 20580 }, { "epoch": 0.7332418903753318, "grad_norm": 1.25039734276359, "learning_rate": 3.5056291999171754e-06, "loss": 0.4373, "step": 20581 }, { "epoch": 0.7332775175018259, "grad_norm": 1.1768543005948169, "learning_rate": 3.5047517720007283e-06, "loss": 0.4501, "step": 20582 }, { "epoch": 0.73331314462832, "grad_norm": 1.3415453197453473, "learning_rate": 3.5038744305730566e-06, "loss": 0.5311, "step": 20583 }, { "epoch": 0.7333487717548142, "grad_norm": 1.6392370401311955, "learning_rate": 3.5029971756458323e-06, "loss": 0.5352, "step": 20584 }, { "epoch": 0.7333843988813082, "grad_norm": 1.7664620035205736, "learning_rate": 3.502120007230747e-06, "loss": 0.5843, "step": 20585 }, { "epoch": 0.7334200260078023, "grad_norm": 1.1876685950495656, "learning_rate": 3.5012429253394707e-06, "loss": 0.418, "step": 20586 }, { "epoch": 0.7334556531342965, "grad_norm": 1.1541150155865212, "learning_rate": 3.5003659299836877e-06, "loss": 0.4123, "step": 20587 }, { "epoch": 0.7334912802607906, "grad_norm": 1.6300198015187033, "learning_rate": 3.4994890211750754e-06, "loss": 0.4903, "step": 20588 }, { "epoch": 0.7335269073872847, "grad_norm": 1.195712181105712, "learning_rate": 3.498612198925313e-06, "loss": 0.4494, "step": 20589 }, { "epoch": 0.7335625345137788, "grad_norm": 1.3265819678960438, "learning_rate": 3.4977354632460715e-06, "loss": 0.5435, "step": 20590 }, { "epoch": 0.7335981616402729, "grad_norm": 1.2348152728570072, "learning_rate": 3.4968588141490255e-06, "loss": 0.4295, "step": 20591 }, { "epoch": 0.733633788766767, "grad_norm": 1.6119148915524883, "learning_rate": 3.4959822516458543e-06, "loss": 0.5375, "step": 20592 }, { "epoch": 0.7336694158932612, "grad_norm": 1.3030684014482898, "learning_rate": 3.495105775748221e-06, "loss": 0.4345, "step": 20593 }, { "epoch": 0.7337050430197553, "grad_norm": 1.4863879732788865, "learning_rate": 3.4942293864678023e-06, "loss": 0.5534, "step": 20594 }, { "epoch": 0.7337406701462493, "grad_norm": 1.7210954488723054, "learning_rate": 3.4933530838162654e-06, "loss": 0.5674, "step": 20595 }, { "epoch": 0.7337762972727435, "grad_norm": 1.215769569016453, "learning_rate": 3.4924768678052845e-06, "loss": 0.4102, "step": 20596 }, { "epoch": 0.7338119243992376, "grad_norm": 1.566087901023681, "learning_rate": 3.491600738446519e-06, "loss": 0.4741, "step": 20597 }, { "epoch": 0.7338475515257317, "grad_norm": 1.151620286235139, "learning_rate": 3.4907246957516416e-06, "loss": 0.4312, "step": 20598 }, { "epoch": 0.7338831786522259, "grad_norm": 1.5848829795194097, "learning_rate": 3.489848739732311e-06, "loss": 0.5453, "step": 20599 }, { "epoch": 0.7339188057787199, "grad_norm": 1.8024770808544432, "learning_rate": 3.488972870400199e-06, "loss": 0.4967, "step": 20600 }, { "epoch": 0.733954432905214, "grad_norm": 1.0866187052665832, "learning_rate": 3.4880970877669594e-06, "loss": 0.4002, "step": 20601 }, { "epoch": 0.7339900600317082, "grad_norm": 1.2307816844636288, "learning_rate": 3.48722139184426e-06, "loss": 0.398, "step": 20602 }, { "epoch": 0.7340256871582023, "grad_norm": 1.1826556454123733, "learning_rate": 3.4863457826437597e-06, "loss": 0.3969, "step": 20603 }, { "epoch": 0.7340613142846963, "grad_norm": 1.6378898728476525, "learning_rate": 3.485470260177117e-06, "loss": 0.5664, "step": 20604 }, { "epoch": 0.7340969414111905, "grad_norm": 1.0588689647620952, "learning_rate": 3.484594824455996e-06, "loss": 0.41, "step": 20605 }, { "epoch": 0.7341325685376846, "grad_norm": 1.10340376621703, "learning_rate": 3.4837194754920446e-06, "loss": 0.3885, "step": 20606 }, { "epoch": 0.7341681956641787, "grad_norm": 1.111037287208271, "learning_rate": 3.4828442132969275e-06, "loss": 0.4394, "step": 20607 }, { "epoch": 0.7342038227906729, "grad_norm": 1.313680052774258, "learning_rate": 3.48196903788229e-06, "loss": 0.4862, "step": 20608 }, { "epoch": 0.7342394499171669, "grad_norm": 1.2991290645640434, "learning_rate": 3.4810939492597918e-06, "loss": 0.4055, "step": 20609 }, { "epoch": 0.734275077043661, "grad_norm": 1.1333115615005473, "learning_rate": 3.480218947441083e-06, "loss": 0.4057, "step": 20610 }, { "epoch": 0.7343107041701552, "grad_norm": 1.402391293699279, "learning_rate": 3.4793440324378203e-06, "loss": 0.4058, "step": 20611 }, { "epoch": 0.7343463312966493, "grad_norm": 1.1781177017807323, "learning_rate": 3.4784692042616463e-06, "loss": 0.4289, "step": 20612 }, { "epoch": 0.7343819584231434, "grad_norm": 1.1430340231271234, "learning_rate": 3.477594462924212e-06, "loss": 0.4059, "step": 20613 }, { "epoch": 0.7344175855496375, "grad_norm": 1.140569011877755, "learning_rate": 3.476719808437171e-06, "loss": 0.3601, "step": 20614 }, { "epoch": 0.7344532126761316, "grad_norm": 1.7915186408688049, "learning_rate": 3.475845240812161e-06, "loss": 0.5253, "step": 20615 }, { "epoch": 0.7344888398026257, "grad_norm": 1.2045998396465687, "learning_rate": 3.4749707600608318e-06, "loss": 0.4748, "step": 20616 }, { "epoch": 0.7345244669291199, "grad_norm": 1.331801571499206, "learning_rate": 3.474096366194828e-06, "loss": 0.442, "step": 20617 }, { "epoch": 0.734560094055614, "grad_norm": 1.4313976806539928, "learning_rate": 3.473222059225795e-06, "loss": 0.5186, "step": 20618 }, { "epoch": 0.734595721182108, "grad_norm": 1.2526232957487007, "learning_rate": 3.47234783916537e-06, "loss": 0.4415, "step": 20619 }, { "epoch": 0.7346313483086022, "grad_norm": 1.7975687297822347, "learning_rate": 3.471473706025199e-06, "loss": 0.4782, "step": 20620 }, { "epoch": 0.7346669754350963, "grad_norm": 4.07676956221046, "learning_rate": 3.4705996598169146e-06, "loss": 0.4071, "step": 20621 }, { "epoch": 0.7347026025615904, "grad_norm": 1.6841628058000224, "learning_rate": 3.4697257005521635e-06, "loss": 0.5587, "step": 20622 }, { "epoch": 0.7347382296880846, "grad_norm": 1.3724002096639112, "learning_rate": 3.4688518282425764e-06, "loss": 0.4318, "step": 20623 }, { "epoch": 0.7347738568145786, "grad_norm": 1.681846051985988, "learning_rate": 3.467978042899791e-06, "loss": 0.5353, "step": 20624 }, { "epoch": 0.7348094839410727, "grad_norm": 1.7512626311503483, "learning_rate": 3.4671043445354446e-06, "loss": 0.5332, "step": 20625 }, { "epoch": 0.7348451110675669, "grad_norm": 1.3818274229265963, "learning_rate": 3.4662307331611687e-06, "loss": 0.5537, "step": 20626 }, { "epoch": 0.734880738194061, "grad_norm": 1.210070112549278, "learning_rate": 3.465357208788602e-06, "loss": 0.4094, "step": 20627 }, { "epoch": 0.734916365320555, "grad_norm": 1.547458690782865, "learning_rate": 3.464483771429368e-06, "loss": 0.5189, "step": 20628 }, { "epoch": 0.7349519924470492, "grad_norm": 1.2050130775706316, "learning_rate": 3.4636104210951038e-06, "loss": 0.4418, "step": 20629 }, { "epoch": 0.7349876195735433, "grad_norm": 1.5026095044016234, "learning_rate": 3.4627371577974323e-06, "loss": 0.5931, "step": 20630 }, { "epoch": 0.7350232467000374, "grad_norm": 1.1614589280858394, "learning_rate": 3.461863981547985e-06, "loss": 0.4238, "step": 20631 }, { "epoch": 0.7350588738265316, "grad_norm": 1.4803990486990435, "learning_rate": 3.460990892358388e-06, "loss": 0.4507, "step": 20632 }, { "epoch": 0.7350945009530256, "grad_norm": 1.302022109577569, "learning_rate": 3.460117890240271e-06, "loss": 0.4049, "step": 20633 }, { "epoch": 0.7351301280795197, "grad_norm": 1.3504498518610069, "learning_rate": 3.4592449752052526e-06, "loss": 0.5285, "step": 20634 }, { "epoch": 0.7351657552060139, "grad_norm": 1.2585808147592161, "learning_rate": 3.458372147264962e-06, "loss": 0.4081, "step": 20635 }, { "epoch": 0.735201382332508, "grad_norm": 2.460418232181238, "learning_rate": 3.457499406431016e-06, "loss": 0.5059, "step": 20636 }, { "epoch": 0.735237009459002, "grad_norm": 1.447626519471603, "learning_rate": 3.4566267527150387e-06, "loss": 0.5795, "step": 20637 }, { "epoch": 0.7352726365854962, "grad_norm": 1.1342680651517243, "learning_rate": 3.4557541861286512e-06, "loss": 0.3815, "step": 20638 }, { "epoch": 0.7353082637119903, "grad_norm": 1.0371661236611187, "learning_rate": 3.4548817066834693e-06, "loss": 0.4175, "step": 20639 }, { "epoch": 0.7353438908384844, "grad_norm": 1.5858293787212967, "learning_rate": 3.454009314391111e-06, "loss": 0.4793, "step": 20640 }, { "epoch": 0.7353795179649786, "grad_norm": 1.3871243195013332, "learning_rate": 3.4531370092631955e-06, "loss": 0.5225, "step": 20641 }, { "epoch": 0.7354151450914727, "grad_norm": 1.5614532245326929, "learning_rate": 3.4522647913113382e-06, "loss": 0.5104, "step": 20642 }, { "epoch": 0.7354507722179667, "grad_norm": 1.1867220191093628, "learning_rate": 3.4513926605471504e-06, "loss": 0.4161, "step": 20643 }, { "epoch": 0.7354863993444609, "grad_norm": 1.3390259468316446, "learning_rate": 3.450520616982248e-06, "loss": 0.4751, "step": 20644 }, { "epoch": 0.735522026470955, "grad_norm": 2.1540341407166737, "learning_rate": 3.44964866062824e-06, "loss": 0.4236, "step": 20645 }, { "epoch": 0.7355576535974491, "grad_norm": 1.3775754068334174, "learning_rate": 3.4487767914967375e-06, "loss": 0.3905, "step": 20646 }, { "epoch": 0.7355932807239433, "grad_norm": 2.2839500503489027, "learning_rate": 3.4479050095993507e-06, "loss": 0.5446, "step": 20647 }, { "epoch": 0.7356289078504373, "grad_norm": 1.4886687004804917, "learning_rate": 3.4470333149476885e-06, "loss": 0.3969, "step": 20648 }, { "epoch": 0.7356645349769314, "grad_norm": 1.4191558404078026, "learning_rate": 3.4461617075533614e-06, "loss": 0.4011, "step": 20649 }, { "epoch": 0.7357001621034256, "grad_norm": 1.53373972829467, "learning_rate": 3.445290187427969e-06, "loss": 0.5682, "step": 20650 }, { "epoch": 0.7357357892299197, "grad_norm": 1.8425385664474556, "learning_rate": 3.444418754583122e-06, "loss": 0.3976, "step": 20651 }, { "epoch": 0.7357714163564137, "grad_norm": 1.5466979751104255, "learning_rate": 3.443547409030418e-06, "loss": 0.4642, "step": 20652 }, { "epoch": 0.7358070434829079, "grad_norm": 1.3276716788103604, "learning_rate": 3.4426761507814664e-06, "loss": 0.4238, "step": 20653 }, { "epoch": 0.735842670609402, "grad_norm": 1.432662242519631, "learning_rate": 3.4418049798478594e-06, "loss": 0.5381, "step": 20654 }, { "epoch": 0.7358782977358961, "grad_norm": 1.6552944184265264, "learning_rate": 3.440933896241209e-06, "loss": 0.461, "step": 20655 }, { "epoch": 0.7359139248623903, "grad_norm": 1.6913163638754725, "learning_rate": 3.4400628999731046e-06, "loss": 0.5908, "step": 20656 }, { "epoch": 0.7359495519888843, "grad_norm": 2.0499100471741905, "learning_rate": 3.4391919910551506e-06, "loss": 0.5405, "step": 20657 }, { "epoch": 0.7359851791153784, "grad_norm": 1.3644593087107597, "learning_rate": 3.4383211694989395e-06, "loss": 0.5159, "step": 20658 }, { "epoch": 0.7360208062418726, "grad_norm": 1.5648613388287054, "learning_rate": 3.4374504353160665e-06, "loss": 0.432, "step": 20659 }, { "epoch": 0.7360564333683667, "grad_norm": 1.607687406355964, "learning_rate": 3.436579788518132e-06, "loss": 0.5014, "step": 20660 }, { "epoch": 0.7360920604948608, "grad_norm": 1.3001151437907044, "learning_rate": 3.4357092291167227e-06, "loss": 0.4613, "step": 20661 }, { "epoch": 0.7361276876213549, "grad_norm": 1.7702553583323348, "learning_rate": 3.434838757123432e-06, "loss": 0.4394, "step": 20662 }, { "epoch": 0.736163314747849, "grad_norm": 1.3575964920346217, "learning_rate": 3.433968372549852e-06, "loss": 0.4396, "step": 20663 }, { "epoch": 0.7361989418743431, "grad_norm": 1.1636182381623539, "learning_rate": 3.433098075407576e-06, "loss": 0.4174, "step": 20664 }, { "epoch": 0.7362345690008373, "grad_norm": 1.342781809274627, "learning_rate": 3.4322278657081866e-06, "loss": 0.3962, "step": 20665 }, { "epoch": 0.7362701961273314, "grad_norm": 1.5688410581274848, "learning_rate": 3.431357743463276e-06, "loss": 0.4625, "step": 20666 }, { "epoch": 0.7363058232538254, "grad_norm": 1.3090040366551599, "learning_rate": 3.430487708684426e-06, "loss": 0.4601, "step": 20667 }, { "epoch": 0.7363414503803196, "grad_norm": 1.7810615394792273, "learning_rate": 3.429617761383223e-06, "loss": 0.4935, "step": 20668 }, { "epoch": 0.7363770775068137, "grad_norm": 1.4962780628032541, "learning_rate": 3.428747901571252e-06, "loss": 0.559, "step": 20669 }, { "epoch": 0.7364127046333078, "grad_norm": 1.3317433054758996, "learning_rate": 3.4278781292601005e-06, "loss": 0.4084, "step": 20670 }, { "epoch": 0.736448331759802, "grad_norm": 1.6406926802693855, "learning_rate": 3.4270084444613404e-06, "loss": 0.5792, "step": 20671 }, { "epoch": 0.736483958886296, "grad_norm": 1.5324167346279214, "learning_rate": 3.4261388471865574e-06, "loss": 0.5358, "step": 20672 }, { "epoch": 0.7365195860127901, "grad_norm": 1.307050074704735, "learning_rate": 3.4252693374473343e-06, "loss": 0.5001, "step": 20673 }, { "epoch": 0.7365552131392843, "grad_norm": 1.2677680776926874, "learning_rate": 3.424399915255242e-06, "loss": 0.4473, "step": 20674 }, { "epoch": 0.7365908402657784, "grad_norm": 1.313099979797037, "learning_rate": 3.423530580621863e-06, "loss": 0.4308, "step": 20675 }, { "epoch": 0.7366264673922724, "grad_norm": 1.5659629866520417, "learning_rate": 3.42266133355877e-06, "loss": 0.5242, "step": 20676 }, { "epoch": 0.7366620945187666, "grad_norm": 1.4114582880200977, "learning_rate": 3.421792174077537e-06, "loss": 0.4276, "step": 20677 }, { "epoch": 0.7366977216452607, "grad_norm": 1.3373687166753088, "learning_rate": 3.4209231021897395e-06, "loss": 0.4645, "step": 20678 }, { "epoch": 0.7367333487717548, "grad_norm": 1.2754866588709128, "learning_rate": 3.4200541179069525e-06, "loss": 0.4293, "step": 20679 }, { "epoch": 0.736768975898249, "grad_norm": 1.0457254168820203, "learning_rate": 3.4191852212407406e-06, "loss": 0.4208, "step": 20680 }, { "epoch": 0.736804603024743, "grad_norm": 1.1376942430055021, "learning_rate": 3.418316412202681e-06, "loss": 0.391, "step": 20681 }, { "epoch": 0.7368402301512371, "grad_norm": 1.2681095283985995, "learning_rate": 3.4174476908043354e-06, "loss": 0.45, "step": 20682 }, { "epoch": 0.7368758572777313, "grad_norm": 1.2977008860385324, "learning_rate": 3.4165790570572744e-06, "loss": 0.4065, "step": 20683 }, { "epoch": 0.7369114844042254, "grad_norm": 1.2779250310173154, "learning_rate": 3.4157105109730646e-06, "loss": 0.3766, "step": 20684 }, { "epoch": 0.7369471115307195, "grad_norm": 1.4438161653768036, "learning_rate": 3.414842052563272e-06, "loss": 0.5036, "step": 20685 }, { "epoch": 0.7369827386572136, "grad_norm": 1.302131206602512, "learning_rate": 3.413973681839463e-06, "loss": 0.4204, "step": 20686 }, { "epoch": 0.7370183657837077, "grad_norm": 1.3206299055677735, "learning_rate": 3.4131053988131947e-06, "loss": 0.4522, "step": 20687 }, { "epoch": 0.7370539929102018, "grad_norm": 1.2114278993179313, "learning_rate": 3.412237203496036e-06, "loss": 0.4478, "step": 20688 }, { "epoch": 0.737089620036696, "grad_norm": 1.8291177285759035, "learning_rate": 3.4113690958995395e-06, "loss": 0.5595, "step": 20689 }, { "epoch": 0.7371252471631901, "grad_norm": 2.1589554350317184, "learning_rate": 3.410501076035272e-06, "loss": 0.5655, "step": 20690 }, { "epoch": 0.7371608742896841, "grad_norm": 1.2880233171432651, "learning_rate": 3.409633143914781e-06, "loss": 0.4169, "step": 20691 }, { "epoch": 0.7371965014161783, "grad_norm": 1.4227526357575746, "learning_rate": 3.408765299549639e-06, "loss": 0.3715, "step": 20692 }, { "epoch": 0.7372321285426724, "grad_norm": 1.6647075979542887, "learning_rate": 3.4078975429513906e-06, "loss": 0.5451, "step": 20693 }, { "epoch": 0.7372677556691665, "grad_norm": 1.3111222150057165, "learning_rate": 3.4070298741315932e-06, "loss": 0.3963, "step": 20694 }, { "epoch": 0.7373033827956607, "grad_norm": 1.1941743612890048, "learning_rate": 3.4061622931018057e-06, "loss": 0.3878, "step": 20695 }, { "epoch": 0.7373390099221547, "grad_norm": 1.130779749778992, "learning_rate": 3.405294799873572e-06, "loss": 0.3861, "step": 20696 }, { "epoch": 0.7373746370486488, "grad_norm": 1.4035386939173928, "learning_rate": 3.4044273944584507e-06, "loss": 0.4122, "step": 20697 }, { "epoch": 0.737410264175143, "grad_norm": 1.4368565353272342, "learning_rate": 3.4035600768679855e-06, "loss": 0.4573, "step": 20698 }, { "epoch": 0.7374458913016371, "grad_norm": 1.3533809013406815, "learning_rate": 3.4026928471137286e-06, "loss": 0.4292, "step": 20699 }, { "epoch": 0.7374815184281311, "grad_norm": 1.3612564564695584, "learning_rate": 3.401825705207228e-06, "loss": 0.3841, "step": 20700 }, { "epoch": 0.7375171455546253, "grad_norm": 1.1458869899172708, "learning_rate": 3.4009586511600335e-06, "loss": 0.3862, "step": 20701 }, { "epoch": 0.7375527726811194, "grad_norm": 1.6337297938956095, "learning_rate": 3.400091684983683e-06, "loss": 0.4996, "step": 20702 }, { "epoch": 0.7375883998076135, "grad_norm": 2.3767639702048586, "learning_rate": 3.3992248066897283e-06, "loss": 0.5215, "step": 20703 }, { "epoch": 0.7376240269341077, "grad_norm": 1.2716293662740263, "learning_rate": 3.3983580162897057e-06, "loss": 0.4329, "step": 20704 }, { "epoch": 0.7376596540606017, "grad_norm": 1.3296039391655383, "learning_rate": 3.3974913137951605e-06, "loss": 0.3823, "step": 20705 }, { "epoch": 0.7376952811870958, "grad_norm": 1.9463978086614568, "learning_rate": 3.396624699217632e-06, "loss": 0.5247, "step": 20706 }, { "epoch": 0.73773090831359, "grad_norm": 1.3317814641628258, "learning_rate": 3.395758172568663e-06, "loss": 0.4954, "step": 20707 }, { "epoch": 0.7377665354400841, "grad_norm": 2.003382957709096, "learning_rate": 3.394891733859792e-06, "loss": 0.5091, "step": 20708 }, { "epoch": 0.7378021625665782, "grad_norm": 1.8382217657564655, "learning_rate": 3.3940253831025514e-06, "loss": 0.5136, "step": 20709 }, { "epoch": 0.7378377896930723, "grad_norm": 1.5516752675569956, "learning_rate": 3.3931591203084834e-06, "loss": 0.5931, "step": 20710 }, { "epoch": 0.7378734168195664, "grad_norm": 1.2219514724428382, "learning_rate": 3.3922929454891173e-06, "loss": 0.404, "step": 20711 }, { "epoch": 0.7379090439460605, "grad_norm": 1.3582509703612669, "learning_rate": 3.391426858655992e-06, "loss": 0.4356, "step": 20712 }, { "epoch": 0.7379446710725547, "grad_norm": 1.794283286835411, "learning_rate": 3.390560859820633e-06, "loss": 0.5471, "step": 20713 }, { "epoch": 0.7379802981990488, "grad_norm": 1.704529218448583, "learning_rate": 3.389694948994576e-06, "loss": 0.5269, "step": 20714 }, { "epoch": 0.7380159253255428, "grad_norm": 1.9557169355348354, "learning_rate": 3.3888291261893513e-06, "loss": 0.432, "step": 20715 }, { "epoch": 0.738051552452037, "grad_norm": 1.487771104035327, "learning_rate": 3.387963391416491e-06, "loss": 0.4715, "step": 20716 }, { "epoch": 0.7380871795785311, "grad_norm": 1.6861747146387267, "learning_rate": 3.3870977446875176e-06, "loss": 0.5455, "step": 20717 }, { "epoch": 0.7381228067050252, "grad_norm": 1.3461135490555163, "learning_rate": 3.3862321860139578e-06, "loss": 0.3994, "step": 20718 }, { "epoch": 0.7381584338315194, "grad_norm": 1.5071017392805013, "learning_rate": 3.3853667154073444e-06, "loss": 0.4003, "step": 20719 }, { "epoch": 0.7381940609580134, "grad_norm": 1.4579759644412646, "learning_rate": 3.384501332879192e-06, "loss": 0.4122, "step": 20720 }, { "epoch": 0.7382296880845075, "grad_norm": 1.1651845830814207, "learning_rate": 3.383636038441028e-06, "loss": 0.4202, "step": 20721 }, { "epoch": 0.7382653152110017, "grad_norm": 1.6304185733293723, "learning_rate": 3.382770832104376e-06, "loss": 0.5515, "step": 20722 }, { "epoch": 0.7383009423374958, "grad_norm": 1.656059890913331, "learning_rate": 3.381905713880759e-06, "loss": 0.5074, "step": 20723 }, { "epoch": 0.7383365694639898, "grad_norm": 1.6728776391981781, "learning_rate": 3.381040683781689e-06, "loss": 0.3879, "step": 20724 }, { "epoch": 0.738372196590484, "grad_norm": 1.4674354657516098, "learning_rate": 3.380175741818692e-06, "loss": 0.4414, "step": 20725 }, { "epoch": 0.7384078237169781, "grad_norm": 1.5921002118894176, "learning_rate": 3.379310888003279e-06, "loss": 0.5377, "step": 20726 }, { "epoch": 0.7384434508434722, "grad_norm": 1.371213877128883, "learning_rate": 3.3784461223469713e-06, "loss": 0.4768, "step": 20727 }, { "epoch": 0.7384790779699664, "grad_norm": 1.7439332485847803, "learning_rate": 3.3775814448612797e-06, "loss": 0.4437, "step": 20728 }, { "epoch": 0.7385147050964604, "grad_norm": 1.407210892573426, "learning_rate": 3.3767168555577158e-06, "loss": 0.5133, "step": 20729 }, { "epoch": 0.7385503322229545, "grad_norm": 1.187564067470263, "learning_rate": 3.3758523544478038e-06, "loss": 0.3953, "step": 20730 }, { "epoch": 0.7385859593494487, "grad_norm": 1.1924643957573737, "learning_rate": 3.374987941543043e-06, "loss": 0.3721, "step": 20731 }, { "epoch": 0.7386215864759428, "grad_norm": 1.202373728295088, "learning_rate": 3.374123616854953e-06, "loss": 0.4064, "step": 20732 }, { "epoch": 0.7386572136024369, "grad_norm": 1.2508470592485033, "learning_rate": 3.3732593803950354e-06, "loss": 0.4589, "step": 20733 }, { "epoch": 0.738692840728931, "grad_norm": 1.9591471695962912, "learning_rate": 3.372395232174803e-06, "loss": 0.4476, "step": 20734 }, { "epoch": 0.7387284678554251, "grad_norm": 1.2485559511482789, "learning_rate": 3.3715311722057577e-06, "loss": 0.4461, "step": 20735 }, { "epoch": 0.7387640949819192, "grad_norm": 1.2858786779889668, "learning_rate": 3.370667200499408e-06, "loss": 0.5259, "step": 20736 }, { "epoch": 0.7387997221084134, "grad_norm": 1.7250523919609662, "learning_rate": 3.3698033170672574e-06, "loss": 0.4943, "step": 20737 }, { "epoch": 0.7388353492349075, "grad_norm": 1.705282549796487, "learning_rate": 3.3689395219208143e-06, "loss": 0.5145, "step": 20738 }, { "epoch": 0.7388709763614015, "grad_norm": 1.8336194679369007, "learning_rate": 3.368075815071572e-06, "loss": 0.5319, "step": 20739 }, { "epoch": 0.7389066034878957, "grad_norm": 1.5606662353934924, "learning_rate": 3.3672121965310358e-06, "loss": 0.4712, "step": 20740 }, { "epoch": 0.7389422306143898, "grad_norm": 1.3522805869762675, "learning_rate": 3.366348666310709e-06, "loss": 0.442, "step": 20741 }, { "epoch": 0.7389778577408839, "grad_norm": 1.338176800149327, "learning_rate": 3.3654852244220824e-06, "loss": 0.4299, "step": 20742 }, { "epoch": 0.7390134848673781, "grad_norm": 1.8016648498906864, "learning_rate": 3.3646218708766586e-06, "loss": 0.4154, "step": 20743 }, { "epoch": 0.7390491119938721, "grad_norm": 1.3266053065174261, "learning_rate": 3.363758605685932e-06, "loss": 0.402, "step": 20744 }, { "epoch": 0.7390847391203662, "grad_norm": 1.7074731643287524, "learning_rate": 3.3628954288614003e-06, "loss": 0.5284, "step": 20745 }, { "epoch": 0.7391203662468604, "grad_norm": 1.3870866307729823, "learning_rate": 3.362032340414553e-06, "loss": 0.4165, "step": 20746 }, { "epoch": 0.7391559933733545, "grad_norm": 1.5889072459170457, "learning_rate": 3.3611693403568866e-06, "loss": 0.4498, "step": 20747 }, { "epoch": 0.7391916204998485, "grad_norm": 1.6614408674795016, "learning_rate": 3.3603064286998888e-06, "loss": 0.5105, "step": 20748 }, { "epoch": 0.7392272476263427, "grad_norm": 5.166423115294193, "learning_rate": 3.3594436054550537e-06, "loss": 0.4105, "step": 20749 }, { "epoch": 0.7392628747528368, "grad_norm": 1.4310797152392283, "learning_rate": 3.3585808706338663e-06, "loss": 0.3938, "step": 20750 }, { "epoch": 0.7392985018793309, "grad_norm": 1.3482831666161186, "learning_rate": 3.357718224247817e-06, "loss": 0.4136, "step": 20751 }, { "epoch": 0.7393341290058251, "grad_norm": 1.305307952520123, "learning_rate": 3.3568556663083906e-06, "loss": 0.4183, "step": 20752 }, { "epoch": 0.7393697561323191, "grad_norm": 1.3309718069406167, "learning_rate": 3.355993196827075e-06, "loss": 0.4367, "step": 20753 }, { "epoch": 0.7394053832588132, "grad_norm": 1.332182188283679, "learning_rate": 3.3551308158153585e-06, "loss": 0.4425, "step": 20754 }, { "epoch": 0.7394410103853074, "grad_norm": 1.3224260860887151, "learning_rate": 3.354268523284715e-06, "loss": 0.4517, "step": 20755 }, { "epoch": 0.7394766375118015, "grad_norm": 1.3802010759481675, "learning_rate": 3.353406319246635e-06, "loss": 0.4304, "step": 20756 }, { "epoch": 0.7395122646382956, "grad_norm": 1.5182682333748645, "learning_rate": 3.352544203712591e-06, "loss": 0.4102, "step": 20757 }, { "epoch": 0.7395478917647897, "grad_norm": 1.3716118097549013, "learning_rate": 3.3516821766940687e-06, "loss": 0.3933, "step": 20758 }, { "epoch": 0.7395835188912838, "grad_norm": 1.6828433204094344, "learning_rate": 3.3508202382025447e-06, "loss": 0.548, "step": 20759 }, { "epoch": 0.7396191460177779, "grad_norm": 1.4672565846352235, "learning_rate": 3.3499583882495003e-06, "loss": 0.4671, "step": 20760 }, { "epoch": 0.7396547731442721, "grad_norm": 1.6480492139748828, "learning_rate": 3.349096626846404e-06, "loss": 0.5374, "step": 20761 }, { "epoch": 0.7396904002707662, "grad_norm": 1.1598925032412264, "learning_rate": 3.3482349540047387e-06, "loss": 0.4311, "step": 20762 }, { "epoch": 0.7397260273972602, "grad_norm": 1.553995456238394, "learning_rate": 3.347373369735971e-06, "loss": 0.5173, "step": 20763 }, { "epoch": 0.7397616545237544, "grad_norm": 1.214536587601477, "learning_rate": 3.346511874051577e-06, "loss": 0.3556, "step": 20764 }, { "epoch": 0.7397972816502485, "grad_norm": 1.6365498868880155, "learning_rate": 3.345650466963031e-06, "loss": 0.5416, "step": 20765 }, { "epoch": 0.7398329087767426, "grad_norm": 1.6034144360436267, "learning_rate": 3.3447891484817963e-06, "loss": 0.5527, "step": 20766 }, { "epoch": 0.7398685359032368, "grad_norm": 1.2977607085439253, "learning_rate": 3.3439279186193464e-06, "loss": 0.3834, "step": 20767 }, { "epoch": 0.7399041630297308, "grad_norm": 1.2564821744254695, "learning_rate": 3.343066777387148e-06, "loss": 0.4308, "step": 20768 }, { "epoch": 0.7399397901562249, "grad_norm": 1.4139443017572675, "learning_rate": 3.342205724796672e-06, "loss": 0.4154, "step": 20769 }, { "epoch": 0.7399754172827191, "grad_norm": 1.1176931161914172, "learning_rate": 3.3413447608593774e-06, "loss": 0.3898, "step": 20770 }, { "epoch": 0.7400110444092132, "grad_norm": 1.5579691993473332, "learning_rate": 3.3404838855867337e-06, "loss": 0.4166, "step": 20771 }, { "epoch": 0.7400466715357072, "grad_norm": 1.7820647052943224, "learning_rate": 3.3396230989901988e-06, "loss": 0.5419, "step": 20772 }, { "epoch": 0.7400822986622014, "grad_norm": 1.36114883353138, "learning_rate": 3.338762401081238e-06, "loss": 0.4418, "step": 20773 }, { "epoch": 0.7401179257886955, "grad_norm": 1.7694536844355546, "learning_rate": 3.337901791871312e-06, "loss": 0.5458, "step": 20774 }, { "epoch": 0.7401535529151896, "grad_norm": 1.3123817990649542, "learning_rate": 3.3370412713718804e-06, "loss": 0.4317, "step": 20775 }, { "epoch": 0.7401891800416838, "grad_norm": 1.3689838500082516, "learning_rate": 3.336180839594404e-06, "loss": 0.4682, "step": 20776 }, { "epoch": 0.7402248071681778, "grad_norm": 1.4706200126418159, "learning_rate": 3.335320496550335e-06, "loss": 0.46, "step": 20777 }, { "epoch": 0.7402604342946719, "grad_norm": 1.3630068898115477, "learning_rate": 3.3344602422511343e-06, "loss": 0.3996, "step": 20778 }, { "epoch": 0.7402960614211661, "grad_norm": 1.268552048919342, "learning_rate": 3.333600076708251e-06, "loss": 0.379, "step": 20779 }, { "epoch": 0.7403316885476602, "grad_norm": 1.5893002182237765, "learning_rate": 3.3327399999331434e-06, "loss": 0.526, "step": 20780 }, { "epoch": 0.7403673156741543, "grad_norm": 1.4006393578174003, "learning_rate": 3.3318800119372617e-06, "loss": 0.4909, "step": 20781 }, { "epoch": 0.7404029428006484, "grad_norm": 1.5125320075465563, "learning_rate": 3.331020112732062e-06, "loss": 0.3896, "step": 20782 }, { "epoch": 0.7404385699271425, "grad_norm": 1.8238265566125664, "learning_rate": 3.3301603023289873e-06, "loss": 0.4446, "step": 20783 }, { "epoch": 0.7404741970536366, "grad_norm": 1.476465589500655, "learning_rate": 3.329300580739494e-06, "loss": 0.4436, "step": 20784 }, { "epoch": 0.7405098241801308, "grad_norm": 1.1280193202071593, "learning_rate": 3.3284409479750223e-06, "loss": 0.4553, "step": 20785 }, { "epoch": 0.7405454513066249, "grad_norm": 1.93244051045365, "learning_rate": 3.3275814040470235e-06, "loss": 0.5503, "step": 20786 }, { "epoch": 0.7405810784331189, "grad_norm": 1.4995084453181862, "learning_rate": 3.3267219489669434e-06, "loss": 0.5378, "step": 20787 }, { "epoch": 0.7406167055596131, "grad_norm": 1.39866496735369, "learning_rate": 3.325862582746223e-06, "loss": 0.4474, "step": 20788 }, { "epoch": 0.7406523326861072, "grad_norm": 1.3936509561750294, "learning_rate": 3.325003305396307e-06, "loss": 0.4603, "step": 20789 }, { "epoch": 0.7406879598126013, "grad_norm": 1.6924781758979222, "learning_rate": 3.324144116928637e-06, "loss": 0.4193, "step": 20790 }, { "epoch": 0.7407235869390955, "grad_norm": 1.2015627950546695, "learning_rate": 3.3232850173546584e-06, "loss": 0.3841, "step": 20791 }, { "epoch": 0.7407592140655895, "grad_norm": 1.5155163340822722, "learning_rate": 3.3224260066858026e-06, "loss": 0.453, "step": 20792 }, { "epoch": 0.7407948411920836, "grad_norm": 1.4129547806002327, "learning_rate": 3.3215670849335156e-06, "loss": 0.4496, "step": 20793 }, { "epoch": 0.7408304683185778, "grad_norm": 1.6732425441982766, "learning_rate": 3.320708252109227e-06, "loss": 0.513, "step": 20794 }, { "epoch": 0.7408660954450719, "grad_norm": 2.034599200452422, "learning_rate": 3.3198495082243777e-06, "loss": 0.5115, "step": 20795 }, { "epoch": 0.740901722571566, "grad_norm": 12.700683247152316, "learning_rate": 3.3189908532904002e-06, "loss": 0.5503, "step": 20796 }, { "epoch": 0.7409373496980601, "grad_norm": 1.3080151728343932, "learning_rate": 3.3181322873187325e-06, "loss": 0.4398, "step": 20797 }, { "epoch": 0.7409729768245542, "grad_norm": 1.3091460054672184, "learning_rate": 3.3172738103208015e-06, "loss": 0.3887, "step": 20798 }, { "epoch": 0.7410086039510483, "grad_norm": 2.2538175212234233, "learning_rate": 3.316415422308039e-06, "loss": 0.5407, "step": 20799 }, { "epoch": 0.7410442310775425, "grad_norm": 1.6871745757454892, "learning_rate": 3.3155571232918817e-06, "loss": 0.4472, "step": 20800 }, { "epoch": 0.7410798582040365, "grad_norm": 1.302102524096281, "learning_rate": 3.3146989132837492e-06, "loss": 0.3735, "step": 20801 }, { "epoch": 0.7411154853305306, "grad_norm": 1.568865752854228, "learning_rate": 3.3138407922950767e-06, "loss": 0.3893, "step": 20802 }, { "epoch": 0.7411511124570248, "grad_norm": 1.4500149889706824, "learning_rate": 3.3129827603372843e-06, "loss": 0.4233, "step": 20803 }, { "epoch": 0.7411867395835189, "grad_norm": 1.3252404752808626, "learning_rate": 3.3121248174218e-06, "loss": 0.3919, "step": 20804 }, { "epoch": 0.741222366710013, "grad_norm": 3.5653660215327143, "learning_rate": 3.3112669635600494e-06, "loss": 0.6506, "step": 20805 }, { "epoch": 0.7412579938365071, "grad_norm": 1.7053244064738196, "learning_rate": 3.310409198763457e-06, "loss": 0.5335, "step": 20806 }, { "epoch": 0.7412936209630012, "grad_norm": 1.582145196795056, "learning_rate": 3.3095515230434384e-06, "loss": 0.4103, "step": 20807 }, { "epoch": 0.7413292480894953, "grad_norm": 1.4860461326973573, "learning_rate": 3.308693936411421e-06, "loss": 0.3749, "step": 20808 }, { "epoch": 0.7413648752159895, "grad_norm": 2.3864339312384617, "learning_rate": 3.3078364388788176e-06, "loss": 0.5481, "step": 20809 }, { "epoch": 0.7414005023424836, "grad_norm": 1.4737691617944952, "learning_rate": 3.3069790304570495e-06, "loss": 0.4805, "step": 20810 }, { "epoch": 0.7414361294689776, "grad_norm": 1.6023967091626363, "learning_rate": 3.306121711157534e-06, "loss": 0.5243, "step": 20811 }, { "epoch": 0.7414717565954718, "grad_norm": 1.6304493985981645, "learning_rate": 3.3052644809916857e-06, "loss": 0.6174, "step": 20812 }, { "epoch": 0.7415073837219659, "grad_norm": 1.408428825611207, "learning_rate": 3.3044073399709243e-06, "loss": 0.3811, "step": 20813 }, { "epoch": 0.74154301084846, "grad_norm": 1.5210737517435644, "learning_rate": 3.303550288106656e-06, "loss": 0.4218, "step": 20814 }, { "epoch": 0.7415786379749542, "grad_norm": 1.5207320718611186, "learning_rate": 3.302693325410299e-06, "loss": 0.4376, "step": 20815 }, { "epoch": 0.7416142651014482, "grad_norm": 1.88393649065839, "learning_rate": 3.3018364518932577e-06, "loss": 0.4785, "step": 20816 }, { "epoch": 0.7416498922279423, "grad_norm": 1.4472232159579403, "learning_rate": 3.3009796675669504e-06, "loss": 0.4478, "step": 20817 }, { "epoch": 0.7416855193544365, "grad_norm": 2.2074303233904526, "learning_rate": 3.300122972442773e-06, "loss": 0.4073, "step": 20818 }, { "epoch": 0.7417211464809306, "grad_norm": 1.590866314214251, "learning_rate": 3.299266366532149e-06, "loss": 0.5073, "step": 20819 }, { "epoch": 0.7417567736074246, "grad_norm": 2.065595922531475, "learning_rate": 3.2984098498464736e-06, "loss": 0.4129, "step": 20820 }, { "epoch": 0.7417924007339188, "grad_norm": 1.4882820469951177, "learning_rate": 3.297553422397154e-06, "loss": 0.4521, "step": 20821 }, { "epoch": 0.7418280278604129, "grad_norm": 1.3753749557277362, "learning_rate": 3.2966970841956004e-06, "loss": 0.4375, "step": 20822 }, { "epoch": 0.741863654986907, "grad_norm": 1.6772595259575531, "learning_rate": 3.2958408352532055e-06, "loss": 0.4934, "step": 20823 }, { "epoch": 0.7418992821134012, "grad_norm": 1.84282522267872, "learning_rate": 3.2949846755813795e-06, "loss": 0.565, "step": 20824 }, { "epoch": 0.7419349092398952, "grad_norm": 1.9013720766995283, "learning_rate": 3.2941286051915165e-06, "loss": 0.406, "step": 20825 }, { "epoch": 0.7419705363663893, "grad_norm": 1.584689368047494, "learning_rate": 3.2932726240950173e-06, "loss": 0.4283, "step": 20826 }, { "epoch": 0.7420061634928835, "grad_norm": 1.4793108559974997, "learning_rate": 3.292416732303282e-06, "loss": 0.3885, "step": 20827 }, { "epoch": 0.7420417906193776, "grad_norm": 1.6417404356726448, "learning_rate": 3.2915609298277083e-06, "loss": 0.5076, "step": 20828 }, { "epoch": 0.7420774177458717, "grad_norm": 1.787429722494506, "learning_rate": 3.290705216679687e-06, "loss": 0.5178, "step": 20829 }, { "epoch": 0.7421130448723658, "grad_norm": 1.4735895150552873, "learning_rate": 3.2898495928706187e-06, "loss": 0.4037, "step": 20830 }, { "epoch": 0.7421486719988599, "grad_norm": 2.0146030399605923, "learning_rate": 3.2889940584118897e-06, "loss": 0.5975, "step": 20831 }, { "epoch": 0.742184299125354, "grad_norm": 1.4203228105279835, "learning_rate": 3.288138613314895e-06, "loss": 0.4381, "step": 20832 }, { "epoch": 0.7422199262518482, "grad_norm": 1.4931993786012394, "learning_rate": 3.287283257591026e-06, "loss": 0.4527, "step": 20833 }, { "epoch": 0.7422555533783423, "grad_norm": 1.8035429478915534, "learning_rate": 3.286427991251674e-06, "loss": 0.5799, "step": 20834 }, { "epoch": 0.7422911805048363, "grad_norm": 1.6616441112578595, "learning_rate": 3.285572814308228e-06, "loss": 0.4243, "step": 20835 }, { "epoch": 0.7423268076313305, "grad_norm": 1.4119367074564735, "learning_rate": 3.2847177267720686e-06, "loss": 0.3959, "step": 20836 }, { "epoch": 0.7423624347578246, "grad_norm": 1.5328405439440327, "learning_rate": 3.283862728654591e-06, "loss": 0.5113, "step": 20837 }, { "epoch": 0.7423980618843187, "grad_norm": 1.543958246493, "learning_rate": 3.283007819967172e-06, "loss": 0.5357, "step": 20838 }, { "epoch": 0.7424336890108129, "grad_norm": 1.5734881564889165, "learning_rate": 3.2821530007212023e-06, "loss": 0.4715, "step": 20839 }, { "epoch": 0.7424693161373069, "grad_norm": 2.818281922863503, "learning_rate": 3.2812982709280573e-06, "loss": 0.3959, "step": 20840 }, { "epoch": 0.742504943263801, "grad_norm": 1.349508032050058, "learning_rate": 3.2804436305991217e-06, "loss": 0.4067, "step": 20841 }, { "epoch": 0.7425405703902952, "grad_norm": 1.6468865706125166, "learning_rate": 3.2795890797457763e-06, "loss": 0.5585, "step": 20842 }, { "epoch": 0.7425761975167893, "grad_norm": 1.7899243379028367, "learning_rate": 3.278734618379402e-06, "loss": 0.4873, "step": 20843 }, { "epoch": 0.7426118246432833, "grad_norm": 1.5064786543867064, "learning_rate": 3.2778802465113725e-06, "loss": 0.3967, "step": 20844 }, { "epoch": 0.7426474517697775, "grad_norm": 2.0413829842438433, "learning_rate": 3.2770259641530642e-06, "loss": 0.487, "step": 20845 }, { "epoch": 0.7426830788962716, "grad_norm": 1.896436912182407, "learning_rate": 3.276171771315858e-06, "loss": 0.5668, "step": 20846 }, { "epoch": 0.7427187060227657, "grad_norm": 1.4611786717595068, "learning_rate": 3.275317668011122e-06, "loss": 0.4461, "step": 20847 }, { "epoch": 0.7427543331492599, "grad_norm": 1.5694027662251608, "learning_rate": 3.2744636542502307e-06, "loss": 0.5435, "step": 20848 }, { "epoch": 0.742789960275754, "grad_norm": 1.7625595259083102, "learning_rate": 3.2736097300445567e-06, "loss": 0.5964, "step": 20849 }, { "epoch": 0.742825587402248, "grad_norm": 1.5593785427042417, "learning_rate": 3.272755895405474e-06, "loss": 0.562, "step": 20850 }, { "epoch": 0.7428612145287422, "grad_norm": 1.9873055806524575, "learning_rate": 3.271902150344346e-06, "loss": 0.5042, "step": 20851 }, { "epoch": 0.7428968416552363, "grad_norm": 2.254272555093561, "learning_rate": 3.2710484948725465e-06, "loss": 0.4898, "step": 20852 }, { "epoch": 0.7429324687817304, "grad_norm": 1.610733173304067, "learning_rate": 3.2701949290014355e-06, "loss": 0.4785, "step": 20853 }, { "epoch": 0.7429680959082245, "grad_norm": 1.3690491577990516, "learning_rate": 3.2693414527423873e-06, "loss": 0.4505, "step": 20854 }, { "epoch": 0.7430037230347186, "grad_norm": 1.2373246926470467, "learning_rate": 3.2684880661067573e-06, "loss": 0.3898, "step": 20855 }, { "epoch": 0.7430393501612127, "grad_norm": 1.4424237962023216, "learning_rate": 3.267634769105912e-06, "loss": 0.4888, "step": 20856 }, { "epoch": 0.7430749772877069, "grad_norm": 1.4104244037075824, "learning_rate": 3.266781561751221e-06, "loss": 0.4204, "step": 20857 }, { "epoch": 0.743110604414201, "grad_norm": 1.426185091419472, "learning_rate": 3.2659284440540375e-06, "loss": 0.5406, "step": 20858 }, { "epoch": 0.743146231540695, "grad_norm": 1.2651140058688173, "learning_rate": 3.2650754160257257e-06, "loss": 0.4533, "step": 20859 }, { "epoch": 0.7431818586671892, "grad_norm": 1.2217177805557904, "learning_rate": 3.2642224776776388e-06, "loss": 0.3797, "step": 20860 }, { "epoch": 0.7432174857936833, "grad_norm": 1.456725810292675, "learning_rate": 3.2633696290211413e-06, "loss": 0.3946, "step": 20861 }, { "epoch": 0.7432531129201774, "grad_norm": 1.9378246953597027, "learning_rate": 3.2625168700675836e-06, "loss": 0.5816, "step": 20862 }, { "epoch": 0.7432887400466716, "grad_norm": 1.2908391760437765, "learning_rate": 3.2616642008283218e-06, "loss": 0.4255, "step": 20863 }, { "epoch": 0.7433243671731656, "grad_norm": 1.2462933894346362, "learning_rate": 3.260811621314711e-06, "loss": 0.4193, "step": 20864 }, { "epoch": 0.7433599942996597, "grad_norm": 1.1904249851918574, "learning_rate": 3.2599591315381075e-06, "loss": 0.405, "step": 20865 }, { "epoch": 0.7433956214261539, "grad_norm": 1.3759144150816043, "learning_rate": 3.2591067315098556e-06, "loss": 0.4241, "step": 20866 }, { "epoch": 0.743431248552648, "grad_norm": 1.28967886991159, "learning_rate": 3.2582544212413103e-06, "loss": 0.4967, "step": 20867 }, { "epoch": 0.743466875679142, "grad_norm": 1.4786506050375774, "learning_rate": 3.257402200743821e-06, "loss": 0.5123, "step": 20868 }, { "epoch": 0.7435025028056362, "grad_norm": 2.0707816192542086, "learning_rate": 3.2565500700287324e-06, "loss": 0.5236, "step": 20869 }, { "epoch": 0.7435381299321303, "grad_norm": 1.7053332507756025, "learning_rate": 3.2556980291073927e-06, "loss": 0.5166, "step": 20870 }, { "epoch": 0.7435737570586244, "grad_norm": 1.205716566072611, "learning_rate": 3.2548460779911475e-06, "loss": 0.3921, "step": 20871 }, { "epoch": 0.7436093841851186, "grad_norm": 1.6282534092036922, "learning_rate": 3.253994216691344e-06, "loss": 0.5657, "step": 20872 }, { "epoch": 0.7436450113116126, "grad_norm": 1.3823531225297914, "learning_rate": 3.2531424452193205e-06, "loss": 0.5134, "step": 20873 }, { "epoch": 0.7436806384381067, "grad_norm": 1.5151438864653783, "learning_rate": 3.2522907635864244e-06, "loss": 0.4205, "step": 20874 }, { "epoch": 0.7437162655646009, "grad_norm": 1.2475421668932394, "learning_rate": 3.2514391718039882e-06, "loss": 0.4277, "step": 20875 }, { "epoch": 0.743751892691095, "grad_norm": 1.5992033193576352, "learning_rate": 3.2505876698833617e-06, "loss": 0.5117, "step": 20876 }, { "epoch": 0.7437875198175891, "grad_norm": 1.7315173444215264, "learning_rate": 3.249736257835874e-06, "loss": 0.5192, "step": 20877 }, { "epoch": 0.7438231469440832, "grad_norm": 1.6812440275946139, "learning_rate": 3.2488849356728656e-06, "loss": 0.5425, "step": 20878 }, { "epoch": 0.7438587740705773, "grad_norm": 2.1383063255137498, "learning_rate": 3.248033703405673e-06, "loss": 0.4969, "step": 20879 }, { "epoch": 0.7438944011970714, "grad_norm": 1.6246463849936221, "learning_rate": 3.247182561045632e-06, "loss": 0.5231, "step": 20880 }, { "epoch": 0.7439300283235656, "grad_norm": 1.4192166640624104, "learning_rate": 3.246331508604078e-06, "loss": 0.468, "step": 20881 }, { "epoch": 0.7439656554500597, "grad_norm": 1.423076316604947, "learning_rate": 3.245480546092337e-06, "loss": 0.4205, "step": 20882 }, { "epoch": 0.7440012825765537, "grad_norm": 1.4102567357862315, "learning_rate": 3.2446296735217462e-06, "loss": 0.4516, "step": 20883 }, { "epoch": 0.7440369097030479, "grad_norm": 1.2470541870648355, "learning_rate": 3.2437788909036317e-06, "loss": 0.4128, "step": 20884 }, { "epoch": 0.744072536829542, "grad_norm": 1.4842284279798457, "learning_rate": 3.242928198249322e-06, "loss": 0.5304, "step": 20885 }, { "epoch": 0.7441081639560361, "grad_norm": 1.1950688443769946, "learning_rate": 3.2420775955701468e-06, "loss": 0.4439, "step": 20886 }, { "epoch": 0.7441437910825303, "grad_norm": 1.5597623449839304, "learning_rate": 3.241227082877435e-06, "loss": 0.4684, "step": 20887 }, { "epoch": 0.7441794182090243, "grad_norm": 1.4453019401744671, "learning_rate": 3.2403766601825046e-06, "loss": 0.462, "step": 20888 }, { "epoch": 0.7442150453355184, "grad_norm": 4.041790425350629, "learning_rate": 3.2395263274966882e-06, "loss": 0.4232, "step": 20889 }, { "epoch": 0.7442506724620126, "grad_norm": 1.494186422490136, "learning_rate": 3.2386760848313002e-06, "loss": 0.5302, "step": 20890 }, { "epoch": 0.7442862995885067, "grad_norm": 1.6218823763403467, "learning_rate": 3.2378259321976668e-06, "loss": 0.493, "step": 20891 }, { "epoch": 0.7443219267150007, "grad_norm": 1.461861810044047, "learning_rate": 3.2369758696071097e-06, "loss": 0.511, "step": 20892 }, { "epoch": 0.7443575538414949, "grad_norm": 1.224510976017136, "learning_rate": 3.23612589707094e-06, "loss": 0.4684, "step": 20893 }, { "epoch": 0.744393180967989, "grad_norm": 1.25666138824429, "learning_rate": 3.235276014600488e-06, "loss": 0.4054, "step": 20894 }, { "epoch": 0.7444288080944831, "grad_norm": 1.54715774842169, "learning_rate": 3.234426222207061e-06, "loss": 0.5328, "step": 20895 }, { "epoch": 0.7444644352209773, "grad_norm": 1.6403568217324722, "learning_rate": 3.2335765199019807e-06, "loss": 0.4266, "step": 20896 }, { "epoch": 0.7445000623474713, "grad_norm": 1.586151132033364, "learning_rate": 3.232726907696556e-06, "loss": 0.4755, "step": 20897 }, { "epoch": 0.7445356894739654, "grad_norm": 1.1718523077138547, "learning_rate": 3.2318773856021057e-06, "loss": 0.4, "step": 20898 }, { "epoch": 0.7445713166004596, "grad_norm": 1.7251154385919028, "learning_rate": 3.2310279536299347e-06, "loss": 0.4086, "step": 20899 }, { "epoch": 0.7446069437269537, "grad_norm": 2.3114162361149946, "learning_rate": 3.230178611791358e-06, "loss": 0.5598, "step": 20900 }, { "epoch": 0.7446425708534478, "grad_norm": 1.9004283285385433, "learning_rate": 3.229329360097685e-06, "loss": 0.5419, "step": 20901 }, { "epoch": 0.744678197979942, "grad_norm": 1.2285402522816025, "learning_rate": 3.2284801985602242e-06, "loss": 0.4056, "step": 20902 }, { "epoch": 0.744713825106436, "grad_norm": 1.1894677894807053, "learning_rate": 3.227631127190285e-06, "loss": 0.4052, "step": 20903 }, { "epoch": 0.7447494522329301, "grad_norm": 2.2256692748414095, "learning_rate": 3.2267821459991686e-06, "loss": 0.4401, "step": 20904 }, { "epoch": 0.7447850793594243, "grad_norm": 1.6925560467030714, "learning_rate": 3.2259332549981847e-06, "loss": 0.5934, "step": 20905 }, { "epoch": 0.7448207064859184, "grad_norm": 1.1540475412144764, "learning_rate": 3.2250844541986313e-06, "loss": 0.4042, "step": 20906 }, { "epoch": 0.7448563336124124, "grad_norm": 1.6337728012626778, "learning_rate": 3.224235743611814e-06, "loss": 0.4821, "step": 20907 }, { "epoch": 0.7448919607389066, "grad_norm": 1.4835943232463047, "learning_rate": 3.2233871232490344e-06, "loss": 0.5271, "step": 20908 }, { "epoch": 0.7449275878654007, "grad_norm": 1.640989439525396, "learning_rate": 3.2225385931215935e-06, "loss": 0.5538, "step": 20909 }, { "epoch": 0.7449632149918948, "grad_norm": 1.1382995681564791, "learning_rate": 3.2216901532407862e-06, "loss": 0.3987, "step": 20910 }, { "epoch": 0.744998842118389, "grad_norm": 1.102625980381061, "learning_rate": 3.2208418036179155e-06, "loss": 0.3924, "step": 20911 }, { "epoch": 0.745034469244883, "grad_norm": 1.8654275754882, "learning_rate": 3.2199935442642704e-06, "loss": 0.5118, "step": 20912 }, { "epoch": 0.7450700963713772, "grad_norm": 1.3088259342128044, "learning_rate": 3.2191453751911505e-06, "loss": 0.4845, "step": 20913 }, { "epoch": 0.7451057234978713, "grad_norm": 1.2208592208186329, "learning_rate": 3.2182972964098535e-06, "loss": 0.4276, "step": 20914 }, { "epoch": 0.7451413506243654, "grad_norm": 1.2505599105573393, "learning_rate": 3.217449307931665e-06, "loss": 0.4366, "step": 20915 }, { "epoch": 0.7451769777508596, "grad_norm": 1.1970617458141932, "learning_rate": 3.2166014097678787e-06, "loss": 0.4621, "step": 20916 }, { "epoch": 0.7452126048773536, "grad_norm": 1.3214083548775752, "learning_rate": 3.2157536019297863e-06, "loss": 0.4697, "step": 20917 }, { "epoch": 0.7452482320038477, "grad_norm": 1.2427581213585517, "learning_rate": 3.2149058844286796e-06, "loss": 0.4351, "step": 20918 }, { "epoch": 0.7452838591303419, "grad_norm": 1.2150585565872822, "learning_rate": 3.2140582572758405e-06, "loss": 0.4386, "step": 20919 }, { "epoch": 0.745319486256836, "grad_norm": 1.4000628973857527, "learning_rate": 3.2132107204825626e-06, "loss": 0.5617, "step": 20920 }, { "epoch": 0.74535511338333, "grad_norm": 1.2142075398175416, "learning_rate": 3.212363274060124e-06, "loss": 0.3881, "step": 20921 }, { "epoch": 0.7453907405098242, "grad_norm": 1.4787206722411772, "learning_rate": 3.2115159180198117e-06, "loss": 0.473, "step": 20922 }, { "epoch": 0.7454263676363183, "grad_norm": 1.4189891242002044, "learning_rate": 3.2106686523729116e-06, "loss": 0.5792, "step": 20923 }, { "epoch": 0.7454619947628124, "grad_norm": 1.1889261457088782, "learning_rate": 3.2098214771307057e-06, "loss": 0.3894, "step": 20924 }, { "epoch": 0.7454976218893066, "grad_norm": 1.326206625070086, "learning_rate": 3.20897439230447e-06, "loss": 0.4887, "step": 20925 }, { "epoch": 0.7455332490158006, "grad_norm": 1.5323892274434143, "learning_rate": 3.208127397905487e-06, "loss": 0.5444, "step": 20926 }, { "epoch": 0.7455688761422947, "grad_norm": 1.2182864759455203, "learning_rate": 3.2072804939450385e-06, "loss": 0.4107, "step": 20927 }, { "epoch": 0.7456045032687889, "grad_norm": 1.3916206301505127, "learning_rate": 3.206433680434394e-06, "loss": 0.442, "step": 20928 }, { "epoch": 0.745640130395283, "grad_norm": 1.5377991959340243, "learning_rate": 3.2055869573848374e-06, "loss": 0.5116, "step": 20929 }, { "epoch": 0.7456757575217771, "grad_norm": 1.490627995784205, "learning_rate": 3.204740324807636e-06, "loss": 0.3855, "step": 20930 }, { "epoch": 0.7457113846482712, "grad_norm": 1.3560323699961145, "learning_rate": 3.203893782714067e-06, "loss": 0.5197, "step": 20931 }, { "epoch": 0.7457470117747653, "grad_norm": 1.6102478605606598, "learning_rate": 3.203047331115402e-06, "loss": 0.602, "step": 20932 }, { "epoch": 0.7457826389012594, "grad_norm": 1.3406219325383968, "learning_rate": 3.202200970022916e-06, "loss": 0.4709, "step": 20933 }, { "epoch": 0.7458182660277536, "grad_norm": 1.3577476931301307, "learning_rate": 3.201354699447873e-06, "loss": 0.3893, "step": 20934 }, { "epoch": 0.7458538931542477, "grad_norm": 1.3273466043487423, "learning_rate": 3.2005085194015463e-06, "loss": 0.4353, "step": 20935 }, { "epoch": 0.7458895202807417, "grad_norm": 1.2105012009775915, "learning_rate": 3.1996624298951982e-06, "loss": 0.4034, "step": 20936 }, { "epoch": 0.7459251474072359, "grad_norm": 1.2586274572051477, "learning_rate": 3.1988164309400993e-06, "loss": 0.4098, "step": 20937 }, { "epoch": 0.74596077453373, "grad_norm": 1.1848814178047689, "learning_rate": 3.197970522547512e-06, "loss": 0.4484, "step": 20938 }, { "epoch": 0.7459964016602241, "grad_norm": 1.6196367947194024, "learning_rate": 3.1971247047287013e-06, "loss": 0.5429, "step": 20939 }, { "epoch": 0.7460320287867183, "grad_norm": 1.4725287154065507, "learning_rate": 3.196278977494934e-06, "loss": 0.5335, "step": 20940 }, { "epoch": 0.7460676559132123, "grad_norm": 1.30435400383562, "learning_rate": 3.1954333408574643e-06, "loss": 0.4425, "step": 20941 }, { "epoch": 0.7461032830397064, "grad_norm": 1.5588374088999344, "learning_rate": 3.1945877948275595e-06, "loss": 0.4564, "step": 20942 }, { "epoch": 0.7461389101662006, "grad_norm": 1.8145284207115162, "learning_rate": 3.19374233941647e-06, "loss": 0.5067, "step": 20943 }, { "epoch": 0.7461745372926947, "grad_norm": 1.2705258457481279, "learning_rate": 3.19289697463546e-06, "loss": 0.4306, "step": 20944 }, { "epoch": 0.7462101644191887, "grad_norm": 1.8012519655733195, "learning_rate": 3.192051700495784e-06, "loss": 0.494, "step": 20945 }, { "epoch": 0.7462457915456829, "grad_norm": 1.260845088406188, "learning_rate": 3.1912065170087027e-06, "loss": 0.3923, "step": 20946 }, { "epoch": 0.746281418672177, "grad_norm": 1.7582907479269867, "learning_rate": 3.1903614241854606e-06, "loss": 0.5139, "step": 20947 }, { "epoch": 0.7463170457986711, "grad_norm": 1.5418698651443299, "learning_rate": 3.1895164220373167e-06, "loss": 0.4299, "step": 20948 }, { "epoch": 0.7463526729251653, "grad_norm": 2.026757632531608, "learning_rate": 3.188671510575525e-06, "loss": 0.5421, "step": 20949 }, { "epoch": 0.7463883000516593, "grad_norm": 1.2120099608980477, "learning_rate": 3.1878266898113296e-06, "loss": 0.4081, "step": 20950 }, { "epoch": 0.7464239271781534, "grad_norm": 1.7514315688158846, "learning_rate": 3.186981959755987e-06, "loss": 0.4851, "step": 20951 }, { "epoch": 0.7464595543046476, "grad_norm": 1.2348540920907727, "learning_rate": 3.186137320420738e-06, "loss": 0.4427, "step": 20952 }, { "epoch": 0.7464951814311417, "grad_norm": 1.8232542134742884, "learning_rate": 3.1852927718168324e-06, "loss": 0.5645, "step": 20953 }, { "epoch": 0.7465308085576358, "grad_norm": 1.5465290484019887, "learning_rate": 3.184448313955518e-06, "loss": 0.4647, "step": 20954 }, { "epoch": 0.74656643568413, "grad_norm": 1.5192863192543813, "learning_rate": 3.1836039468480405e-06, "loss": 0.4114, "step": 20955 }, { "epoch": 0.746602062810624, "grad_norm": 2.269733937269131, "learning_rate": 3.1827596705056383e-06, "loss": 0.4379, "step": 20956 }, { "epoch": 0.7466376899371181, "grad_norm": 1.4771578797873308, "learning_rate": 3.1819154849395597e-06, "loss": 0.4176, "step": 20957 }, { "epoch": 0.7466733170636123, "grad_norm": 1.3685095479851654, "learning_rate": 3.1810713901610367e-06, "loss": 0.4109, "step": 20958 }, { "epoch": 0.7467089441901064, "grad_norm": 1.3786105715004138, "learning_rate": 3.180227386181316e-06, "loss": 0.4452, "step": 20959 }, { "epoch": 0.7467445713166004, "grad_norm": 1.5970170703672766, "learning_rate": 3.1793834730116337e-06, "loss": 0.4735, "step": 20960 }, { "epoch": 0.7467801984430946, "grad_norm": 5.445209758864433, "learning_rate": 3.1785396506632283e-06, "loss": 0.4292, "step": 20961 }, { "epoch": 0.7468158255695887, "grad_norm": 1.3084426314686657, "learning_rate": 3.177695919147339e-06, "loss": 0.4656, "step": 20962 }, { "epoch": 0.7468514526960828, "grad_norm": 1.3601553327437264, "learning_rate": 3.176852278475192e-06, "loss": 0.4457, "step": 20963 }, { "epoch": 0.746887079822577, "grad_norm": 2.0347146457607965, "learning_rate": 3.176008728658031e-06, "loss": 0.5561, "step": 20964 }, { "epoch": 0.746922706949071, "grad_norm": 1.2840679228348308, "learning_rate": 3.1751652697070788e-06, "loss": 0.4444, "step": 20965 }, { "epoch": 0.7469583340755651, "grad_norm": 1.6530227997973743, "learning_rate": 3.174321901633576e-06, "loss": 0.5986, "step": 20966 }, { "epoch": 0.7469939612020593, "grad_norm": 1.39835504048223, "learning_rate": 3.173478624448744e-06, "loss": 0.5724, "step": 20967 }, { "epoch": 0.7470295883285534, "grad_norm": 1.155216981174921, "learning_rate": 3.1726354381638157e-06, "loss": 0.4128, "step": 20968 }, { "epoch": 0.7470652154550474, "grad_norm": 1.1277960594955796, "learning_rate": 3.171792342790019e-06, "loss": 0.4179, "step": 20969 }, { "epoch": 0.7471008425815416, "grad_norm": 1.6630066973415658, "learning_rate": 3.1709493383385824e-06, "loss": 0.5151, "step": 20970 }, { "epoch": 0.7471364697080357, "grad_norm": 1.2280335964579032, "learning_rate": 3.170106424820725e-06, "loss": 0.4272, "step": 20971 }, { "epoch": 0.7471720968345298, "grad_norm": 1.4199026333072755, "learning_rate": 3.169263602247675e-06, "loss": 0.543, "step": 20972 }, { "epoch": 0.747207723961024, "grad_norm": 1.1736500439435043, "learning_rate": 3.1684208706306572e-06, "loss": 0.3834, "step": 20973 }, { "epoch": 0.747243351087518, "grad_norm": 1.498664027601012, "learning_rate": 3.167578229980889e-06, "loss": 0.5337, "step": 20974 }, { "epoch": 0.7472789782140121, "grad_norm": 1.058532758041189, "learning_rate": 3.1667356803095905e-06, "loss": 0.4087, "step": 20975 }, { "epoch": 0.7473146053405063, "grad_norm": 1.3385800090536293, "learning_rate": 3.1658932216279825e-06, "loss": 0.4054, "step": 20976 }, { "epoch": 0.7473502324670004, "grad_norm": 1.138803774637818, "learning_rate": 3.165050853947287e-06, "loss": 0.417, "step": 20977 }, { "epoch": 0.7473858595934945, "grad_norm": 1.2963991343725065, "learning_rate": 3.1642085772787134e-06, "loss": 0.4213, "step": 20978 }, { "epoch": 0.7474214867199886, "grad_norm": 1.6694153625392307, "learning_rate": 3.1633663916334843e-06, "loss": 0.552, "step": 20979 }, { "epoch": 0.7474571138464827, "grad_norm": 1.1895316465519832, "learning_rate": 3.1625242970228065e-06, "loss": 0.3834, "step": 20980 }, { "epoch": 0.7474927409729768, "grad_norm": 1.0701045567345293, "learning_rate": 3.1616822934578994e-06, "loss": 0.4111, "step": 20981 }, { "epoch": 0.747528368099471, "grad_norm": 1.2482833646403793, "learning_rate": 3.1608403809499667e-06, "loss": 0.4052, "step": 20982 }, { "epoch": 0.7475639952259651, "grad_norm": 1.5063578947568492, "learning_rate": 3.1599985595102266e-06, "loss": 0.5418, "step": 20983 }, { "epoch": 0.7475996223524591, "grad_norm": 1.20035604475411, "learning_rate": 3.1591568291498907e-06, "loss": 0.3691, "step": 20984 }, { "epoch": 0.7476352494789533, "grad_norm": 1.219820512121702, "learning_rate": 3.1583151898801602e-06, "loss": 0.4391, "step": 20985 }, { "epoch": 0.7476708766054474, "grad_norm": 1.1071870089364613, "learning_rate": 3.1574736417122474e-06, "loss": 0.3894, "step": 20986 }, { "epoch": 0.7477065037319415, "grad_norm": 1.5967449597288461, "learning_rate": 3.1566321846573533e-06, "loss": 0.5355, "step": 20987 }, { "epoch": 0.7477421308584357, "grad_norm": 1.3942702851498339, "learning_rate": 3.1557908187266872e-06, "loss": 0.5392, "step": 20988 }, { "epoch": 0.7477777579849297, "grad_norm": 1.5883058309265028, "learning_rate": 3.154949543931448e-06, "loss": 0.4586, "step": 20989 }, { "epoch": 0.7478133851114238, "grad_norm": 1.8293170062211328, "learning_rate": 3.1541083602828394e-06, "loss": 0.589, "step": 20990 }, { "epoch": 0.747849012237918, "grad_norm": 1.4225716968406854, "learning_rate": 3.153267267792063e-06, "loss": 0.5256, "step": 20991 }, { "epoch": 0.7478846393644121, "grad_norm": 2.031219167828381, "learning_rate": 3.152426266470322e-06, "loss": 0.5383, "step": 20992 }, { "epoch": 0.7479202664909061, "grad_norm": 1.2079173143757813, "learning_rate": 3.151585356328808e-06, "loss": 0.4572, "step": 20993 }, { "epoch": 0.7479558936174003, "grad_norm": 1.5631370476611324, "learning_rate": 3.1507445373787214e-06, "loss": 0.5328, "step": 20994 }, { "epoch": 0.7479915207438944, "grad_norm": 1.1543279555076982, "learning_rate": 3.1499038096312617e-06, "loss": 0.411, "step": 20995 }, { "epoch": 0.7480271478703885, "grad_norm": 1.1322236251866602, "learning_rate": 3.1490631730976186e-06, "loss": 0.3936, "step": 20996 }, { "epoch": 0.7480627749968827, "grad_norm": 1.1366930244365487, "learning_rate": 3.1482226277889873e-06, "loss": 0.4133, "step": 20997 }, { "epoch": 0.7480984021233767, "grad_norm": 1.3382075935599027, "learning_rate": 3.14738217371656e-06, "loss": 0.5134, "step": 20998 }, { "epoch": 0.7481340292498708, "grad_norm": 1.1611424829500157, "learning_rate": 3.1465418108915336e-06, "loss": 0.4333, "step": 20999 }, { "epoch": 0.748169656376365, "grad_norm": 1.0607406963050552, "learning_rate": 3.145701539325089e-06, "loss": 0.3977, "step": 21000 }, { "epoch": 0.7482052835028591, "grad_norm": 1.2028221255321645, "learning_rate": 3.144861359028423e-06, "loss": 0.4288, "step": 21001 }, { "epoch": 0.7482409106293532, "grad_norm": 1.256222691588662, "learning_rate": 3.144021270012715e-06, "loss": 0.4202, "step": 21002 }, { "epoch": 0.7482765377558473, "grad_norm": 1.1673550682656848, "learning_rate": 3.1431812722891598e-06, "loss": 0.4068, "step": 21003 }, { "epoch": 0.7483121648823414, "grad_norm": 1.5386938952266502, "learning_rate": 3.142341365868936e-06, "loss": 0.5176, "step": 21004 }, { "epoch": 0.7483477920088355, "grad_norm": 1.4877216094389303, "learning_rate": 3.1415015507632294e-06, "loss": 0.4888, "step": 21005 }, { "epoch": 0.7483834191353297, "grad_norm": 1.1133126065132053, "learning_rate": 3.1406618269832232e-06, "loss": 0.418, "step": 21006 }, { "epoch": 0.7484190462618238, "grad_norm": 1.4823908656526665, "learning_rate": 3.1398221945401e-06, "loss": 0.4495, "step": 21007 }, { "epoch": 0.7484546733883178, "grad_norm": 1.4765786197661683, "learning_rate": 3.1389826534450417e-06, "loss": 0.5409, "step": 21008 }, { "epoch": 0.748490300514812, "grad_norm": 1.152796739329474, "learning_rate": 3.138143203709223e-06, "loss": 0.4172, "step": 21009 }, { "epoch": 0.7485259276413061, "grad_norm": 1.8434571567017506, "learning_rate": 3.1373038453438254e-06, "loss": 0.387, "step": 21010 }, { "epoch": 0.7485615547678002, "grad_norm": 1.410587289543102, "learning_rate": 3.136464578360021e-06, "loss": 0.473, "step": 21011 }, { "epoch": 0.7485971818942944, "grad_norm": 1.1120222182202866, "learning_rate": 3.135625402768989e-06, "loss": 0.3866, "step": 21012 }, { "epoch": 0.7486328090207884, "grad_norm": 1.5812543264386265, "learning_rate": 3.1347863185819016e-06, "loss": 0.5236, "step": 21013 }, { "epoch": 0.7486684361472825, "grad_norm": 1.2115402643105189, "learning_rate": 3.1339473258099374e-06, "loss": 0.3702, "step": 21014 }, { "epoch": 0.7487040632737767, "grad_norm": 1.118464871484476, "learning_rate": 3.1331084244642585e-06, "loss": 0.3798, "step": 21015 }, { "epoch": 0.7487396904002708, "grad_norm": 1.538624966064091, "learning_rate": 3.132269614556045e-06, "loss": 0.546, "step": 21016 }, { "epoch": 0.7487753175267648, "grad_norm": 1.1372222281732502, "learning_rate": 3.131430896096459e-06, "loss": 0.3929, "step": 21017 }, { "epoch": 0.748810944653259, "grad_norm": 1.162457621653799, "learning_rate": 3.1305922690966705e-06, "loss": 0.398, "step": 21018 }, { "epoch": 0.7488465717797531, "grad_norm": 1.0865079801343638, "learning_rate": 3.1297537335678507e-06, "loss": 0.4322, "step": 21019 }, { "epoch": 0.7488821989062472, "grad_norm": 1.4473064153568562, "learning_rate": 3.128915289521156e-06, "loss": 0.5117, "step": 21020 }, { "epoch": 0.7489178260327414, "grad_norm": 1.6096723521510647, "learning_rate": 3.1280769369677635e-06, "loss": 0.5414, "step": 21021 }, { "epoch": 0.7489534531592354, "grad_norm": 1.1482256453015345, "learning_rate": 3.1272386759188254e-06, "loss": 0.381, "step": 21022 }, { "epoch": 0.7489890802857295, "grad_norm": 1.0956364147187905, "learning_rate": 3.126400506385513e-06, "loss": 0.3794, "step": 21023 }, { "epoch": 0.7490247074122237, "grad_norm": 1.202486753902861, "learning_rate": 3.1255624283789776e-06, "loss": 0.4458, "step": 21024 }, { "epoch": 0.7490603345387178, "grad_norm": 1.2371030685904854, "learning_rate": 3.124724441910387e-06, "loss": 0.4373, "step": 21025 }, { "epoch": 0.7490959616652119, "grad_norm": 1.4815857087172177, "learning_rate": 3.123886546990893e-06, "loss": 0.5163, "step": 21026 }, { "epoch": 0.749131588791706, "grad_norm": 1.263725249499658, "learning_rate": 3.123048743631656e-06, "loss": 0.4148, "step": 21027 }, { "epoch": 0.7491672159182001, "grad_norm": 1.148266073002192, "learning_rate": 3.12221103184383e-06, "loss": 0.3885, "step": 21028 }, { "epoch": 0.7492028430446942, "grad_norm": 2.0034836612792426, "learning_rate": 3.121373411638573e-06, "loss": 0.5501, "step": 21029 }, { "epoch": 0.7492384701711884, "grad_norm": 1.6329918243327395, "learning_rate": 3.1205358830270405e-06, "loss": 0.5157, "step": 21030 }, { "epoch": 0.7492740972976825, "grad_norm": 2.8518932406890944, "learning_rate": 3.119698446020377e-06, "loss": 0.4536, "step": 21031 }, { "epoch": 0.7493097244241765, "grad_norm": 1.1557716924817514, "learning_rate": 3.118861100629741e-06, "loss": 0.4703, "step": 21032 }, { "epoch": 0.7493453515506707, "grad_norm": 1.2067791311505496, "learning_rate": 3.118023846866276e-06, "loss": 0.4177, "step": 21033 }, { "epoch": 0.7493809786771648, "grad_norm": 1.3997666133852422, "learning_rate": 3.1171866847411335e-06, "loss": 0.5021, "step": 21034 }, { "epoch": 0.7494166058036589, "grad_norm": 1.593065131777444, "learning_rate": 3.1163496142654613e-06, "loss": 0.5635, "step": 21035 }, { "epoch": 0.7494522329301531, "grad_norm": 1.13250149501277, "learning_rate": 3.1155126354504096e-06, "loss": 0.3931, "step": 21036 }, { "epoch": 0.7494878600566471, "grad_norm": 1.796281441592185, "learning_rate": 3.1146757483071144e-06, "loss": 0.5155, "step": 21037 }, { "epoch": 0.7495234871831412, "grad_norm": 1.2065072682926086, "learning_rate": 3.1138389528467284e-06, "loss": 0.4204, "step": 21038 }, { "epoch": 0.7495591143096354, "grad_norm": 1.171123339481286, "learning_rate": 3.1130022490803856e-06, "loss": 0.361, "step": 21039 }, { "epoch": 0.7495947414361295, "grad_norm": 2.0138250208037927, "learning_rate": 3.112165637019232e-06, "loss": 0.5239, "step": 21040 }, { "epoch": 0.7496303685626236, "grad_norm": 1.2447776635692747, "learning_rate": 3.111329116674411e-06, "loss": 0.4517, "step": 21041 }, { "epoch": 0.7496659956891177, "grad_norm": 1.2749902969858673, "learning_rate": 3.110492688057054e-06, "loss": 0.4451, "step": 21042 }, { "epoch": 0.7497016228156118, "grad_norm": 2.0900240139245665, "learning_rate": 3.109656351178302e-06, "loss": 0.5159, "step": 21043 }, { "epoch": 0.7497372499421059, "grad_norm": 1.1734814460187322, "learning_rate": 3.1088201060492927e-06, "loss": 0.3804, "step": 21044 }, { "epoch": 0.7497728770686001, "grad_norm": 1.1627662517142061, "learning_rate": 3.1079839526811627e-06, "loss": 0.3947, "step": 21045 }, { "epoch": 0.7498085041950941, "grad_norm": 1.7180762134921723, "learning_rate": 3.1071478910850405e-06, "loss": 0.503, "step": 21046 }, { "epoch": 0.7498441313215882, "grad_norm": 2.5357173470618304, "learning_rate": 3.106311921272066e-06, "loss": 0.5137, "step": 21047 }, { "epoch": 0.7498797584480824, "grad_norm": 1.4983910926742754, "learning_rate": 3.1054760432533626e-06, "loss": 0.5453, "step": 21048 }, { "epoch": 0.7499153855745765, "grad_norm": 1.1796286071702191, "learning_rate": 3.1046402570400648e-06, "loss": 0.4191, "step": 21049 }, { "epoch": 0.7499510127010706, "grad_norm": 1.7633898237470893, "learning_rate": 3.103804562643302e-06, "loss": 0.5385, "step": 21050 }, { "epoch": 0.7499866398275647, "grad_norm": 2.0899903467405623, "learning_rate": 3.102968960074204e-06, "loss": 0.4939, "step": 21051 }, { "epoch": 0.7500222669540588, "grad_norm": 1.6025039051537755, "learning_rate": 3.1021334493438924e-06, "loss": 0.522, "step": 21052 }, { "epoch": 0.7500578940805529, "grad_norm": 1.6479492908890128, "learning_rate": 3.1012980304634953e-06, "loss": 0.4796, "step": 21053 }, { "epoch": 0.7500935212070471, "grad_norm": 1.385215301066751, "learning_rate": 3.1004627034441393e-06, "loss": 0.4208, "step": 21054 }, { "epoch": 0.7501291483335412, "grad_norm": 1.2204285345094708, "learning_rate": 3.099627468296943e-06, "loss": 0.394, "step": 21055 }, { "epoch": 0.7501647754600352, "grad_norm": 1.2099634055887927, "learning_rate": 3.098792325033032e-06, "loss": 0.3887, "step": 21056 }, { "epoch": 0.7502004025865294, "grad_norm": 1.5071468442481961, "learning_rate": 3.0979572736635187e-06, "loss": 0.4886, "step": 21057 }, { "epoch": 0.7502360297130235, "grad_norm": 1.6044044418832162, "learning_rate": 3.0971223141995354e-06, "loss": 0.4631, "step": 21058 }, { "epoch": 0.7502716568395176, "grad_norm": 1.4683073571127747, "learning_rate": 3.09628744665219e-06, "loss": 0.4374, "step": 21059 }, { "epoch": 0.7503072839660118, "grad_norm": 1.532011854212291, "learning_rate": 3.095452671032606e-06, "loss": 0.482, "step": 21060 }, { "epoch": 0.7503429110925058, "grad_norm": 1.4505571804162511, "learning_rate": 3.0946179873518923e-06, "loss": 0.5388, "step": 21061 }, { "epoch": 0.7503785382189999, "grad_norm": 1.4621059937633532, "learning_rate": 3.0937833956211694e-06, "loss": 0.508, "step": 21062 }, { "epoch": 0.7504141653454941, "grad_norm": 1.2029786553934965, "learning_rate": 3.092948895851545e-06, "loss": 0.4041, "step": 21063 }, { "epoch": 0.7504497924719882, "grad_norm": 1.5662525806082874, "learning_rate": 3.0921144880541343e-06, "loss": 0.514, "step": 21064 }, { "epoch": 0.7504854195984823, "grad_norm": 1.1465436561268234, "learning_rate": 3.0912801722400466e-06, "loss": 0.4151, "step": 21065 }, { "epoch": 0.7505210467249764, "grad_norm": 1.8105547090492502, "learning_rate": 3.090445948420392e-06, "loss": 0.505, "step": 21066 }, { "epoch": 0.7505566738514705, "grad_norm": 1.2350329120895855, "learning_rate": 3.089611816606284e-06, "loss": 0.43, "step": 21067 }, { "epoch": 0.7505923009779646, "grad_norm": 1.8101472026631023, "learning_rate": 3.08877777680882e-06, "loss": 0.4624, "step": 21068 }, { "epoch": 0.7506279281044588, "grad_norm": 1.1723918601493464, "learning_rate": 3.087943829039114e-06, "loss": 0.4043, "step": 21069 }, { "epoch": 0.7506635552309529, "grad_norm": 1.4266130941977018, "learning_rate": 3.0871099733082642e-06, "loss": 0.4463, "step": 21070 }, { "epoch": 0.7506991823574469, "grad_norm": 1.3125011266409037, "learning_rate": 3.086276209627377e-06, "loss": 0.4524, "step": 21071 }, { "epoch": 0.7507348094839411, "grad_norm": 1.3280426853070986, "learning_rate": 3.0854425380075548e-06, "loss": 0.417, "step": 21072 }, { "epoch": 0.7507704366104352, "grad_norm": 1.564448813593944, "learning_rate": 3.0846089584599005e-06, "loss": 0.4904, "step": 21073 }, { "epoch": 0.7508060637369293, "grad_norm": 3.239930116927052, "learning_rate": 3.083775470995509e-06, "loss": 0.5448, "step": 21074 }, { "epoch": 0.7508416908634234, "grad_norm": 1.9361679636359383, "learning_rate": 3.0829420756254813e-06, "loss": 0.5353, "step": 21075 }, { "epoch": 0.7508773179899175, "grad_norm": 1.2530615698620449, "learning_rate": 3.0821087723609167e-06, "loss": 0.4136, "step": 21076 }, { "epoch": 0.7509129451164116, "grad_norm": 1.3743903627179894, "learning_rate": 3.0812755612129075e-06, "loss": 0.5374, "step": 21077 }, { "epoch": 0.7509485722429058, "grad_norm": 1.07898354767468, "learning_rate": 3.0804424421925515e-06, "loss": 0.4056, "step": 21078 }, { "epoch": 0.7509841993693999, "grad_norm": 1.481373937735576, "learning_rate": 3.079609415310939e-06, "loss": 0.6354, "step": 21079 }, { "epoch": 0.7510198264958939, "grad_norm": 1.1616143482528998, "learning_rate": 3.078776480579164e-06, "loss": 0.4244, "step": 21080 }, { "epoch": 0.7510554536223881, "grad_norm": 1.1758538082139172, "learning_rate": 3.077943638008318e-06, "loss": 0.4296, "step": 21081 }, { "epoch": 0.7510910807488822, "grad_norm": 1.1727410656697155, "learning_rate": 3.0771108876094935e-06, "loss": 0.3809, "step": 21082 }, { "epoch": 0.7511267078753763, "grad_norm": 1.2143941334155572, "learning_rate": 3.076278229393773e-06, "loss": 0.3437, "step": 21083 }, { "epoch": 0.7511623350018705, "grad_norm": 1.409648156525456, "learning_rate": 3.0754456633722506e-06, "loss": 0.5125, "step": 21084 }, { "epoch": 0.7511979621283645, "grad_norm": 1.1426531290895738, "learning_rate": 3.074613189556006e-06, "loss": 0.3667, "step": 21085 }, { "epoch": 0.7512335892548586, "grad_norm": 1.5982165954015612, "learning_rate": 3.0737808079561283e-06, "loss": 0.5723, "step": 21086 }, { "epoch": 0.7512692163813528, "grad_norm": 1.9427447657412396, "learning_rate": 3.0729485185836983e-06, "loss": 0.5437, "step": 21087 }, { "epoch": 0.7513048435078469, "grad_norm": 1.063646248903345, "learning_rate": 3.0721163214498016e-06, "loss": 0.4253, "step": 21088 }, { "epoch": 0.751340470634341, "grad_norm": 1.3834806715211347, "learning_rate": 3.071284216565522e-06, "loss": 0.4316, "step": 21089 }, { "epoch": 0.7513760977608351, "grad_norm": 1.1297856066630039, "learning_rate": 3.070452203941933e-06, "loss": 0.4354, "step": 21090 }, { "epoch": 0.7514117248873292, "grad_norm": 1.827985300882613, "learning_rate": 3.069620283590118e-06, "loss": 0.4159, "step": 21091 }, { "epoch": 0.7514473520138233, "grad_norm": 1.241158554440659, "learning_rate": 3.0687884555211513e-06, "loss": 0.4298, "step": 21092 }, { "epoch": 0.7514829791403175, "grad_norm": 1.2858774792117917, "learning_rate": 3.0679567197461135e-06, "loss": 0.4254, "step": 21093 }, { "epoch": 0.7515186062668116, "grad_norm": 1.6367104602343454, "learning_rate": 3.0671250762760728e-06, "loss": 0.4403, "step": 21094 }, { "epoch": 0.7515542333933056, "grad_norm": 1.7621710673436977, "learning_rate": 3.0662935251221094e-06, "loss": 0.5259, "step": 21095 }, { "epoch": 0.7515898605197998, "grad_norm": 1.1263695266423965, "learning_rate": 3.0654620662952926e-06, "loss": 0.3635, "step": 21096 }, { "epoch": 0.7516254876462939, "grad_norm": 1.3711436056699777, "learning_rate": 3.0646306998066997e-06, "loss": 0.4107, "step": 21097 }, { "epoch": 0.751661114772788, "grad_norm": 1.1298013488319854, "learning_rate": 3.063799425667393e-06, "loss": 0.39, "step": 21098 }, { "epoch": 0.7516967418992822, "grad_norm": 1.6763078821474808, "learning_rate": 3.0629682438884455e-06, "loss": 0.5378, "step": 21099 }, { "epoch": 0.7517323690257762, "grad_norm": 1.6847728525294048, "learning_rate": 3.0621371544809264e-06, "loss": 0.4755, "step": 21100 }, { "epoch": 0.7517679961522703, "grad_norm": 1.3897394364044386, "learning_rate": 3.061306157455899e-06, "loss": 0.3996, "step": 21101 }, { "epoch": 0.7518036232787645, "grad_norm": 1.2828926990163303, "learning_rate": 3.060475252824429e-06, "loss": 0.3871, "step": 21102 }, { "epoch": 0.7518392504052586, "grad_norm": 1.5090086649413583, "learning_rate": 3.059644440597581e-06, "loss": 0.4307, "step": 21103 }, { "epoch": 0.7518748775317526, "grad_norm": 1.5171614738036634, "learning_rate": 3.0588137207864223e-06, "loss": 0.5305, "step": 21104 }, { "epoch": 0.7519105046582468, "grad_norm": 1.0842161601516642, "learning_rate": 3.0579830934020062e-06, "loss": 0.404, "step": 21105 }, { "epoch": 0.7519461317847409, "grad_norm": 1.1610526232790295, "learning_rate": 3.057152558455402e-06, "loss": 0.4511, "step": 21106 }, { "epoch": 0.751981758911235, "grad_norm": 1.5962679323251303, "learning_rate": 3.0563221159576594e-06, "loss": 0.4715, "step": 21107 }, { "epoch": 0.7520173860377292, "grad_norm": 2.270804744576084, "learning_rate": 3.055491765919846e-06, "loss": 0.5389, "step": 21108 }, { "epoch": 0.7520530131642232, "grad_norm": 1.479887610806196, "learning_rate": 3.0546615083530064e-06, "loss": 0.4942, "step": 21109 }, { "epoch": 0.7520886402907173, "grad_norm": 1.5556311587731224, "learning_rate": 3.0538313432682064e-06, "loss": 0.4933, "step": 21110 }, { "epoch": 0.7521242674172115, "grad_norm": 2.349834830418124, "learning_rate": 3.053001270676501e-06, "loss": 0.4797, "step": 21111 }, { "epoch": 0.7521598945437056, "grad_norm": 1.2465648571391497, "learning_rate": 3.0521712905889357e-06, "loss": 0.425, "step": 21112 }, { "epoch": 0.7521955216701997, "grad_norm": 1.1383800095404721, "learning_rate": 3.051341403016569e-06, "loss": 0.4367, "step": 21113 }, { "epoch": 0.7522311487966938, "grad_norm": 2.009394196609214, "learning_rate": 3.0505116079704454e-06, "loss": 0.4603, "step": 21114 }, { "epoch": 0.7522667759231879, "grad_norm": 1.3112373975195255, "learning_rate": 3.04968190546162e-06, "loss": 0.5334, "step": 21115 }, { "epoch": 0.752302403049682, "grad_norm": 2.020209638365, "learning_rate": 3.0488522955011356e-06, "loss": 0.4577, "step": 21116 }, { "epoch": 0.7523380301761762, "grad_norm": 1.3017864037581797, "learning_rate": 3.0480227781000406e-06, "loss": 0.3889, "step": 21117 }, { "epoch": 0.7523736573026703, "grad_norm": 1.2950398423760814, "learning_rate": 3.047193353269382e-06, "loss": 0.3908, "step": 21118 }, { "epoch": 0.7524092844291643, "grad_norm": 1.2188572175295993, "learning_rate": 3.046364021020206e-06, "loss": 0.441, "step": 21119 }, { "epoch": 0.7524449115556585, "grad_norm": 1.2220884324564887, "learning_rate": 3.045534781363552e-06, "loss": 0.4146, "step": 21120 }, { "epoch": 0.7524805386821526, "grad_norm": 1.2211785399924873, "learning_rate": 3.044705634310462e-06, "loss": 0.4042, "step": 21121 }, { "epoch": 0.7525161658086467, "grad_norm": 1.7639620356662828, "learning_rate": 3.0438765798719807e-06, "loss": 0.4205, "step": 21122 }, { "epoch": 0.7525517929351409, "grad_norm": 1.1467118357657875, "learning_rate": 3.0430476180591416e-06, "loss": 0.4307, "step": 21123 }, { "epoch": 0.7525874200616349, "grad_norm": 1.4000574595958917, "learning_rate": 3.042218748882987e-06, "loss": 0.424, "step": 21124 }, { "epoch": 0.752623047188129, "grad_norm": 1.7119590224119667, "learning_rate": 3.0413899723545526e-06, "loss": 0.5871, "step": 21125 }, { "epoch": 0.7526586743146232, "grad_norm": 1.6434171969311404, "learning_rate": 3.040561288484878e-06, "loss": 0.4936, "step": 21126 }, { "epoch": 0.7526943014411173, "grad_norm": 1.3753653532812702, "learning_rate": 3.0397326972849895e-06, "loss": 0.4562, "step": 21127 }, { "epoch": 0.7527299285676113, "grad_norm": 1.4339614197412032, "learning_rate": 3.0389041987659296e-06, "loss": 0.4835, "step": 21128 }, { "epoch": 0.7527655556941055, "grad_norm": 1.078347889986674, "learning_rate": 3.0380757929387226e-06, "loss": 0.3539, "step": 21129 }, { "epoch": 0.7528011828205996, "grad_norm": 1.0553585001437051, "learning_rate": 3.037247479814406e-06, "loss": 0.4156, "step": 21130 }, { "epoch": 0.7528368099470937, "grad_norm": 1.1715937353277524, "learning_rate": 3.036419259404002e-06, "loss": 0.3877, "step": 21131 }, { "epoch": 0.7528724370735879, "grad_norm": 1.1843142850054944, "learning_rate": 3.0355911317185392e-06, "loss": 0.4382, "step": 21132 }, { "epoch": 0.7529080642000819, "grad_norm": 1.7587405041846738, "learning_rate": 3.0347630967690556e-06, "loss": 0.549, "step": 21133 }, { "epoch": 0.752943691326576, "grad_norm": 1.295124960679049, "learning_rate": 3.0339351545665665e-06, "loss": 0.3948, "step": 21134 }, { "epoch": 0.7529793184530702, "grad_norm": 1.564533754272988, "learning_rate": 3.0331073051221027e-06, "loss": 0.6131, "step": 21135 }, { "epoch": 0.7530149455795643, "grad_norm": 2.3456733173767215, "learning_rate": 3.0322795484466817e-06, "loss": 0.4242, "step": 21136 }, { "epoch": 0.7530505727060584, "grad_norm": 1.1226779678007341, "learning_rate": 3.031451884551331e-06, "loss": 0.4506, "step": 21137 }, { "epoch": 0.7530861998325525, "grad_norm": 1.1610059623753939, "learning_rate": 3.0306243134470668e-06, "loss": 0.4058, "step": 21138 }, { "epoch": 0.7531218269590466, "grad_norm": 1.5243412461400436, "learning_rate": 3.0297968351449115e-06, "loss": 0.4725, "step": 21139 }, { "epoch": 0.7531574540855407, "grad_norm": 1.5294744161447456, "learning_rate": 3.028969449655883e-06, "loss": 0.499, "step": 21140 }, { "epoch": 0.7531930812120349, "grad_norm": 1.307442564031567, "learning_rate": 3.028142156991001e-06, "loss": 0.4619, "step": 21141 }, { "epoch": 0.753228708338529, "grad_norm": 1.2335811593681736, "learning_rate": 3.027314957161277e-06, "loss": 0.4413, "step": 21142 }, { "epoch": 0.753264335465023, "grad_norm": 1.5388711057293292, "learning_rate": 3.0264878501777307e-06, "loss": 0.5398, "step": 21143 }, { "epoch": 0.7532999625915172, "grad_norm": 1.335680162205778, "learning_rate": 3.0256608360513708e-06, "loss": 0.4815, "step": 21144 }, { "epoch": 0.7533355897180113, "grad_norm": 1.1523555083919665, "learning_rate": 3.0248339147932106e-06, "loss": 0.4046, "step": 21145 }, { "epoch": 0.7533712168445054, "grad_norm": 1.8714349970306006, "learning_rate": 3.024007086414261e-06, "loss": 0.4934, "step": 21146 }, { "epoch": 0.7534068439709996, "grad_norm": 1.1676129965178594, "learning_rate": 3.0231803509255353e-06, "loss": 0.4131, "step": 21147 }, { "epoch": 0.7534424710974936, "grad_norm": 1.353693147003788, "learning_rate": 3.022353708338042e-06, "loss": 0.5385, "step": 21148 }, { "epoch": 0.7534780982239877, "grad_norm": 1.3542225215211, "learning_rate": 3.0215271586627824e-06, "loss": 0.4058, "step": 21149 }, { "epoch": 0.7535137253504819, "grad_norm": 1.0501976678073957, "learning_rate": 3.0207007019107705e-06, "loss": 0.3476, "step": 21150 }, { "epoch": 0.753549352476976, "grad_norm": 1.2367638551992959, "learning_rate": 3.019874338093003e-06, "loss": 0.4156, "step": 21151 }, { "epoch": 0.75358497960347, "grad_norm": 1.1307057786411125, "learning_rate": 3.01904806722049e-06, "loss": 0.4264, "step": 21152 }, { "epoch": 0.7536206067299642, "grad_norm": 1.3598740327241927, "learning_rate": 3.0182218893042304e-06, "loss": 0.5376, "step": 21153 }, { "epoch": 0.7536562338564583, "grad_norm": 1.3654236152762815, "learning_rate": 3.017395804355224e-06, "loss": 0.4559, "step": 21154 }, { "epoch": 0.7536918609829524, "grad_norm": 1.710634345793599, "learning_rate": 3.0165698123844744e-06, "loss": 0.5075, "step": 21155 }, { "epoch": 0.7537274881094466, "grad_norm": 1.3183154866203044, "learning_rate": 3.0157439134029786e-06, "loss": 0.508, "step": 21156 }, { "epoch": 0.7537631152359406, "grad_norm": 1.2760220358102445, "learning_rate": 3.014918107421738e-06, "loss": 0.4538, "step": 21157 }, { "epoch": 0.7537987423624347, "grad_norm": 1.284123112736947, "learning_rate": 3.0140923944517407e-06, "loss": 0.5097, "step": 21158 }, { "epoch": 0.7538343694889289, "grad_norm": 1.6105606031881028, "learning_rate": 3.01326677450399e-06, "loss": 0.5206, "step": 21159 }, { "epoch": 0.753869996615423, "grad_norm": 1.0771325518205546, "learning_rate": 3.0124412475894726e-06, "loss": 0.3542, "step": 21160 }, { "epoch": 0.753905623741917, "grad_norm": 1.2476487910522338, "learning_rate": 3.011615813719184e-06, "loss": 0.4137, "step": 21161 }, { "epoch": 0.7539412508684112, "grad_norm": 1.6609187251871311, "learning_rate": 3.0107904729041147e-06, "loss": 0.4297, "step": 21162 }, { "epoch": 0.7539768779949053, "grad_norm": 1.1590626468504561, "learning_rate": 3.009965225155259e-06, "loss": 0.4161, "step": 21163 }, { "epoch": 0.7540125051213994, "grad_norm": 1.3316009125409785, "learning_rate": 3.0091400704835995e-06, "loss": 0.4735, "step": 21164 }, { "epoch": 0.7540481322478936, "grad_norm": 1.6645518049947565, "learning_rate": 3.008315008900129e-06, "loss": 0.6055, "step": 21165 }, { "epoch": 0.7540837593743877, "grad_norm": 1.7017467246067386, "learning_rate": 3.007490040415827e-06, "loss": 0.538, "step": 21166 }, { "epoch": 0.7541193865008817, "grad_norm": 1.23107296028568, "learning_rate": 3.0066651650416834e-06, "loss": 0.3777, "step": 21167 }, { "epoch": 0.7541550136273759, "grad_norm": 1.355796586006863, "learning_rate": 3.005840382788685e-06, "loss": 0.4348, "step": 21168 }, { "epoch": 0.75419064075387, "grad_norm": 1.1717221676204923, "learning_rate": 3.0050156936678064e-06, "loss": 0.4146, "step": 21169 }, { "epoch": 0.7542262678803641, "grad_norm": 1.6094305446284654, "learning_rate": 3.0041910976900335e-06, "loss": 0.5582, "step": 21170 }, { "epoch": 0.7542618950068583, "grad_norm": 1.8867536801716822, "learning_rate": 3.003366594866345e-06, "loss": 0.5507, "step": 21171 }, { "epoch": 0.7542975221333523, "grad_norm": 1.4548027855911123, "learning_rate": 3.002542185207725e-06, "loss": 0.4339, "step": 21172 }, { "epoch": 0.7543331492598464, "grad_norm": 1.5754685085404376, "learning_rate": 3.0017178687251424e-06, "loss": 0.5233, "step": 21173 }, { "epoch": 0.7543687763863406, "grad_norm": 1.239333652547365, "learning_rate": 3.000893645429582e-06, "loss": 0.3913, "step": 21174 }, { "epoch": 0.7544044035128347, "grad_norm": 1.1438758148808394, "learning_rate": 3.0000695153320114e-06, "loss": 0.4053, "step": 21175 }, { "epoch": 0.7544400306393287, "grad_norm": 2.013163734291235, "learning_rate": 2.999245478443408e-06, "loss": 0.4271, "step": 21176 }, { "epoch": 0.7544756577658229, "grad_norm": 1.250122216731784, "learning_rate": 2.9984215347747437e-06, "loss": 0.3793, "step": 21177 }, { "epoch": 0.754511284892317, "grad_norm": 2.944327318039864, "learning_rate": 2.9975976843369946e-06, "loss": 0.5506, "step": 21178 }, { "epoch": 0.7545469120188111, "grad_norm": 1.4701936865016587, "learning_rate": 2.9967739271411235e-06, "loss": 0.5005, "step": 21179 }, { "epoch": 0.7545825391453053, "grad_norm": 1.6067111736645043, "learning_rate": 2.995950263198102e-06, "loss": 0.4964, "step": 21180 }, { "epoch": 0.7546181662717993, "grad_norm": 1.4781882603409373, "learning_rate": 2.9951266925189006e-06, "loss": 0.5362, "step": 21181 }, { "epoch": 0.7546537933982934, "grad_norm": 1.917718559457962, "learning_rate": 2.9943032151144814e-06, "loss": 0.4974, "step": 21182 }, { "epoch": 0.7546894205247876, "grad_norm": 1.642777172433649, "learning_rate": 2.993479830995815e-06, "loss": 0.4396, "step": 21183 }, { "epoch": 0.7547250476512817, "grad_norm": 1.6625811902004572, "learning_rate": 2.992656540173855e-06, "loss": 0.4988, "step": 21184 }, { "epoch": 0.7547606747777758, "grad_norm": 1.1157746060006968, "learning_rate": 2.991833342659577e-06, "loss": 0.4096, "step": 21185 }, { "epoch": 0.7547963019042699, "grad_norm": 1.2560537722840464, "learning_rate": 2.991010238463934e-06, "loss": 0.4034, "step": 21186 }, { "epoch": 0.754831929030764, "grad_norm": 1.6215021513800634, "learning_rate": 2.9901872275978915e-06, "loss": 0.6258, "step": 21187 }, { "epoch": 0.7548675561572581, "grad_norm": 1.498582862731227, "learning_rate": 2.9893643100724023e-06, "loss": 0.5519, "step": 21188 }, { "epoch": 0.7549031832837523, "grad_norm": 1.2562830692509388, "learning_rate": 2.9885414858984308e-06, "loss": 0.4188, "step": 21189 }, { "epoch": 0.7549388104102464, "grad_norm": 1.4093297388463149, "learning_rate": 2.987718755086927e-06, "loss": 0.4796, "step": 21190 }, { "epoch": 0.7549744375367404, "grad_norm": 1.307479973987058, "learning_rate": 2.9868961176488488e-06, "loss": 0.4139, "step": 21191 }, { "epoch": 0.7550100646632346, "grad_norm": 1.28018944124329, "learning_rate": 2.986073573595152e-06, "loss": 0.5374, "step": 21192 }, { "epoch": 0.7550456917897287, "grad_norm": 1.1851293808713101, "learning_rate": 2.9852511229367862e-06, "loss": 0.3968, "step": 21193 }, { "epoch": 0.7550813189162228, "grad_norm": 1.559996210416683, "learning_rate": 2.9844287656847095e-06, "loss": 0.535, "step": 21194 }, { "epoch": 0.755116946042717, "grad_norm": 1.080246687272268, "learning_rate": 2.9836065018498628e-06, "loss": 0.4262, "step": 21195 }, { "epoch": 0.755152573169211, "grad_norm": 1.2673859573392217, "learning_rate": 2.9827843314432037e-06, "loss": 0.4394, "step": 21196 }, { "epoch": 0.7551882002957051, "grad_norm": 1.181845484283403, "learning_rate": 2.981962254475673e-06, "loss": 0.4245, "step": 21197 }, { "epoch": 0.7552238274221993, "grad_norm": 1.3687976983202315, "learning_rate": 2.9811402709582195e-06, "loss": 0.4993, "step": 21198 }, { "epoch": 0.7552594545486934, "grad_norm": 1.5000901956086528, "learning_rate": 2.98031838090179e-06, "loss": 0.4903, "step": 21199 }, { "epoch": 0.7552950816751874, "grad_norm": 1.3740475747099479, "learning_rate": 2.9794965843173294e-06, "loss": 0.424, "step": 21200 }, { "epoch": 0.7553307088016816, "grad_norm": 1.6314741579980863, "learning_rate": 2.978674881215777e-06, "loss": 0.5275, "step": 21201 }, { "epoch": 0.7553663359281757, "grad_norm": 1.7162601025142101, "learning_rate": 2.9778532716080764e-06, "loss": 0.5383, "step": 21202 }, { "epoch": 0.7554019630546698, "grad_norm": 1.277010314639615, "learning_rate": 2.9770317555051697e-06, "loss": 0.4499, "step": 21203 }, { "epoch": 0.755437590181164, "grad_norm": 1.209610993739746, "learning_rate": 2.9762103329179916e-06, "loss": 0.4115, "step": 21204 }, { "epoch": 0.755473217307658, "grad_norm": 1.1603507677781668, "learning_rate": 2.9753890038574853e-06, "loss": 0.4243, "step": 21205 }, { "epoch": 0.7555088444341521, "grad_norm": 1.6104768606798257, "learning_rate": 2.974567768334581e-06, "loss": 0.5176, "step": 21206 }, { "epoch": 0.7555444715606463, "grad_norm": 2.1640681419924, "learning_rate": 2.9737466263602176e-06, "loss": 0.5272, "step": 21207 }, { "epoch": 0.7555800986871404, "grad_norm": 1.4688567823216174, "learning_rate": 2.972925577945328e-06, "loss": 0.5187, "step": 21208 }, { "epoch": 0.7556157258136345, "grad_norm": 1.8475130197033847, "learning_rate": 2.97210462310085e-06, "loss": 0.4837, "step": 21209 }, { "epoch": 0.7556513529401286, "grad_norm": 1.2799384635323132, "learning_rate": 2.9712837618377076e-06, "loss": 0.3894, "step": 21210 }, { "epoch": 0.7556869800666227, "grad_norm": 1.145344555756704, "learning_rate": 2.9704629941668384e-06, "loss": 0.4194, "step": 21211 }, { "epoch": 0.7557226071931168, "grad_norm": 1.51149066650839, "learning_rate": 2.9696423200991644e-06, "loss": 0.5135, "step": 21212 }, { "epoch": 0.755758234319611, "grad_norm": 1.4371465260888252, "learning_rate": 2.9688217396456166e-06, "loss": 0.4822, "step": 21213 }, { "epoch": 0.755793861446105, "grad_norm": 1.498904997928416, "learning_rate": 2.968001252817122e-06, "loss": 0.5937, "step": 21214 }, { "epoch": 0.7558294885725991, "grad_norm": 1.321790731394836, "learning_rate": 2.9671808596246053e-06, "loss": 0.4442, "step": 21215 }, { "epoch": 0.7558651156990933, "grad_norm": 1.7115569021220485, "learning_rate": 2.9663605600789957e-06, "loss": 0.4953, "step": 21216 }, { "epoch": 0.7559007428255874, "grad_norm": 1.1879618947167405, "learning_rate": 2.9655403541912065e-06, "loss": 0.3973, "step": 21217 }, { "epoch": 0.7559363699520815, "grad_norm": 1.2190905101442013, "learning_rate": 2.9647202419721687e-06, "loss": 0.4676, "step": 21218 }, { "epoch": 0.7559719970785757, "grad_norm": 1.5612313207224167, "learning_rate": 2.963900223432794e-06, "loss": 0.4947, "step": 21219 }, { "epoch": 0.7560076242050697, "grad_norm": 1.1861795714630257, "learning_rate": 2.96308029858401e-06, "loss": 0.4074, "step": 21220 }, { "epoch": 0.7560432513315638, "grad_norm": 1.6010959178340185, "learning_rate": 2.962260467436724e-06, "loss": 0.5062, "step": 21221 }, { "epoch": 0.756078878458058, "grad_norm": 1.2533802553418116, "learning_rate": 2.961440730001864e-06, "loss": 0.4399, "step": 21222 }, { "epoch": 0.7561145055845521, "grad_norm": 1.5193953192697258, "learning_rate": 2.9606210862903385e-06, "loss": 0.5251, "step": 21223 }, { "epoch": 0.7561501327110461, "grad_norm": 1.0797955817545692, "learning_rate": 2.959801536313066e-06, "loss": 0.3678, "step": 21224 }, { "epoch": 0.7561857598375403, "grad_norm": 1.2074874672247584, "learning_rate": 2.9589820800809545e-06, "loss": 0.3844, "step": 21225 }, { "epoch": 0.7562213869640344, "grad_norm": 1.1742065200757017, "learning_rate": 2.9581627176049167e-06, "loss": 0.4329, "step": 21226 }, { "epoch": 0.7562570140905285, "grad_norm": 1.2621195795050726, "learning_rate": 2.9573434488958687e-06, "loss": 0.4255, "step": 21227 }, { "epoch": 0.7562926412170227, "grad_norm": 1.4882977741031906, "learning_rate": 2.9565242739647115e-06, "loss": 0.5002, "step": 21228 }, { "epoch": 0.7563282683435167, "grad_norm": 1.1327937623279833, "learning_rate": 2.9557051928223556e-06, "loss": 0.3535, "step": 21229 }, { "epoch": 0.7563638954700108, "grad_norm": 1.5580450131565904, "learning_rate": 2.95488620547971e-06, "loss": 0.553, "step": 21230 }, { "epoch": 0.756399522596505, "grad_norm": 1.2888109097298488, "learning_rate": 2.9540673119476816e-06, "loss": 0.4239, "step": 21231 }, { "epoch": 0.7564351497229991, "grad_norm": 1.6881612646579613, "learning_rate": 2.953248512237169e-06, "loss": 0.4541, "step": 21232 }, { "epoch": 0.7564707768494932, "grad_norm": 2.060817496235729, "learning_rate": 2.9524298063590808e-06, "loss": 0.4599, "step": 21233 }, { "epoch": 0.7565064039759873, "grad_norm": 1.1382943384305695, "learning_rate": 2.951611194324312e-06, "loss": 0.4003, "step": 21234 }, { "epoch": 0.7565420311024814, "grad_norm": 1.3077876183697537, "learning_rate": 2.9507926761437676e-06, "loss": 0.4227, "step": 21235 }, { "epoch": 0.7565776582289755, "grad_norm": 1.2400931675241051, "learning_rate": 2.949974251828345e-06, "loss": 0.3912, "step": 21236 }, { "epoch": 0.7566132853554697, "grad_norm": 1.143850442075043, "learning_rate": 2.9491559213889433e-06, "loss": 0.4105, "step": 21237 }, { "epoch": 0.7566489124819638, "grad_norm": 1.1391591765342421, "learning_rate": 2.948337684836462e-06, "loss": 0.3824, "step": 21238 }, { "epoch": 0.7566845396084578, "grad_norm": 1.278113154876323, "learning_rate": 2.9475195421817895e-06, "loss": 0.3913, "step": 21239 }, { "epoch": 0.756720166734952, "grad_norm": 1.4645205621939748, "learning_rate": 2.946701493435827e-06, "loss": 0.5648, "step": 21240 }, { "epoch": 0.7567557938614461, "grad_norm": 1.847043642759678, "learning_rate": 2.945883538609462e-06, "loss": 0.536, "step": 21241 }, { "epoch": 0.7567914209879402, "grad_norm": 1.4570463917378491, "learning_rate": 2.94506567771359e-06, "loss": 0.5784, "step": 21242 }, { "epoch": 0.7568270481144344, "grad_norm": 1.1072482626759828, "learning_rate": 2.944247910759097e-06, "loss": 0.4043, "step": 21243 }, { "epoch": 0.7568626752409284, "grad_norm": 1.0572290713338832, "learning_rate": 2.9434302377568747e-06, "loss": 0.3292, "step": 21244 }, { "epoch": 0.7568983023674225, "grad_norm": 1.2611396640113537, "learning_rate": 2.9426126587178117e-06, "loss": 0.4539, "step": 21245 }, { "epoch": 0.7569339294939167, "grad_norm": 1.8560563893716417, "learning_rate": 2.941795173652796e-06, "loss": 0.4846, "step": 21246 }, { "epoch": 0.7569695566204108, "grad_norm": 1.3340982566899702, "learning_rate": 2.940977782572708e-06, "loss": 0.4215, "step": 21247 }, { "epoch": 0.7570051837469048, "grad_norm": 1.265728970674573, "learning_rate": 2.940160485488436e-06, "loss": 0.412, "step": 21248 }, { "epoch": 0.757040810873399, "grad_norm": 1.5790318275789326, "learning_rate": 2.9393432824108636e-06, "loss": 0.5052, "step": 21249 }, { "epoch": 0.7570764379998931, "grad_norm": 1.7573208922834869, "learning_rate": 2.9385261733508676e-06, "loss": 0.5197, "step": 21250 }, { "epoch": 0.7571120651263872, "grad_norm": 1.3211026448326868, "learning_rate": 2.9377091583193317e-06, "loss": 0.4173, "step": 21251 }, { "epoch": 0.7571476922528814, "grad_norm": 1.263429529639403, "learning_rate": 2.9368922373271338e-06, "loss": 0.3984, "step": 21252 }, { "epoch": 0.7571833193793754, "grad_norm": 2.8378035134139625, "learning_rate": 2.936075410385156e-06, "loss": 0.4629, "step": 21253 }, { "epoch": 0.7572189465058695, "grad_norm": 1.3787650711190231, "learning_rate": 2.9352586775042692e-06, "loss": 0.4515, "step": 21254 }, { "epoch": 0.7572545736323637, "grad_norm": 1.7767588644352823, "learning_rate": 2.934442038695353e-06, "loss": 0.4896, "step": 21255 }, { "epoch": 0.7572902007588578, "grad_norm": 1.1332159726486024, "learning_rate": 2.9336254939692765e-06, "loss": 0.4236, "step": 21256 }, { "epoch": 0.7573258278853519, "grad_norm": 1.8112141319957098, "learning_rate": 2.932809043336919e-06, "loss": 0.6008, "step": 21257 }, { "epoch": 0.757361455011846, "grad_norm": 2.2326389123112875, "learning_rate": 2.9319926868091464e-06, "loss": 0.5175, "step": 21258 }, { "epoch": 0.7573970821383401, "grad_norm": 1.306641433962172, "learning_rate": 2.9311764243968278e-06, "loss": 0.4479, "step": 21259 }, { "epoch": 0.7574327092648342, "grad_norm": 1.7034956606634455, "learning_rate": 2.9303602561108425e-06, "loss": 0.5868, "step": 21260 }, { "epoch": 0.7574683363913284, "grad_norm": 1.8659751911237146, "learning_rate": 2.9295441819620483e-06, "loss": 0.4822, "step": 21261 }, { "epoch": 0.7575039635178225, "grad_norm": 1.531346037923888, "learning_rate": 2.928728201961318e-06, "loss": 0.5254, "step": 21262 }, { "epoch": 0.7575395906443165, "grad_norm": 1.4925857605567154, "learning_rate": 2.927912316119511e-06, "loss": 0.5208, "step": 21263 }, { "epoch": 0.7575752177708107, "grad_norm": 1.2554493454712536, "learning_rate": 2.9270965244474994e-06, "loss": 0.4078, "step": 21264 }, { "epoch": 0.7576108448973048, "grad_norm": 1.4255577573605287, "learning_rate": 2.926280826956136e-06, "loss": 0.4728, "step": 21265 }, { "epoch": 0.7576464720237989, "grad_norm": 1.3212593129776706, "learning_rate": 2.9254652236562887e-06, "loss": 0.362, "step": 21266 }, { "epoch": 0.757682099150293, "grad_norm": 2.090365016525876, "learning_rate": 2.924649714558817e-06, "loss": 0.5251, "step": 21267 }, { "epoch": 0.7577177262767871, "grad_norm": 1.377544230076587, "learning_rate": 2.9238342996745816e-06, "loss": 0.4957, "step": 21268 }, { "epoch": 0.7577533534032812, "grad_norm": 1.6313757059565317, "learning_rate": 2.923018979014437e-06, "loss": 0.5241, "step": 21269 }, { "epoch": 0.7577889805297754, "grad_norm": 1.1924236951381308, "learning_rate": 2.9222037525892433e-06, "loss": 0.4269, "step": 21270 }, { "epoch": 0.7578246076562695, "grad_norm": 1.1527562016397543, "learning_rate": 2.9213886204098508e-06, "loss": 0.4008, "step": 21271 }, { "epoch": 0.7578602347827635, "grad_norm": 1.5907554395865442, "learning_rate": 2.920573582487116e-06, "loss": 0.5244, "step": 21272 }, { "epoch": 0.7578958619092577, "grad_norm": 1.5267039870064396, "learning_rate": 2.919758638831893e-06, "loss": 0.4395, "step": 21273 }, { "epoch": 0.7579314890357518, "grad_norm": 1.4513476988264375, "learning_rate": 2.918943789455032e-06, "loss": 0.5694, "step": 21274 }, { "epoch": 0.7579671161622459, "grad_norm": 1.4646523100449005, "learning_rate": 2.9181290343673876e-06, "loss": 0.5216, "step": 21275 }, { "epoch": 0.7580027432887401, "grad_norm": 1.2606866648898496, "learning_rate": 2.9173143735798016e-06, "loss": 0.405, "step": 21276 }, { "epoch": 0.7580383704152341, "grad_norm": 1.5067166991499525, "learning_rate": 2.916499807103128e-06, "loss": 0.5467, "step": 21277 }, { "epoch": 0.7580739975417282, "grad_norm": 1.1684196758439354, "learning_rate": 2.9156853349482075e-06, "loss": 0.4333, "step": 21278 }, { "epoch": 0.7581096246682224, "grad_norm": 1.1936382088029365, "learning_rate": 2.9148709571258914e-06, "loss": 0.4155, "step": 21279 }, { "epoch": 0.7581452517947165, "grad_norm": 1.523628939842736, "learning_rate": 2.9140566736470176e-06, "loss": 0.4963, "step": 21280 }, { "epoch": 0.7581808789212106, "grad_norm": 1.1585176081337687, "learning_rate": 2.9132424845224316e-06, "loss": 0.3797, "step": 21281 }, { "epoch": 0.7582165060477047, "grad_norm": 1.1612218640387721, "learning_rate": 2.912428389762977e-06, "loss": 0.4368, "step": 21282 }, { "epoch": 0.7582521331741988, "grad_norm": 1.724584780516177, "learning_rate": 2.91161438937949e-06, "loss": 0.5607, "step": 21283 }, { "epoch": 0.7582877603006929, "grad_norm": 1.2959847852867978, "learning_rate": 2.9108004833828164e-06, "loss": 0.4555, "step": 21284 }, { "epoch": 0.7583233874271871, "grad_norm": 1.6253073242787701, "learning_rate": 2.9099866717837864e-06, "loss": 0.548, "step": 21285 }, { "epoch": 0.7583590145536812, "grad_norm": 1.7719822727446835, "learning_rate": 2.9091729545932424e-06, "loss": 0.5281, "step": 21286 }, { "epoch": 0.7583946416801753, "grad_norm": 1.6383199890004356, "learning_rate": 2.908359331822013e-06, "loss": 0.4915, "step": 21287 }, { "epoch": 0.7584302688066694, "grad_norm": 1.2229516834147987, "learning_rate": 2.9075458034809368e-06, "loss": 0.3673, "step": 21288 }, { "epoch": 0.7584658959331635, "grad_norm": 1.6854824939148505, "learning_rate": 2.906732369580845e-06, "loss": 0.5474, "step": 21289 }, { "epoch": 0.7585015230596577, "grad_norm": 1.2498836199204006, "learning_rate": 2.9059190301325723e-06, "loss": 0.3959, "step": 21290 }, { "epoch": 0.7585371501861518, "grad_norm": 1.4112395711003103, "learning_rate": 2.9051057851469433e-06, "loss": 0.436, "step": 21291 }, { "epoch": 0.7585727773126458, "grad_norm": 1.4422964649342347, "learning_rate": 2.9042926346347932e-06, "loss": 0.5664, "step": 21292 }, { "epoch": 0.75860840443914, "grad_norm": 1.539773856033462, "learning_rate": 2.9034795786069436e-06, "loss": 0.5346, "step": 21293 }, { "epoch": 0.7586440315656341, "grad_norm": 1.4753603983697539, "learning_rate": 2.902666617074222e-06, "loss": 0.4941, "step": 21294 }, { "epoch": 0.7586796586921282, "grad_norm": 1.4861815676112775, "learning_rate": 2.9018537500474595e-06, "loss": 0.4883, "step": 21295 }, { "epoch": 0.7587152858186224, "grad_norm": 1.16473194253846, "learning_rate": 2.9010409775374692e-06, "loss": 0.4083, "step": 21296 }, { "epoch": 0.7587509129451164, "grad_norm": 1.1396646926263663, "learning_rate": 2.9002282995550866e-06, "loss": 0.4204, "step": 21297 }, { "epoch": 0.7587865400716105, "grad_norm": 1.4898516828053625, "learning_rate": 2.8994157161111235e-06, "loss": 0.5518, "step": 21298 }, { "epoch": 0.7588221671981047, "grad_norm": 1.5305861364029885, "learning_rate": 2.8986032272164068e-06, "loss": 0.5302, "step": 21299 }, { "epoch": 0.7588577943245988, "grad_norm": 1.315476399095239, "learning_rate": 2.897790832881747e-06, "loss": 0.432, "step": 21300 }, { "epoch": 0.7588934214510928, "grad_norm": 1.1749778169131604, "learning_rate": 2.89697853311797e-06, "loss": 0.3565, "step": 21301 }, { "epoch": 0.758929048577587, "grad_norm": 1.7593200611286943, "learning_rate": 2.8961663279358865e-06, "loss": 0.4887, "step": 21302 }, { "epoch": 0.7589646757040811, "grad_norm": 1.2027211293067406, "learning_rate": 2.8953542173463133e-06, "loss": 0.4017, "step": 21303 }, { "epoch": 0.7590003028305752, "grad_norm": 1.082553873319718, "learning_rate": 2.894542201360063e-06, "loss": 0.3823, "step": 21304 }, { "epoch": 0.7590359299570694, "grad_norm": 1.0732752171200304, "learning_rate": 2.8937302799879553e-06, "loss": 0.3319, "step": 21305 }, { "epoch": 0.7590715570835634, "grad_norm": 1.2315129769793214, "learning_rate": 2.8929184532407905e-06, "loss": 0.4173, "step": 21306 }, { "epoch": 0.7591071842100575, "grad_norm": 1.543255502767927, "learning_rate": 2.892106721129385e-06, "loss": 0.4819, "step": 21307 }, { "epoch": 0.7591428113365517, "grad_norm": 1.1777298391190865, "learning_rate": 2.8912950836645505e-06, "loss": 0.3719, "step": 21308 }, { "epoch": 0.7591784384630458, "grad_norm": 1.48593088807121, "learning_rate": 2.8904835408570865e-06, "loss": 0.4925, "step": 21309 }, { "epoch": 0.7592140655895399, "grad_norm": 1.7106332042282326, "learning_rate": 2.8896720927178033e-06, "loss": 0.4183, "step": 21310 }, { "epoch": 0.759249692716034, "grad_norm": 1.2527185601721993, "learning_rate": 2.8888607392575073e-06, "loss": 0.3981, "step": 21311 }, { "epoch": 0.7592853198425281, "grad_norm": 1.175511504053461, "learning_rate": 2.888049480487003e-06, "loss": 0.4069, "step": 21312 }, { "epoch": 0.7593209469690222, "grad_norm": 1.158266608712189, "learning_rate": 2.887238316417088e-06, "loss": 0.349, "step": 21313 }, { "epoch": 0.7593565740955164, "grad_norm": 2.039269251986077, "learning_rate": 2.8864272470585696e-06, "loss": 0.5382, "step": 21314 }, { "epoch": 0.7593922012220105, "grad_norm": 1.6145117336212211, "learning_rate": 2.8856162724222414e-06, "loss": 0.515, "step": 21315 }, { "epoch": 0.7594278283485045, "grad_norm": 1.4279898407789597, "learning_rate": 2.884805392518908e-06, "loss": 0.5072, "step": 21316 }, { "epoch": 0.7594634554749987, "grad_norm": 1.715279159058463, "learning_rate": 2.8839946073593604e-06, "loss": 0.5162, "step": 21317 }, { "epoch": 0.7594990826014928, "grad_norm": 1.1970359797086654, "learning_rate": 2.8831839169543998e-06, "loss": 0.3692, "step": 21318 }, { "epoch": 0.7595347097279869, "grad_norm": 1.2236277767609383, "learning_rate": 2.8823733213148187e-06, "loss": 0.3978, "step": 21319 }, { "epoch": 0.759570336854481, "grad_norm": 1.43873693479557, "learning_rate": 2.8815628204514113e-06, "loss": 0.4664, "step": 21320 }, { "epoch": 0.7596059639809751, "grad_norm": 1.1958664995434631, "learning_rate": 2.8807524143749743e-06, "loss": 0.4314, "step": 21321 }, { "epoch": 0.7596415911074692, "grad_norm": 1.5387274445742405, "learning_rate": 2.8799421030962917e-06, "loss": 0.5803, "step": 21322 }, { "epoch": 0.7596772182339634, "grad_norm": 1.2428225699406488, "learning_rate": 2.8791318866261597e-06, "loss": 0.437, "step": 21323 }, { "epoch": 0.7597128453604575, "grad_norm": 1.4218181058144221, "learning_rate": 2.8783217649753616e-06, "loss": 0.5515, "step": 21324 }, { "epoch": 0.7597484724869515, "grad_norm": 1.8357425927606534, "learning_rate": 2.8775117381546856e-06, "loss": 0.5617, "step": 21325 }, { "epoch": 0.7597840996134457, "grad_norm": 1.2679856516414179, "learning_rate": 2.87670180617492e-06, "loss": 0.4175, "step": 21326 }, { "epoch": 0.7598197267399398, "grad_norm": 1.2704002816751971, "learning_rate": 2.8758919690468514e-06, "loss": 0.3904, "step": 21327 }, { "epoch": 0.7598553538664339, "grad_norm": 1.3071634974973187, "learning_rate": 2.8750822267812574e-06, "loss": 0.4252, "step": 21328 }, { "epoch": 0.7598909809929281, "grad_norm": 1.1460306105711795, "learning_rate": 2.8742725793889235e-06, "loss": 0.3851, "step": 21329 }, { "epoch": 0.7599266081194221, "grad_norm": 1.4511163142264047, "learning_rate": 2.873463026880634e-06, "loss": 0.5178, "step": 21330 }, { "epoch": 0.7599622352459162, "grad_norm": 1.5991255830965603, "learning_rate": 2.872653569267163e-06, "loss": 0.5353, "step": 21331 }, { "epoch": 0.7599978623724104, "grad_norm": 1.457483298944048, "learning_rate": 2.8718442065592944e-06, "loss": 0.5022, "step": 21332 }, { "epoch": 0.7600334894989045, "grad_norm": 2.05783954129972, "learning_rate": 2.871034938767798e-06, "loss": 0.5469, "step": 21333 }, { "epoch": 0.7600691166253986, "grad_norm": 1.121973619748695, "learning_rate": 2.870225765903456e-06, "loss": 0.3872, "step": 21334 }, { "epoch": 0.7601047437518927, "grad_norm": 1.566364133683849, "learning_rate": 2.869416687977041e-06, "loss": 0.4556, "step": 21335 }, { "epoch": 0.7601403708783868, "grad_norm": 1.3595080887398192, "learning_rate": 2.868607704999329e-06, "loss": 0.4965, "step": 21336 }, { "epoch": 0.7601759980048809, "grad_norm": 1.232827767261657, "learning_rate": 2.8677988169810876e-06, "loss": 0.5132, "step": 21337 }, { "epoch": 0.7602116251313751, "grad_norm": 1.5901579217635362, "learning_rate": 2.866990023933093e-06, "loss": 0.5282, "step": 21338 }, { "epoch": 0.7602472522578692, "grad_norm": 1.0368034155167187, "learning_rate": 2.8661813258661087e-06, "loss": 0.3853, "step": 21339 }, { "epoch": 0.7602828793843632, "grad_norm": 1.1116005018428854, "learning_rate": 2.865372722790907e-06, "loss": 0.388, "step": 21340 }, { "epoch": 0.7603185065108574, "grad_norm": 1.7618647137566281, "learning_rate": 2.8645642147182527e-06, "loss": 0.5, "step": 21341 }, { "epoch": 0.7603541336373515, "grad_norm": 1.5134888134640991, "learning_rate": 2.863755801658915e-06, "loss": 0.5177, "step": 21342 }, { "epoch": 0.7603897607638456, "grad_norm": 1.7551337283912591, "learning_rate": 2.862947483623659e-06, "loss": 0.5063, "step": 21343 }, { "epoch": 0.7604253878903398, "grad_norm": 2.5916798606515585, "learning_rate": 2.8621392606232424e-06, "loss": 0.4804, "step": 21344 }, { "epoch": 0.7604610150168338, "grad_norm": 1.5372490699414172, "learning_rate": 2.8613311326684347e-06, "loss": 0.5049, "step": 21345 }, { "epoch": 0.7604966421433279, "grad_norm": 1.3543335651041772, "learning_rate": 2.8605230997699885e-06, "loss": 0.4959, "step": 21346 }, { "epoch": 0.7605322692698221, "grad_norm": 1.5514648068296057, "learning_rate": 2.859715161938671e-06, "loss": 0.5494, "step": 21347 }, { "epoch": 0.7605678963963162, "grad_norm": 1.4309007860301854, "learning_rate": 2.8589073191852303e-06, "loss": 0.5538, "step": 21348 }, { "epoch": 0.7606035235228102, "grad_norm": 1.47525813971556, "learning_rate": 2.858099571520436e-06, "loss": 0.5137, "step": 21349 }, { "epoch": 0.7606391506493044, "grad_norm": 1.7767034751800361, "learning_rate": 2.8572919189550354e-06, "loss": 0.4319, "step": 21350 }, { "epoch": 0.7606747777757985, "grad_norm": 1.1719184664543765, "learning_rate": 2.8564843614997884e-06, "loss": 0.4026, "step": 21351 }, { "epoch": 0.7607104049022926, "grad_norm": 1.3791447264438688, "learning_rate": 2.8556768991654414e-06, "loss": 0.4136, "step": 21352 }, { "epoch": 0.7607460320287868, "grad_norm": 1.4248881614590074, "learning_rate": 2.8548695319627504e-06, "loss": 0.4642, "step": 21353 }, { "epoch": 0.7607816591552808, "grad_norm": 1.6682559644398869, "learning_rate": 2.854062259902469e-06, "loss": 0.4994, "step": 21354 }, { "epoch": 0.7608172862817749, "grad_norm": 1.507156075541955, "learning_rate": 2.8532550829953397e-06, "loss": 0.4884, "step": 21355 }, { "epoch": 0.7608529134082691, "grad_norm": 1.6284635389700464, "learning_rate": 2.852448001252115e-06, "loss": 0.5786, "step": 21356 }, { "epoch": 0.7608885405347632, "grad_norm": 1.8183749146039496, "learning_rate": 2.8516410146835406e-06, "loss": 0.4921, "step": 21357 }, { "epoch": 0.7609241676612573, "grad_norm": 1.74231433607515, "learning_rate": 2.8508341233003656e-06, "loss": 0.5368, "step": 21358 }, { "epoch": 0.7609597947877514, "grad_norm": 1.229173225177737, "learning_rate": 2.8500273271133283e-06, "loss": 0.4304, "step": 21359 }, { "epoch": 0.7609954219142455, "grad_norm": 1.2929294689626998, "learning_rate": 2.849220626133178e-06, "loss": 0.3931, "step": 21360 }, { "epoch": 0.7610310490407396, "grad_norm": 1.1780929129046924, "learning_rate": 2.848414020370649e-06, "loss": 0.4368, "step": 21361 }, { "epoch": 0.7610666761672338, "grad_norm": 1.2790044411225248, "learning_rate": 2.8476075098364875e-06, "loss": 0.4111, "step": 21362 }, { "epoch": 0.7611023032937279, "grad_norm": 1.13199566008805, "learning_rate": 2.84680109454143e-06, "loss": 0.4131, "step": 21363 }, { "epoch": 0.7611379304202219, "grad_norm": 1.1809242212557851, "learning_rate": 2.845994774496216e-06, "loss": 0.4121, "step": 21364 }, { "epoch": 0.7611735575467161, "grad_norm": 1.186706092070618, "learning_rate": 2.845188549711585e-06, "loss": 0.4014, "step": 21365 }, { "epoch": 0.7612091846732102, "grad_norm": 1.1025676391821901, "learning_rate": 2.8443824201982663e-06, "loss": 0.4377, "step": 21366 }, { "epoch": 0.7612448117997043, "grad_norm": 1.1934529795950082, "learning_rate": 2.8435763859670007e-06, "loss": 0.4167, "step": 21367 }, { "epoch": 0.7612804389261985, "grad_norm": 1.6094653011414366, "learning_rate": 2.8427704470285143e-06, "loss": 0.4822, "step": 21368 }, { "epoch": 0.7613160660526925, "grad_norm": 1.219030838242285, "learning_rate": 2.8419646033935443e-06, "loss": 0.4245, "step": 21369 }, { "epoch": 0.7613516931791866, "grad_norm": 1.7855307281827428, "learning_rate": 2.8411588550728164e-06, "loss": 0.4836, "step": 21370 }, { "epoch": 0.7613873203056808, "grad_norm": 1.252224106083625, "learning_rate": 2.840353202077063e-06, "loss": 0.3688, "step": 21371 }, { "epoch": 0.7614229474321749, "grad_norm": 1.5772198973147595, "learning_rate": 2.83954764441701e-06, "loss": 0.4904, "step": 21372 }, { "epoch": 0.7614585745586689, "grad_norm": 1.634046290302994, "learning_rate": 2.838742182103388e-06, "loss": 0.4923, "step": 21373 }, { "epoch": 0.7614942016851631, "grad_norm": 1.5517466332937322, "learning_rate": 2.8379368151469155e-06, "loss": 0.5243, "step": 21374 }, { "epoch": 0.7615298288116572, "grad_norm": 1.2308396186860289, "learning_rate": 2.8371315435583214e-06, "loss": 0.4456, "step": 21375 }, { "epoch": 0.7615654559381513, "grad_norm": 1.2277274872260389, "learning_rate": 2.83632636734833e-06, "loss": 0.4086, "step": 21376 }, { "epoch": 0.7616010830646455, "grad_norm": 1.1804828583308078, "learning_rate": 2.835521286527657e-06, "loss": 0.4206, "step": 21377 }, { "epoch": 0.7616367101911395, "grad_norm": 1.5152941896588585, "learning_rate": 2.8347163011070255e-06, "loss": 0.4791, "step": 21378 }, { "epoch": 0.7616723373176336, "grad_norm": 1.2971745417050886, "learning_rate": 2.8339114110971556e-06, "loss": 0.4563, "step": 21379 }, { "epoch": 0.7617079644441278, "grad_norm": 1.7025924870752018, "learning_rate": 2.833106616508766e-06, "loss": 0.5311, "step": 21380 }, { "epoch": 0.7617435915706219, "grad_norm": 2.1265513043209516, "learning_rate": 2.832301917352569e-06, "loss": 0.5387, "step": 21381 }, { "epoch": 0.761779218697116, "grad_norm": 1.5214025295023232, "learning_rate": 2.8314973136392844e-06, "loss": 0.5144, "step": 21382 }, { "epoch": 0.7618148458236101, "grad_norm": 1.5292177103877822, "learning_rate": 2.830692805379621e-06, "loss": 0.5604, "step": 21383 }, { "epoch": 0.7618504729501042, "grad_norm": 1.4696609139781895, "learning_rate": 2.829888392584297e-06, "loss": 0.5839, "step": 21384 }, { "epoch": 0.7618861000765983, "grad_norm": 1.1200714484066765, "learning_rate": 2.8290840752640135e-06, "loss": 0.4103, "step": 21385 }, { "epoch": 0.7619217272030925, "grad_norm": 1.5963787737983686, "learning_rate": 2.828279853429492e-06, "loss": 0.5352, "step": 21386 }, { "epoch": 0.7619573543295866, "grad_norm": 1.9904234915357866, "learning_rate": 2.82747572709144e-06, "loss": 0.4761, "step": 21387 }, { "epoch": 0.7619929814560806, "grad_norm": 1.2949968581200457, "learning_rate": 2.8266716962605577e-06, "loss": 0.5244, "step": 21388 }, { "epoch": 0.7620286085825748, "grad_norm": 1.1828895033843463, "learning_rate": 2.8258677609475595e-06, "loss": 0.4221, "step": 21389 }, { "epoch": 0.7620642357090689, "grad_norm": 1.1326872168603228, "learning_rate": 2.825063921163144e-06, "loss": 0.4166, "step": 21390 }, { "epoch": 0.762099862835563, "grad_norm": 1.8043599870604272, "learning_rate": 2.82426017691802e-06, "loss": 0.3788, "step": 21391 }, { "epoch": 0.7621354899620572, "grad_norm": 1.1251156157892137, "learning_rate": 2.823456528222883e-06, "loss": 0.4226, "step": 21392 }, { "epoch": 0.7621711170885512, "grad_norm": 1.167603866242926, "learning_rate": 2.8226529750884403e-06, "loss": 0.4132, "step": 21393 }, { "epoch": 0.7622067442150453, "grad_norm": 1.3176641960709923, "learning_rate": 2.821849517525389e-06, "loss": 0.5076, "step": 21394 }, { "epoch": 0.7622423713415395, "grad_norm": 1.6189656073476457, "learning_rate": 2.8210461555444313e-06, "loss": 0.5261, "step": 21395 }, { "epoch": 0.7622779984680336, "grad_norm": 1.4697333064574951, "learning_rate": 2.8202428891562595e-06, "loss": 0.5427, "step": 21396 }, { "epoch": 0.7623136255945276, "grad_norm": 1.9158469217785874, "learning_rate": 2.819439718371575e-06, "loss": 0.5412, "step": 21397 }, { "epoch": 0.7623492527210218, "grad_norm": 1.4462965543010815, "learning_rate": 2.8186366432010658e-06, "loss": 0.4522, "step": 21398 }, { "epoch": 0.7623848798475159, "grad_norm": 1.2066274273030007, "learning_rate": 2.817833663655429e-06, "loss": 0.3728, "step": 21399 }, { "epoch": 0.76242050697401, "grad_norm": 1.5894130099221095, "learning_rate": 2.817030779745358e-06, "loss": 0.4154, "step": 21400 }, { "epoch": 0.7624561341005042, "grad_norm": 1.07176322688374, "learning_rate": 2.8162279914815427e-06, "loss": 0.4283, "step": 21401 }, { "epoch": 0.7624917612269982, "grad_norm": 1.6701335839542273, "learning_rate": 2.8154252988746757e-06, "loss": 0.5822, "step": 21402 }, { "epoch": 0.7625273883534923, "grad_norm": 1.2412315001309622, "learning_rate": 2.814622701935439e-06, "loss": 0.4086, "step": 21403 }, { "epoch": 0.7625630154799865, "grad_norm": 1.1213492491967054, "learning_rate": 2.8138202006745276e-06, "loss": 0.3985, "step": 21404 }, { "epoch": 0.7625986426064806, "grad_norm": 1.0905723167809078, "learning_rate": 2.8130177951026183e-06, "loss": 0.4045, "step": 21405 }, { "epoch": 0.7626342697329747, "grad_norm": 1.3930144522947785, "learning_rate": 2.812215485230405e-06, "loss": 0.4064, "step": 21406 }, { "epoch": 0.7626698968594688, "grad_norm": 1.1480351499262507, "learning_rate": 2.8114132710685625e-06, "loss": 0.3825, "step": 21407 }, { "epoch": 0.7627055239859629, "grad_norm": 1.0586584036586844, "learning_rate": 2.810611152627777e-06, "loss": 0.3992, "step": 21408 }, { "epoch": 0.762741151112457, "grad_norm": 1.1374019136733848, "learning_rate": 2.809809129918729e-06, "loss": 0.4235, "step": 21409 }, { "epoch": 0.7627767782389512, "grad_norm": 1.3081335030789263, "learning_rate": 2.8090072029520987e-06, "loss": 0.5036, "step": 21410 }, { "epoch": 0.7628124053654453, "grad_norm": 1.6272032371045224, "learning_rate": 2.808205371738566e-06, "loss": 0.5126, "step": 21411 }, { "epoch": 0.7628480324919393, "grad_norm": 1.1413678410632886, "learning_rate": 2.8074036362888036e-06, "loss": 0.4322, "step": 21412 }, { "epoch": 0.7628836596184335, "grad_norm": 1.4727916856952978, "learning_rate": 2.8066019966134907e-06, "loss": 0.5859, "step": 21413 }, { "epoch": 0.7629192867449276, "grad_norm": 1.8481148635628983, "learning_rate": 2.8058004527232974e-06, "loss": 0.5938, "step": 21414 }, { "epoch": 0.7629549138714217, "grad_norm": 1.197117221046349, "learning_rate": 2.8049990046289e-06, "loss": 0.419, "step": 21415 }, { "epoch": 0.7629905409979159, "grad_norm": 1.7105284024225385, "learning_rate": 2.80419765234097e-06, "loss": 0.4129, "step": 21416 }, { "epoch": 0.7630261681244099, "grad_norm": 1.5483128977502965, "learning_rate": 2.8033963958701805e-06, "loss": 0.5378, "step": 21417 }, { "epoch": 0.763061795250904, "grad_norm": 1.143859201450746, "learning_rate": 2.8025952352271957e-06, "loss": 0.3341, "step": 21418 }, { "epoch": 0.7630974223773982, "grad_norm": 1.1858457786554635, "learning_rate": 2.801794170422688e-06, "loss": 0.3873, "step": 21419 }, { "epoch": 0.7631330495038923, "grad_norm": 1.430043883725752, "learning_rate": 2.8009932014673193e-06, "loss": 0.5491, "step": 21420 }, { "epoch": 0.7631686766303863, "grad_norm": 1.1876634781197857, "learning_rate": 2.800192328371759e-06, "loss": 0.4429, "step": 21421 }, { "epoch": 0.7632043037568805, "grad_norm": 1.4276882249400502, "learning_rate": 2.799391551146672e-06, "loss": 0.3763, "step": 21422 }, { "epoch": 0.7632399308833746, "grad_norm": 1.1421463373564338, "learning_rate": 2.798590869802714e-06, "loss": 0.4124, "step": 21423 }, { "epoch": 0.7632755580098687, "grad_norm": 1.0453097005201217, "learning_rate": 2.7977902843505578e-06, "loss": 0.3958, "step": 21424 }, { "epoch": 0.7633111851363629, "grad_norm": 1.4890469297602553, "learning_rate": 2.796989794800855e-06, "loss": 0.4826, "step": 21425 }, { "epoch": 0.7633468122628569, "grad_norm": 1.114270763024617, "learning_rate": 2.796189401164271e-06, "loss": 0.4187, "step": 21426 }, { "epoch": 0.763382439389351, "grad_norm": 1.500153791256887, "learning_rate": 2.795389103451457e-06, "loss": 0.5115, "step": 21427 }, { "epoch": 0.7634180665158452, "grad_norm": 1.532261296083087, "learning_rate": 2.7945889016730767e-06, "loss": 0.4136, "step": 21428 }, { "epoch": 0.7634536936423393, "grad_norm": 1.308447481282634, "learning_rate": 2.7937887958397768e-06, "loss": 0.5077, "step": 21429 }, { "epoch": 0.7634893207688334, "grad_norm": 1.8934774126051832, "learning_rate": 2.792988785962217e-06, "loss": 0.4733, "step": 21430 }, { "epoch": 0.7635249478953275, "grad_norm": 1.0876658082845676, "learning_rate": 2.7921888720510494e-06, "loss": 0.3564, "step": 21431 }, { "epoch": 0.7635605750218216, "grad_norm": 1.5296161529340726, "learning_rate": 2.791389054116924e-06, "loss": 0.4875, "step": 21432 }, { "epoch": 0.7635962021483157, "grad_norm": 1.2517638394531831, "learning_rate": 2.7905893321704945e-06, "loss": 0.4107, "step": 21433 }, { "epoch": 0.7636318292748099, "grad_norm": 1.6376506624679903, "learning_rate": 2.7897897062224055e-06, "loss": 0.5994, "step": 21434 }, { "epoch": 0.763667456401304, "grad_norm": 1.2584650589461326, "learning_rate": 2.7889901762833082e-06, "loss": 0.3677, "step": 21435 }, { "epoch": 0.763703083527798, "grad_norm": 1.1710359999629345, "learning_rate": 2.788190742363844e-06, "loss": 0.4367, "step": 21436 }, { "epoch": 0.7637387106542922, "grad_norm": 1.6090359511987768, "learning_rate": 2.7873914044746607e-06, "loss": 0.5567, "step": 21437 }, { "epoch": 0.7637743377807863, "grad_norm": 1.513309429273692, "learning_rate": 2.7865921626264023e-06, "loss": 0.5026, "step": 21438 }, { "epoch": 0.7638099649072804, "grad_norm": 1.6588071460608436, "learning_rate": 2.7857930168297155e-06, "loss": 0.5781, "step": 21439 }, { "epoch": 0.7638455920337746, "grad_norm": 1.1652620286851465, "learning_rate": 2.7849939670952318e-06, "loss": 0.4201, "step": 21440 }, { "epoch": 0.7638812191602686, "grad_norm": 1.1615494641946256, "learning_rate": 2.7841950134336004e-06, "loss": 0.3984, "step": 21441 }, { "epoch": 0.7639168462867627, "grad_norm": 1.2000552713196238, "learning_rate": 2.7833961558554533e-06, "loss": 0.4212, "step": 21442 }, { "epoch": 0.7639524734132569, "grad_norm": 1.2648277012685767, "learning_rate": 2.7825973943714336e-06, "loss": 0.3895, "step": 21443 }, { "epoch": 0.763988100539751, "grad_norm": 1.4607769238995825, "learning_rate": 2.781798728992171e-06, "loss": 0.5206, "step": 21444 }, { "epoch": 0.764023727666245, "grad_norm": 1.3620441081133197, "learning_rate": 2.781000159728303e-06, "loss": 0.4935, "step": 21445 }, { "epoch": 0.7640593547927392, "grad_norm": 1.4641936996389295, "learning_rate": 2.7802016865904646e-06, "loss": 0.4837, "step": 21446 }, { "epoch": 0.7640949819192333, "grad_norm": 1.1750102645277298, "learning_rate": 2.779403309589286e-06, "loss": 0.4471, "step": 21447 }, { "epoch": 0.7641306090457274, "grad_norm": 1.10914929804339, "learning_rate": 2.778605028735403e-06, "loss": 0.3917, "step": 21448 }, { "epoch": 0.7641662361722216, "grad_norm": 1.1515929411791492, "learning_rate": 2.777806844039438e-06, "loss": 0.4232, "step": 21449 }, { "epoch": 0.7642018632987156, "grad_norm": 1.1419174864082624, "learning_rate": 2.7770087555120263e-06, "loss": 0.4152, "step": 21450 }, { "epoch": 0.7642374904252097, "grad_norm": 1.0959753178922946, "learning_rate": 2.776210763163788e-06, "loss": 0.4184, "step": 21451 }, { "epoch": 0.7642731175517039, "grad_norm": 1.110608999616459, "learning_rate": 2.7754128670053536e-06, "loss": 0.4548, "step": 21452 }, { "epoch": 0.764308744678198, "grad_norm": 1.4771785356464202, "learning_rate": 2.774615067047346e-06, "loss": 0.5605, "step": 21453 }, { "epoch": 0.7643443718046921, "grad_norm": 1.4829511391619312, "learning_rate": 2.7738173633003918e-06, "loss": 0.5128, "step": 21454 }, { "epoch": 0.7643799989311862, "grad_norm": 1.1523675331573726, "learning_rate": 2.773019755775107e-06, "loss": 0.4155, "step": 21455 }, { "epoch": 0.7644156260576803, "grad_norm": 1.5377226450339285, "learning_rate": 2.7722222444821156e-06, "loss": 0.6149, "step": 21456 }, { "epoch": 0.7644512531841744, "grad_norm": 1.6373065228662058, "learning_rate": 2.771424829432041e-06, "loss": 0.5277, "step": 21457 }, { "epoch": 0.7644868803106686, "grad_norm": 1.0821364680420906, "learning_rate": 2.770627510635493e-06, "loss": 0.3655, "step": 21458 }, { "epoch": 0.7645225074371627, "grad_norm": 2.7066076286801657, "learning_rate": 2.7698302881030957e-06, "loss": 0.5571, "step": 21459 }, { "epoch": 0.7645581345636567, "grad_norm": 1.1662107643816284, "learning_rate": 2.769033161845457e-06, "loss": 0.4259, "step": 21460 }, { "epoch": 0.7645937616901509, "grad_norm": 1.1860214871361896, "learning_rate": 2.768236131873201e-06, "loss": 0.4255, "step": 21461 }, { "epoch": 0.764629388816645, "grad_norm": 1.268651738909, "learning_rate": 2.7674391981969327e-06, "loss": 0.4303, "step": 21462 }, { "epoch": 0.7646650159431391, "grad_norm": 1.1735108916474652, "learning_rate": 2.76664236082727e-06, "loss": 0.4408, "step": 21463 }, { "epoch": 0.7647006430696333, "grad_norm": 1.1765232563449093, "learning_rate": 2.7658456197748165e-06, "loss": 0.4346, "step": 21464 }, { "epoch": 0.7647362701961273, "grad_norm": 1.474751548712698, "learning_rate": 2.7650489750501885e-06, "loss": 0.5747, "step": 21465 }, { "epoch": 0.7647718973226214, "grad_norm": 1.0584026192975013, "learning_rate": 2.764252426663987e-06, "loss": 0.415, "step": 21466 }, { "epoch": 0.7648075244491156, "grad_norm": 1.4211513780580616, "learning_rate": 2.763455974626822e-06, "loss": 0.4236, "step": 21467 }, { "epoch": 0.7648431515756097, "grad_norm": 1.6411437488971135, "learning_rate": 2.7626596189492983e-06, "loss": 0.5029, "step": 21468 }, { "epoch": 0.7648787787021037, "grad_norm": 1.1211044912241588, "learning_rate": 2.761863359642021e-06, "loss": 0.3813, "step": 21469 }, { "epoch": 0.7649144058285979, "grad_norm": 1.6193617123799213, "learning_rate": 2.7610671967155943e-06, "loss": 0.5785, "step": 21470 }, { "epoch": 0.764950032955092, "grad_norm": 1.6136326806195296, "learning_rate": 2.7602711301806164e-06, "loss": 0.4569, "step": 21471 }, { "epoch": 0.7649856600815861, "grad_norm": 1.256373004121953, "learning_rate": 2.7594751600476898e-06, "loss": 0.5139, "step": 21472 }, { "epoch": 0.7650212872080803, "grad_norm": 1.5881453108127388, "learning_rate": 2.7586792863274104e-06, "loss": 0.5644, "step": 21473 }, { "epoch": 0.7650569143345743, "grad_norm": 1.2060291396387337, "learning_rate": 2.757883509030377e-06, "loss": 0.4087, "step": 21474 }, { "epoch": 0.7650925414610684, "grad_norm": 1.1913238893412281, "learning_rate": 2.7570878281671875e-06, "loss": 0.4048, "step": 21475 }, { "epoch": 0.7651281685875626, "grad_norm": 1.3165483779754887, "learning_rate": 2.7562922437484384e-06, "loss": 0.5425, "step": 21476 }, { "epoch": 0.7651637957140567, "grad_norm": 1.1632263728659067, "learning_rate": 2.755496755784718e-06, "loss": 0.444, "step": 21477 }, { "epoch": 0.7651994228405508, "grad_norm": 1.1395089572421093, "learning_rate": 2.754701364286625e-06, "loss": 0.3938, "step": 21478 }, { "epoch": 0.765235049967045, "grad_norm": 1.0909799784092424, "learning_rate": 2.7539060692647447e-06, "loss": 0.374, "step": 21479 }, { "epoch": 0.765270677093539, "grad_norm": 1.2672094625545622, "learning_rate": 2.7531108707296695e-06, "loss": 0.4344, "step": 21480 }, { "epoch": 0.7653063042200331, "grad_norm": 1.0895853538808602, "learning_rate": 2.752315768691992e-06, "loss": 0.3786, "step": 21481 }, { "epoch": 0.7653419313465273, "grad_norm": 1.5558490889712562, "learning_rate": 2.751520763162292e-06, "loss": 0.5059, "step": 21482 }, { "epoch": 0.7653775584730214, "grad_norm": 1.3088690030228516, "learning_rate": 2.7507258541511604e-06, "loss": 0.453, "step": 21483 }, { "epoch": 0.7654131855995154, "grad_norm": 1.159088608427199, "learning_rate": 2.7499310416691806e-06, "loss": 0.4535, "step": 21484 }, { "epoch": 0.7654488127260096, "grad_norm": 1.5067663392453183, "learning_rate": 2.749136325726939e-06, "loss": 0.4653, "step": 21485 }, { "epoch": 0.7654844398525037, "grad_norm": 1.153283039707779, "learning_rate": 2.7483417063350124e-06, "loss": 0.4, "step": 21486 }, { "epoch": 0.7655200669789978, "grad_norm": 1.0767514352851586, "learning_rate": 2.747547183503988e-06, "loss": 0.4261, "step": 21487 }, { "epoch": 0.765555694105492, "grad_norm": 1.265416578095691, "learning_rate": 2.7467527572444386e-06, "loss": 0.3943, "step": 21488 }, { "epoch": 0.765591321231986, "grad_norm": 1.504834425095821, "learning_rate": 2.745958427566946e-06, "loss": 0.5791, "step": 21489 }, { "epoch": 0.7656269483584801, "grad_norm": 1.2171392589643428, "learning_rate": 2.7451641944820874e-06, "loss": 0.4061, "step": 21490 }, { "epoch": 0.7656625754849743, "grad_norm": 1.5461183849936049, "learning_rate": 2.7443700580004383e-06, "loss": 0.5773, "step": 21491 }, { "epoch": 0.7656982026114684, "grad_norm": 3.043920008810186, "learning_rate": 2.7435760181325767e-06, "loss": 0.5268, "step": 21492 }, { "epoch": 0.7657338297379624, "grad_norm": 1.3715509838857574, "learning_rate": 2.742782074889069e-06, "loss": 0.5212, "step": 21493 }, { "epoch": 0.7657694568644566, "grad_norm": 1.3301047781821087, "learning_rate": 2.741988228280493e-06, "loss": 0.4451, "step": 21494 }, { "epoch": 0.7658050839909507, "grad_norm": 1.243292826316313, "learning_rate": 2.741194478317414e-06, "loss": 0.4854, "step": 21495 }, { "epoch": 0.7658407111174448, "grad_norm": 1.011154902256393, "learning_rate": 2.7404008250104075e-06, "loss": 0.3781, "step": 21496 }, { "epoch": 0.765876338243939, "grad_norm": 1.1849024872027014, "learning_rate": 2.739607268370035e-06, "loss": 0.4117, "step": 21497 }, { "epoch": 0.765911965370433, "grad_norm": 1.1121840669123144, "learning_rate": 2.738813808406866e-06, "loss": 0.3959, "step": 21498 }, { "epoch": 0.7659475924969271, "grad_norm": 1.2557274619900056, "learning_rate": 2.7380204451314674e-06, "loss": 0.3866, "step": 21499 }, { "epoch": 0.7659832196234213, "grad_norm": 2.019774286732689, "learning_rate": 2.7372271785544046e-06, "loss": 0.5037, "step": 21500 }, { "epoch": 0.7660188467499154, "grad_norm": 1.649504602359647, "learning_rate": 2.736434008686235e-06, "loss": 0.58, "step": 21501 }, { "epoch": 0.7660544738764095, "grad_norm": 1.1643003804974497, "learning_rate": 2.7356409355375246e-06, "loss": 0.3643, "step": 21502 }, { "epoch": 0.7660901010029036, "grad_norm": 1.7739030262741133, "learning_rate": 2.734847959118835e-06, "loss": 0.4809, "step": 21503 }, { "epoch": 0.7661257281293977, "grad_norm": 1.3418106877761815, "learning_rate": 2.7340550794407206e-06, "loss": 0.3851, "step": 21504 }, { "epoch": 0.7661613552558918, "grad_norm": 1.583122617744459, "learning_rate": 2.73326229651374e-06, "loss": 0.4972, "step": 21505 }, { "epoch": 0.766196982382386, "grad_norm": 1.582239546513259, "learning_rate": 2.7324696103484517e-06, "loss": 0.5235, "step": 21506 }, { "epoch": 0.7662326095088801, "grad_norm": 1.1879557015916027, "learning_rate": 2.7316770209554135e-06, "loss": 0.4086, "step": 21507 }, { "epoch": 0.7662682366353741, "grad_norm": 1.152305358852738, "learning_rate": 2.7308845283451725e-06, "loss": 0.441, "step": 21508 }, { "epoch": 0.7663038637618683, "grad_norm": 1.4551707360280475, "learning_rate": 2.7300921325282883e-06, "loss": 0.5706, "step": 21509 }, { "epoch": 0.7663394908883624, "grad_norm": 1.7766425297593949, "learning_rate": 2.729299833515305e-06, "loss": 0.5311, "step": 21510 }, { "epoch": 0.7663751180148565, "grad_norm": 1.117291371960036, "learning_rate": 2.7285076313167815e-06, "loss": 0.412, "step": 21511 }, { "epoch": 0.7664107451413507, "grad_norm": 1.280286626871796, "learning_rate": 2.727715525943253e-06, "loss": 0.455, "step": 21512 }, { "epoch": 0.7664463722678447, "grad_norm": 1.5872044168890584, "learning_rate": 2.7269235174052788e-06, "loss": 0.5286, "step": 21513 }, { "epoch": 0.7664819993943388, "grad_norm": 1.2546955193275324, "learning_rate": 2.726131605713405e-06, "loss": 0.4432, "step": 21514 }, { "epoch": 0.766517626520833, "grad_norm": 1.434222430863424, "learning_rate": 2.72533979087817e-06, "loss": 0.587, "step": 21515 }, { "epoch": 0.7665532536473271, "grad_norm": 1.667097451901086, "learning_rate": 2.7245480729101227e-06, "loss": 0.5261, "step": 21516 }, { "epoch": 0.7665888807738211, "grad_norm": 1.5177831567870879, "learning_rate": 2.723756451819801e-06, "loss": 0.532, "step": 21517 }, { "epoch": 0.7666245079003153, "grad_norm": 1.1922284246705164, "learning_rate": 2.7229649276177504e-06, "loss": 0.4347, "step": 21518 }, { "epoch": 0.7666601350268094, "grad_norm": 1.1415117031269961, "learning_rate": 2.7221735003145054e-06, "loss": 0.4371, "step": 21519 }, { "epoch": 0.7666957621533035, "grad_norm": 1.575036850251923, "learning_rate": 2.7213821699206066e-06, "loss": 0.5908, "step": 21520 }, { "epoch": 0.7667313892797977, "grad_norm": 1.133428940322207, "learning_rate": 2.720590936446592e-06, "loss": 0.3988, "step": 21521 }, { "epoch": 0.7667670164062917, "grad_norm": 1.3549327576041397, "learning_rate": 2.7197997999030002e-06, "loss": 0.4789, "step": 21522 }, { "epoch": 0.7668026435327858, "grad_norm": 1.252268181463492, "learning_rate": 2.719008760300359e-06, "loss": 0.4326, "step": 21523 }, { "epoch": 0.76683827065928, "grad_norm": 1.3258392648224315, "learning_rate": 2.718217817649209e-06, "loss": 0.5506, "step": 21524 }, { "epoch": 0.7668738977857741, "grad_norm": 2.2098600801105466, "learning_rate": 2.7174269719600756e-06, "loss": 0.4944, "step": 21525 }, { "epoch": 0.7669095249122682, "grad_norm": 1.5200670219847623, "learning_rate": 2.7166362232434917e-06, "loss": 0.4834, "step": 21526 }, { "epoch": 0.7669451520387623, "grad_norm": 1.1582107467854073, "learning_rate": 2.7158455715099872e-06, "loss": 0.3508, "step": 21527 }, { "epoch": 0.7669807791652564, "grad_norm": 1.4359375800775016, "learning_rate": 2.7150550167700905e-06, "loss": 0.4715, "step": 21528 }, { "epoch": 0.7670164062917505, "grad_norm": 1.469370277022623, "learning_rate": 2.714264559034331e-06, "loss": 0.4602, "step": 21529 }, { "epoch": 0.7670520334182447, "grad_norm": 2.0585999099072008, "learning_rate": 2.7134741983132283e-06, "loss": 0.5567, "step": 21530 }, { "epoch": 0.7670876605447388, "grad_norm": 1.5530913971537434, "learning_rate": 2.7126839346173128e-06, "loss": 0.5433, "step": 21531 }, { "epoch": 0.7671232876712328, "grad_norm": 1.3990098123874228, "learning_rate": 2.7118937679571013e-06, "loss": 0.5497, "step": 21532 }, { "epoch": 0.767158914797727, "grad_norm": 1.3012006011975044, "learning_rate": 2.7111036983431214e-06, "loss": 0.3928, "step": 21533 }, { "epoch": 0.7671945419242211, "grad_norm": 1.1729235249955694, "learning_rate": 2.7103137257858867e-06, "loss": 0.4026, "step": 21534 }, { "epoch": 0.7672301690507152, "grad_norm": 1.9210036053892072, "learning_rate": 2.709523850295921e-06, "loss": 0.5569, "step": 21535 }, { "epoch": 0.7672657961772094, "grad_norm": 1.1497618952682422, "learning_rate": 2.7087340718837396e-06, "loss": 0.3926, "step": 21536 }, { "epoch": 0.7673014233037034, "grad_norm": 1.6212293765015187, "learning_rate": 2.707944390559861e-06, "loss": 0.5259, "step": 21537 }, { "epoch": 0.7673370504301975, "grad_norm": 1.1422357897571611, "learning_rate": 2.707154806334803e-06, "loss": 0.4175, "step": 21538 }, { "epoch": 0.7673726775566917, "grad_norm": 1.1920783542313533, "learning_rate": 2.706365319219073e-06, "loss": 0.4273, "step": 21539 }, { "epoch": 0.7674083046831858, "grad_norm": 1.2172639723566498, "learning_rate": 2.7055759292231896e-06, "loss": 0.4497, "step": 21540 }, { "epoch": 0.7674439318096798, "grad_norm": 1.4559486046607126, "learning_rate": 2.704786636357659e-06, "loss": 0.5001, "step": 21541 }, { "epoch": 0.767479558936174, "grad_norm": 1.7126850857953841, "learning_rate": 2.7039974406329927e-06, "loss": 0.4578, "step": 21542 }, { "epoch": 0.7675151860626681, "grad_norm": 1.191241658892241, "learning_rate": 2.7032083420597e-06, "loss": 0.415, "step": 21543 }, { "epoch": 0.7675508131891622, "grad_norm": 1.217022799464884, "learning_rate": 2.702419340648291e-06, "loss": 0.3881, "step": 21544 }, { "epoch": 0.7675864403156564, "grad_norm": 1.639673184598055, "learning_rate": 2.7016304364092672e-06, "loss": 0.5071, "step": 21545 }, { "epoch": 0.7676220674421504, "grad_norm": 1.1971553223191918, "learning_rate": 2.700841629353137e-06, "loss": 0.4159, "step": 21546 }, { "epoch": 0.7676576945686445, "grad_norm": 1.3832943202946046, "learning_rate": 2.7000529194904e-06, "loss": 0.4688, "step": 21547 }, { "epoch": 0.7676933216951387, "grad_norm": 1.2157597485810003, "learning_rate": 2.6992643068315615e-06, "loss": 0.393, "step": 21548 }, { "epoch": 0.7677289488216328, "grad_norm": 1.2512154905367603, "learning_rate": 2.6984757913871208e-06, "loss": 0.3858, "step": 21549 }, { "epoch": 0.7677645759481269, "grad_norm": 1.5028343056113351, "learning_rate": 2.6976873731675777e-06, "loss": 0.4843, "step": 21550 }, { "epoch": 0.767800203074621, "grad_norm": 1.77296372435771, "learning_rate": 2.696899052183436e-06, "loss": 0.4178, "step": 21551 }, { "epoch": 0.7678358302011151, "grad_norm": 1.1660352389776516, "learning_rate": 2.6961108284451842e-06, "loss": 0.4253, "step": 21552 }, { "epoch": 0.7678714573276092, "grad_norm": 1.2716560816494318, "learning_rate": 2.695322701963324e-06, "loss": 0.4408, "step": 21553 }, { "epoch": 0.7679070844541034, "grad_norm": 1.1780906934989555, "learning_rate": 2.6945346727483458e-06, "loss": 0.3799, "step": 21554 }, { "epoch": 0.7679427115805975, "grad_norm": 1.0909233680983903, "learning_rate": 2.6937467408107476e-06, "loss": 0.3866, "step": 21555 }, { "epoch": 0.7679783387070915, "grad_norm": 1.443275372425425, "learning_rate": 2.692958906161015e-06, "loss": 0.4996, "step": 21556 }, { "epoch": 0.7680139658335857, "grad_norm": 1.4814276863686517, "learning_rate": 2.692171168809642e-06, "loss": 0.5743, "step": 21557 }, { "epoch": 0.7680495929600798, "grad_norm": 1.1955746831376248, "learning_rate": 2.6913835287671177e-06, "loss": 0.3816, "step": 21558 }, { "epoch": 0.7680852200865739, "grad_norm": 1.4746471565152934, "learning_rate": 2.6905959860439303e-06, "loss": 0.4877, "step": 21559 }, { "epoch": 0.7681208472130681, "grad_norm": 1.3990026493355128, "learning_rate": 2.689808540650568e-06, "loss": 0.5019, "step": 21560 }, { "epoch": 0.7681564743395621, "grad_norm": 1.158586845593426, "learning_rate": 2.689021192597513e-06, "loss": 0.5267, "step": 21561 }, { "epoch": 0.7681921014660562, "grad_norm": 1.320145328184931, "learning_rate": 2.6882339418952542e-06, "loss": 0.5247, "step": 21562 }, { "epoch": 0.7682277285925504, "grad_norm": 1.0977685965231554, "learning_rate": 2.6874467885542664e-06, "loss": 0.3852, "step": 21563 }, { "epoch": 0.7682633557190445, "grad_norm": 1.6009006639096062, "learning_rate": 2.6866597325850373e-06, "loss": 0.5029, "step": 21564 }, { "epoch": 0.7682989828455385, "grad_norm": 1.8374057093350054, "learning_rate": 2.685872773998045e-06, "loss": 0.4149, "step": 21565 }, { "epoch": 0.7683346099720327, "grad_norm": 1.5972006401980088, "learning_rate": 2.685085912803772e-06, "loss": 0.5012, "step": 21566 }, { "epoch": 0.7683702370985268, "grad_norm": 1.2806520775254424, "learning_rate": 2.6842991490126903e-06, "loss": 0.3737, "step": 21567 }, { "epoch": 0.7684058642250209, "grad_norm": 1.1901531857244398, "learning_rate": 2.683512482635281e-06, "loss": 0.402, "step": 21568 }, { "epoch": 0.7684414913515151, "grad_norm": 1.2706818411141758, "learning_rate": 2.682725913682014e-06, "loss": 0.4268, "step": 21569 }, { "epoch": 0.7684771184780091, "grad_norm": 1.1406974651783892, "learning_rate": 2.681939442163369e-06, "loss": 0.4058, "step": 21570 }, { "epoch": 0.7685127456045032, "grad_norm": 1.3534600904544283, "learning_rate": 2.6811530680898133e-06, "loss": 0.468, "step": 21571 }, { "epoch": 0.7685483727309974, "grad_norm": 1.8263349088587681, "learning_rate": 2.680366791471819e-06, "loss": 0.5331, "step": 21572 }, { "epoch": 0.7685839998574915, "grad_norm": 1.1226097457222486, "learning_rate": 2.6795806123198575e-06, "loss": 0.3887, "step": 21573 }, { "epoch": 0.7686196269839856, "grad_norm": 1.454396114869252, "learning_rate": 2.6787945306443963e-06, "loss": 0.5507, "step": 21574 }, { "epoch": 0.7686552541104797, "grad_norm": 1.930491181790085, "learning_rate": 2.678008546455907e-06, "loss": 0.4751, "step": 21575 }, { "epoch": 0.7686908812369738, "grad_norm": 1.9077557665548937, "learning_rate": 2.6772226597648476e-06, "loss": 0.5581, "step": 21576 }, { "epoch": 0.7687265083634679, "grad_norm": 1.4001226904452373, "learning_rate": 2.67643687058169e-06, "loss": 0.4703, "step": 21577 }, { "epoch": 0.7687621354899621, "grad_norm": 2.170489684166731, "learning_rate": 2.6756511789168926e-06, "loss": 0.4729, "step": 21578 }, { "epoch": 0.7687977626164562, "grad_norm": 1.0420108259369352, "learning_rate": 2.674865584780918e-06, "loss": 0.3835, "step": 21579 }, { "epoch": 0.7688333897429502, "grad_norm": 1.663353140437724, "learning_rate": 2.6740800881842277e-06, "loss": 0.4989, "step": 21580 }, { "epoch": 0.7688690168694444, "grad_norm": 1.5494045161652252, "learning_rate": 2.673294689137286e-06, "loss": 0.5978, "step": 21581 }, { "epoch": 0.7689046439959385, "grad_norm": 1.1816674964063065, "learning_rate": 2.6725093876505426e-06, "loss": 0.4085, "step": 21582 }, { "epoch": 0.7689402711224326, "grad_norm": 1.6297015745937447, "learning_rate": 2.671724183734459e-06, "loss": 0.5959, "step": 21583 }, { "epoch": 0.7689758982489268, "grad_norm": 1.6289027646324437, "learning_rate": 2.6709390773994915e-06, "loss": 0.4941, "step": 21584 }, { "epoch": 0.7690115253754208, "grad_norm": 1.588365683519425, "learning_rate": 2.670154068656091e-06, "loss": 0.511, "step": 21585 }, { "epoch": 0.7690471525019149, "grad_norm": 1.638972875696349, "learning_rate": 2.6693691575147153e-06, "loss": 0.5621, "step": 21586 }, { "epoch": 0.7690827796284091, "grad_norm": 1.201850237097028, "learning_rate": 2.6685843439858072e-06, "loss": 0.4049, "step": 21587 }, { "epoch": 0.7691184067549032, "grad_norm": 1.0190029686038897, "learning_rate": 2.667799628079829e-06, "loss": 0.3951, "step": 21588 }, { "epoch": 0.7691540338813972, "grad_norm": 1.4189687424518198, "learning_rate": 2.66701500980722e-06, "loss": 0.4494, "step": 21589 }, { "epoch": 0.7691896610078914, "grad_norm": 1.1767471613449225, "learning_rate": 2.666230489178435e-06, "loss": 0.4113, "step": 21590 }, { "epoch": 0.7692252881343855, "grad_norm": 1.1766326259255167, "learning_rate": 2.665446066203914e-06, "loss": 0.4195, "step": 21591 }, { "epoch": 0.7692609152608796, "grad_norm": 1.7531080606592429, "learning_rate": 2.6646617408941087e-06, "loss": 0.5754, "step": 21592 }, { "epoch": 0.7692965423873738, "grad_norm": 1.8315359732306624, "learning_rate": 2.663877513259455e-06, "loss": 0.6256, "step": 21593 }, { "epoch": 0.7693321695138678, "grad_norm": 1.7968912464408158, "learning_rate": 2.663093383310401e-06, "loss": 0.5127, "step": 21594 }, { "epoch": 0.7693677966403619, "grad_norm": 1.2029471207564573, "learning_rate": 2.6623093510573874e-06, "loss": 0.3728, "step": 21595 }, { "epoch": 0.7694034237668561, "grad_norm": 1.5666365622681224, "learning_rate": 2.661525416510853e-06, "loss": 0.5335, "step": 21596 }, { "epoch": 0.7694390508933502, "grad_norm": 1.6257176448928388, "learning_rate": 2.6607415796812385e-06, "loss": 0.5782, "step": 21597 }, { "epoch": 0.7694746780198443, "grad_norm": 1.2672422945859758, "learning_rate": 2.6599578405789783e-06, "loss": 0.4325, "step": 21598 }, { "epoch": 0.7695103051463384, "grad_norm": 1.6178144027791186, "learning_rate": 2.6591741992145128e-06, "loss": 0.5118, "step": 21599 }, { "epoch": 0.7695459322728325, "grad_norm": 1.7526893553998233, "learning_rate": 2.65839065559827e-06, "loss": 0.512, "step": 21600 }, { "epoch": 0.7695815593993266, "grad_norm": 2.3321642502426907, "learning_rate": 2.657607209740688e-06, "loss": 0.5464, "step": 21601 }, { "epoch": 0.7696171865258208, "grad_norm": 1.2442796907420348, "learning_rate": 2.656823861652197e-06, "loss": 0.4263, "step": 21602 }, { "epoch": 0.7696528136523149, "grad_norm": 1.1980397396389468, "learning_rate": 2.6560406113432323e-06, "loss": 0.3969, "step": 21603 }, { "epoch": 0.7696884407788089, "grad_norm": 1.1291839033010052, "learning_rate": 2.6552574588242175e-06, "loss": 0.3803, "step": 21604 }, { "epoch": 0.7697240679053031, "grad_norm": 1.6739104496117783, "learning_rate": 2.6544744041055857e-06, "loss": 0.5098, "step": 21605 }, { "epoch": 0.7697596950317972, "grad_norm": 1.209481885904988, "learning_rate": 2.653691447197758e-06, "loss": 0.4135, "step": 21606 }, { "epoch": 0.7697953221582913, "grad_norm": 1.8834284192582231, "learning_rate": 2.6529085881111636e-06, "loss": 0.5095, "step": 21607 }, { "epoch": 0.7698309492847855, "grad_norm": 1.1933814569437278, "learning_rate": 2.652125826856231e-06, "loss": 0.457, "step": 21608 }, { "epoch": 0.7698665764112795, "grad_norm": 1.159909449376772, "learning_rate": 2.651343163443375e-06, "loss": 0.3966, "step": 21609 }, { "epoch": 0.7699022035377736, "grad_norm": 1.5550117866545745, "learning_rate": 2.6505605978830206e-06, "loss": 0.5328, "step": 21610 }, { "epoch": 0.7699378306642678, "grad_norm": 1.6576865156670553, "learning_rate": 2.6497781301855897e-06, "loss": 0.5177, "step": 21611 }, { "epoch": 0.7699734577907619, "grad_norm": 1.2308464500019314, "learning_rate": 2.6489957603615025e-06, "loss": 0.4291, "step": 21612 }, { "epoch": 0.770009084917256, "grad_norm": 1.1054598638746636, "learning_rate": 2.648213488421173e-06, "loss": 0.4268, "step": 21613 }, { "epoch": 0.7700447120437501, "grad_norm": 1.4102135967949658, "learning_rate": 2.647431314375022e-06, "loss": 0.6406, "step": 21614 }, { "epoch": 0.7700803391702442, "grad_norm": 2.5102366376960092, "learning_rate": 2.6466492382334587e-06, "loss": 0.4904, "step": 21615 }, { "epoch": 0.7701159662967383, "grad_norm": 1.2017190759907301, "learning_rate": 2.6458672600069014e-06, "loss": 0.4241, "step": 21616 }, { "epoch": 0.7701515934232325, "grad_norm": 1.480327026223971, "learning_rate": 2.645085379705761e-06, "loss": 0.5357, "step": 21617 }, { "epoch": 0.7701872205497265, "grad_norm": 1.1885658758635074, "learning_rate": 2.6443035973404497e-06, "loss": 0.4293, "step": 21618 }, { "epoch": 0.7702228476762206, "grad_norm": 1.3002106594144445, "learning_rate": 2.6435219129213796e-06, "loss": 0.4484, "step": 21619 }, { "epoch": 0.7702584748027148, "grad_norm": 1.9747720164022702, "learning_rate": 2.6427403264589556e-06, "loss": 0.4819, "step": 21620 }, { "epoch": 0.7702941019292089, "grad_norm": 1.2816521871798434, "learning_rate": 2.6419588379635884e-06, "loss": 0.4332, "step": 21621 }, { "epoch": 0.770329729055703, "grad_norm": 1.1710235910673708, "learning_rate": 2.6411774474456798e-06, "loss": 0.4395, "step": 21622 }, { "epoch": 0.7703653561821971, "grad_norm": 1.2001536635228645, "learning_rate": 2.6403961549156396e-06, "loss": 0.4348, "step": 21623 }, { "epoch": 0.7704009833086912, "grad_norm": 1.193095579999238, "learning_rate": 2.639614960383864e-06, "loss": 0.4088, "step": 21624 }, { "epoch": 0.7704366104351853, "grad_norm": 1.1649277655132133, "learning_rate": 2.638833863860765e-06, "loss": 0.3776, "step": 21625 }, { "epoch": 0.7704722375616795, "grad_norm": 1.4059321627265644, "learning_rate": 2.638052865356734e-06, "loss": 0.5043, "step": 21626 }, { "epoch": 0.7705078646881736, "grad_norm": 1.4805329868816013, "learning_rate": 2.6372719648821785e-06, "loss": 0.5148, "step": 21627 }, { "epoch": 0.7705434918146676, "grad_norm": 1.075707960050639, "learning_rate": 2.636491162447491e-06, "loss": 0.4126, "step": 21628 }, { "epoch": 0.7705791189411618, "grad_norm": 1.1999720951011579, "learning_rate": 2.6357104580630697e-06, "loss": 0.3845, "step": 21629 }, { "epoch": 0.7706147460676559, "grad_norm": 1.2101362665739561, "learning_rate": 2.634929851739314e-06, "loss": 0.4291, "step": 21630 }, { "epoch": 0.77065037319415, "grad_norm": 1.3205532101764095, "learning_rate": 2.634149343486612e-06, "loss": 0.3911, "step": 21631 }, { "epoch": 0.7706860003206442, "grad_norm": 1.1638833125605035, "learning_rate": 2.63336893331536e-06, "loss": 0.4432, "step": 21632 }, { "epoch": 0.7707216274471382, "grad_norm": 1.2495918093801441, "learning_rate": 2.6325886212359496e-06, "loss": 0.4197, "step": 21633 }, { "epoch": 0.7707572545736323, "grad_norm": 1.0923428109977331, "learning_rate": 2.6318084072587745e-06, "loss": 0.4167, "step": 21634 }, { "epoch": 0.7707928817001265, "grad_norm": 1.199956492299311, "learning_rate": 2.6310282913942177e-06, "loss": 0.4043, "step": 21635 }, { "epoch": 0.7708285088266206, "grad_norm": 1.167748845970819, "learning_rate": 2.6302482736526723e-06, "loss": 0.4573, "step": 21636 }, { "epoch": 0.7708641359531146, "grad_norm": 1.1569275401506285, "learning_rate": 2.62946835404452e-06, "loss": 0.4761, "step": 21637 }, { "epoch": 0.7708997630796088, "grad_norm": 1.5070333192421321, "learning_rate": 2.628688532580148e-06, "loss": 0.4912, "step": 21638 }, { "epoch": 0.7709353902061029, "grad_norm": 1.194057621644597, "learning_rate": 2.62790880926994e-06, "loss": 0.3984, "step": 21639 }, { "epoch": 0.770971017332597, "grad_norm": 1.938026777206164, "learning_rate": 2.6271291841242786e-06, "loss": 0.462, "step": 21640 }, { "epoch": 0.7710066444590912, "grad_norm": 1.5297686638629555, "learning_rate": 2.6263496571535495e-06, "loss": 0.6308, "step": 21641 }, { "epoch": 0.7710422715855852, "grad_norm": 1.154890803546079, "learning_rate": 2.625570228368125e-06, "loss": 0.4094, "step": 21642 }, { "epoch": 0.7710778987120793, "grad_norm": 1.6005150685290364, "learning_rate": 2.624790897778391e-06, "loss": 0.5669, "step": 21643 }, { "epoch": 0.7711135258385735, "grad_norm": 1.6523009173080425, "learning_rate": 2.6240116653947166e-06, "loss": 0.4959, "step": 21644 }, { "epoch": 0.7711491529650676, "grad_norm": 1.4466639954835692, "learning_rate": 2.6232325312274865e-06, "loss": 0.4582, "step": 21645 }, { "epoch": 0.7711847800915617, "grad_norm": 1.184587594739101, "learning_rate": 2.622453495287068e-06, "loss": 0.4228, "step": 21646 }, { "epoch": 0.7712204072180558, "grad_norm": 1.6792591784428372, "learning_rate": 2.621674557583839e-06, "loss": 0.4817, "step": 21647 }, { "epoch": 0.7712560343445499, "grad_norm": 1.08752054283561, "learning_rate": 2.6208957181281693e-06, "loss": 0.416, "step": 21648 }, { "epoch": 0.771291661471044, "grad_norm": 1.7012570467047337, "learning_rate": 2.6201169769304334e-06, "loss": 0.6093, "step": 21649 }, { "epoch": 0.7713272885975382, "grad_norm": 1.1737836488785613, "learning_rate": 2.619338334000996e-06, "loss": 0.4157, "step": 21650 }, { "epoch": 0.7713629157240323, "grad_norm": 1.811649386916491, "learning_rate": 2.6185597893502303e-06, "loss": 0.4849, "step": 21651 }, { "epoch": 0.7713985428505263, "grad_norm": 1.167643114019438, "learning_rate": 2.617781342988497e-06, "loss": 0.417, "step": 21652 }, { "epoch": 0.7714341699770205, "grad_norm": 1.4571224882618532, "learning_rate": 2.6170029949261655e-06, "loss": 0.5916, "step": 21653 }, { "epoch": 0.7714697971035146, "grad_norm": 1.3162006088007447, "learning_rate": 2.616224745173599e-06, "loss": 0.4114, "step": 21654 }, { "epoch": 0.7715054242300087, "grad_norm": 2.1366173413927125, "learning_rate": 2.615446593741161e-06, "loss": 0.555, "step": 21655 }, { "epoch": 0.7715410513565029, "grad_norm": 1.6962121791225033, "learning_rate": 2.614668540639216e-06, "loss": 0.4792, "step": 21656 }, { "epoch": 0.7715766784829969, "grad_norm": 1.0867660126736638, "learning_rate": 2.6138905858781183e-06, "loss": 0.3819, "step": 21657 }, { "epoch": 0.771612305609491, "grad_norm": 1.355652398797795, "learning_rate": 2.6131127294682334e-06, "loss": 0.4321, "step": 21658 }, { "epoch": 0.7716479327359852, "grad_norm": 1.2638244211815293, "learning_rate": 2.612334971419913e-06, "loss": 0.4155, "step": 21659 }, { "epoch": 0.7716835598624793, "grad_norm": 1.170982553108446, "learning_rate": 2.6115573117435187e-06, "loss": 0.3901, "step": 21660 }, { "epoch": 0.7717191869889733, "grad_norm": 1.6533972004508195, "learning_rate": 2.6107797504494e-06, "loss": 0.5063, "step": 21661 }, { "epoch": 0.7717548141154675, "grad_norm": 1.5649190244484352, "learning_rate": 2.6100022875479138e-06, "loss": 0.519, "step": 21662 }, { "epoch": 0.7717904412419616, "grad_norm": 1.3548877763800369, "learning_rate": 2.6092249230494128e-06, "loss": 0.4346, "step": 21663 }, { "epoch": 0.7718260683684558, "grad_norm": 1.1749200318272828, "learning_rate": 2.608447656964247e-06, "loss": 0.3932, "step": 21664 }, { "epoch": 0.7718616954949499, "grad_norm": 1.187274152958268, "learning_rate": 2.6076704893027703e-06, "loss": 0.4054, "step": 21665 }, { "epoch": 0.771897322621444, "grad_norm": 1.6988171901776934, "learning_rate": 2.6068934200753248e-06, "loss": 0.4359, "step": 21666 }, { "epoch": 0.7719329497479381, "grad_norm": 1.1792931469101608, "learning_rate": 2.6061164492922642e-06, "loss": 0.3678, "step": 21667 }, { "epoch": 0.7719685768744322, "grad_norm": 1.4927590631593832, "learning_rate": 2.605339576963929e-06, "loss": 0.5101, "step": 21668 }, { "epoch": 0.7720042040009263, "grad_norm": 1.218127812305156, "learning_rate": 2.6045628031006642e-06, "loss": 0.4444, "step": 21669 }, { "epoch": 0.7720398311274205, "grad_norm": 2.4199679843610387, "learning_rate": 2.603786127712815e-06, "loss": 0.574, "step": 21670 }, { "epoch": 0.7720754582539145, "grad_norm": 1.5521242604591978, "learning_rate": 2.603009550810728e-06, "loss": 0.5247, "step": 21671 }, { "epoch": 0.7721110853804086, "grad_norm": 1.1698953065481552, "learning_rate": 2.602233072404734e-06, "loss": 0.3874, "step": 21672 }, { "epoch": 0.7721467125069028, "grad_norm": 1.060784739751212, "learning_rate": 2.601456692505181e-06, "loss": 0.3564, "step": 21673 }, { "epoch": 0.7721823396333969, "grad_norm": 1.7767942978322275, "learning_rate": 2.6006804111224006e-06, "loss": 0.4827, "step": 21674 }, { "epoch": 0.772217966759891, "grad_norm": 1.440886594639486, "learning_rate": 2.599904228266732e-06, "loss": 0.4809, "step": 21675 }, { "epoch": 0.7722535938863851, "grad_norm": 1.1230531272111712, "learning_rate": 2.5991281439485117e-06, "loss": 0.3826, "step": 21676 }, { "epoch": 0.7722892210128792, "grad_norm": 1.2320008634779698, "learning_rate": 2.5983521581780724e-06, "loss": 0.4472, "step": 21677 }, { "epoch": 0.7723248481393733, "grad_norm": 1.0475271874318488, "learning_rate": 2.5975762709657506e-06, "loss": 0.4019, "step": 21678 }, { "epoch": 0.7723604752658675, "grad_norm": 1.1461022476758795, "learning_rate": 2.596800482321872e-06, "loss": 0.4812, "step": 21679 }, { "epoch": 0.7723961023923616, "grad_norm": 1.1715379036087141, "learning_rate": 2.5960247922567716e-06, "loss": 0.38, "step": 21680 }, { "epoch": 0.7724317295188556, "grad_norm": 1.0557398483848734, "learning_rate": 2.595249200780774e-06, "loss": 0.3931, "step": 21681 }, { "epoch": 0.7724673566453498, "grad_norm": 1.4757636783604584, "learning_rate": 2.5944737079042114e-06, "loss": 0.5418, "step": 21682 }, { "epoch": 0.7725029837718439, "grad_norm": 1.4036691058459585, "learning_rate": 2.593698313637405e-06, "loss": 0.4523, "step": 21683 }, { "epoch": 0.772538610898338, "grad_norm": 1.3263383056912161, "learning_rate": 2.5929230179906827e-06, "loss": 0.4701, "step": 21684 }, { "epoch": 0.7725742380248322, "grad_norm": 1.0087366669509834, "learning_rate": 2.5921478209743665e-06, "loss": 0.4059, "step": 21685 }, { "epoch": 0.7726098651513262, "grad_norm": 2.053234344749883, "learning_rate": 2.5913727225987805e-06, "loss": 0.4669, "step": 21686 }, { "epoch": 0.7726454922778203, "grad_norm": 1.3769913104356397, "learning_rate": 2.5905977228742483e-06, "loss": 0.3879, "step": 21687 }, { "epoch": 0.7726811194043145, "grad_norm": 1.5173142374897168, "learning_rate": 2.5898228218110834e-06, "loss": 0.6014, "step": 21688 }, { "epoch": 0.7727167465308086, "grad_norm": 1.2811310748484797, "learning_rate": 2.58904801941961e-06, "loss": 0.4765, "step": 21689 }, { "epoch": 0.7727523736573026, "grad_norm": 1.1530459059366451, "learning_rate": 2.588273315710138e-06, "loss": 0.4186, "step": 21690 }, { "epoch": 0.7727880007837968, "grad_norm": 1.7523566553989165, "learning_rate": 2.587498710692988e-06, "loss": 0.4827, "step": 21691 }, { "epoch": 0.7728236279102909, "grad_norm": 1.1269304314134585, "learning_rate": 2.5867242043784745e-06, "loss": 0.4015, "step": 21692 }, { "epoch": 0.772859255036785, "grad_norm": 1.415560967245766, "learning_rate": 2.585949796776912e-06, "loss": 0.5686, "step": 21693 }, { "epoch": 0.7728948821632792, "grad_norm": 1.6612234785339917, "learning_rate": 2.585175487898607e-06, "loss": 0.5085, "step": 21694 }, { "epoch": 0.7729305092897732, "grad_norm": 1.8709751873061222, "learning_rate": 2.5844012777538763e-06, "loss": 0.5603, "step": 21695 }, { "epoch": 0.7729661364162673, "grad_norm": 1.2621422421661141, "learning_rate": 2.583627166353022e-06, "loss": 0.4365, "step": 21696 }, { "epoch": 0.7730017635427615, "grad_norm": 1.343964716271236, "learning_rate": 2.5828531537063595e-06, "loss": 0.4242, "step": 21697 }, { "epoch": 0.7730373906692556, "grad_norm": 1.1539192172617752, "learning_rate": 2.582079239824188e-06, "loss": 0.416, "step": 21698 }, { "epoch": 0.7730730177957497, "grad_norm": 1.290907502075983, "learning_rate": 2.581305424716817e-06, "loss": 0.3878, "step": 21699 }, { "epoch": 0.7731086449222438, "grad_norm": 1.6026783982285009, "learning_rate": 2.580531708394549e-06, "loss": 0.5318, "step": 21700 }, { "epoch": 0.7731442720487379, "grad_norm": 1.1549903917034525, "learning_rate": 2.579758090867688e-06, "loss": 0.3764, "step": 21701 }, { "epoch": 0.773179899175232, "grad_norm": 1.6872307538523812, "learning_rate": 2.578984572146537e-06, "loss": 0.4153, "step": 21702 }, { "epoch": 0.7732155263017262, "grad_norm": 1.0362120025309176, "learning_rate": 2.57821115224139e-06, "loss": 0.4067, "step": 21703 }, { "epoch": 0.7732511534282203, "grad_norm": 1.9402100939959694, "learning_rate": 2.5774378311625526e-06, "loss": 0.4866, "step": 21704 }, { "epoch": 0.7732867805547143, "grad_norm": 1.1064930060844627, "learning_rate": 2.576664608920315e-06, "loss": 0.3631, "step": 21705 }, { "epoch": 0.7733224076812085, "grad_norm": 1.5923138457422894, "learning_rate": 2.575891485524977e-06, "loss": 0.5886, "step": 21706 }, { "epoch": 0.7733580348077026, "grad_norm": 1.578655749343617, "learning_rate": 2.575118460986833e-06, "loss": 0.5306, "step": 21707 }, { "epoch": 0.7733936619341967, "grad_norm": 1.1122153970513853, "learning_rate": 2.5743455353161794e-06, "loss": 0.3963, "step": 21708 }, { "epoch": 0.7734292890606909, "grad_norm": 1.100583538496388, "learning_rate": 2.573572708523302e-06, "loss": 0.4258, "step": 21709 }, { "epoch": 0.7734649161871849, "grad_norm": 1.526556040506416, "learning_rate": 2.5727999806184947e-06, "loss": 0.4781, "step": 21710 }, { "epoch": 0.773500543313679, "grad_norm": 1.2668620509676658, "learning_rate": 2.5720273516120507e-06, "loss": 0.3779, "step": 21711 }, { "epoch": 0.7735361704401732, "grad_norm": 1.3769270885424665, "learning_rate": 2.571254821514251e-06, "loss": 0.5032, "step": 21712 }, { "epoch": 0.7735717975666673, "grad_norm": 1.15166568955024, "learning_rate": 2.570482390335385e-06, "loss": 0.4239, "step": 21713 }, { "epoch": 0.7736074246931613, "grad_norm": 1.1862648069736896, "learning_rate": 2.5697100580857392e-06, "loss": 0.4545, "step": 21714 }, { "epoch": 0.7736430518196555, "grad_norm": 1.4118736184335317, "learning_rate": 2.5689378247756015e-06, "loss": 0.5245, "step": 21715 }, { "epoch": 0.7736786789461496, "grad_norm": 1.2279783860197224, "learning_rate": 2.568165690415246e-06, "loss": 0.4264, "step": 21716 }, { "epoch": 0.7737143060726437, "grad_norm": 2.1511095082334446, "learning_rate": 2.567393655014964e-06, "loss": 0.5166, "step": 21717 }, { "epoch": 0.7737499331991379, "grad_norm": 1.7440691304758176, "learning_rate": 2.566621718585026e-06, "loss": 0.5418, "step": 21718 }, { "epoch": 0.773785560325632, "grad_norm": 1.1450342285977018, "learning_rate": 2.565849881135719e-06, "loss": 0.3924, "step": 21719 }, { "epoch": 0.773821187452126, "grad_norm": 1.4692288803088613, "learning_rate": 2.565078142677313e-06, "loss": 0.382, "step": 21720 }, { "epoch": 0.7738568145786202, "grad_norm": 1.0985608129292974, "learning_rate": 2.56430650322009e-06, "loss": 0.3894, "step": 21721 }, { "epoch": 0.7738924417051143, "grad_norm": 1.1319333916929584, "learning_rate": 2.563534962774321e-06, "loss": 0.3739, "step": 21722 }, { "epoch": 0.7739280688316084, "grad_norm": 1.893310314473046, "learning_rate": 2.5627635213502832e-06, "loss": 0.5823, "step": 21723 }, { "epoch": 0.7739636959581025, "grad_norm": 1.1617747737606627, "learning_rate": 2.56199217895825e-06, "loss": 0.3943, "step": 21724 }, { "epoch": 0.7739993230845966, "grad_norm": 1.232065548950433, "learning_rate": 2.5612209356084873e-06, "loss": 0.3826, "step": 21725 }, { "epoch": 0.7740349502110907, "grad_norm": 1.4833346339394529, "learning_rate": 2.56044979131127e-06, "loss": 0.5219, "step": 21726 }, { "epoch": 0.7740705773375849, "grad_norm": 1.40603457401747, "learning_rate": 2.55967874607686e-06, "loss": 0.5027, "step": 21727 }, { "epoch": 0.774106204464079, "grad_norm": 1.6683470735293797, "learning_rate": 2.558907799915529e-06, "loss": 0.4805, "step": 21728 }, { "epoch": 0.774141831590573, "grad_norm": 1.1766598227717595, "learning_rate": 2.558136952837541e-06, "loss": 0.4252, "step": 21729 }, { "epoch": 0.7741774587170672, "grad_norm": 1.0718447426307873, "learning_rate": 2.5573662048531643e-06, "loss": 0.4346, "step": 21730 }, { "epoch": 0.7742130858435613, "grad_norm": 1.3957066498856865, "learning_rate": 2.556595555972655e-06, "loss": 0.4522, "step": 21731 }, { "epoch": 0.7742487129700554, "grad_norm": 1.1548490725330243, "learning_rate": 2.5558250062062826e-06, "loss": 0.3779, "step": 21732 }, { "epoch": 0.7742843400965496, "grad_norm": 1.1322939138547126, "learning_rate": 2.555054555564299e-06, "loss": 0.4301, "step": 21733 }, { "epoch": 0.7743199672230436, "grad_norm": 1.6394116836804133, "learning_rate": 2.5542842040569695e-06, "loss": 0.6039, "step": 21734 }, { "epoch": 0.7743555943495377, "grad_norm": 1.8578761908713852, "learning_rate": 2.553513951694552e-06, "loss": 0.4863, "step": 21735 }, { "epoch": 0.7743912214760319, "grad_norm": 1.0823391560721674, "learning_rate": 2.552743798487298e-06, "loss": 0.4032, "step": 21736 }, { "epoch": 0.774426848602526, "grad_norm": 1.1103181795178008, "learning_rate": 2.551973744445466e-06, "loss": 0.3472, "step": 21737 }, { "epoch": 0.77446247572902, "grad_norm": 2.033467348859881, "learning_rate": 2.5512037895793086e-06, "loss": 0.6322, "step": 21738 }, { "epoch": 0.7744981028555142, "grad_norm": 1.2948439008878512, "learning_rate": 2.5504339338990835e-06, "loss": 0.4302, "step": 21739 }, { "epoch": 0.7745337299820083, "grad_norm": 1.6086511411330966, "learning_rate": 2.5496641774150333e-06, "loss": 0.5181, "step": 21740 }, { "epoch": 0.7745693571085024, "grad_norm": 1.1575705511225542, "learning_rate": 2.548894520137416e-06, "loss": 0.4375, "step": 21741 }, { "epoch": 0.7746049842349966, "grad_norm": 1.8508395659404502, "learning_rate": 2.5481249620764726e-06, "loss": 0.5323, "step": 21742 }, { "epoch": 0.7746406113614906, "grad_norm": 2.1699381613836493, "learning_rate": 2.5473555032424534e-06, "loss": 0.5787, "step": 21743 }, { "epoch": 0.7746762384879847, "grad_norm": 1.1993687842706147, "learning_rate": 2.5465861436456053e-06, "loss": 0.4204, "step": 21744 }, { "epoch": 0.7747118656144789, "grad_norm": 1.6928550141001517, "learning_rate": 2.545816883296173e-06, "loss": 0.4702, "step": 21745 }, { "epoch": 0.774747492740973, "grad_norm": 1.0623047975780027, "learning_rate": 2.545047722204401e-06, "loss": 0.3814, "step": 21746 }, { "epoch": 0.7747831198674671, "grad_norm": 1.6134300035922173, "learning_rate": 2.544278660380527e-06, "loss": 0.5797, "step": 21747 }, { "epoch": 0.7748187469939612, "grad_norm": 1.6259303735615178, "learning_rate": 2.543509697834796e-06, "loss": 0.5059, "step": 21748 }, { "epoch": 0.7748543741204553, "grad_norm": 1.3245459963112596, "learning_rate": 2.542740834577443e-06, "loss": 0.4129, "step": 21749 }, { "epoch": 0.7748900012469494, "grad_norm": 1.917249018227299, "learning_rate": 2.541972070618711e-06, "loss": 0.4526, "step": 21750 }, { "epoch": 0.7749256283734436, "grad_norm": 1.494863857293577, "learning_rate": 2.5412034059688273e-06, "loss": 0.537, "step": 21751 }, { "epoch": 0.7749612554999377, "grad_norm": 1.656421228360197, "learning_rate": 2.54043484063804e-06, "loss": 0.5279, "step": 21752 }, { "epoch": 0.7749968826264317, "grad_norm": 1.2525364306230666, "learning_rate": 2.5396663746365734e-06, "loss": 0.4193, "step": 21753 }, { "epoch": 0.7750325097529259, "grad_norm": 1.40494401774426, "learning_rate": 2.5388980079746673e-06, "loss": 0.4793, "step": 21754 }, { "epoch": 0.77506813687942, "grad_norm": 1.6267830155175254, "learning_rate": 2.5381297406625447e-06, "loss": 0.4727, "step": 21755 }, { "epoch": 0.7751037640059141, "grad_norm": 1.1314236968419351, "learning_rate": 2.5373615727104417e-06, "loss": 0.3918, "step": 21756 }, { "epoch": 0.7751393911324083, "grad_norm": 1.2215495051247207, "learning_rate": 2.5365935041285882e-06, "loss": 0.4475, "step": 21757 }, { "epoch": 0.7751750182589023, "grad_norm": 1.1842191117314647, "learning_rate": 2.5358255349272055e-06, "loss": 0.3932, "step": 21758 }, { "epoch": 0.7752106453853964, "grad_norm": 1.4952286844088254, "learning_rate": 2.5350576651165227e-06, "loss": 0.4385, "step": 21759 }, { "epoch": 0.7752462725118906, "grad_norm": 1.595781466333299, "learning_rate": 2.534289894706765e-06, "loss": 0.5133, "step": 21760 }, { "epoch": 0.7752818996383847, "grad_norm": 1.2097890641811029, "learning_rate": 2.533522223708158e-06, "loss": 0.4244, "step": 21761 }, { "epoch": 0.7753175267648788, "grad_norm": 1.1323475765431112, "learning_rate": 2.5327546521309186e-06, "loss": 0.3875, "step": 21762 }, { "epoch": 0.7753531538913729, "grad_norm": 1.1534644425890932, "learning_rate": 2.5319871799852736e-06, "loss": 0.3759, "step": 21763 }, { "epoch": 0.775388781017867, "grad_norm": 1.0413397846324641, "learning_rate": 2.5312198072814364e-06, "loss": 0.3986, "step": 21764 }, { "epoch": 0.7754244081443611, "grad_norm": 1.2759257994080466, "learning_rate": 2.5304525340296272e-06, "loss": 0.4666, "step": 21765 }, { "epoch": 0.7754600352708553, "grad_norm": 1.2818418105295204, "learning_rate": 2.529685360240064e-06, "loss": 0.4199, "step": 21766 }, { "epoch": 0.7754956623973494, "grad_norm": 1.7926577279498763, "learning_rate": 2.528918285922961e-06, "loss": 0.4958, "step": 21767 }, { "epoch": 0.7755312895238434, "grad_norm": 2.0679499806992525, "learning_rate": 2.528151311088537e-06, "loss": 0.487, "step": 21768 }, { "epoch": 0.7755669166503376, "grad_norm": 1.455293776653555, "learning_rate": 2.5273844357469968e-06, "loss": 0.526, "step": 21769 }, { "epoch": 0.7756025437768317, "grad_norm": 1.5424263157028868, "learning_rate": 2.526617659908559e-06, "loss": 0.5749, "step": 21770 }, { "epoch": 0.7756381709033258, "grad_norm": 1.1551442325795176, "learning_rate": 2.525850983583428e-06, "loss": 0.4102, "step": 21771 }, { "epoch": 0.77567379802982, "grad_norm": 1.142115674960841, "learning_rate": 2.525084406781818e-06, "loss": 0.4088, "step": 21772 }, { "epoch": 0.775709425156314, "grad_norm": 1.179686070097942, "learning_rate": 2.5243179295139307e-06, "loss": 0.4141, "step": 21773 }, { "epoch": 0.7757450522828081, "grad_norm": 1.6133810033581575, "learning_rate": 2.5235515517899768e-06, "loss": 0.4585, "step": 21774 }, { "epoch": 0.7757806794093023, "grad_norm": 1.6896843679552427, "learning_rate": 2.5227852736201573e-06, "loss": 0.542, "step": 21775 }, { "epoch": 0.7758163065357964, "grad_norm": 1.393486793624346, "learning_rate": 2.522019095014683e-06, "loss": 0.503, "step": 21776 }, { "epoch": 0.7758519336622904, "grad_norm": 1.3855405221010786, "learning_rate": 2.5212530159837477e-06, "loss": 0.5169, "step": 21777 }, { "epoch": 0.7758875607887846, "grad_norm": 1.2024141086352769, "learning_rate": 2.5204870365375587e-06, "loss": 0.3963, "step": 21778 }, { "epoch": 0.7759231879152787, "grad_norm": 1.5816192977697374, "learning_rate": 2.51972115668631e-06, "loss": 0.553, "step": 21779 }, { "epoch": 0.7759588150417728, "grad_norm": 1.608241686278838, "learning_rate": 2.518955376440202e-06, "loss": 0.4979, "step": 21780 }, { "epoch": 0.775994442168267, "grad_norm": 1.4181683048608729, "learning_rate": 2.518189695809432e-06, "loss": 0.5097, "step": 21781 }, { "epoch": 0.776030069294761, "grad_norm": 1.3047064840516365, "learning_rate": 2.5174241148041958e-06, "loss": 0.3899, "step": 21782 }, { "epoch": 0.7760656964212551, "grad_norm": 1.7179995550676213, "learning_rate": 2.516658633434691e-06, "loss": 0.4749, "step": 21783 }, { "epoch": 0.7761013235477493, "grad_norm": 1.1562130196346116, "learning_rate": 2.5158932517111033e-06, "loss": 0.4321, "step": 21784 }, { "epoch": 0.7761369506742434, "grad_norm": 1.0342021770894556, "learning_rate": 2.5151279696436315e-06, "loss": 0.34, "step": 21785 }, { "epoch": 0.7761725778007375, "grad_norm": 2.1804607126538986, "learning_rate": 2.514362787242459e-06, "loss": 0.5456, "step": 21786 }, { "epoch": 0.7762082049272316, "grad_norm": 1.6284012025758177, "learning_rate": 2.5135977045177816e-06, "loss": 0.4774, "step": 21787 }, { "epoch": 0.7762438320537257, "grad_norm": 1.2350539685243862, "learning_rate": 2.512832721479781e-06, "loss": 0.466, "step": 21788 }, { "epoch": 0.7762794591802198, "grad_norm": 1.230987533126897, "learning_rate": 2.5120678381386454e-06, "loss": 0.4258, "step": 21789 }, { "epoch": 0.776315086306714, "grad_norm": 1.7383106879271213, "learning_rate": 2.5113030545045604e-06, "loss": 0.5644, "step": 21790 }, { "epoch": 0.776350713433208, "grad_norm": 1.5113514701926012, "learning_rate": 2.5105383705877083e-06, "loss": 0.5082, "step": 21791 }, { "epoch": 0.7763863405597021, "grad_norm": 1.1720685150859382, "learning_rate": 2.5097737863982775e-06, "loss": 0.408, "step": 21792 }, { "epoch": 0.7764219676861963, "grad_norm": 1.8788101547586271, "learning_rate": 2.50900930194644e-06, "loss": 0.5001, "step": 21793 }, { "epoch": 0.7764575948126904, "grad_norm": 1.5449157775962727, "learning_rate": 2.508244917242383e-06, "loss": 0.4993, "step": 21794 }, { "epoch": 0.7764932219391845, "grad_norm": 1.1476589147281584, "learning_rate": 2.5074806322962775e-06, "loss": 0.418, "step": 21795 }, { "epoch": 0.7765288490656787, "grad_norm": 1.2938871446512534, "learning_rate": 2.506716447118305e-06, "loss": 0.4487, "step": 21796 }, { "epoch": 0.7765644761921727, "grad_norm": 1.5811716991972542, "learning_rate": 2.505952361718641e-06, "loss": 0.5504, "step": 21797 }, { "epoch": 0.7766001033186668, "grad_norm": 1.0338113829278313, "learning_rate": 2.5051883761074613e-06, "loss": 0.4035, "step": 21798 }, { "epoch": 0.776635730445161, "grad_norm": 1.7359979227978468, "learning_rate": 2.5044244902949346e-06, "loss": 0.5044, "step": 21799 }, { "epoch": 0.7766713575716551, "grad_norm": 1.4500982561253324, "learning_rate": 2.5036607042912375e-06, "loss": 0.4504, "step": 21800 }, { "epoch": 0.7767069846981491, "grad_norm": 1.0756498409691886, "learning_rate": 2.502897018106535e-06, "loss": 0.4002, "step": 21801 }, { "epoch": 0.7767426118246433, "grad_norm": 1.7469215350979646, "learning_rate": 2.502133431750998e-06, "loss": 0.5289, "step": 21802 }, { "epoch": 0.7767782389511374, "grad_norm": 1.5689294084144938, "learning_rate": 2.5013699452347963e-06, "loss": 0.5142, "step": 21803 }, { "epoch": 0.7768138660776315, "grad_norm": 1.3667399792278099, "learning_rate": 2.500606558568095e-06, "loss": 0.4129, "step": 21804 }, { "epoch": 0.7768494932041257, "grad_norm": 1.0998007433487587, "learning_rate": 2.499843271761061e-06, "loss": 0.4003, "step": 21805 }, { "epoch": 0.7768851203306197, "grad_norm": 1.3565187819232316, "learning_rate": 2.499080084823854e-06, "loss": 0.5197, "step": 21806 }, { "epoch": 0.7769207474571138, "grad_norm": 1.1098125972079083, "learning_rate": 2.498316997766641e-06, "loss": 0.397, "step": 21807 }, { "epoch": 0.776956374583608, "grad_norm": 1.4815648380043125, "learning_rate": 2.497554010599579e-06, "loss": 0.4903, "step": 21808 }, { "epoch": 0.7769920017101021, "grad_norm": 1.4038474811058492, "learning_rate": 2.4967911233328302e-06, "loss": 0.4793, "step": 21809 }, { "epoch": 0.7770276288365962, "grad_norm": 1.82754860042483, "learning_rate": 2.49602833597655e-06, "loss": 0.5626, "step": 21810 }, { "epoch": 0.7770632559630903, "grad_norm": 1.2460831560325898, "learning_rate": 2.495265648540898e-06, "loss": 0.3983, "step": 21811 }, { "epoch": 0.7770988830895844, "grad_norm": 1.6917532390703962, "learning_rate": 2.494503061036029e-06, "loss": 0.4761, "step": 21812 }, { "epoch": 0.7771345102160785, "grad_norm": 1.5261363355238666, "learning_rate": 2.4937405734720964e-06, "loss": 0.5298, "step": 21813 }, { "epoch": 0.7771701373425727, "grad_norm": 1.8330096772412487, "learning_rate": 2.4929781858592595e-06, "loss": 0.4948, "step": 21814 }, { "epoch": 0.7772057644690668, "grad_norm": 2.1016136310372233, "learning_rate": 2.4922158982076615e-06, "loss": 0.5441, "step": 21815 }, { "epoch": 0.7772413915955608, "grad_norm": 1.274970755084242, "learning_rate": 2.4914537105274595e-06, "loss": 0.4128, "step": 21816 }, { "epoch": 0.777277018722055, "grad_norm": 1.1885121448556397, "learning_rate": 2.490691622828796e-06, "loss": 0.4092, "step": 21817 }, { "epoch": 0.7773126458485491, "grad_norm": 1.1342650443928846, "learning_rate": 2.489929635121823e-06, "loss": 0.4317, "step": 21818 }, { "epoch": 0.7773482729750432, "grad_norm": 1.494444669011866, "learning_rate": 2.4891677474166854e-06, "loss": 0.4929, "step": 21819 }, { "epoch": 0.7773839001015374, "grad_norm": 1.5866434693463345, "learning_rate": 2.488405959723532e-06, "loss": 0.4113, "step": 21820 }, { "epoch": 0.7774195272280314, "grad_norm": 1.3575493067837747, "learning_rate": 2.4876442720525006e-06, "loss": 0.4481, "step": 21821 }, { "epoch": 0.7774551543545255, "grad_norm": 1.762237425101647, "learning_rate": 2.4868826844137406e-06, "loss": 0.5268, "step": 21822 }, { "epoch": 0.7774907814810197, "grad_norm": 1.578880135847394, "learning_rate": 2.4861211968173847e-06, "loss": 0.5505, "step": 21823 }, { "epoch": 0.7775264086075138, "grad_norm": 1.3467365847217863, "learning_rate": 2.485359809273581e-06, "loss": 0.5159, "step": 21824 }, { "epoch": 0.7775620357340078, "grad_norm": 1.1178752069815174, "learning_rate": 2.4845985217924594e-06, "loss": 0.4106, "step": 21825 }, { "epoch": 0.777597662860502, "grad_norm": 1.9272286229900926, "learning_rate": 2.4838373343841603e-06, "loss": 0.5139, "step": 21826 }, { "epoch": 0.7776332899869961, "grad_norm": 1.2720118121084578, "learning_rate": 2.4830762470588255e-06, "loss": 0.4469, "step": 21827 }, { "epoch": 0.7776689171134902, "grad_norm": 1.2517101128471888, "learning_rate": 2.482315259826581e-06, "loss": 0.4253, "step": 21828 }, { "epoch": 0.7777045442399844, "grad_norm": 1.2359934911462969, "learning_rate": 2.481554372697568e-06, "loss": 0.457, "step": 21829 }, { "epoch": 0.7777401713664784, "grad_norm": 1.194191092995395, "learning_rate": 2.480793585681909e-06, "loss": 0.4189, "step": 21830 }, { "epoch": 0.7777757984929725, "grad_norm": 1.1067377556391502, "learning_rate": 2.4800328987897427e-06, "loss": 0.4117, "step": 21831 }, { "epoch": 0.7778114256194667, "grad_norm": 1.102489262448578, "learning_rate": 2.479272312031191e-06, "loss": 0.4079, "step": 21832 }, { "epoch": 0.7778470527459608, "grad_norm": 1.1738774138330108, "learning_rate": 2.4785118254163867e-06, "loss": 0.4061, "step": 21833 }, { "epoch": 0.7778826798724549, "grad_norm": 1.200568365764813, "learning_rate": 2.4777514389554524e-06, "loss": 0.3657, "step": 21834 }, { "epoch": 0.777918306998949, "grad_norm": 1.3346268372798995, "learning_rate": 2.47699115265852e-06, "loss": 0.4549, "step": 21835 }, { "epoch": 0.7779539341254431, "grad_norm": 1.359152595309614, "learning_rate": 2.4762309665357056e-06, "loss": 0.4207, "step": 21836 }, { "epoch": 0.7779895612519372, "grad_norm": 1.7478986044611515, "learning_rate": 2.4754708805971338e-06, "loss": 0.5444, "step": 21837 }, { "epoch": 0.7780251883784314, "grad_norm": 1.3226770414991391, "learning_rate": 2.4747108948529297e-06, "loss": 0.4165, "step": 21838 }, { "epoch": 0.7780608155049255, "grad_norm": 1.4176284760245104, "learning_rate": 2.473951009313208e-06, "loss": 0.5502, "step": 21839 }, { "epoch": 0.7780964426314195, "grad_norm": 1.553252765676944, "learning_rate": 2.473191223988087e-06, "loss": 0.4988, "step": 21840 }, { "epoch": 0.7781320697579137, "grad_norm": 1.3813678684468496, "learning_rate": 2.4724315388876872e-06, "loss": 0.4835, "step": 21841 }, { "epoch": 0.7781676968844078, "grad_norm": 1.1096766400362545, "learning_rate": 2.4716719540221267e-06, "loss": 0.4113, "step": 21842 }, { "epoch": 0.7782033240109019, "grad_norm": 1.8190258601810874, "learning_rate": 2.470912469401512e-06, "loss": 0.5598, "step": 21843 }, { "epoch": 0.778238951137396, "grad_norm": 1.3347993191865035, "learning_rate": 2.4701530850359634e-06, "loss": 0.4822, "step": 21844 }, { "epoch": 0.7782745782638901, "grad_norm": 2.3606439709179763, "learning_rate": 2.469393800935587e-06, "loss": 0.4978, "step": 21845 }, { "epoch": 0.7783102053903842, "grad_norm": 1.626808937983092, "learning_rate": 2.468634617110498e-06, "loss": 0.4943, "step": 21846 }, { "epoch": 0.7783458325168784, "grad_norm": 1.5707435398830023, "learning_rate": 2.4678755335708005e-06, "loss": 0.5323, "step": 21847 }, { "epoch": 0.7783814596433725, "grad_norm": 1.6483580544894396, "learning_rate": 2.4671165503266046e-06, "loss": 0.562, "step": 21848 }, { "epoch": 0.7784170867698665, "grad_norm": 1.9877656166230544, "learning_rate": 2.4663576673880174e-06, "loss": 0.5593, "step": 21849 }, { "epoch": 0.7784527138963607, "grad_norm": 1.122643831122988, "learning_rate": 2.4655988847651434e-06, "loss": 0.4032, "step": 21850 }, { "epoch": 0.7784883410228548, "grad_norm": 1.572324026285995, "learning_rate": 2.4648402024680895e-06, "loss": 0.4596, "step": 21851 }, { "epoch": 0.7785239681493489, "grad_norm": 1.2402104754339374, "learning_rate": 2.464081620506952e-06, "loss": 0.3937, "step": 21852 }, { "epoch": 0.7785595952758431, "grad_norm": 1.3245936778925427, "learning_rate": 2.4633231388918377e-06, "loss": 0.4927, "step": 21853 }, { "epoch": 0.7785952224023371, "grad_norm": 1.6114765309561767, "learning_rate": 2.4625647576328415e-06, "loss": 0.5357, "step": 21854 }, { "epoch": 0.7786308495288312, "grad_norm": 1.575159020545549, "learning_rate": 2.461806476740063e-06, "loss": 0.5449, "step": 21855 }, { "epoch": 0.7786664766553254, "grad_norm": 1.1296295572632737, "learning_rate": 2.4610482962236003e-06, "loss": 0.4034, "step": 21856 }, { "epoch": 0.7787021037818195, "grad_norm": 1.1256327991048145, "learning_rate": 2.4602902160935527e-06, "loss": 0.3977, "step": 21857 }, { "epoch": 0.7787377309083136, "grad_norm": 1.5341091605886508, "learning_rate": 2.459532236360007e-06, "loss": 0.5192, "step": 21858 }, { "epoch": 0.7787733580348077, "grad_norm": 1.5340786890047244, "learning_rate": 2.4587743570330615e-06, "loss": 0.5513, "step": 21859 }, { "epoch": 0.7788089851613018, "grad_norm": 1.197964451579936, "learning_rate": 2.458016578122808e-06, "loss": 0.3855, "step": 21860 }, { "epoch": 0.7788446122877959, "grad_norm": 1.3726794113469465, "learning_rate": 2.4572588996393323e-06, "loss": 0.4698, "step": 21861 }, { "epoch": 0.7788802394142901, "grad_norm": 1.7326895919824303, "learning_rate": 2.45650132159273e-06, "loss": 0.4774, "step": 21862 }, { "epoch": 0.7789158665407842, "grad_norm": 1.2541634599411557, "learning_rate": 2.4557438439930813e-06, "loss": 0.4126, "step": 21863 }, { "epoch": 0.7789514936672782, "grad_norm": 1.647129206193551, "learning_rate": 2.4549864668504774e-06, "loss": 0.5469, "step": 21864 }, { "epoch": 0.7789871207937724, "grad_norm": 1.4565326917186658, "learning_rate": 2.4542291901750025e-06, "loss": 0.5409, "step": 21865 }, { "epoch": 0.7790227479202665, "grad_norm": 1.1892881339932067, "learning_rate": 2.453472013976742e-06, "loss": 0.3938, "step": 21866 }, { "epoch": 0.7790583750467606, "grad_norm": 1.3826706251343854, "learning_rate": 2.4527149382657744e-06, "loss": 0.4203, "step": 21867 }, { "epoch": 0.7790940021732548, "grad_norm": 1.154817363727204, "learning_rate": 2.451957963052185e-06, "loss": 0.3899, "step": 21868 }, { "epoch": 0.7791296292997488, "grad_norm": 1.7405858537912506, "learning_rate": 2.451201088346048e-06, "loss": 0.5552, "step": 21869 }, { "epoch": 0.7791652564262429, "grad_norm": 1.256871373215068, "learning_rate": 2.450444314157445e-06, "loss": 0.398, "step": 21870 }, { "epoch": 0.7792008835527371, "grad_norm": 1.2156171500110204, "learning_rate": 2.4496876404964533e-06, "loss": 0.4111, "step": 21871 }, { "epoch": 0.7792365106792312, "grad_norm": 1.1698680289680963, "learning_rate": 2.448931067373147e-06, "loss": 0.4376, "step": 21872 }, { "epoch": 0.7792721378057252, "grad_norm": 1.1921025957839564, "learning_rate": 2.4481745947976053e-06, "loss": 0.4379, "step": 21873 }, { "epoch": 0.7793077649322194, "grad_norm": 1.4210368328654592, "learning_rate": 2.4474182227798936e-06, "loss": 0.5036, "step": 21874 }, { "epoch": 0.7793433920587135, "grad_norm": 1.1468756868607888, "learning_rate": 2.4466619513300917e-06, "loss": 0.4316, "step": 21875 }, { "epoch": 0.7793790191852076, "grad_norm": 1.1626662107323105, "learning_rate": 2.445905780458261e-06, "loss": 0.4116, "step": 21876 }, { "epoch": 0.7794146463117018, "grad_norm": 1.6760749225823646, "learning_rate": 2.4451497101744746e-06, "loss": 0.5636, "step": 21877 }, { "epoch": 0.7794502734381958, "grad_norm": 1.7047431892516465, "learning_rate": 2.4443937404888007e-06, "loss": 0.5961, "step": 21878 }, { "epoch": 0.7794859005646899, "grad_norm": 1.1007335565008778, "learning_rate": 2.4436378714113087e-06, "loss": 0.4, "step": 21879 }, { "epoch": 0.7795215276911841, "grad_norm": 1.1164549514484907, "learning_rate": 2.4428821029520566e-06, "loss": 0.3991, "step": 21880 }, { "epoch": 0.7795571548176782, "grad_norm": 1.5409614046327913, "learning_rate": 2.442126435121115e-06, "loss": 0.505, "step": 21881 }, { "epoch": 0.7795927819441723, "grad_norm": 1.4785182409460909, "learning_rate": 2.4413708679285386e-06, "loss": 0.5134, "step": 21882 }, { "epoch": 0.7796284090706664, "grad_norm": 1.5500859508393567, "learning_rate": 2.440615401384393e-06, "loss": 0.5598, "step": 21883 }, { "epoch": 0.7796640361971605, "grad_norm": 1.2825884721930072, "learning_rate": 2.439860035498739e-06, "loss": 0.4124, "step": 21884 }, { "epoch": 0.7796996633236546, "grad_norm": 1.5454047363223422, "learning_rate": 2.4391047702816306e-06, "loss": 0.4635, "step": 21885 }, { "epoch": 0.7797352904501488, "grad_norm": 1.1831946893853202, "learning_rate": 2.4383496057431264e-06, "loss": 0.4216, "step": 21886 }, { "epoch": 0.7797709175766429, "grad_norm": 1.42714356081577, "learning_rate": 2.4375945418932832e-06, "loss": 0.4385, "step": 21887 }, { "epoch": 0.7798065447031369, "grad_norm": 1.113596100333004, "learning_rate": 2.4368395787421575e-06, "loss": 0.4045, "step": 21888 }, { "epoch": 0.7798421718296311, "grad_norm": 1.144000177218889, "learning_rate": 2.4360847162997958e-06, "loss": 0.3661, "step": 21889 }, { "epoch": 0.7798777989561252, "grad_norm": 1.4168034738225441, "learning_rate": 2.4353299545762567e-06, "loss": 0.442, "step": 21890 }, { "epoch": 0.7799134260826193, "grad_norm": 1.2065358213086208, "learning_rate": 2.4345752935815826e-06, "loss": 0.4219, "step": 21891 }, { "epoch": 0.7799490532091135, "grad_norm": 1.1630787467784875, "learning_rate": 2.433820733325827e-06, "loss": 0.3892, "step": 21892 }, { "epoch": 0.7799846803356075, "grad_norm": 1.8182830327441992, "learning_rate": 2.433066273819037e-06, "loss": 0.4614, "step": 21893 }, { "epoch": 0.7800203074621016, "grad_norm": 1.435885376231196, "learning_rate": 2.432311915071257e-06, "loss": 0.4895, "step": 21894 }, { "epoch": 0.7800559345885958, "grad_norm": 1.3680960938039957, "learning_rate": 2.431557657092538e-06, "loss": 0.537, "step": 21895 }, { "epoch": 0.7800915617150899, "grad_norm": 1.0934968813752364, "learning_rate": 2.430803499892915e-06, "loss": 0.3958, "step": 21896 }, { "epoch": 0.7801271888415839, "grad_norm": 1.193178094451964, "learning_rate": 2.4300494434824373e-06, "loss": 0.4033, "step": 21897 }, { "epoch": 0.7801628159680781, "grad_norm": 1.3115528982559104, "learning_rate": 2.4292954878711395e-06, "loss": 0.4643, "step": 21898 }, { "epoch": 0.7801984430945722, "grad_norm": 1.137949176803723, "learning_rate": 2.428541633069066e-06, "loss": 0.4161, "step": 21899 }, { "epoch": 0.7802340702210663, "grad_norm": 1.6608193553935913, "learning_rate": 2.427787879086251e-06, "loss": 0.5256, "step": 21900 }, { "epoch": 0.7802696973475605, "grad_norm": 1.090799163487575, "learning_rate": 2.427034225932733e-06, "loss": 0.4239, "step": 21901 }, { "epoch": 0.7803053244740545, "grad_norm": 1.3893324152288138, "learning_rate": 2.4262806736185463e-06, "loss": 0.5133, "step": 21902 }, { "epoch": 0.7803409516005486, "grad_norm": 1.6792000832814118, "learning_rate": 2.4255272221537295e-06, "loss": 0.4896, "step": 21903 }, { "epoch": 0.7803765787270428, "grad_norm": 1.1970804394478158, "learning_rate": 2.4247738715483092e-06, "loss": 0.4276, "step": 21904 }, { "epoch": 0.7804122058535369, "grad_norm": 1.609962982158213, "learning_rate": 2.424020621812323e-06, "loss": 0.5707, "step": 21905 }, { "epoch": 0.780447832980031, "grad_norm": 1.6081901660040236, "learning_rate": 2.4232674729557936e-06, "loss": 0.4722, "step": 21906 }, { "epoch": 0.7804834601065251, "grad_norm": 1.4349916613495806, "learning_rate": 2.4225144249887546e-06, "loss": 0.4734, "step": 21907 }, { "epoch": 0.7805190872330192, "grad_norm": 1.0497589372846698, "learning_rate": 2.421761477921232e-06, "loss": 0.402, "step": 21908 }, { "epoch": 0.7805547143595133, "grad_norm": 1.1507182844741088, "learning_rate": 2.421008631763252e-06, "loss": 0.4036, "step": 21909 }, { "epoch": 0.7805903414860075, "grad_norm": 1.172106851735018, "learning_rate": 2.4202558865248426e-06, "loss": 0.4327, "step": 21910 }, { "epoch": 0.7806259686125016, "grad_norm": 1.1253782542025843, "learning_rate": 2.4195032422160213e-06, "loss": 0.381, "step": 21911 }, { "epoch": 0.7806615957389956, "grad_norm": 1.6881171097375494, "learning_rate": 2.418750698846816e-06, "loss": 0.5242, "step": 21912 }, { "epoch": 0.7806972228654898, "grad_norm": 1.6107073682822353, "learning_rate": 2.4179982564272396e-06, "loss": 0.5653, "step": 21913 }, { "epoch": 0.7807328499919839, "grad_norm": 2.0032909453676417, "learning_rate": 2.41724591496732e-06, "loss": 0.5698, "step": 21914 }, { "epoch": 0.780768477118478, "grad_norm": 1.8070042872453034, "learning_rate": 2.4164936744770653e-06, "loss": 0.4889, "step": 21915 }, { "epoch": 0.7808041042449722, "grad_norm": 1.038460835961667, "learning_rate": 2.4157415349665037e-06, "loss": 0.3952, "step": 21916 }, { "epoch": 0.7808397313714662, "grad_norm": 1.183533817455296, "learning_rate": 2.414989496445641e-06, "loss": 0.4224, "step": 21917 }, { "epoch": 0.7808753584979603, "grad_norm": 1.2665217201169798, "learning_rate": 2.414237558924496e-06, "loss": 0.4004, "step": 21918 }, { "epoch": 0.7809109856244545, "grad_norm": 1.0919943302713206, "learning_rate": 2.4134857224130814e-06, "loss": 0.4393, "step": 21919 }, { "epoch": 0.7809466127509486, "grad_norm": 1.7376709962448253, "learning_rate": 2.412733986921405e-06, "loss": 0.5307, "step": 21920 }, { "epoch": 0.7809822398774426, "grad_norm": 1.4429834863841902, "learning_rate": 2.4119823524594822e-06, "loss": 0.4398, "step": 21921 }, { "epoch": 0.7810178670039368, "grad_norm": 1.2861940402695458, "learning_rate": 2.4112308190373147e-06, "loss": 0.4771, "step": 21922 }, { "epoch": 0.7810534941304309, "grad_norm": 1.1586191706939253, "learning_rate": 2.4104793866649124e-06, "loss": 0.3579, "step": 21923 }, { "epoch": 0.781089121256925, "grad_norm": 1.100682294985839, "learning_rate": 2.409728055352283e-06, "loss": 0.3974, "step": 21924 }, { "epoch": 0.7811247483834192, "grad_norm": 1.4563393449016875, "learning_rate": 2.4089768251094314e-06, "loss": 0.5853, "step": 21925 }, { "epoch": 0.7811603755099132, "grad_norm": 1.1738441842074476, "learning_rate": 2.408225695946358e-06, "loss": 0.41, "step": 21926 }, { "epoch": 0.7811960026364073, "grad_norm": 1.5963962570407055, "learning_rate": 2.4074746678730677e-06, "loss": 0.5038, "step": 21927 }, { "epoch": 0.7812316297629015, "grad_norm": 1.1983552224911636, "learning_rate": 2.4067237408995557e-06, "loss": 0.4254, "step": 21928 }, { "epoch": 0.7812672568893956, "grad_norm": 1.4938136474114063, "learning_rate": 2.4059729150358244e-06, "loss": 0.5016, "step": 21929 }, { "epoch": 0.7813028840158897, "grad_norm": 1.5417396241334826, "learning_rate": 2.405222190291873e-06, "loss": 0.5695, "step": 21930 }, { "epoch": 0.7813385111423838, "grad_norm": 1.0600876558396566, "learning_rate": 2.4044715666776954e-06, "loss": 0.3716, "step": 21931 }, { "epoch": 0.7813741382688779, "grad_norm": 1.1455118935495028, "learning_rate": 2.403721044203291e-06, "loss": 0.3487, "step": 21932 }, { "epoch": 0.781409765395372, "grad_norm": 1.1749276055243778, "learning_rate": 2.4029706228786477e-06, "loss": 0.3619, "step": 21933 }, { "epoch": 0.7814453925218662, "grad_norm": 1.8376226138688856, "learning_rate": 2.4022203027137635e-06, "loss": 0.5328, "step": 21934 }, { "epoch": 0.7814810196483603, "grad_norm": 1.6835133287069626, "learning_rate": 2.4014700837186234e-06, "loss": 0.503, "step": 21935 }, { "epoch": 0.7815166467748543, "grad_norm": 1.4206516700652683, "learning_rate": 2.400719965903223e-06, "loss": 0.4627, "step": 21936 }, { "epoch": 0.7815522739013485, "grad_norm": 1.1498237250380774, "learning_rate": 2.399969949277545e-06, "loss": 0.4001, "step": 21937 }, { "epoch": 0.7815879010278426, "grad_norm": 1.6355019958403796, "learning_rate": 2.3992200338515794e-06, "loss": 0.4774, "step": 21938 }, { "epoch": 0.7816235281543367, "grad_norm": 1.1762936828484103, "learning_rate": 2.3984702196353116e-06, "loss": 0.3941, "step": 21939 }, { "epoch": 0.7816591552808309, "grad_norm": 1.5403855257389782, "learning_rate": 2.3977205066387255e-06, "loss": 0.4879, "step": 21940 }, { "epoch": 0.7816947824073249, "grad_norm": 1.1140696856947698, "learning_rate": 2.3969708948718084e-06, "loss": 0.391, "step": 21941 }, { "epoch": 0.781730409533819, "grad_norm": 1.5561966767306585, "learning_rate": 2.396221384344535e-06, "loss": 0.5418, "step": 21942 }, { "epoch": 0.7817660366603132, "grad_norm": 1.513739106358771, "learning_rate": 2.395471975066891e-06, "loss": 0.545, "step": 21943 }, { "epoch": 0.7818016637868073, "grad_norm": 1.5256680667072073, "learning_rate": 2.39472266704885e-06, "loss": 0.5144, "step": 21944 }, { "epoch": 0.7818372909133013, "grad_norm": 1.230193957514121, "learning_rate": 2.3939734603003928e-06, "loss": 0.3922, "step": 21945 }, { "epoch": 0.7818729180397955, "grad_norm": 1.1716328410407477, "learning_rate": 2.393224354831495e-06, "loss": 0.4036, "step": 21946 }, { "epoch": 0.7819085451662896, "grad_norm": 1.12953076651696, "learning_rate": 2.3924753506521346e-06, "loss": 0.4615, "step": 21947 }, { "epoch": 0.7819441722927837, "grad_norm": 1.1332723896782237, "learning_rate": 2.391726447772279e-06, "loss": 0.4206, "step": 21948 }, { "epoch": 0.7819797994192779, "grad_norm": 1.357025646989567, "learning_rate": 2.3909776462019073e-06, "loss": 0.5318, "step": 21949 }, { "epoch": 0.7820154265457719, "grad_norm": 1.0937371195802224, "learning_rate": 2.3902289459509832e-06, "loss": 0.3728, "step": 21950 }, { "epoch": 0.782051053672266, "grad_norm": 1.1449855145621868, "learning_rate": 2.3894803470294824e-06, "loss": 0.4385, "step": 21951 }, { "epoch": 0.7820866807987602, "grad_norm": 1.0485852299669012, "learning_rate": 2.3887318494473676e-06, "loss": 0.3597, "step": 21952 }, { "epoch": 0.7821223079252543, "grad_norm": 1.588090112002644, "learning_rate": 2.387983453214605e-06, "loss": 0.5219, "step": 21953 }, { "epoch": 0.7821579350517484, "grad_norm": 1.4062461512576723, "learning_rate": 2.38723515834117e-06, "loss": 0.5331, "step": 21954 }, { "epoch": 0.7821935621782425, "grad_norm": 1.1559701003078586, "learning_rate": 2.386486964837016e-06, "loss": 0.3925, "step": 21955 }, { "epoch": 0.7822291893047366, "grad_norm": 1.167901476591407, "learning_rate": 2.3857388727121123e-06, "loss": 0.4077, "step": 21956 }, { "epoch": 0.7822648164312307, "grad_norm": 1.6043269868589407, "learning_rate": 2.3849908819764154e-06, "loss": 0.4639, "step": 21957 }, { "epoch": 0.7823004435577249, "grad_norm": 1.3810690545080822, "learning_rate": 2.3842429926398903e-06, "loss": 0.4537, "step": 21958 }, { "epoch": 0.782336070684219, "grad_norm": 1.1180051160468063, "learning_rate": 2.3834952047124895e-06, "loss": 0.4044, "step": 21959 }, { "epoch": 0.782371697810713, "grad_norm": 1.706125435267809, "learning_rate": 2.3827475182041738e-06, "loss": 0.5523, "step": 21960 }, { "epoch": 0.7824073249372072, "grad_norm": 1.293365451480639, "learning_rate": 2.3819999331248987e-06, "loss": 0.5576, "step": 21961 }, { "epoch": 0.7824429520637013, "grad_norm": 1.1638752798610155, "learning_rate": 2.3812524494846224e-06, "loss": 0.3985, "step": 21962 }, { "epoch": 0.7824785791901954, "grad_norm": 1.6612280530136339, "learning_rate": 2.380505067293293e-06, "loss": 0.465, "step": 21963 }, { "epoch": 0.7825142063166896, "grad_norm": 1.1146627009611767, "learning_rate": 2.3797577865608633e-06, "loss": 0.4115, "step": 21964 }, { "epoch": 0.7825498334431836, "grad_norm": 1.2080972543414046, "learning_rate": 2.3790106072972874e-06, "loss": 0.4139, "step": 21965 }, { "epoch": 0.7825854605696777, "grad_norm": 1.2647122677634894, "learning_rate": 2.37826352951251e-06, "loss": 0.3861, "step": 21966 }, { "epoch": 0.7826210876961719, "grad_norm": 1.050917227250699, "learning_rate": 2.37751655321648e-06, "loss": 0.4323, "step": 21967 }, { "epoch": 0.782656714822666, "grad_norm": 2.042665408407441, "learning_rate": 2.3767696784191464e-06, "loss": 0.6042, "step": 21968 }, { "epoch": 0.78269234194916, "grad_norm": 1.1799052489246677, "learning_rate": 2.376022905130454e-06, "loss": 0.3571, "step": 21969 }, { "epoch": 0.7827279690756542, "grad_norm": 1.7289575585620693, "learning_rate": 2.375276233360344e-06, "loss": 0.5253, "step": 21970 }, { "epoch": 0.7827635962021483, "grad_norm": 1.4048376234508508, "learning_rate": 2.374529663118762e-06, "loss": 0.5074, "step": 21971 }, { "epoch": 0.7827992233286424, "grad_norm": 1.0819438607336542, "learning_rate": 2.3737831944156453e-06, "loss": 0.4228, "step": 21972 }, { "epoch": 0.7828348504551366, "grad_norm": 1.492308734753954, "learning_rate": 2.3730368272609394e-06, "loss": 0.4875, "step": 21973 }, { "epoch": 0.7828704775816306, "grad_norm": 1.5574326095850752, "learning_rate": 2.3722905616645754e-06, "loss": 0.5253, "step": 21974 }, { "epoch": 0.7829061047081247, "grad_norm": 1.9519033460351587, "learning_rate": 2.3715443976364937e-06, "loss": 0.4909, "step": 21975 }, { "epoch": 0.7829417318346189, "grad_norm": 1.2115072202214652, "learning_rate": 2.3707983351866304e-06, "loss": 0.3954, "step": 21976 }, { "epoch": 0.782977358961113, "grad_norm": 1.8766630715260644, "learning_rate": 2.370052374324919e-06, "loss": 0.4295, "step": 21977 }, { "epoch": 0.783012986087607, "grad_norm": 1.123831349411893, "learning_rate": 2.369306515061298e-06, "loss": 0.4181, "step": 21978 }, { "epoch": 0.7830486132141012, "grad_norm": 1.7335060190932168, "learning_rate": 2.36856075740569e-06, "loss": 0.4872, "step": 21979 }, { "epoch": 0.7830842403405953, "grad_norm": 1.5259386011941323, "learning_rate": 2.367815101368033e-06, "loss": 0.5416, "step": 21980 }, { "epoch": 0.7831198674670894, "grad_norm": 1.4131071964816881, "learning_rate": 2.3670695469582492e-06, "loss": 0.5667, "step": 21981 }, { "epoch": 0.7831554945935836, "grad_norm": 1.7316877777333461, "learning_rate": 2.36632409418627e-06, "loss": 0.5065, "step": 21982 }, { "epoch": 0.7831911217200777, "grad_norm": 1.7260438653213208, "learning_rate": 2.3655787430620214e-06, "loss": 0.5534, "step": 21983 }, { "epoch": 0.7832267488465717, "grad_norm": 1.9766505414771394, "learning_rate": 2.36483349359543e-06, "loss": 0.455, "step": 21984 }, { "epoch": 0.7832623759730659, "grad_norm": 1.1072048413838522, "learning_rate": 2.3640883457964146e-06, "loss": 0.3489, "step": 21985 }, { "epoch": 0.78329800309956, "grad_norm": 1.1666203789819274, "learning_rate": 2.3633432996749005e-06, "loss": 0.4138, "step": 21986 }, { "epoch": 0.7833336302260541, "grad_norm": 1.4761534604237119, "learning_rate": 2.3625983552408103e-06, "loss": 0.4929, "step": 21987 }, { "epoch": 0.7833692573525483, "grad_norm": 1.0942649505054445, "learning_rate": 2.361853512504059e-06, "loss": 0.438, "step": 21988 }, { "epoch": 0.7834048844790423, "grad_norm": 1.101968012327184, "learning_rate": 2.3611087714745696e-06, "loss": 0.422, "step": 21989 }, { "epoch": 0.7834405116055364, "grad_norm": 1.7585139083124655, "learning_rate": 2.3603641321622506e-06, "loss": 0.4847, "step": 21990 }, { "epoch": 0.7834761387320306, "grad_norm": 1.5446996446179453, "learning_rate": 2.3596195945770284e-06, "loss": 0.4967, "step": 21991 }, { "epoch": 0.7835117658585247, "grad_norm": 1.6995286718754996, "learning_rate": 2.358875158728808e-06, "loss": 0.5363, "step": 21992 }, { "epoch": 0.7835473929850187, "grad_norm": 1.1942840232116956, "learning_rate": 2.3581308246275103e-06, "loss": 0.3705, "step": 21993 }, { "epoch": 0.7835830201115129, "grad_norm": 1.1960483949384657, "learning_rate": 2.3573865922830387e-06, "loss": 0.4184, "step": 21994 }, { "epoch": 0.783618647238007, "grad_norm": 1.133527080293152, "learning_rate": 2.3566424617053084e-06, "loss": 0.3695, "step": 21995 }, { "epoch": 0.7836542743645011, "grad_norm": 1.4354857297659873, "learning_rate": 2.3558984329042233e-06, "loss": 0.4988, "step": 21996 }, { "epoch": 0.7836899014909953, "grad_norm": 1.4115218996279988, "learning_rate": 2.355154505889694e-06, "loss": 0.4252, "step": 21997 }, { "epoch": 0.7837255286174893, "grad_norm": 1.2349948582402206, "learning_rate": 2.354410680671626e-06, "loss": 0.4349, "step": 21998 }, { "epoch": 0.7837611557439834, "grad_norm": 1.3938046491191796, "learning_rate": 2.3536669572599234e-06, "loss": 0.5033, "step": 21999 }, { "epoch": 0.7837967828704776, "grad_norm": 1.1396220942280848, "learning_rate": 2.352923335664493e-06, "loss": 0.4241, "step": 22000 }, { "epoch": 0.7838324099969717, "grad_norm": 1.4084463882488145, "learning_rate": 2.3521798158952302e-06, "loss": 0.4646, "step": 22001 }, { "epoch": 0.7838680371234658, "grad_norm": 1.2757439924461822, "learning_rate": 2.3514363979620412e-06, "loss": 0.4101, "step": 22002 }, { "epoch": 0.7839036642499599, "grad_norm": 1.2041602651770558, "learning_rate": 2.3506930818748207e-06, "loss": 0.3914, "step": 22003 }, { "epoch": 0.783939291376454, "grad_norm": 1.1681139258197126, "learning_rate": 2.3499498676434675e-06, "loss": 0.3738, "step": 22004 }, { "epoch": 0.7839749185029481, "grad_norm": 1.4315407436179761, "learning_rate": 2.3492067552778784e-06, "loss": 0.5599, "step": 22005 }, { "epoch": 0.7840105456294423, "grad_norm": 1.1690363464693438, "learning_rate": 2.348463744787953e-06, "loss": 0.3749, "step": 22006 }, { "epoch": 0.7840461727559364, "grad_norm": 1.2129609383065432, "learning_rate": 2.347720836183578e-06, "loss": 0.3989, "step": 22007 }, { "epoch": 0.7840817998824304, "grad_norm": 1.5836042991147001, "learning_rate": 2.34697802947465e-06, "loss": 0.564, "step": 22008 }, { "epoch": 0.7841174270089246, "grad_norm": 1.246723438188216, "learning_rate": 2.346235324671058e-06, "loss": 0.4215, "step": 22009 }, { "epoch": 0.7841530541354187, "grad_norm": 1.7142433768882828, "learning_rate": 2.3454927217826906e-06, "loss": 0.5085, "step": 22010 }, { "epoch": 0.7841886812619128, "grad_norm": 1.5669636805522744, "learning_rate": 2.3447502208194415e-06, "loss": 0.5821, "step": 22011 }, { "epoch": 0.784224308388407, "grad_norm": 1.1768123167704339, "learning_rate": 2.344007821791191e-06, "loss": 0.3888, "step": 22012 }, { "epoch": 0.784259935514901, "grad_norm": 1.6403200532970754, "learning_rate": 2.3432655247078273e-06, "loss": 0.4588, "step": 22013 }, { "epoch": 0.7842955626413951, "grad_norm": 1.6575293418783514, "learning_rate": 2.342523329579235e-06, "loss": 0.6315, "step": 22014 }, { "epoch": 0.7843311897678893, "grad_norm": 1.5471308175256024, "learning_rate": 2.3417812364153e-06, "loss": 0.5605, "step": 22015 }, { "epoch": 0.7843668168943834, "grad_norm": 1.1279847994882592, "learning_rate": 2.341039245225898e-06, "loss": 0.3889, "step": 22016 }, { "epoch": 0.7844024440208774, "grad_norm": 1.138753916302711, "learning_rate": 2.340297356020913e-06, "loss": 0.3757, "step": 22017 }, { "epoch": 0.7844380711473716, "grad_norm": 1.4054933742523505, "learning_rate": 2.339555568810221e-06, "loss": 0.5335, "step": 22018 }, { "epoch": 0.7844736982738657, "grad_norm": 1.5951980205471592, "learning_rate": 2.338813883603701e-06, "loss": 0.4724, "step": 22019 }, { "epoch": 0.7845093254003598, "grad_norm": 1.46048836128704, "learning_rate": 2.338072300411228e-06, "loss": 0.5418, "step": 22020 }, { "epoch": 0.784544952526854, "grad_norm": 1.7175454120341271, "learning_rate": 2.3373308192426793e-06, "loss": 0.4577, "step": 22021 }, { "epoch": 0.784580579653348, "grad_norm": 1.3773473920013817, "learning_rate": 2.336589440107928e-06, "loss": 0.5005, "step": 22022 }, { "epoch": 0.7846162067798421, "grad_norm": 1.3555053988148553, "learning_rate": 2.3358481630168416e-06, "loss": 0.4228, "step": 22023 }, { "epoch": 0.7846518339063363, "grad_norm": 1.220645476447731, "learning_rate": 2.3351069879792975e-06, "loss": 0.3948, "step": 22024 }, { "epoch": 0.7846874610328304, "grad_norm": 1.6006439747401149, "learning_rate": 2.334365915005159e-06, "loss": 0.5153, "step": 22025 }, { "epoch": 0.7847230881593245, "grad_norm": 1.1605955766936178, "learning_rate": 2.333624944104298e-06, "loss": 0.3805, "step": 22026 }, { "epoch": 0.7847587152858186, "grad_norm": 1.910519977149334, "learning_rate": 2.3328840752865776e-06, "loss": 0.4959, "step": 22027 }, { "epoch": 0.7847943424123127, "grad_norm": 1.5756746077573722, "learning_rate": 2.3321433085618637e-06, "loss": 0.5057, "step": 22028 }, { "epoch": 0.7848299695388068, "grad_norm": 1.573529219111098, "learning_rate": 2.3314026439400217e-06, "loss": 0.4845, "step": 22029 }, { "epoch": 0.784865596665301, "grad_norm": 1.1450128523023888, "learning_rate": 2.3306620814309165e-06, "loss": 0.4251, "step": 22030 }, { "epoch": 0.784901223791795, "grad_norm": 1.6344744463828766, "learning_rate": 2.329921621044403e-06, "loss": 0.5908, "step": 22031 }, { "epoch": 0.7849368509182891, "grad_norm": 1.5087438525240515, "learning_rate": 2.3291812627903475e-06, "loss": 0.5171, "step": 22032 }, { "epoch": 0.7849724780447833, "grad_norm": 1.1455656063000776, "learning_rate": 2.3284410066786025e-06, "loss": 0.3842, "step": 22033 }, { "epoch": 0.7850081051712774, "grad_norm": 1.3270285790822811, "learning_rate": 2.327700852719026e-06, "loss": 0.4706, "step": 22034 }, { "epoch": 0.7850437322977715, "grad_norm": 1.1748819424531936, "learning_rate": 2.326960800921476e-06, "loss": 0.4347, "step": 22035 }, { "epoch": 0.7850793594242657, "grad_norm": 1.1939202656579937, "learning_rate": 2.3262208512958074e-06, "loss": 0.3697, "step": 22036 }, { "epoch": 0.7851149865507597, "grad_norm": 1.3672334656064407, "learning_rate": 2.3254810038518727e-06, "loss": 0.4606, "step": 22037 }, { "epoch": 0.7851506136772538, "grad_norm": 1.3984199269843518, "learning_rate": 2.324741258599521e-06, "loss": 0.4687, "step": 22038 }, { "epoch": 0.785186240803748, "grad_norm": 1.1138099589461536, "learning_rate": 2.324001615548607e-06, "loss": 0.4283, "step": 22039 }, { "epoch": 0.7852218679302421, "grad_norm": 1.1572306694092511, "learning_rate": 2.323262074708973e-06, "loss": 0.4278, "step": 22040 }, { "epoch": 0.7852574950567363, "grad_norm": 1.308317553458708, "learning_rate": 2.322522636090474e-06, "loss": 0.4207, "step": 22041 }, { "epoch": 0.7852931221832303, "grad_norm": 1.1842888771501543, "learning_rate": 2.3217832997029467e-06, "loss": 0.3602, "step": 22042 }, { "epoch": 0.7853287493097244, "grad_norm": 1.2885947504971202, "learning_rate": 2.321044065556246e-06, "loss": 0.4026, "step": 22043 }, { "epoch": 0.7853643764362186, "grad_norm": 1.4691071863427254, "learning_rate": 2.320304933660209e-06, "loss": 0.489, "step": 22044 }, { "epoch": 0.7854000035627127, "grad_norm": 1.6981197782545856, "learning_rate": 2.319565904024679e-06, "loss": 0.5472, "step": 22045 }, { "epoch": 0.7854356306892067, "grad_norm": 1.499554944584819, "learning_rate": 2.3188269766595004e-06, "loss": 0.5691, "step": 22046 }, { "epoch": 0.7854712578157009, "grad_norm": 1.2470447041409554, "learning_rate": 2.3180881515745067e-06, "loss": 0.3894, "step": 22047 }, { "epoch": 0.785506884942195, "grad_norm": 1.1253691302542117, "learning_rate": 2.3173494287795405e-06, "loss": 0.3519, "step": 22048 }, { "epoch": 0.7855425120686891, "grad_norm": 1.6173927987860253, "learning_rate": 2.316610808284435e-06, "loss": 0.5619, "step": 22049 }, { "epoch": 0.7855781391951833, "grad_norm": 1.5635467583611342, "learning_rate": 2.315872290099026e-06, "loss": 0.4852, "step": 22050 }, { "epoch": 0.7856137663216773, "grad_norm": 1.3049261334943396, "learning_rate": 2.3151338742331496e-06, "loss": 0.5241, "step": 22051 }, { "epoch": 0.7856493934481714, "grad_norm": 1.4395782090463909, "learning_rate": 2.3143955606966382e-06, "loss": 0.5051, "step": 22052 }, { "epoch": 0.7856850205746656, "grad_norm": 1.196244409236831, "learning_rate": 2.313657349499321e-06, "loss": 0.4283, "step": 22053 }, { "epoch": 0.7857206477011597, "grad_norm": 1.4794619073883208, "learning_rate": 2.31291924065103e-06, "loss": 0.4721, "step": 22054 }, { "epoch": 0.7857562748276538, "grad_norm": 1.1396534081411536, "learning_rate": 2.31218123416159e-06, "loss": 0.4262, "step": 22055 }, { "epoch": 0.7857919019541479, "grad_norm": 1.4092994775425491, "learning_rate": 2.3114433300408302e-06, "loss": 0.4852, "step": 22056 }, { "epoch": 0.785827529080642, "grad_norm": 1.1827514891217354, "learning_rate": 2.3107055282985756e-06, "loss": 0.4043, "step": 22057 }, { "epoch": 0.7858631562071361, "grad_norm": 1.2274001347667574, "learning_rate": 2.3099678289446527e-06, "loss": 0.4285, "step": 22058 }, { "epoch": 0.7858987833336303, "grad_norm": 1.215820910551869, "learning_rate": 2.309230231988885e-06, "loss": 0.4588, "step": 22059 }, { "epoch": 0.7859344104601244, "grad_norm": 1.5611577264315317, "learning_rate": 2.3084927374410905e-06, "loss": 0.5158, "step": 22060 }, { "epoch": 0.7859700375866184, "grad_norm": 1.9532712940972135, "learning_rate": 2.307755345311092e-06, "loss": 0.5419, "step": 22061 }, { "epoch": 0.7860056647131126, "grad_norm": 1.3868808859763395, "learning_rate": 2.3070180556087076e-06, "loss": 0.5101, "step": 22062 }, { "epoch": 0.7860412918396067, "grad_norm": 1.1796467207398391, "learning_rate": 2.3062808683437555e-06, "loss": 0.3844, "step": 22063 }, { "epoch": 0.7860769189661008, "grad_norm": 2.0050960539033635, "learning_rate": 2.3055437835260497e-06, "loss": 0.5311, "step": 22064 }, { "epoch": 0.786112546092595, "grad_norm": 2.004037038795611, "learning_rate": 2.3048068011654056e-06, "loss": 0.5446, "step": 22065 }, { "epoch": 0.786148173219089, "grad_norm": 1.1599370855100404, "learning_rate": 2.304069921271639e-06, "loss": 0.419, "step": 22066 }, { "epoch": 0.7861838003455831, "grad_norm": 1.5356535370873612, "learning_rate": 2.3033331438545593e-06, "loss": 0.4476, "step": 22067 }, { "epoch": 0.7862194274720773, "grad_norm": 1.4645400904809251, "learning_rate": 2.302596468923981e-06, "loss": 0.5237, "step": 22068 }, { "epoch": 0.7862550545985714, "grad_norm": 1.8082502705567791, "learning_rate": 2.301859896489709e-06, "loss": 0.4501, "step": 22069 }, { "epoch": 0.7862906817250654, "grad_norm": 1.2846877428834587, "learning_rate": 2.3011234265615557e-06, "loss": 0.3949, "step": 22070 }, { "epoch": 0.7863263088515596, "grad_norm": 2.056515009209613, "learning_rate": 2.3003870591493215e-06, "loss": 0.4627, "step": 22071 }, { "epoch": 0.7863619359780537, "grad_norm": 1.3767609926857345, "learning_rate": 2.2996507942628167e-06, "loss": 0.4625, "step": 22072 }, { "epoch": 0.7863975631045478, "grad_norm": 1.2129643803168266, "learning_rate": 2.2989146319118428e-06, "loss": 0.3993, "step": 22073 }, { "epoch": 0.786433190231042, "grad_norm": 1.2999490169050685, "learning_rate": 2.2981785721062066e-06, "loss": 0.4391, "step": 22074 }, { "epoch": 0.786468817357536, "grad_norm": 1.2826896418874982, "learning_rate": 2.2974426148557028e-06, "loss": 0.4052, "step": 22075 }, { "epoch": 0.7865044444840301, "grad_norm": 1.8430022406788498, "learning_rate": 2.2967067601701375e-06, "loss": 0.4577, "step": 22076 }, { "epoch": 0.7865400716105243, "grad_norm": 1.1961495241149926, "learning_rate": 2.2959710080593033e-06, "loss": 0.4028, "step": 22077 }, { "epoch": 0.7865756987370184, "grad_norm": 1.152060777183626, "learning_rate": 2.2952353585330025e-06, "loss": 0.4031, "step": 22078 }, { "epoch": 0.7866113258635125, "grad_norm": 1.5699521492653123, "learning_rate": 2.294499811601022e-06, "loss": 0.4383, "step": 22079 }, { "epoch": 0.7866469529900066, "grad_norm": 1.1372731530723328, "learning_rate": 2.2937643672731656e-06, "loss": 0.4081, "step": 22080 }, { "epoch": 0.7866825801165007, "grad_norm": 1.1378113329975048, "learning_rate": 2.2930290255592256e-06, "loss": 0.4346, "step": 22081 }, { "epoch": 0.7867182072429948, "grad_norm": 1.1740900485964374, "learning_rate": 2.2922937864689888e-06, "loss": 0.3733, "step": 22082 }, { "epoch": 0.786753834369489, "grad_norm": 1.1044915723855135, "learning_rate": 2.29155865001225e-06, "loss": 0.3513, "step": 22083 }, { "epoch": 0.7867894614959831, "grad_norm": 1.1422162471728075, "learning_rate": 2.290823616198793e-06, "loss": 0.3785, "step": 22084 }, { "epoch": 0.7868250886224771, "grad_norm": 1.4235748184140924, "learning_rate": 2.2900886850384107e-06, "loss": 0.4683, "step": 22085 }, { "epoch": 0.7868607157489713, "grad_norm": 1.9328779239773854, "learning_rate": 2.2893538565408834e-06, "loss": 0.5455, "step": 22086 }, { "epoch": 0.7868963428754654, "grad_norm": 1.4426816138956848, "learning_rate": 2.2886191307159988e-06, "loss": 0.4971, "step": 22087 }, { "epoch": 0.7869319700019595, "grad_norm": 1.2801653994322537, "learning_rate": 2.2878845075735412e-06, "loss": 0.4674, "step": 22088 }, { "epoch": 0.7869675971284537, "grad_norm": 1.1195444952379032, "learning_rate": 2.287149987123294e-06, "loss": 0.393, "step": 22089 }, { "epoch": 0.7870032242549477, "grad_norm": 1.477048261072406, "learning_rate": 2.2864155693750334e-06, "loss": 0.5196, "step": 22090 }, { "epoch": 0.7870388513814418, "grad_norm": 1.6973641853514274, "learning_rate": 2.28568125433854e-06, "loss": 0.467, "step": 22091 }, { "epoch": 0.787074478507936, "grad_norm": 1.1810053683073485, "learning_rate": 2.2849470420235963e-06, "loss": 0.3725, "step": 22092 }, { "epoch": 0.7871101056344301, "grad_norm": 1.203017044567399, "learning_rate": 2.284212932439972e-06, "loss": 0.432, "step": 22093 }, { "epoch": 0.7871457327609241, "grad_norm": 1.4689780198130495, "learning_rate": 2.2834789255974466e-06, "loss": 0.4705, "step": 22094 }, { "epoch": 0.7871813598874183, "grad_norm": 1.501099627979217, "learning_rate": 2.2827450215057912e-06, "loss": 0.5194, "step": 22095 }, { "epoch": 0.7872169870139124, "grad_norm": 1.4292310810393352, "learning_rate": 2.2820112201747835e-06, "loss": 0.4935, "step": 22096 }, { "epoch": 0.7872526141404065, "grad_norm": 1.019365148183776, "learning_rate": 2.281277521614188e-06, "loss": 0.392, "step": 22097 }, { "epoch": 0.7872882412669007, "grad_norm": 1.496755192888161, "learning_rate": 2.2805439258337803e-06, "loss": 0.502, "step": 22098 }, { "epoch": 0.7873238683933947, "grad_norm": 1.1708487754314447, "learning_rate": 2.2798104328433237e-06, "loss": 0.3845, "step": 22099 }, { "epoch": 0.7873594955198888, "grad_norm": 1.1464359621935611, "learning_rate": 2.2790770426525888e-06, "loss": 0.3976, "step": 22100 }, { "epoch": 0.787395122646383, "grad_norm": 1.2701929678861519, "learning_rate": 2.278343755271337e-06, "loss": 0.3787, "step": 22101 }, { "epoch": 0.7874307497728771, "grad_norm": 1.1321893363516171, "learning_rate": 2.2776105707093368e-06, "loss": 0.3974, "step": 22102 }, { "epoch": 0.7874663768993712, "grad_norm": 1.2042439092971966, "learning_rate": 2.276877488976348e-06, "loss": 0.3892, "step": 22103 }, { "epoch": 0.7875020040258653, "grad_norm": 1.2504442818444552, "learning_rate": 2.2761445100821334e-06, "loss": 0.4072, "step": 22104 }, { "epoch": 0.7875376311523594, "grad_norm": 1.2008132969211973, "learning_rate": 2.275411634036456e-06, "loss": 0.4121, "step": 22105 }, { "epoch": 0.7875732582788535, "grad_norm": 1.8182809479256972, "learning_rate": 2.27467886084907e-06, "loss": 0.5285, "step": 22106 }, { "epoch": 0.7876088854053477, "grad_norm": 1.3301960686168057, "learning_rate": 2.273946190529738e-06, "loss": 0.4495, "step": 22107 }, { "epoch": 0.7876445125318418, "grad_norm": 1.3093485421340851, "learning_rate": 2.2732136230882084e-06, "loss": 0.4076, "step": 22108 }, { "epoch": 0.7876801396583358, "grad_norm": 1.7297975568126773, "learning_rate": 2.272481158534241e-06, "loss": 0.509, "step": 22109 }, { "epoch": 0.78771576678483, "grad_norm": 1.5356535637696718, "learning_rate": 2.2717487968775877e-06, "loss": 0.4554, "step": 22110 }, { "epoch": 0.7877513939113241, "grad_norm": 1.2780120327319253, "learning_rate": 2.2710165381280036e-06, "loss": 0.4751, "step": 22111 }, { "epoch": 0.7877870210378182, "grad_norm": 1.7060708413734873, "learning_rate": 2.2702843822952337e-06, "loss": 0.5154, "step": 22112 }, { "epoch": 0.7878226481643124, "grad_norm": 2.364542878653559, "learning_rate": 2.269552329389031e-06, "loss": 0.4852, "step": 22113 }, { "epoch": 0.7878582752908064, "grad_norm": 2.335567186334282, "learning_rate": 2.268820379419144e-06, "loss": 0.4642, "step": 22114 }, { "epoch": 0.7878939024173005, "grad_norm": 1.196656709677512, "learning_rate": 2.268088532395315e-06, "loss": 0.4069, "step": 22115 }, { "epoch": 0.7879295295437947, "grad_norm": 1.6132649497851776, "learning_rate": 2.267356788327294e-06, "loss": 0.5631, "step": 22116 }, { "epoch": 0.7879651566702888, "grad_norm": 2.4507424276262437, "learning_rate": 2.266625147224817e-06, "loss": 0.5499, "step": 22117 }, { "epoch": 0.7880007837967828, "grad_norm": 1.2260197674272397, "learning_rate": 2.265893609097637e-06, "loss": 0.3873, "step": 22118 }, { "epoch": 0.788036410923277, "grad_norm": 1.152524989061263, "learning_rate": 2.2651621739554864e-06, "loss": 0.4093, "step": 22119 }, { "epoch": 0.7880720380497711, "grad_norm": 1.2956569003100564, "learning_rate": 2.2644308418081108e-06, "loss": 0.3826, "step": 22120 }, { "epoch": 0.7881076651762652, "grad_norm": 1.1214733710705156, "learning_rate": 2.263699612665242e-06, "loss": 0.3963, "step": 22121 }, { "epoch": 0.7881432923027594, "grad_norm": 1.3441084217223151, "learning_rate": 2.262968486536624e-06, "loss": 0.5128, "step": 22122 }, { "epoch": 0.7881789194292534, "grad_norm": 1.1442768097265272, "learning_rate": 2.2622374634319853e-06, "loss": 0.3805, "step": 22123 }, { "epoch": 0.7882145465557475, "grad_norm": 1.5858144521515078, "learning_rate": 2.2615065433610626e-06, "loss": 0.5282, "step": 22124 }, { "epoch": 0.7882501736822417, "grad_norm": 1.2001526983553563, "learning_rate": 2.2607757263335894e-06, "loss": 0.4249, "step": 22125 }, { "epoch": 0.7882858008087358, "grad_norm": 1.850084794994197, "learning_rate": 2.2600450123592976e-06, "loss": 0.4907, "step": 22126 }, { "epoch": 0.7883214279352299, "grad_norm": 1.524143050619379, "learning_rate": 2.2593144014479184e-06, "loss": 0.5747, "step": 22127 }, { "epoch": 0.788357055061724, "grad_norm": 1.59578264424135, "learning_rate": 2.2585838936091753e-06, "loss": 0.5339, "step": 22128 }, { "epoch": 0.7883926821882181, "grad_norm": 1.1680758670050368, "learning_rate": 2.2578534888528024e-06, "loss": 0.383, "step": 22129 }, { "epoch": 0.7884283093147122, "grad_norm": 1.5418138271532404, "learning_rate": 2.257123187188518e-06, "loss": 0.4625, "step": 22130 }, { "epoch": 0.7884639364412064, "grad_norm": 1.3772366429181164, "learning_rate": 2.2563929886260506e-06, "loss": 0.4369, "step": 22131 }, { "epoch": 0.7884995635677005, "grad_norm": 1.2161520498205578, "learning_rate": 2.255662893175122e-06, "loss": 0.4458, "step": 22132 }, { "epoch": 0.7885351906941945, "grad_norm": 1.4081481740456225, "learning_rate": 2.25493290084546e-06, "loss": 0.4529, "step": 22133 }, { "epoch": 0.7885708178206887, "grad_norm": 1.1584603677611418, "learning_rate": 2.254203011646775e-06, "loss": 0.3739, "step": 22134 }, { "epoch": 0.7886064449471828, "grad_norm": 1.4689413519823984, "learning_rate": 2.253473225588795e-06, "loss": 0.5052, "step": 22135 }, { "epoch": 0.7886420720736769, "grad_norm": 1.6166438455039973, "learning_rate": 2.2527435426812307e-06, "loss": 0.5841, "step": 22136 }, { "epoch": 0.7886776992001711, "grad_norm": 1.1460661122298217, "learning_rate": 2.2520139629338e-06, "loss": 0.4312, "step": 22137 }, { "epoch": 0.7887133263266651, "grad_norm": 1.3446544769916604, "learning_rate": 2.251284486356222e-06, "loss": 0.4329, "step": 22138 }, { "epoch": 0.7887489534531592, "grad_norm": 2.0136742092418687, "learning_rate": 2.2505551129582047e-06, "loss": 0.5822, "step": 22139 }, { "epoch": 0.7887845805796534, "grad_norm": 1.2579979111547432, "learning_rate": 2.2498258427494636e-06, "loss": 0.4592, "step": 22140 }, { "epoch": 0.7888202077061475, "grad_norm": 1.0991010740184848, "learning_rate": 2.249096675739707e-06, "loss": 0.3991, "step": 22141 }, { "epoch": 0.7888558348326415, "grad_norm": 2.3992217536366716, "learning_rate": 2.2483676119386487e-06, "loss": 0.5059, "step": 22142 }, { "epoch": 0.7888914619591357, "grad_norm": 1.1356579189678293, "learning_rate": 2.247638651355991e-06, "loss": 0.4085, "step": 22143 }, { "epoch": 0.7889270890856298, "grad_norm": 1.7123547579022456, "learning_rate": 2.2469097940014474e-06, "loss": 0.5683, "step": 22144 }, { "epoch": 0.7889627162121239, "grad_norm": 1.0899540146146598, "learning_rate": 2.2461810398847138e-06, "loss": 0.3984, "step": 22145 }, { "epoch": 0.7889983433386181, "grad_norm": 1.5461788088485802, "learning_rate": 2.245452389015501e-06, "loss": 0.5276, "step": 22146 }, { "epoch": 0.7890339704651121, "grad_norm": 1.1555005324842458, "learning_rate": 2.244723841403509e-06, "loss": 0.3915, "step": 22147 }, { "epoch": 0.7890695975916062, "grad_norm": 1.298811894021113, "learning_rate": 2.2439953970584404e-06, "loss": 0.4301, "step": 22148 }, { "epoch": 0.7891052247181004, "grad_norm": 1.3201316024779093, "learning_rate": 2.2432670559899962e-06, "loss": 0.4966, "step": 22149 }, { "epoch": 0.7891408518445945, "grad_norm": 1.1974281382541478, "learning_rate": 2.24253881820787e-06, "loss": 0.4382, "step": 22150 }, { "epoch": 0.7891764789710886, "grad_norm": 1.4544401802182185, "learning_rate": 2.2418106837217646e-06, "loss": 0.5102, "step": 22151 }, { "epoch": 0.7892121060975827, "grad_norm": 1.1915656938494306, "learning_rate": 2.24108265254137e-06, "loss": 0.3914, "step": 22152 }, { "epoch": 0.7892477332240768, "grad_norm": 1.2444392127228898, "learning_rate": 2.2403547246763857e-06, "loss": 0.4288, "step": 22153 }, { "epoch": 0.7892833603505709, "grad_norm": 1.244085511238952, "learning_rate": 2.239626900136496e-06, "loss": 0.3971, "step": 22154 }, { "epoch": 0.7893189874770651, "grad_norm": 1.3932774746757115, "learning_rate": 2.2388991789314043e-06, "loss": 0.4722, "step": 22155 }, { "epoch": 0.7893546146035592, "grad_norm": 1.2177963666627485, "learning_rate": 2.238171561070792e-06, "loss": 0.4123, "step": 22156 }, { "epoch": 0.7893902417300532, "grad_norm": 1.259628268409304, "learning_rate": 2.2374440465643533e-06, "loss": 0.3978, "step": 22157 }, { "epoch": 0.7894258688565474, "grad_norm": 1.208512516201401, "learning_rate": 2.2367166354217695e-06, "loss": 0.4389, "step": 22158 }, { "epoch": 0.7894614959830415, "grad_norm": 1.1099624460053168, "learning_rate": 2.2359893276527335e-06, "loss": 0.3869, "step": 22159 }, { "epoch": 0.7894971231095356, "grad_norm": 1.1713856642193363, "learning_rate": 2.2352621232669225e-06, "loss": 0.4427, "step": 22160 }, { "epoch": 0.7895327502360298, "grad_norm": 1.3493560969442338, "learning_rate": 2.2345350222740246e-06, "loss": 0.502, "step": 22161 }, { "epoch": 0.7895683773625238, "grad_norm": 1.1785480599526057, "learning_rate": 2.2338080246837203e-06, "loss": 0.414, "step": 22162 }, { "epoch": 0.7896040044890179, "grad_norm": 1.1295780829967783, "learning_rate": 2.23308113050569e-06, "loss": 0.396, "step": 22163 }, { "epoch": 0.7896396316155121, "grad_norm": 1.2071758233424124, "learning_rate": 2.232354339749616e-06, "loss": 0.4507, "step": 22164 }, { "epoch": 0.7896752587420062, "grad_norm": 1.4512338180123732, "learning_rate": 2.231627652425171e-06, "loss": 0.4905, "step": 22165 }, { "epoch": 0.7897108858685002, "grad_norm": 1.1742329733498387, "learning_rate": 2.2309010685420365e-06, "loss": 0.3536, "step": 22166 }, { "epoch": 0.7897465129949944, "grad_norm": 1.3033395868545206, "learning_rate": 2.23017458810988e-06, "loss": 0.5053, "step": 22167 }, { "epoch": 0.7897821401214885, "grad_norm": 1.9514932380163925, "learning_rate": 2.2294482111383817e-06, "loss": 0.5877, "step": 22168 }, { "epoch": 0.7898177672479826, "grad_norm": 2.0478581861248384, "learning_rate": 2.2287219376372104e-06, "loss": 0.561, "step": 22169 }, { "epoch": 0.7898533943744768, "grad_norm": 1.068958494217097, "learning_rate": 2.2279957676160425e-06, "loss": 0.4292, "step": 22170 }, { "epoch": 0.7898890215009708, "grad_norm": 1.2016406749113027, "learning_rate": 2.227269701084539e-06, "loss": 0.4216, "step": 22171 }, { "epoch": 0.7899246486274649, "grad_norm": 1.4641292601103373, "learning_rate": 2.2265437380523736e-06, "loss": 0.4671, "step": 22172 }, { "epoch": 0.7899602757539591, "grad_norm": 2.640031452524128, "learning_rate": 2.225817878529214e-06, "loss": 0.4816, "step": 22173 }, { "epoch": 0.7899959028804532, "grad_norm": 1.5958050248693982, "learning_rate": 2.2250921225247192e-06, "loss": 0.5063, "step": 22174 }, { "epoch": 0.7900315300069473, "grad_norm": 1.130570111563176, "learning_rate": 2.224366470048561e-06, "loss": 0.3763, "step": 22175 }, { "epoch": 0.7900671571334414, "grad_norm": 1.7419773488292398, "learning_rate": 2.223640921110395e-06, "loss": 0.5416, "step": 22176 }, { "epoch": 0.7901027842599355, "grad_norm": 1.2380402412277156, "learning_rate": 2.2229154757198855e-06, "loss": 0.3838, "step": 22177 }, { "epoch": 0.7901384113864296, "grad_norm": 1.1611956150765963, "learning_rate": 2.2221901338866925e-06, "loss": 0.4225, "step": 22178 }, { "epoch": 0.7901740385129238, "grad_norm": 1.245082835336362, "learning_rate": 2.2214648956204775e-06, "loss": 0.4179, "step": 22179 }, { "epoch": 0.7902096656394179, "grad_norm": 1.1768305035477793, "learning_rate": 2.2207397609308913e-06, "loss": 0.4076, "step": 22180 }, { "epoch": 0.7902452927659119, "grad_norm": 1.8746718373030773, "learning_rate": 2.2200147298275943e-06, "loss": 0.5244, "step": 22181 }, { "epoch": 0.7902809198924061, "grad_norm": 1.422120387901694, "learning_rate": 2.219289802320237e-06, "loss": 0.4305, "step": 22182 }, { "epoch": 0.7903165470189002, "grad_norm": 1.2100908313923548, "learning_rate": 2.218564978418475e-06, "loss": 0.4249, "step": 22183 }, { "epoch": 0.7903521741453943, "grad_norm": 2.821888832202367, "learning_rate": 2.217840258131958e-06, "loss": 0.467, "step": 22184 }, { "epoch": 0.7903878012718885, "grad_norm": 1.5737467724021599, "learning_rate": 2.2171156414703377e-06, "loss": 0.4924, "step": 22185 }, { "epoch": 0.7904234283983825, "grad_norm": 1.946511883592582, "learning_rate": 2.216391128443266e-06, "loss": 0.4924, "step": 22186 }, { "epoch": 0.7904590555248766, "grad_norm": 1.5982690073604546, "learning_rate": 2.215666719060383e-06, "loss": 0.4872, "step": 22187 }, { "epoch": 0.7904946826513708, "grad_norm": 1.8611053223622749, "learning_rate": 2.2149424133313425e-06, "loss": 0.5017, "step": 22188 }, { "epoch": 0.7905303097778649, "grad_norm": 1.6111081562196947, "learning_rate": 2.2142182112657816e-06, "loss": 0.5459, "step": 22189 }, { "epoch": 0.790565936904359, "grad_norm": 1.3647684197315986, "learning_rate": 2.21349411287335e-06, "loss": 0.4961, "step": 22190 }, { "epoch": 0.7906015640308531, "grad_norm": 1.1573346624015717, "learning_rate": 2.2127701181636853e-06, "loss": 0.4382, "step": 22191 }, { "epoch": 0.7906371911573472, "grad_norm": 1.918902213772092, "learning_rate": 2.2120462271464273e-06, "loss": 0.4622, "step": 22192 }, { "epoch": 0.7906728182838413, "grad_norm": 1.463486695584827, "learning_rate": 2.2113224398312185e-06, "loss": 0.4282, "step": 22193 }, { "epoch": 0.7907084454103355, "grad_norm": 1.7850301478265482, "learning_rate": 2.2105987562276954e-06, "loss": 0.467, "step": 22194 }, { "epoch": 0.7907440725368295, "grad_norm": 1.407157166826106, "learning_rate": 2.2098751763454962e-06, "loss": 0.4739, "step": 22195 }, { "epoch": 0.7907796996633236, "grad_norm": 1.8398565879523379, "learning_rate": 2.209151700194252e-06, "loss": 0.4745, "step": 22196 }, { "epoch": 0.7908153267898178, "grad_norm": 1.2362064282919258, "learning_rate": 2.2084283277836015e-06, "loss": 0.5201, "step": 22197 }, { "epoch": 0.7908509539163119, "grad_norm": 1.394447602572702, "learning_rate": 2.2077050591231695e-06, "loss": 0.4637, "step": 22198 }, { "epoch": 0.790886581042806, "grad_norm": 1.598178592190232, "learning_rate": 2.206981894222593e-06, "loss": 0.4773, "step": 22199 }, { "epoch": 0.7909222081693001, "grad_norm": 1.5037356441494818, "learning_rate": 2.206258833091498e-06, "loss": 0.5771, "step": 22200 }, { "epoch": 0.7909578352957942, "grad_norm": 2.9907177077612483, "learning_rate": 2.2055358757395183e-06, "loss": 0.5653, "step": 22201 }, { "epoch": 0.7909934624222883, "grad_norm": 1.6507177047541333, "learning_rate": 2.204813022176273e-06, "loss": 0.4057, "step": 22202 }, { "epoch": 0.7910290895487825, "grad_norm": 1.117828430941551, "learning_rate": 2.2040902724113934e-06, "loss": 0.399, "step": 22203 }, { "epoch": 0.7910647166752766, "grad_norm": 1.6484112969326896, "learning_rate": 2.203367626454498e-06, "loss": 0.5665, "step": 22204 }, { "epoch": 0.7911003438017706, "grad_norm": 1.655187878980319, "learning_rate": 2.2026450843152156e-06, "loss": 0.6235, "step": 22205 }, { "epoch": 0.7911359709282648, "grad_norm": 1.2483723746405606, "learning_rate": 2.201922646003157e-06, "loss": 0.3917, "step": 22206 }, { "epoch": 0.7911715980547589, "grad_norm": 1.170829924042233, "learning_rate": 2.2012003115279533e-06, "loss": 0.3605, "step": 22207 }, { "epoch": 0.791207225181253, "grad_norm": 1.4683386217271783, "learning_rate": 2.200478080899221e-06, "loss": 0.5362, "step": 22208 }, { "epoch": 0.7912428523077472, "grad_norm": 1.1633098648307596, "learning_rate": 2.199755954126571e-06, "loss": 0.3881, "step": 22209 }, { "epoch": 0.7912784794342412, "grad_norm": 2.2463932434508034, "learning_rate": 2.1990339312196263e-06, "loss": 0.5183, "step": 22210 }, { "epoch": 0.7913141065607353, "grad_norm": 1.3259060731859018, "learning_rate": 2.198312012187993e-06, "loss": 0.4535, "step": 22211 }, { "epoch": 0.7913497336872295, "grad_norm": 1.243310182719733, "learning_rate": 2.1975901970412927e-06, "loss": 0.436, "step": 22212 }, { "epoch": 0.7913853608137236, "grad_norm": 1.2232453736198676, "learning_rate": 2.1968684857891285e-06, "loss": 0.4231, "step": 22213 }, { "epoch": 0.7914209879402176, "grad_norm": 1.2043330488559112, "learning_rate": 2.1961468784411153e-06, "loss": 0.4638, "step": 22214 }, { "epoch": 0.7914566150667118, "grad_norm": 1.3329721898718625, "learning_rate": 2.195425375006861e-06, "loss": 0.4599, "step": 22215 }, { "epoch": 0.7914922421932059, "grad_norm": 1.2140208938614003, "learning_rate": 2.1947039754959755e-06, "loss": 0.4366, "step": 22216 }, { "epoch": 0.7915278693197, "grad_norm": 1.22500304203211, "learning_rate": 2.1939826799180584e-06, "loss": 0.4519, "step": 22217 }, { "epoch": 0.7915634964461942, "grad_norm": 1.4867523752661667, "learning_rate": 2.1932614882827196e-06, "loss": 0.508, "step": 22218 }, { "epoch": 0.7915991235726882, "grad_norm": 1.272717652790631, "learning_rate": 2.192540400599562e-06, "loss": 0.4517, "step": 22219 }, { "epoch": 0.7916347506991823, "grad_norm": 1.345183407143902, "learning_rate": 2.1918194168781847e-06, "loss": 0.4174, "step": 22220 }, { "epoch": 0.7916703778256765, "grad_norm": 1.5397482470522783, "learning_rate": 2.191098537128189e-06, "loss": 0.5002, "step": 22221 }, { "epoch": 0.7917060049521706, "grad_norm": 1.6987473538639852, "learning_rate": 2.1903777613591737e-06, "loss": 0.5285, "step": 22222 }, { "epoch": 0.7917416320786647, "grad_norm": 1.368369885750249, "learning_rate": 2.1896570895807403e-06, "loss": 0.5004, "step": 22223 }, { "epoch": 0.7917772592051588, "grad_norm": 1.233771312298083, "learning_rate": 2.1889365218024796e-06, "loss": 0.4529, "step": 22224 }, { "epoch": 0.7918128863316529, "grad_norm": 1.1582270144526767, "learning_rate": 2.1882160580339907e-06, "loss": 0.4116, "step": 22225 }, { "epoch": 0.791848513458147, "grad_norm": 1.0935752285755829, "learning_rate": 2.1874956982848628e-06, "loss": 0.3781, "step": 22226 }, { "epoch": 0.7918841405846412, "grad_norm": 1.4608824871062953, "learning_rate": 2.1867754425646923e-06, "loss": 0.5889, "step": 22227 }, { "epoch": 0.7919197677111353, "grad_norm": 1.100447581437407, "learning_rate": 2.1860552908830657e-06, "loss": 0.4062, "step": 22228 }, { "epoch": 0.7919553948376293, "grad_norm": 1.150015254450188, "learning_rate": 2.1853352432495724e-06, "loss": 0.4603, "step": 22229 }, { "epoch": 0.7919910219641235, "grad_norm": 1.2245432402669947, "learning_rate": 2.184615299673807e-06, "loss": 0.4725, "step": 22230 }, { "epoch": 0.7920266490906176, "grad_norm": 1.1495507873757198, "learning_rate": 2.1838954601653485e-06, "loss": 0.4116, "step": 22231 }, { "epoch": 0.7920622762171117, "grad_norm": 1.2415925205797813, "learning_rate": 2.1831757247337895e-06, "loss": 0.4672, "step": 22232 }, { "epoch": 0.7920979033436059, "grad_norm": 1.3412100246984489, "learning_rate": 2.1824560933887053e-06, "loss": 0.462, "step": 22233 }, { "epoch": 0.7921335304700999, "grad_norm": 1.1825118092942388, "learning_rate": 2.181736566139685e-06, "loss": 0.372, "step": 22234 }, { "epoch": 0.792169157596594, "grad_norm": 1.907501763083151, "learning_rate": 2.1810171429963044e-06, "loss": 0.4723, "step": 22235 }, { "epoch": 0.7922047847230882, "grad_norm": 2.5690755910798027, "learning_rate": 2.180297823968146e-06, "loss": 0.4329, "step": 22236 }, { "epoch": 0.7922404118495823, "grad_norm": 1.5861912735721526, "learning_rate": 2.1795786090647875e-06, "loss": 0.5246, "step": 22237 }, { "epoch": 0.7922760389760763, "grad_norm": 1.2071662065098197, "learning_rate": 2.1788594982958087e-06, "loss": 0.4152, "step": 22238 }, { "epoch": 0.7923116661025705, "grad_norm": 1.1862555461266042, "learning_rate": 2.1781404916707793e-06, "loss": 0.4193, "step": 22239 }, { "epoch": 0.7923472932290646, "grad_norm": 1.7660109409886866, "learning_rate": 2.1774215891992767e-06, "loss": 0.5042, "step": 22240 }, { "epoch": 0.7923829203555587, "grad_norm": 1.6714396644559395, "learning_rate": 2.1767027908908767e-06, "loss": 0.4903, "step": 22241 }, { "epoch": 0.7924185474820529, "grad_norm": 1.4688415570888254, "learning_rate": 2.1759840967551438e-06, "loss": 0.4148, "step": 22242 }, { "epoch": 0.792454174608547, "grad_norm": 1.225011490833696, "learning_rate": 2.1752655068016516e-06, "loss": 0.434, "step": 22243 }, { "epoch": 0.792489801735041, "grad_norm": 1.4057473283856645, "learning_rate": 2.1745470210399676e-06, "loss": 0.4237, "step": 22244 }, { "epoch": 0.7925254288615352, "grad_norm": 1.316405121062433, "learning_rate": 2.1738286394796625e-06, "loss": 0.4528, "step": 22245 }, { "epoch": 0.7925610559880293, "grad_norm": 1.5890674483323437, "learning_rate": 2.1731103621302975e-06, "loss": 0.4916, "step": 22246 }, { "epoch": 0.7925966831145234, "grad_norm": 1.573923005011609, "learning_rate": 2.172392189001441e-06, "loss": 0.5858, "step": 22247 }, { "epoch": 0.7926323102410175, "grad_norm": 1.1997311446050969, "learning_rate": 2.1716741201026503e-06, "loss": 0.4169, "step": 22248 }, { "epoch": 0.7926679373675116, "grad_norm": 1.6767886729753103, "learning_rate": 2.1709561554434933e-06, "loss": 0.4106, "step": 22249 }, { "epoch": 0.7927035644940057, "grad_norm": 1.2011485153452832, "learning_rate": 2.1702382950335257e-06, "loss": 0.4273, "step": 22250 }, { "epoch": 0.7927391916204999, "grad_norm": 1.08996097514467, "learning_rate": 2.169520538882307e-06, "loss": 0.3932, "step": 22251 }, { "epoch": 0.792774818746994, "grad_norm": 1.1620394482704184, "learning_rate": 2.168802886999395e-06, "loss": 0.4073, "step": 22252 }, { "epoch": 0.792810445873488, "grad_norm": 1.1237928103512906, "learning_rate": 2.1680853393943467e-06, "loss": 0.4256, "step": 22253 }, { "epoch": 0.7928460729999822, "grad_norm": 1.5375135271301874, "learning_rate": 2.16736789607672e-06, "loss": 0.5483, "step": 22254 }, { "epoch": 0.7928817001264763, "grad_norm": 1.8256633648496294, "learning_rate": 2.1666505570560605e-06, "loss": 0.5536, "step": 22255 }, { "epoch": 0.7929173272529704, "grad_norm": 1.6068394542578697, "learning_rate": 2.1659333223419267e-06, "loss": 0.4696, "step": 22256 }, { "epoch": 0.7929529543794646, "grad_norm": 1.395308424309485, "learning_rate": 2.165216191943864e-06, "loss": 0.3988, "step": 22257 }, { "epoch": 0.7929885815059586, "grad_norm": 1.2059486500267167, "learning_rate": 2.1644991658714233e-06, "loss": 0.3603, "step": 22258 }, { "epoch": 0.7930242086324527, "grad_norm": 1.5208369448527501, "learning_rate": 2.163782244134154e-06, "loss": 0.5627, "step": 22259 }, { "epoch": 0.7930598357589469, "grad_norm": 1.2457156589470944, "learning_rate": 2.163065426741603e-06, "loss": 0.3888, "step": 22260 }, { "epoch": 0.793095462885441, "grad_norm": 1.2898073462328836, "learning_rate": 2.162348713703312e-06, "loss": 0.456, "step": 22261 }, { "epoch": 0.793131090011935, "grad_norm": 1.0893066507338414, "learning_rate": 2.161632105028828e-06, "loss": 0.4022, "step": 22262 }, { "epoch": 0.7931667171384292, "grad_norm": 1.297583990965679, "learning_rate": 2.160915600727688e-06, "loss": 0.4712, "step": 22263 }, { "epoch": 0.7932023442649233, "grad_norm": 1.3520736449199824, "learning_rate": 2.1601992008094363e-06, "loss": 0.4958, "step": 22264 }, { "epoch": 0.7932379713914174, "grad_norm": 1.2063947014094873, "learning_rate": 2.1594829052836154e-06, "loss": 0.3762, "step": 22265 }, { "epoch": 0.7932735985179116, "grad_norm": 1.1654125149505996, "learning_rate": 2.158766714159757e-06, "loss": 0.3364, "step": 22266 }, { "epoch": 0.7933092256444056, "grad_norm": 1.286968375609547, "learning_rate": 2.1580506274474e-06, "loss": 0.4444, "step": 22267 }, { "epoch": 0.7933448527708997, "grad_norm": 1.325584708357765, "learning_rate": 2.1573346451560797e-06, "loss": 0.4836, "step": 22268 }, { "epoch": 0.7933804798973939, "grad_norm": 1.1476731733047847, "learning_rate": 2.156618767295333e-06, "loss": 0.4068, "step": 22269 }, { "epoch": 0.793416107023888, "grad_norm": 1.1838053544304628, "learning_rate": 2.155902993874688e-06, "loss": 0.4083, "step": 22270 }, { "epoch": 0.7934517341503821, "grad_norm": 1.6373459320635053, "learning_rate": 2.155187324903679e-06, "loss": 0.5183, "step": 22271 }, { "epoch": 0.7934873612768762, "grad_norm": 1.265370881131509, "learning_rate": 2.154471760391832e-06, "loss": 0.4484, "step": 22272 }, { "epoch": 0.7935229884033703, "grad_norm": 1.2289336178498553, "learning_rate": 2.153756300348677e-06, "loss": 0.4427, "step": 22273 }, { "epoch": 0.7935586155298644, "grad_norm": 1.523834488685316, "learning_rate": 2.153040944783741e-06, "loss": 0.5512, "step": 22274 }, { "epoch": 0.7935942426563586, "grad_norm": 1.257574708394857, "learning_rate": 2.1523256937065496e-06, "loss": 0.4661, "step": 22275 }, { "epoch": 0.7936298697828527, "grad_norm": 1.2041552284766648, "learning_rate": 2.151610547126629e-06, "loss": 0.3648, "step": 22276 }, { "epoch": 0.7936654969093467, "grad_norm": 1.0914195070039205, "learning_rate": 2.1508955050534975e-06, "loss": 0.43, "step": 22277 }, { "epoch": 0.7937011240358409, "grad_norm": 1.6073079077522743, "learning_rate": 2.150180567496681e-06, "loss": 0.5067, "step": 22278 }, { "epoch": 0.793736751162335, "grad_norm": 1.9405112773509143, "learning_rate": 2.1494657344656944e-06, "loss": 0.525, "step": 22279 }, { "epoch": 0.7937723782888291, "grad_norm": 1.0601426035125492, "learning_rate": 2.1487510059700612e-06, "loss": 0.4306, "step": 22280 }, { "epoch": 0.7938080054153233, "grad_norm": 1.2164848471297574, "learning_rate": 2.148036382019291e-06, "loss": 0.4508, "step": 22281 }, { "epoch": 0.7938436325418173, "grad_norm": 1.16667904754494, "learning_rate": 2.14732186262291e-06, "loss": 0.393, "step": 22282 }, { "epoch": 0.7938792596683114, "grad_norm": 1.274135424450319, "learning_rate": 2.1466074477904244e-06, "loss": 0.4296, "step": 22283 }, { "epoch": 0.7939148867948056, "grad_norm": 1.566874820841883, "learning_rate": 2.145893137531352e-06, "loss": 0.4742, "step": 22284 }, { "epoch": 0.7939505139212997, "grad_norm": 1.5595522585093893, "learning_rate": 2.1451789318551998e-06, "loss": 0.5866, "step": 22285 }, { "epoch": 0.7939861410477937, "grad_norm": 1.2960324791478592, "learning_rate": 2.144464830771483e-06, "loss": 0.3997, "step": 22286 }, { "epoch": 0.7940217681742879, "grad_norm": 1.292645989267546, "learning_rate": 2.143750834289705e-06, "loss": 0.4242, "step": 22287 }, { "epoch": 0.794057395300782, "grad_norm": 1.2387523388141295, "learning_rate": 2.1430369424193754e-06, "loss": 0.4069, "step": 22288 }, { "epoch": 0.7940930224272761, "grad_norm": 1.2515627955893514, "learning_rate": 2.142323155170001e-06, "loss": 0.4772, "step": 22289 }, { "epoch": 0.7941286495537703, "grad_norm": 1.572465595999255, "learning_rate": 2.141609472551086e-06, "loss": 0.5015, "step": 22290 }, { "epoch": 0.7941642766802643, "grad_norm": 1.8060594459180779, "learning_rate": 2.140895894572136e-06, "loss": 0.5111, "step": 22291 }, { "epoch": 0.7941999038067584, "grad_norm": 1.1489726697906204, "learning_rate": 2.140182421242648e-06, "loss": 0.4236, "step": 22292 }, { "epoch": 0.7942355309332526, "grad_norm": 1.1982733411350608, "learning_rate": 2.1394690525721275e-06, "loss": 0.4171, "step": 22293 }, { "epoch": 0.7942711580597467, "grad_norm": 1.0477318452337117, "learning_rate": 2.1387557885700683e-06, "loss": 0.3927, "step": 22294 }, { "epoch": 0.7943067851862408, "grad_norm": 1.153121107574704, "learning_rate": 2.13804262924597e-06, "loss": 0.4093, "step": 22295 }, { "epoch": 0.794342412312735, "grad_norm": 1.2722893294854654, "learning_rate": 2.137329574609329e-06, "loss": 0.3995, "step": 22296 }, { "epoch": 0.794378039439229, "grad_norm": 1.1652057870265116, "learning_rate": 2.136616624669643e-06, "loss": 0.3815, "step": 22297 }, { "epoch": 0.7944136665657231, "grad_norm": 1.4491218736452507, "learning_rate": 2.1359037794364003e-06, "loss": 0.5596, "step": 22298 }, { "epoch": 0.7944492936922173, "grad_norm": 1.6745892187782234, "learning_rate": 2.1351910389190953e-06, "loss": 0.5358, "step": 22299 }, { "epoch": 0.7944849208187114, "grad_norm": 1.3821813762700923, "learning_rate": 2.1344784031272224e-06, "loss": 0.4128, "step": 22300 }, { "epoch": 0.7945205479452054, "grad_norm": 1.1315732445478703, "learning_rate": 2.1337658720702636e-06, "loss": 0.3733, "step": 22301 }, { "epoch": 0.7945561750716996, "grad_norm": 1.1709571054169934, "learning_rate": 2.1330534457577133e-06, "loss": 0.3861, "step": 22302 }, { "epoch": 0.7945918021981937, "grad_norm": 1.5771790587065038, "learning_rate": 2.132341124199053e-06, "loss": 0.5512, "step": 22303 }, { "epoch": 0.7946274293246878, "grad_norm": 1.855420271023554, "learning_rate": 2.1316289074037687e-06, "loss": 0.5325, "step": 22304 }, { "epoch": 0.794663056451182, "grad_norm": 1.1636312131962805, "learning_rate": 2.130916795381346e-06, "loss": 0.3804, "step": 22305 }, { "epoch": 0.794698683577676, "grad_norm": 1.5643868413636108, "learning_rate": 2.130204788141269e-06, "loss": 0.4758, "step": 22306 }, { "epoch": 0.7947343107041701, "grad_norm": 1.1782564434776852, "learning_rate": 2.129492885693013e-06, "loss": 0.4165, "step": 22307 }, { "epoch": 0.7947699378306643, "grad_norm": 1.1929964348702562, "learning_rate": 2.1287810880460636e-06, "loss": 0.412, "step": 22308 }, { "epoch": 0.7948055649571584, "grad_norm": 1.3329411637186448, "learning_rate": 2.128069395209892e-06, "loss": 0.44, "step": 22309 }, { "epoch": 0.7948411920836524, "grad_norm": 1.1515607140651696, "learning_rate": 2.127357807193979e-06, "loss": 0.3801, "step": 22310 }, { "epoch": 0.7948768192101466, "grad_norm": 1.2447490722268724, "learning_rate": 2.1266463240077995e-06, "loss": 0.456, "step": 22311 }, { "epoch": 0.7949124463366407, "grad_norm": 1.1569733896854384, "learning_rate": 2.125934945660827e-06, "loss": 0.4228, "step": 22312 }, { "epoch": 0.7949480734631348, "grad_norm": 1.1786772856494352, "learning_rate": 2.1252236721625375e-06, "loss": 0.4359, "step": 22313 }, { "epoch": 0.794983700589629, "grad_norm": 1.182653483348488, "learning_rate": 2.1245125035223956e-06, "loss": 0.4044, "step": 22314 }, { "epoch": 0.795019327716123, "grad_norm": 1.1929682362839382, "learning_rate": 2.1238014397498776e-06, "loss": 0.4214, "step": 22315 }, { "epoch": 0.7950549548426171, "grad_norm": 1.485018235178689, "learning_rate": 2.123090480854446e-06, "loss": 0.5276, "step": 22316 }, { "epoch": 0.7950905819691113, "grad_norm": 1.0736569351395202, "learning_rate": 2.1223796268455733e-06, "loss": 0.394, "step": 22317 }, { "epoch": 0.7951262090956054, "grad_norm": 1.2586178814427171, "learning_rate": 2.1216688777327153e-06, "loss": 0.4022, "step": 22318 }, { "epoch": 0.7951618362220995, "grad_norm": 1.1679461648767295, "learning_rate": 2.1209582335253486e-06, "loss": 0.4011, "step": 22319 }, { "epoch": 0.7951974633485936, "grad_norm": 1.1772092983189173, "learning_rate": 2.1202476942329286e-06, "loss": 0.4189, "step": 22320 }, { "epoch": 0.7952330904750877, "grad_norm": 1.32237493713906, "learning_rate": 2.1195372598649167e-06, "loss": 0.4312, "step": 22321 }, { "epoch": 0.7952687176015818, "grad_norm": 1.2535783483798766, "learning_rate": 2.1188269304307785e-06, "loss": 0.3777, "step": 22322 }, { "epoch": 0.795304344728076, "grad_norm": 1.5920812085400042, "learning_rate": 2.1181167059399643e-06, "loss": 0.4752, "step": 22323 }, { "epoch": 0.7953399718545701, "grad_norm": 1.4563842443591122, "learning_rate": 2.117406586401939e-06, "loss": 0.5434, "step": 22324 }, { "epoch": 0.7953755989810641, "grad_norm": 1.2810229943350846, "learning_rate": 2.116696571826151e-06, "loss": 0.4129, "step": 22325 }, { "epoch": 0.7954112261075583, "grad_norm": 1.814770910984648, "learning_rate": 2.115986662222058e-06, "loss": 0.5668, "step": 22326 }, { "epoch": 0.7954468532340524, "grad_norm": 1.9214651690640228, "learning_rate": 2.1152768575991146e-06, "loss": 0.5259, "step": 22327 }, { "epoch": 0.7954824803605465, "grad_norm": 1.1868831391664099, "learning_rate": 2.1145671579667727e-06, "loss": 0.3927, "step": 22328 }, { "epoch": 0.7955181074870407, "grad_norm": 1.4048437438682033, "learning_rate": 2.113857563334478e-06, "loss": 0.4574, "step": 22329 }, { "epoch": 0.7955537346135347, "grad_norm": 1.6779844928313823, "learning_rate": 2.113148073711685e-06, "loss": 0.5386, "step": 22330 }, { "epoch": 0.7955893617400288, "grad_norm": 2.3229337648122947, "learning_rate": 2.112438689107834e-06, "loss": 0.4631, "step": 22331 }, { "epoch": 0.795624988866523, "grad_norm": 1.767142906520336, "learning_rate": 2.111729409532377e-06, "loss": 0.5558, "step": 22332 }, { "epoch": 0.7956606159930171, "grad_norm": 1.2610436444602715, "learning_rate": 2.111020234994755e-06, "loss": 0.3973, "step": 22333 }, { "epoch": 0.7956962431195111, "grad_norm": 1.2807057849099726, "learning_rate": 2.110311165504413e-06, "loss": 0.4166, "step": 22334 }, { "epoch": 0.7957318702460053, "grad_norm": 1.8167247219005487, "learning_rate": 2.109602201070795e-06, "loss": 0.5437, "step": 22335 }, { "epoch": 0.7957674973724994, "grad_norm": 1.5683254651470384, "learning_rate": 2.1088933417033365e-06, "loss": 0.5006, "step": 22336 }, { "epoch": 0.7958031244989935, "grad_norm": 1.176212418852213, "learning_rate": 2.1081845874114815e-06, "loss": 0.3905, "step": 22337 }, { "epoch": 0.7958387516254877, "grad_norm": 2.0175053362560025, "learning_rate": 2.1074759382046617e-06, "loss": 0.4603, "step": 22338 }, { "epoch": 0.7958743787519817, "grad_norm": 2.185735982097827, "learning_rate": 2.1067673940923185e-06, "loss": 0.4359, "step": 22339 }, { "epoch": 0.7959100058784758, "grad_norm": 1.4839535068318523, "learning_rate": 2.106058955083883e-06, "loss": 0.5124, "step": 22340 }, { "epoch": 0.79594563300497, "grad_norm": 1.7456515839467752, "learning_rate": 2.1053506211887885e-06, "loss": 0.5043, "step": 22341 }, { "epoch": 0.7959812601314641, "grad_norm": 1.6147904682933347, "learning_rate": 2.1046423924164693e-06, "loss": 0.522, "step": 22342 }, { "epoch": 0.7960168872579582, "grad_norm": 1.7544672497558569, "learning_rate": 2.1039342687763587e-06, "loss": 0.5246, "step": 22343 }, { "epoch": 0.7960525143844523, "grad_norm": 1.1229336998225397, "learning_rate": 2.1032262502778787e-06, "loss": 0.3732, "step": 22344 }, { "epoch": 0.7960881415109464, "grad_norm": 1.5272318166375363, "learning_rate": 2.1025183369304604e-06, "loss": 0.43, "step": 22345 }, { "epoch": 0.7961237686374405, "grad_norm": 1.4228018584355737, "learning_rate": 2.101810528743534e-06, "loss": 0.4274, "step": 22346 }, { "epoch": 0.7961593957639347, "grad_norm": 1.2568355689555024, "learning_rate": 2.1011028257265176e-06, "loss": 0.411, "step": 22347 }, { "epoch": 0.7961950228904288, "grad_norm": 1.5577648505512232, "learning_rate": 2.1003952278888382e-06, "loss": 0.5237, "step": 22348 }, { "epoch": 0.7962306500169228, "grad_norm": 1.2502363966931929, "learning_rate": 2.099687735239918e-06, "loss": 0.4603, "step": 22349 }, { "epoch": 0.796266277143417, "grad_norm": 1.1811341193841953, "learning_rate": 2.0989803477891803e-06, "loss": 0.4195, "step": 22350 }, { "epoch": 0.7963019042699111, "grad_norm": 1.5173344757334153, "learning_rate": 2.0982730655460393e-06, "loss": 0.5213, "step": 22351 }, { "epoch": 0.7963375313964052, "grad_norm": 1.3247610551882725, "learning_rate": 2.097565888519917e-06, "loss": 0.4361, "step": 22352 }, { "epoch": 0.7963731585228994, "grad_norm": 1.5912232279760357, "learning_rate": 2.0968588167202265e-06, "loss": 0.4958, "step": 22353 }, { "epoch": 0.7964087856493934, "grad_norm": 1.155870715850831, "learning_rate": 2.096151850156388e-06, "loss": 0.3575, "step": 22354 }, { "epoch": 0.7964444127758875, "grad_norm": 1.6439217572672586, "learning_rate": 2.095444988837808e-06, "loss": 0.5227, "step": 22355 }, { "epoch": 0.7964800399023817, "grad_norm": 1.5637326486193863, "learning_rate": 2.094738232773902e-06, "loss": 0.4755, "step": 22356 }, { "epoch": 0.7965156670288758, "grad_norm": 1.6961287102513543, "learning_rate": 2.0940315819740865e-06, "loss": 0.5871, "step": 22357 }, { "epoch": 0.7965512941553698, "grad_norm": 1.708095096273223, "learning_rate": 2.093325036447763e-06, "loss": 0.581, "step": 22358 }, { "epoch": 0.796586921281864, "grad_norm": 1.5770804688107092, "learning_rate": 2.0926185962043465e-06, "loss": 0.4399, "step": 22359 }, { "epoch": 0.7966225484083581, "grad_norm": 1.4019942593659434, "learning_rate": 2.091912261253237e-06, "loss": 0.4643, "step": 22360 }, { "epoch": 0.7966581755348522, "grad_norm": 1.208442146131404, "learning_rate": 2.0912060316038463e-06, "loss": 0.4187, "step": 22361 }, { "epoch": 0.7966938026613464, "grad_norm": 1.195903172029667, "learning_rate": 2.090499907265573e-06, "loss": 0.4231, "step": 22362 }, { "epoch": 0.7967294297878404, "grad_norm": 1.4194919743811192, "learning_rate": 2.089793888247822e-06, "loss": 0.5005, "step": 22363 }, { "epoch": 0.7967650569143345, "grad_norm": 1.2518408444757434, "learning_rate": 2.0890879745599936e-06, "loss": 0.4463, "step": 22364 }, { "epoch": 0.7968006840408287, "grad_norm": 1.3664137568500574, "learning_rate": 2.0883821662114902e-06, "loss": 0.3953, "step": 22365 }, { "epoch": 0.7968363111673228, "grad_norm": 1.3594443547819843, "learning_rate": 2.0876764632117064e-06, "loss": 0.4103, "step": 22366 }, { "epoch": 0.7968719382938169, "grad_norm": 1.1862170590129908, "learning_rate": 2.0869708655700406e-06, "loss": 0.372, "step": 22367 }, { "epoch": 0.796907565420311, "grad_norm": 1.9087651361898839, "learning_rate": 2.0862653732958914e-06, "loss": 0.4346, "step": 22368 }, { "epoch": 0.7969431925468051, "grad_norm": 1.5806886126052508, "learning_rate": 2.0855599863986475e-06, "loss": 0.5306, "step": 22369 }, { "epoch": 0.7969788196732992, "grad_norm": 1.4217884292350116, "learning_rate": 2.0848547048877042e-06, "loss": 0.5424, "step": 22370 }, { "epoch": 0.7970144467997934, "grad_norm": 1.5020203482082857, "learning_rate": 2.084149528772453e-06, "loss": 0.5218, "step": 22371 }, { "epoch": 0.7970500739262875, "grad_norm": 2.5250258861939026, "learning_rate": 2.083444458062286e-06, "loss": 0.5302, "step": 22372 }, { "epoch": 0.7970857010527815, "grad_norm": 1.130311580110667, "learning_rate": 2.0827394927665865e-06, "loss": 0.4231, "step": 22373 }, { "epoch": 0.7971213281792757, "grad_norm": 1.1830070427510728, "learning_rate": 2.0820346328947484e-06, "loss": 0.3792, "step": 22374 }, { "epoch": 0.7971569553057698, "grad_norm": 1.1240996639185172, "learning_rate": 2.0813298784561497e-06, "loss": 0.361, "step": 22375 }, { "epoch": 0.7971925824322639, "grad_norm": 1.596759258993487, "learning_rate": 2.0806252294601815e-06, "loss": 0.5291, "step": 22376 }, { "epoch": 0.7972282095587581, "grad_norm": 1.1824934776997655, "learning_rate": 2.0799206859162213e-06, "loss": 0.4584, "step": 22377 }, { "epoch": 0.7972638366852521, "grad_norm": 1.5985535150320946, "learning_rate": 2.079216247833652e-06, "loss": 0.5288, "step": 22378 }, { "epoch": 0.7972994638117462, "grad_norm": 1.2793882104705199, "learning_rate": 2.078511915221856e-06, "loss": 0.4094, "step": 22379 }, { "epoch": 0.7973350909382404, "grad_norm": 1.1984934786741626, "learning_rate": 2.0778076880902096e-06, "loss": 0.3837, "step": 22380 }, { "epoch": 0.7973707180647345, "grad_norm": 1.1308748111761964, "learning_rate": 2.0771035664480944e-06, "loss": 0.4494, "step": 22381 }, { "epoch": 0.7974063451912285, "grad_norm": 1.421115070828686, "learning_rate": 2.0763995503048793e-06, "loss": 0.5001, "step": 22382 }, { "epoch": 0.7974419723177227, "grad_norm": 1.5118323247642405, "learning_rate": 2.0756956396699457e-06, "loss": 0.5598, "step": 22383 }, { "epoch": 0.7974775994442168, "grad_norm": 1.1645957177728439, "learning_rate": 2.074991834552661e-06, "loss": 0.3946, "step": 22384 }, { "epoch": 0.7975132265707109, "grad_norm": 1.2314758659688578, "learning_rate": 2.0742881349624e-06, "loss": 0.4052, "step": 22385 }, { "epoch": 0.7975488536972051, "grad_norm": 1.396500560072568, "learning_rate": 2.073584540908531e-06, "loss": 0.5263, "step": 22386 }, { "epoch": 0.7975844808236991, "grad_norm": 1.2020916672309863, "learning_rate": 2.072881052400427e-06, "loss": 0.4281, "step": 22387 }, { "epoch": 0.7976201079501932, "grad_norm": 1.3283610486194637, "learning_rate": 2.0721776694474503e-06, "loss": 0.4423, "step": 22388 }, { "epoch": 0.7976557350766874, "grad_norm": 1.102945470511091, "learning_rate": 2.0714743920589707e-06, "loss": 0.3713, "step": 22389 }, { "epoch": 0.7976913622031815, "grad_norm": 1.6038403315570524, "learning_rate": 2.07077122024435e-06, "loss": 0.5477, "step": 22390 }, { "epoch": 0.7977269893296756, "grad_norm": 1.2750611974143453, "learning_rate": 2.0700681540129517e-06, "loss": 0.3989, "step": 22391 }, { "epoch": 0.7977626164561697, "grad_norm": 1.2809265024957674, "learning_rate": 2.069365193374142e-06, "loss": 0.3862, "step": 22392 }, { "epoch": 0.7977982435826638, "grad_norm": 1.2202265089411755, "learning_rate": 2.0686623383372717e-06, "loss": 0.4122, "step": 22393 }, { "epoch": 0.7978338707091579, "grad_norm": 1.21360841552163, "learning_rate": 2.0679595889117123e-06, "loss": 0.4285, "step": 22394 }, { "epoch": 0.7978694978356521, "grad_norm": 1.4143415275612958, "learning_rate": 2.0672569451068126e-06, "loss": 0.5115, "step": 22395 }, { "epoch": 0.7979051249621462, "grad_norm": 1.818344025030573, "learning_rate": 2.066554406931933e-06, "loss": 0.4689, "step": 22396 }, { "epoch": 0.7979407520886402, "grad_norm": 1.4943162574698896, "learning_rate": 2.0658519743964246e-06, "loss": 0.5082, "step": 22397 }, { "epoch": 0.7979763792151344, "grad_norm": 1.3579512425041516, "learning_rate": 2.0651496475096455e-06, "loss": 0.5025, "step": 22398 }, { "epoch": 0.7980120063416285, "grad_norm": 1.1477156554925683, "learning_rate": 2.064447426280942e-06, "loss": 0.43, "step": 22399 }, { "epoch": 0.7980476334681226, "grad_norm": 1.0461069795002043, "learning_rate": 2.063745310719668e-06, "loss": 0.4117, "step": 22400 }, { "epoch": 0.7980832605946168, "grad_norm": 1.441673517661604, "learning_rate": 2.0630433008351723e-06, "loss": 0.4655, "step": 22401 }, { "epoch": 0.7981188877211108, "grad_norm": 1.4821696033412217, "learning_rate": 2.0623413966368023e-06, "loss": 0.4111, "step": 22402 }, { "epoch": 0.7981545148476049, "grad_norm": 1.1704937170042127, "learning_rate": 2.0616395981339076e-06, "loss": 0.4028, "step": 22403 }, { "epoch": 0.7981901419740991, "grad_norm": 1.2030772499374451, "learning_rate": 2.0609379053358268e-06, "loss": 0.393, "step": 22404 }, { "epoch": 0.7982257691005932, "grad_norm": 1.4531803200548925, "learning_rate": 2.06023631825191e-06, "loss": 0.4906, "step": 22405 }, { "epoch": 0.7982613962270872, "grad_norm": 1.6814243701611526, "learning_rate": 2.0595348368914946e-06, "loss": 0.5453, "step": 22406 }, { "epoch": 0.7982970233535814, "grad_norm": 1.4238858131976846, "learning_rate": 2.0588334612639217e-06, "loss": 0.4996, "step": 22407 }, { "epoch": 0.7983326504800755, "grad_norm": 1.0660361636239868, "learning_rate": 2.0581321913785325e-06, "loss": 0.3902, "step": 22408 }, { "epoch": 0.7983682776065696, "grad_norm": 1.4405410424200304, "learning_rate": 2.0574310272446663e-06, "loss": 0.5071, "step": 22409 }, { "epoch": 0.7984039047330638, "grad_norm": 1.2118270392782797, "learning_rate": 2.0567299688716545e-06, "loss": 0.3801, "step": 22410 }, { "epoch": 0.7984395318595578, "grad_norm": 1.1737265821382166, "learning_rate": 2.056029016268839e-06, "loss": 0.3727, "step": 22411 }, { "epoch": 0.7984751589860519, "grad_norm": 1.6170493650597992, "learning_rate": 2.055328169445546e-06, "loss": 0.5005, "step": 22412 }, { "epoch": 0.7985107861125461, "grad_norm": 1.1548254870935168, "learning_rate": 2.0546274284111123e-06, "loss": 0.3821, "step": 22413 }, { "epoch": 0.7985464132390402, "grad_norm": 1.2995840865533472, "learning_rate": 2.0539267931748696e-06, "loss": 0.4443, "step": 22414 }, { "epoch": 0.7985820403655343, "grad_norm": 1.616599125156024, "learning_rate": 2.053226263746143e-06, "loss": 0.606, "step": 22415 }, { "epoch": 0.7986176674920284, "grad_norm": 1.5620207278700073, "learning_rate": 2.052525840134264e-06, "loss": 0.5398, "step": 22416 }, { "epoch": 0.7986532946185225, "grad_norm": 1.231684771905509, "learning_rate": 2.051825522348557e-06, "loss": 0.4253, "step": 22417 }, { "epoch": 0.7986889217450167, "grad_norm": 1.518962420037965, "learning_rate": 2.051125310398353e-06, "loss": 0.5393, "step": 22418 }, { "epoch": 0.7987245488715108, "grad_norm": 1.2452335652248594, "learning_rate": 2.0504252042929686e-06, "loss": 0.4154, "step": 22419 }, { "epoch": 0.7987601759980049, "grad_norm": 1.407975266496028, "learning_rate": 2.0497252040417303e-06, "loss": 0.5351, "step": 22420 }, { "epoch": 0.798795803124499, "grad_norm": 1.3200539752451719, "learning_rate": 2.049025309653956e-06, "loss": 0.442, "step": 22421 }, { "epoch": 0.7988314302509931, "grad_norm": 1.8834027972754883, "learning_rate": 2.0483255211389673e-06, "loss": 0.3912, "step": 22422 }, { "epoch": 0.7988670573774872, "grad_norm": 2.1080490858258227, "learning_rate": 2.0476258385060823e-06, "loss": 0.4861, "step": 22423 }, { "epoch": 0.7989026845039814, "grad_norm": 1.1766461000970556, "learning_rate": 2.046926261764619e-06, "loss": 0.3855, "step": 22424 }, { "epoch": 0.7989383116304755, "grad_norm": 1.516065985506621, "learning_rate": 2.0462267909238897e-06, "loss": 0.4987, "step": 22425 }, { "epoch": 0.7989739387569695, "grad_norm": 1.5236274530743188, "learning_rate": 2.04552742599321e-06, "loss": 0.5081, "step": 22426 }, { "epoch": 0.7990095658834637, "grad_norm": 1.396117440527452, "learning_rate": 2.0448281669818948e-06, "loss": 0.5803, "step": 22427 }, { "epoch": 0.7990451930099578, "grad_norm": 1.7252223007139185, "learning_rate": 2.044129013899251e-06, "loss": 0.5907, "step": 22428 }, { "epoch": 0.7990808201364519, "grad_norm": 1.186688063069793, "learning_rate": 2.043429966754592e-06, "loss": 0.4185, "step": 22429 }, { "epoch": 0.7991164472629461, "grad_norm": 1.2992095881206402, "learning_rate": 2.042731025557222e-06, "loss": 0.485, "step": 22430 }, { "epoch": 0.7991520743894401, "grad_norm": 1.1989525019612977, "learning_rate": 2.042032190316451e-06, "loss": 0.4102, "step": 22431 }, { "epoch": 0.7991877015159342, "grad_norm": 1.6988248906543684, "learning_rate": 2.0413334610415825e-06, "loss": 0.5381, "step": 22432 }, { "epoch": 0.7992233286424284, "grad_norm": 1.3297582751004342, "learning_rate": 2.040634837741925e-06, "loss": 0.4257, "step": 22433 }, { "epoch": 0.7992589557689225, "grad_norm": 1.1983706446009497, "learning_rate": 2.039936320426774e-06, "loss": 0.3777, "step": 22434 }, { "epoch": 0.7992945828954165, "grad_norm": 1.32296383460919, "learning_rate": 2.039237909105439e-06, "loss": 0.5904, "step": 22435 }, { "epoch": 0.7993302100219107, "grad_norm": 1.739379755450717, "learning_rate": 2.038539603787212e-06, "loss": 0.511, "step": 22436 }, { "epoch": 0.7993658371484048, "grad_norm": 1.634295097116594, "learning_rate": 2.0378414044813943e-06, "loss": 0.5379, "step": 22437 }, { "epoch": 0.7994014642748989, "grad_norm": 1.4916219058929772, "learning_rate": 2.037143311197285e-06, "loss": 0.5381, "step": 22438 }, { "epoch": 0.7994370914013931, "grad_norm": 1.421167618099037, "learning_rate": 2.0364453239441773e-06, "loss": 0.4461, "step": 22439 }, { "epoch": 0.7994727185278871, "grad_norm": 1.2312678973339606, "learning_rate": 2.0357474427313695e-06, "loss": 0.4163, "step": 22440 }, { "epoch": 0.7995083456543812, "grad_norm": 1.1011385135412055, "learning_rate": 2.0350496675681476e-06, "loss": 0.3842, "step": 22441 }, { "epoch": 0.7995439727808754, "grad_norm": 1.549739465565006, "learning_rate": 2.0343519984638104e-06, "loss": 0.4358, "step": 22442 }, { "epoch": 0.7995795999073695, "grad_norm": 1.3811756382567635, "learning_rate": 2.03365443542764e-06, "loss": 0.3696, "step": 22443 }, { "epoch": 0.7996152270338636, "grad_norm": 1.370052146583936, "learning_rate": 2.032956978468933e-06, "loss": 0.6083, "step": 22444 }, { "epoch": 0.7996508541603577, "grad_norm": 1.1047369393544302, "learning_rate": 2.0322596275969666e-06, "loss": 0.4019, "step": 22445 }, { "epoch": 0.7996864812868518, "grad_norm": 1.5166771044383687, "learning_rate": 2.0315623828210374e-06, "loss": 0.5313, "step": 22446 }, { "epoch": 0.7997221084133459, "grad_norm": 1.6528131009045235, "learning_rate": 2.0308652441504216e-06, "loss": 0.5336, "step": 22447 }, { "epoch": 0.7997577355398401, "grad_norm": 1.424790922507686, "learning_rate": 2.0301682115944054e-06, "loss": 0.3858, "step": 22448 }, { "epoch": 0.7997933626663342, "grad_norm": 1.1941818849182853, "learning_rate": 2.0294712851622734e-06, "loss": 0.4188, "step": 22449 }, { "epoch": 0.7998289897928282, "grad_norm": 1.2179687428792176, "learning_rate": 2.028774464863299e-06, "loss": 0.3873, "step": 22450 }, { "epoch": 0.7998646169193224, "grad_norm": 1.599671543051311, "learning_rate": 2.0280777507067673e-06, "loss": 0.5317, "step": 22451 }, { "epoch": 0.7999002440458165, "grad_norm": 1.1860552010033116, "learning_rate": 2.0273811427019496e-06, "loss": 0.3913, "step": 22452 }, { "epoch": 0.7999358711723106, "grad_norm": 1.8470388006710778, "learning_rate": 2.026684640858124e-06, "loss": 0.4672, "step": 22453 }, { "epoch": 0.7999714982988048, "grad_norm": 1.1732478122307817, "learning_rate": 2.0259882451845657e-06, "loss": 0.4117, "step": 22454 }, { "epoch": 0.8000071254252988, "grad_norm": 1.2051034442605362, "learning_rate": 2.0252919556905504e-06, "loss": 0.4316, "step": 22455 }, { "epoch": 0.8000427525517929, "grad_norm": 1.185049494227112, "learning_rate": 2.0245957723853436e-06, "loss": 0.4165, "step": 22456 }, { "epoch": 0.8000783796782871, "grad_norm": 1.4296585055117144, "learning_rate": 2.023899695278222e-06, "loss": 0.5349, "step": 22457 }, { "epoch": 0.8001140068047812, "grad_norm": 1.1332328896978627, "learning_rate": 2.0232037243784475e-06, "loss": 0.4091, "step": 22458 }, { "epoch": 0.8001496339312753, "grad_norm": 1.2439079392327361, "learning_rate": 2.0225078596952907e-06, "loss": 0.4644, "step": 22459 }, { "epoch": 0.8001852610577694, "grad_norm": 1.1584550804227338, "learning_rate": 2.021812101238019e-06, "loss": 0.3742, "step": 22460 }, { "epoch": 0.8002208881842635, "grad_norm": 1.448994669358085, "learning_rate": 2.021116449015894e-06, "loss": 0.5434, "step": 22461 }, { "epoch": 0.8002565153107576, "grad_norm": 1.4240230647104586, "learning_rate": 2.0204209030381837e-06, "loss": 0.4642, "step": 22462 }, { "epoch": 0.8002921424372518, "grad_norm": 1.2979002269459148, "learning_rate": 2.019725463314144e-06, "loss": 0.4155, "step": 22463 }, { "epoch": 0.8003277695637458, "grad_norm": 1.1860868820755528, "learning_rate": 2.01903012985304e-06, "loss": 0.4035, "step": 22464 }, { "epoch": 0.8003633966902399, "grad_norm": 1.7695775561338352, "learning_rate": 2.018334902664125e-06, "loss": 0.5335, "step": 22465 }, { "epoch": 0.8003990238167341, "grad_norm": 1.6927381685071294, "learning_rate": 2.017639781756664e-06, "loss": 0.4156, "step": 22466 }, { "epoch": 0.8004346509432282, "grad_norm": 1.8615565383801644, "learning_rate": 2.0169447671399044e-06, "loss": 0.4547, "step": 22467 }, { "epoch": 0.8004702780697223, "grad_norm": 1.4626376670103465, "learning_rate": 2.016249858823106e-06, "loss": 0.5447, "step": 22468 }, { "epoch": 0.8005059051962164, "grad_norm": 1.1385384881207836, "learning_rate": 2.01555505681552e-06, "loss": 0.4002, "step": 22469 }, { "epoch": 0.8005415323227105, "grad_norm": 2.4155566175687126, "learning_rate": 2.014860361126403e-06, "loss": 0.3883, "step": 22470 }, { "epoch": 0.8005771594492046, "grad_norm": 1.1813488430756873, "learning_rate": 2.0141657717649975e-06, "loss": 0.3816, "step": 22471 }, { "epoch": 0.8006127865756988, "grad_norm": 1.6259603527926614, "learning_rate": 2.0134712887405583e-06, "loss": 0.4873, "step": 22472 }, { "epoch": 0.8006484137021929, "grad_norm": 1.4739897508401936, "learning_rate": 2.012776912062332e-06, "loss": 0.4782, "step": 22473 }, { "epoch": 0.8006840408286869, "grad_norm": 1.910891091386262, "learning_rate": 2.012082641739562e-06, "loss": 0.4837, "step": 22474 }, { "epoch": 0.8007196679551811, "grad_norm": 1.5675881806397391, "learning_rate": 2.011388477781495e-06, "loss": 0.4922, "step": 22475 }, { "epoch": 0.8007552950816752, "grad_norm": 1.6386783851204254, "learning_rate": 2.0106944201973745e-06, "loss": 0.5381, "step": 22476 }, { "epoch": 0.8007909222081693, "grad_norm": 1.4686695561918246, "learning_rate": 2.0100004689964434e-06, "loss": 0.4734, "step": 22477 }, { "epoch": 0.8008265493346635, "grad_norm": 1.5001040224557423, "learning_rate": 2.00930662418794e-06, "loss": 0.5921, "step": 22478 }, { "epoch": 0.8008621764611575, "grad_norm": 1.6858105856165582, "learning_rate": 2.0086128857811047e-06, "loss": 0.4993, "step": 22479 }, { "epoch": 0.8008978035876516, "grad_norm": 1.59121750043791, "learning_rate": 2.0079192537851734e-06, "loss": 0.5394, "step": 22480 }, { "epoch": 0.8009334307141458, "grad_norm": 1.6079695119608024, "learning_rate": 2.0072257282093855e-06, "loss": 0.4903, "step": 22481 }, { "epoch": 0.8009690578406399, "grad_norm": 1.2169149042218321, "learning_rate": 2.0065323090629675e-06, "loss": 0.4068, "step": 22482 }, { "epoch": 0.801004684967134, "grad_norm": 1.2981272865977147, "learning_rate": 2.005838996355163e-06, "loss": 0.4237, "step": 22483 }, { "epoch": 0.8010403120936281, "grad_norm": 1.63053019203251, "learning_rate": 2.005145790095201e-06, "loss": 0.53, "step": 22484 }, { "epoch": 0.8010759392201222, "grad_norm": 1.622652652997769, "learning_rate": 2.004452690292309e-06, "loss": 0.4611, "step": 22485 }, { "epoch": 0.8011115663466163, "grad_norm": 1.5368922271303782, "learning_rate": 2.0037596969557205e-06, "loss": 0.5156, "step": 22486 }, { "epoch": 0.8011471934731105, "grad_norm": 1.641356398480107, "learning_rate": 2.003066810094658e-06, "loss": 0.6072, "step": 22487 }, { "epoch": 0.8011828205996046, "grad_norm": 1.1494616771520325, "learning_rate": 2.0023740297183536e-06, "loss": 0.4228, "step": 22488 }, { "epoch": 0.8012184477260986, "grad_norm": 1.7065393787819665, "learning_rate": 2.001681355836026e-06, "loss": 0.5726, "step": 22489 }, { "epoch": 0.8012540748525928, "grad_norm": 1.2301885173421845, "learning_rate": 2.0009887884569014e-06, "loss": 0.4396, "step": 22490 }, { "epoch": 0.8012897019790869, "grad_norm": 1.5028838816712247, "learning_rate": 2.000296327590202e-06, "loss": 0.5362, "step": 22491 }, { "epoch": 0.801325329105581, "grad_norm": 1.531092706245966, "learning_rate": 1.999603973245151e-06, "loss": 0.5167, "step": 22492 }, { "epoch": 0.8013609562320752, "grad_norm": 1.7543905221832876, "learning_rate": 1.9989117254309633e-06, "loss": 0.5277, "step": 22493 }, { "epoch": 0.8013965833585692, "grad_norm": 1.3203036111036073, "learning_rate": 1.9982195841568575e-06, "loss": 0.4067, "step": 22494 }, { "epoch": 0.8014322104850633, "grad_norm": 1.1728473616215371, "learning_rate": 1.9975275494320535e-06, "loss": 0.4485, "step": 22495 }, { "epoch": 0.8014678376115575, "grad_norm": 2.4435444592386597, "learning_rate": 1.9968356212657613e-06, "loss": 0.4509, "step": 22496 }, { "epoch": 0.8015034647380516, "grad_norm": 1.225070488969098, "learning_rate": 1.9961437996671973e-06, "loss": 0.4317, "step": 22497 }, { "epoch": 0.8015390918645456, "grad_norm": 1.438867844503704, "learning_rate": 1.9954520846455717e-06, "loss": 0.5352, "step": 22498 }, { "epoch": 0.8015747189910398, "grad_norm": 1.9330836841079522, "learning_rate": 1.9947604762101003e-06, "loss": 0.546, "step": 22499 }, { "epoch": 0.8016103461175339, "grad_norm": 1.354942878893816, "learning_rate": 1.994068974369985e-06, "loss": 0.4806, "step": 22500 }, { "epoch": 0.801645973244028, "grad_norm": 1.4569652655960048, "learning_rate": 1.99337757913444e-06, "loss": 0.4966, "step": 22501 }, { "epoch": 0.8016816003705222, "grad_norm": 1.403697428488519, "learning_rate": 1.9926862905126663e-06, "loss": 0.5239, "step": 22502 }, { "epoch": 0.8017172274970162, "grad_norm": 1.169222249259749, "learning_rate": 1.9919951085138735e-06, "loss": 0.3676, "step": 22503 }, { "epoch": 0.8017528546235103, "grad_norm": 1.4770272790010064, "learning_rate": 1.991304033147261e-06, "loss": 0.5084, "step": 22504 }, { "epoch": 0.8017884817500045, "grad_norm": 1.2071798209487596, "learning_rate": 1.9906130644220333e-06, "loss": 0.4104, "step": 22505 }, { "epoch": 0.8018241088764986, "grad_norm": 1.5836491763290537, "learning_rate": 1.9899222023473895e-06, "loss": 0.5395, "step": 22506 }, { "epoch": 0.8018597360029927, "grad_norm": 1.5891019795906542, "learning_rate": 1.9892314469325315e-06, "loss": 0.4764, "step": 22507 }, { "epoch": 0.8018953631294868, "grad_norm": 1.5620745774945164, "learning_rate": 1.988540798186659e-06, "loss": 0.5202, "step": 22508 }, { "epoch": 0.8019309902559809, "grad_norm": 1.6497734313569723, "learning_rate": 1.9878502561189616e-06, "loss": 0.5249, "step": 22509 }, { "epoch": 0.801966617382475, "grad_norm": 1.1876079994996904, "learning_rate": 1.987159820738642e-06, "loss": 0.3958, "step": 22510 }, { "epoch": 0.8020022445089692, "grad_norm": 1.2827902765743566, "learning_rate": 1.9864694920548867e-06, "loss": 0.373, "step": 22511 }, { "epoch": 0.8020378716354633, "grad_norm": 1.274593160776876, "learning_rate": 1.9857792700768908e-06, "loss": 0.3787, "step": 22512 }, { "epoch": 0.8020734987619573, "grad_norm": 1.1546075031907168, "learning_rate": 1.9850891548138463e-06, "loss": 0.4, "step": 22513 }, { "epoch": 0.8021091258884515, "grad_norm": 1.5361997315789415, "learning_rate": 1.9843991462749444e-06, "loss": 0.4694, "step": 22514 }, { "epoch": 0.8021447530149456, "grad_norm": 1.3572543499151386, "learning_rate": 1.9837092444693675e-06, "loss": 0.423, "step": 22515 }, { "epoch": 0.8021803801414397, "grad_norm": 1.6794090326321571, "learning_rate": 1.983019449406308e-06, "loss": 0.3667, "step": 22516 }, { "epoch": 0.8022160072679339, "grad_norm": 1.5362726852543525, "learning_rate": 1.9823297610949455e-06, "loss": 0.4922, "step": 22517 }, { "epoch": 0.8022516343944279, "grad_norm": 1.5695090022361304, "learning_rate": 1.9816401795444664e-06, "loss": 0.4761, "step": 22518 }, { "epoch": 0.802287261520922, "grad_norm": 1.3263005074136398, "learning_rate": 1.9809507047640543e-06, "loss": 0.4154, "step": 22519 }, { "epoch": 0.8023228886474162, "grad_norm": 1.5467160519699956, "learning_rate": 1.9802613367628843e-06, "loss": 0.5128, "step": 22520 }, { "epoch": 0.8023585157739103, "grad_norm": 1.1685143269856848, "learning_rate": 1.9795720755501446e-06, "loss": 0.3781, "step": 22521 }, { "epoch": 0.8023941429004043, "grad_norm": 1.1959764524039151, "learning_rate": 1.978882921135007e-06, "loss": 0.4052, "step": 22522 }, { "epoch": 0.8024297700268985, "grad_norm": 1.1208592632833345, "learning_rate": 1.9781938735266513e-06, "loss": 0.3864, "step": 22523 }, { "epoch": 0.8024653971533926, "grad_norm": 1.5149269176869222, "learning_rate": 1.9775049327342486e-06, "loss": 0.6111, "step": 22524 }, { "epoch": 0.8025010242798867, "grad_norm": 1.5816857699444473, "learning_rate": 1.976816098766978e-06, "loss": 0.5587, "step": 22525 }, { "epoch": 0.8025366514063809, "grad_norm": 1.5031128316189997, "learning_rate": 1.976127371634007e-06, "loss": 0.5141, "step": 22526 }, { "epoch": 0.8025722785328749, "grad_norm": 1.5693654009849265, "learning_rate": 1.9754387513445085e-06, "loss": 0.4981, "step": 22527 }, { "epoch": 0.802607905659369, "grad_norm": 1.2211518775423678, "learning_rate": 1.97475023790765e-06, "loss": 0.3981, "step": 22528 }, { "epoch": 0.8026435327858632, "grad_norm": 1.7213725491661396, "learning_rate": 1.974061831332603e-06, "loss": 0.4857, "step": 22529 }, { "epoch": 0.8026791599123573, "grad_norm": 1.1226473293348906, "learning_rate": 1.973373531628535e-06, "loss": 0.3599, "step": 22530 }, { "epoch": 0.8027147870388514, "grad_norm": 1.3615039486934195, "learning_rate": 1.9726853388046052e-06, "loss": 0.4273, "step": 22531 }, { "epoch": 0.8027504141653455, "grad_norm": 1.3445268391507827, "learning_rate": 1.9719972528699847e-06, "loss": 0.4149, "step": 22532 }, { "epoch": 0.8027860412918396, "grad_norm": 1.712198524639257, "learning_rate": 1.971309273833828e-06, "loss": 0.5717, "step": 22533 }, { "epoch": 0.8028216684183337, "grad_norm": 1.7156972462543014, "learning_rate": 1.9706214017053017e-06, "loss": 0.6103, "step": 22534 }, { "epoch": 0.8028572955448279, "grad_norm": 1.3429010805618953, "learning_rate": 1.969933636493562e-06, "loss": 0.5444, "step": 22535 }, { "epoch": 0.802892922671322, "grad_norm": 1.2157638434857903, "learning_rate": 1.9692459782077724e-06, "loss": 0.4075, "step": 22536 }, { "epoch": 0.802928549797816, "grad_norm": 1.122796722688459, "learning_rate": 1.9685584268570824e-06, "loss": 0.4019, "step": 22537 }, { "epoch": 0.8029641769243102, "grad_norm": 1.2293832495062358, "learning_rate": 1.9678709824506535e-06, "loss": 0.4295, "step": 22538 }, { "epoch": 0.8029998040508043, "grad_norm": 1.5320437238882472, "learning_rate": 1.9671836449976345e-06, "loss": 0.5196, "step": 22539 }, { "epoch": 0.8030354311772984, "grad_norm": 1.283924497487239, "learning_rate": 1.9664964145071784e-06, "loss": 0.4196, "step": 22540 }, { "epoch": 0.8030710583037926, "grad_norm": 1.4741633246089187, "learning_rate": 1.965809290988442e-06, "loss": 0.5178, "step": 22541 }, { "epoch": 0.8031066854302866, "grad_norm": 1.0930593581090313, "learning_rate": 1.965122274450567e-06, "loss": 0.4342, "step": 22542 }, { "epoch": 0.8031423125567807, "grad_norm": 1.1805269932624254, "learning_rate": 1.9644353649027048e-06, "loss": 0.4281, "step": 22543 }, { "epoch": 0.8031779396832749, "grad_norm": 1.48157593101052, "learning_rate": 1.963748562354002e-06, "loss": 0.4816, "step": 22544 }, { "epoch": 0.803213566809769, "grad_norm": 1.6868063001019826, "learning_rate": 1.963061866813608e-06, "loss": 0.5625, "step": 22545 }, { "epoch": 0.803249193936263, "grad_norm": 1.0795149504024228, "learning_rate": 1.9623752782906593e-06, "loss": 0.3664, "step": 22546 }, { "epoch": 0.8032848210627572, "grad_norm": 1.1972264046120025, "learning_rate": 1.9616887967943045e-06, "loss": 0.5002, "step": 22547 }, { "epoch": 0.8033204481892513, "grad_norm": 1.434747014604474, "learning_rate": 1.961002422333679e-06, "loss": 0.4914, "step": 22548 }, { "epoch": 0.8033560753157454, "grad_norm": 1.3772011502398274, "learning_rate": 1.960316154917925e-06, "loss": 0.4053, "step": 22549 }, { "epoch": 0.8033917024422396, "grad_norm": 1.1370125246849079, "learning_rate": 1.9596299945561805e-06, "loss": 0.363, "step": 22550 }, { "epoch": 0.8034273295687336, "grad_norm": 1.294192610934626, "learning_rate": 1.958943941257586e-06, "loss": 0.4099, "step": 22551 }, { "epoch": 0.8034629566952277, "grad_norm": 1.5010295266694127, "learning_rate": 1.9582579950312696e-06, "loss": 0.5159, "step": 22552 }, { "epoch": 0.8034985838217219, "grad_norm": 1.3060062308899565, "learning_rate": 1.9575721558863704e-06, "loss": 0.3579, "step": 22553 }, { "epoch": 0.803534210948216, "grad_norm": 1.3345734248326255, "learning_rate": 1.956886423832021e-06, "loss": 0.4114, "step": 22554 }, { "epoch": 0.80356983807471, "grad_norm": 1.285404333497543, "learning_rate": 1.956200798877348e-06, "loss": 0.4195, "step": 22555 }, { "epoch": 0.8036054652012042, "grad_norm": 1.643430320258848, "learning_rate": 1.955515281031486e-06, "loss": 0.5016, "step": 22556 }, { "epoch": 0.8036410923276983, "grad_norm": 1.1804589804993975, "learning_rate": 1.954829870303555e-06, "loss": 0.4106, "step": 22557 }, { "epoch": 0.8036767194541924, "grad_norm": 1.278385270192439, "learning_rate": 1.9541445667026936e-06, "loss": 0.4316, "step": 22558 }, { "epoch": 0.8037123465806866, "grad_norm": 1.1004149571978667, "learning_rate": 1.953459370238019e-06, "loss": 0.3829, "step": 22559 }, { "epoch": 0.8037479737071807, "grad_norm": 1.778317850220779, "learning_rate": 1.952774280918659e-06, "loss": 0.4658, "step": 22560 }, { "epoch": 0.8037836008336747, "grad_norm": 1.2620445761342767, "learning_rate": 1.952089298753732e-06, "loss": 0.4365, "step": 22561 }, { "epoch": 0.8038192279601689, "grad_norm": 1.2787580099942155, "learning_rate": 1.951404423752363e-06, "loss": 0.4786, "step": 22562 }, { "epoch": 0.803854855086663, "grad_norm": 1.8756857217917715, "learning_rate": 1.9507196559236675e-06, "loss": 0.4301, "step": 22563 }, { "epoch": 0.8038904822131571, "grad_norm": 1.115310996123692, "learning_rate": 1.9500349952767663e-06, "loss": 0.4142, "step": 22564 }, { "epoch": 0.8039261093396513, "grad_norm": 1.5131290883248785, "learning_rate": 1.9493504418207754e-06, "loss": 0.4959, "step": 22565 }, { "epoch": 0.8039617364661453, "grad_norm": 1.5181751654851343, "learning_rate": 1.94866599556481e-06, "loss": 0.591, "step": 22566 }, { "epoch": 0.8039973635926394, "grad_norm": 1.320966104069924, "learning_rate": 1.9479816565179875e-06, "loss": 0.5316, "step": 22567 }, { "epoch": 0.8040329907191336, "grad_norm": 1.0270636614442632, "learning_rate": 1.947297424689414e-06, "loss": 0.3619, "step": 22568 }, { "epoch": 0.8040686178456277, "grad_norm": 1.5906390351977413, "learning_rate": 1.946613300088207e-06, "loss": 0.4949, "step": 22569 }, { "epoch": 0.8041042449721217, "grad_norm": 1.6609119614951962, "learning_rate": 1.94592928272347e-06, "loss": 0.5875, "step": 22570 }, { "epoch": 0.8041398720986159, "grad_norm": 1.4991347847872438, "learning_rate": 1.9452453726043143e-06, "loss": 0.4133, "step": 22571 }, { "epoch": 0.80417549922511, "grad_norm": 1.504190046497318, "learning_rate": 1.9445615697398467e-06, "loss": 0.4689, "step": 22572 }, { "epoch": 0.8042111263516041, "grad_norm": 1.2258107633923372, "learning_rate": 1.943877874139174e-06, "loss": 0.4361, "step": 22573 }, { "epoch": 0.8042467534780983, "grad_norm": 1.1612327303076555, "learning_rate": 1.943194285811396e-06, "loss": 0.4006, "step": 22574 }, { "epoch": 0.8042823806045923, "grad_norm": 1.4453674811369672, "learning_rate": 1.942510804765618e-06, "loss": 0.464, "step": 22575 }, { "epoch": 0.8043180077310864, "grad_norm": 1.504382659330609, "learning_rate": 1.941827431010943e-06, "loss": 0.4971, "step": 22576 }, { "epoch": 0.8043536348575806, "grad_norm": 1.3009939492339426, "learning_rate": 1.9411441645564655e-06, "loss": 0.4207, "step": 22577 }, { "epoch": 0.8043892619840747, "grad_norm": 1.2089437546018127, "learning_rate": 1.940461005411288e-06, "loss": 0.4402, "step": 22578 }, { "epoch": 0.8044248891105688, "grad_norm": 1.5859162507394455, "learning_rate": 1.939777953584503e-06, "loss": 0.4931, "step": 22579 }, { "epoch": 0.8044605162370629, "grad_norm": 1.266743708579459, "learning_rate": 1.939095009085209e-06, "loss": 0.4479, "step": 22580 }, { "epoch": 0.804496143363557, "grad_norm": 1.799587669476834, "learning_rate": 1.938412171922499e-06, "loss": 0.5016, "step": 22581 }, { "epoch": 0.8045317704900511, "grad_norm": 1.3903465783971094, "learning_rate": 1.9377294421054683e-06, "loss": 0.418, "step": 22582 }, { "epoch": 0.8045673976165453, "grad_norm": 1.6431276844641858, "learning_rate": 1.9370468196432025e-06, "loss": 0.5165, "step": 22583 }, { "epoch": 0.8046030247430394, "grad_norm": 1.5351929296656257, "learning_rate": 1.9363643045447967e-06, "loss": 0.5012, "step": 22584 }, { "epoch": 0.8046386518695334, "grad_norm": 1.7625479743443688, "learning_rate": 1.9356818968193336e-06, "loss": 0.6086, "step": 22585 }, { "epoch": 0.8046742789960276, "grad_norm": 1.2626216171705105, "learning_rate": 1.934999596475903e-06, "loss": 0.3919, "step": 22586 }, { "epoch": 0.8047099061225217, "grad_norm": 1.1626842538125002, "learning_rate": 1.934317403523589e-06, "loss": 0.4077, "step": 22587 }, { "epoch": 0.8047455332490158, "grad_norm": 1.7219702588472916, "learning_rate": 1.9336353179714774e-06, "loss": 0.5065, "step": 22588 }, { "epoch": 0.80478116037551, "grad_norm": 1.2784194925204706, "learning_rate": 1.9329533398286517e-06, "loss": 0.3958, "step": 22589 }, { "epoch": 0.804816787502004, "grad_norm": 1.108294080828547, "learning_rate": 1.932271469104188e-06, "loss": 0.4108, "step": 22590 }, { "epoch": 0.8048524146284981, "grad_norm": 1.4249776406006607, "learning_rate": 1.931589705807172e-06, "loss": 0.431, "step": 22591 }, { "epoch": 0.8048880417549923, "grad_norm": 1.31771723588706, "learning_rate": 1.930908049946675e-06, "loss": 0.3862, "step": 22592 }, { "epoch": 0.8049236688814864, "grad_norm": 1.2527396452196142, "learning_rate": 1.93022650153178e-06, "loss": 0.4486, "step": 22593 }, { "epoch": 0.8049592960079804, "grad_norm": 1.064260308404603, "learning_rate": 1.929545060571557e-06, "loss": 0.4376, "step": 22594 }, { "epoch": 0.8049949231344746, "grad_norm": 1.118499332659272, "learning_rate": 1.9288637270750832e-06, "loss": 0.4021, "step": 22595 }, { "epoch": 0.8050305502609687, "grad_norm": 1.1656954898985736, "learning_rate": 1.92818250105143e-06, "loss": 0.4203, "step": 22596 }, { "epoch": 0.8050661773874628, "grad_norm": 1.4453314970369426, "learning_rate": 1.927501382509671e-06, "loss": 0.4833, "step": 22597 }, { "epoch": 0.805101804513957, "grad_norm": 1.9196648769652245, "learning_rate": 1.9268203714588706e-06, "loss": 0.4747, "step": 22598 }, { "epoch": 0.805137431640451, "grad_norm": 1.133218391891732, "learning_rate": 1.9261394679081004e-06, "loss": 0.3908, "step": 22599 }, { "epoch": 0.8051730587669451, "grad_norm": 1.6856516527768446, "learning_rate": 1.9254586718664293e-06, "loss": 0.3806, "step": 22600 }, { "epoch": 0.8052086858934393, "grad_norm": 1.2393368431504146, "learning_rate": 1.924777983342916e-06, "loss": 0.4206, "step": 22601 }, { "epoch": 0.8052443130199334, "grad_norm": 1.1010089500656255, "learning_rate": 1.9240974023466284e-06, "loss": 0.384, "step": 22602 }, { "epoch": 0.8052799401464275, "grad_norm": 1.2000153607277118, "learning_rate": 1.923416928886629e-06, "loss": 0.416, "step": 22603 }, { "epoch": 0.8053155672729216, "grad_norm": 1.7059815534766078, "learning_rate": 1.9227365629719806e-06, "loss": 0.5541, "step": 22604 }, { "epoch": 0.8053511943994157, "grad_norm": 1.1197099591107944, "learning_rate": 1.922056304611739e-06, "loss": 0.4202, "step": 22605 }, { "epoch": 0.8053868215259098, "grad_norm": 1.8699075276933987, "learning_rate": 1.9213761538149657e-06, "loss": 0.5335, "step": 22606 }, { "epoch": 0.805422448652404, "grad_norm": 1.3540480446695347, "learning_rate": 1.920696110590713e-06, "loss": 0.5185, "step": 22607 }, { "epoch": 0.805458075778898, "grad_norm": 1.7796299121983794, "learning_rate": 1.9200161749480406e-06, "loss": 0.5271, "step": 22608 }, { "epoch": 0.8054937029053921, "grad_norm": 1.2817786130393656, "learning_rate": 1.9193363468959956e-06, "loss": 0.4161, "step": 22609 }, { "epoch": 0.8055293300318863, "grad_norm": 1.6123170845024497, "learning_rate": 1.918656626443638e-06, "loss": 0.4871, "step": 22610 }, { "epoch": 0.8055649571583804, "grad_norm": 1.5673859902310987, "learning_rate": 1.9179770136000187e-06, "loss": 0.5137, "step": 22611 }, { "epoch": 0.8056005842848745, "grad_norm": 1.1053374333919184, "learning_rate": 1.917297508374182e-06, "loss": 0.3732, "step": 22612 }, { "epoch": 0.8056362114113687, "grad_norm": 1.2361040213621761, "learning_rate": 1.9166181107751802e-06, "loss": 0.3981, "step": 22613 }, { "epoch": 0.8056718385378627, "grad_norm": 1.5488328666017233, "learning_rate": 1.9159388208120568e-06, "loss": 0.4653, "step": 22614 }, { "epoch": 0.8057074656643568, "grad_norm": 1.966244702563125, "learning_rate": 1.915259638493859e-06, "loss": 0.4574, "step": 22615 }, { "epoch": 0.805743092790851, "grad_norm": 1.1572951801400564, "learning_rate": 1.9145805638296287e-06, "loss": 0.4207, "step": 22616 }, { "epoch": 0.8057787199173451, "grad_norm": 1.9953270083734187, "learning_rate": 1.9139015968284104e-06, "loss": 0.5232, "step": 22617 }, { "epoch": 0.8058143470438391, "grad_norm": 1.4085648635669232, "learning_rate": 1.913222737499243e-06, "loss": 0.4882, "step": 22618 }, { "epoch": 0.8058499741703333, "grad_norm": 1.2718717683436007, "learning_rate": 1.912543985851171e-06, "loss": 0.393, "step": 22619 }, { "epoch": 0.8058856012968274, "grad_norm": 1.6579588667591796, "learning_rate": 1.9118653418932244e-06, "loss": 0.5162, "step": 22620 }, { "epoch": 0.8059212284233215, "grad_norm": 1.2413567054305101, "learning_rate": 1.911186805634445e-06, "loss": 0.4846, "step": 22621 }, { "epoch": 0.8059568555498157, "grad_norm": 1.1050156694703817, "learning_rate": 1.9105083770838694e-06, "loss": 0.3723, "step": 22622 }, { "epoch": 0.8059924826763097, "grad_norm": 1.245030491215069, "learning_rate": 1.9098300562505266e-06, "loss": 0.4252, "step": 22623 }, { "epoch": 0.8060281098028038, "grad_norm": 1.6394365729806177, "learning_rate": 1.9091518431434507e-06, "loss": 0.4648, "step": 22624 }, { "epoch": 0.806063736929298, "grad_norm": 1.0770891917870604, "learning_rate": 1.9084737377716732e-06, "loss": 0.4589, "step": 22625 }, { "epoch": 0.8060993640557921, "grad_norm": 1.4328081073745567, "learning_rate": 1.9077957401442272e-06, "loss": 0.52, "step": 22626 }, { "epoch": 0.8061349911822862, "grad_norm": 1.7739921998154693, "learning_rate": 1.9071178502701337e-06, "loss": 0.5155, "step": 22627 }, { "epoch": 0.8061706183087803, "grad_norm": 1.1967816102436055, "learning_rate": 1.906440068158425e-06, "loss": 0.4439, "step": 22628 }, { "epoch": 0.8062062454352744, "grad_norm": 1.420184536187711, "learning_rate": 1.9057623938181213e-06, "loss": 0.4983, "step": 22629 }, { "epoch": 0.8062418725617685, "grad_norm": 1.66422680593355, "learning_rate": 1.9050848272582522e-06, "loss": 0.5668, "step": 22630 }, { "epoch": 0.8062774996882627, "grad_norm": 1.0316249287742332, "learning_rate": 1.904407368487834e-06, "loss": 0.3518, "step": 22631 }, { "epoch": 0.8063131268147568, "grad_norm": 1.5365784083888416, "learning_rate": 1.903730017515889e-06, "loss": 0.6034, "step": 22632 }, { "epoch": 0.8063487539412508, "grad_norm": 1.4086598413138274, "learning_rate": 1.9030527743514382e-06, "loss": 0.4896, "step": 22633 }, { "epoch": 0.806384381067745, "grad_norm": 1.2884844653268388, "learning_rate": 1.9023756390034998e-06, "loss": 0.399, "step": 22634 }, { "epoch": 0.8064200081942391, "grad_norm": 1.2278797034962479, "learning_rate": 1.901698611481092e-06, "loss": 0.403, "step": 22635 }, { "epoch": 0.8064556353207332, "grad_norm": 1.5697274019424723, "learning_rate": 1.9010216917932244e-06, "loss": 0.528, "step": 22636 }, { "epoch": 0.8064912624472274, "grad_norm": 1.256241626856647, "learning_rate": 1.9003448799489165e-06, "loss": 0.3965, "step": 22637 }, { "epoch": 0.8065268895737214, "grad_norm": 1.6346013533938102, "learning_rate": 1.8996681759571755e-06, "loss": 0.4786, "step": 22638 }, { "epoch": 0.8065625167002155, "grad_norm": 1.2188686888790012, "learning_rate": 1.8989915798270152e-06, "loss": 0.4048, "step": 22639 }, { "epoch": 0.8065981438267097, "grad_norm": 1.359812587317488, "learning_rate": 1.8983150915674431e-06, "loss": 0.3877, "step": 22640 }, { "epoch": 0.8066337709532038, "grad_norm": 1.4821168913436413, "learning_rate": 1.8976387111874717e-06, "loss": 0.4559, "step": 22641 }, { "epoch": 0.8066693980796978, "grad_norm": 1.2978711808054633, "learning_rate": 1.8969624386961015e-06, "loss": 0.4325, "step": 22642 }, { "epoch": 0.806705025206192, "grad_norm": 1.5029180840488638, "learning_rate": 1.8962862741023425e-06, "loss": 0.495, "step": 22643 }, { "epoch": 0.8067406523326861, "grad_norm": 1.1088979395960603, "learning_rate": 1.8956102174151935e-06, "loss": 0.4343, "step": 22644 }, { "epoch": 0.8067762794591802, "grad_norm": 1.4703004436317209, "learning_rate": 1.894934268643659e-06, "loss": 0.5369, "step": 22645 }, { "epoch": 0.8068119065856744, "grad_norm": 2.667155422439506, "learning_rate": 1.894258427796739e-06, "loss": 0.5066, "step": 22646 }, { "epoch": 0.8068475337121684, "grad_norm": 1.2307439341350883, "learning_rate": 1.8935826948834347e-06, "loss": 0.4148, "step": 22647 }, { "epoch": 0.8068831608386625, "grad_norm": 1.8133736787730532, "learning_rate": 1.892907069912745e-06, "loss": 0.5493, "step": 22648 }, { "epoch": 0.8069187879651567, "grad_norm": 1.293351508187312, "learning_rate": 1.8922315528936619e-06, "loss": 0.4766, "step": 22649 }, { "epoch": 0.8069544150916508, "grad_norm": 1.2966641439443192, "learning_rate": 1.8915561438351847e-06, "loss": 0.4277, "step": 22650 }, { "epoch": 0.8069900422181449, "grad_norm": 1.087452561541402, "learning_rate": 1.890880842746302e-06, "loss": 0.4071, "step": 22651 }, { "epoch": 0.807025669344639, "grad_norm": 1.3749484571426482, "learning_rate": 1.8902056496360122e-06, "loss": 0.4765, "step": 22652 }, { "epoch": 0.8070612964711331, "grad_norm": 1.1855821143086165, "learning_rate": 1.8895305645133e-06, "loss": 0.4252, "step": 22653 }, { "epoch": 0.8070969235976272, "grad_norm": 1.2723326808987705, "learning_rate": 1.8888555873871573e-06, "loss": 0.4445, "step": 22654 }, { "epoch": 0.8071325507241214, "grad_norm": 1.5634314722540503, "learning_rate": 1.8881807182665711e-06, "loss": 0.5308, "step": 22655 }, { "epoch": 0.8071681778506155, "grad_norm": 2.134536593633739, "learning_rate": 1.8875059571605293e-06, "loss": 0.5448, "step": 22656 }, { "epoch": 0.8072038049771095, "grad_norm": 1.7432119271116697, "learning_rate": 1.8868313040780184e-06, "loss": 0.5198, "step": 22657 }, { "epoch": 0.8072394321036037, "grad_norm": 1.7727169771745193, "learning_rate": 1.886156759028016e-06, "loss": 0.4935, "step": 22658 }, { "epoch": 0.8072750592300978, "grad_norm": 1.5502031350851009, "learning_rate": 1.8854823220195106e-06, "loss": 0.505, "step": 22659 }, { "epoch": 0.8073106863565919, "grad_norm": 1.269350014270382, "learning_rate": 1.8848079930614783e-06, "loss": 0.4134, "step": 22660 }, { "epoch": 0.807346313483086, "grad_norm": 1.0737937922853409, "learning_rate": 1.8841337721628982e-06, "loss": 0.3836, "step": 22661 }, { "epoch": 0.8073819406095801, "grad_norm": 1.1422843399205045, "learning_rate": 1.8834596593327503e-06, "loss": 0.3954, "step": 22662 }, { "epoch": 0.8074175677360742, "grad_norm": 1.1782099141241544, "learning_rate": 1.8827856545800126e-06, "loss": 0.4019, "step": 22663 }, { "epoch": 0.8074531948625684, "grad_norm": 1.5603034276427574, "learning_rate": 1.882111757913655e-06, "loss": 0.4226, "step": 22664 }, { "epoch": 0.8074888219890625, "grad_norm": 1.2119875352022598, "learning_rate": 1.8814379693426554e-06, "loss": 0.3966, "step": 22665 }, { "epoch": 0.8075244491155565, "grad_norm": 1.1298418109486614, "learning_rate": 1.8807642888759825e-06, "loss": 0.4176, "step": 22666 }, { "epoch": 0.8075600762420507, "grad_norm": 1.2762730057438392, "learning_rate": 1.8800907165226068e-06, "loss": 0.3992, "step": 22667 }, { "epoch": 0.8075957033685448, "grad_norm": 1.2394310061700624, "learning_rate": 1.8794172522915022e-06, "loss": 0.3805, "step": 22668 }, { "epoch": 0.8076313304950389, "grad_norm": 1.2394878999624888, "learning_rate": 1.8787438961916292e-06, "loss": 0.4256, "step": 22669 }, { "epoch": 0.8076669576215331, "grad_norm": 1.1301345855472116, "learning_rate": 1.8780706482319588e-06, "loss": 0.3535, "step": 22670 }, { "epoch": 0.8077025847480271, "grad_norm": 1.1588991142952483, "learning_rate": 1.8773975084214547e-06, "loss": 0.3576, "step": 22671 }, { "epoch": 0.8077382118745212, "grad_norm": 1.7077782022578698, "learning_rate": 1.8767244767690818e-06, "loss": 0.5768, "step": 22672 }, { "epoch": 0.8077738390010154, "grad_norm": 1.8276073022656205, "learning_rate": 1.8760515532837976e-06, "loss": 0.3799, "step": 22673 }, { "epoch": 0.8078094661275095, "grad_norm": 1.1231635162762121, "learning_rate": 1.8753787379745691e-06, "loss": 0.4318, "step": 22674 }, { "epoch": 0.8078450932540036, "grad_norm": 1.467021475803222, "learning_rate": 1.8747060308503473e-06, "loss": 0.4771, "step": 22675 }, { "epoch": 0.8078807203804977, "grad_norm": 1.6234287957754443, "learning_rate": 1.8740334319200948e-06, "loss": 0.4693, "step": 22676 }, { "epoch": 0.8079163475069918, "grad_norm": 1.4385244740616485, "learning_rate": 1.8733609411927678e-06, "loss": 0.4687, "step": 22677 }, { "epoch": 0.8079519746334859, "grad_norm": 1.0063347016109467, "learning_rate": 1.8726885586773213e-06, "loss": 0.3868, "step": 22678 }, { "epoch": 0.8079876017599801, "grad_norm": 1.1458641299636412, "learning_rate": 1.872016284382705e-06, "loss": 0.3989, "step": 22679 }, { "epoch": 0.8080232288864742, "grad_norm": 1.8582208079445102, "learning_rate": 1.8713441183178728e-06, "loss": 0.488, "step": 22680 }, { "epoch": 0.8080588560129682, "grad_norm": 1.5462470944461835, "learning_rate": 1.8706720604917782e-06, "loss": 0.4492, "step": 22681 }, { "epoch": 0.8080944831394624, "grad_norm": 1.1229266360103993, "learning_rate": 1.8700001109133648e-06, "loss": 0.4333, "step": 22682 }, { "epoch": 0.8081301102659565, "grad_norm": 1.2411503451112869, "learning_rate": 1.869328269591585e-06, "loss": 0.3876, "step": 22683 }, { "epoch": 0.8081657373924506, "grad_norm": 1.2099323156292905, "learning_rate": 1.868656536535376e-06, "loss": 0.3935, "step": 22684 }, { "epoch": 0.8082013645189448, "grad_norm": 1.4900868184117817, "learning_rate": 1.8679849117536952e-06, "loss": 0.5034, "step": 22685 }, { "epoch": 0.8082369916454388, "grad_norm": 1.162073301697175, "learning_rate": 1.8673133952554756e-06, "loss": 0.3779, "step": 22686 }, { "epoch": 0.8082726187719329, "grad_norm": 1.635348711938693, "learning_rate": 1.8666419870496666e-06, "loss": 0.4886, "step": 22687 }, { "epoch": 0.8083082458984271, "grad_norm": 1.3348525700070042, "learning_rate": 1.865970687145201e-06, "loss": 0.4443, "step": 22688 }, { "epoch": 0.8083438730249212, "grad_norm": 2.364095112425346, "learning_rate": 1.8652994955510228e-06, "loss": 0.5447, "step": 22689 }, { "epoch": 0.8083795001514152, "grad_norm": 1.2673514810147166, "learning_rate": 1.864628412276066e-06, "loss": 0.4589, "step": 22690 }, { "epoch": 0.8084151272779094, "grad_norm": 1.1419926302995929, "learning_rate": 1.863957437329268e-06, "loss": 0.3733, "step": 22691 }, { "epoch": 0.8084507544044035, "grad_norm": 1.164891848998885, "learning_rate": 1.8632865707195636e-06, "loss": 0.3828, "step": 22692 }, { "epoch": 0.8084863815308976, "grad_norm": 1.1855726110057059, "learning_rate": 1.8626158124558858e-06, "loss": 0.4081, "step": 22693 }, { "epoch": 0.8085220086573918, "grad_norm": 1.1567261019455932, "learning_rate": 1.8619451625471685e-06, "loss": 0.4144, "step": 22694 }, { "epoch": 0.8085576357838858, "grad_norm": 1.2163763862138286, "learning_rate": 1.861274621002337e-06, "loss": 0.479, "step": 22695 }, { "epoch": 0.8085932629103799, "grad_norm": 1.1961116798673515, "learning_rate": 1.8606041878303249e-06, "loss": 0.3361, "step": 22696 }, { "epoch": 0.8086288900368741, "grad_norm": 1.2783514184708973, "learning_rate": 1.8599338630400544e-06, "loss": 0.4083, "step": 22697 }, { "epoch": 0.8086645171633682, "grad_norm": 1.1746163850765852, "learning_rate": 1.8592636466404545e-06, "loss": 0.3442, "step": 22698 }, { "epoch": 0.8087001442898623, "grad_norm": 1.6555145334290482, "learning_rate": 1.8585935386404496e-06, "loss": 0.5749, "step": 22699 }, { "epoch": 0.8087357714163564, "grad_norm": 1.2608348228963095, "learning_rate": 1.8579235390489636e-06, "loss": 0.4899, "step": 22700 }, { "epoch": 0.8087713985428505, "grad_norm": 1.2157468562254055, "learning_rate": 1.8572536478749148e-06, "loss": 0.3792, "step": 22701 }, { "epoch": 0.8088070256693446, "grad_norm": 1.3515783237159773, "learning_rate": 1.856583865127226e-06, "loss": 0.4167, "step": 22702 }, { "epoch": 0.8088426527958388, "grad_norm": 1.215045533977688, "learning_rate": 1.855914190814816e-06, "loss": 0.3944, "step": 22703 }, { "epoch": 0.8088782799223329, "grad_norm": 1.4861134603060757, "learning_rate": 1.8552446249465995e-06, "loss": 0.4882, "step": 22704 }, { "epoch": 0.8089139070488269, "grad_norm": 1.2820759110824957, "learning_rate": 1.854575167531496e-06, "loss": 0.3505, "step": 22705 }, { "epoch": 0.8089495341753211, "grad_norm": 1.1322823771720332, "learning_rate": 1.8539058185784143e-06, "loss": 0.3506, "step": 22706 }, { "epoch": 0.8089851613018152, "grad_norm": 1.5071359947794571, "learning_rate": 1.8532365780962702e-06, "loss": 0.4788, "step": 22707 }, { "epoch": 0.8090207884283093, "grad_norm": 1.1957165856407521, "learning_rate": 1.8525674460939757e-06, "loss": 0.3831, "step": 22708 }, { "epoch": 0.8090564155548035, "grad_norm": 2.1109289827137294, "learning_rate": 1.8518984225804437e-06, "loss": 0.5441, "step": 22709 }, { "epoch": 0.8090920426812975, "grad_norm": 1.1905749704166901, "learning_rate": 1.8512295075645758e-06, "loss": 0.398, "step": 22710 }, { "epoch": 0.8091276698077916, "grad_norm": 1.2166554185650338, "learning_rate": 1.8505607010552862e-06, "loss": 0.3764, "step": 22711 }, { "epoch": 0.8091632969342858, "grad_norm": 1.300781223214799, "learning_rate": 1.8498920030614732e-06, "loss": 0.4204, "step": 22712 }, { "epoch": 0.8091989240607799, "grad_norm": 1.1057901593485946, "learning_rate": 1.849223413592046e-06, "loss": 0.354, "step": 22713 }, { "epoch": 0.8092345511872739, "grad_norm": 1.797889466417195, "learning_rate": 1.848554932655905e-06, "loss": 0.4969, "step": 22714 }, { "epoch": 0.8092701783137681, "grad_norm": 1.6543393605178451, "learning_rate": 1.8478865602619533e-06, "loss": 0.5124, "step": 22715 }, { "epoch": 0.8093058054402622, "grad_norm": 1.2210804064766405, "learning_rate": 1.8472182964190922e-06, "loss": 0.4173, "step": 22716 }, { "epoch": 0.8093414325667563, "grad_norm": 1.4127499543089757, "learning_rate": 1.8465501411362153e-06, "loss": 0.4541, "step": 22717 }, { "epoch": 0.8093770596932505, "grad_norm": 1.183225304033765, "learning_rate": 1.8458820944222256e-06, "loss": 0.4216, "step": 22718 }, { "epoch": 0.8094126868197445, "grad_norm": 1.3360837392879288, "learning_rate": 1.8452141562860115e-06, "loss": 0.4354, "step": 22719 }, { "epoch": 0.8094483139462386, "grad_norm": 1.6847871798148883, "learning_rate": 1.8445463267364738e-06, "loss": 0.5766, "step": 22720 }, { "epoch": 0.8094839410727328, "grad_norm": 1.2371205744499947, "learning_rate": 1.8438786057824999e-06, "loss": 0.4147, "step": 22721 }, { "epoch": 0.8095195681992269, "grad_norm": 2.377066100577527, "learning_rate": 1.8432109934329833e-06, "loss": 0.4836, "step": 22722 }, { "epoch": 0.809555195325721, "grad_norm": 1.3836879314199744, "learning_rate": 1.8425434896968132e-06, "loss": 0.4176, "step": 22723 }, { "epoch": 0.8095908224522151, "grad_norm": 1.2986407526307024, "learning_rate": 1.8418760945828806e-06, "loss": 0.3862, "step": 22724 }, { "epoch": 0.8096264495787092, "grad_norm": 1.2689602729338867, "learning_rate": 1.8412088081000668e-06, "loss": 0.4373, "step": 22725 }, { "epoch": 0.8096620767052033, "grad_norm": 1.4752305646146207, "learning_rate": 1.8405416302572621e-06, "loss": 0.6066, "step": 22726 }, { "epoch": 0.8096977038316975, "grad_norm": 1.0715014746709062, "learning_rate": 1.8398745610633495e-06, "loss": 0.3572, "step": 22727 }, { "epoch": 0.8097333309581916, "grad_norm": 1.3650120602309626, "learning_rate": 1.8392076005272098e-06, "loss": 0.4203, "step": 22728 }, { "epoch": 0.8097689580846856, "grad_norm": 1.322210891578016, "learning_rate": 1.8385407486577233e-06, "loss": 0.5345, "step": 22729 }, { "epoch": 0.8098045852111798, "grad_norm": 1.047311793578611, "learning_rate": 1.8378740054637722e-06, "loss": 0.4198, "step": 22730 }, { "epoch": 0.8098402123376739, "grad_norm": 1.1172112750280525, "learning_rate": 1.837207370954236e-06, "loss": 0.4064, "step": 22731 }, { "epoch": 0.809875839464168, "grad_norm": 1.7727423940827969, "learning_rate": 1.8365408451379873e-06, "loss": 0.541, "step": 22732 }, { "epoch": 0.8099114665906622, "grad_norm": 1.8622541285360767, "learning_rate": 1.8358744280239048e-06, "loss": 0.5041, "step": 22733 }, { "epoch": 0.8099470937171562, "grad_norm": 1.268504732188282, "learning_rate": 1.8352081196208581e-06, "loss": 0.4344, "step": 22734 }, { "epoch": 0.8099827208436503, "grad_norm": 1.344985512275361, "learning_rate": 1.8345419199377223e-06, "loss": 0.446, "step": 22735 }, { "epoch": 0.8100183479701445, "grad_norm": 1.2309527960981959, "learning_rate": 1.8338758289833692e-06, "loss": 0.3646, "step": 22736 }, { "epoch": 0.8100539750966386, "grad_norm": 1.1534041018409675, "learning_rate": 1.8332098467666658e-06, "loss": 0.4432, "step": 22737 }, { "epoch": 0.8100896022231326, "grad_norm": 1.1219525029948194, "learning_rate": 1.8325439732964844e-06, "loss": 0.3675, "step": 22738 }, { "epoch": 0.8101252293496268, "grad_norm": 1.5365484148028752, "learning_rate": 1.8318782085816867e-06, "loss": 0.5239, "step": 22739 }, { "epoch": 0.8101608564761209, "grad_norm": 1.1657936565845848, "learning_rate": 1.831212552631142e-06, "loss": 0.41, "step": 22740 }, { "epoch": 0.810196483602615, "grad_norm": 1.3324477231395262, "learning_rate": 1.830547005453709e-06, "loss": 0.5399, "step": 22741 }, { "epoch": 0.8102321107291092, "grad_norm": 1.2764053381212301, "learning_rate": 1.829881567058256e-06, "loss": 0.4458, "step": 22742 }, { "epoch": 0.8102677378556032, "grad_norm": 1.1944116276971093, "learning_rate": 1.829216237453637e-06, "loss": 0.4237, "step": 22743 }, { "epoch": 0.8103033649820973, "grad_norm": 1.2406658538901687, "learning_rate": 1.8285510166487154e-06, "loss": 0.3558, "step": 22744 }, { "epoch": 0.8103389921085915, "grad_norm": 1.8245496153394842, "learning_rate": 1.827885904652349e-06, "loss": 0.4371, "step": 22745 }, { "epoch": 0.8103746192350856, "grad_norm": 1.6028036073597747, "learning_rate": 1.8272209014733955e-06, "loss": 0.5359, "step": 22746 }, { "epoch": 0.8104102463615797, "grad_norm": 1.168101098400854, "learning_rate": 1.8265560071207056e-06, "loss": 0.3897, "step": 22747 }, { "epoch": 0.8104458734880738, "grad_norm": 1.173161245764079, "learning_rate": 1.8258912216031355e-06, "loss": 0.4283, "step": 22748 }, { "epoch": 0.8104815006145679, "grad_norm": 1.6158580146548431, "learning_rate": 1.8252265449295404e-06, "loss": 0.5307, "step": 22749 }, { "epoch": 0.810517127741062, "grad_norm": 1.0681064183876345, "learning_rate": 1.8245619771087642e-06, "loss": 0.4092, "step": 22750 }, { "epoch": 0.8105527548675562, "grad_norm": 1.2301593601481924, "learning_rate": 1.8238975181496598e-06, "loss": 0.4124, "step": 22751 }, { "epoch": 0.8105883819940503, "grad_norm": 1.7294590868560173, "learning_rate": 1.8232331680610749e-06, "loss": 0.478, "step": 22752 }, { "epoch": 0.8106240091205443, "grad_norm": 1.1961835556890539, "learning_rate": 1.8225689268518575e-06, "loss": 0.3864, "step": 22753 }, { "epoch": 0.8106596362470385, "grad_norm": 1.2257943541041387, "learning_rate": 1.8219047945308488e-06, "loss": 0.4264, "step": 22754 }, { "epoch": 0.8106952633735326, "grad_norm": 1.4891863949529809, "learning_rate": 1.8212407711068958e-06, "loss": 0.5501, "step": 22755 }, { "epoch": 0.8107308905000267, "grad_norm": 1.2531147519349708, "learning_rate": 1.820576856588837e-06, "loss": 0.4222, "step": 22756 }, { "epoch": 0.8107665176265209, "grad_norm": 1.126597672978765, "learning_rate": 1.8199130509855157e-06, "loss": 0.4108, "step": 22757 }, { "epoch": 0.8108021447530149, "grad_norm": 1.2599427909453103, "learning_rate": 1.8192493543057676e-06, "loss": 0.4179, "step": 22758 }, { "epoch": 0.810837771879509, "grad_norm": 1.5469989130890174, "learning_rate": 1.8185857665584328e-06, "loss": 0.5473, "step": 22759 }, { "epoch": 0.8108733990060032, "grad_norm": 1.3513353993251715, "learning_rate": 1.8179222877523472e-06, "loss": 0.4102, "step": 22760 }, { "epoch": 0.8109090261324973, "grad_norm": 1.1742924161630217, "learning_rate": 1.817258917896344e-06, "loss": 0.3709, "step": 22761 }, { "epoch": 0.8109446532589913, "grad_norm": 1.7313754695049692, "learning_rate": 1.8165956569992615e-06, "loss": 0.5717, "step": 22762 }, { "epoch": 0.8109802803854855, "grad_norm": 1.1232575151021302, "learning_rate": 1.815932505069925e-06, "loss": 0.431, "step": 22763 }, { "epoch": 0.8110159075119796, "grad_norm": 1.252546612236902, "learning_rate": 1.8152694621171696e-06, "loss": 0.4219, "step": 22764 }, { "epoch": 0.8110515346384737, "grad_norm": 1.2866030360554515, "learning_rate": 1.8146065281498193e-06, "loss": 0.4022, "step": 22765 }, { "epoch": 0.8110871617649679, "grad_norm": 1.4381903670571465, "learning_rate": 1.8139437031767048e-06, "loss": 0.5219, "step": 22766 }, { "epoch": 0.8111227888914619, "grad_norm": 1.6145885955199566, "learning_rate": 1.8132809872066515e-06, "loss": 0.5151, "step": 22767 }, { "epoch": 0.811158416017956, "grad_norm": 1.3770621939310472, "learning_rate": 1.8126183802484864e-06, "loss": 0.407, "step": 22768 }, { "epoch": 0.8111940431444502, "grad_norm": 1.1309007862362983, "learning_rate": 1.8119558823110284e-06, "loss": 0.3645, "step": 22769 }, { "epoch": 0.8112296702709443, "grad_norm": 1.1729078560340473, "learning_rate": 1.8112934934031023e-06, "loss": 0.3864, "step": 22770 }, { "epoch": 0.8112652973974384, "grad_norm": 1.5474109567126686, "learning_rate": 1.8106312135335246e-06, "loss": 0.5121, "step": 22771 }, { "epoch": 0.8113009245239325, "grad_norm": 1.780410284685206, "learning_rate": 1.8099690427111172e-06, "loss": 0.6022, "step": 22772 }, { "epoch": 0.8113365516504266, "grad_norm": 1.6823449562142079, "learning_rate": 1.8093069809446962e-06, "loss": 0.4639, "step": 22773 }, { "epoch": 0.8113721787769207, "grad_norm": 1.4964034867499736, "learning_rate": 1.8086450282430766e-06, "loss": 0.5533, "step": 22774 }, { "epoch": 0.8114078059034149, "grad_norm": 1.3154732254098467, "learning_rate": 1.8079831846150775e-06, "loss": 0.4304, "step": 22775 }, { "epoch": 0.811443433029909, "grad_norm": 1.153222405769571, "learning_rate": 1.8073214500695058e-06, "loss": 0.4428, "step": 22776 }, { "epoch": 0.811479060156403, "grad_norm": 1.8660065884727126, "learning_rate": 1.8066598246151767e-06, "loss": 0.4862, "step": 22777 }, { "epoch": 0.8115146872828972, "grad_norm": 1.3605577519775356, "learning_rate": 1.8059983082608977e-06, "loss": 0.4455, "step": 22778 }, { "epoch": 0.8115503144093913, "grad_norm": 1.8162479557828488, "learning_rate": 1.8053369010154808e-06, "loss": 0.5346, "step": 22779 }, { "epoch": 0.8115859415358854, "grad_norm": 1.6672185993029585, "learning_rate": 1.8046756028877277e-06, "loss": 0.5203, "step": 22780 }, { "epoch": 0.8116215686623796, "grad_norm": 1.5084231783056665, "learning_rate": 1.8040144138864468e-06, "loss": 0.4942, "step": 22781 }, { "epoch": 0.8116571957888736, "grad_norm": 1.7790601433622535, "learning_rate": 1.8033533340204434e-06, "loss": 0.5186, "step": 22782 }, { "epoch": 0.8116928229153677, "grad_norm": 1.8133124737100625, "learning_rate": 1.8026923632985195e-06, "loss": 0.4878, "step": 22783 }, { "epoch": 0.8117284500418619, "grad_norm": 1.4032736162115522, "learning_rate": 1.8020315017294776e-06, "loss": 0.3853, "step": 22784 }, { "epoch": 0.811764077168356, "grad_norm": 1.4366626207474686, "learning_rate": 1.8013707493221155e-06, "loss": 0.4359, "step": 22785 }, { "epoch": 0.81179970429485, "grad_norm": 1.6115517299421909, "learning_rate": 1.800710106085234e-06, "loss": 0.5143, "step": 22786 }, { "epoch": 0.8118353314213442, "grad_norm": 1.8057061788617632, "learning_rate": 1.8000495720276257e-06, "loss": 0.5577, "step": 22787 }, { "epoch": 0.8118709585478383, "grad_norm": 1.4439340656528008, "learning_rate": 1.7993891471580894e-06, "loss": 0.3975, "step": 22788 }, { "epoch": 0.8119065856743324, "grad_norm": 1.2533198382385617, "learning_rate": 1.798728831485418e-06, "loss": 0.4129, "step": 22789 }, { "epoch": 0.8119422128008266, "grad_norm": 1.1853835597093192, "learning_rate": 1.7980686250184077e-06, "loss": 0.4244, "step": 22790 }, { "epoch": 0.8119778399273206, "grad_norm": 1.2680205375526294, "learning_rate": 1.7974085277658426e-06, "loss": 0.4113, "step": 22791 }, { "epoch": 0.8120134670538147, "grad_norm": 1.3260900432122258, "learning_rate": 1.7967485397365202e-06, "loss": 0.3655, "step": 22792 }, { "epoch": 0.8120490941803089, "grad_norm": 1.6601948810136047, "learning_rate": 1.7960886609392215e-06, "loss": 0.4846, "step": 22793 }, { "epoch": 0.812084721306803, "grad_norm": 1.4410697963542491, "learning_rate": 1.7954288913827366e-06, "loss": 0.4715, "step": 22794 }, { "epoch": 0.8121203484332972, "grad_norm": 1.2810072230062424, "learning_rate": 1.7947692310758535e-06, "loss": 0.4086, "step": 22795 }, { "epoch": 0.8121559755597912, "grad_norm": 2.0805128324102364, "learning_rate": 1.7941096800273495e-06, "loss": 0.5398, "step": 22796 }, { "epoch": 0.8121916026862853, "grad_norm": 1.389037855776806, "learning_rate": 1.7934502382460118e-06, "loss": 0.4767, "step": 22797 }, { "epoch": 0.8122272298127795, "grad_norm": 1.1942930145025772, "learning_rate": 1.7927909057406201e-06, "loss": 0.4012, "step": 22798 }, { "epoch": 0.8122628569392736, "grad_norm": 1.7187025173565007, "learning_rate": 1.7921316825199564e-06, "loss": 0.3906, "step": 22799 }, { "epoch": 0.8122984840657677, "grad_norm": 1.2263963387855044, "learning_rate": 1.7914725685927936e-06, "loss": 0.4242, "step": 22800 }, { "epoch": 0.8123341111922618, "grad_norm": 2.0418537273197113, "learning_rate": 1.7908135639679137e-06, "loss": 0.5798, "step": 22801 }, { "epoch": 0.8123697383187559, "grad_norm": 2.092432856788505, "learning_rate": 1.7901546686540872e-06, "loss": 0.5352, "step": 22802 }, { "epoch": 0.81240536544525, "grad_norm": 1.475797277442829, "learning_rate": 1.7894958826600884e-06, "loss": 0.3813, "step": 22803 }, { "epoch": 0.8124409925717442, "grad_norm": 1.6117270828405506, "learning_rate": 1.7888372059946912e-06, "loss": 0.478, "step": 22804 }, { "epoch": 0.8124766196982383, "grad_norm": 1.6182453414703035, "learning_rate": 1.7881786386666688e-06, "loss": 0.5104, "step": 22805 }, { "epoch": 0.8125122468247323, "grad_norm": 1.164296962051241, "learning_rate": 1.7875201806847842e-06, "loss": 0.4238, "step": 22806 }, { "epoch": 0.8125478739512265, "grad_norm": 1.2676085460906084, "learning_rate": 1.786861832057809e-06, "loss": 0.3921, "step": 22807 }, { "epoch": 0.8125835010777206, "grad_norm": 1.1316928353608529, "learning_rate": 1.786203592794512e-06, "loss": 0.4152, "step": 22808 }, { "epoch": 0.8126191282042147, "grad_norm": 1.1986228026577634, "learning_rate": 1.7855454629036516e-06, "loss": 0.3584, "step": 22809 }, { "epoch": 0.8126547553307089, "grad_norm": 1.1633589662690382, "learning_rate": 1.7848874423939966e-06, "loss": 0.4511, "step": 22810 }, { "epoch": 0.8126903824572029, "grad_norm": 1.5058262343051192, "learning_rate": 1.7842295312743064e-06, "loss": 0.5199, "step": 22811 }, { "epoch": 0.812726009583697, "grad_norm": 1.2555110893836379, "learning_rate": 1.783571729553345e-06, "loss": 0.3762, "step": 22812 }, { "epoch": 0.8127616367101912, "grad_norm": 1.3172489329277208, "learning_rate": 1.7829140372398667e-06, "loss": 0.4089, "step": 22813 }, { "epoch": 0.8127972638366853, "grad_norm": 2.005983437213542, "learning_rate": 1.7822564543426336e-06, "loss": 0.4993, "step": 22814 }, { "epoch": 0.8128328909631793, "grad_norm": 1.256406742793532, "learning_rate": 1.781598980870397e-06, "loss": 0.417, "step": 22815 }, { "epoch": 0.8128685180896735, "grad_norm": 1.398997211157984, "learning_rate": 1.7809416168319172e-06, "loss": 0.4367, "step": 22816 }, { "epoch": 0.8129041452161676, "grad_norm": 1.1964790845560558, "learning_rate": 1.7802843622359422e-06, "loss": 0.4194, "step": 22817 }, { "epoch": 0.8129397723426617, "grad_norm": 1.4702730962712907, "learning_rate": 1.7796272170912255e-06, "loss": 0.4837, "step": 22818 }, { "epoch": 0.8129753994691559, "grad_norm": 1.5486605840470618, "learning_rate": 1.7789701814065186e-06, "loss": 0.5586, "step": 22819 }, { "epoch": 0.8130110265956499, "grad_norm": 1.247385022505396, "learning_rate": 1.7783132551905702e-06, "loss": 0.4372, "step": 22820 }, { "epoch": 0.813046653722144, "grad_norm": 1.9063159219259498, "learning_rate": 1.777656438452129e-06, "loss": 0.5544, "step": 22821 }, { "epoch": 0.8130822808486382, "grad_norm": 1.2285176638719078, "learning_rate": 1.7769997311999375e-06, "loss": 0.4644, "step": 22822 }, { "epoch": 0.8131179079751323, "grad_norm": 1.7714924262874994, "learning_rate": 1.7763431334427438e-06, "loss": 0.5207, "step": 22823 }, { "epoch": 0.8131535351016264, "grad_norm": 2.043492386238122, "learning_rate": 1.7756866451892885e-06, "loss": 0.4629, "step": 22824 }, { "epoch": 0.8131891622281205, "grad_norm": 1.1957570881092132, "learning_rate": 1.7750302664483121e-06, "loss": 0.38, "step": 22825 }, { "epoch": 0.8132247893546146, "grad_norm": 1.6351343458924998, "learning_rate": 1.7743739972285578e-06, "loss": 0.6699, "step": 22826 }, { "epoch": 0.8132604164811087, "grad_norm": 1.8232127155727353, "learning_rate": 1.7737178375387653e-06, "loss": 0.5355, "step": 22827 }, { "epoch": 0.8132960436076029, "grad_norm": 1.8088963268645202, "learning_rate": 1.7730617873876666e-06, "loss": 0.5139, "step": 22828 }, { "epoch": 0.813331670734097, "grad_norm": 1.243302342845214, "learning_rate": 1.772405846784001e-06, "loss": 0.4326, "step": 22829 }, { "epoch": 0.813367297860591, "grad_norm": 1.589947308592836, "learning_rate": 1.7717500157365053e-06, "loss": 0.4663, "step": 22830 }, { "epoch": 0.8134029249870852, "grad_norm": 1.2893595281874066, "learning_rate": 1.7710942942539055e-06, "loss": 0.3748, "step": 22831 }, { "epoch": 0.8134385521135793, "grad_norm": 1.431886143823492, "learning_rate": 1.7704386823449404e-06, "loss": 0.4085, "step": 22832 }, { "epoch": 0.8134741792400734, "grad_norm": 1.345678900844729, "learning_rate": 1.769783180018334e-06, "loss": 0.5243, "step": 22833 }, { "epoch": 0.8135098063665676, "grad_norm": 1.4081713930678281, "learning_rate": 1.769127787282816e-06, "loss": 0.512, "step": 22834 }, { "epoch": 0.8135454334930616, "grad_norm": 1.2119874407881837, "learning_rate": 1.7684725041471162e-06, "loss": 0.3787, "step": 22835 }, { "epoch": 0.8135810606195557, "grad_norm": 1.158590961863388, "learning_rate": 1.7678173306199597e-06, "loss": 0.4005, "step": 22836 }, { "epoch": 0.8136166877460499, "grad_norm": 1.5622893173073942, "learning_rate": 1.7671622667100674e-06, "loss": 0.533, "step": 22837 }, { "epoch": 0.813652314872544, "grad_norm": 1.5088240203397008, "learning_rate": 1.7665073124261666e-06, "loss": 0.5217, "step": 22838 }, { "epoch": 0.813687941999038, "grad_norm": 1.5291529946242564, "learning_rate": 1.765852467776973e-06, "loss": 0.5188, "step": 22839 }, { "epoch": 0.8137235691255322, "grad_norm": 1.265913819582954, "learning_rate": 1.765197732771209e-06, "loss": 0.394, "step": 22840 }, { "epoch": 0.8137591962520263, "grad_norm": 1.3213171413946068, "learning_rate": 1.7645431074175923e-06, "loss": 0.4184, "step": 22841 }, { "epoch": 0.8137948233785204, "grad_norm": 1.7061325625041484, "learning_rate": 1.7638885917248416e-06, "loss": 0.5267, "step": 22842 }, { "epoch": 0.8138304505050146, "grad_norm": 1.5162710389991825, "learning_rate": 1.7632341857016733e-06, "loss": 0.4996, "step": 22843 }, { "epoch": 0.8138660776315086, "grad_norm": 1.1786144051582843, "learning_rate": 1.7625798893567958e-06, "loss": 0.3905, "step": 22844 }, { "epoch": 0.8139017047580027, "grad_norm": 1.492569684854169, "learning_rate": 1.7619257026989278e-06, "loss": 0.5301, "step": 22845 }, { "epoch": 0.8139373318844969, "grad_norm": 1.80702217110704, "learning_rate": 1.7612716257367745e-06, "loss": 0.4994, "step": 22846 }, { "epoch": 0.813972959010991, "grad_norm": 1.4425610781335225, "learning_rate": 1.76061765847905e-06, "loss": 0.4774, "step": 22847 }, { "epoch": 0.8140085861374851, "grad_norm": 1.1866553548278054, "learning_rate": 1.7599638009344566e-06, "loss": 0.4204, "step": 22848 }, { "epoch": 0.8140442132639792, "grad_norm": 1.087903832655352, "learning_rate": 1.759310053111709e-06, "loss": 0.4512, "step": 22849 }, { "epoch": 0.8140798403904733, "grad_norm": 1.5133873437267509, "learning_rate": 1.7586564150195052e-06, "loss": 0.5051, "step": 22850 }, { "epoch": 0.8141154675169674, "grad_norm": 1.7958315126551558, "learning_rate": 1.7580028866665544e-06, "loss": 0.5289, "step": 22851 }, { "epoch": 0.8141510946434616, "grad_norm": 1.227014689115959, "learning_rate": 1.7573494680615532e-06, "loss": 0.4156, "step": 22852 }, { "epoch": 0.8141867217699557, "grad_norm": 1.206059892695721, "learning_rate": 1.7566961592132047e-06, "loss": 0.3786, "step": 22853 }, { "epoch": 0.8142223488964497, "grad_norm": 1.254573547301708, "learning_rate": 1.756042960130212e-06, "loss": 0.4483, "step": 22854 }, { "epoch": 0.8142579760229439, "grad_norm": 1.9321241160362932, "learning_rate": 1.7553898708212658e-06, "loss": 0.5422, "step": 22855 }, { "epoch": 0.814293603149438, "grad_norm": 1.2148715634260336, "learning_rate": 1.7547368912950669e-06, "loss": 0.3825, "step": 22856 }, { "epoch": 0.8143292302759321, "grad_norm": 1.447079279781165, "learning_rate": 1.7540840215603094e-06, "loss": 0.4783, "step": 22857 }, { "epoch": 0.8143648574024263, "grad_norm": 1.9176964601757482, "learning_rate": 1.7534312616256889e-06, "loss": 0.5243, "step": 22858 }, { "epoch": 0.8144004845289203, "grad_norm": 1.1800770947610488, "learning_rate": 1.7527786114998934e-06, "loss": 0.4018, "step": 22859 }, { "epoch": 0.8144361116554144, "grad_norm": 1.6145663464512954, "learning_rate": 1.7521260711916166e-06, "loss": 0.544, "step": 22860 }, { "epoch": 0.8144717387819086, "grad_norm": 1.2645470119549114, "learning_rate": 1.7514736407095444e-06, "loss": 0.4085, "step": 22861 }, { "epoch": 0.8145073659084027, "grad_norm": 1.1980289393371752, "learning_rate": 1.7508213200623658e-06, "loss": 0.3895, "step": 22862 }, { "epoch": 0.8145429930348967, "grad_norm": 1.2131779025014815, "learning_rate": 1.750169109258767e-06, "loss": 0.3936, "step": 22863 }, { "epoch": 0.8145786201613909, "grad_norm": 1.1895138801047052, "learning_rate": 1.7495170083074332e-06, "loss": 0.4507, "step": 22864 }, { "epoch": 0.814614247287885, "grad_norm": 1.7245137216491144, "learning_rate": 1.7488650172170496e-06, "loss": 0.5378, "step": 22865 }, { "epoch": 0.8146498744143791, "grad_norm": 1.080153697103402, "learning_rate": 1.7482131359962917e-06, "loss": 0.3843, "step": 22866 }, { "epoch": 0.8146855015408733, "grad_norm": 1.9961143340289944, "learning_rate": 1.747561364653848e-06, "loss": 0.4813, "step": 22867 }, { "epoch": 0.8147211286673673, "grad_norm": 1.2606760614071104, "learning_rate": 1.7469097031983894e-06, "loss": 0.4121, "step": 22868 }, { "epoch": 0.8147567557938614, "grad_norm": 1.409405256443239, "learning_rate": 1.746258151638599e-06, "loss": 0.4862, "step": 22869 }, { "epoch": 0.8147923829203556, "grad_norm": 1.51881765651804, "learning_rate": 1.7456067099831475e-06, "loss": 0.4976, "step": 22870 }, { "epoch": 0.8148280100468497, "grad_norm": 1.176561805481948, "learning_rate": 1.7449553782407136e-06, "loss": 0.4248, "step": 22871 }, { "epoch": 0.8148636371733438, "grad_norm": 1.5404536724729403, "learning_rate": 1.744304156419967e-06, "loss": 0.5314, "step": 22872 }, { "epoch": 0.8148992642998379, "grad_norm": 1.2763934787638642, "learning_rate": 1.7436530445295852e-06, "loss": 0.3862, "step": 22873 }, { "epoch": 0.814934891426332, "grad_norm": 2.013398776354166, "learning_rate": 1.74300204257823e-06, "loss": 0.4717, "step": 22874 }, { "epoch": 0.8149705185528261, "grad_norm": 2.2851130926695906, "learning_rate": 1.742351150574575e-06, "loss": 0.6168, "step": 22875 }, { "epoch": 0.8150061456793203, "grad_norm": 1.2937295630259735, "learning_rate": 1.7417003685272882e-06, "loss": 0.3995, "step": 22876 }, { "epoch": 0.8150417728058144, "grad_norm": 1.2128267074513432, "learning_rate": 1.7410496964450296e-06, "loss": 0.3524, "step": 22877 }, { "epoch": 0.8150773999323084, "grad_norm": 1.2021263729715945, "learning_rate": 1.740399134336469e-06, "loss": 0.4031, "step": 22878 }, { "epoch": 0.8151130270588026, "grad_norm": 1.5947994956970308, "learning_rate": 1.739748682210266e-06, "loss": 0.5705, "step": 22879 }, { "epoch": 0.8151486541852967, "grad_norm": 1.0758662312160705, "learning_rate": 1.739098340075085e-06, "loss": 0.3395, "step": 22880 }, { "epoch": 0.8151842813117908, "grad_norm": 1.6036782309453255, "learning_rate": 1.738448107939581e-06, "loss": 0.4977, "step": 22881 }, { "epoch": 0.815219908438285, "grad_norm": 1.6580652521664576, "learning_rate": 1.7377979858124184e-06, "loss": 0.4927, "step": 22882 }, { "epoch": 0.815255535564779, "grad_norm": 1.3727946823994182, "learning_rate": 1.7371479737022478e-06, "loss": 0.4032, "step": 22883 }, { "epoch": 0.8152911626912731, "grad_norm": 1.6170047073289464, "learning_rate": 1.7364980716177294e-06, "loss": 0.5045, "step": 22884 }, { "epoch": 0.8153267898177673, "grad_norm": 1.7808362258520105, "learning_rate": 1.7358482795675124e-06, "loss": 0.5301, "step": 22885 }, { "epoch": 0.8153624169442614, "grad_norm": 1.3232665758279072, "learning_rate": 1.7351985975602526e-06, "loss": 0.3968, "step": 22886 }, { "epoch": 0.8153980440707554, "grad_norm": 1.261912589839385, "learning_rate": 1.7345490256045993e-06, "loss": 0.3894, "step": 22887 }, { "epoch": 0.8154336711972496, "grad_norm": 1.080219256053501, "learning_rate": 1.7338995637092038e-06, "loss": 0.3953, "step": 22888 }, { "epoch": 0.8154692983237437, "grad_norm": 1.5813496171528632, "learning_rate": 1.7332502118827154e-06, "loss": 0.5221, "step": 22889 }, { "epoch": 0.8155049254502378, "grad_norm": 1.5705094973524998, "learning_rate": 1.7326009701337764e-06, "loss": 0.5246, "step": 22890 }, { "epoch": 0.815540552576732, "grad_norm": 1.4885728996697196, "learning_rate": 1.7319518384710353e-06, "loss": 0.5789, "step": 22891 }, { "epoch": 0.815576179703226, "grad_norm": 1.4277057080210727, "learning_rate": 1.731302816903133e-06, "loss": 0.4489, "step": 22892 }, { "epoch": 0.8156118068297201, "grad_norm": 1.953976087219353, "learning_rate": 1.730653905438714e-06, "loss": 0.5206, "step": 22893 }, { "epoch": 0.8156474339562143, "grad_norm": 1.2258376382314966, "learning_rate": 1.7300051040864163e-06, "loss": 0.4366, "step": 22894 }, { "epoch": 0.8156830610827084, "grad_norm": 1.1695188056234675, "learning_rate": 1.7293564128548845e-06, "loss": 0.416, "step": 22895 }, { "epoch": 0.8157186882092025, "grad_norm": 1.2262513660074659, "learning_rate": 1.7287078317527505e-06, "loss": 0.4223, "step": 22896 }, { "epoch": 0.8157543153356966, "grad_norm": 1.5714936204169625, "learning_rate": 1.7280593607886554e-06, "loss": 0.5074, "step": 22897 }, { "epoch": 0.8157899424621907, "grad_norm": 1.4699386449915561, "learning_rate": 1.7274109999712295e-06, "loss": 0.3889, "step": 22898 }, { "epoch": 0.8158255695886848, "grad_norm": 1.1228371750443331, "learning_rate": 1.7267627493091077e-06, "loss": 0.3747, "step": 22899 }, { "epoch": 0.815861196715179, "grad_norm": 1.4438828967941493, "learning_rate": 1.7261146088109226e-06, "loss": 0.5193, "step": 22900 }, { "epoch": 0.8158968238416731, "grad_norm": 1.5731308241901478, "learning_rate": 1.7254665784853053e-06, "loss": 0.5143, "step": 22901 }, { "epoch": 0.8159324509681671, "grad_norm": 1.8212471949201108, "learning_rate": 1.7248186583408856e-06, "loss": 0.5131, "step": 22902 }, { "epoch": 0.8159680780946613, "grad_norm": 1.6653718354519726, "learning_rate": 1.7241708483862873e-06, "loss": 0.4786, "step": 22903 }, { "epoch": 0.8160037052211554, "grad_norm": 1.2046862563831042, "learning_rate": 1.723523148630142e-06, "loss": 0.4029, "step": 22904 }, { "epoch": 0.8160393323476495, "grad_norm": 1.4802124699771322, "learning_rate": 1.7228755590810674e-06, "loss": 0.497, "step": 22905 }, { "epoch": 0.8160749594741437, "grad_norm": 1.1833127369519036, "learning_rate": 1.722228079747692e-06, "loss": 0.4317, "step": 22906 }, { "epoch": 0.8161105866006377, "grad_norm": 1.674973444749172, "learning_rate": 1.721580710638634e-06, "loss": 0.4908, "step": 22907 }, { "epoch": 0.8161462137271318, "grad_norm": 1.1947019114882513, "learning_rate": 1.7209334517625154e-06, "loss": 0.3984, "step": 22908 }, { "epoch": 0.816181840853626, "grad_norm": 1.373376070552769, "learning_rate": 1.7202863031279537e-06, "loss": 0.4136, "step": 22909 }, { "epoch": 0.8162174679801201, "grad_norm": 1.1202362958527152, "learning_rate": 1.7196392647435678e-06, "loss": 0.4073, "step": 22910 }, { "epoch": 0.8162530951066141, "grad_norm": 1.0996233583091226, "learning_rate": 1.718992336617975e-06, "loss": 0.4285, "step": 22911 }, { "epoch": 0.8162887222331083, "grad_norm": 1.4942344948749036, "learning_rate": 1.7183455187597842e-06, "loss": 0.5071, "step": 22912 }, { "epoch": 0.8163243493596024, "grad_norm": 1.1279670439472138, "learning_rate": 1.7176988111776149e-06, "loss": 0.3669, "step": 22913 }, { "epoch": 0.8163599764860965, "grad_norm": 1.202093462134913, "learning_rate": 1.7170522138800717e-06, "loss": 0.4032, "step": 22914 }, { "epoch": 0.8163956036125907, "grad_norm": 1.6209168693655973, "learning_rate": 1.7164057268757683e-06, "loss": 0.537, "step": 22915 }, { "epoch": 0.8164312307390847, "grad_norm": 1.4815197819982269, "learning_rate": 1.7157593501733117e-06, "loss": 0.4115, "step": 22916 }, { "epoch": 0.8164668578655788, "grad_norm": 1.5785468456751774, "learning_rate": 1.7151130837813123e-06, "loss": 0.4993, "step": 22917 }, { "epoch": 0.816502484992073, "grad_norm": 1.9406415520697347, "learning_rate": 1.7144669277083714e-06, "loss": 0.551, "step": 22918 }, { "epoch": 0.8165381121185671, "grad_norm": 1.3009528429971193, "learning_rate": 1.7138208819630964e-06, "loss": 0.3866, "step": 22919 }, { "epoch": 0.8165737392450612, "grad_norm": 1.1052545692890348, "learning_rate": 1.7131749465540859e-06, "loss": 0.3813, "step": 22920 }, { "epoch": 0.8166093663715553, "grad_norm": 1.2306499650676974, "learning_rate": 1.7125291214899431e-06, "loss": 0.4198, "step": 22921 }, { "epoch": 0.8166449934980494, "grad_norm": 1.7819194975716937, "learning_rate": 1.7118834067792688e-06, "loss": 0.5119, "step": 22922 }, { "epoch": 0.8166806206245435, "grad_norm": 1.2908100627839105, "learning_rate": 1.7112378024306565e-06, "loss": 0.4037, "step": 22923 }, { "epoch": 0.8167162477510377, "grad_norm": 1.157934517352459, "learning_rate": 1.710592308452711e-06, "loss": 0.4078, "step": 22924 }, { "epoch": 0.8167518748775318, "grad_norm": 1.5184409508510557, "learning_rate": 1.7099469248540202e-06, "loss": 0.4692, "step": 22925 }, { "epoch": 0.8167875020040258, "grad_norm": 1.5027644611531796, "learning_rate": 1.7093016516431837e-06, "loss": 0.4919, "step": 22926 }, { "epoch": 0.81682312913052, "grad_norm": 1.565724436507605, "learning_rate": 1.7086564888287872e-06, "loss": 0.479, "step": 22927 }, { "epoch": 0.8168587562570141, "grad_norm": 1.1203450607591992, "learning_rate": 1.7080114364194278e-06, "loss": 0.392, "step": 22928 }, { "epoch": 0.8168943833835082, "grad_norm": 1.6868170278873584, "learning_rate": 1.707366494423689e-06, "loss": 0.4843, "step": 22929 }, { "epoch": 0.8169300105100024, "grad_norm": 1.2037054511771563, "learning_rate": 1.7067216628501616e-06, "loss": 0.3599, "step": 22930 }, { "epoch": 0.8169656376364964, "grad_norm": 1.0963536585274398, "learning_rate": 1.706076941707432e-06, "loss": 0.3818, "step": 22931 }, { "epoch": 0.8170012647629905, "grad_norm": 1.237965889286416, "learning_rate": 1.7054323310040877e-06, "loss": 0.4254, "step": 22932 }, { "epoch": 0.8170368918894847, "grad_norm": 1.1185081197153703, "learning_rate": 1.7047878307487065e-06, "loss": 0.4118, "step": 22933 }, { "epoch": 0.8170725190159788, "grad_norm": 1.255254404685596, "learning_rate": 1.7041434409498737e-06, "loss": 0.3898, "step": 22934 }, { "epoch": 0.8171081461424728, "grad_norm": 1.26522947340772, "learning_rate": 1.7034991616161723e-06, "loss": 0.4312, "step": 22935 }, { "epoch": 0.817143773268967, "grad_norm": 1.084060280659275, "learning_rate": 1.7028549927561756e-06, "loss": 0.3776, "step": 22936 }, { "epoch": 0.8171794003954611, "grad_norm": 1.0940474036390146, "learning_rate": 1.7022109343784643e-06, "loss": 0.3969, "step": 22937 }, { "epoch": 0.8172150275219552, "grad_norm": 1.152131024188537, "learning_rate": 1.701566986491614e-06, "loss": 0.3907, "step": 22938 }, { "epoch": 0.8172506546484494, "grad_norm": 1.182170160301443, "learning_rate": 1.700923149104202e-06, "loss": 0.423, "step": 22939 }, { "epoch": 0.8172862817749434, "grad_norm": 1.358067281751834, "learning_rate": 1.7002794222247966e-06, "loss": 0.4007, "step": 22940 }, { "epoch": 0.8173219089014375, "grad_norm": 1.153127389089649, "learning_rate": 1.6996358058619733e-06, "loss": 0.4007, "step": 22941 }, { "epoch": 0.8173575360279317, "grad_norm": 1.3719653091830672, "learning_rate": 1.6989923000243003e-06, "loss": 0.4499, "step": 22942 }, { "epoch": 0.8173931631544258, "grad_norm": 1.5004620380110087, "learning_rate": 1.6983489047203482e-06, "loss": 0.5405, "step": 22943 }, { "epoch": 0.8174287902809199, "grad_norm": 1.1845972292263727, "learning_rate": 1.6977056199586806e-06, "loss": 0.4086, "step": 22944 }, { "epoch": 0.817464417407414, "grad_norm": 1.1365614909776813, "learning_rate": 1.6970624457478658e-06, "loss": 0.4368, "step": 22945 }, { "epoch": 0.8175000445339081, "grad_norm": 1.2495241950941, "learning_rate": 1.6964193820964669e-06, "loss": 0.437, "step": 22946 }, { "epoch": 0.8175356716604022, "grad_norm": 1.3553708998545115, "learning_rate": 1.6957764290130484e-06, "loss": 0.4785, "step": 22947 }, { "epoch": 0.8175712987868964, "grad_norm": 1.4236599879535479, "learning_rate": 1.6951335865061746e-06, "loss": 0.5355, "step": 22948 }, { "epoch": 0.8176069259133905, "grad_norm": 1.423453270887764, "learning_rate": 1.6944908545843986e-06, "loss": 0.4687, "step": 22949 }, { "epoch": 0.8176425530398845, "grad_norm": 1.2521552586369251, "learning_rate": 1.6938482332562844e-06, "loss": 0.4082, "step": 22950 }, { "epoch": 0.8176781801663787, "grad_norm": 1.7565842995888086, "learning_rate": 1.6932057225303844e-06, "loss": 0.4551, "step": 22951 }, { "epoch": 0.8177138072928728, "grad_norm": 1.2785246204434064, "learning_rate": 1.692563322415256e-06, "loss": 0.4411, "step": 22952 }, { "epoch": 0.8177494344193669, "grad_norm": 1.598396622543715, "learning_rate": 1.6919210329194535e-06, "loss": 0.4963, "step": 22953 }, { "epoch": 0.8177850615458611, "grad_norm": 1.8985609302424313, "learning_rate": 1.6912788540515324e-06, "loss": 0.5168, "step": 22954 }, { "epoch": 0.8178206886723551, "grad_norm": 1.5022642039680387, "learning_rate": 1.6906367858200379e-06, "loss": 0.5371, "step": 22955 }, { "epoch": 0.8178563157988492, "grad_norm": 1.4051436604935479, "learning_rate": 1.6899948282335221e-06, "loss": 0.4094, "step": 22956 }, { "epoch": 0.8178919429253434, "grad_norm": 1.1975507593203114, "learning_rate": 1.6893529813005371e-06, "loss": 0.44, "step": 22957 }, { "epoch": 0.8179275700518375, "grad_norm": 1.5171218407065223, "learning_rate": 1.6887112450296227e-06, "loss": 0.504, "step": 22958 }, { "epoch": 0.8179631971783315, "grad_norm": 1.6368064454916975, "learning_rate": 1.68806961942933e-06, "loss": 0.5255, "step": 22959 }, { "epoch": 0.8179988243048257, "grad_norm": 1.6133024801075093, "learning_rate": 1.6874281045081976e-06, "loss": 0.5567, "step": 22960 }, { "epoch": 0.8180344514313198, "grad_norm": 1.4840123977959114, "learning_rate": 1.6867867002747696e-06, "loss": 0.5005, "step": 22961 }, { "epoch": 0.8180700785578139, "grad_norm": 1.2473190444766709, "learning_rate": 1.6861454067375882e-06, "loss": 0.4097, "step": 22962 }, { "epoch": 0.8181057056843081, "grad_norm": 1.2369625391718126, "learning_rate": 1.6855042239051934e-06, "loss": 0.3835, "step": 22963 }, { "epoch": 0.8181413328108021, "grad_norm": 1.4108306917373885, "learning_rate": 1.6848631517861192e-06, "loss": 0.5457, "step": 22964 }, { "epoch": 0.8181769599372962, "grad_norm": 1.1693407274311614, "learning_rate": 1.6842221903889066e-06, "loss": 0.3882, "step": 22965 }, { "epoch": 0.8182125870637904, "grad_norm": 1.6494612161626638, "learning_rate": 1.6835813397220857e-06, "loss": 0.4905, "step": 22966 }, { "epoch": 0.8182482141902845, "grad_norm": 1.161509991999899, "learning_rate": 1.6829405997941917e-06, "loss": 0.3798, "step": 22967 }, { "epoch": 0.8182838413167786, "grad_norm": 1.2646109710085793, "learning_rate": 1.6822999706137565e-06, "loss": 0.3871, "step": 22968 }, { "epoch": 0.8183194684432727, "grad_norm": 1.1587144535336806, "learning_rate": 1.6816594521893114e-06, "loss": 0.3902, "step": 22969 }, { "epoch": 0.8183550955697668, "grad_norm": 1.3974472948974666, "learning_rate": 1.6810190445293873e-06, "loss": 0.5252, "step": 22970 }, { "epoch": 0.8183907226962609, "grad_norm": 2.0833692762823564, "learning_rate": 1.680378747642507e-06, "loss": 0.5479, "step": 22971 }, { "epoch": 0.8184263498227551, "grad_norm": 1.2220205766994274, "learning_rate": 1.6797385615372008e-06, "loss": 0.388, "step": 22972 }, { "epoch": 0.8184619769492492, "grad_norm": 1.6524890773318475, "learning_rate": 1.6790984862219895e-06, "loss": 0.5001, "step": 22973 }, { "epoch": 0.8184976040757432, "grad_norm": 1.3595200593549386, "learning_rate": 1.6784585217053994e-06, "loss": 0.5279, "step": 22974 }, { "epoch": 0.8185332312022374, "grad_norm": 1.2365168399686666, "learning_rate": 1.6778186679959453e-06, "loss": 0.4252, "step": 22975 }, { "epoch": 0.8185688583287315, "grad_norm": 1.3098604895994617, "learning_rate": 1.6771789251021586e-06, "loss": 0.4197, "step": 22976 }, { "epoch": 0.8186044854552256, "grad_norm": 1.3861722313451408, "learning_rate": 1.6765392930325486e-06, "loss": 0.5015, "step": 22977 }, { "epoch": 0.8186401125817198, "grad_norm": 1.6008052976144838, "learning_rate": 1.6758997717956382e-06, "loss": 0.5583, "step": 22978 }, { "epoch": 0.8186757397082138, "grad_norm": 1.1980507392460042, "learning_rate": 1.6752603613999384e-06, "loss": 0.4019, "step": 22979 }, { "epoch": 0.8187113668347079, "grad_norm": 1.1858308370829354, "learning_rate": 1.674621061853965e-06, "loss": 0.3791, "step": 22980 }, { "epoch": 0.8187469939612021, "grad_norm": 1.1508552399973637, "learning_rate": 1.6739818731662338e-06, "loss": 0.4013, "step": 22981 }, { "epoch": 0.8187826210876962, "grad_norm": 1.1284175930178937, "learning_rate": 1.673342795345251e-06, "loss": 0.4163, "step": 22982 }, { "epoch": 0.8188182482141902, "grad_norm": 1.801653043534616, "learning_rate": 1.672703828399529e-06, "loss": 0.5039, "step": 22983 }, { "epoch": 0.8188538753406844, "grad_norm": 1.1862805126640672, "learning_rate": 1.672064972337576e-06, "loss": 0.3565, "step": 22984 }, { "epoch": 0.8188895024671785, "grad_norm": 1.7964880419204232, "learning_rate": 1.6714262271679016e-06, "loss": 0.5512, "step": 22985 }, { "epoch": 0.8189251295936726, "grad_norm": 1.1244811341481975, "learning_rate": 1.6707875928990059e-06, "loss": 0.3936, "step": 22986 }, { "epoch": 0.8189607567201668, "grad_norm": 1.4013841652872714, "learning_rate": 1.6701490695393973e-06, "loss": 0.4823, "step": 22987 }, { "epoch": 0.8189963838466608, "grad_norm": 1.6051901712685228, "learning_rate": 1.6695106570975728e-06, "loss": 0.5715, "step": 22988 }, { "epoch": 0.8190320109731549, "grad_norm": 1.6247915757737648, "learning_rate": 1.6688723555820374e-06, "loss": 0.4719, "step": 22989 }, { "epoch": 0.8190676380996491, "grad_norm": 1.700674948310583, "learning_rate": 1.6682341650012902e-06, "loss": 0.5079, "step": 22990 }, { "epoch": 0.8191032652261432, "grad_norm": 1.3888588415498782, "learning_rate": 1.6675960853638273e-06, "loss": 0.4561, "step": 22991 }, { "epoch": 0.8191388923526373, "grad_norm": 2.041680893591298, "learning_rate": 1.66695811667815e-06, "loss": 0.5175, "step": 22992 }, { "epoch": 0.8191745194791314, "grad_norm": 1.5448492646528775, "learning_rate": 1.6663202589527471e-06, "loss": 0.409, "step": 22993 }, { "epoch": 0.8192101466056255, "grad_norm": 1.4738470129547532, "learning_rate": 1.6656825121961172e-06, "loss": 0.5169, "step": 22994 }, { "epoch": 0.8192457737321196, "grad_norm": 1.2844635602396792, "learning_rate": 1.6650448764167481e-06, "loss": 0.4, "step": 22995 }, { "epoch": 0.8192814008586138, "grad_norm": 1.0386526851278504, "learning_rate": 1.664407351623134e-06, "loss": 0.4035, "step": 22996 }, { "epoch": 0.8193170279851079, "grad_norm": 1.1717249736483024, "learning_rate": 1.6637699378237605e-06, "loss": 0.3944, "step": 22997 }, { "epoch": 0.8193526551116019, "grad_norm": 1.225909303436093, "learning_rate": 1.663132635027116e-06, "loss": 0.366, "step": 22998 }, { "epoch": 0.8193882822380961, "grad_norm": 1.1649193317017417, "learning_rate": 1.6624954432416884e-06, "loss": 0.4106, "step": 22999 }, { "epoch": 0.8194239093645902, "grad_norm": 1.2838684967694025, "learning_rate": 1.6618583624759631e-06, "loss": 0.4421, "step": 23000 }, { "epoch": 0.8194595364910843, "grad_norm": 1.50838399497429, "learning_rate": 1.6612213927384202e-06, "loss": 0.4428, "step": 23001 }, { "epoch": 0.8194951636175785, "grad_norm": 1.4123154502302535, "learning_rate": 1.6605845340375427e-06, "loss": 0.4748, "step": 23002 }, { "epoch": 0.8195307907440725, "grad_norm": 2.1904768543104387, "learning_rate": 1.6599477863818125e-06, "loss": 0.5565, "step": 23003 }, { "epoch": 0.8195664178705666, "grad_norm": 1.7825375140510262, "learning_rate": 1.6593111497797055e-06, "loss": 0.4741, "step": 23004 }, { "epoch": 0.8196020449970608, "grad_norm": 1.5378485246601663, "learning_rate": 1.6586746242397001e-06, "loss": 0.5121, "step": 23005 }, { "epoch": 0.8196376721235549, "grad_norm": 1.7183439726584384, "learning_rate": 1.658038209770273e-06, "loss": 0.5423, "step": 23006 }, { "epoch": 0.819673299250049, "grad_norm": 1.222075698987307, "learning_rate": 1.6574019063798997e-06, "loss": 0.3802, "step": 23007 }, { "epoch": 0.8197089263765431, "grad_norm": 1.5672852608114136, "learning_rate": 1.6567657140770477e-06, "loss": 0.5656, "step": 23008 }, { "epoch": 0.8197445535030372, "grad_norm": 1.7451168747119812, "learning_rate": 1.656129632870196e-06, "loss": 0.5541, "step": 23009 }, { "epoch": 0.8197801806295313, "grad_norm": 1.176323008153411, "learning_rate": 1.6554936627678065e-06, "loss": 0.4012, "step": 23010 }, { "epoch": 0.8198158077560255, "grad_norm": 1.1265479645585461, "learning_rate": 1.6548578037783536e-06, "loss": 0.4184, "step": 23011 }, { "epoch": 0.8198514348825195, "grad_norm": 1.6894162991764432, "learning_rate": 1.6542220559102983e-06, "loss": 0.4947, "step": 23012 }, { "epoch": 0.8198870620090136, "grad_norm": 1.4340034330681781, "learning_rate": 1.653586419172114e-06, "loss": 0.5146, "step": 23013 }, { "epoch": 0.8199226891355078, "grad_norm": 1.177513253773017, "learning_rate": 1.652950893572257e-06, "loss": 0.3803, "step": 23014 }, { "epoch": 0.8199583162620019, "grad_norm": 1.5849662909944922, "learning_rate": 1.6523154791191942e-06, "loss": 0.5386, "step": 23015 }, { "epoch": 0.819993943388496, "grad_norm": 1.4669327615123322, "learning_rate": 1.6516801758213873e-06, "loss": 0.4503, "step": 23016 }, { "epoch": 0.8200295705149901, "grad_norm": 1.287947025321927, "learning_rate": 1.6510449836872922e-06, "loss": 0.451, "step": 23017 }, { "epoch": 0.8200651976414842, "grad_norm": 1.711008294464168, "learning_rate": 1.6504099027253707e-06, "loss": 0.4965, "step": 23018 }, { "epoch": 0.8201008247679783, "grad_norm": 1.5401306320684585, "learning_rate": 1.649774932944075e-06, "loss": 0.5559, "step": 23019 }, { "epoch": 0.8201364518944725, "grad_norm": 1.40081632035638, "learning_rate": 1.6491400743518627e-06, "loss": 0.5133, "step": 23020 }, { "epoch": 0.8201720790209666, "grad_norm": 1.4247609407911057, "learning_rate": 1.648505326957187e-06, "loss": 0.5153, "step": 23021 }, { "epoch": 0.8202077061474606, "grad_norm": 1.2431540439431001, "learning_rate": 1.6478706907685027e-06, "loss": 0.4129, "step": 23022 }, { "epoch": 0.8202433332739548, "grad_norm": 1.202092783338, "learning_rate": 1.647236165794256e-06, "loss": 0.3787, "step": 23023 }, { "epoch": 0.8202789604004489, "grad_norm": 1.199489574616011, "learning_rate": 1.6466017520429001e-06, "loss": 0.3956, "step": 23024 }, { "epoch": 0.820314587526943, "grad_norm": 1.7044373000230344, "learning_rate": 1.6459674495228772e-06, "loss": 0.5139, "step": 23025 }, { "epoch": 0.8203502146534372, "grad_norm": 1.3831039637066513, "learning_rate": 1.6453332582426384e-06, "loss": 0.3881, "step": 23026 }, { "epoch": 0.8203858417799312, "grad_norm": 1.5512252938590287, "learning_rate": 1.6446991782106258e-06, "loss": 0.5454, "step": 23027 }, { "epoch": 0.8204214689064253, "grad_norm": 1.282129576037768, "learning_rate": 1.6440652094352838e-06, "loss": 0.4255, "step": 23028 }, { "epoch": 0.8204570960329195, "grad_norm": 1.1410474684991587, "learning_rate": 1.6434313519250567e-06, "loss": 0.3777, "step": 23029 }, { "epoch": 0.8204927231594136, "grad_norm": 1.411449919699555, "learning_rate": 1.6427976056883787e-06, "loss": 0.4781, "step": 23030 }, { "epoch": 0.8205283502859076, "grad_norm": 1.4381705185013351, "learning_rate": 1.6421639707336956e-06, "loss": 0.4016, "step": 23031 }, { "epoch": 0.8205639774124018, "grad_norm": 1.2613336961312076, "learning_rate": 1.6415304470694382e-06, "loss": 0.4318, "step": 23032 }, { "epoch": 0.8205996045388959, "grad_norm": 1.2721023162066902, "learning_rate": 1.6408970347040475e-06, "loss": 0.4094, "step": 23033 }, { "epoch": 0.82063523166539, "grad_norm": 1.7293879985022942, "learning_rate": 1.6402637336459527e-06, "loss": 0.6156, "step": 23034 }, { "epoch": 0.8206708587918842, "grad_norm": 1.711877561995497, "learning_rate": 1.6396305439035898e-06, "loss": 0.4606, "step": 23035 }, { "epoch": 0.8207064859183782, "grad_norm": 1.1397942574561852, "learning_rate": 1.6389974654853892e-06, "loss": 0.36, "step": 23036 }, { "epoch": 0.8207421130448723, "grad_norm": 1.7388991205833955, "learning_rate": 1.6383644983997816e-06, "loss": 0.547, "step": 23037 }, { "epoch": 0.8207777401713665, "grad_norm": 1.7073548022474865, "learning_rate": 1.637731642655197e-06, "loss": 0.4874, "step": 23038 }, { "epoch": 0.8208133672978606, "grad_norm": 1.4241936774446047, "learning_rate": 1.6370988982600588e-06, "loss": 0.4593, "step": 23039 }, { "epoch": 0.8208489944243547, "grad_norm": 1.2147935184780183, "learning_rate": 1.636466265222797e-06, "loss": 0.4101, "step": 23040 }, { "epoch": 0.8208846215508488, "grad_norm": 5.0681510421184885, "learning_rate": 1.635833743551829e-06, "loss": 0.5564, "step": 23041 }, { "epoch": 0.8209202486773429, "grad_norm": 1.0937872640982516, "learning_rate": 1.6352013332555817e-06, "loss": 0.4282, "step": 23042 }, { "epoch": 0.820955875803837, "grad_norm": 1.2469187523688778, "learning_rate": 1.6345690343424758e-06, "loss": 0.4213, "step": 23043 }, { "epoch": 0.8209915029303312, "grad_norm": 1.3666381726769339, "learning_rate": 1.6339368468209316e-06, "loss": 0.4235, "step": 23044 }, { "epoch": 0.8210271300568253, "grad_norm": 1.5723456032729983, "learning_rate": 1.6333047706993632e-06, "loss": 0.5818, "step": 23045 }, { "epoch": 0.8210627571833193, "grad_norm": 1.1575608345477824, "learning_rate": 1.632672805986193e-06, "loss": 0.4482, "step": 23046 }, { "epoch": 0.8210983843098135, "grad_norm": 1.49111665120134, "learning_rate": 1.6320409526898307e-06, "loss": 0.4775, "step": 23047 }, { "epoch": 0.8211340114363076, "grad_norm": 1.1560666899372434, "learning_rate": 1.6314092108186907e-06, "loss": 0.3606, "step": 23048 }, { "epoch": 0.8211696385628017, "grad_norm": 1.4350266090080879, "learning_rate": 1.63077758038119e-06, "loss": 0.5172, "step": 23049 }, { "epoch": 0.8212052656892959, "grad_norm": 1.3077639899978286, "learning_rate": 1.63014606138573e-06, "loss": 0.4012, "step": 23050 }, { "epoch": 0.8212408928157899, "grad_norm": 1.0981901279749113, "learning_rate": 1.6295146538407303e-06, "loss": 0.3807, "step": 23051 }, { "epoch": 0.821276519942284, "grad_norm": 1.7345287964804517, "learning_rate": 1.6288833577545914e-06, "loss": 0.4758, "step": 23052 }, { "epoch": 0.8213121470687782, "grad_norm": 1.4942504091159907, "learning_rate": 1.6282521731357237e-06, "loss": 0.5838, "step": 23053 }, { "epoch": 0.8213477741952723, "grad_norm": 1.3264191106752214, "learning_rate": 1.6276210999925279e-06, "loss": 0.3821, "step": 23054 }, { "epoch": 0.8213834013217663, "grad_norm": 1.461157607701735, "learning_rate": 1.626990138333412e-06, "loss": 0.4835, "step": 23055 }, { "epoch": 0.8214190284482605, "grad_norm": 1.153155455449656, "learning_rate": 1.6263592881667712e-06, "loss": 0.3966, "step": 23056 }, { "epoch": 0.8214546555747546, "grad_norm": 1.3692722519101774, "learning_rate": 1.62572854950101e-06, "loss": 0.4582, "step": 23057 }, { "epoch": 0.8214902827012487, "grad_norm": 1.201524508005727, "learning_rate": 1.6250979223445262e-06, "loss": 0.3961, "step": 23058 }, { "epoch": 0.8215259098277429, "grad_norm": 2.013392582836533, "learning_rate": 1.6244674067057197e-06, "loss": 0.4941, "step": 23059 }, { "epoch": 0.821561536954237, "grad_norm": 1.5372912375989283, "learning_rate": 1.6238370025929807e-06, "loss": 0.5062, "step": 23060 }, { "epoch": 0.821597164080731, "grad_norm": 1.2954833938013885, "learning_rate": 1.6232067100147076e-06, "loss": 0.4112, "step": 23061 }, { "epoch": 0.8216327912072252, "grad_norm": 1.9792472595242565, "learning_rate": 1.6225765289792939e-06, "loss": 0.5022, "step": 23062 }, { "epoch": 0.8216684183337193, "grad_norm": 2.0161251266120654, "learning_rate": 1.6219464594951273e-06, "loss": 0.5605, "step": 23063 }, { "epoch": 0.8217040454602134, "grad_norm": 1.2587342031171735, "learning_rate": 1.6213165015706e-06, "loss": 0.4339, "step": 23064 }, { "epoch": 0.8217396725867075, "grad_norm": 1.6345885737441013, "learning_rate": 1.6206866552140988e-06, "loss": 0.4151, "step": 23065 }, { "epoch": 0.8217752997132016, "grad_norm": 1.1765328003068278, "learning_rate": 1.6200569204340143e-06, "loss": 0.3786, "step": 23066 }, { "epoch": 0.8218109268396957, "grad_norm": 1.6624990107625666, "learning_rate": 1.619427297238727e-06, "loss": 0.5053, "step": 23067 }, { "epoch": 0.8218465539661899, "grad_norm": 1.5126972535567416, "learning_rate": 1.6187977856366255e-06, "loss": 0.4899, "step": 23068 }, { "epoch": 0.821882181092684, "grad_norm": 1.073828777501228, "learning_rate": 1.6181683856360864e-06, "loss": 0.325, "step": 23069 }, { "epoch": 0.821917808219178, "grad_norm": 1.133048618868472, "learning_rate": 1.6175390972454962e-06, "loss": 0.4414, "step": 23070 }, { "epoch": 0.8219534353456722, "grad_norm": 1.692115091207222, "learning_rate": 1.6169099204732309e-06, "loss": 0.4878, "step": 23071 }, { "epoch": 0.8219890624721663, "grad_norm": 1.5110029260887474, "learning_rate": 1.6162808553276677e-06, "loss": 0.45, "step": 23072 }, { "epoch": 0.8220246895986604, "grad_norm": 1.1474398460460624, "learning_rate": 1.6156519018171856e-06, "loss": 0.3658, "step": 23073 }, { "epoch": 0.8220603167251546, "grad_norm": 1.1605552913861474, "learning_rate": 1.6150230599501593e-06, "loss": 0.4391, "step": 23074 }, { "epoch": 0.8220959438516486, "grad_norm": 1.705925226875111, "learning_rate": 1.6143943297349629e-06, "loss": 0.5749, "step": 23075 }, { "epoch": 0.8221315709781427, "grad_norm": 1.686658402160825, "learning_rate": 1.6137657111799654e-06, "loss": 0.4665, "step": 23076 }, { "epoch": 0.8221671981046369, "grad_norm": 1.2673844997400021, "learning_rate": 1.613137204293541e-06, "loss": 0.4484, "step": 23077 }, { "epoch": 0.822202825231131, "grad_norm": 1.299178064076469, "learning_rate": 1.6125088090840556e-06, "loss": 0.3903, "step": 23078 }, { "epoch": 0.822238452357625, "grad_norm": 1.9377854424636756, "learning_rate": 1.6118805255598769e-06, "loss": 0.4955, "step": 23079 }, { "epoch": 0.8222740794841192, "grad_norm": 1.286106702642408, "learning_rate": 1.6112523537293722e-06, "loss": 0.359, "step": 23080 }, { "epoch": 0.8223097066106133, "grad_norm": 1.4898283162075359, "learning_rate": 1.6106242936009087e-06, "loss": 0.5395, "step": 23081 }, { "epoch": 0.8223453337371074, "grad_norm": 2.1183436069751007, "learning_rate": 1.6099963451828427e-06, "loss": 0.4171, "step": 23082 }, { "epoch": 0.8223809608636016, "grad_norm": 1.6395643917220803, "learning_rate": 1.609368508483541e-06, "loss": 0.5021, "step": 23083 }, { "epoch": 0.8224165879900956, "grad_norm": 1.2753734384776765, "learning_rate": 1.6087407835113644e-06, "loss": 0.3742, "step": 23084 }, { "epoch": 0.8224522151165897, "grad_norm": 1.6004309369125855, "learning_rate": 1.6081131702746666e-06, "loss": 0.5427, "step": 23085 }, { "epoch": 0.8224878422430839, "grad_norm": 1.3511096343497078, "learning_rate": 1.6074856687818107e-06, "loss": 0.45, "step": 23086 }, { "epoch": 0.822523469369578, "grad_norm": 1.276401577548251, "learning_rate": 1.6068582790411435e-06, "loss": 0.4012, "step": 23087 }, { "epoch": 0.8225590964960721, "grad_norm": 1.1413905745859716, "learning_rate": 1.6062310010610294e-06, "loss": 0.3901, "step": 23088 }, { "epoch": 0.8225947236225662, "grad_norm": 1.283576730712524, "learning_rate": 1.6056038348498149e-06, "loss": 0.4015, "step": 23089 }, { "epoch": 0.8226303507490603, "grad_norm": 1.3063499576521096, "learning_rate": 1.6049767804158546e-06, "loss": 0.4307, "step": 23090 }, { "epoch": 0.8226659778755544, "grad_norm": 1.43082491595982, "learning_rate": 1.6043498377674937e-06, "loss": 0.5034, "step": 23091 }, { "epoch": 0.8227016050020486, "grad_norm": 1.778404527854873, "learning_rate": 1.603723006913086e-06, "loss": 0.5783, "step": 23092 }, { "epoch": 0.8227372321285427, "grad_norm": 1.320381698232315, "learning_rate": 1.6030962878609723e-06, "loss": 0.3864, "step": 23093 }, { "epoch": 0.8227728592550367, "grad_norm": 1.1981621378562959, "learning_rate": 1.6024696806195007e-06, "loss": 0.4104, "step": 23094 }, { "epoch": 0.8228084863815309, "grad_norm": 1.4062792933789716, "learning_rate": 1.601843185197014e-06, "loss": 0.5132, "step": 23095 }, { "epoch": 0.822844113508025, "grad_norm": 1.6418885254852889, "learning_rate": 1.601216801601857e-06, "loss": 0.5443, "step": 23096 }, { "epoch": 0.8228797406345191, "grad_norm": 1.441982146500118, "learning_rate": 1.6005905298423696e-06, "loss": 0.4826, "step": 23097 }, { "epoch": 0.8229153677610133, "grad_norm": 1.2571714245456784, "learning_rate": 1.5999643699268886e-06, "loss": 0.4239, "step": 23098 }, { "epoch": 0.8229509948875073, "grad_norm": 1.2290360987447113, "learning_rate": 1.5993383218637547e-06, "loss": 0.4405, "step": 23099 }, { "epoch": 0.8229866220140014, "grad_norm": 1.8052651820728067, "learning_rate": 1.5987123856613006e-06, "loss": 0.5949, "step": 23100 }, { "epoch": 0.8230222491404956, "grad_norm": 1.3419980837383654, "learning_rate": 1.598086561327864e-06, "loss": 0.3848, "step": 23101 }, { "epoch": 0.8230578762669897, "grad_norm": 1.1805287042579862, "learning_rate": 1.597460848871777e-06, "loss": 0.3768, "step": 23102 }, { "epoch": 0.8230935033934837, "grad_norm": 1.6242523561318125, "learning_rate": 1.5968352483013738e-06, "loss": 0.4956, "step": 23103 }, { "epoch": 0.8231291305199779, "grad_norm": 1.5421424057911062, "learning_rate": 1.5962097596249816e-06, "loss": 0.4127, "step": 23104 }, { "epoch": 0.823164757646472, "grad_norm": 1.7322005253947925, "learning_rate": 1.595584382850931e-06, "loss": 0.4597, "step": 23105 }, { "epoch": 0.8232003847729661, "grad_norm": 1.607334255980643, "learning_rate": 1.5949591179875479e-06, "loss": 0.5004, "step": 23106 }, { "epoch": 0.8232360118994603, "grad_norm": 1.4981577180647063, "learning_rate": 1.5943339650431578e-06, "loss": 0.4475, "step": 23107 }, { "epoch": 0.8232716390259543, "grad_norm": 1.405706298361662, "learning_rate": 1.5937089240260883e-06, "loss": 0.5003, "step": 23108 }, { "epoch": 0.8233072661524484, "grad_norm": 1.4254362412795967, "learning_rate": 1.5930839949446586e-06, "loss": 0.4681, "step": 23109 }, { "epoch": 0.8233428932789426, "grad_norm": 1.370576637058883, "learning_rate": 1.5924591778071917e-06, "loss": 0.4332, "step": 23110 }, { "epoch": 0.8233785204054367, "grad_norm": 2.5528129974864275, "learning_rate": 1.591834472622007e-06, "loss": 0.4832, "step": 23111 }, { "epoch": 0.8234141475319308, "grad_norm": 1.3609124102029786, "learning_rate": 1.5912098793974252e-06, "loss": 0.4392, "step": 23112 }, { "epoch": 0.823449774658425, "grad_norm": 1.3293903008422268, "learning_rate": 1.5905853981417597e-06, "loss": 0.4281, "step": 23113 }, { "epoch": 0.823485401784919, "grad_norm": 1.3014810895209972, "learning_rate": 1.5899610288633293e-06, "loss": 0.4226, "step": 23114 }, { "epoch": 0.8235210289114131, "grad_norm": 1.7524900402865218, "learning_rate": 1.5893367715704443e-06, "loss": 0.5014, "step": 23115 }, { "epoch": 0.8235566560379073, "grad_norm": 1.225229832463602, "learning_rate": 1.5887126262714181e-06, "loss": 0.4292, "step": 23116 }, { "epoch": 0.8235922831644014, "grad_norm": 1.3750058668430931, "learning_rate": 1.5880885929745627e-06, "loss": 0.4444, "step": 23117 }, { "epoch": 0.8236279102908954, "grad_norm": 1.2077707765429126, "learning_rate": 1.587464671688187e-06, "loss": 0.412, "step": 23118 }, { "epoch": 0.8236635374173896, "grad_norm": 1.3040320587746173, "learning_rate": 1.5868408624206022e-06, "loss": 0.4115, "step": 23119 }, { "epoch": 0.8236991645438837, "grad_norm": 1.2840664706783516, "learning_rate": 1.5862171651801096e-06, "loss": 0.4044, "step": 23120 }, { "epoch": 0.8237347916703778, "grad_norm": 1.542672441869673, "learning_rate": 1.5855935799750177e-06, "loss": 0.4889, "step": 23121 }, { "epoch": 0.823770418796872, "grad_norm": 1.5680932562887295, "learning_rate": 1.5849701068136269e-06, "loss": 0.445, "step": 23122 }, { "epoch": 0.823806045923366, "grad_norm": 1.1934277137136613, "learning_rate": 1.5843467457042427e-06, "loss": 0.4074, "step": 23123 }, { "epoch": 0.8238416730498601, "grad_norm": 1.2090384498918887, "learning_rate": 1.5837234966551619e-06, "loss": 0.3952, "step": 23124 }, { "epoch": 0.8238773001763543, "grad_norm": 1.1799732169337172, "learning_rate": 1.5831003596746853e-06, "loss": 0.4272, "step": 23125 }, { "epoch": 0.8239129273028484, "grad_norm": 2.6383793923028858, "learning_rate": 1.5824773347711098e-06, "loss": 0.4829, "step": 23126 }, { "epoch": 0.8239485544293425, "grad_norm": 1.1804646060695199, "learning_rate": 1.5818544219527354e-06, "loss": 0.4055, "step": 23127 }, { "epoch": 0.8239841815558366, "grad_norm": 2.23349228371035, "learning_rate": 1.581231621227851e-06, "loss": 0.4979, "step": 23128 }, { "epoch": 0.8240198086823307, "grad_norm": 2.1467132293902176, "learning_rate": 1.5806089326047514e-06, "loss": 0.5433, "step": 23129 }, { "epoch": 0.8240554358088248, "grad_norm": 1.216161342019031, "learning_rate": 1.5799863560917307e-06, "loss": 0.4285, "step": 23130 }, { "epoch": 0.824091062935319, "grad_norm": 1.2138394742646152, "learning_rate": 1.5793638916970756e-06, "loss": 0.418, "step": 23131 }, { "epoch": 0.824126690061813, "grad_norm": 1.2940232644548484, "learning_rate": 1.578741539429075e-06, "loss": 0.4115, "step": 23132 }, { "epoch": 0.8241623171883071, "grad_norm": 1.4050936585893665, "learning_rate": 1.578119299296017e-06, "loss": 0.4329, "step": 23133 }, { "epoch": 0.8241979443148013, "grad_norm": 1.5095290915237374, "learning_rate": 1.5774971713061905e-06, "loss": 0.4631, "step": 23134 }, { "epoch": 0.8242335714412954, "grad_norm": 1.616909443038284, "learning_rate": 1.5768751554678719e-06, "loss": 0.4662, "step": 23135 }, { "epoch": 0.8242691985677895, "grad_norm": 1.7285606230673345, "learning_rate": 1.5762532517893515e-06, "loss": 0.5561, "step": 23136 }, { "epoch": 0.8243048256942836, "grad_norm": 1.480108197768763, "learning_rate": 1.5756314602789046e-06, "loss": 0.3909, "step": 23137 }, { "epoch": 0.8243404528207777, "grad_norm": 1.2574965944850711, "learning_rate": 1.5750097809448162e-06, "loss": 0.4212, "step": 23138 }, { "epoch": 0.8243760799472718, "grad_norm": 1.8157767686147797, "learning_rate": 1.5743882137953549e-06, "loss": 0.4154, "step": 23139 }, { "epoch": 0.824411707073766, "grad_norm": 1.1905462696580222, "learning_rate": 1.5737667588388095e-06, "loss": 0.3803, "step": 23140 }, { "epoch": 0.8244473342002601, "grad_norm": 1.8681645942078315, "learning_rate": 1.573145416083447e-06, "loss": 0.5693, "step": 23141 }, { "epoch": 0.8244829613267541, "grad_norm": 1.192464151786402, "learning_rate": 1.572524185537543e-06, "loss": 0.3842, "step": 23142 }, { "epoch": 0.8245185884532483, "grad_norm": 1.2067190215380699, "learning_rate": 1.5719030672093717e-06, "loss": 0.4123, "step": 23143 }, { "epoch": 0.8245542155797424, "grad_norm": 1.6983867234450711, "learning_rate": 1.5712820611072e-06, "loss": 0.5961, "step": 23144 }, { "epoch": 0.8245898427062365, "grad_norm": 1.270338197955283, "learning_rate": 1.5706611672393013e-06, "loss": 0.428, "step": 23145 }, { "epoch": 0.8246254698327307, "grad_norm": 1.2929089846534478, "learning_rate": 1.570040385613939e-06, "loss": 0.4049, "step": 23146 }, { "epoch": 0.8246610969592247, "grad_norm": 1.1965049124173635, "learning_rate": 1.569419716239381e-06, "loss": 0.4061, "step": 23147 }, { "epoch": 0.8246967240857188, "grad_norm": 1.5708116253433895, "learning_rate": 1.5687991591238916e-06, "loss": 0.5109, "step": 23148 }, { "epoch": 0.824732351212213, "grad_norm": 1.8025570808999198, "learning_rate": 1.5681787142757375e-06, "loss": 0.4934, "step": 23149 }, { "epoch": 0.8247679783387071, "grad_norm": 1.2890530575901962, "learning_rate": 1.5675583817031747e-06, "loss": 0.424, "step": 23150 }, { "epoch": 0.8248036054652012, "grad_norm": 1.6906014520162784, "learning_rate": 1.5669381614144685e-06, "loss": 0.4789, "step": 23151 }, { "epoch": 0.8248392325916953, "grad_norm": 1.3397130597666982, "learning_rate": 1.5663180534178724e-06, "loss": 0.4162, "step": 23152 }, { "epoch": 0.8248748597181894, "grad_norm": 1.901411197517174, "learning_rate": 1.5656980577216464e-06, "loss": 0.4984, "step": 23153 }, { "epoch": 0.8249104868446835, "grad_norm": 1.1549532197193775, "learning_rate": 1.5650781743340449e-06, "loss": 0.4104, "step": 23154 }, { "epoch": 0.8249461139711777, "grad_norm": 1.4128387414848744, "learning_rate": 1.5644584032633248e-06, "loss": 0.4586, "step": 23155 }, { "epoch": 0.8249817410976718, "grad_norm": 1.9234664226314435, "learning_rate": 1.5638387445177384e-06, "loss": 0.5175, "step": 23156 }, { "epoch": 0.8250173682241658, "grad_norm": 1.994224008072765, "learning_rate": 1.5632191981055334e-06, "loss": 0.382, "step": 23157 }, { "epoch": 0.82505299535066, "grad_norm": 1.3073447565578653, "learning_rate": 1.5625997640349643e-06, "loss": 0.3979, "step": 23158 }, { "epoch": 0.8250886224771541, "grad_norm": 1.3164826257695545, "learning_rate": 1.5619804423142749e-06, "loss": 0.3836, "step": 23159 }, { "epoch": 0.8251242496036482, "grad_norm": 1.696269769654723, "learning_rate": 1.5613612329517158e-06, "loss": 0.4557, "step": 23160 }, { "epoch": 0.8251598767301423, "grad_norm": 1.2863085690690057, "learning_rate": 1.5607421359555275e-06, "loss": 0.3983, "step": 23161 }, { "epoch": 0.8251955038566364, "grad_norm": 1.5340119301306532, "learning_rate": 1.5601231513339566e-06, "loss": 0.5333, "step": 23162 }, { "epoch": 0.8252311309831305, "grad_norm": 1.0568128701213944, "learning_rate": 1.5595042790952442e-06, "loss": 0.3754, "step": 23163 }, { "epoch": 0.8252667581096247, "grad_norm": 1.3886229939468016, "learning_rate": 1.5588855192476327e-06, "loss": 0.4238, "step": 23164 }, { "epoch": 0.8253023852361188, "grad_norm": 1.4629029166466798, "learning_rate": 1.5582668717993632e-06, "loss": 0.4477, "step": 23165 }, { "epoch": 0.8253380123626128, "grad_norm": 1.5832582153163557, "learning_rate": 1.557648336758668e-06, "loss": 0.4534, "step": 23166 }, { "epoch": 0.825373639489107, "grad_norm": 1.2316263763870974, "learning_rate": 1.5570299141337886e-06, "loss": 0.4011, "step": 23167 }, { "epoch": 0.8254092666156011, "grad_norm": 1.2640137678418817, "learning_rate": 1.5564116039329546e-06, "loss": 0.4272, "step": 23168 }, { "epoch": 0.8254448937420952, "grad_norm": 1.5015920085228167, "learning_rate": 1.5557934061644008e-06, "loss": 0.4272, "step": 23169 }, { "epoch": 0.8254805208685894, "grad_norm": 2.0285340915135017, "learning_rate": 1.5551753208363618e-06, "loss": 0.4769, "step": 23170 }, { "epoch": 0.8255161479950834, "grad_norm": 1.2393521860609753, "learning_rate": 1.5545573479570675e-06, "loss": 0.3912, "step": 23171 }, { "epoch": 0.8255517751215776, "grad_norm": 1.3876535046888983, "learning_rate": 1.5539394875347425e-06, "loss": 0.4126, "step": 23172 }, { "epoch": 0.8255874022480717, "grad_norm": 1.4404917601085323, "learning_rate": 1.553321739577619e-06, "loss": 0.4964, "step": 23173 }, { "epoch": 0.8256230293745658, "grad_norm": 1.7278000996168963, "learning_rate": 1.5527041040939184e-06, "loss": 0.627, "step": 23174 }, { "epoch": 0.82565865650106, "grad_norm": 1.922944597336518, "learning_rate": 1.5520865810918673e-06, "loss": 0.514, "step": 23175 }, { "epoch": 0.825694283627554, "grad_norm": 1.6319857011727692, "learning_rate": 1.551469170579687e-06, "loss": 0.4181, "step": 23176 }, { "epoch": 0.8257299107540481, "grad_norm": 1.5784954126445938, "learning_rate": 1.5508518725656007e-06, "loss": 0.5029, "step": 23177 }, { "epoch": 0.8257655378805423, "grad_norm": 1.633643005957345, "learning_rate": 1.55023468705783e-06, "loss": 0.4266, "step": 23178 }, { "epoch": 0.8258011650070364, "grad_norm": 1.2480201318631163, "learning_rate": 1.5496176140645868e-06, "loss": 0.3785, "step": 23179 }, { "epoch": 0.8258367921335305, "grad_norm": 1.2941553169753712, "learning_rate": 1.549000653594095e-06, "loss": 0.4144, "step": 23180 }, { "epoch": 0.8258724192600246, "grad_norm": 1.3211947107319602, "learning_rate": 1.548383805654564e-06, "loss": 0.3761, "step": 23181 }, { "epoch": 0.8259080463865187, "grad_norm": 1.4624114411043017, "learning_rate": 1.5477670702542113e-06, "loss": 0.5365, "step": 23182 }, { "epoch": 0.8259436735130128, "grad_norm": 1.8578367449408706, "learning_rate": 1.5471504474012466e-06, "loss": 0.4723, "step": 23183 }, { "epoch": 0.825979300639507, "grad_norm": 1.2097066482217815, "learning_rate": 1.5465339371038813e-06, "loss": 0.3992, "step": 23184 }, { "epoch": 0.826014927766001, "grad_norm": 1.161582781331796, "learning_rate": 1.5459175393703263e-06, "loss": 0.3886, "step": 23185 }, { "epoch": 0.8260505548924951, "grad_norm": 1.4375859956238128, "learning_rate": 1.5453012542087898e-06, "loss": 0.5208, "step": 23186 }, { "epoch": 0.8260861820189893, "grad_norm": 1.605996858267609, "learning_rate": 1.544685081627474e-06, "loss": 0.443, "step": 23187 }, { "epoch": 0.8261218091454834, "grad_norm": 1.3274568852319448, "learning_rate": 1.5440690216345866e-06, "loss": 0.4087, "step": 23188 }, { "epoch": 0.8261574362719775, "grad_norm": 1.4046251500557314, "learning_rate": 1.5434530742383324e-06, "loss": 0.4255, "step": 23189 }, { "epoch": 0.8261930633984716, "grad_norm": 1.7959515705554618, "learning_rate": 1.5428372394469094e-06, "loss": 0.4748, "step": 23190 }, { "epoch": 0.8262286905249657, "grad_norm": 1.9576898702412935, "learning_rate": 1.5422215172685194e-06, "loss": 0.4781, "step": 23191 }, { "epoch": 0.8262643176514598, "grad_norm": 1.1668125379866092, "learning_rate": 1.5416059077113621e-06, "loss": 0.3643, "step": 23192 }, { "epoch": 0.826299944777954, "grad_norm": 1.8558894769119239, "learning_rate": 1.540990410783636e-06, "loss": 0.3824, "step": 23193 }, { "epoch": 0.8263355719044481, "grad_norm": 1.721479436699229, "learning_rate": 1.5403750264935336e-06, "loss": 0.4851, "step": 23194 }, { "epoch": 0.8263711990309421, "grad_norm": 1.4681724884688274, "learning_rate": 1.5397597548492515e-06, "loss": 0.3944, "step": 23195 }, { "epoch": 0.8264068261574363, "grad_norm": 1.6324859756625985, "learning_rate": 1.53914459585898e-06, "loss": 0.5021, "step": 23196 }, { "epoch": 0.8264424532839304, "grad_norm": 1.2837260001861197, "learning_rate": 1.5385295495309149e-06, "loss": 0.4644, "step": 23197 }, { "epoch": 0.8264780804104245, "grad_norm": 1.7330861271543272, "learning_rate": 1.5379146158732405e-06, "loss": 0.4027, "step": 23198 }, { "epoch": 0.8265137075369187, "grad_norm": 1.5501648414715468, "learning_rate": 1.537299794894147e-06, "loss": 0.3993, "step": 23199 }, { "epoch": 0.8265493346634127, "grad_norm": 1.3484663320103192, "learning_rate": 1.5366850866018224e-06, "loss": 0.4053, "step": 23200 }, { "epoch": 0.8265849617899068, "grad_norm": 1.3638305343484263, "learning_rate": 1.536070491004451e-06, "loss": 0.4009, "step": 23201 }, { "epoch": 0.826620588916401, "grad_norm": 1.1728779041117303, "learning_rate": 1.53545600811022e-06, "loss": 0.3821, "step": 23202 }, { "epoch": 0.8266562160428951, "grad_norm": 1.3199787170323678, "learning_rate": 1.5348416379273057e-06, "loss": 0.4288, "step": 23203 }, { "epoch": 0.8266918431693892, "grad_norm": 1.4214935561226536, "learning_rate": 1.5342273804638942e-06, "loss": 0.4161, "step": 23204 }, { "epoch": 0.8267274702958833, "grad_norm": 1.7524354783516887, "learning_rate": 1.5336132357281598e-06, "loss": 0.4906, "step": 23205 }, { "epoch": 0.8267630974223774, "grad_norm": 1.639411083599122, "learning_rate": 1.5329992037282827e-06, "loss": 0.4686, "step": 23206 }, { "epoch": 0.8267987245488715, "grad_norm": 1.4128695232800081, "learning_rate": 1.5323852844724395e-06, "loss": 0.4312, "step": 23207 }, { "epoch": 0.8268343516753657, "grad_norm": 1.5222717757427586, "learning_rate": 1.5317714779688076e-06, "loss": 0.5215, "step": 23208 }, { "epoch": 0.8268699788018598, "grad_norm": 1.4228202542251849, "learning_rate": 1.5311577842255543e-06, "loss": 0.5267, "step": 23209 }, { "epoch": 0.8269056059283538, "grad_norm": 1.7785635453931874, "learning_rate": 1.5305442032508555e-06, "loss": 0.5148, "step": 23210 }, { "epoch": 0.826941233054848, "grad_norm": 1.2066783187431258, "learning_rate": 1.5299307350528813e-06, "loss": 0.379, "step": 23211 }, { "epoch": 0.8269768601813421, "grad_norm": 1.3686827356106084, "learning_rate": 1.5293173796397987e-06, "loss": 0.4085, "step": 23212 }, { "epoch": 0.8270124873078362, "grad_norm": 1.8443394304057368, "learning_rate": 1.5287041370197775e-06, "loss": 0.5304, "step": 23213 }, { "epoch": 0.8270481144343304, "grad_norm": 1.3039228723602143, "learning_rate": 1.528091007200977e-06, "loss": 0.5143, "step": 23214 }, { "epoch": 0.8270837415608244, "grad_norm": 2.2018930782324, "learning_rate": 1.5274779901915715e-06, "loss": 0.5433, "step": 23215 }, { "epoch": 0.8271193686873185, "grad_norm": 1.6879194659709222, "learning_rate": 1.5268650859997147e-06, "loss": 0.4446, "step": 23216 }, { "epoch": 0.8271549958138127, "grad_norm": 1.2190571364793623, "learning_rate": 1.5262522946335756e-06, "loss": 0.4094, "step": 23217 }, { "epoch": 0.8271906229403068, "grad_norm": 1.3806630136965374, "learning_rate": 1.5256396161013077e-06, "loss": 0.4263, "step": 23218 }, { "epoch": 0.8272262500668008, "grad_norm": 1.365817765890752, "learning_rate": 1.5250270504110732e-06, "loss": 0.42, "step": 23219 }, { "epoch": 0.827261877193295, "grad_norm": 1.5064335180110224, "learning_rate": 1.5244145975710246e-06, "loss": 0.4005, "step": 23220 }, { "epoch": 0.8272975043197891, "grad_norm": 1.9033251252515047, "learning_rate": 1.5238022575893197e-06, "loss": 0.5097, "step": 23221 }, { "epoch": 0.8273331314462832, "grad_norm": 1.4978595731611137, "learning_rate": 1.523190030474112e-06, "loss": 0.4508, "step": 23222 }, { "epoch": 0.8273687585727774, "grad_norm": 1.673454896236792, "learning_rate": 1.5225779162335551e-06, "loss": 0.4183, "step": 23223 }, { "epoch": 0.8274043856992714, "grad_norm": 1.3595538935859481, "learning_rate": 1.5219659148758003e-06, "loss": 0.3814, "step": 23224 }, { "epoch": 0.8274400128257655, "grad_norm": 1.4833847623587992, "learning_rate": 1.5213540264089933e-06, "loss": 0.4073, "step": 23225 }, { "epoch": 0.8274756399522597, "grad_norm": 1.8498895730075047, "learning_rate": 1.520742250841285e-06, "loss": 0.514, "step": 23226 }, { "epoch": 0.8275112670787538, "grad_norm": 1.7507064183928733, "learning_rate": 1.520130588180818e-06, "loss": 0.4519, "step": 23227 }, { "epoch": 0.8275468942052479, "grad_norm": 1.615795628627866, "learning_rate": 1.5195190384357405e-06, "loss": 0.4644, "step": 23228 }, { "epoch": 0.827582521331742, "grad_norm": 3.21767845745457, "learning_rate": 1.5189076016141935e-06, "loss": 0.5319, "step": 23229 }, { "epoch": 0.8276181484582361, "grad_norm": 1.8124268083378419, "learning_rate": 1.5182962777243227e-06, "loss": 0.3883, "step": 23230 }, { "epoch": 0.8276537755847302, "grad_norm": 1.4734994697041388, "learning_rate": 1.5176850667742626e-06, "loss": 0.5666, "step": 23231 }, { "epoch": 0.8276894027112244, "grad_norm": 1.608340973329578, "learning_rate": 1.517073968772157e-06, "loss": 0.5656, "step": 23232 }, { "epoch": 0.8277250298377185, "grad_norm": 1.7912104495112338, "learning_rate": 1.516462983726138e-06, "loss": 0.5665, "step": 23233 }, { "epoch": 0.8277606569642125, "grad_norm": 1.6370024981097302, "learning_rate": 1.5158521116443447e-06, "loss": 0.3892, "step": 23234 }, { "epoch": 0.8277962840907067, "grad_norm": 1.5031143537539928, "learning_rate": 1.515241352534913e-06, "loss": 0.4285, "step": 23235 }, { "epoch": 0.8278319112172008, "grad_norm": 1.444424054558984, "learning_rate": 1.5146307064059696e-06, "loss": 0.4487, "step": 23236 }, { "epoch": 0.8278675383436949, "grad_norm": 1.5533740064202883, "learning_rate": 1.5140201732656501e-06, "loss": 0.4095, "step": 23237 }, { "epoch": 0.827903165470189, "grad_norm": 1.5513776518783553, "learning_rate": 1.5134097531220838e-06, "loss": 0.4368, "step": 23238 }, { "epoch": 0.8279387925966831, "grad_norm": 1.7398681342180355, "learning_rate": 1.5127994459834005e-06, "loss": 0.4466, "step": 23239 }, { "epoch": 0.8279744197231772, "grad_norm": 1.8475736112391479, "learning_rate": 1.5121892518577218e-06, "loss": 0.5606, "step": 23240 }, { "epoch": 0.8280100468496714, "grad_norm": 1.407838720825534, "learning_rate": 1.511579170753179e-06, "loss": 0.3989, "step": 23241 }, { "epoch": 0.8280456739761655, "grad_norm": 1.547021362723336, "learning_rate": 1.5109692026778899e-06, "loss": 0.4499, "step": 23242 }, { "epoch": 0.8280813011026595, "grad_norm": 1.6751033530052089, "learning_rate": 1.5103593476399791e-06, "loss": 0.4084, "step": 23243 }, { "epoch": 0.8281169282291537, "grad_norm": 1.4204235754610208, "learning_rate": 1.509749605647568e-06, "loss": 0.4068, "step": 23244 }, { "epoch": 0.8281525553556478, "grad_norm": 1.452227105121711, "learning_rate": 1.5091399767087743e-06, "loss": 0.4502, "step": 23245 }, { "epoch": 0.8281881824821419, "grad_norm": 1.6673164245787413, "learning_rate": 1.5085304608317196e-06, "loss": 0.4611, "step": 23246 }, { "epoch": 0.8282238096086361, "grad_norm": 1.6815350464825969, "learning_rate": 1.5079210580245152e-06, "loss": 0.5258, "step": 23247 }, { "epoch": 0.8282594367351301, "grad_norm": 1.4394337898402665, "learning_rate": 1.5073117682952787e-06, "loss": 0.386, "step": 23248 }, { "epoch": 0.8282950638616242, "grad_norm": 1.5675578077958305, "learning_rate": 1.5067025916521206e-06, "loss": 0.4784, "step": 23249 }, { "epoch": 0.8283306909881184, "grad_norm": 1.4018331131364876, "learning_rate": 1.5060935281031564e-06, "loss": 0.3948, "step": 23250 }, { "epoch": 0.8283663181146125, "grad_norm": 2.2522768056841005, "learning_rate": 1.5054845776564885e-06, "loss": 0.4607, "step": 23251 }, { "epoch": 0.8284019452411066, "grad_norm": 1.4444481531250555, "learning_rate": 1.5048757403202353e-06, "loss": 0.352, "step": 23252 }, { "epoch": 0.8284375723676007, "grad_norm": 1.520996468007606, "learning_rate": 1.5042670161024975e-06, "loss": 0.4925, "step": 23253 }, { "epoch": 0.8284731994940948, "grad_norm": 1.868179175848776, "learning_rate": 1.5036584050113844e-06, "loss": 0.4589, "step": 23254 }, { "epoch": 0.8285088266205889, "grad_norm": 1.871670388011835, "learning_rate": 1.5030499070549965e-06, "loss": 0.4795, "step": 23255 }, { "epoch": 0.8285444537470831, "grad_norm": 1.8049181733935986, "learning_rate": 1.5024415222414369e-06, "loss": 0.528, "step": 23256 }, { "epoch": 0.8285800808735772, "grad_norm": 1.837007798272455, "learning_rate": 1.5018332505788114e-06, "loss": 0.5171, "step": 23257 }, { "epoch": 0.8286157080000712, "grad_norm": 1.497323905093679, "learning_rate": 1.5012250920752137e-06, "loss": 0.4836, "step": 23258 }, { "epoch": 0.8286513351265654, "grad_norm": 1.5154614813576164, "learning_rate": 1.5006170467387437e-06, "loss": 0.4394, "step": 23259 }, { "epoch": 0.8286869622530595, "grad_norm": 2.293121057023345, "learning_rate": 1.5000091145774987e-06, "loss": 0.5308, "step": 23260 }, { "epoch": 0.8287225893795536, "grad_norm": 1.2909443245234697, "learning_rate": 1.4994012955995752e-06, "loss": 0.4511, "step": 23261 }, { "epoch": 0.8287582165060478, "grad_norm": 1.3210041945166993, "learning_rate": 1.4987935898130634e-06, "loss": 0.3941, "step": 23262 }, { "epoch": 0.8287938436325418, "grad_norm": 1.518332683467589, "learning_rate": 1.4981859972260594e-06, "loss": 0.4132, "step": 23263 }, { "epoch": 0.8288294707590359, "grad_norm": 2.017258208397614, "learning_rate": 1.4975785178466484e-06, "loss": 0.5209, "step": 23264 }, { "epoch": 0.8288650978855301, "grad_norm": 2.008606790893019, "learning_rate": 1.496971151682922e-06, "loss": 0.4494, "step": 23265 }, { "epoch": 0.8289007250120242, "grad_norm": 1.8078949180277248, "learning_rate": 1.4963638987429685e-06, "loss": 0.517, "step": 23266 }, { "epoch": 0.8289363521385182, "grad_norm": 1.254179552722992, "learning_rate": 1.495756759034873e-06, "loss": 0.4036, "step": 23267 }, { "epoch": 0.8289719792650124, "grad_norm": 1.3910580775497559, "learning_rate": 1.495149732566723e-06, "loss": 0.4274, "step": 23268 }, { "epoch": 0.8290076063915065, "grad_norm": 1.379862039371098, "learning_rate": 1.4945428193465972e-06, "loss": 0.4035, "step": 23269 }, { "epoch": 0.8290432335180006, "grad_norm": 2.3295107303493374, "learning_rate": 1.4939360193825803e-06, "loss": 0.5179, "step": 23270 }, { "epoch": 0.8290788606444948, "grad_norm": 1.3897819995572744, "learning_rate": 1.493329332682748e-06, "loss": 0.373, "step": 23271 }, { "epoch": 0.8291144877709888, "grad_norm": 1.386545521574309, "learning_rate": 1.4927227592551841e-06, "loss": 0.3596, "step": 23272 }, { "epoch": 0.8291501148974829, "grad_norm": 1.8871651250710597, "learning_rate": 1.4921162991079607e-06, "loss": 0.3861, "step": 23273 }, { "epoch": 0.8291857420239771, "grad_norm": 1.3879532850748693, "learning_rate": 1.4915099522491561e-06, "loss": 0.4205, "step": 23274 }, { "epoch": 0.8292213691504712, "grad_norm": 1.7415501554471953, "learning_rate": 1.4909037186868424e-06, "loss": 0.3802, "step": 23275 }, { "epoch": 0.8292569962769653, "grad_norm": 1.6092846557104579, "learning_rate": 1.4902975984290958e-06, "loss": 0.51, "step": 23276 }, { "epoch": 0.8292926234034594, "grad_norm": 1.9502200969191457, "learning_rate": 1.4896915914839828e-06, "loss": 0.532, "step": 23277 }, { "epoch": 0.8293282505299535, "grad_norm": 1.3847115038653317, "learning_rate": 1.4890856978595768e-06, "loss": 0.4527, "step": 23278 }, { "epoch": 0.8293638776564476, "grad_norm": 1.435424536371749, "learning_rate": 1.4884799175639409e-06, "loss": 0.4569, "step": 23279 }, { "epoch": 0.8293995047829418, "grad_norm": 1.6433761371534268, "learning_rate": 1.4878742506051436e-06, "loss": 0.4346, "step": 23280 }, { "epoch": 0.8294351319094359, "grad_norm": 1.356580472480095, "learning_rate": 1.4872686969912508e-06, "loss": 0.4308, "step": 23281 }, { "epoch": 0.8294707590359299, "grad_norm": 1.50504108518196, "learning_rate": 1.4866632567303252e-06, "loss": 0.4215, "step": 23282 }, { "epoch": 0.8295063861624241, "grad_norm": 1.3000962323465581, "learning_rate": 1.4860579298304311e-06, "loss": 0.4571, "step": 23283 }, { "epoch": 0.8295420132889182, "grad_norm": 1.5688097345405665, "learning_rate": 1.4854527162996246e-06, "loss": 0.5394, "step": 23284 }, { "epoch": 0.8295776404154123, "grad_norm": 1.6203213786406447, "learning_rate": 1.484847616145969e-06, "loss": 0.5846, "step": 23285 }, { "epoch": 0.8296132675419065, "grad_norm": 1.6388096034047144, "learning_rate": 1.484242629377517e-06, "loss": 0.3715, "step": 23286 }, { "epoch": 0.8296488946684005, "grad_norm": 1.670382057525798, "learning_rate": 1.4836377560023275e-06, "loss": 0.5294, "step": 23287 }, { "epoch": 0.8296845217948946, "grad_norm": 1.2386948811689908, "learning_rate": 1.4830329960284528e-06, "loss": 0.4171, "step": 23288 }, { "epoch": 0.8297201489213888, "grad_norm": 1.7797319249707497, "learning_rate": 1.4824283494639458e-06, "loss": 0.4367, "step": 23289 }, { "epoch": 0.8297557760478829, "grad_norm": 1.37836774937039, "learning_rate": 1.4818238163168597e-06, "loss": 0.5134, "step": 23290 }, { "epoch": 0.8297914031743769, "grad_norm": 1.545092036076703, "learning_rate": 1.4812193965952426e-06, "loss": 0.539, "step": 23291 }, { "epoch": 0.8298270303008711, "grad_norm": 1.6961134196509988, "learning_rate": 1.480615090307146e-06, "loss": 0.4772, "step": 23292 }, { "epoch": 0.8298626574273652, "grad_norm": 1.6895975849842382, "learning_rate": 1.480010897460612e-06, "loss": 0.4983, "step": 23293 }, { "epoch": 0.8298982845538593, "grad_norm": 1.6476186188565962, "learning_rate": 1.479406818063691e-06, "loss": 0.4848, "step": 23294 }, { "epoch": 0.8299339116803535, "grad_norm": 1.1988354744027592, "learning_rate": 1.4788028521244201e-06, "loss": 0.4106, "step": 23295 }, { "epoch": 0.8299695388068475, "grad_norm": 1.6507909282978812, "learning_rate": 1.478198999650846e-06, "loss": 0.454, "step": 23296 }, { "epoch": 0.8300051659333416, "grad_norm": 1.6037990722291333, "learning_rate": 1.477595260651008e-06, "loss": 0.4857, "step": 23297 }, { "epoch": 0.8300407930598358, "grad_norm": 1.5913681765779302, "learning_rate": 1.4769916351329495e-06, "loss": 0.3865, "step": 23298 }, { "epoch": 0.8300764201863299, "grad_norm": 1.9477834734688726, "learning_rate": 1.4763881231047007e-06, "loss": 0.4785, "step": 23299 }, { "epoch": 0.830112047312824, "grad_norm": 1.6246990726904207, "learning_rate": 1.475784724574305e-06, "loss": 0.5174, "step": 23300 }, { "epoch": 0.8301476744393181, "grad_norm": 1.6204918636681735, "learning_rate": 1.475181439549791e-06, "loss": 0.4387, "step": 23301 }, { "epoch": 0.8301833015658122, "grad_norm": 1.3356179050915924, "learning_rate": 1.4745782680391952e-06, "loss": 0.3865, "step": 23302 }, { "epoch": 0.8302189286923063, "grad_norm": 1.5073930888991942, "learning_rate": 1.4739752100505478e-06, "loss": 0.4028, "step": 23303 }, { "epoch": 0.8302545558188005, "grad_norm": 1.5518087690225246, "learning_rate": 1.4733722655918804e-06, "loss": 0.5037, "step": 23304 }, { "epoch": 0.8302901829452946, "grad_norm": 1.282194450015806, "learning_rate": 1.472769434671224e-06, "loss": 0.3802, "step": 23305 }, { "epoch": 0.8303258100717886, "grad_norm": 1.2817612105648553, "learning_rate": 1.4721667172965993e-06, "loss": 0.4001, "step": 23306 }, { "epoch": 0.8303614371982828, "grad_norm": 1.2628783985486327, "learning_rate": 1.4715641134760384e-06, "loss": 0.3973, "step": 23307 }, { "epoch": 0.8303970643247769, "grad_norm": 1.6538476365919448, "learning_rate": 1.4709616232175593e-06, "loss": 0.4008, "step": 23308 }, { "epoch": 0.830432691451271, "grad_norm": 1.5779342795841897, "learning_rate": 1.4703592465291905e-06, "loss": 0.4142, "step": 23309 }, { "epoch": 0.8304683185777652, "grad_norm": 1.479363313600988, "learning_rate": 1.4697569834189474e-06, "loss": 0.4799, "step": 23310 }, { "epoch": 0.8305039457042592, "grad_norm": 1.5971492052084255, "learning_rate": 1.4691548338948537e-06, "loss": 0.603, "step": 23311 }, { "epoch": 0.8305395728307533, "grad_norm": 1.2530467154145843, "learning_rate": 1.4685527979649238e-06, "loss": 0.4378, "step": 23312 }, { "epoch": 0.8305751999572475, "grad_norm": 1.6150523742083138, "learning_rate": 1.4679508756371807e-06, "loss": 0.4752, "step": 23313 }, { "epoch": 0.8306108270837416, "grad_norm": 1.3004990519491093, "learning_rate": 1.4673490669196322e-06, "loss": 0.3673, "step": 23314 }, { "epoch": 0.8306464542102356, "grad_norm": 2.5181248398974367, "learning_rate": 1.4667473718202952e-06, "loss": 0.4783, "step": 23315 }, { "epoch": 0.8306820813367298, "grad_norm": 1.5009158415643524, "learning_rate": 1.4661457903471833e-06, "loss": 0.4395, "step": 23316 }, { "epoch": 0.8307177084632239, "grad_norm": 1.2185780197116702, "learning_rate": 1.4655443225083032e-06, "loss": 0.3561, "step": 23317 }, { "epoch": 0.830753335589718, "grad_norm": 1.5541976057071696, "learning_rate": 1.4649429683116645e-06, "loss": 0.4724, "step": 23318 }, { "epoch": 0.8307889627162122, "grad_norm": 1.4891908259526918, "learning_rate": 1.464341727765277e-06, "loss": 0.4822, "step": 23319 }, { "epoch": 0.8308245898427062, "grad_norm": 1.7004272110058227, "learning_rate": 1.463740600877147e-06, "loss": 0.4617, "step": 23320 }, { "epoch": 0.8308602169692003, "grad_norm": 1.5885444695773392, "learning_rate": 1.4631395876552756e-06, "loss": 0.5784, "step": 23321 }, { "epoch": 0.8308958440956945, "grad_norm": 1.4698221062637313, "learning_rate": 1.4625386881076697e-06, "loss": 0.5218, "step": 23322 }, { "epoch": 0.8309314712221886, "grad_norm": 1.6402292214276, "learning_rate": 1.4619379022423264e-06, "loss": 0.475, "step": 23323 }, { "epoch": 0.8309670983486827, "grad_norm": 2.8248478693629337, "learning_rate": 1.4613372300672502e-06, "loss": 0.4966, "step": 23324 }, { "epoch": 0.8310027254751768, "grad_norm": 1.307966325065747, "learning_rate": 1.4607366715904336e-06, "loss": 0.4195, "step": 23325 }, { "epoch": 0.8310383526016709, "grad_norm": 1.1375950767236342, "learning_rate": 1.4601362268198749e-06, "loss": 0.3892, "step": 23326 }, { "epoch": 0.831073979728165, "grad_norm": 1.1695513025111604, "learning_rate": 1.4595358957635763e-06, "loss": 0.3924, "step": 23327 }, { "epoch": 0.8311096068546592, "grad_norm": 1.3099803823557523, "learning_rate": 1.458935678429525e-06, "loss": 0.4304, "step": 23328 }, { "epoch": 0.8311452339811533, "grad_norm": 1.6727005011766154, "learning_rate": 1.4583355748257156e-06, "loss": 0.5445, "step": 23329 }, { "epoch": 0.8311808611076473, "grad_norm": 2.1049666272957497, "learning_rate": 1.4577355849601372e-06, "loss": 0.4863, "step": 23330 }, { "epoch": 0.8312164882341415, "grad_norm": 2.3769465154253644, "learning_rate": 1.4571357088407822e-06, "loss": 0.4641, "step": 23331 }, { "epoch": 0.8312521153606356, "grad_norm": 1.9705870834341361, "learning_rate": 1.4565359464756345e-06, "loss": 0.5729, "step": 23332 }, { "epoch": 0.8312877424871297, "grad_norm": 1.3295710085621735, "learning_rate": 1.455936297872681e-06, "loss": 0.403, "step": 23333 }, { "epoch": 0.8313233696136239, "grad_norm": 1.6198531112613677, "learning_rate": 1.4553367630399085e-06, "loss": 0.4885, "step": 23334 }, { "epoch": 0.8313589967401179, "grad_norm": 1.655487807166873, "learning_rate": 1.4547373419853016e-06, "loss": 0.5115, "step": 23335 }, { "epoch": 0.831394623866612, "grad_norm": 1.7371957823650042, "learning_rate": 1.4541380347168377e-06, "loss": 0.5502, "step": 23336 }, { "epoch": 0.8314302509931062, "grad_norm": 1.6916286573247357, "learning_rate": 1.4535388412424989e-06, "loss": 0.5524, "step": 23337 }, { "epoch": 0.8314658781196003, "grad_norm": 1.6742167320218901, "learning_rate": 1.4529397615702656e-06, "loss": 0.5374, "step": 23338 }, { "epoch": 0.8315015052460943, "grad_norm": 1.6379598271451672, "learning_rate": 1.4523407957081114e-06, "loss": 0.5519, "step": 23339 }, { "epoch": 0.8315371323725885, "grad_norm": 1.7692614297317615, "learning_rate": 1.4517419436640145e-06, "loss": 0.5307, "step": 23340 }, { "epoch": 0.8315727594990826, "grad_norm": 1.4181274050525503, "learning_rate": 1.4511432054459485e-06, "loss": 0.4135, "step": 23341 }, { "epoch": 0.8316083866255767, "grad_norm": 1.371769346522623, "learning_rate": 1.4505445810618878e-06, "loss": 0.416, "step": 23342 }, { "epoch": 0.8316440137520709, "grad_norm": 1.2357023637762479, "learning_rate": 1.4499460705198e-06, "loss": 0.4126, "step": 23343 }, { "epoch": 0.8316796408785649, "grad_norm": 1.653428438221411, "learning_rate": 1.449347673827659e-06, "loss": 0.5725, "step": 23344 }, { "epoch": 0.831715268005059, "grad_norm": 1.3695025501467, "learning_rate": 1.448749390993427e-06, "loss": 0.4067, "step": 23345 }, { "epoch": 0.8317508951315532, "grad_norm": 1.4224114705893847, "learning_rate": 1.448151222025076e-06, "loss": 0.5191, "step": 23346 }, { "epoch": 0.8317865222580473, "grad_norm": 1.140934011470266, "learning_rate": 1.4475531669305676e-06, "loss": 0.3786, "step": 23347 }, { "epoch": 0.8318221493845414, "grad_norm": 1.191689673842899, "learning_rate": 1.4469552257178655e-06, "loss": 0.3967, "step": 23348 }, { "epoch": 0.8318577765110355, "grad_norm": 1.5550236463679648, "learning_rate": 1.446357398394934e-06, "loss": 0.4264, "step": 23349 }, { "epoch": 0.8318934036375296, "grad_norm": 1.6327436339870134, "learning_rate": 1.4457596849697319e-06, "loss": 0.5123, "step": 23350 }, { "epoch": 0.8319290307640237, "grad_norm": 1.2086757477621144, "learning_rate": 1.4451620854502202e-06, "loss": 0.3994, "step": 23351 }, { "epoch": 0.8319646578905179, "grad_norm": 1.661872773282797, "learning_rate": 1.4445645998443537e-06, "loss": 0.5497, "step": 23352 }, { "epoch": 0.832000285017012, "grad_norm": 1.2943513595169975, "learning_rate": 1.4439672281600903e-06, "loss": 0.4414, "step": 23353 }, { "epoch": 0.832035912143506, "grad_norm": 1.5707063112513893, "learning_rate": 1.4433699704053827e-06, "loss": 0.4461, "step": 23354 }, { "epoch": 0.8320715392700002, "grad_norm": 1.2809542268159426, "learning_rate": 1.4427728265881836e-06, "loss": 0.3977, "step": 23355 }, { "epoch": 0.8321071663964943, "grad_norm": 1.5690248718567907, "learning_rate": 1.4421757967164463e-06, "loss": 0.4237, "step": 23356 }, { "epoch": 0.8321427935229884, "grad_norm": 1.3079510803065895, "learning_rate": 1.4415788807981212e-06, "loss": 0.3522, "step": 23357 }, { "epoch": 0.8321784206494826, "grad_norm": 1.9026810395182363, "learning_rate": 1.4409820788411544e-06, "loss": 0.5144, "step": 23358 }, { "epoch": 0.8322140477759766, "grad_norm": 1.3951231411572662, "learning_rate": 1.440385390853495e-06, "loss": 0.4431, "step": 23359 }, { "epoch": 0.8322496749024707, "grad_norm": 1.4148920188439507, "learning_rate": 1.4397888168430841e-06, "loss": 0.4175, "step": 23360 }, { "epoch": 0.8322853020289649, "grad_norm": 1.536787608783426, "learning_rate": 1.439192356817869e-06, "loss": 0.3437, "step": 23361 }, { "epoch": 0.832320929155459, "grad_norm": 1.2857586103787242, "learning_rate": 1.4385960107857933e-06, "loss": 0.4173, "step": 23362 }, { "epoch": 0.832356556281953, "grad_norm": 1.351496949294432, "learning_rate": 1.4379997787547939e-06, "loss": 0.3766, "step": 23363 }, { "epoch": 0.8323921834084472, "grad_norm": 1.6122902770747842, "learning_rate": 1.4374036607328113e-06, "loss": 0.601, "step": 23364 }, { "epoch": 0.8324278105349413, "grad_norm": 1.5640581910318787, "learning_rate": 1.4368076567277845e-06, "loss": 0.3834, "step": 23365 }, { "epoch": 0.8324634376614354, "grad_norm": 3.644129589404706, "learning_rate": 1.4362117667476517e-06, "loss": 0.5485, "step": 23366 }, { "epoch": 0.8324990647879296, "grad_norm": 1.3731465018728608, "learning_rate": 1.435615990800342e-06, "loss": 0.391, "step": 23367 }, { "epoch": 0.8325346919144236, "grad_norm": 1.4110125658060018, "learning_rate": 1.4350203288937937e-06, "loss": 0.4328, "step": 23368 }, { "epoch": 0.8325703190409177, "grad_norm": 1.2575742643884014, "learning_rate": 1.4344247810359347e-06, "loss": 0.4279, "step": 23369 }, { "epoch": 0.8326059461674119, "grad_norm": 1.6620621868348324, "learning_rate": 1.4338293472346954e-06, "loss": 0.4844, "step": 23370 }, { "epoch": 0.832641573293906, "grad_norm": 1.9611434464659208, "learning_rate": 1.4332340274980073e-06, "loss": 0.5827, "step": 23371 }, { "epoch": 0.8326772004204, "grad_norm": 1.2177958515830798, "learning_rate": 1.432638821833795e-06, "loss": 0.3733, "step": 23372 }, { "epoch": 0.8327128275468942, "grad_norm": 3.0604743865279196, "learning_rate": 1.4320437302499878e-06, "loss": 0.5504, "step": 23373 }, { "epoch": 0.8327484546733883, "grad_norm": 1.4620000657712493, "learning_rate": 1.431448752754504e-06, "loss": 0.4811, "step": 23374 }, { "epoch": 0.8327840817998824, "grad_norm": 1.8022649425599626, "learning_rate": 1.4308538893552715e-06, "loss": 0.5273, "step": 23375 }, { "epoch": 0.8328197089263766, "grad_norm": 1.416550995434569, "learning_rate": 1.4302591400602072e-06, "loss": 0.4945, "step": 23376 }, { "epoch": 0.8328553360528707, "grad_norm": 2.0469289618569606, "learning_rate": 1.429664504877234e-06, "loss": 0.4444, "step": 23377 }, { "epoch": 0.8328909631793647, "grad_norm": 1.4601898693100284, "learning_rate": 1.429069983814264e-06, "loss": 0.5329, "step": 23378 }, { "epoch": 0.8329265903058589, "grad_norm": 1.764769464309435, "learning_rate": 1.4284755768792224e-06, "loss": 0.5588, "step": 23379 }, { "epoch": 0.832962217432353, "grad_norm": 1.572996998648309, "learning_rate": 1.4278812840800184e-06, "loss": 0.394, "step": 23380 }, { "epoch": 0.8329978445588471, "grad_norm": 1.7821694211336607, "learning_rate": 1.4272871054245674e-06, "loss": 0.4752, "step": 23381 }, { "epoch": 0.8330334716853413, "grad_norm": 1.3161485466601677, "learning_rate": 1.426693040920779e-06, "loss": 0.406, "step": 23382 }, { "epoch": 0.8330690988118353, "grad_norm": 1.2353970860432486, "learning_rate": 1.4260990905765648e-06, "loss": 0.4487, "step": 23383 }, { "epoch": 0.8331047259383294, "grad_norm": 1.6798994123661175, "learning_rate": 1.4255052543998372e-06, "loss": 0.5468, "step": 23384 }, { "epoch": 0.8331403530648236, "grad_norm": 1.3492097687650049, "learning_rate": 1.4249115323984975e-06, "loss": 0.4218, "step": 23385 }, { "epoch": 0.8331759801913177, "grad_norm": 1.6583368048095617, "learning_rate": 1.424317924580454e-06, "loss": 0.5378, "step": 23386 }, { "epoch": 0.8332116073178117, "grad_norm": 1.6188517055819134, "learning_rate": 1.4237244309536114e-06, "loss": 0.5188, "step": 23387 }, { "epoch": 0.8332472344443059, "grad_norm": 1.3821002328599312, "learning_rate": 1.4231310515258745e-06, "loss": 0.4643, "step": 23388 }, { "epoch": 0.8332828615708, "grad_norm": 1.585891421493645, "learning_rate": 1.4225377863051415e-06, "loss": 0.5026, "step": 23389 }, { "epoch": 0.8333184886972941, "grad_norm": 2.462517456270899, "learning_rate": 1.4219446352993138e-06, "loss": 0.6189, "step": 23390 }, { "epoch": 0.8333541158237883, "grad_norm": 1.8672848933413113, "learning_rate": 1.4213515985162874e-06, "loss": 0.4617, "step": 23391 }, { "epoch": 0.8333897429502823, "grad_norm": 1.4660655522481536, "learning_rate": 1.4207586759639613e-06, "loss": 0.3423, "step": 23392 }, { "epoch": 0.8334253700767764, "grad_norm": 1.2677837995136636, "learning_rate": 1.4201658676502294e-06, "loss": 0.4433, "step": 23393 }, { "epoch": 0.8334609972032706, "grad_norm": 1.6773186105243658, "learning_rate": 1.4195731735829854e-06, "loss": 0.5013, "step": 23394 }, { "epoch": 0.8334966243297647, "grad_norm": 1.6188202205888826, "learning_rate": 1.418980593770125e-06, "loss": 0.505, "step": 23395 }, { "epoch": 0.8335322514562588, "grad_norm": 1.5392779561445071, "learning_rate": 1.4183881282195333e-06, "loss": 0.432, "step": 23396 }, { "epoch": 0.8335678785827529, "grad_norm": 1.6357572192514134, "learning_rate": 1.4177957769391037e-06, "loss": 0.4212, "step": 23397 }, { "epoch": 0.833603505709247, "grad_norm": 1.2511626304026366, "learning_rate": 1.4172035399367202e-06, "loss": 0.4016, "step": 23398 }, { "epoch": 0.8336391328357411, "grad_norm": 1.2151451208882622, "learning_rate": 1.4166114172202728e-06, "loss": 0.4161, "step": 23399 }, { "epoch": 0.8336747599622353, "grad_norm": 1.481130709141993, "learning_rate": 1.416019408797642e-06, "loss": 0.4633, "step": 23400 }, { "epoch": 0.8337103870887294, "grad_norm": 6.203693573458239, "learning_rate": 1.4154275146767117e-06, "loss": 0.4117, "step": 23401 }, { "epoch": 0.8337460142152234, "grad_norm": 1.541282878027584, "learning_rate": 1.414835734865364e-06, "loss": 0.4601, "step": 23402 }, { "epoch": 0.8337816413417176, "grad_norm": 1.343672981837633, "learning_rate": 1.414244069371482e-06, "loss": 0.4217, "step": 23403 }, { "epoch": 0.8338172684682117, "grad_norm": 1.5868432523219924, "learning_rate": 1.4136525182029392e-06, "loss": 0.5141, "step": 23404 }, { "epoch": 0.8338528955947058, "grad_norm": 1.7153643042073845, "learning_rate": 1.4130610813676172e-06, "loss": 0.4426, "step": 23405 }, { "epoch": 0.8338885227212, "grad_norm": 1.8731590158918305, "learning_rate": 1.412469758873386e-06, "loss": 0.4249, "step": 23406 }, { "epoch": 0.833924149847694, "grad_norm": 1.5491112058542442, "learning_rate": 1.411878550728123e-06, "loss": 0.5961, "step": 23407 }, { "epoch": 0.8339597769741881, "grad_norm": 1.3961279486070621, "learning_rate": 1.4112874569396983e-06, "loss": 0.5513, "step": 23408 }, { "epoch": 0.8339954041006823, "grad_norm": 1.27651456269699, "learning_rate": 1.410696477515986e-06, "loss": 0.3872, "step": 23409 }, { "epoch": 0.8340310312271764, "grad_norm": 4.2017521664184985, "learning_rate": 1.410105612464855e-06, "loss": 0.5021, "step": 23410 }, { "epoch": 0.8340666583536704, "grad_norm": 1.7757875639822918, "learning_rate": 1.409514861794169e-06, "loss": 0.6081, "step": 23411 }, { "epoch": 0.8341022854801646, "grad_norm": 1.4658042438682521, "learning_rate": 1.4089242255118007e-06, "loss": 0.4311, "step": 23412 }, { "epoch": 0.8341379126066587, "grad_norm": 1.497348325095195, "learning_rate": 1.4083337036256072e-06, "loss": 0.4132, "step": 23413 }, { "epoch": 0.8341735397331528, "grad_norm": 1.2719143533047796, "learning_rate": 1.407743296143459e-06, "loss": 0.3866, "step": 23414 }, { "epoch": 0.834209166859647, "grad_norm": 1.2181436760240738, "learning_rate": 1.4071530030732094e-06, "loss": 0.4438, "step": 23415 }, { "epoch": 0.834244793986141, "grad_norm": 1.4088270529301574, "learning_rate": 1.4065628244227292e-06, "loss": 0.3917, "step": 23416 }, { "epoch": 0.8342804211126351, "grad_norm": 2.361281318277322, "learning_rate": 1.4059727601998685e-06, "loss": 0.5059, "step": 23417 }, { "epoch": 0.8343160482391293, "grad_norm": 2.105198019590953, "learning_rate": 1.4053828104124867e-06, "loss": 0.4254, "step": 23418 }, { "epoch": 0.8343516753656234, "grad_norm": 1.5806886151627095, "learning_rate": 1.4047929750684431e-06, "loss": 0.5335, "step": 23419 }, { "epoch": 0.8343873024921175, "grad_norm": 1.5558102749191713, "learning_rate": 1.4042032541755867e-06, "loss": 0.4832, "step": 23420 }, { "epoch": 0.8344229296186116, "grad_norm": 1.5120160333334907, "learning_rate": 1.403613647741774e-06, "loss": 0.4593, "step": 23421 }, { "epoch": 0.8344585567451057, "grad_norm": 2.5823054179914298, "learning_rate": 1.4030241557748514e-06, "loss": 0.4653, "step": 23422 }, { "epoch": 0.8344941838715998, "grad_norm": 1.3829432737681464, "learning_rate": 1.4024347782826708e-06, "loss": 0.4076, "step": 23423 }, { "epoch": 0.834529810998094, "grad_norm": 1.7048930659250665, "learning_rate": 1.4018455152730814e-06, "loss": 0.5336, "step": 23424 }, { "epoch": 0.834565438124588, "grad_norm": 1.4263263757931748, "learning_rate": 1.4012563667539302e-06, "loss": 0.4202, "step": 23425 }, { "epoch": 0.8346010652510821, "grad_norm": 1.657686565722889, "learning_rate": 1.4006673327330588e-06, "loss": 0.5088, "step": 23426 }, { "epoch": 0.8346366923775763, "grad_norm": 1.5735840341078908, "learning_rate": 1.4000784132183154e-06, "loss": 0.6035, "step": 23427 }, { "epoch": 0.8346723195040704, "grad_norm": 1.611296026899132, "learning_rate": 1.3994896082175357e-06, "loss": 0.4414, "step": 23428 }, { "epoch": 0.8347079466305645, "grad_norm": 1.540684081219133, "learning_rate": 1.3989009177385637e-06, "loss": 0.4029, "step": 23429 }, { "epoch": 0.8347435737570587, "grad_norm": 1.2530161215662077, "learning_rate": 1.3983123417892376e-06, "loss": 0.3465, "step": 23430 }, { "epoch": 0.8347792008835527, "grad_norm": 1.3566546695740613, "learning_rate": 1.397723880377395e-06, "loss": 0.3708, "step": 23431 }, { "epoch": 0.8348148280100468, "grad_norm": 1.939611086291748, "learning_rate": 1.397135533510874e-06, "loss": 0.5265, "step": 23432 }, { "epoch": 0.834850455136541, "grad_norm": 1.5004346962964794, "learning_rate": 1.396547301197504e-06, "loss": 0.3886, "step": 23433 }, { "epoch": 0.8348860822630351, "grad_norm": 1.2473474655203638, "learning_rate": 1.3959591834451225e-06, "loss": 0.3624, "step": 23434 }, { "epoch": 0.8349217093895291, "grad_norm": 1.713220462280173, "learning_rate": 1.3953711802615556e-06, "loss": 0.5456, "step": 23435 }, { "epoch": 0.8349573365160233, "grad_norm": 1.5937306100575497, "learning_rate": 1.3947832916546377e-06, "loss": 0.4604, "step": 23436 }, { "epoch": 0.8349929636425174, "grad_norm": 1.677908204706223, "learning_rate": 1.394195517632193e-06, "loss": 0.4688, "step": 23437 }, { "epoch": 0.8350285907690115, "grad_norm": 1.5589532421224674, "learning_rate": 1.3936078582020506e-06, "loss": 0.421, "step": 23438 }, { "epoch": 0.8350642178955057, "grad_norm": 1.3956182959053955, "learning_rate": 1.3930203133720354e-06, "loss": 0.4002, "step": 23439 }, { "epoch": 0.8350998450219997, "grad_norm": 1.4315763698787978, "learning_rate": 1.3924328831499722e-06, "loss": 0.4847, "step": 23440 }, { "epoch": 0.8351354721484938, "grad_norm": 1.3464146368119552, "learning_rate": 1.3918455675436793e-06, "loss": 0.382, "step": 23441 }, { "epoch": 0.835171099274988, "grad_norm": 1.1910255716887899, "learning_rate": 1.3912583665609792e-06, "loss": 0.4081, "step": 23442 }, { "epoch": 0.8352067264014821, "grad_norm": 1.7409404065196041, "learning_rate": 1.3906712802096933e-06, "loss": 0.4827, "step": 23443 }, { "epoch": 0.8352423535279762, "grad_norm": 1.2302074722935896, "learning_rate": 1.3900843084976357e-06, "loss": 0.377, "step": 23444 }, { "epoch": 0.8352779806544703, "grad_norm": 1.4463518325717273, "learning_rate": 1.389497451432622e-06, "loss": 0.4148, "step": 23445 }, { "epoch": 0.8353136077809644, "grad_norm": 1.2946293928119663, "learning_rate": 1.3889107090224696e-06, "loss": 0.4024, "step": 23446 }, { "epoch": 0.8353492349074585, "grad_norm": 1.6147455560763835, "learning_rate": 1.388324081274991e-06, "loss": 0.439, "step": 23447 }, { "epoch": 0.8353848620339527, "grad_norm": 1.984720293242324, "learning_rate": 1.3877375681979944e-06, "loss": 0.5245, "step": 23448 }, { "epoch": 0.8354204891604468, "grad_norm": 1.6325316559262275, "learning_rate": 1.3871511697992946e-06, "loss": 0.4447, "step": 23449 }, { "epoch": 0.8354561162869408, "grad_norm": 1.7524250779300605, "learning_rate": 1.3865648860866943e-06, "loss": 0.5101, "step": 23450 }, { "epoch": 0.835491743413435, "grad_norm": 1.339683989298373, "learning_rate": 1.385978717068005e-06, "loss": 0.38, "step": 23451 }, { "epoch": 0.8355273705399291, "grad_norm": 1.552728999836166, "learning_rate": 1.3853926627510283e-06, "loss": 0.5284, "step": 23452 }, { "epoch": 0.8355629976664232, "grad_norm": 1.3627150184916978, "learning_rate": 1.384806723143567e-06, "loss": 0.4097, "step": 23453 }, { "epoch": 0.8355986247929174, "grad_norm": 1.6169295000167456, "learning_rate": 1.38422089825343e-06, "loss": 0.5118, "step": 23454 }, { "epoch": 0.8356342519194114, "grad_norm": 1.4544905877345153, "learning_rate": 1.3836351880884113e-06, "loss": 0.505, "step": 23455 }, { "epoch": 0.8356698790459055, "grad_norm": 1.7878407145886743, "learning_rate": 1.383049592656316e-06, "loss": 0.4849, "step": 23456 }, { "epoch": 0.8357055061723997, "grad_norm": 2.3106584775729715, "learning_rate": 1.382464111964935e-06, "loss": 0.5221, "step": 23457 }, { "epoch": 0.8357411332988938, "grad_norm": 1.6310658936278937, "learning_rate": 1.3818787460220695e-06, "loss": 0.5271, "step": 23458 }, { "epoch": 0.8357767604253878, "grad_norm": 1.5716513345597893, "learning_rate": 1.3812934948355117e-06, "loss": 0.5731, "step": 23459 }, { "epoch": 0.835812387551882, "grad_norm": 2.1013479304181972, "learning_rate": 1.380708358413053e-06, "loss": 0.5053, "step": 23460 }, { "epoch": 0.8358480146783761, "grad_norm": 1.7990542876306415, "learning_rate": 1.380123336762489e-06, "loss": 0.5508, "step": 23461 }, { "epoch": 0.8358836418048702, "grad_norm": 1.3313199591270284, "learning_rate": 1.3795384298916092e-06, "loss": 0.4188, "step": 23462 }, { "epoch": 0.8359192689313644, "grad_norm": 1.4743365617174797, "learning_rate": 1.378953637808198e-06, "loss": 0.4737, "step": 23463 }, { "epoch": 0.8359548960578584, "grad_norm": 2.1564698999789145, "learning_rate": 1.3783689605200446e-06, "loss": 0.526, "step": 23464 }, { "epoch": 0.8359905231843525, "grad_norm": 1.452333524554912, "learning_rate": 1.377784398034937e-06, "loss": 0.4043, "step": 23465 }, { "epoch": 0.8360261503108467, "grad_norm": 2.766634472389365, "learning_rate": 1.3771999503606547e-06, "loss": 0.5248, "step": 23466 }, { "epoch": 0.8360617774373408, "grad_norm": 1.4148321339818732, "learning_rate": 1.3766156175049827e-06, "loss": 0.3978, "step": 23467 }, { "epoch": 0.8360974045638349, "grad_norm": 1.3362952538508654, "learning_rate": 1.3760313994757002e-06, "loss": 0.4292, "step": 23468 }, { "epoch": 0.836133031690329, "grad_norm": 1.7210321331561338, "learning_rate": 1.3754472962805899e-06, "loss": 0.4076, "step": 23469 }, { "epoch": 0.8361686588168231, "grad_norm": 1.6720824885663608, "learning_rate": 1.3748633079274254e-06, "loss": 0.4794, "step": 23470 }, { "epoch": 0.8362042859433172, "grad_norm": 1.4853641103253012, "learning_rate": 1.3742794344239862e-06, "loss": 0.4189, "step": 23471 }, { "epoch": 0.8362399130698114, "grad_norm": 1.5710295622449935, "learning_rate": 1.373695675778043e-06, "loss": 0.5147, "step": 23472 }, { "epoch": 0.8362755401963055, "grad_norm": 1.4162571616572155, "learning_rate": 1.3731120319973745e-06, "loss": 0.3989, "step": 23473 }, { "epoch": 0.8363111673227995, "grad_norm": 1.2490180481354702, "learning_rate": 1.3725285030897463e-06, "loss": 0.3718, "step": 23474 }, { "epoch": 0.8363467944492937, "grad_norm": 1.4649449604578904, "learning_rate": 1.3719450890629304e-06, "loss": 0.4438, "step": 23475 }, { "epoch": 0.8363824215757878, "grad_norm": 1.6071657528678165, "learning_rate": 1.3713617899246977e-06, "loss": 0.6074, "step": 23476 }, { "epoch": 0.8364180487022819, "grad_norm": 2.099381384283177, "learning_rate": 1.3707786056828132e-06, "loss": 0.464, "step": 23477 }, { "epoch": 0.836453675828776, "grad_norm": 1.2882018885139834, "learning_rate": 1.3701955363450447e-06, "loss": 0.3622, "step": 23478 }, { "epoch": 0.8364893029552701, "grad_norm": 1.7419465401791747, "learning_rate": 1.3696125819191531e-06, "loss": 0.5237, "step": 23479 }, { "epoch": 0.8365249300817642, "grad_norm": 5.3219002620151485, "learning_rate": 1.369029742412904e-06, "loss": 0.4284, "step": 23480 }, { "epoch": 0.8365605572082584, "grad_norm": 1.3846611586645503, "learning_rate": 1.3684470178340548e-06, "loss": 0.4215, "step": 23481 }, { "epoch": 0.8365961843347525, "grad_norm": 1.6256813996633643, "learning_rate": 1.367864408190367e-06, "loss": 0.5112, "step": 23482 }, { "epoch": 0.8366318114612465, "grad_norm": 1.4998152097555073, "learning_rate": 1.3672819134895964e-06, "loss": 0.3856, "step": 23483 }, { "epoch": 0.8366674385877407, "grad_norm": 1.6619307908645682, "learning_rate": 1.366699533739505e-06, "loss": 0.531, "step": 23484 }, { "epoch": 0.8367030657142348, "grad_norm": 1.4515613013222022, "learning_rate": 1.3661172689478408e-06, "loss": 0.4391, "step": 23485 }, { "epoch": 0.8367386928407289, "grad_norm": 1.2761061859520886, "learning_rate": 1.365535119122362e-06, "loss": 0.3766, "step": 23486 }, { "epoch": 0.8367743199672231, "grad_norm": 2.462701271940064, "learning_rate": 1.3649530842708158e-06, "loss": 0.5146, "step": 23487 }, { "epoch": 0.8368099470937171, "grad_norm": 1.7015406869445713, "learning_rate": 1.364371164400955e-06, "loss": 0.3591, "step": 23488 }, { "epoch": 0.8368455742202112, "grad_norm": 1.2672558997141994, "learning_rate": 1.363789359520531e-06, "loss": 0.4095, "step": 23489 }, { "epoch": 0.8368812013467054, "grad_norm": 1.3667912435471317, "learning_rate": 1.3632076696372831e-06, "loss": 0.3866, "step": 23490 }, { "epoch": 0.8369168284731995, "grad_norm": 1.3783770201052434, "learning_rate": 1.3626260947589664e-06, "loss": 0.4267, "step": 23491 }, { "epoch": 0.8369524555996936, "grad_norm": 1.3562403197613884, "learning_rate": 1.362044634893318e-06, "loss": 0.4094, "step": 23492 }, { "epoch": 0.8369880827261877, "grad_norm": 1.7638483317909475, "learning_rate": 1.361463290048085e-06, "loss": 0.4859, "step": 23493 }, { "epoch": 0.8370237098526818, "grad_norm": 1.4118034627701883, "learning_rate": 1.3608820602310036e-06, "loss": 0.517, "step": 23494 }, { "epoch": 0.8370593369791759, "grad_norm": 1.3142275644645682, "learning_rate": 1.3603009454498184e-06, "loss": 0.3904, "step": 23495 }, { "epoch": 0.8370949641056701, "grad_norm": 1.3440248835024278, "learning_rate": 1.3597199457122633e-06, "loss": 0.4122, "step": 23496 }, { "epoch": 0.8371305912321642, "grad_norm": 1.3480815536897277, "learning_rate": 1.3591390610260757e-06, "loss": 0.4048, "step": 23497 }, { "epoch": 0.8371662183586582, "grad_norm": 1.4662195781542584, "learning_rate": 1.3585582913989915e-06, "loss": 0.4268, "step": 23498 }, { "epoch": 0.8372018454851524, "grad_norm": 1.4413278268336418, "learning_rate": 1.3579776368387442e-06, "loss": 0.433, "step": 23499 }, { "epoch": 0.8372374726116465, "grad_norm": 1.7890737991558094, "learning_rate": 1.3573970973530659e-06, "loss": 0.492, "step": 23500 }, { "epoch": 0.8372730997381406, "grad_norm": 1.462389356542442, "learning_rate": 1.3568166729496847e-06, "loss": 0.424, "step": 23501 }, { "epoch": 0.8373087268646348, "grad_norm": 1.4960892280158333, "learning_rate": 1.356236363636333e-06, "loss": 0.4515, "step": 23502 }, { "epoch": 0.8373443539911288, "grad_norm": 1.5763076756518224, "learning_rate": 1.3556561694207337e-06, "loss": 0.5072, "step": 23503 }, { "epoch": 0.8373799811176229, "grad_norm": 2.722024431621964, "learning_rate": 1.3550760903106142e-06, "loss": 0.656, "step": 23504 }, { "epoch": 0.8374156082441171, "grad_norm": 1.5091301348854396, "learning_rate": 1.3544961263136991e-06, "loss": 0.4353, "step": 23505 }, { "epoch": 0.8374512353706112, "grad_norm": 1.3852727618237668, "learning_rate": 1.3539162774377125e-06, "loss": 0.4169, "step": 23506 }, { "epoch": 0.8374868624971052, "grad_norm": 1.4526072516294817, "learning_rate": 1.3533365436903722e-06, "loss": 0.4122, "step": 23507 }, { "epoch": 0.8375224896235994, "grad_norm": 1.2490606461076368, "learning_rate": 1.3527569250794014e-06, "loss": 0.3926, "step": 23508 }, { "epoch": 0.8375581167500935, "grad_norm": 1.5107659670476397, "learning_rate": 1.352177421612515e-06, "loss": 0.4773, "step": 23509 }, { "epoch": 0.8375937438765876, "grad_norm": 1.6606440936388074, "learning_rate": 1.35159803329743e-06, "loss": 0.4964, "step": 23510 }, { "epoch": 0.8376293710030818, "grad_norm": 1.4836562578972914, "learning_rate": 1.3510187601418646e-06, "loss": 0.4141, "step": 23511 }, { "epoch": 0.8376649981295758, "grad_norm": 1.3175471131023078, "learning_rate": 1.3504396021535259e-06, "loss": 0.4033, "step": 23512 }, { "epoch": 0.8377006252560699, "grad_norm": 1.383776000797216, "learning_rate": 1.3498605593401315e-06, "loss": 0.4338, "step": 23513 }, { "epoch": 0.8377362523825641, "grad_norm": 1.4602818305934988, "learning_rate": 1.3492816317093894e-06, "loss": 0.4219, "step": 23514 }, { "epoch": 0.8377718795090582, "grad_norm": 1.224839916469011, "learning_rate": 1.348702819269011e-06, "loss": 0.4332, "step": 23515 }, { "epoch": 0.8378075066355523, "grad_norm": 1.9530507901788157, "learning_rate": 1.3481241220266984e-06, "loss": 0.4284, "step": 23516 }, { "epoch": 0.8378431337620464, "grad_norm": 1.4670942077979685, "learning_rate": 1.3475455399901639e-06, "loss": 0.4526, "step": 23517 }, { "epoch": 0.8378787608885405, "grad_norm": 1.4542664588594592, "learning_rate": 1.3469670731671046e-06, "loss": 0.5622, "step": 23518 }, { "epoch": 0.8379143880150346, "grad_norm": 1.2756073776570651, "learning_rate": 1.3463887215652283e-06, "loss": 0.4344, "step": 23519 }, { "epoch": 0.8379500151415288, "grad_norm": 1.5118047400388974, "learning_rate": 1.3458104851922338e-06, "loss": 0.5788, "step": 23520 }, { "epoch": 0.8379856422680229, "grad_norm": 1.790931843320203, "learning_rate": 1.3452323640558217e-06, "loss": 0.5105, "step": 23521 }, { "epoch": 0.8380212693945169, "grad_norm": 1.8966190929188833, "learning_rate": 1.3446543581636928e-06, "loss": 0.5343, "step": 23522 }, { "epoch": 0.8380568965210111, "grad_norm": 1.2385921389857268, "learning_rate": 1.3440764675235384e-06, "loss": 0.3763, "step": 23523 }, { "epoch": 0.8380925236475052, "grad_norm": 1.2119510611355042, "learning_rate": 1.3434986921430582e-06, "loss": 0.3868, "step": 23524 }, { "epoch": 0.8381281507739993, "grad_norm": 1.4727533973508782, "learning_rate": 1.3429210320299402e-06, "loss": 0.4766, "step": 23525 }, { "epoch": 0.8381637779004935, "grad_norm": 1.2935084166764272, "learning_rate": 1.3423434871918828e-06, "loss": 0.3895, "step": 23526 }, { "epoch": 0.8381994050269875, "grad_norm": 1.4071828556914963, "learning_rate": 1.341766057636571e-06, "loss": 0.4003, "step": 23527 }, { "epoch": 0.8382350321534816, "grad_norm": 1.245981615773337, "learning_rate": 1.3411887433716953e-06, "loss": 0.4163, "step": 23528 }, { "epoch": 0.8382706592799758, "grad_norm": 1.5740064101615208, "learning_rate": 1.3406115444049439e-06, "loss": 0.5291, "step": 23529 }, { "epoch": 0.8383062864064699, "grad_norm": 1.3178840675495445, "learning_rate": 1.3400344607440053e-06, "loss": 0.3985, "step": 23530 }, { "epoch": 0.838341913532964, "grad_norm": 1.1600354650370936, "learning_rate": 1.3394574923965575e-06, "loss": 0.4143, "step": 23531 }, { "epoch": 0.8383775406594581, "grad_norm": 1.397370225914632, "learning_rate": 1.3388806393702891e-06, "loss": 0.4521, "step": 23532 }, { "epoch": 0.8384131677859522, "grad_norm": 1.5067802778659911, "learning_rate": 1.338303901672877e-06, "loss": 0.4371, "step": 23533 }, { "epoch": 0.8384487949124463, "grad_norm": 1.3732678979879365, "learning_rate": 1.3377272793120021e-06, "loss": 0.4036, "step": 23534 }, { "epoch": 0.8384844220389405, "grad_norm": 1.468598919210863, "learning_rate": 1.3371507722953437e-06, "loss": 0.4669, "step": 23535 }, { "epoch": 0.8385200491654345, "grad_norm": 1.3771713348477468, "learning_rate": 1.3365743806305776e-06, "loss": 0.4153, "step": 23536 }, { "epoch": 0.8385556762919286, "grad_norm": 1.787335726714151, "learning_rate": 1.33599810432538e-06, "loss": 0.5233, "step": 23537 }, { "epoch": 0.8385913034184228, "grad_norm": 1.3491219017441147, "learning_rate": 1.3354219433874227e-06, "loss": 0.4212, "step": 23538 }, { "epoch": 0.8386269305449169, "grad_norm": 1.477656704946695, "learning_rate": 1.3348458978243795e-06, "loss": 0.5058, "step": 23539 }, { "epoch": 0.838662557671411, "grad_norm": 1.6879258836122697, "learning_rate": 1.3342699676439186e-06, "loss": 0.5559, "step": 23540 }, { "epoch": 0.8386981847979051, "grad_norm": 1.5491307658839508, "learning_rate": 1.3336941528537118e-06, "loss": 0.4694, "step": 23541 }, { "epoch": 0.8387338119243992, "grad_norm": 1.51973956848805, "learning_rate": 1.3331184534614194e-06, "loss": 0.4098, "step": 23542 }, { "epoch": 0.8387694390508933, "grad_norm": 1.5352854251338204, "learning_rate": 1.3325428694747178e-06, "loss": 0.4607, "step": 23543 }, { "epoch": 0.8388050661773875, "grad_norm": 1.3357370700962485, "learning_rate": 1.331967400901264e-06, "loss": 0.4207, "step": 23544 }, { "epoch": 0.8388406933038816, "grad_norm": 1.596777562713325, "learning_rate": 1.331392047748723e-06, "loss": 0.5232, "step": 23545 }, { "epoch": 0.8388763204303756, "grad_norm": 1.7160741057040414, "learning_rate": 1.3308168100247575e-06, "loss": 0.3694, "step": 23546 }, { "epoch": 0.8389119475568698, "grad_norm": 1.5405082163525912, "learning_rate": 1.3302416877370238e-06, "loss": 0.5022, "step": 23547 }, { "epoch": 0.8389475746833639, "grad_norm": 1.2789229741164563, "learning_rate": 1.3296666808931846e-06, "loss": 0.4297, "step": 23548 }, { "epoch": 0.8389832018098581, "grad_norm": 1.4550068476733249, "learning_rate": 1.3290917895008903e-06, "loss": 0.3774, "step": 23549 }, { "epoch": 0.8390188289363522, "grad_norm": 1.6740946828679495, "learning_rate": 1.3285170135678005e-06, "loss": 0.3908, "step": 23550 }, { "epoch": 0.8390544560628462, "grad_norm": 1.4079100641108084, "learning_rate": 1.3279423531015667e-06, "loss": 0.4501, "step": 23551 }, { "epoch": 0.8390900831893404, "grad_norm": 1.3111849830398548, "learning_rate": 1.327367808109845e-06, "loss": 0.3958, "step": 23552 }, { "epoch": 0.8391257103158345, "grad_norm": 1.361702812700728, "learning_rate": 1.3267933786002796e-06, "loss": 0.4098, "step": 23553 }, { "epoch": 0.8391613374423286, "grad_norm": 1.29858873746444, "learning_rate": 1.326219064580525e-06, "loss": 0.3753, "step": 23554 }, { "epoch": 0.8391969645688228, "grad_norm": 1.4016340537095058, "learning_rate": 1.3256448660582244e-06, "loss": 0.3919, "step": 23555 }, { "epoch": 0.8392325916953168, "grad_norm": 1.3107606107808627, "learning_rate": 1.3250707830410259e-06, "loss": 0.3712, "step": 23556 }, { "epoch": 0.8392682188218109, "grad_norm": 1.3273347349794118, "learning_rate": 1.3244968155365723e-06, "loss": 0.391, "step": 23557 }, { "epoch": 0.8393038459483051, "grad_norm": 1.4602097718942824, "learning_rate": 1.3239229635525074e-06, "loss": 0.6017, "step": 23558 }, { "epoch": 0.8393394730747992, "grad_norm": 1.8487718160938436, "learning_rate": 1.3233492270964754e-06, "loss": 0.5057, "step": 23559 }, { "epoch": 0.8393751002012932, "grad_norm": 1.2153156918228152, "learning_rate": 1.322775606176111e-06, "loss": 0.3943, "step": 23560 }, { "epoch": 0.8394107273277874, "grad_norm": 1.7379941215185417, "learning_rate": 1.322202100799056e-06, "loss": 0.5528, "step": 23561 }, { "epoch": 0.8394463544542815, "grad_norm": 1.2242242053088506, "learning_rate": 1.321628710972943e-06, "loss": 0.3687, "step": 23562 }, { "epoch": 0.8394819815807756, "grad_norm": 1.3958835677919266, "learning_rate": 1.3210554367054119e-06, "loss": 0.409, "step": 23563 }, { "epoch": 0.8395176087072698, "grad_norm": 1.2817358106142835, "learning_rate": 1.3204822780040916e-06, "loss": 0.4045, "step": 23564 }, { "epoch": 0.8395532358337638, "grad_norm": 2.0434345659997213, "learning_rate": 1.3199092348766174e-06, "loss": 0.5257, "step": 23565 }, { "epoch": 0.8395888629602579, "grad_norm": 1.2716763370952298, "learning_rate": 1.3193363073306176e-06, "loss": 0.3609, "step": 23566 }, { "epoch": 0.8396244900867521, "grad_norm": 1.893072035442626, "learning_rate": 1.318763495373725e-06, "loss": 0.3717, "step": 23567 }, { "epoch": 0.8396601172132462, "grad_norm": 1.2975315364419273, "learning_rate": 1.3181907990135624e-06, "loss": 0.3673, "step": 23568 }, { "epoch": 0.8396957443397403, "grad_norm": 2.002323489375965, "learning_rate": 1.317618218257757e-06, "loss": 0.5108, "step": 23569 }, { "epoch": 0.8397313714662344, "grad_norm": 1.7000670354230731, "learning_rate": 1.3170457531139358e-06, "loss": 0.5214, "step": 23570 }, { "epoch": 0.8397669985927285, "grad_norm": 1.4963913896084353, "learning_rate": 1.3164734035897175e-06, "loss": 0.524, "step": 23571 }, { "epoch": 0.8398026257192226, "grad_norm": 1.5564163628785403, "learning_rate": 1.3159011696927248e-06, "loss": 0.4737, "step": 23572 }, { "epoch": 0.8398382528457168, "grad_norm": 1.4859307658707435, "learning_rate": 1.3153290514305784e-06, "loss": 0.5273, "step": 23573 }, { "epoch": 0.8398738799722109, "grad_norm": 1.2627365531355312, "learning_rate": 1.3147570488108985e-06, "loss": 0.4139, "step": 23574 }, { "epoch": 0.8399095070987049, "grad_norm": 1.3740652492196512, "learning_rate": 1.314185161841296e-06, "loss": 0.4431, "step": 23575 }, { "epoch": 0.8399451342251991, "grad_norm": 1.2567199077249194, "learning_rate": 1.3136133905293912e-06, "loss": 0.4285, "step": 23576 }, { "epoch": 0.8399807613516932, "grad_norm": 1.3389173944388189, "learning_rate": 1.313041734882794e-06, "loss": 0.4141, "step": 23577 }, { "epoch": 0.8400163884781873, "grad_norm": 2.362440020166253, "learning_rate": 1.31247019490912e-06, "loss": 0.3844, "step": 23578 }, { "epoch": 0.8400520156046815, "grad_norm": 1.8185389225786732, "learning_rate": 1.3118987706159724e-06, "loss": 0.4568, "step": 23579 }, { "epoch": 0.8400876427311755, "grad_norm": 1.3800202756098578, "learning_rate": 1.3113274620109684e-06, "loss": 0.3709, "step": 23580 }, { "epoch": 0.8401232698576696, "grad_norm": 1.3333634491509578, "learning_rate": 1.310756269101714e-06, "loss": 0.3937, "step": 23581 }, { "epoch": 0.8401588969841638, "grad_norm": 1.2110890358439, "learning_rate": 1.310185191895812e-06, "loss": 0.3776, "step": 23582 }, { "epoch": 0.8401945241106579, "grad_norm": 1.7201602172761359, "learning_rate": 1.3096142304008696e-06, "loss": 0.3922, "step": 23583 }, { "epoch": 0.840230151237152, "grad_norm": 1.358970553997903, "learning_rate": 1.3090433846244853e-06, "loss": 0.3891, "step": 23584 }, { "epoch": 0.8402657783636461, "grad_norm": 1.5515048645470586, "learning_rate": 1.3084726545742654e-06, "loss": 0.5581, "step": 23585 }, { "epoch": 0.8403014054901402, "grad_norm": 1.2774443582476314, "learning_rate": 1.3079020402578046e-06, "loss": 0.3997, "step": 23586 }, { "epoch": 0.8403370326166343, "grad_norm": 2.536034944901478, "learning_rate": 1.3073315416827036e-06, "loss": 0.4875, "step": 23587 }, { "epoch": 0.8403726597431285, "grad_norm": 1.8822520715957811, "learning_rate": 1.3067611588565598e-06, "loss": 0.5929, "step": 23588 }, { "epoch": 0.8404082868696225, "grad_norm": 1.173911655918736, "learning_rate": 1.3061908917869692e-06, "loss": 0.3333, "step": 23589 }, { "epoch": 0.8404439139961166, "grad_norm": 1.3679533076278234, "learning_rate": 1.3056207404815214e-06, "loss": 0.4848, "step": 23590 }, { "epoch": 0.8404795411226108, "grad_norm": 1.7651644006564404, "learning_rate": 1.30505070494781e-06, "loss": 0.4907, "step": 23591 }, { "epoch": 0.8405151682491049, "grad_norm": 1.881416738758459, "learning_rate": 1.3044807851934283e-06, "loss": 0.5313, "step": 23592 }, { "epoch": 0.840550795375599, "grad_norm": 1.4322235606327645, "learning_rate": 1.3039109812259598e-06, "loss": 0.4479, "step": 23593 }, { "epoch": 0.8405864225020931, "grad_norm": 1.6698240988898272, "learning_rate": 1.3033412930529953e-06, "loss": 0.4593, "step": 23594 }, { "epoch": 0.8406220496285872, "grad_norm": 1.3941974804382928, "learning_rate": 1.302771720682121e-06, "loss": 0.448, "step": 23595 }, { "epoch": 0.8406576767550813, "grad_norm": 1.5096692870725998, "learning_rate": 1.302202264120922e-06, "loss": 0.4571, "step": 23596 }, { "epoch": 0.8406933038815755, "grad_norm": 1.7083084366581298, "learning_rate": 1.3016329233769765e-06, "loss": 0.5363, "step": 23597 }, { "epoch": 0.8407289310080696, "grad_norm": 1.395183262518693, "learning_rate": 1.3010636984578706e-06, "loss": 0.4087, "step": 23598 }, { "epoch": 0.8407645581345636, "grad_norm": 1.5262290562692822, "learning_rate": 1.3004945893711796e-06, "loss": 0.4984, "step": 23599 }, { "epoch": 0.8408001852610578, "grad_norm": 2.190189142629379, "learning_rate": 1.299925596124486e-06, "loss": 0.5178, "step": 23600 }, { "epoch": 0.8408358123875519, "grad_norm": 1.3400045187210647, "learning_rate": 1.2993567187253619e-06, "loss": 0.4333, "step": 23601 }, { "epoch": 0.840871439514046, "grad_norm": 1.352733461365349, "learning_rate": 1.2987879571813855e-06, "loss": 0.4395, "step": 23602 }, { "epoch": 0.8409070666405402, "grad_norm": 1.5000211192478736, "learning_rate": 1.2982193115001285e-06, "loss": 0.5945, "step": 23603 }, { "epoch": 0.8409426937670342, "grad_norm": 1.733457639095759, "learning_rate": 1.2976507816891627e-06, "loss": 0.4528, "step": 23604 }, { "epoch": 0.8409783208935283, "grad_norm": 1.5817515090107368, "learning_rate": 1.2970823677560628e-06, "loss": 0.4201, "step": 23605 }, { "epoch": 0.8410139480200225, "grad_norm": 1.347461810775989, "learning_rate": 1.2965140697083911e-06, "loss": 0.3937, "step": 23606 }, { "epoch": 0.8410495751465166, "grad_norm": 1.41964671151227, "learning_rate": 1.2959458875537201e-06, "loss": 0.4695, "step": 23607 }, { "epoch": 0.8410852022730106, "grad_norm": 1.3222115070216278, "learning_rate": 1.2953778212996115e-06, "loss": 0.3971, "step": 23608 }, { "epoch": 0.8411208293995048, "grad_norm": 1.3148362879166307, "learning_rate": 1.2948098709536317e-06, "loss": 0.3828, "step": 23609 }, { "epoch": 0.8411564565259989, "grad_norm": 1.3978155167946673, "learning_rate": 1.2942420365233421e-06, "loss": 0.4461, "step": 23610 }, { "epoch": 0.841192083652493, "grad_norm": 1.560407256897863, "learning_rate": 1.2936743180163081e-06, "loss": 0.4684, "step": 23611 }, { "epoch": 0.8412277107789872, "grad_norm": 1.5923095174084563, "learning_rate": 1.2931067154400834e-06, "loss": 0.4815, "step": 23612 }, { "epoch": 0.8412633379054812, "grad_norm": 2.018392604630685, "learning_rate": 1.2925392288022299e-06, "loss": 0.5205, "step": 23613 }, { "epoch": 0.8412989650319753, "grad_norm": 1.3398951766102367, "learning_rate": 1.2919718581103013e-06, "loss": 0.4082, "step": 23614 }, { "epoch": 0.8413345921584695, "grad_norm": 1.2778237794476417, "learning_rate": 1.291404603371853e-06, "loss": 0.4123, "step": 23615 }, { "epoch": 0.8413702192849636, "grad_norm": 1.8774045151496297, "learning_rate": 1.2908374645944432e-06, "loss": 0.5375, "step": 23616 }, { "epoch": 0.8414058464114577, "grad_norm": 1.531702117187667, "learning_rate": 1.2902704417856137e-06, "loss": 0.5618, "step": 23617 }, { "epoch": 0.8414414735379518, "grad_norm": 1.211092608690411, "learning_rate": 1.2897035349529263e-06, "loss": 0.3914, "step": 23618 }, { "epoch": 0.8414771006644459, "grad_norm": 1.5966062978464557, "learning_rate": 1.2891367441039226e-06, "loss": 0.4487, "step": 23619 }, { "epoch": 0.84151272779094, "grad_norm": 1.4253450732229025, "learning_rate": 1.2885700692461533e-06, "loss": 0.4184, "step": 23620 }, { "epoch": 0.8415483549174342, "grad_norm": 1.3039586434075867, "learning_rate": 1.2880035103871614e-06, "loss": 0.3902, "step": 23621 }, { "epoch": 0.8415839820439283, "grad_norm": 1.5622514817695068, "learning_rate": 1.2874370675344927e-06, "loss": 0.5484, "step": 23622 }, { "epoch": 0.8416196091704223, "grad_norm": 1.4185142140090612, "learning_rate": 1.2868707406956881e-06, "loss": 0.3881, "step": 23623 }, { "epoch": 0.8416552362969165, "grad_norm": 2.057436326631339, "learning_rate": 1.2863045298782894e-06, "loss": 0.4816, "step": 23624 }, { "epoch": 0.8416908634234106, "grad_norm": 1.9143992834335166, "learning_rate": 1.2857384350898361e-06, "loss": 0.5386, "step": 23625 }, { "epoch": 0.8417264905499047, "grad_norm": 1.4755240813619621, "learning_rate": 1.2851724563378664e-06, "loss": 0.4238, "step": 23626 }, { "epoch": 0.8417621176763989, "grad_norm": 1.5293362300309317, "learning_rate": 1.284606593629919e-06, "loss": 0.5134, "step": 23627 }, { "epoch": 0.8417977448028929, "grad_norm": 1.4889401490032375, "learning_rate": 1.2840408469735245e-06, "loss": 0.4232, "step": 23628 }, { "epoch": 0.841833371929387, "grad_norm": 1.4225475547388848, "learning_rate": 1.2834752163762198e-06, "loss": 0.4285, "step": 23629 }, { "epoch": 0.8418689990558812, "grad_norm": 1.8145034612959363, "learning_rate": 1.2829097018455339e-06, "loss": 0.4946, "step": 23630 }, { "epoch": 0.8419046261823753, "grad_norm": 1.7133744569226228, "learning_rate": 1.2823443033889983e-06, "loss": 0.4647, "step": 23631 }, { "epoch": 0.8419402533088693, "grad_norm": 1.8102475612229112, "learning_rate": 1.2817790210141401e-06, "loss": 0.5935, "step": 23632 }, { "epoch": 0.8419758804353635, "grad_norm": 1.4219363047630436, "learning_rate": 1.2812138547284913e-06, "loss": 0.4344, "step": 23633 }, { "epoch": 0.8420115075618576, "grad_norm": 1.8139750187474144, "learning_rate": 1.2806488045395715e-06, "loss": 0.4884, "step": 23634 }, { "epoch": 0.8420471346883517, "grad_norm": 1.473592637814055, "learning_rate": 1.28008387045491e-06, "loss": 0.4607, "step": 23635 }, { "epoch": 0.8420827618148459, "grad_norm": 1.2120772128776107, "learning_rate": 1.2795190524820245e-06, "loss": 0.3902, "step": 23636 }, { "epoch": 0.84211838894134, "grad_norm": 1.5607580953835707, "learning_rate": 1.2789543506284374e-06, "loss": 0.4264, "step": 23637 }, { "epoch": 0.842154016067834, "grad_norm": 1.4877234420533616, "learning_rate": 1.2783897649016709e-06, "loss": 0.5412, "step": 23638 }, { "epoch": 0.8421896431943282, "grad_norm": 1.3477506191628361, "learning_rate": 1.2778252953092384e-06, "loss": 0.3901, "step": 23639 }, { "epoch": 0.8422252703208223, "grad_norm": 1.1672197952032157, "learning_rate": 1.277260941858659e-06, "loss": 0.4152, "step": 23640 }, { "epoch": 0.8422608974473164, "grad_norm": 1.2481853368343188, "learning_rate": 1.2766967045574475e-06, "loss": 0.4109, "step": 23641 }, { "epoch": 0.8422965245738105, "grad_norm": 1.6829969381885181, "learning_rate": 1.2761325834131177e-06, "loss": 0.5457, "step": 23642 }, { "epoch": 0.8423321517003046, "grad_norm": 1.331943040964934, "learning_rate": 1.2755685784331784e-06, "loss": 0.429, "step": 23643 }, { "epoch": 0.8423677788267987, "grad_norm": 1.7823148745665125, "learning_rate": 1.2750046896251444e-06, "loss": 0.5509, "step": 23644 }, { "epoch": 0.8424034059532929, "grad_norm": 1.5064414749322272, "learning_rate": 1.2744409169965188e-06, "loss": 0.4993, "step": 23645 }, { "epoch": 0.842439033079787, "grad_norm": 1.7362173062583595, "learning_rate": 1.27387726055481e-06, "loss": 0.5283, "step": 23646 }, { "epoch": 0.842474660206281, "grad_norm": 1.7617969950573111, "learning_rate": 1.2733137203075263e-06, "loss": 0.512, "step": 23647 }, { "epoch": 0.8425102873327752, "grad_norm": 1.648231518490731, "learning_rate": 1.2727502962621696e-06, "loss": 0.4698, "step": 23648 }, { "epoch": 0.8425459144592693, "grad_norm": 1.9175995442847078, "learning_rate": 1.2721869884262438e-06, "loss": 0.4168, "step": 23649 }, { "epoch": 0.8425815415857634, "grad_norm": 1.673678661482528, "learning_rate": 1.2716237968072476e-06, "loss": 0.5575, "step": 23650 }, { "epoch": 0.8426171687122576, "grad_norm": 1.5110380876585847, "learning_rate": 1.2710607214126835e-06, "loss": 0.416, "step": 23651 }, { "epoch": 0.8426527958387516, "grad_norm": 1.6845287276685523, "learning_rate": 1.2704977622500436e-06, "loss": 0.4123, "step": 23652 }, { "epoch": 0.8426884229652457, "grad_norm": 1.5351250085281558, "learning_rate": 1.2699349193268307e-06, "loss": 0.5327, "step": 23653 }, { "epoch": 0.8427240500917399, "grad_norm": 1.5395811151010053, "learning_rate": 1.2693721926505342e-06, "loss": 0.3864, "step": 23654 }, { "epoch": 0.842759677218234, "grad_norm": 1.6441366672237792, "learning_rate": 1.2688095822286494e-06, "loss": 0.4295, "step": 23655 }, { "epoch": 0.842795304344728, "grad_norm": 1.3726819922584153, "learning_rate": 1.2682470880686682e-06, "loss": 0.4279, "step": 23656 }, { "epoch": 0.8428309314712222, "grad_norm": 1.432677925785278, "learning_rate": 1.267684710178081e-06, "loss": 0.458, "step": 23657 }, { "epoch": 0.8428665585977163, "grad_norm": 1.8070484873981547, "learning_rate": 1.267122448564374e-06, "loss": 0.5752, "step": 23658 }, { "epoch": 0.8429021857242104, "grad_norm": 1.3044847053657374, "learning_rate": 1.266560303235037e-06, "loss": 0.3929, "step": 23659 }, { "epoch": 0.8429378128507046, "grad_norm": 1.6043971817356562, "learning_rate": 1.265998274197553e-06, "loss": 0.4828, "step": 23660 }, { "epoch": 0.8429734399771986, "grad_norm": 2.214795889248181, "learning_rate": 1.2654363614594055e-06, "loss": 0.4728, "step": 23661 }, { "epoch": 0.8430090671036927, "grad_norm": 1.2463506923108356, "learning_rate": 1.264874565028079e-06, "loss": 0.3688, "step": 23662 }, { "epoch": 0.8430446942301869, "grad_norm": 1.827946988167343, "learning_rate": 1.264312884911052e-06, "loss": 0.5144, "step": 23663 }, { "epoch": 0.843080321356681, "grad_norm": 1.5596798909367868, "learning_rate": 1.263751321115808e-06, "loss": 0.515, "step": 23664 }, { "epoch": 0.8431159484831751, "grad_norm": 2.5536044751473392, "learning_rate": 1.2631898736498183e-06, "loss": 0.3976, "step": 23665 }, { "epoch": 0.8431515756096692, "grad_norm": 1.667471332347825, "learning_rate": 1.2626285425205652e-06, "loss": 0.5542, "step": 23666 }, { "epoch": 0.8431872027361633, "grad_norm": 1.2663133394083772, "learning_rate": 1.2620673277355178e-06, "loss": 0.4601, "step": 23667 }, { "epoch": 0.8432228298626574, "grad_norm": 1.7343451155483838, "learning_rate": 1.2615062293021508e-06, "loss": 0.5107, "step": 23668 }, { "epoch": 0.8432584569891516, "grad_norm": 1.2290710447209503, "learning_rate": 1.260945247227936e-06, "loss": 0.3991, "step": 23669 }, { "epoch": 0.8432940841156457, "grad_norm": 1.6202099488578308, "learning_rate": 1.2603843815203466e-06, "loss": 0.5322, "step": 23670 }, { "epoch": 0.8433297112421397, "grad_norm": 1.68265044381419, "learning_rate": 1.259823632186845e-06, "loss": 0.5253, "step": 23671 }, { "epoch": 0.8433653383686339, "grad_norm": 1.5595350843921783, "learning_rate": 1.2592629992349003e-06, "loss": 0.3607, "step": 23672 }, { "epoch": 0.843400965495128, "grad_norm": 1.7472280798499893, "learning_rate": 1.2587024826719817e-06, "loss": 0.453, "step": 23673 }, { "epoch": 0.8434365926216221, "grad_norm": 1.489240736849272, "learning_rate": 1.2581420825055457e-06, "loss": 0.3768, "step": 23674 }, { "epoch": 0.8434722197481163, "grad_norm": 1.6591741042013508, "learning_rate": 1.2575817987430616e-06, "loss": 0.515, "step": 23675 }, { "epoch": 0.8435078468746103, "grad_norm": 1.503238838177545, "learning_rate": 1.2570216313919836e-06, "loss": 0.4079, "step": 23676 }, { "epoch": 0.8435434740011044, "grad_norm": 1.4527183122956508, "learning_rate": 1.2564615804597734e-06, "loss": 0.4873, "step": 23677 }, { "epoch": 0.8435791011275986, "grad_norm": 2.12189226088771, "learning_rate": 1.2559016459538885e-06, "loss": 0.5061, "step": 23678 }, { "epoch": 0.8436147282540927, "grad_norm": 1.7121973020383843, "learning_rate": 1.2553418278817874e-06, "loss": 0.4119, "step": 23679 }, { "epoch": 0.8436503553805867, "grad_norm": 1.3819564328904805, "learning_rate": 1.2547821262509197e-06, "loss": 0.3946, "step": 23680 }, { "epoch": 0.8436859825070809, "grad_norm": 1.3811593459233389, "learning_rate": 1.2542225410687436e-06, "loss": 0.5155, "step": 23681 }, { "epoch": 0.843721609633575, "grad_norm": 1.3983455913821516, "learning_rate": 1.2536630723427045e-06, "loss": 0.4146, "step": 23682 }, { "epoch": 0.8437572367600691, "grad_norm": 1.443408043667627, "learning_rate": 1.2531037200802543e-06, "loss": 0.4507, "step": 23683 }, { "epoch": 0.8437928638865633, "grad_norm": 1.5956077644709101, "learning_rate": 1.2525444842888423e-06, "loss": 0.5342, "step": 23684 }, { "epoch": 0.8438284910130573, "grad_norm": 1.7666331708489835, "learning_rate": 1.251985364975915e-06, "loss": 0.4553, "step": 23685 }, { "epoch": 0.8438641181395514, "grad_norm": 1.4990389051551019, "learning_rate": 1.2514263621489187e-06, "loss": 0.4096, "step": 23686 }, { "epoch": 0.8438997452660456, "grad_norm": 1.5613788120127536, "learning_rate": 1.2508674758152927e-06, "loss": 0.4973, "step": 23687 }, { "epoch": 0.8439353723925397, "grad_norm": 1.7021463290453278, "learning_rate": 1.2503087059824848e-06, "loss": 0.5083, "step": 23688 }, { "epoch": 0.8439709995190338, "grad_norm": 1.9330528598665053, "learning_rate": 1.2497500526579288e-06, "loss": 0.4582, "step": 23689 }, { "epoch": 0.844006626645528, "grad_norm": 1.360505011133693, "learning_rate": 1.2491915158490697e-06, "loss": 0.4318, "step": 23690 }, { "epoch": 0.844042253772022, "grad_norm": 1.3662780423271454, "learning_rate": 1.2486330955633397e-06, "loss": 0.4143, "step": 23691 }, { "epoch": 0.8440778808985161, "grad_norm": 1.2821160563267275, "learning_rate": 1.248074791808177e-06, "loss": 0.3831, "step": 23692 }, { "epoch": 0.8441135080250103, "grad_norm": 1.9484051637385282, "learning_rate": 1.2475166045910158e-06, "loss": 0.5852, "step": 23693 }, { "epoch": 0.8441491351515044, "grad_norm": 1.5443814725319502, "learning_rate": 1.2469585339192892e-06, "loss": 0.4032, "step": 23694 }, { "epoch": 0.8441847622779984, "grad_norm": 1.7374281241779537, "learning_rate": 1.2464005798004287e-06, "loss": 0.5102, "step": 23695 }, { "epoch": 0.8442203894044926, "grad_norm": 1.3949978627483464, "learning_rate": 1.2458427422418618e-06, "loss": 0.4276, "step": 23696 }, { "epoch": 0.8442560165309867, "grad_norm": 1.9457273875868926, "learning_rate": 1.2452850212510182e-06, "loss": 0.3759, "step": 23697 }, { "epoch": 0.8442916436574808, "grad_norm": 1.4659535583362713, "learning_rate": 1.244727416835323e-06, "loss": 0.4342, "step": 23698 }, { "epoch": 0.844327270783975, "grad_norm": 1.4187004123614204, "learning_rate": 1.2441699290022014e-06, "loss": 0.3976, "step": 23699 }, { "epoch": 0.844362897910469, "grad_norm": 1.65410845682591, "learning_rate": 1.2436125577590774e-06, "loss": 0.4275, "step": 23700 }, { "epoch": 0.8443985250369631, "grad_norm": 1.648943892567342, "learning_rate": 1.243055303113374e-06, "loss": 0.4223, "step": 23701 }, { "epoch": 0.8444341521634573, "grad_norm": 1.97838613261827, "learning_rate": 1.2424981650725077e-06, "loss": 0.4369, "step": 23702 }, { "epoch": 0.8444697792899514, "grad_norm": 1.7705704746788093, "learning_rate": 1.2419411436439021e-06, "loss": 0.5336, "step": 23703 }, { "epoch": 0.8445054064164454, "grad_norm": 1.4050917157515617, "learning_rate": 1.2413842388349696e-06, "loss": 0.4489, "step": 23704 }, { "epoch": 0.8445410335429396, "grad_norm": 2.320367426437186, "learning_rate": 1.2408274506531305e-06, "loss": 0.5869, "step": 23705 }, { "epoch": 0.8445766606694337, "grad_norm": 1.8204791512280185, "learning_rate": 1.240270779105791e-06, "loss": 0.5391, "step": 23706 }, { "epoch": 0.8446122877959278, "grad_norm": 1.3472052959753182, "learning_rate": 1.2397142242003713e-06, "loss": 0.3919, "step": 23707 }, { "epoch": 0.844647914922422, "grad_norm": 1.4163117671615986, "learning_rate": 1.2391577859442816e-06, "loss": 0.4087, "step": 23708 }, { "epoch": 0.844683542048916, "grad_norm": 1.6985735090673155, "learning_rate": 1.2386014643449284e-06, "loss": 0.4678, "step": 23709 }, { "epoch": 0.8447191691754101, "grad_norm": 1.415897088201427, "learning_rate": 1.2380452594097215e-06, "loss": 0.3815, "step": 23710 }, { "epoch": 0.8447547963019043, "grad_norm": 1.446893738263467, "learning_rate": 1.2374891711460645e-06, "loss": 0.381, "step": 23711 }, { "epoch": 0.8447904234283984, "grad_norm": 1.361752371618919, "learning_rate": 1.2369331995613664e-06, "loss": 0.3696, "step": 23712 }, { "epoch": 0.8448260505548925, "grad_norm": 1.6032071433569584, "learning_rate": 1.2363773446630245e-06, "loss": 0.456, "step": 23713 }, { "epoch": 0.8448616776813866, "grad_norm": 1.4547934689169375, "learning_rate": 1.235821606458445e-06, "loss": 0.4631, "step": 23714 }, { "epoch": 0.8448973048078807, "grad_norm": 1.5493864814803042, "learning_rate": 1.2352659849550253e-06, "loss": 0.4201, "step": 23715 }, { "epoch": 0.8449329319343748, "grad_norm": 1.7885359544837887, "learning_rate": 1.2347104801601683e-06, "loss": 0.4886, "step": 23716 }, { "epoch": 0.844968559060869, "grad_norm": 1.438494511407654, "learning_rate": 1.234155092081265e-06, "loss": 0.3948, "step": 23717 }, { "epoch": 0.8450041861873631, "grad_norm": 1.9071791537696352, "learning_rate": 1.2335998207257138e-06, "loss": 0.5474, "step": 23718 }, { "epoch": 0.8450398133138571, "grad_norm": 1.4424436429568053, "learning_rate": 1.233044666100911e-06, "loss": 0.368, "step": 23719 }, { "epoch": 0.8450754404403513, "grad_norm": 1.548704431704921, "learning_rate": 1.232489628214244e-06, "loss": 0.458, "step": 23720 }, { "epoch": 0.8451110675668454, "grad_norm": 1.3697797459434047, "learning_rate": 1.2319347070731047e-06, "loss": 0.3975, "step": 23721 }, { "epoch": 0.8451466946933395, "grad_norm": 1.6444852241150314, "learning_rate": 1.2313799026848839e-06, "loss": 0.4721, "step": 23722 }, { "epoch": 0.8451823218198337, "grad_norm": 1.563810550292934, "learning_rate": 1.230825215056971e-06, "loss": 0.435, "step": 23723 }, { "epoch": 0.8452179489463277, "grad_norm": 2.012692955307848, "learning_rate": 1.230270644196747e-06, "loss": 0.5022, "step": 23724 }, { "epoch": 0.8452535760728218, "grad_norm": 1.6911232463005406, "learning_rate": 1.2297161901116018e-06, "loss": 0.4281, "step": 23725 }, { "epoch": 0.845289203199316, "grad_norm": 1.5767437679933105, "learning_rate": 1.2291618528089134e-06, "loss": 0.4671, "step": 23726 }, { "epoch": 0.8453248303258101, "grad_norm": 1.8590137478302495, "learning_rate": 1.2286076322960672e-06, "loss": 0.4288, "step": 23727 }, { "epoch": 0.8453604574523041, "grad_norm": 1.5350522171037473, "learning_rate": 1.2280535285804396e-06, "loss": 0.3652, "step": 23728 }, { "epoch": 0.8453960845787983, "grad_norm": 1.5039222924038713, "learning_rate": 1.2274995416694102e-06, "loss": 0.3991, "step": 23729 }, { "epoch": 0.8454317117052924, "grad_norm": 1.4911148496391542, "learning_rate": 1.2269456715703554e-06, "loss": 0.4816, "step": 23730 }, { "epoch": 0.8454673388317865, "grad_norm": 1.8365125549718788, "learning_rate": 1.2263919182906514e-06, "loss": 0.3579, "step": 23731 }, { "epoch": 0.8455029659582807, "grad_norm": 1.264446657281761, "learning_rate": 1.2258382818376735e-06, "loss": 0.4229, "step": 23732 }, { "epoch": 0.8455385930847747, "grad_norm": 1.892441579177308, "learning_rate": 1.2252847622187902e-06, "loss": 0.4911, "step": 23733 }, { "epoch": 0.8455742202112688, "grad_norm": 1.516997182763921, "learning_rate": 1.2247313594413745e-06, "loss": 0.3874, "step": 23734 }, { "epoch": 0.845609847337763, "grad_norm": 1.607704954228983, "learning_rate": 1.2241780735127927e-06, "loss": 0.4641, "step": 23735 }, { "epoch": 0.8456454744642571, "grad_norm": 1.9541974789789482, "learning_rate": 1.2236249044404137e-06, "loss": 0.5516, "step": 23736 }, { "epoch": 0.8456811015907512, "grad_norm": 1.9861114411136713, "learning_rate": 1.2230718522316031e-06, "loss": 0.3947, "step": 23737 }, { "epoch": 0.8457167287172453, "grad_norm": 1.2375033414146066, "learning_rate": 1.222518916893728e-06, "loss": 0.3517, "step": 23738 }, { "epoch": 0.8457523558437394, "grad_norm": 1.6524516498430464, "learning_rate": 1.2219660984341463e-06, "loss": 0.4682, "step": 23739 }, { "epoch": 0.8457879829702335, "grad_norm": 1.4366612299116757, "learning_rate": 1.2214133968602238e-06, "loss": 0.465, "step": 23740 }, { "epoch": 0.8458236100967277, "grad_norm": 1.5883815734718634, "learning_rate": 1.2208608121793164e-06, "loss": 0.5197, "step": 23741 }, { "epoch": 0.8458592372232218, "grad_norm": 1.6921730704458078, "learning_rate": 1.2203083443987818e-06, "loss": 0.5078, "step": 23742 }, { "epoch": 0.8458948643497158, "grad_norm": 2.6685484788153198, "learning_rate": 1.2197559935259796e-06, "loss": 0.4349, "step": 23743 }, { "epoch": 0.84593049147621, "grad_norm": 1.5012432078006386, "learning_rate": 1.2192037595682626e-06, "loss": 0.3982, "step": 23744 }, { "epoch": 0.8459661186027041, "grad_norm": 1.348621328842103, "learning_rate": 1.2186516425329874e-06, "loss": 0.3972, "step": 23745 }, { "epoch": 0.8460017457291982, "grad_norm": 1.808488539693235, "learning_rate": 1.2180996424275016e-06, "loss": 0.5039, "step": 23746 }, { "epoch": 0.8460373728556924, "grad_norm": 1.5835648527549122, "learning_rate": 1.217547759259159e-06, "loss": 0.3964, "step": 23747 }, { "epoch": 0.8460729999821864, "grad_norm": 1.4254385174430209, "learning_rate": 1.2169959930353049e-06, "loss": 0.4183, "step": 23748 }, { "epoch": 0.8461086271086805, "grad_norm": 1.6265850057446243, "learning_rate": 1.2164443437632901e-06, "loss": 0.4465, "step": 23749 }, { "epoch": 0.8461442542351747, "grad_norm": 1.25280361676664, "learning_rate": 1.2158928114504552e-06, "loss": 0.3793, "step": 23750 }, { "epoch": 0.8461798813616688, "grad_norm": 1.2942339023027571, "learning_rate": 1.2153413961041483e-06, "loss": 0.3721, "step": 23751 }, { "epoch": 0.8462155084881628, "grad_norm": 1.6688251225780577, "learning_rate": 1.2147900977317096e-06, "loss": 0.5151, "step": 23752 }, { "epoch": 0.846251135614657, "grad_norm": 1.4036053029832813, "learning_rate": 1.2142389163404821e-06, "loss": 0.3834, "step": 23753 }, { "epoch": 0.8462867627411511, "grad_norm": 1.4671782411340386, "learning_rate": 1.213687851937807e-06, "loss": 0.4956, "step": 23754 }, { "epoch": 0.8463223898676452, "grad_norm": 1.5856364185795584, "learning_rate": 1.2131369045310159e-06, "loss": 0.4541, "step": 23755 }, { "epoch": 0.8463580169941394, "grad_norm": 1.529838016145333, "learning_rate": 1.212586074127452e-06, "loss": 0.5214, "step": 23756 }, { "epoch": 0.8463936441206334, "grad_norm": 1.4372519013440879, "learning_rate": 1.2120353607344438e-06, "loss": 0.3978, "step": 23757 }, { "epoch": 0.8464292712471275, "grad_norm": 1.1868882692327953, "learning_rate": 1.2114847643593264e-06, "loss": 0.3992, "step": 23758 }, { "epoch": 0.8464648983736217, "grad_norm": 1.8691042745122957, "learning_rate": 1.210934285009433e-06, "loss": 0.4917, "step": 23759 }, { "epoch": 0.8465005255001158, "grad_norm": 1.5700924939500858, "learning_rate": 1.2103839226920954e-06, "loss": 0.3951, "step": 23760 }, { "epoch": 0.8465361526266099, "grad_norm": 1.314464600826151, "learning_rate": 1.209833677414637e-06, "loss": 0.3689, "step": 23761 }, { "epoch": 0.846571779753104, "grad_norm": 2.0494021210062483, "learning_rate": 1.209283549184389e-06, "loss": 0.5547, "step": 23762 }, { "epoch": 0.8466074068795981, "grad_norm": 1.372650173640056, "learning_rate": 1.2087335380086728e-06, "loss": 0.3945, "step": 23763 }, { "epoch": 0.8466430340060922, "grad_norm": 1.4025972048614852, "learning_rate": 1.2081836438948147e-06, "loss": 0.415, "step": 23764 }, { "epoch": 0.8466786611325864, "grad_norm": 1.388431006991934, "learning_rate": 1.2076338668501374e-06, "loss": 0.4914, "step": 23765 }, { "epoch": 0.8467142882590805, "grad_norm": 1.8938575447031991, "learning_rate": 1.2070842068819599e-06, "loss": 0.5644, "step": 23766 }, { "epoch": 0.8467499153855745, "grad_norm": 1.3427976961904196, "learning_rate": 1.2065346639976016e-06, "loss": 0.3993, "step": 23767 }, { "epoch": 0.8467855425120687, "grad_norm": 1.5023227791199236, "learning_rate": 1.2059852382043802e-06, "loss": 0.4215, "step": 23768 }, { "epoch": 0.8468211696385628, "grad_norm": 1.5278168647000732, "learning_rate": 1.2054359295096141e-06, "loss": 0.409, "step": 23769 }, { "epoch": 0.8468567967650569, "grad_norm": 1.2203221574430714, "learning_rate": 1.2048867379206142e-06, "loss": 0.3505, "step": 23770 }, { "epoch": 0.8468924238915511, "grad_norm": 1.5859827944165201, "learning_rate": 1.2043376634446967e-06, "loss": 0.4349, "step": 23771 }, { "epoch": 0.8469280510180451, "grad_norm": 1.3422850946581566, "learning_rate": 1.2037887060891683e-06, "loss": 0.4043, "step": 23772 }, { "epoch": 0.8469636781445392, "grad_norm": 1.755838269619522, "learning_rate": 1.203239865861342e-06, "loss": 0.4793, "step": 23773 }, { "epoch": 0.8469993052710334, "grad_norm": 1.226464232442988, "learning_rate": 1.202691142768525e-06, "loss": 0.407, "step": 23774 }, { "epoch": 0.8470349323975275, "grad_norm": 1.3905084592186534, "learning_rate": 1.202142536818025e-06, "loss": 0.4354, "step": 23775 }, { "epoch": 0.8470705595240215, "grad_norm": 1.4734420597318199, "learning_rate": 1.2015940480171485e-06, "loss": 0.4705, "step": 23776 }, { "epoch": 0.8471061866505157, "grad_norm": 1.500314627120158, "learning_rate": 1.2010456763731948e-06, "loss": 0.4099, "step": 23777 }, { "epoch": 0.8471418137770098, "grad_norm": 1.561611767846821, "learning_rate": 1.2004974218934695e-06, "loss": 0.3858, "step": 23778 }, { "epoch": 0.8471774409035039, "grad_norm": 1.6391344414503777, "learning_rate": 1.1999492845852701e-06, "loss": 0.574, "step": 23779 }, { "epoch": 0.8472130680299981, "grad_norm": 1.3993110010449912, "learning_rate": 1.1994012644558984e-06, "loss": 0.4838, "step": 23780 }, { "epoch": 0.8472486951564921, "grad_norm": 1.4771250026319143, "learning_rate": 1.1988533615126463e-06, "loss": 0.4373, "step": 23781 }, { "epoch": 0.8472843222829862, "grad_norm": 2.2919404537923525, "learning_rate": 1.198305575762817e-06, "loss": 0.5291, "step": 23782 }, { "epoch": 0.8473199494094804, "grad_norm": 1.2867693912632472, "learning_rate": 1.1977579072137001e-06, "loss": 0.3946, "step": 23783 }, { "epoch": 0.8473555765359745, "grad_norm": 1.249801683320352, "learning_rate": 1.19721035587259e-06, "loss": 0.4215, "step": 23784 }, { "epoch": 0.8473912036624686, "grad_norm": 1.5700747449567587, "learning_rate": 1.1966629217467751e-06, "loss": 0.4161, "step": 23785 }, { "epoch": 0.8474268307889627, "grad_norm": 1.2760856335536686, "learning_rate": 1.1961156048435484e-06, "loss": 0.4262, "step": 23786 }, { "epoch": 0.8474624579154568, "grad_norm": 1.5089825932398262, "learning_rate": 1.1955684051701944e-06, "loss": 0.4716, "step": 23787 }, { "epoch": 0.8474980850419509, "grad_norm": 1.286945196568718, "learning_rate": 1.1950213227340013e-06, "loss": 0.4032, "step": 23788 }, { "epoch": 0.8475337121684451, "grad_norm": 1.5477308515156116, "learning_rate": 1.1944743575422523e-06, "loss": 0.546, "step": 23789 }, { "epoch": 0.8475693392949392, "grad_norm": 1.3216407661227418, "learning_rate": 1.1939275096022329e-06, "loss": 0.4407, "step": 23790 }, { "epoch": 0.8476049664214332, "grad_norm": 1.232237471765541, "learning_rate": 1.1933807789212259e-06, "loss": 0.4236, "step": 23791 }, { "epoch": 0.8476405935479274, "grad_norm": 1.3934275848582467, "learning_rate": 1.1928341655065068e-06, "loss": 0.3878, "step": 23792 }, { "epoch": 0.8476762206744215, "grad_norm": 1.4837553952315217, "learning_rate": 1.1922876693653584e-06, "loss": 0.3773, "step": 23793 }, { "epoch": 0.8477118478009156, "grad_norm": 1.6099477208640456, "learning_rate": 1.191741290505055e-06, "loss": 0.5138, "step": 23794 }, { "epoch": 0.8477474749274098, "grad_norm": 1.287077756457178, "learning_rate": 1.1911950289328722e-06, "loss": 0.3808, "step": 23795 }, { "epoch": 0.8477831020539038, "grad_norm": 1.538687365223874, "learning_rate": 1.1906488846560837e-06, "loss": 0.5413, "step": 23796 }, { "epoch": 0.8478187291803979, "grad_norm": 1.8956493779666093, "learning_rate": 1.1901028576819663e-06, "loss": 0.5505, "step": 23797 }, { "epoch": 0.8478543563068921, "grad_norm": 1.465866323869279, "learning_rate": 1.1895569480177838e-06, "loss": 0.5952, "step": 23798 }, { "epoch": 0.8478899834333862, "grad_norm": 1.1093569972975343, "learning_rate": 1.1890111556708095e-06, "loss": 0.3955, "step": 23799 }, { "epoch": 0.8479256105598802, "grad_norm": 1.3660442551989902, "learning_rate": 1.188465480648312e-06, "loss": 0.3691, "step": 23800 }, { "epoch": 0.8479612376863744, "grad_norm": 1.3794706577677416, "learning_rate": 1.1879199229575533e-06, "loss": 0.3658, "step": 23801 }, { "epoch": 0.8479968648128685, "grad_norm": 1.2734471022076805, "learning_rate": 1.187374482605802e-06, "loss": 0.377, "step": 23802 }, { "epoch": 0.8480324919393626, "grad_norm": 1.301391384991389, "learning_rate": 1.1868291596003179e-06, "loss": 0.4803, "step": 23803 }, { "epoch": 0.8480681190658568, "grad_norm": 1.7259560922785868, "learning_rate": 1.1862839539483628e-06, "loss": 0.6565, "step": 23804 }, { "epoch": 0.8481037461923508, "grad_norm": 1.325900233481851, "learning_rate": 1.1857388656571978e-06, "loss": 0.4193, "step": 23805 }, { "epoch": 0.8481393733188449, "grad_norm": 1.339970160197695, "learning_rate": 1.1851938947340824e-06, "loss": 0.4255, "step": 23806 }, { "epoch": 0.8481750004453391, "grad_norm": 1.3129710042301674, "learning_rate": 1.1846490411862689e-06, "loss": 0.4781, "step": 23807 }, { "epoch": 0.8482106275718332, "grad_norm": 1.403778583020441, "learning_rate": 1.1841043050210177e-06, "loss": 0.5048, "step": 23808 }, { "epoch": 0.8482462546983273, "grad_norm": 1.1576718266657595, "learning_rate": 1.1835596862455767e-06, "loss": 0.3497, "step": 23809 }, { "epoch": 0.8482818818248214, "grad_norm": 1.395475573298136, "learning_rate": 1.1830151848672022e-06, "loss": 0.36, "step": 23810 }, { "epoch": 0.8483175089513155, "grad_norm": 1.404659535393034, "learning_rate": 1.1824708008931419e-06, "loss": 0.4906, "step": 23811 }, { "epoch": 0.8483531360778096, "grad_norm": 1.7398811737516213, "learning_rate": 1.1819265343306463e-06, "loss": 0.5201, "step": 23812 }, { "epoch": 0.8483887632043038, "grad_norm": 1.2262833410273795, "learning_rate": 1.1813823851869656e-06, "loss": 0.405, "step": 23813 }, { "epoch": 0.8484243903307979, "grad_norm": 1.1843551504368937, "learning_rate": 1.1808383534693391e-06, "loss": 0.4278, "step": 23814 }, { "epoch": 0.8484600174572919, "grad_norm": 1.4652469219678732, "learning_rate": 1.1802944391850157e-06, "loss": 0.4338, "step": 23815 }, { "epoch": 0.8484956445837861, "grad_norm": 1.5861629399415416, "learning_rate": 1.1797506423412353e-06, "loss": 0.5473, "step": 23816 }, { "epoch": 0.8485312717102802, "grad_norm": 1.9740456849242063, "learning_rate": 1.1792069629452419e-06, "loss": 0.4853, "step": 23817 }, { "epoch": 0.8485668988367743, "grad_norm": 1.5440921489957655, "learning_rate": 1.178663401004272e-06, "loss": 0.4208, "step": 23818 }, { "epoch": 0.8486025259632685, "grad_norm": 1.591497090121232, "learning_rate": 1.178119956525563e-06, "loss": 0.4283, "step": 23819 }, { "epoch": 0.8486381530897625, "grad_norm": 1.2845309288368632, "learning_rate": 1.1775766295163538e-06, "loss": 0.4174, "step": 23820 }, { "epoch": 0.8486737802162566, "grad_norm": 1.5251747382308978, "learning_rate": 1.1770334199838784e-06, "loss": 0.495, "step": 23821 }, { "epoch": 0.8487094073427508, "grad_norm": 1.333544205989912, "learning_rate": 1.1764903279353711e-06, "loss": 0.3567, "step": 23822 }, { "epoch": 0.8487450344692449, "grad_norm": 2.104533436527174, "learning_rate": 1.175947353378062e-06, "loss": 0.5078, "step": 23823 }, { "epoch": 0.848780661595739, "grad_norm": 1.3815543849952472, "learning_rate": 1.1754044963191824e-06, "loss": 0.4496, "step": 23824 }, { "epoch": 0.8488162887222331, "grad_norm": 1.2813064032852146, "learning_rate": 1.174861756765957e-06, "loss": 0.4219, "step": 23825 }, { "epoch": 0.8488519158487272, "grad_norm": 1.1548925812299018, "learning_rate": 1.1743191347256178e-06, "loss": 0.4146, "step": 23826 }, { "epoch": 0.8488875429752213, "grad_norm": 1.5613916713671996, "learning_rate": 1.1737766302053867e-06, "loss": 0.533, "step": 23827 }, { "epoch": 0.8489231701017155, "grad_norm": 1.5474627615516074, "learning_rate": 1.173234243212491e-06, "loss": 0.4728, "step": 23828 }, { "epoch": 0.8489587972282095, "grad_norm": 1.639948021897988, "learning_rate": 1.1726919737541499e-06, "loss": 0.4838, "step": 23829 }, { "epoch": 0.8489944243547036, "grad_norm": 1.728774088799548, "learning_rate": 1.172149821837586e-06, "loss": 0.5653, "step": 23830 }, { "epoch": 0.8490300514811978, "grad_norm": 1.2103477732057313, "learning_rate": 1.171607787470016e-06, "loss": 0.3929, "step": 23831 }, { "epoch": 0.8490656786076919, "grad_norm": 1.5726968532729655, "learning_rate": 1.1710658706586587e-06, "loss": 0.3844, "step": 23832 }, { "epoch": 0.849101305734186, "grad_norm": 1.4570959636912446, "learning_rate": 1.1705240714107301e-06, "loss": 0.4197, "step": 23833 }, { "epoch": 0.8491369328606801, "grad_norm": 1.7278843357138043, "learning_rate": 1.169982389733445e-06, "loss": 0.4706, "step": 23834 }, { "epoch": 0.8491725599871742, "grad_norm": 1.7066885971856316, "learning_rate": 1.1694408256340196e-06, "loss": 0.4315, "step": 23835 }, { "epoch": 0.8492081871136683, "grad_norm": 1.5485877962910264, "learning_rate": 1.168899379119658e-06, "loss": 0.4159, "step": 23836 }, { "epoch": 0.8492438142401625, "grad_norm": 1.644889169247175, "learning_rate": 1.1683580501975766e-06, "loss": 0.5903, "step": 23837 }, { "epoch": 0.8492794413666566, "grad_norm": 1.4876377019245912, "learning_rate": 1.1678168388749788e-06, "loss": 0.5011, "step": 23838 }, { "epoch": 0.8493150684931506, "grad_norm": 1.5524829147241581, "learning_rate": 1.1672757451590756e-06, "loss": 0.5304, "step": 23839 }, { "epoch": 0.8493506956196448, "grad_norm": 1.3018746797250003, "learning_rate": 1.1667347690570663e-06, "loss": 0.4021, "step": 23840 }, { "epoch": 0.8493863227461389, "grad_norm": 1.3131206759908902, "learning_rate": 1.1661939105761576e-06, "loss": 0.4256, "step": 23841 }, { "epoch": 0.849421949872633, "grad_norm": 1.29848716002744, "learning_rate": 1.1656531697235529e-06, "loss": 0.3663, "step": 23842 }, { "epoch": 0.8494575769991272, "grad_norm": 1.3708555709792976, "learning_rate": 1.1651125465064516e-06, "loss": 0.4641, "step": 23843 }, { "epoch": 0.8494932041256212, "grad_norm": 1.3564068603306128, "learning_rate": 1.1645720409320505e-06, "loss": 0.4096, "step": 23844 }, { "epoch": 0.8495288312521153, "grad_norm": 1.4360984963683021, "learning_rate": 1.1640316530075491e-06, "loss": 0.4936, "step": 23845 }, { "epoch": 0.8495644583786095, "grad_norm": 1.335858171108477, "learning_rate": 1.163491382740143e-06, "loss": 0.3987, "step": 23846 }, { "epoch": 0.8496000855051036, "grad_norm": 1.7611351180062462, "learning_rate": 1.162951230137024e-06, "loss": 0.5021, "step": 23847 }, { "epoch": 0.8496357126315977, "grad_norm": 1.307290008339028, "learning_rate": 1.1624111952053862e-06, "loss": 0.4493, "step": 23848 }, { "epoch": 0.8496713397580918, "grad_norm": 1.6885222994003704, "learning_rate": 1.1618712779524199e-06, "loss": 0.499, "step": 23849 }, { "epoch": 0.8497069668845859, "grad_norm": 1.4832340366289378, "learning_rate": 1.161331478385318e-06, "loss": 0.401, "step": 23850 }, { "epoch": 0.84974259401108, "grad_norm": 1.190696743042029, "learning_rate": 1.1607917965112626e-06, "loss": 0.3962, "step": 23851 }, { "epoch": 0.8497782211375742, "grad_norm": 1.219443512969425, "learning_rate": 1.1602522323374455e-06, "loss": 0.3788, "step": 23852 }, { "epoch": 0.8498138482640683, "grad_norm": 1.649482628850766, "learning_rate": 1.1597127858710455e-06, "loss": 0.4036, "step": 23853 }, { "epoch": 0.8498494753905623, "grad_norm": 1.3323131523962657, "learning_rate": 1.1591734571192515e-06, "loss": 0.3858, "step": 23854 }, { "epoch": 0.8498851025170565, "grad_norm": 1.6398369077252297, "learning_rate": 1.1586342460892397e-06, "loss": 0.4963, "step": 23855 }, { "epoch": 0.8499207296435506, "grad_norm": 1.2142981222542664, "learning_rate": 1.1580951527881902e-06, "loss": 0.4347, "step": 23856 }, { "epoch": 0.8499563567700447, "grad_norm": 1.4508924687929592, "learning_rate": 1.157556177223289e-06, "loss": 0.4673, "step": 23857 }, { "epoch": 0.8499919838965388, "grad_norm": 1.5046558670973607, "learning_rate": 1.1570173194017053e-06, "loss": 0.4461, "step": 23858 }, { "epoch": 0.8500276110230329, "grad_norm": 1.7186844606131015, "learning_rate": 1.1564785793306187e-06, "loss": 0.5069, "step": 23859 }, { "epoch": 0.850063238149527, "grad_norm": 1.2923432892756914, "learning_rate": 1.1559399570171992e-06, "loss": 0.4414, "step": 23860 }, { "epoch": 0.8500988652760212, "grad_norm": 1.45427291301565, "learning_rate": 1.155401452468623e-06, "loss": 0.4231, "step": 23861 }, { "epoch": 0.8501344924025153, "grad_norm": 1.4016662192720337, "learning_rate": 1.1548630656920567e-06, "loss": 0.4006, "step": 23862 }, { "epoch": 0.8501701195290093, "grad_norm": 1.11973621372035, "learning_rate": 1.1543247966946702e-06, "loss": 0.3779, "step": 23863 }, { "epoch": 0.8502057466555035, "grad_norm": 1.2234710383961394, "learning_rate": 1.1537866454836321e-06, "loss": 0.4349, "step": 23864 }, { "epoch": 0.8502413737819976, "grad_norm": 3.591378775853382, "learning_rate": 1.1532486120661103e-06, "loss": 0.5271, "step": 23865 }, { "epoch": 0.8502770009084917, "grad_norm": 1.6369872822006961, "learning_rate": 1.152710696449264e-06, "loss": 0.4697, "step": 23866 }, { "epoch": 0.8503126280349859, "grad_norm": 1.861351410752151, "learning_rate": 1.1521728986402613e-06, "loss": 0.4446, "step": 23867 }, { "epoch": 0.8503482551614799, "grad_norm": 2.0161277946211293, "learning_rate": 1.1516352186462588e-06, "loss": 0.4032, "step": 23868 }, { "epoch": 0.850383882287974, "grad_norm": 1.2402119728872285, "learning_rate": 1.1510976564744181e-06, "loss": 0.3918, "step": 23869 }, { "epoch": 0.8504195094144682, "grad_norm": 1.2810376649915298, "learning_rate": 1.1505602121318971e-06, "loss": 0.3881, "step": 23870 }, { "epoch": 0.8504551365409623, "grad_norm": 1.4086146447831744, "learning_rate": 1.1500228856258522e-06, "loss": 0.4107, "step": 23871 }, { "epoch": 0.8504907636674564, "grad_norm": 1.4128762751361488, "learning_rate": 1.14948567696344e-06, "loss": 0.413, "step": 23872 }, { "epoch": 0.8505263907939505, "grad_norm": 1.8137177451628927, "learning_rate": 1.14894858615181e-06, "loss": 0.565, "step": 23873 }, { "epoch": 0.8505620179204446, "grad_norm": 2.0821729109895823, "learning_rate": 1.1484116131981182e-06, "loss": 0.5642, "step": 23874 }, { "epoch": 0.8505976450469387, "grad_norm": 1.5734259876177459, "learning_rate": 1.1478747581095107e-06, "loss": 0.4412, "step": 23875 }, { "epoch": 0.8506332721734329, "grad_norm": 1.441519154990192, "learning_rate": 1.1473380208931406e-06, "loss": 0.4202, "step": 23876 }, { "epoch": 0.850668899299927, "grad_norm": 1.4155230615065213, "learning_rate": 1.1468014015561503e-06, "loss": 0.4507, "step": 23877 }, { "epoch": 0.850704526426421, "grad_norm": 1.5568002468980748, "learning_rate": 1.146264900105687e-06, "loss": 0.4127, "step": 23878 }, { "epoch": 0.8507401535529152, "grad_norm": 1.5601763618745175, "learning_rate": 1.1457285165488951e-06, "loss": 0.5115, "step": 23879 }, { "epoch": 0.8507757806794093, "grad_norm": 1.4510818702678656, "learning_rate": 1.145192250892917e-06, "loss": 0.4056, "step": 23880 }, { "epoch": 0.8508114078059034, "grad_norm": 1.693201594512594, "learning_rate": 1.1446561031448944e-06, "loss": 0.4188, "step": 23881 }, { "epoch": 0.8508470349323976, "grad_norm": 1.501946220058774, "learning_rate": 1.144120073311964e-06, "loss": 0.5434, "step": 23882 }, { "epoch": 0.8508826620588916, "grad_norm": 1.6411440376758697, "learning_rate": 1.1435841614012666e-06, "loss": 0.5218, "step": 23883 }, { "epoch": 0.8509182891853857, "grad_norm": 1.4391536108231342, "learning_rate": 1.1430483674199345e-06, "loss": 0.4432, "step": 23884 }, { "epoch": 0.8509539163118799, "grad_norm": 1.6090280095789253, "learning_rate": 1.1425126913751029e-06, "loss": 0.4069, "step": 23885 }, { "epoch": 0.850989543438374, "grad_norm": 1.2533659656948068, "learning_rate": 1.1419771332739072e-06, "loss": 0.4014, "step": 23886 }, { "epoch": 0.851025170564868, "grad_norm": 1.3542339239044223, "learning_rate": 1.1414416931234785e-06, "loss": 0.4567, "step": 23887 }, { "epoch": 0.8510607976913622, "grad_norm": 1.7364089844213506, "learning_rate": 1.1409063709309442e-06, "loss": 0.5746, "step": 23888 }, { "epoch": 0.8510964248178563, "grad_norm": 1.4118527172562563, "learning_rate": 1.1403711667034356e-06, "loss": 0.4783, "step": 23889 }, { "epoch": 0.8511320519443504, "grad_norm": 1.4461023971812001, "learning_rate": 1.1398360804480746e-06, "loss": 0.4171, "step": 23890 }, { "epoch": 0.8511676790708446, "grad_norm": 1.6079845764555598, "learning_rate": 1.1393011121719899e-06, "loss": 0.5352, "step": 23891 }, { "epoch": 0.8512033061973386, "grad_norm": 1.2271495119944584, "learning_rate": 1.1387662618823058e-06, "loss": 0.4089, "step": 23892 }, { "epoch": 0.8512389333238327, "grad_norm": 1.6819107478921127, "learning_rate": 1.1382315295861401e-06, "loss": 0.4225, "step": 23893 }, { "epoch": 0.8512745604503269, "grad_norm": 1.3202535718302946, "learning_rate": 1.137696915290617e-06, "loss": 0.4461, "step": 23894 }, { "epoch": 0.851310187576821, "grad_norm": 1.2222118959520853, "learning_rate": 1.137162419002853e-06, "loss": 0.4071, "step": 23895 }, { "epoch": 0.851345814703315, "grad_norm": 2.0305384275207645, "learning_rate": 1.1366280407299678e-06, "loss": 0.4286, "step": 23896 }, { "epoch": 0.8513814418298092, "grad_norm": 1.6918733721610169, "learning_rate": 1.136093780479074e-06, "loss": 0.4685, "step": 23897 }, { "epoch": 0.8514170689563033, "grad_norm": 1.2241362187891547, "learning_rate": 1.1355596382572887e-06, "loss": 0.3871, "step": 23898 }, { "epoch": 0.8514526960827974, "grad_norm": 1.3254937257585204, "learning_rate": 1.1350256140717196e-06, "loss": 0.4049, "step": 23899 }, { "epoch": 0.8514883232092916, "grad_norm": 1.3465866293278095, "learning_rate": 1.1344917079294827e-06, "loss": 0.4136, "step": 23900 }, { "epoch": 0.8515239503357857, "grad_norm": 1.7032558994471254, "learning_rate": 1.1339579198376838e-06, "loss": 0.4714, "step": 23901 }, { "epoch": 0.8515595774622797, "grad_norm": 1.2240416559574934, "learning_rate": 1.1334242498034332e-06, "loss": 0.3807, "step": 23902 }, { "epoch": 0.8515952045887739, "grad_norm": 1.487992641549122, "learning_rate": 1.1328906978338372e-06, "loss": 0.4789, "step": 23903 }, { "epoch": 0.851630831715268, "grad_norm": 1.73310611261115, "learning_rate": 1.1323572639359981e-06, "loss": 0.5418, "step": 23904 }, { "epoch": 0.8516664588417621, "grad_norm": 1.5001355653824657, "learning_rate": 1.1318239481170224e-06, "loss": 0.534, "step": 23905 }, { "epoch": 0.8517020859682563, "grad_norm": 1.501740948773232, "learning_rate": 1.1312907503840064e-06, "loss": 0.4851, "step": 23906 }, { "epoch": 0.8517377130947503, "grad_norm": 1.888241965812904, "learning_rate": 1.130757670744055e-06, "loss": 0.5108, "step": 23907 }, { "epoch": 0.8517733402212444, "grad_norm": 1.4930397260761843, "learning_rate": 1.1302247092042607e-06, "loss": 0.5421, "step": 23908 }, { "epoch": 0.8518089673477386, "grad_norm": 1.148975739843306, "learning_rate": 1.1296918657717281e-06, "loss": 0.4171, "step": 23909 }, { "epoch": 0.8518445944742327, "grad_norm": 1.288377284245601, "learning_rate": 1.1291591404535463e-06, "loss": 0.4231, "step": 23910 }, { "epoch": 0.8518802216007267, "grad_norm": 1.214096136975192, "learning_rate": 1.1286265332568124e-06, "loss": 0.4059, "step": 23911 }, { "epoch": 0.8519158487272209, "grad_norm": 1.470242450727635, "learning_rate": 1.1280940441886156e-06, "loss": 0.4711, "step": 23912 }, { "epoch": 0.851951475853715, "grad_norm": 1.2922420557313847, "learning_rate": 1.127561673256049e-06, "loss": 0.415, "step": 23913 }, { "epoch": 0.8519871029802091, "grad_norm": 1.59600632948843, "learning_rate": 1.127029420466199e-06, "loss": 0.4214, "step": 23914 }, { "epoch": 0.8520227301067033, "grad_norm": 1.9439498855745636, "learning_rate": 1.1264972858261535e-06, "loss": 0.5712, "step": 23915 }, { "epoch": 0.8520583572331973, "grad_norm": 1.5439289876208304, "learning_rate": 1.125965269342999e-06, "loss": 0.4905, "step": 23916 }, { "epoch": 0.8520939843596914, "grad_norm": 2.1103554280485906, "learning_rate": 1.1254333710238197e-06, "loss": 0.5324, "step": 23917 }, { "epoch": 0.8521296114861856, "grad_norm": 1.2995288479051037, "learning_rate": 1.1249015908757e-06, "loss": 0.382, "step": 23918 }, { "epoch": 0.8521652386126797, "grad_norm": 1.41733832143283, "learning_rate": 1.1243699289057164e-06, "loss": 0.4403, "step": 23919 }, { "epoch": 0.8522008657391738, "grad_norm": 1.5806235381303024, "learning_rate": 1.1238383851209533e-06, "loss": 0.5362, "step": 23920 }, { "epoch": 0.8522364928656679, "grad_norm": 1.478544226111353, "learning_rate": 1.123306959528484e-06, "loss": 0.5458, "step": 23921 }, { "epoch": 0.852272119992162, "grad_norm": 1.971291476869298, "learning_rate": 1.1227756521353873e-06, "loss": 0.487, "step": 23922 }, { "epoch": 0.8523077471186561, "grad_norm": 1.7700961503406316, "learning_rate": 1.1222444629487372e-06, "loss": 0.5335, "step": 23923 }, { "epoch": 0.8523433742451503, "grad_norm": 1.3271709373613199, "learning_rate": 1.1217133919756096e-06, "loss": 0.4233, "step": 23924 }, { "epoch": 0.8523790013716444, "grad_norm": 1.4640803499592898, "learning_rate": 1.121182439223072e-06, "loss": 0.4852, "step": 23925 }, { "epoch": 0.8524146284981385, "grad_norm": 1.393662010224438, "learning_rate": 1.1206516046981952e-06, "loss": 0.3895, "step": 23926 }, { "epoch": 0.8524502556246326, "grad_norm": 1.2559998860201689, "learning_rate": 1.1201208884080516e-06, "loss": 0.4133, "step": 23927 }, { "epoch": 0.8524858827511267, "grad_norm": 1.3623745541063585, "learning_rate": 1.1195902903597023e-06, "loss": 0.3598, "step": 23928 }, { "epoch": 0.8525215098776209, "grad_norm": 1.136035708136864, "learning_rate": 1.119059810560218e-06, "loss": 0.3972, "step": 23929 }, { "epoch": 0.852557137004115, "grad_norm": 1.4615598826844238, "learning_rate": 1.1185294490166564e-06, "loss": 0.4884, "step": 23930 }, { "epoch": 0.852592764130609, "grad_norm": 1.5163650319924118, "learning_rate": 1.1179992057360844e-06, "loss": 0.5137, "step": 23931 }, { "epoch": 0.8526283912571032, "grad_norm": 1.48776258494057, "learning_rate": 1.1174690807255605e-06, "loss": 0.38, "step": 23932 }, { "epoch": 0.8526640183835973, "grad_norm": 1.2628687318799863, "learning_rate": 1.116939073992146e-06, "loss": 0.418, "step": 23933 }, { "epoch": 0.8526996455100914, "grad_norm": 1.2448174696192722, "learning_rate": 1.116409185542896e-06, "loss": 0.4141, "step": 23934 }, { "epoch": 0.8527352726365856, "grad_norm": 1.4796420871694627, "learning_rate": 1.1158794153848674e-06, "loss": 0.592, "step": 23935 }, { "epoch": 0.8527708997630796, "grad_norm": 1.7411903772296444, "learning_rate": 1.1153497635251131e-06, "loss": 0.4918, "step": 23936 }, { "epoch": 0.8528065268895737, "grad_norm": 1.471230824891407, "learning_rate": 1.1148202299706857e-06, "loss": 0.4163, "step": 23937 }, { "epoch": 0.8528421540160679, "grad_norm": 1.3291426702207156, "learning_rate": 1.114290814728638e-06, "loss": 0.4065, "step": 23938 }, { "epoch": 0.852877781142562, "grad_norm": 1.5294672802055043, "learning_rate": 1.113761517806019e-06, "loss": 0.4103, "step": 23939 }, { "epoch": 0.852913408269056, "grad_norm": 1.5764733397126864, "learning_rate": 1.1132323392098777e-06, "loss": 0.5479, "step": 23940 }, { "epoch": 0.8529490353955502, "grad_norm": 1.5400016849576916, "learning_rate": 1.112703278947258e-06, "loss": 0.4751, "step": 23941 }, { "epoch": 0.8529846625220443, "grad_norm": 1.303652151328938, "learning_rate": 1.112174337025207e-06, "loss": 0.4123, "step": 23942 }, { "epoch": 0.8530202896485384, "grad_norm": 1.5903223965302113, "learning_rate": 1.1116455134507665e-06, "loss": 0.5591, "step": 23943 }, { "epoch": 0.8530559167750326, "grad_norm": 1.3270402669312726, "learning_rate": 1.1111168082309798e-06, "loss": 0.3991, "step": 23944 }, { "epoch": 0.8530915439015266, "grad_norm": 1.803169248940873, "learning_rate": 1.1105882213728814e-06, "loss": 0.4772, "step": 23945 }, { "epoch": 0.8531271710280207, "grad_norm": 1.2893221295461603, "learning_rate": 1.1100597528835177e-06, "loss": 0.3827, "step": 23946 }, { "epoch": 0.8531627981545149, "grad_norm": 1.165852819850651, "learning_rate": 1.109531402769921e-06, "loss": 0.4521, "step": 23947 }, { "epoch": 0.853198425281009, "grad_norm": 1.5028655948376126, "learning_rate": 1.1090031710391281e-06, "loss": 0.4266, "step": 23948 }, { "epoch": 0.853234052407503, "grad_norm": 1.516770605190683, "learning_rate": 1.108475057698174e-06, "loss": 0.5411, "step": 23949 }, { "epoch": 0.8532696795339972, "grad_norm": 1.4962784598947463, "learning_rate": 1.1079470627540866e-06, "loss": 0.5753, "step": 23950 }, { "epoch": 0.8533053066604913, "grad_norm": 1.3415302740717796, "learning_rate": 1.1074191862139016e-06, "loss": 0.4342, "step": 23951 }, { "epoch": 0.8533409337869854, "grad_norm": 1.232003128264486, "learning_rate": 1.1068914280846444e-06, "loss": 0.4446, "step": 23952 }, { "epoch": 0.8533765609134796, "grad_norm": 1.7831961544869093, "learning_rate": 1.1063637883733436e-06, "loss": 0.476, "step": 23953 }, { "epoch": 0.8534121880399737, "grad_norm": 1.5888900103359749, "learning_rate": 1.1058362670870248e-06, "loss": 0.5722, "step": 23954 }, { "epoch": 0.8534478151664677, "grad_norm": 1.4003300190757706, "learning_rate": 1.1053088642327147e-06, "loss": 0.4581, "step": 23955 }, { "epoch": 0.8534834422929619, "grad_norm": 1.5718325193385234, "learning_rate": 1.104781579817432e-06, "loss": 0.3928, "step": 23956 }, { "epoch": 0.853519069419456, "grad_norm": 1.3102334275086662, "learning_rate": 1.1042544138482026e-06, "loss": 0.4428, "step": 23957 }, { "epoch": 0.8535546965459501, "grad_norm": 1.3181544291982432, "learning_rate": 1.1037273663320402e-06, "loss": 0.3791, "step": 23958 }, { "epoch": 0.8535903236724443, "grad_norm": 1.5972380389739504, "learning_rate": 1.1032004372759674e-06, "loss": 0.4314, "step": 23959 }, { "epoch": 0.8536259507989383, "grad_norm": 1.6676528350099908, "learning_rate": 1.1026736266869987e-06, "loss": 0.4888, "step": 23960 }, { "epoch": 0.8536615779254324, "grad_norm": 1.6089274225925365, "learning_rate": 1.1021469345721492e-06, "loss": 0.5621, "step": 23961 }, { "epoch": 0.8536972050519266, "grad_norm": 1.5120447439689015, "learning_rate": 1.1016203609384347e-06, "loss": 0.5532, "step": 23962 }, { "epoch": 0.8537328321784207, "grad_norm": 1.3159716353489852, "learning_rate": 1.1010939057928626e-06, "loss": 0.3994, "step": 23963 }, { "epoch": 0.8537684593049147, "grad_norm": 1.7606376511333484, "learning_rate": 1.1005675691424477e-06, "loss": 0.5155, "step": 23964 }, { "epoch": 0.8538040864314089, "grad_norm": 1.5161281043881758, "learning_rate": 1.1000413509941932e-06, "loss": 0.463, "step": 23965 }, { "epoch": 0.853839713557903, "grad_norm": 1.3947195951071012, "learning_rate": 1.0995152513551111e-06, "loss": 0.4434, "step": 23966 }, { "epoch": 0.8538753406843971, "grad_norm": 1.4488962680378057, "learning_rate": 1.0989892702322025e-06, "loss": 0.4313, "step": 23967 }, { "epoch": 0.8539109678108913, "grad_norm": 1.6712198366843625, "learning_rate": 1.098463407632474e-06, "loss": 0.5249, "step": 23968 }, { "epoch": 0.8539465949373853, "grad_norm": 1.3891208968013276, "learning_rate": 1.0979376635629256e-06, "loss": 0.3774, "step": 23969 }, { "epoch": 0.8539822220638794, "grad_norm": 1.333632069643454, "learning_rate": 1.097412038030563e-06, "loss": 0.3874, "step": 23970 }, { "epoch": 0.8540178491903736, "grad_norm": 1.579557739038478, "learning_rate": 1.096886531042378e-06, "loss": 0.5124, "step": 23971 }, { "epoch": 0.8540534763168677, "grad_norm": 1.2785169687090252, "learning_rate": 1.0963611426053722e-06, "loss": 0.4119, "step": 23972 }, { "epoch": 0.8540891034433618, "grad_norm": 1.4268537054851802, "learning_rate": 1.0958358727265438e-06, "loss": 0.4708, "step": 23973 }, { "epoch": 0.8541247305698559, "grad_norm": 1.4274056935453594, "learning_rate": 1.0953107214128821e-06, "loss": 0.4991, "step": 23974 }, { "epoch": 0.85416035769635, "grad_norm": 1.5007567340904826, "learning_rate": 1.0947856886713814e-06, "loss": 0.4735, "step": 23975 }, { "epoch": 0.8541959848228441, "grad_norm": 1.7403715140938483, "learning_rate": 1.094260774509035e-06, "loss": 0.6055, "step": 23976 }, { "epoch": 0.8542316119493383, "grad_norm": 1.356470806288448, "learning_rate": 1.0937359789328317e-06, "loss": 0.3611, "step": 23977 }, { "epoch": 0.8542672390758324, "grad_norm": 1.476448240047751, "learning_rate": 1.0932113019497581e-06, "loss": 0.4231, "step": 23978 }, { "epoch": 0.8543028662023264, "grad_norm": 1.6665616524542317, "learning_rate": 1.0926867435668031e-06, "loss": 0.4798, "step": 23979 }, { "epoch": 0.8543384933288206, "grad_norm": 1.3251791666997887, "learning_rate": 1.092162303790948e-06, "loss": 0.4184, "step": 23980 }, { "epoch": 0.8543741204553147, "grad_norm": 1.6126704937994916, "learning_rate": 1.091637982629179e-06, "loss": 0.5102, "step": 23981 }, { "epoch": 0.8544097475818088, "grad_norm": 1.6439709744907667, "learning_rate": 1.0911137800884764e-06, "loss": 0.5224, "step": 23982 }, { "epoch": 0.854445374708303, "grad_norm": 1.6524869923916747, "learning_rate": 1.0905896961758179e-06, "loss": 0.5074, "step": 23983 }, { "epoch": 0.854481001834797, "grad_norm": 1.1864207896509058, "learning_rate": 1.0900657308981888e-06, "loss": 0.367, "step": 23984 }, { "epoch": 0.8545166289612911, "grad_norm": 1.5687211133980563, "learning_rate": 1.0895418842625605e-06, "loss": 0.4746, "step": 23985 }, { "epoch": 0.8545522560877853, "grad_norm": 1.330773963464693, "learning_rate": 1.0890181562759117e-06, "loss": 0.4137, "step": 23986 }, { "epoch": 0.8545878832142794, "grad_norm": 1.4618890719262498, "learning_rate": 1.0884945469452125e-06, "loss": 0.3783, "step": 23987 }, { "epoch": 0.8546235103407734, "grad_norm": 1.6112744704211155, "learning_rate": 1.0879710562774404e-06, "loss": 0.5299, "step": 23988 }, { "epoch": 0.8546591374672676, "grad_norm": 2.208918386088696, "learning_rate": 1.087447684279559e-06, "loss": 0.5912, "step": 23989 }, { "epoch": 0.8546947645937617, "grad_norm": 1.395780070463416, "learning_rate": 1.0869244309585425e-06, "loss": 0.3742, "step": 23990 }, { "epoch": 0.8547303917202558, "grad_norm": 1.8587579003918175, "learning_rate": 1.0864012963213566e-06, "loss": 0.5002, "step": 23991 }, { "epoch": 0.85476601884675, "grad_norm": 2.0823200214700437, "learning_rate": 1.08587828037497e-06, "loss": 0.5171, "step": 23992 }, { "epoch": 0.854801645973244, "grad_norm": 2.655295663427108, "learning_rate": 1.085355383126342e-06, "loss": 0.4481, "step": 23993 }, { "epoch": 0.8548372730997381, "grad_norm": 1.4518247374103481, "learning_rate": 1.084832604582441e-06, "loss": 0.4201, "step": 23994 }, { "epoch": 0.8548729002262323, "grad_norm": 1.543369331241899, "learning_rate": 1.0843099447502226e-06, "loss": 0.3311, "step": 23995 }, { "epoch": 0.8549085273527264, "grad_norm": 1.4199092761089107, "learning_rate": 1.083787403636649e-06, "loss": 0.3835, "step": 23996 }, { "epoch": 0.8549441544792205, "grad_norm": 1.304640433034588, "learning_rate": 1.0832649812486785e-06, "loss": 0.4382, "step": 23997 }, { "epoch": 0.8549797816057146, "grad_norm": 1.4046991081439488, "learning_rate": 1.0827426775932658e-06, "loss": 0.4204, "step": 23998 }, { "epoch": 0.8550154087322087, "grad_norm": 1.6726125916904075, "learning_rate": 1.0822204926773706e-06, "loss": 0.6127, "step": 23999 }, { "epoch": 0.8550510358587028, "grad_norm": 1.4508879325768236, "learning_rate": 1.081698426507941e-06, "loss": 0.3769, "step": 24000 }, { "epoch": 0.855086662985197, "grad_norm": 1.491259585047024, "learning_rate": 1.081176479091931e-06, "loss": 0.5522, "step": 24001 }, { "epoch": 0.855122290111691, "grad_norm": 1.7138684925214445, "learning_rate": 1.0806546504362892e-06, "loss": 0.4712, "step": 24002 }, { "epoch": 0.8551579172381851, "grad_norm": 1.205857530247718, "learning_rate": 1.0801329405479667e-06, "loss": 0.3883, "step": 24003 }, { "epoch": 0.8551935443646793, "grad_norm": 1.3966930831195092, "learning_rate": 1.0796113494339066e-06, "loss": 0.3799, "step": 24004 }, { "epoch": 0.8552291714911734, "grad_norm": 1.457232258240267, "learning_rate": 1.079089877101056e-06, "loss": 0.384, "step": 24005 }, { "epoch": 0.8552647986176675, "grad_norm": 1.4442212196357398, "learning_rate": 1.0785685235563592e-06, "loss": 0.4895, "step": 24006 }, { "epoch": 0.8553004257441617, "grad_norm": 1.2372806754888037, "learning_rate": 1.0780472888067583e-06, "loss": 0.3838, "step": 24007 }, { "epoch": 0.8553360528706557, "grad_norm": 1.549587598401101, "learning_rate": 1.077526172859197e-06, "loss": 0.4004, "step": 24008 }, { "epoch": 0.8553716799971498, "grad_norm": 1.6845637106733413, "learning_rate": 1.0770051757206078e-06, "loss": 0.4817, "step": 24009 }, { "epoch": 0.855407307123644, "grad_norm": 1.4048639279250081, "learning_rate": 1.0764842973979338e-06, "loss": 0.4208, "step": 24010 }, { "epoch": 0.8554429342501381, "grad_norm": 2.9411116525596457, "learning_rate": 1.075963537898106e-06, "loss": 0.5211, "step": 24011 }, { "epoch": 0.8554785613766321, "grad_norm": 1.884468769758892, "learning_rate": 1.0754428972280617e-06, "loss": 0.4209, "step": 24012 }, { "epoch": 0.8555141885031263, "grad_norm": 1.6137558827574008, "learning_rate": 1.0749223753947335e-06, "loss": 0.4307, "step": 24013 }, { "epoch": 0.8555498156296204, "grad_norm": 1.7174410759219074, "learning_rate": 1.0744019724050537e-06, "loss": 0.5503, "step": 24014 }, { "epoch": 0.8555854427561145, "grad_norm": 1.2367361377311847, "learning_rate": 1.0738816882659487e-06, "loss": 0.3815, "step": 24015 }, { "epoch": 0.8556210698826087, "grad_norm": 1.2825660169915167, "learning_rate": 1.0733615229843486e-06, "loss": 0.4117, "step": 24016 }, { "epoch": 0.8556566970091027, "grad_norm": 1.5922721331268657, "learning_rate": 1.072841476567179e-06, "loss": 0.5132, "step": 24017 }, { "epoch": 0.8556923241355968, "grad_norm": 1.1438955526560548, "learning_rate": 1.0723215490213635e-06, "loss": 0.3878, "step": 24018 }, { "epoch": 0.855727951262091, "grad_norm": 1.4494005792825455, "learning_rate": 1.0718017403538294e-06, "loss": 0.438, "step": 24019 }, { "epoch": 0.8557635783885851, "grad_norm": 1.7791355148793988, "learning_rate": 1.0712820505714916e-06, "loss": 0.4645, "step": 24020 }, { "epoch": 0.8557992055150792, "grad_norm": 1.2938730399827287, "learning_rate": 1.0707624796812776e-06, "loss": 0.3836, "step": 24021 }, { "epoch": 0.8558348326415733, "grad_norm": 1.3884741951255783, "learning_rate": 1.070243027690101e-06, "loss": 0.4285, "step": 24022 }, { "epoch": 0.8558704597680674, "grad_norm": 1.1991265031690628, "learning_rate": 1.0697236946048805e-06, "loss": 0.4194, "step": 24023 }, { "epoch": 0.8559060868945615, "grad_norm": 1.7341788524881898, "learning_rate": 1.0692044804325308e-06, "loss": 0.5003, "step": 24024 }, { "epoch": 0.8559417140210557, "grad_norm": 1.2904849981853144, "learning_rate": 1.0686853851799672e-06, "loss": 0.4245, "step": 24025 }, { "epoch": 0.8559773411475498, "grad_norm": 1.6512174753304694, "learning_rate": 1.0681664088540977e-06, "loss": 0.5225, "step": 24026 }, { "epoch": 0.8560129682740438, "grad_norm": 1.612988415579735, "learning_rate": 1.0676475514618356e-06, "loss": 0.4027, "step": 24027 }, { "epoch": 0.856048595400538, "grad_norm": 1.2411321895999166, "learning_rate": 1.0671288130100898e-06, "loss": 0.4051, "step": 24028 }, { "epoch": 0.8560842225270321, "grad_norm": 1.4224770105663995, "learning_rate": 1.066610193505767e-06, "loss": 0.4184, "step": 24029 }, { "epoch": 0.8561198496535262, "grad_norm": 1.1692177631643716, "learning_rate": 1.0660916929557763e-06, "loss": 0.3818, "step": 24030 }, { "epoch": 0.8561554767800204, "grad_norm": 1.665941056594407, "learning_rate": 1.0655733113670175e-06, "loss": 0.5572, "step": 24031 }, { "epoch": 0.8561911039065144, "grad_norm": 1.3735664157593215, "learning_rate": 1.0650550487463963e-06, "loss": 0.417, "step": 24032 }, { "epoch": 0.8562267310330085, "grad_norm": 1.800633217497544, "learning_rate": 1.0645369051008103e-06, "loss": 0.5855, "step": 24033 }, { "epoch": 0.8562623581595027, "grad_norm": 1.5358148966340162, "learning_rate": 1.0640188804371609e-06, "loss": 0.5727, "step": 24034 }, { "epoch": 0.8562979852859968, "grad_norm": 1.5249675823486502, "learning_rate": 1.063500974762346e-06, "loss": 0.4538, "step": 24035 }, { "epoch": 0.8563336124124908, "grad_norm": 1.5940259253613744, "learning_rate": 1.0629831880832642e-06, "loss": 0.452, "step": 24036 }, { "epoch": 0.856369239538985, "grad_norm": 2.0255295596054195, "learning_rate": 1.0624655204068056e-06, "loss": 0.409, "step": 24037 }, { "epoch": 0.8564048666654791, "grad_norm": 1.380697316973704, "learning_rate": 1.0619479717398674e-06, "loss": 0.4161, "step": 24038 }, { "epoch": 0.8564404937919732, "grad_norm": 1.537815542058978, "learning_rate": 1.0614305420893367e-06, "loss": 0.4773, "step": 24039 }, { "epoch": 0.8564761209184674, "grad_norm": 1.3890588924325, "learning_rate": 1.0609132314621095e-06, "loss": 0.4137, "step": 24040 }, { "epoch": 0.8565117480449614, "grad_norm": 1.5603281066635417, "learning_rate": 1.060396039865067e-06, "loss": 0.5188, "step": 24041 }, { "epoch": 0.8565473751714555, "grad_norm": 1.5008092107320523, "learning_rate": 1.0598789673051013e-06, "loss": 0.5543, "step": 24042 }, { "epoch": 0.8565830022979497, "grad_norm": 1.7336096917202215, "learning_rate": 1.0593620137890948e-06, "loss": 0.5196, "step": 24043 }, { "epoch": 0.8566186294244438, "grad_norm": 1.465396310223348, "learning_rate": 1.058845179323933e-06, "loss": 0.5674, "step": 24044 }, { "epoch": 0.8566542565509379, "grad_norm": 1.632104150300188, "learning_rate": 1.058328463916498e-06, "loss": 0.4948, "step": 24045 }, { "epoch": 0.856689883677432, "grad_norm": 1.355552102235732, "learning_rate": 1.0578118675736681e-06, "loss": 0.3884, "step": 24046 }, { "epoch": 0.8567255108039261, "grad_norm": 1.285193699051179, "learning_rate": 1.0572953903023254e-06, "loss": 0.3848, "step": 24047 }, { "epoch": 0.8567611379304202, "grad_norm": 1.5990446200759962, "learning_rate": 1.056779032109343e-06, "loss": 0.4757, "step": 24048 }, { "epoch": 0.8567967650569144, "grad_norm": 1.3224893563477, "learning_rate": 1.056262793001599e-06, "loss": 0.4052, "step": 24049 }, { "epoch": 0.8568323921834085, "grad_norm": 1.8066353731976073, "learning_rate": 1.0557466729859667e-06, "loss": 0.5444, "step": 24050 }, { "epoch": 0.8568680193099025, "grad_norm": 1.8258374106630924, "learning_rate": 1.0552306720693218e-06, "loss": 0.5815, "step": 24051 }, { "epoch": 0.8569036464363967, "grad_norm": 1.4218369778233981, "learning_rate": 1.0547147902585298e-06, "loss": 0.5243, "step": 24052 }, { "epoch": 0.8569392735628908, "grad_norm": 1.2244181861843317, "learning_rate": 1.054199027560463e-06, "loss": 0.3552, "step": 24053 }, { "epoch": 0.8569749006893849, "grad_norm": 1.439376504136338, "learning_rate": 1.0536833839819915e-06, "loss": 0.472, "step": 24054 }, { "epoch": 0.857010527815879, "grad_norm": 1.8592635639865385, "learning_rate": 1.0531678595299766e-06, "loss": 0.5077, "step": 24055 }, { "epoch": 0.8570461549423731, "grad_norm": 1.4034533450537876, "learning_rate": 1.052652454211286e-06, "loss": 0.4237, "step": 24056 }, { "epoch": 0.8570817820688672, "grad_norm": 1.2682376315948145, "learning_rate": 1.0521371680327808e-06, "loss": 0.4024, "step": 24057 }, { "epoch": 0.8571174091953614, "grad_norm": 1.7142719602426693, "learning_rate": 1.0516220010013234e-06, "loss": 0.5353, "step": 24058 }, { "epoch": 0.8571530363218555, "grad_norm": 1.317277108399878, "learning_rate": 1.0511069531237739e-06, "loss": 0.4039, "step": 24059 }, { "epoch": 0.8571886634483495, "grad_norm": 1.8456642565936365, "learning_rate": 1.050592024406991e-06, "loss": 0.4785, "step": 24060 }, { "epoch": 0.8572242905748437, "grad_norm": 1.4255772947180068, "learning_rate": 1.0500772148578297e-06, "loss": 0.3924, "step": 24061 }, { "epoch": 0.8572599177013378, "grad_norm": 1.427280721420955, "learning_rate": 1.0495625244831475e-06, "loss": 0.4156, "step": 24062 }, { "epoch": 0.8572955448278319, "grad_norm": 1.2927363725973775, "learning_rate": 1.0490479532897946e-06, "loss": 0.4209, "step": 24063 }, { "epoch": 0.8573311719543261, "grad_norm": 1.1207901304576695, "learning_rate": 1.0485335012846243e-06, "loss": 0.3764, "step": 24064 }, { "epoch": 0.8573667990808201, "grad_norm": 1.328439695244451, "learning_rate": 1.0480191684744867e-06, "loss": 0.4389, "step": 24065 }, { "epoch": 0.8574024262073142, "grad_norm": 1.9325407566768384, "learning_rate": 1.0475049548662308e-06, "loss": 0.5351, "step": 24066 }, { "epoch": 0.8574380533338084, "grad_norm": 1.4696603985556909, "learning_rate": 1.0469908604667067e-06, "loss": 0.5001, "step": 24067 }, { "epoch": 0.8574736804603025, "grad_norm": 1.4060709509538298, "learning_rate": 1.0464768852827544e-06, "loss": 0.411, "step": 24068 }, { "epoch": 0.8575093075867966, "grad_norm": 1.3487254372698334, "learning_rate": 1.0459630293212231e-06, "loss": 0.3716, "step": 24069 }, { "epoch": 0.8575449347132907, "grad_norm": 1.3951421546773417, "learning_rate": 1.0454492925889503e-06, "loss": 0.4296, "step": 24070 }, { "epoch": 0.8575805618397848, "grad_norm": 1.5690408396870614, "learning_rate": 1.0449356750927809e-06, "loss": 0.5227, "step": 24071 }, { "epoch": 0.8576161889662789, "grad_norm": 1.2622145713619666, "learning_rate": 1.044422176839549e-06, "loss": 0.367, "step": 24072 }, { "epoch": 0.8576518160927731, "grad_norm": 1.9204389822602554, "learning_rate": 1.0439087978360985e-06, "loss": 0.4437, "step": 24073 }, { "epoch": 0.8576874432192672, "grad_norm": 1.2513900446771646, "learning_rate": 1.0433955380892603e-06, "loss": 0.3674, "step": 24074 }, { "epoch": 0.8577230703457612, "grad_norm": 2.8490890195110263, "learning_rate": 1.042882397605871e-06, "loss": 0.4988, "step": 24075 }, { "epoch": 0.8577586974722554, "grad_norm": 2.0108358665718975, "learning_rate": 1.0423693763927656e-06, "loss": 0.5507, "step": 24076 }, { "epoch": 0.8577943245987495, "grad_norm": 5.150228239357274, "learning_rate": 1.0418564744567706e-06, "loss": 0.3949, "step": 24077 }, { "epoch": 0.8578299517252436, "grad_norm": 2.1040265661093676, "learning_rate": 1.0413436918047193e-06, "loss": 0.5592, "step": 24078 }, { "epoch": 0.8578655788517378, "grad_norm": 1.2807054565034195, "learning_rate": 1.0408310284434376e-06, "loss": 0.3697, "step": 24079 }, { "epoch": 0.8579012059782318, "grad_norm": 1.364888018772299, "learning_rate": 1.0403184843797521e-06, "loss": 0.3937, "step": 24080 }, { "epoch": 0.8579368331047259, "grad_norm": 1.4543917648907758, "learning_rate": 1.0398060596204885e-06, "loss": 0.3611, "step": 24081 }, { "epoch": 0.8579724602312201, "grad_norm": 1.2361131076714624, "learning_rate": 1.0392937541724724e-06, "loss": 0.3692, "step": 24082 }, { "epoch": 0.8580080873577142, "grad_norm": 1.2572876338379473, "learning_rate": 1.0387815680425206e-06, "loss": 0.3818, "step": 24083 }, { "epoch": 0.8580437144842082, "grad_norm": 1.2825483423639477, "learning_rate": 1.0382695012374588e-06, "loss": 0.3877, "step": 24084 }, { "epoch": 0.8580793416107024, "grad_norm": 1.4309917820412454, "learning_rate": 1.0377575537640994e-06, "loss": 0.4918, "step": 24085 }, { "epoch": 0.8581149687371965, "grad_norm": 1.43513663680797, "learning_rate": 1.0372457256292635e-06, "loss": 0.4262, "step": 24086 }, { "epoch": 0.8581505958636906, "grad_norm": 1.667850070609414, "learning_rate": 1.0367340168397645e-06, "loss": 0.4963, "step": 24087 }, { "epoch": 0.8581862229901848, "grad_norm": 1.398094275068987, "learning_rate": 1.036222427402417e-06, "loss": 0.4044, "step": 24088 }, { "epoch": 0.8582218501166788, "grad_norm": 2.6827184866373086, "learning_rate": 1.0357109573240353e-06, "loss": 0.5181, "step": 24089 }, { "epoch": 0.8582574772431729, "grad_norm": 1.705535199433679, "learning_rate": 1.0351996066114256e-06, "loss": 0.5034, "step": 24090 }, { "epoch": 0.8582931043696671, "grad_norm": 1.3637582262531063, "learning_rate": 1.0346883752714009e-06, "loss": 0.4999, "step": 24091 }, { "epoch": 0.8583287314961612, "grad_norm": 1.3249431986594695, "learning_rate": 1.034177263310766e-06, "loss": 0.3847, "step": 24092 }, { "epoch": 0.8583643586226553, "grad_norm": 1.2918418588767444, "learning_rate": 1.0336662707363287e-06, "loss": 0.4211, "step": 24093 }, { "epoch": 0.8583999857491494, "grad_norm": 1.6716897078208415, "learning_rate": 1.0331553975548902e-06, "loss": 0.4686, "step": 24094 }, { "epoch": 0.8584356128756435, "grad_norm": 1.640417014674472, "learning_rate": 1.0326446437732552e-06, "loss": 0.5303, "step": 24095 }, { "epoch": 0.8584712400021376, "grad_norm": 1.444058639937954, "learning_rate": 1.0321340093982245e-06, "loss": 0.3995, "step": 24096 }, { "epoch": 0.8585068671286318, "grad_norm": 1.5472307137600312, "learning_rate": 1.0316234944365998e-06, "loss": 0.4669, "step": 24097 }, { "epoch": 0.8585424942551259, "grad_norm": 1.8084245316023717, "learning_rate": 1.0311130988951756e-06, "loss": 0.4837, "step": 24098 }, { "epoch": 0.8585781213816199, "grad_norm": 1.6539206255413272, "learning_rate": 1.0306028227807486e-06, "loss": 0.55, "step": 24099 }, { "epoch": 0.8586137485081141, "grad_norm": 2.1886393426280137, "learning_rate": 1.0300926661001154e-06, "loss": 0.4126, "step": 24100 }, { "epoch": 0.8586493756346082, "grad_norm": 1.4342826261402817, "learning_rate": 1.0295826288600664e-06, "loss": 0.5216, "step": 24101 }, { "epoch": 0.8586850027611023, "grad_norm": 2.169012261410705, "learning_rate": 1.0290727110673948e-06, "loss": 0.4621, "step": 24102 }, { "epoch": 0.8587206298875965, "grad_norm": 2.1664240340667402, "learning_rate": 1.028562912728891e-06, "loss": 0.4707, "step": 24103 }, { "epoch": 0.8587562570140905, "grad_norm": 1.7634409434537872, "learning_rate": 1.0280532338513437e-06, "loss": 0.3906, "step": 24104 }, { "epoch": 0.8587918841405846, "grad_norm": 1.3707849116696498, "learning_rate": 1.0275436744415379e-06, "loss": 0.4519, "step": 24105 }, { "epoch": 0.8588275112670788, "grad_norm": 2.3005015535123787, "learning_rate": 1.0270342345062601e-06, "loss": 0.5013, "step": 24106 }, { "epoch": 0.8588631383935729, "grad_norm": 1.2842844139148182, "learning_rate": 1.0265249140522916e-06, "loss": 0.3726, "step": 24107 }, { "epoch": 0.8588987655200669, "grad_norm": 1.2605938688898295, "learning_rate": 1.0260157130864178e-06, "loss": 0.4012, "step": 24108 }, { "epoch": 0.8589343926465611, "grad_norm": 1.375200918515667, "learning_rate": 1.0255066316154139e-06, "loss": 0.3612, "step": 24109 }, { "epoch": 0.8589700197730552, "grad_norm": 1.5687889591016597, "learning_rate": 1.024997669646064e-06, "loss": 0.4385, "step": 24110 }, { "epoch": 0.8590056468995493, "grad_norm": 2.1616031382772216, "learning_rate": 1.024488827185144e-06, "loss": 0.4955, "step": 24111 }, { "epoch": 0.8590412740260435, "grad_norm": 1.2451774495022834, "learning_rate": 1.0239801042394282e-06, "loss": 0.404, "step": 24112 }, { "epoch": 0.8590769011525375, "grad_norm": 1.4232131991354058, "learning_rate": 1.0234715008156925e-06, "loss": 0.3778, "step": 24113 }, { "epoch": 0.8591125282790316, "grad_norm": 1.2437619815662633, "learning_rate": 1.022963016920706e-06, "loss": 0.4357, "step": 24114 }, { "epoch": 0.8591481554055258, "grad_norm": 2.1199776818280816, "learning_rate": 1.0224546525612434e-06, "loss": 0.4609, "step": 24115 }, { "epoch": 0.8591837825320199, "grad_norm": 1.9327453874428857, "learning_rate": 1.02194640774407e-06, "loss": 0.4714, "step": 24116 }, { "epoch": 0.859219409658514, "grad_norm": 1.334338150552986, "learning_rate": 1.0214382824759561e-06, "loss": 0.4206, "step": 24117 }, { "epoch": 0.8592550367850081, "grad_norm": 1.4213898661223616, "learning_rate": 1.0209302767636664e-06, "loss": 0.4522, "step": 24118 }, { "epoch": 0.8592906639115022, "grad_norm": 1.3671844044919312, "learning_rate": 1.0204223906139688e-06, "loss": 0.358, "step": 24119 }, { "epoch": 0.8593262910379963, "grad_norm": 1.5364483112798974, "learning_rate": 1.0199146240336211e-06, "loss": 0.522, "step": 24120 }, { "epoch": 0.8593619181644905, "grad_norm": 1.5782944032297974, "learning_rate": 1.019406977029388e-06, "loss": 0.5014, "step": 24121 }, { "epoch": 0.8593975452909846, "grad_norm": 1.2496939290827933, "learning_rate": 1.0188994496080273e-06, "loss": 0.3673, "step": 24122 }, { "epoch": 0.8594331724174786, "grad_norm": 1.5760290315670265, "learning_rate": 1.0183920417762971e-06, "loss": 0.5062, "step": 24123 }, { "epoch": 0.8594687995439728, "grad_norm": 1.6236867328758458, "learning_rate": 1.0178847535409554e-06, "loss": 0.5234, "step": 24124 }, { "epoch": 0.8595044266704669, "grad_norm": 1.3142685656958306, "learning_rate": 1.0173775849087552e-06, "loss": 0.4164, "step": 24125 }, { "epoch": 0.859540053796961, "grad_norm": 3.044099277417311, "learning_rate": 1.016870535886454e-06, "loss": 0.3915, "step": 24126 }, { "epoch": 0.8595756809234552, "grad_norm": 1.7520131242435706, "learning_rate": 1.016363606480798e-06, "loss": 0.4783, "step": 24127 }, { "epoch": 0.8596113080499492, "grad_norm": 1.9323687863893864, "learning_rate": 1.0158567966985412e-06, "loss": 0.4317, "step": 24128 }, { "epoch": 0.8596469351764433, "grad_norm": 1.2432963616600827, "learning_rate": 1.01535010654643e-06, "loss": 0.4243, "step": 24129 }, { "epoch": 0.8596825623029375, "grad_norm": 1.2700371021950576, "learning_rate": 1.0148435360312125e-06, "loss": 0.436, "step": 24130 }, { "epoch": 0.8597181894294316, "grad_norm": 1.4430763083455496, "learning_rate": 1.0143370851596324e-06, "loss": 0.4344, "step": 24131 }, { "epoch": 0.8597538165559256, "grad_norm": 1.3925711436171555, "learning_rate": 1.0138307539384352e-06, "loss": 0.4588, "step": 24132 }, { "epoch": 0.8597894436824198, "grad_norm": 1.4022319842397024, "learning_rate": 1.013324542374361e-06, "loss": 0.4174, "step": 24133 }, { "epoch": 0.8598250708089139, "grad_norm": 1.7266362181522186, "learning_rate": 1.0128184504741535e-06, "loss": 0.5316, "step": 24134 }, { "epoch": 0.859860697935408, "grad_norm": 2.1509346368945392, "learning_rate": 1.0123124782445505e-06, "loss": 0.4388, "step": 24135 }, { "epoch": 0.8598963250619022, "grad_norm": 1.2797599800274964, "learning_rate": 1.0118066256922876e-06, "loss": 0.4062, "step": 24136 }, { "epoch": 0.8599319521883962, "grad_norm": 1.4886776964077935, "learning_rate": 1.0113008928241042e-06, "loss": 0.4456, "step": 24137 }, { "epoch": 0.8599675793148903, "grad_norm": 1.4978604887918818, "learning_rate": 1.010795279646729e-06, "loss": 0.5016, "step": 24138 }, { "epoch": 0.8600032064413845, "grad_norm": 1.8450620986726154, "learning_rate": 1.0102897861668993e-06, "loss": 0.4008, "step": 24139 }, { "epoch": 0.8600388335678786, "grad_norm": 1.5971944837267107, "learning_rate": 1.0097844123913437e-06, "loss": 0.4963, "step": 24140 }, { "epoch": 0.8600744606943727, "grad_norm": 1.3119753896371493, "learning_rate": 1.0092791583267936e-06, "loss": 0.4201, "step": 24141 }, { "epoch": 0.8601100878208668, "grad_norm": 1.7320757452424786, "learning_rate": 1.0087740239799749e-06, "loss": 0.4126, "step": 24142 }, { "epoch": 0.8601457149473609, "grad_norm": 1.4290883815156825, "learning_rate": 1.0082690093576163e-06, "loss": 0.4638, "step": 24143 }, { "epoch": 0.860181342073855, "grad_norm": 1.6904461474930985, "learning_rate": 1.0077641144664386e-06, "loss": 0.4968, "step": 24144 }, { "epoch": 0.8602169692003492, "grad_norm": 1.6729381852854108, "learning_rate": 1.0072593393131668e-06, "loss": 0.6133, "step": 24145 }, { "epoch": 0.8602525963268433, "grad_norm": 1.7274418754935892, "learning_rate": 1.006754683904524e-06, "loss": 0.4851, "step": 24146 }, { "epoch": 0.8602882234533373, "grad_norm": 2.430159845976032, "learning_rate": 1.0062501482472254e-06, "loss": 0.401, "step": 24147 }, { "epoch": 0.8603238505798315, "grad_norm": 1.6502641102795583, "learning_rate": 1.0057457323479957e-06, "loss": 0.4835, "step": 24148 }, { "epoch": 0.8603594777063256, "grad_norm": 1.5130891558780364, "learning_rate": 1.0052414362135466e-06, "loss": 0.4651, "step": 24149 }, { "epoch": 0.8603951048328197, "grad_norm": 1.4691479523767932, "learning_rate": 1.0047372598505956e-06, "loss": 0.4138, "step": 24150 }, { "epoch": 0.8604307319593139, "grad_norm": 2.020296730621217, "learning_rate": 1.0042332032658541e-06, "loss": 0.5115, "step": 24151 }, { "epoch": 0.8604663590858079, "grad_norm": 1.3414697300145877, "learning_rate": 1.003729266466038e-06, "loss": 0.3972, "step": 24152 }, { "epoch": 0.860501986212302, "grad_norm": 1.5981637179056296, "learning_rate": 1.0032254494578519e-06, "loss": 0.5491, "step": 24153 }, { "epoch": 0.8605376133387962, "grad_norm": 1.3261199283265888, "learning_rate": 1.002721752248007e-06, "loss": 0.3771, "step": 24154 }, { "epoch": 0.8605732404652903, "grad_norm": 1.2630489669990796, "learning_rate": 1.0022181748432113e-06, "loss": 0.41, "step": 24155 }, { "epoch": 0.8606088675917843, "grad_norm": 1.2779685433621701, "learning_rate": 1.0017147172501695e-06, "loss": 0.4183, "step": 24156 }, { "epoch": 0.8606444947182785, "grad_norm": 1.4282367070637536, "learning_rate": 1.0012113794755873e-06, "loss": 0.5182, "step": 24157 }, { "epoch": 0.8606801218447726, "grad_norm": 2.1834728145799867, "learning_rate": 1.0007081615261638e-06, "loss": 0.475, "step": 24158 }, { "epoch": 0.8607157489712667, "grad_norm": 1.5016001194887587, "learning_rate": 1.0002050634086036e-06, "loss": 0.4185, "step": 24159 }, { "epoch": 0.8607513760977609, "grad_norm": 1.3361607102292505, "learning_rate": 9.997020851296003e-07, "loss": 0.4096, "step": 24160 }, { "epoch": 0.8607870032242549, "grad_norm": 1.2912726174983513, "learning_rate": 9.99199226695856e-07, "loss": 0.3969, "step": 24161 }, { "epoch": 0.860822630350749, "grad_norm": 1.9318497866781497, "learning_rate": 9.986964881140638e-07, "loss": 0.6543, "step": 24162 }, { "epoch": 0.8608582574772432, "grad_norm": 1.2844575125212896, "learning_rate": 9.981938693909221e-07, "loss": 0.3521, "step": 24163 }, { "epoch": 0.8608938846037373, "grad_norm": 1.6314870516631619, "learning_rate": 9.976913705331192e-07, "loss": 0.5019, "step": 24164 }, { "epoch": 0.8609295117302314, "grad_norm": 2.0423000656701467, "learning_rate": 9.971889915473499e-07, "loss": 0.4113, "step": 24165 }, { "epoch": 0.8609651388567255, "grad_norm": 1.3345595084775976, "learning_rate": 9.966867324402997e-07, "loss": 0.4118, "step": 24166 }, { "epoch": 0.8610007659832196, "grad_norm": 1.6854806451933573, "learning_rate": 9.961845932186599e-07, "loss": 0.3989, "step": 24167 }, { "epoch": 0.8610363931097137, "grad_norm": 1.570906942956617, "learning_rate": 9.95682573889114e-07, "loss": 0.5327, "step": 24168 }, { "epoch": 0.8610720202362079, "grad_norm": 1.4718693292026392, "learning_rate": 9.951806744583482e-07, "loss": 0.5075, "step": 24169 }, { "epoch": 0.861107647362702, "grad_norm": 1.6742410178680365, "learning_rate": 9.946788949330454e-07, "loss": 0.4049, "step": 24170 }, { "epoch": 0.861143274489196, "grad_norm": 1.4415208929283865, "learning_rate": 9.941772353198886e-07, "loss": 0.5439, "step": 24171 }, { "epoch": 0.8611789016156902, "grad_norm": 1.5283349762461313, "learning_rate": 9.936756956255577e-07, "loss": 0.4512, "step": 24172 }, { "epoch": 0.8612145287421843, "grad_norm": 1.3478962118598754, "learning_rate": 9.931742758567275e-07, "loss": 0.4342, "step": 24173 }, { "epoch": 0.8612501558686784, "grad_norm": 1.7235460519193495, "learning_rate": 9.926729760200805e-07, "loss": 0.4739, "step": 24174 }, { "epoch": 0.8612857829951726, "grad_norm": 2.3411352992018957, "learning_rate": 9.921717961222855e-07, "loss": 0.4932, "step": 24175 }, { "epoch": 0.8613214101216666, "grad_norm": 1.4456159993055278, "learning_rate": 9.916707361700195e-07, "loss": 0.423, "step": 24176 }, { "epoch": 0.8613570372481607, "grad_norm": 2.894906213898653, "learning_rate": 9.91169796169954e-07, "loss": 0.5156, "step": 24177 }, { "epoch": 0.8613926643746549, "grad_norm": 1.4108211411445746, "learning_rate": 9.906689761287613e-07, "loss": 0.3452, "step": 24178 }, { "epoch": 0.861428291501149, "grad_norm": 1.3914169046174565, "learning_rate": 9.901682760531072e-07, "loss": 0.4242, "step": 24179 }, { "epoch": 0.861463918627643, "grad_norm": 1.2705039676949685, "learning_rate": 9.896676959496587e-07, "loss": 0.3941, "step": 24180 }, { "epoch": 0.8614995457541372, "grad_norm": 1.4131929329864947, "learning_rate": 9.89167235825086e-07, "loss": 0.4217, "step": 24181 }, { "epoch": 0.8615351728806313, "grad_norm": 1.695661446010639, "learning_rate": 9.886668956860467e-07, "loss": 0.5081, "step": 24182 }, { "epoch": 0.8615708000071254, "grad_norm": 1.5396252298479993, "learning_rate": 9.881666755392083e-07, "loss": 0.4562, "step": 24183 }, { "epoch": 0.8616064271336196, "grad_norm": 1.4833855652800676, "learning_rate": 9.876665753912262e-07, "loss": 0.5741, "step": 24184 }, { "epoch": 0.8616420542601136, "grad_norm": 1.4329860557974536, "learning_rate": 9.871665952487663e-07, "loss": 0.4156, "step": 24185 }, { "epoch": 0.8616776813866077, "grad_norm": 1.3958593533168335, "learning_rate": 9.86666735118481e-07, "loss": 0.435, "step": 24186 }, { "epoch": 0.8617133085131019, "grad_norm": 1.3950749839701004, "learning_rate": 9.861669950070308e-07, "loss": 0.4066, "step": 24187 }, { "epoch": 0.861748935639596, "grad_norm": 1.3082096730410866, "learning_rate": 9.856673749210644e-07, "loss": 0.4071, "step": 24188 }, { "epoch": 0.8617845627660901, "grad_norm": 1.475733273129701, "learning_rate": 9.8516787486724e-07, "loss": 0.4552, "step": 24189 }, { "epoch": 0.8618201898925842, "grad_norm": 1.5438078206800148, "learning_rate": 9.846684948522034e-07, "loss": 0.5791, "step": 24190 }, { "epoch": 0.8618558170190783, "grad_norm": 1.5558671887744329, "learning_rate": 9.841692348826082e-07, "loss": 0.5411, "step": 24191 }, { "epoch": 0.8618914441455724, "grad_norm": 2.159325514447547, "learning_rate": 9.836700949651012e-07, "loss": 0.4507, "step": 24192 }, { "epoch": 0.8619270712720666, "grad_norm": 1.727285618326966, "learning_rate": 9.831710751063284e-07, "loss": 0.3879, "step": 24193 }, { "epoch": 0.8619626983985607, "grad_norm": 1.4080486547758309, "learning_rate": 9.826721753129375e-07, "loss": 0.4327, "step": 24194 }, { "epoch": 0.8619983255250547, "grad_norm": 1.2288457677707247, "learning_rate": 9.82173395591568e-07, "loss": 0.3812, "step": 24195 }, { "epoch": 0.8620339526515489, "grad_norm": 1.5026523445846707, "learning_rate": 9.816747359488632e-07, "loss": 0.3873, "step": 24196 }, { "epoch": 0.862069579778043, "grad_norm": 1.2530969291699632, "learning_rate": 9.811761963914623e-07, "loss": 0.445, "step": 24197 }, { "epoch": 0.8621052069045371, "grad_norm": 1.3599717921932584, "learning_rate": 9.806777769260034e-07, "loss": 0.4252, "step": 24198 }, { "epoch": 0.8621408340310313, "grad_norm": 1.2705039788698345, "learning_rate": 9.801794775591234e-07, "loss": 0.3961, "step": 24199 }, { "epoch": 0.8621764611575253, "grad_norm": 1.6007281688483894, "learning_rate": 9.796812982974613e-07, "loss": 0.4384, "step": 24200 }, { "epoch": 0.8622120882840194, "grad_norm": 1.303867785569097, "learning_rate": 9.791832391476452e-07, "loss": 0.4329, "step": 24201 }, { "epoch": 0.8622477154105136, "grad_norm": 1.2742061747153723, "learning_rate": 9.786853001163087e-07, "loss": 0.4457, "step": 24202 }, { "epoch": 0.8622833425370077, "grad_norm": 2.4087243961582185, "learning_rate": 9.781874812100856e-07, "loss": 0.5355, "step": 24203 }, { "epoch": 0.8623189696635017, "grad_norm": 1.6576428915755375, "learning_rate": 9.776897824356002e-07, "loss": 0.486, "step": 24204 }, { "epoch": 0.8623545967899959, "grad_norm": 1.3106270207276236, "learning_rate": 9.77192203799483e-07, "loss": 0.3727, "step": 24205 }, { "epoch": 0.86239022391649, "grad_norm": 1.664910027268915, "learning_rate": 9.766947453083564e-07, "loss": 0.3609, "step": 24206 }, { "epoch": 0.8624258510429841, "grad_norm": 1.7147136300178, "learning_rate": 9.761974069688462e-07, "loss": 0.5358, "step": 24207 }, { "epoch": 0.8624614781694783, "grad_norm": 1.2978333212141502, "learning_rate": 9.75700188787575e-07, "loss": 0.4187, "step": 24208 }, { "epoch": 0.8624971052959723, "grad_norm": 1.3668835819376344, "learning_rate": 9.752030907711652e-07, "loss": 0.4248, "step": 24209 }, { "epoch": 0.8625327324224664, "grad_norm": 2.0140482522087946, "learning_rate": 9.747061129262325e-07, "loss": 0.443, "step": 24210 }, { "epoch": 0.8625683595489606, "grad_norm": 1.3983812073722626, "learning_rate": 9.742092552593985e-07, "loss": 0.4536, "step": 24211 }, { "epoch": 0.8626039866754547, "grad_norm": 1.758441654030102, "learning_rate": 9.737125177772744e-07, "loss": 0.3547, "step": 24212 }, { "epoch": 0.8626396138019488, "grad_norm": 1.4871660071125696, "learning_rate": 9.732159004864772e-07, "loss": 0.51, "step": 24213 }, { "epoch": 0.8626752409284429, "grad_norm": 1.315844013246792, "learning_rate": 9.727194033936194e-07, "loss": 0.3968, "step": 24214 }, { "epoch": 0.862710868054937, "grad_norm": 1.2832220295618841, "learning_rate": 9.722230265053124e-07, "loss": 0.3938, "step": 24215 }, { "epoch": 0.8627464951814311, "grad_norm": 1.2417416904634468, "learning_rate": 9.717267698281684e-07, "loss": 0.3879, "step": 24216 }, { "epoch": 0.8627821223079253, "grad_norm": 1.4152931576881058, "learning_rate": 9.712306333687905e-07, "loss": 0.4119, "step": 24217 }, { "epoch": 0.8628177494344194, "grad_norm": 1.3985611294758153, "learning_rate": 9.707346171337895e-07, "loss": 0.3914, "step": 24218 }, { "epoch": 0.8628533765609134, "grad_norm": 1.3008907396703524, "learning_rate": 9.702387211297659e-07, "loss": 0.3826, "step": 24219 }, { "epoch": 0.8628890036874076, "grad_norm": 1.416564340794315, "learning_rate": 9.697429453633268e-07, "loss": 0.3968, "step": 24220 }, { "epoch": 0.8629246308139017, "grad_norm": 1.5783682342447916, "learning_rate": 9.69247289841071e-07, "loss": 0.5594, "step": 24221 }, { "epoch": 0.8629602579403958, "grad_norm": 1.360173280509377, "learning_rate": 9.68751754569599e-07, "loss": 0.4265, "step": 24222 }, { "epoch": 0.86299588506689, "grad_norm": 1.8092896556413969, "learning_rate": 9.6825633955551e-07, "loss": 0.5661, "step": 24223 }, { "epoch": 0.863031512193384, "grad_norm": 1.3362478472552444, "learning_rate": 9.677610448054032e-07, "loss": 0.4274, "step": 24224 }, { "epoch": 0.8630671393198781, "grad_norm": 1.3752458484981398, "learning_rate": 9.672658703258685e-07, "loss": 0.5044, "step": 24225 }, { "epoch": 0.8631027664463723, "grad_norm": 1.565547538331964, "learning_rate": 9.667708161235033e-07, "loss": 0.4588, "step": 24226 }, { "epoch": 0.8631383935728664, "grad_norm": 1.5114534526548917, "learning_rate": 9.662758822049002e-07, "loss": 0.385, "step": 24227 }, { "epoch": 0.8631740206993604, "grad_norm": 1.7299006927160148, "learning_rate": 9.657810685766455e-07, "loss": 0.4792, "step": 24228 }, { "epoch": 0.8632096478258546, "grad_norm": 1.787201193285283, "learning_rate": 9.6528637524533e-07, "loss": 0.561, "step": 24229 }, { "epoch": 0.8632452749523487, "grad_norm": 1.7053621223728284, "learning_rate": 9.647918022175417e-07, "loss": 0.5777, "step": 24230 }, { "epoch": 0.8632809020788428, "grad_norm": 1.4679202593458982, "learning_rate": 9.642973494998687e-07, "loss": 0.4891, "step": 24231 }, { "epoch": 0.863316529205337, "grad_norm": 1.437037004886372, "learning_rate": 9.638030170988888e-07, "loss": 0.3696, "step": 24232 }, { "epoch": 0.863352156331831, "grad_norm": 1.2928664917494668, "learning_rate": 9.633088050211902e-07, "loss": 0.4055, "step": 24233 }, { "epoch": 0.8633877834583251, "grad_norm": 1.795109772325864, "learning_rate": 9.62814713273349e-07, "loss": 0.5098, "step": 24234 }, { "epoch": 0.8634234105848193, "grad_norm": 1.5553866490836783, "learning_rate": 9.623207418619484e-07, "loss": 0.4908, "step": 24235 }, { "epoch": 0.8634590377113134, "grad_norm": 1.8821037263222873, "learning_rate": 9.618268907935602e-07, "loss": 0.4672, "step": 24236 }, { "epoch": 0.8634946648378075, "grad_norm": 1.3293347046425514, "learning_rate": 9.613331600747666e-07, "loss": 0.4277, "step": 24237 }, { "epoch": 0.8635302919643016, "grad_norm": 1.3446955714523003, "learning_rate": 9.608395497121414e-07, "loss": 0.3894, "step": 24238 }, { "epoch": 0.8635659190907957, "grad_norm": 1.6914002508754573, "learning_rate": 9.603460597122527e-07, "loss": 0.5047, "step": 24239 }, { "epoch": 0.8636015462172898, "grad_norm": 1.3395840845017042, "learning_rate": 9.598526900816773e-07, "loss": 0.4046, "step": 24240 }, { "epoch": 0.863637173343784, "grad_norm": 1.4412006544390832, "learning_rate": 9.593594408269802e-07, "loss": 0.4321, "step": 24241 }, { "epoch": 0.8636728004702781, "grad_norm": 1.554215890310534, "learning_rate": 9.588663119547337e-07, "loss": 0.5135, "step": 24242 }, { "epoch": 0.8637084275967721, "grad_norm": 1.28117835685788, "learning_rate": 9.583733034714982e-07, "loss": 0.3954, "step": 24243 }, { "epoch": 0.8637440547232663, "grad_norm": 1.3926981798282354, "learning_rate": 9.57880415383843e-07, "loss": 0.4262, "step": 24244 }, { "epoch": 0.8637796818497604, "grad_norm": 1.6855134448128575, "learning_rate": 9.5738764769833e-07, "loss": 0.5077, "step": 24245 }, { "epoch": 0.8638153089762545, "grad_norm": 1.6495539307002962, "learning_rate": 9.568950004215227e-07, "loss": 0.5596, "step": 24246 }, { "epoch": 0.8638509361027487, "grad_norm": 2.0942514976286435, "learning_rate": 9.564024735599775e-07, "loss": 0.5063, "step": 24247 }, { "epoch": 0.8638865632292427, "grad_norm": 1.2538363813444005, "learning_rate": 9.559100671202537e-07, "loss": 0.4019, "step": 24248 }, { "epoch": 0.8639221903557368, "grad_norm": 1.9116450079409077, "learning_rate": 9.554177811089115e-07, "loss": 0.4597, "step": 24249 }, { "epoch": 0.863957817482231, "grad_norm": 1.7685133454604676, "learning_rate": 9.549256155325015e-07, "loss": 0.5325, "step": 24250 }, { "epoch": 0.8639934446087251, "grad_norm": 1.4779127731393698, "learning_rate": 9.544335703975781e-07, "loss": 0.3856, "step": 24251 }, { "epoch": 0.8640290717352191, "grad_norm": 1.7358716412577133, "learning_rate": 9.539416457106954e-07, "loss": 0.3878, "step": 24252 }, { "epoch": 0.8640646988617133, "grad_norm": 1.3557701219626452, "learning_rate": 9.534498414784033e-07, "loss": 0.4281, "step": 24253 }, { "epoch": 0.8641003259882074, "grad_norm": 1.5409437061020064, "learning_rate": 9.529581577072478e-07, "loss": 0.5095, "step": 24254 }, { "epoch": 0.8641359531147015, "grad_norm": 1.445836237437555, "learning_rate": 9.524665944037803e-07, "loss": 0.4623, "step": 24255 }, { "epoch": 0.8641715802411957, "grad_norm": 1.6816489115706816, "learning_rate": 9.519751515745424e-07, "loss": 0.5002, "step": 24256 }, { "epoch": 0.8642072073676897, "grad_norm": 1.6404506433194705, "learning_rate": 9.514838292260808e-07, "loss": 0.4789, "step": 24257 }, { "epoch": 0.8642428344941838, "grad_norm": 1.5235491850511187, "learning_rate": 9.509926273649351e-07, "loss": 0.5268, "step": 24258 }, { "epoch": 0.864278461620678, "grad_norm": 1.6170874687216787, "learning_rate": 9.505015459976474e-07, "loss": 0.521, "step": 24259 }, { "epoch": 0.8643140887471721, "grad_norm": 1.5140718127180577, "learning_rate": 9.500105851307573e-07, "loss": 0.4799, "step": 24260 }, { "epoch": 0.8643497158736662, "grad_norm": 1.4915395939271365, "learning_rate": 9.495197447708004e-07, "loss": 0.4475, "step": 24261 }, { "epoch": 0.8643853430001603, "grad_norm": 1.6603916361714166, "learning_rate": 9.490290249243173e-07, "loss": 0.5464, "step": 24262 }, { "epoch": 0.8644209701266544, "grad_norm": 1.6841856794106076, "learning_rate": 9.48538425597837e-07, "loss": 0.4803, "step": 24263 }, { "epoch": 0.8644565972531485, "grad_norm": 1.6785678061889855, "learning_rate": 9.480479467978953e-07, "loss": 0.4793, "step": 24264 }, { "epoch": 0.8644922243796427, "grad_norm": 1.423085477138127, "learning_rate": 9.475575885310207e-07, "loss": 0.4264, "step": 24265 }, { "epoch": 0.8645278515061368, "grad_norm": 1.5980301393869525, "learning_rate": 9.470673508037431e-07, "loss": 0.6351, "step": 24266 }, { "epoch": 0.8645634786326308, "grad_norm": 1.2732663990953037, "learning_rate": 9.465772336225931e-07, "loss": 0.4175, "step": 24267 }, { "epoch": 0.864599105759125, "grad_norm": 1.3605392064070025, "learning_rate": 9.460872369940954e-07, "loss": 0.4058, "step": 24268 }, { "epoch": 0.8646347328856191, "grad_norm": 1.4058040663850193, "learning_rate": 9.455973609247737e-07, "loss": 0.4198, "step": 24269 }, { "epoch": 0.8646703600121132, "grad_norm": 1.2449111537712654, "learning_rate": 9.451076054211527e-07, "loss": 0.4158, "step": 24270 }, { "epoch": 0.8647059871386074, "grad_norm": 1.4813132211666218, "learning_rate": 9.446179704897518e-07, "loss": 0.473, "step": 24271 }, { "epoch": 0.8647416142651014, "grad_norm": 1.4153243409297493, "learning_rate": 9.441284561370923e-07, "loss": 0.4515, "step": 24272 }, { "epoch": 0.8647772413915955, "grad_norm": 1.5613758096248875, "learning_rate": 9.436390623696911e-07, "loss": 0.4527, "step": 24273 }, { "epoch": 0.8648128685180897, "grad_norm": 1.2820104039025382, "learning_rate": 9.431497891940667e-07, "loss": 0.4061, "step": 24274 }, { "epoch": 0.8648484956445838, "grad_norm": 1.824417560309394, "learning_rate": 9.426606366167346e-07, "loss": 0.5227, "step": 24275 }, { "epoch": 0.8648841227710778, "grad_norm": 1.9078926327183554, "learning_rate": 9.421716046442054e-07, "loss": 0.4995, "step": 24276 }, { "epoch": 0.864919749897572, "grad_norm": 1.2713493427177232, "learning_rate": 9.416826932829936e-07, "loss": 0.3916, "step": 24277 }, { "epoch": 0.8649553770240661, "grad_norm": 1.2967506874590604, "learning_rate": 9.411939025396067e-07, "loss": 0.3935, "step": 24278 }, { "epoch": 0.8649910041505602, "grad_norm": 1.389676197668142, "learning_rate": 9.407052324205556e-07, "loss": 0.3704, "step": 24279 }, { "epoch": 0.8650266312770544, "grad_norm": 1.5328733065079037, "learning_rate": 9.402166829323455e-07, "loss": 0.4615, "step": 24280 }, { "epoch": 0.8650622584035484, "grad_norm": 1.3119767631900425, "learning_rate": 9.397282540814823e-07, "loss": 0.3924, "step": 24281 }, { "epoch": 0.8650978855300425, "grad_norm": 1.7936722999666306, "learning_rate": 9.392399458744706e-07, "loss": 0.5009, "step": 24282 }, { "epoch": 0.8651335126565367, "grad_norm": 1.3441562185154996, "learning_rate": 9.387517583178119e-07, "loss": 0.4491, "step": 24283 }, { "epoch": 0.8651691397830308, "grad_norm": 1.510823557054029, "learning_rate": 9.382636914180087e-07, "loss": 0.4992, "step": 24284 }, { "epoch": 0.8652047669095249, "grad_norm": 1.9059086526998956, "learning_rate": 9.377757451815562e-07, "loss": 0.4714, "step": 24285 }, { "epoch": 0.865240394036019, "grad_norm": 1.6365160905115275, "learning_rate": 9.372879196149564e-07, "loss": 0.4589, "step": 24286 }, { "epoch": 0.8652760211625131, "grad_norm": 1.2700978571722645, "learning_rate": 9.368002147247002e-07, "loss": 0.3869, "step": 24287 }, { "epoch": 0.8653116482890072, "grad_norm": 1.3279752617850424, "learning_rate": 9.363126305172831e-07, "loss": 0.4409, "step": 24288 }, { "epoch": 0.8653472754155014, "grad_norm": 1.3221650939506224, "learning_rate": 9.358251669991992e-07, "loss": 0.4349, "step": 24289 }, { "epoch": 0.8653829025419955, "grad_norm": 1.2623232288799564, "learning_rate": 9.353378241769407e-07, "loss": 0.4011, "step": 24290 }, { "epoch": 0.8654185296684895, "grad_norm": 1.4576992776828839, "learning_rate": 9.348506020569925e-07, "loss": 0.4087, "step": 24291 }, { "epoch": 0.8654541567949837, "grad_norm": 1.5671614742817455, "learning_rate": 9.343635006458473e-07, "loss": 0.4798, "step": 24292 }, { "epoch": 0.8654897839214778, "grad_norm": 1.3531916978749972, "learning_rate": 9.338765199499855e-07, "loss": 0.4544, "step": 24293 }, { "epoch": 0.8655254110479719, "grad_norm": 1.5718930969909612, "learning_rate": 9.333896599758974e-07, "loss": 0.4222, "step": 24294 }, { "epoch": 0.8655610381744661, "grad_norm": 1.5193263000124697, "learning_rate": 9.329029207300611e-07, "loss": 0.5235, "step": 24295 }, { "epoch": 0.8655966653009601, "grad_norm": 1.4958765529510245, "learning_rate": 9.324163022189603e-07, "loss": 0.509, "step": 24296 }, { "epoch": 0.8656322924274542, "grad_norm": 1.5831312507559752, "learning_rate": 9.319298044490744e-07, "loss": 0.5281, "step": 24297 }, { "epoch": 0.8656679195539484, "grad_norm": 1.3778402075530007, "learning_rate": 9.314434274268802e-07, "loss": 0.3743, "step": 24298 }, { "epoch": 0.8657035466804425, "grad_norm": 1.9689211041411665, "learning_rate": 9.309571711588583e-07, "loss": 0.561, "step": 24299 }, { "epoch": 0.8657391738069365, "grad_norm": 1.6066595656242402, "learning_rate": 9.304710356514779e-07, "loss": 0.4652, "step": 24300 }, { "epoch": 0.8657748009334307, "grad_norm": 1.2797210229543645, "learning_rate": 9.299850209112182e-07, "loss": 0.4062, "step": 24301 }, { "epoch": 0.8658104280599248, "grad_norm": 1.7587264827754272, "learning_rate": 9.294991269445442e-07, "loss": 0.5423, "step": 24302 }, { "epoch": 0.865846055186419, "grad_norm": 1.4648551359708237, "learning_rate": 9.290133537579294e-07, "loss": 0.4647, "step": 24303 }, { "epoch": 0.8658816823129131, "grad_norm": 1.2915636167776043, "learning_rate": 9.28527701357842e-07, "loss": 0.3846, "step": 24304 }, { "epoch": 0.8659173094394071, "grad_norm": 1.2848321226774189, "learning_rate": 9.280421697507514e-07, "loss": 0.3822, "step": 24305 }, { "epoch": 0.8659529365659013, "grad_norm": 1.580467551676254, "learning_rate": 9.275567589431178e-07, "loss": 0.4312, "step": 24306 }, { "epoch": 0.8659885636923954, "grad_norm": 1.4093221948179557, "learning_rate": 9.270714689414073e-07, "loss": 0.4909, "step": 24307 }, { "epoch": 0.8660241908188895, "grad_norm": 1.2733745272367456, "learning_rate": 9.265862997520836e-07, "loss": 0.3569, "step": 24308 }, { "epoch": 0.8660598179453837, "grad_norm": 1.226868560160994, "learning_rate": 9.261012513816037e-07, "loss": 0.3945, "step": 24309 }, { "epoch": 0.8660954450718777, "grad_norm": 1.6652337534534483, "learning_rate": 9.256163238364302e-07, "loss": 0.4388, "step": 24310 }, { "epoch": 0.8661310721983718, "grad_norm": 2.2154871972360115, "learning_rate": 9.251315171230136e-07, "loss": 0.5148, "step": 24311 }, { "epoch": 0.866166699324866, "grad_norm": 1.6731211317014654, "learning_rate": 9.246468312478174e-07, "loss": 0.5254, "step": 24312 }, { "epoch": 0.8662023264513601, "grad_norm": 1.797432936185769, "learning_rate": 9.2416226621729e-07, "loss": 0.5364, "step": 24313 }, { "epoch": 0.8662379535778542, "grad_norm": 1.5022228136777478, "learning_rate": 9.236778220378872e-07, "loss": 0.5405, "step": 24314 }, { "epoch": 0.8662735807043483, "grad_norm": 1.2709616490298623, "learning_rate": 9.231934987160573e-07, "loss": 0.3836, "step": 24315 }, { "epoch": 0.8663092078308424, "grad_norm": 1.1994431040891544, "learning_rate": 9.227092962582518e-07, "loss": 0.3735, "step": 24316 }, { "epoch": 0.8663448349573365, "grad_norm": 1.4786845061932261, "learning_rate": 9.222252146709143e-07, "loss": 0.4211, "step": 24317 }, { "epoch": 0.8663804620838307, "grad_norm": 1.4791836387874409, "learning_rate": 9.217412539604942e-07, "loss": 0.386, "step": 24318 }, { "epoch": 0.8664160892103248, "grad_norm": 1.5564713568913846, "learning_rate": 9.21257414133433e-07, "loss": 0.5171, "step": 24319 }, { "epoch": 0.8664517163368188, "grad_norm": 1.3653791859988929, "learning_rate": 9.207736951961766e-07, "loss": 0.3787, "step": 24320 }, { "epoch": 0.866487343463313, "grad_norm": 2.827645160430618, "learning_rate": 9.202900971551654e-07, "loss": 0.5456, "step": 24321 }, { "epoch": 0.8665229705898071, "grad_norm": 1.9933207136127689, "learning_rate": 9.198066200168365e-07, "loss": 0.4573, "step": 24322 }, { "epoch": 0.8665585977163012, "grad_norm": 1.2370025670346927, "learning_rate": 9.193232637876303e-07, "loss": 0.3763, "step": 24323 }, { "epoch": 0.8665942248427954, "grad_norm": 1.508109885268103, "learning_rate": 9.188400284739807e-07, "loss": 0.5266, "step": 24324 }, { "epoch": 0.8666298519692894, "grad_norm": 1.3088236079678819, "learning_rate": 9.183569140823223e-07, "loss": 0.4082, "step": 24325 }, { "epoch": 0.8666654790957835, "grad_norm": 1.175853060111255, "learning_rate": 9.178739206190912e-07, "loss": 0.3746, "step": 24326 }, { "epoch": 0.8667011062222777, "grad_norm": 1.3249984586380736, "learning_rate": 9.173910480907166e-07, "loss": 0.4083, "step": 24327 }, { "epoch": 0.8667367333487718, "grad_norm": 1.2600313752963264, "learning_rate": 9.16908296503628e-07, "loss": 0.3715, "step": 24328 }, { "epoch": 0.8667723604752658, "grad_norm": 1.4321815707779886, "learning_rate": 9.164256658642546e-07, "loss": 0.6048, "step": 24329 }, { "epoch": 0.86680798760176, "grad_norm": 1.3993539344071284, "learning_rate": 9.159431561790233e-07, "loss": 0.4354, "step": 24330 }, { "epoch": 0.8668436147282541, "grad_norm": 1.6681817748041872, "learning_rate": 9.15460767454357e-07, "loss": 0.3759, "step": 24331 }, { "epoch": 0.8668792418547482, "grad_norm": 1.4537463542249511, "learning_rate": 9.149784996966815e-07, "loss": 0.4834, "step": 24332 }, { "epoch": 0.8669148689812424, "grad_norm": 1.368115400811596, "learning_rate": 9.144963529124163e-07, "loss": 0.4255, "step": 24333 }, { "epoch": 0.8669504961077364, "grad_norm": 1.5988571187397171, "learning_rate": 9.140143271079815e-07, "loss": 0.5156, "step": 24334 }, { "epoch": 0.8669861232342305, "grad_norm": 1.4754474890188718, "learning_rate": 9.135324222897979e-07, "loss": 0.4742, "step": 24335 }, { "epoch": 0.8670217503607247, "grad_norm": 1.5193209272405304, "learning_rate": 9.130506384642823e-07, "loss": 0.4233, "step": 24336 }, { "epoch": 0.8670573774872188, "grad_norm": 1.5705698392179435, "learning_rate": 9.125689756378464e-07, "loss": 0.5744, "step": 24337 }, { "epoch": 0.8670930046137129, "grad_norm": 1.6585400940668082, "learning_rate": 9.120874338169095e-07, "loss": 0.4917, "step": 24338 }, { "epoch": 0.867128631740207, "grad_norm": 1.3746976445865693, "learning_rate": 9.116060130078774e-07, "loss": 0.4054, "step": 24339 }, { "epoch": 0.8671642588667011, "grad_norm": 2.5361491645137435, "learning_rate": 9.111247132171652e-07, "loss": 0.5008, "step": 24340 }, { "epoch": 0.8671998859931952, "grad_norm": 1.6956208664797363, "learning_rate": 9.106435344511788e-07, "loss": 0.449, "step": 24341 }, { "epoch": 0.8672355131196894, "grad_norm": 1.3521905964458292, "learning_rate": 9.101624767163275e-07, "loss": 0.4236, "step": 24342 }, { "epoch": 0.8672711402461835, "grad_norm": 1.5985965961188242, "learning_rate": 9.096815400190173e-07, "loss": 0.4407, "step": 24343 }, { "epoch": 0.8673067673726775, "grad_norm": 1.3411569577341733, "learning_rate": 9.092007243656497e-07, "loss": 0.4274, "step": 24344 }, { "epoch": 0.8673423944991717, "grad_norm": 1.2220965538760291, "learning_rate": 9.087200297626309e-07, "loss": 0.4271, "step": 24345 }, { "epoch": 0.8673780216256658, "grad_norm": 1.6121880494907854, "learning_rate": 9.082394562163576e-07, "loss": 0.6705, "step": 24346 }, { "epoch": 0.8674136487521599, "grad_norm": 3.237128121315383, "learning_rate": 9.077590037332318e-07, "loss": 0.5296, "step": 24347 }, { "epoch": 0.8674492758786541, "grad_norm": 1.487165385759348, "learning_rate": 9.07278672319647e-07, "loss": 0.4147, "step": 24348 }, { "epoch": 0.8674849030051481, "grad_norm": 1.3588718203603862, "learning_rate": 9.067984619820048e-07, "loss": 0.4066, "step": 24349 }, { "epoch": 0.8675205301316422, "grad_norm": 1.3170862431764312, "learning_rate": 9.063183727266956e-07, "loss": 0.4056, "step": 24350 }, { "epoch": 0.8675561572581364, "grad_norm": 1.4772658285927662, "learning_rate": 9.058384045601154e-07, "loss": 0.4453, "step": 24351 }, { "epoch": 0.8675917843846305, "grad_norm": 1.4581412918506877, "learning_rate": 9.053585574886503e-07, "loss": 0.399, "step": 24352 }, { "epoch": 0.8676274115111245, "grad_norm": 1.1764393483719464, "learning_rate": 9.048788315186941e-07, "loss": 0.3727, "step": 24353 }, { "epoch": 0.8676630386376187, "grad_norm": 1.7737035932741485, "learning_rate": 9.043992266566348e-07, "loss": 0.4163, "step": 24354 }, { "epoch": 0.8676986657641128, "grad_norm": 1.2726355268221405, "learning_rate": 9.039197429088553e-07, "loss": 0.3859, "step": 24355 }, { "epoch": 0.8677342928906069, "grad_norm": 1.2943284127587944, "learning_rate": 9.034403802817426e-07, "loss": 0.4037, "step": 24356 }, { "epoch": 0.8677699200171011, "grad_norm": 1.5045034987204762, "learning_rate": 9.029611387816794e-07, "loss": 0.3745, "step": 24357 }, { "epoch": 0.8678055471435951, "grad_norm": 1.8646185310048056, "learning_rate": 9.024820184150485e-07, "loss": 0.3988, "step": 24358 }, { "epoch": 0.8678411742700892, "grad_norm": 1.2684276584644476, "learning_rate": 9.020030191882256e-07, "loss": 0.3832, "step": 24359 }, { "epoch": 0.8678768013965834, "grad_norm": 1.606146979811594, "learning_rate": 9.015241411075948e-07, "loss": 0.3942, "step": 24360 }, { "epoch": 0.8679124285230775, "grad_norm": 1.5029698572661812, "learning_rate": 9.010453841795275e-07, "loss": 0.3976, "step": 24361 }, { "epoch": 0.8679480556495716, "grad_norm": 1.3677735465550698, "learning_rate": 9.005667484104008e-07, "loss": 0.3852, "step": 24362 }, { "epoch": 0.8679836827760657, "grad_norm": 1.3135341840714185, "learning_rate": 9.000882338065875e-07, "loss": 0.4084, "step": 24363 }, { "epoch": 0.8680193099025598, "grad_norm": 3.3010551165023845, "learning_rate": 8.996098403744602e-07, "loss": 0.4248, "step": 24364 }, { "epoch": 0.8680549370290539, "grad_norm": 2.143598342086639, "learning_rate": 8.991315681203894e-07, "loss": 0.5717, "step": 24365 }, { "epoch": 0.8680905641555481, "grad_norm": 1.7109295671067788, "learning_rate": 8.986534170507421e-07, "loss": 0.5182, "step": 24366 }, { "epoch": 0.8681261912820422, "grad_norm": 1.3907606009985243, "learning_rate": 8.981753871718879e-07, "loss": 0.3912, "step": 24367 }, { "epoch": 0.8681618184085362, "grad_norm": 1.7004249666938354, "learning_rate": 8.976974784901882e-07, "loss": 0.3867, "step": 24368 }, { "epoch": 0.8681974455350304, "grad_norm": 1.321185360243687, "learning_rate": 8.972196910120101e-07, "loss": 0.4134, "step": 24369 }, { "epoch": 0.8682330726615245, "grad_norm": 1.4080919158122005, "learning_rate": 8.96742024743713e-07, "loss": 0.3912, "step": 24370 }, { "epoch": 0.8682686997880186, "grad_norm": 1.3800072040531595, "learning_rate": 8.962644796916586e-07, "loss": 0.3753, "step": 24371 }, { "epoch": 0.8683043269145128, "grad_norm": 1.5971135172783144, "learning_rate": 8.957870558622062e-07, "loss": 0.5629, "step": 24372 }, { "epoch": 0.8683399540410068, "grad_norm": 1.5817318350304235, "learning_rate": 8.953097532617138e-07, "loss": 0.4692, "step": 24373 }, { "epoch": 0.8683755811675009, "grad_norm": 1.53467955255048, "learning_rate": 8.948325718965334e-07, "loss": 0.4402, "step": 24374 }, { "epoch": 0.8684112082939951, "grad_norm": 1.289779754244785, "learning_rate": 8.94355511773023e-07, "loss": 0.3897, "step": 24375 }, { "epoch": 0.8684468354204892, "grad_norm": 1.339206370937171, "learning_rate": 8.938785728975341e-07, "loss": 0.4104, "step": 24376 }, { "epoch": 0.8684824625469832, "grad_norm": 1.3599170449090745, "learning_rate": 8.934017552764151e-07, "loss": 0.4132, "step": 24377 }, { "epoch": 0.8685180896734774, "grad_norm": 1.8184472700169514, "learning_rate": 8.929250589160166e-07, "loss": 0.5158, "step": 24378 }, { "epoch": 0.8685537167999715, "grad_norm": 1.2621155519113303, "learning_rate": 8.924484838226877e-07, "loss": 0.3771, "step": 24379 }, { "epoch": 0.8685893439264656, "grad_norm": 1.4856789557805987, "learning_rate": 8.919720300027735e-07, "loss": 0.4366, "step": 24380 }, { "epoch": 0.8686249710529598, "grad_norm": 1.461877779172311, "learning_rate": 8.914956974626166e-07, "loss": 0.4736, "step": 24381 }, { "epoch": 0.8686605981794538, "grad_norm": 1.3498217699752229, "learning_rate": 8.91019486208563e-07, "loss": 0.4499, "step": 24382 }, { "epoch": 0.8686962253059479, "grad_norm": 1.8099217897671287, "learning_rate": 8.905433962469489e-07, "loss": 0.4612, "step": 24383 }, { "epoch": 0.8687318524324421, "grad_norm": 1.5089205543456599, "learning_rate": 8.900674275841193e-07, "loss": 0.5493, "step": 24384 }, { "epoch": 0.8687674795589362, "grad_norm": 1.5995209929607845, "learning_rate": 8.895915802264076e-07, "loss": 0.5442, "step": 24385 }, { "epoch": 0.8688031066854303, "grad_norm": 1.4874979797750847, "learning_rate": 8.891158541801514e-07, "loss": 0.5123, "step": 24386 }, { "epoch": 0.8688387338119244, "grad_norm": 1.54691882350894, "learning_rate": 8.886402494516866e-07, "loss": 0.4459, "step": 24387 }, { "epoch": 0.8688743609384185, "grad_norm": 1.296318665940317, "learning_rate": 8.881647660473447e-07, "loss": 0.413, "step": 24388 }, { "epoch": 0.8689099880649126, "grad_norm": 1.31344679564452, "learning_rate": 8.876894039734585e-07, "loss": 0.419, "step": 24389 }, { "epoch": 0.8689456151914068, "grad_norm": 1.6280510426370074, "learning_rate": 8.872141632363563e-07, "loss": 0.5151, "step": 24390 }, { "epoch": 0.8689812423179009, "grad_norm": 1.5871864664531463, "learning_rate": 8.867390438423695e-07, "loss": 0.568, "step": 24391 }, { "epoch": 0.8690168694443949, "grad_norm": 1.6605071054371903, "learning_rate": 8.862640457978189e-07, "loss": 0.5193, "step": 24392 }, { "epoch": 0.8690524965708891, "grad_norm": 1.3267302348508745, "learning_rate": 8.857891691090336e-07, "loss": 0.395, "step": 24393 }, { "epoch": 0.8690881236973832, "grad_norm": 1.7820266855824347, "learning_rate": 8.853144137823355e-07, "loss": 0.4793, "step": 24394 }, { "epoch": 0.8691237508238773, "grad_norm": 1.2216848827020863, "learning_rate": 8.848397798240483e-07, "loss": 0.3736, "step": 24395 }, { "epoch": 0.8691593779503715, "grad_norm": 2.487917850570214, "learning_rate": 8.84365267240489e-07, "loss": 0.4795, "step": 24396 }, { "epoch": 0.8691950050768655, "grad_norm": 1.334685651785929, "learning_rate": 8.838908760379805e-07, "loss": 0.4121, "step": 24397 }, { "epoch": 0.8692306322033596, "grad_norm": 1.505701669888522, "learning_rate": 8.834166062228333e-07, "loss": 0.4824, "step": 24398 }, { "epoch": 0.8692662593298538, "grad_norm": 1.6681485764580721, "learning_rate": 8.829424578013668e-07, "loss": 0.5815, "step": 24399 }, { "epoch": 0.8693018864563479, "grad_norm": 1.6875351488066876, "learning_rate": 8.824684307798937e-07, "loss": 0.4488, "step": 24400 }, { "epoch": 0.869337513582842, "grad_norm": 1.3404530553092218, "learning_rate": 8.819945251647267e-07, "loss": 0.3991, "step": 24401 }, { "epoch": 0.8693731407093361, "grad_norm": 1.681200250955159, "learning_rate": 8.815207409621773e-07, "loss": 0.5605, "step": 24402 }, { "epoch": 0.8694087678358302, "grad_norm": 1.670061454635183, "learning_rate": 8.810470781785507e-07, "loss": 0.5289, "step": 24403 }, { "epoch": 0.8694443949623243, "grad_norm": 1.298516522138387, "learning_rate": 8.805735368201584e-07, "loss": 0.3745, "step": 24404 }, { "epoch": 0.8694800220888185, "grad_norm": 1.4730718885852687, "learning_rate": 8.80100116893301e-07, "loss": 0.4066, "step": 24405 }, { "epoch": 0.8695156492153125, "grad_norm": 1.247271433039886, "learning_rate": 8.796268184042867e-07, "loss": 0.3771, "step": 24406 }, { "epoch": 0.8695512763418066, "grad_norm": 1.3354921805818496, "learning_rate": 8.791536413594149e-07, "loss": 0.5383, "step": 24407 }, { "epoch": 0.8695869034683008, "grad_norm": 1.399652047540217, "learning_rate": 8.786805857649871e-07, "loss": 0.4249, "step": 24408 }, { "epoch": 0.8696225305947949, "grad_norm": 1.3470645187141588, "learning_rate": 8.78207651627303e-07, "loss": 0.453, "step": 24409 }, { "epoch": 0.869658157721289, "grad_norm": 1.3187844470534584, "learning_rate": 8.777348389526597e-07, "loss": 0.4047, "step": 24410 }, { "epoch": 0.8696937848477831, "grad_norm": 1.6814325906263787, "learning_rate": 8.772621477473553e-07, "loss": 0.5641, "step": 24411 }, { "epoch": 0.8697294119742772, "grad_norm": 1.343909962250731, "learning_rate": 8.767895780176794e-07, "loss": 0.3977, "step": 24412 }, { "epoch": 0.8697650391007713, "grad_norm": 1.4849422270059487, "learning_rate": 8.76317129769928e-07, "loss": 0.5365, "step": 24413 }, { "epoch": 0.8698006662272655, "grad_norm": 1.7146266266300516, "learning_rate": 8.758448030103905e-07, "loss": 0.5059, "step": 24414 }, { "epoch": 0.8698362933537596, "grad_norm": 1.4602676854789796, "learning_rate": 8.753725977453564e-07, "loss": 0.3945, "step": 24415 }, { "epoch": 0.8698719204802536, "grad_norm": 1.29053881581653, "learning_rate": 8.74900513981114e-07, "loss": 0.4009, "step": 24416 }, { "epoch": 0.8699075476067478, "grad_norm": 2.1064275656047604, "learning_rate": 8.744285517239515e-07, "loss": 0.5194, "step": 24417 }, { "epoch": 0.8699431747332419, "grad_norm": 1.3551034916524076, "learning_rate": 8.739567109801495e-07, "loss": 0.5527, "step": 24418 }, { "epoch": 0.869978801859736, "grad_norm": 3.161385631270051, "learning_rate": 8.73484991755994e-07, "loss": 0.5107, "step": 24419 }, { "epoch": 0.8700144289862302, "grad_norm": 1.4311102185161302, "learning_rate": 8.730133940577623e-07, "loss": 0.3879, "step": 24420 }, { "epoch": 0.8700500561127242, "grad_norm": 5.195816881508192, "learning_rate": 8.725419178917394e-07, "loss": 0.545, "step": 24421 }, { "epoch": 0.8700856832392183, "grad_norm": 1.313841172959757, "learning_rate": 8.72070563264199e-07, "loss": 0.4227, "step": 24422 }, { "epoch": 0.8701213103657125, "grad_norm": 1.8278678474476864, "learning_rate": 8.715993301814174e-07, "loss": 0.4465, "step": 24423 }, { "epoch": 0.8701569374922066, "grad_norm": 1.339511037863039, "learning_rate": 8.711282186496739e-07, "loss": 0.3979, "step": 24424 }, { "epoch": 0.8701925646187006, "grad_norm": 1.5021415394502091, "learning_rate": 8.706572286752368e-07, "loss": 0.5144, "step": 24425 }, { "epoch": 0.8702281917451948, "grad_norm": 1.519561324714024, "learning_rate": 8.701863602643823e-07, "loss": 0.5676, "step": 24426 }, { "epoch": 0.8702638188716889, "grad_norm": 1.7126175445269178, "learning_rate": 8.697156134233764e-07, "loss": 0.4037, "step": 24427 }, { "epoch": 0.870299445998183, "grad_norm": 1.5110064189101127, "learning_rate": 8.692449881584897e-07, "loss": 0.4448, "step": 24428 }, { "epoch": 0.8703350731246772, "grad_norm": 1.4199195622801895, "learning_rate": 8.68774484475986e-07, "loss": 0.4412, "step": 24429 }, { "epoch": 0.8703707002511712, "grad_norm": 2.8871634284371748, "learning_rate": 8.683041023821326e-07, "loss": 0.5163, "step": 24430 }, { "epoch": 0.8704063273776653, "grad_norm": 1.6528272888652624, "learning_rate": 8.678338418831933e-07, "loss": 0.5152, "step": 24431 }, { "epoch": 0.8704419545041595, "grad_norm": 1.519825560089347, "learning_rate": 8.673637029854309e-07, "loss": 0.3849, "step": 24432 }, { "epoch": 0.8704775816306536, "grad_norm": 1.397455787379669, "learning_rate": 8.668936856951026e-07, "loss": 0.4019, "step": 24433 }, { "epoch": 0.8705132087571477, "grad_norm": 1.3044126467582629, "learning_rate": 8.66423790018468e-07, "loss": 0.4387, "step": 24434 }, { "epoch": 0.8705488358836418, "grad_norm": 1.7503746454649343, "learning_rate": 8.659540159617863e-07, "loss": 0.5537, "step": 24435 }, { "epoch": 0.8705844630101359, "grad_norm": 1.2406341485598888, "learning_rate": 8.654843635313104e-07, "loss": 0.3932, "step": 24436 }, { "epoch": 0.87062009013663, "grad_norm": 1.5427771882784458, "learning_rate": 8.650148327332941e-07, "loss": 0.4052, "step": 24437 }, { "epoch": 0.8706557172631242, "grad_norm": 1.5243604687912242, "learning_rate": 8.645454235739903e-07, "loss": 0.4616, "step": 24438 }, { "epoch": 0.8706913443896183, "grad_norm": 2.276209176373623, "learning_rate": 8.640761360596517e-07, "loss": 0.3885, "step": 24439 }, { "epoch": 0.8707269715161123, "grad_norm": 1.9090060242742775, "learning_rate": 8.636069701965222e-07, "loss": 0.3955, "step": 24440 }, { "epoch": 0.8707625986426065, "grad_norm": 1.5612753725909436, "learning_rate": 8.631379259908546e-07, "loss": 0.4456, "step": 24441 }, { "epoch": 0.8707982257691006, "grad_norm": 1.4233712680869801, "learning_rate": 8.626690034488894e-07, "loss": 0.3885, "step": 24442 }, { "epoch": 0.8708338528955947, "grad_norm": 1.5874911901350823, "learning_rate": 8.622002025768749e-07, "loss": 0.4764, "step": 24443 }, { "epoch": 0.8708694800220889, "grad_norm": 1.7360012322344895, "learning_rate": 8.617315233810508e-07, "loss": 0.5165, "step": 24444 }, { "epoch": 0.8709051071485829, "grad_norm": 1.6200794936142704, "learning_rate": 8.612629658676575e-07, "loss": 0.4227, "step": 24445 }, { "epoch": 0.870940734275077, "grad_norm": 1.56040502791085, "learning_rate": 8.607945300429366e-07, "loss": 0.388, "step": 24446 }, { "epoch": 0.8709763614015712, "grad_norm": 1.8314189550466595, "learning_rate": 8.603262159131232e-07, "loss": 0.5677, "step": 24447 }, { "epoch": 0.8710119885280653, "grad_norm": 1.4977927307288015, "learning_rate": 8.598580234844567e-07, "loss": 0.3711, "step": 24448 }, { "epoch": 0.8710476156545593, "grad_norm": 1.2393350396741285, "learning_rate": 8.593899527631677e-07, "loss": 0.3968, "step": 24449 }, { "epoch": 0.8710832427810535, "grad_norm": 1.5528832652868994, "learning_rate": 8.589220037554924e-07, "loss": 0.4201, "step": 24450 }, { "epoch": 0.8711188699075476, "grad_norm": 1.4759600793810503, "learning_rate": 8.584541764676579e-07, "loss": 0.4105, "step": 24451 }, { "epoch": 0.8711544970340417, "grad_norm": 1.7494228686760143, "learning_rate": 8.579864709058949e-07, "loss": 0.4975, "step": 24452 }, { "epoch": 0.8711901241605359, "grad_norm": 1.54822362896805, "learning_rate": 8.575188870764328e-07, "loss": 0.43, "step": 24453 }, { "epoch": 0.87122575128703, "grad_norm": 1.3623610992645152, "learning_rate": 8.570514249854977e-07, "loss": 0.3876, "step": 24454 }, { "epoch": 0.871261378413524, "grad_norm": 1.5697881599805694, "learning_rate": 8.565840846393126e-07, "loss": 0.5575, "step": 24455 }, { "epoch": 0.8712970055400182, "grad_norm": 1.5745264667453145, "learning_rate": 8.56116866044101e-07, "loss": 0.49, "step": 24456 }, { "epoch": 0.8713326326665123, "grad_norm": 1.4606347619286386, "learning_rate": 8.556497692060861e-07, "loss": 0.5069, "step": 24457 }, { "epoch": 0.8713682597930064, "grad_norm": 1.475111798009423, "learning_rate": 8.55182794131485e-07, "loss": 0.4103, "step": 24458 }, { "epoch": 0.8714038869195005, "grad_norm": 1.3547032205846965, "learning_rate": 8.547159408265171e-07, "loss": 0.389, "step": 24459 }, { "epoch": 0.8714395140459946, "grad_norm": 1.8550052948991476, "learning_rate": 8.542492092973987e-07, "loss": 0.5593, "step": 24460 }, { "epoch": 0.8714751411724887, "grad_norm": 1.387780132059395, "learning_rate": 8.537825995503435e-07, "loss": 0.4097, "step": 24461 }, { "epoch": 0.8715107682989829, "grad_norm": 1.420005097900833, "learning_rate": 8.533161115915655e-07, "loss": 0.4819, "step": 24462 }, { "epoch": 0.871546395425477, "grad_norm": 1.1461693967168078, "learning_rate": 8.528497454272788e-07, "loss": 0.348, "step": 24463 }, { "epoch": 0.871582022551971, "grad_norm": 3.004143441025792, "learning_rate": 8.523835010636894e-07, "loss": 0.4701, "step": 24464 }, { "epoch": 0.8716176496784652, "grad_norm": 1.3052091866544777, "learning_rate": 8.519173785070101e-07, "loss": 0.3729, "step": 24465 }, { "epoch": 0.8716532768049593, "grad_norm": 1.3273161300004632, "learning_rate": 8.514513777634426e-07, "loss": 0.5326, "step": 24466 }, { "epoch": 0.8716889039314534, "grad_norm": 1.422222481028166, "learning_rate": 8.509854988391941e-07, "loss": 0.379, "step": 24467 }, { "epoch": 0.8717245310579476, "grad_norm": 1.650837072506684, "learning_rate": 8.505197417404687e-07, "loss": 0.5532, "step": 24468 }, { "epoch": 0.8717601581844416, "grad_norm": 2.35195411345989, "learning_rate": 8.50054106473468e-07, "loss": 0.5139, "step": 24469 }, { "epoch": 0.8717957853109357, "grad_norm": 1.4712993131075571, "learning_rate": 8.495885930443936e-07, "loss": 0.4884, "step": 24470 }, { "epoch": 0.8718314124374299, "grad_norm": 1.5453576709381707, "learning_rate": 8.491232014594419e-07, "loss": 0.3766, "step": 24471 }, { "epoch": 0.871867039563924, "grad_norm": 1.8312457585557296, "learning_rate": 8.486579317248122e-07, "loss": 0.5808, "step": 24472 }, { "epoch": 0.871902666690418, "grad_norm": 1.7964618667022083, "learning_rate": 8.481927838466975e-07, "loss": 0.5598, "step": 24473 }, { "epoch": 0.8719382938169122, "grad_norm": 1.234255109106011, "learning_rate": 8.477277578312936e-07, "loss": 0.3566, "step": 24474 }, { "epoch": 0.8719739209434063, "grad_norm": 1.5488572483863396, "learning_rate": 8.472628536847882e-07, "loss": 0.4635, "step": 24475 }, { "epoch": 0.8720095480699004, "grad_norm": 1.4900927914084985, "learning_rate": 8.467980714133795e-07, "loss": 0.5314, "step": 24476 }, { "epoch": 0.8720451751963946, "grad_norm": 1.1977054170519945, "learning_rate": 8.463334110232491e-07, "loss": 0.3418, "step": 24477 }, { "epoch": 0.8720808023228886, "grad_norm": 1.3630834991799496, "learning_rate": 8.458688725205899e-07, "loss": 0.429, "step": 24478 }, { "epoch": 0.8721164294493827, "grad_norm": 1.8240825167662742, "learning_rate": 8.454044559115837e-07, "loss": 0.4938, "step": 24479 }, { "epoch": 0.8721520565758769, "grad_norm": 1.3276821989792937, "learning_rate": 8.449401612024155e-07, "loss": 0.3753, "step": 24480 }, { "epoch": 0.872187683702371, "grad_norm": 1.764638022339674, "learning_rate": 8.444759883992693e-07, "loss": 0.605, "step": 24481 }, { "epoch": 0.8722233108288651, "grad_norm": 2.028590987359237, "learning_rate": 8.440119375083233e-07, "loss": 0.4985, "step": 24482 }, { "epoch": 0.8722589379553592, "grad_norm": 1.9952392985398888, "learning_rate": 8.435480085357583e-07, "loss": 0.5141, "step": 24483 }, { "epoch": 0.8722945650818533, "grad_norm": 1.2654862337043231, "learning_rate": 8.430842014877505e-07, "loss": 0.348, "step": 24484 }, { "epoch": 0.8723301922083474, "grad_norm": 1.58906781677653, "learning_rate": 8.426205163704793e-07, "loss": 0.5317, "step": 24485 }, { "epoch": 0.8723658193348416, "grad_norm": 1.3445475381001002, "learning_rate": 8.421569531901153e-07, "loss": 0.4779, "step": 24486 }, { "epoch": 0.8724014464613357, "grad_norm": 1.457456246654829, "learning_rate": 8.416935119528336e-07, "loss": 0.3878, "step": 24487 }, { "epoch": 0.8724370735878297, "grad_norm": 1.5679130434232524, "learning_rate": 8.412301926648025e-07, "loss": 0.5241, "step": 24488 }, { "epoch": 0.8724727007143239, "grad_norm": 1.9244805986209919, "learning_rate": 8.407669953321929e-07, "loss": 0.5113, "step": 24489 }, { "epoch": 0.872508327840818, "grad_norm": 1.2077930048365193, "learning_rate": 8.403039199611729e-07, "loss": 0.4071, "step": 24490 }, { "epoch": 0.8725439549673121, "grad_norm": 1.647502746697543, "learning_rate": 8.398409665579077e-07, "loss": 0.4689, "step": 24491 }, { "epoch": 0.8725795820938063, "grad_norm": 2.2199461279963146, "learning_rate": 8.393781351285657e-07, "loss": 0.5099, "step": 24492 }, { "epoch": 0.8726152092203003, "grad_norm": 1.369674605780055, "learning_rate": 8.389154256793042e-07, "loss": 0.3643, "step": 24493 }, { "epoch": 0.8726508363467944, "grad_norm": 1.7280766252209818, "learning_rate": 8.384528382162882e-07, "loss": 0.5983, "step": 24494 }, { "epoch": 0.8726864634732886, "grad_norm": 1.3616897935463292, "learning_rate": 8.379903727456751e-07, "loss": 0.4101, "step": 24495 }, { "epoch": 0.8727220905997827, "grad_norm": 1.5492229349449875, "learning_rate": 8.375280292736254e-07, "loss": 0.4852, "step": 24496 }, { "epoch": 0.8727577177262767, "grad_norm": 1.743859325987834, "learning_rate": 8.370658078062933e-07, "loss": 0.4789, "step": 24497 }, { "epoch": 0.8727933448527709, "grad_norm": 1.9732945388701981, "learning_rate": 8.366037083498335e-07, "loss": 0.4894, "step": 24498 }, { "epoch": 0.872828971979265, "grad_norm": 2.978926321399384, "learning_rate": 8.361417309104003e-07, "loss": 0.47, "step": 24499 }, { "epoch": 0.8728645991057591, "grad_norm": 1.34319487414639, "learning_rate": 8.356798754941475e-07, "loss": 0.4292, "step": 24500 }, { "epoch": 0.8729002262322533, "grad_norm": 1.4896895852085419, "learning_rate": 8.352181421072203e-07, "loss": 0.3947, "step": 24501 }, { "epoch": 0.8729358533587473, "grad_norm": 1.2387866657406832, "learning_rate": 8.347565307557693e-07, "loss": 0.3811, "step": 24502 }, { "epoch": 0.8729714804852414, "grad_norm": 1.1752029998996367, "learning_rate": 8.342950414459439e-07, "loss": 0.361, "step": 24503 }, { "epoch": 0.8730071076117356, "grad_norm": 1.3888888592406192, "learning_rate": 8.338336741838837e-07, "loss": 0.3753, "step": 24504 }, { "epoch": 0.8730427347382297, "grad_norm": 1.4255880151414004, "learning_rate": 8.33372428975735e-07, "loss": 0.3949, "step": 24505 }, { "epoch": 0.8730783618647238, "grad_norm": 1.6550642390220993, "learning_rate": 8.329113058276394e-07, "loss": 0.4308, "step": 24506 }, { "epoch": 0.873113988991218, "grad_norm": 1.2549691743111513, "learning_rate": 8.3245030474574e-07, "loss": 0.393, "step": 24507 }, { "epoch": 0.873149616117712, "grad_norm": 1.5110650130026448, "learning_rate": 8.319894257361694e-07, "loss": 0.5079, "step": 24508 }, { "epoch": 0.8731852432442061, "grad_norm": 1.9714002915521633, "learning_rate": 8.315286688050705e-07, "loss": 0.4958, "step": 24509 }, { "epoch": 0.8732208703707003, "grad_norm": 1.6890484631498746, "learning_rate": 8.310680339585731e-07, "loss": 0.4669, "step": 24510 }, { "epoch": 0.8732564974971944, "grad_norm": 1.5090669090331814, "learning_rate": 8.306075212028153e-07, "loss": 0.4489, "step": 24511 }, { "epoch": 0.8732921246236884, "grad_norm": 1.485895656042163, "learning_rate": 8.301471305439246e-07, "loss": 0.3884, "step": 24512 }, { "epoch": 0.8733277517501826, "grad_norm": 1.6817140924766034, "learning_rate": 8.296868619880372e-07, "loss": 0.4837, "step": 24513 }, { "epoch": 0.8733633788766767, "grad_norm": 1.6772266998382666, "learning_rate": 8.29226715541277e-07, "loss": 0.4584, "step": 24514 }, { "epoch": 0.8733990060031708, "grad_norm": 1.7897057994628027, "learning_rate": 8.287666912097736e-07, "loss": 0.5437, "step": 24515 }, { "epoch": 0.873434633129665, "grad_norm": 1.7453398818672605, "learning_rate": 8.283067889996543e-07, "loss": 0.5485, "step": 24516 }, { "epoch": 0.873470260256159, "grad_norm": 1.326293926750796, "learning_rate": 8.278470089170377e-07, "loss": 0.399, "step": 24517 }, { "epoch": 0.8735058873826531, "grad_norm": 1.4603643354223315, "learning_rate": 8.273873509680519e-07, "loss": 0.3828, "step": 24518 }, { "epoch": 0.8735415145091473, "grad_norm": 1.558894010304302, "learning_rate": 8.269278151588123e-07, "loss": 0.5195, "step": 24519 }, { "epoch": 0.8735771416356414, "grad_norm": 2.3413794063039113, "learning_rate": 8.264684014954405e-07, "loss": 0.4877, "step": 24520 }, { "epoch": 0.8736127687621354, "grad_norm": 1.707145020313542, "learning_rate": 8.260091099840539e-07, "loss": 0.4067, "step": 24521 }, { "epoch": 0.8736483958886296, "grad_norm": 1.6447829216818053, "learning_rate": 8.255499406307698e-07, "loss": 0.4181, "step": 24522 }, { "epoch": 0.8736840230151237, "grad_norm": 1.2444840873485394, "learning_rate": 8.250908934417001e-07, "loss": 0.3625, "step": 24523 }, { "epoch": 0.8737196501416178, "grad_norm": 1.147268509565355, "learning_rate": 8.246319684229586e-07, "loss": 0.3565, "step": 24524 }, { "epoch": 0.873755277268112, "grad_norm": 1.3791676911832855, "learning_rate": 8.241731655806539e-07, "loss": 0.3509, "step": 24525 }, { "epoch": 0.873790904394606, "grad_norm": 1.5794007398074206, "learning_rate": 8.237144849208978e-07, "loss": 0.4799, "step": 24526 }, { "epoch": 0.8738265315211001, "grad_norm": 1.6396186188326884, "learning_rate": 8.232559264497975e-07, "loss": 0.5458, "step": 24527 }, { "epoch": 0.8738621586475943, "grad_norm": 1.22607996602278, "learning_rate": 8.22797490173457e-07, "loss": 0.4042, "step": 24528 }, { "epoch": 0.8738977857740884, "grad_norm": 1.3028818688447918, "learning_rate": 8.22339176097986e-07, "loss": 0.5178, "step": 24529 }, { "epoch": 0.8739334129005825, "grad_norm": 2.142737551816529, "learning_rate": 8.218809842294806e-07, "loss": 0.3809, "step": 24530 }, { "epoch": 0.8739690400270766, "grad_norm": 1.5573511488503446, "learning_rate": 8.214229145740471e-07, "loss": 0.496, "step": 24531 }, { "epoch": 0.8740046671535707, "grad_norm": 1.2951906281047103, "learning_rate": 8.209649671377817e-07, "loss": 0.3968, "step": 24532 }, { "epoch": 0.8740402942800648, "grad_norm": 2.889803858425146, "learning_rate": 8.205071419267851e-07, "loss": 0.3891, "step": 24533 }, { "epoch": 0.874075921406559, "grad_norm": 2.066745438570073, "learning_rate": 8.200494389471503e-07, "loss": 0.4324, "step": 24534 }, { "epoch": 0.8741115485330531, "grad_norm": 1.4060173686419333, "learning_rate": 8.195918582049734e-07, "loss": 0.5018, "step": 24535 }, { "epoch": 0.8741471756595471, "grad_norm": 1.3282757980566495, "learning_rate": 8.191343997063472e-07, "loss": 0.4195, "step": 24536 }, { "epoch": 0.8741828027860413, "grad_norm": 1.3701870160053604, "learning_rate": 8.186770634573637e-07, "loss": 0.4078, "step": 24537 }, { "epoch": 0.8742184299125354, "grad_norm": 1.4025786632117108, "learning_rate": 8.182198494641147e-07, "loss": 0.4243, "step": 24538 }, { "epoch": 0.8742540570390295, "grad_norm": 1.5446799625894738, "learning_rate": 8.177627577326841e-07, "loss": 0.5515, "step": 24539 }, { "epoch": 0.8742896841655237, "grad_norm": 1.2548275600446985, "learning_rate": 8.173057882691615e-07, "loss": 0.39, "step": 24540 }, { "epoch": 0.8743253112920177, "grad_norm": 1.7010884974865599, "learning_rate": 8.168489410796298e-07, "loss": 0.3956, "step": 24541 }, { "epoch": 0.8743609384185118, "grad_norm": 1.6359016311080403, "learning_rate": 8.163922161701732e-07, "loss": 0.5145, "step": 24542 }, { "epoch": 0.874396565545006, "grad_norm": 1.570240903474626, "learning_rate": 8.159356135468721e-07, "loss": 0.5166, "step": 24543 }, { "epoch": 0.8744321926715001, "grad_norm": 1.60070056858566, "learning_rate": 8.154791332158107e-07, "loss": 0.5358, "step": 24544 }, { "epoch": 0.8744678197979942, "grad_norm": 1.5466898652945937, "learning_rate": 8.15022775183062e-07, "loss": 0.3772, "step": 24545 }, { "epoch": 0.8745034469244883, "grad_norm": 1.92979532364058, "learning_rate": 8.145665394547064e-07, "loss": 0.5328, "step": 24546 }, { "epoch": 0.8745390740509824, "grad_norm": 1.612199272079609, "learning_rate": 8.14110426036816e-07, "loss": 0.5287, "step": 24547 }, { "epoch": 0.8745747011774765, "grad_norm": 1.2802851394635597, "learning_rate": 8.136544349354669e-07, "loss": 0.3821, "step": 24548 }, { "epoch": 0.8746103283039707, "grad_norm": 2.575304283090421, "learning_rate": 8.131985661567287e-07, "loss": 0.415, "step": 24549 }, { "epoch": 0.8746459554304647, "grad_norm": 1.405188287885117, "learning_rate": 8.127428197066711e-07, "loss": 0.3776, "step": 24550 }, { "epoch": 0.8746815825569588, "grad_norm": 1.4153386025546884, "learning_rate": 8.122871955913681e-07, "loss": 0.4936, "step": 24551 }, { "epoch": 0.874717209683453, "grad_norm": 1.582229594786983, "learning_rate": 8.118316938168813e-07, "loss": 0.5257, "step": 24552 }, { "epoch": 0.8747528368099471, "grad_norm": 1.5211108274595182, "learning_rate": 8.113763143892784e-07, "loss": 0.4547, "step": 24553 }, { "epoch": 0.8747884639364412, "grad_norm": 1.3881531686732644, "learning_rate": 8.109210573146209e-07, "loss": 0.4102, "step": 24554 }, { "epoch": 0.8748240910629353, "grad_norm": 1.372746703600935, "learning_rate": 8.104659225989753e-07, "loss": 0.4105, "step": 24555 }, { "epoch": 0.8748597181894294, "grad_norm": 2.012516599548114, "learning_rate": 8.100109102483955e-07, "loss": 0.5366, "step": 24556 }, { "epoch": 0.8748953453159235, "grad_norm": 1.3352708635515658, "learning_rate": 8.095560202689446e-07, "loss": 0.421, "step": 24557 }, { "epoch": 0.8749309724424177, "grad_norm": 1.8159952505535386, "learning_rate": 8.091012526666797e-07, "loss": 0.5207, "step": 24558 }, { "epoch": 0.8749665995689118, "grad_norm": 1.2691143455937468, "learning_rate": 8.086466074476562e-07, "loss": 0.4464, "step": 24559 }, { "epoch": 0.8750022266954058, "grad_norm": 1.514647221562793, "learning_rate": 8.08192084617927e-07, "loss": 0.4837, "step": 24560 }, { "epoch": 0.8750378538219, "grad_norm": 1.7660895143007518, "learning_rate": 8.077376841835438e-07, "loss": 0.5535, "step": 24561 }, { "epoch": 0.8750734809483941, "grad_norm": 1.4035332272671617, "learning_rate": 8.072834061505608e-07, "loss": 0.4257, "step": 24562 }, { "epoch": 0.8751091080748882, "grad_norm": 1.7388951835465276, "learning_rate": 8.068292505250231e-07, "loss": 0.4739, "step": 24563 }, { "epoch": 0.8751447352013824, "grad_norm": 1.5983512808014673, "learning_rate": 8.063752173129791e-07, "loss": 0.5004, "step": 24564 }, { "epoch": 0.8751803623278764, "grad_norm": 2.2933009583316593, "learning_rate": 8.059213065204741e-07, "loss": 0.4791, "step": 24565 }, { "epoch": 0.8752159894543705, "grad_norm": 1.2689915003199184, "learning_rate": 8.054675181535565e-07, "loss": 0.4067, "step": 24566 }, { "epoch": 0.8752516165808647, "grad_norm": 1.4638756106457582, "learning_rate": 8.050138522182627e-07, "loss": 0.4853, "step": 24567 }, { "epoch": 0.8752872437073588, "grad_norm": 1.2109646825940117, "learning_rate": 8.045603087206388e-07, "loss": 0.3611, "step": 24568 }, { "epoch": 0.8753228708338529, "grad_norm": 1.234164616494682, "learning_rate": 8.041068876667191e-07, "loss": 0.3589, "step": 24569 }, { "epoch": 0.875358497960347, "grad_norm": 1.4181867298184074, "learning_rate": 8.036535890625463e-07, "loss": 0.4373, "step": 24570 }, { "epoch": 0.8753941250868411, "grad_norm": 1.1849137044393367, "learning_rate": 8.032004129141513e-07, "loss": 0.3928, "step": 24571 }, { "epoch": 0.8754297522133352, "grad_norm": 1.2924938042948373, "learning_rate": 8.027473592275714e-07, "loss": 0.3802, "step": 24572 }, { "epoch": 0.8754653793398294, "grad_norm": 1.1512843512720219, "learning_rate": 8.022944280088385e-07, "loss": 0.3673, "step": 24573 }, { "epoch": 0.8755010064663235, "grad_norm": 1.7578310172361777, "learning_rate": 8.018416192639844e-07, "loss": 0.5127, "step": 24574 }, { "epoch": 0.8755366335928175, "grad_norm": 1.3506126913304193, "learning_rate": 8.01388932999041e-07, "loss": 0.4123, "step": 24575 }, { "epoch": 0.8755722607193117, "grad_norm": 1.2975429243889791, "learning_rate": 8.009363692200301e-07, "loss": 0.4853, "step": 24576 }, { "epoch": 0.8756078878458058, "grad_norm": 1.3354336538227092, "learning_rate": 8.004839279329834e-07, "loss": 0.3787, "step": 24577 }, { "epoch": 0.8756435149722999, "grad_norm": 1.3256083171047588, "learning_rate": 8.00031609143922e-07, "loss": 0.4291, "step": 24578 }, { "epoch": 0.875679142098794, "grad_norm": 1.614171290529822, "learning_rate": 7.995794128588696e-07, "loss": 0.5394, "step": 24579 }, { "epoch": 0.8757147692252881, "grad_norm": 1.8127878299154432, "learning_rate": 7.991273390838483e-07, "loss": 0.5694, "step": 24580 }, { "epoch": 0.8757503963517822, "grad_norm": 1.3912105011947176, "learning_rate": 7.986753878248798e-07, "loss": 0.3981, "step": 24581 }, { "epoch": 0.8757860234782764, "grad_norm": 1.7166375702123786, "learning_rate": 7.982235590879772e-07, "loss": 0.4749, "step": 24582 }, { "epoch": 0.8758216506047705, "grad_norm": 1.336974222271873, "learning_rate": 7.9777185287916e-07, "loss": 0.4383, "step": 24583 }, { "epoch": 0.8758572777312645, "grad_norm": 1.570092142675194, "learning_rate": 7.973202692044447e-07, "loss": 0.4223, "step": 24584 }, { "epoch": 0.8758929048577587, "grad_norm": 1.2098750251392665, "learning_rate": 7.968688080698394e-07, "loss": 0.3685, "step": 24585 }, { "epoch": 0.8759285319842528, "grad_norm": 1.7564156091768652, "learning_rate": 7.964174694813609e-07, "loss": 0.5106, "step": 24586 }, { "epoch": 0.8759641591107469, "grad_norm": 1.3902900886113285, "learning_rate": 7.959662534450152e-07, "loss": 0.5042, "step": 24587 }, { "epoch": 0.8759997862372411, "grad_norm": 1.6260742246622426, "learning_rate": 7.955151599668121e-07, "loss": 0.4695, "step": 24588 }, { "epoch": 0.8760354133637351, "grad_norm": 1.6523070309192007, "learning_rate": 7.950641890527577e-07, "loss": 0.5637, "step": 24589 }, { "epoch": 0.8760710404902292, "grad_norm": 1.4249364895515282, "learning_rate": 7.946133407088596e-07, "loss": 0.3924, "step": 24590 }, { "epoch": 0.8761066676167234, "grad_norm": 1.6369826380614807, "learning_rate": 7.941626149411174e-07, "loss": 0.4776, "step": 24591 }, { "epoch": 0.8761422947432175, "grad_norm": 1.7595667548092413, "learning_rate": 7.937120117555363e-07, "loss": 0.5206, "step": 24592 }, { "epoch": 0.8761779218697116, "grad_norm": 1.9043582604276927, "learning_rate": 7.932615311581127e-07, "loss": 0.5974, "step": 24593 }, { "epoch": 0.8762135489962057, "grad_norm": 1.57378160967585, "learning_rate": 7.92811173154846e-07, "loss": 0.482, "step": 24594 }, { "epoch": 0.8762491761226998, "grad_norm": 2.0304576735776627, "learning_rate": 7.92360937751735e-07, "loss": 0.5309, "step": 24595 }, { "epoch": 0.8762848032491939, "grad_norm": 1.8963066412514982, "learning_rate": 7.919108249547736e-07, "loss": 0.5358, "step": 24596 }, { "epoch": 0.8763204303756881, "grad_norm": 1.2293545556008674, "learning_rate": 7.914608347699571e-07, "loss": 0.4426, "step": 24597 }, { "epoch": 0.8763560575021822, "grad_norm": 1.6375466652752249, "learning_rate": 7.910109672032751e-07, "loss": 0.3948, "step": 24598 }, { "epoch": 0.8763916846286762, "grad_norm": 1.2166596330398245, "learning_rate": 7.905612222607195e-07, "loss": 0.3616, "step": 24599 }, { "epoch": 0.8764273117551704, "grad_norm": 1.8566126231979743, "learning_rate": 7.901115999482767e-07, "loss": 0.5075, "step": 24600 }, { "epoch": 0.8764629388816645, "grad_norm": 1.8417397321035358, "learning_rate": 7.896621002719362e-07, "loss": 0.5297, "step": 24601 }, { "epoch": 0.8764985660081586, "grad_norm": 1.48573528803719, "learning_rate": 7.892127232376811e-07, "loss": 0.401, "step": 24602 }, { "epoch": 0.8765341931346528, "grad_norm": 1.549822236290135, "learning_rate": 7.887634688515e-07, "loss": 0.387, "step": 24603 }, { "epoch": 0.8765698202611468, "grad_norm": 1.2256133879817555, "learning_rate": 7.88314337119368e-07, "loss": 0.4026, "step": 24604 }, { "epoch": 0.8766054473876409, "grad_norm": 1.612128933845853, "learning_rate": 7.878653280472726e-07, "loss": 0.4139, "step": 24605 }, { "epoch": 0.8766410745141351, "grad_norm": 1.3707125093340369, "learning_rate": 7.874164416411856e-07, "loss": 0.4336, "step": 24606 }, { "epoch": 0.8766767016406292, "grad_norm": 1.5877581808238663, "learning_rate": 7.86967677907089e-07, "loss": 0.5439, "step": 24607 }, { "epoch": 0.8767123287671232, "grad_norm": 1.118346254914379, "learning_rate": 7.86519036850959e-07, "loss": 0.4141, "step": 24608 }, { "epoch": 0.8767479558936174, "grad_norm": 1.83410033623019, "learning_rate": 7.860705184787665e-07, "loss": 0.6541, "step": 24609 }, { "epoch": 0.8767835830201115, "grad_norm": 1.4830281702616708, "learning_rate": 7.856221227964844e-07, "loss": 0.5462, "step": 24610 }, { "epoch": 0.8768192101466056, "grad_norm": 1.4147816232205408, "learning_rate": 7.851738498100848e-07, "loss": 0.3797, "step": 24611 }, { "epoch": 0.8768548372730998, "grad_norm": 1.9778619159225774, "learning_rate": 7.847256995255381e-07, "loss": 0.5218, "step": 24612 }, { "epoch": 0.8768904643995938, "grad_norm": 1.6428900986367212, "learning_rate": 7.842776719488076e-07, "loss": 0.4902, "step": 24613 }, { "epoch": 0.8769260915260879, "grad_norm": 1.370092544901333, "learning_rate": 7.83829767085863e-07, "loss": 0.3889, "step": 24614 }, { "epoch": 0.8769617186525821, "grad_norm": 1.3787760739118564, "learning_rate": 7.833819849426649e-07, "loss": 0.4003, "step": 24615 }, { "epoch": 0.8769973457790762, "grad_norm": 1.6597713613197331, "learning_rate": 7.829343255251787e-07, "loss": 0.4993, "step": 24616 }, { "epoch": 0.8770329729055703, "grad_norm": 1.4458404209034856, "learning_rate": 7.824867888393628e-07, "loss": 0.428, "step": 24617 }, { "epoch": 0.8770686000320644, "grad_norm": 1.626521911786727, "learning_rate": 7.820393748911792e-07, "loss": 0.5145, "step": 24618 }, { "epoch": 0.8771042271585585, "grad_norm": 1.491972508473421, "learning_rate": 7.815920836865854e-07, "loss": 0.5105, "step": 24619 }, { "epoch": 0.8771398542850526, "grad_norm": 1.1726977612039315, "learning_rate": 7.811449152315342e-07, "loss": 0.3571, "step": 24620 }, { "epoch": 0.8771754814115468, "grad_norm": 1.6183239619635776, "learning_rate": 7.806978695319844e-07, "loss": 0.522, "step": 24621 }, { "epoch": 0.8772111085380409, "grad_norm": 1.4519003477721273, "learning_rate": 7.802509465938846e-07, "loss": 0.3499, "step": 24622 }, { "epoch": 0.8772467356645349, "grad_norm": 1.4224555937538252, "learning_rate": 7.798041464231898e-07, "loss": 0.4272, "step": 24623 }, { "epoch": 0.8772823627910291, "grad_norm": 1.3578450235314563, "learning_rate": 7.793574690258454e-07, "loss": 0.4459, "step": 24624 }, { "epoch": 0.8773179899175232, "grad_norm": 1.4925742477538173, "learning_rate": 7.789109144078e-07, "loss": 0.4304, "step": 24625 }, { "epoch": 0.8773536170440173, "grad_norm": 2.047351742700763, "learning_rate": 7.78464482575002e-07, "loss": 0.4919, "step": 24626 }, { "epoch": 0.8773892441705115, "grad_norm": 1.715131811739932, "learning_rate": 7.780181735333969e-07, "loss": 0.4921, "step": 24627 }, { "epoch": 0.8774248712970055, "grad_norm": 1.2665682705773569, "learning_rate": 7.775719872889232e-07, "loss": 0.4012, "step": 24628 }, { "epoch": 0.8774604984234996, "grad_norm": 2.4552409265728823, "learning_rate": 7.771259238475248e-07, "loss": 0.5375, "step": 24629 }, { "epoch": 0.8774961255499938, "grad_norm": 1.2949797972333223, "learning_rate": 7.766799832151428e-07, "loss": 0.3837, "step": 24630 }, { "epoch": 0.8775317526764879, "grad_norm": 2.0512847129819645, "learning_rate": 7.762341653977112e-07, "loss": 0.5736, "step": 24631 }, { "epoch": 0.8775673798029819, "grad_norm": 1.736111101905954, "learning_rate": 7.757884704011698e-07, "loss": 0.4939, "step": 24632 }, { "epoch": 0.8776030069294761, "grad_norm": 1.7893718210020129, "learning_rate": 7.753428982314515e-07, "loss": 0.5596, "step": 24633 }, { "epoch": 0.8776386340559702, "grad_norm": 2.0779174036350705, "learning_rate": 7.748974488944927e-07, "loss": 0.4753, "step": 24634 }, { "epoch": 0.8776742611824643, "grad_norm": 1.2974415858832218, "learning_rate": 7.744521223962198e-07, "loss": 0.3949, "step": 24635 }, { "epoch": 0.8777098883089585, "grad_norm": 1.4144131542431333, "learning_rate": 7.740069187425669e-07, "loss": 0.4635, "step": 24636 }, { "epoch": 0.8777455154354525, "grad_norm": 1.5338922868228266, "learning_rate": 7.735618379394594e-07, "loss": 0.4933, "step": 24637 }, { "epoch": 0.8777811425619466, "grad_norm": 1.6435996449089418, "learning_rate": 7.731168799928257e-07, "loss": 0.5018, "step": 24638 }, { "epoch": 0.8778167696884408, "grad_norm": 1.4444337769781832, "learning_rate": 7.726720449085878e-07, "loss": 0.4275, "step": 24639 }, { "epoch": 0.8778523968149349, "grad_norm": 1.8309411735873735, "learning_rate": 7.722273326926733e-07, "loss": 0.4444, "step": 24640 }, { "epoch": 0.877888023941429, "grad_norm": 1.4213328467913628, "learning_rate": 7.717827433510005e-07, "loss": 0.3853, "step": 24641 }, { "epoch": 0.8779236510679231, "grad_norm": 1.6890420488081548, "learning_rate": 7.713382768894905e-07, "loss": 0.5371, "step": 24642 }, { "epoch": 0.8779592781944172, "grad_norm": 1.2242398345622287, "learning_rate": 7.708939333140642e-07, "loss": 0.4276, "step": 24643 }, { "epoch": 0.8779949053209113, "grad_norm": 1.332689205687046, "learning_rate": 7.704497126306332e-07, "loss": 0.4144, "step": 24644 }, { "epoch": 0.8780305324474055, "grad_norm": 1.1559627038254716, "learning_rate": 7.700056148451174e-07, "loss": 0.3805, "step": 24645 }, { "epoch": 0.8780661595738996, "grad_norm": 1.4132288559185398, "learning_rate": 7.695616399634265e-07, "loss": 0.5744, "step": 24646 }, { "epoch": 0.8781017867003936, "grad_norm": 2.0245895510083227, "learning_rate": 7.691177879914746e-07, "loss": 0.4961, "step": 24647 }, { "epoch": 0.8781374138268878, "grad_norm": 1.5663522147221252, "learning_rate": 7.686740589351704e-07, "loss": 0.492, "step": 24648 }, { "epoch": 0.8781730409533819, "grad_norm": 1.5597154513637255, "learning_rate": 7.682304528004258e-07, "loss": 0.4106, "step": 24649 }, { "epoch": 0.878208668079876, "grad_norm": 1.7116480852159666, "learning_rate": 7.67786969593144e-07, "loss": 0.5451, "step": 24650 }, { "epoch": 0.8782442952063702, "grad_norm": 1.4948235396266818, "learning_rate": 7.673436093192322e-07, "loss": 0.4842, "step": 24651 }, { "epoch": 0.8782799223328642, "grad_norm": 1.665689522556496, "learning_rate": 7.669003719845925e-07, "loss": 0.4241, "step": 24652 }, { "epoch": 0.8783155494593583, "grad_norm": 1.3656692810030167, "learning_rate": 7.664572575951279e-07, "loss": 0.4288, "step": 24653 }, { "epoch": 0.8783511765858525, "grad_norm": 1.4099781100692101, "learning_rate": 7.660142661567383e-07, "loss": 0.3588, "step": 24654 }, { "epoch": 0.8783868037123466, "grad_norm": 2.292882454760567, "learning_rate": 7.655713976753232e-07, "loss": 0.5326, "step": 24655 }, { "epoch": 0.8784224308388406, "grad_norm": 1.5724729046517394, "learning_rate": 7.651286521567813e-07, "loss": 0.4235, "step": 24656 }, { "epoch": 0.8784580579653348, "grad_norm": 1.558322113726462, "learning_rate": 7.646860296070047e-07, "loss": 0.4388, "step": 24657 }, { "epoch": 0.8784936850918289, "grad_norm": 1.3265074938567327, "learning_rate": 7.642435300318906e-07, "loss": 0.4188, "step": 24658 }, { "epoch": 0.878529312218323, "grad_norm": 1.4276984593822497, "learning_rate": 7.638011534373269e-07, "loss": 0.5533, "step": 24659 }, { "epoch": 0.8785649393448172, "grad_norm": 1.5509280252427058, "learning_rate": 7.633588998292086e-07, "loss": 0.4203, "step": 24660 }, { "epoch": 0.8786005664713112, "grad_norm": 1.4570861359586813, "learning_rate": 7.629167692134199e-07, "loss": 0.5298, "step": 24661 }, { "epoch": 0.8786361935978053, "grad_norm": 1.4595610393946843, "learning_rate": 7.624747615958516e-07, "loss": 0.4428, "step": 24662 }, { "epoch": 0.8786718207242995, "grad_norm": 1.837215929106079, "learning_rate": 7.620328769823892e-07, "loss": 0.4891, "step": 24663 }, { "epoch": 0.8787074478507936, "grad_norm": 1.513561984305335, "learning_rate": 7.615911153789146e-07, "loss": 0.5842, "step": 24664 }, { "epoch": 0.8787430749772877, "grad_norm": 1.3323891290089558, "learning_rate": 7.611494767913141e-07, "loss": 0.4147, "step": 24665 }, { "epoch": 0.8787787021037818, "grad_norm": 1.2899218660669678, "learning_rate": 7.607079612254642e-07, "loss": 0.427, "step": 24666 }, { "epoch": 0.8788143292302759, "grad_norm": 1.5243832478334838, "learning_rate": 7.602665686872468e-07, "loss": 0.5115, "step": 24667 }, { "epoch": 0.87884995635677, "grad_norm": 1.3368630318432178, "learning_rate": 7.598252991825372e-07, "loss": 0.4195, "step": 24668 }, { "epoch": 0.8788855834832642, "grad_norm": 1.7024519599211427, "learning_rate": 7.593841527172119e-07, "loss": 0.4923, "step": 24669 }, { "epoch": 0.8789212106097583, "grad_norm": 1.9414304641544557, "learning_rate": 7.589431292971461e-07, "loss": 0.4063, "step": 24670 }, { "epoch": 0.8789568377362523, "grad_norm": 2.0362663932773315, "learning_rate": 7.585022289282129e-07, "loss": 0.5193, "step": 24671 }, { "epoch": 0.8789924648627465, "grad_norm": 1.3433145337347179, "learning_rate": 7.580614516162799e-07, "loss": 0.4102, "step": 24672 }, { "epoch": 0.8790280919892406, "grad_norm": 1.6290648444512044, "learning_rate": 7.5762079736722e-07, "loss": 0.5098, "step": 24673 }, { "epoch": 0.8790637191157347, "grad_norm": 1.5689393605172597, "learning_rate": 7.571802661868976e-07, "loss": 0.5503, "step": 24674 }, { "epoch": 0.8790993462422289, "grad_norm": 1.5825315915422855, "learning_rate": 7.567398580811802e-07, "loss": 0.4428, "step": 24675 }, { "epoch": 0.8791349733687229, "grad_norm": 1.296528948943493, "learning_rate": 7.562995730559331e-07, "loss": 0.4192, "step": 24676 }, { "epoch": 0.879170600495217, "grad_norm": 2.338027391982109, "learning_rate": 7.55859411117017e-07, "loss": 0.4919, "step": 24677 }, { "epoch": 0.8792062276217112, "grad_norm": 1.295556483688086, "learning_rate": 7.554193722702962e-07, "loss": 0.3628, "step": 24678 }, { "epoch": 0.8792418547482053, "grad_norm": 1.3922518065060658, "learning_rate": 7.549794565216273e-07, "loss": 0.3867, "step": 24679 }, { "epoch": 0.8792774818746995, "grad_norm": 2.2108225574934446, "learning_rate": 7.545396638768698e-07, "loss": 0.5664, "step": 24680 }, { "epoch": 0.8793131090011935, "grad_norm": 1.2409404348449777, "learning_rate": 7.54099994341877e-07, "loss": 0.4252, "step": 24681 }, { "epoch": 0.8793487361276876, "grad_norm": 1.189205781013433, "learning_rate": 7.536604479225074e-07, "loss": 0.3928, "step": 24682 }, { "epoch": 0.8793843632541818, "grad_norm": 1.3597853578567478, "learning_rate": 7.532210246246097e-07, "loss": 0.3817, "step": 24683 }, { "epoch": 0.8794199903806759, "grad_norm": 1.3533855111343511, "learning_rate": 7.52781724454037e-07, "loss": 0.4355, "step": 24684 }, { "epoch": 0.8794556175071699, "grad_norm": 1.5498001363369958, "learning_rate": 7.52342547416639e-07, "loss": 0.5121, "step": 24685 }, { "epoch": 0.8794912446336641, "grad_norm": 1.5484075843549616, "learning_rate": 7.519034935182656e-07, "loss": 0.4093, "step": 24686 }, { "epoch": 0.8795268717601582, "grad_norm": 1.5656936411812183, "learning_rate": 7.514645627647599e-07, "loss": 0.5174, "step": 24687 }, { "epoch": 0.8795624988866523, "grad_norm": 1.4395928898434824, "learning_rate": 7.51025755161967e-07, "loss": 0.4185, "step": 24688 }, { "epoch": 0.8795981260131465, "grad_norm": 1.3293811703962257, "learning_rate": 7.505870707157337e-07, "loss": 0.4019, "step": 24689 }, { "epoch": 0.8796337531396405, "grad_norm": 1.2242197418765797, "learning_rate": 7.501485094318961e-07, "loss": 0.4074, "step": 24690 }, { "epoch": 0.8796693802661346, "grad_norm": 1.3199288644658664, "learning_rate": 7.497100713162964e-07, "loss": 0.376, "step": 24691 }, { "epoch": 0.8797050073926288, "grad_norm": 1.4809928200862474, "learning_rate": 7.492717563747721e-07, "loss": 0.4338, "step": 24692 }, { "epoch": 0.8797406345191229, "grad_norm": 1.4870338766275644, "learning_rate": 7.488335646131628e-07, "loss": 0.4112, "step": 24693 }, { "epoch": 0.879776261645617, "grad_norm": 1.736486933197218, "learning_rate": 7.483954960372985e-07, "loss": 0.4832, "step": 24694 }, { "epoch": 0.8798118887721111, "grad_norm": 1.3853941254222142, "learning_rate": 7.479575506530168e-07, "loss": 0.4142, "step": 24695 }, { "epoch": 0.8798475158986052, "grad_norm": 1.4002624530601626, "learning_rate": 7.47519728466145e-07, "loss": 0.4707, "step": 24696 }, { "epoch": 0.8798831430250993, "grad_norm": 2.1040462831080706, "learning_rate": 7.470820294825177e-07, "loss": 0.5133, "step": 24697 }, { "epoch": 0.8799187701515935, "grad_norm": 1.7443619342237073, "learning_rate": 7.466444537079575e-07, "loss": 0.4681, "step": 24698 }, { "epoch": 0.8799543972780876, "grad_norm": 1.4562024035377985, "learning_rate": 7.462070011482958e-07, "loss": 0.3882, "step": 24699 }, { "epoch": 0.8799900244045816, "grad_norm": 1.6439862187133414, "learning_rate": 7.457696718093543e-07, "loss": 0.5344, "step": 24700 }, { "epoch": 0.8800256515310758, "grad_norm": 1.335469532983361, "learning_rate": 7.453324656969585e-07, "loss": 0.4062, "step": 24701 }, { "epoch": 0.8800612786575699, "grad_norm": 1.7422289946417704, "learning_rate": 7.448953828169314e-07, "loss": 0.4907, "step": 24702 }, { "epoch": 0.880096905784064, "grad_norm": 1.5052183393085181, "learning_rate": 7.444584231750895e-07, "loss": 0.508, "step": 24703 }, { "epoch": 0.8801325329105582, "grad_norm": 1.8772815638748825, "learning_rate": 7.440215867772548e-07, "loss": 0.5998, "step": 24704 }, { "epoch": 0.8801681600370522, "grad_norm": 1.4477731355024641, "learning_rate": 7.435848736292406e-07, "loss": 0.4181, "step": 24705 }, { "epoch": 0.8802037871635463, "grad_norm": 1.5641156124011795, "learning_rate": 7.431482837368631e-07, "loss": 0.4123, "step": 24706 }, { "epoch": 0.8802394142900405, "grad_norm": 1.464137644478004, "learning_rate": 7.427118171059367e-07, "loss": 0.4136, "step": 24707 }, { "epoch": 0.8802750414165346, "grad_norm": 1.6864196534942708, "learning_rate": 7.422754737422744e-07, "loss": 0.5233, "step": 24708 }, { "epoch": 0.8803106685430286, "grad_norm": 1.6763168077827706, "learning_rate": 7.418392536516839e-07, "loss": 0.5308, "step": 24709 }, { "epoch": 0.8803462956695228, "grad_norm": 2.2206939451686596, "learning_rate": 7.414031568399738e-07, "loss": 0.6054, "step": 24710 }, { "epoch": 0.8803819227960169, "grad_norm": 1.5616909802965198, "learning_rate": 7.409671833129539e-07, "loss": 0.4907, "step": 24711 }, { "epoch": 0.880417549922511, "grad_norm": 1.6216432635866733, "learning_rate": 7.405313330764263e-07, "loss": 0.5301, "step": 24712 }, { "epoch": 0.8804531770490052, "grad_norm": 1.6751178077314939, "learning_rate": 7.400956061361975e-07, "loss": 0.5082, "step": 24713 }, { "epoch": 0.8804888041754992, "grad_norm": 1.3392719925008707, "learning_rate": 7.39660002498065e-07, "loss": 0.3788, "step": 24714 }, { "epoch": 0.8805244313019933, "grad_norm": 1.2249306167480565, "learning_rate": 7.392245221678351e-07, "loss": 0.3888, "step": 24715 }, { "epoch": 0.8805600584284875, "grad_norm": 1.7508118830717907, "learning_rate": 7.387891651513024e-07, "loss": 0.5217, "step": 24716 }, { "epoch": 0.8805956855549816, "grad_norm": 1.4424991060493098, "learning_rate": 7.383539314542665e-07, "loss": 0.4277, "step": 24717 }, { "epoch": 0.8806313126814757, "grad_norm": 1.388879084117896, "learning_rate": 7.379188210825195e-07, "loss": 0.3723, "step": 24718 }, { "epoch": 0.8806669398079698, "grad_norm": 1.4533392089785762, "learning_rate": 7.374838340418588e-07, "loss": 0.4406, "step": 24719 }, { "epoch": 0.8807025669344639, "grad_norm": 2.128907902962419, "learning_rate": 7.370489703380745e-07, "loss": 0.471, "step": 24720 }, { "epoch": 0.880738194060958, "grad_norm": 1.3768283232894973, "learning_rate": 7.366142299769563e-07, "loss": 0.3686, "step": 24721 }, { "epoch": 0.8807738211874522, "grad_norm": 1.3593468500151564, "learning_rate": 7.361796129642951e-07, "loss": 0.4111, "step": 24722 }, { "epoch": 0.8808094483139463, "grad_norm": 1.5781800141161995, "learning_rate": 7.357451193058773e-07, "loss": 0.5153, "step": 24723 }, { "epoch": 0.8808450754404403, "grad_norm": 5.122482117142119, "learning_rate": 7.353107490074896e-07, "loss": 0.4856, "step": 24724 }, { "epoch": 0.8808807025669345, "grad_norm": 1.7403477986264555, "learning_rate": 7.348765020749138e-07, "loss": 0.509, "step": 24725 }, { "epoch": 0.8809163296934286, "grad_norm": 1.36292254384193, "learning_rate": 7.344423785139343e-07, "loss": 0.4613, "step": 24726 }, { "epoch": 0.8809519568199227, "grad_norm": 1.5181587226301712, "learning_rate": 7.340083783303298e-07, "loss": 0.5487, "step": 24727 }, { "epoch": 0.8809875839464169, "grad_norm": 1.4922585330156064, "learning_rate": 7.335745015298801e-07, "loss": 0.4466, "step": 24728 }, { "epoch": 0.8810232110729109, "grad_norm": 1.2925856400975824, "learning_rate": 7.331407481183628e-07, "loss": 0.403, "step": 24729 }, { "epoch": 0.881058838199405, "grad_norm": 1.4377917821485307, "learning_rate": 7.327071181015555e-07, "loss": 0.421, "step": 24730 }, { "epoch": 0.8810944653258992, "grad_norm": 1.2616066960416312, "learning_rate": 7.322736114852281e-07, "loss": 0.3805, "step": 24731 }, { "epoch": 0.8811300924523933, "grad_norm": 1.402690279581025, "learning_rate": 7.31840228275158e-07, "loss": 0.3663, "step": 24732 }, { "epoch": 0.8811657195788873, "grad_norm": 1.4085346682550544, "learning_rate": 7.314069684771108e-07, "loss": 0.4133, "step": 24733 }, { "epoch": 0.8812013467053815, "grad_norm": 1.5713853364267385, "learning_rate": 7.309738320968584e-07, "loss": 0.4951, "step": 24734 }, { "epoch": 0.8812369738318756, "grad_norm": 1.5948397549946236, "learning_rate": 7.305408191401697e-07, "loss": 0.5691, "step": 24735 }, { "epoch": 0.8812726009583697, "grad_norm": 1.9112473705458677, "learning_rate": 7.301079296128066e-07, "loss": 0.4767, "step": 24736 }, { "epoch": 0.8813082280848639, "grad_norm": 1.448709973562237, "learning_rate": 7.296751635205368e-07, "loss": 0.5334, "step": 24737 }, { "epoch": 0.8813438552113579, "grad_norm": 1.7561701087717017, "learning_rate": 7.292425208691212e-07, "loss": 0.5246, "step": 24738 }, { "epoch": 0.881379482337852, "grad_norm": 1.539190900501155, "learning_rate": 7.288100016643229e-07, "loss": 0.5013, "step": 24739 }, { "epoch": 0.8814151094643462, "grad_norm": 1.3330289512582234, "learning_rate": 7.283776059118986e-07, "loss": 0.3736, "step": 24740 }, { "epoch": 0.8814507365908403, "grad_norm": 1.6934552649506998, "learning_rate": 7.279453336176079e-07, "loss": 0.454, "step": 24741 }, { "epoch": 0.8814863637173344, "grad_norm": 1.4392035049126983, "learning_rate": 7.275131847872041e-07, "loss": 0.4809, "step": 24742 }, { "epoch": 0.8815219908438285, "grad_norm": 1.634193046103306, "learning_rate": 7.270811594264438e-07, "loss": 0.5279, "step": 24743 }, { "epoch": 0.8815576179703226, "grad_norm": 1.2221007738886602, "learning_rate": 7.2664925754108e-07, "loss": 0.3456, "step": 24744 }, { "epoch": 0.8815932450968167, "grad_norm": 1.378122989646878, "learning_rate": 7.262174791368615e-07, "loss": 0.5215, "step": 24745 }, { "epoch": 0.8816288722233109, "grad_norm": 1.3739887212975022, "learning_rate": 7.257858242195414e-07, "loss": 0.3978, "step": 24746 }, { "epoch": 0.881664499349805, "grad_norm": 2.5878548725199337, "learning_rate": 7.253542927948643e-07, "loss": 0.4999, "step": 24747 }, { "epoch": 0.881700126476299, "grad_norm": 1.484806010646509, "learning_rate": 7.249228848685785e-07, "loss": 0.3919, "step": 24748 }, { "epoch": 0.8817357536027932, "grad_norm": 1.3176608743352443, "learning_rate": 7.244916004464253e-07, "loss": 0.4115, "step": 24749 }, { "epoch": 0.8817713807292873, "grad_norm": 1.9107023773436873, "learning_rate": 7.240604395341521e-07, "loss": 0.446, "step": 24750 }, { "epoch": 0.8818070078557814, "grad_norm": 1.7145812761722252, "learning_rate": 7.236294021374957e-07, "loss": 0.4683, "step": 24751 }, { "epoch": 0.8818426349822756, "grad_norm": 1.4098821997250217, "learning_rate": 7.231984882621978e-07, "loss": 0.4077, "step": 24752 }, { "epoch": 0.8818782621087696, "grad_norm": 1.3690612863560612, "learning_rate": 7.227676979139953e-07, "loss": 0.3946, "step": 24753 }, { "epoch": 0.8819138892352637, "grad_norm": 1.3186168233017366, "learning_rate": 7.223370310986277e-07, "loss": 0.4112, "step": 24754 }, { "epoch": 0.8819495163617579, "grad_norm": 1.56613884530674, "learning_rate": 7.219064878218263e-07, "loss": 0.4894, "step": 24755 }, { "epoch": 0.881985143488252, "grad_norm": 1.2803635739470378, "learning_rate": 7.214760680893241e-07, "loss": 0.3786, "step": 24756 }, { "epoch": 0.882020770614746, "grad_norm": 1.4584343054924034, "learning_rate": 7.210457719068542e-07, "loss": 0.4462, "step": 24757 }, { "epoch": 0.8820563977412402, "grad_norm": 1.3629178392782253, "learning_rate": 7.206155992801455e-07, "loss": 0.3904, "step": 24758 }, { "epoch": 0.8820920248677343, "grad_norm": 1.5286253665440626, "learning_rate": 7.201855502149247e-07, "loss": 0.5947, "step": 24759 }, { "epoch": 0.8821276519942284, "grad_norm": 1.2839521642072238, "learning_rate": 7.197556247169202e-07, "loss": 0.3778, "step": 24760 }, { "epoch": 0.8821632791207226, "grad_norm": 1.6578485127646851, "learning_rate": 7.193258227918575e-07, "loss": 0.5094, "step": 24761 }, { "epoch": 0.8821989062472166, "grad_norm": 1.3290986091679293, "learning_rate": 7.188961444454567e-07, "loss": 0.4151, "step": 24762 }, { "epoch": 0.8822345333737107, "grad_norm": 1.2933174367622728, "learning_rate": 7.18466589683442e-07, "loss": 0.3864, "step": 24763 }, { "epoch": 0.8822701605002049, "grad_norm": 1.2725656156927394, "learning_rate": 7.180371585115309e-07, "loss": 0.3519, "step": 24764 }, { "epoch": 0.882305787626699, "grad_norm": 1.2303084400700697, "learning_rate": 7.176078509354422e-07, "loss": 0.3746, "step": 24765 }, { "epoch": 0.882341414753193, "grad_norm": 1.2659230458697042, "learning_rate": 7.171786669608938e-07, "loss": 0.3832, "step": 24766 }, { "epoch": 0.8823770418796872, "grad_norm": 1.4118262079722808, "learning_rate": 7.167496065936008e-07, "loss": 0.425, "step": 24767 }, { "epoch": 0.8824126690061813, "grad_norm": 1.4111741735619054, "learning_rate": 7.163206698392744e-07, "loss": 0.4382, "step": 24768 }, { "epoch": 0.8824482961326754, "grad_norm": 1.5043564628539359, "learning_rate": 7.158918567036266e-07, "loss": 0.3803, "step": 24769 }, { "epoch": 0.8824839232591696, "grad_norm": 1.2592384355901303, "learning_rate": 7.154631671923695e-07, "loss": 0.3953, "step": 24770 }, { "epoch": 0.8825195503856637, "grad_norm": 1.2756819081716293, "learning_rate": 7.150346013112086e-07, "loss": 0.3438, "step": 24771 }, { "epoch": 0.8825551775121577, "grad_norm": 1.265866628418385, "learning_rate": 7.146061590658526e-07, "loss": 0.3945, "step": 24772 }, { "epoch": 0.8825908046386519, "grad_norm": 1.3270933075169238, "learning_rate": 7.141778404620048e-07, "loss": 0.4136, "step": 24773 }, { "epoch": 0.882626431765146, "grad_norm": 1.9346590742860565, "learning_rate": 7.137496455053694e-07, "loss": 0.4467, "step": 24774 }, { "epoch": 0.8826620588916401, "grad_norm": 1.8114467867047228, "learning_rate": 7.133215742016486e-07, "loss": 0.4283, "step": 24775 }, { "epoch": 0.8826976860181343, "grad_norm": 1.8121005987846024, "learning_rate": 7.128936265565434e-07, "loss": 0.4833, "step": 24776 }, { "epoch": 0.8827333131446283, "grad_norm": 1.2025323108938941, "learning_rate": 7.124658025757481e-07, "loss": 0.4195, "step": 24777 }, { "epoch": 0.8827689402711224, "grad_norm": 1.9491395034330392, "learning_rate": 7.120381022649658e-07, "loss": 0.5183, "step": 24778 }, { "epoch": 0.8828045673976166, "grad_norm": 1.690306755297343, "learning_rate": 7.116105256298845e-07, "loss": 0.4824, "step": 24779 }, { "epoch": 0.8828401945241107, "grad_norm": 1.398358116295505, "learning_rate": 7.111830726762026e-07, "loss": 0.4096, "step": 24780 }, { "epoch": 0.8828758216506047, "grad_norm": 1.795891855065064, "learning_rate": 7.107557434096102e-07, "loss": 0.4564, "step": 24781 }, { "epoch": 0.8829114487770989, "grad_norm": 1.3071726681796347, "learning_rate": 7.103285378357983e-07, "loss": 0.3933, "step": 24782 }, { "epoch": 0.882947075903593, "grad_norm": 2.0653095007453453, "learning_rate": 7.099014559604556e-07, "loss": 0.5567, "step": 24783 }, { "epoch": 0.8829827030300871, "grad_norm": 1.3663444838253618, "learning_rate": 7.094744977892676e-07, "loss": 0.3787, "step": 24784 }, { "epoch": 0.8830183301565813, "grad_norm": 1.6092345664032108, "learning_rate": 7.090476633279209e-07, "loss": 0.4992, "step": 24785 }, { "epoch": 0.8830539572830753, "grad_norm": 1.616238008590103, "learning_rate": 7.086209525820986e-07, "loss": 0.5085, "step": 24786 }, { "epoch": 0.8830895844095694, "grad_norm": 1.441449544083088, "learning_rate": 7.08194365557483e-07, "loss": 0.4255, "step": 24787 }, { "epoch": 0.8831252115360636, "grad_norm": 1.390866337520255, "learning_rate": 7.077679022597528e-07, "loss": 0.4133, "step": 24788 }, { "epoch": 0.8831608386625577, "grad_norm": 1.760922766294427, "learning_rate": 7.073415626945879e-07, "loss": 0.4466, "step": 24789 }, { "epoch": 0.8831964657890518, "grad_norm": 1.3410265446964669, "learning_rate": 7.069153468676648e-07, "loss": 0.4143, "step": 24790 }, { "epoch": 0.8832320929155459, "grad_norm": 1.5493655749974795, "learning_rate": 7.064892547846602e-07, "loss": 0.5927, "step": 24791 }, { "epoch": 0.88326772004204, "grad_norm": 1.6539167790691065, "learning_rate": 7.060632864512473e-07, "loss": 0.4489, "step": 24792 }, { "epoch": 0.8833033471685341, "grad_norm": 1.5478511135847064, "learning_rate": 7.056374418730971e-07, "loss": 0.5698, "step": 24793 }, { "epoch": 0.8833389742950283, "grad_norm": 1.2445050408518608, "learning_rate": 7.052117210558829e-07, "loss": 0.4178, "step": 24794 }, { "epoch": 0.8833746014215224, "grad_norm": 1.2884097830427497, "learning_rate": 7.047861240052689e-07, "loss": 0.3907, "step": 24795 }, { "epoch": 0.8834102285480164, "grad_norm": 1.6861176892885188, "learning_rate": 7.043606507269241e-07, "loss": 0.4603, "step": 24796 }, { "epoch": 0.8834458556745106, "grad_norm": 1.4191901193001943, "learning_rate": 7.03935301226516e-07, "loss": 0.435, "step": 24797 }, { "epoch": 0.8834814828010047, "grad_norm": 1.346592117322251, "learning_rate": 7.035100755097069e-07, "loss": 0.5153, "step": 24798 }, { "epoch": 0.8835171099274988, "grad_norm": 1.2554261798682702, "learning_rate": 7.030849735821588e-07, "loss": 0.3919, "step": 24799 }, { "epoch": 0.883552737053993, "grad_norm": 1.797726597035234, "learning_rate": 7.026599954495328e-07, "loss": 0.3933, "step": 24800 }, { "epoch": 0.883588364180487, "grad_norm": 1.2795579310313463, "learning_rate": 7.022351411174866e-07, "loss": 0.4052, "step": 24801 }, { "epoch": 0.8836239913069811, "grad_norm": 1.572986323875498, "learning_rate": 7.018104105916779e-07, "loss": 0.6211, "step": 24802 }, { "epoch": 0.8836596184334753, "grad_norm": 1.403609399683716, "learning_rate": 7.013858038777632e-07, "loss": 0.4352, "step": 24803 }, { "epoch": 0.8836952455599694, "grad_norm": 1.308215928866179, "learning_rate": 7.009613209813948e-07, "loss": 0.3763, "step": 24804 }, { "epoch": 0.8837308726864634, "grad_norm": 1.99478295920783, "learning_rate": 7.005369619082292e-07, "loss": 0.5543, "step": 24805 }, { "epoch": 0.8837664998129576, "grad_norm": 1.6490093267551158, "learning_rate": 7.001127266639108e-07, "loss": 0.3999, "step": 24806 }, { "epoch": 0.8838021269394517, "grad_norm": 1.545045308883624, "learning_rate": 6.996886152540938e-07, "loss": 0.4697, "step": 24807 }, { "epoch": 0.8838377540659458, "grad_norm": 1.6223826336075537, "learning_rate": 6.992646276844217e-07, "loss": 0.5164, "step": 24808 }, { "epoch": 0.88387338119244, "grad_norm": 1.8066516293949797, "learning_rate": 6.988407639605443e-07, "loss": 0.5733, "step": 24809 }, { "epoch": 0.883909008318934, "grad_norm": 1.3422508449759512, "learning_rate": 6.984170240881005e-07, "loss": 0.3912, "step": 24810 }, { "epoch": 0.8839446354454281, "grad_norm": 1.9945558524765763, "learning_rate": 6.979934080727369e-07, "loss": 0.5063, "step": 24811 }, { "epoch": 0.8839802625719223, "grad_norm": 1.4455187389227604, "learning_rate": 6.97569915920091e-07, "loss": 0.4499, "step": 24812 }, { "epoch": 0.8840158896984164, "grad_norm": 2.0448104723038347, "learning_rate": 6.971465476358064e-07, "loss": 0.4876, "step": 24813 }, { "epoch": 0.8840515168249105, "grad_norm": 1.2001601856715827, "learning_rate": 6.96723303225516e-07, "loss": 0.4023, "step": 24814 }, { "epoch": 0.8840871439514046, "grad_norm": 1.2000070439911599, "learning_rate": 6.963001826948579e-07, "loss": 0.365, "step": 24815 }, { "epoch": 0.8841227710778987, "grad_norm": 1.610243595040168, "learning_rate": 6.958771860494662e-07, "loss": 0.4969, "step": 24816 }, { "epoch": 0.8841583982043928, "grad_norm": 1.4306828720647613, "learning_rate": 6.95454313294972e-07, "loss": 0.502, "step": 24817 }, { "epoch": 0.884194025330887, "grad_norm": 2.0195751125648385, "learning_rate": 6.950315644370075e-07, "loss": 0.454, "step": 24818 }, { "epoch": 0.884229652457381, "grad_norm": 1.5213686358659297, "learning_rate": 6.946089394812006e-07, "loss": 0.5388, "step": 24819 }, { "epoch": 0.8842652795838751, "grad_norm": 1.2430866912446508, "learning_rate": 6.94186438433182e-07, "loss": 0.3612, "step": 24820 }, { "epoch": 0.8843009067103693, "grad_norm": 1.2931282304917369, "learning_rate": 6.937640612985741e-07, "loss": 0.3829, "step": 24821 }, { "epoch": 0.8843365338368634, "grad_norm": 1.4291995546399736, "learning_rate": 6.933418080830034e-07, "loss": 0.4621, "step": 24822 }, { "epoch": 0.8843721609633575, "grad_norm": 1.3991911296654558, "learning_rate": 6.9291967879209e-07, "loss": 0.3708, "step": 24823 }, { "epoch": 0.8844077880898517, "grad_norm": 4.318622116021979, "learning_rate": 6.924976734314581e-07, "loss": 0.5361, "step": 24824 }, { "epoch": 0.8844434152163457, "grad_norm": 1.7445297538362414, "learning_rate": 6.920757920067234e-07, "loss": 0.5924, "step": 24825 }, { "epoch": 0.8844790423428398, "grad_norm": 1.7953047757422276, "learning_rate": 6.916540345235056e-07, "loss": 0.4807, "step": 24826 }, { "epoch": 0.884514669469334, "grad_norm": 1.1938466751957244, "learning_rate": 6.912324009874205e-07, "loss": 0.3986, "step": 24827 }, { "epoch": 0.8845502965958281, "grad_norm": 1.2418516047471078, "learning_rate": 6.908108914040823e-07, "loss": 0.3893, "step": 24828 }, { "epoch": 0.8845859237223221, "grad_norm": 1.8159947301663453, "learning_rate": 6.903895057791043e-07, "loss": 0.532, "step": 24829 }, { "epoch": 0.8846215508488163, "grad_norm": 1.314106743228032, "learning_rate": 6.899682441180966e-07, "loss": 0.4603, "step": 24830 }, { "epoch": 0.8846571779753104, "grad_norm": 1.5617167873971394, "learning_rate": 6.895471064266713e-07, "loss": 0.5007, "step": 24831 }, { "epoch": 0.8846928051018045, "grad_norm": 1.64986781641435, "learning_rate": 6.891260927104315e-07, "loss": 0.5143, "step": 24832 }, { "epoch": 0.8847284322282987, "grad_norm": 1.2634355193525288, "learning_rate": 6.887052029749852e-07, "loss": 0.3929, "step": 24833 }, { "epoch": 0.8847640593547927, "grad_norm": 1.3503758555365981, "learning_rate": 6.882844372259378e-07, "loss": 0.4074, "step": 24834 }, { "epoch": 0.8847996864812868, "grad_norm": 1.547022439514652, "learning_rate": 6.878637954688938e-07, "loss": 0.5311, "step": 24835 }, { "epoch": 0.884835313607781, "grad_norm": 1.1377264547404171, "learning_rate": 6.874432777094498e-07, "loss": 0.3843, "step": 24836 }, { "epoch": 0.8848709407342751, "grad_norm": 1.343024079655563, "learning_rate": 6.870228839532078e-07, "loss": 0.4062, "step": 24837 }, { "epoch": 0.8849065678607692, "grad_norm": 1.3049186350232194, "learning_rate": 6.86602614205768e-07, "loss": 0.4052, "step": 24838 }, { "epoch": 0.8849421949872633, "grad_norm": 1.6968074160304834, "learning_rate": 6.861824684727214e-07, "loss": 0.3732, "step": 24839 }, { "epoch": 0.8849778221137574, "grad_norm": 1.584736179179667, "learning_rate": 6.857624467596669e-07, "loss": 0.502, "step": 24840 }, { "epoch": 0.8850134492402515, "grad_norm": 2.1816871316678603, "learning_rate": 6.853425490721921e-07, "loss": 0.5535, "step": 24841 }, { "epoch": 0.8850490763667457, "grad_norm": 1.2773295920159324, "learning_rate": 6.849227754158961e-07, "loss": 0.3878, "step": 24842 }, { "epoch": 0.8850847034932398, "grad_norm": 1.5772253423672147, "learning_rate": 6.845031257963619e-07, "loss": 0.5656, "step": 24843 }, { "epoch": 0.8851203306197338, "grad_norm": 1.379076776895669, "learning_rate": 6.84083600219182e-07, "loss": 0.3865, "step": 24844 }, { "epoch": 0.885155957746228, "grad_norm": 1.32616579917447, "learning_rate": 6.836641986899384e-07, "loss": 0.3987, "step": 24845 }, { "epoch": 0.8851915848727221, "grad_norm": 1.300055092330517, "learning_rate": 6.83244921214219e-07, "loss": 0.4279, "step": 24846 }, { "epoch": 0.8852272119992162, "grad_norm": 1.4720124620660315, "learning_rate": 6.828257677976035e-07, "loss": 0.467, "step": 24847 }, { "epoch": 0.8852628391257104, "grad_norm": 1.7057814594320233, "learning_rate": 6.824067384456756e-07, "loss": 0.5093, "step": 24848 }, { "epoch": 0.8852984662522044, "grad_norm": 1.6427681937453293, "learning_rate": 6.81987833164015e-07, "loss": 0.4662, "step": 24849 }, { "epoch": 0.8853340933786985, "grad_norm": 1.3165958123555013, "learning_rate": 6.815690519581997e-07, "loss": 0.3836, "step": 24850 }, { "epoch": 0.8853697205051927, "grad_norm": 1.4410159571901666, "learning_rate": 6.811503948338061e-07, "loss": 0.3957, "step": 24851 }, { "epoch": 0.8854053476316868, "grad_norm": 1.1892514548187265, "learning_rate": 6.807318617964077e-07, "loss": 0.406, "step": 24852 }, { "epoch": 0.8854409747581808, "grad_norm": 1.4576076801758735, "learning_rate": 6.803134528515798e-07, "loss": 0.472, "step": 24853 }, { "epoch": 0.885476601884675, "grad_norm": 1.2664377723455695, "learning_rate": 6.798951680048904e-07, "loss": 0.4036, "step": 24854 }, { "epoch": 0.8855122290111691, "grad_norm": 1.5325931861254922, "learning_rate": 6.794770072619116e-07, "loss": 0.4714, "step": 24855 }, { "epoch": 0.8855478561376632, "grad_norm": 1.1889910729556101, "learning_rate": 6.790589706282113e-07, "loss": 0.3646, "step": 24856 }, { "epoch": 0.8855834832641574, "grad_norm": 1.4290145043831164, "learning_rate": 6.786410581093572e-07, "loss": 0.3776, "step": 24857 }, { "epoch": 0.8856191103906514, "grad_norm": 1.5336011842177721, "learning_rate": 6.782232697109104e-07, "loss": 0.4132, "step": 24858 }, { "epoch": 0.8856547375171455, "grad_norm": 1.309311435259357, "learning_rate": 6.778056054384385e-07, "loss": 0.4125, "step": 24859 }, { "epoch": 0.8856903646436397, "grad_norm": 1.3056283337592829, "learning_rate": 6.773880652974985e-07, "loss": 0.4184, "step": 24860 }, { "epoch": 0.8857259917701338, "grad_norm": 1.4696556491181658, "learning_rate": 6.769706492936534e-07, "loss": 0.5366, "step": 24861 }, { "epoch": 0.8857616188966279, "grad_norm": 1.618160008535342, "learning_rate": 6.765533574324623e-07, "loss": 0.4135, "step": 24862 }, { "epoch": 0.885797246023122, "grad_norm": 1.3462407979960054, "learning_rate": 6.761361897194774e-07, "loss": 0.4214, "step": 24863 }, { "epoch": 0.8858328731496161, "grad_norm": 1.2669162375290421, "learning_rate": 6.757191461602563e-07, "loss": 0.399, "step": 24864 }, { "epoch": 0.8858685002761102, "grad_norm": 1.677290715197334, "learning_rate": 6.753022267603526e-07, "loss": 0.574, "step": 24865 }, { "epoch": 0.8859041274026044, "grad_norm": 1.3464326528080044, "learning_rate": 6.748854315253184e-07, "loss": 0.393, "step": 24866 }, { "epoch": 0.8859397545290985, "grad_norm": 1.4118288815055469, "learning_rate": 6.744687604607015e-07, "loss": 0.4076, "step": 24867 }, { "epoch": 0.8859753816555925, "grad_norm": 1.2517022598237006, "learning_rate": 6.740522135720518e-07, "loss": 0.4192, "step": 24868 }, { "epoch": 0.8860110087820867, "grad_norm": 1.3040015119106003, "learning_rate": 6.736357908649149e-07, "loss": 0.4103, "step": 24869 }, { "epoch": 0.8860466359085808, "grad_norm": 1.2726403634858325, "learning_rate": 6.732194923448354e-07, "loss": 0.4114, "step": 24870 }, { "epoch": 0.8860822630350749, "grad_norm": 1.4888618497866017, "learning_rate": 6.728033180173576e-07, "loss": 0.4046, "step": 24871 }, { "epoch": 0.886117890161569, "grad_norm": 1.1969793321055795, "learning_rate": 6.723872678880228e-07, "loss": 0.3866, "step": 24872 }, { "epoch": 0.8861535172880631, "grad_norm": 1.246572517344467, "learning_rate": 6.71971341962373e-07, "loss": 0.3562, "step": 24873 }, { "epoch": 0.8861891444145572, "grad_norm": 1.3773807189614582, "learning_rate": 6.715555402459439e-07, "loss": 0.4785, "step": 24874 }, { "epoch": 0.8862247715410514, "grad_norm": 1.7477829163628542, "learning_rate": 6.711398627442733e-07, "loss": 0.5329, "step": 24875 }, { "epoch": 0.8862603986675455, "grad_norm": 1.45545691411362, "learning_rate": 6.707243094628946e-07, "loss": 0.4678, "step": 24876 }, { "epoch": 0.8862960257940395, "grad_norm": 1.4012887498897437, "learning_rate": 6.703088804073443e-07, "loss": 0.3845, "step": 24877 }, { "epoch": 0.8863316529205337, "grad_norm": 1.2976220780614836, "learning_rate": 6.698935755831493e-07, "loss": 0.4077, "step": 24878 }, { "epoch": 0.8863672800470278, "grad_norm": 1.2324492159746399, "learning_rate": 6.69478394995845e-07, "loss": 0.4507, "step": 24879 }, { "epoch": 0.8864029071735219, "grad_norm": 1.309165285140505, "learning_rate": 6.690633386509571e-07, "loss": 0.3559, "step": 24880 }, { "epoch": 0.8864385343000161, "grad_norm": 1.2256403940543958, "learning_rate": 6.686484065540133e-07, "loss": 0.3564, "step": 24881 }, { "epoch": 0.8864741614265101, "grad_norm": 1.2759607210696344, "learning_rate": 6.682335987105371e-07, "loss": 0.3734, "step": 24882 }, { "epoch": 0.8865097885530042, "grad_norm": 1.4673068245877074, "learning_rate": 6.67818915126054e-07, "loss": 0.4958, "step": 24883 }, { "epoch": 0.8865454156794984, "grad_norm": 1.852209409815365, "learning_rate": 6.67404355806085e-07, "loss": 0.4859, "step": 24884 }, { "epoch": 0.8865810428059925, "grad_norm": 1.580850326653761, "learning_rate": 6.669899207561492e-07, "loss": 0.4625, "step": 24885 }, { "epoch": 0.8866166699324866, "grad_norm": 1.402107287870909, "learning_rate": 6.665756099817655e-07, "loss": 0.4168, "step": 24886 }, { "epoch": 0.8866522970589807, "grad_norm": 1.4414854630016294, "learning_rate": 6.661614234884517e-07, "loss": 0.5053, "step": 24887 }, { "epoch": 0.8866879241854748, "grad_norm": 1.341750693356378, "learning_rate": 6.657473612817244e-07, "loss": 0.4025, "step": 24888 }, { "epoch": 0.8867235513119689, "grad_norm": 1.722573912890725, "learning_rate": 6.653334233670928e-07, "loss": 0.4897, "step": 24889 }, { "epoch": 0.8867591784384631, "grad_norm": 1.2793826138884132, "learning_rate": 6.649196097500732e-07, "loss": 0.3857, "step": 24890 }, { "epoch": 0.8867948055649572, "grad_norm": 1.633271988436812, "learning_rate": 6.645059204361726e-07, "loss": 0.4803, "step": 24891 }, { "epoch": 0.8868304326914512, "grad_norm": 1.1665036676226763, "learning_rate": 6.640923554308998e-07, "loss": 0.3799, "step": 24892 }, { "epoch": 0.8868660598179454, "grad_norm": 1.6305423003274577, "learning_rate": 6.636789147397638e-07, "loss": 0.5383, "step": 24893 }, { "epoch": 0.8869016869444395, "grad_norm": 1.4057905001315543, "learning_rate": 6.63265598368269e-07, "loss": 0.5438, "step": 24894 }, { "epoch": 0.8869373140709336, "grad_norm": 1.5360667445575122, "learning_rate": 6.628524063219189e-07, "loss": 0.5292, "step": 24895 }, { "epoch": 0.8869729411974278, "grad_norm": 1.1128848975490464, "learning_rate": 6.624393386062145e-07, "loss": 0.3535, "step": 24896 }, { "epoch": 0.8870085683239218, "grad_norm": 1.375545675409676, "learning_rate": 6.620263952266581e-07, "loss": 0.3862, "step": 24897 }, { "epoch": 0.8870441954504159, "grad_norm": 1.4438456313947363, "learning_rate": 6.616135761887455e-07, "loss": 0.3975, "step": 24898 }, { "epoch": 0.8870798225769101, "grad_norm": 1.5218629901651157, "learning_rate": 6.612008814979776e-07, "loss": 0.3887, "step": 24899 }, { "epoch": 0.8871154497034042, "grad_norm": 1.2909566456258343, "learning_rate": 6.607883111598446e-07, "loss": 0.3877, "step": 24900 }, { "epoch": 0.8871510768298982, "grad_norm": 1.5424574370873074, "learning_rate": 6.603758651798442e-07, "loss": 0.5292, "step": 24901 }, { "epoch": 0.8871867039563924, "grad_norm": 1.3264932977391775, "learning_rate": 6.599635435634666e-07, "loss": 0.4136, "step": 24902 }, { "epoch": 0.8872223310828865, "grad_norm": 1.3360829265652694, "learning_rate": 6.59551346316204e-07, "loss": 0.3829, "step": 24903 }, { "epoch": 0.8872579582093806, "grad_norm": 1.2408287977148529, "learning_rate": 6.591392734435431e-07, "loss": 0.3745, "step": 24904 }, { "epoch": 0.8872935853358748, "grad_norm": 1.610798066468219, "learning_rate": 6.587273249509718e-07, "loss": 0.4473, "step": 24905 }, { "epoch": 0.8873292124623688, "grad_norm": 1.2492003100501028, "learning_rate": 6.583155008439745e-07, "loss": 0.4034, "step": 24906 }, { "epoch": 0.8873648395888629, "grad_norm": 1.8789446844162292, "learning_rate": 6.57903801128037e-07, "loss": 0.4984, "step": 24907 }, { "epoch": 0.8874004667153571, "grad_norm": 1.460385612007093, "learning_rate": 6.57492225808638e-07, "loss": 0.4002, "step": 24908 }, { "epoch": 0.8874360938418512, "grad_norm": 1.2915038741267832, "learning_rate": 6.570807748912611e-07, "loss": 0.4421, "step": 24909 }, { "epoch": 0.8874717209683453, "grad_norm": 1.300390752943099, "learning_rate": 6.566694483813862e-07, "loss": 0.3824, "step": 24910 }, { "epoch": 0.8875073480948394, "grad_norm": 1.443041950018323, "learning_rate": 6.562582462844858e-07, "loss": 0.4634, "step": 24911 }, { "epoch": 0.8875429752213335, "grad_norm": 1.478189403866647, "learning_rate": 6.558471686060397e-07, "loss": 0.4796, "step": 24912 }, { "epoch": 0.8875786023478276, "grad_norm": 1.6651124943811686, "learning_rate": 6.554362153515171e-07, "loss": 0.4695, "step": 24913 }, { "epoch": 0.8876142294743218, "grad_norm": 1.4098423722189863, "learning_rate": 6.550253865263956e-07, "loss": 0.3743, "step": 24914 }, { "epoch": 0.8876498566008159, "grad_norm": 1.2340635944057292, "learning_rate": 6.54614682136141e-07, "loss": 0.3797, "step": 24915 }, { "epoch": 0.8876854837273099, "grad_norm": 1.6042936959434648, "learning_rate": 6.542041021862233e-07, "loss": 0.5448, "step": 24916 }, { "epoch": 0.8877211108538041, "grad_norm": 1.3703251779953454, "learning_rate": 6.537936466821105e-07, "loss": 0.3831, "step": 24917 }, { "epoch": 0.8877567379802982, "grad_norm": 1.2625313108533962, "learning_rate": 6.53383315629268e-07, "loss": 0.3956, "step": 24918 }, { "epoch": 0.8877923651067923, "grad_norm": 1.403343857537963, "learning_rate": 6.529731090331615e-07, "loss": 0.4225, "step": 24919 }, { "epoch": 0.8878279922332865, "grad_norm": 1.6507122163940895, "learning_rate": 6.5256302689925e-07, "loss": 0.5007, "step": 24920 }, { "epoch": 0.8878636193597805, "grad_norm": 1.9044164807349533, "learning_rate": 6.521530692329958e-07, "loss": 0.5546, "step": 24921 }, { "epoch": 0.8878992464862746, "grad_norm": 1.2116436705749458, "learning_rate": 6.517432360398556e-07, "loss": 0.405, "step": 24922 }, { "epoch": 0.8879348736127688, "grad_norm": 1.813601050810753, "learning_rate": 6.513335273252896e-07, "loss": 0.5377, "step": 24923 }, { "epoch": 0.8879705007392629, "grad_norm": 1.6088934625502653, "learning_rate": 6.509239430947511e-07, "loss": 0.4836, "step": 24924 }, { "epoch": 0.888006127865757, "grad_norm": 1.8286258545536973, "learning_rate": 6.505144833536958e-07, "loss": 0.4719, "step": 24925 }, { "epoch": 0.8880417549922511, "grad_norm": 1.2149335050185122, "learning_rate": 6.501051481075748e-07, "loss": 0.3763, "step": 24926 }, { "epoch": 0.8880773821187452, "grad_norm": 1.3449483731791636, "learning_rate": 6.496959373618394e-07, "loss": 0.395, "step": 24927 }, { "epoch": 0.8881130092452393, "grad_norm": 1.0909314147798774, "learning_rate": 6.492868511219374e-07, "loss": 0.3444, "step": 24928 }, { "epoch": 0.8881486363717335, "grad_norm": 1.8410347099260909, "learning_rate": 6.488778893933168e-07, "loss": 0.5091, "step": 24929 }, { "epoch": 0.8881842634982275, "grad_norm": 1.6734085558984508, "learning_rate": 6.484690521814218e-07, "loss": 0.446, "step": 24930 }, { "epoch": 0.8882198906247216, "grad_norm": 1.240877738103617, "learning_rate": 6.480603394916996e-07, "loss": 0.3927, "step": 24931 }, { "epoch": 0.8882555177512158, "grad_norm": 1.3545366798041112, "learning_rate": 6.476517513295911e-07, "loss": 0.3696, "step": 24932 }, { "epoch": 0.8882911448777099, "grad_norm": 1.3950929761864643, "learning_rate": 6.472432877005341e-07, "loss": 0.4337, "step": 24933 }, { "epoch": 0.888326772004204, "grad_norm": 1.2113925298137551, "learning_rate": 6.468349486099723e-07, "loss": 0.3996, "step": 24934 }, { "epoch": 0.8883623991306981, "grad_norm": 1.2010742983400142, "learning_rate": 6.464267340633401e-07, "loss": 0.3925, "step": 24935 }, { "epoch": 0.8883980262571922, "grad_norm": 1.8476286790996064, "learning_rate": 6.460186440660743e-07, "loss": 0.5391, "step": 24936 }, { "epoch": 0.8884336533836863, "grad_norm": 1.8899180366435067, "learning_rate": 6.456106786236061e-07, "loss": 0.5343, "step": 24937 }, { "epoch": 0.8884692805101805, "grad_norm": 1.5896796054741937, "learning_rate": 6.452028377413711e-07, "loss": 0.542, "step": 24938 }, { "epoch": 0.8885049076366746, "grad_norm": 1.6908667224491565, "learning_rate": 6.447951214247983e-07, "loss": 0.4283, "step": 24939 }, { "epoch": 0.8885405347631686, "grad_norm": 1.954408091608515, "learning_rate": 6.443875296793189e-07, "loss": 0.5596, "step": 24940 }, { "epoch": 0.8885761618896628, "grad_norm": 1.2642482244085016, "learning_rate": 6.439800625103576e-07, "loss": 0.4194, "step": 24941 }, { "epoch": 0.8886117890161569, "grad_norm": 1.236942991200686, "learning_rate": 6.43572719923341e-07, "loss": 0.4287, "step": 24942 }, { "epoch": 0.888647416142651, "grad_norm": 1.5896249224448924, "learning_rate": 6.431655019236949e-07, "loss": 0.5623, "step": 24943 }, { "epoch": 0.8886830432691452, "grad_norm": 1.5583554943876048, "learning_rate": 6.427584085168381e-07, "loss": 0.5317, "step": 24944 }, { "epoch": 0.8887186703956392, "grad_norm": 1.204749038295596, "learning_rate": 6.423514397081942e-07, "loss": 0.4323, "step": 24945 }, { "epoch": 0.8887542975221333, "grad_norm": 1.2997968862792113, "learning_rate": 6.419445955031811e-07, "loss": 0.3885, "step": 24946 }, { "epoch": 0.8887899246486275, "grad_norm": 1.537248789493205, "learning_rate": 6.415378759072177e-07, "loss": 0.5434, "step": 24947 }, { "epoch": 0.8888255517751216, "grad_norm": 1.2072766379005018, "learning_rate": 6.411312809257187e-07, "loss": 0.3426, "step": 24948 }, { "epoch": 0.8888611789016156, "grad_norm": 1.2247726884069927, "learning_rate": 6.407248105640984e-07, "loss": 0.4143, "step": 24949 }, { "epoch": 0.8888968060281098, "grad_norm": 1.5308890119257184, "learning_rate": 6.403184648277672e-07, "loss": 0.4355, "step": 24950 }, { "epoch": 0.8889324331546039, "grad_norm": 1.4186369467317552, "learning_rate": 6.399122437221405e-07, "loss": 0.3666, "step": 24951 }, { "epoch": 0.888968060281098, "grad_norm": 1.27571156636367, "learning_rate": 6.39506147252622e-07, "loss": 0.3628, "step": 24952 }, { "epoch": 0.8890036874075922, "grad_norm": 1.6046027083810823, "learning_rate": 6.391001754246206e-07, "loss": 0.4388, "step": 24953 }, { "epoch": 0.8890393145340862, "grad_norm": 1.4784897751708765, "learning_rate": 6.386943282435454e-07, "loss": 0.4221, "step": 24954 }, { "epoch": 0.8890749416605803, "grad_norm": 1.4571093635268435, "learning_rate": 6.382886057147986e-07, "loss": 0.4145, "step": 24955 }, { "epoch": 0.8891105687870745, "grad_norm": 1.1776909464479828, "learning_rate": 6.378830078437825e-07, "loss": 0.4094, "step": 24956 }, { "epoch": 0.8891461959135686, "grad_norm": 1.2541349491257734, "learning_rate": 6.374775346358975e-07, "loss": 0.4267, "step": 24957 }, { "epoch": 0.8891818230400627, "grad_norm": 1.5033036790155223, "learning_rate": 6.370721860965435e-07, "loss": 0.5081, "step": 24958 }, { "epoch": 0.8892174501665568, "grad_norm": 1.2800278601025672, "learning_rate": 6.366669622311161e-07, "loss": 0.4072, "step": 24959 }, { "epoch": 0.8892530772930509, "grad_norm": 1.3604357280120074, "learning_rate": 6.362618630450134e-07, "loss": 0.4402, "step": 24960 }, { "epoch": 0.889288704419545, "grad_norm": 1.370425798622976, "learning_rate": 6.358568885436289e-07, "loss": 0.3952, "step": 24961 }, { "epoch": 0.8893243315460392, "grad_norm": 1.392090302114177, "learning_rate": 6.35452038732357e-07, "loss": 0.3964, "step": 24962 }, { "epoch": 0.8893599586725333, "grad_norm": 1.6842080275311326, "learning_rate": 6.350473136165836e-07, "loss": 0.5425, "step": 24963 }, { "epoch": 0.8893955857990273, "grad_norm": 1.424054199111972, "learning_rate": 6.346427132017019e-07, "loss": 0.4362, "step": 24964 }, { "epoch": 0.8894312129255215, "grad_norm": 1.2971731594923785, "learning_rate": 6.342382374931e-07, "loss": 0.3983, "step": 24965 }, { "epoch": 0.8894668400520156, "grad_norm": 2.2496090364481667, "learning_rate": 6.338338864961613e-07, "loss": 0.5575, "step": 24966 }, { "epoch": 0.8895024671785097, "grad_norm": 1.2082875547842142, "learning_rate": 6.334296602162703e-07, "loss": 0.417, "step": 24967 }, { "epoch": 0.8895380943050039, "grad_norm": 1.3624370350382902, "learning_rate": 6.330255586588096e-07, "loss": 0.4195, "step": 24968 }, { "epoch": 0.8895737214314979, "grad_norm": 1.3051087979699263, "learning_rate": 6.326215818291636e-07, "loss": 0.4064, "step": 24969 }, { "epoch": 0.889609348557992, "grad_norm": 1.6600181827505565, "learning_rate": 6.322177297327059e-07, "loss": 0.503, "step": 24970 }, { "epoch": 0.8896449756844862, "grad_norm": 1.228280002566463, "learning_rate": 6.318140023748188e-07, "loss": 0.4009, "step": 24971 }, { "epoch": 0.8896806028109803, "grad_norm": 1.389622994330293, "learning_rate": 6.314103997608745e-07, "loss": 0.4086, "step": 24972 }, { "epoch": 0.8897162299374743, "grad_norm": 1.4138099008947034, "learning_rate": 6.310069218962512e-07, "loss": 0.38, "step": 24973 }, { "epoch": 0.8897518570639685, "grad_norm": 1.2400112500805391, "learning_rate": 6.306035687863165e-07, "loss": 0.3862, "step": 24974 }, { "epoch": 0.8897874841904626, "grad_norm": 1.2840441312430269, "learning_rate": 6.302003404364454e-07, "loss": 0.4155, "step": 24975 }, { "epoch": 0.8898231113169567, "grad_norm": 1.8701028485654387, "learning_rate": 6.297972368520056e-07, "loss": 0.4271, "step": 24976 }, { "epoch": 0.8898587384434509, "grad_norm": 1.5965173319247314, "learning_rate": 6.29394258038365e-07, "loss": 0.3914, "step": 24977 }, { "epoch": 0.889894365569945, "grad_norm": 1.2592709146503225, "learning_rate": 6.289914040008904e-07, "loss": 0.4119, "step": 24978 }, { "epoch": 0.889929992696439, "grad_norm": 1.6525034147947988, "learning_rate": 6.285886747449443e-07, "loss": 0.4527, "step": 24979 }, { "epoch": 0.8899656198229332, "grad_norm": 1.3708458627235498, "learning_rate": 6.281860702758913e-07, "loss": 0.4144, "step": 24980 }, { "epoch": 0.8900012469494273, "grad_norm": 1.2451822478326342, "learning_rate": 6.277835905990892e-07, "loss": 0.3559, "step": 24981 }, { "epoch": 0.8900368740759214, "grad_norm": 1.353629947487349, "learning_rate": 6.273812357199005e-07, "loss": 0.3744, "step": 24982 }, { "epoch": 0.8900725012024155, "grad_norm": 1.5625231156878692, "learning_rate": 6.269790056436808e-07, "loss": 0.5593, "step": 24983 }, { "epoch": 0.8901081283289096, "grad_norm": 1.1876626207883059, "learning_rate": 6.265769003757894e-07, "loss": 0.393, "step": 24984 }, { "epoch": 0.8901437554554037, "grad_norm": 1.4579989468201877, "learning_rate": 6.261749199215761e-07, "loss": 0.4308, "step": 24985 }, { "epoch": 0.8901793825818979, "grad_norm": 1.706366135673703, "learning_rate": 6.25773064286398e-07, "loss": 0.5919, "step": 24986 }, { "epoch": 0.890215009708392, "grad_norm": 1.5612451155833973, "learning_rate": 6.253713334756018e-07, "loss": 0.5514, "step": 24987 }, { "epoch": 0.890250636834886, "grad_norm": 1.241004513443938, "learning_rate": 6.249697274945377e-07, "loss": 0.4258, "step": 24988 }, { "epoch": 0.8902862639613802, "grad_norm": 1.6947962107418182, "learning_rate": 6.245682463485569e-07, "loss": 0.5356, "step": 24989 }, { "epoch": 0.8903218910878743, "grad_norm": 1.315095939094925, "learning_rate": 6.24166890043002e-07, "loss": 0.4133, "step": 24990 }, { "epoch": 0.8903575182143684, "grad_norm": 1.4623452267995727, "learning_rate": 6.237656585832174e-07, "loss": 0.5225, "step": 24991 }, { "epoch": 0.8903931453408626, "grad_norm": 1.5370592455092953, "learning_rate": 6.233645519745469e-07, "loss": 0.5374, "step": 24992 }, { "epoch": 0.8904287724673566, "grad_norm": 1.244926413088799, "learning_rate": 6.229635702223325e-07, "loss": 0.3924, "step": 24993 }, { "epoch": 0.8904643995938507, "grad_norm": 1.835968162358538, "learning_rate": 6.225627133319112e-07, "loss": 0.5329, "step": 24994 }, { "epoch": 0.8905000267203449, "grad_norm": 1.2029850655215086, "learning_rate": 6.221619813086232e-07, "loss": 0.3786, "step": 24995 }, { "epoch": 0.890535653846839, "grad_norm": 1.5445897692841288, "learning_rate": 6.217613741578011e-07, "loss": 0.5041, "step": 24996 }, { "epoch": 0.890571280973333, "grad_norm": 1.5319761929439375, "learning_rate": 6.213608918847825e-07, "loss": 0.4974, "step": 24997 }, { "epoch": 0.8906069080998272, "grad_norm": 1.6384494015042868, "learning_rate": 6.209605344948977e-07, "loss": 0.4974, "step": 24998 }, { "epoch": 0.8906425352263213, "grad_norm": 1.2626201827647985, "learning_rate": 6.205603019934791e-07, "loss": 0.4219, "step": 24999 }, { "epoch": 0.8906781623528154, "grad_norm": 1.6178656607527417, "learning_rate": 6.201601943858582e-07, "loss": 0.4794, "step": 25000 }, { "epoch": 0.8907137894793096, "grad_norm": 1.486989570054343, "learning_rate": 6.197602116773582e-07, "loss": 0.4992, "step": 25001 }, { "epoch": 0.8907494166058036, "grad_norm": 1.2421457887559684, "learning_rate": 6.193603538733095e-07, "loss": 0.3847, "step": 25002 }, { "epoch": 0.8907850437322977, "grad_norm": 1.8213026772670313, "learning_rate": 6.189606209790322e-07, "loss": 0.4468, "step": 25003 }, { "epoch": 0.8908206708587919, "grad_norm": 1.320816136837772, "learning_rate": 6.185610129998532e-07, "loss": 0.4297, "step": 25004 }, { "epoch": 0.890856297985286, "grad_norm": 1.5468195021453686, "learning_rate": 6.181615299410882e-07, "loss": 0.4876, "step": 25005 }, { "epoch": 0.8908919251117801, "grad_norm": 1.1939226204421218, "learning_rate": 6.177621718080629e-07, "loss": 0.3704, "step": 25006 }, { "epoch": 0.8909275522382742, "grad_norm": 1.2943499098641678, "learning_rate": 6.173629386060909e-07, "loss": 0.3685, "step": 25007 }, { "epoch": 0.8909631793647683, "grad_norm": 1.2381013886245205, "learning_rate": 6.169638303404912e-07, "loss": 0.3839, "step": 25008 }, { "epoch": 0.8909988064912624, "grad_norm": 1.2683380044143713, "learning_rate": 6.165648470165741e-07, "loss": 0.4389, "step": 25009 }, { "epoch": 0.8910344336177566, "grad_norm": 1.4275256988612703, "learning_rate": 6.161659886396553e-07, "loss": 0.361, "step": 25010 }, { "epoch": 0.8910700607442507, "grad_norm": 1.7460014196455675, "learning_rate": 6.157672552150473e-07, "loss": 0.5083, "step": 25011 }, { "epoch": 0.8911056878707447, "grad_norm": 1.4802972087311819, "learning_rate": 6.153686467480547e-07, "loss": 0.4755, "step": 25012 }, { "epoch": 0.8911413149972389, "grad_norm": 1.305031037672973, "learning_rate": 6.149701632439886e-07, "loss": 0.4054, "step": 25013 }, { "epoch": 0.891176942123733, "grad_norm": 1.7138724402965246, "learning_rate": 6.14571804708155e-07, "loss": 0.526, "step": 25014 }, { "epoch": 0.8912125692502271, "grad_norm": 1.1647183047414378, "learning_rate": 6.141735711458585e-07, "loss": 0.3915, "step": 25015 }, { "epoch": 0.8912481963767213, "grad_norm": 1.4672801218790081, "learning_rate": 6.137754625624004e-07, "loss": 0.4345, "step": 25016 }, { "epoch": 0.8912838235032153, "grad_norm": 1.400558521938509, "learning_rate": 6.133774789630842e-07, "loss": 0.3775, "step": 25017 }, { "epoch": 0.8913194506297094, "grad_norm": 1.1991452675055854, "learning_rate": 6.129796203532057e-07, "loss": 0.4009, "step": 25018 }, { "epoch": 0.8913550777562036, "grad_norm": 1.1194297234162152, "learning_rate": 6.12581886738064e-07, "loss": 0.347, "step": 25019 }, { "epoch": 0.8913907048826977, "grad_norm": 1.7021155736162834, "learning_rate": 6.12184278122957e-07, "loss": 0.508, "step": 25020 }, { "epoch": 0.8914263320091917, "grad_norm": 1.1781103394220938, "learning_rate": 6.117867945131794e-07, "loss": 0.4246, "step": 25021 }, { "epoch": 0.8914619591356859, "grad_norm": 1.5614467730639103, "learning_rate": 6.113894359140204e-07, "loss": 0.4497, "step": 25022 }, { "epoch": 0.89149758626218, "grad_norm": 1.8324792788065822, "learning_rate": 6.109922023307746e-07, "loss": 0.5185, "step": 25023 }, { "epoch": 0.8915332133886741, "grad_norm": 1.581677273884064, "learning_rate": 6.10595093768731e-07, "loss": 0.5018, "step": 25024 }, { "epoch": 0.8915688405151683, "grad_norm": 1.6408436446976815, "learning_rate": 6.101981102331744e-07, "loss": 0.5278, "step": 25025 }, { "epoch": 0.8916044676416623, "grad_norm": 1.1273572873218514, "learning_rate": 6.09801251729395e-07, "loss": 0.3474, "step": 25026 }, { "epoch": 0.8916400947681564, "grad_norm": 1.4160316684961372, "learning_rate": 6.09404518262674e-07, "loss": 0.404, "step": 25027 }, { "epoch": 0.8916757218946506, "grad_norm": 2.2328249029483755, "learning_rate": 6.090079098382951e-07, "loss": 0.4688, "step": 25028 }, { "epoch": 0.8917113490211447, "grad_norm": 2.219613921169176, "learning_rate": 6.086114264615406e-07, "loss": 0.5201, "step": 25029 }, { "epoch": 0.8917469761476388, "grad_norm": 1.2951446379047789, "learning_rate": 6.082150681376897e-07, "loss": 0.3912, "step": 25030 }, { "epoch": 0.891782603274133, "grad_norm": 1.2609066390486965, "learning_rate": 6.078188348720183e-07, "loss": 0.407, "step": 25031 }, { "epoch": 0.891818230400627, "grad_norm": 1.7099651088772945, "learning_rate": 6.074227266698063e-07, "loss": 0.5411, "step": 25032 }, { "epoch": 0.8918538575271211, "grad_norm": 1.6614886677515954, "learning_rate": 6.07026743536323e-07, "loss": 0.4737, "step": 25033 }, { "epoch": 0.8918894846536153, "grad_norm": 1.1930957449729924, "learning_rate": 6.066308854768444e-07, "loss": 0.4251, "step": 25034 }, { "epoch": 0.8919251117801094, "grad_norm": 1.835598687560921, "learning_rate": 6.062351524966414e-07, "loss": 0.5514, "step": 25035 }, { "epoch": 0.8919607389066034, "grad_norm": 1.1790857536421284, "learning_rate": 6.058395446009823e-07, "loss": 0.365, "step": 25036 }, { "epoch": 0.8919963660330976, "grad_norm": 1.5908190214859408, "learning_rate": 6.054440617951385e-07, "loss": 0.516, "step": 25037 }, { "epoch": 0.8920319931595917, "grad_norm": 1.8192719467014957, "learning_rate": 6.050487040843711e-07, "loss": 0.389, "step": 25038 }, { "epoch": 0.8920676202860858, "grad_norm": 1.399217262017182, "learning_rate": 6.046534714739483e-07, "loss": 0.4113, "step": 25039 }, { "epoch": 0.89210324741258, "grad_norm": 1.7286375705234482, "learning_rate": 6.042583639691301e-07, "loss": 0.4847, "step": 25040 }, { "epoch": 0.892138874539074, "grad_norm": 1.1534824224986284, "learning_rate": 6.038633815751815e-07, "loss": 0.4214, "step": 25041 }, { "epoch": 0.8921745016655681, "grad_norm": 1.8494359558223348, "learning_rate": 6.034685242973548e-07, "loss": 0.5056, "step": 25042 }, { "epoch": 0.8922101287920623, "grad_norm": 1.2206159305969417, "learning_rate": 6.030737921409169e-07, "loss": 0.4345, "step": 25043 }, { "epoch": 0.8922457559185564, "grad_norm": 1.5262148725068818, "learning_rate": 6.026791851111169e-07, "loss": 0.472, "step": 25044 }, { "epoch": 0.8922813830450504, "grad_norm": 1.1628091970424905, "learning_rate": 6.022847032132128e-07, "loss": 0.3671, "step": 25045 }, { "epoch": 0.8923170101715446, "grad_norm": 2.1047315267150832, "learning_rate": 6.018903464524573e-07, "loss": 0.5288, "step": 25046 }, { "epoch": 0.8923526372980387, "grad_norm": 1.6875724593359618, "learning_rate": 6.014961148341003e-07, "loss": 0.4396, "step": 25047 }, { "epoch": 0.8923882644245328, "grad_norm": 1.190618704897795, "learning_rate": 6.011020083633923e-07, "loss": 0.4474, "step": 25048 }, { "epoch": 0.892423891551027, "grad_norm": 1.5782015353521253, "learning_rate": 6.00708027045579e-07, "loss": 0.4428, "step": 25049 }, { "epoch": 0.892459518677521, "grad_norm": 1.2462769750147185, "learning_rate": 6.003141708859084e-07, "loss": 0.4474, "step": 25050 }, { "epoch": 0.8924951458040151, "grad_norm": 1.3828488467127606, "learning_rate": 5.999204398896241e-07, "loss": 0.4434, "step": 25051 }, { "epoch": 0.8925307729305093, "grad_norm": 1.2650006797824962, "learning_rate": 5.995268340619709e-07, "loss": 0.4383, "step": 25052 }, { "epoch": 0.8925664000570034, "grad_norm": 1.4915854443921843, "learning_rate": 5.991333534081878e-07, "loss": 0.4152, "step": 25053 }, { "epoch": 0.8926020271834975, "grad_norm": 1.6587278782196853, "learning_rate": 5.987399979335151e-07, "loss": 0.5262, "step": 25054 }, { "epoch": 0.8926376543099916, "grad_norm": 1.2423517757996574, "learning_rate": 5.983467676431898e-07, "loss": 0.4331, "step": 25055 }, { "epoch": 0.8926732814364857, "grad_norm": 1.6397410786444724, "learning_rate": 5.979536625424487e-07, "loss": 0.4996, "step": 25056 }, { "epoch": 0.8927089085629799, "grad_norm": 1.3855060241061938, "learning_rate": 5.975606826365265e-07, "loss": 0.5415, "step": 25057 }, { "epoch": 0.892744535689474, "grad_norm": 1.1074752123501366, "learning_rate": 5.971678279306547e-07, "loss": 0.3785, "step": 25058 }, { "epoch": 0.8927801628159681, "grad_norm": 1.1891795854901435, "learning_rate": 5.96775098430068e-07, "loss": 0.3803, "step": 25059 }, { "epoch": 0.8928157899424622, "grad_norm": 1.5114780157961805, "learning_rate": 5.96382494139991e-07, "loss": 0.5323, "step": 25060 }, { "epoch": 0.8928514170689563, "grad_norm": 1.627290193168227, "learning_rate": 5.959900150656562e-07, "loss": 0.5149, "step": 25061 }, { "epoch": 0.8928870441954504, "grad_norm": 1.7963955591493674, "learning_rate": 5.955976612122849e-07, "loss": 0.5673, "step": 25062 }, { "epoch": 0.8929226713219446, "grad_norm": 1.1825227153113484, "learning_rate": 5.952054325851053e-07, "loss": 0.3957, "step": 25063 }, { "epoch": 0.8929582984484387, "grad_norm": 1.7947566160063906, "learning_rate": 5.948133291893387e-07, "loss": 0.5375, "step": 25064 }, { "epoch": 0.8929939255749327, "grad_norm": 1.5877361597872892, "learning_rate": 5.944213510302054e-07, "loss": 0.475, "step": 25065 }, { "epoch": 0.8930295527014269, "grad_norm": 1.532843148615962, "learning_rate": 5.940294981129269e-07, "loss": 0.4255, "step": 25066 }, { "epoch": 0.893065179827921, "grad_norm": 1.2414861531581363, "learning_rate": 5.93637770442721e-07, "loss": 0.4033, "step": 25067 }, { "epoch": 0.8931008069544151, "grad_norm": 1.439699408004174, "learning_rate": 5.932461680248014e-07, "loss": 0.4404, "step": 25068 }, { "epoch": 0.8931364340809093, "grad_norm": 1.2902292826982165, "learning_rate": 5.928546908643839e-07, "loss": 0.4375, "step": 25069 }, { "epoch": 0.8931720612074033, "grad_norm": 1.4573214092713695, "learning_rate": 5.924633389666823e-07, "loss": 0.43, "step": 25070 }, { "epoch": 0.8932076883338974, "grad_norm": 1.4619964439765485, "learning_rate": 5.920721123369067e-07, "loss": 0.3871, "step": 25071 }, { "epoch": 0.8932433154603916, "grad_norm": 1.270045678101702, "learning_rate": 5.916810109802651e-07, "loss": 0.3925, "step": 25072 }, { "epoch": 0.8932789425868857, "grad_norm": 1.8590159736476766, "learning_rate": 5.912900349019679e-07, "loss": 0.5292, "step": 25073 }, { "epoch": 0.8933145697133797, "grad_norm": 1.5622099824123905, "learning_rate": 5.90899184107221e-07, "loss": 0.5095, "step": 25074 }, { "epoch": 0.8933501968398739, "grad_norm": 1.2837275819142673, "learning_rate": 5.905084586012278e-07, "loss": 0.3976, "step": 25075 }, { "epoch": 0.893385823966368, "grad_norm": 2.0842874777296654, "learning_rate": 5.901178583891908e-07, "loss": 0.4979, "step": 25076 }, { "epoch": 0.8934214510928621, "grad_norm": 1.2910309913519817, "learning_rate": 5.897273834763118e-07, "loss": 0.3997, "step": 25077 }, { "epoch": 0.8934570782193563, "grad_norm": 1.2930634204082863, "learning_rate": 5.893370338677917e-07, "loss": 0.4449, "step": 25078 }, { "epoch": 0.8934927053458503, "grad_norm": 1.210450871691782, "learning_rate": 5.889468095688244e-07, "loss": 0.3897, "step": 25079 }, { "epoch": 0.8935283324723444, "grad_norm": 1.8126728171527613, "learning_rate": 5.885567105846068e-07, "loss": 0.5178, "step": 25080 }, { "epoch": 0.8935639595988386, "grad_norm": 4.93921170458246, "learning_rate": 5.881667369203381e-07, "loss": 0.5772, "step": 25081 }, { "epoch": 0.8935995867253327, "grad_norm": 1.498035601208525, "learning_rate": 5.877768885812052e-07, "loss": 0.448, "step": 25082 }, { "epoch": 0.8936352138518268, "grad_norm": 1.7697857153675491, "learning_rate": 5.873871655724051e-07, "loss": 0.6222, "step": 25083 }, { "epoch": 0.893670840978321, "grad_norm": 1.1448694574410656, "learning_rate": 5.869975678991214e-07, "loss": 0.3794, "step": 25084 }, { "epoch": 0.893706468104815, "grad_norm": 1.7735660011746128, "learning_rate": 5.866080955665455e-07, "loss": 0.4364, "step": 25085 }, { "epoch": 0.8937420952313091, "grad_norm": 1.221966812867791, "learning_rate": 5.862187485798609e-07, "loss": 0.3694, "step": 25086 }, { "epoch": 0.8937777223578033, "grad_norm": 1.9693152568649261, "learning_rate": 5.858295269442537e-07, "loss": 0.6295, "step": 25087 }, { "epoch": 0.8938133494842974, "grad_norm": 1.125369852075395, "learning_rate": 5.854404306649075e-07, "loss": 0.4029, "step": 25088 }, { "epoch": 0.8938489766107914, "grad_norm": 1.2464796454657354, "learning_rate": 5.850514597470025e-07, "loss": 0.4236, "step": 25089 }, { "epoch": 0.8938846037372856, "grad_norm": 1.953564472568551, "learning_rate": 5.846626141957168e-07, "loss": 0.5375, "step": 25090 }, { "epoch": 0.8939202308637797, "grad_norm": 1.6707516224785501, "learning_rate": 5.842738940162296e-07, "loss": 0.4556, "step": 25091 }, { "epoch": 0.8939558579902738, "grad_norm": 1.5526897601884009, "learning_rate": 5.83885299213719e-07, "loss": 0.3869, "step": 25092 }, { "epoch": 0.893991485116768, "grad_norm": 1.3412299962036696, "learning_rate": 5.834968297933541e-07, "loss": 0.459, "step": 25093 }, { "epoch": 0.894027112243262, "grad_norm": 1.4967093686755253, "learning_rate": 5.831084857603109e-07, "loss": 0.5151, "step": 25094 }, { "epoch": 0.8940627393697561, "grad_norm": 1.272632941303667, "learning_rate": 5.827202671197607e-07, "loss": 0.3736, "step": 25095 }, { "epoch": 0.8940983664962503, "grad_norm": 1.1928500250554632, "learning_rate": 5.823321738768739e-07, "loss": 0.417, "step": 25096 }, { "epoch": 0.8941339936227444, "grad_norm": 1.239570459652354, "learning_rate": 5.819442060368153e-07, "loss": 0.3488, "step": 25097 }, { "epoch": 0.8941696207492384, "grad_norm": 1.8586580440797467, "learning_rate": 5.815563636047539e-07, "loss": 0.5571, "step": 25098 }, { "epoch": 0.8942052478757326, "grad_norm": 1.7891777397450601, "learning_rate": 5.811686465858513e-07, "loss": 0.4881, "step": 25099 }, { "epoch": 0.8942408750022267, "grad_norm": 1.1921007582465863, "learning_rate": 5.807810549852721e-07, "loss": 0.3805, "step": 25100 }, { "epoch": 0.8942765021287208, "grad_norm": 1.3560319166178723, "learning_rate": 5.803935888081769e-07, "loss": 0.4478, "step": 25101 }, { "epoch": 0.894312129255215, "grad_norm": 1.3269663367570956, "learning_rate": 5.800062480597235e-07, "loss": 0.4176, "step": 25102 }, { "epoch": 0.894347756381709, "grad_norm": 1.4499029852439256, "learning_rate": 5.796190327450713e-07, "loss": 0.4179, "step": 25103 }, { "epoch": 0.8943833835082031, "grad_norm": 1.088928118934563, "learning_rate": 5.792319428693771e-07, "loss": 0.4057, "step": 25104 }, { "epoch": 0.8944190106346973, "grad_norm": 1.4572057683831896, "learning_rate": 5.788449784377948e-07, "loss": 0.3877, "step": 25105 }, { "epoch": 0.8944546377611914, "grad_norm": 1.6663907283724162, "learning_rate": 5.784581394554756e-07, "loss": 0.5018, "step": 25106 }, { "epoch": 0.8944902648876855, "grad_norm": 1.3783918099009802, "learning_rate": 5.780714259275733e-07, "loss": 0.4252, "step": 25107 }, { "epoch": 0.8945258920141796, "grad_norm": 1.7367653924653434, "learning_rate": 5.776848378592337e-07, "loss": 0.5636, "step": 25108 }, { "epoch": 0.8945615191406737, "grad_norm": 1.7987238617308943, "learning_rate": 5.772983752556061e-07, "loss": 0.549, "step": 25109 }, { "epoch": 0.8945971462671678, "grad_norm": 1.2190529178624736, "learning_rate": 5.769120381218362e-07, "loss": 0.3684, "step": 25110 }, { "epoch": 0.894632773393662, "grad_norm": 1.5981024225387792, "learning_rate": 5.765258264630713e-07, "loss": 0.4977, "step": 25111 }, { "epoch": 0.8946684005201561, "grad_norm": 2.420443341322408, "learning_rate": 5.761397402844493e-07, "loss": 0.5803, "step": 25112 }, { "epoch": 0.8947040276466501, "grad_norm": 1.2340129072870154, "learning_rate": 5.757537795911161e-07, "loss": 0.3995, "step": 25113 }, { "epoch": 0.8947396547731443, "grad_norm": 1.2560626250812081, "learning_rate": 5.753679443882065e-07, "loss": 0.3463, "step": 25114 }, { "epoch": 0.8947752818996384, "grad_norm": 1.1800937609314133, "learning_rate": 5.749822346808598e-07, "loss": 0.3676, "step": 25115 }, { "epoch": 0.8948109090261325, "grad_norm": 1.5774513330601396, "learning_rate": 5.745966504742139e-07, "loss": 0.4494, "step": 25116 }, { "epoch": 0.8948465361526267, "grad_norm": 1.1045710594365434, "learning_rate": 5.742111917733994e-07, "loss": 0.3876, "step": 25117 }, { "epoch": 0.8948821632791207, "grad_norm": 2.02389640559612, "learning_rate": 5.738258585835532e-07, "loss": 0.4449, "step": 25118 }, { "epoch": 0.8949177904056148, "grad_norm": 1.260202697690292, "learning_rate": 5.734406509098033e-07, "loss": 0.3953, "step": 25119 }, { "epoch": 0.894953417532109, "grad_norm": 1.341752097258712, "learning_rate": 5.730555687572803e-07, "loss": 0.4091, "step": 25120 }, { "epoch": 0.8949890446586031, "grad_norm": 1.4913808261001253, "learning_rate": 5.72670612131111e-07, "loss": 0.5008, "step": 25121 }, { "epoch": 0.8950246717850971, "grad_norm": 1.2985500753405481, "learning_rate": 5.722857810364235e-07, "loss": 0.4022, "step": 25122 }, { "epoch": 0.8950602989115913, "grad_norm": 1.268302208400486, "learning_rate": 5.719010754783383e-07, "loss": 0.3869, "step": 25123 }, { "epoch": 0.8950959260380854, "grad_norm": 1.4980777648011239, "learning_rate": 5.715164954619801e-07, "loss": 0.543, "step": 25124 }, { "epoch": 0.8951315531645795, "grad_norm": 1.536812704212596, "learning_rate": 5.711320409924703e-07, "loss": 0.5347, "step": 25125 }, { "epoch": 0.8951671802910737, "grad_norm": 1.3009238307154283, "learning_rate": 5.707477120749272e-07, "loss": 0.4027, "step": 25126 }, { "epoch": 0.8952028074175677, "grad_norm": 2.0939359559841635, "learning_rate": 5.70363508714471e-07, "loss": 0.5765, "step": 25127 }, { "epoch": 0.8952384345440618, "grad_norm": 1.2835692544183566, "learning_rate": 5.699794309162132e-07, "loss": 0.3726, "step": 25128 }, { "epoch": 0.895274061670556, "grad_norm": 1.2058428819642881, "learning_rate": 5.695954786852731e-07, "loss": 0.3646, "step": 25129 }, { "epoch": 0.8953096887970501, "grad_norm": 1.527214770392567, "learning_rate": 5.692116520267587e-07, "loss": 0.4964, "step": 25130 }, { "epoch": 0.8953453159235442, "grad_norm": 1.2611792200875307, "learning_rate": 5.688279509457828e-07, "loss": 0.4358, "step": 25131 }, { "epoch": 0.8953809430500383, "grad_norm": 1.7192602274708801, "learning_rate": 5.684443754474544e-07, "loss": 0.4526, "step": 25132 }, { "epoch": 0.8954165701765324, "grad_norm": 1.3193202412442995, "learning_rate": 5.68060925536884e-07, "loss": 0.3475, "step": 25133 }, { "epoch": 0.8954521973030265, "grad_norm": 1.631232216045992, "learning_rate": 5.67677601219172e-07, "loss": 0.5007, "step": 25134 }, { "epoch": 0.8954878244295207, "grad_norm": 1.6004754837363395, "learning_rate": 5.672944024994287e-07, "loss": 0.5374, "step": 25135 }, { "epoch": 0.8955234515560148, "grad_norm": 2.251813542788475, "learning_rate": 5.66911329382751e-07, "loss": 0.4771, "step": 25136 }, { "epoch": 0.8955590786825088, "grad_norm": 1.9997733538089453, "learning_rate": 5.665283818742418e-07, "loss": 0.4892, "step": 25137 }, { "epoch": 0.895594705809003, "grad_norm": 1.557941646345664, "learning_rate": 5.661455599790033e-07, "loss": 0.4932, "step": 25138 }, { "epoch": 0.8956303329354971, "grad_norm": 1.3867484690669107, "learning_rate": 5.657628637021284e-07, "loss": 0.4177, "step": 25139 }, { "epoch": 0.8956659600619912, "grad_norm": 1.3166769414414747, "learning_rate": 5.653802930487151e-07, "loss": 0.4595, "step": 25140 }, { "epoch": 0.8957015871884854, "grad_norm": 1.2438071061027778, "learning_rate": 5.649978480238583e-07, "loss": 0.4032, "step": 25141 }, { "epoch": 0.8957372143149794, "grad_norm": 1.211551193353474, "learning_rate": 5.646155286326505e-07, "loss": 0.4103, "step": 25142 }, { "epoch": 0.8957728414414735, "grad_norm": 2.7531094436629266, "learning_rate": 5.64233334880181e-07, "loss": 0.5102, "step": 25143 }, { "epoch": 0.8958084685679677, "grad_norm": 1.8670181084399613, "learning_rate": 5.638512667715401e-07, "loss": 0.4374, "step": 25144 }, { "epoch": 0.8958440956944618, "grad_norm": 3.123068771689417, "learning_rate": 5.63469324311815e-07, "loss": 0.5383, "step": 25145 }, { "epoch": 0.8958797228209558, "grad_norm": 1.510634299271531, "learning_rate": 5.630875075060915e-07, "loss": 0.4652, "step": 25146 }, { "epoch": 0.89591534994745, "grad_norm": 1.4498002049707295, "learning_rate": 5.627058163594534e-07, "loss": 0.4703, "step": 25147 }, { "epoch": 0.8959509770739441, "grad_norm": 1.3350260844854984, "learning_rate": 5.623242508769844e-07, "loss": 0.372, "step": 25148 }, { "epoch": 0.8959866042004382, "grad_norm": 1.5705844910010145, "learning_rate": 5.619428110637637e-07, "loss": 0.5341, "step": 25149 }, { "epoch": 0.8960222313269324, "grad_norm": 1.613318946991629, "learning_rate": 5.615614969248717e-07, "loss": 0.4728, "step": 25150 }, { "epoch": 0.8960578584534264, "grad_norm": 1.782802199427053, "learning_rate": 5.611803084653866e-07, "loss": 0.6191, "step": 25151 }, { "epoch": 0.8960934855799205, "grad_norm": 1.2623294137808345, "learning_rate": 5.607992456903822e-07, "loss": 0.4498, "step": 25152 }, { "epoch": 0.8961291127064147, "grad_norm": 1.2938688877227003, "learning_rate": 5.604183086049342e-07, "loss": 0.4305, "step": 25153 }, { "epoch": 0.8961647398329088, "grad_norm": 1.6662184613395616, "learning_rate": 5.60037497214112e-07, "loss": 0.6115, "step": 25154 }, { "epoch": 0.8962003669594029, "grad_norm": 1.513753381346748, "learning_rate": 5.596568115229906e-07, "loss": 0.3903, "step": 25155 }, { "epoch": 0.896235994085897, "grad_norm": 1.1078552347123103, "learning_rate": 5.592762515366356e-07, "loss": 0.3558, "step": 25156 }, { "epoch": 0.8962716212123911, "grad_norm": 1.323814203712401, "learning_rate": 5.588958172601178e-07, "loss": 0.3928, "step": 25157 }, { "epoch": 0.8963072483388852, "grad_norm": 1.676278567158088, "learning_rate": 5.585155086985006e-07, "loss": 0.4855, "step": 25158 }, { "epoch": 0.8963428754653794, "grad_norm": 1.2980615230214012, "learning_rate": 5.581353258568489e-07, "loss": 0.4056, "step": 25159 }, { "epoch": 0.8963785025918735, "grad_norm": 1.265105739472973, "learning_rate": 5.577552687402233e-07, "loss": 0.4264, "step": 25160 }, { "epoch": 0.8964141297183675, "grad_norm": 1.6653863467665864, "learning_rate": 5.573753373536872e-07, "loss": 0.4807, "step": 25161 }, { "epoch": 0.8964497568448617, "grad_norm": 1.5705223315508599, "learning_rate": 5.569955317022968e-07, "loss": 0.4854, "step": 25162 }, { "epoch": 0.8964853839713558, "grad_norm": 1.3203949799614056, "learning_rate": 5.566158517911125e-07, "loss": 0.4926, "step": 25163 }, { "epoch": 0.8965210110978499, "grad_norm": 1.4148849733290532, "learning_rate": 5.562362976251901e-07, "loss": 0.4532, "step": 25164 }, { "epoch": 0.8965566382243441, "grad_norm": 1.313767710211903, "learning_rate": 5.558568692095801e-07, "loss": 0.4225, "step": 25165 }, { "epoch": 0.8965922653508381, "grad_norm": 1.4130871975905106, "learning_rate": 5.554775665493384e-07, "loss": 0.3995, "step": 25166 }, { "epoch": 0.8966278924773322, "grad_norm": 1.8444100019678717, "learning_rate": 5.550983896495132e-07, "loss": 0.497, "step": 25167 }, { "epoch": 0.8966635196038264, "grad_norm": 1.595028543111628, "learning_rate": 5.547193385151561e-07, "loss": 0.5111, "step": 25168 }, { "epoch": 0.8966991467303205, "grad_norm": 1.5503459931685128, "learning_rate": 5.543404131513097e-07, "loss": 0.4908, "step": 25169 }, { "epoch": 0.8967347738568145, "grad_norm": 1.2910500605352997, "learning_rate": 5.539616135630254e-07, "loss": 0.3685, "step": 25170 }, { "epoch": 0.8967704009833087, "grad_norm": 1.3906077090063038, "learning_rate": 5.535829397553438e-07, "loss": 0.4483, "step": 25171 }, { "epoch": 0.8968060281098028, "grad_norm": 1.275388233321342, "learning_rate": 5.532043917333074e-07, "loss": 0.4397, "step": 25172 }, { "epoch": 0.8968416552362969, "grad_norm": 1.6582447129118032, "learning_rate": 5.5282596950196e-07, "loss": 0.5349, "step": 25173 }, { "epoch": 0.8968772823627911, "grad_norm": 1.308063702446276, "learning_rate": 5.524476730663353e-07, "loss": 0.3487, "step": 25174 }, { "epoch": 0.8969129094892851, "grad_norm": 1.816958492658774, "learning_rate": 5.520695024314749e-07, "loss": 0.4369, "step": 25175 }, { "epoch": 0.8969485366157792, "grad_norm": 1.2433375131324649, "learning_rate": 5.516914576024124e-07, "loss": 0.3968, "step": 25176 }, { "epoch": 0.8969841637422734, "grad_norm": 1.7162284939091645, "learning_rate": 5.513135385841806e-07, "loss": 0.5335, "step": 25177 }, { "epoch": 0.8970197908687675, "grad_norm": 1.3050307533772005, "learning_rate": 5.509357453818142e-07, "loss": 0.4255, "step": 25178 }, { "epoch": 0.8970554179952616, "grad_norm": 1.7873578758716258, "learning_rate": 5.505580780003439e-07, "loss": 0.4944, "step": 25179 }, { "epoch": 0.8970910451217557, "grad_norm": 1.4669617500457135, "learning_rate": 5.501805364447965e-07, "loss": 0.469, "step": 25180 }, { "epoch": 0.8971266722482498, "grad_norm": 1.3606961929368642, "learning_rate": 5.498031207202015e-07, "loss": 0.4403, "step": 25181 }, { "epoch": 0.8971622993747439, "grad_norm": 1.647819293469498, "learning_rate": 5.494258308315815e-07, "loss": 0.5839, "step": 25182 }, { "epoch": 0.8971979265012381, "grad_norm": 1.8226745471143933, "learning_rate": 5.490486667839623e-07, "loss": 0.463, "step": 25183 }, { "epoch": 0.8972335536277322, "grad_norm": 1.1954116338377705, "learning_rate": 5.486716285823667e-07, "loss": 0.4117, "step": 25184 }, { "epoch": 0.8972691807542262, "grad_norm": 1.1918976315810337, "learning_rate": 5.482947162318131e-07, "loss": 0.376, "step": 25185 }, { "epoch": 0.8973048078807204, "grad_norm": 1.799938228466763, "learning_rate": 5.479179297373239e-07, "loss": 0.4528, "step": 25186 }, { "epoch": 0.8973404350072145, "grad_norm": 1.533768558508369, "learning_rate": 5.475412691039128e-07, "loss": 0.4889, "step": 25187 }, { "epoch": 0.8973760621337086, "grad_norm": 1.4639457791283395, "learning_rate": 5.471647343365982e-07, "loss": 0.4911, "step": 25188 }, { "epoch": 0.8974116892602028, "grad_norm": 1.3210997483747549, "learning_rate": 5.467883254403905e-07, "loss": 0.3747, "step": 25189 }, { "epoch": 0.8974473163866968, "grad_norm": 1.2988069704760299, "learning_rate": 5.464120424203046e-07, "loss": 0.3934, "step": 25190 }, { "epoch": 0.8974829435131909, "grad_norm": 1.448385104594138, "learning_rate": 5.460358852813496e-07, "loss": 0.4613, "step": 25191 }, { "epoch": 0.8975185706396851, "grad_norm": 1.4545336712024433, "learning_rate": 5.456598540285341e-07, "loss": 0.3985, "step": 25192 }, { "epoch": 0.8975541977661792, "grad_norm": 1.5940842081687465, "learning_rate": 5.45283948666866e-07, "loss": 0.4191, "step": 25193 }, { "epoch": 0.8975898248926732, "grad_norm": 1.834345614881076, "learning_rate": 5.449081692013513e-07, "loss": 0.5131, "step": 25194 }, { "epoch": 0.8976254520191674, "grad_norm": 1.3064840913596902, "learning_rate": 5.445325156369918e-07, "loss": 0.4337, "step": 25195 }, { "epoch": 0.8976610791456615, "grad_norm": 1.2924276556995573, "learning_rate": 5.441569879787911e-07, "loss": 0.3904, "step": 25196 }, { "epoch": 0.8976967062721556, "grad_norm": 1.4138824543270292, "learning_rate": 5.437815862317519e-07, "loss": 0.3723, "step": 25197 }, { "epoch": 0.8977323333986498, "grad_norm": 1.3243065115875992, "learning_rate": 5.43406310400868e-07, "loss": 0.3478, "step": 25198 }, { "epoch": 0.8977679605251438, "grad_norm": 1.6265779709097965, "learning_rate": 5.430311604911387e-07, "loss": 0.4944, "step": 25199 }, { "epoch": 0.8978035876516379, "grad_norm": 1.4832937412018012, "learning_rate": 5.426561365075589e-07, "loss": 0.5102, "step": 25200 }, { "epoch": 0.8978392147781321, "grad_norm": 1.8469123679972925, "learning_rate": 5.422812384551257e-07, "loss": 0.524, "step": 25201 }, { "epoch": 0.8978748419046262, "grad_norm": 1.5059932719463065, "learning_rate": 5.419064663388263e-07, "loss": 0.4734, "step": 25202 }, { "epoch": 0.8979104690311203, "grad_norm": 1.2026093719547881, "learning_rate": 5.415318201636543e-07, "loss": 0.4377, "step": 25203 }, { "epoch": 0.8979460961576144, "grad_norm": 1.4733947601027504, "learning_rate": 5.411572999345959e-07, "loss": 0.5481, "step": 25204 }, { "epoch": 0.8979817232841085, "grad_norm": 1.2648442416603887, "learning_rate": 5.407829056566405e-07, "loss": 0.4158, "step": 25205 }, { "epoch": 0.8980173504106026, "grad_norm": 1.2331287463020755, "learning_rate": 5.404086373347705e-07, "loss": 0.3991, "step": 25206 }, { "epoch": 0.8980529775370968, "grad_norm": 1.4777375401606427, "learning_rate": 5.400344949739722e-07, "loss": 0.4756, "step": 25207 }, { "epoch": 0.8980886046635909, "grad_norm": 1.3670357160808269, "learning_rate": 5.396604785792281e-07, "loss": 0.3861, "step": 25208 }, { "epoch": 0.8981242317900849, "grad_norm": 1.3054037859582963, "learning_rate": 5.392865881555154e-07, "loss": 0.6126, "step": 25209 }, { "epoch": 0.8981598589165791, "grad_norm": 1.4982867112478497, "learning_rate": 5.389128237078156e-07, "loss": 0.4746, "step": 25210 }, { "epoch": 0.8981954860430732, "grad_norm": 1.584005737668445, "learning_rate": 5.385391852411038e-07, "loss": 0.4747, "step": 25211 }, { "epoch": 0.8982311131695673, "grad_norm": 1.4465167431944523, "learning_rate": 5.38165672760358e-07, "loss": 0.366, "step": 25212 }, { "epoch": 0.8982667402960615, "grad_norm": 1.5548670713243185, "learning_rate": 5.377922862705465e-07, "loss": 0.4471, "step": 25213 }, { "epoch": 0.8983023674225555, "grad_norm": 1.460049672085683, "learning_rate": 5.374190257766454e-07, "loss": 0.4171, "step": 25214 }, { "epoch": 0.8983379945490496, "grad_norm": 1.35260029208553, "learning_rate": 5.37045891283624e-07, "loss": 0.4423, "step": 25215 }, { "epoch": 0.8983736216755438, "grad_norm": 1.374929862586487, "learning_rate": 5.366728827964518e-07, "loss": 0.4269, "step": 25216 }, { "epoch": 0.8984092488020379, "grad_norm": 1.525434026987093, "learning_rate": 5.363000003200936e-07, "loss": 0.5043, "step": 25217 }, { "epoch": 0.898444875928532, "grad_norm": 1.6687900856142646, "learning_rate": 5.359272438595153e-07, "loss": 0.5881, "step": 25218 }, { "epoch": 0.8984805030550261, "grad_norm": 5.4216373331044645, "learning_rate": 5.355546134196821e-07, "loss": 0.5058, "step": 25219 }, { "epoch": 0.8985161301815202, "grad_norm": 1.6663880192145457, "learning_rate": 5.351821090055531e-07, "loss": 0.6093, "step": 25220 }, { "epoch": 0.8985517573080143, "grad_norm": 1.2147958576786597, "learning_rate": 5.348097306220901e-07, "loss": 0.4239, "step": 25221 }, { "epoch": 0.8985873844345085, "grad_norm": 1.9621718803137336, "learning_rate": 5.344374782742512e-07, "loss": 0.5078, "step": 25222 }, { "epoch": 0.8986230115610025, "grad_norm": 1.5458169916719189, "learning_rate": 5.340653519669947e-07, "loss": 0.5033, "step": 25223 }, { "epoch": 0.8986586386874966, "grad_norm": 1.405696145002926, "learning_rate": 5.336933517052734e-07, "loss": 0.4589, "step": 25224 }, { "epoch": 0.8986942658139908, "grad_norm": 1.4131418275755134, "learning_rate": 5.33321477494042e-07, "loss": 0.5605, "step": 25225 }, { "epoch": 0.8987298929404849, "grad_norm": 1.241331890278598, "learning_rate": 5.329497293382513e-07, "loss": 0.4198, "step": 25226 }, { "epoch": 0.898765520066979, "grad_norm": 1.559478002041158, "learning_rate": 5.325781072428538e-07, "loss": 0.413, "step": 25227 }, { "epoch": 0.8988011471934731, "grad_norm": 1.3621726061031703, "learning_rate": 5.322066112127932e-07, "loss": 0.4208, "step": 25228 }, { "epoch": 0.8988367743199672, "grad_norm": 1.3681375361932298, "learning_rate": 5.318352412530204e-07, "loss": 0.4491, "step": 25229 }, { "epoch": 0.8988724014464613, "grad_norm": 1.2726179431943954, "learning_rate": 5.314639973684788e-07, "loss": 0.3882, "step": 25230 }, { "epoch": 0.8989080285729555, "grad_norm": 1.219251215518051, "learning_rate": 5.310928795641124e-07, "loss": 0.3848, "step": 25231 }, { "epoch": 0.8989436556994496, "grad_norm": 1.9043520822003512, "learning_rate": 5.307218878448638e-07, "loss": 0.5633, "step": 25232 }, { "epoch": 0.8989792828259436, "grad_norm": 1.5330994016719284, "learning_rate": 5.303510222156716e-07, "loss": 0.3745, "step": 25233 }, { "epoch": 0.8990149099524378, "grad_norm": 1.6320639578707585, "learning_rate": 5.299802826814748e-07, "loss": 0.5269, "step": 25234 }, { "epoch": 0.8990505370789319, "grad_norm": 1.1956704528998525, "learning_rate": 5.296096692472086e-07, "loss": 0.3881, "step": 25235 }, { "epoch": 0.899086164205426, "grad_norm": 1.1383486981939617, "learning_rate": 5.292391819178089e-07, "loss": 0.3509, "step": 25236 }, { "epoch": 0.8991217913319202, "grad_norm": 1.900376355164566, "learning_rate": 5.288688206982084e-07, "loss": 0.5848, "step": 25237 }, { "epoch": 0.8991574184584142, "grad_norm": 1.7195243352448129, "learning_rate": 5.284985855933422e-07, "loss": 0.4782, "step": 25238 }, { "epoch": 0.8991930455849083, "grad_norm": 1.4310831418765488, "learning_rate": 5.281284766081352e-07, "loss": 0.4331, "step": 25239 }, { "epoch": 0.8992286727114025, "grad_norm": 1.453917244434436, "learning_rate": 5.2775849374752e-07, "loss": 0.3955, "step": 25240 }, { "epoch": 0.8992642998378966, "grad_norm": 1.6408573277675584, "learning_rate": 5.273886370164194e-07, "loss": 0.5373, "step": 25241 }, { "epoch": 0.8992999269643907, "grad_norm": 1.6705018101871278, "learning_rate": 5.270189064197606e-07, "loss": 0.5196, "step": 25242 }, { "epoch": 0.8993355540908848, "grad_norm": 1.286162490515479, "learning_rate": 5.266493019624664e-07, "loss": 0.4023, "step": 25243 }, { "epoch": 0.8993711812173789, "grad_norm": 2.4975994481533843, "learning_rate": 5.26279823649457e-07, "loss": 0.4763, "step": 25244 }, { "epoch": 0.899406808343873, "grad_norm": 1.3286806263038862, "learning_rate": 5.259104714856555e-07, "loss": 0.4609, "step": 25245 }, { "epoch": 0.8994424354703672, "grad_norm": 1.618095561648502, "learning_rate": 5.255412454759767e-07, "loss": 0.5684, "step": 25246 }, { "epoch": 0.8994780625968612, "grad_norm": 1.6458447447203908, "learning_rate": 5.2517214562534e-07, "loss": 0.5268, "step": 25247 }, { "epoch": 0.8995136897233553, "grad_norm": 1.2344868887590508, "learning_rate": 5.24803171938657e-07, "loss": 0.3311, "step": 25248 }, { "epoch": 0.8995493168498495, "grad_norm": 1.234759271969437, "learning_rate": 5.244343244208439e-07, "loss": 0.416, "step": 25249 }, { "epoch": 0.8995849439763436, "grad_norm": 1.5138366718213494, "learning_rate": 5.240656030768098e-07, "loss": 0.4888, "step": 25250 }, { "epoch": 0.8996205711028377, "grad_norm": 1.452789058471421, "learning_rate": 5.236970079114645e-07, "loss": 0.4127, "step": 25251 }, { "epoch": 0.8996561982293318, "grad_norm": 1.4694395524222252, "learning_rate": 5.233285389297182e-07, "loss": 0.4221, "step": 25252 }, { "epoch": 0.8996918253558259, "grad_norm": 1.5987052368755905, "learning_rate": 5.229601961364761e-07, "loss": 0.5264, "step": 25253 }, { "epoch": 0.89972745248232, "grad_norm": 2.089462959755359, "learning_rate": 5.225919795366441e-07, "loss": 0.5128, "step": 25254 }, { "epoch": 0.8997630796088142, "grad_norm": 1.9057123382346217, "learning_rate": 5.222238891351228e-07, "loss": 0.4975, "step": 25255 }, { "epoch": 0.8997987067353083, "grad_norm": 1.2453104941660897, "learning_rate": 5.21855924936816e-07, "loss": 0.3638, "step": 25256 }, { "epoch": 0.8998343338618023, "grad_norm": 1.5729817728009932, "learning_rate": 5.214880869466221e-07, "loss": 0.4797, "step": 25257 }, { "epoch": 0.8998699609882965, "grad_norm": 1.3250492715662667, "learning_rate": 5.211203751694382e-07, "loss": 0.4369, "step": 25258 }, { "epoch": 0.8999055881147906, "grad_norm": 1.1204845965849293, "learning_rate": 5.207527896101627e-07, "loss": 0.3853, "step": 25259 }, { "epoch": 0.8999412152412847, "grad_norm": 1.5405160087222591, "learning_rate": 5.203853302736905e-07, "loss": 0.5737, "step": 25260 }, { "epoch": 0.8999768423677789, "grad_norm": 1.6196569788602022, "learning_rate": 5.200179971649122e-07, "loss": 0.427, "step": 25261 }, { "epoch": 0.9000124694942729, "grad_norm": 1.5657434524843548, "learning_rate": 5.196507902887215e-07, "loss": 0.5436, "step": 25262 }, { "epoch": 0.900048096620767, "grad_norm": 1.2402126904031958, "learning_rate": 5.192837096500058e-07, "loss": 0.3692, "step": 25263 }, { "epoch": 0.9000837237472612, "grad_norm": 1.3243907391226195, "learning_rate": 5.189167552536534e-07, "loss": 0.3803, "step": 25264 }, { "epoch": 0.9001193508737553, "grad_norm": 1.6940021701405816, "learning_rate": 5.185499271045536e-07, "loss": 0.4234, "step": 25265 }, { "epoch": 0.9001549780002494, "grad_norm": 1.2491087254379791, "learning_rate": 5.181832252075869e-07, "loss": 0.3801, "step": 25266 }, { "epoch": 0.9001906051267435, "grad_norm": 1.6601505837883117, "learning_rate": 5.178166495676385e-07, "loss": 0.452, "step": 25267 }, { "epoch": 0.9002262322532376, "grad_norm": 1.6195641415532092, "learning_rate": 5.174502001895887e-07, "loss": 0.4628, "step": 25268 }, { "epoch": 0.9002618593797317, "grad_norm": 1.5636047290318233, "learning_rate": 5.170838770783182e-07, "loss": 0.4628, "step": 25269 }, { "epoch": 0.9002974865062259, "grad_norm": 1.4275794850803019, "learning_rate": 5.16717680238703e-07, "loss": 0.4184, "step": 25270 }, { "epoch": 0.90033311363272, "grad_norm": 1.7849997483592406, "learning_rate": 5.163516096756216e-07, "loss": 0.5104, "step": 25271 }, { "epoch": 0.900368740759214, "grad_norm": 2.045575616597075, "learning_rate": 5.159856653939466e-07, "loss": 0.6012, "step": 25272 }, { "epoch": 0.9004043678857082, "grad_norm": 1.2793267773651122, "learning_rate": 5.156198473985508e-07, "loss": 0.3856, "step": 25273 }, { "epoch": 0.9004399950122023, "grad_norm": 1.2522440540400563, "learning_rate": 5.152541556943058e-07, "loss": 0.3964, "step": 25274 }, { "epoch": 0.9004756221386964, "grad_norm": 1.3127615262257661, "learning_rate": 5.148885902860822e-07, "loss": 0.4798, "step": 25275 }, { "epoch": 0.9005112492651905, "grad_norm": 1.4723022211356458, "learning_rate": 5.145231511787463e-07, "loss": 0.4458, "step": 25276 }, { "epoch": 0.9005468763916846, "grad_norm": 8.021057460541925, "learning_rate": 5.141578383771639e-07, "loss": 0.4068, "step": 25277 }, { "epoch": 0.9005825035181787, "grad_norm": 1.4705385150791366, "learning_rate": 5.137926518862013e-07, "loss": 0.4144, "step": 25278 }, { "epoch": 0.9006181306446729, "grad_norm": 2.1945926765733317, "learning_rate": 5.134275917107189e-07, "loss": 0.5066, "step": 25279 }, { "epoch": 0.900653757771167, "grad_norm": 1.8933570421826276, "learning_rate": 5.130626578555809e-07, "loss": 0.4353, "step": 25280 }, { "epoch": 0.900689384897661, "grad_norm": 1.5931867969873421, "learning_rate": 5.126978503256408e-07, "loss": 0.5143, "step": 25281 }, { "epoch": 0.9007250120241552, "grad_norm": 2.0034612506988005, "learning_rate": 5.123331691257627e-07, "loss": 0.6353, "step": 25282 }, { "epoch": 0.9007606391506493, "grad_norm": 1.274765587214785, "learning_rate": 5.119686142607993e-07, "loss": 0.3645, "step": 25283 }, { "epoch": 0.9007962662771434, "grad_norm": 1.4281602731437426, "learning_rate": 5.116041857356069e-07, "loss": 0.4251, "step": 25284 }, { "epoch": 0.9008318934036376, "grad_norm": 1.2486731630262171, "learning_rate": 5.112398835550348e-07, "loss": 0.4312, "step": 25285 }, { "epoch": 0.9008675205301316, "grad_norm": 1.9176895041638695, "learning_rate": 5.108757077239379e-07, "loss": 0.5258, "step": 25286 }, { "epoch": 0.9009031476566257, "grad_norm": 1.470491936165789, "learning_rate": 5.105116582471614e-07, "loss": 0.4796, "step": 25287 }, { "epoch": 0.9009387747831199, "grad_norm": 2.038051208271709, "learning_rate": 5.101477351295558e-07, "loss": 0.5184, "step": 25288 }, { "epoch": 0.900974401909614, "grad_norm": 1.463170773020496, "learning_rate": 5.09783938375965e-07, "loss": 0.4548, "step": 25289 }, { "epoch": 0.901010029036108, "grad_norm": 1.6626548392567588, "learning_rate": 5.094202679912352e-07, "loss": 0.4877, "step": 25290 }, { "epoch": 0.9010456561626022, "grad_norm": 2.310255329773862, "learning_rate": 5.090567239802091e-07, "loss": 0.5094, "step": 25291 }, { "epoch": 0.9010812832890963, "grad_norm": 1.2033626825415895, "learning_rate": 5.086933063477251e-07, "loss": 0.3572, "step": 25292 }, { "epoch": 0.9011169104155904, "grad_norm": 1.221638151795907, "learning_rate": 5.083300150986259e-07, "loss": 0.3701, "step": 25293 }, { "epoch": 0.9011525375420846, "grad_norm": 1.1272510303087773, "learning_rate": 5.079668502377444e-07, "loss": 0.3625, "step": 25294 }, { "epoch": 0.9011881646685787, "grad_norm": 1.7633977224696797, "learning_rate": 5.076038117699189e-07, "loss": 0.5283, "step": 25295 }, { "epoch": 0.9012237917950727, "grad_norm": 1.474214785600131, "learning_rate": 5.072408996999844e-07, "loss": 0.3845, "step": 25296 }, { "epoch": 0.9012594189215669, "grad_norm": 1.3059042883461298, "learning_rate": 5.068781140327727e-07, "loss": 0.4843, "step": 25297 }, { "epoch": 0.901295046048061, "grad_norm": 1.797909523705621, "learning_rate": 5.065154547731132e-07, "loss": 0.5495, "step": 25298 }, { "epoch": 0.9013306731745551, "grad_norm": 1.5509801610623721, "learning_rate": 5.061529219258366e-07, "loss": 0.5844, "step": 25299 }, { "epoch": 0.9013663003010493, "grad_norm": 1.2096744891598232, "learning_rate": 5.057905154957699e-07, "loss": 0.3864, "step": 25300 }, { "epoch": 0.9014019274275433, "grad_norm": 1.258383882533276, "learning_rate": 5.054282354877382e-07, "loss": 0.4339, "step": 25301 }, { "epoch": 0.9014375545540374, "grad_norm": 1.5398522236925696, "learning_rate": 5.050660819065667e-07, "loss": 0.5227, "step": 25302 }, { "epoch": 0.9014731816805316, "grad_norm": 1.8472138579541089, "learning_rate": 5.047040547570748e-07, "loss": 0.5541, "step": 25303 }, { "epoch": 0.9015088088070257, "grad_norm": 1.6154190710296874, "learning_rate": 5.043421540440863e-07, "loss": 0.5161, "step": 25304 }, { "epoch": 0.9015444359335197, "grad_norm": 1.2062803956123773, "learning_rate": 5.039803797724185e-07, "loss": 0.4385, "step": 25305 }, { "epoch": 0.9015800630600139, "grad_norm": 1.2620551184823723, "learning_rate": 5.03618731946891e-07, "loss": 0.4054, "step": 25306 }, { "epoch": 0.901615690186508, "grad_norm": 1.2502825191683413, "learning_rate": 5.032572105723166e-07, "loss": 0.4202, "step": 25307 }, { "epoch": 0.9016513173130021, "grad_norm": 1.3817056408355592, "learning_rate": 5.028958156535113e-07, "loss": 0.4048, "step": 25308 }, { "epoch": 0.9016869444394963, "grad_norm": 1.7229785046094983, "learning_rate": 5.025345471952847e-07, "loss": 0.5202, "step": 25309 }, { "epoch": 0.9017225715659903, "grad_norm": 1.4413734325414245, "learning_rate": 5.021734052024485e-07, "loss": 0.4228, "step": 25310 }, { "epoch": 0.9017581986924844, "grad_norm": 1.1986604396347256, "learning_rate": 5.018123896798132e-07, "loss": 0.3837, "step": 25311 }, { "epoch": 0.9017938258189786, "grad_norm": 1.5235675615627018, "learning_rate": 5.01451500632184e-07, "loss": 0.5047, "step": 25312 }, { "epoch": 0.9018294529454727, "grad_norm": 1.287583361840993, "learning_rate": 5.010907380643682e-07, "loss": 0.3859, "step": 25313 }, { "epoch": 0.9018650800719668, "grad_norm": 1.2623884663959943, "learning_rate": 5.007301019811683e-07, "loss": 0.3969, "step": 25314 }, { "epoch": 0.9019007071984609, "grad_norm": 1.301469843060884, "learning_rate": 5.003695923873875e-07, "loss": 0.4274, "step": 25315 }, { "epoch": 0.901936334324955, "grad_norm": 2.1518466202471283, "learning_rate": 5.000092092878228e-07, "loss": 0.521, "step": 25316 }, { "epoch": 0.9019719614514491, "grad_norm": 1.2639134120119553, "learning_rate": 4.996489526872783e-07, "loss": 0.4268, "step": 25317 }, { "epoch": 0.9020075885779433, "grad_norm": 1.3240248691894645, "learning_rate": 4.992888225905467e-07, "loss": 0.3885, "step": 25318 }, { "epoch": 0.9020432157044374, "grad_norm": 1.3032826992831403, "learning_rate": 4.989288190024255e-07, "loss": 0.4127, "step": 25319 }, { "epoch": 0.9020788428309314, "grad_norm": 1.567983484880034, "learning_rate": 4.985689419277073e-07, "loss": 0.4812, "step": 25320 }, { "epoch": 0.9021144699574256, "grad_norm": 1.155398569533862, "learning_rate": 4.982091913711862e-07, "loss": 0.3727, "step": 25321 }, { "epoch": 0.9021500970839197, "grad_norm": 2.0615637930537307, "learning_rate": 4.978495673376505e-07, "loss": 0.5251, "step": 25322 }, { "epoch": 0.9021857242104138, "grad_norm": 1.2466969747696452, "learning_rate": 4.974900698318885e-07, "loss": 0.3637, "step": 25323 }, { "epoch": 0.902221351336908, "grad_norm": 1.591830544847314, "learning_rate": 4.9713069885869e-07, "loss": 0.4031, "step": 25324 }, { "epoch": 0.902256978463402, "grad_norm": 1.6165884945170026, "learning_rate": 4.967714544228364e-07, "loss": 0.3963, "step": 25325 }, { "epoch": 0.9022926055898961, "grad_norm": 1.1204008297943755, "learning_rate": 4.964123365291129e-07, "loss": 0.3683, "step": 25326 }, { "epoch": 0.9023282327163903, "grad_norm": 1.5190237743386075, "learning_rate": 4.960533451823024e-07, "loss": 0.4687, "step": 25327 }, { "epoch": 0.9023638598428844, "grad_norm": 1.1567669585243494, "learning_rate": 4.956944803871866e-07, "loss": 0.3987, "step": 25328 }, { "epoch": 0.9023994869693784, "grad_norm": 1.1936754581579683, "learning_rate": 4.953357421485394e-07, "loss": 0.3978, "step": 25329 }, { "epoch": 0.9024351140958726, "grad_norm": 1.2490069780873423, "learning_rate": 4.949771304711415e-07, "loss": 0.4513, "step": 25330 }, { "epoch": 0.9024707412223667, "grad_norm": 1.2910554478224678, "learning_rate": 4.946186453597657e-07, "loss": 0.414, "step": 25331 }, { "epoch": 0.9025063683488608, "grad_norm": 1.3093907991403202, "learning_rate": 4.942602868191881e-07, "loss": 0.3968, "step": 25332 }, { "epoch": 0.902541995475355, "grad_norm": 2.150150971648059, "learning_rate": 4.939020548541762e-07, "loss": 0.5214, "step": 25333 }, { "epoch": 0.902577622601849, "grad_norm": 1.285888219565237, "learning_rate": 4.935439494695038e-07, "loss": 0.4119, "step": 25334 }, { "epoch": 0.9026132497283431, "grad_norm": 1.341753094005171, "learning_rate": 4.931859706699394e-07, "loss": 0.4086, "step": 25335 }, { "epoch": 0.9026488768548373, "grad_norm": 1.6070870597360711, "learning_rate": 4.92828118460248e-07, "loss": 0.4985, "step": 25336 }, { "epoch": 0.9026845039813314, "grad_norm": 1.3796760072937129, "learning_rate": 4.924703928451968e-07, "loss": 0.4011, "step": 25337 }, { "epoch": 0.9027201311078255, "grad_norm": 1.2129604874374795, "learning_rate": 4.921127938295456e-07, "loss": 0.4307, "step": 25338 }, { "epoch": 0.9027557582343196, "grad_norm": 1.71134830249512, "learning_rate": 4.917553214180603e-07, "loss": 0.4506, "step": 25339 }, { "epoch": 0.9027913853608137, "grad_norm": 1.624642831520648, "learning_rate": 4.913979756154963e-07, "loss": 0.5378, "step": 25340 }, { "epoch": 0.9028270124873078, "grad_norm": 1.4452930378088225, "learning_rate": 4.910407564266151e-07, "loss": 0.363, "step": 25341 }, { "epoch": 0.902862639613802, "grad_norm": 1.3443984760245566, "learning_rate": 4.906836638561729e-07, "loss": 0.4667, "step": 25342 }, { "epoch": 0.902898266740296, "grad_norm": 1.2898583015979304, "learning_rate": 4.90326697908925e-07, "loss": 0.4116, "step": 25343 }, { "epoch": 0.9029338938667901, "grad_norm": 1.8067502954259418, "learning_rate": 4.899698585896229e-07, "loss": 0.42, "step": 25344 }, { "epoch": 0.9029695209932843, "grad_norm": 1.7994574180726093, "learning_rate": 4.896131459030196e-07, "loss": 0.5103, "step": 25345 }, { "epoch": 0.9030051481197784, "grad_norm": 1.4461483945485643, "learning_rate": 4.892565598538656e-07, "loss": 0.4193, "step": 25346 }, { "epoch": 0.9030407752462725, "grad_norm": 1.917943708799173, "learning_rate": 4.889001004469063e-07, "loss": 0.4854, "step": 25347 }, { "epoch": 0.9030764023727667, "grad_norm": 1.735593483204116, "learning_rate": 4.885437676868909e-07, "loss": 0.5477, "step": 25348 }, { "epoch": 0.9031120294992607, "grad_norm": 1.3824914769817085, "learning_rate": 4.881875615785637e-07, "loss": 0.4161, "step": 25349 }, { "epoch": 0.9031476566257548, "grad_norm": 1.9628037134764957, "learning_rate": 4.878314821266672e-07, "loss": 0.4506, "step": 25350 }, { "epoch": 0.903183283752249, "grad_norm": 1.3827751520707068, "learning_rate": 4.874755293359435e-07, "loss": 0.396, "step": 25351 }, { "epoch": 0.9032189108787431, "grad_norm": 1.8858700810512514, "learning_rate": 4.871197032111319e-07, "loss": 0.5017, "step": 25352 }, { "epoch": 0.9032545380052371, "grad_norm": 1.45456287972408, "learning_rate": 4.867640037569699e-07, "loss": 0.5019, "step": 25353 }, { "epoch": 0.9032901651317313, "grad_norm": 1.5304592331177223, "learning_rate": 4.864084309781958e-07, "loss": 0.3964, "step": 25354 }, { "epoch": 0.9033257922582254, "grad_norm": 1.2378897491375274, "learning_rate": 4.860529848795414e-07, "loss": 0.4191, "step": 25355 }, { "epoch": 0.9033614193847195, "grad_norm": 1.8174175729627504, "learning_rate": 4.856976654657408e-07, "loss": 0.4513, "step": 25356 }, { "epoch": 0.9033970465112137, "grad_norm": 1.2849192147101625, "learning_rate": 4.85342472741528e-07, "loss": 0.4059, "step": 25357 }, { "epoch": 0.9034326736377077, "grad_norm": 1.538263176750516, "learning_rate": 4.849874067116289e-07, "loss": 0.5156, "step": 25358 }, { "epoch": 0.9034683007642018, "grad_norm": 1.4275655424842488, "learning_rate": 4.846324673807756e-07, "loss": 0.4486, "step": 25359 }, { "epoch": 0.903503927890696, "grad_norm": 1.3240048740778525, "learning_rate": 4.842776547536898e-07, "loss": 0.3395, "step": 25360 }, { "epoch": 0.9035395550171901, "grad_norm": 1.6351239108471276, "learning_rate": 4.839229688350988e-07, "loss": 0.5724, "step": 25361 }, { "epoch": 0.9035751821436842, "grad_norm": 1.2877892602268441, "learning_rate": 4.835684096297244e-07, "loss": 0.4145, "step": 25362 }, { "epoch": 0.9036108092701783, "grad_norm": 1.547673319507371, "learning_rate": 4.832139771422872e-07, "loss": 0.4644, "step": 25363 }, { "epoch": 0.9036464363966724, "grad_norm": 1.5789976776449528, "learning_rate": 4.828596713775091e-07, "loss": 0.538, "step": 25364 }, { "epoch": 0.9036820635231665, "grad_norm": 1.1043256469532263, "learning_rate": 4.825054923401074e-07, "loss": 0.354, "step": 25365 }, { "epoch": 0.9037176906496607, "grad_norm": 1.8889974596992003, "learning_rate": 4.821514400347949e-07, "loss": 0.5097, "step": 25366 }, { "epoch": 0.9037533177761548, "grad_norm": 1.3461634489693706, "learning_rate": 4.817975144662912e-07, "loss": 0.3678, "step": 25367 }, { "epoch": 0.9037889449026488, "grad_norm": 1.3945315858086516, "learning_rate": 4.814437156393048e-07, "loss": 0.3848, "step": 25368 }, { "epoch": 0.903824572029143, "grad_norm": 1.5546759638679097, "learning_rate": 4.810900435585486e-07, "loss": 0.5105, "step": 25369 }, { "epoch": 0.9038601991556371, "grad_norm": 1.7849404906073358, "learning_rate": 4.80736498228731e-07, "loss": 0.5134, "step": 25370 }, { "epoch": 0.9038958262821312, "grad_norm": 1.3755639352521185, "learning_rate": 4.803830796545605e-07, "loss": 0.3982, "step": 25371 }, { "epoch": 0.9039314534086254, "grad_norm": 1.3817534211915121, "learning_rate": 4.800297878407456e-07, "loss": 0.4968, "step": 25372 }, { "epoch": 0.9039670805351194, "grad_norm": 1.276832959024542, "learning_rate": 4.796766227919858e-07, "loss": 0.3988, "step": 25373 }, { "epoch": 0.9040027076616135, "grad_norm": 1.4332042704515637, "learning_rate": 4.793235845129885e-07, "loss": 0.4214, "step": 25374 }, { "epoch": 0.9040383347881077, "grad_norm": 1.8961477443508705, "learning_rate": 4.789706730084498e-07, "loss": 0.5015, "step": 25375 }, { "epoch": 0.9040739619146018, "grad_norm": 1.2929960322602096, "learning_rate": 4.786178882830727e-07, "loss": 0.4125, "step": 25376 }, { "epoch": 0.9041095890410958, "grad_norm": 1.756417136766746, "learning_rate": 4.782652303415525e-07, "loss": 0.526, "step": 25377 }, { "epoch": 0.90414521616759, "grad_norm": 1.2649831976962445, "learning_rate": 4.779126991885863e-07, "loss": 0.3687, "step": 25378 }, { "epoch": 0.9041808432940841, "grad_norm": 1.7196212358574416, "learning_rate": 4.775602948288683e-07, "loss": 0.4842, "step": 25379 }, { "epoch": 0.9042164704205782, "grad_norm": 1.3853567993907474, "learning_rate": 4.772080172670901e-07, "loss": 0.4053, "step": 25380 }, { "epoch": 0.9042520975470724, "grad_norm": 1.278564104612699, "learning_rate": 4.768558665079448e-07, "loss": 0.3791, "step": 25381 }, { "epoch": 0.9042877246735664, "grad_norm": 1.9051166018192704, "learning_rate": 4.765038425561186e-07, "loss": 0.4808, "step": 25382 }, { "epoch": 0.9043233518000605, "grad_norm": 1.7237338964817837, "learning_rate": 4.761519454163011e-07, "loss": 0.4744, "step": 25383 }, { "epoch": 0.9043589789265547, "grad_norm": 1.4406054226685057, "learning_rate": 4.7580017509317623e-07, "loss": 0.3928, "step": 25384 }, { "epoch": 0.9043946060530488, "grad_norm": 1.9570541096863587, "learning_rate": 4.754485315914292e-07, "loss": 0.525, "step": 25385 }, { "epoch": 0.9044302331795429, "grad_norm": 1.336876216458214, "learning_rate": 4.750970149157419e-07, "loss": 0.3813, "step": 25386 }, { "epoch": 0.904465860306037, "grad_norm": 1.5094558839004553, "learning_rate": 4.7474562507079713e-07, "loss": 0.4591, "step": 25387 }, { "epoch": 0.9045014874325311, "grad_norm": 1.5671878977309377, "learning_rate": 4.7439436206127007e-07, "loss": 0.5271, "step": 25388 }, { "epoch": 0.9045371145590252, "grad_norm": 1.6627703180604727, "learning_rate": 4.740432258918415e-07, "loss": 0.4633, "step": 25389 }, { "epoch": 0.9045727416855194, "grad_norm": 3.0423667166792656, "learning_rate": 4.736922165671853e-07, "loss": 0.5686, "step": 25390 }, { "epoch": 0.9046083688120135, "grad_norm": 1.3257669795103537, "learning_rate": 4.733413340919746e-07, "loss": 0.4185, "step": 25391 }, { "epoch": 0.9046439959385075, "grad_norm": 1.593601649356715, "learning_rate": 4.7299057847088545e-07, "loss": 0.5749, "step": 25392 }, { "epoch": 0.9046796230650017, "grad_norm": 1.3840635555010168, "learning_rate": 4.726399497085832e-07, "loss": 0.4078, "step": 25393 }, { "epoch": 0.9047152501914958, "grad_norm": 1.4561906782393383, "learning_rate": 4.722894478097395e-07, "loss": 0.3914, "step": 25394 }, { "epoch": 0.9047508773179899, "grad_norm": 1.877618469638891, "learning_rate": 4.7193907277902186e-07, "loss": 0.4066, "step": 25395 }, { "epoch": 0.904786504444484, "grad_norm": 1.1577035530609638, "learning_rate": 4.7158882462109535e-07, "loss": 0.4092, "step": 25396 }, { "epoch": 0.9048221315709781, "grad_norm": 1.3705645410250384, "learning_rate": 4.7123870334062295e-07, "loss": 0.3707, "step": 25397 }, { "epoch": 0.9048577586974722, "grad_norm": 1.5318246638339796, "learning_rate": 4.708887089422687e-07, "loss": 0.4798, "step": 25398 }, { "epoch": 0.9048933858239664, "grad_norm": 1.4824406656487628, "learning_rate": 4.7053884143069106e-07, "loss": 0.5313, "step": 25399 }, { "epoch": 0.9049290129504605, "grad_norm": 1.2266203393690458, "learning_rate": 4.7018910081054856e-07, "loss": 0.3777, "step": 25400 }, { "epoch": 0.9049646400769545, "grad_norm": 1.6746437688010674, "learning_rate": 4.698394870864997e-07, "loss": 0.5163, "step": 25401 }, { "epoch": 0.9050002672034487, "grad_norm": 2.1322835527151227, "learning_rate": 4.6949000026319966e-07, "loss": 0.4867, "step": 25402 }, { "epoch": 0.9050358943299428, "grad_norm": 1.524428029313757, "learning_rate": 4.6914064034530137e-07, "loss": 0.4833, "step": 25403 }, { "epoch": 0.9050715214564369, "grad_norm": 1.3451901942993512, "learning_rate": 4.6879140733745663e-07, "loss": 0.4273, "step": 25404 }, { "epoch": 0.9051071485829311, "grad_norm": 1.4688621752360047, "learning_rate": 4.6844230124431735e-07, "loss": 0.4043, "step": 25405 }, { "epoch": 0.9051427757094251, "grad_norm": 1.3594786802132077, "learning_rate": 4.6809332207053083e-07, "loss": 0.3836, "step": 25406 }, { "epoch": 0.9051784028359192, "grad_norm": 1.6191158860014945, "learning_rate": 4.6774446982074453e-07, "loss": 0.4021, "step": 25407 }, { "epoch": 0.9052140299624134, "grad_norm": 1.5321708609084366, "learning_rate": 4.6739574449960024e-07, "loss": 0.4926, "step": 25408 }, { "epoch": 0.9052496570889075, "grad_norm": 1.6097919119598012, "learning_rate": 4.6704714611174764e-07, "loss": 0.4454, "step": 25409 }, { "epoch": 0.9052852842154016, "grad_norm": 2.591734756262429, "learning_rate": 4.666986746618241e-07, "loss": 0.5363, "step": 25410 }, { "epoch": 0.9053209113418957, "grad_norm": 1.73141554760694, "learning_rate": 4.663503301544725e-07, "loss": 0.492, "step": 25411 }, { "epoch": 0.9053565384683898, "grad_norm": 1.5705856713739343, "learning_rate": 4.6600211259432813e-07, "loss": 0.4296, "step": 25412 }, { "epoch": 0.9053921655948839, "grad_norm": 1.513724944926964, "learning_rate": 4.656540219860317e-07, "loss": 0.5304, "step": 25413 }, { "epoch": 0.9054277927213781, "grad_norm": 1.5277692508339145, "learning_rate": 4.653060583342139e-07, "loss": 0.5541, "step": 25414 }, { "epoch": 0.9054634198478722, "grad_norm": 1.78190111864411, "learning_rate": 4.6495822164351e-07, "loss": 0.5623, "step": 25415 }, { "epoch": 0.9054990469743662, "grad_norm": 1.231793866089482, "learning_rate": 4.646105119185529e-07, "loss": 0.4038, "step": 25416 }, { "epoch": 0.9055346741008604, "grad_norm": 1.3056185032772436, "learning_rate": 4.642629291639711e-07, "loss": 0.4202, "step": 25417 }, { "epoch": 0.9055703012273545, "grad_norm": 1.3280192025227575, "learning_rate": 4.6391547338439533e-07, "loss": 0.3766, "step": 25418 }, { "epoch": 0.9056059283538486, "grad_norm": 1.4926265028143484, "learning_rate": 4.6356814458444866e-07, "loss": 0.4937, "step": 25419 }, { "epoch": 0.9056415554803428, "grad_norm": 1.7160044536412267, "learning_rate": 4.632209427687595e-07, "loss": 0.5114, "step": 25420 }, { "epoch": 0.9056771826068368, "grad_norm": 1.378157119574228, "learning_rate": 4.6287386794194756e-07, "loss": 0.409, "step": 25421 }, { "epoch": 0.9057128097333309, "grad_norm": 1.6926988313244287, "learning_rate": 4.6252692010863573e-07, "loss": 0.5262, "step": 25422 }, { "epoch": 0.9057484368598251, "grad_norm": 1.5522915299201574, "learning_rate": 4.6218009927344486e-07, "loss": 0.4892, "step": 25423 }, { "epoch": 0.9057840639863192, "grad_norm": 2.3084092995843157, "learning_rate": 4.6183340544099343e-07, "loss": 0.4119, "step": 25424 }, { "epoch": 0.9058196911128132, "grad_norm": 1.2864390629888784, "learning_rate": 4.614868386158955e-07, "loss": 0.4126, "step": 25425 }, { "epoch": 0.9058553182393074, "grad_norm": 1.2237784755346415, "learning_rate": 4.611403988027674e-07, "loss": 0.4074, "step": 25426 }, { "epoch": 0.9058909453658015, "grad_norm": 1.8578514761191707, "learning_rate": 4.6079408600622435e-07, "loss": 0.4636, "step": 25427 }, { "epoch": 0.9059265724922956, "grad_norm": 1.287845845787166, "learning_rate": 4.6044790023087373e-07, "loss": 0.395, "step": 25428 }, { "epoch": 0.9059621996187898, "grad_norm": 1.4574046353875787, "learning_rate": 4.6010184148132854e-07, "loss": 0.424, "step": 25429 }, { "epoch": 0.9059978267452838, "grad_norm": 1.3302143921326914, "learning_rate": 4.597559097621929e-07, "loss": 0.4325, "step": 25430 }, { "epoch": 0.9060334538717779, "grad_norm": 1.6871554528406876, "learning_rate": 4.594101050780764e-07, "loss": 0.5087, "step": 25431 }, { "epoch": 0.9060690809982721, "grad_norm": 1.310090543782897, "learning_rate": 4.5906442743358314e-07, "loss": 0.4315, "step": 25432 }, { "epoch": 0.9061047081247662, "grad_norm": 1.4113300809733986, "learning_rate": 4.587188768333173e-07, "loss": 0.5175, "step": 25433 }, { "epoch": 0.9061403352512604, "grad_norm": 1.1820035162087403, "learning_rate": 4.5837345328187623e-07, "loss": 0.3626, "step": 25434 }, { "epoch": 0.9061759623777544, "grad_norm": 1.2481409265606003, "learning_rate": 4.5802815678386405e-07, "loss": 0.3434, "step": 25435 }, { "epoch": 0.9062115895042485, "grad_norm": 1.3371985708288054, "learning_rate": 4.576829873438737e-07, "loss": 0.4036, "step": 25436 }, { "epoch": 0.9062472166307427, "grad_norm": 1.5776800850762065, "learning_rate": 4.573379449665061e-07, "loss": 0.5731, "step": 25437 }, { "epoch": 0.9062828437572368, "grad_norm": 1.5996653214796925, "learning_rate": 4.5699302965635185e-07, "loss": 0.5256, "step": 25438 }, { "epoch": 0.9063184708837309, "grad_norm": 1.2732963318969468, "learning_rate": 4.5664824141800623e-07, "loss": 0.384, "step": 25439 }, { "epoch": 0.906354098010225, "grad_norm": 1.9045691672270935, "learning_rate": 4.563035802560611e-07, "loss": 0.3753, "step": 25440 }, { "epoch": 0.9063897251367191, "grad_norm": 2.5632265880680634, "learning_rate": 4.559590461751029e-07, "loss": 0.4937, "step": 25441 }, { "epoch": 0.9064253522632132, "grad_norm": 1.9631144275426464, "learning_rate": 4.556146391797234e-07, "loss": 0.4891, "step": 25442 }, { "epoch": 0.9064609793897074, "grad_norm": 1.9500988436874378, "learning_rate": 4.5527035927450337e-07, "loss": 0.5766, "step": 25443 }, { "epoch": 0.9064966065162015, "grad_norm": 1.5680702513942755, "learning_rate": 4.549262064640314e-07, "loss": 0.5192, "step": 25444 }, { "epoch": 0.9065322336426955, "grad_norm": 1.140634626131165, "learning_rate": 4.545821807528861e-07, "loss": 0.4013, "step": 25445 }, { "epoch": 0.9065678607691897, "grad_norm": 1.5891181251932103, "learning_rate": 4.542382821456537e-07, "loss": 0.4909, "step": 25446 }, { "epoch": 0.9066034878956838, "grad_norm": 1.394910895208202, "learning_rate": 4.538945106469095e-07, "loss": 0.3772, "step": 25447 }, { "epoch": 0.9066391150221779, "grad_norm": 1.479404816090729, "learning_rate": 4.535508662612331e-07, "loss": 0.4757, "step": 25448 }, { "epoch": 0.906674742148672, "grad_norm": 2.0407546578854348, "learning_rate": 4.5320734899319873e-07, "loss": 0.4951, "step": 25449 }, { "epoch": 0.9067103692751661, "grad_norm": 1.3771223311224974, "learning_rate": 4.5286395884738045e-07, "loss": 0.5084, "step": 25450 }, { "epoch": 0.9067459964016602, "grad_norm": 1.1942177523621293, "learning_rate": 4.5252069582835345e-07, "loss": 0.3814, "step": 25451 }, { "epoch": 0.9067816235281544, "grad_norm": 1.38250773049305, "learning_rate": 4.5217755994068525e-07, "loss": 0.3811, "step": 25452 }, { "epoch": 0.9068172506546485, "grad_norm": 2.2744077165793755, "learning_rate": 4.518345511889466e-07, "loss": 0.5687, "step": 25453 }, { "epoch": 0.9068528777811425, "grad_norm": 1.7602744520467628, "learning_rate": 4.5149166957770495e-07, "loss": 0.4851, "step": 25454 }, { "epoch": 0.9068885049076367, "grad_norm": 1.740819069263611, "learning_rate": 4.511489151115267e-07, "loss": 0.4906, "step": 25455 }, { "epoch": 0.9069241320341308, "grad_norm": 1.305862710990622, "learning_rate": 4.508062877949737e-07, "loss": 0.3783, "step": 25456 }, { "epoch": 0.9069597591606249, "grad_norm": 1.4321849039110026, "learning_rate": 4.504637876326101e-07, "loss": 0.4638, "step": 25457 }, { "epoch": 0.9069953862871191, "grad_norm": 1.3169410903074465, "learning_rate": 4.501214146289956e-07, "loss": 0.3925, "step": 25458 }, { "epoch": 0.9070310134136131, "grad_norm": 1.5505732242897685, "learning_rate": 4.497791687886888e-07, "loss": 0.4686, "step": 25459 }, { "epoch": 0.9070666405401072, "grad_norm": 1.3088461570746304, "learning_rate": 4.4943705011624704e-07, "loss": 0.3696, "step": 25460 }, { "epoch": 0.9071022676666014, "grad_norm": 1.35970994715551, "learning_rate": 4.4909505861622794e-07, "loss": 0.4008, "step": 25461 }, { "epoch": 0.9071378947930955, "grad_norm": 2.5571324452299162, "learning_rate": 4.4875319429318334e-07, "loss": 0.5551, "step": 25462 }, { "epoch": 0.9071735219195896, "grad_norm": 1.7017893476917583, "learning_rate": 4.4841145715166625e-07, "loss": 0.4055, "step": 25463 }, { "epoch": 0.9072091490460837, "grad_norm": 1.7475816641250101, "learning_rate": 4.480698471962264e-07, "loss": 0.5108, "step": 25464 }, { "epoch": 0.9072447761725778, "grad_norm": 1.7853216339236921, "learning_rate": 4.477283644314134e-07, "loss": 0.5661, "step": 25465 }, { "epoch": 0.9072804032990719, "grad_norm": 1.3901521489084254, "learning_rate": 4.473870088617738e-07, "loss": 0.4073, "step": 25466 }, { "epoch": 0.9073160304255661, "grad_norm": 1.3831107271467304, "learning_rate": 4.4704578049185267e-07, "loss": 0.4862, "step": 25467 }, { "epoch": 0.9073516575520602, "grad_norm": 1.1943356209191611, "learning_rate": 4.467046793261931e-07, "loss": 0.392, "step": 25468 }, { "epoch": 0.9073872846785542, "grad_norm": 1.7758192492754954, "learning_rate": 4.4636370536933927e-07, "loss": 0.4944, "step": 25469 }, { "epoch": 0.9074229118050484, "grad_norm": 1.183355075723453, "learning_rate": 4.4602285862583083e-07, "loss": 0.3468, "step": 25470 }, { "epoch": 0.9074585389315425, "grad_norm": 1.2600942879257888, "learning_rate": 4.456821391002042e-07, "loss": 0.3887, "step": 25471 }, { "epoch": 0.9074941660580366, "grad_norm": 1.7408489284459956, "learning_rate": 4.45341546796999e-07, "loss": 0.4159, "step": 25472 }, { "epoch": 0.9075297931845308, "grad_norm": 1.1223580263846855, "learning_rate": 4.450010817207495e-07, "loss": 0.3572, "step": 25473 }, { "epoch": 0.9075654203110248, "grad_norm": 1.3603502194792092, "learning_rate": 4.4466074387598867e-07, "loss": 0.4046, "step": 25474 }, { "epoch": 0.9076010474375189, "grad_norm": 1.7229526840682585, "learning_rate": 4.4432053326724845e-07, "loss": 0.5272, "step": 25475 }, { "epoch": 0.9076366745640131, "grad_norm": 1.6009401944307327, "learning_rate": 4.4398044989905964e-07, "loss": 0.5213, "step": 25476 }, { "epoch": 0.9076723016905072, "grad_norm": 1.6595811577602668, "learning_rate": 4.4364049377595196e-07, "loss": 0.4039, "step": 25477 }, { "epoch": 0.9077079288170012, "grad_norm": 1.6795357012977041, "learning_rate": 4.433006649024485e-07, "loss": 0.5735, "step": 25478 }, { "epoch": 0.9077435559434954, "grad_norm": 2.331569513713018, "learning_rate": 4.4296096328307893e-07, "loss": 0.4995, "step": 25479 }, { "epoch": 0.9077791830699895, "grad_norm": 1.4885608174599898, "learning_rate": 4.4262138892236185e-07, "loss": 0.4878, "step": 25480 }, { "epoch": 0.9078148101964836, "grad_norm": 2.108186053954757, "learning_rate": 4.4228194182482254e-07, "loss": 0.5162, "step": 25481 }, { "epoch": 0.9078504373229778, "grad_norm": 1.236929184972626, "learning_rate": 4.419426219949785e-07, "loss": 0.3779, "step": 25482 }, { "epoch": 0.9078860644494718, "grad_norm": 1.6310820762157885, "learning_rate": 4.4160342943734723e-07, "loss": 0.5684, "step": 25483 }, { "epoch": 0.9079216915759659, "grad_norm": 1.2482249273826171, "learning_rate": 4.4126436415645066e-07, "loss": 0.3791, "step": 25484 }, { "epoch": 0.9079573187024601, "grad_norm": 1.9886756386565818, "learning_rate": 4.4092542615679855e-07, "loss": 0.4572, "step": 25485 }, { "epoch": 0.9079929458289542, "grad_norm": 1.213910566924844, "learning_rate": 4.405866154429073e-07, "loss": 0.3381, "step": 25486 }, { "epoch": 0.9080285729554483, "grad_norm": 1.224841007846643, "learning_rate": 4.402479320192854e-07, "loss": 0.3599, "step": 25487 }, { "epoch": 0.9080642000819424, "grad_norm": 1.621997664165071, "learning_rate": 4.39909375890446e-07, "loss": 0.5193, "step": 25488 }, { "epoch": 0.9080998272084365, "grad_norm": 1.5350402289302967, "learning_rate": 4.395709470608933e-07, "loss": 0.5169, "step": 25489 }, { "epoch": 0.9081354543349306, "grad_norm": 1.5823512927527597, "learning_rate": 4.392326455351359e-07, "loss": 0.5742, "step": 25490 }, { "epoch": 0.9081710814614248, "grad_norm": 1.336502266891588, "learning_rate": 4.38894471317679e-07, "loss": 0.4305, "step": 25491 }, { "epoch": 0.9082067085879189, "grad_norm": 1.7556772126491835, "learning_rate": 4.3855642441302584e-07, "loss": 0.4789, "step": 25492 }, { "epoch": 0.9082423357144129, "grad_norm": 1.3778658515400295, "learning_rate": 4.3821850482567595e-07, "loss": 0.4034, "step": 25493 }, { "epoch": 0.9082779628409071, "grad_norm": 1.5969485609616072, "learning_rate": 4.3788071256013033e-07, "loss": 0.4369, "step": 25494 }, { "epoch": 0.9083135899674012, "grad_norm": 1.445435502328923, "learning_rate": 4.375430476208864e-07, "loss": 0.3994, "step": 25495 }, { "epoch": 0.9083492170938953, "grad_norm": 1.6871714415144268, "learning_rate": 4.3720551001243947e-07, "loss": 0.5909, "step": 25496 }, { "epoch": 0.9083848442203895, "grad_norm": 1.74226041508625, "learning_rate": 4.36868099739286e-07, "loss": 0.518, "step": 25497 }, { "epoch": 0.9084204713468835, "grad_norm": 1.4009073159314442, "learning_rate": 4.365308168059179e-07, "loss": 0.419, "step": 25498 }, { "epoch": 0.9084560984733776, "grad_norm": 1.7760133585463433, "learning_rate": 4.361936612168272e-07, "loss": 0.5678, "step": 25499 }, { "epoch": 0.9084917255998718, "grad_norm": 1.302948672061173, "learning_rate": 4.3585663297650236e-07, "loss": 0.3835, "step": 25500 }, { "epoch": 0.9085273527263659, "grad_norm": 1.3803930053476368, "learning_rate": 4.3551973208943223e-07, "loss": 0.4139, "step": 25501 }, { "epoch": 0.9085629798528599, "grad_norm": 1.3297058756802576, "learning_rate": 4.3518295856009974e-07, "loss": 0.4083, "step": 25502 }, { "epoch": 0.9085986069793541, "grad_norm": 3.0175648021481223, "learning_rate": 4.3484631239299356e-07, "loss": 0.5056, "step": 25503 }, { "epoch": 0.9086342341058482, "grad_norm": 1.1835544815019694, "learning_rate": 4.345097935925935e-07, "loss": 0.3919, "step": 25504 }, { "epoch": 0.9086698612323423, "grad_norm": 1.4058303319320726, "learning_rate": 4.3417340216338147e-07, "loss": 0.395, "step": 25505 }, { "epoch": 0.9087054883588365, "grad_norm": 1.7029978769515977, "learning_rate": 4.3383713810983605e-07, "loss": 0.5076, "step": 25506 }, { "epoch": 0.9087411154853305, "grad_norm": 1.4126326890184853, "learning_rate": 4.3350100143643604e-07, "loss": 0.4562, "step": 25507 }, { "epoch": 0.9087767426118246, "grad_norm": 1.528116595076645, "learning_rate": 4.3316499214765775e-07, "loss": 0.5331, "step": 25508 }, { "epoch": 0.9088123697383188, "grad_norm": 1.2943122994571958, "learning_rate": 4.328291102479731e-07, "loss": 0.4681, "step": 25509 }, { "epoch": 0.9088479968648129, "grad_norm": 1.679277620722535, "learning_rate": 4.324933557418576e-07, "loss": 0.5142, "step": 25510 }, { "epoch": 0.908883623991307, "grad_norm": 1.5082039832817926, "learning_rate": 4.321577286337786e-07, "loss": 0.475, "step": 25511 }, { "epoch": 0.9089192511178011, "grad_norm": 1.3932635434675995, "learning_rate": 4.31822228928207e-07, "loss": 0.4899, "step": 25512 }, { "epoch": 0.9089548782442952, "grad_norm": 1.2590378198981385, "learning_rate": 4.3148685662961043e-07, "loss": 0.3939, "step": 25513 }, { "epoch": 0.9089905053707893, "grad_norm": 1.3354980135411194, "learning_rate": 4.3115161174245526e-07, "loss": 0.3841, "step": 25514 }, { "epoch": 0.9090261324972835, "grad_norm": 1.5930908585819006, "learning_rate": 4.308164942712034e-07, "loss": 0.5834, "step": 25515 }, { "epoch": 0.9090617596237776, "grad_norm": 1.232709997854536, "learning_rate": 4.304815042203203e-07, "loss": 0.4126, "step": 25516 }, { "epoch": 0.9090973867502716, "grad_norm": 1.7310682881172863, "learning_rate": 4.3014664159426234e-07, "loss": 0.5214, "step": 25517 }, { "epoch": 0.9091330138767658, "grad_norm": 1.5276199470149576, "learning_rate": 4.298119063974915e-07, "loss": 0.4072, "step": 25518 }, { "epoch": 0.9091686410032599, "grad_norm": 1.6124456661805524, "learning_rate": 4.294772986344653e-07, "loss": 0.5075, "step": 25519 }, { "epoch": 0.909204268129754, "grad_norm": 1.694570681898326, "learning_rate": 4.291428183096369e-07, "loss": 0.5065, "step": 25520 }, { "epoch": 0.9092398952562482, "grad_norm": 1.3433743336969164, "learning_rate": 4.2880846542746156e-07, "loss": 0.4027, "step": 25521 }, { "epoch": 0.9092755223827422, "grad_norm": 1.6755821740366532, "learning_rate": 4.284742399923925e-07, "loss": 0.4784, "step": 25522 }, { "epoch": 0.9093111495092363, "grad_norm": 1.2047255872576448, "learning_rate": 4.2814014200887935e-07, "loss": 0.3746, "step": 25523 }, { "epoch": 0.9093467766357305, "grad_norm": 1.8091095150752423, "learning_rate": 4.278061714813686e-07, "loss": 0.5017, "step": 25524 }, { "epoch": 0.9093824037622246, "grad_norm": 1.1123398993301046, "learning_rate": 4.274723284143112e-07, "loss": 0.3214, "step": 25525 }, { "epoch": 0.9094180308887186, "grad_norm": 1.5550833902550905, "learning_rate": 4.2713861281214905e-07, "loss": 0.469, "step": 25526 }, { "epoch": 0.9094536580152128, "grad_norm": 1.1267076121419584, "learning_rate": 4.268050246793276e-07, "loss": 0.4198, "step": 25527 }, { "epoch": 0.9094892851417069, "grad_norm": 1.3126979104671277, "learning_rate": 4.2647156402028876e-07, "loss": 0.4192, "step": 25528 }, { "epoch": 0.909524912268201, "grad_norm": 1.6550489465358715, "learning_rate": 4.2613823083947346e-07, "loss": 0.5159, "step": 25529 }, { "epoch": 0.9095605393946952, "grad_norm": 1.459968170290356, "learning_rate": 4.2580502514131705e-07, "loss": 0.4596, "step": 25530 }, { "epoch": 0.9095961665211892, "grad_norm": 1.5847722564069249, "learning_rate": 4.2547194693026037e-07, "loss": 0.5395, "step": 25531 }, { "epoch": 0.9096317936476833, "grad_norm": 1.5807000710567503, "learning_rate": 4.2513899621073774e-07, "loss": 0.4384, "step": 25532 }, { "epoch": 0.9096674207741775, "grad_norm": 1.43200196458308, "learning_rate": 4.2480617298718e-07, "loss": 0.5079, "step": 25533 }, { "epoch": 0.9097030479006716, "grad_norm": 1.2653130286082512, "learning_rate": 4.2447347726402134e-07, "loss": 0.4117, "step": 25534 }, { "epoch": 0.9097386750271657, "grad_norm": 1.9642516690237461, "learning_rate": 4.241409090456905e-07, "loss": 0.5737, "step": 25535 }, { "epoch": 0.9097743021536598, "grad_norm": 1.2784729277662197, "learning_rate": 4.238084683366184e-07, "loss": 0.4128, "step": 25536 }, { "epoch": 0.9098099292801539, "grad_norm": 1.7150433589044247, "learning_rate": 4.234761551412281e-07, "loss": 0.5606, "step": 25537 }, { "epoch": 0.909845556406648, "grad_norm": 1.1796745124891128, "learning_rate": 4.2314396946394833e-07, "loss": 0.3746, "step": 25538 }, { "epoch": 0.9098811835331422, "grad_norm": 1.4787922895656413, "learning_rate": 4.2281191130919885e-07, "loss": 0.5002, "step": 25539 }, { "epoch": 0.9099168106596363, "grad_norm": 1.8306770174917812, "learning_rate": 4.2247998068140395e-07, "loss": 0.4816, "step": 25540 }, { "epoch": 0.9099524377861303, "grad_norm": 1.272302504746533, "learning_rate": 4.221481775849812e-07, "loss": 0.375, "step": 25541 }, { "epoch": 0.9099880649126245, "grad_norm": 1.3453089037346404, "learning_rate": 4.218165020243503e-07, "loss": 0.4137, "step": 25542 }, { "epoch": 0.9100236920391186, "grad_norm": 1.6673065123104265, "learning_rate": 4.2148495400392675e-07, "loss": 0.5652, "step": 25543 }, { "epoch": 0.9100593191656127, "grad_norm": 1.4588994762797702, "learning_rate": 4.2115353352812694e-07, "loss": 0.5048, "step": 25544 }, { "epoch": 0.9100949462921069, "grad_norm": 1.6906790917364714, "learning_rate": 4.20822240601364e-07, "loss": 0.5743, "step": 25545 }, { "epoch": 0.9101305734186009, "grad_norm": 1.3346950857952486, "learning_rate": 4.2049107522804664e-07, "loss": 0.4252, "step": 25546 }, { "epoch": 0.910166200545095, "grad_norm": 1.4297725419947014, "learning_rate": 4.2016003741258805e-07, "loss": 0.4143, "step": 25547 }, { "epoch": 0.9102018276715892, "grad_norm": 1.42699944938734, "learning_rate": 4.198291271593924e-07, "loss": 0.3792, "step": 25548 }, { "epoch": 0.9102374547980833, "grad_norm": 1.3834093522143032, "learning_rate": 4.194983444728684e-07, "loss": 0.4084, "step": 25549 }, { "epoch": 0.9102730819245773, "grad_norm": 2.0259164608750213, "learning_rate": 4.191676893574215e-07, "loss": 0.5554, "step": 25550 }, { "epoch": 0.9103087090510715, "grad_norm": 1.398878515662295, "learning_rate": 4.188371618174536e-07, "loss": 0.3963, "step": 25551 }, { "epoch": 0.9103443361775656, "grad_norm": 1.5970956918796613, "learning_rate": 4.185067618573646e-07, "loss": 0.4423, "step": 25552 }, { "epoch": 0.9103799633040597, "grad_norm": 1.679097508917313, "learning_rate": 4.181764894815543e-07, "loss": 0.5256, "step": 25553 }, { "epoch": 0.9104155904305539, "grad_norm": 1.2521511464923507, "learning_rate": 4.1784634469442363e-07, "loss": 0.407, "step": 25554 }, { "epoch": 0.9104512175570479, "grad_norm": 1.3440624315422958, "learning_rate": 4.175163275003658e-07, "loss": 0.4043, "step": 25555 }, { "epoch": 0.910486844683542, "grad_norm": 1.2872317704008989, "learning_rate": 4.17186437903776e-07, "loss": 0.4072, "step": 25556 }, { "epoch": 0.9105224718100362, "grad_norm": 1.769874310511838, "learning_rate": 4.1685667590904534e-07, "loss": 0.5265, "step": 25557 }, { "epoch": 0.9105580989365303, "grad_norm": 1.4309794925479353, "learning_rate": 4.165270415205669e-07, "loss": 0.4869, "step": 25558 }, { "epoch": 0.9105937260630244, "grad_norm": 1.20545371477474, "learning_rate": 4.161975347427294e-07, "loss": 0.3788, "step": 25559 }, { "epoch": 0.9106293531895185, "grad_norm": 2.0887877997952393, "learning_rate": 4.1586815557992043e-07, "loss": 0.4748, "step": 25560 }, { "epoch": 0.9106649803160126, "grad_norm": 1.243651625265481, "learning_rate": 4.155389040365265e-07, "loss": 0.4059, "step": 25561 }, { "epoch": 0.9107006074425067, "grad_norm": 1.4383580476464128, "learning_rate": 4.152097801169308e-07, "loss": 0.4007, "step": 25562 }, { "epoch": 0.9107362345690009, "grad_norm": 1.6174570739943346, "learning_rate": 4.1488078382551646e-07, "loss": 0.4933, "step": 25563 }, { "epoch": 0.910771861695495, "grad_norm": 2.097523127586087, "learning_rate": 4.145519151666633e-07, "loss": 0.5084, "step": 25564 }, { "epoch": 0.910807488821989, "grad_norm": 1.5995721895335366, "learning_rate": 4.142231741447511e-07, "loss": 0.5101, "step": 25565 }, { "epoch": 0.9108431159484832, "grad_norm": 1.677864281855745, "learning_rate": 4.138945607641576e-07, "loss": 0.4775, "step": 25566 }, { "epoch": 0.9108787430749773, "grad_norm": 1.5639245641983694, "learning_rate": 4.135660750292592e-07, "loss": 0.4252, "step": 25567 }, { "epoch": 0.9109143702014714, "grad_norm": 1.5591273007591144, "learning_rate": 4.1323771694442796e-07, "loss": 0.4319, "step": 25568 }, { "epoch": 0.9109499973279656, "grad_norm": 1.3439057134946129, "learning_rate": 4.129094865140382e-07, "loss": 0.3932, "step": 25569 }, { "epoch": 0.9109856244544596, "grad_norm": 1.108148822631067, "learning_rate": 4.125813837424597e-07, "loss": 0.3475, "step": 25570 }, { "epoch": 0.9110212515809537, "grad_norm": 1.9166185690755877, "learning_rate": 4.1225340863406126e-07, "loss": 0.5303, "step": 25571 }, { "epoch": 0.9110568787074479, "grad_norm": 1.3415368156636234, "learning_rate": 4.1192556119320827e-07, "loss": 0.4404, "step": 25572 }, { "epoch": 0.911092505833942, "grad_norm": 1.3788146406948667, "learning_rate": 4.115978414242694e-07, "loss": 0.3732, "step": 25573 }, { "epoch": 0.911128132960436, "grad_norm": 1.7337203886904684, "learning_rate": 4.1127024933160785e-07, "loss": 0.5214, "step": 25574 }, { "epoch": 0.9111637600869302, "grad_norm": 1.629605015667337, "learning_rate": 4.109427849195846e-07, "loss": 0.4944, "step": 25575 }, { "epoch": 0.9111993872134243, "grad_norm": 1.4192037306983651, "learning_rate": 4.106154481925606e-07, "loss": 0.4955, "step": 25576 }, { "epoch": 0.9112350143399184, "grad_norm": 1.7017901227488217, "learning_rate": 4.1028823915489344e-07, "loss": 0.4937, "step": 25577 }, { "epoch": 0.9112706414664126, "grad_norm": 2.035236359815326, "learning_rate": 4.0996115781094306e-07, "loss": 0.5008, "step": 25578 }, { "epoch": 0.9113062685929066, "grad_norm": 1.1381108276578518, "learning_rate": 4.0963420416506136e-07, "loss": 0.3823, "step": 25579 }, { "epoch": 0.9113418957194007, "grad_norm": 1.404911182501558, "learning_rate": 4.0930737822160285e-07, "loss": 0.4939, "step": 25580 }, { "epoch": 0.9113775228458949, "grad_norm": 2.046042960880395, "learning_rate": 4.0898067998492163e-07, "loss": 0.6305, "step": 25581 }, { "epoch": 0.911413149972389, "grad_norm": 1.6308279783906605, "learning_rate": 4.086541094593666e-07, "loss": 0.5268, "step": 25582 }, { "epoch": 0.9114487770988831, "grad_norm": 3.303074320278533, "learning_rate": 4.083276666492841e-07, "loss": 0.4989, "step": 25583 }, { "epoch": 0.9114844042253772, "grad_norm": 1.7861026458330251, "learning_rate": 4.080013515590253e-07, "loss": 0.4071, "step": 25584 }, { "epoch": 0.9115200313518713, "grad_norm": 1.5430650185342678, "learning_rate": 4.076751641929322e-07, "loss": 0.3949, "step": 25585 }, { "epoch": 0.9115556584783654, "grad_norm": 1.8415551465577384, "learning_rate": 4.073491045553479e-07, "loss": 0.5654, "step": 25586 }, { "epoch": 0.9115912856048596, "grad_norm": 1.573842996843854, "learning_rate": 4.0702317265061466e-07, "loss": 0.5275, "step": 25587 }, { "epoch": 0.9116269127313537, "grad_norm": 1.4755580496085126, "learning_rate": 4.066973684830744e-07, "loss": 0.5351, "step": 25588 }, { "epoch": 0.9116625398578477, "grad_norm": 1.818936078671388, "learning_rate": 4.063716920570637e-07, "loss": 0.5771, "step": 25589 }, { "epoch": 0.9116981669843419, "grad_norm": 1.2282304302588891, "learning_rate": 4.060461433769192e-07, "loss": 0.4454, "step": 25590 }, { "epoch": 0.911733794110836, "grad_norm": 1.206503872066195, "learning_rate": 4.0572072244697726e-07, "loss": 0.3738, "step": 25591 }, { "epoch": 0.9117694212373301, "grad_norm": 1.1861778341954914, "learning_rate": 4.0539542927156893e-07, "loss": 0.3868, "step": 25592 }, { "epoch": 0.9118050483638243, "grad_norm": 1.5746477598514175, "learning_rate": 4.0507026385502747e-07, "loss": 0.5481, "step": 25593 }, { "epoch": 0.9118406754903183, "grad_norm": 1.569555684573907, "learning_rate": 4.0474522620168047e-07, "loss": 0.5127, "step": 25594 }, { "epoch": 0.9118763026168124, "grad_norm": 1.6878974119512529, "learning_rate": 4.0442031631585774e-07, "loss": 0.5342, "step": 25595 }, { "epoch": 0.9119119297433066, "grad_norm": 1.2521998683066644, "learning_rate": 4.040955342018849e-07, "loss": 0.356, "step": 25596 }, { "epoch": 0.9119475568698007, "grad_norm": 1.2433828271709724, "learning_rate": 4.0377087986408937e-07, "loss": 0.3824, "step": 25597 }, { "epoch": 0.9119831839962947, "grad_norm": 1.824365425273221, "learning_rate": 4.034463533067901e-07, "loss": 0.5671, "step": 25598 }, { "epoch": 0.9120188111227889, "grad_norm": 1.8668698533389771, "learning_rate": 4.0312195453431015e-07, "loss": 0.5099, "step": 25599 }, { "epoch": 0.912054438249283, "grad_norm": 1.427314581107843, "learning_rate": 4.0279768355097057e-07, "loss": 0.4399, "step": 25600 }, { "epoch": 0.9120900653757771, "grad_norm": 1.7772160273992423, "learning_rate": 4.0247354036108687e-07, "loss": 0.467, "step": 25601 }, { "epoch": 0.9121256925022713, "grad_norm": 1.356815355115047, "learning_rate": 4.0214952496897663e-07, "loss": 0.4119, "step": 25602 }, { "epoch": 0.9121613196287653, "grad_norm": 1.332981708064535, "learning_rate": 4.0182563737895306e-07, "loss": 0.4304, "step": 25603 }, { "epoch": 0.9121969467552594, "grad_norm": 1.1353734040504095, "learning_rate": 4.015018775953317e-07, "loss": 0.3415, "step": 25604 }, { "epoch": 0.9122325738817536, "grad_norm": 1.1895688936926365, "learning_rate": 4.0117824562242e-07, "loss": 0.3676, "step": 25605 }, { "epoch": 0.9122682010082477, "grad_norm": 1.3354713574332246, "learning_rate": 4.0085474146453027e-07, "loss": 0.4523, "step": 25606 }, { "epoch": 0.9123038281347418, "grad_norm": 1.300956761536, "learning_rate": 4.00531365125969e-07, "loss": 0.413, "step": 25607 }, { "epoch": 0.9123394552612359, "grad_norm": 1.2780904859676845, "learning_rate": 4.0020811661104163e-07, "loss": 0.4039, "step": 25608 }, { "epoch": 0.91237508238773, "grad_norm": 1.5481064262158388, "learning_rate": 3.998849959240525e-07, "loss": 0.4831, "step": 25609 }, { "epoch": 0.9124107095142241, "grad_norm": 1.5818508957806872, "learning_rate": 3.995620030693048e-07, "loss": 0.5048, "step": 25610 }, { "epoch": 0.9124463366407183, "grad_norm": 1.4620506514215628, "learning_rate": 3.992391380511007e-07, "loss": 0.4189, "step": 25611 }, { "epoch": 0.9124819637672124, "grad_norm": 1.1981289332868965, "learning_rate": 3.989164008737367e-07, "loss": 0.4146, "step": 25612 }, { "epoch": 0.9125175908937064, "grad_norm": 1.6826807427425843, "learning_rate": 3.985937915415128e-07, "loss": 0.4745, "step": 25613 }, { "epoch": 0.9125532180202006, "grad_norm": 1.51482978452562, "learning_rate": 3.982713100587232e-07, "loss": 0.4721, "step": 25614 }, { "epoch": 0.9125888451466947, "grad_norm": 1.3978703360098432, "learning_rate": 3.9794895642966234e-07, "loss": 0.4289, "step": 25615 }, { "epoch": 0.9126244722731888, "grad_norm": 1.5983122458735919, "learning_rate": 3.976267306586223e-07, "loss": 0.4843, "step": 25616 }, { "epoch": 0.912660099399683, "grad_norm": 1.3747897719701723, "learning_rate": 3.9730463274989305e-07, "loss": 0.3943, "step": 25617 }, { "epoch": 0.912695726526177, "grad_norm": 1.7795605726108834, "learning_rate": 3.9698266270776556e-07, "loss": 0.5071, "step": 25618 }, { "epoch": 0.9127313536526711, "grad_norm": 1.7509356930683653, "learning_rate": 3.966608205365263e-07, "loss": 0.4323, "step": 25619 }, { "epoch": 0.9127669807791653, "grad_norm": 1.4127808196716964, "learning_rate": 3.963391062404598e-07, "loss": 0.361, "step": 25620 }, { "epoch": 0.9128026079056594, "grad_norm": 1.5816439492959524, "learning_rate": 3.960175198238525e-07, "loss": 0.4108, "step": 25621 }, { "epoch": 0.9128382350321534, "grad_norm": 1.4390199859347594, "learning_rate": 3.9569606129098216e-07, "loss": 0.4691, "step": 25622 }, { "epoch": 0.9128738621586476, "grad_norm": 1.286440991882476, "learning_rate": 3.9537473064613307e-07, "loss": 0.4532, "step": 25623 }, { "epoch": 0.9129094892851417, "grad_norm": 1.4459589741434358, "learning_rate": 3.9505352789358187e-07, "loss": 0.5724, "step": 25624 }, { "epoch": 0.9129451164116358, "grad_norm": 1.6739764985629975, "learning_rate": 3.947324530376062e-07, "loss": 0.567, "step": 25625 }, { "epoch": 0.91298074353813, "grad_norm": 1.807520058225435, "learning_rate": 3.944115060824827e-07, "loss": 0.5274, "step": 25626 }, { "epoch": 0.913016370664624, "grad_norm": 1.4146363227836458, "learning_rate": 3.9409068703248346e-07, "loss": 0.5519, "step": 25627 }, { "epoch": 0.9130519977911181, "grad_norm": 1.5046928254525629, "learning_rate": 3.937699958918817e-07, "loss": 0.4476, "step": 25628 }, { "epoch": 0.9130876249176123, "grad_norm": 1.312570127155109, "learning_rate": 3.934494326649463e-07, "loss": 0.4155, "step": 25629 }, { "epoch": 0.9131232520441064, "grad_norm": 1.6305028561677146, "learning_rate": 3.93128997355946e-07, "loss": 0.4922, "step": 25630 }, { "epoch": 0.9131588791706005, "grad_norm": 2.2250518046909957, "learning_rate": 3.9280868996914745e-07, "loss": 0.3979, "step": 25631 }, { "epoch": 0.9131945062970946, "grad_norm": 1.1936856913034255, "learning_rate": 3.924885105088172e-07, "loss": 0.3938, "step": 25632 }, { "epoch": 0.9132301334235887, "grad_norm": 1.3379912343985099, "learning_rate": 3.921684589792163e-07, "loss": 0.4034, "step": 25633 }, { "epoch": 0.9132657605500828, "grad_norm": 1.2180650222818368, "learning_rate": 3.91848535384608e-07, "loss": 0.4082, "step": 25634 }, { "epoch": 0.913301387676577, "grad_norm": 1.288918709817101, "learning_rate": 3.9152873972925444e-07, "loss": 0.3776, "step": 25635 }, { "epoch": 0.9133370148030711, "grad_norm": 1.308559237867223, "learning_rate": 3.912090720174089e-07, "loss": 0.4072, "step": 25636 }, { "epoch": 0.9133726419295651, "grad_norm": 1.6450298718851755, "learning_rate": 3.908895322533335e-07, "loss": 0.5153, "step": 25637 }, { "epoch": 0.9134082690560593, "grad_norm": 1.2144677079929251, "learning_rate": 3.9057012044127817e-07, "loss": 0.3658, "step": 25638 }, { "epoch": 0.9134438961825534, "grad_norm": 2.2082678566952163, "learning_rate": 3.9025083658549845e-07, "loss": 0.5156, "step": 25639 }, { "epoch": 0.9134795233090475, "grad_norm": 1.5377411820494742, "learning_rate": 3.8993168069024535e-07, "loss": 0.4811, "step": 25640 }, { "epoch": 0.9135151504355417, "grad_norm": 1.42182637961998, "learning_rate": 3.896126527597699e-07, "loss": 0.4077, "step": 25641 }, { "epoch": 0.9135507775620357, "grad_norm": 1.3144199293109047, "learning_rate": 3.892937527983187e-07, "loss": 0.445, "step": 25642 }, { "epoch": 0.9135864046885298, "grad_norm": 1.7162382969542775, "learning_rate": 3.889749808101395e-07, "loss": 0.4848, "step": 25643 }, { "epoch": 0.913622031815024, "grad_norm": 1.6446437020986042, "learning_rate": 3.886563367994756e-07, "loss": 0.4513, "step": 25644 }, { "epoch": 0.9136576589415181, "grad_norm": 1.2731704477680512, "learning_rate": 3.8833782077057015e-07, "loss": 0.3853, "step": 25645 }, { "epoch": 0.9136932860680121, "grad_norm": 1.2566132953431843, "learning_rate": 3.880194327276654e-07, "loss": 0.4069, "step": 25646 }, { "epoch": 0.9137289131945063, "grad_norm": 1.5655161420908421, "learning_rate": 3.8770117267499905e-07, "loss": 0.4339, "step": 25647 }, { "epoch": 0.9137645403210004, "grad_norm": 1.33668647709869, "learning_rate": 3.8738304061681107e-07, "loss": 0.4201, "step": 25648 }, { "epoch": 0.9138001674474945, "grad_norm": 1.2508512342827793, "learning_rate": 3.8706503655733696e-07, "loss": 0.392, "step": 25649 }, { "epoch": 0.9138357945739887, "grad_norm": 1.3066525238427893, "learning_rate": 3.8674716050081105e-07, "loss": 0.4053, "step": 25650 }, { "epoch": 0.9138714217004827, "grad_norm": 1.1858326207341001, "learning_rate": 3.864294124514656e-07, "loss": 0.3328, "step": 25651 }, { "epoch": 0.9139070488269768, "grad_norm": 1.30790898577746, "learning_rate": 3.861117924135327e-07, "loss": 0.3488, "step": 25652 }, { "epoch": 0.913942675953471, "grad_norm": 1.519440906593778, "learning_rate": 3.857943003912401e-07, "loss": 0.5756, "step": 25653 }, { "epoch": 0.9139783030799651, "grad_norm": 1.4170682526258318, "learning_rate": 3.854769363888156e-07, "loss": 0.3694, "step": 25654 }, { "epoch": 0.9140139302064592, "grad_norm": 1.7189541363249423, "learning_rate": 3.851597004104868e-07, "loss": 0.573, "step": 25655 }, { "epoch": 0.9140495573329533, "grad_norm": 2.073799769963916, "learning_rate": 3.8484259246047705e-07, "loss": 0.4861, "step": 25656 }, { "epoch": 0.9140851844594474, "grad_norm": 1.628997540692834, "learning_rate": 3.8452561254300856e-07, "loss": 0.4812, "step": 25657 }, { "epoch": 0.9141208115859415, "grad_norm": 1.3701620128412384, "learning_rate": 3.842087606623024e-07, "loss": 0.425, "step": 25658 }, { "epoch": 0.9141564387124357, "grad_norm": 1.3262132566260112, "learning_rate": 3.838920368225785e-07, "loss": 0.4097, "step": 25659 }, { "epoch": 0.9141920658389298, "grad_norm": 1.4532809535722522, "learning_rate": 3.8357544102805233e-07, "loss": 0.5235, "step": 25660 }, { "epoch": 0.9142276929654238, "grad_norm": 1.5080112182740983, "learning_rate": 3.832589732829406e-07, "loss": 0.437, "step": 25661 }, { "epoch": 0.914263320091918, "grad_norm": 1.7993360725520517, "learning_rate": 3.8294263359145657e-07, "loss": 0.4863, "step": 25662 }, { "epoch": 0.9142989472184121, "grad_norm": 1.3093033442298152, "learning_rate": 3.826264219578157e-07, "loss": 0.3802, "step": 25663 }, { "epoch": 0.9143345743449062, "grad_norm": 1.1864399040554188, "learning_rate": 3.823103383862248e-07, "loss": 0.4268, "step": 25664 }, { "epoch": 0.9143702014714004, "grad_norm": 1.5717363776659579, "learning_rate": 3.8199438288089474e-07, "loss": 0.5235, "step": 25665 }, { "epoch": 0.9144058285978944, "grad_norm": 1.842313326806974, "learning_rate": 3.8167855544603115e-07, "loss": 0.4844, "step": 25666 }, { "epoch": 0.9144414557243885, "grad_norm": 1.448292216104316, "learning_rate": 3.8136285608584065e-07, "loss": 0.5762, "step": 25667 }, { "epoch": 0.9144770828508827, "grad_norm": 1.2938828543415866, "learning_rate": 3.810472848045266e-07, "loss": 0.4053, "step": 25668 }, { "epoch": 0.9145127099773768, "grad_norm": 1.2136577385141856, "learning_rate": 3.8073184160629105e-07, "loss": 0.3548, "step": 25669 }, { "epoch": 0.9145483371038708, "grad_norm": 1.4526932949662017, "learning_rate": 3.8041652649533413e-07, "loss": 0.4418, "step": 25670 }, { "epoch": 0.914583964230365, "grad_norm": 1.7329950090079793, "learning_rate": 3.801013394758557e-07, "loss": 0.5216, "step": 25671 }, { "epoch": 0.9146195913568591, "grad_norm": 1.338484080044863, "learning_rate": 3.7978628055205245e-07, "loss": 0.4015, "step": 25672 }, { "epoch": 0.9146552184833532, "grad_norm": 1.2791116067163826, "learning_rate": 3.7947134972811775e-07, "loss": 0.36, "step": 25673 }, { "epoch": 0.9146908456098474, "grad_norm": 1.4286441831175178, "learning_rate": 3.791565470082481e-07, "loss": 0.4414, "step": 25674 }, { "epoch": 0.9147264727363414, "grad_norm": 1.1641438787735345, "learning_rate": 3.7884187239663137e-07, "loss": 0.4289, "step": 25675 }, { "epoch": 0.9147620998628355, "grad_norm": 1.8160994377881932, "learning_rate": 3.7852732589746087e-07, "loss": 0.5164, "step": 25676 }, { "epoch": 0.9147977269893297, "grad_norm": 1.618272219018775, "learning_rate": 3.7821290751492325e-07, "loss": 0.4974, "step": 25677 }, { "epoch": 0.9148333541158238, "grad_norm": 1.7944813254723917, "learning_rate": 3.778986172532073e-07, "loss": 0.5874, "step": 25678 }, { "epoch": 0.9148689812423179, "grad_norm": 1.2871203497224804, "learning_rate": 3.7758445511649645e-07, "loss": 0.409, "step": 25679 }, { "epoch": 0.914904608368812, "grad_norm": 1.8117752486170775, "learning_rate": 3.7727042110897283e-07, "loss": 0.5263, "step": 25680 }, { "epoch": 0.9149402354953061, "grad_norm": 1.289143526747492, "learning_rate": 3.76956515234822e-07, "loss": 0.4476, "step": 25681 }, { "epoch": 0.9149758626218002, "grad_norm": 1.4226338327861798, "learning_rate": 3.766427374982196e-07, "loss": 0.378, "step": 25682 }, { "epoch": 0.9150114897482944, "grad_norm": 1.250580415512432, "learning_rate": 3.7632908790334656e-07, "loss": 0.3821, "step": 25683 }, { "epoch": 0.9150471168747885, "grad_norm": 1.2045670682302339, "learning_rate": 3.7601556645437633e-07, "loss": 0.4188, "step": 25684 }, { "epoch": 0.9150827440012825, "grad_norm": 1.2770292901374505, "learning_rate": 3.757021731554866e-07, "loss": 0.4219, "step": 25685 }, { "epoch": 0.9151183711277767, "grad_norm": 1.8794553045222844, "learning_rate": 3.753889080108497e-07, "loss": 0.5287, "step": 25686 }, { "epoch": 0.9151539982542708, "grad_norm": 1.3439386683341992, "learning_rate": 3.7507577102463665e-07, "loss": 0.3804, "step": 25687 }, { "epoch": 0.9151896253807649, "grad_norm": 1.5173172316960943, "learning_rate": 3.747627622010164e-07, "loss": 0.499, "step": 25688 }, { "epoch": 0.9152252525072591, "grad_norm": 3.0755939258689664, "learning_rate": 3.7444988154415997e-07, "loss": 0.4782, "step": 25689 }, { "epoch": 0.9152608796337531, "grad_norm": 1.5932481258042048, "learning_rate": 3.7413712905822854e-07, "loss": 0.5328, "step": 25690 }, { "epoch": 0.9152965067602472, "grad_norm": 1.7963460865869167, "learning_rate": 3.7382450474738987e-07, "loss": 0.414, "step": 25691 }, { "epoch": 0.9153321338867414, "grad_norm": 1.3522809744460291, "learning_rate": 3.735120086158062e-07, "loss": 0.3907, "step": 25692 }, { "epoch": 0.9153677610132355, "grad_norm": 1.3024632914084135, "learning_rate": 3.7319964066763857e-07, "loss": 0.4217, "step": 25693 }, { "epoch": 0.9154033881397295, "grad_norm": 1.1729354572332316, "learning_rate": 3.7288740090704824e-07, "loss": 0.3813, "step": 25694 }, { "epoch": 0.9154390152662237, "grad_norm": 1.2296568872193672, "learning_rate": 3.725752893381884e-07, "loss": 0.3818, "step": 25695 }, { "epoch": 0.9154746423927178, "grad_norm": 1.7025838908221742, "learning_rate": 3.722633059652203e-07, "loss": 0.4393, "step": 25696 }, { "epoch": 0.9155102695192119, "grad_norm": 1.4988432583458786, "learning_rate": 3.719514507922939e-07, "loss": 0.4723, "step": 25697 }, { "epoch": 0.9155458966457061, "grad_norm": 1.958515340666172, "learning_rate": 3.7163972382356364e-07, "loss": 0.5263, "step": 25698 }, { "epoch": 0.9155815237722001, "grad_norm": 1.6099636463917835, "learning_rate": 3.713281250631806e-07, "loss": 0.4756, "step": 25699 }, { "epoch": 0.9156171508986942, "grad_norm": 1.6808871134156664, "learning_rate": 3.710166545152949e-07, "loss": 0.4831, "step": 25700 }, { "epoch": 0.9156527780251884, "grad_norm": 1.286992745614722, "learning_rate": 3.7070531218405095e-07, "loss": 0.4168, "step": 25701 }, { "epoch": 0.9156884051516825, "grad_norm": 1.496488341258018, "learning_rate": 3.703940980735976e-07, "loss": 0.5108, "step": 25702 }, { "epoch": 0.9157240322781766, "grad_norm": 1.4092538893660718, "learning_rate": 3.7008301218807716e-07, "loss": 0.4569, "step": 25703 }, { "epoch": 0.9157596594046707, "grad_norm": 2.3832089641402687, "learning_rate": 3.697720545316308e-07, "loss": 0.5392, "step": 25704 }, { "epoch": 0.9157952865311648, "grad_norm": 1.5145617014822421, "learning_rate": 3.694612251084029e-07, "loss": 0.4327, "step": 25705 }, { "epoch": 0.9158309136576589, "grad_norm": 1.5707759791652827, "learning_rate": 3.6915052392252903e-07, "loss": 0.5059, "step": 25706 }, { "epoch": 0.9158665407841531, "grad_norm": 1.3947016193768764, "learning_rate": 3.6883995097814707e-07, "loss": 0.4415, "step": 25707 }, { "epoch": 0.9159021679106472, "grad_norm": 1.5302001309921134, "learning_rate": 3.6852950627939253e-07, "loss": 0.5069, "step": 25708 }, { "epoch": 0.9159377950371412, "grad_norm": 1.1371559255531203, "learning_rate": 3.6821918983040105e-07, "loss": 0.3572, "step": 25709 }, { "epoch": 0.9159734221636354, "grad_norm": 1.3945950206716673, "learning_rate": 3.6790900163530266e-07, "loss": 0.3678, "step": 25710 }, { "epoch": 0.9160090492901295, "grad_norm": 1.3889284225213543, "learning_rate": 3.6759894169822954e-07, "loss": 0.3632, "step": 25711 }, { "epoch": 0.9160446764166236, "grad_norm": 1.299937687791465, "learning_rate": 3.672890100233073e-07, "loss": 0.438, "step": 25712 }, { "epoch": 0.9160803035431178, "grad_norm": 1.4695466866744067, "learning_rate": 3.669792066146649e-07, "loss": 0.4834, "step": 25713 }, { "epoch": 0.9161159306696118, "grad_norm": 1.3995671527212983, "learning_rate": 3.666695314764279e-07, "loss": 0.3781, "step": 25714 }, { "epoch": 0.9161515577961059, "grad_norm": 1.9573494098387776, "learning_rate": 3.663599846127186e-07, "loss": 0.4733, "step": 25715 }, { "epoch": 0.9161871849226001, "grad_norm": 1.3063713448470864, "learning_rate": 3.660505660276614e-07, "loss": 0.3928, "step": 25716 }, { "epoch": 0.9162228120490942, "grad_norm": 1.502862464154037, "learning_rate": 3.657412757253731e-07, "loss": 0.3895, "step": 25717 }, { "epoch": 0.9162584391755882, "grad_norm": 1.4804606266229445, "learning_rate": 3.6543211370997587e-07, "loss": 0.4771, "step": 25718 }, { "epoch": 0.9162940663020824, "grad_norm": 2.5759052376359457, "learning_rate": 3.6512307998558205e-07, "loss": 0.5382, "step": 25719 }, { "epoch": 0.9163296934285765, "grad_norm": 1.6483661508094558, "learning_rate": 3.6481417455631054e-07, "loss": 0.4288, "step": 25720 }, { "epoch": 0.9163653205550706, "grad_norm": 1.4795746822088156, "learning_rate": 3.6450539742627135e-07, "loss": 0.4445, "step": 25721 }, { "epoch": 0.9164009476815648, "grad_norm": 1.0889260618616008, "learning_rate": 3.6419674859957677e-07, "loss": 0.4131, "step": 25722 }, { "epoch": 0.9164365748080588, "grad_norm": 1.2481489544871809, "learning_rate": 3.6388822808033905e-07, "loss": 0.4006, "step": 25723 }, { "epoch": 0.9164722019345529, "grad_norm": 1.2061366892561824, "learning_rate": 3.6357983587266497e-07, "loss": 0.3939, "step": 25724 }, { "epoch": 0.9165078290610471, "grad_norm": 1.3833168049966473, "learning_rate": 3.632715719806601e-07, "loss": 0.4095, "step": 25725 }, { "epoch": 0.9165434561875412, "grad_norm": 1.333499358067132, "learning_rate": 3.6296343640843e-07, "loss": 0.3983, "step": 25726 }, { "epoch": 0.9165790833140353, "grad_norm": 1.4291313065095488, "learning_rate": 3.6265542916007814e-07, "loss": 0.4516, "step": 25727 }, { "epoch": 0.9166147104405294, "grad_norm": 1.687770318273225, "learning_rate": 3.6234755023970447e-07, "loss": 0.4071, "step": 25728 }, { "epoch": 0.9166503375670235, "grad_norm": 1.4259346896659144, "learning_rate": 3.6203979965141025e-07, "loss": 0.4001, "step": 25729 }, { "epoch": 0.9166859646935176, "grad_norm": 1.7367218066026957, "learning_rate": 3.61732177399291e-07, "loss": 0.5625, "step": 25730 }, { "epoch": 0.9167215918200118, "grad_norm": 1.8882213592827788, "learning_rate": 3.614246834874469e-07, "loss": 0.4447, "step": 25731 }, { "epoch": 0.9167572189465059, "grad_norm": 1.7707700451638786, "learning_rate": 3.6111731791996894e-07, "loss": 0.5568, "step": 25732 }, { "epoch": 0.9167928460729999, "grad_norm": 1.1122498917635515, "learning_rate": 3.608100807009518e-07, "loss": 0.3771, "step": 25733 }, { "epoch": 0.9168284731994941, "grad_norm": 1.197168429089869, "learning_rate": 3.6050297183448543e-07, "loss": 0.4059, "step": 25734 }, { "epoch": 0.9168641003259882, "grad_norm": 1.2614347798982668, "learning_rate": 3.6019599132465997e-07, "loss": 0.4208, "step": 25735 }, { "epoch": 0.9168997274524823, "grad_norm": 1.2469659809203593, "learning_rate": 3.59889139175561e-07, "loss": 0.3752, "step": 25736 }, { "epoch": 0.9169353545789765, "grad_norm": 1.4498882323175748, "learning_rate": 3.5958241539127635e-07, "loss": 0.3731, "step": 25737 }, { "epoch": 0.9169709817054705, "grad_norm": 1.2758715850445093, "learning_rate": 3.5927581997589276e-07, "loss": 0.4307, "step": 25738 }, { "epoch": 0.9170066088319646, "grad_norm": 1.3658271851470178, "learning_rate": 3.589693529334881e-07, "loss": 0.3979, "step": 25739 }, { "epoch": 0.9170422359584588, "grad_norm": 1.2967555303269755, "learning_rate": 3.5866301426814573e-07, "loss": 0.4159, "step": 25740 }, { "epoch": 0.9170778630849529, "grad_norm": 1.4061260705179077, "learning_rate": 3.583568039839436e-07, "loss": 0.4149, "step": 25741 }, { "epoch": 0.917113490211447, "grad_norm": 1.1812607286886834, "learning_rate": 3.580507220849616e-07, "loss": 0.3529, "step": 25742 }, { "epoch": 0.9171491173379411, "grad_norm": 1.4134822070975255, "learning_rate": 3.5774476857527107e-07, "loss": 0.4734, "step": 25743 }, { "epoch": 0.9171847444644352, "grad_norm": 1.3539177060798002, "learning_rate": 3.574389434589487e-07, "loss": 0.4241, "step": 25744 }, { "epoch": 0.9172203715909293, "grad_norm": 1.6116002669132254, "learning_rate": 3.5713324674006787e-07, "loss": 0.3769, "step": 25745 }, { "epoch": 0.9172559987174235, "grad_norm": 1.2623620150549792, "learning_rate": 3.568276784226976e-07, "loss": 0.4151, "step": 25746 }, { "epoch": 0.9172916258439175, "grad_norm": 1.4908765192813296, "learning_rate": 3.5652223851090686e-07, "loss": 0.4714, "step": 25747 }, { "epoch": 0.9173272529704116, "grad_norm": 1.6528035800861214, "learning_rate": 3.562169270087634e-07, "loss": 0.5149, "step": 25748 }, { "epoch": 0.9173628800969058, "grad_norm": 1.4530604579468203, "learning_rate": 3.5591174392033077e-07, "loss": 0.4762, "step": 25749 }, { "epoch": 0.9173985072233999, "grad_norm": 1.294083520613841, "learning_rate": 3.5560668924967455e-07, "loss": 0.5095, "step": 25750 }, { "epoch": 0.917434134349894, "grad_norm": 1.2258922437406645, "learning_rate": 3.5530176300085593e-07, "loss": 0.3981, "step": 25751 }, { "epoch": 0.9174697614763881, "grad_norm": 1.3518235979727615, "learning_rate": 3.5499696517793615e-07, "loss": 0.4013, "step": 25752 }, { "epoch": 0.9175053886028822, "grad_norm": 2.9859754901555036, "learning_rate": 3.546922957849741e-07, "loss": 0.509, "step": 25753 }, { "epoch": 0.9175410157293763, "grad_norm": 1.6631106508594797, "learning_rate": 3.543877548260244e-07, "loss": 0.4754, "step": 25754 }, { "epoch": 0.9175766428558705, "grad_norm": 1.2174253831343356, "learning_rate": 3.5408334230514484e-07, "loss": 0.4017, "step": 25755 }, { "epoch": 0.9176122699823646, "grad_norm": 1.3551901515459703, "learning_rate": 3.537790582263867e-07, "loss": 0.3558, "step": 25756 }, { "epoch": 0.9176478971088586, "grad_norm": 1.2781144190286724, "learning_rate": 3.5347490259380334e-07, "loss": 0.4571, "step": 25757 }, { "epoch": 0.9176835242353528, "grad_norm": 1.3027546603393094, "learning_rate": 3.531708754114438e-07, "loss": 0.4116, "step": 25758 }, { "epoch": 0.9177191513618469, "grad_norm": 1.3124184232725122, "learning_rate": 3.528669766833559e-07, "loss": 0.3842, "step": 25759 }, { "epoch": 0.917754778488341, "grad_norm": 1.3161134605084128, "learning_rate": 3.5256320641358756e-07, "loss": 0.4202, "step": 25760 }, { "epoch": 0.9177904056148352, "grad_norm": 2.093651161811943, "learning_rate": 3.522595646061844e-07, "loss": 0.4262, "step": 25761 }, { "epoch": 0.9178260327413292, "grad_norm": 1.490506117775274, "learning_rate": 3.519560512651887e-07, "loss": 0.3963, "step": 25762 }, { "epoch": 0.9178616598678233, "grad_norm": 1.4408282214384402, "learning_rate": 3.5165266639463957e-07, "loss": 0.4246, "step": 25763 }, { "epoch": 0.9178972869943175, "grad_norm": 1.1935332785724686, "learning_rate": 3.513494099985815e-07, "loss": 0.4137, "step": 25764 }, { "epoch": 0.9179329141208116, "grad_norm": 1.3625038394706956, "learning_rate": 3.5104628208104895e-07, "loss": 0.3713, "step": 25765 }, { "epoch": 0.9179685412473056, "grad_norm": 1.8734461023607205, "learning_rate": 3.507432826460788e-07, "loss": 0.5454, "step": 25766 }, { "epoch": 0.9180041683737998, "grad_norm": 1.2041376795799936, "learning_rate": 3.504404116977067e-07, "loss": 0.3267, "step": 25767 }, { "epoch": 0.9180397955002939, "grad_norm": 1.4617170490911149, "learning_rate": 3.501376692399661e-07, "loss": 0.451, "step": 25768 }, { "epoch": 0.918075422626788, "grad_norm": 1.5415011839575585, "learning_rate": 3.498350552768859e-07, "loss": 0.5228, "step": 25769 }, { "epoch": 0.9181110497532822, "grad_norm": 1.768185062124364, "learning_rate": 3.495325698124985e-07, "loss": 0.5269, "step": 25770 }, { "epoch": 0.9181466768797762, "grad_norm": 1.4975665855825382, "learning_rate": 3.492302128508285e-07, "loss": 0.5052, "step": 25771 }, { "epoch": 0.9181823040062703, "grad_norm": 1.2120008395155069, "learning_rate": 3.489279843959037e-07, "loss": 0.3838, "step": 25772 }, { "epoch": 0.9182179311327645, "grad_norm": 1.355770572747098, "learning_rate": 3.4862588445174985e-07, "loss": 0.3725, "step": 25773 }, { "epoch": 0.9182535582592586, "grad_norm": 1.1613812857243604, "learning_rate": 3.48323913022387e-07, "loss": 0.4331, "step": 25774 }, { "epoch": 0.9182891853857527, "grad_norm": 1.337881647903485, "learning_rate": 3.4802207011183864e-07, "loss": 0.4082, "step": 25775 }, { "epoch": 0.9183248125122468, "grad_norm": 1.4181753365227283, "learning_rate": 3.4772035572412154e-07, "loss": 0.4129, "step": 25776 }, { "epoch": 0.9183604396387409, "grad_norm": 1.199941845459326, "learning_rate": 3.4741876986325586e-07, "loss": 0.3727, "step": 25777 }, { "epoch": 0.918396066765235, "grad_norm": 1.765606890156701, "learning_rate": 3.47117312533255e-07, "loss": 0.503, "step": 25778 }, { "epoch": 0.9184316938917292, "grad_norm": 1.3160855190557557, "learning_rate": 3.4681598373813465e-07, "loss": 0.4005, "step": 25779 }, { "epoch": 0.9184673210182233, "grad_norm": 1.3198047970556122, "learning_rate": 3.4651478348190604e-07, "loss": 0.4211, "step": 25780 }, { "epoch": 0.9185029481447173, "grad_norm": 1.543366924282256, "learning_rate": 3.4621371176858043e-07, "loss": 0.5422, "step": 25781 }, { "epoch": 0.9185385752712115, "grad_norm": 1.605264994239046, "learning_rate": 3.459127686021657e-07, "loss": 0.5448, "step": 25782 }, { "epoch": 0.9185742023977056, "grad_norm": 5.6518618167120485, "learning_rate": 3.45611953986672e-07, "loss": 0.5991, "step": 25783 }, { "epoch": 0.9186098295241997, "grad_norm": 1.4032324582048343, "learning_rate": 3.453112679261028e-07, "loss": 0.3856, "step": 25784 }, { "epoch": 0.9186454566506939, "grad_norm": 1.5031657727574954, "learning_rate": 3.4501071042446263e-07, "loss": 0.5402, "step": 25785 }, { "epoch": 0.9186810837771879, "grad_norm": 1.2949588584806369, "learning_rate": 3.447102814857539e-07, "loss": 0.4826, "step": 25786 }, { "epoch": 0.918716710903682, "grad_norm": 1.164918571625459, "learning_rate": 3.444099811139745e-07, "loss": 0.427, "step": 25787 }, { "epoch": 0.9187523380301762, "grad_norm": 1.191434597766761, "learning_rate": 3.4410980931312566e-07, "loss": 0.3615, "step": 25788 }, { "epoch": 0.9187879651566703, "grad_norm": 2.3583631629362505, "learning_rate": 3.4380976608720416e-07, "loss": 0.5371, "step": 25789 }, { "epoch": 0.9188235922831643, "grad_norm": 1.4635655199318227, "learning_rate": 3.4350985144020575e-07, "loss": 0.4439, "step": 25790 }, { "epoch": 0.9188592194096585, "grad_norm": 1.3061164140651191, "learning_rate": 3.432100653761217e-07, "loss": 0.4611, "step": 25791 }, { "epoch": 0.9188948465361526, "grad_norm": 1.7701631742895036, "learning_rate": 3.4291040789894647e-07, "loss": 0.5303, "step": 25792 }, { "epoch": 0.9189304736626467, "grad_norm": 1.5298712276776043, "learning_rate": 3.426108790126681e-07, "loss": 0.4665, "step": 25793 }, { "epoch": 0.9189661007891409, "grad_norm": 1.31750427120419, "learning_rate": 3.423114787212767e-07, "loss": 0.3742, "step": 25794 }, { "epoch": 0.919001727915635, "grad_norm": 1.1758667589545686, "learning_rate": 3.420122070287579e-07, "loss": 0.398, "step": 25795 }, { "epoch": 0.919037355042129, "grad_norm": 1.5239014267710254, "learning_rate": 3.4171306393909643e-07, "loss": 0.5594, "step": 25796 }, { "epoch": 0.9190729821686232, "grad_norm": 1.076933087131094, "learning_rate": 3.4141404945627675e-07, "loss": 0.4158, "step": 25797 }, { "epoch": 0.9191086092951173, "grad_norm": 1.4045611395737159, "learning_rate": 3.411151635842802e-07, "loss": 0.4026, "step": 25798 }, { "epoch": 0.9191442364216114, "grad_norm": 1.4618446344506955, "learning_rate": 3.408164063270869e-07, "loss": 0.4017, "step": 25799 }, { "epoch": 0.9191798635481055, "grad_norm": 1.313292247261961, "learning_rate": 3.4051777768867364e-07, "loss": 0.5011, "step": 25800 }, { "epoch": 0.9192154906745996, "grad_norm": 1.8825912859778609, "learning_rate": 3.4021927767301843e-07, "loss": 0.4013, "step": 25801 }, { "epoch": 0.9192511178010937, "grad_norm": 1.5400656528328656, "learning_rate": 3.399209062840947e-07, "loss": 0.4587, "step": 25802 }, { "epoch": 0.9192867449275879, "grad_norm": 1.7118934098651675, "learning_rate": 3.396226635258759e-07, "loss": 0.5009, "step": 25803 }, { "epoch": 0.919322372054082, "grad_norm": 1.1599146949973247, "learning_rate": 3.393245494023345e-07, "loss": 0.4058, "step": 25804 }, { "epoch": 0.919357999180576, "grad_norm": 1.3165681866073926, "learning_rate": 3.390265639174384e-07, "loss": 0.4828, "step": 25805 }, { "epoch": 0.9193936263070702, "grad_norm": 1.2773303374054084, "learning_rate": 3.387287070751566e-07, "loss": 0.4114, "step": 25806 }, { "epoch": 0.9194292534335643, "grad_norm": 1.7114178023433853, "learning_rate": 3.384309788794549e-07, "loss": 0.519, "step": 25807 }, { "epoch": 0.9194648805600585, "grad_norm": 1.9915991979092558, "learning_rate": 3.381333793342978e-07, "loss": 0.5741, "step": 25808 }, { "epoch": 0.9195005076865526, "grad_norm": 2.0792026460874897, "learning_rate": 3.378359084436478e-07, "loss": 0.5218, "step": 25809 }, { "epoch": 0.9195361348130466, "grad_norm": 1.3423693599854025, "learning_rate": 3.375385662114661e-07, "loss": 0.3835, "step": 25810 }, { "epoch": 0.9195717619395408, "grad_norm": 1.7234560289411436, "learning_rate": 3.3724135264171063e-07, "loss": 0.4285, "step": 25811 }, { "epoch": 0.9196073890660349, "grad_norm": 1.2944109452612809, "learning_rate": 3.3694426773834276e-07, "loss": 0.4179, "step": 25812 }, { "epoch": 0.919643016192529, "grad_norm": 1.280339228527001, "learning_rate": 3.3664731150531484e-07, "loss": 0.3974, "step": 25813 }, { "epoch": 0.9196786433190232, "grad_norm": 1.1824195074718, "learning_rate": 3.3635048394658364e-07, "loss": 0.4414, "step": 25814 }, { "epoch": 0.9197142704455172, "grad_norm": 1.1763556888879119, "learning_rate": 3.3605378506609834e-07, "loss": 0.3891, "step": 25815 }, { "epoch": 0.9197498975720113, "grad_norm": 1.5597880688820407, "learning_rate": 3.357572148678134e-07, "loss": 0.3868, "step": 25816 }, { "epoch": 0.9197855246985055, "grad_norm": 1.3814461393425543, "learning_rate": 3.354607733556747e-07, "loss": 0.385, "step": 25817 }, { "epoch": 0.9198211518249996, "grad_norm": 1.3867589788222001, "learning_rate": 3.3516446053363015e-07, "loss": 0.4142, "step": 25818 }, { "epoch": 0.9198567789514936, "grad_norm": 1.719136355201504, "learning_rate": 3.348682764056277e-07, "loss": 0.502, "step": 25819 }, { "epoch": 0.9198924060779878, "grad_norm": 1.7296462131377164, "learning_rate": 3.345722209756086e-07, "loss": 0.5144, "step": 25820 }, { "epoch": 0.9199280332044819, "grad_norm": 1.5603642990599347, "learning_rate": 3.342762942475175e-07, "loss": 0.5057, "step": 25821 }, { "epoch": 0.919963660330976, "grad_norm": 1.3248689837413046, "learning_rate": 3.3398049622529237e-07, "loss": 0.4225, "step": 25822 }, { "epoch": 0.9199992874574702, "grad_norm": 1.465964583042622, "learning_rate": 3.336848269128734e-07, "loss": 0.5077, "step": 25823 }, { "epoch": 0.9200349145839642, "grad_norm": 1.453002070780707, "learning_rate": 3.3338928631419745e-07, "loss": 0.3786, "step": 25824 }, { "epoch": 0.9200705417104583, "grad_norm": 1.519795208151597, "learning_rate": 3.3309387443319906e-07, "loss": 0.5303, "step": 25825 }, { "epoch": 0.9201061688369525, "grad_norm": 1.3138255911483792, "learning_rate": 3.3279859127381297e-07, "loss": 0.4021, "step": 25826 }, { "epoch": 0.9201417959634466, "grad_norm": 1.593974906955409, "learning_rate": 3.3250343683997046e-07, "loss": 0.5063, "step": 25827 }, { "epoch": 0.9201774230899407, "grad_norm": 1.4098739005886751, "learning_rate": 3.3220841113560165e-07, "loss": 0.4868, "step": 25828 }, { "epoch": 0.9202130502164348, "grad_norm": 1.1328268963315555, "learning_rate": 3.319135141646368e-07, "loss": 0.4148, "step": 25829 }, { "epoch": 0.9202486773429289, "grad_norm": 1.2325797548388508, "learning_rate": 3.3161874593099943e-07, "loss": 0.3503, "step": 25830 }, { "epoch": 0.920284304469423, "grad_norm": 1.7037157147207573, "learning_rate": 3.3132410643861637e-07, "loss": 0.5398, "step": 25831 }, { "epoch": 0.9203199315959172, "grad_norm": 1.2287257005765992, "learning_rate": 3.310295956914122e-07, "loss": 0.3767, "step": 25832 }, { "epoch": 0.9203555587224113, "grad_norm": 1.172593687152576, "learning_rate": 3.3073521369330616e-07, "loss": 0.4085, "step": 25833 }, { "epoch": 0.9203911858489053, "grad_norm": 1.4060002706606993, "learning_rate": 3.3044096044821836e-07, "loss": 0.4224, "step": 25834 }, { "epoch": 0.9204268129753995, "grad_norm": 1.2779845687357152, "learning_rate": 3.301468359600679e-07, "loss": 0.4354, "step": 25835 }, { "epoch": 0.9204624401018936, "grad_norm": 1.9945668571932602, "learning_rate": 3.2985284023277274e-07, "loss": 0.4736, "step": 25836 }, { "epoch": 0.9204980672283877, "grad_norm": 1.9545596311136701, "learning_rate": 3.295589732702453e-07, "loss": 0.4934, "step": 25837 }, { "epoch": 0.9205336943548819, "grad_norm": 1.3285979019341383, "learning_rate": 3.292652350763992e-07, "loss": 0.415, "step": 25838 }, { "epoch": 0.9205693214813759, "grad_norm": 1.7035585121728292, "learning_rate": 3.289716256551456e-07, "loss": 0.5165, "step": 25839 }, { "epoch": 0.92060494860787, "grad_norm": 1.5003405323627292, "learning_rate": 3.2867814501039484e-07, "loss": 0.588, "step": 25840 }, { "epoch": 0.9206405757343642, "grad_norm": 1.5336082587504833, "learning_rate": 3.2838479314605373e-07, "loss": 0.3724, "step": 25841 }, { "epoch": 0.9206762028608583, "grad_norm": 1.2347568274247047, "learning_rate": 3.280915700660292e-07, "loss": 0.3932, "step": 25842 }, { "epoch": 0.9207118299873523, "grad_norm": 1.2402804098451954, "learning_rate": 3.2779847577422696e-07, "loss": 0.4077, "step": 25843 }, { "epoch": 0.9207474571138465, "grad_norm": 1.6193742715762514, "learning_rate": 3.2750551027454727e-07, "loss": 0.5106, "step": 25844 }, { "epoch": 0.9207830842403406, "grad_norm": 1.2867203887745793, "learning_rate": 3.272126735708936e-07, "loss": 0.4423, "step": 25845 }, { "epoch": 0.9208187113668347, "grad_norm": 1.4821822264675182, "learning_rate": 3.269199656671629e-07, "loss": 0.5478, "step": 25846 }, { "epoch": 0.9208543384933289, "grad_norm": 1.5709523409165893, "learning_rate": 3.266273865672553e-07, "loss": 0.4935, "step": 25847 }, { "epoch": 0.920889965619823, "grad_norm": 1.2522996039649275, "learning_rate": 3.263349362750645e-07, "loss": 0.3933, "step": 25848 }, { "epoch": 0.920925592746317, "grad_norm": 1.3909296383866459, "learning_rate": 3.2604261479448507e-07, "loss": 0.386, "step": 25849 }, { "epoch": 0.9209612198728112, "grad_norm": 1.5731232730790532, "learning_rate": 3.2575042212940944e-07, "loss": 0.5123, "step": 25850 }, { "epoch": 0.9209968469993053, "grad_norm": 1.5345577341213434, "learning_rate": 3.254583582837312e-07, "loss": 0.4748, "step": 25851 }, { "epoch": 0.9210324741257994, "grad_norm": 1.5740321067056255, "learning_rate": 3.2516642326133496e-07, "loss": 0.5129, "step": 25852 }, { "epoch": 0.9210681012522935, "grad_norm": 1.1885527688030981, "learning_rate": 3.248746170661099e-07, "loss": 0.4437, "step": 25853 }, { "epoch": 0.9211037283787876, "grad_norm": 1.395511508879974, "learning_rate": 3.2458293970194285e-07, "loss": 0.3777, "step": 25854 }, { "epoch": 0.9211393555052817, "grad_norm": 1.6297610637216826, "learning_rate": 3.2429139117271634e-07, "loss": 0.4848, "step": 25855 }, { "epoch": 0.9211749826317759, "grad_norm": 1.3264263253370372, "learning_rate": 3.239999714823128e-07, "loss": 0.3774, "step": 25856 }, { "epoch": 0.92121060975827, "grad_norm": 1.8740403489475406, "learning_rate": 3.2370868063461234e-07, "loss": 0.472, "step": 25857 }, { "epoch": 0.921246236884764, "grad_norm": 1.7495794700364107, "learning_rate": 3.234175186334954e-07, "loss": 0.5346, "step": 25858 }, { "epoch": 0.9212818640112582, "grad_norm": 1.3194733443754558, "learning_rate": 3.2312648548283643e-07, "loss": 0.4008, "step": 25859 }, { "epoch": 0.9213174911377523, "grad_norm": 1.3050299213189156, "learning_rate": 3.228355811865136e-07, "loss": 0.4343, "step": 25860 }, { "epoch": 0.9213531182642464, "grad_norm": 1.5569706811479576, "learning_rate": 3.2254480574839706e-07, "loss": 0.4586, "step": 25861 }, { "epoch": 0.9213887453907406, "grad_norm": 1.591218254348737, "learning_rate": 3.222541591723627e-07, "loss": 0.6024, "step": 25862 }, { "epoch": 0.9214243725172346, "grad_norm": 1.2346018179829545, "learning_rate": 3.219636414622751e-07, "loss": 0.379, "step": 25863 }, { "epoch": 0.9214599996437287, "grad_norm": 1.224610840982425, "learning_rate": 3.216732526220079e-07, "loss": 0.4276, "step": 25864 }, { "epoch": 0.9214956267702229, "grad_norm": 1.1890228208560771, "learning_rate": 3.2138299265542794e-07, "loss": 0.392, "step": 25865 }, { "epoch": 0.921531253896717, "grad_norm": 1.6277278899584573, "learning_rate": 3.210928615663966e-07, "loss": 0.5094, "step": 25866 }, { "epoch": 0.921566881023211, "grad_norm": 2.297702017512701, "learning_rate": 3.2080285935877977e-07, "loss": 0.4994, "step": 25867 }, { "epoch": 0.9216025081497052, "grad_norm": 1.689941540244281, "learning_rate": 3.2051298603643754e-07, "loss": 0.4953, "step": 25868 }, { "epoch": 0.9216381352761993, "grad_norm": 1.5354510573988625, "learning_rate": 3.202232416032314e-07, "loss": 0.4369, "step": 25869 }, { "epoch": 0.9216737624026934, "grad_norm": 1.830272742939986, "learning_rate": 3.1993362606301703e-07, "loss": 0.4855, "step": 25870 }, { "epoch": 0.9217093895291876, "grad_norm": 1.5601814199561939, "learning_rate": 3.196441394196537e-07, "loss": 0.5341, "step": 25871 }, { "epoch": 0.9217450166556816, "grad_norm": 1.468741203726282, "learning_rate": 3.193547816769937e-07, "loss": 0.3873, "step": 25872 }, { "epoch": 0.9217806437821757, "grad_norm": 1.7268031921176143, "learning_rate": 3.19065552838892e-07, "loss": 0.4813, "step": 25873 }, { "epoch": 0.9218162709086699, "grad_norm": 1.278405445146193, "learning_rate": 3.1877645290919854e-07, "loss": 0.5171, "step": 25874 }, { "epoch": 0.921851898035164, "grad_norm": 1.4630410448605036, "learning_rate": 3.184874818917649e-07, "loss": 0.4201, "step": 25875 }, { "epoch": 0.9218875251616581, "grad_norm": 1.7166256985200248, "learning_rate": 3.1819863979043576e-07, "loss": 0.3765, "step": 25876 }, { "epoch": 0.9219231522881522, "grad_norm": 1.099066982360887, "learning_rate": 3.1790992660905906e-07, "loss": 0.3847, "step": 25877 }, { "epoch": 0.9219587794146463, "grad_norm": 1.689896169072331, "learning_rate": 3.1762134235147844e-07, "loss": 0.4246, "step": 25878 }, { "epoch": 0.9219944065411404, "grad_norm": 1.5377673772971379, "learning_rate": 3.1733288702153755e-07, "loss": 0.558, "step": 25879 }, { "epoch": 0.9220300336676346, "grad_norm": 1.2434569051591884, "learning_rate": 3.1704456062307877e-07, "loss": 0.3978, "step": 25880 }, { "epoch": 0.9220656607941287, "grad_norm": 1.583794510307887, "learning_rate": 3.1675636315993795e-07, "loss": 0.4215, "step": 25881 }, { "epoch": 0.9221012879206227, "grad_norm": 1.7137366798134115, "learning_rate": 3.164682946359554e-07, "loss": 0.5392, "step": 25882 }, { "epoch": 0.9221369150471169, "grad_norm": 1.1885314680360617, "learning_rate": 3.161803550549647e-07, "loss": 0.39, "step": 25883 }, { "epoch": 0.922172542173611, "grad_norm": 1.1852214830616499, "learning_rate": 3.1589254442080276e-07, "loss": 0.3985, "step": 25884 }, { "epoch": 0.9222081693001051, "grad_norm": 1.370278884107626, "learning_rate": 3.156048627372987e-07, "loss": 0.454, "step": 25885 }, { "epoch": 0.9222437964265993, "grad_norm": 1.590893177316494, "learning_rate": 3.1531731000828624e-07, "loss": 0.524, "step": 25886 }, { "epoch": 0.9222794235530933, "grad_norm": 1.3632448927258936, "learning_rate": 3.1502988623759115e-07, "loss": 0.3805, "step": 25887 }, { "epoch": 0.9223150506795874, "grad_norm": 1.3184812179789063, "learning_rate": 3.1474259142904363e-07, "loss": 0.407, "step": 25888 }, { "epoch": 0.9223506778060816, "grad_norm": 1.492919846052294, "learning_rate": 3.1445542558646846e-07, "loss": 0.4691, "step": 25889 }, { "epoch": 0.9223863049325757, "grad_norm": 1.7251754841847065, "learning_rate": 3.1416838871368925e-07, "loss": 0.4859, "step": 25890 }, { "epoch": 0.9224219320590697, "grad_norm": 2.1972761547080446, "learning_rate": 3.138814808145274e-07, "loss": 0.5428, "step": 25891 }, { "epoch": 0.9224575591855639, "grad_norm": 1.429877596435214, "learning_rate": 3.135947018928043e-07, "loss": 0.3943, "step": 25892 }, { "epoch": 0.922493186312058, "grad_norm": 1.465763165175735, "learning_rate": 3.1330805195233684e-07, "loss": 0.445, "step": 25893 }, { "epoch": 0.9225288134385521, "grad_norm": 1.421738990033907, "learning_rate": 3.130215309969442e-07, "loss": 0.3672, "step": 25894 }, { "epoch": 0.9225644405650463, "grad_norm": 1.4396529888249616, "learning_rate": 3.127351390304412e-07, "loss": 0.4025, "step": 25895 }, { "epoch": 0.9226000676915403, "grad_norm": 1.3470180918330341, "learning_rate": 3.124488760566402e-07, "loss": 0.5552, "step": 25896 }, { "epoch": 0.9226356948180344, "grad_norm": 1.8827710205256794, "learning_rate": 3.121627420793538e-07, "loss": 0.5495, "step": 25897 }, { "epoch": 0.9226713219445286, "grad_norm": 1.3974840828703567, "learning_rate": 3.118767371023912e-07, "loss": 0.3848, "step": 25898 }, { "epoch": 0.9227069490710227, "grad_norm": 1.5652798075363756, "learning_rate": 3.115908611295626e-07, "loss": 0.4743, "step": 25899 }, { "epoch": 0.9227425761975168, "grad_norm": 1.2522217851542838, "learning_rate": 3.113051141646717e-07, "loss": 0.3852, "step": 25900 }, { "epoch": 0.922778203324011, "grad_norm": 1.4540516165851702, "learning_rate": 3.110194962115265e-07, "loss": 0.4235, "step": 25901 }, { "epoch": 0.922813830450505, "grad_norm": 1.4838518612456297, "learning_rate": 3.1073400727392956e-07, "loss": 0.4709, "step": 25902 }, { "epoch": 0.9228494575769991, "grad_norm": 1.2116956215421308, "learning_rate": 3.1044864735568003e-07, "loss": 0.4125, "step": 25903 }, { "epoch": 0.9228850847034933, "grad_norm": 1.447361243490796, "learning_rate": 3.1016341646058155e-07, "loss": 0.5187, "step": 25904 }, { "epoch": 0.9229207118299874, "grad_norm": 1.6142573577560988, "learning_rate": 3.098783145924278e-07, "loss": 0.5369, "step": 25905 }, { "epoch": 0.9229563389564814, "grad_norm": 1.531705826053827, "learning_rate": 3.0959334175501897e-07, "loss": 0.4548, "step": 25906 }, { "epoch": 0.9229919660829756, "grad_norm": 1.5998945697125728, "learning_rate": 3.0930849795214657e-07, "loss": 0.5074, "step": 25907 }, { "epoch": 0.9230275932094697, "grad_norm": 1.4509322918373229, "learning_rate": 3.090237831876053e-07, "loss": 0.5075, "step": 25908 }, { "epoch": 0.9230632203359638, "grad_norm": 1.6350023913508303, "learning_rate": 3.0873919746518546e-07, "loss": 0.4864, "step": 25909 }, { "epoch": 0.923098847462458, "grad_norm": 1.3324040079380324, "learning_rate": 3.0845474078867733e-07, "loss": 0.4192, "step": 25910 }, { "epoch": 0.923134474588952, "grad_norm": 1.4812075791467683, "learning_rate": 3.081704131618701e-07, "loss": 0.5095, "step": 25911 }, { "epoch": 0.9231701017154461, "grad_norm": 1.305655001909833, "learning_rate": 3.078862145885453e-07, "loss": 0.392, "step": 25912 }, { "epoch": 0.9232057288419403, "grad_norm": 1.282764422225668, "learning_rate": 3.0760214507249197e-07, "loss": 0.4664, "step": 25913 }, { "epoch": 0.9232413559684344, "grad_norm": 1.5890729518601991, "learning_rate": 3.073182046174894e-07, "loss": 0.4859, "step": 25914 }, { "epoch": 0.9232769830949284, "grad_norm": 1.2433255661628924, "learning_rate": 3.070343932273201e-07, "loss": 0.3391, "step": 25915 }, { "epoch": 0.9233126102214226, "grad_norm": 1.966358998375993, "learning_rate": 3.0675071090576325e-07, "loss": 0.5346, "step": 25916 }, { "epoch": 0.9233482373479167, "grad_norm": 1.464249007922151, "learning_rate": 3.0646715765659583e-07, "loss": 0.5164, "step": 25917 }, { "epoch": 0.9233838644744108, "grad_norm": 1.6707157930637917, "learning_rate": 3.0618373348359265e-07, "loss": 0.5208, "step": 25918 }, { "epoch": 0.923419491600905, "grad_norm": 1.2573445323894645, "learning_rate": 3.0590043839053064e-07, "loss": 0.3778, "step": 25919 }, { "epoch": 0.923455118727399, "grad_norm": 1.6263196995414058, "learning_rate": 3.056172723811779e-07, "loss": 0.4843, "step": 25920 }, { "epoch": 0.9234907458538931, "grad_norm": 1.2842809386597633, "learning_rate": 3.0533423545930807e-07, "loss": 0.4214, "step": 25921 }, { "epoch": 0.9235263729803873, "grad_norm": 1.6064860057610346, "learning_rate": 3.050513276286893e-07, "loss": 0.444, "step": 25922 }, { "epoch": 0.9235620001068814, "grad_norm": 1.4116465541690293, "learning_rate": 3.0476854889308737e-07, "loss": 0.4623, "step": 25923 }, { "epoch": 0.9235976272333755, "grad_norm": 1.1757172899171247, "learning_rate": 3.044858992562694e-07, "loss": 0.3668, "step": 25924 }, { "epoch": 0.9236332543598696, "grad_norm": 1.44620195905738, "learning_rate": 3.0420337872199776e-07, "loss": 0.4826, "step": 25925 }, { "epoch": 0.9236688814863637, "grad_norm": 1.5106269788297386, "learning_rate": 3.039209872940363e-07, "loss": 0.4997, "step": 25926 }, { "epoch": 0.9237045086128578, "grad_norm": 2.0683775350023113, "learning_rate": 3.0363872497614413e-07, "loss": 0.5454, "step": 25927 }, { "epoch": 0.923740135739352, "grad_norm": 1.4791495311579412, "learning_rate": 3.033565917720793e-07, "loss": 0.4305, "step": 25928 }, { "epoch": 0.9237757628658461, "grad_norm": 1.584159090687879, "learning_rate": 3.0307458768559894e-07, "loss": 0.4714, "step": 25929 }, { "epoch": 0.9238113899923401, "grad_norm": 1.578510572748006, "learning_rate": 3.0279271272045776e-07, "loss": 0.5291, "step": 25930 }, { "epoch": 0.9238470171188343, "grad_norm": 1.5727199014081519, "learning_rate": 3.0251096688040936e-07, "loss": 0.5575, "step": 25931 }, { "epoch": 0.9238826442453284, "grad_norm": 1.2954875812293198, "learning_rate": 3.022293501692064e-07, "loss": 0.4339, "step": 25932 }, { "epoch": 0.9239182713718225, "grad_norm": 1.4888015511777586, "learning_rate": 3.0194786259059806e-07, "loss": 0.4258, "step": 25933 }, { "epoch": 0.9239538984983167, "grad_norm": 1.5917720899838506, "learning_rate": 3.016665041483313e-07, "loss": 0.5179, "step": 25934 }, { "epoch": 0.9239895256248107, "grad_norm": 1.358666059369882, "learning_rate": 3.013852748461554e-07, "loss": 0.4199, "step": 25935 }, { "epoch": 0.9240251527513048, "grad_norm": 1.291524530445176, "learning_rate": 3.0110417468781184e-07, "loss": 0.4032, "step": 25936 }, { "epoch": 0.924060779877799, "grad_norm": 1.5887402014790444, "learning_rate": 3.0082320367704756e-07, "loss": 0.4679, "step": 25937 }, { "epoch": 0.9240964070042931, "grad_norm": 1.4572318129723592, "learning_rate": 3.0054236181759845e-07, "loss": 0.5181, "step": 25938 }, { "epoch": 0.9241320341307871, "grad_norm": 1.3145364178793704, "learning_rate": 3.002616491132093e-07, "loss": 0.4138, "step": 25939 }, { "epoch": 0.9241676612572813, "grad_norm": 1.5628512786607358, "learning_rate": 2.9998106556761606e-07, "loss": 0.5842, "step": 25940 }, { "epoch": 0.9242032883837754, "grad_norm": 1.6173115961292575, "learning_rate": 2.9970061118455573e-07, "loss": 0.4735, "step": 25941 }, { "epoch": 0.9242389155102695, "grad_norm": 1.4023078514495462, "learning_rate": 2.9942028596775975e-07, "loss": 0.4145, "step": 25942 }, { "epoch": 0.9242745426367637, "grad_norm": 2.1733192797332954, "learning_rate": 2.9914008992096513e-07, "loss": 0.4437, "step": 25943 }, { "epoch": 0.9243101697632577, "grad_norm": 2.2089696294838737, "learning_rate": 2.9886002304789886e-07, "loss": 0.4868, "step": 25944 }, { "epoch": 0.9243457968897518, "grad_norm": 1.4034255686435133, "learning_rate": 2.9858008535229245e-07, "loss": 0.4612, "step": 25945 }, { "epoch": 0.924381424016246, "grad_norm": 1.405740675062127, "learning_rate": 2.983002768378729e-07, "loss": 0.5605, "step": 25946 }, { "epoch": 0.9244170511427401, "grad_norm": 2.128078932329392, "learning_rate": 2.980205975083661e-07, "loss": 0.5662, "step": 25947 }, { "epoch": 0.9244526782692342, "grad_norm": 1.6767841827912364, "learning_rate": 2.97741047367498e-07, "loss": 0.5815, "step": 25948 }, { "epoch": 0.9244883053957283, "grad_norm": 1.7363369390854635, "learning_rate": 2.974616264189878e-07, "loss": 0.4732, "step": 25949 }, { "epoch": 0.9245239325222224, "grad_norm": 1.2315033423917365, "learning_rate": 2.971823346665581e-07, "loss": 0.3721, "step": 25950 }, { "epoch": 0.9245595596487165, "grad_norm": 1.2346384886473447, "learning_rate": 2.9690317211392814e-07, "loss": 0.4353, "step": 25951 }, { "epoch": 0.9245951867752107, "grad_norm": 1.8662923814788592, "learning_rate": 2.966241387648128e-07, "loss": 0.5914, "step": 25952 }, { "epoch": 0.9246308139017048, "grad_norm": 1.6194395628685452, "learning_rate": 2.9634523462293005e-07, "loss": 0.5387, "step": 25953 }, { "epoch": 0.9246664410281988, "grad_norm": 1.2908597714019283, "learning_rate": 2.960664596919938e-07, "loss": 0.3879, "step": 25954 }, { "epoch": 0.924702068154693, "grad_norm": 1.7673056484680063, "learning_rate": 2.9578781397571534e-07, "loss": 0.6405, "step": 25955 }, { "epoch": 0.9247376952811871, "grad_norm": 1.5596103536435406, "learning_rate": 2.955092974778051e-07, "loss": 0.4578, "step": 25956 }, { "epoch": 0.9247733224076812, "grad_norm": 1.8542783531575233, "learning_rate": 2.9523091020197013e-07, "loss": 0.5587, "step": 25957 }, { "epoch": 0.9248089495341754, "grad_norm": 1.5691802242688169, "learning_rate": 2.949526521519197e-07, "loss": 0.4108, "step": 25958 }, { "epoch": 0.9248445766606694, "grad_norm": 1.5678058966852502, "learning_rate": 2.946745233313586e-07, "loss": 0.434, "step": 25959 }, { "epoch": 0.9248802037871635, "grad_norm": 1.5745152976685275, "learning_rate": 2.943965237439883e-07, "loss": 0.4609, "step": 25960 }, { "epoch": 0.9249158309136577, "grad_norm": 1.5769487678663818, "learning_rate": 2.9411865339351253e-07, "loss": 0.4822, "step": 25961 }, { "epoch": 0.9249514580401518, "grad_norm": 2.4178562797719496, "learning_rate": 2.9384091228363164e-07, "loss": 0.4365, "step": 25962 }, { "epoch": 0.9249870851666459, "grad_norm": 1.8231874584017318, "learning_rate": 2.9356330041804384e-07, "loss": 0.5397, "step": 25963 }, { "epoch": 0.92502271229314, "grad_norm": 1.6195503921062147, "learning_rate": 2.9328581780044383e-07, "loss": 0.4115, "step": 25964 }, { "epoch": 0.9250583394196341, "grad_norm": 1.2088247801872118, "learning_rate": 2.930084644345288e-07, "loss": 0.3859, "step": 25965 }, { "epoch": 0.9250939665461282, "grad_norm": 1.4031725293115922, "learning_rate": 2.9273124032399015e-07, "loss": 0.4404, "step": 25966 }, { "epoch": 0.9251295936726224, "grad_norm": 1.4398010048646768, "learning_rate": 2.924541454725194e-07, "loss": 0.4126, "step": 25967 }, { "epoch": 0.9251652207991165, "grad_norm": 1.3442481617668167, "learning_rate": 2.921771798838069e-07, "loss": 0.4111, "step": 25968 }, { "epoch": 0.9252008479256105, "grad_norm": 1.4875597223681305, "learning_rate": 2.9190034356154085e-07, "loss": 0.4897, "step": 25969 }, { "epoch": 0.9252364750521047, "grad_norm": 1.120558622482751, "learning_rate": 2.916236365094083e-07, "loss": 0.424, "step": 25970 }, { "epoch": 0.9252721021785988, "grad_norm": 1.6153435979793307, "learning_rate": 2.913470587310918e-07, "loss": 0.5127, "step": 25971 }, { "epoch": 0.9253077293050929, "grad_norm": 1.3022437627529286, "learning_rate": 2.910706102302763e-07, "loss": 0.4007, "step": 25972 }, { "epoch": 0.925343356431587, "grad_norm": 1.3287441933450093, "learning_rate": 2.9079429101064093e-07, "loss": 0.4268, "step": 25973 }, { "epoch": 0.9253789835580811, "grad_norm": 1.3338012929949405, "learning_rate": 2.905181010758662e-07, "loss": 0.3995, "step": 25974 }, { "epoch": 0.9254146106845752, "grad_norm": 1.2789284797506892, "learning_rate": 2.90242040429628e-07, "loss": 0.3853, "step": 25975 }, { "epoch": 0.9254502378110694, "grad_norm": 1.7956309290848809, "learning_rate": 2.8996610907560565e-07, "loss": 0.5465, "step": 25976 }, { "epoch": 0.9254858649375635, "grad_norm": 1.3100534547514078, "learning_rate": 2.896903070174706e-07, "loss": 0.4088, "step": 25977 }, { "epoch": 0.9255214920640575, "grad_norm": 1.3196602492139968, "learning_rate": 2.894146342588977e-07, "loss": 0.4203, "step": 25978 }, { "epoch": 0.9255571191905517, "grad_norm": 1.1767092202783804, "learning_rate": 2.8913909080355405e-07, "loss": 0.4085, "step": 25979 }, { "epoch": 0.9255927463170458, "grad_norm": 1.878777285217501, "learning_rate": 2.8886367665511116e-07, "loss": 0.4463, "step": 25980 }, { "epoch": 0.9256283734435399, "grad_norm": 1.8103118029252068, "learning_rate": 2.885883918172383e-07, "loss": 0.4776, "step": 25981 }, { "epoch": 0.9256640005700341, "grad_norm": 1.4823955789956529, "learning_rate": 2.88313236293597e-07, "loss": 0.4797, "step": 25982 }, { "epoch": 0.9256996276965281, "grad_norm": 1.2454761570420938, "learning_rate": 2.8803821008785203e-07, "loss": 0.4218, "step": 25983 }, { "epoch": 0.9257352548230222, "grad_norm": 1.3880391190017007, "learning_rate": 2.8776331320366836e-07, "loss": 0.5342, "step": 25984 }, { "epoch": 0.9257708819495164, "grad_norm": 1.355775535932896, "learning_rate": 2.874885456447041e-07, "loss": 0.3919, "step": 25985 }, { "epoch": 0.9258065090760105, "grad_norm": 1.2263761190194105, "learning_rate": 2.872139074146185e-07, "loss": 0.3765, "step": 25986 }, { "epoch": 0.9258421362025046, "grad_norm": 1.6469901254221888, "learning_rate": 2.8693939851706876e-07, "loss": 0.5344, "step": 25987 }, { "epoch": 0.9258777633289987, "grad_norm": 1.4756714330220357, "learning_rate": 2.8666501895570965e-07, "loss": 0.5258, "step": 25988 }, { "epoch": 0.9259133904554928, "grad_norm": 1.8408932139258782, "learning_rate": 2.863907687341949e-07, "loss": 0.4224, "step": 25989 }, { "epoch": 0.9259490175819869, "grad_norm": 1.4723992483078938, "learning_rate": 2.861166478561761e-07, "loss": 0.5145, "step": 25990 }, { "epoch": 0.9259846447084811, "grad_norm": 1.3193000097985528, "learning_rate": 2.858426563253036e-07, "loss": 0.4227, "step": 25991 }, { "epoch": 0.9260202718349752, "grad_norm": 1.1932876160536978, "learning_rate": 2.855687941452279e-07, "loss": 0.3585, "step": 25992 }, { "epoch": 0.9260558989614692, "grad_norm": 1.552864699631115, "learning_rate": 2.852950613195915e-07, "loss": 0.5167, "step": 25993 }, { "epoch": 0.9260915260879634, "grad_norm": 1.3633731229306212, "learning_rate": 2.850214578520438e-07, "loss": 0.3797, "step": 25994 }, { "epoch": 0.9261271532144575, "grad_norm": 1.4516759378977384, "learning_rate": 2.8474798374622416e-07, "loss": 0.427, "step": 25995 }, { "epoch": 0.9261627803409516, "grad_norm": 1.2335747834232786, "learning_rate": 2.8447463900577623e-07, "loss": 0.4215, "step": 25996 }, { "epoch": 0.9261984074674458, "grad_norm": 1.3453826058733809, "learning_rate": 2.842014236343382e-07, "loss": 0.4675, "step": 25997 }, { "epoch": 0.9262340345939398, "grad_norm": 1.5469516106445194, "learning_rate": 2.839283376355506e-07, "loss": 0.559, "step": 25998 }, { "epoch": 0.9262696617204339, "grad_norm": 1.7316806999448795, "learning_rate": 2.8365538101304713e-07, "loss": 0.4842, "step": 25999 }, { "epoch": 0.9263052888469281, "grad_norm": 1.2409415990520878, "learning_rate": 2.8338255377046375e-07, "loss": 0.3667, "step": 26000 }, { "epoch": 0.9263409159734222, "grad_norm": 1.7058479900010672, "learning_rate": 2.8310985591143313e-07, "loss": 0.4573, "step": 26001 }, { "epoch": 0.9263765430999162, "grad_norm": 1.828807204358881, "learning_rate": 2.8283728743958793e-07, "loss": 0.6038, "step": 26002 }, { "epoch": 0.9264121702264104, "grad_norm": 1.7849986963232383, "learning_rate": 2.8256484835855526e-07, "loss": 0.4602, "step": 26003 }, { "epoch": 0.9264477973529045, "grad_norm": 1.288042596815674, "learning_rate": 2.8229253867196216e-07, "loss": 0.3744, "step": 26004 }, { "epoch": 0.9264834244793986, "grad_norm": 1.5035149901721747, "learning_rate": 2.82020358383438e-07, "loss": 0.3799, "step": 26005 }, { "epoch": 0.9265190516058928, "grad_norm": 1.501029086309732, "learning_rate": 2.817483074966043e-07, "loss": 0.4315, "step": 26006 }, { "epoch": 0.9265546787323868, "grad_norm": 1.1985535797953875, "learning_rate": 2.814763860150849e-07, "loss": 0.3905, "step": 26007 }, { "epoch": 0.9265903058588809, "grad_norm": 1.3953343531121711, "learning_rate": 2.812045939425001e-07, "loss": 0.4024, "step": 26008 }, { "epoch": 0.9266259329853751, "grad_norm": 1.3742800045409016, "learning_rate": 2.8093293128247046e-07, "loss": 0.3823, "step": 26009 }, { "epoch": 0.9266615601118692, "grad_norm": 1.2271759857807287, "learning_rate": 2.806613980386108e-07, "loss": 0.4149, "step": 26010 }, { "epoch": 0.9266971872383633, "grad_norm": 1.3763952465741658, "learning_rate": 2.8038999421453827e-07, "loss": 0.4433, "step": 26011 }, { "epoch": 0.9267328143648574, "grad_norm": 1.9057760861047393, "learning_rate": 2.8011871981386664e-07, "loss": 0.4821, "step": 26012 }, { "epoch": 0.9267684414913515, "grad_norm": 1.2675235163627887, "learning_rate": 2.798475748402074e-07, "loss": 0.3508, "step": 26013 }, { "epoch": 0.9268040686178456, "grad_norm": 1.9563453947188174, "learning_rate": 2.795765592971722e-07, "loss": 0.4929, "step": 26014 }, { "epoch": 0.9268396957443398, "grad_norm": 1.699984248871944, "learning_rate": 2.7930567318836923e-07, "loss": 0.5495, "step": 26015 }, { "epoch": 0.9268753228708339, "grad_norm": 1.4376133196156506, "learning_rate": 2.7903491651740555e-07, "loss": 0.4915, "step": 26016 }, { "epoch": 0.9269109499973279, "grad_norm": 1.3458406139202401, "learning_rate": 2.787642892878861e-07, "loss": 0.4108, "step": 26017 }, { "epoch": 0.9269465771238221, "grad_norm": 1.56836832170468, "learning_rate": 2.784937915034169e-07, "loss": 0.4953, "step": 26018 }, { "epoch": 0.9269822042503162, "grad_norm": 1.8230726791454541, "learning_rate": 2.782234231675951e-07, "loss": 0.5036, "step": 26019 }, { "epoch": 0.9270178313768103, "grad_norm": 1.560900663772141, "learning_rate": 2.779531842840244e-07, "loss": 0.523, "step": 26020 }, { "epoch": 0.9270534585033045, "grad_norm": 1.580950573340333, "learning_rate": 2.776830748563031e-07, "loss": 0.5479, "step": 26021 }, { "epoch": 0.9270890856297985, "grad_norm": 1.2687629621132264, "learning_rate": 2.7741309488802717e-07, "loss": 0.3599, "step": 26022 }, { "epoch": 0.9271247127562926, "grad_norm": 1.2407489323288377, "learning_rate": 2.771432443827904e-07, "loss": 0.3819, "step": 26023 }, { "epoch": 0.9271603398827868, "grad_norm": 1.3045131063522981, "learning_rate": 2.768735233441888e-07, "loss": 0.3855, "step": 26024 }, { "epoch": 0.9271959670092809, "grad_norm": 1.4518437320281694, "learning_rate": 2.766039317758107e-07, "loss": 0.3867, "step": 26025 }, { "epoch": 0.9272315941357749, "grad_norm": 1.5077269043332653, "learning_rate": 2.7633446968124864e-07, "loss": 0.5453, "step": 26026 }, { "epoch": 0.9272672212622691, "grad_norm": 1.3187829113356007, "learning_rate": 2.760651370640888e-07, "loss": 0.4071, "step": 26027 }, { "epoch": 0.9273028483887632, "grad_norm": 1.246911402633859, "learning_rate": 2.757959339279181e-07, "loss": 0.3985, "step": 26028 }, { "epoch": 0.9273384755152573, "grad_norm": 1.2788713709843078, "learning_rate": 2.755268602763217e-07, "loss": 0.4003, "step": 26029 }, { "epoch": 0.9273741026417515, "grad_norm": 1.3726907716246732, "learning_rate": 2.752579161128821e-07, "loss": 0.4244, "step": 26030 }, { "epoch": 0.9274097297682455, "grad_norm": 1.3881689277270326, "learning_rate": 2.7498910144118206e-07, "loss": 0.3828, "step": 26031 }, { "epoch": 0.9274453568947396, "grad_norm": 1.2437938633575822, "learning_rate": 2.7472041626479763e-07, "loss": 0.3851, "step": 26032 }, { "epoch": 0.9274809840212338, "grad_norm": 1.1750166146201966, "learning_rate": 2.744518605873092e-07, "loss": 0.4353, "step": 26033 }, { "epoch": 0.9275166111477279, "grad_norm": 1.555736277919922, "learning_rate": 2.741834344122918e-07, "loss": 0.4484, "step": 26034 }, { "epoch": 0.927552238274222, "grad_norm": 1.2472962068900164, "learning_rate": 2.739151377433191e-07, "loss": 0.3617, "step": 26035 }, { "epoch": 0.9275878654007161, "grad_norm": 2.1319027523218637, "learning_rate": 2.7364697058396504e-07, "loss": 0.5018, "step": 26036 }, { "epoch": 0.9276234925272102, "grad_norm": 1.6840525387037637, "learning_rate": 2.7337893293780007e-07, "loss": 0.5154, "step": 26037 }, { "epoch": 0.9276591196537043, "grad_norm": 1.208449725728887, "learning_rate": 2.7311102480839347e-07, "loss": 0.3772, "step": 26038 }, { "epoch": 0.9276947467801985, "grad_norm": 1.2064852277841254, "learning_rate": 2.728432461993114e-07, "loss": 0.3831, "step": 26039 }, { "epoch": 0.9277303739066926, "grad_norm": 1.7927436699201058, "learning_rate": 2.7257559711412195e-07, "loss": 0.5121, "step": 26040 }, { "epoch": 0.9277660010331866, "grad_norm": 1.3094052401735878, "learning_rate": 2.7230807755638576e-07, "loss": 0.4489, "step": 26041 }, { "epoch": 0.9278016281596808, "grad_norm": 1.2340662737116925, "learning_rate": 2.720406875296677e-07, "loss": 0.3741, "step": 26042 }, { "epoch": 0.9278372552861749, "grad_norm": 1.220385514431608, "learning_rate": 2.717734270375272e-07, "loss": 0.4454, "step": 26043 }, { "epoch": 0.927872882412669, "grad_norm": 1.3277547149660647, "learning_rate": 2.715062960835235e-07, "loss": 0.3927, "step": 26044 }, { "epoch": 0.9279085095391632, "grad_norm": 1.8089928740923689, "learning_rate": 2.712392946712117e-07, "loss": 0.5629, "step": 26045 }, { "epoch": 0.9279441366656572, "grad_norm": 4.105690957044076, "learning_rate": 2.7097242280415105e-07, "loss": 0.4741, "step": 26046 }, { "epoch": 0.9279797637921513, "grad_norm": 1.2978553137993674, "learning_rate": 2.7070568048589094e-07, "loss": 0.3721, "step": 26047 }, { "epoch": 0.9280153909186455, "grad_norm": 1.6074921741442785, "learning_rate": 2.704390677199864e-07, "loss": 0.5364, "step": 26048 }, { "epoch": 0.9280510180451396, "grad_norm": 1.516587802603575, "learning_rate": 2.701725845099856e-07, "loss": 0.4505, "step": 26049 }, { "epoch": 0.9280866451716336, "grad_norm": 2.050121327669299, "learning_rate": 2.699062308594358e-07, "loss": 0.4678, "step": 26050 }, { "epoch": 0.9281222722981278, "grad_norm": 2.140584247414891, "learning_rate": 2.696400067718874e-07, "loss": 0.4154, "step": 26051 }, { "epoch": 0.9281578994246219, "grad_norm": 1.4774189255963412, "learning_rate": 2.6937391225088315e-07, "loss": 0.45, "step": 26052 }, { "epoch": 0.928193526551116, "grad_norm": 1.7441929879781366, "learning_rate": 2.691079472999669e-07, "loss": 0.5404, "step": 26053 }, { "epoch": 0.9282291536776102, "grad_norm": 1.7239518180525368, "learning_rate": 2.688421119226792e-07, "loss": 0.4962, "step": 26054 }, { "epoch": 0.9282647808041042, "grad_norm": 1.261442537778406, "learning_rate": 2.685764061225615e-07, "loss": 0.4452, "step": 26055 }, { "epoch": 0.9283004079305983, "grad_norm": 1.4739678245654857, "learning_rate": 2.6831082990315003e-07, "loss": 0.3777, "step": 26056 }, { "epoch": 0.9283360350570925, "grad_norm": 1.6423556518917817, "learning_rate": 2.680453832679808e-07, "loss": 0.5604, "step": 26057 }, { "epoch": 0.9283716621835866, "grad_norm": 1.4923366740003683, "learning_rate": 2.6778006622059095e-07, "loss": 0.478, "step": 26058 }, { "epoch": 0.9284072893100807, "grad_norm": 1.236458375780129, "learning_rate": 2.675148787645121e-07, "loss": 0.3843, "step": 26059 }, { "epoch": 0.9284429164365748, "grad_norm": 1.1935697320590684, "learning_rate": 2.6724982090327477e-07, "loss": 0.3812, "step": 26060 }, { "epoch": 0.9284785435630689, "grad_norm": 1.571843671063289, "learning_rate": 2.669848926404084e-07, "loss": 0.4999, "step": 26061 }, { "epoch": 0.928514170689563, "grad_norm": 1.3676866755352943, "learning_rate": 2.667200939794423e-07, "loss": 0.3951, "step": 26062 }, { "epoch": 0.9285497978160572, "grad_norm": 1.2909410867299527, "learning_rate": 2.664554249238993e-07, "loss": 0.4095, "step": 26063 }, { "epoch": 0.9285854249425513, "grad_norm": 1.4675758437816677, "learning_rate": 2.6619088547730763e-07, "loss": 0.4714, "step": 26064 }, { "epoch": 0.9286210520690453, "grad_norm": 1.3874996961367765, "learning_rate": 2.6592647564318676e-07, "loss": 0.388, "step": 26065 }, { "epoch": 0.9286566791955395, "grad_norm": 1.3529811048421019, "learning_rate": 2.656621954250593e-07, "loss": 0.4184, "step": 26066 }, { "epoch": 0.9286923063220336, "grad_norm": 1.4004722085113996, "learning_rate": 2.653980448264426e-07, "loss": 0.4596, "step": 26067 }, { "epoch": 0.9287279334485277, "grad_norm": 1.18170119403104, "learning_rate": 2.6513402385085706e-07, "loss": 0.3639, "step": 26068 }, { "epoch": 0.9287635605750219, "grad_norm": 1.5572908086026904, "learning_rate": 2.6487013250181436e-07, "loss": 0.5325, "step": 26069 }, { "epoch": 0.9287991877015159, "grad_norm": 1.9005669363151505, "learning_rate": 2.646063707828317e-07, "loss": 0.5069, "step": 26070 }, { "epoch": 0.92883481482801, "grad_norm": 1.2039289951661483, "learning_rate": 2.643427386974184e-07, "loss": 0.414, "step": 26071 }, { "epoch": 0.9288704419545042, "grad_norm": 1.2636134204836764, "learning_rate": 2.640792362490863e-07, "loss": 0.3958, "step": 26072 }, { "epoch": 0.9289060690809983, "grad_norm": 1.4579548681382002, "learning_rate": 2.638158634413446e-07, "loss": 0.5655, "step": 26073 }, { "epoch": 0.9289416962074923, "grad_norm": 1.2395009552747982, "learning_rate": 2.6355262027769946e-07, "loss": 0.4166, "step": 26074 }, { "epoch": 0.9289773233339865, "grad_norm": 1.2300523004571084, "learning_rate": 2.632895067616581e-07, "loss": 0.4415, "step": 26075 }, { "epoch": 0.9290129504604806, "grad_norm": 1.50329437706621, "learning_rate": 2.6302652289672104e-07, "loss": 0.4911, "step": 26076 }, { "epoch": 0.9290485775869747, "grad_norm": 1.3577005033841738, "learning_rate": 2.627636686863932e-07, "loss": 0.3862, "step": 26077 }, { "epoch": 0.9290842047134689, "grad_norm": 1.396766369776004, "learning_rate": 2.625009441341708e-07, "loss": 0.4456, "step": 26078 }, { "epoch": 0.9291198318399629, "grad_norm": 1.7420483122219828, "learning_rate": 2.622383492435554e-07, "loss": 0.5279, "step": 26079 }, { "epoch": 0.929155458966457, "grad_norm": 1.654694029521528, "learning_rate": 2.619758840180431e-07, "loss": 0.4236, "step": 26080 }, { "epoch": 0.9291910860929512, "grad_norm": 1.3152959935840909, "learning_rate": 2.6171354846112885e-07, "loss": 0.4243, "step": 26081 }, { "epoch": 0.9292267132194453, "grad_norm": 1.3588671683465723, "learning_rate": 2.6145134257630433e-07, "loss": 0.4627, "step": 26082 }, { "epoch": 0.9292623403459394, "grad_norm": 1.2702399523147403, "learning_rate": 2.611892663670612e-07, "loss": 0.4014, "step": 26083 }, { "epoch": 0.9292979674724335, "grad_norm": 1.470181045284227, "learning_rate": 2.609273198368922e-07, "loss": 0.3766, "step": 26084 }, { "epoch": 0.9293335945989276, "grad_norm": 1.7027735703518785, "learning_rate": 2.606655029892813e-07, "loss": 0.4845, "step": 26085 }, { "epoch": 0.9293692217254217, "grad_norm": 1.3798748980488376, "learning_rate": 2.604038158277189e-07, "loss": 0.5276, "step": 26086 }, { "epoch": 0.9294048488519159, "grad_norm": 1.2214288795505783, "learning_rate": 2.601422583556845e-07, "loss": 0.3867, "step": 26087 }, { "epoch": 0.92944047597841, "grad_norm": 1.3109524678243858, "learning_rate": 2.5988083057666534e-07, "loss": 0.4134, "step": 26088 }, { "epoch": 0.929476103104904, "grad_norm": 1.8618766134164157, "learning_rate": 2.5961953249413974e-07, "loss": 0.5151, "step": 26089 }, { "epoch": 0.9295117302313982, "grad_norm": 1.4381478803961236, "learning_rate": 2.5935836411158935e-07, "loss": 0.4859, "step": 26090 }, { "epoch": 0.9295473573578923, "grad_norm": 1.4089698677759859, "learning_rate": 2.5909732543249023e-07, "loss": 0.4064, "step": 26091 }, { "epoch": 0.9295829844843864, "grad_norm": 1.5108708281328225, "learning_rate": 2.5883641646031966e-07, "loss": 0.3902, "step": 26092 }, { "epoch": 0.9296186116108806, "grad_norm": 1.355949269173858, "learning_rate": 2.5857563719854927e-07, "loss": 0.4606, "step": 26093 }, { "epoch": 0.9296542387373746, "grad_norm": 1.81200875490147, "learning_rate": 2.58314987650653e-07, "loss": 0.4915, "step": 26094 }, { "epoch": 0.9296898658638687, "grad_norm": 1.4144620074128151, "learning_rate": 2.5805446782010245e-07, "loss": 0.4114, "step": 26095 }, { "epoch": 0.9297254929903629, "grad_norm": 1.329552478281994, "learning_rate": 2.577940777103649e-07, "loss": 0.4246, "step": 26096 }, { "epoch": 0.929761120116857, "grad_norm": 1.5230192998157936, "learning_rate": 2.575338173249109e-07, "loss": 0.4486, "step": 26097 }, { "epoch": 0.929796747243351, "grad_norm": 1.4419493293275913, "learning_rate": 2.572736866672021e-07, "loss": 0.4089, "step": 26098 }, { "epoch": 0.9298323743698452, "grad_norm": 1.2898616713964917, "learning_rate": 2.570136857407046e-07, "loss": 0.3936, "step": 26099 }, { "epoch": 0.9298680014963393, "grad_norm": 1.6285349833597162, "learning_rate": 2.5675381454887903e-07, "loss": 0.4735, "step": 26100 }, { "epoch": 0.9299036286228334, "grad_norm": 1.8316528012378863, "learning_rate": 2.56494073095187e-07, "loss": 0.4876, "step": 26101 }, { "epoch": 0.9299392557493276, "grad_norm": 1.6411978917885726, "learning_rate": 2.5623446138308584e-07, "loss": 0.4977, "step": 26102 }, { "epoch": 0.9299748828758216, "grad_norm": 1.4564214775234066, "learning_rate": 2.559749794160349e-07, "loss": 0.4142, "step": 26103 }, { "epoch": 0.9300105100023157, "grad_norm": 1.5558263384169053, "learning_rate": 2.5571562719748587e-07, "loss": 0.5004, "step": 26104 }, { "epoch": 0.9300461371288099, "grad_norm": 1.9420398672631773, "learning_rate": 2.5545640473089607e-07, "loss": 0.513, "step": 26105 }, { "epoch": 0.930081764255304, "grad_norm": 1.3997491667070092, "learning_rate": 2.5519731201971374e-07, "loss": 0.4002, "step": 26106 }, { "epoch": 0.930117391381798, "grad_norm": 1.1936716762594817, "learning_rate": 2.5493834906739065e-07, "loss": 0.4254, "step": 26107 }, { "epoch": 0.9301530185082922, "grad_norm": 1.4476152187995903, "learning_rate": 2.5467951587737625e-07, "loss": 0.5025, "step": 26108 }, { "epoch": 0.9301886456347863, "grad_norm": 1.2776919968140017, "learning_rate": 2.5442081245311446e-07, "loss": 0.3928, "step": 26109 }, { "epoch": 0.9302242727612804, "grad_norm": 1.2594342397848148, "learning_rate": 2.541622387980514e-07, "loss": 0.4069, "step": 26110 }, { "epoch": 0.9302598998877746, "grad_norm": 1.5533320663430104, "learning_rate": 2.5390379491562977e-07, "loss": 0.4991, "step": 26111 }, { "epoch": 0.9302955270142687, "grad_norm": 1.2259420592319181, "learning_rate": 2.536454808092914e-07, "loss": 0.3932, "step": 26112 }, { "epoch": 0.9303311541407627, "grad_norm": 1.1993977014726567, "learning_rate": 2.5338729648247574e-07, "loss": 0.3831, "step": 26113 }, { "epoch": 0.9303667812672569, "grad_norm": 1.6391344645823325, "learning_rate": 2.531292419386222e-07, "loss": 0.5926, "step": 26114 }, { "epoch": 0.930402408393751, "grad_norm": 1.1796737140715137, "learning_rate": 2.5287131718116365e-07, "loss": 0.438, "step": 26115 }, { "epoch": 0.9304380355202451, "grad_norm": 1.440649845175448, "learning_rate": 2.526135222135362e-07, "loss": 0.4039, "step": 26116 }, { "epoch": 0.9304736626467393, "grad_norm": 1.4712448620723984, "learning_rate": 2.5235585703917265e-07, "loss": 0.5466, "step": 26117 }, { "epoch": 0.9305092897732333, "grad_norm": 1.8484069588373353, "learning_rate": 2.520983216615047e-07, "loss": 0.5392, "step": 26118 }, { "epoch": 0.9305449168997274, "grad_norm": 1.2119100028548035, "learning_rate": 2.518409160839619e-07, "loss": 0.338, "step": 26119 }, { "epoch": 0.9305805440262216, "grad_norm": 1.4357321206965534, "learning_rate": 2.515836403099703e-07, "loss": 0.3863, "step": 26120 }, { "epoch": 0.9306161711527157, "grad_norm": 1.321918276875367, "learning_rate": 2.5132649434295607e-07, "loss": 0.4313, "step": 26121 }, { "epoch": 0.9306517982792097, "grad_norm": 1.7783188001338397, "learning_rate": 2.510694781863432e-07, "loss": 0.5485, "step": 26122 }, { "epoch": 0.9306874254057039, "grad_norm": 1.6250427509353533, "learning_rate": 2.5081259184355554e-07, "loss": 0.4363, "step": 26123 }, { "epoch": 0.930723052532198, "grad_norm": 1.243135706877789, "learning_rate": 2.5055583531801044e-07, "loss": 0.3927, "step": 26124 }, { "epoch": 0.9307586796586921, "grad_norm": 2.2431340437959277, "learning_rate": 2.502992086131295e-07, "loss": 0.3783, "step": 26125 }, { "epoch": 0.9307943067851863, "grad_norm": 1.3416976423911762, "learning_rate": 2.500427117323301e-07, "loss": 0.4517, "step": 26126 }, { "epoch": 0.9308299339116803, "grad_norm": 1.3867890502487843, "learning_rate": 2.4978634467902607e-07, "loss": 0.4096, "step": 26127 }, { "epoch": 0.9308655610381744, "grad_norm": 1.981230093393055, "learning_rate": 2.4953010745663255e-07, "loss": 0.4865, "step": 26128 }, { "epoch": 0.9309011881646686, "grad_norm": 1.7124451853044942, "learning_rate": 2.492740000685612e-07, "loss": 0.4802, "step": 26129 }, { "epoch": 0.9309368152911627, "grad_norm": 1.1894029236589907, "learning_rate": 2.490180225182204e-07, "loss": 0.3652, "step": 26130 }, { "epoch": 0.9309724424176568, "grad_norm": 1.3584811455864394, "learning_rate": 2.4876217480902077e-07, "loss": 0.4406, "step": 26131 }, { "epoch": 0.9310080695441509, "grad_norm": 1.5250930248553147, "learning_rate": 2.4850645694436736e-07, "loss": 0.5123, "step": 26132 }, { "epoch": 0.931043696670645, "grad_norm": 2.423485514992703, "learning_rate": 2.4825086892766745e-07, "loss": 0.5323, "step": 26133 }, { "epoch": 0.9310793237971391, "grad_norm": 1.870385961725495, "learning_rate": 2.479954107623239e-07, "loss": 0.5525, "step": 26134 }, { "epoch": 0.9311149509236333, "grad_norm": 1.327182679815844, "learning_rate": 2.477400824517373e-07, "loss": 0.5207, "step": 26135 }, { "epoch": 0.9311505780501274, "grad_norm": 1.2197741777633209, "learning_rate": 2.4748488399930826e-07, "loss": 0.3762, "step": 26136 }, { "epoch": 0.9311862051766214, "grad_norm": 1.1949559796451235, "learning_rate": 2.4722981540843403e-07, "loss": 0.3979, "step": 26137 }, { "epoch": 0.9312218323031156, "grad_norm": 1.1829813420137334, "learning_rate": 2.46974876682512e-07, "loss": 0.3521, "step": 26138 }, { "epoch": 0.9312574594296097, "grad_norm": 1.4043343387331244, "learning_rate": 2.4672006782493487e-07, "loss": 0.431, "step": 26139 }, { "epoch": 0.9312930865561038, "grad_norm": 1.5606178490116367, "learning_rate": 2.4646538883909885e-07, "loss": 0.4621, "step": 26140 }, { "epoch": 0.931328713682598, "grad_norm": 1.4946199949962888, "learning_rate": 2.4621083972839244e-07, "loss": 0.4228, "step": 26141 }, { "epoch": 0.931364340809092, "grad_norm": 1.7977691941083658, "learning_rate": 2.4595642049620725e-07, "loss": 0.501, "step": 26142 }, { "epoch": 0.9313999679355861, "grad_norm": 1.3574046426693633, "learning_rate": 2.4570213114592957e-07, "loss": 0.4326, "step": 26143 }, { "epoch": 0.9314355950620803, "grad_norm": 1.6157563154352519, "learning_rate": 2.4544797168094547e-07, "loss": 0.5396, "step": 26144 }, { "epoch": 0.9314712221885744, "grad_norm": 1.337250468543103, "learning_rate": 2.451939421046412e-07, "loss": 0.4306, "step": 26145 }, { "epoch": 0.9315068493150684, "grad_norm": 1.2339364188498587, "learning_rate": 2.449400424203963e-07, "loss": 0.3833, "step": 26146 }, { "epoch": 0.9315424764415626, "grad_norm": 1.7593398289700846, "learning_rate": 2.446862726315935e-07, "loss": 0.4885, "step": 26147 }, { "epoch": 0.9315781035680567, "grad_norm": 1.541242557827598, "learning_rate": 2.444326327416102e-07, "loss": 0.5149, "step": 26148 }, { "epoch": 0.9316137306945508, "grad_norm": 1.2603147739577378, "learning_rate": 2.4417912275382703e-07, "loss": 0.4062, "step": 26149 }, { "epoch": 0.931649357821045, "grad_norm": 1.1992976731170726, "learning_rate": 2.439257426716168e-07, "loss": 0.4261, "step": 26150 }, { "epoch": 0.931684984947539, "grad_norm": 1.2367047709317411, "learning_rate": 2.436724924983558e-07, "loss": 0.3669, "step": 26151 }, { "epoch": 0.9317206120740331, "grad_norm": 1.8392401142766819, "learning_rate": 2.434193722374134e-07, "loss": 0.5227, "step": 26152 }, { "epoch": 0.9317562392005273, "grad_norm": 1.209340339898778, "learning_rate": 2.4316638189216146e-07, "loss": 0.39, "step": 26153 }, { "epoch": 0.9317918663270214, "grad_norm": 1.4412851168641843, "learning_rate": 2.4291352146596946e-07, "loss": 0.3908, "step": 26154 }, { "epoch": 0.9318274934535155, "grad_norm": 1.554884578802433, "learning_rate": 2.426607909622025e-07, "loss": 0.5305, "step": 26155 }, { "epoch": 0.9318631205800096, "grad_norm": 1.2045233790751435, "learning_rate": 2.4240819038422903e-07, "loss": 0.3987, "step": 26156 }, { "epoch": 0.9318987477065037, "grad_norm": 2.2280811149398936, "learning_rate": 2.4215571973540854e-07, "loss": 0.4808, "step": 26157 }, { "epoch": 0.9319343748329978, "grad_norm": 1.392302830077714, "learning_rate": 2.419033790191072e-07, "loss": 0.388, "step": 26158 }, { "epoch": 0.931970001959492, "grad_norm": 1.340615474612722, "learning_rate": 2.416511682386813e-07, "loss": 0.3945, "step": 26159 }, { "epoch": 0.932005629085986, "grad_norm": 1.77251264124837, "learning_rate": 2.413990873974914e-07, "loss": 0.4724, "step": 26160 }, { "epoch": 0.9320412562124801, "grad_norm": 1.3808198380296621, "learning_rate": 2.411471364988938e-07, "loss": 0.4008, "step": 26161 }, { "epoch": 0.9320768833389743, "grad_norm": 1.3834511381230044, "learning_rate": 2.4089531554624233e-07, "loss": 0.4233, "step": 26162 }, { "epoch": 0.9321125104654684, "grad_norm": 1.5327289268811441, "learning_rate": 2.4064362454289e-07, "loss": 0.4509, "step": 26163 }, { "epoch": 0.9321481375919625, "grad_norm": 1.483211060899908, "learning_rate": 2.403920634921908e-07, "loss": 0.4388, "step": 26164 }, { "epoch": 0.9321837647184567, "grad_norm": 1.7549812762903194, "learning_rate": 2.401406323974931e-07, "loss": 0.4649, "step": 26165 }, { "epoch": 0.9322193918449507, "grad_norm": 1.7003561196740782, "learning_rate": 2.398893312621431e-07, "loss": 0.5063, "step": 26166 }, { "epoch": 0.9322550189714448, "grad_norm": 1.706080866988777, "learning_rate": 2.3963816008949034e-07, "loss": 0.5521, "step": 26167 }, { "epoch": 0.932290646097939, "grad_norm": 2.48016484769803, "learning_rate": 2.393871188828767e-07, "loss": 0.5612, "step": 26168 }, { "epoch": 0.9323262732244331, "grad_norm": 1.6024761301059347, "learning_rate": 2.3913620764564495e-07, "loss": 0.5118, "step": 26169 }, { "epoch": 0.9323619003509271, "grad_norm": 1.3639188303928094, "learning_rate": 2.38885426381138e-07, "loss": 0.4133, "step": 26170 }, { "epoch": 0.9323975274774213, "grad_norm": 1.6693733194792015, "learning_rate": 2.386347750926943e-07, "loss": 0.537, "step": 26171 }, { "epoch": 0.9324331546039154, "grad_norm": 1.164539618421571, "learning_rate": 2.3838425378365116e-07, "loss": 0.39, "step": 26172 }, { "epoch": 0.9324687817304095, "grad_norm": 1.3585305167724349, "learning_rate": 2.3813386245734483e-07, "loss": 0.3626, "step": 26173 }, { "epoch": 0.9325044088569037, "grad_norm": 1.630747657185869, "learning_rate": 2.3788360111710818e-07, "loss": 0.5275, "step": 26174 }, { "epoch": 0.9325400359833977, "grad_norm": 1.8039959741738336, "learning_rate": 2.3763346976627632e-07, "loss": 0.5262, "step": 26175 }, { "epoch": 0.9325756631098918, "grad_norm": 1.2097304413923045, "learning_rate": 2.3738346840817772e-07, "loss": 0.413, "step": 26176 }, { "epoch": 0.932611290236386, "grad_norm": 1.4738961482045978, "learning_rate": 2.371335970461408e-07, "loss": 0.4323, "step": 26177 }, { "epoch": 0.9326469173628801, "grad_norm": 1.558810807539576, "learning_rate": 2.3688385568349515e-07, "loss": 0.5104, "step": 26178 }, { "epoch": 0.9326825444893742, "grad_norm": 1.40847422903674, "learning_rate": 2.3663424432356363e-07, "loss": 0.4159, "step": 26179 }, { "epoch": 0.9327181716158683, "grad_norm": 1.3897636202573849, "learning_rate": 2.363847629696736e-07, "loss": 0.4052, "step": 26180 }, { "epoch": 0.9327537987423624, "grad_norm": 1.2940260575540352, "learning_rate": 2.3613541162514242e-07, "loss": 0.3886, "step": 26181 }, { "epoch": 0.9327894258688565, "grad_norm": 1.506480892323333, "learning_rate": 2.358861902932952e-07, "loss": 0.5173, "step": 26182 }, { "epoch": 0.9328250529953507, "grad_norm": 1.4925337433374155, "learning_rate": 2.3563709897744592e-07, "loss": 0.4209, "step": 26183 }, { "epoch": 0.9328606801218448, "grad_norm": 1.344035996743241, "learning_rate": 2.3538813768091418e-07, "loss": 0.3978, "step": 26184 }, { "epoch": 0.9328963072483389, "grad_norm": 1.6001007105732736, "learning_rate": 2.35139306407014e-07, "loss": 0.4128, "step": 26185 }, { "epoch": 0.932931934374833, "grad_norm": 1.2797109519440482, "learning_rate": 2.3489060515906048e-07, "loss": 0.4329, "step": 26186 }, { "epoch": 0.9329675615013271, "grad_norm": 1.4464344746992057, "learning_rate": 2.3464203394036322e-07, "loss": 0.469, "step": 26187 }, { "epoch": 0.9330031886278213, "grad_norm": 1.4297805855830494, "learning_rate": 2.343935927542329e-07, "loss": 0.4137, "step": 26188 }, { "epoch": 0.9330388157543154, "grad_norm": 1.6611048037805196, "learning_rate": 2.3414528160397798e-07, "loss": 0.5838, "step": 26189 }, { "epoch": 0.9330744428808094, "grad_norm": 1.2705694108623016, "learning_rate": 2.338971004929036e-07, "loss": 0.4619, "step": 26190 }, { "epoch": 0.9331100700073036, "grad_norm": 1.420305658589675, "learning_rate": 2.33649049424316e-07, "loss": 0.4171, "step": 26191 }, { "epoch": 0.9331456971337977, "grad_norm": 1.470849809590369, "learning_rate": 2.3340112840151696e-07, "loss": 0.4757, "step": 26192 }, { "epoch": 0.9331813242602918, "grad_norm": 1.4832184646518867, "learning_rate": 2.3315333742780942e-07, "loss": 0.4326, "step": 26193 }, { "epoch": 0.933216951386786, "grad_norm": 1.3736091502563919, "learning_rate": 2.3290567650649187e-07, "loss": 0.4005, "step": 26194 }, { "epoch": 0.93325257851328, "grad_norm": 1.7901012133660066, "learning_rate": 2.3265814564086165e-07, "loss": 0.5224, "step": 26195 }, { "epoch": 0.9332882056397741, "grad_norm": 1.2984390732081215, "learning_rate": 2.3241074483421498e-07, "loss": 0.3984, "step": 26196 }, { "epoch": 0.9333238327662683, "grad_norm": 1.4582989671998552, "learning_rate": 2.3216347408984707e-07, "loss": 0.435, "step": 26197 }, { "epoch": 0.9333594598927624, "grad_norm": 1.991360309261189, "learning_rate": 2.3191633341104859e-07, "loss": 0.4195, "step": 26198 }, { "epoch": 0.9333950870192564, "grad_norm": 1.3375411539186433, "learning_rate": 2.3166932280111242e-07, "loss": 0.3814, "step": 26199 }, { "epoch": 0.9334307141457506, "grad_norm": 1.2808632819831813, "learning_rate": 2.31422442263326e-07, "loss": 0.413, "step": 26200 }, { "epoch": 0.9334663412722447, "grad_norm": 2.4518110102942776, "learning_rate": 2.3117569180097778e-07, "loss": 0.5081, "step": 26201 }, { "epoch": 0.9335019683987388, "grad_norm": 1.4565638193829435, "learning_rate": 2.309290714173551e-07, "loss": 0.4859, "step": 26202 }, { "epoch": 0.933537595525233, "grad_norm": 1.2421919210415449, "learning_rate": 2.306825811157376e-07, "loss": 0.4261, "step": 26203 }, { "epoch": 0.933573222651727, "grad_norm": 1.509210373003284, "learning_rate": 2.3043622089941152e-07, "loss": 0.4179, "step": 26204 }, { "epoch": 0.9336088497782211, "grad_norm": 1.2979817941555747, "learning_rate": 2.3018999077165537e-07, "loss": 0.3609, "step": 26205 }, { "epoch": 0.9336444769047153, "grad_norm": 1.1442063300791154, "learning_rate": 2.2994389073574762e-07, "loss": 0.408, "step": 26206 }, { "epoch": 0.9336801040312094, "grad_norm": 1.5711360727548846, "learning_rate": 2.2969792079496567e-07, "loss": 0.5037, "step": 26207 }, { "epoch": 0.9337157311577035, "grad_norm": 1.3672165948855235, "learning_rate": 2.2945208095258575e-07, "loss": 0.3741, "step": 26208 }, { "epoch": 0.9337513582841976, "grad_norm": 1.3259590229503297, "learning_rate": 2.2920637121187972e-07, "loss": 0.4506, "step": 26209 }, { "epoch": 0.9337869854106917, "grad_norm": 1.2190422380572177, "learning_rate": 2.289607915761205e-07, "loss": 0.3732, "step": 26210 }, { "epoch": 0.9338226125371858, "grad_norm": 1.7278062243134884, "learning_rate": 2.287153420485788e-07, "loss": 0.4977, "step": 26211 }, { "epoch": 0.93385823966368, "grad_norm": 1.164617648953049, "learning_rate": 2.2847002263252093e-07, "loss": 0.441, "step": 26212 }, { "epoch": 0.933893866790174, "grad_norm": 1.5959150378304228, "learning_rate": 2.2822483333121537e-07, "loss": 0.4523, "step": 26213 }, { "epoch": 0.9339294939166681, "grad_norm": 1.2696580558177955, "learning_rate": 2.2797977414792505e-07, "loss": 0.3697, "step": 26214 }, { "epoch": 0.9339651210431623, "grad_norm": 1.4617171014101276, "learning_rate": 2.2773484508591515e-07, "loss": 0.4314, "step": 26215 }, { "epoch": 0.9340007481696564, "grad_norm": 1.4481856829101079, "learning_rate": 2.2749004614844638e-07, "loss": 0.3966, "step": 26216 }, { "epoch": 0.9340363752961505, "grad_norm": 1.2513306442297996, "learning_rate": 2.2724537733877838e-07, "loss": 0.4244, "step": 26217 }, { "epoch": 0.9340720024226447, "grad_norm": 1.881054100401116, "learning_rate": 2.2700083866016854e-07, "loss": 0.4616, "step": 26218 }, { "epoch": 0.9341076295491387, "grad_norm": 1.9466018705681358, "learning_rate": 2.2675643011587312e-07, "loss": 0.4572, "step": 26219 }, { "epoch": 0.9341432566756328, "grad_norm": 1.2091316953873223, "learning_rate": 2.2651215170914731e-07, "loss": 0.3745, "step": 26220 }, { "epoch": 0.934178883802127, "grad_norm": 1.1848680993554923, "learning_rate": 2.2626800344324295e-07, "loss": 0.3935, "step": 26221 }, { "epoch": 0.9342145109286211, "grad_norm": 1.5797170838101482, "learning_rate": 2.260239853214119e-07, "loss": 0.5758, "step": 26222 }, { "epoch": 0.9342501380551151, "grad_norm": 2.0724727940906096, "learning_rate": 2.2578009734690264e-07, "loss": 0.5495, "step": 26223 }, { "epoch": 0.9342857651816093, "grad_norm": 1.3575188933177285, "learning_rate": 2.2553633952296484e-07, "loss": 0.4354, "step": 26224 }, { "epoch": 0.9343213923081034, "grad_norm": 1.4063453735392852, "learning_rate": 2.252927118528403e-07, "loss": 0.5137, "step": 26225 }, { "epoch": 0.9343570194345975, "grad_norm": 1.4518846820838056, "learning_rate": 2.2504921433977755e-07, "loss": 0.4819, "step": 26226 }, { "epoch": 0.9343926465610917, "grad_norm": 1.3208837403714533, "learning_rate": 2.2480584698701513e-07, "loss": 0.4369, "step": 26227 }, { "epoch": 0.9344282736875857, "grad_norm": 2.223845852850768, "learning_rate": 2.24562609797796e-07, "loss": 0.5607, "step": 26228 }, { "epoch": 0.9344639008140798, "grad_norm": 1.3596545832968405, "learning_rate": 2.2431950277535753e-07, "loss": 0.428, "step": 26229 }, { "epoch": 0.934499527940574, "grad_norm": 1.4227007097949382, "learning_rate": 2.2407652592293938e-07, "loss": 0.4208, "step": 26230 }, { "epoch": 0.9345351550670681, "grad_norm": 1.3762650389981719, "learning_rate": 2.2383367924377454e-07, "loss": 0.3773, "step": 26231 }, { "epoch": 0.9345707821935622, "grad_norm": 1.2601331829318638, "learning_rate": 2.2359096274109705e-07, "loss": 0.3488, "step": 26232 }, { "epoch": 0.9346064093200563, "grad_norm": 1.3103282230376718, "learning_rate": 2.233483764181399e-07, "loss": 0.4051, "step": 26233 }, { "epoch": 0.9346420364465504, "grad_norm": 1.4334539714241212, "learning_rate": 2.231059202781316e-07, "loss": 0.4611, "step": 26234 }, { "epoch": 0.9346776635730445, "grad_norm": 1.5530105605945186, "learning_rate": 2.2286359432430182e-07, "loss": 0.4041, "step": 26235 }, { "epoch": 0.9347132906995387, "grad_norm": 1.4023895308360619, "learning_rate": 2.2262139855987796e-07, "loss": 0.4275, "step": 26236 }, { "epoch": 0.9347489178260328, "grad_norm": 2.339202170498241, "learning_rate": 2.2237933298808301e-07, "loss": 0.5001, "step": 26237 }, { "epoch": 0.9347845449525268, "grad_norm": 1.3680955520414326, "learning_rate": 2.2213739761214103e-07, "loss": 0.3678, "step": 26238 }, { "epoch": 0.934820172079021, "grad_norm": 1.283604990182913, "learning_rate": 2.2189559243527614e-07, "loss": 0.3928, "step": 26239 }, { "epoch": 0.9348557992055151, "grad_norm": 1.4882553055469352, "learning_rate": 2.2165391746070352e-07, "loss": 0.4447, "step": 26240 }, { "epoch": 0.9348914263320092, "grad_norm": 1.8727638566710745, "learning_rate": 2.2141237269164507e-07, "loss": 0.492, "step": 26241 }, { "epoch": 0.9349270534585034, "grad_norm": 1.1500506662349188, "learning_rate": 2.2117095813131485e-07, "loss": 0.3986, "step": 26242 }, { "epoch": 0.9349626805849974, "grad_norm": 1.2314117416559478, "learning_rate": 2.209296737829292e-07, "loss": 0.4288, "step": 26243 }, { "epoch": 0.9349983077114915, "grad_norm": 1.4319078178873441, "learning_rate": 2.2068851964969884e-07, "loss": 0.5095, "step": 26244 }, { "epoch": 0.9350339348379857, "grad_norm": 1.349375803959829, "learning_rate": 2.2044749573483683e-07, "loss": 0.5234, "step": 26245 }, { "epoch": 0.9350695619644798, "grad_norm": 1.2885375905967174, "learning_rate": 2.2020660204155276e-07, "loss": 0.4368, "step": 26246 }, { "epoch": 0.9351051890909738, "grad_norm": 2.4303039262975616, "learning_rate": 2.19965838573053e-07, "loss": 0.5342, "step": 26247 }, { "epoch": 0.935140816217468, "grad_norm": 1.2109079100628994, "learning_rate": 2.1972520533254383e-07, "loss": 0.3697, "step": 26248 }, { "epoch": 0.9351764433439621, "grad_norm": 1.2854226938779982, "learning_rate": 2.1948470232322937e-07, "loss": 0.3664, "step": 26249 }, { "epoch": 0.9352120704704562, "grad_norm": 1.4403648325025205, "learning_rate": 2.1924432954831265e-07, "loss": 0.4944, "step": 26250 }, { "epoch": 0.9352476975969504, "grad_norm": 1.25198723623767, "learning_rate": 2.190040870109933e-07, "loss": 0.4074, "step": 26251 }, { "epoch": 0.9352833247234444, "grad_norm": 1.545095397235113, "learning_rate": 2.1876397471447097e-07, "loss": 0.513, "step": 26252 }, { "epoch": 0.9353189518499385, "grad_norm": 1.8855372361882718, "learning_rate": 2.1852399266194312e-07, "loss": 0.4929, "step": 26253 }, { "epoch": 0.9353545789764327, "grad_norm": 1.2979932097450009, "learning_rate": 2.182841408566061e-07, "loss": 0.3974, "step": 26254 }, { "epoch": 0.9353902061029268, "grad_norm": 1.4308141559741012, "learning_rate": 2.1804441930165175e-07, "loss": 0.5283, "step": 26255 }, { "epoch": 0.9354258332294209, "grad_norm": 1.3942160995119588, "learning_rate": 2.1780482800027425e-07, "loss": 0.4734, "step": 26256 }, { "epoch": 0.935461460355915, "grad_norm": 1.2691764122368305, "learning_rate": 2.175653669556621e-07, "loss": 0.4008, "step": 26257 }, { "epoch": 0.9354970874824091, "grad_norm": 1.296982522225476, "learning_rate": 2.173260361710039e-07, "loss": 0.3878, "step": 26258 }, { "epoch": 0.9355327146089032, "grad_norm": 1.2029883618015387, "learning_rate": 2.170868356494882e-07, "loss": 0.3749, "step": 26259 }, { "epoch": 0.9355683417353974, "grad_norm": 1.385877225974282, "learning_rate": 2.1684776539429797e-07, "loss": 0.3423, "step": 26260 }, { "epoch": 0.9356039688618915, "grad_norm": 1.2523236088048106, "learning_rate": 2.166088254086196e-07, "loss": 0.445, "step": 26261 }, { "epoch": 0.9356395959883855, "grad_norm": 1.6603989771495355, "learning_rate": 2.1637001569563166e-07, "loss": 0.426, "step": 26262 }, { "epoch": 0.9356752231148797, "grad_norm": 1.4699616386596261, "learning_rate": 2.16131336258516e-07, "loss": 0.4726, "step": 26263 }, { "epoch": 0.9357108502413738, "grad_norm": 1.5435649187915534, "learning_rate": 2.1589278710045015e-07, "loss": 0.4439, "step": 26264 }, { "epoch": 0.9357464773678679, "grad_norm": 1.2105188880978044, "learning_rate": 2.156543682246104e-07, "loss": 0.3606, "step": 26265 }, { "epoch": 0.935782104494362, "grad_norm": 1.2723214948901083, "learning_rate": 2.1541607963416978e-07, "loss": 0.4061, "step": 26266 }, { "epoch": 0.9358177316208561, "grad_norm": 2.4212908447329515, "learning_rate": 2.1517792133230574e-07, "loss": 0.411, "step": 26267 }, { "epoch": 0.9358533587473502, "grad_norm": 1.9282325348009943, "learning_rate": 2.1493989332218468e-07, "loss": 0.4981, "step": 26268 }, { "epoch": 0.9358889858738444, "grad_norm": 1.2419347337510511, "learning_rate": 2.1470199560697957e-07, "loss": 0.3856, "step": 26269 }, { "epoch": 0.9359246130003385, "grad_norm": 1.2014267560524245, "learning_rate": 2.1446422818985678e-07, "loss": 0.3853, "step": 26270 }, { "epoch": 0.9359602401268325, "grad_norm": 1.5762569916376978, "learning_rate": 2.1422659107398158e-07, "loss": 0.4603, "step": 26271 }, { "epoch": 0.9359958672533267, "grad_norm": 1.2274415483163303, "learning_rate": 2.139890842625214e-07, "loss": 0.4037, "step": 26272 }, { "epoch": 0.9360314943798208, "grad_norm": 2.298725615698657, "learning_rate": 2.1375170775863375e-07, "loss": 0.4646, "step": 26273 }, { "epoch": 0.9360671215063149, "grad_norm": 1.6867486313496955, "learning_rate": 2.1351446156548382e-07, "loss": 0.4844, "step": 26274 }, { "epoch": 0.9361027486328091, "grad_norm": 1.2743371473539402, "learning_rate": 2.1327734568622916e-07, "loss": 0.4014, "step": 26275 }, { "epoch": 0.9361383757593031, "grad_norm": 2.0772336421017665, "learning_rate": 2.1304036012402719e-07, "loss": 0.5695, "step": 26276 }, { "epoch": 0.9361740028857972, "grad_norm": 1.2164145238340285, "learning_rate": 2.1280350488203317e-07, "loss": 0.3449, "step": 26277 }, { "epoch": 0.9362096300122914, "grad_norm": 1.3118319497134447, "learning_rate": 2.1256677996340124e-07, "loss": 0.4224, "step": 26278 }, { "epoch": 0.9362452571387855, "grad_norm": 1.197297139155592, "learning_rate": 2.1233018537128446e-07, "loss": 0.3846, "step": 26279 }, { "epoch": 0.9362808842652796, "grad_norm": 1.6589615975181649, "learning_rate": 2.120937211088314e-07, "loss": 0.5703, "step": 26280 }, { "epoch": 0.9363165113917737, "grad_norm": 2.060663700666887, "learning_rate": 2.118573871791918e-07, "loss": 0.5096, "step": 26281 }, { "epoch": 0.9363521385182678, "grad_norm": 1.6553720314400782, "learning_rate": 2.11621183585512e-07, "loss": 0.4541, "step": 26282 }, { "epoch": 0.9363877656447619, "grad_norm": 1.2945386491458868, "learning_rate": 2.1138511033093944e-07, "loss": 0.3917, "step": 26283 }, { "epoch": 0.9364233927712561, "grad_norm": 1.5209697273162845, "learning_rate": 2.1114916741861502e-07, "loss": 0.3518, "step": 26284 }, { "epoch": 0.9364590198977502, "grad_norm": 1.5908406351392406, "learning_rate": 2.1091335485168286e-07, "loss": 0.3793, "step": 26285 }, { "epoch": 0.9364946470242442, "grad_norm": 1.4290008957529101, "learning_rate": 2.1067767263327932e-07, "loss": 0.5972, "step": 26286 }, { "epoch": 0.9365302741507384, "grad_norm": 2.328721251899364, "learning_rate": 2.1044212076654746e-07, "loss": 0.4681, "step": 26287 }, { "epoch": 0.9365659012772325, "grad_norm": 1.3097091755628851, "learning_rate": 2.1020669925461923e-07, "loss": 0.3804, "step": 26288 }, { "epoch": 0.9366015284037266, "grad_norm": 1.9845663050530562, "learning_rate": 2.09971408100631e-07, "loss": 0.454, "step": 26289 }, { "epoch": 0.9366371555302208, "grad_norm": 1.3570871491943195, "learning_rate": 2.0973624730771803e-07, "loss": 0.423, "step": 26290 }, { "epoch": 0.9366727826567148, "grad_norm": 1.3941597797943102, "learning_rate": 2.095012168790078e-07, "loss": 0.4903, "step": 26291 }, { "epoch": 0.9367084097832089, "grad_norm": 1.4093517169061631, "learning_rate": 2.0926631681763455e-07, "loss": 0.3814, "step": 26292 }, { "epoch": 0.9367440369097031, "grad_norm": 1.6153700623576237, "learning_rate": 2.0903154712672236e-07, "loss": 0.5679, "step": 26293 }, { "epoch": 0.9367796640361972, "grad_norm": 1.4482939476598902, "learning_rate": 2.0879690780939876e-07, "loss": 0.3961, "step": 26294 }, { "epoch": 0.9368152911626912, "grad_norm": 1.3363927550697317, "learning_rate": 2.0856239886878793e-07, "loss": 0.4157, "step": 26295 }, { "epoch": 0.9368509182891854, "grad_norm": 1.4192688186640272, "learning_rate": 2.083280203080118e-07, "loss": 0.4342, "step": 26296 }, { "epoch": 0.9368865454156795, "grad_norm": 1.1459531606973177, "learning_rate": 2.0809377213019233e-07, "loss": 0.4057, "step": 26297 }, { "epoch": 0.9369221725421736, "grad_norm": 1.7679268509649502, "learning_rate": 2.0785965433845036e-07, "loss": 0.542, "step": 26298 }, { "epoch": 0.9369577996686678, "grad_norm": 1.3544870642051194, "learning_rate": 2.0762566693589892e-07, "loss": 0.3945, "step": 26299 }, { "epoch": 0.9369934267951618, "grad_norm": 1.151972458097989, "learning_rate": 2.0739180992565778e-07, "loss": 0.3698, "step": 26300 }, { "epoch": 0.9370290539216559, "grad_norm": 1.505036527159219, "learning_rate": 2.0715808331083887e-07, "loss": 0.4493, "step": 26301 }, { "epoch": 0.9370646810481501, "grad_norm": 1.405528395166641, "learning_rate": 2.0692448709455527e-07, "loss": 0.4625, "step": 26302 }, { "epoch": 0.9371003081746442, "grad_norm": 1.2939481158036528, "learning_rate": 2.066910212799156e-07, "loss": 0.3811, "step": 26303 }, { "epoch": 0.9371359353011383, "grad_norm": 1.7237010626285791, "learning_rate": 2.064576858700307e-07, "loss": 0.5251, "step": 26304 }, { "epoch": 0.9371715624276324, "grad_norm": 1.6747350122388642, "learning_rate": 2.0622448086800694e-07, "loss": 0.5085, "step": 26305 }, { "epoch": 0.9372071895541265, "grad_norm": 1.6462553129612207, "learning_rate": 2.059914062769497e-07, "loss": 0.6383, "step": 26306 }, { "epoch": 0.9372428166806206, "grad_norm": 1.4979785801878613, "learning_rate": 2.057584620999642e-07, "loss": 0.4683, "step": 26307 }, { "epoch": 0.9372784438071148, "grad_norm": 1.469026193009071, "learning_rate": 2.0552564834014797e-07, "loss": 0.5097, "step": 26308 }, { "epoch": 0.9373140709336089, "grad_norm": 1.328870683909295, "learning_rate": 2.0529296500060636e-07, "loss": 0.3997, "step": 26309 }, { "epoch": 0.9373496980601029, "grad_norm": 1.483875934802958, "learning_rate": 2.050604120844324e-07, "loss": 0.4542, "step": 26310 }, { "epoch": 0.9373853251865971, "grad_norm": 1.4512323715361437, "learning_rate": 2.0482798959472584e-07, "loss": 0.3825, "step": 26311 }, { "epoch": 0.9374209523130912, "grad_norm": 1.0971029855246444, "learning_rate": 2.0459569753458198e-07, "loss": 0.4034, "step": 26312 }, { "epoch": 0.9374565794395853, "grad_norm": 1.3679123214464273, "learning_rate": 2.043635359070928e-07, "loss": 0.4324, "step": 26313 }, { "epoch": 0.9374922065660795, "grad_norm": 1.362107875887183, "learning_rate": 2.0413150471534915e-07, "loss": 0.4003, "step": 26314 }, { "epoch": 0.9375278336925735, "grad_norm": 2.455575943637099, "learning_rate": 2.038996039624419e-07, "loss": 0.4502, "step": 26315 }, { "epoch": 0.9375634608190676, "grad_norm": 1.4070419234262022, "learning_rate": 2.036678336514586e-07, "loss": 0.4432, "step": 26316 }, { "epoch": 0.9375990879455618, "grad_norm": 1.8374943177746075, "learning_rate": 2.0343619378548564e-07, "loss": 0.5347, "step": 26317 }, { "epoch": 0.9376347150720559, "grad_norm": 1.7603962783857465, "learning_rate": 2.032046843676061e-07, "loss": 0.5247, "step": 26318 }, { "epoch": 0.9376703421985499, "grad_norm": 1.5453892982543957, "learning_rate": 2.029733054009042e-07, "loss": 0.5524, "step": 26319 }, { "epoch": 0.9377059693250441, "grad_norm": 1.2590906958398036, "learning_rate": 2.0274205688846082e-07, "loss": 0.417, "step": 26320 }, { "epoch": 0.9377415964515382, "grad_norm": 1.2886136938675, "learning_rate": 2.025109388333546e-07, "loss": 0.386, "step": 26321 }, { "epoch": 0.9377772235780323, "grad_norm": 1.3949900981476229, "learning_rate": 2.0227995123866418e-07, "loss": 0.5178, "step": 26322 }, { "epoch": 0.9378128507045265, "grad_norm": 1.5283905558391158, "learning_rate": 2.0204909410746265e-07, "loss": 0.4716, "step": 26323 }, { "epoch": 0.9378484778310205, "grad_norm": 1.6408224077356086, "learning_rate": 2.0181836744282758e-07, "loss": 0.5013, "step": 26324 }, { "epoch": 0.9378841049575146, "grad_norm": 1.3970689436389359, "learning_rate": 2.015877712478287e-07, "loss": 0.5082, "step": 26325 }, { "epoch": 0.9379197320840088, "grad_norm": 1.3749455485470923, "learning_rate": 2.013573055255369e-07, "loss": 0.4478, "step": 26326 }, { "epoch": 0.9379553592105029, "grad_norm": 2.0889831874398666, "learning_rate": 2.0112697027902083e-07, "loss": 0.4126, "step": 26327 }, { "epoch": 0.937990986336997, "grad_norm": 1.7403043316225497, "learning_rate": 2.0089676551134918e-07, "loss": 0.4526, "step": 26328 }, { "epoch": 0.9380266134634911, "grad_norm": 1.2325911292120926, "learning_rate": 2.0066669122558613e-07, "loss": 0.4378, "step": 26329 }, { "epoch": 0.9380622405899852, "grad_norm": 1.4465852976411095, "learning_rate": 2.004367474247948e-07, "loss": 0.4625, "step": 26330 }, { "epoch": 0.9380978677164793, "grad_norm": 1.4501849546588856, "learning_rate": 2.002069341120394e-07, "loss": 0.3565, "step": 26331 }, { "epoch": 0.9381334948429735, "grad_norm": 1.307182047496881, "learning_rate": 1.9997725129037637e-07, "loss": 0.3731, "step": 26332 }, { "epoch": 0.9381691219694676, "grad_norm": 1.2000081709160937, "learning_rate": 1.9974769896286662e-07, "loss": 0.3814, "step": 26333 }, { "epoch": 0.9382047490959616, "grad_norm": 1.5191933816013565, "learning_rate": 1.9951827713256656e-07, "loss": 0.5168, "step": 26334 }, { "epoch": 0.9382403762224558, "grad_norm": 1.5968938330566933, "learning_rate": 1.9928898580253153e-07, "loss": 0.4682, "step": 26335 }, { "epoch": 0.9382760033489499, "grad_norm": 1.1308752615904076, "learning_rate": 1.9905982497581246e-07, "loss": 0.4068, "step": 26336 }, { "epoch": 0.938311630475444, "grad_norm": 2.012555401555208, "learning_rate": 1.988307946554624e-07, "loss": 0.4593, "step": 26337 }, { "epoch": 0.9383472576019382, "grad_norm": 1.291090478474707, "learning_rate": 1.986018948445323e-07, "loss": 0.3548, "step": 26338 }, { "epoch": 0.9383828847284322, "grad_norm": 1.2696004508002627, "learning_rate": 1.9837312554606748e-07, "loss": 0.4198, "step": 26339 }, { "epoch": 0.9384185118549263, "grad_norm": 1.39102730050066, "learning_rate": 1.9814448676311548e-07, "loss": 0.4914, "step": 26340 }, { "epoch": 0.9384541389814205, "grad_norm": 1.7490941723601796, "learning_rate": 1.9791597849872057e-07, "loss": 0.4859, "step": 26341 }, { "epoch": 0.9384897661079146, "grad_norm": 1.700963124649915, "learning_rate": 1.9768760075592698e-07, "loss": 0.4721, "step": 26342 }, { "epoch": 0.9385253932344086, "grad_norm": 1.2455564221452091, "learning_rate": 1.9745935353777222e-07, "loss": 0.3882, "step": 26343 }, { "epoch": 0.9385610203609028, "grad_norm": 1.5792465560563447, "learning_rate": 1.972312368472995e-07, "loss": 0.5107, "step": 26344 }, { "epoch": 0.9385966474873969, "grad_norm": 1.4156474613163696, "learning_rate": 1.9700325068754412e-07, "loss": 0.5117, "step": 26345 }, { "epoch": 0.938632274613891, "grad_norm": 1.506410415724756, "learning_rate": 1.9677539506154363e-07, "loss": 0.4316, "step": 26346 }, { "epoch": 0.9386679017403852, "grad_norm": 1.6178815208520587, "learning_rate": 1.9654766997232898e-07, "loss": 0.4233, "step": 26347 }, { "epoch": 0.9387035288668792, "grad_norm": 1.3772712599785226, "learning_rate": 1.963200754229344e-07, "loss": 0.3988, "step": 26348 }, { "epoch": 0.9387391559933733, "grad_norm": 1.642151049824022, "learning_rate": 1.960926114163908e-07, "loss": 0.471, "step": 26349 }, { "epoch": 0.9387747831198675, "grad_norm": 1.3509275797357618, "learning_rate": 1.9586527795572574e-07, "loss": 0.3689, "step": 26350 }, { "epoch": 0.9388104102463616, "grad_norm": 1.294607620324575, "learning_rate": 1.9563807504396903e-07, "loss": 0.4184, "step": 26351 }, { "epoch": 0.9388460373728557, "grad_norm": 1.4454107059118124, "learning_rate": 1.954110026841427e-07, "loss": 0.4001, "step": 26352 }, { "epoch": 0.9388816644993498, "grad_norm": 1.6681336093075945, "learning_rate": 1.951840608792732e-07, "loss": 0.5348, "step": 26353 }, { "epoch": 0.9389172916258439, "grad_norm": 1.3087901110493343, "learning_rate": 1.9495724963237928e-07, "loss": 0.4107, "step": 26354 }, { "epoch": 0.938952918752338, "grad_norm": 1.1952367735504252, "learning_rate": 1.94730568946484e-07, "loss": 0.3631, "step": 26355 }, { "epoch": 0.9389885458788322, "grad_norm": 1.272388911593981, "learning_rate": 1.945040188246039e-07, "loss": 0.4209, "step": 26356 }, { "epoch": 0.9390241730053263, "grad_norm": 1.6756731923871386, "learning_rate": 1.9427759926975764e-07, "loss": 0.562, "step": 26357 }, { "epoch": 0.9390598001318203, "grad_norm": 1.1693900192184161, "learning_rate": 1.9405131028495838e-07, "loss": 0.3639, "step": 26358 }, { "epoch": 0.9390954272583145, "grad_norm": 1.6355949917189174, "learning_rate": 1.9382515187322037e-07, "loss": 0.4744, "step": 26359 }, { "epoch": 0.9391310543848086, "grad_norm": 1.6120008383075257, "learning_rate": 1.9359912403755343e-07, "loss": 0.5152, "step": 26360 }, { "epoch": 0.9391666815113027, "grad_norm": 1.4948950055795909, "learning_rate": 1.9337322678096847e-07, "loss": 0.5041, "step": 26361 }, { "epoch": 0.9392023086377969, "grad_norm": 2.1187387732247056, "learning_rate": 1.9314746010647533e-07, "loss": 0.5271, "step": 26362 }, { "epoch": 0.9392379357642909, "grad_norm": 1.6964546282832584, "learning_rate": 1.9292182401707603e-07, "loss": 0.5895, "step": 26363 }, { "epoch": 0.939273562890785, "grad_norm": 1.6314851594604438, "learning_rate": 1.926963185157782e-07, "loss": 0.5476, "step": 26364 }, { "epoch": 0.9393091900172792, "grad_norm": 1.892192414417458, "learning_rate": 1.9247094360558383e-07, "loss": 0.5388, "step": 26365 }, { "epoch": 0.9393448171437733, "grad_norm": 1.2527585769587464, "learning_rate": 1.9224569928949498e-07, "loss": 0.3802, "step": 26366 }, { "epoch": 0.9393804442702673, "grad_norm": 1.4064972223430137, "learning_rate": 1.920205855705082e-07, "loss": 0.447, "step": 26367 }, { "epoch": 0.9394160713967615, "grad_norm": 2.19670351659351, "learning_rate": 1.9179560245162433e-07, "loss": 0.6642, "step": 26368 }, { "epoch": 0.9394516985232556, "grad_norm": 1.5607819252644848, "learning_rate": 1.9157074993583546e-07, "loss": 0.4193, "step": 26369 }, { "epoch": 0.9394873256497497, "grad_norm": 1.2222000165976787, "learning_rate": 1.9134602802613922e-07, "loss": 0.4027, "step": 26370 }, { "epoch": 0.9395229527762439, "grad_norm": 1.6367479658412434, "learning_rate": 1.911214367255254e-07, "loss": 0.4877, "step": 26371 }, { "epoch": 0.939558579902738, "grad_norm": 2.4122385435386913, "learning_rate": 1.908969760369861e-07, "loss": 0.5854, "step": 26372 }, { "epoch": 0.939594207029232, "grad_norm": 1.6501450093228427, "learning_rate": 1.906726459635111e-07, "loss": 0.5155, "step": 26373 }, { "epoch": 0.9396298341557262, "grad_norm": 1.2451031749491392, "learning_rate": 1.9044844650808468e-07, "loss": 0.437, "step": 26374 }, { "epoch": 0.9396654612822203, "grad_norm": 1.271623067733289, "learning_rate": 1.902243776736945e-07, "loss": 0.438, "step": 26375 }, { "epoch": 0.9397010884087144, "grad_norm": 1.3757780529149963, "learning_rate": 1.900004394633226e-07, "loss": 0.4123, "step": 26376 }, { "epoch": 0.9397367155352085, "grad_norm": 1.213366008256077, "learning_rate": 1.897766318799521e-07, "loss": 0.4111, "step": 26377 }, { "epoch": 0.9397723426617026, "grad_norm": 1.2494686593999758, "learning_rate": 1.8955295492656068e-07, "loss": 0.4166, "step": 26378 }, { "epoch": 0.9398079697881967, "grad_norm": 1.4332925473558813, "learning_rate": 1.8932940860613147e-07, "loss": 0.4234, "step": 26379 }, { "epoch": 0.9398435969146909, "grad_norm": 1.3810142404301218, "learning_rate": 1.8910599292163767e-07, "loss": 0.3989, "step": 26380 }, { "epoch": 0.939879224041185, "grad_norm": 1.489125051541775, "learning_rate": 1.8888270787605466e-07, "loss": 0.5311, "step": 26381 }, { "epoch": 0.939914851167679, "grad_norm": 2.096638366093842, "learning_rate": 1.8865955347235564e-07, "loss": 0.4988, "step": 26382 }, { "epoch": 0.9399504782941732, "grad_norm": 1.1946193790084563, "learning_rate": 1.884365297135138e-07, "loss": 0.3867, "step": 26383 }, { "epoch": 0.9399861054206673, "grad_norm": 2.165640125885208, "learning_rate": 1.882136366024956e-07, "loss": 0.4761, "step": 26384 }, { "epoch": 0.9400217325471614, "grad_norm": 1.1613869892865452, "learning_rate": 1.8799087414227202e-07, "loss": 0.3677, "step": 26385 }, { "epoch": 0.9400573596736556, "grad_norm": 1.2888887382216914, "learning_rate": 1.8776824233580737e-07, "loss": 0.4081, "step": 26386 }, { "epoch": 0.9400929868001496, "grad_norm": 2.0389352182955545, "learning_rate": 1.8754574118606706e-07, "loss": 0.5933, "step": 26387 }, { "epoch": 0.9401286139266437, "grad_norm": 1.7954292275208088, "learning_rate": 1.8732337069601426e-07, "loss": 0.519, "step": 26388 }, { "epoch": 0.9401642410531379, "grad_norm": 1.4360274194549436, "learning_rate": 1.871011308686088e-07, "loss": 0.3886, "step": 26389 }, { "epoch": 0.940199868179632, "grad_norm": 1.8360205135135343, "learning_rate": 1.868790217068117e-07, "loss": 0.5942, "step": 26390 }, { "epoch": 0.940235495306126, "grad_norm": 1.4360538058567485, "learning_rate": 1.8665704321357835e-07, "loss": 0.4498, "step": 26391 }, { "epoch": 0.9402711224326202, "grad_norm": 1.3241395485582421, "learning_rate": 1.8643519539186527e-07, "loss": 0.4137, "step": 26392 }, { "epoch": 0.9403067495591143, "grad_norm": 1.408375859436557, "learning_rate": 1.8621347824462787e-07, "loss": 0.3896, "step": 26393 }, { "epoch": 0.9403423766856084, "grad_norm": 1.438919257243901, "learning_rate": 1.8599189177481713e-07, "loss": 0.4098, "step": 26394 }, { "epoch": 0.9403780038121026, "grad_norm": 1.3959073859378455, "learning_rate": 1.8577043598538402e-07, "loss": 0.4246, "step": 26395 }, { "epoch": 0.9404136309385966, "grad_norm": 1.55927556887919, "learning_rate": 1.8554911087927617e-07, "loss": 0.4108, "step": 26396 }, { "epoch": 0.9404492580650907, "grad_norm": 1.3511475679423652, "learning_rate": 1.8532791645944347e-07, "loss": 0.3784, "step": 26397 }, { "epoch": 0.9404848851915849, "grad_norm": 1.5886680857633735, "learning_rate": 1.851068527288291e-07, "loss": 0.4615, "step": 26398 }, { "epoch": 0.940520512318079, "grad_norm": 1.744335184874433, "learning_rate": 1.8488591969037738e-07, "loss": 0.5081, "step": 26399 }, { "epoch": 0.9405561394445731, "grad_norm": 1.319295252271651, "learning_rate": 1.8466511734702929e-07, "loss": 0.4615, "step": 26400 }, { "epoch": 0.9405917665710672, "grad_norm": 1.2655687383006173, "learning_rate": 1.8444444570172583e-07, "loss": 0.3948, "step": 26401 }, { "epoch": 0.9406273936975613, "grad_norm": 1.4777676049425035, "learning_rate": 1.842239047574046e-07, "loss": 0.4579, "step": 26402 }, { "epoch": 0.9406630208240554, "grad_norm": 1.492703129957755, "learning_rate": 1.8400349451700438e-07, "loss": 0.3907, "step": 26403 }, { "epoch": 0.9406986479505496, "grad_norm": 1.1393727964954523, "learning_rate": 1.837832149834573e-07, "loss": 0.4062, "step": 26404 }, { "epoch": 0.9407342750770437, "grad_norm": 1.736980922366669, "learning_rate": 1.8356306615969876e-07, "loss": 0.5167, "step": 26405 }, { "epoch": 0.9407699022035377, "grad_norm": 1.2373744557524975, "learning_rate": 1.8334304804865865e-07, "loss": 0.412, "step": 26406 }, { "epoch": 0.9408055293300319, "grad_norm": 1.3002508163654471, "learning_rate": 1.8312316065326796e-07, "loss": 0.4218, "step": 26407 }, { "epoch": 0.940841156456526, "grad_norm": 1.4864958253805638, "learning_rate": 1.8290340397645324e-07, "loss": 0.3995, "step": 26408 }, { "epoch": 0.9408767835830201, "grad_norm": 1.6256368936899757, "learning_rate": 1.82683778021141e-07, "loss": 0.4273, "step": 26409 }, { "epoch": 0.9409124107095143, "grad_norm": 1.3191727838955976, "learning_rate": 1.8246428279025784e-07, "loss": 0.4298, "step": 26410 }, { "epoch": 0.9409480378360083, "grad_norm": 1.8486565350616924, "learning_rate": 1.8224491828672364e-07, "loss": 0.5236, "step": 26411 }, { "epoch": 0.9409836649625024, "grad_norm": 1.576099654209055, "learning_rate": 1.820256845134616e-07, "loss": 0.4315, "step": 26412 }, { "epoch": 0.9410192920889966, "grad_norm": 1.2431461656185727, "learning_rate": 1.8180658147338826e-07, "loss": 0.377, "step": 26413 }, { "epoch": 0.9410549192154907, "grad_norm": 1.5436833985116603, "learning_rate": 1.8158760916942465e-07, "loss": 0.5365, "step": 26414 }, { "epoch": 0.9410905463419847, "grad_norm": 1.7305534201018302, "learning_rate": 1.8136876760448286e-07, "loss": 0.5813, "step": 26415 }, { "epoch": 0.9411261734684789, "grad_norm": 1.4427828154256621, "learning_rate": 1.8115005678147945e-07, "loss": 0.4976, "step": 26416 }, { "epoch": 0.941161800594973, "grad_norm": 1.2940408300739956, "learning_rate": 1.8093147670332544e-07, "loss": 0.3749, "step": 26417 }, { "epoch": 0.9411974277214671, "grad_norm": 1.2661613752678924, "learning_rate": 1.8071302737293294e-07, "loss": 0.3851, "step": 26418 }, { "epoch": 0.9412330548479613, "grad_norm": 1.4606938682973056, "learning_rate": 1.8049470879320962e-07, "loss": 0.4337, "step": 26419 }, { "epoch": 0.9412686819744553, "grad_norm": 1.252876974896675, "learning_rate": 1.8027652096706204e-07, "loss": 0.3706, "step": 26420 }, { "epoch": 0.9413043091009494, "grad_norm": 1.5430765753643543, "learning_rate": 1.800584638973979e-07, "loss": 0.4816, "step": 26421 }, { "epoch": 0.9413399362274436, "grad_norm": 1.6281719531127639, "learning_rate": 1.7984053758711707e-07, "loss": 0.4947, "step": 26422 }, { "epoch": 0.9413755633539377, "grad_norm": 1.4503888459441632, "learning_rate": 1.7962274203912388e-07, "loss": 0.5065, "step": 26423 }, { "epoch": 0.9414111904804318, "grad_norm": 1.218741498136827, "learning_rate": 1.7940507725631717e-07, "loss": 0.3912, "step": 26424 }, { "epoch": 0.941446817606926, "grad_norm": 1.4598706857503057, "learning_rate": 1.7918754324159792e-07, "loss": 0.4995, "step": 26425 }, { "epoch": 0.94148244473342, "grad_norm": 2.3120996734816517, "learning_rate": 1.789701399978594e-07, "loss": 0.4614, "step": 26426 }, { "epoch": 0.9415180718599141, "grad_norm": 1.3082211871015903, "learning_rate": 1.7875286752799926e-07, "loss": 0.4441, "step": 26427 }, { "epoch": 0.9415536989864083, "grad_norm": 1.1821492100658557, "learning_rate": 1.7853572583490853e-07, "loss": 0.3846, "step": 26428 }, { "epoch": 0.9415893261129024, "grad_norm": 1.825822248244254, "learning_rate": 1.7831871492147935e-07, "loss": 0.4548, "step": 26429 }, { "epoch": 0.9416249532393964, "grad_norm": 1.4917125289660003, "learning_rate": 1.781018347906005e-07, "loss": 0.5966, "step": 26430 }, { "epoch": 0.9416605803658906, "grad_norm": 1.395225894188698, "learning_rate": 1.7788508544516192e-07, "loss": 0.3599, "step": 26431 }, { "epoch": 0.9416962074923847, "grad_norm": 1.48109837749759, "learning_rate": 1.7766846688804906e-07, "loss": 0.4773, "step": 26432 }, { "epoch": 0.9417318346188788, "grad_norm": 1.6179515220151746, "learning_rate": 1.774519791221452e-07, "loss": 0.4939, "step": 26433 }, { "epoch": 0.941767461745373, "grad_norm": 1.465653284494475, "learning_rate": 1.7723562215033463e-07, "loss": 0.4793, "step": 26434 }, { "epoch": 0.941803088871867, "grad_norm": 1.3579858654768213, "learning_rate": 1.7701939597549622e-07, "loss": 0.4211, "step": 26435 }, { "epoch": 0.9418387159983611, "grad_norm": 1.385610302568747, "learning_rate": 1.76803300600511e-07, "loss": 0.3766, "step": 26436 }, { "epoch": 0.9418743431248553, "grad_norm": 1.1506921641817591, "learning_rate": 1.7658733602825661e-07, "loss": 0.3833, "step": 26437 }, { "epoch": 0.9419099702513494, "grad_norm": 2.4048589448850297, "learning_rate": 1.7637150226160637e-07, "loss": 0.4111, "step": 26438 }, { "epoch": 0.9419455973778434, "grad_norm": 1.389718559789061, "learning_rate": 1.7615579930343684e-07, "loss": 0.4161, "step": 26439 }, { "epoch": 0.9419812245043376, "grad_norm": 1.7070511387967848, "learning_rate": 1.7594022715661907e-07, "loss": 0.4751, "step": 26440 }, { "epoch": 0.9420168516308317, "grad_norm": 1.3346954369013897, "learning_rate": 1.7572478582402407e-07, "loss": 0.3622, "step": 26441 }, { "epoch": 0.9420524787573258, "grad_norm": 1.2985711379803724, "learning_rate": 1.755094753085196e-07, "loss": 0.4038, "step": 26442 }, { "epoch": 0.94208810588382, "grad_norm": 1.4130782438874645, "learning_rate": 1.752942956129744e-07, "loss": 0.4024, "step": 26443 }, { "epoch": 0.942123733010314, "grad_norm": 1.778315207205326, "learning_rate": 1.750792467402529e-07, "loss": 0.5756, "step": 26444 }, { "epoch": 0.9421593601368081, "grad_norm": 1.8593213231897645, "learning_rate": 1.7486432869321724e-07, "loss": 0.5019, "step": 26445 }, { "epoch": 0.9421949872633023, "grad_norm": 1.313422944686627, "learning_rate": 1.7464954147473067e-07, "loss": 0.4287, "step": 26446 }, { "epoch": 0.9422306143897964, "grad_norm": 1.3789433039957704, "learning_rate": 1.7443488508765316e-07, "loss": 0.433, "step": 26447 }, { "epoch": 0.9422662415162905, "grad_norm": 1.1286492621116593, "learning_rate": 1.742203595348435e-07, "loss": 0.4135, "step": 26448 }, { "epoch": 0.9423018686427846, "grad_norm": 1.642616360110738, "learning_rate": 1.740059648191572e-07, "loss": 0.4086, "step": 26449 }, { "epoch": 0.9423374957692787, "grad_norm": 1.2438300941972058, "learning_rate": 1.7379170094344867e-07, "loss": 0.4149, "step": 26450 }, { "epoch": 0.9423731228957728, "grad_norm": 1.2884665303062675, "learning_rate": 1.7357756791057335e-07, "loss": 0.4022, "step": 26451 }, { "epoch": 0.942408750022267, "grad_norm": 1.1765362980485607, "learning_rate": 1.7336356572337897e-07, "loss": 0.3984, "step": 26452 }, { "epoch": 0.9424443771487611, "grad_norm": 1.3723917840407862, "learning_rate": 1.7314969438471775e-07, "loss": 0.4057, "step": 26453 }, { "epoch": 0.9424800042752551, "grad_norm": 1.2347962669093142, "learning_rate": 1.7293595389743733e-07, "loss": 0.4508, "step": 26454 }, { "epoch": 0.9425156314017493, "grad_norm": 1.33053121996979, "learning_rate": 1.727223442643833e-07, "loss": 0.4237, "step": 26455 }, { "epoch": 0.9425512585282434, "grad_norm": 1.733669212302085, "learning_rate": 1.7250886548839995e-07, "loss": 0.4939, "step": 26456 }, { "epoch": 0.9425868856547375, "grad_norm": 1.3937205908021433, "learning_rate": 1.7229551757233065e-07, "loss": 0.4403, "step": 26457 }, { "epoch": 0.9426225127812317, "grad_norm": 1.6677344579198112, "learning_rate": 1.720823005190153e-07, "loss": 0.4808, "step": 26458 }, { "epoch": 0.9426581399077257, "grad_norm": 1.6164625194972952, "learning_rate": 1.7186921433129277e-07, "loss": 0.5553, "step": 26459 }, { "epoch": 0.9426937670342198, "grad_norm": 1.293332385510007, "learning_rate": 1.716562590120019e-07, "loss": 0.4199, "step": 26460 }, { "epoch": 0.942729394160714, "grad_norm": 1.8633340516396957, "learning_rate": 1.7144343456397706e-07, "loss": 0.5486, "step": 26461 }, { "epoch": 0.9427650212872081, "grad_norm": 1.3486074738588683, "learning_rate": 1.7123074099005376e-07, "loss": 0.4229, "step": 26462 }, { "epoch": 0.9428006484137021, "grad_norm": 1.1835441054495672, "learning_rate": 1.71018178293062e-07, "loss": 0.4125, "step": 26463 }, { "epoch": 0.9428362755401963, "grad_norm": 1.1892001865522495, "learning_rate": 1.708057464758328e-07, "loss": 0.3937, "step": 26464 }, { "epoch": 0.9428719026666904, "grad_norm": 1.391264245776674, "learning_rate": 1.7059344554119616e-07, "loss": 0.3863, "step": 26465 }, { "epoch": 0.9429075297931845, "grad_norm": 2.6248024022318655, "learning_rate": 1.7038127549197758e-07, "loss": 0.3825, "step": 26466 }, { "epoch": 0.9429431569196787, "grad_norm": 1.8047683195354927, "learning_rate": 1.701692363310037e-07, "loss": 0.5228, "step": 26467 }, { "epoch": 0.9429787840461727, "grad_norm": 1.3664080046811733, "learning_rate": 1.6995732806109555e-07, "loss": 0.4497, "step": 26468 }, { "epoch": 0.9430144111726668, "grad_norm": 1.7915449068521094, "learning_rate": 1.697455506850776e-07, "loss": 0.441, "step": 26469 }, { "epoch": 0.943050038299161, "grad_norm": 1.6582853918926383, "learning_rate": 1.6953390420576864e-07, "loss": 0.5517, "step": 26470 }, { "epoch": 0.9430856654256551, "grad_norm": 1.2849457287634778, "learning_rate": 1.6932238862598648e-07, "loss": 0.3429, "step": 26471 }, { "epoch": 0.9431212925521492, "grad_norm": 1.3575077109829865, "learning_rate": 1.6911100394854772e-07, "loss": 0.4044, "step": 26472 }, { "epoch": 0.9431569196786433, "grad_norm": 1.249533373043684, "learning_rate": 1.6889975017626902e-07, "loss": 0.3753, "step": 26473 }, { "epoch": 0.9431925468051374, "grad_norm": 1.4392504035579814, "learning_rate": 1.6868862731195922e-07, "loss": 0.4005, "step": 26474 }, { "epoch": 0.9432281739316315, "grad_norm": 1.9295919092137002, "learning_rate": 1.6847763535843275e-07, "loss": 0.5323, "step": 26475 }, { "epoch": 0.9432638010581257, "grad_norm": 1.2811789139923064, "learning_rate": 1.6826677431849848e-07, "loss": 0.3659, "step": 26476 }, { "epoch": 0.9432994281846198, "grad_norm": 1.585598076472987, "learning_rate": 1.6805604419496412e-07, "loss": 0.5477, "step": 26477 }, { "epoch": 0.9433350553111138, "grad_norm": 1.5840660279802352, "learning_rate": 1.6784544499063637e-07, "loss": 0.5514, "step": 26478 }, { "epoch": 0.943370682437608, "grad_norm": 1.2774574534985617, "learning_rate": 1.676349767083174e-07, "loss": 0.435, "step": 26479 }, { "epoch": 0.9434063095641021, "grad_norm": 1.5844961495792764, "learning_rate": 1.6742463935081277e-07, "loss": 0.5791, "step": 26480 }, { "epoch": 0.9434419366905962, "grad_norm": 1.2660155061817664, "learning_rate": 1.6721443292092022e-07, "loss": 0.3703, "step": 26481 }, { "epoch": 0.9434775638170904, "grad_norm": 1.2438388881438356, "learning_rate": 1.6700435742144082e-07, "loss": 0.3528, "step": 26482 }, { "epoch": 0.9435131909435844, "grad_norm": 1.2223232843527714, "learning_rate": 1.6679441285517128e-07, "loss": 0.4213, "step": 26483 }, { "epoch": 0.9435488180700785, "grad_norm": 1.8180584727693774, "learning_rate": 1.665845992249071e-07, "loss": 0.4506, "step": 26484 }, { "epoch": 0.9435844451965727, "grad_norm": 1.737340765942341, "learning_rate": 1.6637491653344273e-07, "loss": 0.5348, "step": 26485 }, { "epoch": 0.9436200723230668, "grad_norm": 1.5380134007779134, "learning_rate": 1.6616536478356927e-07, "loss": 0.5289, "step": 26486 }, { "epoch": 0.9436556994495608, "grad_norm": 1.6746416390477867, "learning_rate": 1.6595594397807669e-07, "loss": 0.5101, "step": 26487 }, { "epoch": 0.943691326576055, "grad_norm": 1.1844367867041938, "learning_rate": 1.6574665411975389e-07, "loss": 0.453, "step": 26488 }, { "epoch": 0.9437269537025491, "grad_norm": 1.2085350941554207, "learning_rate": 1.6553749521138973e-07, "loss": 0.4139, "step": 26489 }, { "epoch": 0.9437625808290432, "grad_norm": 1.3246900332767892, "learning_rate": 1.6532846725576645e-07, "loss": 0.4149, "step": 26490 }, { "epoch": 0.9437982079555374, "grad_norm": 1.3684246261094928, "learning_rate": 1.651195702556696e-07, "loss": 0.4522, "step": 26491 }, { "epoch": 0.9438338350820314, "grad_norm": 1.601945666489209, "learning_rate": 1.6491080421387917e-07, "loss": 0.5053, "step": 26492 }, { "epoch": 0.9438694622085255, "grad_norm": 1.4073175030337306, "learning_rate": 1.6470216913317628e-07, "loss": 0.3666, "step": 26493 }, { "epoch": 0.9439050893350197, "grad_norm": 1.381858961577423, "learning_rate": 1.6449366501633756e-07, "loss": 0.4833, "step": 26494 }, { "epoch": 0.9439407164615138, "grad_norm": 1.5404596876451266, "learning_rate": 1.6428529186614195e-07, "loss": 0.4168, "step": 26495 }, { "epoch": 0.9439763435880079, "grad_norm": 1.2298718760871625, "learning_rate": 1.6407704968536052e-07, "loss": 0.397, "step": 26496 }, { "epoch": 0.944011970714502, "grad_norm": 1.2156610356833177, "learning_rate": 1.638689384767689e-07, "loss": 0.396, "step": 26497 }, { "epoch": 0.9440475978409961, "grad_norm": 1.431879899669285, "learning_rate": 1.63660958243137e-07, "loss": 0.5109, "step": 26498 }, { "epoch": 0.9440832249674902, "grad_norm": 1.3349377667397586, "learning_rate": 1.6345310898723487e-07, "loss": 0.4151, "step": 26499 }, { "epoch": 0.9441188520939844, "grad_norm": 1.5818372576876472, "learning_rate": 1.6324539071183033e-07, "loss": 0.5, "step": 26500 }, { "epoch": 0.9441544792204785, "grad_norm": 1.5279227234211474, "learning_rate": 1.630378034196889e-07, "loss": 0.4879, "step": 26501 }, { "epoch": 0.9441901063469725, "grad_norm": 1.6141293839885487, "learning_rate": 1.6283034711357505e-07, "loss": 0.528, "step": 26502 }, { "epoch": 0.9442257334734667, "grad_norm": 1.3373305935099349, "learning_rate": 1.626230217962499e-07, "loss": 0.4193, "step": 26503 }, { "epoch": 0.9442613605999608, "grad_norm": 1.341035764249931, "learning_rate": 1.624158274704757e-07, "loss": 0.4151, "step": 26504 }, { "epoch": 0.9442969877264549, "grad_norm": 1.2220997844975583, "learning_rate": 1.6220876413901021e-07, "loss": 0.3677, "step": 26505 }, { "epoch": 0.9443326148529491, "grad_norm": 1.2294208897729046, "learning_rate": 1.6200183180461237e-07, "loss": 0.3924, "step": 26506 }, { "epoch": 0.9443682419794431, "grad_norm": 1.3163696990877614, "learning_rate": 1.6179503047003553e-07, "loss": 0.4102, "step": 26507 }, { "epoch": 0.9444038691059372, "grad_norm": 1.6925462198591665, "learning_rate": 1.6158836013803525e-07, "loss": 0.4697, "step": 26508 }, { "epoch": 0.9444394962324314, "grad_norm": 1.2586139080436483, "learning_rate": 1.6138182081136155e-07, "loss": 0.4037, "step": 26509 }, { "epoch": 0.9444751233589255, "grad_norm": 2.1093913492536225, "learning_rate": 1.6117541249276557e-07, "loss": 0.5274, "step": 26510 }, { "epoch": 0.9445107504854195, "grad_norm": 1.2966311845997582, "learning_rate": 1.6096913518499735e-07, "loss": 0.4114, "step": 26511 }, { "epoch": 0.9445463776119137, "grad_norm": 1.4987372649921518, "learning_rate": 1.6076298889080134e-07, "loss": 0.4482, "step": 26512 }, { "epoch": 0.9445820047384078, "grad_norm": 1.4623853073171507, "learning_rate": 1.605569736129231e-07, "loss": 0.5311, "step": 26513 }, { "epoch": 0.9446176318649019, "grad_norm": 1.3526380468333874, "learning_rate": 1.6035108935410605e-07, "loss": 0.3886, "step": 26514 }, { "epoch": 0.9446532589913961, "grad_norm": 1.801014693245571, "learning_rate": 1.601453361170924e-07, "loss": 0.6058, "step": 26515 }, { "epoch": 0.9446888861178901, "grad_norm": 1.2318474918328894, "learning_rate": 1.5993971390461994e-07, "loss": 0.4041, "step": 26516 }, { "epoch": 0.9447245132443842, "grad_norm": 1.7167584377260023, "learning_rate": 1.5973422271942985e-07, "loss": 0.5326, "step": 26517 }, { "epoch": 0.9447601403708784, "grad_norm": 1.4250949321739679, "learning_rate": 1.5952886256425547e-07, "loss": 0.4867, "step": 26518 }, { "epoch": 0.9447957674973725, "grad_norm": 1.4922382063117137, "learning_rate": 1.5932363344183245e-07, "loss": 0.4221, "step": 26519 }, { "epoch": 0.9448313946238666, "grad_norm": 1.3120404877278187, "learning_rate": 1.5911853535489298e-07, "loss": 0.3717, "step": 26520 }, { "epoch": 0.9448670217503607, "grad_norm": 1.6072209818855883, "learning_rate": 1.5891356830616933e-07, "loss": 0.5527, "step": 26521 }, { "epoch": 0.9449026488768548, "grad_norm": 1.5445848294791, "learning_rate": 1.5870873229838935e-07, "loss": 0.4914, "step": 26522 }, { "epoch": 0.9449382760033489, "grad_norm": 1.867513903752641, "learning_rate": 1.5850402733428194e-07, "loss": 0.4751, "step": 26523 }, { "epoch": 0.9449739031298431, "grad_norm": 1.3712904543649114, "learning_rate": 1.582994534165716e-07, "loss": 0.3677, "step": 26524 }, { "epoch": 0.9450095302563372, "grad_norm": 1.1699917485628113, "learning_rate": 1.5809501054798394e-07, "loss": 0.3957, "step": 26525 }, { "epoch": 0.9450451573828312, "grad_norm": 1.6412162213302166, "learning_rate": 1.5789069873124009e-07, "loss": 0.4431, "step": 26526 }, { "epoch": 0.9450807845093254, "grad_norm": 1.185379919362322, "learning_rate": 1.5768651796906008e-07, "loss": 0.4259, "step": 26527 }, { "epoch": 0.9451164116358195, "grad_norm": 1.2572657423838842, "learning_rate": 1.574824682641629e-07, "loss": 0.4146, "step": 26528 }, { "epoch": 0.9451520387623136, "grad_norm": 1.2975641154196202, "learning_rate": 1.5727854961926748e-07, "loss": 0.3975, "step": 26529 }, { "epoch": 0.9451876658888078, "grad_norm": 1.444423725915244, "learning_rate": 1.5707476203708716e-07, "loss": 0.4436, "step": 26530 }, { "epoch": 0.9452232930153018, "grad_norm": 1.5810189644283994, "learning_rate": 1.5687110552033645e-07, "loss": 0.4895, "step": 26531 }, { "epoch": 0.9452589201417959, "grad_norm": 1.382610311829333, "learning_rate": 1.5666758007172767e-07, "loss": 0.3923, "step": 26532 }, { "epoch": 0.9452945472682901, "grad_norm": 2.021960824176198, "learning_rate": 1.564641856939686e-07, "loss": 0.5256, "step": 26533 }, { "epoch": 0.9453301743947842, "grad_norm": 1.2682432671907242, "learning_rate": 1.562609223897693e-07, "loss": 0.4263, "step": 26534 }, { "epoch": 0.9453658015212782, "grad_norm": 1.7095833779991654, "learning_rate": 1.5605779016183654e-07, "loss": 0.3931, "step": 26535 }, { "epoch": 0.9454014286477724, "grad_norm": 1.3895579133133509, "learning_rate": 1.5585478901287476e-07, "loss": 0.3999, "step": 26536 }, { "epoch": 0.9454370557742665, "grad_norm": 1.638999921843951, "learning_rate": 1.556519189455885e-07, "loss": 0.4566, "step": 26537 }, { "epoch": 0.9454726829007606, "grad_norm": 1.3841927369488551, "learning_rate": 1.5544917996267562e-07, "loss": 0.4493, "step": 26538 }, { "epoch": 0.9455083100272548, "grad_norm": 1.22495205560035, "learning_rate": 1.5524657206683947e-07, "loss": 0.3974, "step": 26539 }, { "epoch": 0.9455439371537488, "grad_norm": 1.1473188642619194, "learning_rate": 1.5504409526077569e-07, "loss": 0.3563, "step": 26540 }, { "epoch": 0.9455795642802429, "grad_norm": 1.7051286506620666, "learning_rate": 1.54841749547181e-07, "loss": 0.557, "step": 26541 }, { "epoch": 0.9456151914067371, "grad_norm": 1.7276316807078176, "learning_rate": 1.5463953492874883e-07, "loss": 0.4196, "step": 26542 }, { "epoch": 0.9456508185332312, "grad_norm": 1.249830158439636, "learning_rate": 1.5443745140817368e-07, "loss": 0.3984, "step": 26543 }, { "epoch": 0.9456864456597253, "grad_norm": 1.125637346599389, "learning_rate": 1.542354989881456e-07, "loss": 0.3968, "step": 26544 }, { "epoch": 0.9457220727862194, "grad_norm": 1.272038256684415, "learning_rate": 1.5403367767135358e-07, "loss": 0.3955, "step": 26545 }, { "epoch": 0.9457576999127135, "grad_norm": 1.5176548782914552, "learning_rate": 1.5383198746048544e-07, "loss": 0.494, "step": 26546 }, { "epoch": 0.9457933270392076, "grad_norm": 2.927781898732173, "learning_rate": 1.536304283582246e-07, "loss": 0.5466, "step": 26547 }, { "epoch": 0.9458289541657018, "grad_norm": 1.2825666473740875, "learning_rate": 1.5342900036725894e-07, "loss": 0.3968, "step": 26548 }, { "epoch": 0.9458645812921959, "grad_norm": 1.453172288776725, "learning_rate": 1.532277034902674e-07, "loss": 0.5408, "step": 26549 }, { "epoch": 0.9459002084186899, "grad_norm": 1.708359084528658, "learning_rate": 1.5302653772993226e-07, "loss": 0.4809, "step": 26550 }, { "epoch": 0.9459358355451841, "grad_norm": 1.278653034867117, "learning_rate": 1.528255030889303e-07, "loss": 0.3979, "step": 26551 }, { "epoch": 0.9459714626716782, "grad_norm": 1.8031476109922444, "learning_rate": 1.5262459956994047e-07, "loss": 0.412, "step": 26552 }, { "epoch": 0.9460070897981723, "grad_norm": 2.3497444661997977, "learning_rate": 1.524238271756362e-07, "loss": 0.4677, "step": 26553 }, { "epoch": 0.9460427169246665, "grad_norm": 1.2019888772762795, "learning_rate": 1.5222318590869312e-07, "loss": 0.4046, "step": 26554 }, { "epoch": 0.9460783440511605, "grad_norm": 1.4804954307663516, "learning_rate": 1.520226757717802e-07, "loss": 0.4034, "step": 26555 }, { "epoch": 0.9461139711776546, "grad_norm": 1.267882498486659, "learning_rate": 1.5182229676756867e-07, "loss": 0.4658, "step": 26556 }, { "epoch": 0.9461495983041488, "grad_norm": 1.3202561625610838, "learning_rate": 1.5162204889872633e-07, "loss": 0.4086, "step": 26557 }, { "epoch": 0.9461852254306429, "grad_norm": 1.5172774425203979, "learning_rate": 1.5142193216792111e-07, "loss": 0.3539, "step": 26558 }, { "epoch": 0.946220852557137, "grad_norm": 1.3530229181179503, "learning_rate": 1.512219465778164e-07, "loss": 0.399, "step": 26559 }, { "epoch": 0.9462564796836311, "grad_norm": 1.2085260540710991, "learning_rate": 1.5102209213107454e-07, "loss": 0.3823, "step": 26560 }, { "epoch": 0.9462921068101252, "grad_norm": 1.612593826816412, "learning_rate": 1.5082236883035896e-07, "loss": 0.5185, "step": 26561 }, { "epoch": 0.9463277339366194, "grad_norm": 1.9321003071551583, "learning_rate": 1.5062277667832637e-07, "loss": 0.4729, "step": 26562 }, { "epoch": 0.9463633610631135, "grad_norm": 1.737835311839016, "learning_rate": 1.504233156776369e-07, "loss": 0.4716, "step": 26563 }, { "epoch": 0.9463989881896075, "grad_norm": 1.281528556791231, "learning_rate": 1.502239858309451e-07, "loss": 0.4035, "step": 26564 }, { "epoch": 0.9464346153161017, "grad_norm": 1.445926542734625, "learning_rate": 1.500247871409055e-07, "loss": 0.5187, "step": 26565 }, { "epoch": 0.9464702424425958, "grad_norm": 1.4581680118488962, "learning_rate": 1.4982571961016934e-07, "loss": 0.4915, "step": 26566 }, { "epoch": 0.9465058695690899, "grad_norm": 1.6926783231789786, "learning_rate": 1.4962678324139002e-07, "loss": 0.6132, "step": 26567 }, { "epoch": 0.9465414966955841, "grad_norm": 1.249650209250641, "learning_rate": 1.4942797803721543e-07, "loss": 0.4145, "step": 26568 }, { "epoch": 0.9465771238220781, "grad_norm": 1.3171864428882702, "learning_rate": 1.4922930400029123e-07, "loss": 0.4048, "step": 26569 }, { "epoch": 0.9466127509485722, "grad_norm": 1.5112784972206523, "learning_rate": 1.4903076113326532e-07, "loss": 0.491, "step": 26570 }, { "epoch": 0.9466483780750664, "grad_norm": 1.5552179437476121, "learning_rate": 1.4883234943877888e-07, "loss": 0.4068, "step": 26571 }, { "epoch": 0.9466840052015605, "grad_norm": 1.7386946866778776, "learning_rate": 1.486340689194754e-07, "loss": 0.4776, "step": 26572 }, { "epoch": 0.9467196323280546, "grad_norm": 1.176719389308354, "learning_rate": 1.4843591957799497e-07, "loss": 0.4206, "step": 26573 }, { "epoch": 0.9467552594545487, "grad_norm": 1.3389844616877586, "learning_rate": 1.482379014169777e-07, "loss": 0.4046, "step": 26574 }, { "epoch": 0.9467908865810428, "grad_norm": 1.4796204048564034, "learning_rate": 1.4804001443905703e-07, "loss": 0.478, "step": 26575 }, { "epoch": 0.9468265137075369, "grad_norm": 1.6782967304791645, "learning_rate": 1.4784225864687085e-07, "loss": 0.4998, "step": 26576 }, { "epoch": 0.9468621408340311, "grad_norm": 1.2989047568278085, "learning_rate": 1.4764463404305042e-07, "loss": 0.3762, "step": 26577 }, { "epoch": 0.9468977679605252, "grad_norm": 1.1877390792418074, "learning_rate": 1.4744714063022914e-07, "loss": 0.4016, "step": 26578 }, { "epoch": 0.9469333950870192, "grad_norm": 2.005817401868605, "learning_rate": 1.4724977841103493e-07, "loss": 0.5015, "step": 26579 }, { "epoch": 0.9469690222135134, "grad_norm": 1.5152175543092534, "learning_rate": 1.470525473880957e-07, "loss": 0.5505, "step": 26580 }, { "epoch": 0.9470046493400075, "grad_norm": 1.5786926137916322, "learning_rate": 1.4685544756403935e-07, "loss": 0.4937, "step": 26581 }, { "epoch": 0.9470402764665016, "grad_norm": 1.5205006830763788, "learning_rate": 1.4665847894148933e-07, "loss": 0.4525, "step": 26582 }, { "epoch": 0.9470759035929958, "grad_norm": 1.3154246943645826, "learning_rate": 1.464616415230702e-07, "loss": 0.3778, "step": 26583 }, { "epoch": 0.9471115307194898, "grad_norm": 1.3633197246792799, "learning_rate": 1.4626493531139985e-07, "loss": 0.4785, "step": 26584 }, { "epoch": 0.9471471578459839, "grad_norm": 1.3303965569234726, "learning_rate": 1.4606836030910066e-07, "loss": 0.4058, "step": 26585 }, { "epoch": 0.9471827849724781, "grad_norm": 1.4301920932113195, "learning_rate": 1.458719165187883e-07, "loss": 0.3782, "step": 26586 }, { "epoch": 0.9472184120989722, "grad_norm": 1.179946052696396, "learning_rate": 1.4567560394307844e-07, "loss": 0.3941, "step": 26587 }, { "epoch": 0.9472540392254662, "grad_norm": 1.2386883596252418, "learning_rate": 1.454794225845857e-07, "loss": 0.3767, "step": 26588 }, { "epoch": 0.9472896663519604, "grad_norm": 1.4854143717317754, "learning_rate": 1.452833724459224e-07, "loss": 0.3986, "step": 26589 }, { "epoch": 0.9473252934784545, "grad_norm": 1.7591296086466828, "learning_rate": 1.4508745352969865e-07, "loss": 0.5081, "step": 26590 }, { "epoch": 0.9473609206049486, "grad_norm": 1.5151012967955397, "learning_rate": 1.4489166583852466e-07, "loss": 0.5038, "step": 26591 }, { "epoch": 0.9473965477314428, "grad_norm": 2.26599087677079, "learning_rate": 1.4469600937500604e-07, "loss": 0.5861, "step": 26592 }, { "epoch": 0.9474321748579368, "grad_norm": 1.2462035952091397, "learning_rate": 1.4450048414174854e-07, "loss": 0.4249, "step": 26593 }, { "epoch": 0.9474678019844309, "grad_norm": 1.3085985803209275, "learning_rate": 1.4430509014135564e-07, "loss": 0.4073, "step": 26594 }, { "epoch": 0.9475034291109251, "grad_norm": 1.5262270527357698, "learning_rate": 1.4410982737642853e-07, "loss": 0.5263, "step": 26595 }, { "epoch": 0.9475390562374192, "grad_norm": 1.3795151259226608, "learning_rate": 1.4391469584956852e-07, "loss": 0.3665, "step": 26596 }, { "epoch": 0.9475746833639133, "grad_norm": 1.4356791824466584, "learning_rate": 1.437196955633735e-07, "loss": 0.4738, "step": 26597 }, { "epoch": 0.9476103104904074, "grad_norm": 1.7049596928637838, "learning_rate": 1.435248265204403e-07, "loss": 0.4594, "step": 26598 }, { "epoch": 0.9476459376169015, "grad_norm": 1.7352776645049712, "learning_rate": 1.433300887233624e-07, "loss": 0.5249, "step": 26599 }, { "epoch": 0.9476815647433956, "grad_norm": 1.1721595979381, "learning_rate": 1.431354821747344e-07, "loss": 0.4083, "step": 26600 }, { "epoch": 0.9477171918698898, "grad_norm": 1.8545540170289674, "learning_rate": 1.429410068771464e-07, "loss": 0.5467, "step": 26601 }, { "epoch": 0.9477528189963839, "grad_norm": 1.2611894911091055, "learning_rate": 1.427466628331886e-07, "loss": 0.4017, "step": 26602 }, { "epoch": 0.9477884461228779, "grad_norm": 1.5358341839900462, "learning_rate": 1.4255245004544894e-07, "loss": 0.4907, "step": 26603 }, { "epoch": 0.9478240732493721, "grad_norm": 1.6190901184804145, "learning_rate": 1.423583685165131e-07, "loss": 0.4165, "step": 26604 }, { "epoch": 0.9478597003758662, "grad_norm": 1.3122508312921484, "learning_rate": 1.4216441824896565e-07, "loss": 0.4353, "step": 26605 }, { "epoch": 0.9478953275023603, "grad_norm": 1.5858329530838864, "learning_rate": 1.4197059924539015e-07, "loss": 0.5309, "step": 26606 }, { "epoch": 0.9479309546288545, "grad_norm": 2.5764905689775555, "learning_rate": 1.417769115083656e-07, "loss": 0.5644, "step": 26607 }, { "epoch": 0.9479665817553485, "grad_norm": 1.5204164205340778, "learning_rate": 1.4158335504047216e-07, "loss": 0.4196, "step": 26608 }, { "epoch": 0.9480022088818426, "grad_norm": 1.7123353173369986, "learning_rate": 1.4138992984428669e-07, "loss": 0.5637, "step": 26609 }, { "epoch": 0.9480378360083368, "grad_norm": 1.1915672586282282, "learning_rate": 1.4119663592238486e-07, "loss": 0.4055, "step": 26610 }, { "epoch": 0.9480734631348309, "grad_norm": 1.507703059169715, "learning_rate": 1.4100347327734244e-07, "loss": 0.5135, "step": 26611 }, { "epoch": 0.948109090261325, "grad_norm": 1.3298500012746568, "learning_rate": 1.4081044191172733e-07, "loss": 0.4169, "step": 26612 }, { "epoch": 0.9481447173878191, "grad_norm": 1.2120310793365645, "learning_rate": 1.4061754182811415e-07, "loss": 0.384, "step": 26613 }, { "epoch": 0.9481803445143132, "grad_norm": 1.9475489038223792, "learning_rate": 1.4042477302906976e-07, "loss": 0.4544, "step": 26614 }, { "epoch": 0.9482159716408073, "grad_norm": 1.839039477501917, "learning_rate": 1.4023213551715985e-07, "loss": 0.5101, "step": 26615 }, { "epoch": 0.9482515987673015, "grad_norm": 1.9709801315799529, "learning_rate": 1.400396292949513e-07, "loss": 0.5254, "step": 26616 }, { "epoch": 0.9482872258937955, "grad_norm": 1.909090305636544, "learning_rate": 1.3984725436500645e-07, "loss": 0.5522, "step": 26617 }, { "epoch": 0.9483228530202896, "grad_norm": 2.2750623969282397, "learning_rate": 1.3965501072988662e-07, "loss": 0.4638, "step": 26618 }, { "epoch": 0.9483584801467838, "grad_norm": 1.4367806947724624, "learning_rate": 1.394628983921531e-07, "loss": 0.5382, "step": 26619 }, { "epoch": 0.9483941072732779, "grad_norm": 1.6180807305716678, "learning_rate": 1.3927091735436382e-07, "loss": 0.5687, "step": 26620 }, { "epoch": 0.948429734399772, "grad_norm": 1.4546106644472405, "learning_rate": 1.390790676190734e-07, "loss": 0.4339, "step": 26621 }, { "epoch": 0.9484653615262661, "grad_norm": 1.4883320660972392, "learning_rate": 1.3888734918883873e-07, "loss": 0.4781, "step": 26622 }, { "epoch": 0.9485009886527602, "grad_norm": 1.2405373595879707, "learning_rate": 1.3869576206620994e-07, "loss": 0.3948, "step": 26623 }, { "epoch": 0.9485366157792543, "grad_norm": 1.7509142453402275, "learning_rate": 1.3850430625374056e-07, "loss": 0.4972, "step": 26624 }, { "epoch": 0.9485722429057485, "grad_norm": 1.4533935729467933, "learning_rate": 1.3831298175397856e-07, "loss": 0.5188, "step": 26625 }, { "epoch": 0.9486078700322426, "grad_norm": 3.092101909164977, "learning_rate": 1.38121788569473e-07, "loss": 0.5817, "step": 26626 }, { "epoch": 0.9486434971587366, "grad_norm": 1.6388252136883308, "learning_rate": 1.3793072670276852e-07, "loss": 0.4904, "step": 26627 }, { "epoch": 0.9486791242852308, "grad_norm": 1.3836058812897938, "learning_rate": 1.3773979615640976e-07, "loss": 0.4292, "step": 26628 }, { "epoch": 0.9487147514117249, "grad_norm": 1.6386924948079045, "learning_rate": 1.375489969329391e-07, "loss": 0.5244, "step": 26629 }, { "epoch": 0.948750378538219, "grad_norm": 1.272264195156502, "learning_rate": 1.3735832903489676e-07, "loss": 0.3518, "step": 26630 }, { "epoch": 0.9487860056647132, "grad_norm": 2.4524797132675538, "learning_rate": 1.371677924648207e-07, "loss": 0.4769, "step": 26631 }, { "epoch": 0.9488216327912072, "grad_norm": 1.6702652104111513, "learning_rate": 1.3697738722525e-07, "loss": 0.5951, "step": 26632 }, { "epoch": 0.9488572599177013, "grad_norm": 1.590755240468009, "learning_rate": 1.367871133187204e-07, "loss": 0.574, "step": 26633 }, { "epoch": 0.9488928870441955, "grad_norm": 1.3937883160066094, "learning_rate": 1.3659697074776324e-07, "loss": 0.5552, "step": 26634 }, { "epoch": 0.9489285141706896, "grad_norm": 1.3131397139564536, "learning_rate": 1.36406959514912e-07, "loss": 0.3932, "step": 26635 }, { "epoch": 0.9489641412971836, "grad_norm": 1.2668350718577022, "learning_rate": 1.3621707962269582e-07, "loss": 0.3762, "step": 26636 }, { "epoch": 0.9489997684236778, "grad_norm": 1.2529301474431827, "learning_rate": 1.3602733107364373e-07, "loss": 0.435, "step": 26637 }, { "epoch": 0.9490353955501719, "grad_norm": 2.041688462989803, "learning_rate": 1.3583771387028267e-07, "loss": 0.5083, "step": 26638 }, { "epoch": 0.949071022676666, "grad_norm": 1.4757805288076637, "learning_rate": 1.356482280151361e-07, "loss": 0.4141, "step": 26639 }, { "epoch": 0.9491066498031602, "grad_norm": 1.1681799456764632, "learning_rate": 1.3545887351072872e-07, "loss": 0.4041, "step": 26640 }, { "epoch": 0.9491422769296542, "grad_norm": 1.6301660388355155, "learning_rate": 1.3526965035958073e-07, "loss": 0.4143, "step": 26641 }, { "epoch": 0.9491779040561483, "grad_norm": 1.6304976275181238, "learning_rate": 1.3508055856421452e-07, "loss": 0.4833, "step": 26642 }, { "epoch": 0.9492135311826425, "grad_norm": 1.2560648121165445, "learning_rate": 1.348915981271437e-07, "loss": 0.3906, "step": 26643 }, { "epoch": 0.9492491583091366, "grad_norm": 1.2150204534580735, "learning_rate": 1.347027690508873e-07, "loss": 0.396, "step": 26644 }, { "epoch": 0.9492847854356307, "grad_norm": 2.3208334036894995, "learning_rate": 1.345140713379589e-07, "loss": 0.4651, "step": 26645 }, { "epoch": 0.9493204125621248, "grad_norm": 1.518092725130852, "learning_rate": 1.3432550499087093e-07, "loss": 0.5105, "step": 26646 }, { "epoch": 0.9493560396886189, "grad_norm": 1.635495880935108, "learning_rate": 1.3413707001213472e-07, "loss": 0.4722, "step": 26647 }, { "epoch": 0.949391666815113, "grad_norm": 1.4334720576730196, "learning_rate": 1.3394876640425937e-07, "loss": 0.3904, "step": 26648 }, { "epoch": 0.9494272939416072, "grad_norm": 1.3099933321945771, "learning_rate": 1.3376059416975174e-07, "loss": 0.4023, "step": 26649 }, { "epoch": 0.9494629210681013, "grad_norm": 1.7829731809698797, "learning_rate": 1.3357255331111874e-07, "loss": 0.5859, "step": 26650 }, { "epoch": 0.9494985481945953, "grad_norm": 1.1972900408405187, "learning_rate": 1.333846438308628e-07, "loss": 0.4128, "step": 26651 }, { "epoch": 0.9495341753210895, "grad_norm": 1.346413740721514, "learning_rate": 1.331968657314864e-07, "loss": 0.438, "step": 26652 }, { "epoch": 0.9495698024475836, "grad_norm": 1.3122480557051, "learning_rate": 1.330092190154919e-07, "loss": 0.4203, "step": 26653 }, { "epoch": 0.9496054295740777, "grad_norm": 1.3924250239912892, "learning_rate": 1.3282170368537407e-07, "loss": 0.3995, "step": 26654 }, { "epoch": 0.9496410567005719, "grad_norm": 1.4924803641111364, "learning_rate": 1.3263431974363306e-07, "loss": 0.4493, "step": 26655 }, { "epoch": 0.9496766838270659, "grad_norm": 2.056217700837137, "learning_rate": 1.3244706719276246e-07, "loss": 0.4907, "step": 26656 }, { "epoch": 0.94971231095356, "grad_norm": 2.0135579838071727, "learning_rate": 1.3225994603525694e-07, "loss": 0.5999, "step": 26657 }, { "epoch": 0.9497479380800542, "grad_norm": 1.3088859722556505, "learning_rate": 1.3207295627360672e-07, "loss": 0.4352, "step": 26658 }, { "epoch": 0.9497835652065483, "grad_norm": 1.2125385166399847, "learning_rate": 1.3188609791030316e-07, "loss": 0.3733, "step": 26659 }, { "epoch": 0.9498191923330424, "grad_norm": 1.165081110132848, "learning_rate": 1.3169937094783314e-07, "loss": 0.4121, "step": 26660 }, { "epoch": 0.9498548194595365, "grad_norm": 1.230780912794077, "learning_rate": 1.3151277538868245e-07, "loss": 0.4324, "step": 26661 }, { "epoch": 0.9498904465860306, "grad_norm": 1.6699818576485186, "learning_rate": 1.31326311235338e-07, "loss": 0.4993, "step": 26662 }, { "epoch": 0.9499260737125247, "grad_norm": 1.368516224471376, "learning_rate": 1.3113997849028114e-07, "loss": 0.3616, "step": 26663 }, { "epoch": 0.9499617008390189, "grad_norm": 1.1623010978127508, "learning_rate": 1.3095377715599433e-07, "loss": 0.3907, "step": 26664 }, { "epoch": 0.949997327965513, "grad_norm": 2.1070788045310667, "learning_rate": 1.3076770723495446e-07, "loss": 0.4588, "step": 26665 }, { "epoch": 0.950032955092007, "grad_norm": 1.4869756650039085, "learning_rate": 1.3058176872964178e-07, "loss": 0.4506, "step": 26666 }, { "epoch": 0.9500685822185012, "grad_norm": 1.4785959584857225, "learning_rate": 1.30395961642531e-07, "loss": 0.4199, "step": 26667 }, { "epoch": 0.9501042093449953, "grad_norm": 1.4224953672089915, "learning_rate": 1.3021028597609676e-07, "loss": 0.5042, "step": 26668 }, { "epoch": 0.9501398364714894, "grad_norm": 1.3950358126731988, "learning_rate": 1.3002474173280932e-07, "loss": 0.3668, "step": 26669 }, { "epoch": 0.9501754635979835, "grad_norm": 1.5270592161133858, "learning_rate": 1.298393289151434e-07, "loss": 0.5124, "step": 26670 }, { "epoch": 0.9502110907244776, "grad_norm": 1.3377500285488233, "learning_rate": 1.2965404752556477e-07, "loss": 0.512, "step": 26671 }, { "epoch": 0.9502467178509717, "grad_norm": 1.464216091173787, "learning_rate": 1.2946889756654146e-07, "loss": 0.4533, "step": 26672 }, { "epoch": 0.9502823449774659, "grad_norm": 1.3654932172008527, "learning_rate": 1.292838790405393e-07, "loss": 0.4925, "step": 26673 }, { "epoch": 0.95031797210396, "grad_norm": 1.6857471091784544, "learning_rate": 1.2909899195002072e-07, "loss": 0.4819, "step": 26674 }, { "epoch": 0.950353599230454, "grad_norm": 1.3236335373762367, "learning_rate": 1.2891423629744938e-07, "loss": 0.4071, "step": 26675 }, { "epoch": 0.9503892263569482, "grad_norm": 1.5295713466815308, "learning_rate": 1.2872961208528323e-07, "loss": 0.5066, "step": 26676 }, { "epoch": 0.9504248534834423, "grad_norm": 1.4735506925926631, "learning_rate": 1.2854511931598257e-07, "loss": 0.4643, "step": 26677 }, { "epoch": 0.9504604806099364, "grad_norm": 2.00398406999462, "learning_rate": 1.283607579920032e-07, "loss": 0.5272, "step": 26678 }, { "epoch": 0.9504961077364306, "grad_norm": 1.3056801715964035, "learning_rate": 1.2817652811579984e-07, "loss": 0.4585, "step": 26679 }, { "epoch": 0.9505317348629246, "grad_norm": 1.3606469388288231, "learning_rate": 1.2799242968982717e-07, "loss": 0.4209, "step": 26680 }, { "epoch": 0.9505673619894187, "grad_norm": 1.511134670544407, "learning_rate": 1.2780846271653435e-07, "loss": 0.4896, "step": 26681 }, { "epoch": 0.9506029891159129, "grad_norm": 1.1899015969770568, "learning_rate": 1.2762462719837276e-07, "loss": 0.4071, "step": 26682 }, { "epoch": 0.950638616242407, "grad_norm": 1.56655573509852, "learning_rate": 1.2744092313778933e-07, "loss": 0.4955, "step": 26683 }, { "epoch": 0.950674243368901, "grad_norm": 1.2745288017797705, "learning_rate": 1.2725735053722986e-07, "loss": 0.4205, "step": 26684 }, { "epoch": 0.9507098704953952, "grad_norm": 1.2133567915925794, "learning_rate": 1.2707390939914022e-07, "loss": 0.3766, "step": 26685 }, { "epoch": 0.9507454976218893, "grad_norm": 1.527351841970926, "learning_rate": 1.2689059972596284e-07, "loss": 0.5628, "step": 26686 }, { "epoch": 0.9507811247483834, "grad_norm": 1.6836712255672297, "learning_rate": 1.2670742152013693e-07, "loss": 0.4679, "step": 26687 }, { "epoch": 0.9508167518748776, "grad_norm": 1.3704071540323057, "learning_rate": 1.2652437478410272e-07, "loss": 0.3913, "step": 26688 }, { "epoch": 0.9508523790013717, "grad_norm": 1.745992561723217, "learning_rate": 1.263414595202983e-07, "loss": 0.4665, "step": 26689 }, { "epoch": 0.9508880061278657, "grad_norm": 1.3048539392096639, "learning_rate": 1.2615867573115836e-07, "loss": 0.4551, "step": 26690 }, { "epoch": 0.9509236332543599, "grad_norm": 1.6371827886402548, "learning_rate": 1.2597602341911764e-07, "loss": 0.5086, "step": 26691 }, { "epoch": 0.950959260380854, "grad_norm": 1.4791267420133631, "learning_rate": 1.2579350258660638e-07, "loss": 0.4036, "step": 26692 }, { "epoch": 0.9509948875073481, "grad_norm": 1.2269238846274977, "learning_rate": 1.2561111323605714e-07, "loss": 0.408, "step": 26693 }, { "epoch": 0.9510305146338422, "grad_norm": 1.9330614778935775, "learning_rate": 1.254288553698979e-07, "loss": 0.5425, "step": 26694 }, { "epoch": 0.9510661417603363, "grad_norm": 1.5406347620460126, "learning_rate": 1.2524672899055456e-07, "loss": 0.4582, "step": 26695 }, { "epoch": 0.9511017688868304, "grad_norm": 1.588735639247557, "learning_rate": 1.2506473410045295e-07, "loss": 0.4733, "step": 26696 }, { "epoch": 0.9511373960133246, "grad_norm": 1.5153053625835249, "learning_rate": 1.2488287070201776e-07, "loss": 0.4887, "step": 26697 }, { "epoch": 0.9511730231398187, "grad_norm": 1.6430303985189407, "learning_rate": 1.247011387976682e-07, "loss": 0.5114, "step": 26698 }, { "epoch": 0.9512086502663127, "grad_norm": 1.2260800151343425, "learning_rate": 1.2451953838982566e-07, "loss": 0.3519, "step": 26699 }, { "epoch": 0.9512442773928069, "grad_norm": 1.2726878508232538, "learning_rate": 1.2433806948090822e-07, "loss": 0.4509, "step": 26700 }, { "epoch": 0.951279904519301, "grad_norm": 1.340254101415552, "learning_rate": 1.2415673207333278e-07, "loss": 0.3734, "step": 26701 }, { "epoch": 0.9513155316457951, "grad_norm": 1.2970445334820948, "learning_rate": 1.2397552616951302e-07, "loss": 0.4167, "step": 26702 }, { "epoch": 0.9513511587722893, "grad_norm": 2.0219839045981267, "learning_rate": 1.2379445177186144e-07, "loss": 0.587, "step": 26703 }, { "epoch": 0.9513867858987833, "grad_norm": 1.2062317784783902, "learning_rate": 1.2361350888279055e-07, "loss": 0.3642, "step": 26704 }, { "epoch": 0.9514224130252774, "grad_norm": 1.3635171190838602, "learning_rate": 1.2343269750470842e-07, "loss": 0.4421, "step": 26705 }, { "epoch": 0.9514580401517716, "grad_norm": 1.477156592737761, "learning_rate": 1.2325201764002316e-07, "loss": 0.4662, "step": 26706 }, { "epoch": 0.9514936672782657, "grad_norm": 1.6278949465204358, "learning_rate": 1.2307146929113945e-07, "loss": 0.5228, "step": 26707 }, { "epoch": 0.9515292944047598, "grad_norm": 1.770067592820918, "learning_rate": 1.228910524604654e-07, "loss": 0.503, "step": 26708 }, { "epoch": 0.9515649215312539, "grad_norm": 1.438107422844557, "learning_rate": 1.2271076715039909e-07, "loss": 0.4078, "step": 26709 }, { "epoch": 0.951600548657748, "grad_norm": 1.6614325977967321, "learning_rate": 1.2253061336334304e-07, "loss": 0.5187, "step": 26710 }, { "epoch": 0.9516361757842421, "grad_norm": 1.4545500406477918, "learning_rate": 1.2235059110169534e-07, "loss": 0.4076, "step": 26711 }, { "epoch": 0.9516718029107363, "grad_norm": 1.0406830352584167, "learning_rate": 1.2217070036785406e-07, "loss": 0.3682, "step": 26712 }, { "epoch": 0.9517074300372304, "grad_norm": 1.3917886428654, "learning_rate": 1.2199094116421395e-07, "loss": 0.4271, "step": 26713 }, { "epoch": 0.9517430571637244, "grad_norm": 1.6285486395153423, "learning_rate": 1.2181131349316866e-07, "loss": 0.5163, "step": 26714 }, { "epoch": 0.9517786842902186, "grad_norm": 1.4514642055722706, "learning_rate": 1.2163181735711072e-07, "loss": 0.4018, "step": 26715 }, { "epoch": 0.9518143114167127, "grad_norm": 1.3870066936023646, "learning_rate": 1.2145245275842932e-07, "loss": 0.3838, "step": 26716 }, { "epoch": 0.9518499385432068, "grad_norm": 1.4876036366993932, "learning_rate": 1.2127321969951367e-07, "loss": 0.4025, "step": 26717 }, { "epoch": 0.951885565669701, "grad_norm": 1.7261272764166324, "learning_rate": 1.2109411818274851e-07, "loss": 0.4883, "step": 26718 }, { "epoch": 0.951921192796195, "grad_norm": 1.6577327512347664, "learning_rate": 1.2091514821052197e-07, "loss": 0.5254, "step": 26719 }, { "epoch": 0.9519568199226891, "grad_norm": 1.634064184125572, "learning_rate": 1.2073630978521432e-07, "loss": 0.5103, "step": 26720 }, { "epoch": 0.9519924470491833, "grad_norm": 1.487801938030208, "learning_rate": 1.2055760290920704e-07, "loss": 0.392, "step": 26721 }, { "epoch": 0.9520280741756774, "grad_norm": 1.5911023730191385, "learning_rate": 1.203790275848815e-07, "loss": 0.5528, "step": 26722 }, { "epoch": 0.9520637013021714, "grad_norm": 1.9808838534474353, "learning_rate": 1.2020058381461586e-07, "loss": 0.5675, "step": 26723 }, { "epoch": 0.9520993284286656, "grad_norm": 1.2073632689188631, "learning_rate": 1.2002227160078262e-07, "loss": 0.3777, "step": 26724 }, { "epoch": 0.9521349555551597, "grad_norm": 1.7934587352925844, "learning_rate": 1.19844090945761e-07, "loss": 0.6002, "step": 26725 }, { "epoch": 0.9521705826816538, "grad_norm": 1.5646396663008162, "learning_rate": 1.196660418519191e-07, "loss": 0.5003, "step": 26726 }, { "epoch": 0.952206209808148, "grad_norm": 1.3761032756573428, "learning_rate": 1.1948812432163172e-07, "loss": 0.4525, "step": 26727 }, { "epoch": 0.952241836934642, "grad_norm": 1.214195350266517, "learning_rate": 1.193103383572647e-07, "loss": 0.4023, "step": 26728 }, { "epoch": 0.9522774640611361, "grad_norm": 1.161917390016886, "learning_rate": 1.1913268396118616e-07, "loss": 0.387, "step": 26729 }, { "epoch": 0.9523130911876303, "grad_norm": 1.4717045102470345, "learning_rate": 1.1895516113576311e-07, "loss": 0.5028, "step": 26730 }, { "epoch": 0.9523487183141244, "grad_norm": 2.602581334555664, "learning_rate": 1.1877776988335809e-07, "loss": 0.4994, "step": 26731 }, { "epoch": 0.9523843454406185, "grad_norm": 1.358530727273829, "learning_rate": 1.1860051020633368e-07, "loss": 0.4146, "step": 26732 }, { "epoch": 0.9524199725671126, "grad_norm": 1.7075986950297308, "learning_rate": 1.1842338210705018e-07, "loss": 0.5489, "step": 26733 }, { "epoch": 0.9524555996936067, "grad_norm": 1.2890351795823733, "learning_rate": 1.1824638558786682e-07, "loss": 0.3969, "step": 26734 }, { "epoch": 0.9524912268201008, "grad_norm": 1.1807331129058152, "learning_rate": 1.1806952065113952e-07, "loss": 0.3607, "step": 26735 }, { "epoch": 0.952526853946595, "grad_norm": 1.3996433396660861, "learning_rate": 1.1789278729922305e-07, "loss": 0.4263, "step": 26736 }, { "epoch": 0.952562481073089, "grad_norm": 1.648574934289604, "learning_rate": 1.1771618553447217e-07, "loss": 0.5924, "step": 26737 }, { "epoch": 0.9525981081995831, "grad_norm": 1.464545244630163, "learning_rate": 1.1753971535923725e-07, "loss": 0.3919, "step": 26738 }, { "epoch": 0.9526337353260773, "grad_norm": 1.5899378812875136, "learning_rate": 1.173633767758675e-07, "loss": 0.3724, "step": 26739 }, { "epoch": 0.9526693624525714, "grad_norm": 1.624442243328657, "learning_rate": 1.1718716978671329e-07, "loss": 0.5144, "step": 26740 }, { "epoch": 0.9527049895790655, "grad_norm": 1.8971237903271125, "learning_rate": 1.170110943941194e-07, "loss": 0.4779, "step": 26741 }, { "epoch": 0.9527406167055597, "grad_norm": 1.5982814586337994, "learning_rate": 1.168351506004295e-07, "loss": 0.5541, "step": 26742 }, { "epoch": 0.9527762438320537, "grad_norm": 1.3277317809829956, "learning_rate": 1.1665933840798838e-07, "loss": 0.4258, "step": 26743 }, { "epoch": 0.9528118709585478, "grad_norm": 1.7997475329071528, "learning_rate": 1.1648365781913528e-07, "loss": 0.5134, "step": 26744 }, { "epoch": 0.952847498085042, "grad_norm": 1.2647621397894704, "learning_rate": 1.1630810883621168e-07, "loss": 0.3769, "step": 26745 }, { "epoch": 0.9528831252115361, "grad_norm": 1.5607620689400683, "learning_rate": 1.1613269146155348e-07, "loss": 0.494, "step": 26746 }, { "epoch": 0.9529187523380301, "grad_norm": 1.1583751203807566, "learning_rate": 1.1595740569749658e-07, "loss": 0.4436, "step": 26747 }, { "epoch": 0.9529543794645243, "grad_norm": 1.5274387203331605, "learning_rate": 1.1578225154637579e-07, "loss": 0.4913, "step": 26748 }, { "epoch": 0.9529900065910184, "grad_norm": 1.3395671117332602, "learning_rate": 1.1560722901052368e-07, "loss": 0.4458, "step": 26749 }, { "epoch": 0.9530256337175125, "grad_norm": 1.3598470769864341, "learning_rate": 1.1543233809226839e-07, "loss": 0.4709, "step": 26750 }, { "epoch": 0.9530612608440067, "grad_norm": 1.5646894249168608, "learning_rate": 1.1525757879394139e-07, "loss": 0.4595, "step": 26751 }, { "epoch": 0.9530968879705007, "grad_norm": 1.241358709867866, "learning_rate": 1.150829511178686e-07, "loss": 0.3661, "step": 26752 }, { "epoch": 0.9531325150969948, "grad_norm": 1.3444412661354217, "learning_rate": 1.1490845506637482e-07, "loss": 0.4272, "step": 26753 }, { "epoch": 0.953168142223489, "grad_norm": 1.386005655599603, "learning_rate": 1.1473409064178487e-07, "loss": 0.4034, "step": 26754 }, { "epoch": 0.9532037693499831, "grad_norm": 1.6842656621758805, "learning_rate": 1.1455985784642021e-07, "loss": 0.5621, "step": 26755 }, { "epoch": 0.9532393964764772, "grad_norm": 1.3619535774183824, "learning_rate": 1.143857566826001e-07, "loss": 0.4248, "step": 26756 }, { "epoch": 0.9532750236029713, "grad_norm": 1.28222300165381, "learning_rate": 1.142117871526438e-07, "loss": 0.465, "step": 26757 }, { "epoch": 0.9533106507294654, "grad_norm": 1.3750426155026394, "learning_rate": 1.1403794925886614e-07, "loss": 0.4092, "step": 26758 }, { "epoch": 0.9533462778559595, "grad_norm": 1.3135504364368347, "learning_rate": 1.1386424300358412e-07, "loss": 0.4871, "step": 26759 }, { "epoch": 0.9533819049824537, "grad_norm": 2.5556788661562715, "learning_rate": 1.1369066838910925e-07, "loss": 0.5347, "step": 26760 }, { "epoch": 0.9534175321089478, "grad_norm": 1.823186142766817, "learning_rate": 1.1351722541775411e-07, "loss": 0.5666, "step": 26761 }, { "epoch": 0.9534531592354418, "grad_norm": 6.968632484619587, "learning_rate": 1.1334391409182687e-07, "loss": 0.5316, "step": 26762 }, { "epoch": 0.953488786361936, "grad_norm": 1.4590635061272827, "learning_rate": 1.1317073441363458e-07, "loss": 0.429, "step": 26763 }, { "epoch": 0.9535244134884301, "grad_norm": 1.4667412816760572, "learning_rate": 1.1299768638548536e-07, "loss": 0.4006, "step": 26764 }, { "epoch": 0.9535600406149242, "grad_norm": 1.692163112970843, "learning_rate": 1.1282477000968296e-07, "loss": 0.5435, "step": 26765 }, { "epoch": 0.9535956677414184, "grad_norm": 1.2249681880267997, "learning_rate": 1.1265198528852884e-07, "loss": 0.3598, "step": 26766 }, { "epoch": 0.9536312948679124, "grad_norm": 1.5898822588907509, "learning_rate": 1.124793322243245e-07, "loss": 0.5077, "step": 26767 }, { "epoch": 0.9536669219944065, "grad_norm": 1.7179516645158177, "learning_rate": 1.1230681081936922e-07, "loss": 0.537, "step": 26768 }, { "epoch": 0.9537025491209007, "grad_norm": 1.2139246387045373, "learning_rate": 1.1213442107596007e-07, "loss": 0.3894, "step": 26769 }, { "epoch": 0.9537381762473948, "grad_norm": 1.221621434574463, "learning_rate": 1.1196216299639073e-07, "loss": 0.3835, "step": 26770 }, { "epoch": 0.9537738033738888, "grad_norm": 1.4088284210536564, "learning_rate": 1.1179003658295828e-07, "loss": 0.4178, "step": 26771 }, { "epoch": 0.953809430500383, "grad_norm": 1.223686252789013, "learning_rate": 1.1161804183795199e-07, "loss": 0.4086, "step": 26772 }, { "epoch": 0.9538450576268771, "grad_norm": 1.114559419203005, "learning_rate": 1.1144617876366337e-07, "loss": 0.365, "step": 26773 }, { "epoch": 0.9538806847533712, "grad_norm": 1.9584184863665395, "learning_rate": 1.1127444736238057e-07, "loss": 0.504, "step": 26774 }, { "epoch": 0.9539163118798654, "grad_norm": 1.2349320597711955, "learning_rate": 1.1110284763639068e-07, "loss": 0.4006, "step": 26775 }, { "epoch": 0.9539519390063594, "grad_norm": 1.7333326831775324, "learning_rate": 1.109313795879774e-07, "loss": 0.5673, "step": 26776 }, { "epoch": 0.9539875661328535, "grad_norm": 1.8460347976724742, "learning_rate": 1.1076004321942557e-07, "loss": 0.5747, "step": 26777 }, { "epoch": 0.9540231932593477, "grad_norm": 1.859635282470314, "learning_rate": 1.105888385330156e-07, "loss": 0.5589, "step": 26778 }, { "epoch": 0.9540588203858418, "grad_norm": 1.487189679995833, "learning_rate": 1.104177655310279e-07, "loss": 0.4308, "step": 26779 }, { "epoch": 0.9540944475123359, "grad_norm": 1.4900388012879047, "learning_rate": 1.1024682421573951e-07, "loss": 0.4734, "step": 26780 }, { "epoch": 0.95413007463883, "grad_norm": 1.3792555836587232, "learning_rate": 1.1007601458942751e-07, "loss": 0.3938, "step": 26781 }, { "epoch": 0.9541657017653241, "grad_norm": 1.6161835921038943, "learning_rate": 1.0990533665436565e-07, "loss": 0.5855, "step": 26782 }, { "epoch": 0.9542013288918182, "grad_norm": 1.2468621999650373, "learning_rate": 1.0973479041282654e-07, "loss": 0.352, "step": 26783 }, { "epoch": 0.9542369560183124, "grad_norm": 1.1659883511822546, "learning_rate": 1.095643758670828e-07, "loss": 0.402, "step": 26784 }, { "epoch": 0.9542725831448065, "grad_norm": 1.1854721715995593, "learning_rate": 1.0939409301940151e-07, "loss": 0.4286, "step": 26785 }, { "epoch": 0.9543082102713005, "grad_norm": 1.2737122326588488, "learning_rate": 1.0922394187205198e-07, "loss": 0.332, "step": 26786 }, { "epoch": 0.9543438373977947, "grad_norm": 1.4182316241419564, "learning_rate": 1.0905392242729796e-07, "loss": 0.4707, "step": 26787 }, { "epoch": 0.9543794645242888, "grad_norm": 1.6971654164340029, "learning_rate": 1.0888403468740538e-07, "loss": 0.3883, "step": 26788 }, { "epoch": 0.9544150916507829, "grad_norm": 1.4307697436901525, "learning_rate": 1.0871427865463357e-07, "loss": 0.4737, "step": 26789 }, { "epoch": 0.954450718777277, "grad_norm": 1.8859177104910625, "learning_rate": 1.0854465433124628e-07, "loss": 0.4239, "step": 26790 }, { "epoch": 0.9544863459037711, "grad_norm": 1.3227440686893528, "learning_rate": 1.0837516171950057e-07, "loss": 0.3824, "step": 26791 }, { "epoch": 0.9545219730302652, "grad_norm": 1.2595377544906465, "learning_rate": 1.0820580082165244e-07, "loss": 0.3501, "step": 26792 }, { "epoch": 0.9545576001567594, "grad_norm": 2.1628740989053923, "learning_rate": 1.0803657163995896e-07, "loss": 0.5097, "step": 26793 }, { "epoch": 0.9545932272832535, "grad_norm": 1.239670751459682, "learning_rate": 1.0786747417667165e-07, "loss": 0.4227, "step": 26794 }, { "epoch": 0.9546288544097475, "grad_norm": 1.7657599918431839, "learning_rate": 1.0769850843404428e-07, "loss": 0.5247, "step": 26795 }, { "epoch": 0.9546644815362417, "grad_norm": 1.566813145114718, "learning_rate": 1.0752967441432394e-07, "loss": 0.4639, "step": 26796 }, { "epoch": 0.9547001086627358, "grad_norm": 1.3151148178897993, "learning_rate": 1.0736097211976216e-07, "loss": 0.3827, "step": 26797 }, { "epoch": 0.9547357357892299, "grad_norm": 1.343734059987567, "learning_rate": 1.0719240155260269e-07, "loss": 0.3715, "step": 26798 }, { "epoch": 0.9547713629157241, "grad_norm": 2.036540998041112, "learning_rate": 1.0702396271509153e-07, "loss": 0.4899, "step": 26799 }, { "epoch": 0.9548069900422181, "grad_norm": 1.799672744189902, "learning_rate": 1.0685565560947131e-07, "loss": 0.4903, "step": 26800 }, { "epoch": 0.9548426171687122, "grad_norm": 1.2547780608292214, "learning_rate": 1.0668748023798137e-07, "loss": 0.3729, "step": 26801 }, { "epoch": 0.9548782442952064, "grad_norm": 1.4623132824721379, "learning_rate": 1.0651943660286434e-07, "loss": 0.5669, "step": 26802 }, { "epoch": 0.9549138714217005, "grad_norm": 1.3005232857619677, "learning_rate": 1.0635152470635513e-07, "loss": 0.4048, "step": 26803 }, { "epoch": 0.9549494985481946, "grad_norm": 1.3163370328572512, "learning_rate": 1.061837445506908e-07, "loss": 0.3841, "step": 26804 }, { "epoch": 0.9549851256746887, "grad_norm": 1.2974271394229762, "learning_rate": 1.0601609613810515e-07, "loss": 0.4317, "step": 26805 }, { "epoch": 0.9550207528011828, "grad_norm": 1.2942923477267831, "learning_rate": 1.0584857947083194e-07, "loss": 0.3847, "step": 26806 }, { "epoch": 0.9550563799276769, "grad_norm": 1.6056165168757206, "learning_rate": 1.0568119455109938e-07, "loss": 0.4921, "step": 26807 }, { "epoch": 0.9550920070541711, "grad_norm": 1.6869815878828518, "learning_rate": 1.055139413811379e-07, "loss": 0.5366, "step": 26808 }, { "epoch": 0.9551276341806652, "grad_norm": 1.2312749082106667, "learning_rate": 1.0534681996317353e-07, "loss": 0.4451, "step": 26809 }, { "epoch": 0.9551632613071592, "grad_norm": 1.4753254590006792, "learning_rate": 1.0517983029943223e-07, "loss": 0.3826, "step": 26810 }, { "epoch": 0.9551988884336534, "grad_norm": 1.4568513207849159, "learning_rate": 1.0501297239213781e-07, "loss": 0.4788, "step": 26811 }, { "epoch": 0.9552345155601475, "grad_norm": 1.4711711492250519, "learning_rate": 1.048462462435107e-07, "loss": 0.5136, "step": 26812 }, { "epoch": 0.9552701426866416, "grad_norm": 1.6353237231390478, "learning_rate": 1.0467965185577356e-07, "loss": 0.4626, "step": 26813 }, { "epoch": 0.9553057698131358, "grad_norm": 1.4232467179049373, "learning_rate": 1.0451318923114351e-07, "loss": 0.4001, "step": 26814 }, { "epoch": 0.9553413969396298, "grad_norm": 1.1835864370203364, "learning_rate": 1.0434685837183656e-07, "loss": 0.4265, "step": 26815 }, { "epoch": 0.9553770240661239, "grad_norm": 1.219475033682205, "learning_rate": 1.041806592800676e-07, "loss": 0.3781, "step": 26816 }, { "epoch": 0.9554126511926181, "grad_norm": 1.7747372346353012, "learning_rate": 1.0401459195805041e-07, "loss": 0.5245, "step": 26817 }, { "epoch": 0.9554482783191122, "grad_norm": 1.431383973821308, "learning_rate": 1.0384865640799436e-07, "loss": 0.4375, "step": 26818 }, { "epoch": 0.9554839054456062, "grad_norm": 1.6970810941563337, "learning_rate": 1.0368285263211097e-07, "loss": 0.5233, "step": 26819 }, { "epoch": 0.9555195325721004, "grad_norm": 1.7911053388060885, "learning_rate": 1.0351718063260852e-07, "loss": 0.4968, "step": 26820 }, { "epoch": 0.9555551596985945, "grad_norm": 1.3104423016965843, "learning_rate": 1.0335164041169188e-07, "loss": 0.3763, "step": 26821 }, { "epoch": 0.9555907868250886, "grad_norm": 1.655444178105173, "learning_rate": 1.0318623197156486e-07, "loss": 0.5414, "step": 26822 }, { "epoch": 0.9556264139515828, "grad_norm": 1.3146171921215555, "learning_rate": 1.0302095531443013e-07, "loss": 0.4169, "step": 26823 }, { "epoch": 0.9556620410780768, "grad_norm": 1.3416730873168143, "learning_rate": 1.0285581044249037e-07, "loss": 0.399, "step": 26824 }, { "epoch": 0.9556976682045709, "grad_norm": 1.5986814135041303, "learning_rate": 1.0269079735794163e-07, "loss": 0.5219, "step": 26825 }, { "epoch": 0.9557332953310651, "grad_norm": 2.0635057137057253, "learning_rate": 1.0252591606298434e-07, "loss": 0.6029, "step": 26826 }, { "epoch": 0.9557689224575592, "grad_norm": 1.3550641013296925, "learning_rate": 1.0236116655981121e-07, "loss": 0.5017, "step": 26827 }, { "epoch": 0.9558045495840533, "grad_norm": 1.5745174694008282, "learning_rate": 1.0219654885061825e-07, "loss": 0.449, "step": 26828 }, { "epoch": 0.9558401767105474, "grad_norm": 1.2481751609477936, "learning_rate": 1.0203206293759482e-07, "loss": 0.3917, "step": 26829 }, { "epoch": 0.9558758038370415, "grad_norm": 1.3535210601183758, "learning_rate": 1.0186770882293474e-07, "loss": 0.4, "step": 26830 }, { "epoch": 0.9559114309635356, "grad_norm": 1.2732191003182558, "learning_rate": 1.0170348650882289e-07, "loss": 0.422, "step": 26831 }, { "epoch": 0.9559470580900298, "grad_norm": 1.5486252842828916, "learning_rate": 1.0153939599744866e-07, "loss": 0.4398, "step": 26832 }, { "epoch": 0.9559826852165239, "grad_norm": 1.516081963903114, "learning_rate": 1.0137543729099474e-07, "loss": 0.5106, "step": 26833 }, { "epoch": 0.9560183123430179, "grad_norm": 1.4586232054632349, "learning_rate": 1.0121161039164606e-07, "loss": 0.3791, "step": 26834 }, { "epoch": 0.9560539394695121, "grad_norm": 1.6536357560346315, "learning_rate": 1.010479153015842e-07, "loss": 0.5544, "step": 26835 }, { "epoch": 0.9560895665960062, "grad_norm": 1.6684848967828498, "learning_rate": 1.0088435202298741e-07, "loss": 0.5226, "step": 26836 }, { "epoch": 0.9561251937225003, "grad_norm": 1.2326130475452173, "learning_rate": 1.0072092055803618e-07, "loss": 0.3994, "step": 26837 }, { "epoch": 0.9561608208489945, "grad_norm": 1.7594592279771775, "learning_rate": 1.005576209089032e-07, "loss": 0.5045, "step": 26838 }, { "epoch": 0.9561964479754885, "grad_norm": 1.5357671572209954, "learning_rate": 1.0039445307776563e-07, "loss": 0.4338, "step": 26839 }, { "epoch": 0.9562320751019826, "grad_norm": 1.3684897952643782, "learning_rate": 1.0023141706679507e-07, "loss": 0.3798, "step": 26840 }, { "epoch": 0.9562677022284768, "grad_norm": 1.4268062049689894, "learning_rate": 1.000685128781631e-07, "loss": 0.4695, "step": 26841 }, { "epoch": 0.9563033293549709, "grad_norm": 1.181430932607078, "learning_rate": 9.9905740514038e-08, "loss": 0.4144, "step": 26842 }, { "epoch": 0.9563389564814649, "grad_norm": 2.0335971670215924, "learning_rate": 9.974309997658915e-08, "loss": 0.5282, "step": 26843 }, { "epoch": 0.9563745836079591, "grad_norm": 1.3970924497219779, "learning_rate": 9.958059126797925e-08, "loss": 0.5667, "step": 26844 }, { "epoch": 0.9564102107344532, "grad_norm": 2.8089992608262575, "learning_rate": 9.941821439037436e-08, "loss": 0.5164, "step": 26845 }, { "epoch": 0.9564458378609473, "grad_norm": 1.5859032770545036, "learning_rate": 9.925596934593607e-08, "loss": 0.5597, "step": 26846 }, { "epoch": 0.9564814649874415, "grad_norm": 1.3038751494666905, "learning_rate": 9.909385613682377e-08, "loss": 0.3953, "step": 26847 }, { "epoch": 0.9565170921139355, "grad_norm": 1.186727613993909, "learning_rate": 9.893187476519794e-08, "loss": 0.3917, "step": 26848 }, { "epoch": 0.9565527192404296, "grad_norm": 1.2114197715138393, "learning_rate": 9.877002523321466e-08, "loss": 0.4493, "step": 26849 }, { "epoch": 0.9565883463669238, "grad_norm": 1.5861445710756052, "learning_rate": 9.860830754302886e-08, "loss": 0.4756, "step": 26850 }, { "epoch": 0.9566239734934179, "grad_norm": 1.3165668993135706, "learning_rate": 9.844672169679437e-08, "loss": 0.4138, "step": 26851 }, { "epoch": 0.956659600619912, "grad_norm": 1.3257736571642202, "learning_rate": 9.82852676966628e-08, "loss": 0.372, "step": 26852 }, { "epoch": 0.9566952277464061, "grad_norm": 1.333534753953325, "learning_rate": 9.812394554478355e-08, "loss": 0.4099, "step": 26853 }, { "epoch": 0.9567308548729002, "grad_norm": 1.2841677571116674, "learning_rate": 9.79627552433049e-08, "loss": 0.4301, "step": 26854 }, { "epoch": 0.9567664819993943, "grad_norm": 1.3015318086787835, "learning_rate": 9.78016967943729e-08, "loss": 0.3865, "step": 26855 }, { "epoch": 0.9568021091258885, "grad_norm": 1.3085992311826997, "learning_rate": 9.764077020013252e-08, "loss": 0.4016, "step": 26856 }, { "epoch": 0.9568377362523826, "grad_norm": 1.2975164475270136, "learning_rate": 9.747997546272647e-08, "loss": 0.4289, "step": 26857 }, { "epoch": 0.9568733633788766, "grad_norm": 1.332018212611391, "learning_rate": 9.731931258429638e-08, "loss": 0.4259, "step": 26858 }, { "epoch": 0.9569089905053708, "grad_norm": 1.6198979999140555, "learning_rate": 9.715878156698166e-08, "loss": 0.5584, "step": 26859 }, { "epoch": 0.9569446176318649, "grad_norm": 1.596614170492006, "learning_rate": 9.699838241291948e-08, "loss": 0.5291, "step": 26860 }, { "epoch": 0.956980244758359, "grad_norm": 1.5550500608158921, "learning_rate": 9.68381151242448e-08, "loss": 0.5016, "step": 26861 }, { "epoch": 0.9570158718848532, "grad_norm": 1.2340240430306177, "learning_rate": 9.667797970309257e-08, "loss": 0.3345, "step": 26862 }, { "epoch": 0.9570514990113472, "grad_norm": 1.4250535741349193, "learning_rate": 9.651797615159552e-08, "loss": 0.4152, "step": 26863 }, { "epoch": 0.9570871261378413, "grad_norm": 1.500225668029574, "learning_rate": 9.63581044718831e-08, "loss": 0.4509, "step": 26864 }, { "epoch": 0.9571227532643355, "grad_norm": 1.3948967158383605, "learning_rate": 9.619836466608578e-08, "loss": 0.4466, "step": 26865 }, { "epoch": 0.9571583803908296, "grad_norm": 1.392975164146357, "learning_rate": 9.603875673632968e-08, "loss": 0.4688, "step": 26866 }, { "epoch": 0.9571940075173236, "grad_norm": 1.335299284138298, "learning_rate": 9.587928068473973e-08, "loss": 0.3968, "step": 26867 }, { "epoch": 0.9572296346438178, "grad_norm": 1.669737742445538, "learning_rate": 9.571993651343869e-08, "loss": 0.4922, "step": 26868 }, { "epoch": 0.9572652617703119, "grad_norm": 1.3024901878251143, "learning_rate": 9.556072422454932e-08, "loss": 0.4055, "step": 26869 }, { "epoch": 0.957300888896806, "grad_norm": 1.3148073278423313, "learning_rate": 9.540164382019324e-08, "loss": 0.421, "step": 26870 }, { "epoch": 0.9573365160233002, "grad_norm": 1.5409102712526437, "learning_rate": 9.524269530248542e-08, "loss": 0.4325, "step": 26871 }, { "epoch": 0.9573721431497942, "grad_norm": 1.5058957215253375, "learning_rate": 9.508387867354529e-08, "loss": 0.5266, "step": 26872 }, { "epoch": 0.9574077702762883, "grad_norm": 1.2937298712154854, "learning_rate": 9.492519393548672e-08, "loss": 0.396, "step": 26873 }, { "epoch": 0.9574433974027825, "grad_norm": 1.4656020769406624, "learning_rate": 9.476664109042134e-08, "loss": 0.4508, "step": 26874 }, { "epoch": 0.9574790245292766, "grad_norm": 2.237247003866749, "learning_rate": 9.46082201404619e-08, "loss": 0.535, "step": 26875 }, { "epoch": 0.9575146516557707, "grad_norm": 1.159403131519723, "learning_rate": 9.444993108771782e-08, "loss": 0.4034, "step": 26876 }, { "epoch": 0.9575502787822648, "grad_norm": 1.6756367400050731, "learning_rate": 9.429177393429634e-08, "loss": 0.4834, "step": 26877 }, { "epoch": 0.9575859059087589, "grad_norm": 1.3613483899684529, "learning_rate": 9.413374868230352e-08, "loss": 0.5089, "step": 26878 }, { "epoch": 0.957621533035253, "grad_norm": 1.247949181153694, "learning_rate": 9.397585533384324e-08, "loss": 0.3557, "step": 26879 }, { "epoch": 0.9576571601617472, "grad_norm": 1.4660017114278294, "learning_rate": 9.381809389101826e-08, "loss": 0.4626, "step": 26880 }, { "epoch": 0.9576927872882413, "grad_norm": 1.16834558281417, "learning_rate": 9.366046435593024e-08, "loss": 0.4217, "step": 26881 }, { "epoch": 0.9577284144147353, "grad_norm": 1.5316625305363019, "learning_rate": 9.350296673067638e-08, "loss": 0.5548, "step": 26882 }, { "epoch": 0.9577640415412295, "grad_norm": 1.4845034183774468, "learning_rate": 9.334560101735612e-08, "loss": 0.4939, "step": 26883 }, { "epoch": 0.9577996686677236, "grad_norm": 1.3710339336184285, "learning_rate": 9.318836721806223e-08, "loss": 0.4422, "step": 26884 }, { "epoch": 0.9578352957942177, "grad_norm": 1.2971753826530132, "learning_rate": 9.303126533488971e-08, "loss": 0.3934, "step": 26885 }, { "epoch": 0.9578709229207119, "grad_norm": 1.5100444874916699, "learning_rate": 9.28742953699302e-08, "loss": 0.4978, "step": 26886 }, { "epoch": 0.9579065500472059, "grad_norm": 1.868353523463661, "learning_rate": 9.271745732527537e-08, "loss": 0.5358, "step": 26887 }, { "epoch": 0.9579421771737, "grad_norm": 1.3570754419060456, "learning_rate": 9.256075120301134e-08, "loss": 0.477, "step": 26888 }, { "epoch": 0.9579778043001942, "grad_norm": 1.215651849352795, "learning_rate": 9.240417700522641e-08, "loss": 0.4174, "step": 26889 }, { "epoch": 0.9580134314266883, "grad_norm": 1.1354795380194251, "learning_rate": 9.224773473400451e-08, "loss": 0.4293, "step": 26890 }, { "epoch": 0.9580490585531823, "grad_norm": 1.3841079291163416, "learning_rate": 9.209142439142948e-08, "loss": 0.4479, "step": 26891 }, { "epoch": 0.9580846856796765, "grad_norm": 1.5986033397929855, "learning_rate": 9.193524597958193e-08, "loss": 0.518, "step": 26892 }, { "epoch": 0.9581203128061706, "grad_norm": 1.8503222563461057, "learning_rate": 9.177919950054237e-08, "loss": 0.5439, "step": 26893 }, { "epoch": 0.9581559399326647, "grad_norm": 1.1259329758043226, "learning_rate": 9.162328495638695e-08, "loss": 0.4245, "step": 26894 }, { "epoch": 0.9581915670591589, "grad_norm": 1.21897926287847, "learning_rate": 9.1467502349194e-08, "loss": 0.4079, "step": 26895 }, { "epoch": 0.9582271941856529, "grad_norm": 1.7057595177079694, "learning_rate": 9.13118516810374e-08, "loss": 0.4609, "step": 26896 }, { "epoch": 0.958262821312147, "grad_norm": 1.6521409971512022, "learning_rate": 9.115633295398884e-08, "loss": 0.529, "step": 26897 }, { "epoch": 0.9582984484386412, "grad_norm": 2.767964325237129, "learning_rate": 9.10009461701189e-08, "loss": 0.4752, "step": 26898 }, { "epoch": 0.9583340755651353, "grad_norm": 1.2812718527781959, "learning_rate": 9.084569133149812e-08, "loss": 0.3969, "step": 26899 }, { "epoch": 0.9583697026916294, "grad_norm": 1.352652437992868, "learning_rate": 9.069056844019264e-08, "loss": 0.4027, "step": 26900 }, { "epoch": 0.9584053298181235, "grad_norm": 1.6492245695773902, "learning_rate": 9.053557749826857e-08, "loss": 0.5243, "step": 26901 }, { "epoch": 0.9584409569446176, "grad_norm": 1.2318155440537741, "learning_rate": 9.038071850778985e-08, "loss": 0.4237, "step": 26902 }, { "epoch": 0.9584765840711117, "grad_norm": 1.6198262578344846, "learning_rate": 9.022599147081813e-08, "loss": 0.4932, "step": 26903 }, { "epoch": 0.9585122111976059, "grad_norm": 1.5832008004261133, "learning_rate": 9.007139638941287e-08, "loss": 0.5726, "step": 26904 }, { "epoch": 0.9585478383241, "grad_norm": 1.388338738029812, "learning_rate": 8.991693326563466e-08, "loss": 0.4174, "step": 26905 }, { "epoch": 0.958583465450594, "grad_norm": 1.669739900473776, "learning_rate": 8.976260210153964e-08, "loss": 0.4853, "step": 26906 }, { "epoch": 0.9586190925770882, "grad_norm": 1.202771498790581, "learning_rate": 8.960840289918172e-08, "loss": 0.3969, "step": 26907 }, { "epoch": 0.9586547197035823, "grad_norm": 2.1819764188309074, "learning_rate": 8.945433566061479e-08, "loss": 0.5371, "step": 26908 }, { "epoch": 0.9586903468300764, "grad_norm": 1.2449572972173248, "learning_rate": 8.930040038789057e-08, "loss": 0.3605, "step": 26909 }, { "epoch": 0.9587259739565706, "grad_norm": 1.1957835755063733, "learning_rate": 8.914659708305851e-08, "loss": 0.3946, "step": 26910 }, { "epoch": 0.9587616010830646, "grad_norm": 1.461668272291315, "learning_rate": 8.899292574816698e-08, "loss": 0.435, "step": 26911 }, { "epoch": 0.9587972282095587, "grad_norm": 1.2208443249936665, "learning_rate": 8.883938638526213e-08, "loss": 0.407, "step": 26912 }, { "epoch": 0.9588328553360529, "grad_norm": 1.2900611262012849, "learning_rate": 8.868597899638897e-08, "loss": 0.4347, "step": 26913 }, { "epoch": 0.958868482462547, "grad_norm": 1.2827433860056061, "learning_rate": 8.853270358358812e-08, "loss": 0.4052, "step": 26914 }, { "epoch": 0.958904109589041, "grad_norm": 1.935513343958655, "learning_rate": 8.837956014890236e-08, "loss": 0.5569, "step": 26915 }, { "epoch": 0.9589397367155352, "grad_norm": 1.4781480401894938, "learning_rate": 8.82265486943712e-08, "loss": 0.5148, "step": 26916 }, { "epoch": 0.9589753638420293, "grad_norm": 1.9569090430569858, "learning_rate": 8.807366922203075e-08, "loss": 0.6211, "step": 26917 }, { "epoch": 0.9590109909685234, "grad_norm": 2.3541917829062453, "learning_rate": 8.79209217339183e-08, "loss": 0.4512, "step": 26918 }, { "epoch": 0.9590466180950176, "grad_norm": 1.20266285928364, "learning_rate": 8.776830623206667e-08, "loss": 0.392, "step": 26919 }, { "epoch": 0.9590822452215116, "grad_norm": 1.3368848168029361, "learning_rate": 8.761582271850865e-08, "loss": 0.4659, "step": 26920 }, { "epoch": 0.9591178723480057, "grad_norm": 1.6741416506917213, "learning_rate": 8.746347119527377e-08, "loss": 0.5106, "step": 26921 }, { "epoch": 0.9591534994744999, "grad_norm": 1.31125642185202, "learning_rate": 8.731125166439036e-08, "loss": 0.4292, "step": 26922 }, { "epoch": 0.959189126600994, "grad_norm": 1.1595470128466818, "learning_rate": 8.715916412788683e-08, "loss": 0.3709, "step": 26923 }, { "epoch": 0.959224753727488, "grad_norm": 1.4856735482267396, "learning_rate": 8.700720858778822e-08, "loss": 0.4074, "step": 26924 }, { "epoch": 0.9592603808539822, "grad_norm": 1.242665846009789, "learning_rate": 8.685538504611734e-08, "loss": 0.4414, "step": 26925 }, { "epoch": 0.9592960079804763, "grad_norm": 1.429353017318209, "learning_rate": 8.670369350489482e-08, "loss": 0.3916, "step": 26926 }, { "epoch": 0.9593316351069704, "grad_norm": 1.7475634872762227, "learning_rate": 8.655213396614349e-08, "loss": 0.5183, "step": 26927 }, { "epoch": 0.9593672622334646, "grad_norm": 1.2132556659729976, "learning_rate": 8.640070643187837e-08, "loss": 0.3898, "step": 26928 }, { "epoch": 0.9594028893599587, "grad_norm": 1.2877980800917104, "learning_rate": 8.624941090411676e-08, "loss": 0.4057, "step": 26929 }, { "epoch": 0.9594385164864527, "grad_norm": 1.2732542789766255, "learning_rate": 8.609824738487371e-08, "loss": 0.3991, "step": 26930 }, { "epoch": 0.9594741436129469, "grad_norm": 1.4308940571443938, "learning_rate": 8.594721587616206e-08, "loss": 0.5577, "step": 26931 }, { "epoch": 0.959509770739441, "grad_norm": 1.3431929742308184, "learning_rate": 8.579631637999242e-08, "loss": 0.4159, "step": 26932 }, { "epoch": 0.9595453978659351, "grad_norm": 1.484864999843712, "learning_rate": 8.56455488983754e-08, "loss": 0.5067, "step": 26933 }, { "epoch": 0.9595810249924293, "grad_norm": 1.2834404779585478, "learning_rate": 8.549491343331718e-08, "loss": 0.3828, "step": 26934 }, { "epoch": 0.9596166521189233, "grad_norm": 1.564619711508438, "learning_rate": 8.534440998682392e-08, "loss": 0.5164, "step": 26935 }, { "epoch": 0.9596522792454174, "grad_norm": 1.8731867092464372, "learning_rate": 8.519403856089959e-08, "loss": 0.4809, "step": 26936 }, { "epoch": 0.9596879063719116, "grad_norm": 1.8003472363551016, "learning_rate": 8.504379915754702e-08, "loss": 0.5139, "step": 26937 }, { "epoch": 0.9597235334984057, "grad_norm": 1.1749542882874346, "learning_rate": 8.489369177876571e-08, "loss": 0.4141, "step": 26938 }, { "epoch": 0.9597591606248999, "grad_norm": 1.4027424527024892, "learning_rate": 8.474371642655521e-08, "loss": 0.4055, "step": 26939 }, { "epoch": 0.9597947877513939, "grad_norm": 1.6392314184781103, "learning_rate": 8.459387310291389e-08, "loss": 0.4884, "step": 26940 }, { "epoch": 0.959830414877888, "grad_norm": 1.3047483528572092, "learning_rate": 8.44441618098335e-08, "loss": 0.3734, "step": 26941 }, { "epoch": 0.9598660420043822, "grad_norm": 1.508415217690079, "learning_rate": 8.429458254931133e-08, "loss": 0.5844, "step": 26942 }, { "epoch": 0.9599016691308763, "grad_norm": 1.787819016995659, "learning_rate": 8.41451353233369e-08, "loss": 0.502, "step": 26943 }, { "epoch": 0.9599372962573703, "grad_norm": 1.808350049005673, "learning_rate": 8.399582013390084e-08, "loss": 0.4943, "step": 26944 }, { "epoch": 0.9599729233838645, "grad_norm": 1.6181728174748085, "learning_rate": 8.384663698299156e-08, "loss": 0.5307, "step": 26945 }, { "epoch": 0.9600085505103586, "grad_norm": 1.3352660353436423, "learning_rate": 8.369758587259413e-08, "loss": 0.4421, "step": 26946 }, { "epoch": 0.9600441776368527, "grad_norm": 1.3033445782482305, "learning_rate": 8.354866680469587e-08, "loss": 0.4107, "step": 26947 }, { "epoch": 0.9600798047633469, "grad_norm": 1.5325692572624903, "learning_rate": 8.33998797812774e-08, "loss": 0.5166, "step": 26948 }, { "epoch": 0.9601154318898409, "grad_norm": 1.2983973838831224, "learning_rate": 8.325122480432046e-08, "loss": 0.5152, "step": 26949 }, { "epoch": 0.960151059016335, "grad_norm": 1.1483285751877703, "learning_rate": 8.310270187580572e-08, "loss": 0.3805, "step": 26950 }, { "epoch": 0.9601866861428292, "grad_norm": 1.4979702587767916, "learning_rate": 8.295431099771045e-08, "loss": 0.4377, "step": 26951 }, { "epoch": 0.9602223132693233, "grad_norm": 1.7725117350170003, "learning_rate": 8.280605217200866e-08, "loss": 0.4862, "step": 26952 }, { "epoch": 0.9602579403958174, "grad_norm": 1.2959345428518143, "learning_rate": 8.265792540067653e-08, "loss": 0.3991, "step": 26953 }, { "epoch": 0.9602935675223115, "grad_norm": 1.2246052967397618, "learning_rate": 8.25099306856858e-08, "loss": 0.4013, "step": 26954 }, { "epoch": 0.9603291946488056, "grad_norm": 1.3716433591929098, "learning_rate": 8.236206802900826e-08, "loss": 0.3902, "step": 26955 }, { "epoch": 0.9603648217752997, "grad_norm": 1.427759646717892, "learning_rate": 8.22143374326112e-08, "loss": 0.5312, "step": 26956 }, { "epoch": 0.9604004489017939, "grad_norm": 1.154761215146799, "learning_rate": 8.206673889846306e-08, "loss": 0.3681, "step": 26957 }, { "epoch": 0.960436076028288, "grad_norm": 1.2867139451949678, "learning_rate": 8.191927242852671e-08, "loss": 0.4082, "step": 26958 }, { "epoch": 0.960471703154782, "grad_norm": 1.1956147780521207, "learning_rate": 8.177193802476945e-08, "loss": 0.421, "step": 26959 }, { "epoch": 0.9605073302812762, "grad_norm": 1.4227711458876602, "learning_rate": 8.162473568914975e-08, "loss": 0.4137, "step": 26960 }, { "epoch": 0.9605429574077703, "grad_norm": 1.0970329678093373, "learning_rate": 8.147766542362934e-08, "loss": 0.3528, "step": 26961 }, { "epoch": 0.9605785845342644, "grad_norm": 1.466788167994413, "learning_rate": 8.133072723016777e-08, "loss": 0.4304, "step": 26962 }, { "epoch": 0.9606142116607586, "grad_norm": 1.541454916894632, "learning_rate": 8.118392111071905e-08, "loss": 0.4622, "step": 26963 }, { "epoch": 0.9606498387872526, "grad_norm": 1.7075657224946357, "learning_rate": 8.103724706724048e-08, "loss": 0.5582, "step": 26964 }, { "epoch": 0.9606854659137467, "grad_norm": 1.6059568521927445, "learning_rate": 8.089070510168273e-08, "loss": 0.4093, "step": 26965 }, { "epoch": 0.9607210930402409, "grad_norm": 1.3477864768800034, "learning_rate": 8.074429521599869e-08, "loss": 0.3903, "step": 26966 }, { "epoch": 0.960756720166735, "grad_norm": 1.8053149069906644, "learning_rate": 8.059801741213791e-08, "loss": 0.5061, "step": 26967 }, { "epoch": 0.960792347293229, "grad_norm": 1.697371156966558, "learning_rate": 8.04518716920466e-08, "loss": 0.4998, "step": 26968 }, { "epoch": 0.9608279744197232, "grad_norm": 1.5174104209388186, "learning_rate": 8.030585805767321e-08, "loss": 0.4599, "step": 26969 }, { "epoch": 0.9608636015462173, "grad_norm": 1.4124091766362021, "learning_rate": 8.015997651096064e-08, "loss": 0.3848, "step": 26970 }, { "epoch": 0.9608992286727114, "grad_norm": 1.3757878489245803, "learning_rate": 8.001422705385064e-08, "loss": 0.4141, "step": 26971 }, { "epoch": 0.9609348557992056, "grad_norm": 1.3628299191506725, "learning_rate": 7.986860968828613e-08, "loss": 0.4411, "step": 26972 }, { "epoch": 0.9609704829256996, "grad_norm": 1.0853120047747487, "learning_rate": 7.972312441620444e-08, "loss": 0.3847, "step": 26973 }, { "epoch": 0.9610061100521937, "grad_norm": 1.3559368937035337, "learning_rate": 7.95777712395429e-08, "loss": 0.3842, "step": 26974 }, { "epoch": 0.9610417371786879, "grad_norm": 2.0328106649569966, "learning_rate": 7.943255016023665e-08, "loss": 0.4524, "step": 26975 }, { "epoch": 0.961077364305182, "grad_norm": 1.3834463072864027, "learning_rate": 7.92874611802208e-08, "loss": 0.3969, "step": 26976 }, { "epoch": 0.9611129914316761, "grad_norm": 1.2322667767436528, "learning_rate": 7.914250430142711e-08, "loss": 0.3792, "step": 26977 }, { "epoch": 0.9611486185581702, "grad_norm": 1.29046607715333, "learning_rate": 7.899767952578407e-08, "loss": 0.4057, "step": 26978 }, { "epoch": 0.9611842456846643, "grad_norm": 1.9894828906715187, "learning_rate": 7.885298685522235e-08, "loss": 0.4446, "step": 26979 }, { "epoch": 0.9612198728111584, "grad_norm": 1.4960936583904143, "learning_rate": 7.870842629166709e-08, "loss": 0.5132, "step": 26980 }, { "epoch": 0.9612554999376526, "grad_norm": 1.3807004354611114, "learning_rate": 7.856399783704339e-08, "loss": 0.4191, "step": 26981 }, { "epoch": 0.9612911270641467, "grad_norm": 1.2263096555266013, "learning_rate": 7.841970149327527e-08, "loss": 0.3902, "step": 26982 }, { "epoch": 0.9613267541906407, "grad_norm": 1.364699696771957, "learning_rate": 7.827553726228343e-08, "loss": 0.4392, "step": 26983 }, { "epoch": 0.9613623813171349, "grad_norm": 2.049568963602851, "learning_rate": 7.813150514598744e-08, "loss": 0.5388, "step": 26984 }, { "epoch": 0.961398008443629, "grad_norm": 1.9117726081535558, "learning_rate": 7.798760514630466e-08, "loss": 0.5419, "step": 26985 }, { "epoch": 0.9614336355701231, "grad_norm": 1.2109917317205765, "learning_rate": 7.784383726515355e-08, "loss": 0.384, "step": 26986 }, { "epoch": 0.9614692626966173, "grad_norm": 1.4731116125552886, "learning_rate": 7.770020150444591e-08, "loss": 0.5179, "step": 26987 }, { "epoch": 0.9615048898231113, "grad_norm": 1.320717631322011, "learning_rate": 7.755669786609688e-08, "loss": 0.4984, "step": 26988 }, { "epoch": 0.9615405169496054, "grad_norm": 1.3784169085050775, "learning_rate": 7.741332635201381e-08, "loss": 0.3735, "step": 26989 }, { "epoch": 0.9615761440760996, "grad_norm": 1.312613816959116, "learning_rate": 7.727008696410854e-08, "loss": 0.4206, "step": 26990 }, { "epoch": 0.9616117712025937, "grad_norm": 1.784354739095203, "learning_rate": 7.712697970428729e-08, "loss": 0.4604, "step": 26991 }, { "epoch": 0.9616473983290877, "grad_norm": 1.5611290386434808, "learning_rate": 7.698400457445742e-08, "loss": 0.5252, "step": 26992 }, { "epoch": 0.9616830254555819, "grad_norm": 1.3005094883410697, "learning_rate": 7.684116157651967e-08, "loss": 0.4161, "step": 26993 }, { "epoch": 0.961718652582076, "grad_norm": 1.5785253430755282, "learning_rate": 7.669845071237913e-08, "loss": 0.5527, "step": 26994 }, { "epoch": 0.9617542797085701, "grad_norm": 1.2855045823935762, "learning_rate": 7.655587198393321e-08, "loss": 0.401, "step": 26995 }, { "epoch": 0.9617899068350643, "grad_norm": 1.8741099520191737, "learning_rate": 7.641342539308261e-08, "loss": 0.4985, "step": 26996 }, { "epoch": 0.9618255339615583, "grad_norm": 1.2708880927091402, "learning_rate": 7.627111094172356e-08, "loss": 0.4191, "step": 26997 }, { "epoch": 0.9618611610880524, "grad_norm": 1.6810305034226236, "learning_rate": 7.612892863175014e-08, "loss": 0.4888, "step": 26998 }, { "epoch": 0.9618967882145466, "grad_norm": 1.504514994574101, "learning_rate": 7.598687846505748e-08, "loss": 0.396, "step": 26999 }, { "epoch": 0.9619324153410407, "grad_norm": 1.509851570952804, "learning_rate": 7.584496044353517e-08, "loss": 0.5348, "step": 27000 }, { "epoch": 0.9619680424675348, "grad_norm": 1.2856872407422406, "learning_rate": 7.570317456907506e-08, "loss": 0.3761, "step": 27001 }, { "epoch": 0.9620036695940289, "grad_norm": 1.562799908722585, "learning_rate": 7.556152084356227e-08, "loss": 0.5073, "step": 27002 }, { "epoch": 0.962039296720523, "grad_norm": 1.7942144006577694, "learning_rate": 7.541999926888643e-08, "loss": 0.5516, "step": 27003 }, { "epoch": 0.9620749238470171, "grad_norm": 1.321741634392953, "learning_rate": 7.527860984692825e-08, "loss": 0.4205, "step": 27004 }, { "epoch": 0.9621105509735113, "grad_norm": 1.1455338896268925, "learning_rate": 7.513735257957288e-08, "loss": 0.4024, "step": 27005 }, { "epoch": 0.9621461781000054, "grad_norm": 1.2824626024510992, "learning_rate": 7.499622746870106e-08, "loss": 0.4073, "step": 27006 }, { "epoch": 0.9621818052264994, "grad_norm": 1.4604625742166255, "learning_rate": 7.485523451619125e-08, "loss": 0.5116, "step": 27007 }, { "epoch": 0.9622174323529936, "grad_norm": 1.571133282516712, "learning_rate": 7.471437372392198e-08, "loss": 0.5218, "step": 27008 }, { "epoch": 0.9622530594794877, "grad_norm": 1.5132149346342998, "learning_rate": 7.457364509376841e-08, "loss": 0.528, "step": 27009 }, { "epoch": 0.9622886866059818, "grad_norm": 1.3691509868497487, "learning_rate": 7.443304862760348e-08, "loss": 0.4212, "step": 27010 }, { "epoch": 0.962324313732476, "grad_norm": 2.092611435696499, "learning_rate": 7.429258432730014e-08, "loss": 0.5357, "step": 27011 }, { "epoch": 0.96235994085897, "grad_norm": 1.3389720329175474, "learning_rate": 7.415225219472799e-08, "loss": 0.3808, "step": 27012 }, { "epoch": 0.9623955679854641, "grad_norm": 1.2976581984973816, "learning_rate": 7.401205223175777e-08, "loss": 0.3914, "step": 27013 }, { "epoch": 0.9624311951119583, "grad_norm": 1.3009442817390964, "learning_rate": 7.387198444025467e-08, "loss": 0.393, "step": 27014 }, { "epoch": 0.9624668222384524, "grad_norm": 1.473187135145427, "learning_rate": 7.373204882208273e-08, "loss": 0.5814, "step": 27015 }, { "epoch": 0.9625024493649464, "grad_norm": 1.44402669706914, "learning_rate": 7.359224537910825e-08, "loss": 0.3924, "step": 27016 }, { "epoch": 0.9625380764914406, "grad_norm": 1.5875605796829104, "learning_rate": 7.345257411318973e-08, "loss": 0.4997, "step": 27017 }, { "epoch": 0.9625737036179347, "grad_norm": 1.2294589085282943, "learning_rate": 7.331303502618903e-08, "loss": 0.3732, "step": 27018 }, { "epoch": 0.9626093307444288, "grad_norm": 1.514214847364217, "learning_rate": 7.317362811996465e-08, "loss": 0.4742, "step": 27019 }, { "epoch": 0.962644957870923, "grad_norm": 1.5924451873182377, "learning_rate": 7.303435339637066e-08, "loss": 0.5038, "step": 27020 }, { "epoch": 0.962680584997417, "grad_norm": 1.4749201041957365, "learning_rate": 7.289521085726226e-08, "loss": 0.4775, "step": 27021 }, { "epoch": 0.9627162121239111, "grad_norm": 1.2883739127655205, "learning_rate": 7.27562005044935e-08, "loss": 0.3796, "step": 27022 }, { "epoch": 0.9627518392504053, "grad_norm": 1.7104652330914343, "learning_rate": 7.261732233991514e-08, "loss": 0.5052, "step": 27023 }, { "epoch": 0.9627874663768994, "grad_norm": 2.246544426181968, "learning_rate": 7.24785763653757e-08, "loss": 0.4819, "step": 27024 }, { "epoch": 0.9628230935033935, "grad_norm": 1.208265792261317, "learning_rate": 7.233996258272258e-08, "loss": 0.39, "step": 27025 }, { "epoch": 0.9628587206298876, "grad_norm": 1.6448679976295106, "learning_rate": 7.220148099380208e-08, "loss": 0.4608, "step": 27026 }, { "epoch": 0.9628943477563817, "grad_norm": 1.752881882936397, "learning_rate": 7.206313160045831e-08, "loss": 0.4879, "step": 27027 }, { "epoch": 0.9629299748828758, "grad_norm": 1.2724429700743733, "learning_rate": 7.192491440453197e-08, "loss": 0.4309, "step": 27028 }, { "epoch": 0.96296560200937, "grad_norm": 1.5799885432734149, "learning_rate": 7.178682940786608e-08, "loss": 0.539, "step": 27029 }, { "epoch": 0.9630012291358641, "grad_norm": 1.3305283255002576, "learning_rate": 7.164887661229691e-08, "loss": 0.4334, "step": 27030 }, { "epoch": 0.9630368562623581, "grad_norm": 1.2921687730695282, "learning_rate": 7.151105601966301e-08, "loss": 0.3849, "step": 27031 }, { "epoch": 0.9630724833888523, "grad_norm": 1.3554039555546056, "learning_rate": 7.137336763179958e-08, "loss": 0.4168, "step": 27032 }, { "epoch": 0.9631081105153464, "grad_norm": 1.325071684838973, "learning_rate": 7.123581145053849e-08, "loss": 0.3965, "step": 27033 }, { "epoch": 0.9631437376418405, "grad_norm": 1.8010852653381204, "learning_rate": 7.109838747771269e-08, "loss": 0.5133, "step": 27034 }, { "epoch": 0.9631793647683347, "grad_norm": 1.1832774922246607, "learning_rate": 7.096109571515075e-08, "loss": 0.3873, "step": 27035 }, { "epoch": 0.9632149918948287, "grad_norm": 2.572306286609882, "learning_rate": 7.082393616468342e-08, "loss": 0.4858, "step": 27036 }, { "epoch": 0.9632506190213228, "grad_norm": 1.2988305534966105, "learning_rate": 7.068690882813478e-08, "loss": 0.4109, "step": 27037 }, { "epoch": 0.963286246147817, "grad_norm": 1.5923621641365733, "learning_rate": 7.055001370733117e-08, "loss": 0.4769, "step": 27038 }, { "epoch": 0.9633218732743111, "grad_norm": 1.5925894798258569, "learning_rate": 7.041325080409223e-08, "loss": 0.5916, "step": 27039 }, { "epoch": 0.9633575004008051, "grad_norm": 1.2793457134410735, "learning_rate": 7.027662012024317e-08, "loss": 0.4507, "step": 27040 }, { "epoch": 0.9633931275272993, "grad_norm": 1.1825374380121974, "learning_rate": 7.014012165760031e-08, "loss": 0.3517, "step": 27041 }, { "epoch": 0.9634287546537934, "grad_norm": 1.6496843675177728, "learning_rate": 7.00037554179822e-08, "loss": 0.4832, "step": 27042 }, { "epoch": 0.9634643817802875, "grad_norm": 1.2428225866430727, "learning_rate": 6.986752140320519e-08, "loss": 0.4403, "step": 27043 }, { "epoch": 0.9635000089067817, "grad_norm": 1.1767841787322382, "learning_rate": 6.973141961508223e-08, "loss": 0.3831, "step": 27044 }, { "epoch": 0.9635356360332757, "grad_norm": 1.3899782305174748, "learning_rate": 6.959545005542745e-08, "loss": 0.4151, "step": 27045 }, { "epoch": 0.9635712631597698, "grad_norm": 1.8719609210353305, "learning_rate": 6.94596127260494e-08, "loss": 0.596, "step": 27046 }, { "epoch": 0.963606890286264, "grad_norm": 1.602892923490993, "learning_rate": 6.932390762875884e-08, "loss": 0.4485, "step": 27047 }, { "epoch": 0.9636425174127581, "grad_norm": 1.6764414096236633, "learning_rate": 6.91883347653599e-08, "loss": 0.4512, "step": 27048 }, { "epoch": 0.9636781445392522, "grad_norm": 1.6211207123041864, "learning_rate": 6.905289413766004e-08, "loss": 0.5391, "step": 27049 }, { "epoch": 0.9637137716657463, "grad_norm": 1.1952500042118992, "learning_rate": 6.891758574746221e-08, "loss": 0.3714, "step": 27050 }, { "epoch": 0.9637493987922404, "grad_norm": 1.1336293986003905, "learning_rate": 6.878240959656835e-08, "loss": 0.3822, "step": 27051 }, { "epoch": 0.9637850259187345, "grad_norm": 1.4034695406388766, "learning_rate": 6.86473656867781e-08, "loss": 0.3878, "step": 27052 }, { "epoch": 0.9638206530452287, "grad_norm": 1.2915814373730314, "learning_rate": 6.851245401989004e-08, "loss": 0.3912, "step": 27053 }, { "epoch": 0.9638562801717228, "grad_norm": 1.3090900715222358, "learning_rate": 6.83776745977005e-08, "loss": 0.4134, "step": 27054 }, { "epoch": 0.9638919072982168, "grad_norm": 1.3199741350464589, "learning_rate": 6.824302742200251e-08, "loss": 0.3857, "step": 27055 }, { "epoch": 0.963927534424711, "grad_norm": 1.358111394144254, "learning_rate": 6.810851249459238e-08, "loss": 0.4295, "step": 27056 }, { "epoch": 0.9639631615512051, "grad_norm": 1.4503253121012258, "learning_rate": 6.797412981725871e-08, "loss": 0.3852, "step": 27057 }, { "epoch": 0.9639987886776992, "grad_norm": 1.238070273715133, "learning_rate": 6.783987939179115e-08, "loss": 0.3987, "step": 27058 }, { "epoch": 0.9640344158041934, "grad_norm": 1.2550703203328017, "learning_rate": 6.77057612199783e-08, "loss": 0.4388, "step": 27059 }, { "epoch": 0.9640700429306874, "grad_norm": 3.052977056965119, "learning_rate": 6.757177530360536e-08, "loss": 0.5112, "step": 27060 }, { "epoch": 0.9641056700571815, "grad_norm": 1.7416357036766468, "learning_rate": 6.743792164445651e-08, "loss": 0.5573, "step": 27061 }, { "epoch": 0.9641412971836757, "grad_norm": 1.49694377784755, "learning_rate": 6.73042002443136e-08, "loss": 0.5118, "step": 27062 }, { "epoch": 0.9641769243101698, "grad_norm": 1.327838902008981, "learning_rate": 6.71706111049586e-08, "loss": 0.3935, "step": 27063 }, { "epoch": 0.9642125514366638, "grad_norm": 1.6387365815342922, "learning_rate": 6.703715422816893e-08, "loss": 0.4978, "step": 27064 }, { "epoch": 0.964248178563158, "grad_norm": 1.5772121293411034, "learning_rate": 6.690382961572317e-08, "loss": 0.4325, "step": 27065 }, { "epoch": 0.9642838056896521, "grad_norm": 1.8465764579487012, "learning_rate": 6.677063726939437e-08, "loss": 0.465, "step": 27066 }, { "epoch": 0.9643194328161462, "grad_norm": 1.2719732020400893, "learning_rate": 6.663757719095887e-08, "loss": 0.4118, "step": 27067 }, { "epoch": 0.9643550599426404, "grad_norm": 1.3628293508800842, "learning_rate": 6.650464938218637e-08, "loss": 0.415, "step": 27068 }, { "epoch": 0.9643906870691344, "grad_norm": 1.7819702033292957, "learning_rate": 6.637185384484767e-08, "loss": 0.5364, "step": 27069 }, { "epoch": 0.9644263141956285, "grad_norm": 1.433282373923711, "learning_rate": 6.623919058071138e-08, "loss": 0.3995, "step": 27070 }, { "epoch": 0.9644619413221227, "grad_norm": 1.2971666154755455, "learning_rate": 6.610665959154272e-08, "loss": 0.3912, "step": 27071 }, { "epoch": 0.9644975684486168, "grad_norm": 1.2540743918676063, "learning_rate": 6.597426087910697e-08, "loss": 0.4238, "step": 27072 }, { "epoch": 0.9645331955751109, "grad_norm": 1.18722044756073, "learning_rate": 6.584199444516826e-08, "loss": 0.3879, "step": 27073 }, { "epoch": 0.964568822701605, "grad_norm": 1.2542262367998156, "learning_rate": 6.57098602914874e-08, "loss": 0.3922, "step": 27074 }, { "epoch": 0.9646044498280991, "grad_norm": 1.225010351949252, "learning_rate": 6.557785841982301e-08, "loss": 0.3949, "step": 27075 }, { "epoch": 0.9646400769545932, "grad_norm": 1.571112670020042, "learning_rate": 6.544598883193253e-08, "loss": 0.5185, "step": 27076 }, { "epoch": 0.9646757040810874, "grad_norm": 1.7162449001536784, "learning_rate": 6.531425152957238e-08, "loss": 0.4827, "step": 27077 }, { "epoch": 0.9647113312075815, "grad_norm": 1.1906163797750111, "learning_rate": 6.51826465144978e-08, "loss": 0.4156, "step": 27078 }, { "epoch": 0.9647469583340755, "grad_norm": 1.6083555958917404, "learning_rate": 6.505117378846071e-08, "loss": 0.3866, "step": 27079 }, { "epoch": 0.9647825854605697, "grad_norm": 1.6929323637499205, "learning_rate": 6.491983335320973e-08, "loss": 0.4447, "step": 27080 }, { "epoch": 0.9648182125870638, "grad_norm": 1.536513577131366, "learning_rate": 6.478862521049678e-08, "loss": 0.4595, "step": 27081 }, { "epoch": 0.9648538397135579, "grad_norm": 1.5925153573639643, "learning_rate": 6.465754936206715e-08, "loss": 0.4927, "step": 27082 }, { "epoch": 0.9648894668400521, "grad_norm": 1.4938327979226338, "learning_rate": 6.452660580966608e-08, "loss": 0.408, "step": 27083 }, { "epoch": 0.9649250939665461, "grad_norm": 1.971287582115843, "learning_rate": 6.439579455503886e-08, "loss": 0.4994, "step": 27084 }, { "epoch": 0.9649607210930402, "grad_norm": 1.6980548840807885, "learning_rate": 6.426511559992521e-08, "loss": 0.5137, "step": 27085 }, { "epoch": 0.9649963482195344, "grad_norm": 1.1771930900641514, "learning_rate": 6.413456894606595e-08, "loss": 0.3976, "step": 27086 }, { "epoch": 0.9650319753460285, "grad_norm": 1.5606212661432466, "learning_rate": 6.400415459520081e-08, "loss": 0.4271, "step": 27087 }, { "epoch": 0.9650676024725225, "grad_norm": 1.5548054132071603, "learning_rate": 6.387387254906397e-08, "loss": 0.4243, "step": 27088 }, { "epoch": 0.9651032295990167, "grad_norm": 1.4355500065957196, "learning_rate": 6.37437228093929e-08, "loss": 0.3949, "step": 27089 }, { "epoch": 0.9651388567255108, "grad_norm": 1.5887219092812719, "learning_rate": 6.361370537791845e-08, "loss": 0.5001, "step": 27090 }, { "epoch": 0.9651744838520049, "grad_norm": 1.3036239348104353, "learning_rate": 6.348382025637256e-08, "loss": 0.3989, "step": 27091 }, { "epoch": 0.9652101109784991, "grad_norm": 1.559192314461536, "learning_rate": 6.335406744648498e-08, "loss": 0.3963, "step": 27092 }, { "epoch": 0.9652457381049931, "grad_norm": 1.2674160519306792, "learning_rate": 6.32244469499832e-08, "loss": 0.3537, "step": 27093 }, { "epoch": 0.9652813652314872, "grad_norm": 1.2894922677851162, "learning_rate": 6.309495876859362e-08, "loss": 0.4167, "step": 27094 }, { "epoch": 0.9653169923579814, "grad_norm": 2.131469493150862, "learning_rate": 6.29656029040393e-08, "loss": 0.4991, "step": 27095 }, { "epoch": 0.9653526194844755, "grad_norm": 1.4523098949651305, "learning_rate": 6.283637935804443e-08, "loss": 0.3853, "step": 27096 }, { "epoch": 0.9653882466109696, "grad_norm": 1.9713323298170184, "learning_rate": 6.270728813232873e-08, "loss": 0.5733, "step": 27097 }, { "epoch": 0.9654238737374637, "grad_norm": 1.2859597949662023, "learning_rate": 6.257832922861084e-08, "loss": 0.4226, "step": 27098 }, { "epoch": 0.9654595008639578, "grad_norm": 1.3573090450467589, "learning_rate": 6.244950264860828e-08, "loss": 0.3991, "step": 27099 }, { "epoch": 0.9654951279904519, "grad_norm": 1.628005437462236, "learning_rate": 6.232080839403631e-08, "loss": 0.4793, "step": 27100 }, { "epoch": 0.9655307551169461, "grad_norm": 1.9581299718359355, "learning_rate": 6.219224646660915e-08, "loss": 0.4856, "step": 27101 }, { "epoch": 0.9655663822434402, "grad_norm": 1.2891275390749586, "learning_rate": 6.206381686803876e-08, "loss": 0.4536, "step": 27102 }, { "epoch": 0.9656020093699342, "grad_norm": 1.3093290082299958, "learning_rate": 6.193551960003374e-08, "loss": 0.4045, "step": 27103 }, { "epoch": 0.9656376364964284, "grad_norm": 1.6015813169223825, "learning_rate": 6.180735466430498e-08, "loss": 0.4623, "step": 27104 }, { "epoch": 0.9656732636229225, "grad_norm": 1.222031590395764, "learning_rate": 6.167932206255666e-08, "loss": 0.4398, "step": 27105 }, { "epoch": 0.9657088907494166, "grad_norm": 1.286378913986048, "learning_rate": 6.15514217964952e-08, "loss": 0.4008, "step": 27106 }, { "epoch": 0.9657445178759108, "grad_norm": 2.2553241825197556, "learning_rate": 6.142365386782257e-08, "loss": 0.4088, "step": 27107 }, { "epoch": 0.9657801450024048, "grad_norm": 1.3127096304647132, "learning_rate": 6.129601827824183e-08, "loss": 0.4064, "step": 27108 }, { "epoch": 0.9658157721288989, "grad_norm": 1.36326954889097, "learning_rate": 6.116851502945054e-08, "loss": 0.3932, "step": 27109 }, { "epoch": 0.9658513992553931, "grad_norm": 1.3982806492923725, "learning_rate": 6.10411441231462e-08, "loss": 0.444, "step": 27110 }, { "epoch": 0.9658870263818872, "grad_norm": 1.5061602613413956, "learning_rate": 6.091390556102749e-08, "loss": 0.4754, "step": 27111 }, { "epoch": 0.9659226535083812, "grad_norm": 1.9221350735827183, "learning_rate": 6.078679934478637e-08, "loss": 0.5482, "step": 27112 }, { "epoch": 0.9659582806348754, "grad_norm": 1.2461221350038194, "learning_rate": 6.065982547611704e-08, "loss": 0.3892, "step": 27113 }, { "epoch": 0.9659939077613695, "grad_norm": 1.2743217172143455, "learning_rate": 6.053298395670814e-08, "loss": 0.3882, "step": 27114 }, { "epoch": 0.9660295348878636, "grad_norm": 1.622231349222723, "learning_rate": 6.040627478825056e-08, "loss": 0.469, "step": 27115 }, { "epoch": 0.9660651620143578, "grad_norm": 1.6227163154989819, "learning_rate": 6.027969797243072e-08, "loss": 0.5134, "step": 27116 }, { "epoch": 0.9661007891408518, "grad_norm": 1.79242679722813, "learning_rate": 6.015325351093393e-08, "loss": 0.4672, "step": 27117 }, { "epoch": 0.9661364162673459, "grad_norm": 1.2688269252412905, "learning_rate": 6.00269414054433e-08, "loss": 0.3724, "step": 27118 }, { "epoch": 0.9661720433938401, "grad_norm": 1.834849832790707, "learning_rate": 5.990076165764302e-08, "loss": 0.5217, "step": 27119 }, { "epoch": 0.9662076705203342, "grad_norm": 1.5389251440377023, "learning_rate": 5.977471426921067e-08, "loss": 0.4954, "step": 27120 }, { "epoch": 0.9662432976468283, "grad_norm": 1.1610131826922512, "learning_rate": 5.964879924182709e-08, "loss": 0.3759, "step": 27121 }, { "epoch": 0.9662789247733224, "grad_norm": 1.2704842764638795, "learning_rate": 5.9523016577166525e-08, "loss": 0.4121, "step": 27122 }, { "epoch": 0.9663145518998165, "grad_norm": 1.5350355940658218, "learning_rate": 5.93973662769054e-08, "loss": 0.4546, "step": 27123 }, { "epoch": 0.9663501790263106, "grad_norm": 1.6636784405168665, "learning_rate": 5.927184834271571e-08, "loss": 0.4621, "step": 27124 }, { "epoch": 0.9663858061528048, "grad_norm": 1.9913939756904562, "learning_rate": 5.9146462776270565e-08, "loss": 0.456, "step": 27125 }, { "epoch": 0.9664214332792989, "grad_norm": 2.227392035045058, "learning_rate": 5.9021209579238624e-08, "loss": 0.5035, "step": 27126 }, { "epoch": 0.9664570604057929, "grad_norm": 1.2508280627176074, "learning_rate": 5.889608875328634e-08, "loss": 0.4113, "step": 27127 }, { "epoch": 0.9664926875322871, "grad_norm": 2.2568990052840743, "learning_rate": 5.8771100300082375e-08, "loss": 0.5579, "step": 27128 }, { "epoch": 0.9665283146587812, "grad_norm": 1.2824912629212795, "learning_rate": 5.864624422128873e-08, "loss": 0.4003, "step": 27129 }, { "epoch": 0.9665639417852753, "grad_norm": 1.7294209089487438, "learning_rate": 5.852152051856963e-08, "loss": 0.5116, "step": 27130 }, { "epoch": 0.9665995689117695, "grad_norm": 1.3523614602159824, "learning_rate": 5.8396929193584864e-08, "loss": 0.4326, "step": 27131 }, { "epoch": 0.9666351960382635, "grad_norm": 1.2501011620111302, "learning_rate": 5.8272470247994205e-08, "loss": 0.3318, "step": 27132 }, { "epoch": 0.9666708231647576, "grad_norm": 3.394019162155023, "learning_rate": 5.814814368345412e-08, "loss": 0.5244, "step": 27133 }, { "epoch": 0.9667064502912518, "grad_norm": 1.3354951124890602, "learning_rate": 5.802394950161994e-08, "loss": 0.4015, "step": 27134 }, { "epoch": 0.9667420774177459, "grad_norm": 1.2289194836061743, "learning_rate": 5.7899887704145897e-08, "loss": 0.3822, "step": 27135 }, { "epoch": 0.96677770454424, "grad_norm": 1.353464642323093, "learning_rate": 5.777595829268512e-08, "loss": 0.4157, "step": 27136 }, { "epoch": 0.9668133316707341, "grad_norm": 1.618261469361684, "learning_rate": 5.765216126888629e-08, "loss": 0.5433, "step": 27137 }, { "epoch": 0.9668489587972282, "grad_norm": 1.5540842219364255, "learning_rate": 5.752849663439697e-08, "loss": 0.4743, "step": 27138 }, { "epoch": 0.9668845859237223, "grad_norm": 1.5044118189422915, "learning_rate": 5.740496439086585e-08, "loss": 0.5409, "step": 27139 }, { "epoch": 0.9669202130502165, "grad_norm": 1.4687703435694943, "learning_rate": 5.7281564539937165e-08, "loss": 0.5227, "step": 27140 }, { "epoch": 0.9669558401767105, "grad_norm": 1.5728423113927614, "learning_rate": 5.715829708325404e-08, "loss": 0.3737, "step": 27141 }, { "epoch": 0.9669914673032046, "grad_norm": 1.208901101015, "learning_rate": 5.703516202245851e-08, "loss": 0.3786, "step": 27142 }, { "epoch": 0.9670270944296988, "grad_norm": 1.5500141447349478, "learning_rate": 5.691215935918815e-08, "loss": 0.5067, "step": 27143 }, { "epoch": 0.9670627215561929, "grad_norm": 1.500414969050556, "learning_rate": 5.6789289095082746e-08, "loss": 0.4403, "step": 27144 }, { "epoch": 0.967098348682687, "grad_norm": 1.3402296545904921, "learning_rate": 5.666655123177767e-08, "loss": 0.3933, "step": 27145 }, { "epoch": 0.9671339758091811, "grad_norm": 1.2072290604662625, "learning_rate": 5.654394577090827e-08, "loss": 0.3507, "step": 27146 }, { "epoch": 0.9671696029356752, "grad_norm": 1.4226222102179997, "learning_rate": 5.642147271410548e-08, "loss": 0.4716, "step": 27147 }, { "epoch": 0.9672052300621693, "grad_norm": 1.4352965689295647, "learning_rate": 5.629913206300131e-08, "loss": 0.4324, "step": 27148 }, { "epoch": 0.9672408571886635, "grad_norm": 1.2292460434513246, "learning_rate": 5.617692381922446e-08, "loss": 0.4008, "step": 27149 }, { "epoch": 0.9672764843151576, "grad_norm": 1.6619507860390446, "learning_rate": 5.605484798440253e-08, "loss": 0.4882, "step": 27150 }, { "epoch": 0.9673121114416516, "grad_norm": 1.5469342118495275, "learning_rate": 5.593290456015976e-08, "loss": 0.5214, "step": 27151 }, { "epoch": 0.9673477385681458, "grad_norm": 1.758714196660663, "learning_rate": 5.5811093548121534e-08, "loss": 0.5251, "step": 27152 }, { "epoch": 0.9673833656946399, "grad_norm": 1.308142781980903, "learning_rate": 5.568941494990876e-08, "loss": 0.383, "step": 27153 }, { "epoch": 0.967418992821134, "grad_norm": 1.8367611458886235, "learning_rate": 5.556786876714238e-08, "loss": 0.5932, "step": 27154 }, { "epoch": 0.9674546199476282, "grad_norm": 1.6165127667417458, "learning_rate": 5.544645500143997e-08, "loss": 0.4822, "step": 27155 }, { "epoch": 0.9674902470741222, "grad_norm": 1.7374779047770899, "learning_rate": 5.532517365441914e-08, "loss": 0.5095, "step": 27156 }, { "epoch": 0.9675258742006163, "grad_norm": 1.2573041651501067, "learning_rate": 5.520402472769526e-08, "loss": 0.4317, "step": 27157 }, { "epoch": 0.9675615013271105, "grad_norm": 1.6141783453956722, "learning_rate": 5.5083008222880375e-08, "loss": 0.5419, "step": 27158 }, { "epoch": 0.9675971284536046, "grad_norm": 1.466355172585831, "learning_rate": 5.4962124141586526e-08, "loss": 0.4884, "step": 27159 }, { "epoch": 0.9676327555800986, "grad_norm": 1.195576314739005, "learning_rate": 5.4841372485422426e-08, "loss": 0.4115, "step": 27160 }, { "epoch": 0.9676683827065928, "grad_norm": 1.7886292247618172, "learning_rate": 5.47207532559968e-08, "loss": 0.6237, "step": 27161 }, { "epoch": 0.9677040098330869, "grad_norm": 1.6468880972509923, "learning_rate": 5.4600266454916115e-08, "loss": 0.5271, "step": 27162 }, { "epoch": 0.967739636959581, "grad_norm": 1.209452455665789, "learning_rate": 5.4479912083783556e-08, "loss": 0.3972, "step": 27163 }, { "epoch": 0.9677752640860752, "grad_norm": 1.4886888042172635, "learning_rate": 5.4359690144203394e-08, "loss": 0.4356, "step": 27164 }, { "epoch": 0.9678108912125692, "grad_norm": 1.7579716225426827, "learning_rate": 5.423960063777545e-08, "loss": 0.3971, "step": 27165 }, { "epoch": 0.9678465183390633, "grad_norm": 1.3834331314721215, "learning_rate": 5.411964356609844e-08, "loss": 0.4441, "step": 27166 }, { "epoch": 0.9678821454655575, "grad_norm": 1.1851634584058865, "learning_rate": 5.399981893077111e-08, "loss": 0.3977, "step": 27167 }, { "epoch": 0.9679177725920516, "grad_norm": 1.5697199396252177, "learning_rate": 5.388012673338661e-08, "loss": 0.4934, "step": 27168 }, { "epoch": 0.9679533997185457, "grad_norm": 1.3604053361679354, "learning_rate": 5.376056697554033e-08, "loss": 0.3805, "step": 27169 }, { "epoch": 0.9679890268450398, "grad_norm": 1.218937869438883, "learning_rate": 5.364113965882545e-08, "loss": 0.4013, "step": 27170 }, { "epoch": 0.9680246539715339, "grad_norm": 1.2763772977016203, "learning_rate": 5.352184478482958e-08, "loss": 0.4308, "step": 27171 }, { "epoch": 0.968060281098028, "grad_norm": 1.22146576387794, "learning_rate": 5.340268235514257e-08, "loss": 0.4193, "step": 27172 }, { "epoch": 0.9680959082245222, "grad_norm": 1.2020538338718256, "learning_rate": 5.3283652371352026e-08, "loss": 0.3793, "step": 27173 }, { "epoch": 0.9681315353510163, "grad_norm": 1.6996418377311375, "learning_rate": 5.316475483504113e-08, "loss": 0.5575, "step": 27174 }, { "epoch": 0.9681671624775103, "grad_norm": 1.5108358742527017, "learning_rate": 5.304598974779418e-08, "loss": 0.4221, "step": 27175 }, { "epoch": 0.9682027896040045, "grad_norm": 1.3471970068819963, "learning_rate": 5.2927357111192125e-08, "loss": 0.4132, "step": 27176 }, { "epoch": 0.9682384167304986, "grad_norm": 1.429232667881782, "learning_rate": 5.280885692681592e-08, "loss": 0.4318, "step": 27177 }, { "epoch": 0.9682740438569927, "grad_norm": 1.3470096409805006, "learning_rate": 5.2690489196242086e-08, "loss": 0.3484, "step": 27178 }, { "epoch": 0.9683096709834869, "grad_norm": 1.2183667770521731, "learning_rate": 5.257225392104603e-08, "loss": 0.3952, "step": 27179 }, { "epoch": 0.9683452981099809, "grad_norm": 1.5665953236469676, "learning_rate": 5.2454151102804276e-08, "loss": 0.4984, "step": 27180 }, { "epoch": 0.968380925236475, "grad_norm": 1.9511069859566241, "learning_rate": 5.233618074308888e-08, "loss": 0.5286, "step": 27181 }, { "epoch": 0.9684165523629692, "grad_norm": 1.3834869621644097, "learning_rate": 5.221834284346861e-08, "loss": 0.377, "step": 27182 }, { "epoch": 0.9684521794894633, "grad_norm": 1.2535036212883086, "learning_rate": 5.210063740551552e-08, "loss": 0.4117, "step": 27183 }, { "epoch": 0.9684878066159573, "grad_norm": 1.1442330593976013, "learning_rate": 5.198306443079504e-08, "loss": 0.3956, "step": 27184 }, { "epoch": 0.9685234337424515, "grad_norm": 1.2791870732615649, "learning_rate": 5.1865623920873685e-08, "loss": 0.4345, "step": 27185 }, { "epoch": 0.9685590608689456, "grad_norm": 1.7185783337689282, "learning_rate": 5.174831587731466e-08, "loss": 0.5134, "step": 27186 }, { "epoch": 0.9685946879954397, "grad_norm": 1.6163197620371135, "learning_rate": 5.163114030168004e-08, "loss": 0.4728, "step": 27187 }, { "epoch": 0.9686303151219339, "grad_norm": 2.2992015836493356, "learning_rate": 5.15140971955308e-08, "loss": 0.5019, "step": 27188 }, { "epoch": 0.968665942248428, "grad_norm": 1.1826997713165934, "learning_rate": 5.1397186560424586e-08, "loss": 0.3899, "step": 27189 }, { "epoch": 0.968701569374922, "grad_norm": 1.2432747517911207, "learning_rate": 5.1280408397919036e-08, "loss": 0.417, "step": 27190 }, { "epoch": 0.9687371965014162, "grad_norm": 1.2377755348684132, "learning_rate": 5.116376270956735e-08, "loss": 0.3946, "step": 27191 }, { "epoch": 0.9687728236279103, "grad_norm": 1.2277320844866562, "learning_rate": 5.1047249496924965e-08, "loss": 0.4066, "step": 27192 }, { "epoch": 0.9688084507544044, "grad_norm": 1.6498611193914454, "learning_rate": 5.093086876154174e-08, "loss": 0.4489, "step": 27193 }, { "epoch": 0.9688440778808985, "grad_norm": 1.3998224131125026, "learning_rate": 5.081462050496866e-08, "loss": 0.4923, "step": 27194 }, { "epoch": 0.9688797050073926, "grad_norm": 1.2323568337496116, "learning_rate": 5.0698504728752264e-08, "loss": 0.4403, "step": 27195 }, { "epoch": 0.9689153321338867, "grad_norm": 1.3925835627991674, "learning_rate": 5.05825214344402e-08, "loss": 0.4352, "step": 27196 }, { "epoch": 0.9689509592603809, "grad_norm": 1.623292283798236, "learning_rate": 5.046667062357569e-08, "loss": 0.5712, "step": 27197 }, { "epoch": 0.968986586386875, "grad_norm": 1.0838156551955838, "learning_rate": 5.035095229770193e-08, "loss": 0.3673, "step": 27198 }, { "epoch": 0.969022213513369, "grad_norm": 1.9596121510343765, "learning_rate": 5.02353664583588e-08, "loss": 0.5455, "step": 27199 }, { "epoch": 0.9690578406398632, "grad_norm": 1.4160290023229964, "learning_rate": 5.01199131070873e-08, "loss": 0.3962, "step": 27200 }, { "epoch": 0.9690934677663573, "grad_norm": 1.4399476192412017, "learning_rate": 5.0004592245422866e-08, "loss": 0.4782, "step": 27201 }, { "epoch": 0.9691290948928514, "grad_norm": 1.8223384240987006, "learning_rate": 4.9889403874903154e-08, "loss": 0.4774, "step": 27202 }, { "epoch": 0.9691647220193456, "grad_norm": 1.6309059495329659, "learning_rate": 4.977434799705916e-08, "loss": 0.5835, "step": 27203 }, { "epoch": 0.9692003491458396, "grad_norm": 1.2993289946266757, "learning_rate": 4.965942461342521e-08, "loss": 0.4102, "step": 27204 }, { "epoch": 0.9692359762723337, "grad_norm": 1.3675672728452808, "learning_rate": 4.9544633725531196e-08, "loss": 0.4022, "step": 27205 }, { "epoch": 0.9692716033988279, "grad_norm": 1.392287236842384, "learning_rate": 4.94299753349059e-08, "loss": 0.3689, "step": 27206 }, { "epoch": 0.969307230525322, "grad_norm": 1.6630390232066, "learning_rate": 4.931544944307476e-08, "loss": 0.4564, "step": 27207 }, { "epoch": 0.969342857651816, "grad_norm": 1.144354856553022, "learning_rate": 4.9201056051563226e-08, "loss": 0.3657, "step": 27208 }, { "epoch": 0.9693784847783102, "grad_norm": 1.5037910845450941, "learning_rate": 4.908679516189452e-08, "loss": 0.4354, "step": 27209 }, { "epoch": 0.9694141119048043, "grad_norm": 1.7466727316916426, "learning_rate": 4.897266677559187e-08, "loss": 0.5437, "step": 27210 }, { "epoch": 0.9694497390312984, "grad_norm": 1.3642621523670502, "learning_rate": 4.885867089417295e-08, "loss": 0.4386, "step": 27211 }, { "epoch": 0.9694853661577926, "grad_norm": 1.5407586411384122, "learning_rate": 4.8744807519155446e-08, "loss": 0.4442, "step": 27212 }, { "epoch": 0.9695209932842866, "grad_norm": 1.3819801855921665, "learning_rate": 4.863107665205702e-08, "loss": 0.4678, "step": 27213 }, { "epoch": 0.9695566204107807, "grad_norm": 2.319490969303594, "learning_rate": 4.851747829439091e-08, "loss": 0.5237, "step": 27214 }, { "epoch": 0.9695922475372749, "grad_norm": 1.5508025246225248, "learning_rate": 4.8404012447670344e-08, "loss": 0.4968, "step": 27215 }, { "epoch": 0.969627874663769, "grad_norm": 1.0867763111128472, "learning_rate": 4.829067911340635e-08, "loss": 0.4037, "step": 27216 }, { "epoch": 0.9696635017902631, "grad_norm": 1.5324648544411077, "learning_rate": 4.8177478293107707e-08, "loss": 0.5267, "step": 27217 }, { "epoch": 0.9696991289167572, "grad_norm": 1.4094249104701855, "learning_rate": 4.8064409988282104e-08, "loss": 0.437, "step": 27218 }, { "epoch": 0.9697347560432513, "grad_norm": 1.2115372209767936, "learning_rate": 4.7951474200435e-08, "loss": 0.4371, "step": 27219 }, { "epoch": 0.9697703831697454, "grad_norm": 1.3985080845418218, "learning_rate": 4.7838670931069643e-08, "loss": 0.4275, "step": 27220 }, { "epoch": 0.9698060102962396, "grad_norm": 1.3404122843720643, "learning_rate": 4.772600018168816e-08, "loss": 0.5214, "step": 27221 }, { "epoch": 0.9698416374227337, "grad_norm": 1.400223342393788, "learning_rate": 4.761346195379157e-08, "loss": 0.3946, "step": 27222 }, { "epoch": 0.9698772645492277, "grad_norm": 1.5084419124041741, "learning_rate": 4.750105624887868e-08, "loss": 0.5536, "step": 27223 }, { "epoch": 0.9699128916757219, "grad_norm": 1.2377411821155275, "learning_rate": 4.738878306844496e-08, "loss": 0.3911, "step": 27224 }, { "epoch": 0.969948518802216, "grad_norm": 2.255603088081723, "learning_rate": 4.727664241398588e-08, "loss": 0.5396, "step": 27225 }, { "epoch": 0.9699841459287101, "grad_norm": 1.17488204053295, "learning_rate": 4.7164634286994696e-08, "loss": 0.3848, "step": 27226 }, { "epoch": 0.9700197730552043, "grad_norm": 1.4210403143289545, "learning_rate": 4.705275868896464e-08, "loss": 0.4176, "step": 27227 }, { "epoch": 0.9700554001816983, "grad_norm": 1.3185300242242395, "learning_rate": 4.6941015621382316e-08, "loss": 0.4051, "step": 27228 }, { "epoch": 0.9700910273081924, "grad_norm": 1.3329464602871204, "learning_rate": 4.6829405085736525e-08, "loss": 0.4073, "step": 27229 }, { "epoch": 0.9701266544346866, "grad_norm": 1.586601142354915, "learning_rate": 4.671792708351497e-08, "loss": 0.4976, "step": 27230 }, { "epoch": 0.9701622815611807, "grad_norm": 1.2793526346406445, "learning_rate": 4.660658161620202e-08, "loss": 0.4283, "step": 27231 }, { "epoch": 0.9701979086876747, "grad_norm": 1.724319281290923, "learning_rate": 4.649536868527871e-08, "loss": 0.3972, "step": 27232 }, { "epoch": 0.9702335358141689, "grad_norm": 1.3815375604580489, "learning_rate": 4.6384288292226075e-08, "loss": 0.4852, "step": 27233 }, { "epoch": 0.970269162940663, "grad_norm": 1.863110358333816, "learning_rate": 4.627334043852405e-08, "loss": 0.4818, "step": 27234 }, { "epoch": 0.9703047900671571, "grad_norm": 1.516071628878074, "learning_rate": 4.6162525125650335e-08, "loss": 0.419, "step": 27235 }, { "epoch": 0.9703404171936513, "grad_norm": 1.2362136848598448, "learning_rate": 4.605184235507931e-08, "loss": 0.4333, "step": 27236 }, { "epoch": 0.9703760443201453, "grad_norm": 1.307437370611741, "learning_rate": 4.594129212828535e-08, "loss": 0.4207, "step": 27237 }, { "epoch": 0.9704116714466394, "grad_norm": 1.4412445540284498, "learning_rate": 4.5830874446740616e-08, "loss": 0.4494, "step": 27238 }, { "epoch": 0.9704472985731336, "grad_norm": 1.5093990399142314, "learning_rate": 4.572058931191614e-08, "loss": 0.4563, "step": 27239 }, { "epoch": 0.9704829256996277, "grad_norm": 1.4111017930147383, "learning_rate": 4.561043672527854e-08, "loss": 0.5565, "step": 27240 }, { "epoch": 0.9705185528261218, "grad_norm": 1.3194916307083706, "learning_rate": 4.5500416688296634e-08, "loss": 0.3926, "step": 27241 }, { "epoch": 0.970554179952616, "grad_norm": 2.0515070451480466, "learning_rate": 4.53905292024337e-08, "loss": 0.4933, "step": 27242 }, { "epoch": 0.97058980707911, "grad_norm": 1.3053327289949768, "learning_rate": 4.528077426915412e-08, "loss": 0.3992, "step": 27243 }, { "epoch": 0.9706254342056041, "grad_norm": 1.50050833705355, "learning_rate": 4.517115188991894e-08, "loss": 0.4835, "step": 27244 }, { "epoch": 0.9706610613320983, "grad_norm": 1.6145314422192014, "learning_rate": 4.5061662066187004e-08, "loss": 0.427, "step": 27245 }, { "epoch": 0.9706966884585924, "grad_norm": 1.4570886219323622, "learning_rate": 4.4952304799418256e-08, "loss": 0.4634, "step": 27246 }, { "epoch": 0.9707323155850864, "grad_norm": 1.2083259223351694, "learning_rate": 4.484308009106819e-08, "loss": 0.4379, "step": 27247 }, { "epoch": 0.9707679427115806, "grad_norm": 1.6014793313537885, "learning_rate": 4.47339879425901e-08, "loss": 0.4729, "step": 27248 }, { "epoch": 0.9708035698380747, "grad_norm": 1.3676809711134597, "learning_rate": 4.462502835543614e-08, "loss": 0.4301, "step": 27249 }, { "epoch": 0.9708391969645688, "grad_norm": 1.4732784089812578, "learning_rate": 4.4516201331059606e-08, "loss": 0.4569, "step": 27250 }, { "epoch": 0.970874824091063, "grad_norm": 1.2489134625469678, "learning_rate": 4.440750687090822e-08, "loss": 0.3829, "step": 27251 }, { "epoch": 0.970910451217557, "grad_norm": 1.544723224888114, "learning_rate": 4.429894497642861e-08, "loss": 0.5946, "step": 27252 }, { "epoch": 0.9709460783440511, "grad_norm": 1.6963936074218893, "learning_rate": 4.419051564906851e-08, "loss": 0.5993, "step": 27253 }, { "epoch": 0.9709817054705453, "grad_norm": 1.6056924801805876, "learning_rate": 4.408221889026898e-08, "loss": 0.4609, "step": 27254 }, { "epoch": 0.9710173325970394, "grad_norm": 1.4218239828573782, "learning_rate": 4.397405470147442e-08, "loss": 0.4905, "step": 27255 }, { "epoch": 0.9710529597235334, "grad_norm": 1.721403767378411, "learning_rate": 4.386602308412369e-08, "loss": 0.4964, "step": 27256 }, { "epoch": 0.9710885868500276, "grad_norm": 1.393769183998509, "learning_rate": 4.375812403965673e-08, "loss": 0.4254, "step": 27257 }, { "epoch": 0.9711242139765217, "grad_norm": 1.504714722352559, "learning_rate": 4.365035756950797e-08, "loss": 0.5337, "step": 27258 }, { "epoch": 0.9711598411030158, "grad_norm": 1.2210463876094848, "learning_rate": 4.354272367511514e-08, "loss": 0.3774, "step": 27259 }, { "epoch": 0.97119546822951, "grad_norm": 1.247944984781959, "learning_rate": 4.34352223579082e-08, "loss": 0.3909, "step": 27260 }, { "epoch": 0.971231095356004, "grad_norm": 1.5904883422824356, "learning_rate": 4.332785361932157e-08, "loss": 0.5476, "step": 27261 }, { "epoch": 0.9712667224824981, "grad_norm": 1.7703096045663582, "learning_rate": 4.322061746078521e-08, "loss": 0.5417, "step": 27262 }, { "epoch": 0.9713023496089923, "grad_norm": 1.2015907737371707, "learning_rate": 4.3113513883723534e-08, "loss": 0.3926, "step": 27263 }, { "epoch": 0.9713379767354864, "grad_norm": 1.696994570832596, "learning_rate": 4.300654288956652e-08, "loss": 0.5049, "step": 27264 }, { "epoch": 0.9713736038619805, "grad_norm": 1.2908906843267545, "learning_rate": 4.2899704479736345e-08, "loss": 0.4115, "step": 27265 }, { "epoch": 0.9714092309884746, "grad_norm": 1.2282807152290844, "learning_rate": 4.279299865565523e-08, "loss": 0.3656, "step": 27266 }, { "epoch": 0.9714448581149687, "grad_norm": 1.2482421216793456, "learning_rate": 4.2686425418745345e-08, "loss": 0.4069, "step": 27267 }, { "epoch": 0.9714804852414628, "grad_norm": 1.259998091042391, "learning_rate": 4.257998477042669e-08, "loss": 0.3924, "step": 27268 }, { "epoch": 0.971516112367957, "grad_norm": 1.216629304251198, "learning_rate": 4.2473676712114776e-08, "loss": 0.4231, "step": 27269 }, { "epoch": 0.9715517394944511, "grad_norm": 1.3792535459667141, "learning_rate": 4.236750124522515e-08, "loss": 0.4013, "step": 27270 }, { "epoch": 0.9715873666209451, "grad_norm": 1.395311770520043, "learning_rate": 4.226145837117335e-08, "loss": 0.3951, "step": 27271 }, { "epoch": 0.9716229937474393, "grad_norm": 1.3529808423319498, "learning_rate": 4.215554809136935e-08, "loss": 0.4439, "step": 27272 }, { "epoch": 0.9716586208739334, "grad_norm": 1.2464684923024099, "learning_rate": 4.204977040722425e-08, "loss": 0.3809, "step": 27273 }, { "epoch": 0.9716942480004275, "grad_norm": 1.2119121167012799, "learning_rate": 4.1944125320146914e-08, "loss": 0.4069, "step": 27274 }, { "epoch": 0.9717298751269217, "grad_norm": 1.7463543371358055, "learning_rate": 4.1838612831544e-08, "loss": 0.487, "step": 27275 }, { "epoch": 0.9717655022534157, "grad_norm": 1.457373758355003, "learning_rate": 4.173323294281995e-08, "loss": 0.4565, "step": 27276 }, { "epoch": 0.9718011293799098, "grad_norm": 1.3115120641831726, "learning_rate": 4.1627985655378064e-08, "loss": 0.5147, "step": 27277 }, { "epoch": 0.971836756506404, "grad_norm": 1.454830109667797, "learning_rate": 4.152287097062058e-08, "loss": 0.5115, "step": 27278 }, { "epoch": 0.9718723836328981, "grad_norm": 1.7436946719285547, "learning_rate": 4.1417888889947466e-08, "loss": 0.5337, "step": 27279 }, { "epoch": 0.9719080107593921, "grad_norm": 1.277368005552174, "learning_rate": 4.1313039414754285e-08, "loss": 0.3677, "step": 27280 }, { "epoch": 0.9719436378858863, "grad_norm": 1.483432048982574, "learning_rate": 4.1208322546438806e-08, "loss": 0.4788, "step": 27281 }, { "epoch": 0.9719792650123804, "grad_norm": 1.4820410106262532, "learning_rate": 4.110373828639547e-08, "loss": 0.3781, "step": 27282 }, { "epoch": 0.9720148921388745, "grad_norm": 1.155421800417669, "learning_rate": 4.09992866360176e-08, "loss": 0.3426, "step": 27283 }, { "epoch": 0.9720505192653687, "grad_norm": 1.328002457200314, "learning_rate": 4.089496759669409e-08, "loss": 0.3945, "step": 27284 }, { "epoch": 0.9720861463918627, "grad_norm": 1.3644801206037367, "learning_rate": 4.0790781169816055e-08, "loss": 0.429, "step": 27285 }, { "epoch": 0.9721217735183568, "grad_norm": 1.274665537769071, "learning_rate": 4.068672735676904e-08, "loss": 0.4261, "step": 27286 }, { "epoch": 0.972157400644851, "grad_norm": 1.224683802109629, "learning_rate": 4.058280615893972e-08, "loss": 0.444, "step": 27287 }, { "epoch": 0.9721930277713451, "grad_norm": 1.2609405296389846, "learning_rate": 4.0479017577711447e-08, "loss": 0.3736, "step": 27288 }, { "epoch": 0.9722286548978392, "grad_norm": 1.2846783865467286, "learning_rate": 4.037536161446642e-08, "loss": 0.374, "step": 27289 }, { "epoch": 0.9722642820243333, "grad_norm": 2.0266148862211337, "learning_rate": 4.0271838270584675e-08, "loss": 0.5213, "step": 27290 }, { "epoch": 0.9722999091508274, "grad_norm": 1.1794059928678209, "learning_rate": 4.016844754744509e-08, "loss": 0.3398, "step": 27291 }, { "epoch": 0.9723355362773215, "grad_norm": 1.4392682724877017, "learning_rate": 4.0065189446424346e-08, "loss": 0.5222, "step": 27292 }, { "epoch": 0.9723711634038157, "grad_norm": 1.7708596651930368, "learning_rate": 3.99620639688969e-08, "loss": 0.5642, "step": 27293 }, { "epoch": 0.9724067905303098, "grad_norm": 1.6550436163265607, "learning_rate": 3.98590711162361e-08, "loss": 0.5842, "step": 27294 }, { "epoch": 0.9724424176568038, "grad_norm": 1.569199143134176, "learning_rate": 3.975621088981308e-08, "loss": 0.5337, "step": 27295 }, { "epoch": 0.972478044783298, "grad_norm": 1.3897090956899025, "learning_rate": 3.965348329099894e-08, "loss": 0.4624, "step": 27296 }, { "epoch": 0.9725136719097921, "grad_norm": 1.2647995639406453, "learning_rate": 3.955088832115928e-08, "loss": 0.3867, "step": 27297 }, { "epoch": 0.9725492990362862, "grad_norm": 1.37260589099566, "learning_rate": 3.9448425981661876e-08, "loss": 0.4072, "step": 27298 }, { "epoch": 0.9725849261627804, "grad_norm": 2.4936088480779195, "learning_rate": 3.934609627387121e-08, "loss": 0.5651, "step": 27299 }, { "epoch": 0.9726205532892744, "grad_norm": 1.2576271692167538, "learning_rate": 3.9243899199149504e-08, "loss": 0.3696, "step": 27300 }, { "epoch": 0.9726561804157685, "grad_norm": 1.4168390983788617, "learning_rate": 3.91418347588568e-08, "loss": 0.4249, "step": 27301 }, { "epoch": 0.9726918075422627, "grad_norm": 1.668126628147437, "learning_rate": 3.903990295435311e-08, "loss": 0.5196, "step": 27302 }, { "epoch": 0.9727274346687568, "grad_norm": 1.3294637437894072, "learning_rate": 3.8938103786995144e-08, "loss": 0.3868, "step": 27303 }, { "epoch": 0.9727630617952508, "grad_norm": 1.3900925089503686, "learning_rate": 3.883643725813846e-08, "loss": 0.3936, "step": 27304 }, { "epoch": 0.972798688921745, "grad_norm": 1.208237224284269, "learning_rate": 3.8734903369138654e-08, "loss": 0.4152, "step": 27305 }, { "epoch": 0.9728343160482391, "grad_norm": 1.1673408028737682, "learning_rate": 3.8633502121344643e-08, "loss": 0.3687, "step": 27306 }, { "epoch": 0.9728699431747332, "grad_norm": 2.068109971433898, "learning_rate": 3.853223351610757e-08, "loss": 0.4837, "step": 27307 }, { "epoch": 0.9729055703012274, "grad_norm": 4.188289222541249, "learning_rate": 3.8431097554778585e-08, "loss": 0.5117, "step": 27308 }, { "epoch": 0.9729411974277214, "grad_norm": 1.400709009660452, "learning_rate": 3.8330094238699936e-08, "loss": 0.4315, "step": 27309 }, { "epoch": 0.9729768245542155, "grad_norm": 1.6880934451859904, "learning_rate": 3.822922356922054e-08, "loss": 0.5081, "step": 27310 }, { "epoch": 0.9730124516807097, "grad_norm": 1.9195314543437545, "learning_rate": 3.812848554768045e-08, "loss": 0.4971, "step": 27311 }, { "epoch": 0.9730480788072038, "grad_norm": 1.6557148471809255, "learning_rate": 3.802788017542414e-08, "loss": 0.395, "step": 27312 }, { "epoch": 0.9730837059336979, "grad_norm": 1.3256405385602597, "learning_rate": 3.792740745378831e-08, "loss": 0.4347, "step": 27313 }, { "epoch": 0.973119333060192, "grad_norm": 1.4737017493675808, "learning_rate": 3.7827067384113014e-08, "loss": 0.4703, "step": 27314 }, { "epoch": 0.9731549601866861, "grad_norm": 1.1470559907968927, "learning_rate": 3.772685996773273e-08, "loss": 0.4025, "step": 27315 }, { "epoch": 0.9731905873131803, "grad_norm": 1.145070187951178, "learning_rate": 3.7626785205983065e-08, "loss": 0.3794, "step": 27316 }, { "epoch": 0.9732262144396744, "grad_norm": 1.3209409931183271, "learning_rate": 3.752684310019628e-08, "loss": 0.3681, "step": 27317 }, { "epoch": 0.9732618415661685, "grad_norm": 1.5922491826600154, "learning_rate": 3.7427033651702414e-08, "loss": 0.4685, "step": 27318 }, { "epoch": 0.9732974686926626, "grad_norm": 1.527287429620405, "learning_rate": 3.732735686183042e-08, "loss": 0.5002, "step": 27319 }, { "epoch": 0.9733330958191567, "grad_norm": 1.2512715625949944, "learning_rate": 3.722781273190923e-08, "loss": 0.3569, "step": 27320 }, { "epoch": 0.9733687229456508, "grad_norm": 1.5618732403053952, "learning_rate": 3.7128401263262226e-08, "loss": 0.5203, "step": 27321 }, { "epoch": 0.973404350072145, "grad_norm": 1.36202980456674, "learning_rate": 3.702912245721502e-08, "loss": 0.4157, "step": 27322 }, { "epoch": 0.9734399771986391, "grad_norm": 1.2637362684486781, "learning_rate": 3.6929976315088765e-08, "loss": 0.4179, "step": 27323 }, { "epoch": 0.9734756043251331, "grad_norm": 1.7900428449980044, "learning_rate": 3.683096283820353e-08, "loss": 0.5705, "step": 27324 }, { "epoch": 0.9735112314516273, "grad_norm": 1.1805806441162712, "learning_rate": 3.6732082027877145e-08, "loss": 0.4128, "step": 27325 }, { "epoch": 0.9735468585781214, "grad_norm": 1.650605219417937, "learning_rate": 3.663333388542745e-08, "loss": 0.5045, "step": 27326 }, { "epoch": 0.9735824857046155, "grad_norm": 1.740937468888553, "learning_rate": 3.653471841217004e-08, "loss": 0.5234, "step": 27327 }, { "epoch": 0.9736181128311097, "grad_norm": 1.372647735594715, "learning_rate": 3.643623560941611e-08, "loss": 0.4041, "step": 27328 }, { "epoch": 0.9736537399576037, "grad_norm": 1.4480125221130349, "learning_rate": 3.633788547847683e-08, "loss": 0.4911, "step": 27329 }, { "epoch": 0.9736893670840978, "grad_norm": 1.4320935945115634, "learning_rate": 3.623966802066447e-08, "loss": 0.4948, "step": 27330 }, { "epoch": 0.973724994210592, "grad_norm": 1.392628753719109, "learning_rate": 3.6141583237283564e-08, "loss": 0.5035, "step": 27331 }, { "epoch": 0.9737606213370861, "grad_norm": 1.720104755112299, "learning_rate": 3.6043631129643044e-08, "loss": 0.4377, "step": 27332 }, { "epoch": 0.9737962484635801, "grad_norm": 1.7924150317851666, "learning_rate": 3.5945811699045206e-08, "loss": 0.5212, "step": 27333 }, { "epoch": 0.9738318755900743, "grad_norm": 1.3574144749472072, "learning_rate": 3.584812494679346e-08, "loss": 0.4209, "step": 27334 }, { "epoch": 0.9738675027165684, "grad_norm": 1.4177885764815246, "learning_rate": 3.5750570874188985e-08, "loss": 0.3881, "step": 27335 }, { "epoch": 0.9739031298430625, "grad_norm": 1.2840680638613025, "learning_rate": 3.565314948252962e-08, "loss": 0.4117, "step": 27336 }, { "epoch": 0.9739387569695567, "grad_norm": 1.6532656926992086, "learning_rate": 3.5555860773113235e-08, "loss": 0.5073, "step": 27337 }, { "epoch": 0.9739743840960507, "grad_norm": 1.2172737989769256, "learning_rate": 3.545870474723545e-08, "loss": 0.4108, "step": 27338 }, { "epoch": 0.9740100112225448, "grad_norm": 1.3156174273613988, "learning_rate": 3.536168140618968e-08, "loss": 0.3641, "step": 27339 }, { "epoch": 0.974045638349039, "grad_norm": 1.2108806865710107, "learning_rate": 3.526479075126821e-08, "loss": 0.3659, "step": 27340 }, { "epoch": 0.9740812654755331, "grad_norm": 1.7540374300820398, "learning_rate": 3.516803278376113e-08, "loss": 0.4698, "step": 27341 }, { "epoch": 0.9741168926020272, "grad_norm": 2.1634825404565645, "learning_rate": 3.50714075049563e-08, "loss": 0.5576, "step": 27342 }, { "epoch": 0.9741525197285213, "grad_norm": 1.206542189150535, "learning_rate": 3.497491491614158e-08, "loss": 0.4257, "step": 27343 }, { "epoch": 0.9741881468550154, "grad_norm": 1.6493180823060796, "learning_rate": 3.487855501860038e-08, "loss": 0.4999, "step": 27344 }, { "epoch": 0.9742237739815095, "grad_norm": 1.3644253566467623, "learning_rate": 3.478232781361612e-08, "loss": 0.4602, "step": 27345 }, { "epoch": 0.9742594011080037, "grad_norm": 1.4542023426797466, "learning_rate": 3.468623330247112e-08, "loss": 0.3799, "step": 27346 }, { "epoch": 0.9742950282344978, "grad_norm": 1.2773036801892428, "learning_rate": 3.4590271486444336e-08, "loss": 0.3638, "step": 27347 }, { "epoch": 0.9743306553609918, "grad_norm": 1.7212033874679875, "learning_rate": 3.449444236681254e-08, "loss": 0.5442, "step": 27348 }, { "epoch": 0.974366282487486, "grad_norm": 1.6887398130820965, "learning_rate": 3.4398745944853597e-08, "loss": 0.4303, "step": 27349 }, { "epoch": 0.9744019096139801, "grad_norm": 1.2280319857201034, "learning_rate": 3.430318222183981e-08, "loss": 0.5239, "step": 27350 }, { "epoch": 0.9744375367404742, "grad_norm": 1.4106433626071126, "learning_rate": 3.4207751199045736e-08, "loss": 0.4337, "step": 27351 }, { "epoch": 0.9744731638669684, "grad_norm": 1.847666638673412, "learning_rate": 3.411245287774034e-08, "loss": 0.5111, "step": 27352 }, { "epoch": 0.9745087909934624, "grad_norm": 1.867957697847662, "learning_rate": 3.401728725919373e-08, "loss": 0.4996, "step": 27353 }, { "epoch": 0.9745444181199565, "grad_norm": 1.6311970007912395, "learning_rate": 3.392225434467378e-08, "loss": 0.5377, "step": 27354 }, { "epoch": 0.9745800452464507, "grad_norm": 1.3131671255797666, "learning_rate": 3.382735413544391e-08, "loss": 0.4038, "step": 27355 }, { "epoch": 0.9746156723729448, "grad_norm": 1.2719293674631247, "learning_rate": 3.373258663276868e-08, "loss": 0.3956, "step": 27356 }, { "epoch": 0.9746512994994389, "grad_norm": 1.6808869065587457, "learning_rate": 3.363795183791041e-08, "loss": 0.4992, "step": 27357 }, { "epoch": 0.974686926625933, "grad_norm": 2.1297889685836093, "learning_rate": 3.35434497521292e-08, "loss": 0.5387, "step": 27358 }, { "epoch": 0.9747225537524271, "grad_norm": 1.7057976696408998, "learning_rate": 3.344908037668182e-08, "loss": 0.504, "step": 27359 }, { "epoch": 0.9747581808789212, "grad_norm": 1.1613067740061263, "learning_rate": 3.335484371282727e-08, "loss": 0.396, "step": 27360 }, { "epoch": 0.9747938080054154, "grad_norm": 1.3324850749952672, "learning_rate": 3.3260739761818986e-08, "loss": 0.3878, "step": 27361 }, { "epoch": 0.9748294351319094, "grad_norm": 1.2776639458382768, "learning_rate": 3.316676852491041e-08, "loss": 0.366, "step": 27362 }, { "epoch": 0.9748650622584035, "grad_norm": 1.5544280778178998, "learning_rate": 3.3072930003351655e-08, "loss": 0.5464, "step": 27363 }, { "epoch": 0.9749006893848977, "grad_norm": 1.3351280374059533, "learning_rate": 3.2979224198395054e-08, "loss": 0.4072, "step": 27364 }, { "epoch": 0.9749363165113918, "grad_norm": 1.2806407072639732, "learning_rate": 3.2885651111285166e-08, "loss": 0.4175, "step": 27365 }, { "epoch": 0.9749719436378859, "grad_norm": 1.312152089380949, "learning_rate": 3.2792210743269883e-08, "loss": 0.3749, "step": 27366 }, { "epoch": 0.97500757076438, "grad_norm": 1.3177605614662113, "learning_rate": 3.269890309559376e-08, "loss": 0.3966, "step": 27367 }, { "epoch": 0.9750431978908741, "grad_norm": 1.1486773150822156, "learning_rate": 3.260572816949692e-08, "loss": 0.3628, "step": 27368 }, { "epoch": 0.9750788250173682, "grad_norm": 1.7108712263023185, "learning_rate": 3.2512685966222815e-08, "loss": 0.5287, "step": 27369 }, { "epoch": 0.9751144521438624, "grad_norm": 2.047372324183129, "learning_rate": 3.241977648700934e-08, "loss": 0.4903, "step": 27370 }, { "epoch": 0.9751500792703565, "grad_norm": 1.225752530777049, "learning_rate": 3.2326999733092165e-08, "loss": 0.3878, "step": 27371 }, { "epoch": 0.9751857063968505, "grad_norm": 1.1767443334039063, "learning_rate": 3.2234355705709206e-08, "loss": 0.4016, "step": 27372 }, { "epoch": 0.9752213335233447, "grad_norm": 1.2448298253626715, "learning_rate": 3.2141844406091696e-08, "loss": 0.4013, "step": 27373 }, { "epoch": 0.9752569606498388, "grad_norm": 1.3163535990469128, "learning_rate": 3.2049465835474194e-08, "loss": 0.4238, "step": 27374 }, { "epoch": 0.9752925877763329, "grad_norm": 1.316648754132574, "learning_rate": 3.195721999508461e-08, "loss": 0.4039, "step": 27375 }, { "epoch": 0.9753282149028271, "grad_norm": 1.7171307632665205, "learning_rate": 3.1865106886150856e-08, "loss": 0.4538, "step": 27376 }, { "epoch": 0.9753638420293211, "grad_norm": 1.1675877002405781, "learning_rate": 3.1773126509900834e-08, "loss": 0.4259, "step": 27377 }, { "epoch": 0.9753994691558152, "grad_norm": 1.5623035870493986, "learning_rate": 3.168127886756023e-08, "loss": 0.5036, "step": 27378 }, { "epoch": 0.9754350962823094, "grad_norm": 1.743506716299073, "learning_rate": 3.158956396034918e-08, "loss": 0.4923, "step": 27379 }, { "epoch": 0.9754707234088035, "grad_norm": 1.3176573509942162, "learning_rate": 3.149798178949226e-08, "loss": 0.3843, "step": 27380 }, { "epoch": 0.9755063505352976, "grad_norm": 1.8189758336103905, "learning_rate": 3.1406532356206274e-08, "loss": 0.528, "step": 27381 }, { "epoch": 0.9755419776617917, "grad_norm": 1.6519259153667927, "learning_rate": 3.131521566171025e-08, "loss": 0.4741, "step": 27382 }, { "epoch": 0.9755776047882858, "grad_norm": 1.9490198853074143, "learning_rate": 3.1224031707219885e-08, "loss": 0.5647, "step": 27383 }, { "epoch": 0.9756132319147799, "grad_norm": 1.524396894523851, "learning_rate": 3.113298049394975e-08, "loss": 0.5196, "step": 27384 }, { "epoch": 0.9756488590412741, "grad_norm": 1.5669067727729253, "learning_rate": 3.104206202311111e-08, "loss": 0.392, "step": 27385 }, { "epoch": 0.9756844861677682, "grad_norm": 1.6559084005829108, "learning_rate": 3.0951276295915214e-08, "loss": 0.4876, "step": 27386 }, { "epoch": 0.9757201132942622, "grad_norm": 1.5903011076555447, "learning_rate": 3.086062331357109e-08, "loss": 0.5018, "step": 27387 }, { "epoch": 0.9757557404207564, "grad_norm": 1.2804828214528992, "learning_rate": 3.077010307728556e-08, "loss": 0.3838, "step": 27388 }, { "epoch": 0.9757913675472505, "grad_norm": 1.246593496543192, "learning_rate": 3.067971558826433e-08, "loss": 0.3949, "step": 27389 }, { "epoch": 0.9758269946737446, "grad_norm": 1.28994688444033, "learning_rate": 3.058946084771086e-08, "loss": 0.4069, "step": 27390 }, { "epoch": 0.9758626218002387, "grad_norm": 1.2023264607141444, "learning_rate": 3.049933885682754e-08, "loss": 0.3611, "step": 27391 }, { "epoch": 0.9758982489267328, "grad_norm": 1.2770362099284585, "learning_rate": 3.040934961681341e-08, "loss": 0.3816, "step": 27392 }, { "epoch": 0.9759338760532269, "grad_norm": 1.6007515604043572, "learning_rate": 3.03194931288664e-08, "loss": 0.5754, "step": 27393 }, { "epoch": 0.9759695031797211, "grad_norm": 1.4629274846715659, "learning_rate": 3.0229769394184427e-08, "loss": 0.4908, "step": 27394 }, { "epoch": 0.9760051303062152, "grad_norm": 1.5437662507260959, "learning_rate": 3.014017841396211e-08, "loss": 0.4655, "step": 27395 }, { "epoch": 0.9760407574327092, "grad_norm": 1.634826916640628, "learning_rate": 3.00507201893907e-08, "loss": 0.4122, "step": 27396 }, { "epoch": 0.9760763845592034, "grad_norm": 1.422156016184796, "learning_rate": 2.9961394721663706e-08, "loss": 0.5118, "step": 27397 }, { "epoch": 0.9761120116856975, "grad_norm": 1.4915683622493319, "learning_rate": 2.987220201196794e-08, "loss": 0.4105, "step": 27398 }, { "epoch": 0.9761476388121916, "grad_norm": 1.3464960875948615, "learning_rate": 2.9783142061493575e-08, "loss": 0.5098, "step": 27399 }, { "epoch": 0.9761832659386858, "grad_norm": 2.0463106282108723, "learning_rate": 2.969421487142521e-08, "loss": 0.4605, "step": 27400 }, { "epoch": 0.9762188930651798, "grad_norm": 1.8214130553326966, "learning_rate": 2.9605420442946342e-08, "loss": 0.487, "step": 27401 }, { "epoch": 0.9762545201916739, "grad_norm": 1.2167028354252554, "learning_rate": 2.9516758777241583e-08, "loss": 0.387, "step": 27402 }, { "epoch": 0.9762901473181681, "grad_norm": 1.7572680533457397, "learning_rate": 2.9428229875488878e-08, "loss": 0.4207, "step": 27403 }, { "epoch": 0.9763257744446622, "grad_norm": 1.975570148270247, "learning_rate": 2.9339833738869506e-08, "loss": 0.5414, "step": 27404 }, { "epoch": 0.9763614015711563, "grad_norm": 1.3493982302675054, "learning_rate": 2.925157036855919e-08, "loss": 0.4347, "step": 27405 }, { "epoch": 0.9763970286976504, "grad_norm": 1.3314873182170586, "learning_rate": 2.916343976573255e-08, "loss": 0.3699, "step": 27406 }, { "epoch": 0.9764326558241445, "grad_norm": 1.844950463496298, "learning_rate": 2.9075441931564197e-08, "loss": 0.5072, "step": 27407 }, { "epoch": 0.9764682829506386, "grad_norm": 2.420936068936217, "learning_rate": 2.898757686722542e-08, "loss": 0.4215, "step": 27408 }, { "epoch": 0.9765039100771328, "grad_norm": 1.2068517264546412, "learning_rate": 2.8899844573886392e-08, "loss": 0.3929, "step": 27409 }, { "epoch": 0.9765395372036269, "grad_norm": 1.2312265223628533, "learning_rate": 2.881224505271618e-08, "loss": 0.423, "step": 27410 }, { "epoch": 0.9765751643301209, "grad_norm": 1.446390870262022, "learning_rate": 2.8724778304879407e-08, "loss": 0.4866, "step": 27411 }, { "epoch": 0.9766107914566151, "grad_norm": 1.3879924107954802, "learning_rate": 2.8637444331541808e-08, "loss": 0.3712, "step": 27412 }, { "epoch": 0.9766464185831092, "grad_norm": 2.005153213096876, "learning_rate": 2.8550243133866896e-08, "loss": 0.5577, "step": 27413 }, { "epoch": 0.9766820457096033, "grad_norm": 1.2847743490491783, "learning_rate": 2.8463174713014852e-08, "loss": 0.407, "step": 27414 }, { "epoch": 0.9767176728360975, "grad_norm": 1.8396250192152035, "learning_rate": 2.837623907014475e-08, "loss": 0.5775, "step": 27415 }, { "epoch": 0.9767532999625915, "grad_norm": 1.4990173961946227, "learning_rate": 2.8289436206415665e-08, "loss": 0.4547, "step": 27416 }, { "epoch": 0.9767889270890856, "grad_norm": 1.3159093822237204, "learning_rate": 2.820276612298223e-08, "loss": 0.4437, "step": 27417 }, { "epoch": 0.9768245542155798, "grad_norm": 1.1666142925710756, "learning_rate": 2.811622882099796e-08, "loss": 0.376, "step": 27418 }, { "epoch": 0.9768601813420739, "grad_norm": 2.1779991504061837, "learning_rate": 2.802982430161749e-08, "loss": 0.517, "step": 27419 }, { "epoch": 0.9768958084685679, "grad_norm": 1.2309628789353833, "learning_rate": 2.7943552565988796e-08, "loss": 0.4091, "step": 27420 }, { "epoch": 0.9769314355950621, "grad_norm": 1.6775316729148653, "learning_rate": 2.7857413615262063e-08, "loss": 0.4901, "step": 27421 }, { "epoch": 0.9769670627215562, "grad_norm": 1.329337526812184, "learning_rate": 2.7771407450584154e-08, "loss": 0.405, "step": 27422 }, { "epoch": 0.9770026898480503, "grad_norm": 1.7825705310704227, "learning_rate": 2.7685534073099705e-08, "loss": 0.4835, "step": 27423 }, { "epoch": 0.9770383169745445, "grad_norm": 1.470249840168145, "learning_rate": 2.7599793483952254e-08, "loss": 0.4649, "step": 27424 }, { "epoch": 0.9770739441010385, "grad_norm": 1.4603791000703523, "learning_rate": 2.7514185684284212e-08, "loss": 0.4044, "step": 27425 }, { "epoch": 0.9771095712275326, "grad_norm": 1.5433301109591804, "learning_rate": 2.742871067523578e-08, "loss": 0.5302, "step": 27426 }, { "epoch": 0.9771451983540268, "grad_norm": 1.3004200182741887, "learning_rate": 2.734336845794383e-08, "loss": 0.4145, "step": 27427 }, { "epoch": 0.9771808254805209, "grad_norm": 1.6518769294000961, "learning_rate": 2.725815903354634e-08, "loss": 0.4851, "step": 27428 }, { "epoch": 0.977216452607015, "grad_norm": 1.2472902909624017, "learning_rate": 2.7173082403175733e-08, "loss": 0.3374, "step": 27429 }, { "epoch": 0.9772520797335091, "grad_norm": 1.3357308872742237, "learning_rate": 2.708813856796666e-08, "loss": 0.4057, "step": 27430 }, { "epoch": 0.9772877068600032, "grad_norm": 1.3081904773645148, "learning_rate": 2.7003327529050438e-08, "loss": 0.3875, "step": 27431 }, { "epoch": 0.9773233339864973, "grad_norm": 1.3282056647796168, "learning_rate": 2.6918649287555055e-08, "loss": 0.4225, "step": 27432 }, { "epoch": 0.9773589611129915, "grad_norm": 1.6785622175832509, "learning_rate": 2.6834103844607383e-08, "loss": 0.5007, "step": 27433 }, { "epoch": 0.9773945882394856, "grad_norm": 1.3953440989785542, "learning_rate": 2.6749691201335416e-08, "loss": 0.4499, "step": 27434 }, { "epoch": 0.9774302153659796, "grad_norm": 1.2422380550826713, "learning_rate": 2.6665411358862693e-08, "loss": 0.4121, "step": 27435 }, { "epoch": 0.9774658424924738, "grad_norm": 2.3839359466447143, "learning_rate": 2.658126431831054e-08, "loss": 0.5653, "step": 27436 }, { "epoch": 0.9775014696189679, "grad_norm": 1.3713928967563018, "learning_rate": 2.6497250080800285e-08, "loss": 0.4092, "step": 27437 }, { "epoch": 0.977537096745462, "grad_norm": 1.5625110298443086, "learning_rate": 2.641336864744992e-08, "loss": 0.4607, "step": 27438 }, { "epoch": 0.9775727238719562, "grad_norm": 1.9215564548735575, "learning_rate": 2.6329620019376335e-08, "loss": 0.4906, "step": 27439 }, { "epoch": 0.9776083509984502, "grad_norm": 1.583086400108033, "learning_rate": 2.6246004197695296e-08, "loss": 0.5334, "step": 27440 }, { "epoch": 0.9776439781249443, "grad_norm": 1.7781038152831141, "learning_rate": 2.6162521183520362e-08, "loss": 0.5676, "step": 27441 }, { "epoch": 0.9776796052514385, "grad_norm": 1.4535814225341996, "learning_rate": 2.6079170977961754e-08, "loss": 0.3745, "step": 27442 }, { "epoch": 0.9777152323779326, "grad_norm": 1.5233411674995498, "learning_rate": 2.5995953582130806e-08, "loss": 0.5583, "step": 27443 }, { "epoch": 0.9777508595044266, "grad_norm": 1.7839872698196282, "learning_rate": 2.591286899713552e-08, "loss": 0.544, "step": 27444 }, { "epoch": 0.9777864866309208, "grad_norm": 1.345712762621164, "learning_rate": 2.582991722408168e-08, "loss": 0.4114, "step": 27445 }, { "epoch": 0.9778221137574149, "grad_norm": 1.3862134381545557, "learning_rate": 2.5747098264073957e-08, "loss": 0.3708, "step": 27446 }, { "epoch": 0.977857740883909, "grad_norm": 1.6389591668933692, "learning_rate": 2.56644121182148e-08, "loss": 0.5151, "step": 27447 }, { "epoch": 0.9778933680104032, "grad_norm": 1.307759993395998, "learning_rate": 2.558185878760555e-08, "loss": 0.4279, "step": 27448 }, { "epoch": 0.9779289951368972, "grad_norm": 1.554173355142195, "learning_rate": 2.5499438273346445e-08, "loss": 0.4882, "step": 27449 }, { "epoch": 0.9779646222633913, "grad_norm": 1.2082726063308005, "learning_rate": 2.5417150576533267e-08, "loss": 0.3846, "step": 27450 }, { "epoch": 0.9780002493898855, "grad_norm": 1.3508124619311426, "learning_rate": 2.533499569826292e-08, "loss": 0.4319, "step": 27451 }, { "epoch": 0.9780358765163796, "grad_norm": 1.55452301833711, "learning_rate": 2.5252973639628975e-08, "loss": 0.5818, "step": 27452 }, { "epoch": 0.9780715036428737, "grad_norm": 1.2908221511777171, "learning_rate": 2.517108440172389e-08, "loss": 0.4058, "step": 27453 }, { "epoch": 0.9781071307693678, "grad_norm": 1.6192949704662545, "learning_rate": 2.50893279856379e-08, "loss": 0.517, "step": 27454 }, { "epoch": 0.9781427578958619, "grad_norm": 1.180913753830969, "learning_rate": 2.500770439245903e-08, "loss": 0.3743, "step": 27455 }, { "epoch": 0.978178385022356, "grad_norm": 1.2850939879419605, "learning_rate": 2.4926213623274185e-08, "loss": 0.4381, "step": 27456 }, { "epoch": 0.9782140121488502, "grad_norm": 1.511973635665445, "learning_rate": 2.4844855679170276e-08, "loss": 0.4811, "step": 27457 }, { "epoch": 0.9782496392753443, "grad_norm": 1.5899031809596709, "learning_rate": 2.476363056122866e-08, "loss": 0.5177, "step": 27458 }, { "epoch": 0.9782852664018383, "grad_norm": 1.6616633756845385, "learning_rate": 2.4682538270531797e-08, "loss": 0.4953, "step": 27459 }, { "epoch": 0.9783208935283325, "grad_norm": 1.3555341732473933, "learning_rate": 2.4601578808158832e-08, "loss": 0.4263, "step": 27460 }, { "epoch": 0.9783565206548266, "grad_norm": 1.2483290098029514, "learning_rate": 2.452075217518779e-08, "loss": 0.3868, "step": 27461 }, { "epoch": 0.9783921477813207, "grad_norm": 1.2855438112507331, "learning_rate": 2.4440058372695586e-08, "loss": 0.425, "step": 27462 }, { "epoch": 0.9784277749078149, "grad_norm": 1.1408444553864772, "learning_rate": 2.4359497401758026e-08, "loss": 0.3834, "step": 27463 }, { "epoch": 0.9784634020343089, "grad_norm": 1.3763627955255497, "learning_rate": 2.4279069263444254e-08, "loss": 0.4046, "step": 27464 }, { "epoch": 0.978499029160803, "grad_norm": 1.533557145164996, "learning_rate": 2.419877395882897e-08, "loss": 0.4685, "step": 27465 }, { "epoch": 0.9785346562872972, "grad_norm": 1.3058559032648325, "learning_rate": 2.4118611488979095e-08, "loss": 0.3925, "step": 27466 }, { "epoch": 0.9785702834137913, "grad_norm": 1.2369487325279438, "learning_rate": 2.403858185496266e-08, "loss": 0.3854, "step": 27467 }, { "epoch": 0.9786059105402853, "grad_norm": 1.6532879028389529, "learning_rate": 2.3958685057844378e-08, "loss": 0.5681, "step": 27468 }, { "epoch": 0.9786415376667795, "grad_norm": 1.4170596611080968, "learning_rate": 2.3878921098690055e-08, "loss": 0.4413, "step": 27469 }, { "epoch": 0.9786771647932736, "grad_norm": 1.4090745240708868, "learning_rate": 2.379928997856107e-08, "loss": 0.4392, "step": 27470 }, { "epoch": 0.9787127919197677, "grad_norm": 1.486702112898935, "learning_rate": 2.371979169851657e-08, "loss": 0.4933, "step": 27471 }, { "epoch": 0.9787484190462619, "grad_norm": 1.837985133743345, "learning_rate": 2.364042625961682e-08, "loss": 0.5372, "step": 27472 }, { "epoch": 0.9787840461727559, "grad_norm": 2.036743788520289, "learning_rate": 2.3561193662916536e-08, "loss": 0.4157, "step": 27473 }, { "epoch": 0.97881967329925, "grad_norm": 1.423904556619929, "learning_rate": 2.348209390947376e-08, "loss": 0.5208, "step": 27474 }, { "epoch": 0.9788553004257442, "grad_norm": 1.1747628412997493, "learning_rate": 2.3403127000338756e-08, "loss": 0.38, "step": 27475 }, { "epoch": 0.9788909275522383, "grad_norm": 1.5775869495031098, "learning_rate": 2.332429293656513e-08, "loss": 0.5336, "step": 27476 }, { "epoch": 0.9789265546787324, "grad_norm": 1.61464459913324, "learning_rate": 2.3245591719200932e-08, "loss": 0.5033, "step": 27477 }, { "epoch": 0.9789621818052265, "grad_norm": 1.4168130637286431, "learning_rate": 2.3167023349296435e-08, "loss": 0.3911, "step": 27478 }, { "epoch": 0.9789978089317206, "grad_norm": 1.7589895088086398, "learning_rate": 2.3088587827895247e-08, "loss": 0.3455, "step": 27479 }, { "epoch": 0.9790334360582147, "grad_norm": 1.6650566284121049, "learning_rate": 2.30102851560432e-08, "loss": 0.5269, "step": 27480 }, { "epoch": 0.9790690631847089, "grad_norm": 1.4973103661017855, "learning_rate": 2.29321153347839e-08, "loss": 0.5175, "step": 27481 }, { "epoch": 0.979104690311203, "grad_norm": 1.8700646320693222, "learning_rate": 2.2854078365155406e-08, "loss": 0.5254, "step": 27482 }, { "epoch": 0.979140317437697, "grad_norm": 1.4928187600447922, "learning_rate": 2.2776174248199114e-08, "loss": 0.4426, "step": 27483 }, { "epoch": 0.9791759445641912, "grad_norm": 1.660958168270324, "learning_rate": 2.2698402984951962e-08, "loss": 0.481, "step": 27484 }, { "epoch": 0.9792115716906853, "grad_norm": 1.531730328140928, "learning_rate": 2.2620764576448685e-08, "loss": 0.4583, "step": 27485 }, { "epoch": 0.9792471988171794, "grad_norm": 1.1952787459329166, "learning_rate": 2.2543259023724005e-08, "loss": 0.3617, "step": 27486 }, { "epoch": 0.9792828259436736, "grad_norm": 1.4030231672315723, "learning_rate": 2.246588632780933e-08, "loss": 0.3795, "step": 27487 }, { "epoch": 0.9793184530701676, "grad_norm": 1.2760831499118594, "learning_rate": 2.2388646489736043e-08, "loss": 0.3777, "step": 27488 }, { "epoch": 0.9793540801966617, "grad_norm": 1.4555683343978818, "learning_rate": 2.231153951053111e-08, "loss": 0.4757, "step": 27489 }, { "epoch": 0.9793897073231559, "grad_norm": 1.2168052458630554, "learning_rate": 2.2234565391222595e-08, "loss": 0.4173, "step": 27490 }, { "epoch": 0.97942533444965, "grad_norm": 1.4775855146926964, "learning_rate": 2.2157724132834126e-08, "loss": 0.4972, "step": 27491 }, { "epoch": 0.979460961576144, "grad_norm": 1.8991072191132614, "learning_rate": 2.2081015736390432e-08, "loss": 0.462, "step": 27492 }, { "epoch": 0.9794965887026382, "grad_norm": 1.361062994956421, "learning_rate": 2.2004440202911816e-08, "loss": 0.3811, "step": 27493 }, { "epoch": 0.9795322158291323, "grad_norm": 1.503371219095311, "learning_rate": 2.1927997533418567e-08, "loss": 0.4695, "step": 27494 }, { "epoch": 0.9795678429556264, "grad_norm": 1.2742957343252124, "learning_rate": 2.185168772892765e-08, "loss": 0.3787, "step": 27495 }, { "epoch": 0.9796034700821206, "grad_norm": 1.239892520039253, "learning_rate": 2.1775510790456035e-08, "loss": 0.3837, "step": 27496 }, { "epoch": 0.9796390972086146, "grad_norm": 1.2949747977911628, "learning_rate": 2.1699466719018457e-08, "loss": 0.3512, "step": 27497 }, { "epoch": 0.9796747243351087, "grad_norm": 1.720665237007915, "learning_rate": 2.1623555515626336e-08, "loss": 0.5502, "step": 27498 }, { "epoch": 0.9797103514616029, "grad_norm": 1.8361333705424476, "learning_rate": 2.1547777181291084e-08, "loss": 0.5087, "step": 27499 }, { "epoch": 0.979745978588097, "grad_norm": 1.567832561245413, "learning_rate": 2.1472131717021894e-08, "loss": 0.5137, "step": 27500 }, { "epoch": 0.979781605714591, "grad_norm": 1.7290604663384312, "learning_rate": 2.1396619123825735e-08, "loss": 0.4926, "step": 27501 }, { "epoch": 0.9798172328410852, "grad_norm": 1.309174650525538, "learning_rate": 2.132123940270847e-08, "loss": 0.5345, "step": 27502 }, { "epoch": 0.9798528599675793, "grad_norm": 1.7329930340547373, "learning_rate": 2.1245992554673743e-08, "loss": 0.4721, "step": 27503 }, { "epoch": 0.9798884870940734, "grad_norm": 1.2661931899621435, "learning_rate": 2.1170878580721867e-08, "loss": 0.3793, "step": 27504 }, { "epoch": 0.9799241142205676, "grad_norm": 1.1967042401171792, "learning_rate": 2.1095897481856476e-08, "loss": 0.3743, "step": 27505 }, { "epoch": 0.9799597413470617, "grad_norm": 1.7099810069198051, "learning_rate": 2.1021049259072335e-08, "loss": 0.4074, "step": 27506 }, { "epoch": 0.9799953684735557, "grad_norm": 1.2388549316199597, "learning_rate": 2.0946333913368645e-08, "loss": 0.4487, "step": 27507 }, { "epoch": 0.9800309956000499, "grad_norm": 1.2002533908859598, "learning_rate": 2.0871751445740163e-08, "loss": 0.3553, "step": 27508 }, { "epoch": 0.980066622726544, "grad_norm": 1.3299461555438155, "learning_rate": 2.079730185717832e-08, "loss": 0.423, "step": 27509 }, { "epoch": 0.9801022498530381, "grad_norm": 1.1277291835528287, "learning_rate": 2.0722985148675655e-08, "loss": 0.3812, "step": 27510 }, { "epoch": 0.9801378769795323, "grad_norm": 1.845992692893874, "learning_rate": 2.064880132122249e-08, "loss": 0.5404, "step": 27511 }, { "epoch": 0.9801735041060263, "grad_norm": 1.312990385647662, "learning_rate": 2.0574750375804696e-08, "loss": 0.4375, "step": 27512 }, { "epoch": 0.9802091312325204, "grad_norm": 1.6919731566949572, "learning_rate": 2.050083231340927e-08, "loss": 0.5448, "step": 27513 }, { "epoch": 0.9802447583590146, "grad_norm": 1.6094482547277498, "learning_rate": 2.042704713501986e-08, "loss": 0.4955, "step": 27514 }, { "epoch": 0.9802803854855087, "grad_norm": 1.3901224715812106, "learning_rate": 2.0353394841621244e-08, "loss": 0.4291, "step": 27515 }, { "epoch": 0.9803160126120027, "grad_norm": 1.5133609443827096, "learning_rate": 2.027987543419152e-08, "loss": 0.4486, "step": 27516 }, { "epoch": 0.9803516397384969, "grad_norm": 1.5545354801707507, "learning_rate": 2.020648891371102e-08, "loss": 0.6059, "step": 27517 }, { "epoch": 0.980387266864991, "grad_norm": 1.701586321715084, "learning_rate": 2.013323528115674e-08, "loss": 0.5944, "step": 27518 }, { "epoch": 0.9804228939914851, "grad_norm": 2.8928943655801387, "learning_rate": 2.006011453750345e-08, "loss": 0.4307, "step": 27519 }, { "epoch": 0.9804585211179793, "grad_norm": 1.0749101336084554, "learning_rate": 1.9987126683724822e-08, "loss": 0.3694, "step": 27520 }, { "epoch": 0.9804941482444733, "grad_norm": 1.557399790383912, "learning_rate": 1.991427172079452e-08, "loss": 0.4663, "step": 27521 }, { "epoch": 0.9805297753709674, "grad_norm": 1.866613661490973, "learning_rate": 1.9841549649680658e-08, "loss": 0.5662, "step": 27522 }, { "epoch": 0.9805654024974616, "grad_norm": 1.1888991338515535, "learning_rate": 1.976896047135246e-08, "loss": 0.406, "step": 27523 }, { "epoch": 0.9806010296239557, "grad_norm": 1.9126749489077957, "learning_rate": 1.9696504186775822e-08, "loss": 0.4372, "step": 27524 }, { "epoch": 0.9806366567504498, "grad_norm": 2.02565420847985, "learning_rate": 1.9624180796916637e-08, "loss": 0.4336, "step": 27525 }, { "epoch": 0.9806722838769439, "grad_norm": 1.2446231744324794, "learning_rate": 1.9551990302736357e-08, "loss": 0.4142, "step": 27526 }, { "epoch": 0.980707911003438, "grad_norm": 1.4878763246247608, "learning_rate": 1.9479932705198658e-08, "loss": 0.4146, "step": 27527 }, { "epoch": 0.9807435381299321, "grad_norm": 1.5136647999928545, "learning_rate": 1.9408008005260548e-08, "loss": 0.4671, "step": 27528 }, { "epoch": 0.9807791652564263, "grad_norm": 1.1408568929286078, "learning_rate": 1.9336216203881265e-08, "loss": 0.344, "step": 27529 }, { "epoch": 0.9808147923829204, "grad_norm": 1.492389987695497, "learning_rate": 1.9264557302016706e-08, "loss": 0.3477, "step": 27530 }, { "epoch": 0.9808504195094144, "grad_norm": 1.5620795614013925, "learning_rate": 1.9193031300620558e-08, "loss": 0.4966, "step": 27531 }, { "epoch": 0.9808860466359086, "grad_norm": 1.6218107928340595, "learning_rate": 1.9121638200644276e-08, "loss": 0.5471, "step": 27532 }, { "epoch": 0.9809216737624027, "grad_norm": 1.6714807789454762, "learning_rate": 1.9050378003040438e-08, "loss": 0.4511, "step": 27533 }, { "epoch": 0.9809573008888968, "grad_norm": 1.2748484840934284, "learning_rate": 1.897925070875717e-08, "loss": 0.3728, "step": 27534 }, { "epoch": 0.980992928015391, "grad_norm": 1.2322287733570547, "learning_rate": 1.890825631874149e-08, "loss": 0.4087, "step": 27535 }, { "epoch": 0.981028555141885, "grad_norm": 1.6711205693557807, "learning_rate": 1.8837394833938207e-08, "loss": 0.533, "step": 27536 }, { "epoch": 0.9810641822683791, "grad_norm": 1.4019545870489087, "learning_rate": 1.876666625529211e-08, "loss": 0.4159, "step": 27537 }, { "epoch": 0.9810998093948733, "grad_norm": 1.6811676311772472, "learning_rate": 1.8696070583744674e-08, "loss": 0.5409, "step": 27538 }, { "epoch": 0.9811354365213674, "grad_norm": 1.2098894271013543, "learning_rate": 1.8625607820234037e-08, "loss": 0.4213, "step": 27539 }, { "epoch": 0.9811710636478614, "grad_norm": 1.5596817923688828, "learning_rate": 1.855527796570167e-08, "loss": 0.5288, "step": 27540 }, { "epoch": 0.9812066907743556, "grad_norm": 1.2059034944238354, "learning_rate": 1.8485081021081263e-08, "loss": 0.3931, "step": 27541 }, { "epoch": 0.9812423179008497, "grad_norm": 1.1669142245213833, "learning_rate": 1.8415016987308742e-08, "loss": 0.3765, "step": 27542 }, { "epoch": 0.9812779450273438, "grad_norm": 1.691384840175415, "learning_rate": 1.83450858653178e-08, "loss": 0.5122, "step": 27543 }, { "epoch": 0.981313572153838, "grad_norm": 1.5309644705852548, "learning_rate": 1.82752876560377e-08, "loss": 0.5037, "step": 27544 }, { "epoch": 0.981349199280332, "grad_norm": 1.4156879910981555, "learning_rate": 1.820562236039991e-08, "loss": 0.4427, "step": 27545 }, { "epoch": 0.9813848264068261, "grad_norm": 1.5234768988447238, "learning_rate": 1.8136089979331473e-08, "loss": 0.3586, "step": 27546 }, { "epoch": 0.9814204535333203, "grad_norm": 1.8371890903902106, "learning_rate": 1.806669051375609e-08, "loss": 0.5082, "step": 27547 }, { "epoch": 0.9814560806598144, "grad_norm": 1.8740103324906119, "learning_rate": 1.7997423964601913e-08, "loss": 0.4984, "step": 27548 }, { "epoch": 0.9814917077863085, "grad_norm": 1.3896965769838514, "learning_rate": 1.7928290332787092e-08, "loss": 0.4129, "step": 27549 }, { "epoch": 0.9815273349128026, "grad_norm": 1.4790208097616886, "learning_rate": 1.7859289619235332e-08, "loss": 0.4762, "step": 27550 }, { "epoch": 0.9815629620392967, "grad_norm": 2.5832363659792414, "learning_rate": 1.779042182486368e-08, "loss": 0.6098, "step": 27551 }, { "epoch": 0.9815985891657908, "grad_norm": 1.2775977014268947, "learning_rate": 1.7721686950589178e-08, "loss": 0.4159, "step": 27552 }, { "epoch": 0.981634216292285, "grad_norm": 1.2279971789948865, "learning_rate": 1.765308499732887e-08, "loss": 0.3753, "step": 27553 }, { "epoch": 0.981669843418779, "grad_norm": 1.2864630589033113, "learning_rate": 1.7584615965994256e-08, "loss": 0.3946, "step": 27554 }, { "epoch": 0.9817054705452731, "grad_norm": 1.6724722264625769, "learning_rate": 1.7516279857497933e-08, "loss": 0.4639, "step": 27555 }, { "epoch": 0.9817410976717673, "grad_norm": 1.4563368185209369, "learning_rate": 1.744807667275028e-08, "loss": 0.3561, "step": 27556 }, { "epoch": 0.9817767247982614, "grad_norm": 1.697086930361373, "learning_rate": 1.7380006412657246e-08, "loss": 0.3979, "step": 27557 }, { "epoch": 0.9818123519247555, "grad_norm": 1.4392697171835989, "learning_rate": 1.73120690781281e-08, "loss": 0.4445, "step": 27558 }, { "epoch": 0.9818479790512497, "grad_norm": 1.2622423372866383, "learning_rate": 1.724426467006657e-08, "loss": 0.4204, "step": 27559 }, { "epoch": 0.9818836061777437, "grad_norm": 1.4693728196061224, "learning_rate": 1.7176593189374145e-08, "loss": 0.5017, "step": 27560 }, { "epoch": 0.9819192333042378, "grad_norm": 1.535784641511379, "learning_rate": 1.710905463695345e-08, "loss": 0.4935, "step": 27561 }, { "epoch": 0.981954860430732, "grad_norm": 1.2618567987138207, "learning_rate": 1.7041649013703754e-08, "loss": 0.3787, "step": 27562 }, { "epoch": 0.9819904875572261, "grad_norm": 1.4224292622251526, "learning_rate": 1.6974376320521014e-08, "loss": 0.4988, "step": 27563 }, { "epoch": 0.9820261146837201, "grad_norm": 1.4524060181704528, "learning_rate": 1.6907236558303397e-08, "loss": 0.5595, "step": 27564 }, { "epoch": 0.9820617418102143, "grad_norm": 1.1709251024528344, "learning_rate": 1.6840229727942415e-08, "loss": 0.3953, "step": 27565 }, { "epoch": 0.9820973689367084, "grad_norm": 1.2634543125230493, "learning_rate": 1.6773355830332903e-08, "loss": 0.3564, "step": 27566 }, { "epoch": 0.9821329960632025, "grad_norm": 1.7980334476083182, "learning_rate": 1.6706614866363048e-08, "loss": 0.4901, "step": 27567 }, { "epoch": 0.9821686231896967, "grad_norm": 1.781252291702833, "learning_rate": 1.664000683692324e-08, "loss": 0.5334, "step": 27568 }, { "epoch": 0.9822042503161907, "grad_norm": 1.4670370064282199, "learning_rate": 1.657353174289944e-08, "loss": 0.6115, "step": 27569 }, { "epoch": 0.9822398774426848, "grad_norm": 1.6331822928581896, "learning_rate": 1.6507189585177606e-08, "loss": 0.4816, "step": 27570 }, { "epoch": 0.982275504569179, "grad_norm": 1.2896693118102331, "learning_rate": 1.6440980364639258e-08, "loss": 0.4224, "step": 27571 }, { "epoch": 0.9823111316956731, "grad_norm": 1.5638930946512317, "learning_rate": 1.637490408216813e-08, "loss": 0.5277, "step": 27572 }, { "epoch": 0.9823467588221672, "grad_norm": 1.4495671217756356, "learning_rate": 1.630896073864352e-08, "loss": 0.5115, "step": 27573 }, { "epoch": 0.9823823859486613, "grad_norm": 1.133704688718348, "learning_rate": 1.6243150334942503e-08, "loss": 0.3871, "step": 27574 }, { "epoch": 0.9824180130751554, "grad_norm": 1.3593989583118877, "learning_rate": 1.6177472871943268e-08, "loss": 0.3901, "step": 27575 }, { "epoch": 0.9824536402016495, "grad_norm": 1.676287593172337, "learning_rate": 1.6111928350518445e-08, "loss": 0.5073, "step": 27576 }, { "epoch": 0.9824892673281437, "grad_norm": 1.5425672412048972, "learning_rate": 1.6046516771541787e-08, "loss": 0.5529, "step": 27577 }, { "epoch": 0.9825248944546378, "grad_norm": 1.8049281770946737, "learning_rate": 1.5981238135883705e-08, "loss": 0.541, "step": 27578 }, { "epoch": 0.9825605215811318, "grad_norm": 1.1873294169535367, "learning_rate": 1.591609244441461e-08, "loss": 0.3845, "step": 27579 }, { "epoch": 0.982596148707626, "grad_norm": 1.6849639117080684, "learning_rate": 1.585107969800048e-08, "loss": 0.4323, "step": 27580 }, { "epoch": 0.9826317758341201, "grad_norm": 1.3448809907514645, "learning_rate": 1.5786199897508402e-08, "loss": 0.3675, "step": 27581 }, { "epoch": 0.9826674029606142, "grad_norm": 1.6527403643982954, "learning_rate": 1.5721453043801014e-08, "loss": 0.5178, "step": 27582 }, { "epoch": 0.9827030300871084, "grad_norm": 1.2234975357305633, "learning_rate": 1.5656839137742073e-08, "loss": 0.4041, "step": 27583 }, { "epoch": 0.9827386572136024, "grad_norm": 1.7843909373349132, "learning_rate": 1.5592358180189782e-08, "loss": 0.5514, "step": 27584 }, { "epoch": 0.9827742843400965, "grad_norm": 1.3814486398446424, "learning_rate": 1.552801017200456e-08, "loss": 0.4136, "step": 27585 }, { "epoch": 0.9828099114665907, "grad_norm": 1.144149488789787, "learning_rate": 1.546379511404239e-08, "loss": 0.4089, "step": 27586 }, { "epoch": 0.9828455385930848, "grad_norm": 1.2564155624649254, "learning_rate": 1.5399713007159255e-08, "loss": 0.4107, "step": 27587 }, { "epoch": 0.9828811657195788, "grad_norm": 1.2825134516034362, "learning_rate": 1.5335763852206697e-08, "loss": 0.3995, "step": 27588 }, { "epoch": 0.982916792846073, "grad_norm": 1.390604420014176, "learning_rate": 1.5271947650037365e-08, "loss": 0.4108, "step": 27589 }, { "epoch": 0.9829524199725671, "grad_norm": 1.5012127074122528, "learning_rate": 1.5208264401501695e-08, "loss": 0.4278, "step": 27590 }, { "epoch": 0.9829880470990612, "grad_norm": 1.3888226545639968, "learning_rate": 1.5144714107446777e-08, "loss": 0.4126, "step": 27591 }, { "epoch": 0.9830236742255554, "grad_norm": 1.2124345864680592, "learning_rate": 1.5081296768718613e-08, "loss": 0.3865, "step": 27592 }, { "epoch": 0.9830593013520494, "grad_norm": 1.3318555924415234, "learning_rate": 1.5018012386162072e-08, "loss": 0.4073, "step": 27593 }, { "epoch": 0.9830949284785435, "grad_norm": 1.4759959868935706, "learning_rate": 1.4954860960619823e-08, "loss": 0.4963, "step": 27594 }, { "epoch": 0.9831305556050377, "grad_norm": 1.3547644657928637, "learning_rate": 1.4891842492933406e-08, "loss": 0.3676, "step": 27595 }, { "epoch": 0.9831661827315318, "grad_norm": 1.205482600493113, "learning_rate": 1.4828956983941046e-08, "loss": 0.3778, "step": 27596 }, { "epoch": 0.9832018098580259, "grad_norm": 1.797281325573637, "learning_rate": 1.4766204434479848e-08, "loss": 0.431, "step": 27597 }, { "epoch": 0.98323743698452, "grad_norm": 1.2449163265412462, "learning_rate": 1.470358484538581e-08, "loss": 0.3668, "step": 27598 }, { "epoch": 0.9832730641110141, "grad_norm": 1.3334751708828192, "learning_rate": 1.4641098217493821e-08, "loss": 0.3817, "step": 27599 }, { "epoch": 0.9833086912375082, "grad_norm": 1.7303818007511238, "learning_rate": 1.4578744551633217e-08, "loss": 0.4895, "step": 27600 }, { "epoch": 0.9833443183640024, "grad_norm": 1.5794434808938163, "learning_rate": 1.4516523848637776e-08, "loss": 0.4368, "step": 27601 }, { "epoch": 0.9833799454904965, "grad_norm": 1.5578115992385568, "learning_rate": 1.4454436109333502e-08, "loss": 0.4788, "step": 27602 }, { "epoch": 0.9834155726169905, "grad_norm": 1.3069932211900583, "learning_rate": 1.4392481334547515e-08, "loss": 0.4248, "step": 27603 }, { "epoch": 0.9834511997434847, "grad_norm": 1.5297828900547286, "learning_rate": 1.4330659525105817e-08, "loss": 0.5772, "step": 27604 }, { "epoch": 0.9834868268699788, "grad_norm": 1.4218723050602526, "learning_rate": 1.4268970681831085e-08, "loss": 0.4321, "step": 27605 }, { "epoch": 0.9835224539964729, "grad_norm": 1.4614039248172095, "learning_rate": 1.4207414805543774e-08, "loss": 0.4707, "step": 27606 }, { "epoch": 0.983558081122967, "grad_norm": 1.3036871296408077, "learning_rate": 1.4145991897064336e-08, "loss": 0.4182, "step": 27607 }, { "epoch": 0.9835937082494611, "grad_norm": 1.470431724546257, "learning_rate": 1.4084701957212122e-08, "loss": 0.496, "step": 27608 }, { "epoch": 0.9836293353759552, "grad_norm": 1.2072322768622321, "learning_rate": 1.4023544986800918e-08, "loss": 0.3745, "step": 27609 }, { "epoch": 0.9836649625024494, "grad_norm": 1.756170156159951, "learning_rate": 1.3962520986645634e-08, "loss": 0.4402, "step": 27610 }, { "epoch": 0.9837005896289435, "grad_norm": 1.5157093598206712, "learning_rate": 1.3901629957558948e-08, "loss": 0.5245, "step": 27611 }, { "epoch": 0.9837362167554375, "grad_norm": 1.3947245456986304, "learning_rate": 1.3840871900353547e-08, "loss": 0.4007, "step": 27612 }, { "epoch": 0.9837718438819317, "grad_norm": 1.5664436403033517, "learning_rate": 1.3780246815835452e-08, "loss": 0.5573, "step": 27613 }, { "epoch": 0.9838074710084258, "grad_norm": 3.0292180477501875, "learning_rate": 1.3719754704814015e-08, "loss": 0.4448, "step": 27614 }, { "epoch": 0.9838430981349199, "grad_norm": 1.283805923453075, "learning_rate": 1.3659395568093037e-08, "loss": 0.419, "step": 27615 }, { "epoch": 0.9838787252614141, "grad_norm": 1.645967114999942, "learning_rate": 1.359916940647854e-08, "loss": 0.5631, "step": 27616 }, { "epoch": 0.9839143523879081, "grad_norm": 1.3970867528443067, "learning_rate": 1.3539076220769887e-08, "loss": 0.3711, "step": 27617 }, { "epoch": 0.9839499795144022, "grad_norm": 1.5427556188749085, "learning_rate": 1.3479116011769766e-08, "loss": 0.4541, "step": 27618 }, { "epoch": 0.9839856066408964, "grad_norm": 1.2823109933256722, "learning_rate": 1.341928878027421e-08, "loss": 0.3856, "step": 27619 }, { "epoch": 0.9840212337673905, "grad_norm": 1.3714093760398822, "learning_rate": 1.3359594527082575e-08, "loss": 0.3565, "step": 27620 }, { "epoch": 0.9840568608938846, "grad_norm": 1.1566701278889595, "learning_rate": 1.3300033252986454e-08, "loss": 0.3988, "step": 27621 }, { "epoch": 0.9840924880203787, "grad_norm": 2.1545459945243834, "learning_rate": 1.3240604958781877e-08, "loss": 0.4731, "step": 27622 }, { "epoch": 0.9841281151468728, "grad_norm": 1.5872192672655787, "learning_rate": 1.3181309645259321e-08, "loss": 0.4496, "step": 27623 }, { "epoch": 0.9841637422733669, "grad_norm": 1.2596214410370958, "learning_rate": 1.3122147313207046e-08, "loss": 0.4177, "step": 27624 }, { "epoch": 0.9841993693998611, "grad_norm": 1.2067580079070461, "learning_rate": 1.3063117963413307e-08, "loss": 0.4106, "step": 27625 }, { "epoch": 0.9842349965263552, "grad_norm": 1.3643013572207043, "learning_rate": 1.3004221596665256e-08, "loss": 0.5183, "step": 27626 }, { "epoch": 0.9842706236528492, "grad_norm": 2.0135298842924567, "learning_rate": 1.294545821374671e-08, "loss": 0.4115, "step": 27627 }, { "epoch": 0.9843062507793434, "grad_norm": 1.5886727325642178, "learning_rate": 1.2886827815440373e-08, "loss": 0.429, "step": 27628 }, { "epoch": 0.9843418779058375, "grad_norm": 1.7313859043635809, "learning_rate": 1.2828330402525624e-08, "loss": 0.4582, "step": 27629 }, { "epoch": 0.9843775050323316, "grad_norm": 1.0611425699130748, "learning_rate": 1.276996597578295e-08, "loss": 0.3608, "step": 27630 }, { "epoch": 0.9844131321588258, "grad_norm": 1.1760561260404347, "learning_rate": 1.2711734535988396e-08, "loss": 0.375, "step": 27631 }, { "epoch": 0.9844487592853198, "grad_norm": 1.6966774368502198, "learning_rate": 1.2653636083916898e-08, "loss": 0.5835, "step": 27632 }, { "epoch": 0.9844843864118139, "grad_norm": 1.4099116906069975, "learning_rate": 1.2595670620344502e-08, "loss": 0.4762, "step": 27633 }, { "epoch": 0.9845200135383081, "grad_norm": 1.7859562110576934, "learning_rate": 1.2537838146040592e-08, "loss": 0.5843, "step": 27634 }, { "epoch": 0.9845556406648022, "grad_norm": 1.3891854202759684, "learning_rate": 1.2480138661776776e-08, "loss": 0.3996, "step": 27635 }, { "epoch": 0.9845912677912962, "grad_norm": 1.4508749145377369, "learning_rate": 1.2422572168320212e-08, "loss": 0.4309, "step": 27636 }, { "epoch": 0.9846268949177904, "grad_norm": 1.7109722047279043, "learning_rate": 1.236513866643807e-08, "loss": 0.3872, "step": 27637 }, { "epoch": 0.9846625220442845, "grad_norm": 1.5639003126775437, "learning_rate": 1.230783815689529e-08, "loss": 0.4919, "step": 27638 }, { "epoch": 0.9846981491707786, "grad_norm": 1.5901880471617187, "learning_rate": 1.2250670640454599e-08, "loss": 0.5682, "step": 27639 }, { "epoch": 0.9847337762972728, "grad_norm": 1.6647982781093544, "learning_rate": 1.2193636117877604e-08, "loss": 0.4866, "step": 27640 }, { "epoch": 0.9847694034237668, "grad_norm": 1.2369590800408274, "learning_rate": 1.2136734589922594e-08, "loss": 0.4438, "step": 27641 }, { "epoch": 0.9848050305502609, "grad_norm": 1.642454271779185, "learning_rate": 1.2079966057348957e-08, "loss": 0.4887, "step": 27642 }, { "epoch": 0.9848406576767551, "grad_norm": 1.2224410639615233, "learning_rate": 1.2023330520911648e-08, "loss": 0.4131, "step": 27643 }, { "epoch": 0.9848762848032492, "grad_norm": 1.586675878287297, "learning_rate": 1.1966827981364505e-08, "loss": 0.4968, "step": 27644 }, { "epoch": 0.9849119119297433, "grad_norm": 1.7189887597023623, "learning_rate": 1.191045843946026e-08, "loss": 0.4334, "step": 27645 }, { "epoch": 0.9849475390562374, "grad_norm": 1.3262422513316052, "learning_rate": 1.1854221895949425e-08, "loss": 0.3951, "step": 27646 }, { "epoch": 0.9849831661827315, "grad_norm": 2.231323644376963, "learning_rate": 1.179811835158029e-08, "loss": 0.434, "step": 27647 }, { "epoch": 0.9850187933092256, "grad_norm": 1.344148336456536, "learning_rate": 1.1742147807101145e-08, "loss": 0.3714, "step": 27648 }, { "epoch": 0.9850544204357198, "grad_norm": 1.6125854069404595, "learning_rate": 1.168631026325695e-08, "loss": 0.5371, "step": 27649 }, { "epoch": 0.9850900475622139, "grad_norm": 1.7339921417262365, "learning_rate": 1.1630605720791555e-08, "loss": 0.5647, "step": 27650 }, { "epoch": 0.9851256746887079, "grad_norm": 1.2510748328031136, "learning_rate": 1.1575034180445477e-08, "loss": 0.4054, "step": 27651 }, { "epoch": 0.9851613018152021, "grad_norm": 1.3484516020578847, "learning_rate": 1.1519595642959236e-08, "loss": 0.4245, "step": 27652 }, { "epoch": 0.9851969289416962, "grad_norm": 1.5711917860311535, "learning_rate": 1.146429010907002e-08, "loss": 0.5039, "step": 27653 }, { "epoch": 0.9852325560681903, "grad_norm": 1.6015233222117586, "learning_rate": 1.1409117579517237e-08, "loss": 0.5289, "step": 27654 }, { "epoch": 0.9852681831946845, "grad_norm": 1.2409978325608926, "learning_rate": 1.1354078055032525e-08, "loss": 0.3879, "step": 27655 }, { "epoch": 0.9853038103211785, "grad_norm": 1.6055322034433794, "learning_rate": 1.129917153634974e-08, "loss": 0.5231, "step": 27656 }, { "epoch": 0.9853394374476726, "grad_norm": 1.2410653917942818, "learning_rate": 1.124439802420052e-08, "loss": 0.4033, "step": 27657 }, { "epoch": 0.9853750645741668, "grad_norm": 1.1805513883767105, "learning_rate": 1.1189757519314282e-08, "loss": 0.3919, "step": 27658 }, { "epoch": 0.9854106917006609, "grad_norm": 1.3707833524065665, "learning_rate": 1.113525002241822e-08, "loss": 0.5464, "step": 27659 }, { "epoch": 0.9854463188271549, "grad_norm": 1.2364498423252415, "learning_rate": 1.1080875534237312e-08, "loss": 0.3759, "step": 27660 }, { "epoch": 0.9854819459536491, "grad_norm": 1.9878442771304496, "learning_rate": 1.1026634055497643e-08, "loss": 0.5451, "step": 27661 }, { "epoch": 0.9855175730801432, "grad_norm": 1.408235968748902, "learning_rate": 1.0972525586918636e-08, "loss": 0.3922, "step": 27662 }, { "epoch": 0.9855532002066373, "grad_norm": 1.6765182001796977, "learning_rate": 1.0918550129223049e-08, "loss": 0.4695, "step": 27663 }, { "epoch": 0.9855888273331315, "grad_norm": 2.0423198487926455, "learning_rate": 1.0864707683129194e-08, "loss": 0.5077, "step": 27664 }, { "epoch": 0.9856244544596255, "grad_norm": 1.6058187871993668, "learning_rate": 1.0810998249353167e-08, "loss": 0.4659, "step": 27665 }, { "epoch": 0.9856600815861196, "grad_norm": 1.302842281916427, "learning_rate": 1.075742182861106e-08, "loss": 0.3762, "step": 27666 }, { "epoch": 0.9856957087126138, "grad_norm": 1.3491749467025538, "learning_rate": 1.0703978421616745e-08, "loss": 0.4034, "step": 27667 }, { "epoch": 0.9857313358391079, "grad_norm": 1.6721420595485228, "learning_rate": 1.0650668029079657e-08, "loss": 0.4921, "step": 27668 }, { "epoch": 0.985766962965602, "grad_norm": 1.3970138478613778, "learning_rate": 1.059749065171145e-08, "loss": 0.3934, "step": 27669 }, { "epoch": 0.9858025900920961, "grad_norm": 1.4066279751559678, "learning_rate": 1.0544446290220445e-08, "loss": 0.4331, "step": 27670 }, { "epoch": 0.9858382172185902, "grad_norm": 1.3753890665378286, "learning_rate": 1.0491534945312742e-08, "loss": 0.43, "step": 27671 }, { "epoch": 0.9858738443450843, "grad_norm": 1.2618690695244486, "learning_rate": 1.0438756617691115e-08, "loss": 0.366, "step": 27672 }, { "epoch": 0.9859094714715785, "grad_norm": 1.5995394649548365, "learning_rate": 1.0386111308060554e-08, "loss": 0.5359, "step": 27673 }, { "epoch": 0.9859450985980726, "grad_norm": 1.4703823817368358, "learning_rate": 1.03335990171205e-08, "loss": 0.4933, "step": 27674 }, { "epoch": 0.9859807257245666, "grad_norm": 1.6764930102115931, "learning_rate": 1.0281219745571503e-08, "loss": 0.5038, "step": 27675 }, { "epoch": 0.9860163528510608, "grad_norm": 1.3621781443135392, "learning_rate": 1.0228973494110783e-08, "loss": 0.4163, "step": 27676 }, { "epoch": 0.9860519799775549, "grad_norm": 1.376397679005195, "learning_rate": 1.0176860263433342e-08, "loss": 0.4739, "step": 27677 }, { "epoch": 0.986087607104049, "grad_norm": 1.5873916103156065, "learning_rate": 1.0124880054233066e-08, "loss": 0.559, "step": 27678 }, { "epoch": 0.9861232342305432, "grad_norm": 1.437807956262821, "learning_rate": 1.0073032867203846e-08, "loss": 0.4907, "step": 27679 }, { "epoch": 0.9861588613570372, "grad_norm": 1.701384934851131, "learning_rate": 1.0021318703034022e-08, "loss": 0.5011, "step": 27680 }, { "epoch": 0.9861944884835313, "grad_norm": 1.2934961461992898, "learning_rate": 9.96973756241193e-09, "loss": 0.3603, "step": 27681 }, { "epoch": 0.9862301156100255, "grad_norm": 1.763130674331258, "learning_rate": 9.918289446027019e-09, "loss": 0.5206, "step": 27682 }, { "epoch": 0.9862657427365196, "grad_norm": 1.5462601369929152, "learning_rate": 9.866974354560966e-09, "loss": 0.4826, "step": 27683 }, { "epoch": 0.9863013698630136, "grad_norm": 1.435610700172088, "learning_rate": 9.815792288699887e-09, "loss": 0.3275, "step": 27684 }, { "epoch": 0.9863369969895078, "grad_norm": 1.2527883462854996, "learning_rate": 9.764743249124353e-09, "loss": 0.373, "step": 27685 }, { "epoch": 0.9863726241160019, "grad_norm": 1.359277931153971, "learning_rate": 9.713827236513818e-09, "loss": 0.3711, "step": 27686 }, { "epoch": 0.986408251242496, "grad_norm": 1.2145586373209438, "learning_rate": 9.663044251546628e-09, "loss": 0.3906, "step": 27687 }, { "epoch": 0.9864438783689902, "grad_norm": 2.856173051560131, "learning_rate": 9.612394294898908e-09, "loss": 0.58, "step": 27688 }, { "epoch": 0.9864795054954842, "grad_norm": 1.174122458460476, "learning_rate": 9.561877367244565e-09, "loss": 0.3821, "step": 27689 }, { "epoch": 0.9865151326219783, "grad_norm": 1.4146433790144035, "learning_rate": 9.511493469256395e-09, "loss": 0.4006, "step": 27690 }, { "epoch": 0.9865507597484725, "grad_norm": 1.8371687599603514, "learning_rate": 9.46124260160608e-09, "loss": 0.5426, "step": 27691 }, { "epoch": 0.9865863868749666, "grad_norm": 1.3312050653794347, "learning_rate": 9.411124764963087e-09, "loss": 0.3933, "step": 27692 }, { "epoch": 0.9866220140014608, "grad_norm": 1.5324046555504427, "learning_rate": 9.361139959993549e-09, "loss": 0.513, "step": 27693 }, { "epoch": 0.9866576411279548, "grad_norm": 1.5411866501664526, "learning_rate": 9.311288187362487e-09, "loss": 0.4596, "step": 27694 }, { "epoch": 0.9866932682544489, "grad_norm": 1.3888893769221673, "learning_rate": 9.261569447736041e-09, "loss": 0.4195, "step": 27695 }, { "epoch": 0.9867288953809431, "grad_norm": 1.4192644128781504, "learning_rate": 9.211983741773678e-09, "loss": 0.4822, "step": 27696 }, { "epoch": 0.9867645225074372, "grad_norm": 1.4828734134120887, "learning_rate": 9.162531070138203e-09, "loss": 0.5258, "step": 27697 }, { "epoch": 0.9868001496339313, "grad_norm": 1.3236286282882337, "learning_rate": 9.113211433485758e-09, "loss": 0.4163, "step": 27698 }, { "epoch": 0.9868357767604254, "grad_norm": 1.687618472842527, "learning_rate": 9.064024832474704e-09, "loss": 0.5404, "step": 27699 }, { "epoch": 0.9868714038869195, "grad_norm": 1.2989232515892946, "learning_rate": 9.014971267758965e-09, "loss": 0.39, "step": 27700 }, { "epoch": 0.9869070310134136, "grad_norm": 1.6306916354554633, "learning_rate": 8.96605073999246e-09, "loss": 0.5167, "step": 27701 }, { "epoch": 0.9869426581399078, "grad_norm": 1.2916421268059741, "learning_rate": 8.917263249825781e-09, "loss": 0.412, "step": 27702 }, { "epoch": 0.9869782852664019, "grad_norm": 1.2918356743728552, "learning_rate": 8.868608797909517e-09, "loss": 0.4386, "step": 27703 }, { "epoch": 0.9870139123928959, "grad_norm": 1.3901628977565756, "learning_rate": 8.82008738489093e-09, "loss": 0.3956, "step": 27704 }, { "epoch": 0.9870495395193901, "grad_norm": 1.233243696507393, "learning_rate": 8.771699011416169e-09, "loss": 0.3364, "step": 27705 }, { "epoch": 0.9870851666458842, "grad_norm": 1.4454630136235656, "learning_rate": 8.723443678130272e-09, "loss": 0.3669, "step": 27706 }, { "epoch": 0.9871207937723783, "grad_norm": 1.1234382733393158, "learning_rate": 8.67532138567495e-09, "loss": 0.3871, "step": 27707 }, { "epoch": 0.9871564208988725, "grad_norm": 2.0084297752570848, "learning_rate": 8.627332134690802e-09, "loss": 0.4618, "step": 27708 }, { "epoch": 0.9871920480253665, "grad_norm": 1.2096397136748478, "learning_rate": 8.579475925818426e-09, "loss": 0.3983, "step": 27709 }, { "epoch": 0.9872276751518606, "grad_norm": 1.445481599403251, "learning_rate": 8.531752759692869e-09, "loss": 0.4018, "step": 27710 }, { "epoch": 0.9872633022783548, "grad_norm": 1.1720425284201532, "learning_rate": 8.4841626369514e-09, "loss": 0.3896, "step": 27711 }, { "epoch": 0.9872989294048489, "grad_norm": 1.6316330261346197, "learning_rate": 8.436705558225733e-09, "loss": 0.5187, "step": 27712 }, { "epoch": 0.9873345565313429, "grad_norm": 1.6014510107487008, "learning_rate": 8.389381524149808e-09, "loss": 0.608, "step": 27713 }, { "epoch": 0.9873701836578371, "grad_norm": 1.2052623907056377, "learning_rate": 8.342190535353123e-09, "loss": 0.416, "step": 27714 }, { "epoch": 0.9874058107843312, "grad_norm": 1.2892775058916994, "learning_rate": 8.295132592464062e-09, "loss": 0.3977, "step": 27715 }, { "epoch": 0.9874414379108253, "grad_norm": 1.300574126908433, "learning_rate": 8.248207696108789e-09, "loss": 0.4141, "step": 27716 }, { "epoch": 0.9874770650373195, "grad_norm": 1.4331431106282404, "learning_rate": 8.201415846912364e-09, "loss": 0.4126, "step": 27717 }, { "epoch": 0.9875126921638135, "grad_norm": 1.3631362840522379, "learning_rate": 8.154757045497618e-09, "loss": 0.3289, "step": 27718 }, { "epoch": 0.9875483192903076, "grad_norm": 1.6929332134571804, "learning_rate": 8.10823129248739e-09, "loss": 0.4652, "step": 27719 }, { "epoch": 0.9875839464168018, "grad_norm": 1.2899044184476733, "learning_rate": 8.06183858849896e-09, "loss": 0.425, "step": 27720 }, { "epoch": 0.9876195735432959, "grad_norm": 1.2970028616115503, "learning_rate": 8.015578934150726e-09, "loss": 0.4121, "step": 27721 }, { "epoch": 0.98765520066979, "grad_norm": 1.8129378176314448, "learning_rate": 7.969452330059969e-09, "loss": 0.4661, "step": 27722 }, { "epoch": 0.9876908277962841, "grad_norm": 1.724165465138258, "learning_rate": 7.923458776839532e-09, "loss": 0.4822, "step": 27723 }, { "epoch": 0.9877264549227782, "grad_norm": 1.2762918159106031, "learning_rate": 7.877598275102261e-09, "loss": 0.3751, "step": 27724 }, { "epoch": 0.9877620820492723, "grad_norm": 1.3840326477230311, "learning_rate": 7.831870825458777e-09, "loss": 0.412, "step": 27725 }, { "epoch": 0.9877977091757665, "grad_norm": 1.2097660422367251, "learning_rate": 7.786276428518592e-09, "loss": 0.4164, "step": 27726 }, { "epoch": 0.9878333363022606, "grad_norm": 1.7282812826406695, "learning_rate": 7.740815084887887e-09, "loss": 0.5946, "step": 27727 }, { "epoch": 0.9878689634287546, "grad_norm": 1.6561159682383058, "learning_rate": 7.695486795171735e-09, "loss": 0.5067, "step": 27728 }, { "epoch": 0.9879045905552488, "grad_norm": 1.556395497204198, "learning_rate": 7.65029155997521e-09, "loss": 0.4055, "step": 27729 }, { "epoch": 0.9879402176817429, "grad_norm": 1.9566445093222087, "learning_rate": 7.605229379898937e-09, "loss": 0.5742, "step": 27730 }, { "epoch": 0.987975844808237, "grad_norm": 1.2531698033040255, "learning_rate": 7.560300255543551e-09, "loss": 0.424, "step": 27731 }, { "epoch": 0.9880114719347312, "grad_norm": 1.570802730034654, "learning_rate": 7.51550418750746e-09, "loss": 0.4639, "step": 27732 }, { "epoch": 0.9880470990612252, "grad_norm": 1.2708845638895696, "learning_rate": 7.470841176385746e-09, "loss": 0.3536, "step": 27733 }, { "epoch": 0.9880827261877193, "grad_norm": 1.2990435799504407, "learning_rate": 7.426311222774596e-09, "loss": 0.3974, "step": 27734 }, { "epoch": 0.9881183533142135, "grad_norm": 1.621048893540038, "learning_rate": 7.38191432726687e-09, "loss": 0.4961, "step": 27735 }, { "epoch": 0.9881539804407076, "grad_norm": 1.7280854365924574, "learning_rate": 7.337650490453208e-09, "loss": 0.4358, "step": 27736 }, { "epoch": 0.9881896075672016, "grad_norm": 1.6797119684825896, "learning_rate": 7.293519712923136e-09, "loss": 0.4568, "step": 27737 }, { "epoch": 0.9882252346936958, "grad_norm": 1.3067439473055917, "learning_rate": 7.2495219952639636e-09, "loss": 0.3883, "step": 27738 }, { "epoch": 0.9882608618201899, "grad_norm": 1.7111191310245013, "learning_rate": 7.205657338062999e-09, "loss": 0.5657, "step": 27739 }, { "epoch": 0.988296488946684, "grad_norm": 1.2435903830346666, "learning_rate": 7.161925741901998e-09, "loss": 0.3638, "step": 27740 }, { "epoch": 0.9883321160731782, "grad_norm": 1.6340288151518827, "learning_rate": 7.118327207364939e-09, "loss": 0.4957, "step": 27741 }, { "epoch": 0.9883677431996722, "grad_norm": 1.5290417134612557, "learning_rate": 7.074861735032468e-09, "loss": 0.4946, "step": 27742 }, { "epoch": 0.9884033703261663, "grad_norm": 1.4035787400732092, "learning_rate": 7.0315293254819e-09, "loss": 0.3994, "step": 27743 }, { "epoch": 0.9884389974526605, "grad_norm": 1.8756168215924256, "learning_rate": 6.988329979291664e-09, "loss": 0.5419, "step": 27744 }, { "epoch": 0.9884746245791546, "grad_norm": 1.1993126665284453, "learning_rate": 6.945263697036853e-09, "loss": 0.439, "step": 27745 }, { "epoch": 0.9885102517056487, "grad_norm": 1.3684536862533927, "learning_rate": 6.902330479290342e-09, "loss": 0.3985, "step": 27746 }, { "epoch": 0.9885458788321428, "grad_norm": 1.3402628310957876, "learning_rate": 6.859530326623898e-09, "loss": 0.4366, "step": 27747 }, { "epoch": 0.9885815059586369, "grad_norm": 1.5204535088923885, "learning_rate": 6.816863239608173e-09, "loss": 0.5421, "step": 27748 }, { "epoch": 0.988617133085131, "grad_norm": 1.521984228430374, "learning_rate": 6.774329218810494e-09, "loss": 0.5016, "step": 27749 }, { "epoch": 0.9886527602116252, "grad_norm": 1.2400172765385507, "learning_rate": 6.7319282647970715e-09, "loss": 0.4087, "step": 27750 }, { "epoch": 0.9886883873381193, "grad_norm": 2.2015667597206257, "learning_rate": 6.6896603781330115e-09, "loss": 0.5429, "step": 27751 }, { "epoch": 0.9887240144646133, "grad_norm": 1.8202048335780847, "learning_rate": 6.647525559381196e-09, "loss": 0.4981, "step": 27752 }, { "epoch": 0.9887596415911075, "grad_norm": 1.4015733670717112, "learning_rate": 6.605523809102288e-09, "loss": 0.4109, "step": 27753 }, { "epoch": 0.9887952687176016, "grad_norm": 1.2610204064026456, "learning_rate": 6.563655127855839e-09, "loss": 0.3799, "step": 27754 }, { "epoch": 0.9888308958440957, "grad_norm": 1.2953832823453924, "learning_rate": 6.521919516199182e-09, "loss": 0.4135, "step": 27755 }, { "epoch": 0.9888665229705899, "grad_norm": 1.1691185220378075, "learning_rate": 6.480316974688539e-09, "loss": 0.3839, "step": 27756 }, { "epoch": 0.9889021500970839, "grad_norm": 1.3325157199314488, "learning_rate": 6.4388475038779095e-09, "loss": 0.3869, "step": 27757 }, { "epoch": 0.988937777223578, "grad_norm": 1.5415263113719804, "learning_rate": 6.3975111043179665e-09, "loss": 0.5763, "step": 27758 }, { "epoch": 0.9889734043500722, "grad_norm": 1.4200122869302925, "learning_rate": 6.356307776560489e-09, "loss": 0.3883, "step": 27759 }, { "epoch": 0.9890090314765663, "grad_norm": 1.3717910252854402, "learning_rate": 6.315237521155038e-09, "loss": 0.4109, "step": 27760 }, { "epoch": 0.9890446586030603, "grad_norm": 1.3305893041442909, "learning_rate": 6.274300338646733e-09, "loss": 0.3956, "step": 27761 }, { "epoch": 0.9890802857295545, "grad_norm": 1.3508567854792628, "learning_rate": 6.233496229580693e-09, "loss": 0.4067, "step": 27762 }, { "epoch": 0.9891159128560486, "grad_norm": 1.5472568003728295, "learning_rate": 6.192825194502039e-09, "loss": 0.4727, "step": 27763 }, { "epoch": 0.9891515399825427, "grad_norm": 1.4771674367560197, "learning_rate": 6.152287233950338e-09, "loss": 0.438, "step": 27764 }, { "epoch": 0.9891871671090369, "grad_norm": 3.9634867223633243, "learning_rate": 6.111882348466269e-09, "loss": 0.5556, "step": 27765 }, { "epoch": 0.9892227942355309, "grad_norm": 1.296065129774835, "learning_rate": 6.07161053858829e-09, "loss": 0.4717, "step": 27766 }, { "epoch": 0.989258421362025, "grad_norm": 1.5556590896664142, "learning_rate": 6.0314718048515295e-09, "loss": 0.5465, "step": 27767 }, { "epoch": 0.9892940484885192, "grad_norm": 1.4951519528648207, "learning_rate": 5.991466147791114e-09, "loss": 0.4775, "step": 27768 }, { "epoch": 0.9893296756150133, "grad_norm": 1.5469097537437593, "learning_rate": 5.951593567939951e-09, "loss": 0.5602, "step": 27769 }, { "epoch": 0.9893653027415074, "grad_norm": 1.423398167383194, "learning_rate": 5.911854065828726e-09, "loss": 0.4997, "step": 27770 }, { "epoch": 0.9894009298680015, "grad_norm": 1.5634430619932147, "learning_rate": 5.872247641987017e-09, "loss": 0.5457, "step": 27771 }, { "epoch": 0.9894365569944956, "grad_norm": 1.6913726938197189, "learning_rate": 5.832774296941068e-09, "loss": 0.5067, "step": 27772 }, { "epoch": 0.9894721841209897, "grad_norm": 1.659981033021868, "learning_rate": 5.793434031218237e-09, "loss": 0.5169, "step": 27773 }, { "epoch": 0.9895078112474839, "grad_norm": 2.066102442359218, "learning_rate": 5.754226845340327e-09, "loss": 0.5812, "step": 27774 }, { "epoch": 0.989543438373978, "grad_norm": 1.5122721773692478, "learning_rate": 5.7151527398302545e-09, "loss": 0.4928, "step": 27775 }, { "epoch": 0.989579065500472, "grad_norm": 1.4479573506192172, "learning_rate": 5.676211715209823e-09, "loss": 0.501, "step": 27776 }, { "epoch": 0.9896146926269662, "grad_norm": 1.6247559840495678, "learning_rate": 5.637403771995287e-09, "loss": 0.5536, "step": 27777 }, { "epoch": 0.9896503197534603, "grad_norm": 1.3878447872728468, "learning_rate": 5.59872891070512e-09, "loss": 0.4568, "step": 27778 }, { "epoch": 0.9896859468799544, "grad_norm": 1.3663459353515583, "learning_rate": 5.560187131852246e-09, "loss": 0.4092, "step": 27779 }, { "epoch": 0.9897215740064486, "grad_norm": 1.2313178652395114, "learning_rate": 5.521778435952918e-09, "loss": 0.4054, "step": 27780 }, { "epoch": 0.9897572011329426, "grad_norm": 1.4940683442367897, "learning_rate": 5.483502823515618e-09, "loss": 0.4324, "step": 27781 }, { "epoch": 0.9897928282594367, "grad_norm": 1.6642904892685793, "learning_rate": 5.4453602950510495e-09, "loss": 0.4693, "step": 27782 }, { "epoch": 0.9898284553859309, "grad_norm": 1.5526729781172, "learning_rate": 5.407350851067694e-09, "loss": 0.6255, "step": 27783 }, { "epoch": 0.989864082512425, "grad_norm": 1.4597539061071756, "learning_rate": 5.369474492071814e-09, "loss": 0.4956, "step": 27784 }, { "epoch": 0.989899709638919, "grad_norm": 1.5868499654221553, "learning_rate": 5.331731218566338e-09, "loss": 0.5181, "step": 27785 }, { "epoch": 0.9899353367654132, "grad_norm": 1.4092497986415722, "learning_rate": 5.2941210310553106e-09, "loss": 0.3968, "step": 27786 }, { "epoch": 0.9899709638919073, "grad_norm": 1.2879520381145275, "learning_rate": 5.2566439300383296e-09, "loss": 0.4101, "step": 27787 }, { "epoch": 0.9900065910184014, "grad_norm": 1.4884482738034732, "learning_rate": 5.2192999160161075e-09, "loss": 0.4756, "step": 27788 }, { "epoch": 0.9900422181448956, "grad_norm": 1.417461149265468, "learning_rate": 5.182088989483802e-09, "loss": 0.471, "step": 27789 }, { "epoch": 0.9900778452713896, "grad_norm": 1.199531067382333, "learning_rate": 5.145011150937685e-09, "loss": 0.3664, "step": 27790 }, { "epoch": 0.9901134723978837, "grad_norm": 1.5048338935253214, "learning_rate": 5.108066400872913e-09, "loss": 0.5285, "step": 27791 }, { "epoch": 0.9901490995243779, "grad_norm": 1.6095869917588939, "learning_rate": 5.071254739780207e-09, "loss": 0.4522, "step": 27792 }, { "epoch": 0.990184726650872, "grad_norm": 1.2301220028286917, "learning_rate": 5.034576168149175e-09, "loss": 0.3555, "step": 27793 }, { "epoch": 0.9902203537773661, "grad_norm": 1.475747327357051, "learning_rate": 4.998030686468314e-09, "loss": 0.5949, "step": 27794 }, { "epoch": 0.9902559809038602, "grad_norm": 1.3737299519785195, "learning_rate": 4.961618295225012e-09, "loss": 0.4458, "step": 27795 }, { "epoch": 0.9902916080303543, "grad_norm": 1.2904642611764063, "learning_rate": 4.925338994904438e-09, "loss": 0.4252, "step": 27796 }, { "epoch": 0.9903272351568484, "grad_norm": 1.6969949992307216, "learning_rate": 4.8891927859895384e-09, "loss": 0.5611, "step": 27797 }, { "epoch": 0.9903628622833426, "grad_norm": 1.6791257768401324, "learning_rate": 4.853179668959928e-09, "loss": 0.5061, "step": 27798 }, { "epoch": 0.9903984894098367, "grad_norm": 1.4127753193423274, "learning_rate": 4.8172996442974464e-09, "loss": 0.4628, "step": 27799 }, { "epoch": 0.9904341165363307, "grad_norm": 1.6386999343606379, "learning_rate": 4.781552712478377e-09, "loss": 0.5337, "step": 27800 }, { "epoch": 0.9904697436628249, "grad_norm": 2.1415734347309012, "learning_rate": 4.745938873979006e-09, "loss": 0.504, "step": 27801 }, { "epoch": 0.990505370789319, "grad_norm": 1.973232742772193, "learning_rate": 4.7104581292745085e-09, "loss": 0.4931, "step": 27802 }, { "epoch": 0.9905409979158131, "grad_norm": 1.2183770630476274, "learning_rate": 4.6751104788356205e-09, "loss": 0.3938, "step": 27803 }, { "epoch": 0.9905766250423073, "grad_norm": 1.273796069409651, "learning_rate": 4.639895923134186e-09, "loss": 0.3839, "step": 27804 }, { "epoch": 0.9906122521688013, "grad_norm": 1.282270118330145, "learning_rate": 4.6048144626387185e-09, "loss": 0.4173, "step": 27805 }, { "epoch": 0.9906478792952954, "grad_norm": 1.3698387659982691, "learning_rate": 4.569866097816622e-09, "loss": 0.4138, "step": 27806 }, { "epoch": 0.9906835064217896, "grad_norm": 1.6845114470403628, "learning_rate": 4.5350508291341914e-09, "loss": 0.4849, "step": 27807 }, { "epoch": 0.9907191335482837, "grad_norm": 1.2499934133667692, "learning_rate": 4.500368657053278e-09, "loss": 0.3756, "step": 27808 }, { "epoch": 0.9907547606747777, "grad_norm": 1.4915372559903186, "learning_rate": 4.465819582035735e-09, "loss": 0.4858, "step": 27809 }, { "epoch": 0.9907903878012719, "grad_norm": 2.1402487245008324, "learning_rate": 4.431403604542306e-09, "loss": 0.4921, "step": 27810 }, { "epoch": 0.990826014927766, "grad_norm": 1.2536686670908175, "learning_rate": 4.397120725032622e-09, "loss": 0.3885, "step": 27811 }, { "epoch": 0.9908616420542601, "grad_norm": 1.6941547156823595, "learning_rate": 4.362970943960765e-09, "loss": 0.5073, "step": 27812 }, { "epoch": 0.9908972691807543, "grad_norm": 1.3287634798408192, "learning_rate": 4.328954261781926e-09, "loss": 0.4019, "step": 27813 }, { "epoch": 0.9909328963072483, "grad_norm": 1.6088532771621817, "learning_rate": 4.2950706789501865e-09, "loss": 0.5079, "step": 27814 }, { "epoch": 0.9909685234337424, "grad_norm": 1.5128788502617911, "learning_rate": 4.2613201959162966e-09, "loss": 0.5268, "step": 27815 }, { "epoch": 0.9910041505602366, "grad_norm": 1.3434433910372339, "learning_rate": 4.227702813129897e-09, "loss": 0.4322, "step": 27816 }, { "epoch": 0.9910397776867307, "grad_norm": 1.4757734466178256, "learning_rate": 4.194218531037298e-09, "loss": 0.5219, "step": 27817 }, { "epoch": 0.9910754048132248, "grad_norm": 1.6252606225821031, "learning_rate": 4.1608673500859175e-09, "loss": 0.5526, "step": 27818 }, { "epoch": 0.991111031939719, "grad_norm": 1.3956438582564876, "learning_rate": 4.1276492707187366e-09, "loss": 0.4011, "step": 27819 }, { "epoch": 0.991146659066213, "grad_norm": 1.7564833650401208, "learning_rate": 4.094564293379843e-09, "loss": 0.5056, "step": 27820 }, { "epoch": 0.9911822861927071, "grad_norm": 1.5509248151720392, "learning_rate": 4.0616124185077765e-09, "loss": 0.4667, "step": 27821 }, { "epoch": 0.9912179133192013, "grad_norm": 1.7726782814521755, "learning_rate": 4.028793646542184e-09, "loss": 0.5191, "step": 27822 }, { "epoch": 0.9912535404456954, "grad_norm": 1.2158776306798258, "learning_rate": 3.996107977919383e-09, "loss": 0.3808, "step": 27823 }, { "epoch": 0.9912891675721894, "grad_norm": 1.4405475448262763, "learning_rate": 3.963555413075693e-09, "loss": 0.4493, "step": 27824 }, { "epoch": 0.9913247946986836, "grad_norm": 1.3087306992530574, "learning_rate": 3.931135952444099e-09, "loss": 0.4169, "step": 27825 }, { "epoch": 0.9913604218251777, "grad_norm": 1.5280599924959144, "learning_rate": 3.898849596456477e-09, "loss": 0.4994, "step": 27826 }, { "epoch": 0.9913960489516718, "grad_norm": 1.2477736861415805, "learning_rate": 3.866696345542487e-09, "loss": 0.3879, "step": 27827 }, { "epoch": 0.991431676078166, "grad_norm": 1.525451312193208, "learning_rate": 3.8346762001295616e-09, "loss": 0.4997, "step": 27828 }, { "epoch": 0.99146730320466, "grad_norm": 2.1163667803213824, "learning_rate": 3.802789160645137e-09, "loss": 0.5108, "step": 27829 }, { "epoch": 0.9915029303311541, "grad_norm": 1.7107887054726894, "learning_rate": 3.77103522751443e-09, "loss": 0.4543, "step": 27830 }, { "epoch": 0.9915385574576483, "grad_norm": 1.5602919696224307, "learning_rate": 3.739414401158214e-09, "loss": 0.4671, "step": 27831 }, { "epoch": 0.9915741845841424, "grad_norm": 1.3033021090951855, "learning_rate": 3.707926681999485e-09, "loss": 0.4879, "step": 27832 }, { "epoch": 0.9916098117106364, "grad_norm": 1.5786259649315597, "learning_rate": 3.676572070455686e-09, "loss": 0.4826, "step": 27833 }, { "epoch": 0.9916454388371306, "grad_norm": 1.3658000899916705, "learning_rate": 3.6453505669464817e-09, "loss": 0.4353, "step": 27834 }, { "epoch": 0.9916810659636247, "grad_norm": 1.2135969899634371, "learning_rate": 3.614262171885985e-09, "loss": 0.4125, "step": 27835 }, { "epoch": 0.9917166930901188, "grad_norm": 1.2365686016897293, "learning_rate": 3.583306885688309e-09, "loss": 0.4132, "step": 27836 }, { "epoch": 0.991752320216613, "grad_norm": 1.4755418346362603, "learning_rate": 3.552484708766457e-09, "loss": 0.5312, "step": 27837 }, { "epoch": 0.991787947343107, "grad_norm": 1.4375613556227491, "learning_rate": 3.5217956415301013e-09, "loss": 0.4329, "step": 27838 }, { "epoch": 0.9918235744696011, "grad_norm": 1.5375705972184779, "learning_rate": 3.491239684387804e-09, "loss": 0.5081, "step": 27839 }, { "epoch": 0.9918592015960953, "grad_norm": 1.2018953038315023, "learning_rate": 3.4608168377470165e-09, "loss": 0.4276, "step": 27840 }, { "epoch": 0.9918948287225894, "grad_norm": 1.3107222873458104, "learning_rate": 3.4305271020129705e-09, "loss": 0.4136, "step": 27841 }, { "epoch": 0.9919304558490835, "grad_norm": 1.2099187137500762, "learning_rate": 3.400370477588677e-09, "loss": 0.3998, "step": 27842 }, { "epoch": 0.9919660829755776, "grad_norm": 1.1625140717530693, "learning_rate": 3.3703469648760367e-09, "loss": 0.3927, "step": 27843 }, { "epoch": 0.9920017101020717, "grad_norm": 1.2962581620855371, "learning_rate": 3.3404565642736198e-09, "loss": 0.3963, "step": 27844 }, { "epoch": 0.9920373372285658, "grad_norm": 2.267271792211975, "learning_rate": 3.3106992761799962e-09, "loss": 0.4876, "step": 27845 }, { "epoch": 0.99207296435506, "grad_norm": 1.9555060940529472, "learning_rate": 3.2810751009926255e-09, "loss": 0.5142, "step": 27846 }, { "epoch": 0.9921085914815541, "grad_norm": 1.416522373437282, "learning_rate": 3.2515840391045272e-09, "loss": 0.4903, "step": 27847 }, { "epoch": 0.9921442186080481, "grad_norm": 1.6934352689311984, "learning_rate": 3.22222609090872e-09, "loss": 0.5609, "step": 27848 }, { "epoch": 0.9921798457345423, "grad_norm": 1.600892190081085, "learning_rate": 3.1930012567971123e-09, "loss": 0.5848, "step": 27849 }, { "epoch": 0.9922154728610364, "grad_norm": 1.2000433372006978, "learning_rate": 3.1639095371571727e-09, "loss": 0.4383, "step": 27850 }, { "epoch": 0.9922510999875305, "grad_norm": 1.2433367265109074, "learning_rate": 3.1349509323774784e-09, "loss": 0.385, "step": 27851 }, { "epoch": 0.9922867271140247, "grad_norm": 1.720088671503054, "learning_rate": 3.1061254428432773e-09, "loss": 0.5637, "step": 27852 }, { "epoch": 0.9923223542405187, "grad_norm": 1.4349250893764143, "learning_rate": 3.0774330689387066e-09, "loss": 0.4228, "step": 27853 }, { "epoch": 0.9923579813670128, "grad_norm": 1.4356390448800873, "learning_rate": 3.0488738110456827e-09, "loss": 0.4862, "step": 27854 }, { "epoch": 0.992393608493507, "grad_norm": 1.4185502258482134, "learning_rate": 3.020447669543902e-09, "loss": 0.4189, "step": 27855 }, { "epoch": 0.9924292356200011, "grad_norm": 3.8838622183696154, "learning_rate": 2.992154644811951e-09, "loss": 0.4816, "step": 27856 }, { "epoch": 0.9924648627464951, "grad_norm": 1.540902309247193, "learning_rate": 2.963994737227305e-09, "loss": 0.4724, "step": 27857 }, { "epoch": 0.9925004898729893, "grad_norm": 1.1564636530484287, "learning_rate": 2.9359679471652193e-09, "loss": 0.3566, "step": 27858 }, { "epoch": 0.9925361169994834, "grad_norm": 1.2931675017254434, "learning_rate": 2.908074274997619e-09, "loss": 0.4296, "step": 27859 }, { "epoch": 0.9925717441259775, "grad_norm": 1.2425719884136341, "learning_rate": 2.8803137210964283e-09, "loss": 0.4028, "step": 27860 }, { "epoch": 0.9926073712524717, "grad_norm": 1.7721849720446863, "learning_rate": 2.852686285831352e-09, "loss": 0.4701, "step": 27861 }, { "epoch": 0.9926429983789657, "grad_norm": 1.248949956914821, "learning_rate": 2.825191969570984e-09, "loss": 0.3752, "step": 27862 }, { "epoch": 0.9926786255054598, "grad_norm": 1.6448257617795812, "learning_rate": 2.7978307726805875e-09, "loss": 0.4983, "step": 27863 }, { "epoch": 0.992714252631954, "grad_norm": 1.6321924817332187, "learning_rate": 2.770602695524316e-09, "loss": 0.5519, "step": 27864 }, { "epoch": 0.9927498797584481, "grad_norm": 1.261628147006781, "learning_rate": 2.743507738465212e-09, "loss": 0.4312, "step": 27865 }, { "epoch": 0.9927855068849422, "grad_norm": 1.2445853470240453, "learning_rate": 2.7165459018640983e-09, "loss": 0.3711, "step": 27866 }, { "epoch": 0.9928211340114363, "grad_norm": 1.2336341405255185, "learning_rate": 2.689717186079577e-09, "loss": 0.4277, "step": 27867 }, { "epoch": 0.9928567611379304, "grad_norm": 2.313598902419576, "learning_rate": 2.6630215914702495e-09, "loss": 0.5454, "step": 27868 }, { "epoch": 0.9928923882644245, "grad_norm": 1.6122638567212801, "learning_rate": 2.6364591183891675e-09, "loss": 0.5465, "step": 27869 }, { "epoch": 0.9929280153909187, "grad_norm": 1.292204785599101, "learning_rate": 2.610029767191602e-09, "loss": 0.4217, "step": 27870 }, { "epoch": 0.9929636425174128, "grad_norm": 1.2305875170969085, "learning_rate": 2.5837335382294937e-09, "loss": 0.4303, "step": 27871 }, { "epoch": 0.9929992696439068, "grad_norm": 1.4161442914202886, "learning_rate": 2.5575704318536732e-09, "loss": 0.4466, "step": 27872 }, { "epoch": 0.993034896770401, "grad_norm": 1.6203659036145208, "learning_rate": 2.53154044841053e-09, "loss": 0.4599, "step": 27873 }, { "epoch": 0.9930705238968951, "grad_norm": 1.4579243039779273, "learning_rate": 2.505643588247564e-09, "loss": 0.4957, "step": 27874 }, { "epoch": 0.9931061510233892, "grad_norm": 1.377337009182933, "learning_rate": 2.4798798517100543e-09, "loss": 0.3841, "step": 27875 }, { "epoch": 0.9931417781498834, "grad_norm": 1.3037058771434376, "learning_rate": 2.45424923914106e-09, "loss": 0.3906, "step": 27876 }, { "epoch": 0.9931774052763774, "grad_norm": 1.2751993926476304, "learning_rate": 2.4287517508814197e-09, "loss": 0.3592, "step": 27877 }, { "epoch": 0.9932130324028715, "grad_norm": 1.399369646149711, "learning_rate": 2.403387387270861e-09, "loss": 0.415, "step": 27878 }, { "epoch": 0.9932486595293657, "grad_norm": 1.5681435890838784, "learning_rate": 2.378156148648003e-09, "loss": 0.4949, "step": 27879 }, { "epoch": 0.9932842866558598, "grad_norm": 1.5587008913876956, "learning_rate": 2.3530580353470223e-09, "loss": 0.4542, "step": 27880 }, { "epoch": 0.9933199137823538, "grad_norm": 1.6997320425122084, "learning_rate": 2.328093047703206e-09, "loss": 0.5431, "step": 27881 }, { "epoch": 0.993355540908848, "grad_norm": 1.271526813125098, "learning_rate": 2.303261186048511e-09, "loss": 0.3629, "step": 27882 }, { "epoch": 0.9933911680353421, "grad_norm": 1.2785474112854183, "learning_rate": 2.2785624507137837e-09, "loss": 0.355, "step": 27883 }, { "epoch": 0.9934267951618362, "grad_norm": 1.4185757784532884, "learning_rate": 2.2539968420276503e-09, "loss": 0.4296, "step": 27884 }, { "epoch": 0.9934624222883304, "grad_norm": 1.4492863146500334, "learning_rate": 2.2295643603187366e-09, "loss": 0.5158, "step": 27885 }, { "epoch": 0.9934980494148244, "grad_norm": 1.9275353309613985, "learning_rate": 2.2052650059101176e-09, "loss": 0.5242, "step": 27886 }, { "epoch": 0.9935336765413185, "grad_norm": 1.5656056048944287, "learning_rate": 2.181098779127089e-09, "loss": 0.4576, "step": 27887 }, { "epoch": 0.9935693036678127, "grad_norm": 1.1751774482242905, "learning_rate": 2.1570656802905042e-09, "loss": 0.3609, "step": 27888 }, { "epoch": 0.9936049307943068, "grad_norm": 1.4342518624953147, "learning_rate": 2.1331657097201087e-09, "loss": 0.5085, "step": 27889 }, { "epoch": 0.9936405579208009, "grad_norm": 1.8191636970907374, "learning_rate": 2.109398867735646e-09, "loss": 0.4441, "step": 27890 }, { "epoch": 0.993676185047295, "grad_norm": 1.3493742227148868, "learning_rate": 2.0857651546524195e-09, "loss": 0.3841, "step": 27891 }, { "epoch": 0.9937118121737891, "grad_norm": 1.4398748085837547, "learning_rate": 2.0622645707857324e-09, "loss": 0.4774, "step": 27892 }, { "epoch": 0.9937474393002832, "grad_norm": 1.1642175589808417, "learning_rate": 2.038897116447558e-09, "loss": 0.3853, "step": 27893 }, { "epoch": 0.9937830664267774, "grad_norm": 1.3768871766625281, "learning_rate": 2.015662791948758e-09, "loss": 0.4363, "step": 27894 }, { "epoch": 0.9938186935532715, "grad_norm": 1.2737411159167056, "learning_rate": 1.9925615976013058e-09, "loss": 0.3946, "step": 27895 }, { "epoch": 0.9938543206797655, "grad_norm": 1.4589219812532221, "learning_rate": 1.969593533709402e-09, "loss": 0.4213, "step": 27896 }, { "epoch": 0.9938899478062597, "grad_norm": 1.2732664316779845, "learning_rate": 1.9467586005816887e-09, "loss": 0.3949, "step": 27897 }, { "epoch": 0.9939255749327538, "grad_norm": 1.7811851017028322, "learning_rate": 1.9240567985212565e-09, "loss": 0.4637, "step": 27898 }, { "epoch": 0.9939612020592479, "grad_norm": 1.2834657051284206, "learning_rate": 1.901488127830087e-09, "loss": 0.4023, "step": 27899 }, { "epoch": 0.9939968291857421, "grad_norm": 1.7371660498011232, "learning_rate": 1.879052588807939e-09, "loss": 0.4348, "step": 27900 }, { "epoch": 0.9940324563122361, "grad_norm": 1.741626678822079, "learning_rate": 1.8567501817556842e-09, "loss": 0.5323, "step": 27901 }, { "epoch": 0.9940680834387302, "grad_norm": 1.3526866396380823, "learning_rate": 1.8345809069686416e-09, "loss": 0.3847, "step": 27902 }, { "epoch": 0.9941037105652244, "grad_norm": 1.8633026105446995, "learning_rate": 1.8125447647421302e-09, "loss": 0.5261, "step": 27903 }, { "epoch": 0.9941393376917185, "grad_norm": 1.2807374302769796, "learning_rate": 1.7906417553703592e-09, "loss": 0.477, "step": 27904 }, { "epoch": 0.9941749648182125, "grad_norm": 1.2300716192777907, "learning_rate": 1.7688718791442072e-09, "loss": 0.39, "step": 27905 }, { "epoch": 0.9942105919447067, "grad_norm": 1.9777228711908938, "learning_rate": 1.7472351363545525e-09, "loss": 0.5344, "step": 27906 }, { "epoch": 0.9942462190712008, "grad_norm": 1.6983260525233534, "learning_rate": 1.7257315272878327e-09, "loss": 0.4908, "step": 27907 }, { "epoch": 0.9942818461976949, "grad_norm": 1.7428383623032344, "learning_rate": 1.7043610522315956e-09, "loss": 0.4376, "step": 27908 }, { "epoch": 0.9943174733241891, "grad_norm": 1.3235349845259416, "learning_rate": 1.6831237114700582e-09, "loss": 0.4165, "step": 27909 }, { "epoch": 0.9943531004506831, "grad_norm": 1.6726693329071372, "learning_rate": 1.6620195052874378e-09, "loss": 0.4511, "step": 27910 }, { "epoch": 0.9943887275771772, "grad_norm": 1.3878867968118804, "learning_rate": 1.6410484339624e-09, "loss": 0.3759, "step": 27911 }, { "epoch": 0.9944243547036714, "grad_norm": 1.2552518466899878, "learning_rate": 1.6202104977747213e-09, "loss": 0.3898, "step": 27912 }, { "epoch": 0.9944599818301655, "grad_norm": 1.341646875451936, "learning_rate": 1.5995056970030676e-09, "loss": 0.3943, "step": 27913 }, { "epoch": 0.9944956089566596, "grad_norm": 1.3927124031045683, "learning_rate": 1.5789340319227741e-09, "loss": 0.484, "step": 27914 }, { "epoch": 0.9945312360831537, "grad_norm": 1.5256305196763404, "learning_rate": 1.5584955028069559e-09, "loss": 0.5403, "step": 27915 }, { "epoch": 0.9945668632096478, "grad_norm": 2.015281105116892, "learning_rate": 1.5381901099287278e-09, "loss": 0.573, "step": 27916 }, { "epoch": 0.9946024903361419, "grad_norm": 1.5416657799675928, "learning_rate": 1.5180178535567636e-09, "loss": 0.5078, "step": 27917 }, { "epoch": 0.9946381174626361, "grad_norm": 1.2605333568267338, "learning_rate": 1.497978733961958e-09, "loss": 0.3753, "step": 27918 }, { "epoch": 0.9946737445891302, "grad_norm": 1.1379458462566239, "learning_rate": 1.4780727514096538e-09, "loss": 0.3973, "step": 27919 }, { "epoch": 0.9947093717156242, "grad_norm": 2.271836852683545, "learning_rate": 1.458299906166305e-09, "loss": 0.3786, "step": 27920 }, { "epoch": 0.9947449988421184, "grad_norm": 1.5112292250024977, "learning_rate": 1.438660198493924e-09, "loss": 0.5198, "step": 27921 }, { "epoch": 0.9947806259686125, "grad_norm": 1.5684168719413467, "learning_rate": 1.4191536286534135e-09, "loss": 0.499, "step": 27922 }, { "epoch": 0.9948162530951066, "grad_norm": 1.3086820431245556, "learning_rate": 1.3997801969056757e-09, "loss": 0.4154, "step": 27923 }, { "epoch": 0.9948518802216008, "grad_norm": 1.6279901708509619, "learning_rate": 1.3805399035093926e-09, "loss": 0.5413, "step": 27924 }, { "epoch": 0.9948875073480948, "grad_norm": 1.278552019695892, "learning_rate": 1.361432748718805e-09, "loss": 0.3883, "step": 27925 }, { "epoch": 0.9949231344745889, "grad_norm": 1.5201896770565722, "learning_rate": 1.342458732790375e-09, "loss": 0.4444, "step": 27926 }, { "epoch": 0.9949587616010831, "grad_norm": 1.6100197533173153, "learning_rate": 1.3236178559750124e-09, "loss": 0.4699, "step": 27927 }, { "epoch": 0.9949943887275772, "grad_norm": 1.8108254474944045, "learning_rate": 1.3049101185247382e-09, "loss": 0.479, "step": 27928 }, { "epoch": 0.9950300158540712, "grad_norm": 1.2890649924108446, "learning_rate": 1.2863355206882422e-09, "loss": 0.422, "step": 27929 }, { "epoch": 0.9950656429805654, "grad_norm": 1.3436474458177268, "learning_rate": 1.2678940627131042e-09, "loss": 0.386, "step": 27930 }, { "epoch": 0.9951012701070595, "grad_norm": 1.7362896272088844, "learning_rate": 1.2495857448435733e-09, "loss": 0.5335, "step": 27931 }, { "epoch": 0.9951368972335536, "grad_norm": 1.3307750804297496, "learning_rate": 1.2314105673250088e-09, "loss": 0.4193, "step": 27932 }, { "epoch": 0.9951725243600478, "grad_norm": 1.3280007576647936, "learning_rate": 1.213368530399439e-09, "loss": 0.4028, "step": 27933 }, { "epoch": 0.9952081514865418, "grad_norm": 1.4644115797537691, "learning_rate": 1.1954596343066726e-09, "loss": 0.4978, "step": 27934 }, { "epoch": 0.9952437786130359, "grad_norm": 4.556659527972236, "learning_rate": 1.177683879284297e-09, "loss": 0.4747, "step": 27935 }, { "epoch": 0.9952794057395301, "grad_norm": 1.2579742848277586, "learning_rate": 1.1600412655710102e-09, "loss": 0.3886, "step": 27936 }, { "epoch": 0.9953150328660242, "grad_norm": 1.6239559818438238, "learning_rate": 1.1425317933988488e-09, "loss": 0.4786, "step": 27937 }, { "epoch": 0.9953506599925183, "grad_norm": 1.751583029733818, "learning_rate": 1.1251554630042905e-09, "loss": 0.54, "step": 27938 }, { "epoch": 0.9953862871190124, "grad_norm": 2.1606804629467087, "learning_rate": 1.1079122746160408e-09, "loss": 0.5318, "step": 27939 }, { "epoch": 0.9954219142455065, "grad_norm": 1.4949557197403454, "learning_rate": 1.0908022284650266e-09, "loss": 0.4071, "step": 27940 }, { "epoch": 0.9954575413720006, "grad_norm": 1.2838048570807292, "learning_rate": 1.0738253247788433e-09, "loss": 0.4232, "step": 27941 }, { "epoch": 0.9954931684984948, "grad_norm": 1.1903814993535224, "learning_rate": 1.0569815637839765e-09, "loss": 0.4214, "step": 27942 }, { "epoch": 0.9955287956249889, "grad_norm": 1.183118004606522, "learning_rate": 1.040270945703581e-09, "loss": 0.3656, "step": 27943 }, { "epoch": 0.9955644227514829, "grad_norm": 1.8598238893595962, "learning_rate": 1.0236934707608115e-09, "loss": 0.4957, "step": 27944 }, { "epoch": 0.9956000498779771, "grad_norm": 1.780761787453305, "learning_rate": 1.0072491391766026e-09, "loss": 0.5794, "step": 27945 }, { "epoch": 0.9956356770044712, "grad_norm": 1.294450063484168, "learning_rate": 9.909379511696682e-10, "loss": 0.3818, "step": 27946 }, { "epoch": 0.9956713041309653, "grad_norm": 1.2484793820974238, "learning_rate": 9.74759906957612e-10, "loss": 0.4181, "step": 27947 }, { "epoch": 0.9957069312574595, "grad_norm": 1.5818012225548836, "learning_rate": 9.587150067558172e-10, "loss": 0.5538, "step": 27948 }, { "epoch": 0.9957425583839535, "grad_norm": 1.3193818651507545, "learning_rate": 9.428032507774465e-10, "loss": 0.4953, "step": 27949 }, { "epoch": 0.9957781855104476, "grad_norm": 1.7814333893118057, "learning_rate": 9.270246392334426e-10, "loss": 0.551, "step": 27950 }, { "epoch": 0.9958138126369418, "grad_norm": 1.2968317807502092, "learning_rate": 9.113791723358578e-10, "loss": 0.3781, "step": 27951 }, { "epoch": 0.9958494397634359, "grad_norm": 1.3487986107548742, "learning_rate": 8.95866850292304e-10, "loss": 0.4145, "step": 27952 }, { "epoch": 0.99588506688993, "grad_norm": 1.594540194780518, "learning_rate": 8.804876733092826e-10, "loss": 0.5065, "step": 27953 }, { "epoch": 0.9959206940164241, "grad_norm": 1.4187298894397733, "learning_rate": 8.652416415921849e-10, "loss": 0.4526, "step": 27954 }, { "epoch": 0.9959563211429182, "grad_norm": 1.415537926025848, "learning_rate": 8.501287553419613e-10, "loss": 0.4377, "step": 27955 }, { "epoch": 0.9959919482694123, "grad_norm": 1.4066674554565863, "learning_rate": 8.351490147617824e-10, "loss": 0.4084, "step": 27956 }, { "epoch": 0.9960275753959065, "grad_norm": 1.4327053879262668, "learning_rate": 8.203024200514887e-10, "loss": 0.46, "step": 27957 }, { "epoch": 0.9960632025224005, "grad_norm": 1.6908683665735083, "learning_rate": 8.05588971406479e-10, "loss": 0.5236, "step": 27958 }, { "epoch": 0.9960988296488946, "grad_norm": 1.4119191950162815, "learning_rate": 7.910086690254836e-10, "loss": 0.3993, "step": 27959 }, { "epoch": 0.9961344567753888, "grad_norm": 1.2750716092626584, "learning_rate": 7.76561513100571e-10, "loss": 0.4205, "step": 27960 }, { "epoch": 0.9961700839018829, "grad_norm": 1.4760317210377139, "learning_rate": 7.622475038238097e-10, "loss": 0.4096, "step": 27961 }, { "epoch": 0.996205711028377, "grad_norm": 1.71557994209142, "learning_rate": 7.480666413883786e-10, "loss": 0.5073, "step": 27962 }, { "epoch": 0.9962413381548711, "grad_norm": 1.4765616016202479, "learning_rate": 7.340189259807951e-10, "loss": 0.3801, "step": 27963 }, { "epoch": 0.9962769652813652, "grad_norm": 1.380117238922094, "learning_rate": 7.201043577886868e-10, "loss": 0.4051, "step": 27964 }, { "epoch": 0.9963125924078593, "grad_norm": 1.983472589980022, "learning_rate": 7.063229369974611e-10, "loss": 0.4915, "step": 27965 }, { "epoch": 0.9963482195343535, "grad_norm": 1.3871818718049733, "learning_rate": 6.926746637903048e-10, "loss": 0.4207, "step": 27966 }, { "epoch": 0.9963838466608476, "grad_norm": 1.4240564869298056, "learning_rate": 6.791595383504046e-10, "loss": 0.5338, "step": 27967 }, { "epoch": 0.9964194737873416, "grad_norm": 1.2868125891782094, "learning_rate": 6.657775608553962e-10, "loss": 0.4137, "step": 27968 }, { "epoch": 0.9964551009138358, "grad_norm": 1.3250442833868872, "learning_rate": 6.525287314851358e-10, "loss": 0.4111, "step": 27969 }, { "epoch": 0.9964907280403299, "grad_norm": 1.33151281591695, "learning_rate": 6.394130504161489e-10, "loss": 0.406, "step": 27970 }, { "epoch": 0.996526355166824, "grad_norm": 1.6346890888042276, "learning_rate": 6.264305178216301e-10, "loss": 0.55, "step": 27971 }, { "epoch": 0.9965619822933182, "grad_norm": 1.1619415722533344, "learning_rate": 6.135811338758846e-10, "loss": 0.4276, "step": 27972 }, { "epoch": 0.9965976094198122, "grad_norm": 1.986357313377877, "learning_rate": 6.008648987487764e-10, "loss": 0.4938, "step": 27973 }, { "epoch": 0.9966332365463063, "grad_norm": 1.2131512704617646, "learning_rate": 5.882818126112799e-10, "loss": 0.3637, "step": 27974 }, { "epoch": 0.9966688636728005, "grad_norm": 1.2521127016083275, "learning_rate": 5.758318756299286e-10, "loss": 0.4438, "step": 27975 }, { "epoch": 0.9967044907992946, "grad_norm": 1.4469666630024502, "learning_rate": 5.635150879701456e-10, "loss": 0.5262, "step": 27976 }, { "epoch": 0.9967401179257886, "grad_norm": 1.7402315632245606, "learning_rate": 5.513314497962441e-10, "loss": 0.4664, "step": 27977 }, { "epoch": 0.9967757450522828, "grad_norm": 1.5596188337475663, "learning_rate": 5.392809612703165e-10, "loss": 0.5564, "step": 27978 }, { "epoch": 0.9968113721787769, "grad_norm": 1.7200378807685774, "learning_rate": 5.273636225533452e-10, "loss": 0.4823, "step": 27979 }, { "epoch": 0.996846999305271, "grad_norm": 1.2828921326956193, "learning_rate": 5.15579433804092e-10, "loss": 0.3741, "step": 27980 }, { "epoch": 0.9968826264317652, "grad_norm": 1.2535409371503294, "learning_rate": 5.039283951790985e-10, "loss": 0.4191, "step": 27981 }, { "epoch": 0.9969182535582592, "grad_norm": 1.2081793503737877, "learning_rate": 4.924105068326857e-10, "loss": 0.4002, "step": 27982 }, { "epoch": 0.9969538806847533, "grad_norm": 1.8797296690316052, "learning_rate": 4.810257689202847e-10, "loss": 0.4295, "step": 27983 }, { "epoch": 0.9969895078112475, "grad_norm": 1.2495057653803563, "learning_rate": 4.697741815906654e-10, "loss": 0.4099, "step": 27984 }, { "epoch": 0.9970251349377416, "grad_norm": 1.4008400509626728, "learning_rate": 4.5865574499703855e-10, "loss": 0.4257, "step": 27985 }, { "epoch": 0.9970607620642357, "grad_norm": 1.5560503435615127, "learning_rate": 4.476704592848435e-10, "loss": 0.4714, "step": 27986 }, { "epoch": 0.9970963891907298, "grad_norm": 1.864802894290197, "learning_rate": 4.368183246006297e-10, "loss": 0.5226, "step": 27987 }, { "epoch": 0.9971320163172239, "grad_norm": 1.185499722761597, "learning_rate": 4.260993410898362e-10, "loss": 0.3658, "step": 27988 }, { "epoch": 0.997167643443718, "grad_norm": 1.7227370249792655, "learning_rate": 4.1551350889568185e-10, "loss": 0.5065, "step": 27989 }, { "epoch": 0.9972032705702122, "grad_norm": 1.2740422416850181, "learning_rate": 4.050608281569446e-10, "loss": 0.3925, "step": 27990 }, { "epoch": 0.9972388976967063, "grad_norm": 1.7489457185307182, "learning_rate": 3.947412990157329e-10, "loss": 0.5556, "step": 27991 }, { "epoch": 0.9972745248232003, "grad_norm": 1.242033063355497, "learning_rate": 3.8455492160638374e-10, "loss": 0.3802, "step": 27992 }, { "epoch": 0.9973101519496945, "grad_norm": 1.404781137400649, "learning_rate": 3.745016960665648e-10, "loss": 0.4664, "step": 27993 }, { "epoch": 0.9973457790761886, "grad_norm": 1.109008575333184, "learning_rate": 3.6458162252950293e-10, "loss": 0.3493, "step": 27994 }, { "epoch": 0.9973814062026827, "grad_norm": 1.8412511703933843, "learning_rate": 3.547947011273145e-10, "loss": 0.505, "step": 27995 }, { "epoch": 0.9974170333291769, "grad_norm": 1.1331464360060157, "learning_rate": 3.4514093199100597e-10, "loss": 0.4052, "step": 27996 }, { "epoch": 0.9974526604556709, "grad_norm": 1.4113681859094798, "learning_rate": 3.356203152471427e-10, "loss": 0.4909, "step": 27997 }, { "epoch": 0.997488287582165, "grad_norm": 1.4773701204402114, "learning_rate": 3.2623285102451054e-10, "loss": 0.5257, "step": 27998 }, { "epoch": 0.9975239147086592, "grad_norm": 1.7460115794435058, "learning_rate": 3.169785394474545e-10, "loss": 0.6012, "step": 27999 }, { "epoch": 0.9975595418351533, "grad_norm": 1.483306877554848, "learning_rate": 3.078573806392093e-10, "loss": 0.4171, "step": 28000 }, { "epoch": 0.9975951689616473, "grad_norm": 1.3611505854120138, "learning_rate": 2.988693747218996e-10, "loss": 0.4099, "step": 28001 }, { "epoch": 0.9976307960881415, "grad_norm": 1.1834624735158923, "learning_rate": 2.900145218143191e-10, "loss": 0.4185, "step": 28002 }, { "epoch": 0.9976664232146356, "grad_norm": 1.2053243998134862, "learning_rate": 2.812928220330413e-10, "loss": 0.4191, "step": 28003 }, { "epoch": 0.9977020503411297, "grad_norm": 1.431087522467957, "learning_rate": 2.7270427549686006e-10, "loss": 0.6093, "step": 28004 }, { "epoch": 0.9977376774676239, "grad_norm": 1.3319813153880145, "learning_rate": 2.6424888231901813e-10, "loss": 0.4267, "step": 28005 }, { "epoch": 0.997773304594118, "grad_norm": 1.2359144793227408, "learning_rate": 2.5592664261275826e-10, "loss": 0.3793, "step": 28006 }, { "epoch": 0.997808931720612, "grad_norm": 1.2502968771751077, "learning_rate": 2.4773755648688226e-10, "loss": 0.4134, "step": 28007 }, { "epoch": 0.9978445588471062, "grad_norm": 1.2638462624756068, "learning_rate": 2.3968162405352267e-10, "loss": 0.3793, "step": 28008 }, { "epoch": 0.9978801859736003, "grad_norm": 1.1794597055237304, "learning_rate": 2.317588454170405e-10, "loss": 0.3864, "step": 28009 }, { "epoch": 0.9979158131000944, "grad_norm": 1.4019018352297088, "learning_rate": 2.2396922068401717e-10, "loss": 0.4854, "step": 28010 }, { "epoch": 0.9979514402265885, "grad_norm": 1.311710416432622, "learning_rate": 2.1631274995770336e-10, "loss": 0.3962, "step": 28011 }, { "epoch": 0.9979870673530826, "grad_norm": 1.7842278734080292, "learning_rate": 2.0878943334246005e-10, "loss": 0.5492, "step": 28012 }, { "epoch": 0.9980226944795767, "grad_norm": 1.6204101793704684, "learning_rate": 2.0139927093487666e-10, "loss": 0.4649, "step": 28013 }, { "epoch": 0.9980583216060709, "grad_norm": 1.5133653984891542, "learning_rate": 1.9414226283598348e-10, "loss": 0.4084, "step": 28014 }, { "epoch": 0.998093948732565, "grad_norm": 1.7322878357823441, "learning_rate": 1.8701840914125967e-10, "loss": 0.5527, "step": 28015 }, { "epoch": 0.998129575859059, "grad_norm": 1.7962481064188767, "learning_rate": 1.8002770994618446e-10, "loss": 0.4754, "step": 28016 }, { "epoch": 0.9981652029855532, "grad_norm": 1.5345618855869867, "learning_rate": 1.7317016534290632e-10, "loss": 0.4513, "step": 28017 }, { "epoch": 0.9982008301120473, "grad_norm": 1.5377709532043922, "learning_rate": 1.6644577542357376e-10, "loss": 0.4517, "step": 28018 }, { "epoch": 0.9982364572385414, "grad_norm": 1.4307104337527692, "learning_rate": 1.5985454027811488e-10, "loss": 0.3903, "step": 28019 }, { "epoch": 0.9982720843650356, "grad_norm": 1.3255585651738342, "learning_rate": 1.5339645999312701e-10, "loss": 0.5048, "step": 28020 }, { "epoch": 0.9983077114915296, "grad_norm": 1.335449890221222, "learning_rate": 1.470715346540974e-10, "loss": 0.414, "step": 28021 }, { "epoch": 0.9983433386180237, "grad_norm": 1.313184308575245, "learning_rate": 1.408797643476234e-10, "loss": 0.4239, "step": 28022 }, { "epoch": 0.9983789657445179, "grad_norm": 1.3416742065045515, "learning_rate": 1.3482114915475132e-10, "loss": 0.3599, "step": 28023 }, { "epoch": 0.998414592871012, "grad_norm": 1.2630893744666962, "learning_rate": 1.2889568915541716e-10, "loss": 0.4029, "step": 28024 }, { "epoch": 0.998450219997506, "grad_norm": 1.3167145047419186, "learning_rate": 1.2310338443066728e-10, "loss": 0.4344, "step": 28025 }, { "epoch": 0.9984858471240002, "grad_norm": 1.5882270717797258, "learning_rate": 1.1744423505488657e-10, "loss": 0.5113, "step": 28026 }, { "epoch": 0.9985214742504943, "grad_norm": 1.3116418632912625, "learning_rate": 1.1191824110579063e-10, "loss": 0.4268, "step": 28027 }, { "epoch": 0.9985571013769884, "grad_norm": 1.2192501741429878, "learning_rate": 1.0652540265554401e-10, "loss": 0.3769, "step": 28028 }, { "epoch": 0.9985927285034826, "grad_norm": 1.7660517621088205, "learning_rate": 1.0126571977631117e-10, "loss": 0.4894, "step": 28029 }, { "epoch": 0.9986283556299766, "grad_norm": 1.271835873321042, "learning_rate": 9.613919253914639e-11, "loss": 0.3829, "step": 28030 }, { "epoch": 0.9986639827564707, "grad_norm": 1.783718435656849, "learning_rate": 9.114582101066304e-11, "loss": 0.5086, "step": 28031 }, { "epoch": 0.9986996098829649, "grad_norm": 1.4391939356345413, "learning_rate": 8.628560525747454e-11, "loss": 0.4787, "step": 28032 }, { "epoch": 0.998735237009459, "grad_norm": 1.5821025901093237, "learning_rate": 8.155854534619424e-11, "loss": 0.5137, "step": 28033 }, { "epoch": 0.9987708641359531, "grad_norm": 1.3228928086905947, "learning_rate": 7.696464133788439e-11, "loss": 0.3594, "step": 28034 }, { "epoch": 0.9988064912624472, "grad_norm": 1.82068540180213, "learning_rate": 7.250389329471752e-11, "loss": 0.5797, "step": 28035 }, { "epoch": 0.9988421183889413, "grad_norm": 2.2029407777978127, "learning_rate": 6.81763012744252e-11, "loss": 0.4522, "step": 28036 }, { "epoch": 0.9988777455154354, "grad_norm": 1.7677095034653585, "learning_rate": 6.39818653369595e-11, "loss": 0.4974, "step": 28037 }, { "epoch": 0.9989133726419296, "grad_norm": 1.6384300717815201, "learning_rate": 5.99205855367213e-11, "loss": 0.5246, "step": 28038 }, { "epoch": 0.9989489997684237, "grad_norm": 1.2616600482889324, "learning_rate": 5.599246192922181e-11, "loss": 0.4164, "step": 28039 }, { "epoch": 0.9989846268949177, "grad_norm": 1.3797435334344976, "learning_rate": 5.219749456442102e-11, "loss": 0.4042, "step": 28040 }, { "epoch": 0.9990202540214119, "grad_norm": 1.5507225873064472, "learning_rate": 4.8535683494499443e-11, "loss": 0.3883, "step": 28041 }, { "epoch": 0.999055881147906, "grad_norm": 1.2927401761338577, "learning_rate": 4.500702876830687e-11, "loss": 0.4158, "step": 28042 }, { "epoch": 0.9990915082744001, "grad_norm": 1.5775631886200578, "learning_rate": 4.161153043136246e-11, "loss": 0.5316, "step": 28043 }, { "epoch": 0.9991271354008943, "grad_norm": 1.7345782724650662, "learning_rate": 3.834918853140579e-11, "loss": 0.5389, "step": 28044 }, { "epoch": 0.9991627625273883, "grad_norm": 1.647806687161914, "learning_rate": 3.5220003109515124e-11, "loss": 0.5468, "step": 28045 }, { "epoch": 0.9991983896538824, "grad_norm": 1.674852787824379, "learning_rate": 3.222397420787893e-11, "loss": 0.5864, "step": 28046 }, { "epoch": 0.9992340167803766, "grad_norm": 1.6331292885042858, "learning_rate": 2.936110186757546e-11, "loss": 0.4712, "step": 28047 }, { "epoch": 0.9992696439068707, "grad_norm": 1.76103947055953, "learning_rate": 2.6631386125242077e-11, "loss": 0.566, "step": 28048 }, { "epoch": 0.9993052710333648, "grad_norm": 1.3870515320113825, "learning_rate": 2.403482701862636e-11, "loss": 0.5405, "step": 28049 }, { "epoch": 0.9993408981598589, "grad_norm": 1.2910879386464376, "learning_rate": 2.1571424581035005e-11, "loss": 0.3959, "step": 28050 }, { "epoch": 0.999376525286353, "grad_norm": 1.6736827066554893, "learning_rate": 1.92411788457747e-11, "loss": 0.3981, "step": 28051 }, { "epoch": 0.9994121524128471, "grad_norm": 1.6987674612113108, "learning_rate": 1.7044089843931687e-11, "loss": 0.5671, "step": 28052 }, { "epoch": 0.9994477795393413, "grad_norm": 1.7216781408556054, "learning_rate": 1.498015760437177e-11, "loss": 0.4756, "step": 28053 }, { "epoch": 0.9994834066658353, "grad_norm": 1.173905472581905, "learning_rate": 1.3049382155960744e-11, "loss": 0.3997, "step": 28054 }, { "epoch": 0.9995190337923294, "grad_norm": 1.518315905672697, "learning_rate": 1.1251763522013292e-11, "loss": 0.498, "step": 28055 }, { "epoch": 0.9995546609188236, "grad_norm": 1.229648851474371, "learning_rate": 9.587301728064546e-12, "loss": 0.3801, "step": 28056 }, { "epoch": 0.9995902880453177, "grad_norm": 1.2123616604088836, "learning_rate": 8.055996797429189e-12, "loss": 0.4067, "step": 28057 }, { "epoch": 0.9996259151718118, "grad_norm": 1.6815283211975574, "learning_rate": 6.657848747870788e-12, "loss": 0.4574, "step": 28058 }, { "epoch": 0.999661542298306, "grad_norm": 1.471851350192751, "learning_rate": 5.392857599373358e-12, "loss": 0.5432, "step": 28059 }, { "epoch": 0.9996971694248, "grad_norm": 1.24495545861116, "learning_rate": 4.261023368590245e-12, "loss": 0.383, "step": 28060 }, { "epoch": 0.9997327965512941, "grad_norm": 1.9130613579524358, "learning_rate": 3.2623460710645706e-12, "loss": 0.522, "step": 28061 }, { "epoch": 0.9997684236777883, "grad_norm": 1.5816316429823842, "learning_rate": 2.3968257190087885e-12, "loss": 0.4073, "step": 28062 }, { "epoch": 0.9998040508042824, "grad_norm": 1.1809086577411472, "learning_rate": 1.664462325745575e-12, "loss": 0.3693, "step": 28063 }, { "epoch": 0.9998396779307764, "grad_norm": 1.660044502940141, "learning_rate": 1.0652558990464913e-12, "loss": 0.5148, "step": 28064 }, { "epoch": 0.9998753050572706, "grad_norm": 1.286479835343202, "learning_rate": 5.992064477933212e-13, "loss": 0.3972, "step": 28065 }, { "epoch": 0.9999109321837647, "grad_norm": 1.4657416835028119, "learning_rate": 2.6631397864740336e-13, "loss": 0.4448, "step": 28066 }, { "epoch": 0.9999465593102588, "grad_norm": 1.1060338485047434, "learning_rate": 6.65784949394066e-14, "loss": 0.365, "step": 28067 }, { "epoch": 0.999982186436753, "grad_norm": 1.2521274234197293, "learning_rate": 0.0, "loss": 0.408, "step": 28068 }, { "epoch": 0.999982186436753, "step": 28068, "total_flos": 5397019275853824.0, "train_loss": 0.02262667652967565, "train_runtime": 11575.5776, "train_samples_per_second": 310.374, "train_steps_per_second": 2.425 } ], "logging_steps": 1.0, "max_steps": 28068, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 5397019275853824.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }