{ "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": 31.79031295410919, "learning_rate": 2.3724792408066433e-08, "loss": 1.312, "step": 1 }, { "epoch": 7.125425298822524e-05, "grad_norm": 24.013125325955507, "learning_rate": 4.7449584816132866e-08, "loss": 1.3428, "step": 2 }, { "epoch": 0.00010688137948233786, "grad_norm": 19.356693071488483, "learning_rate": 7.117437722419929e-08, "loss": 1.418, "step": 3 }, { "epoch": 0.00014250850597645047, "grad_norm": 21.31670673682612, "learning_rate": 9.489916963226573e-08, "loss": 1.3743, "step": 4 }, { "epoch": 0.0001781356324705631, "grad_norm": 29.38508697569715, "learning_rate": 1.1862396204033216e-07, "loss": 1.3977, "step": 5 }, { "epoch": 0.0002137627589646757, "grad_norm": 28.52017973691994, "learning_rate": 1.4234875444839858e-07, "loss": 1.2204, "step": 6 }, { "epoch": 0.0002493898854587883, "grad_norm": 24.083718534617578, "learning_rate": 1.66073546856465e-07, "loss": 1.3951, "step": 7 }, { "epoch": 0.00028501701195290095, "grad_norm": 29.89983641208883, "learning_rate": 1.8979833926453146e-07, "loss": 1.2661, "step": 8 }, { "epoch": 0.00032064413844701354, "grad_norm": 22.14323118025869, "learning_rate": 2.135231316725979e-07, "loss": 1.3392, "step": 9 }, { "epoch": 0.0003562712649411262, "grad_norm": 28.252579932007922, "learning_rate": 2.3724792408066432e-07, "loss": 1.2546, "step": 10 }, { "epoch": 0.0003918983914352388, "grad_norm": 22.31152956272301, "learning_rate": 2.6097271648873075e-07, "loss": 1.3939, "step": 11 }, { "epoch": 0.0004275255179293514, "grad_norm": 27.940701972223035, "learning_rate": 2.8469750889679715e-07, "loss": 1.2663, "step": 12 }, { "epoch": 0.000463152644423464, "grad_norm": 22.629530875247767, "learning_rate": 3.084223013048636e-07, "loss": 1.3289, "step": 13 }, { "epoch": 0.0004987797709175766, "grad_norm": 28.695171223241495, "learning_rate": 3.3214709371293e-07, "loss": 1.2224, "step": 14 }, { "epoch": 0.0005344068974116893, "grad_norm": 24.700493586488506, "learning_rate": 3.558718861209965e-07, "loss": 1.1713, "step": 15 }, { "epoch": 0.0005700340239058019, "grad_norm": 24.41005004766158, "learning_rate": 3.795966785290629e-07, "loss": 1.1565, "step": 16 }, { "epoch": 0.0006056611503999145, "grad_norm": 26.48763920822277, "learning_rate": 4.0332147093712933e-07, "loss": 1.1041, "step": 17 }, { "epoch": 0.0006412882768940271, "grad_norm": 25.229814883568437, "learning_rate": 4.270462633451958e-07, "loss": 1.2756, "step": 18 }, { "epoch": 0.0006769154033881398, "grad_norm": 18.528323889623945, "learning_rate": 4.507710557532622e-07, "loss": 1.2538, "step": 19 }, { "epoch": 0.0007125425298822524, "grad_norm": 21.17072862537969, "learning_rate": 4.7449584816132864e-07, "loss": 1.0024, "step": 20 }, { "epoch": 0.000748169656376365, "grad_norm": 16.969636337093664, "learning_rate": 4.982206405693951e-07, "loss": 0.9339, "step": 21 }, { "epoch": 0.0007837967828704776, "grad_norm": 10.440098339577817, "learning_rate": 5.219454329774615e-07, "loss": 0.9365, "step": 22 }, { "epoch": 0.0008194239093645903, "grad_norm": 11.842714056731513, "learning_rate": 5.456702253855279e-07, "loss": 0.9323, "step": 23 }, { "epoch": 0.0008550510358587028, "grad_norm": 15.033939051677713, "learning_rate": 5.693950177935943e-07, "loss": 1.1405, "step": 24 }, { "epoch": 0.0008906781623528154, "grad_norm": 17.096592338643514, "learning_rate": 5.931198102016608e-07, "loss": 0.9918, "step": 25 }, { "epoch": 0.000926305288846928, "grad_norm": 13.54682668203079, "learning_rate": 6.168446026097272e-07, "loss": 1.0357, "step": 26 }, { "epoch": 0.0009619324153410406, "grad_norm": 10.502944228491938, "learning_rate": 6.405693950177936e-07, "loss": 0.9085, "step": 27 }, { "epoch": 0.0009975595418351532, "grad_norm": 9.790975547209923, "learning_rate": 6.6429418742586e-07, "loss": 0.9577, "step": 28 }, { "epoch": 0.0010331866683292658, "grad_norm": 8.167432698819532, "learning_rate": 6.880189798339265e-07, "loss": 0.8228, "step": 29 }, { "epoch": 0.0010688137948233786, "grad_norm": 8.702438721864524, "learning_rate": 7.11743772241993e-07, "loss": 0.9731, "step": 30 }, { "epoch": 0.0011044409213174912, "grad_norm": 5.8932747652684645, "learning_rate": 7.354685646500593e-07, "loss": 0.8131, "step": 31 }, { "epoch": 0.0011400680478116038, "grad_norm": 6.568350462643241, "learning_rate": 7.591933570581258e-07, "loss": 0.9361, "step": 32 }, { "epoch": 0.0011756951743057164, "grad_norm": 5.887045513532825, "learning_rate": 7.829181494661923e-07, "loss": 0.7707, "step": 33 }, { "epoch": 0.001211322300799829, "grad_norm": 6.195800809224235, "learning_rate": 8.066429418742587e-07, "loss": 0.9737, "step": 34 }, { "epoch": 0.0012469494272939416, "grad_norm": 7.081356578613777, "learning_rate": 8.303677342823251e-07, "loss": 0.9442, "step": 35 }, { "epoch": 0.0012825765537880542, "grad_norm": 7.08160306551178, "learning_rate": 8.540925266903916e-07, "loss": 0.891, "step": 36 }, { "epoch": 0.0013182036802821667, "grad_norm": 5.329645978432587, "learning_rate": 8.778173190984579e-07, "loss": 0.9215, "step": 37 }, { "epoch": 0.0013538308067762796, "grad_norm": 4.937156079237347, "learning_rate": 9.015421115065244e-07, "loss": 0.8782, "step": 38 }, { "epoch": 0.0013894579332703921, "grad_norm": 4.5429124920932535, "learning_rate": 9.252669039145908e-07, "loss": 0.71, "step": 39 }, { "epoch": 0.0014250850597645047, "grad_norm": 5.021894580517158, "learning_rate": 9.489916963226573e-07, "loss": 0.7382, "step": 40 }, { "epoch": 0.0014607121862586173, "grad_norm": 5.696697077008865, "learning_rate": 9.727164887307238e-07, "loss": 0.8934, "step": 41 }, { "epoch": 0.00149633931275273, "grad_norm": 4.053723529306746, "learning_rate": 9.964412811387902e-07, "loss": 0.7937, "step": 42 }, { "epoch": 0.0015319664392468425, "grad_norm": 4.595577838065693, "learning_rate": 1.0201660735468566e-06, "loss": 0.9037, "step": 43 }, { "epoch": 0.001567593565740955, "grad_norm": 4.766618235004359, "learning_rate": 1.043890865954923e-06, "loss": 0.9468, "step": 44 }, { "epoch": 0.0016032206922350677, "grad_norm": 4.815854265462876, "learning_rate": 1.0676156583629894e-06, "loss": 0.9375, "step": 45 }, { "epoch": 0.0016388478187291805, "grad_norm": 4.026024075245735, "learning_rate": 1.0913404507710558e-06, "loss": 0.8492, "step": 46 }, { "epoch": 0.001674474945223293, "grad_norm": 4.694708782804313, "learning_rate": 1.1150652431791222e-06, "loss": 0.9638, "step": 47 }, { "epoch": 0.0017101020717174057, "grad_norm": 4.207286572601061, "learning_rate": 1.1387900355871886e-06, "loss": 0.7572, "step": 48 }, { "epoch": 0.0017457291982115183, "grad_norm": 3.826533984698009, "learning_rate": 1.1625148279952552e-06, "loss": 0.7587, "step": 49 }, { "epoch": 0.0017813563247056309, "grad_norm": 3.6044894710245345, "learning_rate": 1.1862396204033216e-06, "loss": 0.7307, "step": 50 }, { "epoch": 0.0018169834511997435, "grad_norm": 3.5181234479895713, "learning_rate": 1.209964412811388e-06, "loss": 0.7991, "step": 51 }, { "epoch": 0.001852610577693856, "grad_norm": 3.7856439195534777, "learning_rate": 1.2336892052194544e-06, "loss": 0.722, "step": 52 }, { "epoch": 0.0018882377041879686, "grad_norm": 3.641959469549171, "learning_rate": 1.257413997627521e-06, "loss": 0.7883, "step": 53 }, { "epoch": 0.0019238648306820812, "grad_norm": 3.1278367555725235, "learning_rate": 1.2811387900355872e-06, "loss": 0.7339, "step": 54 }, { "epoch": 0.001959491957176194, "grad_norm": 3.33598192689969, "learning_rate": 1.3048635824436536e-06, "loss": 0.7046, "step": 55 }, { "epoch": 0.0019951190836703064, "grad_norm": 3.4270296020001165, "learning_rate": 1.32858837485172e-06, "loss": 0.7372, "step": 56 }, { "epoch": 0.0020307462101644192, "grad_norm": 3.5923360542525327, "learning_rate": 1.3523131672597867e-06, "loss": 0.7792, "step": 57 }, { "epoch": 0.0020663733366585316, "grad_norm": 4.192288123934018, "learning_rate": 1.376037959667853e-06, "loss": 0.8963, "step": 58 }, { "epoch": 0.0021020004631526444, "grad_norm": 3.1267798606528374, "learning_rate": 1.3997627520759195e-06, "loss": 0.7527, "step": 59 }, { "epoch": 0.0021376275896467572, "grad_norm": 3.4427054444465788, "learning_rate": 1.423487544483986e-06, "loss": 0.8817, "step": 60 }, { "epoch": 0.0021732547161408696, "grad_norm": 3.949107539790104, "learning_rate": 1.4472123368920523e-06, "loss": 0.8662, "step": 61 }, { "epoch": 0.0022088818426349824, "grad_norm": 5.102702521931529, "learning_rate": 1.4709371293001187e-06, "loss": 0.7902, "step": 62 }, { "epoch": 0.0022445089691290948, "grad_norm": 2.7952452591272157, "learning_rate": 1.494661921708185e-06, "loss": 0.6726, "step": 63 }, { "epoch": 0.0022801360956232076, "grad_norm": 3.4065678529786947, "learning_rate": 1.5183867141162517e-06, "loss": 1.0039, "step": 64 }, { "epoch": 0.00231576322211732, "grad_norm": 3.1801254863215163, "learning_rate": 1.542111506524318e-06, "loss": 0.7786, "step": 65 }, { "epoch": 0.0023513903486114328, "grad_norm": 3.3176094545784407, "learning_rate": 1.5658362989323845e-06, "loss": 0.8292, "step": 66 }, { "epoch": 0.0023870174751055456, "grad_norm": 3.6050629359323514, "learning_rate": 1.5895610913404507e-06, "loss": 0.8165, "step": 67 }, { "epoch": 0.002422644601599658, "grad_norm": 3.3168825505536006, "learning_rate": 1.6132858837485173e-06, "loss": 0.8595, "step": 68 }, { "epoch": 0.0024582717280937708, "grad_norm": 2.8183607713340852, "learning_rate": 1.6370106761565837e-06, "loss": 0.7321, "step": 69 }, { "epoch": 0.002493898854587883, "grad_norm": 2.8698376898050553, "learning_rate": 1.6607354685646501e-06, "loss": 0.7025, "step": 70 }, { "epoch": 0.002529525981081996, "grad_norm": 3.359980692463656, "learning_rate": 1.6844602609727167e-06, "loss": 0.7553, "step": 71 }, { "epoch": 0.0025651531075761083, "grad_norm": 3.6131767902688483, "learning_rate": 1.7081850533807831e-06, "loss": 0.7946, "step": 72 }, { "epoch": 0.002600780234070221, "grad_norm": 2.7626390562421355, "learning_rate": 1.7319098457888495e-06, "loss": 0.6305, "step": 73 }, { "epoch": 0.0026364073605643335, "grad_norm": 3.670736089580687, "learning_rate": 1.7556346381969157e-06, "loss": 0.7694, "step": 74 }, { "epoch": 0.0026720344870584463, "grad_norm": 3.059522847194626, "learning_rate": 1.7793594306049826e-06, "loss": 0.6797, "step": 75 }, { "epoch": 0.002707661613552559, "grad_norm": 3.820517479318242, "learning_rate": 1.8030842230130487e-06, "loss": 0.6975, "step": 76 }, { "epoch": 0.0027432887400466715, "grad_norm": 2.800147565614958, "learning_rate": 1.8268090154211151e-06, "loss": 0.6365, "step": 77 }, { "epoch": 0.0027789158665407843, "grad_norm": 3.7806722694836354, "learning_rate": 1.8505338078291815e-06, "loss": 0.6908, "step": 78 }, { "epoch": 0.0028145429930348967, "grad_norm": 3.2194569257751553, "learning_rate": 1.8742586002372482e-06, "loss": 0.7937, "step": 79 }, { "epoch": 0.0028501701195290095, "grad_norm": 3.153910621549058, "learning_rate": 1.8979833926453146e-06, "loss": 0.8272, "step": 80 }, { "epoch": 0.002885797246023122, "grad_norm": 3.658011281927994, "learning_rate": 1.921708185053381e-06, "loss": 0.8705, "step": 81 }, { "epoch": 0.0029214243725172347, "grad_norm": 2.6898599470716347, "learning_rate": 1.9454329774614476e-06, "loss": 0.6587, "step": 82 }, { "epoch": 0.002957051499011347, "grad_norm": 2.3549196515063864, "learning_rate": 1.9691577698695138e-06, "loss": 0.6936, "step": 83 }, { "epoch": 0.00299267862550546, "grad_norm": 3.430260450451571, "learning_rate": 1.9928825622775804e-06, "loss": 0.8291, "step": 84 }, { "epoch": 0.0030283057519995727, "grad_norm": 2.9270441327364347, "learning_rate": 2.0166073546856466e-06, "loss": 0.6729, "step": 85 }, { "epoch": 0.003063932878493685, "grad_norm": 2.445284825565081, "learning_rate": 2.040332147093713e-06, "loss": 0.6634, "step": 86 }, { "epoch": 0.003099560004987798, "grad_norm": 3.0624978121248834, "learning_rate": 2.0640569395017794e-06, "loss": 0.7331, "step": 87 }, { "epoch": 0.00313518713148191, "grad_norm": 2.397963378193327, "learning_rate": 2.087781731909846e-06, "loss": 0.6389, "step": 88 }, { "epoch": 0.003170814257976023, "grad_norm": 2.5591338357989253, "learning_rate": 2.111506524317912e-06, "loss": 0.7175, "step": 89 }, { "epoch": 0.0032064413844701354, "grad_norm": 2.4030319277342707, "learning_rate": 2.135231316725979e-06, "loss": 0.7009, "step": 90 }, { "epoch": 0.003242068510964248, "grad_norm": 3.4720454579908613, "learning_rate": 2.1589561091340454e-06, "loss": 0.8054, "step": 91 }, { "epoch": 0.003277695637458361, "grad_norm": 2.5196998852250267, "learning_rate": 2.1826809015421116e-06, "loss": 0.6581, "step": 92 }, { "epoch": 0.0033133227639524734, "grad_norm": 2.2555234318531503, "learning_rate": 2.2064056939501782e-06, "loss": 0.6287, "step": 93 }, { "epoch": 0.003348949890446586, "grad_norm": 3.340669457374696, "learning_rate": 2.2301304863582444e-06, "loss": 0.7369, "step": 94 }, { "epoch": 0.0033845770169406986, "grad_norm": 3.776820062986242, "learning_rate": 2.253855278766311e-06, "loss": 0.8218, "step": 95 }, { "epoch": 0.0034202041434348114, "grad_norm": 3.3721804764229466, "learning_rate": 2.2775800711743772e-06, "loss": 0.8184, "step": 96 }, { "epoch": 0.0034558312699289238, "grad_norm": 2.2375905842089323, "learning_rate": 2.301304863582444e-06, "loss": 0.6651, "step": 97 }, { "epoch": 0.0034914583964230366, "grad_norm": 3.16341065313947, "learning_rate": 2.3250296559905105e-06, "loss": 0.6345, "step": 98 }, { "epoch": 0.003527085522917149, "grad_norm": 2.2614009737018277, "learning_rate": 2.3487544483985766e-06, "loss": 0.6492, "step": 99 }, { "epoch": 0.0035627126494112617, "grad_norm": 2.5347615824910283, "learning_rate": 2.3724792408066433e-06, "loss": 0.7502, "step": 100 }, { "epoch": 0.0035983397759053746, "grad_norm": 2.2311007391355018, "learning_rate": 2.3962040332147095e-06, "loss": 0.6519, "step": 101 }, { "epoch": 0.003633966902399487, "grad_norm": 2.3098194236190706, "learning_rate": 2.419928825622776e-06, "loss": 0.6649, "step": 102 }, { "epoch": 0.0036695940288935997, "grad_norm": 2.456805640459855, "learning_rate": 2.4436536180308423e-06, "loss": 0.6899, "step": 103 }, { "epoch": 0.003705221155387712, "grad_norm": 3.5338907777892894, "learning_rate": 2.467378410438909e-06, "loss": 0.8625, "step": 104 }, { "epoch": 0.003740848281881825, "grad_norm": 2.326785867422014, "learning_rate": 2.4911032028469755e-06, "loss": 0.6667, "step": 105 }, { "epoch": 0.0037764754083759373, "grad_norm": 3.1142867673827968, "learning_rate": 2.514827995255042e-06, "loss": 0.7845, "step": 106 }, { "epoch": 0.00381210253487005, "grad_norm": 2.2826054875966624, "learning_rate": 2.5385527876631083e-06, "loss": 0.6766, "step": 107 }, { "epoch": 0.0038477296613641625, "grad_norm": 3.2843301293888074, "learning_rate": 2.5622775800711745e-06, "loss": 0.6944, "step": 108 }, { "epoch": 0.0038833567878582753, "grad_norm": 3.535159932268127, "learning_rate": 2.586002372479241e-06, "loss": 0.8235, "step": 109 }, { "epoch": 0.003918983914352388, "grad_norm": 3.2859298204124676, "learning_rate": 2.6097271648873073e-06, "loss": 0.8237, "step": 110 }, { "epoch": 0.003954611040846501, "grad_norm": 3.1315496174196102, "learning_rate": 2.633451957295374e-06, "loss": 0.6756, "step": 111 }, { "epoch": 0.003990238167340613, "grad_norm": 2.216300346326018, "learning_rate": 2.65717674970344e-06, "loss": 0.6346, "step": 112 }, { "epoch": 0.004025865293834726, "grad_norm": 2.4508103591887975, "learning_rate": 2.680901542111507e-06, "loss": 0.715, "step": 113 }, { "epoch": 0.0040614924203288385, "grad_norm": 3.125545877812168, "learning_rate": 2.7046263345195733e-06, "loss": 0.7834, "step": 114 }, { "epoch": 0.004097119546822951, "grad_norm": 2.347474741728873, "learning_rate": 2.7283511269276395e-06, "loss": 0.6737, "step": 115 }, { "epoch": 0.004132746673317063, "grad_norm": 1.9748830842045495, "learning_rate": 2.752075919335706e-06, "loss": 0.5638, "step": 116 }, { "epoch": 0.004168373799811176, "grad_norm": 3.926725155716171, "learning_rate": 2.7758007117437723e-06, "loss": 0.8223, "step": 117 }, { "epoch": 0.004204000926305289, "grad_norm": 3.7108021745350777, "learning_rate": 2.799525504151839e-06, "loss": 0.7796, "step": 118 }, { "epoch": 0.004239628052799402, "grad_norm": 3.0103693851726403, "learning_rate": 2.823250296559905e-06, "loss": 0.7866, "step": 119 }, { "epoch": 0.0042752551792935144, "grad_norm": 4.203169916360009, "learning_rate": 2.846975088967972e-06, "loss": 0.7799, "step": 120 }, { "epoch": 0.004310882305787626, "grad_norm": 2.1013590507697124, "learning_rate": 2.8706998813760384e-06, "loss": 0.6112, "step": 121 }, { "epoch": 0.004346509432281739, "grad_norm": 2.264593551525973, "learning_rate": 2.8944246737841046e-06, "loss": 0.6715, "step": 122 }, { "epoch": 0.004382136558775852, "grad_norm": 5.080188769702261, "learning_rate": 2.918149466192171e-06, "loss": 0.7564, "step": 123 }, { "epoch": 0.004417763685269965, "grad_norm": 4.0455618721697215, "learning_rate": 2.9418742586002374e-06, "loss": 0.7755, "step": 124 }, { "epoch": 0.004453390811764077, "grad_norm": 2.331296292561826, "learning_rate": 2.965599051008304e-06, "loss": 0.6333, "step": 125 }, { "epoch": 0.0044890179382581896, "grad_norm": 3.3842874228072617, "learning_rate": 2.98932384341637e-06, "loss": 0.736, "step": 126 }, { "epoch": 0.004524645064752302, "grad_norm": 2.2690627346127705, "learning_rate": 3.0130486358244364e-06, "loss": 0.6422, "step": 127 }, { "epoch": 0.004560272191246415, "grad_norm": 4.868216325600808, "learning_rate": 3.0367734282325034e-06, "loss": 0.7462, "step": 128 }, { "epoch": 0.004595899317740528, "grad_norm": 2.2584573689436183, "learning_rate": 3.0604982206405696e-06, "loss": 0.5959, "step": 129 }, { "epoch": 0.00463152644423464, "grad_norm": 2.514103961316917, "learning_rate": 3.084223013048636e-06, "loss": 0.6461, "step": 130 }, { "epoch": 0.004667153570728753, "grad_norm": 3.201187446970141, "learning_rate": 3.1079478054567024e-06, "loss": 0.794, "step": 131 }, { "epoch": 0.0047027806972228655, "grad_norm": 3.008371824100112, "learning_rate": 3.131672597864769e-06, "loss": 0.7864, "step": 132 }, { "epoch": 0.004738407823716978, "grad_norm": 2.218094394480356, "learning_rate": 3.155397390272835e-06, "loss": 0.6273, "step": 133 }, { "epoch": 0.004774034950211091, "grad_norm": 2.2461530515508046, "learning_rate": 3.1791221826809014e-06, "loss": 0.5988, "step": 134 }, { "epoch": 0.004809662076705203, "grad_norm": 3.2511186603839057, "learning_rate": 3.2028469750889684e-06, "loss": 0.7971, "step": 135 }, { "epoch": 0.004845289203199316, "grad_norm": 2.4152560695640033, "learning_rate": 3.2265717674970346e-06, "loss": 0.7043, "step": 136 }, { "epoch": 0.004880916329693429, "grad_norm": 3.854924377239634, "learning_rate": 3.2502965599051012e-06, "loss": 0.8384, "step": 137 }, { "epoch": 0.0049165434561875415, "grad_norm": 2.6029088907714466, "learning_rate": 3.2740213523131674e-06, "loss": 0.6815, "step": 138 }, { "epoch": 0.0049521705826816535, "grad_norm": 3.416100353262435, "learning_rate": 3.297746144721234e-06, "loss": 0.8349, "step": 139 }, { "epoch": 0.004987797709175766, "grad_norm": 4.007964580506819, "learning_rate": 3.3214709371293002e-06, "loss": 0.6671, "step": 140 }, { "epoch": 0.005023424835669879, "grad_norm": 2.0192875028954056, "learning_rate": 3.3451957295373664e-06, "loss": 0.6623, "step": 141 }, { "epoch": 0.005059051962163992, "grad_norm": 3.4203511923273013, "learning_rate": 3.3689205219454335e-06, "loss": 0.7091, "step": 142 }, { "epoch": 0.005094679088658105, "grad_norm": 2.3456688029743864, "learning_rate": 3.3926453143535e-06, "loss": 0.6221, "step": 143 }, { "epoch": 0.005130306215152217, "grad_norm": 4.002967486871383, "learning_rate": 3.4163701067615663e-06, "loss": 0.7747, "step": 144 }, { "epoch": 0.0051659333416463294, "grad_norm": 3.5181275320756753, "learning_rate": 3.4400948991696325e-06, "loss": 0.697, "step": 145 }, { "epoch": 0.005201560468140442, "grad_norm": 3.4101895739978088, "learning_rate": 3.463819691577699e-06, "loss": 0.7669, "step": 146 }, { "epoch": 0.005237187594634555, "grad_norm": 2.1854139977723457, "learning_rate": 3.4875444839857653e-06, "loss": 0.6404, "step": 147 }, { "epoch": 0.005272814721128667, "grad_norm": 3.6815658393213826, "learning_rate": 3.5112692763938315e-06, "loss": 0.7411, "step": 148 }, { "epoch": 0.00530844184762278, "grad_norm": 3.1361062294622113, "learning_rate": 3.5349940688018985e-06, "loss": 0.7466, "step": 149 }, { "epoch": 0.005344068974116893, "grad_norm": 2.5600738074423557, "learning_rate": 3.558718861209965e-06, "loss": 0.5953, "step": 150 }, { "epoch": 0.005379696100611005, "grad_norm": 3.611403018617861, "learning_rate": 3.5824436536180313e-06, "loss": 0.7884, "step": 151 }, { "epoch": 0.005415323227105118, "grad_norm": 3.5166127834661536, "learning_rate": 3.6061684460260975e-06, "loss": 0.7339, "step": 152 }, { "epoch": 0.00545095035359923, "grad_norm": 2.2594696640567817, "learning_rate": 3.629893238434164e-06, "loss": 0.6302, "step": 153 }, { "epoch": 0.005486577480093343, "grad_norm": 3.2948581968966684, "learning_rate": 3.6536180308422303e-06, "loss": 0.7591, "step": 154 }, { "epoch": 0.005522204606587456, "grad_norm": 3.080289306843851, "learning_rate": 3.6773428232502965e-06, "loss": 0.6733, "step": 155 }, { "epoch": 0.005557831733081569, "grad_norm": 2.152447554544054, "learning_rate": 3.701067615658363e-06, "loss": 0.5954, "step": 156 }, { "epoch": 0.0055934588595756805, "grad_norm": 3.2321040769363556, "learning_rate": 3.72479240806643e-06, "loss": 0.8563, "step": 157 }, { "epoch": 0.005629085986069793, "grad_norm": 2.3066542342790837, "learning_rate": 3.7485172004744963e-06, "loss": 0.6653, "step": 158 }, { "epoch": 0.005664713112563906, "grad_norm": 4.261400533307933, "learning_rate": 3.7722419928825625e-06, "loss": 0.6353, "step": 159 }, { "epoch": 0.005700340239058019, "grad_norm": 2.9539791834434377, "learning_rate": 3.795966785290629e-06, "loss": 0.6915, "step": 160 }, { "epoch": 0.005735967365552132, "grad_norm": 2.0467389782165504, "learning_rate": 3.819691577698696e-06, "loss": 0.6424, "step": 161 }, { "epoch": 0.005771594492046244, "grad_norm": 2.3124550373170094, "learning_rate": 3.843416370106762e-06, "loss": 0.6138, "step": 162 }, { "epoch": 0.0058072216185403565, "grad_norm": 3.7653450305565723, "learning_rate": 3.867141162514828e-06, "loss": 0.7397, "step": 163 }, { "epoch": 0.005842848745034469, "grad_norm": 3.033419979018472, "learning_rate": 3.890865954922895e-06, "loss": 0.7566, "step": 164 }, { "epoch": 0.005878475871528582, "grad_norm": 5.027494617084082, "learning_rate": 3.914590747330961e-06, "loss": 0.7642, "step": 165 }, { "epoch": 0.005914102998022694, "grad_norm": 2.4498591035757826, "learning_rate": 3.9383155397390276e-06, "loss": 0.6386, "step": 166 }, { "epoch": 0.005949730124516807, "grad_norm": 2.95876561892247, "learning_rate": 3.962040332147094e-06, "loss": 0.674, "step": 167 }, { "epoch": 0.00598535725101092, "grad_norm": 3.7326408322290647, "learning_rate": 3.985765124555161e-06, "loss": 0.662, "step": 168 }, { "epoch": 0.0060209843775050325, "grad_norm": 2.8198926309143895, "learning_rate": 4.009489916963227e-06, "loss": 0.5879, "step": 169 }, { "epoch": 0.006056611503999145, "grad_norm": 2.205308478520408, "learning_rate": 4.033214709371293e-06, "loss": 0.6383, "step": 170 }, { "epoch": 0.006092238630493257, "grad_norm": 2.1659440367704037, "learning_rate": 4.05693950177936e-06, "loss": 0.6336, "step": 171 }, { "epoch": 0.00612786575698737, "grad_norm": 4.504736784382706, "learning_rate": 4.080664294187426e-06, "loss": 0.6817, "step": 172 }, { "epoch": 0.006163492883481483, "grad_norm": 3.7134798323033587, "learning_rate": 4.104389086595493e-06, "loss": 0.7887, "step": 173 }, { "epoch": 0.006199120009975596, "grad_norm": 3.78791441152315, "learning_rate": 4.128113879003559e-06, "loss": 0.6071, "step": 174 }, { "epoch": 0.006234747136469708, "grad_norm": 2.459279478984198, "learning_rate": 4.151838671411626e-06, "loss": 0.6053, "step": 175 }, { "epoch": 0.00627037426296382, "grad_norm": 2.2283490546326954, "learning_rate": 4.175563463819692e-06, "loss": 0.6266, "step": 176 }, { "epoch": 0.006306001389457933, "grad_norm": 2.978605008532578, "learning_rate": 4.199288256227758e-06, "loss": 0.7008, "step": 177 }, { "epoch": 0.006341628515952046, "grad_norm": 2.3031495828971633, "learning_rate": 4.223013048635824e-06, "loss": 0.6216, "step": 178 }, { "epoch": 0.006377255642446159, "grad_norm": 2.584190128862883, "learning_rate": 4.2467378410438914e-06, "loss": 0.6179, "step": 179 }, { "epoch": 0.006412882768940271, "grad_norm": 3.20826749606522, "learning_rate": 4.270462633451958e-06, "loss": 0.7484, "step": 180 }, { "epoch": 0.006448509895434384, "grad_norm": 3.3825551174313775, "learning_rate": 4.294187425860024e-06, "loss": 0.7189, "step": 181 }, { "epoch": 0.006484137021928496, "grad_norm": 2.4254187440342285, "learning_rate": 4.317912218268091e-06, "loss": 0.6587, "step": 182 }, { "epoch": 0.006519764148422609, "grad_norm": 3.5262607680412037, "learning_rate": 4.341637010676157e-06, "loss": 0.6838, "step": 183 }, { "epoch": 0.006555391274916722, "grad_norm": 3.9353348544734863, "learning_rate": 4.365361803084223e-06, "loss": 0.67, "step": 184 }, { "epoch": 0.006591018401410834, "grad_norm": 2.3590503616367617, "learning_rate": 4.389086595492289e-06, "loss": 0.6707, "step": 185 }, { "epoch": 0.006626645527904947, "grad_norm": 3.3294009372786815, "learning_rate": 4.4128113879003565e-06, "loss": 0.763, "step": 186 }, { "epoch": 0.00666227265439906, "grad_norm": 2.276499492791831, "learning_rate": 4.436536180308423e-06, "loss": 0.6117, "step": 187 }, { "epoch": 0.006697899780893172, "grad_norm": 2.26219625352971, "learning_rate": 4.460260972716489e-06, "loss": 0.6334, "step": 188 }, { "epoch": 0.006733526907387284, "grad_norm": 2.261881316883183, "learning_rate": 4.483985765124556e-06, "loss": 0.6356, "step": 189 }, { "epoch": 0.006769154033881397, "grad_norm": 3.9383929043396715, "learning_rate": 4.507710557532622e-06, "loss": 0.7768, "step": 190 }, { "epoch": 0.00680478116037551, "grad_norm": 4.089183373549742, "learning_rate": 4.531435349940688e-06, "loss": 0.8049, "step": 191 }, { "epoch": 0.006840408286869623, "grad_norm": 2.8301202290038754, "learning_rate": 4.5551601423487545e-06, "loss": 0.6515, "step": 192 }, { "epoch": 0.0068760354133637356, "grad_norm": 2.2261480568953482, "learning_rate": 4.5788849347568215e-06, "loss": 0.632, "step": 193 }, { "epoch": 0.0069116625398578475, "grad_norm": 4.206288798903984, "learning_rate": 4.602609727164888e-06, "loss": 0.6751, "step": 194 }, { "epoch": 0.00694728966635196, "grad_norm": 2.4711588502407844, "learning_rate": 4.626334519572954e-06, "loss": 0.5888, "step": 195 }, { "epoch": 0.006982916792846073, "grad_norm": 2.8155721824756776, "learning_rate": 4.650059311981021e-06, "loss": 0.7401, "step": 196 }, { "epoch": 0.007018543919340186, "grad_norm": 2.2505156980000423, "learning_rate": 4.673784104389087e-06, "loss": 0.5893, "step": 197 }, { "epoch": 0.007054171045834298, "grad_norm": 2.7911960151031368, "learning_rate": 4.697508896797153e-06, "loss": 0.6257, "step": 198 }, { "epoch": 0.007089798172328411, "grad_norm": 2.3244377653189594, "learning_rate": 4.7212336892052195e-06, "loss": 0.6533, "step": 199 }, { "epoch": 0.0071254252988225235, "grad_norm": 3.8419062589286157, "learning_rate": 4.7449584816132865e-06, "loss": 0.7503, "step": 200 }, { "epoch": 0.007161052425316636, "grad_norm": 2.501523861652408, "learning_rate": 4.768683274021353e-06, "loss": 0.5864, "step": 201 }, { "epoch": 0.007196679551810749, "grad_norm": 3.184020525876615, "learning_rate": 4.792408066429419e-06, "loss": 0.7028, "step": 202 }, { "epoch": 0.007232306678304861, "grad_norm": 2.316954333615399, "learning_rate": 4.816132858837486e-06, "loss": 0.6361, "step": 203 }, { "epoch": 0.007267933804798974, "grad_norm": 3.1579625579720054, "learning_rate": 4.839857651245552e-06, "loss": 0.702, "step": 204 }, { "epoch": 0.007303560931293087, "grad_norm": 3.390066269049058, "learning_rate": 4.863582443653618e-06, "loss": 0.7344, "step": 205 }, { "epoch": 0.0073391880577871995, "grad_norm": 2.216347091091069, "learning_rate": 4.8873072360616845e-06, "loss": 0.5839, "step": 206 }, { "epoch": 0.007374815184281311, "grad_norm": 3.9791493504312925, "learning_rate": 4.911032028469751e-06, "loss": 0.6257, "step": 207 }, { "epoch": 0.007410442310775424, "grad_norm": 2.7435665402347227, "learning_rate": 4.934756820877818e-06, "loss": 0.668, "step": 208 }, { "epoch": 0.007446069437269537, "grad_norm": 3.6624522904153003, "learning_rate": 4.958481613285884e-06, "loss": 0.734, "step": 209 }, { "epoch": 0.00748169656376365, "grad_norm": 3.1163894729333106, "learning_rate": 4.982206405693951e-06, "loss": 0.6828, "step": 210 }, { "epoch": 0.007517323690257763, "grad_norm": 2.137758070190628, "learning_rate": 5.005931198102017e-06, "loss": 0.601, "step": 211 }, { "epoch": 0.007552950816751875, "grad_norm": 3.8287565790157494, "learning_rate": 5.029655990510084e-06, "loss": 0.7558, "step": 212 }, { "epoch": 0.007588577943245987, "grad_norm": 3.311909789832912, "learning_rate": 5.05338078291815e-06, "loss": 0.7208, "step": 213 }, { "epoch": 0.0076242050697401, "grad_norm": 2.319875582687343, "learning_rate": 5.077105575326217e-06, "loss": 0.6012, "step": 214 }, { "epoch": 0.007659832196234213, "grad_norm": 3.958229120039053, "learning_rate": 5.100830367734283e-06, "loss": 0.7093, "step": 215 }, { "epoch": 0.007695459322728325, "grad_norm": 2.552897455303845, "learning_rate": 5.124555160142349e-06, "loss": 0.7184, "step": 216 }, { "epoch": 0.007731086449222438, "grad_norm": 3.4151709838060054, "learning_rate": 5.148279952550416e-06, "loss": 0.7525, "step": 217 }, { "epoch": 0.0077667135757165506, "grad_norm": 2.3087949172337026, "learning_rate": 5.172004744958482e-06, "loss": 0.6043, "step": 218 }, { "epoch": 0.007802340702210663, "grad_norm": 5.153041505021954, "learning_rate": 5.195729537366548e-06, "loss": 0.8225, "step": 219 }, { "epoch": 0.007837967828704776, "grad_norm": 2.479736354127771, "learning_rate": 5.219454329774615e-06, "loss": 0.5792, "step": 220 }, { "epoch": 0.007873594955198888, "grad_norm": 3.99682007646162, "learning_rate": 5.243179122182681e-06, "loss": 0.7146, "step": 221 }, { "epoch": 0.007909222081693002, "grad_norm": 2.9192071248296654, "learning_rate": 5.266903914590748e-06, "loss": 0.7294, "step": 222 }, { "epoch": 0.007944849208187114, "grad_norm": 2.5324128782479023, "learning_rate": 5.290628706998814e-06, "loss": 0.594, "step": 223 }, { "epoch": 0.007980476334681226, "grad_norm": 3.5374524138668293, "learning_rate": 5.31435349940688e-06, "loss": 0.6721, "step": 224 }, { "epoch": 0.00801610346117534, "grad_norm": 3.933808810421925, "learning_rate": 5.338078291814946e-06, "loss": 0.6956, "step": 225 }, { "epoch": 0.008051730587669451, "grad_norm": 2.134693748477244, "learning_rate": 5.361803084223014e-06, "loss": 0.6264, "step": 226 }, { "epoch": 0.008087357714163565, "grad_norm": 3.0039045656349095, "learning_rate": 5.3855278766310805e-06, "loss": 0.5585, "step": 227 }, { "epoch": 0.008122984840657677, "grad_norm": 7.3241848447043045, "learning_rate": 5.409252669039147e-06, "loss": 0.7372, "step": 228 }, { "epoch": 0.008158611967151789, "grad_norm": 2.0728792353829553, "learning_rate": 5.432977461447213e-06, "loss": 0.6561, "step": 229 }, { "epoch": 0.008194239093645903, "grad_norm": 3.2871007025406973, "learning_rate": 5.456702253855279e-06, "loss": 0.6299, "step": 230 }, { "epoch": 0.008229866220140014, "grad_norm": 2.2099077045293614, "learning_rate": 5.480427046263346e-06, "loss": 0.6075, "step": 231 }, { "epoch": 0.008265493346634126, "grad_norm": 2.128070811933608, "learning_rate": 5.504151838671412e-06, "loss": 0.589, "step": 232 }, { "epoch": 0.00830112047312824, "grad_norm": 2.141130262932371, "learning_rate": 5.5278766310794785e-06, "loss": 0.5698, "step": 233 }, { "epoch": 0.008336747599622352, "grad_norm": 4.3459514708754075, "learning_rate": 5.551601423487545e-06, "loss": 0.6182, "step": 234 }, { "epoch": 0.008372374726116466, "grad_norm": 2.291456457186016, "learning_rate": 5.575326215895611e-06, "loss": 0.5503, "step": 235 }, { "epoch": 0.008408001852610578, "grad_norm": 2.231225536848269, "learning_rate": 5.599051008303678e-06, "loss": 0.5978, "step": 236 }, { "epoch": 0.00844362897910469, "grad_norm": 4.009650871216636, "learning_rate": 5.622775800711744e-06, "loss": 0.6281, "step": 237 }, { "epoch": 0.008479256105598803, "grad_norm": 8.460100308978818, "learning_rate": 5.64650059311981e-06, "loss": 0.7512, "step": 238 }, { "epoch": 0.008514883232092915, "grad_norm": 3.525640484638202, "learning_rate": 5.6702253855278765e-06, "loss": 0.721, "step": 239 }, { "epoch": 0.008550510358587029, "grad_norm": 3.6152131146547495, "learning_rate": 5.693950177935944e-06, "loss": 0.6798, "step": 240 }, { "epoch": 0.00858613748508114, "grad_norm": 3.9379055026917795, "learning_rate": 5.7176749703440105e-06, "loss": 0.6694, "step": 241 }, { "epoch": 0.008621764611575253, "grad_norm": 3.7136054772775435, "learning_rate": 5.741399762752077e-06, "loss": 0.7789, "step": 242 }, { "epoch": 0.008657391738069366, "grad_norm": 2.591731809687898, "learning_rate": 5.765124555160143e-06, "loss": 0.5394, "step": 243 }, { "epoch": 0.008693018864563478, "grad_norm": 7.7670829159618995, "learning_rate": 5.788849347568209e-06, "loss": 0.7011, "step": 244 }, { "epoch": 0.008728645991057592, "grad_norm": 2.77959326619853, "learning_rate": 5.812574139976276e-06, "loss": 0.6163, "step": 245 }, { "epoch": 0.008764273117551704, "grad_norm": 4.390614683188382, "learning_rate": 5.836298932384342e-06, "loss": 0.7234, "step": 246 }, { "epoch": 0.008799900244045816, "grad_norm": 3.089017651606292, "learning_rate": 5.8600237247924085e-06, "loss": 0.772, "step": 247 }, { "epoch": 0.00883552737053993, "grad_norm": 3.2594266554617075, "learning_rate": 5.883748517200475e-06, "loss": 0.7145, "step": 248 }, { "epoch": 0.008871154497034042, "grad_norm": 4.174088439338432, "learning_rate": 5.907473309608541e-06, "loss": 0.645, "step": 249 }, { "epoch": 0.008906781623528153, "grad_norm": 2.2232798318590237, "learning_rate": 5.931198102016608e-06, "loss": 0.6118, "step": 250 }, { "epoch": 0.008942408750022267, "grad_norm": 2.2231593502611733, "learning_rate": 5.954922894424674e-06, "loss": 0.6281, "step": 251 }, { "epoch": 0.008978035876516379, "grad_norm": 3.0677728404785714, "learning_rate": 5.97864768683274e-06, "loss": 0.675, "step": 252 }, { "epoch": 0.009013663003010493, "grad_norm": 2.8108504408489186, "learning_rate": 6.0023724792408065e-06, "loss": 0.5443, "step": 253 }, { "epoch": 0.009049290129504605, "grad_norm": 2.4315731694210214, "learning_rate": 6.026097271648873e-06, "loss": 0.5562, "step": 254 }, { "epoch": 0.009084917255998717, "grad_norm": 8.145127270673852, "learning_rate": 6.049822064056941e-06, "loss": 0.7025, "step": 255 }, { "epoch": 0.00912054438249283, "grad_norm": 2.5198390020346206, "learning_rate": 6.073546856465007e-06, "loss": 0.581, "step": 256 }, { "epoch": 0.009156171508986942, "grad_norm": 2.4086306228774896, "learning_rate": 6.097271648873073e-06, "loss": 0.6198, "step": 257 }, { "epoch": 0.009191798635481056, "grad_norm": 2.494905401473409, "learning_rate": 6.120996441281139e-06, "loss": 0.5714, "step": 258 }, { "epoch": 0.009227425761975168, "grad_norm": 2.5183604037142135, "learning_rate": 6.144721233689206e-06, "loss": 0.6285, "step": 259 }, { "epoch": 0.00926305288846928, "grad_norm": 2.2569531967522316, "learning_rate": 6.168446026097272e-06, "loss": 0.6116, "step": 260 }, { "epoch": 0.009298680014963394, "grad_norm": 2.133693409788748, "learning_rate": 6.192170818505339e-06, "loss": 0.6081, "step": 261 }, { "epoch": 0.009334307141457505, "grad_norm": 2.4615605437650534, "learning_rate": 6.215895610913405e-06, "loss": 0.5719, "step": 262 }, { "epoch": 0.009369934267951619, "grad_norm": 4.909133933974998, "learning_rate": 6.239620403321471e-06, "loss": 0.6399, "step": 263 }, { "epoch": 0.009405561394445731, "grad_norm": 2.386881497976247, "learning_rate": 6.263345195729538e-06, "loss": 0.5749, "step": 264 }, { "epoch": 0.009441188520939843, "grad_norm": 3.1142888816885534, "learning_rate": 6.287069988137604e-06, "loss": 0.6213, "step": 265 }, { "epoch": 0.009476815647433957, "grad_norm": 4.0815500410798675, "learning_rate": 6.31079478054567e-06, "loss": 0.6296, "step": 266 }, { "epoch": 0.009512442773928069, "grad_norm": 2.666010130484388, "learning_rate": 6.334519572953737e-06, "loss": 0.628, "step": 267 }, { "epoch": 0.009548069900422182, "grad_norm": 3.3572400474017696, "learning_rate": 6.358244365361803e-06, "loss": 0.7088, "step": 268 }, { "epoch": 0.009583697026916294, "grad_norm": 2.4447310508674605, "learning_rate": 6.381969157769871e-06, "loss": 0.599, "step": 269 }, { "epoch": 0.009619324153410406, "grad_norm": 2.4695833103989235, "learning_rate": 6.405693950177937e-06, "loss": 0.6035, "step": 270 }, { "epoch": 0.00965495127990452, "grad_norm": 2.4132659160428234, "learning_rate": 6.429418742586003e-06, "loss": 0.5713, "step": 271 }, { "epoch": 0.009690578406398632, "grad_norm": 3.6790749730022694, "learning_rate": 6.453143534994069e-06, "loss": 0.7317, "step": 272 }, { "epoch": 0.009726205532892744, "grad_norm": 3.6643455920144143, "learning_rate": 6.476868327402136e-06, "loss": 0.7093, "step": 273 }, { "epoch": 0.009761832659386857, "grad_norm": 2.4488576984710515, "learning_rate": 6.5005931198102025e-06, "loss": 0.5986, "step": 274 }, { "epoch": 0.00979745978588097, "grad_norm": 2.516618734408368, "learning_rate": 6.524317912218269e-06, "loss": 0.555, "step": 275 }, { "epoch": 0.009833086912375083, "grad_norm": 3.3756079121182756, "learning_rate": 6.548042704626335e-06, "loss": 0.6637, "step": 276 }, { "epoch": 0.009868714038869195, "grad_norm": 2.419730188384377, "learning_rate": 6.571767497034401e-06, "loss": 0.6042, "step": 277 }, { "epoch": 0.009904341165363307, "grad_norm": 3.308177030053978, "learning_rate": 6.595492289442468e-06, "loss": 0.7576, "step": 278 }, { "epoch": 0.00993996829185742, "grad_norm": 2.6127040514408617, "learning_rate": 6.619217081850534e-06, "loss": 0.5754, "step": 279 }, { "epoch": 0.009975595418351533, "grad_norm": 2.178392892671189, "learning_rate": 6.6429418742586005e-06, "loss": 0.6049, "step": 280 }, { "epoch": 0.010011222544845646, "grad_norm": 2.2010681363652096, "learning_rate": 6.666666666666667e-06, "loss": 0.5647, "step": 281 }, { "epoch": 0.010046849671339758, "grad_norm": 4.134121105403518, "learning_rate": 6.690391459074733e-06, "loss": 0.6654, "step": 282 }, { "epoch": 0.01008247679783387, "grad_norm": 2.900317370454384, "learning_rate": 6.7141162514828e-06, "loss": 0.7368, "step": 283 }, { "epoch": 0.010118103924327984, "grad_norm": 2.580147486086389, "learning_rate": 6.737841043890867e-06, "loss": 0.5675, "step": 284 }, { "epoch": 0.010153731050822096, "grad_norm": 3.945201070968827, "learning_rate": 6.761565836298933e-06, "loss": 0.6856, "step": 285 }, { "epoch": 0.01018935817731621, "grad_norm": 2.499493622386912, "learning_rate": 6.785290628707e-06, "loss": 0.5511, "step": 286 }, { "epoch": 0.010224985303810321, "grad_norm": 3.1195609121129255, "learning_rate": 6.809015421115066e-06, "loss": 0.7526, "step": 287 }, { "epoch": 0.010260612430304433, "grad_norm": 4.26228755903888, "learning_rate": 6.8327402135231325e-06, "loss": 0.6502, "step": 288 }, { "epoch": 0.010296239556798547, "grad_norm": 2.3206131973439286, "learning_rate": 6.856465005931199e-06, "loss": 0.5659, "step": 289 }, { "epoch": 0.010331866683292659, "grad_norm": 2.8171095390728462, "learning_rate": 6.880189798339265e-06, "loss": 0.6702, "step": 290 }, { "epoch": 0.01036749380978677, "grad_norm": 2.1570864499107794, "learning_rate": 6.903914590747331e-06, "loss": 0.514, "step": 291 }, { "epoch": 0.010403120936280884, "grad_norm": 3.5053616980944104, "learning_rate": 6.927639383155398e-06, "loss": 0.6102, "step": 292 }, { "epoch": 0.010438748062774996, "grad_norm": 2.550635323705399, "learning_rate": 6.951364175563464e-06, "loss": 0.6368, "step": 293 }, { "epoch": 0.01047437518926911, "grad_norm": 3.3973163702985527, "learning_rate": 6.9750889679715305e-06, "loss": 0.7717, "step": 294 }, { "epoch": 0.010510002315763222, "grad_norm": 2.8400791450569867, "learning_rate": 6.998813760379597e-06, "loss": 0.6847, "step": 295 }, { "epoch": 0.010545629442257334, "grad_norm": 2.189665433675379, "learning_rate": 7.022538552787663e-06, "loss": 0.6501, "step": 296 }, { "epoch": 0.010581256568751448, "grad_norm": 2.0086273123612703, "learning_rate": 7.04626334519573e-06, "loss": 0.6375, "step": 297 }, { "epoch": 0.01061688369524556, "grad_norm": 2.3444253914786057, "learning_rate": 7.069988137603797e-06, "loss": 0.6212, "step": 298 }, { "epoch": 0.010652510821739673, "grad_norm": 5.180590685271567, "learning_rate": 7.093712930011863e-06, "loss": 0.6827, "step": 299 }, { "epoch": 0.010688137948233785, "grad_norm": 2.528500968074749, "learning_rate": 7.11743772241993e-06, "loss": 0.5928, "step": 300 }, { "epoch": 0.010723765074727897, "grad_norm": 2.7519847868000276, "learning_rate": 7.141162514827996e-06, "loss": 0.673, "step": 301 }, { "epoch": 0.01075939220122201, "grad_norm": 5.982917429602854, "learning_rate": 7.164887307236063e-06, "loss": 0.7955, "step": 302 }, { "epoch": 0.010795019327716123, "grad_norm": 3.899465947114399, "learning_rate": 7.188612099644129e-06, "loss": 0.6276, "step": 303 }, { "epoch": 0.010830646454210236, "grad_norm": 3.353735530377907, "learning_rate": 7.212336892052195e-06, "loss": 0.6442, "step": 304 }, { "epoch": 0.010866273580704348, "grad_norm": 9.865934823304675, "learning_rate": 7.236061684460261e-06, "loss": 0.6906, "step": 305 }, { "epoch": 0.01090190070719846, "grad_norm": 3.0926291341921193, "learning_rate": 7.259786476868328e-06, "loss": 0.6043, "step": 306 }, { "epoch": 0.010937527833692574, "grad_norm": 3.437280502990903, "learning_rate": 7.283511269276394e-06, "loss": 0.6064, "step": 307 }, { "epoch": 0.010973154960186686, "grad_norm": 4.9634359444960845, "learning_rate": 7.307236061684461e-06, "loss": 0.7412, "step": 308 }, { "epoch": 0.011008782086680798, "grad_norm": 3.2736224820268993, "learning_rate": 7.330960854092527e-06, "loss": 0.6788, "step": 309 }, { "epoch": 0.011044409213174912, "grad_norm": 2.849984518516134, "learning_rate": 7.354685646500593e-06, "loss": 0.6211, "step": 310 }, { "epoch": 0.011080036339669024, "grad_norm": 2.8203428821413086, "learning_rate": 7.37841043890866e-06, "loss": 0.6283, "step": 311 }, { "epoch": 0.011115663466163137, "grad_norm": 1.991160538559669, "learning_rate": 7.402135231316726e-06, "loss": 0.5481, "step": 312 }, { "epoch": 0.011151290592657249, "grad_norm": 3.679735956654739, "learning_rate": 7.425860023724793e-06, "loss": 0.7602, "step": 313 }, { "epoch": 0.011186917719151361, "grad_norm": 3.55847697969927, "learning_rate": 7.44958481613286e-06, "loss": 0.7077, "step": 314 }, { "epoch": 0.011222544845645475, "grad_norm": 3.579937320841086, "learning_rate": 7.4733096085409265e-06, "loss": 0.6044, "step": 315 }, { "epoch": 0.011258171972139587, "grad_norm": 7.294830767557365, "learning_rate": 7.497034400948993e-06, "loss": 0.7066, "step": 316 }, { "epoch": 0.0112937990986337, "grad_norm": 2.9937245146661726, "learning_rate": 7.520759193357059e-06, "loss": 0.6564, "step": 317 }, { "epoch": 0.011329426225127812, "grad_norm": 2.3785988636656645, "learning_rate": 7.544483985765125e-06, "loss": 0.5211, "step": 318 }, { "epoch": 0.011365053351621924, "grad_norm": 4.009103860597076, "learning_rate": 7.568208778173192e-06, "loss": 0.6606, "step": 319 }, { "epoch": 0.011400680478116038, "grad_norm": 2.9125935624359287, "learning_rate": 7.591933570581258e-06, "loss": 0.6383, "step": 320 }, { "epoch": 0.01143630760461015, "grad_norm": 3.312474376348779, "learning_rate": 7.6156583629893245e-06, "loss": 0.7358, "step": 321 }, { "epoch": 0.011471934731104264, "grad_norm": 3.721660881263654, "learning_rate": 7.639383155397391e-06, "loss": 0.7075, "step": 322 }, { "epoch": 0.011507561857598375, "grad_norm": 3.6403861330178704, "learning_rate": 7.663107947805457e-06, "loss": 0.6673, "step": 323 }, { "epoch": 0.011543188984092487, "grad_norm": 3.8936419766093304, "learning_rate": 7.686832740213524e-06, "loss": 0.7267, "step": 324 }, { "epoch": 0.011578816110586601, "grad_norm": 4.444049274282742, "learning_rate": 7.71055753262159e-06, "loss": 0.6581, "step": 325 }, { "epoch": 0.011614443237080713, "grad_norm": 2.438281540431582, "learning_rate": 7.734282325029656e-06, "loss": 0.6346, "step": 326 }, { "epoch": 0.011650070363574827, "grad_norm": 2.5348441804556168, "learning_rate": 7.758007117437723e-06, "loss": 0.5871, "step": 327 }, { "epoch": 0.011685697490068939, "grad_norm": 2.880779663932656, "learning_rate": 7.78173190984579e-06, "loss": 0.5506, "step": 328 }, { "epoch": 0.01172132461656305, "grad_norm": 6.864026671367517, "learning_rate": 7.805456702253856e-06, "loss": 0.729, "step": 329 }, { "epoch": 0.011756951743057164, "grad_norm": 3.6677644231087716, "learning_rate": 7.829181494661923e-06, "loss": 0.7098, "step": 330 }, { "epoch": 0.011792578869551276, "grad_norm": 3.5984337022656425, "learning_rate": 7.85290628706999e-06, "loss": 0.6661, "step": 331 }, { "epoch": 0.011828205996045388, "grad_norm": 3.892485981948745, "learning_rate": 7.876631079478055e-06, "loss": 0.6704, "step": 332 }, { "epoch": 0.011863833122539502, "grad_norm": 2.7250416261842494, "learning_rate": 7.900355871886122e-06, "loss": 0.6067, "step": 333 }, { "epoch": 0.011899460249033614, "grad_norm": 3.1868198275711124, "learning_rate": 7.924080664294187e-06, "loss": 0.7295, "step": 334 }, { "epoch": 0.011935087375527727, "grad_norm": 3.520929382091062, "learning_rate": 7.947805456702255e-06, "loss": 0.7285, "step": 335 }, { "epoch": 0.01197071450202184, "grad_norm": 2.7184440330983484, "learning_rate": 7.971530249110322e-06, "loss": 0.6018, "step": 336 }, { "epoch": 0.012006341628515951, "grad_norm": 2.878086976134045, "learning_rate": 7.995255041518387e-06, "loss": 0.6857, "step": 337 }, { "epoch": 0.012041968755010065, "grad_norm": 4.192889929581826, "learning_rate": 8.018979833926454e-06, "loss": 0.6856, "step": 338 }, { "epoch": 0.012077595881504177, "grad_norm": 5.172184318515023, "learning_rate": 8.04270462633452e-06, "loss": 0.7721, "step": 339 }, { "epoch": 0.01211322300799829, "grad_norm": 2.4759651678795627, "learning_rate": 8.066429418742586e-06, "loss": 0.6399, "step": 340 }, { "epoch": 0.012148850134492403, "grad_norm": 2.0266925976187204, "learning_rate": 8.090154211150653e-06, "loss": 0.5941, "step": 341 }, { "epoch": 0.012184477260986514, "grad_norm": 3.9037681106297772, "learning_rate": 8.11387900355872e-06, "loss": 0.864, "step": 342 }, { "epoch": 0.012220104387480628, "grad_norm": 3.49326532532958, "learning_rate": 8.137603795966786e-06, "loss": 0.6202, "step": 343 }, { "epoch": 0.01225573151397474, "grad_norm": 2.642510871531117, "learning_rate": 8.161328588374853e-06, "loss": 0.705, "step": 344 }, { "epoch": 0.012291358640468854, "grad_norm": 2.4095608034111526, "learning_rate": 8.18505338078292e-06, "loss": 0.619, "step": 345 }, { "epoch": 0.012326985766962966, "grad_norm": 2.0872913422112336, "learning_rate": 8.208778173190985e-06, "loss": 0.549, "step": 346 }, { "epoch": 0.012362612893457078, "grad_norm": 2.0268798728565236, "learning_rate": 8.232502965599052e-06, "loss": 0.5071, "step": 347 }, { "epoch": 0.012398240019951191, "grad_norm": 2.5385700236668227, "learning_rate": 8.256227758007118e-06, "loss": 0.6395, "step": 348 }, { "epoch": 0.012433867146445303, "grad_norm": 2.089905608881449, "learning_rate": 8.279952550415185e-06, "loss": 0.6329, "step": 349 }, { "epoch": 0.012469494272939415, "grad_norm": 2.4682596288736858, "learning_rate": 8.303677342823252e-06, "loss": 0.5975, "step": 350 }, { "epoch": 0.012505121399433529, "grad_norm": 3.441842901996113, "learning_rate": 8.327402135231317e-06, "loss": 0.6673, "step": 351 }, { "epoch": 0.01254074852592764, "grad_norm": 2.100951576158508, "learning_rate": 8.351126927639384e-06, "loss": 0.5653, "step": 352 }, { "epoch": 0.012576375652421755, "grad_norm": 3.095113434058388, "learning_rate": 8.37485172004745e-06, "loss": 0.6669, "step": 353 }, { "epoch": 0.012612002778915866, "grad_norm": 2.928602456703605, "learning_rate": 8.398576512455516e-06, "loss": 0.7243, "step": 354 }, { "epoch": 0.012647629905409978, "grad_norm": 2.6918811721119025, "learning_rate": 8.422301304863583e-06, "loss": 0.7415, "step": 355 }, { "epoch": 0.012683257031904092, "grad_norm": 2.3427334819504653, "learning_rate": 8.446026097271649e-06, "loss": 0.6218, "step": 356 }, { "epoch": 0.012718884158398204, "grad_norm": 3.1551656466045355, "learning_rate": 8.469750889679716e-06, "loss": 0.7654, "step": 357 }, { "epoch": 0.012754511284892318, "grad_norm": 3.3736498769434125, "learning_rate": 8.493475682087783e-06, "loss": 0.577, "step": 358 }, { "epoch": 0.01279013841138643, "grad_norm": 3.2266330577176734, "learning_rate": 8.51720047449585e-06, "loss": 0.6851, "step": 359 }, { "epoch": 0.012825765537880542, "grad_norm": 2.318562896532133, "learning_rate": 8.540925266903915e-06, "loss": 0.5619, "step": 360 }, { "epoch": 0.012861392664374655, "grad_norm": 3.281573435793028, "learning_rate": 8.564650059311982e-06, "loss": 0.7358, "step": 361 }, { "epoch": 0.012897019790868767, "grad_norm": 3.050400067046054, "learning_rate": 8.588374851720048e-06, "loss": 0.6887, "step": 362 }, { "epoch": 0.012932646917362881, "grad_norm": 2.268103515114853, "learning_rate": 8.612099644128115e-06, "loss": 0.5941, "step": 363 }, { "epoch": 0.012968274043856993, "grad_norm": 2.055601525926488, "learning_rate": 8.635824436536182e-06, "loss": 0.5398, "step": 364 }, { "epoch": 0.013003901170351105, "grad_norm": 1.9654907585550216, "learning_rate": 8.659549228944247e-06, "loss": 0.6011, "step": 365 }, { "epoch": 0.013039528296845218, "grad_norm": 2.8027961342651713, "learning_rate": 8.683274021352314e-06, "loss": 0.5645, "step": 366 }, { "epoch": 0.01307515542333933, "grad_norm": 3.07498345548557, "learning_rate": 8.70699881376038e-06, "loss": 0.737, "step": 367 }, { "epoch": 0.013110782549833444, "grad_norm": 2.304214511396199, "learning_rate": 8.730723606168446e-06, "loss": 0.5642, "step": 368 }, { "epoch": 0.013146409676327556, "grad_norm": 2.935052564197731, "learning_rate": 8.754448398576513e-06, "loss": 0.6343, "step": 369 }, { "epoch": 0.013182036802821668, "grad_norm": 1.8753424476316471, "learning_rate": 8.778173190984579e-06, "loss": 0.5692, "step": 370 }, { "epoch": 0.013217663929315782, "grad_norm": 3.2781030407585696, "learning_rate": 8.801897983392646e-06, "loss": 0.6448, "step": 371 }, { "epoch": 0.013253291055809894, "grad_norm": 2.8699624531631986, "learning_rate": 8.825622775800713e-06, "loss": 0.6276, "step": 372 }, { "epoch": 0.013288918182304005, "grad_norm": 2.0728491721424476, "learning_rate": 8.84934756820878e-06, "loss": 0.6133, "step": 373 }, { "epoch": 0.01332454530879812, "grad_norm": 1.9381049177503027, "learning_rate": 8.873072360616845e-06, "loss": 0.5928, "step": 374 }, { "epoch": 0.013360172435292231, "grad_norm": 1.749849422246167, "learning_rate": 8.896797153024912e-06, "loss": 0.5637, "step": 375 }, { "epoch": 0.013395799561786345, "grad_norm": 1.938310736861352, "learning_rate": 8.920521945432978e-06, "loss": 0.5905, "step": 376 }, { "epoch": 0.013431426688280457, "grad_norm": 2.846372738489191, "learning_rate": 8.944246737841045e-06, "loss": 0.6619, "step": 377 }, { "epoch": 0.013467053814774569, "grad_norm": 2.9065823680211187, "learning_rate": 8.967971530249112e-06, "loss": 0.6372, "step": 378 }, { "epoch": 0.013502680941268682, "grad_norm": 1.9914535915748723, "learning_rate": 8.991696322657177e-06, "loss": 0.5421, "step": 379 }, { "epoch": 0.013538308067762794, "grad_norm": 1.9452984250716494, "learning_rate": 9.015421115065244e-06, "loss": 0.5802, "step": 380 }, { "epoch": 0.013573935194256908, "grad_norm": 2.552365341648953, "learning_rate": 9.03914590747331e-06, "loss": 0.721, "step": 381 }, { "epoch": 0.01360956232075102, "grad_norm": 3.092192582526139, "learning_rate": 9.062870699881377e-06, "loss": 0.7488, "step": 382 }, { "epoch": 0.013645189447245132, "grad_norm": 2.1989918188774866, "learning_rate": 9.086595492289444e-06, "loss": 0.5985, "step": 383 }, { "epoch": 0.013680816573739246, "grad_norm": 2.9502111012138896, "learning_rate": 9.110320284697509e-06, "loss": 0.6414, "step": 384 }, { "epoch": 0.013716443700233357, "grad_norm": 2.3848037632656625, "learning_rate": 9.134045077105576e-06, "loss": 0.5982, "step": 385 }, { "epoch": 0.013752070826727471, "grad_norm": 3.2939930255689513, "learning_rate": 9.157769869513643e-06, "loss": 0.5855, "step": 386 }, { "epoch": 0.013787697953221583, "grad_norm": 3.012032877334934, "learning_rate": 9.18149466192171e-06, "loss": 0.6869, "step": 387 }, { "epoch": 0.013823325079715695, "grad_norm": 2.3640653009385635, "learning_rate": 9.205219454329775e-06, "loss": 0.5643, "step": 388 }, { "epoch": 0.013858952206209809, "grad_norm": 1.7970839619779742, "learning_rate": 9.228944246737842e-06, "loss": 0.5714, "step": 389 }, { "epoch": 0.01389457933270392, "grad_norm": 1.948120835693504, "learning_rate": 9.252669039145908e-06, "loss": 0.6167, "step": 390 }, { "epoch": 0.013930206459198033, "grad_norm": 6.259288611309589, "learning_rate": 9.276393831553975e-06, "loss": 0.5709, "step": 391 }, { "epoch": 0.013965833585692146, "grad_norm": 1.8860520144113322, "learning_rate": 9.300118623962042e-06, "loss": 0.5602, "step": 392 }, { "epoch": 0.014001460712186258, "grad_norm": 1.991714883960291, "learning_rate": 9.323843416370107e-06, "loss": 0.6112, "step": 393 }, { "epoch": 0.014037087838680372, "grad_norm": 2.6383604246115113, "learning_rate": 9.347568208778174e-06, "loss": 0.6852, "step": 394 }, { "epoch": 0.014072714965174484, "grad_norm": 1.9945247020623404, "learning_rate": 9.37129300118624e-06, "loss": 0.6284, "step": 395 }, { "epoch": 0.014108342091668596, "grad_norm": 3.41258771273078, "learning_rate": 9.395017793594307e-06, "loss": 0.6269, "step": 396 }, { "epoch": 0.01414396921816271, "grad_norm": 3.476703598400501, "learning_rate": 9.418742586002374e-06, "loss": 0.6179, "step": 397 }, { "epoch": 0.014179596344656821, "grad_norm": 3.349937503082234, "learning_rate": 9.442467378410439e-06, "loss": 0.6196, "step": 398 }, { "epoch": 0.014215223471150935, "grad_norm": 1.816270646981821, "learning_rate": 9.466192170818506e-06, "loss": 0.6064, "step": 399 }, { "epoch": 0.014250850597645047, "grad_norm": 3.261591759454169, "learning_rate": 9.489916963226573e-06, "loss": 0.775, "step": 400 }, { "epoch": 0.014286477724139159, "grad_norm": 2.221024712340117, "learning_rate": 9.51364175563464e-06, "loss": 0.6101, "step": 401 }, { "epoch": 0.014322104850633273, "grad_norm": 1.9611890596970853, "learning_rate": 9.537366548042705e-06, "loss": 0.598, "step": 402 }, { "epoch": 0.014357731977127385, "grad_norm": 2.01720021843294, "learning_rate": 9.561091340450772e-06, "loss": 0.607, "step": 403 }, { "epoch": 0.014393359103621498, "grad_norm": 2.052655067453154, "learning_rate": 9.584816132858838e-06, "loss": 0.5904, "step": 404 }, { "epoch": 0.01442898623011561, "grad_norm": 2.811861237794079, "learning_rate": 9.608540925266905e-06, "loss": 0.5982, "step": 405 }, { "epoch": 0.014464613356609722, "grad_norm": 2.0277968184200073, "learning_rate": 9.632265717674972e-06, "loss": 0.5656, "step": 406 }, { "epoch": 0.014500240483103836, "grad_norm": 3.030816213521878, "learning_rate": 9.655990510083037e-06, "loss": 0.614, "step": 407 }, { "epoch": 0.014535867609597948, "grad_norm": 1.8028746297819265, "learning_rate": 9.679715302491104e-06, "loss": 0.5166, "step": 408 }, { "epoch": 0.01457149473609206, "grad_norm": 2.6733424791995564, "learning_rate": 9.70344009489917e-06, "loss": 0.7469, "step": 409 }, { "epoch": 0.014607121862586173, "grad_norm": 2.0269686521673607, "learning_rate": 9.727164887307237e-06, "loss": 0.5655, "step": 410 }, { "epoch": 0.014642748989080285, "grad_norm": 3.2321210068787414, "learning_rate": 9.750889679715304e-06, "loss": 0.6406, "step": 411 }, { "epoch": 0.014678376115574399, "grad_norm": 2.06026212015437, "learning_rate": 9.774614472123369e-06, "loss": 0.5721, "step": 412 }, { "epoch": 0.014714003242068511, "grad_norm": 1.8295250164409143, "learning_rate": 9.798339264531436e-06, "loss": 0.5595, "step": 413 }, { "epoch": 0.014749630368562623, "grad_norm": 3.8030917236628943, "learning_rate": 9.822064056939501e-06, "loss": 0.7319, "step": 414 }, { "epoch": 0.014785257495056736, "grad_norm": 1.9834115170942024, "learning_rate": 9.84578884934757e-06, "loss": 0.5737, "step": 415 }, { "epoch": 0.014820884621550848, "grad_norm": 1.8724379787548506, "learning_rate": 9.869513641755635e-06, "loss": 0.6216, "step": 416 }, { "epoch": 0.014856511748044962, "grad_norm": 2.0140691520001774, "learning_rate": 9.893238434163703e-06, "loss": 0.5934, "step": 417 }, { "epoch": 0.014892138874539074, "grad_norm": 2.1466847302403935, "learning_rate": 9.916963226571768e-06, "loss": 0.5359, "step": 418 }, { "epoch": 0.014927766001033186, "grad_norm": 4.252566285795933, "learning_rate": 9.940688018979835e-06, "loss": 0.7198, "step": 419 }, { "epoch": 0.0149633931275273, "grad_norm": 2.8304113157249016, "learning_rate": 9.964412811387902e-06, "loss": 0.6476, "step": 420 }, { "epoch": 0.014999020254021412, "grad_norm": 3.0441748745101656, "learning_rate": 9.988137603795967e-06, "loss": 0.6727, "step": 421 }, { "epoch": 0.015034647380515525, "grad_norm": 2.6980763340176552, "learning_rate": 1.0011862396204034e-05, "loss": 0.6592, "step": 422 }, { "epoch": 0.015070274507009637, "grad_norm": 2.0931838719718763, "learning_rate": 1.0035587188612101e-05, "loss": 0.6015, "step": 423 }, { "epoch": 0.01510590163350375, "grad_norm": 4.714715801900258, "learning_rate": 1.0059311981020168e-05, "loss": 0.6357, "step": 424 }, { "epoch": 0.015141528759997863, "grad_norm": 2.198828194912598, "learning_rate": 1.0083036773428234e-05, "loss": 0.5442, "step": 425 }, { "epoch": 0.015177155886491975, "grad_norm": 3.2208063463645495, "learning_rate": 1.01067615658363e-05, "loss": 0.6521, "step": 426 }, { "epoch": 0.015212783012986088, "grad_norm": 1.9135631883509603, "learning_rate": 1.0130486358244366e-05, "loss": 0.6032, "step": 427 }, { "epoch": 0.0152484101394802, "grad_norm": 1.7288229569216063, "learning_rate": 1.0154211150652433e-05, "loss": 0.5609, "step": 428 }, { "epoch": 0.015284037265974312, "grad_norm": 3.6088163661734303, "learning_rate": 1.01779359430605e-05, "loss": 0.6377, "step": 429 }, { "epoch": 0.015319664392468426, "grad_norm": 2.1042991044807526, "learning_rate": 1.0201660735468566e-05, "loss": 0.6424, "step": 430 }, { "epoch": 0.015355291518962538, "grad_norm": 2.718960531067694, "learning_rate": 1.0225385527876633e-05, "loss": 0.6652, "step": 431 }, { "epoch": 0.01539091864545665, "grad_norm": 2.576667087617043, "learning_rate": 1.0249110320284698e-05, "loss": 0.6485, "step": 432 }, { "epoch": 0.015426545771950764, "grad_norm": 2.630829280054204, "learning_rate": 1.0272835112692765e-05, "loss": 0.6973, "step": 433 }, { "epoch": 0.015462172898444876, "grad_norm": 1.9846843937194205, "learning_rate": 1.0296559905100832e-05, "loss": 0.5268, "step": 434 }, { "epoch": 0.01549780002493899, "grad_norm": 1.6927241722328763, "learning_rate": 1.0320284697508897e-05, "loss": 0.521, "step": 435 }, { "epoch": 0.015533427151433101, "grad_norm": 2.8331220248832274, "learning_rate": 1.0344009489916964e-05, "loss": 0.7183, "step": 436 }, { "epoch": 0.015569054277927213, "grad_norm": 2.0355099095668776, "learning_rate": 1.036773428232503e-05, "loss": 0.5351, "step": 437 }, { "epoch": 0.015604681404421327, "grad_norm": 2.6830805516658254, "learning_rate": 1.0391459074733097e-05, "loss": 0.7045, "step": 438 }, { "epoch": 0.01564030853091544, "grad_norm": 1.8218842372033939, "learning_rate": 1.0415183867141164e-05, "loss": 0.5539, "step": 439 }, { "epoch": 0.015675935657409552, "grad_norm": 2.469663893223558, "learning_rate": 1.043890865954923e-05, "loss": 0.678, "step": 440 }, { "epoch": 0.015711562783903663, "grad_norm": 2.039740981451196, "learning_rate": 1.0462633451957296e-05, "loss": 0.5864, "step": 441 }, { "epoch": 0.015747189910397776, "grad_norm": 1.9038273710418796, "learning_rate": 1.0486358244365362e-05, "loss": 0.6357, "step": 442 }, { "epoch": 0.01578281703689189, "grad_norm": 1.994082814132985, "learning_rate": 1.0510083036773429e-05, "loss": 0.5804, "step": 443 }, { "epoch": 0.015818444163386004, "grad_norm": 2.0121480928684017, "learning_rate": 1.0533807829181496e-05, "loss": 0.5612, "step": 444 }, { "epoch": 0.015854071289880114, "grad_norm": 2.1123097513838878, "learning_rate": 1.0557532621589561e-05, "loss": 0.5913, "step": 445 }, { "epoch": 0.015889698416374227, "grad_norm": 2.007529403515734, "learning_rate": 1.0581257413997628e-05, "loss": 0.59, "step": 446 }, { "epoch": 0.01592532554286834, "grad_norm": 1.9951654537404702, "learning_rate": 1.0604982206405693e-05, "loss": 0.576, "step": 447 }, { "epoch": 0.01596095266936245, "grad_norm": 1.9242316883658714, "learning_rate": 1.062870699881376e-05, "loss": 0.5745, "step": 448 }, { "epoch": 0.015996579795856565, "grad_norm": 1.9707351672709252, "learning_rate": 1.0652431791221827e-05, "loss": 0.5921, "step": 449 }, { "epoch": 0.01603220692235068, "grad_norm": 3.318289848274285, "learning_rate": 1.0676156583629893e-05, "loss": 0.7643, "step": 450 }, { "epoch": 0.01606783404884479, "grad_norm": 5.556962959867927, "learning_rate": 1.0699881376037962e-05, "loss": 0.6014, "step": 451 }, { "epoch": 0.016103461175338903, "grad_norm": 2.6194795388913903, "learning_rate": 1.0723606168446029e-05, "loss": 0.6362, "step": 452 }, { "epoch": 0.016139088301833016, "grad_norm": 1.8907161430664694, "learning_rate": 1.0747330960854094e-05, "loss": 0.5989, "step": 453 }, { "epoch": 0.01617471542832713, "grad_norm": 6.0587580439736035, "learning_rate": 1.0771055753262161e-05, "loss": 0.6524, "step": 454 }, { "epoch": 0.01621034255482124, "grad_norm": 2.9272632003848993, "learning_rate": 1.0794780545670226e-05, "loss": 0.6515, "step": 455 }, { "epoch": 0.016245969681315354, "grad_norm": 2.232905827193425, "learning_rate": 1.0818505338078293e-05, "loss": 0.5959, "step": 456 }, { "epoch": 0.016281596807809467, "grad_norm": 6.214122795993087, "learning_rate": 1.084223013048636e-05, "loss": 0.7036, "step": 457 }, { "epoch": 0.016317223934303578, "grad_norm": 3.0436110540177794, "learning_rate": 1.0865954922894426e-05, "loss": 0.7153, "step": 458 }, { "epoch": 0.01635285106079769, "grad_norm": 2.138484395190815, "learning_rate": 1.0889679715302493e-05, "loss": 0.5773, "step": 459 }, { "epoch": 0.016388478187291805, "grad_norm": 2.5404082653726823, "learning_rate": 1.0913404507710558e-05, "loss": 0.6329, "step": 460 }, { "epoch": 0.016424105313785915, "grad_norm": 9.643622237307513, "learning_rate": 1.0937129300118625e-05, "loss": 0.6592, "step": 461 }, { "epoch": 0.01645973244028003, "grad_norm": 3.4233436478620827, "learning_rate": 1.0960854092526692e-05, "loss": 0.6107, "step": 462 }, { "epoch": 0.016495359566774143, "grad_norm": 2.0474200351200422, "learning_rate": 1.0984578884934757e-05, "loss": 0.5588, "step": 463 }, { "epoch": 0.016530986693268253, "grad_norm": 2.685696845852546, "learning_rate": 1.1008303677342825e-05, "loss": 0.6488, "step": 464 }, { "epoch": 0.016566613819762366, "grad_norm": 3.6064290407970794, "learning_rate": 1.103202846975089e-05, "loss": 0.6962, "step": 465 }, { "epoch": 0.01660224094625648, "grad_norm": 1.75501689412755, "learning_rate": 1.1055753262158957e-05, "loss": 0.5877, "step": 466 }, { "epoch": 0.016637868072750594, "grad_norm": 2.6083850765628003, "learning_rate": 1.1079478054567024e-05, "loss": 0.6066, "step": 467 }, { "epoch": 0.016673495199244704, "grad_norm": 2.212466812667502, "learning_rate": 1.110320284697509e-05, "loss": 0.5923, "step": 468 }, { "epoch": 0.016709122325738818, "grad_norm": 2.994989134751467, "learning_rate": 1.1126927639383156e-05, "loss": 0.7105, "step": 469 }, { "epoch": 0.01674474945223293, "grad_norm": 3.101730192027918, "learning_rate": 1.1150652431791222e-05, "loss": 0.6853, "step": 470 }, { "epoch": 0.01678037657872704, "grad_norm": 2.453934796582532, "learning_rate": 1.1174377224199289e-05, "loss": 0.6229, "step": 471 }, { "epoch": 0.016816003705221155, "grad_norm": 2.542389839734008, "learning_rate": 1.1198102016607356e-05, "loss": 0.627, "step": 472 }, { "epoch": 0.01685163083171527, "grad_norm": 3.0858037267961143, "learning_rate": 1.1221826809015421e-05, "loss": 0.6048, "step": 473 }, { "epoch": 0.01688725795820938, "grad_norm": 2.89721375133124, "learning_rate": 1.1245551601423488e-05, "loss": 0.7243, "step": 474 }, { "epoch": 0.016922885084703493, "grad_norm": 2.084444977237843, "learning_rate": 1.1269276393831553e-05, "loss": 0.5317, "step": 475 }, { "epoch": 0.016958512211197607, "grad_norm": 2.7772169620745366, "learning_rate": 1.129300118623962e-05, "loss": 0.7456, "step": 476 }, { "epoch": 0.01699413933769172, "grad_norm": 1.9991228465984812, "learning_rate": 1.1316725978647688e-05, "loss": 0.5628, "step": 477 }, { "epoch": 0.01702976646418583, "grad_norm": 2.127932828145017, "learning_rate": 1.1340450771055753e-05, "loss": 0.5516, "step": 478 }, { "epoch": 0.017065393590679944, "grad_norm": 1.9216599232259264, "learning_rate": 1.136417556346382e-05, "loss": 0.5365, "step": 479 }, { "epoch": 0.017101020717174058, "grad_norm": 1.9171448981281445, "learning_rate": 1.1387900355871889e-05, "loss": 0.5194, "step": 480 }, { "epoch": 0.017136647843668168, "grad_norm": 1.9500109904502596, "learning_rate": 1.1411625148279954e-05, "loss": 0.5714, "step": 481 }, { "epoch": 0.01717227497016228, "grad_norm": 6.302344033414598, "learning_rate": 1.1435349940688021e-05, "loss": 0.6778, "step": 482 }, { "epoch": 0.017207902096656395, "grad_norm": 2.708600343364102, "learning_rate": 1.1459074733096086e-05, "loss": 0.717, "step": 483 }, { "epoch": 0.017243529223150506, "grad_norm": 3.2306557785008736, "learning_rate": 1.1482799525504153e-05, "loss": 0.6976, "step": 484 }, { "epoch": 0.01727915634964462, "grad_norm": 3.742192701093716, "learning_rate": 1.150652431791222e-05, "loss": 0.6504, "step": 485 }, { "epoch": 0.017314783476138733, "grad_norm": 2.424848266630896, "learning_rate": 1.1530249110320286e-05, "loss": 0.6138, "step": 486 }, { "epoch": 0.017350410602632843, "grad_norm": 1.9169125274283287, "learning_rate": 1.1553973902728353e-05, "loss": 0.5492, "step": 487 }, { "epoch": 0.017386037729126957, "grad_norm": 2.008940559863694, "learning_rate": 1.1577698695136418e-05, "loss": 0.592, "step": 488 }, { "epoch": 0.01742166485562107, "grad_norm": 2.8690268660122524, "learning_rate": 1.1601423487544485e-05, "loss": 0.6955, "step": 489 }, { "epoch": 0.017457291982115184, "grad_norm": 1.9971841940150412, "learning_rate": 1.1625148279952552e-05, "loss": 0.4998, "step": 490 }, { "epoch": 0.017492919108609294, "grad_norm": 3.7452907965766884, "learning_rate": 1.1648873072360618e-05, "loss": 0.5822, "step": 491 }, { "epoch": 0.017528546235103408, "grad_norm": 2.735556333081387, "learning_rate": 1.1672597864768685e-05, "loss": 0.7171, "step": 492 }, { "epoch": 0.01756417336159752, "grad_norm": 2.8529926402665278, "learning_rate": 1.169632265717675e-05, "loss": 0.636, "step": 493 }, { "epoch": 0.017599800488091632, "grad_norm": 3.288698582084974, "learning_rate": 1.1720047449584817e-05, "loss": 0.5759, "step": 494 }, { "epoch": 0.017635427614585746, "grad_norm": 2.978447447500666, "learning_rate": 1.1743772241992884e-05, "loss": 0.6473, "step": 495 }, { "epoch": 0.01767105474107986, "grad_norm": 2.149559807260798, "learning_rate": 1.176749703440095e-05, "loss": 0.5976, "step": 496 }, { "epoch": 0.01770668186757397, "grad_norm": 2.0091560563598185, "learning_rate": 1.1791221826809016e-05, "loss": 0.5409, "step": 497 }, { "epoch": 0.017742308994068083, "grad_norm": 2.1598645350630297, "learning_rate": 1.1814946619217082e-05, "loss": 0.5653, "step": 498 }, { "epoch": 0.017777936120562197, "grad_norm": 1.9345573401622183, "learning_rate": 1.1838671411625149e-05, "loss": 0.5269, "step": 499 }, { "epoch": 0.017813563247056307, "grad_norm": 2.5692661560728642, "learning_rate": 1.1862396204033216e-05, "loss": 0.6013, "step": 500 }, { "epoch": 0.01784919037355042, "grad_norm": 2.0121242407536126, "learning_rate": 1.1886120996441281e-05, "loss": 0.5577, "step": 501 }, { "epoch": 0.017884817500044534, "grad_norm": 2.2564032751605088, "learning_rate": 1.1909845788849348e-05, "loss": 0.6344, "step": 502 }, { "epoch": 0.017920444626538648, "grad_norm": 3.5367335379437668, "learning_rate": 1.1933570581257414e-05, "loss": 0.6611, "step": 503 }, { "epoch": 0.017956071753032758, "grad_norm": 3.501763294361195, "learning_rate": 1.195729537366548e-05, "loss": 0.7581, "step": 504 }, { "epoch": 0.017991698879526872, "grad_norm": 2.5388603060204367, "learning_rate": 1.1981020166073548e-05, "loss": 0.6237, "step": 505 }, { "epoch": 0.018027326006020986, "grad_norm": 2.585731833812994, "learning_rate": 1.2004744958481613e-05, "loss": 0.6232, "step": 506 }, { "epoch": 0.018062953132515096, "grad_norm": 2.1364407786261297, "learning_rate": 1.202846975088968e-05, "loss": 0.5896, "step": 507 }, { "epoch": 0.01809858025900921, "grad_norm": 2.242211418529812, "learning_rate": 1.2052194543297745e-05, "loss": 0.5653, "step": 508 }, { "epoch": 0.018134207385503323, "grad_norm": 2.0676553322319466, "learning_rate": 1.2075919335705814e-05, "loss": 0.5704, "step": 509 }, { "epoch": 0.018169834511997433, "grad_norm": 2.829129986683506, "learning_rate": 1.2099644128113881e-05, "loss": 0.6321, "step": 510 }, { "epoch": 0.018205461638491547, "grad_norm": 2.1085004297017056, "learning_rate": 1.2123368920521947e-05, "loss": 0.5948, "step": 511 }, { "epoch": 0.01824108876498566, "grad_norm": 2.3809046919079813, "learning_rate": 1.2147093712930014e-05, "loss": 0.5888, "step": 512 }, { "epoch": 0.018276715891479774, "grad_norm": 1.9987532991667714, "learning_rate": 1.217081850533808e-05, "loss": 0.5255, "step": 513 }, { "epoch": 0.018312343017973885, "grad_norm": 1.8746314575010599, "learning_rate": 1.2194543297746146e-05, "loss": 0.636, "step": 514 }, { "epoch": 0.018347970144467998, "grad_norm": 1.9603931866895687, "learning_rate": 1.2218268090154213e-05, "loss": 0.6514, "step": 515 }, { "epoch": 0.018383597270962112, "grad_norm": 2.0413255138647886, "learning_rate": 1.2241992882562278e-05, "loss": 0.5635, "step": 516 }, { "epoch": 0.018419224397456222, "grad_norm": 2.4002092138591093, "learning_rate": 1.2265717674970345e-05, "loss": 0.601, "step": 517 }, { "epoch": 0.018454851523950336, "grad_norm": 2.799242717774119, "learning_rate": 1.2289442467378412e-05, "loss": 0.6781, "step": 518 }, { "epoch": 0.01849047865044445, "grad_norm": 2.3510867830663362, "learning_rate": 1.2313167259786478e-05, "loss": 0.7038, "step": 519 }, { "epoch": 0.01852610577693856, "grad_norm": 3.0091560648666564, "learning_rate": 1.2336892052194545e-05, "loss": 0.615, "step": 520 }, { "epoch": 0.018561732903432673, "grad_norm": 2.6721785661311657, "learning_rate": 1.236061684460261e-05, "loss": 0.5753, "step": 521 }, { "epoch": 0.018597360029926787, "grad_norm": 2.8001047071288445, "learning_rate": 1.2384341637010677e-05, "loss": 0.5645, "step": 522 }, { "epoch": 0.018632987156420897, "grad_norm": 2.856125816636998, "learning_rate": 1.2408066429418744e-05, "loss": 0.7468, "step": 523 }, { "epoch": 0.01866861428291501, "grad_norm": 3.0025758528170097, "learning_rate": 1.243179122182681e-05, "loss": 0.6724, "step": 524 }, { "epoch": 0.018704241409409125, "grad_norm": 2.831495493728899, "learning_rate": 1.2455516014234877e-05, "loss": 0.5959, "step": 525 }, { "epoch": 0.018739868535903238, "grad_norm": 3.021886993132296, "learning_rate": 1.2479240806642942e-05, "loss": 0.734, "step": 526 }, { "epoch": 0.01877549566239735, "grad_norm": 3.3711896967313866, "learning_rate": 1.2502965599051009e-05, "loss": 0.6997, "step": 527 }, { "epoch": 0.018811122788891462, "grad_norm": 3.5939606401224373, "learning_rate": 1.2526690391459076e-05, "loss": 0.6755, "step": 528 }, { "epoch": 0.018846749915385576, "grad_norm": 3.211008176839121, "learning_rate": 1.2550415183867141e-05, "loss": 0.5965, "step": 529 }, { "epoch": 0.018882377041879686, "grad_norm": 2.190616655569529, "learning_rate": 1.2574139976275208e-05, "loss": 0.5804, "step": 530 }, { "epoch": 0.0189180041683738, "grad_norm": 2.676318322810922, "learning_rate": 1.2597864768683274e-05, "loss": 0.6875, "step": 531 }, { "epoch": 0.018953631294867913, "grad_norm": 3.275259551823701, "learning_rate": 1.262158956109134e-05, "loss": 0.6589, "step": 532 }, { "epoch": 0.018989258421362024, "grad_norm": 3.8122683072949712, "learning_rate": 1.2645314353499408e-05, "loss": 0.7112, "step": 533 }, { "epoch": 0.019024885547856137, "grad_norm": 2.6441882601032805, "learning_rate": 1.2669039145907473e-05, "loss": 0.6558, "step": 534 }, { "epoch": 0.01906051267435025, "grad_norm": 2.6472541726947854, "learning_rate": 1.269276393831554e-05, "loss": 0.675, "step": 535 }, { "epoch": 0.019096139800844365, "grad_norm": 2.8732662246676264, "learning_rate": 1.2716488730723606e-05, "loss": 0.7891, "step": 536 }, { "epoch": 0.019131766927338475, "grad_norm": 3.1372549842346005, "learning_rate": 1.2740213523131673e-05, "loss": 0.7524, "step": 537 }, { "epoch": 0.01916739405383259, "grad_norm": 2.6254348433756705, "learning_rate": 1.2763938315539741e-05, "loss": 0.7525, "step": 538 }, { "epoch": 0.019203021180326702, "grad_norm": 4.773369338245711, "learning_rate": 1.2787663107947808e-05, "loss": 0.6698, "step": 539 }, { "epoch": 0.019238648306820812, "grad_norm": 7.278129603815351, "learning_rate": 1.2811387900355874e-05, "loss": 0.6088, "step": 540 }, { "epoch": 0.019274275433314926, "grad_norm": 2.2809715410630487, "learning_rate": 1.283511269276394e-05, "loss": 0.6485, "step": 541 }, { "epoch": 0.01930990255980904, "grad_norm": 2.990483419098431, "learning_rate": 1.2858837485172006e-05, "loss": 0.6147, "step": 542 }, { "epoch": 0.01934552968630315, "grad_norm": 2.551971068328214, "learning_rate": 1.2882562277580073e-05, "loss": 0.558, "step": 543 }, { "epoch": 0.019381156812797264, "grad_norm": 4.144128252410374, "learning_rate": 1.2906287069988138e-05, "loss": 0.6571, "step": 544 }, { "epoch": 0.019416783939291377, "grad_norm": 2.1465343308814324, "learning_rate": 1.2930011862396206e-05, "loss": 0.6156, "step": 545 }, { "epoch": 0.019452411065785487, "grad_norm": 3.1470850480511645, "learning_rate": 1.2953736654804273e-05, "loss": 0.6726, "step": 546 }, { "epoch": 0.0194880381922796, "grad_norm": 2.537861939095209, "learning_rate": 1.2977461447212338e-05, "loss": 0.5998, "step": 547 }, { "epoch": 0.019523665318773715, "grad_norm": 1.9860615492730311, "learning_rate": 1.3001186239620405e-05, "loss": 0.6256, "step": 548 }, { "epoch": 0.01955929244526783, "grad_norm": 2.130064322495938, "learning_rate": 1.302491103202847e-05, "loss": 0.6251, "step": 549 }, { "epoch": 0.01959491957176194, "grad_norm": 3.6663223024406597, "learning_rate": 1.3048635824436537e-05, "loss": 0.729, "step": 550 }, { "epoch": 0.019630546698256052, "grad_norm": 4.326681865342257, "learning_rate": 1.3072360616844604e-05, "loss": 0.7179, "step": 551 }, { "epoch": 0.019666173824750166, "grad_norm": 2.200571563937244, "learning_rate": 1.309608540925267e-05, "loss": 0.6178, "step": 552 }, { "epoch": 0.019701800951244276, "grad_norm": 2.421016758764712, "learning_rate": 1.3119810201660737e-05, "loss": 0.5983, "step": 553 }, { "epoch": 0.01973742807773839, "grad_norm": 1.7929342100100343, "learning_rate": 1.3143534994068802e-05, "loss": 0.5469, "step": 554 }, { "epoch": 0.019773055204232504, "grad_norm": 2.0480556539142043, "learning_rate": 1.3167259786476869e-05, "loss": 0.5479, "step": 555 }, { "epoch": 0.019808682330726614, "grad_norm": 3.313951358075255, "learning_rate": 1.3190984578884936e-05, "loss": 0.7022, "step": 556 }, { "epoch": 0.019844309457220728, "grad_norm": 2.583080393042542, "learning_rate": 1.3214709371293001e-05, "loss": 0.6424, "step": 557 }, { "epoch": 0.01987993658371484, "grad_norm": 4.450658242725031, "learning_rate": 1.3238434163701069e-05, "loss": 0.6231, "step": 558 }, { "epoch": 0.01991556371020895, "grad_norm": 2.420916954559214, "learning_rate": 1.3262158956109134e-05, "loss": 0.6686, "step": 559 }, { "epoch": 0.019951190836703065, "grad_norm": 2.754316824377658, "learning_rate": 1.3285883748517201e-05, "loss": 0.7232, "step": 560 }, { "epoch": 0.01998681796319718, "grad_norm": 2.5842799667179426, "learning_rate": 1.3309608540925268e-05, "loss": 0.6276, "step": 561 }, { "epoch": 0.020022445089691292, "grad_norm": 3.245490251911885, "learning_rate": 1.3333333333333333e-05, "loss": 0.6816, "step": 562 }, { "epoch": 0.020058072216185403, "grad_norm": 2.136314568843926, "learning_rate": 1.33570581257414e-05, "loss": 0.5885, "step": 563 }, { "epoch": 0.020093699342679516, "grad_norm": 1.8432305011299284, "learning_rate": 1.3380782918149466e-05, "loss": 0.5789, "step": 564 }, { "epoch": 0.02012932646917363, "grad_norm": 3.0300568215011987, "learning_rate": 1.3404507710557533e-05, "loss": 0.5681, "step": 565 }, { "epoch": 0.02016495359566774, "grad_norm": 4.163783099907701, "learning_rate": 1.34282325029656e-05, "loss": 0.6983, "step": 566 }, { "epoch": 0.020200580722161854, "grad_norm": 2.275101532779074, "learning_rate": 1.3451957295373668e-05, "loss": 0.5493, "step": 567 }, { "epoch": 0.020236207848655968, "grad_norm": 2.6264035355739694, "learning_rate": 1.3475682087781734e-05, "loss": 0.625, "step": 568 }, { "epoch": 0.020271834975150078, "grad_norm": 6.727959790263543, "learning_rate": 1.3499406880189801e-05, "loss": 0.5949, "step": 569 }, { "epoch": 0.02030746210164419, "grad_norm": 1.8894441214113007, "learning_rate": 1.3523131672597866e-05, "loss": 0.5446, "step": 570 }, { "epoch": 0.020343089228138305, "grad_norm": 1.7437399386106012, "learning_rate": 1.3546856465005933e-05, "loss": 0.5449, "step": 571 }, { "epoch": 0.02037871635463242, "grad_norm": 2.060140857076962, "learning_rate": 1.3570581257414e-05, "loss": 0.5979, "step": 572 }, { "epoch": 0.02041434348112653, "grad_norm": 2.1047380660913895, "learning_rate": 1.3594306049822066e-05, "loss": 0.5749, "step": 573 }, { "epoch": 0.020449970607620643, "grad_norm": 3.0581305514452364, "learning_rate": 1.3618030842230133e-05, "loss": 0.5807, "step": 574 }, { "epoch": 0.020485597734114756, "grad_norm": 2.1590353673874194, "learning_rate": 1.3641755634638198e-05, "loss": 0.5734, "step": 575 }, { "epoch": 0.020521224860608867, "grad_norm": 1.9909932012906104, "learning_rate": 1.3665480427046265e-05, "loss": 0.5584, "step": 576 }, { "epoch": 0.02055685198710298, "grad_norm": 1.8735062989361686, "learning_rate": 1.368920521945433e-05, "loss": 0.5956, "step": 577 }, { "epoch": 0.020592479113597094, "grad_norm": 2.7557236159796292, "learning_rate": 1.3712930011862397e-05, "loss": 0.6571, "step": 578 }, { "epoch": 0.020628106240091204, "grad_norm": 2.078331833760456, "learning_rate": 1.3736654804270464e-05, "loss": 0.5886, "step": 579 }, { "epoch": 0.020663733366585318, "grad_norm": 1.968465680384235, "learning_rate": 1.376037959667853e-05, "loss": 0.5274, "step": 580 }, { "epoch": 0.02069936049307943, "grad_norm": 1.971148602047193, "learning_rate": 1.3784104389086597e-05, "loss": 0.5657, "step": 581 }, { "epoch": 0.02073498761957354, "grad_norm": 2.7859028221917335, "learning_rate": 1.3807829181494662e-05, "loss": 0.5343, "step": 582 }, { "epoch": 0.020770614746067655, "grad_norm": 1.7963720643991052, "learning_rate": 1.383155397390273e-05, "loss": 0.5467, "step": 583 }, { "epoch": 0.02080624187256177, "grad_norm": 1.7573846989626043, "learning_rate": 1.3855278766310796e-05, "loss": 0.5706, "step": 584 }, { "epoch": 0.020841868999055883, "grad_norm": 1.6919341855874956, "learning_rate": 1.3879003558718862e-05, "loss": 0.5643, "step": 585 }, { "epoch": 0.020877496125549993, "grad_norm": 3.072698234207706, "learning_rate": 1.3902728351126929e-05, "loss": 0.6732, "step": 586 }, { "epoch": 0.020913123252044107, "grad_norm": 2.7982659853837677, "learning_rate": 1.3926453143534994e-05, "loss": 0.7211, "step": 587 }, { "epoch": 0.02094875037853822, "grad_norm": 3.573691379544987, "learning_rate": 1.3950177935943061e-05, "loss": 0.6211, "step": 588 }, { "epoch": 0.02098437750503233, "grad_norm": 2.1582991847801862, "learning_rate": 1.3973902728351128e-05, "loss": 0.5877, "step": 589 }, { "epoch": 0.021020004631526444, "grad_norm": 2.4164313407964095, "learning_rate": 1.3997627520759193e-05, "loss": 0.7909, "step": 590 }, { "epoch": 0.021055631758020558, "grad_norm": 2.5561757446229985, "learning_rate": 1.402135231316726e-05, "loss": 0.5666, "step": 591 }, { "epoch": 0.021091258884514668, "grad_norm": 1.9989409459703777, "learning_rate": 1.4045077105575326e-05, "loss": 0.5619, "step": 592 }, { "epoch": 0.02112688601100878, "grad_norm": 2.818210364684643, "learning_rate": 1.4068801897983393e-05, "loss": 0.5904, "step": 593 }, { "epoch": 0.021162513137502895, "grad_norm": 1.9195902146540915, "learning_rate": 1.409252669039146e-05, "loss": 0.5301, "step": 594 }, { "epoch": 0.02119814026399701, "grad_norm": 2.021512675334311, "learning_rate": 1.4116251482799525e-05, "loss": 0.588, "step": 595 }, { "epoch": 0.02123376739049112, "grad_norm": 2.466689222306675, "learning_rate": 1.4139976275207594e-05, "loss": 0.5013, "step": 596 }, { "epoch": 0.021269394516985233, "grad_norm": 2.7438524893873217, "learning_rate": 1.4163701067615661e-05, "loss": 0.6095, "step": 597 }, { "epoch": 0.021305021643479347, "grad_norm": 4.967611938830713, "learning_rate": 1.4187425860023726e-05, "loss": 0.6856, "step": 598 }, { "epoch": 0.021340648769973457, "grad_norm": 1.853877718277599, "learning_rate": 1.4211150652431793e-05, "loss": 0.5827, "step": 599 }, { "epoch": 0.02137627589646757, "grad_norm": 2.6764997496154086, "learning_rate": 1.423487544483986e-05, "loss": 0.6189, "step": 600 }, { "epoch": 0.021411903022961684, "grad_norm": 3.407992117416997, "learning_rate": 1.4258600237247926e-05, "loss": 0.7175, "step": 601 }, { "epoch": 0.021447530149455794, "grad_norm": 2.432832235102616, "learning_rate": 1.4282325029655993e-05, "loss": 0.6092, "step": 602 }, { "epoch": 0.021483157275949908, "grad_norm": 2.0744565181471573, "learning_rate": 1.4306049822064058e-05, "loss": 0.5507, "step": 603 }, { "epoch": 0.02151878440244402, "grad_norm": 2.0030979025210316, "learning_rate": 1.4329774614472125e-05, "loss": 0.6097, "step": 604 }, { "epoch": 0.021554411528938132, "grad_norm": 2.8395521229543075, "learning_rate": 1.4353499406880192e-05, "loss": 0.698, "step": 605 }, { "epoch": 0.021590038655432246, "grad_norm": 1.8185709746348557, "learning_rate": 1.4377224199288258e-05, "loss": 0.5488, "step": 606 }, { "epoch": 0.02162566578192636, "grad_norm": 3.3530557025372127, "learning_rate": 1.4400948991696325e-05, "loss": 0.6671, "step": 607 }, { "epoch": 0.021661292908420473, "grad_norm": 1.7996798277794028, "learning_rate": 1.442467378410439e-05, "loss": 0.5795, "step": 608 }, { "epoch": 0.021696920034914583, "grad_norm": 2.771577410845119, "learning_rate": 1.4448398576512457e-05, "loss": 0.6881, "step": 609 }, { "epoch": 0.021732547161408697, "grad_norm": 2.1777174850826495, "learning_rate": 1.4472123368920522e-05, "loss": 0.5907, "step": 610 }, { "epoch": 0.02176817428790281, "grad_norm": 3.626410427129873, "learning_rate": 1.449584816132859e-05, "loss": 0.68, "step": 611 }, { "epoch": 0.02180380141439692, "grad_norm": 3.937168196874977, "learning_rate": 1.4519572953736656e-05, "loss": 0.707, "step": 612 }, { "epoch": 0.021839428540891034, "grad_norm": 1.7925682576716455, "learning_rate": 1.4543297746144722e-05, "loss": 0.591, "step": 613 }, { "epoch": 0.021875055667385148, "grad_norm": 2.119970301403296, "learning_rate": 1.4567022538552789e-05, "loss": 0.6188, "step": 614 }, { "epoch": 0.021910682793879258, "grad_norm": 2.1891135522586134, "learning_rate": 1.4590747330960854e-05, "loss": 0.6484, "step": 615 }, { "epoch": 0.021946309920373372, "grad_norm": 1.8583675191689495, "learning_rate": 1.4614472123368921e-05, "loss": 0.5763, "step": 616 }, { "epoch": 0.021981937046867486, "grad_norm": 2.417254464639028, "learning_rate": 1.4638196915776988e-05, "loss": 0.6502, "step": 617 }, { "epoch": 0.022017564173361596, "grad_norm": 1.8783839001092943, "learning_rate": 1.4661921708185054e-05, "loss": 0.5864, "step": 618 }, { "epoch": 0.02205319129985571, "grad_norm": 1.8717616288202574, "learning_rate": 1.468564650059312e-05, "loss": 0.6005, "step": 619 }, { "epoch": 0.022088818426349823, "grad_norm": 2.766650055196217, "learning_rate": 1.4709371293001186e-05, "loss": 0.6572, "step": 620 }, { "epoch": 0.022124445552843937, "grad_norm": 2.8806912132195874, "learning_rate": 1.4733096085409253e-05, "loss": 0.5647, "step": 621 }, { "epoch": 0.022160072679338047, "grad_norm": 1.7432962866374775, "learning_rate": 1.475682087781732e-05, "loss": 0.556, "step": 622 }, { "epoch": 0.02219569980583216, "grad_norm": 1.9253083519080445, "learning_rate": 1.4780545670225385e-05, "loss": 0.5569, "step": 623 }, { "epoch": 0.022231326932326274, "grad_norm": 1.8931349598628016, "learning_rate": 1.4804270462633452e-05, "loss": 0.6153, "step": 624 }, { "epoch": 0.022266954058820385, "grad_norm": 1.920582079819275, "learning_rate": 1.4827995255041521e-05, "loss": 0.5497, "step": 625 }, { "epoch": 0.022302581185314498, "grad_norm": 1.8191265962408472, "learning_rate": 1.4851720047449586e-05, "loss": 0.546, "step": 626 }, { "epoch": 0.022338208311808612, "grad_norm": 1.772746739540775, "learning_rate": 1.4875444839857654e-05, "loss": 0.6152, "step": 627 }, { "epoch": 0.022373835438302722, "grad_norm": 1.7383683298671693, "learning_rate": 1.489916963226572e-05, "loss": 0.594, "step": 628 }, { "epoch": 0.022409462564796836, "grad_norm": 2.2891607716769062, "learning_rate": 1.4922894424673786e-05, "loss": 0.6936, "step": 629 }, { "epoch": 0.02244508969129095, "grad_norm": 3.775576592664761, "learning_rate": 1.4946619217081853e-05, "loss": 0.6675, "step": 630 }, { "epoch": 0.022480716817785063, "grad_norm": 1.9336252912673662, "learning_rate": 1.4970344009489918e-05, "loss": 0.6226, "step": 631 }, { "epoch": 0.022516343944279173, "grad_norm": 1.8188802211208879, "learning_rate": 1.4994068801897985e-05, "loss": 0.5902, "step": 632 }, { "epoch": 0.022551971070773287, "grad_norm": 1.708678859774494, "learning_rate": 1.5017793594306052e-05, "loss": 0.5863, "step": 633 }, { "epoch": 0.0225875981972674, "grad_norm": 2.3403968247037508, "learning_rate": 1.5041518386714118e-05, "loss": 0.5693, "step": 634 }, { "epoch": 0.02262322532376151, "grad_norm": 3.836015821185222, "learning_rate": 1.5065243179122185e-05, "loss": 0.6712, "step": 635 }, { "epoch": 0.022658852450255625, "grad_norm": 3.301506741097682, "learning_rate": 1.508896797153025e-05, "loss": 0.7806, "step": 636 }, { "epoch": 0.02269447957674974, "grad_norm": 1.9823672958901741, "learning_rate": 1.5112692763938317e-05, "loss": 0.5543, "step": 637 }, { "epoch": 0.02273010670324385, "grad_norm": 3.108004782351458, "learning_rate": 1.5136417556346384e-05, "loss": 0.6864, "step": 638 }, { "epoch": 0.022765733829737962, "grad_norm": 1.9834234513870252, "learning_rate": 1.516014234875445e-05, "loss": 0.598, "step": 639 }, { "epoch": 0.022801360956232076, "grad_norm": 2.737232781271813, "learning_rate": 1.5183867141162517e-05, "loss": 0.5781, "step": 640 }, { "epoch": 0.022836988082726186, "grad_norm": 2.7563359097474085, "learning_rate": 1.5207591933570582e-05, "loss": 0.7307, "step": 641 }, { "epoch": 0.0228726152092203, "grad_norm": 1.8417109893264187, "learning_rate": 1.5231316725978649e-05, "loss": 0.5605, "step": 642 }, { "epoch": 0.022908242335714413, "grad_norm": 2.499194009710437, "learning_rate": 1.5255041518386714e-05, "loss": 0.6339, "step": 643 }, { "epoch": 0.022943869462208527, "grad_norm": 2.883864968910172, "learning_rate": 1.5278766310794783e-05, "loss": 0.6921, "step": 644 }, { "epoch": 0.022979496588702637, "grad_norm": 3.059642268912012, "learning_rate": 1.5302491103202847e-05, "loss": 0.6808, "step": 645 }, { "epoch": 0.02301512371519675, "grad_norm": 1.9923765589510096, "learning_rate": 1.5326215895610914e-05, "loss": 0.5793, "step": 646 }, { "epoch": 0.023050750841690865, "grad_norm": 2.5904169809946445, "learning_rate": 1.534994068801898e-05, "loss": 0.6715, "step": 647 }, { "epoch": 0.023086377968184975, "grad_norm": 2.117937511715003, "learning_rate": 1.5373665480427048e-05, "loss": 0.5295, "step": 648 }, { "epoch": 0.02312200509467909, "grad_norm": 1.9318272773554712, "learning_rate": 1.5397390272835115e-05, "loss": 0.5871, "step": 649 }, { "epoch": 0.023157632221173202, "grad_norm": 1.900019622573953, "learning_rate": 1.542111506524318e-05, "loss": 0.5688, "step": 650 }, { "epoch": 0.023193259347667312, "grad_norm": 2.684799277199338, "learning_rate": 1.5444839857651245e-05, "loss": 0.7116, "step": 651 }, { "epoch": 0.023228886474161426, "grad_norm": 2.5861053232540896, "learning_rate": 1.5468564650059313e-05, "loss": 0.64, "step": 652 }, { "epoch": 0.02326451360065554, "grad_norm": 2.0589149537504334, "learning_rate": 1.549228944246738e-05, "loss": 0.5799, "step": 653 }, { "epoch": 0.023300140727149653, "grad_norm": 2.8331650722256083, "learning_rate": 1.5516014234875447e-05, "loss": 0.6389, "step": 654 }, { "epoch": 0.023335767853643764, "grad_norm": 1.760336962449126, "learning_rate": 1.5539739027283514e-05, "loss": 0.5653, "step": 655 }, { "epoch": 0.023371394980137877, "grad_norm": 1.8010573555451779, "learning_rate": 1.556346381969158e-05, "loss": 0.5705, "step": 656 }, { "epoch": 0.02340702210663199, "grad_norm": 3.182779474827196, "learning_rate": 1.5587188612099648e-05, "loss": 0.6079, "step": 657 }, { "epoch": 0.0234426492331261, "grad_norm": 1.9262982162482207, "learning_rate": 1.561091340450771e-05, "loss": 0.5606, "step": 658 }, { "epoch": 0.023478276359620215, "grad_norm": 2.4074222426827157, "learning_rate": 1.563463819691578e-05, "loss": 0.6067, "step": 659 }, { "epoch": 0.02351390348611433, "grad_norm": 1.9838622328024667, "learning_rate": 1.5658362989323845e-05, "loss": 0.5511, "step": 660 }, { "epoch": 0.02354953061260844, "grad_norm": 2.588782301410212, "learning_rate": 1.5682087781731912e-05, "loss": 0.6595, "step": 661 }, { "epoch": 0.023585157739102552, "grad_norm": 2.288091146471363, "learning_rate": 1.570581257413998e-05, "loss": 0.6915, "step": 662 }, { "epoch": 0.023620784865596666, "grad_norm": 2.502946031634256, "learning_rate": 1.5729537366548043e-05, "loss": 0.716, "step": 663 }, { "epoch": 0.023656411992090776, "grad_norm": 1.992900846894705, "learning_rate": 1.575326215895611e-05, "loss": 0.5718, "step": 664 }, { "epoch": 0.02369203911858489, "grad_norm": 1.9839899753961285, "learning_rate": 1.5776986951364177e-05, "loss": 0.6595, "step": 665 }, { "epoch": 0.023727666245079004, "grad_norm": 1.9227774515173126, "learning_rate": 1.5800711743772244e-05, "loss": 0.567, "step": 666 }, { "epoch": 0.023763293371573117, "grad_norm": 1.8575224937325139, "learning_rate": 1.582443653618031e-05, "loss": 0.5922, "step": 667 }, { "epoch": 0.023798920498067228, "grad_norm": 2.4583719423059547, "learning_rate": 1.5848161328588375e-05, "loss": 0.6554, "step": 668 }, { "epoch": 0.02383454762456134, "grad_norm": 3.0514362925888783, "learning_rate": 1.5871886120996442e-05, "loss": 0.6894, "step": 669 }, { "epoch": 0.023870174751055455, "grad_norm": 1.9196612208377686, "learning_rate": 1.589561091340451e-05, "loss": 0.5493, "step": 670 }, { "epoch": 0.023905801877549565, "grad_norm": 3.2689845125971653, "learning_rate": 1.5919335705812576e-05, "loss": 0.6918, "step": 671 }, { "epoch": 0.02394142900404368, "grad_norm": 2.3452407259934582, "learning_rate": 1.5943060498220643e-05, "loss": 0.638, "step": 672 }, { "epoch": 0.023977056130537792, "grad_norm": 2.530006637045938, "learning_rate": 1.5966785290628707e-05, "loss": 0.6794, "step": 673 }, { "epoch": 0.024012683257031903, "grad_norm": 1.817888217180465, "learning_rate": 1.5990510083036774e-05, "loss": 0.5433, "step": 674 }, { "epoch": 0.024048310383526016, "grad_norm": 1.8429137131053848, "learning_rate": 1.601423487544484e-05, "loss": 0.5698, "step": 675 }, { "epoch": 0.02408393751002013, "grad_norm": 2.7093257518469027, "learning_rate": 1.6037959667852908e-05, "loss": 0.7644, "step": 676 }, { "epoch": 0.024119564636514244, "grad_norm": 2.5169436918586765, "learning_rate": 1.6061684460260975e-05, "loss": 0.5733, "step": 677 }, { "epoch": 0.024155191763008354, "grad_norm": 1.6565454773642452, "learning_rate": 1.608540925266904e-05, "loss": 0.5867, "step": 678 }, { "epoch": 0.024190818889502468, "grad_norm": 1.8454792651288479, "learning_rate": 1.6109134045077106e-05, "loss": 0.5991, "step": 679 }, { "epoch": 0.02422644601599658, "grad_norm": 1.7518097758206954, "learning_rate": 1.6132858837485173e-05, "loss": 0.5549, "step": 680 }, { "epoch": 0.02426207314249069, "grad_norm": 2.6546811918851576, "learning_rate": 1.615658362989324e-05, "loss": 0.6691, "step": 681 }, { "epoch": 0.024297700268984805, "grad_norm": 1.7063671819282618, "learning_rate": 1.6180308422301307e-05, "loss": 0.556, "step": 682 }, { "epoch": 0.02433332739547892, "grad_norm": 3.0227738744768495, "learning_rate": 1.620403321470937e-05, "loss": 0.7331, "step": 683 }, { "epoch": 0.02436895452197303, "grad_norm": 2.0439490344752835, "learning_rate": 1.622775800711744e-05, "loss": 0.5665, "step": 684 }, { "epoch": 0.024404581648467143, "grad_norm": 2.4866468631556105, "learning_rate": 1.6251482799525508e-05, "loss": 0.6027, "step": 685 }, { "epoch": 0.024440208774961256, "grad_norm": 2.34293043534793, "learning_rate": 1.627520759193357e-05, "loss": 0.6226, "step": 686 }, { "epoch": 0.024475835901455367, "grad_norm": 1.7073548890298644, "learning_rate": 1.629893238434164e-05, "loss": 0.5302, "step": 687 }, { "epoch": 0.02451146302794948, "grad_norm": 2.033701633487404, "learning_rate": 1.6322657176749706e-05, "loss": 0.5584, "step": 688 }, { "epoch": 0.024547090154443594, "grad_norm": 1.8048878467137563, "learning_rate": 1.6346381969157773e-05, "loss": 0.5617, "step": 689 }, { "epoch": 0.024582717280937708, "grad_norm": 1.6629616630308879, "learning_rate": 1.637010676156584e-05, "loss": 0.5937, "step": 690 }, { "epoch": 0.024618344407431818, "grad_norm": 1.6798593881774606, "learning_rate": 1.6393831553973903e-05, "loss": 0.6002, "step": 691 }, { "epoch": 0.02465397153392593, "grad_norm": 1.8011080578979153, "learning_rate": 1.641755634638197e-05, "loss": 0.5831, "step": 692 }, { "epoch": 0.024689598660420045, "grad_norm": 1.897325035941249, "learning_rate": 1.6441281138790037e-05, "loss": 0.5163, "step": 693 }, { "epoch": 0.024725225786914155, "grad_norm": 2.5688743708066917, "learning_rate": 1.6465005931198104e-05, "loss": 0.706, "step": 694 }, { "epoch": 0.02476085291340827, "grad_norm": 2.9773678615363894, "learning_rate": 1.648873072360617e-05, "loss": 0.6769, "step": 695 }, { "epoch": 0.024796480039902383, "grad_norm": 2.471001756543344, "learning_rate": 1.6512455516014235e-05, "loss": 0.7458, "step": 696 }, { "epoch": 0.024832107166396493, "grad_norm": 1.7408326587998155, "learning_rate": 1.6536180308422302e-05, "loss": 0.5602, "step": 697 }, { "epoch": 0.024867734292890607, "grad_norm": 1.784384937573898, "learning_rate": 1.655990510083037e-05, "loss": 0.5999, "step": 698 }, { "epoch": 0.02490336141938472, "grad_norm": 2.704082223919804, "learning_rate": 1.6583629893238436e-05, "loss": 0.7294, "step": 699 }, { "epoch": 0.02493898854587883, "grad_norm": 1.685734254823008, "learning_rate": 1.6607354685646503e-05, "loss": 0.5454, "step": 700 }, { "epoch": 0.024974615672372944, "grad_norm": 1.6884511723273916, "learning_rate": 1.6631079478054567e-05, "loss": 0.5501, "step": 701 }, { "epoch": 0.025010242798867058, "grad_norm": 3.294320975925651, "learning_rate": 1.6654804270462634e-05, "loss": 0.7213, "step": 702 }, { "epoch": 0.02504586992536117, "grad_norm": 1.8730699665259838, "learning_rate": 1.66785290628707e-05, "loss": 0.5434, "step": 703 }, { "epoch": 0.02508149705185528, "grad_norm": 2.2381868865014645, "learning_rate": 1.6702253855278768e-05, "loss": 0.716, "step": 704 }, { "epoch": 0.025117124178349395, "grad_norm": 1.7599659641506493, "learning_rate": 1.6725978647686835e-05, "loss": 0.5222, "step": 705 }, { "epoch": 0.02515275130484351, "grad_norm": 1.7554811479550123, "learning_rate": 1.67497034400949e-05, "loss": 0.5757, "step": 706 }, { "epoch": 0.02518837843133762, "grad_norm": 2.8897906900946357, "learning_rate": 1.6773428232502966e-05, "loss": 0.6685, "step": 707 }, { "epoch": 0.025224005557831733, "grad_norm": 2.3769998728833284, "learning_rate": 1.6797153024911033e-05, "loss": 0.699, "step": 708 }, { "epoch": 0.025259632684325847, "grad_norm": 3.1327524162978175, "learning_rate": 1.68208778173191e-05, "loss": 0.6491, "step": 709 }, { "epoch": 0.025295259810819957, "grad_norm": 2.756884084979343, "learning_rate": 1.6844602609727167e-05, "loss": 0.6481, "step": 710 }, { "epoch": 0.02533088693731407, "grad_norm": 1.8156162394639168, "learning_rate": 1.686832740213523e-05, "loss": 0.5652, "step": 711 }, { "epoch": 0.025366514063808184, "grad_norm": 1.862936558842172, "learning_rate": 1.6892052194543298e-05, "loss": 0.5859, "step": 712 }, { "epoch": 0.025402141190302298, "grad_norm": 2.6318517198884965, "learning_rate": 1.6915776986951368e-05, "loss": 0.6071, "step": 713 }, { "epoch": 0.025437768316796408, "grad_norm": 1.6240788393545396, "learning_rate": 1.693950177935943e-05, "loss": 0.5646, "step": 714 }, { "epoch": 0.02547339544329052, "grad_norm": 2.3089740583696146, "learning_rate": 1.69632265717675e-05, "loss": 0.6584, "step": 715 }, { "epoch": 0.025509022569784635, "grad_norm": 2.261068304545678, "learning_rate": 1.6986951364175566e-05, "loss": 0.6141, "step": 716 }, { "epoch": 0.025544649696278746, "grad_norm": 1.8189251765809658, "learning_rate": 1.7010676156583633e-05, "loss": 0.6269, "step": 717 }, { "epoch": 0.02558027682277286, "grad_norm": 2.7873509371004612, "learning_rate": 1.70344009489917e-05, "loss": 0.6855, "step": 718 }, { "epoch": 0.025615903949266973, "grad_norm": 1.8844452271053957, "learning_rate": 1.7058125741399763e-05, "loss": 0.6319, "step": 719 }, { "epoch": 0.025651531075761083, "grad_norm": 2.9420365395535626, "learning_rate": 1.708185053380783e-05, "loss": 0.6261, "step": 720 }, { "epoch": 0.025687158202255197, "grad_norm": 2.4866192674116525, "learning_rate": 1.7105575326215898e-05, "loss": 0.6934, "step": 721 }, { "epoch": 0.02572278532874931, "grad_norm": 2.740172078432155, "learning_rate": 1.7129300118623965e-05, "loss": 0.6518, "step": 722 }, { "epoch": 0.02575841245524342, "grad_norm": 1.7812533797634311, "learning_rate": 1.715302491103203e-05, "loss": 0.6049, "step": 723 }, { "epoch": 0.025794039581737534, "grad_norm": 1.9303473644405622, "learning_rate": 1.7176749703440095e-05, "loss": 0.5742, "step": 724 }, { "epoch": 0.025829666708231648, "grad_norm": 1.6387776967960364, "learning_rate": 1.7200474495848162e-05, "loss": 0.5159, "step": 725 }, { "epoch": 0.025865293834725762, "grad_norm": 1.756124391426539, "learning_rate": 1.722419928825623e-05, "loss": 0.6059, "step": 726 }, { "epoch": 0.025900920961219872, "grad_norm": 1.9178042124423926, "learning_rate": 1.7247924080664296e-05, "loss": 0.5495, "step": 727 }, { "epoch": 0.025936548087713986, "grad_norm": 2.6945790717491658, "learning_rate": 1.7271648873072363e-05, "loss": 0.6515, "step": 728 }, { "epoch": 0.0259721752142081, "grad_norm": 3.0566673055131823, "learning_rate": 1.7295373665480427e-05, "loss": 0.7616, "step": 729 }, { "epoch": 0.02600780234070221, "grad_norm": 1.9345572671053541, "learning_rate": 1.7319098457888494e-05, "loss": 0.5748, "step": 730 }, { "epoch": 0.026043429467196323, "grad_norm": 1.822752922311343, "learning_rate": 1.734282325029656e-05, "loss": 0.5898, "step": 731 }, { "epoch": 0.026079056593690437, "grad_norm": 1.5183698786579494, "learning_rate": 1.7366548042704628e-05, "loss": 0.5663, "step": 732 }, { "epoch": 0.026114683720184547, "grad_norm": 1.6609404799353449, "learning_rate": 1.7390272835112695e-05, "loss": 0.617, "step": 733 }, { "epoch": 0.02615031084667866, "grad_norm": 2.38471060596249, "learning_rate": 1.741399762752076e-05, "loss": 0.6957, "step": 734 }, { "epoch": 0.026185937973172774, "grad_norm": 2.222036074977046, "learning_rate": 1.7437722419928826e-05, "loss": 0.6571, "step": 735 }, { "epoch": 0.026221565099666888, "grad_norm": 2.7614836903834927, "learning_rate": 1.7461447212336893e-05, "loss": 0.6883, "step": 736 }, { "epoch": 0.026257192226161, "grad_norm": 1.8154270800077923, "learning_rate": 1.748517200474496e-05, "loss": 0.5783, "step": 737 }, { "epoch": 0.026292819352655112, "grad_norm": 1.7597316310966082, "learning_rate": 1.7508896797153027e-05, "loss": 0.5166, "step": 738 }, { "epoch": 0.026328446479149226, "grad_norm": 3.0823525565533805, "learning_rate": 1.753262158956109e-05, "loss": 0.696, "step": 739 }, { "epoch": 0.026364073605643336, "grad_norm": 1.6381563288819467, "learning_rate": 1.7556346381969158e-05, "loss": 0.5421, "step": 740 }, { "epoch": 0.02639970073213745, "grad_norm": 1.8034211314092798, "learning_rate": 1.7580071174377225e-05, "loss": 0.5414, "step": 741 }, { "epoch": 0.026435327858631563, "grad_norm": 1.9034468652455203, "learning_rate": 1.7603795966785292e-05, "loss": 0.5499, "step": 742 }, { "epoch": 0.026470954985125673, "grad_norm": 1.8008232613099602, "learning_rate": 1.762752075919336e-05, "loss": 0.5408, "step": 743 }, { "epoch": 0.026506582111619787, "grad_norm": 3.0620979338194503, "learning_rate": 1.7651245551601426e-05, "loss": 0.6432, "step": 744 }, { "epoch": 0.0265422092381139, "grad_norm": 2.3800448108591405, "learning_rate": 1.7674970344009493e-05, "loss": 0.6541, "step": 745 }, { "epoch": 0.02657783636460801, "grad_norm": 1.8806970583582159, "learning_rate": 1.769869513641756e-05, "loss": 0.6077, "step": 746 }, { "epoch": 0.026613463491102125, "grad_norm": 2.516908207239252, "learning_rate": 1.7722419928825624e-05, "loss": 0.6939, "step": 747 }, { "epoch": 0.02664909061759624, "grad_norm": 1.8980128739802409, "learning_rate": 1.774614472123369e-05, "loss": 0.5191, "step": 748 }, { "epoch": 0.026684717744090352, "grad_norm": 1.8484824724221665, "learning_rate": 1.7769869513641758e-05, "loss": 0.6606, "step": 749 }, { "epoch": 0.026720344870584462, "grad_norm": 1.7777363743878323, "learning_rate": 1.7793594306049825e-05, "loss": 0.5679, "step": 750 }, { "epoch": 0.026755971997078576, "grad_norm": 2.333219250871948, "learning_rate": 1.7817319098457892e-05, "loss": 0.7557, "step": 751 }, { "epoch": 0.02679159912357269, "grad_norm": 1.8093371225156587, "learning_rate": 1.7841043890865955e-05, "loss": 0.5506, "step": 752 }, { "epoch": 0.0268272262500668, "grad_norm": 2.900828175187249, "learning_rate": 1.7864768683274022e-05, "loss": 0.6472, "step": 753 }, { "epoch": 0.026862853376560913, "grad_norm": 1.7653789608069212, "learning_rate": 1.788849347568209e-05, "loss": 0.6149, "step": 754 }, { "epoch": 0.026898480503055027, "grad_norm": 2.586913775592621, "learning_rate": 1.7912218268090156e-05, "loss": 0.7245, "step": 755 }, { "epoch": 0.026934107629549137, "grad_norm": 1.813389113351238, "learning_rate": 1.7935943060498224e-05, "loss": 0.6151, "step": 756 }, { "epoch": 0.02696973475604325, "grad_norm": 2.127014716615834, "learning_rate": 1.7959667852906287e-05, "loss": 0.644, "step": 757 }, { "epoch": 0.027005361882537365, "grad_norm": 3.1247704742985327, "learning_rate": 1.7983392645314354e-05, "loss": 0.666, "step": 758 }, { "epoch": 0.027040989009031475, "grad_norm": 1.912379743104086, "learning_rate": 1.800711743772242e-05, "loss": 0.5897, "step": 759 }, { "epoch": 0.02707661613552559, "grad_norm": 3.3080083338753865, "learning_rate": 1.8030842230130488e-05, "loss": 0.6036, "step": 760 }, { "epoch": 0.027112243262019702, "grad_norm": 2.7739679169547125, "learning_rate": 1.8054567022538555e-05, "loss": 0.6103, "step": 761 }, { "epoch": 0.027147870388513816, "grad_norm": 2.4791348446653094, "learning_rate": 1.807829181494662e-05, "loss": 0.6814, "step": 762 }, { "epoch": 0.027183497515007926, "grad_norm": 1.7740844406337661, "learning_rate": 1.8102016607354686e-05, "loss": 0.6033, "step": 763 }, { "epoch": 0.02721912464150204, "grad_norm": 2.6623473504322894, "learning_rate": 1.8125741399762753e-05, "loss": 0.7146, "step": 764 }, { "epoch": 0.027254751767996153, "grad_norm": 1.638871101472601, "learning_rate": 1.814946619217082e-05, "loss": 0.5048, "step": 765 }, { "epoch": 0.027290378894490264, "grad_norm": 1.7384896502508203, "learning_rate": 1.8173190984578887e-05, "loss": 0.5184, "step": 766 }, { "epoch": 0.027326006020984377, "grad_norm": 1.5839210230501646, "learning_rate": 1.819691577698695e-05, "loss": 0.5894, "step": 767 }, { "epoch": 0.02736163314747849, "grad_norm": 2.0361766013354625, "learning_rate": 1.8220640569395018e-05, "loss": 0.5641, "step": 768 }, { "epoch": 0.0273972602739726, "grad_norm": 1.7420084151589008, "learning_rate": 1.8244365361803085e-05, "loss": 0.5405, "step": 769 }, { "epoch": 0.027432887400466715, "grad_norm": 1.6119599375311908, "learning_rate": 1.8268090154211152e-05, "loss": 0.526, "step": 770 }, { "epoch": 0.02746851452696083, "grad_norm": 2.8720257113656245, "learning_rate": 1.829181494661922e-05, "loss": 0.7003, "step": 771 }, { "epoch": 0.027504141653454942, "grad_norm": 1.8076958401196774, "learning_rate": 1.8315539739027286e-05, "loss": 0.5611, "step": 772 }, { "epoch": 0.027539768779949052, "grad_norm": 1.607391567702599, "learning_rate": 1.8339264531435353e-05, "loss": 0.5396, "step": 773 }, { "epoch": 0.027575395906443166, "grad_norm": 2.6009652096555875, "learning_rate": 1.836298932384342e-05, "loss": 0.7346, "step": 774 }, { "epoch": 0.02761102303293728, "grad_norm": 1.8453813097400518, "learning_rate": 1.8386714116251484e-05, "loss": 0.5487, "step": 775 }, { "epoch": 0.02764665015943139, "grad_norm": 2.9191460458093776, "learning_rate": 1.841043890865955e-05, "loss": 0.6803, "step": 776 }, { "epoch": 0.027682277285925504, "grad_norm": 1.7318805840696345, "learning_rate": 1.8434163701067618e-05, "loss": 0.4906, "step": 777 }, { "epoch": 0.027717904412419617, "grad_norm": 2.644441867167321, "learning_rate": 1.8457888493475685e-05, "loss": 0.6859, "step": 778 }, { "epoch": 0.027753531538913728, "grad_norm": 2.6684123916638858, "learning_rate": 1.8481613285883752e-05, "loss": 0.6293, "step": 779 }, { "epoch": 0.02778915866540784, "grad_norm": 2.8854431909312117, "learning_rate": 1.8505338078291815e-05, "loss": 0.7212, "step": 780 }, { "epoch": 0.027824785791901955, "grad_norm": 1.785839649088226, "learning_rate": 1.8529062870699883e-05, "loss": 0.5847, "step": 781 }, { "epoch": 0.027860412918396065, "grad_norm": 2.3951066234102107, "learning_rate": 1.855278766310795e-05, "loss": 0.5808, "step": 782 }, { "epoch": 0.02789604004489018, "grad_norm": 4.573192497915985, "learning_rate": 1.8576512455516017e-05, "loss": 0.6866, "step": 783 }, { "epoch": 0.027931667171384292, "grad_norm": 2.2918930153697015, "learning_rate": 1.8600237247924084e-05, "loss": 0.5621, "step": 784 }, { "epoch": 0.027967294297878406, "grad_norm": 2.71073685370842, "learning_rate": 1.8623962040332147e-05, "loss": 0.6734, "step": 785 }, { "epoch": 0.028002921424372516, "grad_norm": 2.3351118050846424, "learning_rate": 1.8647686832740214e-05, "loss": 0.6736, "step": 786 }, { "epoch": 0.02803854855086663, "grad_norm": 1.8564229536452956, "learning_rate": 1.867141162514828e-05, "loss": 0.569, "step": 787 }, { "epoch": 0.028074175677360744, "grad_norm": 1.6625058614508417, "learning_rate": 1.869513641755635e-05, "loss": 0.5571, "step": 788 }, { "epoch": 0.028109802803854854, "grad_norm": 2.064997928491524, "learning_rate": 1.8718861209964415e-05, "loss": 0.6657, "step": 789 }, { "epoch": 0.028145429930348968, "grad_norm": 2.4007565604656764, "learning_rate": 1.874258600237248e-05, "loss": 0.6247, "step": 790 }, { "epoch": 0.02818105705684308, "grad_norm": 1.6256741811628967, "learning_rate": 1.8766310794780546e-05, "loss": 0.5233, "step": 791 }, { "epoch": 0.02821668418333719, "grad_norm": 2.8364021344897004, "learning_rate": 1.8790035587188613e-05, "loss": 0.6453, "step": 792 }, { "epoch": 0.028252311309831305, "grad_norm": 2.9343297037952634, "learning_rate": 1.881376037959668e-05, "loss": 0.6144, "step": 793 }, { "epoch": 0.02828793843632542, "grad_norm": 1.748168882343687, "learning_rate": 1.8837485172004747e-05, "loss": 0.6052, "step": 794 }, { "epoch": 0.028323565562819532, "grad_norm": 1.6943358995003115, "learning_rate": 1.886120996441281e-05, "loss": 0.594, "step": 795 }, { "epoch": 0.028359192689313643, "grad_norm": 1.7636559231257418, "learning_rate": 1.8884934756820878e-05, "loss": 0.545, "step": 796 }, { "epoch": 0.028394819815807756, "grad_norm": 3.3152695559276135, "learning_rate": 1.8908659549228945e-05, "loss": 0.6782, "step": 797 }, { "epoch": 0.02843044694230187, "grad_norm": 1.8995752465055347, "learning_rate": 1.8932384341637012e-05, "loss": 0.5469, "step": 798 }, { "epoch": 0.02846607406879598, "grad_norm": 1.6881930699412688, "learning_rate": 1.895610913404508e-05, "loss": 0.6134, "step": 799 }, { "epoch": 0.028501701195290094, "grad_norm": 1.5152530607026207, "learning_rate": 1.8979833926453146e-05, "loss": 0.5986, "step": 800 }, { "epoch": 0.028537328321784208, "grad_norm": 1.8895350149816823, "learning_rate": 1.9003558718861213e-05, "loss": 0.5809, "step": 801 }, { "epoch": 0.028572955448278318, "grad_norm": 2.546324805757951, "learning_rate": 1.902728351126928e-05, "loss": 0.6452, "step": 802 }, { "epoch": 0.02860858257477243, "grad_norm": 1.6524280112172407, "learning_rate": 1.9051008303677344e-05, "loss": 0.5722, "step": 803 }, { "epoch": 0.028644209701266545, "grad_norm": 1.7269441805348347, "learning_rate": 1.907473309608541e-05, "loss": 0.575, "step": 804 }, { "epoch": 0.028679836827760655, "grad_norm": 1.9980744384473257, "learning_rate": 1.9098457888493478e-05, "loss": 0.6439, "step": 805 }, { "epoch": 0.02871546395425477, "grad_norm": 1.7983681340043116, "learning_rate": 1.9122182680901545e-05, "loss": 0.5973, "step": 806 }, { "epoch": 0.028751091080748883, "grad_norm": 1.5859304997331098, "learning_rate": 1.9145907473309612e-05, "loss": 0.5625, "step": 807 }, { "epoch": 0.028786718207242996, "grad_norm": 1.701810447312772, "learning_rate": 1.9169632265717676e-05, "loss": 0.6128, "step": 808 }, { "epoch": 0.028822345333737107, "grad_norm": 2.5463696873910457, "learning_rate": 1.9193357058125743e-05, "loss": 0.7554, "step": 809 }, { "epoch": 0.02885797246023122, "grad_norm": 1.7688198087394453, "learning_rate": 1.921708185053381e-05, "loss": 0.549, "step": 810 }, { "epoch": 0.028893599586725334, "grad_norm": 2.8627267051466694, "learning_rate": 1.9240806642941877e-05, "loss": 0.7375, "step": 811 }, { "epoch": 0.028929226713219444, "grad_norm": 2.0841925739388443, "learning_rate": 1.9264531435349944e-05, "loss": 0.6699, "step": 812 }, { "epoch": 0.028964853839713558, "grad_norm": 3.0175149366401652, "learning_rate": 1.9288256227758007e-05, "loss": 0.6539, "step": 813 }, { "epoch": 0.02900048096620767, "grad_norm": 2.0767269266707102, "learning_rate": 1.9311981020166074e-05, "loss": 0.6384, "step": 814 }, { "epoch": 0.02903610809270178, "grad_norm": 2.622933532137662, "learning_rate": 1.933570581257414e-05, "loss": 0.6457, "step": 815 }, { "epoch": 0.029071735219195895, "grad_norm": 1.8182618415626306, "learning_rate": 1.935943060498221e-05, "loss": 0.6, "step": 816 }, { "epoch": 0.02910736234569001, "grad_norm": 1.6251064256719123, "learning_rate": 1.9383155397390276e-05, "loss": 0.5354, "step": 817 }, { "epoch": 0.02914298947218412, "grad_norm": 2.6040981280284, "learning_rate": 1.940688018979834e-05, "loss": 0.676, "step": 818 }, { "epoch": 0.029178616598678233, "grad_norm": 2.5202917271025598, "learning_rate": 1.9430604982206406e-05, "loss": 0.6881, "step": 819 }, { "epoch": 0.029214243725172347, "grad_norm": 1.7552054729582227, "learning_rate": 1.9454329774614473e-05, "loss": 0.5945, "step": 820 }, { "epoch": 0.02924987085166646, "grad_norm": 1.8766262612978535, "learning_rate": 1.947805456702254e-05, "loss": 0.562, "step": 821 }, { "epoch": 0.02928549797816057, "grad_norm": 2.1899410177475453, "learning_rate": 1.9501779359430607e-05, "loss": 0.6413, "step": 822 }, { "epoch": 0.029321125104654684, "grad_norm": 1.6708180024954493, "learning_rate": 1.952550415183867e-05, "loss": 0.5426, "step": 823 }, { "epoch": 0.029356752231148798, "grad_norm": 1.9710487040772, "learning_rate": 1.9549228944246738e-05, "loss": 0.6034, "step": 824 }, { "epoch": 0.029392379357642908, "grad_norm": 1.8691745241540632, "learning_rate": 1.9572953736654805e-05, "loss": 0.5599, "step": 825 }, { "epoch": 0.029428006484137022, "grad_norm": 1.5831188683308377, "learning_rate": 1.9596678529062872e-05, "loss": 0.5777, "step": 826 }, { "epoch": 0.029463633610631135, "grad_norm": 1.5882625347660493, "learning_rate": 1.962040332147094e-05, "loss": 0.5899, "step": 827 }, { "epoch": 0.029499260737125246, "grad_norm": 1.7552200153535145, "learning_rate": 1.9644128113879003e-05, "loss": 0.5387, "step": 828 }, { "epoch": 0.02953488786361936, "grad_norm": 2.244836095694088, "learning_rate": 1.9667852906287073e-05, "loss": 0.6919, "step": 829 }, { "epoch": 0.029570514990113473, "grad_norm": 2.0140951397899864, "learning_rate": 1.969157769869514e-05, "loss": 0.617, "step": 830 }, { "epoch": 0.029606142116607587, "grad_norm": 2.7332617389949685, "learning_rate": 1.9715302491103204e-05, "loss": 0.6284, "step": 831 }, { "epoch": 0.029641769243101697, "grad_norm": 1.8229411505011768, "learning_rate": 1.973902728351127e-05, "loss": 0.6115, "step": 832 }, { "epoch": 0.02967739636959581, "grad_norm": 1.6766244175728802, "learning_rate": 1.9762752075919338e-05, "loss": 0.5508, "step": 833 }, { "epoch": 0.029713023496089924, "grad_norm": 2.3385518273808654, "learning_rate": 1.9786476868327405e-05, "loss": 0.6203, "step": 834 }, { "epoch": 0.029748650622584034, "grad_norm": 1.5722222350233055, "learning_rate": 1.9810201660735472e-05, "loss": 0.5288, "step": 835 }, { "epoch": 0.029784277749078148, "grad_norm": 1.6976710472090955, "learning_rate": 1.9833926453143536e-05, "loss": 0.6168, "step": 836 }, { "epoch": 0.029819904875572262, "grad_norm": 3.0666845143567554, "learning_rate": 1.9857651245551603e-05, "loss": 0.6321, "step": 837 }, { "epoch": 0.029855532002066372, "grad_norm": 1.611564982431032, "learning_rate": 1.988137603795967e-05, "loss": 0.5141, "step": 838 }, { "epoch": 0.029891159128560486, "grad_norm": 1.7715462076241004, "learning_rate": 1.9905100830367737e-05, "loss": 0.5849, "step": 839 }, { "epoch": 0.0299267862550546, "grad_norm": 1.6509805033959173, "learning_rate": 1.9928825622775804e-05, "loss": 0.5131, "step": 840 }, { "epoch": 0.02996241338154871, "grad_norm": 1.9859722854347555, "learning_rate": 1.9952550415183868e-05, "loss": 0.5984, "step": 841 }, { "epoch": 0.029998040508042823, "grad_norm": 3.1338946359782702, "learning_rate": 1.9976275207591935e-05, "loss": 0.6603, "step": 842 }, { "epoch": 0.030033667634536937, "grad_norm": 3.442312809287929, "learning_rate": 2e-05, "loss": 0.6276, "step": 843 }, { "epoch": 0.03006929476103105, "grad_norm": 1.771708653041108, "learning_rate": 1.9999999933421508e-05, "loss": 0.5392, "step": 844 }, { "epoch": 0.03010492188752516, "grad_norm": 1.7751812755992504, "learning_rate": 1.9999999733686023e-05, "loss": 0.5928, "step": 845 }, { "epoch": 0.030140549014019274, "grad_norm": 1.595784493308538, "learning_rate": 1.9999999400793553e-05, "loss": 0.5646, "step": 846 }, { "epoch": 0.030176176140513388, "grad_norm": 2.6219717814358123, "learning_rate": 1.9999998934744103e-05, "loss": 0.6573, "step": 847 }, { "epoch": 0.0302118032670075, "grad_norm": 2.2214825564422167, "learning_rate": 1.9999998335537675e-05, "loss": 0.6656, "step": 848 }, { "epoch": 0.030247430393501612, "grad_norm": 2.807690874757704, "learning_rate": 1.9999997603174283e-05, "loss": 0.6974, "step": 849 }, { "epoch": 0.030283057519995726, "grad_norm": 2.3550655739778428, "learning_rate": 1.999999673765393e-05, "loss": 0.6932, "step": 850 }, { "epoch": 0.030318684646489836, "grad_norm": 2.0952010934184204, "learning_rate": 1.9999995738976634e-05, "loss": 0.5962, "step": 851 }, { "epoch": 0.03035431177298395, "grad_norm": 1.9626917684421648, "learning_rate": 1.99999946071424e-05, "loss": 0.5729, "step": 852 }, { "epoch": 0.030389938899478063, "grad_norm": 1.594362858228399, "learning_rate": 1.9999993342151254e-05, "loss": 0.5708, "step": 853 }, { "epoch": 0.030425566025972177, "grad_norm": 1.9999430808048175, "learning_rate": 1.9999991944003204e-05, "loss": 0.5656, "step": 854 }, { "epoch": 0.030461193152466287, "grad_norm": 2.0881857593249107, "learning_rate": 1.9999990412698275e-05, "loss": 0.6159, "step": 855 }, { "epoch": 0.0304968202789604, "grad_norm": 2.2367349735282787, "learning_rate": 1.999998874823648e-05, "loss": 0.6036, "step": 856 }, { "epoch": 0.030532447405454514, "grad_norm": 2.807850476653829, "learning_rate": 1.9999986950617843e-05, "loss": 0.6052, "step": 857 }, { "epoch": 0.030568074531948625, "grad_norm": 1.7188868631672112, "learning_rate": 1.9999985019842398e-05, "loss": 0.5857, "step": 858 }, { "epoch": 0.03060370165844274, "grad_norm": 1.6054759563099392, "learning_rate": 1.9999982955910158e-05, "loss": 0.5471, "step": 859 }, { "epoch": 0.030639328784936852, "grad_norm": 1.82085946670356, "learning_rate": 1.9999980758821154e-05, "loss": 0.5581, "step": 860 }, { "epoch": 0.030674955911430962, "grad_norm": 1.784183511226183, "learning_rate": 1.9999978428575423e-05, "loss": 0.5892, "step": 861 }, { "epoch": 0.030710583037925076, "grad_norm": 2.8334321603761974, "learning_rate": 1.9999975965172983e-05, "loss": 0.7617, "step": 862 }, { "epoch": 0.03074621016441919, "grad_norm": 1.7337994139068345, "learning_rate": 1.9999973368613876e-05, "loss": 0.538, "step": 863 }, { "epoch": 0.0307818372909133, "grad_norm": 1.9654486642962736, "learning_rate": 1.9999970638898134e-05, "loss": 0.7424, "step": 864 }, { "epoch": 0.030817464417407413, "grad_norm": 1.6903526732349279, "learning_rate": 1.9999967776025794e-05, "loss": 0.5759, "step": 865 }, { "epoch": 0.030853091543901527, "grad_norm": 1.797012981191305, "learning_rate": 1.9999964779996893e-05, "loss": 0.5557, "step": 866 }, { "epoch": 0.03088871867039564, "grad_norm": 4.184281821862162, "learning_rate": 1.999996165081147e-05, "loss": 0.7124, "step": 867 }, { "epoch": 0.03092434579688975, "grad_norm": 1.9464700997183746, "learning_rate": 1.999995838846957e-05, "loss": 0.6021, "step": 868 }, { "epoch": 0.030959972923383865, "grad_norm": 1.966699946871856, "learning_rate": 1.9999954992971234e-05, "loss": 0.5446, "step": 869 }, { "epoch": 0.03099560004987798, "grad_norm": 1.8295339398992967, "learning_rate": 1.9999951464316507e-05, "loss": 0.5795, "step": 870 }, { "epoch": 0.03103122717637209, "grad_norm": 1.9912742431158517, "learning_rate": 1.9999947802505438e-05, "loss": 0.6273, "step": 871 }, { "epoch": 0.031066854302866202, "grad_norm": 1.959520346947956, "learning_rate": 1.9999944007538073e-05, "loss": 0.5692, "step": 872 }, { "epoch": 0.031102481429360316, "grad_norm": 2.40136692659842, "learning_rate": 1.9999940079414462e-05, "loss": 0.6829, "step": 873 }, { "epoch": 0.031138108555854426, "grad_norm": 2.090629880146182, "learning_rate": 1.9999936018134664e-05, "loss": 0.5792, "step": 874 }, { "epoch": 0.03117373568234854, "grad_norm": 1.9496187866574972, "learning_rate": 1.9999931823698727e-05, "loss": 0.6831, "step": 875 }, { "epoch": 0.031209362808842653, "grad_norm": 1.7246984622552444, "learning_rate": 1.9999927496106708e-05, "loss": 0.5191, "step": 876 }, { "epoch": 0.031244989935336764, "grad_norm": 2.5387892895049697, "learning_rate": 1.9999923035358663e-05, "loss": 0.7333, "step": 877 }, { "epoch": 0.03128061706183088, "grad_norm": 1.8051206481674074, "learning_rate": 1.9999918441454657e-05, "loss": 0.5613, "step": 878 }, { "epoch": 0.03131624418832499, "grad_norm": 1.7050389311973513, "learning_rate": 1.9999913714394744e-05, "loss": 0.5546, "step": 879 }, { "epoch": 0.031351871314819105, "grad_norm": 2.5991898156623723, "learning_rate": 1.999990885417899e-05, "loss": 0.6295, "step": 880 }, { "epoch": 0.031387498441313215, "grad_norm": 1.8577654787489295, "learning_rate": 1.999990386080746e-05, "loss": 0.5543, "step": 881 }, { "epoch": 0.031423125567807325, "grad_norm": 1.7042194485105233, "learning_rate": 1.9999898734280227e-05, "loss": 0.6001, "step": 882 }, { "epoch": 0.03145875269430144, "grad_norm": 2.5170365756892075, "learning_rate": 1.9999893474597347e-05, "loss": 0.616, "step": 883 }, { "epoch": 0.03149437982079555, "grad_norm": 2.761773559953114, "learning_rate": 1.9999888081758895e-05, "loss": 0.6477, "step": 884 }, { "epoch": 0.03153000694728967, "grad_norm": 1.8097108015470602, "learning_rate": 1.9999882555764946e-05, "loss": 0.5394, "step": 885 }, { "epoch": 0.03156563407378378, "grad_norm": 2.307045477527686, "learning_rate": 1.999987689661557e-05, "loss": 0.6595, "step": 886 }, { "epoch": 0.03160126120027789, "grad_norm": 1.6889192708055873, "learning_rate": 1.9999871104310846e-05, "loss": 0.5414, "step": 887 }, { "epoch": 0.03163688832677201, "grad_norm": 1.6860982544332408, "learning_rate": 1.9999865178850847e-05, "loss": 0.565, "step": 888 }, { "epoch": 0.03167251545326612, "grad_norm": 3.6288820131143154, "learning_rate": 1.9999859120235655e-05, "loss": 0.6514, "step": 889 }, { "epoch": 0.03170814257976023, "grad_norm": 2.6086381796650757, "learning_rate": 1.999985292846535e-05, "loss": 0.6435, "step": 890 }, { "epoch": 0.031743769706254345, "grad_norm": 1.8794961850717473, "learning_rate": 1.9999846603540008e-05, "loss": 0.5901, "step": 891 }, { "epoch": 0.031779396832748455, "grad_norm": 1.8526498860309102, "learning_rate": 1.9999840145459725e-05, "loss": 0.5922, "step": 892 }, { "epoch": 0.031815023959242565, "grad_norm": 1.6851509802549398, "learning_rate": 1.9999833554224575e-05, "loss": 0.5451, "step": 893 }, { "epoch": 0.03185065108573668, "grad_norm": 2.552457643441991, "learning_rate": 1.9999826829834657e-05, "loss": 0.6821, "step": 894 }, { "epoch": 0.03188627821223079, "grad_norm": 1.720504161725621, "learning_rate": 1.9999819972290055e-05, "loss": 0.5861, "step": 895 }, { "epoch": 0.0319219053387249, "grad_norm": 2.8349727061695424, "learning_rate": 1.999981298159086e-05, "loss": 0.7371, "step": 896 }, { "epoch": 0.03195753246521902, "grad_norm": 3.706313456531034, "learning_rate": 1.999980585773717e-05, "loss": 0.6532, "step": 897 }, { "epoch": 0.03199315959171313, "grad_norm": 2.5438075293839937, "learning_rate": 1.9999798600729067e-05, "loss": 0.6682, "step": 898 }, { "epoch": 0.03202878671820724, "grad_norm": 1.7266526948817333, "learning_rate": 1.9999791210566658e-05, "loss": 0.5906, "step": 899 }, { "epoch": 0.03206441384470136, "grad_norm": 1.592838252179967, "learning_rate": 1.9999783687250046e-05, "loss": 0.5932, "step": 900 }, { "epoch": 0.03210004097119547, "grad_norm": 3.1135929154333817, "learning_rate": 1.9999776030779316e-05, "loss": 0.6041, "step": 901 }, { "epoch": 0.03213566809768958, "grad_norm": 1.502975079162287, "learning_rate": 1.9999768241154583e-05, "loss": 0.5649, "step": 902 }, { "epoch": 0.032171295224183695, "grad_norm": 2.7128089511927294, "learning_rate": 1.999976031837595e-05, "loss": 0.5661, "step": 903 }, { "epoch": 0.032206922350677805, "grad_norm": 1.8036172166628714, "learning_rate": 1.9999752262443516e-05, "loss": 0.5826, "step": 904 }, { "epoch": 0.032242549477171915, "grad_norm": 2.185765660430288, "learning_rate": 1.999974407335739e-05, "loss": 0.7474, "step": 905 }, { "epoch": 0.03227817660366603, "grad_norm": 1.7003788407967964, "learning_rate": 1.9999735751117684e-05, "loss": 0.5669, "step": 906 }, { "epoch": 0.03231380373016014, "grad_norm": 1.6694018711014118, "learning_rate": 1.9999727295724506e-05, "loss": 0.5103, "step": 907 }, { "epoch": 0.03234943085665426, "grad_norm": 3.7230012504110412, "learning_rate": 1.999971870717797e-05, "loss": 0.5704, "step": 908 }, { "epoch": 0.03238505798314837, "grad_norm": 2.107877572038869, "learning_rate": 1.9999709985478188e-05, "loss": 0.7141, "step": 909 }, { "epoch": 0.03242068510964248, "grad_norm": 2.3962788379762223, "learning_rate": 1.999970113062528e-05, "loss": 0.6909, "step": 910 }, { "epoch": 0.0324563122361366, "grad_norm": 2.2623657379362117, "learning_rate": 1.999969214261936e-05, "loss": 0.5863, "step": 911 }, { "epoch": 0.03249193936263071, "grad_norm": 1.8683602657362888, "learning_rate": 1.9999683021460553e-05, "loss": 0.5799, "step": 912 }, { "epoch": 0.03252756648912482, "grad_norm": 1.7323860576424117, "learning_rate": 1.9999673767148976e-05, "loss": 0.6172, "step": 913 }, { "epoch": 0.032563193615618935, "grad_norm": 1.8413639749553485, "learning_rate": 1.9999664379684754e-05, "loss": 0.5788, "step": 914 }, { "epoch": 0.032598820742113045, "grad_norm": 3.4477076463353304, "learning_rate": 1.9999654859068013e-05, "loss": 0.6979, "step": 915 }, { "epoch": 0.032634447868607155, "grad_norm": 2.639888609986981, "learning_rate": 1.9999645205298876e-05, "loss": 0.7113, "step": 916 }, { "epoch": 0.03267007499510127, "grad_norm": 1.7940449793151667, "learning_rate": 1.9999635418377474e-05, "loss": 0.5607, "step": 917 }, { "epoch": 0.03270570212159538, "grad_norm": 1.8940027939483826, "learning_rate": 1.9999625498303936e-05, "loss": 0.5749, "step": 918 }, { "epoch": 0.03274132924808949, "grad_norm": 2.000287018467016, "learning_rate": 1.9999615445078394e-05, "loss": 0.6006, "step": 919 }, { "epoch": 0.03277695637458361, "grad_norm": 1.6078650323625994, "learning_rate": 1.9999605258700983e-05, "loss": 0.5415, "step": 920 }, { "epoch": 0.03281258350107772, "grad_norm": 3.0437112507078377, "learning_rate": 1.9999594939171844e-05, "loss": 0.6927, "step": 921 }, { "epoch": 0.03284821062757183, "grad_norm": 1.7048466318441904, "learning_rate": 1.9999584486491107e-05, "loss": 0.5258, "step": 922 }, { "epoch": 0.03288383775406595, "grad_norm": 1.688544831878011, "learning_rate": 1.999957390065891e-05, "loss": 0.6046, "step": 923 }, { "epoch": 0.03291946488056006, "grad_norm": 1.65276592243681, "learning_rate": 1.99995631816754e-05, "loss": 0.5315, "step": 924 }, { "epoch": 0.03295509200705417, "grad_norm": 1.9144309535769708, "learning_rate": 1.9999552329540715e-05, "loss": 0.587, "step": 925 }, { "epoch": 0.032990719133548285, "grad_norm": 2.527896713902534, "learning_rate": 1.9999541344255006e-05, "loss": 0.6347, "step": 926 }, { "epoch": 0.033026346260042395, "grad_norm": 3.0491290712059085, "learning_rate": 1.999953022581841e-05, "loss": 0.6264, "step": 927 }, { "epoch": 0.033061973386536506, "grad_norm": 1.933222805402801, "learning_rate": 1.9999518974231082e-05, "loss": 0.5667, "step": 928 }, { "epoch": 0.03309760051303062, "grad_norm": 1.6932226769038607, "learning_rate": 1.999950758949317e-05, "loss": 0.5836, "step": 929 }, { "epoch": 0.03313322763952473, "grad_norm": 2.6060396869576037, "learning_rate": 1.999949607160482e-05, "loss": 0.6844, "step": 930 }, { "epoch": 0.03316885476601885, "grad_norm": 2.2998649381572185, "learning_rate": 1.9999484420566197e-05, "loss": 0.7285, "step": 931 }, { "epoch": 0.03320448189251296, "grad_norm": 3.1050915820416147, "learning_rate": 1.999947263637745e-05, "loss": 0.6363, "step": 932 }, { "epoch": 0.03324010901900707, "grad_norm": 1.8204219905904289, "learning_rate": 1.999946071903873e-05, "loss": 0.5923, "step": 933 }, { "epoch": 0.03327573614550119, "grad_norm": 1.7890083198579236, "learning_rate": 1.9999448668550205e-05, "loss": 0.6044, "step": 934 }, { "epoch": 0.0333113632719953, "grad_norm": 1.783411356023264, "learning_rate": 1.9999436484912033e-05, "loss": 0.5179, "step": 935 }, { "epoch": 0.03334699039848941, "grad_norm": 1.6229360474051078, "learning_rate": 1.999942416812437e-05, "loss": 0.5385, "step": 936 }, { "epoch": 0.033382617524983525, "grad_norm": 1.5383223574289548, "learning_rate": 1.999941171818739e-05, "loss": 0.5161, "step": 937 }, { "epoch": 0.033418244651477635, "grad_norm": 1.5409378324245377, "learning_rate": 1.9999399135101255e-05, "loss": 0.4923, "step": 938 }, { "epoch": 0.033453871777971746, "grad_norm": 1.6512173869440059, "learning_rate": 1.9999386418866126e-05, "loss": 0.5532, "step": 939 }, { "epoch": 0.03348949890446586, "grad_norm": 1.7128002161966636, "learning_rate": 1.9999373569482178e-05, "loss": 0.5744, "step": 940 }, { "epoch": 0.03352512603095997, "grad_norm": 1.7054688158506766, "learning_rate": 1.9999360586949584e-05, "loss": 0.5333, "step": 941 }, { "epoch": 0.03356075315745408, "grad_norm": 3.2251728194026157, "learning_rate": 1.9999347471268517e-05, "loss": 0.6275, "step": 942 }, { "epoch": 0.0335963802839482, "grad_norm": 2.605903595455308, "learning_rate": 1.9999334222439146e-05, "loss": 0.5699, "step": 943 }, { "epoch": 0.03363200741044231, "grad_norm": 2.059564237815413, "learning_rate": 1.999932084046165e-05, "loss": 0.5666, "step": 944 }, { "epoch": 0.03366763453693642, "grad_norm": 1.9637765006001398, "learning_rate": 1.999930732533621e-05, "loss": 0.5576, "step": 945 }, { "epoch": 0.03370326166343054, "grad_norm": 2.2170187496314413, "learning_rate": 1.9999293677063004e-05, "loss": 0.6167, "step": 946 }, { "epoch": 0.03373888878992465, "grad_norm": 1.7258932222092895, "learning_rate": 1.9999279895642212e-05, "loss": 0.5629, "step": 947 }, { "epoch": 0.03377451591641876, "grad_norm": 1.718304504753369, "learning_rate": 1.999926598107402e-05, "loss": 0.5933, "step": 948 }, { "epoch": 0.033810143042912875, "grad_norm": 2.535999840062178, "learning_rate": 1.9999251933358614e-05, "loss": 0.6037, "step": 949 }, { "epoch": 0.033845770169406986, "grad_norm": 1.7885302875080784, "learning_rate": 1.999923775249618e-05, "loss": 0.6093, "step": 950 }, { "epoch": 0.033881397295901096, "grad_norm": 2.805924699741052, "learning_rate": 1.99992234384869e-05, "loss": 0.6124, "step": 951 }, { "epoch": 0.03391702442239521, "grad_norm": 1.8304610814836266, "learning_rate": 1.9999208991330975e-05, "loss": 0.5904, "step": 952 }, { "epoch": 0.03395265154888932, "grad_norm": 1.6726002613751276, "learning_rate": 1.9999194411028596e-05, "loss": 0.5478, "step": 953 }, { "epoch": 0.03398827867538344, "grad_norm": 1.7011518938533658, "learning_rate": 1.999917969757995e-05, "loss": 0.5716, "step": 954 }, { "epoch": 0.03402390580187755, "grad_norm": 1.6525426447549165, "learning_rate": 1.9999164850985238e-05, "loss": 0.5846, "step": 955 }, { "epoch": 0.03405953292837166, "grad_norm": 2.5438612592224588, "learning_rate": 1.999914987124466e-05, "loss": 0.6578, "step": 956 }, { "epoch": 0.03409516005486578, "grad_norm": 1.9709332950002951, "learning_rate": 1.999913475835841e-05, "loss": 0.5507, "step": 957 }, { "epoch": 0.03413078718135989, "grad_norm": 1.7108204049844855, "learning_rate": 1.9999119512326694e-05, "loss": 0.5656, "step": 958 }, { "epoch": 0.034166414307854, "grad_norm": 2.330178466035195, "learning_rate": 1.999910413314971e-05, "loss": 0.6963, "step": 959 }, { "epoch": 0.034202041434348116, "grad_norm": 2.223172755328051, "learning_rate": 1.9999088620827666e-05, "loss": 0.6054, "step": 960 }, { "epoch": 0.034237668560842226, "grad_norm": 2.263973730733286, "learning_rate": 1.9999072975360766e-05, "loss": 0.6895, "step": 961 }, { "epoch": 0.034273295687336336, "grad_norm": 2.1299474042074964, "learning_rate": 1.9999057196749225e-05, "loss": 0.5707, "step": 962 }, { "epoch": 0.03430892281383045, "grad_norm": 1.737868481929294, "learning_rate": 1.9999041284993246e-05, "loss": 0.5574, "step": 963 }, { "epoch": 0.03434454994032456, "grad_norm": 2.3754420631154676, "learning_rate": 1.9999025240093045e-05, "loss": 0.6534, "step": 964 }, { "epoch": 0.034380177066818673, "grad_norm": 1.7099948052157101, "learning_rate": 1.999900906204883e-05, "loss": 0.5117, "step": 965 }, { "epoch": 0.03441580419331279, "grad_norm": 1.5431653989093628, "learning_rate": 1.9998992750860827e-05, "loss": 0.494, "step": 966 }, { "epoch": 0.0344514313198069, "grad_norm": 1.9878240300150507, "learning_rate": 1.999897630652924e-05, "loss": 0.7075, "step": 967 }, { "epoch": 0.03448705844630101, "grad_norm": 1.6292398899647318, "learning_rate": 1.9998959729054296e-05, "loss": 0.6018, "step": 968 }, { "epoch": 0.03452268557279513, "grad_norm": 3.3349314242143926, "learning_rate": 1.9998943018436218e-05, "loss": 0.7234, "step": 969 }, { "epoch": 0.03455831269928924, "grad_norm": 2.8529716674696393, "learning_rate": 1.9998926174675223e-05, "loss": 0.7075, "step": 970 }, { "epoch": 0.03459393982578335, "grad_norm": 1.783608035698952, "learning_rate": 1.9998909197771538e-05, "loss": 0.5922, "step": 971 }, { "epoch": 0.034629566952277466, "grad_norm": 2.4230313994141177, "learning_rate": 1.9998892087725387e-05, "loss": 0.6494, "step": 972 }, { "epoch": 0.034665194078771576, "grad_norm": 2.5463194098015682, "learning_rate": 1.9998874844536997e-05, "loss": 0.6737, "step": 973 }, { "epoch": 0.034700821205265686, "grad_norm": 1.5630442450162416, "learning_rate": 1.9998857468206604e-05, "loss": 0.5468, "step": 974 }, { "epoch": 0.0347364483317598, "grad_norm": 1.5059803676165369, "learning_rate": 1.999883995873443e-05, "loss": 0.5559, "step": 975 }, { "epoch": 0.034772075458253913, "grad_norm": 2.1044535897784464, "learning_rate": 1.9998822316120718e-05, "loss": 0.6439, "step": 976 }, { "epoch": 0.034807702584748024, "grad_norm": 1.4639498107998674, "learning_rate": 1.9998804540365695e-05, "loss": 0.5657, "step": 977 }, { "epoch": 0.03484332971124214, "grad_norm": 1.6345155332696863, "learning_rate": 1.9998786631469602e-05, "loss": 0.5895, "step": 978 }, { "epoch": 0.03487895683773625, "grad_norm": 1.624525203315038, "learning_rate": 1.999876858943268e-05, "loss": 0.5745, "step": 979 }, { "epoch": 0.03491458396423037, "grad_norm": 1.6328140619134706, "learning_rate": 1.999875041425516e-05, "loss": 0.6134, "step": 980 }, { "epoch": 0.03495021109072448, "grad_norm": 1.5597442532875188, "learning_rate": 1.999873210593729e-05, "loss": 0.563, "step": 981 }, { "epoch": 0.03498583821721859, "grad_norm": 1.4669643701806878, "learning_rate": 1.9998713664479313e-05, "loss": 0.574, "step": 982 }, { "epoch": 0.035021465343712706, "grad_norm": 2.0620684954856787, "learning_rate": 1.9998695089881477e-05, "loss": 0.7151, "step": 983 }, { "epoch": 0.035057092470206816, "grad_norm": 3.1958614271336434, "learning_rate": 1.9998676382144027e-05, "loss": 0.7315, "step": 984 }, { "epoch": 0.035092719596700926, "grad_norm": 1.742691052539258, "learning_rate": 1.999865754126721e-05, "loss": 0.5314, "step": 985 }, { "epoch": 0.03512834672319504, "grad_norm": 2.2801585961012387, "learning_rate": 1.9998638567251283e-05, "loss": 0.7123, "step": 986 }, { "epoch": 0.035163973849689154, "grad_norm": 2.9977583747599748, "learning_rate": 1.999861946009649e-05, "loss": 0.7284, "step": 987 }, { "epoch": 0.035199600976183264, "grad_norm": 1.6384528794927955, "learning_rate": 1.9998600219803097e-05, "loss": 0.5839, "step": 988 }, { "epoch": 0.03523522810267738, "grad_norm": 2.383444236625157, "learning_rate": 1.999858084637135e-05, "loss": 0.6991, "step": 989 }, { "epoch": 0.03527085522917149, "grad_norm": 1.760018509370371, "learning_rate": 1.999856133980151e-05, "loss": 0.5653, "step": 990 }, { "epoch": 0.0353064823556656, "grad_norm": 1.43952869204012, "learning_rate": 1.9998541700093836e-05, "loss": 0.5445, "step": 991 }, { "epoch": 0.03534210948215972, "grad_norm": 1.7370042884810775, "learning_rate": 1.999852192724859e-05, "loss": 0.6023, "step": 992 }, { "epoch": 0.03537773660865383, "grad_norm": 1.5466446350967633, "learning_rate": 1.999850202126604e-05, "loss": 0.5446, "step": 993 }, { "epoch": 0.03541336373514794, "grad_norm": 2.370995718260016, "learning_rate": 1.9998481982146445e-05, "loss": 0.586, "step": 994 }, { "epoch": 0.035448990861642056, "grad_norm": 2.114797236281453, "learning_rate": 1.9998461809890073e-05, "loss": 0.6029, "step": 995 }, { "epoch": 0.035484617988136166, "grad_norm": 1.6435668764517604, "learning_rate": 1.9998441504497194e-05, "loss": 0.543, "step": 996 }, { "epoch": 0.035520245114630276, "grad_norm": 1.7478749988764661, "learning_rate": 1.999842106596808e-05, "loss": 0.6091, "step": 997 }, { "epoch": 0.035555872241124394, "grad_norm": 2.4419834166373398, "learning_rate": 1.9998400494302998e-05, "loss": 0.7032, "step": 998 }, { "epoch": 0.035591499367618504, "grad_norm": 2.217895764725458, "learning_rate": 1.9998379789502226e-05, "loss": 0.752, "step": 999 }, { "epoch": 0.035627126494112614, "grad_norm": 2.027132238261744, "learning_rate": 1.999835895156604e-05, "loss": 0.7402, "step": 1000 }, { "epoch": 0.03566275362060673, "grad_norm": 2.633966569191458, "learning_rate": 1.9998337980494714e-05, "loss": 0.7022, "step": 1001 }, { "epoch": 0.03569838074710084, "grad_norm": 2.3105196275260473, "learning_rate": 1.9998316876288532e-05, "loss": 0.6897, "step": 1002 }, { "epoch": 0.03573400787359496, "grad_norm": 2.4297759657280804, "learning_rate": 1.9998295638947768e-05, "loss": 0.7517, "step": 1003 }, { "epoch": 0.03576963500008907, "grad_norm": 2.7058028144808848, "learning_rate": 1.9998274268472714e-05, "loss": 0.6639, "step": 1004 }, { "epoch": 0.03580526212658318, "grad_norm": 2.918126084854446, "learning_rate": 1.9998252764863646e-05, "loss": 0.6913, "step": 1005 }, { "epoch": 0.035840889253077296, "grad_norm": 1.964697511798385, "learning_rate": 1.9998231128120858e-05, "loss": 0.5837, "step": 1006 }, { "epoch": 0.035876516379571406, "grad_norm": 1.791275224197727, "learning_rate": 1.9998209358244632e-05, "loss": 0.5449, "step": 1007 }, { "epoch": 0.035912143506065516, "grad_norm": 2.2328647970788222, "learning_rate": 1.999818745523526e-05, "loss": 0.6925, "step": 1008 }, { "epoch": 0.035947770632559634, "grad_norm": 2.386154128414624, "learning_rate": 1.9998165419093032e-05, "loss": 0.7175, "step": 1009 }, { "epoch": 0.035983397759053744, "grad_norm": 2.1442183085904403, "learning_rate": 1.9998143249818245e-05, "loss": 0.6092, "step": 1010 }, { "epoch": 0.036019024885547854, "grad_norm": 2.606296107236684, "learning_rate": 1.9998120947411195e-05, "loss": 0.6742, "step": 1011 }, { "epoch": 0.03605465201204197, "grad_norm": 1.822668872214218, "learning_rate": 1.9998098511872173e-05, "loss": 0.6493, "step": 1012 }, { "epoch": 0.03609027913853608, "grad_norm": 1.6606150737619845, "learning_rate": 1.999807594320148e-05, "loss": 0.4965, "step": 1013 }, { "epoch": 0.03612590626503019, "grad_norm": 2.932398922491734, "learning_rate": 1.999805324139942e-05, "loss": 0.696, "step": 1014 }, { "epoch": 0.03616153339152431, "grad_norm": 2.5307913842867777, "learning_rate": 1.9998030406466293e-05, "loss": 0.5815, "step": 1015 }, { "epoch": 0.03619716051801842, "grad_norm": 1.431156955004804, "learning_rate": 1.99980074384024e-05, "loss": 0.5343, "step": 1016 }, { "epoch": 0.03623278764451253, "grad_norm": 2.222157058110697, "learning_rate": 1.9997984337208055e-05, "loss": 0.643, "step": 1017 }, { "epoch": 0.036268414771006646, "grad_norm": 2.4663354299447096, "learning_rate": 1.9997961102883553e-05, "loss": 0.6747, "step": 1018 }, { "epoch": 0.036304041897500756, "grad_norm": 3.7350647643291666, "learning_rate": 1.999793773542922e-05, "loss": 0.5706, "step": 1019 }, { "epoch": 0.03633966902399487, "grad_norm": 1.93546032059128, "learning_rate": 1.999791423484535e-05, "loss": 0.6371, "step": 1020 }, { "epoch": 0.036375296150488984, "grad_norm": 3.405509977232439, "learning_rate": 1.9997890601132266e-05, "loss": 0.673, "step": 1021 }, { "epoch": 0.036410923276983094, "grad_norm": 3.1612366940145447, "learning_rate": 1.999786683429028e-05, "loss": 0.6958, "step": 1022 }, { "epoch": 0.036446550403477204, "grad_norm": 2.2783844062285468, "learning_rate": 1.999784293431971e-05, "loss": 0.6366, "step": 1023 }, { "epoch": 0.03648217752997132, "grad_norm": 2.1972353451738567, "learning_rate": 1.9997818901220873e-05, "loss": 0.6457, "step": 1024 }, { "epoch": 0.03651780465646543, "grad_norm": 1.5792924604570475, "learning_rate": 1.999779473499409e-05, "loss": 0.5966, "step": 1025 }, { "epoch": 0.03655343178295955, "grad_norm": 2.6330162418418768, "learning_rate": 1.999777043563968e-05, "loss": 0.7002, "step": 1026 }, { "epoch": 0.03658905890945366, "grad_norm": 2.729002616110577, "learning_rate": 1.9997746003157975e-05, "loss": 0.668, "step": 1027 }, { "epoch": 0.03662468603594777, "grad_norm": 1.826680716604144, "learning_rate": 1.9997721437549286e-05, "loss": 0.5842, "step": 1028 }, { "epoch": 0.036660313162441886, "grad_norm": 3.6251025975919764, "learning_rate": 1.9997696738813954e-05, "loss": 0.6316, "step": 1029 }, { "epoch": 0.036695940288935996, "grad_norm": 2.7062932581175936, "learning_rate": 1.99976719069523e-05, "loss": 0.6252, "step": 1030 }, { "epoch": 0.03673156741543011, "grad_norm": 2.4387370801469235, "learning_rate": 1.9997646941964655e-05, "loss": 0.6898, "step": 1031 }, { "epoch": 0.036767194541924224, "grad_norm": 1.8665654465404253, "learning_rate": 1.9997621843851353e-05, "loss": 0.5487, "step": 1032 }, { "epoch": 0.036802821668418334, "grad_norm": 2.273893113270332, "learning_rate": 1.9997596612612732e-05, "loss": 0.7125, "step": 1033 }, { "epoch": 0.036838448794912444, "grad_norm": 2.4834705047467716, "learning_rate": 1.999757124824912e-05, "loss": 0.6303, "step": 1034 }, { "epoch": 0.03687407592140656, "grad_norm": 2.5471835319859, "learning_rate": 1.999754575076086e-05, "loss": 0.6427, "step": 1035 }, { "epoch": 0.03690970304790067, "grad_norm": 3.6773224476781485, "learning_rate": 1.9997520120148294e-05, "loss": 0.669, "step": 1036 }, { "epoch": 0.03694533017439478, "grad_norm": 2.551260662295294, "learning_rate": 1.9997494356411752e-05, "loss": 0.6425, "step": 1037 }, { "epoch": 0.0369809573008889, "grad_norm": 2.0348227389950786, "learning_rate": 1.999746845955159e-05, "loss": 0.7188, "step": 1038 }, { "epoch": 0.03701658442738301, "grad_norm": 1.5619423015097187, "learning_rate": 1.999744242956815e-05, "loss": 0.5336, "step": 1039 }, { "epoch": 0.03705221155387712, "grad_norm": 2.5584091490625367, "learning_rate": 1.9997416266461772e-05, "loss": 0.622, "step": 1040 }, { "epoch": 0.037087838680371236, "grad_norm": 1.6825469336920373, "learning_rate": 1.999738997023281e-05, "loss": 0.5604, "step": 1041 }, { "epoch": 0.03712346580686535, "grad_norm": 2.6384172741262883, "learning_rate": 1.999736354088161e-05, "loss": 0.5983, "step": 1042 }, { "epoch": 0.03715909293335946, "grad_norm": 2.577551710627498, "learning_rate": 1.999733697840853e-05, "loss": 0.6469, "step": 1043 }, { "epoch": 0.037194720059853574, "grad_norm": 1.5165748103310537, "learning_rate": 1.9997310282813923e-05, "loss": 0.5229, "step": 1044 }, { "epoch": 0.037230347186347684, "grad_norm": 2.8164534226878932, "learning_rate": 1.9997283454098137e-05, "loss": 0.6553, "step": 1045 }, { "epoch": 0.037265974312841794, "grad_norm": 1.781186968654212, "learning_rate": 1.9997256492261538e-05, "loss": 0.6362, "step": 1046 }, { "epoch": 0.03730160143933591, "grad_norm": 1.901209267693804, "learning_rate": 1.9997229397304475e-05, "loss": 0.5696, "step": 1047 }, { "epoch": 0.03733722856583002, "grad_norm": 3.154305467464335, "learning_rate": 1.999720216922732e-05, "loss": 0.6383, "step": 1048 }, { "epoch": 0.03737285569232414, "grad_norm": 1.6619894181240822, "learning_rate": 1.9997174808030432e-05, "loss": 0.5972, "step": 1049 }, { "epoch": 0.03740848281881825, "grad_norm": 2.0881909341875695, "learning_rate": 1.999714731371417e-05, "loss": 0.6722, "step": 1050 }, { "epoch": 0.03744410994531236, "grad_norm": 1.5288915018396678, "learning_rate": 1.9997119686278905e-05, "loss": 0.5618, "step": 1051 }, { "epoch": 0.037479737071806477, "grad_norm": 1.5067770979813744, "learning_rate": 1.9997091925725006e-05, "loss": 0.5605, "step": 1052 }, { "epoch": 0.03751536419830059, "grad_norm": 1.49923465576017, "learning_rate": 1.9997064032052838e-05, "loss": 0.5644, "step": 1053 }, { "epoch": 0.0375509913247947, "grad_norm": 1.481545472492944, "learning_rate": 1.9997036005262775e-05, "loss": 0.5704, "step": 1054 }, { "epoch": 0.037586618451288814, "grad_norm": 2.277231569989921, "learning_rate": 1.999700784535519e-05, "loss": 0.7624, "step": 1055 }, { "epoch": 0.037622245577782924, "grad_norm": 1.5647463479009938, "learning_rate": 1.9996979552330457e-05, "loss": 0.5425, "step": 1056 }, { "epoch": 0.037657872704277034, "grad_norm": 3.2696448884530374, "learning_rate": 1.9996951126188957e-05, "loss": 0.6352, "step": 1057 }, { "epoch": 0.03769349983077115, "grad_norm": 2.5566131123677764, "learning_rate": 1.9996922566931063e-05, "loss": 0.6011, "step": 1058 }, { "epoch": 0.03772912695726526, "grad_norm": 3.1017830965336928, "learning_rate": 1.9996893874557157e-05, "loss": 0.7655, "step": 1059 }, { "epoch": 0.03776475408375937, "grad_norm": 1.5091947998980297, "learning_rate": 1.9996865049067625e-05, "loss": 0.561, "step": 1060 }, { "epoch": 0.03780038121025349, "grad_norm": 2.203314331305332, "learning_rate": 1.9996836090462844e-05, "loss": 0.7436, "step": 1061 }, { "epoch": 0.0378360083367476, "grad_norm": 1.786592150466159, "learning_rate": 1.9996806998743207e-05, "loss": 0.5781, "step": 1062 }, { "epoch": 0.03787163546324171, "grad_norm": 2.1432130053334464, "learning_rate": 1.9996777773909093e-05, "loss": 0.7434, "step": 1063 }, { "epoch": 0.03790726258973583, "grad_norm": 2.232366911346092, "learning_rate": 1.9996748415960895e-05, "loss": 0.742, "step": 1064 }, { "epoch": 0.03794288971622994, "grad_norm": 1.5437521193413397, "learning_rate": 1.999671892489901e-05, "loss": 0.5581, "step": 1065 }, { "epoch": 0.03797851684272405, "grad_norm": 2.1755008771262463, "learning_rate": 1.9996689300723823e-05, "loss": 0.5988, "step": 1066 }, { "epoch": 0.038014143969218164, "grad_norm": 1.4790525912219021, "learning_rate": 1.9996659543435728e-05, "loss": 0.5751, "step": 1067 }, { "epoch": 0.038049771095712275, "grad_norm": 3.374199496726535, "learning_rate": 1.9996629653035128e-05, "loss": 0.7304, "step": 1068 }, { "epoch": 0.038085398222206385, "grad_norm": 2.8114654462673188, "learning_rate": 1.9996599629522413e-05, "loss": 0.6917, "step": 1069 }, { "epoch": 0.0381210253487005, "grad_norm": 1.4834365843325548, "learning_rate": 1.999656947289799e-05, "loss": 0.5594, "step": 1070 }, { "epoch": 0.03815665247519461, "grad_norm": 2.5854385279174354, "learning_rate": 1.9996539183162253e-05, "loss": 0.626, "step": 1071 }, { "epoch": 0.03819227960168873, "grad_norm": 1.7993335062203242, "learning_rate": 1.9996508760315614e-05, "loss": 0.5456, "step": 1072 }, { "epoch": 0.03822790672818284, "grad_norm": 2.232576760221879, "learning_rate": 1.9996478204358473e-05, "loss": 0.6898, "step": 1073 }, { "epoch": 0.03826353385467695, "grad_norm": 2.283160602848741, "learning_rate": 1.9996447515291236e-05, "loss": 0.6399, "step": 1074 }, { "epoch": 0.03829916098117107, "grad_norm": 1.650408043266593, "learning_rate": 1.9996416693114312e-05, "loss": 0.5563, "step": 1075 }, { "epoch": 0.03833478810766518, "grad_norm": 2.4401836466532756, "learning_rate": 1.9996385737828116e-05, "loss": 0.6492, "step": 1076 }, { "epoch": 0.03837041523415929, "grad_norm": 1.521589721435809, "learning_rate": 1.9996354649433054e-05, "loss": 0.6098, "step": 1077 }, { "epoch": 0.038406042360653404, "grad_norm": 1.6704042641994115, "learning_rate": 1.9996323427929546e-05, "loss": 0.5836, "step": 1078 }, { "epoch": 0.038441669487147515, "grad_norm": 1.5138431031840163, "learning_rate": 1.9996292073318e-05, "loss": 0.5603, "step": 1079 }, { "epoch": 0.038477296613641625, "grad_norm": 1.5606398155123353, "learning_rate": 1.9996260585598845e-05, "loss": 0.5207, "step": 1080 }, { "epoch": 0.03851292374013574, "grad_norm": 2.513386565015468, "learning_rate": 1.9996228964772485e-05, "loss": 0.6879, "step": 1081 }, { "epoch": 0.03854855086662985, "grad_norm": 2.399775476134365, "learning_rate": 1.9996197210839357e-05, "loss": 0.6739, "step": 1082 }, { "epoch": 0.03858417799312396, "grad_norm": 2.8731687637668113, "learning_rate": 1.9996165323799873e-05, "loss": 0.6277, "step": 1083 }, { "epoch": 0.03861980511961808, "grad_norm": 2.2085970618113886, "learning_rate": 1.9996133303654458e-05, "loss": 0.6816, "step": 1084 }, { "epoch": 0.03865543224611219, "grad_norm": 2.285827254914098, "learning_rate": 1.9996101150403543e-05, "loss": 0.7425, "step": 1085 }, { "epoch": 0.0386910593726063, "grad_norm": 1.9317978830024767, "learning_rate": 1.9996068864047558e-05, "loss": 0.5813, "step": 1086 }, { "epoch": 0.03872668649910042, "grad_norm": 1.6734522759349595, "learning_rate": 1.9996036444586925e-05, "loss": 0.5818, "step": 1087 }, { "epoch": 0.03876231362559453, "grad_norm": 1.7424863586205894, "learning_rate": 1.9996003892022084e-05, "loss": 0.6146, "step": 1088 }, { "epoch": 0.03879794075208864, "grad_norm": 1.637820806789744, "learning_rate": 1.999597120635346e-05, "loss": 0.6539, "step": 1089 }, { "epoch": 0.038833567878582755, "grad_norm": 1.6286403270505465, "learning_rate": 1.9995938387581494e-05, "loss": 0.5246, "step": 1090 }, { "epoch": 0.038869195005076865, "grad_norm": 2.15663454155808, "learning_rate": 1.9995905435706623e-05, "loss": 0.7235, "step": 1091 }, { "epoch": 0.038904822131570975, "grad_norm": 2.722528056992288, "learning_rate": 1.9995872350729282e-05, "loss": 0.7022, "step": 1092 }, { "epoch": 0.03894044925806509, "grad_norm": 2.620246847578743, "learning_rate": 1.9995839132649916e-05, "loss": 0.618, "step": 1093 }, { "epoch": 0.0389760763845592, "grad_norm": 1.5912236331055036, "learning_rate": 1.9995805781468965e-05, "loss": 0.5908, "step": 1094 }, { "epoch": 0.03901170351105332, "grad_norm": 2.3107895343574456, "learning_rate": 1.9995772297186872e-05, "loss": 0.7367, "step": 1095 }, { "epoch": 0.03904733063754743, "grad_norm": 2.480710228078086, "learning_rate": 1.9995738679804086e-05, "loss": 0.6565, "step": 1096 }, { "epoch": 0.03908295776404154, "grad_norm": 1.9594568846606584, "learning_rate": 1.9995704929321053e-05, "loss": 0.5218, "step": 1097 }, { "epoch": 0.03911858489053566, "grad_norm": 2.8378443975693926, "learning_rate": 1.999567104573822e-05, "loss": 0.6986, "step": 1098 }, { "epoch": 0.03915421201702977, "grad_norm": 1.3993210427753686, "learning_rate": 1.9995637029056042e-05, "loss": 0.5176, "step": 1099 }, { "epoch": 0.03918983914352388, "grad_norm": 1.6919934442627367, "learning_rate": 1.999560287927497e-05, "loss": 0.5727, "step": 1100 }, { "epoch": 0.039225466270017995, "grad_norm": 2.5219591212730084, "learning_rate": 1.999556859639546e-05, "loss": 0.7115, "step": 1101 }, { "epoch": 0.039261093396512105, "grad_norm": 2.542640020259864, "learning_rate": 1.9995534180417965e-05, "loss": 0.6302, "step": 1102 }, { "epoch": 0.039296720523006215, "grad_norm": 2.4429776966307593, "learning_rate": 1.999549963134295e-05, "loss": 0.693, "step": 1103 }, { "epoch": 0.03933234764950033, "grad_norm": 1.9365253465332917, "learning_rate": 1.9995464949170867e-05, "loss": 0.5901, "step": 1104 }, { "epoch": 0.03936797477599444, "grad_norm": 1.999998407041065, "learning_rate": 1.9995430133902184e-05, "loss": 0.5275, "step": 1105 }, { "epoch": 0.03940360190248855, "grad_norm": 1.7620283409938433, "learning_rate": 1.9995395185537364e-05, "loss": 0.5451, "step": 1106 }, { "epoch": 0.03943922902898267, "grad_norm": 1.6152191887579284, "learning_rate": 1.999536010407687e-05, "loss": 0.5673, "step": 1107 }, { "epoch": 0.03947485615547678, "grad_norm": 2.751009702011775, "learning_rate": 1.9995324889521167e-05, "loss": 0.7044, "step": 1108 }, { "epoch": 0.03951048328197089, "grad_norm": 3.702396929691065, "learning_rate": 1.999528954187073e-05, "loss": 0.6416, "step": 1109 }, { "epoch": 0.03954611040846501, "grad_norm": 2.2538478574752823, "learning_rate": 1.9995254061126023e-05, "loss": 0.6262, "step": 1110 }, { "epoch": 0.03958173753495912, "grad_norm": 3.7552306866658784, "learning_rate": 1.9995218447287522e-05, "loss": 0.6763, "step": 1111 }, { "epoch": 0.03961736466145323, "grad_norm": 1.8452950821353846, "learning_rate": 1.9995182700355703e-05, "loss": 0.6742, "step": 1112 }, { "epoch": 0.039652991787947345, "grad_norm": 1.7703521805233098, "learning_rate": 1.999514682033104e-05, "loss": 0.5746, "step": 1113 }, { "epoch": 0.039688618914441455, "grad_norm": 1.7888900252279858, "learning_rate": 1.9995110807214015e-05, "loss": 0.5791, "step": 1114 }, { "epoch": 0.039724246040935565, "grad_norm": 3.0554555393863443, "learning_rate": 1.9995074661005098e-05, "loss": 0.728, "step": 1115 }, { "epoch": 0.03975987316742968, "grad_norm": 2.45246169039751, "learning_rate": 1.9995038381704777e-05, "loss": 0.5807, "step": 1116 }, { "epoch": 0.03979550029392379, "grad_norm": 1.8095334146689661, "learning_rate": 1.9995001969313534e-05, "loss": 0.5604, "step": 1117 }, { "epoch": 0.0398311274204179, "grad_norm": 11.555993739280474, "learning_rate": 1.9994965423831853e-05, "loss": 0.7008, "step": 1118 }, { "epoch": 0.03986675454691202, "grad_norm": 1.6870284536812616, "learning_rate": 1.999492874526022e-05, "loss": 0.5691, "step": 1119 }, { "epoch": 0.03990238167340613, "grad_norm": 3.9712061851850784, "learning_rate": 1.9994891933599128e-05, "loss": 0.662, "step": 1120 }, { "epoch": 0.03993800879990025, "grad_norm": 2.918011804747919, "learning_rate": 1.9994854988849063e-05, "loss": 0.7589, "step": 1121 }, { "epoch": 0.03997363592639436, "grad_norm": 1.7792608595483106, "learning_rate": 1.9994817911010518e-05, "loss": 0.5388, "step": 1122 }, { "epoch": 0.04000926305288847, "grad_norm": 2.024359155055614, "learning_rate": 1.9994780700083985e-05, "loss": 0.5902, "step": 1123 }, { "epoch": 0.040044890179382585, "grad_norm": 2.796824301946557, "learning_rate": 1.9994743356069962e-05, "loss": 0.6735, "step": 1124 }, { "epoch": 0.040080517305876695, "grad_norm": 1.6190377710336006, "learning_rate": 1.9994705878968946e-05, "loss": 0.5468, "step": 1125 }, { "epoch": 0.040116144432370805, "grad_norm": 1.5977872747960145, "learning_rate": 1.9994668268781433e-05, "loss": 0.5932, "step": 1126 }, { "epoch": 0.04015177155886492, "grad_norm": 1.8164257846413228, "learning_rate": 1.999463052550793e-05, "loss": 0.6282, "step": 1127 }, { "epoch": 0.04018739868535903, "grad_norm": 1.6527005437986144, "learning_rate": 1.9994592649148935e-05, "loss": 0.63, "step": 1128 }, { "epoch": 0.04022302581185314, "grad_norm": 2.3262937575344087, "learning_rate": 1.999455463970495e-05, "loss": 0.5935, "step": 1129 }, { "epoch": 0.04025865293834726, "grad_norm": 1.7537160490208121, "learning_rate": 1.9994516497176485e-05, "loss": 0.612, "step": 1130 }, { "epoch": 0.04029428006484137, "grad_norm": 2.876315442268658, "learning_rate": 1.999447822156405e-05, "loss": 0.6047, "step": 1131 }, { "epoch": 0.04032990719133548, "grad_norm": 2.387868371215932, "learning_rate": 1.999443981286815e-05, "loss": 0.759, "step": 1132 }, { "epoch": 0.0403655343178296, "grad_norm": 2.7759835642464674, "learning_rate": 1.9994401271089295e-05, "loss": 0.6353, "step": 1133 }, { "epoch": 0.04040116144432371, "grad_norm": 2.31556888116092, "learning_rate": 1.9994362596228005e-05, "loss": 0.6794, "step": 1134 }, { "epoch": 0.04043678857081782, "grad_norm": 1.64543473713631, "learning_rate": 1.999432378828479e-05, "loss": 0.5421, "step": 1135 }, { "epoch": 0.040472415697311935, "grad_norm": 1.7263819579646895, "learning_rate": 1.999428484726017e-05, "loss": 0.5425, "step": 1136 }, { "epoch": 0.040508042823806045, "grad_norm": 1.7161810968577647, "learning_rate": 1.999424577315466e-05, "loss": 0.5593, "step": 1137 }, { "epoch": 0.040543669950300155, "grad_norm": 3.116804114025337, "learning_rate": 1.9994206565968783e-05, "loss": 0.7079, "step": 1138 }, { "epoch": 0.04057929707679427, "grad_norm": 2.2413039835311146, "learning_rate": 1.999416722570306e-05, "loss": 0.6581, "step": 1139 }, { "epoch": 0.04061492420328838, "grad_norm": 3.0408413616408794, "learning_rate": 1.9994127752358014e-05, "loss": 0.6538, "step": 1140 }, { "epoch": 0.04065055132978249, "grad_norm": 2.0928382985728606, "learning_rate": 1.999408814593417e-05, "loss": 0.6137, "step": 1141 }, { "epoch": 0.04068617845627661, "grad_norm": 2.8559136150695066, "learning_rate": 1.999404840643206e-05, "loss": 0.621, "step": 1142 }, { "epoch": 0.04072180558277072, "grad_norm": 1.604700100348843, "learning_rate": 1.999400853385221e-05, "loss": 0.5171, "step": 1143 }, { "epoch": 0.04075743270926484, "grad_norm": 2.3540658719300804, "learning_rate": 1.999396852819515e-05, "loss": 0.7508, "step": 1144 }, { "epoch": 0.04079305983575895, "grad_norm": 2.005477196832939, "learning_rate": 1.9993928389461414e-05, "loss": 0.58, "step": 1145 }, { "epoch": 0.04082868696225306, "grad_norm": 1.778443497383349, "learning_rate": 1.9993888117651537e-05, "loss": 0.656, "step": 1146 }, { "epoch": 0.040864314088747175, "grad_norm": 1.4634835666214472, "learning_rate": 1.9993847712766052e-05, "loss": 0.5612, "step": 1147 }, { "epoch": 0.040899941215241285, "grad_norm": 1.83258655414563, "learning_rate": 1.99938071748055e-05, "loss": 0.5392, "step": 1148 }, { "epoch": 0.040935568341735395, "grad_norm": 1.7469831636134663, "learning_rate": 1.999376650377042e-05, "loss": 0.5804, "step": 1149 }, { "epoch": 0.04097119546822951, "grad_norm": 1.650958057427587, "learning_rate": 1.9993725699661354e-05, "loss": 0.543, "step": 1150 }, { "epoch": 0.04100682259472362, "grad_norm": 1.5265247630840162, "learning_rate": 1.9993684762478846e-05, "loss": 0.6182, "step": 1151 }, { "epoch": 0.04104244972121773, "grad_norm": 2.459603325247118, "learning_rate": 1.999364369222344e-05, "loss": 0.648, "step": 1152 }, { "epoch": 0.04107807684771185, "grad_norm": 2.489621752779811, "learning_rate": 1.9993602488895684e-05, "loss": 0.6541, "step": 1153 }, { "epoch": 0.04111370397420596, "grad_norm": 2.5971758265145626, "learning_rate": 1.9993561152496123e-05, "loss": 0.6868, "step": 1154 }, { "epoch": 0.04114933110070007, "grad_norm": 1.4456315714344279, "learning_rate": 1.999351968302531e-05, "loss": 0.5534, "step": 1155 }, { "epoch": 0.04118495822719419, "grad_norm": 1.5004469706210586, "learning_rate": 1.99934780804838e-05, "loss": 0.592, "step": 1156 }, { "epoch": 0.0412205853536883, "grad_norm": 3.3806307901568204, "learning_rate": 1.9993436344872147e-05, "loss": 0.7088, "step": 1157 }, { "epoch": 0.04125621248018241, "grad_norm": 2.061633849546533, "learning_rate": 1.99933944761909e-05, "loss": 0.6985, "step": 1158 }, { "epoch": 0.041291839606676525, "grad_norm": 1.4959419097784006, "learning_rate": 1.9993352474440622e-05, "loss": 0.6058, "step": 1159 }, { "epoch": 0.041327466733170636, "grad_norm": 1.5662794231702182, "learning_rate": 1.9993310339621866e-05, "loss": 0.5512, "step": 1160 }, { "epoch": 0.041363093859664746, "grad_norm": 2.4942879978791903, "learning_rate": 1.9993268071735206e-05, "loss": 0.673, "step": 1161 }, { "epoch": 0.04139872098615886, "grad_norm": 2.8102134893432975, "learning_rate": 1.9993225670781192e-05, "loss": 0.6578, "step": 1162 }, { "epoch": 0.04143434811265297, "grad_norm": 2.46988233322553, "learning_rate": 1.9993183136760392e-05, "loss": 0.639, "step": 1163 }, { "epoch": 0.04146997523914708, "grad_norm": 1.7861126570573296, "learning_rate": 1.999314046967338e-05, "loss": 0.6032, "step": 1164 }, { "epoch": 0.0415056023656412, "grad_norm": 3.2552411951443703, "learning_rate": 1.9993097669520712e-05, "loss": 0.626, "step": 1165 }, { "epoch": 0.04154122949213531, "grad_norm": 2.5978490799642344, "learning_rate": 1.9993054736302967e-05, "loss": 0.6826, "step": 1166 }, { "epoch": 0.04157685661862943, "grad_norm": 1.7487010140869588, "learning_rate": 1.999301167002071e-05, "loss": 0.5587, "step": 1167 }, { "epoch": 0.04161248374512354, "grad_norm": 1.503722548784248, "learning_rate": 1.999296847067452e-05, "loss": 0.5182, "step": 1168 }, { "epoch": 0.04164811087161765, "grad_norm": 3.867025321653274, "learning_rate": 1.999292513826497e-05, "loss": 0.6661, "step": 1169 }, { "epoch": 0.041683737998111765, "grad_norm": 2.0817802293546936, "learning_rate": 1.9992881672792637e-05, "loss": 0.6042, "step": 1170 }, { "epoch": 0.041719365124605876, "grad_norm": 1.8488190693451125, "learning_rate": 1.99928380742581e-05, "loss": 0.5632, "step": 1171 }, { "epoch": 0.041754992251099986, "grad_norm": 3.0416869551954813, "learning_rate": 1.9992794342661937e-05, "loss": 0.6641, "step": 1172 }, { "epoch": 0.0417906193775941, "grad_norm": 1.7246171050938188, "learning_rate": 1.999275047800474e-05, "loss": 0.635, "step": 1173 }, { "epoch": 0.04182624650408821, "grad_norm": 3.9029694949669063, "learning_rate": 1.9992706480287077e-05, "loss": 0.6442, "step": 1174 }, { "epoch": 0.04186187363058232, "grad_norm": 1.9373339122498874, "learning_rate": 1.9992662349509547e-05, "loss": 0.6743, "step": 1175 }, { "epoch": 0.04189750075707644, "grad_norm": 2.6046439154130585, "learning_rate": 1.9992618085672736e-05, "loss": 0.6084, "step": 1176 }, { "epoch": 0.04193312788357055, "grad_norm": 1.720146289570249, "learning_rate": 1.9992573688777226e-05, "loss": 0.5232, "step": 1177 }, { "epoch": 0.04196875501006466, "grad_norm": 1.6110675719888272, "learning_rate": 1.9992529158823616e-05, "loss": 0.5523, "step": 1178 }, { "epoch": 0.04200438213655878, "grad_norm": 1.5596941564056461, "learning_rate": 1.9992484495812496e-05, "loss": 0.5435, "step": 1179 }, { "epoch": 0.04204000926305289, "grad_norm": 2.0866990025369043, "learning_rate": 1.9992439699744457e-05, "loss": 0.6477, "step": 1180 }, { "epoch": 0.042075636389547, "grad_norm": 2.7835534857811486, "learning_rate": 1.99923947706201e-05, "loss": 0.6092, "step": 1181 }, { "epoch": 0.042111263516041116, "grad_norm": 1.8312672577901692, "learning_rate": 1.9992349708440026e-05, "loss": 0.5685, "step": 1182 }, { "epoch": 0.042146890642535226, "grad_norm": 2.034546391172599, "learning_rate": 1.999230451320483e-05, "loss": 0.6885, "step": 1183 }, { "epoch": 0.042182517769029336, "grad_norm": 1.6042529825061356, "learning_rate": 1.9992259184915115e-05, "loss": 0.5496, "step": 1184 }, { "epoch": 0.04221814489552345, "grad_norm": 2.5520714453653803, "learning_rate": 1.9992213723571482e-05, "loss": 0.6223, "step": 1185 }, { "epoch": 0.04225377202201756, "grad_norm": 2.5499029101800628, "learning_rate": 1.9992168129174544e-05, "loss": 0.6606, "step": 1186 }, { "epoch": 0.042289399148511674, "grad_norm": 1.7753715029116413, "learning_rate": 1.99921224017249e-05, "loss": 0.5602, "step": 1187 }, { "epoch": 0.04232502627500579, "grad_norm": 1.3696812515707033, "learning_rate": 1.9992076541223163e-05, "loss": 0.4569, "step": 1188 }, { "epoch": 0.0423606534014999, "grad_norm": 3.1965936680580924, "learning_rate": 1.999203054766994e-05, "loss": 0.6246, "step": 1189 }, { "epoch": 0.04239628052799402, "grad_norm": 1.4129570675561323, "learning_rate": 1.9991984421065852e-05, "loss": 0.5756, "step": 1190 }, { "epoch": 0.04243190765448813, "grad_norm": 2.7663171616745093, "learning_rate": 1.9991938161411505e-05, "loss": 0.7122, "step": 1191 }, { "epoch": 0.04246753478098224, "grad_norm": 1.542850176935414, "learning_rate": 1.9991891768707515e-05, "loss": 0.5429, "step": 1192 }, { "epoch": 0.042503161907476356, "grad_norm": 1.9215097874910991, "learning_rate": 1.9991845242954505e-05, "loss": 0.6305, "step": 1193 }, { "epoch": 0.042538789033970466, "grad_norm": 2.466821568427294, "learning_rate": 1.999179858415309e-05, "loss": 0.6304, "step": 1194 }, { "epoch": 0.042574416160464576, "grad_norm": 1.4786315871611544, "learning_rate": 1.9991751792303892e-05, "loss": 0.5909, "step": 1195 }, { "epoch": 0.04261004328695869, "grad_norm": 2.5756164336667378, "learning_rate": 1.9991704867407537e-05, "loss": 0.5909, "step": 1196 }, { "epoch": 0.0426456704134528, "grad_norm": 2.692575126354171, "learning_rate": 1.999165780946465e-05, "loss": 0.6443, "step": 1197 }, { "epoch": 0.042681297539946914, "grad_norm": 1.7261459331547326, "learning_rate": 1.9991610618475852e-05, "loss": 0.6123, "step": 1198 }, { "epoch": 0.04271692466644103, "grad_norm": 1.5833251429344297, "learning_rate": 1.9991563294441777e-05, "loss": 0.5625, "step": 1199 }, { "epoch": 0.04275255179293514, "grad_norm": 2.0424430962534506, "learning_rate": 1.999151583736305e-05, "loss": 0.5792, "step": 1200 }, { "epoch": 0.04278817891942925, "grad_norm": 1.882609199004718, "learning_rate": 1.9991468247240307e-05, "loss": 0.6811, "step": 1201 }, { "epoch": 0.04282380604592337, "grad_norm": 2.082346762133716, "learning_rate": 1.9991420524074183e-05, "loss": 0.5891, "step": 1202 }, { "epoch": 0.04285943317241748, "grad_norm": 2.413222571329888, "learning_rate": 1.999137266786531e-05, "loss": 0.6722, "step": 1203 }, { "epoch": 0.04289506029891159, "grad_norm": 1.5784807490225834, "learning_rate": 1.999132467861433e-05, "loss": 0.6161, "step": 1204 }, { "epoch": 0.042930687425405706, "grad_norm": 1.5685553439539666, "learning_rate": 1.999127655632187e-05, "loss": 0.5549, "step": 1205 }, { "epoch": 0.042966314551899816, "grad_norm": 1.8808184701533033, "learning_rate": 1.9991228300988586e-05, "loss": 0.7187, "step": 1206 }, { "epoch": 0.043001941678393926, "grad_norm": 2.905636283850455, "learning_rate": 1.9991179912615112e-05, "loss": 0.616, "step": 1207 }, { "epoch": 0.04303756880488804, "grad_norm": 1.5157355185753425, "learning_rate": 1.9991131391202093e-05, "loss": 0.55, "step": 1208 }, { "epoch": 0.043073195931382154, "grad_norm": 2.027208354059122, "learning_rate": 1.9991082736750176e-05, "loss": 0.655, "step": 1209 }, { "epoch": 0.043108823057876264, "grad_norm": 2.4280794442527514, "learning_rate": 1.9991033949260008e-05, "loss": 0.6645, "step": 1210 }, { "epoch": 0.04314445018437038, "grad_norm": 2.2842223281899807, "learning_rate": 1.9990985028732243e-05, "loss": 0.5758, "step": 1211 }, { "epoch": 0.04318007731086449, "grad_norm": 1.5177629440102516, "learning_rate": 1.999093597516753e-05, "loss": 0.5526, "step": 1212 }, { "epoch": 0.04321570443735861, "grad_norm": 1.4590279713943508, "learning_rate": 1.9990886788566515e-05, "loss": 0.5085, "step": 1213 }, { "epoch": 0.04325133156385272, "grad_norm": 2.2466704625837775, "learning_rate": 1.9990837468929864e-05, "loss": 0.7087, "step": 1214 }, { "epoch": 0.04328695869034683, "grad_norm": 2.479109396508944, "learning_rate": 1.9990788016258227e-05, "loss": 0.7302, "step": 1215 }, { "epoch": 0.043322585816840946, "grad_norm": 2.9071288691165553, "learning_rate": 1.9990738430552267e-05, "loss": 0.6671, "step": 1216 }, { "epoch": 0.043358212943335056, "grad_norm": 1.6624514879547059, "learning_rate": 1.999068871181264e-05, "loss": 0.5958, "step": 1217 }, { "epoch": 0.043393840069829166, "grad_norm": 1.5862046657818545, "learning_rate": 1.9990638860040007e-05, "loss": 0.6378, "step": 1218 }, { "epoch": 0.04342946719632328, "grad_norm": 1.5467405825845313, "learning_rate": 1.999058887523504e-05, "loss": 0.5772, "step": 1219 }, { "epoch": 0.043465094322817394, "grad_norm": 1.4979004704428556, "learning_rate": 1.9990538757398394e-05, "loss": 0.5409, "step": 1220 }, { "epoch": 0.043500721449311504, "grad_norm": 2.7156075205884282, "learning_rate": 1.9990488506530746e-05, "loss": 0.7206, "step": 1221 }, { "epoch": 0.04353634857580562, "grad_norm": 2.232361201878065, "learning_rate": 1.9990438122632757e-05, "loss": 0.6701, "step": 1222 }, { "epoch": 0.04357197570229973, "grad_norm": 2.2876724146374845, "learning_rate": 1.9990387605705104e-05, "loss": 0.664, "step": 1223 }, { "epoch": 0.04360760282879384, "grad_norm": 2.4740950161758555, "learning_rate": 1.9990336955748456e-05, "loss": 0.6315, "step": 1224 }, { "epoch": 0.04364322995528796, "grad_norm": 1.465836068265387, "learning_rate": 1.999028617276349e-05, "loss": 0.5369, "step": 1225 }, { "epoch": 0.04367885708178207, "grad_norm": 2.227181431207388, "learning_rate": 1.9990235256750878e-05, "loss": 0.6536, "step": 1226 }, { "epoch": 0.04371448420827618, "grad_norm": 2.4906110691648555, "learning_rate": 1.99901842077113e-05, "loss": 0.6017, "step": 1227 }, { "epoch": 0.043750111334770296, "grad_norm": 1.590419746935351, "learning_rate": 1.999013302564544e-05, "loss": 0.5855, "step": 1228 }, { "epoch": 0.043785738461264406, "grad_norm": 2.61987748056039, "learning_rate": 1.9990081710553976e-05, "loss": 0.6787, "step": 1229 }, { "epoch": 0.043821365587758516, "grad_norm": 2.6305743657563454, "learning_rate": 1.999003026243759e-05, "loss": 0.642, "step": 1230 }, { "epoch": 0.043856992714252634, "grad_norm": 1.714285890137128, "learning_rate": 1.9989978681296968e-05, "loss": 0.5522, "step": 1231 }, { "epoch": 0.043892619840746744, "grad_norm": 1.651122830451584, "learning_rate": 1.9989926967132798e-05, "loss": 0.5809, "step": 1232 }, { "epoch": 0.043928246967240854, "grad_norm": 1.622625653933105, "learning_rate": 1.998987511994577e-05, "loss": 0.6362, "step": 1233 }, { "epoch": 0.04396387409373497, "grad_norm": 2.49152923342253, "learning_rate": 1.998982313973657e-05, "loss": 0.7006, "step": 1234 }, { "epoch": 0.04399950122022908, "grad_norm": 1.553048436164289, "learning_rate": 1.998977102650589e-05, "loss": 0.5545, "step": 1235 }, { "epoch": 0.04403512834672319, "grad_norm": 2.42747776905402, "learning_rate": 1.998971878025443e-05, "loss": 0.6928, "step": 1236 }, { "epoch": 0.04407075547321731, "grad_norm": 2.654422233776336, "learning_rate": 1.998966640098288e-05, "loss": 0.7216, "step": 1237 }, { "epoch": 0.04410638259971142, "grad_norm": 1.817496854399247, "learning_rate": 1.998961388869194e-05, "loss": 0.4674, "step": 1238 }, { "epoch": 0.044142009726205536, "grad_norm": 1.594070927640967, "learning_rate": 1.9989561243382313e-05, "loss": 0.6226, "step": 1239 }, { "epoch": 0.044177636852699646, "grad_norm": 2.299866971666634, "learning_rate": 1.998950846505469e-05, "loss": 0.7063, "step": 1240 }, { "epoch": 0.044213263979193757, "grad_norm": 2.4445699687066855, "learning_rate": 1.9989455553709784e-05, "loss": 0.6525, "step": 1241 }, { "epoch": 0.044248891105687874, "grad_norm": 2.808609663454424, "learning_rate": 1.998940250934829e-05, "loss": 0.6011, "step": 1242 }, { "epoch": 0.044284518232181984, "grad_norm": 2.5718455077937477, "learning_rate": 1.998934933197092e-05, "loss": 0.7311, "step": 1243 }, { "epoch": 0.044320145358676094, "grad_norm": 2.0786424835828714, "learning_rate": 1.9989296021578382e-05, "loss": 0.5741, "step": 1244 }, { "epoch": 0.04435577248517021, "grad_norm": 1.741609464808086, "learning_rate": 1.998924257817139e-05, "loss": 0.5558, "step": 1245 }, { "epoch": 0.04439139961166432, "grad_norm": 1.6037117833956818, "learning_rate": 1.998918900175065e-05, "loss": 0.5625, "step": 1246 }, { "epoch": 0.04442702673815843, "grad_norm": 2.4561456313716183, "learning_rate": 1.998913529231687e-05, "loss": 0.6594, "step": 1247 }, { "epoch": 0.04446265386465255, "grad_norm": 2.19368701541492, "learning_rate": 1.998908144987078e-05, "loss": 0.656, "step": 1248 }, { "epoch": 0.04449828099114666, "grad_norm": 2.8096001050103983, "learning_rate": 1.9989027474413083e-05, "loss": 0.6778, "step": 1249 }, { "epoch": 0.04453390811764077, "grad_norm": 2.6280014053264096, "learning_rate": 1.9988973365944505e-05, "loss": 0.8227, "step": 1250 }, { "epoch": 0.044569535244134886, "grad_norm": 1.7193276586122197, "learning_rate": 1.9988919124465765e-05, "loss": 0.5709, "step": 1251 }, { "epoch": 0.044605162370628997, "grad_norm": 3.1190750670410705, "learning_rate": 1.9988864749977584e-05, "loss": 0.7179, "step": 1252 }, { "epoch": 0.04464078949712311, "grad_norm": 1.5378909473439173, "learning_rate": 1.9988810242480686e-05, "loss": 0.5382, "step": 1253 }, { "epoch": 0.044676416623617224, "grad_norm": 2.2758424157387576, "learning_rate": 1.99887556019758e-05, "loss": 0.6244, "step": 1254 }, { "epoch": 0.044712043750111334, "grad_norm": 2.4222193622908543, "learning_rate": 1.998870082846365e-05, "loss": 0.5974, "step": 1255 }, { "epoch": 0.044747670876605444, "grad_norm": 1.8392815215877298, "learning_rate": 1.9988645921944968e-05, "loss": 0.5579, "step": 1256 }, { "epoch": 0.04478329800309956, "grad_norm": 2.464290267588288, "learning_rate": 1.9988590882420486e-05, "loss": 0.574, "step": 1257 }, { "epoch": 0.04481892512959367, "grad_norm": 2.0727527137544013, "learning_rate": 1.998853570989093e-05, "loss": 0.4806, "step": 1258 }, { "epoch": 0.04485455225608778, "grad_norm": 1.836638806517764, "learning_rate": 1.9988480404357044e-05, "loss": 0.5973, "step": 1259 }, { "epoch": 0.0448901793825819, "grad_norm": 2.8162748458585347, "learning_rate": 1.9988424965819556e-05, "loss": 0.6993, "step": 1260 }, { "epoch": 0.04492580650907601, "grad_norm": 1.8064630300771403, "learning_rate": 1.998836939427921e-05, "loss": 0.5351, "step": 1261 }, { "epoch": 0.044961433635570126, "grad_norm": 1.5840377569070785, "learning_rate": 1.9988313689736746e-05, "loss": 0.573, "step": 1262 }, { "epoch": 0.04499706076206424, "grad_norm": 3.166300502625975, "learning_rate": 1.99882578521929e-05, "loss": 0.6934, "step": 1263 }, { "epoch": 0.04503268788855835, "grad_norm": 2.707986253851071, "learning_rate": 1.998820188164842e-05, "loss": 0.6114, "step": 1264 }, { "epoch": 0.045068315015052464, "grad_norm": 1.5971613286052404, "learning_rate": 1.9988145778104055e-05, "loss": 0.5183, "step": 1265 }, { "epoch": 0.045103942141546574, "grad_norm": 2.1373811986552846, "learning_rate": 1.9988089541560543e-05, "loss": 0.6121, "step": 1266 }, { "epoch": 0.045139569268040684, "grad_norm": 2.149268864290413, "learning_rate": 1.998803317201864e-05, "loss": 0.5583, "step": 1267 }, { "epoch": 0.0451751963945348, "grad_norm": 2.373791045656053, "learning_rate": 1.998797666947909e-05, "loss": 0.6988, "step": 1268 }, { "epoch": 0.04521082352102891, "grad_norm": 1.7317352946877864, "learning_rate": 1.9987920033942653e-05, "loss": 0.5483, "step": 1269 }, { "epoch": 0.04524645064752302, "grad_norm": 1.8653461844142303, "learning_rate": 1.998786326541008e-05, "loss": 0.5683, "step": 1270 }, { "epoch": 0.04528207777401714, "grad_norm": 1.7200773402336453, "learning_rate": 1.9987806363882124e-05, "loss": 0.5516, "step": 1271 }, { "epoch": 0.04531770490051125, "grad_norm": 1.9744438904429835, "learning_rate": 1.998774932935955e-05, "loss": 0.5606, "step": 1272 }, { "epoch": 0.04535333202700536, "grad_norm": 3.9076942234131944, "learning_rate": 1.9987692161843107e-05, "loss": 0.6685, "step": 1273 }, { "epoch": 0.04538895915349948, "grad_norm": 3.2952744079648943, "learning_rate": 1.998763486133356e-05, "loss": 0.6566, "step": 1274 }, { "epoch": 0.04542458627999359, "grad_norm": 3.3081079902075263, "learning_rate": 1.998757742783168e-05, "loss": 0.6304, "step": 1275 }, { "epoch": 0.0454602134064877, "grad_norm": 2.462858625633849, "learning_rate": 1.9987519861338226e-05, "loss": 0.6813, "step": 1276 }, { "epoch": 0.045495840532981814, "grad_norm": 2.9411268938339914, "learning_rate": 1.998746216185396e-05, "loss": 0.7278, "step": 1277 }, { "epoch": 0.045531467659475924, "grad_norm": 3.330017485143286, "learning_rate": 1.998740432937966e-05, "loss": 0.6413, "step": 1278 }, { "epoch": 0.045567094785970035, "grad_norm": 2.483982704265938, "learning_rate": 1.9987346363916085e-05, "loss": 0.6395, "step": 1279 }, { "epoch": 0.04560272191246415, "grad_norm": 3.0845111302223556, "learning_rate": 1.9987288265464017e-05, "loss": 0.7181, "step": 1280 }, { "epoch": 0.04563834903895826, "grad_norm": 2.0506816311308347, "learning_rate": 1.9987230034024222e-05, "loss": 0.5419, "step": 1281 }, { "epoch": 0.04567397616545237, "grad_norm": 2.022436442131202, "learning_rate": 1.9987171669597476e-05, "loss": 0.5823, "step": 1282 }, { "epoch": 0.04570960329194649, "grad_norm": 2.4447251990723298, "learning_rate": 1.9987113172184562e-05, "loss": 0.6577, "step": 1283 }, { "epoch": 0.0457452304184406, "grad_norm": 1.7787789633500044, "learning_rate": 1.9987054541786256e-05, "loss": 0.5717, "step": 1284 }, { "epoch": 0.04578085754493472, "grad_norm": 1.4310244283481084, "learning_rate": 1.9986995778403336e-05, "loss": 0.5374, "step": 1285 }, { "epoch": 0.04581648467142883, "grad_norm": 1.6014069140018405, "learning_rate": 1.998693688203659e-05, "loss": 0.5467, "step": 1286 }, { "epoch": 0.04585211179792294, "grad_norm": 1.7114563697191598, "learning_rate": 1.9986877852686794e-05, "loss": 0.5601, "step": 1287 }, { "epoch": 0.045887738924417054, "grad_norm": 1.6126609742748914, "learning_rate": 1.9986818690354745e-05, "loss": 0.5468, "step": 1288 }, { "epoch": 0.045923366050911164, "grad_norm": 2.6897469340943636, "learning_rate": 1.998675939504122e-05, "loss": 0.6302, "step": 1289 }, { "epoch": 0.045958993177405275, "grad_norm": 1.565773362142335, "learning_rate": 1.9986699966747018e-05, "loss": 0.5844, "step": 1290 }, { "epoch": 0.04599462030389939, "grad_norm": 1.591766344392843, "learning_rate": 1.998664040547292e-05, "loss": 0.5946, "step": 1291 }, { "epoch": 0.0460302474303935, "grad_norm": 1.3714745077589807, "learning_rate": 1.9986580711219726e-05, "loss": 0.5491, "step": 1292 }, { "epoch": 0.04606587455688761, "grad_norm": 1.4754171001205696, "learning_rate": 1.9986520883988233e-05, "loss": 0.536, "step": 1293 }, { "epoch": 0.04610150168338173, "grad_norm": 1.4535645368004992, "learning_rate": 1.998646092377923e-05, "loss": 0.5585, "step": 1294 }, { "epoch": 0.04613712880987584, "grad_norm": 2.245338684481994, "learning_rate": 1.9986400830593524e-05, "loss": 0.6635, "step": 1295 }, { "epoch": 0.04617275593636995, "grad_norm": 1.5443522911072718, "learning_rate": 1.9986340604431908e-05, "loss": 0.5718, "step": 1296 }, { "epoch": 0.04620838306286407, "grad_norm": 2.5589722932075407, "learning_rate": 1.9986280245295186e-05, "loss": 0.6267, "step": 1297 }, { "epoch": 0.04624401018935818, "grad_norm": 1.5359813023465252, "learning_rate": 1.9986219753184167e-05, "loss": 0.5276, "step": 1298 }, { "epoch": 0.04627963731585229, "grad_norm": 8.936750341342691, "learning_rate": 1.998615912809965e-05, "loss": 0.6568, "step": 1299 }, { "epoch": 0.046315264442346404, "grad_norm": 1.696037576943752, "learning_rate": 1.9986098370042443e-05, "loss": 0.5677, "step": 1300 }, { "epoch": 0.046350891568840515, "grad_norm": 1.6682040076287972, "learning_rate": 1.9986037479013357e-05, "loss": 0.6179, "step": 1301 }, { "epoch": 0.046386518695334625, "grad_norm": 4.741272180346173, "learning_rate": 1.9985976455013203e-05, "loss": 0.745, "step": 1302 }, { "epoch": 0.04642214582182874, "grad_norm": 2.8463123711756335, "learning_rate": 1.9985915298042792e-05, "loss": 0.6101, "step": 1303 }, { "epoch": 0.04645777294832285, "grad_norm": 4.178259145175184, "learning_rate": 1.998585400810294e-05, "loss": 0.6958, "step": 1304 }, { "epoch": 0.04649340007481696, "grad_norm": 3.84299473463075, "learning_rate": 1.998579258519446e-05, "loss": 0.6799, "step": 1305 }, { "epoch": 0.04652902720131108, "grad_norm": 1.8837707656017177, "learning_rate": 1.998573102931817e-05, "loss": 0.5355, "step": 1306 }, { "epoch": 0.04656465432780519, "grad_norm": 2.056609148758401, "learning_rate": 1.9985669340474894e-05, "loss": 0.5054, "step": 1307 }, { "epoch": 0.04660028145429931, "grad_norm": 3.5823351156389696, "learning_rate": 1.9985607518665456e-05, "loss": 0.7376, "step": 1308 }, { "epoch": 0.04663590858079342, "grad_norm": 4.257565437616267, "learning_rate": 1.998554556389067e-05, "loss": 0.6157, "step": 1309 }, { "epoch": 0.04667153570728753, "grad_norm": 2.324255517706289, "learning_rate": 1.9985483476151362e-05, "loss": 0.4767, "step": 1310 }, { "epoch": 0.046707162833781644, "grad_norm": 1.5974989434756106, "learning_rate": 1.998542125544837e-05, "loss": 0.5538, "step": 1311 }, { "epoch": 0.046742789960275755, "grad_norm": 2.178494766020782, "learning_rate": 1.998535890178251e-05, "loss": 0.7217, "step": 1312 }, { "epoch": 0.046778417086769865, "grad_norm": 2.9812762326272364, "learning_rate": 1.9985296415154614e-05, "loss": 0.6958, "step": 1313 }, { "epoch": 0.04681404421326398, "grad_norm": 2.1790608823172093, "learning_rate": 1.998523379556552e-05, "loss": 0.6658, "step": 1314 }, { "epoch": 0.04684967133975809, "grad_norm": 1.9053078734790012, "learning_rate": 1.998517104301606e-05, "loss": 0.5945, "step": 1315 }, { "epoch": 0.0468852984662522, "grad_norm": 1.614308829617795, "learning_rate": 1.998510815750707e-05, "loss": 0.645, "step": 1316 }, { "epoch": 0.04692092559274632, "grad_norm": 1.7849192154209297, "learning_rate": 1.9985045139039383e-05, "loss": 0.6157, "step": 1317 }, { "epoch": 0.04695655271924043, "grad_norm": 1.4369963173387073, "learning_rate": 1.998498198761384e-05, "loss": 0.544, "step": 1318 }, { "epoch": 0.04699217984573454, "grad_norm": 1.444176263180057, "learning_rate": 1.9984918703231285e-05, "loss": 0.5896, "step": 1319 }, { "epoch": 0.04702780697222866, "grad_norm": 1.751272797020499, "learning_rate": 1.9984855285892555e-05, "loss": 0.568, "step": 1320 }, { "epoch": 0.04706343409872277, "grad_norm": 4.184987517626978, "learning_rate": 1.99847917355985e-05, "loss": 0.6389, "step": 1321 }, { "epoch": 0.04709906122521688, "grad_norm": 2.2463248095822976, "learning_rate": 1.9984728052349965e-05, "loss": 0.6929, "step": 1322 }, { "epoch": 0.047134688351710995, "grad_norm": 2.7406014883066776, "learning_rate": 1.9984664236147795e-05, "loss": 0.6393, "step": 1323 }, { "epoch": 0.047170315478205105, "grad_norm": 2.290510301337982, "learning_rate": 1.998460028699284e-05, "loss": 0.6815, "step": 1324 }, { "epoch": 0.047205942604699215, "grad_norm": 2.1693057152612085, "learning_rate": 1.9984536204885957e-05, "loss": 0.5815, "step": 1325 }, { "epoch": 0.04724156973119333, "grad_norm": 2.5172434250167206, "learning_rate": 1.9984471989827996e-05, "loss": 0.6918, "step": 1326 }, { "epoch": 0.04727719685768744, "grad_norm": 1.7577320547055404, "learning_rate": 1.9984407641819812e-05, "loss": 0.6177, "step": 1327 }, { "epoch": 0.04731282398418155, "grad_norm": 2.302484709109473, "learning_rate": 1.998434316086226e-05, "loss": 0.6122, "step": 1328 }, { "epoch": 0.04734845111067567, "grad_norm": 1.6926849093064744, "learning_rate": 1.9984278546956202e-05, "loss": 0.5797, "step": 1329 }, { "epoch": 0.04738407823716978, "grad_norm": 2.0090506903792487, "learning_rate": 1.9984213800102494e-05, "loss": 0.6075, "step": 1330 }, { "epoch": 0.0474197053636639, "grad_norm": 1.9924599253843416, "learning_rate": 1.9984148920302e-05, "loss": 0.585, "step": 1331 }, { "epoch": 0.04745533249015801, "grad_norm": 1.594139607675426, "learning_rate": 1.9984083907555587e-05, "loss": 0.5553, "step": 1332 }, { "epoch": 0.04749095961665212, "grad_norm": 2.0877322601694095, "learning_rate": 1.998401876186412e-05, "loss": 0.6117, "step": 1333 }, { "epoch": 0.047526586743146235, "grad_norm": 1.6173978216465144, "learning_rate": 1.998395348322846e-05, "loss": 0.5166, "step": 1334 }, { "epoch": 0.047562213869640345, "grad_norm": 2.094858569752342, "learning_rate": 1.998388807164948e-05, "loss": 0.6081, "step": 1335 }, { "epoch": 0.047597840996134455, "grad_norm": 2.4813563653498276, "learning_rate": 1.9983822527128058e-05, "loss": 0.7018, "step": 1336 }, { "epoch": 0.04763346812262857, "grad_norm": 2.007083932877898, "learning_rate": 1.9983756849665058e-05, "loss": 0.6555, "step": 1337 }, { "epoch": 0.04766909524912268, "grad_norm": 1.5519639884161647, "learning_rate": 1.9983691039261358e-05, "loss": 0.5881, "step": 1338 }, { "epoch": 0.04770472237561679, "grad_norm": 1.6377932689448809, "learning_rate": 1.9983625095917833e-05, "loss": 0.5819, "step": 1339 }, { "epoch": 0.04774034950211091, "grad_norm": 2.1329553837910784, "learning_rate": 1.9983559019635363e-05, "loss": 0.6203, "step": 1340 }, { "epoch": 0.04777597662860502, "grad_norm": 2.2956386954807644, "learning_rate": 1.9983492810414823e-05, "loss": 0.676, "step": 1341 }, { "epoch": 0.04781160375509913, "grad_norm": 1.5808113721743962, "learning_rate": 1.99834264682571e-05, "loss": 0.587, "step": 1342 }, { "epoch": 0.04784723088159325, "grad_norm": 1.3732687100420031, "learning_rate": 1.998335999316308e-05, "loss": 0.556, "step": 1343 }, { "epoch": 0.04788285800808736, "grad_norm": 1.99679724048243, "learning_rate": 1.9983293385133636e-05, "loss": 0.6132, "step": 1344 }, { "epoch": 0.04791848513458147, "grad_norm": 1.9704347809647307, "learning_rate": 1.9983226644169668e-05, "loss": 0.6605, "step": 1345 }, { "epoch": 0.047954112261075585, "grad_norm": 2.3682254276174652, "learning_rate": 1.9983159770272057e-05, "loss": 0.6601, "step": 1346 }, { "epoch": 0.047989739387569695, "grad_norm": 2.002402569581471, "learning_rate": 1.9983092763441698e-05, "loss": 0.6423, "step": 1347 }, { "epoch": 0.048025366514063805, "grad_norm": 1.7011531552881225, "learning_rate": 1.9983025623679483e-05, "loss": 0.6033, "step": 1348 }, { "epoch": 0.04806099364055792, "grad_norm": 1.463165289366805, "learning_rate": 1.9982958350986296e-05, "loss": 0.5453, "step": 1349 }, { "epoch": 0.04809662076705203, "grad_norm": 1.5027484166982747, "learning_rate": 1.9982890945363048e-05, "loss": 0.5453, "step": 1350 }, { "epoch": 0.04813224789354614, "grad_norm": 2.0799487706972135, "learning_rate": 1.9982823406810627e-05, "loss": 0.6728, "step": 1351 }, { "epoch": 0.04816787502004026, "grad_norm": 1.5075385859529544, "learning_rate": 1.9982755735329937e-05, "loss": 0.5689, "step": 1352 }, { "epoch": 0.04820350214653437, "grad_norm": 1.9885894233372279, "learning_rate": 1.9982687930921873e-05, "loss": 0.6416, "step": 1353 }, { "epoch": 0.04823912927302849, "grad_norm": 1.4746197599993442, "learning_rate": 1.9982619993587343e-05, "loss": 0.5435, "step": 1354 }, { "epoch": 0.0482747563995226, "grad_norm": 1.955224398953386, "learning_rate": 1.9982551923327255e-05, "loss": 0.5676, "step": 1355 }, { "epoch": 0.04831038352601671, "grad_norm": 2.0964502194899377, "learning_rate": 1.9982483720142506e-05, "loss": 0.5941, "step": 1356 }, { "epoch": 0.048346010652510825, "grad_norm": 5.156310834604046, "learning_rate": 1.998241538403401e-05, "loss": 0.6004, "step": 1357 }, { "epoch": 0.048381637779004935, "grad_norm": 1.5033634038899804, "learning_rate": 1.998234691500267e-05, "loss": 0.5674, "step": 1358 }, { "epoch": 0.048417264905499045, "grad_norm": 1.3680837011885096, "learning_rate": 1.998227831304941e-05, "loss": 0.5343, "step": 1359 }, { "epoch": 0.04845289203199316, "grad_norm": 2.1975874685609385, "learning_rate": 1.9982209578175136e-05, "loss": 0.6448, "step": 1360 }, { "epoch": 0.04848851915848727, "grad_norm": 1.5888247978927104, "learning_rate": 1.9982140710380765e-05, "loss": 0.5945, "step": 1361 }, { "epoch": 0.04852414628498138, "grad_norm": 1.27133330737476, "learning_rate": 1.9982071709667216e-05, "loss": 0.5406, "step": 1362 }, { "epoch": 0.0485597734114755, "grad_norm": 1.2798213784495467, "learning_rate": 1.9982002576035403e-05, "loss": 0.5307, "step": 1363 }, { "epoch": 0.04859540053796961, "grad_norm": 1.3986794855386326, "learning_rate": 1.9981933309486245e-05, "loss": 0.5263, "step": 1364 }, { "epoch": 0.04863102766446372, "grad_norm": 1.323684411088962, "learning_rate": 1.998186391002067e-05, "loss": 0.5424, "step": 1365 }, { "epoch": 0.04866665479095784, "grad_norm": 1.3941986403212812, "learning_rate": 1.9981794377639603e-05, "loss": 0.5359, "step": 1366 }, { "epoch": 0.04870228191745195, "grad_norm": 2.5408309958207878, "learning_rate": 1.9981724712343963e-05, "loss": 0.6225, "step": 1367 }, { "epoch": 0.04873790904394606, "grad_norm": 2.360086036085224, "learning_rate": 1.9981654914134684e-05, "loss": 0.658, "step": 1368 }, { "epoch": 0.048773536170440175, "grad_norm": 2.1739970019970287, "learning_rate": 1.9981584983012694e-05, "loss": 0.6669, "step": 1369 }, { "epoch": 0.048809163296934285, "grad_norm": 2.083140927001827, "learning_rate": 1.998151491897892e-05, "loss": 0.6363, "step": 1370 }, { "epoch": 0.048844790423428396, "grad_norm": 2.8802840497129365, "learning_rate": 1.99814447220343e-05, "loss": 0.6579, "step": 1371 }, { "epoch": 0.04888041754992251, "grad_norm": 1.7976948401444937, "learning_rate": 1.9981374392179765e-05, "loss": 0.5541, "step": 1372 }, { "epoch": 0.04891604467641662, "grad_norm": 1.5757913202578104, "learning_rate": 1.998130392941626e-05, "loss": 0.5868, "step": 1373 }, { "epoch": 0.04895167180291073, "grad_norm": 2.113022308995278, "learning_rate": 1.9981233333744708e-05, "loss": 0.7421, "step": 1374 }, { "epoch": 0.04898729892940485, "grad_norm": 1.554047427704514, "learning_rate": 1.998116260516606e-05, "loss": 0.5511, "step": 1375 }, { "epoch": 0.04902292605589896, "grad_norm": 1.7956991752733746, "learning_rate": 1.998109174368126e-05, "loss": 0.6004, "step": 1376 }, { "epoch": 0.04905855318239307, "grad_norm": 1.8540772874325324, "learning_rate": 1.9981020749291246e-05, "loss": 0.5347, "step": 1377 }, { "epoch": 0.04909418030888719, "grad_norm": 1.9850289042397251, "learning_rate": 1.9980949621996962e-05, "loss": 0.602, "step": 1378 }, { "epoch": 0.0491298074353813, "grad_norm": 1.605201208792769, "learning_rate": 1.9980878361799356e-05, "loss": 0.5844, "step": 1379 }, { "epoch": 0.049165434561875415, "grad_norm": 3.1493616765935837, "learning_rate": 1.998080696869938e-05, "loss": 0.5831, "step": 1380 }, { "epoch": 0.049201061688369525, "grad_norm": 2.299995878717782, "learning_rate": 1.9980735442697984e-05, "loss": 0.6527, "step": 1381 }, { "epoch": 0.049236688814863636, "grad_norm": 2.5708548507717484, "learning_rate": 1.998066378379612e-05, "loss": 0.6056, "step": 1382 }, { "epoch": 0.04927231594135775, "grad_norm": 3.2473324661277827, "learning_rate": 1.998059199199474e-05, "loss": 0.6844, "step": 1383 }, { "epoch": 0.04930794306785186, "grad_norm": 2.311568531247717, "learning_rate": 1.9980520067294803e-05, "loss": 0.6665, "step": 1384 }, { "epoch": 0.04934357019434597, "grad_norm": 1.4380291795341877, "learning_rate": 1.9980448009697264e-05, "loss": 0.4981, "step": 1385 }, { "epoch": 0.04937919732084009, "grad_norm": 1.4338741904949435, "learning_rate": 1.9980375819203084e-05, "loss": 0.5505, "step": 1386 }, { "epoch": 0.0494148244473342, "grad_norm": 1.3672751590129157, "learning_rate": 1.9980303495813227e-05, "loss": 0.5728, "step": 1387 }, { "epoch": 0.04945045157382831, "grad_norm": 1.5960248622623148, "learning_rate": 1.998023103952865e-05, "loss": 0.5988, "step": 1388 }, { "epoch": 0.04948607870032243, "grad_norm": 2.139137243076183, "learning_rate": 1.998015845035032e-05, "loss": 0.6435, "step": 1389 }, { "epoch": 0.04952170582681654, "grad_norm": 1.584232345215362, "learning_rate": 1.9980085728279207e-05, "loss": 0.5858, "step": 1390 }, { "epoch": 0.04955733295331065, "grad_norm": 1.2794653118955066, "learning_rate": 1.9980012873316276e-05, "loss": 0.5466, "step": 1391 }, { "epoch": 0.049592960079804765, "grad_norm": 2.5414166907177225, "learning_rate": 1.9979939885462498e-05, "loss": 0.6723, "step": 1392 }, { "epoch": 0.049628587206298876, "grad_norm": 1.5440701827275385, "learning_rate": 1.9979866764718846e-05, "loss": 0.6039, "step": 1393 }, { "epoch": 0.049664214332792986, "grad_norm": 2.6267235225863903, "learning_rate": 1.997979351108629e-05, "loss": 0.7021, "step": 1394 }, { "epoch": 0.0496998414592871, "grad_norm": 2.1767897416764495, "learning_rate": 1.997972012456581e-05, "loss": 0.6223, "step": 1395 }, { "epoch": 0.04973546858578121, "grad_norm": 1.6221528633846254, "learning_rate": 1.997964660515838e-05, "loss": 0.5724, "step": 1396 }, { "epoch": 0.04977109571227532, "grad_norm": 1.572035763287513, "learning_rate": 1.9979572952864982e-05, "loss": 0.5574, "step": 1397 }, { "epoch": 0.04980672283876944, "grad_norm": 1.392849856192076, "learning_rate": 1.9979499167686592e-05, "loss": 0.5391, "step": 1398 }, { "epoch": 0.04984234996526355, "grad_norm": 1.3957154504241338, "learning_rate": 1.9979425249624196e-05, "loss": 0.4801, "step": 1399 }, { "epoch": 0.04987797709175766, "grad_norm": 2.2881806401069396, "learning_rate": 1.997935119867878e-05, "loss": 0.6871, "step": 1400 }, { "epoch": 0.04991360421825178, "grad_norm": 2.5205773083555187, "learning_rate": 1.9979277014851328e-05, "loss": 0.687, "step": 1401 }, { "epoch": 0.04994923134474589, "grad_norm": 1.9356093655812745, "learning_rate": 1.9979202698142823e-05, "loss": 0.6165, "step": 1402 }, { "epoch": 0.049984858471240005, "grad_norm": 1.9312577886568418, "learning_rate": 1.9979128248554263e-05, "loss": 0.5507, "step": 1403 }, { "epoch": 0.050020485597734116, "grad_norm": 2.4135157473793587, "learning_rate": 1.9979053666086633e-05, "loss": 0.585, "step": 1404 }, { "epoch": 0.050056112724228226, "grad_norm": 2.940071729955509, "learning_rate": 1.9978978950740927e-05, "loss": 0.6593, "step": 1405 }, { "epoch": 0.05009173985072234, "grad_norm": 1.915838652892532, "learning_rate": 1.9978904102518143e-05, "loss": 0.6458, "step": 1406 }, { "epoch": 0.05012736697721645, "grad_norm": 1.5162165748907193, "learning_rate": 1.997882912141928e-05, "loss": 0.5523, "step": 1407 }, { "epoch": 0.05016299410371056, "grad_norm": 2.0441619086047087, "learning_rate": 1.9978754007445328e-05, "loss": 0.7254, "step": 1408 }, { "epoch": 0.05019862123020468, "grad_norm": 2.530934561126216, "learning_rate": 1.9978678760597293e-05, "loss": 0.6259, "step": 1409 }, { "epoch": 0.05023424835669879, "grad_norm": 2.0013411510867165, "learning_rate": 1.9978603380876175e-05, "loss": 0.6379, "step": 1410 }, { "epoch": 0.0502698754831929, "grad_norm": 1.632281263773251, "learning_rate": 1.997852786828298e-05, "loss": 0.5993, "step": 1411 }, { "epoch": 0.05030550260968702, "grad_norm": 1.9928525960480203, "learning_rate": 1.9978452222818712e-05, "loss": 0.6367, "step": 1412 }, { "epoch": 0.05034112973618113, "grad_norm": 1.9197062063506887, "learning_rate": 1.9978376444484374e-05, "loss": 0.5925, "step": 1413 }, { "epoch": 0.05037675686267524, "grad_norm": 1.3682826432871298, "learning_rate": 1.9978300533280984e-05, "loss": 0.5229, "step": 1414 }, { "epoch": 0.050412383989169356, "grad_norm": 1.4751760799943703, "learning_rate": 1.9978224489209547e-05, "loss": 0.5309, "step": 1415 }, { "epoch": 0.050448011115663466, "grad_norm": 1.461302034393572, "learning_rate": 1.9978148312271072e-05, "loss": 0.5537, "step": 1416 }, { "epoch": 0.050483638242157576, "grad_norm": 1.5430123981307964, "learning_rate": 1.9978072002466585e-05, "loss": 0.5904, "step": 1417 }, { "epoch": 0.05051926536865169, "grad_norm": 1.3250808206021472, "learning_rate": 1.997799555979709e-05, "loss": 0.5423, "step": 1418 }, { "epoch": 0.0505548924951458, "grad_norm": 1.381211278533058, "learning_rate": 1.997791898426361e-05, "loss": 0.6125, "step": 1419 }, { "epoch": 0.050590519621639914, "grad_norm": 1.8759392792301919, "learning_rate": 1.9977842275867167e-05, "loss": 0.5844, "step": 1420 }, { "epoch": 0.05062614674813403, "grad_norm": 1.410984276987067, "learning_rate": 1.9977765434608778e-05, "loss": 0.5582, "step": 1421 }, { "epoch": 0.05066177387462814, "grad_norm": 1.57433675578162, "learning_rate": 1.997768846048947e-05, "loss": 0.59, "step": 1422 }, { "epoch": 0.05069740100112225, "grad_norm": 1.4417770687881248, "learning_rate": 1.9977611353510268e-05, "loss": 0.5607, "step": 1423 }, { "epoch": 0.05073302812761637, "grad_norm": 2.3923981288549974, "learning_rate": 1.9977534113672194e-05, "loss": 0.6476, "step": 1424 }, { "epoch": 0.05076865525411048, "grad_norm": 2.03276216316513, "learning_rate": 1.9977456740976277e-05, "loss": 0.6759, "step": 1425 }, { "epoch": 0.050804282380604596, "grad_norm": 2.2009171961307095, "learning_rate": 1.9977379235423553e-05, "loss": 0.6826, "step": 1426 }, { "epoch": 0.050839909507098706, "grad_norm": 4.62127572415103, "learning_rate": 1.9977301597015047e-05, "loss": 0.6309, "step": 1427 }, { "epoch": 0.050875536633592816, "grad_norm": 1.4174693740212498, "learning_rate": 1.9977223825751802e-05, "loss": 0.5406, "step": 1428 }, { "epoch": 0.05091116376008693, "grad_norm": 2.044735555696429, "learning_rate": 1.9977145921634848e-05, "loss": 0.569, "step": 1429 }, { "epoch": 0.05094679088658104, "grad_norm": 1.4772513830992775, "learning_rate": 1.997706788466522e-05, "loss": 0.5665, "step": 1430 }, { "epoch": 0.050982418013075154, "grad_norm": 1.488093350338321, "learning_rate": 1.997698971484396e-05, "loss": 0.6231, "step": 1431 }, { "epoch": 0.05101804513956927, "grad_norm": 2.0834938917466133, "learning_rate": 1.9976911412172106e-05, "loss": 0.6947, "step": 1432 }, { "epoch": 0.05105367226606338, "grad_norm": 1.448001715696406, "learning_rate": 1.9976832976650705e-05, "loss": 0.5587, "step": 1433 }, { "epoch": 0.05108929939255749, "grad_norm": 2.0018332507363104, "learning_rate": 1.99767544082808e-05, "loss": 0.6032, "step": 1434 }, { "epoch": 0.05112492651905161, "grad_norm": 2.049515941349042, "learning_rate": 1.9976675707063436e-05, "loss": 0.6779, "step": 1435 }, { "epoch": 0.05116055364554572, "grad_norm": 1.452044332827533, "learning_rate": 1.9976596872999664e-05, "loss": 0.4866, "step": 1436 }, { "epoch": 0.05119618077203983, "grad_norm": 2.0794158988889073, "learning_rate": 1.9976517906090528e-05, "loss": 0.6602, "step": 1437 }, { "epoch": 0.051231807898533946, "grad_norm": 1.5877585518085027, "learning_rate": 1.9976438806337085e-05, "loss": 0.5337, "step": 1438 }, { "epoch": 0.051267435025028056, "grad_norm": 2.011688199413847, "learning_rate": 1.9976359573740388e-05, "loss": 0.5514, "step": 1439 }, { "epoch": 0.051303062151522166, "grad_norm": 1.4943933189170682, "learning_rate": 1.9976280208301488e-05, "loss": 0.5935, "step": 1440 }, { "epoch": 0.05133868927801628, "grad_norm": 1.9647916255738718, "learning_rate": 1.997620071002144e-05, "loss": 0.6927, "step": 1441 }, { "epoch": 0.051374316404510394, "grad_norm": 1.4668462682043772, "learning_rate": 1.9976121078901313e-05, "loss": 0.592, "step": 1442 }, { "epoch": 0.051409943531004504, "grad_norm": 2.139741112454947, "learning_rate": 1.9976041314942156e-05, "loss": 0.6489, "step": 1443 }, { "epoch": 0.05144557065749862, "grad_norm": 2.245145602984398, "learning_rate": 1.997596141814504e-05, "loss": 0.6185, "step": 1444 }, { "epoch": 0.05148119778399273, "grad_norm": 2.028910653153294, "learning_rate": 1.9975881388511022e-05, "loss": 0.6951, "step": 1445 }, { "epoch": 0.05151682491048684, "grad_norm": 1.8412521972853457, "learning_rate": 1.9975801226041173e-05, "loss": 0.6418, "step": 1446 }, { "epoch": 0.05155245203698096, "grad_norm": 1.607837661096566, "learning_rate": 1.9975720930736557e-05, "loss": 0.5482, "step": 1447 }, { "epoch": 0.05158807916347507, "grad_norm": 1.719256690077516, "learning_rate": 1.9975640502598243e-05, "loss": 0.6127, "step": 1448 }, { "epoch": 0.051623706289969186, "grad_norm": 2.577202185052275, "learning_rate": 1.9975559941627303e-05, "loss": 0.6002, "step": 1449 }, { "epoch": 0.051659333416463296, "grad_norm": 2.1745641004344223, "learning_rate": 1.9975479247824813e-05, "loss": 0.6107, "step": 1450 }, { "epoch": 0.051694960542957406, "grad_norm": 2.3301831027264823, "learning_rate": 1.9975398421191842e-05, "loss": 0.6163, "step": 1451 }, { "epoch": 0.051730587669451523, "grad_norm": 2.153860926130477, "learning_rate": 1.9975317461729472e-05, "loss": 0.6059, "step": 1452 }, { "epoch": 0.051766214795945634, "grad_norm": 1.5020683174819751, "learning_rate": 1.9975236369438773e-05, "loss": 0.5852, "step": 1453 }, { "epoch": 0.051801841922439744, "grad_norm": 1.799635653417546, "learning_rate": 1.9975155144320833e-05, "loss": 0.6506, "step": 1454 }, { "epoch": 0.05183746904893386, "grad_norm": 2.355734837044571, "learning_rate": 1.9975073786376725e-05, "loss": 0.7114, "step": 1455 }, { "epoch": 0.05187309617542797, "grad_norm": 2.4935806677908947, "learning_rate": 1.9974992295607542e-05, "loss": 0.6484, "step": 1456 }, { "epoch": 0.05190872330192208, "grad_norm": 1.292784072420685, "learning_rate": 1.9974910672014366e-05, "loss": 0.4964, "step": 1457 }, { "epoch": 0.0519443504284162, "grad_norm": 3.101384014999326, "learning_rate": 1.9974828915598276e-05, "loss": 0.6406, "step": 1458 }, { "epoch": 0.05197997755491031, "grad_norm": 2.165028770572609, "learning_rate": 1.9974747026360372e-05, "loss": 0.595, "step": 1459 }, { "epoch": 0.05201560468140442, "grad_norm": 2.0078186694393745, "learning_rate": 1.997466500430174e-05, "loss": 0.6078, "step": 1460 }, { "epoch": 0.052051231807898536, "grad_norm": 2.083923510852802, "learning_rate": 1.9974582849423468e-05, "loss": 0.6954, "step": 1461 }, { "epoch": 0.052086858934392646, "grad_norm": 1.5803643483749161, "learning_rate": 1.9974500561726657e-05, "loss": 0.58, "step": 1462 }, { "epoch": 0.05212248606088676, "grad_norm": 1.468976556378538, "learning_rate": 1.99744181412124e-05, "loss": 0.5804, "step": 1463 }, { "epoch": 0.052158113187380874, "grad_norm": 1.806963378912346, "learning_rate": 1.9974335587881786e-05, "loss": 0.7756, "step": 1464 }, { "epoch": 0.052193740313874984, "grad_norm": 2.310677537215292, "learning_rate": 1.9974252901735926e-05, "loss": 0.7159, "step": 1465 }, { "epoch": 0.052229367440369094, "grad_norm": 1.3323430244772407, "learning_rate": 1.9974170082775923e-05, "loss": 0.6196, "step": 1466 }, { "epoch": 0.05226499456686321, "grad_norm": 2.202780173867658, "learning_rate": 1.9974087131002865e-05, "loss": 0.6521, "step": 1467 }, { "epoch": 0.05230062169335732, "grad_norm": 1.8252095676300621, "learning_rate": 1.9974004046417873e-05, "loss": 0.6028, "step": 1468 }, { "epoch": 0.05233624881985143, "grad_norm": 2.114452203937842, "learning_rate": 1.997392082902204e-05, "loss": 0.5965, "step": 1469 }, { "epoch": 0.05237187594634555, "grad_norm": 1.4773298416419303, "learning_rate": 1.9973837478816483e-05, "loss": 0.5499, "step": 1470 }, { "epoch": 0.05240750307283966, "grad_norm": 2.292894479274995, "learning_rate": 1.9973753995802307e-05, "loss": 0.6528, "step": 1471 }, { "epoch": 0.052443130199333776, "grad_norm": 1.377759134083893, "learning_rate": 1.9973670379980626e-05, "loss": 0.536, "step": 1472 }, { "epoch": 0.052478757325827886, "grad_norm": 1.9517162097978296, "learning_rate": 1.997358663135255e-05, "loss": 0.6762, "step": 1473 }, { "epoch": 0.052514384452322, "grad_norm": 1.4342200415877904, "learning_rate": 1.99735027499192e-05, "loss": 0.5499, "step": 1474 }, { "epoch": 0.052550011578816114, "grad_norm": 2.380535751046246, "learning_rate": 1.997341873568169e-05, "loss": 0.7066, "step": 1475 }, { "epoch": 0.052585638705310224, "grad_norm": 1.964640886106765, "learning_rate": 1.997333458864114e-05, "loss": 0.6079, "step": 1476 }, { "epoch": 0.052621265831804334, "grad_norm": 1.506785892518624, "learning_rate": 1.9973250308798665e-05, "loss": 0.5818, "step": 1477 }, { "epoch": 0.05265689295829845, "grad_norm": 1.6066245904970362, "learning_rate": 1.9973165896155395e-05, "loss": 0.6217, "step": 1478 }, { "epoch": 0.05269252008479256, "grad_norm": 2.75402876318358, "learning_rate": 1.997308135071245e-05, "loss": 0.6604, "step": 1479 }, { "epoch": 0.05272814721128667, "grad_norm": 1.917216441336962, "learning_rate": 1.997299667247095e-05, "loss": 0.6313, "step": 1480 }, { "epoch": 0.05276377433778079, "grad_norm": 1.8890683193776945, "learning_rate": 1.9972911861432033e-05, "loss": 0.6084, "step": 1481 }, { "epoch": 0.0527994014642749, "grad_norm": 1.4097214081126013, "learning_rate": 1.9972826917596827e-05, "loss": 0.5804, "step": 1482 }, { "epoch": 0.05283502859076901, "grad_norm": 2.4278929959344344, "learning_rate": 1.9972741840966455e-05, "loss": 0.6298, "step": 1483 }, { "epoch": 0.052870655717263126, "grad_norm": 1.403082619966321, "learning_rate": 1.997265663154206e-05, "loss": 0.5522, "step": 1484 }, { "epoch": 0.05290628284375724, "grad_norm": 1.4220002802797937, "learning_rate": 1.9972571289324767e-05, "loss": 0.5777, "step": 1485 }, { "epoch": 0.05294190997025135, "grad_norm": 1.4889797735646197, "learning_rate": 1.9972485814315715e-05, "loss": 0.5152, "step": 1486 }, { "epoch": 0.052977537096745464, "grad_norm": 2.6731524262276265, "learning_rate": 1.997240020651605e-05, "loss": 0.6313, "step": 1487 }, { "epoch": 0.053013164223239574, "grad_norm": 1.909734754852532, "learning_rate": 1.9972314465926902e-05, "loss": 0.6092, "step": 1488 }, { "epoch": 0.053048791349733684, "grad_norm": 1.366540974427449, "learning_rate": 1.9972228592549418e-05, "loss": 0.5233, "step": 1489 }, { "epoch": 0.0530844184762278, "grad_norm": 1.727063814742558, "learning_rate": 1.997214258638474e-05, "loss": 0.6566, "step": 1490 }, { "epoch": 0.05312004560272191, "grad_norm": 1.527865736288272, "learning_rate": 1.9972056447434012e-05, "loss": 0.5687, "step": 1491 }, { "epoch": 0.05315567272921602, "grad_norm": 2.2102456078480746, "learning_rate": 1.9971970175698387e-05, "loss": 0.5986, "step": 1492 }, { "epoch": 0.05319129985571014, "grad_norm": 2.3783457325735853, "learning_rate": 1.9971883771179003e-05, "loss": 0.6251, "step": 1493 }, { "epoch": 0.05322692698220425, "grad_norm": 2.0679486507189213, "learning_rate": 1.997179723387702e-05, "loss": 0.6926, "step": 1494 }, { "epoch": 0.05326255410869836, "grad_norm": 1.5353200923078212, "learning_rate": 1.9971710563793584e-05, "loss": 0.5441, "step": 1495 }, { "epoch": 0.05329818123519248, "grad_norm": 2.1113625925591535, "learning_rate": 1.9971623760929856e-05, "loss": 0.6601, "step": 1496 }, { "epoch": 0.05333380836168659, "grad_norm": 2.247772333441751, "learning_rate": 1.9971536825286984e-05, "loss": 0.6222, "step": 1497 }, { "epoch": 0.053369435488180704, "grad_norm": 1.4034811903523934, "learning_rate": 1.9971449756866135e-05, "loss": 0.589, "step": 1498 }, { "epoch": 0.053405062614674814, "grad_norm": 1.467537139349677, "learning_rate": 1.997136255566846e-05, "loss": 0.6017, "step": 1499 }, { "epoch": 0.053440689741168924, "grad_norm": 1.8840563900908365, "learning_rate": 1.997127522169512e-05, "loss": 0.6502, "step": 1500 }, { "epoch": 0.05347631686766304, "grad_norm": 1.8200929564867645, "learning_rate": 1.9971187754947287e-05, "loss": 0.5609, "step": 1501 }, { "epoch": 0.05351194399415715, "grad_norm": 2.203523996168465, "learning_rate": 1.9971100155426117e-05, "loss": 0.6411, "step": 1502 }, { "epoch": 0.05354757112065126, "grad_norm": 1.3575718735798716, "learning_rate": 1.9971012423132776e-05, "loss": 0.6029, "step": 1503 }, { "epoch": 0.05358319824714538, "grad_norm": 2.0573273813868753, "learning_rate": 1.9970924558068438e-05, "loss": 0.6056, "step": 1504 }, { "epoch": 0.05361882537363949, "grad_norm": 1.924487655374349, "learning_rate": 1.9970836560234266e-05, "loss": 0.5776, "step": 1505 }, { "epoch": 0.0536544525001336, "grad_norm": 2.201680164732566, "learning_rate": 1.9970748429631445e-05, "loss": 0.7056, "step": 1506 }, { "epoch": 0.05369007962662772, "grad_norm": 1.599638435437494, "learning_rate": 1.997066016626113e-05, "loss": 0.5553, "step": 1507 }, { "epoch": 0.05372570675312183, "grad_norm": 1.4616526201261277, "learning_rate": 1.9970571770124513e-05, "loss": 0.5557, "step": 1508 }, { "epoch": 0.05376133387961594, "grad_norm": 2.916344820246249, "learning_rate": 1.997048324122276e-05, "loss": 0.6801, "step": 1509 }, { "epoch": 0.053796961006110054, "grad_norm": 1.3704390075383526, "learning_rate": 1.9970394579557054e-05, "loss": 0.5391, "step": 1510 }, { "epoch": 0.053832588132604164, "grad_norm": 2.1645692619980745, "learning_rate": 1.9970305785128576e-05, "loss": 0.668, "step": 1511 }, { "epoch": 0.053868215259098275, "grad_norm": 2.706397418705658, "learning_rate": 1.9970216857938508e-05, "loss": 0.5998, "step": 1512 }, { "epoch": 0.05390384238559239, "grad_norm": 1.4285669615426597, "learning_rate": 1.9970127797988033e-05, "loss": 0.5347, "step": 1513 }, { "epoch": 0.0539394695120865, "grad_norm": 2.609723926525505, "learning_rate": 1.997003860527834e-05, "loss": 0.6304, "step": 1514 }, { "epoch": 0.05397509663858061, "grad_norm": 1.4870077825687424, "learning_rate": 1.996994927981061e-05, "loss": 0.5541, "step": 1515 }, { "epoch": 0.05401072376507473, "grad_norm": 1.4866316224677933, "learning_rate": 1.996985982158604e-05, "loss": 0.6085, "step": 1516 }, { "epoch": 0.05404635089156884, "grad_norm": 1.5978126565572368, "learning_rate": 1.996977023060582e-05, "loss": 0.5813, "step": 1517 }, { "epoch": 0.05408197801806295, "grad_norm": 1.353090973307863, "learning_rate": 1.9969680506871138e-05, "loss": 0.5915, "step": 1518 }, { "epoch": 0.05411760514455707, "grad_norm": 1.284569279790849, "learning_rate": 1.996959065038319e-05, "loss": 0.501, "step": 1519 }, { "epoch": 0.05415323227105118, "grad_norm": 2.1929502257246347, "learning_rate": 1.9969500661143174e-05, "loss": 0.6377, "step": 1520 }, { "epoch": 0.054188859397545294, "grad_norm": 1.4015233452825948, "learning_rate": 1.996941053915229e-05, "loss": 0.6278, "step": 1521 }, { "epoch": 0.054224486524039404, "grad_norm": 2.633542514022256, "learning_rate": 1.9969320284411736e-05, "loss": 0.7181, "step": 1522 }, { "epoch": 0.054260113650533515, "grad_norm": 1.5489471167258582, "learning_rate": 1.9969229896922715e-05, "loss": 0.5074, "step": 1523 }, { "epoch": 0.05429574077702763, "grad_norm": 2.2990379775077487, "learning_rate": 1.996913937668643e-05, "loss": 0.7055, "step": 1524 }, { "epoch": 0.05433136790352174, "grad_norm": 2.0370254918271957, "learning_rate": 1.996904872370409e-05, "loss": 0.6285, "step": 1525 }, { "epoch": 0.05436699503001585, "grad_norm": 1.362815856027215, "learning_rate": 1.9968957937976893e-05, "loss": 0.5784, "step": 1526 }, { "epoch": 0.05440262215650997, "grad_norm": 1.7027480492592935, "learning_rate": 1.996886701950605e-05, "loss": 0.5813, "step": 1527 }, { "epoch": 0.05443824928300408, "grad_norm": 1.3257976085588574, "learning_rate": 1.996877596829278e-05, "loss": 0.5558, "step": 1528 }, { "epoch": 0.05447387640949819, "grad_norm": 1.7619434653865154, "learning_rate": 1.9968684784338293e-05, "loss": 0.6307, "step": 1529 }, { "epoch": 0.05450950353599231, "grad_norm": 1.3990236999125207, "learning_rate": 1.9968593467643795e-05, "loss": 0.5349, "step": 1530 }, { "epoch": 0.05454513066248642, "grad_norm": 1.3484570980389954, "learning_rate": 1.996850201821051e-05, "loss": 0.5392, "step": 1531 }, { "epoch": 0.05458075778898053, "grad_norm": 2.4003802542054213, "learning_rate": 1.996841043603965e-05, "loss": 0.6206, "step": 1532 }, { "epoch": 0.054616384915474644, "grad_norm": 2.8210303012136078, "learning_rate": 1.9968318721132442e-05, "loss": 0.677, "step": 1533 }, { "epoch": 0.054652012041968755, "grad_norm": 2.084299754828923, "learning_rate": 1.9968226873490103e-05, "loss": 0.6926, "step": 1534 }, { "epoch": 0.054687639168462865, "grad_norm": 2.681231458606225, "learning_rate": 1.996813489311385e-05, "loss": 0.6758, "step": 1535 }, { "epoch": 0.05472326629495698, "grad_norm": 1.6141748330344954, "learning_rate": 1.9968042780004917e-05, "loss": 0.5944, "step": 1536 }, { "epoch": 0.05475889342145109, "grad_norm": 2.485975192310891, "learning_rate": 1.996795053416453e-05, "loss": 0.5877, "step": 1537 }, { "epoch": 0.0547945205479452, "grad_norm": 1.2517212751146098, "learning_rate": 1.996785815559391e-05, "loss": 0.5427, "step": 1538 }, { "epoch": 0.05483014767443932, "grad_norm": 2.4388383120255317, "learning_rate": 1.9967765644294293e-05, "loss": 0.6999, "step": 1539 }, { "epoch": 0.05486577480093343, "grad_norm": 2.339876908388795, "learning_rate": 1.9967673000266908e-05, "loss": 0.6265, "step": 1540 }, { "epoch": 0.05490140192742754, "grad_norm": 1.4833768198519501, "learning_rate": 1.9967580223512994e-05, "loss": 0.5661, "step": 1541 }, { "epoch": 0.05493702905392166, "grad_norm": 1.4807342612591485, "learning_rate": 1.9967487314033778e-05, "loss": 0.5917, "step": 1542 }, { "epoch": 0.05497265618041577, "grad_norm": 1.6480445314305612, "learning_rate": 1.9967394271830506e-05, "loss": 0.5455, "step": 1543 }, { "epoch": 0.055008283306909884, "grad_norm": 1.676670051317551, "learning_rate": 1.996730109690441e-05, "loss": 0.5592, "step": 1544 }, { "epoch": 0.055043910433403995, "grad_norm": 1.3664359157028936, "learning_rate": 1.996720778925673e-05, "loss": 0.5576, "step": 1545 }, { "epoch": 0.055079537559898105, "grad_norm": 1.3193802988386216, "learning_rate": 1.9967114348888716e-05, "loss": 0.5381, "step": 1546 }, { "epoch": 0.05511516468639222, "grad_norm": 2.4031707847011807, "learning_rate": 1.9967020775801605e-05, "loss": 0.6663, "step": 1547 }, { "epoch": 0.05515079181288633, "grad_norm": 2.2429022916099903, "learning_rate": 1.9966927069996652e-05, "loss": 0.6424, "step": 1548 }, { "epoch": 0.05518641893938044, "grad_norm": 2.1328711583650932, "learning_rate": 1.996683323147509e-05, "loss": 0.6766, "step": 1549 }, { "epoch": 0.05522204606587456, "grad_norm": 1.4377491999485956, "learning_rate": 1.996673926023818e-05, "loss": 0.5938, "step": 1550 }, { "epoch": 0.05525767319236867, "grad_norm": 1.6666284259542306, "learning_rate": 1.9966645156287173e-05, "loss": 0.6409, "step": 1551 }, { "epoch": 0.05529330031886278, "grad_norm": 1.4002572773360968, "learning_rate": 1.996655091962332e-05, "loss": 0.4802, "step": 1552 }, { "epoch": 0.0553289274453569, "grad_norm": 1.7517036206792822, "learning_rate": 1.9966456550247875e-05, "loss": 0.6344, "step": 1553 }, { "epoch": 0.05536455457185101, "grad_norm": 1.4422984672707593, "learning_rate": 1.9966362048162092e-05, "loss": 0.5534, "step": 1554 }, { "epoch": 0.05540018169834512, "grad_norm": 1.5595967020605987, "learning_rate": 1.9966267413367232e-05, "loss": 0.5696, "step": 1555 }, { "epoch": 0.055435808824839235, "grad_norm": 1.5757420255855628, "learning_rate": 1.9966172645864556e-05, "loss": 0.5304, "step": 1556 }, { "epoch": 0.055471435951333345, "grad_norm": 1.5326989158424205, "learning_rate": 1.9966077745655327e-05, "loss": 0.4937, "step": 1557 }, { "epoch": 0.055507063077827455, "grad_norm": 2.0528286026557345, "learning_rate": 1.996598271274081e-05, "loss": 0.6785, "step": 1558 }, { "epoch": 0.05554269020432157, "grad_norm": 2.2909850349037013, "learning_rate": 1.9965887547122263e-05, "loss": 0.5836, "step": 1559 }, { "epoch": 0.05557831733081568, "grad_norm": 1.3676136528014537, "learning_rate": 1.9965792248800955e-05, "loss": 0.4985, "step": 1560 }, { "epoch": 0.05561394445730979, "grad_norm": 1.5580498934154232, "learning_rate": 1.996569681777816e-05, "loss": 0.538, "step": 1561 }, { "epoch": 0.05564957158380391, "grad_norm": 2.171113178950075, "learning_rate": 1.996560125405515e-05, "loss": 0.6651, "step": 1562 }, { "epoch": 0.05568519871029802, "grad_norm": 1.5405922132313272, "learning_rate": 1.9965505557633188e-05, "loss": 0.5905, "step": 1563 }, { "epoch": 0.05572082583679213, "grad_norm": 1.3062511487850346, "learning_rate": 1.9965409728513558e-05, "loss": 0.591, "step": 1564 }, { "epoch": 0.05575645296328625, "grad_norm": 2.537090708530661, "learning_rate": 1.996531376669753e-05, "loss": 0.6015, "step": 1565 }, { "epoch": 0.05579208008978036, "grad_norm": 1.5127716156637396, "learning_rate": 1.9965217672186387e-05, "loss": 0.5968, "step": 1566 }, { "epoch": 0.055827707216274475, "grad_norm": 1.4839875257259278, "learning_rate": 1.99651214449814e-05, "loss": 0.5363, "step": 1567 }, { "epoch": 0.055863334342768585, "grad_norm": 1.42937418927783, "learning_rate": 1.996502508508386e-05, "loss": 0.5662, "step": 1568 }, { "epoch": 0.055898961469262695, "grad_norm": 1.3052007518895086, "learning_rate": 1.9964928592495046e-05, "loss": 0.4916, "step": 1569 }, { "epoch": 0.05593458859575681, "grad_norm": 1.4513285250403363, "learning_rate": 1.996483196721624e-05, "loss": 0.4884, "step": 1570 }, { "epoch": 0.05597021572225092, "grad_norm": 2.1881642841605102, "learning_rate": 1.9964735209248734e-05, "loss": 0.6121, "step": 1571 }, { "epoch": 0.05600584284874503, "grad_norm": 1.5356386094637302, "learning_rate": 1.996463831859381e-05, "loss": 0.5403, "step": 1572 }, { "epoch": 0.05604146997523915, "grad_norm": 1.2819843826921988, "learning_rate": 1.9964541295252765e-05, "loss": 0.5145, "step": 1573 }, { "epoch": 0.05607709710173326, "grad_norm": 2.1901023592212674, "learning_rate": 1.996444413922689e-05, "loss": 0.6176, "step": 1574 }, { "epoch": 0.05611272422822737, "grad_norm": 1.3635979040737554, "learning_rate": 1.9964346850517472e-05, "loss": 0.5617, "step": 1575 }, { "epoch": 0.05614835135472149, "grad_norm": 2.2825416603724893, "learning_rate": 1.9964249429125812e-05, "loss": 0.6327, "step": 1576 }, { "epoch": 0.0561839784812156, "grad_norm": 1.5750926002951182, "learning_rate": 1.9964151875053207e-05, "loss": 0.5404, "step": 1577 }, { "epoch": 0.05621960560770971, "grad_norm": 1.3749696958007531, "learning_rate": 1.9964054188300955e-05, "loss": 0.5418, "step": 1578 }, { "epoch": 0.056255232734203825, "grad_norm": 1.8470217912077045, "learning_rate": 1.996395636887036e-05, "loss": 0.6587, "step": 1579 }, { "epoch": 0.056290859860697935, "grad_norm": 2.8865378434248443, "learning_rate": 1.996385841676272e-05, "loss": 0.635, "step": 1580 }, { "epoch": 0.056326486987192045, "grad_norm": 1.4618753492140142, "learning_rate": 1.9963760331979338e-05, "loss": 0.545, "step": 1581 }, { "epoch": 0.05636211411368616, "grad_norm": 1.4893759052000637, "learning_rate": 1.9963662114521524e-05, "loss": 0.5647, "step": 1582 }, { "epoch": 0.05639774124018027, "grad_norm": 1.3974362639128863, "learning_rate": 1.9963563764390584e-05, "loss": 0.5495, "step": 1583 }, { "epoch": 0.05643336836667438, "grad_norm": 1.9866664434211838, "learning_rate": 1.996346528158783e-05, "loss": 0.6534, "step": 1584 }, { "epoch": 0.0564689954931685, "grad_norm": 1.3513954615473325, "learning_rate": 1.9963366666114573e-05, "loss": 0.5454, "step": 1585 }, { "epoch": 0.05650462261966261, "grad_norm": 1.4644530019846416, "learning_rate": 1.9963267917972125e-05, "loss": 0.5465, "step": 1586 }, { "epoch": 0.05654024974615672, "grad_norm": 1.305902414251368, "learning_rate": 1.99631690371618e-05, "loss": 0.5685, "step": 1587 }, { "epoch": 0.05657587687265084, "grad_norm": 1.5205842707125266, "learning_rate": 1.996307002368491e-05, "loss": 0.583, "step": 1588 }, { "epoch": 0.05661150399914495, "grad_norm": 1.3990444206699308, "learning_rate": 1.9962970877542786e-05, "loss": 0.5717, "step": 1589 }, { "epoch": 0.056647131125639065, "grad_norm": 1.8291883177352977, "learning_rate": 1.996287159873674e-05, "loss": 0.6182, "step": 1590 }, { "epoch": 0.056682758252133175, "grad_norm": 1.3791622722644055, "learning_rate": 1.9962772187268093e-05, "loss": 0.5699, "step": 1591 }, { "epoch": 0.056718385378627285, "grad_norm": 1.7568493661554294, "learning_rate": 1.9962672643138173e-05, "loss": 0.5842, "step": 1592 }, { "epoch": 0.0567540125051214, "grad_norm": 1.9185260991055175, "learning_rate": 1.99625729663483e-05, "loss": 0.6767, "step": 1593 }, { "epoch": 0.05678963963161551, "grad_norm": 1.5287480310007073, "learning_rate": 1.9962473156899803e-05, "loss": 0.5507, "step": 1594 }, { "epoch": 0.05682526675810962, "grad_norm": 1.870540298154605, "learning_rate": 1.9962373214794016e-05, "loss": 0.6481, "step": 1595 }, { "epoch": 0.05686089388460374, "grad_norm": 2.0947586331809123, "learning_rate": 1.996227314003227e-05, "loss": 0.6961, "step": 1596 }, { "epoch": 0.05689652101109785, "grad_norm": 1.286817559616957, "learning_rate": 1.996217293261589e-05, "loss": 0.4916, "step": 1597 }, { "epoch": 0.05693214813759196, "grad_norm": 2.175120944037233, "learning_rate": 1.9962072592546215e-05, "loss": 0.691, "step": 1598 }, { "epoch": 0.05696777526408608, "grad_norm": 1.6825649387781532, "learning_rate": 1.9961972119824577e-05, "loss": 0.5404, "step": 1599 }, { "epoch": 0.05700340239058019, "grad_norm": 2.0328054111009926, "learning_rate": 1.996187151445232e-05, "loss": 0.5912, "step": 1600 }, { "epoch": 0.0570390295170743, "grad_norm": 1.5971745482464172, "learning_rate": 1.9961770776430783e-05, "loss": 0.5833, "step": 1601 }, { "epoch": 0.057074656643568415, "grad_norm": 1.540744905915454, "learning_rate": 1.9961669905761303e-05, "loss": 0.5507, "step": 1602 }, { "epoch": 0.057110283770062525, "grad_norm": 1.9488819614470654, "learning_rate": 1.9961568902445224e-05, "loss": 0.6882, "step": 1603 }, { "epoch": 0.057145910896556636, "grad_norm": 3.26570641475607, "learning_rate": 1.9961467766483893e-05, "loss": 0.5871, "step": 1604 }, { "epoch": 0.05718153802305075, "grad_norm": 2.1698616527848116, "learning_rate": 1.9961366497878657e-05, "loss": 0.6792, "step": 1605 }, { "epoch": 0.05721716514954486, "grad_norm": 2.2909821970552655, "learning_rate": 1.9961265096630863e-05, "loss": 0.6593, "step": 1606 }, { "epoch": 0.05725279227603897, "grad_norm": 2.4113170795976395, "learning_rate": 1.996116356274186e-05, "loss": 0.6173, "step": 1607 }, { "epoch": 0.05728841940253309, "grad_norm": 1.4619070731183057, "learning_rate": 1.9961061896213006e-05, "loss": 0.5943, "step": 1608 }, { "epoch": 0.0573240465290272, "grad_norm": 1.384911085275159, "learning_rate": 1.996096009704565e-05, "loss": 0.4991, "step": 1609 }, { "epoch": 0.05735967365552131, "grad_norm": 1.4991909801442613, "learning_rate": 1.9960858165241145e-05, "loss": 0.5756, "step": 1610 }, { "epoch": 0.05739530078201543, "grad_norm": 2.101505449411462, "learning_rate": 1.996075610080085e-05, "loss": 0.6714, "step": 1611 }, { "epoch": 0.05743092790850954, "grad_norm": 1.4606926822421897, "learning_rate": 1.996065390372613e-05, "loss": 0.5796, "step": 1612 }, { "epoch": 0.057466555035003655, "grad_norm": 1.4213209064095715, "learning_rate": 1.996055157401834e-05, "loss": 0.5301, "step": 1613 }, { "epoch": 0.057502182161497765, "grad_norm": 1.8843583540438535, "learning_rate": 1.9960449111678845e-05, "loss": 0.6874, "step": 1614 }, { "epoch": 0.057537809287991876, "grad_norm": 1.3547695458979716, "learning_rate": 1.9960346516709005e-05, "loss": 0.5464, "step": 1615 }, { "epoch": 0.05757343641448599, "grad_norm": 3.0525514505275866, "learning_rate": 1.9960243789110188e-05, "loss": 0.5684, "step": 1616 }, { "epoch": 0.0576090635409801, "grad_norm": 2.4203945715193242, "learning_rate": 1.9960140928883765e-05, "loss": 0.5036, "step": 1617 }, { "epoch": 0.05764469066747421, "grad_norm": 1.607233510858677, "learning_rate": 1.9960037936031105e-05, "loss": 0.5865, "step": 1618 }, { "epoch": 0.05768031779396833, "grad_norm": 1.4505327967261388, "learning_rate": 1.9959934810553577e-05, "loss": 0.5339, "step": 1619 }, { "epoch": 0.05771594492046244, "grad_norm": 1.5441663096837008, "learning_rate": 1.9959831552452557e-05, "loss": 0.5996, "step": 1620 }, { "epoch": 0.05775157204695655, "grad_norm": 1.4352592204341703, "learning_rate": 1.9959728161729416e-05, "loss": 0.603, "step": 1621 }, { "epoch": 0.05778719917345067, "grad_norm": 2.057585094634906, "learning_rate": 1.9959624638385534e-05, "loss": 0.6463, "step": 1622 }, { "epoch": 0.05782282629994478, "grad_norm": 1.4437811385068746, "learning_rate": 1.9959520982422293e-05, "loss": 0.5071, "step": 1623 }, { "epoch": 0.05785845342643889, "grad_norm": 2.0148745807641735, "learning_rate": 1.995941719384106e-05, "loss": 0.6121, "step": 1624 }, { "epoch": 0.057894080552933005, "grad_norm": 1.7810717391579947, "learning_rate": 1.9959313272643232e-05, "loss": 0.562, "step": 1625 }, { "epoch": 0.057929707679427116, "grad_norm": 2.2596060069404937, "learning_rate": 1.9959209218830184e-05, "loss": 0.6784, "step": 1626 }, { "epoch": 0.057965334805921226, "grad_norm": 2.2832203949520373, "learning_rate": 1.9959105032403307e-05, "loss": 0.628, "step": 1627 }, { "epoch": 0.05800096193241534, "grad_norm": 1.9624874364825697, "learning_rate": 1.9959000713363985e-05, "loss": 0.6095, "step": 1628 }, { "epoch": 0.05803658905890945, "grad_norm": 2.5112838962423236, "learning_rate": 1.9958896261713606e-05, "loss": 0.645, "step": 1629 }, { "epoch": 0.05807221618540356, "grad_norm": 1.4173615889096067, "learning_rate": 1.9958791677453565e-05, "loss": 0.6158, "step": 1630 }, { "epoch": 0.05810784331189768, "grad_norm": 1.3350967455952025, "learning_rate": 1.9958686960585247e-05, "loss": 0.5603, "step": 1631 }, { "epoch": 0.05814347043839179, "grad_norm": 1.5597067523766968, "learning_rate": 1.9958582111110055e-05, "loss": 0.6015, "step": 1632 }, { "epoch": 0.0581790975648859, "grad_norm": 2.188334436434864, "learning_rate": 1.9958477129029382e-05, "loss": 0.7008, "step": 1633 }, { "epoch": 0.05821472469138002, "grad_norm": 5.960779642570997, "learning_rate": 1.9958372014344624e-05, "loss": 0.6236, "step": 1634 }, { "epoch": 0.05825035181787413, "grad_norm": 1.4641413247922865, "learning_rate": 1.9958266767057183e-05, "loss": 0.5001, "step": 1635 }, { "epoch": 0.05828597894436824, "grad_norm": 1.840127263104607, "learning_rate": 1.995816138716846e-05, "loss": 0.6577, "step": 1636 }, { "epoch": 0.058321606070862356, "grad_norm": 2.0551339256901775, "learning_rate": 1.9958055874679853e-05, "loss": 0.6044, "step": 1637 }, { "epoch": 0.058357233197356466, "grad_norm": 2.1416744662732943, "learning_rate": 1.9957950229592777e-05, "loss": 0.6393, "step": 1638 }, { "epoch": 0.05839286032385058, "grad_norm": 2.099664906971355, "learning_rate": 1.9957844451908633e-05, "loss": 0.6882, "step": 1639 }, { "epoch": 0.05842848745034469, "grad_norm": 1.442432095804462, "learning_rate": 1.995773854162883e-05, "loss": 0.5494, "step": 1640 }, { "epoch": 0.058464114576838803, "grad_norm": 1.5858134842483707, "learning_rate": 1.9957632498754776e-05, "loss": 0.5347, "step": 1641 }, { "epoch": 0.05849974170333292, "grad_norm": 2.30057009623096, "learning_rate": 1.9957526323287886e-05, "loss": 0.5378, "step": 1642 }, { "epoch": 0.05853536882982703, "grad_norm": 1.972243269020946, "learning_rate": 1.9957420015229572e-05, "loss": 0.5973, "step": 1643 }, { "epoch": 0.05857099595632114, "grad_norm": 2.026658026899643, "learning_rate": 1.9957313574581258e-05, "loss": 0.6764, "step": 1644 }, { "epoch": 0.05860662308281526, "grad_norm": 1.4452296650907115, "learning_rate": 1.9957207001344345e-05, "loss": 0.5681, "step": 1645 }, { "epoch": 0.05864225020930937, "grad_norm": 1.415470296080671, "learning_rate": 1.9957100295520267e-05, "loss": 0.5574, "step": 1646 }, { "epoch": 0.05867787733580348, "grad_norm": 1.3368825667479955, "learning_rate": 1.9956993457110437e-05, "loss": 0.5431, "step": 1647 }, { "epoch": 0.058713504462297596, "grad_norm": 1.9789801056139005, "learning_rate": 1.9956886486116278e-05, "loss": 0.7218, "step": 1648 }, { "epoch": 0.058749131588791706, "grad_norm": 1.3561579529678496, "learning_rate": 1.9956779382539216e-05, "loss": 0.5466, "step": 1649 }, { "epoch": 0.058784758715285816, "grad_norm": 1.4310419622757509, "learning_rate": 1.995667214638068e-05, "loss": 0.567, "step": 1650 }, { "epoch": 0.05882038584177993, "grad_norm": 3.161512076519061, "learning_rate": 1.9956564777642093e-05, "loss": 0.6857, "step": 1651 }, { "epoch": 0.058856012968274043, "grad_norm": 1.6276930297759813, "learning_rate": 1.995645727632489e-05, "loss": 0.5544, "step": 1652 }, { "epoch": 0.058891640094768154, "grad_norm": 1.5354386894318146, "learning_rate": 1.9956349642430494e-05, "loss": 0.5518, "step": 1653 }, { "epoch": 0.05892726722126227, "grad_norm": 1.7271924834068118, "learning_rate": 1.9956241875960348e-05, "loss": 0.6608, "step": 1654 }, { "epoch": 0.05896289434775638, "grad_norm": 2.3386252918534316, "learning_rate": 1.995613397691588e-05, "loss": 0.5722, "step": 1655 }, { "epoch": 0.05899852147425049, "grad_norm": 2.1157121325707338, "learning_rate": 1.995602594529853e-05, "loss": 0.6186, "step": 1656 }, { "epoch": 0.05903414860074461, "grad_norm": 1.7775552716067442, "learning_rate": 1.995591778110973e-05, "loss": 0.6955, "step": 1657 }, { "epoch": 0.05906977572723872, "grad_norm": 1.3962079380942471, "learning_rate": 1.9955809484350933e-05, "loss": 0.5924, "step": 1658 }, { "epoch": 0.05910540285373283, "grad_norm": 1.6123030800069096, "learning_rate": 1.9955701055023574e-05, "loss": 0.5779, "step": 1659 }, { "epoch": 0.059141029980226946, "grad_norm": 2.3416195003568308, "learning_rate": 1.995559249312909e-05, "loss": 0.5985, "step": 1660 }, { "epoch": 0.059176657106721056, "grad_norm": 1.7226654146587315, "learning_rate": 1.995548379866894e-05, "loss": 0.5398, "step": 1661 }, { "epoch": 0.05921228423321517, "grad_norm": 2.117974051292949, "learning_rate": 1.9955374971644566e-05, "loss": 0.6635, "step": 1662 }, { "epoch": 0.059247911359709284, "grad_norm": 1.541969237951431, "learning_rate": 1.995526601205741e-05, "loss": 0.5778, "step": 1663 }, { "epoch": 0.059283538486203394, "grad_norm": 2.476465982611358, "learning_rate": 1.9955156919908932e-05, "loss": 0.7427, "step": 1664 }, { "epoch": 0.05931916561269751, "grad_norm": 1.3843084725660744, "learning_rate": 1.9955047695200584e-05, "loss": 0.5038, "step": 1665 }, { "epoch": 0.05935479273919162, "grad_norm": 1.4175544646202058, "learning_rate": 1.9954938337933815e-05, "loss": 0.5705, "step": 1666 }, { "epoch": 0.05939041986568573, "grad_norm": 1.4080506953336762, "learning_rate": 1.9954828848110082e-05, "loss": 0.635, "step": 1667 }, { "epoch": 0.05942604699217985, "grad_norm": 2.301219489494383, "learning_rate": 1.9954719225730847e-05, "loss": 0.6201, "step": 1668 }, { "epoch": 0.05946167411867396, "grad_norm": 1.6383189320653677, "learning_rate": 1.9954609470797566e-05, "loss": 0.5613, "step": 1669 }, { "epoch": 0.05949730124516807, "grad_norm": 2.325567040864267, "learning_rate": 1.9954499583311703e-05, "loss": 0.655, "step": 1670 }, { "epoch": 0.059532928371662186, "grad_norm": 1.3788854411359788, "learning_rate": 1.9954389563274724e-05, "loss": 0.5497, "step": 1671 }, { "epoch": 0.059568555498156296, "grad_norm": 1.4080713124435034, "learning_rate": 1.9954279410688087e-05, "loss": 0.5898, "step": 1672 }, { "epoch": 0.059604182624650406, "grad_norm": 1.6549221620220587, "learning_rate": 1.9954169125553262e-05, "loss": 0.6678, "step": 1673 }, { "epoch": 0.059639809751144524, "grad_norm": 1.3800015882386787, "learning_rate": 1.9954058707871717e-05, "loss": 0.5164, "step": 1674 }, { "epoch": 0.059675436877638634, "grad_norm": 1.72109603527121, "learning_rate": 1.9953948157644924e-05, "loss": 0.5359, "step": 1675 }, { "epoch": 0.059711064004132744, "grad_norm": 2.832488882324694, "learning_rate": 1.995383747487435e-05, "loss": 0.6105, "step": 1676 }, { "epoch": 0.05974669113062686, "grad_norm": 1.343081489629975, "learning_rate": 1.995372665956148e-05, "loss": 0.5407, "step": 1677 }, { "epoch": 0.05978231825712097, "grad_norm": 1.58815316853109, "learning_rate": 1.9953615711707775e-05, "loss": 0.5131, "step": 1678 }, { "epoch": 0.05981794538361508, "grad_norm": 2.5963221323308834, "learning_rate": 1.9953504631314722e-05, "loss": 0.6165, "step": 1679 }, { "epoch": 0.0598535725101092, "grad_norm": 1.3432772063332, "learning_rate": 1.9953393418383798e-05, "loss": 0.5643, "step": 1680 }, { "epoch": 0.05988919963660331, "grad_norm": 1.8224468127601583, "learning_rate": 1.9953282072916487e-05, "loss": 0.6603, "step": 1681 }, { "epoch": 0.05992482676309742, "grad_norm": 1.8045834745040499, "learning_rate": 1.9953170594914267e-05, "loss": 0.5983, "step": 1682 }, { "epoch": 0.059960453889591536, "grad_norm": 1.4599973309186498, "learning_rate": 1.995305898437862e-05, "loss": 0.5268, "step": 1683 }, { "epoch": 0.059996081016085646, "grad_norm": 1.4560778304395325, "learning_rate": 1.995294724131104e-05, "loss": 0.5488, "step": 1684 }, { "epoch": 0.060031708142579764, "grad_norm": 1.309515676884185, "learning_rate": 1.9952835365713006e-05, "loss": 0.5392, "step": 1685 }, { "epoch": 0.060067335269073874, "grad_norm": 1.936183468245959, "learning_rate": 1.9952723357586013e-05, "loss": 0.6951, "step": 1686 }, { "epoch": 0.060102962395567984, "grad_norm": 2.6393092381567196, "learning_rate": 1.9952611216931558e-05, "loss": 0.7244, "step": 1687 }, { "epoch": 0.0601385895220621, "grad_norm": 1.5803605458782553, "learning_rate": 1.9952498943751123e-05, "loss": 0.4695, "step": 1688 }, { "epoch": 0.06017421664855621, "grad_norm": 1.4520753067061005, "learning_rate": 1.9952386538046208e-05, "loss": 0.5692, "step": 1689 }, { "epoch": 0.06020984377505032, "grad_norm": 1.382188622890282, "learning_rate": 1.9952273999818312e-05, "loss": 0.5832, "step": 1690 }, { "epoch": 0.06024547090154444, "grad_norm": 2.047003525754574, "learning_rate": 1.9952161329068934e-05, "loss": 0.6578, "step": 1691 }, { "epoch": 0.06028109802803855, "grad_norm": 1.629716378347668, "learning_rate": 1.9952048525799567e-05, "loss": 0.5817, "step": 1692 }, { "epoch": 0.06031672515453266, "grad_norm": 1.2848284716051486, "learning_rate": 1.995193559001172e-05, "loss": 0.5629, "step": 1693 }, { "epoch": 0.060352352281026776, "grad_norm": 1.4644795257360812, "learning_rate": 1.9951822521706896e-05, "loss": 0.5443, "step": 1694 }, { "epoch": 0.060387979407520886, "grad_norm": 1.419728822666075, "learning_rate": 1.9951709320886596e-05, "loss": 0.5442, "step": 1695 }, { "epoch": 0.060423606534015, "grad_norm": 2.1122786393989825, "learning_rate": 1.9951595987552332e-05, "loss": 0.5643, "step": 1696 }, { "epoch": 0.060459233660509114, "grad_norm": 1.4354099543900831, "learning_rate": 1.9951482521705608e-05, "loss": 0.535, "step": 1697 }, { "epoch": 0.060494860787003224, "grad_norm": 2.5091110859950687, "learning_rate": 1.9951368923347945e-05, "loss": 0.6327, "step": 1698 }, { "epoch": 0.060530487913497334, "grad_norm": 1.7543071656769005, "learning_rate": 1.9951255192480844e-05, "loss": 0.5967, "step": 1699 }, { "epoch": 0.06056611503999145, "grad_norm": 1.478517867819926, "learning_rate": 1.995114132910583e-05, "loss": 0.565, "step": 1700 }, { "epoch": 0.06060174216648556, "grad_norm": 1.3307393712717097, "learning_rate": 1.995102733322441e-05, "loss": 0.5381, "step": 1701 }, { "epoch": 0.06063736929297967, "grad_norm": 2.4588090018365034, "learning_rate": 1.9950913204838108e-05, "loss": 0.6042, "step": 1702 }, { "epoch": 0.06067299641947379, "grad_norm": 2.1519579682261765, "learning_rate": 1.9950798943948438e-05, "loss": 0.6523, "step": 1703 }, { "epoch": 0.0607086235459679, "grad_norm": 1.4175344043832572, "learning_rate": 1.9950684550556926e-05, "loss": 0.5632, "step": 1704 }, { "epoch": 0.06074425067246201, "grad_norm": 2.539910676685275, "learning_rate": 1.9950570024665094e-05, "loss": 0.6172, "step": 1705 }, { "epoch": 0.060779877798956126, "grad_norm": 2.0063353280574256, "learning_rate": 1.995045536627447e-05, "loss": 0.6633, "step": 1706 }, { "epoch": 0.06081550492545024, "grad_norm": 1.5090617101974355, "learning_rate": 1.9950340575386575e-05, "loss": 0.5615, "step": 1707 }, { "epoch": 0.060851132051944354, "grad_norm": 2.2458758195315833, "learning_rate": 1.9950225652002943e-05, "loss": 0.5613, "step": 1708 }, { "epoch": 0.060886759178438464, "grad_norm": 2.0648024629439137, "learning_rate": 1.99501105961251e-05, "loss": 0.7526, "step": 1709 }, { "epoch": 0.060922386304932574, "grad_norm": 4.135304532200817, "learning_rate": 1.994999540775458e-05, "loss": 0.7237, "step": 1710 }, { "epoch": 0.06095801343142669, "grad_norm": 2.0248261258696636, "learning_rate": 1.9949880086892914e-05, "loss": 0.5524, "step": 1711 }, { "epoch": 0.0609936405579208, "grad_norm": 2.137872244331626, "learning_rate": 1.9949764633541645e-05, "loss": 0.6934, "step": 1712 }, { "epoch": 0.06102926768441491, "grad_norm": 2.7677515434521327, "learning_rate": 1.99496490477023e-05, "loss": 0.6242, "step": 1713 }, { "epoch": 0.06106489481090903, "grad_norm": 2.3425348172292764, "learning_rate": 1.994953332937643e-05, "loss": 0.6213, "step": 1714 }, { "epoch": 0.06110052193740314, "grad_norm": 2.1429666560237033, "learning_rate": 1.9949417478565564e-05, "loss": 0.598, "step": 1715 }, { "epoch": 0.06113614906389725, "grad_norm": 1.8822695997478474, "learning_rate": 1.994930149527125e-05, "loss": 0.6423, "step": 1716 }, { "epoch": 0.061171776190391366, "grad_norm": 2.561739371707006, "learning_rate": 1.9949185379495034e-05, "loss": 0.6637, "step": 1717 }, { "epoch": 0.06120740331688548, "grad_norm": 2.365923505709006, "learning_rate": 1.994906913123846e-05, "loss": 0.7029, "step": 1718 }, { "epoch": 0.06124303044337959, "grad_norm": 3.058526664682775, "learning_rate": 1.9948952750503077e-05, "loss": 0.6454, "step": 1719 }, { "epoch": 0.061278657569873704, "grad_norm": 2.2793914610239914, "learning_rate": 1.9948836237290435e-05, "loss": 0.5554, "step": 1720 }, { "epoch": 0.061314284696367814, "grad_norm": 1.4920541782781043, "learning_rate": 1.9948719591602084e-05, "loss": 0.5566, "step": 1721 }, { "epoch": 0.061349911822861924, "grad_norm": 2.6954606223329494, "learning_rate": 1.9948602813439576e-05, "loss": 0.6674, "step": 1722 }, { "epoch": 0.06138553894935604, "grad_norm": 4.059230189282602, "learning_rate": 1.9948485902804472e-05, "loss": 0.6385, "step": 1723 }, { "epoch": 0.06142116607585015, "grad_norm": 1.4728134334412881, "learning_rate": 1.9948368859698322e-05, "loss": 0.5985, "step": 1724 }, { "epoch": 0.06145679320234426, "grad_norm": 1.5032054061580804, "learning_rate": 1.9948251684122687e-05, "loss": 0.5591, "step": 1725 }, { "epoch": 0.06149242032883838, "grad_norm": 6.893585354197282, "learning_rate": 1.9948134376079128e-05, "loss": 0.6871, "step": 1726 }, { "epoch": 0.06152804745533249, "grad_norm": 2.368133428908768, "learning_rate": 1.9948016935569208e-05, "loss": 0.6444, "step": 1727 }, { "epoch": 0.0615636745818266, "grad_norm": 1.495942922143509, "learning_rate": 1.9947899362594486e-05, "loss": 0.5641, "step": 1728 }, { "epoch": 0.06159930170832072, "grad_norm": 6.569763039859467, "learning_rate": 1.9947781657156533e-05, "loss": 0.6055, "step": 1729 }, { "epoch": 0.06163492883481483, "grad_norm": 37.97818715663566, "learning_rate": 1.9947663819256915e-05, "loss": 0.7436, "step": 1730 }, { "epoch": 0.061670555961308944, "grad_norm": 76.62582106912963, "learning_rate": 1.9947545848897197e-05, "loss": 2.9193, "step": 1731 }, { "epoch": 0.061706183087803054, "grad_norm": 55.492360380968535, "learning_rate": 1.9947427746078955e-05, "loss": 1.0002, "step": 1732 }, { "epoch": 0.061741810214297164, "grad_norm": 57.24134945372312, "learning_rate": 1.994730951080376e-05, "loss": 1.1248, "step": 1733 }, { "epoch": 0.06177743734079128, "grad_norm": 6.830250877401925, "learning_rate": 1.9947191143073185e-05, "loss": 0.711, "step": 1734 }, { "epoch": 0.06181306446728539, "grad_norm": 14.226796710417233, "learning_rate": 1.994707264288881e-05, "loss": 0.9022, "step": 1735 }, { "epoch": 0.0618486915937795, "grad_norm": 22.999585668636254, "learning_rate": 1.9946954010252207e-05, "loss": 1.0178, "step": 1736 }, { "epoch": 0.06188431872027362, "grad_norm": 14.378615974820697, "learning_rate": 1.9946835245164958e-05, "loss": 0.7742, "step": 1737 }, { "epoch": 0.06191994584676773, "grad_norm": 12.499904042555473, "learning_rate": 1.994671634762865e-05, "loss": 0.9161, "step": 1738 }, { "epoch": 0.06195557297326184, "grad_norm": 10.08784364151024, "learning_rate": 1.994659731764486e-05, "loss": 0.7344, "step": 1739 }, { "epoch": 0.06199120009975596, "grad_norm": 10.320508656157566, "learning_rate": 1.9946478155215172e-05, "loss": 0.7221, "step": 1740 }, { "epoch": 0.06202682722625007, "grad_norm": 4.197032752619728, "learning_rate": 1.9946358860341174e-05, "loss": 0.626, "step": 1741 }, { "epoch": 0.06206245435274418, "grad_norm": 5.007431960821024, "learning_rate": 1.994623943302446e-05, "loss": 0.7064, "step": 1742 }, { "epoch": 0.062098081479238294, "grad_norm": 29.73020576943956, "learning_rate": 1.9946119873266615e-05, "loss": 0.8017, "step": 1743 }, { "epoch": 0.062133708605732405, "grad_norm": 16.35765203349077, "learning_rate": 1.9946000181069232e-05, "loss": 0.6489, "step": 1744 }, { "epoch": 0.062169335732226515, "grad_norm": 28.157241701532648, "learning_rate": 1.9945880356433904e-05, "loss": 0.8654, "step": 1745 }, { "epoch": 0.06220496285872063, "grad_norm": 51.09773641113586, "learning_rate": 1.9945760399362227e-05, "loss": 2.0126, "step": 1746 }, { "epoch": 0.06224058998521474, "grad_norm": 25.554541588317257, "learning_rate": 1.99456403098558e-05, "loss": 0.8381, "step": 1747 }, { "epoch": 0.06227621711170885, "grad_norm": 5.086762903060995, "learning_rate": 1.9945520087916218e-05, "loss": 0.8434, "step": 1748 }, { "epoch": 0.06231184423820297, "grad_norm": 3.198716959665418, "learning_rate": 1.9945399733545085e-05, "loss": 0.6794, "step": 1749 }, { "epoch": 0.06234747136469708, "grad_norm": 2.572431985292631, "learning_rate": 1.9945279246744004e-05, "loss": 0.7592, "step": 1750 }, { "epoch": 0.06238309849119119, "grad_norm": 3.509791780645515, "learning_rate": 1.994515862751458e-05, "loss": 0.6846, "step": 1751 }, { "epoch": 0.06241872561768531, "grad_norm": 2.167181893341427, "learning_rate": 1.9945037875858414e-05, "loss": 0.7198, "step": 1752 }, { "epoch": 0.06245435274417942, "grad_norm": 2.05302218347827, "learning_rate": 1.9944916991777122e-05, "loss": 0.5673, "step": 1753 }, { "epoch": 0.06248997987067353, "grad_norm": 2.483643887533083, "learning_rate": 1.9944795975272308e-05, "loss": 0.7065, "step": 1754 }, { "epoch": 0.06252560699716764, "grad_norm": 1.6826265924704975, "learning_rate": 1.994467482634558e-05, "loss": 0.6533, "step": 1755 }, { "epoch": 0.06256123412366175, "grad_norm": 1.5721593212849883, "learning_rate": 1.9944553544998563e-05, "loss": 0.6097, "step": 1756 }, { "epoch": 0.06259686125015587, "grad_norm": 2.591553868144003, "learning_rate": 1.9944432131232857e-05, "loss": 0.6652, "step": 1757 }, { "epoch": 0.06263248837664998, "grad_norm": 1.61356539412425, "learning_rate": 1.9944310585050094e-05, "loss": 0.6202, "step": 1758 }, { "epoch": 0.06266811550314409, "grad_norm": 1.7075504479129866, "learning_rate": 1.994418890645188e-05, "loss": 0.6048, "step": 1759 }, { "epoch": 0.06270374262963821, "grad_norm": 1.4196305609520135, "learning_rate": 1.9944067095439845e-05, "loss": 0.5554, "step": 1760 }, { "epoch": 0.06273936975613231, "grad_norm": 1.5934528789147595, "learning_rate": 1.99439451520156e-05, "loss": 0.5537, "step": 1761 }, { "epoch": 0.06277499688262643, "grad_norm": 4.015185356756166, "learning_rate": 1.9943823076180778e-05, "loss": 0.579, "step": 1762 }, { "epoch": 0.06281062400912055, "grad_norm": 1.4681257517175905, "learning_rate": 1.9943700867937e-05, "loss": 0.5506, "step": 1763 }, { "epoch": 0.06284625113561465, "grad_norm": 1.5947407774009899, "learning_rate": 1.9943578527285897e-05, "loss": 0.621, "step": 1764 }, { "epoch": 0.06288187826210877, "grad_norm": 1.3756954025641344, "learning_rate": 1.994345605422909e-05, "loss": 0.5407, "step": 1765 }, { "epoch": 0.06291750538860288, "grad_norm": 1.439954844777577, "learning_rate": 1.9943333448768226e-05, "loss": 0.5996, "step": 1766 }, { "epoch": 0.062953132515097, "grad_norm": 2.5654395168028334, "learning_rate": 1.994321071090492e-05, "loss": 0.7592, "step": 1767 }, { "epoch": 0.0629887596415911, "grad_norm": 2.172991952316255, "learning_rate": 1.9943087840640813e-05, "loss": 0.6534, "step": 1768 }, { "epoch": 0.06302438676808522, "grad_norm": 1.3753675538717005, "learning_rate": 1.9942964837977542e-05, "loss": 0.5434, "step": 1769 }, { "epoch": 0.06306001389457934, "grad_norm": 2.6039318020427946, "learning_rate": 1.9942841702916746e-05, "loss": 0.6814, "step": 1770 }, { "epoch": 0.06309564102107344, "grad_norm": 2.227242186421801, "learning_rate": 1.9942718435460066e-05, "loss": 0.7582, "step": 1771 }, { "epoch": 0.06313126814756756, "grad_norm": 1.363147668396615, "learning_rate": 1.9942595035609137e-05, "loss": 0.5922, "step": 1772 }, { "epoch": 0.06316689527406168, "grad_norm": 1.4623008639523292, "learning_rate": 1.9942471503365604e-05, "loss": 0.6244, "step": 1773 }, { "epoch": 0.06320252240055578, "grad_norm": 1.4548348885130091, "learning_rate": 1.9942347838731115e-05, "loss": 0.5772, "step": 1774 }, { "epoch": 0.0632381495270499, "grad_norm": 1.4292635760279409, "learning_rate": 1.9942224041707315e-05, "loss": 0.6036, "step": 1775 }, { "epoch": 0.06327377665354401, "grad_norm": 1.829083334933875, "learning_rate": 1.9942100112295857e-05, "loss": 0.6247, "step": 1776 }, { "epoch": 0.06330940378003812, "grad_norm": 2.9321475978955744, "learning_rate": 1.994197605049838e-05, "loss": 0.6606, "step": 1777 }, { "epoch": 0.06334503090653223, "grad_norm": 1.389943425700481, "learning_rate": 1.994185185631655e-05, "loss": 0.5526, "step": 1778 }, { "epoch": 0.06338065803302635, "grad_norm": 1.6585630282694137, "learning_rate": 1.994172752975201e-05, "loss": 0.6132, "step": 1779 }, { "epoch": 0.06341628515952046, "grad_norm": 1.3148114523275534, "learning_rate": 1.9941603070806415e-05, "loss": 0.5742, "step": 1780 }, { "epoch": 0.06345191228601457, "grad_norm": 2.1131661005328812, "learning_rate": 1.9941478479481433e-05, "loss": 0.6515, "step": 1781 }, { "epoch": 0.06348753941250869, "grad_norm": 2.167640392941399, "learning_rate": 1.9941353755778713e-05, "loss": 0.7104, "step": 1782 }, { "epoch": 0.06352316653900279, "grad_norm": 2.4636238481085306, "learning_rate": 1.994122889969992e-05, "loss": 0.7132, "step": 1783 }, { "epoch": 0.06355879366549691, "grad_norm": 1.3995260782595653, "learning_rate": 1.9941103911246714e-05, "loss": 0.6052, "step": 1784 }, { "epoch": 0.06359442079199103, "grad_norm": 1.4851192232979675, "learning_rate": 1.9940978790420762e-05, "loss": 0.5235, "step": 1785 }, { "epoch": 0.06363004791848513, "grad_norm": 2.217212173817568, "learning_rate": 1.994085353722373e-05, "loss": 0.6844, "step": 1786 }, { "epoch": 0.06366567504497925, "grad_norm": 1.2701439537021417, "learning_rate": 1.9940728151657285e-05, "loss": 0.6183, "step": 1787 }, { "epoch": 0.06370130217147336, "grad_norm": 1.510487356226408, "learning_rate": 1.9940602633723097e-05, "loss": 0.5101, "step": 1788 }, { "epoch": 0.06373692929796747, "grad_norm": 1.4212475449581161, "learning_rate": 1.9940476983422834e-05, "loss": 0.5567, "step": 1789 }, { "epoch": 0.06377255642446159, "grad_norm": 2.4841775203813596, "learning_rate": 1.9940351200758175e-05, "loss": 0.6632, "step": 1790 }, { "epoch": 0.0638081835509557, "grad_norm": 1.1710942107708944, "learning_rate": 1.994022528573079e-05, "loss": 0.5445, "step": 1791 }, { "epoch": 0.0638438106774498, "grad_norm": 1.3473977401042374, "learning_rate": 1.994009923834236e-05, "loss": 0.5579, "step": 1792 }, { "epoch": 0.06387943780394392, "grad_norm": 1.4264064484475187, "learning_rate": 1.993997305859456e-05, "loss": 0.543, "step": 1793 }, { "epoch": 0.06391506493043804, "grad_norm": 2.2267454218397567, "learning_rate": 1.993984674648907e-05, "loss": 0.6541, "step": 1794 }, { "epoch": 0.06395069205693214, "grad_norm": 1.3332993714972348, "learning_rate": 1.993972030202757e-05, "loss": 0.5161, "step": 1795 }, { "epoch": 0.06398631918342626, "grad_norm": 1.4083517263745475, "learning_rate": 1.9939593725211753e-05, "loss": 0.6583, "step": 1796 }, { "epoch": 0.06402194630992038, "grad_norm": 1.9872388691035636, "learning_rate": 1.9939467016043293e-05, "loss": 0.6215, "step": 1797 }, { "epoch": 0.06405757343641448, "grad_norm": 1.8447702545679259, "learning_rate": 1.9939340174523885e-05, "loss": 0.6758, "step": 1798 }, { "epoch": 0.0640932005629086, "grad_norm": 1.931204435356586, "learning_rate": 1.9939213200655216e-05, "loss": 0.6547, "step": 1799 }, { "epoch": 0.06412882768940271, "grad_norm": 1.516959531463243, "learning_rate": 1.9939086094438975e-05, "loss": 0.6106, "step": 1800 }, { "epoch": 0.06416445481589682, "grad_norm": 1.382301465343679, "learning_rate": 1.9938958855876854e-05, "loss": 0.5818, "step": 1801 }, { "epoch": 0.06420008194239094, "grad_norm": 2.636466258694311, "learning_rate": 1.993883148497055e-05, "loss": 0.6511, "step": 1802 }, { "epoch": 0.06423570906888505, "grad_norm": 2.342399546466757, "learning_rate": 1.993870398172176e-05, "loss": 0.6514, "step": 1803 }, { "epoch": 0.06427133619537916, "grad_norm": 3.5787073231877806, "learning_rate": 1.993857634613218e-05, "loss": 0.5555, "step": 1804 }, { "epoch": 0.06430696332187327, "grad_norm": 1.988765245739922, "learning_rate": 1.9938448578203508e-05, "loss": 0.5591, "step": 1805 }, { "epoch": 0.06434259044836739, "grad_norm": 1.419760372422666, "learning_rate": 1.9938320677937442e-05, "loss": 0.5435, "step": 1806 }, { "epoch": 0.0643782175748615, "grad_norm": 2.2564455951553257, "learning_rate": 1.9938192645335696e-05, "loss": 0.6736, "step": 1807 }, { "epoch": 0.06441384470135561, "grad_norm": 2.2756599792536156, "learning_rate": 1.9938064480399967e-05, "loss": 0.7074, "step": 1808 }, { "epoch": 0.06444947182784973, "grad_norm": 2.129271229025057, "learning_rate": 1.993793618313196e-05, "loss": 0.6665, "step": 1809 }, { "epoch": 0.06448509895434383, "grad_norm": 2.2613190967594714, "learning_rate": 1.993780775353339e-05, "loss": 0.667, "step": 1810 }, { "epoch": 0.06452072608083795, "grad_norm": 1.3201098892106524, "learning_rate": 1.9937679191605964e-05, "loss": 0.5684, "step": 1811 }, { "epoch": 0.06455635320733207, "grad_norm": 1.3153275365489219, "learning_rate": 1.9937550497351394e-05, "loss": 0.513, "step": 1812 }, { "epoch": 0.06459198033382617, "grad_norm": 1.3899483510382984, "learning_rate": 1.9937421670771392e-05, "loss": 0.5314, "step": 1813 }, { "epoch": 0.06462760746032029, "grad_norm": 1.4409310617007989, "learning_rate": 1.9937292711867676e-05, "loss": 0.5513, "step": 1814 }, { "epoch": 0.0646632345868144, "grad_norm": 1.4072421219235582, "learning_rate": 1.9937163620641957e-05, "loss": 0.5619, "step": 1815 }, { "epoch": 0.06469886171330852, "grad_norm": 1.2990137969893658, "learning_rate": 1.9937034397095963e-05, "loss": 0.4628, "step": 1816 }, { "epoch": 0.06473448883980262, "grad_norm": 2.597765076719649, "learning_rate": 1.9936905041231408e-05, "loss": 0.74, "step": 1817 }, { "epoch": 0.06477011596629674, "grad_norm": 1.2751489060764296, "learning_rate": 1.9936775553050017e-05, "loss": 0.5617, "step": 1818 }, { "epoch": 0.06480574309279086, "grad_norm": 1.9588867606574805, "learning_rate": 1.9936645932553515e-05, "loss": 0.7353, "step": 1819 }, { "epoch": 0.06484137021928496, "grad_norm": 2.221569066971491, "learning_rate": 1.993651617974363e-05, "loss": 0.6702, "step": 1820 }, { "epoch": 0.06487699734577908, "grad_norm": 2.2864490164244495, "learning_rate": 1.9936386294622084e-05, "loss": 0.6453, "step": 1821 }, { "epoch": 0.0649126244722732, "grad_norm": 1.4096456775520037, "learning_rate": 1.9936256277190608e-05, "loss": 0.539, "step": 1822 }, { "epoch": 0.0649482515987673, "grad_norm": 2.051532711949651, "learning_rate": 1.9936126127450935e-05, "loss": 0.6692, "step": 1823 }, { "epoch": 0.06498387872526142, "grad_norm": 2.039374092215643, "learning_rate": 1.99359958454048e-05, "loss": 0.679, "step": 1824 }, { "epoch": 0.06501950585175553, "grad_norm": 1.6811859419429829, "learning_rate": 1.9935865431053936e-05, "loss": 0.6051, "step": 1825 }, { "epoch": 0.06505513297824964, "grad_norm": 2.005721694424544, "learning_rate": 1.9935734884400076e-05, "loss": 0.5856, "step": 1826 }, { "epoch": 0.06509076010474375, "grad_norm": 1.4016116067420257, "learning_rate": 1.9935604205444962e-05, "loss": 0.5877, "step": 1827 }, { "epoch": 0.06512638723123787, "grad_norm": 1.7385721763846405, "learning_rate": 1.9935473394190335e-05, "loss": 0.6522, "step": 1828 }, { "epoch": 0.06516201435773197, "grad_norm": 1.427304180843595, "learning_rate": 1.9935342450637934e-05, "loss": 0.596, "step": 1829 }, { "epoch": 0.06519764148422609, "grad_norm": 1.8922737539000836, "learning_rate": 1.9935211374789506e-05, "loss": 0.6202, "step": 1830 }, { "epoch": 0.06523326861072021, "grad_norm": 2.282619323264976, "learning_rate": 1.993508016664679e-05, "loss": 0.5783, "step": 1831 }, { "epoch": 0.06526889573721431, "grad_norm": 1.814966275145972, "learning_rate": 1.993494882621154e-05, "loss": 0.6045, "step": 1832 }, { "epoch": 0.06530452286370843, "grad_norm": 2.177463763655771, "learning_rate": 1.99348173534855e-05, "loss": 0.6606, "step": 1833 }, { "epoch": 0.06534014999020255, "grad_norm": 1.4665148420688248, "learning_rate": 1.993468574847043e-05, "loss": 0.6164, "step": 1834 }, { "epoch": 0.06537577711669665, "grad_norm": 1.433746967134289, "learning_rate": 1.993455401116807e-05, "loss": 0.5524, "step": 1835 }, { "epoch": 0.06541140424319077, "grad_norm": 2.8163247760014074, "learning_rate": 1.993442214158018e-05, "loss": 0.6891, "step": 1836 }, { "epoch": 0.06544703136968488, "grad_norm": 1.3722998680515623, "learning_rate": 1.9934290139708516e-05, "loss": 0.5864, "step": 1837 }, { "epoch": 0.06548265849617899, "grad_norm": 1.8163497143256564, "learning_rate": 1.9934158005554833e-05, "loss": 0.7018, "step": 1838 }, { "epoch": 0.0655182856226731, "grad_norm": 1.339463528443824, "learning_rate": 1.9934025739120893e-05, "loss": 0.5444, "step": 1839 }, { "epoch": 0.06555391274916722, "grad_norm": 1.3458387231207922, "learning_rate": 1.993389334040846e-05, "loss": 0.579, "step": 1840 }, { "epoch": 0.06558953987566132, "grad_norm": 2.4247519527384083, "learning_rate": 1.993376080941929e-05, "loss": 0.6909, "step": 1841 }, { "epoch": 0.06562516700215544, "grad_norm": 2.4983477722001957, "learning_rate": 1.9933628146155154e-05, "loss": 0.6512, "step": 1842 }, { "epoch": 0.06566079412864956, "grad_norm": 1.422595297411537, "learning_rate": 1.9933495350617814e-05, "loss": 0.5365, "step": 1843 }, { "epoch": 0.06569642125514366, "grad_norm": 2.1481438944907714, "learning_rate": 1.9933362422809043e-05, "loss": 0.5879, "step": 1844 }, { "epoch": 0.06573204838163778, "grad_norm": 2.3433666749174487, "learning_rate": 1.9933229362730608e-05, "loss": 0.6475, "step": 1845 }, { "epoch": 0.0657676755081319, "grad_norm": 1.280965968802409, "learning_rate": 1.993309617038428e-05, "loss": 0.5623, "step": 1846 }, { "epoch": 0.065803302634626, "grad_norm": 2.306500000052155, "learning_rate": 1.9932962845771834e-05, "loss": 0.7191, "step": 1847 }, { "epoch": 0.06583892976112012, "grad_norm": 1.5093548086344615, "learning_rate": 1.993282938889504e-05, "loss": 0.57, "step": 1848 }, { "epoch": 0.06587455688761423, "grad_norm": 1.3559106310277655, "learning_rate": 1.9932695799755688e-05, "loss": 0.5503, "step": 1849 }, { "epoch": 0.06591018401410834, "grad_norm": 1.862637251352559, "learning_rate": 1.9932562078355544e-05, "loss": 0.693, "step": 1850 }, { "epoch": 0.06594581114060245, "grad_norm": 1.450426915030344, "learning_rate": 1.9932428224696396e-05, "loss": 0.469, "step": 1851 }, { "epoch": 0.06598143826709657, "grad_norm": 1.8975618439764173, "learning_rate": 1.9932294238780024e-05, "loss": 0.6329, "step": 1852 }, { "epoch": 0.06601706539359067, "grad_norm": 1.3995722973021245, "learning_rate": 1.993216012060821e-05, "loss": 0.5223, "step": 1853 }, { "epoch": 0.06605269252008479, "grad_norm": 2.3869663956961134, "learning_rate": 1.9932025870182744e-05, "loss": 0.6417, "step": 1854 }, { "epoch": 0.06608831964657891, "grad_norm": 1.3914417693008239, "learning_rate": 1.993189148750541e-05, "loss": 0.5808, "step": 1855 }, { "epoch": 0.06612394677307301, "grad_norm": 1.536574520741775, "learning_rate": 1.9931756972578e-05, "loss": 0.5829, "step": 1856 }, { "epoch": 0.06615957389956713, "grad_norm": 2.192871139001819, "learning_rate": 1.9931622325402304e-05, "loss": 0.6355, "step": 1857 }, { "epoch": 0.06619520102606125, "grad_norm": 1.5799877707878658, "learning_rate": 1.993148754598011e-05, "loss": 0.6087, "step": 1858 }, { "epoch": 0.06623082815255535, "grad_norm": 1.6980135373597232, "learning_rate": 1.9931352634313222e-05, "loss": 0.7238, "step": 1859 }, { "epoch": 0.06626645527904947, "grad_norm": 1.9724958384406228, "learning_rate": 1.9931217590403435e-05, "loss": 0.591, "step": 1860 }, { "epoch": 0.06630208240554358, "grad_norm": 1.7521889188953446, "learning_rate": 1.993108241425254e-05, "loss": 0.6895, "step": 1861 }, { "epoch": 0.0663377095320377, "grad_norm": 2.6252901503404034, "learning_rate": 1.993094710586234e-05, "loss": 0.6642, "step": 1862 }, { "epoch": 0.0663733366585318, "grad_norm": 2.9109971859813752, "learning_rate": 1.993081166523464e-05, "loss": 0.5446, "step": 1863 }, { "epoch": 0.06640896378502592, "grad_norm": 1.603690354018156, "learning_rate": 1.9930676092371243e-05, "loss": 0.6438, "step": 1864 }, { "epoch": 0.06644459091152004, "grad_norm": 1.9882940032888097, "learning_rate": 1.993054038727395e-05, "loss": 0.5852, "step": 1865 }, { "epoch": 0.06648021803801414, "grad_norm": 1.4351821336283697, "learning_rate": 1.9930404549944577e-05, "loss": 0.5967, "step": 1866 }, { "epoch": 0.06651584516450826, "grad_norm": 1.40698730890877, "learning_rate": 1.993026858038492e-05, "loss": 0.5371, "step": 1867 }, { "epoch": 0.06655147229100238, "grad_norm": 2.0523702527783043, "learning_rate": 1.9930132478596795e-05, "loss": 0.6314, "step": 1868 }, { "epoch": 0.06658709941749648, "grad_norm": 1.4889394395120257, "learning_rate": 1.992999624458202e-05, "loss": 0.5202, "step": 1869 }, { "epoch": 0.0666227265439906, "grad_norm": 1.340944575488283, "learning_rate": 1.99298598783424e-05, "loss": 0.5566, "step": 1870 }, { "epoch": 0.06665835367048471, "grad_norm": 1.3264551910505298, "learning_rate": 1.9929723379879758e-05, "loss": 0.556, "step": 1871 }, { "epoch": 0.06669398079697882, "grad_norm": 1.448618301901145, "learning_rate": 1.9929586749195907e-05, "loss": 0.5399, "step": 1872 }, { "epoch": 0.06672960792347293, "grad_norm": 2.0124424647151753, "learning_rate": 1.9929449986292672e-05, "loss": 0.7345, "step": 1873 }, { "epoch": 0.06676523504996705, "grad_norm": 1.6724412418001473, "learning_rate": 1.9929313091171866e-05, "loss": 0.5358, "step": 1874 }, { "epoch": 0.06680086217646115, "grad_norm": 1.4251937065467064, "learning_rate": 1.9929176063835317e-05, "loss": 0.5465, "step": 1875 }, { "epoch": 0.06683648930295527, "grad_norm": 2.2543130577619, "learning_rate": 1.992903890428485e-05, "loss": 0.6653, "step": 1876 }, { "epoch": 0.06687211642944939, "grad_norm": 1.4599897745257593, "learning_rate": 1.992890161252229e-05, "loss": 0.5428, "step": 1877 }, { "epoch": 0.06690774355594349, "grad_norm": 2.61082070305713, "learning_rate": 1.9928764188549462e-05, "loss": 0.6656, "step": 1878 }, { "epoch": 0.06694337068243761, "grad_norm": 2.4673199697261694, "learning_rate": 1.9928626632368204e-05, "loss": 0.6702, "step": 1879 }, { "epoch": 0.06697899780893173, "grad_norm": 3.0008477090559578, "learning_rate": 1.9928488943980338e-05, "loss": 0.6843, "step": 1880 }, { "epoch": 0.06701462493542583, "grad_norm": 1.4913069314848708, "learning_rate": 1.9928351123387703e-05, "loss": 0.5911, "step": 1881 }, { "epoch": 0.06705025206191995, "grad_norm": 3.5107885153737293, "learning_rate": 1.9928213170592133e-05, "loss": 0.7111, "step": 1882 }, { "epoch": 0.06708587918841406, "grad_norm": 1.5781609917530812, "learning_rate": 1.992807508559547e-05, "loss": 0.5959, "step": 1883 }, { "epoch": 0.06712150631490817, "grad_norm": 2.3984632459093937, "learning_rate": 1.992793686839954e-05, "loss": 0.4743, "step": 1884 }, { "epoch": 0.06715713344140228, "grad_norm": 1.8726455479209165, "learning_rate": 1.99277985190062e-05, "loss": 0.5882, "step": 1885 }, { "epoch": 0.0671927605678964, "grad_norm": 1.482955695230956, "learning_rate": 1.992766003741728e-05, "loss": 0.5885, "step": 1886 }, { "epoch": 0.0672283876943905, "grad_norm": 2.360386865173907, "learning_rate": 1.9927521423634627e-05, "loss": 0.6848, "step": 1887 }, { "epoch": 0.06726401482088462, "grad_norm": 2.196354020720704, "learning_rate": 1.992738267766009e-05, "loss": 0.7128, "step": 1888 }, { "epoch": 0.06729964194737874, "grad_norm": 1.4749973273317245, "learning_rate": 1.992724379949551e-05, "loss": 0.5759, "step": 1889 }, { "epoch": 0.06733526907387284, "grad_norm": 1.9958844657888974, "learning_rate": 1.992710478914274e-05, "loss": 0.8002, "step": 1890 }, { "epoch": 0.06737089620036696, "grad_norm": 2.1382642391958697, "learning_rate": 1.9926965646603632e-05, "loss": 0.602, "step": 1891 }, { "epoch": 0.06740652332686108, "grad_norm": 1.828074257776077, "learning_rate": 1.9926826371880035e-05, "loss": 0.5508, "step": 1892 }, { "epoch": 0.06744215045335518, "grad_norm": 1.513955982965228, "learning_rate": 1.9926686964973813e-05, "loss": 0.573, "step": 1893 }, { "epoch": 0.0674777775798493, "grad_norm": 1.2735811848204455, "learning_rate": 1.9926547425886812e-05, "loss": 0.574, "step": 1894 }, { "epoch": 0.06751340470634341, "grad_norm": 1.6010128333112876, "learning_rate": 1.9926407754620892e-05, "loss": 0.6134, "step": 1895 }, { "epoch": 0.06754903183283752, "grad_norm": 1.4139487974054512, "learning_rate": 1.9926267951177917e-05, "loss": 0.5472, "step": 1896 }, { "epoch": 0.06758465895933163, "grad_norm": 1.4659725358457343, "learning_rate": 1.992612801555975e-05, "loss": 0.5955, "step": 1897 }, { "epoch": 0.06762028608582575, "grad_norm": 2.77346125204732, "learning_rate": 1.9925987947768244e-05, "loss": 0.6177, "step": 1898 }, { "epoch": 0.06765591321231985, "grad_norm": 2.094475746198554, "learning_rate": 1.9925847747805274e-05, "loss": 0.5762, "step": 1899 }, { "epoch": 0.06769154033881397, "grad_norm": 2.2649082934562172, "learning_rate": 1.99257074156727e-05, "loss": 0.5599, "step": 1900 }, { "epoch": 0.06772716746530809, "grad_norm": 1.510369609006074, "learning_rate": 1.99255669513724e-05, "loss": 0.5953, "step": 1901 }, { "epoch": 0.06776279459180219, "grad_norm": 5.19220066414617, "learning_rate": 1.992542635490623e-05, "loss": 0.6101, "step": 1902 }, { "epoch": 0.06779842171829631, "grad_norm": 1.718616004651328, "learning_rate": 1.992528562627608e-05, "loss": 0.7008, "step": 1903 }, { "epoch": 0.06783404884479043, "grad_norm": 2.077949746655911, "learning_rate": 1.9925144765483807e-05, "loss": 0.6032, "step": 1904 }, { "epoch": 0.06786967597128453, "grad_norm": 2.4256268969369326, "learning_rate": 1.99250037725313e-05, "loss": 0.6556, "step": 1905 }, { "epoch": 0.06790530309777865, "grad_norm": 1.906744163934658, "learning_rate": 1.992486264742043e-05, "loss": 0.6006, "step": 1906 }, { "epoch": 0.06794093022427276, "grad_norm": 2.219175664646434, "learning_rate": 1.9924721390153072e-05, "loss": 0.656, "step": 1907 }, { "epoch": 0.06797655735076688, "grad_norm": 1.5150412842011773, "learning_rate": 1.9924580000731116e-05, "loss": 0.5209, "step": 1908 }, { "epoch": 0.06801218447726098, "grad_norm": 1.4832615129721138, "learning_rate": 1.9924438479156437e-05, "loss": 0.5848, "step": 1909 }, { "epoch": 0.0680478116037551, "grad_norm": 1.3795740304505906, "learning_rate": 1.9924296825430928e-05, "loss": 0.5323, "step": 1910 }, { "epoch": 0.06808343873024922, "grad_norm": 1.352180401840993, "learning_rate": 1.9924155039556464e-05, "loss": 0.5605, "step": 1911 }, { "epoch": 0.06811906585674332, "grad_norm": 2.307082200470907, "learning_rate": 1.9924013121534943e-05, "loss": 0.6705, "step": 1912 }, { "epoch": 0.06815469298323744, "grad_norm": 1.3579403717720533, "learning_rate": 1.9923871071368253e-05, "loss": 0.5927, "step": 1913 }, { "epoch": 0.06819032010973156, "grad_norm": 2.2146935657139206, "learning_rate": 1.9923728889058277e-05, "loss": 0.6274, "step": 1914 }, { "epoch": 0.06822594723622566, "grad_norm": 1.9334300921804732, "learning_rate": 1.992358657460692e-05, "loss": 0.5868, "step": 1915 }, { "epoch": 0.06826157436271978, "grad_norm": 1.6932844486028948, "learning_rate": 1.992344412801607e-05, "loss": 0.6084, "step": 1916 }, { "epoch": 0.0682972014892139, "grad_norm": 2.085658667035152, "learning_rate": 1.9923301549287624e-05, "loss": 0.5375, "step": 1917 }, { "epoch": 0.068332828615708, "grad_norm": 1.6942455997583734, "learning_rate": 1.9923158838423482e-05, "loss": 0.6248, "step": 1918 }, { "epoch": 0.06836845574220211, "grad_norm": 1.5992046535037152, "learning_rate": 1.9923015995425547e-05, "loss": 0.5457, "step": 1919 }, { "epoch": 0.06840408286869623, "grad_norm": 1.4846168143853413, "learning_rate": 1.9922873020295714e-05, "loss": 0.5582, "step": 1920 }, { "epoch": 0.06843970999519033, "grad_norm": 2.6110489914107258, "learning_rate": 1.9922729913035893e-05, "loss": 0.6536, "step": 1921 }, { "epoch": 0.06847533712168445, "grad_norm": 1.7816051042426564, "learning_rate": 1.992258667364799e-05, "loss": 0.6432, "step": 1922 }, { "epoch": 0.06851096424817857, "grad_norm": 1.3655328012955, "learning_rate": 1.9922443302133906e-05, "loss": 0.47, "step": 1923 }, { "epoch": 0.06854659137467267, "grad_norm": 1.545226503376536, "learning_rate": 1.9922299798495555e-05, "loss": 0.6038, "step": 1924 }, { "epoch": 0.06858221850116679, "grad_norm": 1.4400032027079448, "learning_rate": 1.9922156162734846e-05, "loss": 0.5674, "step": 1925 }, { "epoch": 0.0686178456276609, "grad_norm": 1.415654629028422, "learning_rate": 1.9922012394853695e-05, "loss": 0.5713, "step": 1926 }, { "epoch": 0.06865347275415501, "grad_norm": 1.4047024436522526, "learning_rate": 1.9921868494854014e-05, "loss": 0.5606, "step": 1927 }, { "epoch": 0.06868909988064913, "grad_norm": 2.0451500989201703, "learning_rate": 1.992172446273772e-05, "loss": 0.606, "step": 1928 }, { "epoch": 0.06872472700714324, "grad_norm": 1.9247769370256087, "learning_rate": 1.9921580298506728e-05, "loss": 0.541, "step": 1929 }, { "epoch": 0.06876035413363735, "grad_norm": 1.8965179555932639, "learning_rate": 1.992143600216296e-05, "loss": 0.7148, "step": 1930 }, { "epoch": 0.06879598126013146, "grad_norm": 1.7587909739851302, "learning_rate": 1.9921291573708338e-05, "loss": 0.6717, "step": 1931 }, { "epoch": 0.06883160838662558, "grad_norm": 1.4666243570463098, "learning_rate": 1.9921147013144782e-05, "loss": 0.5505, "step": 1932 }, { "epoch": 0.06886723551311968, "grad_norm": 1.3600973403763876, "learning_rate": 1.9921002320474218e-05, "loss": 0.5365, "step": 1933 }, { "epoch": 0.0689028626396138, "grad_norm": 1.2543703184305872, "learning_rate": 1.9920857495698575e-05, "loss": 0.5371, "step": 1934 }, { "epoch": 0.06893848976610792, "grad_norm": 2.2886746655813464, "learning_rate": 1.992071253881978e-05, "loss": 0.7141, "step": 1935 }, { "epoch": 0.06897411689260202, "grad_norm": 1.3131990621146474, "learning_rate": 1.9920567449839764e-05, "loss": 0.5495, "step": 1936 }, { "epoch": 0.06900974401909614, "grad_norm": 1.3397775252107798, "learning_rate": 1.9920422228760458e-05, "loss": 0.4915, "step": 1937 }, { "epoch": 0.06904537114559026, "grad_norm": 1.4686525862027602, "learning_rate": 1.99202768755838e-05, "loss": 0.5178, "step": 1938 }, { "epoch": 0.06908099827208436, "grad_norm": 2.457376681578538, "learning_rate": 1.9920131390311716e-05, "loss": 0.7452, "step": 1939 }, { "epoch": 0.06911662539857848, "grad_norm": 1.4863756253268954, "learning_rate": 1.991998577294615e-05, "loss": 0.5948, "step": 1940 }, { "epoch": 0.0691522525250726, "grad_norm": 1.7668353105817922, "learning_rate": 1.991984002348904e-05, "loss": 0.539, "step": 1941 }, { "epoch": 0.0691878796515667, "grad_norm": 1.5665728465641902, "learning_rate": 1.991969414194233e-05, "loss": 0.5865, "step": 1942 }, { "epoch": 0.06922350677806081, "grad_norm": 1.4106589820668551, "learning_rate": 1.9919548128307954e-05, "loss": 0.5375, "step": 1943 }, { "epoch": 0.06925913390455493, "grad_norm": 1.2376917316996459, "learning_rate": 1.9919401982587865e-05, "loss": 0.4901, "step": 1944 }, { "epoch": 0.06929476103104903, "grad_norm": 1.3257171567468196, "learning_rate": 1.9919255704784004e-05, "loss": 0.5323, "step": 1945 }, { "epoch": 0.06933038815754315, "grad_norm": 2.480490337231705, "learning_rate": 1.9919109294898317e-05, "loss": 0.6111, "step": 1946 }, { "epoch": 0.06936601528403727, "grad_norm": 2.2352065201382967, "learning_rate": 1.991896275293276e-05, "loss": 0.6327, "step": 1947 }, { "epoch": 0.06940164241053137, "grad_norm": 1.804732330255047, "learning_rate": 1.991881607888928e-05, "loss": 0.6551, "step": 1948 }, { "epoch": 0.06943726953702549, "grad_norm": 2.174413530500234, "learning_rate": 1.9918669272769832e-05, "loss": 0.7119, "step": 1949 }, { "epoch": 0.0694728966635196, "grad_norm": 2.4326140284936746, "learning_rate": 1.991852233457637e-05, "loss": 0.6513, "step": 1950 }, { "epoch": 0.06950852379001371, "grad_norm": 1.353018635585256, "learning_rate": 1.991837526431085e-05, "loss": 0.4893, "step": 1951 }, { "epoch": 0.06954415091650783, "grad_norm": 1.5358528541836094, "learning_rate": 1.9918228061975234e-05, "loss": 0.5094, "step": 1952 }, { "epoch": 0.06957977804300194, "grad_norm": 1.4194772977843484, "learning_rate": 1.9918080727571473e-05, "loss": 0.5328, "step": 1953 }, { "epoch": 0.06961540516949605, "grad_norm": 2.111174545486803, "learning_rate": 1.991793326110154e-05, "loss": 0.6038, "step": 1954 }, { "epoch": 0.06965103229599016, "grad_norm": 1.9472903729779136, "learning_rate": 1.991778566256739e-05, "loss": 0.6894, "step": 1955 }, { "epoch": 0.06968665942248428, "grad_norm": 2.0865255453134908, "learning_rate": 1.9917637931970994e-05, "loss": 0.5954, "step": 1956 }, { "epoch": 0.0697222865489784, "grad_norm": 1.861922854584134, "learning_rate": 1.9917490069314315e-05, "loss": 0.5904, "step": 1957 }, { "epoch": 0.0697579136754725, "grad_norm": 1.5606712798027902, "learning_rate": 1.9917342074599325e-05, "loss": 0.5623, "step": 1958 }, { "epoch": 0.06979354080196662, "grad_norm": 1.5337883294603074, "learning_rate": 1.991719394782799e-05, "loss": 0.5412, "step": 1959 }, { "epoch": 0.06982916792846074, "grad_norm": 2.6888681741937077, "learning_rate": 1.9917045689002294e-05, "loss": 0.6531, "step": 1960 }, { "epoch": 0.06986479505495484, "grad_norm": 1.905226715598018, "learning_rate": 1.9916897298124197e-05, "loss": 0.6933, "step": 1961 }, { "epoch": 0.06990042218144896, "grad_norm": 1.3788548369547708, "learning_rate": 1.991674877519568e-05, "loss": 0.5498, "step": 1962 }, { "epoch": 0.06993604930794307, "grad_norm": 1.9399499999408443, "learning_rate": 1.9916600120218724e-05, "loss": 0.6377, "step": 1963 }, { "epoch": 0.06997167643443718, "grad_norm": 2.1834479975277574, "learning_rate": 1.9916451333195305e-05, "loss": 0.557, "step": 1964 }, { "epoch": 0.0700073035609313, "grad_norm": 2.31881347612397, "learning_rate": 1.9916302414127408e-05, "loss": 0.7095, "step": 1965 }, { "epoch": 0.07004293068742541, "grad_norm": 2.2198104354476382, "learning_rate": 1.9916153363017012e-05, "loss": 0.6452, "step": 1966 }, { "epoch": 0.07007855781391951, "grad_norm": 2.1821257191083703, "learning_rate": 1.99160041798661e-05, "loss": 0.5941, "step": 1967 }, { "epoch": 0.07011418494041363, "grad_norm": 1.6554453011171042, "learning_rate": 1.9915854864676665e-05, "loss": 0.6028, "step": 1968 }, { "epoch": 0.07014981206690775, "grad_norm": 1.4731732258733286, "learning_rate": 1.991570541745069e-05, "loss": 0.5125, "step": 1969 }, { "epoch": 0.07018543919340185, "grad_norm": 1.539602735562771, "learning_rate": 1.9915555838190167e-05, "loss": 0.5625, "step": 1970 }, { "epoch": 0.07022106631989597, "grad_norm": 1.3142228340510786, "learning_rate": 1.991540612689709e-05, "loss": 0.5741, "step": 1971 }, { "epoch": 0.07025669344639009, "grad_norm": 3.595895090392528, "learning_rate": 1.9915256283573446e-05, "loss": 0.6326, "step": 1972 }, { "epoch": 0.07029232057288419, "grad_norm": 1.7678575460544674, "learning_rate": 1.9915106308221234e-05, "loss": 0.6198, "step": 1973 }, { "epoch": 0.07032794769937831, "grad_norm": 2.473316475106005, "learning_rate": 1.9914956200842453e-05, "loss": 0.6167, "step": 1974 }, { "epoch": 0.07036357482587242, "grad_norm": 1.376857593070322, "learning_rate": 1.99148059614391e-05, "loss": 0.5553, "step": 1975 }, { "epoch": 0.07039920195236653, "grad_norm": 1.5630375627693651, "learning_rate": 1.9914655590013177e-05, "loss": 0.6123, "step": 1976 }, { "epoch": 0.07043482907886064, "grad_norm": 1.819843706204592, "learning_rate": 1.9914505086566683e-05, "loss": 0.6077, "step": 1977 }, { "epoch": 0.07047045620535476, "grad_norm": 1.402008870987646, "learning_rate": 1.9914354451101627e-05, "loss": 0.5401, "step": 1978 }, { "epoch": 0.07050608333184887, "grad_norm": 2.325646821069176, "learning_rate": 1.991420368362001e-05, "loss": 0.5745, "step": 1979 }, { "epoch": 0.07054171045834298, "grad_norm": 1.943068105674672, "learning_rate": 1.991405278412384e-05, "loss": 0.6059, "step": 1980 }, { "epoch": 0.0705773375848371, "grad_norm": 1.896325500196238, "learning_rate": 1.991390175261513e-05, "loss": 0.6511, "step": 1981 }, { "epoch": 0.0706129647113312, "grad_norm": 1.3281913784584078, "learning_rate": 1.9913750589095886e-05, "loss": 0.6099, "step": 1982 }, { "epoch": 0.07064859183782532, "grad_norm": 1.934431566326525, "learning_rate": 1.9913599293568124e-05, "loss": 0.5855, "step": 1983 }, { "epoch": 0.07068421896431944, "grad_norm": 1.770240581406861, "learning_rate": 1.9913447866033858e-05, "loss": 0.686, "step": 1984 }, { "epoch": 0.07071984609081354, "grad_norm": 2.887409236886243, "learning_rate": 1.9913296306495108e-05, "loss": 0.6566, "step": 1985 }, { "epoch": 0.07075547321730766, "grad_norm": 1.3573361926295664, "learning_rate": 1.9913144614953883e-05, "loss": 0.519, "step": 1986 }, { "epoch": 0.07079110034380177, "grad_norm": 1.3773689957835205, "learning_rate": 1.9912992791412212e-05, "loss": 0.5823, "step": 1987 }, { "epoch": 0.07082672747029588, "grad_norm": 2.5584263239286082, "learning_rate": 1.9912840835872112e-05, "loss": 0.6321, "step": 1988 }, { "epoch": 0.07086235459679, "grad_norm": 1.2912642786625965, "learning_rate": 1.991268874833561e-05, "loss": 0.5458, "step": 1989 }, { "epoch": 0.07089798172328411, "grad_norm": 1.7228109302787542, "learning_rate": 1.991253652880473e-05, "loss": 0.6067, "step": 1990 }, { "epoch": 0.07093360884977822, "grad_norm": 2.317454553904374, "learning_rate": 1.9912384177281494e-05, "loss": 0.6365, "step": 1991 }, { "epoch": 0.07096923597627233, "grad_norm": 2.4389374949634783, "learning_rate": 1.9912231693767933e-05, "loss": 0.6951, "step": 1992 }, { "epoch": 0.07100486310276645, "grad_norm": 1.3352760150108913, "learning_rate": 1.9912079078266086e-05, "loss": 0.487, "step": 1993 }, { "epoch": 0.07104049022926055, "grad_norm": 2.2061182446255065, "learning_rate": 1.991192633077797e-05, "loss": 0.6268, "step": 1994 }, { "epoch": 0.07107611735575467, "grad_norm": 1.8499235850471538, "learning_rate": 1.991177345130563e-05, "loss": 0.6447, "step": 1995 }, { "epoch": 0.07111174448224879, "grad_norm": 1.5278819744771768, "learning_rate": 1.99116204398511e-05, "loss": 0.6238, "step": 1996 }, { "epoch": 0.07114737160874289, "grad_norm": 2.1799557114339425, "learning_rate": 1.9911467296416412e-05, "loss": 0.6206, "step": 1997 }, { "epoch": 0.07118299873523701, "grad_norm": 2.2969765651926233, "learning_rate": 1.9911314021003614e-05, "loss": 0.7002, "step": 1998 }, { "epoch": 0.07121862586173112, "grad_norm": 1.3361232777828405, "learning_rate": 1.991116061361474e-05, "loss": 0.5081, "step": 1999 }, { "epoch": 0.07125425298822523, "grad_norm": 1.8870221117669004, "learning_rate": 1.9911007074251832e-05, "loss": 0.5606, "step": 2000 }, { "epoch": 0.07128988011471935, "grad_norm": 2.5802739727190285, "learning_rate": 1.991085340291694e-05, "loss": 0.6255, "step": 2001 }, { "epoch": 0.07132550724121346, "grad_norm": 1.9700317478536191, "learning_rate": 1.9910699599612112e-05, "loss": 0.6513, "step": 2002 }, { "epoch": 0.07136113436770758, "grad_norm": 1.4000445202986913, "learning_rate": 1.9910545664339387e-05, "loss": 0.4705, "step": 2003 }, { "epoch": 0.07139676149420168, "grad_norm": 1.38813757167113, "learning_rate": 1.991039159710082e-05, "loss": 0.549, "step": 2004 }, { "epoch": 0.0714323886206958, "grad_norm": 1.7086355088862148, "learning_rate": 1.9910237397898463e-05, "loss": 0.5548, "step": 2005 }, { "epoch": 0.07146801574718992, "grad_norm": 1.312229922090176, "learning_rate": 1.9910083066734366e-05, "loss": 0.5579, "step": 2006 }, { "epoch": 0.07150364287368402, "grad_norm": 1.353319299307791, "learning_rate": 1.990992860361059e-05, "loss": 0.5177, "step": 2007 }, { "epoch": 0.07153927000017814, "grad_norm": 2.0203154572717006, "learning_rate": 1.9909774008529184e-05, "loss": 0.6467, "step": 2008 }, { "epoch": 0.07157489712667225, "grad_norm": 1.6013332921690744, "learning_rate": 1.990961928149221e-05, "loss": 0.5893, "step": 2009 }, { "epoch": 0.07161052425316636, "grad_norm": 1.9885775718031107, "learning_rate": 1.9909464422501733e-05, "loss": 0.6765, "step": 2010 }, { "epoch": 0.07164615137966047, "grad_norm": 2.1500153910223307, "learning_rate": 1.9909309431559808e-05, "loss": 0.5597, "step": 2011 }, { "epoch": 0.07168177850615459, "grad_norm": 1.7342222087695423, "learning_rate": 1.9909154308668503e-05, "loss": 0.5676, "step": 2012 }, { "epoch": 0.0717174056326487, "grad_norm": 2.2954919999685903, "learning_rate": 1.990899905382988e-05, "loss": 0.6134, "step": 2013 }, { "epoch": 0.07175303275914281, "grad_norm": 1.8438401890240017, "learning_rate": 1.9908843667046014e-05, "loss": 0.6632, "step": 2014 }, { "epoch": 0.07178865988563693, "grad_norm": 1.2927844753367503, "learning_rate": 1.9908688148318964e-05, "loss": 0.5349, "step": 2015 }, { "epoch": 0.07182428701213103, "grad_norm": 1.4047128936443165, "learning_rate": 1.9908532497650808e-05, "loss": 0.5551, "step": 2016 }, { "epoch": 0.07185991413862515, "grad_norm": 1.392096277942647, "learning_rate": 1.9908376715043616e-05, "loss": 0.5478, "step": 2017 }, { "epoch": 0.07189554126511927, "grad_norm": 1.5298128056207336, "learning_rate": 1.990822080049946e-05, "loss": 0.5786, "step": 2018 }, { "epoch": 0.07193116839161337, "grad_norm": 1.2388196641261553, "learning_rate": 1.990806475402042e-05, "loss": 0.5426, "step": 2019 }, { "epoch": 0.07196679551810749, "grad_norm": 3.5649312775302278, "learning_rate": 1.9907908575608573e-05, "loss": 0.6007, "step": 2020 }, { "epoch": 0.0720024226446016, "grad_norm": 1.307682199835335, "learning_rate": 1.9907752265265997e-05, "loss": 0.5258, "step": 2021 }, { "epoch": 0.07203804977109571, "grad_norm": 1.6900916569683162, "learning_rate": 1.9907595822994774e-05, "loss": 0.6539, "step": 2022 }, { "epoch": 0.07207367689758983, "grad_norm": 1.3593195431550897, "learning_rate": 1.990743924879699e-05, "loss": 0.545, "step": 2023 }, { "epoch": 0.07210930402408394, "grad_norm": 1.477249960772742, "learning_rate": 1.9907282542674726e-05, "loss": 0.5814, "step": 2024 }, { "epoch": 0.07214493115057805, "grad_norm": 1.8972389483013161, "learning_rate": 1.990712570463007e-05, "loss": 0.7216, "step": 2025 }, { "epoch": 0.07218055827707216, "grad_norm": 2.1981862817440403, "learning_rate": 1.990696873466511e-05, "loss": 0.6777, "step": 2026 }, { "epoch": 0.07221618540356628, "grad_norm": 1.2275265869884735, "learning_rate": 1.990681163278194e-05, "loss": 0.5667, "step": 2027 }, { "epoch": 0.07225181253006038, "grad_norm": 2.229063102716507, "learning_rate": 1.9906654398982647e-05, "loss": 0.5623, "step": 2028 }, { "epoch": 0.0722874396565545, "grad_norm": 2.1564723878596226, "learning_rate": 1.9906497033269325e-05, "loss": 0.6845, "step": 2029 }, { "epoch": 0.07232306678304862, "grad_norm": 1.4579819454125524, "learning_rate": 1.990633953564407e-05, "loss": 0.5558, "step": 2030 }, { "epoch": 0.07235869390954272, "grad_norm": 1.7878317907463523, "learning_rate": 1.9906181906108983e-05, "loss": 0.6594, "step": 2031 }, { "epoch": 0.07239432103603684, "grad_norm": 1.3864094634739028, "learning_rate": 1.9906024144666158e-05, "loss": 0.4964, "step": 2032 }, { "epoch": 0.07242994816253095, "grad_norm": 1.9879838457379202, "learning_rate": 1.9905866251317697e-05, "loss": 0.6613, "step": 2033 }, { "epoch": 0.07246557528902506, "grad_norm": 1.8496864073694894, "learning_rate": 1.9905708226065704e-05, "loss": 0.6643, "step": 2034 }, { "epoch": 0.07250120241551918, "grad_norm": 1.2286792492091207, "learning_rate": 1.9905550068912284e-05, "loss": 0.5066, "step": 2035 }, { "epoch": 0.07253682954201329, "grad_norm": 1.3410016592947618, "learning_rate": 1.990539177985954e-05, "loss": 0.5485, "step": 2036 }, { "epoch": 0.0725724566685074, "grad_norm": 1.9143804196464838, "learning_rate": 1.990523335890958e-05, "loss": 0.6363, "step": 2037 }, { "epoch": 0.07260808379500151, "grad_norm": 1.726798398744202, "learning_rate": 1.9905074806064514e-05, "loss": 0.6677, "step": 2038 }, { "epoch": 0.07264371092149563, "grad_norm": 1.3572658581305665, "learning_rate": 1.9904916121326455e-05, "loss": 0.5354, "step": 2039 }, { "epoch": 0.07267933804798973, "grad_norm": 1.929614489706408, "learning_rate": 1.9904757304697516e-05, "loss": 0.6268, "step": 2040 }, { "epoch": 0.07271496517448385, "grad_norm": 1.6861528195472575, "learning_rate": 1.9904598356179808e-05, "loss": 0.6437, "step": 2041 }, { "epoch": 0.07275059230097797, "grad_norm": 1.3839509283756535, "learning_rate": 1.990443927577545e-05, "loss": 0.5069, "step": 2042 }, { "epoch": 0.07278621942747207, "grad_norm": 1.7979271091574733, "learning_rate": 1.9904280063486563e-05, "loss": 0.6048, "step": 2043 }, { "epoch": 0.07282184655396619, "grad_norm": 1.4015396533605773, "learning_rate": 1.9904120719315262e-05, "loss": 0.5581, "step": 2044 }, { "epoch": 0.0728574736804603, "grad_norm": 2.6981304520294795, "learning_rate": 1.9903961243263673e-05, "loss": 0.6366, "step": 2045 }, { "epoch": 0.07289310080695441, "grad_norm": 1.908710228150243, "learning_rate": 1.9903801635333915e-05, "loss": 0.5486, "step": 2046 }, { "epoch": 0.07292872793344853, "grad_norm": 1.3393089586061075, "learning_rate": 1.9903641895528117e-05, "loss": 0.5218, "step": 2047 }, { "epoch": 0.07296435505994264, "grad_norm": 1.7918933432642083, "learning_rate": 1.9903482023848407e-05, "loss": 0.648, "step": 2048 }, { "epoch": 0.07299998218643676, "grad_norm": 1.4532060981187842, "learning_rate": 1.9903322020296908e-05, "loss": 0.5315, "step": 2049 }, { "epoch": 0.07303560931293086, "grad_norm": 2.0355855804087066, "learning_rate": 1.9903161884875756e-05, "loss": 0.7231, "step": 2050 }, { "epoch": 0.07307123643942498, "grad_norm": 1.6970558408652958, "learning_rate": 1.9903001617587082e-05, "loss": 0.6251, "step": 2051 }, { "epoch": 0.0731068635659191, "grad_norm": 1.6383443458400593, "learning_rate": 1.990284121843302e-05, "loss": 0.562, "step": 2052 }, { "epoch": 0.0731424906924132, "grad_norm": 1.3820485486678884, "learning_rate": 1.9902680687415704e-05, "loss": 0.5218, "step": 2053 }, { "epoch": 0.07317811781890732, "grad_norm": 1.312472455518063, "learning_rate": 1.9902520024537277e-05, "loss": 0.5197, "step": 2054 }, { "epoch": 0.07321374494540143, "grad_norm": 1.2881240148190158, "learning_rate": 1.990235922979987e-05, "loss": 0.4612, "step": 2055 }, { "epoch": 0.07324937207189554, "grad_norm": 2.0282745000339877, "learning_rate": 1.990219830320563e-05, "loss": 0.6107, "step": 2056 }, { "epoch": 0.07328499919838966, "grad_norm": 1.4719237820632936, "learning_rate": 1.9902037244756697e-05, "loss": 0.5246, "step": 2057 }, { "epoch": 0.07332062632488377, "grad_norm": 1.472506813948263, "learning_rate": 1.9901876054455217e-05, "loss": 0.5158, "step": 2058 }, { "epoch": 0.07335625345137788, "grad_norm": 2.0383273910877957, "learning_rate": 1.990171473230334e-05, "loss": 0.7045, "step": 2059 }, { "epoch": 0.07339188057787199, "grad_norm": 1.3971632963643965, "learning_rate": 1.9901553278303206e-05, "loss": 0.5484, "step": 2060 }, { "epoch": 0.07342750770436611, "grad_norm": 1.4178586750125637, "learning_rate": 1.9901391692456974e-05, "loss": 0.4699, "step": 2061 }, { "epoch": 0.07346313483086021, "grad_norm": 1.761101813948154, "learning_rate": 1.9901229974766787e-05, "loss": 0.6586, "step": 2062 }, { "epoch": 0.07349876195735433, "grad_norm": 1.8537164425218378, "learning_rate": 1.9901068125234805e-05, "loss": 0.598, "step": 2063 }, { "epoch": 0.07353438908384845, "grad_norm": 1.842038569131198, "learning_rate": 1.990090614386318e-05, "loss": 0.6723, "step": 2064 }, { "epoch": 0.07357001621034255, "grad_norm": 1.5232333589972924, "learning_rate": 1.9900744030654066e-05, "loss": 0.6063, "step": 2065 }, { "epoch": 0.07360564333683667, "grad_norm": 1.942727187805144, "learning_rate": 1.9900581785609627e-05, "loss": 0.6396, "step": 2066 }, { "epoch": 0.07364127046333079, "grad_norm": 1.2592002028997475, "learning_rate": 1.9900419408732024e-05, "loss": 0.5369, "step": 2067 }, { "epoch": 0.07367689758982489, "grad_norm": 1.4478042395317872, "learning_rate": 1.9900256900023413e-05, "loss": 0.5581, "step": 2068 }, { "epoch": 0.073712524716319, "grad_norm": 1.5566288837686442, "learning_rate": 1.990009425948596e-05, "loss": 0.6395, "step": 2069 }, { "epoch": 0.07374815184281312, "grad_norm": 1.3841565709305155, "learning_rate": 1.9899931487121838e-05, "loss": 0.5832, "step": 2070 }, { "epoch": 0.07378377896930723, "grad_norm": 1.315367633443702, "learning_rate": 1.9899768582933205e-05, "loss": 0.5862, "step": 2071 }, { "epoch": 0.07381940609580134, "grad_norm": 2.3820786354258647, "learning_rate": 1.9899605546922237e-05, "loss": 0.6775, "step": 2072 }, { "epoch": 0.07385503322229546, "grad_norm": 3.4256889442172405, "learning_rate": 1.98994423790911e-05, "loss": 0.6668, "step": 2073 }, { "epoch": 0.07389066034878956, "grad_norm": 1.3822153572185711, "learning_rate": 1.9899279079441967e-05, "loss": 0.5626, "step": 2074 }, { "epoch": 0.07392628747528368, "grad_norm": 1.307319888452981, "learning_rate": 1.9899115647977015e-05, "loss": 0.5734, "step": 2075 }, { "epoch": 0.0739619146017778, "grad_norm": 2.139603315769266, "learning_rate": 1.9898952084698418e-05, "loss": 0.6808, "step": 2076 }, { "epoch": 0.0739975417282719, "grad_norm": 1.3101396807553036, "learning_rate": 1.9898788389608356e-05, "loss": 0.5509, "step": 2077 }, { "epoch": 0.07403316885476602, "grad_norm": 1.2654601806306218, "learning_rate": 1.9898624562709007e-05, "loss": 0.5589, "step": 2078 }, { "epoch": 0.07406879598126014, "grad_norm": 2.105528597664333, "learning_rate": 1.9898460604002555e-05, "loss": 0.648, "step": 2079 }, { "epoch": 0.07410442310775424, "grad_norm": 1.2961742693783027, "learning_rate": 1.9898296513491178e-05, "loss": 0.5428, "step": 2080 }, { "epoch": 0.07414005023424836, "grad_norm": 2.1212328513115777, "learning_rate": 1.989813229117707e-05, "loss": 0.6447, "step": 2081 }, { "epoch": 0.07417567736074247, "grad_norm": 1.9191176308086932, "learning_rate": 1.9897967937062407e-05, "loss": 0.6137, "step": 2082 }, { "epoch": 0.07421130448723658, "grad_norm": 3.5221539962903075, "learning_rate": 1.9897803451149385e-05, "loss": 0.6603, "step": 2083 }, { "epoch": 0.0742469316137307, "grad_norm": 2.7960875573751274, "learning_rate": 1.989763883344019e-05, "loss": 0.651, "step": 2084 }, { "epoch": 0.07428255874022481, "grad_norm": 1.5417590508938337, "learning_rate": 1.9897474083937017e-05, "loss": 0.531, "step": 2085 }, { "epoch": 0.07431818586671891, "grad_norm": 2.1086796052186725, "learning_rate": 1.989730920264206e-05, "loss": 0.679, "step": 2086 }, { "epoch": 0.07435381299321303, "grad_norm": 1.5517798598420878, "learning_rate": 1.9897144189557513e-05, "loss": 0.5299, "step": 2087 }, { "epoch": 0.07438944011970715, "grad_norm": 2.082791518892542, "learning_rate": 1.989697904468557e-05, "loss": 0.6571, "step": 2088 }, { "epoch": 0.07442506724620125, "grad_norm": 1.922703160751861, "learning_rate": 1.9896813768028437e-05, "loss": 0.6196, "step": 2089 }, { "epoch": 0.07446069437269537, "grad_norm": 1.9719529340314057, "learning_rate": 1.9896648359588312e-05, "loss": 0.6375, "step": 2090 }, { "epoch": 0.07449632149918949, "grad_norm": 1.7748338258374623, "learning_rate": 1.9896482819367393e-05, "loss": 0.5874, "step": 2091 }, { "epoch": 0.07453194862568359, "grad_norm": 2.4578443380676807, "learning_rate": 1.989631714736789e-05, "loss": 0.7011, "step": 2092 }, { "epoch": 0.0745675757521777, "grad_norm": 2.018925968568235, "learning_rate": 1.9896151343592007e-05, "loss": 0.6728, "step": 2093 }, { "epoch": 0.07460320287867182, "grad_norm": 2.919902577694286, "learning_rate": 1.9895985408041954e-05, "loss": 0.6665, "step": 2094 }, { "epoch": 0.07463883000516593, "grad_norm": 2.0979975061732348, "learning_rate": 1.9895819340719934e-05, "loss": 0.5946, "step": 2095 }, { "epoch": 0.07467445713166004, "grad_norm": 2.1002542876059627, "learning_rate": 1.9895653141628166e-05, "loss": 0.7055, "step": 2096 }, { "epoch": 0.07471008425815416, "grad_norm": 2.5103300430173183, "learning_rate": 1.989548681076886e-05, "loss": 0.6657, "step": 2097 }, { "epoch": 0.07474571138464828, "grad_norm": 1.236157170107569, "learning_rate": 1.9895320348144226e-05, "loss": 0.4995, "step": 2098 }, { "epoch": 0.07478133851114238, "grad_norm": 1.98454264717889, "learning_rate": 1.989515375375649e-05, "loss": 0.6448, "step": 2099 }, { "epoch": 0.0748169656376365, "grad_norm": 1.3610911066347593, "learning_rate": 1.9894987027607863e-05, "loss": 0.5202, "step": 2100 }, { "epoch": 0.07485259276413062, "grad_norm": 1.4423727402421542, "learning_rate": 1.9894820169700567e-05, "loss": 0.639, "step": 2101 }, { "epoch": 0.07488821989062472, "grad_norm": 1.9724862566892123, "learning_rate": 1.989465318003683e-05, "loss": 0.6453, "step": 2102 }, { "epoch": 0.07492384701711884, "grad_norm": 1.78547847986173, "learning_rate": 1.9894486058618863e-05, "loss": 0.6655, "step": 2103 }, { "epoch": 0.07495947414361295, "grad_norm": 1.351281695978665, "learning_rate": 1.98943188054489e-05, "loss": 0.513, "step": 2104 }, { "epoch": 0.07499510127010706, "grad_norm": 1.444292939394339, "learning_rate": 1.9894151420529167e-05, "loss": 0.5277, "step": 2105 }, { "epoch": 0.07503072839660117, "grad_norm": 1.4971508608313853, "learning_rate": 1.98939839038619e-05, "loss": 0.5287, "step": 2106 }, { "epoch": 0.07506635552309529, "grad_norm": 1.4435224480421882, "learning_rate": 1.989381625544931e-05, "loss": 0.5438, "step": 2107 }, { "epoch": 0.0751019826495894, "grad_norm": 1.3804521410223025, "learning_rate": 1.9893648475293646e-05, "loss": 0.5858, "step": 2108 }, { "epoch": 0.07513760977608351, "grad_norm": 1.346751246447011, "learning_rate": 1.9893480563397135e-05, "loss": 0.5189, "step": 2109 }, { "epoch": 0.07517323690257763, "grad_norm": 1.3427005330228035, "learning_rate": 1.989331251976202e-05, "loss": 0.5433, "step": 2110 }, { "epoch": 0.07520886402907173, "grad_norm": 2.2457464285122293, "learning_rate": 1.9893144344390533e-05, "loss": 0.6501, "step": 2111 }, { "epoch": 0.07524449115556585, "grad_norm": 1.9519956782244068, "learning_rate": 1.989297603728491e-05, "loss": 0.6693, "step": 2112 }, { "epoch": 0.07528011828205997, "grad_norm": 1.481872395047167, "learning_rate": 1.9892807598447398e-05, "loss": 0.5704, "step": 2113 }, { "epoch": 0.07531574540855407, "grad_norm": 1.821612279052408, "learning_rate": 1.989263902788024e-05, "loss": 0.6012, "step": 2114 }, { "epoch": 0.07535137253504819, "grad_norm": 1.4776521075817268, "learning_rate": 1.9892470325585676e-05, "loss": 0.5514, "step": 2115 }, { "epoch": 0.0753869996615423, "grad_norm": 1.2857286204009497, "learning_rate": 1.9892301491565957e-05, "loss": 0.505, "step": 2116 }, { "epoch": 0.0754226267880364, "grad_norm": 1.442354008241517, "learning_rate": 1.989213252582333e-05, "loss": 0.5563, "step": 2117 }, { "epoch": 0.07545825391453052, "grad_norm": 1.7581432882346493, "learning_rate": 1.9891963428360043e-05, "loss": 0.6125, "step": 2118 }, { "epoch": 0.07549388104102464, "grad_norm": 1.7858857582911363, "learning_rate": 1.989179419917835e-05, "loss": 0.5104, "step": 2119 }, { "epoch": 0.07552950816751874, "grad_norm": 1.246264047565588, "learning_rate": 1.98916248382805e-05, "loss": 0.5321, "step": 2120 }, { "epoch": 0.07556513529401286, "grad_norm": 1.3975500552695446, "learning_rate": 1.9891455345668756e-05, "loss": 0.5563, "step": 2121 }, { "epoch": 0.07560076242050698, "grad_norm": 1.2256924688533197, "learning_rate": 1.9891285721345366e-05, "loss": 0.5289, "step": 2122 }, { "epoch": 0.07563638954700108, "grad_norm": 1.3730400824158933, "learning_rate": 1.98911159653126e-05, "loss": 0.5303, "step": 2123 }, { "epoch": 0.0756720166734952, "grad_norm": 2.3446258626624616, "learning_rate": 1.9890946077572702e-05, "loss": 0.6192, "step": 2124 }, { "epoch": 0.07570764379998932, "grad_norm": 1.3090550500572173, "learning_rate": 1.9890776058127948e-05, "loss": 0.5617, "step": 2125 }, { "epoch": 0.07574327092648342, "grad_norm": 2.2786336613076092, "learning_rate": 1.98906059069806e-05, "loss": 0.7058, "step": 2126 }, { "epoch": 0.07577889805297754, "grad_norm": 2.0317617302125806, "learning_rate": 1.9890435624132918e-05, "loss": 0.5858, "step": 2127 }, { "epoch": 0.07581452517947165, "grad_norm": 1.8327808925054105, "learning_rate": 1.9890265209587175e-05, "loss": 0.6202, "step": 2128 }, { "epoch": 0.07585015230596576, "grad_norm": 1.9381755003396066, "learning_rate": 1.9890094663345635e-05, "loss": 0.6725, "step": 2129 }, { "epoch": 0.07588577943245987, "grad_norm": 1.5044203571595858, "learning_rate": 1.9889923985410576e-05, "loss": 0.5082, "step": 2130 }, { "epoch": 0.07592140655895399, "grad_norm": 2.068024468464399, "learning_rate": 1.988975317578426e-05, "loss": 0.7154, "step": 2131 }, { "epoch": 0.0759570336854481, "grad_norm": 1.536278364568742, "learning_rate": 1.9889582234468973e-05, "loss": 0.5872, "step": 2132 }, { "epoch": 0.07599266081194221, "grad_norm": 1.312186456035071, "learning_rate": 1.9889411161466987e-05, "loss": 0.5328, "step": 2133 }, { "epoch": 0.07602828793843633, "grad_norm": 1.5030021402965763, "learning_rate": 1.9889239956780576e-05, "loss": 0.5097, "step": 2134 }, { "epoch": 0.07606391506493043, "grad_norm": 1.6152036408428203, "learning_rate": 1.9889068620412025e-05, "loss": 0.623, "step": 2135 }, { "epoch": 0.07609954219142455, "grad_norm": 1.3701586191901605, "learning_rate": 1.9888897152363613e-05, "loss": 0.6391, "step": 2136 }, { "epoch": 0.07613516931791867, "grad_norm": 2.04989002086961, "learning_rate": 1.988872555263762e-05, "loss": 0.6203, "step": 2137 }, { "epoch": 0.07617079644441277, "grad_norm": 1.2426934772622222, "learning_rate": 1.9888553821236338e-05, "loss": 0.5564, "step": 2138 }, { "epoch": 0.07620642357090689, "grad_norm": 1.7264996946154563, "learning_rate": 1.988838195816205e-05, "loss": 0.6151, "step": 2139 }, { "epoch": 0.076242050697401, "grad_norm": 1.3335349815896402, "learning_rate": 1.9888209963417045e-05, "loss": 0.5734, "step": 2140 }, { "epoch": 0.07627767782389511, "grad_norm": 1.4681958896222422, "learning_rate": 1.988803783700361e-05, "loss": 0.5936, "step": 2141 }, { "epoch": 0.07631330495038922, "grad_norm": 1.888654745219897, "learning_rate": 1.9887865578924042e-05, "loss": 0.6263, "step": 2142 }, { "epoch": 0.07634893207688334, "grad_norm": 2.124144183954037, "learning_rate": 1.988769318918063e-05, "loss": 0.6522, "step": 2143 }, { "epoch": 0.07638455920337746, "grad_norm": 2.421877894004826, "learning_rate": 1.9887520667775677e-05, "loss": 0.6077, "step": 2144 }, { "epoch": 0.07642018632987156, "grad_norm": 1.3041031404691277, "learning_rate": 1.9887348014711475e-05, "loss": 0.5443, "step": 2145 }, { "epoch": 0.07645581345636568, "grad_norm": 1.356223333771857, "learning_rate": 1.988717522999032e-05, "loss": 0.5536, "step": 2146 }, { "epoch": 0.0764914405828598, "grad_norm": 1.2213561062830822, "learning_rate": 1.9887002313614518e-05, "loss": 0.5408, "step": 2147 }, { "epoch": 0.0765270677093539, "grad_norm": 1.7972370112276983, "learning_rate": 1.9886829265586368e-05, "loss": 0.6014, "step": 2148 }, { "epoch": 0.07656269483584802, "grad_norm": 1.908365618094179, "learning_rate": 1.9886656085908176e-05, "loss": 0.6573, "step": 2149 }, { "epoch": 0.07659832196234213, "grad_norm": 2.5404225409475663, "learning_rate": 1.988648277458225e-05, "loss": 0.6333, "step": 2150 }, { "epoch": 0.07663394908883624, "grad_norm": 1.4164640050928117, "learning_rate": 1.9886309331610893e-05, "loss": 0.6029, "step": 2151 }, { "epoch": 0.07666957621533035, "grad_norm": 1.5274912978343504, "learning_rate": 1.988613575699642e-05, "loss": 0.5556, "step": 2152 }, { "epoch": 0.07670520334182447, "grad_norm": 1.9280215879663285, "learning_rate": 1.9885962050741138e-05, "loss": 0.6282, "step": 2153 }, { "epoch": 0.07674083046831857, "grad_norm": 1.2590303434854266, "learning_rate": 1.9885788212847358e-05, "loss": 0.5138, "step": 2154 }, { "epoch": 0.07677645759481269, "grad_norm": 1.419227383824992, "learning_rate": 1.98856142433174e-05, "loss": 0.53, "step": 2155 }, { "epoch": 0.07681208472130681, "grad_norm": 1.3767146498429255, "learning_rate": 1.988544014215358e-05, "loss": 0.5885, "step": 2156 }, { "epoch": 0.07684771184780091, "grad_norm": 1.995290453216958, "learning_rate": 1.9885265909358216e-05, "loss": 0.6092, "step": 2157 }, { "epoch": 0.07688333897429503, "grad_norm": 1.3574648870777315, "learning_rate": 1.9885091544933628e-05, "loss": 0.5484, "step": 2158 }, { "epoch": 0.07691896610078915, "grad_norm": 2.2167545112288236, "learning_rate": 1.9884917048882132e-05, "loss": 0.6405, "step": 2159 }, { "epoch": 0.07695459322728325, "grad_norm": 2.086441043623901, "learning_rate": 1.988474242120606e-05, "loss": 0.6671, "step": 2160 }, { "epoch": 0.07699022035377737, "grad_norm": 2.445449902345114, "learning_rate": 1.9884567661907733e-05, "loss": 0.6133, "step": 2161 }, { "epoch": 0.07702584748027148, "grad_norm": 1.4394157186611434, "learning_rate": 1.988439277098948e-05, "loss": 0.5624, "step": 2162 }, { "epoch": 0.07706147460676559, "grad_norm": 1.4091286276820167, "learning_rate": 1.9884217748453625e-05, "loss": 0.542, "step": 2163 }, { "epoch": 0.0770971017332597, "grad_norm": 2.1514721187165033, "learning_rate": 1.9884042594302503e-05, "loss": 0.6494, "step": 2164 }, { "epoch": 0.07713272885975382, "grad_norm": 2.815586812886685, "learning_rate": 1.988386730853845e-05, "loss": 0.6128, "step": 2165 }, { "epoch": 0.07716835598624792, "grad_norm": 1.3564317703306794, "learning_rate": 1.988369189116379e-05, "loss": 0.5328, "step": 2166 }, { "epoch": 0.07720398311274204, "grad_norm": 2.396325432085148, "learning_rate": 1.9883516342180867e-05, "loss": 0.5943, "step": 2167 }, { "epoch": 0.07723961023923616, "grad_norm": 1.9266995006911176, "learning_rate": 1.9883340661592013e-05, "loss": 0.6639, "step": 2168 }, { "epoch": 0.07727523736573026, "grad_norm": 1.8398284498466209, "learning_rate": 1.9883164849399572e-05, "loss": 0.5801, "step": 2169 }, { "epoch": 0.07731086449222438, "grad_norm": 2.0096881735278473, "learning_rate": 1.9882988905605885e-05, "loss": 0.6197, "step": 2170 }, { "epoch": 0.0773464916187185, "grad_norm": 1.4070626293156934, "learning_rate": 1.988281283021329e-05, "loss": 0.5663, "step": 2171 }, { "epoch": 0.0773821187452126, "grad_norm": 1.4091987472237208, "learning_rate": 1.9882636623224135e-05, "loss": 0.5308, "step": 2172 }, { "epoch": 0.07741774587170672, "grad_norm": 1.2794049416737152, "learning_rate": 1.9882460284640762e-05, "loss": 0.5088, "step": 2173 }, { "epoch": 0.07745337299820083, "grad_norm": 2.1927693991855413, "learning_rate": 1.988228381446553e-05, "loss": 0.6312, "step": 2174 }, { "epoch": 0.07748900012469494, "grad_norm": 1.431762366701626, "learning_rate": 1.9882107212700778e-05, "loss": 0.535, "step": 2175 }, { "epoch": 0.07752462725118905, "grad_norm": 1.3133794918420216, "learning_rate": 1.9881930479348863e-05, "loss": 0.5239, "step": 2176 }, { "epoch": 0.07756025437768317, "grad_norm": 1.1830035999621187, "learning_rate": 1.9881753614412135e-05, "loss": 0.5039, "step": 2177 }, { "epoch": 0.07759588150417727, "grad_norm": 1.363055915641001, "learning_rate": 1.988157661789295e-05, "loss": 0.4922, "step": 2178 }, { "epoch": 0.07763150863067139, "grad_norm": 2.238808792706384, "learning_rate": 1.9881399489793667e-05, "loss": 0.6612, "step": 2179 }, { "epoch": 0.07766713575716551, "grad_norm": 1.2229216170794044, "learning_rate": 1.9881222230116645e-05, "loss": 0.5457, "step": 2180 }, { "epoch": 0.07770276288365961, "grad_norm": 1.398254672495796, "learning_rate": 1.9881044838864238e-05, "loss": 0.5305, "step": 2181 }, { "epoch": 0.07773839001015373, "grad_norm": 1.8837321243163037, "learning_rate": 1.9880867316038815e-05, "loss": 0.5682, "step": 2182 }, { "epoch": 0.07777401713664785, "grad_norm": 1.3976824569174269, "learning_rate": 1.988068966164274e-05, "loss": 0.5423, "step": 2183 }, { "epoch": 0.07780964426314195, "grad_norm": 2.0681617034387707, "learning_rate": 1.988051187567837e-05, "loss": 0.6757, "step": 2184 }, { "epoch": 0.07784527138963607, "grad_norm": 2.1024064848709045, "learning_rate": 1.9880333958148085e-05, "loss": 0.563, "step": 2185 }, { "epoch": 0.07788089851613018, "grad_norm": 1.3729471043420658, "learning_rate": 1.988015590905424e-05, "loss": 0.5478, "step": 2186 }, { "epoch": 0.07791652564262429, "grad_norm": 1.539033487856092, "learning_rate": 1.9879977728399217e-05, "loss": 0.572, "step": 2187 }, { "epoch": 0.0779521527691184, "grad_norm": 1.4000595732079253, "learning_rate": 1.9879799416185386e-05, "loss": 0.571, "step": 2188 }, { "epoch": 0.07798777989561252, "grad_norm": 1.6458869764093476, "learning_rate": 1.9879620972415118e-05, "loss": 0.6102, "step": 2189 }, { "epoch": 0.07802340702210664, "grad_norm": 2.4951346525726295, "learning_rate": 1.9879442397090797e-05, "loss": 0.627, "step": 2190 }, { "epoch": 0.07805903414860074, "grad_norm": 1.2318853858404886, "learning_rate": 1.987926369021479e-05, "loss": 0.5435, "step": 2191 }, { "epoch": 0.07809466127509486, "grad_norm": 1.3577166508201717, "learning_rate": 1.9879084851789478e-05, "loss": 0.5491, "step": 2192 }, { "epoch": 0.07813028840158898, "grad_norm": 1.2624096323920149, "learning_rate": 1.9878905881817254e-05, "loss": 0.5381, "step": 2193 }, { "epoch": 0.07816591552808308, "grad_norm": 1.6448689025451142, "learning_rate": 1.987872678030049e-05, "loss": 0.5704, "step": 2194 }, { "epoch": 0.0782015426545772, "grad_norm": 1.400196739796017, "learning_rate": 1.987854754724157e-05, "loss": 0.5895, "step": 2195 }, { "epoch": 0.07823716978107131, "grad_norm": 1.9897459700229465, "learning_rate": 1.987836818264289e-05, "loss": 0.5396, "step": 2196 }, { "epoch": 0.07827279690756542, "grad_norm": 1.2098689600743715, "learning_rate": 1.9878188686506834e-05, "loss": 0.5137, "step": 2197 }, { "epoch": 0.07830842403405953, "grad_norm": 1.428421524864691, "learning_rate": 1.987800905883579e-05, "loss": 0.5613, "step": 2198 }, { "epoch": 0.07834405116055365, "grad_norm": 1.348122032732944, "learning_rate": 1.9877829299632146e-05, "loss": 0.5406, "step": 2199 }, { "epoch": 0.07837967828704775, "grad_norm": 1.9487504185953, "learning_rate": 1.9877649408898306e-05, "loss": 0.5977, "step": 2200 }, { "epoch": 0.07841530541354187, "grad_norm": 1.4268825036608754, "learning_rate": 1.9877469386636658e-05, "loss": 0.5157, "step": 2201 }, { "epoch": 0.07845093254003599, "grad_norm": 4.2602244903553, "learning_rate": 1.9877289232849605e-05, "loss": 0.7133, "step": 2202 }, { "epoch": 0.07848655966653009, "grad_norm": 1.7430356271271945, "learning_rate": 1.9877108947539537e-05, "loss": 0.6752, "step": 2203 }, { "epoch": 0.07852218679302421, "grad_norm": 1.9799114989554576, "learning_rate": 1.9876928530708862e-05, "loss": 0.63, "step": 2204 }, { "epoch": 0.07855781391951833, "grad_norm": 2.0522241702964457, "learning_rate": 1.987674798235998e-05, "loss": 0.6436, "step": 2205 }, { "epoch": 0.07859344104601243, "grad_norm": 1.4061953850619082, "learning_rate": 1.9876567302495295e-05, "loss": 0.5587, "step": 2206 }, { "epoch": 0.07862906817250655, "grad_norm": 2.1522486521662887, "learning_rate": 1.987638649111721e-05, "loss": 0.5433, "step": 2207 }, { "epoch": 0.07866469529900066, "grad_norm": 1.3907325950876597, "learning_rate": 1.987620554822814e-05, "loss": 0.5349, "step": 2208 }, { "epoch": 0.07870032242549477, "grad_norm": 1.5227543161585275, "learning_rate": 1.9876024473830488e-05, "loss": 0.5906, "step": 2209 }, { "epoch": 0.07873594955198888, "grad_norm": 1.3634788452691016, "learning_rate": 1.987584326792667e-05, "loss": 0.5164, "step": 2210 }, { "epoch": 0.078771576678483, "grad_norm": 1.9980591442881404, "learning_rate": 1.9875661930519093e-05, "loss": 0.5156, "step": 2211 }, { "epoch": 0.0788072038049771, "grad_norm": 1.2130118333966515, "learning_rate": 1.9875480461610176e-05, "loss": 0.5352, "step": 2212 }, { "epoch": 0.07884283093147122, "grad_norm": 2.070148614134971, "learning_rate": 1.9875298861202332e-05, "loss": 0.6397, "step": 2213 }, { "epoch": 0.07887845805796534, "grad_norm": 1.1934973218876186, "learning_rate": 1.9875117129297986e-05, "loss": 0.5821, "step": 2214 }, { "epoch": 0.07891408518445944, "grad_norm": 2.005453148734325, "learning_rate": 1.987493526589955e-05, "loss": 0.6728, "step": 2215 }, { "epoch": 0.07894971231095356, "grad_norm": 3.8278647274312885, "learning_rate": 1.9874753271009445e-05, "loss": 0.6996, "step": 2216 }, { "epoch": 0.07898533943744768, "grad_norm": 1.6937294041811484, "learning_rate": 1.9874571144630105e-05, "loss": 0.5443, "step": 2217 }, { "epoch": 0.07902096656394178, "grad_norm": 1.4697965848959376, "learning_rate": 1.9874388886763944e-05, "loss": 0.5414, "step": 2218 }, { "epoch": 0.0790565936904359, "grad_norm": 1.3121579934645418, "learning_rate": 1.9874206497413397e-05, "loss": 0.5315, "step": 2219 }, { "epoch": 0.07909222081693001, "grad_norm": 1.3648522018112357, "learning_rate": 1.9874023976580886e-05, "loss": 0.5108, "step": 2220 }, { "epoch": 0.07912784794342412, "grad_norm": 1.372577722805055, "learning_rate": 1.9873841324268842e-05, "loss": 0.5352, "step": 2221 }, { "epoch": 0.07916347506991823, "grad_norm": 1.744482205425277, "learning_rate": 1.9873658540479703e-05, "loss": 0.6045, "step": 2222 }, { "epoch": 0.07919910219641235, "grad_norm": 1.3710128573537985, "learning_rate": 1.98734756252159e-05, "loss": 0.5316, "step": 2223 }, { "epoch": 0.07923472932290646, "grad_norm": 1.3784193009459407, "learning_rate": 1.9873292578479867e-05, "loss": 0.5914, "step": 2224 }, { "epoch": 0.07927035644940057, "grad_norm": 1.3621834143159308, "learning_rate": 1.987310940027404e-05, "loss": 0.4981, "step": 2225 }, { "epoch": 0.07930598357589469, "grad_norm": 2.1995180511763976, "learning_rate": 1.9872926090600862e-05, "loss": 0.6612, "step": 2226 }, { "epoch": 0.07934161070238879, "grad_norm": 1.3136711089025603, "learning_rate": 1.9872742649462772e-05, "loss": 0.5458, "step": 2227 }, { "epoch": 0.07937723782888291, "grad_norm": 1.7415807367575113, "learning_rate": 1.9872559076862214e-05, "loss": 0.7043, "step": 2228 }, { "epoch": 0.07941286495537703, "grad_norm": 1.9089053581473336, "learning_rate": 1.9872375372801627e-05, "loss": 0.6158, "step": 2229 }, { "epoch": 0.07944849208187113, "grad_norm": 1.305818018627099, "learning_rate": 1.9872191537283466e-05, "loss": 0.4995, "step": 2230 }, { "epoch": 0.07948411920836525, "grad_norm": 2.278705258949364, "learning_rate": 1.9872007570310175e-05, "loss": 0.6201, "step": 2231 }, { "epoch": 0.07951974633485936, "grad_norm": 1.967337872319488, "learning_rate": 1.98718234718842e-05, "loss": 0.6102, "step": 2232 }, { "epoch": 0.07955537346135347, "grad_norm": 2.3700482891251693, "learning_rate": 1.9871639242007997e-05, "loss": 0.6467, "step": 2233 }, { "epoch": 0.07959100058784759, "grad_norm": 1.8530670771989066, "learning_rate": 1.9871454880684016e-05, "loss": 0.6138, "step": 2234 }, { "epoch": 0.0796266277143417, "grad_norm": 1.4128343364225036, "learning_rate": 1.987127038791472e-05, "loss": 0.5912, "step": 2235 }, { "epoch": 0.0796622548408358, "grad_norm": 1.804108941933111, "learning_rate": 1.9871085763702553e-05, "loss": 0.5784, "step": 2236 }, { "epoch": 0.07969788196732992, "grad_norm": 1.9905799205266204, "learning_rate": 1.987090100804998e-05, "loss": 0.6795, "step": 2237 }, { "epoch": 0.07973350909382404, "grad_norm": 2.0538208554547324, "learning_rate": 1.9870716120959462e-05, "loss": 0.5581, "step": 2238 }, { "epoch": 0.07976913622031816, "grad_norm": 1.456408889797915, "learning_rate": 1.987053110243346e-05, "loss": 0.6047, "step": 2239 }, { "epoch": 0.07980476334681226, "grad_norm": 2.0026062625877104, "learning_rate": 1.9870345952474436e-05, "loss": 0.6576, "step": 2240 }, { "epoch": 0.07984039047330638, "grad_norm": 1.7125503054909974, "learning_rate": 1.9870160671084857e-05, "loss": 0.6724, "step": 2241 }, { "epoch": 0.0798760175998005, "grad_norm": 1.947952273982146, "learning_rate": 1.9869975258267192e-05, "loss": 0.6748, "step": 2242 }, { "epoch": 0.0799116447262946, "grad_norm": 1.2447187652937752, "learning_rate": 1.9869789714023906e-05, "loss": 0.4983, "step": 2243 }, { "epoch": 0.07994727185278871, "grad_norm": 1.2246771567307897, "learning_rate": 1.986960403835747e-05, "loss": 0.5188, "step": 2244 }, { "epoch": 0.07998289897928283, "grad_norm": 2.01261907122453, "learning_rate": 1.9869418231270358e-05, "loss": 0.6081, "step": 2245 }, { "epoch": 0.08001852610577694, "grad_norm": 1.9384008055463018, "learning_rate": 1.9869232292765046e-05, "loss": 0.5347, "step": 2246 }, { "epoch": 0.08005415323227105, "grad_norm": 1.422955992311979, "learning_rate": 1.986904622284401e-05, "loss": 0.5735, "step": 2247 }, { "epoch": 0.08008978035876517, "grad_norm": 2.002298085704396, "learning_rate": 1.986886002150972e-05, "loss": 0.6712, "step": 2248 }, { "epoch": 0.08012540748525927, "grad_norm": 2.5800342405643693, "learning_rate": 1.9868673688764663e-05, "loss": 0.6361, "step": 2249 }, { "epoch": 0.08016103461175339, "grad_norm": 1.4917079571988736, "learning_rate": 1.986848722461132e-05, "loss": 0.5614, "step": 2250 }, { "epoch": 0.08019666173824751, "grad_norm": 1.8636070066236445, "learning_rate": 1.986830062905217e-05, "loss": 0.6603, "step": 2251 }, { "epoch": 0.08023228886474161, "grad_norm": 2.5152731386160143, "learning_rate": 1.9868113902089702e-05, "loss": 0.6597, "step": 2252 }, { "epoch": 0.08026791599123573, "grad_norm": 1.4683544324655646, "learning_rate": 1.9867927043726393e-05, "loss": 0.5965, "step": 2253 }, { "epoch": 0.08030354311772984, "grad_norm": 1.7978267266314207, "learning_rate": 1.9867740053964746e-05, "loss": 0.6234, "step": 2254 }, { "epoch": 0.08033917024422395, "grad_norm": 1.3292189713843865, "learning_rate": 1.986755293280724e-05, "loss": 0.5464, "step": 2255 }, { "epoch": 0.08037479737071807, "grad_norm": 1.7756070254073943, "learning_rate": 1.986736568025637e-05, "loss": 0.5619, "step": 2256 }, { "epoch": 0.08041042449721218, "grad_norm": 1.3940549863491374, "learning_rate": 1.9867178296314627e-05, "loss": 0.5624, "step": 2257 }, { "epoch": 0.08044605162370629, "grad_norm": 1.3283814027300787, "learning_rate": 1.986699078098451e-05, "loss": 0.4671, "step": 2258 }, { "epoch": 0.0804816787502004, "grad_norm": 1.6253475848850651, "learning_rate": 1.9866803134268514e-05, "loss": 0.6036, "step": 2259 }, { "epoch": 0.08051730587669452, "grad_norm": 2.2740003776685604, "learning_rate": 1.986661535616914e-05, "loss": 0.5936, "step": 2260 }, { "epoch": 0.08055293300318862, "grad_norm": 1.4185583963944586, "learning_rate": 1.9866427446688883e-05, "loss": 0.5369, "step": 2261 }, { "epoch": 0.08058856012968274, "grad_norm": 1.8050830226288623, "learning_rate": 1.986623940583025e-05, "loss": 0.6113, "step": 2262 }, { "epoch": 0.08062418725617686, "grad_norm": 1.4233236671428782, "learning_rate": 1.986605123359574e-05, "loss": 0.5659, "step": 2263 }, { "epoch": 0.08065981438267096, "grad_norm": 1.339263894075227, "learning_rate": 1.9865862929987866e-05, "loss": 0.5856, "step": 2264 }, { "epoch": 0.08069544150916508, "grad_norm": 1.370556547529804, "learning_rate": 1.986567449500913e-05, "loss": 0.5545, "step": 2265 }, { "epoch": 0.0807310686356592, "grad_norm": 2.018179371569776, "learning_rate": 1.986548592866204e-05, "loss": 0.656, "step": 2266 }, { "epoch": 0.0807666957621533, "grad_norm": 2.023799596313918, "learning_rate": 1.9865297230949113e-05, "loss": 0.6171, "step": 2267 }, { "epoch": 0.08080232288864742, "grad_norm": 1.8290297476726611, "learning_rate": 1.9865108401872856e-05, "loss": 0.6988, "step": 2268 }, { "epoch": 0.08083795001514153, "grad_norm": 2.1365878159005365, "learning_rate": 1.986491944143579e-05, "loss": 0.6371, "step": 2269 }, { "epoch": 0.08087357714163564, "grad_norm": 1.9337603024594627, "learning_rate": 1.986473034964042e-05, "loss": 0.5414, "step": 2270 }, { "epoch": 0.08090920426812975, "grad_norm": 1.356339584289113, "learning_rate": 1.9864541126489275e-05, "loss": 0.556, "step": 2271 }, { "epoch": 0.08094483139462387, "grad_norm": 1.9637873138953148, "learning_rate": 1.9864351771984864e-05, "loss": 0.706, "step": 2272 }, { "epoch": 0.08098045852111797, "grad_norm": 2.5999402391763806, "learning_rate": 1.986416228612972e-05, "loss": 0.6515, "step": 2273 }, { "epoch": 0.08101608564761209, "grad_norm": 1.4592619673093228, "learning_rate": 1.986397266892636e-05, "loss": 0.6046, "step": 2274 }, { "epoch": 0.08105171277410621, "grad_norm": 1.3501544047412728, "learning_rate": 1.9863782920377305e-05, "loss": 0.5509, "step": 2275 }, { "epoch": 0.08108733990060031, "grad_norm": 1.3632181657694216, "learning_rate": 1.986359304048509e-05, "loss": 0.55, "step": 2276 }, { "epoch": 0.08112296702709443, "grad_norm": 1.7348419356723945, "learning_rate": 1.986340302925224e-05, "loss": 0.6419, "step": 2277 }, { "epoch": 0.08115859415358855, "grad_norm": 1.2687053845913425, "learning_rate": 1.986321288668128e-05, "loss": 0.4648, "step": 2278 }, { "epoch": 0.08119422128008265, "grad_norm": 1.286717353037396, "learning_rate": 1.986302261277475e-05, "loss": 0.4956, "step": 2279 }, { "epoch": 0.08122984840657677, "grad_norm": 2.110523301515872, "learning_rate": 1.986283220753518e-05, "loss": 0.6407, "step": 2280 }, { "epoch": 0.08126547553307088, "grad_norm": 1.3649356665992547, "learning_rate": 1.9862641670965105e-05, "loss": 0.5737, "step": 2281 }, { "epoch": 0.08130110265956499, "grad_norm": 2.0406888988769674, "learning_rate": 1.9862451003067063e-05, "loss": 0.6333, "step": 2282 }, { "epoch": 0.0813367297860591, "grad_norm": 1.4010450302586135, "learning_rate": 1.986226020384359e-05, "loss": 0.6061, "step": 2283 }, { "epoch": 0.08137235691255322, "grad_norm": 1.271489103405085, "learning_rate": 1.9862069273297233e-05, "loss": 0.5239, "step": 2284 }, { "epoch": 0.08140798403904734, "grad_norm": 1.9448029669501172, "learning_rate": 1.986187821143053e-05, "loss": 0.6841, "step": 2285 }, { "epoch": 0.08144361116554144, "grad_norm": 1.3460699428534586, "learning_rate": 1.9861687018246022e-05, "loss": 0.532, "step": 2286 }, { "epoch": 0.08147923829203556, "grad_norm": 3.2465666598834924, "learning_rate": 1.986149569374626e-05, "loss": 0.5458, "step": 2287 }, { "epoch": 0.08151486541852968, "grad_norm": 2.322322136612061, "learning_rate": 1.986130423793379e-05, "loss": 0.6403, "step": 2288 }, { "epoch": 0.08155049254502378, "grad_norm": 1.3050579517079823, "learning_rate": 1.9861112650811166e-05, "loss": 0.5125, "step": 2289 }, { "epoch": 0.0815861196715179, "grad_norm": 1.7953254085494077, "learning_rate": 1.986092093238093e-05, "loss": 0.5721, "step": 2290 }, { "epoch": 0.08162174679801201, "grad_norm": 1.3789258225647627, "learning_rate": 1.986072908264564e-05, "loss": 0.5194, "step": 2291 }, { "epoch": 0.08165737392450612, "grad_norm": 1.6675315829484365, "learning_rate": 1.9860537101607848e-05, "loss": 0.6306, "step": 2292 }, { "epoch": 0.08169300105100023, "grad_norm": 1.4031975674864827, "learning_rate": 1.9860344989270115e-05, "loss": 0.6189, "step": 2293 }, { "epoch": 0.08172862817749435, "grad_norm": 1.3804389502977243, "learning_rate": 1.9860152745634994e-05, "loss": 0.5914, "step": 2294 }, { "epoch": 0.08176425530398845, "grad_norm": 1.6097572744372834, "learning_rate": 1.985996037070505e-05, "loss": 0.6725, "step": 2295 }, { "epoch": 0.08179988243048257, "grad_norm": 1.8174357426866838, "learning_rate": 1.985976786448284e-05, "loss": 0.6251, "step": 2296 }, { "epoch": 0.08183550955697669, "grad_norm": 2.2030254003951515, "learning_rate": 1.9859575226970932e-05, "loss": 0.6631, "step": 2297 }, { "epoch": 0.08187113668347079, "grad_norm": 1.861840143850897, "learning_rate": 1.9859382458171887e-05, "loss": 0.6344, "step": 2298 }, { "epoch": 0.08190676380996491, "grad_norm": 1.5231402096330884, "learning_rate": 1.9859189558088275e-05, "loss": 0.5324, "step": 2299 }, { "epoch": 0.08194239093645903, "grad_norm": 1.9437047767007718, "learning_rate": 1.985899652672266e-05, "loss": 0.6587, "step": 2300 }, { "epoch": 0.08197801806295313, "grad_norm": 2.106893727538784, "learning_rate": 1.9858803364077616e-05, "loss": 0.6371, "step": 2301 }, { "epoch": 0.08201364518944725, "grad_norm": 1.376560769618341, "learning_rate": 1.9858610070155715e-05, "loss": 0.4922, "step": 2302 }, { "epoch": 0.08204927231594136, "grad_norm": 1.756566577797297, "learning_rate": 1.985841664495953e-05, "loss": 0.6412, "step": 2303 }, { "epoch": 0.08208489944243547, "grad_norm": 2.2178158532826213, "learning_rate": 1.9858223088491635e-05, "loss": 0.6734, "step": 2304 }, { "epoch": 0.08212052656892958, "grad_norm": 2.248107654275402, "learning_rate": 1.9858029400754613e-05, "loss": 0.5657, "step": 2305 }, { "epoch": 0.0821561536954237, "grad_norm": 1.3574961649999673, "learning_rate": 1.9857835581751038e-05, "loss": 0.5306, "step": 2306 }, { "epoch": 0.0821917808219178, "grad_norm": 1.8824132367191457, "learning_rate": 1.985764163148349e-05, "loss": 0.6195, "step": 2307 }, { "epoch": 0.08222740794841192, "grad_norm": 1.2960459888450526, "learning_rate": 1.985744754995455e-05, "loss": 0.5505, "step": 2308 }, { "epoch": 0.08226303507490604, "grad_norm": 1.229381581188702, "learning_rate": 1.9857253337166813e-05, "loss": 0.5228, "step": 2309 }, { "epoch": 0.08229866220140014, "grad_norm": 1.9807622417354596, "learning_rate": 1.9857058993122856e-05, "loss": 0.6043, "step": 2310 }, { "epoch": 0.08233428932789426, "grad_norm": 1.6988868005693152, "learning_rate": 1.985686451782527e-05, "loss": 0.6053, "step": 2311 }, { "epoch": 0.08236991645438838, "grad_norm": 1.8007810270286708, "learning_rate": 1.9856669911276637e-05, "loss": 0.6977, "step": 2312 }, { "epoch": 0.08240554358088248, "grad_norm": 1.975498119734629, "learning_rate": 1.9856475173479565e-05, "loss": 0.6255, "step": 2313 }, { "epoch": 0.0824411707073766, "grad_norm": 1.5111272828406606, "learning_rate": 1.9856280304436628e-05, "loss": 0.6142, "step": 2314 }, { "epoch": 0.08247679783387071, "grad_norm": 1.3492204046516556, "learning_rate": 1.9856085304150433e-05, "loss": 0.5387, "step": 2315 }, { "epoch": 0.08251242496036482, "grad_norm": 2.254689935606166, "learning_rate": 1.9855890172623573e-05, "loss": 0.595, "step": 2316 }, { "epoch": 0.08254805208685893, "grad_norm": 1.3817653784082051, "learning_rate": 1.9855694909858645e-05, "loss": 0.5381, "step": 2317 }, { "epoch": 0.08258367921335305, "grad_norm": 2.145421513271894, "learning_rate": 1.985549951585825e-05, "loss": 0.6468, "step": 2318 }, { "epoch": 0.08261930633984715, "grad_norm": 1.3008946611339067, "learning_rate": 1.9855303990624996e-05, "loss": 0.4959, "step": 2319 }, { "epoch": 0.08265493346634127, "grad_norm": 1.3259624811939776, "learning_rate": 1.9855108334161478e-05, "loss": 0.5198, "step": 2320 }, { "epoch": 0.08269056059283539, "grad_norm": 2.526555515896046, "learning_rate": 1.9854912546470304e-05, "loss": 0.6996, "step": 2321 }, { "epoch": 0.08272618771932949, "grad_norm": 1.5173703658639988, "learning_rate": 1.985471662755408e-05, "loss": 0.5371, "step": 2322 }, { "epoch": 0.08276181484582361, "grad_norm": 1.3397725428562053, "learning_rate": 1.9854520577415414e-05, "loss": 0.6176, "step": 2323 }, { "epoch": 0.08279744197231773, "grad_norm": 1.3822402972301855, "learning_rate": 1.9854324396056922e-05, "loss": 0.5176, "step": 2324 }, { "epoch": 0.08283306909881183, "grad_norm": 1.8057686404356281, "learning_rate": 1.9854128083481216e-05, "loss": 0.6578, "step": 2325 }, { "epoch": 0.08286869622530595, "grad_norm": 1.337831058446699, "learning_rate": 1.98539316396909e-05, "loss": 0.5088, "step": 2326 }, { "epoch": 0.08290432335180006, "grad_norm": 1.32677319527056, "learning_rate": 1.9853735064688603e-05, "loss": 0.5677, "step": 2327 }, { "epoch": 0.08293995047829417, "grad_norm": 1.3266194861793665, "learning_rate": 1.9853538358476933e-05, "loss": 0.4849, "step": 2328 }, { "epoch": 0.08297557760478828, "grad_norm": 1.2714681890220905, "learning_rate": 1.985334152105851e-05, "loss": 0.5392, "step": 2329 }, { "epoch": 0.0830112047312824, "grad_norm": 1.257058564536769, "learning_rate": 1.985314455243596e-05, "loss": 0.5306, "step": 2330 }, { "epoch": 0.0830468318577765, "grad_norm": 1.449727242235049, "learning_rate": 1.9852947452611906e-05, "loss": 0.5351, "step": 2331 }, { "epoch": 0.08308245898427062, "grad_norm": 1.2616838947378644, "learning_rate": 1.9852750221588966e-05, "loss": 0.5282, "step": 2332 }, { "epoch": 0.08311808611076474, "grad_norm": 1.9335945884845998, "learning_rate": 1.9852552859369774e-05, "loss": 0.5733, "step": 2333 }, { "epoch": 0.08315371323725886, "grad_norm": 1.3858425161374357, "learning_rate": 1.985235536595695e-05, "loss": 0.599, "step": 2334 }, { "epoch": 0.08318934036375296, "grad_norm": 1.331371906333134, "learning_rate": 1.985215774135313e-05, "loss": 0.5794, "step": 2335 }, { "epoch": 0.08322496749024708, "grad_norm": 1.3509133502883583, "learning_rate": 1.9851959985560944e-05, "loss": 0.5287, "step": 2336 }, { "epoch": 0.0832605946167412, "grad_norm": 2.053984661767953, "learning_rate": 1.9851762098583025e-05, "loss": 0.6191, "step": 2337 }, { "epoch": 0.0832962217432353, "grad_norm": 2.3052138281315036, "learning_rate": 1.9851564080422007e-05, "loss": 0.5984, "step": 2338 }, { "epoch": 0.08333184886972941, "grad_norm": 2.624302219832074, "learning_rate": 1.9851365931080525e-05, "loss": 0.6394, "step": 2339 }, { "epoch": 0.08336747599622353, "grad_norm": 1.355288631735426, "learning_rate": 1.9851167650561223e-05, "loss": 0.5414, "step": 2340 }, { "epoch": 0.08340310312271763, "grad_norm": 2.4783165552092887, "learning_rate": 1.985096923886674e-05, "loss": 0.7037, "step": 2341 }, { "epoch": 0.08343873024921175, "grad_norm": 1.2102270808783497, "learning_rate": 1.985077069599971e-05, "loss": 0.5099, "step": 2342 }, { "epoch": 0.08347435737570587, "grad_norm": 1.3937901458472473, "learning_rate": 1.9850572021962788e-05, "loss": 0.5371, "step": 2343 }, { "epoch": 0.08350998450219997, "grad_norm": 1.2749023924234855, "learning_rate": 1.9850373216758613e-05, "loss": 0.4775, "step": 2344 }, { "epoch": 0.08354561162869409, "grad_norm": 1.6755460822218156, "learning_rate": 1.985017428038983e-05, "loss": 0.642, "step": 2345 }, { "epoch": 0.0835812387551882, "grad_norm": 1.7750725327251022, "learning_rate": 1.9849975212859095e-05, "loss": 0.6403, "step": 2346 }, { "epoch": 0.08361686588168231, "grad_norm": 2.733770490172308, "learning_rate": 1.9849776014169054e-05, "loss": 0.6411, "step": 2347 }, { "epoch": 0.08365249300817643, "grad_norm": 1.3764129079778458, "learning_rate": 1.9849576684322365e-05, "loss": 0.529, "step": 2348 }, { "epoch": 0.08368812013467054, "grad_norm": 1.339461019144313, "learning_rate": 1.9849377223321674e-05, "loss": 0.5492, "step": 2349 }, { "epoch": 0.08372374726116465, "grad_norm": 1.7949932786045435, "learning_rate": 1.9849177631169643e-05, "loss": 0.6458, "step": 2350 }, { "epoch": 0.08375937438765876, "grad_norm": 2.429913621137589, "learning_rate": 1.9848977907868927e-05, "loss": 0.5974, "step": 2351 }, { "epoch": 0.08379500151415288, "grad_norm": 2.2365634653939925, "learning_rate": 1.9848778053422184e-05, "loss": 0.6733, "step": 2352 }, { "epoch": 0.08383062864064698, "grad_norm": 1.3259854166700937, "learning_rate": 1.984857806783208e-05, "loss": 0.4949, "step": 2353 }, { "epoch": 0.0838662557671411, "grad_norm": 1.952157087138693, "learning_rate": 1.9848377951101274e-05, "loss": 0.5964, "step": 2354 }, { "epoch": 0.08390188289363522, "grad_norm": 2.240443277171897, "learning_rate": 1.9848177703232433e-05, "loss": 0.7388, "step": 2355 }, { "epoch": 0.08393751002012932, "grad_norm": 1.4541751166741774, "learning_rate": 1.984797732422822e-05, "loss": 0.5676, "step": 2356 }, { "epoch": 0.08397313714662344, "grad_norm": 1.4482789636795783, "learning_rate": 1.984777681409131e-05, "loss": 0.5344, "step": 2357 }, { "epoch": 0.08400876427311756, "grad_norm": 1.3700746107135573, "learning_rate": 1.9847576172824365e-05, "loss": 0.5299, "step": 2358 }, { "epoch": 0.08404439139961166, "grad_norm": 1.3180107099722256, "learning_rate": 1.984737540043006e-05, "loss": 0.5327, "step": 2359 }, { "epoch": 0.08408001852610578, "grad_norm": 1.7318098275237748, "learning_rate": 1.984717449691107e-05, "loss": 0.6158, "step": 2360 }, { "epoch": 0.0841156456525999, "grad_norm": 1.4724644088497811, "learning_rate": 1.984697346227007e-05, "loss": 0.5206, "step": 2361 }, { "epoch": 0.084151272779094, "grad_norm": 1.3239153614958208, "learning_rate": 1.9846772296509736e-05, "loss": 0.5148, "step": 2362 }, { "epoch": 0.08418689990558811, "grad_norm": 1.4181336427864453, "learning_rate": 1.984657099963274e-05, "loss": 0.5475, "step": 2363 }, { "epoch": 0.08422252703208223, "grad_norm": 1.8414597085857263, "learning_rate": 1.9846369571641776e-05, "loss": 0.5968, "step": 2364 }, { "epoch": 0.08425815415857633, "grad_norm": 1.3049398187885892, "learning_rate": 1.984616801253952e-05, "loss": 0.5094, "step": 2365 }, { "epoch": 0.08429378128507045, "grad_norm": 2.4341152819025234, "learning_rate": 1.9845966322328647e-05, "loss": 0.646, "step": 2366 }, { "epoch": 0.08432940841156457, "grad_norm": 2.1524488494395997, "learning_rate": 1.9845764501011854e-05, "loss": 0.563, "step": 2367 }, { "epoch": 0.08436503553805867, "grad_norm": 1.4998045209219508, "learning_rate": 1.9845562548591826e-05, "loss": 0.5113, "step": 2368 }, { "epoch": 0.08440066266455279, "grad_norm": 2.1120649573921844, "learning_rate": 1.984536046507125e-05, "loss": 0.6464, "step": 2369 }, { "epoch": 0.0844362897910469, "grad_norm": 1.4942055509965007, "learning_rate": 1.9845158250452822e-05, "loss": 0.5471, "step": 2370 }, { "epoch": 0.08447191691754101, "grad_norm": 2.7130284943222613, "learning_rate": 1.9844955904739226e-05, "loss": 0.6637, "step": 2371 }, { "epoch": 0.08450754404403513, "grad_norm": 1.2386846266108757, "learning_rate": 1.9844753427933163e-05, "loss": 0.4886, "step": 2372 }, { "epoch": 0.08454317117052924, "grad_norm": 1.3729072153124195, "learning_rate": 1.9844550820037326e-05, "loss": 0.5457, "step": 2373 }, { "epoch": 0.08457879829702335, "grad_norm": 1.2352690492630645, "learning_rate": 1.9844348081054413e-05, "loss": 0.4655, "step": 2374 }, { "epoch": 0.08461442542351746, "grad_norm": 1.93385768214817, "learning_rate": 1.9844145210987123e-05, "loss": 0.7131, "step": 2375 }, { "epoch": 0.08465005255001158, "grad_norm": 1.5113999239291727, "learning_rate": 1.9843942209838165e-05, "loss": 0.5337, "step": 2376 }, { "epoch": 0.08468567967650568, "grad_norm": 1.5417434533892216, "learning_rate": 1.9843739077610232e-05, "loss": 0.5365, "step": 2377 }, { "epoch": 0.0847213068029998, "grad_norm": 1.2540182648207694, "learning_rate": 1.9843535814306033e-05, "loss": 0.5248, "step": 2378 }, { "epoch": 0.08475693392949392, "grad_norm": 2.0349336389474098, "learning_rate": 1.9843332419928277e-05, "loss": 0.6836, "step": 2379 }, { "epoch": 0.08479256105598804, "grad_norm": 2.0855294424188338, "learning_rate": 1.9843128894479667e-05, "loss": 0.6643, "step": 2380 }, { "epoch": 0.08482818818248214, "grad_norm": 2.5748857859996686, "learning_rate": 1.9842925237962913e-05, "loss": 0.6054, "step": 2381 }, { "epoch": 0.08486381530897626, "grad_norm": 1.934180182632415, "learning_rate": 1.9842721450380737e-05, "loss": 0.6686, "step": 2382 }, { "epoch": 0.08489944243547037, "grad_norm": 1.748547795804549, "learning_rate": 1.9842517531735837e-05, "loss": 0.5597, "step": 2383 }, { "epoch": 0.08493506956196448, "grad_norm": 1.8562850709897971, "learning_rate": 1.9842313482030943e-05, "loss": 0.6123, "step": 2384 }, { "epoch": 0.0849706966884586, "grad_norm": 1.4359686357758699, "learning_rate": 1.984210930126876e-05, "loss": 0.5163, "step": 2385 }, { "epoch": 0.08500632381495271, "grad_norm": 1.8112527978737096, "learning_rate": 1.9841904989452018e-05, "loss": 0.6934, "step": 2386 }, { "epoch": 0.08504195094144681, "grad_norm": 2.142842774171888, "learning_rate": 1.9841700546583432e-05, "loss": 0.6164, "step": 2387 }, { "epoch": 0.08507757806794093, "grad_norm": 1.3519369629569504, "learning_rate": 1.9841495972665717e-05, "loss": 0.5713, "step": 2388 }, { "epoch": 0.08511320519443505, "grad_norm": 1.3745084590154804, "learning_rate": 1.9841291267701613e-05, "loss": 0.4955, "step": 2389 }, { "epoch": 0.08514883232092915, "grad_norm": 2.0436603530150745, "learning_rate": 1.9841086431693832e-05, "loss": 0.6308, "step": 2390 }, { "epoch": 0.08518445944742327, "grad_norm": 1.3064952813711066, "learning_rate": 1.9840881464645107e-05, "loss": 0.5101, "step": 2391 }, { "epoch": 0.08522008657391739, "grad_norm": 1.9577913955338722, "learning_rate": 1.9840676366558172e-05, "loss": 0.7159, "step": 2392 }, { "epoch": 0.08525571370041149, "grad_norm": 2.028584518597362, "learning_rate": 1.9840471137435746e-05, "loss": 0.6312, "step": 2393 }, { "epoch": 0.0852913408269056, "grad_norm": 2.259513974847013, "learning_rate": 1.984026577728057e-05, "loss": 0.5813, "step": 2394 }, { "epoch": 0.08532696795339972, "grad_norm": 1.9578766268072965, "learning_rate": 1.984006028609538e-05, "loss": 0.6745, "step": 2395 }, { "epoch": 0.08536259507989383, "grad_norm": 2.18738483175596, "learning_rate": 1.983985466388291e-05, "loss": 0.5832, "step": 2396 }, { "epoch": 0.08539822220638794, "grad_norm": 1.4701448085710485, "learning_rate": 1.9839648910645898e-05, "loss": 0.5853, "step": 2397 }, { "epoch": 0.08543384933288206, "grad_norm": 1.5349898371879582, "learning_rate": 1.9839443026387078e-05, "loss": 0.5724, "step": 2398 }, { "epoch": 0.08546947645937616, "grad_norm": 1.372101407626276, "learning_rate": 1.98392370111092e-05, "loss": 0.5627, "step": 2399 }, { "epoch": 0.08550510358587028, "grad_norm": 1.872984237459853, "learning_rate": 1.9839030864815004e-05, "loss": 0.6199, "step": 2400 }, { "epoch": 0.0855407307123644, "grad_norm": 1.2384204525051692, "learning_rate": 1.9838824587507236e-05, "loss": 0.5324, "step": 2401 }, { "epoch": 0.0855763578388585, "grad_norm": 1.3336396876758727, "learning_rate": 1.983861817918864e-05, "loss": 0.5526, "step": 2402 }, { "epoch": 0.08561198496535262, "grad_norm": 1.3802106578843234, "learning_rate": 1.9838411639861966e-05, "loss": 0.5649, "step": 2403 }, { "epoch": 0.08564761209184674, "grad_norm": 1.8993079118573981, "learning_rate": 1.9838204969529965e-05, "loss": 0.5234, "step": 2404 }, { "epoch": 0.08568323921834084, "grad_norm": 1.9768455958711677, "learning_rate": 1.983799816819539e-05, "loss": 0.6412, "step": 2405 }, { "epoch": 0.08571886634483496, "grad_norm": 1.8005617572593493, "learning_rate": 1.983779123586099e-05, "loss": 0.5721, "step": 2406 }, { "epoch": 0.08575449347132907, "grad_norm": 2.186833447202334, "learning_rate": 1.9837584172529526e-05, "loss": 0.5852, "step": 2407 }, { "epoch": 0.08579012059782318, "grad_norm": 1.3495123459359397, "learning_rate": 1.983737697820375e-05, "loss": 0.5122, "step": 2408 }, { "epoch": 0.0858257477243173, "grad_norm": 1.375538491742851, "learning_rate": 1.983716965288643e-05, "loss": 0.5636, "step": 2409 }, { "epoch": 0.08586137485081141, "grad_norm": 1.413125154306008, "learning_rate": 1.983696219658031e-05, "loss": 0.5893, "step": 2410 }, { "epoch": 0.08589700197730551, "grad_norm": 1.729522752737004, "learning_rate": 1.983675460928817e-05, "loss": 0.6014, "step": 2411 }, { "epoch": 0.08593262910379963, "grad_norm": 1.9267549766026493, "learning_rate": 1.9836546891012766e-05, "loss": 0.6053, "step": 2412 }, { "epoch": 0.08596825623029375, "grad_norm": 1.2791335369296495, "learning_rate": 1.9836339041756866e-05, "loss": 0.51, "step": 2413 }, { "epoch": 0.08600388335678785, "grad_norm": 1.28684620257003, "learning_rate": 1.983613106152323e-05, "loss": 0.4844, "step": 2414 }, { "epoch": 0.08603951048328197, "grad_norm": 1.5393408300451148, "learning_rate": 1.983592295031464e-05, "loss": 0.5423, "step": 2415 }, { "epoch": 0.08607513760977609, "grad_norm": 1.4094167781059428, "learning_rate": 1.983571470813386e-05, "loss": 0.5524, "step": 2416 }, { "epoch": 0.08611076473627019, "grad_norm": 1.4340286108275002, "learning_rate": 1.9835506334983665e-05, "loss": 0.574, "step": 2417 }, { "epoch": 0.08614639186276431, "grad_norm": 1.433547511402864, "learning_rate": 1.9835297830866827e-05, "loss": 0.5361, "step": 2418 }, { "epoch": 0.08618201898925842, "grad_norm": 1.5750879424880788, "learning_rate": 1.9835089195786122e-05, "loss": 0.5047, "step": 2419 }, { "epoch": 0.08621764611575253, "grad_norm": 2.345764465557652, "learning_rate": 1.983488042974433e-05, "loss": 0.643, "step": 2420 }, { "epoch": 0.08625327324224664, "grad_norm": 2.017176892095656, "learning_rate": 1.9834671532744238e-05, "loss": 0.6704, "step": 2421 }, { "epoch": 0.08628890036874076, "grad_norm": 2.1821596283682654, "learning_rate": 1.983446250478861e-05, "loss": 0.6281, "step": 2422 }, { "epoch": 0.08632452749523487, "grad_norm": 1.8595798486274275, "learning_rate": 1.9834253345880248e-05, "loss": 0.6347, "step": 2423 }, { "epoch": 0.08636015462172898, "grad_norm": 1.3792922005725676, "learning_rate": 1.9834044056021926e-05, "loss": 0.5424, "step": 2424 }, { "epoch": 0.0863957817482231, "grad_norm": 1.3423568046583236, "learning_rate": 1.9833834635216434e-05, "loss": 0.5279, "step": 2425 }, { "epoch": 0.08643140887471722, "grad_norm": 1.2400181302676458, "learning_rate": 1.983362508346656e-05, "loss": 0.5561, "step": 2426 }, { "epoch": 0.08646703600121132, "grad_norm": 1.4266746420952934, "learning_rate": 1.9833415400775092e-05, "loss": 0.5625, "step": 2427 }, { "epoch": 0.08650266312770544, "grad_norm": 2.6547729319508684, "learning_rate": 1.983320558714483e-05, "loss": 0.6436, "step": 2428 }, { "epoch": 0.08653829025419955, "grad_norm": 2.1121978072868974, "learning_rate": 1.983299564257856e-05, "loss": 0.6798, "step": 2429 }, { "epoch": 0.08657391738069366, "grad_norm": 2.2152213163240364, "learning_rate": 1.983278556707908e-05, "loss": 0.6088, "step": 2430 }, { "epoch": 0.08660954450718777, "grad_norm": 1.3451652915113435, "learning_rate": 1.983257536064919e-05, "loss": 0.5363, "step": 2431 }, { "epoch": 0.08664517163368189, "grad_norm": 2.2299216621501428, "learning_rate": 1.9832365023291686e-05, "loss": 0.614, "step": 2432 }, { "epoch": 0.086680798760176, "grad_norm": 1.3348079600361493, "learning_rate": 1.9832154555009367e-05, "loss": 0.5586, "step": 2433 }, { "epoch": 0.08671642588667011, "grad_norm": 1.4580659421241653, "learning_rate": 1.983194395580504e-05, "loss": 0.5158, "step": 2434 }, { "epoch": 0.08675205301316423, "grad_norm": 1.2496764836840963, "learning_rate": 1.9831733225681506e-05, "loss": 0.5501, "step": 2435 }, { "epoch": 0.08678768013965833, "grad_norm": 1.3106643873513475, "learning_rate": 1.983152236464157e-05, "loss": 0.4646, "step": 2436 }, { "epoch": 0.08682330726615245, "grad_norm": 1.802148965157533, "learning_rate": 1.9831311372688044e-05, "loss": 0.6039, "step": 2437 }, { "epoch": 0.08685893439264657, "grad_norm": 1.3126667712146045, "learning_rate": 1.9831100249823732e-05, "loss": 0.4592, "step": 2438 }, { "epoch": 0.08689456151914067, "grad_norm": 1.7958851955264907, "learning_rate": 1.9830888996051454e-05, "loss": 0.5891, "step": 2439 }, { "epoch": 0.08693018864563479, "grad_norm": 2.385523682034339, "learning_rate": 1.9830677611374012e-05, "loss": 0.6655, "step": 2440 }, { "epoch": 0.0869658157721289, "grad_norm": 2.8804929225382385, "learning_rate": 1.9830466095794235e-05, "loss": 0.622, "step": 2441 }, { "epoch": 0.08700144289862301, "grad_norm": 1.2892150387398098, "learning_rate": 1.9830254449314922e-05, "loss": 0.5557, "step": 2442 }, { "epoch": 0.08703707002511712, "grad_norm": 1.3892953217997466, "learning_rate": 1.9830042671938904e-05, "loss": 0.54, "step": 2443 }, { "epoch": 0.08707269715161124, "grad_norm": 1.3735031780254012, "learning_rate": 1.9829830763668998e-05, "loss": 0.5443, "step": 2444 }, { "epoch": 0.08710832427810535, "grad_norm": 1.9637560207572764, "learning_rate": 1.9829618724508025e-05, "loss": 0.6799, "step": 2445 }, { "epoch": 0.08714395140459946, "grad_norm": 1.5444912362486523, "learning_rate": 1.9829406554458806e-05, "loss": 0.5402, "step": 2446 }, { "epoch": 0.08717957853109358, "grad_norm": 2.5409621380229326, "learning_rate": 1.9829194253524168e-05, "loss": 0.7738, "step": 2447 }, { "epoch": 0.08721520565758768, "grad_norm": 1.9161087226831615, "learning_rate": 1.9828981821706937e-05, "loss": 0.6193, "step": 2448 }, { "epoch": 0.0872508327840818, "grad_norm": 1.3009819269089953, "learning_rate": 1.9828769259009947e-05, "loss": 0.5499, "step": 2449 }, { "epoch": 0.08728645991057592, "grad_norm": 1.404451061235646, "learning_rate": 1.9828556565436026e-05, "loss": 0.5205, "step": 2450 }, { "epoch": 0.08732208703707002, "grad_norm": 1.4022563155337813, "learning_rate": 1.9828343740988002e-05, "loss": 0.5853, "step": 2451 }, { "epoch": 0.08735771416356414, "grad_norm": 1.3587368044462798, "learning_rate": 1.982813078566871e-05, "loss": 0.5252, "step": 2452 }, { "epoch": 0.08739334129005825, "grad_norm": 1.2362136011062175, "learning_rate": 1.9827917699480984e-05, "loss": 0.47, "step": 2453 }, { "epoch": 0.08742896841655236, "grad_norm": 1.9131363450013146, "learning_rate": 1.982770448242767e-05, "loss": 0.5885, "step": 2454 }, { "epoch": 0.08746459554304647, "grad_norm": 1.2647672502148433, "learning_rate": 1.98274911345116e-05, "loss": 0.4869, "step": 2455 }, { "epoch": 0.08750022266954059, "grad_norm": 2.188677483004102, "learning_rate": 1.9827277655735617e-05, "loss": 0.6976, "step": 2456 }, { "epoch": 0.0875358497960347, "grad_norm": 1.3963621673033255, "learning_rate": 1.9827064046102564e-05, "loss": 0.5046, "step": 2457 }, { "epoch": 0.08757147692252881, "grad_norm": 2.0861187408376343, "learning_rate": 1.9826850305615283e-05, "loss": 0.724, "step": 2458 }, { "epoch": 0.08760710404902293, "grad_norm": 5.7114036844146066, "learning_rate": 1.9826636434276624e-05, "loss": 0.7371, "step": 2459 }, { "epoch": 0.08764273117551703, "grad_norm": 1.3319432015277215, "learning_rate": 1.982642243208943e-05, "loss": 0.5494, "step": 2460 }, { "epoch": 0.08767835830201115, "grad_norm": 1.3555681915202757, "learning_rate": 1.9826208299056554e-05, "loss": 0.581, "step": 2461 }, { "epoch": 0.08771398542850527, "grad_norm": 1.439548653419518, "learning_rate": 1.9825994035180846e-05, "loss": 0.5037, "step": 2462 }, { "epoch": 0.08774961255499937, "grad_norm": 1.7078718179034778, "learning_rate": 1.9825779640465157e-05, "loss": 0.5915, "step": 2463 }, { "epoch": 0.08778523968149349, "grad_norm": 2.4833925524548075, "learning_rate": 1.982556511491235e-05, "loss": 0.6335, "step": 2464 }, { "epoch": 0.0878208668079876, "grad_norm": 1.9106058523395144, "learning_rate": 1.9825350458525272e-05, "loss": 0.586, "step": 2465 }, { "epoch": 0.08785649393448171, "grad_norm": 1.7126620987257644, "learning_rate": 1.9825135671306784e-05, "loss": 0.6169, "step": 2466 }, { "epoch": 0.08789212106097583, "grad_norm": 1.7500569573748417, "learning_rate": 1.982492075325975e-05, "loss": 0.5645, "step": 2467 }, { "epoch": 0.08792774818746994, "grad_norm": 3.336764040231692, "learning_rate": 1.982470570438703e-05, "loss": 0.6205, "step": 2468 }, { "epoch": 0.08796337531396405, "grad_norm": 1.3049759935352558, "learning_rate": 1.9824490524691482e-05, "loss": 0.4955, "step": 2469 }, { "epoch": 0.08799900244045816, "grad_norm": 2.238374942533461, "learning_rate": 1.9824275214175978e-05, "loss": 0.5896, "step": 2470 }, { "epoch": 0.08803462956695228, "grad_norm": 1.3897008714088825, "learning_rate": 1.982405977284338e-05, "loss": 0.5248, "step": 2471 }, { "epoch": 0.08807025669344638, "grad_norm": 2.7953884815749777, "learning_rate": 1.9823844200696564e-05, "loss": 0.6693, "step": 2472 }, { "epoch": 0.0881058838199405, "grad_norm": 1.434162453941207, "learning_rate": 1.9823628497738397e-05, "loss": 0.5723, "step": 2473 }, { "epoch": 0.08814151094643462, "grad_norm": 1.5624209297243372, "learning_rate": 1.9823412663971746e-05, "loss": 0.5631, "step": 2474 }, { "epoch": 0.08817713807292873, "grad_norm": 2.766330534964731, "learning_rate": 1.9823196699399488e-05, "loss": 0.6594, "step": 2475 }, { "epoch": 0.08821276519942284, "grad_norm": 2.2878789025624027, "learning_rate": 1.9822980604024506e-05, "loss": 0.6499, "step": 2476 }, { "epoch": 0.08824839232591696, "grad_norm": 1.7190059743847053, "learning_rate": 1.982276437784967e-05, "loss": 0.5098, "step": 2477 }, { "epoch": 0.08828401945241107, "grad_norm": 1.409868818991232, "learning_rate": 1.9822548020877858e-05, "loss": 0.5405, "step": 2478 }, { "epoch": 0.08831964657890518, "grad_norm": 2.270392021458229, "learning_rate": 1.9822331533111952e-05, "loss": 0.6355, "step": 2479 }, { "epoch": 0.08835527370539929, "grad_norm": 1.7285401538937932, "learning_rate": 1.982211491455484e-05, "loss": 0.5657, "step": 2480 }, { "epoch": 0.08839090083189341, "grad_norm": 3.4512632091639093, "learning_rate": 1.98218981652094e-05, "loss": 0.5991, "step": 2481 }, { "epoch": 0.08842652795838751, "grad_norm": 2.8992065856650084, "learning_rate": 1.9821681285078522e-05, "loss": 0.611, "step": 2482 }, { "epoch": 0.08846215508488163, "grad_norm": 1.7150837008214332, "learning_rate": 1.9821464274165095e-05, "loss": 0.6496, "step": 2483 }, { "epoch": 0.08849778221137575, "grad_norm": 1.7710253998843788, "learning_rate": 1.9821247132472004e-05, "loss": 0.5924, "step": 2484 }, { "epoch": 0.08853340933786985, "grad_norm": 1.7520504672926405, "learning_rate": 1.9821029860002145e-05, "loss": 0.5085, "step": 2485 }, { "epoch": 0.08856903646436397, "grad_norm": 1.8467516267288155, "learning_rate": 1.98208124567584e-05, "loss": 0.6328, "step": 2486 }, { "epoch": 0.08860466359085808, "grad_norm": 1.3248627397698924, "learning_rate": 1.9820594922743684e-05, "loss": 0.5303, "step": 2487 }, { "epoch": 0.08864029071735219, "grad_norm": 1.3200461213318653, "learning_rate": 1.982037725796088e-05, "loss": 0.5016, "step": 2488 }, { "epoch": 0.0886759178438463, "grad_norm": 2.1374184120232815, "learning_rate": 1.9820159462412885e-05, "loss": 0.6747, "step": 2489 }, { "epoch": 0.08871154497034042, "grad_norm": 2.5290236996152284, "learning_rate": 1.9819941536102604e-05, "loss": 0.6273, "step": 2490 }, { "epoch": 0.08874717209683453, "grad_norm": 1.3513454486506462, "learning_rate": 1.981972347903294e-05, "loss": 0.5333, "step": 2491 }, { "epoch": 0.08878279922332864, "grad_norm": 1.392258093807407, "learning_rate": 1.9819505291206792e-05, "loss": 0.59, "step": 2492 }, { "epoch": 0.08881842634982276, "grad_norm": 1.4694632406510781, "learning_rate": 1.9819286972627066e-05, "loss": 0.5612, "step": 2493 }, { "epoch": 0.08885405347631686, "grad_norm": 2.1386682196970233, "learning_rate": 1.9819068523296677e-05, "loss": 0.7115, "step": 2494 }, { "epoch": 0.08888968060281098, "grad_norm": 2.554744054214384, "learning_rate": 1.981884994321852e-05, "loss": 0.5628, "step": 2495 }, { "epoch": 0.0889253077293051, "grad_norm": 3.6738727137043554, "learning_rate": 1.981863123239552e-05, "loss": 0.7337, "step": 2496 }, { "epoch": 0.0889609348557992, "grad_norm": 1.6108649731469904, "learning_rate": 1.9818412390830576e-05, "loss": 0.5611, "step": 2497 }, { "epoch": 0.08899656198229332, "grad_norm": 1.509268244440054, "learning_rate": 1.9818193418526613e-05, "loss": 0.5616, "step": 2498 }, { "epoch": 0.08903218910878744, "grad_norm": 1.95458751738047, "learning_rate": 1.981797431548654e-05, "loss": 0.5697, "step": 2499 }, { "epoch": 0.08906781623528154, "grad_norm": 2.489731903438686, "learning_rate": 1.9817755081713277e-05, "loss": 0.5601, "step": 2500 }, { "epoch": 0.08910344336177566, "grad_norm": 1.722231735813237, "learning_rate": 1.9817535717209743e-05, "loss": 0.5915, "step": 2501 }, { "epoch": 0.08913907048826977, "grad_norm": 1.2599205657340473, "learning_rate": 1.9817316221978863e-05, "loss": 0.4524, "step": 2502 }, { "epoch": 0.08917469761476388, "grad_norm": 1.4439343992249165, "learning_rate": 1.981709659602355e-05, "loss": 0.5553, "step": 2503 }, { "epoch": 0.08921032474125799, "grad_norm": 1.6533594297730994, "learning_rate": 1.9816876839346734e-05, "loss": 0.5341, "step": 2504 }, { "epoch": 0.08924595186775211, "grad_norm": 1.4664889855334793, "learning_rate": 1.9816656951951343e-05, "loss": 0.5584, "step": 2505 }, { "epoch": 0.08928157899424621, "grad_norm": 2.1105855106566125, "learning_rate": 1.9816436933840303e-05, "loss": 0.6179, "step": 2506 }, { "epoch": 0.08931720612074033, "grad_norm": 1.3670506315349245, "learning_rate": 1.9816216785016545e-05, "loss": 0.5274, "step": 2507 }, { "epoch": 0.08935283324723445, "grad_norm": 2.117004318105296, "learning_rate": 1.9815996505483e-05, "loss": 0.6156, "step": 2508 }, { "epoch": 0.08938846037372855, "grad_norm": 2.581805177693072, "learning_rate": 1.98157760952426e-05, "loss": 0.5751, "step": 2509 }, { "epoch": 0.08942408750022267, "grad_norm": 2.0971415098401276, "learning_rate": 1.9815555554298277e-05, "loss": 0.6438, "step": 2510 }, { "epoch": 0.08945971462671679, "grad_norm": 1.9852184689737071, "learning_rate": 1.9815334882652976e-05, "loss": 0.558, "step": 2511 }, { "epoch": 0.08949534175321089, "grad_norm": 1.9600543465114788, "learning_rate": 1.9815114080309624e-05, "loss": 0.6031, "step": 2512 }, { "epoch": 0.089530968879705, "grad_norm": 1.484048221177659, "learning_rate": 1.9814893147271172e-05, "loss": 0.5349, "step": 2513 }, { "epoch": 0.08956659600619912, "grad_norm": 1.4668608477750766, "learning_rate": 1.9814672083540558e-05, "loss": 0.5828, "step": 2514 }, { "epoch": 0.08960222313269323, "grad_norm": 1.9132322103606008, "learning_rate": 1.9814450889120725e-05, "loss": 0.5994, "step": 2515 }, { "epoch": 0.08963785025918734, "grad_norm": 1.9161981866673647, "learning_rate": 1.9814229564014618e-05, "loss": 0.61, "step": 2516 }, { "epoch": 0.08967347738568146, "grad_norm": 2.178970459805342, "learning_rate": 1.9814008108225185e-05, "loss": 0.585, "step": 2517 }, { "epoch": 0.08970910451217556, "grad_norm": 1.2623784175579469, "learning_rate": 1.9813786521755373e-05, "loss": 0.5235, "step": 2518 }, { "epoch": 0.08974473163866968, "grad_norm": 1.4380070922576609, "learning_rate": 1.9813564804608138e-05, "loss": 0.5272, "step": 2519 }, { "epoch": 0.0897803587651638, "grad_norm": 2.1539224616544996, "learning_rate": 1.9813342956786422e-05, "loss": 0.6015, "step": 2520 }, { "epoch": 0.08981598589165792, "grad_norm": 2.012301190993882, "learning_rate": 1.981312097829319e-05, "loss": 0.6582, "step": 2521 }, { "epoch": 0.08985161301815202, "grad_norm": 1.3538633175111514, "learning_rate": 1.9812898869131392e-05, "loss": 0.5331, "step": 2522 }, { "epoch": 0.08988724014464614, "grad_norm": 1.8503518998900075, "learning_rate": 1.9812676629303987e-05, "loss": 0.6729, "step": 2523 }, { "epoch": 0.08992286727114025, "grad_norm": 1.379147639083531, "learning_rate": 1.9812454258813934e-05, "loss": 0.4986, "step": 2524 }, { "epoch": 0.08995849439763436, "grad_norm": 2.8702469338000265, "learning_rate": 1.9812231757664193e-05, "loss": 0.6231, "step": 2525 }, { "epoch": 0.08999412152412847, "grad_norm": 1.8274872741361692, "learning_rate": 1.981200912585773e-05, "loss": 0.6022, "step": 2526 }, { "epoch": 0.09002974865062259, "grad_norm": 1.8186447131123737, "learning_rate": 1.9811786363397508e-05, "loss": 0.627, "step": 2527 }, { "epoch": 0.0900653757771167, "grad_norm": 1.4419847240160304, "learning_rate": 1.981156347028649e-05, "loss": 0.5065, "step": 2528 }, { "epoch": 0.09010100290361081, "grad_norm": 1.410364288892441, "learning_rate": 1.9811340446527646e-05, "loss": 0.5305, "step": 2529 }, { "epoch": 0.09013663003010493, "grad_norm": 1.4576064645618956, "learning_rate": 1.9811117292123948e-05, "loss": 0.5593, "step": 2530 }, { "epoch": 0.09017225715659903, "grad_norm": 2.312445070678533, "learning_rate": 1.9810894007078364e-05, "loss": 0.572, "step": 2531 }, { "epoch": 0.09020788428309315, "grad_norm": 2.6868138737093825, "learning_rate": 1.981067059139387e-05, "loss": 0.619, "step": 2532 }, { "epoch": 0.09024351140958727, "grad_norm": 1.8559143520242802, "learning_rate": 1.981044704507344e-05, "loss": 0.6681, "step": 2533 }, { "epoch": 0.09027913853608137, "grad_norm": 1.5177150249739852, "learning_rate": 1.981022336812005e-05, "loss": 0.5779, "step": 2534 }, { "epoch": 0.09031476566257549, "grad_norm": 1.5709749521074519, "learning_rate": 1.9809999560536683e-05, "loss": 0.5584, "step": 2535 }, { "epoch": 0.0903503927890696, "grad_norm": 1.5189804118821675, "learning_rate": 1.9809775622326307e-05, "loss": 0.5315, "step": 2536 }, { "epoch": 0.0903860199155637, "grad_norm": 2.701495346216973, "learning_rate": 1.9809551553491918e-05, "loss": 0.547, "step": 2537 }, { "epoch": 0.09042164704205782, "grad_norm": 2.3179227300074388, "learning_rate": 1.9809327354036493e-05, "loss": 0.7229, "step": 2538 }, { "epoch": 0.09045727416855194, "grad_norm": 1.4203560210390618, "learning_rate": 1.9809103023963017e-05, "loss": 0.5676, "step": 2539 }, { "epoch": 0.09049290129504604, "grad_norm": 1.4428566444598818, "learning_rate": 1.9808878563274476e-05, "loss": 0.5478, "step": 2540 }, { "epoch": 0.09052852842154016, "grad_norm": 1.4244496986599058, "learning_rate": 1.9808653971973865e-05, "loss": 0.5932, "step": 2541 }, { "epoch": 0.09056415554803428, "grad_norm": 2.0488920175365677, "learning_rate": 1.9808429250064164e-05, "loss": 0.5556, "step": 2542 }, { "epoch": 0.09059978267452838, "grad_norm": 1.8661433787564157, "learning_rate": 1.9808204397548377e-05, "loss": 0.7134, "step": 2543 }, { "epoch": 0.0906354098010225, "grad_norm": 1.5851850257679072, "learning_rate": 1.9807979414429494e-05, "loss": 0.5181, "step": 2544 }, { "epoch": 0.09067103692751662, "grad_norm": 1.4782264375842329, "learning_rate": 1.9807754300710507e-05, "loss": 0.5756, "step": 2545 }, { "epoch": 0.09070666405401072, "grad_norm": 1.319644886710472, "learning_rate": 1.9807529056394416e-05, "loss": 0.5675, "step": 2546 }, { "epoch": 0.09074229118050484, "grad_norm": 1.3337749946576667, "learning_rate": 1.9807303681484224e-05, "loss": 0.6008, "step": 2547 }, { "epoch": 0.09077791830699895, "grad_norm": 1.3140720143189273, "learning_rate": 1.9807078175982925e-05, "loss": 0.5101, "step": 2548 }, { "epoch": 0.09081354543349306, "grad_norm": 1.799296130668542, "learning_rate": 1.9806852539893526e-05, "loss": 0.5918, "step": 2549 }, { "epoch": 0.09084917255998717, "grad_norm": 1.4027497417240502, "learning_rate": 1.9806626773219032e-05, "loss": 0.5336, "step": 2550 }, { "epoch": 0.09088479968648129, "grad_norm": 1.3251691431439891, "learning_rate": 1.9806400875962447e-05, "loss": 0.575, "step": 2551 }, { "epoch": 0.0909204268129754, "grad_norm": 1.8066572268697287, "learning_rate": 1.980617484812678e-05, "loss": 0.6815, "step": 2552 }, { "epoch": 0.09095605393946951, "grad_norm": 1.9811080616104084, "learning_rate": 1.9805948689715043e-05, "loss": 0.6303, "step": 2553 }, { "epoch": 0.09099168106596363, "grad_norm": 2.5971404873087907, "learning_rate": 1.9805722400730243e-05, "loss": 0.6564, "step": 2554 }, { "epoch": 0.09102730819245773, "grad_norm": 2.054520021842595, "learning_rate": 1.98054959811754e-05, "loss": 0.6086, "step": 2555 }, { "epoch": 0.09106293531895185, "grad_norm": 1.828995637384066, "learning_rate": 1.980526943105352e-05, "loss": 0.6482, "step": 2556 }, { "epoch": 0.09109856244544597, "grad_norm": 1.3808098962581385, "learning_rate": 1.9805042750367622e-05, "loss": 0.4977, "step": 2557 }, { "epoch": 0.09113418957194007, "grad_norm": 2.416019992693956, "learning_rate": 1.980481593912073e-05, "loss": 0.6713, "step": 2558 }, { "epoch": 0.09116981669843419, "grad_norm": 1.257963172630653, "learning_rate": 1.9804588997315858e-05, "loss": 0.5124, "step": 2559 }, { "epoch": 0.0912054438249283, "grad_norm": 1.439648322467675, "learning_rate": 1.9804361924956034e-05, "loss": 0.5355, "step": 2560 }, { "epoch": 0.0912410709514224, "grad_norm": 1.6728212291289262, "learning_rate": 1.9804134722044276e-05, "loss": 0.5278, "step": 2561 }, { "epoch": 0.09127669807791652, "grad_norm": 1.5823070347922312, "learning_rate": 1.9803907388583613e-05, "loss": 0.6691, "step": 2562 }, { "epoch": 0.09131232520441064, "grad_norm": 1.7481372479224133, "learning_rate": 1.9803679924577066e-05, "loss": 0.5839, "step": 2563 }, { "epoch": 0.09134795233090474, "grad_norm": 1.8928427454634444, "learning_rate": 1.980345233002767e-05, "loss": 0.6715, "step": 2564 }, { "epoch": 0.09138357945739886, "grad_norm": 1.9528322591486516, "learning_rate": 1.980322460493846e-05, "loss": 0.5843, "step": 2565 }, { "epoch": 0.09141920658389298, "grad_norm": 2.3391669406489717, "learning_rate": 1.9802996749312456e-05, "loss": 0.6685, "step": 2566 }, { "epoch": 0.0914548337103871, "grad_norm": 1.3237695867966257, "learning_rate": 1.9802768763152703e-05, "loss": 0.4874, "step": 2567 }, { "epoch": 0.0914904608368812, "grad_norm": 2.2171622503675157, "learning_rate": 1.980254064646223e-05, "loss": 0.5976, "step": 2568 }, { "epoch": 0.09152608796337532, "grad_norm": 1.8267561016843343, "learning_rate": 1.9802312399244075e-05, "loss": 0.5255, "step": 2569 }, { "epoch": 0.09156171508986943, "grad_norm": 1.4654846364289877, "learning_rate": 1.9802084021501282e-05, "loss": 0.5732, "step": 2570 }, { "epoch": 0.09159734221636354, "grad_norm": 1.7291334602542594, "learning_rate": 1.9801855513236888e-05, "loss": 0.6328, "step": 2571 }, { "epoch": 0.09163296934285765, "grad_norm": 1.9640771381670576, "learning_rate": 1.9801626874453933e-05, "loss": 0.729, "step": 2572 }, { "epoch": 0.09166859646935177, "grad_norm": 1.368095590095667, "learning_rate": 1.9801398105155472e-05, "loss": 0.5414, "step": 2573 }, { "epoch": 0.09170422359584587, "grad_norm": 1.7331760540672336, "learning_rate": 1.980116920534454e-05, "loss": 0.5676, "step": 2574 }, { "epoch": 0.09173985072233999, "grad_norm": 1.3493367959500393, "learning_rate": 1.980094017502419e-05, "loss": 0.5354, "step": 2575 }, { "epoch": 0.09177547784883411, "grad_norm": 2.044789574791432, "learning_rate": 1.980071101419747e-05, "loss": 0.5621, "step": 2576 }, { "epoch": 0.09181110497532821, "grad_norm": 1.6643275375224809, "learning_rate": 1.9800481722867434e-05, "loss": 0.5238, "step": 2577 }, { "epoch": 0.09184673210182233, "grad_norm": 2.027005873396015, "learning_rate": 1.9800252301037135e-05, "loss": 0.612, "step": 2578 }, { "epoch": 0.09188235922831645, "grad_norm": 1.3408285919770233, "learning_rate": 1.9800022748709624e-05, "loss": 0.5388, "step": 2579 }, { "epoch": 0.09191798635481055, "grad_norm": 1.483228000687004, "learning_rate": 1.9799793065887966e-05, "loss": 0.5776, "step": 2580 }, { "epoch": 0.09195361348130467, "grad_norm": 1.4202199159555198, "learning_rate": 1.9799563252575208e-05, "loss": 0.514, "step": 2581 }, { "epoch": 0.09198924060779878, "grad_norm": 1.3522687388902932, "learning_rate": 1.9799333308774415e-05, "loss": 0.55, "step": 2582 }, { "epoch": 0.09202486773429289, "grad_norm": 1.4292039189128967, "learning_rate": 1.9799103234488654e-05, "loss": 0.5915, "step": 2583 }, { "epoch": 0.092060494860787, "grad_norm": 2.1078280041185753, "learning_rate": 1.979887302972098e-05, "loss": 0.7566, "step": 2584 }, { "epoch": 0.09209612198728112, "grad_norm": 1.4050654518126056, "learning_rate": 1.9798642694474465e-05, "loss": 0.5753, "step": 2585 }, { "epoch": 0.09213174911377522, "grad_norm": 1.3316094833531686, "learning_rate": 1.9798412228752174e-05, "loss": 0.5252, "step": 2586 }, { "epoch": 0.09216737624026934, "grad_norm": 1.2512550386959305, "learning_rate": 1.9798181632557176e-05, "loss": 0.5155, "step": 2587 }, { "epoch": 0.09220300336676346, "grad_norm": 1.8535864349426499, "learning_rate": 1.9797950905892537e-05, "loss": 0.6262, "step": 2588 }, { "epoch": 0.09223863049325756, "grad_norm": 1.8762088275305187, "learning_rate": 1.979772004876134e-05, "loss": 0.5753, "step": 2589 }, { "epoch": 0.09227425761975168, "grad_norm": 1.3976418421436019, "learning_rate": 1.979748906116665e-05, "loss": 0.571, "step": 2590 }, { "epoch": 0.0923098847462458, "grad_norm": 1.3347559170003738, "learning_rate": 1.9797257943111542e-05, "loss": 0.4592, "step": 2591 }, { "epoch": 0.0923455118727399, "grad_norm": 1.8528902509595497, "learning_rate": 1.9797026694599097e-05, "loss": 0.6249, "step": 2592 }, { "epoch": 0.09238113899923402, "grad_norm": 1.2972554375239447, "learning_rate": 1.9796795315632394e-05, "loss": 0.5312, "step": 2593 }, { "epoch": 0.09241676612572813, "grad_norm": 1.845436748134775, "learning_rate": 1.9796563806214516e-05, "loss": 0.6171, "step": 2594 }, { "epoch": 0.09245239325222224, "grad_norm": 2.693479768401358, "learning_rate": 1.9796332166348543e-05, "loss": 0.6809, "step": 2595 }, { "epoch": 0.09248802037871635, "grad_norm": 2.3243322175552956, "learning_rate": 1.979610039603756e-05, "loss": 0.5917, "step": 2596 }, { "epoch": 0.09252364750521047, "grad_norm": 1.3996678848686868, "learning_rate": 1.9795868495284654e-05, "loss": 0.5203, "step": 2597 }, { "epoch": 0.09255927463170457, "grad_norm": 2.046465447120759, "learning_rate": 1.979563646409291e-05, "loss": 0.632, "step": 2598 }, { "epoch": 0.09259490175819869, "grad_norm": 1.6522846284987394, "learning_rate": 1.979540430246542e-05, "loss": 0.6753, "step": 2599 }, { "epoch": 0.09263052888469281, "grad_norm": 1.5143983660096427, "learning_rate": 1.9795172010405274e-05, "loss": 0.5388, "step": 2600 }, { "epoch": 0.09266615601118691, "grad_norm": 2.0452576714713877, "learning_rate": 1.979493958791557e-05, "loss": 0.666, "step": 2601 }, { "epoch": 0.09270178313768103, "grad_norm": 1.380767470236677, "learning_rate": 1.9794707034999396e-05, "loss": 0.5843, "step": 2602 }, { "epoch": 0.09273741026417515, "grad_norm": 1.3522221952926916, "learning_rate": 1.9794474351659854e-05, "loss": 0.542, "step": 2603 }, { "epoch": 0.09277303739066925, "grad_norm": 1.9035729696386967, "learning_rate": 1.979424153790004e-05, "loss": 0.6504, "step": 2604 }, { "epoch": 0.09280866451716337, "grad_norm": 1.9600414622807072, "learning_rate": 1.979400859372305e-05, "loss": 0.694, "step": 2605 }, { "epoch": 0.09284429164365748, "grad_norm": 1.3348310467446478, "learning_rate": 1.9793775519131994e-05, "loss": 0.494, "step": 2606 }, { "epoch": 0.09287991877015159, "grad_norm": 1.3206899890723667, "learning_rate": 1.9793542314129972e-05, "loss": 0.5154, "step": 2607 }, { "epoch": 0.0929155458966457, "grad_norm": 1.3309794539436401, "learning_rate": 1.9793308978720086e-05, "loss": 0.5319, "step": 2608 }, { "epoch": 0.09295117302313982, "grad_norm": 1.2776037561746598, "learning_rate": 1.979307551290545e-05, "loss": 0.5508, "step": 2609 }, { "epoch": 0.09298680014963392, "grad_norm": 1.7726957646400334, "learning_rate": 1.9792841916689163e-05, "loss": 0.6375, "step": 2610 }, { "epoch": 0.09302242727612804, "grad_norm": 2.136459607626097, "learning_rate": 1.9792608190074345e-05, "loss": 0.5363, "step": 2611 }, { "epoch": 0.09305805440262216, "grad_norm": 2.1029757755600635, "learning_rate": 1.9792374333064105e-05, "loss": 0.6557, "step": 2612 }, { "epoch": 0.09309368152911626, "grad_norm": 1.33006068131764, "learning_rate": 1.979214034566155e-05, "loss": 0.5505, "step": 2613 }, { "epoch": 0.09312930865561038, "grad_norm": 1.3013794571620383, "learning_rate": 1.979190622786981e-05, "loss": 0.4937, "step": 2614 }, { "epoch": 0.0931649357821045, "grad_norm": 1.6111399480013457, "learning_rate": 1.9791671979691988e-05, "loss": 0.5873, "step": 2615 }, { "epoch": 0.09320056290859861, "grad_norm": 1.369320360661967, "learning_rate": 1.9791437601131215e-05, "loss": 0.5572, "step": 2616 }, { "epoch": 0.09323619003509272, "grad_norm": 2.1435265002478325, "learning_rate": 1.9791203092190605e-05, "loss": 0.6356, "step": 2617 }, { "epoch": 0.09327181716158683, "grad_norm": 1.333116050586332, "learning_rate": 1.979096845287328e-05, "loss": 0.5546, "step": 2618 }, { "epoch": 0.09330744428808095, "grad_norm": 2.3209905347810933, "learning_rate": 1.9790733683182366e-05, "loss": 0.6976, "step": 2619 }, { "epoch": 0.09334307141457505, "grad_norm": 1.450089731669035, "learning_rate": 1.9790498783120992e-05, "loss": 0.5095, "step": 2620 }, { "epoch": 0.09337869854106917, "grad_norm": 1.9470043977614309, "learning_rate": 1.9790263752692286e-05, "loss": 0.6617, "step": 2621 }, { "epoch": 0.09341432566756329, "grad_norm": 2.1149481124237757, "learning_rate": 1.979002859189937e-05, "loss": 0.5283, "step": 2622 }, { "epoch": 0.09344995279405739, "grad_norm": 1.2317386616067332, "learning_rate": 1.9789793300745383e-05, "loss": 0.4771, "step": 2623 }, { "epoch": 0.09348557992055151, "grad_norm": 1.2134615891526257, "learning_rate": 1.9789557879233454e-05, "loss": 0.5154, "step": 2624 }, { "epoch": 0.09352120704704563, "grad_norm": 2.464775246027429, "learning_rate": 1.9789322327366722e-05, "loss": 0.6862, "step": 2625 }, { "epoch": 0.09355683417353973, "grad_norm": 1.3577340890480236, "learning_rate": 1.9789086645148318e-05, "loss": 0.5592, "step": 2626 }, { "epoch": 0.09359246130003385, "grad_norm": 1.1261304217653627, "learning_rate": 1.978885083258139e-05, "loss": 0.5452, "step": 2627 }, { "epoch": 0.09362808842652796, "grad_norm": 1.2924209339716226, "learning_rate": 1.9788614889669063e-05, "loss": 0.5151, "step": 2628 }, { "epoch": 0.09366371555302207, "grad_norm": 2.0254353962160554, "learning_rate": 1.978837881641449e-05, "loss": 0.5927, "step": 2629 }, { "epoch": 0.09369934267951618, "grad_norm": 1.2711992820256848, "learning_rate": 1.9788142612820808e-05, "loss": 0.5116, "step": 2630 }, { "epoch": 0.0937349698060103, "grad_norm": 1.22829907860389, "learning_rate": 1.9787906278891172e-05, "loss": 0.5191, "step": 2631 }, { "epoch": 0.0937705969325044, "grad_norm": 1.3837822701203435, "learning_rate": 1.9787669814628718e-05, "loss": 0.5804, "step": 2632 }, { "epoch": 0.09380622405899852, "grad_norm": 1.3260430126684948, "learning_rate": 1.97874332200366e-05, "loss": 0.5852, "step": 2633 }, { "epoch": 0.09384185118549264, "grad_norm": 1.2254871071507047, "learning_rate": 1.978719649511797e-05, "loss": 0.4871, "step": 2634 }, { "epoch": 0.09387747831198674, "grad_norm": 1.4062776128560488, "learning_rate": 1.9786959639875976e-05, "loss": 0.5247, "step": 2635 }, { "epoch": 0.09391310543848086, "grad_norm": 1.1931460098657514, "learning_rate": 1.9786722654313773e-05, "loss": 0.4938, "step": 2636 }, { "epoch": 0.09394873256497498, "grad_norm": 1.2386755789550628, "learning_rate": 1.9786485538434518e-05, "loss": 0.4572, "step": 2637 }, { "epoch": 0.09398435969146908, "grad_norm": 2.114248548817853, "learning_rate": 1.9786248292241368e-05, "loss": 0.5805, "step": 2638 }, { "epoch": 0.0940199868179632, "grad_norm": 2.0882784052675096, "learning_rate": 1.978601091573748e-05, "loss": 0.6008, "step": 2639 }, { "epoch": 0.09405561394445731, "grad_norm": 1.517956682702297, "learning_rate": 1.978577340892602e-05, "loss": 0.5001, "step": 2640 }, { "epoch": 0.09409124107095142, "grad_norm": 1.7733973263819225, "learning_rate": 1.9785535771810145e-05, "loss": 0.619, "step": 2641 }, { "epoch": 0.09412686819744553, "grad_norm": 2.3225494960906463, "learning_rate": 1.978529800439302e-05, "loss": 0.5751, "step": 2642 }, { "epoch": 0.09416249532393965, "grad_norm": 1.5328430351768192, "learning_rate": 1.9785060106677818e-05, "loss": 0.5792, "step": 2643 }, { "epoch": 0.09419812245043375, "grad_norm": 1.8384642586919486, "learning_rate": 1.9784822078667698e-05, "loss": 0.5989, "step": 2644 }, { "epoch": 0.09423374957692787, "grad_norm": 1.2979463854478472, "learning_rate": 1.978458392036583e-05, "loss": 0.5248, "step": 2645 }, { "epoch": 0.09426937670342199, "grad_norm": 1.3592067784207604, "learning_rate": 1.9784345631775393e-05, "loss": 0.5281, "step": 2646 }, { "epoch": 0.09430500382991609, "grad_norm": 1.7235650807395095, "learning_rate": 1.978410721289955e-05, "loss": 0.6112, "step": 2647 }, { "epoch": 0.09434063095641021, "grad_norm": 1.6178669642356176, "learning_rate": 1.9783868663741487e-05, "loss": 0.6028, "step": 2648 }, { "epoch": 0.09437625808290433, "grad_norm": 1.2992037501736549, "learning_rate": 1.9783629984304368e-05, "loss": 0.518, "step": 2649 }, { "epoch": 0.09441188520939843, "grad_norm": 1.2486267196578713, "learning_rate": 1.9783391174591383e-05, "loss": 0.4779, "step": 2650 }, { "epoch": 0.09444751233589255, "grad_norm": 1.4191761593812289, "learning_rate": 1.9783152234605703e-05, "loss": 0.5105, "step": 2651 }, { "epoch": 0.09448313946238666, "grad_norm": 2.1177948549774213, "learning_rate": 1.9782913164350513e-05, "loss": 0.6306, "step": 2652 }, { "epoch": 0.09451876658888077, "grad_norm": 2.150951792447335, "learning_rate": 1.9782673963828998e-05, "loss": 0.6263, "step": 2653 }, { "epoch": 0.09455439371537488, "grad_norm": 1.5839217125496396, "learning_rate": 1.978243463304434e-05, "loss": 0.5112, "step": 2654 }, { "epoch": 0.094590020841869, "grad_norm": 1.491947218621465, "learning_rate": 1.9782195171999728e-05, "loss": 0.5443, "step": 2655 }, { "epoch": 0.0946256479683631, "grad_norm": 1.724282910314135, "learning_rate": 1.978195558069835e-05, "loss": 0.6652, "step": 2656 }, { "epoch": 0.09466127509485722, "grad_norm": 1.4651789736406102, "learning_rate": 1.9781715859143396e-05, "loss": 0.5984, "step": 2657 }, { "epoch": 0.09469690222135134, "grad_norm": 1.8395006347504363, "learning_rate": 1.9781476007338058e-05, "loss": 0.6869, "step": 2658 }, { "epoch": 0.09473252934784544, "grad_norm": 1.5914286109917344, "learning_rate": 1.978123602528553e-05, "loss": 0.6969, "step": 2659 }, { "epoch": 0.09476815647433956, "grad_norm": 1.946656107866648, "learning_rate": 1.978099591298901e-05, "loss": 0.695, "step": 2660 }, { "epoch": 0.09480378360083368, "grad_norm": 2.075790311039734, "learning_rate": 1.978075567045169e-05, "loss": 0.555, "step": 2661 }, { "epoch": 0.0948394107273278, "grad_norm": 1.9761538359117148, "learning_rate": 1.9780515297676772e-05, "loss": 0.6467, "step": 2662 }, { "epoch": 0.0948750378538219, "grad_norm": 1.451977035319678, "learning_rate": 1.978027479466746e-05, "loss": 0.5516, "step": 2663 }, { "epoch": 0.09491066498031601, "grad_norm": 1.8352061319201505, "learning_rate": 1.978003416142695e-05, "loss": 0.5516, "step": 2664 }, { "epoch": 0.09494629210681013, "grad_norm": 1.9432352183750048, "learning_rate": 1.9779793397958448e-05, "loss": 0.6122, "step": 2665 }, { "epoch": 0.09498191923330424, "grad_norm": 1.502929836971656, "learning_rate": 1.9779552504265163e-05, "loss": 0.5635, "step": 2666 }, { "epoch": 0.09501754635979835, "grad_norm": 1.3555060900154325, "learning_rate": 1.9779311480350303e-05, "loss": 0.5473, "step": 2667 }, { "epoch": 0.09505317348629247, "grad_norm": 1.9585718620464103, "learning_rate": 1.9779070326217074e-05, "loss": 0.5545, "step": 2668 }, { "epoch": 0.09508880061278657, "grad_norm": 2.003130461298475, "learning_rate": 1.9778829041868687e-05, "loss": 0.6121, "step": 2669 }, { "epoch": 0.09512442773928069, "grad_norm": 1.3522127261922885, "learning_rate": 1.9778587627308357e-05, "loss": 0.4953, "step": 2670 }, { "epoch": 0.0951600548657748, "grad_norm": 2.2211012446133913, "learning_rate": 1.9778346082539298e-05, "loss": 0.6516, "step": 2671 }, { "epoch": 0.09519568199226891, "grad_norm": 2.9331337469108765, "learning_rate": 1.9778104407564726e-05, "loss": 0.6635, "step": 2672 }, { "epoch": 0.09523130911876303, "grad_norm": 1.397930871290027, "learning_rate": 1.9777862602387862e-05, "loss": 0.5349, "step": 2673 }, { "epoch": 0.09526693624525714, "grad_norm": 2.7529821903151075, "learning_rate": 1.977762066701192e-05, "loss": 0.6573, "step": 2674 }, { "epoch": 0.09530256337175125, "grad_norm": 1.4460584023118785, "learning_rate": 1.9777378601440126e-05, "loss": 0.6082, "step": 2675 }, { "epoch": 0.09533819049824536, "grad_norm": 1.3769208307543772, "learning_rate": 1.97771364056757e-05, "loss": 0.5625, "step": 2676 }, { "epoch": 0.09537381762473948, "grad_norm": 2.7914691265880127, "learning_rate": 1.977689407972187e-05, "loss": 0.6138, "step": 2677 }, { "epoch": 0.09540944475123359, "grad_norm": 1.4903874192424091, "learning_rate": 1.977665162358186e-05, "loss": 0.4856, "step": 2678 }, { "epoch": 0.0954450718777277, "grad_norm": 1.4244689410023146, "learning_rate": 1.9776409037258903e-05, "loss": 0.5645, "step": 2679 }, { "epoch": 0.09548069900422182, "grad_norm": 1.4330301985395957, "learning_rate": 1.9776166320756228e-05, "loss": 0.561, "step": 2680 }, { "epoch": 0.09551632613071592, "grad_norm": 1.3536759786777182, "learning_rate": 1.9775923474077064e-05, "loss": 0.5523, "step": 2681 }, { "epoch": 0.09555195325721004, "grad_norm": 2.139380094974181, "learning_rate": 1.9775680497224643e-05, "loss": 0.6262, "step": 2682 }, { "epoch": 0.09558758038370416, "grad_norm": 2.008993512218728, "learning_rate": 1.9775437390202205e-05, "loss": 0.6811, "step": 2683 }, { "epoch": 0.09562320751019826, "grad_norm": 1.6594565259425402, "learning_rate": 1.9775194153012985e-05, "loss": 0.5305, "step": 2684 }, { "epoch": 0.09565883463669238, "grad_norm": 6.069813537125953, "learning_rate": 1.9774950785660226e-05, "loss": 0.6291, "step": 2685 }, { "epoch": 0.0956944617631865, "grad_norm": 1.720419062704256, "learning_rate": 1.9774707288147163e-05, "loss": 0.5521, "step": 2686 }, { "epoch": 0.0957300888896806, "grad_norm": 1.8925150372677324, "learning_rate": 1.9774463660477038e-05, "loss": 0.6164, "step": 2687 }, { "epoch": 0.09576571601617472, "grad_norm": 4.589354414549632, "learning_rate": 1.97742199026531e-05, "loss": 0.6778, "step": 2688 }, { "epoch": 0.09580134314266883, "grad_norm": 2.5609176668952887, "learning_rate": 1.977397601467859e-05, "loss": 0.5441, "step": 2689 }, { "epoch": 0.09583697026916294, "grad_norm": 1.6810179214140244, "learning_rate": 1.9773731996556758e-05, "loss": 0.6318, "step": 2690 }, { "epoch": 0.09587259739565705, "grad_norm": 4.03681506687367, "learning_rate": 1.9773487848290856e-05, "loss": 0.57, "step": 2691 }, { "epoch": 0.09590822452215117, "grad_norm": 2.0379369259679065, "learning_rate": 1.977324356988413e-05, "loss": 0.5212, "step": 2692 }, { "epoch": 0.09594385164864527, "grad_norm": 1.4650202319233965, "learning_rate": 1.9772999161339833e-05, "loss": 0.538, "step": 2693 }, { "epoch": 0.09597947877513939, "grad_norm": 1.4591908180564201, "learning_rate": 1.9772754622661223e-05, "loss": 0.5732, "step": 2694 }, { "epoch": 0.09601510590163351, "grad_norm": 2.1314638461464672, "learning_rate": 1.9772509953851556e-05, "loss": 0.6238, "step": 2695 }, { "epoch": 0.09605073302812761, "grad_norm": 2.762460013485686, "learning_rate": 1.9772265154914087e-05, "loss": 0.6686, "step": 2696 }, { "epoch": 0.09608636015462173, "grad_norm": 1.9029179641181304, "learning_rate": 1.9772020225852077e-05, "loss": 0.582, "step": 2697 }, { "epoch": 0.09612198728111584, "grad_norm": 1.9204486265724643, "learning_rate": 1.9771775166668787e-05, "loss": 0.6653, "step": 2698 }, { "epoch": 0.09615761440760995, "grad_norm": 1.33356447329936, "learning_rate": 1.977152997736748e-05, "loss": 0.4971, "step": 2699 }, { "epoch": 0.09619324153410407, "grad_norm": 1.8672737190778301, "learning_rate": 1.977128465795142e-05, "loss": 0.5396, "step": 2700 }, { "epoch": 0.09622886866059818, "grad_norm": 3.130896358498028, "learning_rate": 1.9771039208423882e-05, "loss": 0.5996, "step": 2701 }, { "epoch": 0.09626449578709229, "grad_norm": 1.6038328723351214, "learning_rate": 1.9770793628788123e-05, "loss": 0.4977, "step": 2702 }, { "epoch": 0.0963001229135864, "grad_norm": 1.4419837937705788, "learning_rate": 1.9770547919047416e-05, "loss": 0.591, "step": 2703 }, { "epoch": 0.09633575004008052, "grad_norm": 1.3849727721784824, "learning_rate": 1.9770302079205036e-05, "loss": 0.537, "step": 2704 }, { "epoch": 0.09637137716657462, "grad_norm": 1.8565842594991881, "learning_rate": 1.9770056109264254e-05, "loss": 0.6169, "step": 2705 }, { "epoch": 0.09640700429306874, "grad_norm": 1.6282519934320336, "learning_rate": 1.976981000922835e-05, "loss": 0.562, "step": 2706 }, { "epoch": 0.09644263141956286, "grad_norm": 1.5657873507882685, "learning_rate": 1.976956377910059e-05, "loss": 0.5353, "step": 2707 }, { "epoch": 0.09647825854605697, "grad_norm": 1.7984163400031, "learning_rate": 1.9769317418884264e-05, "loss": 0.5376, "step": 2708 }, { "epoch": 0.09651388567255108, "grad_norm": 2.917528713028027, "learning_rate": 1.9769070928582648e-05, "loss": 0.6099, "step": 2709 }, { "epoch": 0.0965495127990452, "grad_norm": 1.3662828642026508, "learning_rate": 1.9768824308199025e-05, "loss": 0.4798, "step": 2710 }, { "epoch": 0.09658513992553931, "grad_norm": 1.6807415002230714, "learning_rate": 1.9768577557736677e-05, "loss": 0.5405, "step": 2711 }, { "epoch": 0.09662076705203342, "grad_norm": 1.2623958182119794, "learning_rate": 1.976833067719889e-05, "loss": 0.5138, "step": 2712 }, { "epoch": 0.09665639417852753, "grad_norm": 2.252821929164554, "learning_rate": 1.9768083666588954e-05, "loss": 0.6952, "step": 2713 }, { "epoch": 0.09669202130502165, "grad_norm": 1.4604776940236504, "learning_rate": 1.9767836525910156e-05, "loss": 0.5809, "step": 2714 }, { "epoch": 0.09672764843151575, "grad_norm": 1.406438155659452, "learning_rate": 1.9767589255165785e-05, "loss": 0.5762, "step": 2715 }, { "epoch": 0.09676327555800987, "grad_norm": 1.6367355584954708, "learning_rate": 1.976734185435914e-05, "loss": 0.6418, "step": 2716 }, { "epoch": 0.09679890268450399, "grad_norm": 1.4271103326438996, "learning_rate": 1.976709432349351e-05, "loss": 0.5058, "step": 2717 }, { "epoch": 0.09683452981099809, "grad_norm": 1.4556939090626408, "learning_rate": 1.976684666257219e-05, "loss": 0.5023, "step": 2718 }, { "epoch": 0.09687015693749221, "grad_norm": 1.3460003096685742, "learning_rate": 1.9766598871598484e-05, "loss": 0.5425, "step": 2719 }, { "epoch": 0.09690578406398632, "grad_norm": 1.284459587771661, "learning_rate": 1.9766350950575685e-05, "loss": 0.5495, "step": 2720 }, { "epoch": 0.09694141119048043, "grad_norm": 1.726649354836342, "learning_rate": 1.9766102899507098e-05, "loss": 0.6205, "step": 2721 }, { "epoch": 0.09697703831697455, "grad_norm": 2.0612666171264347, "learning_rate": 1.9765854718396026e-05, "loss": 0.5962, "step": 2722 }, { "epoch": 0.09701266544346866, "grad_norm": 1.8857807895575, "learning_rate": 1.9765606407245766e-05, "loss": 0.5818, "step": 2723 }, { "epoch": 0.09704829256996277, "grad_norm": 1.8801014943117844, "learning_rate": 1.9765357966059638e-05, "loss": 0.6242, "step": 2724 }, { "epoch": 0.09708391969645688, "grad_norm": 2.3872970328129064, "learning_rate": 1.976510939484094e-05, "loss": 0.5763, "step": 2725 }, { "epoch": 0.097119546822951, "grad_norm": 1.3785173928805254, "learning_rate": 1.9764860693592988e-05, "loss": 0.5367, "step": 2726 }, { "epoch": 0.0971551739494451, "grad_norm": 1.2658265733931204, "learning_rate": 1.9764611862319087e-05, "loss": 0.5015, "step": 2727 }, { "epoch": 0.09719080107593922, "grad_norm": 2.0529556212921176, "learning_rate": 1.9764362901022556e-05, "loss": 0.6027, "step": 2728 }, { "epoch": 0.09722642820243334, "grad_norm": 1.9613718681541148, "learning_rate": 1.9764113809706707e-05, "loss": 0.6596, "step": 2729 }, { "epoch": 0.09726205532892744, "grad_norm": 3.185499909404259, "learning_rate": 1.976386458837486e-05, "loss": 0.6248, "step": 2730 }, { "epoch": 0.09729768245542156, "grad_norm": 1.4069591413313702, "learning_rate": 1.9763615237030328e-05, "loss": 0.5786, "step": 2731 }, { "epoch": 0.09733330958191568, "grad_norm": 1.8157119274213915, "learning_rate": 1.9763365755676438e-05, "loss": 0.6134, "step": 2732 }, { "epoch": 0.09736893670840978, "grad_norm": 1.727642744169814, "learning_rate": 1.9763116144316506e-05, "loss": 0.5478, "step": 2733 }, { "epoch": 0.0974045638349039, "grad_norm": 1.4746435718477673, "learning_rate": 1.976286640295386e-05, "loss": 0.5647, "step": 2734 }, { "epoch": 0.09744019096139801, "grad_norm": 1.477813359559155, "learning_rate": 1.9762616531591826e-05, "loss": 0.5561, "step": 2735 }, { "epoch": 0.09747581808789212, "grad_norm": 2.270720116677809, "learning_rate": 1.9762366530233724e-05, "loss": 0.6767, "step": 2736 }, { "epoch": 0.09751144521438623, "grad_norm": 1.3659161639708115, "learning_rate": 1.9762116398882894e-05, "loss": 0.5499, "step": 2737 }, { "epoch": 0.09754707234088035, "grad_norm": 2.035598670576744, "learning_rate": 1.976186613754266e-05, "loss": 0.611, "step": 2738 }, { "epoch": 0.09758269946737445, "grad_norm": 1.2941812449345702, "learning_rate": 1.976161574621635e-05, "loss": 0.5572, "step": 2739 }, { "epoch": 0.09761832659386857, "grad_norm": 2.3163078022540726, "learning_rate": 1.9761365224907306e-05, "loss": 0.5713, "step": 2740 }, { "epoch": 0.09765395372036269, "grad_norm": 1.870733574960614, "learning_rate": 1.976111457361886e-05, "loss": 0.6601, "step": 2741 }, { "epoch": 0.09768958084685679, "grad_norm": 1.9243147854508076, "learning_rate": 1.9760863792354356e-05, "loss": 0.683, "step": 2742 }, { "epoch": 0.09772520797335091, "grad_norm": 2.4525596474937443, "learning_rate": 1.9760612881117124e-05, "loss": 0.7233, "step": 2743 }, { "epoch": 0.09776083509984503, "grad_norm": 1.3281462459866238, "learning_rate": 1.9760361839910513e-05, "loss": 0.5502, "step": 2744 }, { "epoch": 0.09779646222633913, "grad_norm": 1.678001900885201, "learning_rate": 1.976011066873786e-05, "loss": 0.544, "step": 2745 }, { "epoch": 0.09783208935283325, "grad_norm": 1.3079474483166633, "learning_rate": 1.975985936760251e-05, "loss": 0.5479, "step": 2746 }, { "epoch": 0.09786771647932736, "grad_norm": 1.6398537956098864, "learning_rate": 1.975960793650781e-05, "loss": 0.6225, "step": 2747 }, { "epoch": 0.09790334360582147, "grad_norm": 1.535174106781017, "learning_rate": 1.9759356375457114e-05, "loss": 0.5298, "step": 2748 }, { "epoch": 0.09793897073231558, "grad_norm": 1.4372977951687844, "learning_rate": 1.975910468445376e-05, "loss": 0.5482, "step": 2749 }, { "epoch": 0.0979745978588097, "grad_norm": 2.0857773872683203, "learning_rate": 1.9758852863501108e-05, "loss": 0.5628, "step": 2750 }, { "epoch": 0.0980102249853038, "grad_norm": 2.574358401662527, "learning_rate": 1.975860091260251e-05, "loss": 0.6591, "step": 2751 }, { "epoch": 0.09804585211179792, "grad_norm": 1.223063356565086, "learning_rate": 1.975834883176132e-05, "loss": 0.488, "step": 2752 }, { "epoch": 0.09808147923829204, "grad_norm": 2.403712450809696, "learning_rate": 1.9758096620980896e-05, "loss": 0.5704, "step": 2753 }, { "epoch": 0.09811710636478614, "grad_norm": 2.0330981034870828, "learning_rate": 1.9757844280264594e-05, "loss": 0.6147, "step": 2754 }, { "epoch": 0.09815273349128026, "grad_norm": 2.191473447218173, "learning_rate": 1.9757591809615772e-05, "loss": 0.6043, "step": 2755 }, { "epoch": 0.09818836061777438, "grad_norm": 1.6415562225921454, "learning_rate": 1.97573392090378e-05, "loss": 0.5793, "step": 2756 }, { "epoch": 0.09822398774426849, "grad_norm": 2.4537449025223013, "learning_rate": 1.975708647853403e-05, "loss": 0.7097, "step": 2757 }, { "epoch": 0.0982596148707626, "grad_norm": 1.4987562285263587, "learning_rate": 1.975683361810784e-05, "loss": 0.5134, "step": 2758 }, { "epoch": 0.09829524199725671, "grad_norm": 1.3568698959274963, "learning_rate": 1.9756580627762587e-05, "loss": 0.5621, "step": 2759 }, { "epoch": 0.09833086912375083, "grad_norm": 1.1963115301163063, "learning_rate": 1.9756327507501647e-05, "loss": 0.4708, "step": 2760 }, { "epoch": 0.09836649625024493, "grad_norm": 1.382090035466048, "learning_rate": 1.9756074257328387e-05, "loss": 0.5562, "step": 2761 }, { "epoch": 0.09840212337673905, "grad_norm": 1.4650425834669627, "learning_rate": 1.9755820877246175e-05, "loss": 0.5558, "step": 2762 }, { "epoch": 0.09843775050323317, "grad_norm": 1.8334839076051923, "learning_rate": 1.975556736725839e-05, "loss": 0.6032, "step": 2763 }, { "epoch": 0.09847337762972727, "grad_norm": 1.9286806961695033, "learning_rate": 1.9755313727368408e-05, "loss": 0.6255, "step": 2764 }, { "epoch": 0.09850900475622139, "grad_norm": 1.6652099449364102, "learning_rate": 1.9755059957579605e-05, "loss": 0.5562, "step": 2765 }, { "epoch": 0.0985446318827155, "grad_norm": 1.1963539963374894, "learning_rate": 1.975480605789536e-05, "loss": 0.5214, "step": 2766 }, { "epoch": 0.09858025900920961, "grad_norm": 1.2946528573855407, "learning_rate": 1.9754552028319057e-05, "loss": 0.5511, "step": 2767 }, { "epoch": 0.09861588613570373, "grad_norm": 1.4158848289008668, "learning_rate": 1.9754297868854075e-05, "loss": 0.5384, "step": 2768 }, { "epoch": 0.09865151326219784, "grad_norm": 1.9915623240289084, "learning_rate": 1.9754043579503793e-05, "loss": 0.6382, "step": 2769 }, { "epoch": 0.09868714038869195, "grad_norm": 1.6129950438646397, "learning_rate": 1.975378916027161e-05, "loss": 0.5341, "step": 2770 }, { "epoch": 0.09872276751518606, "grad_norm": 1.2559635627611125, "learning_rate": 1.97535346111609e-05, "loss": 0.4829, "step": 2771 }, { "epoch": 0.09875839464168018, "grad_norm": 1.3595771989101588, "learning_rate": 1.9753279932175068e-05, "loss": 0.5252, "step": 2772 }, { "epoch": 0.09879402176817428, "grad_norm": 1.2749173460068637, "learning_rate": 1.975302512331749e-05, "loss": 0.5515, "step": 2773 }, { "epoch": 0.0988296488946684, "grad_norm": 3.1325280215656086, "learning_rate": 1.975277018459157e-05, "loss": 0.6548, "step": 2774 }, { "epoch": 0.09886527602116252, "grad_norm": 1.2989375479454859, "learning_rate": 1.9752515116000693e-05, "loss": 0.5808, "step": 2775 }, { "epoch": 0.09890090314765662, "grad_norm": 1.3135285428757246, "learning_rate": 1.9752259917548266e-05, "loss": 0.5102, "step": 2776 }, { "epoch": 0.09893653027415074, "grad_norm": 1.3970013488402506, "learning_rate": 1.9752004589237678e-05, "loss": 0.5605, "step": 2777 }, { "epoch": 0.09897215740064486, "grad_norm": 1.2650124521096269, "learning_rate": 1.9751749131072335e-05, "loss": 0.5266, "step": 2778 }, { "epoch": 0.09900778452713896, "grad_norm": 1.349075387847801, "learning_rate": 1.9751493543055634e-05, "loss": 0.5356, "step": 2779 }, { "epoch": 0.09904341165363308, "grad_norm": 1.3250969977696054, "learning_rate": 1.9751237825190982e-05, "loss": 0.5387, "step": 2780 }, { "epoch": 0.0990790387801272, "grad_norm": 1.4868675196215329, "learning_rate": 1.9750981977481783e-05, "loss": 0.535, "step": 2781 }, { "epoch": 0.0991146659066213, "grad_norm": 1.255713617230585, "learning_rate": 1.9750725999931443e-05, "loss": 0.5627, "step": 2782 }, { "epoch": 0.09915029303311541, "grad_norm": 2.1267920859383334, "learning_rate": 1.9750469892543372e-05, "loss": 0.6585, "step": 2783 }, { "epoch": 0.09918592015960953, "grad_norm": 2.201557051720242, "learning_rate": 1.9750213655320977e-05, "loss": 0.6148, "step": 2784 }, { "epoch": 0.09922154728610363, "grad_norm": 1.304178788913647, "learning_rate": 1.974995728826767e-05, "loss": 0.5365, "step": 2785 }, { "epoch": 0.09925717441259775, "grad_norm": 1.868797423674463, "learning_rate": 1.9749700791386872e-05, "loss": 0.7332, "step": 2786 }, { "epoch": 0.09929280153909187, "grad_norm": 1.3752147279582998, "learning_rate": 1.974944416468199e-05, "loss": 0.5193, "step": 2787 }, { "epoch": 0.09932842866558597, "grad_norm": 1.3615373063267349, "learning_rate": 1.9749187408156448e-05, "loss": 0.553, "step": 2788 }, { "epoch": 0.09936405579208009, "grad_norm": 2.4502169068168773, "learning_rate": 1.974893052181366e-05, "loss": 0.6625, "step": 2789 }, { "epoch": 0.0993996829185742, "grad_norm": 1.3688956356426436, "learning_rate": 1.9748673505657045e-05, "loss": 0.4878, "step": 2790 }, { "epoch": 0.09943531004506831, "grad_norm": 1.2989361700346043, "learning_rate": 1.974841635969003e-05, "loss": 0.4744, "step": 2791 }, { "epoch": 0.09947093717156243, "grad_norm": 1.3529521476320814, "learning_rate": 1.9748159083916038e-05, "loss": 0.5631, "step": 2792 }, { "epoch": 0.09950656429805654, "grad_norm": 1.3510237646406542, "learning_rate": 1.9747901678338496e-05, "loss": 0.5528, "step": 2793 }, { "epoch": 0.09954219142455065, "grad_norm": 3.9836648993798964, "learning_rate": 1.974764414296083e-05, "loss": 0.6605, "step": 2794 }, { "epoch": 0.09957781855104476, "grad_norm": 1.321620127218296, "learning_rate": 1.9747386477786466e-05, "loss": 0.5409, "step": 2795 }, { "epoch": 0.09961344567753888, "grad_norm": 2.2110596384288472, "learning_rate": 1.974712868281884e-05, "loss": 0.6744, "step": 2796 }, { "epoch": 0.09964907280403298, "grad_norm": 1.9919667258469733, "learning_rate": 1.974687075806138e-05, "loss": 0.5631, "step": 2797 }, { "epoch": 0.0996846999305271, "grad_norm": 1.2289177567539875, "learning_rate": 1.9746612703517524e-05, "loss": 0.5141, "step": 2798 }, { "epoch": 0.09972032705702122, "grad_norm": 1.4683504855729679, "learning_rate": 1.974635451919071e-05, "loss": 0.5408, "step": 2799 }, { "epoch": 0.09975595418351532, "grad_norm": 2.5085739809898326, "learning_rate": 1.9746096205084372e-05, "loss": 0.6128, "step": 2800 }, { "epoch": 0.09979158131000944, "grad_norm": 1.824616598456682, "learning_rate": 1.9745837761201948e-05, "loss": 0.6595, "step": 2801 }, { "epoch": 0.09982720843650356, "grad_norm": 1.977078509122343, "learning_rate": 1.974557918754689e-05, "loss": 0.6644, "step": 2802 }, { "epoch": 0.09986283556299767, "grad_norm": 2.108525372998671, "learning_rate": 1.9745320484122626e-05, "loss": 0.6897, "step": 2803 }, { "epoch": 0.09989846268949178, "grad_norm": 2.67147881268546, "learning_rate": 1.974506165093261e-05, "loss": 0.6614, "step": 2804 }, { "epoch": 0.0999340898159859, "grad_norm": 1.3088083133294952, "learning_rate": 1.9744802687980285e-05, "loss": 0.5562, "step": 2805 }, { "epoch": 0.09996971694248001, "grad_norm": 1.7093788285709517, "learning_rate": 1.974454359526911e-05, "loss": 0.5707, "step": 2806 }, { "epoch": 0.10000534406897411, "grad_norm": 1.3910620616139338, "learning_rate": 1.9744284372802517e-05, "loss": 0.5113, "step": 2807 }, { "epoch": 0.10004097119546823, "grad_norm": 1.3126512307267153, "learning_rate": 1.974402502058397e-05, "loss": 0.523, "step": 2808 }, { "epoch": 0.10007659832196235, "grad_norm": 1.9436966999200596, "learning_rate": 1.9743765538616918e-05, "loss": 0.6044, "step": 2809 }, { "epoch": 0.10011222544845645, "grad_norm": 2.014454923237537, "learning_rate": 1.9743505926904815e-05, "loss": 0.613, "step": 2810 }, { "epoch": 0.10014785257495057, "grad_norm": 1.2586355981193738, "learning_rate": 1.9743246185451125e-05, "loss": 0.5358, "step": 2811 }, { "epoch": 0.10018347970144469, "grad_norm": 2.1830539759319745, "learning_rate": 1.97429863142593e-05, "loss": 0.6426, "step": 2812 }, { "epoch": 0.10021910682793879, "grad_norm": 1.380186885043208, "learning_rate": 1.97427263133328e-05, "loss": 0.5453, "step": 2813 }, { "epoch": 0.1002547339544329, "grad_norm": 1.6037833321706134, "learning_rate": 1.9742466182675093e-05, "loss": 0.5019, "step": 2814 }, { "epoch": 0.10029036108092702, "grad_norm": 1.6289397672498491, "learning_rate": 1.974220592228964e-05, "loss": 0.5712, "step": 2815 }, { "epoch": 0.10032598820742113, "grad_norm": 2.1832555616602036, "learning_rate": 1.97419455321799e-05, "loss": 0.6462, "step": 2816 }, { "epoch": 0.10036161533391524, "grad_norm": 1.8787840844283286, "learning_rate": 1.9741685012349347e-05, "loss": 0.5799, "step": 2817 }, { "epoch": 0.10039724246040936, "grad_norm": 1.2983856165963048, "learning_rate": 1.9741424362801453e-05, "loss": 0.581, "step": 2818 }, { "epoch": 0.10043286958690346, "grad_norm": 1.3293929971328942, "learning_rate": 1.9741163583539685e-05, "loss": 0.5534, "step": 2819 }, { "epoch": 0.10046849671339758, "grad_norm": 1.3966376370031321, "learning_rate": 1.974090267456751e-05, "loss": 0.5804, "step": 2820 }, { "epoch": 0.1005041238398917, "grad_norm": 1.3478205659260198, "learning_rate": 1.974064163588841e-05, "loss": 0.4955, "step": 2821 }, { "epoch": 0.1005397509663858, "grad_norm": 1.4794756878508297, "learning_rate": 1.9740380467505863e-05, "loss": 0.4888, "step": 2822 }, { "epoch": 0.10057537809287992, "grad_norm": 1.3614451025664862, "learning_rate": 1.9740119169423337e-05, "loss": 0.5587, "step": 2823 }, { "epoch": 0.10061100521937404, "grad_norm": 2.2396804605127354, "learning_rate": 1.9739857741644318e-05, "loss": 0.5799, "step": 2824 }, { "epoch": 0.10064663234586814, "grad_norm": 1.6247562253780834, "learning_rate": 1.9739596184172284e-05, "loss": 0.5587, "step": 2825 }, { "epoch": 0.10068225947236226, "grad_norm": 1.4077446026129443, "learning_rate": 1.973933449701072e-05, "loss": 0.5272, "step": 2826 }, { "epoch": 0.10071788659885637, "grad_norm": 1.266073641822671, "learning_rate": 1.973907268016311e-05, "loss": 0.4932, "step": 2827 }, { "epoch": 0.10075351372535048, "grad_norm": 1.1728175339396307, "learning_rate": 1.9738810733632937e-05, "loss": 0.557, "step": 2828 }, { "epoch": 0.1007891408518446, "grad_norm": 1.2323640418658353, "learning_rate": 1.9738548657423696e-05, "loss": 0.5266, "step": 2829 }, { "epoch": 0.10082476797833871, "grad_norm": 1.9590001878069798, "learning_rate": 1.9738286451538874e-05, "loss": 0.6762, "step": 2830 }, { "epoch": 0.10086039510483281, "grad_norm": 2.189822319979819, "learning_rate": 1.9738024115981957e-05, "loss": 0.6675, "step": 2831 }, { "epoch": 0.10089602223132693, "grad_norm": 1.3498686607452677, "learning_rate": 1.9737761650756446e-05, "loss": 0.5286, "step": 2832 }, { "epoch": 0.10093164935782105, "grad_norm": 1.2313400300616055, "learning_rate": 1.973749905586583e-05, "loss": 0.4685, "step": 2833 }, { "epoch": 0.10096727648431515, "grad_norm": 1.229971634116217, "learning_rate": 1.973723633131361e-05, "loss": 0.4893, "step": 2834 }, { "epoch": 0.10100290361080927, "grad_norm": 2.68908577623413, "learning_rate": 1.973697347710328e-05, "loss": 0.6639, "step": 2835 }, { "epoch": 0.10103853073730339, "grad_norm": 1.1801496667278661, "learning_rate": 1.9736710493238345e-05, "loss": 0.5115, "step": 2836 }, { "epoch": 0.10107415786379749, "grad_norm": 1.5143436133311095, "learning_rate": 1.97364473797223e-05, "loss": 0.5489, "step": 2837 }, { "epoch": 0.1011097849902916, "grad_norm": 1.4612245785627376, "learning_rate": 1.9736184136558657e-05, "loss": 0.5265, "step": 2838 }, { "epoch": 0.10114541211678572, "grad_norm": 1.1157665705972655, "learning_rate": 1.9735920763750916e-05, "loss": 0.4618, "step": 2839 }, { "epoch": 0.10118103924327983, "grad_norm": 1.2191673347772294, "learning_rate": 1.9735657261302587e-05, "loss": 0.5505, "step": 2840 }, { "epoch": 0.10121666636977394, "grad_norm": 1.833147850284087, "learning_rate": 1.973539362921717e-05, "loss": 0.6778, "step": 2841 }, { "epoch": 0.10125229349626806, "grad_norm": 2.1609315500814374, "learning_rate": 1.9735129867498188e-05, "loss": 0.5145, "step": 2842 }, { "epoch": 0.10128792062276216, "grad_norm": 1.2515750431817485, "learning_rate": 1.9734865976149145e-05, "loss": 0.5088, "step": 2843 }, { "epoch": 0.10132354774925628, "grad_norm": 1.2531486205614137, "learning_rate": 1.973460195517356e-05, "loss": 0.523, "step": 2844 }, { "epoch": 0.1013591748757504, "grad_norm": 2.282223754305505, "learning_rate": 1.9734337804574943e-05, "loss": 0.6648, "step": 2845 }, { "epoch": 0.1013948020022445, "grad_norm": 1.435644835204534, "learning_rate": 1.9734073524356817e-05, "loss": 0.5953, "step": 2846 }, { "epoch": 0.10143042912873862, "grad_norm": 2.0179932837743677, "learning_rate": 1.9733809114522695e-05, "loss": 0.6248, "step": 2847 }, { "epoch": 0.10146605625523274, "grad_norm": 1.3278516429869929, "learning_rate": 1.9733544575076103e-05, "loss": 0.5938, "step": 2848 }, { "epoch": 0.10150168338172684, "grad_norm": 1.815089237473661, "learning_rate": 1.9733279906020558e-05, "loss": 0.6599, "step": 2849 }, { "epoch": 0.10153731050822096, "grad_norm": 1.2538752571891667, "learning_rate": 1.9733015107359593e-05, "loss": 0.456, "step": 2850 }, { "epoch": 0.10157293763471507, "grad_norm": 2.3302106657511157, "learning_rate": 1.9732750179096726e-05, "loss": 0.5932, "step": 2851 }, { "epoch": 0.10160856476120919, "grad_norm": 1.273702617309784, "learning_rate": 1.973248512123549e-05, "loss": 0.5569, "step": 2852 }, { "epoch": 0.1016441918877033, "grad_norm": 1.2456757837723806, "learning_rate": 1.973221993377941e-05, "loss": 0.5386, "step": 2853 }, { "epoch": 0.10167981901419741, "grad_norm": 2.028879697468395, "learning_rate": 1.9731954616732022e-05, "loss": 0.678, "step": 2854 }, { "epoch": 0.10171544614069153, "grad_norm": 1.8612130777029656, "learning_rate": 1.9731689170096853e-05, "loss": 0.7227, "step": 2855 }, { "epoch": 0.10175107326718563, "grad_norm": 1.7982028380851396, "learning_rate": 1.973142359387744e-05, "loss": 0.6201, "step": 2856 }, { "epoch": 0.10178670039367975, "grad_norm": 1.3988508505084478, "learning_rate": 1.9731157888077322e-05, "loss": 0.5321, "step": 2857 }, { "epoch": 0.10182232752017387, "grad_norm": 1.348442334472639, "learning_rate": 1.9730892052700034e-05, "loss": 0.5237, "step": 2858 }, { "epoch": 0.10185795464666797, "grad_norm": 2.1123150480646933, "learning_rate": 1.973062608774912e-05, "loss": 0.6723, "step": 2859 }, { "epoch": 0.10189358177316209, "grad_norm": 1.8549440148379073, "learning_rate": 1.9730359993228114e-05, "loss": 0.6699, "step": 2860 }, { "epoch": 0.1019292088996562, "grad_norm": 1.9618401660268525, "learning_rate": 1.9730093769140567e-05, "loss": 0.6312, "step": 2861 }, { "epoch": 0.10196483602615031, "grad_norm": 1.3212171147729785, "learning_rate": 1.9729827415490016e-05, "loss": 0.5207, "step": 2862 }, { "epoch": 0.10200046315264442, "grad_norm": 1.1320314971323668, "learning_rate": 1.9729560932280016e-05, "loss": 0.4627, "step": 2863 }, { "epoch": 0.10203609027913854, "grad_norm": 2.2193220621452445, "learning_rate": 1.972929431951411e-05, "loss": 0.6078, "step": 2864 }, { "epoch": 0.10207171740563264, "grad_norm": 1.3862047764771104, "learning_rate": 1.972902757719585e-05, "loss": 0.5547, "step": 2865 }, { "epoch": 0.10210734453212676, "grad_norm": 3.264272509143845, "learning_rate": 1.972876070532879e-05, "loss": 0.6136, "step": 2866 }, { "epoch": 0.10214297165862088, "grad_norm": 1.9045378252700342, "learning_rate": 1.972849370391648e-05, "loss": 0.712, "step": 2867 }, { "epoch": 0.10217859878511498, "grad_norm": 2.4461242658860085, "learning_rate": 1.9728226572962474e-05, "loss": 0.6219, "step": 2868 }, { "epoch": 0.1022142259116091, "grad_norm": 1.328984205593817, "learning_rate": 1.9727959312470335e-05, "loss": 0.5222, "step": 2869 }, { "epoch": 0.10224985303810322, "grad_norm": 1.908603632410569, "learning_rate": 1.9727691922443616e-05, "loss": 0.6224, "step": 2870 }, { "epoch": 0.10228548016459732, "grad_norm": 1.4126500024015294, "learning_rate": 1.972742440288588e-05, "loss": 0.5168, "step": 2871 }, { "epoch": 0.10232110729109144, "grad_norm": 2.7685208192873496, "learning_rate": 1.972715675380069e-05, "loss": 0.5504, "step": 2872 }, { "epoch": 0.10235673441758555, "grad_norm": 1.246505103557168, "learning_rate": 1.9726888975191607e-05, "loss": 0.5382, "step": 2873 }, { "epoch": 0.10239236154407966, "grad_norm": 1.3764533574218671, "learning_rate": 1.9726621067062203e-05, "loss": 0.5337, "step": 2874 }, { "epoch": 0.10242798867057377, "grad_norm": 1.8501955794742857, "learning_rate": 1.9726353029416037e-05, "loss": 0.6416, "step": 2875 }, { "epoch": 0.10246361579706789, "grad_norm": 2.8597362391295222, "learning_rate": 1.972608486225668e-05, "loss": 0.7334, "step": 2876 }, { "epoch": 0.102499242923562, "grad_norm": 1.2795166135253198, "learning_rate": 1.972581656558771e-05, "loss": 0.5432, "step": 2877 }, { "epoch": 0.10253487005005611, "grad_norm": 1.8844774442332586, "learning_rate": 1.9725548139412693e-05, "loss": 0.6974, "step": 2878 }, { "epoch": 0.10257049717655023, "grad_norm": 1.2924491986241358, "learning_rate": 1.9725279583735203e-05, "loss": 0.5034, "step": 2879 }, { "epoch": 0.10260612430304433, "grad_norm": 1.3825330653321348, "learning_rate": 1.972501089855882e-05, "loss": 0.5505, "step": 2880 }, { "epoch": 0.10264175142953845, "grad_norm": 1.689752909990971, "learning_rate": 1.9724742083887117e-05, "loss": 0.6605, "step": 2881 }, { "epoch": 0.10267737855603257, "grad_norm": 1.636340534540158, "learning_rate": 1.9724473139723682e-05, "loss": 0.5706, "step": 2882 }, { "epoch": 0.10271300568252667, "grad_norm": 1.7443243637525105, "learning_rate": 1.9724204066072084e-05, "loss": 0.5867, "step": 2883 }, { "epoch": 0.10274863280902079, "grad_norm": 1.27264408602609, "learning_rate": 1.9723934862935912e-05, "loss": 0.5369, "step": 2884 }, { "epoch": 0.1027842599355149, "grad_norm": 1.7884004620999652, "learning_rate": 1.9723665530318753e-05, "loss": 0.6085, "step": 2885 }, { "epoch": 0.10281988706200901, "grad_norm": 2.156527562122695, "learning_rate": 1.972339606822419e-05, "loss": 0.6689, "step": 2886 }, { "epoch": 0.10285551418850312, "grad_norm": 2.074187494505538, "learning_rate": 1.9723126476655813e-05, "loss": 0.6107, "step": 2887 }, { "epoch": 0.10289114131499724, "grad_norm": 1.2460100199358413, "learning_rate": 1.972285675561721e-05, "loss": 0.5096, "step": 2888 }, { "epoch": 0.10292676844149135, "grad_norm": 1.2516507593132675, "learning_rate": 1.9722586905111975e-05, "loss": 0.5061, "step": 2889 }, { "epoch": 0.10296239556798546, "grad_norm": 1.2047555610854304, "learning_rate": 1.97223169251437e-05, "loss": 0.5253, "step": 2890 }, { "epoch": 0.10299802269447958, "grad_norm": 2.369075849899234, "learning_rate": 1.9722046815715975e-05, "loss": 0.6622, "step": 2891 }, { "epoch": 0.10303364982097368, "grad_norm": 1.377945510988541, "learning_rate": 1.9721776576832408e-05, "loss": 0.5585, "step": 2892 }, { "epoch": 0.1030692769474678, "grad_norm": 1.7820818649713268, "learning_rate": 1.9721506208496585e-05, "loss": 0.6165, "step": 2893 }, { "epoch": 0.10310490407396192, "grad_norm": 1.6367947752058205, "learning_rate": 1.9721235710712118e-05, "loss": 0.6125, "step": 2894 }, { "epoch": 0.10314053120045602, "grad_norm": 1.4333064334745842, "learning_rate": 1.9720965083482596e-05, "loss": 0.5552, "step": 2895 }, { "epoch": 0.10317615832695014, "grad_norm": 1.3836251681943001, "learning_rate": 1.9720694326811633e-05, "loss": 0.5208, "step": 2896 }, { "epoch": 0.10321178545344425, "grad_norm": 2.5103893382108216, "learning_rate": 1.972042344070283e-05, "loss": 0.6615, "step": 2897 }, { "epoch": 0.10324741257993837, "grad_norm": 1.4124549692361081, "learning_rate": 1.9720152425159792e-05, "loss": 0.5627, "step": 2898 }, { "epoch": 0.10328303970643248, "grad_norm": 1.1330163791525842, "learning_rate": 1.9719881280186138e-05, "loss": 0.4949, "step": 2899 }, { "epoch": 0.10331866683292659, "grad_norm": 1.3228164878779505, "learning_rate": 1.9719610005785466e-05, "loss": 0.5425, "step": 2900 }, { "epoch": 0.10335429395942071, "grad_norm": 2.081874473843628, "learning_rate": 1.9719338601961393e-05, "loss": 0.6305, "step": 2901 }, { "epoch": 0.10338992108591481, "grad_norm": 1.33092640191011, "learning_rate": 1.9719067068717532e-05, "loss": 0.5167, "step": 2902 }, { "epoch": 0.10342554821240893, "grad_norm": 1.8627205620309504, "learning_rate": 1.97187954060575e-05, "loss": 0.6431, "step": 2903 }, { "epoch": 0.10346117533890305, "grad_norm": 1.636453893726158, "learning_rate": 1.9718523613984917e-05, "loss": 0.5886, "step": 2904 }, { "epoch": 0.10349680246539715, "grad_norm": 1.3625680458742684, "learning_rate": 1.9718251692503397e-05, "loss": 0.5516, "step": 2905 }, { "epoch": 0.10353242959189127, "grad_norm": 2.564210715842821, "learning_rate": 1.9717979641616564e-05, "loss": 0.6539, "step": 2906 }, { "epoch": 0.10356805671838538, "grad_norm": 1.6229177109100155, "learning_rate": 1.971770746132804e-05, "loss": 0.6152, "step": 2907 }, { "epoch": 0.10360368384487949, "grad_norm": 1.2818266094479263, "learning_rate": 1.971743515164145e-05, "loss": 0.4762, "step": 2908 }, { "epoch": 0.1036393109713736, "grad_norm": 1.4639157710048003, "learning_rate": 1.9717162712560417e-05, "loss": 0.5012, "step": 2909 }, { "epoch": 0.10367493809786772, "grad_norm": 1.7848399959361139, "learning_rate": 1.9716890144088568e-05, "loss": 0.5737, "step": 2910 }, { "epoch": 0.10371056522436183, "grad_norm": 1.2828629275165135, "learning_rate": 1.9716617446229537e-05, "loss": 0.581, "step": 2911 }, { "epoch": 0.10374619235085594, "grad_norm": 1.9007433397501488, "learning_rate": 1.9716344618986955e-05, "loss": 0.5891, "step": 2912 }, { "epoch": 0.10378181947735006, "grad_norm": 2.117575711733484, "learning_rate": 1.9716071662364454e-05, "loss": 0.6033, "step": 2913 }, { "epoch": 0.10381744660384416, "grad_norm": 1.3794616798483308, "learning_rate": 1.9715798576365663e-05, "loss": 0.5249, "step": 2914 }, { "epoch": 0.10385307373033828, "grad_norm": 2.616049711616082, "learning_rate": 1.9715525360994227e-05, "loss": 0.5516, "step": 2915 }, { "epoch": 0.1038887008568324, "grad_norm": 1.7157029788877665, "learning_rate": 1.9715252016253776e-05, "loss": 0.7308, "step": 2916 }, { "epoch": 0.1039243279833265, "grad_norm": 1.5078977891874565, "learning_rate": 1.9714978542147958e-05, "loss": 0.6043, "step": 2917 }, { "epoch": 0.10395995510982062, "grad_norm": 2.332696957907907, "learning_rate": 1.9714704938680408e-05, "loss": 0.6783, "step": 2918 }, { "epoch": 0.10399558223631473, "grad_norm": 1.8905647061256865, "learning_rate": 1.9714431205854773e-05, "loss": 0.6131, "step": 2919 }, { "epoch": 0.10403120936280884, "grad_norm": 2.0046996180869288, "learning_rate": 1.9714157343674695e-05, "loss": 0.595, "step": 2920 }, { "epoch": 0.10406683648930296, "grad_norm": 2.099980637513715, "learning_rate": 1.9713883352143825e-05, "loss": 0.6374, "step": 2921 }, { "epoch": 0.10410246361579707, "grad_norm": 1.3171324135385596, "learning_rate": 1.9713609231265807e-05, "loss": 0.5685, "step": 2922 }, { "epoch": 0.10413809074229118, "grad_norm": 1.3742617188763213, "learning_rate": 1.9713334981044293e-05, "loss": 0.5286, "step": 2923 }, { "epoch": 0.10417371786878529, "grad_norm": 2.0775073325913453, "learning_rate": 1.9713060601482934e-05, "loss": 0.6928, "step": 2924 }, { "epoch": 0.10420934499527941, "grad_norm": 1.2026317728972182, "learning_rate": 1.9712786092585384e-05, "loss": 0.5081, "step": 2925 }, { "epoch": 0.10424497212177351, "grad_norm": 1.8529975992233887, "learning_rate": 1.9712511454355298e-05, "loss": 0.6308, "step": 2926 }, { "epoch": 0.10428059924826763, "grad_norm": 1.167780274753364, "learning_rate": 1.9712236686796332e-05, "loss": 0.5192, "step": 2927 }, { "epoch": 0.10431622637476175, "grad_norm": 1.389180367252183, "learning_rate": 1.971196178991215e-05, "loss": 0.5696, "step": 2928 }, { "epoch": 0.10435185350125585, "grad_norm": 1.7355425378935543, "learning_rate": 1.9711686763706404e-05, "loss": 0.5459, "step": 2929 }, { "epoch": 0.10438748062774997, "grad_norm": 1.334182198106215, "learning_rate": 1.9711411608182764e-05, "loss": 0.5223, "step": 2930 }, { "epoch": 0.10442310775424408, "grad_norm": 1.3069723189973679, "learning_rate": 1.971113632334489e-05, "loss": 0.5116, "step": 2931 }, { "epoch": 0.10445873488073819, "grad_norm": 2.138801398221858, "learning_rate": 1.9710860909196446e-05, "loss": 0.6319, "step": 2932 }, { "epoch": 0.1044943620072323, "grad_norm": 1.348749197311975, "learning_rate": 1.9710585365741105e-05, "loss": 0.5229, "step": 2933 }, { "epoch": 0.10452998913372642, "grad_norm": 2.0451691146308635, "learning_rate": 1.971030969298253e-05, "loss": 0.5237, "step": 2934 }, { "epoch": 0.10456561626022053, "grad_norm": 1.2298777046916385, "learning_rate": 1.9710033890924395e-05, "loss": 0.5352, "step": 2935 }, { "epoch": 0.10460124338671464, "grad_norm": 1.9056005180829199, "learning_rate": 1.9709757959570373e-05, "loss": 0.6047, "step": 2936 }, { "epoch": 0.10463687051320876, "grad_norm": 1.1804247210236365, "learning_rate": 1.9709481898924133e-05, "loss": 0.5475, "step": 2937 }, { "epoch": 0.10467249763970286, "grad_norm": 1.3111464761198117, "learning_rate": 1.970920570898936e-05, "loss": 0.5136, "step": 2938 }, { "epoch": 0.10470812476619698, "grad_norm": 1.2575313776289896, "learning_rate": 1.9708929389769725e-05, "loss": 0.5108, "step": 2939 }, { "epoch": 0.1047437518926911, "grad_norm": 1.528699839600359, "learning_rate": 1.970865294126891e-05, "loss": 0.5418, "step": 2940 }, { "epoch": 0.1047793790191852, "grad_norm": 1.206171562128153, "learning_rate": 1.9708376363490593e-05, "loss": 0.5585, "step": 2941 }, { "epoch": 0.10481500614567932, "grad_norm": 2.1603488432196536, "learning_rate": 1.970809965643846e-05, "loss": 0.6625, "step": 2942 }, { "epoch": 0.10485063327217344, "grad_norm": 1.3029942800176986, "learning_rate": 1.9707822820116193e-05, "loss": 0.5684, "step": 2943 }, { "epoch": 0.10488626039866755, "grad_norm": 2.2188665613760623, "learning_rate": 1.970754585452748e-05, "loss": 0.6475, "step": 2944 }, { "epoch": 0.10492188752516166, "grad_norm": 2.3823905980233056, "learning_rate": 1.9707268759676013e-05, "loss": 0.6318, "step": 2945 }, { "epoch": 0.10495751465165577, "grad_norm": 1.3502016834857484, "learning_rate": 1.9706991535565473e-05, "loss": 0.5187, "step": 2946 }, { "epoch": 0.10499314177814989, "grad_norm": 1.8098895165600937, "learning_rate": 1.970671418219956e-05, "loss": 0.6253, "step": 2947 }, { "epoch": 0.105028768904644, "grad_norm": 2.0745259313896303, "learning_rate": 1.970643669958196e-05, "loss": 0.6353, "step": 2948 }, { "epoch": 0.10506439603113811, "grad_norm": 1.2313316070542057, "learning_rate": 1.970615908771637e-05, "loss": 0.5361, "step": 2949 }, { "epoch": 0.10510002315763223, "grad_norm": 1.8661128133157143, "learning_rate": 1.9705881346606487e-05, "loss": 0.6898, "step": 2950 }, { "epoch": 0.10513565028412633, "grad_norm": 1.7999710844799435, "learning_rate": 1.9705603476256012e-05, "loss": 0.5985, "step": 2951 }, { "epoch": 0.10517127741062045, "grad_norm": 1.7177758944789072, "learning_rate": 1.9705325476668646e-05, "loss": 0.6168, "step": 2952 }, { "epoch": 0.10520690453711457, "grad_norm": 1.9721709669512884, "learning_rate": 1.9705047347848083e-05, "loss": 0.6826, "step": 2953 }, { "epoch": 0.10524253166360867, "grad_norm": 3.0525013955068774, "learning_rate": 1.9704769089798032e-05, "loss": 0.5659, "step": 2954 }, { "epoch": 0.10527815879010279, "grad_norm": 1.6183739426359893, "learning_rate": 1.9704490702522198e-05, "loss": 0.654, "step": 2955 }, { "epoch": 0.1053137859165969, "grad_norm": 1.4047158654226566, "learning_rate": 1.9704212186024288e-05, "loss": 0.5645, "step": 2956 }, { "epoch": 0.105349413043091, "grad_norm": 2.8729488556224796, "learning_rate": 1.970393354030801e-05, "loss": 0.6392, "step": 2957 }, { "epoch": 0.10538504016958512, "grad_norm": 2.2502013626142814, "learning_rate": 1.970365476537707e-05, "loss": 0.6582, "step": 2958 }, { "epoch": 0.10542066729607924, "grad_norm": 1.7657816578484113, "learning_rate": 1.970337586123519e-05, "loss": 0.6135, "step": 2959 }, { "epoch": 0.10545629442257334, "grad_norm": 2.362073246214802, "learning_rate": 1.9703096827886076e-05, "loss": 0.6408, "step": 2960 }, { "epoch": 0.10549192154906746, "grad_norm": 1.960932304122304, "learning_rate": 1.9702817665333443e-05, "loss": 0.5255, "step": 2961 }, { "epoch": 0.10552754867556158, "grad_norm": 2.2592323307944255, "learning_rate": 1.9702538373581012e-05, "loss": 0.6354, "step": 2962 }, { "epoch": 0.10556317580205568, "grad_norm": 2.208049304682393, "learning_rate": 1.9702258952632505e-05, "loss": 0.6281, "step": 2963 }, { "epoch": 0.1055988029285498, "grad_norm": 2.3774118742093853, "learning_rate": 1.9701979402491637e-05, "loss": 0.6707, "step": 2964 }, { "epoch": 0.10563443005504392, "grad_norm": 2.0432509366745863, "learning_rate": 1.970169972316213e-05, "loss": 0.5362, "step": 2965 }, { "epoch": 0.10567005718153802, "grad_norm": 2.1438142735113734, "learning_rate": 1.970141991464771e-05, "loss": 0.6722, "step": 2966 }, { "epoch": 0.10570568430803214, "grad_norm": 1.6983760718286072, "learning_rate": 1.9701139976952104e-05, "loss": 0.55, "step": 2967 }, { "epoch": 0.10574131143452625, "grad_norm": 1.2153937667042958, "learning_rate": 1.9700859910079037e-05, "loss": 0.5112, "step": 2968 }, { "epoch": 0.10577693856102036, "grad_norm": 1.2268656810595893, "learning_rate": 1.9700579714032242e-05, "loss": 0.5198, "step": 2969 }, { "epoch": 0.10581256568751447, "grad_norm": 1.7049091145640833, "learning_rate": 1.9700299388815447e-05, "loss": 0.5297, "step": 2970 }, { "epoch": 0.10584819281400859, "grad_norm": 1.9528804441935612, "learning_rate": 1.9700018934432385e-05, "loss": 0.5944, "step": 2971 }, { "epoch": 0.1058838199405027, "grad_norm": 1.3596651992340176, "learning_rate": 1.969973835088679e-05, "loss": 0.5986, "step": 2972 }, { "epoch": 0.10591944706699681, "grad_norm": 1.7522133354110105, "learning_rate": 1.9699457638182404e-05, "loss": 0.5916, "step": 2973 }, { "epoch": 0.10595507419349093, "grad_norm": 1.9618793609122378, "learning_rate": 1.9699176796322955e-05, "loss": 0.6175, "step": 2974 }, { "epoch": 0.10599070131998503, "grad_norm": 1.7696972693687334, "learning_rate": 1.969889582531219e-05, "loss": 0.6961, "step": 2975 }, { "epoch": 0.10602632844647915, "grad_norm": 1.8364789972704891, "learning_rate": 1.9698614725153847e-05, "loss": 0.7343, "step": 2976 }, { "epoch": 0.10606195557297327, "grad_norm": 1.6266155338669732, "learning_rate": 1.9698333495851672e-05, "loss": 0.6112, "step": 2977 }, { "epoch": 0.10609758269946737, "grad_norm": 1.734899283593026, "learning_rate": 1.9698052137409404e-05, "loss": 0.5455, "step": 2978 }, { "epoch": 0.10613320982596149, "grad_norm": 1.9963261787529103, "learning_rate": 1.9697770649830796e-05, "loss": 0.5291, "step": 2979 }, { "epoch": 0.1061688369524556, "grad_norm": 1.44415028469315, "learning_rate": 1.9697489033119594e-05, "loss": 0.5829, "step": 2980 }, { "epoch": 0.1062044640789497, "grad_norm": 1.874756724697698, "learning_rate": 1.9697207287279543e-05, "loss": 0.5866, "step": 2981 }, { "epoch": 0.10624009120544382, "grad_norm": 1.7251150940609479, "learning_rate": 1.9696925412314405e-05, "loss": 0.682, "step": 2982 }, { "epoch": 0.10627571833193794, "grad_norm": 2.6075466257577182, "learning_rate": 1.9696643408227923e-05, "loss": 0.7119, "step": 2983 }, { "epoch": 0.10631134545843204, "grad_norm": 2.167162572307654, "learning_rate": 1.9696361275023856e-05, "loss": 0.6552, "step": 2984 }, { "epoch": 0.10634697258492616, "grad_norm": 1.41968015363925, "learning_rate": 1.9696079012705964e-05, "loss": 0.5405, "step": 2985 }, { "epoch": 0.10638259971142028, "grad_norm": 1.282550116617018, "learning_rate": 1.9695796621278004e-05, "loss": 0.557, "step": 2986 }, { "epoch": 0.10641822683791438, "grad_norm": 1.403149817629937, "learning_rate": 1.9695514100743734e-05, "loss": 0.5859, "step": 2987 }, { "epoch": 0.1064538539644085, "grad_norm": 1.2252826288520977, "learning_rate": 1.9695231451106914e-05, "loss": 0.5038, "step": 2988 }, { "epoch": 0.10648948109090262, "grad_norm": 1.2676684025688967, "learning_rate": 1.9694948672371313e-05, "loss": 0.5145, "step": 2989 }, { "epoch": 0.10652510821739672, "grad_norm": 1.255907355638448, "learning_rate": 1.9694665764540694e-05, "loss": 0.4431, "step": 2990 }, { "epoch": 0.10656073534389084, "grad_norm": 1.3930878418153247, "learning_rate": 1.9694382727618825e-05, "loss": 0.5279, "step": 2991 }, { "epoch": 0.10659636247038495, "grad_norm": 1.7481944222079535, "learning_rate": 1.9694099561609474e-05, "loss": 0.6295, "step": 2992 }, { "epoch": 0.10663198959687907, "grad_norm": 1.3235881821980975, "learning_rate": 1.9693816266516408e-05, "loss": 0.5574, "step": 2993 }, { "epoch": 0.10666761672337317, "grad_norm": 1.9236273291286448, "learning_rate": 1.9693532842343405e-05, "loss": 0.6792, "step": 2994 }, { "epoch": 0.10670324384986729, "grad_norm": 1.3188377587314561, "learning_rate": 1.969324928909424e-05, "loss": 0.5024, "step": 2995 }, { "epoch": 0.10673887097636141, "grad_norm": 2.0500345071075547, "learning_rate": 1.9692965606772682e-05, "loss": 0.5748, "step": 2996 }, { "epoch": 0.10677449810285551, "grad_norm": 2.589594173957828, "learning_rate": 1.969268179538251e-05, "loss": 0.6302, "step": 2997 }, { "epoch": 0.10681012522934963, "grad_norm": 1.7588577533271854, "learning_rate": 1.969239785492751e-05, "loss": 0.5646, "step": 2998 }, { "epoch": 0.10684575235584375, "grad_norm": 1.930151026668985, "learning_rate": 1.9692113785411456e-05, "loss": 0.6607, "step": 2999 }, { "epoch": 0.10688137948233785, "grad_norm": 1.701811676138835, "learning_rate": 1.9691829586838134e-05, "loss": 0.5707, "step": 3000 }, { "epoch": 0.10691700660883197, "grad_norm": 1.3032929447823376, "learning_rate": 1.9691545259211322e-05, "loss": 0.5605, "step": 3001 }, { "epoch": 0.10695263373532608, "grad_norm": 1.717575086445554, "learning_rate": 1.9691260802534816e-05, "loss": 0.6591, "step": 3002 }, { "epoch": 0.10698826086182019, "grad_norm": 1.8938871997016147, "learning_rate": 1.9690976216812397e-05, "loss": 0.5821, "step": 3003 }, { "epoch": 0.1070238879883143, "grad_norm": 2.135064460953161, "learning_rate": 1.9690691502047852e-05, "loss": 0.6973, "step": 3004 }, { "epoch": 0.10705951511480842, "grad_norm": 1.4915671439930474, "learning_rate": 1.9690406658244984e-05, "loss": 0.5236, "step": 3005 }, { "epoch": 0.10709514224130252, "grad_norm": 1.3219350523752678, "learning_rate": 1.9690121685407572e-05, "loss": 0.5452, "step": 3006 }, { "epoch": 0.10713076936779664, "grad_norm": 1.645609018361605, "learning_rate": 1.968983658353942e-05, "loss": 0.5801, "step": 3007 }, { "epoch": 0.10716639649429076, "grad_norm": 1.3814169148991444, "learning_rate": 1.968955135264432e-05, "loss": 0.532, "step": 3008 }, { "epoch": 0.10720202362078486, "grad_norm": 1.3421192133119435, "learning_rate": 1.9689265992726072e-05, "loss": 0.5325, "step": 3009 }, { "epoch": 0.10723765074727898, "grad_norm": 1.224028060037796, "learning_rate": 1.9688980503788474e-05, "loss": 0.5371, "step": 3010 }, { "epoch": 0.1072732778737731, "grad_norm": 1.2139126468138652, "learning_rate": 1.968869488583533e-05, "loss": 0.5524, "step": 3011 }, { "epoch": 0.1073089050002672, "grad_norm": 1.3366359541696313, "learning_rate": 1.9688409138870443e-05, "loss": 0.4827, "step": 3012 }, { "epoch": 0.10734453212676132, "grad_norm": 1.2284727051548399, "learning_rate": 1.968812326289761e-05, "loss": 0.5609, "step": 3013 }, { "epoch": 0.10738015925325543, "grad_norm": 1.4557055225136755, "learning_rate": 1.968783725792065e-05, "loss": 0.5775, "step": 3014 }, { "epoch": 0.10741578637974954, "grad_norm": 1.828034035900903, "learning_rate": 1.968755112394336e-05, "loss": 0.5894, "step": 3015 }, { "epoch": 0.10745141350624365, "grad_norm": 2.2539894280291404, "learning_rate": 1.968726486096956e-05, "loss": 0.6102, "step": 3016 }, { "epoch": 0.10748704063273777, "grad_norm": 1.325138705366776, "learning_rate": 1.9686978469003057e-05, "loss": 0.5429, "step": 3017 }, { "epoch": 0.10752266775923187, "grad_norm": 2.8214724703351006, "learning_rate": 1.9686691948047665e-05, "loss": 0.6685, "step": 3018 }, { "epoch": 0.10755829488572599, "grad_norm": 1.37971649844773, "learning_rate": 1.96864052981072e-05, "loss": 0.5355, "step": 3019 }, { "epoch": 0.10759392201222011, "grad_norm": 2.73527449052987, "learning_rate": 1.9686118519185474e-05, "loss": 0.7943, "step": 3020 }, { "epoch": 0.10762954913871421, "grad_norm": 1.9340984855404784, "learning_rate": 1.9685831611286312e-05, "loss": 0.5598, "step": 3021 }, { "epoch": 0.10766517626520833, "grad_norm": 1.3319915275085585, "learning_rate": 1.9685544574413533e-05, "loss": 0.5537, "step": 3022 }, { "epoch": 0.10770080339170245, "grad_norm": 1.4391145039220476, "learning_rate": 1.968525740857096e-05, "loss": 0.5071, "step": 3023 }, { "epoch": 0.10773643051819655, "grad_norm": 1.370921374043089, "learning_rate": 1.968497011376241e-05, "loss": 0.5384, "step": 3024 }, { "epoch": 0.10777205764469067, "grad_norm": 1.3125214471595357, "learning_rate": 1.9684682689991716e-05, "loss": 0.5114, "step": 3025 }, { "epoch": 0.10780768477118478, "grad_norm": 1.3509291084197652, "learning_rate": 1.9684395137262703e-05, "loss": 0.5379, "step": 3026 }, { "epoch": 0.10784331189767889, "grad_norm": 1.8720052873435868, "learning_rate": 1.96841074555792e-05, "loss": 0.5893, "step": 3027 }, { "epoch": 0.107878939024173, "grad_norm": 1.9601527740896552, "learning_rate": 1.9683819644945036e-05, "loss": 0.5983, "step": 3028 }, { "epoch": 0.10791456615066712, "grad_norm": 1.9276518434645848, "learning_rate": 1.9683531705364048e-05, "loss": 0.5969, "step": 3029 }, { "epoch": 0.10795019327716122, "grad_norm": 1.7331441607737088, "learning_rate": 1.9683243636840064e-05, "loss": 0.6411, "step": 3030 }, { "epoch": 0.10798582040365534, "grad_norm": 1.2810428313116116, "learning_rate": 1.9682955439376923e-05, "loss": 0.4835, "step": 3031 }, { "epoch": 0.10802144753014946, "grad_norm": 2.102767050952131, "learning_rate": 1.9682667112978464e-05, "loss": 0.4873, "step": 3032 }, { "epoch": 0.10805707465664356, "grad_norm": 1.650029673665717, "learning_rate": 1.9682378657648523e-05, "loss": 0.6748, "step": 3033 }, { "epoch": 0.10809270178313768, "grad_norm": 2.446650296043552, "learning_rate": 1.968209007339094e-05, "loss": 0.6051, "step": 3034 }, { "epoch": 0.1081283289096318, "grad_norm": 1.8663860669515584, "learning_rate": 1.9681801360209568e-05, "loss": 0.6248, "step": 3035 }, { "epoch": 0.1081639560361259, "grad_norm": 1.26175548753573, "learning_rate": 1.968151251810824e-05, "loss": 0.521, "step": 3036 }, { "epoch": 0.10819958316262002, "grad_norm": 2.121611420739744, "learning_rate": 1.96812235470908e-05, "loss": 0.6493, "step": 3037 }, { "epoch": 0.10823521028911413, "grad_norm": 2.189913399406004, "learning_rate": 1.968093444716111e-05, "loss": 0.5575, "step": 3038 }, { "epoch": 0.10827083741560825, "grad_norm": 1.45370470140716, "learning_rate": 1.9680645218323005e-05, "loss": 0.5068, "step": 3039 }, { "epoch": 0.10830646454210235, "grad_norm": 1.9499514293099103, "learning_rate": 1.968035586058035e-05, "loss": 0.5187, "step": 3040 }, { "epoch": 0.10834209166859647, "grad_norm": 1.2339135792740228, "learning_rate": 1.9680066373936984e-05, "loss": 0.5054, "step": 3041 }, { "epoch": 0.10837771879509059, "grad_norm": 1.1506813388906048, "learning_rate": 1.9679776758396772e-05, "loss": 0.5245, "step": 3042 }, { "epoch": 0.10841334592158469, "grad_norm": 2.113008600085375, "learning_rate": 1.9679487013963566e-05, "loss": 0.6459, "step": 3043 }, { "epoch": 0.10844897304807881, "grad_norm": 1.2098173940819061, "learning_rate": 1.9679197140641223e-05, "loss": 0.5385, "step": 3044 }, { "epoch": 0.10848460017457293, "grad_norm": 1.8574781254202375, "learning_rate": 1.9678907138433605e-05, "loss": 0.7161, "step": 3045 }, { "epoch": 0.10852022730106703, "grad_norm": 1.723996516593752, "learning_rate": 1.9678617007344573e-05, "loss": 0.6237, "step": 3046 }, { "epoch": 0.10855585442756115, "grad_norm": 1.7239527486364938, "learning_rate": 1.967832674737799e-05, "loss": 0.6546, "step": 3047 }, { "epoch": 0.10859148155405526, "grad_norm": 1.9269653654674868, "learning_rate": 1.9678036358537726e-05, "loss": 0.6156, "step": 3048 }, { "epoch": 0.10862710868054937, "grad_norm": 1.9562855952769345, "learning_rate": 1.967774584082764e-05, "loss": 0.6276, "step": 3049 }, { "epoch": 0.10866273580704348, "grad_norm": 2.0507357498191046, "learning_rate": 1.9677455194251603e-05, "loss": 0.6406, "step": 3050 }, { "epoch": 0.1086983629335376, "grad_norm": 1.966103775108146, "learning_rate": 1.967716441881349e-05, "loss": 0.663, "step": 3051 }, { "epoch": 0.1087339900600317, "grad_norm": 2.4595234802588624, "learning_rate": 1.9676873514517164e-05, "loss": 0.6328, "step": 3052 }, { "epoch": 0.10876961718652582, "grad_norm": 2.023077030199123, "learning_rate": 1.9676582481366508e-05, "loss": 0.5776, "step": 3053 }, { "epoch": 0.10880524431301994, "grad_norm": 2.389692326297695, "learning_rate": 1.967629131936539e-05, "loss": 0.5525, "step": 3054 }, { "epoch": 0.10884087143951404, "grad_norm": 2.052009698642209, "learning_rate": 1.9676000028517688e-05, "loss": 0.6214, "step": 3055 }, { "epoch": 0.10887649856600816, "grad_norm": 1.237358313432261, "learning_rate": 1.9675708608827286e-05, "loss": 0.5886, "step": 3056 }, { "epoch": 0.10891212569250228, "grad_norm": 1.2569011961560994, "learning_rate": 1.9675417060298058e-05, "loss": 0.5066, "step": 3057 }, { "epoch": 0.10894775281899638, "grad_norm": 3.4842132868521123, "learning_rate": 1.9675125382933894e-05, "loss": 0.6089, "step": 3058 }, { "epoch": 0.1089833799454905, "grad_norm": 1.322388821586823, "learning_rate": 1.967483357673867e-05, "loss": 0.5215, "step": 3059 }, { "epoch": 0.10901900707198461, "grad_norm": 1.503479460074899, "learning_rate": 1.9674541641716272e-05, "loss": 0.56, "step": 3060 }, { "epoch": 0.10905463419847872, "grad_norm": 1.400673133478176, "learning_rate": 1.9674249577870593e-05, "loss": 0.5488, "step": 3061 }, { "epoch": 0.10909026132497283, "grad_norm": 1.7891549880864541, "learning_rate": 1.9673957385205516e-05, "loss": 0.6059, "step": 3062 }, { "epoch": 0.10912588845146695, "grad_norm": 1.4215961952791245, "learning_rate": 1.9673665063724937e-05, "loss": 0.579, "step": 3063 }, { "epoch": 0.10916151557796105, "grad_norm": 1.2653365812226187, "learning_rate": 1.967337261343275e-05, "loss": 0.5321, "step": 3064 }, { "epoch": 0.10919714270445517, "grad_norm": 2.334102464612805, "learning_rate": 1.967308003433284e-05, "loss": 0.7169, "step": 3065 }, { "epoch": 0.10923276983094929, "grad_norm": 2.0551340522661183, "learning_rate": 1.9672787326429107e-05, "loss": 0.6099, "step": 3066 }, { "epoch": 0.10926839695744339, "grad_norm": 1.9397873100602683, "learning_rate": 1.9672494489725453e-05, "loss": 0.5637, "step": 3067 }, { "epoch": 0.10930402408393751, "grad_norm": 1.3134533303644342, "learning_rate": 1.9672201524225774e-05, "loss": 0.533, "step": 3068 }, { "epoch": 0.10933965121043163, "grad_norm": 1.1977570254655177, "learning_rate": 1.967190842993397e-05, "loss": 0.5353, "step": 3069 }, { "epoch": 0.10937527833692573, "grad_norm": 1.4104068593782206, "learning_rate": 1.9671615206853947e-05, "loss": 0.6002, "step": 3070 }, { "epoch": 0.10941090546341985, "grad_norm": 1.8868339642773824, "learning_rate": 1.9671321854989605e-05, "loss": 0.6982, "step": 3071 }, { "epoch": 0.10944653258991396, "grad_norm": 1.7655827168298666, "learning_rate": 1.9671028374344858e-05, "loss": 0.616, "step": 3072 }, { "epoch": 0.10948215971640807, "grad_norm": 1.5138552092419675, "learning_rate": 1.9670734764923604e-05, "loss": 0.59, "step": 3073 }, { "epoch": 0.10951778684290218, "grad_norm": 1.2016852278714576, "learning_rate": 1.9670441026729757e-05, "loss": 0.4754, "step": 3074 }, { "epoch": 0.1095534139693963, "grad_norm": 2.0579235163780605, "learning_rate": 1.9670147159767228e-05, "loss": 0.5953, "step": 3075 }, { "epoch": 0.1095890410958904, "grad_norm": 1.2524183343285336, "learning_rate": 1.9669853164039935e-05, "loss": 0.5111, "step": 3076 }, { "epoch": 0.10962466822238452, "grad_norm": 2.040383059375501, "learning_rate": 1.9669559039551783e-05, "loss": 0.5834, "step": 3077 }, { "epoch": 0.10966029534887864, "grad_norm": 1.496777187679646, "learning_rate": 1.9669264786306695e-05, "loss": 0.5418, "step": 3078 }, { "epoch": 0.10969592247537274, "grad_norm": 1.9961818205890896, "learning_rate": 1.966897040430859e-05, "loss": 0.5526, "step": 3079 }, { "epoch": 0.10973154960186686, "grad_norm": 1.2227128262491294, "learning_rate": 1.9668675893561384e-05, "loss": 0.5844, "step": 3080 }, { "epoch": 0.10976717672836098, "grad_norm": 1.1948900731283283, "learning_rate": 1.9668381254069e-05, "loss": 0.5232, "step": 3081 }, { "epoch": 0.10980280385485508, "grad_norm": 2.2468801798323703, "learning_rate": 1.9668086485835368e-05, "loss": 0.6429, "step": 3082 }, { "epoch": 0.1098384309813492, "grad_norm": 1.3758029046114626, "learning_rate": 1.96677915888644e-05, "loss": 0.5617, "step": 3083 }, { "epoch": 0.10987405810784331, "grad_norm": 1.2729348888648935, "learning_rate": 1.966749656316003e-05, "loss": 0.4686, "step": 3084 }, { "epoch": 0.10990968523433743, "grad_norm": 1.2065481906684423, "learning_rate": 1.9667201408726187e-05, "loss": 0.4981, "step": 3085 }, { "epoch": 0.10994531236083153, "grad_norm": 1.8705458658501748, "learning_rate": 1.9666906125566803e-05, "loss": 0.5576, "step": 3086 }, { "epoch": 0.10998093948732565, "grad_norm": 1.145175690703991, "learning_rate": 1.9666610713685804e-05, "loss": 0.5415, "step": 3087 }, { "epoch": 0.11001656661381977, "grad_norm": 1.2443185127748513, "learning_rate": 1.966631517308713e-05, "loss": 0.5577, "step": 3088 }, { "epoch": 0.11005219374031387, "grad_norm": 2.281512075044854, "learning_rate": 1.966601950377471e-05, "loss": 0.6349, "step": 3089 }, { "epoch": 0.11008782086680799, "grad_norm": 1.8146853042161093, "learning_rate": 1.9665723705752483e-05, "loss": 0.6139, "step": 3090 }, { "epoch": 0.1101234479933021, "grad_norm": 1.549536049675488, "learning_rate": 1.9665427779024394e-05, "loss": 0.6492, "step": 3091 }, { "epoch": 0.11015907511979621, "grad_norm": 2.22988996846683, "learning_rate": 1.9665131723594376e-05, "loss": 0.6479, "step": 3092 }, { "epoch": 0.11019470224629033, "grad_norm": 1.2623041097346508, "learning_rate": 1.966483553946637e-05, "loss": 0.4945, "step": 3093 }, { "epoch": 0.11023032937278444, "grad_norm": 1.256942655481055, "learning_rate": 1.9664539226644327e-05, "loss": 0.5347, "step": 3094 }, { "epoch": 0.11026595649927855, "grad_norm": 1.354934227831858, "learning_rate": 1.9664242785132188e-05, "loss": 0.6179, "step": 3095 }, { "epoch": 0.11030158362577266, "grad_norm": 1.1836996433654625, "learning_rate": 1.9663946214933905e-05, "loss": 0.4979, "step": 3096 }, { "epoch": 0.11033721075226678, "grad_norm": 2.006608025511706, "learning_rate": 1.9663649516053417e-05, "loss": 0.563, "step": 3097 }, { "epoch": 0.11037283787876088, "grad_norm": 1.3163243086880732, "learning_rate": 1.9663352688494686e-05, "loss": 0.517, "step": 3098 }, { "epoch": 0.110408465005255, "grad_norm": 1.2084079922581412, "learning_rate": 1.966305573226166e-05, "loss": 0.5133, "step": 3099 }, { "epoch": 0.11044409213174912, "grad_norm": 2.139413391546505, "learning_rate": 1.966275864735829e-05, "loss": 0.624, "step": 3100 }, { "epoch": 0.11047971925824322, "grad_norm": 2.09181606366276, "learning_rate": 1.9662461433788537e-05, "loss": 0.5327, "step": 3101 }, { "epoch": 0.11051534638473734, "grad_norm": 1.9095310543259763, "learning_rate": 1.9662164091556355e-05, "loss": 0.5744, "step": 3102 }, { "epoch": 0.11055097351123146, "grad_norm": 1.7095141147776687, "learning_rate": 1.9661866620665705e-05, "loss": 0.6245, "step": 3103 }, { "epoch": 0.11058660063772556, "grad_norm": 1.3462165422696222, "learning_rate": 1.9661569021120546e-05, "loss": 0.5311, "step": 3104 }, { "epoch": 0.11062222776421968, "grad_norm": 1.752992474231438, "learning_rate": 1.9661271292924845e-05, "loss": 0.6047, "step": 3105 }, { "epoch": 0.1106578548907138, "grad_norm": 1.8697861930442221, "learning_rate": 1.9660973436082563e-05, "loss": 0.6052, "step": 3106 }, { "epoch": 0.1106934820172079, "grad_norm": 1.8769907942555104, "learning_rate": 1.9660675450597667e-05, "loss": 0.6459, "step": 3107 }, { "epoch": 0.11072910914370201, "grad_norm": 1.3086950479768926, "learning_rate": 1.9660377336474127e-05, "loss": 0.4981, "step": 3108 }, { "epoch": 0.11076473627019613, "grad_norm": 1.2275540948799641, "learning_rate": 1.9660079093715906e-05, "loss": 0.498, "step": 3109 }, { "epoch": 0.11080036339669024, "grad_norm": 1.4630917779601476, "learning_rate": 1.9659780722326983e-05, "loss": 0.573, "step": 3110 }, { "epoch": 0.11083599052318435, "grad_norm": 2.31852419087001, "learning_rate": 1.9659482222311327e-05, "loss": 0.6763, "step": 3111 }, { "epoch": 0.11087161764967847, "grad_norm": 1.2374080673819823, "learning_rate": 1.9659183593672914e-05, "loss": 0.5593, "step": 3112 }, { "epoch": 0.11090724477617257, "grad_norm": 1.2635029746803252, "learning_rate": 1.965888483641572e-05, "loss": 0.5194, "step": 3113 }, { "epoch": 0.11094287190266669, "grad_norm": 2.5935892833947745, "learning_rate": 1.9658585950543725e-05, "loss": 0.6366, "step": 3114 }, { "epoch": 0.11097849902916081, "grad_norm": 1.6418053056640314, "learning_rate": 1.9658286936060907e-05, "loss": 0.6225, "step": 3115 }, { "epoch": 0.11101412615565491, "grad_norm": 1.6887325986464963, "learning_rate": 1.9657987792971244e-05, "loss": 0.5771, "step": 3116 }, { "epoch": 0.11104975328214903, "grad_norm": 1.408889573583188, "learning_rate": 1.9657688521278726e-05, "loss": 0.4866, "step": 3117 }, { "epoch": 0.11108538040864314, "grad_norm": 1.1604313823941388, "learning_rate": 1.9657389120987332e-05, "loss": 0.4881, "step": 3118 }, { "epoch": 0.11112100753513725, "grad_norm": 1.3259774007661216, "learning_rate": 1.9657089592101057e-05, "loss": 0.5428, "step": 3119 }, { "epoch": 0.11115663466163136, "grad_norm": 2.511968068143354, "learning_rate": 1.965678993462388e-05, "loss": 0.6543, "step": 3120 }, { "epoch": 0.11119226178812548, "grad_norm": 1.4072621071704705, "learning_rate": 1.9656490148559797e-05, "loss": 0.5897, "step": 3121 }, { "epoch": 0.11122788891461959, "grad_norm": 2.0651070111381027, "learning_rate": 1.9656190233912796e-05, "loss": 0.6359, "step": 3122 }, { "epoch": 0.1112635160411137, "grad_norm": 1.270428447073583, "learning_rate": 1.9655890190686878e-05, "loss": 0.5459, "step": 3123 }, { "epoch": 0.11129914316760782, "grad_norm": 1.5120321776976349, "learning_rate": 1.9655590018886027e-05, "loss": 0.5428, "step": 3124 }, { "epoch": 0.11133477029410192, "grad_norm": 1.3039341414400845, "learning_rate": 1.965528971851425e-05, "loss": 0.5296, "step": 3125 }, { "epoch": 0.11137039742059604, "grad_norm": 1.4854061646984043, "learning_rate": 1.965498928957554e-05, "loss": 0.6009, "step": 3126 }, { "epoch": 0.11140602454709016, "grad_norm": 2.1617011654633402, "learning_rate": 1.96546887320739e-05, "loss": 0.6211, "step": 3127 }, { "epoch": 0.11144165167358426, "grad_norm": 1.3565718586359636, "learning_rate": 1.965438804601333e-05, "loss": 0.5596, "step": 3128 }, { "epoch": 0.11147727880007838, "grad_norm": 1.307156709034943, "learning_rate": 1.9654087231397834e-05, "loss": 0.5398, "step": 3129 }, { "epoch": 0.1115129059265725, "grad_norm": 1.1752141461474535, "learning_rate": 1.9653786288231424e-05, "loss": 0.4992, "step": 3130 }, { "epoch": 0.1115485330530666, "grad_norm": 1.4201400112226594, "learning_rate": 1.9653485216518094e-05, "loss": 0.6209, "step": 3131 }, { "epoch": 0.11158416017956072, "grad_norm": 1.780597443449136, "learning_rate": 1.9653184016261867e-05, "loss": 0.6215, "step": 3132 }, { "epoch": 0.11161978730605483, "grad_norm": 2.722818261355748, "learning_rate": 1.9652882687466746e-05, "loss": 0.5699, "step": 3133 }, { "epoch": 0.11165541443254895, "grad_norm": 1.2679641129195551, "learning_rate": 1.9652581230136748e-05, "loss": 0.5982, "step": 3134 }, { "epoch": 0.11169104155904305, "grad_norm": 1.3041157191345052, "learning_rate": 1.965227964427588e-05, "loss": 0.5508, "step": 3135 }, { "epoch": 0.11172666868553717, "grad_norm": 1.4628890993907173, "learning_rate": 1.9651977929888163e-05, "loss": 0.4952, "step": 3136 }, { "epoch": 0.11176229581203129, "grad_norm": 1.2731459610874782, "learning_rate": 1.9651676086977613e-05, "loss": 0.5461, "step": 3137 }, { "epoch": 0.11179792293852539, "grad_norm": 2.026497066821899, "learning_rate": 1.9651374115548255e-05, "loss": 0.6917, "step": 3138 }, { "epoch": 0.11183355006501951, "grad_norm": 2.009328320445993, "learning_rate": 1.9651072015604098e-05, "loss": 0.6401, "step": 3139 }, { "epoch": 0.11186917719151362, "grad_norm": 2.5986194417575654, "learning_rate": 1.9650769787149175e-05, "loss": 0.7533, "step": 3140 }, { "epoch": 0.11190480431800773, "grad_norm": 2.408001079101615, "learning_rate": 1.9650467430187503e-05, "loss": 0.6813, "step": 3141 }, { "epoch": 0.11194043144450185, "grad_norm": 1.8046323101235036, "learning_rate": 1.9650164944723116e-05, "loss": 0.5999, "step": 3142 }, { "epoch": 0.11197605857099596, "grad_norm": 1.8033315706078308, "learning_rate": 1.9649862330760036e-05, "loss": 0.4977, "step": 3143 }, { "epoch": 0.11201168569749007, "grad_norm": 1.7542739664973033, "learning_rate": 1.9649559588302295e-05, "loss": 0.6581, "step": 3144 }, { "epoch": 0.11204731282398418, "grad_norm": 2.376312680173263, "learning_rate": 1.9649256717353926e-05, "loss": 0.6744, "step": 3145 }, { "epoch": 0.1120829399504783, "grad_norm": 1.6896441109461793, "learning_rate": 1.9648953717918955e-05, "loss": 0.5275, "step": 3146 }, { "epoch": 0.1121185670769724, "grad_norm": 1.219690916652234, "learning_rate": 1.964865059000142e-05, "loss": 0.498, "step": 3147 }, { "epoch": 0.11215419420346652, "grad_norm": 1.406538618183878, "learning_rate": 1.964834733360536e-05, "loss": 0.4847, "step": 3148 }, { "epoch": 0.11218982132996064, "grad_norm": 1.3262721159034394, "learning_rate": 1.9648043948734814e-05, "loss": 0.5401, "step": 3149 }, { "epoch": 0.11222544845645474, "grad_norm": 1.7412442804520296, "learning_rate": 1.9647740435393818e-05, "loss": 0.5779, "step": 3150 }, { "epoch": 0.11226107558294886, "grad_norm": 1.3918443810580068, "learning_rate": 1.9647436793586415e-05, "loss": 0.5688, "step": 3151 }, { "epoch": 0.11229670270944297, "grad_norm": 1.3200277368994313, "learning_rate": 1.9647133023316646e-05, "loss": 0.5642, "step": 3152 }, { "epoch": 0.11233232983593708, "grad_norm": 1.879773854274217, "learning_rate": 1.964682912458856e-05, "loss": 0.5515, "step": 3153 }, { "epoch": 0.1123679569624312, "grad_norm": 1.264936379157477, "learning_rate": 1.9646525097406198e-05, "loss": 0.4888, "step": 3154 }, { "epoch": 0.11240358408892531, "grad_norm": 1.5472774544770422, "learning_rate": 1.9646220941773616e-05, "loss": 0.572, "step": 3155 }, { "epoch": 0.11243921121541942, "grad_norm": 1.578934617282185, "learning_rate": 1.9645916657694858e-05, "loss": 0.5861, "step": 3156 }, { "epoch": 0.11247483834191353, "grad_norm": 1.8771357137554887, "learning_rate": 1.9645612245173978e-05, "loss": 0.6211, "step": 3157 }, { "epoch": 0.11251046546840765, "grad_norm": 1.341341681563147, "learning_rate": 1.964530770421503e-05, "loss": 0.5189, "step": 3158 }, { "epoch": 0.11254609259490175, "grad_norm": 2.1709549709331566, "learning_rate": 1.9645003034822064e-05, "loss": 0.6134, "step": 3159 }, { "epoch": 0.11258171972139587, "grad_norm": 1.2678698028121045, "learning_rate": 1.9644698236999144e-05, "loss": 0.5339, "step": 3160 }, { "epoch": 0.11261734684788999, "grad_norm": 2.0711815887532947, "learning_rate": 1.964439331075033e-05, "loss": 0.5744, "step": 3161 }, { "epoch": 0.11265297397438409, "grad_norm": 1.9833832593280012, "learning_rate": 1.964408825607967e-05, "loss": 0.6439, "step": 3162 }, { "epoch": 0.11268860110087821, "grad_norm": 2.7066992368293397, "learning_rate": 1.964378307299124e-05, "loss": 0.6082, "step": 3163 }, { "epoch": 0.11272422822737233, "grad_norm": 1.7447540811799978, "learning_rate": 1.9643477761489097e-05, "loss": 0.5773, "step": 3164 }, { "epoch": 0.11275985535386643, "grad_norm": 1.3174862061673154, "learning_rate": 1.9643172321577304e-05, "loss": 0.5574, "step": 3165 }, { "epoch": 0.11279548248036055, "grad_norm": 1.4838329894482796, "learning_rate": 1.9642866753259934e-05, "loss": 0.5108, "step": 3166 }, { "epoch": 0.11283110960685466, "grad_norm": 1.2054368615462505, "learning_rate": 1.9642561056541052e-05, "loss": 0.4505, "step": 3167 }, { "epoch": 0.11286673673334877, "grad_norm": 1.3090367458421037, "learning_rate": 1.964225523142473e-05, "loss": 0.492, "step": 3168 }, { "epoch": 0.11290236385984288, "grad_norm": 1.2493090264043802, "learning_rate": 1.9641949277915044e-05, "loss": 0.4726, "step": 3169 }, { "epoch": 0.112937990986337, "grad_norm": 1.9231864887473578, "learning_rate": 1.9641643196016058e-05, "loss": 0.7253, "step": 3170 }, { "epoch": 0.1129736181128311, "grad_norm": 1.4234426907033981, "learning_rate": 1.9641336985731854e-05, "loss": 0.5478, "step": 3171 }, { "epoch": 0.11300924523932522, "grad_norm": 1.1940168498842718, "learning_rate": 1.9641030647066516e-05, "loss": 0.5095, "step": 3172 }, { "epoch": 0.11304487236581934, "grad_norm": 1.6504519485762583, "learning_rate": 1.964072418002411e-05, "loss": 0.6438, "step": 3173 }, { "epoch": 0.11308049949231344, "grad_norm": 1.775836033918482, "learning_rate": 1.9640417584608726e-05, "loss": 0.6579, "step": 3174 }, { "epoch": 0.11311612661880756, "grad_norm": 2.0984007207316973, "learning_rate": 1.9640110860824442e-05, "loss": 0.601, "step": 3175 }, { "epoch": 0.11315175374530168, "grad_norm": 1.2885815449030087, "learning_rate": 1.9639804008675344e-05, "loss": 0.5678, "step": 3176 }, { "epoch": 0.11318738087179578, "grad_norm": 1.9801060915557305, "learning_rate": 1.9639497028165518e-05, "loss": 0.6429, "step": 3177 }, { "epoch": 0.1132230079982899, "grad_norm": 1.2218316190714937, "learning_rate": 1.963918991929905e-05, "loss": 0.5574, "step": 3178 }, { "epoch": 0.11325863512478401, "grad_norm": 1.5988402726171917, "learning_rate": 1.9638882682080033e-05, "loss": 0.6493, "step": 3179 }, { "epoch": 0.11329426225127813, "grad_norm": 1.3301656095613614, "learning_rate": 1.9638575316512557e-05, "loss": 0.5399, "step": 3180 }, { "epoch": 0.11332988937777223, "grad_norm": 1.4822445380917268, "learning_rate": 1.963826782260071e-05, "loss": 0.5405, "step": 3181 }, { "epoch": 0.11336551650426635, "grad_norm": 2.109091606444912, "learning_rate": 1.9637960200348592e-05, "loss": 0.6526, "step": 3182 }, { "epoch": 0.11340114363076047, "grad_norm": 2.246464356674048, "learning_rate": 1.9637652449760297e-05, "loss": 0.6745, "step": 3183 }, { "epoch": 0.11343677075725457, "grad_norm": 2.025951073803596, "learning_rate": 1.9637344570839924e-05, "loss": 0.6742, "step": 3184 }, { "epoch": 0.11347239788374869, "grad_norm": 1.2623657852974892, "learning_rate": 1.9637036563591573e-05, "loss": 0.4945, "step": 3185 }, { "epoch": 0.1135080250102428, "grad_norm": 1.2359084500915, "learning_rate": 1.963672842801934e-05, "loss": 0.5258, "step": 3186 }, { "epoch": 0.11354365213673691, "grad_norm": 1.2690487618982473, "learning_rate": 1.9636420164127337e-05, "loss": 0.5129, "step": 3187 }, { "epoch": 0.11357927926323103, "grad_norm": 2.0629357535213835, "learning_rate": 1.9636111771919665e-05, "loss": 0.6424, "step": 3188 }, { "epoch": 0.11361490638972514, "grad_norm": 2.0802296041912545, "learning_rate": 1.9635803251400427e-05, "loss": 0.6108, "step": 3189 }, { "epoch": 0.11365053351621925, "grad_norm": 1.1498131163386882, "learning_rate": 1.9635494602573732e-05, "loss": 0.4864, "step": 3190 }, { "epoch": 0.11368616064271336, "grad_norm": 1.3215378891450529, "learning_rate": 1.9635185825443694e-05, "loss": 0.5546, "step": 3191 }, { "epoch": 0.11372178776920748, "grad_norm": 1.6670475173107417, "learning_rate": 1.963487692001442e-05, "loss": 0.6288, "step": 3192 }, { "epoch": 0.11375741489570158, "grad_norm": 1.259144571454654, "learning_rate": 1.9634567886290027e-05, "loss": 0.5721, "step": 3193 }, { "epoch": 0.1137930420221957, "grad_norm": 1.475163898629637, "learning_rate": 1.963425872427463e-05, "loss": 0.5041, "step": 3194 }, { "epoch": 0.11382866914868982, "grad_norm": 1.8744541244088706, "learning_rate": 1.963394943397234e-05, "loss": 0.5831, "step": 3195 }, { "epoch": 0.11386429627518392, "grad_norm": 2.0892477503899283, "learning_rate": 1.9633640015387282e-05, "loss": 0.6413, "step": 3196 }, { "epoch": 0.11389992340167804, "grad_norm": 1.6978637097463902, "learning_rate": 1.9633330468523572e-05, "loss": 0.6531, "step": 3197 }, { "epoch": 0.11393555052817216, "grad_norm": 2.053713522978595, "learning_rate": 1.9633020793385338e-05, "loss": 0.5946, "step": 3198 }, { "epoch": 0.11397117765466626, "grad_norm": 1.818979341805036, "learning_rate": 1.9632710989976696e-05, "loss": 0.6388, "step": 3199 }, { "epoch": 0.11400680478116038, "grad_norm": 1.3357006919694652, "learning_rate": 1.9632401058301774e-05, "loss": 0.5122, "step": 3200 }, { "epoch": 0.11404243190765449, "grad_norm": 1.9264372814206512, "learning_rate": 1.9632090998364698e-05, "loss": 0.5721, "step": 3201 }, { "epoch": 0.1140780590341486, "grad_norm": 1.8623119745328716, "learning_rate": 1.9631780810169602e-05, "loss": 0.6298, "step": 3202 }, { "epoch": 0.11411368616064271, "grad_norm": 1.3465893410468732, "learning_rate": 1.963147049372061e-05, "loss": 0.5228, "step": 3203 }, { "epoch": 0.11414931328713683, "grad_norm": 1.7558941670306356, "learning_rate": 1.9631160049021852e-05, "loss": 0.6464, "step": 3204 }, { "epoch": 0.11418494041363093, "grad_norm": 1.8212500421128683, "learning_rate": 1.9630849476077473e-05, "loss": 0.5652, "step": 3205 }, { "epoch": 0.11422056754012505, "grad_norm": 1.8222759963789537, "learning_rate": 1.96305387748916e-05, "loss": 0.532, "step": 3206 }, { "epoch": 0.11425619466661917, "grad_norm": 1.9348152903779998, "learning_rate": 1.9630227945468368e-05, "loss": 0.5624, "step": 3207 }, { "epoch": 0.11429182179311327, "grad_norm": 1.8933550954545941, "learning_rate": 1.9629916987811924e-05, "loss": 0.641, "step": 3208 }, { "epoch": 0.11432744891960739, "grad_norm": 1.7964322921705222, "learning_rate": 1.9629605901926404e-05, "loss": 0.6353, "step": 3209 }, { "epoch": 0.1143630760461015, "grad_norm": 1.3100240578657372, "learning_rate": 1.962929468781595e-05, "loss": 0.5397, "step": 3210 }, { "epoch": 0.11439870317259561, "grad_norm": 1.1886853152979509, "learning_rate": 1.962898334548471e-05, "loss": 0.5273, "step": 3211 }, { "epoch": 0.11443433029908973, "grad_norm": 1.3533938769376435, "learning_rate": 1.9628671874936823e-05, "loss": 0.6189, "step": 3212 }, { "epoch": 0.11446995742558384, "grad_norm": 1.1634692584716453, "learning_rate": 1.9628360276176438e-05, "loss": 0.4918, "step": 3213 }, { "epoch": 0.11450558455207795, "grad_norm": 1.7640843236284103, "learning_rate": 1.9628048549207707e-05, "loss": 0.6088, "step": 3214 }, { "epoch": 0.11454121167857206, "grad_norm": 1.1954203265642356, "learning_rate": 1.9627736694034782e-05, "loss": 0.474, "step": 3215 }, { "epoch": 0.11457683880506618, "grad_norm": 1.9964132084631219, "learning_rate": 1.9627424710661812e-05, "loss": 0.6414, "step": 3216 }, { "epoch": 0.11461246593156028, "grad_norm": 2.248270801559651, "learning_rate": 1.9627112599092956e-05, "loss": 0.6114, "step": 3217 }, { "epoch": 0.1146480930580544, "grad_norm": 1.3170092319892708, "learning_rate": 1.9626800359332362e-05, "loss": 0.5371, "step": 3218 }, { "epoch": 0.11468372018454852, "grad_norm": 1.359784875797939, "learning_rate": 1.9626487991384194e-05, "loss": 0.5535, "step": 3219 }, { "epoch": 0.11471934731104262, "grad_norm": 1.7619816834779716, "learning_rate": 1.9626175495252612e-05, "loss": 0.7598, "step": 3220 }, { "epoch": 0.11475497443753674, "grad_norm": 1.6961436031402135, "learning_rate": 1.9625862870941774e-05, "loss": 0.5722, "step": 3221 }, { "epoch": 0.11479060156403086, "grad_norm": 1.5230298653072094, "learning_rate": 1.9625550118455844e-05, "loss": 0.601, "step": 3222 }, { "epoch": 0.11482622869052496, "grad_norm": 1.3333053943931288, "learning_rate": 1.9625237237798985e-05, "loss": 0.4824, "step": 3223 }, { "epoch": 0.11486185581701908, "grad_norm": 2.0258101412303797, "learning_rate": 1.9624924228975365e-05, "loss": 0.714, "step": 3224 }, { "epoch": 0.1148974829435132, "grad_norm": 1.4277077628766879, "learning_rate": 1.9624611091989152e-05, "loss": 0.5842, "step": 3225 }, { "epoch": 0.11493311007000731, "grad_norm": 1.3052162948993857, "learning_rate": 1.9624297826844513e-05, "loss": 0.522, "step": 3226 }, { "epoch": 0.11496873719650141, "grad_norm": 1.1431532082583127, "learning_rate": 1.9623984433545625e-05, "loss": 0.4803, "step": 3227 }, { "epoch": 0.11500436432299553, "grad_norm": 1.3087549514702674, "learning_rate": 1.9623670912096656e-05, "loss": 0.5101, "step": 3228 }, { "epoch": 0.11503999144948965, "grad_norm": 1.4206126234364806, "learning_rate": 1.9623357262501783e-05, "loss": 0.553, "step": 3229 }, { "epoch": 0.11507561857598375, "grad_norm": 2.008151891488644, "learning_rate": 1.962304348476518e-05, "loss": 0.6226, "step": 3230 }, { "epoch": 0.11511124570247787, "grad_norm": 1.6700818174828957, "learning_rate": 1.9622729578891028e-05, "loss": 0.6397, "step": 3231 }, { "epoch": 0.11514687282897199, "grad_norm": 1.6834230267765762, "learning_rate": 1.9622415544883506e-05, "loss": 0.6411, "step": 3232 }, { "epoch": 0.11518249995546609, "grad_norm": 1.3864784687230258, "learning_rate": 1.9622101382746795e-05, "loss": 0.5417, "step": 3233 }, { "epoch": 0.1152181270819602, "grad_norm": 1.332027688985768, "learning_rate": 1.9621787092485078e-05, "loss": 0.5583, "step": 3234 }, { "epoch": 0.11525375420845432, "grad_norm": 1.2370525736964448, "learning_rate": 1.962147267410254e-05, "loss": 0.5214, "step": 3235 }, { "epoch": 0.11528938133494843, "grad_norm": 1.9026588278585956, "learning_rate": 1.9621158127603372e-05, "loss": 0.6424, "step": 3236 }, { "epoch": 0.11532500846144254, "grad_norm": 1.3531265701091257, "learning_rate": 1.9620843452991754e-05, "loss": 0.5099, "step": 3237 }, { "epoch": 0.11536063558793666, "grad_norm": 1.529153265107617, "learning_rate": 1.9620528650271882e-05, "loss": 0.4976, "step": 3238 }, { "epoch": 0.11539626271443076, "grad_norm": 1.767488209528548, "learning_rate": 1.962021371944795e-05, "loss": 0.5206, "step": 3239 }, { "epoch": 0.11543188984092488, "grad_norm": 2.3037832507984755, "learning_rate": 1.9619898660524145e-05, "loss": 0.5623, "step": 3240 }, { "epoch": 0.115467516967419, "grad_norm": 1.588849297114736, "learning_rate": 1.961958347350467e-05, "loss": 0.553, "step": 3241 }, { "epoch": 0.1155031440939131, "grad_norm": 1.4717329052010992, "learning_rate": 1.961926815839371e-05, "loss": 0.5357, "step": 3242 }, { "epoch": 0.11553877122040722, "grad_norm": 2.1374187501223267, "learning_rate": 1.9618952715195476e-05, "loss": 0.5987, "step": 3243 }, { "epoch": 0.11557439834690134, "grad_norm": 1.3132263207004247, "learning_rate": 1.961863714391416e-05, "loss": 0.5204, "step": 3244 }, { "epoch": 0.11561002547339544, "grad_norm": 1.260525383752187, "learning_rate": 1.9618321444553972e-05, "loss": 0.5238, "step": 3245 }, { "epoch": 0.11564565259988956, "grad_norm": 1.701273583581614, "learning_rate": 1.9618005617119106e-05, "loss": 0.6288, "step": 3246 }, { "epoch": 0.11568127972638367, "grad_norm": 2.73091086381851, "learning_rate": 1.961768966161378e-05, "loss": 0.701, "step": 3247 }, { "epoch": 0.11571690685287778, "grad_norm": 1.2810288699570629, "learning_rate": 1.9617373578042187e-05, "loss": 0.494, "step": 3248 }, { "epoch": 0.1157525339793719, "grad_norm": 1.4024085275878353, "learning_rate": 1.9617057366408545e-05, "loss": 0.5511, "step": 3249 }, { "epoch": 0.11578816110586601, "grad_norm": 1.3215692108608454, "learning_rate": 1.9616741026717063e-05, "loss": 0.5736, "step": 3250 }, { "epoch": 0.11582378823236011, "grad_norm": 2.116099793541012, "learning_rate": 1.961642455897195e-05, "loss": 0.6853, "step": 3251 }, { "epoch": 0.11585941535885423, "grad_norm": 1.2886348555892897, "learning_rate": 1.9616107963177423e-05, "loss": 0.5091, "step": 3252 }, { "epoch": 0.11589504248534835, "grad_norm": 1.6303635187533025, "learning_rate": 1.96157912393377e-05, "loss": 0.6299, "step": 3253 }, { "epoch": 0.11593066961184245, "grad_norm": 1.6339870352498014, "learning_rate": 1.9615474387456995e-05, "loss": 0.5362, "step": 3254 }, { "epoch": 0.11596629673833657, "grad_norm": 1.3163720121344646, "learning_rate": 1.9615157407539525e-05, "loss": 0.5476, "step": 3255 }, { "epoch": 0.11600192386483069, "grad_norm": 1.7811931230092253, "learning_rate": 1.9614840299589516e-05, "loss": 0.6241, "step": 3256 }, { "epoch": 0.11603755099132479, "grad_norm": 2.148240981111019, "learning_rate": 1.9614523063611185e-05, "loss": 0.6034, "step": 3257 }, { "epoch": 0.1160731781178189, "grad_norm": 1.2324188602583688, "learning_rate": 1.9614205699608763e-05, "loss": 0.5703, "step": 3258 }, { "epoch": 0.11610880524431302, "grad_norm": 1.3059503411922697, "learning_rate": 1.961388820758647e-05, "loss": 0.5663, "step": 3259 }, { "epoch": 0.11614443237080713, "grad_norm": 2.510593696589128, "learning_rate": 1.9613570587548537e-05, "loss": 0.6358, "step": 3260 }, { "epoch": 0.11618005949730124, "grad_norm": 1.8031294117615662, "learning_rate": 1.9613252839499192e-05, "loss": 0.6362, "step": 3261 }, { "epoch": 0.11621568662379536, "grad_norm": 2.0087288007720194, "learning_rate": 1.9612934963442663e-05, "loss": 0.6479, "step": 3262 }, { "epoch": 0.11625131375028946, "grad_norm": 1.2763946488818452, "learning_rate": 1.961261695938319e-05, "loss": 0.5309, "step": 3263 }, { "epoch": 0.11628694087678358, "grad_norm": 1.6279348376717284, "learning_rate": 1.9612298827325e-05, "loss": 0.5565, "step": 3264 }, { "epoch": 0.1163225680032777, "grad_norm": 2.2131440801896107, "learning_rate": 1.9611980567272337e-05, "loss": 0.6131, "step": 3265 }, { "epoch": 0.1163581951297718, "grad_norm": 1.3091389556230257, "learning_rate": 1.961166217922943e-05, "loss": 0.5558, "step": 3266 }, { "epoch": 0.11639382225626592, "grad_norm": 1.2638849882884322, "learning_rate": 1.9611343663200527e-05, "loss": 0.5951, "step": 3267 }, { "epoch": 0.11642944938276004, "grad_norm": 1.9434858247265687, "learning_rate": 1.961102501918986e-05, "loss": 0.7525, "step": 3268 }, { "epoch": 0.11646507650925414, "grad_norm": 1.3098474368285604, "learning_rate": 1.961070624720168e-05, "loss": 0.5376, "step": 3269 }, { "epoch": 0.11650070363574826, "grad_norm": 1.2262060911742025, "learning_rate": 1.961038734724023e-05, "loss": 0.5294, "step": 3270 }, { "epoch": 0.11653633076224237, "grad_norm": 1.8854204340523846, "learning_rate": 1.9610068319309757e-05, "loss": 0.655, "step": 3271 }, { "epoch": 0.11657195788873648, "grad_norm": 1.3252583446887995, "learning_rate": 1.9609749163414503e-05, "loss": 0.5273, "step": 3272 }, { "epoch": 0.1166075850152306, "grad_norm": 1.174833388438559, "learning_rate": 1.9609429879558726e-05, "loss": 0.4969, "step": 3273 }, { "epoch": 0.11664321214172471, "grad_norm": 1.4302124634976696, "learning_rate": 1.960911046774667e-05, "loss": 0.6411, "step": 3274 }, { "epoch": 0.11667883926821883, "grad_norm": 1.3944616538267531, "learning_rate": 1.9608790927982592e-05, "loss": 0.5713, "step": 3275 }, { "epoch": 0.11671446639471293, "grad_norm": 2.303361890095246, "learning_rate": 1.960847126027075e-05, "loss": 0.5949, "step": 3276 }, { "epoch": 0.11675009352120705, "grad_norm": 1.3810474977493816, "learning_rate": 1.9608151464615392e-05, "loss": 0.5003, "step": 3277 }, { "epoch": 0.11678572064770117, "grad_norm": 1.2075532584578286, "learning_rate": 1.9607831541020784e-05, "loss": 0.5333, "step": 3278 }, { "epoch": 0.11682134777419527, "grad_norm": 2.18082268748292, "learning_rate": 1.9607511489491187e-05, "loss": 0.6625, "step": 3279 }, { "epoch": 0.11685697490068939, "grad_norm": 1.312264105857379, "learning_rate": 1.9607191310030857e-05, "loss": 0.5585, "step": 3280 }, { "epoch": 0.1168926020271835, "grad_norm": 2.030205565727516, "learning_rate": 1.960687100264406e-05, "loss": 0.6513, "step": 3281 }, { "epoch": 0.11692822915367761, "grad_norm": 1.4725479822885525, "learning_rate": 1.9606550567335058e-05, "loss": 0.607, "step": 3282 }, { "epoch": 0.11696385628017172, "grad_norm": 1.3025670462355243, "learning_rate": 1.960623000410812e-05, "loss": 0.5279, "step": 3283 }, { "epoch": 0.11699948340666584, "grad_norm": 1.386266505038889, "learning_rate": 1.9605909312967516e-05, "loss": 0.5775, "step": 3284 }, { "epoch": 0.11703511053315994, "grad_norm": 1.4422630198476916, "learning_rate": 1.960558849391752e-05, "loss": 0.562, "step": 3285 }, { "epoch": 0.11707073765965406, "grad_norm": 1.697889257646645, "learning_rate": 1.9605267546962395e-05, "loss": 0.5462, "step": 3286 }, { "epoch": 0.11710636478614818, "grad_norm": 1.2194063881355666, "learning_rate": 1.960494647210642e-05, "loss": 0.547, "step": 3287 }, { "epoch": 0.11714199191264228, "grad_norm": 2.126635900780195, "learning_rate": 1.960462526935387e-05, "loss": 0.5779, "step": 3288 }, { "epoch": 0.1171776190391364, "grad_norm": 2.1313268498423454, "learning_rate": 1.960430393870902e-05, "loss": 0.597, "step": 3289 }, { "epoch": 0.11721324616563052, "grad_norm": 1.3101815192528588, "learning_rate": 1.9603982480176152e-05, "loss": 0.5004, "step": 3290 }, { "epoch": 0.11724887329212462, "grad_norm": 1.282724027428445, "learning_rate": 1.960366089375954e-05, "loss": 0.5056, "step": 3291 }, { "epoch": 0.11728450041861874, "grad_norm": 1.984258181768845, "learning_rate": 1.9603339179463475e-05, "loss": 0.5903, "step": 3292 }, { "epoch": 0.11732012754511285, "grad_norm": 1.3898848126396, "learning_rate": 1.9603017337292235e-05, "loss": 0.5283, "step": 3293 }, { "epoch": 0.11735575467160696, "grad_norm": 1.3019784073674558, "learning_rate": 1.960269536725011e-05, "loss": 0.5544, "step": 3294 }, { "epoch": 0.11739138179810107, "grad_norm": 1.6220754374636404, "learning_rate": 1.960237326934138e-05, "loss": 0.6515, "step": 3295 }, { "epoch": 0.11742700892459519, "grad_norm": 1.3677583436707452, "learning_rate": 1.960205104357034e-05, "loss": 0.4788, "step": 3296 }, { "epoch": 0.1174626360510893, "grad_norm": 1.317508910802289, "learning_rate": 1.960172868994128e-05, "loss": 0.5501, "step": 3297 }, { "epoch": 0.11749826317758341, "grad_norm": 1.3077029370880038, "learning_rate": 1.960140620845849e-05, "loss": 0.4932, "step": 3298 }, { "epoch": 0.11753389030407753, "grad_norm": 1.7157369601631587, "learning_rate": 1.9601083599126266e-05, "loss": 0.6544, "step": 3299 }, { "epoch": 0.11756951743057163, "grad_norm": 1.6784035046409655, "learning_rate": 1.96007608619489e-05, "loss": 0.6329, "step": 3300 }, { "epoch": 0.11760514455706575, "grad_norm": 1.254519247678451, "learning_rate": 1.9600437996930695e-05, "loss": 0.502, "step": 3301 }, { "epoch": 0.11764077168355987, "grad_norm": 1.3672557530698164, "learning_rate": 1.960011500407595e-05, "loss": 0.5659, "step": 3302 }, { "epoch": 0.11767639881005397, "grad_norm": 1.9523467993045154, "learning_rate": 1.959979188338896e-05, "loss": 0.6407, "step": 3303 }, { "epoch": 0.11771202593654809, "grad_norm": 1.3961707965840475, "learning_rate": 1.9599468634874033e-05, "loss": 0.5538, "step": 3304 }, { "epoch": 0.1177476530630422, "grad_norm": 1.5849566648041897, "learning_rate": 1.9599145258535472e-05, "loss": 0.5207, "step": 3305 }, { "epoch": 0.11778328018953631, "grad_norm": 1.9155867107813291, "learning_rate": 1.9598821754377582e-05, "loss": 0.6212, "step": 3306 }, { "epoch": 0.11781890731603042, "grad_norm": 1.695608667230396, "learning_rate": 1.9598498122404674e-05, "loss": 0.529, "step": 3307 }, { "epoch": 0.11785453444252454, "grad_norm": 1.3214308869530242, "learning_rate": 1.959817436262105e-05, "loss": 0.5195, "step": 3308 }, { "epoch": 0.11789016156901864, "grad_norm": 1.3195865016136983, "learning_rate": 1.9597850475031026e-05, "loss": 0.5352, "step": 3309 }, { "epoch": 0.11792578869551276, "grad_norm": 1.7202164474700745, "learning_rate": 1.9597526459638913e-05, "loss": 0.6204, "step": 3310 }, { "epoch": 0.11796141582200688, "grad_norm": 1.6218049557861418, "learning_rate": 1.959720231644903e-05, "loss": 0.6518, "step": 3311 }, { "epoch": 0.11799704294850098, "grad_norm": 1.6453041039650182, "learning_rate": 1.959687804546569e-05, "loss": 0.6471, "step": 3312 }, { "epoch": 0.1180326700749951, "grad_norm": 2.016603815345135, "learning_rate": 1.9596553646693212e-05, "loss": 0.6459, "step": 3313 }, { "epoch": 0.11806829720148922, "grad_norm": 2.152790036745947, "learning_rate": 1.9596229120135916e-05, "loss": 0.695, "step": 3314 }, { "epoch": 0.11810392432798332, "grad_norm": 1.3089776482320086, "learning_rate": 1.9595904465798116e-05, "loss": 0.5098, "step": 3315 }, { "epoch": 0.11813955145447744, "grad_norm": 1.34593094906989, "learning_rate": 1.9595579683684143e-05, "loss": 0.5275, "step": 3316 }, { "epoch": 0.11817517858097155, "grad_norm": 2.6540677005207067, "learning_rate": 1.9595254773798322e-05, "loss": 0.6729, "step": 3317 }, { "epoch": 0.11821080570746566, "grad_norm": 1.2804611198357612, "learning_rate": 1.9594929736144978e-05, "loss": 0.527, "step": 3318 }, { "epoch": 0.11824643283395977, "grad_norm": 1.3860858929577227, "learning_rate": 1.9594604570728433e-05, "loss": 0.5305, "step": 3319 }, { "epoch": 0.11828205996045389, "grad_norm": 2.0256087655605834, "learning_rate": 1.9594279277553024e-05, "loss": 0.6969, "step": 3320 }, { "epoch": 0.11831768708694801, "grad_norm": 1.637900777090271, "learning_rate": 1.9593953856623082e-05, "loss": 0.6068, "step": 3321 }, { "epoch": 0.11835331421344211, "grad_norm": 1.446445330813898, "learning_rate": 1.959362830794294e-05, "loss": 0.5564, "step": 3322 }, { "epoch": 0.11838894133993623, "grad_norm": 1.3071743686787032, "learning_rate": 1.9593302631516926e-05, "loss": 0.5137, "step": 3323 }, { "epoch": 0.11842456846643035, "grad_norm": 1.8331460605298815, "learning_rate": 1.9592976827349387e-05, "loss": 0.5847, "step": 3324 }, { "epoch": 0.11846019559292445, "grad_norm": 1.2707060730239834, "learning_rate": 1.9592650895444654e-05, "loss": 0.4746, "step": 3325 }, { "epoch": 0.11849582271941857, "grad_norm": 1.923075842997857, "learning_rate": 1.959232483580707e-05, "loss": 0.6371, "step": 3326 }, { "epoch": 0.11853144984591268, "grad_norm": 1.8763661310959323, "learning_rate": 1.9591998648440976e-05, "loss": 0.4989, "step": 3327 }, { "epoch": 0.11856707697240679, "grad_norm": 1.5571844710371803, "learning_rate": 1.959167233335072e-05, "loss": 0.4943, "step": 3328 }, { "epoch": 0.1186027040989009, "grad_norm": 1.6728821671202851, "learning_rate": 1.9591345890540635e-05, "loss": 0.6049, "step": 3329 }, { "epoch": 0.11863833122539502, "grad_norm": 1.6617946684100584, "learning_rate": 1.959101932001508e-05, "loss": 0.5318, "step": 3330 }, { "epoch": 0.11867395835188913, "grad_norm": 1.3332417505557543, "learning_rate": 1.95906926217784e-05, "loss": 0.5203, "step": 3331 }, { "epoch": 0.11870958547838324, "grad_norm": 1.2552204506701599, "learning_rate": 1.959036579583494e-05, "loss": 0.5173, "step": 3332 }, { "epoch": 0.11874521260487736, "grad_norm": 1.901819432443116, "learning_rate": 1.959003884218906e-05, "loss": 0.6072, "step": 3333 }, { "epoch": 0.11878083973137146, "grad_norm": 1.959829529421848, "learning_rate": 1.958971176084511e-05, "loss": 0.635, "step": 3334 }, { "epoch": 0.11881646685786558, "grad_norm": 1.281687314716188, "learning_rate": 1.9589384551807442e-05, "loss": 0.5338, "step": 3335 }, { "epoch": 0.1188520939843597, "grad_norm": 1.2203032247416892, "learning_rate": 1.958905721508042e-05, "loss": 0.5401, "step": 3336 }, { "epoch": 0.1188877211108538, "grad_norm": 1.1792586519704051, "learning_rate": 1.9588729750668394e-05, "loss": 0.5266, "step": 3337 }, { "epoch": 0.11892334823734792, "grad_norm": 1.822254879273839, "learning_rate": 1.958840215857573e-05, "loss": 0.5276, "step": 3338 }, { "epoch": 0.11895897536384203, "grad_norm": 1.5997346449908096, "learning_rate": 1.9588074438806792e-05, "loss": 0.5854, "step": 3339 }, { "epoch": 0.11899460249033614, "grad_norm": 1.3141392835687395, "learning_rate": 1.958774659136594e-05, "loss": 0.5167, "step": 3340 }, { "epoch": 0.11903022961683025, "grad_norm": 1.2356131514514341, "learning_rate": 1.9587418616257544e-05, "loss": 0.5738, "step": 3341 }, { "epoch": 0.11906585674332437, "grad_norm": 1.3776323632721286, "learning_rate": 1.9587090513485966e-05, "loss": 0.5634, "step": 3342 }, { "epoch": 0.11910148386981848, "grad_norm": 2.0212329573222787, "learning_rate": 1.9586762283055575e-05, "loss": 0.7264, "step": 3343 }, { "epoch": 0.11913711099631259, "grad_norm": 1.241939402559687, "learning_rate": 1.9586433924970742e-05, "loss": 0.5463, "step": 3344 }, { "epoch": 0.11917273812280671, "grad_norm": 1.768795174302345, "learning_rate": 1.9586105439235842e-05, "loss": 0.6477, "step": 3345 }, { "epoch": 0.11920836524930081, "grad_norm": 1.2684993702293426, "learning_rate": 1.9585776825855252e-05, "loss": 0.5463, "step": 3346 }, { "epoch": 0.11924399237579493, "grad_norm": 1.1709283563762491, "learning_rate": 1.958544808483334e-05, "loss": 0.496, "step": 3347 }, { "epoch": 0.11927961950228905, "grad_norm": 1.2175313041541342, "learning_rate": 1.9585119216174486e-05, "loss": 0.4934, "step": 3348 }, { "epoch": 0.11931524662878315, "grad_norm": 1.3149654308289231, "learning_rate": 1.958479021988307e-05, "loss": 0.5576, "step": 3349 }, { "epoch": 0.11935087375527727, "grad_norm": 1.3014365327769644, "learning_rate": 1.9584461095963475e-05, "loss": 0.5033, "step": 3350 }, { "epoch": 0.11938650088177138, "grad_norm": 1.924263998059608, "learning_rate": 1.9584131844420084e-05, "loss": 0.6018, "step": 3351 }, { "epoch": 0.11942212800826549, "grad_norm": 1.7926187896478456, "learning_rate": 1.958380246525727e-05, "loss": 0.5933, "step": 3352 }, { "epoch": 0.1194577551347596, "grad_norm": 1.7082091944775697, "learning_rate": 1.9583472958479434e-05, "loss": 0.7129, "step": 3353 }, { "epoch": 0.11949338226125372, "grad_norm": 1.4277419545116372, "learning_rate": 1.9583143324090957e-05, "loss": 0.4974, "step": 3354 }, { "epoch": 0.11952900938774783, "grad_norm": 1.3457905704235873, "learning_rate": 1.9582813562096228e-05, "loss": 0.4804, "step": 3355 }, { "epoch": 0.11956463651424194, "grad_norm": 2.2818075499306767, "learning_rate": 1.9582483672499638e-05, "loss": 0.6705, "step": 3356 }, { "epoch": 0.11960026364073606, "grad_norm": 1.3016638563539822, "learning_rate": 1.9582153655305578e-05, "loss": 0.5295, "step": 3357 }, { "epoch": 0.11963589076723016, "grad_norm": 1.3863045518718793, "learning_rate": 1.9581823510518445e-05, "loss": 0.5307, "step": 3358 }, { "epoch": 0.11967151789372428, "grad_norm": 1.6692324970635433, "learning_rate": 1.958149323814264e-05, "loss": 0.6155, "step": 3359 }, { "epoch": 0.1197071450202184, "grad_norm": 1.8531807734211958, "learning_rate": 1.9581162838182548e-05, "loss": 0.5456, "step": 3360 }, { "epoch": 0.1197427721467125, "grad_norm": 1.6013491532225939, "learning_rate": 1.958083231064258e-05, "loss": 0.513, "step": 3361 }, { "epoch": 0.11977839927320662, "grad_norm": 2.101558643130671, "learning_rate": 1.9580501655527132e-05, "loss": 0.6054, "step": 3362 }, { "epoch": 0.11981402639970073, "grad_norm": 1.7808522119407342, "learning_rate": 1.958017087284061e-05, "loss": 0.6323, "step": 3363 }, { "epoch": 0.11984965352619484, "grad_norm": 4.011728745339708, "learning_rate": 1.9579839962587413e-05, "loss": 0.6656, "step": 3364 }, { "epoch": 0.11988528065268896, "grad_norm": 1.2926518365674933, "learning_rate": 1.9579508924771954e-05, "loss": 0.5455, "step": 3365 }, { "epoch": 0.11992090777918307, "grad_norm": 1.3507095910305131, "learning_rate": 1.9579177759398638e-05, "loss": 0.4377, "step": 3366 }, { "epoch": 0.11995653490567718, "grad_norm": 2.1286735247263064, "learning_rate": 1.9578846466471875e-05, "loss": 0.6037, "step": 3367 }, { "epoch": 0.11999216203217129, "grad_norm": 1.3319545426558026, "learning_rate": 1.9578515045996074e-05, "loss": 0.5021, "step": 3368 }, { "epoch": 0.12002778915866541, "grad_norm": 2.003955952545767, "learning_rate": 1.9578183497975652e-05, "loss": 0.6689, "step": 3369 }, { "epoch": 0.12006341628515953, "grad_norm": 1.694274656008995, "learning_rate": 1.9577851822415023e-05, "loss": 0.5238, "step": 3370 }, { "epoch": 0.12009904341165363, "grad_norm": 1.3309768165229963, "learning_rate": 1.95775200193186e-05, "loss": 0.539, "step": 3371 }, { "epoch": 0.12013467053814775, "grad_norm": 1.6992702186873694, "learning_rate": 1.9577188088690804e-05, "loss": 0.6206, "step": 3372 }, { "epoch": 0.12017029766464186, "grad_norm": 1.1649824384680918, "learning_rate": 1.9576856030536055e-05, "loss": 0.4846, "step": 3373 }, { "epoch": 0.12020592479113597, "grad_norm": 2.1286581341537616, "learning_rate": 1.9576523844858774e-05, "loss": 0.6858, "step": 3374 }, { "epoch": 0.12024155191763009, "grad_norm": 1.5620678357013877, "learning_rate": 1.9576191531663383e-05, "loss": 0.5128, "step": 3375 }, { "epoch": 0.1202771790441242, "grad_norm": 2.794038355362247, "learning_rate": 1.957585909095431e-05, "loss": 0.692, "step": 3376 }, { "epoch": 0.1203128061706183, "grad_norm": 2.5333003234534472, "learning_rate": 1.9575526522735982e-05, "loss": 0.6043, "step": 3377 }, { "epoch": 0.12034843329711242, "grad_norm": 1.7180567254030523, "learning_rate": 1.957519382701282e-05, "loss": 0.5619, "step": 3378 }, { "epoch": 0.12038406042360654, "grad_norm": 1.2774421377334073, "learning_rate": 1.9574861003789264e-05, "loss": 0.5407, "step": 3379 }, { "epoch": 0.12041968755010064, "grad_norm": 1.9613321887858985, "learning_rate": 1.957452805306974e-05, "loss": 0.6282, "step": 3380 }, { "epoch": 0.12045531467659476, "grad_norm": 1.8579121750527738, "learning_rate": 1.9574194974858686e-05, "loss": 0.6794, "step": 3381 }, { "epoch": 0.12049094180308888, "grad_norm": 1.2217268152799479, "learning_rate": 1.957386176916053e-05, "loss": 0.5077, "step": 3382 }, { "epoch": 0.12052656892958298, "grad_norm": 1.6628578097432811, "learning_rate": 1.9573528435979715e-05, "loss": 0.6416, "step": 3383 }, { "epoch": 0.1205621960560771, "grad_norm": 1.3331663004777374, "learning_rate": 1.9573194975320672e-05, "loss": 0.5397, "step": 3384 }, { "epoch": 0.12059782318257121, "grad_norm": 1.7266232582073862, "learning_rate": 1.9572861387187854e-05, "loss": 0.6216, "step": 3385 }, { "epoch": 0.12063345030906532, "grad_norm": 1.3680944216626572, "learning_rate": 1.957252767158569e-05, "loss": 0.5423, "step": 3386 }, { "epoch": 0.12066907743555944, "grad_norm": 1.3001192561666852, "learning_rate": 1.9572193828518633e-05, "loss": 0.57, "step": 3387 }, { "epoch": 0.12070470456205355, "grad_norm": 1.8939053317242702, "learning_rate": 1.9571859857991124e-05, "loss": 0.6499, "step": 3388 }, { "epoch": 0.12074033168854766, "grad_norm": 1.4825919007175872, "learning_rate": 1.957152576000761e-05, "loss": 0.664, "step": 3389 }, { "epoch": 0.12077595881504177, "grad_norm": 1.313846224818073, "learning_rate": 1.957119153457254e-05, "loss": 0.533, "step": 3390 }, { "epoch": 0.12081158594153589, "grad_norm": 1.149092825774676, "learning_rate": 1.9570857181690365e-05, "loss": 0.4425, "step": 3391 }, { "epoch": 0.12084721306803, "grad_norm": 1.9560854872761522, "learning_rate": 1.957052270136554e-05, "loss": 0.7001, "step": 3392 }, { "epoch": 0.12088284019452411, "grad_norm": 1.7423121185740933, "learning_rate": 1.9570188093602512e-05, "loss": 0.6593, "step": 3393 }, { "epoch": 0.12091846732101823, "grad_norm": 1.471823482943307, "learning_rate": 1.956985335840574e-05, "loss": 0.6212, "step": 3394 }, { "epoch": 0.12095409444751233, "grad_norm": 1.8180791649958705, "learning_rate": 1.9569518495779682e-05, "loss": 0.5873, "step": 3395 }, { "epoch": 0.12098972157400645, "grad_norm": 1.4626378426996518, "learning_rate": 1.9569183505728796e-05, "loss": 0.5861, "step": 3396 }, { "epoch": 0.12102534870050057, "grad_norm": 1.8777890991136308, "learning_rate": 1.9568848388257547e-05, "loss": 0.7259, "step": 3397 }, { "epoch": 0.12106097582699467, "grad_norm": 2.040018240315422, "learning_rate": 1.9568513143370392e-05, "loss": 0.6067, "step": 3398 }, { "epoch": 0.12109660295348879, "grad_norm": 1.6489102506743505, "learning_rate": 1.9568177771071795e-05, "loss": 0.6132, "step": 3399 }, { "epoch": 0.1211322300799829, "grad_norm": 1.3832701204863842, "learning_rate": 1.9567842271366223e-05, "loss": 0.5415, "step": 3400 }, { "epoch": 0.121167857206477, "grad_norm": 1.2631554402661824, "learning_rate": 1.9567506644258144e-05, "loss": 0.5294, "step": 3401 }, { "epoch": 0.12120348433297112, "grad_norm": 1.333228762363253, "learning_rate": 1.9567170889752027e-05, "loss": 0.5313, "step": 3402 }, { "epoch": 0.12123911145946524, "grad_norm": 1.1349222401784256, "learning_rate": 1.9566835007852344e-05, "loss": 0.511, "step": 3403 }, { "epoch": 0.12127473858595934, "grad_norm": 1.9968293769719976, "learning_rate": 1.9566498998563565e-05, "loss": 0.6067, "step": 3404 }, { "epoch": 0.12131036571245346, "grad_norm": 1.2841309780590748, "learning_rate": 1.9566162861890165e-05, "loss": 0.5431, "step": 3405 }, { "epoch": 0.12134599283894758, "grad_norm": 1.4459171402405795, "learning_rate": 1.9565826597836623e-05, "loss": 0.5154, "step": 3406 }, { "epoch": 0.12138161996544168, "grad_norm": 1.21374876291042, "learning_rate": 1.9565490206407408e-05, "loss": 0.5258, "step": 3407 }, { "epoch": 0.1214172470919358, "grad_norm": 1.3796656774555975, "learning_rate": 1.9565153687607006e-05, "loss": 0.5233, "step": 3408 }, { "epoch": 0.12145287421842992, "grad_norm": 2.298105588354057, "learning_rate": 1.95648170414399e-05, "loss": 0.5363, "step": 3409 }, { "epoch": 0.12148850134492402, "grad_norm": 1.734589132445493, "learning_rate": 1.956448026791057e-05, "loss": 0.6062, "step": 3410 }, { "epoch": 0.12152412847141814, "grad_norm": 1.7301180734976007, "learning_rate": 1.9564143367023498e-05, "loss": 0.6777, "step": 3411 }, { "epoch": 0.12155975559791225, "grad_norm": 1.225379704066257, "learning_rate": 1.9563806338783174e-05, "loss": 0.5711, "step": 3412 }, { "epoch": 0.12159538272440636, "grad_norm": 1.3712651064864443, "learning_rate": 1.9563469183194082e-05, "loss": 0.5741, "step": 3413 }, { "epoch": 0.12163100985090047, "grad_norm": 2.500958577741217, "learning_rate": 1.9563131900260715e-05, "loss": 0.6383, "step": 3414 }, { "epoch": 0.12166663697739459, "grad_norm": 1.218667324684152, "learning_rate": 1.9562794489987562e-05, "loss": 0.5223, "step": 3415 }, { "epoch": 0.12170226410388871, "grad_norm": 1.1662145149110452, "learning_rate": 1.9562456952379117e-05, "loss": 0.4932, "step": 3416 }, { "epoch": 0.12173789123038281, "grad_norm": 1.3409959806715193, "learning_rate": 1.9562119287439874e-05, "loss": 0.5298, "step": 3417 }, { "epoch": 0.12177351835687693, "grad_norm": 1.1439015623102535, "learning_rate": 1.9561781495174327e-05, "loss": 0.5002, "step": 3418 }, { "epoch": 0.12180914548337105, "grad_norm": 1.267442669173275, "learning_rate": 1.9561443575586975e-05, "loss": 0.5477, "step": 3419 }, { "epoch": 0.12184477260986515, "grad_norm": 1.4017450577688448, "learning_rate": 1.956110552868232e-05, "loss": 0.534, "step": 3420 }, { "epoch": 0.12188039973635927, "grad_norm": 2.0060170180914088, "learning_rate": 1.9560767354464864e-05, "loss": 0.5865, "step": 3421 }, { "epoch": 0.12191602686285338, "grad_norm": 1.1481502021694205, "learning_rate": 1.9560429052939108e-05, "loss": 0.4944, "step": 3422 }, { "epoch": 0.12195165398934749, "grad_norm": 1.6445828955681072, "learning_rate": 1.9560090624109554e-05, "loss": 0.6829, "step": 3423 }, { "epoch": 0.1219872811158416, "grad_norm": 2.2041951317633215, "learning_rate": 1.9559752067980717e-05, "loss": 0.743, "step": 3424 }, { "epoch": 0.12202290824233572, "grad_norm": 1.8230580315963998, "learning_rate": 1.9559413384557095e-05, "loss": 0.6773, "step": 3425 }, { "epoch": 0.12205853536882982, "grad_norm": 1.3896344033398131, "learning_rate": 1.95590745738432e-05, "loss": 0.5489, "step": 3426 }, { "epoch": 0.12209416249532394, "grad_norm": 1.2688430084100306, "learning_rate": 1.955873563584355e-05, "loss": 0.5241, "step": 3427 }, { "epoch": 0.12212978962181806, "grad_norm": 1.2979569387754684, "learning_rate": 1.955839657056265e-05, "loss": 0.5208, "step": 3428 }, { "epoch": 0.12216541674831216, "grad_norm": 1.6927823947023601, "learning_rate": 1.9558057378005025e-05, "loss": 0.6508, "step": 3429 }, { "epoch": 0.12220104387480628, "grad_norm": 2.2269354226107447, "learning_rate": 1.955771805817518e-05, "loss": 0.6027, "step": 3430 }, { "epoch": 0.1222366710013004, "grad_norm": 1.2892697927485315, "learning_rate": 1.955737861107764e-05, "loss": 0.5154, "step": 3431 }, { "epoch": 0.1222722981277945, "grad_norm": 1.2784882143518626, "learning_rate": 1.9557039036716926e-05, "loss": 0.4549, "step": 3432 }, { "epoch": 0.12230792525428862, "grad_norm": 1.3344878860209435, "learning_rate": 1.9556699335097553e-05, "loss": 0.5298, "step": 3433 }, { "epoch": 0.12234355238078273, "grad_norm": 1.7959323180419187, "learning_rate": 1.955635950622405e-05, "loss": 0.5698, "step": 3434 }, { "epoch": 0.12237917950727684, "grad_norm": 1.207927608791049, "learning_rate": 1.955601955010094e-05, "loss": 0.5189, "step": 3435 }, { "epoch": 0.12241480663377095, "grad_norm": 1.9816308401494713, "learning_rate": 1.9555679466732754e-05, "loss": 0.6305, "step": 3436 }, { "epoch": 0.12245043376026507, "grad_norm": 1.2105502418828342, "learning_rate": 1.9555339256124012e-05, "loss": 0.5401, "step": 3437 }, { "epoch": 0.12248606088675917, "grad_norm": 2.1964274020513423, "learning_rate": 1.9554998918279253e-05, "loss": 0.5905, "step": 3438 }, { "epoch": 0.12252168801325329, "grad_norm": 2.415021224679646, "learning_rate": 1.9554658453203003e-05, "loss": 0.7198, "step": 3439 }, { "epoch": 0.12255731513974741, "grad_norm": 1.6496121901961704, "learning_rate": 1.9554317860899798e-05, "loss": 0.6172, "step": 3440 }, { "epoch": 0.12259294226624151, "grad_norm": 2.1084952718425245, "learning_rate": 1.9553977141374173e-05, "loss": 0.5528, "step": 3441 }, { "epoch": 0.12262856939273563, "grad_norm": 2.2157575355780086, "learning_rate": 1.9553636294630663e-05, "loss": 0.6922, "step": 3442 }, { "epoch": 0.12266419651922975, "grad_norm": 1.293398338421822, "learning_rate": 1.955329532067381e-05, "loss": 0.5255, "step": 3443 }, { "epoch": 0.12269982364572385, "grad_norm": 1.540398269914083, "learning_rate": 1.955295421950815e-05, "loss": 0.5577, "step": 3444 }, { "epoch": 0.12273545077221797, "grad_norm": 1.6398270273242364, "learning_rate": 1.9552612991138228e-05, "loss": 0.5384, "step": 3445 }, { "epoch": 0.12277107789871208, "grad_norm": 1.1802120384182373, "learning_rate": 1.955227163556859e-05, "loss": 0.4791, "step": 3446 }, { "epoch": 0.12280670502520619, "grad_norm": 1.8992675231452525, "learning_rate": 1.9551930152803775e-05, "loss": 0.6142, "step": 3447 }, { "epoch": 0.1228423321517003, "grad_norm": 1.6899549453215204, "learning_rate": 1.9551588542848337e-05, "loss": 0.5673, "step": 3448 }, { "epoch": 0.12287795927819442, "grad_norm": 1.2394763371239876, "learning_rate": 1.955124680570682e-05, "loss": 0.5123, "step": 3449 }, { "epoch": 0.12291358640468852, "grad_norm": 2.5054717257876478, "learning_rate": 1.955090494138377e-05, "loss": 0.5982, "step": 3450 }, { "epoch": 0.12294921353118264, "grad_norm": 1.278219715717403, "learning_rate": 1.9550562949883755e-05, "loss": 0.5613, "step": 3451 }, { "epoch": 0.12298484065767676, "grad_norm": 1.4300982276016472, "learning_rate": 1.9550220831211313e-05, "loss": 0.5373, "step": 3452 }, { "epoch": 0.12302046778417086, "grad_norm": 1.7726372475679797, "learning_rate": 1.9549878585371006e-05, "loss": 0.6021, "step": 3453 }, { "epoch": 0.12305609491066498, "grad_norm": 1.6620339336649155, "learning_rate": 1.9549536212367394e-05, "loss": 0.6739, "step": 3454 }, { "epoch": 0.1230917220371591, "grad_norm": 1.1625883927393132, "learning_rate": 1.9549193712205028e-05, "loss": 0.5458, "step": 3455 }, { "epoch": 0.1231273491636532, "grad_norm": 1.2290112254136532, "learning_rate": 1.9548851084888478e-05, "loss": 0.496, "step": 3456 }, { "epoch": 0.12316297629014732, "grad_norm": 2.055642350628586, "learning_rate": 1.9548508330422297e-05, "loss": 0.6291, "step": 3457 }, { "epoch": 0.12319860341664143, "grad_norm": 1.2568639152971517, "learning_rate": 1.9548165448811054e-05, "loss": 0.4971, "step": 3458 }, { "epoch": 0.12323423054313554, "grad_norm": 1.6712737540236091, "learning_rate": 1.9547822440059318e-05, "loss": 0.6556, "step": 3459 }, { "epoch": 0.12326985766962965, "grad_norm": 1.6058858392427253, "learning_rate": 1.954747930417165e-05, "loss": 0.6366, "step": 3460 }, { "epoch": 0.12330548479612377, "grad_norm": 1.7594917679235287, "learning_rate": 1.954713604115262e-05, "loss": 0.5616, "step": 3461 }, { "epoch": 0.12334111192261789, "grad_norm": 2.219570194495588, "learning_rate": 1.9546792651006805e-05, "loss": 0.6604, "step": 3462 }, { "epoch": 0.12337673904911199, "grad_norm": 1.3579500779904001, "learning_rate": 1.954644913373877e-05, "loss": 0.5643, "step": 3463 }, { "epoch": 0.12341236617560611, "grad_norm": 1.1785745598040822, "learning_rate": 1.954610548935309e-05, "loss": 0.5196, "step": 3464 }, { "epoch": 0.12344799330210023, "grad_norm": 1.5628588315204681, "learning_rate": 1.9545761717854346e-05, "loss": 0.6424, "step": 3465 }, { "epoch": 0.12348362042859433, "grad_norm": 1.4509413072027726, "learning_rate": 1.9545417819247113e-05, "loss": 0.579, "step": 3466 }, { "epoch": 0.12351924755508845, "grad_norm": 1.9849043498312746, "learning_rate": 1.9545073793535972e-05, "loss": 0.5802, "step": 3467 }, { "epoch": 0.12355487468158256, "grad_norm": 1.2643601272107141, "learning_rate": 1.95447296407255e-05, "loss": 0.5917, "step": 3468 }, { "epoch": 0.12359050180807667, "grad_norm": 1.7242765260320925, "learning_rate": 1.954438536082028e-05, "loss": 0.6558, "step": 3469 }, { "epoch": 0.12362612893457078, "grad_norm": 1.3144585573879624, "learning_rate": 1.95440409538249e-05, "loss": 0.612, "step": 3470 }, { "epoch": 0.1236617560610649, "grad_norm": 1.8992064769458754, "learning_rate": 1.954369641974394e-05, "loss": 0.5869, "step": 3471 }, { "epoch": 0.123697383187559, "grad_norm": 1.1818138115655334, "learning_rate": 1.9543351758581995e-05, "loss": 0.5255, "step": 3472 }, { "epoch": 0.12373301031405312, "grad_norm": 2.0535680379815355, "learning_rate": 1.954300697034365e-05, "loss": 0.6694, "step": 3473 }, { "epoch": 0.12376863744054724, "grad_norm": 1.7457556550394204, "learning_rate": 1.9542662055033497e-05, "loss": 0.5476, "step": 3474 }, { "epoch": 0.12380426456704134, "grad_norm": 1.861456993037634, "learning_rate": 1.954231701265613e-05, "loss": 0.622, "step": 3475 }, { "epoch": 0.12383989169353546, "grad_norm": 1.9093998042374476, "learning_rate": 1.954197184321614e-05, "loss": 0.6129, "step": 3476 }, { "epoch": 0.12387551882002958, "grad_norm": 2.0892466660046556, "learning_rate": 1.9541626546718128e-05, "loss": 0.6126, "step": 3477 }, { "epoch": 0.12391114594652368, "grad_norm": 1.7771979579944952, "learning_rate": 1.9541281123166685e-05, "loss": 0.5654, "step": 3478 }, { "epoch": 0.1239467730730178, "grad_norm": 1.5429325874700677, "learning_rate": 1.954093557256642e-05, "loss": 0.6909, "step": 3479 }, { "epoch": 0.12398240019951191, "grad_norm": 1.4091681479591787, "learning_rate": 1.9540589894921927e-05, "loss": 0.5615, "step": 3480 }, { "epoch": 0.12401802732600602, "grad_norm": 2.8923614619978384, "learning_rate": 1.9540244090237808e-05, "loss": 0.6718, "step": 3481 }, { "epoch": 0.12405365445250013, "grad_norm": 1.3247111920514263, "learning_rate": 1.9539898158518676e-05, "loss": 0.5393, "step": 3482 }, { "epoch": 0.12408928157899425, "grad_norm": 1.4914880579970828, "learning_rate": 1.9539552099769128e-05, "loss": 0.5599, "step": 3483 }, { "epoch": 0.12412490870548835, "grad_norm": 1.1946256709132865, "learning_rate": 1.9539205913993776e-05, "loss": 0.5141, "step": 3484 }, { "epoch": 0.12416053583198247, "grad_norm": 1.455385222772528, "learning_rate": 1.9538859601197235e-05, "loss": 0.5264, "step": 3485 }, { "epoch": 0.12419616295847659, "grad_norm": 1.2439575209180629, "learning_rate": 1.9538513161384105e-05, "loss": 0.4933, "step": 3486 }, { "epoch": 0.12423179008497069, "grad_norm": 1.3808606083056452, "learning_rate": 1.9538166594559008e-05, "loss": 0.5311, "step": 3487 }, { "epoch": 0.12426741721146481, "grad_norm": 1.2271850334501917, "learning_rate": 1.953781990072656e-05, "loss": 0.49, "step": 3488 }, { "epoch": 0.12430304433795893, "grad_norm": 1.780114263639673, "learning_rate": 1.9537473079891365e-05, "loss": 0.62, "step": 3489 }, { "epoch": 0.12433867146445303, "grad_norm": 1.3935122522668242, "learning_rate": 1.9537126132058055e-05, "loss": 0.5928, "step": 3490 }, { "epoch": 0.12437429859094715, "grad_norm": 1.2667291829146334, "learning_rate": 1.9536779057231243e-05, "loss": 0.5184, "step": 3491 }, { "epoch": 0.12440992571744126, "grad_norm": 1.2175199904487966, "learning_rate": 1.953643185541555e-05, "loss": 0.5149, "step": 3492 }, { "epoch": 0.12444555284393537, "grad_norm": 1.1918801135520798, "learning_rate": 1.953608452661561e-05, "loss": 0.4939, "step": 3493 }, { "epoch": 0.12448117997042948, "grad_norm": 1.8624714743906188, "learning_rate": 1.9535737070836028e-05, "loss": 0.5944, "step": 3494 }, { "epoch": 0.1245168070969236, "grad_norm": 1.7610611380941765, "learning_rate": 1.9535389488081447e-05, "loss": 0.6818, "step": 3495 }, { "epoch": 0.1245524342234177, "grad_norm": 1.8204770179717067, "learning_rate": 1.9535041778356492e-05, "loss": 0.6639, "step": 3496 }, { "epoch": 0.12458806134991182, "grad_norm": 1.8286026210958555, "learning_rate": 1.9534693941665788e-05, "loss": 0.6184, "step": 3497 }, { "epoch": 0.12462368847640594, "grad_norm": 1.4150542101282206, "learning_rate": 1.9534345978013972e-05, "loss": 0.5392, "step": 3498 }, { "epoch": 0.12465931560290004, "grad_norm": 1.978593511805678, "learning_rate": 1.9533997887405673e-05, "loss": 0.664, "step": 3499 }, { "epoch": 0.12469494272939416, "grad_norm": 1.2741336404898973, "learning_rate": 1.9533649669845533e-05, "loss": 0.5581, "step": 3500 }, { "epoch": 0.12473056985588828, "grad_norm": 1.333902285433046, "learning_rate": 1.9533301325338176e-05, "loss": 0.5639, "step": 3501 }, { "epoch": 0.12476619698238238, "grad_norm": 2.003669745263457, "learning_rate": 1.9532952853888255e-05, "loss": 0.6108, "step": 3502 }, { "epoch": 0.1248018241088765, "grad_norm": 1.2581964805435548, "learning_rate": 1.95326042555004e-05, "loss": 0.5785, "step": 3503 }, { "epoch": 0.12483745123537061, "grad_norm": 1.2505955790332532, "learning_rate": 1.953225553017926e-05, "loss": 0.5506, "step": 3504 }, { "epoch": 0.12487307836186472, "grad_norm": 1.3147887854376215, "learning_rate": 1.9531906677929472e-05, "loss": 0.4975, "step": 3505 }, { "epoch": 0.12490870548835883, "grad_norm": 1.1976163240872404, "learning_rate": 1.9531557698755686e-05, "loss": 0.5313, "step": 3506 }, { "epoch": 0.12494433261485295, "grad_norm": 1.2672470295585079, "learning_rate": 1.9531208592662545e-05, "loss": 0.6069, "step": 3507 }, { "epoch": 0.12497995974134705, "grad_norm": 1.386359128814792, "learning_rate": 1.95308593596547e-05, "loss": 0.5465, "step": 3508 }, { "epoch": 0.12501558686784117, "grad_norm": 1.6303264250485152, "learning_rate": 1.9530509999736803e-05, "loss": 0.5979, "step": 3509 }, { "epoch": 0.12505121399433528, "grad_norm": 1.9072934676809485, "learning_rate": 1.9530160512913502e-05, "loss": 0.6306, "step": 3510 }, { "epoch": 0.1250868411208294, "grad_norm": 1.411779510309459, "learning_rate": 1.9529810899189453e-05, "loss": 0.5725, "step": 3511 }, { "epoch": 0.1251224682473235, "grad_norm": 1.2146341280133475, "learning_rate": 1.952946115856931e-05, "loss": 0.5491, "step": 3512 }, { "epoch": 0.1251580953738176, "grad_norm": 1.8564056935128397, "learning_rate": 1.9529111291057735e-05, "loss": 0.6546, "step": 3513 }, { "epoch": 0.12519372250031174, "grad_norm": 1.1644658121636362, "learning_rate": 1.952876129665938e-05, "loss": 0.4466, "step": 3514 }, { "epoch": 0.12522934962680585, "grad_norm": 1.19903158821619, "learning_rate": 1.9528411175378905e-05, "loss": 0.4561, "step": 3515 }, { "epoch": 0.12526497675329995, "grad_norm": 1.6305485626696958, "learning_rate": 1.952806092722098e-05, "loss": 0.7259, "step": 3516 }, { "epoch": 0.12530060387979408, "grad_norm": 1.2924905266636353, "learning_rate": 1.9527710552190262e-05, "loss": 0.5025, "step": 3517 }, { "epoch": 0.12533623100628818, "grad_norm": 1.3198633222511185, "learning_rate": 1.952736005029142e-05, "loss": 0.5834, "step": 3518 }, { "epoch": 0.1253718581327823, "grad_norm": 1.3138262547593724, "learning_rate": 1.9527009421529117e-05, "loss": 0.517, "step": 3519 }, { "epoch": 0.12540748525927642, "grad_norm": 1.9591023154123166, "learning_rate": 1.952665866590803e-05, "loss": 0.5631, "step": 3520 }, { "epoch": 0.12544311238577052, "grad_norm": 1.978692096680652, "learning_rate": 1.9526307783432817e-05, "loss": 0.6681, "step": 3521 }, { "epoch": 0.12547873951226463, "grad_norm": 1.9087347499156886, "learning_rate": 1.9525956774108163e-05, "loss": 0.6332, "step": 3522 }, { "epoch": 0.12551436663875876, "grad_norm": 1.626360045552108, "learning_rate": 1.9525605637938733e-05, "loss": 0.501, "step": 3523 }, { "epoch": 0.12554999376525286, "grad_norm": 1.128575039655937, "learning_rate": 1.9525254374929205e-05, "loss": 0.4882, "step": 3524 }, { "epoch": 0.12558562089174696, "grad_norm": 1.3956468033996503, "learning_rate": 1.952490298508426e-05, "loss": 0.5214, "step": 3525 }, { "epoch": 0.1256212480182411, "grad_norm": 1.2749913719033377, "learning_rate": 1.952455146840857e-05, "loss": 0.5064, "step": 3526 }, { "epoch": 0.1256568751447352, "grad_norm": 1.1678027605061845, "learning_rate": 1.9524199824906826e-05, "loss": 0.5066, "step": 3527 }, { "epoch": 0.1256925022712293, "grad_norm": 2.735570435899517, "learning_rate": 1.9523848054583703e-05, "loss": 0.5937, "step": 3528 }, { "epoch": 0.12572812939772343, "grad_norm": 1.7260306044457296, "learning_rate": 1.9523496157443885e-05, "loss": 0.5812, "step": 3529 }, { "epoch": 0.12576375652421753, "grad_norm": 1.197741651655593, "learning_rate": 1.9523144133492057e-05, "loss": 0.5184, "step": 3530 }, { "epoch": 0.12579938365071164, "grad_norm": 2.4873741918335197, "learning_rate": 1.9522791982732912e-05, "loss": 0.6505, "step": 3531 }, { "epoch": 0.12583501077720577, "grad_norm": 1.746091535390245, "learning_rate": 1.9522439705171132e-05, "loss": 0.6659, "step": 3532 }, { "epoch": 0.12587063790369987, "grad_norm": 1.890233403356529, "learning_rate": 1.9522087300811417e-05, "loss": 0.6616, "step": 3533 }, { "epoch": 0.125906265030194, "grad_norm": 1.7236636557460208, "learning_rate": 1.952173476965845e-05, "loss": 0.6011, "step": 3534 }, { "epoch": 0.1259418921566881, "grad_norm": 2.014730509709412, "learning_rate": 1.952138211171693e-05, "loss": 0.6874, "step": 3535 }, { "epoch": 0.1259775192831822, "grad_norm": 1.220142596220602, "learning_rate": 1.952102932699155e-05, "loss": 0.5499, "step": 3536 }, { "epoch": 0.12601314640967634, "grad_norm": 1.280496041144509, "learning_rate": 1.9520676415487013e-05, "loss": 0.5076, "step": 3537 }, { "epoch": 0.12604877353617044, "grad_norm": 2.0556297963603543, "learning_rate": 1.9520323377208017e-05, "loss": 0.7024, "step": 3538 }, { "epoch": 0.12608440066266455, "grad_norm": 1.9174010794378031, "learning_rate": 1.9519970212159257e-05, "loss": 0.6978, "step": 3539 }, { "epoch": 0.12612002778915868, "grad_norm": 1.2951337521235573, "learning_rate": 1.951961692034544e-05, "loss": 0.5342, "step": 3540 }, { "epoch": 0.12615565491565278, "grad_norm": 1.2245490167194006, "learning_rate": 1.951926350177127e-05, "loss": 0.4907, "step": 3541 }, { "epoch": 0.12619128204214689, "grad_norm": 2.5120959102597804, "learning_rate": 1.9518909956441454e-05, "loss": 0.65, "step": 3542 }, { "epoch": 0.12622690916864102, "grad_norm": 1.1291668846089853, "learning_rate": 1.9518556284360696e-05, "loss": 0.5336, "step": 3543 }, { "epoch": 0.12626253629513512, "grad_norm": 1.6216271396720494, "learning_rate": 1.951820248553371e-05, "loss": 0.5977, "step": 3544 }, { "epoch": 0.12629816342162922, "grad_norm": 1.927997145651542, "learning_rate": 1.9517848559965207e-05, "loss": 0.6102, "step": 3545 }, { "epoch": 0.12633379054812335, "grad_norm": 1.4799761477750573, "learning_rate": 1.9517494507659894e-05, "loss": 0.5642, "step": 3546 }, { "epoch": 0.12636941767461746, "grad_norm": 1.2523712516658179, "learning_rate": 1.951714032862249e-05, "loss": 0.4942, "step": 3547 }, { "epoch": 0.12640504480111156, "grad_norm": 1.4955356429865345, "learning_rate": 1.9516786022857714e-05, "loss": 0.5721, "step": 3548 }, { "epoch": 0.1264406719276057, "grad_norm": 2.2502079630983602, "learning_rate": 1.951643159037028e-05, "loss": 0.6772, "step": 3549 }, { "epoch": 0.1264762990540998, "grad_norm": 1.2781815234373728, "learning_rate": 1.9516077031164905e-05, "loss": 0.53, "step": 3550 }, { "epoch": 0.1265119261805939, "grad_norm": 2.1414897365013212, "learning_rate": 1.951572234524631e-05, "loss": 0.6797, "step": 3551 }, { "epoch": 0.12654755330708803, "grad_norm": 1.8885835498707517, "learning_rate": 1.9515367532619226e-05, "loss": 0.5075, "step": 3552 }, { "epoch": 0.12658318043358213, "grad_norm": 1.3503733849721176, "learning_rate": 1.9515012593288373e-05, "loss": 0.574, "step": 3553 }, { "epoch": 0.12661880756007624, "grad_norm": 2.0380416137138546, "learning_rate": 1.9514657527258474e-05, "loss": 0.6438, "step": 3554 }, { "epoch": 0.12665443468657037, "grad_norm": 1.7598638731070808, "learning_rate": 1.951430233453426e-05, "loss": 0.6872, "step": 3555 }, { "epoch": 0.12669006181306447, "grad_norm": 1.747406948285889, "learning_rate": 1.951394701512046e-05, "loss": 0.5326, "step": 3556 }, { "epoch": 0.12672568893955857, "grad_norm": 1.201836994801211, "learning_rate": 1.9513591569021807e-05, "loss": 0.5317, "step": 3557 }, { "epoch": 0.1267613160660527, "grad_norm": 1.5789566299375903, "learning_rate": 1.9513235996243033e-05, "loss": 0.6448, "step": 3558 }, { "epoch": 0.1267969431925468, "grad_norm": 1.3412986286817892, "learning_rate": 1.951288029678887e-05, "loss": 0.5387, "step": 3559 }, { "epoch": 0.1268325703190409, "grad_norm": 1.3768621060420818, "learning_rate": 1.9512524470664058e-05, "loss": 0.5962, "step": 3560 }, { "epoch": 0.12686819744553504, "grad_norm": 2.893735257990952, "learning_rate": 1.9512168517873333e-05, "loss": 0.6677, "step": 3561 }, { "epoch": 0.12690382457202914, "grad_norm": 1.2469058481420048, "learning_rate": 1.951181243842144e-05, "loss": 0.5268, "step": 3562 }, { "epoch": 0.12693945169852325, "grad_norm": 1.7065184889528104, "learning_rate": 1.951145623231311e-05, "loss": 0.5895, "step": 3563 }, { "epoch": 0.12697507882501738, "grad_norm": 1.7918841530175453, "learning_rate": 1.9511099899553097e-05, "loss": 0.6002, "step": 3564 }, { "epoch": 0.12701070595151148, "grad_norm": 1.2872016408972868, "learning_rate": 1.9510743440146136e-05, "loss": 0.5091, "step": 3565 }, { "epoch": 0.12704633307800559, "grad_norm": 1.2467185729337347, "learning_rate": 1.9510386854096982e-05, "loss": 0.5577, "step": 3566 }, { "epoch": 0.12708196020449972, "grad_norm": 1.8460175962453893, "learning_rate": 1.951003014141038e-05, "loss": 0.6245, "step": 3567 }, { "epoch": 0.12711758733099382, "grad_norm": 1.3684340407483766, "learning_rate": 1.9509673302091075e-05, "loss": 0.5736, "step": 3568 }, { "epoch": 0.12715321445748792, "grad_norm": 1.2915668027654663, "learning_rate": 1.950931633614383e-05, "loss": 0.5605, "step": 3569 }, { "epoch": 0.12718884158398205, "grad_norm": 1.24042013429579, "learning_rate": 1.9508959243573387e-05, "loss": 0.5249, "step": 3570 }, { "epoch": 0.12722446871047616, "grad_norm": 2.5236867875957234, "learning_rate": 1.9508602024384507e-05, "loss": 0.6013, "step": 3571 }, { "epoch": 0.12726009583697026, "grad_norm": 1.2661940166242616, "learning_rate": 1.950824467858194e-05, "loss": 0.5102, "step": 3572 }, { "epoch": 0.1272957229634644, "grad_norm": 4.822773187415467, "learning_rate": 1.9507887206170456e-05, "loss": 0.6747, "step": 3573 }, { "epoch": 0.1273313500899585, "grad_norm": 2.2098130617049954, "learning_rate": 1.9507529607154807e-05, "loss": 0.623, "step": 3574 }, { "epoch": 0.1273669772164526, "grad_norm": 1.3895567483001754, "learning_rate": 1.9507171881539754e-05, "loss": 0.5106, "step": 3575 }, { "epoch": 0.12740260434294673, "grad_norm": 2.3627774036593263, "learning_rate": 1.950681402933006e-05, "loss": 0.6705, "step": 3576 }, { "epoch": 0.12743823146944083, "grad_norm": 2.7356017742502634, "learning_rate": 1.95064560505305e-05, "loss": 0.6067, "step": 3577 }, { "epoch": 0.12747385859593494, "grad_norm": 1.4690431919042768, "learning_rate": 1.9506097945145825e-05, "loss": 0.562, "step": 3578 }, { "epoch": 0.12750948572242907, "grad_norm": 1.7013091816656336, "learning_rate": 1.9505739713180815e-05, "loss": 0.5238, "step": 3579 }, { "epoch": 0.12754511284892317, "grad_norm": 2.617395542316548, "learning_rate": 1.9505381354640236e-05, "loss": 0.4689, "step": 3580 }, { "epoch": 0.12758073997541727, "grad_norm": 1.4522610288094575, "learning_rate": 1.9505022869528862e-05, "loss": 0.5691, "step": 3581 }, { "epoch": 0.1276163671019114, "grad_norm": 2.9667664790613566, "learning_rate": 1.9504664257851464e-05, "loss": 0.5986, "step": 3582 }, { "epoch": 0.1276519942284055, "grad_norm": 1.2539339831721048, "learning_rate": 1.9504305519612815e-05, "loss": 0.5333, "step": 3583 }, { "epoch": 0.1276876213548996, "grad_norm": 1.3664569288477855, "learning_rate": 1.95039466548177e-05, "loss": 0.6391, "step": 3584 }, { "epoch": 0.12772324848139374, "grad_norm": 1.2458139676917612, "learning_rate": 1.9503587663470887e-05, "loss": 0.4994, "step": 3585 }, { "epoch": 0.12775887560788785, "grad_norm": 1.263647200599049, "learning_rate": 1.9503228545577167e-05, "loss": 0.5367, "step": 3586 }, { "epoch": 0.12779450273438195, "grad_norm": 2.0698067906962105, "learning_rate": 1.9502869301141315e-05, "loss": 0.6037, "step": 3587 }, { "epoch": 0.12783012986087608, "grad_norm": 1.333533051834814, "learning_rate": 1.9502509930168113e-05, "loss": 0.5817, "step": 3588 }, { "epoch": 0.12786575698737018, "grad_norm": 1.6801494357926297, "learning_rate": 1.9502150432662354e-05, "loss": 0.5778, "step": 3589 }, { "epoch": 0.12790138411386429, "grad_norm": 1.269219339541553, "learning_rate": 1.9501790808628815e-05, "loss": 0.515, "step": 3590 }, { "epoch": 0.12793701124035842, "grad_norm": 1.2840053742882866, "learning_rate": 1.9501431058072295e-05, "loss": 0.5005, "step": 3591 }, { "epoch": 0.12797263836685252, "grad_norm": 1.2458107688735507, "learning_rate": 1.9501071180997577e-05, "loss": 0.5377, "step": 3592 }, { "epoch": 0.12800826549334662, "grad_norm": 1.278712112618362, "learning_rate": 1.9500711177409456e-05, "loss": 0.5047, "step": 3593 }, { "epoch": 0.12804389261984075, "grad_norm": 3.0289006670748724, "learning_rate": 1.9500351047312726e-05, "loss": 0.6024, "step": 3594 }, { "epoch": 0.12807951974633486, "grad_norm": 1.4306246517006773, "learning_rate": 1.9499990790712177e-05, "loss": 0.5484, "step": 3595 }, { "epoch": 0.12811514687282896, "grad_norm": 1.8225825426532198, "learning_rate": 1.9499630407612614e-05, "loss": 0.6181, "step": 3596 }, { "epoch": 0.1281507739993231, "grad_norm": 2.020231451576869, "learning_rate": 1.9499269898018832e-05, "loss": 0.6064, "step": 3597 }, { "epoch": 0.1281864011258172, "grad_norm": 1.208088212243124, "learning_rate": 1.9498909261935634e-05, "loss": 0.5429, "step": 3598 }, { "epoch": 0.1282220282523113, "grad_norm": 1.5555014278158272, "learning_rate": 1.9498548499367818e-05, "loss": 0.5681, "step": 3599 }, { "epoch": 0.12825765537880543, "grad_norm": 2.0965033529766046, "learning_rate": 1.9498187610320188e-05, "loss": 0.6801, "step": 3600 }, { "epoch": 0.12829328250529953, "grad_norm": 2.1828418672173417, "learning_rate": 1.9497826594797553e-05, "loss": 0.6765, "step": 3601 }, { "epoch": 0.12832890963179364, "grad_norm": 1.8321384218253558, "learning_rate": 1.9497465452804717e-05, "loss": 0.5748, "step": 3602 }, { "epoch": 0.12836453675828777, "grad_norm": 3.487625546654763, "learning_rate": 1.949710418434649e-05, "loss": 0.6275, "step": 3603 }, { "epoch": 0.12840016388478187, "grad_norm": 1.624982396274398, "learning_rate": 1.9496742789427684e-05, "loss": 0.6034, "step": 3604 }, { "epoch": 0.12843579101127597, "grad_norm": 1.3908117694914346, "learning_rate": 1.949638126805311e-05, "loss": 0.5396, "step": 3605 }, { "epoch": 0.1284714181377701, "grad_norm": 1.381466473917263, "learning_rate": 1.9496019620227583e-05, "loss": 0.5499, "step": 3606 }, { "epoch": 0.1285070452642642, "grad_norm": 2.410654688119109, "learning_rate": 1.9495657845955914e-05, "loss": 0.5694, "step": 3607 }, { "epoch": 0.1285426723907583, "grad_norm": 1.2405673058357936, "learning_rate": 1.9495295945242927e-05, "loss": 0.5065, "step": 3608 }, { "epoch": 0.12857829951725244, "grad_norm": 2.353807294829631, "learning_rate": 1.9494933918093437e-05, "loss": 0.5785, "step": 3609 }, { "epoch": 0.12861392664374655, "grad_norm": 1.3782431329221085, "learning_rate": 1.9494571764512265e-05, "loss": 0.5934, "step": 3610 }, { "epoch": 0.12864955377024065, "grad_norm": 1.266538424740709, "learning_rate": 1.9494209484504233e-05, "loss": 0.5203, "step": 3611 }, { "epoch": 0.12868518089673478, "grad_norm": 1.6357027877135468, "learning_rate": 1.9493847078074166e-05, "loss": 0.6657, "step": 3612 }, { "epoch": 0.12872080802322888, "grad_norm": 1.3016763587813933, "learning_rate": 1.9493484545226887e-05, "loss": 0.5393, "step": 3613 }, { "epoch": 0.128756435149723, "grad_norm": 1.2335466170503182, "learning_rate": 1.949312188596723e-05, "loss": 0.5814, "step": 3614 }, { "epoch": 0.12879206227621712, "grad_norm": 1.4590781060341418, "learning_rate": 1.949275910030002e-05, "loss": 0.5406, "step": 3615 }, { "epoch": 0.12882768940271122, "grad_norm": 1.29833494388846, "learning_rate": 1.9492396188230083e-05, "loss": 0.5193, "step": 3616 }, { "epoch": 0.12886331652920532, "grad_norm": 1.645696314784017, "learning_rate": 1.9492033149762258e-05, "loss": 0.6741, "step": 3617 }, { "epoch": 0.12889894365569946, "grad_norm": 1.161321403512333, "learning_rate": 1.9491669984901377e-05, "loss": 0.5171, "step": 3618 }, { "epoch": 0.12893457078219356, "grad_norm": 2.227435646425057, "learning_rate": 1.9491306693652276e-05, "loss": 0.5698, "step": 3619 }, { "epoch": 0.12897019790868766, "grad_norm": 1.2172793408597096, "learning_rate": 1.949094327601979e-05, "loss": 0.5197, "step": 3620 }, { "epoch": 0.1290058250351818, "grad_norm": 2.5433386229464796, "learning_rate": 1.9490579732008765e-05, "loss": 0.5594, "step": 3621 }, { "epoch": 0.1290414521616759, "grad_norm": 1.4909488177284722, "learning_rate": 1.9490216061624038e-05, "loss": 0.5349, "step": 3622 }, { "epoch": 0.12907707928817, "grad_norm": 1.1880830910838476, "learning_rate": 1.948985226487045e-05, "loss": 0.5018, "step": 3623 }, { "epoch": 0.12911270641466413, "grad_norm": 1.1424477246409268, "learning_rate": 1.948948834175284e-05, "loss": 0.5141, "step": 3624 }, { "epoch": 0.12914833354115823, "grad_norm": 1.2830007579102722, "learning_rate": 1.9489124292276066e-05, "loss": 0.4708, "step": 3625 }, { "epoch": 0.12918396066765234, "grad_norm": 2.1704897982858933, "learning_rate": 1.9488760116444966e-05, "loss": 0.6699, "step": 3626 }, { "epoch": 0.12921958779414647, "grad_norm": 1.6853400995441696, "learning_rate": 1.9488395814264394e-05, "loss": 0.6202, "step": 3627 }, { "epoch": 0.12925521492064057, "grad_norm": 2.4885634138193584, "learning_rate": 1.9488031385739203e-05, "loss": 0.6266, "step": 3628 }, { "epoch": 0.1292908420471347, "grad_norm": 1.8951609316295153, "learning_rate": 1.948766683087424e-05, "loss": 0.6661, "step": 3629 }, { "epoch": 0.1293264691736288, "grad_norm": 1.8584783867317252, "learning_rate": 1.9487302149674362e-05, "loss": 0.6411, "step": 3630 }, { "epoch": 0.1293620963001229, "grad_norm": 2.964043586568282, "learning_rate": 1.9486937342144423e-05, "loss": 0.6261, "step": 3631 }, { "epoch": 0.12939772342661704, "grad_norm": 1.9722320463134606, "learning_rate": 1.9486572408289285e-05, "loss": 0.524, "step": 3632 }, { "epoch": 0.12943335055311114, "grad_norm": 1.7951877301490178, "learning_rate": 1.9486207348113803e-05, "loss": 0.6665, "step": 3633 }, { "epoch": 0.12946897767960525, "grad_norm": 1.3893434299048781, "learning_rate": 1.9485842161622835e-05, "loss": 0.5142, "step": 3634 }, { "epoch": 0.12950460480609938, "grad_norm": 2.231563146709597, "learning_rate": 1.9485476848821257e-05, "loss": 0.657, "step": 3635 }, { "epoch": 0.12954023193259348, "grad_norm": 1.2265947312777374, "learning_rate": 1.948511140971392e-05, "loss": 0.4563, "step": 3636 }, { "epoch": 0.12957585905908758, "grad_norm": 1.3075518484256228, "learning_rate": 1.9484745844305694e-05, "loss": 0.537, "step": 3637 }, { "epoch": 0.12961148618558171, "grad_norm": 1.191422775473983, "learning_rate": 1.948438015260145e-05, "loss": 0.483, "step": 3638 }, { "epoch": 0.12964711331207582, "grad_norm": 1.7306272278929613, "learning_rate": 1.9484014334606058e-05, "loss": 0.6357, "step": 3639 }, { "epoch": 0.12968274043856992, "grad_norm": 1.2112510336373277, "learning_rate": 1.9483648390324382e-05, "loss": 0.5149, "step": 3640 }, { "epoch": 0.12971836756506405, "grad_norm": 1.2209287582346673, "learning_rate": 1.9483282319761298e-05, "loss": 0.4934, "step": 3641 }, { "epoch": 0.12975399469155816, "grad_norm": 1.8121352968002649, "learning_rate": 1.9482916122921685e-05, "loss": 0.6438, "step": 3642 }, { "epoch": 0.12978962181805226, "grad_norm": 1.9822439929285576, "learning_rate": 1.9482549799810412e-05, "loss": 0.5746, "step": 3643 }, { "epoch": 0.1298252489445464, "grad_norm": 1.2525408676852101, "learning_rate": 1.948218335043236e-05, "loss": 0.518, "step": 3644 }, { "epoch": 0.1298608760710405, "grad_norm": 1.4546699067163318, "learning_rate": 1.9481816774792415e-05, "loss": 0.5021, "step": 3645 }, { "epoch": 0.1298965031975346, "grad_norm": 2.259025572387386, "learning_rate": 1.948145007289545e-05, "loss": 0.637, "step": 3646 }, { "epoch": 0.12993213032402873, "grad_norm": 1.480036240260648, "learning_rate": 1.9481083244746347e-05, "loss": 0.5397, "step": 3647 }, { "epoch": 0.12996775745052283, "grad_norm": 1.3737015322999584, "learning_rate": 1.9480716290349998e-05, "loss": 0.5372, "step": 3648 }, { "epoch": 0.13000338457701693, "grad_norm": 1.360997642929271, "learning_rate": 1.948034920971128e-05, "loss": 0.5124, "step": 3649 }, { "epoch": 0.13003901170351106, "grad_norm": 1.3057300482570495, "learning_rate": 1.947998200283509e-05, "loss": 0.545, "step": 3650 }, { "epoch": 0.13007463883000517, "grad_norm": 1.3497274732160611, "learning_rate": 1.9479614669726313e-05, "loss": 0.5479, "step": 3651 }, { "epoch": 0.13011026595649927, "grad_norm": 1.7155964672871031, "learning_rate": 1.9479247210389838e-05, "loss": 0.5959, "step": 3652 }, { "epoch": 0.1301458930829934, "grad_norm": 1.5512669966543733, "learning_rate": 1.9478879624830564e-05, "loss": 0.6339, "step": 3653 }, { "epoch": 0.1301815202094875, "grad_norm": 1.5041676918757585, "learning_rate": 1.947851191305338e-05, "loss": 0.5474, "step": 3654 }, { "epoch": 0.1302171473359816, "grad_norm": 1.7855921896356917, "learning_rate": 1.947814407506319e-05, "loss": 0.6379, "step": 3655 }, { "epoch": 0.13025277446247574, "grad_norm": 1.605679134508822, "learning_rate": 1.947777611086488e-05, "loss": 0.5574, "step": 3656 }, { "epoch": 0.13028840158896984, "grad_norm": 1.2088854851014308, "learning_rate": 1.9477408020463357e-05, "loss": 0.4808, "step": 3657 }, { "epoch": 0.13032402871546395, "grad_norm": 2.1010309425080673, "learning_rate": 1.9477039803863524e-05, "loss": 0.5611, "step": 3658 }, { "epoch": 0.13035965584195808, "grad_norm": 2.424050215035774, "learning_rate": 1.9476671461070286e-05, "loss": 0.6172, "step": 3659 }, { "epoch": 0.13039528296845218, "grad_norm": 1.3813334517138252, "learning_rate": 1.9476302992088536e-05, "loss": 0.545, "step": 3660 }, { "epoch": 0.13043091009494628, "grad_norm": 2.008254557897687, "learning_rate": 1.9475934396923194e-05, "loss": 0.6156, "step": 3661 }, { "epoch": 0.13046653722144042, "grad_norm": 2.7481603959265706, "learning_rate": 1.947556567557916e-05, "loss": 0.6562, "step": 3662 }, { "epoch": 0.13050216434793452, "grad_norm": 2.235370394177174, "learning_rate": 1.9475196828061346e-05, "loss": 0.6127, "step": 3663 }, { "epoch": 0.13053779147442862, "grad_norm": 2.0546856520177657, "learning_rate": 1.9474827854374662e-05, "loss": 0.6426, "step": 3664 }, { "epoch": 0.13057341860092275, "grad_norm": 1.2310185834538359, "learning_rate": 1.9474458754524023e-05, "loss": 0.5037, "step": 3665 }, { "epoch": 0.13060904572741686, "grad_norm": 1.8952650659680526, "learning_rate": 1.9474089528514346e-05, "loss": 0.6613, "step": 3666 }, { "epoch": 0.13064467285391096, "grad_norm": 2.654849730724204, "learning_rate": 1.9473720176350546e-05, "loss": 0.6047, "step": 3667 }, { "epoch": 0.1306802999804051, "grad_norm": 1.460621863929683, "learning_rate": 1.9473350698037537e-05, "loss": 0.5185, "step": 3668 }, { "epoch": 0.1307159271068992, "grad_norm": 2.5349362906889525, "learning_rate": 1.947298109358024e-05, "loss": 0.582, "step": 3669 }, { "epoch": 0.1307515542333933, "grad_norm": 1.5097725747685333, "learning_rate": 1.9472611362983583e-05, "loss": 0.5217, "step": 3670 }, { "epoch": 0.13078718135988743, "grad_norm": 1.9034413724108357, "learning_rate": 1.9472241506252482e-05, "loss": 0.5624, "step": 3671 }, { "epoch": 0.13082280848638153, "grad_norm": 1.1979477751387948, "learning_rate": 1.9471871523391866e-05, "loss": 0.4798, "step": 3672 }, { "epoch": 0.13085843561287563, "grad_norm": 1.3106894088758563, "learning_rate": 1.9471501414406662e-05, "loss": 0.4906, "step": 3673 }, { "epoch": 0.13089406273936977, "grad_norm": 1.9884662828120143, "learning_rate": 1.9471131179301793e-05, "loss": 0.6518, "step": 3674 }, { "epoch": 0.13092968986586387, "grad_norm": 1.2536179378106207, "learning_rate": 1.9470760818082192e-05, "loss": 0.5319, "step": 3675 }, { "epoch": 0.13096531699235797, "grad_norm": 1.8554752905334155, "learning_rate": 1.947039033075279e-05, "loss": 0.6282, "step": 3676 }, { "epoch": 0.1310009441188521, "grad_norm": 1.4332689061774602, "learning_rate": 1.9470019717318525e-05, "loss": 0.5919, "step": 3677 }, { "epoch": 0.1310365712453462, "grad_norm": 1.561355015123177, "learning_rate": 1.946964897778433e-05, "loss": 0.6476, "step": 3678 }, { "epoch": 0.1310721983718403, "grad_norm": 1.3280167885301835, "learning_rate": 1.9469278112155138e-05, "loss": 0.5235, "step": 3679 }, { "epoch": 0.13110782549833444, "grad_norm": 2.2957635953071756, "learning_rate": 1.946890712043589e-05, "loss": 0.6908, "step": 3680 }, { "epoch": 0.13114345262482854, "grad_norm": 1.27006691521766, "learning_rate": 1.946853600263152e-05, "loss": 0.5364, "step": 3681 }, { "epoch": 0.13117907975132265, "grad_norm": 2.073406799430124, "learning_rate": 1.946816475874698e-05, "loss": 0.7002, "step": 3682 }, { "epoch": 0.13121470687781678, "grad_norm": 1.6606699446565567, "learning_rate": 1.946779338878721e-05, "loss": 0.547, "step": 3683 }, { "epoch": 0.13125033400431088, "grad_norm": 2.202949622987222, "learning_rate": 1.946742189275715e-05, "loss": 0.5698, "step": 3684 }, { "epoch": 0.13128596113080498, "grad_norm": 1.326173845710171, "learning_rate": 1.946705027066175e-05, "loss": 0.5318, "step": 3685 }, { "epoch": 0.13132158825729912, "grad_norm": 1.9105469250605034, "learning_rate": 1.9466678522505963e-05, "loss": 0.5068, "step": 3686 }, { "epoch": 0.13135721538379322, "grad_norm": 1.2367893833658705, "learning_rate": 1.946630664829473e-05, "loss": 0.5263, "step": 3687 }, { "epoch": 0.13139284251028732, "grad_norm": 2.367160556223962, "learning_rate": 1.9465934648033008e-05, "loss": 0.6127, "step": 3688 }, { "epoch": 0.13142846963678145, "grad_norm": 3.100840974921938, "learning_rate": 1.946556252172575e-05, "loss": 0.6246, "step": 3689 }, { "epoch": 0.13146409676327556, "grad_norm": 1.8919425167358204, "learning_rate": 1.946519026937791e-05, "loss": 0.5713, "step": 3690 }, { "epoch": 0.13149972388976966, "grad_norm": 1.300424277209778, "learning_rate": 1.946481789099445e-05, "loss": 0.4955, "step": 3691 }, { "epoch": 0.1315353510162638, "grad_norm": 1.668436678673277, "learning_rate": 1.946444538658032e-05, "loss": 0.6355, "step": 3692 }, { "epoch": 0.1315709781427579, "grad_norm": 2.0246415422903365, "learning_rate": 1.9464072756140487e-05, "loss": 0.6353, "step": 3693 }, { "epoch": 0.131606605269252, "grad_norm": 1.8386696513274035, "learning_rate": 1.9463699999679908e-05, "loss": 0.5983, "step": 3694 }, { "epoch": 0.13164223239574613, "grad_norm": 1.278783532276479, "learning_rate": 1.9463327117203554e-05, "loss": 0.589, "step": 3695 }, { "epoch": 0.13167785952224023, "grad_norm": 1.328770741667428, "learning_rate": 1.9462954108716378e-05, "loss": 0.5168, "step": 3696 }, { "epoch": 0.13171348664873433, "grad_norm": 1.2185472023047859, "learning_rate": 1.9462580974223355e-05, "loss": 0.5189, "step": 3697 }, { "epoch": 0.13174911377522847, "grad_norm": 1.3251779083635993, "learning_rate": 1.9462207713729454e-05, "loss": 0.4632, "step": 3698 }, { "epoch": 0.13178474090172257, "grad_norm": 2.438339636861205, "learning_rate": 1.9461834327239646e-05, "loss": 0.6118, "step": 3699 }, { "epoch": 0.13182036802821667, "grad_norm": 1.6072812260624825, "learning_rate": 1.94614608147589e-05, "loss": 0.6687, "step": 3700 }, { "epoch": 0.1318559951547108, "grad_norm": 2.1062505555296664, "learning_rate": 1.9461087176292185e-05, "loss": 0.5621, "step": 3701 }, { "epoch": 0.1318916222812049, "grad_norm": 1.21551526082137, "learning_rate": 1.946071341184449e-05, "loss": 0.4975, "step": 3702 }, { "epoch": 0.131927249407699, "grad_norm": 1.318262004275422, "learning_rate": 1.946033952142077e-05, "loss": 0.5322, "step": 3703 }, { "epoch": 0.13196287653419314, "grad_norm": 3.643020894163861, "learning_rate": 1.9459965505026028e-05, "loss": 0.5939, "step": 3704 }, { "epoch": 0.13199850366068724, "grad_norm": 1.2242820213734686, "learning_rate": 1.945959136266523e-05, "loss": 0.5035, "step": 3705 }, { "epoch": 0.13203413078718135, "grad_norm": 1.5640162076065645, "learning_rate": 1.945921709434336e-05, "loss": 0.6087, "step": 3706 }, { "epoch": 0.13206975791367548, "grad_norm": 1.6541340186613893, "learning_rate": 1.9458842700065408e-05, "loss": 0.5864, "step": 3707 }, { "epoch": 0.13210538504016958, "grad_norm": 1.236773774294383, "learning_rate": 1.9458468179836346e-05, "loss": 0.5606, "step": 3708 }, { "epoch": 0.13214101216666368, "grad_norm": 1.25704511514338, "learning_rate": 1.9458093533661176e-05, "loss": 0.4977, "step": 3709 }, { "epoch": 0.13217663929315782, "grad_norm": 1.3578414244055774, "learning_rate": 1.9457718761544877e-05, "loss": 0.5394, "step": 3710 }, { "epoch": 0.13221226641965192, "grad_norm": 1.1866556722826782, "learning_rate": 1.9457343863492443e-05, "loss": 0.5531, "step": 3711 }, { "epoch": 0.13224789354614602, "grad_norm": 1.2563870321764246, "learning_rate": 1.945696883950886e-05, "loss": 0.5289, "step": 3712 }, { "epoch": 0.13228352067264015, "grad_norm": 2.1486218330796656, "learning_rate": 1.9456593689599135e-05, "loss": 0.602, "step": 3713 }, { "epoch": 0.13231914779913426, "grad_norm": 1.2235320364700688, "learning_rate": 1.945621841376825e-05, "loss": 0.4858, "step": 3714 }, { "epoch": 0.13235477492562836, "grad_norm": 1.9421086950921285, "learning_rate": 1.945584301202121e-05, "loss": 0.5797, "step": 3715 }, { "epoch": 0.1323904020521225, "grad_norm": 1.2114528187749303, "learning_rate": 1.9455467484363008e-05, "loss": 0.4853, "step": 3716 }, { "epoch": 0.1324260291786166, "grad_norm": 1.8687748501319343, "learning_rate": 1.945509183079865e-05, "loss": 0.646, "step": 3717 }, { "epoch": 0.1324616563051107, "grad_norm": 1.3365574149464259, "learning_rate": 1.9454716051333137e-05, "loss": 0.5195, "step": 3718 }, { "epoch": 0.13249728343160483, "grad_norm": 1.781615559487923, "learning_rate": 1.945434014597147e-05, "loss": 0.5464, "step": 3719 }, { "epoch": 0.13253291055809893, "grad_norm": 1.2650136818091684, "learning_rate": 1.9453964114718655e-05, "loss": 0.5056, "step": 3720 }, { "epoch": 0.13256853768459306, "grad_norm": 1.2968350322879578, "learning_rate": 1.94535879575797e-05, "loss": 0.6087, "step": 3721 }, { "epoch": 0.13260416481108717, "grad_norm": 1.2795632046756114, "learning_rate": 1.9453211674559613e-05, "loss": 0.4994, "step": 3722 }, { "epoch": 0.13263979193758127, "grad_norm": 1.2946159027666566, "learning_rate": 1.9452835265663404e-05, "loss": 0.5388, "step": 3723 }, { "epoch": 0.1326754190640754, "grad_norm": 1.2074160348946916, "learning_rate": 1.945245873089609e-05, "loss": 0.4674, "step": 3724 }, { "epoch": 0.1327110461905695, "grad_norm": 1.991786140737419, "learning_rate": 1.9452082070262678e-05, "loss": 0.6085, "step": 3725 }, { "epoch": 0.1327466733170636, "grad_norm": 1.3683970272962718, "learning_rate": 1.9451705283768186e-05, "loss": 0.5479, "step": 3726 }, { "epoch": 0.13278230044355774, "grad_norm": 2.2173907406449604, "learning_rate": 1.9451328371417636e-05, "loss": 0.595, "step": 3727 }, { "epoch": 0.13281792757005184, "grad_norm": 1.2869627609687837, "learning_rate": 1.945095133321604e-05, "loss": 0.5003, "step": 3728 }, { "epoch": 0.13285355469654594, "grad_norm": 1.8411366128677333, "learning_rate": 1.945057416916842e-05, "loss": 0.5805, "step": 3729 }, { "epoch": 0.13288918182304008, "grad_norm": 1.7533175854343943, "learning_rate": 1.94501968792798e-05, "loss": 0.5148, "step": 3730 }, { "epoch": 0.13292480894953418, "grad_norm": 1.351488031469804, "learning_rate": 1.9449819463555206e-05, "loss": 0.5326, "step": 3731 }, { "epoch": 0.13296043607602828, "grad_norm": 1.377206737871749, "learning_rate": 1.944944192199966e-05, "loss": 0.5806, "step": 3732 }, { "epoch": 0.1329960632025224, "grad_norm": 1.5635848729506574, "learning_rate": 1.944906425461819e-05, "loss": 0.6815, "step": 3733 }, { "epoch": 0.13303169032901652, "grad_norm": 1.6762490633348115, "learning_rate": 1.9448686461415822e-05, "loss": 0.6265, "step": 3734 }, { "epoch": 0.13306731745551062, "grad_norm": 1.336579773736142, "learning_rate": 1.9448308542397592e-05, "loss": 0.5031, "step": 3735 }, { "epoch": 0.13310294458200475, "grad_norm": 1.1714731866403307, "learning_rate": 1.9447930497568528e-05, "loss": 0.5025, "step": 3736 }, { "epoch": 0.13313857170849885, "grad_norm": 2.4078745552733607, "learning_rate": 1.9447552326933666e-05, "loss": 0.6935, "step": 3737 }, { "epoch": 0.13317419883499296, "grad_norm": 1.1851221562053844, "learning_rate": 1.9447174030498043e-05, "loss": 0.5094, "step": 3738 }, { "epoch": 0.1332098259614871, "grad_norm": 1.8694549796759201, "learning_rate": 1.9446795608266693e-05, "loss": 0.6009, "step": 3739 }, { "epoch": 0.1332454530879812, "grad_norm": 1.2771750662429935, "learning_rate": 1.9446417060244656e-05, "loss": 0.5194, "step": 3740 }, { "epoch": 0.1332810802144753, "grad_norm": 1.8993653469852305, "learning_rate": 1.9446038386436977e-05, "loss": 0.5904, "step": 3741 }, { "epoch": 0.13331670734096943, "grad_norm": 1.202955662821598, "learning_rate": 1.9445659586848692e-05, "loss": 0.5317, "step": 3742 }, { "epoch": 0.13335233446746353, "grad_norm": 1.496932010172316, "learning_rate": 1.9445280661484848e-05, "loss": 0.5969, "step": 3743 }, { "epoch": 0.13338796159395763, "grad_norm": 3.3569846444089024, "learning_rate": 1.944490161035049e-05, "loss": 0.6681, "step": 3744 }, { "epoch": 0.13342358872045176, "grad_norm": 1.3286386710575169, "learning_rate": 1.9444522433450662e-05, "loss": 0.5498, "step": 3745 }, { "epoch": 0.13345921584694587, "grad_norm": 1.8581292038935735, "learning_rate": 1.9444143130790422e-05, "loss": 0.6684, "step": 3746 }, { "epoch": 0.13349484297343997, "grad_norm": 1.2455874502106015, "learning_rate": 1.944376370237481e-05, "loss": 0.5139, "step": 3747 }, { "epoch": 0.1335304700999341, "grad_norm": 1.5393291915967462, "learning_rate": 1.944338414820889e-05, "loss": 0.6813, "step": 3748 }, { "epoch": 0.1335660972264282, "grad_norm": 1.7850588285732307, "learning_rate": 1.9443004468297704e-05, "loss": 0.6041, "step": 3749 }, { "epoch": 0.1336017243529223, "grad_norm": 1.672980171252276, "learning_rate": 1.9442624662646317e-05, "loss": 0.6484, "step": 3750 }, { "epoch": 0.13363735147941644, "grad_norm": 1.7676470852411847, "learning_rate": 1.9442244731259778e-05, "loss": 0.5441, "step": 3751 }, { "epoch": 0.13367297860591054, "grad_norm": 1.2031630214978424, "learning_rate": 1.9441864674143158e-05, "loss": 0.5093, "step": 3752 }, { "epoch": 0.13370860573240465, "grad_norm": 1.9551970906647689, "learning_rate": 1.9441484491301505e-05, "loss": 0.6269, "step": 3753 }, { "epoch": 0.13374423285889878, "grad_norm": 1.3601077144507063, "learning_rate": 1.9441104182739884e-05, "loss": 0.5521, "step": 3754 }, { "epoch": 0.13377985998539288, "grad_norm": 1.0854808035871062, "learning_rate": 1.9440723748463365e-05, "loss": 0.4608, "step": 3755 }, { "epoch": 0.13381548711188698, "grad_norm": 1.6109718385168645, "learning_rate": 1.944034318847701e-05, "loss": 0.6927, "step": 3756 }, { "epoch": 0.1338511142383811, "grad_norm": 1.7239222525257387, "learning_rate": 1.943996250278589e-05, "loss": 0.6395, "step": 3757 }, { "epoch": 0.13388674136487522, "grad_norm": 1.6691064616774414, "learning_rate": 1.943958169139507e-05, "loss": 0.5913, "step": 3758 }, { "epoch": 0.13392236849136932, "grad_norm": 1.3319660208640283, "learning_rate": 1.943920075430962e-05, "loss": 0.5363, "step": 3759 }, { "epoch": 0.13395799561786345, "grad_norm": 1.242601715531617, "learning_rate": 1.9438819691534615e-05, "loss": 0.4589, "step": 3760 }, { "epoch": 0.13399362274435755, "grad_norm": 1.680353025571787, "learning_rate": 1.943843850307513e-05, "loss": 0.6285, "step": 3761 }, { "epoch": 0.13402924987085166, "grad_norm": 1.3486204602052738, "learning_rate": 1.9438057188936237e-05, "loss": 0.5301, "step": 3762 }, { "epoch": 0.1340648769973458, "grad_norm": 1.4543041435560882, "learning_rate": 1.9437675749123017e-05, "loss": 0.5111, "step": 3763 }, { "epoch": 0.1341005041238399, "grad_norm": 1.7123153123759862, "learning_rate": 1.943729418364055e-05, "loss": 0.6685, "step": 3764 }, { "epoch": 0.134136131250334, "grad_norm": 1.9910916467553448, "learning_rate": 1.943691249249391e-05, "loss": 0.6377, "step": 3765 }, { "epoch": 0.13417175837682813, "grad_norm": 1.2921468948173684, "learning_rate": 1.943653067568819e-05, "loss": 0.5455, "step": 3766 }, { "epoch": 0.13420738550332223, "grad_norm": 1.2890620609744934, "learning_rate": 1.9436148733228463e-05, "loss": 0.5316, "step": 3767 }, { "epoch": 0.13424301262981633, "grad_norm": 1.6693276985607413, "learning_rate": 1.9435766665119823e-05, "loss": 0.5642, "step": 3768 }, { "epoch": 0.13427863975631046, "grad_norm": 1.6494366151056954, "learning_rate": 1.943538447136735e-05, "loss": 0.5598, "step": 3769 }, { "epoch": 0.13431426688280457, "grad_norm": 1.899167085782676, "learning_rate": 1.9435002151976144e-05, "loss": 0.6603, "step": 3770 }, { "epoch": 0.13434989400929867, "grad_norm": 2.0980540884408687, "learning_rate": 1.9434619706951286e-05, "loss": 0.6119, "step": 3771 }, { "epoch": 0.1343855211357928, "grad_norm": 1.759426926805035, "learning_rate": 1.9434237136297876e-05, "loss": 0.6641, "step": 3772 }, { "epoch": 0.1344211482622869, "grad_norm": 1.593606855361832, "learning_rate": 1.9433854440021e-05, "loss": 0.6395, "step": 3773 }, { "epoch": 0.134456775388781, "grad_norm": 2.4873169648012508, "learning_rate": 1.943347161812576e-05, "loss": 0.6234, "step": 3774 }, { "epoch": 0.13449240251527514, "grad_norm": 2.415141969372676, "learning_rate": 1.9433088670617254e-05, "loss": 0.6782, "step": 3775 }, { "epoch": 0.13452802964176924, "grad_norm": 2.190947931657875, "learning_rate": 1.9432705597500573e-05, "loss": 0.5836, "step": 3776 }, { "epoch": 0.13456365676826335, "grad_norm": 1.426087349976734, "learning_rate": 1.943232239878083e-05, "loss": 0.5351, "step": 3777 }, { "epoch": 0.13459928389475748, "grad_norm": 1.8093179891847162, "learning_rate": 1.943193907446312e-05, "loss": 0.6573, "step": 3778 }, { "epoch": 0.13463491102125158, "grad_norm": 1.3629150430064738, "learning_rate": 1.9431555624552546e-05, "loss": 0.5178, "step": 3779 }, { "epoch": 0.13467053814774568, "grad_norm": 1.2291855328253445, "learning_rate": 1.943117204905422e-05, "loss": 0.5109, "step": 3780 }, { "epoch": 0.13470616527423981, "grad_norm": 1.3035251806894617, "learning_rate": 1.9430788347973245e-05, "loss": 0.5072, "step": 3781 }, { "epoch": 0.13474179240073392, "grad_norm": 1.135205638804586, "learning_rate": 1.943040452131473e-05, "loss": 0.4725, "step": 3782 }, { "epoch": 0.13477741952722802, "grad_norm": 1.3270711691678048, "learning_rate": 1.943002056908379e-05, "loss": 0.56, "step": 3783 }, { "epoch": 0.13481304665372215, "grad_norm": 1.3158900349143556, "learning_rate": 1.942963649128553e-05, "loss": 0.5344, "step": 3784 }, { "epoch": 0.13484867378021625, "grad_norm": 1.8688117738687762, "learning_rate": 1.942925228792507e-05, "loss": 0.5921, "step": 3785 }, { "epoch": 0.13488430090671036, "grad_norm": 1.3074228332165236, "learning_rate": 1.942886795900753e-05, "loss": 0.4994, "step": 3786 }, { "epoch": 0.1349199280332045, "grad_norm": 1.850030115392293, "learning_rate": 1.9428483504538023e-05, "loss": 0.7293, "step": 3787 }, { "epoch": 0.1349555551596986, "grad_norm": 1.2818484141093553, "learning_rate": 1.9428098924521665e-05, "loss": 0.5411, "step": 3788 }, { "epoch": 0.1349911822861927, "grad_norm": 1.2724688890419038, "learning_rate": 1.942771421896358e-05, "loss": 0.5451, "step": 3789 }, { "epoch": 0.13502680941268683, "grad_norm": 2.3318764374273027, "learning_rate": 1.942732938786889e-05, "loss": 0.6119, "step": 3790 }, { "epoch": 0.13506243653918093, "grad_norm": 1.7671965923527078, "learning_rate": 1.942694443124272e-05, "loss": 0.5754, "step": 3791 }, { "epoch": 0.13509806366567503, "grad_norm": 1.640313415921868, "learning_rate": 1.9426559349090198e-05, "loss": 0.6289, "step": 3792 }, { "epoch": 0.13513369079216916, "grad_norm": 2.3393626156071727, "learning_rate": 1.942617414141645e-05, "loss": 0.5743, "step": 3793 }, { "epoch": 0.13516931791866327, "grad_norm": 1.4356150957423222, "learning_rate": 1.9425788808226604e-05, "loss": 0.4919, "step": 3794 }, { "epoch": 0.13520494504515737, "grad_norm": 1.8525346496591892, "learning_rate": 1.9425403349525788e-05, "loss": 0.6276, "step": 3795 }, { "epoch": 0.1352405721716515, "grad_norm": 1.239764837417676, "learning_rate": 1.942501776531914e-05, "loss": 0.5088, "step": 3796 }, { "epoch": 0.1352761992981456, "grad_norm": 1.231820130816862, "learning_rate": 1.9424632055611797e-05, "loss": 0.5272, "step": 3797 }, { "epoch": 0.1353118264246397, "grad_norm": 1.2849429260436176, "learning_rate": 1.9424246220408888e-05, "loss": 0.5228, "step": 3798 }, { "epoch": 0.13534745355113384, "grad_norm": 1.3747609760927462, "learning_rate": 1.942386025971555e-05, "loss": 0.5063, "step": 3799 }, { "epoch": 0.13538308067762794, "grad_norm": 2.198216183619624, "learning_rate": 1.9423474173536932e-05, "loss": 0.6215, "step": 3800 }, { "epoch": 0.13541870780412205, "grad_norm": 1.2990539221274173, "learning_rate": 1.9423087961878165e-05, "loss": 0.5142, "step": 3801 }, { "epoch": 0.13545433493061618, "grad_norm": 1.7181758474272886, "learning_rate": 1.9422701624744396e-05, "loss": 0.6553, "step": 3802 }, { "epoch": 0.13548996205711028, "grad_norm": 1.3478428811061138, "learning_rate": 1.942231516214077e-05, "loss": 0.5468, "step": 3803 }, { "epoch": 0.13552558918360438, "grad_norm": 1.2121719970966958, "learning_rate": 1.942192857407243e-05, "loss": 0.5105, "step": 3804 }, { "epoch": 0.13556121631009851, "grad_norm": 1.7313994941690762, "learning_rate": 1.9421541860544525e-05, "loss": 0.6338, "step": 3805 }, { "epoch": 0.13559684343659262, "grad_norm": 2.0511688498018477, "learning_rate": 1.9421155021562206e-05, "loss": 0.5761, "step": 3806 }, { "epoch": 0.13563247056308672, "grad_norm": 1.267600735475254, "learning_rate": 1.9420768057130622e-05, "loss": 0.551, "step": 3807 }, { "epoch": 0.13566809768958085, "grad_norm": 2.1997031369596027, "learning_rate": 1.9420380967254932e-05, "loss": 0.5408, "step": 3808 }, { "epoch": 0.13570372481607496, "grad_norm": 1.2734392257546214, "learning_rate": 1.941999375194028e-05, "loss": 0.5205, "step": 3809 }, { "epoch": 0.13573935194256906, "grad_norm": 1.1738056237372363, "learning_rate": 1.941960641119183e-05, "loss": 0.4834, "step": 3810 }, { "epoch": 0.1357749790690632, "grad_norm": 1.6696355085272983, "learning_rate": 1.941921894501473e-05, "loss": 0.5964, "step": 3811 }, { "epoch": 0.1358106061955573, "grad_norm": 1.8467020281663569, "learning_rate": 1.941883135341415e-05, "loss": 0.5853, "step": 3812 }, { "epoch": 0.1358462333220514, "grad_norm": 1.7540059815688842, "learning_rate": 1.941844363639525e-05, "loss": 0.5979, "step": 3813 }, { "epoch": 0.13588186044854553, "grad_norm": 2.126094931665548, "learning_rate": 1.9418055793963185e-05, "loss": 0.6341, "step": 3814 }, { "epoch": 0.13591748757503963, "grad_norm": 1.6139589720729701, "learning_rate": 1.9417667826123128e-05, "loss": 0.5596, "step": 3815 }, { "epoch": 0.13595311470153376, "grad_norm": 1.5159963149723346, "learning_rate": 1.941727973288024e-05, "loss": 0.5148, "step": 3816 }, { "epoch": 0.13598874182802786, "grad_norm": 1.2727814999882519, "learning_rate": 1.9416891514239688e-05, "loss": 0.5433, "step": 3817 }, { "epoch": 0.13602436895452197, "grad_norm": 2.044738424443477, "learning_rate": 1.9416503170206647e-05, "loss": 0.5589, "step": 3818 }, { "epoch": 0.1360599960810161, "grad_norm": 1.2327429171152773, "learning_rate": 1.9416114700786285e-05, "loss": 0.5223, "step": 3819 }, { "epoch": 0.1360956232075102, "grad_norm": 1.8884593515198722, "learning_rate": 1.9415726105983773e-05, "loss": 0.6335, "step": 3820 }, { "epoch": 0.1361312503340043, "grad_norm": 1.2203228951695158, "learning_rate": 1.9415337385804285e-05, "loss": 0.5485, "step": 3821 }, { "epoch": 0.13616687746049844, "grad_norm": 1.43325479203116, "learning_rate": 1.9414948540253e-05, "loss": 0.5494, "step": 3822 }, { "epoch": 0.13620250458699254, "grad_norm": 1.3054197690153229, "learning_rate": 1.9414559569335093e-05, "loss": 0.5381, "step": 3823 }, { "epoch": 0.13623813171348664, "grad_norm": 2.0065933837637115, "learning_rate": 1.9414170473055747e-05, "loss": 0.6789, "step": 3824 }, { "epoch": 0.13627375883998077, "grad_norm": 1.4375047211226064, "learning_rate": 1.941378125142014e-05, "loss": 0.5872, "step": 3825 }, { "epoch": 0.13630938596647488, "grad_norm": 1.919816348409809, "learning_rate": 1.9413391904433458e-05, "loss": 0.6227, "step": 3826 }, { "epoch": 0.13634501309296898, "grad_norm": 2.245314418329718, "learning_rate": 1.9413002432100882e-05, "loss": 0.6184, "step": 3827 }, { "epoch": 0.1363806402194631, "grad_norm": 1.1481529671090318, "learning_rate": 1.9412612834427597e-05, "loss": 0.5566, "step": 3828 }, { "epoch": 0.13641626734595722, "grad_norm": 1.2973884351436356, "learning_rate": 1.9412223111418795e-05, "loss": 0.5573, "step": 3829 }, { "epoch": 0.13645189447245132, "grad_norm": 1.899659952907059, "learning_rate": 1.9411833263079665e-05, "loss": 0.6553, "step": 3830 }, { "epoch": 0.13648752159894545, "grad_norm": 1.7118071206383105, "learning_rate": 1.9411443289415396e-05, "loss": 0.7591, "step": 3831 }, { "epoch": 0.13652314872543955, "grad_norm": 1.844466652560302, "learning_rate": 1.9411053190431176e-05, "loss": 0.6016, "step": 3832 }, { "epoch": 0.13655877585193366, "grad_norm": 2.043338887241329, "learning_rate": 1.9410662966132213e-05, "loss": 0.7673, "step": 3833 }, { "epoch": 0.1365944029784278, "grad_norm": 1.2720652864746325, "learning_rate": 1.9410272616523688e-05, "loss": 0.5462, "step": 3834 }, { "epoch": 0.1366300301049219, "grad_norm": 2.054342548465614, "learning_rate": 1.9409882141610813e-05, "loss": 0.5804, "step": 3835 }, { "epoch": 0.136665657231416, "grad_norm": 1.9343452327134072, "learning_rate": 1.9409491541398774e-05, "loss": 0.5385, "step": 3836 }, { "epoch": 0.13670128435791012, "grad_norm": 1.8184658915473733, "learning_rate": 1.940910081589278e-05, "loss": 0.5858, "step": 3837 }, { "epoch": 0.13673691148440423, "grad_norm": 1.828733226564461, "learning_rate": 1.9408709965098035e-05, "loss": 0.6103, "step": 3838 }, { "epoch": 0.13677253861089833, "grad_norm": 1.265397039477372, "learning_rate": 1.9408318989019736e-05, "loss": 0.5516, "step": 3839 }, { "epoch": 0.13680816573739246, "grad_norm": 1.7621524170866716, "learning_rate": 1.9407927887663097e-05, "loss": 0.5709, "step": 3840 }, { "epoch": 0.13684379286388657, "grad_norm": 2.004348002678604, "learning_rate": 1.9407536661033323e-05, "loss": 0.5598, "step": 3841 }, { "epoch": 0.13687941999038067, "grad_norm": 1.2771691404408343, "learning_rate": 1.9407145309135617e-05, "loss": 0.4588, "step": 3842 }, { "epoch": 0.1369150471168748, "grad_norm": 1.2600187214529759, "learning_rate": 1.9406753831975202e-05, "loss": 0.5388, "step": 3843 }, { "epoch": 0.1369506742433689, "grad_norm": 1.2296084201508217, "learning_rate": 1.9406362229557283e-05, "loss": 0.5059, "step": 3844 }, { "epoch": 0.136986301369863, "grad_norm": 3.9907669241251473, "learning_rate": 1.9405970501887074e-05, "loss": 0.6898, "step": 3845 }, { "epoch": 0.13702192849635714, "grad_norm": 1.2129304440839206, "learning_rate": 1.9405578648969796e-05, "loss": 0.5095, "step": 3846 }, { "epoch": 0.13705755562285124, "grad_norm": 2.2253821789963015, "learning_rate": 1.9405186670810667e-05, "loss": 0.6143, "step": 3847 }, { "epoch": 0.13709318274934534, "grad_norm": 1.9308308258956939, "learning_rate": 1.9404794567414897e-05, "loss": 0.5973, "step": 3848 }, { "epoch": 0.13712880987583947, "grad_norm": 1.3777192473587758, "learning_rate": 1.940440233878772e-05, "loss": 0.5109, "step": 3849 }, { "epoch": 0.13716443700233358, "grad_norm": 1.3411426006490839, "learning_rate": 1.9404009984934347e-05, "loss": 0.5634, "step": 3850 }, { "epoch": 0.13720006412882768, "grad_norm": 1.4227525553867906, "learning_rate": 1.940361750586001e-05, "loss": 0.6231, "step": 3851 }, { "epoch": 0.1372356912553218, "grad_norm": 2.070327169518408, "learning_rate": 1.9403224901569936e-05, "loss": 0.6116, "step": 3852 }, { "epoch": 0.13727131838181592, "grad_norm": 1.3720064490912152, "learning_rate": 1.9402832172069347e-05, "loss": 0.5172, "step": 3853 }, { "epoch": 0.13730694550831002, "grad_norm": 1.1966435797928425, "learning_rate": 1.9402439317363475e-05, "loss": 0.4931, "step": 3854 }, { "epoch": 0.13734257263480415, "grad_norm": 1.8433316422071413, "learning_rate": 1.940204633745755e-05, "loss": 0.6088, "step": 3855 }, { "epoch": 0.13737819976129825, "grad_norm": 1.3586359244426554, "learning_rate": 1.9401653232356813e-05, "loss": 0.5391, "step": 3856 }, { "epoch": 0.13741382688779236, "grad_norm": 1.7384537926542492, "learning_rate": 1.940126000206649e-05, "loss": 0.5563, "step": 3857 }, { "epoch": 0.1374494540142865, "grad_norm": 4.034550344646864, "learning_rate": 1.9400866646591816e-05, "loss": 0.5985, "step": 3858 }, { "epoch": 0.1374850811407806, "grad_norm": 2.47729185117951, "learning_rate": 1.940047316593803e-05, "loss": 0.6446, "step": 3859 }, { "epoch": 0.1375207082672747, "grad_norm": 1.2981740195674387, "learning_rate": 1.9400079560110376e-05, "loss": 0.6044, "step": 3860 }, { "epoch": 0.13755633539376882, "grad_norm": 1.9202254425841019, "learning_rate": 1.9399685829114093e-05, "loss": 0.6528, "step": 3861 }, { "epoch": 0.13759196252026293, "grad_norm": 1.2166809976137918, "learning_rate": 1.9399291972954425e-05, "loss": 0.5103, "step": 3862 }, { "epoch": 0.13762758964675703, "grad_norm": 1.9781396015039618, "learning_rate": 1.9398897991636613e-05, "loss": 0.6372, "step": 3863 }, { "epoch": 0.13766321677325116, "grad_norm": 1.3600809855209448, "learning_rate": 1.9398503885165902e-05, "loss": 0.5253, "step": 3864 }, { "epoch": 0.13769884389974527, "grad_norm": 1.5926812438905418, "learning_rate": 1.9398109653547547e-05, "loss": 0.6119, "step": 3865 }, { "epoch": 0.13773447102623937, "grad_norm": 2.1682527339165834, "learning_rate": 1.939771529678679e-05, "loss": 0.6755, "step": 3866 }, { "epoch": 0.1377700981527335, "grad_norm": 1.58378458108719, "learning_rate": 1.9397320814888883e-05, "loss": 0.6459, "step": 3867 }, { "epoch": 0.1378057252792276, "grad_norm": 1.6990639508504664, "learning_rate": 1.9396926207859085e-05, "loss": 0.6157, "step": 3868 }, { "epoch": 0.1378413524057217, "grad_norm": 1.2531031993934445, "learning_rate": 1.9396531475702646e-05, "loss": 0.5334, "step": 3869 }, { "epoch": 0.13787697953221584, "grad_norm": 1.1456629625363763, "learning_rate": 1.9396136618424824e-05, "loss": 0.5109, "step": 3870 }, { "epoch": 0.13791260665870994, "grad_norm": 1.7359364488238203, "learning_rate": 1.939574163603087e-05, "loss": 0.627, "step": 3871 }, { "epoch": 0.13794823378520404, "grad_norm": 1.58438842551967, "learning_rate": 1.9395346528526055e-05, "loss": 0.6126, "step": 3872 }, { "epoch": 0.13798386091169818, "grad_norm": 1.1788834515143465, "learning_rate": 1.9394951295915628e-05, "loss": 0.5386, "step": 3873 }, { "epoch": 0.13801948803819228, "grad_norm": 1.7148816921844803, "learning_rate": 1.9394555938204866e-05, "loss": 0.7109, "step": 3874 }, { "epoch": 0.13805511516468638, "grad_norm": 2.077788360090912, "learning_rate": 1.939416045539902e-05, "loss": 0.6213, "step": 3875 }, { "epoch": 0.1380907422911805, "grad_norm": 1.1335807455243359, "learning_rate": 1.939376484750336e-05, "loss": 0.5423, "step": 3876 }, { "epoch": 0.13812636941767462, "grad_norm": 1.8888856998207586, "learning_rate": 1.939336911452316e-05, "loss": 0.6141, "step": 3877 }, { "epoch": 0.13816199654416872, "grad_norm": 1.8559178102755662, "learning_rate": 1.9392973256463682e-05, "loss": 0.6107, "step": 3878 }, { "epoch": 0.13819762367066285, "grad_norm": 1.636870844611121, "learning_rate": 1.93925772733302e-05, "loss": 0.6383, "step": 3879 }, { "epoch": 0.13823325079715695, "grad_norm": 1.3904083868651236, "learning_rate": 1.9392181165127984e-05, "loss": 0.5678, "step": 3880 }, { "epoch": 0.13826887792365106, "grad_norm": 1.973398314816964, "learning_rate": 1.9391784931862313e-05, "loss": 0.6833, "step": 3881 }, { "epoch": 0.1383045050501452, "grad_norm": 1.2938147543248395, "learning_rate": 1.9391388573538463e-05, "loss": 0.5586, "step": 3882 }, { "epoch": 0.1383401321766393, "grad_norm": 1.8820340308009056, "learning_rate": 1.9390992090161706e-05, "loss": 0.567, "step": 3883 }, { "epoch": 0.1383757593031334, "grad_norm": 1.398905508765506, "learning_rate": 1.9390595481737327e-05, "loss": 0.5677, "step": 3884 }, { "epoch": 0.13841138642962753, "grad_norm": 1.333688392344727, "learning_rate": 1.9390198748270607e-05, "loss": 0.5016, "step": 3885 }, { "epoch": 0.13844701355612163, "grad_norm": 2.319789528217445, "learning_rate": 1.9389801889766828e-05, "loss": 0.634, "step": 3886 }, { "epoch": 0.13848264068261573, "grad_norm": 1.2586936364578862, "learning_rate": 1.9389404906231272e-05, "loss": 0.5043, "step": 3887 }, { "epoch": 0.13851826780910986, "grad_norm": 1.4676878674512934, "learning_rate": 1.9389007797669225e-05, "loss": 0.5252, "step": 3888 }, { "epoch": 0.13855389493560397, "grad_norm": 3.1501725700635475, "learning_rate": 1.9388610564085982e-05, "loss": 0.5717, "step": 3889 }, { "epoch": 0.13858952206209807, "grad_norm": 1.1992841361633184, "learning_rate": 1.9388213205486823e-05, "loss": 0.5178, "step": 3890 }, { "epoch": 0.1386251491885922, "grad_norm": 1.2037910943900496, "learning_rate": 1.9387815721877045e-05, "loss": 0.4752, "step": 3891 }, { "epoch": 0.1386607763150863, "grad_norm": 2.8167740412449676, "learning_rate": 1.9387418113261938e-05, "loss": 0.7085, "step": 3892 }, { "epoch": 0.1386964034415804, "grad_norm": 1.2930226928049864, "learning_rate": 1.9387020379646798e-05, "loss": 0.6022, "step": 3893 }, { "epoch": 0.13873203056807454, "grad_norm": 1.2425054572076695, "learning_rate": 1.9386622521036917e-05, "loss": 0.536, "step": 3894 }, { "epoch": 0.13876765769456864, "grad_norm": 1.6064919699855937, "learning_rate": 1.93862245374376e-05, "loss": 0.6038, "step": 3895 }, { "epoch": 0.13880328482106274, "grad_norm": 1.7683337059868116, "learning_rate": 1.938582642885414e-05, "loss": 0.6785, "step": 3896 }, { "epoch": 0.13883891194755688, "grad_norm": 1.1630992637747988, "learning_rate": 1.9385428195291848e-05, "loss": 0.4629, "step": 3897 }, { "epoch": 0.13887453907405098, "grad_norm": 2.0512199430687814, "learning_rate": 1.9385029836756013e-05, "loss": 0.7444, "step": 3898 }, { "epoch": 0.13891016620054508, "grad_norm": 1.3643940810759305, "learning_rate": 1.938463135325195e-05, "loss": 0.4997, "step": 3899 }, { "epoch": 0.1389457933270392, "grad_norm": 3.507182949549625, "learning_rate": 1.9384232744784955e-05, "loss": 0.4593, "step": 3900 }, { "epoch": 0.13898142045353332, "grad_norm": 1.29104992412631, "learning_rate": 1.9383834011360347e-05, "loss": 0.458, "step": 3901 }, { "epoch": 0.13901704758002742, "grad_norm": 1.3234884631422914, "learning_rate": 1.9383435152983428e-05, "loss": 0.5431, "step": 3902 }, { "epoch": 0.13905267470652155, "grad_norm": 1.2899045145268475, "learning_rate": 1.9383036169659513e-05, "loss": 0.5645, "step": 3903 }, { "epoch": 0.13908830183301565, "grad_norm": 1.6720478311643034, "learning_rate": 1.938263706139391e-05, "loss": 0.5722, "step": 3904 }, { "epoch": 0.13912392895950976, "grad_norm": 1.255652650485346, "learning_rate": 1.938223782819194e-05, "loss": 0.5007, "step": 3905 }, { "epoch": 0.1391595560860039, "grad_norm": 1.202822965770791, "learning_rate": 1.9381838470058913e-05, "loss": 0.4895, "step": 3906 }, { "epoch": 0.139195183212498, "grad_norm": 2.356433782153087, "learning_rate": 1.938143898700015e-05, "loss": 0.5869, "step": 3907 }, { "epoch": 0.1392308103389921, "grad_norm": 1.1722666640410488, "learning_rate": 1.938103937902097e-05, "loss": 0.5591, "step": 3908 }, { "epoch": 0.13926643746548623, "grad_norm": 1.3030548110570948, "learning_rate": 1.9380639646126692e-05, "loss": 0.5068, "step": 3909 }, { "epoch": 0.13930206459198033, "grad_norm": 1.4472666276244113, "learning_rate": 1.9380239788322642e-05, "loss": 0.5127, "step": 3910 }, { "epoch": 0.13933769171847446, "grad_norm": 1.1789766457511166, "learning_rate": 1.9379839805614144e-05, "loss": 0.4813, "step": 3911 }, { "epoch": 0.13937331884496856, "grad_norm": 2.0272605689678453, "learning_rate": 1.9379439698006522e-05, "loss": 0.6256, "step": 3912 }, { "epoch": 0.13940894597146267, "grad_norm": 1.3686977981579915, "learning_rate": 1.9379039465505105e-05, "loss": 0.5355, "step": 3913 }, { "epoch": 0.1394445730979568, "grad_norm": 1.2173758737507794, "learning_rate": 1.937863910811522e-05, "loss": 0.5208, "step": 3914 }, { "epoch": 0.1394802002244509, "grad_norm": 1.2053944428141241, "learning_rate": 1.93782386258422e-05, "loss": 0.5526, "step": 3915 }, { "epoch": 0.139515827350945, "grad_norm": 1.270240940430705, "learning_rate": 1.937783801869138e-05, "loss": 0.5674, "step": 3916 }, { "epoch": 0.13955145447743914, "grad_norm": 3.537730668001064, "learning_rate": 1.937743728666809e-05, "loss": 0.5468, "step": 3917 }, { "epoch": 0.13958708160393324, "grad_norm": 1.3353102267161998, "learning_rate": 1.9377036429777673e-05, "loss": 0.5174, "step": 3918 }, { "epoch": 0.13962270873042734, "grad_norm": 1.6792364198559635, "learning_rate": 1.9376635448025456e-05, "loss": 0.6378, "step": 3919 }, { "epoch": 0.13965833585692147, "grad_norm": 1.2524015168048015, "learning_rate": 1.9376234341416784e-05, "loss": 0.5024, "step": 3920 }, { "epoch": 0.13969396298341558, "grad_norm": 1.1249202810217935, "learning_rate": 1.9375833109957e-05, "loss": 0.4748, "step": 3921 }, { "epoch": 0.13972959010990968, "grad_norm": 1.2764959542007537, "learning_rate": 1.9375431753651447e-05, "loss": 0.5375, "step": 3922 }, { "epoch": 0.1397652172364038, "grad_norm": 1.8756018299296644, "learning_rate": 1.9375030272505463e-05, "loss": 0.6156, "step": 3923 }, { "epoch": 0.1398008443628979, "grad_norm": 1.3766045838758334, "learning_rate": 1.93746286665244e-05, "loss": 0.5497, "step": 3924 }, { "epoch": 0.13983647148939202, "grad_norm": 1.3866675539156865, "learning_rate": 1.9374226935713603e-05, "loss": 0.4959, "step": 3925 }, { "epoch": 0.13987209861588615, "grad_norm": 1.6513967051482923, "learning_rate": 1.9373825080078426e-05, "loss": 0.5995, "step": 3926 }, { "epoch": 0.13990772574238025, "grad_norm": 1.6279559245736202, "learning_rate": 1.9373423099624213e-05, "loss": 0.616, "step": 3927 }, { "epoch": 0.13994335286887435, "grad_norm": 2.4620645717858496, "learning_rate": 1.937302099435632e-05, "loss": 0.6485, "step": 3928 }, { "epoch": 0.13997897999536849, "grad_norm": 1.8640571828344747, "learning_rate": 1.9372618764280102e-05, "loss": 0.5669, "step": 3929 }, { "epoch": 0.1400146071218626, "grad_norm": 1.8295039319906725, "learning_rate": 1.937221640940091e-05, "loss": 0.6168, "step": 3930 }, { "epoch": 0.1400502342483567, "grad_norm": 1.2628866355619401, "learning_rate": 1.937181392972411e-05, "loss": 0.5184, "step": 3931 }, { "epoch": 0.14008586137485082, "grad_norm": 1.2348888189462632, "learning_rate": 1.9371411325255057e-05, "loss": 0.4987, "step": 3932 }, { "epoch": 0.14012148850134493, "grad_norm": 1.9933104777718507, "learning_rate": 1.9371008595999114e-05, "loss": 0.6475, "step": 3933 }, { "epoch": 0.14015711562783903, "grad_norm": 1.2124627627848619, "learning_rate": 1.9370605741961638e-05, "loss": 0.559, "step": 3934 }, { "epoch": 0.14019274275433316, "grad_norm": 1.251396531093879, "learning_rate": 1.9370202763147997e-05, "loss": 0.5055, "step": 3935 }, { "epoch": 0.14022836988082726, "grad_norm": 1.2343187301789464, "learning_rate": 1.9369799659563556e-05, "loss": 0.4965, "step": 3936 }, { "epoch": 0.14026399700732137, "grad_norm": 1.6790416758726212, "learning_rate": 1.9369396431213686e-05, "loss": 0.6553, "step": 3937 }, { "epoch": 0.1402996241338155, "grad_norm": 1.3489286773965208, "learning_rate": 1.936899307810375e-05, "loss": 0.565, "step": 3938 }, { "epoch": 0.1403352512603096, "grad_norm": 1.2944110041583916, "learning_rate": 1.9368589600239127e-05, "loss": 0.5441, "step": 3939 }, { "epoch": 0.1403708783868037, "grad_norm": 1.3584533512268855, "learning_rate": 1.9368185997625184e-05, "loss": 0.5867, "step": 3940 }, { "epoch": 0.14040650551329784, "grad_norm": 1.2525773687848136, "learning_rate": 1.9367782270267294e-05, "loss": 0.5643, "step": 3941 }, { "epoch": 0.14044213263979194, "grad_norm": 1.5948083175684915, "learning_rate": 1.936737841817084e-05, "loss": 0.6907, "step": 3942 }, { "epoch": 0.14047775976628604, "grad_norm": 2.1127575798272913, "learning_rate": 1.936697444134119e-05, "loss": 0.7161, "step": 3943 }, { "epoch": 0.14051338689278017, "grad_norm": 1.8066722116267009, "learning_rate": 1.936657033978373e-05, "loss": 0.5439, "step": 3944 }, { "epoch": 0.14054901401927428, "grad_norm": 2.0325045917256848, "learning_rate": 1.9366166113503843e-05, "loss": 0.6341, "step": 3945 }, { "epoch": 0.14058464114576838, "grad_norm": 1.2752670267655335, "learning_rate": 1.9365761762506903e-05, "loss": 0.5402, "step": 3946 }, { "epoch": 0.1406202682722625, "grad_norm": 1.2681690334953257, "learning_rate": 1.93653572867983e-05, "loss": 0.532, "step": 3947 }, { "epoch": 0.14065589539875661, "grad_norm": 2.2237133850490296, "learning_rate": 1.936495268638342e-05, "loss": 0.6296, "step": 3948 }, { "epoch": 0.14069152252525072, "grad_norm": 1.3072786448847717, "learning_rate": 1.9364547961267645e-05, "loss": 0.5854, "step": 3949 }, { "epoch": 0.14072714965174485, "grad_norm": 1.9538243713592514, "learning_rate": 1.9364143111456374e-05, "loss": 0.5889, "step": 3950 }, { "epoch": 0.14076277677823895, "grad_norm": 1.7687911721613108, "learning_rate": 1.9363738136954988e-05, "loss": 0.6072, "step": 3951 }, { "epoch": 0.14079840390473305, "grad_norm": 1.2066268758065193, "learning_rate": 1.9363333037768886e-05, "loss": 0.5428, "step": 3952 }, { "epoch": 0.14083403103122719, "grad_norm": 1.253679218541803, "learning_rate": 1.9362927813903457e-05, "loss": 0.5741, "step": 3953 }, { "epoch": 0.1408696581577213, "grad_norm": 2.3794168917056835, "learning_rate": 1.9362522465364106e-05, "loss": 0.6262, "step": 3954 }, { "epoch": 0.1409052852842154, "grad_norm": 1.7471834518233267, "learning_rate": 1.9362116992156223e-05, "loss": 0.6721, "step": 3955 }, { "epoch": 0.14094091241070952, "grad_norm": 1.205149668328003, "learning_rate": 1.9361711394285202e-05, "loss": 0.5183, "step": 3956 }, { "epoch": 0.14097653953720363, "grad_norm": 1.3377293550214666, "learning_rate": 1.9361305671756458e-05, "loss": 0.4768, "step": 3957 }, { "epoch": 0.14101216666369773, "grad_norm": 1.252610962992069, "learning_rate": 1.936089982457538e-05, "loss": 0.5285, "step": 3958 }, { "epoch": 0.14104779379019186, "grad_norm": 1.299393756427583, "learning_rate": 1.9360493852747383e-05, "loss": 0.5239, "step": 3959 }, { "epoch": 0.14108342091668596, "grad_norm": 1.626793769685862, "learning_rate": 1.9360087756277862e-05, "loss": 0.5312, "step": 3960 }, { "epoch": 0.14111904804318007, "grad_norm": 1.4637090386271252, "learning_rate": 1.9359681535172233e-05, "loss": 0.6283, "step": 3961 }, { "epoch": 0.1411546751696742, "grad_norm": 1.202518297443932, "learning_rate": 1.9359275189435904e-05, "loss": 0.5271, "step": 3962 }, { "epoch": 0.1411903022961683, "grad_norm": 2.010417478586903, "learning_rate": 1.9358868719074283e-05, "loss": 0.6675, "step": 3963 }, { "epoch": 0.1412259294226624, "grad_norm": 1.257592185876636, "learning_rate": 1.9358462124092784e-05, "loss": 0.4908, "step": 3964 }, { "epoch": 0.14126155654915654, "grad_norm": 1.1524556305865137, "learning_rate": 1.935805540449682e-05, "loss": 0.5194, "step": 3965 }, { "epoch": 0.14129718367565064, "grad_norm": 1.9008040001114166, "learning_rate": 1.9357648560291807e-05, "loss": 0.617, "step": 3966 }, { "epoch": 0.14133281080214474, "grad_norm": 1.3653026195987346, "learning_rate": 1.9357241591483165e-05, "loss": 0.5919, "step": 3967 }, { "epoch": 0.14136843792863887, "grad_norm": 1.8743681628220434, "learning_rate": 1.935683449807631e-05, "loss": 0.6352, "step": 3968 }, { "epoch": 0.14140406505513298, "grad_norm": 1.5973091949046045, "learning_rate": 1.935642728007666e-05, "loss": 0.5988, "step": 3969 }, { "epoch": 0.14143969218162708, "grad_norm": 1.372480693158351, "learning_rate": 1.9356019937489647e-05, "loss": 0.5197, "step": 3970 }, { "epoch": 0.1414753193081212, "grad_norm": 1.578790711796239, "learning_rate": 1.9355612470320683e-05, "loss": 0.6546, "step": 3971 }, { "epoch": 0.14151094643461531, "grad_norm": 1.3144527904096601, "learning_rate": 1.9355204878575202e-05, "loss": 0.5424, "step": 3972 }, { "epoch": 0.14154657356110942, "grad_norm": 2.372856852982992, "learning_rate": 1.9354797162258632e-05, "loss": 0.6538, "step": 3973 }, { "epoch": 0.14158220068760355, "grad_norm": 1.4375805124366865, "learning_rate": 1.9354389321376398e-05, "loss": 0.5249, "step": 3974 }, { "epoch": 0.14161782781409765, "grad_norm": 1.173220969021677, "learning_rate": 1.935398135593393e-05, "loss": 0.5346, "step": 3975 }, { "epoch": 0.14165345494059176, "grad_norm": 1.9012960528648621, "learning_rate": 1.9353573265936663e-05, "loss": 0.6812, "step": 3976 }, { "epoch": 0.1416890820670859, "grad_norm": 1.898761023833307, "learning_rate": 1.935316505139003e-05, "loss": 0.5818, "step": 3977 }, { "epoch": 0.14172470919358, "grad_norm": 1.9413777605177018, "learning_rate": 1.9352756712299467e-05, "loss": 0.6182, "step": 3978 }, { "epoch": 0.1417603363200741, "grad_norm": 1.3101132150610963, "learning_rate": 1.935234824867041e-05, "loss": 0.5456, "step": 3979 }, { "epoch": 0.14179596344656822, "grad_norm": 1.1275056196282192, "learning_rate": 1.93519396605083e-05, "loss": 0.5594, "step": 3980 }, { "epoch": 0.14183159057306233, "grad_norm": 2.0818900525523207, "learning_rate": 1.9351530947818577e-05, "loss": 0.5999, "step": 3981 }, { "epoch": 0.14186721769955643, "grad_norm": 1.2373599958785582, "learning_rate": 1.9351122110606682e-05, "loss": 0.4687, "step": 3982 }, { "epoch": 0.14190284482605056, "grad_norm": 1.122388203312123, "learning_rate": 1.9350713148878065e-05, "loss": 0.467, "step": 3983 }, { "epoch": 0.14193847195254466, "grad_norm": 2.038622217491876, "learning_rate": 1.9350304062638162e-05, "loss": 0.568, "step": 3984 }, { "epoch": 0.14197409907903877, "grad_norm": 1.7060312939860942, "learning_rate": 1.9349894851892426e-05, "loss": 0.5919, "step": 3985 }, { "epoch": 0.1420097262055329, "grad_norm": 1.6785814572240625, "learning_rate": 1.9349485516646303e-05, "loss": 0.6075, "step": 3986 }, { "epoch": 0.142045353332027, "grad_norm": 1.2471006960754707, "learning_rate": 1.934907605690525e-05, "loss": 0.5525, "step": 3987 }, { "epoch": 0.1420809804585211, "grad_norm": 1.797264928479911, "learning_rate": 1.9348666472674712e-05, "loss": 0.6526, "step": 3988 }, { "epoch": 0.14211660758501524, "grad_norm": 2.4751303738603485, "learning_rate": 1.9348256763960146e-05, "loss": 0.5055, "step": 3989 }, { "epoch": 0.14215223471150934, "grad_norm": 1.3183163230689816, "learning_rate": 1.9347846930767006e-05, "loss": 0.5003, "step": 3990 }, { "epoch": 0.14218786183800344, "grad_norm": 1.2490783236241836, "learning_rate": 1.9347436973100753e-05, "loss": 0.5633, "step": 3991 }, { "epoch": 0.14222348896449757, "grad_norm": 1.1978219428005925, "learning_rate": 1.934702689096684e-05, "loss": 0.5229, "step": 3992 }, { "epoch": 0.14225911609099168, "grad_norm": 2.1682558079311613, "learning_rate": 1.934661668437073e-05, "loss": 0.6957, "step": 3993 }, { "epoch": 0.14229474321748578, "grad_norm": 1.2803722315553112, "learning_rate": 1.934620635331789e-05, "loss": 0.5342, "step": 3994 }, { "epoch": 0.1423303703439799, "grad_norm": 2.1930102878327875, "learning_rate": 1.934579589781378e-05, "loss": 0.605, "step": 3995 }, { "epoch": 0.14236599747047401, "grad_norm": 1.6231680963879693, "learning_rate": 1.934538531786386e-05, "loss": 0.5562, "step": 3996 }, { "epoch": 0.14240162459696812, "grad_norm": 1.6283616721913081, "learning_rate": 1.9344974613473608e-05, "loss": 0.6919, "step": 3997 }, { "epoch": 0.14243725172346225, "grad_norm": 1.735481682429743, "learning_rate": 1.9344563784648486e-05, "loss": 0.6098, "step": 3998 }, { "epoch": 0.14247287884995635, "grad_norm": 1.251961905627344, "learning_rate": 1.9344152831393963e-05, "loss": 0.4946, "step": 3999 }, { "epoch": 0.14250850597645046, "grad_norm": 1.918978406235816, "learning_rate": 1.934374175371552e-05, "loss": 0.7121, "step": 4000 }, { "epoch": 0.1425441331029446, "grad_norm": 1.8650998466626347, "learning_rate": 1.9343330551618616e-05, "loss": 0.6018, "step": 4001 }, { "epoch": 0.1425797602294387, "grad_norm": 1.315879370132805, "learning_rate": 1.934291922510874e-05, "loss": 0.5227, "step": 4002 }, { "epoch": 0.1426153873559328, "grad_norm": 2.531018008138239, "learning_rate": 1.9342507774191362e-05, "loss": 0.5776, "step": 4003 }, { "epoch": 0.14265101448242692, "grad_norm": 2.3560765551226117, "learning_rate": 1.9342096198871963e-05, "loss": 0.6067, "step": 4004 }, { "epoch": 0.14268664160892103, "grad_norm": 2.6741720704033654, "learning_rate": 1.934168449915603e-05, "loss": 0.6279, "step": 4005 }, { "epoch": 0.14272226873541516, "grad_norm": 1.6572640011915651, "learning_rate": 1.934127267504903e-05, "loss": 0.5791, "step": 4006 }, { "epoch": 0.14275789586190926, "grad_norm": 1.2028709336396224, "learning_rate": 1.934086072655646e-05, "loss": 0.4635, "step": 4007 }, { "epoch": 0.14279352298840337, "grad_norm": 1.7618928217935566, "learning_rate": 1.93404486536838e-05, "loss": 0.6719, "step": 4008 }, { "epoch": 0.1428291501148975, "grad_norm": 1.2464297770694952, "learning_rate": 1.9340036456436534e-05, "loss": 0.5045, "step": 4009 }, { "epoch": 0.1428647772413916, "grad_norm": 1.6893032232506746, "learning_rate": 1.933962413482016e-05, "loss": 0.5652, "step": 4010 }, { "epoch": 0.1429004043678857, "grad_norm": 2.6530703692320485, "learning_rate": 1.933921168884016e-05, "loss": 0.5932, "step": 4011 }, { "epoch": 0.14293603149437983, "grad_norm": 1.4122628888562774, "learning_rate": 1.9338799118502026e-05, "loss": 0.5106, "step": 4012 }, { "epoch": 0.14297165862087394, "grad_norm": 1.864745149914943, "learning_rate": 1.9338386423811254e-05, "loss": 0.5273, "step": 4013 }, { "epoch": 0.14300728574736804, "grad_norm": 1.329453613928598, "learning_rate": 1.9337973604773345e-05, "loss": 0.5432, "step": 4014 }, { "epoch": 0.14304291287386217, "grad_norm": 1.3267365534003523, "learning_rate": 1.9337560661393788e-05, "loss": 0.5258, "step": 4015 }, { "epoch": 0.14307854000035627, "grad_norm": 1.669064630193205, "learning_rate": 1.9337147593678085e-05, "loss": 0.6439, "step": 4016 }, { "epoch": 0.14311416712685038, "grad_norm": 3.0718286327499174, "learning_rate": 1.9336734401631735e-05, "loss": 0.6489, "step": 4017 }, { "epoch": 0.1431497942533445, "grad_norm": 1.3047271854190798, "learning_rate": 1.9336321085260236e-05, "loss": 0.5772, "step": 4018 }, { "epoch": 0.1431854213798386, "grad_norm": 1.1804253265982936, "learning_rate": 1.93359076445691e-05, "loss": 0.5182, "step": 4019 }, { "epoch": 0.14322104850633272, "grad_norm": 1.6050735155376448, "learning_rate": 1.933549407956383e-05, "loss": 0.6289, "step": 4020 }, { "epoch": 0.14325667563282685, "grad_norm": 1.618153760277658, "learning_rate": 1.9335080390249928e-05, "loss": 0.5788, "step": 4021 }, { "epoch": 0.14329230275932095, "grad_norm": 1.754204713897076, "learning_rate": 1.933466657663291e-05, "loss": 0.6429, "step": 4022 }, { "epoch": 0.14332792988581505, "grad_norm": 2.1033984115935134, "learning_rate": 1.933425263871828e-05, "loss": 0.6857, "step": 4023 }, { "epoch": 0.14336355701230918, "grad_norm": 1.718608816168316, "learning_rate": 1.933383857651155e-05, "loss": 0.5794, "step": 4024 }, { "epoch": 0.1433991841388033, "grad_norm": 1.9741644554781979, "learning_rate": 1.9333424390018237e-05, "loss": 0.6499, "step": 4025 }, { "epoch": 0.1434348112652974, "grad_norm": 1.7190353757355965, "learning_rate": 1.9333010079243856e-05, "loss": 0.5808, "step": 4026 }, { "epoch": 0.14347043839179152, "grad_norm": 2.290789888620692, "learning_rate": 1.9332595644193918e-05, "loss": 0.5995, "step": 4027 }, { "epoch": 0.14350606551828562, "grad_norm": 1.9182712271948932, "learning_rate": 1.9332181084873947e-05, "loss": 0.6483, "step": 4028 }, { "epoch": 0.14354169264477973, "grad_norm": 2.0726296590286077, "learning_rate": 1.9331766401289466e-05, "loss": 0.6409, "step": 4029 }, { "epoch": 0.14357731977127386, "grad_norm": 1.1940583288113165, "learning_rate": 1.933135159344599e-05, "loss": 0.4833, "step": 4030 }, { "epoch": 0.14361294689776796, "grad_norm": 1.2991981761294356, "learning_rate": 1.9330936661349044e-05, "loss": 0.5139, "step": 4031 }, { "epoch": 0.14364857402426207, "grad_norm": 1.8852286046900995, "learning_rate": 1.9330521605004157e-05, "loss": 0.5839, "step": 4032 }, { "epoch": 0.1436842011507562, "grad_norm": 1.1843747334941268, "learning_rate": 1.9330106424416852e-05, "loss": 0.5254, "step": 4033 }, { "epoch": 0.1437198282772503, "grad_norm": 1.2639290776414263, "learning_rate": 1.932969111959266e-05, "loss": 0.5128, "step": 4034 }, { "epoch": 0.1437554554037444, "grad_norm": 2.378289371150489, "learning_rate": 1.9329275690537108e-05, "loss": 0.5671, "step": 4035 }, { "epoch": 0.14379108253023853, "grad_norm": 1.277085031253293, "learning_rate": 1.932886013725573e-05, "loss": 0.5399, "step": 4036 }, { "epoch": 0.14382670965673264, "grad_norm": 2.156602153351545, "learning_rate": 1.9328444459754057e-05, "loss": 0.5689, "step": 4037 }, { "epoch": 0.14386233678322674, "grad_norm": 1.8103787752025122, "learning_rate": 1.932802865803763e-05, "loss": 0.6629, "step": 4038 }, { "epoch": 0.14389796390972087, "grad_norm": 1.2390230820346224, "learning_rate": 1.9327612732111976e-05, "loss": 0.5546, "step": 4039 }, { "epoch": 0.14393359103621498, "grad_norm": 1.6791382090114437, "learning_rate": 1.9327196681982644e-05, "loss": 0.6814, "step": 4040 }, { "epoch": 0.14396921816270908, "grad_norm": 1.3165155295009576, "learning_rate": 1.9326780507655165e-05, "loss": 0.4966, "step": 4041 }, { "epoch": 0.1440048452892032, "grad_norm": 1.7382401050552314, "learning_rate": 1.9326364209135088e-05, "loss": 0.683, "step": 4042 }, { "epoch": 0.1440404724156973, "grad_norm": 1.2840782486984637, "learning_rate": 1.9325947786427953e-05, "loss": 0.5701, "step": 4043 }, { "epoch": 0.14407609954219142, "grad_norm": 1.1667149754881483, "learning_rate": 1.9325531239539303e-05, "loss": 0.5097, "step": 4044 }, { "epoch": 0.14411172666868555, "grad_norm": 1.8671579330804176, "learning_rate": 1.9325114568474684e-05, "loss": 0.6902, "step": 4045 }, { "epoch": 0.14414735379517965, "grad_norm": 1.8033431142691854, "learning_rate": 1.932469777323965e-05, "loss": 0.5518, "step": 4046 }, { "epoch": 0.14418298092167375, "grad_norm": 2.0008819784997858, "learning_rate": 1.9324280853839744e-05, "loss": 0.7015, "step": 4047 }, { "epoch": 0.14421860804816788, "grad_norm": 1.1665712550709824, "learning_rate": 1.9323863810280525e-05, "loss": 0.4996, "step": 4048 }, { "epoch": 0.144254235174662, "grad_norm": 1.169384087483984, "learning_rate": 1.9323446642567543e-05, "loss": 0.5183, "step": 4049 }, { "epoch": 0.1442898623011561, "grad_norm": 1.830746136153869, "learning_rate": 1.932302935070635e-05, "loss": 0.6315, "step": 4050 }, { "epoch": 0.14432548942765022, "grad_norm": 1.7400641472549696, "learning_rate": 1.93226119347025e-05, "loss": 0.6893, "step": 4051 }, { "epoch": 0.14436111655414433, "grad_norm": 1.8100542673914366, "learning_rate": 1.9322194394561563e-05, "loss": 0.6157, "step": 4052 }, { "epoch": 0.14439674368063843, "grad_norm": 1.2523112785610484, "learning_rate": 1.932177673028909e-05, "loss": 0.5125, "step": 4053 }, { "epoch": 0.14443237080713256, "grad_norm": 1.935321406348868, "learning_rate": 1.9321358941890645e-05, "loss": 0.6279, "step": 4054 }, { "epoch": 0.14446799793362666, "grad_norm": 1.8829742529754054, "learning_rate": 1.932094102937179e-05, "loss": 0.5479, "step": 4055 }, { "epoch": 0.14450362506012077, "grad_norm": 1.114697259053382, "learning_rate": 1.932052299273809e-05, "loss": 0.5092, "step": 4056 }, { "epoch": 0.1445392521866149, "grad_norm": 1.6081515269940345, "learning_rate": 1.9320104831995113e-05, "loss": 0.6588, "step": 4057 }, { "epoch": 0.144574879313109, "grad_norm": 1.3219569230343036, "learning_rate": 1.9319686547148422e-05, "loss": 0.5172, "step": 4058 }, { "epoch": 0.1446105064396031, "grad_norm": 1.859458467379262, "learning_rate": 1.9319268138203595e-05, "loss": 0.5716, "step": 4059 }, { "epoch": 0.14464613356609723, "grad_norm": 1.55156242319248, "learning_rate": 1.9318849605166195e-05, "loss": 0.5146, "step": 4060 }, { "epoch": 0.14468176069259134, "grad_norm": 1.2084182195220294, "learning_rate": 1.9318430948041802e-05, "loss": 0.5309, "step": 4061 }, { "epoch": 0.14471738781908544, "grad_norm": 1.3398961167617134, "learning_rate": 1.9318012166835987e-05, "loss": 0.5649, "step": 4062 }, { "epoch": 0.14475301494557957, "grad_norm": 1.8067635839414393, "learning_rate": 1.9317593261554325e-05, "loss": 0.5506, "step": 4063 }, { "epoch": 0.14478864207207368, "grad_norm": 1.4147772008645467, "learning_rate": 1.9317174232202398e-05, "loss": 0.5813, "step": 4064 }, { "epoch": 0.14482426919856778, "grad_norm": 2.0422155236464348, "learning_rate": 1.931675507878578e-05, "loss": 0.5995, "step": 4065 }, { "epoch": 0.1448598963250619, "grad_norm": 1.3262653767161807, "learning_rate": 1.9316335801310064e-05, "loss": 0.5284, "step": 4066 }, { "epoch": 0.144895523451556, "grad_norm": 1.5521516521964474, "learning_rate": 1.931591639978082e-05, "loss": 0.6755, "step": 4067 }, { "epoch": 0.14493115057805012, "grad_norm": 1.325458381961559, "learning_rate": 1.9315496874203637e-05, "loss": 0.4925, "step": 4068 }, { "epoch": 0.14496677770454425, "grad_norm": 1.7961105888447733, "learning_rate": 1.9315077224584106e-05, "loss": 0.6367, "step": 4069 }, { "epoch": 0.14500240483103835, "grad_norm": 1.2830921331964873, "learning_rate": 1.9314657450927807e-05, "loss": 0.5643, "step": 4070 }, { "epoch": 0.14503803195753245, "grad_norm": 1.872941059188133, "learning_rate": 1.9314237553240335e-05, "loss": 0.611, "step": 4071 }, { "epoch": 0.14507365908402658, "grad_norm": 2.0245359161920558, "learning_rate": 1.931381753152728e-05, "loss": 0.6393, "step": 4072 }, { "epoch": 0.1451092862105207, "grad_norm": 1.5297801468170258, "learning_rate": 1.9313397385794236e-05, "loss": 0.6681, "step": 4073 }, { "epoch": 0.1451449133370148, "grad_norm": 2.695235371149907, "learning_rate": 1.9312977116046793e-05, "loss": 0.6847, "step": 4074 }, { "epoch": 0.14518054046350892, "grad_norm": 1.6717826475039241, "learning_rate": 1.9312556722290553e-05, "loss": 0.6298, "step": 4075 }, { "epoch": 0.14521616759000303, "grad_norm": 1.4890426663465584, "learning_rate": 1.931213620453111e-05, "loss": 0.5552, "step": 4076 }, { "epoch": 0.14525179471649713, "grad_norm": 1.1913576155710572, "learning_rate": 1.931171556277406e-05, "loss": 0.4861, "step": 4077 }, { "epoch": 0.14528742184299126, "grad_norm": 1.7820144637917326, "learning_rate": 1.9311294797025016e-05, "loss": 0.58, "step": 4078 }, { "epoch": 0.14532304896948536, "grad_norm": 1.1953988339257515, "learning_rate": 1.931087390728957e-05, "loss": 0.497, "step": 4079 }, { "epoch": 0.14535867609597947, "grad_norm": 2.1833209214884697, "learning_rate": 1.931045289357333e-05, "loss": 0.5647, "step": 4080 }, { "epoch": 0.1453943032224736, "grad_norm": 1.91951704676152, "learning_rate": 1.9310031755881903e-05, "loss": 0.6228, "step": 4081 }, { "epoch": 0.1454299303489677, "grad_norm": 2.0103532664082695, "learning_rate": 1.9309610494220895e-05, "loss": 0.543, "step": 4082 }, { "epoch": 0.1454655574754618, "grad_norm": 1.1433143639254075, "learning_rate": 1.930918910859592e-05, "loss": 0.4849, "step": 4083 }, { "epoch": 0.14550118460195594, "grad_norm": 1.2928622059416623, "learning_rate": 1.930876759901258e-05, "loss": 0.5259, "step": 4084 }, { "epoch": 0.14553681172845004, "grad_norm": 1.3296193496522164, "learning_rate": 1.9308345965476498e-05, "loss": 0.5539, "step": 4085 }, { "epoch": 0.14557243885494414, "grad_norm": 1.3486680389423622, "learning_rate": 1.9307924207993278e-05, "loss": 0.5261, "step": 4086 }, { "epoch": 0.14560806598143827, "grad_norm": 1.3681131995670766, "learning_rate": 1.9307502326568545e-05, "loss": 0.5271, "step": 4087 }, { "epoch": 0.14564369310793238, "grad_norm": 2.248463003828932, "learning_rate": 1.9307080321207913e-05, "loss": 0.5468, "step": 4088 }, { "epoch": 0.14567932023442648, "grad_norm": 1.1840103737019494, "learning_rate": 1.9306658191917e-05, "loss": 0.49, "step": 4089 }, { "epoch": 0.1457149473609206, "grad_norm": 2.2952708447486017, "learning_rate": 1.930623593870143e-05, "loss": 0.665, "step": 4090 }, { "epoch": 0.1457505744874147, "grad_norm": 1.410357849935325, "learning_rate": 1.930581356156682e-05, "loss": 0.5069, "step": 4091 }, { "epoch": 0.14578620161390882, "grad_norm": 1.7107996118577022, "learning_rate": 1.93053910605188e-05, "loss": 0.5958, "step": 4092 }, { "epoch": 0.14582182874040295, "grad_norm": 1.2091373434864732, "learning_rate": 1.9304968435562995e-05, "loss": 0.5108, "step": 4093 }, { "epoch": 0.14585745586689705, "grad_norm": 1.5641641984922583, "learning_rate": 1.930454568670503e-05, "loss": 0.6186, "step": 4094 }, { "epoch": 0.14589308299339115, "grad_norm": 1.2212937552262455, "learning_rate": 1.9304122813950538e-05, "loss": 0.5057, "step": 4095 }, { "epoch": 0.14592871011988529, "grad_norm": 1.307724031523014, "learning_rate": 1.9303699817305144e-05, "loss": 0.5611, "step": 4096 }, { "epoch": 0.1459643372463794, "grad_norm": 1.2572210374365322, "learning_rate": 1.9303276696774486e-05, "loss": 0.5127, "step": 4097 }, { "epoch": 0.14599996437287352, "grad_norm": 1.542830524164249, "learning_rate": 1.9302853452364196e-05, "loss": 0.5519, "step": 4098 }, { "epoch": 0.14603559149936762, "grad_norm": 1.2168741697110168, "learning_rate": 1.930243008407991e-05, "loss": 0.5212, "step": 4099 }, { "epoch": 0.14607121862586173, "grad_norm": 1.2231286307543476, "learning_rate": 1.9302006591927266e-05, "loss": 0.5176, "step": 4100 }, { "epoch": 0.14610684575235586, "grad_norm": 1.6119496348196427, "learning_rate": 1.9301582975911903e-05, "loss": 0.6201, "step": 4101 }, { "epoch": 0.14614247287884996, "grad_norm": 1.200989052665491, "learning_rate": 1.9301159236039458e-05, "loss": 0.4781, "step": 4102 }, { "epoch": 0.14617810000534406, "grad_norm": 1.7533310507774302, "learning_rate": 1.930073537231558e-05, "loss": 0.4606, "step": 4103 }, { "epoch": 0.1462137271318382, "grad_norm": 1.3983920333700068, "learning_rate": 1.9300311384745908e-05, "loss": 0.5379, "step": 4104 }, { "epoch": 0.1462493542583323, "grad_norm": 1.3137202042839586, "learning_rate": 1.9299887273336088e-05, "loss": 0.5014, "step": 4105 }, { "epoch": 0.1462849813848264, "grad_norm": 5.097423480838025, "learning_rate": 1.9299463038091772e-05, "loss": 0.5311, "step": 4106 }, { "epoch": 0.14632060851132053, "grad_norm": 2.3911018647183875, "learning_rate": 1.9299038679018606e-05, "loss": 0.5518, "step": 4107 }, { "epoch": 0.14635623563781464, "grad_norm": 1.5684736908705086, "learning_rate": 1.929861419612224e-05, "loss": 0.5629, "step": 4108 }, { "epoch": 0.14639186276430874, "grad_norm": 1.3233619509317995, "learning_rate": 1.929818958940832e-05, "loss": 0.4737, "step": 4109 }, { "epoch": 0.14642748989080287, "grad_norm": 1.3466557107166162, "learning_rate": 1.9297764858882516e-05, "loss": 0.5529, "step": 4110 }, { "epoch": 0.14646311701729697, "grad_norm": 1.419451878300541, "learning_rate": 1.929734000455047e-05, "loss": 0.5788, "step": 4111 }, { "epoch": 0.14649874414379108, "grad_norm": 1.2527631278597584, "learning_rate": 1.9296915026417846e-05, "loss": 0.512, "step": 4112 }, { "epoch": 0.1465343712702852, "grad_norm": 1.2913187385270966, "learning_rate": 1.9296489924490296e-05, "loss": 0.506, "step": 4113 }, { "epoch": 0.1465699983967793, "grad_norm": 1.7602651861685035, "learning_rate": 1.9296064698773485e-05, "loss": 0.684, "step": 4114 }, { "epoch": 0.14660562552327341, "grad_norm": 2.4208522403606003, "learning_rate": 1.9295639349273076e-05, "loss": 0.6697, "step": 4115 }, { "epoch": 0.14664125264976755, "grad_norm": 1.249192746580075, "learning_rate": 1.9295213875994735e-05, "loss": 0.5454, "step": 4116 }, { "epoch": 0.14667687977626165, "grad_norm": 1.6185938534007187, "learning_rate": 1.929478827894412e-05, "loss": 0.5737, "step": 4117 }, { "epoch": 0.14671250690275575, "grad_norm": 1.1779663377989855, "learning_rate": 1.9294362558126904e-05, "loss": 0.5361, "step": 4118 }, { "epoch": 0.14674813402924988, "grad_norm": 1.8289896413584326, "learning_rate": 1.9293936713548754e-05, "loss": 0.6031, "step": 4119 }, { "epoch": 0.14678376115574399, "grad_norm": 1.809116287268873, "learning_rate": 1.929351074521534e-05, "loss": 0.6363, "step": 4120 }, { "epoch": 0.1468193882822381, "grad_norm": 1.3452561235938858, "learning_rate": 1.9293084653132336e-05, "loss": 0.5223, "step": 4121 }, { "epoch": 0.14685501540873222, "grad_norm": 1.2779250225403984, "learning_rate": 1.9292658437305415e-05, "loss": 0.5023, "step": 4122 }, { "epoch": 0.14689064253522632, "grad_norm": 1.2463091502524941, "learning_rate": 1.929223209774025e-05, "loss": 0.4909, "step": 4123 }, { "epoch": 0.14692626966172043, "grad_norm": 1.8240134433464008, "learning_rate": 1.929180563444252e-05, "loss": 0.6261, "step": 4124 }, { "epoch": 0.14696189678821456, "grad_norm": 1.9832235904780822, "learning_rate": 1.9291379047417903e-05, "loss": 0.592, "step": 4125 }, { "epoch": 0.14699752391470866, "grad_norm": 1.8584125045541369, "learning_rate": 1.929095233667208e-05, "loss": 0.6145, "step": 4126 }, { "epoch": 0.14703315104120276, "grad_norm": 2.7926292603222986, "learning_rate": 1.9290525502210733e-05, "loss": 0.6141, "step": 4127 }, { "epoch": 0.1470687781676969, "grad_norm": 1.3599696701648556, "learning_rate": 1.9290098544039546e-05, "loss": 0.5306, "step": 4128 }, { "epoch": 0.147104405294191, "grad_norm": 1.8411976171958564, "learning_rate": 1.9289671462164202e-05, "loss": 0.5898, "step": 4129 }, { "epoch": 0.1471400324206851, "grad_norm": 1.9188761765988571, "learning_rate": 1.928924425659039e-05, "loss": 0.6164, "step": 4130 }, { "epoch": 0.14717565954717923, "grad_norm": 1.674677812117265, "learning_rate": 1.9288816927323798e-05, "loss": 0.6081, "step": 4131 }, { "epoch": 0.14721128667367334, "grad_norm": 2.125555732547844, "learning_rate": 1.9288389474370116e-05, "loss": 0.6851, "step": 4132 }, { "epoch": 0.14724691380016744, "grad_norm": 1.7537113966290738, "learning_rate": 1.928796189773504e-05, "loss": 0.6213, "step": 4133 }, { "epoch": 0.14728254092666157, "grad_norm": 2.2323734158862774, "learning_rate": 1.9287534197424255e-05, "loss": 0.5553, "step": 4134 }, { "epoch": 0.14731816805315567, "grad_norm": 2.067580642087445, "learning_rate": 1.928710637344346e-05, "loss": 0.6479, "step": 4135 }, { "epoch": 0.14735379517964978, "grad_norm": 1.3557381380027609, "learning_rate": 1.9286678425798353e-05, "loss": 0.5158, "step": 4136 }, { "epoch": 0.1473894223061439, "grad_norm": 1.798324480068252, "learning_rate": 1.9286250354494634e-05, "loss": 0.6667, "step": 4137 }, { "epoch": 0.147425049432638, "grad_norm": 1.9829950262545297, "learning_rate": 1.9285822159537996e-05, "loss": 0.6041, "step": 4138 }, { "epoch": 0.14746067655913211, "grad_norm": 1.849837338937673, "learning_rate": 1.9285393840934152e-05, "loss": 0.5976, "step": 4139 }, { "epoch": 0.14749630368562625, "grad_norm": 1.887308015904021, "learning_rate": 1.9284965398688793e-05, "loss": 0.5307, "step": 4140 }, { "epoch": 0.14753193081212035, "grad_norm": 2.118224345631908, "learning_rate": 1.9284536832807635e-05, "loss": 0.6091, "step": 4141 }, { "epoch": 0.14756755793861445, "grad_norm": 1.6212301723587614, "learning_rate": 1.9284108143296378e-05, "loss": 0.5792, "step": 4142 }, { "epoch": 0.14760318506510858, "grad_norm": 3.6408532609449153, "learning_rate": 1.9283679330160726e-05, "loss": 0.6344, "step": 4143 }, { "epoch": 0.1476388121916027, "grad_norm": 1.297954014476582, "learning_rate": 1.92832503934064e-05, "loss": 0.5021, "step": 4144 }, { "epoch": 0.1476744393180968, "grad_norm": 1.9282502788703664, "learning_rate": 1.9282821333039106e-05, "loss": 0.6097, "step": 4145 }, { "epoch": 0.14771006644459092, "grad_norm": 1.5615457449425918, "learning_rate": 1.928239214906456e-05, "loss": 0.5379, "step": 4146 }, { "epoch": 0.14774569357108502, "grad_norm": 1.4611234712921923, "learning_rate": 1.928196284148847e-05, "loss": 0.4991, "step": 4147 }, { "epoch": 0.14778132069757913, "grad_norm": 1.2434695438487913, "learning_rate": 1.9281533410316558e-05, "loss": 0.506, "step": 4148 }, { "epoch": 0.14781694782407326, "grad_norm": 1.2616419977888351, "learning_rate": 1.9281103855554545e-05, "loss": 0.5627, "step": 4149 }, { "epoch": 0.14785257495056736, "grad_norm": 1.9013875200792782, "learning_rate": 1.9280674177208145e-05, "loss": 0.5963, "step": 4150 }, { "epoch": 0.14788820207706146, "grad_norm": 1.742584385411796, "learning_rate": 1.9280244375283083e-05, "loss": 0.5622, "step": 4151 }, { "epoch": 0.1479238292035556, "grad_norm": 1.4121006363031974, "learning_rate": 1.9279814449785076e-05, "loss": 0.5741, "step": 4152 }, { "epoch": 0.1479594563300497, "grad_norm": 1.6322190704112531, "learning_rate": 1.927938440071986e-05, "loss": 0.5296, "step": 4153 }, { "epoch": 0.1479950834565438, "grad_norm": 1.5182531186877226, "learning_rate": 1.927895422809315e-05, "loss": 0.5181, "step": 4154 }, { "epoch": 0.14803071058303793, "grad_norm": 1.317656545444537, "learning_rate": 1.9278523931910678e-05, "loss": 0.5276, "step": 4155 }, { "epoch": 0.14806633770953204, "grad_norm": 2.184779805098414, "learning_rate": 1.9278093512178175e-05, "loss": 0.5938, "step": 4156 }, { "epoch": 0.14810196483602614, "grad_norm": 1.7978106901569106, "learning_rate": 1.9277662968901377e-05, "loss": 0.6071, "step": 4157 }, { "epoch": 0.14813759196252027, "grad_norm": 1.2489321418443684, "learning_rate": 1.9277232302086006e-05, "loss": 0.5165, "step": 4158 }, { "epoch": 0.14817321908901437, "grad_norm": 1.7296047880041776, "learning_rate": 1.9276801511737805e-05, "loss": 0.5826, "step": 4159 }, { "epoch": 0.14820884621550848, "grad_norm": 1.3435348539745233, "learning_rate": 1.9276370597862507e-05, "loss": 0.523, "step": 4160 }, { "epoch": 0.1482444733420026, "grad_norm": 1.7845842098136313, "learning_rate": 1.927593956046585e-05, "loss": 0.7041, "step": 4161 }, { "epoch": 0.1482801004684967, "grad_norm": 1.3208215767013867, "learning_rate": 1.9275508399553577e-05, "loss": 0.5024, "step": 4162 }, { "epoch": 0.14831572759499081, "grad_norm": 1.3245501811881404, "learning_rate": 1.9275077115131424e-05, "loss": 0.4918, "step": 4163 }, { "epoch": 0.14835135472148495, "grad_norm": 1.2977648145668694, "learning_rate": 1.9274645707205138e-05, "loss": 0.5492, "step": 4164 }, { "epoch": 0.14838698184797905, "grad_norm": 1.310377027853668, "learning_rate": 1.927421417578046e-05, "loss": 0.5587, "step": 4165 }, { "epoch": 0.14842260897447315, "grad_norm": 1.3625529472053381, "learning_rate": 1.9273782520863137e-05, "loss": 0.5409, "step": 4166 }, { "epoch": 0.14845823610096728, "grad_norm": 1.14449602079648, "learning_rate": 1.927335074245892e-05, "loss": 0.5338, "step": 4167 }, { "epoch": 0.1484938632274614, "grad_norm": 1.5760256093958382, "learning_rate": 1.9272918840573558e-05, "loss": 0.595, "step": 4168 }, { "epoch": 0.1485294903539555, "grad_norm": 1.2539303621250426, "learning_rate": 1.9272486815212797e-05, "loss": 0.5509, "step": 4169 }, { "epoch": 0.14856511748044962, "grad_norm": 1.2419531724904278, "learning_rate": 1.9272054666382397e-05, "loss": 0.5321, "step": 4170 }, { "epoch": 0.14860074460694372, "grad_norm": 1.2259973482654432, "learning_rate": 1.9271622394088103e-05, "loss": 0.5369, "step": 4171 }, { "epoch": 0.14863637173343783, "grad_norm": 1.8944153862687416, "learning_rate": 1.9271189998335677e-05, "loss": 0.6472, "step": 4172 }, { "epoch": 0.14867199885993196, "grad_norm": 1.8372612138723872, "learning_rate": 1.927075747913088e-05, "loss": 0.6042, "step": 4173 }, { "epoch": 0.14870762598642606, "grad_norm": 1.2006945435411558, "learning_rate": 1.9270324836479465e-05, "loss": 0.5098, "step": 4174 }, { "epoch": 0.14874325311292017, "grad_norm": 1.2654685502938843, "learning_rate": 1.9269892070387196e-05, "loss": 0.5263, "step": 4175 }, { "epoch": 0.1487788802394143, "grad_norm": 1.224996335283906, "learning_rate": 1.9269459180859834e-05, "loss": 0.4697, "step": 4176 }, { "epoch": 0.1488145073659084, "grad_norm": 1.2937687670263012, "learning_rate": 1.9269026167903142e-05, "loss": 0.4946, "step": 4177 }, { "epoch": 0.1488501344924025, "grad_norm": 1.4198464376532784, "learning_rate": 1.926859303152289e-05, "loss": 0.4719, "step": 4178 }, { "epoch": 0.14888576161889663, "grad_norm": 1.8459203293874242, "learning_rate": 1.9268159771724845e-05, "loss": 0.6374, "step": 4179 }, { "epoch": 0.14892138874539074, "grad_norm": 1.2496134988430705, "learning_rate": 1.9267726388514775e-05, "loss": 0.5351, "step": 4180 }, { "epoch": 0.14895701587188484, "grad_norm": 1.159260703190745, "learning_rate": 1.926729288189845e-05, "loss": 0.475, "step": 4181 }, { "epoch": 0.14899264299837897, "grad_norm": 1.952023631973376, "learning_rate": 1.9266859251881637e-05, "loss": 0.6097, "step": 4182 }, { "epoch": 0.14902827012487307, "grad_norm": 1.7020575105392226, "learning_rate": 1.9266425498470122e-05, "loss": 0.6281, "step": 4183 }, { "epoch": 0.14906389725136718, "grad_norm": 1.5858921391468044, "learning_rate": 1.926599162166967e-05, "loss": 0.5962, "step": 4184 }, { "epoch": 0.1490995243778613, "grad_norm": 1.2827990510962417, "learning_rate": 1.9265557621486067e-05, "loss": 0.5147, "step": 4185 }, { "epoch": 0.1491351515043554, "grad_norm": 1.260272508469557, "learning_rate": 1.9265123497925088e-05, "loss": 0.5007, "step": 4186 }, { "epoch": 0.14917077863084952, "grad_norm": 1.2467178214863102, "learning_rate": 1.9264689250992514e-05, "loss": 0.5174, "step": 4187 }, { "epoch": 0.14920640575734365, "grad_norm": 1.8825878160341019, "learning_rate": 1.9264254880694123e-05, "loss": 0.6198, "step": 4188 }, { "epoch": 0.14924203288383775, "grad_norm": 1.2698190035201966, "learning_rate": 1.9263820387035705e-05, "loss": 0.5415, "step": 4189 }, { "epoch": 0.14927766001033185, "grad_norm": 1.2315720201580154, "learning_rate": 1.9263385770023048e-05, "loss": 0.5687, "step": 4190 }, { "epoch": 0.14931328713682598, "grad_norm": 1.6308280936844393, "learning_rate": 1.9262951029661926e-05, "loss": 0.6843, "step": 4191 }, { "epoch": 0.1493489142633201, "grad_norm": 1.5197425995873044, "learning_rate": 1.9262516165958143e-05, "loss": 0.6469, "step": 4192 }, { "epoch": 0.14938454138981422, "grad_norm": 1.5804949769027947, "learning_rate": 1.9262081178917483e-05, "loss": 0.6185, "step": 4193 }, { "epoch": 0.14942016851630832, "grad_norm": 1.7111266126338933, "learning_rate": 1.9261646068545736e-05, "loss": 0.6905, "step": 4194 }, { "epoch": 0.14945579564280242, "grad_norm": 1.3793818753882752, "learning_rate": 1.92612108348487e-05, "loss": 0.5229, "step": 4195 }, { "epoch": 0.14949142276929656, "grad_norm": 1.2379175271543355, "learning_rate": 1.9260775477832165e-05, "loss": 0.4923, "step": 4196 }, { "epoch": 0.14952704989579066, "grad_norm": 1.8882187299301463, "learning_rate": 1.9260339997501936e-05, "loss": 0.5847, "step": 4197 }, { "epoch": 0.14956267702228476, "grad_norm": 1.2793345896958714, "learning_rate": 1.9259904393863804e-05, "loss": 0.5292, "step": 4198 }, { "epoch": 0.1495983041487789, "grad_norm": 1.201166699978239, "learning_rate": 1.9259468666923575e-05, "loss": 0.5078, "step": 4199 }, { "epoch": 0.149633931275273, "grad_norm": 1.2715372225995438, "learning_rate": 1.9259032816687047e-05, "loss": 0.5442, "step": 4200 }, { "epoch": 0.1496695584017671, "grad_norm": 1.3114301728682092, "learning_rate": 1.9258596843160026e-05, "loss": 0.5215, "step": 4201 }, { "epoch": 0.14970518552826123, "grad_norm": 1.8082141218148906, "learning_rate": 1.925816074634832e-05, "loss": 0.6328, "step": 4202 }, { "epoch": 0.14974081265475533, "grad_norm": 1.588258937450341, "learning_rate": 1.9257724526257727e-05, "loss": 0.67, "step": 4203 }, { "epoch": 0.14977643978124944, "grad_norm": 1.7181171463668927, "learning_rate": 1.9257288182894064e-05, "loss": 0.5772, "step": 4204 }, { "epoch": 0.14981206690774357, "grad_norm": 1.7123923284448497, "learning_rate": 1.925685171626314e-05, "loss": 0.6448, "step": 4205 }, { "epoch": 0.14984769403423767, "grad_norm": 1.5561970746754479, "learning_rate": 1.925641512637076e-05, "loss": 0.5963, "step": 4206 }, { "epoch": 0.14988332116073178, "grad_norm": 1.4520930625974, "learning_rate": 1.925597841322275e-05, "loss": 0.5971, "step": 4207 }, { "epoch": 0.1499189482872259, "grad_norm": 1.2943367979765243, "learning_rate": 1.925554157682491e-05, "loss": 0.5504, "step": 4208 }, { "epoch": 0.14995457541372, "grad_norm": 2.099696899617957, "learning_rate": 1.9255104617183068e-05, "loss": 0.6348, "step": 4209 }, { "epoch": 0.1499902025402141, "grad_norm": 1.2279238216345334, "learning_rate": 1.9254667534303044e-05, "loss": 0.5625, "step": 4210 }, { "epoch": 0.15002582966670824, "grad_norm": 2.1335671468415516, "learning_rate": 1.925423032819065e-05, "loss": 0.5855, "step": 4211 }, { "epoch": 0.15006145679320235, "grad_norm": 1.726518787244953, "learning_rate": 1.9253792998851707e-05, "loss": 0.614, "step": 4212 }, { "epoch": 0.15009708391969645, "grad_norm": 1.8043392765315647, "learning_rate": 1.9253355546292044e-05, "loss": 0.7455, "step": 4213 }, { "epoch": 0.15013271104619058, "grad_norm": 1.734610994041747, "learning_rate": 1.9252917970517486e-05, "loss": 0.6016, "step": 4214 }, { "epoch": 0.15016833817268468, "grad_norm": 1.6974763220568363, "learning_rate": 1.9252480271533858e-05, "loss": 0.716, "step": 4215 }, { "epoch": 0.1502039652991788, "grad_norm": 1.3014820932833477, "learning_rate": 1.9252042449346987e-05, "loss": 0.5446, "step": 4216 }, { "epoch": 0.15023959242567292, "grad_norm": 1.3622233133840131, "learning_rate": 1.92516045039627e-05, "loss": 0.5638, "step": 4217 }, { "epoch": 0.15027521955216702, "grad_norm": 1.584694464387494, "learning_rate": 1.9251166435386837e-05, "loss": 0.6, "step": 4218 }, { "epoch": 0.15031084667866113, "grad_norm": 1.3705765459440142, "learning_rate": 1.925072824362523e-05, "loss": 0.5192, "step": 4219 }, { "epoch": 0.15034647380515526, "grad_norm": 1.18106827902831, "learning_rate": 1.9250289928683706e-05, "loss": 0.4638, "step": 4220 }, { "epoch": 0.15038210093164936, "grad_norm": 1.3462955812770652, "learning_rate": 1.9249851490568106e-05, "loss": 0.5423, "step": 4221 }, { "epoch": 0.15041772805814346, "grad_norm": 2.015959005364491, "learning_rate": 1.924941292928427e-05, "loss": 0.6374, "step": 4222 }, { "epoch": 0.1504533551846376, "grad_norm": 2.201828085164381, "learning_rate": 1.9248974244838035e-05, "loss": 0.6811, "step": 4223 }, { "epoch": 0.1504889823111317, "grad_norm": 1.2183169364159827, "learning_rate": 1.9248535437235242e-05, "loss": 0.5164, "step": 4224 }, { "epoch": 0.1505246094376258, "grad_norm": 1.8851467521240122, "learning_rate": 1.9248096506481737e-05, "loss": 0.5939, "step": 4225 }, { "epoch": 0.15056023656411993, "grad_norm": 2.390692169474342, "learning_rate": 1.9247657452583362e-05, "loss": 0.6546, "step": 4226 }, { "epoch": 0.15059586369061403, "grad_norm": 3.006733569755348, "learning_rate": 1.9247218275545964e-05, "loss": 0.5765, "step": 4227 }, { "epoch": 0.15063149081710814, "grad_norm": 1.4719979682160367, "learning_rate": 1.9246778975375393e-05, "loss": 0.563, "step": 4228 }, { "epoch": 0.15066711794360227, "grad_norm": 1.514222401681045, "learning_rate": 1.9246339552077497e-05, "loss": 0.6052, "step": 4229 }, { "epoch": 0.15070274507009637, "grad_norm": 1.832175713084836, "learning_rate": 1.9245900005658124e-05, "loss": 0.6401, "step": 4230 }, { "epoch": 0.15073837219659048, "grad_norm": 1.7799381502154625, "learning_rate": 1.9245460336123136e-05, "loss": 0.6828, "step": 4231 }, { "epoch": 0.1507739993230846, "grad_norm": 2.0642417385139034, "learning_rate": 1.9245020543478373e-05, "loss": 0.6458, "step": 4232 }, { "epoch": 0.1508096264495787, "grad_norm": 1.8830107465823218, "learning_rate": 1.9244580627729704e-05, "loss": 0.6574, "step": 4233 }, { "epoch": 0.1508452535760728, "grad_norm": 1.7931995606615632, "learning_rate": 1.9244140588882984e-05, "loss": 0.6597, "step": 4234 }, { "epoch": 0.15088088070256694, "grad_norm": 1.280668355837954, "learning_rate": 1.924370042694407e-05, "loss": 0.5093, "step": 4235 }, { "epoch": 0.15091650782906105, "grad_norm": 1.37848076114777, "learning_rate": 1.9243260141918822e-05, "loss": 0.5515, "step": 4236 }, { "epoch": 0.15095213495555515, "grad_norm": 1.3084058788637152, "learning_rate": 1.9242819733813106e-05, "loss": 0.5353, "step": 4237 }, { "epoch": 0.15098776208204928, "grad_norm": 1.9295402036635227, "learning_rate": 1.9242379202632785e-05, "loss": 0.6618, "step": 4238 }, { "epoch": 0.15102338920854338, "grad_norm": 1.3175001937352608, "learning_rate": 1.9241938548383723e-05, "loss": 0.5104, "step": 4239 }, { "epoch": 0.1510590163350375, "grad_norm": 1.3457998528608235, "learning_rate": 1.9241497771071792e-05, "loss": 0.5356, "step": 4240 }, { "epoch": 0.15109464346153162, "grad_norm": 1.26864550143844, "learning_rate": 1.9241056870702857e-05, "loss": 0.5615, "step": 4241 }, { "epoch": 0.15113027058802572, "grad_norm": 1.2775438647087518, "learning_rate": 1.924061584728279e-05, "loss": 0.5152, "step": 4242 }, { "epoch": 0.15116589771451983, "grad_norm": 2.2395319949386763, "learning_rate": 1.9240174700817464e-05, "loss": 0.6622, "step": 4243 }, { "epoch": 0.15120152484101396, "grad_norm": 1.3207896610265375, "learning_rate": 1.9239733431312756e-05, "loss": 0.4682, "step": 4244 }, { "epoch": 0.15123715196750806, "grad_norm": 2.136669420634468, "learning_rate": 1.9239292038774538e-05, "loss": 0.6403, "step": 4245 }, { "epoch": 0.15127277909400216, "grad_norm": 2.104491673867839, "learning_rate": 1.9238850523208687e-05, "loss": 0.6535, "step": 4246 }, { "epoch": 0.1513084062204963, "grad_norm": 1.3396865371688682, "learning_rate": 1.923840888462109e-05, "loss": 0.5456, "step": 4247 }, { "epoch": 0.1513440333469904, "grad_norm": 1.3248591779241374, "learning_rate": 1.923796712301761e-05, "loss": 0.565, "step": 4248 }, { "epoch": 0.1513796604734845, "grad_norm": 1.3986913570172677, "learning_rate": 1.923752523840415e-05, "loss": 0.5638, "step": 4249 }, { "epoch": 0.15141528759997863, "grad_norm": 2.219735353050198, "learning_rate": 1.923708323078658e-05, "loss": 0.6103, "step": 4250 }, { "epoch": 0.15145091472647274, "grad_norm": 1.4224110338114135, "learning_rate": 1.9236641100170798e-05, "loss": 0.5973, "step": 4251 }, { "epoch": 0.15148654185296684, "grad_norm": 1.98723466025817, "learning_rate": 1.9236198846562676e-05, "loss": 0.5924, "step": 4252 }, { "epoch": 0.15152216897946097, "grad_norm": 1.2272544922895712, "learning_rate": 1.9235756469968112e-05, "loss": 0.5447, "step": 4253 }, { "epoch": 0.15155779610595507, "grad_norm": 1.2250233503739654, "learning_rate": 1.9235313970392997e-05, "loss": 0.5126, "step": 4254 }, { "epoch": 0.15159342323244918, "grad_norm": 1.8064043594051014, "learning_rate": 1.923487134784322e-05, "loss": 0.6578, "step": 4255 }, { "epoch": 0.1516290503589433, "grad_norm": 1.2566676774198056, "learning_rate": 1.9234428602324678e-05, "loss": 0.5135, "step": 4256 }, { "epoch": 0.1516646774854374, "grad_norm": 1.8260469639882326, "learning_rate": 1.9233985733843262e-05, "loss": 0.6528, "step": 4257 }, { "epoch": 0.1517003046119315, "grad_norm": 1.4322946862319148, "learning_rate": 1.9233542742404872e-05, "loss": 0.5987, "step": 4258 }, { "epoch": 0.15173593173842564, "grad_norm": 1.2050271161288517, "learning_rate": 1.923309962801541e-05, "loss": 0.5127, "step": 4259 }, { "epoch": 0.15177155886491975, "grad_norm": 1.643136507576774, "learning_rate": 1.923265639068077e-05, "loss": 0.6031, "step": 4260 }, { "epoch": 0.15180718599141385, "grad_norm": 1.278026682184555, "learning_rate": 1.9232213030406857e-05, "loss": 0.5237, "step": 4261 }, { "epoch": 0.15184281311790798, "grad_norm": 1.374844827353045, "learning_rate": 1.9231769547199577e-05, "loss": 0.5384, "step": 4262 }, { "epoch": 0.15187844024440209, "grad_norm": 1.2903357238708886, "learning_rate": 1.923132594106483e-05, "loss": 0.5273, "step": 4263 }, { "epoch": 0.1519140673708962, "grad_norm": 1.6154307880874392, "learning_rate": 1.9230882212008528e-05, "loss": 0.565, "step": 4264 }, { "epoch": 0.15194969449739032, "grad_norm": 1.6058402390076651, "learning_rate": 1.9230438360036577e-05, "loss": 0.6954, "step": 4265 }, { "epoch": 0.15198532162388442, "grad_norm": 1.4689560205745493, "learning_rate": 1.9229994385154884e-05, "loss": 0.5475, "step": 4266 }, { "epoch": 0.15202094875037853, "grad_norm": 1.6826547179437625, "learning_rate": 1.922955028736937e-05, "loss": 0.5959, "step": 4267 }, { "epoch": 0.15205657587687266, "grad_norm": 1.1399909622155144, "learning_rate": 1.922910606668594e-05, "loss": 0.4962, "step": 4268 }, { "epoch": 0.15209220300336676, "grad_norm": 1.8195827282233275, "learning_rate": 1.922866172311051e-05, "loss": 0.5506, "step": 4269 }, { "epoch": 0.15212783012986086, "grad_norm": 1.7274928562519944, "learning_rate": 1.9228217256649e-05, "loss": 0.6853, "step": 4270 }, { "epoch": 0.152163457256355, "grad_norm": 1.3434969038082631, "learning_rate": 1.922777266730733e-05, "loss": 0.5146, "step": 4271 }, { "epoch": 0.1521990843828491, "grad_norm": 1.2931915281904591, "learning_rate": 1.9227327955091412e-05, "loss": 0.5596, "step": 4272 }, { "epoch": 0.1522347115093432, "grad_norm": 1.5889882204765544, "learning_rate": 1.9226883120007178e-05, "loss": 0.5808, "step": 4273 }, { "epoch": 0.15227033863583733, "grad_norm": 2.016350048135003, "learning_rate": 1.9226438162060544e-05, "loss": 0.5677, "step": 4274 }, { "epoch": 0.15230596576233144, "grad_norm": 1.3214342389756715, "learning_rate": 1.9225993081257435e-05, "loss": 0.4772, "step": 4275 }, { "epoch": 0.15234159288882554, "grad_norm": 2.804424896317604, "learning_rate": 1.9225547877603782e-05, "loss": 0.5319, "step": 4276 }, { "epoch": 0.15237722001531967, "grad_norm": 1.5577140890963441, "learning_rate": 1.922510255110551e-05, "loss": 0.5692, "step": 4277 }, { "epoch": 0.15241284714181377, "grad_norm": 1.32612387351478, "learning_rate": 1.922465710176855e-05, "loss": 0.5005, "step": 4278 }, { "epoch": 0.15244847426830788, "grad_norm": 3.581599084480133, "learning_rate": 1.922421152959883e-05, "loss": 0.6912, "step": 4279 }, { "epoch": 0.152484101394802, "grad_norm": 2.1141643208443397, "learning_rate": 1.922376583460229e-05, "loss": 0.5446, "step": 4280 }, { "epoch": 0.1525197285212961, "grad_norm": 2.3508698187147488, "learning_rate": 1.922332001678486e-05, "loss": 0.6123, "step": 4281 }, { "epoch": 0.15255535564779021, "grad_norm": 1.7014548781675676, "learning_rate": 1.9222874076152478e-05, "loss": 0.6298, "step": 4282 }, { "epoch": 0.15259098277428434, "grad_norm": 1.8137865563236388, "learning_rate": 1.9222428012711082e-05, "loss": 0.6103, "step": 4283 }, { "epoch": 0.15262660990077845, "grad_norm": 1.9604673506976582, "learning_rate": 1.9221981826466606e-05, "loss": 0.5948, "step": 4284 }, { "epoch": 0.15266223702727255, "grad_norm": 1.3829763509849373, "learning_rate": 1.9221535517425002e-05, "loss": 0.5323, "step": 4285 }, { "epoch": 0.15269786415376668, "grad_norm": 1.159247122702666, "learning_rate": 1.9221089085592203e-05, "loss": 0.487, "step": 4286 }, { "epoch": 0.15273349128026079, "grad_norm": 1.8926071848388153, "learning_rate": 1.922064253097416e-05, "loss": 0.6359, "step": 4287 }, { "epoch": 0.15276911840675492, "grad_norm": 2.5565830153298243, "learning_rate": 1.9220195853576815e-05, "loss": 0.6878, "step": 4288 }, { "epoch": 0.15280474553324902, "grad_norm": 1.8199180578157665, "learning_rate": 1.9219749053406118e-05, "loss": 0.6654, "step": 4289 }, { "epoch": 0.15284037265974312, "grad_norm": 1.333643956495342, "learning_rate": 1.9219302130468017e-05, "loss": 0.5385, "step": 4290 }, { "epoch": 0.15287599978623725, "grad_norm": 1.186805071288085, "learning_rate": 1.9218855084768466e-05, "loss": 0.5472, "step": 4291 }, { "epoch": 0.15291162691273136, "grad_norm": 1.1268474436583447, "learning_rate": 1.9218407916313417e-05, "loss": 0.4927, "step": 4292 }, { "epoch": 0.15294725403922546, "grad_norm": 1.8728525598477972, "learning_rate": 1.921796062510882e-05, "loss": 0.6266, "step": 4293 }, { "epoch": 0.1529828811657196, "grad_norm": 1.3993564919795465, "learning_rate": 1.9217513211160637e-05, "loss": 0.5424, "step": 4294 }, { "epoch": 0.1530185082922137, "grad_norm": 1.3080017854060821, "learning_rate": 1.9217065674474823e-05, "loss": 0.4679, "step": 4295 }, { "epoch": 0.1530541354187078, "grad_norm": 2.322320516595878, "learning_rate": 1.9216618015057336e-05, "loss": 0.6813, "step": 4296 }, { "epoch": 0.15308976254520193, "grad_norm": 1.6415662900543098, "learning_rate": 1.921617023291414e-05, "loss": 0.573, "step": 4297 }, { "epoch": 0.15312538967169603, "grad_norm": 1.3828674726367798, "learning_rate": 1.9215722328051195e-05, "loss": 0.6646, "step": 4298 }, { "epoch": 0.15316101679819014, "grad_norm": 1.3913752196174063, "learning_rate": 1.9215274300474463e-05, "loss": 0.6086, "step": 4299 }, { "epoch": 0.15319664392468427, "grad_norm": 1.2427820246938588, "learning_rate": 1.9214826150189915e-05, "loss": 0.5326, "step": 4300 }, { "epoch": 0.15323227105117837, "grad_norm": 1.220621599096604, "learning_rate": 1.921437787720352e-05, "loss": 0.5786, "step": 4301 }, { "epoch": 0.15326789817767247, "grad_norm": 1.8597877199870558, "learning_rate": 1.9213929481521237e-05, "loss": 0.7129, "step": 4302 }, { "epoch": 0.1533035253041666, "grad_norm": 1.263705658936549, "learning_rate": 1.9213480963149042e-05, "loss": 0.5494, "step": 4303 }, { "epoch": 0.1533391524306607, "grad_norm": 1.205791992585914, "learning_rate": 1.9213032322092913e-05, "loss": 0.5452, "step": 4304 }, { "epoch": 0.1533747795571548, "grad_norm": 1.6550982055883832, "learning_rate": 1.9212583558358816e-05, "loss": 0.6409, "step": 4305 }, { "epoch": 0.15341040668364894, "grad_norm": 2.8127560371166687, "learning_rate": 1.9212134671952733e-05, "loss": 0.6214, "step": 4306 }, { "epoch": 0.15344603381014305, "grad_norm": 1.2830449204276049, "learning_rate": 1.9211685662880636e-05, "loss": 0.5612, "step": 4307 }, { "epoch": 0.15348166093663715, "grad_norm": 1.3341047769245111, "learning_rate": 1.92112365311485e-05, "loss": 0.5345, "step": 4308 }, { "epoch": 0.15351728806313128, "grad_norm": 1.3030556533548303, "learning_rate": 1.921078727676232e-05, "loss": 0.4944, "step": 4309 }, { "epoch": 0.15355291518962538, "grad_norm": 2.110030252407842, "learning_rate": 1.9210337899728066e-05, "loss": 0.5936, "step": 4310 }, { "epoch": 0.1535885423161195, "grad_norm": 1.314566835463649, "learning_rate": 1.9209888400051727e-05, "loss": 0.5379, "step": 4311 }, { "epoch": 0.15362416944261362, "grad_norm": 1.4809507793085879, "learning_rate": 1.9209438777739283e-05, "loss": 0.6093, "step": 4312 }, { "epoch": 0.15365979656910772, "grad_norm": 1.3172646476605356, "learning_rate": 1.9208989032796728e-05, "loss": 0.5088, "step": 4313 }, { "epoch": 0.15369542369560182, "grad_norm": 1.417936105951382, "learning_rate": 1.9208539165230044e-05, "loss": 0.5635, "step": 4314 }, { "epoch": 0.15373105082209595, "grad_norm": 1.731237073246927, "learning_rate": 1.9208089175045227e-05, "loss": 0.6591, "step": 4315 }, { "epoch": 0.15376667794859006, "grad_norm": 2.1468147627853242, "learning_rate": 1.9207639062248267e-05, "loss": 0.6457, "step": 4316 }, { "epoch": 0.15380230507508416, "grad_norm": 1.905620731974873, "learning_rate": 1.9207188826845156e-05, "loss": 0.6587, "step": 4317 }, { "epoch": 0.1538379322015783, "grad_norm": 1.8925207743418122, "learning_rate": 1.920673846884189e-05, "loss": 0.6794, "step": 4318 }, { "epoch": 0.1538735593280724, "grad_norm": 1.954378867375657, "learning_rate": 1.9206287988244467e-05, "loss": 0.7151, "step": 4319 }, { "epoch": 0.1539091864545665, "grad_norm": 2.0848856921418624, "learning_rate": 1.9205837385058884e-05, "loss": 0.605, "step": 4320 }, { "epoch": 0.15394481358106063, "grad_norm": 1.721281302388163, "learning_rate": 1.9205386659291143e-05, "loss": 0.5705, "step": 4321 }, { "epoch": 0.15398044070755473, "grad_norm": 2.316579099515091, "learning_rate": 1.9204935810947245e-05, "loss": 0.5706, "step": 4322 }, { "epoch": 0.15401606783404884, "grad_norm": 1.4692305510457144, "learning_rate": 1.9204484840033187e-05, "loss": 0.5257, "step": 4323 }, { "epoch": 0.15405169496054297, "grad_norm": 1.3075117434221657, "learning_rate": 1.9204033746554987e-05, "loss": 0.5797, "step": 4324 }, { "epoch": 0.15408732208703707, "grad_norm": 1.7781844295094236, "learning_rate": 1.9203582530518644e-05, "loss": 0.6041, "step": 4325 }, { "epoch": 0.15412294921353117, "grad_norm": 2.119921045288369, "learning_rate": 1.9203131191930162e-05, "loss": 0.5815, "step": 4326 }, { "epoch": 0.1541585763400253, "grad_norm": 1.3255644583210184, "learning_rate": 1.920267973079556e-05, "loss": 0.5112, "step": 4327 }, { "epoch": 0.1541942034665194, "grad_norm": 1.3849512237333557, "learning_rate": 1.920222814712084e-05, "loss": 0.5039, "step": 4328 }, { "epoch": 0.1542298305930135, "grad_norm": 1.4442161305871304, "learning_rate": 1.9201776440912024e-05, "loss": 0.5602, "step": 4329 }, { "epoch": 0.15426545771950764, "grad_norm": 1.6464263304731934, "learning_rate": 1.9201324612175123e-05, "loss": 0.5361, "step": 4330 }, { "epoch": 0.15430108484600175, "grad_norm": 1.3884531132948517, "learning_rate": 1.920087266091615e-05, "loss": 0.477, "step": 4331 }, { "epoch": 0.15433671197249585, "grad_norm": 1.6893231653103706, "learning_rate": 1.9200420587141133e-05, "loss": 0.6599, "step": 4332 }, { "epoch": 0.15437233909898998, "grad_norm": 2.400862967941973, "learning_rate": 1.919996839085608e-05, "loss": 0.6832, "step": 4333 }, { "epoch": 0.15440796622548408, "grad_norm": 1.7963719508297196, "learning_rate": 1.919951607206702e-05, "loss": 0.7213, "step": 4334 }, { "epoch": 0.1544435933519782, "grad_norm": 2.2602819135215095, "learning_rate": 1.9199063630779973e-05, "loss": 0.6529, "step": 4335 }, { "epoch": 0.15447922047847232, "grad_norm": 1.6830360563321296, "learning_rate": 1.919861106700096e-05, "loss": 0.5522, "step": 4336 }, { "epoch": 0.15451484760496642, "grad_norm": 1.5018085718731935, "learning_rate": 1.9198158380736015e-05, "loss": 0.6605, "step": 4337 }, { "epoch": 0.15455047473146052, "grad_norm": 1.60770218370812, "learning_rate": 1.9197705571991162e-05, "loss": 0.6069, "step": 4338 }, { "epoch": 0.15458610185795466, "grad_norm": 1.5368007731644435, "learning_rate": 1.919725264077243e-05, "loss": 0.4639, "step": 4339 }, { "epoch": 0.15462172898444876, "grad_norm": 1.4532256125240472, "learning_rate": 1.919679958708585e-05, "loss": 0.5353, "step": 4340 }, { "epoch": 0.15465735611094286, "grad_norm": 3.310512029047644, "learning_rate": 1.9196346410937455e-05, "loss": 0.692, "step": 4341 }, { "epoch": 0.154692983237437, "grad_norm": 1.2404561854357696, "learning_rate": 1.9195893112333284e-05, "loss": 0.4731, "step": 4342 }, { "epoch": 0.1547286103639311, "grad_norm": 2.871143135914645, "learning_rate": 1.9195439691279363e-05, "loss": 0.6781, "step": 4343 }, { "epoch": 0.1547642374904252, "grad_norm": 1.194348435581215, "learning_rate": 1.9194986147781737e-05, "loss": 0.532, "step": 4344 }, { "epoch": 0.15479986461691933, "grad_norm": 1.3256560526415435, "learning_rate": 1.9194532481846447e-05, "loss": 0.4941, "step": 4345 }, { "epoch": 0.15483549174341343, "grad_norm": 2.520564635454961, "learning_rate": 1.9194078693479524e-05, "loss": 0.7214, "step": 4346 }, { "epoch": 0.15487111886990754, "grad_norm": 1.8404449757947396, "learning_rate": 1.9193624782687023e-05, "loss": 0.6058, "step": 4347 }, { "epoch": 0.15490674599640167, "grad_norm": 1.6583410571490502, "learning_rate": 1.919317074947498e-05, "loss": 0.6554, "step": 4348 }, { "epoch": 0.15494237312289577, "grad_norm": 1.9336573291591164, "learning_rate": 1.9192716593849443e-05, "loss": 0.6347, "step": 4349 }, { "epoch": 0.15497800024938987, "grad_norm": 1.5620584348511777, "learning_rate": 1.9192262315816457e-05, "loss": 0.5759, "step": 4350 }, { "epoch": 0.155013627375884, "grad_norm": 1.3558400742684407, "learning_rate": 1.9191807915382076e-05, "loss": 0.4972, "step": 4351 }, { "epoch": 0.1550492545023781, "grad_norm": 1.23978378927118, "learning_rate": 1.9191353392552346e-05, "loss": 0.5484, "step": 4352 }, { "epoch": 0.1550848816288722, "grad_norm": 2.0125826798496487, "learning_rate": 1.919089874733332e-05, "loss": 0.6653, "step": 4353 }, { "epoch": 0.15512050875536634, "grad_norm": 1.6726386851504884, "learning_rate": 1.9190443979731056e-05, "loss": 0.6334, "step": 4354 }, { "epoch": 0.15515613588186045, "grad_norm": 1.186303323195709, "learning_rate": 1.9189989089751605e-05, "loss": 0.5288, "step": 4355 }, { "epoch": 0.15519176300835455, "grad_norm": 1.8494446502289001, "learning_rate": 1.918953407740103e-05, "loss": 0.5875, "step": 4356 }, { "epoch": 0.15522739013484868, "grad_norm": 2.0123522704222077, "learning_rate": 1.9189078942685378e-05, "loss": 0.6238, "step": 4357 }, { "epoch": 0.15526301726134278, "grad_norm": 1.205322842870603, "learning_rate": 1.9188623685610725e-05, "loss": 0.5279, "step": 4358 }, { "epoch": 0.1552986443878369, "grad_norm": 1.2434375472553263, "learning_rate": 1.918816830618312e-05, "loss": 0.55, "step": 4359 }, { "epoch": 0.15533427151433102, "grad_norm": 2.036531940758613, "learning_rate": 1.9187712804408632e-05, "loss": 0.6218, "step": 4360 }, { "epoch": 0.15536989864082512, "grad_norm": 1.2050282307324671, "learning_rate": 1.918725718029333e-05, "loss": 0.5057, "step": 4361 }, { "epoch": 0.15540552576731922, "grad_norm": 1.8405104163382788, "learning_rate": 1.9186801433843272e-05, "loss": 0.6035, "step": 4362 }, { "epoch": 0.15544115289381336, "grad_norm": 1.187414278952704, "learning_rate": 1.918634556506454e-05, "loss": 0.5053, "step": 4363 }, { "epoch": 0.15547678002030746, "grad_norm": 1.7082044006988093, "learning_rate": 1.9185889573963185e-05, "loss": 0.6391, "step": 4364 }, { "epoch": 0.15551240714680156, "grad_norm": 1.3283067358335503, "learning_rate": 1.9185433460545296e-05, "loss": 0.5507, "step": 4365 }, { "epoch": 0.1555480342732957, "grad_norm": 1.1053310934287304, "learning_rate": 1.918497722481694e-05, "loss": 0.5212, "step": 4366 }, { "epoch": 0.1555836613997898, "grad_norm": 1.2259013705119546, "learning_rate": 1.918452086678419e-05, "loss": 0.515, "step": 4367 }, { "epoch": 0.1556192885262839, "grad_norm": 1.218379443842607, "learning_rate": 1.9184064386453127e-05, "loss": 0.5036, "step": 4368 }, { "epoch": 0.15565491565277803, "grad_norm": 1.5508304366485377, "learning_rate": 1.918360778382983e-05, "loss": 0.6361, "step": 4369 }, { "epoch": 0.15569054277927213, "grad_norm": 1.2082178479763943, "learning_rate": 1.9183151058920373e-05, "loss": 0.4682, "step": 4370 }, { "epoch": 0.15572616990576624, "grad_norm": 2.089182178109286, "learning_rate": 1.9182694211730842e-05, "loss": 0.6506, "step": 4371 }, { "epoch": 0.15576179703226037, "grad_norm": 1.2530471208791365, "learning_rate": 1.918223724226732e-05, "loss": 0.5114, "step": 4372 }, { "epoch": 0.15579742415875447, "grad_norm": 1.3213720512462732, "learning_rate": 1.9181780150535886e-05, "loss": 0.5251, "step": 4373 }, { "epoch": 0.15583305128524857, "grad_norm": 1.1270752510121178, "learning_rate": 1.9181322936542638e-05, "loss": 0.4898, "step": 4374 }, { "epoch": 0.1558686784117427, "grad_norm": 2.154865880267943, "learning_rate": 1.9180865600293655e-05, "loss": 0.664, "step": 4375 }, { "epoch": 0.1559043055382368, "grad_norm": 1.7785371220459374, "learning_rate": 1.9180408141795032e-05, "loss": 0.6971, "step": 4376 }, { "epoch": 0.1559399326647309, "grad_norm": 1.2807763603214115, "learning_rate": 1.9179950561052855e-05, "loss": 0.566, "step": 4377 }, { "epoch": 0.15597555979122504, "grad_norm": 1.191180210216974, "learning_rate": 1.917949285807322e-05, "loss": 0.5281, "step": 4378 }, { "epoch": 0.15601118691771915, "grad_norm": 1.76127394336565, "learning_rate": 1.917903503286222e-05, "loss": 0.5979, "step": 4379 }, { "epoch": 0.15604681404421328, "grad_norm": 1.4168778365877617, "learning_rate": 1.9178577085425956e-05, "loss": 0.5303, "step": 4380 }, { "epoch": 0.15608244117070738, "grad_norm": 1.3979469670564477, "learning_rate": 1.917811901577052e-05, "loss": 0.5472, "step": 4381 }, { "epoch": 0.15611806829720148, "grad_norm": 1.7277237736576998, "learning_rate": 1.9177660823902017e-05, "loss": 0.6868, "step": 4382 }, { "epoch": 0.15615369542369562, "grad_norm": 1.95566820408277, "learning_rate": 1.9177202509826545e-05, "loss": 0.6624, "step": 4383 }, { "epoch": 0.15618932255018972, "grad_norm": 1.2377812528582428, "learning_rate": 1.9176744073550204e-05, "loss": 0.5045, "step": 4384 }, { "epoch": 0.15622494967668382, "grad_norm": 1.7210731344988381, "learning_rate": 1.9176285515079102e-05, "loss": 0.6546, "step": 4385 }, { "epoch": 0.15626057680317795, "grad_norm": 1.1963894216786295, "learning_rate": 1.9175826834419347e-05, "loss": 0.5446, "step": 4386 }, { "epoch": 0.15629620392967206, "grad_norm": 1.3246183514973435, "learning_rate": 1.9175368031577045e-05, "loss": 0.5012, "step": 4387 }, { "epoch": 0.15633183105616616, "grad_norm": 1.293300532629654, "learning_rate": 1.9174909106558303e-05, "loss": 0.495, "step": 4388 }, { "epoch": 0.1563674581826603, "grad_norm": 1.2718543317710915, "learning_rate": 1.917445005936923e-05, "loss": 0.5087, "step": 4389 }, { "epoch": 0.1564030853091544, "grad_norm": 2.0703612362815482, "learning_rate": 1.9173990890015948e-05, "loss": 0.6314, "step": 4390 }, { "epoch": 0.1564387124356485, "grad_norm": 1.272302239467899, "learning_rate": 1.917353159850456e-05, "loss": 0.5067, "step": 4391 }, { "epoch": 0.15647433956214263, "grad_norm": 1.2035089307429419, "learning_rate": 1.917307218484119e-05, "loss": 0.4974, "step": 4392 }, { "epoch": 0.15650996668863673, "grad_norm": 1.7806749014481233, "learning_rate": 1.917261264903195e-05, "loss": 0.7071, "step": 4393 }, { "epoch": 0.15654559381513083, "grad_norm": 1.20996319827631, "learning_rate": 1.9172152991082963e-05, "loss": 0.5241, "step": 4394 }, { "epoch": 0.15658122094162497, "grad_norm": 1.983863220253152, "learning_rate": 1.9171693211000348e-05, "loss": 0.6508, "step": 4395 }, { "epoch": 0.15661684806811907, "grad_norm": 1.9667337116906545, "learning_rate": 1.9171233308790225e-05, "loss": 0.6196, "step": 4396 }, { "epoch": 0.15665247519461317, "grad_norm": 1.8872503841381492, "learning_rate": 1.9170773284458725e-05, "loss": 0.5846, "step": 4397 }, { "epoch": 0.1566881023211073, "grad_norm": 1.2525687512375985, "learning_rate": 1.9170313138011964e-05, "loss": 0.5519, "step": 4398 }, { "epoch": 0.1567237294476014, "grad_norm": 1.625966769372747, "learning_rate": 1.9169852869456077e-05, "loss": 0.657, "step": 4399 }, { "epoch": 0.1567593565740955, "grad_norm": 1.8298657135915337, "learning_rate": 1.916939247879719e-05, "loss": 0.6201, "step": 4400 }, { "epoch": 0.15679498370058964, "grad_norm": 1.2876478506879032, "learning_rate": 1.916893196604143e-05, "loss": 0.5527, "step": 4401 }, { "epoch": 0.15683061082708374, "grad_norm": 1.2510987161057394, "learning_rate": 1.9168471331194933e-05, "loss": 0.502, "step": 4402 }, { "epoch": 0.15686623795357785, "grad_norm": 1.2760928073099997, "learning_rate": 1.9168010574263833e-05, "loss": 0.5352, "step": 4403 }, { "epoch": 0.15690186508007198, "grad_norm": 1.2012860551180917, "learning_rate": 1.9167549695254265e-05, "loss": 0.5491, "step": 4404 }, { "epoch": 0.15693749220656608, "grad_norm": 2.7665753822824874, "learning_rate": 1.916708869417236e-05, "loss": 0.5914, "step": 4405 }, { "epoch": 0.15697311933306018, "grad_norm": 1.7863189275152083, "learning_rate": 1.9166627571024267e-05, "loss": 0.6662, "step": 4406 }, { "epoch": 0.15700874645955432, "grad_norm": 1.6147663326913195, "learning_rate": 1.916616632581612e-05, "loss": 0.5907, "step": 4407 }, { "epoch": 0.15704437358604842, "grad_norm": 1.2735977405533991, "learning_rate": 1.916570495855406e-05, "loss": 0.4862, "step": 4408 }, { "epoch": 0.15708000071254252, "grad_norm": 1.7808613767750545, "learning_rate": 1.9165243469244233e-05, "loss": 0.5839, "step": 4409 }, { "epoch": 0.15711562783903665, "grad_norm": 1.9248237704358953, "learning_rate": 1.916478185789278e-05, "loss": 0.6575, "step": 4410 }, { "epoch": 0.15715125496553076, "grad_norm": 1.4433448076834627, "learning_rate": 1.9164320124505854e-05, "loss": 0.5248, "step": 4411 }, { "epoch": 0.15718688209202486, "grad_norm": 1.331962149100042, "learning_rate": 1.91638582690896e-05, "loss": 0.54, "step": 4412 }, { "epoch": 0.157222509218519, "grad_norm": 1.4543736236333813, "learning_rate": 1.916339629165017e-05, "loss": 0.5548, "step": 4413 }, { "epoch": 0.1572581363450131, "grad_norm": 2.232978374377222, "learning_rate": 1.916293419219371e-05, "loss": 0.7297, "step": 4414 }, { "epoch": 0.1572937634715072, "grad_norm": 1.6474226619037364, "learning_rate": 1.9162471970726377e-05, "loss": 0.6241, "step": 4415 }, { "epoch": 0.15732939059800133, "grad_norm": 1.3897997966882627, "learning_rate": 1.9162009627254327e-05, "loss": 0.5202, "step": 4416 }, { "epoch": 0.15736501772449543, "grad_norm": 2.9355267728063943, "learning_rate": 1.9161547161783714e-05, "loss": 0.707, "step": 4417 }, { "epoch": 0.15740064485098954, "grad_norm": 1.2937703890928067, "learning_rate": 1.9161084574320696e-05, "loss": 0.5449, "step": 4418 }, { "epoch": 0.15743627197748367, "grad_norm": 1.695092768503539, "learning_rate": 1.916062186487144e-05, "loss": 0.625, "step": 4419 }, { "epoch": 0.15747189910397777, "grad_norm": 1.2611192940922016, "learning_rate": 1.9160159033442092e-05, "loss": 0.4949, "step": 4420 }, { "epoch": 0.15750752623047187, "grad_norm": 1.158299526929694, "learning_rate": 1.915969608003883e-05, "loss": 0.4683, "step": 4421 }, { "epoch": 0.157543153356966, "grad_norm": 1.2261867097191934, "learning_rate": 1.9159233004667807e-05, "loss": 0.5536, "step": 4422 }, { "epoch": 0.1575787804834601, "grad_norm": 1.8327906485823404, "learning_rate": 1.91587698073352e-05, "loss": 0.6468, "step": 4423 }, { "epoch": 0.1576144076099542, "grad_norm": 1.8434868748630244, "learning_rate": 1.915830648804717e-05, "loss": 0.6575, "step": 4424 }, { "epoch": 0.15765003473644834, "grad_norm": 2.2781061388576904, "learning_rate": 1.9157843046809886e-05, "loss": 0.5977, "step": 4425 }, { "epoch": 0.15768566186294244, "grad_norm": 1.9463007173389637, "learning_rate": 1.9157379483629524e-05, "loss": 0.6348, "step": 4426 }, { "epoch": 0.15772128898943655, "grad_norm": 1.5697479740896587, "learning_rate": 1.915691579851225e-05, "loss": 0.6409, "step": 4427 }, { "epoch": 0.15775691611593068, "grad_norm": 2.0079157955808946, "learning_rate": 1.9156451991464243e-05, "loss": 0.6739, "step": 4428 }, { "epoch": 0.15779254324242478, "grad_norm": 1.7672959429349988, "learning_rate": 1.915598806249168e-05, "loss": 0.6227, "step": 4429 }, { "epoch": 0.15782817036891889, "grad_norm": 1.9733832678202985, "learning_rate": 1.915552401160073e-05, "loss": 0.6946, "step": 4430 }, { "epoch": 0.15786379749541302, "grad_norm": 1.4732111335358262, "learning_rate": 1.9155059838797588e-05, "loss": 0.5645, "step": 4431 }, { "epoch": 0.15789942462190712, "grad_norm": 1.5409397442959145, "learning_rate": 1.915459554408842e-05, "loss": 0.577, "step": 4432 }, { "epoch": 0.15793505174840122, "grad_norm": 1.7267109532294935, "learning_rate": 1.9154131127479416e-05, "loss": 0.7566, "step": 4433 }, { "epoch": 0.15797067887489535, "grad_norm": 1.1295781191077146, "learning_rate": 1.9153666588976753e-05, "loss": 0.4714, "step": 4434 }, { "epoch": 0.15800630600138946, "grad_norm": 1.2917308908365186, "learning_rate": 1.9153201928586625e-05, "loss": 0.5761, "step": 4435 }, { "epoch": 0.15804193312788356, "grad_norm": 1.2978469097306267, "learning_rate": 1.9152737146315213e-05, "loss": 0.5474, "step": 4436 }, { "epoch": 0.1580775602543777, "grad_norm": 1.1864161302639387, "learning_rate": 1.915227224216871e-05, "loss": 0.5792, "step": 4437 }, { "epoch": 0.1581131873808718, "grad_norm": 1.1239839408217698, "learning_rate": 1.9151807216153305e-05, "loss": 0.5055, "step": 4438 }, { "epoch": 0.1581488145073659, "grad_norm": 1.1918887138025136, "learning_rate": 1.9151342068275188e-05, "loss": 0.4974, "step": 4439 }, { "epoch": 0.15818444163386003, "grad_norm": 1.5753016266183657, "learning_rate": 1.915087679854056e-05, "loss": 0.5992, "step": 4440 }, { "epoch": 0.15822006876035413, "grad_norm": 2.1423448963293548, "learning_rate": 1.9150411406955606e-05, "loss": 0.6999, "step": 4441 }, { "epoch": 0.15825569588684824, "grad_norm": 1.9425096014063057, "learning_rate": 1.9149945893526532e-05, "loss": 0.5229, "step": 4442 }, { "epoch": 0.15829132301334237, "grad_norm": 1.3070848829081108, "learning_rate": 1.9149480258259535e-05, "loss": 0.4847, "step": 4443 }, { "epoch": 0.15832695013983647, "grad_norm": 1.3521526705707696, "learning_rate": 1.914901450116081e-05, "loss": 0.5097, "step": 4444 }, { "epoch": 0.15836257726633057, "grad_norm": 1.3062262586882953, "learning_rate": 1.914854862223656e-05, "loss": 0.5193, "step": 4445 }, { "epoch": 0.1583982043928247, "grad_norm": 1.9772281796095066, "learning_rate": 1.9148082621492994e-05, "loss": 0.6098, "step": 4446 }, { "epoch": 0.1584338315193188, "grad_norm": 1.770044099478643, "learning_rate": 1.914761649893631e-05, "loss": 0.5936, "step": 4447 }, { "epoch": 0.1584694586458129, "grad_norm": 1.9594124739465526, "learning_rate": 1.9147150254572725e-05, "loss": 0.5971, "step": 4448 }, { "epoch": 0.15850508577230704, "grad_norm": 1.1519443943148346, "learning_rate": 1.9146683888408434e-05, "loss": 0.4685, "step": 4449 }, { "epoch": 0.15854071289880114, "grad_norm": 1.3199998335701038, "learning_rate": 1.914621740044966e-05, "loss": 0.5357, "step": 4450 }, { "epoch": 0.15857634002529525, "grad_norm": 1.275683289941817, "learning_rate": 1.9145750790702606e-05, "loss": 0.5215, "step": 4451 }, { "epoch": 0.15861196715178938, "grad_norm": 1.9080487392072358, "learning_rate": 1.9145284059173485e-05, "loss": 0.6549, "step": 4452 }, { "epoch": 0.15864759427828348, "grad_norm": 1.2037915366518022, "learning_rate": 1.9144817205868515e-05, "loss": 0.5554, "step": 4453 }, { "epoch": 0.15868322140477759, "grad_norm": 1.2353402807678688, "learning_rate": 1.9144350230793916e-05, "loss": 0.5524, "step": 4454 }, { "epoch": 0.15871884853127172, "grad_norm": 2.1257622092945865, "learning_rate": 1.91438831339559e-05, "loss": 0.5935, "step": 4455 }, { "epoch": 0.15875447565776582, "grad_norm": 1.7929261193533625, "learning_rate": 1.914341591536069e-05, "loss": 0.6026, "step": 4456 }, { "epoch": 0.15879010278425992, "grad_norm": 1.8666357246973644, "learning_rate": 1.9142948575014504e-05, "loss": 0.6793, "step": 4457 }, { "epoch": 0.15882572991075405, "grad_norm": 1.2309157178510266, "learning_rate": 1.914248111292357e-05, "loss": 0.5013, "step": 4458 }, { "epoch": 0.15886135703724816, "grad_norm": 1.6396511526516317, "learning_rate": 1.9142013529094105e-05, "loss": 0.6993, "step": 4459 }, { "epoch": 0.15889698416374226, "grad_norm": 1.6312937929841134, "learning_rate": 1.9141545823532346e-05, "loss": 0.6273, "step": 4460 }, { "epoch": 0.1589326112902364, "grad_norm": 1.9003087195447705, "learning_rate": 1.9141077996244512e-05, "loss": 0.6716, "step": 4461 }, { "epoch": 0.1589682384167305, "grad_norm": 1.1330353613529238, "learning_rate": 1.9140610047236834e-05, "loss": 0.4676, "step": 4462 }, { "epoch": 0.1590038655432246, "grad_norm": 1.6738982461005594, "learning_rate": 1.9140141976515545e-05, "loss": 0.5981, "step": 4463 }, { "epoch": 0.15903949266971873, "grad_norm": 2.0989570119856076, "learning_rate": 1.9139673784086875e-05, "loss": 0.6039, "step": 4464 }, { "epoch": 0.15907511979621283, "grad_norm": 1.4833175583635727, "learning_rate": 1.9139205469957066e-05, "loss": 0.5879, "step": 4465 }, { "epoch": 0.15911074692270694, "grad_norm": 2.16175663972466, "learning_rate": 1.9138737034132343e-05, "loss": 0.621, "step": 4466 }, { "epoch": 0.15914637404920107, "grad_norm": 1.6310073957965956, "learning_rate": 1.9138268476618953e-05, "loss": 0.634, "step": 4467 }, { "epoch": 0.15918200117569517, "grad_norm": 1.172054317635294, "learning_rate": 1.9137799797423128e-05, "loss": 0.562, "step": 4468 }, { "epoch": 0.15921762830218927, "grad_norm": 1.7552143584368507, "learning_rate": 1.913733099655111e-05, "loss": 0.622, "step": 4469 }, { "epoch": 0.1592532554286834, "grad_norm": 1.2567692702552602, "learning_rate": 1.9136862074009147e-05, "loss": 0.5598, "step": 4470 }, { "epoch": 0.1592888825551775, "grad_norm": 2.1401096453810595, "learning_rate": 1.913639302980348e-05, "loss": 0.7503, "step": 4471 }, { "epoch": 0.1593245096816716, "grad_norm": 1.7963661044605943, "learning_rate": 1.9135923863940352e-05, "loss": 0.6515, "step": 4472 }, { "epoch": 0.15936013680816574, "grad_norm": 1.9579497289240628, "learning_rate": 1.913545457642601e-05, "loss": 0.6761, "step": 4473 }, { "epoch": 0.15939576393465985, "grad_norm": 1.2316460205833502, "learning_rate": 1.9134985167266708e-05, "loss": 0.5012, "step": 4474 }, { "epoch": 0.15943139106115398, "grad_norm": 2.0307804482158756, "learning_rate": 1.913451563646869e-05, "loss": 0.6642, "step": 4475 }, { "epoch": 0.15946701818764808, "grad_norm": 1.1546967539684656, "learning_rate": 1.9134045984038216e-05, "loss": 0.5472, "step": 4476 }, { "epoch": 0.15950264531414218, "grad_norm": 1.9537060977584513, "learning_rate": 1.9133576209981535e-05, "loss": 0.566, "step": 4477 }, { "epoch": 0.15953827244063631, "grad_norm": 1.7087108973363012, "learning_rate": 1.91331063143049e-05, "loss": 0.5599, "step": 4478 }, { "epoch": 0.15957389956713042, "grad_norm": 1.7759919396858463, "learning_rate": 1.9132636297014572e-05, "loss": 0.5682, "step": 4479 }, { "epoch": 0.15960952669362452, "grad_norm": 1.5763386871408493, "learning_rate": 1.9132166158116806e-05, "loss": 0.6878, "step": 4480 }, { "epoch": 0.15964515382011865, "grad_norm": 1.7414166250460024, "learning_rate": 1.913169589761787e-05, "loss": 0.6662, "step": 4481 }, { "epoch": 0.15968078094661275, "grad_norm": 2.100092861799113, "learning_rate": 1.9131225515524017e-05, "loss": 0.6331, "step": 4482 }, { "epoch": 0.15971640807310686, "grad_norm": 1.234510549055534, "learning_rate": 1.9130755011841514e-05, "loss": 0.5717, "step": 4483 }, { "epoch": 0.159752035199601, "grad_norm": 1.8469950540259616, "learning_rate": 1.9130284386576627e-05, "loss": 0.6002, "step": 4484 }, { "epoch": 0.1597876623260951, "grad_norm": 2.1809123602914133, "learning_rate": 1.912981363973562e-05, "loss": 0.5746, "step": 4485 }, { "epoch": 0.1598232894525892, "grad_norm": 1.2173926116604752, "learning_rate": 1.9129342771324765e-05, "loss": 0.5482, "step": 4486 }, { "epoch": 0.15985891657908333, "grad_norm": 1.92690496803322, "learning_rate": 1.912887178135033e-05, "loss": 0.6956, "step": 4487 }, { "epoch": 0.15989454370557743, "grad_norm": 1.5895062553964583, "learning_rate": 1.9128400669818586e-05, "loss": 0.5755, "step": 4488 }, { "epoch": 0.15993017083207153, "grad_norm": 1.2361610391507076, "learning_rate": 1.9127929436735805e-05, "loss": 0.5426, "step": 4489 }, { "epoch": 0.15996579795856566, "grad_norm": 1.6394220951743408, "learning_rate": 1.9127458082108265e-05, "loss": 0.7123, "step": 4490 }, { "epoch": 0.16000142508505977, "grad_norm": 1.5271251707453617, "learning_rate": 1.912698660594224e-05, "loss": 0.5976, "step": 4491 }, { "epoch": 0.16003705221155387, "grad_norm": 1.4110166247739944, "learning_rate": 1.912651500824401e-05, "loss": 0.6045, "step": 4492 }, { "epoch": 0.160072679338048, "grad_norm": 1.2651609003316275, "learning_rate": 1.9126043289019853e-05, "loss": 0.4795, "step": 4493 }, { "epoch": 0.1601083064645421, "grad_norm": 1.5469250837957893, "learning_rate": 1.912557144827605e-05, "loss": 0.5981, "step": 4494 }, { "epoch": 0.1601439335910362, "grad_norm": 1.099929550690477, "learning_rate": 1.9125099486018887e-05, "loss": 0.4905, "step": 4495 }, { "epoch": 0.16017956071753034, "grad_norm": 1.9260297396993777, "learning_rate": 1.9124627402254646e-05, "loss": 0.5547, "step": 4496 }, { "epoch": 0.16021518784402444, "grad_norm": 1.4048142121955178, "learning_rate": 1.912415519698961e-05, "loss": 0.5882, "step": 4497 }, { "epoch": 0.16025081497051855, "grad_norm": 1.174282680912372, "learning_rate": 1.9123682870230075e-05, "loss": 0.4688, "step": 4498 }, { "epoch": 0.16028644209701268, "grad_norm": 1.202405491589872, "learning_rate": 1.912321042198232e-05, "loss": 0.5108, "step": 4499 }, { "epoch": 0.16032206922350678, "grad_norm": 1.0877969706057502, "learning_rate": 1.912273785225265e-05, "loss": 0.5043, "step": 4500 }, { "epoch": 0.16035769635000088, "grad_norm": 1.6278110673094197, "learning_rate": 1.9122265161047342e-05, "loss": 0.5783, "step": 4501 }, { "epoch": 0.16039332347649501, "grad_norm": 1.8460223069839439, "learning_rate": 1.91217923483727e-05, "loss": 0.6671, "step": 4502 }, { "epoch": 0.16042895060298912, "grad_norm": 1.2624715516046612, "learning_rate": 1.9121319414235017e-05, "loss": 0.5049, "step": 4503 }, { "epoch": 0.16046457772948322, "grad_norm": 1.3387056584887436, "learning_rate": 1.9120846358640588e-05, "loss": 0.5524, "step": 4504 }, { "epoch": 0.16050020485597735, "grad_norm": 1.1754408440845887, "learning_rate": 1.9120373181595716e-05, "loss": 0.5182, "step": 4505 }, { "epoch": 0.16053583198247146, "grad_norm": 1.6929181229236863, "learning_rate": 1.9119899883106702e-05, "loss": 0.6088, "step": 4506 }, { "epoch": 0.16057145910896556, "grad_norm": 1.190373176633866, "learning_rate": 1.9119426463179845e-05, "loss": 0.5176, "step": 4507 }, { "epoch": 0.1606070862354597, "grad_norm": 1.110134954030141, "learning_rate": 1.9118952921821453e-05, "loss": 0.5056, "step": 4508 }, { "epoch": 0.1606427133619538, "grad_norm": 2.0673401162461666, "learning_rate": 1.9118479259037826e-05, "loss": 0.5661, "step": 4509 }, { "epoch": 0.1606783404884479, "grad_norm": 2.515971815226948, "learning_rate": 1.9118005474835275e-05, "loss": 0.6321, "step": 4510 }, { "epoch": 0.16071396761494203, "grad_norm": 1.2217824550179228, "learning_rate": 1.9117531569220106e-05, "loss": 0.5431, "step": 4511 }, { "epoch": 0.16074959474143613, "grad_norm": 1.2703593550100478, "learning_rate": 1.9117057542198637e-05, "loss": 0.5125, "step": 4512 }, { "epoch": 0.16078522186793023, "grad_norm": 1.6716885873154725, "learning_rate": 1.911658339377717e-05, "loss": 0.5892, "step": 4513 }, { "epoch": 0.16082084899442436, "grad_norm": 2.0449358310340298, "learning_rate": 1.9116109123962024e-05, "loss": 0.6673, "step": 4514 }, { "epoch": 0.16085647612091847, "grad_norm": 1.529557852036601, "learning_rate": 1.9115634732759512e-05, "loss": 0.5528, "step": 4515 }, { "epoch": 0.16089210324741257, "grad_norm": 1.1850093486593007, "learning_rate": 1.9115160220175952e-05, "loss": 0.5814, "step": 4516 }, { "epoch": 0.1609277303739067, "grad_norm": 1.1842331483360944, "learning_rate": 1.9114685586217666e-05, "loss": 0.5532, "step": 4517 }, { "epoch": 0.1609633575004008, "grad_norm": 1.1670639499083322, "learning_rate": 1.911421083089097e-05, "loss": 0.4994, "step": 4518 }, { "epoch": 0.1609989846268949, "grad_norm": 1.8069821164713455, "learning_rate": 1.9113735954202184e-05, "loss": 0.6385, "step": 4519 }, { "epoch": 0.16103461175338904, "grad_norm": 1.9073510660457536, "learning_rate": 1.9113260956157634e-05, "loss": 0.581, "step": 4520 }, { "epoch": 0.16107023887988314, "grad_norm": 1.5616367149555481, "learning_rate": 1.9112785836763645e-05, "loss": 0.594, "step": 4521 }, { "epoch": 0.16110586600637725, "grad_norm": 1.6592291500817227, "learning_rate": 1.9112310596026545e-05, "loss": 0.6566, "step": 4522 }, { "epoch": 0.16114149313287138, "grad_norm": 1.1757880203762519, "learning_rate": 1.911183523395266e-05, "loss": 0.5071, "step": 4523 }, { "epoch": 0.16117712025936548, "grad_norm": 1.2455212279774104, "learning_rate": 1.9111359750548314e-05, "loss": 0.5547, "step": 4524 }, { "epoch": 0.16121274738585958, "grad_norm": 1.2430590582075332, "learning_rate": 1.9110884145819852e-05, "loss": 0.5024, "step": 4525 }, { "epoch": 0.16124837451235371, "grad_norm": 1.1954593808540093, "learning_rate": 1.9110408419773595e-05, "loss": 0.5335, "step": 4526 }, { "epoch": 0.16128400163884782, "grad_norm": 1.70599889862863, "learning_rate": 1.9109932572415886e-05, "loss": 0.6547, "step": 4527 }, { "epoch": 0.16131962876534192, "grad_norm": 1.1677954683735838, "learning_rate": 1.910945660375305e-05, "loss": 0.5376, "step": 4528 }, { "epoch": 0.16135525589183605, "grad_norm": 1.1286279152077994, "learning_rate": 1.910898051379144e-05, "loss": 0.5657, "step": 4529 }, { "epoch": 0.16139088301833016, "grad_norm": 1.144362041945165, "learning_rate": 1.9108504302537386e-05, "loss": 0.5077, "step": 4530 }, { "epoch": 0.16142651014482426, "grad_norm": 1.2976599038428254, "learning_rate": 1.910802796999723e-05, "loss": 0.5694, "step": 4531 }, { "epoch": 0.1614621372713184, "grad_norm": 1.1333714942569726, "learning_rate": 1.9107551516177313e-05, "loss": 0.4852, "step": 4532 }, { "epoch": 0.1614977643978125, "grad_norm": 2.2116166539102635, "learning_rate": 1.9107074941083987e-05, "loss": 0.5783, "step": 4533 }, { "epoch": 0.1615333915243066, "grad_norm": 1.141851572904511, "learning_rate": 1.910659824472359e-05, "loss": 0.5933, "step": 4534 }, { "epoch": 0.16156901865080073, "grad_norm": 1.635902364248032, "learning_rate": 1.910612142710247e-05, "loss": 0.6615, "step": 4535 }, { "epoch": 0.16160464577729483, "grad_norm": 1.0798621036447307, "learning_rate": 1.9105644488226978e-05, "loss": 0.4781, "step": 4536 }, { "epoch": 0.16164027290378893, "grad_norm": 1.1571638464316523, "learning_rate": 1.910516742810347e-05, "loss": 0.5176, "step": 4537 }, { "epoch": 0.16167590003028307, "grad_norm": 1.3205536023815314, "learning_rate": 1.9104690246738288e-05, "loss": 0.5449, "step": 4538 }, { "epoch": 0.16171152715677717, "grad_norm": 1.3239030991719605, "learning_rate": 1.9104212944137796e-05, "loss": 0.5458, "step": 4539 }, { "epoch": 0.16174715428327127, "grad_norm": 1.172518610183418, "learning_rate": 1.9103735520308345e-05, "loss": 0.542, "step": 4540 }, { "epoch": 0.1617827814097654, "grad_norm": 1.7776728612334665, "learning_rate": 1.910325797525629e-05, "loss": 0.6748, "step": 4541 }, { "epoch": 0.1618184085362595, "grad_norm": 2.0573894491775966, "learning_rate": 1.910278030898799e-05, "loss": 0.6081, "step": 4542 }, { "epoch": 0.1618540356627536, "grad_norm": 1.2651336122715773, "learning_rate": 1.9102302521509814e-05, "loss": 0.5643, "step": 4543 }, { "epoch": 0.16188966278924774, "grad_norm": 1.5261897791944226, "learning_rate": 1.9101824612828113e-05, "loss": 0.618, "step": 4544 }, { "epoch": 0.16192528991574184, "grad_norm": 1.3515937983523851, "learning_rate": 1.910134658294926e-05, "loss": 0.5651, "step": 4545 }, { "epoch": 0.16196091704223595, "grad_norm": 1.790777490842301, "learning_rate": 1.910086843187961e-05, "loss": 0.6087, "step": 4546 }, { "epoch": 0.16199654416873008, "grad_norm": 1.2297245708534812, "learning_rate": 1.910039015962554e-05, "loss": 0.5168, "step": 4547 }, { "epoch": 0.16203217129522418, "grad_norm": 1.2045006003801209, "learning_rate": 1.9099911766193414e-05, "loss": 0.5146, "step": 4548 }, { "epoch": 0.16206779842171828, "grad_norm": 1.5085675709987443, "learning_rate": 1.90994332515896e-05, "loss": 0.7013, "step": 4549 }, { "epoch": 0.16210342554821242, "grad_norm": 1.2029894384989015, "learning_rate": 1.9098954615820477e-05, "loss": 0.5155, "step": 4550 }, { "epoch": 0.16213905267470652, "grad_norm": 1.8676325879549767, "learning_rate": 1.909847585889241e-05, "loss": 0.6312, "step": 4551 }, { "epoch": 0.16217467980120062, "grad_norm": 1.166958344228037, "learning_rate": 1.9097996980811774e-05, "loss": 0.4829, "step": 4552 }, { "epoch": 0.16221030692769475, "grad_norm": 1.2233180502710592, "learning_rate": 1.9097517981584955e-05, "loss": 0.5137, "step": 4553 }, { "epoch": 0.16224593405418886, "grad_norm": 1.8833154896319992, "learning_rate": 1.909703886121832e-05, "loss": 0.5611, "step": 4554 }, { "epoch": 0.16228156118068296, "grad_norm": 1.8083827023520127, "learning_rate": 1.9096559619718256e-05, "loss": 0.6172, "step": 4555 }, { "epoch": 0.1623171883071771, "grad_norm": 1.0428587050428526, "learning_rate": 1.909608025709115e-05, "loss": 0.4222, "step": 4556 }, { "epoch": 0.1623528154336712, "grad_norm": 1.3465111312423854, "learning_rate": 1.909560077334337e-05, "loss": 0.5276, "step": 4557 }, { "epoch": 0.1623884425601653, "grad_norm": 1.6876890929466168, "learning_rate": 1.9095121168481308e-05, "loss": 0.6139, "step": 4558 }, { "epoch": 0.16242406968665943, "grad_norm": 1.311144444830105, "learning_rate": 1.909464144251135e-05, "loss": 0.5609, "step": 4559 }, { "epoch": 0.16245969681315353, "grad_norm": 1.7243927693951338, "learning_rate": 1.9094161595439887e-05, "loss": 0.5889, "step": 4560 }, { "epoch": 0.16249532393964763, "grad_norm": 2.5943335915029766, "learning_rate": 1.9093681627273306e-05, "loss": 0.5472, "step": 4561 }, { "epoch": 0.16253095106614177, "grad_norm": 1.7857211869941807, "learning_rate": 1.9093201538018e-05, "loss": 0.6298, "step": 4562 }, { "epoch": 0.16256657819263587, "grad_norm": 1.7653452101423013, "learning_rate": 1.9092721327680356e-05, "loss": 0.6075, "step": 4563 }, { "epoch": 0.16260220531912997, "grad_norm": 2.3488112914638597, "learning_rate": 1.9092240996266772e-05, "loss": 0.6062, "step": 4564 }, { "epoch": 0.1626378324456241, "grad_norm": 1.3734022412094211, "learning_rate": 1.9091760543783646e-05, "loss": 0.499, "step": 4565 }, { "epoch": 0.1626734595721182, "grad_norm": 1.233851020721285, "learning_rate": 1.9091279970237373e-05, "loss": 0.5394, "step": 4566 }, { "epoch": 0.1627090866986123, "grad_norm": 1.1127646175189576, "learning_rate": 1.9090799275634354e-05, "loss": 0.5003, "step": 4567 }, { "epoch": 0.16274471382510644, "grad_norm": 2.338878343107524, "learning_rate": 1.9090318459980986e-05, "loss": 0.6955, "step": 4568 }, { "epoch": 0.16278034095160054, "grad_norm": 1.7613419051176415, "learning_rate": 1.9089837523283673e-05, "loss": 0.5265, "step": 4569 }, { "epoch": 0.16281596807809467, "grad_norm": 1.8201613873954572, "learning_rate": 1.9089356465548824e-05, "loss": 0.6553, "step": 4570 }, { "epoch": 0.16285159520458878, "grad_norm": 1.2113424058276967, "learning_rate": 1.9088875286782836e-05, "loss": 0.5516, "step": 4571 }, { "epoch": 0.16288722233108288, "grad_norm": 1.805215103166201, "learning_rate": 1.9088393986992124e-05, "loss": 0.5516, "step": 4572 }, { "epoch": 0.162922849457577, "grad_norm": 1.254694225076401, "learning_rate": 1.9087912566183094e-05, "loss": 0.5893, "step": 4573 }, { "epoch": 0.16295847658407112, "grad_norm": 1.1725830294457236, "learning_rate": 1.9087431024362156e-05, "loss": 0.5272, "step": 4574 }, { "epoch": 0.16299410371056522, "grad_norm": 1.1727043617111506, "learning_rate": 1.908694936153572e-05, "loss": 0.4802, "step": 4575 }, { "epoch": 0.16302973083705935, "grad_norm": 1.1449214117797455, "learning_rate": 1.9086467577710203e-05, "loss": 0.5379, "step": 4576 }, { "epoch": 0.16306535796355345, "grad_norm": 2.2554579173686324, "learning_rate": 1.908598567289202e-05, "loss": 0.6433, "step": 4577 }, { "epoch": 0.16310098509004756, "grad_norm": 1.1812238620716207, "learning_rate": 1.9085503647087588e-05, "loss": 0.4791, "step": 4578 }, { "epoch": 0.1631366122165417, "grad_norm": 1.2124975343398936, "learning_rate": 1.9085021500303322e-05, "loss": 0.5275, "step": 4579 }, { "epoch": 0.1631722393430358, "grad_norm": 1.4941632612866682, "learning_rate": 1.9084539232545645e-05, "loss": 0.5804, "step": 4580 }, { "epoch": 0.1632078664695299, "grad_norm": 2.005223809292249, "learning_rate": 1.908405684382098e-05, "loss": 0.5852, "step": 4581 }, { "epoch": 0.16324349359602403, "grad_norm": 1.9209190495719315, "learning_rate": 1.9083574334135745e-05, "loss": 0.6213, "step": 4582 }, { "epoch": 0.16327912072251813, "grad_norm": 1.1799330196168503, "learning_rate": 1.9083091703496373e-05, "loss": 0.5166, "step": 4583 }, { "epoch": 0.16331474784901223, "grad_norm": 1.25364220870605, "learning_rate": 1.9082608951909285e-05, "loss": 0.546, "step": 4584 }, { "epoch": 0.16335037497550636, "grad_norm": 1.2729194255546634, "learning_rate": 1.9082126079380908e-05, "loss": 0.5168, "step": 4585 }, { "epoch": 0.16338600210200047, "grad_norm": 1.7615500058235958, "learning_rate": 1.908164308591768e-05, "loss": 0.6256, "step": 4586 }, { "epoch": 0.16342162922849457, "grad_norm": 1.5895535337374598, "learning_rate": 1.9081159971526022e-05, "loss": 0.5953, "step": 4587 }, { "epoch": 0.1634572563549887, "grad_norm": 1.3051398364546296, "learning_rate": 1.908067673621237e-05, "loss": 0.5511, "step": 4588 }, { "epoch": 0.1634928834814828, "grad_norm": 2.982004687127888, "learning_rate": 1.9080193379983166e-05, "loss": 0.6736, "step": 4589 }, { "epoch": 0.1635285106079769, "grad_norm": 1.5731820872947235, "learning_rate": 1.9079709902844837e-05, "loss": 0.6126, "step": 4590 }, { "epoch": 0.16356413773447104, "grad_norm": 1.2160647138805112, "learning_rate": 1.9079226304803823e-05, "loss": 0.4757, "step": 4591 }, { "epoch": 0.16359976486096514, "grad_norm": 1.9861525639204798, "learning_rate": 1.9078742585866566e-05, "loss": 0.5141, "step": 4592 }, { "epoch": 0.16363539198745924, "grad_norm": 1.2604547253563245, "learning_rate": 1.907825874603951e-05, "loss": 0.5019, "step": 4593 }, { "epoch": 0.16367101911395338, "grad_norm": 1.6029312061774472, "learning_rate": 1.907777478532909e-05, "loss": 0.5925, "step": 4594 }, { "epoch": 0.16370664624044748, "grad_norm": 1.1900471263297832, "learning_rate": 1.9077290703741753e-05, "loss": 0.5123, "step": 4595 }, { "epoch": 0.16374227336694158, "grad_norm": 1.7399104918051087, "learning_rate": 1.9076806501283944e-05, "loss": 0.6712, "step": 4596 }, { "epoch": 0.1637779004934357, "grad_norm": 1.9547557102066069, "learning_rate": 1.9076322177962116e-05, "loss": 0.5475, "step": 4597 }, { "epoch": 0.16381352761992982, "grad_norm": 1.1480145135130924, "learning_rate": 1.907583773378271e-05, "loss": 0.5786, "step": 4598 }, { "epoch": 0.16384915474642392, "grad_norm": 1.2336135487808357, "learning_rate": 1.9075353168752184e-05, "loss": 0.473, "step": 4599 }, { "epoch": 0.16388478187291805, "grad_norm": 1.2376986247171555, "learning_rate": 1.9074868482876988e-05, "loss": 0.4899, "step": 4600 }, { "epoch": 0.16392040899941215, "grad_norm": 1.1209647812910914, "learning_rate": 1.9074383676163574e-05, "loss": 0.5148, "step": 4601 }, { "epoch": 0.16395603612590626, "grad_norm": 1.319883467894948, "learning_rate": 1.9073898748618396e-05, "loss": 0.5267, "step": 4602 }, { "epoch": 0.1639916632524004, "grad_norm": 1.1601977506183483, "learning_rate": 1.907341370024792e-05, "loss": 0.5109, "step": 4603 }, { "epoch": 0.1640272903788945, "grad_norm": 2.2699188883276187, "learning_rate": 1.9072928531058594e-05, "loss": 0.6218, "step": 4604 }, { "epoch": 0.1640629175053886, "grad_norm": 1.6614161854946072, "learning_rate": 1.9072443241056884e-05, "loss": 0.6366, "step": 4605 }, { "epoch": 0.16409854463188273, "grad_norm": 1.9957156324202736, "learning_rate": 1.9071957830249252e-05, "loss": 0.5828, "step": 4606 }, { "epoch": 0.16413417175837683, "grad_norm": 1.331391152922452, "learning_rate": 1.907147229864216e-05, "loss": 0.5181, "step": 4607 }, { "epoch": 0.16416979888487093, "grad_norm": 1.2784645889747734, "learning_rate": 1.9070986646242072e-05, "loss": 0.5433, "step": 4608 }, { "epoch": 0.16420542601136506, "grad_norm": 1.9870340889460611, "learning_rate": 1.907050087305546e-05, "loss": 0.6012, "step": 4609 }, { "epoch": 0.16424105313785917, "grad_norm": 1.329690476769323, "learning_rate": 1.9070014979088786e-05, "loss": 0.5677, "step": 4610 }, { "epoch": 0.16427668026435327, "grad_norm": 2.3021909482708467, "learning_rate": 1.9069528964348522e-05, "loss": 0.6349, "step": 4611 }, { "epoch": 0.1643123073908474, "grad_norm": 1.9404009721457227, "learning_rate": 1.9069042828841144e-05, "loss": 0.5851, "step": 4612 }, { "epoch": 0.1643479345173415, "grad_norm": 1.856316516883301, "learning_rate": 1.906855657257312e-05, "loss": 0.5723, "step": 4613 }, { "epoch": 0.1643835616438356, "grad_norm": 2.268803436268835, "learning_rate": 1.9068070195550928e-05, "loss": 0.6241, "step": 4614 }, { "epoch": 0.16441918877032974, "grad_norm": 3.370969187761064, "learning_rate": 1.9067583697781043e-05, "loss": 0.6074, "step": 4615 }, { "epoch": 0.16445481589682384, "grad_norm": 1.803384101861854, "learning_rate": 1.9067097079269942e-05, "loss": 0.627, "step": 4616 }, { "epoch": 0.16449044302331794, "grad_norm": 1.5108751882755385, "learning_rate": 1.9066610340024106e-05, "loss": 0.5589, "step": 4617 }, { "epoch": 0.16452607014981208, "grad_norm": 1.9012685056058292, "learning_rate": 1.9066123480050017e-05, "loss": 0.5723, "step": 4618 }, { "epoch": 0.16456169727630618, "grad_norm": 2.8181114372768437, "learning_rate": 1.9065636499354157e-05, "loss": 0.6263, "step": 4619 }, { "epoch": 0.16459732440280028, "grad_norm": 1.807726913931101, "learning_rate": 1.906514939794301e-05, "loss": 0.5187, "step": 4620 }, { "epoch": 0.1646329515292944, "grad_norm": 1.4986147544597321, "learning_rate": 1.906466217582306e-05, "loss": 0.6164, "step": 4621 }, { "epoch": 0.16466857865578852, "grad_norm": 1.6356801119236801, "learning_rate": 1.90641748330008e-05, "loss": 0.5842, "step": 4622 }, { "epoch": 0.16470420578228262, "grad_norm": 1.2249004779023047, "learning_rate": 1.906368736948272e-05, "loss": 0.5698, "step": 4623 }, { "epoch": 0.16473983290877675, "grad_norm": 1.9397540357234735, "learning_rate": 1.9063199785275302e-05, "loss": 0.6895, "step": 4624 }, { "epoch": 0.16477546003527085, "grad_norm": 1.9417127606616225, "learning_rate": 1.9062712080385045e-05, "loss": 0.7043, "step": 4625 }, { "epoch": 0.16481108716176496, "grad_norm": 1.5241158467642757, "learning_rate": 1.9062224254818445e-05, "loss": 0.6187, "step": 4626 }, { "epoch": 0.1648467142882591, "grad_norm": 1.1722688951391587, "learning_rate": 1.9061736308581996e-05, "loss": 0.5192, "step": 4627 }, { "epoch": 0.1648823414147532, "grad_norm": 1.424309626626546, "learning_rate": 1.9061248241682187e-05, "loss": 0.6233, "step": 4628 }, { "epoch": 0.1649179685412473, "grad_norm": 1.2633413714816357, "learning_rate": 1.906076005412553e-05, "loss": 0.5935, "step": 4629 }, { "epoch": 0.16495359566774143, "grad_norm": 1.2560750712565067, "learning_rate": 1.906027174591852e-05, "loss": 0.5203, "step": 4630 }, { "epoch": 0.16498922279423553, "grad_norm": 1.2381105272664648, "learning_rate": 1.9059783317067658e-05, "loss": 0.5471, "step": 4631 }, { "epoch": 0.16502484992072963, "grad_norm": 1.9682503169813816, "learning_rate": 1.9059294767579447e-05, "loss": 0.7018, "step": 4632 }, { "epoch": 0.16506047704722376, "grad_norm": 1.254629053461198, "learning_rate": 1.9058806097460395e-05, "loss": 0.5448, "step": 4633 }, { "epoch": 0.16509610417371787, "grad_norm": 1.8108429418092207, "learning_rate": 1.905831730671701e-05, "loss": 0.6557, "step": 4634 }, { "epoch": 0.16513173130021197, "grad_norm": 1.532238323725556, "learning_rate": 1.90578283953558e-05, "loss": 0.7174, "step": 4635 }, { "epoch": 0.1651673584267061, "grad_norm": 1.6550054508433807, "learning_rate": 1.9057339363383268e-05, "loss": 0.6226, "step": 4636 }, { "epoch": 0.1652029855532002, "grad_norm": 2.2457230682092972, "learning_rate": 1.9056850210805935e-05, "loss": 0.6253, "step": 4637 }, { "epoch": 0.1652386126796943, "grad_norm": 2.2124524192749533, "learning_rate": 1.905636093763031e-05, "loss": 0.5821, "step": 4638 }, { "epoch": 0.16527423980618844, "grad_norm": 2.3168911905058183, "learning_rate": 1.905587154386291e-05, "loss": 0.6136, "step": 4639 }, { "epoch": 0.16530986693268254, "grad_norm": 1.4093864552552589, "learning_rate": 1.905538202951025e-05, "loss": 0.5321, "step": 4640 }, { "epoch": 0.16534549405917665, "grad_norm": 1.3224414446471708, "learning_rate": 1.905489239457885e-05, "loss": 0.5809, "step": 4641 }, { "epoch": 0.16538112118567078, "grad_norm": 1.7198264377279617, "learning_rate": 1.9054402639075226e-05, "loss": 0.6491, "step": 4642 }, { "epoch": 0.16541674831216488, "grad_norm": 1.1913373362316368, "learning_rate": 1.9053912763005907e-05, "loss": 0.4759, "step": 4643 }, { "epoch": 0.16545237543865898, "grad_norm": 1.8115255048348076, "learning_rate": 1.905342276637741e-05, "loss": 0.5911, "step": 4644 }, { "epoch": 0.16548800256515311, "grad_norm": 2.2020042839564855, "learning_rate": 1.905293264919626e-05, "loss": 0.6468, "step": 4645 }, { "epoch": 0.16552362969164722, "grad_norm": 1.9479568481064353, "learning_rate": 1.9052442411468984e-05, "loss": 0.6281, "step": 4646 }, { "epoch": 0.16555925681814132, "grad_norm": 1.3964791868526218, "learning_rate": 1.9051952053202107e-05, "loss": 0.5606, "step": 4647 }, { "epoch": 0.16559488394463545, "grad_norm": 4.263950574115031, "learning_rate": 1.9051461574402165e-05, "loss": 0.6261, "step": 4648 }, { "epoch": 0.16563051107112955, "grad_norm": 2.0970353012546172, "learning_rate": 1.9050970975075685e-05, "loss": 0.5779, "step": 4649 }, { "epoch": 0.16566613819762366, "grad_norm": 1.2428672671888779, "learning_rate": 1.90504802552292e-05, "loss": 0.5095, "step": 4650 }, { "epoch": 0.1657017653241178, "grad_norm": 1.257892618106835, "learning_rate": 1.9049989414869245e-05, "loss": 0.5558, "step": 4651 }, { "epoch": 0.1657373924506119, "grad_norm": 1.3428408099969802, "learning_rate": 1.9049498454002354e-05, "loss": 0.4954, "step": 4652 }, { "epoch": 0.165773019577106, "grad_norm": 2.60381211112289, "learning_rate": 1.9049007372635068e-05, "loss": 0.6328, "step": 4653 }, { "epoch": 0.16580864670360013, "grad_norm": 2.6862559057017226, "learning_rate": 1.9048516170773924e-05, "loss": 0.6664, "step": 4654 }, { "epoch": 0.16584427383009423, "grad_norm": 1.1680617705032244, "learning_rate": 1.904802484842546e-05, "loss": 0.4802, "step": 4655 }, { "epoch": 0.16587990095658833, "grad_norm": 1.1881540822981311, "learning_rate": 1.9047533405596225e-05, "loss": 0.4761, "step": 4656 }, { "epoch": 0.16591552808308246, "grad_norm": 1.5214286791178915, "learning_rate": 1.9047041842292752e-05, "loss": 0.6079, "step": 4657 }, { "epoch": 0.16595115520957657, "grad_norm": 1.3932639429575977, "learning_rate": 1.90465501585216e-05, "loss": 0.5555, "step": 4658 }, { "epoch": 0.16598678233607067, "grad_norm": 2.2340928729525182, "learning_rate": 1.9046058354289307e-05, "loss": 0.6089, "step": 4659 }, { "epoch": 0.1660224094625648, "grad_norm": 1.543114341487036, "learning_rate": 1.9045566429602425e-05, "loss": 0.5193, "step": 4660 }, { "epoch": 0.1660580365890589, "grad_norm": 1.389988396439979, "learning_rate": 1.90450743844675e-05, "loss": 0.5018, "step": 4661 }, { "epoch": 0.166093663715553, "grad_norm": 1.38091767469365, "learning_rate": 1.904458221889109e-05, "loss": 0.5291, "step": 4662 }, { "epoch": 0.16612929084204714, "grad_norm": 1.3221599998324323, "learning_rate": 1.904408993287975e-05, "loss": 0.5078, "step": 4663 }, { "epoch": 0.16616491796854124, "grad_norm": 2.1390364606510492, "learning_rate": 1.9043597526440027e-05, "loss": 0.6648, "step": 4664 }, { "epoch": 0.16620054509503537, "grad_norm": 1.2994746229955745, "learning_rate": 1.904310499957848e-05, "loss": 0.5411, "step": 4665 }, { "epoch": 0.16623617222152948, "grad_norm": 1.7760580446317684, "learning_rate": 1.904261235230167e-05, "loss": 0.6237, "step": 4666 }, { "epoch": 0.16627179934802358, "grad_norm": 1.1786952157240513, "learning_rate": 1.9042119584616157e-05, "loss": 0.5597, "step": 4667 }, { "epoch": 0.1663074264745177, "grad_norm": 1.3443861879616916, "learning_rate": 1.9041626696528503e-05, "loss": 0.5018, "step": 4668 }, { "epoch": 0.16634305360101181, "grad_norm": 1.914779676927586, "learning_rate": 1.904113368804527e-05, "loss": 0.6453, "step": 4669 }, { "epoch": 0.16637868072750592, "grad_norm": 1.9508309890378386, "learning_rate": 1.904064055917302e-05, "loss": 0.6833, "step": 4670 }, { "epoch": 0.16641430785400005, "grad_norm": 1.8717467063916964, "learning_rate": 1.9040147309918326e-05, "loss": 0.6332, "step": 4671 }, { "epoch": 0.16644993498049415, "grad_norm": 1.2541106693345028, "learning_rate": 1.903965394028775e-05, "loss": 0.5225, "step": 4672 }, { "epoch": 0.16648556210698826, "grad_norm": 2.1336086945017017, "learning_rate": 1.903916045028786e-05, "loss": 0.6027, "step": 4673 }, { "epoch": 0.1665211892334824, "grad_norm": 1.4524082290390592, "learning_rate": 1.9038666839925234e-05, "loss": 0.5106, "step": 4674 }, { "epoch": 0.1665568163599765, "grad_norm": 1.262749391343364, "learning_rate": 1.9038173109206443e-05, "loss": 0.4716, "step": 4675 }, { "epoch": 0.1665924434864706, "grad_norm": 1.563608529279172, "learning_rate": 1.9037679258138056e-05, "loss": 0.5977, "step": 4676 }, { "epoch": 0.16662807061296472, "grad_norm": 1.3122280063229177, "learning_rate": 1.9037185286726653e-05, "loss": 0.4699, "step": 4677 }, { "epoch": 0.16666369773945883, "grad_norm": 1.3166975228543256, "learning_rate": 1.9036691194978815e-05, "loss": 0.496, "step": 4678 }, { "epoch": 0.16669932486595293, "grad_norm": 1.3381612742201112, "learning_rate": 1.9036196982901113e-05, "loss": 0.5633, "step": 4679 }, { "epoch": 0.16673495199244706, "grad_norm": 1.77090218811871, "learning_rate": 1.9035702650500135e-05, "loss": 0.6573, "step": 4680 }, { "epoch": 0.16677057911894116, "grad_norm": 1.1653986563412042, "learning_rate": 1.9035208197782456e-05, "loss": 0.523, "step": 4681 }, { "epoch": 0.16680620624543527, "grad_norm": 1.3659485437506853, "learning_rate": 1.903471362475467e-05, "loss": 0.5441, "step": 4682 }, { "epoch": 0.1668418333719294, "grad_norm": 2.1163493381675345, "learning_rate": 1.9034218931423356e-05, "loss": 0.6139, "step": 4683 }, { "epoch": 0.1668774604984235, "grad_norm": 1.4458048580280758, "learning_rate": 1.9033724117795103e-05, "loss": 0.5644, "step": 4684 }, { "epoch": 0.1669130876249176, "grad_norm": 1.2504139192506487, "learning_rate": 1.90332291838765e-05, "loss": 0.574, "step": 4685 }, { "epoch": 0.16694871475141174, "grad_norm": 2.134656607388088, "learning_rate": 1.9032734129674132e-05, "loss": 0.6866, "step": 4686 }, { "epoch": 0.16698434187790584, "grad_norm": 2.6374047331984483, "learning_rate": 1.90322389551946e-05, "loss": 0.6442, "step": 4687 }, { "epoch": 0.16701996900439994, "grad_norm": 2.5757320006617737, "learning_rate": 1.903174366044449e-05, "loss": 0.6836, "step": 4688 }, { "epoch": 0.16705559613089407, "grad_norm": 1.4253115632632665, "learning_rate": 1.90312482454304e-05, "loss": 0.515, "step": 4689 }, { "epoch": 0.16709122325738818, "grad_norm": 1.3758024495298893, "learning_rate": 1.9030752710158934e-05, "loss": 0.5794, "step": 4690 }, { "epoch": 0.16712685038388228, "grad_norm": 1.3410501199516807, "learning_rate": 1.9030257054636677e-05, "loss": 0.5345, "step": 4691 }, { "epoch": 0.1671624775103764, "grad_norm": 2.306305635497717, "learning_rate": 1.9029761278870235e-05, "loss": 0.6138, "step": 4692 }, { "epoch": 0.16719810463687051, "grad_norm": 1.2505250768418688, "learning_rate": 1.9029265382866216e-05, "loss": 0.5623, "step": 4693 }, { "epoch": 0.16723373176336462, "grad_norm": 2.1100128185357785, "learning_rate": 1.902876936663121e-05, "loss": 0.6919, "step": 4694 }, { "epoch": 0.16726935888985875, "grad_norm": 1.6469003179151862, "learning_rate": 1.9028273230171833e-05, "loss": 0.4765, "step": 4695 }, { "epoch": 0.16730498601635285, "grad_norm": 1.8443876449552805, "learning_rate": 1.902777697349469e-05, "loss": 0.5399, "step": 4696 }, { "epoch": 0.16734061314284696, "grad_norm": 2.3767914528982255, "learning_rate": 1.902728059660638e-05, "loss": 0.6936, "step": 4697 }, { "epoch": 0.1673762402693411, "grad_norm": 2.092685719316296, "learning_rate": 1.9026784099513525e-05, "loss": 0.6298, "step": 4698 }, { "epoch": 0.1674118673958352, "grad_norm": 1.259991890093332, "learning_rate": 1.9026287482222727e-05, "loss": 0.5293, "step": 4699 }, { "epoch": 0.1674474945223293, "grad_norm": 1.2020470328006962, "learning_rate": 1.9025790744740606e-05, "loss": 0.4733, "step": 4700 }, { "epoch": 0.16748312164882342, "grad_norm": 2.2103085915082197, "learning_rate": 1.902529388707377e-05, "loss": 0.6078, "step": 4701 }, { "epoch": 0.16751874877531753, "grad_norm": 2.7522629898900437, "learning_rate": 1.9024796909228835e-05, "loss": 0.7118, "step": 4702 }, { "epoch": 0.16755437590181163, "grad_norm": 1.4313225047062463, "learning_rate": 1.9024299811212426e-05, "loss": 0.5029, "step": 4703 }, { "epoch": 0.16759000302830576, "grad_norm": 1.1488941025483812, "learning_rate": 1.9023802593031156e-05, "loss": 0.516, "step": 4704 }, { "epoch": 0.16762563015479987, "grad_norm": 1.636158337338499, "learning_rate": 1.9023305254691645e-05, "loss": 0.6876, "step": 4705 }, { "epoch": 0.16766125728129397, "grad_norm": 1.9168808484910953, "learning_rate": 1.902280779620052e-05, "loss": 0.6261, "step": 4706 }, { "epoch": 0.1676968844077881, "grad_norm": 1.279176520328019, "learning_rate": 1.90223102175644e-05, "loss": 0.5327, "step": 4707 }, { "epoch": 0.1677325115342822, "grad_norm": 1.7701406589500552, "learning_rate": 1.9021812518789917e-05, "loss": 0.5749, "step": 4708 }, { "epoch": 0.1677681386607763, "grad_norm": 2.181430283947941, "learning_rate": 1.9021314699883692e-05, "loss": 0.6875, "step": 4709 }, { "epoch": 0.16780376578727044, "grad_norm": 1.2702977130327606, "learning_rate": 1.902081676085236e-05, "loss": 0.4864, "step": 4710 }, { "epoch": 0.16783939291376454, "grad_norm": 1.165994425746499, "learning_rate": 1.902031870170254e-05, "loss": 0.5361, "step": 4711 }, { "epoch": 0.16787502004025864, "grad_norm": 1.7426602432835474, "learning_rate": 1.9019820522440877e-05, "loss": 0.5361, "step": 4712 }, { "epoch": 0.16791064716675277, "grad_norm": 1.994456117765953, "learning_rate": 1.9019322223073997e-05, "loss": 0.6135, "step": 4713 }, { "epoch": 0.16794627429324688, "grad_norm": 3.2700702159137576, "learning_rate": 1.901882380360854e-05, "loss": 0.6902, "step": 4714 }, { "epoch": 0.16798190141974098, "grad_norm": 1.718528810623801, "learning_rate": 1.901832526405114e-05, "loss": 0.7158, "step": 4715 }, { "epoch": 0.1680175285462351, "grad_norm": 1.4528311636554045, "learning_rate": 1.9017826604408434e-05, "loss": 0.4374, "step": 4716 }, { "epoch": 0.16805315567272922, "grad_norm": 1.2909531383833888, "learning_rate": 1.9017327824687065e-05, "loss": 0.5239, "step": 4717 }, { "epoch": 0.16808878279922332, "grad_norm": 1.596195157633055, "learning_rate": 1.901682892489367e-05, "loss": 0.535, "step": 4718 }, { "epoch": 0.16812440992571745, "grad_norm": 1.8246447442434048, "learning_rate": 1.90163299050349e-05, "loss": 0.5536, "step": 4719 }, { "epoch": 0.16816003705221155, "grad_norm": 1.2465147653665136, "learning_rate": 1.9015830765117394e-05, "loss": 0.5222, "step": 4720 }, { "epoch": 0.16819566417870566, "grad_norm": 1.7109699429630623, "learning_rate": 1.90153315051478e-05, "loss": 0.6531, "step": 4721 }, { "epoch": 0.1682312913051998, "grad_norm": 1.7123263528884354, "learning_rate": 1.9014832125132764e-05, "loss": 0.5661, "step": 4722 }, { "epoch": 0.1682669184316939, "grad_norm": 1.2376752388915004, "learning_rate": 1.9014332625078937e-05, "loss": 0.4652, "step": 4723 }, { "epoch": 0.168302545558188, "grad_norm": 1.2318016844746567, "learning_rate": 1.9013833004992972e-05, "loss": 0.5619, "step": 4724 }, { "epoch": 0.16833817268468212, "grad_norm": 1.0714792967557867, "learning_rate": 1.901333326488152e-05, "loss": 0.4922, "step": 4725 }, { "epoch": 0.16837379981117623, "grad_norm": 2.200935065160131, "learning_rate": 1.9012833404751237e-05, "loss": 0.5924, "step": 4726 }, { "epoch": 0.16840942693767033, "grad_norm": 1.6884512551611275, "learning_rate": 1.9012333424608775e-05, "loss": 0.5955, "step": 4727 }, { "epoch": 0.16844505406416446, "grad_norm": 2.002783748567862, "learning_rate": 1.9011833324460792e-05, "loss": 0.5412, "step": 4728 }, { "epoch": 0.16848068119065857, "grad_norm": 1.138775301603969, "learning_rate": 1.9011333104313956e-05, "loss": 0.4866, "step": 4729 }, { "epoch": 0.16851630831715267, "grad_norm": 1.7077193517935194, "learning_rate": 1.9010832764174916e-05, "loss": 0.585, "step": 4730 }, { "epoch": 0.1685519354436468, "grad_norm": 1.2387435420045452, "learning_rate": 1.9010332304050344e-05, "loss": 0.4747, "step": 4731 }, { "epoch": 0.1685875625701409, "grad_norm": 1.2361296278910088, "learning_rate": 1.9009831723946894e-05, "loss": 0.5727, "step": 4732 }, { "epoch": 0.168623189696635, "grad_norm": 1.2668848098938414, "learning_rate": 1.900933102387124e-05, "loss": 0.5734, "step": 4733 }, { "epoch": 0.16865881682312914, "grad_norm": 1.2816367448102304, "learning_rate": 1.9008830203830048e-05, "loss": 0.5819, "step": 4734 }, { "epoch": 0.16869444394962324, "grad_norm": 1.7745926975006043, "learning_rate": 1.9008329263829982e-05, "loss": 0.6583, "step": 4735 }, { "epoch": 0.16873007107611734, "grad_norm": 1.6391135446714975, "learning_rate": 1.9007828203877718e-05, "loss": 0.6485, "step": 4736 }, { "epoch": 0.16876569820261147, "grad_norm": 1.8702829322420063, "learning_rate": 1.9007327023979924e-05, "loss": 0.6383, "step": 4737 }, { "epoch": 0.16880132532910558, "grad_norm": 1.5360496324568393, "learning_rate": 1.9006825724143274e-05, "loss": 0.6444, "step": 4738 }, { "epoch": 0.16883695245559968, "grad_norm": 1.189033775041139, "learning_rate": 1.9006324304374442e-05, "loss": 0.4864, "step": 4739 }, { "epoch": 0.1688725795820938, "grad_norm": 1.2236181263629586, "learning_rate": 1.9005822764680112e-05, "loss": 0.5041, "step": 4740 }, { "epoch": 0.16890820670858792, "grad_norm": 1.5646827960235108, "learning_rate": 1.9005321105066956e-05, "loss": 0.5823, "step": 4741 }, { "epoch": 0.16894383383508202, "grad_norm": 1.193244784530457, "learning_rate": 1.9004819325541653e-05, "loss": 0.5064, "step": 4742 }, { "epoch": 0.16897946096157615, "grad_norm": 1.6774423734419743, "learning_rate": 1.9004317426110888e-05, "loss": 0.6586, "step": 4743 }, { "epoch": 0.16901508808807025, "grad_norm": 1.5897259775158679, "learning_rate": 1.9003815406781342e-05, "loss": 0.6001, "step": 4744 }, { "epoch": 0.16905071521456436, "grad_norm": 1.3144308478834337, "learning_rate": 1.9003313267559704e-05, "loss": 0.4997, "step": 4745 }, { "epoch": 0.1690863423410585, "grad_norm": 1.3656571580896575, "learning_rate": 1.9002811008452654e-05, "loss": 0.5321, "step": 4746 }, { "epoch": 0.1691219694675526, "grad_norm": 1.7914802028672487, "learning_rate": 1.900230862946688e-05, "loss": 0.6144, "step": 4747 }, { "epoch": 0.1691575965940467, "grad_norm": 1.361223048293876, "learning_rate": 1.900180613060908e-05, "loss": 0.5438, "step": 4748 }, { "epoch": 0.16919322372054083, "grad_norm": 3.4735672858785227, "learning_rate": 1.900130351188594e-05, "loss": 0.66, "step": 4749 }, { "epoch": 0.16922885084703493, "grad_norm": 2.523253463764378, "learning_rate": 1.9000800773304146e-05, "loss": 0.6098, "step": 4750 }, { "epoch": 0.16926447797352903, "grad_norm": 1.2150379629169137, "learning_rate": 1.90002979148704e-05, "loss": 0.4917, "step": 4751 }, { "epoch": 0.16930010510002316, "grad_norm": 1.6457275694540534, "learning_rate": 1.8999794936591402e-05, "loss": 0.661, "step": 4752 }, { "epoch": 0.16933573222651727, "grad_norm": 1.224507801342171, "learning_rate": 1.899929183847384e-05, "loss": 0.5483, "step": 4753 }, { "epoch": 0.16937135935301137, "grad_norm": 1.1690419610555176, "learning_rate": 1.8998788620524413e-05, "loss": 0.5385, "step": 4754 }, { "epoch": 0.1694069864795055, "grad_norm": 1.3476885465556583, "learning_rate": 1.8998285282749833e-05, "loss": 0.4942, "step": 4755 }, { "epoch": 0.1694426136059996, "grad_norm": 1.185227300419007, "learning_rate": 1.899778182515679e-05, "loss": 0.5879, "step": 4756 }, { "epoch": 0.16947824073249373, "grad_norm": 2.1337400626844487, "learning_rate": 1.8997278247751992e-05, "loss": 0.7117, "step": 4757 }, { "epoch": 0.16951386785898784, "grad_norm": 1.1985173639044404, "learning_rate": 1.899677455054215e-05, "loss": 0.5004, "step": 4758 }, { "epoch": 0.16954949498548194, "grad_norm": 1.8864358376745987, "learning_rate": 1.8996270733533965e-05, "loss": 0.6809, "step": 4759 }, { "epoch": 0.16958512211197607, "grad_norm": 1.4207982449473584, "learning_rate": 1.8995766796734148e-05, "loss": 0.5324, "step": 4760 }, { "epoch": 0.16962074923847018, "grad_norm": 1.292654237269404, "learning_rate": 1.8995262740149407e-05, "loss": 0.5867, "step": 4761 }, { "epoch": 0.16965637636496428, "grad_norm": 1.05166384953221, "learning_rate": 1.8994758563786456e-05, "loss": 0.494, "step": 4762 }, { "epoch": 0.1696920034914584, "grad_norm": 1.19961389121927, "learning_rate": 1.8994254267652006e-05, "loss": 0.505, "step": 4763 }, { "epoch": 0.1697276306179525, "grad_norm": 1.9866313504786786, "learning_rate": 1.8993749851752776e-05, "loss": 0.6297, "step": 4764 }, { "epoch": 0.16976325774444662, "grad_norm": 2.0057714325494755, "learning_rate": 1.899324531609548e-05, "loss": 0.6967, "step": 4765 }, { "epoch": 0.16979888487094075, "grad_norm": 2.3085440448012444, "learning_rate": 1.899274066068684e-05, "loss": 0.6407, "step": 4766 }, { "epoch": 0.16983451199743485, "grad_norm": 1.2640968389726792, "learning_rate": 1.8992235885533566e-05, "loss": 0.4848, "step": 4767 }, { "epoch": 0.16987013912392895, "grad_norm": 1.8847586425937606, "learning_rate": 1.8991730990642388e-05, "loss": 0.5521, "step": 4768 }, { "epoch": 0.16990576625042308, "grad_norm": 2.1404764911865235, "learning_rate": 1.8991225976020026e-05, "loss": 0.5785, "step": 4769 }, { "epoch": 0.1699413933769172, "grad_norm": 2.8545374287333813, "learning_rate": 1.8990720841673206e-05, "loss": 0.6451, "step": 4770 }, { "epoch": 0.1699770205034113, "grad_norm": 1.4501990364058324, "learning_rate": 1.8990215587608657e-05, "loss": 0.5535, "step": 4771 }, { "epoch": 0.17001264762990542, "grad_norm": 1.8022764772254725, "learning_rate": 1.8989710213833104e-05, "loss": 0.6309, "step": 4772 }, { "epoch": 0.17004827475639953, "grad_norm": 1.8590423604812043, "learning_rate": 1.8989204720353273e-05, "loss": 0.5541, "step": 4773 }, { "epoch": 0.17008390188289363, "grad_norm": 1.22684362407528, "learning_rate": 1.89886991071759e-05, "loss": 0.5478, "step": 4774 }, { "epoch": 0.17011952900938776, "grad_norm": 1.3300432550294679, "learning_rate": 1.8988193374307715e-05, "loss": 0.5408, "step": 4775 }, { "epoch": 0.17015515613588186, "grad_norm": 2.1226024757844666, "learning_rate": 1.8987687521755452e-05, "loss": 0.6158, "step": 4776 }, { "epoch": 0.17019078326237597, "grad_norm": 2.1503949932662905, "learning_rate": 1.898718154952585e-05, "loss": 0.5983, "step": 4777 }, { "epoch": 0.1702264103888701, "grad_norm": 1.2529668603289532, "learning_rate": 1.898667545762564e-05, "loss": 0.5348, "step": 4778 }, { "epoch": 0.1702620375153642, "grad_norm": 1.953205264868667, "learning_rate": 1.8986169246061567e-05, "loss": 0.6723, "step": 4779 }, { "epoch": 0.1702976646418583, "grad_norm": 1.3050627943880337, "learning_rate": 1.898566291484037e-05, "loss": 0.4986, "step": 4780 }, { "epoch": 0.17033329176835244, "grad_norm": 1.8508396718709719, "learning_rate": 1.898515646396879e-05, "loss": 0.6608, "step": 4781 }, { "epoch": 0.17036891889484654, "grad_norm": 2.1160597696642545, "learning_rate": 1.8984649893453573e-05, "loss": 0.524, "step": 4782 }, { "epoch": 0.17040454602134064, "grad_norm": 1.996491221102918, "learning_rate": 1.898414320330146e-05, "loss": 0.6785, "step": 4783 }, { "epoch": 0.17044017314783477, "grad_norm": 1.8849628675485852, "learning_rate": 1.8983636393519202e-05, "loss": 0.6267, "step": 4784 }, { "epoch": 0.17047580027432888, "grad_norm": 1.3764510019559748, "learning_rate": 1.898312946411355e-05, "loss": 0.4942, "step": 4785 }, { "epoch": 0.17051142740082298, "grad_norm": 1.24596555291985, "learning_rate": 1.8982622415091244e-05, "loss": 0.5155, "step": 4786 }, { "epoch": 0.1705470545273171, "grad_norm": 2.1821711669104267, "learning_rate": 1.8982115246459046e-05, "loss": 0.6112, "step": 4787 }, { "epoch": 0.1705826816538112, "grad_norm": 1.4102234202997879, "learning_rate": 1.8981607958223705e-05, "loss": 0.5492, "step": 4788 }, { "epoch": 0.17061830878030532, "grad_norm": 1.2305608443108305, "learning_rate": 1.8981100550391974e-05, "loss": 0.5189, "step": 4789 }, { "epoch": 0.17065393590679945, "grad_norm": 1.5143241369366418, "learning_rate": 1.8980593022970615e-05, "loss": 0.559, "step": 4790 }, { "epoch": 0.17068956303329355, "grad_norm": 3.664441573450843, "learning_rate": 1.8980085375966382e-05, "loss": 0.6462, "step": 4791 }, { "epoch": 0.17072519015978765, "grad_norm": 1.1510206179571925, "learning_rate": 1.8979577609386033e-05, "loss": 0.5352, "step": 4792 }, { "epoch": 0.17076081728628179, "grad_norm": 1.1509830254221531, "learning_rate": 1.8979069723236335e-05, "loss": 0.512, "step": 4793 }, { "epoch": 0.1707964444127759, "grad_norm": 1.2253434021703375, "learning_rate": 1.8978561717524044e-05, "loss": 0.4871, "step": 4794 }, { "epoch": 0.17083207153927, "grad_norm": 1.1355560694610043, "learning_rate": 1.8978053592255935e-05, "loss": 0.5304, "step": 4795 }, { "epoch": 0.17086769866576412, "grad_norm": 1.3764279743381642, "learning_rate": 1.8977545347438758e-05, "loss": 0.5224, "step": 4796 }, { "epoch": 0.17090332579225823, "grad_norm": 1.303399402178599, "learning_rate": 1.8977036983079296e-05, "loss": 0.4885, "step": 4797 }, { "epoch": 0.17093895291875233, "grad_norm": 2.667944509582036, "learning_rate": 1.897652849918431e-05, "loss": 0.657, "step": 4798 }, { "epoch": 0.17097458004524646, "grad_norm": 1.2651066208461739, "learning_rate": 1.8976019895760572e-05, "loss": 0.5317, "step": 4799 }, { "epoch": 0.17101020717174056, "grad_norm": 1.4091941859321984, "learning_rate": 1.897551117281486e-05, "loss": 0.5511, "step": 4800 }, { "epoch": 0.17104583429823467, "grad_norm": 1.364407193174834, "learning_rate": 1.8975002330353938e-05, "loss": 0.5827, "step": 4801 }, { "epoch": 0.1710814614247288, "grad_norm": 1.1342331110552113, "learning_rate": 1.8974493368384587e-05, "loss": 0.4787, "step": 4802 }, { "epoch": 0.1711170885512229, "grad_norm": 1.8928201983021304, "learning_rate": 1.8973984286913584e-05, "loss": 0.5889, "step": 4803 }, { "epoch": 0.171152715677717, "grad_norm": 2.5297149952397997, "learning_rate": 1.8973475085947714e-05, "loss": 0.6539, "step": 4804 }, { "epoch": 0.17118834280421114, "grad_norm": 2.7902077571398025, "learning_rate": 1.897296576549374e-05, "loss": 0.7208, "step": 4805 }, { "epoch": 0.17122396993070524, "grad_norm": 1.2435968096881067, "learning_rate": 1.8972456325558463e-05, "loss": 0.5557, "step": 4806 }, { "epoch": 0.17125959705719934, "grad_norm": 1.2696302484376867, "learning_rate": 1.897194676614866e-05, "loss": 0.5883, "step": 4807 }, { "epoch": 0.17129522418369347, "grad_norm": 2.0631284537964576, "learning_rate": 1.8971437087271113e-05, "loss": 0.5073, "step": 4808 }, { "epoch": 0.17133085131018758, "grad_norm": 1.2120395397683472, "learning_rate": 1.8970927288932607e-05, "loss": 0.5483, "step": 4809 }, { "epoch": 0.17136647843668168, "grad_norm": 1.438621110318484, "learning_rate": 1.8970417371139934e-05, "loss": 0.6237, "step": 4810 }, { "epoch": 0.1714021055631758, "grad_norm": 2.223767245391675, "learning_rate": 1.8969907333899885e-05, "loss": 0.6668, "step": 4811 }, { "epoch": 0.1714377326896699, "grad_norm": 1.413098272607865, "learning_rate": 1.8969397177219252e-05, "loss": 0.5013, "step": 4812 }, { "epoch": 0.17147335981616402, "grad_norm": 1.2736197472074238, "learning_rate": 1.896888690110483e-05, "loss": 0.558, "step": 4813 }, { "epoch": 0.17150898694265815, "grad_norm": 1.8401859203226953, "learning_rate": 1.8968376505563402e-05, "loss": 0.6772, "step": 4814 }, { "epoch": 0.17154461406915225, "grad_norm": 1.9234938221388145, "learning_rate": 1.896786599060178e-05, "loss": 0.6469, "step": 4815 }, { "epoch": 0.17158024119564635, "grad_norm": 1.2759354078785403, "learning_rate": 1.8967355356226747e-05, "loss": 0.4945, "step": 4816 }, { "epoch": 0.17161586832214049, "grad_norm": 1.2762767863015165, "learning_rate": 1.8966844602445112e-05, "loss": 0.4942, "step": 4817 }, { "epoch": 0.1716514954486346, "grad_norm": 1.812180251209877, "learning_rate": 1.8966333729263674e-05, "loss": 0.6095, "step": 4818 }, { "epoch": 0.1716871225751287, "grad_norm": 1.9481397535695846, "learning_rate": 1.8965822736689238e-05, "loss": 0.591, "step": 4819 }, { "epoch": 0.17172274970162282, "grad_norm": 1.374892900095084, "learning_rate": 1.89653116247286e-05, "loss": 0.523, "step": 4820 }, { "epoch": 0.17175837682811693, "grad_norm": 1.1786996327451997, "learning_rate": 1.8964800393388577e-05, "loss": 0.5182, "step": 4821 }, { "epoch": 0.17179400395461103, "grad_norm": 1.2700786823267538, "learning_rate": 1.8964289042675967e-05, "loss": 0.5835, "step": 4822 }, { "epoch": 0.17182963108110516, "grad_norm": 1.368495703596872, "learning_rate": 1.8963777572597583e-05, "loss": 0.5408, "step": 4823 }, { "epoch": 0.17186525820759926, "grad_norm": 4.170953922795035, "learning_rate": 1.896326598316024e-05, "loss": 0.5909, "step": 4824 }, { "epoch": 0.17190088533409337, "grad_norm": 1.710610062534917, "learning_rate": 1.896275427437074e-05, "loss": 0.6159, "step": 4825 }, { "epoch": 0.1719365124605875, "grad_norm": 1.7939904820517232, "learning_rate": 1.8962242446235904e-05, "loss": 0.6021, "step": 4826 }, { "epoch": 0.1719721395870816, "grad_norm": 1.2584878346111694, "learning_rate": 1.8961730498762544e-05, "loss": 0.4948, "step": 4827 }, { "epoch": 0.1720077667135757, "grad_norm": 2.176754492707027, "learning_rate": 1.896121843195748e-05, "loss": 0.6022, "step": 4828 }, { "epoch": 0.17204339384006984, "grad_norm": 1.3563397295507764, "learning_rate": 1.896070624582753e-05, "loss": 0.5611, "step": 4829 }, { "epoch": 0.17207902096656394, "grad_norm": 1.2043578281877956, "learning_rate": 1.896019394037951e-05, "loss": 0.4979, "step": 4830 }, { "epoch": 0.17211464809305804, "grad_norm": 1.1757872861462786, "learning_rate": 1.895968151562025e-05, "loss": 0.5285, "step": 4831 }, { "epoch": 0.17215027521955217, "grad_norm": 1.7354031854612735, "learning_rate": 1.8959168971556564e-05, "loss": 0.5315, "step": 4832 }, { "epoch": 0.17218590234604628, "grad_norm": 2.195423949097724, "learning_rate": 1.8958656308195284e-05, "loss": 0.6806, "step": 4833 }, { "epoch": 0.17222152947254038, "grad_norm": 2.954825648839896, "learning_rate": 1.895814352554323e-05, "loss": 0.6827, "step": 4834 }, { "epoch": 0.1722571565990345, "grad_norm": 2.0828613688645548, "learning_rate": 1.8957630623607236e-05, "loss": 0.6765, "step": 4835 }, { "epoch": 0.17229278372552861, "grad_norm": 1.704020406696391, "learning_rate": 1.895711760239413e-05, "loss": 0.5486, "step": 4836 }, { "epoch": 0.17232841085202272, "grad_norm": 1.8144182793499157, "learning_rate": 1.895660446191074e-05, "loss": 0.5967, "step": 4837 }, { "epoch": 0.17236403797851685, "grad_norm": 1.4921052959115988, "learning_rate": 1.8956091202163903e-05, "loss": 0.5934, "step": 4838 }, { "epoch": 0.17239966510501095, "grad_norm": 1.847386951675971, "learning_rate": 1.895557782316045e-05, "loss": 0.6751, "step": 4839 }, { "epoch": 0.17243529223150506, "grad_norm": 1.3022287385803517, "learning_rate": 1.895506432490722e-05, "loss": 0.5001, "step": 4840 }, { "epoch": 0.1724709193579992, "grad_norm": 1.1401953337887694, "learning_rate": 1.895455070741105e-05, "loss": 0.4962, "step": 4841 }, { "epoch": 0.1725065464844933, "grad_norm": 1.7583717025006391, "learning_rate": 1.895403697067878e-05, "loss": 0.6603, "step": 4842 }, { "epoch": 0.1725421736109874, "grad_norm": 1.832142023038341, "learning_rate": 1.8953523114717247e-05, "loss": 0.6114, "step": 4843 }, { "epoch": 0.17257780073748152, "grad_norm": 2.275758049238087, "learning_rate": 1.8953009139533297e-05, "loss": 0.6478, "step": 4844 }, { "epoch": 0.17261342786397563, "grad_norm": 1.2558664488898406, "learning_rate": 1.895249504513377e-05, "loss": 0.5474, "step": 4845 }, { "epoch": 0.17264905499046973, "grad_norm": 1.178068025089184, "learning_rate": 1.8951980831525515e-05, "loss": 0.4895, "step": 4846 }, { "epoch": 0.17268468211696386, "grad_norm": 2.725199745353643, "learning_rate": 1.8951466498715378e-05, "loss": 0.7074, "step": 4847 }, { "epoch": 0.17272030924345796, "grad_norm": 1.7727389202861858, "learning_rate": 1.895095204671021e-05, "loss": 0.659, "step": 4848 }, { "epoch": 0.17275593636995207, "grad_norm": 1.1043129306706296, "learning_rate": 1.8950437475516855e-05, "loss": 0.5058, "step": 4849 }, { "epoch": 0.1727915634964462, "grad_norm": 1.2712167572050153, "learning_rate": 1.8949922785142174e-05, "loss": 0.5792, "step": 4850 }, { "epoch": 0.1728271906229403, "grad_norm": 1.2383217070603147, "learning_rate": 1.894940797559301e-05, "loss": 0.5315, "step": 4851 }, { "epoch": 0.17286281774943443, "grad_norm": 1.9778322414879723, "learning_rate": 1.8948893046876228e-05, "loss": 0.6582, "step": 4852 }, { "epoch": 0.17289844487592854, "grad_norm": 1.668946793998661, "learning_rate": 1.8948377998998678e-05, "loss": 0.6779, "step": 4853 }, { "epoch": 0.17293407200242264, "grad_norm": 1.7286232342388756, "learning_rate": 1.894786283196722e-05, "loss": 0.5544, "step": 4854 }, { "epoch": 0.17296969912891677, "grad_norm": 1.2934929522084226, "learning_rate": 1.8947347545788718e-05, "loss": 0.5243, "step": 4855 }, { "epoch": 0.17300532625541087, "grad_norm": 2.5850595655116004, "learning_rate": 1.8946832140470027e-05, "loss": 0.61, "step": 4856 }, { "epoch": 0.17304095338190498, "grad_norm": 1.6281148691676004, "learning_rate": 1.894631661601801e-05, "loss": 0.659, "step": 4857 }, { "epoch": 0.1730765805083991, "grad_norm": 1.5245202042788906, "learning_rate": 1.894580097243954e-05, "loss": 0.6232, "step": 4858 }, { "epoch": 0.1731122076348932, "grad_norm": 1.2108787025929846, "learning_rate": 1.8945285209741474e-05, "loss": 0.5395, "step": 4859 }, { "epoch": 0.17314783476138731, "grad_norm": 1.8545488914328192, "learning_rate": 1.894476932793068e-05, "loss": 0.7107, "step": 4860 }, { "epoch": 0.17318346188788145, "grad_norm": 1.2106544862335697, "learning_rate": 1.8944253327014036e-05, "loss": 0.5088, "step": 4861 }, { "epoch": 0.17321908901437555, "grad_norm": 1.7237464259850324, "learning_rate": 1.8943737206998404e-05, "loss": 0.577, "step": 4862 }, { "epoch": 0.17325471614086965, "grad_norm": 1.6514382746817522, "learning_rate": 1.8943220967890657e-05, "loss": 0.5867, "step": 4863 }, { "epoch": 0.17329034326736378, "grad_norm": 1.7417651604202833, "learning_rate": 1.8942704609697678e-05, "loss": 0.5345, "step": 4864 }, { "epoch": 0.1733259703938579, "grad_norm": 1.242883534418543, "learning_rate": 1.8942188132426334e-05, "loss": 0.5377, "step": 4865 }, { "epoch": 0.173361597520352, "grad_norm": 1.7225518987458506, "learning_rate": 1.8941671536083506e-05, "loss": 0.6197, "step": 4866 }, { "epoch": 0.17339722464684612, "grad_norm": 2.42345630834779, "learning_rate": 1.894115482067607e-05, "loss": 0.6503, "step": 4867 }, { "epoch": 0.17343285177334022, "grad_norm": 2.0192116842986714, "learning_rate": 1.8940637986210907e-05, "loss": 0.5533, "step": 4868 }, { "epoch": 0.17346847889983433, "grad_norm": 1.1568308686005409, "learning_rate": 1.8940121032694902e-05, "loss": 0.4768, "step": 4869 }, { "epoch": 0.17350410602632846, "grad_norm": 1.227685475151784, "learning_rate": 1.8939603960134935e-05, "loss": 0.5268, "step": 4870 }, { "epoch": 0.17353973315282256, "grad_norm": 1.1660611115572, "learning_rate": 1.8939086768537894e-05, "loss": 0.5648, "step": 4871 }, { "epoch": 0.17357536027931667, "grad_norm": 1.8198622609190591, "learning_rate": 1.8938569457910664e-05, "loss": 0.6455, "step": 4872 }, { "epoch": 0.1736109874058108, "grad_norm": 1.8146761563354123, "learning_rate": 1.8938052028260136e-05, "loss": 0.614, "step": 4873 }, { "epoch": 0.1736466145323049, "grad_norm": 1.1687688487096555, "learning_rate": 1.8937534479593195e-05, "loss": 0.533, "step": 4874 }, { "epoch": 0.173682241658799, "grad_norm": 2.3266626757178286, "learning_rate": 1.893701681191674e-05, "loss": 0.6061, "step": 4875 }, { "epoch": 0.17371786878529313, "grad_norm": 1.1772579922969162, "learning_rate": 1.8936499025237656e-05, "loss": 0.5287, "step": 4876 }, { "epoch": 0.17375349591178724, "grad_norm": 1.1779856553607524, "learning_rate": 1.8935981119562842e-05, "loss": 0.5115, "step": 4877 }, { "epoch": 0.17378912303828134, "grad_norm": 1.2553998350330227, "learning_rate": 1.8935463094899194e-05, "loss": 0.5481, "step": 4878 }, { "epoch": 0.17382475016477547, "grad_norm": 2.0133884514202167, "learning_rate": 1.8934944951253607e-05, "loss": 0.5474, "step": 4879 }, { "epoch": 0.17386037729126957, "grad_norm": 1.4001658210985275, "learning_rate": 1.8934426688632986e-05, "loss": 0.5569, "step": 4880 }, { "epoch": 0.17389600441776368, "grad_norm": 1.8012081973448835, "learning_rate": 1.8933908307044223e-05, "loss": 0.6625, "step": 4881 }, { "epoch": 0.1739316315442578, "grad_norm": 1.8784972350636697, "learning_rate": 1.8933389806494232e-05, "loss": 0.6521, "step": 4882 }, { "epoch": 0.1739672586707519, "grad_norm": 2.7462091849564034, "learning_rate": 1.8932871186989913e-05, "loss": 0.6295, "step": 4883 }, { "epoch": 0.17400288579724602, "grad_norm": 1.1744295952738957, "learning_rate": 1.8932352448538166e-05, "loss": 0.5384, "step": 4884 }, { "epoch": 0.17403851292374015, "grad_norm": 1.2992167883250982, "learning_rate": 1.8931833591145904e-05, "loss": 0.5052, "step": 4885 }, { "epoch": 0.17407414005023425, "grad_norm": 1.894322957567276, "learning_rate": 1.8931314614820038e-05, "loss": 0.5712, "step": 4886 }, { "epoch": 0.17410976717672835, "grad_norm": 2.032562825225962, "learning_rate": 1.893079551956747e-05, "loss": 0.6273, "step": 4887 }, { "epoch": 0.17414539430322248, "grad_norm": 1.3968632253816518, "learning_rate": 1.8930276305395122e-05, "loss": 0.5831, "step": 4888 }, { "epoch": 0.1741810214297166, "grad_norm": 2.1847989887452406, "learning_rate": 1.8929756972309902e-05, "loss": 0.5929, "step": 4889 }, { "epoch": 0.1742166485562107, "grad_norm": 1.1590205422124797, "learning_rate": 1.8929237520318725e-05, "loss": 0.5271, "step": 4890 }, { "epoch": 0.17425227568270482, "grad_norm": 2.648109221230312, "learning_rate": 1.8928717949428508e-05, "loss": 0.5556, "step": 4891 }, { "epoch": 0.17428790280919892, "grad_norm": 2.1747115793242235, "learning_rate": 1.8928198259646177e-05, "loss": 0.567, "step": 4892 }, { "epoch": 0.17432352993569303, "grad_norm": 1.2717083729909107, "learning_rate": 1.892767845097864e-05, "loss": 0.5409, "step": 4893 }, { "epoch": 0.17435915706218716, "grad_norm": 1.5604551664242574, "learning_rate": 1.892715852343282e-05, "loss": 0.6801, "step": 4894 }, { "epoch": 0.17439478418868126, "grad_norm": 2.0923994303445745, "learning_rate": 1.8926638477015654e-05, "loss": 0.6609, "step": 4895 }, { "epoch": 0.17443041131517537, "grad_norm": 1.240809136340341, "learning_rate": 1.8926118311734055e-05, "loss": 0.5201, "step": 4896 }, { "epoch": 0.1744660384416695, "grad_norm": 1.7494256822349479, "learning_rate": 1.892559802759495e-05, "loss": 0.6227, "step": 4897 }, { "epoch": 0.1745016655681636, "grad_norm": 1.3976902338373987, "learning_rate": 1.8925077624605266e-05, "loss": 0.516, "step": 4898 }, { "epoch": 0.1745372926946577, "grad_norm": 1.1934726913978568, "learning_rate": 1.892455710277194e-05, "loss": 0.5049, "step": 4899 }, { "epoch": 0.17457291982115183, "grad_norm": 1.0980217477841139, "learning_rate": 1.8924036462101896e-05, "loss": 0.4743, "step": 4900 }, { "epoch": 0.17460854694764594, "grad_norm": 1.7316835119383935, "learning_rate": 1.892351570260207e-05, "loss": 0.6457, "step": 4901 }, { "epoch": 0.17464417407414004, "grad_norm": 1.1899695984179168, "learning_rate": 1.8922994824279394e-05, "loss": 0.4875, "step": 4902 }, { "epoch": 0.17467980120063417, "grad_norm": 1.208430596495691, "learning_rate": 1.8922473827140805e-05, "loss": 0.5523, "step": 4903 }, { "epoch": 0.17471542832712827, "grad_norm": 1.389948881547524, "learning_rate": 1.892195271119324e-05, "loss": 0.5556, "step": 4904 }, { "epoch": 0.17475105545362238, "grad_norm": 1.738438272125464, "learning_rate": 1.8921431476443642e-05, "loss": 0.6225, "step": 4905 }, { "epoch": 0.1747866825801165, "grad_norm": 1.16803950303808, "learning_rate": 1.892091012289895e-05, "loss": 0.521, "step": 4906 }, { "epoch": 0.1748223097066106, "grad_norm": 1.0890065841083163, "learning_rate": 1.8920388650566094e-05, "loss": 0.4785, "step": 4907 }, { "epoch": 0.17485793683310472, "grad_norm": 1.7943644860585524, "learning_rate": 1.8919867059452038e-05, "loss": 0.6246, "step": 4908 }, { "epoch": 0.17489356395959885, "grad_norm": 1.7284694687431421, "learning_rate": 1.8919345349563714e-05, "loss": 0.5771, "step": 4909 }, { "epoch": 0.17492919108609295, "grad_norm": 1.3514935889116706, "learning_rate": 1.8918823520908073e-05, "loss": 0.645, "step": 4910 }, { "epoch": 0.17496481821258705, "grad_norm": 1.7998859355285983, "learning_rate": 1.891830157349206e-05, "loss": 0.6135, "step": 4911 }, { "epoch": 0.17500044533908118, "grad_norm": 1.978674649078944, "learning_rate": 1.891777950732263e-05, "loss": 0.5818, "step": 4912 }, { "epoch": 0.1750360724655753, "grad_norm": 1.522814256571241, "learning_rate": 1.8917257322406735e-05, "loss": 0.6467, "step": 4913 }, { "epoch": 0.1750716995920694, "grad_norm": 1.6416244850451218, "learning_rate": 1.891673501875132e-05, "loss": 0.6466, "step": 4914 }, { "epoch": 0.17510732671856352, "grad_norm": 1.7086083976770534, "learning_rate": 1.8916212596363355e-05, "loss": 0.6053, "step": 4915 }, { "epoch": 0.17514295384505763, "grad_norm": 1.5662838717638659, "learning_rate": 1.891569005524978e-05, "loss": 0.6444, "step": 4916 }, { "epoch": 0.17517858097155173, "grad_norm": 1.6223566179905542, "learning_rate": 1.8915167395417563e-05, "loss": 0.614, "step": 4917 }, { "epoch": 0.17521420809804586, "grad_norm": 2.275689286774848, "learning_rate": 1.891464461687366e-05, "loss": 0.6057, "step": 4918 }, { "epoch": 0.17524983522453996, "grad_norm": 2.3323090058180957, "learning_rate": 1.8914121719625033e-05, "loss": 0.5217, "step": 4919 }, { "epoch": 0.17528546235103407, "grad_norm": 2.05238890710082, "learning_rate": 1.8913598703678648e-05, "loss": 0.6491, "step": 4920 }, { "epoch": 0.1753210894775282, "grad_norm": 1.9069731430695906, "learning_rate": 1.891307556904146e-05, "loss": 0.5972, "step": 4921 }, { "epoch": 0.1753567166040223, "grad_norm": 1.880777069870327, "learning_rate": 1.8912552315720446e-05, "loss": 0.4909, "step": 4922 }, { "epoch": 0.1753923437305164, "grad_norm": 2.019916190867171, "learning_rate": 1.8912028943722562e-05, "loss": 0.5684, "step": 4923 }, { "epoch": 0.17542797085701053, "grad_norm": 1.460214806335107, "learning_rate": 1.891150545305479e-05, "loss": 0.5921, "step": 4924 }, { "epoch": 0.17546359798350464, "grad_norm": 1.2460035568589527, "learning_rate": 1.891098184372409e-05, "loss": 0.5437, "step": 4925 }, { "epoch": 0.17549922510999874, "grad_norm": 1.225980776123508, "learning_rate": 1.891045811573744e-05, "loss": 0.5191, "step": 4926 }, { "epoch": 0.17553485223649287, "grad_norm": 1.779362320933239, "learning_rate": 1.890993426910181e-05, "loss": 0.6426, "step": 4927 }, { "epoch": 0.17557047936298698, "grad_norm": 2.247965018717166, "learning_rate": 1.8909410303824182e-05, "loss": 0.6725, "step": 4928 }, { "epoch": 0.17560610648948108, "grad_norm": 1.8459518544965412, "learning_rate": 1.8908886219911528e-05, "loss": 0.5101, "step": 4929 }, { "epoch": 0.1756417336159752, "grad_norm": 1.2546448174793294, "learning_rate": 1.8908362017370826e-05, "loss": 0.5284, "step": 4930 }, { "epoch": 0.1756773607424693, "grad_norm": 3.1018570232617337, "learning_rate": 1.8907837696209055e-05, "loss": 0.6248, "step": 4931 }, { "epoch": 0.17571298786896342, "grad_norm": 1.2926082027108992, "learning_rate": 1.89073132564332e-05, "loss": 0.5573, "step": 4932 }, { "epoch": 0.17574861499545755, "grad_norm": 2.0846515835554023, "learning_rate": 1.8906788698050244e-05, "loss": 0.6134, "step": 4933 }, { "epoch": 0.17578424212195165, "grad_norm": 1.916484353294405, "learning_rate": 1.890626402106717e-05, "loss": 0.5634, "step": 4934 }, { "epoch": 0.17581986924844575, "grad_norm": 1.3621890213885912, "learning_rate": 1.8905739225490966e-05, "loss": 0.5181, "step": 4935 }, { "epoch": 0.17585549637493988, "grad_norm": 1.9296497332881775, "learning_rate": 1.890521431132862e-05, "loss": 0.663, "step": 4936 }, { "epoch": 0.175891123501434, "grad_norm": 3.5132367849141946, "learning_rate": 1.890468927858712e-05, "loss": 0.6089, "step": 4937 }, { "epoch": 0.1759267506279281, "grad_norm": 1.1151495465862755, "learning_rate": 1.890416412727346e-05, "loss": 0.5181, "step": 4938 }, { "epoch": 0.17596237775442222, "grad_norm": 1.8643419854753365, "learning_rate": 1.8903638857394627e-05, "loss": 0.6498, "step": 4939 }, { "epoch": 0.17599800488091633, "grad_norm": 1.407874448657441, "learning_rate": 1.8903113468957624e-05, "loss": 0.5957, "step": 4940 }, { "epoch": 0.17603363200741043, "grad_norm": 1.2784266489929286, "learning_rate": 1.8902587961969442e-05, "loss": 0.5716, "step": 4941 }, { "epoch": 0.17606925913390456, "grad_norm": 1.8650229231881592, "learning_rate": 1.8902062336437078e-05, "loss": 0.5476, "step": 4942 }, { "epoch": 0.17610488626039866, "grad_norm": 1.1440418720909002, "learning_rate": 1.890153659236753e-05, "loss": 0.5119, "step": 4943 }, { "epoch": 0.17614051338689277, "grad_norm": 1.2509455156350913, "learning_rate": 1.89010107297678e-05, "loss": 0.4881, "step": 4944 }, { "epoch": 0.1761761405133869, "grad_norm": 1.0345615807122353, "learning_rate": 1.8900484748644893e-05, "loss": 0.5442, "step": 4945 }, { "epoch": 0.176211767639881, "grad_norm": 1.8868343021118112, "learning_rate": 1.889995864900581e-05, "loss": 0.6129, "step": 4946 }, { "epoch": 0.17624739476637513, "grad_norm": 1.9363695588460255, "learning_rate": 1.8899432430857554e-05, "loss": 0.5406, "step": 4947 }, { "epoch": 0.17628302189286923, "grad_norm": 1.8694322245323234, "learning_rate": 1.889890609420714e-05, "loss": 0.636, "step": 4948 }, { "epoch": 0.17631864901936334, "grad_norm": 1.7398513407743827, "learning_rate": 1.889837963906157e-05, "loss": 0.6003, "step": 4949 }, { "epoch": 0.17635427614585747, "grad_norm": 1.5790912709112215, "learning_rate": 1.8897853065427853e-05, "loss": 0.611, "step": 4950 }, { "epoch": 0.17638990327235157, "grad_norm": 2.0885559588221807, "learning_rate": 1.8897326373313004e-05, "loss": 0.5884, "step": 4951 }, { "epoch": 0.17642553039884568, "grad_norm": 1.164223852966284, "learning_rate": 1.8896799562724037e-05, "loss": 0.5091, "step": 4952 }, { "epoch": 0.1764611575253398, "grad_norm": 1.9916996198095511, "learning_rate": 1.889627263366796e-05, "loss": 0.7904, "step": 4953 }, { "epoch": 0.1764967846518339, "grad_norm": 1.1126303074469843, "learning_rate": 1.8895745586151803e-05, "loss": 0.4873, "step": 4954 }, { "epoch": 0.176532411778328, "grad_norm": 1.6290679078087473, "learning_rate": 1.889521842018257e-05, "loss": 0.6654, "step": 4955 }, { "epoch": 0.17656803890482214, "grad_norm": 1.8015176650663955, "learning_rate": 1.889469113576729e-05, "loss": 0.5937, "step": 4956 }, { "epoch": 0.17660366603131625, "grad_norm": 2.098229336223921, "learning_rate": 1.889416373291298e-05, "loss": 0.6172, "step": 4957 }, { "epoch": 0.17663929315781035, "grad_norm": 1.2205031608325536, "learning_rate": 1.889363621162666e-05, "loss": 0.5193, "step": 4958 }, { "epoch": 0.17667492028430448, "grad_norm": 1.8982873263767595, "learning_rate": 1.8893108571915357e-05, "loss": 0.6146, "step": 4959 }, { "epoch": 0.17671054741079859, "grad_norm": 1.1488900663588075, "learning_rate": 1.8892580813786102e-05, "loss": 0.5039, "step": 4960 }, { "epoch": 0.1767461745372927, "grad_norm": 1.7848638056302744, "learning_rate": 1.8892052937245914e-05, "loss": 0.6629, "step": 4961 }, { "epoch": 0.17678180166378682, "grad_norm": 1.6785297495755123, "learning_rate": 1.889152494230183e-05, "loss": 0.588, "step": 4962 }, { "epoch": 0.17681742879028092, "grad_norm": 1.5685362172458088, "learning_rate": 1.8890996828960874e-05, "loss": 0.5038, "step": 4963 }, { "epoch": 0.17685305591677503, "grad_norm": 1.5926690730847641, "learning_rate": 1.889046859723008e-05, "loss": 0.5347, "step": 4964 }, { "epoch": 0.17688868304326916, "grad_norm": 1.2162220237791157, "learning_rate": 1.888994024711648e-05, "loss": 0.502, "step": 4965 }, { "epoch": 0.17692431016976326, "grad_norm": 2.103541614723526, "learning_rate": 1.888941177862712e-05, "loss": 0.6761, "step": 4966 }, { "epoch": 0.17695993729625736, "grad_norm": 1.8620261937816094, "learning_rate": 1.8888883191769027e-05, "loss": 0.5642, "step": 4967 }, { "epoch": 0.1769955644227515, "grad_norm": 1.7116998238542276, "learning_rate": 1.8888354486549238e-05, "loss": 0.6638, "step": 4968 }, { "epoch": 0.1770311915492456, "grad_norm": 1.3361368208760283, "learning_rate": 1.8887825662974796e-05, "loss": 0.5601, "step": 4969 }, { "epoch": 0.1770668186757397, "grad_norm": 1.6930137002167387, "learning_rate": 1.8887296721052745e-05, "loss": 0.6456, "step": 4970 }, { "epoch": 0.17710244580223383, "grad_norm": 1.3788373316326927, "learning_rate": 1.8886767660790128e-05, "loss": 0.5228, "step": 4971 }, { "epoch": 0.17713807292872794, "grad_norm": 1.2005574504049392, "learning_rate": 1.8886238482193983e-05, "loss": 0.5256, "step": 4972 }, { "epoch": 0.17717370005522204, "grad_norm": 1.1529311699898686, "learning_rate": 1.8885709185271363e-05, "loss": 0.5358, "step": 4973 }, { "epoch": 0.17720932718171617, "grad_norm": 1.7935891589675081, "learning_rate": 1.8885179770029317e-05, "loss": 0.6691, "step": 4974 }, { "epoch": 0.17724495430821027, "grad_norm": 1.8528663769571503, "learning_rate": 1.888465023647489e-05, "loss": 0.6243, "step": 4975 }, { "epoch": 0.17728058143470438, "grad_norm": 1.1689072425516307, "learning_rate": 1.8884120584615136e-05, "loss": 0.4863, "step": 4976 }, { "epoch": 0.1773162085611985, "grad_norm": 1.621799660241363, "learning_rate": 1.8883590814457106e-05, "loss": 0.5335, "step": 4977 }, { "epoch": 0.1773518356876926, "grad_norm": 1.1970269166733873, "learning_rate": 1.8883060926007856e-05, "loss": 0.4925, "step": 4978 }, { "epoch": 0.1773874628141867, "grad_norm": 1.3736465153875828, "learning_rate": 1.8882530919274442e-05, "loss": 0.5298, "step": 4979 }, { "epoch": 0.17742308994068084, "grad_norm": 1.264931871368554, "learning_rate": 1.8882000794263916e-05, "loss": 0.5451, "step": 4980 }, { "epoch": 0.17745871706717495, "grad_norm": 1.2549420794757333, "learning_rate": 1.8881470550983345e-05, "loss": 0.5101, "step": 4981 }, { "epoch": 0.17749434419366905, "grad_norm": 1.3541669720832918, "learning_rate": 1.8880940189439786e-05, "loss": 0.5703, "step": 4982 }, { "epoch": 0.17752997132016318, "grad_norm": 2.2072328904419645, "learning_rate": 1.88804097096403e-05, "loss": 0.5636, "step": 4983 }, { "epoch": 0.17756559844665729, "grad_norm": 1.1700905155772818, "learning_rate": 1.8879879111591954e-05, "loss": 0.5067, "step": 4984 }, { "epoch": 0.1776012255731514, "grad_norm": 1.6242136519569554, "learning_rate": 1.8879348395301805e-05, "loss": 0.5894, "step": 4985 }, { "epoch": 0.17763685269964552, "grad_norm": 2.089798433887911, "learning_rate": 1.8878817560776932e-05, "loss": 0.623, "step": 4986 }, { "epoch": 0.17767247982613962, "grad_norm": 1.7430346238313943, "learning_rate": 1.8878286608024394e-05, "loss": 0.6133, "step": 4987 }, { "epoch": 0.17770810695263373, "grad_norm": 1.1977550319222197, "learning_rate": 1.8877755537051263e-05, "loss": 0.5008, "step": 4988 }, { "epoch": 0.17774373407912786, "grad_norm": 1.2236839076703412, "learning_rate": 1.8877224347864612e-05, "loss": 0.5156, "step": 4989 }, { "epoch": 0.17777936120562196, "grad_norm": 1.5206398867150386, "learning_rate": 1.8876693040471518e-05, "loss": 0.5795, "step": 4990 }, { "epoch": 0.17781498833211606, "grad_norm": 1.893325883362192, "learning_rate": 1.887616161487905e-05, "loss": 0.6594, "step": 4991 }, { "epoch": 0.1778506154586102, "grad_norm": 1.9797739200270075, "learning_rate": 1.8875630071094285e-05, "loss": 0.6693, "step": 4992 }, { "epoch": 0.1778862425851043, "grad_norm": 1.3413315424889443, "learning_rate": 1.8875098409124302e-05, "loss": 0.4653, "step": 4993 }, { "epoch": 0.1779218697115984, "grad_norm": 1.910473994287628, "learning_rate": 1.887456662897618e-05, "loss": 0.6374, "step": 4994 }, { "epoch": 0.17795749683809253, "grad_norm": 2.037353395598097, "learning_rate": 1.8874034730657002e-05, "loss": 0.5761, "step": 4995 }, { "epoch": 0.17799312396458664, "grad_norm": 1.1885812919159886, "learning_rate": 1.8873502714173848e-05, "loss": 0.5555, "step": 4996 }, { "epoch": 0.17802875109108074, "grad_norm": 1.8186761099513604, "learning_rate": 1.8872970579533805e-05, "loss": 0.6045, "step": 4997 }, { "epoch": 0.17806437821757487, "grad_norm": 1.4274273440539318, "learning_rate": 1.8872438326743955e-05, "loss": 0.5662, "step": 4998 }, { "epoch": 0.17810000534406897, "grad_norm": 1.6412344406920554, "learning_rate": 1.8871905955811387e-05, "loss": 0.5985, "step": 4999 }, { "epoch": 0.17813563247056308, "grad_norm": 1.2179121912075426, "learning_rate": 1.887137346674319e-05, "loss": 0.4889, "step": 5000 }, { "epoch": 0.1781712595970572, "grad_norm": 1.8546176838871937, "learning_rate": 1.8870840859546455e-05, "loss": 0.6531, "step": 5001 }, { "epoch": 0.1782068867235513, "grad_norm": 1.9055824239810808, "learning_rate": 1.8870308134228274e-05, "loss": 0.6049, "step": 5002 }, { "epoch": 0.17824251385004541, "grad_norm": 2.0706237910618395, "learning_rate": 1.886977529079574e-05, "loss": 0.6017, "step": 5003 }, { "epoch": 0.17827814097653955, "grad_norm": 1.176976840169998, "learning_rate": 1.886924232925595e-05, "loss": 0.5507, "step": 5004 }, { "epoch": 0.17831376810303365, "grad_norm": 1.7162132528424636, "learning_rate": 1.8868709249615994e-05, "loss": 0.5772, "step": 5005 }, { "epoch": 0.17834939522952775, "grad_norm": 1.4266931226226087, "learning_rate": 1.886817605188298e-05, "loss": 0.4961, "step": 5006 }, { "epoch": 0.17838502235602188, "grad_norm": 1.5798162711052017, "learning_rate": 1.8867642736064e-05, "loss": 0.6208, "step": 5007 }, { "epoch": 0.17842064948251599, "grad_norm": 1.6516809698550752, "learning_rate": 1.8867109302166165e-05, "loss": 0.547, "step": 5008 }, { "epoch": 0.1784562766090101, "grad_norm": 2.2305518620086358, "learning_rate": 1.8866575750196568e-05, "loss": 0.5983, "step": 5009 }, { "epoch": 0.17849190373550422, "grad_norm": 1.8438062189010016, "learning_rate": 1.8866042080162317e-05, "loss": 0.5737, "step": 5010 }, { "epoch": 0.17852753086199832, "grad_norm": 1.2311381502485388, "learning_rate": 1.886550829207052e-05, "loss": 0.5376, "step": 5011 }, { "epoch": 0.17856315798849243, "grad_norm": 1.649962523631145, "learning_rate": 1.8864974385928284e-05, "loss": 0.6428, "step": 5012 }, { "epoch": 0.17859878511498656, "grad_norm": 1.241640799209089, "learning_rate": 1.8864440361742714e-05, "loss": 0.5768, "step": 5013 }, { "epoch": 0.17863441224148066, "grad_norm": 1.1935407986792108, "learning_rate": 1.886390621952093e-05, "loss": 0.5463, "step": 5014 }, { "epoch": 0.17867003936797476, "grad_norm": 1.7913068869523898, "learning_rate": 1.8863371959270034e-05, "loss": 0.6367, "step": 5015 }, { "epoch": 0.1787056664944689, "grad_norm": 1.8985043182840333, "learning_rate": 1.8862837580997147e-05, "loss": 0.6903, "step": 5016 }, { "epoch": 0.178741293620963, "grad_norm": 1.1935774858796508, "learning_rate": 1.8862303084709385e-05, "loss": 0.5149, "step": 5017 }, { "epoch": 0.1787769207474571, "grad_norm": 2.2757186101346787, "learning_rate": 1.886176847041386e-05, "loss": 0.6306, "step": 5018 }, { "epoch": 0.17881254787395123, "grad_norm": 2.334868306743952, "learning_rate": 1.8861233738117695e-05, "loss": 0.6617, "step": 5019 }, { "epoch": 0.17884817500044534, "grad_norm": 2.375176912613868, "learning_rate": 1.8860698887828014e-05, "loss": 0.6787, "step": 5020 }, { "epoch": 0.17888380212693944, "grad_norm": 1.9327011081548748, "learning_rate": 1.8860163919551925e-05, "loss": 0.6891, "step": 5021 }, { "epoch": 0.17891942925343357, "grad_norm": 1.819895486836048, "learning_rate": 1.8859628833296568e-05, "loss": 0.7566, "step": 5022 }, { "epoch": 0.17895505637992767, "grad_norm": 2.035508059723014, "learning_rate": 1.8859093629069057e-05, "loss": 0.6815, "step": 5023 }, { "epoch": 0.17899068350642178, "grad_norm": 1.6648546764956853, "learning_rate": 1.8858558306876524e-05, "loss": 0.6581, "step": 5024 }, { "epoch": 0.1790263106329159, "grad_norm": 2.0784950307435985, "learning_rate": 1.8858022866726092e-05, "loss": 0.6045, "step": 5025 }, { "epoch": 0.17906193775941, "grad_norm": 1.5567618595415493, "learning_rate": 1.88574873086249e-05, "loss": 0.6986, "step": 5026 }, { "epoch": 0.17909756488590411, "grad_norm": 1.2304402741174438, "learning_rate": 1.885695163258007e-05, "loss": 0.4887, "step": 5027 }, { "epoch": 0.17913319201239825, "grad_norm": 1.7240864518869745, "learning_rate": 1.8856415838598738e-05, "loss": 0.6124, "step": 5028 }, { "epoch": 0.17916881913889235, "grad_norm": 1.1960285595684206, "learning_rate": 1.885587992668804e-05, "loss": 0.5634, "step": 5029 }, { "epoch": 0.17920444626538645, "grad_norm": 1.72539303104562, "learning_rate": 1.8855343896855107e-05, "loss": 0.6513, "step": 5030 }, { "epoch": 0.17924007339188058, "grad_norm": 1.4233795807799614, "learning_rate": 1.8854807749107086e-05, "loss": 0.5987, "step": 5031 }, { "epoch": 0.1792757005183747, "grad_norm": 1.2793780382414837, "learning_rate": 1.8854271483451106e-05, "loss": 0.519, "step": 5032 }, { "epoch": 0.1793113276448688, "grad_norm": 1.1221687462653327, "learning_rate": 1.8853735099894317e-05, "loss": 0.4629, "step": 5033 }, { "epoch": 0.17934695477136292, "grad_norm": 1.2464979205640259, "learning_rate": 1.885319859844385e-05, "loss": 0.5703, "step": 5034 }, { "epoch": 0.17938258189785702, "grad_norm": 1.8263107386241382, "learning_rate": 1.8852661979106864e-05, "loss": 0.6341, "step": 5035 }, { "epoch": 0.17941820902435113, "grad_norm": 1.2023813517992277, "learning_rate": 1.8852125241890492e-05, "loss": 0.606, "step": 5036 }, { "epoch": 0.17945383615084526, "grad_norm": 2.1235409535888725, "learning_rate": 1.8851588386801886e-05, "loss": 0.4992, "step": 5037 }, { "epoch": 0.17948946327733936, "grad_norm": 1.5707538162234518, "learning_rate": 1.885105141384819e-05, "loss": 0.6021, "step": 5038 }, { "epoch": 0.17952509040383346, "grad_norm": 1.794982737590698, "learning_rate": 1.8850514323036564e-05, "loss": 0.6605, "step": 5039 }, { "epoch": 0.1795607175303276, "grad_norm": 1.4957015085222556, "learning_rate": 1.884997711437415e-05, "loss": 0.6124, "step": 5040 }, { "epoch": 0.1795963446568217, "grad_norm": 1.157852692372449, "learning_rate": 1.8849439787868105e-05, "loss": 0.5169, "step": 5041 }, { "epoch": 0.17963197178331583, "grad_norm": 1.8149232682412157, "learning_rate": 1.8848902343525583e-05, "loss": 0.6834, "step": 5042 }, { "epoch": 0.17966759890980993, "grad_norm": 1.4070628772498752, "learning_rate": 1.8848364781353744e-05, "loss": 0.5461, "step": 5043 }, { "epoch": 0.17970322603630404, "grad_norm": 1.2136205015797423, "learning_rate": 1.884782710135974e-05, "loss": 0.5032, "step": 5044 }, { "epoch": 0.17973885316279817, "grad_norm": 1.2158252704119468, "learning_rate": 1.8847289303550738e-05, "loss": 0.4535, "step": 5045 }, { "epoch": 0.17977448028929227, "grad_norm": 1.933186167166573, "learning_rate": 1.8846751387933893e-05, "loss": 0.6104, "step": 5046 }, { "epoch": 0.17981010741578637, "grad_norm": 1.8108932330879806, "learning_rate": 1.8846213354516368e-05, "loss": 0.6008, "step": 5047 }, { "epoch": 0.1798457345422805, "grad_norm": 1.3948253507300978, "learning_rate": 1.8845675203305332e-05, "loss": 0.5638, "step": 5048 }, { "epoch": 0.1798813616687746, "grad_norm": 1.3164767708413656, "learning_rate": 1.8845136934307944e-05, "loss": 0.5149, "step": 5049 }, { "epoch": 0.1799169887952687, "grad_norm": 1.9538713789381417, "learning_rate": 1.884459854753138e-05, "loss": 0.6744, "step": 5050 }, { "epoch": 0.17995261592176284, "grad_norm": 1.2997916972399088, "learning_rate": 1.8844060042982804e-05, "loss": 0.5455, "step": 5051 }, { "epoch": 0.17998824304825695, "grad_norm": 1.3258130977552107, "learning_rate": 1.8843521420669383e-05, "loss": 0.5354, "step": 5052 }, { "epoch": 0.18002387017475105, "grad_norm": 1.1629795085202117, "learning_rate": 1.8842982680598295e-05, "loss": 0.4864, "step": 5053 }, { "epoch": 0.18005949730124518, "grad_norm": 1.2512865410405345, "learning_rate": 1.8842443822776714e-05, "loss": 0.5299, "step": 5054 }, { "epoch": 0.18009512442773928, "grad_norm": 1.251381693856659, "learning_rate": 1.8841904847211808e-05, "loss": 0.5444, "step": 5055 }, { "epoch": 0.1801307515542334, "grad_norm": 1.2446076792210976, "learning_rate": 1.8841365753910765e-05, "loss": 0.5409, "step": 5056 }, { "epoch": 0.18016637868072752, "grad_norm": 7.097334158048928, "learning_rate": 1.884082654288075e-05, "loss": 0.5377, "step": 5057 }, { "epoch": 0.18020200580722162, "grad_norm": 2.7295822710266813, "learning_rate": 1.8840287214128958e-05, "loss": 0.6395, "step": 5058 }, { "epoch": 0.18023763293371572, "grad_norm": 1.9056380345123074, "learning_rate": 1.883974776766256e-05, "loss": 0.5631, "step": 5059 }, { "epoch": 0.18027326006020986, "grad_norm": 1.3921274181987386, "learning_rate": 1.8839208203488738e-05, "loss": 0.4985, "step": 5060 }, { "epoch": 0.18030888718670396, "grad_norm": 1.2567676508416745, "learning_rate": 1.8838668521614685e-05, "loss": 0.51, "step": 5061 }, { "epoch": 0.18034451431319806, "grad_norm": 1.3339237592938291, "learning_rate": 1.8838128722047583e-05, "loss": 0.5181, "step": 5062 }, { "epoch": 0.1803801414396922, "grad_norm": 1.0968890710548795, "learning_rate": 1.8837588804794614e-05, "loss": 0.5372, "step": 5063 }, { "epoch": 0.1804157685661863, "grad_norm": 1.7501926282938483, "learning_rate": 1.8837048769862978e-05, "loss": 0.6494, "step": 5064 }, { "epoch": 0.1804513956926804, "grad_norm": 1.9070851130927768, "learning_rate": 1.883650861725986e-05, "loss": 0.644, "step": 5065 }, { "epoch": 0.18048702281917453, "grad_norm": 1.322398115286473, "learning_rate": 1.8835968346992454e-05, "loss": 0.5121, "step": 5066 }, { "epoch": 0.18052264994566863, "grad_norm": 2.0233504978881145, "learning_rate": 1.8835427959067952e-05, "loss": 0.5773, "step": 5067 }, { "epoch": 0.18055827707216274, "grad_norm": 1.4984930241100343, "learning_rate": 1.883488745349355e-05, "loss": 0.6435, "step": 5068 }, { "epoch": 0.18059390419865687, "grad_norm": 1.40478907645929, "learning_rate": 1.883434683027645e-05, "loss": 0.5486, "step": 5069 }, { "epoch": 0.18062953132515097, "grad_norm": 1.3431195299701162, "learning_rate": 1.8833806089423845e-05, "loss": 0.5633, "step": 5070 }, { "epoch": 0.18066515845164507, "grad_norm": 1.4846870619637238, "learning_rate": 1.8833265230942936e-05, "loss": 0.5555, "step": 5071 }, { "epoch": 0.1807007855781392, "grad_norm": 1.5960445235192677, "learning_rate": 1.8832724254840927e-05, "loss": 0.6456, "step": 5072 }, { "epoch": 0.1807364127046333, "grad_norm": 1.8521757539800054, "learning_rate": 1.8832183161125026e-05, "loss": 0.5726, "step": 5073 }, { "epoch": 0.1807720398311274, "grad_norm": 2.166318324693676, "learning_rate": 1.8831641949802425e-05, "loss": 0.5307, "step": 5074 }, { "epoch": 0.18080766695762154, "grad_norm": 1.176304712233278, "learning_rate": 1.883110062088034e-05, "loss": 0.5259, "step": 5075 }, { "epoch": 0.18084329408411565, "grad_norm": 1.3113379000007352, "learning_rate": 1.8830559174365984e-05, "loss": 0.5192, "step": 5076 }, { "epoch": 0.18087892121060975, "grad_norm": 1.303860205832684, "learning_rate": 1.8830017610266555e-05, "loss": 0.5396, "step": 5077 }, { "epoch": 0.18091454833710388, "grad_norm": 1.2154546536030846, "learning_rate": 1.8829475928589272e-05, "loss": 0.5159, "step": 5078 }, { "epoch": 0.18095017546359798, "grad_norm": 1.4275826117768509, "learning_rate": 1.8828934129341343e-05, "loss": 0.5717, "step": 5079 }, { "epoch": 0.1809858025900921, "grad_norm": 1.1557692037340672, "learning_rate": 1.8828392212529988e-05, "loss": 0.5034, "step": 5080 }, { "epoch": 0.18102142971658622, "grad_norm": 1.1753605465372243, "learning_rate": 1.8827850178162416e-05, "loss": 0.5514, "step": 5081 }, { "epoch": 0.18105705684308032, "grad_norm": 1.5749724377092627, "learning_rate": 1.882730802624585e-05, "loss": 0.6249, "step": 5082 }, { "epoch": 0.18109268396957443, "grad_norm": 1.912268413169164, "learning_rate": 1.882676575678751e-05, "loss": 0.5103, "step": 5083 }, { "epoch": 0.18112831109606856, "grad_norm": 1.6901559845150624, "learning_rate": 1.8826223369794614e-05, "loss": 0.5756, "step": 5084 }, { "epoch": 0.18116393822256266, "grad_norm": 1.0857206888825752, "learning_rate": 1.8825680865274385e-05, "loss": 0.5155, "step": 5085 }, { "epoch": 0.18119956534905676, "grad_norm": 1.7678589616520015, "learning_rate": 1.8825138243234047e-05, "loss": 0.6749, "step": 5086 }, { "epoch": 0.1812351924755509, "grad_norm": 1.9737611904313943, "learning_rate": 1.8824595503680823e-05, "loss": 0.5585, "step": 5087 }, { "epoch": 0.181270819602045, "grad_norm": 2.028720764244438, "learning_rate": 1.882405264662194e-05, "loss": 0.6948, "step": 5088 }, { "epoch": 0.1813064467285391, "grad_norm": 1.6843744700268557, "learning_rate": 1.882350967206463e-05, "loss": 0.59, "step": 5089 }, { "epoch": 0.18134207385503323, "grad_norm": 1.9775285838222136, "learning_rate": 1.8822966580016124e-05, "loss": 0.5272, "step": 5090 }, { "epoch": 0.18137770098152733, "grad_norm": 1.3248610248272565, "learning_rate": 1.8822423370483647e-05, "loss": 0.5202, "step": 5091 }, { "epoch": 0.18141332810802144, "grad_norm": 1.240355146712944, "learning_rate": 1.8821880043474443e-05, "loss": 0.463, "step": 5092 }, { "epoch": 0.18144895523451557, "grad_norm": 1.1934026259839163, "learning_rate": 1.882133659899573e-05, "loss": 0.5211, "step": 5093 }, { "epoch": 0.18148458236100967, "grad_norm": 1.4287448661621787, "learning_rate": 1.882079303705476e-05, "loss": 0.5128, "step": 5094 }, { "epoch": 0.18152020948750378, "grad_norm": 1.7255857594259543, "learning_rate": 1.8820249357658768e-05, "loss": 0.5804, "step": 5095 }, { "epoch": 0.1815558366139979, "grad_norm": 1.1147698502338428, "learning_rate": 1.8819705560814987e-05, "loss": 0.4544, "step": 5096 }, { "epoch": 0.181591463740492, "grad_norm": 1.9666997636596504, "learning_rate": 1.8819161646530665e-05, "loss": 0.6069, "step": 5097 }, { "epoch": 0.1816270908669861, "grad_norm": 2.2986939507978894, "learning_rate": 1.8818617614813037e-05, "loss": 0.5862, "step": 5098 }, { "epoch": 0.18166271799348024, "grad_norm": 1.1915076856052187, "learning_rate": 1.8818073465669354e-05, "loss": 0.5116, "step": 5099 }, { "epoch": 0.18169834511997435, "grad_norm": 1.9024520638520295, "learning_rate": 1.8817529199106858e-05, "loss": 0.6633, "step": 5100 }, { "epoch": 0.18173397224646845, "grad_norm": 1.7214918520062867, "learning_rate": 1.88169848151328e-05, "loss": 0.6325, "step": 5101 }, { "epoch": 0.18176959937296258, "grad_norm": 1.22163666285858, "learning_rate": 1.8816440313754425e-05, "loss": 0.5424, "step": 5102 }, { "epoch": 0.18180522649945668, "grad_norm": 2.5577273517053154, "learning_rate": 1.8815895694978987e-05, "loss": 0.5786, "step": 5103 }, { "epoch": 0.1818408536259508, "grad_norm": 2.160342062111809, "learning_rate": 1.8815350958813733e-05, "loss": 0.6878, "step": 5104 }, { "epoch": 0.18187648075244492, "grad_norm": 1.3220258574186143, "learning_rate": 1.881480610526592e-05, "loss": 0.5128, "step": 5105 }, { "epoch": 0.18191210787893902, "grad_norm": 1.7892169539689777, "learning_rate": 1.8814261134342804e-05, "loss": 0.6274, "step": 5106 }, { "epoch": 0.18194773500543313, "grad_norm": 1.9043665177478437, "learning_rate": 1.881371604605164e-05, "loss": 0.5686, "step": 5107 }, { "epoch": 0.18198336213192726, "grad_norm": 1.3834199569307626, "learning_rate": 1.8813170840399686e-05, "loss": 0.5467, "step": 5108 }, { "epoch": 0.18201898925842136, "grad_norm": 1.2852769180185366, "learning_rate": 1.88126255173942e-05, "loss": 0.5119, "step": 5109 }, { "epoch": 0.18205461638491546, "grad_norm": 1.094715215782606, "learning_rate": 1.881208007704245e-05, "loss": 0.49, "step": 5110 }, { "epoch": 0.1820902435114096, "grad_norm": 1.113657062751258, "learning_rate": 1.881153451935169e-05, "loss": 0.5361, "step": 5111 }, { "epoch": 0.1821258706379037, "grad_norm": 1.4207907814549177, "learning_rate": 1.881098884432919e-05, "loss": 0.5214, "step": 5112 }, { "epoch": 0.1821614977643978, "grad_norm": 1.538541347651352, "learning_rate": 1.8810443051982215e-05, "loss": 0.607, "step": 5113 }, { "epoch": 0.18219712489089193, "grad_norm": 1.381225885591163, "learning_rate": 1.8809897142318038e-05, "loss": 0.5121, "step": 5114 }, { "epoch": 0.18223275201738603, "grad_norm": 1.2437079228536896, "learning_rate": 1.8809351115343918e-05, "loss": 0.5317, "step": 5115 }, { "epoch": 0.18226837914388014, "grad_norm": 1.811704155607464, "learning_rate": 1.8808804971067132e-05, "loss": 0.6271, "step": 5116 }, { "epoch": 0.18230400627037427, "grad_norm": 1.2629597242087212, "learning_rate": 1.8808258709494945e-05, "loss": 0.5059, "step": 5117 }, { "epoch": 0.18233963339686837, "grad_norm": 1.2747521329048561, "learning_rate": 1.8807712330634645e-05, "loss": 0.492, "step": 5118 }, { "epoch": 0.18237526052336248, "grad_norm": 1.4972080246193609, "learning_rate": 1.8807165834493495e-05, "loss": 0.6198, "step": 5119 }, { "epoch": 0.1824108876498566, "grad_norm": 1.3062402144776772, "learning_rate": 1.8806619221078776e-05, "loss": 0.5267, "step": 5120 }, { "epoch": 0.1824465147763507, "grad_norm": 1.8048382923868378, "learning_rate": 1.8806072490397768e-05, "loss": 0.5277, "step": 5121 }, { "epoch": 0.1824821419028448, "grad_norm": 1.8207373590682001, "learning_rate": 1.880552564245775e-05, "loss": 0.7241, "step": 5122 }, { "epoch": 0.18251776902933894, "grad_norm": 1.4035285732787781, "learning_rate": 1.8804978677266003e-05, "loss": 0.545, "step": 5123 }, { "epoch": 0.18255339615583305, "grad_norm": 2.3714651822348944, "learning_rate": 1.880443159482981e-05, "loss": 0.6079, "step": 5124 }, { "epoch": 0.18258902328232715, "grad_norm": 1.2107329523660928, "learning_rate": 1.8803884395156456e-05, "loss": 0.5211, "step": 5125 }, { "epoch": 0.18262465040882128, "grad_norm": 1.1358056772328362, "learning_rate": 1.8803337078253228e-05, "loss": 0.4925, "step": 5126 }, { "epoch": 0.18266027753531539, "grad_norm": 1.3256401012816172, "learning_rate": 1.8802789644127413e-05, "loss": 0.5835, "step": 5127 }, { "epoch": 0.1826959046618095, "grad_norm": 1.118823653335115, "learning_rate": 1.8802242092786302e-05, "loss": 0.4807, "step": 5128 }, { "epoch": 0.18273153178830362, "grad_norm": 1.8769070296482397, "learning_rate": 1.8801694424237184e-05, "loss": 0.6866, "step": 5129 }, { "epoch": 0.18276715891479772, "grad_norm": 1.6343107460570443, "learning_rate": 1.8801146638487355e-05, "loss": 0.5833, "step": 5130 }, { "epoch": 0.18280278604129183, "grad_norm": 1.2359653111029656, "learning_rate": 1.8800598735544105e-05, "loss": 0.4856, "step": 5131 }, { "epoch": 0.18283841316778596, "grad_norm": 1.656156058848968, "learning_rate": 1.8800050715414733e-05, "loss": 0.6556, "step": 5132 }, { "epoch": 0.18287404029428006, "grad_norm": 1.169154980728192, "learning_rate": 1.8799502578106533e-05, "loss": 0.5431, "step": 5133 }, { "epoch": 0.1829096674207742, "grad_norm": 1.3004000657782688, "learning_rate": 1.8798954323626808e-05, "loss": 0.5054, "step": 5134 }, { "epoch": 0.1829452945472683, "grad_norm": 1.6094380629895764, "learning_rate": 1.8798405951982853e-05, "loss": 0.6051, "step": 5135 }, { "epoch": 0.1829809216737624, "grad_norm": 1.3388107556562816, "learning_rate": 1.8797857463181977e-05, "loss": 0.5178, "step": 5136 }, { "epoch": 0.18301654880025653, "grad_norm": 1.1505386245914417, "learning_rate": 1.8797308857231477e-05, "loss": 0.5121, "step": 5137 }, { "epoch": 0.18305217592675063, "grad_norm": 1.1533163007758105, "learning_rate": 1.879676013413866e-05, "loss": 0.4928, "step": 5138 }, { "epoch": 0.18308780305324474, "grad_norm": 1.7523357756825553, "learning_rate": 1.8796211293910836e-05, "loss": 0.5753, "step": 5139 }, { "epoch": 0.18312343017973887, "grad_norm": 1.2017325647164678, "learning_rate": 1.8795662336555307e-05, "loss": 0.472, "step": 5140 }, { "epoch": 0.18315905730623297, "grad_norm": 1.2571949799340705, "learning_rate": 1.8795113262079388e-05, "loss": 0.5362, "step": 5141 }, { "epoch": 0.18319468443272707, "grad_norm": 1.355592557841485, "learning_rate": 1.879456407049039e-05, "loss": 0.5418, "step": 5142 }, { "epoch": 0.1832303115592212, "grad_norm": 1.2262710724923735, "learning_rate": 1.8794014761795624e-05, "loss": 0.5131, "step": 5143 }, { "epoch": 0.1832659386857153, "grad_norm": 2.2879024758021314, "learning_rate": 1.87934653360024e-05, "loss": 0.5616, "step": 5144 }, { "epoch": 0.1833015658122094, "grad_norm": 1.9945449405111286, "learning_rate": 1.8792915793118042e-05, "loss": 0.636, "step": 5145 }, { "epoch": 0.18333719293870354, "grad_norm": 1.5586122078737283, "learning_rate": 1.8792366133149866e-05, "loss": 0.5933, "step": 5146 }, { "epoch": 0.18337282006519764, "grad_norm": 1.286992553789423, "learning_rate": 1.8791816356105186e-05, "loss": 0.5494, "step": 5147 }, { "epoch": 0.18340844719169175, "grad_norm": 1.217305815778783, "learning_rate": 1.879126646199133e-05, "loss": 0.5059, "step": 5148 }, { "epoch": 0.18344407431818588, "grad_norm": 1.8651168845101078, "learning_rate": 1.8790716450815614e-05, "loss": 0.6914, "step": 5149 }, { "epoch": 0.18347970144467998, "grad_norm": 1.357226510839253, "learning_rate": 1.8790166322585365e-05, "loss": 0.5408, "step": 5150 }, { "epoch": 0.18351532857117409, "grad_norm": 1.4866708439439065, "learning_rate": 1.8789616077307905e-05, "loss": 0.5151, "step": 5151 }, { "epoch": 0.18355095569766822, "grad_norm": 1.293572455544473, "learning_rate": 1.8789065714990567e-05, "loss": 0.4809, "step": 5152 }, { "epoch": 0.18358658282416232, "grad_norm": 1.2637404851748897, "learning_rate": 1.8788515235640675e-05, "loss": 0.5269, "step": 5153 }, { "epoch": 0.18362220995065642, "grad_norm": 2.8105611406616493, "learning_rate": 1.878796463926556e-05, "loss": 0.5898, "step": 5154 }, { "epoch": 0.18365783707715055, "grad_norm": 1.5345386988525909, "learning_rate": 1.878741392587255e-05, "loss": 0.541, "step": 5155 }, { "epoch": 0.18369346420364466, "grad_norm": 1.470874977839249, "learning_rate": 1.8786863095468985e-05, "loss": 0.5383, "step": 5156 }, { "epoch": 0.18372909133013876, "grad_norm": 1.4497827988159966, "learning_rate": 1.8786312148062196e-05, "loss": 0.512, "step": 5157 }, { "epoch": 0.1837647184566329, "grad_norm": 1.7856046663977818, "learning_rate": 1.878576108365952e-05, "loss": 0.6677, "step": 5158 }, { "epoch": 0.183800345583127, "grad_norm": 1.4087535441731431, "learning_rate": 1.878520990226829e-05, "loss": 0.5431, "step": 5159 }, { "epoch": 0.1838359727096211, "grad_norm": 1.3256027559607, "learning_rate": 1.8784658603895855e-05, "loss": 0.515, "step": 5160 }, { "epoch": 0.18387159983611523, "grad_norm": 1.2596966943908898, "learning_rate": 1.878410718854955e-05, "loss": 0.5204, "step": 5161 }, { "epoch": 0.18390722696260933, "grad_norm": 2.7926607777136425, "learning_rate": 1.8783555656236714e-05, "loss": 0.6494, "step": 5162 }, { "epoch": 0.18394285408910344, "grad_norm": 1.207158964634589, "learning_rate": 1.87830040069647e-05, "loss": 0.4961, "step": 5163 }, { "epoch": 0.18397848121559757, "grad_norm": 1.3379476412116227, "learning_rate": 1.8782452240740845e-05, "loss": 0.5461, "step": 5164 }, { "epoch": 0.18401410834209167, "grad_norm": 1.7252790464746353, "learning_rate": 1.87819003575725e-05, "loss": 0.6097, "step": 5165 }, { "epoch": 0.18404973546858577, "grad_norm": 1.30137348924417, "learning_rate": 1.8781348357467017e-05, "loss": 0.5769, "step": 5166 }, { "epoch": 0.1840853625950799, "grad_norm": 1.7850016661345525, "learning_rate": 1.878079624043174e-05, "loss": 0.562, "step": 5167 }, { "epoch": 0.184120989721574, "grad_norm": 2.0393064886463472, "learning_rate": 1.878024400647402e-05, "loss": 0.6371, "step": 5168 }, { "epoch": 0.1841566168480681, "grad_norm": 1.5209068021748955, "learning_rate": 1.877969165560122e-05, "loss": 0.5079, "step": 5169 }, { "epoch": 0.18419224397456224, "grad_norm": 1.9709808799942046, "learning_rate": 1.8779139187820687e-05, "loss": 0.5717, "step": 5170 }, { "epoch": 0.18422787110105635, "grad_norm": 1.2759006946589735, "learning_rate": 1.877858660313978e-05, "loss": 0.4983, "step": 5171 }, { "epoch": 0.18426349822755045, "grad_norm": 1.9759368243927713, "learning_rate": 1.8778033901565856e-05, "loss": 0.5545, "step": 5172 }, { "epoch": 0.18429912535404458, "grad_norm": 1.4964379093725417, "learning_rate": 1.8777481083106274e-05, "loss": 0.5143, "step": 5173 }, { "epoch": 0.18433475248053868, "grad_norm": 2.241090190143908, "learning_rate": 1.8776928147768397e-05, "loss": 0.6034, "step": 5174 }, { "epoch": 0.18437037960703279, "grad_norm": 1.9021989788463156, "learning_rate": 1.8776375095559588e-05, "loss": 0.5264, "step": 5175 }, { "epoch": 0.18440600673352692, "grad_norm": 1.5845337848830612, "learning_rate": 1.877582192648721e-05, "loss": 0.6344, "step": 5176 }, { "epoch": 0.18444163386002102, "grad_norm": 2.0390927189512764, "learning_rate": 1.877526864055863e-05, "loss": 0.6556, "step": 5177 }, { "epoch": 0.18447726098651512, "grad_norm": 2.1303595737250944, "learning_rate": 1.8774715237781213e-05, "loss": 0.7302, "step": 5178 }, { "epoch": 0.18451288811300925, "grad_norm": 1.8962528956456122, "learning_rate": 1.877416171816233e-05, "loss": 0.5927, "step": 5179 }, { "epoch": 0.18454851523950336, "grad_norm": 1.4233631452105504, "learning_rate": 1.8773608081709353e-05, "loss": 0.5486, "step": 5180 }, { "epoch": 0.18458414236599746, "grad_norm": 1.5639787493713873, "learning_rate": 1.8773054328429645e-05, "loss": 0.6275, "step": 5181 }, { "epoch": 0.1846197694924916, "grad_norm": 1.352564226722957, "learning_rate": 1.8772500458330595e-05, "loss": 0.4808, "step": 5182 }, { "epoch": 0.1846553966189857, "grad_norm": 1.3438307375220466, "learning_rate": 1.8771946471419563e-05, "loss": 0.5096, "step": 5183 }, { "epoch": 0.1846910237454798, "grad_norm": 1.9975917874969158, "learning_rate": 1.8771392367703937e-05, "loss": 0.5764, "step": 5184 }, { "epoch": 0.18472665087197393, "grad_norm": 1.634227161132273, "learning_rate": 1.877083814719109e-05, "loss": 0.5041, "step": 5185 }, { "epoch": 0.18476227799846803, "grad_norm": 1.3654506553747068, "learning_rate": 1.87702838098884e-05, "loss": 0.4939, "step": 5186 }, { "epoch": 0.18479790512496214, "grad_norm": 1.807549982117144, "learning_rate": 1.8769729355803254e-05, "loss": 0.5099, "step": 5187 }, { "epoch": 0.18483353225145627, "grad_norm": 1.4383511459282599, "learning_rate": 1.8769174784943032e-05, "loss": 0.527, "step": 5188 }, { "epoch": 0.18486915937795037, "grad_norm": 1.423895434548049, "learning_rate": 1.8768620097315117e-05, "loss": 0.5406, "step": 5189 }, { "epoch": 0.18490478650444447, "grad_norm": 2.2055834916786208, "learning_rate": 1.8768065292926895e-05, "loss": 0.5922, "step": 5190 }, { "epoch": 0.1849404136309386, "grad_norm": 1.3734618145353903, "learning_rate": 1.876751037178576e-05, "loss": 0.5516, "step": 5191 }, { "epoch": 0.1849760407574327, "grad_norm": 1.3240559243597387, "learning_rate": 1.8766955333899092e-05, "loss": 0.5476, "step": 5192 }, { "epoch": 0.1850116678839268, "grad_norm": 1.785771079474848, "learning_rate": 1.8766400179274287e-05, "loss": 0.5809, "step": 5193 }, { "epoch": 0.18504729501042094, "grad_norm": 1.4436189967145514, "learning_rate": 1.876584490791874e-05, "loss": 0.4897, "step": 5194 }, { "epoch": 0.18508292213691505, "grad_norm": 1.614238361904024, "learning_rate": 1.8765289519839832e-05, "loss": 0.6304, "step": 5195 }, { "epoch": 0.18511854926340915, "grad_norm": 1.439259743879681, "learning_rate": 1.8764734015044975e-05, "loss": 0.5915, "step": 5196 }, { "epoch": 0.18515417638990328, "grad_norm": 1.5067682008583185, "learning_rate": 1.8764178393541555e-05, "loss": 0.5429, "step": 5197 }, { "epoch": 0.18518980351639738, "grad_norm": 1.1956629211904768, "learning_rate": 1.876362265533698e-05, "loss": 0.5016, "step": 5198 }, { "epoch": 0.1852254306428915, "grad_norm": 1.7511215802578148, "learning_rate": 1.8763066800438638e-05, "loss": 0.5308, "step": 5199 }, { "epoch": 0.18526105776938562, "grad_norm": 1.2993701283596075, "learning_rate": 1.8762510828853936e-05, "loss": 0.5105, "step": 5200 }, { "epoch": 0.18529668489587972, "grad_norm": 1.300638696535689, "learning_rate": 1.8761954740590283e-05, "loss": 0.5624, "step": 5201 }, { "epoch": 0.18533231202237382, "grad_norm": 1.24813144626778, "learning_rate": 1.876139853565507e-05, "loss": 0.5279, "step": 5202 }, { "epoch": 0.18536793914886796, "grad_norm": 1.1296366939428308, "learning_rate": 1.876084221405572e-05, "loss": 0.5224, "step": 5203 }, { "epoch": 0.18540356627536206, "grad_norm": 1.345058848401211, "learning_rate": 1.876028577579963e-05, "loss": 0.4803, "step": 5204 }, { "epoch": 0.18543919340185616, "grad_norm": 1.223451073706973, "learning_rate": 1.875972922089421e-05, "loss": 0.5618, "step": 5205 }, { "epoch": 0.1854748205283503, "grad_norm": 1.7338916468756291, "learning_rate": 1.8759172549346872e-05, "loss": 0.6936, "step": 5206 }, { "epoch": 0.1855104476548444, "grad_norm": 2.4196055749417953, "learning_rate": 1.8758615761165034e-05, "loss": 0.5795, "step": 5207 }, { "epoch": 0.1855460747813385, "grad_norm": 2.0250498636249428, "learning_rate": 1.87580588563561e-05, "loss": 0.5946, "step": 5208 }, { "epoch": 0.18558170190783263, "grad_norm": 1.9473093791003266, "learning_rate": 1.8757501834927495e-05, "loss": 0.6264, "step": 5209 }, { "epoch": 0.18561732903432673, "grad_norm": 1.8021435312719034, "learning_rate": 1.875694469688663e-05, "loss": 0.6432, "step": 5210 }, { "epoch": 0.18565295616082084, "grad_norm": 1.2744390699300225, "learning_rate": 1.8756387442240923e-05, "loss": 0.5088, "step": 5211 }, { "epoch": 0.18568858328731497, "grad_norm": 1.193769947194177, "learning_rate": 1.87558300709978e-05, "loss": 0.4742, "step": 5212 }, { "epoch": 0.18572421041380907, "grad_norm": 1.3143607314926182, "learning_rate": 1.8755272583164678e-05, "loss": 0.4973, "step": 5213 }, { "epoch": 0.18575983754030317, "grad_norm": 1.3056003748351435, "learning_rate": 1.8754714978748983e-05, "loss": 0.5305, "step": 5214 }, { "epoch": 0.1857954646667973, "grad_norm": 1.1921215543837118, "learning_rate": 1.875415725775814e-05, "loss": 0.5172, "step": 5215 }, { "epoch": 0.1858310917932914, "grad_norm": 1.1452703079700581, "learning_rate": 1.8753599420199574e-05, "loss": 0.5219, "step": 5216 }, { "epoch": 0.1858667189197855, "grad_norm": 1.576403676457364, "learning_rate": 1.8753041466080713e-05, "loss": 0.5332, "step": 5217 }, { "epoch": 0.18590234604627964, "grad_norm": 1.1320395966192354, "learning_rate": 1.8752483395408987e-05, "loss": 0.5353, "step": 5218 }, { "epoch": 0.18593797317277375, "grad_norm": 1.3761050616075, "learning_rate": 1.8751925208191822e-05, "loss": 0.5353, "step": 5219 }, { "epoch": 0.18597360029926785, "grad_norm": 1.2345686377455942, "learning_rate": 1.8751366904436663e-05, "loss": 0.5505, "step": 5220 }, { "epoch": 0.18600922742576198, "grad_norm": 1.2554849755476694, "learning_rate": 1.8750808484150936e-05, "loss": 0.5431, "step": 5221 }, { "epoch": 0.18604485455225608, "grad_norm": 2.1402262144638944, "learning_rate": 1.8750249947342075e-05, "loss": 0.6804, "step": 5222 }, { "epoch": 0.1860804816787502, "grad_norm": 2.11013585264349, "learning_rate": 1.874969129401752e-05, "loss": 0.6467, "step": 5223 }, { "epoch": 0.18611610880524432, "grad_norm": 1.8450886925290362, "learning_rate": 1.874913252418471e-05, "loss": 0.6571, "step": 5224 }, { "epoch": 0.18615173593173842, "grad_norm": 1.3403813558837303, "learning_rate": 1.8748573637851085e-05, "loss": 0.537, "step": 5225 }, { "epoch": 0.18618736305823252, "grad_norm": 1.4656545383983561, "learning_rate": 1.8748014635024086e-05, "loss": 0.5918, "step": 5226 }, { "epoch": 0.18622299018472666, "grad_norm": 1.3776184638116085, "learning_rate": 1.8747455515711158e-05, "loss": 0.5635, "step": 5227 }, { "epoch": 0.18625861731122076, "grad_norm": 1.2847544701270825, "learning_rate": 1.8746896279919745e-05, "loss": 0.5101, "step": 5228 }, { "epoch": 0.1862942444377149, "grad_norm": 1.6168321725600017, "learning_rate": 1.87463369276573e-05, "loss": 0.5331, "step": 5229 }, { "epoch": 0.186329871564209, "grad_norm": 1.3288938999728641, "learning_rate": 1.874577745893126e-05, "loss": 0.5305, "step": 5230 }, { "epoch": 0.1863654986907031, "grad_norm": 2.156392506634316, "learning_rate": 1.8745217873749083e-05, "loss": 0.5234, "step": 5231 }, { "epoch": 0.18640112581719723, "grad_norm": 1.186469420235896, "learning_rate": 1.8744658172118215e-05, "loss": 0.5175, "step": 5232 }, { "epoch": 0.18643675294369133, "grad_norm": 1.5291905784501796, "learning_rate": 1.874409835404611e-05, "loss": 0.5481, "step": 5233 }, { "epoch": 0.18647238007018543, "grad_norm": 1.7370334371059692, "learning_rate": 1.8743538419540228e-05, "loss": 0.6367, "step": 5234 }, { "epoch": 0.18650800719667956, "grad_norm": 1.3819960387799493, "learning_rate": 1.874297836860802e-05, "loss": 0.57, "step": 5235 }, { "epoch": 0.18654363432317367, "grad_norm": 1.3429548246076544, "learning_rate": 1.8742418201256944e-05, "loss": 0.5569, "step": 5236 }, { "epoch": 0.18657926144966777, "grad_norm": 1.6560404394425245, "learning_rate": 1.8741857917494454e-05, "loss": 0.6168, "step": 5237 }, { "epoch": 0.1866148885761619, "grad_norm": 1.9839332512138284, "learning_rate": 1.874129751732802e-05, "loss": 0.6163, "step": 5238 }, { "epoch": 0.186650515702656, "grad_norm": 2.264193006483764, "learning_rate": 1.87407370007651e-05, "loss": 0.6444, "step": 5239 }, { "epoch": 0.1866861428291501, "grad_norm": 1.444057676145564, "learning_rate": 1.8740176367813156e-05, "loss": 0.483, "step": 5240 }, { "epoch": 0.18672176995564424, "grad_norm": 1.8844767088316952, "learning_rate": 1.8739615618479657e-05, "loss": 0.6011, "step": 5241 }, { "epoch": 0.18675739708213834, "grad_norm": 1.6116921203819892, "learning_rate": 1.8739054752772064e-05, "loss": 0.5512, "step": 5242 }, { "epoch": 0.18679302420863245, "grad_norm": 1.9989608622330761, "learning_rate": 1.873849377069785e-05, "loss": 0.6349, "step": 5243 }, { "epoch": 0.18682865133512658, "grad_norm": 1.3384220149584176, "learning_rate": 1.8737932672264487e-05, "loss": 0.5399, "step": 5244 }, { "epoch": 0.18686427846162068, "grad_norm": 1.4525121239702683, "learning_rate": 1.8737371457479437e-05, "loss": 0.611, "step": 5245 }, { "epoch": 0.18689990558811478, "grad_norm": 1.8020411914802281, "learning_rate": 1.8736810126350184e-05, "loss": 0.5338, "step": 5246 }, { "epoch": 0.18693553271460892, "grad_norm": 1.6606097222804277, "learning_rate": 1.8736248678884195e-05, "loss": 0.6502, "step": 5247 }, { "epoch": 0.18697115984110302, "grad_norm": 1.1724919526579478, "learning_rate": 1.8735687115088947e-05, "loss": 0.4886, "step": 5248 }, { "epoch": 0.18700678696759712, "grad_norm": 1.7531394167930099, "learning_rate": 1.873512543497192e-05, "loss": 0.5795, "step": 5249 }, { "epoch": 0.18704241409409125, "grad_norm": 1.2888535188773396, "learning_rate": 1.8734563638540596e-05, "loss": 0.5157, "step": 5250 }, { "epoch": 0.18707804122058536, "grad_norm": 1.636768474672736, "learning_rate": 1.873400172580245e-05, "loss": 0.6806, "step": 5251 }, { "epoch": 0.18711366834707946, "grad_norm": 1.188063115512165, "learning_rate": 1.8733439696764968e-05, "loss": 0.5664, "step": 5252 }, { "epoch": 0.1871492954735736, "grad_norm": 1.1633753460660043, "learning_rate": 1.873287755143563e-05, "loss": 0.4836, "step": 5253 }, { "epoch": 0.1871849226000677, "grad_norm": 1.9050558773376887, "learning_rate": 1.873231528982192e-05, "loss": 0.6092, "step": 5254 }, { "epoch": 0.1872205497265618, "grad_norm": 1.6368643269800085, "learning_rate": 1.8731752911931332e-05, "loss": 0.5858, "step": 5255 }, { "epoch": 0.18725617685305593, "grad_norm": 1.244690852138939, "learning_rate": 1.873119041777135e-05, "loss": 0.5036, "step": 5256 }, { "epoch": 0.18729180397955003, "grad_norm": 1.6083217207050111, "learning_rate": 1.8730627807349468e-05, "loss": 0.6956, "step": 5257 }, { "epoch": 0.18732743110604413, "grad_norm": 1.69797327129356, "learning_rate": 1.8730065080673173e-05, "loss": 0.6177, "step": 5258 }, { "epoch": 0.18736305823253827, "grad_norm": 1.1810545954946337, "learning_rate": 1.872950223774996e-05, "loss": 0.4913, "step": 5259 }, { "epoch": 0.18739868535903237, "grad_norm": 1.7707842172248311, "learning_rate": 1.872893927858732e-05, "loss": 0.557, "step": 5260 }, { "epoch": 0.18743431248552647, "grad_norm": 1.7225463533247285, "learning_rate": 1.8728376203192755e-05, "loss": 0.624, "step": 5261 }, { "epoch": 0.1874699396120206, "grad_norm": 1.0914338546380247, "learning_rate": 1.872781301157376e-05, "loss": 0.4547, "step": 5262 }, { "epoch": 0.1875055667385147, "grad_norm": 1.554576155149984, "learning_rate": 1.8727249703737832e-05, "loss": 0.6869, "step": 5263 }, { "epoch": 0.1875411938650088, "grad_norm": 1.6340234198725507, "learning_rate": 1.8726686279692475e-05, "loss": 0.6109, "step": 5264 }, { "epoch": 0.18757682099150294, "grad_norm": 1.2282583519657, "learning_rate": 1.872612273944519e-05, "loss": 0.5738, "step": 5265 }, { "epoch": 0.18761244811799704, "grad_norm": 1.9376722437814091, "learning_rate": 1.8725559083003483e-05, "loss": 0.626, "step": 5266 }, { "epoch": 0.18764807524449115, "grad_norm": 1.1163815563213222, "learning_rate": 1.872499531037486e-05, "loss": 0.5296, "step": 5267 }, { "epoch": 0.18768370237098528, "grad_norm": 1.2333381793212406, "learning_rate": 1.8724431421566822e-05, "loss": 0.5334, "step": 5268 }, { "epoch": 0.18771932949747938, "grad_norm": 1.651411206698331, "learning_rate": 1.8723867416586885e-05, "loss": 0.7007, "step": 5269 }, { "epoch": 0.18775495662397348, "grad_norm": 2.1464986525436816, "learning_rate": 1.8723303295442553e-05, "loss": 0.5596, "step": 5270 }, { "epoch": 0.18779058375046762, "grad_norm": 1.2187901460999582, "learning_rate": 1.8722739058141344e-05, "loss": 0.4691, "step": 5271 }, { "epoch": 0.18782621087696172, "grad_norm": 1.3353201960460923, "learning_rate": 1.8722174704690762e-05, "loss": 0.5292, "step": 5272 }, { "epoch": 0.18786183800345582, "grad_norm": 1.2114444720129, "learning_rate": 1.8721610235098332e-05, "loss": 0.5557, "step": 5273 }, { "epoch": 0.18789746512994995, "grad_norm": 1.1785025879156075, "learning_rate": 1.8721045649371567e-05, "loss": 0.5014, "step": 5274 }, { "epoch": 0.18793309225644406, "grad_norm": 1.4820031027603415, "learning_rate": 1.8720480947517977e-05, "loss": 0.6151, "step": 5275 }, { "epoch": 0.18796871938293816, "grad_norm": 1.6357075331188966, "learning_rate": 1.8719916129545094e-05, "loss": 0.542, "step": 5276 }, { "epoch": 0.1880043465094323, "grad_norm": 1.657269870665421, "learning_rate": 1.871935119546043e-05, "loss": 0.5619, "step": 5277 }, { "epoch": 0.1880399736359264, "grad_norm": 1.3381590465074646, "learning_rate": 1.871878614527151e-05, "loss": 0.5505, "step": 5278 }, { "epoch": 0.1880756007624205, "grad_norm": 1.61295136224167, "learning_rate": 1.8718220978985862e-05, "loss": 0.5591, "step": 5279 }, { "epoch": 0.18811122788891463, "grad_norm": 1.3850962396419997, "learning_rate": 1.8717655696611004e-05, "loss": 0.5342, "step": 5280 }, { "epoch": 0.18814685501540873, "grad_norm": 1.3554549408212329, "learning_rate": 1.871709029815447e-05, "loss": 0.5562, "step": 5281 }, { "epoch": 0.18818248214190283, "grad_norm": 1.1929838451532162, "learning_rate": 1.8716524783623785e-05, "loss": 0.5169, "step": 5282 }, { "epoch": 0.18821810926839697, "grad_norm": 1.84362776754148, "learning_rate": 1.8715959153026476e-05, "loss": 0.6028, "step": 5283 }, { "epoch": 0.18825373639489107, "grad_norm": 1.7189024358807758, "learning_rate": 1.8715393406370085e-05, "loss": 0.5603, "step": 5284 }, { "epoch": 0.18828936352138517, "grad_norm": 1.9832526938526447, "learning_rate": 1.8714827543662134e-05, "loss": 0.5631, "step": 5285 }, { "epoch": 0.1883249906478793, "grad_norm": 1.231061298721198, "learning_rate": 1.8714261564910167e-05, "loss": 0.4985, "step": 5286 }, { "epoch": 0.1883606177743734, "grad_norm": 2.1544076540653285, "learning_rate": 1.8713695470121714e-05, "loss": 0.7222, "step": 5287 }, { "epoch": 0.1883962449008675, "grad_norm": 1.2117187123185686, "learning_rate": 1.8713129259304315e-05, "loss": 0.5362, "step": 5288 }, { "epoch": 0.18843187202736164, "grad_norm": 1.3056116829458733, "learning_rate": 1.871256293246551e-05, "loss": 0.4939, "step": 5289 }, { "epoch": 0.18846749915385574, "grad_norm": 1.623116619727941, "learning_rate": 1.8711996489612842e-05, "loss": 0.634, "step": 5290 }, { "epoch": 0.18850312628034985, "grad_norm": 1.5291513744561127, "learning_rate": 1.871142993075385e-05, "loss": 0.6035, "step": 5291 }, { "epoch": 0.18853875340684398, "grad_norm": 1.2298939277562257, "learning_rate": 1.871086325589608e-05, "loss": 0.5196, "step": 5292 }, { "epoch": 0.18857438053333808, "grad_norm": 1.1933456151365018, "learning_rate": 1.8710296465047077e-05, "loss": 0.4671, "step": 5293 }, { "epoch": 0.18861000765983219, "grad_norm": 2.194700717733695, "learning_rate": 1.8709729558214388e-05, "loss": 0.5418, "step": 5294 }, { "epoch": 0.18864563478632632, "grad_norm": 1.3533962494385057, "learning_rate": 1.870916253540556e-05, "loss": 0.5108, "step": 5295 }, { "epoch": 0.18868126191282042, "grad_norm": 2.9403691491811577, "learning_rate": 1.870859539662815e-05, "loss": 0.5519, "step": 5296 }, { "epoch": 0.18871688903931452, "grad_norm": 1.1571235032989342, "learning_rate": 1.87080281418897e-05, "loss": 0.5662, "step": 5297 }, { "epoch": 0.18875251616580865, "grad_norm": 1.210796985068234, "learning_rate": 1.8707460771197773e-05, "loss": 0.5219, "step": 5298 }, { "epoch": 0.18878814329230276, "grad_norm": 2.0980671079464814, "learning_rate": 1.870689328455992e-05, "loss": 0.6349, "step": 5299 }, { "epoch": 0.18882377041879686, "grad_norm": 1.9266494280432964, "learning_rate": 1.8706325681983694e-05, "loss": 0.6107, "step": 5300 }, { "epoch": 0.188859397545291, "grad_norm": 1.1972006932651227, "learning_rate": 1.870575796347666e-05, "loss": 0.4881, "step": 5301 }, { "epoch": 0.1888950246717851, "grad_norm": 1.9773791039679036, "learning_rate": 1.870519012904637e-05, "loss": 0.5429, "step": 5302 }, { "epoch": 0.1889306517982792, "grad_norm": 1.3778754060563174, "learning_rate": 1.870462217870039e-05, "loss": 0.5357, "step": 5303 }, { "epoch": 0.18896627892477333, "grad_norm": 1.4361118016413164, "learning_rate": 1.870405411244628e-05, "loss": 0.5556, "step": 5304 }, { "epoch": 0.18900190605126743, "grad_norm": 1.8361965375320608, "learning_rate": 1.870348593029161e-05, "loss": 0.5571, "step": 5305 }, { "epoch": 0.18903753317776154, "grad_norm": 1.2457961786050118, "learning_rate": 1.870291763224394e-05, "loss": 0.5251, "step": 5306 }, { "epoch": 0.18907316030425567, "grad_norm": 1.3953658467734504, "learning_rate": 1.870234921831084e-05, "loss": 0.5254, "step": 5307 }, { "epoch": 0.18910878743074977, "grad_norm": 1.6969181165622274, "learning_rate": 1.8701780688499872e-05, "loss": 0.6432, "step": 5308 }, { "epoch": 0.18914441455724387, "grad_norm": 1.6810575796855796, "learning_rate": 1.8701212042818616e-05, "loss": 0.6075, "step": 5309 }, { "epoch": 0.189180041683738, "grad_norm": 1.5962790023694504, "learning_rate": 1.8700643281274644e-05, "loss": 0.5572, "step": 5310 }, { "epoch": 0.1892156688102321, "grad_norm": 1.2599143356986826, "learning_rate": 1.8700074403875517e-05, "loss": 0.5591, "step": 5311 }, { "epoch": 0.1892512959367262, "grad_norm": 1.208869505975061, "learning_rate": 1.8699505410628823e-05, "loss": 0.4936, "step": 5312 }, { "epoch": 0.18928692306322034, "grad_norm": 1.8273888702533554, "learning_rate": 1.8698936301542133e-05, "loss": 0.6371, "step": 5313 }, { "epoch": 0.18932255018971444, "grad_norm": 1.6437836246452129, "learning_rate": 1.8698367076623024e-05, "loss": 0.7062, "step": 5314 }, { "epoch": 0.18935817731620855, "grad_norm": 1.2136866869113927, "learning_rate": 1.869779773587908e-05, "loss": 0.5287, "step": 5315 }, { "epoch": 0.18939380444270268, "grad_norm": 1.190990961384996, "learning_rate": 1.869722827931788e-05, "loss": 0.4709, "step": 5316 }, { "epoch": 0.18942943156919678, "grad_norm": 2.0132301668039974, "learning_rate": 1.8696658706947005e-05, "loss": 0.6363, "step": 5317 }, { "epoch": 0.18946505869569089, "grad_norm": 1.285295916958179, "learning_rate": 1.8696089018774043e-05, "loss": 0.5645, "step": 5318 }, { "epoch": 0.18950068582218502, "grad_norm": 1.6951412661891785, "learning_rate": 1.8695519214806576e-05, "loss": 0.5654, "step": 5319 }, { "epoch": 0.18953631294867912, "grad_norm": 1.2641038605356023, "learning_rate": 1.869494929505219e-05, "loss": 0.56, "step": 5320 }, { "epoch": 0.18957194007517322, "grad_norm": 1.3552782934283625, "learning_rate": 1.8694379259518482e-05, "loss": 0.5525, "step": 5321 }, { "epoch": 0.18960756720166735, "grad_norm": 1.3987890648127712, "learning_rate": 1.8693809108213035e-05, "loss": 0.5358, "step": 5322 }, { "epoch": 0.18964319432816146, "grad_norm": 1.114407781834, "learning_rate": 1.8693238841143442e-05, "loss": 0.4929, "step": 5323 }, { "epoch": 0.1896788214546556, "grad_norm": 2.213661935741936, "learning_rate": 1.8692668458317297e-05, "loss": 0.5658, "step": 5324 }, { "epoch": 0.1897144485811497, "grad_norm": 1.2726750759156307, "learning_rate": 1.8692097959742196e-05, "loss": 0.4743, "step": 5325 }, { "epoch": 0.1897500757076438, "grad_norm": 2.113278620009955, "learning_rate": 1.8691527345425737e-05, "loss": 0.6303, "step": 5326 }, { "epoch": 0.18978570283413793, "grad_norm": 1.9249656067398793, "learning_rate": 1.869095661537552e-05, "loss": 0.6634, "step": 5327 }, { "epoch": 0.18982132996063203, "grad_norm": 1.8778778054832022, "learning_rate": 1.8690385769599135e-05, "loss": 0.6242, "step": 5328 }, { "epoch": 0.18985695708712613, "grad_norm": 1.8252259062587968, "learning_rate": 1.868981480810419e-05, "loss": 0.6479, "step": 5329 }, { "epoch": 0.18989258421362026, "grad_norm": 1.37644152723741, "learning_rate": 1.8689243730898288e-05, "loss": 0.5193, "step": 5330 }, { "epoch": 0.18992821134011437, "grad_norm": 1.2157962667200355, "learning_rate": 1.8688672537989032e-05, "loss": 0.524, "step": 5331 }, { "epoch": 0.18996383846660847, "grad_norm": 1.5501794259312722, "learning_rate": 1.8688101229384028e-05, "loss": 0.5343, "step": 5332 }, { "epoch": 0.1899994655931026, "grad_norm": 1.2202757610278558, "learning_rate": 1.8687529805090884e-05, "loss": 0.5036, "step": 5333 }, { "epoch": 0.1900350927195967, "grad_norm": 1.110938610511082, "learning_rate": 1.8686958265117207e-05, "loss": 0.4977, "step": 5334 }, { "epoch": 0.1900707198460908, "grad_norm": 1.5997885648816297, "learning_rate": 1.868638660947061e-05, "loss": 0.5609, "step": 5335 }, { "epoch": 0.19010634697258494, "grad_norm": 1.7488241708275463, "learning_rate": 1.8685814838158703e-05, "loss": 0.6232, "step": 5336 }, { "epoch": 0.19014197409907904, "grad_norm": 1.2383671317526523, "learning_rate": 1.8685242951189106e-05, "loss": 0.5085, "step": 5337 }, { "epoch": 0.19017760122557315, "grad_norm": 2.1795656706196214, "learning_rate": 1.8684670948569425e-05, "loss": 0.6015, "step": 5338 }, { "epoch": 0.19021322835206728, "grad_norm": 2.0388931491182785, "learning_rate": 1.8684098830307276e-05, "loss": 0.5791, "step": 5339 }, { "epoch": 0.19024885547856138, "grad_norm": 1.0939125207107998, "learning_rate": 1.8683526596410284e-05, "loss": 0.4813, "step": 5340 }, { "epoch": 0.19028448260505548, "grad_norm": 2.0642448062551026, "learning_rate": 1.8682954246886067e-05, "loss": 0.6327, "step": 5341 }, { "epoch": 0.1903201097315496, "grad_norm": 1.1968904609954973, "learning_rate": 1.8682381781742246e-05, "loss": 0.5193, "step": 5342 }, { "epoch": 0.19035573685804372, "grad_norm": 1.7797927456319111, "learning_rate": 1.868180920098644e-05, "loss": 0.6067, "step": 5343 }, { "epoch": 0.19039136398453782, "grad_norm": 2.5015760428764002, "learning_rate": 1.8681236504626277e-05, "loss": 0.6996, "step": 5344 }, { "epoch": 0.19042699111103195, "grad_norm": 1.2930639590542217, "learning_rate": 1.8680663692669383e-05, "loss": 0.6205, "step": 5345 }, { "epoch": 0.19046261823752605, "grad_norm": 1.113919173162144, "learning_rate": 1.8680090765123388e-05, "loss": 0.4527, "step": 5346 }, { "epoch": 0.19049824536402016, "grad_norm": 2.203518391421134, "learning_rate": 1.867951772199591e-05, "loss": 0.6103, "step": 5347 }, { "epoch": 0.1905338724905143, "grad_norm": 1.3863078341881871, "learning_rate": 1.867894456329459e-05, "loss": 0.5502, "step": 5348 }, { "epoch": 0.1905694996170084, "grad_norm": 1.8497786568891745, "learning_rate": 1.867837128902706e-05, "loss": 0.5834, "step": 5349 }, { "epoch": 0.1906051267435025, "grad_norm": 1.2558333598827869, "learning_rate": 1.8677797899200945e-05, "loss": 0.5432, "step": 5350 }, { "epoch": 0.19064075386999663, "grad_norm": 1.8183570175644281, "learning_rate": 1.867722439382389e-05, "loss": 0.5403, "step": 5351 }, { "epoch": 0.19067638099649073, "grad_norm": 1.3924613558226029, "learning_rate": 1.8676650772903527e-05, "loss": 0.5494, "step": 5352 }, { "epoch": 0.19071200812298483, "grad_norm": 1.4818917145518284, "learning_rate": 1.867607703644749e-05, "loss": 0.6326, "step": 5353 }, { "epoch": 0.19074763524947896, "grad_norm": 1.649798320551494, "learning_rate": 1.867550318446343e-05, "loss": 0.5512, "step": 5354 }, { "epoch": 0.19078326237597307, "grad_norm": 3.2639265892397957, "learning_rate": 1.8674929216958977e-05, "loss": 0.7203, "step": 5355 }, { "epoch": 0.19081888950246717, "grad_norm": 1.1884993832621988, "learning_rate": 1.867435513394178e-05, "loss": 0.5315, "step": 5356 }, { "epoch": 0.1908545166289613, "grad_norm": 1.0937136547951698, "learning_rate": 1.867378093541948e-05, "loss": 0.506, "step": 5357 }, { "epoch": 0.1908901437554554, "grad_norm": 1.8728162687458798, "learning_rate": 1.8673206621399724e-05, "loss": 0.5562, "step": 5358 }, { "epoch": 0.1909257708819495, "grad_norm": 1.343585224201998, "learning_rate": 1.867263219189016e-05, "loss": 0.6199, "step": 5359 }, { "epoch": 0.19096139800844364, "grad_norm": 1.0773814977947738, "learning_rate": 1.8672057646898435e-05, "loss": 0.5387, "step": 5360 }, { "epoch": 0.19099702513493774, "grad_norm": 1.4900853750933694, "learning_rate": 1.8671482986432202e-05, "loss": 0.5319, "step": 5361 }, { "epoch": 0.19103265226143185, "grad_norm": 1.7599578097797746, "learning_rate": 1.8670908210499112e-05, "loss": 0.7265, "step": 5362 }, { "epoch": 0.19106827938792598, "grad_norm": 1.644191156938214, "learning_rate": 1.867033331910682e-05, "loss": 0.5923, "step": 5363 }, { "epoch": 0.19110390651442008, "grad_norm": 1.200069319162286, "learning_rate": 1.866975831226298e-05, "loss": 0.5226, "step": 5364 }, { "epoch": 0.19113953364091418, "grad_norm": 1.8439811464363929, "learning_rate": 1.8669183189975244e-05, "loss": 0.5437, "step": 5365 }, { "epoch": 0.19117516076740831, "grad_norm": 1.213693684531769, "learning_rate": 1.8668607952251277e-05, "loss": 0.5523, "step": 5366 }, { "epoch": 0.19121078789390242, "grad_norm": 1.7696726871273813, "learning_rate": 1.866803259909874e-05, "loss": 0.5799, "step": 5367 }, { "epoch": 0.19124641502039652, "grad_norm": 1.5669233216817167, "learning_rate": 1.8667457130525286e-05, "loss": 0.4955, "step": 5368 }, { "epoch": 0.19128204214689065, "grad_norm": 1.167707581602057, "learning_rate": 1.866688154653858e-05, "loss": 0.4856, "step": 5369 }, { "epoch": 0.19131766927338476, "grad_norm": 1.2485366041382024, "learning_rate": 1.8666305847146292e-05, "loss": 0.5888, "step": 5370 }, { "epoch": 0.19135329639987886, "grad_norm": 1.8132246029893466, "learning_rate": 1.8665730032356085e-05, "loss": 0.5248, "step": 5371 }, { "epoch": 0.191388923526373, "grad_norm": 1.0352533754484752, "learning_rate": 1.8665154102175626e-05, "loss": 0.4982, "step": 5372 }, { "epoch": 0.1914245506528671, "grad_norm": 1.7263061427299244, "learning_rate": 1.866457805661258e-05, "loss": 0.6051, "step": 5373 }, { "epoch": 0.1914601777793612, "grad_norm": 1.2070152400556877, "learning_rate": 1.8664001895674623e-05, "loss": 0.5217, "step": 5374 }, { "epoch": 0.19149580490585533, "grad_norm": 1.6447293910850542, "learning_rate": 1.8663425619369424e-05, "loss": 0.6628, "step": 5375 }, { "epoch": 0.19153143203234943, "grad_norm": 1.185372121128623, "learning_rate": 1.866284922770466e-05, "loss": 0.4778, "step": 5376 }, { "epoch": 0.19156705915884353, "grad_norm": 1.5784080562560046, "learning_rate": 1.8662272720688e-05, "loss": 0.6521, "step": 5377 }, { "epoch": 0.19160268628533766, "grad_norm": 1.1226693291972536, "learning_rate": 1.8661696098327125e-05, "loss": 0.5414, "step": 5378 }, { "epoch": 0.19163831341183177, "grad_norm": 1.1314501198852087, "learning_rate": 1.8661119360629714e-05, "loss": 0.537, "step": 5379 }, { "epoch": 0.19167394053832587, "grad_norm": 2.01464354477812, "learning_rate": 1.8660542507603444e-05, "loss": 0.6126, "step": 5380 }, { "epoch": 0.19170956766482, "grad_norm": 1.7519721834275659, "learning_rate": 1.8659965539255996e-05, "loss": 0.7166, "step": 5381 }, { "epoch": 0.1917451947913141, "grad_norm": 1.1670389078739503, "learning_rate": 1.8659388455595055e-05, "loss": 0.5145, "step": 5382 }, { "epoch": 0.1917808219178082, "grad_norm": 1.13784868843621, "learning_rate": 1.8658811256628308e-05, "loss": 0.5245, "step": 5383 }, { "epoch": 0.19181644904430234, "grad_norm": 1.5331701733462733, "learning_rate": 1.865823394236343e-05, "loss": 0.5653, "step": 5384 }, { "epoch": 0.19185207617079644, "grad_norm": 1.119072536675661, "learning_rate": 1.865765651280812e-05, "loss": 0.5009, "step": 5385 }, { "epoch": 0.19188770329729055, "grad_norm": 1.1333154351757913, "learning_rate": 1.8657078967970063e-05, "loss": 0.5189, "step": 5386 }, { "epoch": 0.19192333042378468, "grad_norm": 1.2850024354878842, "learning_rate": 1.8656501307856943e-05, "loss": 0.5415, "step": 5387 }, { "epoch": 0.19195895755027878, "grad_norm": 1.5309558101412366, "learning_rate": 1.8655923532476463e-05, "loss": 0.4899, "step": 5388 }, { "epoch": 0.19199458467677288, "grad_norm": 2.269962810731754, "learning_rate": 1.865534564183631e-05, "loss": 0.509, "step": 5389 }, { "epoch": 0.19203021180326701, "grad_norm": 3.1901791353368876, "learning_rate": 1.8654767635944178e-05, "loss": 0.6244, "step": 5390 }, { "epoch": 0.19206583892976112, "grad_norm": 1.2591104892853586, "learning_rate": 1.8654189514807765e-05, "loss": 0.5437, "step": 5391 }, { "epoch": 0.19210146605625522, "grad_norm": 1.2540506380204497, "learning_rate": 1.8653611278434774e-05, "loss": 0.5855, "step": 5392 }, { "epoch": 0.19213709318274935, "grad_norm": 1.185768844002009, "learning_rate": 1.8653032926832897e-05, "loss": 0.5481, "step": 5393 }, { "epoch": 0.19217272030924346, "grad_norm": 1.1377504659837174, "learning_rate": 1.865245446000984e-05, "loss": 0.5059, "step": 5394 }, { "epoch": 0.19220834743573756, "grad_norm": 1.1412160489973875, "learning_rate": 1.86518758779733e-05, "loss": 0.4802, "step": 5395 }, { "epoch": 0.1922439745622317, "grad_norm": 1.9392637346555892, "learning_rate": 1.865129718073099e-05, "loss": 0.6521, "step": 5396 }, { "epoch": 0.1922796016887258, "grad_norm": 1.1719841607533317, "learning_rate": 1.8650718368290613e-05, "loss": 0.5299, "step": 5397 }, { "epoch": 0.1923152288152199, "grad_norm": 1.4731221532935088, "learning_rate": 1.865013944065987e-05, "loss": 0.6592, "step": 5398 }, { "epoch": 0.19235085594171403, "grad_norm": 1.9048289971405163, "learning_rate": 1.8649560397846475e-05, "loss": 0.6119, "step": 5399 }, { "epoch": 0.19238648306820813, "grad_norm": 1.1345017542875058, "learning_rate": 1.864898123985814e-05, "loss": 0.4943, "step": 5400 }, { "epoch": 0.19242211019470223, "grad_norm": 1.2620169291789525, "learning_rate": 1.864840196670257e-05, "loss": 0.5372, "step": 5401 }, { "epoch": 0.19245773732119636, "grad_norm": 1.776152237773278, "learning_rate": 1.8647822578387488e-05, "loss": 0.4904, "step": 5402 }, { "epoch": 0.19249336444769047, "grad_norm": 1.1995749239332685, "learning_rate": 1.8647243074920604e-05, "loss": 0.5221, "step": 5403 }, { "epoch": 0.19252899157418457, "grad_norm": 1.0583450220611785, "learning_rate": 1.864666345630963e-05, "loss": 0.5033, "step": 5404 }, { "epoch": 0.1925646187006787, "grad_norm": 1.1671694668207842, "learning_rate": 1.864608372256229e-05, "loss": 0.5415, "step": 5405 }, { "epoch": 0.1926002458271728, "grad_norm": 1.883663326582901, "learning_rate": 1.86455038736863e-05, "loss": 0.54, "step": 5406 }, { "epoch": 0.1926358729536669, "grad_norm": 1.8945100945709337, "learning_rate": 1.864492390968939e-05, "loss": 0.6499, "step": 5407 }, { "epoch": 0.19267150008016104, "grad_norm": 2.061446861409161, "learning_rate": 1.864434383057927e-05, "loss": 0.6676, "step": 5408 }, { "epoch": 0.19270712720665514, "grad_norm": 1.2630199262936315, "learning_rate": 1.864376363636367e-05, "loss": 0.5515, "step": 5409 }, { "epoch": 0.19274275433314925, "grad_norm": 1.536368012104101, "learning_rate": 1.8643183327050315e-05, "loss": 0.5655, "step": 5410 }, { "epoch": 0.19277838145964338, "grad_norm": 1.6158412530280413, "learning_rate": 1.8642602902646937e-05, "loss": 0.6404, "step": 5411 }, { "epoch": 0.19281400858613748, "grad_norm": 1.784942787875346, "learning_rate": 1.8642022363161255e-05, "loss": 0.6227, "step": 5412 }, { "epoch": 0.19284963571263158, "grad_norm": 1.1944643668166492, "learning_rate": 1.864144170860101e-05, "loss": 0.5396, "step": 5413 }, { "epoch": 0.19288526283912572, "grad_norm": 1.271475172769629, "learning_rate": 1.8640860938973925e-05, "loss": 0.5324, "step": 5414 }, { "epoch": 0.19292088996561982, "grad_norm": 1.795695476435688, "learning_rate": 1.8640280054287738e-05, "loss": 0.6033, "step": 5415 }, { "epoch": 0.19295651709211395, "grad_norm": 1.3985908056075984, "learning_rate": 1.8639699054550187e-05, "loss": 0.4988, "step": 5416 }, { "epoch": 0.19299214421860805, "grad_norm": 1.838779248582993, "learning_rate": 1.8639117939768997e-05, "loss": 0.6365, "step": 5417 }, { "epoch": 0.19302777134510216, "grad_norm": 2.18642960675465, "learning_rate": 1.8638536709951916e-05, "loss": 0.6869, "step": 5418 }, { "epoch": 0.1930633984715963, "grad_norm": 1.7431266579832523, "learning_rate": 1.8637955365106685e-05, "loss": 0.5986, "step": 5419 }, { "epoch": 0.1930990255980904, "grad_norm": 1.2764829554058608, "learning_rate": 1.8637373905241036e-05, "loss": 0.5279, "step": 5420 }, { "epoch": 0.1931346527245845, "grad_norm": 1.7673605261121912, "learning_rate": 1.8636792330362716e-05, "loss": 0.6148, "step": 5421 }, { "epoch": 0.19317027985107862, "grad_norm": 1.7510783866976232, "learning_rate": 1.8636210640479474e-05, "loss": 0.6594, "step": 5422 }, { "epoch": 0.19320590697757273, "grad_norm": 1.1921465302239034, "learning_rate": 1.8635628835599048e-05, "loss": 0.4673, "step": 5423 }, { "epoch": 0.19324153410406683, "grad_norm": 1.3954608045531334, "learning_rate": 1.8635046915729186e-05, "loss": 0.509, "step": 5424 }, { "epoch": 0.19327716123056096, "grad_norm": 1.2545142586999056, "learning_rate": 1.8634464880877644e-05, "loss": 0.4901, "step": 5425 }, { "epoch": 0.19331278835705507, "grad_norm": 1.7120291312104188, "learning_rate": 1.8633882731052163e-05, "loss": 0.5416, "step": 5426 }, { "epoch": 0.19334841548354917, "grad_norm": 1.443390860734936, "learning_rate": 1.8633300466260497e-05, "loss": 0.5397, "step": 5427 }, { "epoch": 0.1933840426100433, "grad_norm": 1.9583108254158645, "learning_rate": 1.8632718086510404e-05, "loss": 0.655, "step": 5428 }, { "epoch": 0.1934196697365374, "grad_norm": 1.2497029054675768, "learning_rate": 1.8632135591809635e-05, "loss": 0.5426, "step": 5429 }, { "epoch": 0.1934552968630315, "grad_norm": 1.1700337785089712, "learning_rate": 1.8631552982165946e-05, "loss": 0.4533, "step": 5430 }, { "epoch": 0.19349092398952564, "grad_norm": 1.2748498069774357, "learning_rate": 1.8630970257587096e-05, "loss": 0.5138, "step": 5431 }, { "epoch": 0.19352655111601974, "grad_norm": 2.2608276119911066, "learning_rate": 1.8630387418080848e-05, "loss": 0.6537, "step": 5432 }, { "epoch": 0.19356217824251384, "grad_norm": 2.3348637275915975, "learning_rate": 1.8629804463654956e-05, "loss": 0.5972, "step": 5433 }, { "epoch": 0.19359780536900797, "grad_norm": 1.3413384147652507, "learning_rate": 1.8629221394317186e-05, "loss": 0.5614, "step": 5434 }, { "epoch": 0.19363343249550208, "grad_norm": 1.6492145921871157, "learning_rate": 1.8628638210075305e-05, "loss": 0.5611, "step": 5435 }, { "epoch": 0.19366905962199618, "grad_norm": 1.362264477022259, "learning_rate": 1.862805491093707e-05, "loss": 0.5787, "step": 5436 }, { "epoch": 0.1937046867484903, "grad_norm": 2.071393054525344, "learning_rate": 1.8627471496910257e-05, "loss": 0.7136, "step": 5437 }, { "epoch": 0.19374031387498442, "grad_norm": 1.8187437259984964, "learning_rate": 1.862688796800263e-05, "loss": 0.6019, "step": 5438 }, { "epoch": 0.19377594100147852, "grad_norm": 1.14982075761629, "learning_rate": 1.862630432422196e-05, "loss": 0.4948, "step": 5439 }, { "epoch": 0.19381156812797265, "grad_norm": 1.1307181032275195, "learning_rate": 1.8625720565576016e-05, "loss": 0.4656, "step": 5440 }, { "epoch": 0.19384719525446675, "grad_norm": 1.9696438524406588, "learning_rate": 1.8625136692072577e-05, "loss": 0.6562, "step": 5441 }, { "epoch": 0.19388282238096086, "grad_norm": 1.0746966502656905, "learning_rate": 1.862455270371941e-05, "loss": 0.4945, "step": 5442 }, { "epoch": 0.193918449507455, "grad_norm": 1.1941483326233007, "learning_rate": 1.8623968600524302e-05, "loss": 0.5004, "step": 5443 }, { "epoch": 0.1939540766339491, "grad_norm": 1.2125236850814896, "learning_rate": 1.862338438249502e-05, "loss": 0.5117, "step": 5444 }, { "epoch": 0.1939897037604432, "grad_norm": 1.0776374910883368, "learning_rate": 1.8622800049639347e-05, "loss": 0.4649, "step": 5445 }, { "epoch": 0.19402533088693732, "grad_norm": 1.5984531926095236, "learning_rate": 1.8622215601965066e-05, "loss": 0.6684, "step": 5446 }, { "epoch": 0.19406095801343143, "grad_norm": 2.075107420814405, "learning_rate": 1.8621631039479954e-05, "loss": 0.6825, "step": 5447 }, { "epoch": 0.19409658513992553, "grad_norm": 1.2073784477722296, "learning_rate": 1.8621046362191806e-05, "loss": 0.4777, "step": 5448 }, { "epoch": 0.19413221226641966, "grad_norm": 2.112204189798392, "learning_rate": 1.8620461570108396e-05, "loss": 0.6419, "step": 5449 }, { "epoch": 0.19416783939291377, "grad_norm": 1.1859502245573985, "learning_rate": 1.8619876663237513e-05, "loss": 0.5345, "step": 5450 }, { "epoch": 0.19420346651940787, "grad_norm": 1.262091602646824, "learning_rate": 1.8619291641586947e-05, "loss": 0.5159, "step": 5451 }, { "epoch": 0.194239093645902, "grad_norm": 2.345223977177434, "learning_rate": 1.861870650516449e-05, "loss": 0.5882, "step": 5452 }, { "epoch": 0.1942747207723961, "grad_norm": 1.8952404471153814, "learning_rate": 1.861812125397793e-05, "loss": 0.672, "step": 5453 }, { "epoch": 0.1943103478988902, "grad_norm": 1.3852230400447614, "learning_rate": 1.8617535888035066e-05, "loss": 0.5292, "step": 5454 }, { "epoch": 0.19434597502538434, "grad_norm": 1.6608038592812842, "learning_rate": 1.8616950407343685e-05, "loss": 0.5979, "step": 5455 }, { "epoch": 0.19438160215187844, "grad_norm": 1.85927036158245, "learning_rate": 1.861636481191159e-05, "loss": 0.6573, "step": 5456 }, { "epoch": 0.19441722927837254, "grad_norm": 1.163637051282601, "learning_rate": 1.861577910174657e-05, "loss": 0.5381, "step": 5457 }, { "epoch": 0.19445285640486668, "grad_norm": 1.2221197539598785, "learning_rate": 1.8615193276856434e-05, "loss": 0.502, "step": 5458 }, { "epoch": 0.19448848353136078, "grad_norm": 1.267886602996937, "learning_rate": 1.8614607337248975e-05, "loss": 0.4899, "step": 5459 }, { "epoch": 0.19452411065785488, "grad_norm": 1.6721238059134094, "learning_rate": 1.8614021282932e-05, "loss": 0.6749, "step": 5460 }, { "epoch": 0.194559737784349, "grad_norm": 2.2916482577344586, "learning_rate": 1.861343511391331e-05, "loss": 0.6454, "step": 5461 }, { "epoch": 0.19459536491084312, "grad_norm": 1.1577500777990208, "learning_rate": 1.861284883020071e-05, "loss": 0.4609, "step": 5462 }, { "epoch": 0.19463099203733722, "grad_norm": 2.7939842074399945, "learning_rate": 1.861226243180201e-05, "loss": 0.6603, "step": 5463 }, { "epoch": 0.19466661916383135, "grad_norm": 1.8029206758005536, "learning_rate": 1.861167591872501e-05, "loss": 0.5952, "step": 5464 }, { "epoch": 0.19470224629032545, "grad_norm": 1.648902462260956, "learning_rate": 1.8611089290977534e-05, "loss": 0.5178, "step": 5465 }, { "epoch": 0.19473787341681956, "grad_norm": 1.5591244039196386, "learning_rate": 1.861050254856738e-05, "loss": 0.5647, "step": 5466 }, { "epoch": 0.1947735005433137, "grad_norm": 1.6959830579839081, "learning_rate": 1.8609915691502367e-05, "loss": 0.6367, "step": 5467 }, { "epoch": 0.1948091276698078, "grad_norm": 1.8763146255031298, "learning_rate": 1.860932871979031e-05, "loss": 0.5189, "step": 5468 }, { "epoch": 0.1948447547963019, "grad_norm": 3.6209486044522885, "learning_rate": 1.8608741633439023e-05, "loss": 0.6024, "step": 5469 }, { "epoch": 0.19488038192279603, "grad_norm": 1.2061806492789469, "learning_rate": 1.8608154432456324e-05, "loss": 0.4662, "step": 5470 }, { "epoch": 0.19491600904929013, "grad_norm": 1.6225239500343214, "learning_rate": 1.860756711685003e-05, "loss": 0.6719, "step": 5471 }, { "epoch": 0.19495163617578423, "grad_norm": 1.5555626995521536, "learning_rate": 1.8606979686627966e-05, "loss": 0.488, "step": 5472 }, { "epoch": 0.19498726330227836, "grad_norm": 1.3864149160263455, "learning_rate": 1.860639214179795e-05, "loss": 0.4947, "step": 5473 }, { "epoch": 0.19502289042877247, "grad_norm": 1.534910982893989, "learning_rate": 1.860580448236781e-05, "loss": 0.5948, "step": 5474 }, { "epoch": 0.19505851755526657, "grad_norm": 2.0012763371069786, "learning_rate": 1.8605216708345363e-05, "loss": 0.6682, "step": 5475 }, { "epoch": 0.1950941446817607, "grad_norm": 1.9690977312416458, "learning_rate": 1.8604628819738447e-05, "loss": 0.6372, "step": 5476 }, { "epoch": 0.1951297718082548, "grad_norm": 1.6394864818122503, "learning_rate": 1.8604040816554882e-05, "loss": 0.5343, "step": 5477 }, { "epoch": 0.1951653989347489, "grad_norm": 1.7188935154104497, "learning_rate": 1.8603452698802498e-05, "loss": 0.6075, "step": 5478 }, { "epoch": 0.19520102606124304, "grad_norm": 1.3000686676033588, "learning_rate": 1.860286446648913e-05, "loss": 0.4737, "step": 5479 }, { "epoch": 0.19523665318773714, "grad_norm": 1.5797113446906292, "learning_rate": 1.8602276119622605e-05, "loss": 0.5614, "step": 5480 }, { "epoch": 0.19527228031423124, "grad_norm": 1.8341460721452012, "learning_rate": 1.8601687658210763e-05, "loss": 0.6288, "step": 5481 }, { "epoch": 0.19530790744072538, "grad_norm": 1.179875375045946, "learning_rate": 1.860109908226144e-05, "loss": 0.5137, "step": 5482 }, { "epoch": 0.19534353456721948, "grad_norm": 1.233499023380893, "learning_rate": 1.8600510391782465e-05, "loss": 0.5024, "step": 5483 }, { "epoch": 0.19537916169371358, "grad_norm": 1.725007028515613, "learning_rate": 1.859992158678169e-05, "loss": 0.6095, "step": 5484 }, { "epoch": 0.1954147888202077, "grad_norm": 1.811957910319111, "learning_rate": 1.8599332667266942e-05, "loss": 0.6694, "step": 5485 }, { "epoch": 0.19545041594670182, "grad_norm": 1.366911057795661, "learning_rate": 1.8598743633246073e-05, "loss": 0.531, "step": 5486 }, { "epoch": 0.19548604307319592, "grad_norm": 1.24146997392195, "learning_rate": 1.859815448472692e-05, "loss": 0.5891, "step": 5487 }, { "epoch": 0.19552167019969005, "grad_norm": 1.2998253818896652, "learning_rate": 1.8597565221717332e-05, "loss": 0.4815, "step": 5488 }, { "epoch": 0.19555729732618415, "grad_norm": 1.8899087028566044, "learning_rate": 1.859697584422515e-05, "loss": 0.6086, "step": 5489 }, { "epoch": 0.19559292445267826, "grad_norm": 1.1764076293697858, "learning_rate": 1.859638635225823e-05, "loss": 0.4974, "step": 5490 }, { "epoch": 0.1956285515791724, "grad_norm": 1.5837074515756862, "learning_rate": 1.8595796745824417e-05, "loss": 0.6505, "step": 5491 }, { "epoch": 0.1956641787056665, "grad_norm": 1.2452395074768863, "learning_rate": 1.8595207024931558e-05, "loss": 0.5291, "step": 5492 }, { "epoch": 0.1956998058321606, "grad_norm": 1.2758180473234644, "learning_rate": 1.8594617189587515e-05, "loss": 0.4589, "step": 5493 }, { "epoch": 0.19573543295865473, "grad_norm": 1.5736738340176535, "learning_rate": 1.8594027239800132e-05, "loss": 0.5486, "step": 5494 }, { "epoch": 0.19577106008514883, "grad_norm": 1.5557000109061982, "learning_rate": 1.8593437175577274e-05, "loss": 0.6112, "step": 5495 }, { "epoch": 0.19580668721164293, "grad_norm": 1.6588309085847697, "learning_rate": 1.8592846996926793e-05, "loss": 0.5741, "step": 5496 }, { "epoch": 0.19584231433813706, "grad_norm": 1.6454013251701576, "learning_rate": 1.8592256703856547e-05, "loss": 0.7239, "step": 5497 }, { "epoch": 0.19587794146463117, "grad_norm": 1.2838617641235452, "learning_rate": 1.8591666296374393e-05, "loss": 0.5064, "step": 5498 }, { "epoch": 0.19591356859112527, "grad_norm": 1.0891397750084357, "learning_rate": 1.8591075774488205e-05, "loss": 0.5203, "step": 5499 }, { "epoch": 0.1959491957176194, "grad_norm": 1.752580439949896, "learning_rate": 1.8590485138205833e-05, "loss": 0.5555, "step": 5500 }, { "epoch": 0.1959848228441135, "grad_norm": 2.2952794531262137, "learning_rate": 1.8589894387535148e-05, "loss": 0.6092, "step": 5501 }, { "epoch": 0.1960204499706076, "grad_norm": 1.3410497334547458, "learning_rate": 1.8589303522484017e-05, "loss": 0.5791, "step": 5502 }, { "epoch": 0.19605607709710174, "grad_norm": 1.2410165531634039, "learning_rate": 1.8588712543060304e-05, "loss": 0.553, "step": 5503 }, { "epoch": 0.19609170422359584, "grad_norm": 1.1011629469967914, "learning_rate": 1.8588121449271882e-05, "loss": 0.506, "step": 5504 }, { "epoch": 0.19612733135008995, "grad_norm": 1.1868774589106945, "learning_rate": 1.8587530241126617e-05, "loss": 0.5473, "step": 5505 }, { "epoch": 0.19616295847658408, "grad_norm": 2.2770034251326092, "learning_rate": 1.8586938918632385e-05, "loss": 0.6071, "step": 5506 }, { "epoch": 0.19619858560307818, "grad_norm": 1.835094491163608, "learning_rate": 1.8586347481797064e-05, "loss": 0.5467, "step": 5507 }, { "epoch": 0.19623421272957228, "grad_norm": 1.6927168249719002, "learning_rate": 1.858575593062852e-05, "loss": 0.5554, "step": 5508 }, { "epoch": 0.1962698398560664, "grad_norm": 1.6588860223767818, "learning_rate": 1.8585164265134636e-05, "loss": 0.5761, "step": 5509 }, { "epoch": 0.19630546698256052, "grad_norm": 1.268450967270528, "learning_rate": 1.858457248532329e-05, "loss": 0.5159, "step": 5510 }, { "epoch": 0.19634109410905465, "grad_norm": 1.2582489444366889, "learning_rate": 1.8583980591202362e-05, "loss": 0.5198, "step": 5511 }, { "epoch": 0.19637672123554875, "grad_norm": 1.2019906226690056, "learning_rate": 1.858338858277973e-05, "loss": 0.519, "step": 5512 }, { "epoch": 0.19641234836204285, "grad_norm": 1.7911463732586173, "learning_rate": 1.8582796460063282e-05, "loss": 0.5875, "step": 5513 }, { "epoch": 0.19644797548853699, "grad_norm": 1.7231220906711737, "learning_rate": 1.85822042230609e-05, "loss": 0.5869, "step": 5514 }, { "epoch": 0.1964836026150311, "grad_norm": 2.271996412636357, "learning_rate": 1.858161187178047e-05, "loss": 0.5075, "step": 5515 }, { "epoch": 0.1965192297415252, "grad_norm": 1.207416912562211, "learning_rate": 1.8581019406229875e-05, "loss": 0.5451, "step": 5516 }, { "epoch": 0.19655485686801932, "grad_norm": 1.309805625981826, "learning_rate": 1.8580426826417016e-05, "loss": 0.5412, "step": 5517 }, { "epoch": 0.19659048399451343, "grad_norm": 1.7715109932096547, "learning_rate": 1.8579834132349773e-05, "loss": 0.6012, "step": 5518 }, { "epoch": 0.19662611112100753, "grad_norm": 1.833620900972974, "learning_rate": 1.857924132403604e-05, "loss": 0.6865, "step": 5519 }, { "epoch": 0.19666173824750166, "grad_norm": 2.096790961978869, "learning_rate": 1.8578648401483713e-05, "loss": 0.5858, "step": 5520 }, { "epoch": 0.19669736537399576, "grad_norm": 1.223689777518327, "learning_rate": 1.857805536470069e-05, "loss": 0.4671, "step": 5521 }, { "epoch": 0.19673299250048987, "grad_norm": 1.6307652390397736, "learning_rate": 1.857746221369486e-05, "loss": 0.5889, "step": 5522 }, { "epoch": 0.196768619626984, "grad_norm": 1.279103819017557, "learning_rate": 1.857686894847413e-05, "loss": 0.5509, "step": 5523 }, { "epoch": 0.1968042467534781, "grad_norm": 1.2189754182385242, "learning_rate": 1.857627556904639e-05, "loss": 0.5017, "step": 5524 }, { "epoch": 0.1968398738799722, "grad_norm": 1.5299754549169609, "learning_rate": 1.857568207541955e-05, "loss": 0.5759, "step": 5525 }, { "epoch": 0.19687550100646634, "grad_norm": 1.8177993158418742, "learning_rate": 1.8575088467601504e-05, "loss": 0.6234, "step": 5526 }, { "epoch": 0.19691112813296044, "grad_norm": 1.4095848691988473, "learning_rate": 1.8574494745600166e-05, "loss": 0.5332, "step": 5527 }, { "epoch": 0.19694675525945454, "grad_norm": 2.109018868953655, "learning_rate": 1.8573900909423438e-05, "loss": 0.5912, "step": 5528 }, { "epoch": 0.19698238238594867, "grad_norm": 1.089612605434481, "learning_rate": 1.857330695907922e-05, "loss": 0.4925, "step": 5529 }, { "epoch": 0.19701800951244278, "grad_norm": 1.2739607165992035, "learning_rate": 1.8572712894575435e-05, "loss": 0.5632, "step": 5530 }, { "epoch": 0.19705363663893688, "grad_norm": 3.798922579795684, "learning_rate": 1.8572118715919983e-05, "loss": 0.7259, "step": 5531 }, { "epoch": 0.197089263765431, "grad_norm": 1.2344468388374774, "learning_rate": 1.857152442312078e-05, "loss": 0.5365, "step": 5532 }, { "epoch": 0.19712489089192511, "grad_norm": 1.2915490488122223, "learning_rate": 1.8570930016185737e-05, "loss": 0.5175, "step": 5533 }, { "epoch": 0.19716051801841922, "grad_norm": 1.528319460099897, "learning_rate": 1.8570335495122772e-05, "loss": 0.619, "step": 5534 }, { "epoch": 0.19719614514491335, "grad_norm": 1.5473391643001893, "learning_rate": 1.8569740859939797e-05, "loss": 0.5669, "step": 5535 }, { "epoch": 0.19723177227140745, "grad_norm": 1.2432524434962104, "learning_rate": 1.856914611064473e-05, "loss": 0.467, "step": 5536 }, { "epoch": 0.19726739939790155, "grad_norm": 1.2257282568344996, "learning_rate": 1.85685512472455e-05, "loss": 0.5269, "step": 5537 }, { "epoch": 0.19730302652439569, "grad_norm": 1.9230722696302187, "learning_rate": 1.8567956269750014e-05, "loss": 0.6004, "step": 5538 }, { "epoch": 0.1973386536508898, "grad_norm": 1.6989253673237272, "learning_rate": 1.8567361178166206e-05, "loss": 0.5893, "step": 5539 }, { "epoch": 0.1973742807773839, "grad_norm": 1.1722814360665044, "learning_rate": 1.8566765972501994e-05, "loss": 0.5586, "step": 5540 }, { "epoch": 0.19740990790387802, "grad_norm": 1.2305587213818179, "learning_rate": 1.8566170652765304e-05, "loss": 0.5326, "step": 5541 }, { "epoch": 0.19744553503037213, "grad_norm": 1.6719933320481604, "learning_rate": 1.8565575218964067e-05, "loss": 0.5451, "step": 5542 }, { "epoch": 0.19748116215686623, "grad_norm": 1.1480135630436057, "learning_rate": 1.8564979671106212e-05, "loss": 0.4942, "step": 5543 }, { "epoch": 0.19751678928336036, "grad_norm": 1.2056284590561244, "learning_rate": 1.856438400919966e-05, "loss": 0.5659, "step": 5544 }, { "epoch": 0.19755241640985446, "grad_norm": 1.5917402175648723, "learning_rate": 1.8563788233252352e-05, "loss": 0.7007, "step": 5545 }, { "epoch": 0.19758804353634857, "grad_norm": 1.1483468387364717, "learning_rate": 1.8563192343272215e-05, "loss": 0.4726, "step": 5546 }, { "epoch": 0.1976236706628427, "grad_norm": 1.5555358419664878, "learning_rate": 1.856259633926719e-05, "loss": 0.5745, "step": 5547 }, { "epoch": 0.1976592977893368, "grad_norm": 1.9181219654002601, "learning_rate": 1.856200022124521e-05, "loss": 0.6347, "step": 5548 }, { "epoch": 0.1976949249158309, "grad_norm": 1.2356959371643514, "learning_rate": 1.8561403989214207e-05, "loss": 0.4732, "step": 5549 }, { "epoch": 0.19773055204232504, "grad_norm": 1.795304387179085, "learning_rate": 1.856080764318213e-05, "loss": 0.6289, "step": 5550 }, { "epoch": 0.19776617916881914, "grad_norm": 1.6685458713552141, "learning_rate": 1.8560211183156918e-05, "loss": 0.6644, "step": 5551 }, { "epoch": 0.19780180629531324, "grad_norm": 1.3598257059493826, "learning_rate": 1.8559614609146508e-05, "loss": 0.5436, "step": 5552 }, { "epoch": 0.19783743342180737, "grad_norm": 1.2403596416259728, "learning_rate": 1.855901792115885e-05, "loss": 0.5243, "step": 5553 }, { "epoch": 0.19787306054830148, "grad_norm": 1.6032019330856129, "learning_rate": 1.8558421119201882e-05, "loss": 0.6772, "step": 5554 }, { "epoch": 0.19790868767479558, "grad_norm": 1.8119875302547417, "learning_rate": 1.8557824203283553e-05, "loss": 0.5971, "step": 5555 }, { "epoch": 0.1979443148012897, "grad_norm": 1.5253116936381563, "learning_rate": 1.8557227173411818e-05, "loss": 0.6181, "step": 5556 }, { "epoch": 0.19797994192778381, "grad_norm": 1.170719014747571, "learning_rate": 1.855663002959462e-05, "loss": 0.4862, "step": 5557 }, { "epoch": 0.19801556905427792, "grad_norm": 2.0278189984257913, "learning_rate": 1.8556032771839912e-05, "loss": 0.58, "step": 5558 }, { "epoch": 0.19805119618077205, "grad_norm": 1.0915759828939187, "learning_rate": 1.8555435400155648e-05, "loss": 0.4904, "step": 5559 }, { "epoch": 0.19808682330726615, "grad_norm": 1.0933560000072997, "learning_rate": 1.8554837914549783e-05, "loss": 0.52, "step": 5560 }, { "epoch": 0.19812245043376026, "grad_norm": 1.169623352936664, "learning_rate": 1.8554240315030268e-05, "loss": 0.5657, "step": 5561 }, { "epoch": 0.1981580775602544, "grad_norm": 1.7730299589511684, "learning_rate": 1.855364260160507e-05, "loss": 0.7788, "step": 5562 }, { "epoch": 0.1981937046867485, "grad_norm": 1.8980948049545234, "learning_rate": 1.8553044774282135e-05, "loss": 0.5767, "step": 5563 }, { "epoch": 0.1982293318132426, "grad_norm": 1.3168268824037133, "learning_rate": 1.8552446833069434e-05, "loss": 0.5042, "step": 5564 }, { "epoch": 0.19826495893973672, "grad_norm": 1.5989905422358877, "learning_rate": 1.855184877797493e-05, "loss": 0.576, "step": 5565 }, { "epoch": 0.19830058606623083, "grad_norm": 1.3066815300319374, "learning_rate": 1.855125060900658e-05, "loss": 0.5522, "step": 5566 }, { "epoch": 0.19833621319272493, "grad_norm": 1.0799644453567152, "learning_rate": 1.8550652326172344e-05, "loss": 0.4753, "step": 5567 }, { "epoch": 0.19837184031921906, "grad_norm": 1.0903706596153837, "learning_rate": 1.8550053929480202e-05, "loss": 0.5117, "step": 5568 }, { "epoch": 0.19840746744571316, "grad_norm": 1.6489995391988257, "learning_rate": 1.8549455418938114e-05, "loss": 0.6122, "step": 5569 }, { "epoch": 0.19844309457220727, "grad_norm": 1.8072121526008704, "learning_rate": 1.8548856794554053e-05, "loss": 0.5649, "step": 5570 }, { "epoch": 0.1984787216987014, "grad_norm": 1.2344562986327479, "learning_rate": 1.8548258056335987e-05, "loss": 0.5684, "step": 5571 }, { "epoch": 0.1985143488251955, "grad_norm": 2.1042681118693447, "learning_rate": 1.854765920429189e-05, "loss": 0.6066, "step": 5572 }, { "epoch": 0.1985499759516896, "grad_norm": 1.1533703657785492, "learning_rate": 1.8547060238429737e-05, "loss": 0.5276, "step": 5573 }, { "epoch": 0.19858560307818374, "grad_norm": 1.6999758481591491, "learning_rate": 1.8546461158757504e-05, "loss": 0.6184, "step": 5574 }, { "epoch": 0.19862123020467784, "grad_norm": 1.7988900184108747, "learning_rate": 1.8545861965283166e-05, "loss": 0.6604, "step": 5575 }, { "epoch": 0.19865685733117194, "grad_norm": 1.952481945204834, "learning_rate": 1.8545262658014702e-05, "loss": 0.5954, "step": 5576 }, { "epoch": 0.19869248445766607, "grad_norm": 1.6893043883417203, "learning_rate": 1.8544663236960094e-05, "loss": 0.6393, "step": 5577 }, { "epoch": 0.19872811158416018, "grad_norm": 1.765850033153062, "learning_rate": 1.854406370212732e-05, "loss": 0.699, "step": 5578 }, { "epoch": 0.19876373871065428, "grad_norm": 1.3308812794910223, "learning_rate": 1.8543464053524365e-05, "loss": 0.5321, "step": 5579 }, { "epoch": 0.1987993658371484, "grad_norm": 2.3170950080795065, "learning_rate": 1.854286429115922e-05, "loss": 0.5906, "step": 5580 }, { "epoch": 0.19883499296364252, "grad_norm": 1.0983562342758595, "learning_rate": 1.8542264415039864e-05, "loss": 0.5041, "step": 5581 }, { "epoch": 0.19887062009013662, "grad_norm": 1.1510570564794882, "learning_rate": 1.8541664425174287e-05, "loss": 0.4982, "step": 5582 }, { "epoch": 0.19890624721663075, "grad_norm": 1.1016909843786407, "learning_rate": 1.8541064321570477e-05, "loss": 0.5104, "step": 5583 }, { "epoch": 0.19894187434312485, "grad_norm": 1.1857596517375433, "learning_rate": 1.8540464104236428e-05, "loss": 0.5047, "step": 5584 }, { "epoch": 0.19897750146961896, "grad_norm": 1.2599290586572869, "learning_rate": 1.8539863773180125e-05, "loss": 0.5397, "step": 5585 }, { "epoch": 0.1990131285961131, "grad_norm": 1.2600377596752388, "learning_rate": 1.853926332840957e-05, "loss": 0.5174, "step": 5586 }, { "epoch": 0.1990487557226072, "grad_norm": 1.1421115718181867, "learning_rate": 1.8538662769932754e-05, "loss": 0.5032, "step": 5587 }, { "epoch": 0.1990843828491013, "grad_norm": 1.2016030279258798, "learning_rate": 1.8538062097757676e-05, "loss": 0.5596, "step": 5588 }, { "epoch": 0.19912000997559542, "grad_norm": 1.3070837105703734, "learning_rate": 1.8537461311892334e-05, "loss": 0.5262, "step": 5589 }, { "epoch": 0.19915563710208953, "grad_norm": 1.2733772628483309, "learning_rate": 1.8536860412344725e-05, "loss": 0.5328, "step": 5590 }, { "epoch": 0.19919126422858363, "grad_norm": 1.198603850595129, "learning_rate": 1.8536259399122855e-05, "loss": 0.5197, "step": 5591 }, { "epoch": 0.19922689135507776, "grad_norm": 1.8629455546562796, "learning_rate": 1.8535658272234724e-05, "loss": 0.5266, "step": 5592 }, { "epoch": 0.19926251848157187, "grad_norm": 1.212813227958278, "learning_rate": 1.8535057031688337e-05, "loss": 0.5348, "step": 5593 }, { "epoch": 0.19929814560806597, "grad_norm": 1.784973068083134, "learning_rate": 1.8534455677491697e-05, "loss": 0.6462, "step": 5594 }, { "epoch": 0.1993337727345601, "grad_norm": 1.7186070472397483, "learning_rate": 1.853385420965282e-05, "loss": 0.6556, "step": 5595 }, { "epoch": 0.1993693998610542, "grad_norm": 1.2471168793517873, "learning_rate": 1.8533252628179705e-05, "loss": 0.4889, "step": 5596 }, { "epoch": 0.1994050269875483, "grad_norm": 1.5736727134603363, "learning_rate": 1.853265093308037e-05, "loss": 0.6829, "step": 5597 }, { "epoch": 0.19944065411404244, "grad_norm": 1.7189824909682585, "learning_rate": 1.853204912436282e-05, "loss": 0.6478, "step": 5598 }, { "epoch": 0.19947628124053654, "grad_norm": 1.8885564006031548, "learning_rate": 1.8531447202035077e-05, "loss": 0.6366, "step": 5599 }, { "epoch": 0.19951190836703064, "grad_norm": 1.6009870083354234, "learning_rate": 1.853084516610515e-05, "loss": 0.617, "step": 5600 }, { "epoch": 0.19954753549352477, "grad_norm": 1.8371804055627612, "learning_rate": 1.8530243016581055e-05, "loss": 0.6233, "step": 5601 }, { "epoch": 0.19958316262001888, "grad_norm": 1.5055488643102823, "learning_rate": 1.8529640753470812e-05, "loss": 0.6061, "step": 5602 }, { "epoch": 0.19961878974651298, "grad_norm": 1.7373740193660816, "learning_rate": 1.8529038376782443e-05, "loss": 0.6171, "step": 5603 }, { "epoch": 0.1996544168730071, "grad_norm": 1.3022106347884923, "learning_rate": 1.8528435886523964e-05, "loss": 0.5318, "step": 5604 }, { "epoch": 0.19969004399950122, "grad_norm": 1.7123743599994026, "learning_rate": 1.85278332827034e-05, "loss": 0.6485, "step": 5605 }, { "epoch": 0.19972567112599535, "grad_norm": 1.9383357157536016, "learning_rate": 1.852723056532878e-05, "loss": 0.5546, "step": 5606 }, { "epoch": 0.19976129825248945, "grad_norm": 1.0637796325148121, "learning_rate": 1.852662773440812e-05, "loss": 0.519, "step": 5607 }, { "epoch": 0.19979692537898355, "grad_norm": 1.0586695629005363, "learning_rate": 1.8526024789949453e-05, "loss": 0.5255, "step": 5608 }, { "epoch": 0.19983255250547768, "grad_norm": 1.180894166544786, "learning_rate": 1.852542173196081e-05, "loss": 0.5201, "step": 5609 }, { "epoch": 0.1998681796319718, "grad_norm": 1.7455554223097098, "learning_rate": 1.8524818560450214e-05, "loss": 0.5617, "step": 5610 }, { "epoch": 0.1999038067584659, "grad_norm": 1.910482333240813, "learning_rate": 1.8524215275425698e-05, "loss": 0.6669, "step": 5611 }, { "epoch": 0.19993943388496002, "grad_norm": 1.2319181137350301, "learning_rate": 1.85236118768953e-05, "loss": 0.5212, "step": 5612 }, { "epoch": 0.19997506101145412, "grad_norm": 1.8638036559377373, "learning_rate": 1.8523008364867056e-05, "loss": 0.6526, "step": 5613 }, { "epoch": 0.20001068813794823, "grad_norm": 1.4860491696047589, "learning_rate": 1.8522404739348993e-05, "loss": 0.5813, "step": 5614 }, { "epoch": 0.20004631526444236, "grad_norm": 1.2884259661546598, "learning_rate": 1.8521801000349156e-05, "loss": 0.4967, "step": 5615 }, { "epoch": 0.20008194239093646, "grad_norm": 1.197719762815932, "learning_rate": 1.852119714787558e-05, "loss": 0.5462, "step": 5616 }, { "epoch": 0.20011756951743057, "grad_norm": 1.5492657035631792, "learning_rate": 1.8520593181936312e-05, "loss": 0.5786, "step": 5617 }, { "epoch": 0.2001531966439247, "grad_norm": 1.4998956657092668, "learning_rate": 1.851998910253939e-05, "loss": 0.68, "step": 5618 }, { "epoch": 0.2001888237704188, "grad_norm": 1.616491258322453, "learning_rate": 1.8519384909692854e-05, "loss": 0.629, "step": 5619 }, { "epoch": 0.2002244508969129, "grad_norm": 1.8994556239669125, "learning_rate": 1.851878060340476e-05, "loss": 0.6222, "step": 5620 }, { "epoch": 0.20026007802340703, "grad_norm": 2.0936045911730385, "learning_rate": 1.851817618368314e-05, "loss": 0.5809, "step": 5621 }, { "epoch": 0.20029570514990114, "grad_norm": 1.39053247108457, "learning_rate": 1.8517571650536056e-05, "loss": 0.6321, "step": 5622 }, { "epoch": 0.20033133227639524, "grad_norm": 1.5610189896639213, "learning_rate": 1.851696700397155e-05, "loss": 0.6084, "step": 5623 }, { "epoch": 0.20036695940288937, "grad_norm": 1.6195959540957456, "learning_rate": 1.8516362243997677e-05, "loss": 0.6224, "step": 5624 }, { "epoch": 0.20040258652938348, "grad_norm": 1.9944269174387896, "learning_rate": 1.8515757370622487e-05, "loss": 0.6373, "step": 5625 }, { "epoch": 0.20043821365587758, "grad_norm": 1.7383603594790003, "learning_rate": 1.8515152383854034e-05, "loss": 0.5712, "step": 5626 }, { "epoch": 0.2004738407823717, "grad_norm": 1.2724334563842146, "learning_rate": 1.8514547283700376e-05, "loss": 0.5625, "step": 5627 }, { "epoch": 0.2005094679088658, "grad_norm": 1.646863717669871, "learning_rate": 1.8513942070169572e-05, "loss": 0.5943, "step": 5628 }, { "epoch": 0.20054509503535992, "grad_norm": 1.303310205693263, "learning_rate": 1.8513336743269676e-05, "loss": 0.57, "step": 5629 }, { "epoch": 0.20058072216185405, "grad_norm": 1.3177127543072171, "learning_rate": 1.8512731303008752e-05, "loss": 0.5141, "step": 5630 }, { "epoch": 0.20061634928834815, "grad_norm": 1.5392512203785096, "learning_rate": 1.8512125749394857e-05, "loss": 0.5442, "step": 5631 }, { "epoch": 0.20065197641484225, "grad_norm": 1.2114528149403185, "learning_rate": 1.8511520082436064e-05, "loss": 0.4717, "step": 5632 }, { "epoch": 0.20068760354133638, "grad_norm": 1.6675131259561125, "learning_rate": 1.8510914302140428e-05, "loss": 0.5406, "step": 5633 }, { "epoch": 0.2007232306678305, "grad_norm": 1.618131287760513, "learning_rate": 1.851030840851602e-05, "loss": 0.6503, "step": 5634 }, { "epoch": 0.2007588577943246, "grad_norm": 2.4842474540631088, "learning_rate": 1.8509702401570906e-05, "loss": 0.6068, "step": 5635 }, { "epoch": 0.20079448492081872, "grad_norm": 1.2226206817203302, "learning_rate": 1.8509096281313158e-05, "loss": 0.5053, "step": 5636 }, { "epoch": 0.20083011204731283, "grad_norm": 1.2915272674064437, "learning_rate": 1.8508490047750845e-05, "loss": 0.5092, "step": 5637 }, { "epoch": 0.20086573917380693, "grad_norm": 1.7190707629169397, "learning_rate": 1.8507883700892042e-05, "loss": 0.6548, "step": 5638 }, { "epoch": 0.20090136630030106, "grad_norm": 1.0271877270905752, "learning_rate": 1.8507277240744818e-05, "loss": 0.4817, "step": 5639 }, { "epoch": 0.20093699342679516, "grad_norm": 1.74806767560868, "learning_rate": 1.8506670667317253e-05, "loss": 0.675, "step": 5640 }, { "epoch": 0.20097262055328927, "grad_norm": 1.5420639796889841, "learning_rate": 1.8506063980617423e-05, "loss": 0.5615, "step": 5641 }, { "epoch": 0.2010082476797834, "grad_norm": 1.6028638080383415, "learning_rate": 1.8505457180653405e-05, "loss": 0.6182, "step": 5642 }, { "epoch": 0.2010438748062775, "grad_norm": 1.4159967219454794, "learning_rate": 1.8504850267433278e-05, "loss": 0.5169, "step": 5643 }, { "epoch": 0.2010795019327716, "grad_norm": 1.1880911555474698, "learning_rate": 1.8504243240965128e-05, "loss": 0.5257, "step": 5644 }, { "epoch": 0.20111512905926573, "grad_norm": 1.7697558835109743, "learning_rate": 1.850363610125703e-05, "loss": 0.5935, "step": 5645 }, { "epoch": 0.20115075618575984, "grad_norm": 1.5856368978138893, "learning_rate": 1.850302884831708e-05, "loss": 0.6019, "step": 5646 }, { "epoch": 0.20118638331225394, "grad_norm": 1.2039150320184007, "learning_rate": 1.8502421482153354e-05, "loss": 0.4934, "step": 5647 }, { "epoch": 0.20122201043874807, "grad_norm": 1.55101851370928, "learning_rate": 1.8501814002773946e-05, "loss": 0.6142, "step": 5648 }, { "epoch": 0.20125763756524218, "grad_norm": 1.1192587645274972, "learning_rate": 1.8501206410186938e-05, "loss": 0.4727, "step": 5649 }, { "epoch": 0.20129326469173628, "grad_norm": 1.7078909838193228, "learning_rate": 1.8500598704400427e-05, "loss": 0.6132, "step": 5650 }, { "epoch": 0.2013288918182304, "grad_norm": 1.1874321576279105, "learning_rate": 1.84999908854225e-05, "loss": 0.5753, "step": 5651 }, { "epoch": 0.2013645189447245, "grad_norm": 1.0282408134403695, "learning_rate": 1.849938295326126e-05, "loss": 0.4676, "step": 5652 }, { "epoch": 0.20140014607121862, "grad_norm": 1.1434372445971748, "learning_rate": 1.849877490792479e-05, "loss": 0.5323, "step": 5653 }, { "epoch": 0.20143577319771275, "grad_norm": 1.1798826619692235, "learning_rate": 1.849816674942119e-05, "loss": 0.5321, "step": 5654 }, { "epoch": 0.20147140032420685, "grad_norm": 1.9105936867176976, "learning_rate": 1.8497558477758565e-05, "loss": 0.5061, "step": 5655 }, { "epoch": 0.20150702745070095, "grad_norm": 1.2145458711521564, "learning_rate": 1.8496950092945006e-05, "loss": 0.5368, "step": 5656 }, { "epoch": 0.20154265457719509, "grad_norm": 5.725704458913702, "learning_rate": 1.8496341594988617e-05, "loss": 0.5463, "step": 5657 }, { "epoch": 0.2015782817036892, "grad_norm": 1.2552411817799396, "learning_rate": 1.8495732983897504e-05, "loss": 0.5274, "step": 5658 }, { "epoch": 0.2016139088301833, "grad_norm": 1.8216723426628523, "learning_rate": 1.8495124259679767e-05, "loss": 0.5597, "step": 5659 }, { "epoch": 0.20164953595667742, "grad_norm": 1.0571422575620313, "learning_rate": 1.849451542234351e-05, "loss": 0.5045, "step": 5660 }, { "epoch": 0.20168516308317153, "grad_norm": 1.2101606834050387, "learning_rate": 1.8493906471896846e-05, "loss": 0.5182, "step": 5661 }, { "epoch": 0.20172079020966563, "grad_norm": 1.800674560851009, "learning_rate": 1.8493297408347882e-05, "loss": 0.633, "step": 5662 }, { "epoch": 0.20175641733615976, "grad_norm": 1.2787262188221005, "learning_rate": 1.849268823170472e-05, "loss": 0.5504, "step": 5663 }, { "epoch": 0.20179204446265386, "grad_norm": 1.181372036192449, "learning_rate": 1.8492078941975486e-05, "loss": 0.5066, "step": 5664 }, { "epoch": 0.20182767158914797, "grad_norm": 1.0596370812063147, "learning_rate": 1.849146953916828e-05, "loss": 0.5102, "step": 5665 }, { "epoch": 0.2018632987156421, "grad_norm": 1.945912485925167, "learning_rate": 1.8490860023291228e-05, "loss": 0.6031, "step": 5666 }, { "epoch": 0.2018989258421362, "grad_norm": 2.190655174825419, "learning_rate": 1.8490250394352433e-05, "loss": 0.6663, "step": 5667 }, { "epoch": 0.2019345529686303, "grad_norm": 1.8750874344742299, "learning_rate": 1.8489640652360025e-05, "loss": 0.4356, "step": 5668 }, { "epoch": 0.20197018009512444, "grad_norm": 1.1899221933658695, "learning_rate": 1.8489030797322112e-05, "loss": 0.5528, "step": 5669 }, { "epoch": 0.20200580722161854, "grad_norm": 1.2450988666599183, "learning_rate": 1.8488420829246826e-05, "loss": 0.5227, "step": 5670 }, { "epoch": 0.20204143434811264, "grad_norm": 1.225792490751891, "learning_rate": 1.848781074814228e-05, "loss": 0.5249, "step": 5671 }, { "epoch": 0.20207706147460677, "grad_norm": 1.9548018814552468, "learning_rate": 1.8487200554016602e-05, "loss": 0.6192, "step": 5672 }, { "epoch": 0.20211268860110088, "grad_norm": 1.1352294967319825, "learning_rate": 1.8486590246877916e-05, "loss": 0.527, "step": 5673 }, { "epoch": 0.20214831572759498, "grad_norm": 1.2858574232383346, "learning_rate": 1.848597982673435e-05, "loss": 0.5471, "step": 5674 }, { "epoch": 0.2021839428540891, "grad_norm": 1.6763140717304248, "learning_rate": 1.8485369293594033e-05, "loss": 0.6534, "step": 5675 }, { "epoch": 0.2022195699805832, "grad_norm": 1.5755737748317542, "learning_rate": 1.848475864746509e-05, "loss": 0.541, "step": 5676 }, { "epoch": 0.20225519710707732, "grad_norm": 1.7011915301947351, "learning_rate": 1.8484147888355656e-05, "loss": 0.5679, "step": 5677 }, { "epoch": 0.20229082423357145, "grad_norm": 1.2346499367714086, "learning_rate": 1.8483537016273866e-05, "loss": 0.549, "step": 5678 }, { "epoch": 0.20232645136006555, "grad_norm": 1.9343869760734835, "learning_rate": 1.8482926031227845e-05, "loss": 0.5655, "step": 5679 }, { "epoch": 0.20236207848655965, "grad_norm": 1.6482542050116171, "learning_rate": 1.848231493322574e-05, "loss": 0.6082, "step": 5680 }, { "epoch": 0.20239770561305379, "grad_norm": 1.2581625719282312, "learning_rate": 1.848170372227568e-05, "loss": 0.5324, "step": 5681 }, { "epoch": 0.2024333327395479, "grad_norm": 1.390815285263767, "learning_rate": 1.8481092398385805e-05, "loss": 0.5696, "step": 5682 }, { "epoch": 0.202468959866042, "grad_norm": 1.259517704317141, "learning_rate": 1.848048096156426e-05, "loss": 0.5827, "step": 5683 }, { "epoch": 0.20250458699253612, "grad_norm": 2.242733120846528, "learning_rate": 1.8479869411819183e-05, "loss": 0.6231, "step": 5684 }, { "epoch": 0.20254021411903023, "grad_norm": 1.750905987489473, "learning_rate": 1.8479257749158717e-05, "loss": 0.6326, "step": 5685 }, { "epoch": 0.20257584124552433, "grad_norm": 1.440109250879688, "learning_rate": 1.847864597359101e-05, "loss": 0.5913, "step": 5686 }, { "epoch": 0.20261146837201846, "grad_norm": 1.6955623150050092, "learning_rate": 1.8478034085124204e-05, "loss": 0.5399, "step": 5687 }, { "epoch": 0.20264709549851256, "grad_norm": 1.8346421094085854, "learning_rate": 1.8477422083766446e-05, "loss": 0.6836, "step": 5688 }, { "epoch": 0.20268272262500667, "grad_norm": 1.1393478467956162, "learning_rate": 1.8476809969525892e-05, "loss": 0.5074, "step": 5689 }, { "epoch": 0.2027183497515008, "grad_norm": 1.586814735474251, "learning_rate": 1.8476197742410683e-05, "loss": 0.6054, "step": 5690 }, { "epoch": 0.2027539768779949, "grad_norm": 2.0741083377885876, "learning_rate": 1.847558540242898e-05, "loss": 0.4943, "step": 5691 }, { "epoch": 0.202789604004489, "grad_norm": 1.3777909087612552, "learning_rate": 1.847497294958893e-05, "loss": 0.5262, "step": 5692 }, { "epoch": 0.20282523113098314, "grad_norm": 1.1092061503959958, "learning_rate": 1.8474360383898694e-05, "loss": 0.5225, "step": 5693 }, { "epoch": 0.20286085825747724, "grad_norm": 1.6058928945704658, "learning_rate": 1.8473747705366427e-05, "loss": 0.487, "step": 5694 }, { "epoch": 0.20289648538397134, "grad_norm": 1.1669305440708344, "learning_rate": 1.8473134914000284e-05, "loss": 0.5235, "step": 5695 }, { "epoch": 0.20293211251046547, "grad_norm": 1.7665958332402254, "learning_rate": 1.847252200980843e-05, "loss": 0.5949, "step": 5696 }, { "epoch": 0.20296773963695958, "grad_norm": 1.191271593679942, "learning_rate": 1.8471908992799024e-05, "loss": 0.5016, "step": 5697 }, { "epoch": 0.20300336676345368, "grad_norm": 1.3369543567541329, "learning_rate": 1.8471295862980226e-05, "loss": 0.5754, "step": 5698 }, { "epoch": 0.2030389938899478, "grad_norm": 1.1661261073834437, "learning_rate": 1.8470682620360202e-05, "loss": 0.485, "step": 5699 }, { "epoch": 0.20307462101644191, "grad_norm": 1.796332198894787, "learning_rate": 1.847006926494712e-05, "loss": 0.6849, "step": 5700 }, { "epoch": 0.20311024814293605, "grad_norm": 1.3456289703371034, "learning_rate": 1.8469455796749148e-05, "loss": 0.5514, "step": 5701 }, { "epoch": 0.20314587526943015, "grad_norm": 1.4479847940486805, "learning_rate": 1.8468842215774447e-05, "loss": 0.6303, "step": 5702 }, { "epoch": 0.20318150239592425, "grad_norm": 1.234587463545484, "learning_rate": 1.8468228522031197e-05, "loss": 0.4872, "step": 5703 }, { "epoch": 0.20321712952241838, "grad_norm": 1.4870061781677457, "learning_rate": 1.8467614715527563e-05, "loss": 0.5615, "step": 5704 }, { "epoch": 0.20325275664891249, "grad_norm": 1.0763402224209193, "learning_rate": 1.8467000796271717e-05, "loss": 0.4769, "step": 5705 }, { "epoch": 0.2032883837754066, "grad_norm": 1.6031600112496898, "learning_rate": 1.8466386764271843e-05, "loss": 0.545, "step": 5706 }, { "epoch": 0.20332401090190072, "grad_norm": 1.1915307059463225, "learning_rate": 1.846577261953611e-05, "loss": 0.5774, "step": 5707 }, { "epoch": 0.20335963802839482, "grad_norm": 1.147144712422336, "learning_rate": 1.8465158362072696e-05, "loss": 0.522, "step": 5708 }, { "epoch": 0.20339526515488893, "grad_norm": 1.1734637813088993, "learning_rate": 1.8464543991889784e-05, "loss": 0.5412, "step": 5709 }, { "epoch": 0.20343089228138306, "grad_norm": 1.8592869322208807, "learning_rate": 1.846392950899555e-05, "loss": 0.6548, "step": 5710 }, { "epoch": 0.20346651940787716, "grad_norm": 1.2817367385755996, "learning_rate": 1.846331491339818e-05, "loss": 0.5517, "step": 5711 }, { "epoch": 0.20350214653437126, "grad_norm": 1.2186348366108026, "learning_rate": 1.8462700205105855e-05, "loss": 0.5105, "step": 5712 }, { "epoch": 0.2035377736608654, "grad_norm": 1.2584807723307934, "learning_rate": 1.846208538412676e-05, "loss": 0.4447, "step": 5713 }, { "epoch": 0.2035734007873595, "grad_norm": 1.5637227860190082, "learning_rate": 1.8461470450469086e-05, "loss": 0.6154, "step": 5714 }, { "epoch": 0.2036090279138536, "grad_norm": 1.1841801081933643, "learning_rate": 1.8460855404141024e-05, "loss": 0.5828, "step": 5715 }, { "epoch": 0.20364465504034773, "grad_norm": 1.0957754740655126, "learning_rate": 1.846024024515075e-05, "loss": 0.4907, "step": 5716 }, { "epoch": 0.20368028216684184, "grad_norm": 1.7059118468578336, "learning_rate": 1.8459624973506467e-05, "loss": 0.6764, "step": 5717 }, { "epoch": 0.20371590929333594, "grad_norm": 1.1442352565427087, "learning_rate": 1.8459009589216364e-05, "loss": 0.5541, "step": 5718 }, { "epoch": 0.20375153641983007, "grad_norm": 1.5458210228092986, "learning_rate": 1.845839409228864e-05, "loss": 0.6396, "step": 5719 }, { "epoch": 0.20378716354632417, "grad_norm": 1.2761667143790765, "learning_rate": 1.8457778482731484e-05, "loss": 0.5249, "step": 5720 }, { "epoch": 0.20382279067281828, "grad_norm": 1.6831717209018995, "learning_rate": 1.8457162760553092e-05, "loss": 0.5848, "step": 5721 }, { "epoch": 0.2038584177993124, "grad_norm": 1.266212042577254, "learning_rate": 1.845654692576167e-05, "loss": 0.5181, "step": 5722 }, { "epoch": 0.2038940449258065, "grad_norm": 1.1694685003385499, "learning_rate": 1.8455930978365415e-05, "loss": 0.5278, "step": 5723 }, { "epoch": 0.20392967205230061, "grad_norm": 1.1770145809825057, "learning_rate": 1.845531491837253e-05, "loss": 0.552, "step": 5724 }, { "epoch": 0.20396529917879475, "grad_norm": 1.5378605334158457, "learning_rate": 1.845469874579121e-05, "loss": 0.617, "step": 5725 }, { "epoch": 0.20400092630528885, "grad_norm": 1.1394144081230242, "learning_rate": 1.8454082460629677e-05, "loss": 0.5354, "step": 5726 }, { "epoch": 0.20403655343178295, "grad_norm": 1.6061517389328803, "learning_rate": 1.8453466062896122e-05, "loss": 0.602, "step": 5727 }, { "epoch": 0.20407218055827708, "grad_norm": 1.274070130577499, "learning_rate": 1.8452849552598757e-05, "loss": 0.5657, "step": 5728 }, { "epoch": 0.2041078076847712, "grad_norm": 1.5233198559955754, "learning_rate": 1.8452232929745793e-05, "loss": 0.5349, "step": 5729 }, { "epoch": 0.2041434348112653, "grad_norm": 1.6001248791858003, "learning_rate": 1.845161619434544e-05, "loss": 0.5837, "step": 5730 }, { "epoch": 0.20417906193775942, "grad_norm": 1.2142709022415936, "learning_rate": 1.845099934640591e-05, "loss": 0.4798, "step": 5731 }, { "epoch": 0.20421468906425352, "grad_norm": 1.1186054329422181, "learning_rate": 1.8450382385935414e-05, "loss": 0.4701, "step": 5732 }, { "epoch": 0.20425031619074763, "grad_norm": 1.498481462512785, "learning_rate": 1.8449765312942172e-05, "loss": 0.5913, "step": 5733 }, { "epoch": 0.20428594331724176, "grad_norm": 1.7680422106575793, "learning_rate": 1.84491481274344e-05, "loss": 0.5801, "step": 5734 }, { "epoch": 0.20432157044373586, "grad_norm": 1.2105604541386572, "learning_rate": 1.8448530829420313e-05, "loss": 0.514, "step": 5735 }, { "epoch": 0.20435719757022996, "grad_norm": 1.2047124531583966, "learning_rate": 1.8447913418908136e-05, "loss": 0.4972, "step": 5736 }, { "epoch": 0.2043928246967241, "grad_norm": 1.5068492753003642, "learning_rate": 1.8447295895906086e-05, "loss": 0.5651, "step": 5737 }, { "epoch": 0.2044284518232182, "grad_norm": 1.5224976835548574, "learning_rate": 1.8446678260422388e-05, "loss": 0.5309, "step": 5738 }, { "epoch": 0.2044640789497123, "grad_norm": 1.1747885254478387, "learning_rate": 1.844606051246526e-05, "loss": 0.5164, "step": 5739 }, { "epoch": 0.20449970607620643, "grad_norm": 1.222652757333165, "learning_rate": 1.8445442652042937e-05, "loss": 0.5259, "step": 5740 }, { "epoch": 0.20453533320270054, "grad_norm": 1.8855503930447075, "learning_rate": 1.844482467916364e-05, "loss": 0.6446, "step": 5741 }, { "epoch": 0.20457096032919464, "grad_norm": 1.610691796368329, "learning_rate": 1.84442065938356e-05, "loss": 0.6209, "step": 5742 }, { "epoch": 0.20460658745568877, "grad_norm": 1.6419379366216382, "learning_rate": 1.844358839606705e-05, "loss": 0.5834, "step": 5743 }, { "epoch": 0.20464221458218287, "grad_norm": 1.4517654903797366, "learning_rate": 1.8442970085866214e-05, "loss": 0.5706, "step": 5744 }, { "epoch": 0.20467784170867698, "grad_norm": 4.702367320693248, "learning_rate": 1.8442351663241337e-05, "loss": 0.6725, "step": 5745 }, { "epoch": 0.2047134688351711, "grad_norm": 1.7327620473515004, "learning_rate": 1.844173312820064e-05, "loss": 0.7174, "step": 5746 }, { "epoch": 0.2047490959616652, "grad_norm": 1.242572690572082, "learning_rate": 1.844111448075237e-05, "loss": 0.4921, "step": 5747 }, { "epoch": 0.20478472308815932, "grad_norm": 1.248290064132767, "learning_rate": 1.8440495720904758e-05, "loss": 0.4861, "step": 5748 }, { "epoch": 0.20482035021465345, "grad_norm": 1.163611879217761, "learning_rate": 1.8439876848666048e-05, "loss": 0.5463, "step": 5749 }, { "epoch": 0.20485597734114755, "grad_norm": 1.9201179381073763, "learning_rate": 1.8439257864044476e-05, "loss": 0.6517, "step": 5750 }, { "epoch": 0.20489160446764165, "grad_norm": 1.16203452633513, "learning_rate": 1.8438638767048287e-05, "loss": 0.4995, "step": 5751 }, { "epoch": 0.20492723159413578, "grad_norm": 1.7960591560965316, "learning_rate": 1.8438019557685727e-05, "loss": 0.6428, "step": 5752 }, { "epoch": 0.2049628587206299, "grad_norm": 1.2643446400292448, "learning_rate": 1.843740023596504e-05, "loss": 0.4949, "step": 5753 }, { "epoch": 0.204998485847124, "grad_norm": 1.6716257903186875, "learning_rate": 1.8436780801894468e-05, "loss": 0.6394, "step": 5754 }, { "epoch": 0.20503411297361812, "grad_norm": 2.2400250258999956, "learning_rate": 1.8436161255482265e-05, "loss": 0.6043, "step": 5755 }, { "epoch": 0.20506974010011222, "grad_norm": 1.2197424243128416, "learning_rate": 1.8435541596736675e-05, "loss": 0.547, "step": 5756 }, { "epoch": 0.20510536722660633, "grad_norm": 2.2672231768429745, "learning_rate": 1.8434921825665958e-05, "loss": 0.6534, "step": 5757 }, { "epoch": 0.20514099435310046, "grad_norm": 1.8670097885832122, "learning_rate": 1.8434301942278357e-05, "loss": 0.6269, "step": 5758 }, { "epoch": 0.20517662147959456, "grad_norm": 1.2596540987887734, "learning_rate": 1.843368194658213e-05, "loss": 0.6002, "step": 5759 }, { "epoch": 0.20521224860608867, "grad_norm": 1.4639438922048857, "learning_rate": 1.8433061838585537e-05, "loss": 0.564, "step": 5760 }, { "epoch": 0.2052478757325828, "grad_norm": 1.3508984369754686, "learning_rate": 1.8432441618296828e-05, "loss": 0.5646, "step": 5761 }, { "epoch": 0.2052835028590769, "grad_norm": 1.2692874684529665, "learning_rate": 1.8431821285724263e-05, "loss": 0.5404, "step": 5762 }, { "epoch": 0.205319129985571, "grad_norm": 1.8002323729577765, "learning_rate": 1.843120084087611e-05, "loss": 0.6223, "step": 5763 }, { "epoch": 0.20535475711206513, "grad_norm": 1.8233813463093855, "learning_rate": 1.8430580283760623e-05, "loss": 0.6529, "step": 5764 }, { "epoch": 0.20539038423855924, "grad_norm": 1.2401369546007541, "learning_rate": 1.8429959614386064e-05, "loss": 0.5611, "step": 5765 }, { "epoch": 0.20542601136505334, "grad_norm": 1.69521952904439, "learning_rate": 1.84293388327607e-05, "loss": 0.5241, "step": 5766 }, { "epoch": 0.20546163849154747, "grad_norm": 1.5028852765185212, "learning_rate": 1.84287179388928e-05, "loss": 0.5359, "step": 5767 }, { "epoch": 0.20549726561804157, "grad_norm": 1.7358035477912483, "learning_rate": 1.842809693279063e-05, "loss": 0.6196, "step": 5768 }, { "epoch": 0.20553289274453568, "grad_norm": 1.1178386195147678, "learning_rate": 1.8427475814462458e-05, "loss": 0.5203, "step": 5769 }, { "epoch": 0.2055685198710298, "grad_norm": 1.7842229492854125, "learning_rate": 1.8426854583916554e-05, "loss": 0.6304, "step": 5770 }, { "epoch": 0.2056041469975239, "grad_norm": 1.3597675105730425, "learning_rate": 1.8426233241161193e-05, "loss": 0.4852, "step": 5771 }, { "epoch": 0.20563977412401802, "grad_norm": 2.2189682761692042, "learning_rate": 1.8425611786204647e-05, "loss": 0.6635, "step": 5772 }, { "epoch": 0.20567540125051215, "grad_norm": 1.9170867471149782, "learning_rate": 1.842499021905519e-05, "loss": 0.6026, "step": 5773 }, { "epoch": 0.20571102837700625, "grad_norm": 1.6136488632137698, "learning_rate": 1.8424368539721098e-05, "loss": 0.6364, "step": 5774 }, { "epoch": 0.20574665550350035, "grad_norm": 1.9530050136971524, "learning_rate": 1.842374674821065e-05, "loss": 0.683, "step": 5775 }, { "epoch": 0.20578228262999448, "grad_norm": 1.239739961899955, "learning_rate": 1.842312484453213e-05, "loss": 0.5255, "step": 5776 }, { "epoch": 0.2058179097564886, "grad_norm": 1.8418230809862222, "learning_rate": 1.8422502828693814e-05, "loss": 0.5293, "step": 5777 }, { "epoch": 0.2058535368829827, "grad_norm": 1.1362918961173298, "learning_rate": 1.8421880700703983e-05, "loss": 0.5161, "step": 5778 }, { "epoch": 0.20588916400947682, "grad_norm": 1.6286714852822513, "learning_rate": 1.8421258460570927e-05, "loss": 0.6174, "step": 5779 }, { "epoch": 0.20592479113597092, "grad_norm": 1.2026565425212126, "learning_rate": 1.8420636108302926e-05, "loss": 0.5264, "step": 5780 }, { "epoch": 0.20596041826246503, "grad_norm": 1.0880730586472613, "learning_rate": 1.8420013643908272e-05, "loss": 0.4961, "step": 5781 }, { "epoch": 0.20599604538895916, "grad_norm": 1.7499859398034354, "learning_rate": 1.8419391067395248e-05, "loss": 0.6055, "step": 5782 }, { "epoch": 0.20603167251545326, "grad_norm": 1.1408616438283181, "learning_rate": 1.841876837877215e-05, "loss": 0.5278, "step": 5783 }, { "epoch": 0.20606729964194737, "grad_norm": 1.8753126084344836, "learning_rate": 1.841814557804727e-05, "loss": 0.6377, "step": 5784 }, { "epoch": 0.2061029267684415, "grad_norm": 1.18217947144773, "learning_rate": 1.8417522665228894e-05, "loss": 0.5193, "step": 5785 }, { "epoch": 0.2061385538949356, "grad_norm": 1.223759274035864, "learning_rate": 1.8416899640325318e-05, "loss": 0.5053, "step": 5786 }, { "epoch": 0.2061741810214297, "grad_norm": 1.1448284631911394, "learning_rate": 1.8416276503344843e-05, "loss": 0.51, "step": 5787 }, { "epoch": 0.20620980814792383, "grad_norm": 1.0588983336331192, "learning_rate": 1.841565325429576e-05, "loss": 0.4701, "step": 5788 }, { "epoch": 0.20624543527441794, "grad_norm": 1.1324855325811047, "learning_rate": 1.8415029893186375e-05, "loss": 0.4892, "step": 5789 }, { "epoch": 0.20628106240091204, "grad_norm": 1.2693751464775538, "learning_rate": 1.8414406420024984e-05, "loss": 0.5464, "step": 5790 }, { "epoch": 0.20631668952740617, "grad_norm": 1.6651747385265645, "learning_rate": 1.8413782834819895e-05, "loss": 0.6323, "step": 5791 }, { "epoch": 0.20635231665390028, "grad_norm": 1.0914971428022453, "learning_rate": 1.84131591375794e-05, "loss": 0.4792, "step": 5792 }, { "epoch": 0.2063879437803944, "grad_norm": 1.1129009020709741, "learning_rate": 1.8412535328311813e-05, "loss": 0.5544, "step": 5793 }, { "epoch": 0.2064235709068885, "grad_norm": 1.6413904532288863, "learning_rate": 1.841191140702544e-05, "loss": 0.59, "step": 5794 }, { "epoch": 0.2064591980333826, "grad_norm": 1.1805684676990924, "learning_rate": 1.8411287373728584e-05, "loss": 0.5002, "step": 5795 }, { "epoch": 0.20649482515987674, "grad_norm": 1.265826440902813, "learning_rate": 1.8410663228429558e-05, "loss": 0.5267, "step": 5796 }, { "epoch": 0.20653045228637085, "grad_norm": 1.1763610473021755, "learning_rate": 1.8410038971136673e-05, "loss": 0.5554, "step": 5797 }, { "epoch": 0.20656607941286495, "grad_norm": 1.75525766549143, "learning_rate": 1.8409414601858243e-05, "loss": 0.6899, "step": 5798 }, { "epoch": 0.20660170653935908, "grad_norm": 1.665312113292891, "learning_rate": 1.840879012060258e-05, "loss": 0.6524, "step": 5799 }, { "epoch": 0.20663733366585318, "grad_norm": 1.688924938863117, "learning_rate": 1.8408165527377995e-05, "loss": 0.5725, "step": 5800 }, { "epoch": 0.2066729607923473, "grad_norm": 2.089762710248748, "learning_rate": 1.840754082219281e-05, "loss": 0.7117, "step": 5801 }, { "epoch": 0.20670858791884142, "grad_norm": 1.281726187246558, "learning_rate": 1.8406916005055345e-05, "loss": 0.5495, "step": 5802 }, { "epoch": 0.20674421504533552, "grad_norm": 1.865586695399867, "learning_rate": 1.8406291075973914e-05, "loss": 0.6625, "step": 5803 }, { "epoch": 0.20677984217182963, "grad_norm": 1.2451121236079836, "learning_rate": 1.8405666034956842e-05, "loss": 0.5464, "step": 5804 }, { "epoch": 0.20681546929832376, "grad_norm": 1.1400068807441752, "learning_rate": 1.8405040882012454e-05, "loss": 0.5535, "step": 5805 }, { "epoch": 0.20685109642481786, "grad_norm": 1.1004552876988924, "learning_rate": 1.8404415617149073e-05, "loss": 0.5339, "step": 5806 }, { "epoch": 0.20688672355131196, "grad_norm": 1.1750637690346981, "learning_rate": 1.8403790240375022e-05, "loss": 0.4621, "step": 5807 }, { "epoch": 0.2069223506778061, "grad_norm": 1.2646536516347824, "learning_rate": 1.840316475169863e-05, "loss": 0.5128, "step": 5808 }, { "epoch": 0.2069579778043002, "grad_norm": 1.2135666763297983, "learning_rate": 1.8402539151128224e-05, "loss": 0.5809, "step": 5809 }, { "epoch": 0.2069936049307943, "grad_norm": 1.6368189295039306, "learning_rate": 1.840191343867214e-05, "loss": 0.5646, "step": 5810 }, { "epoch": 0.20702923205728843, "grad_norm": 1.2506503823965662, "learning_rate": 1.84012876143387e-05, "loss": 0.5162, "step": 5811 }, { "epoch": 0.20706485918378253, "grad_norm": 2.0017437930570217, "learning_rate": 1.840066167813625e-05, "loss": 0.5953, "step": 5812 }, { "epoch": 0.20710048631027664, "grad_norm": 1.1811969214929665, "learning_rate": 1.8400035630073114e-05, "loss": 0.4931, "step": 5813 }, { "epoch": 0.20713611343677077, "grad_norm": 1.1865790653089279, "learning_rate": 1.839940947015763e-05, "loss": 0.5527, "step": 5814 }, { "epoch": 0.20717174056326487, "grad_norm": 1.3318932408596038, "learning_rate": 1.8398783198398146e-05, "loss": 0.5758, "step": 5815 }, { "epoch": 0.20720736768975898, "grad_norm": 1.087055688791588, "learning_rate": 1.8398156814802987e-05, "loss": 0.4889, "step": 5816 }, { "epoch": 0.2072429948162531, "grad_norm": 1.1553576626650532, "learning_rate": 1.83975303193805e-05, "loss": 0.5111, "step": 5817 }, { "epoch": 0.2072786219427472, "grad_norm": 1.0822069283346463, "learning_rate": 1.839690371213903e-05, "loss": 0.4667, "step": 5818 }, { "epoch": 0.2073142490692413, "grad_norm": 1.7710381246996962, "learning_rate": 1.8396276993086916e-05, "loss": 0.5637, "step": 5819 }, { "epoch": 0.20734987619573544, "grad_norm": 1.2211645909561748, "learning_rate": 1.839565016223251e-05, "loss": 0.5272, "step": 5820 }, { "epoch": 0.20738550332222955, "grad_norm": 2.1064501230022405, "learning_rate": 1.8395023219584146e-05, "loss": 0.6252, "step": 5821 }, { "epoch": 0.20742113044872365, "grad_norm": 1.2466549004057659, "learning_rate": 1.8394396165150185e-05, "loss": 0.5103, "step": 5822 }, { "epoch": 0.20745675757521778, "grad_norm": 1.8011128409167791, "learning_rate": 1.839376899893897e-05, "loss": 0.6642, "step": 5823 }, { "epoch": 0.20749238470171188, "grad_norm": 1.1923462271907075, "learning_rate": 1.8393141720958856e-05, "loss": 0.533, "step": 5824 }, { "epoch": 0.207528011828206, "grad_norm": 1.2853023534608194, "learning_rate": 1.8392514331218196e-05, "loss": 0.477, "step": 5825 }, { "epoch": 0.20756363895470012, "grad_norm": 1.168152324856355, "learning_rate": 1.8391886829725334e-05, "loss": 0.482, "step": 5826 }, { "epoch": 0.20759926608119422, "grad_norm": 1.3180506967121957, "learning_rate": 1.8391259216488637e-05, "loss": 0.4904, "step": 5827 }, { "epoch": 0.20763489320768833, "grad_norm": 1.7220511906517464, "learning_rate": 1.839063149151646e-05, "loss": 0.5919, "step": 5828 }, { "epoch": 0.20767052033418246, "grad_norm": 1.151308219792161, "learning_rate": 1.839000365481716e-05, "loss": 0.5118, "step": 5829 }, { "epoch": 0.20770614746067656, "grad_norm": 1.3095178287218925, "learning_rate": 1.8389375706399097e-05, "loss": 0.5267, "step": 5830 }, { "epoch": 0.20774177458717066, "grad_norm": 1.2443710360235845, "learning_rate": 1.838874764627063e-05, "loss": 0.4958, "step": 5831 }, { "epoch": 0.2077774017136648, "grad_norm": 1.071330588733646, "learning_rate": 1.8388119474440125e-05, "loss": 0.4768, "step": 5832 }, { "epoch": 0.2078130288401589, "grad_norm": 1.1142937061069502, "learning_rate": 1.838749119091595e-05, "loss": 0.4487, "step": 5833 }, { "epoch": 0.207848655966653, "grad_norm": 1.1229450247784944, "learning_rate": 1.8386862795706462e-05, "loss": 0.5236, "step": 5834 }, { "epoch": 0.20788428309314713, "grad_norm": 1.272793665144349, "learning_rate": 1.8386234288820034e-05, "loss": 0.5613, "step": 5835 }, { "epoch": 0.20791991021964124, "grad_norm": 1.7599474531386674, "learning_rate": 1.8385605670265038e-05, "loss": 0.6921, "step": 5836 }, { "epoch": 0.20795553734613534, "grad_norm": 1.7967820594487842, "learning_rate": 1.8384976940049842e-05, "loss": 0.6494, "step": 5837 }, { "epoch": 0.20799116447262947, "grad_norm": 1.1940262941939774, "learning_rate": 1.8384348098182815e-05, "loss": 0.5265, "step": 5838 }, { "epoch": 0.20802679159912357, "grad_norm": 1.5889356693910694, "learning_rate": 1.8383719144672333e-05, "loss": 0.6395, "step": 5839 }, { "epoch": 0.20806241872561768, "grad_norm": 1.6070227654350644, "learning_rate": 1.8383090079526774e-05, "loss": 0.6422, "step": 5840 }, { "epoch": 0.2080980458521118, "grad_norm": 1.1912074078259767, "learning_rate": 1.8382460902754508e-05, "loss": 0.4911, "step": 5841 }, { "epoch": 0.2081336729786059, "grad_norm": 1.2693512339388169, "learning_rate": 1.8381831614363915e-05, "loss": 0.5688, "step": 5842 }, { "epoch": 0.2081693001051, "grad_norm": 1.1761366301448806, "learning_rate": 1.838120221436338e-05, "loss": 0.5241, "step": 5843 }, { "epoch": 0.20820492723159414, "grad_norm": 1.196316422512387, "learning_rate": 1.8380572702761276e-05, "loss": 0.5284, "step": 5844 }, { "epoch": 0.20824055435808825, "grad_norm": 1.114859144677694, "learning_rate": 1.8379943079565987e-05, "loss": 0.5104, "step": 5845 }, { "epoch": 0.20827618148458235, "grad_norm": 1.7714476685378977, "learning_rate": 1.83793133447859e-05, "loss": 0.6951, "step": 5846 }, { "epoch": 0.20831180861107648, "grad_norm": 1.244625579241763, "learning_rate": 1.83786834984294e-05, "loss": 0.4749, "step": 5847 }, { "epoch": 0.20834743573757059, "grad_norm": 1.108219447332221, "learning_rate": 1.8378053540504874e-05, "loss": 0.4553, "step": 5848 }, { "epoch": 0.2083830628640647, "grad_norm": 2.084475474410527, "learning_rate": 1.8377423471020708e-05, "loss": 0.6802, "step": 5849 }, { "epoch": 0.20841868999055882, "grad_norm": 1.4040038933133518, "learning_rate": 1.8376793289985295e-05, "loss": 0.5606, "step": 5850 }, { "epoch": 0.20845431711705292, "grad_norm": 1.489528543994619, "learning_rate": 1.837616299740702e-05, "loss": 0.5394, "step": 5851 }, { "epoch": 0.20848994424354703, "grad_norm": 1.3990999574272518, "learning_rate": 1.8375532593294282e-05, "loss": 0.5295, "step": 5852 }, { "epoch": 0.20852557137004116, "grad_norm": 1.0755415394455525, "learning_rate": 1.8374902077655474e-05, "loss": 0.4563, "step": 5853 }, { "epoch": 0.20856119849653526, "grad_norm": 2.00797328594756, "learning_rate": 1.837427145049899e-05, "loss": 0.6629, "step": 5854 }, { "epoch": 0.20859682562302936, "grad_norm": 1.69121982672758, "learning_rate": 1.837364071183323e-05, "loss": 0.567, "step": 5855 }, { "epoch": 0.2086324527495235, "grad_norm": 1.434650924016597, "learning_rate": 1.837300986166659e-05, "loss": 0.5298, "step": 5856 }, { "epoch": 0.2086680798760176, "grad_norm": 1.1233407300663782, "learning_rate": 1.8372378900007476e-05, "loss": 0.5064, "step": 5857 }, { "epoch": 0.2087037070025117, "grad_norm": 1.64416360275373, "learning_rate": 1.837174782686428e-05, "loss": 0.6432, "step": 5858 }, { "epoch": 0.20873933412900583, "grad_norm": 1.8585240275392025, "learning_rate": 1.837111664224541e-05, "loss": 0.6963, "step": 5859 }, { "epoch": 0.20877496125549994, "grad_norm": 1.7366743229435897, "learning_rate": 1.837048534615927e-05, "loss": 0.5922, "step": 5860 }, { "epoch": 0.20881058838199404, "grad_norm": 1.945171961953869, "learning_rate": 1.836985393861427e-05, "loss": 0.5792, "step": 5861 }, { "epoch": 0.20884621550848817, "grad_norm": 1.8542279055661568, "learning_rate": 1.836922241961881e-05, "loss": 0.5446, "step": 5862 }, { "epoch": 0.20888184263498227, "grad_norm": 1.4290955214074579, "learning_rate": 1.836859078918131e-05, "loss": 0.495, "step": 5863 }, { "epoch": 0.20891746976147638, "grad_norm": 2.1341930487016443, "learning_rate": 1.8367959047310174e-05, "loss": 0.5803, "step": 5864 }, { "epoch": 0.2089530968879705, "grad_norm": 1.660117819670938, "learning_rate": 1.8367327194013808e-05, "loss": 0.553, "step": 5865 }, { "epoch": 0.2089887240144646, "grad_norm": 1.4925810538119069, "learning_rate": 1.836669522930064e-05, "loss": 0.606, "step": 5866 }, { "epoch": 0.20902435114095871, "grad_norm": 1.5117194933249658, "learning_rate": 1.8366063153179074e-05, "loss": 0.6121, "step": 5867 }, { "epoch": 0.20905997826745285, "grad_norm": 1.2178219323147497, "learning_rate": 1.8365430965657527e-05, "loss": 0.5027, "step": 5868 }, { "epoch": 0.20909560539394695, "grad_norm": 1.9634097012787377, "learning_rate": 1.836479866674442e-05, "loss": 0.5784, "step": 5869 }, { "epoch": 0.20913123252044105, "grad_norm": 1.7620876355990462, "learning_rate": 1.8364166256448173e-05, "loss": 0.5849, "step": 5870 }, { "epoch": 0.20916685964693518, "grad_norm": 1.2128220217593735, "learning_rate": 1.8363533734777207e-05, "loss": 0.4929, "step": 5871 }, { "epoch": 0.20920248677342929, "grad_norm": 1.1173602321558822, "learning_rate": 1.8362901101739942e-05, "loss": 0.5077, "step": 5872 }, { "epoch": 0.2092381138999234, "grad_norm": 1.6531463760980207, "learning_rate": 1.8362268357344804e-05, "loss": 0.6204, "step": 5873 }, { "epoch": 0.20927374102641752, "grad_norm": 1.813570708656356, "learning_rate": 1.8361635501600222e-05, "loss": 0.5831, "step": 5874 }, { "epoch": 0.20930936815291162, "grad_norm": 1.9911821066350746, "learning_rate": 1.8361002534514613e-05, "loss": 0.6511, "step": 5875 }, { "epoch": 0.20934499527940573, "grad_norm": 1.1888119515881252, "learning_rate": 1.8360369456096413e-05, "loss": 0.4728, "step": 5876 }, { "epoch": 0.20938062240589986, "grad_norm": 1.1370363069396139, "learning_rate": 1.8359736266354052e-05, "loss": 0.5326, "step": 5877 }, { "epoch": 0.20941624953239396, "grad_norm": 1.6789709144193785, "learning_rate": 1.8359102965295956e-05, "loss": 0.6118, "step": 5878 }, { "epoch": 0.20945187665888806, "grad_norm": 1.1842223399522824, "learning_rate": 1.8358469552930564e-05, "loss": 0.4766, "step": 5879 }, { "epoch": 0.2094875037853822, "grad_norm": 1.2316830970391943, "learning_rate": 1.8357836029266307e-05, "loss": 0.5007, "step": 5880 }, { "epoch": 0.2095231309118763, "grad_norm": 1.7604922871939321, "learning_rate": 1.8357202394311624e-05, "loss": 0.6401, "step": 5881 }, { "epoch": 0.2095587580383704, "grad_norm": 1.7526501249357866, "learning_rate": 1.8356568648074946e-05, "loss": 0.68, "step": 5882 }, { "epoch": 0.20959438516486453, "grad_norm": 1.624606709100386, "learning_rate": 1.8355934790564718e-05, "loss": 0.5665, "step": 5883 }, { "epoch": 0.20963001229135864, "grad_norm": 1.168917356145548, "learning_rate": 1.8355300821789373e-05, "loss": 0.5662, "step": 5884 }, { "epoch": 0.20966563941785274, "grad_norm": 1.0336988509986877, "learning_rate": 1.8354666741757366e-05, "loss": 0.5073, "step": 5885 }, { "epoch": 0.20970126654434687, "grad_norm": 1.2180858542409385, "learning_rate": 1.8354032550477126e-05, "loss": 0.5367, "step": 5886 }, { "epoch": 0.20973689367084097, "grad_norm": 2.1651265315181214, "learning_rate": 1.8353398247957105e-05, "loss": 0.58, "step": 5887 }, { "epoch": 0.2097725207973351, "grad_norm": 1.2297782472714252, "learning_rate": 1.8352763834205748e-05, "loss": 0.5781, "step": 5888 }, { "epoch": 0.2098081479238292, "grad_norm": 1.1884750318051318, "learning_rate": 1.83521293092315e-05, "loss": 0.5282, "step": 5889 }, { "epoch": 0.2098437750503233, "grad_norm": 1.0891104198989399, "learning_rate": 1.8351494673042814e-05, "loss": 0.4753, "step": 5890 }, { "epoch": 0.20987940217681744, "grad_norm": 1.2031787301589678, "learning_rate": 1.835085992564814e-05, "loss": 0.5176, "step": 5891 }, { "epoch": 0.20991502930331155, "grad_norm": 1.3100579204241551, "learning_rate": 1.8350225067055927e-05, "loss": 0.5376, "step": 5892 }, { "epoch": 0.20995065642980565, "grad_norm": 1.2646300764609737, "learning_rate": 1.8349590097274633e-05, "loss": 0.4726, "step": 5893 }, { "epoch": 0.20998628355629978, "grad_norm": 1.5978888932819046, "learning_rate": 1.834895501631271e-05, "loss": 0.64, "step": 5894 }, { "epoch": 0.21002191068279388, "grad_norm": 1.6654076509101594, "learning_rate": 1.8348319824178616e-05, "loss": 0.6788, "step": 5895 }, { "epoch": 0.210057537809288, "grad_norm": 1.2389450057514577, "learning_rate": 1.8347684520880807e-05, "loss": 0.5059, "step": 5896 }, { "epoch": 0.21009316493578212, "grad_norm": 1.2146509272583914, "learning_rate": 1.8347049106427745e-05, "loss": 0.5022, "step": 5897 }, { "epoch": 0.21012879206227622, "grad_norm": 1.6834530621080634, "learning_rate": 1.834641358082789e-05, "loss": 0.54, "step": 5898 }, { "epoch": 0.21016441918877032, "grad_norm": 2.027339676707011, "learning_rate": 1.8345777944089703e-05, "loss": 0.6885, "step": 5899 }, { "epoch": 0.21020004631526445, "grad_norm": 1.8664443833220108, "learning_rate": 1.834514219622165e-05, "loss": 0.6627, "step": 5900 }, { "epoch": 0.21023567344175856, "grad_norm": 2.087212254730552, "learning_rate": 1.83445063372322e-05, "loss": 0.6787, "step": 5901 }, { "epoch": 0.21027130056825266, "grad_norm": 1.162181613432918, "learning_rate": 1.834387036712981e-05, "loss": 0.5295, "step": 5902 }, { "epoch": 0.2103069276947468, "grad_norm": 2.2254375772547035, "learning_rate": 1.8343234285922955e-05, "loss": 0.6022, "step": 5903 }, { "epoch": 0.2103425548212409, "grad_norm": 1.1015678779888407, "learning_rate": 1.8342598093620103e-05, "loss": 0.4804, "step": 5904 }, { "epoch": 0.210378181947735, "grad_norm": 1.2992148718252203, "learning_rate": 1.834196179022973e-05, "loss": 0.5609, "step": 5905 }, { "epoch": 0.21041380907422913, "grad_norm": 1.10601432429525, "learning_rate": 1.83413253757603e-05, "loss": 0.4721, "step": 5906 }, { "epoch": 0.21044943620072323, "grad_norm": 2.0722973403331593, "learning_rate": 1.8340688850220298e-05, "loss": 0.6475, "step": 5907 }, { "epoch": 0.21048506332721734, "grad_norm": 1.1852750677104948, "learning_rate": 1.8340052213618192e-05, "loss": 0.4892, "step": 5908 }, { "epoch": 0.21052069045371147, "grad_norm": 1.4519568867836679, "learning_rate": 1.833941546596246e-05, "loss": 0.6932, "step": 5909 }, { "epoch": 0.21055631758020557, "grad_norm": 1.6070140427880968, "learning_rate": 1.8338778607261584e-05, "loss": 0.6021, "step": 5910 }, { "epoch": 0.21059194470669967, "grad_norm": 1.6867021000892606, "learning_rate": 1.8338141637524042e-05, "loss": 0.6235, "step": 5911 }, { "epoch": 0.2106275718331938, "grad_norm": 1.6037107535988808, "learning_rate": 1.8337504556758313e-05, "loss": 0.6215, "step": 5912 }, { "epoch": 0.2106631989596879, "grad_norm": 1.3108569171094286, "learning_rate": 1.8336867364972887e-05, "loss": 0.5487, "step": 5913 }, { "epoch": 0.210698826086182, "grad_norm": 1.2220558238005512, "learning_rate": 1.8336230062176245e-05, "loss": 0.5438, "step": 5914 }, { "epoch": 0.21073445321267614, "grad_norm": 1.1833571362326505, "learning_rate": 1.833559264837687e-05, "loss": 0.5222, "step": 5915 }, { "epoch": 0.21077008033917025, "grad_norm": 1.1627673541339294, "learning_rate": 1.8334955123583252e-05, "loss": 0.5114, "step": 5916 }, { "epoch": 0.21080570746566435, "grad_norm": 1.4136156192642804, "learning_rate": 1.8334317487803884e-05, "loss": 0.5182, "step": 5917 }, { "epoch": 0.21084133459215848, "grad_norm": 1.7470217608064516, "learning_rate": 1.8333679741047255e-05, "loss": 0.5931, "step": 5918 }, { "epoch": 0.21087696171865258, "grad_norm": 1.1745581353402406, "learning_rate": 1.8333041883321852e-05, "loss": 0.5037, "step": 5919 }, { "epoch": 0.2109125888451467, "grad_norm": 1.563889503546069, "learning_rate": 1.833240391463617e-05, "loss": 0.5792, "step": 5920 }, { "epoch": 0.21094821597164082, "grad_norm": 1.6518154624739174, "learning_rate": 1.833176583499871e-05, "loss": 0.6501, "step": 5921 }, { "epoch": 0.21098384309813492, "grad_norm": 2.1427374425867187, "learning_rate": 1.8331127644417965e-05, "loss": 0.5941, "step": 5922 }, { "epoch": 0.21101947022462902, "grad_norm": 1.583950993842889, "learning_rate": 1.833048934290243e-05, "loss": 0.7178, "step": 5923 }, { "epoch": 0.21105509735112316, "grad_norm": 1.455024191145801, "learning_rate": 1.8329850930460606e-05, "loss": 0.604, "step": 5924 }, { "epoch": 0.21109072447761726, "grad_norm": 1.2742446757876174, "learning_rate": 1.8329212407100996e-05, "loss": 0.5412, "step": 5925 }, { "epoch": 0.21112635160411136, "grad_norm": 2.2148428928027064, "learning_rate": 1.83285737728321e-05, "loss": 0.6106, "step": 5926 }, { "epoch": 0.2111619787306055, "grad_norm": 1.1908193945170604, "learning_rate": 1.8327935027662426e-05, "loss": 0.5209, "step": 5927 }, { "epoch": 0.2111976058570996, "grad_norm": 1.0642161756493074, "learning_rate": 1.832729617160047e-05, "loss": 0.5298, "step": 5928 }, { "epoch": 0.2112332329835937, "grad_norm": 1.7239987036154574, "learning_rate": 1.8326657204654752e-05, "loss": 0.6103, "step": 5929 }, { "epoch": 0.21126886011008783, "grad_norm": 2.1881107876517714, "learning_rate": 1.832601812683377e-05, "loss": 0.5667, "step": 5930 }, { "epoch": 0.21130448723658193, "grad_norm": 1.0950437706435736, "learning_rate": 1.8325378938146037e-05, "loss": 0.5, "step": 5931 }, { "epoch": 0.21134011436307604, "grad_norm": 1.5606813620417186, "learning_rate": 1.8324739638600064e-05, "loss": 0.6004, "step": 5932 }, { "epoch": 0.21137574148957017, "grad_norm": 1.223274756834151, "learning_rate": 1.8324100228204367e-05, "loss": 0.5031, "step": 5933 }, { "epoch": 0.21141136861606427, "grad_norm": 1.9466602325138573, "learning_rate": 1.8323460706967453e-05, "loss": 0.6621, "step": 5934 }, { "epoch": 0.21144699574255837, "grad_norm": 1.4093900420388303, "learning_rate": 1.8322821074897846e-05, "loss": 0.4918, "step": 5935 }, { "epoch": 0.2114826228690525, "grad_norm": 1.8008540138519107, "learning_rate": 1.8322181332004057e-05, "loss": 0.6035, "step": 5936 }, { "epoch": 0.2115182499955466, "grad_norm": 1.7499634759277385, "learning_rate": 1.8321541478294605e-05, "loss": 0.6314, "step": 5937 }, { "epoch": 0.2115538771220407, "grad_norm": 1.2088756039384037, "learning_rate": 1.8320901513778014e-05, "loss": 0.5068, "step": 5938 }, { "epoch": 0.21158950424853484, "grad_norm": 1.4974600327421075, "learning_rate": 1.8320261438462802e-05, "loss": 0.6301, "step": 5939 }, { "epoch": 0.21162513137502895, "grad_norm": 1.2419404866077979, "learning_rate": 1.8319621252357492e-05, "loss": 0.5102, "step": 5940 }, { "epoch": 0.21166075850152305, "grad_norm": 1.2857098949567674, "learning_rate": 1.8318980955470614e-05, "loss": 0.516, "step": 5941 }, { "epoch": 0.21169638562801718, "grad_norm": 1.1592155370310433, "learning_rate": 1.8318340547810688e-05, "loss": 0.4873, "step": 5942 }, { "epoch": 0.21173201275451128, "grad_norm": 1.1971103902866007, "learning_rate": 1.8317700029386245e-05, "loss": 0.5168, "step": 5943 }, { "epoch": 0.2117676398810054, "grad_norm": 1.6853443259505216, "learning_rate": 1.831705940020581e-05, "loss": 0.576, "step": 5944 }, { "epoch": 0.21180326700749952, "grad_norm": 1.981167464712136, "learning_rate": 1.8316418660277916e-05, "loss": 0.6253, "step": 5945 }, { "epoch": 0.21183889413399362, "grad_norm": 1.156479970253078, "learning_rate": 1.8315777809611098e-05, "loss": 0.474, "step": 5946 }, { "epoch": 0.21187452126048772, "grad_norm": 1.2375742544933628, "learning_rate": 1.8315136848213883e-05, "loss": 0.5109, "step": 5947 }, { "epoch": 0.21191014838698186, "grad_norm": 1.2586970000929865, "learning_rate": 1.831449577609481e-05, "loss": 0.581, "step": 5948 }, { "epoch": 0.21194577551347596, "grad_norm": 1.835778854956511, "learning_rate": 1.8313854593262413e-05, "loss": 0.5872, "step": 5949 }, { "epoch": 0.21198140263997006, "grad_norm": 1.8957025088582005, "learning_rate": 1.8313213299725232e-05, "loss": 0.58, "step": 5950 }, { "epoch": 0.2120170297664642, "grad_norm": 1.7460737832964393, "learning_rate": 1.8312571895491803e-05, "loss": 0.693, "step": 5951 }, { "epoch": 0.2120526568929583, "grad_norm": 1.4143317855452076, "learning_rate": 1.8311930380570672e-05, "loss": 0.5335, "step": 5952 }, { "epoch": 0.2120882840194524, "grad_norm": 1.1532413879735879, "learning_rate": 1.831128875497038e-05, "loss": 0.4763, "step": 5953 }, { "epoch": 0.21212391114594653, "grad_norm": 1.4787890377928063, "learning_rate": 1.8310647018699464e-05, "loss": 0.6029, "step": 5954 }, { "epoch": 0.21215953827244063, "grad_norm": 1.2341811251604315, "learning_rate": 1.8310005171766477e-05, "loss": 0.5545, "step": 5955 }, { "epoch": 0.21219516539893474, "grad_norm": 1.272473087114129, "learning_rate": 1.8309363214179965e-05, "loss": 0.5062, "step": 5956 }, { "epoch": 0.21223079252542887, "grad_norm": 1.2166087446241631, "learning_rate": 1.830872114594847e-05, "loss": 0.4908, "step": 5957 }, { "epoch": 0.21226641965192297, "grad_norm": 1.1385237396909165, "learning_rate": 1.8308078967080547e-05, "loss": 0.5253, "step": 5958 }, { "epoch": 0.21230204677841708, "grad_norm": 1.5057336757309672, "learning_rate": 1.8307436677584743e-05, "loss": 0.5304, "step": 5959 }, { "epoch": 0.2123376739049112, "grad_norm": 1.951295328687989, "learning_rate": 1.830679427746962e-05, "loss": 0.642, "step": 5960 }, { "epoch": 0.2123733010314053, "grad_norm": 1.115363295007255, "learning_rate": 1.830615176674372e-05, "loss": 0.5089, "step": 5961 }, { "epoch": 0.2124089281578994, "grad_norm": 1.4709298542314495, "learning_rate": 1.8305509145415602e-05, "loss": 0.5328, "step": 5962 }, { "epoch": 0.21244455528439354, "grad_norm": 1.243970331330827, "learning_rate": 1.8304866413493826e-05, "loss": 0.5495, "step": 5963 }, { "epoch": 0.21248018241088765, "grad_norm": 1.2559271305170383, "learning_rate": 1.8304223570986952e-05, "loss": 0.5015, "step": 5964 }, { "epoch": 0.21251580953738175, "grad_norm": 1.1419311194435942, "learning_rate": 1.8303580617903532e-05, "loss": 0.5339, "step": 5965 }, { "epoch": 0.21255143666387588, "grad_norm": 1.1247191975026691, "learning_rate": 1.8302937554252138e-05, "loss": 0.5337, "step": 5966 }, { "epoch": 0.21258706379036998, "grad_norm": 1.175734522807393, "learning_rate": 1.8302294380041325e-05, "loss": 0.4943, "step": 5967 }, { "epoch": 0.2126226909168641, "grad_norm": 1.2400834688585847, "learning_rate": 1.8301651095279657e-05, "loss": 0.5383, "step": 5968 }, { "epoch": 0.21265831804335822, "grad_norm": 1.117062414589223, "learning_rate": 1.8301007699975704e-05, "loss": 0.4715, "step": 5969 }, { "epoch": 0.21269394516985232, "grad_norm": 2.207333170125454, "learning_rate": 1.830036419413803e-05, "loss": 0.6834, "step": 5970 }, { "epoch": 0.21272957229634643, "grad_norm": 1.809476122843566, "learning_rate": 1.8299720577775205e-05, "loss": 0.631, "step": 5971 }, { "epoch": 0.21276519942284056, "grad_norm": 1.127408961121637, "learning_rate": 1.82990768508958e-05, "loss": 0.4974, "step": 5972 }, { "epoch": 0.21280082654933466, "grad_norm": 1.1930866388881838, "learning_rate": 1.8298433013508384e-05, "loss": 0.5173, "step": 5973 }, { "epoch": 0.21283645367582876, "grad_norm": 1.1752877719732129, "learning_rate": 1.8297789065621536e-05, "loss": 0.4825, "step": 5974 }, { "epoch": 0.2128720808023229, "grad_norm": 1.6392771961843415, "learning_rate": 1.8297145007243827e-05, "loss": 0.536, "step": 5975 }, { "epoch": 0.212907707928817, "grad_norm": 1.6631037381909362, "learning_rate": 1.829650083838383e-05, "loss": 0.6002, "step": 5976 }, { "epoch": 0.2129433350553111, "grad_norm": 1.5318049371084785, "learning_rate": 1.8295856559050126e-05, "loss": 0.6225, "step": 5977 }, { "epoch": 0.21297896218180523, "grad_norm": 1.44385024577217, "learning_rate": 1.8295212169251295e-05, "loss": 0.6936, "step": 5978 }, { "epoch": 0.21301458930829933, "grad_norm": 1.0717726955622076, "learning_rate": 1.8294567668995914e-05, "loss": 0.5348, "step": 5979 }, { "epoch": 0.21305021643479344, "grad_norm": 1.1278127396750184, "learning_rate": 1.829392305829257e-05, "loss": 0.5767, "step": 5980 }, { "epoch": 0.21308584356128757, "grad_norm": 1.0696134981824776, "learning_rate": 1.829327833714984e-05, "loss": 0.4588, "step": 5981 }, { "epoch": 0.21312147068778167, "grad_norm": 1.123646229786844, "learning_rate": 1.8292633505576313e-05, "loss": 0.5137, "step": 5982 }, { "epoch": 0.2131570978142758, "grad_norm": 1.2293562824523396, "learning_rate": 1.8291988563580576e-05, "loss": 0.5301, "step": 5983 }, { "epoch": 0.2131927249407699, "grad_norm": 1.1106433066820374, "learning_rate": 1.8291343511171217e-05, "loss": 0.536, "step": 5984 }, { "epoch": 0.213228352067264, "grad_norm": 1.0529997459358147, "learning_rate": 1.829069834835682e-05, "loss": 0.4705, "step": 5985 }, { "epoch": 0.21326397919375814, "grad_norm": 1.0862604719202413, "learning_rate": 1.829005307514598e-05, "loss": 0.4994, "step": 5986 }, { "epoch": 0.21329960632025224, "grad_norm": 1.7295317282429952, "learning_rate": 1.828940769154729e-05, "loss": 0.5472, "step": 5987 }, { "epoch": 0.21333523344674635, "grad_norm": 1.503654450662011, "learning_rate": 1.8288762197569343e-05, "loss": 0.6192, "step": 5988 }, { "epoch": 0.21337086057324048, "grad_norm": 1.5831626992547163, "learning_rate": 1.8288116593220737e-05, "loss": 0.5466, "step": 5989 }, { "epoch": 0.21340648769973458, "grad_norm": 1.3834400874492019, "learning_rate": 1.8287470878510062e-05, "loss": 0.4922, "step": 5990 }, { "epoch": 0.21344211482622868, "grad_norm": 1.5799931395776534, "learning_rate": 1.8286825053445916e-05, "loss": 0.6941, "step": 5991 }, { "epoch": 0.21347774195272282, "grad_norm": 1.8933835527373937, "learning_rate": 1.8286179118036908e-05, "loss": 0.6862, "step": 5992 }, { "epoch": 0.21351336907921692, "grad_norm": 1.8209588074282905, "learning_rate": 1.828553307229163e-05, "loss": 0.6268, "step": 5993 }, { "epoch": 0.21354899620571102, "grad_norm": 1.1999253750417849, "learning_rate": 1.828488691621869e-05, "loss": 0.4741, "step": 5994 }, { "epoch": 0.21358462333220515, "grad_norm": 1.6491428105548365, "learning_rate": 1.828424064982669e-05, "loss": 0.569, "step": 5995 }, { "epoch": 0.21362025045869926, "grad_norm": 1.2327196426978417, "learning_rate": 1.828359427312423e-05, "loss": 0.5367, "step": 5996 }, { "epoch": 0.21365587758519336, "grad_norm": 1.091451371297067, "learning_rate": 1.828294778611993e-05, "loss": 0.5088, "step": 5997 }, { "epoch": 0.2136915047116875, "grad_norm": 1.2378106159327213, "learning_rate": 1.828230118882239e-05, "loss": 0.5199, "step": 5998 }, { "epoch": 0.2137271318381816, "grad_norm": 1.5228175060180245, "learning_rate": 1.828165448124022e-05, "loss": 0.6342, "step": 5999 }, { "epoch": 0.2137627589646757, "grad_norm": 1.2879445620970211, "learning_rate": 1.8281007663382027e-05, "loss": 0.5427, "step": 6000 }, { "epoch": 0.21379838609116983, "grad_norm": 1.1255181771776022, "learning_rate": 1.8280360735256432e-05, "loss": 0.5646, "step": 6001 }, { "epoch": 0.21383401321766393, "grad_norm": 1.8220975087389313, "learning_rate": 1.8279713696872047e-05, "loss": 0.6469, "step": 6002 }, { "epoch": 0.21386964034415804, "grad_norm": 1.1473849215080174, "learning_rate": 1.8279066548237487e-05, "loss": 0.5103, "step": 6003 }, { "epoch": 0.21390526747065217, "grad_norm": 1.186680442998034, "learning_rate": 1.827841928936137e-05, "loss": 0.5427, "step": 6004 }, { "epoch": 0.21394089459714627, "grad_norm": 1.1706733589409328, "learning_rate": 1.8277771920252312e-05, "loss": 0.5309, "step": 6005 }, { "epoch": 0.21397652172364037, "grad_norm": 1.1334677943173577, "learning_rate": 1.8277124440918935e-05, "loss": 0.5098, "step": 6006 }, { "epoch": 0.2140121488501345, "grad_norm": 1.1346890493266277, "learning_rate": 1.8276476851369862e-05, "loss": 0.5236, "step": 6007 }, { "epoch": 0.2140477759766286, "grad_norm": 1.5391594383191156, "learning_rate": 1.8275829151613712e-05, "loss": 0.635, "step": 6008 }, { "epoch": 0.2140834031031227, "grad_norm": 1.24561948357564, "learning_rate": 1.8275181341659117e-05, "loss": 0.5314, "step": 6009 }, { "epoch": 0.21411903022961684, "grad_norm": 2.222258214492884, "learning_rate": 1.8274533421514697e-05, "loss": 0.6318, "step": 6010 }, { "epoch": 0.21415465735611094, "grad_norm": 2.0492906769148354, "learning_rate": 1.827388539118908e-05, "loss": 0.6339, "step": 6011 }, { "epoch": 0.21419028448260505, "grad_norm": 1.8478870419165734, "learning_rate": 1.8273237250690894e-05, "loss": 0.684, "step": 6012 }, { "epoch": 0.21422591160909918, "grad_norm": 1.315781736685705, "learning_rate": 1.8272589000028774e-05, "loss": 0.4713, "step": 6013 }, { "epoch": 0.21426153873559328, "grad_norm": 1.1829280950481715, "learning_rate": 1.8271940639211347e-05, "loss": 0.4678, "step": 6014 }, { "epoch": 0.21429716586208739, "grad_norm": 1.2320866686086682, "learning_rate": 1.8271292168247253e-05, "loss": 0.5226, "step": 6015 }, { "epoch": 0.21433279298858152, "grad_norm": 1.6310804725461705, "learning_rate": 1.8270643587145118e-05, "loss": 0.5591, "step": 6016 }, { "epoch": 0.21436842011507562, "grad_norm": 2.229608349764098, "learning_rate": 1.8269994895913585e-05, "loss": 0.6166, "step": 6017 }, { "epoch": 0.21440404724156972, "grad_norm": 1.77678238523531, "learning_rate": 1.826934609456129e-05, "loss": 0.6152, "step": 6018 }, { "epoch": 0.21443967436806385, "grad_norm": 1.2608083185190113, "learning_rate": 1.826869718309687e-05, "loss": 0.5259, "step": 6019 }, { "epoch": 0.21447530149455796, "grad_norm": 1.1229237998725494, "learning_rate": 1.8268048161528966e-05, "loss": 0.5302, "step": 6020 }, { "epoch": 0.21451092862105206, "grad_norm": 1.5421563766594377, "learning_rate": 1.8267399029866227e-05, "loss": 0.6195, "step": 6021 }, { "epoch": 0.2145465557475462, "grad_norm": 1.7112809000854319, "learning_rate": 1.826674978811729e-05, "loss": 0.5461, "step": 6022 }, { "epoch": 0.2145821828740403, "grad_norm": 1.1957899406864558, "learning_rate": 1.8266100436290796e-05, "loss": 0.4749, "step": 6023 }, { "epoch": 0.2146178100005344, "grad_norm": 2.012548831355216, "learning_rate": 1.8265450974395403e-05, "loss": 0.6179, "step": 6024 }, { "epoch": 0.21465343712702853, "grad_norm": 1.2380335266342883, "learning_rate": 1.826480140243975e-05, "loss": 0.4943, "step": 6025 }, { "epoch": 0.21468906425352263, "grad_norm": 1.5636506007573125, "learning_rate": 1.826415172043249e-05, "loss": 0.6305, "step": 6026 }, { "epoch": 0.21472469138001674, "grad_norm": 1.4213404768404814, "learning_rate": 1.8263501928382274e-05, "loss": 0.6248, "step": 6027 }, { "epoch": 0.21476031850651087, "grad_norm": 1.2163435904357787, "learning_rate": 1.8262852026297754e-05, "loss": 0.5139, "step": 6028 }, { "epoch": 0.21479594563300497, "grad_norm": 1.2423789645892547, "learning_rate": 1.8262202014187583e-05, "loss": 0.5115, "step": 6029 }, { "epoch": 0.21483157275949907, "grad_norm": 1.5212955590938906, "learning_rate": 1.826155189206042e-05, "loss": 0.6163, "step": 6030 }, { "epoch": 0.2148671998859932, "grad_norm": 2.030342024109571, "learning_rate": 1.8260901659924917e-05, "loss": 0.6233, "step": 6031 }, { "epoch": 0.2149028270124873, "grad_norm": 1.8336416309067802, "learning_rate": 1.8260251317789738e-05, "loss": 0.526, "step": 6032 }, { "epoch": 0.2149384541389814, "grad_norm": 1.5353579644070445, "learning_rate": 1.8259600865663532e-05, "loss": 0.5738, "step": 6033 }, { "epoch": 0.21497408126547554, "grad_norm": 2.0971706611824477, "learning_rate": 1.825895030355497e-05, "loss": 0.6555, "step": 6034 }, { "epoch": 0.21500970839196965, "grad_norm": 1.4076299480592362, "learning_rate": 1.8258299631472717e-05, "loss": 0.5022, "step": 6035 }, { "epoch": 0.21504533551846375, "grad_norm": 1.5156022772104207, "learning_rate": 1.8257648849425427e-05, "loss": 0.5363, "step": 6036 }, { "epoch": 0.21508096264495788, "grad_norm": 1.0908627351110503, "learning_rate": 1.8256997957421774e-05, "loss": 0.4934, "step": 6037 }, { "epoch": 0.21511658977145198, "grad_norm": 1.2718562832349016, "learning_rate": 1.825634695547042e-05, "loss": 0.5348, "step": 6038 }, { "epoch": 0.21515221689794609, "grad_norm": 1.2388110182792167, "learning_rate": 1.8255695843580036e-05, "loss": 0.4819, "step": 6039 }, { "epoch": 0.21518784402444022, "grad_norm": 1.1885400141120863, "learning_rate": 1.8255044621759288e-05, "loss": 0.4934, "step": 6040 }, { "epoch": 0.21522347115093432, "grad_norm": 1.6557621284777282, "learning_rate": 1.8254393290016854e-05, "loss": 0.6272, "step": 6041 }, { "epoch": 0.21525909827742842, "grad_norm": 1.1617694029830983, "learning_rate": 1.8253741848361403e-05, "loss": 0.528, "step": 6042 }, { "epoch": 0.21529472540392255, "grad_norm": 1.5278288209832276, "learning_rate": 1.8253090296801614e-05, "loss": 0.5715, "step": 6043 }, { "epoch": 0.21533035253041666, "grad_norm": 1.6062340796916386, "learning_rate": 1.8252438635346155e-05, "loss": 0.7237, "step": 6044 }, { "epoch": 0.21536597965691076, "grad_norm": 1.2720226375142392, "learning_rate": 1.8251786864003712e-05, "loss": 0.5573, "step": 6045 }, { "epoch": 0.2154016067834049, "grad_norm": 1.5790950476667243, "learning_rate": 1.8251134982782952e-05, "loss": 0.5875, "step": 6046 }, { "epoch": 0.215437233909899, "grad_norm": 1.439789428139295, "learning_rate": 1.825048299169257e-05, "loss": 0.6236, "step": 6047 }, { "epoch": 0.2154728610363931, "grad_norm": 1.1640603838543109, "learning_rate": 1.8249830890741234e-05, "loss": 0.5144, "step": 6048 }, { "epoch": 0.21550848816288723, "grad_norm": 1.206604734164143, "learning_rate": 1.8249178679937637e-05, "loss": 0.5611, "step": 6049 }, { "epoch": 0.21554411528938133, "grad_norm": 1.164331573140681, "learning_rate": 1.824852635929046e-05, "loss": 0.5022, "step": 6050 }, { "epoch": 0.21557974241587544, "grad_norm": 1.2103255697609039, "learning_rate": 1.824787392880839e-05, "loss": 0.5629, "step": 6051 }, { "epoch": 0.21561536954236957, "grad_norm": 1.1007359692707661, "learning_rate": 1.8247221388500106e-05, "loss": 0.5031, "step": 6052 }, { "epoch": 0.21565099666886367, "grad_norm": 1.2173808159601127, "learning_rate": 1.8246568738374316e-05, "loss": 0.5045, "step": 6053 }, { "epoch": 0.21568662379535777, "grad_norm": 1.1398429086960735, "learning_rate": 1.824591597843969e-05, "loss": 0.5105, "step": 6054 }, { "epoch": 0.2157222509218519, "grad_norm": 2.0268425003681605, "learning_rate": 1.8245263108704937e-05, "loss": 0.6486, "step": 6055 }, { "epoch": 0.215757878048346, "grad_norm": 1.8171780954613526, "learning_rate": 1.8244610129178737e-05, "loss": 0.6379, "step": 6056 }, { "epoch": 0.2157935051748401, "grad_norm": 1.3899958008842783, "learning_rate": 1.824395703986979e-05, "loss": 0.5813, "step": 6057 }, { "epoch": 0.21582913230133424, "grad_norm": 1.4397625398461236, "learning_rate": 1.8243303840786798e-05, "loss": 0.6376, "step": 6058 }, { "epoch": 0.21586475942782835, "grad_norm": 1.0650620437734168, "learning_rate": 1.824265053193845e-05, "loss": 0.4885, "step": 6059 }, { "epoch": 0.21590038655432245, "grad_norm": 1.2776040190617135, "learning_rate": 1.824199711333345e-05, "loss": 0.5276, "step": 6060 }, { "epoch": 0.21593601368081658, "grad_norm": 1.9425120274874983, "learning_rate": 1.8241343584980495e-05, "loss": 0.5433, "step": 6061 }, { "epoch": 0.21597164080731068, "grad_norm": 1.3986993529776213, "learning_rate": 1.8240689946888296e-05, "loss": 0.6549, "step": 6062 }, { "epoch": 0.2160072679338048, "grad_norm": 1.3000897172322992, "learning_rate": 1.8240036199065546e-05, "loss": 0.5614, "step": 6063 }, { "epoch": 0.21604289506029892, "grad_norm": 1.0936954536618697, "learning_rate": 1.823938234152095e-05, "loss": 0.5405, "step": 6064 }, { "epoch": 0.21607852218679302, "grad_norm": 1.1732625819984477, "learning_rate": 1.8238728374263228e-05, "loss": 0.5609, "step": 6065 }, { "epoch": 0.21611414931328712, "grad_norm": 1.212416731339019, "learning_rate": 1.8238074297301076e-05, "loss": 0.5122, "step": 6066 }, { "epoch": 0.21614977643978125, "grad_norm": 1.1598918107908354, "learning_rate": 1.8237420110643205e-05, "loss": 0.4954, "step": 6067 }, { "epoch": 0.21618540356627536, "grad_norm": 1.6347121226412493, "learning_rate": 1.8236765814298328e-05, "loss": 0.616, "step": 6068 }, { "epoch": 0.21622103069276946, "grad_norm": 1.099072912247118, "learning_rate": 1.823611140827516e-05, "loss": 0.5296, "step": 6069 }, { "epoch": 0.2162566578192636, "grad_norm": 1.1300585717015306, "learning_rate": 1.823545689258241e-05, "loss": 0.4851, "step": 6070 }, { "epoch": 0.2162922849457577, "grad_norm": 1.221946413046572, "learning_rate": 1.8234802267228794e-05, "loss": 0.4984, "step": 6071 }, { "epoch": 0.2163279120722518, "grad_norm": 1.066742385312337, "learning_rate": 1.823414753222303e-05, "loss": 0.4798, "step": 6072 }, { "epoch": 0.21636353919874593, "grad_norm": 1.2045050376626372, "learning_rate": 1.823349268757384e-05, "loss": 0.5043, "step": 6073 }, { "epoch": 0.21639916632524003, "grad_norm": 1.1596805708036102, "learning_rate": 1.8232837733289935e-05, "loss": 0.4748, "step": 6074 }, { "epoch": 0.21643479345173414, "grad_norm": 1.7435064952787978, "learning_rate": 1.8232182669380044e-05, "loss": 0.6373, "step": 6075 }, { "epoch": 0.21647042057822827, "grad_norm": 1.6025740279031129, "learning_rate": 1.8231527495852884e-05, "loss": 0.6917, "step": 6076 }, { "epoch": 0.21650604770472237, "grad_norm": 1.0911466659880638, "learning_rate": 1.8230872212717185e-05, "loss": 0.5024, "step": 6077 }, { "epoch": 0.2165416748312165, "grad_norm": 1.0732797164607961, "learning_rate": 1.8230216819981667e-05, "loss": 0.4746, "step": 6078 }, { "epoch": 0.2165773019577106, "grad_norm": 1.5403607102269545, "learning_rate": 1.822956131765506e-05, "loss": 0.5734, "step": 6079 }, { "epoch": 0.2166129290842047, "grad_norm": 1.3404547226744628, "learning_rate": 1.8228905705746094e-05, "loss": 0.5152, "step": 6080 }, { "epoch": 0.21664855621069884, "grad_norm": 1.8640009640978394, "learning_rate": 1.8228249984263496e-05, "loss": 0.5797, "step": 6081 }, { "epoch": 0.21668418333719294, "grad_norm": 1.1934064358537027, "learning_rate": 1.8227594153216e-05, "loss": 0.5129, "step": 6082 }, { "epoch": 0.21671981046368705, "grad_norm": 1.5862005056476185, "learning_rate": 1.8226938212612335e-05, "loss": 0.6491, "step": 6083 }, { "epoch": 0.21675543759018118, "grad_norm": 1.1354508762435225, "learning_rate": 1.8226282162461237e-05, "loss": 0.4842, "step": 6084 }, { "epoch": 0.21679106471667528, "grad_norm": 1.6775747419539713, "learning_rate": 1.8225626002771443e-05, "loss": 0.5865, "step": 6085 }, { "epoch": 0.21682669184316938, "grad_norm": 1.256022853821849, "learning_rate": 1.822496973355169e-05, "loss": 0.5597, "step": 6086 }, { "epoch": 0.21686231896966351, "grad_norm": 1.989301553801954, "learning_rate": 1.8224313354810714e-05, "loss": 0.5224, "step": 6087 }, { "epoch": 0.21689794609615762, "grad_norm": 1.2298758174295317, "learning_rate": 1.822365686655726e-05, "loss": 0.5238, "step": 6088 }, { "epoch": 0.21693357322265172, "grad_norm": 1.186469713882971, "learning_rate": 1.8223000268800063e-05, "loss": 0.5284, "step": 6089 }, { "epoch": 0.21696920034914585, "grad_norm": 1.177465512699452, "learning_rate": 1.8222343561547876e-05, "loss": 0.4834, "step": 6090 }, { "epoch": 0.21700482747563996, "grad_norm": 2.507946028967624, "learning_rate": 1.8221686744809434e-05, "loss": 0.6241, "step": 6091 }, { "epoch": 0.21704045460213406, "grad_norm": 1.429655453772256, "learning_rate": 1.8221029818593484e-05, "loss": 0.6405, "step": 6092 }, { "epoch": 0.2170760817286282, "grad_norm": 1.1623282343715906, "learning_rate": 1.8220372782908778e-05, "loss": 0.4416, "step": 6093 }, { "epoch": 0.2171117088551223, "grad_norm": 1.464069771895813, "learning_rate": 1.821971563776406e-05, "loss": 0.5463, "step": 6094 }, { "epoch": 0.2171473359816164, "grad_norm": 1.1765145460459412, "learning_rate": 1.8219058383168088e-05, "loss": 0.5503, "step": 6095 }, { "epoch": 0.21718296310811053, "grad_norm": 1.218262886092724, "learning_rate": 1.8218401019129607e-05, "loss": 0.5313, "step": 6096 }, { "epoch": 0.21721859023460463, "grad_norm": 1.1182479290205198, "learning_rate": 1.821774354565737e-05, "loss": 0.4477, "step": 6097 }, { "epoch": 0.21725421736109873, "grad_norm": 1.5332892693361853, "learning_rate": 1.8217085962760136e-05, "loss": 0.568, "step": 6098 }, { "epoch": 0.21728984448759286, "grad_norm": 1.4618643539202398, "learning_rate": 1.8216428270446657e-05, "loss": 0.5727, "step": 6099 }, { "epoch": 0.21732547161408697, "grad_norm": 1.3202044179128787, "learning_rate": 1.8215770468725696e-05, "loss": 0.4994, "step": 6100 }, { "epoch": 0.21736109874058107, "grad_norm": 1.2479506615739882, "learning_rate": 1.8215112557606005e-05, "loss": 0.4548, "step": 6101 }, { "epoch": 0.2173967258670752, "grad_norm": 1.098247354752146, "learning_rate": 1.8214454537096352e-05, "loss": 0.5254, "step": 6102 }, { "epoch": 0.2174323529935693, "grad_norm": 1.3715434228995975, "learning_rate": 1.821379640720549e-05, "loss": 0.4913, "step": 6103 }, { "epoch": 0.2174679801200634, "grad_norm": 1.6337484098067203, "learning_rate": 1.821313816794219e-05, "loss": 0.6185, "step": 6104 }, { "epoch": 0.21750360724655754, "grad_norm": 1.2450137702504593, "learning_rate": 1.821247981931522e-05, "loss": 0.4374, "step": 6105 }, { "epoch": 0.21753923437305164, "grad_norm": 1.1482353751793042, "learning_rate": 1.8211821361333333e-05, "loss": 0.5433, "step": 6106 }, { "epoch": 0.21757486149954575, "grad_norm": 1.2292996752496883, "learning_rate": 1.821116279400531e-05, "loss": 0.5218, "step": 6107 }, { "epoch": 0.21761048862603988, "grad_norm": 2.16186073083917, "learning_rate": 1.8210504117339917e-05, "loss": 0.6317, "step": 6108 }, { "epoch": 0.21764611575253398, "grad_norm": 1.6841334388900402, "learning_rate": 1.8209845331345915e-05, "loss": 0.6224, "step": 6109 }, { "epoch": 0.21768174287902808, "grad_norm": 1.5822059735829048, "learning_rate": 1.820918643603209e-05, "loss": 0.5918, "step": 6110 }, { "epoch": 0.21771737000552221, "grad_norm": 1.2544965545732896, "learning_rate": 1.8208527431407206e-05, "loss": 0.4971, "step": 6111 }, { "epoch": 0.21775299713201632, "grad_norm": 1.1706132115321526, "learning_rate": 1.8207868317480046e-05, "loss": 0.4517, "step": 6112 }, { "epoch": 0.21778862425851042, "grad_norm": 1.1950585967515184, "learning_rate": 1.820720909425938e-05, "loss": 0.4973, "step": 6113 }, { "epoch": 0.21782425138500455, "grad_norm": 1.5770043068925432, "learning_rate": 1.8206549761753992e-05, "loss": 0.5894, "step": 6114 }, { "epoch": 0.21785987851149866, "grad_norm": 1.4378272231538796, "learning_rate": 1.8205890319972653e-05, "loss": 0.6037, "step": 6115 }, { "epoch": 0.21789550563799276, "grad_norm": 1.5543613019384699, "learning_rate": 1.8205230768924148e-05, "loss": 0.5702, "step": 6116 }, { "epoch": 0.2179311327644869, "grad_norm": 1.8470807184244815, "learning_rate": 1.8204571108617263e-05, "loss": 0.6429, "step": 6117 }, { "epoch": 0.217966759890981, "grad_norm": 1.4833151759852574, "learning_rate": 1.8203911339060784e-05, "loss": 0.5746, "step": 6118 }, { "epoch": 0.2180023870174751, "grad_norm": 1.18562903337828, "learning_rate": 1.8203251460263485e-05, "loss": 0.4879, "step": 6119 }, { "epoch": 0.21803801414396923, "grad_norm": 1.5367316770325419, "learning_rate": 1.820259147223416e-05, "loss": 0.5609, "step": 6120 }, { "epoch": 0.21807364127046333, "grad_norm": 1.2843212671638546, "learning_rate": 1.8201931374981595e-05, "loss": 0.5214, "step": 6121 }, { "epoch": 0.21810926839695743, "grad_norm": 1.8437172001872693, "learning_rate": 1.820127116851458e-05, "loss": 0.6438, "step": 6122 }, { "epoch": 0.21814489552345157, "grad_norm": 1.8268434600903707, "learning_rate": 1.8200610852841913e-05, "loss": 0.5872, "step": 6123 }, { "epoch": 0.21818052264994567, "grad_norm": 1.265144439797633, "learning_rate": 1.8199950427972374e-05, "loss": 0.518, "step": 6124 }, { "epoch": 0.21821614977643977, "grad_norm": 1.3360593151416822, "learning_rate": 1.8199289893914768e-05, "loss": 0.6061, "step": 6125 }, { "epoch": 0.2182517769029339, "grad_norm": 1.0648480870504409, "learning_rate": 1.8198629250677884e-05, "loss": 0.4677, "step": 6126 }, { "epoch": 0.218287404029428, "grad_norm": 1.6726015124244218, "learning_rate": 1.8197968498270523e-05, "loss": 0.571, "step": 6127 }, { "epoch": 0.2183230311559221, "grad_norm": 1.2684592201557665, "learning_rate": 1.819730763670148e-05, "loss": 0.5115, "step": 6128 }, { "epoch": 0.21835865828241624, "grad_norm": 1.3659261542223293, "learning_rate": 1.8196646665979557e-05, "loss": 0.5289, "step": 6129 }, { "epoch": 0.21839428540891034, "grad_norm": 1.2071714767832555, "learning_rate": 1.8195985586113557e-05, "loss": 0.5082, "step": 6130 }, { "epoch": 0.21842991253540445, "grad_norm": 1.7759097108032753, "learning_rate": 1.819532439711228e-05, "loss": 0.5992, "step": 6131 }, { "epoch": 0.21846553966189858, "grad_norm": 1.5503032572165556, "learning_rate": 1.8194663098984524e-05, "loss": 0.5769, "step": 6132 }, { "epoch": 0.21850116678839268, "grad_norm": 1.4662903434961532, "learning_rate": 1.8194001691739104e-05, "loss": 0.604, "step": 6133 }, { "epoch": 0.21853679391488678, "grad_norm": 1.6165317133342605, "learning_rate": 1.8193340175384826e-05, "loss": 0.6063, "step": 6134 }, { "epoch": 0.21857242104138092, "grad_norm": 1.5881997263400576, "learning_rate": 1.8192678549930496e-05, "loss": 0.6208, "step": 6135 }, { "epoch": 0.21860804816787502, "grad_norm": 1.4615310502234014, "learning_rate": 1.8192016815384925e-05, "loss": 0.4923, "step": 6136 }, { "epoch": 0.21864367529436912, "grad_norm": 1.2615714992704619, "learning_rate": 1.8191354971756923e-05, "loss": 0.4956, "step": 6137 }, { "epoch": 0.21867930242086325, "grad_norm": 1.3681555199698172, "learning_rate": 1.8190693019055307e-05, "loss": 0.483, "step": 6138 }, { "epoch": 0.21871492954735736, "grad_norm": 1.3385751724232091, "learning_rate": 1.8190030957288884e-05, "loss": 0.467, "step": 6139 }, { "epoch": 0.21875055667385146, "grad_norm": 1.6213726061762965, "learning_rate": 1.8189368786466478e-05, "loss": 0.6465, "step": 6140 }, { "epoch": 0.2187861838003456, "grad_norm": 1.791302207764409, "learning_rate": 1.81887065065969e-05, "loss": 0.6457, "step": 6141 }, { "epoch": 0.2188218109268397, "grad_norm": 1.5203357562711886, "learning_rate": 1.8188044117688973e-05, "loss": 0.6003, "step": 6142 }, { "epoch": 0.2188574380533338, "grad_norm": 1.6534219889514434, "learning_rate": 1.8187381619751516e-05, "loss": 0.6246, "step": 6143 }, { "epoch": 0.21889306517982793, "grad_norm": 1.373218328840413, "learning_rate": 1.818671901279335e-05, "loss": 0.5099, "step": 6144 }, { "epoch": 0.21892869230632203, "grad_norm": 1.62821153544824, "learning_rate": 1.8186056296823297e-05, "loss": 0.5513, "step": 6145 }, { "epoch": 0.21896431943281613, "grad_norm": 1.7249990456906645, "learning_rate": 1.8185393471850184e-05, "loss": 0.694, "step": 6146 }, { "epoch": 0.21899994655931027, "grad_norm": 2.945508200434182, "learning_rate": 1.8184730537882833e-05, "loss": 0.6394, "step": 6147 }, { "epoch": 0.21903557368580437, "grad_norm": 1.3310726384496205, "learning_rate": 1.8184067494930076e-05, "loss": 0.5005, "step": 6148 }, { "epoch": 0.21907120081229847, "grad_norm": 1.288326660409068, "learning_rate": 1.818340434300074e-05, "loss": 0.5519, "step": 6149 }, { "epoch": 0.2191068279387926, "grad_norm": 1.526524859069827, "learning_rate": 1.8182741082103656e-05, "loss": 0.5998, "step": 6150 }, { "epoch": 0.2191424550652867, "grad_norm": 1.25538475626361, "learning_rate": 1.8182077712247654e-05, "loss": 0.5404, "step": 6151 }, { "epoch": 0.2191780821917808, "grad_norm": 1.1554979074697407, "learning_rate": 1.818141423344157e-05, "loss": 0.5029, "step": 6152 }, { "epoch": 0.21921370931827494, "grad_norm": 1.1671855417222954, "learning_rate": 1.8180750645694236e-05, "loss": 0.4948, "step": 6153 }, { "epoch": 0.21924933644476904, "grad_norm": 1.1350016557026612, "learning_rate": 1.8180086949014488e-05, "loss": 0.5202, "step": 6154 }, { "epoch": 0.21928496357126315, "grad_norm": 1.1955779647811335, "learning_rate": 1.8179423143411166e-05, "loss": 0.5429, "step": 6155 }, { "epoch": 0.21932059069775728, "grad_norm": 1.248078530328043, "learning_rate": 1.8178759228893108e-05, "loss": 0.4542, "step": 6156 }, { "epoch": 0.21935621782425138, "grad_norm": 1.7668222688019732, "learning_rate": 1.8178095205469154e-05, "loss": 0.6588, "step": 6157 }, { "epoch": 0.21939184495074548, "grad_norm": 1.617005735983056, "learning_rate": 1.8177431073148144e-05, "loss": 0.5103, "step": 6158 }, { "epoch": 0.21942747207723962, "grad_norm": 1.5168321381170349, "learning_rate": 1.8176766831938927e-05, "loss": 0.6171, "step": 6159 }, { "epoch": 0.21946309920373372, "grad_norm": 1.9185001527469339, "learning_rate": 1.817610248185034e-05, "loss": 0.6428, "step": 6160 }, { "epoch": 0.21949872633022782, "grad_norm": 1.2205213499033236, "learning_rate": 1.8175438022891238e-05, "loss": 0.5148, "step": 6161 }, { "epoch": 0.21953435345672195, "grad_norm": 1.2230361784226365, "learning_rate": 1.8174773455070464e-05, "loss": 0.5466, "step": 6162 }, { "epoch": 0.21956998058321606, "grad_norm": 1.141749141226509, "learning_rate": 1.8174108778396866e-05, "loss": 0.459, "step": 6163 }, { "epoch": 0.21960560770971016, "grad_norm": 1.3859218816224241, "learning_rate": 1.8173443992879297e-05, "loss": 0.563, "step": 6164 }, { "epoch": 0.2196412348362043, "grad_norm": 1.0987073673222978, "learning_rate": 1.8172779098526607e-05, "loss": 0.5029, "step": 6165 }, { "epoch": 0.2196768619626984, "grad_norm": 1.2077302066090079, "learning_rate": 1.8172114095347653e-05, "loss": 0.4923, "step": 6166 }, { "epoch": 0.2197124890891925, "grad_norm": 1.2280779988168402, "learning_rate": 1.8171448983351284e-05, "loss": 0.4793, "step": 6167 }, { "epoch": 0.21974811621568663, "grad_norm": 1.21867015960801, "learning_rate": 1.8170783762546363e-05, "loss": 0.4948, "step": 6168 }, { "epoch": 0.21978374334218073, "grad_norm": 1.5467300582272963, "learning_rate": 1.8170118432941748e-05, "loss": 0.5523, "step": 6169 }, { "epoch": 0.21981937046867486, "grad_norm": 1.54690641778388, "learning_rate": 1.816945299454629e-05, "loss": 0.7079, "step": 6170 }, { "epoch": 0.21985499759516897, "grad_norm": 1.9352652338281187, "learning_rate": 1.816878744736886e-05, "loss": 0.5541, "step": 6171 }, { "epoch": 0.21989062472166307, "grad_norm": 1.2450192294539149, "learning_rate": 1.8168121791418313e-05, "loss": 0.508, "step": 6172 }, { "epoch": 0.2199262518481572, "grad_norm": 1.881709434941844, "learning_rate": 1.8167456026703515e-05, "loss": 0.643, "step": 6173 }, { "epoch": 0.2199618789746513, "grad_norm": 2.522121872290217, "learning_rate": 1.8166790153233335e-05, "loss": 0.6347, "step": 6174 }, { "epoch": 0.2199975061011454, "grad_norm": 1.4893812930287396, "learning_rate": 1.8166124171016634e-05, "loss": 0.6071, "step": 6175 }, { "epoch": 0.22003313322763954, "grad_norm": 1.273237704109442, "learning_rate": 1.816545808006228e-05, "loss": 0.4541, "step": 6176 }, { "epoch": 0.22006876035413364, "grad_norm": 1.7589011302519193, "learning_rate": 1.8164791880379145e-05, "loss": 0.6331, "step": 6177 }, { "epoch": 0.22010438748062774, "grad_norm": 2.204932169841722, "learning_rate": 1.81641255719761e-05, "loss": 0.6724, "step": 6178 }, { "epoch": 0.22014001460712188, "grad_norm": 2.0159762071891185, "learning_rate": 1.8163459154862015e-05, "loss": 0.5904, "step": 6179 }, { "epoch": 0.22017564173361598, "grad_norm": 1.336529907806013, "learning_rate": 1.8162792629045765e-05, "loss": 0.5242, "step": 6180 }, { "epoch": 0.22021126886011008, "grad_norm": 1.263321604304148, "learning_rate": 1.816212599453623e-05, "loss": 0.4779, "step": 6181 }, { "epoch": 0.2202468959866042, "grad_norm": 3.817971135861795, "learning_rate": 1.816145925134228e-05, "loss": 0.6482, "step": 6182 }, { "epoch": 0.22028252311309832, "grad_norm": 2.156082700992583, "learning_rate": 1.816079239947279e-05, "loss": 0.683, "step": 6183 }, { "epoch": 0.22031815023959242, "grad_norm": 2.6005221421623763, "learning_rate": 1.8160125438936652e-05, "loss": 0.5713, "step": 6184 }, { "epoch": 0.22035377736608655, "grad_norm": 1.2973452257713627, "learning_rate": 1.8159458369742742e-05, "loss": 0.4874, "step": 6185 }, { "epoch": 0.22038940449258065, "grad_norm": 1.7972699385608846, "learning_rate": 1.8158791191899934e-05, "loss": 0.6027, "step": 6186 }, { "epoch": 0.22042503161907476, "grad_norm": 1.349583152880101, "learning_rate": 1.8158123905417124e-05, "loss": 0.5031, "step": 6187 }, { "epoch": 0.2204606587455689, "grad_norm": 2.08745935783316, "learning_rate": 1.8157456510303188e-05, "loss": 0.564, "step": 6188 }, { "epoch": 0.220496285872063, "grad_norm": 1.620553014650663, "learning_rate": 1.8156789006567018e-05, "loss": 0.4487, "step": 6189 }, { "epoch": 0.2205319129985571, "grad_norm": 1.7130517009483306, "learning_rate": 1.8156121394217504e-05, "loss": 0.536, "step": 6190 }, { "epoch": 0.22056754012505123, "grad_norm": 1.8393082941158316, "learning_rate": 1.815545367326353e-05, "loss": 0.4916, "step": 6191 }, { "epoch": 0.22060316725154533, "grad_norm": 1.88399882755688, "learning_rate": 1.815478584371399e-05, "loss": 0.7361, "step": 6192 }, { "epoch": 0.22063879437803943, "grad_norm": 2.9018816384608708, "learning_rate": 1.8154117905577777e-05, "loss": 0.7296, "step": 6193 }, { "epoch": 0.22067442150453356, "grad_norm": 1.627915624113406, "learning_rate": 1.8153449858863784e-05, "loss": 0.5691, "step": 6194 }, { "epoch": 0.22071004863102767, "grad_norm": 1.8587679604801624, "learning_rate": 1.815278170358091e-05, "loss": 0.6082, "step": 6195 }, { "epoch": 0.22074567575752177, "grad_norm": 1.5570419058179594, "learning_rate": 1.815211343973805e-05, "loss": 0.5509, "step": 6196 }, { "epoch": 0.2207813028840159, "grad_norm": 2.129247067408093, "learning_rate": 1.8151445067344097e-05, "loss": 0.5859, "step": 6197 }, { "epoch": 0.22081693001051, "grad_norm": 1.2994620338508307, "learning_rate": 1.8150776586407957e-05, "loss": 0.594, "step": 6198 }, { "epoch": 0.2208525571370041, "grad_norm": 3.323154517691828, "learning_rate": 1.815010799693853e-05, "loss": 0.579, "step": 6199 }, { "epoch": 0.22088818426349824, "grad_norm": 1.53372141065062, "learning_rate": 1.8149439298944716e-05, "loss": 0.6169, "step": 6200 }, { "epoch": 0.22092381138999234, "grad_norm": 1.8631176978117565, "learning_rate": 1.8148770492435427e-05, "loss": 0.6993, "step": 6201 }, { "epoch": 0.22095943851648644, "grad_norm": 1.2458775053635784, "learning_rate": 1.8148101577419557e-05, "loss": 0.4922, "step": 6202 }, { "epoch": 0.22099506564298058, "grad_norm": 1.3056033336122848, "learning_rate": 1.8147432553906024e-05, "loss": 0.5112, "step": 6203 }, { "epoch": 0.22103069276947468, "grad_norm": 1.864577926955893, "learning_rate": 1.814676342190373e-05, "loss": 0.6018, "step": 6204 }, { "epoch": 0.22106631989596878, "grad_norm": 1.1573104755047372, "learning_rate": 1.814609418142159e-05, "loss": 0.5137, "step": 6205 }, { "epoch": 0.2211019470224629, "grad_norm": 1.9855733967096114, "learning_rate": 1.8145424832468507e-05, "loss": 0.5973, "step": 6206 }, { "epoch": 0.22113757414895702, "grad_norm": 1.4787559915066002, "learning_rate": 1.8144755375053402e-05, "loss": 0.5441, "step": 6207 }, { "epoch": 0.22117320127545112, "grad_norm": 1.7558396461241614, "learning_rate": 1.8144085809185184e-05, "loss": 0.5949, "step": 6208 }, { "epoch": 0.22120882840194525, "grad_norm": 1.4119688281356453, "learning_rate": 1.8143416134872775e-05, "loss": 0.5139, "step": 6209 }, { "epoch": 0.22124445552843935, "grad_norm": 1.6768186195228076, "learning_rate": 1.814274635212509e-05, "loss": 0.7351, "step": 6210 }, { "epoch": 0.22128008265493346, "grad_norm": 1.524743303021637, "learning_rate": 1.814207646095104e-05, "loss": 0.5174, "step": 6211 }, { "epoch": 0.2213157097814276, "grad_norm": 1.652740543674031, "learning_rate": 1.8141406461359553e-05, "loss": 0.644, "step": 6212 }, { "epoch": 0.2213513369079217, "grad_norm": 1.7834650788284172, "learning_rate": 1.8140736353359546e-05, "loss": 0.5439, "step": 6213 }, { "epoch": 0.2213869640344158, "grad_norm": 1.4122112065581325, "learning_rate": 1.8140066136959945e-05, "loss": 0.6107, "step": 6214 }, { "epoch": 0.22142259116090993, "grad_norm": 2.1838355845921096, "learning_rate": 1.813939581216968e-05, "loss": 0.6877, "step": 6215 }, { "epoch": 0.22145821828740403, "grad_norm": 1.2691702753926322, "learning_rate": 1.8138725378997667e-05, "loss": 0.5242, "step": 6216 }, { "epoch": 0.22149384541389813, "grad_norm": 1.2098980233976857, "learning_rate": 1.8138054837452834e-05, "loss": 0.5139, "step": 6217 }, { "epoch": 0.22152947254039226, "grad_norm": 1.830137725903908, "learning_rate": 1.8137384187544116e-05, "loss": 0.5921, "step": 6218 }, { "epoch": 0.22156509966688637, "grad_norm": 1.4082102050719802, "learning_rate": 1.8136713429280435e-05, "loss": 0.4915, "step": 6219 }, { "epoch": 0.22160072679338047, "grad_norm": 1.5047452729358854, "learning_rate": 1.813604256267073e-05, "loss": 0.483, "step": 6220 }, { "epoch": 0.2216363539198746, "grad_norm": 1.8630168468549524, "learning_rate": 1.8135371587723937e-05, "loss": 0.6141, "step": 6221 }, { "epoch": 0.2216719810463687, "grad_norm": 2.0011523769087214, "learning_rate": 1.8134700504448982e-05, "loss": 0.6668, "step": 6222 }, { "epoch": 0.2217076081728628, "grad_norm": 1.5816460958290914, "learning_rate": 1.8134029312854803e-05, "loss": 0.5232, "step": 6223 }, { "epoch": 0.22174323529935694, "grad_norm": 1.3642098451745828, "learning_rate": 1.8133358012950338e-05, "loss": 0.4981, "step": 6224 }, { "epoch": 0.22177886242585104, "grad_norm": 1.2736786310669288, "learning_rate": 1.8132686604744524e-05, "loss": 0.5922, "step": 6225 }, { "epoch": 0.22181448955234515, "grad_norm": 1.2470315833807686, "learning_rate": 1.8132015088246308e-05, "loss": 0.5088, "step": 6226 }, { "epoch": 0.22185011667883928, "grad_norm": 1.586697137026035, "learning_rate": 1.8131343463464622e-05, "loss": 0.5953, "step": 6227 }, { "epoch": 0.22188574380533338, "grad_norm": 1.4480667964311396, "learning_rate": 1.813067173040842e-05, "loss": 0.5327, "step": 6228 }, { "epoch": 0.22192137093182748, "grad_norm": 1.560054794884218, "learning_rate": 1.8129999889086637e-05, "loss": 0.6632, "step": 6229 }, { "epoch": 0.22195699805832161, "grad_norm": 1.6724630182595477, "learning_rate": 1.8129327939508227e-05, "loss": 0.5265, "step": 6230 }, { "epoch": 0.22199262518481572, "grad_norm": 1.4795277933926168, "learning_rate": 1.812865588168213e-05, "loss": 0.5932, "step": 6231 }, { "epoch": 0.22202825231130982, "grad_norm": 1.4224536811169568, "learning_rate": 1.81279837156173e-05, "loss": 0.5514, "step": 6232 }, { "epoch": 0.22206387943780395, "grad_norm": 1.394230819560723, "learning_rate": 1.812731144132268e-05, "loss": 0.5019, "step": 6233 }, { "epoch": 0.22209950656429805, "grad_norm": 1.3177187683708214, "learning_rate": 1.8126639058807235e-05, "loss": 0.5198, "step": 6234 }, { "epoch": 0.22213513369079216, "grad_norm": 1.5484161747704333, "learning_rate": 1.8125966568079908e-05, "loss": 0.5722, "step": 6235 }, { "epoch": 0.2221707608172863, "grad_norm": 1.3127086611693755, "learning_rate": 1.8125293969149655e-05, "loss": 0.5011, "step": 6236 }, { "epoch": 0.2222063879437804, "grad_norm": 1.2110329189656879, "learning_rate": 1.8124621262025434e-05, "loss": 0.5689, "step": 6237 }, { "epoch": 0.2222420150702745, "grad_norm": 1.7729143867661434, "learning_rate": 1.8123948446716204e-05, "loss": 0.5463, "step": 6238 }, { "epoch": 0.22227764219676863, "grad_norm": 1.6381241558164945, "learning_rate": 1.812327552323092e-05, "loss": 0.5679, "step": 6239 }, { "epoch": 0.22231326932326273, "grad_norm": 1.3127989282240191, "learning_rate": 1.8122602491578546e-05, "loss": 0.5202, "step": 6240 }, { "epoch": 0.22234889644975683, "grad_norm": 1.1031073611379574, "learning_rate": 1.8121929351768042e-05, "loss": 0.533, "step": 6241 }, { "epoch": 0.22238452357625096, "grad_norm": 1.2920552548407134, "learning_rate": 1.8121256103808372e-05, "loss": 0.5935, "step": 6242 }, { "epoch": 0.22242015070274507, "grad_norm": 1.1873405224801779, "learning_rate": 1.8120582747708503e-05, "loss": 0.4654, "step": 6243 }, { "epoch": 0.22245577782923917, "grad_norm": 1.6552800833352095, "learning_rate": 1.8119909283477394e-05, "loss": 0.5385, "step": 6244 }, { "epoch": 0.2224914049557333, "grad_norm": 1.6617521868087817, "learning_rate": 1.811923571112402e-05, "loss": 0.599, "step": 6245 }, { "epoch": 0.2225270320822274, "grad_norm": 1.1777255121358838, "learning_rate": 1.8118562030657347e-05, "loss": 0.4966, "step": 6246 }, { "epoch": 0.2225626592087215, "grad_norm": 1.7346102257344262, "learning_rate": 1.811788824208635e-05, "loss": 0.5471, "step": 6247 }, { "epoch": 0.22259828633521564, "grad_norm": 1.3856162575736017, "learning_rate": 1.811721434541999e-05, "loss": 0.6358, "step": 6248 }, { "epoch": 0.22263391346170974, "grad_norm": 1.3419490076904819, "learning_rate": 1.811654034066725e-05, "loss": 0.5262, "step": 6249 }, { "epoch": 0.22266954058820385, "grad_norm": 1.2534575394240575, "learning_rate": 1.81158662278371e-05, "loss": 0.5058, "step": 6250 }, { "epoch": 0.22270516771469798, "grad_norm": 1.3403550832904705, "learning_rate": 1.8115192006938525e-05, "loss": 0.5283, "step": 6251 }, { "epoch": 0.22274079484119208, "grad_norm": 1.781365529354121, "learning_rate": 1.8114517677980493e-05, "loss": 0.6148, "step": 6252 }, { "epoch": 0.22277642196768618, "grad_norm": 1.6496919930247511, "learning_rate": 1.8113843240971983e-05, "loss": 0.5606, "step": 6253 }, { "epoch": 0.22281204909418031, "grad_norm": 1.6095620120025107, "learning_rate": 1.8113168695921985e-05, "loss": 0.6775, "step": 6254 }, { "epoch": 0.22284767622067442, "grad_norm": 1.1730506742747708, "learning_rate": 1.811249404283947e-05, "loss": 0.4292, "step": 6255 }, { "epoch": 0.22288330334716852, "grad_norm": 1.3907865228436047, "learning_rate": 1.811181928173343e-05, "loss": 0.5182, "step": 6256 }, { "epoch": 0.22291893047366265, "grad_norm": 1.9354988735720176, "learning_rate": 1.8111144412612846e-05, "loss": 0.6565, "step": 6257 }, { "epoch": 0.22295455760015676, "grad_norm": 2.077494305789138, "learning_rate": 1.8110469435486703e-05, "loss": 0.6632, "step": 6258 }, { "epoch": 0.22299018472665086, "grad_norm": 1.6607477336387775, "learning_rate": 1.8109794350363992e-05, "loss": 0.6025, "step": 6259 }, { "epoch": 0.223025811853145, "grad_norm": 1.6103371134907292, "learning_rate": 1.8109119157253697e-05, "loss": 0.5451, "step": 6260 }, { "epoch": 0.2230614389796391, "grad_norm": 2.12026609447225, "learning_rate": 1.8108443856164818e-05, "loss": 0.6209, "step": 6261 }, { "epoch": 0.2230970661061332, "grad_norm": 1.286569882773837, "learning_rate": 1.810776844710634e-05, "loss": 0.4958, "step": 6262 }, { "epoch": 0.22313269323262733, "grad_norm": 2.5664436213908175, "learning_rate": 1.8107092930087258e-05, "loss": 0.6736, "step": 6263 }, { "epoch": 0.22316832035912143, "grad_norm": 1.3109361888040216, "learning_rate": 1.8106417305116565e-05, "loss": 0.5555, "step": 6264 }, { "epoch": 0.22320394748561556, "grad_norm": 1.2216617457254193, "learning_rate": 1.8105741572203262e-05, "loss": 0.5037, "step": 6265 }, { "epoch": 0.22323957461210966, "grad_norm": 1.160389285981704, "learning_rate": 1.8105065731356343e-05, "loss": 0.4737, "step": 6266 }, { "epoch": 0.22327520173860377, "grad_norm": 1.9354069964520773, "learning_rate": 1.810438978258481e-05, "loss": 0.5039, "step": 6267 }, { "epoch": 0.2233108288650979, "grad_norm": 1.6873151767029575, "learning_rate": 1.8103713725897664e-05, "loss": 0.6083, "step": 6268 }, { "epoch": 0.223346455991592, "grad_norm": 1.2145155205870288, "learning_rate": 1.8103037561303903e-05, "loss": 0.5431, "step": 6269 }, { "epoch": 0.2233820831180861, "grad_norm": 1.2595037027840177, "learning_rate": 1.810236128881253e-05, "loss": 0.5356, "step": 6270 }, { "epoch": 0.22341771024458024, "grad_norm": 1.3798716049764874, "learning_rate": 1.810168490843256e-05, "loss": 0.5305, "step": 6271 }, { "epoch": 0.22345333737107434, "grad_norm": 2.1007603057057604, "learning_rate": 1.8101008420172984e-05, "loss": 0.6101, "step": 6272 }, { "epoch": 0.22348896449756844, "grad_norm": 2.497778996338395, "learning_rate": 1.8100331824042828e-05, "loss": 0.5982, "step": 6273 }, { "epoch": 0.22352459162406257, "grad_norm": 1.701147940268256, "learning_rate": 1.8099655120051087e-05, "loss": 0.5743, "step": 6274 }, { "epoch": 0.22356021875055668, "grad_norm": 1.4710695268420224, "learning_rate": 1.8098978308206775e-05, "loss": 0.5586, "step": 6275 }, { "epoch": 0.22359584587705078, "grad_norm": 1.1452123530147875, "learning_rate": 1.809830138851891e-05, "loss": 0.5009, "step": 6276 }, { "epoch": 0.2236314730035449, "grad_norm": 1.4952831580506494, "learning_rate": 1.80976243609965e-05, "loss": 0.6445, "step": 6277 }, { "epoch": 0.22366710013003901, "grad_norm": 1.7094130751781735, "learning_rate": 1.8096947225648566e-05, "loss": 0.5323, "step": 6278 }, { "epoch": 0.22370272725653312, "grad_norm": 1.1398332238368514, "learning_rate": 1.8096269982484114e-05, "loss": 0.5114, "step": 6279 }, { "epoch": 0.22373835438302725, "grad_norm": 1.6258938120303306, "learning_rate": 1.809559263151217e-05, "loss": 0.6574, "step": 6280 }, { "epoch": 0.22377398150952135, "grad_norm": 1.4089423648640191, "learning_rate": 1.809491517274175e-05, "loss": 0.5361, "step": 6281 }, { "epoch": 0.22380960863601546, "grad_norm": 1.6341403835426653, "learning_rate": 1.8094237606181878e-05, "loss": 0.6251, "step": 6282 }, { "epoch": 0.2238452357625096, "grad_norm": 1.9791799445079428, "learning_rate": 1.809355993184158e-05, "loss": 0.5572, "step": 6283 }, { "epoch": 0.2238808628890037, "grad_norm": 1.1335313146878887, "learning_rate": 1.8092882149729867e-05, "loss": 0.4688, "step": 6284 }, { "epoch": 0.2239164900154978, "grad_norm": 1.1915149009060912, "learning_rate": 1.8092204259855773e-05, "loss": 0.5244, "step": 6285 }, { "epoch": 0.22395211714199192, "grad_norm": 1.1633605665186317, "learning_rate": 1.8091526262228328e-05, "loss": 0.5194, "step": 6286 }, { "epoch": 0.22398774426848603, "grad_norm": 1.5305479916187095, "learning_rate": 1.809084815685655e-05, "loss": 0.6394, "step": 6287 }, { "epoch": 0.22402337139498013, "grad_norm": 1.2368112346893665, "learning_rate": 1.8090169943749477e-05, "loss": 0.5579, "step": 6288 }, { "epoch": 0.22405899852147426, "grad_norm": 1.1430463274218743, "learning_rate": 1.8089491622916133e-05, "loss": 0.4906, "step": 6289 }, { "epoch": 0.22409462564796837, "grad_norm": 1.386281849521587, "learning_rate": 1.8088813194365555e-05, "loss": 0.5327, "step": 6290 }, { "epoch": 0.22413025277446247, "grad_norm": 2.2983017230885996, "learning_rate": 1.808813465810678e-05, "loss": 0.5889, "step": 6291 }, { "epoch": 0.2241658799009566, "grad_norm": 1.7095506164230645, "learning_rate": 1.808745601414883e-05, "loss": 0.6677, "step": 6292 }, { "epoch": 0.2242015070274507, "grad_norm": 1.7775596841374228, "learning_rate": 1.808677726250076e-05, "loss": 0.5264, "step": 6293 }, { "epoch": 0.2242371341539448, "grad_norm": 1.9580261832772932, "learning_rate": 1.8086098403171593e-05, "loss": 0.5473, "step": 6294 }, { "epoch": 0.22427276128043894, "grad_norm": 1.719819113134554, "learning_rate": 1.8085419436170373e-05, "loss": 0.6008, "step": 6295 }, { "epoch": 0.22430838840693304, "grad_norm": 1.6808853153345213, "learning_rate": 1.8084740361506143e-05, "loss": 0.5603, "step": 6296 }, { "epoch": 0.22434401553342714, "grad_norm": 1.3216441658234093, "learning_rate": 1.8084061179187945e-05, "loss": 0.5275, "step": 6297 }, { "epoch": 0.22437964265992127, "grad_norm": 1.9431686918735014, "learning_rate": 1.8083381889224822e-05, "loss": 0.5939, "step": 6298 }, { "epoch": 0.22441526978641538, "grad_norm": 1.1894383736035437, "learning_rate": 1.808270249162582e-05, "loss": 0.5139, "step": 6299 }, { "epoch": 0.22445089691290948, "grad_norm": 1.9396300211739406, "learning_rate": 1.808202298639998e-05, "loss": 0.6189, "step": 6300 }, { "epoch": 0.2244865240394036, "grad_norm": 1.3862824782094387, "learning_rate": 1.8081343373556363e-05, "loss": 0.5914, "step": 6301 }, { "epoch": 0.22452215116589772, "grad_norm": 1.279916551853088, "learning_rate": 1.8080663653104005e-05, "loss": 0.497, "step": 6302 }, { "epoch": 0.22455777829239182, "grad_norm": 1.1676780137025033, "learning_rate": 1.8079983825051964e-05, "loss": 0.5077, "step": 6303 }, { "epoch": 0.22459340541888595, "grad_norm": 1.151480794929258, "learning_rate": 1.807930388940929e-05, "loss": 0.5071, "step": 6304 }, { "epoch": 0.22462903254538005, "grad_norm": 1.5668352159987426, "learning_rate": 1.807862384618504e-05, "loss": 0.6358, "step": 6305 }, { "epoch": 0.22466465967187416, "grad_norm": 1.2154561416447611, "learning_rate": 1.8077943695388262e-05, "loss": 0.5469, "step": 6306 }, { "epoch": 0.2247002867983683, "grad_norm": 2.0242513066607715, "learning_rate": 1.807726343702802e-05, "loss": 0.6652, "step": 6307 }, { "epoch": 0.2247359139248624, "grad_norm": 1.3809407688741722, "learning_rate": 1.8076583071113372e-05, "loss": 0.5877, "step": 6308 }, { "epoch": 0.2247715410513565, "grad_norm": 1.318931952874285, "learning_rate": 1.8075902597653372e-05, "loss": 0.5252, "step": 6309 }, { "epoch": 0.22480716817785062, "grad_norm": 1.6805716771626584, "learning_rate": 1.807522201665709e-05, "loss": 0.5561, "step": 6310 }, { "epoch": 0.22484279530434473, "grad_norm": 1.1682995756294832, "learning_rate": 1.8074541328133574e-05, "loss": 0.5261, "step": 6311 }, { "epoch": 0.22487842243083883, "grad_norm": 1.8942961536613994, "learning_rate": 1.80738605320919e-05, "loss": 0.6802, "step": 6312 }, { "epoch": 0.22491404955733296, "grad_norm": 1.7629155286541516, "learning_rate": 1.807317962854113e-05, "loss": 0.6018, "step": 6313 }, { "epoch": 0.22494967668382707, "grad_norm": 1.2027086471789923, "learning_rate": 1.8072498617490334e-05, "loss": 0.4906, "step": 6314 }, { "epoch": 0.22498530381032117, "grad_norm": 1.7802166373693664, "learning_rate": 1.807181749894857e-05, "loss": 0.5761, "step": 6315 }, { "epoch": 0.2250209309368153, "grad_norm": 1.2713243403891579, "learning_rate": 1.807113627292492e-05, "loss": 0.5114, "step": 6316 }, { "epoch": 0.2250565580633094, "grad_norm": 1.532251394447305, "learning_rate": 1.8070454939428446e-05, "loss": 0.5773, "step": 6317 }, { "epoch": 0.2250921851898035, "grad_norm": 1.1623031539747275, "learning_rate": 1.8069773498468224e-05, "loss": 0.5579, "step": 6318 }, { "epoch": 0.22512781231629764, "grad_norm": 1.5328209733109082, "learning_rate": 1.8069091950053324e-05, "loss": 0.6533, "step": 6319 }, { "epoch": 0.22516343944279174, "grad_norm": 1.2438567444687156, "learning_rate": 1.806841029419283e-05, "loss": 0.4574, "step": 6320 }, { "epoch": 0.22519906656928584, "grad_norm": 1.1997672957523375, "learning_rate": 1.8067728530895812e-05, "loss": 0.5465, "step": 6321 }, { "epoch": 0.22523469369577998, "grad_norm": 1.5746242137080992, "learning_rate": 1.806704666017135e-05, "loss": 0.5829, "step": 6322 }, { "epoch": 0.22527032082227408, "grad_norm": 1.0980971778253006, "learning_rate": 1.8066364682028522e-05, "loss": 0.5032, "step": 6323 }, { "epoch": 0.22530594794876818, "grad_norm": 2.104259139580532, "learning_rate": 1.8065682596476413e-05, "loss": 0.6301, "step": 6324 }, { "epoch": 0.2253415750752623, "grad_norm": 1.155376235310686, "learning_rate": 1.80650004035241e-05, "loss": 0.4619, "step": 6325 }, { "epoch": 0.22537720220175642, "grad_norm": 1.47987517524917, "learning_rate": 1.806431810318067e-05, "loss": 0.6282, "step": 6326 }, { "epoch": 0.22541282932825052, "grad_norm": 1.5584308230617645, "learning_rate": 1.806363569545521e-05, "loss": 0.5506, "step": 6327 }, { "epoch": 0.22544845645474465, "grad_norm": 1.57718433089158, "learning_rate": 1.80629531803568e-05, "loss": 0.5443, "step": 6328 }, { "epoch": 0.22548408358123875, "grad_norm": 1.48631322880203, "learning_rate": 1.8062270557894533e-05, "loss": 0.6538, "step": 6329 }, { "epoch": 0.22551971070773286, "grad_norm": 1.1955465443562003, "learning_rate": 1.8061587828077503e-05, "loss": 0.5339, "step": 6330 }, { "epoch": 0.225555337834227, "grad_norm": 1.2253668853188744, "learning_rate": 1.806090499091479e-05, "loss": 0.548, "step": 6331 }, { "epoch": 0.2255909649607211, "grad_norm": 1.1335198917938698, "learning_rate": 1.8060222046415502e-05, "loss": 0.5426, "step": 6332 }, { "epoch": 0.2256265920872152, "grad_norm": 1.6744232261271375, "learning_rate": 1.8059538994588715e-05, "loss": 0.5756, "step": 6333 }, { "epoch": 0.22566221921370933, "grad_norm": 1.7267377235455181, "learning_rate": 1.805885583544354e-05, "loss": 0.579, "step": 6334 }, { "epoch": 0.22569784634020343, "grad_norm": 1.5425204326962878, "learning_rate": 1.8058172568989062e-05, "loss": 0.5684, "step": 6335 }, { "epoch": 0.22573347346669753, "grad_norm": 1.7061341570764499, "learning_rate": 1.8057489195234384e-05, "loss": 0.6051, "step": 6336 }, { "epoch": 0.22576910059319166, "grad_norm": 1.5899841952791862, "learning_rate": 1.8056805714188606e-05, "loss": 0.5533, "step": 6337 }, { "epoch": 0.22580472771968577, "grad_norm": 1.550190349942397, "learning_rate": 1.8056122125860827e-05, "loss": 0.6263, "step": 6338 }, { "epoch": 0.22584035484617987, "grad_norm": 3.964064526606871, "learning_rate": 1.8055438430260154e-05, "loss": 0.6003, "step": 6339 }, { "epoch": 0.225875981972674, "grad_norm": 1.180453813821479, "learning_rate": 1.8054754627395687e-05, "loss": 0.5134, "step": 6340 }, { "epoch": 0.2259116090991681, "grad_norm": 1.932908951512307, "learning_rate": 1.805407071727653e-05, "loss": 0.5899, "step": 6341 }, { "epoch": 0.2259472362256622, "grad_norm": 2.1208570787382817, "learning_rate": 1.8053386699911795e-05, "loss": 0.5932, "step": 6342 }, { "epoch": 0.22598286335215634, "grad_norm": 1.9512686625643563, "learning_rate": 1.8052702575310588e-05, "loss": 0.6603, "step": 6343 }, { "epoch": 0.22601849047865044, "grad_norm": 1.3830969495458834, "learning_rate": 1.8052018343482016e-05, "loss": 0.5263, "step": 6344 }, { "epoch": 0.22605411760514454, "grad_norm": 1.0973668832196217, "learning_rate": 1.805133400443519e-05, "loss": 0.4648, "step": 6345 }, { "epoch": 0.22608974473163868, "grad_norm": 1.184472779323277, "learning_rate": 1.805064955817923e-05, "loss": 0.4889, "step": 6346 }, { "epoch": 0.22612537185813278, "grad_norm": 1.1460150983008295, "learning_rate": 1.8049965004723238e-05, "loss": 0.5135, "step": 6347 }, { "epoch": 0.22616099898462688, "grad_norm": 1.0937380820271285, "learning_rate": 1.8049280344076335e-05, "loss": 0.4701, "step": 6348 }, { "epoch": 0.226196626111121, "grad_norm": 1.1143170038573271, "learning_rate": 1.8048595576247642e-05, "loss": 0.5324, "step": 6349 }, { "epoch": 0.22623225323761512, "grad_norm": 1.1411904879290964, "learning_rate": 1.804791070124627e-05, "loss": 0.5385, "step": 6350 }, { "epoch": 0.22626788036410922, "grad_norm": 1.8142628876863607, "learning_rate": 1.8047225719081344e-05, "loss": 0.6425, "step": 6351 }, { "epoch": 0.22630350749060335, "grad_norm": 1.1936658425744038, "learning_rate": 1.8046540629761983e-05, "loss": 0.5116, "step": 6352 }, { "epoch": 0.22633913461709745, "grad_norm": 1.6876405235476015, "learning_rate": 1.804585543329731e-05, "loss": 0.5806, "step": 6353 }, { "epoch": 0.22637476174359156, "grad_norm": 2.2978040437918503, "learning_rate": 1.8045170129696444e-05, "loss": 0.6564, "step": 6354 }, { "epoch": 0.2264103888700857, "grad_norm": 1.5090685477113912, "learning_rate": 1.8044484718968517e-05, "loss": 0.6248, "step": 6355 }, { "epoch": 0.2264460159965798, "grad_norm": 1.1025652587275028, "learning_rate": 1.8043799201122656e-05, "loss": 0.5569, "step": 6356 }, { "epoch": 0.2264816431230739, "grad_norm": 1.2327893410379125, "learning_rate": 1.8043113576167985e-05, "loss": 0.5021, "step": 6357 }, { "epoch": 0.22651727024956803, "grad_norm": 1.1793718299268823, "learning_rate": 1.804242784411363e-05, "loss": 0.5149, "step": 6358 }, { "epoch": 0.22655289737606213, "grad_norm": 1.1140981039442066, "learning_rate": 1.804174200496873e-05, "loss": 0.5384, "step": 6359 }, { "epoch": 0.22658852450255626, "grad_norm": 1.9480691533041599, "learning_rate": 1.8041056058742418e-05, "loss": 0.6384, "step": 6360 }, { "epoch": 0.22662415162905036, "grad_norm": 1.7282753106704967, "learning_rate": 1.8040370005443822e-05, "loss": 0.6151, "step": 6361 }, { "epoch": 0.22665977875554447, "grad_norm": 1.197136046559964, "learning_rate": 1.8039683845082077e-05, "loss": 0.5418, "step": 6362 }, { "epoch": 0.2266954058820386, "grad_norm": 1.823124375942918, "learning_rate": 1.8038997577666326e-05, "loss": 0.6764, "step": 6363 }, { "epoch": 0.2267310330085327, "grad_norm": 1.2031181155356059, "learning_rate": 1.8038311203205698e-05, "loss": 0.5061, "step": 6364 }, { "epoch": 0.2267666601350268, "grad_norm": 1.217597921361505, "learning_rate": 1.8037624721709344e-05, "loss": 0.5147, "step": 6365 }, { "epoch": 0.22680228726152094, "grad_norm": 1.8018506139021973, "learning_rate": 1.8036938133186397e-05, "loss": 0.637, "step": 6366 }, { "epoch": 0.22683791438801504, "grad_norm": 1.087162805299059, "learning_rate": 1.8036251437645998e-05, "loss": 0.4792, "step": 6367 }, { "epoch": 0.22687354151450914, "grad_norm": 2.306047753005705, "learning_rate": 1.8035564635097296e-05, "loss": 0.6051, "step": 6368 }, { "epoch": 0.22690916864100327, "grad_norm": 1.553035089323771, "learning_rate": 1.8034877725549435e-05, "loss": 0.6599, "step": 6369 }, { "epoch": 0.22694479576749738, "grad_norm": 1.1209947166476981, "learning_rate": 1.803419070901156e-05, "loss": 0.5083, "step": 6370 }, { "epoch": 0.22698042289399148, "grad_norm": 1.4953387558768494, "learning_rate": 1.8033503585492823e-05, "loss": 0.5867, "step": 6371 }, { "epoch": 0.2270160500204856, "grad_norm": 1.45190653732672, "learning_rate": 1.803281635500237e-05, "loss": 0.5886, "step": 6372 }, { "epoch": 0.2270516771469797, "grad_norm": 1.5532447946273793, "learning_rate": 1.8032129017549353e-05, "loss": 0.598, "step": 6373 }, { "epoch": 0.22708730427347382, "grad_norm": 1.1445550362423076, "learning_rate": 1.803144157314292e-05, "loss": 0.5388, "step": 6374 }, { "epoch": 0.22712293139996795, "grad_norm": 1.2737020478074403, "learning_rate": 1.803075402179223e-05, "loss": 0.5618, "step": 6375 }, { "epoch": 0.22715855852646205, "grad_norm": 2.0542165681230857, "learning_rate": 1.8030066363506438e-05, "loss": 0.6711, "step": 6376 }, { "epoch": 0.22719418565295615, "grad_norm": 1.457186894421051, "learning_rate": 1.80293785982947e-05, "loss": 0.5751, "step": 6377 }, { "epoch": 0.22722981277945029, "grad_norm": 1.2318655740103877, "learning_rate": 1.8028690726166172e-05, "loss": 0.4943, "step": 6378 }, { "epoch": 0.2272654399059444, "grad_norm": 1.211490787314067, "learning_rate": 1.802800274713002e-05, "loss": 0.5458, "step": 6379 }, { "epoch": 0.2273010670324385, "grad_norm": 1.1906117020410567, "learning_rate": 1.8027314661195397e-05, "loss": 0.5354, "step": 6380 }, { "epoch": 0.22733669415893262, "grad_norm": 1.4606960668975004, "learning_rate": 1.8026626468371468e-05, "loss": 0.509, "step": 6381 }, { "epoch": 0.22737232128542673, "grad_norm": 1.1676156225692678, "learning_rate": 1.8025938168667398e-05, "loss": 0.537, "step": 6382 }, { "epoch": 0.22740794841192083, "grad_norm": 1.0873167556353343, "learning_rate": 1.802524976209235e-05, "loss": 0.5084, "step": 6383 }, { "epoch": 0.22744357553841496, "grad_norm": 1.12599776906091, "learning_rate": 1.8024561248655494e-05, "loss": 0.5249, "step": 6384 }, { "epoch": 0.22747920266490906, "grad_norm": 2.093456484238456, "learning_rate": 1.8023872628365997e-05, "loss": 0.6159, "step": 6385 }, { "epoch": 0.22751482979140317, "grad_norm": 1.1969041571382504, "learning_rate": 1.8023183901233024e-05, "loss": 0.4997, "step": 6386 }, { "epoch": 0.2275504569178973, "grad_norm": 2.075048696552291, "learning_rate": 1.802249506726575e-05, "loss": 0.6589, "step": 6387 }, { "epoch": 0.2275860840443914, "grad_norm": 1.0868819611727212, "learning_rate": 1.8021806126473353e-05, "loss": 0.5071, "step": 6388 }, { "epoch": 0.2276217111708855, "grad_norm": 1.1478443466279578, "learning_rate": 1.8021117078864994e-05, "loss": 0.4882, "step": 6389 }, { "epoch": 0.22765733829737964, "grad_norm": 1.865118524713993, "learning_rate": 1.802042792444986e-05, "loss": 0.5994, "step": 6390 }, { "epoch": 0.22769296542387374, "grad_norm": 1.7265994664769342, "learning_rate": 1.8019738663237115e-05, "loss": 0.5092, "step": 6391 }, { "epoch": 0.22772859255036784, "grad_norm": 1.0818923895766597, "learning_rate": 1.801904929523595e-05, "loss": 0.4962, "step": 6392 }, { "epoch": 0.22776421967686197, "grad_norm": 1.1799872502959055, "learning_rate": 1.8018359820455535e-05, "loss": 0.6008, "step": 6393 }, { "epoch": 0.22779984680335608, "grad_norm": 1.1748339845308338, "learning_rate": 1.8017670238905057e-05, "loss": 0.5174, "step": 6394 }, { "epoch": 0.22783547392985018, "grad_norm": 1.2769175281557033, "learning_rate": 1.8016980550593695e-05, "loss": 0.4994, "step": 6395 }, { "epoch": 0.2278711010563443, "grad_norm": 2.4637822260048337, "learning_rate": 1.8016290755530636e-05, "loss": 0.6124, "step": 6396 }, { "epoch": 0.22790672818283841, "grad_norm": 1.6096222094156598, "learning_rate": 1.801560085372506e-05, "loss": 0.5609, "step": 6397 }, { "epoch": 0.22794235530933252, "grad_norm": 1.1497020120090906, "learning_rate": 1.8014910845186154e-05, "loss": 0.476, "step": 6398 }, { "epoch": 0.22797798243582665, "grad_norm": 1.1265433820479656, "learning_rate": 1.801422072992311e-05, "loss": 0.5053, "step": 6399 }, { "epoch": 0.22801360956232075, "grad_norm": 1.5511429904425507, "learning_rate": 1.8013530507945115e-05, "loss": 0.6329, "step": 6400 }, { "epoch": 0.22804923668881485, "grad_norm": 1.1437635120526466, "learning_rate": 1.801284017926136e-05, "loss": 0.5329, "step": 6401 }, { "epoch": 0.22808486381530899, "grad_norm": 1.7700750471657531, "learning_rate": 1.801214974388104e-05, "loss": 0.6118, "step": 6402 }, { "epoch": 0.2281204909418031, "grad_norm": 1.2397052561820359, "learning_rate": 1.8011459201813345e-05, "loss": 0.5053, "step": 6403 }, { "epoch": 0.2281561180682972, "grad_norm": 1.509967270162392, "learning_rate": 1.801076855306747e-05, "loss": 0.5412, "step": 6404 }, { "epoch": 0.22819174519479132, "grad_norm": 1.1317988407140793, "learning_rate": 1.8010077797652613e-05, "loss": 0.5018, "step": 6405 }, { "epoch": 0.22822737232128543, "grad_norm": 1.5765708148977549, "learning_rate": 1.800938693557797e-05, "loss": 0.569, "step": 6406 }, { "epoch": 0.22826299944777953, "grad_norm": 1.1571050558239384, "learning_rate": 1.8008695966852743e-05, "loss": 0.522, "step": 6407 }, { "epoch": 0.22829862657427366, "grad_norm": 1.8337033863150227, "learning_rate": 1.800800489148613e-05, "loss": 0.6258, "step": 6408 }, { "epoch": 0.22833425370076776, "grad_norm": 1.0679419375185555, "learning_rate": 1.8007313709487334e-05, "loss": 0.4503, "step": 6409 }, { "epoch": 0.22836988082726187, "grad_norm": 1.129733185069674, "learning_rate": 1.8006622420865564e-05, "loss": 0.5083, "step": 6410 }, { "epoch": 0.228405507953756, "grad_norm": 1.1189665990829856, "learning_rate": 1.8005931025630016e-05, "loss": 0.5335, "step": 6411 }, { "epoch": 0.2284411350802501, "grad_norm": 1.3925905732369515, "learning_rate": 1.8005239523789903e-05, "loss": 0.5622, "step": 6412 }, { "epoch": 0.2284767622067442, "grad_norm": 1.2566897734835913, "learning_rate": 1.800454791535443e-05, "loss": 0.5142, "step": 6413 }, { "epoch": 0.22851238933323834, "grad_norm": 1.1096478465186437, "learning_rate": 1.8003856200332804e-05, "loss": 0.4981, "step": 6414 }, { "epoch": 0.22854801645973244, "grad_norm": 1.6051035826523703, "learning_rate": 1.800316437873424e-05, "loss": 0.5807, "step": 6415 }, { "epoch": 0.22858364358622654, "grad_norm": 1.1454658440473553, "learning_rate": 1.800247245056795e-05, "loss": 0.4603, "step": 6416 }, { "epoch": 0.22861927071272067, "grad_norm": 1.4615690507578585, "learning_rate": 1.800178041584314e-05, "loss": 0.6107, "step": 6417 }, { "epoch": 0.22865489783921478, "grad_norm": 1.189483770320478, "learning_rate": 1.800108827456904e-05, "loss": 0.4688, "step": 6418 }, { "epoch": 0.22869052496570888, "grad_norm": 1.1549199993316304, "learning_rate": 1.800039602675485e-05, "loss": 0.4911, "step": 6419 }, { "epoch": 0.228726152092203, "grad_norm": 1.3025780543442798, "learning_rate": 1.7999703672409802e-05, "loss": 0.5447, "step": 6420 }, { "epoch": 0.22876177921869711, "grad_norm": 1.4626821898395512, "learning_rate": 1.79990112115431e-05, "loss": 0.6662, "step": 6421 }, { "epoch": 0.22879740634519122, "grad_norm": 1.758555642884624, "learning_rate": 1.799831864416398e-05, "loss": 0.5629, "step": 6422 }, { "epoch": 0.22883303347168535, "grad_norm": 2.4148452960596383, "learning_rate": 1.7997625970281652e-05, "loss": 0.5748, "step": 6423 }, { "epoch": 0.22886866059817945, "grad_norm": 1.8311534742762132, "learning_rate": 1.7996933189905347e-05, "loss": 0.6153, "step": 6424 }, { "epoch": 0.22890428772467356, "grad_norm": 1.7432521430063104, "learning_rate": 1.799624030304428e-05, "loss": 0.5739, "step": 6425 }, { "epoch": 0.2289399148511677, "grad_norm": 1.1517650600613911, "learning_rate": 1.799554730970769e-05, "loss": 0.5249, "step": 6426 }, { "epoch": 0.2289755419776618, "grad_norm": 1.620245455502486, "learning_rate": 1.79948542099048e-05, "loss": 0.6138, "step": 6427 }, { "epoch": 0.2290111691041559, "grad_norm": 1.2460289891307923, "learning_rate": 1.7994161003644838e-05, "loss": 0.5301, "step": 6428 }, { "epoch": 0.22904679623065002, "grad_norm": 1.5297092300762156, "learning_rate": 1.7993467690937033e-05, "loss": 0.6135, "step": 6429 }, { "epoch": 0.22908242335714413, "grad_norm": 1.3211736491369328, "learning_rate": 1.799277427179062e-05, "loss": 0.5428, "step": 6430 }, { "epoch": 0.22911805048363823, "grad_norm": 1.2894617663251289, "learning_rate": 1.799208074621483e-05, "loss": 0.5195, "step": 6431 }, { "epoch": 0.22915367761013236, "grad_norm": 1.2312023250233446, "learning_rate": 1.79913871142189e-05, "loss": 0.5366, "step": 6432 }, { "epoch": 0.22918930473662646, "grad_norm": 1.209499485623483, "learning_rate": 1.799069337581206e-05, "loss": 0.5176, "step": 6433 }, { "epoch": 0.22922493186312057, "grad_norm": 1.4964986777156868, "learning_rate": 1.7989999531003558e-05, "loss": 0.5492, "step": 6434 }, { "epoch": 0.2292605589896147, "grad_norm": 1.162841665927398, "learning_rate": 1.7989305579802625e-05, "loss": 0.5214, "step": 6435 }, { "epoch": 0.2292961861161088, "grad_norm": 1.6312764440996748, "learning_rate": 1.7988611522218505e-05, "loss": 0.5401, "step": 6436 }, { "epoch": 0.2293318132426029, "grad_norm": 1.1815910507022065, "learning_rate": 1.798791735826044e-05, "loss": 0.5627, "step": 6437 }, { "epoch": 0.22936744036909704, "grad_norm": 1.6556678790565211, "learning_rate": 1.7987223087937672e-05, "loss": 0.6409, "step": 6438 }, { "epoch": 0.22940306749559114, "grad_norm": 1.7748449156883914, "learning_rate": 1.7986528711259442e-05, "loss": 0.6368, "step": 6439 }, { "epoch": 0.22943869462208524, "grad_norm": 1.1646128845482122, "learning_rate": 1.7985834228235002e-05, "loss": 0.4629, "step": 6440 }, { "epoch": 0.22947432174857937, "grad_norm": 1.8407218747965046, "learning_rate": 1.7985139638873603e-05, "loss": 0.6343, "step": 6441 }, { "epoch": 0.22950994887507348, "grad_norm": 1.719043470739594, "learning_rate": 1.798444494318448e-05, "loss": 0.5411, "step": 6442 }, { "epoch": 0.22954557600156758, "grad_norm": 1.5221045119989798, "learning_rate": 1.7983750141176898e-05, "loss": 0.5838, "step": 6443 }, { "epoch": 0.2295812031280617, "grad_norm": 1.9380234377613461, "learning_rate": 1.79830552328601e-05, "loss": 0.6218, "step": 6444 }, { "epoch": 0.22961683025455581, "grad_norm": 1.1727614687908403, "learning_rate": 1.798236021824334e-05, "loss": 0.5173, "step": 6445 }, { "epoch": 0.22965245738104992, "grad_norm": 1.6770363456915471, "learning_rate": 1.7981665097335875e-05, "loss": 0.5669, "step": 6446 }, { "epoch": 0.22968808450754405, "grad_norm": 1.5124771523385072, "learning_rate": 1.7980969870146963e-05, "loss": 0.6574, "step": 6447 }, { "epoch": 0.22972371163403815, "grad_norm": 1.1558727906325044, "learning_rate": 1.798027453668586e-05, "loss": 0.5268, "step": 6448 }, { "epoch": 0.22975933876053226, "grad_norm": 2.24273654158231, "learning_rate": 1.797957909696182e-05, "loss": 0.5514, "step": 6449 }, { "epoch": 0.2297949658870264, "grad_norm": 1.216377224397039, "learning_rate": 1.797888355098411e-05, "loss": 0.546, "step": 6450 }, { "epoch": 0.2298305930135205, "grad_norm": 1.0863259784564987, "learning_rate": 1.7978187898761982e-05, "loss": 0.5173, "step": 6451 }, { "epoch": 0.22986622014001462, "grad_norm": 1.7897124687839872, "learning_rate": 1.7977492140304713e-05, "loss": 0.6715, "step": 6452 }, { "epoch": 0.22990184726650872, "grad_norm": 1.0665411835963383, "learning_rate": 1.7976796275621556e-05, "loss": 0.5015, "step": 6453 }, { "epoch": 0.22993747439300283, "grad_norm": 1.6211984542930864, "learning_rate": 1.7976100304721786e-05, "loss": 0.6637, "step": 6454 }, { "epoch": 0.22997310151949696, "grad_norm": 1.2669018515282269, "learning_rate": 1.7975404227614658e-05, "loss": 0.4891, "step": 6455 }, { "epoch": 0.23000872864599106, "grad_norm": 1.142230720763729, "learning_rate": 1.7974708044309453e-05, "loss": 0.5009, "step": 6456 }, { "epoch": 0.23004435577248517, "grad_norm": 1.322546146776661, "learning_rate": 1.7974011754815435e-05, "loss": 0.5294, "step": 6457 }, { "epoch": 0.2300799828989793, "grad_norm": 1.7320898698912204, "learning_rate": 1.7973315359141875e-05, "loss": 0.5543, "step": 6458 }, { "epoch": 0.2301156100254734, "grad_norm": 1.55213514690216, "learning_rate": 1.7972618857298055e-05, "loss": 0.5504, "step": 6459 }, { "epoch": 0.2301512371519675, "grad_norm": 2.1620155698825565, "learning_rate": 1.7971922249293238e-05, "loss": 0.6461, "step": 6460 }, { "epoch": 0.23018686427846163, "grad_norm": 1.5044543783278599, "learning_rate": 1.7971225535136703e-05, "loss": 0.6356, "step": 6461 }, { "epoch": 0.23022249140495574, "grad_norm": 1.1171359510471375, "learning_rate": 1.7970528714837728e-05, "loss": 0.4624, "step": 6462 }, { "epoch": 0.23025811853144984, "grad_norm": 1.1225656733548026, "learning_rate": 1.7969831788405592e-05, "loss": 0.4702, "step": 6463 }, { "epoch": 0.23029374565794397, "grad_norm": 2.012703151251212, "learning_rate": 1.796913475584958e-05, "loss": 0.6813, "step": 6464 }, { "epoch": 0.23032937278443807, "grad_norm": 1.6693063252555163, "learning_rate": 1.7968437617178966e-05, "loss": 0.6047, "step": 6465 }, { "epoch": 0.23036499991093218, "grad_norm": 2.077980415983067, "learning_rate": 1.7967740372403036e-05, "loss": 0.6048, "step": 6466 }, { "epoch": 0.2304006270374263, "grad_norm": 2.24401860114394, "learning_rate": 1.7967043021531073e-05, "loss": 0.5884, "step": 6467 }, { "epoch": 0.2304362541639204, "grad_norm": 1.6260055151977093, "learning_rate": 1.796634556457236e-05, "loss": 0.6098, "step": 6468 }, { "epoch": 0.23047188129041452, "grad_norm": 1.2846603297812569, "learning_rate": 1.7965648001536193e-05, "loss": 0.5376, "step": 6469 }, { "epoch": 0.23050750841690865, "grad_norm": 1.2260393237715823, "learning_rate": 1.7964950332431854e-05, "loss": 0.5014, "step": 6470 }, { "epoch": 0.23054313554340275, "grad_norm": 1.363099372460621, "learning_rate": 1.7964252557268633e-05, "loss": 0.5727, "step": 6471 }, { "epoch": 0.23057876266989685, "grad_norm": 1.6100299278190457, "learning_rate": 1.7963554676055823e-05, "loss": 0.6266, "step": 6472 }, { "epoch": 0.23061438979639098, "grad_norm": 1.429152615777261, "learning_rate": 1.7962856688802717e-05, "loss": 0.5241, "step": 6473 }, { "epoch": 0.2306500169228851, "grad_norm": 1.2060001999689804, "learning_rate": 1.7962158595518607e-05, "loss": 0.5354, "step": 6474 }, { "epoch": 0.2306856440493792, "grad_norm": 1.3942499985265593, "learning_rate": 1.796146039621279e-05, "loss": 0.5075, "step": 6475 }, { "epoch": 0.23072127117587332, "grad_norm": 2.0598675143835523, "learning_rate": 1.7960762090894565e-05, "loss": 0.6388, "step": 6476 }, { "epoch": 0.23075689830236742, "grad_norm": 1.2763789281664049, "learning_rate": 1.796006367957323e-05, "loss": 0.5226, "step": 6477 }, { "epoch": 0.23079252542886153, "grad_norm": 1.1903370939552391, "learning_rate": 1.795936516225808e-05, "loss": 0.4708, "step": 6478 }, { "epoch": 0.23082815255535566, "grad_norm": 1.944446879431559, "learning_rate": 1.795866653895842e-05, "loss": 0.6185, "step": 6479 }, { "epoch": 0.23086377968184976, "grad_norm": 1.5809652104649667, "learning_rate": 1.795796780968355e-05, "loss": 0.5897, "step": 6480 }, { "epoch": 0.23089940680834387, "grad_norm": 2.6101896958801314, "learning_rate": 1.795726897444278e-05, "loss": 0.5902, "step": 6481 }, { "epoch": 0.230935033934838, "grad_norm": 1.1299596417041475, "learning_rate": 1.795657003324541e-05, "loss": 0.5167, "step": 6482 }, { "epoch": 0.2309706610613321, "grad_norm": 1.145949028928519, "learning_rate": 1.795587098610075e-05, "loss": 0.5133, "step": 6483 }, { "epoch": 0.2310062881878262, "grad_norm": 1.1275372806330417, "learning_rate": 1.7955171833018107e-05, "loss": 0.4633, "step": 6484 }, { "epoch": 0.23104191531432033, "grad_norm": 1.2175445723121785, "learning_rate": 1.795447257400679e-05, "loss": 0.5613, "step": 6485 }, { "epoch": 0.23107754244081444, "grad_norm": 1.1682870593763057, "learning_rate": 1.795377320907611e-05, "loss": 0.4711, "step": 6486 }, { "epoch": 0.23111316956730854, "grad_norm": 1.2087138728616356, "learning_rate": 1.7953073738235382e-05, "loss": 0.5238, "step": 6487 }, { "epoch": 0.23114879669380267, "grad_norm": 1.6041507052208417, "learning_rate": 1.795237416149392e-05, "loss": 0.6076, "step": 6488 }, { "epoch": 0.23118442382029677, "grad_norm": 1.876337861805797, "learning_rate": 1.7951674478861036e-05, "loss": 0.6328, "step": 6489 }, { "epoch": 0.23122005094679088, "grad_norm": 1.2310976989395084, "learning_rate": 1.795097469034605e-05, "loss": 0.5297, "step": 6490 }, { "epoch": 0.231255678073285, "grad_norm": 1.662956271614409, "learning_rate": 1.7950274795958272e-05, "loss": 0.6469, "step": 6491 }, { "epoch": 0.2312913051997791, "grad_norm": 1.5237473778138548, "learning_rate": 1.7949574795707034e-05, "loss": 0.5432, "step": 6492 }, { "epoch": 0.23132693232627322, "grad_norm": 1.6894863370802742, "learning_rate": 1.794887468960165e-05, "loss": 0.6189, "step": 6493 }, { "epoch": 0.23136255945276735, "grad_norm": 2.9606892961643374, "learning_rate": 1.7948174477651445e-05, "loss": 0.5481, "step": 6494 }, { "epoch": 0.23139818657926145, "grad_norm": 1.5153822256754712, "learning_rate": 1.794747415986574e-05, "loss": 0.7014, "step": 6495 }, { "epoch": 0.23143381370575555, "grad_norm": 1.4430791514180823, "learning_rate": 1.794677373625386e-05, "loss": 0.6611, "step": 6496 }, { "epoch": 0.23146944083224968, "grad_norm": 1.4148571055529018, "learning_rate": 1.7946073206825134e-05, "loss": 0.5587, "step": 6497 }, { "epoch": 0.2315050679587438, "grad_norm": 1.7542656658184157, "learning_rate": 1.794537257158889e-05, "loss": 0.4837, "step": 6498 }, { "epoch": 0.2315406950852379, "grad_norm": 1.220045445617498, "learning_rate": 1.7944671830554455e-05, "loss": 0.5282, "step": 6499 }, { "epoch": 0.23157632221173202, "grad_norm": 1.175693653334799, "learning_rate": 1.7943970983731164e-05, "loss": 0.4871, "step": 6500 }, { "epoch": 0.23161194933822613, "grad_norm": 1.3873611369828034, "learning_rate": 1.7943270031128345e-05, "loss": 0.5737, "step": 6501 }, { "epoch": 0.23164757646472023, "grad_norm": 1.755172935791471, "learning_rate": 1.7942568972755336e-05, "loss": 0.6203, "step": 6502 }, { "epoch": 0.23168320359121436, "grad_norm": 1.8314635407280364, "learning_rate": 1.7941867808621468e-05, "loss": 0.6762, "step": 6503 }, { "epoch": 0.23171883071770846, "grad_norm": 1.1327416297434665, "learning_rate": 1.7941166538736082e-05, "loss": 0.5332, "step": 6504 }, { "epoch": 0.23175445784420257, "grad_norm": 1.70536469030984, "learning_rate": 1.794046516310851e-05, "loss": 0.5856, "step": 6505 }, { "epoch": 0.2317900849706967, "grad_norm": 1.2087554453063583, "learning_rate": 1.7939763681748094e-05, "loss": 0.5213, "step": 6506 }, { "epoch": 0.2318257120971908, "grad_norm": 1.0775228305308227, "learning_rate": 1.7939062094664174e-05, "loss": 0.5029, "step": 6507 }, { "epoch": 0.2318613392236849, "grad_norm": 1.2730735605963022, "learning_rate": 1.7938360401866096e-05, "loss": 0.4901, "step": 6508 }, { "epoch": 0.23189696635017903, "grad_norm": 1.1562633689241788, "learning_rate": 1.7937658603363196e-05, "loss": 0.5423, "step": 6509 }, { "epoch": 0.23193259347667314, "grad_norm": 1.388928681507573, "learning_rate": 1.7936956699164828e-05, "loss": 0.5407, "step": 6510 }, { "epoch": 0.23196822060316724, "grad_norm": 1.2731605303349836, "learning_rate": 1.7936254689280336e-05, "loss": 0.5097, "step": 6511 }, { "epoch": 0.23200384772966137, "grad_norm": 1.146082655213633, "learning_rate": 1.793555257371906e-05, "loss": 0.5078, "step": 6512 }, { "epoch": 0.23203947485615548, "grad_norm": 1.8791418669819184, "learning_rate": 1.793485035249036e-05, "loss": 0.4984, "step": 6513 }, { "epoch": 0.23207510198264958, "grad_norm": 1.5439553376935125, "learning_rate": 1.7934148025603576e-05, "loss": 0.5576, "step": 6514 }, { "epoch": 0.2321107291091437, "grad_norm": 1.559472020359965, "learning_rate": 1.793344559306807e-05, "loss": 0.5307, "step": 6515 }, { "epoch": 0.2321463562356378, "grad_norm": 1.7696299993164213, "learning_rate": 1.793274305489319e-05, "loss": 0.598, "step": 6516 }, { "epoch": 0.23218198336213192, "grad_norm": 1.2970475371780916, "learning_rate": 1.793204041108829e-05, "loss": 0.4872, "step": 6517 }, { "epoch": 0.23221761048862605, "grad_norm": 1.1691654907578148, "learning_rate": 1.7931337661662728e-05, "loss": 0.471, "step": 6518 }, { "epoch": 0.23225323761512015, "grad_norm": 1.264065510273101, "learning_rate": 1.793063480662586e-05, "loss": 0.4977, "step": 6519 }, { "epoch": 0.23228886474161425, "grad_norm": 1.929257831516349, "learning_rate": 1.792993184598705e-05, "loss": 0.6327, "step": 6520 }, { "epoch": 0.23232449186810838, "grad_norm": 1.8898594849695374, "learning_rate": 1.7929228779755652e-05, "loss": 0.6035, "step": 6521 }, { "epoch": 0.2323601189946025, "grad_norm": 1.5957631569676347, "learning_rate": 1.7928525607941033e-05, "loss": 0.5985, "step": 6522 }, { "epoch": 0.2323957461210966, "grad_norm": 1.151181561477765, "learning_rate": 1.7927822330552553e-05, "loss": 0.4645, "step": 6523 }, { "epoch": 0.23243137324759072, "grad_norm": 1.2400223011741738, "learning_rate": 1.7927118947599576e-05, "loss": 0.4945, "step": 6524 }, { "epoch": 0.23246700037408483, "grad_norm": 2.538460149909024, "learning_rate": 1.792641545909147e-05, "loss": 0.6738, "step": 6525 }, { "epoch": 0.23250262750057893, "grad_norm": 1.2526350290539827, "learning_rate": 1.7925711865037602e-05, "loss": 0.4739, "step": 6526 }, { "epoch": 0.23253825462707306, "grad_norm": 1.4532224598275996, "learning_rate": 1.792500816544734e-05, "loss": 0.7088, "step": 6527 }, { "epoch": 0.23257388175356716, "grad_norm": 1.6550924701070886, "learning_rate": 1.7924304360330057e-05, "loss": 0.6471, "step": 6528 }, { "epoch": 0.23260950888006127, "grad_norm": 1.9666505582209985, "learning_rate": 1.7923600449695124e-05, "loss": 0.7241, "step": 6529 }, { "epoch": 0.2326451360065554, "grad_norm": 1.2211581943440588, "learning_rate": 1.792289643355191e-05, "loss": 0.5137, "step": 6530 }, { "epoch": 0.2326807631330495, "grad_norm": 1.5599877495101164, "learning_rate": 1.792219231190979e-05, "loss": 0.5964, "step": 6531 }, { "epoch": 0.2327163902595436, "grad_norm": 1.2532188419316075, "learning_rate": 1.7921488084778147e-05, "loss": 0.5291, "step": 6532 }, { "epoch": 0.23275201738603774, "grad_norm": 1.199516978184538, "learning_rate": 1.7920783752166348e-05, "loss": 0.5016, "step": 6533 }, { "epoch": 0.23278764451253184, "grad_norm": 1.5089170057011445, "learning_rate": 1.792007931408378e-05, "loss": 0.5367, "step": 6534 }, { "epoch": 0.23282327163902594, "grad_norm": 1.2753110436974309, "learning_rate": 1.791937477053982e-05, "loss": 0.5383, "step": 6535 }, { "epoch": 0.23285889876552007, "grad_norm": 1.78812290176683, "learning_rate": 1.7918670121543854e-05, "loss": 0.6505, "step": 6536 }, { "epoch": 0.23289452589201418, "grad_norm": 1.188424646541879, "learning_rate": 1.7917965367105253e-05, "loss": 0.5328, "step": 6537 }, { "epoch": 0.23293015301850828, "grad_norm": 1.4427404497625174, "learning_rate": 1.7917260507233415e-05, "loss": 0.5248, "step": 6538 }, { "epoch": 0.2329657801450024, "grad_norm": 1.239581657622928, "learning_rate": 1.7916555541937715e-05, "loss": 0.4945, "step": 6539 }, { "epoch": 0.2330014072714965, "grad_norm": 1.624884063797184, "learning_rate": 1.7915850471227548e-05, "loss": 0.5998, "step": 6540 }, { "epoch": 0.23303703439799062, "grad_norm": 1.2822798939996856, "learning_rate": 1.7915145295112297e-05, "loss": 0.4911, "step": 6541 }, { "epoch": 0.23307266152448475, "grad_norm": 1.9134157085054542, "learning_rate": 1.7914440013601355e-05, "loss": 0.5909, "step": 6542 }, { "epoch": 0.23310828865097885, "grad_norm": 1.2262811489914751, "learning_rate": 1.791373462670411e-05, "loss": 0.5764, "step": 6543 }, { "epoch": 0.23314391577747295, "grad_norm": 1.6075781594252594, "learning_rate": 1.791302913442996e-05, "loss": 0.6311, "step": 6544 }, { "epoch": 0.23317954290396709, "grad_norm": 1.3995481027508425, "learning_rate": 1.7912323536788297e-05, "loss": 0.6166, "step": 6545 }, { "epoch": 0.2332151700304612, "grad_norm": 1.5373944740002021, "learning_rate": 1.7911617833788513e-05, "loss": 0.619, "step": 6546 }, { "epoch": 0.23325079715695532, "grad_norm": 1.3745794808712193, "learning_rate": 1.791091202544001e-05, "loss": 0.5155, "step": 6547 }, { "epoch": 0.23328642428344942, "grad_norm": 2.4885013902192563, "learning_rate": 1.791020611175218e-05, "loss": 0.5937, "step": 6548 }, { "epoch": 0.23332205140994353, "grad_norm": 1.3625464972014216, "learning_rate": 1.790950009273443e-05, "loss": 0.52, "step": 6549 }, { "epoch": 0.23335767853643766, "grad_norm": 2.310132158294611, "learning_rate": 1.7908793968396158e-05, "loss": 0.553, "step": 6550 }, { "epoch": 0.23339330566293176, "grad_norm": 1.50484658999646, "learning_rate": 1.7908087738746764e-05, "loss": 0.5401, "step": 6551 }, { "epoch": 0.23342893278942586, "grad_norm": 1.5083185114139501, "learning_rate": 1.7907381403795655e-05, "loss": 0.5149, "step": 6552 }, { "epoch": 0.23346455991592, "grad_norm": 2.3238843425585016, "learning_rate": 1.7906674963552237e-05, "loss": 0.4787, "step": 6553 }, { "epoch": 0.2335001870424141, "grad_norm": 2.642301445506248, "learning_rate": 1.7905968418025915e-05, "loss": 0.6157, "step": 6554 }, { "epoch": 0.2335358141689082, "grad_norm": 1.9950027939846713, "learning_rate": 1.79052617672261e-05, "loss": 0.5666, "step": 6555 }, { "epoch": 0.23357144129540233, "grad_norm": 4.101420382007974, "learning_rate": 1.7904555011162195e-05, "loss": 0.6427, "step": 6556 }, { "epoch": 0.23360706842189644, "grad_norm": 1.3431113837080542, "learning_rate": 1.7903848149843616e-05, "loss": 0.5756, "step": 6557 }, { "epoch": 0.23364269554839054, "grad_norm": 2.429116959297666, "learning_rate": 1.7903141183279776e-05, "loss": 0.585, "step": 6558 }, { "epoch": 0.23367832267488467, "grad_norm": 1.5851984648568134, "learning_rate": 1.7902434111480086e-05, "loss": 0.5893, "step": 6559 }, { "epoch": 0.23371394980137877, "grad_norm": 2.187819695121212, "learning_rate": 1.7901726934453964e-05, "loss": 0.6498, "step": 6560 }, { "epoch": 0.23374957692787288, "grad_norm": 1.6028422914902116, "learning_rate": 1.790101965221082e-05, "loss": 0.5461, "step": 6561 }, { "epoch": 0.233785204054367, "grad_norm": 2.4981855057462217, "learning_rate": 1.790031226476008e-05, "loss": 0.5975, "step": 6562 }, { "epoch": 0.2338208311808611, "grad_norm": 1.6719378477514042, "learning_rate": 1.7899604772111163e-05, "loss": 0.7109, "step": 6563 }, { "epoch": 0.23385645830735521, "grad_norm": 1.3579141416955443, "learning_rate": 1.7898897174273486e-05, "loss": 0.6037, "step": 6564 }, { "epoch": 0.23389208543384934, "grad_norm": 1.5529824816393878, "learning_rate": 1.7898189471256468e-05, "loss": 0.5623, "step": 6565 }, { "epoch": 0.23392771256034345, "grad_norm": 1.2260109489226485, "learning_rate": 1.789748166306954e-05, "loss": 0.4993, "step": 6566 }, { "epoch": 0.23396333968683755, "grad_norm": 1.4470044209027275, "learning_rate": 1.7896773749722123e-05, "loss": 0.5179, "step": 6567 }, { "epoch": 0.23399896681333168, "grad_norm": 1.900983529186379, "learning_rate": 1.7896065731223645e-05, "loss": 0.6266, "step": 6568 }, { "epoch": 0.23403459393982579, "grad_norm": 1.687960526527902, "learning_rate": 1.7895357607583534e-05, "loss": 0.5557, "step": 6569 }, { "epoch": 0.2340702210663199, "grad_norm": 1.4340398937480872, "learning_rate": 1.7894649378811214e-05, "loss": 0.51, "step": 6570 }, { "epoch": 0.23410584819281402, "grad_norm": 1.1807706412452088, "learning_rate": 1.789394104491612e-05, "loss": 0.4548, "step": 6571 }, { "epoch": 0.23414147531930812, "grad_norm": 1.5766350315942281, "learning_rate": 1.7893232605907686e-05, "loss": 0.548, "step": 6572 }, { "epoch": 0.23417710244580223, "grad_norm": 1.8663702549630534, "learning_rate": 1.789252406179534e-05, "loss": 0.6182, "step": 6573 }, { "epoch": 0.23421272957229636, "grad_norm": 1.2143102369800494, "learning_rate": 1.7891815412588524e-05, "loss": 0.4772, "step": 6574 }, { "epoch": 0.23424835669879046, "grad_norm": 1.257881249976449, "learning_rate": 1.7891106658296665e-05, "loss": 0.519, "step": 6575 }, { "epoch": 0.23428398382528456, "grad_norm": 1.1581953032595556, "learning_rate": 1.7890397798929207e-05, "loss": 0.4576, "step": 6576 }, { "epoch": 0.2343196109517787, "grad_norm": 1.5126728620749386, "learning_rate": 1.7889688834495585e-05, "loss": 0.6209, "step": 6577 }, { "epoch": 0.2343552380782728, "grad_norm": 1.2523989492125678, "learning_rate": 1.7888979765005247e-05, "loss": 0.5398, "step": 6578 }, { "epoch": 0.2343908652047669, "grad_norm": 1.2310902156799282, "learning_rate": 1.7888270590467627e-05, "loss": 0.5185, "step": 6579 }, { "epoch": 0.23442649233126103, "grad_norm": 1.3012077443220318, "learning_rate": 1.788756131089217e-05, "loss": 0.4993, "step": 6580 }, { "epoch": 0.23446211945775514, "grad_norm": 1.187145171689945, "learning_rate": 1.7886851926288318e-05, "loss": 0.4777, "step": 6581 }, { "epoch": 0.23449774658424924, "grad_norm": 1.6252737824469303, "learning_rate": 1.7886142436665525e-05, "loss": 0.5689, "step": 6582 }, { "epoch": 0.23453337371074337, "grad_norm": 1.235931057736457, "learning_rate": 1.788543284203323e-05, "loss": 0.5287, "step": 6583 }, { "epoch": 0.23456900083723747, "grad_norm": 1.5374988197576962, "learning_rate": 1.7884723142400887e-05, "loss": 0.5315, "step": 6584 }, { "epoch": 0.23460462796373158, "grad_norm": 1.2597814327571353, "learning_rate": 1.7884013337777944e-05, "loss": 0.5379, "step": 6585 }, { "epoch": 0.2346402550902257, "grad_norm": 1.1687794221465417, "learning_rate": 1.7883303428173854e-05, "loss": 0.4522, "step": 6586 }, { "epoch": 0.2346758822167198, "grad_norm": 1.171596230023765, "learning_rate": 1.7882593413598065e-05, "loss": 0.4653, "step": 6587 }, { "epoch": 0.23471150934321391, "grad_norm": 1.197272375669796, "learning_rate": 1.788188329406004e-05, "loss": 0.5132, "step": 6588 }, { "epoch": 0.23474713646970805, "grad_norm": 1.959991504199356, "learning_rate": 1.7881173069569225e-05, "loss": 0.6751, "step": 6589 }, { "epoch": 0.23478276359620215, "grad_norm": 1.598828699011148, "learning_rate": 1.7880462740135084e-05, "loss": 0.5596, "step": 6590 }, { "epoch": 0.23481839072269625, "grad_norm": 1.2356371259624945, "learning_rate": 1.7879752305767073e-05, "loss": 0.5037, "step": 6591 }, { "epoch": 0.23485401784919038, "grad_norm": 1.4573560128735412, "learning_rate": 1.7879041766474654e-05, "loss": 0.5693, "step": 6592 }, { "epoch": 0.2348896449756845, "grad_norm": 1.2120455225968019, "learning_rate": 1.7878331122267284e-05, "loss": 0.4696, "step": 6593 }, { "epoch": 0.2349252721021786, "grad_norm": 2.1265189546121572, "learning_rate": 1.7877620373154433e-05, "loss": 0.5921, "step": 6594 }, { "epoch": 0.23496089922867272, "grad_norm": 1.1747116384564262, "learning_rate": 1.7876909519145558e-05, "loss": 0.4902, "step": 6595 }, { "epoch": 0.23499652635516682, "grad_norm": 1.7454304607687252, "learning_rate": 1.7876198560250127e-05, "loss": 0.5068, "step": 6596 }, { "epoch": 0.23503215348166093, "grad_norm": 1.730873420105077, "learning_rate": 1.7875487496477604e-05, "loss": 0.5896, "step": 6597 }, { "epoch": 0.23506778060815506, "grad_norm": 1.0423840888940505, "learning_rate": 1.7874776327837463e-05, "loss": 0.4368, "step": 6598 }, { "epoch": 0.23510340773464916, "grad_norm": 1.9436553160735275, "learning_rate": 1.7874065054339173e-05, "loss": 0.603, "step": 6599 }, { "epoch": 0.23513903486114326, "grad_norm": 1.397945262673702, "learning_rate": 1.7873353675992203e-05, "loss": 0.5724, "step": 6600 }, { "epoch": 0.2351746619876374, "grad_norm": 1.688689050448768, "learning_rate": 1.787264219280602e-05, "loss": 0.5893, "step": 6601 }, { "epoch": 0.2352102891141315, "grad_norm": 1.219251663872862, "learning_rate": 1.787193060479011e-05, "loss": 0.5459, "step": 6602 }, { "epoch": 0.2352459162406256, "grad_norm": 1.8661948674694353, "learning_rate": 1.7871218911953942e-05, "loss": 0.5714, "step": 6603 }, { "epoch": 0.23528154336711973, "grad_norm": 1.7059499885613374, "learning_rate": 1.7870507114306988e-05, "loss": 0.5599, "step": 6604 }, { "epoch": 0.23531717049361384, "grad_norm": 1.460108184914002, "learning_rate": 1.7869795211858735e-05, "loss": 0.5285, "step": 6605 }, { "epoch": 0.23535279762010794, "grad_norm": 1.1358985415173153, "learning_rate": 1.7869083204618655e-05, "loss": 0.514, "step": 6606 }, { "epoch": 0.23538842474660207, "grad_norm": 1.4068243343295224, "learning_rate": 1.7868371092596233e-05, "loss": 0.6173, "step": 6607 }, { "epoch": 0.23542405187309617, "grad_norm": 1.3369495800226445, "learning_rate": 1.786765887580095e-05, "loss": 0.5268, "step": 6608 }, { "epoch": 0.23545967899959028, "grad_norm": 1.3076202634728713, "learning_rate": 1.786694655424229e-05, "loss": 0.4965, "step": 6609 }, { "epoch": 0.2354953061260844, "grad_norm": 1.597717200139335, "learning_rate": 1.786623412792974e-05, "loss": 0.6373, "step": 6610 }, { "epoch": 0.2355309332525785, "grad_norm": 1.8421044019941495, "learning_rate": 1.7865521596872782e-05, "loss": 0.5608, "step": 6611 }, { "epoch": 0.23556656037907261, "grad_norm": 1.2256011586768825, "learning_rate": 1.7864808961080906e-05, "loss": 0.5371, "step": 6612 }, { "epoch": 0.23560218750556675, "grad_norm": 1.8732057305639067, "learning_rate": 1.78640962205636e-05, "loss": 0.5671, "step": 6613 }, { "epoch": 0.23563781463206085, "grad_norm": 1.2627277061045306, "learning_rate": 1.786338337533036e-05, "loss": 0.4951, "step": 6614 }, { "epoch": 0.23567344175855495, "grad_norm": 1.6230703841490792, "learning_rate": 1.7862670425390672e-05, "loss": 0.6052, "step": 6615 }, { "epoch": 0.23570906888504908, "grad_norm": 1.2793292667264484, "learning_rate": 1.7861957370754034e-05, "loss": 0.5503, "step": 6616 }, { "epoch": 0.2357446960115432, "grad_norm": 1.2010456393911018, "learning_rate": 1.7861244211429935e-05, "loss": 0.5124, "step": 6617 }, { "epoch": 0.2357803231380373, "grad_norm": 1.7448989254709544, "learning_rate": 1.7860530947427878e-05, "loss": 0.6737, "step": 6618 }, { "epoch": 0.23581595026453142, "grad_norm": 1.5589704829103481, "learning_rate": 1.7859817578757353e-05, "loss": 0.657, "step": 6619 }, { "epoch": 0.23585157739102552, "grad_norm": 1.916839230315841, "learning_rate": 1.7859104105427867e-05, "loss": 0.6655, "step": 6620 }, { "epoch": 0.23588720451751963, "grad_norm": 2.0484188724709997, "learning_rate": 1.7858390527448915e-05, "loss": 0.6356, "step": 6621 }, { "epoch": 0.23592283164401376, "grad_norm": 1.563142142959406, "learning_rate": 1.7857676844830002e-05, "loss": 0.6163, "step": 6622 }, { "epoch": 0.23595845877050786, "grad_norm": 3.035035697604134, "learning_rate": 1.7856963057580626e-05, "loss": 0.5376, "step": 6623 }, { "epoch": 0.23599408589700197, "grad_norm": 3.261346867554937, "learning_rate": 1.7856249165710298e-05, "loss": 0.4699, "step": 6624 }, { "epoch": 0.2360297130234961, "grad_norm": 2.1892218094016274, "learning_rate": 1.785553516922852e-05, "loss": 0.6356, "step": 6625 }, { "epoch": 0.2360653401499902, "grad_norm": 1.693898994320547, "learning_rate": 1.78548210681448e-05, "loss": 0.616, "step": 6626 }, { "epoch": 0.2361009672764843, "grad_norm": 2.0262153376575394, "learning_rate": 1.785410686246865e-05, "loss": 0.6246, "step": 6627 }, { "epoch": 0.23613659440297843, "grad_norm": 1.1614248381912093, "learning_rate": 1.785339255220958e-05, "loss": 0.4671, "step": 6628 }, { "epoch": 0.23617222152947254, "grad_norm": 1.4553675822701269, "learning_rate": 1.785267813737709e-05, "loss": 0.5143, "step": 6629 }, { "epoch": 0.23620784865596664, "grad_norm": 1.4837516774825412, "learning_rate": 1.7851963617980707e-05, "loss": 0.6276, "step": 6630 }, { "epoch": 0.23624347578246077, "grad_norm": 1.6013100990283586, "learning_rate": 1.7851248994029943e-05, "loss": 0.6023, "step": 6631 }, { "epoch": 0.23627910290895487, "grad_norm": 1.7266913819437613, "learning_rate": 1.785053426553431e-05, "loss": 0.6266, "step": 6632 }, { "epoch": 0.23631473003544898, "grad_norm": 1.4770152737582556, "learning_rate": 1.7849819432503322e-05, "loss": 0.5926, "step": 6633 }, { "epoch": 0.2363503571619431, "grad_norm": 1.248085422854884, "learning_rate": 1.7849104494946504e-05, "loss": 0.4955, "step": 6634 }, { "epoch": 0.2363859842884372, "grad_norm": 1.8396468437719211, "learning_rate": 1.7848389452873372e-05, "loss": 0.655, "step": 6635 }, { "epoch": 0.23642161141493132, "grad_norm": 1.2609640679458292, "learning_rate": 1.7847674306293454e-05, "loss": 0.4983, "step": 6636 }, { "epoch": 0.23645723854142545, "grad_norm": 1.1917565735515032, "learning_rate": 1.784695905521626e-05, "loss": 0.5153, "step": 6637 }, { "epoch": 0.23649286566791955, "grad_norm": 1.8005987909972443, "learning_rate": 1.7846243699651325e-05, "loss": 0.5958, "step": 6638 }, { "epoch": 0.23652849279441365, "grad_norm": 1.3009664332343145, "learning_rate": 1.7845528239608172e-05, "loss": 0.5412, "step": 6639 }, { "epoch": 0.23656411992090778, "grad_norm": 1.5120090206540868, "learning_rate": 1.7844812675096323e-05, "loss": 0.6002, "step": 6640 }, { "epoch": 0.2365997470474019, "grad_norm": 1.4855394185128423, "learning_rate": 1.7844097006125315e-05, "loss": 0.6187, "step": 6641 }, { "epoch": 0.23663537417389602, "grad_norm": 1.4008869004975175, "learning_rate": 1.7843381232704668e-05, "loss": 0.5303, "step": 6642 }, { "epoch": 0.23667100130039012, "grad_norm": 1.54049333224501, "learning_rate": 1.7842665354843923e-05, "loss": 0.6173, "step": 6643 }, { "epoch": 0.23670662842688422, "grad_norm": 1.6317527153103657, "learning_rate": 1.7841949372552602e-05, "loss": 0.5506, "step": 6644 }, { "epoch": 0.23674225555337836, "grad_norm": 1.3823216768788749, "learning_rate": 1.7841233285840248e-05, "loss": 0.5476, "step": 6645 }, { "epoch": 0.23677788267987246, "grad_norm": 1.6661950635297418, "learning_rate": 1.7840517094716386e-05, "loss": 0.611, "step": 6646 }, { "epoch": 0.23681350980636656, "grad_norm": 1.2795987576263252, "learning_rate": 1.7839800799190566e-05, "loss": 0.5071, "step": 6647 }, { "epoch": 0.2368491369328607, "grad_norm": 1.0432214830580262, "learning_rate": 1.7839084399272317e-05, "loss": 0.4697, "step": 6648 }, { "epoch": 0.2368847640593548, "grad_norm": 1.7276512950815397, "learning_rate": 1.7838367894971175e-05, "loss": 0.569, "step": 6649 }, { "epoch": 0.2369203911858489, "grad_norm": 2.10248660958429, "learning_rate": 1.7837651286296692e-05, "loss": 0.6607, "step": 6650 }, { "epoch": 0.23695601831234303, "grad_norm": 1.306858862420148, "learning_rate": 1.78369345732584e-05, "loss": 0.5161, "step": 6651 }, { "epoch": 0.23699164543883713, "grad_norm": 1.6034171395154673, "learning_rate": 1.7836217755865846e-05, "loss": 0.6081, "step": 6652 }, { "epoch": 0.23702727256533124, "grad_norm": 1.2558806192348302, "learning_rate": 1.783550083412858e-05, "loss": 0.4785, "step": 6653 }, { "epoch": 0.23706289969182537, "grad_norm": 1.650237907821163, "learning_rate": 1.783478380805614e-05, "loss": 0.5742, "step": 6654 }, { "epoch": 0.23709852681831947, "grad_norm": 1.298459718437603, "learning_rate": 1.7834066677658077e-05, "loss": 0.5092, "step": 6655 }, { "epoch": 0.23713415394481357, "grad_norm": 1.2993568254865613, "learning_rate": 1.783334944294394e-05, "loss": 0.5246, "step": 6656 }, { "epoch": 0.2371697810713077, "grad_norm": 1.1386457965767416, "learning_rate": 1.7832632103923286e-05, "loss": 0.4949, "step": 6657 }, { "epoch": 0.2372054081978018, "grad_norm": 1.2623967032069492, "learning_rate": 1.7831914660605654e-05, "loss": 0.4944, "step": 6658 }, { "epoch": 0.2372410353242959, "grad_norm": 1.419832344592709, "learning_rate": 1.7831197113000607e-05, "loss": 0.652, "step": 6659 }, { "epoch": 0.23727666245079004, "grad_norm": 2.3261683010679635, "learning_rate": 1.7830479461117697e-05, "loss": 0.5607, "step": 6660 }, { "epoch": 0.23731228957728415, "grad_norm": 2.2527283952488446, "learning_rate": 1.7829761704966477e-05, "loss": 0.5894, "step": 6661 }, { "epoch": 0.23734791670377825, "grad_norm": 1.336958778571968, "learning_rate": 1.782904384455651e-05, "loss": 0.5361, "step": 6662 }, { "epoch": 0.23738354383027238, "grad_norm": 2.067506828490501, "learning_rate": 1.7828325879897353e-05, "loss": 0.7032, "step": 6663 }, { "epoch": 0.23741917095676648, "grad_norm": 1.36527022040088, "learning_rate": 1.782760781099856e-05, "loss": 0.5022, "step": 6664 }, { "epoch": 0.2374547980832606, "grad_norm": 1.5763353356489618, "learning_rate": 1.7826889637869705e-05, "loss": 0.5841, "step": 6665 }, { "epoch": 0.23749042520975472, "grad_norm": 1.874301285055045, "learning_rate": 1.782617136052034e-05, "loss": 0.6033, "step": 6666 }, { "epoch": 0.23752605233624882, "grad_norm": 2.0693956682000296, "learning_rate": 1.7825452978960032e-05, "loss": 0.5453, "step": 6667 }, { "epoch": 0.23756167946274293, "grad_norm": 2.0056754015210583, "learning_rate": 1.782473449319835e-05, "loss": 0.6039, "step": 6668 }, { "epoch": 0.23759730658923706, "grad_norm": 1.9737180007869974, "learning_rate": 1.7824015903244857e-05, "loss": 0.6955, "step": 6669 }, { "epoch": 0.23763293371573116, "grad_norm": 1.2908463874825495, "learning_rate": 1.782329720910913e-05, "loss": 0.5193, "step": 6670 }, { "epoch": 0.23766856084222526, "grad_norm": 1.6031633881266087, "learning_rate": 1.7822578410800724e-05, "loss": 0.5999, "step": 6671 }, { "epoch": 0.2377041879687194, "grad_norm": 2.6009465230923676, "learning_rate": 1.7821859508329223e-05, "loss": 0.693, "step": 6672 }, { "epoch": 0.2377398150952135, "grad_norm": 1.2517834995605466, "learning_rate": 1.7821140501704195e-05, "loss": 0.5133, "step": 6673 }, { "epoch": 0.2377754422217076, "grad_norm": 1.3219939940607233, "learning_rate": 1.7820421390935213e-05, "loss": 0.5596, "step": 6674 }, { "epoch": 0.23781106934820173, "grad_norm": 2.1602309725026907, "learning_rate": 1.7819702176031858e-05, "loss": 0.6125, "step": 6675 }, { "epoch": 0.23784669647469583, "grad_norm": 1.174854726655521, "learning_rate": 1.7818982857003698e-05, "loss": 0.5406, "step": 6676 }, { "epoch": 0.23788232360118994, "grad_norm": 1.3262754574936333, "learning_rate": 1.781826343386032e-05, "loss": 0.5094, "step": 6677 }, { "epoch": 0.23791795072768407, "grad_norm": 1.2515909831032097, "learning_rate": 1.7817543906611298e-05, "loss": 0.501, "step": 6678 }, { "epoch": 0.23795357785417817, "grad_norm": 1.88081200764147, "learning_rate": 1.7816824275266213e-05, "loss": 0.6205, "step": 6679 }, { "epoch": 0.23798920498067228, "grad_norm": 1.153438989262476, "learning_rate": 1.7816104539834654e-05, "loss": 0.48, "step": 6680 }, { "epoch": 0.2380248321071664, "grad_norm": 1.0900611761497982, "learning_rate": 1.7815384700326195e-05, "loss": 0.489, "step": 6681 }, { "epoch": 0.2380604592336605, "grad_norm": 1.6936657415546474, "learning_rate": 1.781466475675043e-05, "loss": 0.6244, "step": 6682 }, { "epoch": 0.2380960863601546, "grad_norm": 1.5153025305719874, "learning_rate": 1.7813944709116938e-05, "loss": 0.5469, "step": 6683 }, { "epoch": 0.23813171348664874, "grad_norm": 2.025642391260124, "learning_rate": 1.7813224557435313e-05, "loss": 0.5839, "step": 6684 }, { "epoch": 0.23816734061314285, "grad_norm": 1.4394545159139822, "learning_rate": 1.7812504301715142e-05, "loss": 0.535, "step": 6685 }, { "epoch": 0.23820296773963695, "grad_norm": 1.1079941413516694, "learning_rate": 1.7811783941966013e-05, "loss": 0.5107, "step": 6686 }, { "epoch": 0.23823859486613108, "grad_norm": 2.031687777158152, "learning_rate": 1.781106347819752e-05, "loss": 0.6111, "step": 6687 }, { "epoch": 0.23827422199262518, "grad_norm": 1.8240506533864482, "learning_rate": 1.781034291041926e-05, "loss": 0.6927, "step": 6688 }, { "epoch": 0.2383098491191193, "grad_norm": 1.8998293348117317, "learning_rate": 1.7809622238640826e-05, "loss": 0.6425, "step": 6689 }, { "epoch": 0.23834547624561342, "grad_norm": 1.354728384287957, "learning_rate": 1.7808901462871814e-05, "loss": 0.5683, "step": 6690 }, { "epoch": 0.23838110337210752, "grad_norm": 1.5456995886861666, "learning_rate": 1.780818058312182e-05, "loss": 0.6508, "step": 6691 }, { "epoch": 0.23841673049860163, "grad_norm": 1.276055769630043, "learning_rate": 1.7807459599400442e-05, "loss": 0.5464, "step": 6692 }, { "epoch": 0.23845235762509576, "grad_norm": 1.3205265489596036, "learning_rate": 1.780673851171728e-05, "loss": 0.5476, "step": 6693 }, { "epoch": 0.23848798475158986, "grad_norm": 1.2212565519806202, "learning_rate": 1.780601732008194e-05, "loss": 0.49, "step": 6694 }, { "epoch": 0.23852361187808396, "grad_norm": 1.7868013417722606, "learning_rate": 1.7805296024504026e-05, "loss": 0.621, "step": 6695 }, { "epoch": 0.2385592390045781, "grad_norm": 1.3483742581254277, "learning_rate": 1.780457462499314e-05, "loss": 0.5017, "step": 6696 }, { "epoch": 0.2385948661310722, "grad_norm": 1.3750497160156867, "learning_rate": 1.7803853121558886e-05, "loss": 0.5436, "step": 6697 }, { "epoch": 0.2386304932575663, "grad_norm": 2.338105334499511, "learning_rate": 1.7803131514210874e-05, "loss": 0.5342, "step": 6698 }, { "epoch": 0.23866612038406043, "grad_norm": 1.6093409400465484, "learning_rate": 1.780240980295871e-05, "loss": 0.5234, "step": 6699 }, { "epoch": 0.23870174751055453, "grad_norm": 1.2341181320250636, "learning_rate": 1.780168798781201e-05, "loss": 0.4906, "step": 6700 }, { "epoch": 0.23873737463704864, "grad_norm": 1.5490363049871532, "learning_rate": 1.7800966068780378e-05, "loss": 0.5866, "step": 6701 }, { "epoch": 0.23877300176354277, "grad_norm": 1.2816522700351556, "learning_rate": 1.780024404587343e-05, "loss": 0.4731, "step": 6702 }, { "epoch": 0.23880862889003687, "grad_norm": 1.2761004646986753, "learning_rate": 1.779952191910078e-05, "loss": 0.5361, "step": 6703 }, { "epoch": 0.23884425601653098, "grad_norm": 1.3350989539832065, "learning_rate": 1.7798799688472046e-05, "loss": 0.5194, "step": 6704 }, { "epoch": 0.2388798831430251, "grad_norm": 1.3569169453301666, "learning_rate": 1.7798077353996843e-05, "loss": 0.5501, "step": 6705 }, { "epoch": 0.2389155102695192, "grad_norm": 1.13408912664661, "learning_rate": 1.779735491568479e-05, "loss": 0.5173, "step": 6706 }, { "epoch": 0.2389511373960133, "grad_norm": 1.5365962273010507, "learning_rate": 1.7796632373545507e-05, "loss": 0.6155, "step": 6707 }, { "epoch": 0.23898676452250744, "grad_norm": 2.159631242012366, "learning_rate": 1.7795909727588608e-05, "loss": 0.6625, "step": 6708 }, { "epoch": 0.23902239164900155, "grad_norm": 1.170477827729128, "learning_rate": 1.779518697782373e-05, "loss": 0.4964, "step": 6709 }, { "epoch": 0.23905801877549565, "grad_norm": 1.450931485823574, "learning_rate": 1.7794464124260484e-05, "loss": 0.4584, "step": 6710 }, { "epoch": 0.23909364590198978, "grad_norm": 1.4008486069247943, "learning_rate": 1.7793741166908504e-05, "loss": 0.5133, "step": 6711 }, { "epoch": 0.23912927302848389, "grad_norm": 1.4827059368502165, "learning_rate": 1.779301810577741e-05, "loss": 0.6695, "step": 6712 }, { "epoch": 0.239164900154978, "grad_norm": 1.349742073576921, "learning_rate": 1.7792294940876835e-05, "loss": 0.5377, "step": 6713 }, { "epoch": 0.23920052728147212, "grad_norm": 1.6016096359774545, "learning_rate": 1.7791571672216404e-05, "loss": 0.5734, "step": 6714 }, { "epoch": 0.23923615440796622, "grad_norm": 1.240916571940914, "learning_rate": 1.779084829980575e-05, "loss": 0.46, "step": 6715 }, { "epoch": 0.23927178153446033, "grad_norm": 2.555329334155622, "learning_rate": 1.7790124823654506e-05, "loss": 0.5473, "step": 6716 }, { "epoch": 0.23930740866095446, "grad_norm": 1.9219238960186191, "learning_rate": 1.7789401243772307e-05, "loss": 0.6925, "step": 6717 }, { "epoch": 0.23934303578744856, "grad_norm": 2.1506069802631713, "learning_rate": 1.7788677560168782e-05, "loss": 0.5974, "step": 6718 }, { "epoch": 0.23937866291394266, "grad_norm": 2.024102097330683, "learning_rate": 1.7787953772853574e-05, "loss": 0.5604, "step": 6719 }, { "epoch": 0.2394142900404368, "grad_norm": 2.1108385038552724, "learning_rate": 1.778722988183632e-05, "loss": 0.6148, "step": 6720 }, { "epoch": 0.2394499171669309, "grad_norm": 2.411976742428864, "learning_rate": 1.7786505887126652e-05, "loss": 0.6305, "step": 6721 }, { "epoch": 0.239485544293425, "grad_norm": 1.519912855540856, "learning_rate": 1.778578178873422e-05, "loss": 0.6685, "step": 6722 }, { "epoch": 0.23952117141991913, "grad_norm": 1.4295545167048405, "learning_rate": 1.778505758666866e-05, "loss": 0.5086, "step": 6723 }, { "epoch": 0.23955679854641324, "grad_norm": 1.2943526509298808, "learning_rate": 1.7784333280939618e-05, "loss": 0.5427, "step": 6724 }, { "epoch": 0.23959242567290734, "grad_norm": 1.9150491963261511, "learning_rate": 1.7783608871556735e-05, "loss": 0.6555, "step": 6725 }, { "epoch": 0.23962805279940147, "grad_norm": 1.2397528069771515, "learning_rate": 1.778288435852966e-05, "loss": 0.4759, "step": 6726 }, { "epoch": 0.23966367992589557, "grad_norm": 1.6323558907709357, "learning_rate": 1.7782159741868042e-05, "loss": 0.5128, "step": 6727 }, { "epoch": 0.23969930705238968, "grad_norm": 1.3722188635984576, "learning_rate": 1.7781435021581527e-05, "loss": 0.5587, "step": 6728 }, { "epoch": 0.2397349341788838, "grad_norm": 1.2500055396361702, "learning_rate": 1.7780710197679765e-05, "loss": 0.5058, "step": 6729 }, { "epoch": 0.2397705613053779, "grad_norm": 1.9909976773779263, "learning_rate": 1.777998527017241e-05, "loss": 0.5465, "step": 6730 }, { "epoch": 0.239806188431872, "grad_norm": 1.2791309565272913, "learning_rate": 1.777926023906911e-05, "loss": 0.5112, "step": 6731 }, { "epoch": 0.23984181555836614, "grad_norm": 1.2284150657706885, "learning_rate": 1.7778535104379523e-05, "loss": 0.5031, "step": 6732 }, { "epoch": 0.23987744268486025, "grad_norm": 1.1907433517470307, "learning_rate": 1.777780986611331e-05, "loss": 0.4452, "step": 6733 }, { "epoch": 0.23991306981135435, "grad_norm": 1.1980354705045173, "learning_rate": 1.777708452428012e-05, "loss": 0.4879, "step": 6734 }, { "epoch": 0.23994869693784848, "grad_norm": 1.2168911309359014, "learning_rate": 1.7776359078889605e-05, "loss": 0.4645, "step": 6735 }, { "epoch": 0.23998432406434259, "grad_norm": 1.1763048321910408, "learning_rate": 1.7775633529951442e-05, "loss": 0.5135, "step": 6736 }, { "epoch": 0.24001995119083672, "grad_norm": 2.1837888306191675, "learning_rate": 1.777490787747528e-05, "loss": 0.5684, "step": 6737 }, { "epoch": 0.24005557831733082, "grad_norm": 1.1946474919083798, "learning_rate": 1.777418212147079e-05, "loss": 0.5091, "step": 6738 }, { "epoch": 0.24009120544382492, "grad_norm": 1.1703939842394526, "learning_rate": 1.7773456261947627e-05, "loss": 0.5389, "step": 6739 }, { "epoch": 0.24012683257031905, "grad_norm": 1.169473452373879, "learning_rate": 1.7772730298915463e-05, "loss": 0.4882, "step": 6740 }, { "epoch": 0.24016245969681316, "grad_norm": 2.4010234090040177, "learning_rate": 1.7772004232383963e-05, "loss": 0.5776, "step": 6741 }, { "epoch": 0.24019808682330726, "grad_norm": 1.159539483299781, "learning_rate": 1.7771278062362787e-05, "loss": 0.4864, "step": 6742 }, { "epoch": 0.2402337139498014, "grad_norm": 1.9497211059134432, "learning_rate": 1.777055178886162e-05, "loss": 0.6712, "step": 6743 }, { "epoch": 0.2402693410762955, "grad_norm": 1.5267290061860235, "learning_rate": 1.7769825411890122e-05, "loss": 0.549, "step": 6744 }, { "epoch": 0.2403049682027896, "grad_norm": 1.6868969714489306, "learning_rate": 1.7769098931457968e-05, "loss": 0.6087, "step": 6745 }, { "epoch": 0.24034059532928373, "grad_norm": 1.2603356060258029, "learning_rate": 1.7768372347574828e-05, "loss": 0.4887, "step": 6746 }, { "epoch": 0.24037622245577783, "grad_norm": 1.4565159745197398, "learning_rate": 1.7767645660250387e-05, "loss": 0.5387, "step": 6747 }, { "epoch": 0.24041184958227194, "grad_norm": 1.2562379654529157, "learning_rate": 1.776691886949431e-05, "loss": 0.5482, "step": 6748 }, { "epoch": 0.24044747670876607, "grad_norm": 2.2060515950438435, "learning_rate": 1.7766191975316282e-05, "loss": 0.6121, "step": 6749 }, { "epoch": 0.24048310383526017, "grad_norm": 1.595590549211668, "learning_rate": 1.7765464977725977e-05, "loss": 0.5311, "step": 6750 }, { "epoch": 0.24051873096175427, "grad_norm": 1.6948293115660387, "learning_rate": 1.776473787673308e-05, "loss": 0.6227, "step": 6751 }, { "epoch": 0.2405543580882484, "grad_norm": 1.2163549408799912, "learning_rate": 1.776401067234727e-05, "loss": 0.5213, "step": 6752 }, { "epoch": 0.2405899852147425, "grad_norm": 1.1703715572403148, "learning_rate": 1.7763283364578234e-05, "loss": 0.463, "step": 6753 }, { "epoch": 0.2406256123412366, "grad_norm": 1.2950536530474048, "learning_rate": 1.7762555953435653e-05, "loss": 0.5335, "step": 6754 }, { "epoch": 0.24066123946773074, "grad_norm": 1.107261978462621, "learning_rate": 1.776182843892921e-05, "loss": 0.5082, "step": 6755 }, { "epoch": 0.24069686659422485, "grad_norm": 1.473143278815113, "learning_rate": 1.7761100821068597e-05, "loss": 0.7357, "step": 6756 }, { "epoch": 0.24073249372071895, "grad_norm": 1.6645811476082297, "learning_rate": 1.7760373099863504e-05, "loss": 0.6147, "step": 6757 }, { "epoch": 0.24076812084721308, "grad_norm": 1.6309174769665475, "learning_rate": 1.775964527532362e-05, "loss": 0.5697, "step": 6758 }, { "epoch": 0.24080374797370718, "grad_norm": 1.9950095597893656, "learning_rate": 1.7758917347458634e-05, "loss": 0.7001, "step": 6759 }, { "epoch": 0.2408393751002013, "grad_norm": 1.8183528823317625, "learning_rate": 1.775818931627824e-05, "loss": 0.6774, "step": 6760 }, { "epoch": 0.24087500222669542, "grad_norm": 1.1688382733114528, "learning_rate": 1.775746118179213e-05, "loss": 0.4507, "step": 6761 }, { "epoch": 0.24091062935318952, "grad_norm": 1.1423651173266176, "learning_rate": 1.7756732944010007e-05, "loss": 0.4475, "step": 6762 }, { "epoch": 0.24094625647968362, "grad_norm": 1.176902449182575, "learning_rate": 1.775600460294156e-05, "loss": 0.51, "step": 6763 }, { "epoch": 0.24098188360617775, "grad_norm": 1.1571481356291038, "learning_rate": 1.775527615859649e-05, "loss": 0.4611, "step": 6764 }, { "epoch": 0.24101751073267186, "grad_norm": 1.1733484707029371, "learning_rate": 1.7754547610984502e-05, "loss": 0.4738, "step": 6765 }, { "epoch": 0.24105313785916596, "grad_norm": 1.5792075802372267, "learning_rate": 1.775381896011529e-05, "loss": 0.5709, "step": 6766 }, { "epoch": 0.2410887649856601, "grad_norm": 1.696998882946107, "learning_rate": 1.775309020599856e-05, "loss": 0.568, "step": 6767 }, { "epoch": 0.2411243921121542, "grad_norm": 2.1016777662494524, "learning_rate": 1.7752361348644012e-05, "loss": 0.5624, "step": 6768 }, { "epoch": 0.2411600192386483, "grad_norm": 1.0678857190958406, "learning_rate": 1.7751632388061354e-05, "loss": 0.414, "step": 6769 }, { "epoch": 0.24119564636514243, "grad_norm": 1.4034358530622288, "learning_rate": 1.7750903324260294e-05, "loss": 0.6068, "step": 6770 }, { "epoch": 0.24123127349163653, "grad_norm": 1.5339427142892668, "learning_rate": 1.775017415725054e-05, "loss": 0.7027, "step": 6771 }, { "epoch": 0.24126690061813064, "grad_norm": 1.0923771363140269, "learning_rate": 1.7749444887041797e-05, "loss": 0.443, "step": 6772 }, { "epoch": 0.24130252774462477, "grad_norm": 2.091463554789812, "learning_rate": 1.774871551364378e-05, "loss": 0.6727, "step": 6773 }, { "epoch": 0.24133815487111887, "grad_norm": 2.3292224216155684, "learning_rate": 1.77479860370662e-05, "loss": 0.6476, "step": 6774 }, { "epoch": 0.24137378199761297, "grad_norm": 1.172648333181028, "learning_rate": 1.7747256457318772e-05, "loss": 0.5767, "step": 6775 }, { "epoch": 0.2414094091241071, "grad_norm": 1.9938143990603685, "learning_rate": 1.7746526774411207e-05, "loss": 0.5842, "step": 6776 }, { "epoch": 0.2414450362506012, "grad_norm": 1.5638625147634484, "learning_rate": 1.7745796988353226e-05, "loss": 0.5674, "step": 6777 }, { "epoch": 0.2414806633770953, "grad_norm": 1.1815011899398422, "learning_rate": 1.774506709915454e-05, "loss": 0.4785, "step": 6778 }, { "epoch": 0.24151629050358944, "grad_norm": 2.3892078803425556, "learning_rate": 1.7744337106824878e-05, "loss": 0.6983, "step": 6779 }, { "epoch": 0.24155191763008355, "grad_norm": 1.1485499178958034, "learning_rate": 1.774360701137395e-05, "loss": 0.48, "step": 6780 }, { "epoch": 0.24158754475657765, "grad_norm": 1.8984182275881174, "learning_rate": 1.7742876812811485e-05, "loss": 0.6005, "step": 6781 }, { "epoch": 0.24162317188307178, "grad_norm": 1.1838754073058007, "learning_rate": 1.7742146511147202e-05, "loss": 0.5464, "step": 6782 }, { "epoch": 0.24165879900956588, "grad_norm": 1.1934231672250346, "learning_rate": 1.7741416106390828e-05, "loss": 0.4711, "step": 6783 }, { "epoch": 0.24169442613606, "grad_norm": 1.2661393490085642, "learning_rate": 1.774068559855208e-05, "loss": 0.531, "step": 6784 }, { "epoch": 0.24173005326255412, "grad_norm": 1.7631715126590704, "learning_rate": 1.7739954987640705e-05, "loss": 0.6083, "step": 6785 }, { "epoch": 0.24176568038904822, "grad_norm": 1.6949397581400896, "learning_rate": 1.773922427366641e-05, "loss": 0.5701, "step": 6786 }, { "epoch": 0.24180130751554232, "grad_norm": 1.3010955097775943, "learning_rate": 1.773849345663894e-05, "loss": 0.5363, "step": 6787 }, { "epoch": 0.24183693464203646, "grad_norm": 1.871880152631909, "learning_rate": 1.7737762536568017e-05, "loss": 0.665, "step": 6788 }, { "epoch": 0.24187256176853056, "grad_norm": 1.1555438324405953, "learning_rate": 1.773703151346338e-05, "loss": 0.5401, "step": 6789 }, { "epoch": 0.24190818889502466, "grad_norm": 1.1444913951014892, "learning_rate": 1.7736300387334758e-05, "loss": 0.4896, "step": 6790 }, { "epoch": 0.2419438160215188, "grad_norm": 1.5497537344709855, "learning_rate": 1.7735569158191892e-05, "loss": 0.6226, "step": 6791 }, { "epoch": 0.2419794431480129, "grad_norm": 1.0519918255711544, "learning_rate": 1.7734837826044514e-05, "loss": 0.4572, "step": 6792 }, { "epoch": 0.242015070274507, "grad_norm": 2.161735146201848, "learning_rate": 1.7734106390902364e-05, "loss": 0.6713, "step": 6793 }, { "epoch": 0.24205069740100113, "grad_norm": 1.728645706912107, "learning_rate": 1.7733374852775184e-05, "loss": 0.5563, "step": 6794 }, { "epoch": 0.24208632452749523, "grad_norm": 1.1004258082208125, "learning_rate": 1.7732643211672708e-05, "loss": 0.4876, "step": 6795 }, { "epoch": 0.24212195165398934, "grad_norm": 1.1287408161602692, "learning_rate": 1.7731911467604686e-05, "loss": 0.5296, "step": 6796 }, { "epoch": 0.24215757878048347, "grad_norm": 1.562060774967153, "learning_rate": 1.7731179620580857e-05, "loss": 0.6158, "step": 6797 }, { "epoch": 0.24219320590697757, "grad_norm": 1.2717123023644992, "learning_rate": 1.773044767061097e-05, "loss": 0.552, "step": 6798 }, { "epoch": 0.24222883303347167, "grad_norm": 1.2546069474203436, "learning_rate": 1.772971561770477e-05, "loss": 0.4672, "step": 6799 }, { "epoch": 0.2422644601599658, "grad_norm": 1.0769343681574748, "learning_rate": 1.7728983461872002e-05, "loss": 0.5156, "step": 6800 }, { "epoch": 0.2423000872864599, "grad_norm": 1.6053690066505695, "learning_rate": 1.7728251203122414e-05, "loss": 0.6234, "step": 6801 }, { "epoch": 0.242335714412954, "grad_norm": 1.613656656259622, "learning_rate": 1.772751884146576e-05, "loss": 0.6448, "step": 6802 }, { "epoch": 0.24237134153944814, "grad_norm": 1.1982805823748879, "learning_rate": 1.7726786376911795e-05, "loss": 0.486, "step": 6803 }, { "epoch": 0.24240696866594225, "grad_norm": 1.6513324534428335, "learning_rate": 1.7726053809470266e-05, "loss": 0.6307, "step": 6804 }, { "epoch": 0.24244259579243635, "grad_norm": 1.6512019682102714, "learning_rate": 1.772532113915093e-05, "loss": 0.6322, "step": 6805 }, { "epoch": 0.24247822291893048, "grad_norm": 1.199389691350742, "learning_rate": 1.7724588365963546e-05, "loss": 0.5241, "step": 6806 }, { "epoch": 0.24251385004542458, "grad_norm": 1.2663276275746878, "learning_rate": 1.7723855489917868e-05, "loss": 0.4484, "step": 6807 }, { "epoch": 0.2425494771719187, "grad_norm": 1.2278134587959215, "learning_rate": 1.7723122511023654e-05, "loss": 0.5776, "step": 6808 }, { "epoch": 0.24258510429841282, "grad_norm": 1.244414435099792, "learning_rate": 1.7722389429290668e-05, "loss": 0.512, "step": 6809 }, { "epoch": 0.24262073142490692, "grad_norm": 1.579608452085103, "learning_rate": 1.7721656244728665e-05, "loss": 0.6107, "step": 6810 }, { "epoch": 0.24265635855140102, "grad_norm": 1.3132860126277417, "learning_rate": 1.7720922957347417e-05, "loss": 0.4922, "step": 6811 }, { "epoch": 0.24269198567789516, "grad_norm": 1.607366753254715, "learning_rate": 1.772018956715668e-05, "loss": 0.5781, "step": 6812 }, { "epoch": 0.24272761280438926, "grad_norm": 1.6128798395733441, "learning_rate": 1.771945607416622e-05, "loss": 0.6968, "step": 6813 }, { "epoch": 0.24276323993088336, "grad_norm": 1.5552051580958042, "learning_rate": 1.7718722478385815e-05, "loss": 0.639, "step": 6814 }, { "epoch": 0.2427988670573775, "grad_norm": 1.1357797850295053, "learning_rate": 1.771798877982522e-05, "loss": 0.518, "step": 6815 }, { "epoch": 0.2428344941838716, "grad_norm": 2.5438140424419338, "learning_rate": 1.771725497849421e-05, "loss": 0.5827, "step": 6816 }, { "epoch": 0.2428701213103657, "grad_norm": 1.3119845174129272, "learning_rate": 1.7716521074402554e-05, "loss": 0.5158, "step": 6817 }, { "epoch": 0.24290574843685983, "grad_norm": 1.2357783574092378, "learning_rate": 1.771578706756003e-05, "loss": 0.5263, "step": 6818 }, { "epoch": 0.24294137556335393, "grad_norm": 1.1520617906056938, "learning_rate": 1.7715052957976406e-05, "loss": 0.5322, "step": 6819 }, { "epoch": 0.24297700268984804, "grad_norm": 1.0890888949431665, "learning_rate": 1.771431874566146e-05, "loss": 0.4536, "step": 6820 }, { "epoch": 0.24301262981634217, "grad_norm": 1.06192163325191, "learning_rate": 1.771358443062497e-05, "loss": 0.4543, "step": 6821 }, { "epoch": 0.24304825694283627, "grad_norm": 1.392798348616237, "learning_rate": 1.7712850012876708e-05, "loss": 0.502, "step": 6822 }, { "epoch": 0.24308388406933037, "grad_norm": 1.1223723340939442, "learning_rate": 1.771211549242646e-05, "loss": 0.491, "step": 6823 }, { "epoch": 0.2431195111958245, "grad_norm": 1.2755278201019271, "learning_rate": 1.7711380869284e-05, "loss": 0.5245, "step": 6824 }, { "epoch": 0.2431551383223186, "grad_norm": 1.3669868784206332, "learning_rate": 1.771064614345912e-05, "loss": 0.642, "step": 6825 }, { "epoch": 0.2431907654488127, "grad_norm": 1.6746020876452474, "learning_rate": 1.7709911314961593e-05, "loss": 0.6203, "step": 6826 }, { "epoch": 0.24322639257530684, "grad_norm": 1.0947368842270544, "learning_rate": 1.770917638380121e-05, "loss": 0.5019, "step": 6827 }, { "epoch": 0.24326201970180095, "grad_norm": 1.5933353442335878, "learning_rate": 1.7708441349987753e-05, "loss": 0.6104, "step": 6828 }, { "epoch": 0.24329764682829508, "grad_norm": 1.587005874386873, "learning_rate": 1.7707706213531012e-05, "loss": 0.6394, "step": 6829 }, { "epoch": 0.24333327395478918, "grad_norm": 1.378146517037901, "learning_rate": 1.7706970974440776e-05, "loss": 0.5593, "step": 6830 }, { "epoch": 0.24336890108128328, "grad_norm": 1.5667624523934003, "learning_rate": 1.7706235632726834e-05, "loss": 0.583, "step": 6831 }, { "epoch": 0.24340452820777742, "grad_norm": 1.3241896639649728, "learning_rate": 1.7705500188398977e-05, "loss": 0.5345, "step": 6832 }, { "epoch": 0.24344015533427152, "grad_norm": 2.024486294233077, "learning_rate": 1.7704764641467004e-05, "loss": 0.6001, "step": 6833 }, { "epoch": 0.24347578246076562, "grad_norm": 1.1430265566336768, "learning_rate": 1.77040289919407e-05, "loss": 0.4481, "step": 6834 }, { "epoch": 0.24351140958725975, "grad_norm": 1.3523124161132338, "learning_rate": 1.7703293239829865e-05, "loss": 0.538, "step": 6835 }, { "epoch": 0.24354703671375386, "grad_norm": 1.4438323633810597, "learning_rate": 1.77025573851443e-05, "loss": 0.5873, "step": 6836 }, { "epoch": 0.24358266384024796, "grad_norm": 1.7290296088797739, "learning_rate": 1.77018214278938e-05, "loss": 0.594, "step": 6837 }, { "epoch": 0.2436182909667421, "grad_norm": 1.9311914492660671, "learning_rate": 1.7701085368088157e-05, "loss": 0.6168, "step": 6838 }, { "epoch": 0.2436539180932362, "grad_norm": 1.2296815660491411, "learning_rate": 1.7700349205737186e-05, "loss": 0.4821, "step": 6839 }, { "epoch": 0.2436895452197303, "grad_norm": 1.5901793964366402, "learning_rate": 1.769961294085068e-05, "loss": 0.6346, "step": 6840 }, { "epoch": 0.24372517234622443, "grad_norm": 1.0591375358814328, "learning_rate": 1.7698876573438448e-05, "loss": 0.5085, "step": 6841 }, { "epoch": 0.24376079947271853, "grad_norm": 1.6857731265256102, "learning_rate": 1.7698140103510294e-05, "loss": 0.6513, "step": 6842 }, { "epoch": 0.24379642659921263, "grad_norm": 1.2503450892677503, "learning_rate": 1.769740353107602e-05, "loss": 0.5317, "step": 6843 }, { "epoch": 0.24383205372570677, "grad_norm": 1.8133790762009612, "learning_rate": 1.7696666856145443e-05, "loss": 0.6228, "step": 6844 }, { "epoch": 0.24386768085220087, "grad_norm": 1.526184126084776, "learning_rate": 1.769593007872836e-05, "loss": 0.6657, "step": 6845 }, { "epoch": 0.24390330797869497, "grad_norm": 1.136270168668862, "learning_rate": 1.7695193198834598e-05, "loss": 0.4695, "step": 6846 }, { "epoch": 0.2439389351051891, "grad_norm": 1.702205670095369, "learning_rate": 1.7694456216473955e-05, "loss": 0.7148, "step": 6847 }, { "epoch": 0.2439745622316832, "grad_norm": 1.923375883283381, "learning_rate": 1.7693719131656247e-05, "loss": 0.5972, "step": 6848 }, { "epoch": 0.2440101893581773, "grad_norm": 1.203133894260075, "learning_rate": 1.7692981944391293e-05, "loss": 0.4966, "step": 6849 }, { "epoch": 0.24404581648467144, "grad_norm": 1.3464612243282235, "learning_rate": 1.769224465468891e-05, "loss": 0.5773, "step": 6850 }, { "epoch": 0.24408144361116554, "grad_norm": 1.3741640628437903, "learning_rate": 1.769150726255891e-05, "loss": 0.5121, "step": 6851 }, { "epoch": 0.24411707073765965, "grad_norm": 2.1209264340474996, "learning_rate": 1.7690769768011118e-05, "loss": 0.5798, "step": 6852 }, { "epoch": 0.24415269786415378, "grad_norm": 1.3945408551372607, "learning_rate": 1.769003217105535e-05, "loss": 0.5435, "step": 6853 }, { "epoch": 0.24418832499064788, "grad_norm": 1.210872481969244, "learning_rate": 1.7689294471701424e-05, "loss": 0.488, "step": 6854 }, { "epoch": 0.24422395211714198, "grad_norm": 1.040682425937143, "learning_rate": 1.7688556669959173e-05, "loss": 0.4426, "step": 6855 }, { "epoch": 0.24425957924363612, "grad_norm": 2.587768061280059, "learning_rate": 1.7687818765838415e-05, "loss": 0.5702, "step": 6856 }, { "epoch": 0.24429520637013022, "grad_norm": 1.519511607918434, "learning_rate": 1.7687080759348975e-05, "loss": 0.5274, "step": 6857 }, { "epoch": 0.24433083349662432, "grad_norm": 1.6988644808546434, "learning_rate": 1.7686342650500682e-05, "loss": 0.5847, "step": 6858 }, { "epoch": 0.24436646062311845, "grad_norm": 1.3878073696238025, "learning_rate": 1.7685604439303365e-05, "loss": 0.5675, "step": 6859 }, { "epoch": 0.24440208774961256, "grad_norm": 1.5002593892550384, "learning_rate": 1.7684866125766853e-05, "loss": 0.6192, "step": 6860 }, { "epoch": 0.24443771487610666, "grad_norm": 1.160165293317572, "learning_rate": 1.7684127709900976e-05, "loss": 0.5603, "step": 6861 }, { "epoch": 0.2444733420026008, "grad_norm": 1.3039636721668861, "learning_rate": 1.7683389191715568e-05, "loss": 0.4916, "step": 6862 }, { "epoch": 0.2445089691290949, "grad_norm": 1.1304340641828348, "learning_rate": 1.7682650571220462e-05, "loss": 0.5492, "step": 6863 }, { "epoch": 0.244544596255589, "grad_norm": 1.5963038595284609, "learning_rate": 1.7681911848425498e-05, "loss": 0.5617, "step": 6864 }, { "epoch": 0.24458022338208313, "grad_norm": 2.0904671847181886, "learning_rate": 1.76811730233405e-05, "loss": 0.6327, "step": 6865 }, { "epoch": 0.24461585050857723, "grad_norm": 1.992620006200536, "learning_rate": 1.7680434095975322e-05, "loss": 0.6263, "step": 6866 }, { "epoch": 0.24465147763507133, "grad_norm": 1.1187730674040937, "learning_rate": 1.7679695066339792e-05, "loss": 0.481, "step": 6867 }, { "epoch": 0.24468710476156547, "grad_norm": 1.3146947339150528, "learning_rate": 1.7678955934443757e-05, "loss": 0.5018, "step": 6868 }, { "epoch": 0.24472273188805957, "grad_norm": 1.657779842607416, "learning_rate": 1.7678216700297056e-05, "loss": 0.6064, "step": 6869 }, { "epoch": 0.24475835901455367, "grad_norm": 1.1561146561457332, "learning_rate": 1.7677477363909534e-05, "loss": 0.5052, "step": 6870 }, { "epoch": 0.2447939861410478, "grad_norm": 1.5443338152528905, "learning_rate": 1.7676737925291027e-05, "loss": 0.5312, "step": 6871 }, { "epoch": 0.2448296132675419, "grad_norm": 1.4329230323707278, "learning_rate": 1.7675998384451398e-05, "loss": 0.6449, "step": 6872 }, { "epoch": 0.244865240394036, "grad_norm": 1.5374530715538677, "learning_rate": 1.767525874140048e-05, "loss": 0.5628, "step": 6873 }, { "epoch": 0.24490086752053014, "grad_norm": 1.1757213663064279, "learning_rate": 1.7674518996148128e-05, "loss": 0.5361, "step": 6874 }, { "epoch": 0.24493649464702424, "grad_norm": 1.804191346109003, "learning_rate": 1.7673779148704192e-05, "loss": 0.5976, "step": 6875 }, { "epoch": 0.24497212177351835, "grad_norm": 1.5674181355042884, "learning_rate": 1.7673039199078525e-05, "loss": 0.6328, "step": 6876 }, { "epoch": 0.24500774890001248, "grad_norm": 1.6335538626495378, "learning_rate": 1.7672299147280975e-05, "loss": 0.5852, "step": 6877 }, { "epoch": 0.24504337602650658, "grad_norm": 1.3132510503846768, "learning_rate": 1.7671558993321403e-05, "loss": 0.5247, "step": 6878 }, { "epoch": 0.24507900315300069, "grad_norm": 1.1452834806120367, "learning_rate": 1.7670818737209657e-05, "loss": 0.524, "step": 6879 }, { "epoch": 0.24511463027949482, "grad_norm": 1.463724743144088, "learning_rate": 1.76700783789556e-05, "loss": 0.583, "step": 6880 }, { "epoch": 0.24515025740598892, "grad_norm": 1.1991642414679133, "learning_rate": 1.7669337918569086e-05, "loss": 0.5143, "step": 6881 }, { "epoch": 0.24518588453248302, "grad_norm": 1.6240854456465474, "learning_rate": 1.7668597356059977e-05, "loss": 0.5988, "step": 6882 }, { "epoch": 0.24522151165897715, "grad_norm": 1.267013918943131, "learning_rate": 1.766785669143814e-05, "loss": 0.5268, "step": 6883 }, { "epoch": 0.24525713878547126, "grad_norm": 1.1279894718599646, "learning_rate": 1.7667115924713427e-05, "loss": 0.485, "step": 6884 }, { "epoch": 0.24529276591196536, "grad_norm": 1.741412090181701, "learning_rate": 1.7666375055895707e-05, "loss": 0.6441, "step": 6885 }, { "epoch": 0.2453283930384595, "grad_norm": 1.1382445923694098, "learning_rate": 1.7665634084994842e-05, "loss": 0.4087, "step": 6886 }, { "epoch": 0.2453640201649536, "grad_norm": 1.183998136604507, "learning_rate": 1.7664893012020704e-05, "loss": 0.5554, "step": 6887 }, { "epoch": 0.2453996472914477, "grad_norm": 1.6045833211828484, "learning_rate": 1.7664151836983157e-05, "loss": 0.5378, "step": 6888 }, { "epoch": 0.24543527441794183, "grad_norm": 1.699272873259646, "learning_rate": 1.7663410559892073e-05, "loss": 0.6306, "step": 6889 }, { "epoch": 0.24547090154443593, "grad_norm": 1.0955515200359602, "learning_rate": 1.766266918075732e-05, "loss": 0.5182, "step": 6890 }, { "epoch": 0.24550652867093004, "grad_norm": 1.1124799727544845, "learning_rate": 1.7661927699588772e-05, "loss": 0.4828, "step": 6891 }, { "epoch": 0.24554215579742417, "grad_norm": 1.423999398593276, "learning_rate": 1.7661186116396304e-05, "loss": 0.5423, "step": 6892 }, { "epoch": 0.24557778292391827, "grad_norm": 1.1547502299912376, "learning_rate": 1.766044443118978e-05, "loss": 0.4986, "step": 6893 }, { "epoch": 0.24561341005041237, "grad_norm": 1.1681599599833388, "learning_rate": 1.765970264397909e-05, "loss": 0.5055, "step": 6894 }, { "epoch": 0.2456490371769065, "grad_norm": 1.73545349991241, "learning_rate": 1.7658960754774104e-05, "loss": 0.6611, "step": 6895 }, { "epoch": 0.2456846643034006, "grad_norm": 1.1285220112265664, "learning_rate": 1.7658218763584704e-05, "loss": 0.514, "step": 6896 }, { "epoch": 0.2457202914298947, "grad_norm": 1.879870057514506, "learning_rate": 1.7657476670420766e-05, "loss": 0.6262, "step": 6897 }, { "epoch": 0.24575591855638884, "grad_norm": 1.8831878873230323, "learning_rate": 1.7656734475292175e-05, "loss": 0.5726, "step": 6898 }, { "epoch": 0.24579154568288294, "grad_norm": 1.1095169632702477, "learning_rate": 1.7655992178208813e-05, "loss": 0.5218, "step": 6899 }, { "epoch": 0.24582717280937705, "grad_norm": 1.8255590251378462, "learning_rate": 1.765524977918056e-05, "loss": 0.565, "step": 6900 }, { "epoch": 0.24586279993587118, "grad_norm": 1.4868351387639633, "learning_rate": 1.765450727821731e-05, "loss": 0.5622, "step": 6901 }, { "epoch": 0.24589842706236528, "grad_norm": 1.5606888992736117, "learning_rate": 1.7653764675328945e-05, "loss": 0.5431, "step": 6902 }, { "epoch": 0.24593405418885939, "grad_norm": 1.045599031563697, "learning_rate": 1.7653021970525356e-05, "loss": 0.4509, "step": 6903 }, { "epoch": 0.24596968131535352, "grad_norm": 1.5631923820056237, "learning_rate": 1.7652279163816422e-05, "loss": 0.5848, "step": 6904 }, { "epoch": 0.24600530844184762, "grad_norm": 1.6956775072399328, "learning_rate": 1.765153625521205e-05, "loss": 0.5569, "step": 6905 }, { "epoch": 0.24604093556834172, "grad_norm": 1.8745736976259995, "learning_rate": 1.765079324472212e-05, "loss": 0.69, "step": 6906 }, { "epoch": 0.24607656269483585, "grad_norm": 3.2107885160124767, "learning_rate": 1.7650050132356537e-05, "loss": 0.6199, "step": 6907 }, { "epoch": 0.24611218982132996, "grad_norm": 1.5009609368562358, "learning_rate": 1.7649306918125182e-05, "loss": 0.6266, "step": 6908 }, { "epoch": 0.24614781694782406, "grad_norm": 1.651133305305048, "learning_rate": 1.7648563602037962e-05, "loss": 0.5761, "step": 6909 }, { "epoch": 0.2461834440743182, "grad_norm": 1.272118298038986, "learning_rate": 1.764782018410477e-05, "loss": 0.5836, "step": 6910 }, { "epoch": 0.2462190712008123, "grad_norm": 1.1834897893597287, "learning_rate": 1.764707666433551e-05, "loss": 0.4622, "step": 6911 }, { "epoch": 0.2462546983273064, "grad_norm": 1.6684336456995248, "learning_rate": 1.7646333042740077e-05, "loss": 0.5777, "step": 6912 }, { "epoch": 0.24629032545380053, "grad_norm": 1.1778081716726387, "learning_rate": 1.7645589319328374e-05, "loss": 0.5436, "step": 6913 }, { "epoch": 0.24632595258029463, "grad_norm": 1.7590347262020931, "learning_rate": 1.764484549411031e-05, "loss": 0.5826, "step": 6914 }, { "epoch": 0.24636157970678874, "grad_norm": 1.4729080716173577, "learning_rate": 1.764410156709578e-05, "loss": 0.558, "step": 6915 }, { "epoch": 0.24639720683328287, "grad_norm": 1.664243669802097, "learning_rate": 1.7643357538294697e-05, "loss": 0.6379, "step": 6916 }, { "epoch": 0.24643283395977697, "grad_norm": 1.282803127688293, "learning_rate": 1.7642613407716963e-05, "loss": 0.5438, "step": 6917 }, { "epoch": 0.24646846108627107, "grad_norm": 1.2602106854100186, "learning_rate": 1.7641869175372493e-05, "loss": 0.5324, "step": 6918 }, { "epoch": 0.2465040882127652, "grad_norm": 1.8402226632872483, "learning_rate": 1.7641124841271194e-05, "loss": 0.5569, "step": 6919 }, { "epoch": 0.2465397153392593, "grad_norm": 1.287781532516619, "learning_rate": 1.7640380405422976e-05, "loss": 0.5039, "step": 6920 }, { "epoch": 0.2465753424657534, "grad_norm": 1.1812859417216894, "learning_rate": 1.763963586783775e-05, "loss": 0.5242, "step": 6921 }, { "epoch": 0.24661096959224754, "grad_norm": 1.3200041790423664, "learning_rate": 1.7638891228525434e-05, "loss": 0.5191, "step": 6922 }, { "epoch": 0.24664659671874165, "grad_norm": 1.1807416428482764, "learning_rate": 1.763814648749594e-05, "loss": 0.4952, "step": 6923 }, { "epoch": 0.24668222384523578, "grad_norm": 1.6623945026620819, "learning_rate": 1.763740164475919e-05, "loss": 0.7034, "step": 6924 }, { "epoch": 0.24671785097172988, "grad_norm": 1.2725157584438171, "learning_rate": 1.7636656700325102e-05, "loss": 0.4813, "step": 6925 }, { "epoch": 0.24675347809822398, "grad_norm": 1.2235310262772896, "learning_rate": 1.7635911654203588e-05, "loss": 0.5149, "step": 6926 }, { "epoch": 0.2467891052247181, "grad_norm": 1.5947611587981547, "learning_rate": 1.7635166506404574e-05, "loss": 0.5927, "step": 6927 }, { "epoch": 0.24682473235121222, "grad_norm": 1.5343921662642812, "learning_rate": 1.763442125693798e-05, "loss": 0.564, "step": 6928 }, { "epoch": 0.24686035947770632, "grad_norm": 1.1965610735277483, "learning_rate": 1.7633675905813732e-05, "loss": 0.4893, "step": 6929 }, { "epoch": 0.24689598660420045, "grad_norm": 1.819327931448111, "learning_rate": 1.7632930453041752e-05, "loss": 0.6017, "step": 6930 }, { "epoch": 0.24693161373069455, "grad_norm": 1.2195321406450723, "learning_rate": 1.763218489863197e-05, "loss": 0.541, "step": 6931 }, { "epoch": 0.24696724085718866, "grad_norm": 1.5501085135453765, "learning_rate": 1.7631439242594314e-05, "loss": 0.5057, "step": 6932 }, { "epoch": 0.2470028679836828, "grad_norm": 1.8103467369296131, "learning_rate": 1.7630693484938707e-05, "loss": 0.5321, "step": 6933 }, { "epoch": 0.2470384951101769, "grad_norm": 1.3809699586777897, "learning_rate": 1.762994762567508e-05, "loss": 0.4851, "step": 6934 }, { "epoch": 0.247074122236671, "grad_norm": 1.690629396781845, "learning_rate": 1.762920166481337e-05, "loss": 0.543, "step": 6935 }, { "epoch": 0.24710974936316513, "grad_norm": 1.2831611862601617, "learning_rate": 1.762845560236351e-05, "loss": 0.5576, "step": 6936 }, { "epoch": 0.24714537648965923, "grad_norm": 1.8962310105491167, "learning_rate": 1.762770943833543e-05, "loss": 0.7294, "step": 6937 }, { "epoch": 0.24718100361615333, "grad_norm": 1.9110571078942513, "learning_rate": 1.7626963172739066e-05, "loss": 0.5689, "step": 6938 }, { "epoch": 0.24721663074264746, "grad_norm": 1.5909825475287824, "learning_rate": 1.7626216805584357e-05, "loss": 0.6153, "step": 6939 }, { "epoch": 0.24725225786914157, "grad_norm": 1.7668921787318532, "learning_rate": 1.762547033688124e-05, "loss": 0.6496, "step": 6940 }, { "epoch": 0.24728788499563567, "grad_norm": 1.2270696420391356, "learning_rate": 1.7624723766639658e-05, "loss": 0.492, "step": 6941 }, { "epoch": 0.2473235121221298, "grad_norm": 2.2106549551588093, "learning_rate": 1.7623977094869547e-05, "loss": 0.5887, "step": 6942 }, { "epoch": 0.2473591392486239, "grad_norm": 2.323645920715098, "learning_rate": 1.7623230321580856e-05, "loss": 0.6556, "step": 6943 }, { "epoch": 0.247394766375118, "grad_norm": 1.1956844248434966, "learning_rate": 1.7622483446783523e-05, "loss": 0.4993, "step": 6944 }, { "epoch": 0.24743039350161214, "grad_norm": 1.1459721875402669, "learning_rate": 1.7621736470487493e-05, "loss": 0.5153, "step": 6945 }, { "epoch": 0.24746602062810624, "grad_norm": 1.1285894934847966, "learning_rate": 1.7620989392702715e-05, "loss": 0.4954, "step": 6946 }, { "epoch": 0.24750164775460035, "grad_norm": 1.2944379428258765, "learning_rate": 1.762024221343914e-05, "loss": 0.4839, "step": 6947 }, { "epoch": 0.24753727488109448, "grad_norm": 1.2774944873634615, "learning_rate": 1.761949493270671e-05, "loss": 0.5192, "step": 6948 }, { "epoch": 0.24757290200758858, "grad_norm": 1.5534648143958372, "learning_rate": 1.761874755051538e-05, "loss": 0.6495, "step": 6949 }, { "epoch": 0.24760852913408268, "grad_norm": 1.2301267316468778, "learning_rate": 1.7618000066875104e-05, "loss": 0.4959, "step": 6950 }, { "epoch": 0.24764415626057681, "grad_norm": 1.5060118592967238, "learning_rate": 1.761725248179583e-05, "loss": 0.6024, "step": 6951 }, { "epoch": 0.24767978338707092, "grad_norm": 1.5398381109602772, "learning_rate": 1.7616504795287516e-05, "loss": 0.5856, "step": 6952 }, { "epoch": 0.24771541051356502, "grad_norm": 1.6457533801745006, "learning_rate": 1.7615757007360114e-05, "loss": 0.5755, "step": 6953 }, { "epoch": 0.24775103764005915, "grad_norm": 2.101654828073708, "learning_rate": 1.7615009118023587e-05, "loss": 0.5662, "step": 6954 }, { "epoch": 0.24778666476655326, "grad_norm": 1.583627543956879, "learning_rate": 1.761426112728789e-05, "loss": 0.6151, "step": 6955 }, { "epoch": 0.24782229189304736, "grad_norm": 1.1523936148926255, "learning_rate": 1.7613513035162983e-05, "loss": 0.5029, "step": 6956 }, { "epoch": 0.2478579190195415, "grad_norm": 2.3072385749017315, "learning_rate": 1.761276484165883e-05, "loss": 0.6013, "step": 6957 }, { "epoch": 0.2478935461460356, "grad_norm": 1.2025393216958757, "learning_rate": 1.7612016546785396e-05, "loss": 0.5392, "step": 6958 }, { "epoch": 0.2479291732725297, "grad_norm": 1.1224994195752471, "learning_rate": 1.7611268150552637e-05, "loss": 0.489, "step": 6959 }, { "epoch": 0.24796480039902383, "grad_norm": 1.1245759098919144, "learning_rate": 1.7610519652970525e-05, "loss": 0.5126, "step": 6960 }, { "epoch": 0.24800042752551793, "grad_norm": 2.296806058556858, "learning_rate": 1.760977105404902e-05, "loss": 0.6111, "step": 6961 }, { "epoch": 0.24803605465201203, "grad_norm": 2.020500223220003, "learning_rate": 1.7609022353798097e-05, "loss": 0.607, "step": 6962 }, { "epoch": 0.24807168177850616, "grad_norm": 1.5491655779917886, "learning_rate": 1.7608273552227723e-05, "loss": 0.5481, "step": 6963 }, { "epoch": 0.24810730890500027, "grad_norm": 1.1955260900923095, "learning_rate": 1.7607524649347867e-05, "loss": 0.507, "step": 6964 }, { "epoch": 0.24814293603149437, "grad_norm": 1.1187976300619873, "learning_rate": 1.7606775645168506e-05, "loss": 0.4722, "step": 6965 }, { "epoch": 0.2481785631579885, "grad_norm": 1.780147241211203, "learning_rate": 1.760602653969961e-05, "loss": 0.5575, "step": 6966 }, { "epoch": 0.2482141902844826, "grad_norm": 1.5611802432974915, "learning_rate": 1.7605277332951155e-05, "loss": 0.7257, "step": 6967 }, { "epoch": 0.2482498174109767, "grad_norm": 1.2034143756774989, "learning_rate": 1.7604528024933113e-05, "loss": 0.5182, "step": 6968 }, { "epoch": 0.24828544453747084, "grad_norm": 1.2168327925560944, "learning_rate": 1.7603778615655468e-05, "loss": 0.5084, "step": 6969 }, { "epoch": 0.24832107166396494, "grad_norm": 1.988236631101003, "learning_rate": 1.7603029105128193e-05, "loss": 0.6434, "step": 6970 }, { "epoch": 0.24835669879045905, "grad_norm": 1.3022736175994265, "learning_rate": 1.7602279493361277e-05, "loss": 0.552, "step": 6971 }, { "epoch": 0.24839232591695318, "grad_norm": 1.201871231311771, "learning_rate": 1.7601529780364692e-05, "loss": 0.5298, "step": 6972 }, { "epoch": 0.24842795304344728, "grad_norm": 1.8276135535316351, "learning_rate": 1.760077996614842e-05, "loss": 0.6302, "step": 6973 }, { "epoch": 0.24846358016994138, "grad_norm": 1.6478833308006282, "learning_rate": 1.760003005072246e-05, "loss": 0.5752, "step": 6974 }, { "epoch": 0.24849920729643551, "grad_norm": 2.5639439127099326, "learning_rate": 1.759928003409678e-05, "loss": 0.6372, "step": 6975 }, { "epoch": 0.24853483442292962, "grad_norm": 1.6290196328479734, "learning_rate": 1.759852991628138e-05, "loss": 0.5855, "step": 6976 }, { "epoch": 0.24857046154942372, "grad_norm": 1.4392942159140585, "learning_rate": 1.759777969728624e-05, "loss": 0.5673, "step": 6977 }, { "epoch": 0.24860608867591785, "grad_norm": 1.3954094584599075, "learning_rate": 1.7597029377121356e-05, "loss": 0.6218, "step": 6978 }, { "epoch": 0.24864171580241196, "grad_norm": 1.680128114405439, "learning_rate": 1.759627895579671e-05, "loss": 0.586, "step": 6979 }, { "epoch": 0.24867734292890606, "grad_norm": 1.9316228457224296, "learning_rate": 1.7595528433322308e-05, "loss": 0.6301, "step": 6980 }, { "epoch": 0.2487129700554002, "grad_norm": 1.9219032332286365, "learning_rate": 1.759477780970813e-05, "loss": 0.676, "step": 6981 }, { "epoch": 0.2487485971818943, "grad_norm": 1.2700031802448521, "learning_rate": 1.759402708496418e-05, "loss": 0.5186, "step": 6982 }, { "epoch": 0.2487842243083884, "grad_norm": 1.658617246894938, "learning_rate": 1.7593276259100447e-05, "loss": 0.545, "step": 6983 }, { "epoch": 0.24881985143488253, "grad_norm": 1.2724457495544717, "learning_rate": 1.7592525332126938e-05, "loss": 0.5387, "step": 6984 }, { "epoch": 0.24885547856137663, "grad_norm": 1.9768791118558602, "learning_rate": 1.7591774304053646e-05, "loss": 0.6076, "step": 6985 }, { "epoch": 0.24889110568787073, "grad_norm": 1.078793920476475, "learning_rate": 1.759102317489057e-05, "loss": 0.4936, "step": 6986 }, { "epoch": 0.24892673281436486, "grad_norm": 1.2024370264382624, "learning_rate": 1.7590271944647718e-05, "loss": 0.5272, "step": 6987 }, { "epoch": 0.24896235994085897, "grad_norm": 1.1420256679141934, "learning_rate": 1.758952061333509e-05, "loss": 0.4889, "step": 6988 }, { "epoch": 0.24899798706735307, "grad_norm": 1.336655495018078, "learning_rate": 1.758876918096269e-05, "loss": 0.5222, "step": 6989 }, { "epoch": 0.2490336141938472, "grad_norm": 2.0676883323509636, "learning_rate": 1.758801764754052e-05, "loss": 0.5578, "step": 6990 }, { "epoch": 0.2490692413203413, "grad_norm": 1.2378898082744465, "learning_rate": 1.7587266013078596e-05, "loss": 0.4713, "step": 6991 }, { "epoch": 0.2491048684468354, "grad_norm": 1.1687927772997653, "learning_rate": 1.758651427758692e-05, "loss": 0.4488, "step": 6992 }, { "epoch": 0.24914049557332954, "grad_norm": 1.1051932341155137, "learning_rate": 1.7585762441075504e-05, "loss": 0.5391, "step": 6993 }, { "epoch": 0.24917612269982364, "grad_norm": 1.2435537962694594, "learning_rate": 1.758501050355436e-05, "loss": 0.5145, "step": 6994 }, { "epoch": 0.24921174982631775, "grad_norm": 1.2484394198096551, "learning_rate": 1.75842584650335e-05, "loss": 0.5225, "step": 6995 }, { "epoch": 0.24924737695281188, "grad_norm": 1.633043442727366, "learning_rate": 1.7583506325522936e-05, "loss": 0.586, "step": 6996 }, { "epoch": 0.24928300407930598, "grad_norm": 1.2355581236152808, "learning_rate": 1.7582754085032683e-05, "loss": 0.5523, "step": 6997 }, { "epoch": 0.24931863120580008, "grad_norm": 2.107443573809832, "learning_rate": 1.7582001743572763e-05, "loss": 0.5791, "step": 6998 }, { "epoch": 0.24935425833229422, "grad_norm": 1.7983382593143908, "learning_rate": 1.7581249301153187e-05, "loss": 0.6022, "step": 6999 }, { "epoch": 0.24938988545878832, "grad_norm": 1.5499975413654488, "learning_rate": 1.758049675778398e-05, "loss": 0.6798, "step": 7000 }, { "epoch": 0.24942551258528242, "grad_norm": 1.6768341532956312, "learning_rate": 1.757974411347516e-05, "loss": 0.6783, "step": 7001 }, { "epoch": 0.24946113971177655, "grad_norm": 1.589313030362335, "learning_rate": 1.757899136823675e-05, "loss": 0.6485, "step": 7002 }, { "epoch": 0.24949676683827066, "grad_norm": 1.525986585123559, "learning_rate": 1.757823852207877e-05, "loss": 0.6765, "step": 7003 }, { "epoch": 0.24953239396476476, "grad_norm": 1.4513688492242307, "learning_rate": 1.757748557501125e-05, "loss": 0.6356, "step": 7004 }, { "epoch": 0.2495680210912589, "grad_norm": 1.4839417942831659, "learning_rate": 1.757673252704421e-05, "loss": 0.6026, "step": 7005 }, { "epoch": 0.249603648217753, "grad_norm": 1.3043125471077255, "learning_rate": 1.7575979378187683e-05, "loss": 0.5414, "step": 7006 }, { "epoch": 0.2496392753442471, "grad_norm": 1.384720576537372, "learning_rate": 1.757522612845169e-05, "loss": 0.5417, "step": 7007 }, { "epoch": 0.24967490247074123, "grad_norm": 1.1800783364885907, "learning_rate": 1.7574472777846276e-05, "loss": 0.4917, "step": 7008 }, { "epoch": 0.24971052959723533, "grad_norm": 1.5695305760260663, "learning_rate": 1.7573719326381455e-05, "loss": 0.571, "step": 7009 }, { "epoch": 0.24974615672372943, "grad_norm": 1.2367799040021108, "learning_rate": 1.757296577406727e-05, "loss": 0.5026, "step": 7010 }, { "epoch": 0.24978178385022357, "grad_norm": 1.1242250972044212, "learning_rate": 1.757221212091375e-05, "loss": 0.4887, "step": 7011 }, { "epoch": 0.24981741097671767, "grad_norm": 1.4822459738889002, "learning_rate": 1.7571458366930937e-05, "loss": 0.548, "step": 7012 }, { "epoch": 0.24985303810321177, "grad_norm": 2.095667043611195, "learning_rate": 1.7570704512128865e-05, "loss": 0.6583, "step": 7013 }, { "epoch": 0.2498886652297059, "grad_norm": 1.282273330419582, "learning_rate": 1.7569950556517566e-05, "loss": 0.6459, "step": 7014 }, { "epoch": 0.2499242923562, "grad_norm": 1.2032964664411874, "learning_rate": 1.7569196500107087e-05, "loss": 0.511, "step": 7015 }, { "epoch": 0.2499599194826941, "grad_norm": 1.6941483836889406, "learning_rate": 1.7568442342907463e-05, "loss": 0.6511, "step": 7016 }, { "epoch": 0.24999554660918824, "grad_norm": 1.4638720679968373, "learning_rate": 1.7567688084928743e-05, "loss": 0.6721, "step": 7017 }, { "epoch": 0.25003117373568234, "grad_norm": 1.5995632601952843, "learning_rate": 1.7566933726180963e-05, "loss": 0.5852, "step": 7018 }, { "epoch": 0.2500668008621765, "grad_norm": 1.8884168619101611, "learning_rate": 1.7566179266674175e-05, "loss": 0.5286, "step": 7019 }, { "epoch": 0.25010242798867055, "grad_norm": 1.3048457168204535, "learning_rate": 1.7565424706418422e-05, "loss": 0.4996, "step": 7020 }, { "epoch": 0.2501380551151647, "grad_norm": 1.574861372249137, "learning_rate": 1.7564670045423747e-05, "loss": 0.6564, "step": 7021 }, { "epoch": 0.2501736822416588, "grad_norm": 1.4265789677480738, "learning_rate": 1.7563915283700205e-05, "loss": 0.5863, "step": 7022 }, { "epoch": 0.2502093093681529, "grad_norm": 1.1156340010533237, "learning_rate": 1.7563160421257845e-05, "loss": 0.5261, "step": 7023 }, { "epoch": 0.250244936494647, "grad_norm": 1.2881897695548952, "learning_rate": 1.7562405458106718e-05, "loss": 0.5162, "step": 7024 }, { "epoch": 0.25028056362114115, "grad_norm": 1.2612608713097486, "learning_rate": 1.7561650394256874e-05, "loss": 0.5056, "step": 7025 }, { "epoch": 0.2503161907476352, "grad_norm": 1.739388217416117, "learning_rate": 1.7560895229718374e-05, "loss": 0.6119, "step": 7026 }, { "epoch": 0.25035181787412936, "grad_norm": 1.1264873862572726, "learning_rate": 1.7560139964501263e-05, "loss": 0.5397, "step": 7027 }, { "epoch": 0.2503874450006235, "grad_norm": 6.872212052295991, "learning_rate": 1.755938459861561e-05, "loss": 0.654, "step": 7028 }, { "epoch": 0.25042307212711756, "grad_norm": 1.2807759283999018, "learning_rate": 1.7558629132071465e-05, "loss": 0.5563, "step": 7029 }, { "epoch": 0.2504586992536117, "grad_norm": 1.889701081815432, "learning_rate": 1.755787356487889e-05, "loss": 0.6345, "step": 7030 }, { "epoch": 0.2504943263801058, "grad_norm": 1.8945812570665572, "learning_rate": 1.7557117897047944e-05, "loss": 0.6132, "step": 7031 }, { "epoch": 0.2505299535065999, "grad_norm": 1.1557872663374813, "learning_rate": 1.7556362128588695e-05, "loss": 0.5251, "step": 7032 }, { "epoch": 0.25056558063309403, "grad_norm": 2.358507555875459, "learning_rate": 1.75556062595112e-05, "loss": 0.4921, "step": 7033 }, { "epoch": 0.25060120775958816, "grad_norm": 1.575164929138862, "learning_rate": 1.7554850289825527e-05, "loss": 0.5564, "step": 7034 }, { "epoch": 0.25063683488608224, "grad_norm": 1.144128438991776, "learning_rate": 1.7554094219541743e-05, "loss": 0.5463, "step": 7035 }, { "epoch": 0.25067246201257637, "grad_norm": 1.5632310523171322, "learning_rate": 1.755333804866991e-05, "loss": 0.5681, "step": 7036 }, { "epoch": 0.2507080891390705, "grad_norm": 1.2541389712551962, "learning_rate": 1.755258177722011e-05, "loss": 0.4535, "step": 7037 }, { "epoch": 0.2507437162655646, "grad_norm": 1.2574836390553334, "learning_rate": 1.7551825405202397e-05, "loss": 0.4944, "step": 7038 }, { "epoch": 0.2507793433920587, "grad_norm": 1.1873049430822167, "learning_rate": 1.7551068932626854e-05, "loss": 0.5354, "step": 7039 }, { "epoch": 0.25081497051855284, "grad_norm": 1.9226955991592227, "learning_rate": 1.7550312359503547e-05, "loss": 0.6426, "step": 7040 }, { "epoch": 0.2508505976450469, "grad_norm": 6.682801038289492, "learning_rate": 1.7549555685842556e-05, "loss": 0.6352, "step": 7041 }, { "epoch": 0.25088622477154104, "grad_norm": 1.249379601425778, "learning_rate": 1.7548798911653956e-05, "loss": 0.5254, "step": 7042 }, { "epoch": 0.2509218518980352, "grad_norm": 1.1829864115420956, "learning_rate": 1.754804203694782e-05, "loss": 0.4429, "step": 7043 }, { "epoch": 0.25095747902452925, "grad_norm": 1.965125256260391, "learning_rate": 1.7547285061734227e-05, "loss": 0.5909, "step": 7044 }, { "epoch": 0.2509931061510234, "grad_norm": 2.0192013737092434, "learning_rate": 1.754652798602326e-05, "loss": 0.6182, "step": 7045 }, { "epoch": 0.2510287332775175, "grad_norm": 2.01027780190777, "learning_rate": 1.7545770809824997e-05, "loss": 0.6179, "step": 7046 }, { "epoch": 0.2510643604040116, "grad_norm": 1.416272625931895, "learning_rate": 1.7545013533149525e-05, "loss": 0.4817, "step": 7047 }, { "epoch": 0.2510999875305057, "grad_norm": 1.1181794225461736, "learning_rate": 1.754425615600692e-05, "loss": 0.562, "step": 7048 }, { "epoch": 0.25113561465699985, "grad_norm": 1.1776037191998314, "learning_rate": 1.7543498678407275e-05, "loss": 0.5448, "step": 7049 }, { "epoch": 0.2511712417834939, "grad_norm": 1.0925937558216026, "learning_rate": 1.754274110036067e-05, "loss": 0.4649, "step": 7050 }, { "epoch": 0.25120686890998806, "grad_norm": 1.9017601749133808, "learning_rate": 1.7541983421877196e-05, "loss": 0.6125, "step": 7051 }, { "epoch": 0.2512424960364822, "grad_norm": 1.3493396889144575, "learning_rate": 1.754122564296694e-05, "loss": 0.4842, "step": 7052 }, { "epoch": 0.25127812316297626, "grad_norm": 1.1619778023104195, "learning_rate": 1.7540467763639994e-05, "loss": 0.5149, "step": 7053 }, { "epoch": 0.2513137502894704, "grad_norm": 1.9571185722865163, "learning_rate": 1.753970978390645e-05, "loss": 0.5414, "step": 7054 }, { "epoch": 0.2513493774159645, "grad_norm": 1.147111355766433, "learning_rate": 1.75389517037764e-05, "loss": 0.4997, "step": 7055 }, { "epoch": 0.2513850045424586, "grad_norm": 1.2520952980582096, "learning_rate": 1.7538193523259938e-05, "loss": 0.503, "step": 7056 }, { "epoch": 0.25142063166895273, "grad_norm": 1.591622494135217, "learning_rate": 1.753743524236716e-05, "loss": 0.5651, "step": 7057 }, { "epoch": 0.25145625879544686, "grad_norm": 1.120975315743805, "learning_rate": 1.7536676861108167e-05, "loss": 0.4686, "step": 7058 }, { "epoch": 0.25149188592194094, "grad_norm": 1.9192744765161471, "learning_rate": 1.753591837949305e-05, "loss": 0.6687, "step": 7059 }, { "epoch": 0.25152751304843507, "grad_norm": 1.9920870938790392, "learning_rate": 1.7535159797531914e-05, "loss": 0.4968, "step": 7060 }, { "epoch": 0.2515631401749292, "grad_norm": 1.1895671602986624, "learning_rate": 1.7534401115234857e-05, "loss": 0.5612, "step": 7061 }, { "epoch": 0.2515987673014233, "grad_norm": 1.8807479973103884, "learning_rate": 1.7533642332611986e-05, "loss": 0.573, "step": 7062 }, { "epoch": 0.2516343944279174, "grad_norm": 1.576119073254555, "learning_rate": 1.7532883449673396e-05, "loss": 0.6417, "step": 7063 }, { "epoch": 0.25167002155441154, "grad_norm": 1.1516911422636946, "learning_rate": 1.7532124466429202e-05, "loss": 0.4875, "step": 7064 }, { "epoch": 0.25170564868090567, "grad_norm": 2.0541331623222816, "learning_rate": 1.7531365382889508e-05, "loss": 0.6123, "step": 7065 }, { "epoch": 0.25174127580739974, "grad_norm": 1.1872810637479392, "learning_rate": 1.7530606199064416e-05, "loss": 0.5112, "step": 7066 }, { "epoch": 0.2517769029338939, "grad_norm": 1.2201292676716633, "learning_rate": 1.752984691496404e-05, "loss": 0.487, "step": 7067 }, { "epoch": 0.251812530060388, "grad_norm": 1.1478995334090123, "learning_rate": 1.752908753059849e-05, "loss": 0.4781, "step": 7068 }, { "epoch": 0.2518481571868821, "grad_norm": 1.1402232113439523, "learning_rate": 1.7528328045977876e-05, "loss": 0.4395, "step": 7069 }, { "epoch": 0.2518837843133762, "grad_norm": 2.173403887709897, "learning_rate": 1.7527568461112313e-05, "loss": 0.681, "step": 7070 }, { "epoch": 0.25191941143987034, "grad_norm": 1.1052701566786822, "learning_rate": 1.7526808776011913e-05, "loss": 0.499, "step": 7071 }, { "epoch": 0.2519550385663644, "grad_norm": 1.6480360817123878, "learning_rate": 1.7526048990686793e-05, "loss": 0.5466, "step": 7072 }, { "epoch": 0.25199066569285855, "grad_norm": 1.669361166155445, "learning_rate": 1.752528910514707e-05, "loss": 0.5599, "step": 7073 }, { "epoch": 0.2520262928193527, "grad_norm": 1.1200561400693145, "learning_rate": 1.7524529119402867e-05, "loss": 0.5078, "step": 7074 }, { "epoch": 0.25206191994584676, "grad_norm": 1.2990044418913709, "learning_rate": 1.7523769033464298e-05, "loss": 0.5146, "step": 7075 }, { "epoch": 0.2520975470723409, "grad_norm": 1.469800012013257, "learning_rate": 1.7523008847341484e-05, "loss": 0.6166, "step": 7076 }, { "epoch": 0.252133174198835, "grad_norm": 1.0782610463502587, "learning_rate": 1.752224856104455e-05, "loss": 0.4894, "step": 7077 }, { "epoch": 0.2521688013253291, "grad_norm": 1.0390898194696467, "learning_rate": 1.7521488174583616e-05, "loss": 0.4691, "step": 7078 }, { "epoch": 0.2522044284518232, "grad_norm": 1.4887426677316171, "learning_rate": 1.752072768796881e-05, "loss": 0.6016, "step": 7079 }, { "epoch": 0.25224005557831736, "grad_norm": 1.1633443082851573, "learning_rate": 1.7519967101210264e-05, "loss": 0.4931, "step": 7080 }, { "epoch": 0.25227568270481143, "grad_norm": 1.2138827919322164, "learning_rate": 1.7519206414318092e-05, "loss": 0.5358, "step": 7081 }, { "epoch": 0.25231130983130556, "grad_norm": 1.6771731782867805, "learning_rate": 1.7518445627302438e-05, "loss": 0.5915, "step": 7082 }, { "epoch": 0.2523469369577997, "grad_norm": 1.4163174597228032, "learning_rate": 1.7517684740173418e-05, "loss": 0.5092, "step": 7083 }, { "epoch": 0.25238256408429377, "grad_norm": 1.1727304288101463, "learning_rate": 1.7516923752941174e-05, "loss": 0.4741, "step": 7084 }, { "epoch": 0.2524181912107879, "grad_norm": 1.4528512147558734, "learning_rate": 1.751616266561584e-05, "loss": 0.6024, "step": 7085 }, { "epoch": 0.25245381833728203, "grad_norm": 1.3791403203516301, "learning_rate": 1.7515401478207542e-05, "loss": 0.6148, "step": 7086 }, { "epoch": 0.2524894454637761, "grad_norm": 1.8508454073880554, "learning_rate": 1.7514640190726425e-05, "loss": 0.6203, "step": 7087 }, { "epoch": 0.25252507259027024, "grad_norm": 1.5715589524996914, "learning_rate": 1.751387880318262e-05, "loss": 0.632, "step": 7088 }, { "epoch": 0.25256069971676437, "grad_norm": 1.6592686916375516, "learning_rate": 1.7513117315586265e-05, "loss": 0.6255, "step": 7089 }, { "epoch": 0.25259632684325845, "grad_norm": 1.133889959994574, "learning_rate": 1.75123557279475e-05, "loss": 0.4977, "step": 7090 }, { "epoch": 0.2526319539697526, "grad_norm": 1.1946266983116671, "learning_rate": 1.7511594040276468e-05, "loss": 0.4779, "step": 7091 }, { "epoch": 0.2526675810962467, "grad_norm": 2.3081879853842966, "learning_rate": 1.7510832252583317e-05, "loss": 0.6581, "step": 7092 }, { "epoch": 0.2527032082227408, "grad_norm": 1.2148514023536825, "learning_rate": 1.7510070364878177e-05, "loss": 0.5335, "step": 7093 }, { "epoch": 0.2527388353492349, "grad_norm": 1.7831377510280515, "learning_rate": 1.7509308377171205e-05, "loss": 0.5544, "step": 7094 }, { "epoch": 0.25277446247572904, "grad_norm": 1.8713314622310293, "learning_rate": 1.7508546289472543e-05, "loss": 0.6474, "step": 7095 }, { "epoch": 0.2528100896022231, "grad_norm": 1.5426827630364215, "learning_rate": 1.7507784101792342e-05, "loss": 0.6398, "step": 7096 }, { "epoch": 0.25284571672871725, "grad_norm": 1.1952369920161299, "learning_rate": 1.7507021814140745e-05, "loss": 0.4832, "step": 7097 }, { "epoch": 0.2528813438552114, "grad_norm": 1.2902689563791592, "learning_rate": 1.7506259426527903e-05, "loss": 0.4708, "step": 7098 }, { "epoch": 0.25291697098170546, "grad_norm": 1.4873147255443613, "learning_rate": 1.7505496938963974e-05, "loss": 0.5, "step": 7099 }, { "epoch": 0.2529525981081996, "grad_norm": 1.2032750861226793, "learning_rate": 1.7504734351459105e-05, "loss": 0.5315, "step": 7100 }, { "epoch": 0.2529882252346937, "grad_norm": 1.3747349726103708, "learning_rate": 1.7503971664023454e-05, "loss": 0.5563, "step": 7101 }, { "epoch": 0.2530238523611878, "grad_norm": 1.2401369672248286, "learning_rate": 1.7503208876667175e-05, "loss": 0.4868, "step": 7102 }, { "epoch": 0.2530594794876819, "grad_norm": 1.114387641937704, "learning_rate": 1.7502445989400426e-05, "loss": 0.4616, "step": 7103 }, { "epoch": 0.25309510661417606, "grad_norm": 1.1721566577662172, "learning_rate": 1.750168300223336e-05, "loss": 0.5184, "step": 7104 }, { "epoch": 0.25313073374067013, "grad_norm": 1.0961320919987474, "learning_rate": 1.7500919915176147e-05, "loss": 0.4642, "step": 7105 }, { "epoch": 0.25316636086716426, "grad_norm": 1.1864465997050866, "learning_rate": 1.750015672823894e-05, "loss": 0.4723, "step": 7106 }, { "epoch": 0.2532019879936584, "grad_norm": 1.427356987463339, "learning_rate": 1.7499393441431903e-05, "loss": 0.664, "step": 7107 }, { "epoch": 0.25323761512015247, "grad_norm": 1.3902242471800539, "learning_rate": 1.7498630054765204e-05, "loss": 0.6055, "step": 7108 }, { "epoch": 0.2532732422466466, "grad_norm": 1.591967643725237, "learning_rate": 1.7497866568249e-05, "loss": 0.5218, "step": 7109 }, { "epoch": 0.25330886937314073, "grad_norm": 1.4893396864353678, "learning_rate": 1.749710298189347e-05, "loss": 0.5302, "step": 7110 }, { "epoch": 0.2533444964996348, "grad_norm": 1.1486465036601006, "learning_rate": 1.7496339295708767e-05, "loss": 0.5026, "step": 7111 }, { "epoch": 0.25338012362612894, "grad_norm": 1.2055148497087536, "learning_rate": 1.749557550970507e-05, "loss": 0.5361, "step": 7112 }, { "epoch": 0.25341575075262307, "grad_norm": 1.2428549006008045, "learning_rate": 1.7494811623892543e-05, "loss": 0.4936, "step": 7113 }, { "epoch": 0.25345137787911715, "grad_norm": 1.9141143293265175, "learning_rate": 1.7494047638281363e-05, "loss": 0.5758, "step": 7114 }, { "epoch": 0.2534870050056113, "grad_norm": 1.7552304839687354, "learning_rate": 1.7493283552881696e-05, "loss": 0.6365, "step": 7115 }, { "epoch": 0.2535226321321054, "grad_norm": 1.2404769720488829, "learning_rate": 1.7492519367703725e-05, "loss": 0.4853, "step": 7116 }, { "epoch": 0.2535582592585995, "grad_norm": 1.7709505244202104, "learning_rate": 1.749175508275762e-05, "loss": 0.591, "step": 7117 }, { "epoch": 0.2535938863850936, "grad_norm": 1.119690490489698, "learning_rate": 1.7490990698053562e-05, "loss": 0.5382, "step": 7118 }, { "epoch": 0.25362951351158775, "grad_norm": 1.524304385965376, "learning_rate": 1.7490226213601725e-05, "loss": 0.5759, "step": 7119 }, { "epoch": 0.2536651406380818, "grad_norm": 1.8901215515096574, "learning_rate": 1.7489461629412293e-05, "loss": 0.5164, "step": 7120 }, { "epoch": 0.25370076776457595, "grad_norm": 1.223209210916301, "learning_rate": 1.748869694549544e-05, "loss": 0.5464, "step": 7121 }, { "epoch": 0.2537363948910701, "grad_norm": 2.1359020141900698, "learning_rate": 1.748793216186136e-05, "loss": 0.5664, "step": 7122 }, { "epoch": 0.25377202201756416, "grad_norm": 1.8491207615872856, "learning_rate": 1.7487167278520222e-05, "loss": 0.6039, "step": 7123 }, { "epoch": 0.2538076491440583, "grad_norm": 1.1106529650588977, "learning_rate": 1.7486402295482224e-05, "loss": 0.5148, "step": 7124 }, { "epoch": 0.2538432762705524, "grad_norm": 1.409012989721411, "learning_rate": 1.748563721275754e-05, "loss": 0.5954, "step": 7125 }, { "epoch": 0.2538789033970465, "grad_norm": 1.0898625832334576, "learning_rate": 1.7484872030356373e-05, "loss": 0.5419, "step": 7126 }, { "epoch": 0.2539145305235406, "grad_norm": 1.226518631867718, "learning_rate": 1.74841067482889e-05, "loss": 0.5783, "step": 7127 }, { "epoch": 0.25395015765003476, "grad_norm": 1.2830500225876988, "learning_rate": 1.748334136656531e-05, "loss": 0.5347, "step": 7128 }, { "epoch": 0.25398578477652883, "grad_norm": 1.4125460976211102, "learning_rate": 1.7482575885195806e-05, "loss": 0.6216, "step": 7129 }, { "epoch": 0.25402141190302296, "grad_norm": 1.7049622699031224, "learning_rate": 1.7481810304190568e-05, "loss": 0.5921, "step": 7130 }, { "epoch": 0.2540570390295171, "grad_norm": 1.223360985958, "learning_rate": 1.74810446235598e-05, "loss": 0.562, "step": 7131 }, { "epoch": 0.25409266615601117, "grad_norm": 1.1671795879491045, "learning_rate": 1.7480278843313692e-05, "loss": 0.4885, "step": 7132 }, { "epoch": 0.2541282932825053, "grad_norm": 1.1167741926023225, "learning_rate": 1.7479512963462448e-05, "loss": 0.5431, "step": 7133 }, { "epoch": 0.25416392040899943, "grad_norm": 1.8856396269146827, "learning_rate": 1.7478746984016256e-05, "loss": 0.6794, "step": 7134 }, { "epoch": 0.2541995475354935, "grad_norm": 1.5517573671981038, "learning_rate": 1.747798090498532e-05, "loss": 0.6082, "step": 7135 }, { "epoch": 0.25423517466198764, "grad_norm": 1.168387215190993, "learning_rate": 1.7477214726379844e-05, "loss": 0.507, "step": 7136 }, { "epoch": 0.25427080178848177, "grad_norm": 1.1698283145659996, "learning_rate": 1.7476448448210027e-05, "loss": 0.5395, "step": 7137 }, { "epoch": 0.25430642891497585, "grad_norm": 3.0945237844348834, "learning_rate": 1.747568207048607e-05, "loss": 0.6469, "step": 7138 }, { "epoch": 0.25434205604147, "grad_norm": 1.3737593586468904, "learning_rate": 1.7474915593218185e-05, "loss": 0.5645, "step": 7139 }, { "epoch": 0.2543776831679641, "grad_norm": 2.936957096566085, "learning_rate": 1.7474149016416576e-05, "loss": 0.6179, "step": 7140 }, { "epoch": 0.2544133102944582, "grad_norm": 1.7123010598069988, "learning_rate": 1.747338234009144e-05, "loss": 0.6702, "step": 7141 }, { "epoch": 0.2544489374209523, "grad_norm": 1.2168145654616989, "learning_rate": 1.7472615564253004e-05, "loss": 0.4934, "step": 7142 }, { "epoch": 0.25448456454744645, "grad_norm": 1.2900538681064069, "learning_rate": 1.7471848688911465e-05, "loss": 0.4937, "step": 7143 }, { "epoch": 0.2545201916739405, "grad_norm": 1.2761567904481417, "learning_rate": 1.747108171407704e-05, "loss": 0.5109, "step": 7144 }, { "epoch": 0.25455581880043465, "grad_norm": 1.4601593302176508, "learning_rate": 1.7470314639759936e-05, "loss": 0.5378, "step": 7145 }, { "epoch": 0.2545914459269288, "grad_norm": 1.4278013763334791, "learning_rate": 1.7469547465970373e-05, "loss": 0.5758, "step": 7146 }, { "epoch": 0.25462707305342286, "grad_norm": 1.1497818951348575, "learning_rate": 1.7468780192718566e-05, "loss": 0.4428, "step": 7147 }, { "epoch": 0.254662700179917, "grad_norm": 2.6482692336123574, "learning_rate": 1.746801282001473e-05, "loss": 0.6069, "step": 7148 }, { "epoch": 0.2546983273064111, "grad_norm": 1.2839197980081274, "learning_rate": 1.7467245347869083e-05, "loss": 0.5442, "step": 7149 }, { "epoch": 0.2547339544329052, "grad_norm": 1.13942718258029, "learning_rate": 1.7466477776291845e-05, "loss": 0.5167, "step": 7150 }, { "epoch": 0.2547695815593993, "grad_norm": 1.223944495541115, "learning_rate": 1.7465710105293238e-05, "loss": 0.5178, "step": 7151 }, { "epoch": 0.25480520868589346, "grad_norm": 1.8274196144620438, "learning_rate": 1.746494233488348e-05, "loss": 0.5588, "step": 7152 }, { "epoch": 0.25484083581238753, "grad_norm": 1.297763421822816, "learning_rate": 1.7464174465072798e-05, "loss": 0.5547, "step": 7153 }, { "epoch": 0.25487646293888166, "grad_norm": 1.4932994652755707, "learning_rate": 1.7463406495871416e-05, "loss": 0.5831, "step": 7154 }, { "epoch": 0.2549120900653758, "grad_norm": 1.3161912572724483, "learning_rate": 1.746263842728956e-05, "loss": 0.5144, "step": 7155 }, { "epoch": 0.25494771719186987, "grad_norm": 1.777443766759931, "learning_rate": 1.7461870259337457e-05, "loss": 0.593, "step": 7156 }, { "epoch": 0.254983344318364, "grad_norm": 1.8330097727041712, "learning_rate": 1.7461101992025334e-05, "loss": 0.5549, "step": 7157 }, { "epoch": 0.25501897144485813, "grad_norm": 1.1808667529084615, "learning_rate": 1.746033362536343e-05, "loss": 0.5028, "step": 7158 }, { "epoch": 0.2550545985713522, "grad_norm": 1.1770344539930784, "learning_rate": 1.7459565159361963e-05, "loss": 0.5068, "step": 7159 }, { "epoch": 0.25509022569784634, "grad_norm": 1.315745557082713, "learning_rate": 1.745879659403117e-05, "loss": 0.4946, "step": 7160 }, { "epoch": 0.25512585282434047, "grad_norm": 1.572333045157177, "learning_rate": 1.7458027929381292e-05, "loss": 0.6038, "step": 7161 }, { "epoch": 0.25516147995083455, "grad_norm": 1.1653034483175777, "learning_rate": 1.7457259165422557e-05, "loss": 0.5162, "step": 7162 }, { "epoch": 0.2551971070773287, "grad_norm": 1.7085890397942294, "learning_rate": 1.7456490302165208e-05, "loss": 0.621, "step": 7163 }, { "epoch": 0.2552327342038228, "grad_norm": 1.1947795814507893, "learning_rate": 1.7455721339619477e-05, "loss": 0.5549, "step": 7164 }, { "epoch": 0.2552683613303169, "grad_norm": 1.1028457572931785, "learning_rate": 1.7454952277795602e-05, "loss": 0.5054, "step": 7165 }, { "epoch": 0.255303988456811, "grad_norm": 2.365492076759505, "learning_rate": 1.745418311670383e-05, "loss": 0.5955, "step": 7166 }, { "epoch": 0.25533961558330515, "grad_norm": 1.2140809903259957, "learning_rate": 1.7453413856354396e-05, "loss": 0.5159, "step": 7167 }, { "epoch": 0.2553752427097992, "grad_norm": 1.2271809075682742, "learning_rate": 1.745264449675755e-05, "loss": 0.4921, "step": 7168 }, { "epoch": 0.25541086983629335, "grad_norm": 2.049507859405456, "learning_rate": 1.745187503792353e-05, "loss": 0.5223, "step": 7169 }, { "epoch": 0.2554464969627875, "grad_norm": 2.282094753966084, "learning_rate": 1.745110547986259e-05, "loss": 0.685, "step": 7170 }, { "epoch": 0.25548212408928156, "grad_norm": 1.3029691151481064, "learning_rate": 1.7450335822584968e-05, "loss": 0.5511, "step": 7171 }, { "epoch": 0.2555177512157757, "grad_norm": 2.1557909001158513, "learning_rate": 1.744956606610092e-05, "loss": 0.591, "step": 7172 }, { "epoch": 0.2555533783422698, "grad_norm": 1.7403868889446812, "learning_rate": 1.744879621042069e-05, "loss": 0.62, "step": 7173 }, { "epoch": 0.2555890054687639, "grad_norm": 1.667767614703634, "learning_rate": 1.7448026255554537e-05, "loss": 0.548, "step": 7174 }, { "epoch": 0.25562463259525803, "grad_norm": 1.5195980804937634, "learning_rate": 1.7447256201512707e-05, "loss": 0.5464, "step": 7175 }, { "epoch": 0.25566025972175216, "grad_norm": 1.33849671370866, "learning_rate": 1.7446486048305454e-05, "loss": 0.5039, "step": 7176 }, { "epoch": 0.25569588684824623, "grad_norm": 1.562009097090772, "learning_rate": 1.7445715795943033e-05, "loss": 0.6064, "step": 7177 }, { "epoch": 0.25573151397474037, "grad_norm": 1.179659087493634, "learning_rate": 1.7444945444435703e-05, "loss": 0.5004, "step": 7178 }, { "epoch": 0.2557671411012345, "grad_norm": 1.8219632506329864, "learning_rate": 1.744417499379372e-05, "loss": 0.5581, "step": 7179 }, { "epoch": 0.25580276822772857, "grad_norm": 1.1382550313844313, "learning_rate": 1.7443404444027347e-05, "loss": 0.5235, "step": 7180 }, { "epoch": 0.2558383953542227, "grad_norm": 1.060518650020486, "learning_rate": 1.7442633795146838e-05, "loss": 0.5033, "step": 7181 }, { "epoch": 0.25587402248071683, "grad_norm": 1.2795197903603743, "learning_rate": 1.744186304716246e-05, "loss": 0.5625, "step": 7182 }, { "epoch": 0.2559096496072109, "grad_norm": 1.1376084219317333, "learning_rate": 1.7441092200084472e-05, "loss": 0.5212, "step": 7183 }, { "epoch": 0.25594527673370504, "grad_norm": 1.7791926639122897, "learning_rate": 1.744032125392314e-05, "loss": 0.6196, "step": 7184 }, { "epoch": 0.25598090386019917, "grad_norm": 1.5331268129741957, "learning_rate": 1.7439550208688737e-05, "loss": 0.603, "step": 7185 }, { "epoch": 0.25601653098669325, "grad_norm": 1.2289277193093722, "learning_rate": 1.7438779064391515e-05, "loss": 0.4767, "step": 7186 }, { "epoch": 0.2560521581131874, "grad_norm": 2.113474301871056, "learning_rate": 1.7438007821041752e-05, "loss": 0.7603, "step": 7187 }, { "epoch": 0.2560877852396815, "grad_norm": 1.0020257602328502, "learning_rate": 1.7437236478649718e-05, "loss": 0.4813, "step": 7188 }, { "epoch": 0.2561234123661756, "grad_norm": 1.1417680610887668, "learning_rate": 1.743646503722568e-05, "loss": 0.4977, "step": 7189 }, { "epoch": 0.2561590394926697, "grad_norm": 1.8953984705396119, "learning_rate": 1.743569349677991e-05, "loss": 0.6075, "step": 7190 }, { "epoch": 0.25619466661916385, "grad_norm": 1.745288668003622, "learning_rate": 1.7434921857322692e-05, "loss": 0.5234, "step": 7191 }, { "epoch": 0.2562302937456579, "grad_norm": 1.104686044998859, "learning_rate": 1.7434150118864286e-05, "loss": 0.501, "step": 7192 }, { "epoch": 0.25626592087215205, "grad_norm": 1.155245181490845, "learning_rate": 1.7433378281414975e-05, "loss": 0.4829, "step": 7193 }, { "epoch": 0.2563015479986462, "grad_norm": 1.771911017317097, "learning_rate": 1.743260634498504e-05, "loss": 0.5685, "step": 7194 }, { "epoch": 0.25633717512514026, "grad_norm": 1.1503568200853236, "learning_rate": 1.7431834309584753e-05, "loss": 0.5424, "step": 7195 }, { "epoch": 0.2563728022516344, "grad_norm": 1.5364046910673324, "learning_rate": 1.7431062175224404e-05, "loss": 0.6486, "step": 7196 }, { "epoch": 0.2564084293781285, "grad_norm": 1.7710030938533343, "learning_rate": 1.743028994191426e-05, "loss": 0.679, "step": 7197 }, { "epoch": 0.2564440565046226, "grad_norm": 1.6998223413549087, "learning_rate": 1.7429517609664618e-05, "loss": 0.5396, "step": 7198 }, { "epoch": 0.25647968363111673, "grad_norm": 1.1325964263085584, "learning_rate": 1.7428745178485754e-05, "loss": 0.5343, "step": 7199 }, { "epoch": 0.25651531075761086, "grad_norm": 1.9231860123969322, "learning_rate": 1.7427972648387953e-05, "loss": 0.602, "step": 7200 }, { "epoch": 0.25655093788410493, "grad_norm": 1.6436104347107088, "learning_rate": 1.7427200019381504e-05, "loss": 0.61, "step": 7201 }, { "epoch": 0.25658656501059907, "grad_norm": 1.307358721090924, "learning_rate": 1.74264272914767e-05, "loss": 0.5141, "step": 7202 }, { "epoch": 0.2566221921370932, "grad_norm": 1.498781608421995, "learning_rate": 1.7425654464683822e-05, "loss": 0.6523, "step": 7203 }, { "epoch": 0.2566578192635873, "grad_norm": 1.2305210336753842, "learning_rate": 1.742488153901317e-05, "loss": 0.5169, "step": 7204 }, { "epoch": 0.2566934463900814, "grad_norm": 1.074740617546368, "learning_rate": 1.7424108514475025e-05, "loss": 0.521, "step": 7205 }, { "epoch": 0.25672907351657553, "grad_norm": 1.6584156879304914, "learning_rate": 1.742333539107969e-05, "loss": 0.6146, "step": 7206 }, { "epoch": 0.2567647006430696, "grad_norm": 1.4372297798421592, "learning_rate": 1.7422562168837452e-05, "loss": 0.5, "step": 7207 }, { "epoch": 0.25680032776956374, "grad_norm": 1.5201531396679002, "learning_rate": 1.742178884775861e-05, "loss": 0.5966, "step": 7208 }, { "epoch": 0.25683595489605787, "grad_norm": 1.709557150012629, "learning_rate": 1.7421015427853468e-05, "loss": 0.5904, "step": 7209 }, { "epoch": 0.25687158202255195, "grad_norm": 1.1544347997020687, "learning_rate": 1.7420241909132312e-05, "loss": 0.5155, "step": 7210 }, { "epoch": 0.2569072091490461, "grad_norm": 1.1686677872915503, "learning_rate": 1.741946829160545e-05, "loss": 0.5335, "step": 7211 }, { "epoch": 0.2569428362755402, "grad_norm": 1.6138831497553674, "learning_rate": 1.7418694575283185e-05, "loss": 0.6105, "step": 7212 }, { "epoch": 0.2569784634020343, "grad_norm": 1.064326063421843, "learning_rate": 1.7417920760175817e-05, "loss": 0.4905, "step": 7213 }, { "epoch": 0.2570140905285284, "grad_norm": 1.3585678656464986, "learning_rate": 1.7417146846293646e-05, "loss": 0.6206, "step": 7214 }, { "epoch": 0.25704971765502255, "grad_norm": 1.258562547208527, "learning_rate": 1.741637283364698e-05, "loss": 0.5336, "step": 7215 }, { "epoch": 0.2570853447815166, "grad_norm": 1.8220685364358815, "learning_rate": 1.7415598722246127e-05, "loss": 0.5737, "step": 7216 }, { "epoch": 0.25712097190801075, "grad_norm": 1.9128317756218034, "learning_rate": 1.7414824512101395e-05, "loss": 0.5902, "step": 7217 }, { "epoch": 0.2571565990345049, "grad_norm": 1.659652486234925, "learning_rate": 1.7414050203223092e-05, "loss": 0.6241, "step": 7218 }, { "epoch": 0.25719222616099896, "grad_norm": 1.072558982634365, "learning_rate": 1.7413275795621527e-05, "loss": 0.4718, "step": 7219 }, { "epoch": 0.2572278532874931, "grad_norm": 1.5931978414348822, "learning_rate": 1.7412501289307013e-05, "loss": 0.596, "step": 7220 }, { "epoch": 0.2572634804139872, "grad_norm": 2.938317091464382, "learning_rate": 1.7411726684289863e-05, "loss": 0.5892, "step": 7221 }, { "epoch": 0.2572991075404813, "grad_norm": 1.1414095914112623, "learning_rate": 1.7410951980580396e-05, "loss": 0.4482, "step": 7222 }, { "epoch": 0.25733473466697543, "grad_norm": 1.3200173843649936, "learning_rate": 1.7410177178188917e-05, "loss": 0.5001, "step": 7223 }, { "epoch": 0.25737036179346956, "grad_norm": 1.199767823568117, "learning_rate": 1.7409402277125753e-05, "loss": 0.5402, "step": 7224 }, { "epoch": 0.25740598891996364, "grad_norm": 1.6064709829526627, "learning_rate": 1.740862727740122e-05, "loss": 0.6327, "step": 7225 }, { "epoch": 0.25744161604645777, "grad_norm": 1.6863781883363567, "learning_rate": 1.7407852179025636e-05, "loss": 0.5615, "step": 7226 }, { "epoch": 0.2574772431729519, "grad_norm": 2.222566008740006, "learning_rate": 1.740707698200932e-05, "loss": 0.562, "step": 7227 }, { "epoch": 0.257512870299446, "grad_norm": 1.1642977861442492, "learning_rate": 1.74063016863626e-05, "loss": 0.4735, "step": 7228 }, { "epoch": 0.2575484974259401, "grad_norm": 1.223182903925115, "learning_rate": 1.7405526292095794e-05, "loss": 0.5715, "step": 7229 }, { "epoch": 0.25758412455243423, "grad_norm": 1.840160684431836, "learning_rate": 1.7404750799219228e-05, "loss": 0.6186, "step": 7230 }, { "epoch": 0.2576197516789283, "grad_norm": 1.6691496186993717, "learning_rate": 1.7403975207743233e-05, "loss": 0.6213, "step": 7231 }, { "epoch": 0.25765537880542244, "grad_norm": 1.7072247868594945, "learning_rate": 1.740319951767813e-05, "loss": 0.5018, "step": 7232 }, { "epoch": 0.2576910059319166, "grad_norm": 1.3807836735914822, "learning_rate": 1.7402423729034252e-05, "loss": 0.5415, "step": 7233 }, { "epoch": 0.25772663305841065, "grad_norm": 1.5333487159438426, "learning_rate": 1.7401647841821928e-05, "loss": 0.5922, "step": 7234 }, { "epoch": 0.2577622601849048, "grad_norm": 1.1013279308204311, "learning_rate": 1.7400871856051492e-05, "loss": 0.5273, "step": 7235 }, { "epoch": 0.2577978873113989, "grad_norm": 1.1113344292803968, "learning_rate": 1.7400095771733273e-05, "loss": 0.4775, "step": 7236 }, { "epoch": 0.257833514437893, "grad_norm": 1.5131185428564589, "learning_rate": 1.7399319588877603e-05, "loss": 0.5416, "step": 7237 }, { "epoch": 0.2578691415643871, "grad_norm": 1.5327834959063804, "learning_rate": 1.7398543307494824e-05, "loss": 0.5663, "step": 7238 }, { "epoch": 0.25790476869088125, "grad_norm": 1.6057069442259453, "learning_rate": 1.739776692759527e-05, "loss": 0.6211, "step": 7239 }, { "epoch": 0.2579403958173753, "grad_norm": 1.7104015568926556, "learning_rate": 1.7396990449189275e-05, "loss": 0.6731, "step": 7240 }, { "epoch": 0.25797602294386945, "grad_norm": 1.4004872237061414, "learning_rate": 1.7396213872287187e-05, "loss": 0.5947, "step": 7241 }, { "epoch": 0.2580116500703636, "grad_norm": 1.1054353389550178, "learning_rate": 1.7395437196899338e-05, "loss": 0.5199, "step": 7242 }, { "epoch": 0.25804727719685766, "grad_norm": 1.2834093942546212, "learning_rate": 1.7394660423036076e-05, "loss": 0.4784, "step": 7243 }, { "epoch": 0.2580829043233518, "grad_norm": 1.5801653877912285, "learning_rate": 1.739388355070774e-05, "loss": 0.5907, "step": 7244 }, { "epoch": 0.2581185314498459, "grad_norm": 1.095836423806451, "learning_rate": 1.7393106579924676e-05, "loss": 0.4961, "step": 7245 }, { "epoch": 0.25815415857634, "grad_norm": 1.814947659398449, "learning_rate": 1.7392329510697233e-05, "loss": 0.585, "step": 7246 }, { "epoch": 0.25818978570283413, "grad_norm": 2.193699331312022, "learning_rate": 1.7391552343035753e-05, "loss": 0.5592, "step": 7247 }, { "epoch": 0.25822541282932826, "grad_norm": 1.5072934231062953, "learning_rate": 1.739077507695059e-05, "loss": 0.5082, "step": 7248 }, { "epoch": 0.25826103995582234, "grad_norm": 1.0253730486888906, "learning_rate": 1.738999771245209e-05, "loss": 0.4456, "step": 7249 }, { "epoch": 0.25829666708231647, "grad_norm": 1.8318493914156078, "learning_rate": 1.7389220249550605e-05, "loss": 0.5375, "step": 7250 }, { "epoch": 0.2583322942088106, "grad_norm": 1.4792826226239753, "learning_rate": 1.7388442688256485e-05, "loss": 0.6738, "step": 7251 }, { "epoch": 0.2583679213353047, "grad_norm": 1.1615061806809577, "learning_rate": 1.7387665028580088e-05, "loss": 0.4745, "step": 7252 }, { "epoch": 0.2584035484617988, "grad_norm": 1.5346006752112984, "learning_rate": 1.7386887270531768e-05, "loss": 0.5364, "step": 7253 }, { "epoch": 0.25843917558829294, "grad_norm": 1.3015834428971882, "learning_rate": 1.7386109414121883e-05, "loss": 0.5559, "step": 7254 }, { "epoch": 0.25847480271478707, "grad_norm": 1.7466036662806663, "learning_rate": 1.7385331459360786e-05, "loss": 0.5826, "step": 7255 }, { "epoch": 0.25851042984128114, "grad_norm": 1.0145261198236504, "learning_rate": 1.7384553406258842e-05, "loss": 0.4451, "step": 7256 }, { "epoch": 0.2585460569677753, "grad_norm": 1.520726309345615, "learning_rate": 1.7383775254826402e-05, "loss": 0.5437, "step": 7257 }, { "epoch": 0.2585816840942694, "grad_norm": 1.1537161588792861, "learning_rate": 1.7382997005073837e-05, "loss": 0.4863, "step": 7258 }, { "epoch": 0.2586173112207635, "grad_norm": 1.5647139931391196, "learning_rate": 1.7382218657011507e-05, "loss": 0.6296, "step": 7259 }, { "epoch": 0.2586529383472576, "grad_norm": 1.6244372153775117, "learning_rate": 1.7381440210649774e-05, "loss": 0.5796, "step": 7260 }, { "epoch": 0.25868856547375174, "grad_norm": 1.9581322085593527, "learning_rate": 1.7380661665999006e-05, "loss": 0.5914, "step": 7261 }, { "epoch": 0.2587241926002458, "grad_norm": 1.6430586549927584, "learning_rate": 1.737988302306957e-05, "loss": 0.5372, "step": 7262 }, { "epoch": 0.25875981972673995, "grad_norm": 1.2713355508021114, "learning_rate": 1.737910428187183e-05, "loss": 0.569, "step": 7263 }, { "epoch": 0.2587954468532341, "grad_norm": 1.6943878974577817, "learning_rate": 1.7378325442416165e-05, "loss": 0.6334, "step": 7264 }, { "epoch": 0.25883107397972815, "grad_norm": 1.5236255197529915, "learning_rate": 1.7377546504712933e-05, "loss": 0.5824, "step": 7265 }, { "epoch": 0.2588667011062223, "grad_norm": 1.7421719545801149, "learning_rate": 1.7376767468772518e-05, "loss": 0.6287, "step": 7266 }, { "epoch": 0.2589023282327164, "grad_norm": 1.294242830695332, "learning_rate": 1.7375988334605283e-05, "loss": 0.4603, "step": 7267 }, { "epoch": 0.2589379553592105, "grad_norm": 1.976799629733336, "learning_rate": 1.7375209102221614e-05, "loss": 0.5159, "step": 7268 }, { "epoch": 0.2589735824857046, "grad_norm": 1.149169883129811, "learning_rate": 1.7374429771631877e-05, "loss": 0.5445, "step": 7269 }, { "epoch": 0.25900920961219875, "grad_norm": 1.2162170271504684, "learning_rate": 1.737365034284645e-05, "loss": 0.4972, "step": 7270 }, { "epoch": 0.25904483673869283, "grad_norm": 1.2210680202522892, "learning_rate": 1.7372870815875723e-05, "loss": 0.5212, "step": 7271 }, { "epoch": 0.25908046386518696, "grad_norm": 1.3575414972226985, "learning_rate": 1.7372091190730062e-05, "loss": 0.5313, "step": 7272 }, { "epoch": 0.2591160909916811, "grad_norm": 1.3052034861017026, "learning_rate": 1.7371311467419856e-05, "loss": 0.4563, "step": 7273 }, { "epoch": 0.25915171811817517, "grad_norm": 1.1994422383488204, "learning_rate": 1.7370531645955485e-05, "loss": 0.4857, "step": 7274 }, { "epoch": 0.2591873452446693, "grad_norm": 1.177543952257496, "learning_rate": 1.736975172634733e-05, "loss": 0.5274, "step": 7275 }, { "epoch": 0.25922297237116343, "grad_norm": 1.7804788522275907, "learning_rate": 1.7368971708605785e-05, "loss": 0.633, "step": 7276 }, { "epoch": 0.2592585994976575, "grad_norm": 1.5162095291027478, "learning_rate": 1.736819159274123e-05, "loss": 0.6247, "step": 7277 }, { "epoch": 0.25929422662415164, "grad_norm": 2.948031784798415, "learning_rate": 1.736741137876405e-05, "loss": 0.6126, "step": 7278 }, { "epoch": 0.25932985375064577, "grad_norm": 1.1845085311606411, "learning_rate": 1.736663106668464e-05, "loss": 0.4725, "step": 7279 }, { "epoch": 0.25936548087713984, "grad_norm": 1.198723232291498, "learning_rate": 1.736585065651339e-05, "loss": 0.5122, "step": 7280 }, { "epoch": 0.259401108003634, "grad_norm": 1.6368800771620327, "learning_rate": 1.736507014826069e-05, "loss": 0.6093, "step": 7281 }, { "epoch": 0.2594367351301281, "grad_norm": 2.1702209294169736, "learning_rate": 1.736428954193693e-05, "loss": 0.604, "step": 7282 }, { "epoch": 0.2594723622566222, "grad_norm": 2.201845251758551, "learning_rate": 1.7363508837552513e-05, "loss": 0.6581, "step": 7283 }, { "epoch": 0.2595079893831163, "grad_norm": 1.9316826347323512, "learning_rate": 1.7362728035117826e-05, "loss": 0.6078, "step": 7284 }, { "epoch": 0.25954361650961044, "grad_norm": 1.5133943395245095, "learning_rate": 1.7361947134643265e-05, "loss": 0.5043, "step": 7285 }, { "epoch": 0.2595792436361045, "grad_norm": 1.7474439523446947, "learning_rate": 1.7361166136139238e-05, "loss": 0.662, "step": 7286 }, { "epoch": 0.25961487076259865, "grad_norm": 1.653230871216056, "learning_rate": 1.7360385039616137e-05, "loss": 0.6542, "step": 7287 }, { "epoch": 0.2596504978890928, "grad_norm": 1.156037998936561, "learning_rate": 1.7359603845084365e-05, "loss": 0.5432, "step": 7288 }, { "epoch": 0.25968612501558686, "grad_norm": 1.7025692926556175, "learning_rate": 1.735882255255432e-05, "loss": 0.6173, "step": 7289 }, { "epoch": 0.259721752142081, "grad_norm": 1.742499525054707, "learning_rate": 1.7358041162036414e-05, "loss": 0.6157, "step": 7290 }, { "epoch": 0.2597573792685751, "grad_norm": 1.4937326611923483, "learning_rate": 1.7357259673541045e-05, "loss": 0.5764, "step": 7291 }, { "epoch": 0.2597930063950692, "grad_norm": 1.1458397607187638, "learning_rate": 1.7356478087078625e-05, "loss": 0.4893, "step": 7292 }, { "epoch": 0.2598286335215633, "grad_norm": 1.123020680778429, "learning_rate": 1.735569640265955e-05, "loss": 0.4528, "step": 7293 }, { "epoch": 0.25986426064805745, "grad_norm": 1.3397289112065347, "learning_rate": 1.735491462029424e-05, "loss": 0.5152, "step": 7294 }, { "epoch": 0.25989988777455153, "grad_norm": 1.2200163022629908, "learning_rate": 1.73541327399931e-05, "loss": 0.478, "step": 7295 }, { "epoch": 0.25993551490104566, "grad_norm": 2.7142728947563355, "learning_rate": 1.7353350761766544e-05, "loss": 0.5847, "step": 7296 }, { "epoch": 0.2599711420275398, "grad_norm": 1.8875424410102577, "learning_rate": 1.735256868562498e-05, "loss": 0.6016, "step": 7297 }, { "epoch": 0.26000676915403387, "grad_norm": 1.1031302859580812, "learning_rate": 1.7351786511578827e-05, "loss": 0.4969, "step": 7298 }, { "epoch": 0.260042396280528, "grad_norm": 1.224351250993291, "learning_rate": 1.73510042396385e-05, "loss": 0.5346, "step": 7299 }, { "epoch": 0.26007802340702213, "grad_norm": 1.6768842913254784, "learning_rate": 1.7350221869814413e-05, "loss": 0.7284, "step": 7300 }, { "epoch": 0.2601136505335162, "grad_norm": 1.959362430099398, "learning_rate": 1.7349439402116983e-05, "loss": 0.6629, "step": 7301 }, { "epoch": 0.26014927766001034, "grad_norm": 1.5872658142480998, "learning_rate": 1.734865683655663e-05, "loss": 0.6926, "step": 7302 }, { "epoch": 0.26018490478650447, "grad_norm": 1.7049949395421067, "learning_rate": 1.7347874173143773e-05, "loss": 0.6065, "step": 7303 }, { "epoch": 0.26022053191299854, "grad_norm": 1.4453479967255034, "learning_rate": 1.734709141188884e-05, "loss": 0.5005, "step": 7304 }, { "epoch": 0.2602561590394927, "grad_norm": 1.246772719750159, "learning_rate": 1.7346308552802245e-05, "loss": 0.5042, "step": 7305 }, { "epoch": 0.2602917861659868, "grad_norm": 1.6611403816926311, "learning_rate": 1.7345525595894418e-05, "loss": 0.6775, "step": 7306 }, { "epoch": 0.2603274132924809, "grad_norm": 1.7415344331430145, "learning_rate": 1.7344742541175782e-05, "loss": 0.5702, "step": 7307 }, { "epoch": 0.260363040418975, "grad_norm": 1.417048297690385, "learning_rate": 1.734395938865677e-05, "loss": 0.5522, "step": 7308 }, { "epoch": 0.26039866754546914, "grad_norm": 2.073384191039143, "learning_rate": 1.7343176138347803e-05, "loss": 0.6263, "step": 7309 }, { "epoch": 0.2604342946719632, "grad_norm": 1.7162102253047922, "learning_rate": 1.7342392790259314e-05, "loss": 0.6515, "step": 7310 }, { "epoch": 0.26046992179845735, "grad_norm": 1.2244924373321544, "learning_rate": 1.734160934440173e-05, "loss": 0.4962, "step": 7311 }, { "epoch": 0.2605055489249515, "grad_norm": 1.5904264876317722, "learning_rate": 1.7340825800785494e-05, "loss": 0.5442, "step": 7312 }, { "epoch": 0.26054117605144556, "grad_norm": 1.3565544705270463, "learning_rate": 1.734004215942102e-05, "loss": 0.5507, "step": 7313 }, { "epoch": 0.2605768031779397, "grad_norm": 2.8112151004936505, "learning_rate": 1.7339258420318765e-05, "loss": 0.6907, "step": 7314 }, { "epoch": 0.2606124303044338, "grad_norm": 1.5416051705966731, "learning_rate": 1.7338474583489148e-05, "loss": 0.5516, "step": 7315 }, { "epoch": 0.2606480574309279, "grad_norm": 1.0789096726996128, "learning_rate": 1.7337690648942615e-05, "loss": 0.4795, "step": 7316 }, { "epoch": 0.260683684557422, "grad_norm": 2.03024358253407, "learning_rate": 1.7336906616689598e-05, "loss": 0.5976, "step": 7317 }, { "epoch": 0.26071931168391616, "grad_norm": 1.1693430370811637, "learning_rate": 1.7336122486740546e-05, "loss": 0.533, "step": 7318 }, { "epoch": 0.26075493881041023, "grad_norm": 1.1880545602219477, "learning_rate": 1.7335338259105895e-05, "loss": 0.499, "step": 7319 }, { "epoch": 0.26079056593690436, "grad_norm": 1.0842388825590052, "learning_rate": 1.7334553933796087e-05, "loss": 0.4853, "step": 7320 }, { "epoch": 0.2608261930633985, "grad_norm": 1.1613951057958023, "learning_rate": 1.7333769510821567e-05, "loss": 0.4778, "step": 7321 }, { "epoch": 0.26086182018989257, "grad_norm": 2.091548707350279, "learning_rate": 1.7332984990192782e-05, "loss": 0.6218, "step": 7322 }, { "epoch": 0.2608974473163867, "grad_norm": 1.3185604196890204, "learning_rate": 1.7332200371920173e-05, "loss": 0.5325, "step": 7323 }, { "epoch": 0.26093307444288083, "grad_norm": 1.4935877966647673, "learning_rate": 1.7331415656014196e-05, "loss": 0.5457, "step": 7324 }, { "epoch": 0.2609687015693749, "grad_norm": 1.749925320185492, "learning_rate": 1.7330630842485287e-05, "loss": 0.6864, "step": 7325 }, { "epoch": 0.26100432869586904, "grad_norm": 1.3987300182520612, "learning_rate": 1.732984593134391e-05, "loss": 0.4591, "step": 7326 }, { "epoch": 0.26103995582236317, "grad_norm": 1.444167175375011, "learning_rate": 1.732906092260051e-05, "loss": 0.5048, "step": 7327 }, { "epoch": 0.26107558294885724, "grad_norm": 1.7707745280093004, "learning_rate": 1.7328275816265542e-05, "loss": 0.6213, "step": 7328 }, { "epoch": 0.2611112100753514, "grad_norm": 1.6476809252598048, "learning_rate": 1.732749061234946e-05, "loss": 0.4978, "step": 7329 }, { "epoch": 0.2611468372018455, "grad_norm": 1.6048406951933516, "learning_rate": 1.7326705310862716e-05, "loss": 0.6006, "step": 7330 }, { "epoch": 0.2611824643283396, "grad_norm": 1.1221179697566563, "learning_rate": 1.7325919911815774e-05, "loss": 0.4884, "step": 7331 }, { "epoch": 0.2612180914548337, "grad_norm": 1.297539268269246, "learning_rate": 1.7325134415219085e-05, "loss": 0.5013, "step": 7332 }, { "epoch": 0.26125371858132784, "grad_norm": 1.8710013170486208, "learning_rate": 1.732434882108311e-05, "loss": 0.6613, "step": 7333 }, { "epoch": 0.2612893457078219, "grad_norm": 1.116893425383874, "learning_rate": 1.7323563129418317e-05, "loss": 0.4978, "step": 7334 }, { "epoch": 0.26132497283431605, "grad_norm": 2.549236232434644, "learning_rate": 1.7322777340235154e-05, "loss": 0.6322, "step": 7335 }, { "epoch": 0.2613605999608102, "grad_norm": 1.7880415827672345, "learning_rate": 1.7321991453544096e-05, "loss": 0.5666, "step": 7336 }, { "epoch": 0.26139622708730426, "grad_norm": 2.383458561273571, "learning_rate": 1.7321205469355603e-05, "loss": 0.7286, "step": 7337 }, { "epoch": 0.2614318542137984, "grad_norm": 1.4367232822120846, "learning_rate": 1.7320419387680142e-05, "loss": 0.5897, "step": 7338 }, { "epoch": 0.2614674813402925, "grad_norm": 1.4148344076608388, "learning_rate": 1.7319633208528185e-05, "loss": 0.5428, "step": 7339 }, { "epoch": 0.2615031084667866, "grad_norm": 1.4871519295278877, "learning_rate": 1.7318846931910192e-05, "loss": 0.5241, "step": 7340 }, { "epoch": 0.2615387355932807, "grad_norm": 1.189139281335132, "learning_rate": 1.7318060557836634e-05, "loss": 0.5625, "step": 7341 }, { "epoch": 0.26157436271977486, "grad_norm": 1.5030220337040383, "learning_rate": 1.731727408631799e-05, "loss": 0.5651, "step": 7342 }, { "epoch": 0.26160998984626893, "grad_norm": 1.645894242915789, "learning_rate": 1.7316487517364723e-05, "loss": 0.5143, "step": 7343 }, { "epoch": 0.26164561697276306, "grad_norm": 3.2962454226618023, "learning_rate": 1.731570085098731e-05, "loss": 0.4532, "step": 7344 }, { "epoch": 0.2616812440992572, "grad_norm": 2.175960991713541, "learning_rate": 1.7314914087196232e-05, "loss": 0.6235, "step": 7345 }, { "epoch": 0.26171687122575127, "grad_norm": 1.9352859937683096, "learning_rate": 1.7314127226001954e-05, "loss": 0.6515, "step": 7346 }, { "epoch": 0.2617524983522454, "grad_norm": 1.8957319010049718, "learning_rate": 1.7313340267414964e-05, "loss": 0.5578, "step": 7347 }, { "epoch": 0.26178812547873953, "grad_norm": 1.2184465123284207, "learning_rate": 1.7312553211445738e-05, "loss": 0.5363, "step": 7348 }, { "epoch": 0.2618237526052336, "grad_norm": 1.8653189267038137, "learning_rate": 1.7311766058104752e-05, "loss": 0.5802, "step": 7349 }, { "epoch": 0.26185937973172774, "grad_norm": 1.7613576550387362, "learning_rate": 1.731097880740249e-05, "loss": 0.637, "step": 7350 }, { "epoch": 0.26189500685822187, "grad_norm": 1.8020238429560678, "learning_rate": 1.731019145934943e-05, "loss": 0.6384, "step": 7351 }, { "epoch": 0.26193063398471594, "grad_norm": 1.5246334406346749, "learning_rate": 1.730940401395607e-05, "loss": 0.6397, "step": 7352 }, { "epoch": 0.2619662611112101, "grad_norm": 1.6150379045815377, "learning_rate": 1.7308616471232885e-05, "loss": 0.6344, "step": 7353 }, { "epoch": 0.2620018882377042, "grad_norm": 1.129939112738512, "learning_rate": 1.7307828831190362e-05, "loss": 0.498, "step": 7354 }, { "epoch": 0.2620375153641983, "grad_norm": 1.186828723067529, "learning_rate": 1.7307041093838986e-05, "loss": 0.4851, "step": 7355 }, { "epoch": 0.2620731424906924, "grad_norm": 1.1524520980665642, "learning_rate": 1.7306253259189255e-05, "loss": 0.5071, "step": 7356 }, { "epoch": 0.26210876961718654, "grad_norm": 1.4981340094725737, "learning_rate": 1.7305465327251657e-05, "loss": 0.5684, "step": 7357 }, { "epoch": 0.2621443967436806, "grad_norm": 1.9289386102576276, "learning_rate": 1.730467729803668e-05, "loss": 0.618, "step": 7358 }, { "epoch": 0.26218002387017475, "grad_norm": 2.0178366584334997, "learning_rate": 1.7303889171554816e-05, "loss": 0.5903, "step": 7359 }, { "epoch": 0.2622156509966689, "grad_norm": 1.8909085769802685, "learning_rate": 1.7303100947816567e-05, "loss": 0.5217, "step": 7360 }, { "epoch": 0.26225127812316296, "grad_norm": 1.673382240183979, "learning_rate": 1.7302312626832423e-05, "loss": 0.6537, "step": 7361 }, { "epoch": 0.2622869052496571, "grad_norm": 1.2191489480449509, "learning_rate": 1.730152420861288e-05, "loss": 0.4807, "step": 7362 }, { "epoch": 0.2623225323761512, "grad_norm": 1.5024067372873167, "learning_rate": 1.730073569316844e-05, "loss": 0.6216, "step": 7363 }, { "epoch": 0.2623581595026453, "grad_norm": 2.615854281800979, "learning_rate": 1.7299947080509603e-05, "loss": 0.6323, "step": 7364 }, { "epoch": 0.2623937866291394, "grad_norm": 1.1089205538764257, "learning_rate": 1.7299158370646866e-05, "loss": 0.4709, "step": 7365 }, { "epoch": 0.26242941375563356, "grad_norm": 1.0716743175360994, "learning_rate": 1.7298369563590734e-05, "loss": 0.5083, "step": 7366 }, { "epoch": 0.26246504088212763, "grad_norm": 1.2409211492772523, "learning_rate": 1.729758065935171e-05, "loss": 0.5069, "step": 7367 }, { "epoch": 0.26250066800862176, "grad_norm": 1.0807207228170075, "learning_rate": 1.72967916579403e-05, "loss": 0.5483, "step": 7368 }, { "epoch": 0.2625362951351159, "grad_norm": 1.2196445164074234, "learning_rate": 1.729600255936701e-05, "loss": 0.5409, "step": 7369 }, { "epoch": 0.26257192226160997, "grad_norm": 1.0120444736376764, "learning_rate": 1.7295213363642346e-05, "loss": 0.4998, "step": 7370 }, { "epoch": 0.2626075493881041, "grad_norm": 1.4737305514314925, "learning_rate": 1.7294424070776813e-05, "loss": 0.6408, "step": 7371 }, { "epoch": 0.26264317651459823, "grad_norm": 1.1673369557552096, "learning_rate": 1.729363468078093e-05, "loss": 0.4969, "step": 7372 }, { "epoch": 0.2626788036410923, "grad_norm": 1.4761394556400667, "learning_rate": 1.72928451936652e-05, "loss": 0.5714, "step": 7373 }, { "epoch": 0.26271443076758644, "grad_norm": 3.080854027519809, "learning_rate": 1.7292055609440138e-05, "loss": 0.6173, "step": 7374 }, { "epoch": 0.26275005789408057, "grad_norm": 1.3058019379304575, "learning_rate": 1.729126592811626e-05, "loss": 0.5089, "step": 7375 }, { "epoch": 0.26278568502057464, "grad_norm": 1.2565151470705413, "learning_rate": 1.7290476149704082e-05, "loss": 0.4923, "step": 7376 }, { "epoch": 0.2628213121470688, "grad_norm": 1.350390874870983, "learning_rate": 1.7289686274214116e-05, "loss": 0.6305, "step": 7377 }, { "epoch": 0.2628569392735629, "grad_norm": 1.2381890070304071, "learning_rate": 1.7288896301656885e-05, "loss": 0.5328, "step": 7378 }, { "epoch": 0.262892566400057, "grad_norm": 1.449106014232291, "learning_rate": 1.72881062320429e-05, "loss": 0.6462, "step": 7379 }, { "epoch": 0.2629281935265511, "grad_norm": 1.1778935276249751, "learning_rate": 1.728731606538269e-05, "loss": 0.514, "step": 7380 }, { "epoch": 0.26296382065304524, "grad_norm": 1.4485199213281532, "learning_rate": 1.7286525801686774e-05, "loss": 0.6586, "step": 7381 }, { "epoch": 0.2629994477795393, "grad_norm": 1.1530264414333111, "learning_rate": 1.7285735440965673e-05, "loss": 0.5174, "step": 7382 }, { "epoch": 0.26303507490603345, "grad_norm": 1.255217592763896, "learning_rate": 1.728494498322991e-05, "loss": 0.5523, "step": 7383 }, { "epoch": 0.2630707020325276, "grad_norm": 1.7988545677364698, "learning_rate": 1.7284154428490014e-05, "loss": 0.6365, "step": 7384 }, { "epoch": 0.26310632915902166, "grad_norm": 1.294362089633865, "learning_rate": 1.728336377675651e-05, "loss": 0.5156, "step": 7385 }, { "epoch": 0.2631419562855158, "grad_norm": 1.7244420462251726, "learning_rate": 1.7282573028039927e-05, "loss": 0.6247, "step": 7386 }, { "epoch": 0.2631775834120099, "grad_norm": 1.1742520216915677, "learning_rate": 1.7281782182350796e-05, "loss": 0.4706, "step": 7387 }, { "epoch": 0.263213210538504, "grad_norm": 2.0007472960728294, "learning_rate": 1.7280991239699643e-05, "loss": 0.5967, "step": 7388 }, { "epoch": 0.2632488376649981, "grad_norm": 1.4618197821030343, "learning_rate": 1.7280200200097003e-05, "loss": 0.5709, "step": 7389 }, { "epoch": 0.26328446479149226, "grad_norm": 2.0685019459949507, "learning_rate": 1.727940906355341e-05, "loss": 0.6786, "step": 7390 }, { "epoch": 0.26332009191798633, "grad_norm": 1.3610436695177497, "learning_rate": 1.7278617830079394e-05, "loss": 0.5221, "step": 7391 }, { "epoch": 0.26335571904448046, "grad_norm": 1.4251242048253958, "learning_rate": 1.72778264996855e-05, "loss": 0.6214, "step": 7392 }, { "epoch": 0.2633913461709746, "grad_norm": 1.3995095963313535, "learning_rate": 1.727703507238225e-05, "loss": 0.6622, "step": 7393 }, { "epoch": 0.26342697329746867, "grad_norm": 1.482979426236248, "learning_rate": 1.72762435481802e-05, "loss": 0.5585, "step": 7394 }, { "epoch": 0.2634626004239628, "grad_norm": 1.3620413980667563, "learning_rate": 1.727545192708988e-05, "loss": 0.5506, "step": 7395 }, { "epoch": 0.26349822755045693, "grad_norm": 1.9940469190192274, "learning_rate": 1.727466020912183e-05, "loss": 0.6258, "step": 7396 }, { "epoch": 0.263533854676951, "grad_norm": 1.5136345294345588, "learning_rate": 1.7273868394286596e-05, "loss": 0.633, "step": 7397 }, { "epoch": 0.26356948180344514, "grad_norm": 1.8365468154824782, "learning_rate": 1.727307648259472e-05, "loss": 0.5904, "step": 7398 }, { "epoch": 0.26360510892993927, "grad_norm": 1.1506636949144133, "learning_rate": 1.7272284474056748e-05, "loss": 0.5427, "step": 7399 }, { "epoch": 0.26364073605643334, "grad_norm": 1.5796219904669697, "learning_rate": 1.7271492368683224e-05, "loss": 0.6362, "step": 7400 }, { "epoch": 0.2636763631829275, "grad_norm": 1.1189054631036286, "learning_rate": 1.72707001664847e-05, "loss": 0.5024, "step": 7401 }, { "epoch": 0.2637119903094216, "grad_norm": 1.6140315203104685, "learning_rate": 1.7269907867471715e-05, "loss": 0.5876, "step": 7402 }, { "epoch": 0.2637476174359157, "grad_norm": 1.3689760376284832, "learning_rate": 1.7269115471654828e-05, "loss": 0.5418, "step": 7403 }, { "epoch": 0.2637832445624098, "grad_norm": 1.7382807427066087, "learning_rate": 1.726832297904459e-05, "loss": 0.5974, "step": 7404 }, { "epoch": 0.26381887168890394, "grad_norm": 1.1754839269275195, "learning_rate": 1.7267530389651546e-05, "loss": 0.477, "step": 7405 }, { "epoch": 0.263854498815398, "grad_norm": 1.163090324783868, "learning_rate": 1.7266737703486263e-05, "loss": 0.5123, "step": 7406 }, { "epoch": 0.26389012594189215, "grad_norm": 1.4983740103688004, "learning_rate": 1.726594492055928e-05, "loss": 0.6101, "step": 7407 }, { "epoch": 0.2639257530683863, "grad_norm": 1.9231151616450595, "learning_rate": 1.726515204088117e-05, "loss": 0.5622, "step": 7408 }, { "epoch": 0.26396138019488036, "grad_norm": 1.2517325425862693, "learning_rate": 1.7264359064462478e-05, "loss": 0.5299, "step": 7409 }, { "epoch": 0.2639970073213745, "grad_norm": 1.7629025311261168, "learning_rate": 1.7263565991313767e-05, "loss": 0.6345, "step": 7410 }, { "epoch": 0.2640326344478686, "grad_norm": 1.8117377465913365, "learning_rate": 1.72627728214456e-05, "loss": 0.6399, "step": 7411 }, { "epoch": 0.2640682615743627, "grad_norm": 1.2569617398014215, "learning_rate": 1.7261979554868533e-05, "loss": 0.4993, "step": 7412 }, { "epoch": 0.2641038887008568, "grad_norm": 1.5582049014581432, "learning_rate": 1.7261186191593135e-05, "loss": 0.5455, "step": 7413 }, { "epoch": 0.26413951582735096, "grad_norm": 1.5045465109126863, "learning_rate": 1.726039273162997e-05, "loss": 0.5861, "step": 7414 }, { "epoch": 0.26417514295384503, "grad_norm": 1.1895352884633401, "learning_rate": 1.7259599174989598e-05, "loss": 0.5076, "step": 7415 }, { "epoch": 0.26421077008033916, "grad_norm": 1.9368735221130966, "learning_rate": 1.725880552168259e-05, "loss": 0.5818, "step": 7416 }, { "epoch": 0.2642463972068333, "grad_norm": 1.7130773077315606, "learning_rate": 1.725801177171951e-05, "loss": 0.6479, "step": 7417 }, { "epoch": 0.26428202433332737, "grad_norm": 2.880360406959819, "learning_rate": 1.7257217925110934e-05, "loss": 0.5895, "step": 7418 }, { "epoch": 0.2643176514598215, "grad_norm": 1.6045726119316817, "learning_rate": 1.7256423981867425e-05, "loss": 0.6612, "step": 7419 }, { "epoch": 0.26435327858631563, "grad_norm": 1.6463552991205572, "learning_rate": 1.7255629941999565e-05, "loss": 0.5939, "step": 7420 }, { "epoch": 0.2643889057128097, "grad_norm": 1.4942652967499939, "learning_rate": 1.7254835805517913e-05, "loss": 0.617, "step": 7421 }, { "epoch": 0.26442453283930384, "grad_norm": 1.2925014026250803, "learning_rate": 1.7254041572433058e-05, "loss": 0.5219, "step": 7422 }, { "epoch": 0.26446015996579797, "grad_norm": 1.8179314883492015, "learning_rate": 1.7253247242755565e-05, "loss": 0.6751, "step": 7423 }, { "epoch": 0.26449578709229205, "grad_norm": 1.0834008416494236, "learning_rate": 1.7252452816496017e-05, "loss": 0.4943, "step": 7424 }, { "epoch": 0.2645314142187862, "grad_norm": 1.953024421753667, "learning_rate": 1.725165829366499e-05, "loss": 0.6395, "step": 7425 }, { "epoch": 0.2645670413452803, "grad_norm": 1.4704583692413131, "learning_rate": 1.7250863674273063e-05, "loss": 0.5528, "step": 7426 }, { "epoch": 0.2646026684717744, "grad_norm": 1.2232333553832255, "learning_rate": 1.7250068958330822e-05, "loss": 0.4927, "step": 7427 }, { "epoch": 0.2646382955982685, "grad_norm": 1.2485035113781675, "learning_rate": 1.724927414584884e-05, "loss": 0.5594, "step": 7428 }, { "epoch": 0.26467392272476264, "grad_norm": 1.1210322095766754, "learning_rate": 1.724847923683771e-05, "loss": 0.4987, "step": 7429 }, { "epoch": 0.2647095498512567, "grad_norm": 1.1053259679871785, "learning_rate": 1.7247684231308012e-05, "loss": 0.4859, "step": 7430 }, { "epoch": 0.26474517697775085, "grad_norm": 1.0467011121583563, "learning_rate": 1.7246889129270332e-05, "loss": 0.5238, "step": 7431 }, { "epoch": 0.264780804104245, "grad_norm": 1.434114757732418, "learning_rate": 1.724609393073526e-05, "loss": 0.643, "step": 7432 }, { "epoch": 0.26481643123073906, "grad_norm": 1.851857722030543, "learning_rate": 1.724529863571338e-05, "loss": 0.6666, "step": 7433 }, { "epoch": 0.2648520583572332, "grad_norm": 1.192205887238758, "learning_rate": 1.7244503244215285e-05, "loss": 0.51, "step": 7434 }, { "epoch": 0.2648876854837273, "grad_norm": 1.7584677755758682, "learning_rate": 1.7243707756251566e-05, "loss": 0.655, "step": 7435 }, { "epoch": 0.2649233126102214, "grad_norm": 1.1860519729993984, "learning_rate": 1.7242912171832814e-05, "loss": 0.4782, "step": 7436 }, { "epoch": 0.2649589397367155, "grad_norm": 1.1250897894189673, "learning_rate": 1.7242116490969623e-05, "loss": 0.4486, "step": 7437 }, { "epoch": 0.26499456686320966, "grad_norm": 1.6406687931856418, "learning_rate": 1.7241320713672592e-05, "loss": 0.7301, "step": 7438 }, { "epoch": 0.26503019398970373, "grad_norm": 1.6832764782996683, "learning_rate": 1.7240524839952315e-05, "loss": 0.6337, "step": 7439 }, { "epoch": 0.26506582111619786, "grad_norm": 1.3269076885749003, "learning_rate": 1.7239728869819384e-05, "loss": 0.5346, "step": 7440 }, { "epoch": 0.265101448242692, "grad_norm": 1.203158161435827, "learning_rate": 1.723893280328441e-05, "loss": 0.4544, "step": 7441 }, { "epoch": 0.2651370753691861, "grad_norm": 1.4955393604806033, "learning_rate": 1.7238136640357977e-05, "loss": 0.52, "step": 7442 }, { "epoch": 0.2651727024956802, "grad_norm": 2.0236373694724894, "learning_rate": 1.72373403810507e-05, "loss": 0.5699, "step": 7443 }, { "epoch": 0.26520832962217433, "grad_norm": 1.0718899853272348, "learning_rate": 1.723654402537318e-05, "loss": 0.5121, "step": 7444 }, { "epoch": 0.26524395674866846, "grad_norm": 1.7605335555192905, "learning_rate": 1.7235747573336016e-05, "loss": 0.7017, "step": 7445 }, { "epoch": 0.26527958387516254, "grad_norm": 3.897479995685828, "learning_rate": 1.723495102494982e-05, "loss": 0.702, "step": 7446 }, { "epoch": 0.26531521100165667, "grad_norm": 1.6160169301536087, "learning_rate": 1.7234154380225185e-05, "loss": 0.6736, "step": 7447 }, { "epoch": 0.2653508381281508, "grad_norm": 1.1766570643727445, "learning_rate": 1.7233357639172735e-05, "loss": 0.503, "step": 7448 }, { "epoch": 0.2653864652546449, "grad_norm": 1.19813241752675, "learning_rate": 1.7232560801803067e-05, "loss": 0.5059, "step": 7449 }, { "epoch": 0.265422092381139, "grad_norm": 1.3719493746109697, "learning_rate": 1.7231763868126804e-05, "loss": 0.6128, "step": 7450 }, { "epoch": 0.26545771950763314, "grad_norm": 1.6889369183559733, "learning_rate": 1.7230966838154543e-05, "loss": 0.6049, "step": 7451 }, { "epoch": 0.2654933466341272, "grad_norm": 1.432738166674917, "learning_rate": 1.723016971189691e-05, "loss": 0.5502, "step": 7452 }, { "epoch": 0.26552897376062135, "grad_norm": 1.7686138710665302, "learning_rate": 1.722937248936451e-05, "loss": 0.5926, "step": 7453 }, { "epoch": 0.2655646008871155, "grad_norm": 1.195899410069825, "learning_rate": 1.722857517056796e-05, "loss": 0.4713, "step": 7454 }, { "epoch": 0.26560022801360955, "grad_norm": 1.5434841741877927, "learning_rate": 1.7227777755517886e-05, "loss": 0.6524, "step": 7455 }, { "epoch": 0.2656358551401037, "grad_norm": 1.1944547706802955, "learning_rate": 1.7226980244224896e-05, "loss": 0.5014, "step": 7456 }, { "epoch": 0.2656714822665978, "grad_norm": 1.0771931318886843, "learning_rate": 1.722618263669961e-05, "loss": 0.523, "step": 7457 }, { "epoch": 0.2657071093930919, "grad_norm": 1.1952266341196705, "learning_rate": 1.7225384932952655e-05, "loss": 0.4968, "step": 7458 }, { "epoch": 0.265742736519586, "grad_norm": 1.1289144867695673, "learning_rate": 1.7224587132994648e-05, "loss": 0.524, "step": 7459 }, { "epoch": 0.26577836364608015, "grad_norm": 1.17892720759147, "learning_rate": 1.7223789236836212e-05, "loss": 0.5403, "step": 7460 }, { "epoch": 0.2658139907725742, "grad_norm": 1.2078679270789843, "learning_rate": 1.7222991244487976e-05, "loss": 0.5431, "step": 7461 }, { "epoch": 0.26584961789906836, "grad_norm": 1.955501634767991, "learning_rate": 1.7222193155960565e-05, "loss": 0.6728, "step": 7462 }, { "epoch": 0.2658852450255625, "grad_norm": 1.0985347886499846, "learning_rate": 1.72213949712646e-05, "loss": 0.5243, "step": 7463 }, { "epoch": 0.26592087215205656, "grad_norm": 1.7414267770470546, "learning_rate": 1.7220596690410712e-05, "loss": 0.5514, "step": 7464 }, { "epoch": 0.2659564992785507, "grad_norm": 1.5032390288319561, "learning_rate": 1.7219798313409538e-05, "loss": 0.5219, "step": 7465 }, { "epoch": 0.2659921264050448, "grad_norm": 1.6365471674875725, "learning_rate": 1.72189998402717e-05, "loss": 0.6195, "step": 7466 }, { "epoch": 0.2660277535315389, "grad_norm": 1.5288691366770266, "learning_rate": 1.7218201271007832e-05, "loss": 0.6404, "step": 7467 }, { "epoch": 0.26606338065803303, "grad_norm": 1.1863364965618362, "learning_rate": 1.7217402605628572e-05, "loss": 0.5354, "step": 7468 }, { "epoch": 0.26609900778452716, "grad_norm": 1.503529850279497, "learning_rate": 1.721660384414455e-05, "loss": 0.5972, "step": 7469 }, { "epoch": 0.26613463491102124, "grad_norm": 1.2974354353823632, "learning_rate": 1.7215804986566405e-05, "loss": 0.5595, "step": 7470 }, { "epoch": 0.26617026203751537, "grad_norm": 2.123175692211424, "learning_rate": 1.721500603290477e-05, "loss": 0.6171, "step": 7471 }, { "epoch": 0.2662058891640095, "grad_norm": 1.5764064948576462, "learning_rate": 1.7214206983170287e-05, "loss": 0.5907, "step": 7472 }, { "epoch": 0.2662415162905036, "grad_norm": 1.4800575407261103, "learning_rate": 1.7213407837373596e-05, "loss": 0.5813, "step": 7473 }, { "epoch": 0.2662771434169977, "grad_norm": 1.1325902537054253, "learning_rate": 1.7212608595525342e-05, "loss": 0.5177, "step": 7474 }, { "epoch": 0.26631277054349184, "grad_norm": 1.6637876238180467, "learning_rate": 1.7211809257636158e-05, "loss": 0.6928, "step": 7475 }, { "epoch": 0.2663483976699859, "grad_norm": 1.423488936844951, "learning_rate": 1.7211009823716695e-05, "loss": 0.4932, "step": 7476 }, { "epoch": 0.26638402479648005, "grad_norm": 1.2738244837265167, "learning_rate": 1.7210210293777597e-05, "loss": 0.5183, "step": 7477 }, { "epoch": 0.2664196519229742, "grad_norm": 1.8264338823778457, "learning_rate": 1.7209410667829506e-05, "loss": 0.6312, "step": 7478 }, { "epoch": 0.26645527904946825, "grad_norm": 1.0816800237735673, "learning_rate": 1.720861094588308e-05, "loss": 0.4507, "step": 7479 }, { "epoch": 0.2664909061759624, "grad_norm": 1.1606396864499526, "learning_rate": 1.7207811127948953e-05, "loss": 0.5138, "step": 7480 }, { "epoch": 0.2665265333024565, "grad_norm": 1.5747264288041474, "learning_rate": 1.7207011214037785e-05, "loss": 0.619, "step": 7481 }, { "epoch": 0.2665621604289506, "grad_norm": 1.665997290660049, "learning_rate": 1.7206211204160227e-05, "loss": 0.6043, "step": 7482 }, { "epoch": 0.2665977875554447, "grad_norm": 1.5764924590555178, "learning_rate": 1.720541109832693e-05, "loss": 0.5993, "step": 7483 }, { "epoch": 0.26663341468193885, "grad_norm": 1.679592842788531, "learning_rate": 1.7204610896548547e-05, "loss": 0.6124, "step": 7484 }, { "epoch": 0.2666690418084329, "grad_norm": 1.3218505407640282, "learning_rate": 1.720381059883573e-05, "loss": 0.5218, "step": 7485 }, { "epoch": 0.26670466893492706, "grad_norm": 1.143021209590672, "learning_rate": 1.7203010205199144e-05, "loss": 0.5324, "step": 7486 }, { "epoch": 0.2667402960614212, "grad_norm": 1.244513156314951, "learning_rate": 1.7202209715649444e-05, "loss": 0.5228, "step": 7487 }, { "epoch": 0.26677592318791526, "grad_norm": 1.9143160147534475, "learning_rate": 1.7201409130197288e-05, "loss": 0.6002, "step": 7488 }, { "epoch": 0.2668115503144094, "grad_norm": 1.1323709190637765, "learning_rate": 1.720060844885333e-05, "loss": 0.5268, "step": 7489 }, { "epoch": 0.2668471774409035, "grad_norm": 1.10714139930119, "learning_rate": 1.7199807671628243e-05, "loss": 0.5442, "step": 7490 }, { "epoch": 0.2668828045673976, "grad_norm": 1.112396075100672, "learning_rate": 1.7199006798532683e-05, "loss": 0.4784, "step": 7491 }, { "epoch": 0.26691843169389173, "grad_norm": 1.2343409926149542, "learning_rate": 1.7198205829577314e-05, "loss": 0.4771, "step": 7492 }, { "epoch": 0.26695405882038586, "grad_norm": 1.1312670472801696, "learning_rate": 1.7197404764772807e-05, "loss": 0.4955, "step": 7493 }, { "epoch": 0.26698968594687994, "grad_norm": 1.1615929703446979, "learning_rate": 1.7196603604129823e-05, "loss": 0.5078, "step": 7494 }, { "epoch": 0.26702531307337407, "grad_norm": 1.261787010527124, "learning_rate": 1.719580234765903e-05, "loss": 0.5155, "step": 7495 }, { "epoch": 0.2670609401998682, "grad_norm": 1.1623289540362252, "learning_rate": 1.7195000995371104e-05, "loss": 0.5186, "step": 7496 }, { "epoch": 0.2670965673263623, "grad_norm": 1.1305556853368672, "learning_rate": 1.7194199547276708e-05, "loss": 0.4774, "step": 7497 }, { "epoch": 0.2671321944528564, "grad_norm": 1.7121578415392684, "learning_rate": 1.7193398003386514e-05, "loss": 0.5762, "step": 7498 }, { "epoch": 0.26716782157935054, "grad_norm": 1.9648588813571193, "learning_rate": 1.71925963637112e-05, "loss": 0.6181, "step": 7499 }, { "epoch": 0.2672034487058446, "grad_norm": 1.1412600034282445, "learning_rate": 1.7191794628261437e-05, "loss": 0.4787, "step": 7500 }, { "epoch": 0.26723907583233875, "grad_norm": 1.6467329526955183, "learning_rate": 1.71909927970479e-05, "loss": 0.5985, "step": 7501 }, { "epoch": 0.2672747029588329, "grad_norm": 1.448137871121686, "learning_rate": 1.7190190870081272e-05, "loss": 0.6029, "step": 7502 }, { "epoch": 0.26731033008532695, "grad_norm": 1.8214387937257246, "learning_rate": 1.7189388847372227e-05, "loss": 0.6251, "step": 7503 }, { "epoch": 0.2673459572118211, "grad_norm": 2.6825745839809647, "learning_rate": 1.718858672893144e-05, "loss": 0.5517, "step": 7504 }, { "epoch": 0.2673815843383152, "grad_norm": 1.5987428020256407, "learning_rate": 1.7187784514769598e-05, "loss": 0.6131, "step": 7505 }, { "epoch": 0.2674172114648093, "grad_norm": 1.2338362573300272, "learning_rate": 1.7186982204897386e-05, "loss": 0.4798, "step": 7506 }, { "epoch": 0.2674528385913034, "grad_norm": 1.705581902719292, "learning_rate": 1.7186179799325477e-05, "loss": 0.5442, "step": 7507 }, { "epoch": 0.26748846571779755, "grad_norm": 1.276029555938727, "learning_rate": 1.7185377298064562e-05, "loss": 0.5106, "step": 7508 }, { "epoch": 0.2675240928442916, "grad_norm": 2.2746264785479378, "learning_rate": 1.718457470112533e-05, "loss": 0.5991, "step": 7509 }, { "epoch": 0.26755971997078576, "grad_norm": 1.5636105306013401, "learning_rate": 1.7183772008518458e-05, "loss": 0.673, "step": 7510 }, { "epoch": 0.2675953470972799, "grad_norm": 1.1996490776164643, "learning_rate": 1.718296922025464e-05, "loss": 0.5245, "step": 7511 }, { "epoch": 0.26763097422377397, "grad_norm": 1.603292318516953, "learning_rate": 1.7182166336344572e-05, "loss": 0.6139, "step": 7512 }, { "epoch": 0.2676666013502681, "grad_norm": 2.0027039934662363, "learning_rate": 1.7181363356798938e-05, "loss": 0.5885, "step": 7513 }, { "epoch": 0.2677022284767622, "grad_norm": 1.591075116945721, "learning_rate": 1.718056028162843e-05, "loss": 0.6271, "step": 7514 }, { "epoch": 0.2677378556032563, "grad_norm": 1.4341504613520872, "learning_rate": 1.717975711084374e-05, "loss": 0.5594, "step": 7515 }, { "epoch": 0.26777348272975043, "grad_norm": 1.1807470057534457, "learning_rate": 1.7178953844455572e-05, "loss": 0.5159, "step": 7516 }, { "epoch": 0.26780910985624456, "grad_norm": 1.2042395516307827, "learning_rate": 1.7178150482474613e-05, "loss": 0.5014, "step": 7517 }, { "epoch": 0.26784473698273864, "grad_norm": 1.2273227398298587, "learning_rate": 1.7177347024911562e-05, "loss": 0.5158, "step": 7518 }, { "epoch": 0.26788036410923277, "grad_norm": 1.1662544856482573, "learning_rate": 1.717654347177712e-05, "loss": 0.5672, "step": 7519 }, { "epoch": 0.2679159912357269, "grad_norm": 1.085392607255058, "learning_rate": 1.7175739823081985e-05, "loss": 0.5277, "step": 7520 }, { "epoch": 0.267951618362221, "grad_norm": 1.5751673603467404, "learning_rate": 1.717493607883686e-05, "loss": 0.5447, "step": 7521 }, { "epoch": 0.2679872454887151, "grad_norm": 1.4862613010218853, "learning_rate": 1.7174132239052443e-05, "loss": 0.5714, "step": 7522 }, { "epoch": 0.26802287261520924, "grad_norm": 1.2281817489484714, "learning_rate": 1.7173328303739442e-05, "loss": 0.5023, "step": 7523 }, { "epoch": 0.2680584997417033, "grad_norm": 1.321534441318988, "learning_rate": 1.7172524272908562e-05, "loss": 0.4575, "step": 7524 }, { "epoch": 0.26809412686819745, "grad_norm": 1.645454934047751, "learning_rate": 1.717172014657051e-05, "loss": 0.6555, "step": 7525 }, { "epoch": 0.2681297539946916, "grad_norm": 1.132394853300206, "learning_rate": 1.717091592473599e-05, "loss": 0.4743, "step": 7526 }, { "epoch": 0.26816538112118565, "grad_norm": 1.508328387499303, "learning_rate": 1.7170111607415706e-05, "loss": 0.6156, "step": 7527 }, { "epoch": 0.2682010082476798, "grad_norm": 1.244901437967981, "learning_rate": 1.7169307194620383e-05, "loss": 0.5295, "step": 7528 }, { "epoch": 0.2682366353741739, "grad_norm": 1.292352744347934, "learning_rate": 1.716850268636072e-05, "loss": 0.4969, "step": 7529 }, { "epoch": 0.268272262500668, "grad_norm": 1.5544701577823326, "learning_rate": 1.7167698082647433e-05, "loss": 0.5484, "step": 7530 }, { "epoch": 0.2683078896271621, "grad_norm": 1.4284622240431228, "learning_rate": 1.7166893383491237e-05, "loss": 0.6209, "step": 7531 }, { "epoch": 0.26834351675365625, "grad_norm": 1.1112728493589303, "learning_rate": 1.7166088588902843e-05, "loss": 0.502, "step": 7532 }, { "epoch": 0.26837914388015033, "grad_norm": 1.4151527999514326, "learning_rate": 1.7165283698892977e-05, "loss": 0.5384, "step": 7533 }, { "epoch": 0.26841477100664446, "grad_norm": 2.381248698840307, "learning_rate": 1.7164478713472346e-05, "loss": 0.606, "step": 7534 }, { "epoch": 0.2684503981331386, "grad_norm": 1.1756938667555707, "learning_rate": 1.7163673632651676e-05, "loss": 0.4958, "step": 7535 }, { "epoch": 0.26848602525963267, "grad_norm": 1.781323764309269, "learning_rate": 1.7162868456441678e-05, "loss": 0.5152, "step": 7536 }, { "epoch": 0.2685216523861268, "grad_norm": 1.6966410526641613, "learning_rate": 1.7162063184853087e-05, "loss": 0.6058, "step": 7537 }, { "epoch": 0.26855727951262093, "grad_norm": 1.1245114951031066, "learning_rate": 1.7161257817896613e-05, "loss": 0.4984, "step": 7538 }, { "epoch": 0.268592906639115, "grad_norm": 1.6070363473913671, "learning_rate": 1.7160452355582993e-05, "loss": 0.4825, "step": 7539 }, { "epoch": 0.26862853376560913, "grad_norm": 1.0804898610981737, "learning_rate": 1.715964679792294e-05, "loss": 0.5251, "step": 7540 }, { "epoch": 0.26866416089210327, "grad_norm": 1.7147749840662365, "learning_rate": 1.7158841144927186e-05, "loss": 0.5617, "step": 7541 }, { "epoch": 0.26869978801859734, "grad_norm": 1.0965790019022374, "learning_rate": 1.7158035396606457e-05, "loss": 0.5058, "step": 7542 }, { "epoch": 0.26873541514509147, "grad_norm": 1.6307099316585671, "learning_rate": 1.715722955297149e-05, "loss": 0.6319, "step": 7543 }, { "epoch": 0.2687710422715856, "grad_norm": 1.0771901157424415, "learning_rate": 1.7156423614033e-05, "loss": 0.465, "step": 7544 }, { "epoch": 0.2688066693980797, "grad_norm": 1.6257666894419176, "learning_rate": 1.7155617579801735e-05, "loss": 0.5638, "step": 7545 }, { "epoch": 0.2688422965245738, "grad_norm": 1.1131880667553737, "learning_rate": 1.7154811450288414e-05, "loss": 0.5252, "step": 7546 }, { "epoch": 0.26887792365106794, "grad_norm": 1.1118173783875476, "learning_rate": 1.7154005225503785e-05, "loss": 0.5042, "step": 7547 }, { "epoch": 0.268913550777562, "grad_norm": 2.2595901520772594, "learning_rate": 1.715319890545857e-05, "loss": 0.6434, "step": 7548 }, { "epoch": 0.26894917790405615, "grad_norm": 1.4694124265847188, "learning_rate": 1.7152392490163516e-05, "loss": 0.6706, "step": 7549 }, { "epoch": 0.2689848050305503, "grad_norm": 1.7639973025497446, "learning_rate": 1.7151585979629354e-05, "loss": 0.5564, "step": 7550 }, { "epoch": 0.26902043215704435, "grad_norm": 1.1464565814133605, "learning_rate": 1.7150779373866825e-05, "loss": 0.5079, "step": 7551 }, { "epoch": 0.2690560592835385, "grad_norm": 1.3848118635372606, "learning_rate": 1.7149972672886673e-05, "loss": 0.6442, "step": 7552 }, { "epoch": 0.2690916864100326, "grad_norm": 2.174267647779358, "learning_rate": 1.7149165876699635e-05, "loss": 0.6072, "step": 7553 }, { "epoch": 0.2691273135365267, "grad_norm": 1.2380890573380643, "learning_rate": 1.714835898531646e-05, "loss": 0.4878, "step": 7554 }, { "epoch": 0.2691629406630208, "grad_norm": 1.6824726207590257, "learning_rate": 1.7147551998747886e-05, "loss": 0.5979, "step": 7555 }, { "epoch": 0.26919856778951495, "grad_norm": 1.6724009404236009, "learning_rate": 1.7146744917004662e-05, "loss": 0.6001, "step": 7556 }, { "epoch": 0.26923419491600903, "grad_norm": 1.5543576700521475, "learning_rate": 1.7145937740097537e-05, "loss": 0.6492, "step": 7557 }, { "epoch": 0.26926982204250316, "grad_norm": 1.3490693979093888, "learning_rate": 1.7145130468037252e-05, "loss": 0.5639, "step": 7558 }, { "epoch": 0.2693054491689973, "grad_norm": 1.0944290178634837, "learning_rate": 1.714432310083456e-05, "loss": 0.5073, "step": 7559 }, { "epoch": 0.26934107629549137, "grad_norm": 1.4368555942260937, "learning_rate": 1.7143515638500216e-05, "loss": 0.5575, "step": 7560 }, { "epoch": 0.2693767034219855, "grad_norm": 1.6285609362699864, "learning_rate": 1.714270808104497e-05, "loss": 0.6324, "step": 7561 }, { "epoch": 0.26941233054847963, "grad_norm": 1.1364442768481795, "learning_rate": 1.7141900428479568e-05, "loss": 0.5304, "step": 7562 }, { "epoch": 0.2694479576749737, "grad_norm": 1.1242581327897088, "learning_rate": 1.7141092680814772e-05, "loss": 0.5032, "step": 7563 }, { "epoch": 0.26948358480146783, "grad_norm": 1.1003526830982526, "learning_rate": 1.7140284838061336e-05, "loss": 0.502, "step": 7564 }, { "epoch": 0.26951921192796197, "grad_norm": 1.921649586623591, "learning_rate": 1.7139476900230013e-05, "loss": 0.6014, "step": 7565 }, { "epoch": 0.26955483905445604, "grad_norm": 1.1182786161985105, "learning_rate": 1.713866886733157e-05, "loss": 0.499, "step": 7566 }, { "epoch": 0.2695904661809502, "grad_norm": 1.1486221801775187, "learning_rate": 1.713786073937676e-05, "loss": 0.5094, "step": 7567 }, { "epoch": 0.2696260933074443, "grad_norm": 2.2698091459598886, "learning_rate": 1.7137052516376344e-05, "loss": 0.5524, "step": 7568 }, { "epoch": 0.2696617204339384, "grad_norm": 1.1059130976020055, "learning_rate": 1.7136244198341087e-05, "loss": 0.5149, "step": 7569 }, { "epoch": 0.2696973475604325, "grad_norm": 1.1745945793539545, "learning_rate": 1.713543578528175e-05, "loss": 0.5154, "step": 7570 }, { "epoch": 0.26973297468692664, "grad_norm": 1.9581927363101383, "learning_rate": 1.7134627277209097e-05, "loss": 0.5986, "step": 7571 }, { "epoch": 0.2697686018134207, "grad_norm": 1.1031979977255666, "learning_rate": 1.7133818674133893e-05, "loss": 0.4918, "step": 7572 }, { "epoch": 0.26980422893991485, "grad_norm": 1.3233732898448103, "learning_rate": 1.7133009976066912e-05, "loss": 0.5156, "step": 7573 }, { "epoch": 0.269839856066409, "grad_norm": 1.1520124775107206, "learning_rate": 1.7132201183018915e-05, "loss": 0.4749, "step": 7574 }, { "epoch": 0.26987548319290305, "grad_norm": 1.7984844743005555, "learning_rate": 1.7131392295000676e-05, "loss": 0.6873, "step": 7575 }, { "epoch": 0.2699111103193972, "grad_norm": 1.1225926254998027, "learning_rate": 1.7130583312022962e-05, "loss": 0.4999, "step": 7576 }, { "epoch": 0.2699467374458913, "grad_norm": 2.122642940222253, "learning_rate": 1.7129774234096544e-05, "loss": 0.5306, "step": 7577 }, { "epoch": 0.2699823645723854, "grad_norm": 1.1399727669264081, "learning_rate": 1.7128965061232204e-05, "loss": 0.5175, "step": 7578 }, { "epoch": 0.2700179916988795, "grad_norm": 1.0648273357031195, "learning_rate": 1.7128155793440712e-05, "loss": 0.4743, "step": 7579 }, { "epoch": 0.27005361882537365, "grad_norm": 1.2792168756367372, "learning_rate": 1.712734643073284e-05, "loss": 0.4958, "step": 7580 }, { "epoch": 0.27008924595186773, "grad_norm": 1.2342513639664587, "learning_rate": 1.7126536973119366e-05, "loss": 0.4914, "step": 7581 }, { "epoch": 0.27012487307836186, "grad_norm": 1.613751538856344, "learning_rate": 1.7125727420611074e-05, "loss": 0.5595, "step": 7582 }, { "epoch": 0.270160500204856, "grad_norm": 1.5254101313541804, "learning_rate": 1.7124917773218745e-05, "loss": 0.5861, "step": 7583 }, { "epoch": 0.27019612733135007, "grad_norm": 1.5373079301257897, "learning_rate": 1.712410803095315e-05, "loss": 0.6674, "step": 7584 }, { "epoch": 0.2702317544578442, "grad_norm": 2.786429697980696, "learning_rate": 1.7123298193825083e-05, "loss": 0.5931, "step": 7585 }, { "epoch": 0.27026738158433833, "grad_norm": 1.5857677148187665, "learning_rate": 1.7122488261845316e-05, "loss": 0.6515, "step": 7586 }, { "epoch": 0.2703030087108324, "grad_norm": 1.7612963915745654, "learning_rate": 1.7121678235024642e-05, "loss": 0.5418, "step": 7587 }, { "epoch": 0.27033863583732654, "grad_norm": 1.1453920859229578, "learning_rate": 1.7120868113373842e-05, "loss": 0.4429, "step": 7588 }, { "epoch": 0.27037426296382067, "grad_norm": 1.6039585745749814, "learning_rate": 1.712005789690371e-05, "loss": 0.5247, "step": 7589 }, { "epoch": 0.27040989009031474, "grad_norm": 1.0411492624491263, "learning_rate": 1.711924758562503e-05, "loss": 0.475, "step": 7590 }, { "epoch": 0.2704455172168089, "grad_norm": 1.6289101115504696, "learning_rate": 1.711843717954859e-05, "loss": 0.5551, "step": 7591 }, { "epoch": 0.270481144343303, "grad_norm": 1.8888284645969018, "learning_rate": 1.711762667868518e-05, "loss": 0.6053, "step": 7592 }, { "epoch": 0.2705167714697971, "grad_norm": 1.297151386623788, "learning_rate": 1.7116816083045603e-05, "loss": 0.4989, "step": 7593 }, { "epoch": 0.2705523985962912, "grad_norm": 1.7151848489234673, "learning_rate": 1.7116005392640646e-05, "loss": 0.6368, "step": 7594 }, { "epoch": 0.27058802572278534, "grad_norm": 1.7338394061739772, "learning_rate": 1.7115194607481095e-05, "loss": 0.6258, "step": 7595 }, { "epoch": 0.2706236528492794, "grad_norm": 1.856294609526379, "learning_rate": 1.711438372757776e-05, "loss": 0.7185, "step": 7596 }, { "epoch": 0.27065927997577355, "grad_norm": 1.4524102685121978, "learning_rate": 1.7113572752941436e-05, "loss": 0.5726, "step": 7597 }, { "epoch": 0.2706949071022677, "grad_norm": 1.8784680180786821, "learning_rate": 1.7112761683582914e-05, "loss": 0.5657, "step": 7598 }, { "epoch": 0.27073053422876175, "grad_norm": 1.2784460974077596, "learning_rate": 1.7111950519513e-05, "loss": 0.5235, "step": 7599 }, { "epoch": 0.2707661613552559, "grad_norm": 1.4002242816529724, "learning_rate": 1.7111139260742494e-05, "loss": 0.6093, "step": 7600 }, { "epoch": 0.27080178848175, "grad_norm": 1.253578615690583, "learning_rate": 1.71103279072822e-05, "loss": 0.4896, "step": 7601 }, { "epoch": 0.2708374156082441, "grad_norm": 1.1844645852882507, "learning_rate": 1.7109516459142916e-05, "loss": 0.5186, "step": 7602 }, { "epoch": 0.2708730427347382, "grad_norm": 1.3524858917021236, "learning_rate": 1.7108704916335454e-05, "loss": 0.5018, "step": 7603 }, { "epoch": 0.27090866986123235, "grad_norm": 1.5696715714566705, "learning_rate": 1.7107893278870616e-05, "loss": 0.5168, "step": 7604 }, { "epoch": 0.27094429698772643, "grad_norm": 1.438864661540073, "learning_rate": 1.710708154675921e-05, "loss": 0.4726, "step": 7605 }, { "epoch": 0.27097992411422056, "grad_norm": 1.2777171778827716, "learning_rate": 1.710626972001205e-05, "loss": 0.4964, "step": 7606 }, { "epoch": 0.2710155512407147, "grad_norm": 1.5966277874913588, "learning_rate": 1.710545779863994e-05, "loss": 0.6831, "step": 7607 }, { "epoch": 0.27105117836720877, "grad_norm": 1.0712492329590277, "learning_rate": 1.710464578265369e-05, "loss": 0.5007, "step": 7608 }, { "epoch": 0.2710868054937029, "grad_norm": 1.1361050418987226, "learning_rate": 1.7103833672064117e-05, "loss": 0.4834, "step": 7609 }, { "epoch": 0.27112243262019703, "grad_norm": 1.1615175182242434, "learning_rate": 1.7103021466882036e-05, "loss": 0.466, "step": 7610 }, { "epoch": 0.2711580597466911, "grad_norm": 1.205404072584097, "learning_rate": 1.7102209167118257e-05, "loss": 0.5396, "step": 7611 }, { "epoch": 0.27119368687318524, "grad_norm": 1.302431217610619, "learning_rate": 1.7101396772783595e-05, "loss": 0.5106, "step": 7612 }, { "epoch": 0.27122931399967937, "grad_norm": 1.241259537420175, "learning_rate": 1.7100584283888876e-05, "loss": 0.4859, "step": 7613 }, { "epoch": 0.27126494112617344, "grad_norm": 1.691543175261082, "learning_rate": 1.7099771700444916e-05, "loss": 0.5193, "step": 7614 }, { "epoch": 0.2713005682526676, "grad_norm": 1.1423524191815086, "learning_rate": 1.709895902246253e-05, "loss": 0.4724, "step": 7615 }, { "epoch": 0.2713361953791617, "grad_norm": 1.1321173750196443, "learning_rate": 1.7098146249952545e-05, "loss": 0.4839, "step": 7616 }, { "epoch": 0.2713718225056558, "grad_norm": 1.2479560196901502, "learning_rate": 1.709733338292578e-05, "loss": 0.5209, "step": 7617 }, { "epoch": 0.2714074496321499, "grad_norm": 2.244986511500244, "learning_rate": 1.709652042139306e-05, "loss": 0.5453, "step": 7618 }, { "epoch": 0.27144307675864404, "grad_norm": 1.2834481772255941, "learning_rate": 1.709570736536521e-05, "loss": 0.5026, "step": 7619 }, { "epoch": 0.2714787038851381, "grad_norm": 1.8475406014776323, "learning_rate": 1.709489421485306e-05, "loss": 0.6367, "step": 7620 }, { "epoch": 0.27151433101163225, "grad_norm": 1.1076054106651168, "learning_rate": 1.709408096986743e-05, "loss": 0.4541, "step": 7621 }, { "epoch": 0.2715499581381264, "grad_norm": 1.16735066177669, "learning_rate": 1.7093267630419156e-05, "loss": 0.4992, "step": 7622 }, { "epoch": 0.27158558526462045, "grad_norm": 1.2667248337664223, "learning_rate": 1.7092454196519064e-05, "loss": 0.4354, "step": 7623 }, { "epoch": 0.2716212123911146, "grad_norm": 1.2243889458909818, "learning_rate": 1.709164066817799e-05, "loss": 0.5161, "step": 7624 }, { "epoch": 0.2716568395176087, "grad_norm": 2.3338970094432496, "learning_rate": 1.7090827045406763e-05, "loss": 0.6212, "step": 7625 }, { "epoch": 0.2716924666441028, "grad_norm": 1.2441731172871808, "learning_rate": 1.7090013328216214e-05, "loss": 0.4834, "step": 7626 }, { "epoch": 0.2717280937705969, "grad_norm": 1.206959855066639, "learning_rate": 1.7089199516617188e-05, "loss": 0.5504, "step": 7627 }, { "epoch": 0.27176372089709105, "grad_norm": 1.5737928104555758, "learning_rate": 1.7088385610620508e-05, "loss": 0.6383, "step": 7628 }, { "epoch": 0.27179934802358513, "grad_norm": 1.4699789987494232, "learning_rate": 1.7087571610237026e-05, "loss": 0.5719, "step": 7629 }, { "epoch": 0.27183497515007926, "grad_norm": 2.064771495145675, "learning_rate": 1.708675751547757e-05, "loss": 0.5318, "step": 7630 }, { "epoch": 0.2718706022765734, "grad_norm": 1.2755244821321639, "learning_rate": 1.7085943326352988e-05, "loss": 0.5221, "step": 7631 }, { "epoch": 0.2719062294030675, "grad_norm": 1.7687430180117465, "learning_rate": 1.708512904287411e-05, "loss": 0.701, "step": 7632 }, { "epoch": 0.2719418565295616, "grad_norm": 1.076537797475349, "learning_rate": 1.7084314665051795e-05, "loss": 0.4692, "step": 7633 }, { "epoch": 0.27197748365605573, "grad_norm": 1.5869609141331658, "learning_rate": 1.7083500192896876e-05, "loss": 0.6056, "step": 7634 }, { "epoch": 0.27201311078254986, "grad_norm": 1.2367078955058646, "learning_rate": 1.70826856264202e-05, "loss": 0.5256, "step": 7635 }, { "epoch": 0.27204873790904394, "grad_norm": 1.1714418016161694, "learning_rate": 1.7081870965632618e-05, "loss": 0.478, "step": 7636 }, { "epoch": 0.27208436503553807, "grad_norm": 1.4310687742024175, "learning_rate": 1.7081056210544968e-05, "loss": 0.5376, "step": 7637 }, { "epoch": 0.2721199921620322, "grad_norm": 1.3819414556608516, "learning_rate": 1.7080241361168108e-05, "loss": 0.619, "step": 7638 }, { "epoch": 0.2721556192885263, "grad_norm": 1.0977033803412233, "learning_rate": 1.7079426417512886e-05, "loss": 0.5226, "step": 7639 }, { "epoch": 0.2721912464150204, "grad_norm": 1.1235288239721255, "learning_rate": 1.7078611379590154e-05, "loss": 0.4715, "step": 7640 }, { "epoch": 0.27222687354151454, "grad_norm": 1.7038773111433876, "learning_rate": 1.707779624741076e-05, "loss": 0.633, "step": 7641 }, { "epoch": 0.2722625006680086, "grad_norm": 2.094360474913865, "learning_rate": 1.7076981020985567e-05, "loss": 0.5885, "step": 7642 }, { "epoch": 0.27229812779450274, "grad_norm": 1.7809123578220083, "learning_rate": 1.707616570032542e-05, "loss": 0.6061, "step": 7643 }, { "epoch": 0.2723337549209969, "grad_norm": 1.2801327305124208, "learning_rate": 1.7075350285441185e-05, "loss": 0.5111, "step": 7644 }, { "epoch": 0.27236938204749095, "grad_norm": 2.058679045054369, "learning_rate": 1.7074534776343715e-05, "loss": 0.5629, "step": 7645 }, { "epoch": 0.2724050091739851, "grad_norm": 1.2425287120328252, "learning_rate": 1.7073719173043865e-05, "loss": 0.513, "step": 7646 }, { "epoch": 0.2724406363004792, "grad_norm": 1.920926255270034, "learning_rate": 1.7072903475552505e-05, "loss": 0.5456, "step": 7647 }, { "epoch": 0.2724762634269733, "grad_norm": 1.3767129453626237, "learning_rate": 1.707208768388049e-05, "loss": 0.5862, "step": 7648 }, { "epoch": 0.2725118905534674, "grad_norm": 1.1509564453373347, "learning_rate": 1.7071271798038686e-05, "loss": 0.4806, "step": 7649 }, { "epoch": 0.27254751767996155, "grad_norm": 1.458932644841446, "learning_rate": 1.7070455818037955e-05, "loss": 0.5652, "step": 7650 }, { "epoch": 0.2725831448064556, "grad_norm": 1.1607220019853628, "learning_rate": 1.706963974388916e-05, "loss": 0.541, "step": 7651 }, { "epoch": 0.27261877193294975, "grad_norm": 1.7422884332098254, "learning_rate": 1.7068823575603172e-05, "loss": 0.5596, "step": 7652 }, { "epoch": 0.2726543990594439, "grad_norm": 1.060848998931936, "learning_rate": 1.7068007313190858e-05, "loss": 0.457, "step": 7653 }, { "epoch": 0.27269002618593796, "grad_norm": 1.289675866890705, "learning_rate": 1.7067190956663084e-05, "loss": 0.5206, "step": 7654 }, { "epoch": 0.2727256533124321, "grad_norm": 1.1475462209974814, "learning_rate": 1.7066374506030726e-05, "loss": 0.4851, "step": 7655 }, { "epoch": 0.2727612804389262, "grad_norm": 1.4906626389017181, "learning_rate": 1.706555796130465e-05, "loss": 0.6375, "step": 7656 }, { "epoch": 0.2727969075654203, "grad_norm": 1.8411478939819683, "learning_rate": 1.706474132249573e-05, "loss": 0.6099, "step": 7657 }, { "epoch": 0.27283253469191443, "grad_norm": 1.1180236416033293, "learning_rate": 1.7063924589614847e-05, "loss": 0.5457, "step": 7658 }, { "epoch": 0.27286816181840856, "grad_norm": 1.5070114628295015, "learning_rate": 1.7063107762672865e-05, "loss": 0.5191, "step": 7659 }, { "epoch": 0.27290378894490264, "grad_norm": 5.625852506556573, "learning_rate": 1.7062290841680668e-05, "loss": 0.5611, "step": 7660 }, { "epoch": 0.27293941607139677, "grad_norm": 1.1103912180921338, "learning_rate": 1.7061473826649135e-05, "loss": 0.4803, "step": 7661 }, { "epoch": 0.2729750431978909, "grad_norm": 1.5393530115298595, "learning_rate": 1.7060656717589142e-05, "loss": 0.5424, "step": 7662 }, { "epoch": 0.273010670324385, "grad_norm": 1.2712719015367477, "learning_rate": 1.7059839514511565e-05, "loss": 0.5045, "step": 7663 }, { "epoch": 0.2730462974508791, "grad_norm": 1.1119346681596796, "learning_rate": 1.7059022217427293e-05, "loss": 0.48, "step": 7664 }, { "epoch": 0.27308192457737324, "grad_norm": 1.6746691133060345, "learning_rate": 1.705820482634721e-05, "loss": 0.5809, "step": 7665 }, { "epoch": 0.2731175517038673, "grad_norm": 1.1598471040787868, "learning_rate": 1.705738734128219e-05, "loss": 0.5078, "step": 7666 }, { "epoch": 0.27315317883036144, "grad_norm": 1.1012601979422652, "learning_rate": 1.7056569762243127e-05, "loss": 0.5232, "step": 7667 }, { "epoch": 0.2731888059568556, "grad_norm": 1.1194487492149887, "learning_rate": 1.7055752089240907e-05, "loss": 0.4875, "step": 7668 }, { "epoch": 0.27322443308334965, "grad_norm": 1.1894874694414035, "learning_rate": 1.7054934322286415e-05, "loss": 0.458, "step": 7669 }, { "epoch": 0.2732600602098438, "grad_norm": 1.4377996074420611, "learning_rate": 1.705411646139054e-05, "loss": 0.5069, "step": 7670 }, { "epoch": 0.2732956873363379, "grad_norm": 1.6814466252425078, "learning_rate": 1.7053298506564174e-05, "loss": 0.5403, "step": 7671 }, { "epoch": 0.273331314462832, "grad_norm": 1.7601322670180768, "learning_rate": 1.7052480457818213e-05, "loss": 0.6541, "step": 7672 }, { "epoch": 0.2733669415893261, "grad_norm": 1.2826314355822834, "learning_rate": 1.705166231516354e-05, "loss": 0.5203, "step": 7673 }, { "epoch": 0.27340256871582025, "grad_norm": 2.5789086235822154, "learning_rate": 1.7050844078611058e-05, "loss": 0.5491, "step": 7674 }, { "epoch": 0.2734381958423143, "grad_norm": 1.117694291582034, "learning_rate": 1.7050025748171655e-05, "loss": 0.5109, "step": 7675 }, { "epoch": 0.27347382296880846, "grad_norm": 1.0678362825599306, "learning_rate": 1.7049207323856232e-05, "loss": 0.5249, "step": 7676 }, { "epoch": 0.2735094500953026, "grad_norm": 1.3463230881179997, "learning_rate": 1.704838880567569e-05, "loss": 0.6662, "step": 7677 }, { "epoch": 0.27354507722179666, "grad_norm": 2.1805706391654773, "learning_rate": 1.7047570193640924e-05, "loss": 0.5116, "step": 7678 }, { "epoch": 0.2735807043482908, "grad_norm": 2.024361076321896, "learning_rate": 1.7046751487762835e-05, "loss": 0.6459, "step": 7679 }, { "epoch": 0.2736163314747849, "grad_norm": 1.040365547121049, "learning_rate": 1.704593268805232e-05, "loss": 0.5433, "step": 7680 }, { "epoch": 0.273651958601279, "grad_norm": 1.849336529476983, "learning_rate": 1.704511379452029e-05, "loss": 0.6288, "step": 7681 }, { "epoch": 0.27368758572777313, "grad_norm": 1.7647210181204005, "learning_rate": 1.7044294807177647e-05, "loss": 0.4715, "step": 7682 }, { "epoch": 0.27372321285426726, "grad_norm": 1.1853546487633977, "learning_rate": 1.704347572603529e-05, "loss": 0.4913, "step": 7683 }, { "epoch": 0.27375883998076134, "grad_norm": 1.2168644367161856, "learning_rate": 1.7042656551104137e-05, "loss": 0.505, "step": 7684 }, { "epoch": 0.27379446710725547, "grad_norm": 1.9709290760552074, "learning_rate": 1.7041837282395084e-05, "loss": 0.6214, "step": 7685 }, { "epoch": 0.2738300942337496, "grad_norm": 1.5555711811035824, "learning_rate": 1.7041017919919046e-05, "loss": 0.6882, "step": 7686 }, { "epoch": 0.2738657213602437, "grad_norm": 1.5009950668742555, "learning_rate": 1.7040198463686938e-05, "loss": 0.5493, "step": 7687 }, { "epoch": 0.2739013484867378, "grad_norm": 1.661426293456888, "learning_rate": 1.7039378913709664e-05, "loss": 0.6029, "step": 7688 }, { "epoch": 0.27393697561323194, "grad_norm": 1.189215553316187, "learning_rate": 1.703855926999814e-05, "loss": 0.494, "step": 7689 }, { "epoch": 0.273972602739726, "grad_norm": 1.572435145790013, "learning_rate": 1.7037739532563276e-05, "loss": 0.5622, "step": 7690 }, { "epoch": 0.27400822986622014, "grad_norm": 1.2818622948075584, "learning_rate": 1.7036919701415994e-05, "loss": 0.4977, "step": 7691 }, { "epoch": 0.2740438569927143, "grad_norm": 1.7084508497227802, "learning_rate": 1.7036099776567207e-05, "loss": 0.5712, "step": 7692 }, { "epoch": 0.27407948411920835, "grad_norm": 1.751312066129311, "learning_rate": 1.7035279758027835e-05, "loss": 0.6186, "step": 7693 }, { "epoch": 0.2741151112457025, "grad_norm": 1.3873381809284566, "learning_rate": 1.7034459645808794e-05, "loss": 0.5838, "step": 7694 }, { "epoch": 0.2741507383721966, "grad_norm": 1.1328460231750364, "learning_rate": 1.7033639439921007e-05, "loss": 0.5357, "step": 7695 }, { "epoch": 0.2741863654986907, "grad_norm": 1.5597201936200975, "learning_rate": 1.7032819140375395e-05, "loss": 0.6711, "step": 7696 }, { "epoch": 0.2742219926251848, "grad_norm": 1.6274352358113218, "learning_rate": 1.7031998747182877e-05, "loss": 0.5559, "step": 7697 }, { "epoch": 0.27425761975167895, "grad_norm": 1.7640175667677265, "learning_rate": 1.7031178260354388e-05, "loss": 0.6595, "step": 7698 }, { "epoch": 0.274293246878173, "grad_norm": 1.688254956904757, "learning_rate": 1.703035767990084e-05, "loss": 0.506, "step": 7699 }, { "epoch": 0.27432887400466716, "grad_norm": 1.3953238412359128, "learning_rate": 1.702953700583317e-05, "loss": 0.5556, "step": 7700 }, { "epoch": 0.2743645011311613, "grad_norm": 1.0991204575128704, "learning_rate": 1.7028716238162296e-05, "loss": 0.4593, "step": 7701 }, { "epoch": 0.27440012825765536, "grad_norm": 1.0843675323562454, "learning_rate": 1.7027895376899154e-05, "loss": 0.519, "step": 7702 }, { "epoch": 0.2744357553841495, "grad_norm": 1.104816126721023, "learning_rate": 1.702707442205467e-05, "loss": 0.5251, "step": 7703 }, { "epoch": 0.2744713825106436, "grad_norm": 1.1612745645203746, "learning_rate": 1.7026253373639784e-05, "loss": 0.5008, "step": 7704 }, { "epoch": 0.2745070096371377, "grad_norm": 1.706421858657263, "learning_rate": 1.702543223166542e-05, "loss": 0.6333, "step": 7705 }, { "epoch": 0.27454263676363183, "grad_norm": 1.1335357623135478, "learning_rate": 1.702461099614252e-05, "loss": 0.4881, "step": 7706 }, { "epoch": 0.27457826389012596, "grad_norm": 1.1955279676982784, "learning_rate": 1.7023789667082012e-05, "loss": 0.5077, "step": 7707 }, { "epoch": 0.27461389101662004, "grad_norm": 2.0640831871927108, "learning_rate": 1.7022968244494832e-05, "loss": 0.6143, "step": 7708 }, { "epoch": 0.27464951814311417, "grad_norm": 1.1083621198708913, "learning_rate": 1.7022146728391926e-05, "loss": 0.4821, "step": 7709 }, { "epoch": 0.2746851452696083, "grad_norm": 1.1535047252539719, "learning_rate": 1.7021325118784225e-05, "loss": 0.489, "step": 7710 }, { "epoch": 0.2747207723961024, "grad_norm": 1.818021887989936, "learning_rate": 1.7020503415682675e-05, "loss": 0.6894, "step": 7711 }, { "epoch": 0.2747563995225965, "grad_norm": 1.9810173554147648, "learning_rate": 1.701968161909821e-05, "loss": 0.6111, "step": 7712 }, { "epoch": 0.27479202664909064, "grad_norm": 1.6418704687908598, "learning_rate": 1.7018859729041785e-05, "loss": 0.6648, "step": 7713 }, { "epoch": 0.2748276537755847, "grad_norm": 1.2793869064137122, "learning_rate": 1.701803774552433e-05, "loss": 0.5597, "step": 7714 }, { "epoch": 0.27486328090207884, "grad_norm": 1.0978093416191017, "learning_rate": 1.70172156685568e-05, "loss": 0.4899, "step": 7715 }, { "epoch": 0.274898908028573, "grad_norm": 1.5775332697514397, "learning_rate": 1.7016393498150135e-05, "loss": 0.5652, "step": 7716 }, { "epoch": 0.27493453515506705, "grad_norm": 1.4657748883149788, "learning_rate": 1.7015571234315294e-05, "loss": 0.5019, "step": 7717 }, { "epoch": 0.2749701622815612, "grad_norm": 1.2412179902319325, "learning_rate": 1.7014748877063212e-05, "loss": 0.5423, "step": 7718 }, { "epoch": 0.2750057894080553, "grad_norm": 1.0249039223955287, "learning_rate": 1.7013926426404854e-05, "loss": 0.49, "step": 7719 }, { "epoch": 0.2750414165345494, "grad_norm": 1.0795711167765347, "learning_rate": 1.701310388235115e-05, "loss": 0.4559, "step": 7720 }, { "epoch": 0.2750770436610435, "grad_norm": 1.112932822917826, "learning_rate": 1.7012281244913076e-05, "loss": 0.4975, "step": 7721 }, { "epoch": 0.27511267078753765, "grad_norm": 1.1247649708305711, "learning_rate": 1.7011458514101575e-05, "loss": 0.4473, "step": 7722 }, { "epoch": 0.2751482979140317, "grad_norm": 1.134143315822162, "learning_rate": 1.7010635689927602e-05, "loss": 0.4666, "step": 7723 }, { "epoch": 0.27518392504052586, "grad_norm": 1.3230396205324453, "learning_rate": 1.7009812772402115e-05, "loss": 0.6121, "step": 7724 }, { "epoch": 0.27521955216702, "grad_norm": 1.0911807357456762, "learning_rate": 1.7008989761536067e-05, "loss": 0.49, "step": 7725 }, { "epoch": 0.27525517929351406, "grad_norm": 1.2564459373191865, "learning_rate": 1.7008166657340425e-05, "loss": 0.4759, "step": 7726 }, { "epoch": 0.2752908064200082, "grad_norm": 1.7849638310174545, "learning_rate": 1.7007343459826147e-05, "loss": 0.5547, "step": 7727 }, { "epoch": 0.2753264335465023, "grad_norm": 1.1740392266781878, "learning_rate": 1.700652016900419e-05, "loss": 0.4899, "step": 7728 }, { "epoch": 0.2753620606729964, "grad_norm": 1.697328082000088, "learning_rate": 1.7005696784885518e-05, "loss": 0.5491, "step": 7729 }, { "epoch": 0.27539768779949053, "grad_norm": 1.1575747009335264, "learning_rate": 1.70048733074811e-05, "loss": 0.5285, "step": 7730 }, { "epoch": 0.27543331492598466, "grad_norm": 1.145430349109805, "learning_rate": 1.7004049736801898e-05, "loss": 0.5428, "step": 7731 }, { "epoch": 0.27546894205247874, "grad_norm": 1.1929808854991415, "learning_rate": 1.700322607285888e-05, "loss": 0.5408, "step": 7732 }, { "epoch": 0.27550456917897287, "grad_norm": 1.5594592555721227, "learning_rate": 1.700240231566301e-05, "loss": 0.6934, "step": 7733 }, { "epoch": 0.275540196305467, "grad_norm": 1.1528688044893103, "learning_rate": 1.700157846522526e-05, "loss": 0.4411, "step": 7734 }, { "epoch": 0.2755758234319611, "grad_norm": 1.246881273438847, "learning_rate": 1.7000754521556596e-05, "loss": 0.5191, "step": 7735 }, { "epoch": 0.2756114505584552, "grad_norm": 1.8732212669942951, "learning_rate": 1.6999930484667992e-05, "loss": 0.5932, "step": 7736 }, { "epoch": 0.27564707768494934, "grad_norm": 1.834520711084755, "learning_rate": 1.6999106354570422e-05, "loss": 0.6344, "step": 7737 }, { "epoch": 0.2756827048114434, "grad_norm": 1.0984715249071262, "learning_rate": 1.699828213127486e-05, "loss": 0.4675, "step": 7738 }, { "epoch": 0.27571833193793754, "grad_norm": 1.561380598387745, "learning_rate": 1.6997457814792278e-05, "loss": 0.5973, "step": 7739 }, { "epoch": 0.2757539590644317, "grad_norm": 1.3025833895432484, "learning_rate": 1.6996633405133656e-05, "loss": 0.5973, "step": 7740 }, { "epoch": 0.27578958619092575, "grad_norm": 1.5922575265306869, "learning_rate": 1.6995808902309968e-05, "loss": 0.6717, "step": 7741 }, { "epoch": 0.2758252133174199, "grad_norm": 1.648987620070914, "learning_rate": 1.6994984306332195e-05, "loss": 0.5297, "step": 7742 }, { "epoch": 0.275860840443914, "grad_norm": 1.5312643027797714, "learning_rate": 1.6994159617211318e-05, "loss": 0.5702, "step": 7743 }, { "epoch": 0.2758964675704081, "grad_norm": 1.116923639738717, "learning_rate": 1.6993334834958318e-05, "loss": 0.4668, "step": 7744 }, { "epoch": 0.2759320946969022, "grad_norm": 1.7712774426052402, "learning_rate": 1.6992509959584176e-05, "loss": 0.571, "step": 7745 }, { "epoch": 0.27596772182339635, "grad_norm": 2.9173760537717848, "learning_rate": 1.6991684991099873e-05, "loss": 0.558, "step": 7746 }, { "epoch": 0.2760033489498904, "grad_norm": 1.686890283182216, "learning_rate": 1.6990859929516404e-05, "loss": 0.6223, "step": 7747 }, { "epoch": 0.27603897607638456, "grad_norm": 2.468404931928073, "learning_rate": 1.6990034774844745e-05, "loss": 0.537, "step": 7748 }, { "epoch": 0.2760746032028787, "grad_norm": 1.3150899731859211, "learning_rate": 1.6989209527095885e-05, "loss": 0.5146, "step": 7749 }, { "epoch": 0.27611023032937276, "grad_norm": 1.2259142086352954, "learning_rate": 1.698838418628082e-05, "loss": 0.5293, "step": 7750 }, { "epoch": 0.2761458574558669, "grad_norm": 1.1125231780220457, "learning_rate": 1.698755875241053e-05, "loss": 0.5095, "step": 7751 }, { "epoch": 0.276181484582361, "grad_norm": 1.077396601425014, "learning_rate": 1.698673322549601e-05, "loss": 0.4787, "step": 7752 }, { "epoch": 0.2762171117088551, "grad_norm": 1.1414900787457032, "learning_rate": 1.698590760554826e-05, "loss": 0.5236, "step": 7753 }, { "epoch": 0.27625273883534923, "grad_norm": 1.558779979239656, "learning_rate": 1.6985081892578264e-05, "loss": 0.5, "step": 7754 }, { "epoch": 0.27628836596184336, "grad_norm": 1.1094737492023237, "learning_rate": 1.6984256086597022e-05, "loss": 0.4957, "step": 7755 }, { "epoch": 0.27632399308833744, "grad_norm": 1.0739328565572894, "learning_rate": 1.6983430187615527e-05, "loss": 0.4916, "step": 7756 }, { "epoch": 0.27635962021483157, "grad_norm": 1.1593518400142522, "learning_rate": 1.698260419564478e-05, "loss": 0.5404, "step": 7757 }, { "epoch": 0.2763952473413257, "grad_norm": 1.1150674348958949, "learning_rate": 1.6981778110695775e-05, "loss": 0.4696, "step": 7758 }, { "epoch": 0.2764308744678198, "grad_norm": 1.0266177924769135, "learning_rate": 1.6980951932779514e-05, "loss": 0.5295, "step": 7759 }, { "epoch": 0.2764665015943139, "grad_norm": 1.0785714570124592, "learning_rate": 1.6980125661907e-05, "loss": 0.4707, "step": 7760 }, { "epoch": 0.27650212872080804, "grad_norm": 1.0748806239633273, "learning_rate": 1.6979299298089237e-05, "loss": 0.5146, "step": 7761 }, { "epoch": 0.2765377558473021, "grad_norm": 1.1633266026874005, "learning_rate": 1.697847284133722e-05, "loss": 0.4801, "step": 7762 }, { "epoch": 0.27657338297379624, "grad_norm": 1.6436719821287569, "learning_rate": 1.697764629166196e-05, "loss": 0.5787, "step": 7763 }, { "epoch": 0.2766090101002904, "grad_norm": 1.1438326758502377, "learning_rate": 1.6976819649074465e-05, "loss": 0.495, "step": 7764 }, { "epoch": 0.27664463722678445, "grad_norm": 2.5908034927854997, "learning_rate": 1.697599291358574e-05, "loss": 0.6279, "step": 7765 }, { "epoch": 0.2766802643532786, "grad_norm": 1.2003818236096844, "learning_rate": 1.697516608520679e-05, "loss": 0.5243, "step": 7766 }, { "epoch": 0.2767158914797727, "grad_norm": 1.8257881348598484, "learning_rate": 1.6974339163948636e-05, "loss": 0.5094, "step": 7767 }, { "epoch": 0.2767515186062668, "grad_norm": 1.1374512466132425, "learning_rate": 1.6973512149822272e-05, "loss": 0.4967, "step": 7768 }, { "epoch": 0.2767871457327609, "grad_norm": 1.0644084332359807, "learning_rate": 1.6972685042838725e-05, "loss": 0.4332, "step": 7769 }, { "epoch": 0.27682277285925505, "grad_norm": 1.5905267195970563, "learning_rate": 1.6971857843009e-05, "loss": 0.5874, "step": 7770 }, { "epoch": 0.2768583999857491, "grad_norm": 1.2255158633491838, "learning_rate": 1.6971030550344122e-05, "loss": 0.548, "step": 7771 }, { "epoch": 0.27689402711224326, "grad_norm": 1.4087341272527156, "learning_rate": 1.6970203164855092e-05, "loss": 0.5882, "step": 7772 }, { "epoch": 0.2769296542387374, "grad_norm": 1.4958539273741243, "learning_rate": 1.696937568655294e-05, "loss": 0.6901, "step": 7773 }, { "epoch": 0.27696528136523146, "grad_norm": 1.0905712034048207, "learning_rate": 1.696854811544867e-05, "loss": 0.5063, "step": 7774 }, { "epoch": 0.2770009084917256, "grad_norm": 1.6730479163986842, "learning_rate": 1.696772045155332e-05, "loss": 0.6013, "step": 7775 }, { "epoch": 0.2770365356182197, "grad_norm": 1.0731026264787686, "learning_rate": 1.69668926948779e-05, "loss": 0.4483, "step": 7776 }, { "epoch": 0.2770721627447138, "grad_norm": 1.9778779755854061, "learning_rate": 1.6966064845433436e-05, "loss": 0.6864, "step": 7777 }, { "epoch": 0.27710778987120793, "grad_norm": 1.2626593009165727, "learning_rate": 1.6965236903230947e-05, "loss": 0.5694, "step": 7778 }, { "epoch": 0.27714341699770206, "grad_norm": 1.1992858960790955, "learning_rate": 1.696440886828146e-05, "loss": 0.5267, "step": 7779 }, { "epoch": 0.27717904412419614, "grad_norm": 1.164571755874562, "learning_rate": 1.6963580740596003e-05, "loss": 0.5028, "step": 7780 }, { "epoch": 0.27721467125069027, "grad_norm": 1.1133390263507636, "learning_rate": 1.6962752520185598e-05, "loss": 0.4593, "step": 7781 }, { "epoch": 0.2772502983771844, "grad_norm": 1.106124590811905, "learning_rate": 1.6961924207061283e-05, "loss": 0.503, "step": 7782 }, { "epoch": 0.2772859255036785, "grad_norm": 1.6499362504626887, "learning_rate": 1.6961095801234075e-05, "loss": 0.5075, "step": 7783 }, { "epoch": 0.2773215526301726, "grad_norm": 1.1310916097381276, "learning_rate": 1.6960267302715014e-05, "loss": 0.5299, "step": 7784 }, { "epoch": 0.27735717975666674, "grad_norm": 1.9184689397750199, "learning_rate": 1.6959438711515126e-05, "loss": 0.6025, "step": 7785 }, { "epoch": 0.2773928068831608, "grad_norm": 1.013366388717616, "learning_rate": 1.695861002764545e-05, "loss": 0.4972, "step": 7786 }, { "epoch": 0.27742843400965495, "grad_norm": 1.5696366282129977, "learning_rate": 1.6957781251117012e-05, "loss": 0.5526, "step": 7787 }, { "epoch": 0.2774640611361491, "grad_norm": 1.1304638362598949, "learning_rate": 1.695695238194086e-05, "loss": 0.5125, "step": 7788 }, { "epoch": 0.27749968826264315, "grad_norm": 1.1930396341712863, "learning_rate": 1.6956123420128025e-05, "loss": 0.6, "step": 7789 }, { "epoch": 0.2775353153891373, "grad_norm": 1.1234108836785253, "learning_rate": 1.695529436568954e-05, "loss": 0.5569, "step": 7790 }, { "epoch": 0.2775709425156314, "grad_norm": 1.5899127469886698, "learning_rate": 1.6954465218636452e-05, "loss": 0.5228, "step": 7791 }, { "epoch": 0.2776065696421255, "grad_norm": 1.4180590969395208, "learning_rate": 1.6953635978979798e-05, "loss": 0.5828, "step": 7792 }, { "epoch": 0.2776421967686196, "grad_norm": 1.1188624842422403, "learning_rate": 1.695280664673062e-05, "loss": 0.5306, "step": 7793 }, { "epoch": 0.27767782389511375, "grad_norm": 1.4788867616387114, "learning_rate": 1.6951977221899962e-05, "loss": 0.5982, "step": 7794 }, { "epoch": 0.2777134510216078, "grad_norm": 1.1306935154651159, "learning_rate": 1.695114770449887e-05, "loss": 0.4822, "step": 7795 }, { "epoch": 0.27774907814810196, "grad_norm": 1.0592612826945158, "learning_rate": 1.6950318094538386e-05, "loss": 0.4478, "step": 7796 }, { "epoch": 0.2777847052745961, "grad_norm": 1.188274033206082, "learning_rate": 1.6949488392029557e-05, "loss": 0.4779, "step": 7797 }, { "epoch": 0.27782033240109016, "grad_norm": 1.7639265943386326, "learning_rate": 1.6948658596983435e-05, "loss": 0.6497, "step": 7798 }, { "epoch": 0.2778559595275843, "grad_norm": 1.290112028050743, "learning_rate": 1.6947828709411066e-05, "loss": 0.5146, "step": 7799 }, { "epoch": 0.2778915866540784, "grad_norm": 1.2581903752714665, "learning_rate": 1.69469987293235e-05, "loss": 0.5014, "step": 7800 }, { "epoch": 0.2779272137805725, "grad_norm": 1.7036761485548635, "learning_rate": 1.6946168656731793e-05, "loss": 0.571, "step": 7801 }, { "epoch": 0.27796284090706663, "grad_norm": 1.2090464941879375, "learning_rate": 1.6945338491646995e-05, "loss": 0.5165, "step": 7802 }, { "epoch": 0.27799846803356076, "grad_norm": 1.6399331181053503, "learning_rate": 1.694450823408016e-05, "loss": 0.5488, "step": 7803 }, { "epoch": 0.27803409516005484, "grad_norm": 1.1616952010316215, "learning_rate": 1.6943677884042343e-05, "loss": 0.4742, "step": 7804 }, { "epoch": 0.27806972228654897, "grad_norm": 1.9644524286460214, "learning_rate": 1.69428474415446e-05, "loss": 0.5908, "step": 7805 }, { "epoch": 0.2781053494130431, "grad_norm": 1.2095563098156108, "learning_rate": 1.6942016906597997e-05, "loss": 0.4697, "step": 7806 }, { "epoch": 0.2781409765395372, "grad_norm": 3.018455711177002, "learning_rate": 1.694118627921358e-05, "loss": 0.6205, "step": 7807 }, { "epoch": 0.2781766036660313, "grad_norm": 1.284812084827029, "learning_rate": 1.694035555940242e-05, "loss": 0.5222, "step": 7808 }, { "epoch": 0.27821223079252544, "grad_norm": 1.6403157280050047, "learning_rate": 1.6939524747175574e-05, "loss": 0.4905, "step": 7809 }, { "epoch": 0.2782478579190195, "grad_norm": 2.0952068482358417, "learning_rate": 1.6938693842544105e-05, "loss": 0.6438, "step": 7810 }, { "epoch": 0.27828348504551365, "grad_norm": 1.187000149476109, "learning_rate": 1.6937862845519078e-05, "loss": 0.5441, "step": 7811 }, { "epoch": 0.2783191121720078, "grad_norm": 1.139345208289, "learning_rate": 1.693703175611156e-05, "loss": 0.5422, "step": 7812 }, { "epoch": 0.27835473929850185, "grad_norm": 1.6632454101116292, "learning_rate": 1.693620057433261e-05, "loss": 0.5101, "step": 7813 }, { "epoch": 0.278390366424996, "grad_norm": 1.1519836783357706, "learning_rate": 1.6935369300193304e-05, "loss": 0.4976, "step": 7814 }, { "epoch": 0.2784259935514901, "grad_norm": 1.0477860549365938, "learning_rate": 1.693453793370471e-05, "loss": 0.5078, "step": 7815 }, { "epoch": 0.2784616206779842, "grad_norm": 1.739438203862213, "learning_rate": 1.6933706474877895e-05, "loss": 0.6098, "step": 7816 }, { "epoch": 0.2784972478044783, "grad_norm": 1.5413465386790741, "learning_rate": 1.693287492372393e-05, "loss": 0.5863, "step": 7817 }, { "epoch": 0.27853287493097245, "grad_norm": 1.0822619232342545, "learning_rate": 1.6932043280253892e-05, "loss": 0.5171, "step": 7818 }, { "epoch": 0.2785685020574666, "grad_norm": 1.2201673005052271, "learning_rate": 1.6931211544478855e-05, "loss": 0.4726, "step": 7819 }, { "epoch": 0.27860412918396066, "grad_norm": 1.3235118040695781, "learning_rate": 1.6930379716409885e-05, "loss": 0.5557, "step": 7820 }, { "epoch": 0.2786397563104548, "grad_norm": 1.8065024634874258, "learning_rate": 1.692954779605807e-05, "loss": 0.5697, "step": 7821 }, { "epoch": 0.2786753834369489, "grad_norm": 1.2011297649098143, "learning_rate": 1.6928715783434482e-05, "loss": 0.5353, "step": 7822 }, { "epoch": 0.278711010563443, "grad_norm": 1.1308271358308475, "learning_rate": 1.6927883678550197e-05, "loss": 0.484, "step": 7823 }, { "epoch": 0.2787466376899371, "grad_norm": 1.7528156988737793, "learning_rate": 1.6927051481416304e-05, "loss": 0.5477, "step": 7824 }, { "epoch": 0.27878226481643126, "grad_norm": 1.7310325248953287, "learning_rate": 1.6926219192043873e-05, "loss": 0.6264, "step": 7825 }, { "epoch": 0.27881789194292533, "grad_norm": 1.477278114887577, "learning_rate": 1.6925386810443994e-05, "loss": 0.5152, "step": 7826 }, { "epoch": 0.27885351906941946, "grad_norm": 1.2194411485491723, "learning_rate": 1.6924554336627754e-05, "loss": 0.5453, "step": 7827 }, { "epoch": 0.2788891461959136, "grad_norm": 1.5417992520866808, "learning_rate": 1.692372177060623e-05, "loss": 0.5543, "step": 7828 }, { "epoch": 0.27892477332240767, "grad_norm": 1.480569602164381, "learning_rate": 1.692288911239051e-05, "loss": 0.6503, "step": 7829 }, { "epoch": 0.2789604004489018, "grad_norm": 1.073019812671162, "learning_rate": 1.6922056361991685e-05, "loss": 0.4616, "step": 7830 }, { "epoch": 0.27899602757539593, "grad_norm": 1.2926179610756932, "learning_rate": 1.692122351942084e-05, "loss": 0.4935, "step": 7831 }, { "epoch": 0.27903165470189, "grad_norm": 1.1658305826810917, "learning_rate": 1.6920390584689063e-05, "loss": 0.4485, "step": 7832 }, { "epoch": 0.27906728182838414, "grad_norm": 1.5347298389331294, "learning_rate": 1.6919557557807454e-05, "loss": 0.6356, "step": 7833 }, { "epoch": 0.27910290895487827, "grad_norm": 1.2984996878852322, "learning_rate": 1.6918724438787095e-05, "loss": 0.5454, "step": 7834 }, { "epoch": 0.27913853608137235, "grad_norm": 1.1094354806526883, "learning_rate": 1.6917891227639086e-05, "loss": 0.4731, "step": 7835 }, { "epoch": 0.2791741632078665, "grad_norm": 1.1857646533536337, "learning_rate": 1.691705792437452e-05, "loss": 0.5208, "step": 7836 }, { "epoch": 0.2792097903343606, "grad_norm": 1.069410242684284, "learning_rate": 1.691622452900449e-05, "loss": 0.4925, "step": 7837 }, { "epoch": 0.2792454174608547, "grad_norm": 1.1912519267256296, "learning_rate": 1.69153910415401e-05, "loss": 0.5149, "step": 7838 }, { "epoch": 0.2792810445873488, "grad_norm": 1.2312924556075509, "learning_rate": 1.6914557461992445e-05, "loss": 0.4923, "step": 7839 }, { "epoch": 0.27931667171384295, "grad_norm": 1.0912191698952487, "learning_rate": 1.6913723790372626e-05, "loss": 0.4928, "step": 7840 }, { "epoch": 0.279352298840337, "grad_norm": 1.0485965506108696, "learning_rate": 1.691289002669174e-05, "loss": 0.4754, "step": 7841 }, { "epoch": 0.27938792596683115, "grad_norm": 1.1441500388763477, "learning_rate": 1.691205617096089e-05, "loss": 0.5661, "step": 7842 }, { "epoch": 0.2794235530933253, "grad_norm": 1.7452035717515761, "learning_rate": 1.691122222319118e-05, "loss": 0.6591, "step": 7843 }, { "epoch": 0.27945918021981936, "grad_norm": 1.4830705638827395, "learning_rate": 1.691038818339372e-05, "loss": 0.5941, "step": 7844 }, { "epoch": 0.2794948073463135, "grad_norm": 1.766696719116965, "learning_rate": 1.690955405157961e-05, "loss": 0.6239, "step": 7845 }, { "epoch": 0.2795304344728076, "grad_norm": 1.2083187742804922, "learning_rate": 1.6908719827759954e-05, "loss": 0.5019, "step": 7846 }, { "epoch": 0.2795660615993017, "grad_norm": 1.1289590253193493, "learning_rate": 1.690788551194587e-05, "loss": 0.5001, "step": 7847 }, { "epoch": 0.2796016887257958, "grad_norm": 1.8014253249871455, "learning_rate": 1.6907051104148456e-05, "loss": 0.5863, "step": 7848 }, { "epoch": 0.27963731585228996, "grad_norm": 1.2242752698494084, "learning_rate": 1.6906216604378836e-05, "loss": 0.4989, "step": 7849 }, { "epoch": 0.27967294297878403, "grad_norm": 1.4287206150712537, "learning_rate": 1.690538201264811e-05, "loss": 0.5824, "step": 7850 }, { "epoch": 0.27970857010527816, "grad_norm": 2.617135271011798, "learning_rate": 1.69045473289674e-05, "loss": 0.5758, "step": 7851 }, { "epoch": 0.2797441972317723, "grad_norm": 1.5987772974187089, "learning_rate": 1.6903712553347812e-05, "loss": 0.6457, "step": 7852 }, { "epoch": 0.27977982435826637, "grad_norm": 1.9446995599014822, "learning_rate": 1.6902877685800467e-05, "loss": 0.5414, "step": 7853 }, { "epoch": 0.2798154514847605, "grad_norm": 1.1379990683363232, "learning_rate": 1.690204272633648e-05, "loss": 0.5118, "step": 7854 }, { "epoch": 0.27985107861125463, "grad_norm": 1.6694300651623974, "learning_rate": 1.6901207674966972e-05, "loss": 0.6008, "step": 7855 }, { "epoch": 0.2798867057377487, "grad_norm": 1.1678426944125706, "learning_rate": 1.690037253170306e-05, "loss": 0.5068, "step": 7856 }, { "epoch": 0.27992233286424284, "grad_norm": 1.1919654818106646, "learning_rate": 1.6899537296555865e-05, "loss": 0.5458, "step": 7857 }, { "epoch": 0.27995795999073697, "grad_norm": 1.3770319863623979, "learning_rate": 1.6898701969536508e-05, "loss": 0.5436, "step": 7858 }, { "epoch": 0.27999358711723105, "grad_norm": 1.0983030402909955, "learning_rate": 1.689786655065611e-05, "loss": 0.4735, "step": 7859 }, { "epoch": 0.2800292142437252, "grad_norm": 1.094537922358936, "learning_rate": 1.68970310399258e-05, "loss": 0.5216, "step": 7860 }, { "epoch": 0.2800648413702193, "grad_norm": 1.2542533546633596, "learning_rate": 1.68961954373567e-05, "loss": 0.4846, "step": 7861 }, { "epoch": 0.2801004684967134, "grad_norm": 1.3580958957792981, "learning_rate": 1.689535974295994e-05, "loss": 0.4899, "step": 7862 }, { "epoch": 0.2801360956232075, "grad_norm": 1.0870189584611996, "learning_rate": 1.689452395674664e-05, "loss": 0.5157, "step": 7863 }, { "epoch": 0.28017172274970165, "grad_norm": 1.0451180818894816, "learning_rate": 1.689368807872794e-05, "loss": 0.5103, "step": 7864 }, { "epoch": 0.2802073498761957, "grad_norm": 1.8374083091880473, "learning_rate": 1.6892852108914963e-05, "loss": 0.6373, "step": 7865 }, { "epoch": 0.28024297700268985, "grad_norm": 1.0512648889213987, "learning_rate": 1.689201604731884e-05, "loss": 0.4144, "step": 7866 }, { "epoch": 0.280278604129184, "grad_norm": 1.7630115894131002, "learning_rate": 1.689117989395071e-05, "loss": 0.6717, "step": 7867 }, { "epoch": 0.28031423125567806, "grad_norm": 1.3417646660972429, "learning_rate": 1.68903436488217e-05, "loss": 0.4574, "step": 7868 }, { "epoch": 0.2803498583821722, "grad_norm": 1.8705024948616369, "learning_rate": 1.6889507311942946e-05, "loss": 0.6445, "step": 7869 }, { "epoch": 0.2803854855086663, "grad_norm": 1.1809268687198384, "learning_rate": 1.6888670883325593e-05, "loss": 0.5068, "step": 7870 }, { "epoch": 0.2804211126351604, "grad_norm": 1.4699861096391196, "learning_rate": 1.6887834362980768e-05, "loss": 0.5171, "step": 7871 }, { "epoch": 0.2804567397616545, "grad_norm": 2.3011591525801287, "learning_rate": 1.688699775091962e-05, "loss": 0.61, "step": 7872 }, { "epoch": 0.28049236688814866, "grad_norm": 1.161512234816839, "learning_rate": 1.6886161047153278e-05, "loss": 0.4742, "step": 7873 }, { "epoch": 0.28052799401464273, "grad_norm": 1.2978741619816752, "learning_rate": 1.6885324251692887e-05, "loss": 0.4904, "step": 7874 }, { "epoch": 0.28056362114113687, "grad_norm": 1.2187679181409867, "learning_rate": 1.6884487364549596e-05, "loss": 0.485, "step": 7875 }, { "epoch": 0.280599248267631, "grad_norm": 1.0645685483778544, "learning_rate": 1.688365038573454e-05, "loss": 0.4613, "step": 7876 }, { "epoch": 0.28063487539412507, "grad_norm": 1.575411847392629, "learning_rate": 1.688281331525887e-05, "loss": 0.5177, "step": 7877 }, { "epoch": 0.2806705025206192, "grad_norm": 1.7493181282599441, "learning_rate": 1.6881976153133724e-05, "loss": 0.5834, "step": 7878 }, { "epoch": 0.28070612964711333, "grad_norm": 1.37068209975381, "learning_rate": 1.6881138899370263e-05, "loss": 0.5768, "step": 7879 }, { "epoch": 0.2807417567736074, "grad_norm": 1.5872184873960546, "learning_rate": 1.6880301553979627e-05, "loss": 0.5292, "step": 7880 }, { "epoch": 0.28077738390010154, "grad_norm": 2.2882002299173743, "learning_rate": 1.687946411697296e-05, "loss": 0.5832, "step": 7881 }, { "epoch": 0.28081301102659567, "grad_norm": 1.0841058455293313, "learning_rate": 1.6878626588361427e-05, "loss": 0.5039, "step": 7882 }, { "epoch": 0.28084863815308975, "grad_norm": 1.847991582647197, "learning_rate": 1.6877788968156172e-05, "loss": 0.6058, "step": 7883 }, { "epoch": 0.2808842652795839, "grad_norm": 1.6554476469522452, "learning_rate": 1.6876951256368346e-05, "loss": 0.5525, "step": 7884 }, { "epoch": 0.280919892406078, "grad_norm": 1.1438070492582666, "learning_rate": 1.6876113453009112e-05, "loss": 0.479, "step": 7885 }, { "epoch": 0.2809555195325721, "grad_norm": 1.0762497965555986, "learning_rate": 1.6875275558089618e-05, "loss": 0.5364, "step": 7886 }, { "epoch": 0.2809911466590662, "grad_norm": 1.5846483071511963, "learning_rate": 1.6874437571621025e-05, "loss": 0.5656, "step": 7887 }, { "epoch": 0.28102677378556035, "grad_norm": 1.6646445105190066, "learning_rate": 1.6873599493614492e-05, "loss": 0.6229, "step": 7888 }, { "epoch": 0.2810624009120544, "grad_norm": 1.1409211733595224, "learning_rate": 1.6872761324081174e-05, "loss": 0.4933, "step": 7889 }, { "epoch": 0.28109802803854855, "grad_norm": 1.0842230798250803, "learning_rate": 1.6871923063032238e-05, "loss": 0.5192, "step": 7890 }, { "epoch": 0.2811336551650427, "grad_norm": 1.8318325299337885, "learning_rate": 1.6871084710478844e-05, "loss": 0.6567, "step": 7891 }, { "epoch": 0.28116928229153676, "grad_norm": 1.187771824398951, "learning_rate": 1.6870246266432153e-05, "loss": 0.5067, "step": 7892 }, { "epoch": 0.2812049094180309, "grad_norm": 1.1397668992133736, "learning_rate": 1.686940773090333e-05, "loss": 0.5227, "step": 7893 }, { "epoch": 0.281240536544525, "grad_norm": 1.2178283733211583, "learning_rate": 1.6868569103903544e-05, "loss": 0.5304, "step": 7894 }, { "epoch": 0.2812761636710191, "grad_norm": 1.5925741108990679, "learning_rate": 1.686773038544396e-05, "loss": 0.6066, "step": 7895 }, { "epoch": 0.28131179079751323, "grad_norm": 1.111493256994347, "learning_rate": 1.686689157553574e-05, "loss": 0.5125, "step": 7896 }, { "epoch": 0.28134741792400736, "grad_norm": 1.1636141136152556, "learning_rate": 1.6866052674190067e-05, "loss": 0.5109, "step": 7897 }, { "epoch": 0.28138304505050143, "grad_norm": 1.4668958465156732, "learning_rate": 1.68652136814181e-05, "loss": 0.6536, "step": 7898 }, { "epoch": 0.28141867217699557, "grad_norm": 1.158130051370541, "learning_rate": 1.686437459723101e-05, "loss": 0.5226, "step": 7899 }, { "epoch": 0.2814542993034897, "grad_norm": 1.6243476623586839, "learning_rate": 1.6863535421639982e-05, "loss": 0.6671, "step": 7900 }, { "epoch": 0.28148992642998377, "grad_norm": 1.181002288024327, "learning_rate": 1.6862696154656178e-05, "loss": 0.5378, "step": 7901 }, { "epoch": 0.2815255535564779, "grad_norm": 1.7583137869290455, "learning_rate": 1.686185679629078e-05, "loss": 0.5817, "step": 7902 }, { "epoch": 0.28156118068297203, "grad_norm": 1.0380784582260583, "learning_rate": 1.6861017346554958e-05, "loss": 0.4744, "step": 7903 }, { "epoch": 0.2815968078094661, "grad_norm": 1.8097892900372248, "learning_rate": 1.68601778054599e-05, "loss": 0.5711, "step": 7904 }, { "epoch": 0.28163243493596024, "grad_norm": 1.110553809675441, "learning_rate": 1.685933817301678e-05, "loss": 0.5195, "step": 7905 }, { "epoch": 0.28166806206245437, "grad_norm": 1.9178351093360693, "learning_rate": 1.6858498449236774e-05, "loss": 0.559, "step": 7906 }, { "epoch": 0.28170368918894845, "grad_norm": 1.7359507947572852, "learning_rate": 1.6857658634131068e-05, "loss": 0.6069, "step": 7907 }, { "epoch": 0.2817393163154426, "grad_norm": 1.1160271934312018, "learning_rate": 1.6856818727710847e-05, "loss": 0.55, "step": 7908 }, { "epoch": 0.2817749434419367, "grad_norm": 1.1303390167036922, "learning_rate": 1.6855978729987287e-05, "loss": 0.4826, "step": 7909 }, { "epoch": 0.2818105705684308, "grad_norm": 1.1555195699881606, "learning_rate": 1.685513864097158e-05, "loss": 0.4545, "step": 7910 }, { "epoch": 0.2818461976949249, "grad_norm": 1.6525234322872884, "learning_rate": 1.6854298460674915e-05, "loss": 0.6041, "step": 7911 }, { "epoch": 0.28188182482141905, "grad_norm": 1.0366074038216024, "learning_rate": 1.685345818910847e-05, "loss": 0.4718, "step": 7912 }, { "epoch": 0.2819174519479131, "grad_norm": 1.6297412035167802, "learning_rate": 1.685261782628344e-05, "loss": 0.5918, "step": 7913 }, { "epoch": 0.28195307907440725, "grad_norm": 1.0275005783515292, "learning_rate": 1.6851777372211014e-05, "loss": 0.4522, "step": 7914 }, { "epoch": 0.2819887062009014, "grad_norm": 1.991315000654822, "learning_rate": 1.6850936826902386e-05, "loss": 0.6269, "step": 7915 }, { "epoch": 0.28202433332739546, "grad_norm": 1.3084426615925742, "learning_rate": 1.685009619036874e-05, "loss": 0.5168, "step": 7916 }, { "epoch": 0.2820599604538896, "grad_norm": 1.0708484395069784, "learning_rate": 1.684925546262128e-05, "loss": 0.5221, "step": 7917 }, { "epoch": 0.2820955875803837, "grad_norm": 1.6882032809939858, "learning_rate": 1.6848414643671197e-05, "loss": 0.5891, "step": 7918 }, { "epoch": 0.2821312147068778, "grad_norm": 1.1642379922379174, "learning_rate": 1.684757373352968e-05, "loss": 0.4481, "step": 7919 }, { "epoch": 0.28216684183337193, "grad_norm": 1.6332730993473574, "learning_rate": 1.6846732732207938e-05, "loss": 0.6298, "step": 7920 }, { "epoch": 0.28220246895986606, "grad_norm": 1.1036110820619571, "learning_rate": 1.6845891639717164e-05, "loss": 0.5041, "step": 7921 }, { "epoch": 0.28223809608636014, "grad_norm": 1.4524962129561025, "learning_rate": 1.6845050456068555e-05, "loss": 0.6022, "step": 7922 }, { "epoch": 0.28227372321285427, "grad_norm": 1.1962538247476229, "learning_rate": 1.684420918127332e-05, "loss": 0.5564, "step": 7923 }, { "epoch": 0.2823093503393484, "grad_norm": 4.070718861867559, "learning_rate": 1.6843367815342648e-05, "loss": 0.6473, "step": 7924 }, { "epoch": 0.2823449774658425, "grad_norm": 1.1221499986567522, "learning_rate": 1.6842526358287755e-05, "loss": 0.515, "step": 7925 }, { "epoch": 0.2823806045923366, "grad_norm": 1.1581878741515588, "learning_rate": 1.684168481011984e-05, "loss": 0.5186, "step": 7926 }, { "epoch": 0.28241623171883073, "grad_norm": 1.1456591409175934, "learning_rate": 1.684084317085011e-05, "loss": 0.519, "step": 7927 }, { "epoch": 0.2824518588453248, "grad_norm": 1.5931209366048893, "learning_rate": 1.684000144048977e-05, "loss": 0.607, "step": 7928 }, { "epoch": 0.28248748597181894, "grad_norm": 1.5842301323877601, "learning_rate": 1.6839159619050036e-05, "loss": 0.6516, "step": 7929 }, { "epoch": 0.2825231130983131, "grad_norm": 1.9161444905346887, "learning_rate": 1.6838317706542105e-05, "loss": 0.5254, "step": 7930 }, { "epoch": 0.28255874022480715, "grad_norm": 1.2478172901681863, "learning_rate": 1.6837475702977198e-05, "loss": 0.5213, "step": 7931 }, { "epoch": 0.2825943673513013, "grad_norm": 1.5207382757090584, "learning_rate": 1.6836633608366518e-05, "loss": 0.5685, "step": 7932 }, { "epoch": 0.2826299944777954, "grad_norm": 1.6849661473689266, "learning_rate": 1.6835791422721287e-05, "loss": 0.5849, "step": 7933 }, { "epoch": 0.2826656216042895, "grad_norm": 1.7936887867756894, "learning_rate": 1.6834949146052717e-05, "loss": 0.5958, "step": 7934 }, { "epoch": 0.2827012487307836, "grad_norm": 1.072589617572928, "learning_rate": 1.6834106778372018e-05, "loss": 0.4627, "step": 7935 }, { "epoch": 0.28273687585727775, "grad_norm": 1.3723213971502934, "learning_rate": 1.683326431969041e-05, "loss": 0.5697, "step": 7936 }, { "epoch": 0.2827725029837718, "grad_norm": 1.4510485547748337, "learning_rate": 1.6832421770019115e-05, "loss": 0.6193, "step": 7937 }, { "epoch": 0.28280813011026595, "grad_norm": 1.5010498624881994, "learning_rate": 1.6831579129369347e-05, "loss": 0.5755, "step": 7938 }, { "epoch": 0.2828437572367601, "grad_norm": 1.7978049683135156, "learning_rate": 1.6830736397752326e-05, "loss": 0.6081, "step": 7939 }, { "epoch": 0.28287938436325416, "grad_norm": 1.6562867710308251, "learning_rate": 1.682989357517928e-05, "loss": 0.6162, "step": 7940 }, { "epoch": 0.2829150114897483, "grad_norm": 1.6325586053724697, "learning_rate": 1.6829050661661424e-05, "loss": 0.638, "step": 7941 }, { "epoch": 0.2829506386162424, "grad_norm": 1.8280414094841415, "learning_rate": 1.6828207657209982e-05, "loss": 0.6559, "step": 7942 }, { "epoch": 0.2829862657427365, "grad_norm": 1.4354492778472563, "learning_rate": 1.6827364561836187e-05, "loss": 0.4893, "step": 7943 }, { "epoch": 0.28302189286923063, "grad_norm": 1.2386508653050934, "learning_rate": 1.682652137555126e-05, "loss": 0.5293, "step": 7944 }, { "epoch": 0.28305751999572476, "grad_norm": 1.6045849964904355, "learning_rate": 1.6825678098366428e-05, "loss": 0.6183, "step": 7945 }, { "epoch": 0.28309314712221884, "grad_norm": 1.7894328896736693, "learning_rate": 1.6824834730292923e-05, "loss": 0.6477, "step": 7946 }, { "epoch": 0.28312877424871297, "grad_norm": 1.4643709494360395, "learning_rate": 1.6823991271341972e-05, "loss": 0.6276, "step": 7947 }, { "epoch": 0.2831644013752071, "grad_norm": 1.2667024083560234, "learning_rate": 1.682314772152481e-05, "loss": 0.5294, "step": 7948 }, { "epoch": 0.2832000285017012, "grad_norm": 1.0723748510515227, "learning_rate": 1.6822304080852663e-05, "loss": 0.4579, "step": 7949 }, { "epoch": 0.2832356556281953, "grad_norm": 1.6008415479537854, "learning_rate": 1.6821460349336773e-05, "loss": 0.5781, "step": 7950 }, { "epoch": 0.28327128275468944, "grad_norm": 3.255394771892139, "learning_rate": 1.682061652698837e-05, "loss": 0.5821, "step": 7951 }, { "epoch": 0.2833069098811835, "grad_norm": 1.1768962935768026, "learning_rate": 1.6819772613818686e-05, "loss": 0.5205, "step": 7952 }, { "epoch": 0.28334253700767764, "grad_norm": 1.2099544829675484, "learning_rate": 1.6818928609838967e-05, "loss": 0.5451, "step": 7953 }, { "epoch": 0.2833781641341718, "grad_norm": 1.9234379439843718, "learning_rate": 1.6818084515060444e-05, "loss": 0.6785, "step": 7954 }, { "epoch": 0.28341379126066585, "grad_norm": 1.9985919532363965, "learning_rate": 1.6817240329494365e-05, "loss": 0.5471, "step": 7955 }, { "epoch": 0.28344941838716, "grad_norm": 1.1878462120446345, "learning_rate": 1.681639605315196e-05, "loss": 0.5056, "step": 7956 }, { "epoch": 0.2834850455136541, "grad_norm": 1.56532433808758, "learning_rate": 1.681555168604448e-05, "loss": 0.69, "step": 7957 }, { "epoch": 0.2835206726401482, "grad_norm": 1.8721509168752968, "learning_rate": 1.681470722818317e-05, "loss": 0.5352, "step": 7958 }, { "epoch": 0.2835562997666423, "grad_norm": 1.9328540007875272, "learning_rate": 1.6813862679579265e-05, "loss": 0.6222, "step": 7959 }, { "epoch": 0.28359192689313645, "grad_norm": 1.247354483682181, "learning_rate": 1.681301804024402e-05, "loss": 0.5075, "step": 7960 }, { "epoch": 0.2836275540196305, "grad_norm": 1.742356111128351, "learning_rate": 1.6812173310188674e-05, "loss": 0.6638, "step": 7961 }, { "epoch": 0.28366318114612465, "grad_norm": 2.357355097001269, "learning_rate": 1.681132848942448e-05, "loss": 0.6068, "step": 7962 }, { "epoch": 0.2836988082726188, "grad_norm": 1.960075426082506, "learning_rate": 1.6810483577962684e-05, "loss": 0.499, "step": 7963 }, { "epoch": 0.28373443539911286, "grad_norm": 1.2012429458897798, "learning_rate": 1.6809638575814542e-05, "loss": 0.5951, "step": 7964 }, { "epoch": 0.283770062525607, "grad_norm": 1.3298132987684248, "learning_rate": 1.68087934829913e-05, "loss": 0.5352, "step": 7965 }, { "epoch": 0.2838056896521011, "grad_norm": 1.5608163989527886, "learning_rate": 1.6807948299504216e-05, "loss": 0.6318, "step": 7966 }, { "epoch": 0.2838413167785952, "grad_norm": 1.1766745500619904, "learning_rate": 1.6807103025364542e-05, "loss": 0.5048, "step": 7967 }, { "epoch": 0.28387694390508933, "grad_norm": 1.0737056424660503, "learning_rate": 1.6806257660583534e-05, "loss": 0.4586, "step": 7968 }, { "epoch": 0.28391257103158346, "grad_norm": 1.1829010000457603, "learning_rate": 1.6805412205172443e-05, "loss": 0.4809, "step": 7969 }, { "epoch": 0.28394819815807754, "grad_norm": 1.1653958727178104, "learning_rate": 1.6804566659142536e-05, "loss": 0.5238, "step": 7970 }, { "epoch": 0.28398382528457167, "grad_norm": 1.0705724745272374, "learning_rate": 1.680372102250507e-05, "loss": 0.4843, "step": 7971 }, { "epoch": 0.2840194524110658, "grad_norm": 1.4527201108911996, "learning_rate": 1.6802875295271297e-05, "loss": 0.4831, "step": 7972 }, { "epoch": 0.2840550795375599, "grad_norm": 1.6800937064657862, "learning_rate": 1.680202947745249e-05, "loss": 0.517, "step": 7973 }, { "epoch": 0.284090706664054, "grad_norm": 1.2444012719541793, "learning_rate": 1.6801183569059903e-05, "loss": 0.5071, "step": 7974 }, { "epoch": 0.28412633379054814, "grad_norm": 1.13288597765908, "learning_rate": 1.6800337570104806e-05, "loss": 0.5332, "step": 7975 }, { "epoch": 0.2841619609170422, "grad_norm": 1.432356243011847, "learning_rate": 1.679949148059846e-05, "loss": 0.6128, "step": 7976 }, { "epoch": 0.28419758804353634, "grad_norm": 1.461381081516747, "learning_rate": 1.679864530055213e-05, "loss": 0.6435, "step": 7977 }, { "epoch": 0.2842332151700305, "grad_norm": 1.443470936682615, "learning_rate": 1.6797799029977086e-05, "loss": 0.6229, "step": 7978 }, { "epoch": 0.28426884229652455, "grad_norm": 1.5342526539678114, "learning_rate": 1.6796952668884602e-05, "loss": 0.5499, "step": 7979 }, { "epoch": 0.2843044694230187, "grad_norm": 1.339195217898128, "learning_rate": 1.6796106217285933e-05, "loss": 0.5581, "step": 7980 }, { "epoch": 0.2843400965495128, "grad_norm": 1.7080497900993517, "learning_rate": 1.6795259675192367e-05, "loss": 0.604, "step": 7981 }, { "epoch": 0.2843757236760069, "grad_norm": 1.7588993198402252, "learning_rate": 1.6794413042615168e-05, "loss": 0.6279, "step": 7982 }, { "epoch": 0.284411350802501, "grad_norm": 1.63375839114806, "learning_rate": 1.6793566319565607e-05, "loss": 0.5384, "step": 7983 }, { "epoch": 0.28444697792899515, "grad_norm": 1.593117701080039, "learning_rate": 1.6792719506054967e-05, "loss": 0.6079, "step": 7984 }, { "epoch": 0.2844826050554892, "grad_norm": 1.7699354036249482, "learning_rate": 1.6791872602094513e-05, "loss": 0.5094, "step": 7985 }, { "epoch": 0.28451823218198335, "grad_norm": 1.3624659518350137, "learning_rate": 1.6791025607695532e-05, "loss": 0.5505, "step": 7986 }, { "epoch": 0.2845538593084775, "grad_norm": 1.4720325717671892, "learning_rate": 1.67901785228693e-05, "loss": 0.5788, "step": 7987 }, { "epoch": 0.28458948643497156, "grad_norm": 1.5608248291222873, "learning_rate": 1.678933134762709e-05, "loss": 0.5203, "step": 7988 }, { "epoch": 0.2846251135614657, "grad_norm": 1.3417297108450834, "learning_rate": 1.678848408198019e-05, "loss": 0.5046, "step": 7989 }, { "epoch": 0.2846607406879598, "grad_norm": 1.1595637036978992, "learning_rate": 1.678763672593988e-05, "loss": 0.4342, "step": 7990 }, { "epoch": 0.2846963678144539, "grad_norm": 1.144381253472678, "learning_rate": 1.678678927951744e-05, "loss": 0.4927, "step": 7991 }, { "epoch": 0.28473199494094803, "grad_norm": 1.1426355386577423, "learning_rate": 1.678594174272416e-05, "loss": 0.4888, "step": 7992 }, { "epoch": 0.28476762206744216, "grad_norm": 1.2677057024013507, "learning_rate": 1.6785094115571323e-05, "loss": 0.5413, "step": 7993 }, { "epoch": 0.28480324919393624, "grad_norm": 1.8929277573967096, "learning_rate": 1.6784246398070215e-05, "loss": 0.564, "step": 7994 }, { "epoch": 0.28483887632043037, "grad_norm": 1.5341614141580036, "learning_rate": 1.678339859023212e-05, "loss": 0.5923, "step": 7995 }, { "epoch": 0.2848745034469245, "grad_norm": 1.5423315500810477, "learning_rate": 1.6782550692068338e-05, "loss": 0.6092, "step": 7996 }, { "epoch": 0.2849101305734186, "grad_norm": 1.4836321288532368, "learning_rate": 1.678170270359015e-05, "loss": 0.632, "step": 7997 }, { "epoch": 0.2849457576999127, "grad_norm": 1.1737831449979146, "learning_rate": 1.678085462480885e-05, "loss": 0.5015, "step": 7998 }, { "epoch": 0.28498138482640684, "grad_norm": 1.8594884248314518, "learning_rate": 1.6780006455735735e-05, "loss": 0.6155, "step": 7999 }, { "epoch": 0.2850170119529009, "grad_norm": 1.9147012018713057, "learning_rate": 1.677915819638209e-05, "loss": 0.5744, "step": 8000 }, { "epoch": 0.28505263907939504, "grad_norm": 1.3898558252239732, "learning_rate": 1.6778309846759216e-05, "loss": 0.6344, "step": 8001 }, { "epoch": 0.2850882662058892, "grad_norm": 1.4037581713124145, "learning_rate": 1.6777461406878412e-05, "loss": 0.5813, "step": 8002 }, { "epoch": 0.28512389333238325, "grad_norm": 1.106300449115545, "learning_rate": 1.6776612876750967e-05, "loss": 0.5357, "step": 8003 }, { "epoch": 0.2851595204588774, "grad_norm": 1.1751047401315637, "learning_rate": 1.6775764256388187e-05, "loss": 0.4775, "step": 8004 }, { "epoch": 0.2851951475853715, "grad_norm": 1.1842835933482336, "learning_rate": 1.6774915545801372e-05, "loss": 0.5528, "step": 8005 }, { "epoch": 0.2852307747118656, "grad_norm": 1.4239677505036414, "learning_rate": 1.6774066745001818e-05, "loss": 0.5187, "step": 8006 }, { "epoch": 0.2852664018383597, "grad_norm": 2.1666453215671337, "learning_rate": 1.6773217854000834e-05, "loss": 0.5956, "step": 8007 }, { "epoch": 0.28530202896485385, "grad_norm": 1.5508620970535796, "learning_rate": 1.6772368872809716e-05, "loss": 0.5742, "step": 8008 }, { "epoch": 0.285337656091348, "grad_norm": 1.0895057535430388, "learning_rate": 1.6771519801439775e-05, "loss": 0.4864, "step": 8009 }, { "epoch": 0.28537328321784206, "grad_norm": 1.6309073092408424, "learning_rate": 1.6770670639902316e-05, "loss": 0.6049, "step": 8010 }, { "epoch": 0.2854089103443362, "grad_norm": 1.4547119786296248, "learning_rate": 1.6769821388208645e-05, "loss": 0.5996, "step": 8011 }, { "epoch": 0.2854445374708303, "grad_norm": 1.2405241456623632, "learning_rate": 1.6768972046370068e-05, "loss": 0.534, "step": 8012 }, { "epoch": 0.2854801645973244, "grad_norm": 1.085041606897757, "learning_rate": 1.67681226143979e-05, "loss": 0.456, "step": 8013 }, { "epoch": 0.2855157917238185, "grad_norm": 1.8575235478452654, "learning_rate": 1.6767273092303448e-05, "loss": 0.6643, "step": 8014 }, { "epoch": 0.28555141885031265, "grad_norm": 1.1626036647490872, "learning_rate": 1.6766423480098023e-05, "loss": 0.5341, "step": 8015 }, { "epoch": 0.28558704597680673, "grad_norm": 1.2936658319254053, "learning_rate": 1.676557377779294e-05, "loss": 0.654, "step": 8016 }, { "epoch": 0.28562267310330086, "grad_norm": 1.2454383369292397, "learning_rate": 1.6764723985399515e-05, "loss": 0.4643, "step": 8017 }, { "epoch": 0.285658300229795, "grad_norm": 1.1280263000916406, "learning_rate": 1.6763874102929058e-05, "loss": 0.4695, "step": 8018 }, { "epoch": 0.28569392735628907, "grad_norm": 1.720682159715363, "learning_rate": 1.6763024130392893e-05, "loss": 0.6531, "step": 8019 }, { "epoch": 0.2857295544827832, "grad_norm": 2.2058509452908552, "learning_rate": 1.6762174067802334e-05, "loss": 0.6338, "step": 8020 }, { "epoch": 0.28576518160927733, "grad_norm": 1.0718812969345632, "learning_rate": 1.6761323915168702e-05, "loss": 0.4831, "step": 8021 }, { "epoch": 0.2858008087357714, "grad_norm": 1.7500654221565803, "learning_rate": 1.6760473672503316e-05, "loss": 0.6713, "step": 8022 }, { "epoch": 0.28583643586226554, "grad_norm": 1.1405583536755053, "learning_rate": 1.6759623339817497e-05, "loss": 0.4713, "step": 8023 }, { "epoch": 0.28587206298875967, "grad_norm": 1.0823300726032117, "learning_rate": 1.675877291712257e-05, "loss": 0.4816, "step": 8024 }, { "epoch": 0.28590769011525374, "grad_norm": 1.1622289160942336, "learning_rate": 1.6757922404429854e-05, "loss": 0.4785, "step": 8025 }, { "epoch": 0.2859433172417479, "grad_norm": 1.486559790118991, "learning_rate": 1.675707180175068e-05, "loss": 0.5, "step": 8026 }, { "epoch": 0.285978944368242, "grad_norm": 1.3768134771346132, "learning_rate": 1.6756221109096375e-05, "loss": 0.4503, "step": 8027 }, { "epoch": 0.2860145714947361, "grad_norm": 1.6097909943058009, "learning_rate": 1.675537032647826e-05, "loss": 0.5176, "step": 8028 }, { "epoch": 0.2860501986212302, "grad_norm": 2.325823721603234, "learning_rate": 1.6754519453907666e-05, "loss": 0.5566, "step": 8029 }, { "epoch": 0.28608582574772434, "grad_norm": 1.391815611893846, "learning_rate": 1.6753668491395923e-05, "loss": 0.6078, "step": 8030 }, { "epoch": 0.2861214528742184, "grad_norm": 2.293982934404777, "learning_rate": 1.675281743895437e-05, "loss": 0.554, "step": 8031 }, { "epoch": 0.28615708000071255, "grad_norm": 1.1269494025385185, "learning_rate": 1.6751966296594328e-05, "loss": 0.5263, "step": 8032 }, { "epoch": 0.2861927071272067, "grad_norm": 1.0214157782812325, "learning_rate": 1.675111506432714e-05, "loss": 0.481, "step": 8033 }, { "epoch": 0.28622833425370076, "grad_norm": 1.1631636368617784, "learning_rate": 1.675026374216413e-05, "loss": 0.4863, "step": 8034 }, { "epoch": 0.2862639613801949, "grad_norm": 1.0190737063604474, "learning_rate": 1.6749412330116643e-05, "loss": 0.4866, "step": 8035 }, { "epoch": 0.286299588506689, "grad_norm": 1.1318808748545695, "learning_rate": 1.6748560828196013e-05, "loss": 0.5037, "step": 8036 }, { "epoch": 0.2863352156331831, "grad_norm": 1.0752198603225522, "learning_rate": 1.674770923641358e-05, "loss": 0.5195, "step": 8037 }, { "epoch": 0.2863708427596772, "grad_norm": 1.6743672811279915, "learning_rate": 1.674685755478068e-05, "loss": 0.62, "step": 8038 }, { "epoch": 0.28640646988617136, "grad_norm": 1.4702003320418717, "learning_rate": 1.674600578330866e-05, "loss": 0.5668, "step": 8039 }, { "epoch": 0.28644209701266543, "grad_norm": 2.3118190942821504, "learning_rate": 1.6745153922008853e-05, "loss": 0.6456, "step": 8040 }, { "epoch": 0.28647772413915956, "grad_norm": 1.7362579462027952, "learning_rate": 1.6744301970892608e-05, "loss": 0.5956, "step": 8041 }, { "epoch": 0.2865133512656537, "grad_norm": 2.160809378937112, "learning_rate": 1.674344992997127e-05, "loss": 0.5741, "step": 8042 }, { "epoch": 0.28654897839214777, "grad_norm": 1.1234998955646576, "learning_rate": 1.6742597799256182e-05, "loss": 0.4938, "step": 8043 }, { "epoch": 0.2865846055186419, "grad_norm": 1.5898895317744135, "learning_rate": 1.674174557875869e-05, "loss": 0.5615, "step": 8044 }, { "epoch": 0.28662023264513603, "grad_norm": 1.6184509820672954, "learning_rate": 1.6740893268490147e-05, "loss": 0.6187, "step": 8045 }, { "epoch": 0.2866558597716301, "grad_norm": 2.1719226893910735, "learning_rate": 1.6740040868461897e-05, "loss": 0.5883, "step": 8046 }, { "epoch": 0.28669148689812424, "grad_norm": 1.7445872198380354, "learning_rate": 1.673918837868529e-05, "loss": 0.5646, "step": 8047 }, { "epoch": 0.28672711402461837, "grad_norm": 1.0540594334654665, "learning_rate": 1.673833579917168e-05, "loss": 0.4889, "step": 8048 }, { "epoch": 0.28676274115111244, "grad_norm": 1.6151936380481833, "learning_rate": 1.673748312993242e-05, "loss": 0.5632, "step": 8049 }, { "epoch": 0.2867983682776066, "grad_norm": 1.8593041819330327, "learning_rate": 1.6736630370978862e-05, "loss": 0.5794, "step": 8050 }, { "epoch": 0.2868339954041007, "grad_norm": 1.5794374271724643, "learning_rate": 1.6735777522322362e-05, "loss": 0.585, "step": 8051 }, { "epoch": 0.2868696225305948, "grad_norm": 2.309020309374609, "learning_rate": 1.6734924583974275e-05, "loss": 0.6671, "step": 8052 }, { "epoch": 0.2869052496570889, "grad_norm": 1.0655132958250195, "learning_rate": 1.6734071555945965e-05, "loss": 0.4828, "step": 8053 }, { "epoch": 0.28694087678358304, "grad_norm": 2.1160330662920495, "learning_rate": 1.6733218438248782e-05, "loss": 0.6148, "step": 8054 }, { "epoch": 0.2869765039100771, "grad_norm": 1.4451868580576603, "learning_rate": 1.673236523089409e-05, "loss": 0.4944, "step": 8055 }, { "epoch": 0.28701213103657125, "grad_norm": 1.5922380156836615, "learning_rate": 1.6731511933893246e-05, "loss": 0.6295, "step": 8056 }, { "epoch": 0.2870477581630654, "grad_norm": 1.3337749008949706, "learning_rate": 1.6730658547257617e-05, "loss": 0.5971, "step": 8057 }, { "epoch": 0.28708338528955946, "grad_norm": 1.841484093703898, "learning_rate": 1.6729805070998567e-05, "loss": 0.5436, "step": 8058 }, { "epoch": 0.2871190124160536, "grad_norm": 1.8449835502691694, "learning_rate": 1.672895150512746e-05, "loss": 0.62, "step": 8059 }, { "epoch": 0.2871546395425477, "grad_norm": 1.9901959889261414, "learning_rate": 1.6728097849655654e-05, "loss": 0.7328, "step": 8060 }, { "epoch": 0.2871902666690418, "grad_norm": 1.2402501305527553, "learning_rate": 1.672724410459453e-05, "loss": 0.4824, "step": 8061 }, { "epoch": 0.2872258937955359, "grad_norm": 1.1680982345758513, "learning_rate": 1.6726390269955444e-05, "loss": 0.5102, "step": 8062 }, { "epoch": 0.28726152092203006, "grad_norm": 1.1816694086717707, "learning_rate": 1.672553634574977e-05, "loss": 0.5289, "step": 8063 }, { "epoch": 0.28729714804852413, "grad_norm": 1.0768699761139149, "learning_rate": 1.6724682331988882e-05, "loss": 0.5153, "step": 8064 }, { "epoch": 0.28733277517501826, "grad_norm": 1.0111273014269329, "learning_rate": 1.6723828228684144e-05, "loss": 0.4958, "step": 8065 }, { "epoch": 0.2873684023015124, "grad_norm": 1.2281305401820002, "learning_rate": 1.6722974035846937e-05, "loss": 0.4976, "step": 8066 }, { "epoch": 0.28740402942800647, "grad_norm": 1.5017284790792356, "learning_rate": 1.672211975348863e-05, "loss": 0.6289, "step": 8067 }, { "epoch": 0.2874396565545006, "grad_norm": 1.5154013831186406, "learning_rate": 1.67212653816206e-05, "loss": 0.6337, "step": 8068 }, { "epoch": 0.28747528368099473, "grad_norm": 1.595573250422888, "learning_rate": 1.6720410920254227e-05, "loss": 0.5438, "step": 8069 }, { "epoch": 0.2875109108074888, "grad_norm": 1.9384178828051362, "learning_rate": 1.671955636940088e-05, "loss": 0.598, "step": 8070 }, { "epoch": 0.28754653793398294, "grad_norm": 1.1923894127833283, "learning_rate": 1.6718701729071946e-05, "loss": 0.4851, "step": 8071 }, { "epoch": 0.28758216506047707, "grad_norm": 3.548158518201436, "learning_rate": 1.6717846999278805e-05, "loss": 0.5303, "step": 8072 }, { "epoch": 0.28761779218697114, "grad_norm": 1.1771034693004474, "learning_rate": 1.6716992180032832e-05, "loss": 0.4763, "step": 8073 }, { "epoch": 0.2876534193134653, "grad_norm": 1.489037211949083, "learning_rate": 1.671613727134541e-05, "loss": 0.6388, "step": 8074 }, { "epoch": 0.2876890464399594, "grad_norm": 1.6611328790878643, "learning_rate": 1.671528227322793e-05, "loss": 0.5952, "step": 8075 }, { "epoch": 0.2877246735664535, "grad_norm": 1.158752288491693, "learning_rate": 1.6714427185691775e-05, "loss": 0.5102, "step": 8076 }, { "epoch": 0.2877603006929476, "grad_norm": 1.254301842601367, "learning_rate": 1.6713572008748328e-05, "loss": 0.5635, "step": 8077 }, { "epoch": 0.28779592781944174, "grad_norm": 1.7797227013967416, "learning_rate": 1.6712716742408973e-05, "loss": 0.5469, "step": 8078 }, { "epoch": 0.2878315549459358, "grad_norm": 1.047052210668215, "learning_rate": 1.6711861386685108e-05, "loss": 0.444, "step": 8079 }, { "epoch": 0.28786718207242995, "grad_norm": 1.1947256502432118, "learning_rate": 1.6711005941588113e-05, "loss": 0.4757, "step": 8080 }, { "epoch": 0.2879028091989241, "grad_norm": 1.9825928049035941, "learning_rate": 1.6710150407129386e-05, "loss": 0.5588, "step": 8081 }, { "epoch": 0.28793843632541816, "grad_norm": 1.6824863701051311, "learning_rate": 1.6709294783320314e-05, "loss": 0.5448, "step": 8082 }, { "epoch": 0.2879740634519123, "grad_norm": 1.2949686031726813, "learning_rate": 1.6708439070172296e-05, "loss": 0.5071, "step": 8083 }, { "epoch": 0.2880096905784064, "grad_norm": 1.1326699335384658, "learning_rate": 1.6707583267696722e-05, "loss": 0.4872, "step": 8084 }, { "epoch": 0.2880453177049005, "grad_norm": 1.3111707131136339, "learning_rate": 1.6706727375904983e-05, "loss": 0.5514, "step": 8085 }, { "epoch": 0.2880809448313946, "grad_norm": 1.1921807435392922, "learning_rate": 1.6705871394808487e-05, "loss": 0.4758, "step": 8086 }, { "epoch": 0.28811657195788876, "grad_norm": 2.4951771393994995, "learning_rate": 1.6705015324418624e-05, "loss": 0.5474, "step": 8087 }, { "epoch": 0.28815219908438283, "grad_norm": 2.3931635936375053, "learning_rate": 1.6704159164746797e-05, "loss": 0.5939, "step": 8088 }, { "epoch": 0.28818782621087696, "grad_norm": 1.54865158312638, "learning_rate": 1.67033029158044e-05, "loss": 0.5697, "step": 8089 }, { "epoch": 0.2882234533373711, "grad_norm": 1.4309988065942283, "learning_rate": 1.6702446577602846e-05, "loss": 0.7117, "step": 8090 }, { "epoch": 0.28825908046386517, "grad_norm": 1.2352605776516306, "learning_rate": 1.670159015015353e-05, "loss": 0.4897, "step": 8091 }, { "epoch": 0.2882947075903593, "grad_norm": 1.2928469014632322, "learning_rate": 1.6700733633467855e-05, "loss": 0.5346, "step": 8092 }, { "epoch": 0.28833033471685343, "grad_norm": 1.2410201919190718, "learning_rate": 1.6699877027557227e-05, "loss": 0.4722, "step": 8093 }, { "epoch": 0.2883659618433475, "grad_norm": 1.6246139331052687, "learning_rate": 1.6699020332433054e-05, "loss": 0.6329, "step": 8094 }, { "epoch": 0.28840158896984164, "grad_norm": 1.0715513356106097, "learning_rate": 1.6698163548106744e-05, "loss": 0.5042, "step": 8095 }, { "epoch": 0.28843721609633577, "grad_norm": 1.6748829569266674, "learning_rate": 1.6697306674589704e-05, "loss": 0.5735, "step": 8096 }, { "epoch": 0.28847284322282984, "grad_norm": 1.1258126861015245, "learning_rate": 1.6696449711893346e-05, "loss": 0.4939, "step": 8097 }, { "epoch": 0.288508470349324, "grad_norm": 1.1479092232518875, "learning_rate": 1.6695592660029077e-05, "loss": 0.5493, "step": 8098 }, { "epoch": 0.2885440974758181, "grad_norm": 1.9231550256489869, "learning_rate": 1.6694735519008312e-05, "loss": 0.5354, "step": 8099 }, { "epoch": 0.2885797246023122, "grad_norm": 1.0735579577328376, "learning_rate": 1.6693878288842466e-05, "loss": 0.4773, "step": 8100 }, { "epoch": 0.2886153517288063, "grad_norm": 1.414861220548106, "learning_rate": 1.6693020969542953e-05, "loss": 0.5635, "step": 8101 }, { "epoch": 0.28865097885530044, "grad_norm": 1.4877956372773655, "learning_rate": 1.6692163561121185e-05, "loss": 0.4918, "step": 8102 }, { "epoch": 0.2886866059817945, "grad_norm": 1.699347722851152, "learning_rate": 1.6691306063588583e-05, "loss": 0.6616, "step": 8103 }, { "epoch": 0.28872223310828865, "grad_norm": 1.2779640113787347, "learning_rate": 1.6690448476956565e-05, "loss": 0.4766, "step": 8104 }, { "epoch": 0.2887578602347828, "grad_norm": 2.082823894349569, "learning_rate": 1.6689590801236545e-05, "loss": 0.5537, "step": 8105 }, { "epoch": 0.28879348736127686, "grad_norm": 1.393297490690986, "learning_rate": 1.6688733036439952e-05, "loss": 0.6324, "step": 8106 }, { "epoch": 0.288829114487771, "grad_norm": 1.7062761516467284, "learning_rate": 1.66878751825782e-05, "loss": 0.6038, "step": 8107 }, { "epoch": 0.2888647416142651, "grad_norm": 1.1133345877336622, "learning_rate": 1.668701723966272e-05, "loss": 0.5255, "step": 8108 }, { "epoch": 0.2889003687407592, "grad_norm": 2.435282368619578, "learning_rate": 1.6686159207704927e-05, "loss": 0.6231, "step": 8109 }, { "epoch": 0.2889359958672533, "grad_norm": 2.496750219368511, "learning_rate": 1.6685301086716252e-05, "loss": 0.5341, "step": 8110 }, { "epoch": 0.28897162299374746, "grad_norm": 1.2581705322642651, "learning_rate": 1.6684442876708124e-05, "loss": 0.5084, "step": 8111 }, { "epoch": 0.28900725012024153, "grad_norm": 1.1975470287555796, "learning_rate": 1.668358457769196e-05, "loss": 0.5312, "step": 8112 }, { "epoch": 0.28904287724673566, "grad_norm": 1.2884629895598492, "learning_rate": 1.66827261896792e-05, "loss": 0.4843, "step": 8113 }, { "epoch": 0.2890785043732298, "grad_norm": 2.555266940926672, "learning_rate": 1.6681867712681274e-05, "loss": 0.5956, "step": 8114 }, { "epoch": 0.28911413149972387, "grad_norm": 1.6571730296231708, "learning_rate": 1.66810091467096e-05, "loss": 0.5655, "step": 8115 }, { "epoch": 0.289149758626218, "grad_norm": 1.389582964383248, "learning_rate": 1.6680150491775626e-05, "loss": 0.5693, "step": 8116 }, { "epoch": 0.28918538575271213, "grad_norm": 1.1806502410780324, "learning_rate": 1.6679291747890777e-05, "loss": 0.4943, "step": 8117 }, { "epoch": 0.2892210128792062, "grad_norm": 1.60255479913435, "learning_rate": 1.6678432915066488e-05, "loss": 0.5251, "step": 8118 }, { "epoch": 0.28925664000570034, "grad_norm": 1.1248196371248869, "learning_rate": 1.66775739933142e-05, "loss": 0.5185, "step": 8119 }, { "epoch": 0.28929226713219447, "grad_norm": 1.2226925227845833, "learning_rate": 1.6676714982645345e-05, "loss": 0.432, "step": 8120 }, { "epoch": 0.28932789425868854, "grad_norm": 1.5347093879916265, "learning_rate": 1.6675855883071364e-05, "loss": 0.5434, "step": 8121 }, { "epoch": 0.2893635213851827, "grad_norm": 3.43801944132658, "learning_rate": 1.6674996694603694e-05, "loss": 0.6331, "step": 8122 }, { "epoch": 0.2893991485116768, "grad_norm": 1.235492147520218, "learning_rate": 1.667413741725378e-05, "loss": 0.4861, "step": 8123 }, { "epoch": 0.2894347756381709, "grad_norm": 1.1980424515699792, "learning_rate": 1.6673278051033062e-05, "loss": 0.5394, "step": 8124 }, { "epoch": 0.289470402764665, "grad_norm": 1.979595361769577, "learning_rate": 1.667241859595298e-05, "loss": 0.6543, "step": 8125 }, { "epoch": 0.28950602989115914, "grad_norm": 1.6613596818166372, "learning_rate": 1.6671559052024983e-05, "loss": 0.5652, "step": 8126 }, { "epoch": 0.2895416570176532, "grad_norm": 1.5344078905999783, "learning_rate": 1.667069941926051e-05, "loss": 0.5724, "step": 8127 }, { "epoch": 0.28957728414414735, "grad_norm": 1.1301474759509789, "learning_rate": 1.6669839697671014e-05, "loss": 0.4949, "step": 8128 }, { "epoch": 0.2896129112706415, "grad_norm": 1.1804878684844897, "learning_rate": 1.666897988726794e-05, "loss": 0.5242, "step": 8129 }, { "epoch": 0.28964853839713556, "grad_norm": 1.7780556770386784, "learning_rate": 1.666811998806274e-05, "loss": 0.6514, "step": 8130 }, { "epoch": 0.2896841655236297, "grad_norm": 1.2844666455546887, "learning_rate": 1.666726000006686e-05, "loss": 0.4651, "step": 8131 }, { "epoch": 0.2897197926501238, "grad_norm": 1.9451235454655862, "learning_rate": 1.6666399923291748e-05, "loss": 0.5751, "step": 8132 }, { "epoch": 0.2897554197766179, "grad_norm": 1.1851151696839886, "learning_rate": 1.6665539757748866e-05, "loss": 0.5197, "step": 8133 }, { "epoch": 0.289791046903112, "grad_norm": 1.3483987693457453, "learning_rate": 1.6664679503449668e-05, "loss": 0.5089, "step": 8134 }, { "epoch": 0.28982667402960616, "grad_norm": 2.499908596837823, "learning_rate": 1.6663819160405597e-05, "loss": 0.5268, "step": 8135 }, { "epoch": 0.28986230115610023, "grad_norm": 2.048947900212034, "learning_rate": 1.666295872862812e-05, "loss": 0.648, "step": 8136 }, { "epoch": 0.28989792828259436, "grad_norm": 1.5163418926584031, "learning_rate": 1.666209820812869e-05, "loss": 0.5125, "step": 8137 }, { "epoch": 0.2899335554090885, "grad_norm": 1.2214493198383813, "learning_rate": 1.6661237598918766e-05, "loss": 0.5419, "step": 8138 }, { "epoch": 0.28996918253558257, "grad_norm": 2.37533100278956, "learning_rate": 1.66603769010098e-05, "loss": 0.5776, "step": 8139 }, { "epoch": 0.2900048096620767, "grad_norm": 1.627597663603065, "learning_rate": 1.6659516114413273e-05, "loss": 0.6611, "step": 8140 }, { "epoch": 0.29004043678857083, "grad_norm": 1.306163847318933, "learning_rate": 1.6658655239140627e-05, "loss": 0.4709, "step": 8141 }, { "epoch": 0.2900760639150649, "grad_norm": 1.1914451079876003, "learning_rate": 1.665779427520333e-05, "loss": 0.4533, "step": 8142 }, { "epoch": 0.29011169104155904, "grad_norm": 1.13203794945585, "learning_rate": 1.6656933222612856e-05, "loss": 0.4744, "step": 8143 }, { "epoch": 0.29014731816805317, "grad_norm": 1.2211972739923962, "learning_rate": 1.6656072081380657e-05, "loss": 0.5194, "step": 8144 }, { "epoch": 0.29018294529454725, "grad_norm": 2.9559468416645216, "learning_rate": 1.6655210851518207e-05, "loss": 0.6413, "step": 8145 }, { "epoch": 0.2902185724210414, "grad_norm": 1.2171643518862318, "learning_rate": 1.6654349533036975e-05, "loss": 0.521, "step": 8146 }, { "epoch": 0.2902541995475355, "grad_norm": 1.1990216465171708, "learning_rate": 1.6653488125948425e-05, "loss": 0.4704, "step": 8147 }, { "epoch": 0.2902898266740296, "grad_norm": 1.5004841839401166, "learning_rate": 1.665262663026403e-05, "loss": 0.5519, "step": 8148 }, { "epoch": 0.2903254538005237, "grad_norm": 1.42945278535813, "learning_rate": 1.6651765045995264e-05, "loss": 0.6003, "step": 8149 }, { "epoch": 0.29036108092701784, "grad_norm": 1.7245661213559347, "learning_rate": 1.66509033731536e-05, "loss": 0.6322, "step": 8150 }, { "epoch": 0.2903967080535119, "grad_norm": 1.1717567401298876, "learning_rate": 1.6650041611750505e-05, "loss": 0.5044, "step": 8151 }, { "epoch": 0.29043233518000605, "grad_norm": 1.9169880157883783, "learning_rate": 1.6649179761797457e-05, "loss": 0.6329, "step": 8152 }, { "epoch": 0.2904679623065002, "grad_norm": 1.418024786448821, "learning_rate": 1.6648317823305933e-05, "loss": 0.6425, "step": 8153 }, { "epoch": 0.29050358943299426, "grad_norm": 1.1739256797600046, "learning_rate": 1.664745579628741e-05, "loss": 0.465, "step": 8154 }, { "epoch": 0.2905392165594884, "grad_norm": 1.5573755054019363, "learning_rate": 1.6646593680753368e-05, "loss": 0.6536, "step": 8155 }, { "epoch": 0.2905748436859825, "grad_norm": 1.6100959327161426, "learning_rate": 1.6645731476715286e-05, "loss": 0.561, "step": 8156 }, { "epoch": 0.2906104708124766, "grad_norm": 1.2455092609823344, "learning_rate": 1.6644869184184644e-05, "loss": 0.509, "step": 8157 }, { "epoch": 0.2906460979389707, "grad_norm": 1.2169954828581997, "learning_rate": 1.6644006803172926e-05, "loss": 0.5492, "step": 8158 }, { "epoch": 0.29068172506546486, "grad_norm": 1.0923538454733595, "learning_rate": 1.664314433369161e-05, "loss": 0.497, "step": 8159 }, { "epoch": 0.29071735219195893, "grad_norm": 1.6554048534912007, "learning_rate": 1.6642281775752187e-05, "loss": 0.6263, "step": 8160 }, { "epoch": 0.29075297931845306, "grad_norm": 1.8259818243758388, "learning_rate": 1.6641419129366137e-05, "loss": 0.663, "step": 8161 }, { "epoch": 0.2907886064449472, "grad_norm": 1.0425470755938584, "learning_rate": 1.6640556394544953e-05, "loss": 0.491, "step": 8162 }, { "epoch": 0.29082423357144127, "grad_norm": 1.8001983588688737, "learning_rate": 1.6639693571300115e-05, "loss": 0.6487, "step": 8163 }, { "epoch": 0.2908598606979354, "grad_norm": 1.1430055694766528, "learning_rate": 1.6638830659643116e-05, "loss": 0.4563, "step": 8164 }, { "epoch": 0.29089548782442953, "grad_norm": 1.6435830552309083, "learning_rate": 1.663796765958545e-05, "loss": 0.716, "step": 8165 }, { "epoch": 0.2909311149509236, "grad_norm": 1.5164035822977633, "learning_rate": 1.6637104571138603e-05, "loss": 0.55, "step": 8166 }, { "epoch": 0.29096674207741774, "grad_norm": 1.6705876035575404, "learning_rate": 1.663624139431407e-05, "loss": 0.6182, "step": 8167 }, { "epoch": 0.29100236920391187, "grad_norm": 1.1471796653595037, "learning_rate": 1.6635378129123342e-05, "loss": 0.4649, "step": 8168 }, { "epoch": 0.29103799633040595, "grad_norm": 1.7142290954893091, "learning_rate": 1.663451477557792e-05, "loss": 0.6378, "step": 8169 }, { "epoch": 0.2910736234569001, "grad_norm": 2.3200056523136654, "learning_rate": 1.6633651333689296e-05, "loss": 0.6137, "step": 8170 }, { "epoch": 0.2911092505833942, "grad_norm": 1.2583227095933078, "learning_rate": 1.6632787803468968e-05, "loss": 0.4818, "step": 8171 }, { "epoch": 0.2911448777098883, "grad_norm": 1.1944685654551845, "learning_rate": 1.6631924184928432e-05, "loss": 0.5217, "step": 8172 }, { "epoch": 0.2911805048363824, "grad_norm": 1.9170525749919398, "learning_rate": 1.663106047807919e-05, "loss": 0.6686, "step": 8173 }, { "epoch": 0.29121613196287655, "grad_norm": 1.0935603344779907, "learning_rate": 1.6630196682932746e-05, "loss": 0.524, "step": 8174 }, { "epoch": 0.2912517590893706, "grad_norm": 1.4190149119265332, "learning_rate": 1.6629332799500593e-05, "loss": 0.5945, "step": 8175 }, { "epoch": 0.29128738621586475, "grad_norm": 2.039842525744671, "learning_rate": 1.6628468827794245e-05, "loss": 0.5812, "step": 8176 }, { "epoch": 0.2913230133423589, "grad_norm": 1.8322653190399543, "learning_rate": 1.6627604767825203e-05, "loss": 0.6566, "step": 8177 }, { "epoch": 0.29135864046885296, "grad_norm": 2.0512332808950653, "learning_rate": 1.6626740619604967e-05, "loss": 0.5688, "step": 8178 }, { "epoch": 0.2913942675953471, "grad_norm": 1.2424309952907016, "learning_rate": 1.662587638314505e-05, "loss": 0.4905, "step": 8179 }, { "epoch": 0.2914298947218412, "grad_norm": 1.2656279244257032, "learning_rate": 1.6625012058456956e-05, "loss": 0.5071, "step": 8180 }, { "epoch": 0.2914655218483353, "grad_norm": 1.5822809371354747, "learning_rate": 1.66241476455522e-05, "loss": 0.5648, "step": 8181 }, { "epoch": 0.2915011489748294, "grad_norm": 1.1274964131170944, "learning_rate": 1.6623283144442283e-05, "loss": 0.5101, "step": 8182 }, { "epoch": 0.29153677610132356, "grad_norm": 1.5002319791277507, "learning_rate": 1.6622418555138727e-05, "loss": 0.6369, "step": 8183 }, { "epoch": 0.29157240322781763, "grad_norm": 1.490660857852666, "learning_rate": 1.6621553877653034e-05, "loss": 0.527, "step": 8184 }, { "epoch": 0.29160803035431176, "grad_norm": 1.1111926401665555, "learning_rate": 1.6620689111996724e-05, "loss": 0.494, "step": 8185 }, { "epoch": 0.2916436574808059, "grad_norm": 2.2362392185179605, "learning_rate": 1.6619824258181313e-05, "loss": 0.578, "step": 8186 }, { "epoch": 0.29167928460729997, "grad_norm": 1.172919883618499, "learning_rate": 1.6618959316218313e-05, "loss": 0.5102, "step": 8187 }, { "epoch": 0.2917149117337941, "grad_norm": 1.176047567789517, "learning_rate": 1.6618094286119245e-05, "loss": 0.4681, "step": 8188 }, { "epoch": 0.29175053886028823, "grad_norm": 1.069867642743679, "learning_rate": 1.661722916789562e-05, "loss": 0.468, "step": 8189 }, { "epoch": 0.2917861659867823, "grad_norm": 1.5643845389481295, "learning_rate": 1.6616363961558972e-05, "loss": 0.6827, "step": 8190 }, { "epoch": 0.29182179311327644, "grad_norm": 1.3429070588852747, "learning_rate": 1.661549866712081e-05, "loss": 0.5973, "step": 8191 }, { "epoch": 0.29185742023977057, "grad_norm": 1.2763518391295436, "learning_rate": 1.661463328459266e-05, "loss": 0.5027, "step": 8192 }, { "epoch": 0.29189304736626465, "grad_norm": 1.5653770052478337, "learning_rate": 1.6613767813986045e-05, "loss": 0.5921, "step": 8193 }, { "epoch": 0.2919286744927588, "grad_norm": 1.133156618479636, "learning_rate": 1.6612902255312485e-05, "loss": 0.4833, "step": 8194 }, { "epoch": 0.2919643016192529, "grad_norm": 1.5240713347573496, "learning_rate": 1.6612036608583514e-05, "loss": 0.6269, "step": 8195 }, { "epoch": 0.29199992874574704, "grad_norm": 1.9843113984459948, "learning_rate": 1.6611170873810653e-05, "loss": 0.5892, "step": 8196 }, { "epoch": 0.2920355558722411, "grad_norm": 1.56805110168987, "learning_rate": 1.6610305051005427e-05, "loss": 0.5538, "step": 8197 }, { "epoch": 0.29207118299873525, "grad_norm": 1.5046979594990535, "learning_rate": 1.660943914017937e-05, "loss": 0.5667, "step": 8198 }, { "epoch": 0.2921068101252294, "grad_norm": 1.1561115342509871, "learning_rate": 1.6608573141344014e-05, "loss": 0.4679, "step": 8199 }, { "epoch": 0.29214243725172345, "grad_norm": 1.174097886341603, "learning_rate": 1.6607707054510886e-05, "loss": 0.4732, "step": 8200 }, { "epoch": 0.2921780643782176, "grad_norm": 1.8861294471298458, "learning_rate": 1.660684087969152e-05, "loss": 0.6064, "step": 8201 }, { "epoch": 0.2922136915047117, "grad_norm": 1.1902408961047612, "learning_rate": 1.660597461689745e-05, "loss": 0.4384, "step": 8202 }, { "epoch": 0.2922493186312058, "grad_norm": 1.247009063894627, "learning_rate": 1.6605108266140208e-05, "loss": 0.5136, "step": 8203 }, { "epoch": 0.2922849457576999, "grad_norm": 1.6534408688232538, "learning_rate": 1.660424182743134e-05, "loss": 0.6624, "step": 8204 }, { "epoch": 0.29232057288419405, "grad_norm": 1.0194851760449153, "learning_rate": 1.6603375300782368e-05, "loss": 0.4566, "step": 8205 }, { "epoch": 0.2923562000106881, "grad_norm": 1.2522941347938266, "learning_rate": 1.660250868620484e-05, "loss": 0.5771, "step": 8206 }, { "epoch": 0.29239182713718226, "grad_norm": 1.0572426296609627, "learning_rate": 1.6601641983710298e-05, "loss": 0.5168, "step": 8207 }, { "epoch": 0.2924274542636764, "grad_norm": 1.0815934037804869, "learning_rate": 1.6600775193310276e-05, "loss": 0.4408, "step": 8208 }, { "epoch": 0.29246308139017047, "grad_norm": 1.2177423936324252, "learning_rate": 1.659990831501632e-05, "loss": 0.5412, "step": 8209 }, { "epoch": 0.2924987085166646, "grad_norm": 1.2415437622912613, "learning_rate": 1.659904134883997e-05, "loss": 0.4998, "step": 8210 }, { "epoch": 0.2925343356431587, "grad_norm": 1.1840488050947917, "learning_rate": 1.6598174294792774e-05, "loss": 0.4843, "step": 8211 }, { "epoch": 0.2925699627696528, "grad_norm": 1.208698756908257, "learning_rate": 1.6597307152886273e-05, "loss": 0.5283, "step": 8212 }, { "epoch": 0.29260558989614693, "grad_norm": 1.2268830874113388, "learning_rate": 1.6596439923132016e-05, "loss": 0.4984, "step": 8213 }, { "epoch": 0.29264121702264106, "grad_norm": 1.1716338630730043, "learning_rate": 1.659557260554155e-05, "loss": 0.5153, "step": 8214 }, { "epoch": 0.29267684414913514, "grad_norm": 1.1892528424787205, "learning_rate": 1.659470520012643e-05, "loss": 0.4562, "step": 8215 }, { "epoch": 0.29271247127562927, "grad_norm": 1.0563859406372305, "learning_rate": 1.65938377068982e-05, "loss": 0.4874, "step": 8216 }, { "epoch": 0.2927480984021234, "grad_norm": 1.0813335510758897, "learning_rate": 1.6592970125868405e-05, "loss": 0.5095, "step": 8217 }, { "epoch": 0.2927837255286175, "grad_norm": 1.6620402345734255, "learning_rate": 1.6592102457048613e-05, "loss": 0.5194, "step": 8218 }, { "epoch": 0.2928193526551116, "grad_norm": 1.1891916516175682, "learning_rate": 1.6591234700450366e-05, "loss": 0.5004, "step": 8219 }, { "epoch": 0.29285497978160574, "grad_norm": 2.280015147288462, "learning_rate": 1.659036685608522e-05, "loss": 0.6035, "step": 8220 }, { "epoch": 0.2928906069080998, "grad_norm": 1.2182003491264355, "learning_rate": 1.6589498923964734e-05, "loss": 0.4983, "step": 8221 }, { "epoch": 0.29292623403459395, "grad_norm": 1.5260219259752386, "learning_rate": 1.6588630904100465e-05, "loss": 0.5578, "step": 8222 }, { "epoch": 0.2929618611610881, "grad_norm": 1.2217948136409011, "learning_rate": 1.658776279650397e-05, "loss": 0.5302, "step": 8223 }, { "epoch": 0.29299748828758215, "grad_norm": 1.823977113578504, "learning_rate": 1.6586894601186804e-05, "loss": 0.6267, "step": 8224 }, { "epoch": 0.2930331154140763, "grad_norm": 1.4222310479070635, "learning_rate": 1.6586026318160538e-05, "loss": 0.5183, "step": 8225 }, { "epoch": 0.2930687425405704, "grad_norm": 1.568711962526294, "learning_rate": 1.6585157947436727e-05, "loss": 0.6097, "step": 8226 }, { "epoch": 0.2931043696670645, "grad_norm": 1.1280758357159353, "learning_rate": 1.6584289489026937e-05, "loss": 0.4989, "step": 8227 }, { "epoch": 0.2931399967935586, "grad_norm": 1.9557770031935457, "learning_rate": 1.6583420942942728e-05, "loss": 0.617, "step": 8228 }, { "epoch": 0.29317562392005275, "grad_norm": 1.520935321889781, "learning_rate": 1.6582552309195667e-05, "loss": 0.628, "step": 8229 }, { "epoch": 0.29321125104654683, "grad_norm": 2.678908045901135, "learning_rate": 1.6581683587797322e-05, "loss": 0.641, "step": 8230 }, { "epoch": 0.29324687817304096, "grad_norm": 1.4804495042910937, "learning_rate": 1.6580814778759263e-05, "loss": 0.4471, "step": 8231 }, { "epoch": 0.2932825052995351, "grad_norm": 1.2590760897698277, "learning_rate": 1.657994588209305e-05, "loss": 0.4788, "step": 8232 }, { "epoch": 0.29331813242602917, "grad_norm": 1.1273984306508034, "learning_rate": 1.657907689781026e-05, "loss": 0.4765, "step": 8233 }, { "epoch": 0.2933537595525233, "grad_norm": 1.5726702242082158, "learning_rate": 1.6578207825922466e-05, "loss": 0.6436, "step": 8234 }, { "epoch": 0.2933893866790174, "grad_norm": 1.1376854187281513, "learning_rate": 1.6577338666441232e-05, "loss": 0.474, "step": 8235 }, { "epoch": 0.2934250138055115, "grad_norm": 1.169646361435451, "learning_rate": 1.657646941937814e-05, "loss": 0.4986, "step": 8236 }, { "epoch": 0.29346064093200563, "grad_norm": 1.8447062542134705, "learning_rate": 1.657560008474476e-05, "loss": 0.6042, "step": 8237 }, { "epoch": 0.29349626805849977, "grad_norm": 1.1105253263547408, "learning_rate": 1.657473066255267e-05, "loss": 0.4585, "step": 8238 }, { "epoch": 0.29353189518499384, "grad_norm": 1.8528729390100458, "learning_rate": 1.6573861152813445e-05, "loss": 0.6561, "step": 8239 }, { "epoch": 0.29356752231148797, "grad_norm": 1.4111522359875628, "learning_rate": 1.657299155553866e-05, "loss": 0.5313, "step": 8240 }, { "epoch": 0.2936031494379821, "grad_norm": 2.066436643219492, "learning_rate": 1.6572121870739903e-05, "loss": 0.6121, "step": 8241 }, { "epoch": 0.2936387765644762, "grad_norm": 1.4604985876175602, "learning_rate": 1.657125209842875e-05, "loss": 0.6199, "step": 8242 }, { "epoch": 0.2936744036909703, "grad_norm": 1.0988193423329735, "learning_rate": 1.6570382238616778e-05, "loss": 0.4639, "step": 8243 }, { "epoch": 0.29371003081746444, "grad_norm": 1.2647470223892827, "learning_rate": 1.6569512291315577e-05, "loss": 0.5257, "step": 8244 }, { "epoch": 0.2937456579439585, "grad_norm": 1.5348414121932732, "learning_rate": 1.656864225653673e-05, "loss": 0.6478, "step": 8245 }, { "epoch": 0.29378128507045265, "grad_norm": 1.098744862114488, "learning_rate": 1.6567772134291816e-05, "loss": 0.5479, "step": 8246 }, { "epoch": 0.2938169121969468, "grad_norm": 1.6493082814765887, "learning_rate": 1.6566901924592425e-05, "loss": 0.5577, "step": 8247 }, { "epoch": 0.29385253932344085, "grad_norm": 1.2611681744149041, "learning_rate": 1.6566031627450147e-05, "loss": 0.5336, "step": 8248 }, { "epoch": 0.293888166449935, "grad_norm": 1.099725129375624, "learning_rate": 1.656516124287657e-05, "loss": 0.4927, "step": 8249 }, { "epoch": 0.2939237935764291, "grad_norm": 1.6857356313733685, "learning_rate": 1.656429077088328e-05, "loss": 0.6067, "step": 8250 }, { "epoch": 0.2939594207029232, "grad_norm": 1.3520502020530012, "learning_rate": 1.6563420211481873e-05, "loss": 0.5666, "step": 8251 }, { "epoch": 0.2939950478294173, "grad_norm": 1.5604978743685627, "learning_rate": 1.6562549564683937e-05, "loss": 0.6027, "step": 8252 }, { "epoch": 0.29403067495591145, "grad_norm": 1.227954549675144, "learning_rate": 1.6561678830501064e-05, "loss": 0.5288, "step": 8253 }, { "epoch": 0.29406630208240553, "grad_norm": 1.1323181018322002, "learning_rate": 1.6560808008944852e-05, "loss": 0.4579, "step": 8254 }, { "epoch": 0.29410192920889966, "grad_norm": 1.2007491096058822, "learning_rate": 1.65599371000269e-05, "loss": 0.5171, "step": 8255 }, { "epoch": 0.2941375563353938, "grad_norm": 1.2198706507127548, "learning_rate": 1.6559066103758797e-05, "loss": 0.5015, "step": 8256 }, { "epoch": 0.29417318346188787, "grad_norm": 1.1546182775977527, "learning_rate": 1.655819502015214e-05, "loss": 0.4923, "step": 8257 }, { "epoch": 0.294208810588382, "grad_norm": 1.6860802417696548, "learning_rate": 1.655732384921854e-05, "loss": 0.6581, "step": 8258 }, { "epoch": 0.29424443771487613, "grad_norm": 1.1521423840644467, "learning_rate": 1.6556452590969583e-05, "loss": 0.5151, "step": 8259 }, { "epoch": 0.2942800648413702, "grad_norm": 0.9948917849942039, "learning_rate": 1.6555581245416882e-05, "loss": 0.4812, "step": 8260 }, { "epoch": 0.29431569196786433, "grad_norm": 1.0851215358657584, "learning_rate": 1.6554709812572032e-05, "loss": 0.4504, "step": 8261 }, { "epoch": 0.29435131909435847, "grad_norm": 1.3279715053985273, "learning_rate": 1.655383829244664e-05, "loss": 0.5394, "step": 8262 }, { "epoch": 0.29438694622085254, "grad_norm": 1.1613011793619021, "learning_rate": 1.6552966685052312e-05, "loss": 0.5441, "step": 8263 }, { "epoch": 0.29442257334734667, "grad_norm": 1.9553218437100686, "learning_rate": 1.655209499040065e-05, "loss": 0.6294, "step": 8264 }, { "epoch": 0.2944582004738408, "grad_norm": 1.9480095455315856, "learning_rate": 1.6551223208503263e-05, "loss": 0.605, "step": 8265 }, { "epoch": 0.2944938276003349, "grad_norm": 1.567573248839933, "learning_rate": 1.6550351339371763e-05, "loss": 0.6121, "step": 8266 }, { "epoch": 0.294529454726829, "grad_norm": 1.23095555733397, "learning_rate": 1.6549479383017755e-05, "loss": 0.5407, "step": 8267 }, { "epoch": 0.29456508185332314, "grad_norm": 1.6227334217944789, "learning_rate": 1.6548607339452853e-05, "loss": 0.5459, "step": 8268 }, { "epoch": 0.2946007089798172, "grad_norm": 1.070554252811939, "learning_rate": 1.6547735208688664e-05, "loss": 0.4979, "step": 8269 }, { "epoch": 0.29463633610631135, "grad_norm": 1.0778526111311602, "learning_rate": 1.6546862990736805e-05, "loss": 0.5208, "step": 8270 }, { "epoch": 0.2946719632328055, "grad_norm": 1.7788131150321214, "learning_rate": 1.654599068560889e-05, "loss": 0.5571, "step": 8271 }, { "epoch": 0.29470759035929955, "grad_norm": 1.5283408756877894, "learning_rate": 1.6545118293316533e-05, "loss": 0.5614, "step": 8272 }, { "epoch": 0.2947432174857937, "grad_norm": 1.587886125713541, "learning_rate": 1.6544245813871353e-05, "loss": 0.5132, "step": 8273 }, { "epoch": 0.2947788446122878, "grad_norm": 1.1488978712709728, "learning_rate": 1.6543373247284965e-05, "loss": 0.5397, "step": 8274 }, { "epoch": 0.2948144717387819, "grad_norm": 1.5792074759328794, "learning_rate": 1.654250059356899e-05, "loss": 0.6054, "step": 8275 }, { "epoch": 0.294850098865276, "grad_norm": 2.078060427739789, "learning_rate": 1.654162785273504e-05, "loss": 0.5722, "step": 8276 }, { "epoch": 0.29488572599177015, "grad_norm": 1.6842833954644367, "learning_rate": 1.6540755024794748e-05, "loss": 0.5947, "step": 8277 }, { "epoch": 0.29492135311826423, "grad_norm": 1.2134200591607618, "learning_rate": 1.6539882109759734e-05, "loss": 0.4667, "step": 8278 }, { "epoch": 0.29495698024475836, "grad_norm": 1.2262484048458664, "learning_rate": 1.653900910764161e-05, "loss": 0.4957, "step": 8279 }, { "epoch": 0.2949926073712525, "grad_norm": 1.151568182459636, "learning_rate": 1.653813601845202e-05, "loss": 0.5027, "step": 8280 }, { "epoch": 0.29502823449774657, "grad_norm": 1.178270717042096, "learning_rate": 1.653726284220257e-05, "loss": 0.4892, "step": 8281 }, { "epoch": 0.2950638616242407, "grad_norm": 2.0274434939876453, "learning_rate": 1.6536389578904898e-05, "loss": 0.6449, "step": 8282 }, { "epoch": 0.29509948875073483, "grad_norm": 1.136688531947015, "learning_rate": 1.6535516228570633e-05, "loss": 0.4631, "step": 8283 }, { "epoch": 0.2951351158772289, "grad_norm": 1.2670096763286915, "learning_rate": 1.6534642791211403e-05, "loss": 0.493, "step": 8284 }, { "epoch": 0.29517074300372304, "grad_norm": 1.2677316319370902, "learning_rate": 1.653376926683883e-05, "loss": 0.5003, "step": 8285 }, { "epoch": 0.29520637013021717, "grad_norm": 1.197930842766936, "learning_rate": 1.6532895655464556e-05, "loss": 0.4938, "step": 8286 }, { "epoch": 0.29524199725671124, "grad_norm": 1.2541981121617458, "learning_rate": 1.653202195710021e-05, "loss": 0.552, "step": 8287 }, { "epoch": 0.2952776243832054, "grad_norm": 1.09218886577774, "learning_rate": 1.6531148171757427e-05, "loss": 0.4618, "step": 8288 }, { "epoch": 0.2953132515096995, "grad_norm": 1.5896156974663729, "learning_rate": 1.6530274299447843e-05, "loss": 0.5358, "step": 8289 }, { "epoch": 0.2953488786361936, "grad_norm": 1.1221896892872643, "learning_rate": 1.6529400340183087e-05, "loss": 0.5046, "step": 8290 }, { "epoch": 0.2953845057626877, "grad_norm": 1.6957850033262372, "learning_rate": 1.6528526293974806e-05, "loss": 0.5563, "step": 8291 }, { "epoch": 0.29542013288918184, "grad_norm": 1.0834831782507903, "learning_rate": 1.6527652160834633e-05, "loss": 0.4813, "step": 8292 }, { "epoch": 0.2954557600156759, "grad_norm": 1.1490965657295902, "learning_rate": 1.6526777940774205e-05, "loss": 0.488, "step": 8293 }, { "epoch": 0.29549138714217005, "grad_norm": 1.818992566278092, "learning_rate": 1.6525903633805175e-05, "loss": 0.5941, "step": 8294 }, { "epoch": 0.2955270142686642, "grad_norm": 1.3454276256940223, "learning_rate": 1.6525029239939166e-05, "loss": 0.6662, "step": 8295 }, { "epoch": 0.29556264139515825, "grad_norm": 1.0763003655729906, "learning_rate": 1.652415475918784e-05, "loss": 0.4757, "step": 8296 }, { "epoch": 0.2955982685216524, "grad_norm": 1.0932043079773688, "learning_rate": 1.652328019156283e-05, "loss": 0.4455, "step": 8297 }, { "epoch": 0.2956338956481465, "grad_norm": 1.5971591935718334, "learning_rate": 1.652240553707579e-05, "loss": 0.6005, "step": 8298 }, { "epoch": 0.2956695227746406, "grad_norm": 1.1647509007646324, "learning_rate": 1.6521530795738356e-05, "loss": 0.5002, "step": 8299 }, { "epoch": 0.2957051499011347, "grad_norm": 1.685798299930125, "learning_rate": 1.652065596756218e-05, "loss": 0.5375, "step": 8300 }, { "epoch": 0.29574077702762885, "grad_norm": 1.8352630979474358, "learning_rate": 1.6519781052558918e-05, "loss": 0.4895, "step": 8301 }, { "epoch": 0.29577640415412293, "grad_norm": 1.1064031408044523, "learning_rate": 1.651890605074021e-05, "loss": 0.4907, "step": 8302 }, { "epoch": 0.29581203128061706, "grad_norm": 1.207450285689423, "learning_rate": 1.6518030962117713e-05, "loss": 0.4659, "step": 8303 }, { "epoch": 0.2958476584071112, "grad_norm": 1.527033423887102, "learning_rate": 1.651715578670308e-05, "loss": 0.6228, "step": 8304 }, { "epoch": 0.29588328553360527, "grad_norm": 1.481382466388802, "learning_rate": 1.6516280524507956e-05, "loss": 0.5876, "step": 8305 }, { "epoch": 0.2959189126600994, "grad_norm": 1.7182634474201324, "learning_rate": 1.6515405175544005e-05, "loss": 0.64, "step": 8306 }, { "epoch": 0.29595453978659353, "grad_norm": 1.4335716376800962, "learning_rate": 1.651452973982288e-05, "loss": 0.5178, "step": 8307 }, { "epoch": 0.2959901669130876, "grad_norm": 1.6810172036703188, "learning_rate": 1.6513654217356242e-05, "loss": 0.5918, "step": 8308 }, { "epoch": 0.29602579403958174, "grad_norm": 1.5745626009420177, "learning_rate": 1.6512778608155743e-05, "loss": 0.5713, "step": 8309 }, { "epoch": 0.29606142116607587, "grad_norm": 1.6583168700638007, "learning_rate": 1.6511902912233043e-05, "loss": 0.6346, "step": 8310 }, { "epoch": 0.29609704829256994, "grad_norm": 1.204445279849605, "learning_rate": 1.6511027129599807e-05, "loss": 0.5233, "step": 8311 }, { "epoch": 0.2961326754190641, "grad_norm": 1.2392764677393704, "learning_rate": 1.6510151260267694e-05, "loss": 0.5116, "step": 8312 }, { "epoch": 0.2961683025455582, "grad_norm": 1.1372850027742718, "learning_rate": 1.6509275304248366e-05, "loss": 0.4906, "step": 8313 }, { "epoch": 0.2962039296720523, "grad_norm": 1.2106122956581178, "learning_rate": 1.6508399261553484e-05, "loss": 0.4719, "step": 8314 }, { "epoch": 0.2962395567985464, "grad_norm": 1.2873279971187845, "learning_rate": 1.650752313219472e-05, "loss": 0.4964, "step": 8315 }, { "epoch": 0.29627518392504054, "grad_norm": 1.3902595214180278, "learning_rate": 1.6506646916183732e-05, "loss": 0.5653, "step": 8316 }, { "epoch": 0.2963108110515346, "grad_norm": 1.2021128170360789, "learning_rate": 1.6505770613532196e-05, "loss": 0.5035, "step": 8317 }, { "epoch": 0.29634643817802875, "grad_norm": 1.617108798507159, "learning_rate": 1.650489422425178e-05, "loss": 0.6127, "step": 8318 }, { "epoch": 0.2963820653045229, "grad_norm": 1.4235414672432019, "learning_rate": 1.650401774835415e-05, "loss": 0.6322, "step": 8319 }, { "epoch": 0.29641769243101695, "grad_norm": 1.5800803742827956, "learning_rate": 1.650314118585098e-05, "loss": 0.5745, "step": 8320 }, { "epoch": 0.2964533195575111, "grad_norm": 1.190609787308933, "learning_rate": 1.650226453675393e-05, "loss": 0.5037, "step": 8321 }, { "epoch": 0.2964889466840052, "grad_norm": 1.1785383091414927, "learning_rate": 1.650138780107469e-05, "loss": 0.5072, "step": 8322 }, { "epoch": 0.2965245738104993, "grad_norm": 1.9327250476680118, "learning_rate": 1.6500510978824928e-05, "loss": 0.7105, "step": 8323 }, { "epoch": 0.2965602009369934, "grad_norm": 1.1113493079188235, "learning_rate": 1.6499634070016316e-05, "loss": 0.4959, "step": 8324 }, { "epoch": 0.29659582806348755, "grad_norm": 1.5452371561796172, "learning_rate": 1.649875707466053e-05, "loss": 0.6147, "step": 8325 }, { "epoch": 0.29663145518998163, "grad_norm": 1.4679051583583296, "learning_rate": 1.6497879992769258e-05, "loss": 0.558, "step": 8326 }, { "epoch": 0.29666708231647576, "grad_norm": 1.1702846329939016, "learning_rate": 1.6497002824354172e-05, "loss": 0.5206, "step": 8327 }, { "epoch": 0.2967027094429699, "grad_norm": 1.3340173568922449, "learning_rate": 1.6496125569426946e-05, "loss": 0.5202, "step": 8328 }, { "epoch": 0.29673833656946397, "grad_norm": 1.2421142481129572, "learning_rate": 1.6495248227999272e-05, "loss": 0.5683, "step": 8329 }, { "epoch": 0.2967739636959581, "grad_norm": 1.6225431283879865, "learning_rate": 1.6494370800082826e-05, "loss": 0.6332, "step": 8330 }, { "epoch": 0.29680959082245223, "grad_norm": 1.8444127692008763, "learning_rate": 1.6493493285689293e-05, "loss": 0.5017, "step": 8331 }, { "epoch": 0.2968452179489463, "grad_norm": 1.1816282137734209, "learning_rate": 1.6492615684830362e-05, "loss": 0.479, "step": 8332 }, { "epoch": 0.29688084507544044, "grad_norm": 1.6927544993690053, "learning_rate": 1.6491737997517713e-05, "loss": 0.5785, "step": 8333 }, { "epoch": 0.29691647220193457, "grad_norm": 1.2602186063755765, "learning_rate": 1.6490860223763037e-05, "loss": 0.4976, "step": 8334 }, { "epoch": 0.29695209932842864, "grad_norm": 1.28968277645657, "learning_rate": 1.6489982363578017e-05, "loss": 0.5413, "step": 8335 }, { "epoch": 0.2969877264549228, "grad_norm": 1.0763453912693588, "learning_rate": 1.6489104416974346e-05, "loss": 0.5007, "step": 8336 }, { "epoch": 0.2970233535814169, "grad_norm": 1.5980946116197616, "learning_rate": 1.6488226383963715e-05, "loss": 0.6766, "step": 8337 }, { "epoch": 0.297058980707911, "grad_norm": 1.8303982023567706, "learning_rate": 1.6487348264557817e-05, "loss": 0.6307, "step": 8338 }, { "epoch": 0.2970946078344051, "grad_norm": 1.972992152597528, "learning_rate": 1.648647005876834e-05, "loss": 0.5655, "step": 8339 }, { "epoch": 0.29713023496089924, "grad_norm": 1.554177187424283, "learning_rate": 1.6485591766606978e-05, "loss": 0.625, "step": 8340 }, { "epoch": 0.2971658620873933, "grad_norm": 1.2279510140968035, "learning_rate": 1.648471338808543e-05, "loss": 0.4704, "step": 8341 }, { "epoch": 0.29720148921388745, "grad_norm": 1.6396009174955406, "learning_rate": 1.6483834923215395e-05, "loss": 0.6089, "step": 8342 }, { "epoch": 0.2972371163403816, "grad_norm": 1.4949212086378, "learning_rate": 1.648295637200856e-05, "loss": 0.5997, "step": 8343 }, { "epoch": 0.29727274346687566, "grad_norm": 2.3139339470714417, "learning_rate": 1.6482077734476634e-05, "loss": 0.5873, "step": 8344 }, { "epoch": 0.2973083705933698, "grad_norm": 1.292164784375833, "learning_rate": 1.6481199010631312e-05, "loss": 0.5005, "step": 8345 }, { "epoch": 0.2973439977198639, "grad_norm": 1.8275477897404966, "learning_rate": 1.6480320200484293e-05, "loss": 0.5679, "step": 8346 }, { "epoch": 0.297379624846358, "grad_norm": 1.0894684617556378, "learning_rate": 1.647944130404728e-05, "loss": 0.4431, "step": 8347 }, { "epoch": 0.2974152519728521, "grad_norm": 1.271268137974776, "learning_rate": 1.6478562321331982e-05, "loss": 0.4672, "step": 8348 }, { "epoch": 0.29745087909934625, "grad_norm": 1.0772160939173627, "learning_rate": 1.6477683252350095e-05, "loss": 0.4622, "step": 8349 }, { "epoch": 0.29748650622584033, "grad_norm": 1.0718255612819747, "learning_rate": 1.647680409711333e-05, "loss": 0.5339, "step": 8350 }, { "epoch": 0.29752213335233446, "grad_norm": 1.3788946693405397, "learning_rate": 1.6475924855633386e-05, "loss": 0.4786, "step": 8351 }, { "epoch": 0.2975577604788286, "grad_norm": 1.1029836016111076, "learning_rate": 1.647504552792198e-05, "loss": 0.5119, "step": 8352 }, { "epoch": 0.29759338760532267, "grad_norm": 1.1519537531451014, "learning_rate": 1.6474166113990817e-05, "loss": 0.4996, "step": 8353 }, { "epoch": 0.2976290147318168, "grad_norm": 1.1323579365066536, "learning_rate": 1.6473286613851606e-05, "loss": 0.513, "step": 8354 }, { "epoch": 0.29766464185831093, "grad_norm": 1.2766960503392157, "learning_rate": 1.647240702751606e-05, "loss": 0.5234, "step": 8355 }, { "epoch": 0.297700268984805, "grad_norm": 1.7318897792810148, "learning_rate": 1.647152735499589e-05, "loss": 0.5884, "step": 8356 }, { "epoch": 0.29773589611129914, "grad_norm": 1.61434952344049, "learning_rate": 1.6470647596302812e-05, "loss": 0.6079, "step": 8357 }, { "epoch": 0.29777152323779327, "grad_norm": 2.532955978046737, "learning_rate": 1.6469767751448538e-05, "loss": 0.6558, "step": 8358 }, { "epoch": 0.29780715036428734, "grad_norm": 1.6078271342895405, "learning_rate": 1.6468887820444785e-05, "loss": 0.547, "step": 8359 }, { "epoch": 0.2978427774907815, "grad_norm": 1.6678547687313015, "learning_rate": 1.6468007803303267e-05, "loss": 0.5721, "step": 8360 }, { "epoch": 0.2978784046172756, "grad_norm": 1.1762643109931303, "learning_rate": 1.6467127700035703e-05, "loss": 0.4876, "step": 8361 }, { "epoch": 0.2979140317437697, "grad_norm": 1.2012393434112483, "learning_rate": 1.6466247510653813e-05, "loss": 0.4691, "step": 8362 }, { "epoch": 0.2979496588702638, "grad_norm": 1.3781074046959785, "learning_rate": 1.646536723516932e-05, "loss": 0.4944, "step": 8363 }, { "epoch": 0.29798528599675794, "grad_norm": 1.4299771983252128, "learning_rate": 1.6464486873593942e-05, "loss": 0.5928, "step": 8364 }, { "epoch": 0.298020913123252, "grad_norm": 1.189809859988555, "learning_rate": 1.6463606425939404e-05, "loss": 0.4932, "step": 8365 }, { "epoch": 0.29805654024974615, "grad_norm": 1.3881403977948605, "learning_rate": 1.646272589221743e-05, "loss": 0.4894, "step": 8366 }, { "epoch": 0.2980921673762403, "grad_norm": 1.2292889709971297, "learning_rate": 1.6461845272439743e-05, "loss": 0.5085, "step": 8367 }, { "epoch": 0.29812779450273436, "grad_norm": 1.77648376416163, "learning_rate": 1.646096456661807e-05, "loss": 0.5013, "step": 8368 }, { "epoch": 0.2981634216292285, "grad_norm": 1.249025995073841, "learning_rate": 1.6460083774764135e-05, "loss": 0.5049, "step": 8369 }, { "epoch": 0.2981990487557226, "grad_norm": 1.1189124148577438, "learning_rate": 1.645920289688967e-05, "loss": 0.4981, "step": 8370 }, { "epoch": 0.2982346758822167, "grad_norm": 1.3747955824011262, "learning_rate": 1.6458321933006407e-05, "loss": 0.548, "step": 8371 }, { "epoch": 0.2982703030087108, "grad_norm": 2.1318446395918493, "learning_rate": 1.6457440883126074e-05, "loss": 0.5036, "step": 8372 }, { "epoch": 0.29830593013520496, "grad_norm": 1.1454186440908132, "learning_rate": 1.6456559747260396e-05, "loss": 0.5022, "step": 8373 }, { "epoch": 0.29834155726169903, "grad_norm": 1.1933581888468148, "learning_rate": 1.6455678525421116e-05, "loss": 0.4977, "step": 8374 }, { "epoch": 0.29837718438819316, "grad_norm": 1.620704467093014, "learning_rate": 1.645479721761996e-05, "loss": 0.6424, "step": 8375 }, { "epoch": 0.2984128115146873, "grad_norm": 1.4678257468281906, "learning_rate": 1.6453915823868673e-05, "loss": 0.5542, "step": 8376 }, { "epoch": 0.29844843864118137, "grad_norm": 1.3045276861094905, "learning_rate": 1.6453034344178983e-05, "loss": 0.5706, "step": 8377 }, { "epoch": 0.2984840657676755, "grad_norm": 1.7669341515102925, "learning_rate": 1.6452152778562633e-05, "loss": 0.5715, "step": 8378 }, { "epoch": 0.29851969289416963, "grad_norm": 1.5944475586630595, "learning_rate": 1.6451271127031357e-05, "loss": 0.622, "step": 8379 }, { "epoch": 0.2985553200206637, "grad_norm": 1.5034982536016628, "learning_rate": 1.6450389389596898e-05, "loss": 0.5196, "step": 8380 }, { "epoch": 0.29859094714715784, "grad_norm": 1.6045737872969283, "learning_rate": 1.6449507566270993e-05, "loss": 0.6126, "step": 8381 }, { "epoch": 0.29862657427365197, "grad_norm": 1.2271560553005605, "learning_rate": 1.644862565706539e-05, "loss": 0.5243, "step": 8382 }, { "epoch": 0.2986622014001461, "grad_norm": 1.7043318238129448, "learning_rate": 1.6447743661991826e-05, "loss": 0.6186, "step": 8383 }, { "epoch": 0.2986978285266402, "grad_norm": 1.446034521097126, "learning_rate": 1.6446861581062047e-05, "loss": 0.6032, "step": 8384 }, { "epoch": 0.2987334556531343, "grad_norm": 2.3159709552529746, "learning_rate": 1.6445979414287806e-05, "loss": 0.615, "step": 8385 }, { "epoch": 0.29876908277962844, "grad_norm": 1.1419793724833618, "learning_rate": 1.6445097161680837e-05, "loss": 0.5173, "step": 8386 }, { "epoch": 0.2988047099061225, "grad_norm": 1.1456360441049247, "learning_rate": 1.64442148232529e-05, "loss": 0.5336, "step": 8387 }, { "epoch": 0.29884033703261664, "grad_norm": 1.1636859804460786, "learning_rate": 1.6443332399015734e-05, "loss": 0.4936, "step": 8388 }, { "epoch": 0.2988759641591108, "grad_norm": 1.307384635856288, "learning_rate": 1.6442449888981096e-05, "loss": 0.5003, "step": 8389 }, { "epoch": 0.29891159128560485, "grad_norm": 1.395444320313448, "learning_rate": 1.6441567293160737e-05, "loss": 0.5783, "step": 8390 }, { "epoch": 0.298947218412099, "grad_norm": 1.1132776779782623, "learning_rate": 1.6440684611566403e-05, "loss": 0.4788, "step": 8391 }, { "epoch": 0.2989828455385931, "grad_norm": 1.5381817042103314, "learning_rate": 1.643980184420985e-05, "loss": 0.5669, "step": 8392 }, { "epoch": 0.2990184726650872, "grad_norm": 1.4717749914799525, "learning_rate": 1.6438918991102842e-05, "loss": 0.6182, "step": 8393 }, { "epoch": 0.2990540997915813, "grad_norm": 1.2854091270946366, "learning_rate": 1.643803605225712e-05, "loss": 0.5121, "step": 8394 }, { "epoch": 0.29908972691807545, "grad_norm": 1.260599093943091, "learning_rate": 1.6437153027684452e-05, "loss": 0.5329, "step": 8395 }, { "epoch": 0.2991253540445695, "grad_norm": 1.6718630403223067, "learning_rate": 1.6436269917396593e-05, "loss": 0.5906, "step": 8396 }, { "epoch": 0.29916098117106366, "grad_norm": 1.6371345108375952, "learning_rate": 1.6435386721405297e-05, "loss": 0.5531, "step": 8397 }, { "epoch": 0.2991966082975578, "grad_norm": 1.224867703426073, "learning_rate": 1.643450343972233e-05, "loss": 0.4605, "step": 8398 }, { "epoch": 0.29923223542405186, "grad_norm": 1.1143368422543023, "learning_rate": 1.6433620072359452e-05, "loss": 0.4412, "step": 8399 }, { "epoch": 0.299267862550546, "grad_norm": 1.2867760837991875, "learning_rate": 1.6432736619328425e-05, "loss": 0.5009, "step": 8400 }, { "epoch": 0.2993034896770401, "grad_norm": 1.8288495112287007, "learning_rate": 1.643185308064102e-05, "loss": 0.6788, "step": 8401 }, { "epoch": 0.2993391168035342, "grad_norm": 1.8132220860172665, "learning_rate": 1.643096945630899e-05, "loss": 0.5695, "step": 8402 }, { "epoch": 0.29937474393002833, "grad_norm": 1.1814450560440812, "learning_rate": 1.6430085746344107e-05, "loss": 0.5564, "step": 8403 }, { "epoch": 0.29941037105652246, "grad_norm": 1.6249014687980419, "learning_rate": 1.642920195075814e-05, "loss": 0.63, "step": 8404 }, { "epoch": 0.29944599818301654, "grad_norm": 1.1522031394610959, "learning_rate": 1.6428318069562854e-05, "loss": 0.4618, "step": 8405 }, { "epoch": 0.29948162530951067, "grad_norm": 1.2190395833930983, "learning_rate": 1.642743410277002e-05, "loss": 0.528, "step": 8406 }, { "epoch": 0.2995172524360048, "grad_norm": 1.1707749204967368, "learning_rate": 1.642655005039141e-05, "loss": 0.518, "step": 8407 }, { "epoch": 0.2995528795624989, "grad_norm": 1.5651749306612863, "learning_rate": 1.6425665912438794e-05, "loss": 0.6193, "step": 8408 }, { "epoch": 0.299588506688993, "grad_norm": 1.1580061463549298, "learning_rate": 1.642478168892394e-05, "loss": 0.4811, "step": 8409 }, { "epoch": 0.29962413381548714, "grad_norm": 1.4031146234992062, "learning_rate": 1.6423897379858633e-05, "loss": 0.6053, "step": 8410 }, { "epoch": 0.2996597609419812, "grad_norm": 1.2125226145807642, "learning_rate": 1.6423012985254638e-05, "loss": 0.5002, "step": 8411 }, { "epoch": 0.29969538806847534, "grad_norm": 1.936853473089505, "learning_rate": 1.6422128505123737e-05, "loss": 0.5855, "step": 8412 }, { "epoch": 0.2997310151949695, "grad_norm": 2.379162532835234, "learning_rate": 1.642124393947771e-05, "loss": 0.5772, "step": 8413 }, { "epoch": 0.29976664232146355, "grad_norm": 1.1759792571765473, "learning_rate": 1.6420359288328325e-05, "loss": 0.4776, "step": 8414 }, { "epoch": 0.2998022694479577, "grad_norm": 1.2160654389189023, "learning_rate": 1.6419474551687375e-05, "loss": 0.4745, "step": 8415 }, { "epoch": 0.2998378965744518, "grad_norm": 1.724911082184678, "learning_rate": 1.641858972956663e-05, "loss": 0.6696, "step": 8416 }, { "epoch": 0.2998735237009459, "grad_norm": 1.1406462329089386, "learning_rate": 1.641770482197788e-05, "loss": 0.483, "step": 8417 }, { "epoch": 0.29990915082744, "grad_norm": 1.2255867591357605, "learning_rate": 1.64168198289329e-05, "loss": 0.5005, "step": 8418 }, { "epoch": 0.29994477795393415, "grad_norm": 1.1941566217677186, "learning_rate": 1.6415934750443483e-05, "loss": 0.5085, "step": 8419 }, { "epoch": 0.2999804050804282, "grad_norm": 1.1619269175739875, "learning_rate": 1.6415049586521407e-05, "loss": 0.4404, "step": 8420 }, { "epoch": 0.30001603220692236, "grad_norm": 1.6304278816003002, "learning_rate": 1.6414164337178467e-05, "loss": 0.6338, "step": 8421 }, { "epoch": 0.3000516593334165, "grad_norm": 1.143800349615345, "learning_rate": 1.641327900242644e-05, "loss": 0.5416, "step": 8422 }, { "epoch": 0.30008728645991056, "grad_norm": 1.6693183318688705, "learning_rate": 1.6412393582277122e-05, "loss": 0.5817, "step": 8423 }, { "epoch": 0.3001229135864047, "grad_norm": 1.1879403044135433, "learning_rate": 1.6411508076742305e-05, "loss": 0.4783, "step": 8424 }, { "epoch": 0.3001585407128988, "grad_norm": 1.6964971572625236, "learning_rate": 1.6410622485833774e-05, "loss": 0.5815, "step": 8425 }, { "epoch": 0.3001941678393929, "grad_norm": 1.4321398042922817, "learning_rate": 1.6409736809563325e-05, "loss": 0.6709, "step": 8426 }, { "epoch": 0.30022979496588703, "grad_norm": 1.9471472221127137, "learning_rate": 1.640885104794275e-05, "loss": 0.5662, "step": 8427 }, { "epoch": 0.30026542209238116, "grad_norm": 1.1905737790212694, "learning_rate": 1.6407965200983843e-05, "loss": 0.4951, "step": 8428 }, { "epoch": 0.30030104921887524, "grad_norm": 1.1855033921864542, "learning_rate": 1.64070792686984e-05, "loss": 0.5102, "step": 8429 }, { "epoch": 0.30033667634536937, "grad_norm": 1.521131507008281, "learning_rate": 1.6406193251098218e-05, "loss": 0.6687, "step": 8430 }, { "epoch": 0.3003723034718635, "grad_norm": 1.4607007843723443, "learning_rate": 1.6405307148195095e-05, "loss": 0.5663, "step": 8431 }, { "epoch": 0.3004079305983576, "grad_norm": 1.182013437238929, "learning_rate": 1.640442096000083e-05, "loss": 0.5289, "step": 8432 }, { "epoch": 0.3004435577248517, "grad_norm": 3.334654714944662, "learning_rate": 1.6403534686527223e-05, "loss": 0.5956, "step": 8433 }, { "epoch": 0.30047918485134584, "grad_norm": 1.2949836916540556, "learning_rate": 1.6402648327786076e-05, "loss": 0.4651, "step": 8434 }, { "epoch": 0.3005148119778399, "grad_norm": 1.6621132048103016, "learning_rate": 1.6401761883789196e-05, "loss": 0.5671, "step": 8435 }, { "epoch": 0.30055043910433404, "grad_norm": 1.620140419157982, "learning_rate": 1.6400875354548377e-05, "loss": 0.5622, "step": 8436 }, { "epoch": 0.3005860662308282, "grad_norm": 1.8173054142181488, "learning_rate": 1.6399988740075433e-05, "loss": 0.6184, "step": 8437 }, { "epoch": 0.30062169335732225, "grad_norm": 1.4843438002575244, "learning_rate": 1.6399102040382163e-05, "loss": 0.4908, "step": 8438 }, { "epoch": 0.3006573204838164, "grad_norm": 1.7434934059940421, "learning_rate": 1.6398215255480376e-05, "loss": 0.5774, "step": 8439 }, { "epoch": 0.3006929476103105, "grad_norm": 1.2382732327336055, "learning_rate": 1.639732838538188e-05, "loss": 0.5246, "step": 8440 }, { "epoch": 0.3007285747368046, "grad_norm": 1.8161204958280668, "learning_rate": 1.6396441430098485e-05, "loss": 0.6288, "step": 8441 }, { "epoch": 0.3007642018632987, "grad_norm": 1.427325262063663, "learning_rate": 1.6395554389642007e-05, "loss": 0.5604, "step": 8442 }, { "epoch": 0.30079982898979285, "grad_norm": 1.181222131558088, "learning_rate": 1.6394667264024246e-05, "loss": 0.501, "step": 8443 }, { "epoch": 0.3008354561162869, "grad_norm": 1.305870116866637, "learning_rate": 1.639378005325702e-05, "loss": 0.4479, "step": 8444 }, { "epoch": 0.30087108324278106, "grad_norm": 1.1816701445001965, "learning_rate": 1.639289275735215e-05, "loss": 0.4889, "step": 8445 }, { "epoch": 0.3009067103692752, "grad_norm": 1.8359432808452045, "learning_rate": 1.639200537632144e-05, "loss": 0.5945, "step": 8446 }, { "epoch": 0.30094233749576926, "grad_norm": 1.1344187446621035, "learning_rate": 1.6391117910176712e-05, "loss": 0.4754, "step": 8447 }, { "epoch": 0.3009779646222634, "grad_norm": 1.7411541345195476, "learning_rate": 1.639023035892978e-05, "loss": 0.6007, "step": 8448 }, { "epoch": 0.3010135917487575, "grad_norm": 1.3810681049077134, "learning_rate": 1.6389342722592468e-05, "loss": 0.4873, "step": 8449 }, { "epoch": 0.3010492188752516, "grad_norm": 1.7658003647585365, "learning_rate": 1.6388455001176587e-05, "loss": 0.6484, "step": 8450 }, { "epoch": 0.30108484600174573, "grad_norm": 1.6324627390303454, "learning_rate": 1.6387567194693967e-05, "loss": 0.6033, "step": 8451 }, { "epoch": 0.30112047312823986, "grad_norm": 1.162789731905039, "learning_rate": 1.6386679303156422e-05, "loss": 0.4714, "step": 8452 }, { "epoch": 0.30115610025473394, "grad_norm": 1.3570292214045896, "learning_rate": 1.638579132657578e-05, "loss": 0.5099, "step": 8453 }, { "epoch": 0.30119172738122807, "grad_norm": 1.086842779335183, "learning_rate": 1.6384903264963864e-05, "loss": 0.4966, "step": 8454 }, { "epoch": 0.3012273545077222, "grad_norm": 1.3841157598945562, "learning_rate": 1.6384015118332494e-05, "loss": 0.5795, "step": 8455 }, { "epoch": 0.3012629816342163, "grad_norm": 1.708838173361624, "learning_rate": 1.6383126886693507e-05, "loss": 0.7056, "step": 8456 }, { "epoch": 0.3012986087607104, "grad_norm": 1.375511706073571, "learning_rate": 1.6382238570058718e-05, "loss": 0.5896, "step": 8457 }, { "epoch": 0.30133423588720454, "grad_norm": 1.1119284645912688, "learning_rate": 1.638135016843996e-05, "loss": 0.4319, "step": 8458 }, { "epoch": 0.3013698630136986, "grad_norm": 1.50884431564363, "learning_rate": 1.638046168184907e-05, "loss": 0.6678, "step": 8459 }, { "epoch": 0.30140549014019274, "grad_norm": 1.4473315176667316, "learning_rate": 1.637957311029787e-05, "loss": 0.5043, "step": 8460 }, { "epoch": 0.3014411172666869, "grad_norm": 1.1865781646826559, "learning_rate": 1.63786844537982e-05, "loss": 0.4795, "step": 8461 }, { "epoch": 0.30147674439318095, "grad_norm": 1.1965772767531475, "learning_rate": 1.6377795712361882e-05, "loss": 0.4801, "step": 8462 }, { "epoch": 0.3015123715196751, "grad_norm": 1.0719239895965027, "learning_rate": 1.6376906886000758e-05, "loss": 0.4367, "step": 8463 }, { "epoch": 0.3015479986461692, "grad_norm": 1.1261980299599212, "learning_rate": 1.637601797472666e-05, "loss": 0.4606, "step": 8464 }, { "epoch": 0.3015836257726633, "grad_norm": 1.5351255625917406, "learning_rate": 1.637512897855143e-05, "loss": 0.5916, "step": 8465 }, { "epoch": 0.3016192528991574, "grad_norm": 1.1987864144227462, "learning_rate": 1.63742398974869e-05, "loss": 0.4973, "step": 8466 }, { "epoch": 0.30165488002565155, "grad_norm": 1.24052468618699, "learning_rate": 1.637335073154491e-05, "loss": 0.4584, "step": 8467 }, { "epoch": 0.3016905071521456, "grad_norm": 1.1903604895999553, "learning_rate": 1.63724614807373e-05, "loss": 0.5137, "step": 8468 }, { "epoch": 0.30172613427863976, "grad_norm": 1.1562645806183078, "learning_rate": 1.637157214507591e-05, "loss": 0.5168, "step": 8469 }, { "epoch": 0.3017617614051339, "grad_norm": 1.135482262454181, "learning_rate": 1.6370682724572583e-05, "loss": 0.4552, "step": 8470 }, { "epoch": 0.30179738853162796, "grad_norm": 1.725592599239863, "learning_rate": 1.636979321923917e-05, "loss": 0.6762, "step": 8471 }, { "epoch": 0.3018330156581221, "grad_norm": 1.2514503537731914, "learning_rate": 1.63689036290875e-05, "loss": 0.5505, "step": 8472 }, { "epoch": 0.3018686427846162, "grad_norm": 1.43830811399231, "learning_rate": 1.636801395412943e-05, "loss": 0.5447, "step": 8473 }, { "epoch": 0.3019042699111103, "grad_norm": 1.4926431766825643, "learning_rate": 1.6367124194376802e-05, "loss": 0.5916, "step": 8474 }, { "epoch": 0.30193989703760443, "grad_norm": 1.2126321280379682, "learning_rate": 1.6366234349841465e-05, "loss": 0.5499, "step": 8475 }, { "epoch": 0.30197552416409856, "grad_norm": 1.2604388028707947, "learning_rate": 1.636534442053527e-05, "loss": 0.5866, "step": 8476 }, { "epoch": 0.30201115129059264, "grad_norm": 1.4447050693901737, "learning_rate": 1.6364454406470064e-05, "loss": 0.5698, "step": 8477 }, { "epoch": 0.30204677841708677, "grad_norm": 1.4140032856725218, "learning_rate": 1.6363564307657696e-05, "loss": 0.5995, "step": 8478 }, { "epoch": 0.3020824055435809, "grad_norm": 1.2250438112162063, "learning_rate": 1.6362674124110026e-05, "loss": 0.4887, "step": 8479 }, { "epoch": 0.302118032670075, "grad_norm": 2.511802075734569, "learning_rate": 1.6361783855838898e-05, "loss": 0.5996, "step": 8480 }, { "epoch": 0.3021536597965691, "grad_norm": 1.7679626234674577, "learning_rate": 1.6360893502856175e-05, "loss": 0.6504, "step": 8481 }, { "epoch": 0.30218928692306324, "grad_norm": 1.4672010485954006, "learning_rate": 1.6360003065173707e-05, "loss": 0.5422, "step": 8482 }, { "epoch": 0.3022249140495573, "grad_norm": 1.8018846798078085, "learning_rate": 1.6359112542803355e-05, "loss": 0.6041, "step": 8483 }, { "epoch": 0.30226054117605144, "grad_norm": 1.4183939680913502, "learning_rate": 1.6358221935756975e-05, "loss": 0.6044, "step": 8484 }, { "epoch": 0.3022961683025456, "grad_norm": 1.242223024066552, "learning_rate": 1.6357331244046423e-05, "loss": 0.5079, "step": 8485 }, { "epoch": 0.30233179542903965, "grad_norm": 1.7328762017988355, "learning_rate": 1.6356440467683564e-05, "loss": 0.6312, "step": 8486 }, { "epoch": 0.3023674225555338, "grad_norm": 1.7196531865941334, "learning_rate": 1.6355549606680257e-05, "loss": 0.5934, "step": 8487 }, { "epoch": 0.3024030496820279, "grad_norm": 1.1425941046838923, "learning_rate": 1.6354658661048364e-05, "loss": 0.5186, "step": 8488 }, { "epoch": 0.302438676808522, "grad_norm": 1.2500966193179985, "learning_rate": 1.635376763079975e-05, "loss": 0.4529, "step": 8489 }, { "epoch": 0.3024743039350161, "grad_norm": 1.131443178520348, "learning_rate": 1.6352876515946276e-05, "loss": 0.4538, "step": 8490 }, { "epoch": 0.30250993106151025, "grad_norm": 1.458480738304817, "learning_rate": 1.6351985316499815e-05, "loss": 0.6166, "step": 8491 }, { "epoch": 0.3025455581880043, "grad_norm": 1.7637836218275695, "learning_rate": 1.6351094032472228e-05, "loss": 0.54, "step": 8492 }, { "epoch": 0.30258118531449846, "grad_norm": 1.1212855442409597, "learning_rate": 1.6350202663875385e-05, "loss": 0.5294, "step": 8493 }, { "epoch": 0.3026168124409926, "grad_norm": 1.3875063325320065, "learning_rate": 1.6349311210721156e-05, "loss": 0.4875, "step": 8494 }, { "epoch": 0.30265243956748666, "grad_norm": 1.262211821549233, "learning_rate": 1.634841967302141e-05, "loss": 0.4924, "step": 8495 }, { "epoch": 0.3026880666939808, "grad_norm": 5.631683486533858, "learning_rate": 1.6347528050788016e-05, "loss": 0.618, "step": 8496 }, { "epoch": 0.3027236938204749, "grad_norm": 1.2643956903894529, "learning_rate": 1.6346636344032852e-05, "loss": 0.5179, "step": 8497 }, { "epoch": 0.302759320946969, "grad_norm": 1.6346107981745748, "learning_rate": 1.6345744552767788e-05, "loss": 0.6587, "step": 8498 }, { "epoch": 0.30279494807346313, "grad_norm": 1.2091056770546504, "learning_rate": 1.6344852677004702e-05, "loss": 0.5279, "step": 8499 }, { "epoch": 0.30283057519995726, "grad_norm": 1.6107250754480598, "learning_rate": 1.6343960716755467e-05, "loss": 0.628, "step": 8500 }, { "epoch": 0.30286620232645134, "grad_norm": 1.3710265626384988, "learning_rate": 1.6343068672031958e-05, "loss": 0.4752, "step": 8501 }, { "epoch": 0.30290182945294547, "grad_norm": 1.3827318694179687, "learning_rate": 1.6342176542846057e-05, "loss": 0.6187, "step": 8502 }, { "epoch": 0.3029374565794396, "grad_norm": 1.2753040222348613, "learning_rate": 1.6341284329209647e-05, "loss": 0.5985, "step": 8503 }, { "epoch": 0.3029730837059337, "grad_norm": 1.0605345823553443, "learning_rate": 1.63403920311346e-05, "loss": 0.4723, "step": 8504 }, { "epoch": 0.3030087108324278, "grad_norm": 1.6927374514899238, "learning_rate": 1.63394996486328e-05, "loss": 0.4795, "step": 8505 }, { "epoch": 0.30304433795892194, "grad_norm": 1.0709549596415813, "learning_rate": 1.6338607181716132e-05, "loss": 0.5394, "step": 8506 }, { "epoch": 0.303079965085416, "grad_norm": 1.2142779478103172, "learning_rate": 1.633771463039648e-05, "loss": 0.4883, "step": 8507 }, { "epoch": 0.30311559221191015, "grad_norm": 1.7151975618853115, "learning_rate": 1.6336821994685728e-05, "loss": 0.5782, "step": 8508 }, { "epoch": 0.3031512193384043, "grad_norm": 1.055343837303545, "learning_rate": 1.6335929274595762e-05, "loss": 0.529, "step": 8509 }, { "epoch": 0.30318684646489835, "grad_norm": 1.1617589702722073, "learning_rate": 1.633503647013847e-05, "loss": 0.4873, "step": 8510 }, { "epoch": 0.3032224735913925, "grad_norm": 1.4779626588914954, "learning_rate": 1.6334143581325736e-05, "loss": 0.5727, "step": 8511 }, { "epoch": 0.3032581007178866, "grad_norm": 1.508060131573186, "learning_rate": 1.6333250608169454e-05, "loss": 0.574, "step": 8512 }, { "epoch": 0.3032937278443807, "grad_norm": 1.0600996852235205, "learning_rate": 1.6332357550681515e-05, "loss": 0.4914, "step": 8513 }, { "epoch": 0.3033293549708748, "grad_norm": 1.1637525418476722, "learning_rate": 1.633146440887381e-05, "loss": 0.5332, "step": 8514 }, { "epoch": 0.30336498209736895, "grad_norm": 1.210658009266389, "learning_rate": 1.633057118275823e-05, "loss": 0.532, "step": 8515 }, { "epoch": 0.303400609223863, "grad_norm": 1.636528574579591, "learning_rate": 1.6329677872346674e-05, "loss": 0.6214, "step": 8516 }, { "epoch": 0.30343623635035716, "grad_norm": 1.110215022333054, "learning_rate": 1.632878447765103e-05, "loss": 0.477, "step": 8517 }, { "epoch": 0.3034718634768513, "grad_norm": 1.1858975052747633, "learning_rate": 1.632789099868319e-05, "loss": 0.4735, "step": 8518 }, { "epoch": 0.30350749060334536, "grad_norm": 1.3913395774187733, "learning_rate": 1.6326997435455067e-05, "loss": 0.6101, "step": 8519 }, { "epoch": 0.3035431177298395, "grad_norm": 1.2925484619487255, "learning_rate": 1.6326103787978546e-05, "loss": 0.4821, "step": 8520 }, { "epoch": 0.3035787448563336, "grad_norm": 1.397291227222155, "learning_rate": 1.6325210056265537e-05, "loss": 0.5768, "step": 8521 }, { "epoch": 0.3036143719828277, "grad_norm": 1.9279577671356383, "learning_rate": 1.6324316240327932e-05, "loss": 0.4416, "step": 8522 }, { "epoch": 0.30364999910932183, "grad_norm": 1.314205460553701, "learning_rate": 1.6323422340177634e-05, "loss": 0.597, "step": 8523 }, { "epoch": 0.30368562623581596, "grad_norm": 1.4381993194200253, "learning_rate": 1.6322528355826545e-05, "loss": 0.5892, "step": 8524 }, { "epoch": 0.30372125336231004, "grad_norm": 1.1702746752578659, "learning_rate": 1.632163428728658e-05, "loss": 0.505, "step": 8525 }, { "epoch": 0.30375688048880417, "grad_norm": 1.3297845318093804, "learning_rate": 1.632074013456963e-05, "loss": 0.5388, "step": 8526 }, { "epoch": 0.3037925076152983, "grad_norm": 1.1124274739044784, "learning_rate": 1.631984589768761e-05, "loss": 0.4606, "step": 8527 }, { "epoch": 0.3038281347417924, "grad_norm": 1.0684870016187704, "learning_rate": 1.6318951576652424e-05, "loss": 0.4852, "step": 8528 }, { "epoch": 0.3038637618682865, "grad_norm": 1.6056380263605712, "learning_rate": 1.6318057171475977e-05, "loss": 0.6574, "step": 8529 }, { "epoch": 0.30389938899478064, "grad_norm": 1.8613278524585717, "learning_rate": 1.6317162682170188e-05, "loss": 0.6332, "step": 8530 }, { "epoch": 0.3039350161212747, "grad_norm": 1.3489055423665788, "learning_rate": 1.6316268108746963e-05, "loss": 0.5283, "step": 8531 }, { "epoch": 0.30397064324776885, "grad_norm": 1.6490707968970133, "learning_rate": 1.631537345121821e-05, "loss": 0.53, "step": 8532 }, { "epoch": 0.304006270374263, "grad_norm": 1.4735515783971043, "learning_rate": 1.6314478709595845e-05, "loss": 0.6256, "step": 8533 }, { "epoch": 0.30404189750075705, "grad_norm": 1.4703053971889533, "learning_rate": 1.6313583883891785e-05, "loss": 0.6229, "step": 8534 }, { "epoch": 0.3040775246272512, "grad_norm": 1.1656876832591234, "learning_rate": 1.6312688974117943e-05, "loss": 0.5545, "step": 8535 }, { "epoch": 0.3041131517537453, "grad_norm": 1.2119071912629924, "learning_rate": 1.6311793980286232e-05, "loss": 0.4835, "step": 8536 }, { "epoch": 0.3041487788802394, "grad_norm": 1.1204537325929402, "learning_rate": 1.6310898902408574e-05, "loss": 0.4996, "step": 8537 }, { "epoch": 0.3041844060067335, "grad_norm": 1.200761451482401, "learning_rate": 1.6310003740496887e-05, "loss": 0.4917, "step": 8538 }, { "epoch": 0.30422003313322765, "grad_norm": 1.17156437799088, "learning_rate": 1.6309108494563088e-05, "loss": 0.5399, "step": 8539 }, { "epoch": 0.3042556602597217, "grad_norm": 1.0979861302921625, "learning_rate": 1.63082131646191e-05, "loss": 0.5161, "step": 8540 }, { "epoch": 0.30429128738621586, "grad_norm": 1.429012918848517, "learning_rate": 1.6307317750676846e-05, "loss": 0.6201, "step": 8541 }, { "epoch": 0.30432691451271, "grad_norm": 1.083502509616976, "learning_rate": 1.6306422252748245e-05, "loss": 0.468, "step": 8542 }, { "epoch": 0.30436254163920407, "grad_norm": 2.0071326134527903, "learning_rate": 1.6305526670845225e-05, "loss": 0.6365, "step": 8543 }, { "epoch": 0.3043981687656982, "grad_norm": 1.3508727229267534, "learning_rate": 1.630463100497971e-05, "loss": 0.5459, "step": 8544 }, { "epoch": 0.3044337958921923, "grad_norm": 1.1953572370993686, "learning_rate": 1.630373525516363e-05, "loss": 0.5266, "step": 8545 }, { "epoch": 0.3044694230186864, "grad_norm": 1.4186967665611134, "learning_rate": 1.6302839421408906e-05, "loss": 0.5539, "step": 8546 }, { "epoch": 0.30450505014518053, "grad_norm": 1.1647681328653239, "learning_rate": 1.630194350372747e-05, "loss": 0.5003, "step": 8547 }, { "epoch": 0.30454067727167466, "grad_norm": 1.8450347415448904, "learning_rate": 1.630104750213125e-05, "loss": 0.6124, "step": 8548 }, { "epoch": 0.30457630439816874, "grad_norm": 1.1553732869561189, "learning_rate": 1.630015141663218e-05, "loss": 0.4749, "step": 8549 }, { "epoch": 0.30461193152466287, "grad_norm": 1.481328996124066, "learning_rate": 1.629925524724219e-05, "loss": 0.6097, "step": 8550 }, { "epoch": 0.304647558651157, "grad_norm": 1.7262280495393403, "learning_rate": 1.6298358993973212e-05, "loss": 0.6008, "step": 8551 }, { "epoch": 0.3046831857776511, "grad_norm": 1.9367703671863177, "learning_rate": 1.629746265683718e-05, "loss": 0.5564, "step": 8552 }, { "epoch": 0.3047188129041452, "grad_norm": 1.622869933946934, "learning_rate": 1.6296566235846037e-05, "loss": 0.5201, "step": 8553 }, { "epoch": 0.30475444003063934, "grad_norm": 1.7082602970743612, "learning_rate": 1.629566973101171e-05, "loss": 0.6271, "step": 8554 }, { "epoch": 0.3047900671571334, "grad_norm": 1.553365411924767, "learning_rate": 1.629477314234614e-05, "loss": 0.5122, "step": 8555 }, { "epoch": 0.30482569428362755, "grad_norm": 1.518817411359476, "learning_rate": 1.6293876469861266e-05, "loss": 0.6018, "step": 8556 }, { "epoch": 0.3048613214101217, "grad_norm": 1.2223534136556764, "learning_rate": 1.629297971356903e-05, "loss": 0.4388, "step": 8557 }, { "epoch": 0.30489694853661575, "grad_norm": 1.8133286397967892, "learning_rate": 1.6292082873481368e-05, "loss": 0.5824, "step": 8558 }, { "epoch": 0.3049325756631099, "grad_norm": 1.590278309393455, "learning_rate": 1.6291185949610223e-05, "loss": 0.6262, "step": 8559 }, { "epoch": 0.304968202789604, "grad_norm": 1.4345060733644541, "learning_rate": 1.6290288941967546e-05, "loss": 0.5933, "step": 8560 }, { "epoch": 0.3050038299160981, "grad_norm": 1.7294422171613533, "learning_rate": 1.628939185056527e-05, "loss": 0.6054, "step": 8561 }, { "epoch": 0.3050394570425922, "grad_norm": 1.2218277446359873, "learning_rate": 1.6288494675415346e-05, "loss": 0.5131, "step": 8562 }, { "epoch": 0.30507508416908635, "grad_norm": 1.974713937219412, "learning_rate": 1.628759741652972e-05, "loss": 0.5318, "step": 8563 }, { "epoch": 0.30511071129558043, "grad_norm": 1.165916662030523, "learning_rate": 1.6286700073920343e-05, "loss": 0.5273, "step": 8564 }, { "epoch": 0.30514633842207456, "grad_norm": 1.158277947744226, "learning_rate": 1.6285802647599156e-05, "loss": 0.4754, "step": 8565 }, { "epoch": 0.3051819655485687, "grad_norm": 2.2560740361414884, "learning_rate": 1.6284905137578114e-05, "loss": 0.5349, "step": 8566 }, { "epoch": 0.30521759267506277, "grad_norm": 1.0947836838562195, "learning_rate": 1.628400754386917e-05, "loss": 0.5021, "step": 8567 }, { "epoch": 0.3052532198015569, "grad_norm": 1.446693222103191, "learning_rate": 1.628310986648427e-05, "loss": 0.6214, "step": 8568 }, { "epoch": 0.305288846928051, "grad_norm": 1.4014310359047166, "learning_rate": 1.6282212105435372e-05, "loss": 0.5513, "step": 8569 }, { "epoch": 0.3053244740545451, "grad_norm": 1.1860684968252708, "learning_rate": 1.628131426073443e-05, "loss": 0.5088, "step": 8570 }, { "epoch": 0.30536010118103923, "grad_norm": 1.72545493334787, "learning_rate": 1.6280416332393396e-05, "loss": 0.576, "step": 8571 }, { "epoch": 0.30539572830753337, "grad_norm": 1.487868537281278, "learning_rate": 1.627951832042423e-05, "loss": 0.5747, "step": 8572 }, { "epoch": 0.3054313554340275, "grad_norm": 1.2121616864563733, "learning_rate": 1.6278620224838887e-05, "loss": 0.4764, "step": 8573 }, { "epoch": 0.30546698256052157, "grad_norm": 2.0480062167631843, "learning_rate": 1.6277722045649328e-05, "loss": 0.6664, "step": 8574 }, { "epoch": 0.3055026096870157, "grad_norm": 1.3713413459215043, "learning_rate": 1.6276823782867514e-05, "loss": 0.5098, "step": 8575 }, { "epoch": 0.30553823681350983, "grad_norm": 1.731200458271618, "learning_rate": 1.62759254365054e-05, "loss": 0.487, "step": 8576 }, { "epoch": 0.3055738639400039, "grad_norm": 1.2088638915977397, "learning_rate": 1.627502700657496e-05, "loss": 0.5354, "step": 8577 }, { "epoch": 0.30560949106649804, "grad_norm": 1.0633967016854016, "learning_rate": 1.6274128493088142e-05, "loss": 0.4625, "step": 8578 }, { "epoch": 0.30564511819299217, "grad_norm": 1.0972289625085339, "learning_rate": 1.6273229896056922e-05, "loss": 0.4658, "step": 8579 }, { "epoch": 0.30568074531948625, "grad_norm": 1.6876223559116488, "learning_rate": 1.6272331215493257e-05, "loss": 0.6338, "step": 8580 }, { "epoch": 0.3057163724459804, "grad_norm": 1.1837874054134256, "learning_rate": 1.627143245140912e-05, "loss": 0.4775, "step": 8581 }, { "epoch": 0.3057519995724745, "grad_norm": 1.1762812368252247, "learning_rate": 1.6270533603816476e-05, "loss": 0.5275, "step": 8582 }, { "epoch": 0.3057876266989686, "grad_norm": 1.1341986544109643, "learning_rate": 1.6269634672727296e-05, "loss": 0.481, "step": 8583 }, { "epoch": 0.3058232538254627, "grad_norm": 1.5072384385181556, "learning_rate": 1.6268735658153546e-05, "loss": 0.5412, "step": 8584 }, { "epoch": 0.30585888095195685, "grad_norm": 1.1178105698392782, "learning_rate": 1.62678365601072e-05, "loss": 0.4829, "step": 8585 }, { "epoch": 0.3058945080784509, "grad_norm": 1.4838620553627138, "learning_rate": 1.626693737860023e-05, "loss": 0.6214, "step": 8586 }, { "epoch": 0.30593013520494505, "grad_norm": 1.0732799160898439, "learning_rate": 1.6266038113644605e-05, "loss": 0.5461, "step": 8587 }, { "epoch": 0.3059657623314392, "grad_norm": 1.2527824606055997, "learning_rate": 1.6265138765252308e-05, "loss": 0.4962, "step": 8588 }, { "epoch": 0.30600138945793326, "grad_norm": 1.5556036510191276, "learning_rate": 1.6264239333435305e-05, "loss": 0.51, "step": 8589 }, { "epoch": 0.3060370165844274, "grad_norm": 1.2987728113874315, "learning_rate": 1.626333981820558e-05, "loss": 0.6027, "step": 8590 }, { "epoch": 0.3060726437109215, "grad_norm": 1.2234492411209636, "learning_rate": 1.6262440219575106e-05, "loss": 0.4759, "step": 8591 }, { "epoch": 0.3061082708374156, "grad_norm": 1.203923419879661, "learning_rate": 1.6261540537555863e-05, "loss": 0.5044, "step": 8592 }, { "epoch": 0.30614389796390973, "grad_norm": 1.2801499447927962, "learning_rate": 1.626064077215983e-05, "loss": 0.5404, "step": 8593 }, { "epoch": 0.30617952509040386, "grad_norm": 1.5525955012508197, "learning_rate": 1.6259740923398994e-05, "loss": 0.5332, "step": 8594 }, { "epoch": 0.30621515221689793, "grad_norm": 2.150210653451821, "learning_rate": 1.625884099128533e-05, "loss": 0.5805, "step": 8595 }, { "epoch": 0.30625077934339207, "grad_norm": 1.4792828053348102, "learning_rate": 1.625794097583082e-05, "loss": 0.5984, "step": 8596 }, { "epoch": 0.3062864064698862, "grad_norm": 1.230675817374143, "learning_rate": 1.6257040877047454e-05, "loss": 0.4851, "step": 8597 }, { "epoch": 0.30632203359638027, "grad_norm": 1.147502907100173, "learning_rate": 1.6256140694947217e-05, "loss": 0.5131, "step": 8598 }, { "epoch": 0.3063576607228744, "grad_norm": 1.2605252628423569, "learning_rate": 1.6255240429542092e-05, "loss": 0.5721, "step": 8599 }, { "epoch": 0.30639328784936853, "grad_norm": 1.2398312793792425, "learning_rate": 1.6254340080844068e-05, "loss": 0.5111, "step": 8600 }, { "epoch": 0.3064289149758626, "grad_norm": 1.196836794643815, "learning_rate": 1.6253439648865137e-05, "loss": 0.5218, "step": 8601 }, { "epoch": 0.30646454210235674, "grad_norm": 1.053229171059464, "learning_rate": 1.6252539133617283e-05, "loss": 0.5126, "step": 8602 }, { "epoch": 0.30650016922885087, "grad_norm": 1.0966598963629135, "learning_rate": 1.6251638535112505e-05, "loss": 0.5096, "step": 8603 }, { "epoch": 0.30653579635534495, "grad_norm": 1.0809120844364237, "learning_rate": 1.6250737853362785e-05, "loss": 0.4915, "step": 8604 }, { "epoch": 0.3065714234818391, "grad_norm": 1.380931004374544, "learning_rate": 1.6249837088380122e-05, "loss": 0.5804, "step": 8605 }, { "epoch": 0.3066070506083332, "grad_norm": 1.4773224062495505, "learning_rate": 1.6248936240176512e-05, "loss": 0.4712, "step": 8606 }, { "epoch": 0.3066426777348273, "grad_norm": 1.3027916709488183, "learning_rate": 1.6248035308763947e-05, "loss": 0.5219, "step": 8607 }, { "epoch": 0.3066783048613214, "grad_norm": 1.155274726965332, "learning_rate": 1.6247134294154426e-05, "loss": 0.5016, "step": 8608 }, { "epoch": 0.30671393198781555, "grad_norm": 1.3811847208948516, "learning_rate": 1.6246233196359944e-05, "loss": 0.5932, "step": 8609 }, { "epoch": 0.3067495591143096, "grad_norm": 1.2352577025880151, "learning_rate": 1.6245332015392502e-05, "loss": 0.4632, "step": 8610 }, { "epoch": 0.30678518624080375, "grad_norm": 2.1042696711672244, "learning_rate": 1.62444307512641e-05, "loss": 0.5323, "step": 8611 }, { "epoch": 0.3068208133672979, "grad_norm": 1.221971389835972, "learning_rate": 1.6243529403986737e-05, "loss": 0.534, "step": 8612 }, { "epoch": 0.30685644049379196, "grad_norm": 1.5208599455885985, "learning_rate": 1.6242627973572415e-05, "loss": 0.6036, "step": 8613 }, { "epoch": 0.3068920676202861, "grad_norm": 1.2624549043751803, "learning_rate": 1.6241726460033136e-05, "loss": 0.5282, "step": 8614 }, { "epoch": 0.3069276947467802, "grad_norm": 1.5797858949630323, "learning_rate": 1.624082486338091e-05, "loss": 0.5663, "step": 8615 }, { "epoch": 0.3069633218732743, "grad_norm": 1.0939569605668966, "learning_rate": 1.623992318362774e-05, "loss": 0.4952, "step": 8616 }, { "epoch": 0.30699894899976843, "grad_norm": 1.5392673910583676, "learning_rate": 1.623902142078563e-05, "loss": 0.5463, "step": 8617 }, { "epoch": 0.30703457612626256, "grad_norm": 1.2549353292002294, "learning_rate": 1.623811957486659e-05, "loss": 0.496, "step": 8618 }, { "epoch": 0.30707020325275663, "grad_norm": 1.2078069846356396, "learning_rate": 1.6237217645882623e-05, "loss": 0.5169, "step": 8619 }, { "epoch": 0.30710583037925077, "grad_norm": 1.1377819485591634, "learning_rate": 1.623631563384575e-05, "loss": 0.4457, "step": 8620 }, { "epoch": 0.3071414575057449, "grad_norm": 1.1635002131751915, "learning_rate": 1.6235413538767974e-05, "loss": 0.5371, "step": 8621 }, { "epoch": 0.307177084632239, "grad_norm": 1.7983908817648555, "learning_rate": 1.6234511360661305e-05, "loss": 0.599, "step": 8622 }, { "epoch": 0.3072127117587331, "grad_norm": 1.4876432297768112, "learning_rate": 1.623360909953776e-05, "loss": 0.5733, "step": 8623 }, { "epoch": 0.30724833888522723, "grad_norm": 1.1257766624195102, "learning_rate": 1.6232706755409354e-05, "loss": 0.5428, "step": 8624 }, { "epoch": 0.3072839660117213, "grad_norm": 1.6736307904081607, "learning_rate": 1.6231804328288104e-05, "loss": 0.68, "step": 8625 }, { "epoch": 0.30731959313821544, "grad_norm": 1.1255620669358577, "learning_rate": 1.623090181818602e-05, "loss": 0.5187, "step": 8626 }, { "epoch": 0.30735522026470957, "grad_norm": 1.1010722677233735, "learning_rate": 1.6229999225115122e-05, "loss": 0.5126, "step": 8627 }, { "epoch": 0.30739084739120365, "grad_norm": 1.160526708730246, "learning_rate": 1.6229096549087434e-05, "loss": 0.4859, "step": 8628 }, { "epoch": 0.3074264745176978, "grad_norm": 1.769804795126978, "learning_rate": 1.6228193790114968e-05, "loss": 0.585, "step": 8629 }, { "epoch": 0.3074621016441919, "grad_norm": 1.472440155980099, "learning_rate": 1.6227290948209748e-05, "loss": 0.6196, "step": 8630 }, { "epoch": 0.307497728770686, "grad_norm": 1.7462705949611637, "learning_rate": 1.62263880233838e-05, "loss": 0.5762, "step": 8631 }, { "epoch": 0.3075333558971801, "grad_norm": 1.6270990494709867, "learning_rate": 1.6225485015649142e-05, "loss": 0.6291, "step": 8632 }, { "epoch": 0.30756898302367425, "grad_norm": 2.245708688821304, "learning_rate": 1.62245819250178e-05, "loss": 0.5679, "step": 8633 }, { "epoch": 0.3076046101501683, "grad_norm": 1.822130763315796, "learning_rate": 1.6223678751501798e-05, "loss": 0.4907, "step": 8634 }, { "epoch": 0.30764023727666245, "grad_norm": 1.391689782959557, "learning_rate": 1.6222775495113165e-05, "loss": 0.5517, "step": 8635 }, { "epoch": 0.3076758644031566, "grad_norm": 1.1718738456998932, "learning_rate": 1.6221872155863923e-05, "loss": 0.5287, "step": 8636 }, { "epoch": 0.30771149152965066, "grad_norm": 1.1859578318209407, "learning_rate": 1.6220968733766108e-05, "loss": 0.4294, "step": 8637 }, { "epoch": 0.3077471186561448, "grad_norm": 1.4656779363841335, "learning_rate": 1.6220065228831746e-05, "loss": 0.5524, "step": 8638 }, { "epoch": 0.3077827457826389, "grad_norm": 1.4808100760191902, "learning_rate": 1.6219161641072863e-05, "loss": 0.5911, "step": 8639 }, { "epoch": 0.307818372909133, "grad_norm": 1.0342280321797528, "learning_rate": 1.62182579705015e-05, "loss": 0.457, "step": 8640 }, { "epoch": 0.30785400003562713, "grad_norm": 1.6606061157108822, "learning_rate": 1.621735421712969e-05, "loss": 0.5728, "step": 8641 }, { "epoch": 0.30788962716212126, "grad_norm": 1.0170817731404957, "learning_rate": 1.6216450380969456e-05, "loss": 0.5265, "step": 8642 }, { "epoch": 0.30792525428861534, "grad_norm": 1.1398096156056783, "learning_rate": 1.621554646203284e-05, "loss": 0.4459, "step": 8643 }, { "epoch": 0.30796088141510947, "grad_norm": 1.2209618115593914, "learning_rate": 1.6214642460331884e-05, "loss": 0.4828, "step": 8644 }, { "epoch": 0.3079965085416036, "grad_norm": 1.096106328370479, "learning_rate": 1.6213738375878617e-05, "loss": 0.4734, "step": 8645 }, { "epoch": 0.3080321356680977, "grad_norm": 1.0501741971721417, "learning_rate": 1.621283420868508e-05, "loss": 0.4596, "step": 8646 }, { "epoch": 0.3080677627945918, "grad_norm": 1.7354135362799112, "learning_rate": 1.6211929958763314e-05, "loss": 0.6131, "step": 8647 }, { "epoch": 0.30810338992108594, "grad_norm": 1.1655200313101417, "learning_rate": 1.6211025626125358e-05, "loss": 0.5659, "step": 8648 }, { "epoch": 0.30813901704758, "grad_norm": 2.1052371457232053, "learning_rate": 1.621012121078325e-05, "loss": 0.6135, "step": 8649 }, { "epoch": 0.30817464417407414, "grad_norm": 1.0898178654950377, "learning_rate": 1.6209216712749044e-05, "loss": 0.4868, "step": 8650 }, { "epoch": 0.3082102713005683, "grad_norm": 1.088099414086242, "learning_rate": 1.6208312132034776e-05, "loss": 0.4551, "step": 8651 }, { "epoch": 0.30824589842706235, "grad_norm": 1.0696195587945465, "learning_rate": 1.620740746865249e-05, "loss": 0.4567, "step": 8652 }, { "epoch": 0.3082815255535565, "grad_norm": 1.059231826562402, "learning_rate": 1.620650272261424e-05, "loss": 0.5312, "step": 8653 }, { "epoch": 0.3083171526800506, "grad_norm": 1.1180941252562304, "learning_rate": 1.6205597893932065e-05, "loss": 0.4713, "step": 8654 }, { "epoch": 0.3083527798065447, "grad_norm": 2.0590938620427917, "learning_rate": 1.6204692982618015e-05, "loss": 0.6456, "step": 8655 }, { "epoch": 0.3083884069330388, "grad_norm": 1.5812564545189463, "learning_rate": 1.6203787988684144e-05, "loss": 0.5476, "step": 8656 }, { "epoch": 0.30842403405953295, "grad_norm": 1.2046875060365339, "learning_rate": 1.6202882912142497e-05, "loss": 0.5199, "step": 8657 }, { "epoch": 0.308459661186027, "grad_norm": 1.102406046160903, "learning_rate": 1.620197775300513e-05, "loss": 0.5007, "step": 8658 }, { "epoch": 0.30849528831252115, "grad_norm": 1.1908667005848634, "learning_rate": 1.6201072511284096e-05, "loss": 0.4866, "step": 8659 }, { "epoch": 0.3085309154390153, "grad_norm": 1.9836812838622921, "learning_rate": 1.6200167186991445e-05, "loss": 0.6909, "step": 8660 }, { "epoch": 0.30856654256550936, "grad_norm": 1.642354013993545, "learning_rate": 1.6199261780139238e-05, "loss": 0.5785, "step": 8661 }, { "epoch": 0.3086021696920035, "grad_norm": 1.1467163263449456, "learning_rate": 1.619835629073952e-05, "loss": 0.4571, "step": 8662 }, { "epoch": 0.3086377968184976, "grad_norm": 1.7746231727704793, "learning_rate": 1.6197450718804358e-05, "loss": 0.6023, "step": 8663 }, { "epoch": 0.3086734239449917, "grad_norm": 1.555488071052332, "learning_rate": 1.6196545064345813e-05, "loss": 0.6692, "step": 8664 }, { "epoch": 0.30870905107148583, "grad_norm": 1.7342172179378064, "learning_rate": 1.6195639327375935e-05, "loss": 0.4991, "step": 8665 }, { "epoch": 0.30874467819797996, "grad_norm": 1.1566524207650077, "learning_rate": 1.619473350790679e-05, "loss": 0.4908, "step": 8666 }, { "epoch": 0.30878030532447404, "grad_norm": 1.8191274664938861, "learning_rate": 1.6193827605950438e-05, "loss": 0.5889, "step": 8667 }, { "epoch": 0.30881593245096817, "grad_norm": 1.09618217626209, "learning_rate": 1.6192921621518946e-05, "loss": 0.4906, "step": 8668 }, { "epoch": 0.3088515595774623, "grad_norm": 1.8971947622684469, "learning_rate": 1.6192015554624368e-05, "loss": 0.5596, "step": 8669 }, { "epoch": 0.3088871867039564, "grad_norm": 1.994749230668455, "learning_rate": 1.6191109405278776e-05, "loss": 0.5696, "step": 8670 }, { "epoch": 0.3089228138304505, "grad_norm": 1.2207606385721714, "learning_rate": 1.6190203173494236e-05, "loss": 0.4839, "step": 8671 }, { "epoch": 0.30895844095694464, "grad_norm": 8.492061499318153, "learning_rate": 1.6189296859282816e-05, "loss": 0.6397, "step": 8672 }, { "epoch": 0.3089940680834387, "grad_norm": 1.7310823228043348, "learning_rate": 1.618839046265658e-05, "loss": 0.6197, "step": 8673 }, { "epoch": 0.30902969520993284, "grad_norm": 1.1701754932598682, "learning_rate": 1.6187483983627602e-05, "loss": 0.4725, "step": 8674 }, { "epoch": 0.309065322336427, "grad_norm": 1.3449093563437764, "learning_rate": 1.6186577422207946e-05, "loss": 0.5316, "step": 8675 }, { "epoch": 0.30910094946292105, "grad_norm": 1.6986541877139931, "learning_rate": 1.618567077840969e-05, "loss": 0.5526, "step": 8676 }, { "epoch": 0.3091365765894152, "grad_norm": 1.5002989924682484, "learning_rate": 1.6184764052244904e-05, "loss": 0.6058, "step": 8677 }, { "epoch": 0.3091722037159093, "grad_norm": 1.1840428001111891, "learning_rate": 1.6183857243725662e-05, "loss": 0.4852, "step": 8678 }, { "epoch": 0.3092078308424034, "grad_norm": 1.1442718961021003, "learning_rate": 1.618295035286404e-05, "loss": 0.483, "step": 8679 }, { "epoch": 0.3092434579688975, "grad_norm": 5.267411714456613, "learning_rate": 1.618204337967211e-05, "loss": 0.5785, "step": 8680 }, { "epoch": 0.30927908509539165, "grad_norm": 1.1699162110458605, "learning_rate": 1.6181136324161954e-05, "loss": 0.5427, "step": 8681 }, { "epoch": 0.3093147122218857, "grad_norm": 1.1025192322728292, "learning_rate": 1.6180229186345643e-05, "loss": 0.5007, "step": 8682 }, { "epoch": 0.30935033934837985, "grad_norm": 1.5831498841845073, "learning_rate": 1.617932196623526e-05, "loss": 0.6414, "step": 8683 }, { "epoch": 0.309385966474874, "grad_norm": 1.9630445464681525, "learning_rate": 1.617841466384289e-05, "loss": 0.651, "step": 8684 }, { "epoch": 0.30942159360136806, "grad_norm": 2.090948227499963, "learning_rate": 1.617750727918061e-05, "loss": 0.5164, "step": 8685 }, { "epoch": 0.3094572207278622, "grad_norm": 1.4658399555985289, "learning_rate": 1.6176599812260504e-05, "loss": 0.5844, "step": 8686 }, { "epoch": 0.3094928478543563, "grad_norm": 1.2441732112297923, "learning_rate": 1.617569226309465e-05, "loss": 0.5086, "step": 8687 }, { "epoch": 0.3095284749808504, "grad_norm": 1.1681679192307122, "learning_rate": 1.6174784631695138e-05, "loss": 0.4958, "step": 8688 }, { "epoch": 0.30956410210734453, "grad_norm": 1.234263889530989, "learning_rate": 1.6173876918074057e-05, "loss": 0.5226, "step": 8689 }, { "epoch": 0.30959972923383866, "grad_norm": 1.3866360832285816, "learning_rate": 1.6172969122243488e-05, "loss": 0.558, "step": 8690 }, { "epoch": 0.30963535636033274, "grad_norm": 1.092604131774113, "learning_rate": 1.6172061244215517e-05, "loss": 0.4408, "step": 8691 }, { "epoch": 0.30967098348682687, "grad_norm": 1.9116058121980761, "learning_rate": 1.617115328400224e-05, "loss": 0.5861, "step": 8692 }, { "epoch": 0.309706610613321, "grad_norm": 1.7040320099235038, "learning_rate": 1.617024524161574e-05, "loss": 0.577, "step": 8693 }, { "epoch": 0.3097422377398151, "grad_norm": 1.6003438792424134, "learning_rate": 1.6169337117068114e-05, "loss": 0.6799, "step": 8694 }, { "epoch": 0.3097778648663092, "grad_norm": 1.2529353070128686, "learning_rate": 1.6168428910371455e-05, "loss": 0.4798, "step": 8695 }, { "epoch": 0.30981349199280334, "grad_norm": 1.2097242903661156, "learning_rate": 1.6167520621537847e-05, "loss": 0.4818, "step": 8696 }, { "epoch": 0.3098491191192974, "grad_norm": 2.1986827290894495, "learning_rate": 1.6166612250579398e-05, "loss": 0.533, "step": 8697 }, { "epoch": 0.30988474624579154, "grad_norm": 1.1549942254057468, "learning_rate": 1.6165703797508193e-05, "loss": 0.4575, "step": 8698 }, { "epoch": 0.3099203733722857, "grad_norm": 1.125270020340795, "learning_rate": 1.6164795262336334e-05, "loss": 0.4456, "step": 8699 }, { "epoch": 0.30995600049877975, "grad_norm": 1.7774663297861726, "learning_rate": 1.6163886645075916e-05, "loss": 0.6488, "step": 8700 }, { "epoch": 0.3099916276252739, "grad_norm": 1.626747301156057, "learning_rate": 1.6162977945739036e-05, "loss": 0.6212, "step": 8701 }, { "epoch": 0.310027254751768, "grad_norm": 1.1471184196408235, "learning_rate": 1.6162069164337803e-05, "loss": 0.5051, "step": 8702 }, { "epoch": 0.3100628818782621, "grad_norm": 1.305756763796846, "learning_rate": 1.6161160300884308e-05, "loss": 0.499, "step": 8703 }, { "epoch": 0.3100985090047562, "grad_norm": 1.1048538422132703, "learning_rate": 1.6160251355390658e-05, "loss": 0.5001, "step": 8704 }, { "epoch": 0.31013413613125035, "grad_norm": 1.0856906318957293, "learning_rate": 1.615934232786896e-05, "loss": 0.4859, "step": 8705 }, { "epoch": 0.3101697632577444, "grad_norm": 1.0507220795626961, "learning_rate": 1.6158433218331305e-05, "loss": 0.5132, "step": 8706 }, { "epoch": 0.31020539038423856, "grad_norm": 1.5159657732780265, "learning_rate": 1.6157524026789814e-05, "loss": 0.6185, "step": 8707 }, { "epoch": 0.3102410175107327, "grad_norm": 1.7310461696290715, "learning_rate": 1.6156614753256583e-05, "loss": 0.6362, "step": 8708 }, { "epoch": 0.31027664463722676, "grad_norm": 1.1608572351515143, "learning_rate": 1.6155705397743726e-05, "loss": 0.4738, "step": 8709 }, { "epoch": 0.3103122717637209, "grad_norm": 1.2018527409290851, "learning_rate": 1.615479596026335e-05, "loss": 0.5235, "step": 8710 }, { "epoch": 0.310347898890215, "grad_norm": 1.1260635186083703, "learning_rate": 1.615388644082756e-05, "loss": 0.5468, "step": 8711 }, { "epoch": 0.3103835260167091, "grad_norm": 1.420416922300861, "learning_rate": 1.6152976839448472e-05, "loss": 0.5719, "step": 8712 }, { "epoch": 0.31041915314320323, "grad_norm": 1.110744469535698, "learning_rate": 1.6152067156138198e-05, "loss": 0.4397, "step": 8713 }, { "epoch": 0.31045478026969736, "grad_norm": 1.3029222493759047, "learning_rate": 1.615115739090885e-05, "loss": 0.5082, "step": 8714 }, { "epoch": 0.31049040739619144, "grad_norm": 1.2150003375832041, "learning_rate": 1.615024754377254e-05, "loss": 0.4401, "step": 8715 }, { "epoch": 0.31052603452268557, "grad_norm": 1.3679980514747003, "learning_rate": 1.6149337614741388e-05, "loss": 0.5894, "step": 8716 }, { "epoch": 0.3105616616491797, "grad_norm": 1.1316886777239707, "learning_rate": 1.6148427603827504e-05, "loss": 0.4743, "step": 8717 }, { "epoch": 0.3105972887756738, "grad_norm": 1.0636293276285547, "learning_rate": 1.614751751104301e-05, "loss": 0.4778, "step": 8718 }, { "epoch": 0.3106329159021679, "grad_norm": 1.1221630005899108, "learning_rate": 1.6146607336400024e-05, "loss": 0.5382, "step": 8719 }, { "epoch": 0.31066854302866204, "grad_norm": 1.4104196991109001, "learning_rate": 1.6145697079910666e-05, "loss": 0.6456, "step": 8720 }, { "epoch": 0.3107041701551561, "grad_norm": 1.1091062242427616, "learning_rate": 1.6144786741587056e-05, "loss": 0.4517, "step": 8721 }, { "epoch": 0.31073979728165024, "grad_norm": 1.4701144472754377, "learning_rate": 1.6143876321441313e-05, "loss": 0.5354, "step": 8722 }, { "epoch": 0.3107754244081444, "grad_norm": 1.2968155672009736, "learning_rate": 1.6142965819485564e-05, "loss": 0.5579, "step": 8723 }, { "epoch": 0.31081105153463845, "grad_norm": 1.597120346507072, "learning_rate": 1.614205523573193e-05, "loss": 0.6232, "step": 8724 }, { "epoch": 0.3108466786611326, "grad_norm": 1.9235036716702292, "learning_rate": 1.6141144570192538e-05, "loss": 0.5574, "step": 8725 }, { "epoch": 0.3108823057876267, "grad_norm": 1.6370908026370155, "learning_rate": 1.614023382287952e-05, "loss": 0.63, "step": 8726 }, { "epoch": 0.3109179329141208, "grad_norm": 1.1605637134939237, "learning_rate": 1.613932299380499e-05, "loss": 0.5219, "step": 8727 }, { "epoch": 0.3109535600406149, "grad_norm": 1.873239525472853, "learning_rate": 1.6138412082981084e-05, "loss": 0.6136, "step": 8728 }, { "epoch": 0.31098918716710905, "grad_norm": 1.4466482217707843, "learning_rate": 1.6137501090419932e-05, "loss": 0.4953, "step": 8729 }, { "epoch": 0.3110248142936031, "grad_norm": 1.5432941629688193, "learning_rate": 1.6136590016133663e-05, "loss": 0.534, "step": 8730 }, { "epoch": 0.31106044142009726, "grad_norm": 1.3006930643447356, "learning_rate": 1.6135678860134408e-05, "loss": 0.5618, "step": 8731 }, { "epoch": 0.3110960685465914, "grad_norm": 1.5011719010835378, "learning_rate": 1.61347676224343e-05, "loss": 0.6037, "step": 8732 }, { "epoch": 0.31113169567308546, "grad_norm": 1.6511528318430913, "learning_rate": 1.6133856303045476e-05, "loss": 0.5305, "step": 8733 }, { "epoch": 0.3111673227995796, "grad_norm": 1.0726475767668002, "learning_rate": 1.6132944901980062e-05, "loss": 0.4852, "step": 8734 }, { "epoch": 0.3112029499260737, "grad_norm": 1.0891029829541008, "learning_rate": 1.6132033419250203e-05, "loss": 0.4961, "step": 8735 }, { "epoch": 0.3112385770525678, "grad_norm": 1.0049916810567283, "learning_rate": 1.6131121854868035e-05, "loss": 0.4428, "step": 8736 }, { "epoch": 0.31127420417906193, "grad_norm": 1.0848146592477064, "learning_rate": 1.6130210208845695e-05, "loss": 0.5335, "step": 8737 }, { "epoch": 0.31130983130555606, "grad_norm": 2.437395059271125, "learning_rate": 1.612929848119532e-05, "loss": 0.7111, "step": 8738 }, { "epoch": 0.31134545843205014, "grad_norm": 1.3269638352861026, "learning_rate": 1.612838667192905e-05, "loss": 0.4829, "step": 8739 }, { "epoch": 0.31138108555854427, "grad_norm": 1.19359936439657, "learning_rate": 1.6127474781059028e-05, "loss": 0.4544, "step": 8740 }, { "epoch": 0.3114167126850384, "grad_norm": 1.0989423103214262, "learning_rate": 1.61265628085974e-05, "loss": 0.4582, "step": 8741 }, { "epoch": 0.3114523398115325, "grad_norm": 0.9975682132327808, "learning_rate": 1.61256507545563e-05, "loss": 0.4763, "step": 8742 }, { "epoch": 0.3114879669380266, "grad_norm": 1.0945987178816943, "learning_rate": 1.612473861894788e-05, "loss": 0.5202, "step": 8743 }, { "epoch": 0.31152359406452074, "grad_norm": 1.10404842928546, "learning_rate": 1.612382640178429e-05, "loss": 0.5334, "step": 8744 }, { "epoch": 0.3115592211910148, "grad_norm": 0.9973865492079149, "learning_rate": 1.6122914103077665e-05, "loss": 0.4837, "step": 8745 }, { "epoch": 0.31159484831750894, "grad_norm": 1.2145686642910196, "learning_rate": 1.6122001722840162e-05, "loss": 0.4985, "step": 8746 }, { "epoch": 0.3116304754440031, "grad_norm": 1.3208160362794017, "learning_rate": 1.6121089261083927e-05, "loss": 0.5725, "step": 8747 }, { "epoch": 0.31166610257049715, "grad_norm": 1.1010636468104185, "learning_rate": 1.6120176717821107e-05, "loss": 0.4454, "step": 8748 }, { "epoch": 0.3117017296969913, "grad_norm": 1.0605435430756307, "learning_rate": 1.6119264093063864e-05, "loss": 0.5238, "step": 8749 }, { "epoch": 0.3117373568234854, "grad_norm": 1.3901618780466025, "learning_rate": 1.6118351386824335e-05, "loss": 0.4678, "step": 8750 }, { "epoch": 0.3117729839499795, "grad_norm": 1.6366636179972531, "learning_rate": 1.611743859911468e-05, "loss": 0.5363, "step": 8751 }, { "epoch": 0.3118086110764736, "grad_norm": 1.6001920015609148, "learning_rate": 1.611652572994706e-05, "loss": 0.5512, "step": 8752 }, { "epoch": 0.31184423820296775, "grad_norm": 1.201865438430499, "learning_rate": 1.6115612779333624e-05, "loss": 0.5215, "step": 8753 }, { "epoch": 0.3118798653294618, "grad_norm": 1.0707798085189801, "learning_rate": 1.6114699747286528e-05, "loss": 0.518, "step": 8754 }, { "epoch": 0.31191549245595596, "grad_norm": 1.5561057776706912, "learning_rate": 1.6113786633817925e-05, "loss": 0.5855, "step": 8755 }, { "epoch": 0.3119511195824501, "grad_norm": 1.9463873045307665, "learning_rate": 1.6112873438939987e-05, "loss": 0.5702, "step": 8756 }, { "epoch": 0.31198674670894416, "grad_norm": 1.5645027211189506, "learning_rate": 1.6111960162664863e-05, "loss": 0.5239, "step": 8757 }, { "epoch": 0.3120223738354383, "grad_norm": 1.317536566539714, "learning_rate": 1.611104680500472e-05, "loss": 0.6099, "step": 8758 }, { "epoch": 0.3120580009619324, "grad_norm": 1.7318074975786795, "learning_rate": 1.6110133365971715e-05, "loss": 0.4855, "step": 8759 }, { "epoch": 0.31209362808842656, "grad_norm": 1.0437075597888295, "learning_rate": 1.6109219845578016e-05, "loss": 0.4993, "step": 8760 }, { "epoch": 0.31212925521492063, "grad_norm": 1.150496218849545, "learning_rate": 1.6108306243835784e-05, "loss": 0.4973, "step": 8761 }, { "epoch": 0.31216488234141476, "grad_norm": 1.727226067730491, "learning_rate": 1.6107392560757182e-05, "loss": 0.606, "step": 8762 }, { "epoch": 0.3122005094679089, "grad_norm": 1.132500439285876, "learning_rate": 1.6106478796354382e-05, "loss": 0.4774, "step": 8763 }, { "epoch": 0.31223613659440297, "grad_norm": 1.3690481725215422, "learning_rate": 1.610556495063955e-05, "loss": 0.6402, "step": 8764 }, { "epoch": 0.3122717637208971, "grad_norm": 1.5197813482919458, "learning_rate": 1.610465102362485e-05, "loss": 0.5971, "step": 8765 }, { "epoch": 0.31230739084739123, "grad_norm": 1.515364801917477, "learning_rate": 1.6103737015322458e-05, "loss": 0.5798, "step": 8766 }, { "epoch": 0.3123430179738853, "grad_norm": 1.6579146186901905, "learning_rate": 1.610282292574454e-05, "loss": 0.6641, "step": 8767 }, { "epoch": 0.31237864510037944, "grad_norm": 1.9211497125041246, "learning_rate": 1.610190875490327e-05, "loss": 0.5869, "step": 8768 }, { "epoch": 0.31241427222687357, "grad_norm": 1.1222396205536886, "learning_rate": 1.6100994502810813e-05, "loss": 0.4497, "step": 8769 }, { "epoch": 0.31244989935336764, "grad_norm": 1.404592068030386, "learning_rate": 1.6100080169479357e-05, "loss": 0.441, "step": 8770 }, { "epoch": 0.3124855264798618, "grad_norm": 2.4072157880616945, "learning_rate": 1.609916575492107e-05, "loss": 0.5527, "step": 8771 }, { "epoch": 0.3125211536063559, "grad_norm": 1.3836887079537605, "learning_rate": 1.6098251259148125e-05, "loss": 0.5825, "step": 8772 }, { "epoch": 0.31255678073285, "grad_norm": 1.101845039040765, "learning_rate": 1.6097336682172703e-05, "loss": 0.5073, "step": 8773 }, { "epoch": 0.3125924078593441, "grad_norm": 1.7325457318687323, "learning_rate": 1.609642202400698e-05, "loss": 0.6321, "step": 8774 }, { "epoch": 0.31262803498583824, "grad_norm": 1.4821445877548334, "learning_rate": 1.609550728466314e-05, "loss": 0.5761, "step": 8775 }, { "epoch": 0.3126636621123323, "grad_norm": 1.2220596920892757, "learning_rate": 1.609459246415336e-05, "loss": 0.4809, "step": 8776 }, { "epoch": 0.31269928923882645, "grad_norm": 1.5251847027827352, "learning_rate": 1.609367756248982e-05, "loss": 0.5437, "step": 8777 }, { "epoch": 0.3127349163653206, "grad_norm": 1.5942922953569048, "learning_rate": 1.6092762579684703e-05, "loss": 0.6659, "step": 8778 }, { "epoch": 0.31277054349181466, "grad_norm": 1.7537145340104328, "learning_rate": 1.6091847515750192e-05, "loss": 0.6911, "step": 8779 }, { "epoch": 0.3128061706183088, "grad_norm": 1.0854358523224028, "learning_rate": 1.6090932370698477e-05, "loss": 0.4789, "step": 8780 }, { "epoch": 0.3128417977448029, "grad_norm": 1.7084303694403613, "learning_rate": 1.6090017144541737e-05, "loss": 0.551, "step": 8781 }, { "epoch": 0.312877424871297, "grad_norm": 1.764738033589077, "learning_rate": 1.6089101837292168e-05, "loss": 0.5354, "step": 8782 }, { "epoch": 0.3129130519977911, "grad_norm": 1.1583598910911785, "learning_rate": 1.6088186448961948e-05, "loss": 0.4886, "step": 8783 }, { "epoch": 0.31294867912428526, "grad_norm": 1.1816328793925868, "learning_rate": 1.608727097956327e-05, "loss": 0.4757, "step": 8784 }, { "epoch": 0.31298430625077933, "grad_norm": 1.1604658040042763, "learning_rate": 1.608635542910832e-05, "loss": 0.4381, "step": 8785 }, { "epoch": 0.31301993337727346, "grad_norm": 1.2640711692765974, "learning_rate": 1.60854397976093e-05, "loss": 0.5051, "step": 8786 }, { "epoch": 0.3130555605037676, "grad_norm": 1.2591963409382527, "learning_rate": 1.6084524085078392e-05, "loss": 0.5004, "step": 8787 }, { "epoch": 0.31309118763026167, "grad_norm": 1.146822892801752, "learning_rate": 1.6083608291527793e-05, "loss": 0.4672, "step": 8788 }, { "epoch": 0.3131268147567558, "grad_norm": 1.6110524174121827, "learning_rate": 1.6082692416969698e-05, "loss": 0.5486, "step": 8789 }, { "epoch": 0.31316244188324993, "grad_norm": 1.1171929200501958, "learning_rate": 1.6081776461416306e-05, "loss": 0.4707, "step": 8790 }, { "epoch": 0.313198069009744, "grad_norm": 1.213876749291318, "learning_rate": 1.6080860424879805e-05, "loss": 0.4997, "step": 8791 }, { "epoch": 0.31323369613623814, "grad_norm": 1.6216135096851807, "learning_rate": 1.60799443073724e-05, "loss": 0.5436, "step": 8792 }, { "epoch": 0.31326932326273227, "grad_norm": 1.1105861053138841, "learning_rate": 1.607902810890628e-05, "loss": 0.5105, "step": 8793 }, { "epoch": 0.31330495038922634, "grad_norm": 1.1717840263076447, "learning_rate": 1.6078111829493657e-05, "loss": 0.4799, "step": 8794 }, { "epoch": 0.3133405775157205, "grad_norm": 1.084626558947841, "learning_rate": 1.607719546914673e-05, "loss": 0.4939, "step": 8795 }, { "epoch": 0.3133762046422146, "grad_norm": 1.8477222061901868, "learning_rate": 1.6076279027877694e-05, "loss": 0.5885, "step": 8796 }, { "epoch": 0.3134118317687087, "grad_norm": 1.806748911355526, "learning_rate": 1.6075362505698756e-05, "loss": 0.5997, "step": 8797 }, { "epoch": 0.3134474588952028, "grad_norm": 1.1973783584301207, "learning_rate": 1.6074445902622117e-05, "loss": 0.5166, "step": 8798 }, { "epoch": 0.31348308602169694, "grad_norm": 1.1428480721305752, "learning_rate": 1.607352921865999e-05, "loss": 0.4946, "step": 8799 }, { "epoch": 0.313518713148191, "grad_norm": 1.5099186958035558, "learning_rate": 1.6072612453824575e-05, "loss": 0.5818, "step": 8800 }, { "epoch": 0.31355434027468515, "grad_norm": 1.2561194536986993, "learning_rate": 1.6071695608128077e-05, "loss": 0.4658, "step": 8801 }, { "epoch": 0.3135899674011793, "grad_norm": 1.3508196351102768, "learning_rate": 1.607077868158271e-05, "loss": 0.4741, "step": 8802 }, { "epoch": 0.31362559452767336, "grad_norm": 1.9679976843630307, "learning_rate": 1.6069861674200682e-05, "loss": 0.5907, "step": 8803 }, { "epoch": 0.3136612216541675, "grad_norm": 1.6769012699027372, "learning_rate": 1.6068944585994203e-05, "loss": 0.6503, "step": 8804 }, { "epoch": 0.3136968487806616, "grad_norm": 1.1600510885444546, "learning_rate": 1.606802741697549e-05, "loss": 0.4759, "step": 8805 }, { "epoch": 0.3137324759071557, "grad_norm": 1.5417390006651235, "learning_rate": 1.6067110167156743e-05, "loss": 0.7116, "step": 8806 }, { "epoch": 0.3137681030336498, "grad_norm": 2.0833040888389585, "learning_rate": 1.6066192836550186e-05, "loss": 0.5491, "step": 8807 }, { "epoch": 0.31380373016014396, "grad_norm": 1.6071863763029735, "learning_rate": 1.6065275425168034e-05, "loss": 0.5503, "step": 8808 }, { "epoch": 0.31383935728663803, "grad_norm": 1.4103458664660702, "learning_rate": 1.6064357933022496e-05, "loss": 0.51, "step": 8809 }, { "epoch": 0.31387498441313216, "grad_norm": 1.9769931696535812, "learning_rate": 1.6063440360125797e-05, "loss": 0.6244, "step": 8810 }, { "epoch": 0.3139106115396263, "grad_norm": 1.6953132793374492, "learning_rate": 1.6062522706490148e-05, "loss": 0.5513, "step": 8811 }, { "epoch": 0.31394623866612037, "grad_norm": 1.188899011397497, "learning_rate": 1.6061604972127774e-05, "loss": 0.496, "step": 8812 }, { "epoch": 0.3139818657926145, "grad_norm": 1.086262542464883, "learning_rate": 1.6060687157050894e-05, "loss": 0.4447, "step": 8813 }, { "epoch": 0.31401749291910863, "grad_norm": 1.1055219755767178, "learning_rate": 1.6059769261271727e-05, "loss": 0.5187, "step": 8814 }, { "epoch": 0.3140531200456027, "grad_norm": 1.4700008242521683, "learning_rate": 1.6058851284802498e-05, "loss": 0.5549, "step": 8815 }, { "epoch": 0.31408874717209684, "grad_norm": 1.1939636792770638, "learning_rate": 1.6057933227655427e-05, "loss": 0.4721, "step": 8816 }, { "epoch": 0.31412437429859097, "grad_norm": 1.5533799710597516, "learning_rate": 1.605701508984274e-05, "loss": 0.5941, "step": 8817 }, { "epoch": 0.31416000142508504, "grad_norm": 1.0561938730409508, "learning_rate": 1.6056096871376667e-05, "loss": 0.4756, "step": 8818 }, { "epoch": 0.3141956285515792, "grad_norm": 1.7180777626778092, "learning_rate": 1.6055178572269433e-05, "loss": 0.5137, "step": 8819 }, { "epoch": 0.3142312556780733, "grad_norm": 1.130692762899547, "learning_rate": 1.605426019253326e-05, "loss": 0.4986, "step": 8820 }, { "epoch": 0.3142668828045674, "grad_norm": 1.56439631781421, "learning_rate": 1.6053341732180382e-05, "loss": 0.4656, "step": 8821 }, { "epoch": 0.3143025099310615, "grad_norm": 1.4770273292334783, "learning_rate": 1.6052423191223026e-05, "loss": 0.6216, "step": 8822 }, { "epoch": 0.31433813705755564, "grad_norm": 1.1532514982400677, "learning_rate": 1.6051504569673426e-05, "loss": 0.4925, "step": 8823 }, { "epoch": 0.3143737641840497, "grad_norm": 1.827107835435231, "learning_rate": 1.6050585867543814e-05, "loss": 0.5662, "step": 8824 }, { "epoch": 0.31440939131054385, "grad_norm": 1.0501922632622793, "learning_rate": 1.604966708484642e-05, "loss": 0.466, "step": 8825 }, { "epoch": 0.314445018437038, "grad_norm": 1.363204915181126, "learning_rate": 1.604874822159348e-05, "loss": 0.4904, "step": 8826 }, { "epoch": 0.31448064556353206, "grad_norm": 1.751107637801631, "learning_rate": 1.604782927779723e-05, "loss": 0.6042, "step": 8827 }, { "epoch": 0.3145162726900262, "grad_norm": 1.1477912807655013, "learning_rate": 1.6046910253469908e-05, "loss": 0.498, "step": 8828 }, { "epoch": 0.3145518998165203, "grad_norm": 1.7495728137485835, "learning_rate": 1.6045991148623752e-05, "loss": 0.5689, "step": 8829 }, { "epoch": 0.3145875269430144, "grad_norm": 1.129306145501638, "learning_rate": 1.604507196327099e-05, "loss": 0.4897, "step": 8830 }, { "epoch": 0.3146231540695085, "grad_norm": 1.2325601969527062, "learning_rate": 1.6044152697423876e-05, "loss": 0.5003, "step": 8831 }, { "epoch": 0.31465878119600266, "grad_norm": 2.1683682161796827, "learning_rate": 1.604323335109464e-05, "loss": 0.5384, "step": 8832 }, { "epoch": 0.31469440832249673, "grad_norm": 1.847230374036277, "learning_rate": 1.6042313924295535e-05, "loss": 0.6267, "step": 8833 }, { "epoch": 0.31473003544899086, "grad_norm": 1.2398808981464482, "learning_rate": 1.6041394417038788e-05, "loss": 0.4846, "step": 8834 }, { "epoch": 0.314765662575485, "grad_norm": 1.095252187480164, "learning_rate": 1.604047482933666e-05, "loss": 0.4957, "step": 8835 }, { "epoch": 0.31480128970197907, "grad_norm": 2.297972869308704, "learning_rate": 1.603955516120138e-05, "loss": 0.5904, "step": 8836 }, { "epoch": 0.3148369168284732, "grad_norm": 1.2594933300671944, "learning_rate": 1.6038635412645204e-05, "loss": 0.5425, "step": 8837 }, { "epoch": 0.31487254395496733, "grad_norm": 1.1534121786225686, "learning_rate": 1.6037715583680377e-05, "loss": 0.4312, "step": 8838 }, { "epoch": 0.3149081710814614, "grad_norm": 1.2588851949507989, "learning_rate": 1.603679567431915e-05, "loss": 0.5056, "step": 8839 }, { "epoch": 0.31494379820795554, "grad_norm": 1.516896321551513, "learning_rate": 1.6035875684573764e-05, "loss": 0.5109, "step": 8840 }, { "epoch": 0.31497942533444967, "grad_norm": 1.3133320430465953, "learning_rate": 1.603495561445648e-05, "loss": 0.5257, "step": 8841 }, { "epoch": 0.31501505246094375, "grad_norm": 1.169825410373597, "learning_rate": 1.603403546397954e-05, "loss": 0.5166, "step": 8842 }, { "epoch": 0.3150506795874379, "grad_norm": 1.2893324132734456, "learning_rate": 1.6033115233155202e-05, "loss": 0.5314, "step": 8843 }, { "epoch": 0.315086306713932, "grad_norm": 1.1675163520109755, "learning_rate": 1.6032194921995718e-05, "loss": 0.4587, "step": 8844 }, { "epoch": 0.3151219338404261, "grad_norm": 1.1676125535926007, "learning_rate": 1.6031274530513342e-05, "loss": 0.495, "step": 8845 }, { "epoch": 0.3151575609669202, "grad_norm": 1.0209554437874988, "learning_rate": 1.6030354058720328e-05, "loss": 0.4479, "step": 8846 }, { "epoch": 0.31519318809341434, "grad_norm": 1.1823259422609091, "learning_rate": 1.6029433506628932e-05, "loss": 0.4842, "step": 8847 }, { "epoch": 0.3152288152199084, "grad_norm": 1.0400813706950685, "learning_rate": 1.6028512874251423e-05, "loss": 0.4521, "step": 8848 }, { "epoch": 0.31526444234640255, "grad_norm": 1.2079663680325685, "learning_rate": 1.6027592161600046e-05, "loss": 0.506, "step": 8849 }, { "epoch": 0.3153000694728967, "grad_norm": 1.89178211577414, "learning_rate": 1.6026671368687066e-05, "loss": 0.5453, "step": 8850 }, { "epoch": 0.31533569659939076, "grad_norm": 1.1458788292130981, "learning_rate": 1.6025750495524745e-05, "loss": 0.5344, "step": 8851 }, { "epoch": 0.3153713237258849, "grad_norm": 1.1488122341752465, "learning_rate": 1.6024829542125347e-05, "loss": 0.4847, "step": 8852 }, { "epoch": 0.315406950852379, "grad_norm": 1.562184483650575, "learning_rate": 1.602390850850113e-05, "loss": 0.5658, "step": 8853 }, { "epoch": 0.3154425779788731, "grad_norm": 1.42308956083089, "learning_rate": 1.602298739466436e-05, "loss": 0.5212, "step": 8854 }, { "epoch": 0.3154782051053672, "grad_norm": 1.1388746413225095, "learning_rate": 1.6022066200627306e-05, "loss": 0.5188, "step": 8855 }, { "epoch": 0.31551383223186136, "grad_norm": 1.744859018406173, "learning_rate": 1.602114492640223e-05, "loss": 0.6547, "step": 8856 }, { "epoch": 0.31554945935835543, "grad_norm": 1.221478037835506, "learning_rate": 1.6020223572001403e-05, "loss": 0.5245, "step": 8857 }, { "epoch": 0.31558508648484956, "grad_norm": 1.7580923383713527, "learning_rate": 1.6019302137437086e-05, "loss": 0.6103, "step": 8858 }, { "epoch": 0.3156207136113437, "grad_norm": 1.0841634009813539, "learning_rate": 1.6018380622721554e-05, "loss": 0.5259, "step": 8859 }, { "epoch": 0.31565634073783777, "grad_norm": 1.5880158216905296, "learning_rate": 1.601745902786708e-05, "loss": 0.5969, "step": 8860 }, { "epoch": 0.3156919678643319, "grad_norm": 1.122434647913232, "learning_rate": 1.6016537352885936e-05, "loss": 0.489, "step": 8861 }, { "epoch": 0.31572759499082603, "grad_norm": 1.865083892367244, "learning_rate": 1.601561559779039e-05, "loss": 0.595, "step": 8862 }, { "epoch": 0.3157632221173201, "grad_norm": 1.2237593628850774, "learning_rate": 1.6014693762592716e-05, "loss": 0.5206, "step": 8863 }, { "epoch": 0.31579884924381424, "grad_norm": 1.2906010007828572, "learning_rate": 1.6013771847305187e-05, "loss": 0.5563, "step": 8864 }, { "epoch": 0.31583447637030837, "grad_norm": 1.0872605458119133, "learning_rate": 1.6012849851940088e-05, "loss": 0.4954, "step": 8865 }, { "epoch": 0.31587010349680245, "grad_norm": 1.5405862560032515, "learning_rate": 1.601192777650969e-05, "loss": 0.5802, "step": 8866 }, { "epoch": 0.3159057306232966, "grad_norm": 1.0333644480694062, "learning_rate": 1.601100562102627e-05, "loss": 0.4441, "step": 8867 }, { "epoch": 0.3159413577497907, "grad_norm": 2.006490574097823, "learning_rate": 1.601008338550211e-05, "loss": 0.6069, "step": 8868 }, { "epoch": 0.3159769848762848, "grad_norm": 1.7753186468611537, "learning_rate": 1.6009161069949488e-05, "loss": 0.6276, "step": 8869 }, { "epoch": 0.3160126120027789, "grad_norm": 1.461610160129876, "learning_rate": 1.6008238674380688e-05, "loss": 0.5879, "step": 8870 }, { "epoch": 0.31604823912927305, "grad_norm": 1.7293280106904259, "learning_rate": 1.6007316198807986e-05, "loss": 0.5945, "step": 8871 }, { "epoch": 0.3160838662557671, "grad_norm": 1.549488047546829, "learning_rate": 1.6006393643243675e-05, "loss": 0.6136, "step": 8872 }, { "epoch": 0.31611949338226125, "grad_norm": 1.1142983165979605, "learning_rate": 1.6005471007700033e-05, "loss": 0.5112, "step": 8873 }, { "epoch": 0.3161551205087554, "grad_norm": 1.6364000452380945, "learning_rate": 1.6004548292189345e-05, "loss": 0.6058, "step": 8874 }, { "epoch": 0.31619074763524946, "grad_norm": 1.131607953059958, "learning_rate": 1.60036254967239e-05, "loss": 0.4822, "step": 8875 }, { "epoch": 0.3162263747617436, "grad_norm": 2.033920477422371, "learning_rate": 1.6002702621315988e-05, "loss": 0.6053, "step": 8876 }, { "epoch": 0.3162620018882377, "grad_norm": 1.472249973741664, "learning_rate": 1.6001779665977892e-05, "loss": 0.5362, "step": 8877 }, { "epoch": 0.3162976290147318, "grad_norm": 1.5992912514068556, "learning_rate": 1.6000856630721906e-05, "loss": 0.5399, "step": 8878 }, { "epoch": 0.3163332561412259, "grad_norm": 1.1850707715233608, "learning_rate": 1.599993351556032e-05, "loss": 0.4808, "step": 8879 }, { "epoch": 0.31636888326772006, "grad_norm": 1.09485314520998, "learning_rate": 1.5999010320505424e-05, "loss": 0.5107, "step": 8880 }, { "epoch": 0.31640451039421413, "grad_norm": 1.160139738313776, "learning_rate": 1.5998087045569515e-05, "loss": 0.4826, "step": 8881 }, { "epoch": 0.31644013752070826, "grad_norm": 1.1460314135729497, "learning_rate": 1.599716369076488e-05, "loss": 0.536, "step": 8882 }, { "epoch": 0.3164757646472024, "grad_norm": 1.5624868016812417, "learning_rate": 1.5996240256103824e-05, "loss": 0.5303, "step": 8883 }, { "epoch": 0.31651139177369647, "grad_norm": 1.4854233432611461, "learning_rate": 1.5995316741598636e-05, "loss": 0.573, "step": 8884 }, { "epoch": 0.3165470189001906, "grad_norm": 1.5203939582310801, "learning_rate": 1.5994393147261616e-05, "loss": 0.5064, "step": 8885 }, { "epoch": 0.31658264602668473, "grad_norm": 1.6328165153083858, "learning_rate": 1.599346947310506e-05, "loss": 0.5886, "step": 8886 }, { "epoch": 0.3166182731531788, "grad_norm": 1.6404987881158164, "learning_rate": 1.5992545719141268e-05, "loss": 0.5029, "step": 8887 }, { "epoch": 0.31665390027967294, "grad_norm": 1.1719958544245237, "learning_rate": 1.5991621885382542e-05, "loss": 0.4883, "step": 8888 }, { "epoch": 0.31668952740616707, "grad_norm": 1.2842222897496933, "learning_rate": 1.5990697971841183e-05, "loss": 0.5628, "step": 8889 }, { "epoch": 0.31672515453266115, "grad_norm": 1.9129321811744704, "learning_rate": 1.5989773978529496e-05, "loss": 0.6102, "step": 8890 }, { "epoch": 0.3167607816591553, "grad_norm": 2.096719111723016, "learning_rate": 1.5988849905459775e-05, "loss": 0.5381, "step": 8891 }, { "epoch": 0.3167964087856494, "grad_norm": 1.6084590785564319, "learning_rate": 1.5987925752644334e-05, "loss": 0.6634, "step": 8892 }, { "epoch": 0.3168320359121435, "grad_norm": 1.0766619036860998, "learning_rate": 1.598700152009548e-05, "loss": 0.4707, "step": 8893 }, { "epoch": 0.3168676630386376, "grad_norm": 6.255267664608629, "learning_rate": 1.5986077207825513e-05, "loss": 0.5998, "step": 8894 }, { "epoch": 0.31690329016513175, "grad_norm": 1.1785198509786794, "learning_rate": 1.5985152815846745e-05, "loss": 0.5324, "step": 8895 }, { "epoch": 0.3169389172916258, "grad_norm": 1.159235352219337, "learning_rate": 1.5984228344171484e-05, "loss": 0.4958, "step": 8896 }, { "epoch": 0.31697454441811995, "grad_norm": 1.398730804684951, "learning_rate": 1.598330379281204e-05, "loss": 0.5275, "step": 8897 }, { "epoch": 0.3170101715446141, "grad_norm": 1.1291865930986287, "learning_rate": 1.5982379161780722e-05, "loss": 0.4788, "step": 8898 }, { "epoch": 0.31704579867110816, "grad_norm": 1.4945156731755238, "learning_rate": 1.5981454451089846e-05, "loss": 0.6004, "step": 8899 }, { "epoch": 0.3170814257976023, "grad_norm": 1.3985930426007793, "learning_rate": 1.5980529660751723e-05, "loss": 0.5157, "step": 8900 }, { "epoch": 0.3171170529240964, "grad_norm": 1.1928136796706308, "learning_rate": 1.597960479077867e-05, "loss": 0.4759, "step": 8901 }, { "epoch": 0.3171526800505905, "grad_norm": 1.7136977984267212, "learning_rate": 1.5978679841182995e-05, "loss": 0.5052, "step": 8902 }, { "epoch": 0.3171883071770846, "grad_norm": 2.5149641408076167, "learning_rate": 1.597775481197702e-05, "loss": 0.6102, "step": 8903 }, { "epoch": 0.31722393430357876, "grad_norm": 1.734605827444211, "learning_rate": 1.5976829703173063e-05, "loss": 0.5367, "step": 8904 }, { "epoch": 0.31725956143007283, "grad_norm": 1.4701854801409378, "learning_rate": 1.5975904514783442e-05, "loss": 0.4714, "step": 8905 }, { "epoch": 0.31729518855656696, "grad_norm": 1.503156634341884, "learning_rate": 1.5974979246820472e-05, "loss": 0.5429, "step": 8906 }, { "epoch": 0.3173308156830611, "grad_norm": 1.2620084671171317, "learning_rate": 1.597405389929648e-05, "loss": 0.4533, "step": 8907 }, { "epoch": 0.31736644280955517, "grad_norm": 2.054992324252306, "learning_rate": 1.5973128472223783e-05, "loss": 0.5537, "step": 8908 }, { "epoch": 0.3174020699360493, "grad_norm": 2.0927665795774066, "learning_rate": 1.597220296561471e-05, "loss": 0.5392, "step": 8909 }, { "epoch": 0.31743769706254343, "grad_norm": 1.5081659933190723, "learning_rate": 1.5971277379481573e-05, "loss": 0.6265, "step": 8910 }, { "epoch": 0.3174733241890375, "grad_norm": 2.0089003197760156, "learning_rate": 1.597035171383671e-05, "loss": 0.6726, "step": 8911 }, { "epoch": 0.31750895131553164, "grad_norm": 1.2182680196546964, "learning_rate": 1.5969425968692437e-05, "loss": 0.5026, "step": 8912 }, { "epoch": 0.31754457844202577, "grad_norm": 1.4193241184900545, "learning_rate": 1.5968500144061085e-05, "loss": 0.508, "step": 8913 }, { "epoch": 0.31758020556851985, "grad_norm": 1.4059553169444963, "learning_rate": 1.5967574239954983e-05, "loss": 0.4968, "step": 8914 }, { "epoch": 0.317615832695014, "grad_norm": 1.5689150568996755, "learning_rate": 1.596664825638646e-05, "loss": 0.5, "step": 8915 }, { "epoch": 0.3176514598215081, "grad_norm": 1.5973398349824122, "learning_rate": 1.596572219336784e-05, "loss": 0.5256, "step": 8916 }, { "epoch": 0.3176870869480022, "grad_norm": 3.5818427168267775, "learning_rate": 1.5964796050911467e-05, "loss": 0.5473, "step": 8917 }, { "epoch": 0.3177227140744963, "grad_norm": 1.083578602773109, "learning_rate": 1.596386982902966e-05, "loss": 0.4889, "step": 8918 }, { "epoch": 0.31775834120099045, "grad_norm": 1.1980545046455593, "learning_rate": 1.5962943527734757e-05, "loss": 0.4491, "step": 8919 }, { "epoch": 0.3177939683274845, "grad_norm": 3.3046474376822315, "learning_rate": 1.5962017147039096e-05, "loss": 0.4867, "step": 8920 }, { "epoch": 0.31782959545397865, "grad_norm": 2.578128986492339, "learning_rate": 1.596109068695501e-05, "loss": 0.6505, "step": 8921 }, { "epoch": 0.3178652225804728, "grad_norm": 2.3769083044257115, "learning_rate": 1.5960164147494835e-05, "loss": 0.5592, "step": 8922 }, { "epoch": 0.31790084970696686, "grad_norm": 1.278014966387428, "learning_rate": 1.5959237528670907e-05, "loss": 0.4921, "step": 8923 }, { "epoch": 0.317936476833461, "grad_norm": 1.5763565489752684, "learning_rate": 1.5958310830495567e-05, "loss": 0.6144, "step": 8924 }, { "epoch": 0.3179721039599551, "grad_norm": 2.003344133265046, "learning_rate": 1.5957384052981152e-05, "loss": 0.6033, "step": 8925 }, { "epoch": 0.3180077310864492, "grad_norm": 1.15683097310229, "learning_rate": 1.5956457196140008e-05, "loss": 0.4659, "step": 8926 }, { "epoch": 0.31804335821294333, "grad_norm": 1.154534285783258, "learning_rate": 1.595553025998447e-05, "loss": 0.5421, "step": 8927 }, { "epoch": 0.31807898533943746, "grad_norm": 1.3358991783759022, "learning_rate": 1.595460324452688e-05, "loss": 0.5817, "step": 8928 }, { "epoch": 0.31811461246593153, "grad_norm": 1.229846931986318, "learning_rate": 1.595367614977959e-05, "loss": 0.503, "step": 8929 }, { "epoch": 0.31815023959242567, "grad_norm": 1.1582811775179391, "learning_rate": 1.5952748975754944e-05, "loss": 0.5309, "step": 8930 }, { "epoch": 0.3181858667189198, "grad_norm": 2.2731522744298323, "learning_rate": 1.595182172246528e-05, "loss": 0.5426, "step": 8931 }, { "epoch": 0.31822149384541387, "grad_norm": 1.2792513812574482, "learning_rate": 1.5950894389922954e-05, "loss": 0.5321, "step": 8932 }, { "epoch": 0.318257120971908, "grad_norm": 1.1284684696743998, "learning_rate": 1.5949966978140303e-05, "loss": 0.4925, "step": 8933 }, { "epoch": 0.31829274809840213, "grad_norm": 1.5690171742498418, "learning_rate": 1.5949039487129687e-05, "loss": 0.6259, "step": 8934 }, { "epoch": 0.3183283752248962, "grad_norm": 1.8881671122395656, "learning_rate": 1.5948111916903456e-05, "loss": 0.6364, "step": 8935 }, { "epoch": 0.31836400235139034, "grad_norm": 1.1370914271815162, "learning_rate": 1.594718426747395e-05, "loss": 0.4754, "step": 8936 }, { "epoch": 0.31839962947788447, "grad_norm": 1.2245605712511687, "learning_rate": 1.594625653885353e-05, "loss": 0.4779, "step": 8937 }, { "epoch": 0.31843525660437855, "grad_norm": 1.7496446362601947, "learning_rate": 1.594532873105455e-05, "loss": 0.5719, "step": 8938 }, { "epoch": 0.3184708837308727, "grad_norm": 1.7122416536471008, "learning_rate": 1.5944400844089365e-05, "loss": 0.5882, "step": 8939 }, { "epoch": 0.3185065108573668, "grad_norm": 1.7422913486373146, "learning_rate": 1.5943472877970325e-05, "loss": 0.6176, "step": 8940 }, { "epoch": 0.3185421379838609, "grad_norm": 1.089483817534556, "learning_rate": 1.594254483270979e-05, "loss": 0.456, "step": 8941 }, { "epoch": 0.318577765110355, "grad_norm": 1.1341741860066656, "learning_rate": 1.5941616708320115e-05, "loss": 0.475, "step": 8942 }, { "epoch": 0.31861339223684915, "grad_norm": 1.228388061395651, "learning_rate": 1.5940688504813664e-05, "loss": 0.4779, "step": 8943 }, { "epoch": 0.3186490193633432, "grad_norm": 1.8088197190987907, "learning_rate": 1.593976022220279e-05, "loss": 0.6458, "step": 8944 }, { "epoch": 0.31868464648983735, "grad_norm": 1.0096528415194341, "learning_rate": 1.593883186049986e-05, "loss": 0.4742, "step": 8945 }, { "epoch": 0.3187202736163315, "grad_norm": 1.6163763837171956, "learning_rate": 1.593790341971723e-05, "loss": 0.5499, "step": 8946 }, { "epoch": 0.31875590074282556, "grad_norm": 1.631931542705053, "learning_rate": 1.5936974899867267e-05, "loss": 0.5784, "step": 8947 }, { "epoch": 0.3187915278693197, "grad_norm": 1.3549926284690346, "learning_rate": 1.5936046300962336e-05, "loss": 0.46, "step": 8948 }, { "epoch": 0.3188271549958138, "grad_norm": 1.1923270333029288, "learning_rate": 1.5935117623014795e-05, "loss": 0.5538, "step": 8949 }, { "epoch": 0.31886278212230795, "grad_norm": 1.095834096917316, "learning_rate": 1.5934188866037017e-05, "loss": 0.4816, "step": 8950 }, { "epoch": 0.31889840924880203, "grad_norm": 2.1450498947133005, "learning_rate": 1.5933260030041364e-05, "loss": 0.604, "step": 8951 }, { "epoch": 0.31893403637529616, "grad_norm": 1.2782141676225882, "learning_rate": 1.5932331115040208e-05, "loss": 0.5121, "step": 8952 }, { "epoch": 0.3189696635017903, "grad_norm": 1.3942513561792982, "learning_rate": 1.5931402121045918e-05, "loss": 0.5989, "step": 8953 }, { "epoch": 0.31900529062828437, "grad_norm": 2.2366602928703316, "learning_rate": 1.5930473048070863e-05, "loss": 0.6137, "step": 8954 }, { "epoch": 0.3190409177547785, "grad_norm": 1.2876174369975428, "learning_rate": 1.5929543896127412e-05, "loss": 0.4971, "step": 8955 }, { "epoch": 0.31907654488127263, "grad_norm": 1.4676106556615511, "learning_rate": 1.592861466522794e-05, "loss": 0.6723, "step": 8956 }, { "epoch": 0.3191121720077667, "grad_norm": 1.0450566721275951, "learning_rate": 1.5927685355384822e-05, "loss": 0.4445, "step": 8957 }, { "epoch": 0.31914779913426083, "grad_norm": 1.6872783035262413, "learning_rate": 1.5926755966610426e-05, "loss": 0.6411, "step": 8958 }, { "epoch": 0.31918342626075497, "grad_norm": 2.0461905469330257, "learning_rate": 1.5925826498917138e-05, "loss": 0.54, "step": 8959 }, { "epoch": 0.31921905338724904, "grad_norm": 1.2316219905669463, "learning_rate": 1.5924896952317323e-05, "loss": 0.4978, "step": 8960 }, { "epoch": 0.31925468051374317, "grad_norm": 1.5597501041286859, "learning_rate": 1.592396732682337e-05, "loss": 0.597, "step": 8961 }, { "epoch": 0.3192903076402373, "grad_norm": 2.163284528086541, "learning_rate": 1.5923037622447644e-05, "loss": 0.6897, "step": 8962 }, { "epoch": 0.3193259347667314, "grad_norm": 1.1861114535768484, "learning_rate": 1.5922107839202535e-05, "loss": 0.5407, "step": 8963 }, { "epoch": 0.3193615618932255, "grad_norm": 1.2268488127074266, "learning_rate": 1.5921177977100423e-05, "loss": 0.4951, "step": 8964 }, { "epoch": 0.31939718901971964, "grad_norm": 1.2211380618326853, "learning_rate": 1.5920248036153685e-05, "loss": 0.5299, "step": 8965 }, { "epoch": 0.3194328161462137, "grad_norm": 1.2386756436663084, "learning_rate": 1.5919318016374707e-05, "loss": 0.4899, "step": 8966 }, { "epoch": 0.31946844327270785, "grad_norm": 1.7715734001249346, "learning_rate": 1.5918387917775873e-05, "loss": 0.5421, "step": 8967 }, { "epoch": 0.319504070399202, "grad_norm": 2.128658606582575, "learning_rate": 1.5917457740369565e-05, "loss": 0.5786, "step": 8968 }, { "epoch": 0.31953969752569605, "grad_norm": 1.5084369292138835, "learning_rate": 1.5916527484168173e-05, "loss": 0.5882, "step": 8969 }, { "epoch": 0.3195753246521902, "grad_norm": 1.4407548648780573, "learning_rate": 1.5915597149184083e-05, "loss": 0.5363, "step": 8970 }, { "epoch": 0.3196109517786843, "grad_norm": 2.4949484217938003, "learning_rate": 1.591466673542968e-05, "loss": 0.5989, "step": 8971 }, { "epoch": 0.3196465789051784, "grad_norm": 1.203973871213644, "learning_rate": 1.591373624291736e-05, "loss": 0.5308, "step": 8972 }, { "epoch": 0.3196822060316725, "grad_norm": 1.8754981464304983, "learning_rate": 1.59128056716595e-05, "loss": 0.5425, "step": 8973 }, { "epoch": 0.31971783315816665, "grad_norm": 1.2288319563480095, "learning_rate": 1.5911875021668506e-05, "loss": 0.4832, "step": 8974 }, { "epoch": 0.31975346028466073, "grad_norm": 1.1574309287142939, "learning_rate": 1.5910944292956764e-05, "loss": 0.5414, "step": 8975 }, { "epoch": 0.31978908741115486, "grad_norm": 1.7196032313061864, "learning_rate": 1.5910013485536664e-05, "loss": 0.5549, "step": 8976 }, { "epoch": 0.319824714537649, "grad_norm": 1.4644645889145327, "learning_rate": 1.5909082599420606e-05, "loss": 0.5712, "step": 8977 }, { "epoch": 0.31986034166414307, "grad_norm": 1.2267168865622415, "learning_rate": 1.5908151634620985e-05, "loss": 0.5153, "step": 8978 }, { "epoch": 0.3198959687906372, "grad_norm": 1.9944418001513176, "learning_rate": 1.5907220591150186e-05, "loss": 0.6084, "step": 8979 }, { "epoch": 0.31993159591713133, "grad_norm": 1.105380315626386, "learning_rate": 1.5906289469020626e-05, "loss": 0.4437, "step": 8980 }, { "epoch": 0.3199672230436254, "grad_norm": 2.763127110629031, "learning_rate": 1.590535826824469e-05, "loss": 0.5481, "step": 8981 }, { "epoch": 0.32000285017011953, "grad_norm": 1.1193062663252538, "learning_rate": 1.590442698883478e-05, "loss": 0.4706, "step": 8982 }, { "epoch": 0.32003847729661367, "grad_norm": 2.0010771994403482, "learning_rate": 1.5903495630803302e-05, "loss": 0.6166, "step": 8983 }, { "epoch": 0.32007410442310774, "grad_norm": 1.4424448664470406, "learning_rate": 1.590256419416265e-05, "loss": 0.676, "step": 8984 }, { "epoch": 0.3201097315496019, "grad_norm": 1.186138551837288, "learning_rate": 1.590163267892523e-05, "loss": 0.5327, "step": 8985 }, { "epoch": 0.320145358676096, "grad_norm": 1.1614102317225856, "learning_rate": 1.5900701085103444e-05, "loss": 0.4914, "step": 8986 }, { "epoch": 0.3201809858025901, "grad_norm": 1.3060058983776959, "learning_rate": 1.5899769412709704e-05, "loss": 0.5317, "step": 8987 }, { "epoch": 0.3202166129290842, "grad_norm": 1.0849729552286116, "learning_rate": 1.5898837661756405e-05, "loss": 0.4686, "step": 8988 }, { "epoch": 0.32025224005557834, "grad_norm": 1.547212067321967, "learning_rate": 1.5897905832255962e-05, "loss": 0.5887, "step": 8989 }, { "epoch": 0.3202878671820724, "grad_norm": 1.1496997588697755, "learning_rate": 1.589697392422078e-05, "loss": 0.5183, "step": 8990 }, { "epoch": 0.32032349430856655, "grad_norm": 1.1843444246425467, "learning_rate": 1.589604193766327e-05, "loss": 0.4979, "step": 8991 }, { "epoch": 0.3203591214350607, "grad_norm": 1.0252732526919617, "learning_rate": 1.5895109872595837e-05, "loss": 0.5057, "step": 8992 }, { "epoch": 0.32039474856155475, "grad_norm": 1.1808987220067704, "learning_rate": 1.58941777290309e-05, "loss": 0.462, "step": 8993 }, { "epoch": 0.3204303756880489, "grad_norm": 1.5141451977433449, "learning_rate": 1.5893245506980866e-05, "loss": 0.587, "step": 8994 }, { "epoch": 0.320466002814543, "grad_norm": 1.3465761630134858, "learning_rate": 1.5892313206458148e-05, "loss": 0.4959, "step": 8995 }, { "epoch": 0.3205016299410371, "grad_norm": 1.1191408101092, "learning_rate": 1.589138082747516e-05, "loss": 0.5218, "step": 8996 }, { "epoch": 0.3205372570675312, "grad_norm": 1.4910472894321383, "learning_rate": 1.5890448370044318e-05, "loss": 0.5809, "step": 8997 }, { "epoch": 0.32057288419402535, "grad_norm": 1.695667178896448, "learning_rate": 1.588951583417804e-05, "loss": 0.5965, "step": 8998 }, { "epoch": 0.32060851132051943, "grad_norm": 1.105092506387224, "learning_rate": 1.5888583219888743e-05, "loss": 0.4913, "step": 8999 }, { "epoch": 0.32064413844701356, "grad_norm": 1.2645962259001566, "learning_rate": 1.5887650527188844e-05, "loss": 0.5256, "step": 9000 }, { "epoch": 0.3206797655735077, "grad_norm": 1.5813379007481725, "learning_rate": 1.5886717756090765e-05, "loss": 0.4984, "step": 9001 }, { "epoch": 0.32071539270000177, "grad_norm": 1.828971630257236, "learning_rate": 1.588578490660692e-05, "loss": 0.611, "step": 9002 }, { "epoch": 0.3207510198264959, "grad_norm": 1.2031438234272067, "learning_rate": 1.588485197874974e-05, "loss": 0.5399, "step": 9003 }, { "epoch": 0.32078664695299003, "grad_norm": 1.9673328865039792, "learning_rate": 1.5883918972531637e-05, "loss": 0.5737, "step": 9004 }, { "epoch": 0.3208222740794841, "grad_norm": 1.4342429263368246, "learning_rate": 1.5882985887965045e-05, "loss": 0.5828, "step": 9005 }, { "epoch": 0.32085790120597824, "grad_norm": 1.2843015527538748, "learning_rate": 1.5882052725062382e-05, "loss": 0.553, "step": 9006 }, { "epoch": 0.32089352833247237, "grad_norm": 1.0313894638139587, "learning_rate": 1.5881119483836073e-05, "loss": 0.4537, "step": 9007 }, { "epoch": 0.32092915545896644, "grad_norm": 1.1202378148353964, "learning_rate": 1.5880186164298553e-05, "loss": 0.4636, "step": 9008 }, { "epoch": 0.3209647825854606, "grad_norm": 1.183282602302314, "learning_rate": 1.5879252766462243e-05, "loss": 0.4802, "step": 9009 }, { "epoch": 0.3210004097119547, "grad_norm": 1.763213738863808, "learning_rate": 1.587831929033957e-05, "loss": 0.6249, "step": 9010 }, { "epoch": 0.3210360368384488, "grad_norm": 1.1170262676238063, "learning_rate": 1.5877385735942967e-05, "loss": 0.4527, "step": 9011 }, { "epoch": 0.3210716639649429, "grad_norm": 1.1890360474055972, "learning_rate": 1.5876452103284868e-05, "loss": 0.4869, "step": 9012 }, { "epoch": 0.32110729109143704, "grad_norm": 1.539920849998869, "learning_rate": 1.58755183923777e-05, "loss": 0.6427, "step": 9013 }, { "epoch": 0.3211429182179311, "grad_norm": 1.7560657529978132, "learning_rate": 1.58745846032339e-05, "loss": 0.6536, "step": 9014 }, { "epoch": 0.32117854534442525, "grad_norm": 1.113967321511473, "learning_rate": 1.5873650735865894e-05, "loss": 0.4858, "step": 9015 }, { "epoch": 0.3212141724709194, "grad_norm": 1.0785210036322805, "learning_rate": 1.587271679028613e-05, "loss": 0.4552, "step": 9016 }, { "epoch": 0.32124979959741345, "grad_norm": 2.160921307201874, "learning_rate": 1.5871782766507033e-05, "loss": 0.526, "step": 9017 }, { "epoch": 0.3212854267239076, "grad_norm": 1.6762775337954183, "learning_rate": 1.5870848664541046e-05, "loss": 0.6383, "step": 9018 }, { "epoch": 0.3213210538504017, "grad_norm": 1.6552106304645027, "learning_rate": 1.5869914484400606e-05, "loss": 0.612, "step": 9019 }, { "epoch": 0.3213566809768958, "grad_norm": 1.2238247490558245, "learning_rate": 1.586898022609815e-05, "loss": 0.4819, "step": 9020 }, { "epoch": 0.3213923081033899, "grad_norm": 1.2468585363030378, "learning_rate": 1.586804588964612e-05, "loss": 0.4964, "step": 9021 }, { "epoch": 0.32142793522988405, "grad_norm": 1.1805064042023334, "learning_rate": 1.5867111475056957e-05, "loss": 0.4995, "step": 9022 }, { "epoch": 0.32146356235637813, "grad_norm": 2.154405870800773, "learning_rate": 1.5866176982343108e-05, "loss": 0.5581, "step": 9023 }, { "epoch": 0.32149918948287226, "grad_norm": 1.1729833712268323, "learning_rate": 1.586524241151701e-05, "loss": 0.4939, "step": 9024 }, { "epoch": 0.3215348166093664, "grad_norm": 1.2285476609800317, "learning_rate": 1.586430776259111e-05, "loss": 0.4715, "step": 9025 }, { "epoch": 0.32157044373586047, "grad_norm": 1.1360981215012178, "learning_rate": 1.586337303557785e-05, "loss": 0.4845, "step": 9026 }, { "epoch": 0.3216060708623546, "grad_norm": 1.6651771706058092, "learning_rate": 1.5862438230489682e-05, "loss": 0.5992, "step": 9027 }, { "epoch": 0.32164169798884873, "grad_norm": 1.0833657868109696, "learning_rate": 1.5861503347339052e-05, "loss": 0.517, "step": 9028 }, { "epoch": 0.3216773251153428, "grad_norm": 1.5306091470088117, "learning_rate": 1.5860568386138408e-05, "loss": 0.5635, "step": 9029 }, { "epoch": 0.32171295224183694, "grad_norm": 1.043827638829031, "learning_rate": 1.5859633346900202e-05, "loss": 0.4395, "step": 9030 }, { "epoch": 0.32174857936833107, "grad_norm": 1.5714853604206005, "learning_rate": 1.585869822963688e-05, "loss": 0.5485, "step": 9031 }, { "epoch": 0.32178420649482514, "grad_norm": 1.9375091950339949, "learning_rate": 1.5857763034360892e-05, "loss": 0.5424, "step": 9032 }, { "epoch": 0.3218198336213193, "grad_norm": 1.5704582373046907, "learning_rate": 1.5856827761084698e-05, "loss": 0.6039, "step": 9033 }, { "epoch": 0.3218554607478134, "grad_norm": 1.0730121208316081, "learning_rate": 1.585589240982075e-05, "loss": 0.5063, "step": 9034 }, { "epoch": 0.3218910878743075, "grad_norm": 1.7915988514781804, "learning_rate": 1.58549569805815e-05, "loss": 0.6795, "step": 9035 }, { "epoch": 0.3219267150008016, "grad_norm": 1.3770315285076116, "learning_rate": 1.5854021473379408e-05, "loss": 0.5785, "step": 9036 }, { "epoch": 0.32196234212729574, "grad_norm": 1.065815874019045, "learning_rate": 1.5853085888226926e-05, "loss": 0.4843, "step": 9037 }, { "epoch": 0.3219979692537898, "grad_norm": 1.8897119867390526, "learning_rate": 1.585215022513652e-05, "loss": 0.6729, "step": 9038 }, { "epoch": 0.32203359638028395, "grad_norm": 1.1874835006160138, "learning_rate": 1.5851214484120637e-05, "loss": 0.528, "step": 9039 }, { "epoch": 0.3220692235067781, "grad_norm": 1.8021972455248947, "learning_rate": 1.585027866519174e-05, "loss": 0.635, "step": 9040 }, { "epoch": 0.32210485063327216, "grad_norm": 1.4500068119103995, "learning_rate": 1.58493427683623e-05, "loss": 0.5552, "step": 9041 }, { "epoch": 0.3221404777597663, "grad_norm": 1.1989306464138183, "learning_rate": 1.584840679364477e-05, "loss": 0.516, "step": 9042 }, { "epoch": 0.3221761048862604, "grad_norm": 1.5946197572229166, "learning_rate": 1.5847470741051617e-05, "loss": 0.6358, "step": 9043 }, { "epoch": 0.3222117320127545, "grad_norm": 1.31345915734874, "learning_rate": 1.5846534610595306e-05, "loss": 0.4973, "step": 9044 }, { "epoch": 0.3222473591392486, "grad_norm": 1.2832913297282778, "learning_rate": 1.5845598402288297e-05, "loss": 0.5451, "step": 9045 }, { "epoch": 0.32228298626574275, "grad_norm": 1.5373110217501569, "learning_rate": 1.5844662116143057e-05, "loss": 0.4848, "step": 9046 }, { "epoch": 0.32231861339223683, "grad_norm": 1.79076371418134, "learning_rate": 1.584372575217206e-05, "loss": 0.5946, "step": 9047 }, { "epoch": 0.32235424051873096, "grad_norm": 1.2059775395584222, "learning_rate": 1.584278931038777e-05, "loss": 0.5134, "step": 9048 }, { "epoch": 0.3223898676452251, "grad_norm": 1.233898261646432, "learning_rate": 1.5841852790802655e-05, "loss": 0.4999, "step": 9049 }, { "epoch": 0.32242549477171917, "grad_norm": 1.1855647853931937, "learning_rate": 1.584091619342919e-05, "loss": 0.5059, "step": 9050 }, { "epoch": 0.3224611218982133, "grad_norm": 1.2402268416008724, "learning_rate": 1.5839979518279836e-05, "loss": 0.5152, "step": 9051 }, { "epoch": 0.32249674902470743, "grad_norm": 1.8327824151329837, "learning_rate": 1.583904276536708e-05, "loss": 0.5731, "step": 9052 }, { "epoch": 0.3225323761512015, "grad_norm": 1.5013958283356794, "learning_rate": 1.5838105934703385e-05, "loss": 0.6606, "step": 9053 }, { "epoch": 0.32256800327769564, "grad_norm": 1.453185611090331, "learning_rate": 1.583716902630123e-05, "loss": 0.492, "step": 9054 }, { "epoch": 0.32260363040418977, "grad_norm": 1.5514892430210094, "learning_rate": 1.583623204017309e-05, "loss": 0.6076, "step": 9055 }, { "epoch": 0.32263925753068384, "grad_norm": 1.3650385009307782, "learning_rate": 1.583529497633144e-05, "loss": 0.6264, "step": 9056 }, { "epoch": 0.322674884657178, "grad_norm": 2.371900545122227, "learning_rate": 1.583435783478876e-05, "loss": 0.6179, "step": 9057 }, { "epoch": 0.3227105117836721, "grad_norm": 1.5232953087496939, "learning_rate": 1.5833420615557523e-05, "loss": 0.6336, "step": 9058 }, { "epoch": 0.3227461389101662, "grad_norm": 1.58117129279997, "learning_rate": 1.583248331865022e-05, "loss": 0.6094, "step": 9059 }, { "epoch": 0.3227817660366603, "grad_norm": 1.6479283048937785, "learning_rate": 1.583154594407932e-05, "loss": 0.6402, "step": 9060 }, { "epoch": 0.32281739316315444, "grad_norm": 1.1277905736728868, "learning_rate": 1.583060849185731e-05, "loss": 0.4783, "step": 9061 }, { "epoch": 0.3228530202896485, "grad_norm": 1.192366639578619, "learning_rate": 1.5829670961996673e-05, "loss": 0.4829, "step": 9062 }, { "epoch": 0.32288864741614265, "grad_norm": 1.1738844144536307, "learning_rate": 1.5828733354509895e-05, "loss": 0.4934, "step": 9063 }, { "epoch": 0.3229242745426368, "grad_norm": 1.1820935448840963, "learning_rate": 1.5827795669409458e-05, "loss": 0.5011, "step": 9064 }, { "epoch": 0.32295990166913086, "grad_norm": 1.5212239632889617, "learning_rate": 1.5826857906707846e-05, "loss": 0.5792, "step": 9065 }, { "epoch": 0.322995528795625, "grad_norm": 1.0590992660814775, "learning_rate": 1.582592006641755e-05, "loss": 0.463, "step": 9066 }, { "epoch": 0.3230311559221191, "grad_norm": 1.0575768693080396, "learning_rate": 1.5824982148551055e-05, "loss": 0.4978, "step": 9067 }, { "epoch": 0.3230667830486132, "grad_norm": 2.678248596988964, "learning_rate": 1.5824044153120853e-05, "loss": 0.5281, "step": 9068 }, { "epoch": 0.3231024101751073, "grad_norm": 1.8442929023416523, "learning_rate": 1.582310608013943e-05, "loss": 0.5523, "step": 9069 }, { "epoch": 0.32313803730160146, "grad_norm": 1.7779156058885388, "learning_rate": 1.5822167929619282e-05, "loss": 0.6854, "step": 9070 }, { "epoch": 0.32317366442809553, "grad_norm": 1.1694030089917735, "learning_rate": 1.5821229701572897e-05, "loss": 0.4715, "step": 9071 }, { "epoch": 0.32320929155458966, "grad_norm": 1.3761968618372358, "learning_rate": 1.582029139601277e-05, "loss": 0.5367, "step": 9072 }, { "epoch": 0.3232449186810838, "grad_norm": 1.9054449580077732, "learning_rate": 1.5819353012951392e-05, "loss": 0.5653, "step": 9073 }, { "epoch": 0.32328054580757787, "grad_norm": 1.083906164504869, "learning_rate": 1.5818414552401264e-05, "loss": 0.4811, "step": 9074 }, { "epoch": 0.323316172934072, "grad_norm": 1.963291583429146, "learning_rate": 1.581747601437488e-05, "loss": 0.6352, "step": 9075 }, { "epoch": 0.32335180006056613, "grad_norm": 2.044407361379737, "learning_rate": 1.5816537398884737e-05, "loss": 0.6018, "step": 9076 }, { "epoch": 0.3233874271870602, "grad_norm": 1.3124746443457955, "learning_rate": 1.5815598705943333e-05, "loss": 0.5575, "step": 9077 }, { "epoch": 0.32342305431355434, "grad_norm": 1.4608259488947133, "learning_rate": 1.5814659935563165e-05, "loss": 0.6612, "step": 9078 }, { "epoch": 0.32345868144004847, "grad_norm": 1.6970922870253995, "learning_rate": 1.5813721087756735e-05, "loss": 0.6087, "step": 9079 }, { "epoch": 0.32349430856654254, "grad_norm": 1.1417107528644737, "learning_rate": 1.581278216253655e-05, "loss": 0.4956, "step": 9080 }, { "epoch": 0.3235299356930367, "grad_norm": 1.8598642501963691, "learning_rate": 1.5811843159915105e-05, "loss": 0.6447, "step": 9081 }, { "epoch": 0.3235655628195308, "grad_norm": 1.367037305611859, "learning_rate": 1.5810904079904906e-05, "loss": 0.5002, "step": 9082 }, { "epoch": 0.3236011899460249, "grad_norm": 1.695467110406348, "learning_rate": 1.5809964922518458e-05, "loss": 0.5543, "step": 9083 }, { "epoch": 0.323636817072519, "grad_norm": 2.50027782745792, "learning_rate": 1.5809025687768265e-05, "loss": 0.5911, "step": 9084 }, { "epoch": 0.32367244419901314, "grad_norm": 1.0517409288509916, "learning_rate": 1.5808086375666834e-05, "loss": 0.4925, "step": 9085 }, { "epoch": 0.3237080713255072, "grad_norm": 1.2286340231179471, "learning_rate": 1.5807146986226677e-05, "loss": 0.4865, "step": 9086 }, { "epoch": 0.32374369845200135, "grad_norm": 1.2461827142898325, "learning_rate": 1.5806207519460296e-05, "loss": 0.5406, "step": 9087 }, { "epoch": 0.3237793255784955, "grad_norm": 1.504729059303651, "learning_rate": 1.5805267975380206e-05, "loss": 0.5283, "step": 9088 }, { "epoch": 0.32381495270498956, "grad_norm": 1.6501098292450327, "learning_rate": 1.580432835399891e-05, "loss": 0.6074, "step": 9089 }, { "epoch": 0.3238505798314837, "grad_norm": 1.857799283653531, "learning_rate": 1.580338865532893e-05, "loss": 0.5901, "step": 9090 }, { "epoch": 0.3238862069579778, "grad_norm": 1.7507569460305648, "learning_rate": 1.5802448879382773e-05, "loss": 0.5941, "step": 9091 }, { "epoch": 0.3239218340844719, "grad_norm": 1.5266199595554129, "learning_rate": 1.580150902617295e-05, "loss": 0.595, "step": 9092 }, { "epoch": 0.323957461210966, "grad_norm": 1.428417685950892, "learning_rate": 1.5800569095711983e-05, "loss": 0.4876, "step": 9093 }, { "epoch": 0.32399308833746016, "grad_norm": 1.2561722107649629, "learning_rate": 1.5799629088012384e-05, "loss": 0.4741, "step": 9094 }, { "epoch": 0.32402871546395423, "grad_norm": 1.3752181964410015, "learning_rate": 1.5798689003086668e-05, "loss": 0.5929, "step": 9095 }, { "epoch": 0.32406434259044836, "grad_norm": 1.3968959800613523, "learning_rate": 1.579774884094736e-05, "loss": 0.5612, "step": 9096 }, { "epoch": 0.3240999697169425, "grad_norm": 1.4914796903836014, "learning_rate": 1.5796808601606966e-05, "loss": 0.5444, "step": 9097 }, { "epoch": 0.32413559684343657, "grad_norm": 1.2250570005830068, "learning_rate": 1.579586828507802e-05, "loss": 0.5352, "step": 9098 }, { "epoch": 0.3241712239699307, "grad_norm": 2.0484035298518535, "learning_rate": 1.5794927891373032e-05, "loss": 0.5752, "step": 9099 }, { "epoch": 0.32420685109642483, "grad_norm": 2.059287239540848, "learning_rate": 1.5793987420504532e-05, "loss": 0.5768, "step": 9100 }, { "epoch": 0.3242424782229189, "grad_norm": 1.575635526930699, "learning_rate": 1.5793046872485036e-05, "loss": 0.5567, "step": 9101 }, { "epoch": 0.32427810534941304, "grad_norm": 1.895170323594798, "learning_rate": 1.579210624732708e-05, "loss": 0.5028, "step": 9102 }, { "epoch": 0.32431373247590717, "grad_norm": 1.5419261799682025, "learning_rate": 1.5791165545043174e-05, "loss": 0.6003, "step": 9103 }, { "epoch": 0.32434935960240124, "grad_norm": 1.360827656556281, "learning_rate": 1.579022476564585e-05, "loss": 0.5634, "step": 9104 }, { "epoch": 0.3243849867288954, "grad_norm": 1.3890316746478077, "learning_rate": 1.578928390914764e-05, "loss": 0.5585, "step": 9105 }, { "epoch": 0.3244206138553895, "grad_norm": 1.2270569832317526, "learning_rate": 1.5788342975561067e-05, "loss": 0.4454, "step": 9106 }, { "epoch": 0.3244562409818836, "grad_norm": 1.7054005530078478, "learning_rate": 1.578740196489866e-05, "loss": 0.5843, "step": 9107 }, { "epoch": 0.3244918681083777, "grad_norm": 1.495450036768906, "learning_rate": 1.5786460877172954e-05, "loss": 0.4832, "step": 9108 }, { "epoch": 0.32452749523487184, "grad_norm": 1.1635356570749915, "learning_rate": 1.5785519712396473e-05, "loss": 0.5043, "step": 9109 }, { "epoch": 0.3245631223613659, "grad_norm": 1.4393720646041335, "learning_rate": 1.5784578470581755e-05, "loss": 0.6283, "step": 9110 }, { "epoch": 0.32459874948786005, "grad_norm": 1.4684204607783873, "learning_rate": 1.5783637151741335e-05, "loss": 0.5229, "step": 9111 }, { "epoch": 0.3246343766143542, "grad_norm": 1.2519466894999365, "learning_rate": 1.5782695755887737e-05, "loss": 0.5359, "step": 9112 }, { "epoch": 0.32467000374084826, "grad_norm": 1.1076952293999855, "learning_rate": 1.5781754283033504e-05, "loss": 0.4529, "step": 9113 }, { "epoch": 0.3247056308673424, "grad_norm": 1.1110444080151631, "learning_rate": 1.5780812733191176e-05, "loss": 0.5115, "step": 9114 }, { "epoch": 0.3247412579938365, "grad_norm": 1.1595956386776989, "learning_rate": 1.5779871106373283e-05, "loss": 0.5204, "step": 9115 }, { "epoch": 0.3247768851203306, "grad_norm": 1.6735071927392715, "learning_rate": 1.577892940259237e-05, "loss": 0.5663, "step": 9116 }, { "epoch": 0.3248125122468247, "grad_norm": 1.541469647661835, "learning_rate": 1.577798762186097e-05, "loss": 0.4288, "step": 9117 }, { "epoch": 0.32484813937331886, "grad_norm": 1.19046386083053, "learning_rate": 1.5777045764191625e-05, "loss": 0.4744, "step": 9118 }, { "epoch": 0.32488376649981293, "grad_norm": 1.191694740967615, "learning_rate": 1.577610382959688e-05, "loss": 0.5068, "step": 9119 }, { "epoch": 0.32491939362630706, "grad_norm": 1.4772719458078833, "learning_rate": 1.5775161818089273e-05, "loss": 0.6209, "step": 9120 }, { "epoch": 0.3249550207528012, "grad_norm": 1.5565117993809008, "learning_rate": 1.577421972968135e-05, "loss": 0.5865, "step": 9121 }, { "epoch": 0.32499064787929527, "grad_norm": 1.0935270023818615, "learning_rate": 1.577327756438566e-05, "loss": 0.4816, "step": 9122 }, { "epoch": 0.3250262750057894, "grad_norm": 1.8354543565448682, "learning_rate": 1.577233532221474e-05, "loss": 0.6043, "step": 9123 }, { "epoch": 0.32506190213228353, "grad_norm": 1.1232480543463916, "learning_rate": 1.5771393003181138e-05, "loss": 0.4986, "step": 9124 }, { "epoch": 0.3250975292587776, "grad_norm": 1.3246844919223149, "learning_rate": 1.5770450607297406e-05, "loss": 0.5388, "step": 9125 }, { "epoch": 0.32513315638527174, "grad_norm": 1.4027801130684001, "learning_rate": 1.5769508134576094e-05, "loss": 0.4853, "step": 9126 }, { "epoch": 0.32516878351176587, "grad_norm": 1.5106623992788957, "learning_rate": 1.5768565585029747e-05, "loss": 0.5679, "step": 9127 }, { "epoch": 0.32520441063825994, "grad_norm": 1.7099110404551219, "learning_rate": 1.5767622958670915e-05, "loss": 0.634, "step": 9128 }, { "epoch": 0.3252400377647541, "grad_norm": 1.082895203391438, "learning_rate": 1.5766680255512155e-05, "loss": 0.4529, "step": 9129 }, { "epoch": 0.3252756648912482, "grad_norm": 1.9928486760798987, "learning_rate": 1.5765737475566013e-05, "loss": 0.5875, "step": 9130 }, { "epoch": 0.3253112920177423, "grad_norm": 1.0998773517987839, "learning_rate": 1.5764794618845055e-05, "loss": 0.4692, "step": 9131 }, { "epoch": 0.3253469191442364, "grad_norm": 1.6595209085169318, "learning_rate": 1.5763851685361818e-05, "loss": 0.5663, "step": 9132 }, { "epoch": 0.32538254627073054, "grad_norm": 1.007094427141986, "learning_rate": 1.576290867512887e-05, "loss": 0.4815, "step": 9133 }, { "epoch": 0.3254181733972246, "grad_norm": 4.943867199621478, "learning_rate": 1.5761965588158767e-05, "loss": 0.5055, "step": 9134 }, { "epoch": 0.32545380052371875, "grad_norm": 1.1324302928685626, "learning_rate": 1.5761022424464063e-05, "loss": 0.4486, "step": 9135 }, { "epoch": 0.3254894276502129, "grad_norm": 1.037885236832211, "learning_rate": 1.5760079184057316e-05, "loss": 0.4757, "step": 9136 }, { "epoch": 0.325525054776707, "grad_norm": 1.166853746632024, "learning_rate": 1.5759135866951095e-05, "loss": 0.4785, "step": 9137 }, { "epoch": 0.3255606819032011, "grad_norm": 2.290769528998724, "learning_rate": 1.575819247315795e-05, "loss": 0.5401, "step": 9138 }, { "epoch": 0.3255963090296952, "grad_norm": 1.290270160930459, "learning_rate": 1.575724900269045e-05, "loss": 0.5398, "step": 9139 }, { "epoch": 0.32563193615618935, "grad_norm": 1.543187181490067, "learning_rate": 1.5756305455561154e-05, "loss": 0.6536, "step": 9140 }, { "epoch": 0.3256675632826834, "grad_norm": 1.4428608295416865, "learning_rate": 1.5755361831782626e-05, "loss": 0.572, "step": 9141 }, { "epoch": 0.32570319040917756, "grad_norm": 1.3914640091956711, "learning_rate": 1.5754418131367436e-05, "loss": 0.5636, "step": 9142 }, { "epoch": 0.3257388175356717, "grad_norm": 1.2719426254905517, "learning_rate": 1.5753474354328142e-05, "loss": 0.5132, "step": 9143 }, { "epoch": 0.32577444466216576, "grad_norm": 1.4532417644623847, "learning_rate": 1.575253050067732e-05, "loss": 0.5803, "step": 9144 }, { "epoch": 0.3258100717886599, "grad_norm": 1.560692622879562, "learning_rate": 1.575158657042753e-05, "loss": 0.5876, "step": 9145 }, { "epoch": 0.325845698915154, "grad_norm": 1.1174029661437714, "learning_rate": 1.575064256359135e-05, "loss": 0.4921, "step": 9146 }, { "epoch": 0.3258813260416481, "grad_norm": 1.0955355910800375, "learning_rate": 1.5749698480181335e-05, "loss": 0.4954, "step": 9147 }, { "epoch": 0.32591695316814223, "grad_norm": 2.170942608456124, "learning_rate": 1.5748754320210074e-05, "loss": 0.5283, "step": 9148 }, { "epoch": 0.32595258029463636, "grad_norm": 1.698333765248607, "learning_rate": 1.574781008369013e-05, "loss": 0.587, "step": 9149 }, { "epoch": 0.32598820742113044, "grad_norm": 1.0844914338816924, "learning_rate": 1.5746865770634073e-05, "loss": 0.4757, "step": 9150 }, { "epoch": 0.32602383454762457, "grad_norm": 0.9979469163096845, "learning_rate": 1.5745921381054487e-05, "loss": 0.4791, "step": 9151 }, { "epoch": 0.3260594616741187, "grad_norm": 1.711373406187406, "learning_rate": 1.574497691496394e-05, "loss": 0.5458, "step": 9152 }, { "epoch": 0.3260950888006128, "grad_norm": 2.377963450628399, "learning_rate": 1.5744032372375007e-05, "loss": 0.5333, "step": 9153 }, { "epoch": 0.3261307159271069, "grad_norm": 12.718471730346272, "learning_rate": 1.5743087753300274e-05, "loss": 0.6363, "step": 9154 }, { "epoch": 0.32616634305360104, "grad_norm": 1.6010082056810242, "learning_rate": 1.574214305775231e-05, "loss": 0.6086, "step": 9155 }, { "epoch": 0.3262019701800951, "grad_norm": 1.1882623668143315, "learning_rate": 1.5741198285743694e-05, "loss": 0.5168, "step": 9156 }, { "epoch": 0.32623759730658924, "grad_norm": 1.085432199372649, "learning_rate": 1.5740253437287014e-05, "loss": 0.4514, "step": 9157 }, { "epoch": 0.3262732244330834, "grad_norm": 1.07646378707076, "learning_rate": 1.5739308512394847e-05, "loss": 0.4416, "step": 9158 }, { "epoch": 0.32630885155957745, "grad_norm": 1.449478217182969, "learning_rate": 1.5738363511079776e-05, "loss": 0.595, "step": 9159 }, { "epoch": 0.3263444786860716, "grad_norm": 1.1074670358812633, "learning_rate": 1.573741843335438e-05, "loss": 0.5269, "step": 9160 }, { "epoch": 0.3263801058125657, "grad_norm": 1.3656995254937947, "learning_rate": 1.5736473279231255e-05, "loss": 0.4793, "step": 9161 }, { "epoch": 0.3264157329390598, "grad_norm": 1.6744104482173623, "learning_rate": 1.5735528048722974e-05, "loss": 0.6562, "step": 9162 }, { "epoch": 0.3264513600655539, "grad_norm": 1.7461568296259333, "learning_rate": 1.573458274184213e-05, "loss": 0.5875, "step": 9163 }, { "epoch": 0.32648698719204805, "grad_norm": 1.0266033065933051, "learning_rate": 1.5733637358601304e-05, "loss": 0.4776, "step": 9164 }, { "epoch": 0.3265226143185421, "grad_norm": 1.0180894118567816, "learning_rate": 1.5732691899013094e-05, "loss": 0.4518, "step": 9165 }, { "epoch": 0.32655824144503626, "grad_norm": 1.1360240425195518, "learning_rate": 1.573174636309008e-05, "loss": 0.5543, "step": 9166 }, { "epoch": 0.3265938685715304, "grad_norm": 1.4522166669933967, "learning_rate": 1.573080075084486e-05, "loss": 0.5997, "step": 9167 }, { "epoch": 0.32662949569802446, "grad_norm": 1.3598781535562277, "learning_rate": 1.5729855062290024e-05, "loss": 0.4948, "step": 9168 }, { "epoch": 0.3266651228245186, "grad_norm": 1.432396139164749, "learning_rate": 1.572890929743816e-05, "loss": 0.591, "step": 9169 }, { "epoch": 0.3267007499510127, "grad_norm": 1.5765282216172107, "learning_rate": 1.572796345630186e-05, "loss": 0.6195, "step": 9170 }, { "epoch": 0.3267363770775068, "grad_norm": 1.5831598919816632, "learning_rate": 1.5727017538893732e-05, "loss": 0.6335, "step": 9171 }, { "epoch": 0.32677200420400093, "grad_norm": 1.0209531328399717, "learning_rate": 1.572607154522636e-05, "loss": 0.484, "step": 9172 }, { "epoch": 0.32680763133049506, "grad_norm": 1.925584256221216, "learning_rate": 1.572512547531234e-05, "loss": 0.5956, "step": 9173 }, { "epoch": 0.32684325845698914, "grad_norm": 2.0284669754946205, "learning_rate": 1.5724179329164274e-05, "loss": 0.6193, "step": 9174 }, { "epoch": 0.32687888558348327, "grad_norm": 1.3755886990954547, "learning_rate": 1.572323310679476e-05, "loss": 0.5351, "step": 9175 }, { "epoch": 0.3269145127099774, "grad_norm": 1.0829674183828064, "learning_rate": 1.5722286808216397e-05, "loss": 0.475, "step": 9176 }, { "epoch": 0.3269501398364715, "grad_norm": 1.3623161472986671, "learning_rate": 1.5721340433441786e-05, "loss": 0.5985, "step": 9177 }, { "epoch": 0.3269857669629656, "grad_norm": 1.085295872064069, "learning_rate": 1.5720393982483528e-05, "loss": 0.5062, "step": 9178 }, { "epoch": 0.32702139408945974, "grad_norm": 1.6015092578503025, "learning_rate": 1.5719447455354225e-05, "loss": 0.5769, "step": 9179 }, { "epoch": 0.3270570212159538, "grad_norm": 1.1184858561496136, "learning_rate": 1.571850085206648e-05, "loss": 0.5238, "step": 9180 }, { "epoch": 0.32709264834244794, "grad_norm": 1.5782499630500684, "learning_rate": 1.57175541726329e-05, "loss": 0.5643, "step": 9181 }, { "epoch": 0.3271282754689421, "grad_norm": 1.2193266168543717, "learning_rate": 1.571660741706609e-05, "loss": 0.4853, "step": 9182 }, { "epoch": 0.32716390259543615, "grad_norm": 1.0442575311904059, "learning_rate": 1.571566058537866e-05, "loss": 0.4685, "step": 9183 }, { "epoch": 0.3271995297219303, "grad_norm": 1.4738735532731462, "learning_rate": 1.571471367758321e-05, "loss": 0.6123, "step": 9184 }, { "epoch": 0.3272351568484244, "grad_norm": 1.0369604747799877, "learning_rate": 1.5713766693692356e-05, "loss": 0.4883, "step": 9185 }, { "epoch": 0.3272707839749185, "grad_norm": 1.1550113134484286, "learning_rate": 1.5712819633718702e-05, "loss": 0.4623, "step": 9186 }, { "epoch": 0.3273064111014126, "grad_norm": 1.081857029916402, "learning_rate": 1.5711872497674863e-05, "loss": 0.4566, "step": 9187 }, { "epoch": 0.32734203822790675, "grad_norm": 1.3314169502085815, "learning_rate": 1.571092528557345e-05, "loss": 0.6052, "step": 9188 }, { "epoch": 0.3273776653544008, "grad_norm": 1.551627361688068, "learning_rate": 1.5709977997427077e-05, "loss": 0.566, "step": 9189 }, { "epoch": 0.32741329248089496, "grad_norm": 1.131987828439352, "learning_rate": 1.570903063324835e-05, "loss": 0.4877, "step": 9190 }, { "epoch": 0.3274489196073891, "grad_norm": 1.20382121618063, "learning_rate": 1.570808319304989e-05, "loss": 0.5339, "step": 9191 }, { "epoch": 0.32748454673388316, "grad_norm": 1.0930103760698122, "learning_rate": 1.570713567684432e-05, "loss": 0.4761, "step": 9192 }, { "epoch": 0.3275201738603773, "grad_norm": 1.210715640573758, "learning_rate": 1.5706188084644242e-05, "loss": 0.4998, "step": 9193 }, { "epoch": 0.3275558009868714, "grad_norm": 1.0570194007296383, "learning_rate": 1.5705240416462288e-05, "loss": 0.5175, "step": 9194 }, { "epoch": 0.3275914281133655, "grad_norm": 1.7247482081021974, "learning_rate": 1.5704292672311063e-05, "loss": 0.6402, "step": 9195 }, { "epoch": 0.32762705523985963, "grad_norm": 1.5972801865760156, "learning_rate": 1.57033448522032e-05, "loss": 0.5234, "step": 9196 }, { "epoch": 0.32766268236635376, "grad_norm": 1.2493847925719606, "learning_rate": 1.570239695615131e-05, "loss": 0.4906, "step": 9197 }, { "epoch": 0.32769830949284784, "grad_norm": 1.7304497037481905, "learning_rate": 1.5701448984168018e-05, "loss": 0.6212, "step": 9198 }, { "epoch": 0.32773393661934197, "grad_norm": 1.132499784317719, "learning_rate": 1.5700500936265954e-05, "loss": 0.4936, "step": 9199 }, { "epoch": 0.3277695637458361, "grad_norm": 1.193627725870981, "learning_rate": 1.5699552812457726e-05, "loss": 0.4947, "step": 9200 }, { "epoch": 0.3278051908723302, "grad_norm": 1.5460667492246367, "learning_rate": 1.5698604612755977e-05, "loss": 0.5127, "step": 9201 }, { "epoch": 0.3278408179988243, "grad_norm": 1.407761156733449, "learning_rate": 1.569765633717332e-05, "loss": 0.5876, "step": 9202 }, { "epoch": 0.32787644512531844, "grad_norm": 1.2090867518459814, "learning_rate": 1.569670798572239e-05, "loss": 0.4616, "step": 9203 }, { "epoch": 0.3279120722518125, "grad_norm": 1.1115354462720632, "learning_rate": 1.569575955841581e-05, "loss": 0.44, "step": 9204 }, { "epoch": 0.32794769937830665, "grad_norm": 1.1278749064325557, "learning_rate": 1.5694811055266207e-05, "loss": 0.4867, "step": 9205 }, { "epoch": 0.3279833265048008, "grad_norm": 1.1702787744304055, "learning_rate": 1.5693862476286218e-05, "loss": 0.4626, "step": 9206 }, { "epoch": 0.32801895363129485, "grad_norm": 1.824635890314544, "learning_rate": 1.569291382148847e-05, "loss": 0.5598, "step": 9207 }, { "epoch": 0.328054580757789, "grad_norm": 2.1028814127420596, "learning_rate": 1.5691965090885594e-05, "loss": 0.6608, "step": 9208 }, { "epoch": 0.3280902078842831, "grad_norm": 1.1651697524033893, "learning_rate": 1.5691016284490225e-05, "loss": 0.4646, "step": 9209 }, { "epoch": 0.3281258350107772, "grad_norm": 1.7346223564721726, "learning_rate": 1.5690067402314996e-05, "loss": 0.636, "step": 9210 }, { "epoch": 0.3281614621372713, "grad_norm": 1.4934745153490894, "learning_rate": 1.5689118444372544e-05, "loss": 0.6579, "step": 9211 }, { "epoch": 0.32819708926376545, "grad_norm": 1.7713708954093161, "learning_rate": 1.5688169410675503e-05, "loss": 0.6316, "step": 9212 }, { "epoch": 0.3282327163902595, "grad_norm": 1.6466352053631612, "learning_rate": 1.568722030123651e-05, "loss": 0.6274, "step": 9213 }, { "epoch": 0.32826834351675366, "grad_norm": 1.0318471623199863, "learning_rate": 1.5686271116068203e-05, "loss": 0.4613, "step": 9214 }, { "epoch": 0.3283039706432478, "grad_norm": 1.698776163089281, "learning_rate": 1.568532185518322e-05, "loss": 0.6198, "step": 9215 }, { "epoch": 0.32833959776974186, "grad_norm": 1.075057528027302, "learning_rate": 1.5684372518594203e-05, "loss": 0.4486, "step": 9216 }, { "epoch": 0.328375224896236, "grad_norm": 1.7301631738531555, "learning_rate": 1.5683423106313798e-05, "loss": 0.5778, "step": 9217 }, { "epoch": 0.3284108520227301, "grad_norm": 1.4502147004342014, "learning_rate": 1.5682473618354638e-05, "loss": 0.5065, "step": 9218 }, { "epoch": 0.3284464791492242, "grad_norm": 1.5710271066634884, "learning_rate": 1.568152405472937e-05, "loss": 0.6443, "step": 9219 }, { "epoch": 0.32848210627571833, "grad_norm": 0.9494060434501174, "learning_rate": 1.568057441545064e-05, "loss": 0.4512, "step": 9220 }, { "epoch": 0.32851773340221246, "grad_norm": 1.4128367412998504, "learning_rate": 1.5679624700531088e-05, "loss": 0.5715, "step": 9221 }, { "epoch": 0.32855336052870654, "grad_norm": 1.1300324164125355, "learning_rate": 1.5678674909983364e-05, "loss": 0.4781, "step": 9222 }, { "epoch": 0.32858898765520067, "grad_norm": 1.096885643098803, "learning_rate": 1.5677725043820117e-05, "loss": 0.4691, "step": 9223 }, { "epoch": 0.3286246147816948, "grad_norm": 1.0813417391368836, "learning_rate": 1.5676775102053993e-05, "loss": 0.4938, "step": 9224 }, { "epoch": 0.3286602419081889, "grad_norm": 1.4514021567604236, "learning_rate": 1.5675825084697635e-05, "loss": 0.4997, "step": 9225 }, { "epoch": 0.328695869034683, "grad_norm": 0.9759490434380823, "learning_rate": 1.5674874991763705e-05, "loss": 0.4526, "step": 9226 }, { "epoch": 0.32873149616117714, "grad_norm": 1.3110345093066649, "learning_rate": 1.5673924823264845e-05, "loss": 0.6147, "step": 9227 }, { "epoch": 0.3287671232876712, "grad_norm": 1.0729243101844166, "learning_rate": 1.5672974579213705e-05, "loss": 0.4965, "step": 9228 }, { "epoch": 0.32880275041416535, "grad_norm": 1.4599636281301422, "learning_rate": 1.5672024259622955e-05, "loss": 0.5108, "step": 9229 }, { "epoch": 0.3288383775406595, "grad_norm": 2.231944275232639, "learning_rate": 1.5671073864505227e-05, "loss": 0.5144, "step": 9230 }, { "epoch": 0.32887400466715355, "grad_norm": 2.4072645892137223, "learning_rate": 1.5670123393873194e-05, "loss": 0.5567, "step": 9231 }, { "epoch": 0.3289096317936477, "grad_norm": 1.6413315851489212, "learning_rate": 1.56691728477395e-05, "loss": 0.5846, "step": 9232 }, { "epoch": 0.3289452589201418, "grad_norm": 1.4849899292167497, "learning_rate": 1.5668222226116808e-05, "loss": 0.5477, "step": 9233 }, { "epoch": 0.3289808860466359, "grad_norm": 1.28033093756621, "learning_rate": 1.5667271529017777e-05, "loss": 0.4392, "step": 9234 }, { "epoch": 0.32901651317313, "grad_norm": 1.5625209796816666, "learning_rate": 1.5666320756455063e-05, "loss": 0.6173, "step": 9235 }, { "epoch": 0.32905214029962415, "grad_norm": 1.309535924923486, "learning_rate": 1.5665369908441328e-05, "loss": 0.4872, "step": 9236 }, { "epoch": 0.3290877674261182, "grad_norm": 1.2219752514562565, "learning_rate": 1.5664418984989235e-05, "loss": 0.5165, "step": 9237 }, { "epoch": 0.32912339455261236, "grad_norm": 1.9128363956452599, "learning_rate": 1.566346798611144e-05, "loss": 0.5495, "step": 9238 }, { "epoch": 0.3291590216791065, "grad_norm": 1.2414019034743249, "learning_rate": 1.5662516911820612e-05, "loss": 0.488, "step": 9239 }, { "epoch": 0.32919464880560056, "grad_norm": 1.6348907206579566, "learning_rate": 1.5661565762129414e-05, "loss": 0.6764, "step": 9240 }, { "epoch": 0.3292302759320947, "grad_norm": 2.2315243082958016, "learning_rate": 1.566061453705051e-05, "loss": 0.6382, "step": 9241 }, { "epoch": 0.3292659030585888, "grad_norm": 1.1357127386443435, "learning_rate": 1.5659663236596564e-05, "loss": 0.4671, "step": 9242 }, { "epoch": 0.3293015301850829, "grad_norm": 1.4676857699640888, "learning_rate": 1.565871186078025e-05, "loss": 0.6126, "step": 9243 }, { "epoch": 0.32933715731157703, "grad_norm": 1.5647804025855419, "learning_rate": 1.5657760409614232e-05, "loss": 0.5983, "step": 9244 }, { "epoch": 0.32937278443807116, "grad_norm": 1.1840631450920736, "learning_rate": 1.565680888311118e-05, "loss": 0.5112, "step": 9245 }, { "epoch": 0.32940841156456524, "grad_norm": 1.0567648177337339, "learning_rate": 1.5655857281283764e-05, "loss": 0.4759, "step": 9246 }, { "epoch": 0.32944403869105937, "grad_norm": 1.5901511869555656, "learning_rate": 1.5654905604144656e-05, "loss": 0.5416, "step": 9247 }, { "epoch": 0.3294796658175535, "grad_norm": 1.1780624067392713, "learning_rate": 1.5653953851706524e-05, "loss": 0.5313, "step": 9248 }, { "epoch": 0.3295152929440476, "grad_norm": 1.189530873922084, "learning_rate": 1.5653002023982047e-05, "loss": 0.4093, "step": 9249 }, { "epoch": 0.3295509200705417, "grad_norm": 1.3865433756955965, "learning_rate": 1.5652050120983896e-05, "loss": 0.5705, "step": 9250 }, { "epoch": 0.32958654719703584, "grad_norm": 1.1488241145655838, "learning_rate": 1.5651098142724744e-05, "loss": 0.5202, "step": 9251 }, { "epoch": 0.3296221743235299, "grad_norm": 1.999576519860106, "learning_rate": 1.5650146089217276e-05, "loss": 0.5051, "step": 9252 }, { "epoch": 0.32965780145002405, "grad_norm": 1.1239802042336438, "learning_rate": 1.564919396047416e-05, "loss": 0.5323, "step": 9253 }, { "epoch": 0.3296934285765182, "grad_norm": 1.0982944709599283, "learning_rate": 1.564824175650808e-05, "loss": 0.4944, "step": 9254 }, { "epoch": 0.32972905570301225, "grad_norm": 2.168574258982106, "learning_rate": 1.564728947733171e-05, "loss": 0.7148, "step": 9255 }, { "epoch": 0.3297646828295064, "grad_norm": 1.0997196799945863, "learning_rate": 1.5646337122957736e-05, "loss": 0.5161, "step": 9256 }, { "epoch": 0.3298003099560005, "grad_norm": 1.474818355217823, "learning_rate": 1.5645384693398837e-05, "loss": 0.5657, "step": 9257 }, { "epoch": 0.3298359370824946, "grad_norm": 1.5609379209525531, "learning_rate": 1.5644432188667695e-05, "loss": 0.4797, "step": 9258 }, { "epoch": 0.3298715642089887, "grad_norm": 1.7011896913998497, "learning_rate": 1.5643479608776992e-05, "loss": 0.6129, "step": 9259 }, { "epoch": 0.32990719133548285, "grad_norm": 1.3813463452055668, "learning_rate": 1.5642526953739413e-05, "loss": 0.5689, "step": 9260 }, { "epoch": 0.32994281846197693, "grad_norm": 1.4487915349947453, "learning_rate": 1.5641574223567643e-05, "loss": 0.7061, "step": 9261 }, { "epoch": 0.32997844558847106, "grad_norm": 1.1554693635099154, "learning_rate": 1.564062141827437e-05, "loss": 0.5067, "step": 9262 }, { "epoch": 0.3300140727149652, "grad_norm": 1.2096932907771747, "learning_rate": 1.563966853787228e-05, "loss": 0.48, "step": 9263 }, { "epoch": 0.33004969984145927, "grad_norm": 2.482107473422747, "learning_rate": 1.5638715582374065e-05, "loss": 0.5295, "step": 9264 }, { "epoch": 0.3300853269679534, "grad_norm": 1.1010881336216973, "learning_rate": 1.5637762551792404e-05, "loss": 0.4638, "step": 9265 }, { "epoch": 0.3301209540944475, "grad_norm": 1.555910281134409, "learning_rate": 1.563680944614e-05, "loss": 0.5753, "step": 9266 }, { "epoch": 0.3301565812209416, "grad_norm": 1.07100218010757, "learning_rate": 1.5635856265429538e-05, "loss": 0.4824, "step": 9267 }, { "epoch": 0.33019220834743573, "grad_norm": 1.6313583764816857, "learning_rate": 1.5634903009673706e-05, "loss": 0.5625, "step": 9268 }, { "epoch": 0.33022783547392986, "grad_norm": 1.1046078624431448, "learning_rate": 1.5633949678885208e-05, "loss": 0.4594, "step": 9269 }, { "epoch": 0.33026346260042394, "grad_norm": 1.1440165621577258, "learning_rate": 1.5632996273076727e-05, "loss": 0.4514, "step": 9270 }, { "epoch": 0.33029908972691807, "grad_norm": 1.1286085995836979, "learning_rate": 1.5632042792260965e-05, "loss": 0.5134, "step": 9271 }, { "epoch": 0.3303347168534122, "grad_norm": 1.0484888202957638, "learning_rate": 1.5631089236450616e-05, "loss": 0.4684, "step": 9272 }, { "epoch": 0.3303703439799063, "grad_norm": 1.5635577294322212, "learning_rate": 1.563013560565838e-05, "loss": 0.5913, "step": 9273 }, { "epoch": 0.3304059711064004, "grad_norm": 1.3689469023921108, "learning_rate": 1.5629181899896952e-05, "loss": 0.4755, "step": 9274 }, { "epoch": 0.33044159823289454, "grad_norm": 1.4897308036896217, "learning_rate": 1.562822811917903e-05, "loss": 0.628, "step": 9275 }, { "epoch": 0.3304772253593886, "grad_norm": 2.054817270705142, "learning_rate": 1.562727426351732e-05, "loss": 0.5858, "step": 9276 }, { "epoch": 0.33051285248588275, "grad_norm": 1.281267487983948, "learning_rate": 1.562632033292452e-05, "loss": 0.5174, "step": 9277 }, { "epoch": 0.3305484796123769, "grad_norm": 1.1625162391069535, "learning_rate": 1.562536632741333e-05, "loss": 0.4627, "step": 9278 }, { "epoch": 0.33058410673887095, "grad_norm": 1.6219105051506952, "learning_rate": 1.562441224699646e-05, "loss": 0.6874, "step": 9279 }, { "epoch": 0.3306197338653651, "grad_norm": 1.0061637900633924, "learning_rate": 1.5623458091686606e-05, "loss": 0.487, "step": 9280 }, { "epoch": 0.3306553609918592, "grad_norm": 1.2056594220295849, "learning_rate": 1.5622503861496475e-05, "loss": 0.5145, "step": 9281 }, { "epoch": 0.3306909881183533, "grad_norm": 1.6349203169572473, "learning_rate": 1.562154955643878e-05, "loss": 0.615, "step": 9282 }, { "epoch": 0.3307266152448474, "grad_norm": 2.142209710280919, "learning_rate": 1.562059517652622e-05, "loss": 0.6749, "step": 9283 }, { "epoch": 0.33076224237134155, "grad_norm": 1.891582045925313, "learning_rate": 1.5619640721771507e-05, "loss": 0.5463, "step": 9284 }, { "epoch": 0.33079786949783563, "grad_norm": 1.341627213372883, "learning_rate": 1.5618686192187353e-05, "loss": 0.6004, "step": 9285 }, { "epoch": 0.33083349662432976, "grad_norm": 1.1251522744865743, "learning_rate": 1.561773158778646e-05, "loss": 0.4801, "step": 9286 }, { "epoch": 0.3308691237508239, "grad_norm": 1.0647710115295215, "learning_rate": 1.5616776908581546e-05, "loss": 0.5028, "step": 9287 }, { "epoch": 0.33090475087731797, "grad_norm": 1.142757361430613, "learning_rate": 1.5615822154585326e-05, "loss": 0.5054, "step": 9288 }, { "epoch": 0.3309403780038121, "grad_norm": 0.9801109682452486, "learning_rate": 1.5614867325810503e-05, "loss": 0.458, "step": 9289 }, { "epoch": 0.33097600513030623, "grad_norm": 1.4717870359762364, "learning_rate": 1.56139124222698e-05, "loss": 0.6041, "step": 9290 }, { "epoch": 0.3310116322568003, "grad_norm": 1.1615363972337651, "learning_rate": 1.5612957443975928e-05, "loss": 0.4887, "step": 9291 }, { "epoch": 0.33104725938329443, "grad_norm": 1.437931519601504, "learning_rate": 1.5612002390941604e-05, "loss": 0.546, "step": 9292 }, { "epoch": 0.33108288650978857, "grad_norm": 1.107212089640862, "learning_rate": 1.561104726317955e-05, "loss": 0.4513, "step": 9293 }, { "epoch": 0.33111851363628264, "grad_norm": 1.5477215352760834, "learning_rate": 1.5610092060702477e-05, "loss": 0.5828, "step": 9294 }, { "epoch": 0.33115414076277677, "grad_norm": 1.873671243334022, "learning_rate": 1.5609136783523106e-05, "loss": 0.5172, "step": 9295 }, { "epoch": 0.3311897678892709, "grad_norm": 1.18255263824576, "learning_rate": 1.560818143165416e-05, "loss": 0.5211, "step": 9296 }, { "epoch": 0.331225395015765, "grad_norm": 1.553572033678612, "learning_rate": 1.560722600510835e-05, "loss": 0.6317, "step": 9297 }, { "epoch": 0.3312610221422591, "grad_norm": 1.3016138607649124, "learning_rate": 1.5606270503898414e-05, "loss": 0.4896, "step": 9298 }, { "epoch": 0.33129664926875324, "grad_norm": 1.0862576829617452, "learning_rate": 1.560531492803707e-05, "loss": 0.471, "step": 9299 }, { "epoch": 0.3313322763952473, "grad_norm": 1.6125321451308587, "learning_rate": 1.5604359277537037e-05, "loss": 0.6185, "step": 9300 }, { "epoch": 0.33136790352174145, "grad_norm": 1.468555586114983, "learning_rate": 1.560340355241104e-05, "loss": 0.6244, "step": 9301 }, { "epoch": 0.3314035306482356, "grad_norm": 1.098610507933264, "learning_rate": 1.5602447752671813e-05, "loss": 0.5039, "step": 9302 }, { "epoch": 0.33143915777472965, "grad_norm": 1.1292401849280205, "learning_rate": 1.5601491878332077e-05, "loss": 0.4775, "step": 9303 }, { "epoch": 0.3314747849012238, "grad_norm": 2.2399381956282527, "learning_rate": 1.5600535929404563e-05, "loss": 0.6126, "step": 9304 }, { "epoch": 0.3315104120277179, "grad_norm": 1.568390573271197, "learning_rate": 1.5599579905901996e-05, "loss": 0.5948, "step": 9305 }, { "epoch": 0.331546039154212, "grad_norm": 1.3398084689363376, "learning_rate": 1.5598623807837107e-05, "loss": 0.6033, "step": 9306 }, { "epoch": 0.3315816662807061, "grad_norm": 1.1112770485867993, "learning_rate": 1.559766763522263e-05, "loss": 0.5171, "step": 9307 }, { "epoch": 0.33161729340720025, "grad_norm": 1.4144926675202936, "learning_rate": 1.55967113880713e-05, "loss": 0.5372, "step": 9308 }, { "epoch": 0.33165292053369433, "grad_norm": 1.3802599824173831, "learning_rate": 1.559575506639584e-05, "loss": 0.6356, "step": 9309 }, { "epoch": 0.33168854766018846, "grad_norm": 1.1983060503522678, "learning_rate": 1.559479867020899e-05, "loss": 0.4687, "step": 9310 }, { "epoch": 0.3317241747866826, "grad_norm": 1.3400552453532302, "learning_rate": 1.559384219952349e-05, "loss": 0.5805, "step": 9311 }, { "epoch": 0.33175980191317667, "grad_norm": 2.011548976932469, "learning_rate": 1.559288565435207e-05, "loss": 0.5626, "step": 9312 }, { "epoch": 0.3317954290396708, "grad_norm": 2.234645568952513, "learning_rate": 1.5591929034707468e-05, "loss": 0.5695, "step": 9313 }, { "epoch": 0.33183105616616493, "grad_norm": 1.0519485091832828, "learning_rate": 1.5590972340602425e-05, "loss": 0.516, "step": 9314 }, { "epoch": 0.331866683292659, "grad_norm": 1.5962796620439783, "learning_rate": 1.5590015572049673e-05, "loss": 0.6674, "step": 9315 }, { "epoch": 0.33190231041915313, "grad_norm": 1.2246842451425466, "learning_rate": 1.5589058729061957e-05, "loss": 0.5248, "step": 9316 }, { "epoch": 0.33193793754564727, "grad_norm": 1.0794083587585908, "learning_rate": 1.558810181165202e-05, "loss": 0.4851, "step": 9317 }, { "epoch": 0.33197356467214134, "grad_norm": 1.155866106644051, "learning_rate": 1.55871448198326e-05, "loss": 0.465, "step": 9318 }, { "epoch": 0.3320091917986355, "grad_norm": 1.181547968813121, "learning_rate": 1.5586187753616443e-05, "loss": 0.5338, "step": 9319 }, { "epoch": 0.3320448189251296, "grad_norm": 1.3619388666111722, "learning_rate": 1.558523061301629e-05, "loss": 0.5754, "step": 9320 }, { "epoch": 0.3320804460516237, "grad_norm": 1.0427144843360032, "learning_rate": 1.5584273398044887e-05, "loss": 0.4955, "step": 9321 }, { "epoch": 0.3321160731781178, "grad_norm": 1.8567684329992675, "learning_rate": 1.558331610871498e-05, "loss": 0.5909, "step": 9322 }, { "epoch": 0.33215170030461194, "grad_norm": 1.074424832277912, "learning_rate": 1.5582358745039318e-05, "loss": 0.4849, "step": 9323 }, { "epoch": 0.332187327431106, "grad_norm": 1.1571149910775262, "learning_rate": 1.558140130703065e-05, "loss": 0.5017, "step": 9324 }, { "epoch": 0.33222295455760015, "grad_norm": 1.1510351827442338, "learning_rate": 1.5580443794701715e-05, "loss": 0.5162, "step": 9325 }, { "epoch": 0.3322585816840943, "grad_norm": 1.882198144966984, "learning_rate": 1.557948620806528e-05, "loss": 0.5485, "step": 9326 }, { "epoch": 0.3322942088105884, "grad_norm": 1.0800048196179335, "learning_rate": 1.5578528547134078e-05, "loss": 0.4989, "step": 9327 }, { "epoch": 0.3323298359370825, "grad_norm": 1.5503679875474619, "learning_rate": 1.5577570811920873e-05, "loss": 0.5639, "step": 9328 }, { "epoch": 0.3323654630635766, "grad_norm": 1.673903462034291, "learning_rate": 1.5576613002438413e-05, "loss": 0.5775, "step": 9329 }, { "epoch": 0.33240109019007075, "grad_norm": 1.1001094820838186, "learning_rate": 1.5575655118699454e-05, "loss": 0.5082, "step": 9330 }, { "epoch": 0.3324367173165648, "grad_norm": 1.1852243051492006, "learning_rate": 1.557469716071675e-05, "loss": 0.4575, "step": 9331 }, { "epoch": 0.33247234444305895, "grad_norm": 1.365474348329071, "learning_rate": 1.5573739128503053e-05, "loss": 0.6406, "step": 9332 }, { "epoch": 0.3325079715695531, "grad_norm": 1.0434027146061267, "learning_rate": 1.557278102207113e-05, "loss": 0.4548, "step": 9333 }, { "epoch": 0.33254359869604716, "grad_norm": 1.7275487369098375, "learning_rate": 1.557182284143373e-05, "loss": 0.5455, "step": 9334 }, { "epoch": 0.3325792258225413, "grad_norm": 1.8813397359237731, "learning_rate": 1.5570864586603613e-05, "loss": 0.6603, "step": 9335 }, { "epoch": 0.3326148529490354, "grad_norm": 1.7276819700485548, "learning_rate": 1.5569906257593542e-05, "loss": 0.6494, "step": 9336 }, { "epoch": 0.3326504800755295, "grad_norm": 1.8760680540480765, "learning_rate": 1.5568947854416277e-05, "loss": 0.5757, "step": 9337 }, { "epoch": 0.33268610720202363, "grad_norm": 1.401220590557564, "learning_rate": 1.5567989377084574e-05, "loss": 0.5567, "step": 9338 }, { "epoch": 0.33272173432851776, "grad_norm": 1.3500066815344403, "learning_rate": 1.5567030825611203e-05, "loss": 0.5719, "step": 9339 }, { "epoch": 0.33275736145501184, "grad_norm": 1.1759759896199464, "learning_rate": 1.5566072200008928e-05, "loss": 0.5389, "step": 9340 }, { "epoch": 0.33279298858150597, "grad_norm": 1.8480035013853855, "learning_rate": 1.5565113500290513e-05, "loss": 0.5533, "step": 9341 }, { "epoch": 0.3328286157080001, "grad_norm": 1.301839306959893, "learning_rate": 1.5564154726468717e-05, "loss": 0.5963, "step": 9342 }, { "epoch": 0.3328642428344942, "grad_norm": 1.1252796337324023, "learning_rate": 1.5563195878556312e-05, "loss": 0.5267, "step": 9343 }, { "epoch": 0.3328998699609883, "grad_norm": 1.6874882158087663, "learning_rate": 1.5562236956566064e-05, "loss": 0.5749, "step": 9344 }, { "epoch": 0.33293549708748243, "grad_norm": 1.151853729031893, "learning_rate": 1.5561277960510747e-05, "loss": 0.555, "step": 9345 }, { "epoch": 0.3329711242139765, "grad_norm": 1.2202112896139024, "learning_rate": 1.5560318890403125e-05, "loss": 0.54, "step": 9346 }, { "epoch": 0.33300675134047064, "grad_norm": 1.0082332339153162, "learning_rate": 1.5559359746255972e-05, "loss": 0.4439, "step": 9347 }, { "epoch": 0.3330423784669648, "grad_norm": 1.251289582535592, "learning_rate": 1.5558400528082057e-05, "loss": 0.5713, "step": 9348 }, { "epoch": 0.33307800559345885, "grad_norm": 1.6882297736504677, "learning_rate": 1.5557441235894156e-05, "loss": 0.5799, "step": 9349 }, { "epoch": 0.333113632719953, "grad_norm": 1.3660367309492734, "learning_rate": 1.5556481869705038e-05, "loss": 0.5403, "step": 9350 }, { "epoch": 0.3331492598464471, "grad_norm": 1.4637777310525717, "learning_rate": 1.5555522429527484e-05, "loss": 0.5421, "step": 9351 }, { "epoch": 0.3331848869729412, "grad_norm": 1.3033385489411038, "learning_rate": 1.555456291537426e-05, "loss": 0.5259, "step": 9352 }, { "epoch": 0.3332205140994353, "grad_norm": 2.0586912429581146, "learning_rate": 1.5553603327258153e-05, "loss": 0.5836, "step": 9353 }, { "epoch": 0.33325614122592945, "grad_norm": 1.1170148119018102, "learning_rate": 1.5552643665191933e-05, "loss": 0.4967, "step": 9354 }, { "epoch": 0.3332917683524235, "grad_norm": 1.1161516907961788, "learning_rate": 1.5551683929188383e-05, "loss": 0.4798, "step": 9355 }, { "epoch": 0.33332739547891765, "grad_norm": 1.5709355361597717, "learning_rate": 1.555072411926028e-05, "loss": 0.5354, "step": 9356 }, { "epoch": 0.3333630226054118, "grad_norm": 1.3930405163872257, "learning_rate": 1.5549764235420407e-05, "loss": 0.6161, "step": 9357 }, { "epoch": 0.33339864973190586, "grad_norm": 1.69473238063321, "learning_rate": 1.554880427768154e-05, "loss": 0.6047, "step": 9358 }, { "epoch": 0.3334342768584, "grad_norm": 1.6148750633506217, "learning_rate": 1.5547844246056472e-05, "loss": 0.5736, "step": 9359 }, { "epoch": 0.3334699039848941, "grad_norm": 1.5321174866469982, "learning_rate": 1.5546884140557977e-05, "loss": 0.5557, "step": 9360 }, { "epoch": 0.3335055311113882, "grad_norm": 1.066872234542027, "learning_rate": 1.5545923961198842e-05, "loss": 0.4835, "step": 9361 }, { "epoch": 0.33354115823788233, "grad_norm": 1.4982195504859384, "learning_rate": 1.5544963707991855e-05, "loss": 0.6286, "step": 9362 }, { "epoch": 0.33357678536437646, "grad_norm": 1.3744144578182032, "learning_rate": 1.55440033809498e-05, "loss": 0.6178, "step": 9363 }, { "epoch": 0.33361241249087054, "grad_norm": 1.6758292125587224, "learning_rate": 1.5543042980085467e-05, "loss": 0.5846, "step": 9364 }, { "epoch": 0.33364803961736467, "grad_norm": 1.144706150756794, "learning_rate": 1.5542082505411636e-05, "loss": 0.4566, "step": 9365 }, { "epoch": 0.3336836667438588, "grad_norm": 1.6081869969542408, "learning_rate": 1.5541121956941107e-05, "loss": 0.577, "step": 9366 }, { "epoch": 0.3337192938703529, "grad_norm": 1.0862028479834074, "learning_rate": 1.5540161334686665e-05, "loss": 0.4842, "step": 9367 }, { "epoch": 0.333754920996847, "grad_norm": 1.6487715799411726, "learning_rate": 1.5539200638661106e-05, "loss": 0.6307, "step": 9368 }, { "epoch": 0.33379054812334114, "grad_norm": 1.8393692011627802, "learning_rate": 1.5538239868877215e-05, "loss": 0.5493, "step": 9369 }, { "epoch": 0.3338261752498352, "grad_norm": 1.197880054411136, "learning_rate": 1.553727902534779e-05, "loss": 0.4917, "step": 9370 }, { "epoch": 0.33386180237632934, "grad_norm": 2.301690192215823, "learning_rate": 1.5536318108085624e-05, "loss": 0.6127, "step": 9371 }, { "epoch": 0.3338974295028235, "grad_norm": 1.1523451190906289, "learning_rate": 1.5535357117103514e-05, "loss": 0.4966, "step": 9372 }, { "epoch": 0.33393305662931755, "grad_norm": 1.3500795864272765, "learning_rate": 1.5534396052414257e-05, "loss": 0.6121, "step": 9373 }, { "epoch": 0.3339686837558117, "grad_norm": 1.5620418171224133, "learning_rate": 1.5533434914030644e-05, "loss": 0.6092, "step": 9374 }, { "epoch": 0.3340043108823058, "grad_norm": 1.3663818563590453, "learning_rate": 1.5532473701965477e-05, "loss": 0.5766, "step": 9375 }, { "epoch": 0.3340399380087999, "grad_norm": 1.1851450634384784, "learning_rate": 1.553151241623156e-05, "loss": 0.4867, "step": 9376 }, { "epoch": 0.334075565135294, "grad_norm": 1.3565781520648001, "learning_rate": 1.5530551056841687e-05, "loss": 0.6045, "step": 9377 }, { "epoch": 0.33411119226178815, "grad_norm": 1.0449676872893874, "learning_rate": 1.5529589623808657e-05, "loss": 0.5258, "step": 9378 }, { "epoch": 0.3341468193882822, "grad_norm": 1.4951007632499977, "learning_rate": 1.5528628117145282e-05, "loss": 0.6482, "step": 9379 }, { "epoch": 0.33418244651477635, "grad_norm": 1.0932506834322382, "learning_rate": 1.5527666536864357e-05, "loss": 0.5236, "step": 9380 }, { "epoch": 0.3342180736412705, "grad_norm": 1.690584180892532, "learning_rate": 1.552670488297869e-05, "loss": 0.5283, "step": 9381 }, { "epoch": 0.33425370076776456, "grad_norm": 1.0312532607821845, "learning_rate": 1.5525743155501084e-05, "loss": 0.4973, "step": 9382 }, { "epoch": 0.3342893278942587, "grad_norm": 1.2822710702697846, "learning_rate": 1.5524781354444344e-05, "loss": 0.5752, "step": 9383 }, { "epoch": 0.3343249550207528, "grad_norm": 1.1023663435401365, "learning_rate": 1.552381947982128e-05, "loss": 0.51, "step": 9384 }, { "epoch": 0.3343605821472469, "grad_norm": 1.9233037563185305, "learning_rate": 1.5522857531644697e-05, "loss": 0.5568, "step": 9385 }, { "epoch": 0.33439620927374103, "grad_norm": 1.059216356236132, "learning_rate": 1.5521895509927408e-05, "loss": 0.4666, "step": 9386 }, { "epoch": 0.33443183640023516, "grad_norm": 1.8554802062521147, "learning_rate": 1.552093341468222e-05, "loss": 0.5457, "step": 9387 }, { "epoch": 0.33446746352672924, "grad_norm": 1.7037919180365666, "learning_rate": 1.5519971245921946e-05, "loss": 0.6042, "step": 9388 }, { "epoch": 0.33450309065322337, "grad_norm": 1.0855851683342554, "learning_rate": 1.5519009003659392e-05, "loss": 0.531, "step": 9389 }, { "epoch": 0.3345387177797175, "grad_norm": 1.8779291610559474, "learning_rate": 1.5518046687907378e-05, "loss": 0.5423, "step": 9390 }, { "epoch": 0.3345743449062116, "grad_norm": 2.024235365102621, "learning_rate": 1.551708429867872e-05, "loss": 0.5917, "step": 9391 }, { "epoch": 0.3346099720327057, "grad_norm": 1.4224850971871688, "learning_rate": 1.551612183598622e-05, "loss": 0.5734, "step": 9392 }, { "epoch": 0.33464559915919984, "grad_norm": 1.2393690552963519, "learning_rate": 1.551515929984271e-05, "loss": 0.5022, "step": 9393 }, { "epoch": 0.3346812262856939, "grad_norm": 1.0914334769111962, "learning_rate": 1.5514196690260993e-05, "loss": 0.5122, "step": 9394 }, { "epoch": 0.33471685341218804, "grad_norm": 1.1240509173797553, "learning_rate": 1.5513234007253894e-05, "loss": 0.5166, "step": 9395 }, { "epoch": 0.3347524805386822, "grad_norm": 1.1614622852741914, "learning_rate": 1.5512271250834233e-05, "loss": 0.4572, "step": 9396 }, { "epoch": 0.33478810766517625, "grad_norm": 1.5454277962548706, "learning_rate": 1.5511308421014826e-05, "loss": 0.68, "step": 9397 }, { "epoch": 0.3348237347916704, "grad_norm": 1.034006524298408, "learning_rate": 1.55103455178085e-05, "loss": 0.471, "step": 9398 }, { "epoch": 0.3348593619181645, "grad_norm": 1.507911917949664, "learning_rate": 1.5509382541228065e-05, "loss": 0.5938, "step": 9399 }, { "epoch": 0.3348949890446586, "grad_norm": 1.542486999836036, "learning_rate": 1.5508419491286354e-05, "loss": 0.6227, "step": 9400 }, { "epoch": 0.3349306161711527, "grad_norm": 1.6087222456227706, "learning_rate": 1.5507456367996184e-05, "loss": 0.611, "step": 9401 }, { "epoch": 0.33496624329764685, "grad_norm": 1.0049685162247723, "learning_rate": 1.550649317137039e-05, "loss": 0.4925, "step": 9402 }, { "epoch": 0.3350018704241409, "grad_norm": 1.6792441476477098, "learning_rate": 1.5505529901421782e-05, "loss": 0.6411, "step": 9403 }, { "epoch": 0.33503749755063505, "grad_norm": 1.7367905104981598, "learning_rate": 1.5504566558163204e-05, "loss": 0.5963, "step": 9404 }, { "epoch": 0.3350731246771292, "grad_norm": 2.1156910952818264, "learning_rate": 1.550360314160747e-05, "loss": 0.5563, "step": 9405 }, { "epoch": 0.33510875180362326, "grad_norm": 1.287031414413486, "learning_rate": 1.5502639651767414e-05, "loss": 0.4552, "step": 9406 }, { "epoch": 0.3351443789301174, "grad_norm": 3.0644857875972673, "learning_rate": 1.5501676088655864e-05, "loss": 0.5625, "step": 9407 }, { "epoch": 0.3351800060566115, "grad_norm": 1.285205020866677, "learning_rate": 1.550071245228565e-05, "loss": 0.5476, "step": 9408 }, { "epoch": 0.3352156331831056, "grad_norm": 1.8166409027128956, "learning_rate": 1.549974874266961e-05, "loss": 0.6157, "step": 9409 }, { "epoch": 0.33525126030959973, "grad_norm": 1.1226610554664678, "learning_rate": 1.549878495982057e-05, "loss": 0.5166, "step": 9410 }, { "epoch": 0.33528688743609386, "grad_norm": 1.103942492348284, "learning_rate": 1.549782110375136e-05, "loss": 0.4761, "step": 9411 }, { "epoch": 0.33532251456258794, "grad_norm": 1.442610654699587, "learning_rate": 1.5496857174474824e-05, "loss": 0.5525, "step": 9412 }, { "epoch": 0.33535814168908207, "grad_norm": 1.086578616568378, "learning_rate": 1.549589317200379e-05, "loss": 0.4566, "step": 9413 }, { "epoch": 0.3353937688155762, "grad_norm": 1.4997905887683698, "learning_rate": 1.5494929096351098e-05, "loss": 0.6433, "step": 9414 }, { "epoch": 0.3354293959420703, "grad_norm": 1.0836463487641346, "learning_rate": 1.5493964947529584e-05, "loss": 0.5193, "step": 9415 }, { "epoch": 0.3354650230685644, "grad_norm": 1.2026814124773835, "learning_rate": 1.5493000725552087e-05, "loss": 0.4749, "step": 9416 }, { "epoch": 0.33550065019505854, "grad_norm": 1.1242515520128618, "learning_rate": 1.5492036430431446e-05, "loss": 0.4864, "step": 9417 }, { "epoch": 0.3355362773215526, "grad_norm": 1.4669299034780654, "learning_rate": 1.5491072062180504e-05, "loss": 0.6107, "step": 9418 }, { "epoch": 0.33557190444804674, "grad_norm": 1.1233108746849636, "learning_rate": 1.5490107620812096e-05, "loss": 0.4665, "step": 9419 }, { "epoch": 0.3356075315745409, "grad_norm": 1.2760955919019685, "learning_rate": 1.548914310633907e-05, "loss": 0.5095, "step": 9420 }, { "epoch": 0.33564315870103495, "grad_norm": 1.6131853172719564, "learning_rate": 1.5488178518774265e-05, "loss": 0.634, "step": 9421 }, { "epoch": 0.3356787858275291, "grad_norm": 1.4054019767446155, "learning_rate": 1.5487213858130528e-05, "loss": 0.5753, "step": 9422 }, { "epoch": 0.3357144129540232, "grad_norm": 1.5571197641245254, "learning_rate": 1.5486249124420702e-05, "loss": 0.6516, "step": 9423 }, { "epoch": 0.3357500400805173, "grad_norm": 1.7439278491141725, "learning_rate": 1.5485284317657635e-05, "loss": 0.5775, "step": 9424 }, { "epoch": 0.3357856672070114, "grad_norm": 2.274168573616694, "learning_rate": 1.548431943785417e-05, "loss": 0.5555, "step": 9425 }, { "epoch": 0.33582129433350555, "grad_norm": 1.376646946703929, "learning_rate": 1.5483354485023164e-05, "loss": 0.5372, "step": 9426 }, { "epoch": 0.3358569214599996, "grad_norm": 1.4570037836920058, "learning_rate": 1.5482389459177454e-05, "loss": 0.5844, "step": 9427 }, { "epoch": 0.33589254858649376, "grad_norm": 2.874932709004962, "learning_rate": 1.5481424360329906e-05, "loss": 0.5291, "step": 9428 }, { "epoch": 0.3359281757129879, "grad_norm": 1.840180724683306, "learning_rate": 1.5480459188493352e-05, "loss": 0.5165, "step": 9429 }, { "epoch": 0.33596380283948196, "grad_norm": 1.0920013923016305, "learning_rate": 1.5479493943680657e-05, "loss": 0.4656, "step": 9430 }, { "epoch": 0.3359994299659761, "grad_norm": 1.0610298313544957, "learning_rate": 1.547852862590467e-05, "loss": 0.503, "step": 9431 }, { "epoch": 0.3360350570924702, "grad_norm": 1.1272244049442057, "learning_rate": 1.547756323517825e-05, "loss": 0.4524, "step": 9432 }, { "epoch": 0.3360706842189643, "grad_norm": 1.6797011865678773, "learning_rate": 1.5476597771514243e-05, "loss": 0.536, "step": 9433 }, { "epoch": 0.33610631134545843, "grad_norm": 1.0792439402122933, "learning_rate": 1.5475632234925505e-05, "loss": 0.4315, "step": 9434 }, { "epoch": 0.33614193847195256, "grad_norm": 1.518474816485299, "learning_rate": 1.54746666254249e-05, "loss": 0.5459, "step": 9435 }, { "epoch": 0.33617756559844664, "grad_norm": 1.0266019875743673, "learning_rate": 1.547370094302528e-05, "loss": 0.5022, "step": 9436 }, { "epoch": 0.33621319272494077, "grad_norm": 1.2877284140509329, "learning_rate": 1.5472735187739507e-05, "loss": 0.4806, "step": 9437 }, { "epoch": 0.3362488198514349, "grad_norm": 1.9828551615986914, "learning_rate": 1.547176935958044e-05, "loss": 0.5853, "step": 9438 }, { "epoch": 0.336284446977929, "grad_norm": 1.9400024135345528, "learning_rate": 1.5470803458560943e-05, "loss": 0.5401, "step": 9439 }, { "epoch": 0.3363200741044231, "grad_norm": 1.203633185848015, "learning_rate": 1.546983748469387e-05, "loss": 0.4639, "step": 9440 }, { "epoch": 0.33635570123091724, "grad_norm": 1.829399097725552, "learning_rate": 1.546887143799209e-05, "loss": 0.6361, "step": 9441 }, { "epoch": 0.3363913283574113, "grad_norm": 1.0777874321589207, "learning_rate": 1.5467905318468462e-05, "loss": 0.4286, "step": 9442 }, { "epoch": 0.33642695548390544, "grad_norm": 2.0287210691648876, "learning_rate": 1.5466939126135856e-05, "loss": 0.6299, "step": 9443 }, { "epoch": 0.3364625826103996, "grad_norm": 1.6737566413610963, "learning_rate": 1.546597286100713e-05, "loss": 0.6508, "step": 9444 }, { "epoch": 0.33649820973689365, "grad_norm": 1.1416245319193308, "learning_rate": 1.546500652309516e-05, "loss": 0.4473, "step": 9445 }, { "epoch": 0.3365338368633878, "grad_norm": 1.390916762759795, "learning_rate": 1.5464040112412804e-05, "loss": 0.5818, "step": 9446 }, { "epoch": 0.3365694639898819, "grad_norm": 1.1562928636241028, "learning_rate": 1.5463073628972935e-05, "loss": 0.4889, "step": 9447 }, { "epoch": 0.336605091116376, "grad_norm": 3.57595217210084, "learning_rate": 1.5462107072788422e-05, "loss": 0.5876, "step": 9448 }, { "epoch": 0.3366407182428701, "grad_norm": 1.1949002021847586, "learning_rate": 1.546114044387214e-05, "loss": 0.4819, "step": 9449 }, { "epoch": 0.33667634536936425, "grad_norm": 1.9131862001771425, "learning_rate": 1.546017374223695e-05, "loss": 0.565, "step": 9450 }, { "epoch": 0.3367119724958583, "grad_norm": 1.4133682696672527, "learning_rate": 1.5459206967895735e-05, "loss": 0.6435, "step": 9451 }, { "epoch": 0.33674759962235246, "grad_norm": 1.0984513460661214, "learning_rate": 1.5458240120861363e-05, "loss": 0.541, "step": 9452 }, { "epoch": 0.3367832267488466, "grad_norm": 1.925143712365375, "learning_rate": 1.5457273201146708e-05, "loss": 0.6459, "step": 9453 }, { "epoch": 0.33681885387534066, "grad_norm": 2.219778992817623, "learning_rate": 1.5456306208764646e-05, "loss": 0.5804, "step": 9454 }, { "epoch": 0.3368544810018348, "grad_norm": 1.1939716536751477, "learning_rate": 1.545533914372805e-05, "loss": 0.5263, "step": 9455 }, { "epoch": 0.3368901081283289, "grad_norm": 1.4361366725206395, "learning_rate": 1.5454372006049803e-05, "loss": 0.6056, "step": 9456 }, { "epoch": 0.336925735254823, "grad_norm": 1.1450475288427766, "learning_rate": 1.5453404795742783e-05, "loss": 0.4827, "step": 9457 }, { "epoch": 0.33696136238131713, "grad_norm": 1.593324700866972, "learning_rate": 1.545243751281986e-05, "loss": 0.609, "step": 9458 }, { "epoch": 0.33699698950781126, "grad_norm": 1.4903722587641777, "learning_rate": 1.545147015729392e-05, "loss": 0.5928, "step": 9459 }, { "epoch": 0.33703261663430534, "grad_norm": 1.0757698465235537, "learning_rate": 1.5450502729177853e-05, "loss": 0.4708, "step": 9460 }, { "epoch": 0.33706824376079947, "grad_norm": 1.112430077807708, "learning_rate": 1.5449535228484526e-05, "loss": 0.4905, "step": 9461 }, { "epoch": 0.3371038708872936, "grad_norm": 1.1682316600859959, "learning_rate": 1.544856765522683e-05, "loss": 0.4836, "step": 9462 }, { "epoch": 0.3371394980137877, "grad_norm": 1.1364554399804492, "learning_rate": 1.544760000941765e-05, "loss": 0.4764, "step": 9463 }, { "epoch": 0.3371751251402818, "grad_norm": 1.0731252442068597, "learning_rate": 1.544663229106986e-05, "loss": 0.4398, "step": 9464 }, { "epoch": 0.33721075226677594, "grad_norm": 1.0948439526481826, "learning_rate": 1.5445664500196366e-05, "loss": 0.4342, "step": 9465 }, { "epoch": 0.33724637939327, "grad_norm": 1.0748956118322233, "learning_rate": 1.5444696636810036e-05, "loss": 0.4498, "step": 9466 }, { "epoch": 0.33728200651976414, "grad_norm": 1.7175620861211385, "learning_rate": 1.5443728700923764e-05, "loss": 0.5483, "step": 9467 }, { "epoch": 0.3373176336462583, "grad_norm": 1.4144648991522457, "learning_rate": 1.5442760692550443e-05, "loss": 0.5406, "step": 9468 }, { "epoch": 0.33735326077275235, "grad_norm": 1.077774075708504, "learning_rate": 1.5441792611702956e-05, "loss": 0.455, "step": 9469 }, { "epoch": 0.3373888878992465, "grad_norm": 1.093646974265643, "learning_rate": 1.5440824458394196e-05, "loss": 0.47, "step": 9470 }, { "epoch": 0.3374245150257406, "grad_norm": 1.1881901106333976, "learning_rate": 1.5439856232637058e-05, "loss": 0.4464, "step": 9471 }, { "epoch": 0.3374601421522347, "grad_norm": 1.105828132247341, "learning_rate": 1.543888793444443e-05, "loss": 0.4845, "step": 9472 }, { "epoch": 0.3374957692787288, "grad_norm": 2.546375171145984, "learning_rate": 1.543791956382921e-05, "loss": 0.5735, "step": 9473 }, { "epoch": 0.33753139640522295, "grad_norm": 1.0443257993869397, "learning_rate": 1.543695112080429e-05, "loss": 0.5168, "step": 9474 }, { "epoch": 0.337567023531717, "grad_norm": 1.6046368215387998, "learning_rate": 1.5435982605382565e-05, "loss": 0.5801, "step": 9475 }, { "epoch": 0.33760265065821116, "grad_norm": 1.1276344700211214, "learning_rate": 1.5435014017576932e-05, "loss": 0.5646, "step": 9476 }, { "epoch": 0.3376382777847053, "grad_norm": 1.062907503525654, "learning_rate": 1.543404535740029e-05, "loss": 0.4593, "step": 9477 }, { "epoch": 0.33767390491119936, "grad_norm": 1.1892768746773366, "learning_rate": 1.5433076624865533e-05, "loss": 0.5024, "step": 9478 }, { "epoch": 0.3377095320376935, "grad_norm": 1.24672437849315, "learning_rate": 1.5432107819985566e-05, "loss": 0.5182, "step": 9479 }, { "epoch": 0.3377451591641876, "grad_norm": 2.414734064613918, "learning_rate": 1.5431138942773284e-05, "loss": 0.6223, "step": 9480 }, { "epoch": 0.3377807862906817, "grad_norm": 1.2543323593647688, "learning_rate": 1.543016999324159e-05, "loss": 0.5019, "step": 9481 }, { "epoch": 0.33781641341717583, "grad_norm": 1.7118101453839951, "learning_rate": 1.542920097140339e-05, "loss": 0.6142, "step": 9482 }, { "epoch": 0.33785204054366996, "grad_norm": 2.0644125600174794, "learning_rate": 1.5428231877271584e-05, "loss": 0.6784, "step": 9483 }, { "epoch": 0.33788766767016404, "grad_norm": 1.137408302360346, "learning_rate": 1.5427262710859072e-05, "loss": 0.5169, "step": 9484 }, { "epoch": 0.33792329479665817, "grad_norm": 1.3036660403200486, "learning_rate": 1.542629347217877e-05, "loss": 0.564, "step": 9485 }, { "epoch": 0.3379589219231523, "grad_norm": 1.7068056738466864, "learning_rate": 1.5425324161243572e-05, "loss": 0.7092, "step": 9486 }, { "epoch": 0.3379945490496464, "grad_norm": 1.102913543186473, "learning_rate": 1.5424354778066395e-05, "loss": 0.4711, "step": 9487 }, { "epoch": 0.3380301761761405, "grad_norm": 1.1392321962752496, "learning_rate": 1.542338532266014e-05, "loss": 0.469, "step": 9488 }, { "epoch": 0.33806580330263464, "grad_norm": 1.164212096724989, "learning_rate": 1.5422415795037722e-05, "loss": 0.4794, "step": 9489 }, { "epoch": 0.3381014304291287, "grad_norm": 1.5052175481585315, "learning_rate": 1.5421446195212045e-05, "loss": 0.6407, "step": 9490 }, { "epoch": 0.33813705755562284, "grad_norm": 1.553780797204946, "learning_rate": 1.5420476523196023e-05, "loss": 0.6013, "step": 9491 }, { "epoch": 0.338172684682117, "grad_norm": 1.0691461729320257, "learning_rate": 1.5419506779002568e-05, "loss": 0.493, "step": 9492 }, { "epoch": 0.33820831180861105, "grad_norm": 1.083700358182276, "learning_rate": 1.5418536962644593e-05, "loss": 0.5037, "step": 9493 }, { "epoch": 0.3382439389351052, "grad_norm": 1.6931635194541024, "learning_rate": 1.5417567074135006e-05, "loss": 0.6449, "step": 9494 }, { "epoch": 0.3382795660615993, "grad_norm": 1.7806278282241015, "learning_rate": 1.5416597113486734e-05, "loss": 0.5597, "step": 9495 }, { "epoch": 0.3383151931880934, "grad_norm": 1.8361839146512502, "learning_rate": 1.541562708071268e-05, "loss": 0.5084, "step": 9496 }, { "epoch": 0.3383508203145875, "grad_norm": 1.1168931514494032, "learning_rate": 1.5414656975825767e-05, "loss": 0.4667, "step": 9497 }, { "epoch": 0.33838644744108165, "grad_norm": 1.5867932328594716, "learning_rate": 1.5413686798838913e-05, "loss": 0.5513, "step": 9498 }, { "epoch": 0.3384220745675757, "grad_norm": 1.2743605238899371, "learning_rate": 1.541271654976504e-05, "loss": 0.5334, "step": 9499 }, { "epoch": 0.33845770169406986, "grad_norm": 1.7972740495183517, "learning_rate": 1.5411746228617058e-05, "loss": 0.616, "step": 9500 }, { "epoch": 0.338493328820564, "grad_norm": 1.8210142176558974, "learning_rate": 1.541077583540789e-05, "loss": 0.6358, "step": 9501 }, { "epoch": 0.33852895594705806, "grad_norm": 1.1820314949407345, "learning_rate": 1.5409805370150464e-05, "loss": 0.4551, "step": 9502 }, { "epoch": 0.3385645830735522, "grad_norm": 1.496948529608108, "learning_rate": 1.5408834832857695e-05, "loss": 0.6055, "step": 9503 }, { "epoch": 0.3386002102000463, "grad_norm": 1.4872920461929213, "learning_rate": 1.5407864223542512e-05, "loss": 0.5623, "step": 9504 }, { "epoch": 0.3386358373265404, "grad_norm": 1.1644223708991224, "learning_rate": 1.5406893542217835e-05, "loss": 0.4562, "step": 9505 }, { "epoch": 0.33867146445303453, "grad_norm": 1.509662861681542, "learning_rate": 1.540592278889659e-05, "loss": 0.5936, "step": 9506 }, { "epoch": 0.33870709157952866, "grad_norm": 1.686754396766983, "learning_rate": 1.540495196359171e-05, "loss": 0.5928, "step": 9507 }, { "epoch": 0.33874271870602274, "grad_norm": 1.5579527981639556, "learning_rate": 1.5403981066316115e-05, "loss": 0.6701, "step": 9508 }, { "epoch": 0.33877834583251687, "grad_norm": 0.995762328735439, "learning_rate": 1.5403010097082734e-05, "loss": 0.3828, "step": 9509 }, { "epoch": 0.338813972959011, "grad_norm": 1.1841063508034895, "learning_rate": 1.5402039055904496e-05, "loss": 0.4927, "step": 9510 }, { "epoch": 0.3388496000855051, "grad_norm": 1.1485945950229464, "learning_rate": 1.5401067942794332e-05, "loss": 0.5199, "step": 9511 }, { "epoch": 0.3388852272119992, "grad_norm": 1.121679028584158, "learning_rate": 1.5400096757765176e-05, "loss": 0.5007, "step": 9512 }, { "epoch": 0.33892085433849334, "grad_norm": 1.1332562261836479, "learning_rate": 1.5399125500829956e-05, "loss": 0.5354, "step": 9513 }, { "epoch": 0.33895648146498747, "grad_norm": 1.0717303066213257, "learning_rate": 1.5398154172001606e-05, "loss": 0.5273, "step": 9514 }, { "epoch": 0.33899210859148154, "grad_norm": 2.394720062747557, "learning_rate": 1.539718277129306e-05, "loss": 0.5529, "step": 9515 }, { "epoch": 0.3390277357179757, "grad_norm": 1.1340274616435855, "learning_rate": 1.5396211298717248e-05, "loss": 0.466, "step": 9516 }, { "epoch": 0.3390633628444698, "grad_norm": 1.100314525977963, "learning_rate": 1.539523975428712e-05, "loss": 0.479, "step": 9517 }, { "epoch": 0.3390989899709639, "grad_norm": 1.306174798201542, "learning_rate": 1.5394268138015598e-05, "loss": 0.4551, "step": 9518 }, { "epoch": 0.339134617097458, "grad_norm": 1.2285321261263522, "learning_rate": 1.5393296449915627e-05, "loss": 0.4724, "step": 9519 }, { "epoch": 0.33917024422395214, "grad_norm": 1.2913830860434512, "learning_rate": 1.539232469000014e-05, "loss": 0.5533, "step": 9520 }, { "epoch": 0.3392058713504462, "grad_norm": 1.1926415162738817, "learning_rate": 1.539135285828209e-05, "loss": 0.4665, "step": 9521 }, { "epoch": 0.33924149847694035, "grad_norm": 1.1485459580700752, "learning_rate": 1.53903809547744e-05, "loss": 0.4838, "step": 9522 }, { "epoch": 0.3392771256034345, "grad_norm": 1.489170214400921, "learning_rate": 1.5389408979490022e-05, "loss": 0.5779, "step": 9523 }, { "epoch": 0.33931275272992856, "grad_norm": 1.102280486941984, "learning_rate": 1.53884369324419e-05, "loss": 0.4499, "step": 9524 }, { "epoch": 0.3393483798564227, "grad_norm": 1.1076299425793958, "learning_rate": 1.538746481364297e-05, "loss": 0.4694, "step": 9525 }, { "epoch": 0.3393840069829168, "grad_norm": 1.7051754643653607, "learning_rate": 1.538649262310618e-05, "loss": 0.6155, "step": 9526 }, { "epoch": 0.3394196341094109, "grad_norm": 1.1676433101145425, "learning_rate": 1.538552036084448e-05, "loss": 0.5021, "step": 9527 }, { "epoch": 0.339455261235905, "grad_norm": 1.084757675423388, "learning_rate": 1.538454802687081e-05, "loss": 0.5043, "step": 9528 }, { "epoch": 0.33949088836239916, "grad_norm": 1.0396170795665316, "learning_rate": 1.538357562119812e-05, "loss": 0.4393, "step": 9529 }, { "epoch": 0.33952651548889323, "grad_norm": 1.1651398204910073, "learning_rate": 1.5382603143839353e-05, "loss": 0.5264, "step": 9530 }, { "epoch": 0.33956214261538736, "grad_norm": 1.1299494485263677, "learning_rate": 1.5381630594807465e-05, "loss": 0.5227, "step": 9531 }, { "epoch": 0.3395977697418815, "grad_norm": 1.723005390110772, "learning_rate": 1.5380657974115407e-05, "loss": 0.4579, "step": 9532 }, { "epoch": 0.33963339686837557, "grad_norm": 1.0851226588441467, "learning_rate": 1.5379685281776125e-05, "loss": 0.4688, "step": 9533 }, { "epoch": 0.3396690239948697, "grad_norm": 1.7503393547901662, "learning_rate": 1.5378712517802576e-05, "loss": 0.6794, "step": 9534 }, { "epoch": 0.33970465112136383, "grad_norm": 1.0573928622238844, "learning_rate": 1.537773968220771e-05, "loss": 0.5189, "step": 9535 }, { "epoch": 0.3397402782478579, "grad_norm": 1.128681529360066, "learning_rate": 1.5376766775004476e-05, "loss": 0.5019, "step": 9536 }, { "epoch": 0.33977590537435204, "grad_norm": 1.2356599216754365, "learning_rate": 1.537579379620584e-05, "loss": 0.4715, "step": 9537 }, { "epoch": 0.33981153250084617, "grad_norm": 1.0739635568068735, "learning_rate": 1.5374820745824753e-05, "loss": 0.479, "step": 9538 }, { "epoch": 0.33984715962734025, "grad_norm": 1.7868343019394024, "learning_rate": 1.537384762387417e-05, "loss": 0.5757, "step": 9539 }, { "epoch": 0.3398827867538344, "grad_norm": 1.5950593343206423, "learning_rate": 1.5372874430367044e-05, "loss": 0.5689, "step": 9540 }, { "epoch": 0.3399184138803285, "grad_norm": 1.6055409245507064, "learning_rate": 1.5371901165316342e-05, "loss": 0.6374, "step": 9541 }, { "epoch": 0.3399540410068226, "grad_norm": 1.5575579971030824, "learning_rate": 1.5370927828735026e-05, "loss": 0.5555, "step": 9542 }, { "epoch": 0.3399896681333167, "grad_norm": 0.9947230448410794, "learning_rate": 1.5369954420636048e-05, "loss": 0.503, "step": 9543 }, { "epoch": 0.34002529525981084, "grad_norm": 1.686493282331352, "learning_rate": 1.536898094103237e-05, "loss": 0.6059, "step": 9544 }, { "epoch": 0.3400609223863049, "grad_norm": 1.5054355532307828, "learning_rate": 1.5368007389936965e-05, "loss": 0.6155, "step": 9545 }, { "epoch": 0.34009654951279905, "grad_norm": 1.3356367139921592, "learning_rate": 1.5367033767362787e-05, "loss": 0.6376, "step": 9546 }, { "epoch": 0.3401321766392932, "grad_norm": 1.9557551543429725, "learning_rate": 1.5366060073322803e-05, "loss": 0.6358, "step": 9547 }, { "epoch": 0.34016780376578726, "grad_norm": 1.320927276483996, "learning_rate": 1.536508630782998e-05, "loss": 0.5098, "step": 9548 }, { "epoch": 0.3402034308922814, "grad_norm": 0.9837806861007883, "learning_rate": 1.536411247089728e-05, "loss": 0.454, "step": 9549 }, { "epoch": 0.3402390580187755, "grad_norm": 1.813900307615208, "learning_rate": 1.536313856253768e-05, "loss": 0.5717, "step": 9550 }, { "epoch": 0.3402746851452696, "grad_norm": 1.05923373069325, "learning_rate": 1.536216458276413e-05, "loss": 0.4448, "step": 9551 }, { "epoch": 0.3403103122717637, "grad_norm": 1.3114481829315106, "learning_rate": 1.5361190531589615e-05, "loss": 0.5086, "step": 9552 }, { "epoch": 0.34034593939825786, "grad_norm": 1.1231436674228148, "learning_rate": 1.5360216409027104e-05, "loss": 0.4778, "step": 9553 }, { "epoch": 0.34038156652475193, "grad_norm": 1.068059679712143, "learning_rate": 1.5359242215089565e-05, "loss": 0.4488, "step": 9554 }, { "epoch": 0.34041719365124606, "grad_norm": 1.0325879431472325, "learning_rate": 1.5358267949789968e-05, "loss": 0.4314, "step": 9555 }, { "epoch": 0.3404528207777402, "grad_norm": 1.2516755436876195, "learning_rate": 1.535729361314129e-05, "loss": 0.4689, "step": 9556 }, { "epoch": 0.34048844790423427, "grad_norm": 1.385964845835109, "learning_rate": 1.53563192051565e-05, "loss": 0.6001, "step": 9557 }, { "epoch": 0.3405240750307284, "grad_norm": 1.4413546793095675, "learning_rate": 1.535534472584858e-05, "loss": 0.567, "step": 9558 }, { "epoch": 0.34055970215722253, "grad_norm": 0.9810594488323046, "learning_rate": 1.5354370175230498e-05, "loss": 0.4414, "step": 9559 }, { "epoch": 0.3405953292837166, "grad_norm": 1.0803746578554199, "learning_rate": 1.5353395553315234e-05, "loss": 0.4987, "step": 9560 }, { "epoch": 0.34063095641021074, "grad_norm": 1.1780105720491705, "learning_rate": 1.5352420860115768e-05, "loss": 0.503, "step": 9561 }, { "epoch": 0.34066658353670487, "grad_norm": 1.5104342610288999, "learning_rate": 1.5351446095645076e-05, "loss": 0.6039, "step": 9562 }, { "epoch": 0.34070221066319895, "grad_norm": 1.2807050312769948, "learning_rate": 1.535047125991614e-05, "loss": 0.5316, "step": 9563 }, { "epoch": 0.3407378377896931, "grad_norm": 1.4246257861440348, "learning_rate": 1.534949635294194e-05, "loss": 0.5188, "step": 9564 }, { "epoch": 0.3407734649161872, "grad_norm": 1.2113001389028824, "learning_rate": 1.5348521374735458e-05, "loss": 0.5017, "step": 9565 }, { "epoch": 0.3408090920426813, "grad_norm": 1.6607613377816484, "learning_rate": 1.5347546325309672e-05, "loss": 0.638, "step": 9566 }, { "epoch": 0.3408447191691754, "grad_norm": 1.8713861644211536, "learning_rate": 1.534657120467757e-05, "loss": 0.5603, "step": 9567 }, { "epoch": 0.34088034629566955, "grad_norm": 1.4605841219472246, "learning_rate": 1.534559601285214e-05, "loss": 0.5859, "step": 9568 }, { "epoch": 0.3409159734221636, "grad_norm": 1.1635008432287985, "learning_rate": 1.534462074984636e-05, "loss": 0.5239, "step": 9569 }, { "epoch": 0.34095160054865775, "grad_norm": 1.9139088104199866, "learning_rate": 1.5343645415673216e-05, "loss": 0.6103, "step": 9570 }, { "epoch": 0.3409872276751519, "grad_norm": 1.0730488650685384, "learning_rate": 1.53426700103457e-05, "loss": 0.4379, "step": 9571 }, { "epoch": 0.34102285480164596, "grad_norm": 1.3496257728098653, "learning_rate": 1.5341694533876802e-05, "loss": 0.5665, "step": 9572 }, { "epoch": 0.3410584819281401, "grad_norm": 1.1431441727112588, "learning_rate": 1.5340718986279505e-05, "loss": 0.4808, "step": 9573 }, { "epoch": 0.3410941090546342, "grad_norm": 1.2841231138252704, "learning_rate": 1.5339743367566796e-05, "loss": 0.5296, "step": 9574 }, { "epoch": 0.3411297361811283, "grad_norm": 1.1207011633241344, "learning_rate": 1.5338767677751677e-05, "loss": 0.4971, "step": 9575 }, { "epoch": 0.3411653633076224, "grad_norm": 1.7966530537841774, "learning_rate": 1.5337791916847137e-05, "loss": 0.6064, "step": 9576 }, { "epoch": 0.34120099043411656, "grad_norm": 1.554118155191594, "learning_rate": 1.5336816084866165e-05, "loss": 0.565, "step": 9577 }, { "epoch": 0.34123661756061063, "grad_norm": 1.0179435758957727, "learning_rate": 1.5335840181821757e-05, "loss": 0.4889, "step": 9578 }, { "epoch": 0.34127224468710476, "grad_norm": 1.0798558267714813, "learning_rate": 1.5334864207726908e-05, "loss": 0.511, "step": 9579 }, { "epoch": 0.3413078718135989, "grad_norm": 1.7902020521830797, "learning_rate": 1.533388816259461e-05, "loss": 0.6412, "step": 9580 }, { "epoch": 0.34134349894009297, "grad_norm": 1.5522820453875306, "learning_rate": 1.5332912046437866e-05, "loss": 0.7118, "step": 9581 }, { "epoch": 0.3413791260665871, "grad_norm": 1.4995300397507816, "learning_rate": 1.533193585926967e-05, "loss": 0.6562, "step": 9582 }, { "epoch": 0.34141475319308123, "grad_norm": 1.0954273945339852, "learning_rate": 1.533095960110302e-05, "loss": 0.4125, "step": 9583 }, { "epoch": 0.3414503803195753, "grad_norm": 1.1058484711687722, "learning_rate": 1.5329983271950916e-05, "loss": 0.5214, "step": 9584 }, { "epoch": 0.34148600744606944, "grad_norm": 1.1011085457617475, "learning_rate": 1.5329006871826365e-05, "loss": 0.4349, "step": 9585 }, { "epoch": 0.34152163457256357, "grad_norm": 1.4880392743546902, "learning_rate": 1.5328030400742355e-05, "loss": 0.5464, "step": 9586 }, { "epoch": 0.34155726169905765, "grad_norm": 1.1130372142321074, "learning_rate": 1.53270538587119e-05, "loss": 0.4698, "step": 9587 }, { "epoch": 0.3415928888255518, "grad_norm": 1.3220098428857485, "learning_rate": 1.5326077245747998e-05, "loss": 0.5093, "step": 9588 }, { "epoch": 0.3416285159520459, "grad_norm": 1.0758681857540242, "learning_rate": 1.5325100561863656e-05, "loss": 0.4849, "step": 9589 }, { "epoch": 0.34166414307854, "grad_norm": 1.373572574732069, "learning_rate": 1.532412380707188e-05, "loss": 0.623, "step": 9590 }, { "epoch": 0.3416997702050341, "grad_norm": 1.1178262115509605, "learning_rate": 1.5323146981385676e-05, "loss": 0.4629, "step": 9591 }, { "epoch": 0.34173539733152825, "grad_norm": 1.0127688905468522, "learning_rate": 1.532217008481804e-05, "loss": 0.4766, "step": 9592 }, { "epoch": 0.3417710244580223, "grad_norm": 1.051090378256787, "learning_rate": 1.5321193117381998e-05, "loss": 0.5065, "step": 9593 }, { "epoch": 0.34180665158451645, "grad_norm": 1.481196035628271, "learning_rate": 1.5320216079090546e-05, "loss": 0.6331, "step": 9594 }, { "epoch": 0.3418422787110106, "grad_norm": 2.000188072351666, "learning_rate": 1.5319238969956697e-05, "loss": 0.6083, "step": 9595 }, { "epoch": 0.34187790583750466, "grad_norm": 1.1353112486828478, "learning_rate": 1.5318261789993466e-05, "loss": 0.4858, "step": 9596 }, { "epoch": 0.3419135329639988, "grad_norm": 1.3963564874570902, "learning_rate": 1.5317284539213862e-05, "loss": 0.5633, "step": 9597 }, { "epoch": 0.3419491600904929, "grad_norm": 1.914180178192731, "learning_rate": 1.53163072176309e-05, "loss": 0.6159, "step": 9598 }, { "epoch": 0.341984787216987, "grad_norm": 2.2425274957349703, "learning_rate": 1.531532982525759e-05, "loss": 0.4806, "step": 9599 }, { "epoch": 0.3420204143434811, "grad_norm": 1.108146040206999, "learning_rate": 1.5314352362106943e-05, "loss": 0.4927, "step": 9600 }, { "epoch": 0.34205604146997526, "grad_norm": 1.440893144824301, "learning_rate": 1.5313374828191987e-05, "loss": 0.5881, "step": 9601 }, { "epoch": 0.34209166859646933, "grad_norm": 1.3970576775088055, "learning_rate": 1.5312397223525728e-05, "loss": 0.5757, "step": 9602 }, { "epoch": 0.34212729572296346, "grad_norm": 1.3451467660673555, "learning_rate": 1.5311419548121186e-05, "loss": 0.5834, "step": 9603 }, { "epoch": 0.3421629228494576, "grad_norm": 1.4083668985229603, "learning_rate": 1.531044180199138e-05, "loss": 0.5335, "step": 9604 }, { "epoch": 0.34219854997595167, "grad_norm": 1.5049569664130422, "learning_rate": 1.5309463985149332e-05, "loss": 0.574, "step": 9605 }, { "epoch": 0.3422341771024458, "grad_norm": 1.262324630056548, "learning_rate": 1.530848609760806e-05, "loss": 0.4735, "step": 9606 }, { "epoch": 0.34226980422893993, "grad_norm": 1.610624593931538, "learning_rate": 1.5307508139380583e-05, "loss": 0.5532, "step": 9607 }, { "epoch": 0.342305431355434, "grad_norm": 1.2618843915889224, "learning_rate": 1.530653011047993e-05, "loss": 0.5035, "step": 9608 }, { "epoch": 0.34234105848192814, "grad_norm": 1.0839728551615209, "learning_rate": 1.5305552010919115e-05, "loss": 0.4834, "step": 9609 }, { "epoch": 0.34237668560842227, "grad_norm": 1.119561130508303, "learning_rate": 1.5304573840711166e-05, "loss": 0.5079, "step": 9610 }, { "epoch": 0.34241231273491635, "grad_norm": 1.1166249953859095, "learning_rate": 1.530359559986911e-05, "loss": 0.4528, "step": 9611 }, { "epoch": 0.3424479398614105, "grad_norm": 1.2064546192844217, "learning_rate": 1.5302617288405972e-05, "loss": 0.4795, "step": 9612 }, { "epoch": 0.3424835669879046, "grad_norm": 1.36403945754405, "learning_rate": 1.530163890633478e-05, "loss": 0.5986, "step": 9613 }, { "epoch": 0.3425191941143987, "grad_norm": 1.4970099102987369, "learning_rate": 1.5300660453668557e-05, "loss": 0.5939, "step": 9614 }, { "epoch": 0.3425548212408928, "grad_norm": 1.117549213036326, "learning_rate": 1.529968193042034e-05, "loss": 0.4907, "step": 9615 }, { "epoch": 0.34259044836738695, "grad_norm": 1.040319697888139, "learning_rate": 1.529870333660315e-05, "loss": 0.4543, "step": 9616 }, { "epoch": 0.342626075493881, "grad_norm": 1.182700395346081, "learning_rate": 1.5297724672230023e-05, "loss": 0.4534, "step": 9617 }, { "epoch": 0.34266170262037515, "grad_norm": 1.181447303709256, "learning_rate": 1.529674593731399e-05, "loss": 0.5268, "step": 9618 }, { "epoch": 0.3426973297468693, "grad_norm": 1.0605954440909344, "learning_rate": 1.529576713186808e-05, "loss": 0.4436, "step": 9619 }, { "epoch": 0.34273295687336336, "grad_norm": 1.0864683854917114, "learning_rate": 1.5294788255905328e-05, "loss": 0.466, "step": 9620 }, { "epoch": 0.3427685839998575, "grad_norm": 1.0017336750928654, "learning_rate": 1.529380930943877e-05, "loss": 0.4353, "step": 9621 }, { "epoch": 0.3428042111263516, "grad_norm": 1.1196461300967346, "learning_rate": 1.5292830292481445e-05, "loss": 0.489, "step": 9622 }, { "epoch": 0.3428398382528457, "grad_norm": 1.1136069413795708, "learning_rate": 1.5291851205046383e-05, "loss": 0.495, "step": 9623 }, { "epoch": 0.3428754653793398, "grad_norm": 1.9226954194211983, "learning_rate": 1.5290872047146623e-05, "loss": 0.6451, "step": 9624 }, { "epoch": 0.34291109250583396, "grad_norm": 1.378694690321727, "learning_rate": 1.5289892818795205e-05, "loss": 0.5672, "step": 9625 }, { "epoch": 0.34294671963232803, "grad_norm": 1.5896875793939063, "learning_rate": 1.5288913520005162e-05, "loss": 0.6214, "step": 9626 }, { "epoch": 0.34298234675882217, "grad_norm": 1.6067013784801674, "learning_rate": 1.5287934150789543e-05, "loss": 0.5893, "step": 9627 }, { "epoch": 0.3430179738853163, "grad_norm": 1.1159571818869387, "learning_rate": 1.5286954711161387e-05, "loss": 0.4917, "step": 9628 }, { "epoch": 0.34305360101181037, "grad_norm": 1.1127880168236213, "learning_rate": 1.528597520113373e-05, "loss": 0.4567, "step": 9629 }, { "epoch": 0.3430892281383045, "grad_norm": 1.237837264047616, "learning_rate": 1.5284995620719618e-05, "loss": 0.6196, "step": 9630 }, { "epoch": 0.34312485526479863, "grad_norm": 1.5579154466776834, "learning_rate": 1.5284015969932097e-05, "loss": 0.6022, "step": 9631 }, { "epoch": 0.3431604823912927, "grad_norm": 1.4186350577833702, "learning_rate": 1.528303624878421e-05, "loss": 0.6522, "step": 9632 }, { "epoch": 0.34319610951778684, "grad_norm": 1.1185560858610666, "learning_rate": 1.5282056457289002e-05, "loss": 0.5297, "step": 9633 }, { "epoch": 0.34323173664428097, "grad_norm": 1.8077553531470238, "learning_rate": 1.528107659545952e-05, "loss": 0.6737, "step": 9634 }, { "epoch": 0.34326736377077505, "grad_norm": 0.9978506113467525, "learning_rate": 1.5280096663308813e-05, "loss": 0.4587, "step": 9635 }, { "epoch": 0.3433029908972692, "grad_norm": 1.1399198995757802, "learning_rate": 1.5279116660849926e-05, "loss": 0.4783, "step": 9636 }, { "epoch": 0.3433386180237633, "grad_norm": 1.1218991128580733, "learning_rate": 1.5278136588095914e-05, "loss": 0.4661, "step": 9637 }, { "epoch": 0.3433742451502574, "grad_norm": 1.6323372666776954, "learning_rate": 1.527715644505982e-05, "loss": 0.6018, "step": 9638 }, { "epoch": 0.3434098722767515, "grad_norm": 1.2084877022562877, "learning_rate": 1.5276176231754705e-05, "loss": 0.4974, "step": 9639 }, { "epoch": 0.34344549940324565, "grad_norm": 1.1146114713820752, "learning_rate": 1.5275195948193613e-05, "loss": 0.4467, "step": 9640 }, { "epoch": 0.3434811265297397, "grad_norm": 1.1784266770986938, "learning_rate": 1.52742155943896e-05, "loss": 0.4597, "step": 9641 }, { "epoch": 0.34351675365623385, "grad_norm": 1.1962710146930218, "learning_rate": 1.5273235170355717e-05, "loss": 0.4849, "step": 9642 }, { "epoch": 0.343552380782728, "grad_norm": 1.1765862277951888, "learning_rate": 1.5272254676105026e-05, "loss": 0.5279, "step": 9643 }, { "epoch": 0.34358800790922206, "grad_norm": 1.094533686016237, "learning_rate": 1.5271274111650577e-05, "loss": 0.5514, "step": 9644 }, { "epoch": 0.3436236350357162, "grad_norm": 1.1185893863011453, "learning_rate": 1.5270293477005428e-05, "loss": 0.5277, "step": 9645 }, { "epoch": 0.3436592621622103, "grad_norm": 1.8971192244668897, "learning_rate": 1.5269312772182637e-05, "loss": 0.5878, "step": 9646 }, { "epoch": 0.3436948892887044, "grad_norm": 1.4665151710393967, "learning_rate": 1.526833199719527e-05, "loss": 0.5999, "step": 9647 }, { "epoch": 0.34373051641519853, "grad_norm": 1.1086285783831382, "learning_rate": 1.5267351152056375e-05, "loss": 0.4903, "step": 9648 }, { "epoch": 0.34376614354169266, "grad_norm": 1.6123835358212109, "learning_rate": 1.5266370236779015e-05, "loss": 0.57, "step": 9649 }, { "epoch": 0.34380177066818673, "grad_norm": 1.186643772412218, "learning_rate": 1.526538925137626e-05, "loss": 0.4634, "step": 9650 }, { "epoch": 0.34383739779468087, "grad_norm": 1.140229404482159, "learning_rate": 1.5264408195861164e-05, "loss": 0.4677, "step": 9651 }, { "epoch": 0.343873024921175, "grad_norm": 1.214819305450906, "learning_rate": 1.526342707024679e-05, "loss": 0.4952, "step": 9652 }, { "epoch": 0.3439086520476691, "grad_norm": 1.4427685925098297, "learning_rate": 1.526244587454621e-05, "loss": 0.5915, "step": 9653 }, { "epoch": 0.3439442791741632, "grad_norm": 1.0859971396364059, "learning_rate": 1.5261464608772487e-05, "loss": 0.5047, "step": 9654 }, { "epoch": 0.34397990630065733, "grad_norm": 1.6437352300068466, "learning_rate": 1.526048327293868e-05, "loss": 0.6187, "step": 9655 }, { "epoch": 0.3440155334271514, "grad_norm": 1.562502733393399, "learning_rate": 1.5259501867057864e-05, "loss": 0.625, "step": 9656 }, { "epoch": 0.34405116055364554, "grad_norm": 1.2231686564524609, "learning_rate": 1.5258520391143106e-05, "loss": 0.5234, "step": 9657 }, { "epoch": 0.34408678768013967, "grad_norm": 1.0608209193045108, "learning_rate": 1.525753884520747e-05, "loss": 0.4866, "step": 9658 }, { "epoch": 0.34412241480663375, "grad_norm": 1.2323130351402347, "learning_rate": 1.525655722926403e-05, "loss": 0.5604, "step": 9659 }, { "epoch": 0.3441580419331279, "grad_norm": 1.40531106404767, "learning_rate": 1.5255575543325859e-05, "loss": 0.6037, "step": 9660 }, { "epoch": 0.344193669059622, "grad_norm": 1.142508523427906, "learning_rate": 1.5254593787406026e-05, "loss": 0.4822, "step": 9661 }, { "epoch": 0.3442292961861161, "grad_norm": 1.1435041312930363, "learning_rate": 1.5253611961517603e-05, "loss": 0.5159, "step": 9662 }, { "epoch": 0.3442649233126102, "grad_norm": 2.081033352306558, "learning_rate": 1.5252630065673662e-05, "loss": 0.6574, "step": 9663 }, { "epoch": 0.34430055043910435, "grad_norm": 1.5899151036787755, "learning_rate": 1.5251648099887285e-05, "loss": 0.665, "step": 9664 }, { "epoch": 0.3443361775655984, "grad_norm": 1.1304779138036913, "learning_rate": 1.525066606417154e-05, "loss": 0.445, "step": 9665 }, { "epoch": 0.34437180469209255, "grad_norm": 1.1163959004477935, "learning_rate": 1.5249683958539504e-05, "loss": 0.4884, "step": 9666 }, { "epoch": 0.3444074318185867, "grad_norm": 1.1473150762519329, "learning_rate": 1.5248701783004261e-05, "loss": 0.4922, "step": 9667 }, { "epoch": 0.34444305894508076, "grad_norm": 1.183475091338121, "learning_rate": 1.5247719537578882e-05, "loss": 0.5248, "step": 9668 }, { "epoch": 0.3444786860715749, "grad_norm": 1.4712523670304574, "learning_rate": 1.5246737222276454e-05, "loss": 0.5854, "step": 9669 }, { "epoch": 0.344514313198069, "grad_norm": 1.1646803475608023, "learning_rate": 1.5245754837110048e-05, "loss": 0.5448, "step": 9670 }, { "epoch": 0.3445499403245631, "grad_norm": 1.0754055437797618, "learning_rate": 1.5244772382092751e-05, "loss": 0.4926, "step": 9671 }, { "epoch": 0.34458556745105723, "grad_norm": 1.119447045577784, "learning_rate": 1.5243789857237646e-05, "loss": 0.4843, "step": 9672 }, { "epoch": 0.34462119457755136, "grad_norm": 1.0298564601655436, "learning_rate": 1.5242807262557815e-05, "loss": 0.4273, "step": 9673 }, { "epoch": 0.34465682170404544, "grad_norm": 1.0893274070758008, "learning_rate": 1.5241824598066335e-05, "loss": 0.5478, "step": 9674 }, { "epoch": 0.34469244883053957, "grad_norm": 1.3908898076584717, "learning_rate": 1.52408418637763e-05, "loss": 0.4593, "step": 9675 }, { "epoch": 0.3447280759570337, "grad_norm": 2.000549372187778, "learning_rate": 1.5239859059700794e-05, "loss": 0.5605, "step": 9676 }, { "epoch": 0.3447637030835278, "grad_norm": 1.2197599161762644, "learning_rate": 1.52388761858529e-05, "loss": 0.559, "step": 9677 }, { "epoch": 0.3447993302100219, "grad_norm": 1.1173180020379612, "learning_rate": 1.5237893242245707e-05, "loss": 0.4925, "step": 9678 }, { "epoch": 0.34483495733651603, "grad_norm": 1.6400359637534843, "learning_rate": 1.5236910228892306e-05, "loss": 0.5811, "step": 9679 }, { "epoch": 0.3448705844630101, "grad_norm": 1.447316716654232, "learning_rate": 1.5235927145805785e-05, "loss": 0.6577, "step": 9680 }, { "epoch": 0.34490621158950424, "grad_norm": 2.053148209460509, "learning_rate": 1.5234943992999235e-05, "loss": 0.5499, "step": 9681 }, { "epoch": 0.3449418387159984, "grad_norm": 1.256631131602109, "learning_rate": 1.5233960770485742e-05, "loss": 0.4508, "step": 9682 }, { "epoch": 0.34497746584249245, "grad_norm": 1.527170444938723, "learning_rate": 1.523297747827841e-05, "loss": 0.6377, "step": 9683 }, { "epoch": 0.3450130929689866, "grad_norm": 1.1092742538319045, "learning_rate": 1.5231994116390322e-05, "loss": 0.5479, "step": 9684 }, { "epoch": 0.3450487200954807, "grad_norm": 1.0650586991186914, "learning_rate": 1.5231010684834574e-05, "loss": 0.504, "step": 9685 }, { "epoch": 0.3450843472219748, "grad_norm": 1.441216486204654, "learning_rate": 1.5230027183624265e-05, "loss": 0.4991, "step": 9686 }, { "epoch": 0.3451199743484689, "grad_norm": 1.7011707972504115, "learning_rate": 1.5229043612772488e-05, "loss": 0.487, "step": 9687 }, { "epoch": 0.34515560147496305, "grad_norm": 1.5155679278063114, "learning_rate": 1.5228059972292339e-05, "loss": 0.5799, "step": 9688 }, { "epoch": 0.3451912286014571, "grad_norm": 1.1144538928036307, "learning_rate": 1.5227076262196917e-05, "loss": 0.4688, "step": 9689 }, { "epoch": 0.34522685572795125, "grad_norm": 1.0325914821973006, "learning_rate": 1.5226092482499321e-05, "loss": 0.4997, "step": 9690 }, { "epoch": 0.3452624828544454, "grad_norm": 1.4419891833990413, "learning_rate": 1.5225108633212652e-05, "loss": 0.5689, "step": 9691 }, { "epoch": 0.34529810998093946, "grad_norm": 1.893157151646998, "learning_rate": 1.5224124714350011e-05, "loss": 0.5576, "step": 9692 }, { "epoch": 0.3453337371074336, "grad_norm": 1.8700478036936972, "learning_rate": 1.5223140725924494e-05, "loss": 0.6121, "step": 9693 }, { "epoch": 0.3453693642339277, "grad_norm": 1.3091591252836525, "learning_rate": 1.5222156667949212e-05, "loss": 0.5884, "step": 9694 }, { "epoch": 0.3454049913604218, "grad_norm": 1.159447657228819, "learning_rate": 1.522117254043726e-05, "loss": 0.4737, "step": 9695 }, { "epoch": 0.34544061848691593, "grad_norm": 1.0302111318216776, "learning_rate": 1.5220188343401749e-05, "loss": 0.4839, "step": 9696 }, { "epoch": 0.34547624561341006, "grad_norm": 1.1099151470302464, "learning_rate": 1.5219204076855781e-05, "loss": 0.478, "step": 9697 }, { "epoch": 0.34551187273990414, "grad_norm": 1.2952620744190506, "learning_rate": 1.521821974081246e-05, "loss": 0.5205, "step": 9698 }, { "epoch": 0.34554749986639827, "grad_norm": 1.5193056332453885, "learning_rate": 1.5217235335284902e-05, "loss": 0.5474, "step": 9699 }, { "epoch": 0.3455831269928924, "grad_norm": 1.0920680296363574, "learning_rate": 1.5216250860286201e-05, "loss": 0.4861, "step": 9700 }, { "epoch": 0.3456187541193865, "grad_norm": 1.6929115941349686, "learning_rate": 1.5215266315829477e-05, "loss": 0.5783, "step": 9701 }, { "epoch": 0.3456543812458806, "grad_norm": 1.6388472505003318, "learning_rate": 1.5214281701927837e-05, "loss": 0.5525, "step": 9702 }, { "epoch": 0.34569000837237474, "grad_norm": 1.3794164148149943, "learning_rate": 1.5213297018594393e-05, "loss": 0.6242, "step": 9703 }, { "epoch": 0.34572563549886887, "grad_norm": 1.8499699271886874, "learning_rate": 1.5212312265842256e-05, "loss": 0.5692, "step": 9704 }, { "epoch": 0.34576126262536294, "grad_norm": 1.5560249988900974, "learning_rate": 1.5211327443684533e-05, "loss": 0.5608, "step": 9705 }, { "epoch": 0.3457968897518571, "grad_norm": 1.4639005029395253, "learning_rate": 1.5210342552134348e-05, "loss": 0.6293, "step": 9706 }, { "epoch": 0.3458325168783512, "grad_norm": 1.0148081775485507, "learning_rate": 1.520935759120481e-05, "loss": 0.4631, "step": 9707 }, { "epoch": 0.3458681440048453, "grad_norm": 2.309634092889617, "learning_rate": 1.5208372560909031e-05, "loss": 0.5183, "step": 9708 }, { "epoch": 0.3459037711313394, "grad_norm": 1.7087712775212487, "learning_rate": 1.5207387461260132e-05, "loss": 0.6024, "step": 9709 }, { "epoch": 0.34593939825783354, "grad_norm": 1.1887489856206237, "learning_rate": 1.520640229227123e-05, "loss": 0.4663, "step": 9710 }, { "epoch": 0.3459750253843276, "grad_norm": 1.5957818694236676, "learning_rate": 1.520541705395544e-05, "loss": 0.4896, "step": 9711 }, { "epoch": 0.34601065251082175, "grad_norm": 1.334322347110178, "learning_rate": 1.5204431746325888e-05, "loss": 0.5817, "step": 9712 }, { "epoch": 0.3460462796373159, "grad_norm": 1.1140494644997954, "learning_rate": 1.5203446369395685e-05, "loss": 0.4688, "step": 9713 }, { "epoch": 0.34608190676380995, "grad_norm": 1.0472280964618392, "learning_rate": 1.520246092317796e-05, "loss": 0.4354, "step": 9714 }, { "epoch": 0.3461175338903041, "grad_norm": 1.0737209639554577, "learning_rate": 1.5201475407685831e-05, "loss": 0.4912, "step": 9715 }, { "epoch": 0.3461531610167982, "grad_norm": 1.3366598926546949, "learning_rate": 1.520048982293242e-05, "loss": 0.5545, "step": 9716 }, { "epoch": 0.3461887881432923, "grad_norm": 1.7573231021519276, "learning_rate": 1.5199504168930852e-05, "loss": 0.5791, "step": 9717 }, { "epoch": 0.3462244152697864, "grad_norm": 1.3322912789473575, "learning_rate": 1.5198518445694255e-05, "loss": 0.534, "step": 9718 }, { "epoch": 0.34626004239628055, "grad_norm": 1.0880802759176873, "learning_rate": 1.5197532653235747e-05, "loss": 0.4637, "step": 9719 }, { "epoch": 0.34629566952277463, "grad_norm": 1.1130894988870927, "learning_rate": 1.5196546791568461e-05, "loss": 0.5017, "step": 9720 }, { "epoch": 0.34633129664926876, "grad_norm": 1.9013954063782246, "learning_rate": 1.5195560860705522e-05, "loss": 0.5673, "step": 9721 }, { "epoch": 0.3463669237757629, "grad_norm": 1.3710669083679372, "learning_rate": 1.5194574860660058e-05, "loss": 0.5441, "step": 9722 }, { "epoch": 0.34640255090225697, "grad_norm": 2.0830895346125677, "learning_rate": 1.5193588791445199e-05, "loss": 0.6322, "step": 9723 }, { "epoch": 0.3464381780287511, "grad_norm": 1.8825470957840702, "learning_rate": 1.5192602653074077e-05, "loss": 0.6521, "step": 9724 }, { "epoch": 0.34647380515524523, "grad_norm": 1.272888620097273, "learning_rate": 1.519161644555982e-05, "loss": 0.553, "step": 9725 }, { "epoch": 0.3465094322817393, "grad_norm": 1.0833641831873586, "learning_rate": 1.5190630168915563e-05, "loss": 0.5439, "step": 9726 }, { "epoch": 0.34654505940823344, "grad_norm": 1.1356253305481598, "learning_rate": 1.5189643823154435e-05, "loss": 0.4839, "step": 9727 }, { "epoch": 0.34658068653472757, "grad_norm": 1.4706163097113996, "learning_rate": 1.5188657408289574e-05, "loss": 0.5942, "step": 9728 }, { "epoch": 0.34661631366122164, "grad_norm": 1.3620118473049263, "learning_rate": 1.518767092433411e-05, "loss": 0.5511, "step": 9729 }, { "epoch": 0.3466519407877158, "grad_norm": 1.0841134369684975, "learning_rate": 1.5186684371301183e-05, "loss": 0.5258, "step": 9730 }, { "epoch": 0.3466875679142099, "grad_norm": 1.499223733325051, "learning_rate": 1.5185697749203928e-05, "loss": 0.572, "step": 9731 }, { "epoch": 0.346723195040704, "grad_norm": 3.563073002004993, "learning_rate": 1.5184711058055483e-05, "loss": 0.5129, "step": 9732 }, { "epoch": 0.3467588221671981, "grad_norm": 1.120172958515497, "learning_rate": 1.518372429786899e-05, "loss": 0.5131, "step": 9733 }, { "epoch": 0.34679444929369224, "grad_norm": 1.5319532666042457, "learning_rate": 1.518273746865758e-05, "loss": 0.5498, "step": 9734 }, { "epoch": 0.3468300764201863, "grad_norm": 2.0470509429941868, "learning_rate": 1.5181750570434402e-05, "loss": 0.6572, "step": 9735 }, { "epoch": 0.34686570354668045, "grad_norm": 1.065235788997919, "learning_rate": 1.518076360321259e-05, "loss": 0.4838, "step": 9736 }, { "epoch": 0.3469013306731746, "grad_norm": 1.9707180097324686, "learning_rate": 1.5179776567005289e-05, "loss": 0.5983, "step": 9737 }, { "epoch": 0.34693695779966865, "grad_norm": 1.9293092684754332, "learning_rate": 1.5178789461825645e-05, "loss": 0.525, "step": 9738 }, { "epoch": 0.3469725849261628, "grad_norm": 1.4960293107300433, "learning_rate": 1.5177802287686798e-05, "loss": 0.6014, "step": 9739 }, { "epoch": 0.3470082120526569, "grad_norm": 1.4940785127573977, "learning_rate": 1.5176815044601897e-05, "loss": 0.6269, "step": 9740 }, { "epoch": 0.347043839179151, "grad_norm": 1.88613746384881, "learning_rate": 1.5175827732584082e-05, "loss": 0.5533, "step": 9741 }, { "epoch": 0.3470794663056451, "grad_norm": 1.7257296348282851, "learning_rate": 1.5174840351646504e-05, "loss": 0.6352, "step": 9742 }, { "epoch": 0.34711509343213925, "grad_norm": 1.3939905599131408, "learning_rate": 1.517385290180231e-05, "loss": 0.5131, "step": 9743 }, { "epoch": 0.34715072055863333, "grad_norm": 1.7208153457725728, "learning_rate": 1.5172865383064653e-05, "loss": 0.6921, "step": 9744 }, { "epoch": 0.34718634768512746, "grad_norm": 2.108626436504735, "learning_rate": 1.517187779544667e-05, "loss": 0.586, "step": 9745 }, { "epoch": 0.3472219748116216, "grad_norm": 1.7230656635097539, "learning_rate": 1.5170890138961523e-05, "loss": 0.611, "step": 9746 }, { "epoch": 0.34725760193811567, "grad_norm": 1.6383091499421625, "learning_rate": 1.5169902413622359e-05, "loss": 0.5626, "step": 9747 }, { "epoch": 0.3472932290646098, "grad_norm": 1.1600838408940972, "learning_rate": 1.516891461944233e-05, "loss": 0.5343, "step": 9748 }, { "epoch": 0.34732885619110393, "grad_norm": 1.0908389968986523, "learning_rate": 1.516792675643459e-05, "loss": 0.4845, "step": 9749 }, { "epoch": 0.347364483317598, "grad_norm": 1.0553055583185713, "learning_rate": 1.5166938824612296e-05, "loss": 0.4463, "step": 9750 }, { "epoch": 0.34740011044409214, "grad_norm": 1.408882389908502, "learning_rate": 1.5165950823988598e-05, "loss": 0.5713, "step": 9751 }, { "epoch": 0.34743573757058627, "grad_norm": 1.7838711387261055, "learning_rate": 1.5164962754576654e-05, "loss": 0.5077, "step": 9752 }, { "epoch": 0.34747136469708034, "grad_norm": 1.034483273629658, "learning_rate": 1.5163974616389621e-05, "loss": 0.5004, "step": 9753 }, { "epoch": 0.3475069918235745, "grad_norm": 1.7648532129510242, "learning_rate": 1.5162986409440657e-05, "loss": 0.5924, "step": 9754 }, { "epoch": 0.3475426189500686, "grad_norm": 1.1328511073689533, "learning_rate": 1.5161998133742919e-05, "loss": 0.5463, "step": 9755 }, { "epoch": 0.3475782460765627, "grad_norm": 1.0630310041780378, "learning_rate": 1.5161009789309569e-05, "loss": 0.4267, "step": 9756 }, { "epoch": 0.3476138732030568, "grad_norm": 1.5205918228261794, "learning_rate": 1.5160021376153765e-05, "loss": 0.5454, "step": 9757 }, { "epoch": 0.34764950032955094, "grad_norm": 1.1007917148162156, "learning_rate": 1.5159032894288672e-05, "loss": 0.4867, "step": 9758 }, { "epoch": 0.347685127456045, "grad_norm": 1.5948840500233856, "learning_rate": 1.5158044343727447e-05, "loss": 0.604, "step": 9759 }, { "epoch": 0.34772075458253915, "grad_norm": 1.2276357142311334, "learning_rate": 1.5157055724483258e-05, "loss": 0.5284, "step": 9760 }, { "epoch": 0.3477563817090333, "grad_norm": 1.671694225544065, "learning_rate": 1.5156067036569267e-05, "loss": 0.6014, "step": 9761 }, { "epoch": 0.34779200883552736, "grad_norm": 1.2477444292088782, "learning_rate": 1.5155078279998643e-05, "loss": 0.479, "step": 9762 }, { "epoch": 0.3478276359620215, "grad_norm": 1.0865433362968038, "learning_rate": 1.5154089454784547e-05, "loss": 0.4669, "step": 9763 }, { "epoch": 0.3478632630885156, "grad_norm": 1.3440542789279968, "learning_rate": 1.5153100560940143e-05, "loss": 0.6014, "step": 9764 }, { "epoch": 0.3478988902150097, "grad_norm": 1.216886180022315, "learning_rate": 1.5152111598478607e-05, "loss": 0.5173, "step": 9765 }, { "epoch": 0.3479345173415038, "grad_norm": 1.1458862100063736, "learning_rate": 1.5151122567413104e-05, "loss": 0.5026, "step": 9766 }, { "epoch": 0.34797014446799795, "grad_norm": 1.7174125521764942, "learning_rate": 1.51501334677568e-05, "loss": 0.6264, "step": 9767 }, { "epoch": 0.34800577159449203, "grad_norm": 1.5632650845240403, "learning_rate": 1.5149144299522874e-05, "loss": 0.6268, "step": 9768 }, { "epoch": 0.34804139872098616, "grad_norm": 1.7321768978537548, "learning_rate": 1.514815506272449e-05, "loss": 0.6153, "step": 9769 }, { "epoch": 0.3480770258474803, "grad_norm": 1.1732756573752734, "learning_rate": 1.5147165757374822e-05, "loss": 0.5041, "step": 9770 }, { "epoch": 0.34811265297397437, "grad_norm": 1.0900476916346877, "learning_rate": 1.5146176383487045e-05, "loss": 0.4605, "step": 9771 }, { "epoch": 0.3481482801004685, "grad_norm": 1.841626796719257, "learning_rate": 1.5145186941074335e-05, "loss": 0.5977, "step": 9772 }, { "epoch": 0.34818390722696263, "grad_norm": 1.6600291133595908, "learning_rate": 1.5144197430149863e-05, "loss": 0.5599, "step": 9773 }, { "epoch": 0.3482195343534567, "grad_norm": 1.3744599788414342, "learning_rate": 1.5143207850726809e-05, "loss": 0.59, "step": 9774 }, { "epoch": 0.34825516147995084, "grad_norm": 1.2120603994129786, "learning_rate": 1.5142218202818342e-05, "loss": 0.5315, "step": 9775 }, { "epoch": 0.34829078860644497, "grad_norm": 1.5748356601911597, "learning_rate": 1.5141228486437649e-05, "loss": 0.4671, "step": 9776 }, { "epoch": 0.34832641573293904, "grad_norm": 1.1384462206669952, "learning_rate": 1.5140238701597905e-05, "loss": 0.4821, "step": 9777 }, { "epoch": 0.3483620428594332, "grad_norm": 1.129565222136238, "learning_rate": 1.513924884831229e-05, "loss": 0.5311, "step": 9778 }, { "epoch": 0.3483976699859273, "grad_norm": 1.142144739820213, "learning_rate": 1.5138258926593983e-05, "loss": 0.5701, "step": 9779 }, { "epoch": 0.3484332971124214, "grad_norm": 1.068807091788522, "learning_rate": 1.513726893645617e-05, "loss": 0.4781, "step": 9780 }, { "epoch": 0.3484689242389155, "grad_norm": 1.048582827978288, "learning_rate": 1.5136278877912029e-05, "loss": 0.4577, "step": 9781 }, { "epoch": 0.34850455136540964, "grad_norm": 1.1036019094969232, "learning_rate": 1.5135288750974742e-05, "loss": 0.4791, "step": 9782 }, { "epoch": 0.3485401784919037, "grad_norm": 1.1684470197127057, "learning_rate": 1.51342985556575e-05, "loss": 0.51, "step": 9783 }, { "epoch": 0.34857580561839785, "grad_norm": 1.0811741570024431, "learning_rate": 1.5133308291973482e-05, "loss": 0.5018, "step": 9784 }, { "epoch": 0.348611432744892, "grad_norm": 1.6150386923308513, "learning_rate": 1.5132317959935874e-05, "loss": 0.563, "step": 9785 }, { "epoch": 0.34864705987138606, "grad_norm": 1.5886196442205247, "learning_rate": 1.5131327559557867e-05, "loss": 0.5981, "step": 9786 }, { "epoch": 0.3486826869978802, "grad_norm": 1.449230627583627, "learning_rate": 1.513033709085265e-05, "loss": 0.5585, "step": 9787 }, { "epoch": 0.3487183141243743, "grad_norm": 1.1092837895723056, "learning_rate": 1.5129346553833406e-05, "loss": 0.506, "step": 9788 }, { "epoch": 0.3487539412508684, "grad_norm": 1.1624233713507282, "learning_rate": 1.5128355948513326e-05, "loss": 0.5203, "step": 9789 }, { "epoch": 0.3487895683773625, "grad_norm": 1.2038554838404394, "learning_rate": 1.5127365274905603e-05, "loss": 0.4423, "step": 9790 }, { "epoch": 0.34882519550385666, "grad_norm": 1.0058789916330826, "learning_rate": 1.512637453302343e-05, "loss": 0.417, "step": 9791 }, { "epoch": 0.34886082263035073, "grad_norm": 1.5410369940908115, "learning_rate": 1.5125383722879994e-05, "loss": 0.6953, "step": 9792 }, { "epoch": 0.34889644975684486, "grad_norm": 1.0943803354344586, "learning_rate": 1.5124392844488491e-05, "loss": 0.5366, "step": 9793 }, { "epoch": 0.348932076883339, "grad_norm": 1.0100603691820624, "learning_rate": 1.5123401897862118e-05, "loss": 0.46, "step": 9794 }, { "epoch": 0.34896770400983307, "grad_norm": 1.1068882667202036, "learning_rate": 1.5122410883014067e-05, "loss": 0.5426, "step": 9795 }, { "epoch": 0.3490033311363272, "grad_norm": 1.0300394014839938, "learning_rate": 1.5121419799957534e-05, "loss": 0.4765, "step": 9796 }, { "epoch": 0.34903895826282133, "grad_norm": 1.120976523497996, "learning_rate": 1.5120428648705716e-05, "loss": 0.4914, "step": 9797 }, { "epoch": 0.3490745853893154, "grad_norm": 1.1275622314334328, "learning_rate": 1.5119437429271813e-05, "loss": 0.4988, "step": 9798 }, { "epoch": 0.34911021251580954, "grad_norm": 1.2194994990110544, "learning_rate": 1.5118446141669023e-05, "loss": 0.5322, "step": 9799 }, { "epoch": 0.34914583964230367, "grad_norm": 1.7262762573809298, "learning_rate": 1.5117454785910543e-05, "loss": 0.6112, "step": 9800 }, { "epoch": 0.34918146676879774, "grad_norm": 1.0738859458803562, "learning_rate": 1.5116463362009576e-05, "loss": 0.5171, "step": 9801 }, { "epoch": 0.3492170938952919, "grad_norm": 1.3605081996360313, "learning_rate": 1.5115471869979326e-05, "loss": 0.5208, "step": 9802 }, { "epoch": 0.349252721021786, "grad_norm": 1.5170422143863678, "learning_rate": 1.5114480309832991e-05, "loss": 0.6334, "step": 9803 }, { "epoch": 0.3492883481482801, "grad_norm": 1.6319459799293328, "learning_rate": 1.5113488681583773e-05, "loss": 0.5494, "step": 9804 }, { "epoch": 0.3493239752747742, "grad_norm": 1.504074568586315, "learning_rate": 1.511249698524488e-05, "loss": 0.626, "step": 9805 }, { "epoch": 0.34935960240126834, "grad_norm": 1.8851022589208948, "learning_rate": 1.5111505220829519e-05, "loss": 0.5384, "step": 9806 }, { "epoch": 0.3493952295277624, "grad_norm": 1.4808570274760415, "learning_rate": 1.5110513388350892e-05, "loss": 0.5687, "step": 9807 }, { "epoch": 0.34943085665425655, "grad_norm": 1.357778965034223, "learning_rate": 1.5109521487822208e-05, "loss": 0.5834, "step": 9808 }, { "epoch": 0.3494664837807507, "grad_norm": 1.0306527293123313, "learning_rate": 1.5108529519256672e-05, "loss": 0.5091, "step": 9809 }, { "epoch": 0.34950211090724476, "grad_norm": 1.7573962661458695, "learning_rate": 1.5107537482667497e-05, "loss": 0.6431, "step": 9810 }, { "epoch": 0.3495377380337389, "grad_norm": 0.9912401793796761, "learning_rate": 1.5106545378067887e-05, "loss": 0.4105, "step": 9811 }, { "epoch": 0.349573365160233, "grad_norm": 1.589569356875021, "learning_rate": 1.510555320547106e-05, "loss": 0.5238, "step": 9812 }, { "epoch": 0.3496089922867271, "grad_norm": 2.4952252830736636, "learning_rate": 1.5104560964890223e-05, "loss": 0.5231, "step": 9813 }, { "epoch": 0.3496446194132212, "grad_norm": 1.31359331494044, "learning_rate": 1.5103568656338588e-05, "loss": 0.6372, "step": 9814 }, { "epoch": 0.34968024653971536, "grad_norm": 1.0230691176356688, "learning_rate": 1.5102576279829368e-05, "loss": 0.4671, "step": 9815 }, { "epoch": 0.34971587366620943, "grad_norm": 1.6359295382537569, "learning_rate": 1.510158383537578e-05, "loss": 0.6012, "step": 9816 }, { "epoch": 0.34975150079270356, "grad_norm": 1.5994069441963203, "learning_rate": 1.5100591322991033e-05, "loss": 0.6167, "step": 9817 }, { "epoch": 0.3497871279191977, "grad_norm": 1.0329339129886337, "learning_rate": 1.509959874268835e-05, "loss": 0.5044, "step": 9818 }, { "epoch": 0.34982275504569177, "grad_norm": 1.1937193673332664, "learning_rate": 1.509860609448095e-05, "loss": 0.5472, "step": 9819 }, { "epoch": 0.3498583821721859, "grad_norm": 1.2066191344858044, "learning_rate": 1.5097613378382042e-05, "loss": 0.5285, "step": 9820 }, { "epoch": 0.34989400929868003, "grad_norm": 1.0552666047953263, "learning_rate": 1.5096620594404847e-05, "loss": 0.4509, "step": 9821 }, { "epoch": 0.3499296364251741, "grad_norm": 1.3971173562704216, "learning_rate": 1.509562774256259e-05, "loss": 0.6089, "step": 9822 }, { "epoch": 0.34996526355166824, "grad_norm": 1.257192186190012, "learning_rate": 1.5094634822868486e-05, "loss": 0.5321, "step": 9823 }, { "epoch": 0.35000089067816237, "grad_norm": 1.2059572815353448, "learning_rate": 1.5093641835335762e-05, "loss": 0.504, "step": 9824 }, { "epoch": 0.35003651780465644, "grad_norm": 1.1935254526651713, "learning_rate": 1.5092648779977635e-05, "loss": 0.5212, "step": 9825 }, { "epoch": 0.3500721449311506, "grad_norm": 1.7923853684403959, "learning_rate": 1.509165565680733e-05, "loss": 0.5364, "step": 9826 }, { "epoch": 0.3501077720576447, "grad_norm": 1.3921029772821496, "learning_rate": 1.5090662465838067e-05, "loss": 0.4992, "step": 9827 }, { "epoch": 0.3501433991841388, "grad_norm": 1.3621943770836693, "learning_rate": 1.5089669207083075e-05, "loss": 0.4875, "step": 9828 }, { "epoch": 0.3501790263106329, "grad_norm": 1.2806351192258643, "learning_rate": 1.5088675880555586e-05, "loss": 0.4687, "step": 9829 }, { "epoch": 0.35021465343712704, "grad_norm": 1.7646682450604099, "learning_rate": 1.5087682486268818e-05, "loss": 0.5794, "step": 9830 }, { "epoch": 0.3502502805636211, "grad_norm": 1.2403872134743799, "learning_rate": 1.5086689024236e-05, "loss": 0.4697, "step": 9831 }, { "epoch": 0.35028590769011525, "grad_norm": 1.7779382042363179, "learning_rate": 1.5085695494470365e-05, "loss": 0.6361, "step": 9832 }, { "epoch": 0.3503215348166094, "grad_norm": 1.3944322299929088, "learning_rate": 1.5084701896985139e-05, "loss": 0.6258, "step": 9833 }, { "epoch": 0.35035716194310346, "grad_norm": 1.492709664105874, "learning_rate": 1.5083708231793552e-05, "loss": 0.4983, "step": 9834 }, { "epoch": 0.3503927890695976, "grad_norm": 1.3166601307074748, "learning_rate": 1.5082714498908839e-05, "loss": 0.459, "step": 9835 }, { "epoch": 0.3504284161960917, "grad_norm": 1.6382291000029412, "learning_rate": 1.508172069834423e-05, "loss": 0.4386, "step": 9836 }, { "epoch": 0.3504640433225858, "grad_norm": 1.64146285928825, "learning_rate": 1.5080726830112955e-05, "loss": 0.714, "step": 9837 }, { "epoch": 0.3504996704490799, "grad_norm": 1.6418115238138005, "learning_rate": 1.5079732894228252e-05, "loss": 0.5101, "step": 9838 }, { "epoch": 0.35053529757557406, "grad_norm": 1.999667837001182, "learning_rate": 1.5078738890703355e-05, "loss": 0.56, "step": 9839 }, { "epoch": 0.35057092470206813, "grad_norm": 1.3254162931322755, "learning_rate": 1.5077744819551499e-05, "loss": 0.603, "step": 9840 }, { "epoch": 0.35060655182856226, "grad_norm": 2.049502125801963, "learning_rate": 1.5076750680785924e-05, "loss": 0.6556, "step": 9841 }, { "epoch": 0.3506421789550564, "grad_norm": 1.6622175047306416, "learning_rate": 1.5075756474419862e-05, "loss": 0.4992, "step": 9842 }, { "epoch": 0.35067780608155047, "grad_norm": 1.377585794335341, "learning_rate": 1.5074762200466557e-05, "loss": 0.5148, "step": 9843 }, { "epoch": 0.3507134332080446, "grad_norm": 2.0580341375766227, "learning_rate": 1.5073767858939248e-05, "loss": 0.5169, "step": 9844 }, { "epoch": 0.35074906033453873, "grad_norm": 1.155041744512091, "learning_rate": 1.5072773449851172e-05, "loss": 0.4966, "step": 9845 }, { "epoch": 0.3507846874610328, "grad_norm": 1.800510795303508, "learning_rate": 1.5071778973215573e-05, "loss": 0.5379, "step": 9846 }, { "epoch": 0.35082031458752694, "grad_norm": 2.2671360659824993, "learning_rate": 1.5070784429045689e-05, "loss": 0.578, "step": 9847 }, { "epoch": 0.35085594171402107, "grad_norm": 1.2159540805614644, "learning_rate": 1.5069789817354767e-05, "loss": 0.5059, "step": 9848 }, { "epoch": 0.35089156884051514, "grad_norm": 1.268175280158004, "learning_rate": 1.506879513815605e-05, "loss": 0.4916, "step": 9849 }, { "epoch": 0.3509271959670093, "grad_norm": 1.186520506692616, "learning_rate": 1.5067800391462783e-05, "loss": 0.437, "step": 9850 }, { "epoch": 0.3509628230935034, "grad_norm": 1.2631779747564473, "learning_rate": 1.5066805577288212e-05, "loss": 0.4735, "step": 9851 }, { "epoch": 0.3509984502199975, "grad_norm": 1.9772749964772258, "learning_rate": 1.5065810695645584e-05, "loss": 0.6091, "step": 9852 }, { "epoch": 0.3510340773464916, "grad_norm": 1.4869944214493187, "learning_rate": 1.5064815746548144e-05, "loss": 0.4621, "step": 9853 }, { "epoch": 0.35106970447298574, "grad_norm": 1.733920447108093, "learning_rate": 1.5063820730009145e-05, "loss": 0.5738, "step": 9854 }, { "epoch": 0.3511053315994798, "grad_norm": 1.318948810183095, "learning_rate": 1.5062825646041832e-05, "loss": 0.4529, "step": 9855 }, { "epoch": 0.35114095872597395, "grad_norm": 1.4047447823151007, "learning_rate": 1.5061830494659458e-05, "loss": 0.4928, "step": 9856 }, { "epoch": 0.3511765858524681, "grad_norm": 2.8913804354871395, "learning_rate": 1.5060835275875269e-05, "loss": 0.6093, "step": 9857 }, { "epoch": 0.35121221297896216, "grad_norm": 1.2637958037891432, "learning_rate": 1.5059839989702524e-05, "loss": 0.4883, "step": 9858 }, { "epoch": 0.3512478401054563, "grad_norm": 1.2485453340075683, "learning_rate": 1.5058844636154475e-05, "loss": 0.4766, "step": 9859 }, { "epoch": 0.3512834672319504, "grad_norm": 1.5779975842119618, "learning_rate": 1.5057849215244371e-05, "loss": 0.5829, "step": 9860 }, { "epoch": 0.3513190943584445, "grad_norm": 1.2224572909303641, "learning_rate": 1.505685372698547e-05, "loss": 0.5218, "step": 9861 }, { "epoch": 0.3513547214849386, "grad_norm": 1.2426650995986084, "learning_rate": 1.5055858171391026e-05, "loss": 0.5006, "step": 9862 }, { "epoch": 0.35139034861143276, "grad_norm": 1.4063999079081626, "learning_rate": 1.5054862548474298e-05, "loss": 0.554, "step": 9863 }, { "epoch": 0.35142597573792683, "grad_norm": 1.3904503296302841, "learning_rate": 1.5053866858248543e-05, "loss": 0.5879, "step": 9864 }, { "epoch": 0.35146160286442096, "grad_norm": 1.6386952467239646, "learning_rate": 1.5052871100727014e-05, "loss": 0.6276, "step": 9865 }, { "epoch": 0.3514972299909151, "grad_norm": 1.1523852173357076, "learning_rate": 1.5051875275922979e-05, "loss": 0.4599, "step": 9866 }, { "epoch": 0.35153285711740917, "grad_norm": 1.0738311677394419, "learning_rate": 1.5050879383849693e-05, "loss": 0.444, "step": 9867 }, { "epoch": 0.3515684842439033, "grad_norm": 1.115902803418978, "learning_rate": 1.5049883424520416e-05, "loss": 0.5141, "step": 9868 }, { "epoch": 0.35160411137039743, "grad_norm": 1.5099003516182121, "learning_rate": 1.5048887397948413e-05, "loss": 0.5478, "step": 9869 }, { "epoch": 0.3516397384968915, "grad_norm": 1.5761162122568062, "learning_rate": 1.5047891304146945e-05, "loss": 0.5354, "step": 9870 }, { "epoch": 0.35167536562338564, "grad_norm": 1.5980998197539595, "learning_rate": 1.5046895143129277e-05, "loss": 0.5535, "step": 9871 }, { "epoch": 0.35171099274987977, "grad_norm": 1.4489455460885314, "learning_rate": 1.504589891490867e-05, "loss": 0.5775, "step": 9872 }, { "epoch": 0.35174661987637384, "grad_norm": 1.4922520038565497, "learning_rate": 1.5044902619498394e-05, "loss": 0.6026, "step": 9873 }, { "epoch": 0.351782247002868, "grad_norm": 1.7727777648678122, "learning_rate": 1.5043906256911711e-05, "loss": 0.5758, "step": 9874 }, { "epoch": 0.3518178741293621, "grad_norm": 3.1329896795566454, "learning_rate": 1.5042909827161894e-05, "loss": 0.6313, "step": 9875 }, { "epoch": 0.3518535012558562, "grad_norm": 1.446317724068764, "learning_rate": 1.5041913330262205e-05, "loss": 0.5498, "step": 9876 }, { "epoch": 0.3518891283823503, "grad_norm": 1.1691541772294043, "learning_rate": 1.5040916766225917e-05, "loss": 0.498, "step": 9877 }, { "epoch": 0.35192475550884444, "grad_norm": 1.787556155524139, "learning_rate": 1.50399201350663e-05, "loss": 0.5381, "step": 9878 }, { "epoch": 0.3519603826353385, "grad_norm": 1.1541572899724613, "learning_rate": 1.5038923436796623e-05, "loss": 0.4728, "step": 9879 }, { "epoch": 0.35199600976183265, "grad_norm": 1.2098864613860363, "learning_rate": 1.503792667143016e-05, "loss": 0.5351, "step": 9880 }, { "epoch": 0.3520316368883268, "grad_norm": 1.635539512296266, "learning_rate": 1.5036929838980179e-05, "loss": 0.5948, "step": 9881 }, { "epoch": 0.35206726401482086, "grad_norm": 1.5862685284464186, "learning_rate": 1.5035932939459961e-05, "loss": 0.6452, "step": 9882 }, { "epoch": 0.352102891141315, "grad_norm": 0.9881803294291412, "learning_rate": 1.503493597288277e-05, "loss": 0.4898, "step": 9883 }, { "epoch": 0.3521385182678091, "grad_norm": 1.2634252107317452, "learning_rate": 1.5033938939261892e-05, "loss": 0.4722, "step": 9884 }, { "epoch": 0.3521741453943032, "grad_norm": 1.5495274599487794, "learning_rate": 1.5032941838610597e-05, "loss": 0.5799, "step": 9885 }, { "epoch": 0.3522097725207973, "grad_norm": 1.2954166086262433, "learning_rate": 1.503194467094216e-05, "loss": 0.5728, "step": 9886 }, { "epoch": 0.35224539964729146, "grad_norm": 1.0941923302068042, "learning_rate": 1.5030947436269866e-05, "loss": 0.5229, "step": 9887 }, { "epoch": 0.35228102677378553, "grad_norm": 1.1722773270576088, "learning_rate": 1.5029950134606991e-05, "loss": 0.3946, "step": 9888 }, { "epoch": 0.35231665390027966, "grad_norm": 1.437571698734976, "learning_rate": 1.5028952765966815e-05, "loss": 0.5968, "step": 9889 }, { "epoch": 0.3523522810267738, "grad_norm": 1.21911956155894, "learning_rate": 1.5027955330362619e-05, "loss": 0.4893, "step": 9890 }, { "epoch": 0.3523879081532679, "grad_norm": 1.134188701545947, "learning_rate": 1.502695782780768e-05, "loss": 0.4714, "step": 9891 }, { "epoch": 0.352423535279762, "grad_norm": 1.125579585473011, "learning_rate": 1.5025960258315285e-05, "loss": 0.4383, "step": 9892 }, { "epoch": 0.35245916240625613, "grad_norm": 2.201661280328617, "learning_rate": 1.5024962621898715e-05, "loss": 0.5295, "step": 9893 }, { "epoch": 0.35249478953275026, "grad_norm": 1.0800251051868133, "learning_rate": 1.5023964918571256e-05, "loss": 0.4992, "step": 9894 }, { "epoch": 0.35253041665924434, "grad_norm": 1.0853877221280843, "learning_rate": 1.5022967148346194e-05, "loss": 0.4736, "step": 9895 }, { "epoch": 0.35256604378573847, "grad_norm": 1.649744249905101, "learning_rate": 1.5021969311236815e-05, "loss": 0.586, "step": 9896 }, { "epoch": 0.3526016709122326, "grad_norm": 1.0031441676118715, "learning_rate": 1.50209714072564e-05, "loss": 0.4557, "step": 9897 }, { "epoch": 0.3526372980387267, "grad_norm": 1.1647716214103958, "learning_rate": 1.5019973436418242e-05, "loss": 0.5177, "step": 9898 }, { "epoch": 0.3526729251652208, "grad_norm": 1.1491366892924009, "learning_rate": 1.5018975398735632e-05, "loss": 0.4444, "step": 9899 }, { "epoch": 0.35270855229171494, "grad_norm": 1.2138736884867654, "learning_rate": 1.5017977294221854e-05, "loss": 0.4481, "step": 9900 }, { "epoch": 0.352744179418209, "grad_norm": 1.085702851234837, "learning_rate": 1.5016979122890203e-05, "loss": 0.5421, "step": 9901 }, { "epoch": 0.35277980654470315, "grad_norm": 1.0443806549117525, "learning_rate": 1.5015980884753967e-05, "loss": 0.4646, "step": 9902 }, { "epoch": 0.3528154336711973, "grad_norm": 1.3406735455231802, "learning_rate": 1.501498257982644e-05, "loss": 0.487, "step": 9903 }, { "epoch": 0.35285106079769135, "grad_norm": 1.5458331791162607, "learning_rate": 1.5013984208120915e-05, "loss": 0.5667, "step": 9904 }, { "epoch": 0.3528866879241855, "grad_norm": 1.3249080349620947, "learning_rate": 1.5012985769650686e-05, "loss": 0.5563, "step": 9905 }, { "epoch": 0.3529223150506796, "grad_norm": 1.4978708378948715, "learning_rate": 1.5011987264429047e-05, "loss": 0.6561, "step": 9906 }, { "epoch": 0.3529579421771737, "grad_norm": 1.0842780330245974, "learning_rate": 1.5010988692469295e-05, "loss": 0.5011, "step": 9907 }, { "epoch": 0.3529935693036678, "grad_norm": 1.4275004578607902, "learning_rate": 1.5009990053784725e-05, "loss": 0.5601, "step": 9908 }, { "epoch": 0.35302919643016195, "grad_norm": 1.9899688581936439, "learning_rate": 1.5008991348388632e-05, "loss": 0.5865, "step": 9909 }, { "epoch": 0.353064823556656, "grad_norm": 1.3821978335880916, "learning_rate": 1.5007992576294323e-05, "loss": 0.6565, "step": 9910 }, { "epoch": 0.35310045068315016, "grad_norm": 1.276982096830558, "learning_rate": 1.5006993737515092e-05, "loss": 0.4915, "step": 9911 }, { "epoch": 0.3531360778096443, "grad_norm": 1.3120716915028166, "learning_rate": 1.500599483206424e-05, "loss": 0.653, "step": 9912 }, { "epoch": 0.35317170493613836, "grad_norm": 1.4632960859450592, "learning_rate": 1.5004995859955064e-05, "loss": 0.4784, "step": 9913 }, { "epoch": 0.3532073320626325, "grad_norm": 1.163347998855376, "learning_rate": 1.5003996821200875e-05, "loss": 0.5373, "step": 9914 }, { "epoch": 0.3532429591891266, "grad_norm": 2.4559467701048385, "learning_rate": 1.500299771581497e-05, "loss": 0.4765, "step": 9915 }, { "epoch": 0.3532785863156207, "grad_norm": 1.5651939105971227, "learning_rate": 1.5001998543810654e-05, "loss": 0.539, "step": 9916 }, { "epoch": 0.35331421344211483, "grad_norm": 1.4257753791465657, "learning_rate": 1.5000999305201227e-05, "loss": 0.5879, "step": 9917 }, { "epoch": 0.35334984056860896, "grad_norm": 1.3057322809527727, "learning_rate": 1.5000000000000002e-05, "loss": 0.4881, "step": 9918 }, { "epoch": 0.35338546769510304, "grad_norm": 1.473978557597386, "learning_rate": 1.4999000628220281e-05, "loss": 0.4981, "step": 9919 }, { "epoch": 0.35342109482159717, "grad_norm": 1.2241675255915152, "learning_rate": 1.4998001189875373e-05, "loss": 0.5141, "step": 9920 }, { "epoch": 0.3534567219480913, "grad_norm": 1.2036774617717565, "learning_rate": 1.4997001684978585e-05, "loss": 0.4713, "step": 9921 }, { "epoch": 0.3534923490745854, "grad_norm": 1.2905914621133756, "learning_rate": 1.4996002113543227e-05, "loss": 0.5091, "step": 9922 }, { "epoch": 0.3535279762010795, "grad_norm": 1.2750292915306713, "learning_rate": 1.499500247558261e-05, "loss": 0.5122, "step": 9923 }, { "epoch": 0.35356360332757364, "grad_norm": 1.2089592949923806, "learning_rate": 1.4994002771110043e-05, "loss": 0.5261, "step": 9924 }, { "epoch": 0.3535992304540677, "grad_norm": 1.1661553679694423, "learning_rate": 1.4993003000138839e-05, "loss": 0.4882, "step": 9925 }, { "epoch": 0.35363485758056185, "grad_norm": 1.8201804553217948, "learning_rate": 1.499200316268231e-05, "loss": 0.5145, "step": 9926 }, { "epoch": 0.353670484707056, "grad_norm": 1.148946975343555, "learning_rate": 1.4991003258753772e-05, "loss": 0.5267, "step": 9927 }, { "epoch": 0.35370611183355005, "grad_norm": 1.1897391678515123, "learning_rate": 1.4990003288366533e-05, "loss": 0.5694, "step": 9928 }, { "epoch": 0.3537417389600442, "grad_norm": 1.7904638346043749, "learning_rate": 1.4989003251533913e-05, "loss": 0.5349, "step": 9929 }, { "epoch": 0.3537773660865383, "grad_norm": 1.5885191388564375, "learning_rate": 1.498800314826923e-05, "loss": 0.5751, "step": 9930 }, { "epoch": 0.3538129932130324, "grad_norm": 1.8461985573307895, "learning_rate": 1.4987002978585796e-05, "loss": 0.6521, "step": 9931 }, { "epoch": 0.3538486203395265, "grad_norm": 1.521165137734397, "learning_rate": 1.4986002742496932e-05, "loss": 0.4637, "step": 9932 }, { "epoch": 0.35388424746602065, "grad_norm": 1.5587192753306656, "learning_rate": 1.4985002440015959e-05, "loss": 0.6076, "step": 9933 }, { "epoch": 0.3539198745925147, "grad_norm": 1.4650558455939504, "learning_rate": 1.4984002071156192e-05, "loss": 0.5784, "step": 9934 }, { "epoch": 0.35395550171900886, "grad_norm": 1.4219443446905902, "learning_rate": 1.4983001635930954e-05, "loss": 0.5246, "step": 9935 }, { "epoch": 0.353991128845503, "grad_norm": 1.3246233502368538, "learning_rate": 1.4982001134353568e-05, "loss": 0.4537, "step": 9936 }, { "epoch": 0.35402675597199706, "grad_norm": 1.4125100561336357, "learning_rate": 1.4981000566437354e-05, "loss": 0.5279, "step": 9937 }, { "epoch": 0.3540623830984912, "grad_norm": 1.1799476489142215, "learning_rate": 1.4979999932195635e-05, "loss": 0.4843, "step": 9938 }, { "epoch": 0.3540980102249853, "grad_norm": 1.483216872853517, "learning_rate": 1.4978999231641736e-05, "loss": 0.4644, "step": 9939 }, { "epoch": 0.3541336373514794, "grad_norm": 1.2287685022824715, "learning_rate": 1.4977998464788983e-05, "loss": 0.4761, "step": 9940 }, { "epoch": 0.35416926447797353, "grad_norm": 1.9406521176053828, "learning_rate": 1.4976997631650701e-05, "loss": 0.4957, "step": 9941 }, { "epoch": 0.35420489160446766, "grad_norm": 1.6538674114498655, "learning_rate": 1.4975996732240214e-05, "loss": 0.6234, "step": 9942 }, { "epoch": 0.35424051873096174, "grad_norm": 1.5931709759490356, "learning_rate": 1.4974995766570856e-05, "loss": 0.45, "step": 9943 }, { "epoch": 0.35427614585745587, "grad_norm": 1.4844894720221953, "learning_rate": 1.497399473465595e-05, "loss": 0.5569, "step": 9944 }, { "epoch": 0.35431177298395, "grad_norm": 1.210883988588662, "learning_rate": 1.497299363650883e-05, "loss": 0.4706, "step": 9945 }, { "epoch": 0.3543474001104441, "grad_norm": 1.1870313751910293, "learning_rate": 1.4971992472142822e-05, "loss": 0.4575, "step": 9946 }, { "epoch": 0.3543830272369382, "grad_norm": 1.973922527204299, "learning_rate": 1.497099124157126e-05, "loss": 0.6061, "step": 9947 }, { "epoch": 0.35441865436343234, "grad_norm": 1.645666461287765, "learning_rate": 1.4969989944807472e-05, "loss": 0.6125, "step": 9948 }, { "epoch": 0.3544542814899264, "grad_norm": 2.0614017130272777, "learning_rate": 1.4968988581864798e-05, "loss": 0.6126, "step": 9949 }, { "epoch": 0.35448990861642055, "grad_norm": 1.5049153812593654, "learning_rate": 1.4967987152756567e-05, "loss": 0.528, "step": 9950 }, { "epoch": 0.3545255357429147, "grad_norm": 1.1025383853909836, "learning_rate": 1.4966985657496113e-05, "loss": 0.4276, "step": 9951 }, { "epoch": 0.35456116286940875, "grad_norm": 1.0810085455060192, "learning_rate": 1.4965984096096775e-05, "loss": 0.5464, "step": 9952 }, { "epoch": 0.3545967899959029, "grad_norm": 1.6736404669273197, "learning_rate": 1.496498246857189e-05, "loss": 0.6268, "step": 9953 }, { "epoch": 0.354632417122397, "grad_norm": 1.3884602781437072, "learning_rate": 1.4963980774934788e-05, "loss": 0.5786, "step": 9954 }, { "epoch": 0.3546680442488911, "grad_norm": 1.5963138925969242, "learning_rate": 1.4962979015198813e-05, "loss": 0.5426, "step": 9955 }, { "epoch": 0.3547036713753852, "grad_norm": 1.7957144420623152, "learning_rate": 1.4961977189377306e-05, "loss": 0.5864, "step": 9956 }, { "epoch": 0.35473929850187935, "grad_norm": 1.4266642660706597, "learning_rate": 1.4960975297483605e-05, "loss": 0.5475, "step": 9957 }, { "epoch": 0.3547749256283734, "grad_norm": 1.769463246947649, "learning_rate": 1.4959973339531046e-05, "loss": 0.5844, "step": 9958 }, { "epoch": 0.35481055275486756, "grad_norm": 1.8318936756851338, "learning_rate": 1.4958971315532979e-05, "loss": 0.6184, "step": 9959 }, { "epoch": 0.3548461798813617, "grad_norm": 1.2504225247005691, "learning_rate": 1.4957969225502742e-05, "loss": 0.4964, "step": 9960 }, { "epoch": 0.35488180700785577, "grad_norm": 1.466433840087269, "learning_rate": 1.4956967069453678e-05, "loss": 0.5461, "step": 9961 }, { "epoch": 0.3549174341343499, "grad_norm": 1.2997907216259523, "learning_rate": 1.4955964847399135e-05, "loss": 0.4627, "step": 9962 }, { "epoch": 0.354953061260844, "grad_norm": 1.523600478829259, "learning_rate": 1.4954962559352457e-05, "loss": 0.5808, "step": 9963 }, { "epoch": 0.3549886883873381, "grad_norm": 1.2174811745662528, "learning_rate": 1.4953960205326987e-05, "loss": 0.5046, "step": 9964 }, { "epoch": 0.35502431551383223, "grad_norm": 1.5506794770637915, "learning_rate": 1.4952957785336073e-05, "loss": 0.5562, "step": 9965 }, { "epoch": 0.35505994264032636, "grad_norm": 1.5445654316788628, "learning_rate": 1.4951955299393066e-05, "loss": 0.6776, "step": 9966 }, { "epoch": 0.35509556976682044, "grad_norm": 1.78188781510594, "learning_rate": 1.4950952747511311e-05, "loss": 0.532, "step": 9967 }, { "epoch": 0.35513119689331457, "grad_norm": 1.1843295535004115, "learning_rate": 1.4949950129704162e-05, "loss": 0.4668, "step": 9968 }, { "epoch": 0.3551668240198087, "grad_norm": 1.809029153050189, "learning_rate": 1.4948947445984967e-05, "loss": 0.5683, "step": 9969 }, { "epoch": 0.3552024511463028, "grad_norm": 1.0851633675964447, "learning_rate": 1.4947944696367079e-05, "loss": 0.4715, "step": 9970 }, { "epoch": 0.3552380782727969, "grad_norm": 1.4821571998019103, "learning_rate": 1.4946941880863846e-05, "loss": 0.5642, "step": 9971 }, { "epoch": 0.35527370539929104, "grad_norm": 1.1449794321537343, "learning_rate": 1.4945938999488628e-05, "loss": 0.5402, "step": 9972 }, { "epoch": 0.3553093325257851, "grad_norm": 1.1630525542359675, "learning_rate": 1.494493605225477e-05, "loss": 0.4745, "step": 9973 }, { "epoch": 0.35534495965227925, "grad_norm": 1.145372458043885, "learning_rate": 1.4943933039175636e-05, "loss": 0.4568, "step": 9974 }, { "epoch": 0.3553805867787734, "grad_norm": 1.848418820705315, "learning_rate": 1.4942929960264576e-05, "loss": 0.5377, "step": 9975 }, { "epoch": 0.35541621390526745, "grad_norm": 1.1117781718154724, "learning_rate": 1.494192681553495e-05, "loss": 0.4725, "step": 9976 }, { "epoch": 0.3554518410317616, "grad_norm": 1.061775828777854, "learning_rate": 1.4940923605000112e-05, "loss": 0.4805, "step": 9977 }, { "epoch": 0.3554874681582557, "grad_norm": 1.633883172444258, "learning_rate": 1.4939920328673422e-05, "loss": 0.5981, "step": 9978 }, { "epoch": 0.3555230952847498, "grad_norm": 1.08376380745191, "learning_rate": 1.4938916986568246e-05, "loss": 0.5242, "step": 9979 }, { "epoch": 0.3555587224112439, "grad_norm": 1.69686024957702, "learning_rate": 1.4937913578697933e-05, "loss": 0.5035, "step": 9980 }, { "epoch": 0.35559434953773805, "grad_norm": 1.4131764814955958, "learning_rate": 1.493691010507585e-05, "loss": 0.5296, "step": 9981 }, { "epoch": 0.35562997666423213, "grad_norm": 1.7230279953597076, "learning_rate": 1.493590656571536e-05, "loss": 0.6111, "step": 9982 }, { "epoch": 0.35566560379072626, "grad_norm": 1.4369475254802668, "learning_rate": 1.4934902960629825e-05, "loss": 0.5643, "step": 9983 }, { "epoch": 0.3557012309172204, "grad_norm": 1.566579612927797, "learning_rate": 1.4933899289832603e-05, "loss": 0.564, "step": 9984 }, { "epoch": 0.35573685804371447, "grad_norm": 1.0711830897911325, "learning_rate": 1.4932895553337068e-05, "loss": 0.5604, "step": 9985 }, { "epoch": 0.3557724851702086, "grad_norm": 1.0225505856049752, "learning_rate": 1.493189175115658e-05, "loss": 0.5063, "step": 9986 }, { "epoch": 0.3558081122967027, "grad_norm": 1.1325418869270332, "learning_rate": 1.4930887883304503e-05, "loss": 0.4866, "step": 9987 }, { "epoch": 0.3558437394231968, "grad_norm": 1.1072325086928652, "learning_rate": 1.4929883949794213e-05, "loss": 0.4931, "step": 9988 }, { "epoch": 0.35587936654969093, "grad_norm": 1.0571648499287043, "learning_rate": 1.4928879950639068e-05, "loss": 0.4791, "step": 9989 }, { "epoch": 0.35591499367618507, "grad_norm": 2.292009456691323, "learning_rate": 1.4927875885852442e-05, "loss": 0.6081, "step": 9990 }, { "epoch": 0.35595062080267914, "grad_norm": 1.9290591237434003, "learning_rate": 1.4926871755447705e-05, "loss": 0.6008, "step": 9991 }, { "epoch": 0.35598624792917327, "grad_norm": 1.560845297997284, "learning_rate": 1.4925867559438225e-05, "loss": 0.5823, "step": 9992 }, { "epoch": 0.3560218750556674, "grad_norm": 1.3056036148302543, "learning_rate": 1.4924863297837378e-05, "loss": 0.5217, "step": 9993 }, { "epoch": 0.3560575021821615, "grad_norm": 1.687465018059751, "learning_rate": 1.4923858970658534e-05, "loss": 0.5754, "step": 9994 }, { "epoch": 0.3560931293086556, "grad_norm": 1.694003290875995, "learning_rate": 1.4922854577915064e-05, "loss": 0.6027, "step": 9995 }, { "epoch": 0.35612875643514974, "grad_norm": 1.204594753659362, "learning_rate": 1.4921850119620348e-05, "loss": 0.5162, "step": 9996 }, { "epoch": 0.3561643835616438, "grad_norm": 1.6304339486793962, "learning_rate": 1.4920845595787755e-05, "loss": 0.5204, "step": 9997 }, { "epoch": 0.35620001068813795, "grad_norm": 1.5133340450148758, "learning_rate": 1.4919841006430663e-05, "loss": 0.5777, "step": 9998 }, { "epoch": 0.3562356378146321, "grad_norm": 1.1326500411154083, "learning_rate": 1.491883635156245e-05, "loss": 0.4477, "step": 9999 }, { "epoch": 0.35627126494112615, "grad_norm": 1.001359591010271, "learning_rate": 1.4917831631196494e-05, "loss": 0.4326, "step": 10000 }, { "epoch": 0.3563068920676203, "grad_norm": 1.1101158498345205, "learning_rate": 1.4916826845346168e-05, "loss": 0.4918, "step": 10001 }, { "epoch": 0.3563425191941144, "grad_norm": 1.1796178503327799, "learning_rate": 1.4915821994024859e-05, "loss": 0.5309, "step": 10002 }, { "epoch": 0.3563781463206085, "grad_norm": 1.5765593357076684, "learning_rate": 1.4914817077245945e-05, "loss": 0.6021, "step": 10003 }, { "epoch": 0.3564137734471026, "grad_norm": 1.063516770340558, "learning_rate": 1.4913812095022806e-05, "loss": 0.4975, "step": 10004 }, { "epoch": 0.35644940057359675, "grad_norm": 1.4675040733028124, "learning_rate": 1.4912807047368825e-05, "loss": 0.5493, "step": 10005 }, { "epoch": 0.35648502770009083, "grad_norm": 1.151601802025265, "learning_rate": 1.4911801934297381e-05, "loss": 0.564, "step": 10006 }, { "epoch": 0.35652065482658496, "grad_norm": 1.5743613748412146, "learning_rate": 1.4910796755821863e-05, "loss": 0.5831, "step": 10007 }, { "epoch": 0.3565562819530791, "grad_norm": 2.2344397388732733, "learning_rate": 1.4909791511955656e-05, "loss": 0.6455, "step": 10008 }, { "epoch": 0.35659190907957317, "grad_norm": 1.0419395861774026, "learning_rate": 1.4908786202712142e-05, "loss": 0.4857, "step": 10009 }, { "epoch": 0.3566275362060673, "grad_norm": 1.0937936430262039, "learning_rate": 1.4907780828104707e-05, "loss": 0.5039, "step": 10010 }, { "epoch": 0.35666316333256143, "grad_norm": 1.6476818647266323, "learning_rate": 1.4906775388146742e-05, "loss": 0.6407, "step": 10011 }, { "epoch": 0.3566987904590555, "grad_norm": 1.077324801611898, "learning_rate": 1.4905769882851633e-05, "loss": 0.4941, "step": 10012 }, { "epoch": 0.35673441758554963, "grad_norm": 1.0626405311984324, "learning_rate": 1.4904764312232766e-05, "loss": 0.4638, "step": 10013 }, { "epoch": 0.35677004471204377, "grad_norm": 1.4356496420498412, "learning_rate": 1.4903758676303538e-05, "loss": 0.5597, "step": 10014 }, { "epoch": 0.35680567183853784, "grad_norm": 1.1179198478432337, "learning_rate": 1.4902752975077333e-05, "loss": 0.5265, "step": 10015 }, { "epoch": 0.35684129896503197, "grad_norm": 1.5101597123721138, "learning_rate": 1.4901747208567545e-05, "loss": 0.6105, "step": 10016 }, { "epoch": 0.3568769260915261, "grad_norm": 1.7158847515327258, "learning_rate": 1.4900741376787568e-05, "loss": 0.5769, "step": 10017 }, { "epoch": 0.3569125532180202, "grad_norm": 1.4639290807489953, "learning_rate": 1.4899735479750794e-05, "loss": 0.6432, "step": 10018 }, { "epoch": 0.3569481803445143, "grad_norm": 1.0804440336877725, "learning_rate": 1.4898729517470618e-05, "loss": 0.4405, "step": 10019 }, { "epoch": 0.35698380747100844, "grad_norm": 1.2192372545443746, "learning_rate": 1.4897723489960436e-05, "loss": 0.47, "step": 10020 }, { "epoch": 0.3570194345975025, "grad_norm": 1.144562447094412, "learning_rate": 1.489671739723364e-05, "loss": 0.4818, "step": 10021 }, { "epoch": 0.35705506172399665, "grad_norm": 1.631137930451913, "learning_rate": 1.489571123930363e-05, "loss": 0.5433, "step": 10022 }, { "epoch": 0.3570906888504908, "grad_norm": 1.1338362735911545, "learning_rate": 1.4894705016183803e-05, "loss": 0.4671, "step": 10023 }, { "epoch": 0.35712631597698485, "grad_norm": 2.585337374903236, "learning_rate": 1.4893698727887555e-05, "loss": 0.5878, "step": 10024 }, { "epoch": 0.357161943103479, "grad_norm": 1.1579555611610197, "learning_rate": 1.489269237442829e-05, "loss": 0.4969, "step": 10025 }, { "epoch": 0.3571975702299731, "grad_norm": 1.0233068694073277, "learning_rate": 1.4891685955819405e-05, "loss": 0.46, "step": 10026 }, { "epoch": 0.3572331973564672, "grad_norm": 1.154212483925941, "learning_rate": 1.4890679472074308e-05, "loss": 0.5191, "step": 10027 }, { "epoch": 0.3572688244829613, "grad_norm": 1.1154570139008446, "learning_rate": 1.488967292320639e-05, "loss": 0.4708, "step": 10028 }, { "epoch": 0.35730445160945545, "grad_norm": 1.1035598754727398, "learning_rate": 1.4888666309229058e-05, "loss": 0.4604, "step": 10029 }, { "epoch": 0.35734007873594953, "grad_norm": 1.5013726594952852, "learning_rate": 1.4887659630155722e-05, "loss": 0.5512, "step": 10030 }, { "epoch": 0.35737570586244366, "grad_norm": 1.3630566166609372, "learning_rate": 1.488665288599978e-05, "loss": 0.5721, "step": 10031 }, { "epoch": 0.3574113329889378, "grad_norm": 3.0787086206315872, "learning_rate": 1.4885646076774639e-05, "loss": 0.5543, "step": 10032 }, { "epoch": 0.35744696011543187, "grad_norm": 1.1533518423477658, "learning_rate": 1.4884639202493705e-05, "loss": 0.5593, "step": 10033 }, { "epoch": 0.357482587241926, "grad_norm": 1.5724109236411299, "learning_rate": 1.4883632263170387e-05, "loss": 0.571, "step": 10034 }, { "epoch": 0.35751821436842013, "grad_norm": 1.510797658810834, "learning_rate": 1.488262525881809e-05, "loss": 0.6182, "step": 10035 }, { "epoch": 0.3575538414949142, "grad_norm": 2.0061367649869055, "learning_rate": 1.4881618189450224e-05, "loss": 0.5301, "step": 10036 }, { "epoch": 0.35758946862140834, "grad_norm": 1.9632210226186173, "learning_rate": 1.4880611055080202e-05, "loss": 0.5889, "step": 10037 }, { "epoch": 0.35762509574790247, "grad_norm": 1.6292187218501695, "learning_rate": 1.4879603855721434e-05, "loss": 0.4927, "step": 10038 }, { "epoch": 0.35766072287439654, "grad_norm": 1.6100400869800835, "learning_rate": 1.4878596591387329e-05, "loss": 0.661, "step": 10039 }, { "epoch": 0.3576963500008907, "grad_norm": 1.4504693856831066, "learning_rate": 1.4877589262091299e-05, "loss": 0.613, "step": 10040 }, { "epoch": 0.3577319771273848, "grad_norm": 1.4366738965400696, "learning_rate": 1.487658186784676e-05, "loss": 0.5155, "step": 10041 }, { "epoch": 0.3577676042538789, "grad_norm": 1.2500541559790423, "learning_rate": 1.4875574408667125e-05, "loss": 0.5285, "step": 10042 }, { "epoch": 0.357803231380373, "grad_norm": 2.1535969152656604, "learning_rate": 1.4874566884565806e-05, "loss": 0.5836, "step": 10043 }, { "epoch": 0.35783885850686714, "grad_norm": 1.5000191994129588, "learning_rate": 1.4873559295556225e-05, "loss": 0.4727, "step": 10044 }, { "epoch": 0.3578744856333612, "grad_norm": 1.4743028922568806, "learning_rate": 1.4872551641651796e-05, "loss": 0.5097, "step": 10045 }, { "epoch": 0.35791011275985535, "grad_norm": 1.2337671234012775, "learning_rate": 1.4871543922865936e-05, "loss": 0.4448, "step": 10046 }, { "epoch": 0.3579457398863495, "grad_norm": 1.7346119308188932, "learning_rate": 1.4870536139212062e-05, "loss": 0.6143, "step": 10047 }, { "epoch": 0.35798136701284355, "grad_norm": 1.1487341151391839, "learning_rate": 1.4869528290703592e-05, "loss": 0.4595, "step": 10048 }, { "epoch": 0.3580169941393377, "grad_norm": 1.621698391492208, "learning_rate": 1.4868520377353954e-05, "loss": 0.5722, "step": 10049 }, { "epoch": 0.3580526212658318, "grad_norm": 1.6369522957906122, "learning_rate": 1.4867512399176564e-05, "loss": 0.5784, "step": 10050 }, { "epoch": 0.3580882483923259, "grad_norm": 1.0526058835658363, "learning_rate": 1.4866504356184844e-05, "loss": 0.465, "step": 10051 }, { "epoch": 0.35812387551882, "grad_norm": 1.0479183316391614, "learning_rate": 1.4865496248392215e-05, "loss": 0.4483, "step": 10052 }, { "epoch": 0.35815950264531415, "grad_norm": 1.1812125843134802, "learning_rate": 1.4864488075812105e-05, "loss": 0.5098, "step": 10053 }, { "epoch": 0.35819512977180823, "grad_norm": 1.7431379360683588, "learning_rate": 1.4863479838457937e-05, "loss": 0.6199, "step": 10054 }, { "epoch": 0.35823075689830236, "grad_norm": 1.6740432674913668, "learning_rate": 1.486247153634313e-05, "loss": 0.6363, "step": 10055 }, { "epoch": 0.3582663840247965, "grad_norm": 1.1929264569227815, "learning_rate": 1.486146316948112e-05, "loss": 0.4806, "step": 10056 }, { "epoch": 0.35830201115129057, "grad_norm": 1.5703679521823182, "learning_rate": 1.4860454737885332e-05, "loss": 0.5548, "step": 10057 }, { "epoch": 0.3583376382777847, "grad_norm": 1.4155970535744662, "learning_rate": 1.4859446241569188e-05, "loss": 0.4977, "step": 10058 }, { "epoch": 0.35837326540427883, "grad_norm": 1.098946798953436, "learning_rate": 1.485843768054612e-05, "loss": 0.5049, "step": 10059 }, { "epoch": 0.3584088925307729, "grad_norm": 1.5419445099223306, "learning_rate": 1.4857429054829561e-05, "loss": 0.5783, "step": 10060 }, { "epoch": 0.35844451965726704, "grad_norm": 1.5477692384340427, "learning_rate": 1.4856420364432943e-05, "loss": 0.646, "step": 10061 }, { "epoch": 0.35848014678376117, "grad_norm": 1.844231180114217, "learning_rate": 1.4855411609369685e-05, "loss": 0.7092, "step": 10062 }, { "epoch": 0.35851577391025524, "grad_norm": 2.132291676865103, "learning_rate": 1.4854402789653233e-05, "loss": 0.5483, "step": 10063 }, { "epoch": 0.3585514010367494, "grad_norm": 1.5001211984941933, "learning_rate": 1.4853393905297013e-05, "loss": 0.4815, "step": 10064 }, { "epoch": 0.3585870281632435, "grad_norm": 1.130430786043689, "learning_rate": 1.4852384956314463e-05, "loss": 0.5112, "step": 10065 }, { "epoch": 0.3586226552897376, "grad_norm": 2.082866644361547, "learning_rate": 1.4851375942719013e-05, "loss": 0.5904, "step": 10066 }, { "epoch": 0.3586582824162317, "grad_norm": 1.217839733717469, "learning_rate": 1.4850366864524104e-05, "loss": 0.5144, "step": 10067 }, { "epoch": 0.35869390954272584, "grad_norm": 1.4870843287551292, "learning_rate": 1.4849357721743169e-05, "loss": 0.4944, "step": 10068 }, { "epoch": 0.3587295366692199, "grad_norm": 1.53112711294033, "learning_rate": 1.4848348514389646e-05, "loss": 0.6098, "step": 10069 }, { "epoch": 0.35876516379571405, "grad_norm": 1.1582857037724728, "learning_rate": 1.484733924247697e-05, "loss": 0.4661, "step": 10070 }, { "epoch": 0.3588007909222082, "grad_norm": 1.1201346096114106, "learning_rate": 1.4846329906018588e-05, "loss": 0.5477, "step": 10071 }, { "epoch": 0.35883641804870225, "grad_norm": 1.2602866947723674, "learning_rate": 1.4845320505027938e-05, "loss": 0.499, "step": 10072 }, { "epoch": 0.3588720451751964, "grad_norm": 1.0877256155742876, "learning_rate": 1.4844311039518457e-05, "loss": 0.4237, "step": 10073 }, { "epoch": 0.3589076723016905, "grad_norm": 1.0444609068091253, "learning_rate": 1.4843301509503584e-05, "loss": 0.4693, "step": 10074 }, { "epoch": 0.3589432994281846, "grad_norm": 1.100249453092104, "learning_rate": 1.4842291914996771e-05, "loss": 0.4422, "step": 10075 }, { "epoch": 0.3589789265546787, "grad_norm": 1.2827911969157535, "learning_rate": 1.4841282256011458e-05, "loss": 0.5678, "step": 10076 }, { "epoch": 0.35901455368117285, "grad_norm": 1.1105617832801096, "learning_rate": 1.4840272532561084e-05, "loss": 0.5293, "step": 10077 }, { "epoch": 0.35905018080766693, "grad_norm": 1.4269655691997596, "learning_rate": 1.48392627446591e-05, "loss": 0.6461, "step": 10078 }, { "epoch": 0.35908580793416106, "grad_norm": 1.0688877036634234, "learning_rate": 1.4838252892318949e-05, "loss": 0.5034, "step": 10079 }, { "epoch": 0.3591214350606552, "grad_norm": 1.4067515445237369, "learning_rate": 1.4837242975554082e-05, "loss": 0.5669, "step": 10080 }, { "epoch": 0.3591570621871493, "grad_norm": 1.4579113918217173, "learning_rate": 1.483623299437794e-05, "loss": 0.6182, "step": 10081 }, { "epoch": 0.3591926893136434, "grad_norm": 1.1467756832601115, "learning_rate": 1.4835222948803976e-05, "loss": 0.4967, "step": 10082 }, { "epoch": 0.35922831644013753, "grad_norm": 2.069526848076586, "learning_rate": 1.4834212838845639e-05, "loss": 0.5608, "step": 10083 }, { "epoch": 0.35926394356663166, "grad_norm": 1.087690116289131, "learning_rate": 1.4833202664516381e-05, "loss": 0.4603, "step": 10084 }, { "epoch": 0.35929957069312574, "grad_norm": 1.9399470640099186, "learning_rate": 1.483219242582965e-05, "loss": 0.6382, "step": 10085 }, { "epoch": 0.35933519781961987, "grad_norm": 0.9988845014987676, "learning_rate": 1.48311821227989e-05, "loss": 0.4671, "step": 10086 }, { "epoch": 0.359370824946114, "grad_norm": 1.5354036792832573, "learning_rate": 1.483017175543758e-05, "loss": 0.5656, "step": 10087 }, { "epoch": 0.3594064520726081, "grad_norm": 1.0196819070680225, "learning_rate": 1.4829161323759152e-05, "loss": 0.457, "step": 10088 }, { "epoch": 0.3594420791991022, "grad_norm": 0.9627051402830643, "learning_rate": 1.482815082777706e-05, "loss": 0.4874, "step": 10089 }, { "epoch": 0.35947770632559634, "grad_norm": 1.8630035164199528, "learning_rate": 1.4827140267504772e-05, "loss": 0.5848, "step": 10090 }, { "epoch": 0.3595133334520904, "grad_norm": 1.1564595054384268, "learning_rate": 1.4826129642955734e-05, "loss": 0.4716, "step": 10091 }, { "epoch": 0.35954896057858454, "grad_norm": 1.905619045124475, "learning_rate": 1.4825118954143405e-05, "loss": 0.5932, "step": 10092 }, { "epoch": 0.3595845877050787, "grad_norm": 1.1059607254964756, "learning_rate": 1.4824108201081247e-05, "loss": 0.5104, "step": 10093 }, { "epoch": 0.35962021483157275, "grad_norm": 1.6243051346581174, "learning_rate": 1.4823097383782714e-05, "loss": 0.5742, "step": 10094 }, { "epoch": 0.3596558419580669, "grad_norm": 1.2195639923261667, "learning_rate": 1.4822086502261268e-05, "loss": 0.5283, "step": 10095 }, { "epoch": 0.359691469084561, "grad_norm": 1.4218052280061402, "learning_rate": 1.4821075556530371e-05, "loss": 0.4968, "step": 10096 }, { "epoch": 0.3597270962110551, "grad_norm": 1.3006489065710047, "learning_rate": 1.4820064546603484e-05, "loss": 0.4634, "step": 10097 }, { "epoch": 0.3597627233375492, "grad_norm": 1.2514067555237116, "learning_rate": 1.481905347249407e-05, "loss": 0.4846, "step": 10098 }, { "epoch": 0.35979835046404335, "grad_norm": 1.5665939652154004, "learning_rate": 1.4818042334215585e-05, "loss": 0.5692, "step": 10099 }, { "epoch": 0.3598339775905374, "grad_norm": 1.7557769151637865, "learning_rate": 1.4817031131781505e-05, "loss": 0.6252, "step": 10100 }, { "epoch": 0.35986960471703155, "grad_norm": 1.1114388726688882, "learning_rate": 1.4816019865205284e-05, "loss": 0.4547, "step": 10101 }, { "epoch": 0.3599052318435257, "grad_norm": 1.4346527273289738, "learning_rate": 1.4815008534500397e-05, "loss": 0.4913, "step": 10102 }, { "epoch": 0.35994085897001976, "grad_norm": 1.12568592249601, "learning_rate": 1.4813997139680303e-05, "loss": 0.5033, "step": 10103 }, { "epoch": 0.3599764860965139, "grad_norm": 1.3283130953746305, "learning_rate": 1.4812985680758471e-05, "loss": 0.6424, "step": 10104 }, { "epoch": 0.360012113223008, "grad_norm": 1.9019203610242539, "learning_rate": 1.4811974157748372e-05, "loss": 0.4928, "step": 10105 }, { "epoch": 0.3600477403495021, "grad_norm": 1.2023893600391498, "learning_rate": 1.4810962570663475e-05, "loss": 0.5051, "step": 10106 }, { "epoch": 0.36008336747599623, "grad_norm": 1.6740411335818823, "learning_rate": 1.4809950919517245e-05, "loss": 0.5823, "step": 10107 }, { "epoch": 0.36011899460249036, "grad_norm": 1.3568730313523454, "learning_rate": 1.4808939204323158e-05, "loss": 0.5603, "step": 10108 }, { "epoch": 0.36015462172898444, "grad_norm": 1.83931241870977, "learning_rate": 1.4807927425094687e-05, "loss": 0.591, "step": 10109 }, { "epoch": 0.36019024885547857, "grad_norm": 1.400088357565841, "learning_rate": 1.48069155818453e-05, "loss": 0.5816, "step": 10110 }, { "epoch": 0.3602258759819727, "grad_norm": 1.6701449761448062, "learning_rate": 1.4805903674588474e-05, "loss": 0.5641, "step": 10111 }, { "epoch": 0.3602615031084668, "grad_norm": 1.1202653081955292, "learning_rate": 1.4804891703337679e-05, "loss": 0.497, "step": 10112 }, { "epoch": 0.3602971302349609, "grad_norm": 1.5981628938135255, "learning_rate": 1.4803879668106393e-05, "loss": 0.6821, "step": 10113 }, { "epoch": 0.36033275736145504, "grad_norm": 1.2641181244645787, "learning_rate": 1.480286756890809e-05, "loss": 0.5247, "step": 10114 }, { "epoch": 0.3603683844879491, "grad_norm": 1.1272073794835444, "learning_rate": 1.4801855405756251e-05, "loss": 0.4687, "step": 10115 }, { "epoch": 0.36040401161444324, "grad_norm": 1.3957946725469417, "learning_rate": 1.480084317866435e-05, "loss": 0.4933, "step": 10116 }, { "epoch": 0.3604396387409374, "grad_norm": 1.073823397882324, "learning_rate": 1.4799830887645867e-05, "loss": 0.4891, "step": 10117 }, { "epoch": 0.36047526586743145, "grad_norm": 1.171417958253554, "learning_rate": 1.4798818532714278e-05, "loss": 0.5654, "step": 10118 }, { "epoch": 0.3605108929939256, "grad_norm": 1.3563644605978042, "learning_rate": 1.4797806113883069e-05, "loss": 0.6014, "step": 10119 }, { "epoch": 0.3605465201204197, "grad_norm": 1.2805980718054175, "learning_rate": 1.4796793631165718e-05, "loss": 0.4819, "step": 10120 }, { "epoch": 0.3605821472469138, "grad_norm": 2.0926108208250347, "learning_rate": 1.4795781084575707e-05, "loss": 0.7127, "step": 10121 }, { "epoch": 0.3606177743734079, "grad_norm": 1.920577759594922, "learning_rate": 1.4794768474126518e-05, "loss": 0.5508, "step": 10122 }, { "epoch": 0.36065340149990205, "grad_norm": 1.4286229799922674, "learning_rate": 1.4793755799831637e-05, "loss": 0.5612, "step": 10123 }, { "epoch": 0.3606890286263961, "grad_norm": 2.5155899443889704, "learning_rate": 1.4792743061704547e-05, "loss": 0.5956, "step": 10124 }, { "epoch": 0.36072465575289026, "grad_norm": 1.544643874146523, "learning_rate": 1.479173025975873e-05, "loss": 0.5547, "step": 10125 }, { "epoch": 0.3607602828793844, "grad_norm": 2.2843113073275134, "learning_rate": 1.4790717394007676e-05, "loss": 0.6104, "step": 10126 }, { "epoch": 0.36079591000587846, "grad_norm": 1.1041357979415898, "learning_rate": 1.4789704464464873e-05, "loss": 0.5233, "step": 10127 }, { "epoch": 0.3608315371323726, "grad_norm": 1.1850026398359765, "learning_rate": 1.4788691471143808e-05, "loss": 0.4918, "step": 10128 }, { "epoch": 0.3608671642588667, "grad_norm": 1.2743470247177482, "learning_rate": 1.4787678414057969e-05, "loss": 0.488, "step": 10129 }, { "epoch": 0.3609027913853608, "grad_norm": 1.6872030165593976, "learning_rate": 1.478666529322084e-05, "loss": 0.5901, "step": 10130 }, { "epoch": 0.36093841851185493, "grad_norm": 1.5888095123882484, "learning_rate": 1.4785652108645925e-05, "loss": 0.6268, "step": 10131 }, { "epoch": 0.36097404563834906, "grad_norm": 1.7935617317127777, "learning_rate": 1.4784638860346702e-05, "loss": 0.5073, "step": 10132 }, { "epoch": 0.36100967276484314, "grad_norm": 1.4925740312307247, "learning_rate": 1.4783625548336669e-05, "loss": 0.4662, "step": 10133 }, { "epoch": 0.36104529989133727, "grad_norm": 1.4810533659492753, "learning_rate": 1.478261217262932e-05, "loss": 0.558, "step": 10134 }, { "epoch": 0.3610809270178314, "grad_norm": 1.4299618866581822, "learning_rate": 1.4781598733238147e-05, "loss": 0.4746, "step": 10135 }, { "epoch": 0.3611165541443255, "grad_norm": 1.5622685928278608, "learning_rate": 1.4780585230176646e-05, "loss": 0.5552, "step": 10136 }, { "epoch": 0.3611521812708196, "grad_norm": 1.340847982675525, "learning_rate": 1.4779571663458308e-05, "loss": 0.5011, "step": 10137 }, { "epoch": 0.36118780839731374, "grad_norm": 3.2718693745613026, "learning_rate": 1.4778558033096633e-05, "loss": 0.4397, "step": 10138 }, { "epoch": 0.3612234355238078, "grad_norm": 1.8101009653398872, "learning_rate": 1.4777544339105121e-05, "loss": 0.5839, "step": 10139 }, { "epoch": 0.36125906265030194, "grad_norm": 2.0062054626877583, "learning_rate": 1.4776530581497261e-05, "loss": 0.5619, "step": 10140 }, { "epoch": 0.3612946897767961, "grad_norm": 1.7304579256432038, "learning_rate": 1.4775516760286561e-05, "loss": 0.4521, "step": 10141 }, { "epoch": 0.36133031690329015, "grad_norm": 1.6551425394834105, "learning_rate": 1.4774502875486518e-05, "loss": 0.5612, "step": 10142 }, { "epoch": 0.3613659440297843, "grad_norm": 1.7219400543226528, "learning_rate": 1.4773488927110633e-05, "loss": 0.4775, "step": 10143 }, { "epoch": 0.3614015711562784, "grad_norm": 1.4254262217440339, "learning_rate": 1.4772474915172403e-05, "loss": 0.4254, "step": 10144 }, { "epoch": 0.3614371982827725, "grad_norm": 1.7879334302667496, "learning_rate": 1.4771460839685336e-05, "loss": 0.4875, "step": 10145 }, { "epoch": 0.3614728254092666, "grad_norm": 2.2062697444659354, "learning_rate": 1.4770446700662934e-05, "loss": 0.5909, "step": 10146 }, { "epoch": 0.36150845253576075, "grad_norm": 1.926878359906837, "learning_rate": 1.47694324981187e-05, "loss": 0.5925, "step": 10147 }, { "epoch": 0.3615440796622548, "grad_norm": 2.4770043614270087, "learning_rate": 1.4768418232066135e-05, "loss": 0.4576, "step": 10148 }, { "epoch": 0.36157970678874896, "grad_norm": 1.5354083620069454, "learning_rate": 1.476740390251875e-05, "loss": 0.5115, "step": 10149 }, { "epoch": 0.3616153339152431, "grad_norm": 2.603108973964114, "learning_rate": 1.4766389509490051e-05, "loss": 0.5434, "step": 10150 }, { "epoch": 0.36165096104173716, "grad_norm": 1.5856443251255452, "learning_rate": 1.4765375052993543e-05, "loss": 0.5622, "step": 10151 }, { "epoch": 0.3616865881682313, "grad_norm": 2.159639314563698, "learning_rate": 1.4764360533042734e-05, "loss": 0.6665, "step": 10152 }, { "epoch": 0.3617222152947254, "grad_norm": 1.296862153531187, "learning_rate": 1.4763345949651137e-05, "loss": 0.4489, "step": 10153 }, { "epoch": 0.3617578424212195, "grad_norm": 1.5192419549419711, "learning_rate": 1.476233130283226e-05, "loss": 0.5025, "step": 10154 }, { "epoch": 0.36179346954771363, "grad_norm": 1.340818453362798, "learning_rate": 1.4761316592599611e-05, "loss": 0.4724, "step": 10155 }, { "epoch": 0.36182909667420776, "grad_norm": 1.6416469498723263, "learning_rate": 1.4760301818966705e-05, "loss": 0.5217, "step": 10156 }, { "epoch": 0.36186472380070184, "grad_norm": 1.9082320387399625, "learning_rate": 1.4759286981947052e-05, "loss": 0.5557, "step": 10157 }, { "epoch": 0.36190035092719597, "grad_norm": 1.2129328996750848, "learning_rate": 1.4758272081554168e-05, "loss": 0.4749, "step": 10158 }, { "epoch": 0.3619359780536901, "grad_norm": 1.6410029922686333, "learning_rate": 1.475725711780156e-05, "loss": 0.5687, "step": 10159 }, { "epoch": 0.3619716051801842, "grad_norm": 1.9057756383136322, "learning_rate": 1.4756242090702756e-05, "loss": 0.559, "step": 10160 }, { "epoch": 0.3620072323066783, "grad_norm": 1.32948430423653, "learning_rate": 1.475522700027126e-05, "loss": 0.4725, "step": 10161 }, { "epoch": 0.36204285943317244, "grad_norm": 1.2216115650149353, "learning_rate": 1.4754211846520594e-05, "loss": 0.4844, "step": 10162 }, { "epoch": 0.3620784865596665, "grad_norm": 2.2038812181266554, "learning_rate": 1.475319662946427e-05, "loss": 0.5784, "step": 10163 }, { "epoch": 0.36211411368616064, "grad_norm": 1.7887203303635426, "learning_rate": 1.4752181349115816e-05, "loss": 0.5673, "step": 10164 }, { "epoch": 0.3621497408126548, "grad_norm": 1.459971396045361, "learning_rate": 1.4751166005488742e-05, "loss": 0.4994, "step": 10165 }, { "epoch": 0.36218536793914885, "grad_norm": 1.9810123224028717, "learning_rate": 1.4750150598596576e-05, "loss": 0.6319, "step": 10166 }, { "epoch": 0.362220995065643, "grad_norm": 1.6710091662546889, "learning_rate": 1.474913512845283e-05, "loss": 0.6176, "step": 10167 }, { "epoch": 0.3622566221921371, "grad_norm": 3.346799952790287, "learning_rate": 1.4748119595071033e-05, "loss": 0.5885, "step": 10168 }, { "epoch": 0.3622922493186312, "grad_norm": 1.5330152210844115, "learning_rate": 1.4747103998464704e-05, "loss": 0.5389, "step": 10169 }, { "epoch": 0.3623278764451253, "grad_norm": 1.6239954682646742, "learning_rate": 1.4746088338647367e-05, "loss": 0.5271, "step": 10170 }, { "epoch": 0.36236350357161945, "grad_norm": 2.735116824699783, "learning_rate": 1.4745072615632547e-05, "loss": 0.6214, "step": 10171 }, { "epoch": 0.3623991306981135, "grad_norm": 1.233565385740398, "learning_rate": 1.474405682943377e-05, "loss": 0.4601, "step": 10172 }, { "epoch": 0.36243475782460766, "grad_norm": 1.2557466537414672, "learning_rate": 1.4743040980064561e-05, "loss": 0.4923, "step": 10173 }, { "epoch": 0.3624703849511018, "grad_norm": 1.5272915401231604, "learning_rate": 1.4742025067538439e-05, "loss": 0.5395, "step": 10174 }, { "epoch": 0.36250601207759586, "grad_norm": 1.8529801487124338, "learning_rate": 1.4741009091868946e-05, "loss": 0.5322, "step": 10175 }, { "epoch": 0.36254163920409, "grad_norm": 1.270882741506401, "learning_rate": 1.4739993053069601e-05, "loss": 0.4394, "step": 10176 }, { "epoch": 0.3625772663305841, "grad_norm": 1.5658047788855016, "learning_rate": 1.4738976951153935e-05, "loss": 0.5832, "step": 10177 }, { "epoch": 0.3626128934570782, "grad_norm": 1.7782643981364152, "learning_rate": 1.4737960786135479e-05, "loss": 0.603, "step": 10178 }, { "epoch": 0.36264852058357233, "grad_norm": 1.667027936416406, "learning_rate": 1.4736944558027764e-05, "loss": 0.5418, "step": 10179 }, { "epoch": 0.36268414771006646, "grad_norm": 1.3575558707550797, "learning_rate": 1.4735928266844322e-05, "loss": 0.4283, "step": 10180 }, { "epoch": 0.36271977483656054, "grad_norm": 1.5256093380178213, "learning_rate": 1.4734911912598681e-05, "loss": 0.6021, "step": 10181 }, { "epoch": 0.36275540196305467, "grad_norm": 3.378205629055753, "learning_rate": 1.4733895495304382e-05, "loss": 0.6304, "step": 10182 }, { "epoch": 0.3627910290895488, "grad_norm": 1.8524543889018446, "learning_rate": 1.4732879014974956e-05, "loss": 0.5964, "step": 10183 }, { "epoch": 0.3628266562160429, "grad_norm": 1.7541790894971871, "learning_rate": 1.4731862471623938e-05, "loss": 0.5232, "step": 10184 }, { "epoch": 0.362862283342537, "grad_norm": 1.599533947911328, "learning_rate": 1.473084586526486e-05, "loss": 0.5948, "step": 10185 }, { "epoch": 0.36289791046903114, "grad_norm": 1.6499982830896356, "learning_rate": 1.4729829195911265e-05, "loss": 0.5778, "step": 10186 }, { "epoch": 0.3629335375955252, "grad_norm": 2.2472992182289206, "learning_rate": 1.472881246357669e-05, "loss": 0.5966, "step": 10187 }, { "epoch": 0.36296916472201934, "grad_norm": 1.7104704377569333, "learning_rate": 1.472779566827467e-05, "loss": 0.4444, "step": 10188 }, { "epoch": 0.3630047918485135, "grad_norm": 1.5370970174482024, "learning_rate": 1.4726778810018745e-05, "loss": 0.4371, "step": 10189 }, { "epoch": 0.36304041897500755, "grad_norm": 1.5937116374795708, "learning_rate": 1.4725761888822458e-05, "loss": 0.5946, "step": 10190 }, { "epoch": 0.3630760461015017, "grad_norm": 2.056604857006183, "learning_rate": 1.4724744904699349e-05, "loss": 0.4596, "step": 10191 }, { "epoch": 0.3631116732279958, "grad_norm": 1.611678923128894, "learning_rate": 1.472372785766296e-05, "loss": 0.5486, "step": 10192 }, { "epoch": 0.3631473003544899, "grad_norm": 2.2086438170727387, "learning_rate": 1.472271074772683e-05, "loss": 0.4745, "step": 10193 }, { "epoch": 0.363182927480984, "grad_norm": 2.5226330055954818, "learning_rate": 1.4721693574904505e-05, "loss": 0.5282, "step": 10194 }, { "epoch": 0.36321855460747815, "grad_norm": 1.5421588477886992, "learning_rate": 1.4720676339209532e-05, "loss": 0.5062, "step": 10195 }, { "epoch": 0.3632541817339722, "grad_norm": 1.6512366276524837, "learning_rate": 1.4719659040655452e-05, "loss": 0.6351, "step": 10196 }, { "epoch": 0.36328980886046636, "grad_norm": 1.411994510851764, "learning_rate": 1.4718641679255814e-05, "loss": 0.4779, "step": 10197 }, { "epoch": 0.3633254359869605, "grad_norm": 1.4278811182423652, "learning_rate": 1.4717624255024165e-05, "loss": 0.5341, "step": 10198 }, { "epoch": 0.36336106311345456, "grad_norm": 1.4515276695448023, "learning_rate": 1.471660676797405e-05, "loss": 0.4456, "step": 10199 }, { "epoch": 0.3633966902399487, "grad_norm": 1.855254052766971, "learning_rate": 1.4715589218119019e-05, "loss": 0.5725, "step": 10200 }, { "epoch": 0.3634323173664428, "grad_norm": 2.233769134604583, "learning_rate": 1.4714571605472622e-05, "loss": 0.5599, "step": 10201 }, { "epoch": 0.3634679444929369, "grad_norm": 1.5458645776703013, "learning_rate": 1.4713553930048413e-05, "loss": 0.4555, "step": 10202 }, { "epoch": 0.36350357161943103, "grad_norm": 1.463332469194534, "learning_rate": 1.4712536191859934e-05, "loss": 0.5409, "step": 10203 }, { "epoch": 0.36353919874592516, "grad_norm": 1.431691531793984, "learning_rate": 1.4711518390920742e-05, "loss": 0.4946, "step": 10204 }, { "epoch": 0.36357482587241924, "grad_norm": 1.5167082391794877, "learning_rate": 1.4710500527244392e-05, "loss": 0.5494, "step": 10205 }, { "epoch": 0.36361045299891337, "grad_norm": 1.745136951220907, "learning_rate": 1.4709482600844433e-05, "loss": 0.4569, "step": 10206 }, { "epoch": 0.3636460801254075, "grad_norm": 1.9969330167187251, "learning_rate": 1.4708464611734421e-05, "loss": 0.5422, "step": 10207 }, { "epoch": 0.3636817072519016, "grad_norm": 1.7497147745216708, "learning_rate": 1.4707446559927914e-05, "loss": 0.5758, "step": 10208 }, { "epoch": 0.3637173343783957, "grad_norm": 2.0746821661770825, "learning_rate": 1.4706428445438462e-05, "loss": 0.5516, "step": 10209 }, { "epoch": 0.36375296150488984, "grad_norm": 2.4047347290975156, "learning_rate": 1.470541026827963e-05, "loss": 0.4578, "step": 10210 }, { "epoch": 0.3637885886313839, "grad_norm": 2.0875379910760983, "learning_rate": 1.4704392028464968e-05, "loss": 0.4728, "step": 10211 }, { "epoch": 0.36382421575787804, "grad_norm": 2.0422725066250846, "learning_rate": 1.470337372600804e-05, "loss": 0.4552, "step": 10212 }, { "epoch": 0.3638598428843722, "grad_norm": 1.8971306411805615, "learning_rate": 1.4702355360922403e-05, "loss": 0.5573, "step": 10213 }, { "epoch": 0.36389547001086625, "grad_norm": 1.5526796782562322, "learning_rate": 1.4701336933221617e-05, "loss": 0.5446, "step": 10214 }, { "epoch": 0.3639310971373604, "grad_norm": 2.0775955673059734, "learning_rate": 1.4700318442919244e-05, "loss": 0.6396, "step": 10215 }, { "epoch": 0.3639667242638545, "grad_norm": 2.669194950444627, "learning_rate": 1.4699299890028847e-05, "loss": 0.4619, "step": 10216 }, { "epoch": 0.3640023513903486, "grad_norm": 2.6293033020780046, "learning_rate": 1.4698281274563986e-05, "loss": 0.6059, "step": 10217 }, { "epoch": 0.3640379785168427, "grad_norm": 1.952392571742882, "learning_rate": 1.4697262596538227e-05, "loss": 0.5236, "step": 10218 }, { "epoch": 0.36407360564333685, "grad_norm": 3.2020597487971894, "learning_rate": 1.4696243855965131e-05, "loss": 0.5315, "step": 10219 }, { "epoch": 0.3641092327698309, "grad_norm": 2.941270727821815, "learning_rate": 1.4695225052858264e-05, "loss": 0.5352, "step": 10220 }, { "epoch": 0.36414485989632506, "grad_norm": 6.8611818214548945, "learning_rate": 1.4694206187231197e-05, "loss": 0.4823, "step": 10221 }, { "epoch": 0.3641804870228192, "grad_norm": 2.105316188067556, "learning_rate": 1.4693187259097495e-05, "loss": 0.5112, "step": 10222 }, { "epoch": 0.36421611414931326, "grad_norm": 2.1829488832068473, "learning_rate": 1.4692168268470722e-05, "loss": 0.4394, "step": 10223 }, { "epoch": 0.3642517412758074, "grad_norm": 4.967646929864929, "learning_rate": 1.4691149215364449e-05, "loss": 0.4439, "step": 10224 }, { "epoch": 0.3642873684023015, "grad_norm": 1.823167537076341, "learning_rate": 1.4690130099792245e-05, "loss": 0.5252, "step": 10225 }, { "epoch": 0.3643229955287956, "grad_norm": 2.4838739202893514, "learning_rate": 1.468911092176768e-05, "loss": 0.6139, "step": 10226 }, { "epoch": 0.36435862265528973, "grad_norm": 3.3644093607170382, "learning_rate": 1.4688091681304327e-05, "loss": 0.5206, "step": 10227 }, { "epoch": 0.36439424978178386, "grad_norm": 3.9850038759512554, "learning_rate": 1.4687072378415756e-05, "loss": 0.4605, "step": 10228 }, { "epoch": 0.36442987690827794, "grad_norm": 2.7128168796127428, "learning_rate": 1.4686053013115542e-05, "loss": 0.6588, "step": 10229 }, { "epoch": 0.36446550403477207, "grad_norm": 2.009594242143269, "learning_rate": 1.4685033585417255e-05, "loss": 0.5378, "step": 10230 }, { "epoch": 0.3645011311612662, "grad_norm": 2.3604965472781143, "learning_rate": 1.4684014095334472e-05, "loss": 0.505, "step": 10231 }, { "epoch": 0.3645367582877603, "grad_norm": 2.5355770074600397, "learning_rate": 1.4682994542880763e-05, "loss": 0.4865, "step": 10232 }, { "epoch": 0.3645723854142544, "grad_norm": 2.413002748985721, "learning_rate": 1.4681974928069714e-05, "loss": 0.6286, "step": 10233 }, { "epoch": 0.36460801254074854, "grad_norm": 3.5622404525163187, "learning_rate": 1.4680955250914892e-05, "loss": 0.465, "step": 10234 }, { "epoch": 0.3646436396672426, "grad_norm": 3.017523832922467, "learning_rate": 1.4679935511429883e-05, "loss": 0.5529, "step": 10235 }, { "epoch": 0.36467926679373674, "grad_norm": 3.126936596695251, "learning_rate": 1.467891570962826e-05, "loss": 0.5328, "step": 10236 }, { "epoch": 0.3647148939202309, "grad_norm": 9.250730475437091, "learning_rate": 1.4677895845523606e-05, "loss": 0.6063, "step": 10237 }, { "epoch": 0.36475052104672495, "grad_norm": 2.1385909766633415, "learning_rate": 1.4676875919129497e-05, "loss": 0.4312, "step": 10238 }, { "epoch": 0.3647861481732191, "grad_norm": 3.772238146925707, "learning_rate": 1.4675855930459516e-05, "loss": 0.5873, "step": 10239 }, { "epoch": 0.3648217752997132, "grad_norm": 3.0270728981656405, "learning_rate": 1.4674835879527245e-05, "loss": 0.5891, "step": 10240 }, { "epoch": 0.3648574024262073, "grad_norm": 3.0664978524109254, "learning_rate": 1.4673815766346267e-05, "loss": 0.5337, "step": 10241 }, { "epoch": 0.3648930295527014, "grad_norm": 2.396705871941776, "learning_rate": 1.4672795590930165e-05, "loss": 0.545, "step": 10242 }, { "epoch": 0.36492865667919555, "grad_norm": 3.9525849890289884, "learning_rate": 1.4671775353292525e-05, "loss": 0.6423, "step": 10243 }, { "epoch": 0.3649642838056896, "grad_norm": 9.47591531053558, "learning_rate": 1.4670755053446927e-05, "loss": 0.6172, "step": 10244 }, { "epoch": 0.36499991093218376, "grad_norm": 3.7561597427719993, "learning_rate": 1.4669734691406965e-05, "loss": 0.561, "step": 10245 }, { "epoch": 0.3650355380586779, "grad_norm": 2.9825976828154745, "learning_rate": 1.4668714267186219e-05, "loss": 0.6831, "step": 10246 }, { "epoch": 0.36507116518517196, "grad_norm": 3.721633113080071, "learning_rate": 1.4667693780798282e-05, "loss": 0.5112, "step": 10247 }, { "epoch": 0.3651067923116661, "grad_norm": 7.145656676327508, "learning_rate": 1.4666673232256738e-05, "loss": 0.5069, "step": 10248 }, { "epoch": 0.3651424194381602, "grad_norm": 3.221393840844719, "learning_rate": 1.4665652621575177e-05, "loss": 0.5022, "step": 10249 }, { "epoch": 0.3651780465646543, "grad_norm": 3.874810431416625, "learning_rate": 1.4664631948767195e-05, "loss": 0.5682, "step": 10250 }, { "epoch": 0.36521367369114843, "grad_norm": 4.692687831541499, "learning_rate": 1.4663611213846374e-05, "loss": 0.5789, "step": 10251 }, { "epoch": 0.36524930081764256, "grad_norm": 2.68678508647574, "learning_rate": 1.466259041682631e-05, "loss": 0.5498, "step": 10252 }, { "epoch": 0.36528492794413664, "grad_norm": 3.613844728089571, "learning_rate": 1.4661569557720596e-05, "loss": 0.5393, "step": 10253 }, { "epoch": 0.36532055507063077, "grad_norm": 6.543016904284045, "learning_rate": 1.4660548636542828e-05, "loss": 0.44, "step": 10254 }, { "epoch": 0.3653561821971249, "grad_norm": 18.87536730817023, "learning_rate": 1.4659527653306593e-05, "loss": 0.6609, "step": 10255 }, { "epoch": 0.365391809323619, "grad_norm": 8.175576858614244, "learning_rate": 1.4658506608025493e-05, "loss": 0.4787, "step": 10256 }, { "epoch": 0.3654274364501131, "grad_norm": 3.491203231108857, "learning_rate": 1.465748550071312e-05, "loss": 0.4475, "step": 10257 }, { "epoch": 0.36546306357660724, "grad_norm": 2.6522228641946075, "learning_rate": 1.4656464331383073e-05, "loss": 0.4777, "step": 10258 }, { "epoch": 0.3654986907031013, "grad_norm": 3.6628552180683216, "learning_rate": 1.4655443100048951e-05, "loss": 0.5094, "step": 10259 }, { "epoch": 0.36553431782959545, "grad_norm": 8.663871134066628, "learning_rate": 1.4654421806724346e-05, "loss": 0.5177, "step": 10260 }, { "epoch": 0.3655699449560896, "grad_norm": 2.194084667065311, "learning_rate": 1.4653400451422863e-05, "loss": 0.4839, "step": 10261 }, { "epoch": 0.36560557208258365, "grad_norm": 1.971015807479646, "learning_rate": 1.4652379034158104e-05, "loss": 0.5267, "step": 10262 }, { "epoch": 0.3656411992090778, "grad_norm": 2.8742656190686526, "learning_rate": 1.4651357554943662e-05, "loss": 0.4892, "step": 10263 }, { "epoch": 0.3656768263355719, "grad_norm": 10.318378406124827, "learning_rate": 1.4650336013793145e-05, "loss": 0.5778, "step": 10264 }, { "epoch": 0.365712453462066, "grad_norm": 2.6036652041947725, "learning_rate": 1.4649314410720153e-05, "loss": 0.5162, "step": 10265 }, { "epoch": 0.3657480805885601, "grad_norm": 2.1193616595096176, "learning_rate": 1.4648292745738291e-05, "loss": 0.5404, "step": 10266 }, { "epoch": 0.36578370771505425, "grad_norm": 3.5646554214181325, "learning_rate": 1.464727101886116e-05, "loss": 0.5241, "step": 10267 }, { "epoch": 0.3658193348415484, "grad_norm": 3.683286623558898, "learning_rate": 1.4646249230102364e-05, "loss": 0.517, "step": 10268 }, { "epoch": 0.36585496196804246, "grad_norm": 2.5233837045407737, "learning_rate": 1.4645227379475518e-05, "loss": 0.5945, "step": 10269 }, { "epoch": 0.3658905890945366, "grad_norm": 4.374346983524234, "learning_rate": 1.464420546699422e-05, "loss": 0.5832, "step": 10270 }, { "epoch": 0.3659262162210307, "grad_norm": 2.067304307208683, "learning_rate": 1.4643183492672081e-05, "loss": 0.5025, "step": 10271 }, { "epoch": 0.3659618433475248, "grad_norm": 2.5335153248818862, "learning_rate": 1.4642161456522708e-05, "loss": 0.5076, "step": 10272 }, { "epoch": 0.3659974704740189, "grad_norm": 3.289363384267429, "learning_rate": 1.4641139358559714e-05, "loss": 0.6114, "step": 10273 }, { "epoch": 0.36603309760051306, "grad_norm": 1.7280254136757958, "learning_rate": 1.4640117198796702e-05, "loss": 0.5278, "step": 10274 }, { "epoch": 0.36606872472700713, "grad_norm": 3.2997048026569265, "learning_rate": 1.4639094977247284e-05, "loss": 0.6053, "step": 10275 }, { "epoch": 0.36610435185350126, "grad_norm": 1.3581006099848045, "learning_rate": 1.4638072693925076e-05, "loss": 0.4603, "step": 10276 }, { "epoch": 0.3661399789799954, "grad_norm": 1.336756084283539, "learning_rate": 1.4637050348843691e-05, "loss": 0.494, "step": 10277 }, { "epoch": 0.36617560610648947, "grad_norm": 2.250585933090058, "learning_rate": 1.4636027942016736e-05, "loss": 0.5387, "step": 10278 }, { "epoch": 0.3662112332329836, "grad_norm": 2.116329773509793, "learning_rate": 1.4635005473457828e-05, "loss": 0.5477, "step": 10279 }, { "epoch": 0.36624686035947773, "grad_norm": 2.0414897445241107, "learning_rate": 1.4633982943180585e-05, "loss": 0.6078, "step": 10280 }, { "epoch": 0.3662824874859718, "grad_norm": 2.0954704008514278, "learning_rate": 1.463296035119862e-05, "loss": 0.5593, "step": 10281 }, { "epoch": 0.36631811461246594, "grad_norm": 1.559341587389041, "learning_rate": 1.4631937697525546e-05, "loss": 0.448, "step": 10282 }, { "epoch": 0.36635374173896007, "grad_norm": 1.5121003611767048, "learning_rate": 1.4630914982174988e-05, "loss": 0.5053, "step": 10283 }, { "epoch": 0.36638936886545415, "grad_norm": 1.5030165192886473, "learning_rate": 1.462989220516056e-05, "loss": 0.4743, "step": 10284 }, { "epoch": 0.3664249959919483, "grad_norm": 1.9431711780116172, "learning_rate": 1.4628869366495881e-05, "loss": 0.6831, "step": 10285 }, { "epoch": 0.3664606231184424, "grad_norm": 4.966939650455177, "learning_rate": 1.4627846466194569e-05, "loss": 0.6515, "step": 10286 }, { "epoch": 0.3664962502449365, "grad_norm": 1.5972537443692825, "learning_rate": 1.4626823504270248e-05, "loss": 0.5234, "step": 10287 }, { "epoch": 0.3665318773714306, "grad_norm": 1.8962351034326388, "learning_rate": 1.4625800480736539e-05, "loss": 0.4966, "step": 10288 }, { "epoch": 0.36656750449792475, "grad_norm": 1.699720539758608, "learning_rate": 1.4624777395607061e-05, "loss": 0.5617, "step": 10289 }, { "epoch": 0.3666031316244188, "grad_norm": 2.213474392342355, "learning_rate": 1.4623754248895439e-05, "loss": 0.5658, "step": 10290 }, { "epoch": 0.36663875875091295, "grad_norm": 1.9019917329694742, "learning_rate": 1.46227310406153e-05, "loss": 0.512, "step": 10291 }, { "epoch": 0.3666743858774071, "grad_norm": 5.738036724512519, "learning_rate": 1.4621707770780267e-05, "loss": 0.5872, "step": 10292 }, { "epoch": 0.36671001300390116, "grad_norm": 1.4249555509428726, "learning_rate": 1.4620684439403962e-05, "loss": 0.5239, "step": 10293 }, { "epoch": 0.3667456401303953, "grad_norm": 1.9086642304838028, "learning_rate": 1.4619661046500015e-05, "loss": 0.5097, "step": 10294 }, { "epoch": 0.3667812672568894, "grad_norm": 1.3765230254200402, "learning_rate": 1.4618637592082053e-05, "loss": 0.4981, "step": 10295 }, { "epoch": 0.3668168943833835, "grad_norm": 1.3764784873420555, "learning_rate": 1.4617614076163702e-05, "loss": 0.4687, "step": 10296 }, { "epoch": 0.3668525215098776, "grad_norm": 3.4213494213739764, "learning_rate": 1.4616590498758591e-05, "loss": 0.4748, "step": 10297 }, { "epoch": 0.36688814863637176, "grad_norm": 1.7958252647402875, "learning_rate": 1.4615566859880353e-05, "loss": 0.5745, "step": 10298 }, { "epoch": 0.36692377576286583, "grad_norm": 2.474552940770517, "learning_rate": 1.4614543159542618e-05, "loss": 0.5204, "step": 10299 }, { "epoch": 0.36695940288935996, "grad_norm": 1.2773349086684898, "learning_rate": 1.4613519397759013e-05, "loss": 0.4733, "step": 10300 }, { "epoch": 0.3669950300158541, "grad_norm": 1.272852612323146, "learning_rate": 1.4612495574543169e-05, "loss": 0.4792, "step": 10301 }, { "epoch": 0.36703065714234817, "grad_norm": 1.5328098604713767, "learning_rate": 1.4611471689908725e-05, "loss": 0.548, "step": 10302 }, { "epoch": 0.3670662842688423, "grad_norm": 1.5855895645600186, "learning_rate": 1.4610447743869313e-05, "loss": 0.5238, "step": 10303 }, { "epoch": 0.36710191139533643, "grad_norm": 1.7608728647075924, "learning_rate": 1.4609423736438568e-05, "loss": 0.5056, "step": 10304 }, { "epoch": 0.3671375385218305, "grad_norm": 1.5621994783713335, "learning_rate": 1.4608399667630122e-05, "loss": 0.5531, "step": 10305 }, { "epoch": 0.36717316564832464, "grad_norm": 2.3463216684923505, "learning_rate": 1.4607375537457616e-05, "loss": 0.5805, "step": 10306 }, { "epoch": 0.36720879277481877, "grad_norm": 4.12694615254876, "learning_rate": 1.4606351345934684e-05, "loss": 0.4518, "step": 10307 }, { "epoch": 0.36724441990131285, "grad_norm": 1.8456731225800926, "learning_rate": 1.4605327093074961e-05, "loss": 0.5554, "step": 10308 }, { "epoch": 0.367280047027807, "grad_norm": 1.464998283208511, "learning_rate": 1.4604302778892091e-05, "loss": 0.4931, "step": 10309 }, { "epoch": 0.3673156741543011, "grad_norm": 3.010992935427156, "learning_rate": 1.4603278403399712e-05, "loss": 0.6626, "step": 10310 }, { "epoch": 0.3673513012807952, "grad_norm": 1.3191143910450616, "learning_rate": 1.4602253966611462e-05, "loss": 0.4935, "step": 10311 }, { "epoch": 0.3673869284072893, "grad_norm": 2.1565121651760544, "learning_rate": 1.4601229468540985e-05, "loss": 0.5785, "step": 10312 }, { "epoch": 0.36742255553378345, "grad_norm": 4.749659832960932, "learning_rate": 1.4600204909201919e-05, "loss": 0.5169, "step": 10313 }, { "epoch": 0.3674581826602775, "grad_norm": 5.346404822752971, "learning_rate": 1.4599180288607911e-05, "loss": 0.6538, "step": 10314 }, { "epoch": 0.36749380978677165, "grad_norm": 2.2200218654197297, "learning_rate": 1.4598155606772605e-05, "loss": 0.53, "step": 10315 }, { "epoch": 0.3675294369132658, "grad_norm": 4.710366348840039, "learning_rate": 1.4597130863709638e-05, "loss": 0.6036, "step": 10316 }, { "epoch": 0.36756506403975986, "grad_norm": 2.465145629398902, "learning_rate": 1.4596106059432664e-05, "loss": 0.5142, "step": 10317 }, { "epoch": 0.367600691166254, "grad_norm": 1.3160355104342178, "learning_rate": 1.4595081193955325e-05, "loss": 0.505, "step": 10318 }, { "epoch": 0.3676363182927481, "grad_norm": 1.9049084629545563, "learning_rate": 1.4594056267291268e-05, "loss": 0.52, "step": 10319 }, { "epoch": 0.3676719454192422, "grad_norm": 2.9736731158779666, "learning_rate": 1.4593031279454142e-05, "loss": 0.5441, "step": 10320 }, { "epoch": 0.3677075725457363, "grad_norm": 1.6494436204372511, "learning_rate": 1.4592006230457592e-05, "loss": 0.5345, "step": 10321 }, { "epoch": 0.36774319967223046, "grad_norm": 1.4202755492962604, "learning_rate": 1.4590981120315271e-05, "loss": 0.4534, "step": 10322 }, { "epoch": 0.36777882679872453, "grad_norm": 2.218079182454644, "learning_rate": 1.4589955949040823e-05, "loss": 0.5207, "step": 10323 }, { "epoch": 0.36781445392521867, "grad_norm": 1.854255074140005, "learning_rate": 1.458893071664791e-05, "loss": 0.5126, "step": 10324 }, { "epoch": 0.3678500810517128, "grad_norm": 2.013361330530067, "learning_rate": 1.4587905423150173e-05, "loss": 0.5053, "step": 10325 }, { "epoch": 0.36788570817820687, "grad_norm": 2.1533461499434563, "learning_rate": 1.4586880068561269e-05, "loss": 0.5713, "step": 10326 }, { "epoch": 0.367921335304701, "grad_norm": 1.085992628410817, "learning_rate": 1.458585465289485e-05, "loss": 0.4608, "step": 10327 }, { "epoch": 0.36795696243119513, "grad_norm": 2.3939673752476813, "learning_rate": 1.4584829176164575e-05, "loss": 0.5207, "step": 10328 }, { "epoch": 0.3679925895576892, "grad_norm": 1.8087589893433285, "learning_rate": 1.4583803638384093e-05, "loss": 0.4955, "step": 10329 }, { "epoch": 0.36802821668418334, "grad_norm": 2.1087253744054046, "learning_rate": 1.458277803956706e-05, "loss": 0.5035, "step": 10330 }, { "epoch": 0.36806384381067747, "grad_norm": 2.0947034731733116, "learning_rate": 1.4581752379727134e-05, "loss": 0.6428, "step": 10331 }, { "epoch": 0.36809947093717155, "grad_norm": 2.632189702751267, "learning_rate": 1.4580726658877975e-05, "loss": 0.5273, "step": 10332 }, { "epoch": 0.3681350980636657, "grad_norm": 7.075217688000476, "learning_rate": 1.4579700877033238e-05, "loss": 0.5724, "step": 10333 }, { "epoch": 0.3681707251901598, "grad_norm": 1.8878133958111443, "learning_rate": 1.4578675034206582e-05, "loss": 0.5163, "step": 10334 }, { "epoch": 0.3682063523166539, "grad_norm": 1.4818031734450576, "learning_rate": 1.4577649130411669e-05, "loss": 0.5084, "step": 10335 }, { "epoch": 0.368241979443148, "grad_norm": 1.968015494477125, "learning_rate": 1.4576623165662157e-05, "loss": 0.5842, "step": 10336 }, { "epoch": 0.36827760656964215, "grad_norm": 2.0248714506444125, "learning_rate": 1.4575597139971708e-05, "loss": 0.6155, "step": 10337 }, { "epoch": 0.3683132336961362, "grad_norm": 2.12262793625251, "learning_rate": 1.4574571053353987e-05, "loss": 0.6069, "step": 10338 }, { "epoch": 0.36834886082263035, "grad_norm": 1.4095401412483926, "learning_rate": 1.4573544905822652e-05, "loss": 0.4652, "step": 10339 }, { "epoch": 0.3683844879491245, "grad_norm": 2.071132879357885, "learning_rate": 1.4572518697391374e-05, "loss": 0.6306, "step": 10340 }, { "epoch": 0.36842011507561856, "grad_norm": 1.425791923903214, "learning_rate": 1.457149242807381e-05, "loss": 0.4557, "step": 10341 }, { "epoch": 0.3684557422021127, "grad_norm": 3.3215734254491878, "learning_rate": 1.4570466097883631e-05, "loss": 0.6586, "step": 10342 }, { "epoch": 0.3684913693286068, "grad_norm": 1.6835568048239375, "learning_rate": 1.45694397068345e-05, "loss": 0.4788, "step": 10343 }, { "epoch": 0.3685269964551009, "grad_norm": 3.277265155315138, "learning_rate": 1.4568413254940086e-05, "loss": 0.6046, "step": 10344 }, { "epoch": 0.36856262358159503, "grad_norm": 1.501524298725227, "learning_rate": 1.4567386742214055e-05, "loss": 0.4963, "step": 10345 }, { "epoch": 0.36859825070808916, "grad_norm": 3.558247783302487, "learning_rate": 1.456636016867008e-05, "loss": 0.5826, "step": 10346 }, { "epoch": 0.36863387783458323, "grad_norm": 1.500171710610078, "learning_rate": 1.4565333534321827e-05, "loss": 0.4988, "step": 10347 }, { "epoch": 0.36866950496107737, "grad_norm": 1.8657738625034963, "learning_rate": 1.4564306839182965e-05, "loss": 0.5193, "step": 10348 }, { "epoch": 0.3687051320875715, "grad_norm": 1.763166276265315, "learning_rate": 1.4563280083267171e-05, "loss": 0.5311, "step": 10349 }, { "epoch": 0.36874075921406557, "grad_norm": 3.1788846209074966, "learning_rate": 1.4562253266588109e-05, "loss": 0.5655, "step": 10350 }, { "epoch": 0.3687763863405597, "grad_norm": 1.8874312609465778, "learning_rate": 1.4561226389159457e-05, "loss": 0.6268, "step": 10351 }, { "epoch": 0.36881201346705383, "grad_norm": 1.8131237401564755, "learning_rate": 1.456019945099489e-05, "loss": 0.6119, "step": 10352 }, { "epoch": 0.3688476405935479, "grad_norm": 1.9153899639497904, "learning_rate": 1.4559172452108077e-05, "loss": 0.5904, "step": 10353 }, { "epoch": 0.36888326772004204, "grad_norm": 2.5723700499272404, "learning_rate": 1.4558145392512696e-05, "loss": 0.5203, "step": 10354 }, { "epoch": 0.36891889484653617, "grad_norm": 2.453937675051579, "learning_rate": 1.4557118272222426e-05, "loss": 0.5551, "step": 10355 }, { "epoch": 0.36895452197303025, "grad_norm": 2.162211259611808, "learning_rate": 1.4556091091250939e-05, "loss": 0.6086, "step": 10356 }, { "epoch": 0.3689901490995244, "grad_norm": 2.279801839498122, "learning_rate": 1.4555063849611915e-05, "loss": 0.4946, "step": 10357 }, { "epoch": 0.3690257762260185, "grad_norm": 1.3205712018529765, "learning_rate": 1.4554036547319033e-05, "loss": 0.5105, "step": 10358 }, { "epoch": 0.3690614033525126, "grad_norm": 1.298730797297362, "learning_rate": 1.4553009184385966e-05, "loss": 0.4965, "step": 10359 }, { "epoch": 0.3690970304790067, "grad_norm": 1.5323221116713286, "learning_rate": 1.4551981760826407e-05, "loss": 0.4857, "step": 10360 }, { "epoch": 0.36913265760550085, "grad_norm": 1.4690736981187773, "learning_rate": 1.4550954276654023e-05, "loss": 0.5137, "step": 10361 }, { "epoch": 0.3691682847319949, "grad_norm": 1.1447611351983198, "learning_rate": 1.4549926731882505e-05, "loss": 0.4385, "step": 10362 }, { "epoch": 0.36920391185848905, "grad_norm": 2.299243749858713, "learning_rate": 1.4548899126525532e-05, "loss": 0.5641, "step": 10363 }, { "epoch": 0.3692395389849832, "grad_norm": 2.994762776589821, "learning_rate": 1.4547871460596787e-05, "loss": 0.5247, "step": 10364 }, { "epoch": 0.36927516611147726, "grad_norm": 1.1673691870362832, "learning_rate": 1.4546843734109957e-05, "loss": 0.4917, "step": 10365 }, { "epoch": 0.3693107932379714, "grad_norm": 1.3904483928530083, "learning_rate": 1.4545815947078722e-05, "loss": 0.5107, "step": 10366 }, { "epoch": 0.3693464203644655, "grad_norm": 1.31308130453606, "learning_rate": 1.4544788099516773e-05, "loss": 0.4963, "step": 10367 }, { "epoch": 0.3693820474909596, "grad_norm": 1.6464867573608861, "learning_rate": 1.454376019143779e-05, "loss": 0.5633, "step": 10368 }, { "epoch": 0.36941767461745373, "grad_norm": 2.0154483303664223, "learning_rate": 1.454273222285547e-05, "loss": 0.6473, "step": 10369 }, { "epoch": 0.36945330174394786, "grad_norm": 2.3041445224952746, "learning_rate": 1.454170419378349e-05, "loss": 0.5665, "step": 10370 }, { "epoch": 0.36948892887044194, "grad_norm": 1.1415713567570238, "learning_rate": 1.4540676104235545e-05, "loss": 0.4815, "step": 10371 }, { "epoch": 0.36952455599693607, "grad_norm": 1.2083531328867727, "learning_rate": 1.4539647954225325e-05, "loss": 0.4973, "step": 10372 }, { "epoch": 0.3695601831234302, "grad_norm": 2.7239622245336514, "learning_rate": 1.4538619743766518e-05, "loss": 0.6417, "step": 10373 }, { "epoch": 0.3695958102499243, "grad_norm": 1.977691473888382, "learning_rate": 1.4537591472872819e-05, "loss": 0.4974, "step": 10374 }, { "epoch": 0.3696314373764184, "grad_norm": 1.4652104581073815, "learning_rate": 1.4536563141557918e-05, "loss": 0.5239, "step": 10375 }, { "epoch": 0.36966706450291253, "grad_norm": 2.0854058934007473, "learning_rate": 1.4535534749835505e-05, "loss": 0.5526, "step": 10376 }, { "epoch": 0.3697026916294066, "grad_norm": 1.8864389663464334, "learning_rate": 1.453450629771928e-05, "loss": 0.5324, "step": 10377 }, { "epoch": 0.36973831875590074, "grad_norm": 3.544074602169662, "learning_rate": 1.4533477785222933e-05, "loss": 0.6664, "step": 10378 }, { "epoch": 0.36977394588239487, "grad_norm": 1.2017382422301477, "learning_rate": 1.4532449212360158e-05, "loss": 0.4555, "step": 10379 }, { "epoch": 0.36980957300888895, "grad_norm": 2.71040725430372, "learning_rate": 1.4531420579144657e-05, "loss": 0.576, "step": 10380 }, { "epoch": 0.3698452001353831, "grad_norm": 3.809213910246506, "learning_rate": 1.4530391885590125e-05, "loss": 0.5812, "step": 10381 }, { "epoch": 0.3698808272618772, "grad_norm": 2.3759781828249333, "learning_rate": 1.4529363131710254e-05, "loss": 0.621, "step": 10382 }, { "epoch": 0.3699164543883713, "grad_norm": 1.1831886472776265, "learning_rate": 1.452833431751875e-05, "loss": 0.4963, "step": 10383 }, { "epoch": 0.3699520815148654, "grad_norm": 2.7867162298714283, "learning_rate": 1.4527305443029307e-05, "loss": 0.6354, "step": 10384 }, { "epoch": 0.36998770864135955, "grad_norm": 1.1960673278051173, "learning_rate": 1.452627650825563e-05, "loss": 0.493, "step": 10385 }, { "epoch": 0.3700233357678536, "grad_norm": 1.2480425547133422, "learning_rate": 1.4525247513211418e-05, "loss": 0.5315, "step": 10386 }, { "epoch": 0.37005896289434775, "grad_norm": 2.840349047949374, "learning_rate": 1.452421845791037e-05, "loss": 0.4974, "step": 10387 }, { "epoch": 0.3700945900208419, "grad_norm": 1.4637604726934264, "learning_rate": 1.4523189342366192e-05, "loss": 0.5022, "step": 10388 }, { "epoch": 0.37013021714733596, "grad_norm": 1.1696148208076413, "learning_rate": 1.4522160166592589e-05, "loss": 0.4956, "step": 10389 }, { "epoch": 0.3701658442738301, "grad_norm": 2.648753630344482, "learning_rate": 1.4521130930603259e-05, "loss": 0.5986, "step": 10390 }, { "epoch": 0.3702014714003242, "grad_norm": 2.21252896843072, "learning_rate": 1.4520101634411912e-05, "loss": 0.6132, "step": 10391 }, { "epoch": 0.3702370985268183, "grad_norm": 1.2763569644901018, "learning_rate": 1.4519072278032254e-05, "loss": 0.4964, "step": 10392 }, { "epoch": 0.37027272565331243, "grad_norm": 1.2695900896072314, "learning_rate": 1.4518042861477988e-05, "loss": 0.5221, "step": 10393 }, { "epoch": 0.37030835277980656, "grad_norm": 2.107584174811218, "learning_rate": 1.4517013384762821e-05, "loss": 0.5764, "step": 10394 }, { "epoch": 0.37034397990630064, "grad_norm": 1.1855829196605203, "learning_rate": 1.4515983847900466e-05, "loss": 0.4416, "step": 10395 }, { "epoch": 0.37037960703279477, "grad_norm": 1.1635457609670108, "learning_rate": 1.4514954250904632e-05, "loss": 0.5145, "step": 10396 }, { "epoch": 0.3704152341592889, "grad_norm": 2.2228661704456023, "learning_rate": 1.4513924593789025e-05, "loss": 0.6178, "step": 10397 }, { "epoch": 0.370450861285783, "grad_norm": 1.046044695848469, "learning_rate": 1.4512894876567355e-05, "loss": 0.4583, "step": 10398 }, { "epoch": 0.3704864884122771, "grad_norm": 1.227050625615497, "learning_rate": 1.4511865099253335e-05, "loss": 0.4608, "step": 10399 }, { "epoch": 0.37052211553877124, "grad_norm": 1.903651467390167, "learning_rate": 1.451083526186068e-05, "loss": 0.6799, "step": 10400 }, { "epoch": 0.3705577426652653, "grad_norm": 1.869595998069498, "learning_rate": 1.45098053644031e-05, "loss": 0.5599, "step": 10401 }, { "epoch": 0.37059336979175944, "grad_norm": 1.2379167382823582, "learning_rate": 1.4508775406894308e-05, "loss": 0.5136, "step": 10402 }, { "epoch": 0.3706289969182536, "grad_norm": 1.1056513784368998, "learning_rate": 1.4507745389348023e-05, "loss": 0.5046, "step": 10403 }, { "epoch": 0.37066462404474765, "grad_norm": 1.9051448355958827, "learning_rate": 1.4506715311777956e-05, "loss": 0.6294, "step": 10404 }, { "epoch": 0.3707002511712418, "grad_norm": 1.1297756083520658, "learning_rate": 1.4505685174197823e-05, "loss": 0.429, "step": 10405 }, { "epoch": 0.3707358782977359, "grad_norm": 1.251155666570446, "learning_rate": 1.4504654976621342e-05, "loss": 0.5205, "step": 10406 }, { "epoch": 0.37077150542423, "grad_norm": 1.1732600355754474, "learning_rate": 1.4503624719062234e-05, "loss": 0.5017, "step": 10407 }, { "epoch": 0.3708071325507241, "grad_norm": 1.2857733275095895, "learning_rate": 1.4502594401534216e-05, "loss": 0.5134, "step": 10408 }, { "epoch": 0.37084275967721825, "grad_norm": 1.9978738331128716, "learning_rate": 1.4501564024051003e-05, "loss": 0.5266, "step": 10409 }, { "epoch": 0.3708783868037123, "grad_norm": 1.985340605838067, "learning_rate": 1.4500533586626322e-05, "loss": 0.6073, "step": 10410 }, { "epoch": 0.37091401393020645, "grad_norm": 1.2120485111539545, "learning_rate": 1.449950308927389e-05, "loss": 0.5178, "step": 10411 }, { "epoch": 0.3709496410567006, "grad_norm": 2.643062250219355, "learning_rate": 1.4498472532007429e-05, "loss": 0.6734, "step": 10412 }, { "epoch": 0.37098526818319466, "grad_norm": 1.49289440291165, "learning_rate": 1.449744191484066e-05, "loss": 0.4894, "step": 10413 }, { "epoch": 0.3710208953096888, "grad_norm": 1.1623111971544944, "learning_rate": 1.449641123778731e-05, "loss": 0.4813, "step": 10414 }, { "epoch": 0.3710565224361829, "grad_norm": 1.9095828213206545, "learning_rate": 1.44953805008611e-05, "loss": 0.5475, "step": 10415 }, { "epoch": 0.371092149562677, "grad_norm": 4.776106306589949, "learning_rate": 1.4494349704075758e-05, "loss": 0.5636, "step": 10416 }, { "epoch": 0.37112777668917113, "grad_norm": 1.9561003777944128, "learning_rate": 1.4493318847445007e-05, "loss": 0.5612, "step": 10417 }, { "epoch": 0.37116340381566526, "grad_norm": 1.6411799332907815, "learning_rate": 1.4492287930982575e-05, "loss": 0.5125, "step": 10418 }, { "epoch": 0.37119903094215934, "grad_norm": 1.1935338150089498, "learning_rate": 1.4491256954702193e-05, "loss": 0.4765, "step": 10419 }, { "epoch": 0.37123465806865347, "grad_norm": 2.0007329942864645, "learning_rate": 1.4490225918617583e-05, "loss": 0.5483, "step": 10420 }, { "epoch": 0.3712702851951476, "grad_norm": 1.3208060004245796, "learning_rate": 1.4489194822742474e-05, "loss": 0.4916, "step": 10421 }, { "epoch": 0.3713059123216417, "grad_norm": 2.1964239513160693, "learning_rate": 1.4488163667090603e-05, "loss": 0.5666, "step": 10422 }, { "epoch": 0.3713415394481358, "grad_norm": 1.7980691901515622, "learning_rate": 1.4487132451675692e-05, "loss": 0.461, "step": 10423 }, { "epoch": 0.37137716657462994, "grad_norm": 1.138355159040749, "learning_rate": 1.4486101176511476e-05, "loss": 0.5026, "step": 10424 }, { "epoch": 0.371412793701124, "grad_norm": 1.2901791986582072, "learning_rate": 1.448506984161169e-05, "loss": 0.4798, "step": 10425 }, { "epoch": 0.37144842082761814, "grad_norm": 1.982888901310502, "learning_rate": 1.4484038446990063e-05, "loss": 0.5634, "step": 10426 }, { "epoch": 0.3714840479541123, "grad_norm": 1.099929798941041, "learning_rate": 1.4483006992660328e-05, "loss": 0.4831, "step": 10427 }, { "epoch": 0.37151967508060635, "grad_norm": 1.0897401228279373, "learning_rate": 1.448197547863622e-05, "loss": 0.473, "step": 10428 }, { "epoch": 0.3715553022071005, "grad_norm": 2.211201940958813, "learning_rate": 1.4480943904931481e-05, "loss": 0.6801, "step": 10429 }, { "epoch": 0.3715909293335946, "grad_norm": 1.1437835347515193, "learning_rate": 1.447991227155984e-05, "loss": 0.4767, "step": 10430 }, { "epoch": 0.3716265564600887, "grad_norm": 1.153001984694563, "learning_rate": 1.4478880578535038e-05, "loss": 0.4428, "step": 10431 }, { "epoch": 0.3716621835865828, "grad_norm": 1.805793847226323, "learning_rate": 1.4477848825870807e-05, "loss": 0.6073, "step": 10432 }, { "epoch": 0.37169781071307695, "grad_norm": 1.1336379038957234, "learning_rate": 1.4476817013580889e-05, "loss": 0.4747, "step": 10433 }, { "epoch": 0.371733437839571, "grad_norm": 1.881355883961467, "learning_rate": 1.4475785141679027e-05, "loss": 0.5648, "step": 10434 }, { "epoch": 0.37176906496606515, "grad_norm": 1.2675063814150798, "learning_rate": 1.4474753210178953e-05, "loss": 0.4841, "step": 10435 }, { "epoch": 0.3718046920925593, "grad_norm": 1.9153975683460769, "learning_rate": 1.4473721219094414e-05, "loss": 0.5916, "step": 10436 }, { "epoch": 0.37184031921905336, "grad_norm": 1.2585437402616497, "learning_rate": 1.4472689168439152e-05, "loss": 0.5559, "step": 10437 }, { "epoch": 0.3718759463455475, "grad_norm": 4.209229541636941, "learning_rate": 1.4471657058226906e-05, "loss": 0.5788, "step": 10438 }, { "epoch": 0.3719115734720416, "grad_norm": 2.6540355149597685, "learning_rate": 1.447062488847142e-05, "loss": 0.5376, "step": 10439 }, { "epoch": 0.3719472005985357, "grad_norm": 2.237108853359898, "learning_rate": 1.4469592659186438e-05, "loss": 0.5665, "step": 10440 }, { "epoch": 0.37198282772502983, "grad_norm": 1.0859037546595196, "learning_rate": 1.4468560370385706e-05, "loss": 0.5173, "step": 10441 }, { "epoch": 0.37201845485152396, "grad_norm": 2.3046464045732424, "learning_rate": 1.4467528022082972e-05, "loss": 0.6068, "step": 10442 }, { "epoch": 0.37205408197801804, "grad_norm": 1.9191388433319292, "learning_rate": 1.4466495614291977e-05, "loss": 0.5556, "step": 10443 }, { "epoch": 0.37208970910451217, "grad_norm": 1.080109949483522, "learning_rate": 1.4465463147026472e-05, "loss": 0.4433, "step": 10444 }, { "epoch": 0.3721253362310063, "grad_norm": 1.7900414946272811, "learning_rate": 1.4464430620300205e-05, "loss": 0.5306, "step": 10445 }, { "epoch": 0.3721609633575004, "grad_norm": 2.1507468599139865, "learning_rate": 1.446339803412692e-05, "loss": 0.5703, "step": 10446 }, { "epoch": 0.3721965904839945, "grad_norm": 1.2929663987068616, "learning_rate": 1.4462365388520376e-05, "loss": 0.5023, "step": 10447 }, { "epoch": 0.37223221761048864, "grad_norm": 2.03402975490242, "learning_rate": 1.4461332683494317e-05, "loss": 0.5322, "step": 10448 }, { "epoch": 0.3722678447369827, "grad_norm": 1.6608217470500775, "learning_rate": 1.4460299919062495e-05, "loss": 0.5133, "step": 10449 }, { "epoch": 0.37230347186347684, "grad_norm": 1.1150946679151306, "learning_rate": 1.4459267095238659e-05, "loss": 0.4988, "step": 10450 }, { "epoch": 0.372339098989971, "grad_norm": 1.259052394174005, "learning_rate": 1.4458234212036563e-05, "loss": 0.4902, "step": 10451 }, { "epoch": 0.37237472611646505, "grad_norm": 1.618550481712441, "learning_rate": 1.4457201269469967e-05, "loss": 0.5893, "step": 10452 }, { "epoch": 0.3724103532429592, "grad_norm": 2.5198906205340013, "learning_rate": 1.4456168267552619e-05, "loss": 0.527, "step": 10453 }, { "epoch": 0.3724459803694533, "grad_norm": 8.041717191793275, "learning_rate": 1.4455135206298275e-05, "loss": 0.5183, "step": 10454 }, { "epoch": 0.3724816074959474, "grad_norm": 1.0835142021616588, "learning_rate": 1.4454102085720693e-05, "loss": 0.486, "step": 10455 }, { "epoch": 0.3725172346224415, "grad_norm": 1.9958013184497476, "learning_rate": 1.4453068905833629e-05, "loss": 0.7135, "step": 10456 }, { "epoch": 0.37255286174893565, "grad_norm": 1.0317436022596223, "learning_rate": 1.4452035666650841e-05, "loss": 0.4817, "step": 10457 }, { "epoch": 0.3725884888754298, "grad_norm": 1.5462793957836238, "learning_rate": 1.4451002368186084e-05, "loss": 0.6078, "step": 10458 }, { "epoch": 0.37262411600192386, "grad_norm": 1.1340993189368513, "learning_rate": 1.444996901045312e-05, "loss": 0.4812, "step": 10459 }, { "epoch": 0.372659743128418, "grad_norm": 1.632714719609782, "learning_rate": 1.4448935593465709e-05, "loss": 0.6233, "step": 10460 }, { "epoch": 0.3726953702549121, "grad_norm": 1.2849772307527043, "learning_rate": 1.4447902117237608e-05, "loss": 0.4761, "step": 10461 }, { "epoch": 0.3727309973814062, "grad_norm": 1.0662740761836635, "learning_rate": 1.4446868581782586e-05, "loss": 0.4732, "step": 10462 }, { "epoch": 0.3727666245079003, "grad_norm": 1.0903197861589788, "learning_rate": 1.4445834987114395e-05, "loss": 0.409, "step": 10463 }, { "epoch": 0.37280225163439445, "grad_norm": 2.1639214011727366, "learning_rate": 1.4444801333246809e-05, "loss": 0.5717, "step": 10464 }, { "epoch": 0.37283787876088853, "grad_norm": 1.9642407711268435, "learning_rate": 1.4443767620193582e-05, "loss": 0.5877, "step": 10465 }, { "epoch": 0.37287350588738266, "grad_norm": 1.0837359528886983, "learning_rate": 1.4442733847968488e-05, "loss": 0.4525, "step": 10466 }, { "epoch": 0.3729091330138768, "grad_norm": 1.9517977852519497, "learning_rate": 1.4441700016585287e-05, "loss": 0.5983, "step": 10467 }, { "epoch": 0.37294476014037087, "grad_norm": 1.8267615665435346, "learning_rate": 1.4440666126057743e-05, "loss": 0.6012, "step": 10468 }, { "epoch": 0.372980387266865, "grad_norm": 2.6909544229285403, "learning_rate": 1.4439632176399628e-05, "loss": 0.4941, "step": 10469 }, { "epoch": 0.37301601439335913, "grad_norm": 1.1565600602603334, "learning_rate": 1.4438598167624704e-05, "loss": 0.488, "step": 10470 }, { "epoch": 0.3730516415198532, "grad_norm": 1.2553624931397318, "learning_rate": 1.4437564099746746e-05, "loss": 0.5075, "step": 10471 }, { "epoch": 0.37308726864634734, "grad_norm": 1.2892934553650353, "learning_rate": 1.4436529972779519e-05, "loss": 0.5446, "step": 10472 }, { "epoch": 0.37312289577284147, "grad_norm": 1.997623260092794, "learning_rate": 1.4435495786736796e-05, "loss": 0.581, "step": 10473 }, { "epoch": 0.37315852289933554, "grad_norm": 1.84179812154482, "learning_rate": 1.4434461541632344e-05, "loss": 0.5798, "step": 10474 }, { "epoch": 0.3731941500258297, "grad_norm": 4.062174477253899, "learning_rate": 1.4433427237479938e-05, "loss": 0.5247, "step": 10475 }, { "epoch": 0.3732297771523238, "grad_norm": 1.9352028554422285, "learning_rate": 1.443239287429335e-05, "loss": 0.5048, "step": 10476 }, { "epoch": 0.3732654042788179, "grad_norm": 2.1598437739007266, "learning_rate": 1.4431358452086353e-05, "loss": 0.5913, "step": 10477 }, { "epoch": 0.373301031405312, "grad_norm": 1.1373786076807815, "learning_rate": 1.4430323970872721e-05, "loss": 0.4725, "step": 10478 }, { "epoch": 0.37333665853180614, "grad_norm": 1.9791362704875202, "learning_rate": 1.4429289430666229e-05, "loss": 0.5747, "step": 10479 }, { "epoch": 0.3733722856583002, "grad_norm": 1.0973546213233878, "learning_rate": 1.4428254831480651e-05, "loss": 0.4741, "step": 10480 }, { "epoch": 0.37340791278479435, "grad_norm": 1.6691211598817368, "learning_rate": 1.4427220173329766e-05, "loss": 0.5517, "step": 10481 }, { "epoch": 0.3734435399112885, "grad_norm": 1.9034257235566006, "learning_rate": 1.4426185456227348e-05, "loss": 0.6005, "step": 10482 }, { "epoch": 0.37347916703778256, "grad_norm": 1.2405981894448963, "learning_rate": 1.4425150680187181e-05, "loss": 0.48, "step": 10483 }, { "epoch": 0.3735147941642767, "grad_norm": 1.6220197953222437, "learning_rate": 1.4424115845223035e-05, "loss": 0.6824, "step": 10484 }, { "epoch": 0.3735504212907708, "grad_norm": 2.875853824783092, "learning_rate": 1.4423080951348696e-05, "loss": 0.7137, "step": 10485 }, { "epoch": 0.3735860484172649, "grad_norm": 2.0256604772745526, "learning_rate": 1.4422045998577943e-05, "loss": 0.4893, "step": 10486 }, { "epoch": 0.373621675543759, "grad_norm": 1.175401290485698, "learning_rate": 1.4421010986924557e-05, "loss": 0.5003, "step": 10487 }, { "epoch": 0.37365730267025316, "grad_norm": 1.332776743761371, "learning_rate": 1.4419975916402317e-05, "loss": 0.5707, "step": 10488 }, { "epoch": 0.37369292979674723, "grad_norm": 1.9199659388836474, "learning_rate": 1.4418940787025011e-05, "loss": 0.6502, "step": 10489 }, { "epoch": 0.37372855692324136, "grad_norm": 2.082436553249846, "learning_rate": 1.4417905598806422e-05, "loss": 0.5743, "step": 10490 }, { "epoch": 0.3737641840497355, "grad_norm": 1.344897012536935, "learning_rate": 1.4416870351760325e-05, "loss": 0.4744, "step": 10491 }, { "epoch": 0.37379981117622957, "grad_norm": 1.1754035175019253, "learning_rate": 1.4415835045900516e-05, "loss": 0.5175, "step": 10492 }, { "epoch": 0.3738354383027237, "grad_norm": 0.9979747005983994, "learning_rate": 1.4414799681240781e-05, "loss": 0.4643, "step": 10493 }, { "epoch": 0.37387106542921783, "grad_norm": 1.2951331411111529, "learning_rate": 1.4413764257794898e-05, "loss": 0.484, "step": 10494 }, { "epoch": 0.3739066925557119, "grad_norm": 1.063736011815892, "learning_rate": 1.4412728775576658e-05, "loss": 0.4795, "step": 10495 }, { "epoch": 0.37394231968220604, "grad_norm": 1.1299539313627216, "learning_rate": 1.4411693234599853e-05, "loss": 0.5095, "step": 10496 }, { "epoch": 0.37397794680870017, "grad_norm": 1.1158579653613663, "learning_rate": 1.4410657634878267e-05, "loss": 0.4576, "step": 10497 }, { "epoch": 0.37401357393519424, "grad_norm": 7.251298137392651, "learning_rate": 1.4409621976425691e-05, "loss": 0.5996, "step": 10498 }, { "epoch": 0.3740492010616884, "grad_norm": 1.5331394867008337, "learning_rate": 1.4408586259255917e-05, "loss": 0.4337, "step": 10499 }, { "epoch": 0.3740848281881825, "grad_norm": 2.0343008167313092, "learning_rate": 1.4407550483382737e-05, "loss": 0.5291, "step": 10500 }, { "epoch": 0.3741204553146766, "grad_norm": 4.026486050931422, "learning_rate": 1.4406514648819943e-05, "loss": 0.635, "step": 10501 }, { "epoch": 0.3741560824411707, "grad_norm": 1.1451011729935008, "learning_rate": 1.4405478755581323e-05, "loss": 0.4606, "step": 10502 }, { "epoch": 0.37419170956766484, "grad_norm": 1.0125231833879638, "learning_rate": 1.4404442803680673e-05, "loss": 0.4469, "step": 10503 }, { "epoch": 0.3742273366941589, "grad_norm": 2.0117484352962665, "learning_rate": 1.4403406793131792e-05, "loss": 0.5548, "step": 10504 }, { "epoch": 0.37426296382065305, "grad_norm": 3.115258152837156, "learning_rate": 1.4402370723948471e-05, "loss": 0.5973, "step": 10505 }, { "epoch": 0.3742985909471472, "grad_norm": 1.3856983012853128, "learning_rate": 1.4401334596144506e-05, "loss": 0.4936, "step": 10506 }, { "epoch": 0.37433421807364126, "grad_norm": 1.9873071249953687, "learning_rate": 1.4400298409733695e-05, "loss": 0.6841, "step": 10507 }, { "epoch": 0.3743698452001354, "grad_norm": 1.126862117236692, "learning_rate": 1.4399262164729835e-05, "loss": 0.4476, "step": 10508 }, { "epoch": 0.3744054723266295, "grad_norm": 1.7781075942327644, "learning_rate": 1.4398225861146724e-05, "loss": 0.5301, "step": 10509 }, { "epoch": 0.3744410994531236, "grad_norm": 2.16977206060334, "learning_rate": 1.4397189498998161e-05, "loss": 0.6315, "step": 10510 }, { "epoch": 0.3744767265796177, "grad_norm": 1.0580662383233532, "learning_rate": 1.4396153078297948e-05, "loss": 0.4658, "step": 10511 }, { "epoch": 0.37451235370611186, "grad_norm": 2.1948156167524138, "learning_rate": 1.4395116599059882e-05, "loss": 0.53, "step": 10512 }, { "epoch": 0.37454798083260593, "grad_norm": 1.5940162361898669, "learning_rate": 1.439408006129777e-05, "loss": 0.6146, "step": 10513 }, { "epoch": 0.37458360795910006, "grad_norm": 1.0642966800126121, "learning_rate": 1.4393043465025405e-05, "loss": 0.5299, "step": 10514 }, { "epoch": 0.3746192350855942, "grad_norm": 1.0705902797957303, "learning_rate": 1.43920068102566e-05, "loss": 0.4794, "step": 10515 }, { "epoch": 0.37465486221208827, "grad_norm": 0.9516635844274524, "learning_rate": 1.4390970097005155e-05, "loss": 0.4943, "step": 10516 }, { "epoch": 0.3746904893385824, "grad_norm": 1.3861503737164003, "learning_rate": 1.4389933325284871e-05, "loss": 0.5006, "step": 10517 }, { "epoch": 0.37472611646507653, "grad_norm": 1.112778464172716, "learning_rate": 1.438889649510956e-05, "loss": 0.4644, "step": 10518 }, { "epoch": 0.3747617435915706, "grad_norm": 1.5702191470865003, "learning_rate": 1.4387859606493022e-05, "loss": 0.5131, "step": 10519 }, { "epoch": 0.37479737071806474, "grad_norm": 1.1272532078917683, "learning_rate": 1.4386822659449067e-05, "loss": 0.5103, "step": 10520 }, { "epoch": 0.37483299784455887, "grad_norm": 2.0767424928219236, "learning_rate": 1.43857856539915e-05, "loss": 0.6447, "step": 10521 }, { "epoch": 0.37486862497105294, "grad_norm": 1.1279985862185031, "learning_rate": 1.4384748590134134e-05, "loss": 0.5077, "step": 10522 }, { "epoch": 0.3749042520975471, "grad_norm": 1.5822123007512308, "learning_rate": 1.4383711467890776e-05, "loss": 0.5387, "step": 10523 }, { "epoch": 0.3749398792240412, "grad_norm": 1.4932642195653898, "learning_rate": 1.4382674287275235e-05, "loss": 0.564, "step": 10524 }, { "epoch": 0.3749755063505353, "grad_norm": 1.281069172905629, "learning_rate": 1.4381637048301323e-05, "loss": 0.4754, "step": 10525 }, { "epoch": 0.3750111334770294, "grad_norm": 1.5618430868607285, "learning_rate": 1.4380599750982848e-05, "loss": 0.5474, "step": 10526 }, { "epoch": 0.37504676060352354, "grad_norm": 1.5062155044754137, "learning_rate": 1.437956239533363e-05, "loss": 0.5822, "step": 10527 }, { "epoch": 0.3750823877300176, "grad_norm": 1.0898416779120854, "learning_rate": 1.4378524981367474e-05, "loss": 0.4988, "step": 10528 }, { "epoch": 0.37511801485651175, "grad_norm": 1.2445734138898847, "learning_rate": 1.4377487509098203e-05, "loss": 0.5252, "step": 10529 }, { "epoch": 0.3751536419830059, "grad_norm": 1.3734532258651069, "learning_rate": 1.4376449978539621e-05, "loss": 0.5312, "step": 10530 }, { "epoch": 0.37518926910949996, "grad_norm": 1.1467960925399932, "learning_rate": 1.437541238970555e-05, "loss": 0.4382, "step": 10531 }, { "epoch": 0.3752248962359941, "grad_norm": 1.600709128701629, "learning_rate": 1.4374374742609805e-05, "loss": 0.5316, "step": 10532 }, { "epoch": 0.3752605233624882, "grad_norm": 1.9031520111531366, "learning_rate": 1.43733370372662e-05, "loss": 0.6255, "step": 10533 }, { "epoch": 0.3752961504889823, "grad_norm": 1.161693184052014, "learning_rate": 1.437229927368856e-05, "loss": 0.4915, "step": 10534 }, { "epoch": 0.3753317776154764, "grad_norm": 1.7379993774960374, "learning_rate": 1.4371261451890697e-05, "loss": 0.52, "step": 10535 }, { "epoch": 0.37536740474197056, "grad_norm": 1.095191144042231, "learning_rate": 1.4370223571886432e-05, "loss": 0.4569, "step": 10536 }, { "epoch": 0.37540303186846463, "grad_norm": 1.0713632312955428, "learning_rate": 1.4369185633689587e-05, "loss": 0.4898, "step": 10537 }, { "epoch": 0.37543865899495876, "grad_norm": 1.9143375884907097, "learning_rate": 1.436814763731398e-05, "loss": 0.5383, "step": 10538 }, { "epoch": 0.3754742861214529, "grad_norm": 1.2151370572671294, "learning_rate": 1.4367109582773435e-05, "loss": 0.4486, "step": 10539 }, { "epoch": 0.37550991324794697, "grad_norm": 1.56596360043727, "learning_rate": 1.4366071470081773e-05, "loss": 0.6316, "step": 10540 }, { "epoch": 0.3755455403744411, "grad_norm": 1.0469597046886128, "learning_rate": 1.436503329925282e-05, "loss": 0.502, "step": 10541 }, { "epoch": 0.37558116750093523, "grad_norm": 1.3724129065837076, "learning_rate": 1.4363995070300395e-05, "loss": 0.5358, "step": 10542 }, { "epoch": 0.3756167946274293, "grad_norm": 1.0817824677598413, "learning_rate": 1.4362956783238325e-05, "loss": 0.451, "step": 10543 }, { "epoch": 0.37565242175392344, "grad_norm": 1.7279595551648743, "learning_rate": 1.4361918438080436e-05, "loss": 0.5785, "step": 10544 }, { "epoch": 0.37568804888041757, "grad_norm": 1.1710596000789761, "learning_rate": 1.4360880034840554e-05, "loss": 0.4996, "step": 10545 }, { "epoch": 0.37572367600691164, "grad_norm": 1.235589805012612, "learning_rate": 1.4359841573532509e-05, "loss": 0.5141, "step": 10546 }, { "epoch": 0.3757593031334058, "grad_norm": 2.3940390084643806, "learning_rate": 1.4358803054170122e-05, "loss": 0.6604, "step": 10547 }, { "epoch": 0.3757949302598999, "grad_norm": 1.1249125042532009, "learning_rate": 1.435776447676723e-05, "loss": 0.5109, "step": 10548 }, { "epoch": 0.375830557386394, "grad_norm": 1.0336032107054927, "learning_rate": 1.4356725841337655e-05, "loss": 0.4614, "step": 10549 }, { "epoch": 0.3758661845128881, "grad_norm": 1.0307073136248281, "learning_rate": 1.4355687147895232e-05, "loss": 0.5263, "step": 10550 }, { "epoch": 0.37590181163938224, "grad_norm": 1.1054457084173335, "learning_rate": 1.4354648396453789e-05, "loss": 0.4798, "step": 10551 }, { "epoch": 0.3759374387658763, "grad_norm": 1.1054165414785948, "learning_rate": 1.435360958702716e-05, "loss": 0.4867, "step": 10552 }, { "epoch": 0.37597306589237045, "grad_norm": 1.1273819729349572, "learning_rate": 1.4352570719629177e-05, "loss": 0.5473, "step": 10553 }, { "epoch": 0.3760086930188646, "grad_norm": 1.3748933351945174, "learning_rate": 1.435153179427367e-05, "loss": 0.5845, "step": 10554 }, { "epoch": 0.37604432014535866, "grad_norm": 1.1281876092435428, "learning_rate": 1.4350492810974474e-05, "loss": 0.4614, "step": 10555 }, { "epoch": 0.3760799472718528, "grad_norm": 1.067312845087612, "learning_rate": 1.4349453769745431e-05, "loss": 0.4687, "step": 10556 }, { "epoch": 0.3761155743983469, "grad_norm": 1.0955668471475388, "learning_rate": 1.434841467060037e-05, "loss": 0.4914, "step": 10557 }, { "epoch": 0.376151201524841, "grad_norm": 1.0686338164428102, "learning_rate": 1.4347375513553126e-05, "loss": 0.4327, "step": 10558 }, { "epoch": 0.3761868286513351, "grad_norm": 1.4390030840768087, "learning_rate": 1.4346336298617537e-05, "loss": 0.6452, "step": 10559 }, { "epoch": 0.37622245577782926, "grad_norm": 1.16167302050978, "learning_rate": 1.4345297025807447e-05, "loss": 0.4894, "step": 10560 }, { "epoch": 0.37625808290432333, "grad_norm": 2.0280946953735013, "learning_rate": 1.4344257695136687e-05, "loss": 0.564, "step": 10561 }, { "epoch": 0.37629371003081746, "grad_norm": 2.501261693025605, "learning_rate": 1.4343218306619098e-05, "loss": 0.5491, "step": 10562 }, { "epoch": 0.3763293371573116, "grad_norm": 1.5629803093533308, "learning_rate": 1.4342178860268523e-05, "loss": 0.6564, "step": 10563 }, { "epoch": 0.37636496428380567, "grad_norm": 1.6160360428662361, "learning_rate": 1.4341139356098803e-05, "loss": 0.6194, "step": 10564 }, { "epoch": 0.3764005914102998, "grad_norm": 1.845266797577288, "learning_rate": 1.4340099794123779e-05, "loss": 0.5781, "step": 10565 }, { "epoch": 0.37643621853679393, "grad_norm": 1.4455700608086894, "learning_rate": 1.4339060174357289e-05, "loss": 0.5606, "step": 10566 }, { "epoch": 0.376471845663288, "grad_norm": 1.0515686894666887, "learning_rate": 1.433802049681318e-05, "loss": 0.4986, "step": 10567 }, { "epoch": 0.37650747278978214, "grad_norm": 1.397942027543431, "learning_rate": 1.4336980761505297e-05, "loss": 0.5379, "step": 10568 }, { "epoch": 0.37654309991627627, "grad_norm": 1.4247134834590165, "learning_rate": 1.4335940968447486e-05, "loss": 0.6032, "step": 10569 }, { "epoch": 0.37657872704277034, "grad_norm": 2.2117283423211127, "learning_rate": 1.4334901117653588e-05, "loss": 0.5903, "step": 10570 }, { "epoch": 0.3766143541692645, "grad_norm": 1.4471641364485925, "learning_rate": 1.4333861209137453e-05, "loss": 0.6119, "step": 10571 }, { "epoch": 0.3766499812957586, "grad_norm": 1.1334354718670845, "learning_rate": 1.4332821242912927e-05, "loss": 0.4587, "step": 10572 }, { "epoch": 0.3766856084222527, "grad_norm": 1.02922513552082, "learning_rate": 1.4331781218993855e-05, "loss": 0.489, "step": 10573 }, { "epoch": 0.3767212355487468, "grad_norm": 2.060588579409379, "learning_rate": 1.4330741137394091e-05, "loss": 0.5333, "step": 10574 }, { "epoch": 0.37675686267524094, "grad_norm": 1.03751711386482, "learning_rate": 1.432970099812748e-05, "loss": 0.4206, "step": 10575 }, { "epoch": 0.376792489801735, "grad_norm": 1.523314465672292, "learning_rate": 1.4328660801207876e-05, "loss": 0.5982, "step": 10576 }, { "epoch": 0.37682811692822915, "grad_norm": 1.8525060575192074, "learning_rate": 1.4327620546649126e-05, "loss": 0.5715, "step": 10577 }, { "epoch": 0.3768637440547233, "grad_norm": 6.291459801024156, "learning_rate": 1.4326580234465084e-05, "loss": 0.4466, "step": 10578 }, { "epoch": 0.37689937118121736, "grad_norm": 2.5674630408590953, "learning_rate": 1.4325539864669602e-05, "loss": 0.5536, "step": 10579 }, { "epoch": 0.3769349983077115, "grad_norm": 1.4391002255017908, "learning_rate": 1.4324499437276536e-05, "loss": 0.4949, "step": 10580 }, { "epoch": 0.3769706254342056, "grad_norm": 1.0793763490452077, "learning_rate": 1.4323458952299731e-05, "loss": 0.4785, "step": 10581 }, { "epoch": 0.3770062525606997, "grad_norm": 1.6737480810953485, "learning_rate": 1.4322418409753053e-05, "loss": 0.5048, "step": 10582 }, { "epoch": 0.3770418796871938, "grad_norm": 1.7540045561713475, "learning_rate": 1.4321377809650355e-05, "loss": 0.5049, "step": 10583 }, { "epoch": 0.37707750681368796, "grad_norm": 1.1326509715474984, "learning_rate": 1.4320337152005487e-05, "loss": 0.4906, "step": 10584 }, { "epoch": 0.37711313394018203, "grad_norm": 1.648675685046588, "learning_rate": 1.431929643683231e-05, "loss": 0.5667, "step": 10585 }, { "epoch": 0.37714876106667616, "grad_norm": 1.1017460525465796, "learning_rate": 1.4318255664144682e-05, "loss": 0.4713, "step": 10586 }, { "epoch": 0.3771843881931703, "grad_norm": 1.0221160181526614, "learning_rate": 1.4317214833956465e-05, "loss": 0.5005, "step": 10587 }, { "epoch": 0.37722001531966437, "grad_norm": 1.237857278536036, "learning_rate": 1.431617394628151e-05, "loss": 0.5271, "step": 10588 }, { "epoch": 0.3772556424461585, "grad_norm": 1.0965197820226749, "learning_rate": 1.4315133001133686e-05, "loss": 0.5254, "step": 10589 }, { "epoch": 0.37729126957265263, "grad_norm": 1.1316000116559932, "learning_rate": 1.4314091998526846e-05, "loss": 0.434, "step": 10590 }, { "epoch": 0.3773268966991467, "grad_norm": 1.5900943510110346, "learning_rate": 1.431305093847486e-05, "loss": 0.5138, "step": 10591 }, { "epoch": 0.37736252382564084, "grad_norm": 1.5425075268509603, "learning_rate": 1.4312009820991584e-05, "loss": 0.5859, "step": 10592 }, { "epoch": 0.37739815095213497, "grad_norm": 1.1857494617574167, "learning_rate": 1.4310968646090884e-05, "loss": 0.4848, "step": 10593 }, { "epoch": 0.37743377807862905, "grad_norm": 2.074738793483741, "learning_rate": 1.4309927413786624e-05, "loss": 0.6219, "step": 10594 }, { "epoch": 0.3774694052051232, "grad_norm": 1.8945332658538, "learning_rate": 1.4308886124092668e-05, "loss": 0.577, "step": 10595 }, { "epoch": 0.3775050323316173, "grad_norm": 1.4489567318897845, "learning_rate": 1.4307844777022876e-05, "loss": 0.5119, "step": 10596 }, { "epoch": 0.3775406594581114, "grad_norm": 1.4612575979677231, "learning_rate": 1.4306803372591126e-05, "loss": 0.6911, "step": 10597 }, { "epoch": 0.3775762865846055, "grad_norm": 1.1836158776155785, "learning_rate": 1.4305761910811278e-05, "loss": 0.4841, "step": 10598 }, { "epoch": 0.37761191371109964, "grad_norm": 1.134088116188119, "learning_rate": 1.43047203916972e-05, "loss": 0.5149, "step": 10599 }, { "epoch": 0.3776475408375937, "grad_norm": 1.0312157356650735, "learning_rate": 1.430367881526276e-05, "loss": 0.4317, "step": 10600 }, { "epoch": 0.37768316796408785, "grad_norm": 1.30922783607345, "learning_rate": 1.4302637181521828e-05, "loss": 0.6037, "step": 10601 }, { "epoch": 0.377718795090582, "grad_norm": 1.1143174719740678, "learning_rate": 1.4301595490488275e-05, "loss": 0.4429, "step": 10602 }, { "epoch": 0.37775442221707606, "grad_norm": 1.2430943432106842, "learning_rate": 1.4300553742175974e-05, "loss": 0.4958, "step": 10603 }, { "epoch": 0.3777900493435702, "grad_norm": 1.234125306176913, "learning_rate": 1.4299511936598791e-05, "loss": 0.4961, "step": 10604 }, { "epoch": 0.3778256764700643, "grad_norm": 1.252861728312238, "learning_rate": 1.4298470073770605e-05, "loss": 0.4937, "step": 10605 }, { "epoch": 0.3778613035965584, "grad_norm": 1.5879459639377498, "learning_rate": 1.4297428153705282e-05, "loss": 0.6376, "step": 10606 }, { "epoch": 0.3778969307230525, "grad_norm": 1.0268084894414922, "learning_rate": 1.4296386176416701e-05, "loss": 0.3875, "step": 10607 }, { "epoch": 0.37793255784954666, "grad_norm": 1.3895216981467016, "learning_rate": 1.4295344141918734e-05, "loss": 0.6297, "step": 10608 }, { "epoch": 0.37796818497604073, "grad_norm": 1.2063047920846957, "learning_rate": 1.429430205022526e-05, "loss": 0.4464, "step": 10609 }, { "epoch": 0.37800381210253486, "grad_norm": 1.0464858439007678, "learning_rate": 1.429325990135015e-05, "loss": 0.4749, "step": 10610 }, { "epoch": 0.378039439229029, "grad_norm": 1.983913709751919, "learning_rate": 1.4292217695307285e-05, "loss": 0.4851, "step": 10611 }, { "epoch": 0.37807506635552307, "grad_norm": 1.493305634293778, "learning_rate": 1.429117543211054e-05, "loss": 0.6345, "step": 10612 }, { "epoch": 0.3781106934820172, "grad_norm": 1.9130624152972056, "learning_rate": 1.4290133111773794e-05, "loss": 0.5419, "step": 10613 }, { "epoch": 0.37814632060851133, "grad_norm": 1.1048167378143514, "learning_rate": 1.4289090734310929e-05, "loss": 0.4522, "step": 10614 }, { "epoch": 0.3781819477350054, "grad_norm": 1.5848938209027397, "learning_rate": 1.4288048299735821e-05, "loss": 0.5229, "step": 10615 }, { "epoch": 0.37821757486149954, "grad_norm": 1.1087984744619614, "learning_rate": 1.4287005808062355e-05, "loss": 0.4931, "step": 10616 }, { "epoch": 0.37825320198799367, "grad_norm": 8.168037775325471, "learning_rate": 1.4285963259304413e-05, "loss": 0.6415, "step": 10617 }, { "epoch": 0.37828882911448775, "grad_norm": 1.1018788009803178, "learning_rate": 1.4284920653475867e-05, "loss": 0.5113, "step": 10618 }, { "epoch": 0.3783244562409819, "grad_norm": 1.0462755359883138, "learning_rate": 1.4283877990590613e-05, "loss": 0.4649, "step": 10619 }, { "epoch": 0.378360083367476, "grad_norm": 1.7146361765496658, "learning_rate": 1.4282835270662528e-05, "loss": 0.5571, "step": 10620 }, { "epoch": 0.3783957104939701, "grad_norm": 1.9361071622095067, "learning_rate": 1.4281792493705497e-05, "loss": 0.5738, "step": 10621 }, { "epoch": 0.3784313376204642, "grad_norm": 1.0773675274293706, "learning_rate": 1.4280749659733406e-05, "loss": 0.514, "step": 10622 }, { "epoch": 0.37846696474695835, "grad_norm": 0.9911873135478084, "learning_rate": 1.4279706768760142e-05, "loss": 0.4951, "step": 10623 }, { "epoch": 0.3785025918734524, "grad_norm": 1.6255594700114062, "learning_rate": 1.4278663820799591e-05, "loss": 0.5653, "step": 10624 }, { "epoch": 0.37853821899994655, "grad_norm": 1.6323445685159594, "learning_rate": 1.427762081586564e-05, "loss": 0.5598, "step": 10625 }, { "epoch": 0.3785738461264407, "grad_norm": 1.076482005573468, "learning_rate": 1.4276577753972179e-05, "loss": 0.4504, "step": 10626 }, { "epoch": 0.37860947325293476, "grad_norm": 1.0342074330264255, "learning_rate": 1.4275534635133097e-05, "loss": 0.4665, "step": 10627 }, { "epoch": 0.3786451003794289, "grad_norm": 1.6332287941417853, "learning_rate": 1.427449145936228e-05, "loss": 0.5315, "step": 10628 }, { "epoch": 0.378680727505923, "grad_norm": 1.2400059359616291, "learning_rate": 1.4273448226673622e-05, "loss": 0.4758, "step": 10629 }, { "epoch": 0.3787163546324171, "grad_norm": 1.0921131535266695, "learning_rate": 1.4272404937081017e-05, "loss": 0.4604, "step": 10630 }, { "epoch": 0.3787519817589112, "grad_norm": 1.9019875755111373, "learning_rate": 1.4271361590598351e-05, "loss": 0.5805, "step": 10631 }, { "epoch": 0.37878760888540536, "grad_norm": 1.9552431192416506, "learning_rate": 1.4270318187239522e-05, "loss": 0.6326, "step": 10632 }, { "epoch": 0.37882323601189943, "grad_norm": 2.3851328325822334, "learning_rate": 1.4269274727018419e-05, "loss": 0.5875, "step": 10633 }, { "epoch": 0.37885886313839356, "grad_norm": 1.377493526682482, "learning_rate": 1.4268231209948942e-05, "loss": 0.6168, "step": 10634 }, { "epoch": 0.3788944902648877, "grad_norm": 1.1975510300020096, "learning_rate": 1.426718763604498e-05, "loss": 0.46, "step": 10635 }, { "epoch": 0.37893011739138177, "grad_norm": 1.9535117353031484, "learning_rate": 1.4266144005320435e-05, "loss": 0.6142, "step": 10636 }, { "epoch": 0.3789657445178759, "grad_norm": 1.2228554331295367, "learning_rate": 1.4265100317789198e-05, "loss": 0.4977, "step": 10637 }, { "epoch": 0.37900137164437003, "grad_norm": 1.6137088296992637, "learning_rate": 1.426405657346517e-05, "loss": 0.5384, "step": 10638 }, { "epoch": 0.3790369987708641, "grad_norm": 1.023885243827007, "learning_rate": 1.426301277236225e-05, "loss": 0.4514, "step": 10639 }, { "epoch": 0.37907262589735824, "grad_norm": 1.6159139856709108, "learning_rate": 1.4261968914494335e-05, "loss": 0.5544, "step": 10640 }, { "epoch": 0.37910825302385237, "grad_norm": 1.1498311109976203, "learning_rate": 1.4260924999875323e-05, "loss": 0.465, "step": 10641 }, { "epoch": 0.37914388015034645, "grad_norm": 1.6421034381500634, "learning_rate": 1.4259881028519118e-05, "loss": 0.5961, "step": 10642 }, { "epoch": 0.3791795072768406, "grad_norm": 1.079175120886337, "learning_rate": 1.4258837000439618e-05, "loss": 0.4816, "step": 10643 }, { "epoch": 0.3792151344033347, "grad_norm": 1.1917115296234, "learning_rate": 1.4257792915650728e-05, "loss": 0.5464, "step": 10644 }, { "epoch": 0.37925076152982884, "grad_norm": 1.080620834725279, "learning_rate": 1.4256748774166348e-05, "loss": 0.5073, "step": 10645 }, { "epoch": 0.3792863886563229, "grad_norm": 2.233049904510721, "learning_rate": 1.4255704576000387e-05, "loss": 0.5991, "step": 10646 }, { "epoch": 0.37932201578281705, "grad_norm": 1.5537666197956799, "learning_rate": 1.425466032116674e-05, "loss": 0.4952, "step": 10647 }, { "epoch": 0.3793576429093112, "grad_norm": 1.216726997991475, "learning_rate": 1.4253616009679317e-05, "loss": 0.458, "step": 10648 }, { "epoch": 0.37939327003580525, "grad_norm": 1.0977264105861515, "learning_rate": 1.4252571641552026e-05, "loss": 0.4914, "step": 10649 }, { "epoch": 0.3794288971622994, "grad_norm": 1.6589481168105564, "learning_rate": 1.425152721679877e-05, "loss": 0.5677, "step": 10650 }, { "epoch": 0.3794645242887935, "grad_norm": 1.6142043596192661, "learning_rate": 1.4250482735433459e-05, "loss": 0.55, "step": 10651 }, { "epoch": 0.3795001514152876, "grad_norm": 1.0172923930760382, "learning_rate": 1.4249438197469997e-05, "loss": 0.4521, "step": 10652 }, { "epoch": 0.3795357785417817, "grad_norm": 1.4680004378237268, "learning_rate": 1.4248393602922299e-05, "loss": 0.5653, "step": 10653 }, { "epoch": 0.37957140566827585, "grad_norm": 1.2152002601173015, "learning_rate": 1.424734895180427e-05, "loss": 0.4885, "step": 10654 }, { "epoch": 0.3796070327947699, "grad_norm": 1.8503263025895436, "learning_rate": 1.4246304244129819e-05, "loss": 0.5522, "step": 10655 }, { "epoch": 0.37964265992126406, "grad_norm": 1.9438458531859186, "learning_rate": 1.4245259479912859e-05, "loss": 0.4461, "step": 10656 }, { "epoch": 0.3796782870477582, "grad_norm": 1.4219762170933457, "learning_rate": 1.4244214659167304e-05, "loss": 0.4836, "step": 10657 }, { "epoch": 0.37971391417425226, "grad_norm": 1.5049895758738485, "learning_rate": 1.4243169781907062e-05, "loss": 0.6263, "step": 10658 }, { "epoch": 0.3797495413007464, "grad_norm": 2.214355130567507, "learning_rate": 1.4242124848146048e-05, "loss": 0.6017, "step": 10659 }, { "epoch": 0.3797851684272405, "grad_norm": 1.4072979321781052, "learning_rate": 1.4241079857898177e-05, "loss": 0.5971, "step": 10660 }, { "epoch": 0.3798207955537346, "grad_norm": 1.0517384047987806, "learning_rate": 1.4240034811177364e-05, "loss": 0.4455, "step": 10661 }, { "epoch": 0.37985642268022873, "grad_norm": 1.523199546774759, "learning_rate": 1.4238989707997526e-05, "loss": 0.5427, "step": 10662 }, { "epoch": 0.37989204980672286, "grad_norm": 1.7272075784072447, "learning_rate": 1.4237944548372572e-05, "loss": 0.5141, "step": 10663 }, { "epoch": 0.37992767693321694, "grad_norm": 1.1280107219635434, "learning_rate": 1.4236899332316427e-05, "loss": 0.4648, "step": 10664 }, { "epoch": 0.37996330405971107, "grad_norm": 1.8406342661826935, "learning_rate": 1.4235854059843008e-05, "loss": 0.5784, "step": 10665 }, { "epoch": 0.3799989311862052, "grad_norm": 1.9258769988173645, "learning_rate": 1.4234808730966231e-05, "loss": 0.5503, "step": 10666 }, { "epoch": 0.3800345583126993, "grad_norm": 1.270694896255668, "learning_rate": 1.4233763345700014e-05, "loss": 0.5223, "step": 10667 }, { "epoch": 0.3800701854391934, "grad_norm": 1.1253510625823346, "learning_rate": 1.423271790405828e-05, "loss": 0.5166, "step": 10668 }, { "epoch": 0.38010581256568754, "grad_norm": 1.6210670350748606, "learning_rate": 1.4231672406054948e-05, "loss": 0.5785, "step": 10669 }, { "epoch": 0.3801414396921816, "grad_norm": 1.4359645839684287, "learning_rate": 1.423062685170394e-05, "loss": 0.4802, "step": 10670 }, { "epoch": 0.38017706681867575, "grad_norm": 1.630245305427693, "learning_rate": 1.4229581241019176e-05, "loss": 0.5113, "step": 10671 }, { "epoch": 0.3802126939451699, "grad_norm": 1.8066823521573327, "learning_rate": 1.4228535574014586e-05, "loss": 0.551, "step": 10672 }, { "epoch": 0.38024832107166395, "grad_norm": 1.1029160794857273, "learning_rate": 1.4227489850704088e-05, "loss": 0.4779, "step": 10673 }, { "epoch": 0.3802839481981581, "grad_norm": 1.2011246370337476, "learning_rate": 1.4226444071101606e-05, "loss": 0.5143, "step": 10674 }, { "epoch": 0.3803195753246522, "grad_norm": 1.079753488055748, "learning_rate": 1.4225398235221069e-05, "loss": 0.475, "step": 10675 }, { "epoch": 0.3803552024511463, "grad_norm": 1.7191800261531578, "learning_rate": 1.42243523430764e-05, "loss": 0.6177, "step": 10676 }, { "epoch": 0.3803908295776404, "grad_norm": 1.529398407389858, "learning_rate": 1.4223306394681529e-05, "loss": 0.6117, "step": 10677 }, { "epoch": 0.38042645670413455, "grad_norm": 1.0901450466567657, "learning_rate": 1.4222260390050378e-05, "loss": 0.515, "step": 10678 }, { "epoch": 0.38046208383062863, "grad_norm": 1.8743562021449598, "learning_rate": 1.4221214329196882e-05, "loss": 0.5881, "step": 10679 }, { "epoch": 0.38049771095712276, "grad_norm": 1.2510196571052075, "learning_rate": 1.4220168212134966e-05, "loss": 0.5342, "step": 10680 }, { "epoch": 0.3805333380836169, "grad_norm": 1.2058899660853857, "learning_rate": 1.4219122038878557e-05, "loss": 0.54, "step": 10681 }, { "epoch": 0.38056896521011097, "grad_norm": 1.2305266836957298, "learning_rate": 1.4218075809441591e-05, "loss": 0.5291, "step": 10682 }, { "epoch": 0.3806045923366051, "grad_norm": 1.0410236004422897, "learning_rate": 1.4217029523837998e-05, "loss": 0.4358, "step": 10683 }, { "epoch": 0.3806402194630992, "grad_norm": 1.2188594060860822, "learning_rate": 1.421598318208171e-05, "loss": 0.5155, "step": 10684 }, { "epoch": 0.3806758465895933, "grad_norm": 1.1647536710797157, "learning_rate": 1.4214936784186657e-05, "loss": 0.4597, "step": 10685 }, { "epoch": 0.38071147371608743, "grad_norm": 1.734718764981496, "learning_rate": 1.4213890330166777e-05, "loss": 0.5354, "step": 10686 }, { "epoch": 0.38074710084258157, "grad_norm": 1.1844564394848847, "learning_rate": 1.4212843820036005e-05, "loss": 0.5122, "step": 10687 }, { "epoch": 0.38078272796907564, "grad_norm": 2.0552736718448563, "learning_rate": 1.4211797253808268e-05, "loss": 0.59, "step": 10688 }, { "epoch": 0.38081835509556977, "grad_norm": 1.7840635111865677, "learning_rate": 1.4210750631497505e-05, "loss": 0.6082, "step": 10689 }, { "epoch": 0.3808539822220639, "grad_norm": 1.0828974145335772, "learning_rate": 1.420970395311766e-05, "loss": 0.4594, "step": 10690 }, { "epoch": 0.380889609348558, "grad_norm": 1.431207738035494, "learning_rate": 1.4208657218682662e-05, "loss": 0.6204, "step": 10691 }, { "epoch": 0.3809252364750521, "grad_norm": 1.809154161599889, "learning_rate": 1.420761042820645e-05, "loss": 0.6027, "step": 10692 }, { "epoch": 0.38096086360154624, "grad_norm": 1.1494322346482653, "learning_rate": 1.4206563581702966e-05, "loss": 0.5123, "step": 10693 }, { "epoch": 0.3809964907280403, "grad_norm": 1.6311529441148465, "learning_rate": 1.4205516679186144e-05, "loss": 0.4813, "step": 10694 }, { "epoch": 0.38103211785453445, "grad_norm": 1.8135694466511767, "learning_rate": 1.4204469720669934e-05, "loss": 0.4867, "step": 10695 }, { "epoch": 0.3810677449810286, "grad_norm": 1.316449685842178, "learning_rate": 1.4203422706168267e-05, "loss": 0.5862, "step": 10696 }, { "epoch": 0.38110337210752265, "grad_norm": 1.9500974706641556, "learning_rate": 1.4202375635695088e-05, "loss": 0.5913, "step": 10697 }, { "epoch": 0.3811389992340168, "grad_norm": 1.0789068593740558, "learning_rate": 1.420132850926434e-05, "loss": 0.4804, "step": 10698 }, { "epoch": 0.3811746263605109, "grad_norm": 2.0306191699893, "learning_rate": 1.420028132688997e-05, "loss": 0.6149, "step": 10699 }, { "epoch": 0.381210253487005, "grad_norm": 1.4914718587580318, "learning_rate": 1.4199234088585915e-05, "loss": 0.632, "step": 10700 }, { "epoch": 0.3812458806134991, "grad_norm": 1.5415231594359606, "learning_rate": 1.4198186794366125e-05, "loss": 0.4947, "step": 10701 }, { "epoch": 0.38128150773999325, "grad_norm": 1.0981589793497517, "learning_rate": 1.4197139444244544e-05, "loss": 0.4681, "step": 10702 }, { "epoch": 0.38131713486648733, "grad_norm": 1.9718030777586393, "learning_rate": 1.4196092038235116e-05, "loss": 0.6011, "step": 10703 }, { "epoch": 0.38135276199298146, "grad_norm": 1.837246985474578, "learning_rate": 1.419504457635179e-05, "loss": 0.5542, "step": 10704 }, { "epoch": 0.3813883891194756, "grad_norm": 1.9546225757207987, "learning_rate": 1.419399705860851e-05, "loss": 0.5489, "step": 10705 }, { "epoch": 0.38142401624596967, "grad_norm": 1.7288209681568318, "learning_rate": 1.4192949485019231e-05, "loss": 0.5992, "step": 10706 }, { "epoch": 0.3814596433724638, "grad_norm": 1.5878992578374806, "learning_rate": 1.4191901855597902e-05, "loss": 0.5624, "step": 10707 }, { "epoch": 0.38149527049895793, "grad_norm": 1.1261794302875243, "learning_rate": 1.4190854170358464e-05, "loss": 0.5464, "step": 10708 }, { "epoch": 0.381530897625452, "grad_norm": 1.5767304463907927, "learning_rate": 1.4189806429314876e-05, "loss": 0.5109, "step": 10709 }, { "epoch": 0.38156652475194613, "grad_norm": 1.2433635904127742, "learning_rate": 1.418875863248109e-05, "loss": 0.5186, "step": 10710 }, { "epoch": 0.38160215187844027, "grad_norm": 1.2242150924428021, "learning_rate": 1.418771077987105e-05, "loss": 0.5091, "step": 10711 }, { "epoch": 0.38163777900493434, "grad_norm": 1.5306433121684249, "learning_rate": 1.4186662871498718e-05, "loss": 0.5646, "step": 10712 }, { "epoch": 0.38167340613142847, "grad_norm": 1.2160311386294946, "learning_rate": 1.4185614907378043e-05, "loss": 0.4999, "step": 10713 }, { "epoch": 0.3817090332579226, "grad_norm": 1.912559058864747, "learning_rate": 1.4184566887522977e-05, "loss": 0.5574, "step": 10714 }, { "epoch": 0.3817446603844167, "grad_norm": 0.9838044098554605, "learning_rate": 1.4183518811947479e-05, "loss": 0.4263, "step": 10715 }, { "epoch": 0.3817802875109108, "grad_norm": 1.6660565075629585, "learning_rate": 1.4182470680665505e-05, "loss": 0.6063, "step": 10716 }, { "epoch": 0.38181591463740494, "grad_norm": 1.530220775412725, "learning_rate": 1.4181422493691008e-05, "loss": 0.6997, "step": 10717 }, { "epoch": 0.381851541763899, "grad_norm": 1.9210973868076637, "learning_rate": 1.4180374251037951e-05, "loss": 0.5538, "step": 10718 }, { "epoch": 0.38188716889039315, "grad_norm": 1.142078404088568, "learning_rate": 1.4179325952720287e-05, "loss": 0.5046, "step": 10719 }, { "epoch": 0.3819227960168873, "grad_norm": 2.739539625404271, "learning_rate": 1.417827759875198e-05, "loss": 0.6247, "step": 10720 }, { "epoch": 0.38195842314338135, "grad_norm": 1.1367194353431904, "learning_rate": 1.4177229189146983e-05, "loss": 0.5189, "step": 10721 }, { "epoch": 0.3819940502698755, "grad_norm": 1.4690335327320427, "learning_rate": 1.4176180723919262e-05, "loss": 0.5978, "step": 10722 }, { "epoch": 0.3820296773963696, "grad_norm": 1.218632485009184, "learning_rate": 1.4175132203082773e-05, "loss": 0.5168, "step": 10723 }, { "epoch": 0.3820653045228637, "grad_norm": 1.422605542665699, "learning_rate": 1.4174083626651482e-05, "loss": 0.4769, "step": 10724 }, { "epoch": 0.3821009316493578, "grad_norm": 1.1995778062855857, "learning_rate": 1.4173034994639349e-05, "loss": 0.4788, "step": 10725 }, { "epoch": 0.38213655877585195, "grad_norm": 2.0495052228922073, "learning_rate": 1.4171986307060339e-05, "loss": 0.6283, "step": 10726 }, { "epoch": 0.38217218590234603, "grad_norm": 1.1537663613118159, "learning_rate": 1.4170937563928417e-05, "loss": 0.4668, "step": 10727 }, { "epoch": 0.38220781302884016, "grad_norm": 1.8183780228931261, "learning_rate": 1.4169888765257542e-05, "loss": 0.5599, "step": 10728 }, { "epoch": 0.3822434401553343, "grad_norm": 1.3631056705387858, "learning_rate": 1.4168839911061686e-05, "loss": 0.6086, "step": 10729 }, { "epoch": 0.38227906728182837, "grad_norm": 1.738592027569394, "learning_rate": 1.4167791001354811e-05, "loss": 0.5666, "step": 10730 }, { "epoch": 0.3823146944083225, "grad_norm": 1.695059586244692, "learning_rate": 1.416674203615089e-05, "loss": 0.5074, "step": 10731 }, { "epoch": 0.38235032153481663, "grad_norm": 1.507846341892776, "learning_rate": 1.4165693015463884e-05, "loss": 0.6496, "step": 10732 }, { "epoch": 0.3823859486613107, "grad_norm": 1.4428441321741785, "learning_rate": 1.4164643939307765e-05, "loss": 0.5906, "step": 10733 }, { "epoch": 0.38242157578780483, "grad_norm": 1.834578970907149, "learning_rate": 1.4163594807696496e-05, "loss": 0.5833, "step": 10734 }, { "epoch": 0.38245720291429897, "grad_norm": 1.1811469109643644, "learning_rate": 1.4162545620644058e-05, "loss": 0.4757, "step": 10735 }, { "epoch": 0.38249283004079304, "grad_norm": 1.3325945150083736, "learning_rate": 1.4161496378164414e-05, "loss": 0.5948, "step": 10736 }, { "epoch": 0.3825284571672872, "grad_norm": 1.148349631493309, "learning_rate": 1.4160447080271533e-05, "loss": 0.4457, "step": 10737 }, { "epoch": 0.3825640842937813, "grad_norm": 1.1006840013982635, "learning_rate": 1.4159397726979395e-05, "loss": 0.4787, "step": 10738 }, { "epoch": 0.3825997114202754, "grad_norm": 1.0498921716253802, "learning_rate": 1.415834831830197e-05, "loss": 0.4986, "step": 10739 }, { "epoch": 0.3826353385467695, "grad_norm": 1.0152302727175884, "learning_rate": 1.4157298854253228e-05, "loss": 0.468, "step": 10740 }, { "epoch": 0.38267096567326364, "grad_norm": 1.179281992289907, "learning_rate": 1.4156249334847148e-05, "loss": 0.4597, "step": 10741 }, { "epoch": 0.3827065927997577, "grad_norm": 2.111854845682563, "learning_rate": 1.41551997600977e-05, "loss": 0.6076, "step": 10742 }, { "epoch": 0.38274221992625185, "grad_norm": 1.7269154351159333, "learning_rate": 1.4154150130018867e-05, "loss": 0.6347, "step": 10743 }, { "epoch": 0.382777847052746, "grad_norm": 1.9574678172769822, "learning_rate": 1.415310044462462e-05, "loss": 0.564, "step": 10744 }, { "epoch": 0.38281347417924005, "grad_norm": 1.159660616268322, "learning_rate": 1.4152050703928934e-05, "loss": 0.5146, "step": 10745 }, { "epoch": 0.3828491013057342, "grad_norm": 2.0932731798675333, "learning_rate": 1.4151000907945796e-05, "loss": 0.5793, "step": 10746 }, { "epoch": 0.3828847284322283, "grad_norm": 1.1701310771574058, "learning_rate": 1.4149951056689177e-05, "loss": 0.4749, "step": 10747 }, { "epoch": 0.3829203555587224, "grad_norm": 1.2636713670885042, "learning_rate": 1.414890115017306e-05, "loss": 0.4923, "step": 10748 }, { "epoch": 0.3829559826852165, "grad_norm": 1.6473452825044494, "learning_rate": 1.4147851188411423e-05, "loss": 0.6126, "step": 10749 }, { "epoch": 0.38299160981171065, "grad_norm": 1.1417553714141278, "learning_rate": 1.414680117141825e-05, "loss": 0.469, "step": 10750 }, { "epoch": 0.38302723693820473, "grad_norm": 1.1203421126466793, "learning_rate": 1.4145751099207518e-05, "loss": 0.486, "step": 10751 }, { "epoch": 0.38306286406469886, "grad_norm": 1.0435752689039568, "learning_rate": 1.4144700971793215e-05, "loss": 0.4938, "step": 10752 }, { "epoch": 0.383098491191193, "grad_norm": 0.9579788442368785, "learning_rate": 1.4143650789189321e-05, "loss": 0.4504, "step": 10753 }, { "epoch": 0.38313411831768707, "grad_norm": 1.3028363213864744, "learning_rate": 1.414260055140982e-05, "loss": 0.4882, "step": 10754 }, { "epoch": 0.3831697454441812, "grad_norm": 1.3161505252186916, "learning_rate": 1.4141550258468697e-05, "loss": 0.6536, "step": 10755 }, { "epoch": 0.38320537257067533, "grad_norm": 1.1185979243880475, "learning_rate": 1.4140499910379936e-05, "loss": 0.5093, "step": 10756 }, { "epoch": 0.3832409996971694, "grad_norm": 1.0108475078998103, "learning_rate": 1.4139449507157528e-05, "loss": 0.4651, "step": 10757 }, { "epoch": 0.38327662682366354, "grad_norm": 1.6824848693798944, "learning_rate": 1.4138399048815456e-05, "loss": 0.5395, "step": 10758 }, { "epoch": 0.38331225395015767, "grad_norm": 1.4206261744589528, "learning_rate": 1.413734853536771e-05, "loss": 0.5758, "step": 10759 }, { "epoch": 0.38334788107665174, "grad_norm": 1.5319008128863154, "learning_rate": 1.4136297966828272e-05, "loss": 0.5503, "step": 10760 }, { "epoch": 0.3833835082031459, "grad_norm": 1.7342740530695442, "learning_rate": 1.4135247343211138e-05, "loss": 0.625, "step": 10761 }, { "epoch": 0.38341913532964, "grad_norm": 1.3388018077729158, "learning_rate": 1.4134196664530295e-05, "loss": 0.63, "step": 10762 }, { "epoch": 0.3834547624561341, "grad_norm": 1.5284351824583515, "learning_rate": 1.4133145930799732e-05, "loss": 0.5901, "step": 10763 }, { "epoch": 0.3834903895826282, "grad_norm": 1.0738243465206543, "learning_rate": 1.4132095142033442e-05, "loss": 0.4315, "step": 10764 }, { "epoch": 0.38352601670912234, "grad_norm": 2.076049415089399, "learning_rate": 1.413104429824542e-05, "loss": 0.5558, "step": 10765 }, { "epoch": 0.3835616438356164, "grad_norm": 1.1052578068443504, "learning_rate": 1.4129993399449653e-05, "loss": 0.4979, "step": 10766 }, { "epoch": 0.38359727096211055, "grad_norm": 1.8377442041975975, "learning_rate": 1.4128942445660139e-05, "loss": 0.5783, "step": 10767 }, { "epoch": 0.3836328980886047, "grad_norm": 1.811961664282942, "learning_rate": 1.4127891436890867e-05, "loss": 0.6389, "step": 10768 }, { "epoch": 0.38366852521509875, "grad_norm": 1.150610535036012, "learning_rate": 1.4126840373155841e-05, "loss": 0.472, "step": 10769 }, { "epoch": 0.3837041523415929, "grad_norm": 2.3845380335202684, "learning_rate": 1.412578925446905e-05, "loss": 0.5951, "step": 10770 }, { "epoch": 0.383739779468087, "grad_norm": 1.5803588398085036, "learning_rate": 1.4124738080844486e-05, "loss": 0.517, "step": 10771 }, { "epoch": 0.3837754065945811, "grad_norm": 1.0167560378557294, "learning_rate": 1.4123686852296154e-05, "loss": 0.4813, "step": 10772 }, { "epoch": 0.3838110337210752, "grad_norm": 1.0678849812324056, "learning_rate": 1.4122635568838052e-05, "loss": 0.4717, "step": 10773 }, { "epoch": 0.38384666084756935, "grad_norm": 0.9757410610895577, "learning_rate": 1.4121584230484172e-05, "loss": 0.4493, "step": 10774 }, { "epoch": 0.38388228797406343, "grad_norm": 1.0913884308557058, "learning_rate": 1.4120532837248518e-05, "loss": 0.4757, "step": 10775 }, { "epoch": 0.38391791510055756, "grad_norm": 2.1435147172098894, "learning_rate": 1.411948138914509e-05, "loss": 0.6074, "step": 10776 }, { "epoch": 0.3839535422270517, "grad_norm": 0.9531430742815473, "learning_rate": 1.411842988618789e-05, "loss": 0.5251, "step": 10777 }, { "epoch": 0.38398916935354577, "grad_norm": 1.0241978207602744, "learning_rate": 1.4117378328390916e-05, "loss": 0.398, "step": 10778 }, { "epoch": 0.3840247964800399, "grad_norm": 1.0554851689772766, "learning_rate": 1.4116326715768169e-05, "loss": 0.4281, "step": 10779 }, { "epoch": 0.38406042360653403, "grad_norm": 1.6665462170586167, "learning_rate": 1.4115275048333658e-05, "loss": 0.6201, "step": 10780 }, { "epoch": 0.3840960507330281, "grad_norm": 0.9856027768222129, "learning_rate": 1.4114223326101384e-05, "loss": 0.4742, "step": 10781 }, { "epoch": 0.38413167785952224, "grad_norm": 1.0347905509816693, "learning_rate": 1.4113171549085347e-05, "loss": 0.478, "step": 10782 }, { "epoch": 0.38416730498601637, "grad_norm": 2.16644914962831, "learning_rate": 1.411211971729956e-05, "loss": 0.615, "step": 10783 }, { "epoch": 0.38420293211251044, "grad_norm": 1.1034202512566151, "learning_rate": 1.4111067830758023e-05, "loss": 0.5067, "step": 10784 }, { "epoch": 0.3842385592390046, "grad_norm": 1.1424526802261503, "learning_rate": 1.4110015889474744e-05, "loss": 0.4606, "step": 10785 }, { "epoch": 0.3842741863654987, "grad_norm": 1.7308866187092304, "learning_rate": 1.4108963893463728e-05, "loss": 0.598, "step": 10786 }, { "epoch": 0.3843098134919928, "grad_norm": 1.2547777648319245, "learning_rate": 1.4107911842738988e-05, "loss": 0.5181, "step": 10787 }, { "epoch": 0.3843454406184869, "grad_norm": 1.84991836136405, "learning_rate": 1.4106859737314532e-05, "loss": 0.6341, "step": 10788 }, { "epoch": 0.38438106774498104, "grad_norm": 1.1474573052347978, "learning_rate": 1.4105807577204367e-05, "loss": 0.5211, "step": 10789 }, { "epoch": 0.3844166948714751, "grad_norm": 1.749562424935409, "learning_rate": 1.4104755362422502e-05, "loss": 0.5853, "step": 10790 }, { "epoch": 0.38445232199796925, "grad_norm": 1.4492803510232721, "learning_rate": 1.4103703092982954e-05, "loss": 0.5759, "step": 10791 }, { "epoch": 0.3844879491244634, "grad_norm": 1.5127093182095932, "learning_rate": 1.4102650768899731e-05, "loss": 0.5851, "step": 10792 }, { "epoch": 0.38452357625095746, "grad_norm": 1.160762052473242, "learning_rate": 1.4101598390186842e-05, "loss": 0.439, "step": 10793 }, { "epoch": 0.3845592033774516, "grad_norm": 1.003957216664168, "learning_rate": 1.4100545956858307e-05, "loss": 0.4901, "step": 10794 }, { "epoch": 0.3845948305039457, "grad_norm": 2.208038308082999, "learning_rate": 1.4099493468928136e-05, "loss": 0.4634, "step": 10795 }, { "epoch": 0.3846304576304398, "grad_norm": 1.0541217773918614, "learning_rate": 1.4098440926410344e-05, "loss": 0.4745, "step": 10796 }, { "epoch": 0.3846660847569339, "grad_norm": 2.4208384617946757, "learning_rate": 1.4097388329318945e-05, "loss": 0.6002, "step": 10797 }, { "epoch": 0.38470171188342805, "grad_norm": 1.2668287359629993, "learning_rate": 1.4096335677667954e-05, "loss": 0.4883, "step": 10798 }, { "epoch": 0.38473733900992213, "grad_norm": 1.130645285044077, "learning_rate": 1.4095282971471396e-05, "loss": 0.5499, "step": 10799 }, { "epoch": 0.38477296613641626, "grad_norm": 1.125220707335384, "learning_rate": 1.409423021074328e-05, "loss": 0.4296, "step": 10800 }, { "epoch": 0.3848085932629104, "grad_norm": 1.8968760498675514, "learning_rate": 1.4093177395497626e-05, "loss": 0.5774, "step": 10801 }, { "epoch": 0.38484422038940447, "grad_norm": 2.072596468546404, "learning_rate": 1.4092124525748457e-05, "loss": 0.5852, "step": 10802 }, { "epoch": 0.3848798475158986, "grad_norm": 1.5430018299691446, "learning_rate": 1.4091071601509789e-05, "loss": 0.5087, "step": 10803 }, { "epoch": 0.38491547464239273, "grad_norm": 1.8897502951414344, "learning_rate": 1.4090018622795643e-05, "loss": 0.5766, "step": 10804 }, { "epoch": 0.3849511017688868, "grad_norm": 1.8646102692612598, "learning_rate": 1.4088965589620039e-05, "loss": 0.5588, "step": 10805 }, { "epoch": 0.38498672889538094, "grad_norm": 1.69096237482438, "learning_rate": 1.4087912501996999e-05, "loss": 0.5803, "step": 10806 }, { "epoch": 0.38502235602187507, "grad_norm": 1.2027528974694783, "learning_rate": 1.408685935994055e-05, "loss": 0.5539, "step": 10807 }, { "epoch": 0.38505798314836914, "grad_norm": 1.0336404733946396, "learning_rate": 1.4085806163464709e-05, "loss": 0.4446, "step": 10808 }, { "epoch": 0.3850936102748633, "grad_norm": 1.6267061387592912, "learning_rate": 1.4084752912583504e-05, "loss": 0.5965, "step": 10809 }, { "epoch": 0.3851292374013574, "grad_norm": 1.4272519830432588, "learning_rate": 1.4083699607310959e-05, "loss": 0.5913, "step": 10810 }, { "epoch": 0.3851648645278515, "grad_norm": 1.0249957621521888, "learning_rate": 1.4082646247661101e-05, "loss": 0.4763, "step": 10811 }, { "epoch": 0.3852004916543456, "grad_norm": 1.4293391659030434, "learning_rate": 1.4081592833647952e-05, "loss": 0.5361, "step": 10812 }, { "epoch": 0.38523611878083974, "grad_norm": 1.2907264115269261, "learning_rate": 1.4080539365285544e-05, "loss": 0.5508, "step": 10813 }, { "epoch": 0.3852717459073338, "grad_norm": 0.9788092919211355, "learning_rate": 1.4079485842587902e-05, "loss": 0.4264, "step": 10814 }, { "epoch": 0.38530737303382795, "grad_norm": 1.103574149886749, "learning_rate": 1.4078432265569056e-05, "loss": 0.5167, "step": 10815 }, { "epoch": 0.3853430001603221, "grad_norm": 1.1114854548418285, "learning_rate": 1.407737863424303e-05, "loss": 0.4708, "step": 10816 }, { "epoch": 0.38537862728681616, "grad_norm": 1.0357072497311828, "learning_rate": 1.4076324948623858e-05, "loss": 0.4976, "step": 10817 }, { "epoch": 0.3854142544133103, "grad_norm": 1.7734273639878435, "learning_rate": 1.4075271208725572e-05, "loss": 0.6049, "step": 10818 }, { "epoch": 0.3854498815398044, "grad_norm": 1.0105558487281665, "learning_rate": 1.4074217414562198e-05, "loss": 0.4545, "step": 10819 }, { "epoch": 0.3854855086662985, "grad_norm": 0.9852397012878437, "learning_rate": 1.4073163566147775e-05, "loss": 0.4483, "step": 10820 }, { "epoch": 0.3855211357927926, "grad_norm": 1.444065425977309, "learning_rate": 1.4072109663496332e-05, "loss": 0.5954, "step": 10821 }, { "epoch": 0.38555676291928676, "grad_norm": 1.6383016784758975, "learning_rate": 1.4071055706621902e-05, "loss": 0.664, "step": 10822 }, { "epoch": 0.38559239004578083, "grad_norm": 1.225450786844644, "learning_rate": 1.407000169553852e-05, "loss": 0.538, "step": 10823 }, { "epoch": 0.38562801717227496, "grad_norm": 1.1257787480502357, "learning_rate": 1.406894763026022e-05, "loss": 0.5547, "step": 10824 }, { "epoch": 0.3856636442987691, "grad_norm": 1.1021126815810418, "learning_rate": 1.4067893510801039e-05, "loss": 0.4607, "step": 10825 }, { "epoch": 0.38569927142526317, "grad_norm": 2.3410325419531595, "learning_rate": 1.4066839337175013e-05, "loss": 0.6015, "step": 10826 }, { "epoch": 0.3857348985517573, "grad_norm": 1.1669980312133565, "learning_rate": 1.4065785109396174e-05, "loss": 0.5138, "step": 10827 }, { "epoch": 0.38577052567825143, "grad_norm": 1.70152077386412, "learning_rate": 1.4064730827478571e-05, "loss": 0.5721, "step": 10828 }, { "epoch": 0.3858061528047455, "grad_norm": 0.9995496446447875, "learning_rate": 1.4063676491436234e-05, "loss": 0.4356, "step": 10829 }, { "epoch": 0.38584177993123964, "grad_norm": 1.8983886705238275, "learning_rate": 1.4062622101283206e-05, "loss": 0.5437, "step": 10830 }, { "epoch": 0.38587740705773377, "grad_norm": 1.0073476996954889, "learning_rate": 1.4061567657033519e-05, "loss": 0.4214, "step": 10831 }, { "epoch": 0.3859130341842279, "grad_norm": 1.3487797038606821, "learning_rate": 1.4060513158701224e-05, "loss": 0.5945, "step": 10832 }, { "epoch": 0.385948661310722, "grad_norm": 1.1427976693754012, "learning_rate": 1.4059458606300358e-05, "loss": 0.4893, "step": 10833 }, { "epoch": 0.3859842884372161, "grad_norm": 1.0444920182176636, "learning_rate": 1.4058403999844963e-05, "loss": 0.4601, "step": 10834 }, { "epoch": 0.38601991556371024, "grad_norm": 1.6936210536890552, "learning_rate": 1.405734933934908e-05, "loss": 0.5528, "step": 10835 }, { "epoch": 0.3860555426902043, "grad_norm": 1.570006906561053, "learning_rate": 1.4056294624826755e-05, "loss": 0.6427, "step": 10836 }, { "epoch": 0.38609116981669844, "grad_norm": 1.2204686338043054, "learning_rate": 1.4055239856292036e-05, "loss": 0.5513, "step": 10837 }, { "epoch": 0.3861267969431926, "grad_norm": 1.5404025576333313, "learning_rate": 1.405418503375896e-05, "loss": 0.4867, "step": 10838 }, { "epoch": 0.38616242406968665, "grad_norm": 1.0327089171262767, "learning_rate": 1.4053130157241578e-05, "loss": 0.4601, "step": 10839 }, { "epoch": 0.3861980511961808, "grad_norm": 1.1671988326225398, "learning_rate": 1.4052075226753936e-05, "loss": 0.5157, "step": 10840 }, { "epoch": 0.3862336783226749, "grad_norm": 1.6088421498021683, "learning_rate": 1.4051020242310081e-05, "loss": 0.5566, "step": 10841 }, { "epoch": 0.386269305449169, "grad_norm": 1.6227426283699589, "learning_rate": 1.4049965203924055e-05, "loss": 0.5477, "step": 10842 }, { "epoch": 0.3863049325756631, "grad_norm": 1.4034685622692462, "learning_rate": 1.4048910111609914e-05, "loss": 0.5682, "step": 10843 }, { "epoch": 0.38634055970215725, "grad_norm": 1.4769832852333196, "learning_rate": 1.4047854965381706e-05, "loss": 0.6357, "step": 10844 }, { "epoch": 0.3863761868286513, "grad_norm": 1.6694129127171453, "learning_rate": 1.4046799765253481e-05, "loss": 0.6503, "step": 10845 }, { "epoch": 0.38641181395514546, "grad_norm": 1.1321877183553164, "learning_rate": 1.4045744511239284e-05, "loss": 0.5017, "step": 10846 }, { "epoch": 0.3864474410816396, "grad_norm": 1.0276366635588348, "learning_rate": 1.4044689203353174e-05, "loss": 0.4833, "step": 10847 }, { "epoch": 0.38648306820813366, "grad_norm": 1.1009134872394528, "learning_rate": 1.4043633841609201e-05, "loss": 0.4811, "step": 10848 }, { "epoch": 0.3865186953346278, "grad_norm": 1.426293205650119, "learning_rate": 1.4042578426021418e-05, "loss": 0.5319, "step": 10849 }, { "epoch": 0.3865543224611219, "grad_norm": 1.0175830933932106, "learning_rate": 1.4041522956603872e-05, "loss": 0.522, "step": 10850 }, { "epoch": 0.386589949587616, "grad_norm": 1.2463627422901042, "learning_rate": 1.4040467433370628e-05, "loss": 0.4494, "step": 10851 }, { "epoch": 0.38662557671411013, "grad_norm": 1.0578814738736773, "learning_rate": 1.4039411856335736e-05, "loss": 0.5242, "step": 10852 }, { "epoch": 0.38666120384060426, "grad_norm": 1.6164413288391815, "learning_rate": 1.403835622551325e-05, "loss": 0.6515, "step": 10853 }, { "epoch": 0.38669683096709834, "grad_norm": 1.1293177438301258, "learning_rate": 1.403730054091723e-05, "loss": 0.4611, "step": 10854 }, { "epoch": 0.38673245809359247, "grad_norm": 1.4152103986127258, "learning_rate": 1.4036244802561727e-05, "loss": 0.6031, "step": 10855 }, { "epoch": 0.3867680852200866, "grad_norm": 1.145746740210612, "learning_rate": 1.4035189010460808e-05, "loss": 0.4784, "step": 10856 }, { "epoch": 0.3868037123465807, "grad_norm": 1.6092994210304463, "learning_rate": 1.4034133164628523e-05, "loss": 0.5899, "step": 10857 }, { "epoch": 0.3868393394730748, "grad_norm": 1.1013610910576035, "learning_rate": 1.4033077265078939e-05, "loss": 0.4946, "step": 10858 }, { "epoch": 0.38687496659956894, "grad_norm": 1.1051843822491503, "learning_rate": 1.4032021311826112e-05, "loss": 0.4885, "step": 10859 }, { "epoch": 0.386910593726063, "grad_norm": 1.010512707459338, "learning_rate": 1.4030965304884101e-05, "loss": 0.4539, "step": 10860 }, { "epoch": 0.38694622085255714, "grad_norm": 1.5567659390702713, "learning_rate": 1.402990924426697e-05, "loss": 0.5432, "step": 10861 }, { "epoch": 0.3869818479790513, "grad_norm": 1.0137742731460477, "learning_rate": 1.402885312998878e-05, "loss": 0.4792, "step": 10862 }, { "epoch": 0.38701747510554535, "grad_norm": 1.3016287115570366, "learning_rate": 1.4027796962063598e-05, "loss": 0.4861, "step": 10863 }, { "epoch": 0.3870531022320395, "grad_norm": 1.0968709623625137, "learning_rate": 1.402674074050548e-05, "loss": 0.5012, "step": 10864 }, { "epoch": 0.3870887293585336, "grad_norm": 1.2563288804901251, "learning_rate": 1.4025684465328495e-05, "loss": 0.5002, "step": 10865 }, { "epoch": 0.3871243564850277, "grad_norm": 1.1035876294881328, "learning_rate": 1.4024628136546711e-05, "loss": 0.4556, "step": 10866 }, { "epoch": 0.3871599836115218, "grad_norm": 1.6847105375282019, "learning_rate": 1.4023571754174189e-05, "loss": 0.6397, "step": 10867 }, { "epoch": 0.38719561073801595, "grad_norm": 1.2048119141976923, "learning_rate": 1.4022515318224992e-05, "loss": 0.471, "step": 10868 }, { "epoch": 0.38723123786451, "grad_norm": 1.1029149709152868, "learning_rate": 1.4021458828713196e-05, "loss": 0.5103, "step": 10869 }, { "epoch": 0.38726686499100416, "grad_norm": 1.5575070915885885, "learning_rate": 1.4020402285652865e-05, "loss": 0.618, "step": 10870 }, { "epoch": 0.3873024921174983, "grad_norm": 1.720114491491186, "learning_rate": 1.4019345689058068e-05, "loss": 0.4684, "step": 10871 }, { "epoch": 0.38733811924399236, "grad_norm": 0.9938529164240444, "learning_rate": 1.4018289038942871e-05, "loss": 0.4363, "step": 10872 }, { "epoch": 0.3873737463704865, "grad_norm": 1.3682291468151773, "learning_rate": 1.4017232335321351e-05, "loss": 0.5804, "step": 10873 }, { "epoch": 0.3874093734969806, "grad_norm": 1.4216098498315561, "learning_rate": 1.4016175578207573e-05, "loss": 0.5773, "step": 10874 }, { "epoch": 0.3874450006234747, "grad_norm": 1.1418397485525456, "learning_rate": 1.4015118767615607e-05, "loss": 0.4975, "step": 10875 }, { "epoch": 0.38748062774996883, "grad_norm": 1.6606128265014033, "learning_rate": 1.4014061903559532e-05, "loss": 0.7527, "step": 10876 }, { "epoch": 0.38751625487646296, "grad_norm": 1.5874018612521512, "learning_rate": 1.4013004986053416e-05, "loss": 0.5399, "step": 10877 }, { "epoch": 0.38755188200295704, "grad_norm": 1.300547664310909, "learning_rate": 1.4011948015111334e-05, "loss": 0.5432, "step": 10878 }, { "epoch": 0.38758750912945117, "grad_norm": 1.100961216092543, "learning_rate": 1.401089099074736e-05, "loss": 0.4841, "step": 10879 }, { "epoch": 0.3876231362559453, "grad_norm": 1.034469426671951, "learning_rate": 1.4009833912975566e-05, "loss": 0.4032, "step": 10880 }, { "epoch": 0.3876587633824394, "grad_norm": 1.6649563060840151, "learning_rate": 1.4008776781810035e-05, "loss": 0.6049, "step": 10881 }, { "epoch": 0.3876943905089335, "grad_norm": 1.1104680914907172, "learning_rate": 1.4007719597264837e-05, "loss": 0.4582, "step": 10882 }, { "epoch": 0.38773001763542764, "grad_norm": 1.1765648512995435, "learning_rate": 1.400666235935405e-05, "loss": 0.5431, "step": 10883 }, { "epoch": 0.3877656447619217, "grad_norm": 1.035087722072529, "learning_rate": 1.4005605068091756e-05, "loss": 0.4927, "step": 10884 }, { "epoch": 0.38780127188841584, "grad_norm": 1.428006883370146, "learning_rate": 1.400454772349203e-05, "loss": 0.5444, "step": 10885 }, { "epoch": 0.38783689901491, "grad_norm": 1.0163701855637346, "learning_rate": 1.4003490325568953e-05, "loss": 0.4875, "step": 10886 }, { "epoch": 0.38787252614140405, "grad_norm": 1.3505496575536062, "learning_rate": 1.40024328743366e-05, "loss": 0.4928, "step": 10887 }, { "epoch": 0.3879081532678982, "grad_norm": 1.8031662658291199, "learning_rate": 1.400137536980906e-05, "loss": 0.5788, "step": 10888 }, { "epoch": 0.3879437803943923, "grad_norm": 1.4043419606030083, "learning_rate": 1.400031781200041e-05, "loss": 0.4932, "step": 10889 }, { "epoch": 0.3879794075208864, "grad_norm": 1.1183229653435438, "learning_rate": 1.3999260200924728e-05, "loss": 0.4926, "step": 10890 }, { "epoch": 0.3880150346473805, "grad_norm": 1.4647860212447654, "learning_rate": 1.3998202536596104e-05, "loss": 0.6493, "step": 10891 }, { "epoch": 0.38805066177387465, "grad_norm": 1.1368033535582378, "learning_rate": 1.3997144819028619e-05, "loss": 0.4727, "step": 10892 }, { "epoch": 0.3880862889003687, "grad_norm": 1.3529840698116882, "learning_rate": 1.3996087048236357e-05, "loss": 0.5, "step": 10893 }, { "epoch": 0.38812191602686286, "grad_norm": 3.3064432790252467, "learning_rate": 1.39950292242334e-05, "loss": 0.5563, "step": 10894 }, { "epoch": 0.388157543153357, "grad_norm": 2.0505167816549186, "learning_rate": 1.3993971347033841e-05, "loss": 0.5847, "step": 10895 }, { "epoch": 0.38819317027985106, "grad_norm": 1.5149758664484232, "learning_rate": 1.3992913416651761e-05, "loss": 0.575, "step": 10896 }, { "epoch": 0.3882287974063452, "grad_norm": 1.3923961740325497, "learning_rate": 1.3991855433101247e-05, "loss": 0.554, "step": 10897 }, { "epoch": 0.3882644245328393, "grad_norm": 1.1639690499312914, "learning_rate": 1.3990797396396385e-05, "loss": 0.5306, "step": 10898 }, { "epoch": 0.3883000516593334, "grad_norm": 1.0932655901700017, "learning_rate": 1.398973930655127e-05, "loss": 0.5102, "step": 10899 }, { "epoch": 0.38833567878582753, "grad_norm": 1.5103081703947405, "learning_rate": 1.3988681163579987e-05, "loss": 0.6199, "step": 10900 }, { "epoch": 0.38837130591232166, "grad_norm": 1.0181893037682799, "learning_rate": 1.3987622967496623e-05, "loss": 0.4818, "step": 10901 }, { "epoch": 0.38840693303881574, "grad_norm": 1.2027261391456663, "learning_rate": 1.3986564718315277e-05, "loss": 0.5813, "step": 10902 }, { "epoch": 0.38844256016530987, "grad_norm": 1.4579865718387652, "learning_rate": 1.3985506416050032e-05, "loss": 0.5651, "step": 10903 }, { "epoch": 0.388478187291804, "grad_norm": 1.6785325391097594, "learning_rate": 1.3984448060714985e-05, "loss": 0.5978, "step": 10904 }, { "epoch": 0.3885138144182981, "grad_norm": 1.159012858352794, "learning_rate": 1.3983389652324229e-05, "loss": 0.5171, "step": 10905 }, { "epoch": 0.3885494415447922, "grad_norm": 0.9547085050287016, "learning_rate": 1.3982331190891851e-05, "loss": 0.4423, "step": 10906 }, { "epoch": 0.38858506867128634, "grad_norm": 1.1151543553109393, "learning_rate": 1.3981272676431951e-05, "loss": 0.4977, "step": 10907 }, { "epoch": 0.3886206957977804, "grad_norm": 1.098673280812468, "learning_rate": 1.3980214108958626e-05, "loss": 0.4174, "step": 10908 }, { "epoch": 0.38865632292427454, "grad_norm": 1.002321766659649, "learning_rate": 1.3979155488485963e-05, "loss": 0.5013, "step": 10909 }, { "epoch": 0.3886919500507687, "grad_norm": 1.6432914967858023, "learning_rate": 1.3978096815028066e-05, "loss": 0.5784, "step": 10910 }, { "epoch": 0.38872757717726275, "grad_norm": 1.1878692406021343, "learning_rate": 1.397703808859903e-05, "loss": 0.4875, "step": 10911 }, { "epoch": 0.3887632043037569, "grad_norm": 1.103331125929365, "learning_rate": 1.3975979309212952e-05, "loss": 0.4887, "step": 10912 }, { "epoch": 0.388798831430251, "grad_norm": 1.9855400225457664, "learning_rate": 1.3974920476883927e-05, "loss": 0.5777, "step": 10913 }, { "epoch": 0.3888344585567451, "grad_norm": 2.8173452807309864, "learning_rate": 1.3973861591626061e-05, "loss": 0.6026, "step": 10914 }, { "epoch": 0.3888700856832392, "grad_norm": 1.411257123535137, "learning_rate": 1.397280265345345e-05, "loss": 0.5914, "step": 10915 }, { "epoch": 0.38890571280973335, "grad_norm": 1.139885378695043, "learning_rate": 1.3971743662380195e-05, "loss": 0.5095, "step": 10916 }, { "epoch": 0.3889413399362274, "grad_norm": 1.5470015163841349, "learning_rate": 1.3970684618420396e-05, "loss": 0.6397, "step": 10917 }, { "epoch": 0.38897696706272156, "grad_norm": 1.022736546526688, "learning_rate": 1.3969625521588157e-05, "loss": 0.4907, "step": 10918 }, { "epoch": 0.3890125941892157, "grad_norm": 1.5888682903652416, "learning_rate": 1.396856637189758e-05, "loss": 0.6792, "step": 10919 }, { "epoch": 0.38904822131570976, "grad_norm": 1.1712935734330032, "learning_rate": 1.3967507169362766e-05, "loss": 0.4693, "step": 10920 }, { "epoch": 0.3890838484422039, "grad_norm": 1.3663751152293924, "learning_rate": 1.3966447913997823e-05, "loss": 0.5939, "step": 10921 }, { "epoch": 0.389119475568698, "grad_norm": 1.6099498127080947, "learning_rate": 1.3965388605816852e-05, "loss": 0.5416, "step": 10922 }, { "epoch": 0.3891551026951921, "grad_norm": 1.4725690998070744, "learning_rate": 1.396432924483396e-05, "loss": 0.6235, "step": 10923 }, { "epoch": 0.38919072982168623, "grad_norm": 1.5304558396143644, "learning_rate": 1.3963269831063255e-05, "loss": 0.5761, "step": 10924 }, { "epoch": 0.38922635694818036, "grad_norm": 1.1836601963747226, "learning_rate": 1.3962210364518837e-05, "loss": 0.4848, "step": 10925 }, { "epoch": 0.38926198407467444, "grad_norm": 1.2132071641549478, "learning_rate": 1.3961150845214824e-05, "loss": 0.4808, "step": 10926 }, { "epoch": 0.38929761120116857, "grad_norm": 1.3081001149696336, "learning_rate": 1.396009127316532e-05, "loss": 0.6585, "step": 10927 }, { "epoch": 0.3893332383276627, "grad_norm": 2.0976521923004294, "learning_rate": 1.3959031648384427e-05, "loss": 0.5525, "step": 10928 }, { "epoch": 0.3893688654541568, "grad_norm": 2.9084421588060136, "learning_rate": 1.3957971970886264e-05, "loss": 0.5427, "step": 10929 }, { "epoch": 0.3894044925806509, "grad_norm": 1.476838636390714, "learning_rate": 1.3956912240684938e-05, "loss": 0.6166, "step": 10930 }, { "epoch": 0.38944011970714504, "grad_norm": 1.0221764526554702, "learning_rate": 1.395585245779456e-05, "loss": 0.4637, "step": 10931 }, { "epoch": 0.3894757468336391, "grad_norm": 1.457437688932523, "learning_rate": 1.3954792622229238e-05, "loss": 0.4819, "step": 10932 }, { "epoch": 0.38951137396013324, "grad_norm": 1.3324002685316387, "learning_rate": 1.3953732734003094e-05, "loss": 0.5665, "step": 10933 }, { "epoch": 0.3895470010866274, "grad_norm": 1.1390692914300216, "learning_rate": 1.3952672793130231e-05, "loss": 0.4793, "step": 10934 }, { "epoch": 0.38958262821312145, "grad_norm": 1.4707474500755815, "learning_rate": 1.3951612799624767e-05, "loss": 0.5238, "step": 10935 }, { "epoch": 0.3896182553396156, "grad_norm": 2.0079104224270536, "learning_rate": 1.3950552753500817e-05, "loss": 0.62, "step": 10936 }, { "epoch": 0.3896538824661097, "grad_norm": 1.015170598357447, "learning_rate": 1.3949492654772496e-05, "loss": 0.4862, "step": 10937 }, { "epoch": 0.3896895095926038, "grad_norm": 1.028643227689615, "learning_rate": 1.394843250345392e-05, "loss": 0.4604, "step": 10938 }, { "epoch": 0.3897251367190979, "grad_norm": 1.006892290235529, "learning_rate": 1.3947372299559204e-05, "loss": 0.4542, "step": 10939 }, { "epoch": 0.38976076384559205, "grad_norm": 1.0863820229272232, "learning_rate": 1.3946312043102468e-05, "loss": 0.5326, "step": 10940 }, { "epoch": 0.3897963909720861, "grad_norm": 1.0835162028783767, "learning_rate": 1.3945251734097829e-05, "loss": 0.4923, "step": 10941 }, { "epoch": 0.38983201809858026, "grad_norm": 1.5810821768500427, "learning_rate": 1.3944191372559408e-05, "loss": 0.5326, "step": 10942 }, { "epoch": 0.3898676452250744, "grad_norm": 1.06851000680786, "learning_rate": 1.3943130958501318e-05, "loss": 0.4556, "step": 10943 }, { "epoch": 0.38990327235156846, "grad_norm": 1.0973480165614975, "learning_rate": 1.3942070491937685e-05, "loss": 0.5215, "step": 10944 }, { "epoch": 0.3899388994780626, "grad_norm": 1.128368445739927, "learning_rate": 1.3941009972882628e-05, "loss": 0.4798, "step": 10945 }, { "epoch": 0.3899745266045567, "grad_norm": 1.4158151901735898, "learning_rate": 1.3939949401350268e-05, "loss": 0.5841, "step": 10946 }, { "epoch": 0.3900101537310508, "grad_norm": 1.6154960792679292, "learning_rate": 1.3938888777354728e-05, "loss": 0.6067, "step": 10947 }, { "epoch": 0.39004578085754493, "grad_norm": 1.0203903545429174, "learning_rate": 1.3937828100910129e-05, "loss": 0.4844, "step": 10948 }, { "epoch": 0.39008140798403906, "grad_norm": 1.3530497207811538, "learning_rate": 1.3936767372030602e-05, "loss": 0.6413, "step": 10949 }, { "epoch": 0.39011703511053314, "grad_norm": 1.1107019764727142, "learning_rate": 1.393570659073026e-05, "loss": 0.5066, "step": 10950 }, { "epoch": 0.39015266223702727, "grad_norm": 1.6357033524334816, "learning_rate": 1.3934645757023235e-05, "loss": 0.5844, "step": 10951 }, { "epoch": 0.3901882893635214, "grad_norm": 1.0286984913602466, "learning_rate": 1.3933584870923655e-05, "loss": 0.4884, "step": 10952 }, { "epoch": 0.3902239164900155, "grad_norm": 1.6938052485632586, "learning_rate": 1.3932523932445641e-05, "loss": 0.5491, "step": 10953 }, { "epoch": 0.3902595436165096, "grad_norm": 1.0223710127175332, "learning_rate": 1.3931462941603322e-05, "loss": 0.4962, "step": 10954 }, { "epoch": 0.39029517074300374, "grad_norm": 0.9947477197547079, "learning_rate": 1.3930401898410826e-05, "loss": 0.444, "step": 10955 }, { "epoch": 0.3903307978694978, "grad_norm": 2.7618584536806408, "learning_rate": 1.3929340802882283e-05, "loss": 0.5534, "step": 10956 }, { "epoch": 0.39036642499599195, "grad_norm": 1.0269565105755927, "learning_rate": 1.3928279655031818e-05, "loss": 0.4629, "step": 10957 }, { "epoch": 0.3904020521224861, "grad_norm": 1.396701346856746, "learning_rate": 1.3927218454873566e-05, "loss": 0.6024, "step": 10958 }, { "epoch": 0.39043767924898015, "grad_norm": 1.776847248725234, "learning_rate": 1.3926157202421653e-05, "loss": 0.562, "step": 10959 }, { "epoch": 0.3904733063754743, "grad_norm": 1.1334762975218935, "learning_rate": 1.3925095897690213e-05, "loss": 0.5277, "step": 10960 }, { "epoch": 0.3905089335019684, "grad_norm": 2.139547072985769, "learning_rate": 1.392403454069338e-05, "loss": 0.623, "step": 10961 }, { "epoch": 0.3905445606284625, "grad_norm": 1.1727201371001086, "learning_rate": 1.3922973131445281e-05, "loss": 0.462, "step": 10962 }, { "epoch": 0.3905801877549566, "grad_norm": 1.0844584745718473, "learning_rate": 1.3921911669960055e-05, "loss": 0.5171, "step": 10963 }, { "epoch": 0.39061581488145075, "grad_norm": 1.4224395549835251, "learning_rate": 1.3920850156251835e-05, "loss": 0.5567, "step": 10964 }, { "epoch": 0.3906514420079448, "grad_norm": 1.384372714516423, "learning_rate": 1.3919788590334751e-05, "loss": 0.6095, "step": 10965 }, { "epoch": 0.39068706913443896, "grad_norm": 1.044737180266692, "learning_rate": 1.3918726972222947e-05, "loss": 0.473, "step": 10966 }, { "epoch": 0.3907226962609331, "grad_norm": 1.8442348439620686, "learning_rate": 1.3917665301930552e-05, "loss": 0.532, "step": 10967 }, { "epoch": 0.39075832338742716, "grad_norm": 1.3753732909032979, "learning_rate": 1.3916603579471705e-05, "loss": 0.5006, "step": 10968 }, { "epoch": 0.3907939505139213, "grad_norm": 1.0732306527829563, "learning_rate": 1.3915541804860542e-05, "loss": 0.5451, "step": 10969 }, { "epoch": 0.3908295776404154, "grad_norm": 1.7127497062393557, "learning_rate": 1.3914479978111206e-05, "loss": 0.5438, "step": 10970 }, { "epoch": 0.3908652047669095, "grad_norm": 1.475090875726495, "learning_rate": 1.391341809923783e-05, "loss": 0.4799, "step": 10971 }, { "epoch": 0.39090083189340363, "grad_norm": 1.4731030086900565, "learning_rate": 1.3912356168254559e-05, "loss": 0.5765, "step": 10972 }, { "epoch": 0.39093645901989776, "grad_norm": 1.6604633564058404, "learning_rate": 1.391129418517553e-05, "loss": 0.5819, "step": 10973 }, { "epoch": 0.39097208614639184, "grad_norm": 1.378002068612045, "learning_rate": 1.3910232150014887e-05, "loss": 0.5969, "step": 10974 }, { "epoch": 0.39100771327288597, "grad_norm": 1.0339111856958585, "learning_rate": 1.3909170062786768e-05, "loss": 0.4916, "step": 10975 }, { "epoch": 0.3910433403993801, "grad_norm": 1.0943521180134204, "learning_rate": 1.3908107923505317e-05, "loss": 0.4983, "step": 10976 }, { "epoch": 0.3910789675258742, "grad_norm": 1.57912726208701, "learning_rate": 1.3907045732184679e-05, "loss": 0.5405, "step": 10977 }, { "epoch": 0.3911145946523683, "grad_norm": 1.0847166792518708, "learning_rate": 1.3905983488838996e-05, "loss": 0.5267, "step": 10978 }, { "epoch": 0.39115022177886244, "grad_norm": 1.500010242709478, "learning_rate": 1.3904921193482412e-05, "loss": 0.5864, "step": 10979 }, { "epoch": 0.3911858489053565, "grad_norm": 1.4136012086681933, "learning_rate": 1.3903858846129073e-05, "loss": 0.5145, "step": 10980 }, { "epoch": 0.39122147603185065, "grad_norm": 1.7407605159628436, "learning_rate": 1.3902796446793125e-05, "loss": 0.5447, "step": 10981 }, { "epoch": 0.3912571031583448, "grad_norm": 1.7653743375471334, "learning_rate": 1.3901733995488713e-05, "loss": 0.566, "step": 10982 }, { "epoch": 0.39129273028483885, "grad_norm": 1.6110639994431968, "learning_rate": 1.3900671492229986e-05, "loss": 0.6295, "step": 10983 }, { "epoch": 0.391328357411333, "grad_norm": 1.1942370123904593, "learning_rate": 1.3899608937031092e-05, "loss": 0.507, "step": 10984 }, { "epoch": 0.3913639845378271, "grad_norm": 1.156201625785536, "learning_rate": 1.3898546329906182e-05, "loss": 0.4925, "step": 10985 }, { "epoch": 0.3913996116643212, "grad_norm": 1.0481227824043857, "learning_rate": 1.38974836708694e-05, "loss": 0.4865, "step": 10986 }, { "epoch": 0.3914352387908153, "grad_norm": 1.1156117549996944, "learning_rate": 1.38964209599349e-05, "loss": 0.4703, "step": 10987 }, { "epoch": 0.39147086591730945, "grad_norm": 0.9924236093564609, "learning_rate": 1.3895358197116831e-05, "loss": 0.4377, "step": 10988 }, { "epoch": 0.3915064930438035, "grad_norm": 1.027093527340595, "learning_rate": 1.3894295382429346e-05, "loss": 0.5084, "step": 10989 }, { "epoch": 0.39154212017029766, "grad_norm": 1.4836537608807827, "learning_rate": 1.3893232515886594e-05, "loss": 0.602, "step": 10990 }, { "epoch": 0.3915777472967918, "grad_norm": 1.490943357050645, "learning_rate": 1.3892169597502729e-05, "loss": 0.5612, "step": 10991 }, { "epoch": 0.39161337442328586, "grad_norm": 1.0579642410776713, "learning_rate": 1.3891106627291907e-05, "loss": 0.437, "step": 10992 }, { "epoch": 0.39164900154978, "grad_norm": 1.436028219213733, "learning_rate": 1.3890043605268284e-05, "loss": 0.482, "step": 10993 }, { "epoch": 0.3916846286762741, "grad_norm": 1.100490848275293, "learning_rate": 1.3888980531446005e-05, "loss": 0.4367, "step": 10994 }, { "epoch": 0.3917202558027682, "grad_norm": 2.235093647675421, "learning_rate": 1.3887917405839236e-05, "loss": 0.6196, "step": 10995 }, { "epoch": 0.39175588292926233, "grad_norm": 1.128274579653732, "learning_rate": 1.3886854228462128e-05, "loss": 0.4633, "step": 10996 }, { "epoch": 0.39179151005575646, "grad_norm": 1.6534404573535915, "learning_rate": 1.3885790999328839e-05, "loss": 0.5459, "step": 10997 }, { "epoch": 0.39182713718225054, "grad_norm": 1.8101243187189502, "learning_rate": 1.3884727718453527e-05, "loss": 0.6478, "step": 10998 }, { "epoch": 0.39186276430874467, "grad_norm": 1.1759229871305263, "learning_rate": 1.388366438585035e-05, "loss": 0.452, "step": 10999 }, { "epoch": 0.3918983914352388, "grad_norm": 1.6518455933435217, "learning_rate": 1.388260100153347e-05, "loss": 0.5341, "step": 11000 }, { "epoch": 0.3919340185617329, "grad_norm": 1.0497800036345486, "learning_rate": 1.388153756551704e-05, "loss": 0.4028, "step": 11001 }, { "epoch": 0.391969645688227, "grad_norm": 1.8673813408487052, "learning_rate": 1.3880474077815227e-05, "loss": 0.5219, "step": 11002 }, { "epoch": 0.39200527281472114, "grad_norm": 1.4418495524426522, "learning_rate": 1.3879410538442185e-05, "loss": 0.5448, "step": 11003 }, { "epoch": 0.3920408999412152, "grad_norm": 1.4073559599677792, "learning_rate": 1.3878346947412087e-05, "loss": 0.5895, "step": 11004 }, { "epoch": 0.39207652706770935, "grad_norm": 1.7361555452654283, "learning_rate": 1.3877283304739085e-05, "loss": 0.6266, "step": 11005 }, { "epoch": 0.3921121541942035, "grad_norm": 1.77369506416366, "learning_rate": 1.3876219610437344e-05, "loss": 0.521, "step": 11006 }, { "epoch": 0.39214778132069755, "grad_norm": 2.0139290340018428, "learning_rate": 1.3875155864521031e-05, "loss": 0.6817, "step": 11007 }, { "epoch": 0.3921834084471917, "grad_norm": 1.1426878628535762, "learning_rate": 1.387409206700431e-05, "loss": 0.4902, "step": 11008 }, { "epoch": 0.3922190355736858, "grad_norm": 1.8656966624007303, "learning_rate": 1.3873028217901346e-05, "loss": 0.5195, "step": 11009 }, { "epoch": 0.3922546627001799, "grad_norm": 1.0690872606689854, "learning_rate": 1.38719643172263e-05, "loss": 0.4296, "step": 11010 }, { "epoch": 0.392290289826674, "grad_norm": 1.2282574614614403, "learning_rate": 1.3870900364993347e-05, "loss": 0.4947, "step": 11011 }, { "epoch": 0.39232591695316815, "grad_norm": 1.2644918322035612, "learning_rate": 1.3869836361216649e-05, "loss": 0.4757, "step": 11012 }, { "epoch": 0.39236154407966223, "grad_norm": 1.6232334263478772, "learning_rate": 1.3868772305910376e-05, "loss": 0.6259, "step": 11013 }, { "epoch": 0.39239717120615636, "grad_norm": 1.6120285442369535, "learning_rate": 1.3867708199088692e-05, "loss": 0.586, "step": 11014 }, { "epoch": 0.3924327983326505, "grad_norm": 1.2034192403173507, "learning_rate": 1.3866644040765774e-05, "loss": 0.5205, "step": 11015 }, { "epoch": 0.39246842545914457, "grad_norm": 1.1787150416295795, "learning_rate": 1.3865579830955787e-05, "loss": 0.4399, "step": 11016 }, { "epoch": 0.3925040525856387, "grad_norm": 1.2803915707019509, "learning_rate": 1.3864515569672898e-05, "loss": 0.4955, "step": 11017 }, { "epoch": 0.3925396797121328, "grad_norm": 1.119775098300907, "learning_rate": 1.3863451256931286e-05, "loss": 0.4576, "step": 11018 }, { "epoch": 0.3925753068386269, "grad_norm": 1.1602968919750665, "learning_rate": 1.3862386892745123e-05, "loss": 0.5297, "step": 11019 }, { "epoch": 0.39261093396512103, "grad_norm": 1.53248489134117, "learning_rate": 1.3861322477128578e-05, "loss": 0.5427, "step": 11020 }, { "epoch": 0.39264656109161516, "grad_norm": 1.7745053688331103, "learning_rate": 1.3860258010095821e-05, "loss": 0.4948, "step": 11021 }, { "epoch": 0.3926821882181093, "grad_norm": 1.4863737514791504, "learning_rate": 1.3859193491661037e-05, "loss": 0.6183, "step": 11022 }, { "epoch": 0.39271781534460337, "grad_norm": 0.9837590101023993, "learning_rate": 1.385812892183839e-05, "loss": 0.4504, "step": 11023 }, { "epoch": 0.3927534424710975, "grad_norm": 1.1789925269257562, "learning_rate": 1.385706430064206e-05, "loss": 0.4319, "step": 11024 }, { "epoch": 0.39278906959759163, "grad_norm": 1.1561807140038909, "learning_rate": 1.3855999628086223e-05, "loss": 0.5055, "step": 11025 }, { "epoch": 0.3928246967240857, "grad_norm": 1.5622836620675642, "learning_rate": 1.3854934904185059e-05, "loss": 0.4794, "step": 11026 }, { "epoch": 0.39286032385057984, "grad_norm": 1.6561694964557243, "learning_rate": 1.385387012895274e-05, "loss": 0.6617, "step": 11027 }, { "epoch": 0.39289595097707397, "grad_norm": 1.4651383899396626, "learning_rate": 1.3852805302403445e-05, "loss": 0.4956, "step": 11028 }, { "epoch": 0.39293157810356805, "grad_norm": 1.2871255963917914, "learning_rate": 1.3851740424551352e-05, "loss": 0.4802, "step": 11029 }, { "epoch": 0.3929672052300622, "grad_norm": 1.2717453578768274, "learning_rate": 1.3850675495410647e-05, "loss": 0.6315, "step": 11030 }, { "epoch": 0.3930028323565563, "grad_norm": 1.1311177569168687, "learning_rate": 1.3849610514995508e-05, "loss": 0.4721, "step": 11031 }, { "epoch": 0.3930384594830504, "grad_norm": 1.7325540028574402, "learning_rate": 1.3848545483320111e-05, "loss": 0.5996, "step": 11032 }, { "epoch": 0.3930740866095445, "grad_norm": 1.398251306784118, "learning_rate": 1.3847480400398643e-05, "loss": 0.4867, "step": 11033 }, { "epoch": 0.39310971373603865, "grad_norm": 1.2109983358840342, "learning_rate": 1.3846415266245284e-05, "loss": 0.4879, "step": 11034 }, { "epoch": 0.3931453408625327, "grad_norm": 1.216367052386313, "learning_rate": 1.3845350080874217e-05, "loss": 0.4935, "step": 11035 }, { "epoch": 0.39318096798902685, "grad_norm": 1.1139770661289528, "learning_rate": 1.3844284844299625e-05, "loss": 0.4815, "step": 11036 }, { "epoch": 0.393216595115521, "grad_norm": 0.9930237622857887, "learning_rate": 1.3843219556535693e-05, "loss": 0.428, "step": 11037 }, { "epoch": 0.39325222224201506, "grad_norm": 1.1396622551430737, "learning_rate": 1.3842154217596608e-05, "loss": 0.4899, "step": 11038 }, { "epoch": 0.3932878493685092, "grad_norm": 1.0761115414658327, "learning_rate": 1.3841088827496553e-05, "loss": 0.4676, "step": 11039 }, { "epoch": 0.3933234764950033, "grad_norm": 1.1334491645450488, "learning_rate": 1.3840023386249715e-05, "loss": 0.4672, "step": 11040 }, { "epoch": 0.3933591036214974, "grad_norm": 1.5222715770222965, "learning_rate": 1.3838957893870281e-05, "loss": 0.4828, "step": 11041 }, { "epoch": 0.39339473074799153, "grad_norm": 1.673446036975152, "learning_rate": 1.3837892350372444e-05, "loss": 0.6021, "step": 11042 }, { "epoch": 0.39343035787448566, "grad_norm": 1.0568635163069027, "learning_rate": 1.3836826755770386e-05, "loss": 0.4136, "step": 11043 }, { "epoch": 0.39346598500097973, "grad_norm": 1.4445033648412111, "learning_rate": 1.3835761110078293e-05, "loss": 0.613, "step": 11044 }, { "epoch": 0.39350161212747387, "grad_norm": 1.8958666270223155, "learning_rate": 1.3834695413310363e-05, "loss": 0.5455, "step": 11045 }, { "epoch": 0.393537239253968, "grad_norm": 1.225483123132106, "learning_rate": 1.3833629665480785e-05, "loss": 0.519, "step": 11046 }, { "epoch": 0.39357286638046207, "grad_norm": 1.0069874233613187, "learning_rate": 1.3832563866603748e-05, "loss": 0.4783, "step": 11047 }, { "epoch": 0.3936084935069562, "grad_norm": 1.5098823110535453, "learning_rate": 1.3831498016693443e-05, "loss": 0.5323, "step": 11048 }, { "epoch": 0.39364412063345033, "grad_norm": 2.3156600185772875, "learning_rate": 1.3830432115764066e-05, "loss": 0.5648, "step": 11049 }, { "epoch": 0.3936797477599444, "grad_norm": 1.4579178177966357, "learning_rate": 1.3829366163829808e-05, "loss": 0.6014, "step": 11050 }, { "epoch": 0.39371537488643854, "grad_norm": 1.11608805436236, "learning_rate": 1.3828300160904856e-05, "loss": 0.4249, "step": 11051 }, { "epoch": 0.39375100201293267, "grad_norm": 1.1959617923043422, "learning_rate": 1.3827234107003417e-05, "loss": 0.5205, "step": 11052 }, { "epoch": 0.39378662913942675, "grad_norm": 1.1210068698327775, "learning_rate": 1.3826168002139682e-05, "loss": 0.4704, "step": 11053 }, { "epoch": 0.3938222562659209, "grad_norm": 1.1226837866785777, "learning_rate": 1.3825101846327847e-05, "loss": 0.4914, "step": 11054 }, { "epoch": 0.393857883392415, "grad_norm": 1.6706484623021685, "learning_rate": 1.3824035639582101e-05, "loss": 0.5301, "step": 11055 }, { "epoch": 0.3938935105189091, "grad_norm": 1.3529424502528622, "learning_rate": 1.382296938191665e-05, "loss": 0.5439, "step": 11056 }, { "epoch": 0.3939291376454032, "grad_norm": 1.0169526880953508, "learning_rate": 1.3821903073345695e-05, "loss": 0.4674, "step": 11057 }, { "epoch": 0.39396476477189735, "grad_norm": 1.4908348039685915, "learning_rate": 1.3820836713883424e-05, "loss": 0.6451, "step": 11058 }, { "epoch": 0.3940003918983914, "grad_norm": 1.1009173917906794, "learning_rate": 1.3819770303544042e-05, "loss": 0.4848, "step": 11059 }, { "epoch": 0.39403601902488555, "grad_norm": 1.0518918160181479, "learning_rate": 1.381870384234175e-05, "loss": 0.4914, "step": 11060 }, { "epoch": 0.3940716461513797, "grad_norm": 1.7436573363597843, "learning_rate": 1.3817637330290748e-05, "loss": 0.6367, "step": 11061 }, { "epoch": 0.39410727327787376, "grad_norm": 1.591076260416089, "learning_rate": 1.3816570767405234e-05, "loss": 0.596, "step": 11062 }, { "epoch": 0.3941429004043679, "grad_norm": 1.194281731188833, "learning_rate": 1.381550415369941e-05, "loss": 0.4771, "step": 11063 }, { "epoch": 0.394178527530862, "grad_norm": 1.562820854311581, "learning_rate": 1.3814437489187486e-05, "loss": 0.5105, "step": 11064 }, { "epoch": 0.3942141546573561, "grad_norm": 1.0385928713776051, "learning_rate": 1.3813370773883662e-05, "loss": 0.4601, "step": 11065 }, { "epoch": 0.39424978178385023, "grad_norm": 2.0729466908722642, "learning_rate": 1.3812304007802135e-05, "loss": 0.4645, "step": 11066 }, { "epoch": 0.39428540891034436, "grad_norm": 1.0226904226992886, "learning_rate": 1.381123719095712e-05, "loss": 0.4859, "step": 11067 }, { "epoch": 0.39432103603683843, "grad_norm": 1.0434766720219366, "learning_rate": 1.3810170323362817e-05, "loss": 0.4678, "step": 11068 }, { "epoch": 0.39435666316333257, "grad_norm": 1.1255043506978215, "learning_rate": 1.3809103405033435e-05, "loss": 0.5372, "step": 11069 }, { "epoch": 0.3943922902898267, "grad_norm": 0.9846180808789298, "learning_rate": 1.3808036435983175e-05, "loss": 0.4561, "step": 11070 }, { "epoch": 0.3944279174163208, "grad_norm": 1.2637623784116516, "learning_rate": 1.380696941622625e-05, "loss": 0.541, "step": 11071 }, { "epoch": 0.3944635445428149, "grad_norm": 1.4367126127167629, "learning_rate": 1.3805902345776867e-05, "loss": 0.5022, "step": 11072 }, { "epoch": 0.39449917166930903, "grad_norm": 1.0473900319394387, "learning_rate": 1.380483522464923e-05, "loss": 0.516, "step": 11073 }, { "epoch": 0.3945347987958031, "grad_norm": 1.0929239802381925, "learning_rate": 1.3803768052857559e-05, "loss": 0.4867, "step": 11074 }, { "epoch": 0.39457042592229724, "grad_norm": 1.127386122520179, "learning_rate": 1.3802700830416052e-05, "loss": 0.4821, "step": 11075 }, { "epoch": 0.39460605304879137, "grad_norm": 1.5934842206918929, "learning_rate": 1.380163355733893e-05, "loss": 0.6527, "step": 11076 }, { "epoch": 0.39464168017528545, "grad_norm": 1.040019386654083, "learning_rate": 1.3800566233640396e-05, "loss": 0.4841, "step": 11077 }, { "epoch": 0.3946773073017796, "grad_norm": 1.5689730460562223, "learning_rate": 1.3799498859334668e-05, "loss": 0.525, "step": 11078 }, { "epoch": 0.3947129344282737, "grad_norm": 1.0470311880379026, "learning_rate": 1.379843143443596e-05, "loss": 0.4622, "step": 11079 }, { "epoch": 0.3947485615547678, "grad_norm": 1.0764597763788801, "learning_rate": 1.3797363958958478e-05, "loss": 0.4934, "step": 11080 }, { "epoch": 0.3947841886812619, "grad_norm": 1.8729979096442597, "learning_rate": 1.379629643291644e-05, "loss": 0.591, "step": 11081 }, { "epoch": 0.39481981580775605, "grad_norm": 1.1425655687276606, "learning_rate": 1.3795228856324065e-05, "loss": 0.4658, "step": 11082 }, { "epoch": 0.3948554429342501, "grad_norm": 1.6852271736017812, "learning_rate": 1.3794161229195566e-05, "loss": 0.5346, "step": 11083 }, { "epoch": 0.39489107006074425, "grad_norm": 1.1308001185614838, "learning_rate": 1.3793093551545154e-05, "loss": 0.4921, "step": 11084 }, { "epoch": 0.3949266971872384, "grad_norm": 1.1405073376978974, "learning_rate": 1.3792025823387055e-05, "loss": 0.448, "step": 11085 }, { "epoch": 0.39496232431373246, "grad_norm": 1.5392895981023635, "learning_rate": 1.3790958044735477e-05, "loss": 0.6411, "step": 11086 }, { "epoch": 0.3949979514402266, "grad_norm": 1.545200364209389, "learning_rate": 1.3789890215604645e-05, "loss": 0.6022, "step": 11087 }, { "epoch": 0.3950335785667207, "grad_norm": 1.9129996413607313, "learning_rate": 1.3788822336008776e-05, "loss": 0.5926, "step": 11088 }, { "epoch": 0.3950692056932148, "grad_norm": 2.713790584768429, "learning_rate": 1.3787754405962088e-05, "loss": 0.6105, "step": 11089 }, { "epoch": 0.39510483281970893, "grad_norm": 1.196402924876914, "learning_rate": 1.3786686425478806e-05, "loss": 0.4634, "step": 11090 }, { "epoch": 0.39514045994620306, "grad_norm": 1.3622248797394767, "learning_rate": 1.3785618394573144e-05, "loss": 0.6319, "step": 11091 }, { "epoch": 0.39517608707269714, "grad_norm": 2.266786805405171, "learning_rate": 1.3784550313259328e-05, "loss": 0.63, "step": 11092 }, { "epoch": 0.39521171419919127, "grad_norm": 1.254728218797818, "learning_rate": 1.378348218155158e-05, "loss": 0.5146, "step": 11093 }, { "epoch": 0.3952473413256854, "grad_norm": 1.1184149236653378, "learning_rate": 1.3782413999464122e-05, "loss": 0.4643, "step": 11094 }, { "epoch": 0.3952829684521795, "grad_norm": 1.7137212021361308, "learning_rate": 1.3781345767011178e-05, "loss": 0.5501, "step": 11095 }, { "epoch": 0.3953185955786736, "grad_norm": 1.1689901175287927, "learning_rate": 1.3780277484206972e-05, "loss": 0.4818, "step": 11096 }, { "epoch": 0.39535422270516773, "grad_norm": 1.2664736554229754, "learning_rate": 1.3779209151065729e-05, "loss": 0.5471, "step": 11097 }, { "epoch": 0.3953898498316618, "grad_norm": 1.9502999049915446, "learning_rate": 1.3778140767601673e-05, "loss": 0.4978, "step": 11098 }, { "epoch": 0.39542547695815594, "grad_norm": 1.095309307905534, "learning_rate": 1.3777072333829033e-05, "loss": 0.505, "step": 11099 }, { "epoch": 0.3954611040846501, "grad_norm": 1.1856061764175378, "learning_rate": 1.3776003849762036e-05, "loss": 0.459, "step": 11100 }, { "epoch": 0.39549673121114415, "grad_norm": 1.7573915781797371, "learning_rate": 1.3774935315414908e-05, "loss": 0.5704, "step": 11101 }, { "epoch": 0.3955323583376383, "grad_norm": 1.5763749421280875, "learning_rate": 1.3773866730801877e-05, "loss": 0.63, "step": 11102 }, { "epoch": 0.3955679854641324, "grad_norm": 1.0136218048326624, "learning_rate": 1.3772798095937172e-05, "loss": 0.4776, "step": 11103 }, { "epoch": 0.3956036125906265, "grad_norm": 1.1204903677029752, "learning_rate": 1.3771729410835027e-05, "loss": 0.4813, "step": 11104 }, { "epoch": 0.3956392397171206, "grad_norm": 1.1816354779803617, "learning_rate": 1.3770660675509666e-05, "loss": 0.4912, "step": 11105 }, { "epoch": 0.39567486684361475, "grad_norm": 1.2710556935344879, "learning_rate": 1.3769591889975322e-05, "loss": 0.5926, "step": 11106 }, { "epoch": 0.3957104939701088, "grad_norm": 1.0812755419488849, "learning_rate": 1.3768523054246228e-05, "loss": 0.4152, "step": 11107 }, { "epoch": 0.39574612109660295, "grad_norm": 2.210279899483313, "learning_rate": 1.3767454168336616e-05, "loss": 0.5965, "step": 11108 }, { "epoch": 0.3957817482230971, "grad_norm": 1.1479225316643056, "learning_rate": 1.3766385232260717e-05, "loss": 0.4902, "step": 11109 }, { "epoch": 0.39581737534959116, "grad_norm": 1.166626027826119, "learning_rate": 1.3765316246032767e-05, "loss": 0.4478, "step": 11110 }, { "epoch": 0.3958530024760853, "grad_norm": 1.7368652868775352, "learning_rate": 1.3764247209666999e-05, "loss": 0.5734, "step": 11111 }, { "epoch": 0.3958886296025794, "grad_norm": 1.3761972525348742, "learning_rate": 1.376317812317765e-05, "loss": 0.5396, "step": 11112 }, { "epoch": 0.3959242567290735, "grad_norm": 1.0639184736555134, "learning_rate": 1.3762108986578952e-05, "loss": 0.4509, "step": 11113 }, { "epoch": 0.39595988385556763, "grad_norm": 1.1277853500097965, "learning_rate": 1.3761039799885143e-05, "loss": 0.5178, "step": 11114 }, { "epoch": 0.39599551098206176, "grad_norm": 1.1266525089215924, "learning_rate": 1.3759970563110461e-05, "loss": 0.5379, "step": 11115 }, { "epoch": 0.39603113810855584, "grad_norm": 1.3785508860766458, "learning_rate": 1.3758901276269146e-05, "loss": 0.5167, "step": 11116 }, { "epoch": 0.39606676523504997, "grad_norm": 1.7593436327319594, "learning_rate": 1.375783193937543e-05, "loss": 0.6182, "step": 11117 }, { "epoch": 0.3961023923615441, "grad_norm": 0.9962708929549893, "learning_rate": 1.3756762552443555e-05, "loss": 0.463, "step": 11118 }, { "epoch": 0.3961380194880382, "grad_norm": 1.5558320977946554, "learning_rate": 1.375569311548776e-05, "loss": 0.4738, "step": 11119 }, { "epoch": 0.3961736466145323, "grad_norm": 1.2028675219592186, "learning_rate": 1.375462362852229e-05, "loss": 0.4841, "step": 11120 }, { "epoch": 0.39620927374102644, "grad_norm": 1.11170862059315, "learning_rate": 1.3753554091561379e-05, "loss": 0.4617, "step": 11121 }, { "epoch": 0.3962449008675205, "grad_norm": 1.5404485981105327, "learning_rate": 1.375248450461927e-05, "loss": 0.6079, "step": 11122 }, { "epoch": 0.39628052799401464, "grad_norm": 1.1081978226616083, "learning_rate": 1.375141486771021e-05, "loss": 0.4559, "step": 11123 }, { "epoch": 0.3963161551205088, "grad_norm": 1.5174205655823096, "learning_rate": 1.3750345180848438e-05, "loss": 0.6093, "step": 11124 }, { "epoch": 0.39635178224700285, "grad_norm": 1.0558790506160007, "learning_rate": 1.37492754440482e-05, "loss": 0.4849, "step": 11125 }, { "epoch": 0.396387409373497, "grad_norm": 1.4032625379737116, "learning_rate": 1.3748205657323737e-05, "loss": 0.6426, "step": 11126 }, { "epoch": 0.3964230364999911, "grad_norm": 1.1169192695944041, "learning_rate": 1.3747135820689294e-05, "loss": 0.4561, "step": 11127 }, { "epoch": 0.3964586636264852, "grad_norm": 1.4797478914010846, "learning_rate": 1.3746065934159123e-05, "loss": 0.6285, "step": 11128 }, { "epoch": 0.3964942907529793, "grad_norm": 1.5973226192241408, "learning_rate": 1.3744995997747462e-05, "loss": 0.5705, "step": 11129 }, { "epoch": 0.39652991787947345, "grad_norm": 1.070005036266351, "learning_rate": 1.3743926011468565e-05, "loss": 0.4825, "step": 11130 }, { "epoch": 0.3965655450059675, "grad_norm": 1.67422109565451, "learning_rate": 1.3742855975336676e-05, "loss": 0.4404, "step": 11131 }, { "epoch": 0.39660117213246165, "grad_norm": 1.1435608257646885, "learning_rate": 1.374178588936604e-05, "loss": 0.5022, "step": 11132 }, { "epoch": 0.3966367992589558, "grad_norm": 1.3138278634095173, "learning_rate": 1.374071575357091e-05, "loss": 0.589, "step": 11133 }, { "epoch": 0.39667242638544986, "grad_norm": 1.3533046050945774, "learning_rate": 1.3739645567965538e-05, "loss": 0.4059, "step": 11134 }, { "epoch": 0.396708053511944, "grad_norm": 1.0307475159456683, "learning_rate": 1.373857533256417e-05, "loss": 0.4616, "step": 11135 }, { "epoch": 0.3967436806384381, "grad_norm": 1.3777172980691885, "learning_rate": 1.3737505047381058e-05, "loss": 0.4924, "step": 11136 }, { "epoch": 0.3967793077649322, "grad_norm": 1.0814255215895399, "learning_rate": 1.3736434712430455e-05, "loss": 0.4542, "step": 11137 }, { "epoch": 0.39681493489142633, "grad_norm": 1.2581115541621817, "learning_rate": 1.3735364327726613e-05, "loss": 0.5525, "step": 11138 }, { "epoch": 0.39685056201792046, "grad_norm": 1.412565098961856, "learning_rate": 1.3734293893283783e-05, "loss": 0.586, "step": 11139 }, { "epoch": 0.39688618914441454, "grad_norm": 1.9111070347999743, "learning_rate": 1.3733223409116216e-05, "loss": 0.5687, "step": 11140 }, { "epoch": 0.39692181627090867, "grad_norm": 1.385543666997615, "learning_rate": 1.3732152875238176e-05, "loss": 0.5768, "step": 11141 }, { "epoch": 0.3969574433974028, "grad_norm": 1.3476446853302437, "learning_rate": 1.373108229166391e-05, "loss": 0.5577, "step": 11142 }, { "epoch": 0.3969930705238969, "grad_norm": 1.1885957891744174, "learning_rate": 1.3730011658407676e-05, "loss": 0.4736, "step": 11143 }, { "epoch": 0.397028697650391, "grad_norm": 1.144515170363944, "learning_rate": 1.3728940975483726e-05, "loss": 0.4835, "step": 11144 }, { "epoch": 0.39706432477688514, "grad_norm": 1.499728135205504, "learning_rate": 1.372787024290632e-05, "loss": 0.5266, "step": 11145 }, { "epoch": 0.3970999519033792, "grad_norm": 1.0552115383837017, "learning_rate": 1.3726799460689722e-05, "loss": 0.4702, "step": 11146 }, { "epoch": 0.39713557902987334, "grad_norm": 1.2150582737050286, "learning_rate": 1.3725728628848181e-05, "loss": 0.4965, "step": 11147 }, { "epoch": 0.3971712061563675, "grad_norm": 1.0859838688911563, "learning_rate": 1.3724657747395957e-05, "loss": 0.4857, "step": 11148 }, { "epoch": 0.39720683328286155, "grad_norm": 1.6863817469115512, "learning_rate": 1.3723586816347316e-05, "loss": 0.5851, "step": 11149 }, { "epoch": 0.3972424604093557, "grad_norm": 0.9627844076842015, "learning_rate": 1.3722515835716515e-05, "loss": 0.4248, "step": 11150 }, { "epoch": 0.3972780875358498, "grad_norm": 1.1790116166179017, "learning_rate": 1.372144480551781e-05, "loss": 0.5248, "step": 11151 }, { "epoch": 0.3973137146623439, "grad_norm": 1.5546801441853448, "learning_rate": 1.3720373725765466e-05, "loss": 0.5902, "step": 11152 }, { "epoch": 0.397349341788838, "grad_norm": 1.5444506213967975, "learning_rate": 1.3719302596473747e-05, "loss": 0.5591, "step": 11153 }, { "epoch": 0.39738496891533215, "grad_norm": 1.498286762728608, "learning_rate": 1.3718231417656913e-05, "loss": 0.6423, "step": 11154 }, { "epoch": 0.3974205960418262, "grad_norm": 1.2825299707961577, "learning_rate": 1.371716018932923e-05, "loss": 0.483, "step": 11155 }, { "epoch": 0.39745622316832036, "grad_norm": 1.5733165919369034, "learning_rate": 1.371608891150496e-05, "loss": 0.5918, "step": 11156 }, { "epoch": 0.3974918502948145, "grad_norm": 1.1822707979920035, "learning_rate": 1.371501758419837e-05, "loss": 0.5012, "step": 11157 }, { "epoch": 0.39752747742130856, "grad_norm": 1.5530330513211954, "learning_rate": 1.3713946207423724e-05, "loss": 0.5718, "step": 11158 }, { "epoch": 0.3975631045478027, "grad_norm": 1.2509679940928884, "learning_rate": 1.3712874781195286e-05, "loss": 0.6348, "step": 11159 }, { "epoch": 0.3975987316742968, "grad_norm": 1.3952702654697522, "learning_rate": 1.3711803305527326e-05, "loss": 0.5411, "step": 11160 }, { "epoch": 0.3976343588007909, "grad_norm": 1.481276884144756, "learning_rate": 1.3710731780434115e-05, "loss": 0.5701, "step": 11161 }, { "epoch": 0.39766998592728503, "grad_norm": 1.1507419524372084, "learning_rate": 1.3709660205929914e-05, "loss": 0.5315, "step": 11162 }, { "epoch": 0.39770561305377916, "grad_norm": 1.179410667277514, "learning_rate": 1.3708588582028993e-05, "loss": 0.5153, "step": 11163 }, { "epoch": 0.39774124018027324, "grad_norm": 1.0574936522484442, "learning_rate": 1.3707516908745624e-05, "loss": 0.4739, "step": 11164 }, { "epoch": 0.39777686730676737, "grad_norm": 1.1460512038122599, "learning_rate": 1.3706445186094075e-05, "loss": 0.4592, "step": 11165 }, { "epoch": 0.3978124944332615, "grad_norm": 1.0090406337075741, "learning_rate": 1.3705373414088618e-05, "loss": 0.503, "step": 11166 }, { "epoch": 0.3978481215597556, "grad_norm": 0.9859797300099126, "learning_rate": 1.3704301592743522e-05, "loss": 0.4842, "step": 11167 }, { "epoch": 0.3978837486862497, "grad_norm": 1.0333651281948173, "learning_rate": 1.3703229722073065e-05, "loss": 0.4919, "step": 11168 }, { "epoch": 0.39791937581274384, "grad_norm": 1.1769850368511265, "learning_rate": 1.3702157802091515e-05, "loss": 0.4969, "step": 11169 }, { "epoch": 0.3979550029392379, "grad_norm": 1.0266131053223708, "learning_rate": 1.3701085832813145e-05, "loss": 0.4755, "step": 11170 }, { "epoch": 0.39799063006573204, "grad_norm": 1.1666325439220586, "learning_rate": 1.3700013814252232e-05, "loss": 0.5103, "step": 11171 }, { "epoch": 0.3980262571922262, "grad_norm": 1.2319065016591264, "learning_rate": 1.3698941746423047e-05, "loss": 0.5865, "step": 11172 }, { "epoch": 0.39806188431872025, "grad_norm": 2.0565271876145843, "learning_rate": 1.3697869629339869e-05, "loss": 0.5511, "step": 11173 }, { "epoch": 0.3980975114452144, "grad_norm": 1.0972227797440632, "learning_rate": 1.3696797463016972e-05, "loss": 0.4601, "step": 11174 }, { "epoch": 0.3981331385717085, "grad_norm": 1.1531862590121673, "learning_rate": 1.3695725247468633e-05, "loss": 0.439, "step": 11175 }, { "epoch": 0.3981687656982026, "grad_norm": 0.956103244715192, "learning_rate": 1.3694652982709126e-05, "loss": 0.4525, "step": 11176 }, { "epoch": 0.3982043928246967, "grad_norm": 1.2495350693307958, "learning_rate": 1.3693580668752735e-05, "loss": 0.4825, "step": 11177 }, { "epoch": 0.39824001995119085, "grad_norm": 1.5788251460545015, "learning_rate": 1.3692508305613733e-05, "loss": 0.5928, "step": 11178 }, { "epoch": 0.3982756470776849, "grad_norm": 1.5962338068260835, "learning_rate": 1.3691435893306406e-05, "loss": 0.5929, "step": 11179 }, { "epoch": 0.39831127420417906, "grad_norm": 1.6411563586318172, "learning_rate": 1.3690363431845029e-05, "loss": 0.6196, "step": 11180 }, { "epoch": 0.3983469013306732, "grad_norm": 1.7199838047674618, "learning_rate": 1.3689290921243883e-05, "loss": 0.5906, "step": 11181 }, { "epoch": 0.39838252845716726, "grad_norm": 1.0840039037179054, "learning_rate": 1.3688218361517246e-05, "loss": 0.4524, "step": 11182 }, { "epoch": 0.3984181555836614, "grad_norm": 1.148401849480736, "learning_rate": 1.3687145752679409e-05, "loss": 0.4933, "step": 11183 }, { "epoch": 0.3984537827101555, "grad_norm": 2.178210359404784, "learning_rate": 1.3686073094744647e-05, "loss": 0.6405, "step": 11184 }, { "epoch": 0.3984894098366496, "grad_norm": 1.1077879587484407, "learning_rate": 1.3685000387727245e-05, "loss": 0.5032, "step": 11185 }, { "epoch": 0.39852503696314373, "grad_norm": 1.0255501238706064, "learning_rate": 1.3683927631641486e-05, "loss": 0.4378, "step": 11186 }, { "epoch": 0.39856066408963786, "grad_norm": 1.5555101785864447, "learning_rate": 1.368285482650166e-05, "loss": 0.6657, "step": 11187 }, { "epoch": 0.39859629121613194, "grad_norm": 1.385743981606871, "learning_rate": 1.3681781972322046e-05, "loss": 0.6028, "step": 11188 }, { "epoch": 0.39863191834262607, "grad_norm": 1.0429864386893628, "learning_rate": 1.3680709069116929e-05, "loss": 0.4705, "step": 11189 }, { "epoch": 0.3986675454691202, "grad_norm": 1.471398547546885, "learning_rate": 1.3679636116900598e-05, "loss": 0.5791, "step": 11190 }, { "epoch": 0.3987031725956143, "grad_norm": 1.1039327407426465, "learning_rate": 1.3678563115687342e-05, "loss": 0.5027, "step": 11191 }, { "epoch": 0.3987387997221084, "grad_norm": 1.0721537298050055, "learning_rate": 1.3677490065491447e-05, "loss": 0.4767, "step": 11192 }, { "epoch": 0.39877442684860254, "grad_norm": 1.0950191265258877, "learning_rate": 1.3676416966327201e-05, "loss": 0.4931, "step": 11193 }, { "epoch": 0.3988100539750966, "grad_norm": 1.0906732667795216, "learning_rate": 1.3675343818208894e-05, "loss": 0.4487, "step": 11194 }, { "epoch": 0.39884568110159074, "grad_norm": 1.1430240158109128, "learning_rate": 1.3674270621150817e-05, "loss": 0.4476, "step": 11195 }, { "epoch": 0.3988813082280849, "grad_norm": 1.5913109295146526, "learning_rate": 1.3673197375167255e-05, "loss": 0.6031, "step": 11196 }, { "epoch": 0.39891693535457895, "grad_norm": 1.6683996955332285, "learning_rate": 1.3672124080272504e-05, "loss": 0.6081, "step": 11197 }, { "epoch": 0.3989525624810731, "grad_norm": 2.0373012673045423, "learning_rate": 1.3671050736480855e-05, "loss": 0.5942, "step": 11198 }, { "epoch": 0.3989881896075672, "grad_norm": 1.0688729850531355, "learning_rate": 1.3669977343806599e-05, "loss": 0.4683, "step": 11199 }, { "epoch": 0.3990238167340613, "grad_norm": 1.2483298248296653, "learning_rate": 1.3668903902264027e-05, "loss": 0.4795, "step": 11200 }, { "epoch": 0.3990594438605554, "grad_norm": 1.0842716621421444, "learning_rate": 1.3667830411867438e-05, "loss": 0.4531, "step": 11201 }, { "epoch": 0.39909507098704955, "grad_norm": 1.7441591585092318, "learning_rate": 1.3666756872631122e-05, "loss": 0.5877, "step": 11202 }, { "epoch": 0.3991306981135436, "grad_norm": 1.5813435482628502, "learning_rate": 1.3665683284569377e-05, "loss": 0.635, "step": 11203 }, { "epoch": 0.39916632524003776, "grad_norm": 1.1545515697552169, "learning_rate": 1.3664609647696497e-05, "loss": 0.4636, "step": 11204 }, { "epoch": 0.3992019523665319, "grad_norm": 1.11104038726467, "learning_rate": 1.3663535962026777e-05, "loss": 0.4663, "step": 11205 }, { "epoch": 0.39923757949302596, "grad_norm": 1.0665816435700213, "learning_rate": 1.3662462227574515e-05, "loss": 0.4641, "step": 11206 }, { "epoch": 0.3992732066195201, "grad_norm": 2.3538303284513207, "learning_rate": 1.3661388444354009e-05, "loss": 0.5426, "step": 11207 }, { "epoch": 0.3993088337460142, "grad_norm": 0.9693391632897014, "learning_rate": 1.3660314612379554e-05, "loss": 0.4444, "step": 11208 }, { "epoch": 0.39934446087250836, "grad_norm": 1.0464431182048772, "learning_rate": 1.3659240731665455e-05, "loss": 0.4908, "step": 11209 }, { "epoch": 0.39938008799900243, "grad_norm": 1.8681987158201765, "learning_rate": 1.3658166802226007e-05, "loss": 0.5586, "step": 11210 }, { "epoch": 0.39941571512549656, "grad_norm": 1.8257518170118407, "learning_rate": 1.365709282407551e-05, "loss": 0.6216, "step": 11211 }, { "epoch": 0.3994513422519907, "grad_norm": 1.1349887241830656, "learning_rate": 1.3656018797228269e-05, "loss": 0.4932, "step": 11212 }, { "epoch": 0.39948696937848477, "grad_norm": 1.860875601940376, "learning_rate": 1.365494472169858e-05, "loss": 0.6149, "step": 11213 }, { "epoch": 0.3995225965049789, "grad_norm": 1.4567612227985658, "learning_rate": 1.3653870597500747e-05, "loss": 0.5702, "step": 11214 }, { "epoch": 0.39955822363147303, "grad_norm": 1.5335790324107244, "learning_rate": 1.3652796424649073e-05, "loss": 0.5223, "step": 11215 }, { "epoch": 0.3995938507579671, "grad_norm": 1.8554209400277213, "learning_rate": 1.3651722203157863e-05, "loss": 0.5887, "step": 11216 }, { "epoch": 0.39962947788446124, "grad_norm": 1.4369869523608787, "learning_rate": 1.365064793304142e-05, "loss": 0.6048, "step": 11217 }, { "epoch": 0.39966510501095537, "grad_norm": 1.5612992916579942, "learning_rate": 1.3649573614314045e-05, "loss": 0.5594, "step": 11218 }, { "epoch": 0.39970073213744944, "grad_norm": 1.3170699878672134, "learning_rate": 1.3648499246990048e-05, "loss": 0.4682, "step": 11219 }, { "epoch": 0.3997363592639436, "grad_norm": 1.0392970954745864, "learning_rate": 1.3647424831083733e-05, "loss": 0.4664, "step": 11220 }, { "epoch": 0.3997719863904377, "grad_norm": 1.2355228159477458, "learning_rate": 1.3646350366609409e-05, "loss": 0.4848, "step": 11221 }, { "epoch": 0.3998076135169318, "grad_norm": 1.1736720943317251, "learning_rate": 1.3645275853581379e-05, "loss": 0.4895, "step": 11222 }, { "epoch": 0.3998432406434259, "grad_norm": 1.1210813502304373, "learning_rate": 1.3644201292013955e-05, "loss": 0.4637, "step": 11223 }, { "epoch": 0.39987886776992004, "grad_norm": 1.8339729976676162, "learning_rate": 1.3643126681921441e-05, "loss": 0.5257, "step": 11224 }, { "epoch": 0.3999144948964141, "grad_norm": 1.3973204787296116, "learning_rate": 1.3642052023318151e-05, "loss": 0.6133, "step": 11225 }, { "epoch": 0.39995012202290825, "grad_norm": 1.530419396307664, "learning_rate": 1.3640977316218391e-05, "loss": 0.5533, "step": 11226 }, { "epoch": 0.3999857491494024, "grad_norm": 1.6039146110389637, "learning_rate": 1.3639902560636473e-05, "loss": 0.6119, "step": 11227 }, { "epoch": 0.40002137627589646, "grad_norm": 1.805537823466357, "learning_rate": 1.363882775658671e-05, "loss": 0.6169, "step": 11228 }, { "epoch": 0.4000570034023906, "grad_norm": 1.1070179005458467, "learning_rate": 1.3637752904083412e-05, "loss": 0.5068, "step": 11229 }, { "epoch": 0.4000926305288847, "grad_norm": 1.7404165437075538, "learning_rate": 1.3636678003140889e-05, "loss": 0.5806, "step": 11230 }, { "epoch": 0.4001282576553788, "grad_norm": 1.0733944668536612, "learning_rate": 1.363560305377346e-05, "loss": 0.434, "step": 11231 }, { "epoch": 0.4001638847818729, "grad_norm": 1.138316462508332, "learning_rate": 1.3634528055995433e-05, "loss": 0.4745, "step": 11232 }, { "epoch": 0.40019951190836706, "grad_norm": 1.1874786030431104, "learning_rate": 1.3633453009821127e-05, "loss": 0.4932, "step": 11233 }, { "epoch": 0.40023513903486113, "grad_norm": 1.1042746606041385, "learning_rate": 1.3632377915264852e-05, "loss": 0.4357, "step": 11234 }, { "epoch": 0.40027076616135526, "grad_norm": 1.5612936970543114, "learning_rate": 1.3631302772340926e-05, "loss": 0.5859, "step": 11235 }, { "epoch": 0.4003063932878494, "grad_norm": 1.0590757731235325, "learning_rate": 1.3630227581063666e-05, "loss": 0.4687, "step": 11236 }, { "epoch": 0.40034202041434347, "grad_norm": 1.4996139300504658, "learning_rate": 1.3629152341447389e-05, "loss": 0.5456, "step": 11237 }, { "epoch": 0.4003776475408376, "grad_norm": 1.8054740207445348, "learning_rate": 1.362807705350641e-05, "loss": 0.5729, "step": 11238 }, { "epoch": 0.40041327466733173, "grad_norm": 1.1687188533013357, "learning_rate": 1.3627001717255053e-05, "loss": 0.4791, "step": 11239 }, { "epoch": 0.4004489017938258, "grad_norm": 1.2704389875975226, "learning_rate": 1.362592633270763e-05, "loss": 0.4523, "step": 11240 }, { "epoch": 0.40048452892031994, "grad_norm": 1.2402603409641697, "learning_rate": 1.3624850899878463e-05, "loss": 0.5218, "step": 11241 }, { "epoch": 0.40052015604681407, "grad_norm": 1.4958244557278317, "learning_rate": 1.3623775418781874e-05, "loss": 0.6281, "step": 11242 }, { "epoch": 0.40055578317330814, "grad_norm": 2.0141725449432393, "learning_rate": 1.3622699889432184e-05, "loss": 0.5783, "step": 11243 }, { "epoch": 0.4005914102998023, "grad_norm": 2.056327549689897, "learning_rate": 1.362162431184371e-05, "loss": 0.5856, "step": 11244 }, { "epoch": 0.4006270374262964, "grad_norm": 1.0563740912587383, "learning_rate": 1.3620548686030779e-05, "loss": 0.452, "step": 11245 }, { "epoch": 0.4006626645527905, "grad_norm": 2.531570409235725, "learning_rate": 1.3619473012007712e-05, "loss": 0.6127, "step": 11246 }, { "epoch": 0.4006982916792846, "grad_norm": 1.550504057894491, "learning_rate": 1.3618397289788831e-05, "loss": 0.4637, "step": 11247 }, { "epoch": 0.40073391880577874, "grad_norm": 1.4427038876244747, "learning_rate": 1.361732151938846e-05, "loss": 0.5345, "step": 11248 }, { "epoch": 0.4007695459322728, "grad_norm": 1.099594316699964, "learning_rate": 1.3616245700820922e-05, "loss": 0.4868, "step": 11249 }, { "epoch": 0.40080517305876695, "grad_norm": 1.2307701170062593, "learning_rate": 1.361516983410055e-05, "loss": 0.4986, "step": 11250 }, { "epoch": 0.4008408001852611, "grad_norm": 1.5044992250674953, "learning_rate": 1.3614093919241664e-05, "loss": 0.5772, "step": 11251 }, { "epoch": 0.40087642731175516, "grad_norm": 1.6422302721124482, "learning_rate": 1.3613017956258591e-05, "loss": 0.5976, "step": 11252 }, { "epoch": 0.4009120544382493, "grad_norm": 1.47134936261146, "learning_rate": 1.3611941945165656e-05, "loss": 0.5522, "step": 11253 }, { "epoch": 0.4009476815647434, "grad_norm": 1.9182596148445767, "learning_rate": 1.3610865885977194e-05, "loss": 0.5741, "step": 11254 }, { "epoch": 0.4009833086912375, "grad_norm": 1.7029462014948586, "learning_rate": 1.3609789778707527e-05, "loss": 0.6018, "step": 11255 }, { "epoch": 0.4010189358177316, "grad_norm": 0.9639159329929708, "learning_rate": 1.3608713623370983e-05, "loss": 0.4271, "step": 11256 }, { "epoch": 0.40105456294422576, "grad_norm": 1.1107603983175587, "learning_rate": 1.3607637419981898e-05, "loss": 0.4711, "step": 11257 }, { "epoch": 0.40109019007071983, "grad_norm": 1.0772752790836646, "learning_rate": 1.3606561168554597e-05, "loss": 0.4615, "step": 11258 }, { "epoch": 0.40112581719721396, "grad_norm": 1.0226812814213795, "learning_rate": 1.3605484869103416e-05, "loss": 0.4904, "step": 11259 }, { "epoch": 0.4011614443237081, "grad_norm": 1.1130214346004403, "learning_rate": 1.360440852164268e-05, "loss": 0.416, "step": 11260 }, { "epoch": 0.40119707145020217, "grad_norm": 1.0376147332539747, "learning_rate": 1.3603332126186728e-05, "loss": 0.501, "step": 11261 }, { "epoch": 0.4012326985766963, "grad_norm": 1.0809873669053092, "learning_rate": 1.360225568274989e-05, "loss": 0.4143, "step": 11262 }, { "epoch": 0.40126832570319043, "grad_norm": 1.0561236811340786, "learning_rate": 1.3601179191346502e-05, "loss": 0.4782, "step": 11263 }, { "epoch": 0.4013039528296845, "grad_norm": 1.4389820904317454, "learning_rate": 1.360010265199089e-05, "loss": 0.4738, "step": 11264 }, { "epoch": 0.40133957995617864, "grad_norm": 1.0547059289494058, "learning_rate": 1.3599026064697399e-05, "loss": 0.4636, "step": 11265 }, { "epoch": 0.40137520708267277, "grad_norm": 1.1613552033971692, "learning_rate": 1.3597949429480361e-05, "loss": 0.493, "step": 11266 }, { "epoch": 0.40141083420916684, "grad_norm": 1.0465485074760776, "learning_rate": 1.3596872746354108e-05, "loss": 0.4299, "step": 11267 }, { "epoch": 0.401446461335661, "grad_norm": 1.1154634419776603, "learning_rate": 1.3595796015332986e-05, "loss": 0.4959, "step": 11268 }, { "epoch": 0.4014820884621551, "grad_norm": 3.229080567684756, "learning_rate": 1.3594719236431321e-05, "loss": 0.6012, "step": 11269 }, { "epoch": 0.4015177155886492, "grad_norm": 1.840294559296667, "learning_rate": 1.3593642409663461e-05, "loss": 0.5874, "step": 11270 }, { "epoch": 0.4015533427151433, "grad_norm": 2.074907527106967, "learning_rate": 1.3592565535043738e-05, "loss": 0.5799, "step": 11271 }, { "epoch": 0.40158896984163744, "grad_norm": 1.1786500639703161, "learning_rate": 1.3591488612586492e-05, "loss": 0.4409, "step": 11272 }, { "epoch": 0.4016245969681315, "grad_norm": 1.2535350597907822, "learning_rate": 1.3590411642306068e-05, "loss": 0.4926, "step": 11273 }, { "epoch": 0.40166022409462565, "grad_norm": 1.1050513987457573, "learning_rate": 1.3589334624216804e-05, "loss": 0.3843, "step": 11274 }, { "epoch": 0.4016958512211198, "grad_norm": 1.550262185141728, "learning_rate": 1.3588257558333039e-05, "loss": 0.6137, "step": 11275 }, { "epoch": 0.40173147834761386, "grad_norm": 1.9071522094596287, "learning_rate": 1.3587180444669117e-05, "loss": 0.5424, "step": 11276 }, { "epoch": 0.401767105474108, "grad_norm": 1.4963527258514182, "learning_rate": 1.358610328323938e-05, "loss": 0.5862, "step": 11277 }, { "epoch": 0.4018027326006021, "grad_norm": 1.0383487284647155, "learning_rate": 1.358502607405817e-05, "loss": 0.4382, "step": 11278 }, { "epoch": 0.4018383597270962, "grad_norm": 1.869662081399564, "learning_rate": 1.3583948817139833e-05, "loss": 0.5118, "step": 11279 }, { "epoch": 0.4018739868535903, "grad_norm": 1.7093557353470548, "learning_rate": 1.3582871512498714e-05, "loss": 0.6462, "step": 11280 }, { "epoch": 0.40190961398008446, "grad_norm": 1.1455937393762994, "learning_rate": 1.3581794160149158e-05, "loss": 0.5187, "step": 11281 }, { "epoch": 0.40194524110657853, "grad_norm": 1.1905583201242214, "learning_rate": 1.3580716760105504e-05, "loss": 0.4647, "step": 11282 }, { "epoch": 0.40198086823307266, "grad_norm": 1.3576947198336609, "learning_rate": 1.3579639312382105e-05, "loss": 0.6109, "step": 11283 }, { "epoch": 0.4020164953595668, "grad_norm": 1.2307829191191293, "learning_rate": 1.3578561816993307e-05, "loss": 0.5243, "step": 11284 }, { "epoch": 0.40205212248606087, "grad_norm": 1.0828061043766577, "learning_rate": 1.3577484273953458e-05, "loss": 0.4913, "step": 11285 }, { "epoch": 0.402087749612555, "grad_norm": 1.0499285658558222, "learning_rate": 1.3576406683276905e-05, "loss": 0.4391, "step": 11286 }, { "epoch": 0.40212337673904913, "grad_norm": 1.9154465371184222, "learning_rate": 1.3575329044977998e-05, "loss": 0.5107, "step": 11287 }, { "epoch": 0.4021590038655432, "grad_norm": 1.6002967507388453, "learning_rate": 1.3574251359071087e-05, "loss": 0.5057, "step": 11288 }, { "epoch": 0.40219463099203734, "grad_norm": 1.0864731365114113, "learning_rate": 1.357317362557052e-05, "loss": 0.48, "step": 11289 }, { "epoch": 0.40223025811853147, "grad_norm": 1.984757683398237, "learning_rate": 1.3572095844490645e-05, "loss": 0.6411, "step": 11290 }, { "epoch": 0.40226588524502555, "grad_norm": 1.0684986126201257, "learning_rate": 1.357101801584582e-05, "loss": 0.4921, "step": 11291 }, { "epoch": 0.4023015123715197, "grad_norm": 1.5245738696777151, "learning_rate": 1.3569940139650395e-05, "loss": 0.5187, "step": 11292 }, { "epoch": 0.4023371394980138, "grad_norm": 1.5787562853969745, "learning_rate": 1.356886221591872e-05, "loss": 0.613, "step": 11293 }, { "epoch": 0.4023727666245079, "grad_norm": 1.07701771568345, "learning_rate": 1.356778424466515e-05, "loss": 0.5164, "step": 11294 }, { "epoch": 0.402408393751002, "grad_norm": 1.1468702742346781, "learning_rate": 1.3566706225904041e-05, "loss": 0.4448, "step": 11295 }, { "epoch": 0.40244402087749614, "grad_norm": 1.3853353940322426, "learning_rate": 1.3565628159649745e-05, "loss": 0.5775, "step": 11296 }, { "epoch": 0.4024796480039902, "grad_norm": 1.4105452348034695, "learning_rate": 1.3564550045916616e-05, "loss": 0.5516, "step": 11297 }, { "epoch": 0.40251527513048435, "grad_norm": 1.03335241812831, "learning_rate": 1.3563471884719012e-05, "loss": 0.4225, "step": 11298 }, { "epoch": 0.4025509022569785, "grad_norm": 1.0160614895171616, "learning_rate": 1.3562393676071293e-05, "loss": 0.5183, "step": 11299 }, { "epoch": 0.40258652938347256, "grad_norm": 1.0309485953239779, "learning_rate": 1.356131541998781e-05, "loss": 0.4918, "step": 11300 }, { "epoch": 0.4026221565099667, "grad_norm": 1.1812971927368265, "learning_rate": 1.356023711648292e-05, "loss": 0.5208, "step": 11301 }, { "epoch": 0.4026577836364608, "grad_norm": 1.1404021169936167, "learning_rate": 1.3559158765570987e-05, "loss": 0.4791, "step": 11302 }, { "epoch": 0.4026934107629549, "grad_norm": 1.6187828022889608, "learning_rate": 1.355808036726637e-05, "loss": 0.6403, "step": 11303 }, { "epoch": 0.402729037889449, "grad_norm": 1.4904637506651008, "learning_rate": 1.3557001921583419e-05, "loss": 0.5733, "step": 11304 }, { "epoch": 0.40276466501594316, "grad_norm": 1.1615784992670464, "learning_rate": 1.3555923428536506e-05, "loss": 0.4377, "step": 11305 }, { "epoch": 0.40280029214243723, "grad_norm": 1.179041537055507, "learning_rate": 1.3554844888139986e-05, "loss": 0.5228, "step": 11306 }, { "epoch": 0.40283591926893136, "grad_norm": 1.0549163184880759, "learning_rate": 1.3553766300408225e-05, "loss": 0.4938, "step": 11307 }, { "epoch": 0.4028715463954255, "grad_norm": 1.4031161032333437, "learning_rate": 1.3552687665355577e-05, "loss": 0.5225, "step": 11308 }, { "epoch": 0.40290717352191957, "grad_norm": 1.6494684726361841, "learning_rate": 1.3551608982996412e-05, "loss": 0.5808, "step": 11309 }, { "epoch": 0.4029428006484137, "grad_norm": 1.0073791197512023, "learning_rate": 1.3550530253345088e-05, "loss": 0.4471, "step": 11310 }, { "epoch": 0.40297842777490783, "grad_norm": 1.3518164400837978, "learning_rate": 1.3549451476415976e-05, "loss": 0.6068, "step": 11311 }, { "epoch": 0.4030140549014019, "grad_norm": 1.0654445122919407, "learning_rate": 1.3548372652223434e-05, "loss": 0.5133, "step": 11312 }, { "epoch": 0.40304968202789604, "grad_norm": 1.3245040071466667, "learning_rate": 1.3547293780781832e-05, "loss": 0.6023, "step": 11313 }, { "epoch": 0.40308530915439017, "grad_norm": 1.9905620977659495, "learning_rate": 1.3546214862105533e-05, "loss": 0.591, "step": 11314 }, { "epoch": 0.40312093628088425, "grad_norm": 1.409867839813979, "learning_rate": 1.3545135896208904e-05, "loss": 0.5736, "step": 11315 }, { "epoch": 0.4031565634073784, "grad_norm": 1.0068456224584654, "learning_rate": 1.3544056883106313e-05, "loss": 0.4694, "step": 11316 }, { "epoch": 0.4031921905338725, "grad_norm": 1.6401767556224347, "learning_rate": 1.3542977822812126e-05, "loss": 0.6264, "step": 11317 }, { "epoch": 0.4032278176603666, "grad_norm": 1.2950038503380477, "learning_rate": 1.3541898715340716e-05, "loss": 0.4795, "step": 11318 }, { "epoch": 0.4032634447868607, "grad_norm": 1.0521711983931652, "learning_rate": 1.3540819560706446e-05, "loss": 0.4547, "step": 11319 }, { "epoch": 0.40329907191335485, "grad_norm": 1.1791177108005457, "learning_rate": 1.353974035892369e-05, "loss": 0.501, "step": 11320 }, { "epoch": 0.4033346990398489, "grad_norm": 1.4643624744204862, "learning_rate": 1.3538661110006816e-05, "loss": 0.5542, "step": 11321 }, { "epoch": 0.40337032616634305, "grad_norm": 1.1369041732300502, "learning_rate": 1.3537581813970197e-05, "loss": 0.4767, "step": 11322 }, { "epoch": 0.4034059532928372, "grad_norm": 1.1332461322688532, "learning_rate": 1.3536502470828204e-05, "loss": 0.5247, "step": 11323 }, { "epoch": 0.40344158041933126, "grad_norm": 1.493953157138857, "learning_rate": 1.3535423080595209e-05, "loss": 0.4873, "step": 11324 }, { "epoch": 0.4034772075458254, "grad_norm": 1.075229565893983, "learning_rate": 1.3534343643285583e-05, "loss": 0.4887, "step": 11325 }, { "epoch": 0.4035128346723195, "grad_norm": 1.5430995818496505, "learning_rate": 1.3533264158913704e-05, "loss": 0.6026, "step": 11326 }, { "epoch": 0.4035484617988136, "grad_norm": 1.0772148801329011, "learning_rate": 1.3532184627493939e-05, "loss": 0.464, "step": 11327 }, { "epoch": 0.4035840889253077, "grad_norm": 1.167173297107454, "learning_rate": 1.3531105049040667e-05, "loss": 0.4836, "step": 11328 }, { "epoch": 0.40361971605180186, "grad_norm": 1.0521476386472974, "learning_rate": 1.3530025423568266e-05, "loss": 0.477, "step": 11329 }, { "epoch": 0.40365534317829593, "grad_norm": 0.9653877014982069, "learning_rate": 1.3528945751091108e-05, "loss": 0.4479, "step": 11330 }, { "epoch": 0.40369097030479006, "grad_norm": 1.0534752459430674, "learning_rate": 1.3527866031623569e-05, "loss": 0.4535, "step": 11331 }, { "epoch": 0.4037265974312842, "grad_norm": 1.011009517063859, "learning_rate": 1.352678626518003e-05, "loss": 0.4784, "step": 11332 }, { "epoch": 0.40376222455777827, "grad_norm": 1.1158580023870581, "learning_rate": 1.3525706451774866e-05, "loss": 0.483, "step": 11333 }, { "epoch": 0.4037978516842724, "grad_norm": 1.5854730404944317, "learning_rate": 1.3524626591422458e-05, "loss": 0.5176, "step": 11334 }, { "epoch": 0.40383347881076653, "grad_norm": 1.4953897125013549, "learning_rate": 1.3523546684137183e-05, "loss": 0.5889, "step": 11335 }, { "epoch": 0.4038691059372606, "grad_norm": 1.7431748436931624, "learning_rate": 1.352246672993342e-05, "loss": 0.5779, "step": 11336 }, { "epoch": 0.40390473306375474, "grad_norm": 1.3733381122812842, "learning_rate": 1.3521386728825551e-05, "loss": 0.5543, "step": 11337 }, { "epoch": 0.40394036019024887, "grad_norm": 1.2671915690022117, "learning_rate": 1.3520306680827954e-05, "loss": 0.5009, "step": 11338 }, { "epoch": 0.40397598731674295, "grad_norm": 1.7008587947232285, "learning_rate": 1.3519226585955016e-05, "loss": 0.5665, "step": 11339 }, { "epoch": 0.4040116144432371, "grad_norm": 1.0239342372581248, "learning_rate": 1.3518146444221114e-05, "loss": 0.4877, "step": 11340 }, { "epoch": 0.4040472415697312, "grad_norm": 1.864723384431599, "learning_rate": 1.3517066255640635e-05, "loss": 0.4875, "step": 11341 }, { "epoch": 0.4040828686962253, "grad_norm": 1.4849216779436964, "learning_rate": 1.351598602022796e-05, "loss": 0.6168, "step": 11342 }, { "epoch": 0.4041184958227194, "grad_norm": 1.048734025471015, "learning_rate": 1.3514905737997474e-05, "loss": 0.4414, "step": 11343 }, { "epoch": 0.40415412294921355, "grad_norm": 1.0660886798808933, "learning_rate": 1.3513825408963561e-05, "loss": 0.5025, "step": 11344 }, { "epoch": 0.4041897500757076, "grad_norm": 1.1592530094765567, "learning_rate": 1.3512745033140608e-05, "loss": 0.5085, "step": 11345 }, { "epoch": 0.40422537720220175, "grad_norm": 1.6876646046036063, "learning_rate": 1.3511664610542996e-05, "loss": 0.5842, "step": 11346 }, { "epoch": 0.4042610043286959, "grad_norm": 1.7679984530353374, "learning_rate": 1.3510584141185117e-05, "loss": 0.5909, "step": 11347 }, { "epoch": 0.40429663145518996, "grad_norm": 1.0729645676798332, "learning_rate": 1.350950362508136e-05, "loss": 0.5222, "step": 11348 }, { "epoch": 0.4043322585816841, "grad_norm": 1.074627531145874, "learning_rate": 1.3508423062246104e-05, "loss": 0.4591, "step": 11349 }, { "epoch": 0.4043678857081782, "grad_norm": 1.0601089340354455, "learning_rate": 1.3507342452693743e-05, "loss": 0.4795, "step": 11350 }, { "epoch": 0.4044035128346723, "grad_norm": 1.4821159088732603, "learning_rate": 1.3506261796438666e-05, "loss": 0.5968, "step": 11351 }, { "epoch": 0.4044391399611664, "grad_norm": 1.0864583348922463, "learning_rate": 1.3505181093495265e-05, "loss": 0.4725, "step": 11352 }, { "epoch": 0.40447476708766056, "grad_norm": 1.6104765698856365, "learning_rate": 1.3504100343877928e-05, "loss": 0.6097, "step": 11353 }, { "epoch": 0.40451039421415463, "grad_norm": 1.01737519739202, "learning_rate": 1.3503019547601044e-05, "loss": 0.4497, "step": 11354 }, { "epoch": 0.40454602134064876, "grad_norm": 3.0101579773814717, "learning_rate": 1.3501938704679005e-05, "loss": 0.595, "step": 11355 }, { "epoch": 0.4045816484671429, "grad_norm": 1.6267628986537, "learning_rate": 1.350085781512621e-05, "loss": 0.5375, "step": 11356 }, { "epoch": 0.40461727559363697, "grad_norm": 1.0574010986799744, "learning_rate": 1.3499776878957038e-05, "loss": 0.4651, "step": 11357 }, { "epoch": 0.4046529027201311, "grad_norm": 1.2302133896895862, "learning_rate": 1.3498695896185896e-05, "loss": 0.5126, "step": 11358 }, { "epoch": 0.40468852984662523, "grad_norm": 1.348981923055046, "learning_rate": 1.3497614866827173e-05, "loss": 0.583, "step": 11359 }, { "epoch": 0.4047241569731193, "grad_norm": 1.0491997801780109, "learning_rate": 1.3496533790895262e-05, "loss": 0.485, "step": 11360 }, { "epoch": 0.40475978409961344, "grad_norm": 1.1636039564284852, "learning_rate": 1.3495452668404558e-05, "loss": 0.5127, "step": 11361 }, { "epoch": 0.40479541122610757, "grad_norm": 1.0737217611800989, "learning_rate": 1.3494371499369461e-05, "loss": 0.4217, "step": 11362 }, { "epoch": 0.40483103835260165, "grad_norm": 1.713474418874044, "learning_rate": 1.3493290283804363e-05, "loss": 0.5653, "step": 11363 }, { "epoch": 0.4048666654790958, "grad_norm": 1.5078393128930623, "learning_rate": 1.3492209021723665e-05, "loss": 0.5523, "step": 11364 }, { "epoch": 0.4049022926055899, "grad_norm": 1.3732736991654035, "learning_rate": 1.349112771314176e-05, "loss": 0.4751, "step": 11365 }, { "epoch": 0.404937919732084, "grad_norm": 1.523894927955728, "learning_rate": 1.3490046358073055e-05, "loss": 0.5438, "step": 11366 }, { "epoch": 0.4049735468585781, "grad_norm": 1.2191101086989484, "learning_rate": 1.348896495653194e-05, "loss": 0.4729, "step": 11367 }, { "epoch": 0.40500917398507225, "grad_norm": 1.0778547571140153, "learning_rate": 1.3487883508532815e-05, "loss": 0.5042, "step": 11368 }, { "epoch": 0.4050448011115663, "grad_norm": 1.6077618397453857, "learning_rate": 1.3486802014090088e-05, "loss": 0.667, "step": 11369 }, { "epoch": 0.40508042823806045, "grad_norm": 1.5377335594159214, "learning_rate": 1.3485720473218153e-05, "loss": 0.5028, "step": 11370 }, { "epoch": 0.4051160553645546, "grad_norm": 1.7819181591340574, "learning_rate": 1.3484638885931416e-05, "loss": 0.6452, "step": 11371 }, { "epoch": 0.40515168249104866, "grad_norm": 1.8845570987936393, "learning_rate": 1.3483557252244272e-05, "loss": 0.573, "step": 11372 }, { "epoch": 0.4051873096175428, "grad_norm": 1.3506041454901054, "learning_rate": 1.3482475572171132e-05, "loss": 0.6107, "step": 11373 }, { "epoch": 0.4052229367440369, "grad_norm": 1.619493457818303, "learning_rate": 1.3481393845726396e-05, "loss": 0.5036, "step": 11374 }, { "epoch": 0.405258563870531, "grad_norm": 1.644937028506315, "learning_rate": 1.3480312072924465e-05, "loss": 0.5393, "step": 11375 }, { "epoch": 0.4052941909970251, "grad_norm": 1.5953403364886773, "learning_rate": 1.3479230253779745e-05, "loss": 0.6271, "step": 11376 }, { "epoch": 0.40532981812351926, "grad_norm": 1.8119143317190303, "learning_rate": 1.3478148388306644e-05, "loss": 0.651, "step": 11377 }, { "epoch": 0.40536544525001333, "grad_norm": 1.470146617891932, "learning_rate": 1.347706647651957e-05, "loss": 0.53, "step": 11378 }, { "epoch": 0.40540107237650747, "grad_norm": 1.0608409622949475, "learning_rate": 1.3475984518432918e-05, "loss": 0.5006, "step": 11379 }, { "epoch": 0.4054366995030016, "grad_norm": 1.167232168632402, "learning_rate": 1.3474902514061108e-05, "loss": 0.5067, "step": 11380 }, { "epoch": 0.40547232662949567, "grad_norm": 1.598768275974384, "learning_rate": 1.3473820463418542e-05, "loss": 0.5563, "step": 11381 }, { "epoch": 0.4055079537559898, "grad_norm": 1.1379060759929056, "learning_rate": 1.347273836651963e-05, "loss": 0.5106, "step": 11382 }, { "epoch": 0.40554358088248393, "grad_norm": 1.0110282704912104, "learning_rate": 1.3471656223378773e-05, "loss": 0.4623, "step": 11383 }, { "epoch": 0.405579208008978, "grad_norm": 1.671205048583723, "learning_rate": 1.3470574034010394e-05, "loss": 0.5817, "step": 11384 }, { "epoch": 0.40561483513547214, "grad_norm": 1.0412201075694498, "learning_rate": 1.3469491798428891e-05, "loss": 0.4987, "step": 11385 }, { "epoch": 0.40565046226196627, "grad_norm": 1.1181926541886604, "learning_rate": 1.3468409516648679e-05, "loss": 0.4482, "step": 11386 }, { "epoch": 0.40568608938846035, "grad_norm": 1.2149020054540594, "learning_rate": 1.3467327188684171e-05, "loss": 0.5354, "step": 11387 }, { "epoch": 0.4057217165149545, "grad_norm": 1.055526519938658, "learning_rate": 1.3466244814549781e-05, "loss": 0.4723, "step": 11388 }, { "epoch": 0.4057573436414486, "grad_norm": 1.1390817203389396, "learning_rate": 1.3465162394259919e-05, "loss": 0.4776, "step": 11389 }, { "epoch": 0.4057929707679427, "grad_norm": 1.0642776974431913, "learning_rate": 1.3464079927828995e-05, "loss": 0.4978, "step": 11390 }, { "epoch": 0.4058285978944368, "grad_norm": 1.558468611323216, "learning_rate": 1.3462997415271424e-05, "loss": 0.559, "step": 11391 }, { "epoch": 0.40586422502093095, "grad_norm": 1.2207121306596178, "learning_rate": 1.3461914856601625e-05, "loss": 0.4729, "step": 11392 }, { "epoch": 0.405899852147425, "grad_norm": 1.0552698385707262, "learning_rate": 1.346083225183401e-05, "loss": 0.456, "step": 11393 }, { "epoch": 0.40593547927391915, "grad_norm": 1.5772394795847169, "learning_rate": 1.3459749600982991e-05, "loss": 0.5767, "step": 11394 }, { "epoch": 0.4059711064004133, "grad_norm": 1.0744926069545784, "learning_rate": 1.3458666904062992e-05, "loss": 0.5003, "step": 11395 }, { "epoch": 0.40600673352690736, "grad_norm": 2.237174195602042, "learning_rate": 1.3457584161088424e-05, "loss": 0.5944, "step": 11396 }, { "epoch": 0.4060423606534015, "grad_norm": 0.9897732171498479, "learning_rate": 1.3456501372073706e-05, "loss": 0.4203, "step": 11397 }, { "epoch": 0.4060779877798956, "grad_norm": 1.4497118035003558, "learning_rate": 1.3455418537033253e-05, "loss": 0.5289, "step": 11398 }, { "epoch": 0.40611361490638975, "grad_norm": 1.1254093882954475, "learning_rate": 1.3454335655981491e-05, "loss": 0.4807, "step": 11399 }, { "epoch": 0.40614924203288383, "grad_norm": 1.7492581668564615, "learning_rate": 1.3453252728932837e-05, "loss": 0.6318, "step": 11400 }, { "epoch": 0.40618486915937796, "grad_norm": 1.636422961743321, "learning_rate": 1.3452169755901707e-05, "loss": 0.5608, "step": 11401 }, { "epoch": 0.4062204962858721, "grad_norm": 1.4588486523793527, "learning_rate": 1.3451086736902524e-05, "loss": 0.5482, "step": 11402 }, { "epoch": 0.40625612341236617, "grad_norm": 1.5605697381920436, "learning_rate": 1.3450003671949707e-05, "loss": 0.5873, "step": 11403 }, { "epoch": 0.4062917505388603, "grad_norm": 1.1122681304168804, "learning_rate": 1.3448920561057683e-05, "loss": 0.4498, "step": 11404 }, { "epoch": 0.40632737766535443, "grad_norm": 1.6383320878449779, "learning_rate": 1.3447837404240868e-05, "loss": 0.5094, "step": 11405 }, { "epoch": 0.4063630047918485, "grad_norm": 1.3145734252444035, "learning_rate": 1.3446754201513689e-05, "loss": 0.4873, "step": 11406 }, { "epoch": 0.40639863191834263, "grad_norm": 1.4735010637271737, "learning_rate": 1.3445670952890573e-05, "loss": 0.5328, "step": 11407 }, { "epoch": 0.40643425904483677, "grad_norm": 1.16447247306374, "learning_rate": 1.3444587658385934e-05, "loss": 0.4813, "step": 11408 }, { "epoch": 0.40646988617133084, "grad_norm": 1.1840869695199479, "learning_rate": 1.3443504318014204e-05, "loss": 0.4731, "step": 11409 }, { "epoch": 0.40650551329782497, "grad_norm": 1.5638393667460506, "learning_rate": 1.3442420931789808e-05, "loss": 0.5346, "step": 11410 }, { "epoch": 0.4065411404243191, "grad_norm": 1.2172031468209874, "learning_rate": 1.3441337499727171e-05, "loss": 0.4974, "step": 11411 }, { "epoch": 0.4065767675508132, "grad_norm": 1.1641247747892731, "learning_rate": 1.3440254021840722e-05, "loss": 0.4978, "step": 11412 }, { "epoch": 0.4066123946773073, "grad_norm": 1.4992100528481713, "learning_rate": 1.3439170498144882e-05, "loss": 0.5735, "step": 11413 }, { "epoch": 0.40664802180380144, "grad_norm": 1.5904843286803816, "learning_rate": 1.3438086928654087e-05, "loss": 0.5298, "step": 11414 }, { "epoch": 0.4066836489302955, "grad_norm": 2.857797103008537, "learning_rate": 1.343700331338276e-05, "loss": 0.6287, "step": 11415 }, { "epoch": 0.40671927605678965, "grad_norm": 1.1828316405006951, "learning_rate": 1.3435919652345331e-05, "loss": 0.4888, "step": 11416 }, { "epoch": 0.4067549031832838, "grad_norm": 1.5926159127454003, "learning_rate": 1.3434835945556228e-05, "loss": 0.618, "step": 11417 }, { "epoch": 0.40679053030977785, "grad_norm": 1.0904404167781767, "learning_rate": 1.3433752193029888e-05, "loss": 0.4214, "step": 11418 }, { "epoch": 0.406826157436272, "grad_norm": 3.319112838479614, "learning_rate": 1.3432668394780735e-05, "loss": 0.5929, "step": 11419 }, { "epoch": 0.4068617845627661, "grad_norm": 1.2835388628776858, "learning_rate": 1.3431584550823203e-05, "loss": 0.4816, "step": 11420 }, { "epoch": 0.4068974116892602, "grad_norm": 1.139654479824414, "learning_rate": 1.3430500661171721e-05, "loss": 0.4493, "step": 11421 }, { "epoch": 0.4069330388157543, "grad_norm": 1.1785561014980186, "learning_rate": 1.3429416725840731e-05, "loss": 0.4881, "step": 11422 }, { "epoch": 0.40696866594224845, "grad_norm": 1.183772696214164, "learning_rate": 1.3428332744844659e-05, "loss": 0.4965, "step": 11423 }, { "epoch": 0.40700429306874253, "grad_norm": 1.1758446273282275, "learning_rate": 1.3427248718197935e-05, "loss": 0.427, "step": 11424 }, { "epoch": 0.40703992019523666, "grad_norm": 1.3237458370775605, "learning_rate": 1.3426164645915005e-05, "loss": 0.4635, "step": 11425 }, { "epoch": 0.4070755473217308, "grad_norm": 1.0979900209934752, "learning_rate": 1.3425080528010297e-05, "loss": 0.4794, "step": 11426 }, { "epoch": 0.40711117444822487, "grad_norm": 1.137778941411089, "learning_rate": 1.3423996364498248e-05, "loss": 0.4587, "step": 11427 }, { "epoch": 0.407146801574719, "grad_norm": 1.709537819864585, "learning_rate": 1.3422912155393291e-05, "loss": 0.5488, "step": 11428 }, { "epoch": 0.40718242870121313, "grad_norm": 1.2818909706562702, "learning_rate": 1.3421827900709867e-05, "loss": 0.4811, "step": 11429 }, { "epoch": 0.4072180558277072, "grad_norm": 1.0064343087339054, "learning_rate": 1.3420743600462416e-05, "loss": 0.4597, "step": 11430 }, { "epoch": 0.40725368295420133, "grad_norm": 1.6879179068025243, "learning_rate": 1.341965925466537e-05, "loss": 0.4948, "step": 11431 }, { "epoch": 0.40728931008069547, "grad_norm": 2.754503950377206, "learning_rate": 1.3418574863333169e-05, "loss": 0.6014, "step": 11432 }, { "epoch": 0.40732493720718954, "grad_norm": 1.4660418362953067, "learning_rate": 1.3417490426480258e-05, "loss": 0.4529, "step": 11433 }, { "epoch": 0.4073605643336837, "grad_norm": 1.0159943432159868, "learning_rate": 1.3416405944121074e-05, "loss": 0.4288, "step": 11434 }, { "epoch": 0.4073961914601778, "grad_norm": 1.4268507379501716, "learning_rate": 1.3415321416270055e-05, "loss": 0.5298, "step": 11435 }, { "epoch": 0.4074318185866719, "grad_norm": 1.9761491122854389, "learning_rate": 1.3414236842941644e-05, "loss": 0.6382, "step": 11436 }, { "epoch": 0.407467445713166, "grad_norm": 1.1047744941320292, "learning_rate": 1.3413152224150282e-05, "loss": 0.4821, "step": 11437 }, { "epoch": 0.40750307283966014, "grad_norm": 1.0100071497381704, "learning_rate": 1.3412067559910416e-05, "loss": 0.4195, "step": 11438 }, { "epoch": 0.4075386999661542, "grad_norm": 1.0100893492979828, "learning_rate": 1.341098285023648e-05, "loss": 0.4314, "step": 11439 }, { "epoch": 0.40757432709264835, "grad_norm": 1.80618480436793, "learning_rate": 1.3409898095142928e-05, "loss": 0.5781, "step": 11440 }, { "epoch": 0.4076099542191425, "grad_norm": 1.2626677262228692, "learning_rate": 1.3408813294644199e-05, "loss": 0.5243, "step": 11441 }, { "epoch": 0.40764558134563655, "grad_norm": 1.1457668895085766, "learning_rate": 1.3407728448754737e-05, "loss": 0.4428, "step": 11442 }, { "epoch": 0.4076812084721307, "grad_norm": 1.1959671416568791, "learning_rate": 1.340664355748899e-05, "loss": 0.5713, "step": 11443 }, { "epoch": 0.4077168355986248, "grad_norm": 1.107883472117574, "learning_rate": 1.34055586208614e-05, "loss": 0.4396, "step": 11444 }, { "epoch": 0.4077524627251189, "grad_norm": 1.0779182198905068, "learning_rate": 1.340447363888642e-05, "loss": 0.5045, "step": 11445 }, { "epoch": 0.407788089851613, "grad_norm": 1.1240379123343345, "learning_rate": 1.3403388611578491e-05, "loss": 0.4676, "step": 11446 }, { "epoch": 0.40782371697810715, "grad_norm": 1.1212622420604568, "learning_rate": 1.3402303538952065e-05, "loss": 0.4662, "step": 11447 }, { "epoch": 0.40785934410460123, "grad_norm": 1.194604744191447, "learning_rate": 1.3401218421021591e-05, "loss": 0.4895, "step": 11448 }, { "epoch": 0.40789497123109536, "grad_norm": 1.048824624908657, "learning_rate": 1.3400133257801517e-05, "loss": 0.4982, "step": 11449 }, { "epoch": 0.4079305983575895, "grad_norm": 1.1111332916329557, "learning_rate": 1.3399048049306289e-05, "loss": 0.4517, "step": 11450 }, { "epoch": 0.40796622548408357, "grad_norm": 1.095713417786608, "learning_rate": 1.3397962795550363e-05, "loss": 0.4814, "step": 11451 }, { "epoch": 0.4080018526105777, "grad_norm": 1.7803282020221893, "learning_rate": 1.3396877496548185e-05, "loss": 0.5503, "step": 11452 }, { "epoch": 0.40803747973707183, "grad_norm": 1.5480733879191984, "learning_rate": 1.3395792152314211e-05, "loss": 0.548, "step": 11453 }, { "epoch": 0.4080731068635659, "grad_norm": 1.6635724821721907, "learning_rate": 1.3394706762862887e-05, "loss": 0.5621, "step": 11454 }, { "epoch": 0.40810873399006004, "grad_norm": 1.9871764225358541, "learning_rate": 1.3393621328208673e-05, "loss": 0.6264, "step": 11455 }, { "epoch": 0.40814436111655417, "grad_norm": 1.0869381659447828, "learning_rate": 1.339253584836602e-05, "loss": 0.4939, "step": 11456 }, { "epoch": 0.40817998824304824, "grad_norm": 1.6485122752077828, "learning_rate": 1.339145032334938e-05, "loss": 0.5727, "step": 11457 }, { "epoch": 0.4082156153695424, "grad_norm": 1.102893010281956, "learning_rate": 1.3390364753173206e-05, "loss": 0.4583, "step": 11458 }, { "epoch": 0.4082512424960365, "grad_norm": 1.2018321958372111, "learning_rate": 1.3389279137851961e-05, "loss": 0.4533, "step": 11459 }, { "epoch": 0.4082868696225306, "grad_norm": 1.7005641270557839, "learning_rate": 1.3388193477400091e-05, "loss": 0.5916, "step": 11460 }, { "epoch": 0.4083224967490247, "grad_norm": 1.6851057747274234, "learning_rate": 1.3387107771832058e-05, "loss": 0.5591, "step": 11461 }, { "epoch": 0.40835812387551884, "grad_norm": 1.2470548653569586, "learning_rate": 1.3386022021162317e-05, "loss": 0.4806, "step": 11462 }, { "epoch": 0.4083937510020129, "grad_norm": 1.5722819584941659, "learning_rate": 1.3384936225405326e-05, "loss": 0.4875, "step": 11463 }, { "epoch": 0.40842937812850705, "grad_norm": 0.9874541849187612, "learning_rate": 1.3383850384575546e-05, "loss": 0.4352, "step": 11464 }, { "epoch": 0.4084650052550012, "grad_norm": 1.1556228752458744, "learning_rate": 1.3382764498687429e-05, "loss": 0.5157, "step": 11465 }, { "epoch": 0.40850063238149525, "grad_norm": 1.9330095730142618, "learning_rate": 1.3381678567755438e-05, "loss": 0.6545, "step": 11466 }, { "epoch": 0.4085362595079894, "grad_norm": 1.0814232990532746, "learning_rate": 1.3380592591794034e-05, "loss": 0.4833, "step": 11467 }, { "epoch": 0.4085718866344835, "grad_norm": 1.5250433091466915, "learning_rate": 1.337950657081768e-05, "loss": 0.5501, "step": 11468 }, { "epoch": 0.4086075137609776, "grad_norm": 2.654555722534305, "learning_rate": 1.337842050484083e-05, "loss": 0.5567, "step": 11469 }, { "epoch": 0.4086431408874717, "grad_norm": 0.9769780974870027, "learning_rate": 1.3377334393877952e-05, "loss": 0.442, "step": 11470 }, { "epoch": 0.40867876801396585, "grad_norm": 1.0023852127661517, "learning_rate": 1.3376248237943505e-05, "loss": 0.4356, "step": 11471 }, { "epoch": 0.40871439514045993, "grad_norm": 1.077782456429367, "learning_rate": 1.3375162037051955e-05, "loss": 0.4483, "step": 11472 }, { "epoch": 0.40875002226695406, "grad_norm": 1.3833317684922037, "learning_rate": 1.3374075791217758e-05, "loss": 0.5494, "step": 11473 }, { "epoch": 0.4087856493934482, "grad_norm": 1.7338902028220478, "learning_rate": 1.3372989500455389e-05, "loss": 0.5557, "step": 11474 }, { "epoch": 0.40882127651994227, "grad_norm": 1.0992400099957202, "learning_rate": 1.3371903164779306e-05, "loss": 0.4767, "step": 11475 }, { "epoch": 0.4088569036464364, "grad_norm": 1.0526388595158158, "learning_rate": 1.3370816784203973e-05, "loss": 0.4954, "step": 11476 }, { "epoch": 0.40889253077293053, "grad_norm": 0.9948268473309119, "learning_rate": 1.336973035874386e-05, "loss": 0.486, "step": 11477 }, { "epoch": 0.4089281578994246, "grad_norm": 1.0541094249409104, "learning_rate": 1.3368643888413431e-05, "loss": 0.4966, "step": 11478 }, { "epoch": 0.40896378502591874, "grad_norm": 1.093181292259563, "learning_rate": 1.3367557373227156e-05, "loss": 0.4725, "step": 11479 }, { "epoch": 0.40899941215241287, "grad_norm": 1.0751600622923358, "learning_rate": 1.33664708131995e-05, "loss": 0.492, "step": 11480 }, { "epoch": 0.40903503927890694, "grad_norm": 1.7226224594470232, "learning_rate": 1.3365384208344932e-05, "loss": 0.5996, "step": 11481 }, { "epoch": 0.4090706664054011, "grad_norm": 1.1244999827555728, "learning_rate": 1.3364297558677921e-05, "loss": 0.506, "step": 11482 }, { "epoch": 0.4091062935318952, "grad_norm": 1.158351378717956, "learning_rate": 1.3363210864212937e-05, "loss": 0.492, "step": 11483 }, { "epoch": 0.4091419206583893, "grad_norm": 1.9790490454321514, "learning_rate": 1.3362124124964449e-05, "loss": 0.5849, "step": 11484 }, { "epoch": 0.4091775477848834, "grad_norm": 0.985204693006152, "learning_rate": 1.3361037340946927e-05, "loss": 0.4325, "step": 11485 }, { "epoch": 0.40921317491137754, "grad_norm": 1.472507048989686, "learning_rate": 1.3359950512174847e-05, "loss": 0.5654, "step": 11486 }, { "epoch": 0.4092488020378716, "grad_norm": 1.0697430193274802, "learning_rate": 1.3358863638662673e-05, "loss": 0.4483, "step": 11487 }, { "epoch": 0.40928442916436575, "grad_norm": 1.4484268914613856, "learning_rate": 1.3357776720424887e-05, "loss": 0.6108, "step": 11488 }, { "epoch": 0.4093200562908599, "grad_norm": 1.0892883542641412, "learning_rate": 1.3356689757475954e-05, "loss": 0.4588, "step": 11489 }, { "epoch": 0.40935568341735395, "grad_norm": 1.0758529816456364, "learning_rate": 1.335560274983035e-05, "loss": 0.4533, "step": 11490 }, { "epoch": 0.4093913105438481, "grad_norm": 1.0464125428896944, "learning_rate": 1.3354515697502552e-05, "loss": 0.4547, "step": 11491 }, { "epoch": 0.4094269376703422, "grad_norm": 1.093976840159002, "learning_rate": 1.335342860050703e-05, "loss": 0.4709, "step": 11492 }, { "epoch": 0.4094625647968363, "grad_norm": 1.0116001744185878, "learning_rate": 1.3352341458858264e-05, "loss": 0.4739, "step": 11493 }, { "epoch": 0.4094981919233304, "grad_norm": 1.1090005158706153, "learning_rate": 1.3351254272570731e-05, "loss": 0.4573, "step": 11494 }, { "epoch": 0.40953381904982455, "grad_norm": 1.0905416728821078, "learning_rate": 1.3350167041658901e-05, "loss": 0.4529, "step": 11495 }, { "epoch": 0.40956944617631863, "grad_norm": 1.1231367986393965, "learning_rate": 1.3349079766137259e-05, "loss": 0.5041, "step": 11496 }, { "epoch": 0.40960507330281276, "grad_norm": 1.9485676373935172, "learning_rate": 1.3347992446020275e-05, "loss": 0.5935, "step": 11497 }, { "epoch": 0.4096407004293069, "grad_norm": 1.063200133943093, "learning_rate": 1.3346905081322435e-05, "loss": 0.473, "step": 11498 }, { "epoch": 0.40967632755580097, "grad_norm": 1.6767600956057547, "learning_rate": 1.3345817672058212e-05, "loss": 0.5274, "step": 11499 }, { "epoch": 0.4097119546822951, "grad_norm": 0.9960623117711571, "learning_rate": 1.3344730218242088e-05, "loss": 0.4384, "step": 11500 }, { "epoch": 0.40974758180878923, "grad_norm": 1.1040919632693103, "learning_rate": 1.3343642719888545e-05, "loss": 0.4897, "step": 11501 }, { "epoch": 0.4097832089352833, "grad_norm": 1.1229163681347276, "learning_rate": 1.334255517701206e-05, "loss": 0.5238, "step": 11502 }, { "epoch": 0.40981883606177744, "grad_norm": 1.4456647516411807, "learning_rate": 1.3341467589627118e-05, "loss": 0.5656, "step": 11503 }, { "epoch": 0.40985446318827157, "grad_norm": 1.0080360819767094, "learning_rate": 1.33403799577482e-05, "loss": 0.4637, "step": 11504 }, { "epoch": 0.40989009031476564, "grad_norm": 1.1400334107214853, "learning_rate": 1.3339292281389789e-05, "loss": 0.4734, "step": 11505 }, { "epoch": 0.4099257174412598, "grad_norm": 1.5801594070821918, "learning_rate": 1.3338204560566364e-05, "loss": 0.6519, "step": 11506 }, { "epoch": 0.4099613445677539, "grad_norm": 1.5692563359479108, "learning_rate": 1.3337116795292413e-05, "loss": 0.5923, "step": 11507 }, { "epoch": 0.409996971694248, "grad_norm": 1.1299303369652829, "learning_rate": 1.333602898558242e-05, "loss": 0.4737, "step": 11508 }, { "epoch": 0.4100325988207421, "grad_norm": 0.9882427941646937, "learning_rate": 1.333494113145087e-05, "loss": 0.445, "step": 11509 }, { "epoch": 0.41006822594723624, "grad_norm": 1.0126142629867267, "learning_rate": 1.3333853232912245e-05, "loss": 0.4472, "step": 11510 }, { "epoch": 0.4101038530737303, "grad_norm": 1.1946790162661172, "learning_rate": 1.3332765289981038e-05, "loss": 0.4619, "step": 11511 }, { "epoch": 0.41013948020022445, "grad_norm": 1.7126355471010093, "learning_rate": 1.333167730267173e-05, "loss": 0.5769, "step": 11512 }, { "epoch": 0.4101751073267186, "grad_norm": 1.2037693518978598, "learning_rate": 1.3330589270998809e-05, "loss": 0.5156, "step": 11513 }, { "epoch": 0.41021073445321266, "grad_norm": 1.5829143521224183, "learning_rate": 1.3329501194976764e-05, "loss": 0.5825, "step": 11514 }, { "epoch": 0.4102463615797068, "grad_norm": 1.3478463404287355, "learning_rate": 1.3328413074620086e-05, "loss": 0.5264, "step": 11515 }, { "epoch": 0.4102819887062009, "grad_norm": 1.1600569895996562, "learning_rate": 1.3327324909943257e-05, "loss": 0.4461, "step": 11516 }, { "epoch": 0.410317615832695, "grad_norm": 1.088866292273051, "learning_rate": 1.3326236700960777e-05, "loss": 0.4701, "step": 11517 }, { "epoch": 0.4103532429591891, "grad_norm": 1.6061904030311698, "learning_rate": 1.3325148447687126e-05, "loss": 0.6046, "step": 11518 }, { "epoch": 0.41038887008568325, "grad_norm": 1.6500784878122672, "learning_rate": 1.3324060150136799e-05, "loss": 0.5811, "step": 11519 }, { "epoch": 0.41042449721217733, "grad_norm": 1.0563980726772022, "learning_rate": 1.3322971808324292e-05, "loss": 0.445, "step": 11520 }, { "epoch": 0.41046012433867146, "grad_norm": 1.0876977633653429, "learning_rate": 1.3321883422264088e-05, "loss": 0.4952, "step": 11521 }, { "epoch": 0.4104957514651656, "grad_norm": 1.2510994645378377, "learning_rate": 1.3320794991970688e-05, "loss": 0.5487, "step": 11522 }, { "epoch": 0.41053137859165967, "grad_norm": 1.0767532529230313, "learning_rate": 1.3319706517458579e-05, "loss": 0.5138, "step": 11523 }, { "epoch": 0.4105670057181538, "grad_norm": 1.0419370454550394, "learning_rate": 1.331861799874226e-05, "loss": 0.4448, "step": 11524 }, { "epoch": 0.41060263284464793, "grad_norm": 1.1294465046653184, "learning_rate": 1.3317529435836219e-05, "loss": 0.4879, "step": 11525 }, { "epoch": 0.410638259971142, "grad_norm": 1.161243995539889, "learning_rate": 1.3316440828754957e-05, "loss": 0.5041, "step": 11526 }, { "epoch": 0.41067388709763614, "grad_norm": 1.103884869863139, "learning_rate": 1.3315352177512967e-05, "loss": 0.5033, "step": 11527 }, { "epoch": 0.41070951422413027, "grad_norm": 1.0298870968317204, "learning_rate": 1.3314263482124748e-05, "loss": 0.4938, "step": 11528 }, { "epoch": 0.41074514135062434, "grad_norm": 1.102318765231875, "learning_rate": 1.3313174742604788e-05, "loss": 0.484, "step": 11529 }, { "epoch": 0.4107807684771185, "grad_norm": 1.0434164535995143, "learning_rate": 1.3312085958967595e-05, "loss": 0.47, "step": 11530 }, { "epoch": 0.4108163956036126, "grad_norm": 1.112592017015409, "learning_rate": 1.3310997131227664e-05, "loss": 0.4946, "step": 11531 }, { "epoch": 0.4108520227301067, "grad_norm": 2.2901820261267027, "learning_rate": 1.330990825939949e-05, "loss": 0.5668, "step": 11532 }, { "epoch": 0.4108876498566008, "grad_norm": 1.1519039777539752, "learning_rate": 1.3308819343497572e-05, "loss": 0.4815, "step": 11533 }, { "epoch": 0.41092327698309494, "grad_norm": 1.570058942446511, "learning_rate": 1.3307730383536415e-05, "loss": 0.57, "step": 11534 }, { "epoch": 0.410958904109589, "grad_norm": 1.3627721455262876, "learning_rate": 1.3306641379530515e-05, "loss": 0.6328, "step": 11535 }, { "epoch": 0.41099453123608315, "grad_norm": 1.0689474196145747, "learning_rate": 1.3305552331494371e-05, "loss": 0.4372, "step": 11536 }, { "epoch": 0.4110301583625773, "grad_norm": 1.0771706808375952, "learning_rate": 1.3304463239442489e-05, "loss": 0.4476, "step": 11537 }, { "epoch": 0.41106578548907136, "grad_norm": 1.26655386671792, "learning_rate": 1.3303374103389371e-05, "loss": 0.5407, "step": 11538 }, { "epoch": 0.4111014126155655, "grad_norm": 2.0049812737492467, "learning_rate": 1.3302284923349517e-05, "loss": 0.5831, "step": 11539 }, { "epoch": 0.4111370397420596, "grad_norm": 1.1710562981796964, "learning_rate": 1.330119569933743e-05, "loss": 0.4651, "step": 11540 }, { "epoch": 0.4111726668685537, "grad_norm": 1.0916995549611517, "learning_rate": 1.3300106431367617e-05, "loss": 0.4591, "step": 11541 }, { "epoch": 0.4112082939950478, "grad_norm": 1.1583079729462504, "learning_rate": 1.3299017119454579e-05, "loss": 0.4777, "step": 11542 }, { "epoch": 0.41124392112154196, "grad_norm": 1.6000806075531777, "learning_rate": 1.3297927763612822e-05, "loss": 0.6386, "step": 11543 }, { "epoch": 0.41127954824803603, "grad_norm": 1.5303884711244868, "learning_rate": 1.3296838363856852e-05, "loss": 0.5161, "step": 11544 }, { "epoch": 0.41131517537453016, "grad_norm": 1.5867578824093103, "learning_rate": 1.3295748920201175e-05, "loss": 0.5599, "step": 11545 }, { "epoch": 0.4113508025010243, "grad_norm": 1.1230879313680513, "learning_rate": 1.3294659432660297e-05, "loss": 0.5199, "step": 11546 }, { "epoch": 0.41138642962751837, "grad_norm": 1.1073007610982988, "learning_rate": 1.3293569901248727e-05, "loss": 0.4564, "step": 11547 }, { "epoch": 0.4114220567540125, "grad_norm": 2.1623217943396376, "learning_rate": 1.3292480325980967e-05, "loss": 0.5158, "step": 11548 }, { "epoch": 0.41145768388050663, "grad_norm": 1.0305682296238405, "learning_rate": 1.3291390706871536e-05, "loss": 0.4988, "step": 11549 }, { "epoch": 0.4114933110070007, "grad_norm": 1.677170957899249, "learning_rate": 1.3290301043934935e-05, "loss": 0.4976, "step": 11550 }, { "epoch": 0.41152893813349484, "grad_norm": 1.4395416849143499, "learning_rate": 1.3289211337185674e-05, "loss": 0.5499, "step": 11551 }, { "epoch": 0.41156456525998897, "grad_norm": 1.6296430528255315, "learning_rate": 1.3288121586638268e-05, "loss": 0.6068, "step": 11552 }, { "epoch": 0.41160019238648304, "grad_norm": 1.8267376247709206, "learning_rate": 1.3287031792307226e-05, "loss": 0.6191, "step": 11553 }, { "epoch": 0.4116358195129772, "grad_norm": 1.5235699847324533, "learning_rate": 1.3285941954207055e-05, "loss": 0.5671, "step": 11554 }, { "epoch": 0.4116714466394713, "grad_norm": 1.161154626461286, "learning_rate": 1.328485207235227e-05, "loss": 0.4847, "step": 11555 }, { "epoch": 0.4117070737659654, "grad_norm": 2.2618788903962623, "learning_rate": 1.3283762146757387e-05, "loss": 0.5568, "step": 11556 }, { "epoch": 0.4117427008924595, "grad_norm": 1.5577046882571706, "learning_rate": 1.3282672177436913e-05, "loss": 0.5847, "step": 11557 }, { "epoch": 0.41177832801895364, "grad_norm": 1.098741606349478, "learning_rate": 1.3281582164405364e-05, "loss": 0.4992, "step": 11558 }, { "epoch": 0.4118139551454477, "grad_norm": 1.742526848388805, "learning_rate": 1.3280492107677255e-05, "loss": 0.5791, "step": 11559 }, { "epoch": 0.41184958227194185, "grad_norm": 1.3910189553986967, "learning_rate": 1.3279402007267101e-05, "loss": 0.4785, "step": 11560 }, { "epoch": 0.411885209398436, "grad_norm": 1.1272211931630163, "learning_rate": 1.327831186318942e-05, "loss": 0.4739, "step": 11561 }, { "epoch": 0.41192083652493006, "grad_norm": 1.0412929033388902, "learning_rate": 1.327722167545872e-05, "loss": 0.5058, "step": 11562 }, { "epoch": 0.4119564636514242, "grad_norm": 1.2329650608767715, "learning_rate": 1.3276131444089526e-05, "loss": 0.4832, "step": 11563 }, { "epoch": 0.4119920907779183, "grad_norm": 1.145013900809155, "learning_rate": 1.3275041169096351e-05, "loss": 0.4568, "step": 11564 }, { "epoch": 0.4120277179044124, "grad_norm": 1.163959629159746, "learning_rate": 1.3273950850493717e-05, "loss": 0.4815, "step": 11565 }, { "epoch": 0.4120633450309065, "grad_norm": 1.6748652102049701, "learning_rate": 1.3272860488296134e-05, "loss": 0.5178, "step": 11566 }, { "epoch": 0.41209897215740066, "grad_norm": 1.7537687528633672, "learning_rate": 1.327177008251813e-05, "loss": 0.5393, "step": 11567 }, { "epoch": 0.41213459928389473, "grad_norm": 2.1172588004001596, "learning_rate": 1.3270679633174219e-05, "loss": 0.6417, "step": 11568 }, { "epoch": 0.41217022641038886, "grad_norm": 1.1099881856754934, "learning_rate": 1.326958914027892e-05, "loss": 0.4681, "step": 11569 }, { "epoch": 0.412205853536883, "grad_norm": 1.7913066005893845, "learning_rate": 1.326849860384676e-05, "loss": 0.5658, "step": 11570 }, { "epoch": 0.41224148066337707, "grad_norm": 1.3052650175762586, "learning_rate": 1.3267408023892255e-05, "loss": 0.5178, "step": 11571 }, { "epoch": 0.4122771077898712, "grad_norm": 1.2105408966017157, "learning_rate": 1.3266317400429928e-05, "loss": 0.4878, "step": 11572 }, { "epoch": 0.41231273491636533, "grad_norm": 1.5016659630433404, "learning_rate": 1.3265226733474304e-05, "loss": 0.5465, "step": 11573 }, { "epoch": 0.4123483620428594, "grad_norm": 1.088160528133423, "learning_rate": 1.32641360230399e-05, "loss": 0.4487, "step": 11574 }, { "epoch": 0.41238398916935354, "grad_norm": 1.4995382290052959, "learning_rate": 1.3263045269141248e-05, "loss": 0.531, "step": 11575 }, { "epoch": 0.41241961629584767, "grad_norm": 1.8571414693412542, "learning_rate": 1.3261954471792864e-05, "loss": 0.5706, "step": 11576 }, { "epoch": 0.41245524342234174, "grad_norm": 1.0948917764344916, "learning_rate": 1.3260863631009277e-05, "loss": 0.4525, "step": 11577 }, { "epoch": 0.4124908705488359, "grad_norm": 1.2324094837065667, "learning_rate": 1.325977274680501e-05, "loss": 0.4593, "step": 11578 }, { "epoch": 0.41252649767533, "grad_norm": 1.0296605816637412, "learning_rate": 1.3258681819194595e-05, "loss": 0.4353, "step": 11579 }, { "epoch": 0.4125621248018241, "grad_norm": 1.1658419623696816, "learning_rate": 1.3257590848192552e-05, "loss": 0.5597, "step": 11580 }, { "epoch": 0.4125977519283182, "grad_norm": 1.7313153143731341, "learning_rate": 1.3256499833813407e-05, "loss": 0.5878, "step": 11581 }, { "epoch": 0.41263337905481234, "grad_norm": 1.5792417730887505, "learning_rate": 1.3255408776071692e-05, "loss": 0.5373, "step": 11582 }, { "epoch": 0.4126690061813064, "grad_norm": 1.9862786729318178, "learning_rate": 1.3254317674981935e-05, "loss": 0.6142, "step": 11583 }, { "epoch": 0.41270463330780055, "grad_norm": 1.4620142257834148, "learning_rate": 1.3253226530558663e-05, "loss": 0.6093, "step": 11584 }, { "epoch": 0.4127402604342947, "grad_norm": 1.1817343196902634, "learning_rate": 1.3252135342816406e-05, "loss": 0.4803, "step": 11585 }, { "epoch": 0.4127758875607888, "grad_norm": 1.599280640758527, "learning_rate": 1.3251044111769693e-05, "loss": 0.5373, "step": 11586 }, { "epoch": 0.4128115146872829, "grad_norm": 1.184112846997519, "learning_rate": 1.3249952837433058e-05, "loss": 0.514, "step": 11587 }, { "epoch": 0.412847141813777, "grad_norm": 1.127555588172507, "learning_rate": 1.3248861519821026e-05, "loss": 0.4599, "step": 11588 }, { "epoch": 0.41288276894027115, "grad_norm": 1.158341710355111, "learning_rate": 1.3247770158948131e-05, "loss": 0.4839, "step": 11589 }, { "epoch": 0.4129183960667652, "grad_norm": 1.0700529306055497, "learning_rate": 1.3246678754828913e-05, "loss": 0.4652, "step": 11590 }, { "epoch": 0.41295402319325936, "grad_norm": 1.2134523081005701, "learning_rate": 1.3245587307477891e-05, "loss": 0.4824, "step": 11591 }, { "epoch": 0.4129896503197535, "grad_norm": 1.5905683185720763, "learning_rate": 1.3244495816909608e-05, "loss": 0.5543, "step": 11592 }, { "epoch": 0.41302527744624756, "grad_norm": 1.3915300243990925, "learning_rate": 1.3243404283138596e-05, "loss": 0.4379, "step": 11593 }, { "epoch": 0.4130609045727417, "grad_norm": 1.1794217869321237, "learning_rate": 1.3242312706179388e-05, "loss": 0.4853, "step": 11594 }, { "epoch": 0.4130965316992358, "grad_norm": 1.402524970031595, "learning_rate": 1.324122108604652e-05, "loss": 0.5688, "step": 11595 }, { "epoch": 0.4131321588257299, "grad_norm": 1.139198142374718, "learning_rate": 1.3240129422754528e-05, "loss": 0.4567, "step": 11596 }, { "epoch": 0.41316778595222403, "grad_norm": 1.102042178286398, "learning_rate": 1.3239037716317951e-05, "loss": 0.3964, "step": 11597 }, { "epoch": 0.41320341307871816, "grad_norm": 1.6651789332704656, "learning_rate": 1.323794596675132e-05, "loss": 0.479, "step": 11598 }, { "epoch": 0.41323904020521224, "grad_norm": 1.2568402629636288, "learning_rate": 1.3236854174069177e-05, "loss": 0.4903, "step": 11599 }, { "epoch": 0.41327466733170637, "grad_norm": 1.0273659962381874, "learning_rate": 1.3235762338286056e-05, "loss": 0.4485, "step": 11600 }, { "epoch": 0.4133102944582005, "grad_norm": 1.1007681867615573, "learning_rate": 1.3234670459416498e-05, "loss": 0.4619, "step": 11601 }, { "epoch": 0.4133459215846946, "grad_norm": 1.7494552021581558, "learning_rate": 1.3233578537475043e-05, "loss": 0.5067, "step": 11602 }, { "epoch": 0.4133815487111887, "grad_norm": 1.395751206241786, "learning_rate": 1.3232486572476229e-05, "loss": 0.5274, "step": 11603 }, { "epoch": 0.41341717583768284, "grad_norm": 1.0368117131267296, "learning_rate": 1.3231394564434596e-05, "loss": 0.4592, "step": 11604 }, { "epoch": 0.4134528029641769, "grad_norm": 1.2144957372762601, "learning_rate": 1.3230302513364689e-05, "loss": 0.4985, "step": 11605 }, { "epoch": 0.41348843009067104, "grad_norm": 1.6212580691898968, "learning_rate": 1.3229210419281042e-05, "loss": 0.6233, "step": 11606 }, { "epoch": 0.4135240572171652, "grad_norm": 1.2187178632155473, "learning_rate": 1.3228118282198202e-05, "loss": 0.4578, "step": 11607 }, { "epoch": 0.41355968434365925, "grad_norm": 1.2686735485148317, "learning_rate": 1.3227026102130713e-05, "loss": 0.4824, "step": 11608 }, { "epoch": 0.4135953114701534, "grad_norm": 1.2005041581361364, "learning_rate": 1.3225933879093116e-05, "loss": 0.4767, "step": 11609 }, { "epoch": 0.4136309385966475, "grad_norm": 1.5279093693625914, "learning_rate": 1.3224841613099955e-05, "loss": 0.6227, "step": 11610 }, { "epoch": 0.4136665657231416, "grad_norm": 1.802970615789746, "learning_rate": 1.322374930416577e-05, "loss": 0.5574, "step": 11611 }, { "epoch": 0.4137021928496357, "grad_norm": 1.1389514976754613, "learning_rate": 1.3222656952305113e-05, "loss": 0.4608, "step": 11612 }, { "epoch": 0.41373781997612985, "grad_norm": 1.1033812186045757, "learning_rate": 1.3221564557532526e-05, "loss": 0.4599, "step": 11613 }, { "epoch": 0.4137734471026239, "grad_norm": 1.3873860569767142, "learning_rate": 1.3220472119862553e-05, "loss": 0.4878, "step": 11614 }, { "epoch": 0.41380907422911806, "grad_norm": 1.676528716179908, "learning_rate": 1.3219379639309743e-05, "loss": 0.5778, "step": 11615 }, { "epoch": 0.4138447013556122, "grad_norm": 1.4388569237007727, "learning_rate": 1.3218287115888647e-05, "loss": 0.592, "step": 11616 }, { "epoch": 0.41388032848210626, "grad_norm": 1.4555416087995179, "learning_rate": 1.3217194549613804e-05, "loss": 0.5213, "step": 11617 }, { "epoch": 0.4139159556086004, "grad_norm": 1.5340889565294158, "learning_rate": 1.3216101940499768e-05, "loss": 0.511, "step": 11618 }, { "epoch": 0.4139515827350945, "grad_norm": 1.3060618060944031, "learning_rate": 1.3215009288561087e-05, "loss": 0.487, "step": 11619 }, { "epoch": 0.4139872098615886, "grad_norm": 1.0189323967070039, "learning_rate": 1.3213916593812313e-05, "loss": 0.4151, "step": 11620 }, { "epoch": 0.41402283698808273, "grad_norm": 1.113109146953976, "learning_rate": 1.3212823856267991e-05, "loss": 0.4852, "step": 11621 }, { "epoch": 0.41405846411457686, "grad_norm": 1.7059667400328904, "learning_rate": 1.321173107594267e-05, "loss": 0.5361, "step": 11622 }, { "epoch": 0.41409409124107094, "grad_norm": 2.293294411628448, "learning_rate": 1.321063825285091e-05, "loss": 0.5396, "step": 11623 }, { "epoch": 0.41412971836756507, "grad_norm": 1.0258515835093416, "learning_rate": 1.3209545387007257e-05, "loss": 0.4187, "step": 11624 }, { "epoch": 0.4141653454940592, "grad_norm": 1.4945153368527777, "learning_rate": 1.3208452478426261e-05, "loss": 0.6137, "step": 11625 }, { "epoch": 0.4142009726205533, "grad_norm": 1.2055405533503072, "learning_rate": 1.3207359527122479e-05, "loss": 0.5054, "step": 11626 }, { "epoch": 0.4142365997470474, "grad_norm": 1.1990168861677342, "learning_rate": 1.3206266533110466e-05, "loss": 0.485, "step": 11627 }, { "epoch": 0.41427222687354154, "grad_norm": 1.1545447257261747, "learning_rate": 1.3205173496404768e-05, "loss": 0.4663, "step": 11628 }, { "epoch": 0.4143078540000356, "grad_norm": 1.9304242065844675, "learning_rate": 1.320408041701995e-05, "loss": 0.5768, "step": 11629 }, { "epoch": 0.41434348112652974, "grad_norm": 1.1251184801023983, "learning_rate": 1.3202987294970555e-05, "loss": 0.4802, "step": 11630 }, { "epoch": 0.4143791082530239, "grad_norm": 1.0224765935031936, "learning_rate": 1.3201894130271148e-05, "loss": 0.4646, "step": 11631 }, { "epoch": 0.41441473537951795, "grad_norm": 1.105075358355137, "learning_rate": 1.3200800922936287e-05, "loss": 0.4863, "step": 11632 }, { "epoch": 0.4144503625060121, "grad_norm": 1.01089222511058, "learning_rate": 1.319970767298052e-05, "loss": 0.4805, "step": 11633 }, { "epoch": 0.4144859896325062, "grad_norm": 1.1899100583554063, "learning_rate": 1.3198614380418412e-05, "loss": 0.461, "step": 11634 }, { "epoch": 0.4145216167590003, "grad_norm": 1.8614252004170497, "learning_rate": 1.3197521045264517e-05, "loss": 0.6017, "step": 11635 }, { "epoch": 0.4145572438854944, "grad_norm": 1.134962115834102, "learning_rate": 1.3196427667533393e-05, "loss": 0.466, "step": 11636 }, { "epoch": 0.41459287101198855, "grad_norm": 1.8728805457231532, "learning_rate": 1.31953342472396e-05, "loss": 0.5532, "step": 11637 }, { "epoch": 0.4146284981384826, "grad_norm": 1.4256179772963904, "learning_rate": 1.3194240784397701e-05, "loss": 0.5721, "step": 11638 }, { "epoch": 0.41466412526497676, "grad_norm": 1.008724745975045, "learning_rate": 1.319314727902225e-05, "loss": 0.4358, "step": 11639 }, { "epoch": 0.4146997523914709, "grad_norm": 1.1551911941587674, "learning_rate": 1.3192053731127812e-05, "loss": 0.4822, "step": 11640 }, { "epoch": 0.41473537951796496, "grad_norm": 1.0716011991615846, "learning_rate": 1.3190960140728946e-05, "loss": 0.4579, "step": 11641 }, { "epoch": 0.4147710066444591, "grad_norm": 1.142031284560488, "learning_rate": 1.3189866507840217e-05, "loss": 0.4275, "step": 11642 }, { "epoch": 0.4148066337709532, "grad_norm": 1.055187737813547, "learning_rate": 1.318877283247619e-05, "loss": 0.4486, "step": 11643 }, { "epoch": 0.4148422608974473, "grad_norm": 1.2419631880029687, "learning_rate": 1.3187679114651415e-05, "loss": 0.4774, "step": 11644 }, { "epoch": 0.41487788802394143, "grad_norm": 1.0658039083610344, "learning_rate": 1.318658535438047e-05, "loss": 0.421, "step": 11645 }, { "epoch": 0.41491351515043556, "grad_norm": 2.0637503396773287, "learning_rate": 1.3185491551677917e-05, "loss": 0.6575, "step": 11646 }, { "epoch": 0.41494914227692964, "grad_norm": 2.0758775798297435, "learning_rate": 1.3184397706558314e-05, "loss": 0.514, "step": 11647 }, { "epoch": 0.41498476940342377, "grad_norm": 1.5757268231235524, "learning_rate": 1.3183303819036226e-05, "loss": 0.6057, "step": 11648 }, { "epoch": 0.4150203965299179, "grad_norm": 1.6108914515073285, "learning_rate": 1.3182209889126228e-05, "loss": 0.5952, "step": 11649 }, { "epoch": 0.415056023656412, "grad_norm": 1.16408564357264, "learning_rate": 1.318111591684288e-05, "loss": 0.4844, "step": 11650 }, { "epoch": 0.4150916507829061, "grad_norm": 1.1604141755595176, "learning_rate": 1.3180021902200749e-05, "loss": 0.518, "step": 11651 }, { "epoch": 0.41512727790940024, "grad_norm": 1.7521392489042975, "learning_rate": 1.3178927845214402e-05, "loss": 0.569, "step": 11652 }, { "epoch": 0.4151629050358943, "grad_norm": 1.1209705694811416, "learning_rate": 1.3177833745898409e-05, "loss": 0.4232, "step": 11653 }, { "epoch": 0.41519853216238845, "grad_norm": 1.0292477441866035, "learning_rate": 1.3176739604267344e-05, "loss": 0.4318, "step": 11654 }, { "epoch": 0.4152341592888826, "grad_norm": 1.907232413400267, "learning_rate": 1.3175645420335768e-05, "loss": 0.5404, "step": 11655 }, { "epoch": 0.41526978641537665, "grad_norm": 1.0574555253157658, "learning_rate": 1.317455119411825e-05, "loss": 0.4584, "step": 11656 }, { "epoch": 0.4153054135418708, "grad_norm": 1.1217927627157867, "learning_rate": 1.3173456925629368e-05, "loss": 0.4345, "step": 11657 }, { "epoch": 0.4153410406683649, "grad_norm": 2.3746801515426457, "learning_rate": 1.3172362614883687e-05, "loss": 0.5764, "step": 11658 }, { "epoch": 0.415376667794859, "grad_norm": 3.05212918529668, "learning_rate": 1.317126826189578e-05, "loss": 0.6519, "step": 11659 }, { "epoch": 0.4154122949213531, "grad_norm": 1.762155835801615, "learning_rate": 1.317017386668022e-05, "loss": 0.649, "step": 11660 }, { "epoch": 0.41544792204784725, "grad_norm": 1.4230639281401007, "learning_rate": 1.3169079429251581e-05, "loss": 0.5878, "step": 11661 }, { "epoch": 0.4154835491743413, "grad_norm": 1.4368614530755062, "learning_rate": 1.3167984949624436e-05, "loss": 0.5092, "step": 11662 }, { "epoch": 0.41551917630083546, "grad_norm": 1.6934781488737685, "learning_rate": 1.3166890427813348e-05, "loss": 0.5726, "step": 11663 }, { "epoch": 0.4155548034273296, "grad_norm": 1.1027035909896268, "learning_rate": 1.3165795863832909e-05, "loss": 0.441, "step": 11664 }, { "epoch": 0.41559043055382366, "grad_norm": 1.0565531937555752, "learning_rate": 1.3164701257697683e-05, "loss": 0.442, "step": 11665 }, { "epoch": 0.4156260576803178, "grad_norm": 1.490507567293266, "learning_rate": 1.3163606609422249e-05, "loss": 0.5305, "step": 11666 }, { "epoch": 0.4156616848068119, "grad_norm": 1.2244361742902918, "learning_rate": 1.316251191902118e-05, "loss": 0.4738, "step": 11667 }, { "epoch": 0.415697311933306, "grad_norm": 1.759428042571779, "learning_rate": 1.3161417186509053e-05, "loss": 0.5444, "step": 11668 }, { "epoch": 0.41573293905980013, "grad_norm": 1.0268389785870045, "learning_rate": 1.316032241190045e-05, "loss": 0.4275, "step": 11669 }, { "epoch": 0.41576856618629426, "grad_norm": 1.0827118198234467, "learning_rate": 1.3159227595209942e-05, "loss": 0.4994, "step": 11670 }, { "epoch": 0.41580419331278834, "grad_norm": 1.8257803895222953, "learning_rate": 1.3158132736452111e-05, "loss": 0.6491, "step": 11671 }, { "epoch": 0.41583982043928247, "grad_norm": 1.4453620142246761, "learning_rate": 1.3157037835641537e-05, "loss": 0.6115, "step": 11672 }, { "epoch": 0.4158754475657766, "grad_norm": 1.0985454609731953, "learning_rate": 1.3155942892792796e-05, "loss": 0.4768, "step": 11673 }, { "epoch": 0.4159110746922707, "grad_norm": 1.14650096745358, "learning_rate": 1.3154847907920472e-05, "loss": 0.5453, "step": 11674 }, { "epoch": 0.4159467018187648, "grad_norm": 1.1176747298057683, "learning_rate": 1.315375288103914e-05, "loss": 0.5022, "step": 11675 }, { "epoch": 0.41598232894525894, "grad_norm": 1.4870878317552403, "learning_rate": 1.3152657812163385e-05, "loss": 0.6642, "step": 11676 }, { "epoch": 0.416017956071753, "grad_norm": 1.093054188866769, "learning_rate": 1.3151562701307787e-05, "loss": 0.4481, "step": 11677 }, { "epoch": 0.41605358319824715, "grad_norm": 1.6003103095262219, "learning_rate": 1.3150467548486929e-05, "loss": 0.6501, "step": 11678 }, { "epoch": 0.4160892103247413, "grad_norm": 1.5455118487125477, "learning_rate": 1.3149372353715395e-05, "loss": 0.5775, "step": 11679 }, { "epoch": 0.41612483745123535, "grad_norm": 1.8605450216690953, "learning_rate": 1.3148277117007765e-05, "loss": 0.6144, "step": 11680 }, { "epoch": 0.4161604645777295, "grad_norm": 1.083637045479651, "learning_rate": 1.3147181838378628e-05, "loss": 0.466, "step": 11681 }, { "epoch": 0.4161960917042236, "grad_norm": 1.6518606367064441, "learning_rate": 1.314608651784256e-05, "loss": 0.5219, "step": 11682 }, { "epoch": 0.4162317188307177, "grad_norm": 1.1952118651039283, "learning_rate": 1.3144991155414155e-05, "loss": 0.4573, "step": 11683 }, { "epoch": 0.4162673459572118, "grad_norm": 1.5636527262089301, "learning_rate": 1.3143895751107993e-05, "loss": 0.5281, "step": 11684 }, { "epoch": 0.41630297308370595, "grad_norm": 1.3315881612197071, "learning_rate": 1.3142800304938662e-05, "loss": 0.5625, "step": 11685 }, { "epoch": 0.4163386002102, "grad_norm": 0.9624183335100944, "learning_rate": 1.3141704816920746e-05, "loss": 0.4971, "step": 11686 }, { "epoch": 0.41637422733669416, "grad_norm": 1.529454983329393, "learning_rate": 1.3140609287068838e-05, "loss": 0.5679, "step": 11687 }, { "epoch": 0.4164098544631883, "grad_norm": 1.205482274218746, "learning_rate": 1.3139513715397521e-05, "loss": 0.5063, "step": 11688 }, { "epoch": 0.41644548158968236, "grad_norm": 1.3846596302836987, "learning_rate": 1.3138418101921384e-05, "loss": 0.5742, "step": 11689 }, { "epoch": 0.4164811087161765, "grad_norm": 1.4803189968702124, "learning_rate": 1.3137322446655017e-05, "loss": 0.552, "step": 11690 }, { "epoch": 0.4165167358426706, "grad_norm": 1.081186172204036, "learning_rate": 1.3136226749613009e-05, "loss": 0.4736, "step": 11691 }, { "epoch": 0.4165523629691647, "grad_norm": 1.866168999274189, "learning_rate": 1.313513101080995e-05, "loss": 0.4586, "step": 11692 }, { "epoch": 0.41658799009565883, "grad_norm": 1.2168602347835782, "learning_rate": 1.3134035230260427e-05, "loss": 0.5076, "step": 11693 }, { "epoch": 0.41662361722215296, "grad_norm": 1.6053896807300416, "learning_rate": 1.3132939407979038e-05, "loss": 0.5673, "step": 11694 }, { "epoch": 0.41665924434864704, "grad_norm": 1.1166886761007835, "learning_rate": 1.3131843543980369e-05, "loss": 0.4743, "step": 11695 }, { "epoch": 0.41669487147514117, "grad_norm": 2.914860246747415, "learning_rate": 1.3130747638279016e-05, "loss": 0.6002, "step": 11696 }, { "epoch": 0.4167304986016353, "grad_norm": 2.351834541326889, "learning_rate": 1.3129651690889569e-05, "loss": 0.5225, "step": 11697 }, { "epoch": 0.4167661257281294, "grad_norm": 1.2079087256193142, "learning_rate": 1.312855570182662e-05, "loss": 0.4404, "step": 11698 }, { "epoch": 0.4168017528546235, "grad_norm": 1.0985857799275287, "learning_rate": 1.3127459671104767e-05, "loss": 0.3989, "step": 11699 }, { "epoch": 0.41683737998111764, "grad_norm": 1.8166276404344865, "learning_rate": 1.3126363598738602e-05, "loss": 0.5678, "step": 11700 }, { "epoch": 0.4168730071076117, "grad_norm": 1.6101564854387762, "learning_rate": 1.3125267484742721e-05, "loss": 0.6088, "step": 11701 }, { "epoch": 0.41690863423410585, "grad_norm": 1.396646472165336, "learning_rate": 1.312417132913172e-05, "loss": 0.5184, "step": 11702 }, { "epoch": 0.4169442613606, "grad_norm": 1.1091864528906155, "learning_rate": 1.3123075131920196e-05, "loss": 0.4413, "step": 11703 }, { "epoch": 0.41697988848709405, "grad_norm": 1.0355064756824233, "learning_rate": 1.312197889312274e-05, "loss": 0.4218, "step": 11704 }, { "epoch": 0.4170155156135882, "grad_norm": 1.356509110546986, "learning_rate": 1.3120882612753953e-05, "loss": 0.4724, "step": 11705 }, { "epoch": 0.4170511427400823, "grad_norm": 2.202867727370401, "learning_rate": 1.3119786290828437e-05, "loss": 0.5559, "step": 11706 }, { "epoch": 0.4170867698665764, "grad_norm": 1.8477597643078039, "learning_rate": 1.311868992736078e-05, "loss": 0.5777, "step": 11707 }, { "epoch": 0.4171223969930705, "grad_norm": 1.0596033575273203, "learning_rate": 1.3117593522365588e-05, "loss": 0.4767, "step": 11708 }, { "epoch": 0.41715802411956465, "grad_norm": 1.2059689469300794, "learning_rate": 1.311649707585746e-05, "loss": 0.451, "step": 11709 }, { "epoch": 0.4171936512460587, "grad_norm": 1.0769493713730067, "learning_rate": 1.3115400587850996e-05, "loss": 0.4397, "step": 11710 }, { "epoch": 0.41722927837255286, "grad_norm": 1.1087338233568407, "learning_rate": 1.3114304058360795e-05, "loss": 0.433, "step": 11711 }, { "epoch": 0.417264905499047, "grad_norm": 1.1896704085838135, "learning_rate": 1.3113207487401458e-05, "loss": 0.4745, "step": 11712 }, { "epoch": 0.41730053262554107, "grad_norm": 1.1321633708058048, "learning_rate": 1.311211087498759e-05, "loss": 0.4741, "step": 11713 }, { "epoch": 0.4173361597520352, "grad_norm": 1.0456584834415406, "learning_rate": 1.3111014221133789e-05, "loss": 0.4786, "step": 11714 }, { "epoch": 0.4173717868785293, "grad_norm": 1.1199866182282188, "learning_rate": 1.3109917525854657e-05, "loss": 0.4913, "step": 11715 }, { "epoch": 0.4174074140050234, "grad_norm": 1.1277120159086476, "learning_rate": 1.3108820789164802e-05, "loss": 0.4578, "step": 11716 }, { "epoch": 0.41744304113151753, "grad_norm": 1.32515841369976, "learning_rate": 1.3107724011078826e-05, "loss": 0.5125, "step": 11717 }, { "epoch": 0.41747866825801166, "grad_norm": 1.0154670897761415, "learning_rate": 1.3106627191611333e-05, "loss": 0.4314, "step": 11718 }, { "epoch": 0.41751429538450574, "grad_norm": 1.8851319412703285, "learning_rate": 1.3105530330776924e-05, "loss": 0.6222, "step": 11719 }, { "epoch": 0.41754992251099987, "grad_norm": 1.3763101811335317, "learning_rate": 1.3104433428590209e-05, "loss": 0.5859, "step": 11720 }, { "epoch": 0.417585549637494, "grad_norm": 1.0791404298301428, "learning_rate": 1.3103336485065791e-05, "loss": 0.4553, "step": 11721 }, { "epoch": 0.4176211767639881, "grad_norm": 1.6562205764917812, "learning_rate": 1.3102239500218283e-05, "loss": 0.6216, "step": 11722 }, { "epoch": 0.4176568038904822, "grad_norm": 1.089259287810774, "learning_rate": 1.3101142474062284e-05, "loss": 0.4809, "step": 11723 }, { "epoch": 0.41769243101697634, "grad_norm": 1.1413274543920982, "learning_rate": 1.310004540661241e-05, "loss": 0.5381, "step": 11724 }, { "epoch": 0.4177280581434704, "grad_norm": 1.928225072578733, "learning_rate": 1.3098948297883263e-05, "loss": 0.5841, "step": 11725 }, { "epoch": 0.41776368526996455, "grad_norm": 1.0606336325571437, "learning_rate": 1.3097851147889452e-05, "loss": 0.4608, "step": 11726 }, { "epoch": 0.4177993123964587, "grad_norm": 1.641630225889918, "learning_rate": 1.3096753956645586e-05, "loss": 0.6085, "step": 11727 }, { "epoch": 0.41783493952295275, "grad_norm": 1.0787900806547732, "learning_rate": 1.3095656724166282e-05, "loss": 0.4309, "step": 11728 }, { "epoch": 0.4178705666494469, "grad_norm": 1.8707937154741503, "learning_rate": 1.3094559450466143e-05, "loss": 0.5523, "step": 11729 }, { "epoch": 0.417906193775941, "grad_norm": 1.0512741101434704, "learning_rate": 1.3093462135559778e-05, "loss": 0.43, "step": 11730 }, { "epoch": 0.4179418209024351, "grad_norm": 1.0680660665179926, "learning_rate": 1.3092364779461805e-05, "loss": 0.4677, "step": 11731 }, { "epoch": 0.4179774480289292, "grad_norm": 1.0575345613221385, "learning_rate": 1.3091267382186832e-05, "loss": 0.4866, "step": 11732 }, { "epoch": 0.41801307515542335, "grad_norm": 1.7113478546400949, "learning_rate": 1.3090169943749475e-05, "loss": 0.5634, "step": 11733 }, { "epoch": 0.41804870228191743, "grad_norm": 1.7999140308627206, "learning_rate": 1.3089072464164345e-05, "loss": 0.5437, "step": 11734 }, { "epoch": 0.41808432940841156, "grad_norm": 1.6830671329880724, "learning_rate": 1.3087974943446055e-05, "loss": 0.5341, "step": 11735 }, { "epoch": 0.4181199565349057, "grad_norm": 1.4913269714188306, "learning_rate": 1.3086877381609223e-05, "loss": 0.6508, "step": 11736 }, { "epoch": 0.41815558366139977, "grad_norm": 1.6538560432636986, "learning_rate": 1.308577977866846e-05, "loss": 0.564, "step": 11737 }, { "epoch": 0.4181912107878939, "grad_norm": 1.8012932819414311, "learning_rate": 1.308468213463838e-05, "loss": 0.5411, "step": 11738 }, { "epoch": 0.418226837914388, "grad_norm": 1.0238506946904766, "learning_rate": 1.3083584449533605e-05, "loss": 0.4513, "step": 11739 }, { "epoch": 0.4182624650408821, "grad_norm": 1.1156706045699938, "learning_rate": 1.3082486723368746e-05, "loss": 0.487, "step": 11740 }, { "epoch": 0.41829809216737623, "grad_norm": 2.2861324403501837, "learning_rate": 1.3081388956158421e-05, "loss": 0.5759, "step": 11741 }, { "epoch": 0.41833371929387037, "grad_norm": 1.5837845942869329, "learning_rate": 1.308029114791725e-05, "loss": 0.5556, "step": 11742 }, { "epoch": 0.41836934642036444, "grad_norm": 1.0504221673129648, "learning_rate": 1.3079193298659849e-05, "loss": 0.4376, "step": 11743 }, { "epoch": 0.41840497354685857, "grad_norm": 1.9264650929888323, "learning_rate": 1.3078095408400834e-05, "loss": 0.5464, "step": 11744 }, { "epoch": 0.4184406006733527, "grad_norm": 1.074208686738992, "learning_rate": 1.3076997477154829e-05, "loss": 0.4117, "step": 11745 }, { "epoch": 0.4184762277998468, "grad_norm": 1.570240698614983, "learning_rate": 1.3075899504936454e-05, "loss": 0.5588, "step": 11746 }, { "epoch": 0.4185118549263409, "grad_norm": 1.0978974755987245, "learning_rate": 1.3074801491760326e-05, "loss": 0.4635, "step": 11747 }, { "epoch": 0.41854748205283504, "grad_norm": 1.061449452074002, "learning_rate": 1.3073703437641066e-05, "loss": 0.4787, "step": 11748 }, { "epoch": 0.4185831091793291, "grad_norm": 1.4988197574745108, "learning_rate": 1.3072605342593297e-05, "loss": 0.5999, "step": 11749 }, { "epoch": 0.41861873630582325, "grad_norm": 1.2134944475778182, "learning_rate": 1.307150720663164e-05, "loss": 0.4558, "step": 11750 }, { "epoch": 0.4186543634323174, "grad_norm": 1.0503405561344379, "learning_rate": 1.3070409029770717e-05, "loss": 0.5276, "step": 11751 }, { "epoch": 0.41868999055881145, "grad_norm": 1.9425418014352103, "learning_rate": 1.3069310812025151e-05, "loss": 0.5227, "step": 11752 }, { "epoch": 0.4187256176853056, "grad_norm": 1.6741982928161898, "learning_rate": 1.306821255340957e-05, "loss": 0.5198, "step": 11753 }, { "epoch": 0.4187612448117997, "grad_norm": 1.6552412438222865, "learning_rate": 1.3067114253938593e-05, "loss": 0.5555, "step": 11754 }, { "epoch": 0.4187968719382938, "grad_norm": 1.0636081082401936, "learning_rate": 1.3066015913626841e-05, "loss": 0.4525, "step": 11755 }, { "epoch": 0.4188324990647879, "grad_norm": 1.12523577136733, "learning_rate": 1.306491753248895e-05, "loss": 0.4861, "step": 11756 }, { "epoch": 0.41886812619128205, "grad_norm": 1.4560906630295984, "learning_rate": 1.3063819110539537e-05, "loss": 0.5327, "step": 11757 }, { "epoch": 0.41890375331777613, "grad_norm": 1.036276656325476, "learning_rate": 1.3062720647793231e-05, "loss": 0.463, "step": 11758 }, { "epoch": 0.41893938044427026, "grad_norm": 1.0746308367673967, "learning_rate": 1.306162214426466e-05, "loss": 0.4814, "step": 11759 }, { "epoch": 0.4189750075707644, "grad_norm": 1.1742280451468645, "learning_rate": 1.306052359996845e-05, "loss": 0.4235, "step": 11760 }, { "epoch": 0.41901063469725847, "grad_norm": 1.0900579170086564, "learning_rate": 1.3059425014919229e-05, "loss": 0.453, "step": 11761 }, { "epoch": 0.4190462618237526, "grad_norm": 1.6000472818514202, "learning_rate": 1.3058326389131626e-05, "loss": 0.5543, "step": 11762 }, { "epoch": 0.41908188895024673, "grad_norm": 1.0326237735231079, "learning_rate": 1.3057227722620272e-05, "loss": 0.4297, "step": 11763 }, { "epoch": 0.4191175160767408, "grad_norm": 1.4631152172824766, "learning_rate": 1.3056129015399787e-05, "loss": 0.5462, "step": 11764 }, { "epoch": 0.41915314320323493, "grad_norm": 1.0483114024195193, "learning_rate": 1.3055030267484814e-05, "loss": 0.4738, "step": 11765 }, { "epoch": 0.41918877032972907, "grad_norm": 1.1010560960786664, "learning_rate": 1.3053931478889976e-05, "loss": 0.3926, "step": 11766 }, { "epoch": 0.41922439745622314, "grad_norm": 1.1156361831970445, "learning_rate": 1.3052832649629905e-05, "loss": 0.4856, "step": 11767 }, { "epoch": 0.4192600245827173, "grad_norm": 1.4702165543473653, "learning_rate": 1.3051733779719233e-05, "loss": 0.5654, "step": 11768 }, { "epoch": 0.4192956517092114, "grad_norm": 1.0547932597671479, "learning_rate": 1.3050634869172594e-05, "loss": 0.4583, "step": 11769 }, { "epoch": 0.4193312788357055, "grad_norm": 1.0110633214383002, "learning_rate": 1.3049535918004618e-05, "loss": 0.4589, "step": 11770 }, { "epoch": 0.4193669059621996, "grad_norm": 1.1108192012150746, "learning_rate": 1.3048436926229942e-05, "loss": 0.4475, "step": 11771 }, { "epoch": 0.41940253308869374, "grad_norm": 1.2552306796870716, "learning_rate": 1.3047337893863197e-05, "loss": 0.6257, "step": 11772 }, { "epoch": 0.4194381602151878, "grad_norm": 1.0475843589114784, "learning_rate": 1.3046238820919018e-05, "loss": 0.4819, "step": 11773 }, { "epoch": 0.41947378734168195, "grad_norm": 1.1692561115277718, "learning_rate": 1.3045139707412041e-05, "loss": 0.4649, "step": 11774 }, { "epoch": 0.4195094144681761, "grad_norm": 1.192744233058873, "learning_rate": 1.3044040553356896e-05, "loss": 0.4974, "step": 11775 }, { "epoch": 0.4195450415946702, "grad_norm": 1.2054281098554216, "learning_rate": 1.3042941358768227e-05, "loss": 0.5129, "step": 11776 }, { "epoch": 0.4195806687211643, "grad_norm": 1.1213809733829183, "learning_rate": 1.3041842123660667e-05, "loss": 0.488, "step": 11777 }, { "epoch": 0.4196162958476584, "grad_norm": 1.1369335002666439, "learning_rate": 1.304074284804885e-05, "loss": 0.4737, "step": 11778 }, { "epoch": 0.41965192297415255, "grad_norm": 1.4655749458871052, "learning_rate": 1.3039643531947417e-05, "loss": 0.5944, "step": 11779 }, { "epoch": 0.4196875501006466, "grad_norm": 1.5046404871017205, "learning_rate": 1.3038544175371009e-05, "loss": 0.5445, "step": 11780 }, { "epoch": 0.41972317722714075, "grad_norm": 2.666362205582818, "learning_rate": 1.303744477833426e-05, "loss": 0.5816, "step": 11781 }, { "epoch": 0.4197588043536349, "grad_norm": 1.362243054118919, "learning_rate": 1.3036345340851813e-05, "loss": 0.5549, "step": 11782 }, { "epoch": 0.41979443148012896, "grad_norm": 1.1388068774301598, "learning_rate": 1.30352458629383e-05, "loss": 0.5024, "step": 11783 }, { "epoch": 0.4198300586066231, "grad_norm": 1.207646218687078, "learning_rate": 1.3034146344608372e-05, "loss": 0.4709, "step": 11784 }, { "epoch": 0.4198656857331172, "grad_norm": 1.465577264909395, "learning_rate": 1.3033046785876665e-05, "loss": 0.5063, "step": 11785 }, { "epoch": 0.4199013128596113, "grad_norm": 1.5012913715061407, "learning_rate": 1.3031947186757814e-05, "loss": 0.5629, "step": 11786 }, { "epoch": 0.41993693998610543, "grad_norm": 1.4247145424988619, "learning_rate": 1.3030847547266472e-05, "loss": 0.6612, "step": 11787 }, { "epoch": 0.41997256711259956, "grad_norm": 1.183652360477818, "learning_rate": 1.3029747867417275e-05, "loss": 0.4906, "step": 11788 }, { "epoch": 0.42000819423909364, "grad_norm": 1.1510255031030487, "learning_rate": 1.3028648147224867e-05, "loss": 0.4961, "step": 11789 }, { "epoch": 0.42004382136558777, "grad_norm": 1.0459482994337683, "learning_rate": 1.302754838670389e-05, "loss": 0.4668, "step": 11790 }, { "epoch": 0.4200794484920819, "grad_norm": 1.0870825202385752, "learning_rate": 1.3026448585868994e-05, "loss": 0.4199, "step": 11791 }, { "epoch": 0.420115075618576, "grad_norm": 1.0956713032331487, "learning_rate": 1.302534874473482e-05, "loss": 0.4474, "step": 11792 }, { "epoch": 0.4201507027450701, "grad_norm": 1.1430144791935315, "learning_rate": 1.3024248863316012e-05, "loss": 0.4992, "step": 11793 }, { "epoch": 0.42018632987156423, "grad_norm": 1.0788122612671993, "learning_rate": 1.3023148941627215e-05, "loss": 0.5149, "step": 11794 }, { "epoch": 0.4202219569980583, "grad_norm": 1.1977604123750496, "learning_rate": 1.302204897968308e-05, "loss": 0.4395, "step": 11795 }, { "epoch": 0.42025758412455244, "grad_norm": 1.4717392549197095, "learning_rate": 1.3020948977498246e-05, "loss": 0.5188, "step": 11796 }, { "epoch": 0.4202932112510466, "grad_norm": 1.0681799358029813, "learning_rate": 1.3019848935087367e-05, "loss": 0.4498, "step": 11797 }, { "epoch": 0.42032883837754065, "grad_norm": 2.2151390098383366, "learning_rate": 1.3018748852465089e-05, "loss": 0.547, "step": 11798 }, { "epoch": 0.4203644655040348, "grad_norm": 1.4418237450010247, "learning_rate": 1.3017648729646061e-05, "loss": 0.5859, "step": 11799 }, { "epoch": 0.4204000926305289, "grad_norm": 1.419470968037393, "learning_rate": 1.3016548566644931e-05, "loss": 0.5175, "step": 11800 }, { "epoch": 0.420435719757023, "grad_norm": 1.25435251810787, "learning_rate": 1.3015448363476346e-05, "loss": 0.5125, "step": 11801 }, { "epoch": 0.4204713468835171, "grad_norm": 1.1260749986020708, "learning_rate": 1.3014348120154959e-05, "loss": 0.4341, "step": 11802 }, { "epoch": 0.42050697401001125, "grad_norm": 1.0740273313969786, "learning_rate": 1.301324783669542e-05, "loss": 0.4562, "step": 11803 }, { "epoch": 0.4205426011365053, "grad_norm": 1.6978218770942997, "learning_rate": 1.301214751311238e-05, "loss": 0.6257, "step": 11804 }, { "epoch": 0.42057822826299945, "grad_norm": 1.9114360062544826, "learning_rate": 1.3011047149420491e-05, "loss": 0.566, "step": 11805 }, { "epoch": 0.4206138553894936, "grad_norm": 1.1819621894572148, "learning_rate": 1.3009946745634406e-05, "loss": 0.5016, "step": 11806 }, { "epoch": 0.42064948251598766, "grad_norm": 1.3272896663762925, "learning_rate": 1.3008846301768773e-05, "loss": 0.4646, "step": 11807 }, { "epoch": 0.4206851096424818, "grad_norm": 1.4816290111632118, "learning_rate": 1.3007745817838251e-05, "loss": 0.5676, "step": 11808 }, { "epoch": 0.4207207367689759, "grad_norm": 1.1661947335141118, "learning_rate": 1.300664529385749e-05, "loss": 0.447, "step": 11809 }, { "epoch": 0.42075636389547, "grad_norm": 1.1597293041183305, "learning_rate": 1.3005544729841146e-05, "loss": 0.4857, "step": 11810 }, { "epoch": 0.42079199102196413, "grad_norm": 1.0990686936346208, "learning_rate": 1.3004444125803875e-05, "loss": 0.4492, "step": 11811 }, { "epoch": 0.42082761814845826, "grad_norm": 1.0116755273351534, "learning_rate": 1.3003343481760327e-05, "loss": 0.4287, "step": 11812 }, { "epoch": 0.42086324527495234, "grad_norm": 1.906517158733342, "learning_rate": 1.300224279772516e-05, "loss": 0.4846, "step": 11813 }, { "epoch": 0.42089887240144647, "grad_norm": 1.123553663209797, "learning_rate": 1.3001142073713035e-05, "loss": 0.4487, "step": 11814 }, { "epoch": 0.4209344995279406, "grad_norm": 2.267059224357688, "learning_rate": 1.3000041309738607e-05, "loss": 0.6108, "step": 11815 }, { "epoch": 0.4209701266544347, "grad_norm": 2.1181520828916707, "learning_rate": 1.2998940505816529e-05, "loss": 0.5769, "step": 11816 }, { "epoch": 0.4210057537809288, "grad_norm": 1.1082726896861745, "learning_rate": 1.2997839661961464e-05, "loss": 0.4558, "step": 11817 }, { "epoch": 0.42104138090742294, "grad_norm": 1.476840691915702, "learning_rate": 1.2996738778188067e-05, "loss": 0.5754, "step": 11818 }, { "epoch": 0.421077008033917, "grad_norm": 1.6937261189342532, "learning_rate": 1.2995637854511e-05, "loss": 0.5929, "step": 11819 }, { "epoch": 0.42111263516041114, "grad_norm": 1.149257665509339, "learning_rate": 1.299453689094492e-05, "loss": 0.45, "step": 11820 }, { "epoch": 0.4211482622869053, "grad_norm": 1.2075300014507226, "learning_rate": 1.2993435887504489e-05, "loss": 0.5013, "step": 11821 }, { "epoch": 0.42118388941339935, "grad_norm": 1.1303138695419788, "learning_rate": 1.2992334844204368e-05, "loss": 0.4974, "step": 11822 }, { "epoch": 0.4212195165398935, "grad_norm": 1.2107285197516875, "learning_rate": 1.2991233761059214e-05, "loss": 0.5259, "step": 11823 }, { "epoch": 0.4212551436663876, "grad_norm": 1.2079348986853715, "learning_rate": 1.2990132638083694e-05, "loss": 0.5075, "step": 11824 }, { "epoch": 0.4212907707928817, "grad_norm": 1.7477609824281064, "learning_rate": 1.2989031475292464e-05, "loss": 0.5828, "step": 11825 }, { "epoch": 0.4213263979193758, "grad_norm": 1.03108744704823, "learning_rate": 1.2987930272700196e-05, "loss": 0.4655, "step": 11826 }, { "epoch": 0.42136202504586995, "grad_norm": 1.5485288167215343, "learning_rate": 1.2986829030321543e-05, "loss": 0.5631, "step": 11827 }, { "epoch": 0.421397652172364, "grad_norm": 1.6836324078961404, "learning_rate": 1.2985727748171177e-05, "loss": 0.557, "step": 11828 }, { "epoch": 0.42143327929885815, "grad_norm": 1.4966234843365287, "learning_rate": 1.298462642626376e-05, "loss": 0.5529, "step": 11829 }, { "epoch": 0.4214689064253523, "grad_norm": 1.382648907461042, "learning_rate": 1.2983525064613955e-05, "loss": 0.5507, "step": 11830 }, { "epoch": 0.42150453355184636, "grad_norm": 1.0609449540234048, "learning_rate": 1.2982423663236428e-05, "loss": 0.4315, "step": 11831 }, { "epoch": 0.4215401606783405, "grad_norm": 1.2531024088367748, "learning_rate": 1.2981322222145846e-05, "loss": 0.5198, "step": 11832 }, { "epoch": 0.4215757878048346, "grad_norm": 1.3262461636403258, "learning_rate": 1.2980220741356874e-05, "loss": 0.572, "step": 11833 }, { "epoch": 0.4216114149313287, "grad_norm": 1.0391462900339727, "learning_rate": 1.297911922088418e-05, "loss": 0.4641, "step": 11834 }, { "epoch": 0.42164704205782283, "grad_norm": 1.1761741604914688, "learning_rate": 1.2978017660742433e-05, "loss": 0.4621, "step": 11835 }, { "epoch": 0.42168266918431696, "grad_norm": 1.4214731995828773, "learning_rate": 1.2976916060946295e-05, "loss": 0.4359, "step": 11836 }, { "epoch": 0.42171829631081104, "grad_norm": 2.0963240086124926, "learning_rate": 1.2975814421510444e-05, "loss": 0.5413, "step": 11837 }, { "epoch": 0.42175392343730517, "grad_norm": 1.1367913405601264, "learning_rate": 1.2974712742449543e-05, "loss": 0.4544, "step": 11838 }, { "epoch": 0.4217895505637993, "grad_norm": 1.8539303415802282, "learning_rate": 1.2973611023778259e-05, "loss": 0.5686, "step": 11839 }, { "epoch": 0.4218251776902934, "grad_norm": 1.2334030273087588, "learning_rate": 1.2972509265511269e-05, "loss": 0.4544, "step": 11840 }, { "epoch": 0.4218608048167875, "grad_norm": 1.152897111763137, "learning_rate": 1.297140746766324e-05, "loss": 0.5174, "step": 11841 }, { "epoch": 0.42189643194328164, "grad_norm": 1.2358844648604956, "learning_rate": 1.2970305630248841e-05, "loss": 0.4795, "step": 11842 }, { "epoch": 0.4219320590697757, "grad_norm": 2.847251468201768, "learning_rate": 1.296920375328275e-05, "loss": 0.6144, "step": 11843 }, { "epoch": 0.42196768619626984, "grad_norm": 1.6377050235601867, "learning_rate": 1.2968101836779634e-05, "loss": 0.6199, "step": 11844 }, { "epoch": 0.422003313322764, "grad_norm": 1.6112776137089095, "learning_rate": 1.296699988075417e-05, "loss": 0.6897, "step": 11845 }, { "epoch": 0.42203894044925805, "grad_norm": 1.2611602819754368, "learning_rate": 1.2965897885221024e-05, "loss": 0.4035, "step": 11846 }, { "epoch": 0.4220745675757522, "grad_norm": 1.2300766859821153, "learning_rate": 1.2964795850194878e-05, "loss": 0.4977, "step": 11847 }, { "epoch": 0.4221101947022463, "grad_norm": 1.059138350700767, "learning_rate": 1.2963693775690403e-05, "loss": 0.4748, "step": 11848 }, { "epoch": 0.4221458218287404, "grad_norm": 1.5555712084368967, "learning_rate": 1.2962591661722273e-05, "loss": 0.5731, "step": 11849 }, { "epoch": 0.4221814489552345, "grad_norm": 1.0806886665617927, "learning_rate": 1.2961489508305162e-05, "loss": 0.4579, "step": 11850 }, { "epoch": 0.42221707608172865, "grad_norm": 1.4657843973323843, "learning_rate": 1.2960387315453753e-05, "loss": 0.6303, "step": 11851 }, { "epoch": 0.4222527032082227, "grad_norm": 1.4113241400614314, "learning_rate": 1.2959285083182715e-05, "loss": 0.5627, "step": 11852 }, { "epoch": 0.42228833033471685, "grad_norm": 1.4460713386974924, "learning_rate": 1.2958182811506728e-05, "loss": 0.465, "step": 11853 }, { "epoch": 0.422323957461211, "grad_norm": 1.7582046400050915, "learning_rate": 1.2957080500440469e-05, "loss": 0.6195, "step": 11854 }, { "epoch": 0.42235958458770506, "grad_norm": 1.7994299886923766, "learning_rate": 1.2955978149998616e-05, "loss": 0.5346, "step": 11855 }, { "epoch": 0.4223952117141992, "grad_norm": 1.2206663759985317, "learning_rate": 1.2954875760195851e-05, "loss": 0.4684, "step": 11856 }, { "epoch": 0.4224308388406933, "grad_norm": 1.1997609227678603, "learning_rate": 1.2953773331046845e-05, "loss": 0.5066, "step": 11857 }, { "epoch": 0.4224664659671874, "grad_norm": 1.2975663296970494, "learning_rate": 1.2952670862566285e-05, "loss": 0.5002, "step": 11858 }, { "epoch": 0.42250209309368153, "grad_norm": 1.9191985017894249, "learning_rate": 1.2951568354768846e-05, "loss": 0.5684, "step": 11859 }, { "epoch": 0.42253772022017566, "grad_norm": 1.2171890413653887, "learning_rate": 1.2950465807669215e-05, "loss": 0.5042, "step": 11860 }, { "epoch": 0.42257334734666974, "grad_norm": 1.1983441839917943, "learning_rate": 1.2949363221282067e-05, "loss": 0.5072, "step": 11861 }, { "epoch": 0.42260897447316387, "grad_norm": 1.5330144542757564, "learning_rate": 1.2948260595622088e-05, "loss": 0.4965, "step": 11862 }, { "epoch": 0.422644601599658, "grad_norm": 1.2335049065253505, "learning_rate": 1.2947157930703959e-05, "loss": 0.5075, "step": 11863 }, { "epoch": 0.4226802287261521, "grad_norm": 1.3810255146410706, "learning_rate": 1.2946055226542362e-05, "loss": 0.4822, "step": 11864 }, { "epoch": 0.4227158558526462, "grad_norm": 1.098076609798287, "learning_rate": 1.2944952483151978e-05, "loss": 0.4604, "step": 11865 }, { "epoch": 0.42275148297914034, "grad_norm": 1.0738635684135904, "learning_rate": 1.2943849700547495e-05, "loss": 0.4364, "step": 11866 }, { "epoch": 0.4227871101056344, "grad_norm": 1.3404804952608578, "learning_rate": 1.2942746878743597e-05, "loss": 0.5304, "step": 11867 }, { "epoch": 0.42282273723212854, "grad_norm": 1.395117970131665, "learning_rate": 1.2941644017754964e-05, "loss": 0.5911, "step": 11868 }, { "epoch": 0.4228583643586227, "grad_norm": 1.1141338316424518, "learning_rate": 1.2940541117596287e-05, "loss": 0.4718, "step": 11869 }, { "epoch": 0.42289399148511675, "grad_norm": 1.5084645255722893, "learning_rate": 1.2939438178282254e-05, "loss": 0.5595, "step": 11870 }, { "epoch": 0.4229296186116109, "grad_norm": 1.1662585779174255, "learning_rate": 1.2938335199827539e-05, "loss": 0.4981, "step": 11871 }, { "epoch": 0.422965245738105, "grad_norm": 2.382294787541722, "learning_rate": 1.293723218224684e-05, "loss": 0.5088, "step": 11872 }, { "epoch": 0.4230008728645991, "grad_norm": 1.632654368126423, "learning_rate": 1.2936129125554842e-05, "loss": 0.5515, "step": 11873 }, { "epoch": 0.4230364999910932, "grad_norm": 1.9084725507878535, "learning_rate": 1.2935026029766234e-05, "loss": 0.6397, "step": 11874 }, { "epoch": 0.42307212711758735, "grad_norm": 2.3748567391486364, "learning_rate": 1.2933922894895703e-05, "loss": 0.6028, "step": 11875 }, { "epoch": 0.4231077542440814, "grad_norm": 1.6149060173476568, "learning_rate": 1.2932819720957934e-05, "loss": 0.5383, "step": 11876 }, { "epoch": 0.42314338137057556, "grad_norm": 1.421697791987351, "learning_rate": 1.2931716507967624e-05, "loss": 0.525, "step": 11877 }, { "epoch": 0.4231790084970697, "grad_norm": 1.4767976260261921, "learning_rate": 1.2930613255939458e-05, "loss": 0.466, "step": 11878 }, { "epoch": 0.42321463562356376, "grad_norm": 1.4480564660092334, "learning_rate": 1.2929509964888129e-05, "loss": 0.5087, "step": 11879 }, { "epoch": 0.4232502627500579, "grad_norm": 1.3233167976040279, "learning_rate": 1.2928406634828327e-05, "loss": 0.5536, "step": 11880 }, { "epoch": 0.423285889876552, "grad_norm": 1.544772179544989, "learning_rate": 1.2927303265774745e-05, "loss": 0.5957, "step": 11881 }, { "epoch": 0.4233215170030461, "grad_norm": 1.8008866616317003, "learning_rate": 1.292619985774207e-05, "loss": 0.5115, "step": 11882 }, { "epoch": 0.42335714412954023, "grad_norm": 1.2372888835691744, "learning_rate": 1.2925096410745002e-05, "loss": 0.4538, "step": 11883 }, { "epoch": 0.42339277125603436, "grad_norm": 1.1233098539308066, "learning_rate": 1.2923992924798229e-05, "loss": 0.409, "step": 11884 }, { "epoch": 0.42342839838252844, "grad_norm": 1.4894135400398327, "learning_rate": 1.292288939991645e-05, "loss": 0.5304, "step": 11885 }, { "epoch": 0.42346402550902257, "grad_norm": 1.650390858990025, "learning_rate": 1.2921785836114353e-05, "loss": 0.5306, "step": 11886 }, { "epoch": 0.4234996526355167, "grad_norm": 1.2232816155076163, "learning_rate": 1.2920682233406634e-05, "loss": 0.4781, "step": 11887 }, { "epoch": 0.4235352797620108, "grad_norm": 1.2376907923064961, "learning_rate": 1.2919578591807992e-05, "loss": 0.4756, "step": 11888 }, { "epoch": 0.4235709068885049, "grad_norm": 1.4940481597874207, "learning_rate": 1.2918474911333121e-05, "loss": 0.5526, "step": 11889 }, { "epoch": 0.42360653401499904, "grad_norm": 1.2117540957444535, "learning_rate": 1.2917371191996714e-05, "loss": 0.4914, "step": 11890 }, { "epoch": 0.4236421611414931, "grad_norm": 1.5861501885677736, "learning_rate": 1.2916267433813472e-05, "loss": 0.5377, "step": 11891 }, { "epoch": 0.42367778826798724, "grad_norm": 2.022416368732409, "learning_rate": 1.291516363679809e-05, "loss": 0.5475, "step": 11892 }, { "epoch": 0.4237134153944814, "grad_norm": 1.723342211414068, "learning_rate": 1.2914059800965268e-05, "loss": 0.4782, "step": 11893 }, { "epoch": 0.42374904252097545, "grad_norm": 1.5413993517815876, "learning_rate": 1.2912955926329702e-05, "loss": 0.523, "step": 11894 }, { "epoch": 0.4237846696474696, "grad_norm": 1.5921084513396822, "learning_rate": 1.291185201290609e-05, "loss": 0.5952, "step": 11895 }, { "epoch": 0.4238202967739637, "grad_norm": 1.0932482746450758, "learning_rate": 1.2910748060709138e-05, "loss": 0.4601, "step": 11896 }, { "epoch": 0.4238559239004578, "grad_norm": 1.0780163066436543, "learning_rate": 1.2909644069753537e-05, "loss": 0.4586, "step": 11897 }, { "epoch": 0.4238915510269519, "grad_norm": 1.1456369987723476, "learning_rate": 1.2908540040053992e-05, "loss": 0.4691, "step": 11898 }, { "epoch": 0.42392717815344605, "grad_norm": 1.0302601984390343, "learning_rate": 1.2907435971625204e-05, "loss": 0.4075, "step": 11899 }, { "epoch": 0.4239628052799401, "grad_norm": 2.8277242551015167, "learning_rate": 1.2906331864481874e-05, "loss": 0.5226, "step": 11900 }, { "epoch": 0.42399843240643426, "grad_norm": 1.9587169056220246, "learning_rate": 1.2905227718638704e-05, "loss": 0.6101, "step": 11901 }, { "epoch": 0.4240340595329284, "grad_norm": 1.5225947019095694, "learning_rate": 1.2904123534110396e-05, "loss": 0.5716, "step": 11902 }, { "epoch": 0.42406968665942246, "grad_norm": 1.5424826434153236, "learning_rate": 1.2903019310911653e-05, "loss": 0.5568, "step": 11903 }, { "epoch": 0.4241053137859166, "grad_norm": 1.095229501657954, "learning_rate": 1.290191504905718e-05, "loss": 0.4066, "step": 11904 }, { "epoch": 0.4241409409124107, "grad_norm": 1.1198001294935367, "learning_rate": 1.2900810748561678e-05, "loss": 0.4488, "step": 11905 }, { "epoch": 0.4241765680389048, "grad_norm": 1.8090381404948592, "learning_rate": 1.2899706409439854e-05, "loss": 0.5323, "step": 11906 }, { "epoch": 0.42421219516539893, "grad_norm": 1.0520110211464913, "learning_rate": 1.2898602031706413e-05, "loss": 0.481, "step": 11907 }, { "epoch": 0.42424782229189306, "grad_norm": 1.17873678813512, "learning_rate": 1.289749761537606e-05, "loss": 0.4811, "step": 11908 }, { "epoch": 0.42428344941838714, "grad_norm": 1.5025650034337734, "learning_rate": 1.2896393160463499e-05, "loss": 0.6021, "step": 11909 }, { "epoch": 0.42431907654488127, "grad_norm": 1.2059905677173706, "learning_rate": 1.289528866698344e-05, "loss": 0.4696, "step": 11910 }, { "epoch": 0.4243547036713754, "grad_norm": 1.305514916445526, "learning_rate": 1.289418413495059e-05, "loss": 0.5098, "step": 11911 }, { "epoch": 0.4243903307978695, "grad_norm": 2.3638262653411024, "learning_rate": 1.2893079564379657e-05, "loss": 0.6179, "step": 11912 }, { "epoch": 0.4244259579243636, "grad_norm": 1.1169247863884242, "learning_rate": 1.289197495528534e-05, "loss": 0.4824, "step": 11913 }, { "epoch": 0.42446158505085774, "grad_norm": 1.730916720826022, "learning_rate": 1.2890870307682362e-05, "loss": 0.5885, "step": 11914 }, { "epoch": 0.4244972121773518, "grad_norm": 1.071494669133226, "learning_rate": 1.2889765621585424e-05, "loss": 0.437, "step": 11915 }, { "epoch": 0.42453283930384594, "grad_norm": 1.0082801046844498, "learning_rate": 1.2888660897009233e-05, "loss": 0.4888, "step": 11916 }, { "epoch": 0.4245684664303401, "grad_norm": 1.6927080206199818, "learning_rate": 1.2887556133968506e-05, "loss": 0.6836, "step": 11917 }, { "epoch": 0.42460409355683415, "grad_norm": 1.7343069171501515, "learning_rate": 1.2886451332477952e-05, "loss": 0.5748, "step": 11918 }, { "epoch": 0.4246397206833283, "grad_norm": 1.4762456882692518, "learning_rate": 1.288534649255228e-05, "loss": 0.5599, "step": 11919 }, { "epoch": 0.4246753478098224, "grad_norm": 1.2696563333253732, "learning_rate": 1.2884241614206204e-05, "loss": 0.4922, "step": 11920 }, { "epoch": 0.4247109749363165, "grad_norm": 1.1854609275555932, "learning_rate": 1.2883136697454432e-05, "loss": 0.4752, "step": 11921 }, { "epoch": 0.4247466020628106, "grad_norm": 1.2319513996831948, "learning_rate": 1.288203174231168e-05, "loss": 0.4761, "step": 11922 }, { "epoch": 0.42478222918930475, "grad_norm": 1.0695100968444609, "learning_rate": 1.2880926748792663e-05, "loss": 0.457, "step": 11923 }, { "epoch": 0.4248178563157988, "grad_norm": 1.47214581854814, "learning_rate": 1.287982171691209e-05, "loss": 0.6398, "step": 11924 }, { "epoch": 0.42485348344229296, "grad_norm": 1.6402858517509553, "learning_rate": 1.2878716646684681e-05, "loss": 0.5532, "step": 11925 }, { "epoch": 0.4248891105687871, "grad_norm": 1.0198253666313517, "learning_rate": 1.2877611538125148e-05, "loss": 0.4858, "step": 11926 }, { "epoch": 0.42492473769528116, "grad_norm": 2.0808331950938825, "learning_rate": 1.2876506391248206e-05, "loss": 0.6241, "step": 11927 }, { "epoch": 0.4249603648217753, "grad_norm": 1.5097901504937166, "learning_rate": 1.2875401206068567e-05, "loss": 0.5623, "step": 11928 }, { "epoch": 0.4249959919482694, "grad_norm": 1.0887246952376446, "learning_rate": 1.2874295982600953e-05, "loss": 0.4311, "step": 11929 }, { "epoch": 0.4250316190747635, "grad_norm": 1.2593319972464767, "learning_rate": 1.287319072086008e-05, "loss": 0.4489, "step": 11930 }, { "epoch": 0.42506724620125763, "grad_norm": 1.9124575584507073, "learning_rate": 1.2872085420860665e-05, "loss": 0.5538, "step": 11931 }, { "epoch": 0.42510287332775176, "grad_norm": 1.2577746067125835, "learning_rate": 1.2870980082617423e-05, "loss": 0.4486, "step": 11932 }, { "epoch": 0.42513850045424584, "grad_norm": 1.4875352249701912, "learning_rate": 1.2869874706145075e-05, "loss": 0.5491, "step": 11933 }, { "epoch": 0.42517412758073997, "grad_norm": 1.0325714621130169, "learning_rate": 1.2868769291458341e-05, "loss": 0.4644, "step": 11934 }, { "epoch": 0.4252097547072341, "grad_norm": 1.1281013598554632, "learning_rate": 1.2867663838571938e-05, "loss": 0.494, "step": 11935 }, { "epoch": 0.4252453818337282, "grad_norm": 1.1099659259189343, "learning_rate": 1.2866558347500586e-05, "loss": 0.4859, "step": 11936 }, { "epoch": 0.4252810089602223, "grad_norm": 1.3822051339432593, "learning_rate": 1.2865452818259007e-05, "loss": 0.4986, "step": 11937 }, { "epoch": 0.42531663608671644, "grad_norm": 1.0332327143098772, "learning_rate": 1.286434725086192e-05, "loss": 0.4157, "step": 11938 }, { "epoch": 0.4253522632132105, "grad_norm": 2.21509078659096, "learning_rate": 1.2863241645324048e-05, "loss": 0.546, "step": 11939 }, { "epoch": 0.42538789033970464, "grad_norm": 1.7369634184956617, "learning_rate": 1.2862136001660109e-05, "loss": 0.6133, "step": 11940 }, { "epoch": 0.4254235174661988, "grad_norm": 1.085898861675343, "learning_rate": 1.2861030319884832e-05, "loss": 0.4387, "step": 11941 }, { "epoch": 0.42545914459269285, "grad_norm": 1.6739733342466938, "learning_rate": 1.2859924600012937e-05, "loss": 0.6006, "step": 11942 }, { "epoch": 0.425494771719187, "grad_norm": 1.0575934679831949, "learning_rate": 1.2858818842059145e-05, "loss": 0.4047, "step": 11943 }, { "epoch": 0.4255303988456811, "grad_norm": 1.18460960657286, "learning_rate": 1.2857713046038183e-05, "loss": 0.4613, "step": 11944 }, { "epoch": 0.4255660259721752, "grad_norm": 2.1361049853593896, "learning_rate": 1.2856607211964773e-05, "loss": 0.549, "step": 11945 }, { "epoch": 0.4256016530986693, "grad_norm": 1.178456111792034, "learning_rate": 1.2855501339853644e-05, "loss": 0.4919, "step": 11946 }, { "epoch": 0.42563728022516345, "grad_norm": 1.6958847617346675, "learning_rate": 1.2854395429719514e-05, "loss": 0.4629, "step": 11947 }, { "epoch": 0.4256729073516575, "grad_norm": 1.1262882155539087, "learning_rate": 1.2853289481577119e-05, "loss": 0.5015, "step": 11948 }, { "epoch": 0.42570853447815166, "grad_norm": 1.1141179377416026, "learning_rate": 1.2852183495441179e-05, "loss": 0.4465, "step": 11949 }, { "epoch": 0.4257441616046458, "grad_norm": 1.0844560789361153, "learning_rate": 1.285107747132642e-05, "loss": 0.4564, "step": 11950 }, { "epoch": 0.42577978873113986, "grad_norm": 1.172223171125634, "learning_rate": 1.2849971409247575e-05, "loss": 0.4716, "step": 11951 }, { "epoch": 0.425815415857634, "grad_norm": 1.708681778072305, "learning_rate": 1.2848865309219365e-05, "loss": 0.6167, "step": 11952 }, { "epoch": 0.4258510429841281, "grad_norm": 1.6161970090138522, "learning_rate": 1.2847759171256523e-05, "loss": 0.5492, "step": 11953 }, { "epoch": 0.4258866701106222, "grad_norm": 1.1495113693560275, "learning_rate": 1.2846652995373777e-05, "loss": 0.4822, "step": 11954 }, { "epoch": 0.42592229723711633, "grad_norm": 1.0510087462006392, "learning_rate": 1.2845546781585857e-05, "loss": 0.4595, "step": 11955 }, { "epoch": 0.42595792436361046, "grad_norm": 1.0970870147340661, "learning_rate": 1.2844440529907496e-05, "loss": 0.5107, "step": 11956 }, { "epoch": 0.42599355149010454, "grad_norm": 1.1207138525370315, "learning_rate": 1.284333424035342e-05, "loss": 0.488, "step": 11957 }, { "epoch": 0.42602917861659867, "grad_norm": 1.1128170970318247, "learning_rate": 1.284222791293836e-05, "loss": 0.4184, "step": 11958 }, { "epoch": 0.4260648057430928, "grad_norm": 1.564697851341432, "learning_rate": 1.2841121547677047e-05, "loss": 0.5511, "step": 11959 }, { "epoch": 0.4261004328695869, "grad_norm": 1.7679811285745852, "learning_rate": 1.2840015144584216e-05, "loss": 0.6556, "step": 11960 }, { "epoch": 0.426136059996081, "grad_norm": 1.1751411930578068, "learning_rate": 1.28389087036746e-05, "loss": 0.456, "step": 11961 }, { "epoch": 0.42617168712257514, "grad_norm": 1.983299904300903, "learning_rate": 1.2837802224962928e-05, "loss": 0.7041, "step": 11962 }, { "epoch": 0.42620731424906927, "grad_norm": 1.5715523570194367, "learning_rate": 1.2836695708463936e-05, "loss": 0.6088, "step": 11963 }, { "epoch": 0.42624294137556334, "grad_norm": 1.6052006543724016, "learning_rate": 1.2835589154192358e-05, "loss": 0.5365, "step": 11964 }, { "epoch": 0.4262785685020575, "grad_norm": 1.735813786448697, "learning_rate": 1.2834482562162929e-05, "loss": 0.5497, "step": 11965 }, { "epoch": 0.4263141956285516, "grad_norm": 1.674376138763192, "learning_rate": 1.2833375932390383e-05, "loss": 0.6184, "step": 11966 }, { "epoch": 0.4263498227550457, "grad_norm": 1.1534909900775758, "learning_rate": 1.2832269264889459e-05, "loss": 0.4678, "step": 11967 }, { "epoch": 0.4263854498815398, "grad_norm": 1.8758420174407513, "learning_rate": 1.2831162559674888e-05, "loss": 0.5573, "step": 11968 }, { "epoch": 0.42642107700803394, "grad_norm": 1.8112454851303346, "learning_rate": 1.2830055816761409e-05, "loss": 0.6412, "step": 11969 }, { "epoch": 0.426456704134528, "grad_norm": 1.1867265652662378, "learning_rate": 1.2828949036163757e-05, "loss": 0.5107, "step": 11970 }, { "epoch": 0.42649233126102215, "grad_norm": 1.590712865583454, "learning_rate": 1.2827842217896674e-05, "loss": 0.5678, "step": 11971 }, { "epoch": 0.4265279583875163, "grad_norm": 1.1029697751477618, "learning_rate": 1.2826735361974891e-05, "loss": 0.4755, "step": 11972 }, { "epoch": 0.42656358551401036, "grad_norm": 1.146641603632929, "learning_rate": 1.2825628468413156e-05, "loss": 0.4659, "step": 11973 }, { "epoch": 0.4265992126405045, "grad_norm": 1.6309464553664872, "learning_rate": 1.2824521537226203e-05, "loss": 0.573, "step": 11974 }, { "epoch": 0.4266348397669986, "grad_norm": 1.7152401016188346, "learning_rate": 1.2823414568428767e-05, "loss": 0.5909, "step": 11975 }, { "epoch": 0.4266704668934927, "grad_norm": 1.180719308506194, "learning_rate": 1.2822307562035593e-05, "loss": 0.4962, "step": 11976 }, { "epoch": 0.4267060940199868, "grad_norm": 1.013574661991159, "learning_rate": 1.2821200518061422e-05, "loss": 0.4153, "step": 11977 }, { "epoch": 0.42674172114648096, "grad_norm": 1.1383120027365217, "learning_rate": 1.2820093436520998e-05, "loss": 0.4616, "step": 11978 }, { "epoch": 0.42677734827297503, "grad_norm": 1.5915075992451846, "learning_rate": 1.2818986317429057e-05, "loss": 0.5128, "step": 11979 }, { "epoch": 0.42681297539946916, "grad_norm": 1.8849842823184533, "learning_rate": 1.2817879160800337e-05, "loss": 0.5395, "step": 11980 }, { "epoch": 0.4268486025259633, "grad_norm": 1.125966325069642, "learning_rate": 1.2816771966649593e-05, "loss": 0.4913, "step": 11981 }, { "epoch": 0.42688422965245737, "grad_norm": 1.2041920984305063, "learning_rate": 1.2815664734991559e-05, "loss": 0.5147, "step": 11982 }, { "epoch": 0.4269198567789515, "grad_norm": 3.0968857822712463, "learning_rate": 1.2814557465840984e-05, "loss": 0.5575, "step": 11983 }, { "epoch": 0.42695548390544563, "grad_norm": 1.7585806511973963, "learning_rate": 1.2813450159212604e-05, "loss": 0.5871, "step": 11984 }, { "epoch": 0.4269911110319397, "grad_norm": 1.4759763079664385, "learning_rate": 1.2812342815121173e-05, "loss": 0.5368, "step": 11985 }, { "epoch": 0.42702673815843384, "grad_norm": 2.094345860651478, "learning_rate": 1.2811235433581428e-05, "loss": 0.5678, "step": 11986 }, { "epoch": 0.42706236528492797, "grad_norm": 1.072965694628008, "learning_rate": 1.2810128014608121e-05, "loss": 0.4962, "step": 11987 }, { "epoch": 0.42709799241142204, "grad_norm": 1.1423534566734692, "learning_rate": 1.2809020558215992e-05, "loss": 0.5032, "step": 11988 }, { "epoch": 0.4271336195379162, "grad_norm": 1.4373973260278337, "learning_rate": 1.2807913064419797e-05, "loss": 0.5454, "step": 11989 }, { "epoch": 0.4271692466644103, "grad_norm": 1.082192746074176, "learning_rate": 1.2806805533234272e-05, "loss": 0.4605, "step": 11990 }, { "epoch": 0.4272048737909044, "grad_norm": 1.5938536958610403, "learning_rate": 1.280569796467417e-05, "loss": 0.5055, "step": 11991 }, { "epoch": 0.4272405009173985, "grad_norm": 1.7707915958866085, "learning_rate": 1.280459035875424e-05, "loss": 0.6112, "step": 11992 }, { "epoch": 0.42727612804389264, "grad_norm": 1.8693818373007651, "learning_rate": 1.280348271548923e-05, "loss": 0.678, "step": 11993 }, { "epoch": 0.4273117551703867, "grad_norm": 1.1109444419464862, "learning_rate": 1.2802375034893886e-05, "loss": 0.4486, "step": 11994 }, { "epoch": 0.42734738229688085, "grad_norm": 1.205215707069767, "learning_rate": 1.280126731698296e-05, "loss": 0.464, "step": 11995 }, { "epoch": 0.427383009423375, "grad_norm": 1.2154226613849872, "learning_rate": 1.2800159561771201e-05, "loss": 0.4734, "step": 11996 }, { "epoch": 0.42741863654986906, "grad_norm": 1.0993315517109492, "learning_rate": 1.2799051769273363e-05, "loss": 0.4485, "step": 11997 }, { "epoch": 0.4274542636763632, "grad_norm": 1.1680873488814618, "learning_rate": 1.2797943939504192e-05, "loss": 0.502, "step": 11998 }, { "epoch": 0.4274898908028573, "grad_norm": 1.0568594362996244, "learning_rate": 1.279683607247844e-05, "loss": 0.4642, "step": 11999 }, { "epoch": 0.4275255179293514, "grad_norm": 1.1451483006115895, "learning_rate": 1.2795728168210864e-05, "loss": 0.4948, "step": 12000 }, { "epoch": 0.4275611450558455, "grad_norm": 3.357285700044408, "learning_rate": 1.2794620226716213e-05, "loss": 0.6496, "step": 12001 }, { "epoch": 0.42759677218233966, "grad_norm": 1.060675813160372, "learning_rate": 1.2793512248009241e-05, "loss": 0.459, "step": 12002 }, { "epoch": 0.42763239930883373, "grad_norm": 1.3994319595719331, "learning_rate": 1.2792404232104699e-05, "loss": 0.5465, "step": 12003 }, { "epoch": 0.42766802643532786, "grad_norm": 1.2357755432892037, "learning_rate": 1.2791296179017343e-05, "loss": 0.4996, "step": 12004 }, { "epoch": 0.427703653561822, "grad_norm": 1.197144498687819, "learning_rate": 1.279018808876193e-05, "loss": 0.4595, "step": 12005 }, { "epoch": 0.42773928068831607, "grad_norm": 1.5245620531346955, "learning_rate": 1.2789079961353208e-05, "loss": 0.5668, "step": 12006 }, { "epoch": 0.4277749078148102, "grad_norm": 1.5025284858521228, "learning_rate": 1.278797179680594e-05, "loss": 0.5884, "step": 12007 }, { "epoch": 0.42781053494130433, "grad_norm": 1.4504968254245327, "learning_rate": 1.278686359513488e-05, "loss": 0.547, "step": 12008 }, { "epoch": 0.4278461620677984, "grad_norm": 0.9932453866421388, "learning_rate": 1.2785755356354778e-05, "loss": 0.4445, "step": 12009 }, { "epoch": 0.42788178919429254, "grad_norm": 1.0855585789017765, "learning_rate": 1.27846470804804e-05, "loss": 0.4789, "step": 12010 }, { "epoch": 0.42791741632078667, "grad_norm": 1.5950227781006767, "learning_rate": 1.2783538767526497e-05, "loss": 0.6654, "step": 12011 }, { "epoch": 0.42795304344728075, "grad_norm": 1.5331803924571736, "learning_rate": 1.2782430417507836e-05, "loss": 0.5672, "step": 12012 }, { "epoch": 0.4279886705737749, "grad_norm": 1.5740805625258358, "learning_rate": 1.2781322030439165e-05, "loss": 0.5151, "step": 12013 }, { "epoch": 0.428024297700269, "grad_norm": 1.1683058612555448, "learning_rate": 1.2780213606335246e-05, "loss": 0.5253, "step": 12014 }, { "epoch": 0.4280599248267631, "grad_norm": 1.1129843768977066, "learning_rate": 1.2779105145210842e-05, "loss": 0.5133, "step": 12015 }, { "epoch": 0.4280955519532572, "grad_norm": 1.5488800500909652, "learning_rate": 1.2777996647080711e-05, "loss": 0.5621, "step": 12016 }, { "epoch": 0.42813117907975135, "grad_norm": 1.3743835960273687, "learning_rate": 1.2776888111959609e-05, "loss": 0.501, "step": 12017 }, { "epoch": 0.4281668062062454, "grad_norm": 1.477465067243024, "learning_rate": 1.2775779539862305e-05, "loss": 0.5131, "step": 12018 }, { "epoch": 0.42820243333273955, "grad_norm": 1.2552766809399145, "learning_rate": 1.2774670930803555e-05, "loss": 0.531, "step": 12019 }, { "epoch": 0.4282380604592337, "grad_norm": 1.0157548266771974, "learning_rate": 1.2773562284798122e-05, "loss": 0.3935, "step": 12020 }, { "epoch": 0.42827368758572776, "grad_norm": 1.0138200893222373, "learning_rate": 1.2772453601860768e-05, "loss": 0.47, "step": 12021 }, { "epoch": 0.4283093147122219, "grad_norm": 1.2477379968129585, "learning_rate": 1.2771344882006256e-05, "loss": 0.4972, "step": 12022 }, { "epoch": 0.428344941838716, "grad_norm": 1.3068199087535683, "learning_rate": 1.2770236125249351e-05, "loss": 0.5023, "step": 12023 }, { "epoch": 0.4283805689652101, "grad_norm": 1.5819530497454353, "learning_rate": 1.2769127331604817e-05, "loss": 0.6533, "step": 12024 }, { "epoch": 0.4284161960917042, "grad_norm": 1.2254104679719298, "learning_rate": 1.2768018501087412e-05, "loss": 0.46, "step": 12025 }, { "epoch": 0.42845182321819836, "grad_norm": 1.2221017334046007, "learning_rate": 1.276690963371191e-05, "loss": 0.4928, "step": 12026 }, { "epoch": 0.42848745034469243, "grad_norm": 1.1007765529966882, "learning_rate": 1.2765800729493073e-05, "loss": 0.4731, "step": 12027 }, { "epoch": 0.42852307747118656, "grad_norm": 1.057850149560744, "learning_rate": 1.2764691788445664e-05, "loss": 0.4805, "step": 12028 }, { "epoch": 0.4285587045976807, "grad_norm": 1.5227531443199231, "learning_rate": 1.2763582810584452e-05, "loss": 0.4672, "step": 12029 }, { "epoch": 0.42859433172417477, "grad_norm": 1.038970811472338, "learning_rate": 1.2762473795924204e-05, "loss": 0.4762, "step": 12030 }, { "epoch": 0.4286299588506689, "grad_norm": 1.1967971226116603, "learning_rate": 1.2761364744479688e-05, "loss": 0.4836, "step": 12031 }, { "epoch": 0.42866558597716303, "grad_norm": 1.2298731942784782, "learning_rate": 1.2760255656265665e-05, "loss": 0.4428, "step": 12032 }, { "epoch": 0.4287012131036571, "grad_norm": 1.5119566720757636, "learning_rate": 1.275914653129691e-05, "loss": 0.6404, "step": 12033 }, { "epoch": 0.42873684023015124, "grad_norm": 1.6743794437305781, "learning_rate": 1.2758037369588195e-05, "loss": 0.5892, "step": 12034 }, { "epoch": 0.42877246735664537, "grad_norm": 1.0541499194037973, "learning_rate": 1.2756928171154284e-05, "loss": 0.4247, "step": 12035 }, { "epoch": 0.42880809448313945, "grad_norm": 1.573604855261862, "learning_rate": 1.2755818936009943e-05, "loss": 0.5745, "step": 12036 }, { "epoch": 0.4288437216096336, "grad_norm": 1.0988328609227886, "learning_rate": 1.2754709664169948e-05, "loss": 0.4792, "step": 12037 }, { "epoch": 0.4288793487361277, "grad_norm": 1.7048369718239285, "learning_rate": 1.2753600355649071e-05, "loss": 0.5977, "step": 12038 }, { "epoch": 0.4289149758626218, "grad_norm": 2.182510327573021, "learning_rate": 1.275249101046208e-05, "loss": 0.5454, "step": 12039 }, { "epoch": 0.4289506029891159, "grad_norm": 1.0152171148496372, "learning_rate": 1.2751381628623746e-05, "loss": 0.4113, "step": 12040 }, { "epoch": 0.42898623011561005, "grad_norm": 1.1851526735257314, "learning_rate": 1.2750272210148844e-05, "loss": 0.4657, "step": 12041 }, { "epoch": 0.4290218572421041, "grad_norm": 1.4544823937317044, "learning_rate": 1.2749162755052145e-05, "loss": 0.632, "step": 12042 }, { "epoch": 0.42905748436859825, "grad_norm": 1.1473728305028668, "learning_rate": 1.274805326334842e-05, "loss": 0.4755, "step": 12043 }, { "epoch": 0.4290931114950924, "grad_norm": 1.2515340234988919, "learning_rate": 1.2746943735052445e-05, "loss": 0.5068, "step": 12044 }, { "epoch": 0.42912873862158646, "grad_norm": 1.0581858497847416, "learning_rate": 1.2745834170178996e-05, "loss": 0.4644, "step": 12045 }, { "epoch": 0.4291643657480806, "grad_norm": 1.5901039245312059, "learning_rate": 1.2744724568742848e-05, "loss": 0.5904, "step": 12046 }, { "epoch": 0.4291999928745747, "grad_norm": 1.190858755385417, "learning_rate": 1.274361493075877e-05, "loss": 0.4722, "step": 12047 }, { "epoch": 0.4292356200010688, "grad_norm": 1.18470281964778, "learning_rate": 1.2742505256241543e-05, "loss": 0.4641, "step": 12048 }, { "epoch": 0.4292712471275629, "grad_norm": 1.468088141345021, "learning_rate": 1.2741395545205944e-05, "loss": 0.6683, "step": 12049 }, { "epoch": 0.42930687425405706, "grad_norm": 1.555936440071498, "learning_rate": 1.2740285797666742e-05, "loss": 0.6057, "step": 12050 }, { "epoch": 0.42934250138055113, "grad_norm": 1.432270629807401, "learning_rate": 1.2739176013638724e-05, "loss": 0.5704, "step": 12051 }, { "epoch": 0.42937812850704526, "grad_norm": 1.2695803333914997, "learning_rate": 1.2738066193136661e-05, "loss": 0.5153, "step": 12052 }, { "epoch": 0.4294137556335394, "grad_norm": 1.120021744173815, "learning_rate": 1.2736956336175332e-05, "loss": 0.5109, "step": 12053 }, { "epoch": 0.42944938276003347, "grad_norm": 1.1496425534079535, "learning_rate": 1.2735846442769516e-05, "loss": 0.4727, "step": 12054 }, { "epoch": 0.4294850098865276, "grad_norm": 1.177831159861774, "learning_rate": 1.2734736512933992e-05, "loss": 0.4848, "step": 12055 }, { "epoch": 0.42952063701302173, "grad_norm": 1.1381861356027783, "learning_rate": 1.2733626546683542e-05, "loss": 0.4403, "step": 12056 }, { "epoch": 0.4295562641395158, "grad_norm": 1.208918979000799, "learning_rate": 1.2732516544032944e-05, "loss": 0.4598, "step": 12057 }, { "epoch": 0.42959189126600994, "grad_norm": 1.2386005998174303, "learning_rate": 1.2731406504996979e-05, "loss": 0.4873, "step": 12058 }, { "epoch": 0.42962751839250407, "grad_norm": 1.9432059810197189, "learning_rate": 1.2730296429590423e-05, "loss": 0.5479, "step": 12059 }, { "epoch": 0.42966314551899815, "grad_norm": 1.0337457389221252, "learning_rate": 1.2729186317828066e-05, "loss": 0.4525, "step": 12060 }, { "epoch": 0.4296987726454923, "grad_norm": 1.213195528723628, "learning_rate": 1.2728076169724686e-05, "loss": 0.4927, "step": 12061 }, { "epoch": 0.4297343997719864, "grad_norm": 1.6349613206334537, "learning_rate": 1.2726965985295061e-05, "loss": 0.6437, "step": 12062 }, { "epoch": 0.4297700268984805, "grad_norm": 1.6053637317070328, "learning_rate": 1.2725855764553981e-05, "loss": 0.631, "step": 12063 }, { "epoch": 0.4298056540249746, "grad_norm": 2.730347376880651, "learning_rate": 1.2724745507516225e-05, "loss": 0.5706, "step": 12064 }, { "epoch": 0.42984128115146875, "grad_norm": 1.0397982974921154, "learning_rate": 1.2723635214196579e-05, "loss": 0.4538, "step": 12065 }, { "epoch": 0.4298769082779628, "grad_norm": 1.255119884732653, "learning_rate": 1.2722524884609824e-05, "loss": 0.5182, "step": 12066 }, { "epoch": 0.42991253540445695, "grad_norm": 1.6122420213395892, "learning_rate": 1.2721414518770748e-05, "loss": 0.6018, "step": 12067 }, { "epoch": 0.4299481625309511, "grad_norm": 1.2786599852455558, "learning_rate": 1.2720304116694137e-05, "loss": 0.4649, "step": 12068 }, { "epoch": 0.42998378965744516, "grad_norm": 1.7186327138247266, "learning_rate": 1.2719193678394775e-05, "loss": 0.5806, "step": 12069 }, { "epoch": 0.4300194167839393, "grad_norm": 1.524519255538592, "learning_rate": 1.2718083203887445e-05, "loss": 0.6165, "step": 12070 }, { "epoch": 0.4300550439104334, "grad_norm": 1.7169484280838043, "learning_rate": 1.2716972693186941e-05, "loss": 0.4949, "step": 12071 }, { "epoch": 0.4300906710369275, "grad_norm": 1.498099976245011, "learning_rate": 1.2715862146308049e-05, "loss": 0.6402, "step": 12072 }, { "epoch": 0.4301262981634216, "grad_norm": 1.274885432851187, "learning_rate": 1.2714751563265551e-05, "loss": 0.4685, "step": 12073 }, { "epoch": 0.43016192528991576, "grad_norm": 1.144889820710068, "learning_rate": 1.271364094407424e-05, "loss": 0.4426, "step": 12074 }, { "epoch": 0.43019755241640983, "grad_norm": 2.1321000884403856, "learning_rate": 1.2712530288748901e-05, "loss": 0.6334, "step": 12075 }, { "epoch": 0.43023317954290397, "grad_norm": 1.0299236598908472, "learning_rate": 1.2711419597304326e-05, "loss": 0.42, "step": 12076 }, { "epoch": 0.4302688066693981, "grad_norm": 2.077771247706574, "learning_rate": 1.2710308869755303e-05, "loss": 0.5641, "step": 12077 }, { "epoch": 0.43030443379589217, "grad_norm": 1.1681027188098518, "learning_rate": 1.2709198106116624e-05, "loss": 0.5109, "step": 12078 }, { "epoch": 0.4303400609223863, "grad_norm": 1.1755534995979082, "learning_rate": 1.2708087306403078e-05, "loss": 0.4924, "step": 12079 }, { "epoch": 0.43037568804888043, "grad_norm": 1.5051581144339112, "learning_rate": 1.2706976470629457e-05, "loss": 0.6156, "step": 12080 }, { "epoch": 0.4304113151753745, "grad_norm": 1.0706698526563185, "learning_rate": 1.2705865598810551e-05, "loss": 0.4635, "step": 12081 }, { "epoch": 0.43044694230186864, "grad_norm": 2.7879209074094455, "learning_rate": 1.2704754690961157e-05, "loss": 0.6335, "step": 12082 }, { "epoch": 0.43048256942836277, "grad_norm": 1.7908405102041447, "learning_rate": 1.2703643747096064e-05, "loss": 0.5652, "step": 12083 }, { "epoch": 0.43051819655485685, "grad_norm": 1.0185009461159382, "learning_rate": 1.2702532767230062e-05, "loss": 0.4902, "step": 12084 }, { "epoch": 0.430553823681351, "grad_norm": 1.1178692678002806, "learning_rate": 1.2701421751377944e-05, "loss": 0.4339, "step": 12085 }, { "epoch": 0.4305894508078451, "grad_norm": 1.161550614325292, "learning_rate": 1.2700310699554513e-05, "loss": 0.4911, "step": 12086 }, { "epoch": 0.4306250779343392, "grad_norm": 1.5362791664488151, "learning_rate": 1.2699199611774558e-05, "loss": 0.6426, "step": 12087 }, { "epoch": 0.4306607050608333, "grad_norm": 1.6585360363416064, "learning_rate": 1.269808848805287e-05, "loss": 0.6305, "step": 12088 }, { "epoch": 0.43069633218732745, "grad_norm": 1.3900178316069545, "learning_rate": 1.2696977328404248e-05, "loss": 0.5838, "step": 12089 }, { "epoch": 0.4307319593138215, "grad_norm": 1.6960005493607786, "learning_rate": 1.2695866132843486e-05, "loss": 0.5847, "step": 12090 }, { "epoch": 0.43076758644031565, "grad_norm": 1.2545999957196725, "learning_rate": 1.2694754901385386e-05, "loss": 0.4961, "step": 12091 }, { "epoch": 0.4308032135668098, "grad_norm": 1.3445212058726068, "learning_rate": 1.2693643634044738e-05, "loss": 0.5376, "step": 12092 }, { "epoch": 0.43083884069330386, "grad_norm": 1.5451646005764483, "learning_rate": 1.2692532330836346e-05, "loss": 0.5794, "step": 12093 }, { "epoch": 0.430874467819798, "grad_norm": 1.4477972951371714, "learning_rate": 1.2691420991775003e-05, "loss": 0.5938, "step": 12094 }, { "epoch": 0.4309100949462921, "grad_norm": 2.320586854868022, "learning_rate": 1.2690309616875509e-05, "loss": 0.6114, "step": 12095 }, { "epoch": 0.4309457220727862, "grad_norm": 1.053997207677916, "learning_rate": 1.2689198206152657e-05, "loss": 0.453, "step": 12096 }, { "epoch": 0.43098134919928033, "grad_norm": 1.2310787505564198, "learning_rate": 1.2688086759621256e-05, "loss": 0.5139, "step": 12097 }, { "epoch": 0.43101697632577446, "grad_norm": 1.07172766556199, "learning_rate": 1.26869752772961e-05, "loss": 0.4239, "step": 12098 }, { "epoch": 0.43105260345226853, "grad_norm": 1.1484051243026392, "learning_rate": 1.268586375919199e-05, "loss": 0.4887, "step": 12099 }, { "epoch": 0.43108823057876267, "grad_norm": 1.4129165206512853, "learning_rate": 1.2684752205323727e-05, "loss": 0.4871, "step": 12100 }, { "epoch": 0.4311238577052568, "grad_norm": 1.14433365742792, "learning_rate": 1.2683640615706113e-05, "loss": 0.4996, "step": 12101 }, { "epoch": 0.43115948483175087, "grad_norm": 1.3172140378465849, "learning_rate": 1.2682528990353943e-05, "loss": 0.5013, "step": 12102 }, { "epoch": 0.431195111958245, "grad_norm": 1.0380320159669048, "learning_rate": 1.268141732928203e-05, "loss": 0.4535, "step": 12103 }, { "epoch": 0.43123073908473913, "grad_norm": 1.1961326272038333, "learning_rate": 1.2680305632505168e-05, "loss": 0.5153, "step": 12104 }, { "epoch": 0.4312663662112332, "grad_norm": 1.0485560531690932, "learning_rate": 1.2679193900038165e-05, "loss": 0.4589, "step": 12105 }, { "epoch": 0.43130199333772734, "grad_norm": 1.077685465474225, "learning_rate": 1.2678082131895824e-05, "loss": 0.4339, "step": 12106 }, { "epoch": 0.43133762046422147, "grad_norm": 1.366912534704186, "learning_rate": 1.2676970328092944e-05, "loss": 0.4789, "step": 12107 }, { "epoch": 0.43137324759071555, "grad_norm": 1.112730229049727, "learning_rate": 1.2675858488644334e-05, "loss": 0.4365, "step": 12108 }, { "epoch": 0.4314088747172097, "grad_norm": 1.7555147846402868, "learning_rate": 1.2674746613564799e-05, "loss": 0.5552, "step": 12109 }, { "epoch": 0.4314445018437038, "grad_norm": 1.140927302878681, "learning_rate": 1.2673634702869144e-05, "loss": 0.4611, "step": 12110 }, { "epoch": 0.4314801289701979, "grad_norm": 1.1752948500800304, "learning_rate": 1.2672522756572173e-05, "loss": 0.449, "step": 12111 }, { "epoch": 0.431515756096692, "grad_norm": 1.5629236440058842, "learning_rate": 1.2671410774688693e-05, "loss": 0.5391, "step": 12112 }, { "epoch": 0.43155138322318615, "grad_norm": 1.4134775910792334, "learning_rate": 1.267029875723351e-05, "loss": 0.6809, "step": 12113 }, { "epoch": 0.4315870103496802, "grad_norm": 1.2901212501126897, "learning_rate": 1.2669186704221436e-05, "loss": 0.5833, "step": 12114 }, { "epoch": 0.43162263747617435, "grad_norm": 1.4560531670932833, "learning_rate": 1.2668074615667272e-05, "loss": 0.5833, "step": 12115 }, { "epoch": 0.4316582646026685, "grad_norm": 1.5938554831025074, "learning_rate": 1.2666962491585832e-05, "loss": 0.5404, "step": 12116 }, { "epoch": 0.43169389172916256, "grad_norm": 1.3850213537048128, "learning_rate": 1.2665850331991923e-05, "loss": 0.5746, "step": 12117 }, { "epoch": 0.4317295188556567, "grad_norm": 1.4339905751391584, "learning_rate": 1.266473813690035e-05, "loss": 0.5904, "step": 12118 }, { "epoch": 0.4317651459821508, "grad_norm": 1.6550982925827684, "learning_rate": 1.266362590632593e-05, "loss": 0.5731, "step": 12119 }, { "epoch": 0.4318007731086449, "grad_norm": 1.3881123482201607, "learning_rate": 1.2662513640283467e-05, "loss": 0.4866, "step": 12120 }, { "epoch": 0.43183640023513903, "grad_norm": 1.4045585016141406, "learning_rate": 1.2661401338787776e-05, "loss": 0.5805, "step": 12121 }, { "epoch": 0.43187202736163316, "grad_norm": 1.6728326657282286, "learning_rate": 1.2660289001853663e-05, "loss": 0.4873, "step": 12122 }, { "epoch": 0.43190765448812724, "grad_norm": 1.0938987868548975, "learning_rate": 1.2659176629495947e-05, "loss": 0.4811, "step": 12123 }, { "epoch": 0.43194328161462137, "grad_norm": 1.0444559189936848, "learning_rate": 1.265806422172943e-05, "loss": 0.5033, "step": 12124 }, { "epoch": 0.4319789087411155, "grad_norm": 1.124497467399787, "learning_rate": 1.2656951778568933e-05, "loss": 0.3981, "step": 12125 }, { "epoch": 0.4320145358676096, "grad_norm": 1.7012748299036686, "learning_rate": 1.2655839300029265e-05, "loss": 0.5391, "step": 12126 }, { "epoch": 0.4320501629941037, "grad_norm": 1.4250722429573957, "learning_rate": 1.265472678612524e-05, "loss": 0.5807, "step": 12127 }, { "epoch": 0.43208579012059783, "grad_norm": 1.2466503604117076, "learning_rate": 1.2653614236871676e-05, "loss": 0.4615, "step": 12128 }, { "epoch": 0.4321214172470919, "grad_norm": 1.773011620206266, "learning_rate": 1.2652501652283378e-05, "loss": 0.5792, "step": 12129 }, { "epoch": 0.43215704437358604, "grad_norm": 1.624310946253966, "learning_rate": 1.2651389032375172e-05, "loss": 0.6422, "step": 12130 }, { "epoch": 0.43219267150008017, "grad_norm": 1.1546904709859187, "learning_rate": 1.2650276377161865e-05, "loss": 0.4361, "step": 12131 }, { "epoch": 0.43222829862657425, "grad_norm": 1.723799641398856, "learning_rate": 1.2649163686658276e-05, "loss": 0.6253, "step": 12132 }, { "epoch": 0.4322639257530684, "grad_norm": 1.5410750413597492, "learning_rate": 1.2648050960879222e-05, "loss": 0.5336, "step": 12133 }, { "epoch": 0.4322995528795625, "grad_norm": 1.3823036526082975, "learning_rate": 1.2646938199839516e-05, "loss": 0.4416, "step": 12134 }, { "epoch": 0.4323351800060566, "grad_norm": 1.103477731390147, "learning_rate": 1.2645825403553979e-05, "loss": 0.5107, "step": 12135 }, { "epoch": 0.4323708071325507, "grad_norm": 1.133908317248093, "learning_rate": 1.2644712572037427e-05, "loss": 0.4251, "step": 12136 }, { "epoch": 0.43240643425904485, "grad_norm": 1.3264809641646438, "learning_rate": 1.2643599705304678e-05, "loss": 0.5509, "step": 12137 }, { "epoch": 0.4324420613855389, "grad_norm": 1.1370952703596702, "learning_rate": 1.2642486803370553e-05, "loss": 0.4703, "step": 12138 }, { "epoch": 0.43247768851203305, "grad_norm": 1.0238685617302485, "learning_rate": 1.2641373866249868e-05, "loss": 0.4844, "step": 12139 }, { "epoch": 0.4325133156385272, "grad_norm": 1.2059768610458612, "learning_rate": 1.2640260893957445e-05, "loss": 0.432, "step": 12140 }, { "epoch": 0.43254894276502126, "grad_norm": 1.385267873399985, "learning_rate": 1.2639147886508104e-05, "loss": 0.5265, "step": 12141 }, { "epoch": 0.4325845698915154, "grad_norm": 1.116725269793636, "learning_rate": 1.263803484391666e-05, "loss": 0.4589, "step": 12142 }, { "epoch": 0.4326201970180095, "grad_norm": 1.1856870116445912, "learning_rate": 1.2636921766197942e-05, "loss": 0.4691, "step": 12143 }, { "epoch": 0.4326558241445036, "grad_norm": 1.5159451866098086, "learning_rate": 1.2635808653366764e-05, "loss": 0.5921, "step": 12144 }, { "epoch": 0.43269145127099773, "grad_norm": 1.9474035814054358, "learning_rate": 1.2634695505437955e-05, "loss": 0.5004, "step": 12145 }, { "epoch": 0.43272707839749186, "grad_norm": 1.141388348959753, "learning_rate": 1.2633582322426332e-05, "loss": 0.4719, "step": 12146 }, { "epoch": 0.43276270552398594, "grad_norm": 1.503708250059768, "learning_rate": 1.2632469104346722e-05, "loss": 0.5941, "step": 12147 }, { "epoch": 0.43279833265048007, "grad_norm": 1.4903046467519818, "learning_rate": 1.263135585121394e-05, "loss": 0.5576, "step": 12148 }, { "epoch": 0.4328339597769742, "grad_norm": 1.5980178553525224, "learning_rate": 1.263024256304282e-05, "loss": 0.5634, "step": 12149 }, { "epoch": 0.4328695869034683, "grad_norm": 1.2372350492328765, "learning_rate": 1.262912923984818e-05, "loss": 0.4641, "step": 12150 }, { "epoch": 0.4329052140299624, "grad_norm": 1.1698273839203022, "learning_rate": 1.262801588164485e-05, "loss": 0.4925, "step": 12151 }, { "epoch": 0.43294084115645654, "grad_norm": 1.2528208906858584, "learning_rate": 1.262690248844765e-05, "loss": 0.463, "step": 12152 }, { "epoch": 0.43297646828295067, "grad_norm": 1.1054573249673854, "learning_rate": 1.2625789060271405e-05, "loss": 0.4882, "step": 12153 }, { "epoch": 0.43301209540944474, "grad_norm": 1.2826648556913314, "learning_rate": 1.2624675597130946e-05, "loss": 0.5638, "step": 12154 }, { "epoch": 0.4330477225359389, "grad_norm": 1.1398418091878935, "learning_rate": 1.2623562099041096e-05, "loss": 0.4245, "step": 12155 }, { "epoch": 0.433083349662433, "grad_norm": 1.37202132749238, "learning_rate": 1.2622448566016683e-05, "loss": 0.6013, "step": 12156 }, { "epoch": 0.4331189767889271, "grad_norm": 1.025318091253515, "learning_rate": 1.2621334998072534e-05, "loss": 0.4957, "step": 12157 }, { "epoch": 0.4331546039154212, "grad_norm": 1.32077184250389, "learning_rate": 1.262022139522348e-05, "loss": 0.5462, "step": 12158 }, { "epoch": 0.43319023104191534, "grad_norm": 1.1604896905772455, "learning_rate": 1.2619107757484341e-05, "loss": 0.4914, "step": 12159 }, { "epoch": 0.4332258581684094, "grad_norm": 1.3396660277501078, "learning_rate": 1.2617994084869954e-05, "loss": 0.4343, "step": 12160 }, { "epoch": 0.43326148529490355, "grad_norm": 1.1290658875832462, "learning_rate": 1.2616880377395147e-05, "loss": 0.5093, "step": 12161 }, { "epoch": 0.4332971124213977, "grad_norm": 1.4274112099285703, "learning_rate": 1.261576663507475e-05, "loss": 0.5344, "step": 12162 }, { "epoch": 0.43333273954789175, "grad_norm": 1.1027333660119232, "learning_rate": 1.2614652857923589e-05, "loss": 0.4452, "step": 12163 }, { "epoch": 0.4333683666743859, "grad_norm": 1.4332731782773966, "learning_rate": 1.26135390459565e-05, "loss": 0.5391, "step": 12164 }, { "epoch": 0.43340399380088, "grad_norm": 1.051578242598766, "learning_rate": 1.261242519918831e-05, "loss": 0.4818, "step": 12165 }, { "epoch": 0.4334396209273741, "grad_norm": 1.1216442068187518, "learning_rate": 1.2611311317633854e-05, "loss": 0.4492, "step": 12166 }, { "epoch": 0.4334752480538682, "grad_norm": 1.41997940292317, "learning_rate": 1.2610197401307963e-05, "loss": 0.5257, "step": 12167 }, { "epoch": 0.43351087518036235, "grad_norm": 1.0209524448948732, "learning_rate": 1.2609083450225468e-05, "loss": 0.4754, "step": 12168 }, { "epoch": 0.43354650230685643, "grad_norm": 1.5384104294063148, "learning_rate": 1.2607969464401204e-05, "loss": 0.5735, "step": 12169 }, { "epoch": 0.43358212943335056, "grad_norm": 1.0098071772010155, "learning_rate": 1.260685544385e-05, "loss": 0.4304, "step": 12170 }, { "epoch": 0.4336177565598447, "grad_norm": 1.230693861496186, "learning_rate": 1.2605741388586696e-05, "loss": 0.4559, "step": 12171 }, { "epoch": 0.43365338368633877, "grad_norm": 1.297345790994284, "learning_rate": 1.2604627298626126e-05, "loss": 0.4545, "step": 12172 }, { "epoch": 0.4336890108128329, "grad_norm": 1.3987191540215702, "learning_rate": 1.2603513173983121e-05, "loss": 0.5155, "step": 12173 }, { "epoch": 0.43372463793932703, "grad_norm": 1.1683955348034873, "learning_rate": 1.260239901467252e-05, "loss": 0.4394, "step": 12174 }, { "epoch": 0.4337602650658211, "grad_norm": 1.2246840360632905, "learning_rate": 1.2601284820709156e-05, "loss": 0.4496, "step": 12175 }, { "epoch": 0.43379589219231524, "grad_norm": 1.5795201104151766, "learning_rate": 1.2600170592107868e-05, "loss": 0.6347, "step": 12176 }, { "epoch": 0.43383151931880937, "grad_norm": 1.1191263462228753, "learning_rate": 1.259905632888349e-05, "loss": 0.4111, "step": 12177 }, { "epoch": 0.43386714644530344, "grad_norm": 1.1544658450454666, "learning_rate": 1.2597942031050859e-05, "loss": 0.4917, "step": 12178 }, { "epoch": 0.4339027735717976, "grad_norm": 1.0575773608802554, "learning_rate": 1.2596827698624816e-05, "loss": 0.4383, "step": 12179 }, { "epoch": 0.4339384006982917, "grad_norm": 1.3904655943922875, "learning_rate": 1.2595713331620196e-05, "loss": 0.5285, "step": 12180 }, { "epoch": 0.4339740278247858, "grad_norm": 1.034838325420423, "learning_rate": 1.2594598930051838e-05, "loss": 0.4834, "step": 12181 }, { "epoch": 0.4340096549512799, "grad_norm": 1.2991537150641366, "learning_rate": 1.2593484493934582e-05, "loss": 0.5229, "step": 12182 }, { "epoch": 0.43404528207777404, "grad_norm": 1.074086202642936, "learning_rate": 1.2592370023283268e-05, "loss": 0.4998, "step": 12183 }, { "epoch": 0.4340809092042681, "grad_norm": 1.1456792640941351, "learning_rate": 1.2591255518112737e-05, "loss": 0.4315, "step": 12184 }, { "epoch": 0.43411653633076225, "grad_norm": 2.4896500734951954, "learning_rate": 1.2590140978437826e-05, "loss": 0.5493, "step": 12185 }, { "epoch": 0.4341521634572564, "grad_norm": 2.2425454496123356, "learning_rate": 1.2589026404273376e-05, "loss": 0.5676, "step": 12186 }, { "epoch": 0.43418779058375045, "grad_norm": 1.5843948120541451, "learning_rate": 1.2587911795634231e-05, "loss": 0.5539, "step": 12187 }, { "epoch": 0.4342234177102446, "grad_norm": 1.5124270760011556, "learning_rate": 1.2586797152535233e-05, "loss": 0.5771, "step": 12188 }, { "epoch": 0.4342590448367387, "grad_norm": 1.1901319526136744, "learning_rate": 1.2585682474991221e-05, "loss": 0.5506, "step": 12189 }, { "epoch": 0.4342946719632328, "grad_norm": 1.3970931748635593, "learning_rate": 1.258456776301704e-05, "loss": 0.5456, "step": 12190 }, { "epoch": 0.4343302990897269, "grad_norm": 1.116661483051264, "learning_rate": 1.2583453016627533e-05, "loss": 0.4532, "step": 12191 }, { "epoch": 0.43436592621622105, "grad_norm": 1.1124596791372765, "learning_rate": 1.2582338235837544e-05, "loss": 0.519, "step": 12192 }, { "epoch": 0.43440155334271513, "grad_norm": 1.5021075243159647, "learning_rate": 1.2581223420661914e-05, "loss": 0.6106, "step": 12193 }, { "epoch": 0.43443718046920926, "grad_norm": 1.2018946190034854, "learning_rate": 1.258010857111549e-05, "loss": 0.5217, "step": 12194 }, { "epoch": 0.4344728075957034, "grad_norm": 1.0455017636170598, "learning_rate": 1.257899368721312e-05, "loss": 0.4468, "step": 12195 }, { "epoch": 0.43450843472219747, "grad_norm": 1.230908038976415, "learning_rate": 1.2577878768969644e-05, "loss": 0.4494, "step": 12196 }, { "epoch": 0.4345440618486916, "grad_norm": 1.3692895259783653, "learning_rate": 1.2576763816399913e-05, "loss": 0.5702, "step": 12197 }, { "epoch": 0.43457968897518573, "grad_norm": 1.6602174268298393, "learning_rate": 1.2575648829518769e-05, "loss": 0.508, "step": 12198 }, { "epoch": 0.4346153161016798, "grad_norm": 1.178814548776164, "learning_rate": 1.257453380834106e-05, "loss": 0.5178, "step": 12199 }, { "epoch": 0.43465094322817394, "grad_norm": 1.0547352173259184, "learning_rate": 1.2573418752881635e-05, "loss": 0.4521, "step": 12200 }, { "epoch": 0.43468657035466807, "grad_norm": 1.178486481444637, "learning_rate": 1.257230366315534e-05, "loss": 0.4996, "step": 12201 }, { "epoch": 0.43472219748116214, "grad_norm": 1.0738133004298969, "learning_rate": 1.2571188539177027e-05, "loss": 0.4546, "step": 12202 }, { "epoch": 0.4347578246076563, "grad_norm": 1.107830695577512, "learning_rate": 1.2570073380961536e-05, "loss": 0.4663, "step": 12203 }, { "epoch": 0.4347934517341504, "grad_norm": 1.1436316534204978, "learning_rate": 1.2568958188523723e-05, "loss": 0.4995, "step": 12204 }, { "epoch": 0.4348290788606445, "grad_norm": 1.116033440972371, "learning_rate": 1.2567842961878438e-05, "loss": 0.3851, "step": 12205 }, { "epoch": 0.4348647059871386, "grad_norm": 1.1608842995984874, "learning_rate": 1.2566727701040527e-05, "loss": 0.4765, "step": 12206 }, { "epoch": 0.43490033311363274, "grad_norm": 1.1362062381273308, "learning_rate": 1.2565612406024845e-05, "loss": 0.5415, "step": 12207 }, { "epoch": 0.4349359602401268, "grad_norm": 1.302411982727491, "learning_rate": 1.2564497076846239e-05, "loss": 0.5299, "step": 12208 }, { "epoch": 0.43497158736662095, "grad_norm": 1.3570252045520306, "learning_rate": 1.2563381713519562e-05, "loss": 0.5986, "step": 12209 }, { "epoch": 0.4350072144931151, "grad_norm": 1.0850327218185956, "learning_rate": 1.2562266316059665e-05, "loss": 0.4572, "step": 12210 }, { "epoch": 0.43504284161960916, "grad_norm": 1.6994205155664466, "learning_rate": 1.2561150884481403e-05, "loss": 0.5577, "step": 12211 }, { "epoch": 0.4350784687461033, "grad_norm": 1.653732691209012, "learning_rate": 1.2560035418799626e-05, "loss": 0.5832, "step": 12212 }, { "epoch": 0.4351140958725974, "grad_norm": 1.5154724893579337, "learning_rate": 1.2558919919029189e-05, "loss": 0.5253, "step": 12213 }, { "epoch": 0.4351497229990915, "grad_norm": 1.1262431616031896, "learning_rate": 1.2557804385184944e-05, "loss": 0.4362, "step": 12214 }, { "epoch": 0.4351853501255856, "grad_norm": 1.1623103662096086, "learning_rate": 1.2556688817281745e-05, "loss": 0.476, "step": 12215 }, { "epoch": 0.43522097725207975, "grad_norm": 1.635458227454768, "learning_rate": 1.2555573215334449e-05, "loss": 0.5882, "step": 12216 }, { "epoch": 0.43525660437857383, "grad_norm": 1.2112379226637693, "learning_rate": 1.2554457579357906e-05, "loss": 0.5241, "step": 12217 }, { "epoch": 0.43529223150506796, "grad_norm": 1.1417754500631927, "learning_rate": 1.2553341909366979e-05, "loss": 0.477, "step": 12218 }, { "epoch": 0.4353278586315621, "grad_norm": 1.1582816283933965, "learning_rate": 1.2552226205376517e-05, "loss": 0.4995, "step": 12219 }, { "epoch": 0.43536348575805617, "grad_norm": 1.12426803760122, "learning_rate": 1.255111046740138e-05, "loss": 0.4901, "step": 12220 }, { "epoch": 0.4353991128845503, "grad_norm": 1.117631049620709, "learning_rate": 1.2549994695456424e-05, "loss": 0.4805, "step": 12221 }, { "epoch": 0.43543474001104443, "grad_norm": 1.4772313908916717, "learning_rate": 1.2548878889556508e-05, "loss": 0.6104, "step": 12222 }, { "epoch": 0.4354703671375385, "grad_norm": 1.0795125846120623, "learning_rate": 1.2547763049716486e-05, "loss": 0.43, "step": 12223 }, { "epoch": 0.43550599426403264, "grad_norm": 1.531231971101479, "learning_rate": 1.2546647175951218e-05, "loss": 0.5636, "step": 12224 }, { "epoch": 0.43554162139052677, "grad_norm": 1.8370878233581276, "learning_rate": 1.2545531268275565e-05, "loss": 0.6205, "step": 12225 }, { "epoch": 0.43557724851702084, "grad_norm": 1.6051530538481744, "learning_rate": 1.254441532670438e-05, "loss": 0.5272, "step": 12226 }, { "epoch": 0.435612875643515, "grad_norm": 1.505764225933821, "learning_rate": 1.2543299351252532e-05, "loss": 0.6425, "step": 12227 }, { "epoch": 0.4356485027700091, "grad_norm": 1.188316469459248, "learning_rate": 1.2542183341934873e-05, "loss": 0.5021, "step": 12228 }, { "epoch": 0.4356841298965032, "grad_norm": 1.2645710932605678, "learning_rate": 1.2541067298766263e-05, "loss": 0.5725, "step": 12229 }, { "epoch": 0.4357197570229973, "grad_norm": 1.5134215853880113, "learning_rate": 1.2539951221761567e-05, "loss": 0.5048, "step": 12230 }, { "epoch": 0.43575538414949144, "grad_norm": 1.1720059162014111, "learning_rate": 1.2538835110935648e-05, "loss": 0.4896, "step": 12231 }, { "epoch": 0.4357910112759855, "grad_norm": 1.4551274529321394, "learning_rate": 1.2537718966303363e-05, "loss": 0.563, "step": 12232 }, { "epoch": 0.43582663840247965, "grad_norm": 1.1705879440628355, "learning_rate": 1.2536602787879576e-05, "loss": 0.5375, "step": 12233 }, { "epoch": 0.4358622655289738, "grad_norm": 1.3828105542636155, "learning_rate": 1.253548657567915e-05, "loss": 0.5262, "step": 12234 }, { "epoch": 0.43589789265546786, "grad_norm": 1.569698411066532, "learning_rate": 1.2534370329716949e-05, "loss": 0.5705, "step": 12235 }, { "epoch": 0.435933519781962, "grad_norm": 1.484160555162373, "learning_rate": 1.2533254050007837e-05, "loss": 0.5729, "step": 12236 }, { "epoch": 0.4359691469084561, "grad_norm": 1.056622259792043, "learning_rate": 1.2532137736566674e-05, "loss": 0.4532, "step": 12237 }, { "epoch": 0.4360047740349502, "grad_norm": 1.8012599308311161, "learning_rate": 1.2531021389408329e-05, "loss": 0.6154, "step": 12238 }, { "epoch": 0.4360404011614443, "grad_norm": 1.5971770893823563, "learning_rate": 1.2529905008547662e-05, "loss": 0.6345, "step": 12239 }, { "epoch": 0.43607602828793846, "grad_norm": 1.3659184749323918, "learning_rate": 1.2528788593999544e-05, "loss": 0.528, "step": 12240 }, { "epoch": 0.43611165541443253, "grad_norm": 1.2418902952203132, "learning_rate": 1.252767214577884e-05, "loss": 0.4884, "step": 12241 }, { "epoch": 0.43614728254092666, "grad_norm": 1.0433674071879613, "learning_rate": 1.252655566390041e-05, "loss": 0.4336, "step": 12242 }, { "epoch": 0.4361829096674208, "grad_norm": 1.8573456518244433, "learning_rate": 1.2525439148379127e-05, "loss": 0.5559, "step": 12243 }, { "epoch": 0.43621853679391487, "grad_norm": 1.0498356587201818, "learning_rate": 1.2524322599229862e-05, "loss": 0.4154, "step": 12244 }, { "epoch": 0.436254163920409, "grad_norm": 1.1944352348901708, "learning_rate": 1.2523206016467468e-05, "loss": 0.4892, "step": 12245 }, { "epoch": 0.43628979104690313, "grad_norm": 1.6831801204051322, "learning_rate": 1.2522089400106829e-05, "loss": 0.6023, "step": 12246 }, { "epoch": 0.4363254181733972, "grad_norm": 1.1302871178451281, "learning_rate": 1.2520972750162804e-05, "loss": 0.4426, "step": 12247 }, { "epoch": 0.43636104529989134, "grad_norm": 1.4586394427614129, "learning_rate": 1.251985606665027e-05, "loss": 0.4912, "step": 12248 }, { "epoch": 0.43639667242638547, "grad_norm": 1.2808123421708812, "learning_rate": 1.2518739349584083e-05, "loss": 0.4786, "step": 12249 }, { "epoch": 0.43643229955287954, "grad_norm": 1.7346012562359194, "learning_rate": 1.2517622598979127e-05, "loss": 0.5402, "step": 12250 }, { "epoch": 0.4364679266793737, "grad_norm": 1.4494634110011715, "learning_rate": 1.2516505814850264e-05, "loss": 0.6107, "step": 12251 }, { "epoch": 0.4365035538058678, "grad_norm": 1.1922595440881885, "learning_rate": 1.2515388997212366e-05, "loss": 0.4832, "step": 12252 }, { "epoch": 0.4365391809323619, "grad_norm": 1.1145178975473848, "learning_rate": 1.2514272146080307e-05, "loss": 0.4786, "step": 12253 }, { "epoch": 0.436574808058856, "grad_norm": 0.9829796639161541, "learning_rate": 1.2513155261468957e-05, "loss": 0.4478, "step": 12254 }, { "epoch": 0.43661043518535014, "grad_norm": 1.346743680104273, "learning_rate": 1.251203834339319e-05, "loss": 0.5377, "step": 12255 }, { "epoch": 0.4366460623118442, "grad_norm": 2.170886371155723, "learning_rate": 1.251092139186787e-05, "loss": 0.6302, "step": 12256 }, { "epoch": 0.43668168943833835, "grad_norm": 1.0555095139576838, "learning_rate": 1.2509804406907881e-05, "loss": 0.434, "step": 12257 }, { "epoch": 0.4367173165648325, "grad_norm": 1.1132368323487258, "learning_rate": 1.2508687388528094e-05, "loss": 0.4413, "step": 12258 }, { "epoch": 0.43675294369132656, "grad_norm": 1.0700178857933242, "learning_rate": 1.2507570336743383e-05, "loss": 0.4954, "step": 12259 }, { "epoch": 0.4367885708178207, "grad_norm": 1.0268607946070065, "learning_rate": 1.2506453251568614e-05, "loss": 0.4939, "step": 12260 }, { "epoch": 0.4368241979443148, "grad_norm": 1.4283530821786263, "learning_rate": 1.2505336133018671e-05, "loss": 0.6323, "step": 12261 }, { "epoch": 0.4368598250708089, "grad_norm": 1.7421645506314913, "learning_rate": 1.2504218981108424e-05, "loss": 0.5924, "step": 12262 }, { "epoch": 0.436895452197303, "grad_norm": 1.4207875673326276, "learning_rate": 1.2503101795852752e-05, "loss": 0.5734, "step": 12263 }, { "epoch": 0.43693107932379716, "grad_norm": 1.5262833748193654, "learning_rate": 1.2501984577266531e-05, "loss": 0.4523, "step": 12264 }, { "epoch": 0.43696670645029123, "grad_norm": 1.0276388906724199, "learning_rate": 1.2500867325364636e-05, "loss": 0.4295, "step": 12265 }, { "epoch": 0.43700233357678536, "grad_norm": 1.6443042392120413, "learning_rate": 1.2499750040161947e-05, "loss": 0.5539, "step": 12266 }, { "epoch": 0.4370379607032795, "grad_norm": 1.2944014479363135, "learning_rate": 1.2498632721673337e-05, "loss": 0.4987, "step": 12267 }, { "epoch": 0.43707358782977357, "grad_norm": 1.997598624328785, "learning_rate": 1.2497515369913684e-05, "loss": 0.5328, "step": 12268 }, { "epoch": 0.4371092149562677, "grad_norm": 1.2523148886118154, "learning_rate": 1.249639798489787e-05, "loss": 0.4918, "step": 12269 }, { "epoch": 0.43714484208276183, "grad_norm": 1.5406966921864644, "learning_rate": 1.2495280566640773e-05, "loss": 0.5685, "step": 12270 }, { "epoch": 0.4371804692092559, "grad_norm": 1.1293527500280143, "learning_rate": 1.249416311515727e-05, "loss": 0.4234, "step": 12271 }, { "epoch": 0.43721609633575004, "grad_norm": 1.0329296623541822, "learning_rate": 1.2493045630462242e-05, "loss": 0.4984, "step": 12272 }, { "epoch": 0.43725172346224417, "grad_norm": 1.6100945700925597, "learning_rate": 1.2491928112570568e-05, "loss": 0.5763, "step": 12273 }, { "epoch": 0.43728735058873824, "grad_norm": 1.2669653108925296, "learning_rate": 1.249081056149713e-05, "loss": 0.431, "step": 12274 }, { "epoch": 0.4373229777152324, "grad_norm": 1.7204492911440832, "learning_rate": 1.2489692977256807e-05, "loss": 0.5178, "step": 12275 }, { "epoch": 0.4373586048417265, "grad_norm": 1.17985704780096, "learning_rate": 1.2488575359864483e-05, "loss": 0.4437, "step": 12276 }, { "epoch": 0.4373942319682206, "grad_norm": 1.1492897285795642, "learning_rate": 1.248745770933504e-05, "loss": 0.4778, "step": 12277 }, { "epoch": 0.4374298590947147, "grad_norm": 1.4812835222738663, "learning_rate": 1.2486340025683359e-05, "loss": 0.6038, "step": 12278 }, { "epoch": 0.43746548622120884, "grad_norm": 1.1016259199489655, "learning_rate": 1.248522230892432e-05, "loss": 0.4394, "step": 12279 }, { "epoch": 0.4375011133477029, "grad_norm": 1.1437731699719378, "learning_rate": 1.2484104559072812e-05, "loss": 0.4812, "step": 12280 }, { "epoch": 0.43753674047419705, "grad_norm": 1.7536492994386, "learning_rate": 1.2482986776143715e-05, "loss": 0.5565, "step": 12281 }, { "epoch": 0.4375723676006912, "grad_norm": 1.5478233213401336, "learning_rate": 1.2481868960151911e-05, "loss": 0.5674, "step": 12282 }, { "epoch": 0.43760799472718526, "grad_norm": 1.3261638512847613, "learning_rate": 1.248075111111229e-05, "loss": 0.5458, "step": 12283 }, { "epoch": 0.4376436218536794, "grad_norm": 1.3054379916700272, "learning_rate": 1.2479633229039734e-05, "loss": 0.4606, "step": 12284 }, { "epoch": 0.4376792489801735, "grad_norm": 1.8503493772088748, "learning_rate": 1.2478515313949127e-05, "loss": 0.6323, "step": 12285 }, { "epoch": 0.4377148761066676, "grad_norm": 1.1587977818332849, "learning_rate": 1.2477397365855355e-05, "loss": 0.4587, "step": 12286 }, { "epoch": 0.4377505032331617, "grad_norm": 1.446449287077371, "learning_rate": 1.2476279384773307e-05, "loss": 0.5709, "step": 12287 }, { "epoch": 0.43778613035965586, "grad_norm": 1.4213370075297613, "learning_rate": 1.2475161370717868e-05, "loss": 0.6242, "step": 12288 }, { "epoch": 0.43782175748614993, "grad_norm": 1.3185558991207478, "learning_rate": 1.2474043323703926e-05, "loss": 0.6142, "step": 12289 }, { "epoch": 0.43785738461264406, "grad_norm": 0.9718228025549775, "learning_rate": 1.2472925243746365e-05, "loss": 0.4305, "step": 12290 }, { "epoch": 0.4378930117391382, "grad_norm": 1.296515486358367, "learning_rate": 1.2471807130860079e-05, "loss": 0.5395, "step": 12291 }, { "epoch": 0.43792863886563227, "grad_norm": 1.1210125150994688, "learning_rate": 1.2470688985059954e-05, "loss": 0.4864, "step": 12292 }, { "epoch": 0.4379642659921264, "grad_norm": 1.187901665062945, "learning_rate": 1.2469570806360876e-05, "loss": 0.4828, "step": 12293 }, { "epoch": 0.43799989311862053, "grad_norm": 1.239687084280943, "learning_rate": 1.2468452594777738e-05, "loss": 0.5691, "step": 12294 }, { "epoch": 0.4380355202451146, "grad_norm": 0.9943217122721829, "learning_rate": 1.2467334350325429e-05, "loss": 0.424, "step": 12295 }, { "epoch": 0.43807114737160874, "grad_norm": 1.1282271581487635, "learning_rate": 1.2466216073018838e-05, "loss": 0.5206, "step": 12296 }, { "epoch": 0.43810677449810287, "grad_norm": 1.5353981628795443, "learning_rate": 1.2465097762872853e-05, "loss": 0.558, "step": 12297 }, { "epoch": 0.43814240162459694, "grad_norm": 1.073360202124728, "learning_rate": 1.2463979419902371e-05, "loss": 0.468, "step": 12298 }, { "epoch": 0.4381780287510911, "grad_norm": 1.1457166837100092, "learning_rate": 1.2462861044122279e-05, "loss": 0.5114, "step": 12299 }, { "epoch": 0.4382136558775852, "grad_norm": 1.0636138776855466, "learning_rate": 1.2461742635547473e-05, "loss": 0.5061, "step": 12300 }, { "epoch": 0.4382492830040793, "grad_norm": 1.0752911334503519, "learning_rate": 1.2460624194192842e-05, "loss": 0.418, "step": 12301 }, { "epoch": 0.4382849101305734, "grad_norm": 1.728482647735828, "learning_rate": 1.2459505720073281e-05, "loss": 0.6277, "step": 12302 }, { "epoch": 0.43832053725706754, "grad_norm": 1.1266316975860475, "learning_rate": 1.2458387213203682e-05, "loss": 0.5205, "step": 12303 }, { "epoch": 0.4383561643835616, "grad_norm": 1.0503740176556917, "learning_rate": 1.2457268673598939e-05, "loss": 0.462, "step": 12304 }, { "epoch": 0.43839179151005575, "grad_norm": 1.1007935314903525, "learning_rate": 1.2456150101273945e-05, "loss": 0.4615, "step": 12305 }, { "epoch": 0.4384274186365499, "grad_norm": 1.5990297255100865, "learning_rate": 1.2455031496243597e-05, "loss": 0.5803, "step": 12306 }, { "epoch": 0.43846304576304396, "grad_norm": 1.0823535746297028, "learning_rate": 1.245391285852279e-05, "loss": 0.4755, "step": 12307 }, { "epoch": 0.4384986728895381, "grad_norm": 1.3371249787647195, "learning_rate": 1.2452794188126413e-05, "loss": 0.609, "step": 12308 }, { "epoch": 0.4385343000160322, "grad_norm": 1.125725425642206, "learning_rate": 1.245167548506937e-05, "loss": 0.4764, "step": 12309 }, { "epoch": 0.4385699271425263, "grad_norm": 1.1399692442391918, "learning_rate": 1.2450556749366556e-05, "loss": 0.4726, "step": 12310 }, { "epoch": 0.4386055542690204, "grad_norm": 1.795166571402662, "learning_rate": 1.2449437981032864e-05, "loss": 0.6087, "step": 12311 }, { "epoch": 0.43864118139551456, "grad_norm": 1.4876921875672342, "learning_rate": 1.2448319180083193e-05, "loss": 0.5435, "step": 12312 }, { "epoch": 0.43867680852200863, "grad_norm": 1.1197119754384142, "learning_rate": 1.2447200346532443e-05, "loss": 0.4318, "step": 12313 }, { "epoch": 0.43871243564850276, "grad_norm": 1.526519936730432, "learning_rate": 1.244608148039551e-05, "loss": 0.5717, "step": 12314 }, { "epoch": 0.4387480627749969, "grad_norm": 1.0605491858194827, "learning_rate": 1.244496258168729e-05, "loss": 0.4378, "step": 12315 }, { "epoch": 0.43878368990149097, "grad_norm": 1.3047234801899825, "learning_rate": 1.2443843650422684e-05, "loss": 0.4636, "step": 12316 }, { "epoch": 0.4388193170279851, "grad_norm": 1.07801832701988, "learning_rate": 1.2442724686616594e-05, "loss": 0.4388, "step": 12317 }, { "epoch": 0.43885494415447923, "grad_norm": 1.1399112678669494, "learning_rate": 1.2441605690283915e-05, "loss": 0.4655, "step": 12318 }, { "epoch": 0.4388905712809733, "grad_norm": 1.0921678025039836, "learning_rate": 1.244048666143955e-05, "loss": 0.4833, "step": 12319 }, { "epoch": 0.43892619840746744, "grad_norm": 1.0712692922327383, "learning_rate": 1.2439367600098402e-05, "loss": 0.442, "step": 12320 }, { "epoch": 0.43896182553396157, "grad_norm": 1.136828849894783, "learning_rate": 1.2438248506275366e-05, "loss": 0.4916, "step": 12321 }, { "epoch": 0.43899745266045564, "grad_norm": 1.4710315189769425, "learning_rate": 1.2437129379985349e-05, "loss": 0.6012, "step": 12322 }, { "epoch": 0.4390330797869498, "grad_norm": 1.5047871683045821, "learning_rate": 1.243601022124325e-05, "loss": 0.55, "step": 12323 }, { "epoch": 0.4390687069134439, "grad_norm": 1.1251404962753733, "learning_rate": 1.2434891030063971e-05, "loss": 0.4586, "step": 12324 }, { "epoch": 0.439104334039938, "grad_norm": 1.4805039399149054, "learning_rate": 1.243377180646242e-05, "loss": 0.5549, "step": 12325 }, { "epoch": 0.4391399611664321, "grad_norm": 1.0135568518812719, "learning_rate": 1.2432652550453494e-05, "loss": 0.4544, "step": 12326 }, { "epoch": 0.43917558829292624, "grad_norm": 1.1142154773704627, "learning_rate": 1.2431533262052099e-05, "loss": 0.4271, "step": 12327 }, { "epoch": 0.4392112154194203, "grad_norm": 1.0808139963308352, "learning_rate": 1.2430413941273138e-05, "loss": 0.4673, "step": 12328 }, { "epoch": 0.43924684254591445, "grad_norm": 1.5383785431642578, "learning_rate": 1.242929458813152e-05, "loss": 0.5692, "step": 12329 }, { "epoch": 0.4392824696724086, "grad_norm": 1.261951106842101, "learning_rate": 1.2428175202642144e-05, "loss": 0.6074, "step": 12330 }, { "epoch": 0.43931809679890266, "grad_norm": 1.3527383861158284, "learning_rate": 1.2427055784819917e-05, "loss": 0.5855, "step": 12331 }, { "epoch": 0.4393537239253968, "grad_norm": 1.101989269756115, "learning_rate": 1.2425936334679745e-05, "loss": 0.4556, "step": 12332 }, { "epoch": 0.4393893510518909, "grad_norm": 1.0879761189204524, "learning_rate": 1.2424816852236539e-05, "loss": 0.4408, "step": 12333 }, { "epoch": 0.439424978178385, "grad_norm": 1.1847933296801572, "learning_rate": 1.24236973375052e-05, "loss": 0.4392, "step": 12334 }, { "epoch": 0.4394606053048791, "grad_norm": 1.1651775432375298, "learning_rate": 1.2422577790500634e-05, "loss": 0.4659, "step": 12335 }, { "epoch": 0.43949623243137326, "grad_norm": 1.6468997953128008, "learning_rate": 1.2421458211237755e-05, "loss": 0.6228, "step": 12336 }, { "epoch": 0.43953185955786733, "grad_norm": 1.4165723922784519, "learning_rate": 1.2420338599731467e-05, "loss": 0.5484, "step": 12337 }, { "epoch": 0.43956748668436146, "grad_norm": 1.2021884458685443, "learning_rate": 1.2419218955996677e-05, "loss": 0.4656, "step": 12338 }, { "epoch": 0.4396031138108556, "grad_norm": 1.9964154629592479, "learning_rate": 1.24180992800483e-05, "loss": 0.5353, "step": 12339 }, { "epoch": 0.4396387409373497, "grad_norm": 1.2204898543165688, "learning_rate": 1.2416979571901238e-05, "loss": 0.5116, "step": 12340 }, { "epoch": 0.4396743680638438, "grad_norm": 1.0431185927461262, "learning_rate": 1.2415859831570405e-05, "loss": 0.3956, "step": 12341 }, { "epoch": 0.43970999519033793, "grad_norm": 1.244620871210601, "learning_rate": 1.2414740059070707e-05, "loss": 0.4653, "step": 12342 }, { "epoch": 0.43974562231683206, "grad_norm": 1.1346862173193653, "learning_rate": 1.2413620254417058e-05, "loss": 0.4666, "step": 12343 }, { "epoch": 0.43978124944332614, "grad_norm": 1.0733770249373262, "learning_rate": 1.2412500417624367e-05, "loss": 0.4386, "step": 12344 }, { "epoch": 0.43981687656982027, "grad_norm": 1.4800609508085645, "learning_rate": 1.241138054870755e-05, "loss": 0.5547, "step": 12345 }, { "epoch": 0.4398525036963144, "grad_norm": 1.5342622262732837, "learning_rate": 1.2410260647681513e-05, "loss": 0.5373, "step": 12346 }, { "epoch": 0.4398881308228085, "grad_norm": 1.2745196476884073, "learning_rate": 1.2409140714561172e-05, "loss": 0.5892, "step": 12347 }, { "epoch": 0.4399237579493026, "grad_norm": 1.567803302313279, "learning_rate": 1.240802074936144e-05, "loss": 0.5889, "step": 12348 }, { "epoch": 0.43995938507579674, "grad_norm": 1.460818817766026, "learning_rate": 1.2406900752097228e-05, "loss": 0.5926, "step": 12349 }, { "epoch": 0.4399950122022908, "grad_norm": 1.07282661248356, "learning_rate": 1.2405780722783447e-05, "loss": 0.4818, "step": 12350 }, { "epoch": 0.44003063932878494, "grad_norm": 1.0146956250963608, "learning_rate": 1.2404660661435017e-05, "loss": 0.4861, "step": 12351 }, { "epoch": 0.4400662664552791, "grad_norm": 1.0961219856588034, "learning_rate": 1.240354056806685e-05, "loss": 0.4435, "step": 12352 }, { "epoch": 0.44010189358177315, "grad_norm": 1.589807722495065, "learning_rate": 1.2402420442693858e-05, "loss": 0.5285, "step": 12353 }, { "epoch": 0.4401375207082673, "grad_norm": 1.2863240990280649, "learning_rate": 1.2401300285330962e-05, "loss": 0.5425, "step": 12354 }, { "epoch": 0.4401731478347614, "grad_norm": 1.7399985024896574, "learning_rate": 1.2400180095993072e-05, "loss": 0.6075, "step": 12355 }, { "epoch": 0.4402087749612555, "grad_norm": 1.5076031781860424, "learning_rate": 1.2399059874695104e-05, "loss": 0.5445, "step": 12356 }, { "epoch": 0.4402444020877496, "grad_norm": 1.106487055505073, "learning_rate": 1.239793962145198e-05, "loss": 0.5001, "step": 12357 }, { "epoch": 0.44028002921424375, "grad_norm": 1.8364517022611082, "learning_rate": 1.2396819336278614e-05, "loss": 0.6053, "step": 12358 }, { "epoch": 0.4403156563407378, "grad_norm": 1.352880053917248, "learning_rate": 1.2395699019189924e-05, "loss": 0.5282, "step": 12359 }, { "epoch": 0.44035128346723196, "grad_norm": 1.5638486159319822, "learning_rate": 1.2394578670200825e-05, "loss": 0.5186, "step": 12360 }, { "epoch": 0.4403869105937261, "grad_norm": 1.3191816988724547, "learning_rate": 1.2393458289326237e-05, "loss": 0.5042, "step": 12361 }, { "epoch": 0.44042253772022016, "grad_norm": 1.6889771322853335, "learning_rate": 1.2392337876581081e-05, "loss": 0.5335, "step": 12362 }, { "epoch": 0.4404581648467143, "grad_norm": 1.1747235678505512, "learning_rate": 1.2391217431980273e-05, "loss": 0.5301, "step": 12363 }, { "epoch": 0.4404937919732084, "grad_norm": 1.2584386627944273, "learning_rate": 1.2390096955538733e-05, "loss": 0.5644, "step": 12364 }, { "epoch": 0.4405294190997025, "grad_norm": 1.4320014817369506, "learning_rate": 1.2388976447271384e-05, "loss": 0.5285, "step": 12365 }, { "epoch": 0.44056504622619663, "grad_norm": 1.137392950686871, "learning_rate": 1.2387855907193141e-05, "loss": 0.4217, "step": 12366 }, { "epoch": 0.44060067335269076, "grad_norm": 1.229378896628401, "learning_rate": 1.2386735335318926e-05, "loss": 0.4558, "step": 12367 }, { "epoch": 0.44063630047918484, "grad_norm": 1.1096134774210757, "learning_rate": 1.2385614731663667e-05, "loss": 0.4793, "step": 12368 }, { "epoch": 0.44067192760567897, "grad_norm": 1.5131159183269347, "learning_rate": 1.2384494096242277e-05, "loss": 0.5112, "step": 12369 }, { "epoch": 0.4407075547321731, "grad_norm": 1.092022148134043, "learning_rate": 1.2383373429069683e-05, "loss": 0.4825, "step": 12370 }, { "epoch": 0.4407431818586672, "grad_norm": 1.6056218048070652, "learning_rate": 1.2382252730160804e-05, "loss": 0.5495, "step": 12371 }, { "epoch": 0.4407788089851613, "grad_norm": 1.6471828323284927, "learning_rate": 1.2381131999530564e-05, "loss": 0.5461, "step": 12372 }, { "epoch": 0.44081443611165544, "grad_norm": 1.5866556875858897, "learning_rate": 1.238001123719389e-05, "loss": 0.5955, "step": 12373 }, { "epoch": 0.4408500632381495, "grad_norm": 1.1734672584414396, "learning_rate": 1.2378890443165704e-05, "loss": 0.4603, "step": 12374 }, { "epoch": 0.44088569036464365, "grad_norm": 1.3968084791693949, "learning_rate": 1.2377769617460927e-05, "loss": 0.5898, "step": 12375 }, { "epoch": 0.4409213174911378, "grad_norm": 1.1412239790065848, "learning_rate": 1.2376648760094482e-05, "loss": 0.4414, "step": 12376 }, { "epoch": 0.44095694461763185, "grad_norm": 1.14512748597139, "learning_rate": 1.2375527871081303e-05, "loss": 0.4963, "step": 12377 }, { "epoch": 0.440992571744126, "grad_norm": 1.2883463875806602, "learning_rate": 1.2374406950436308e-05, "loss": 0.5773, "step": 12378 }, { "epoch": 0.4410281988706201, "grad_norm": 2.0788495032616883, "learning_rate": 1.2373285998174424e-05, "loss": 0.5324, "step": 12379 }, { "epoch": 0.4410638259971142, "grad_norm": 1.2350808390910617, "learning_rate": 1.2372165014310578e-05, "loss": 0.4714, "step": 12380 }, { "epoch": 0.4410994531236083, "grad_norm": 1.5380022954203374, "learning_rate": 1.2371043998859698e-05, "loss": 0.5666, "step": 12381 }, { "epoch": 0.44113508025010245, "grad_norm": 1.5174034789482536, "learning_rate": 1.236992295183671e-05, "loss": 0.4573, "step": 12382 }, { "epoch": 0.4411707073765965, "grad_norm": 1.366008045111528, "learning_rate": 1.236880187325654e-05, "loss": 0.5612, "step": 12383 }, { "epoch": 0.44120633450309066, "grad_norm": 1.7520301505760036, "learning_rate": 1.2367680763134117e-05, "loss": 0.6481, "step": 12384 }, { "epoch": 0.4412419616295848, "grad_norm": 1.196820965891578, "learning_rate": 1.2366559621484373e-05, "loss": 0.4546, "step": 12385 }, { "epoch": 0.44127758875607886, "grad_norm": 1.1738110031281388, "learning_rate": 1.2365438448322234e-05, "loss": 0.4725, "step": 12386 }, { "epoch": 0.441313215882573, "grad_norm": 1.0262147246918447, "learning_rate": 1.2364317243662623e-05, "loss": 0.4077, "step": 12387 }, { "epoch": 0.4413488430090671, "grad_norm": 1.5206286482808204, "learning_rate": 1.236319600752048e-05, "loss": 0.5475, "step": 12388 }, { "epoch": 0.4413844701355612, "grad_norm": 1.1473473952690967, "learning_rate": 1.236207473991073e-05, "loss": 0.4504, "step": 12389 }, { "epoch": 0.44142009726205533, "grad_norm": 1.0425420683049442, "learning_rate": 1.2360953440848303e-05, "loss": 0.4461, "step": 12390 }, { "epoch": 0.44145572438854946, "grad_norm": 1.284749829234386, "learning_rate": 1.2359832110348126e-05, "loss": 0.4881, "step": 12391 }, { "epoch": 0.44149135151504354, "grad_norm": 1.1174849132312872, "learning_rate": 1.2358710748425142e-05, "loss": 0.4759, "step": 12392 }, { "epoch": 0.44152697864153767, "grad_norm": 1.2033648264665158, "learning_rate": 1.2357589355094275e-05, "loss": 0.4408, "step": 12393 }, { "epoch": 0.4415626057680318, "grad_norm": 1.5438379317537823, "learning_rate": 1.2356467930370455e-05, "loss": 0.5568, "step": 12394 }, { "epoch": 0.4415982328945259, "grad_norm": 1.5857827664914312, "learning_rate": 1.2355346474268619e-05, "loss": 0.469, "step": 12395 }, { "epoch": 0.44163386002102, "grad_norm": 1.2509791528884158, "learning_rate": 1.2354224986803697e-05, "loss": 0.4199, "step": 12396 }, { "epoch": 0.44166948714751414, "grad_norm": 1.2062417372093779, "learning_rate": 1.2353103467990627e-05, "loss": 0.4917, "step": 12397 }, { "epoch": 0.4417051142740082, "grad_norm": 1.2218092231641509, "learning_rate": 1.2351981917844335e-05, "loss": 0.5184, "step": 12398 }, { "epoch": 0.44174074140050235, "grad_norm": 1.0089131330084433, "learning_rate": 1.2350860336379765e-05, "loss": 0.442, "step": 12399 }, { "epoch": 0.4417763685269965, "grad_norm": 1.6368552257041078, "learning_rate": 1.2349738723611845e-05, "loss": 0.5999, "step": 12400 }, { "epoch": 0.44181199565349055, "grad_norm": 1.597505543468287, "learning_rate": 1.2348617079555507e-05, "loss": 0.6075, "step": 12401 }, { "epoch": 0.4418476227799847, "grad_norm": 1.1243589980266944, "learning_rate": 1.2347495404225693e-05, "loss": 0.4542, "step": 12402 }, { "epoch": 0.4418832499064788, "grad_norm": 1.221450944188967, "learning_rate": 1.2346373697637339e-05, "loss": 0.4486, "step": 12403 }, { "epoch": 0.4419188770329729, "grad_norm": 1.3325528283131463, "learning_rate": 1.234525195980538e-05, "loss": 0.4865, "step": 12404 }, { "epoch": 0.441954504159467, "grad_norm": 1.1064759908893629, "learning_rate": 1.234413019074475e-05, "loss": 0.4821, "step": 12405 }, { "epoch": 0.44199013128596115, "grad_norm": 1.7670877970439363, "learning_rate": 1.2343008390470388e-05, "loss": 0.5614, "step": 12406 }, { "epoch": 0.4420257584124552, "grad_norm": 1.1145733988689666, "learning_rate": 1.2341886558997231e-05, "loss": 0.4332, "step": 12407 }, { "epoch": 0.44206138553894936, "grad_norm": 1.5429036811341492, "learning_rate": 1.234076469634022e-05, "loss": 0.5209, "step": 12408 }, { "epoch": 0.4420970126654435, "grad_norm": 1.7345486801372292, "learning_rate": 1.233964280251429e-05, "loss": 0.619, "step": 12409 }, { "epoch": 0.44213263979193757, "grad_norm": 1.2833107128629349, "learning_rate": 1.2338520877534379e-05, "loss": 0.4401, "step": 12410 }, { "epoch": 0.4421682669184317, "grad_norm": 2.0770240281022527, "learning_rate": 1.2337398921415431e-05, "loss": 0.5814, "step": 12411 }, { "epoch": 0.4422038940449258, "grad_norm": 1.4175543290117432, "learning_rate": 1.2336276934172382e-05, "loss": 0.5863, "step": 12412 }, { "epoch": 0.4422395211714199, "grad_norm": 1.2179352383612865, "learning_rate": 1.2335154915820167e-05, "loss": 0.4606, "step": 12413 }, { "epoch": 0.44227514829791403, "grad_norm": 0.906794135696234, "learning_rate": 1.2334032866373737e-05, "loss": 0.4172, "step": 12414 }, { "epoch": 0.44231077542440816, "grad_norm": 1.1006895417395852, "learning_rate": 1.233291078584803e-05, "loss": 0.4806, "step": 12415 }, { "epoch": 0.44234640255090224, "grad_norm": 1.0955680024024945, "learning_rate": 1.2331788674257982e-05, "loss": 0.4572, "step": 12416 }, { "epoch": 0.44238202967739637, "grad_norm": 1.4542480056577423, "learning_rate": 1.2330666531618538e-05, "loss": 0.5359, "step": 12417 }, { "epoch": 0.4424176568038905, "grad_norm": 1.1569192715109002, "learning_rate": 1.2329544357944643e-05, "loss": 0.4793, "step": 12418 }, { "epoch": 0.4424532839303846, "grad_norm": 1.3461875980902915, "learning_rate": 1.2328422153251234e-05, "loss": 0.5675, "step": 12419 }, { "epoch": 0.4424889110568787, "grad_norm": 1.5555877317527156, "learning_rate": 1.2327299917553257e-05, "loss": 0.6421, "step": 12420 }, { "epoch": 0.44252453818337284, "grad_norm": 1.4080492291703137, "learning_rate": 1.2326177650865656e-05, "loss": 0.588, "step": 12421 }, { "epoch": 0.4425601653098669, "grad_norm": 1.0455431853957746, "learning_rate": 1.2325055353203373e-05, "loss": 0.4321, "step": 12422 }, { "epoch": 0.44259579243636105, "grad_norm": 1.4023260826976887, "learning_rate": 1.2323933024581355e-05, "loss": 0.4469, "step": 12423 }, { "epoch": 0.4426314195628552, "grad_norm": 1.6152818516527974, "learning_rate": 1.232281066501454e-05, "loss": 0.5774, "step": 12424 }, { "epoch": 0.44266704668934925, "grad_norm": 1.6117498367861394, "learning_rate": 1.2321688274517877e-05, "loss": 0.6417, "step": 12425 }, { "epoch": 0.4427026738158434, "grad_norm": 1.5416519154708446, "learning_rate": 1.2320565853106317e-05, "loss": 0.4339, "step": 12426 }, { "epoch": 0.4427383009423375, "grad_norm": 1.2298055863842638, "learning_rate": 1.2319443400794796e-05, "loss": 0.5281, "step": 12427 }, { "epoch": 0.4427739280688316, "grad_norm": 1.5597126486410968, "learning_rate": 1.2318320917598268e-05, "loss": 0.5352, "step": 12428 }, { "epoch": 0.4428095551953257, "grad_norm": 1.1379786516558081, "learning_rate": 1.2317198403531676e-05, "loss": 0.4911, "step": 12429 }, { "epoch": 0.44284518232181985, "grad_norm": 1.4233171630538066, "learning_rate": 1.2316075858609969e-05, "loss": 0.591, "step": 12430 }, { "epoch": 0.44288080944831393, "grad_norm": 1.1315625170636832, "learning_rate": 1.231495328284809e-05, "loss": 0.4203, "step": 12431 }, { "epoch": 0.44291643657480806, "grad_norm": 2.04683631541919, "learning_rate": 1.2313830676260989e-05, "loss": 0.4757, "step": 12432 }, { "epoch": 0.4429520637013022, "grad_norm": 1.6920751816535697, "learning_rate": 1.2312708038863618e-05, "loss": 0.5454, "step": 12433 }, { "epoch": 0.44298769082779627, "grad_norm": 1.148386767661205, "learning_rate": 1.2311585370670925e-05, "loss": 0.519, "step": 12434 }, { "epoch": 0.4430233179542904, "grad_norm": 1.3543292737289294, "learning_rate": 1.2310462671697853e-05, "loss": 0.6055, "step": 12435 }, { "epoch": 0.4430589450807845, "grad_norm": 1.6902670381770704, "learning_rate": 1.2309339941959358e-05, "loss": 0.5478, "step": 12436 }, { "epoch": 0.4430945722072786, "grad_norm": 1.0828753722754463, "learning_rate": 1.2308217181470385e-05, "loss": 0.4514, "step": 12437 }, { "epoch": 0.44313019933377273, "grad_norm": 1.327256113120868, "learning_rate": 1.230709439024589e-05, "loss": 0.4946, "step": 12438 }, { "epoch": 0.44316582646026687, "grad_norm": 1.5382063209857468, "learning_rate": 1.230597156830082e-05, "loss": 0.5567, "step": 12439 }, { "epoch": 0.44320145358676094, "grad_norm": 1.1617037275861406, "learning_rate": 1.2304848715650123e-05, "loss": 0.5015, "step": 12440 }, { "epoch": 0.44323708071325507, "grad_norm": 1.8778696704135347, "learning_rate": 1.2303725832308759e-05, "loss": 0.5509, "step": 12441 }, { "epoch": 0.4432727078397492, "grad_norm": 1.726753419918731, "learning_rate": 1.2302602918291674e-05, "loss": 0.5236, "step": 12442 }, { "epoch": 0.4433083349662433, "grad_norm": 1.2127728050085949, "learning_rate": 1.2301479973613822e-05, "loss": 0.4525, "step": 12443 }, { "epoch": 0.4433439620927374, "grad_norm": 1.0886189714107344, "learning_rate": 1.2300356998290155e-05, "loss": 0.4895, "step": 12444 }, { "epoch": 0.44337958921923154, "grad_norm": 1.532740501710109, "learning_rate": 1.2299233992335628e-05, "loss": 0.5457, "step": 12445 }, { "epoch": 0.4434152163457256, "grad_norm": 1.0601764316307418, "learning_rate": 1.229811095576519e-05, "loss": 0.4322, "step": 12446 }, { "epoch": 0.44345084347221975, "grad_norm": 1.0821403353059915, "learning_rate": 1.2296987888593802e-05, "loss": 0.4969, "step": 12447 }, { "epoch": 0.4434864705987139, "grad_norm": 1.5006085356728627, "learning_rate": 1.2295864790836411e-05, "loss": 0.5317, "step": 12448 }, { "epoch": 0.44352209772520795, "grad_norm": 1.7363472681079788, "learning_rate": 1.2294741662507978e-05, "loss": 0.5787, "step": 12449 }, { "epoch": 0.4435577248517021, "grad_norm": 1.1099828043486801, "learning_rate": 1.2293618503623459e-05, "loss": 0.4561, "step": 12450 }, { "epoch": 0.4435933519781962, "grad_norm": 1.5536863575403765, "learning_rate": 1.2292495314197801e-05, "loss": 0.5167, "step": 12451 }, { "epoch": 0.4436289791046903, "grad_norm": 1.3712159918952194, "learning_rate": 1.2291372094245969e-05, "loss": 0.504, "step": 12452 }, { "epoch": 0.4436646062311844, "grad_norm": 1.575998815552805, "learning_rate": 1.2290248843782915e-05, "loss": 0.5149, "step": 12453 }, { "epoch": 0.44370023335767855, "grad_norm": 1.3988270153844689, "learning_rate": 1.2289125562823594e-05, "loss": 0.5601, "step": 12454 }, { "epoch": 0.44373586048417263, "grad_norm": 1.2407665019142993, "learning_rate": 1.2288002251382969e-05, "loss": 0.4854, "step": 12455 }, { "epoch": 0.44377148761066676, "grad_norm": 1.0936780324256368, "learning_rate": 1.2286878909475997e-05, "loss": 0.3704, "step": 12456 }, { "epoch": 0.4438071147371609, "grad_norm": 1.94176852248373, "learning_rate": 1.2285755537117631e-05, "loss": 0.5893, "step": 12457 }, { "epoch": 0.44384274186365497, "grad_norm": 1.1372518537139447, "learning_rate": 1.2284632134322832e-05, "loss": 0.4229, "step": 12458 }, { "epoch": 0.4438783689901491, "grad_norm": 1.1423085356023222, "learning_rate": 1.2283508701106559e-05, "loss": 0.4931, "step": 12459 }, { "epoch": 0.44391399611664323, "grad_norm": 1.8121173230364434, "learning_rate": 1.2282385237483771e-05, "loss": 0.5487, "step": 12460 }, { "epoch": 0.4439496232431373, "grad_norm": 1.18563553166214, "learning_rate": 1.2281261743469433e-05, "loss": 0.4625, "step": 12461 }, { "epoch": 0.44398525036963143, "grad_norm": 1.1272482799380579, "learning_rate": 1.2280138219078494e-05, "loss": 0.4504, "step": 12462 }, { "epoch": 0.44402087749612557, "grad_norm": 1.2792650448469522, "learning_rate": 1.2279014664325925e-05, "loss": 0.6738, "step": 12463 }, { "epoch": 0.44405650462261964, "grad_norm": 1.2136936626376573, "learning_rate": 1.2277891079226683e-05, "loss": 0.4837, "step": 12464 }, { "epoch": 0.44409213174911377, "grad_norm": 1.325752558180582, "learning_rate": 1.2276767463795727e-05, "loss": 0.5877, "step": 12465 }, { "epoch": 0.4441277588756079, "grad_norm": 1.5588379542252797, "learning_rate": 1.2275643818048023e-05, "loss": 0.5065, "step": 12466 }, { "epoch": 0.444163386002102, "grad_norm": 1.2908417780130708, "learning_rate": 1.227452014199853e-05, "loss": 0.5296, "step": 12467 }, { "epoch": 0.4441990131285961, "grad_norm": 1.242270275885174, "learning_rate": 1.2273396435662212e-05, "loss": 0.4944, "step": 12468 }, { "epoch": 0.44423464025509024, "grad_norm": 3.181167014845137, "learning_rate": 1.227227269905403e-05, "loss": 0.5503, "step": 12469 }, { "epoch": 0.4442702673815843, "grad_norm": 6.29507340581783, "learning_rate": 1.227114893218895e-05, "loss": 0.5706, "step": 12470 }, { "epoch": 0.44430589450807845, "grad_norm": 3.751356526208944, "learning_rate": 1.227002513508193e-05, "loss": 0.5636, "step": 12471 }, { "epoch": 0.4443415216345726, "grad_norm": 2.112784372799977, "learning_rate": 1.2268901307747947e-05, "loss": 0.6101, "step": 12472 }, { "epoch": 0.44437714876106665, "grad_norm": 1.5379898855925755, "learning_rate": 1.226777745020195e-05, "loss": 0.595, "step": 12473 }, { "epoch": 0.4444127758875608, "grad_norm": 1.0716563616674046, "learning_rate": 1.2266653562458914e-05, "loss": 0.4526, "step": 12474 }, { "epoch": 0.4444484030140549, "grad_norm": 2.860112215058159, "learning_rate": 1.2265529644533804e-05, "loss": 0.5211, "step": 12475 }, { "epoch": 0.444484030140549, "grad_norm": 1.4860487754516951, "learning_rate": 1.226440569644158e-05, "loss": 0.5657, "step": 12476 }, { "epoch": 0.4445196572670431, "grad_norm": 1.3585041451665243, "learning_rate": 1.2263281718197213e-05, "loss": 0.474, "step": 12477 }, { "epoch": 0.44455528439353725, "grad_norm": 1.0683380791480288, "learning_rate": 1.2262157709815668e-05, "loss": 0.4894, "step": 12478 }, { "epoch": 0.44459091152003133, "grad_norm": 1.4779228157661704, "learning_rate": 1.226103367131191e-05, "loss": 0.5798, "step": 12479 }, { "epoch": 0.44462653864652546, "grad_norm": 1.8473044472307527, "learning_rate": 1.225990960270091e-05, "loss": 0.517, "step": 12480 }, { "epoch": 0.4446621657730196, "grad_norm": 0.9910519591437994, "learning_rate": 1.2258785503997635e-05, "loss": 0.4769, "step": 12481 }, { "epoch": 0.44469779289951367, "grad_norm": 1.1883181705188128, "learning_rate": 1.2257661375217052e-05, "loss": 0.4785, "step": 12482 }, { "epoch": 0.4447334200260078, "grad_norm": 1.625801519928114, "learning_rate": 1.2256537216374127e-05, "loss": 0.4776, "step": 12483 }, { "epoch": 0.44476904715250193, "grad_norm": 1.8962761716026595, "learning_rate": 1.225541302748383e-05, "loss": 0.6114, "step": 12484 }, { "epoch": 0.444804674278996, "grad_norm": 1.1169191828384524, "learning_rate": 1.2254288808561137e-05, "loss": 0.4216, "step": 12485 }, { "epoch": 0.44484030140549013, "grad_norm": 1.0889882468448793, "learning_rate": 1.2253164559621014e-05, "loss": 0.4659, "step": 12486 }, { "epoch": 0.44487592853198427, "grad_norm": 1.7392764556991238, "learning_rate": 1.2252040280678426e-05, "loss": 0.556, "step": 12487 }, { "epoch": 0.44491155565847834, "grad_norm": 1.617114709452934, "learning_rate": 1.225091597174835e-05, "loss": 0.5617, "step": 12488 }, { "epoch": 0.4449471827849725, "grad_norm": 1.524171093243453, "learning_rate": 1.2249791632845753e-05, "loss": 0.4817, "step": 12489 }, { "epoch": 0.4449828099114666, "grad_norm": 1.1520164862212667, "learning_rate": 1.2248667263985609e-05, "loss": 0.4838, "step": 12490 }, { "epoch": 0.4450184370379607, "grad_norm": 1.7926749700381746, "learning_rate": 1.2247542865182887e-05, "loss": 0.6053, "step": 12491 }, { "epoch": 0.4450540641644548, "grad_norm": 1.37348220891453, "learning_rate": 1.2246418436452563e-05, "loss": 0.5654, "step": 12492 }, { "epoch": 0.44508969129094894, "grad_norm": 1.0728962978042529, "learning_rate": 1.2245293977809607e-05, "loss": 0.4297, "step": 12493 }, { "epoch": 0.445125318417443, "grad_norm": 1.3094036570830578, "learning_rate": 1.2244169489268989e-05, "loss": 0.4858, "step": 12494 }, { "epoch": 0.44516094554393715, "grad_norm": 1.0886837457367726, "learning_rate": 1.2243044970845688e-05, "loss": 0.4448, "step": 12495 }, { "epoch": 0.4451965726704313, "grad_norm": 1.4633248548501694, "learning_rate": 1.2241920422554676e-05, "loss": 0.586, "step": 12496 }, { "epoch": 0.44523219979692535, "grad_norm": 1.1889431192991948, "learning_rate": 1.2240795844410926e-05, "loss": 0.5008, "step": 12497 }, { "epoch": 0.4452678269234195, "grad_norm": 1.0079110535958835, "learning_rate": 1.2239671236429413e-05, "loss": 0.4755, "step": 12498 }, { "epoch": 0.4453034540499136, "grad_norm": 1.2250534962550772, "learning_rate": 1.2238546598625112e-05, "loss": 0.4846, "step": 12499 }, { "epoch": 0.4453390811764077, "grad_norm": 1.8515512472751139, "learning_rate": 1.2237421931013001e-05, "loss": 0.5413, "step": 12500 }, { "epoch": 0.4453747083029018, "grad_norm": 1.3807190587019091, "learning_rate": 1.223629723360805e-05, "loss": 0.4158, "step": 12501 }, { "epoch": 0.44541033542939595, "grad_norm": 1.853872153099992, "learning_rate": 1.2235172506425239e-05, "loss": 0.5497, "step": 12502 }, { "epoch": 0.44544596255589003, "grad_norm": 1.0720924630424398, "learning_rate": 1.2234047749479543e-05, "loss": 0.4458, "step": 12503 }, { "epoch": 0.44548158968238416, "grad_norm": 1.2355657031111997, "learning_rate": 1.2232922962785942e-05, "loss": 0.4921, "step": 12504 }, { "epoch": 0.4455172168088783, "grad_norm": 1.5746862225211953, "learning_rate": 1.223179814635941e-05, "loss": 0.58, "step": 12505 }, { "epoch": 0.44555284393537237, "grad_norm": 1.143793800581297, "learning_rate": 1.2230673300214925e-05, "loss": 0.5017, "step": 12506 }, { "epoch": 0.4455884710618665, "grad_norm": 1.4893383391623392, "learning_rate": 1.2229548424367466e-05, "loss": 0.4964, "step": 12507 }, { "epoch": 0.44562409818836063, "grad_norm": 1.1559104680462489, "learning_rate": 1.2228423518832011e-05, "loss": 0.4416, "step": 12508 }, { "epoch": 0.4456597253148547, "grad_norm": 1.384099346698366, "learning_rate": 1.2227298583623541e-05, "loss": 0.4164, "step": 12509 }, { "epoch": 0.44569535244134884, "grad_norm": 1.793723996002002, "learning_rate": 1.2226173618757032e-05, "loss": 0.6518, "step": 12510 }, { "epoch": 0.44573097956784297, "grad_norm": 1.4583029479219132, "learning_rate": 1.2225048624247467e-05, "loss": 0.5736, "step": 12511 }, { "epoch": 0.44576660669433704, "grad_norm": 1.2622019638452395, "learning_rate": 1.2223923600109822e-05, "loss": 0.4469, "step": 12512 }, { "epoch": 0.4458022338208312, "grad_norm": 1.1757655755650953, "learning_rate": 1.2222798546359082e-05, "loss": 0.4939, "step": 12513 }, { "epoch": 0.4458378609473253, "grad_norm": 1.2196061458666323, "learning_rate": 1.2221673463010225e-05, "loss": 0.4787, "step": 12514 }, { "epoch": 0.4458734880738194, "grad_norm": 1.9544283452817208, "learning_rate": 1.2220548350078232e-05, "loss": 0.5471, "step": 12515 }, { "epoch": 0.4459091152003135, "grad_norm": 1.343380596310957, "learning_rate": 1.2219423207578087e-05, "loss": 0.4535, "step": 12516 }, { "epoch": 0.44594474232680764, "grad_norm": 1.231066588228561, "learning_rate": 1.2218298035524768e-05, "loss": 0.4715, "step": 12517 }, { "epoch": 0.4459803694533017, "grad_norm": 1.745949641918138, "learning_rate": 1.2217172833933261e-05, "loss": 0.4765, "step": 12518 }, { "epoch": 0.44601599657979585, "grad_norm": 1.3895504523917879, "learning_rate": 1.221604760281855e-05, "loss": 0.5214, "step": 12519 }, { "epoch": 0.44605162370629, "grad_norm": 2.9194305595907353, "learning_rate": 1.2214922342195616e-05, "loss": 0.6562, "step": 12520 }, { "epoch": 0.44608725083278405, "grad_norm": 2.03765286670992, "learning_rate": 1.2213797052079441e-05, "loss": 0.5826, "step": 12521 }, { "epoch": 0.4461228779592782, "grad_norm": 1.1740584060765458, "learning_rate": 1.2212671732485012e-05, "loss": 0.4921, "step": 12522 }, { "epoch": 0.4461585050857723, "grad_norm": 1.7166105139228973, "learning_rate": 1.2211546383427315e-05, "loss": 0.5454, "step": 12523 }, { "epoch": 0.4461941322122664, "grad_norm": 1.997662194136553, "learning_rate": 1.2210421004921329e-05, "loss": 0.5013, "step": 12524 }, { "epoch": 0.4462297593387605, "grad_norm": 1.3389113141921851, "learning_rate": 1.2209295596982043e-05, "loss": 0.4225, "step": 12525 }, { "epoch": 0.44626538646525465, "grad_norm": 1.7332041094211392, "learning_rate": 1.2208170159624442e-05, "loss": 0.627, "step": 12526 }, { "epoch": 0.44630101359174873, "grad_norm": 1.6135587490992145, "learning_rate": 1.2207044692863512e-05, "loss": 0.6263, "step": 12527 }, { "epoch": 0.44633664071824286, "grad_norm": 1.1320480046390569, "learning_rate": 1.2205919196714237e-05, "loss": 0.4428, "step": 12528 }, { "epoch": 0.446372267844737, "grad_norm": 1.4719150828943197, "learning_rate": 1.2204793671191607e-05, "loss": 0.5239, "step": 12529 }, { "epoch": 0.4464078949712311, "grad_norm": 2.6422615030454866, "learning_rate": 1.220366811631061e-05, "loss": 0.539, "step": 12530 }, { "epoch": 0.4464435220977252, "grad_norm": 1.166898596693075, "learning_rate": 1.2202542532086232e-05, "loss": 0.4556, "step": 12531 }, { "epoch": 0.44647914922421933, "grad_norm": 2.5781922548757166, "learning_rate": 1.2201416918533462e-05, "loss": 0.641, "step": 12532 }, { "epoch": 0.44651477635071346, "grad_norm": 1.101016066361871, "learning_rate": 1.2200291275667282e-05, "loss": 0.4487, "step": 12533 }, { "epoch": 0.44655040347720754, "grad_norm": 1.1514242292441508, "learning_rate": 1.219916560350269e-05, "loss": 0.4889, "step": 12534 }, { "epoch": 0.44658603060370167, "grad_norm": 1.282328501490446, "learning_rate": 1.219803990205467e-05, "loss": 0.5651, "step": 12535 }, { "epoch": 0.4466216577301958, "grad_norm": 1.6616137980328276, "learning_rate": 1.2196914171338209e-05, "loss": 0.5838, "step": 12536 }, { "epoch": 0.4466572848566899, "grad_norm": 1.0969926034696633, "learning_rate": 1.2195788411368305e-05, "loss": 0.4308, "step": 12537 }, { "epoch": 0.446692911983184, "grad_norm": 1.097262584020471, "learning_rate": 1.2194662622159943e-05, "loss": 0.4941, "step": 12538 }, { "epoch": 0.44672853910967814, "grad_norm": 1.2018515956238014, "learning_rate": 1.2193536803728113e-05, "loss": 0.4585, "step": 12539 }, { "epoch": 0.4467641662361722, "grad_norm": 1.5163606149991826, "learning_rate": 1.2192410956087803e-05, "loss": 0.5314, "step": 12540 }, { "epoch": 0.44679979336266634, "grad_norm": 1.4204347491759133, "learning_rate": 1.2191285079254014e-05, "loss": 0.5446, "step": 12541 }, { "epoch": 0.4468354204891605, "grad_norm": 1.7669500874735318, "learning_rate": 1.2190159173241731e-05, "loss": 0.6431, "step": 12542 }, { "epoch": 0.44687104761565455, "grad_norm": 1.026738400128167, "learning_rate": 1.218903323806595e-05, "loss": 0.4567, "step": 12543 }, { "epoch": 0.4469066747421487, "grad_norm": 1.539912076763673, "learning_rate": 1.2187907273741656e-05, "loss": 0.6232, "step": 12544 }, { "epoch": 0.4469423018686428, "grad_norm": 1.256718387096351, "learning_rate": 1.2186781280283854e-05, "loss": 0.4515, "step": 12545 }, { "epoch": 0.4469779289951369, "grad_norm": 2.0919794955611892, "learning_rate": 1.2185655257707527e-05, "loss": 0.5667, "step": 12546 }, { "epoch": 0.447013556121631, "grad_norm": 1.7007828070166027, "learning_rate": 1.2184529206027674e-05, "loss": 0.594, "step": 12547 }, { "epoch": 0.44704918324812515, "grad_norm": 2.0391543432064756, "learning_rate": 1.2183403125259287e-05, "loss": 0.5031, "step": 12548 }, { "epoch": 0.4470848103746192, "grad_norm": 1.0150794395067322, "learning_rate": 1.218227701541736e-05, "loss": 0.5441, "step": 12549 }, { "epoch": 0.44712043750111335, "grad_norm": 1.1445856667010665, "learning_rate": 1.2181150876516891e-05, "loss": 0.4513, "step": 12550 }, { "epoch": 0.4471560646276075, "grad_norm": 1.5008467833805992, "learning_rate": 1.2180024708572873e-05, "loss": 0.5698, "step": 12551 }, { "epoch": 0.44719169175410156, "grad_norm": 2.6619293361503424, "learning_rate": 1.2178898511600301e-05, "loss": 0.6523, "step": 12552 }, { "epoch": 0.4472273188805957, "grad_norm": 1.3745461237751135, "learning_rate": 1.2177772285614176e-05, "loss": 0.5249, "step": 12553 }, { "epoch": 0.4472629460070898, "grad_norm": 1.1290527794317073, "learning_rate": 1.217664603062949e-05, "loss": 0.4508, "step": 12554 }, { "epoch": 0.4472985731335839, "grad_norm": 0.999370177463024, "learning_rate": 1.2175519746661235e-05, "loss": 0.435, "step": 12555 }, { "epoch": 0.44733420026007803, "grad_norm": 1.0783684570871404, "learning_rate": 1.2174393433724419e-05, "loss": 0.4656, "step": 12556 }, { "epoch": 0.44736982738657216, "grad_norm": 1.485848649999847, "learning_rate": 1.2173267091834035e-05, "loss": 0.5274, "step": 12557 }, { "epoch": 0.44740545451306624, "grad_norm": 1.467758642730156, "learning_rate": 1.217214072100508e-05, "loss": 0.5849, "step": 12558 }, { "epoch": 0.44744108163956037, "grad_norm": 1.1213256258416415, "learning_rate": 1.2171014321252552e-05, "loss": 0.4966, "step": 12559 }, { "epoch": 0.4474767087660545, "grad_norm": 1.2024659054585076, "learning_rate": 1.2169887892591452e-05, "loss": 0.4843, "step": 12560 }, { "epoch": 0.4475123358925486, "grad_norm": 1.1043585318609312, "learning_rate": 1.2168761435036778e-05, "loss": 0.4632, "step": 12561 }, { "epoch": 0.4475479630190427, "grad_norm": 1.2018031380811767, "learning_rate": 1.2167634948603529e-05, "loss": 0.5079, "step": 12562 }, { "epoch": 0.44758359014553684, "grad_norm": 1.0421647091144426, "learning_rate": 1.2166508433306704e-05, "loss": 0.4429, "step": 12563 }, { "epoch": 0.4476192172720309, "grad_norm": 1.1490126677288204, "learning_rate": 1.2165381889161305e-05, "loss": 0.4597, "step": 12564 }, { "epoch": 0.44765484439852504, "grad_norm": 1.616099358520216, "learning_rate": 1.2164255316182336e-05, "loss": 0.5267, "step": 12565 }, { "epoch": 0.4476904715250192, "grad_norm": 1.1745193094035296, "learning_rate": 1.2163128714384791e-05, "loss": 0.4233, "step": 12566 }, { "epoch": 0.44772609865151325, "grad_norm": 1.1704664305774977, "learning_rate": 1.2162002083783676e-05, "loss": 0.5132, "step": 12567 }, { "epoch": 0.4477617257780074, "grad_norm": 1.2685466443586535, "learning_rate": 1.2160875424393997e-05, "loss": 0.457, "step": 12568 }, { "epoch": 0.4477973529045015, "grad_norm": 1.1077107068087682, "learning_rate": 1.2159748736230747e-05, "loss": 0.4547, "step": 12569 }, { "epoch": 0.4478329800309956, "grad_norm": 1.4306360060522074, "learning_rate": 1.2158622019308932e-05, "loss": 0.5778, "step": 12570 }, { "epoch": 0.4478686071574897, "grad_norm": 2.688270253134807, "learning_rate": 1.2157495273643558e-05, "loss": 0.5561, "step": 12571 }, { "epoch": 0.44790423428398385, "grad_norm": 1.2656905853694258, "learning_rate": 1.2156368499249629e-05, "loss": 0.4722, "step": 12572 }, { "epoch": 0.4479398614104779, "grad_norm": 1.233199791063724, "learning_rate": 1.2155241696142141e-05, "loss": 0.5072, "step": 12573 }, { "epoch": 0.44797548853697206, "grad_norm": 1.1951404080215633, "learning_rate": 1.2154114864336107e-05, "loss": 0.4422, "step": 12574 }, { "epoch": 0.4480111156634662, "grad_norm": 1.1730809093759231, "learning_rate": 1.2152988003846525e-05, "loss": 0.436, "step": 12575 }, { "epoch": 0.44804674278996026, "grad_norm": 1.146530196724771, "learning_rate": 1.2151861114688406e-05, "loss": 0.4783, "step": 12576 }, { "epoch": 0.4480823699164544, "grad_norm": 1.2366540314648464, "learning_rate": 1.2150734196876747e-05, "loss": 0.5039, "step": 12577 }, { "epoch": 0.4481179970429485, "grad_norm": 1.1898298329536237, "learning_rate": 1.2149607250426565e-05, "loss": 0.4874, "step": 12578 }, { "epoch": 0.4481536241694426, "grad_norm": 1.171757426919591, "learning_rate": 1.2148480275352857e-05, "loss": 0.4345, "step": 12579 }, { "epoch": 0.44818925129593673, "grad_norm": 1.9264460708462428, "learning_rate": 1.2147353271670634e-05, "loss": 0.533, "step": 12580 }, { "epoch": 0.44822487842243086, "grad_norm": 1.284755828299355, "learning_rate": 1.2146226239394898e-05, "loss": 0.5239, "step": 12581 }, { "epoch": 0.44826050554892494, "grad_norm": 1.6438456132398664, "learning_rate": 1.2145099178540665e-05, "loss": 0.5009, "step": 12582 }, { "epoch": 0.44829613267541907, "grad_norm": 1.2716522950732307, "learning_rate": 1.2143972089122932e-05, "loss": 0.5193, "step": 12583 }, { "epoch": 0.4483317598019132, "grad_norm": 1.2387146714914696, "learning_rate": 1.2142844971156714e-05, "loss": 0.4961, "step": 12584 }, { "epoch": 0.4483673869284073, "grad_norm": 1.3497772531617205, "learning_rate": 1.2141717824657018e-05, "loss": 0.4904, "step": 12585 }, { "epoch": 0.4484030140549014, "grad_norm": 1.1065821549127421, "learning_rate": 1.214059064963885e-05, "loss": 0.4605, "step": 12586 }, { "epoch": 0.44843864118139554, "grad_norm": 1.2980773718088747, "learning_rate": 1.2139463446117224e-05, "loss": 0.4235, "step": 12587 }, { "epoch": 0.4484742683078896, "grad_norm": 1.4264691111702497, "learning_rate": 1.2138336214107148e-05, "loss": 0.5255, "step": 12588 }, { "epoch": 0.44850989543438374, "grad_norm": 1.3001998175977536, "learning_rate": 1.2137208953623626e-05, "loss": 0.463, "step": 12589 }, { "epoch": 0.4485455225608779, "grad_norm": 2.3220589535439116, "learning_rate": 1.2136081664681675e-05, "loss": 0.5928, "step": 12590 }, { "epoch": 0.44858114968737195, "grad_norm": 1.0929234932358285, "learning_rate": 1.2134954347296306e-05, "loss": 0.4483, "step": 12591 }, { "epoch": 0.4486167768138661, "grad_norm": 1.3580408392920182, "learning_rate": 1.2133827001482524e-05, "loss": 0.4952, "step": 12592 }, { "epoch": 0.4486524039403602, "grad_norm": 1.4747068189477077, "learning_rate": 1.2132699627255347e-05, "loss": 0.5646, "step": 12593 }, { "epoch": 0.4486880310668543, "grad_norm": 1.9838676481969275, "learning_rate": 1.2131572224629783e-05, "loss": 0.5744, "step": 12594 }, { "epoch": 0.4487236581933484, "grad_norm": 1.7549298416142574, "learning_rate": 1.2130444793620844e-05, "loss": 0.6141, "step": 12595 }, { "epoch": 0.44875928531984255, "grad_norm": 1.5254174414402977, "learning_rate": 1.2129317334243542e-05, "loss": 0.4673, "step": 12596 }, { "epoch": 0.4487949124463366, "grad_norm": 1.2570959122463523, "learning_rate": 1.2128189846512895e-05, "loss": 0.4653, "step": 12597 }, { "epoch": 0.44883053957283076, "grad_norm": 1.2414919754001852, "learning_rate": 1.2127062330443908e-05, "loss": 0.5316, "step": 12598 }, { "epoch": 0.4488661666993249, "grad_norm": 1.122926073920141, "learning_rate": 1.2125934786051605e-05, "loss": 0.4834, "step": 12599 }, { "epoch": 0.44890179382581896, "grad_norm": 1.3142698266991826, "learning_rate": 1.2124807213350991e-05, "loss": 0.569, "step": 12600 }, { "epoch": 0.4489374209523131, "grad_norm": 1.1083450109265574, "learning_rate": 1.2123679612357085e-05, "loss": 0.4725, "step": 12601 }, { "epoch": 0.4489730480788072, "grad_norm": 1.5035163447749964, "learning_rate": 1.21225519830849e-05, "loss": 0.609, "step": 12602 }, { "epoch": 0.4490086752053013, "grad_norm": 1.0686670370448637, "learning_rate": 1.2121424325549449e-05, "loss": 0.3839, "step": 12603 }, { "epoch": 0.44904430233179543, "grad_norm": 1.6615801938233123, "learning_rate": 1.2120296639765754e-05, "loss": 0.6183, "step": 12604 }, { "epoch": 0.44907992945828956, "grad_norm": 1.4553057096318118, "learning_rate": 1.2119168925748827e-05, "loss": 0.4929, "step": 12605 }, { "epoch": 0.44911555658478364, "grad_norm": 1.1321363651627483, "learning_rate": 1.2118041183513686e-05, "loss": 0.4664, "step": 12606 }, { "epoch": 0.44915118371127777, "grad_norm": 1.113371555100543, "learning_rate": 1.2116913413075342e-05, "loss": 0.5135, "step": 12607 }, { "epoch": 0.4491868108377719, "grad_norm": 1.797638240882846, "learning_rate": 1.211578561444882e-05, "loss": 0.5969, "step": 12608 }, { "epoch": 0.449222437964266, "grad_norm": 1.768973431419437, "learning_rate": 1.2114657787649131e-05, "loss": 0.5575, "step": 12609 }, { "epoch": 0.4492580650907601, "grad_norm": 1.2297900509310895, "learning_rate": 1.2113529932691297e-05, "loss": 0.4842, "step": 12610 }, { "epoch": 0.44929369221725424, "grad_norm": 1.04396357379917, "learning_rate": 1.2112402049590333e-05, "loss": 0.4424, "step": 12611 }, { "epoch": 0.4493293193437483, "grad_norm": 1.4880355719485514, "learning_rate": 1.2111274138361262e-05, "loss": 0.4799, "step": 12612 }, { "epoch": 0.44936494647024244, "grad_norm": 1.150410627291645, "learning_rate": 1.2110146199019099e-05, "loss": 0.4755, "step": 12613 }, { "epoch": 0.4494005735967366, "grad_norm": 2.4730388055071164, "learning_rate": 1.2109018231578863e-05, "loss": 0.5927, "step": 12614 }, { "epoch": 0.44943620072323065, "grad_norm": 1.2013849107210164, "learning_rate": 1.2107890236055578e-05, "loss": 0.4875, "step": 12615 }, { "epoch": 0.4494718278497248, "grad_norm": 1.0611547378819886, "learning_rate": 1.2106762212464257e-05, "loss": 0.4532, "step": 12616 }, { "epoch": 0.4495074549762189, "grad_norm": 1.1368913832506453, "learning_rate": 1.210563416081993e-05, "loss": 0.499, "step": 12617 }, { "epoch": 0.449543082102713, "grad_norm": 1.2075877097181738, "learning_rate": 1.2104506081137608e-05, "loss": 0.5392, "step": 12618 }, { "epoch": 0.4495787092292071, "grad_norm": 1.4322148879702459, "learning_rate": 1.2103377973432317e-05, "loss": 0.5506, "step": 12619 }, { "epoch": 0.44961433635570125, "grad_norm": 1.115964628230307, "learning_rate": 1.2102249837719083e-05, "loss": 0.4253, "step": 12620 }, { "epoch": 0.4496499634821953, "grad_norm": 1.7088969078264253, "learning_rate": 1.2101121674012916e-05, "loss": 0.5895, "step": 12621 }, { "epoch": 0.44968559060868946, "grad_norm": 1.2813406566991312, "learning_rate": 1.2099993482328846e-05, "loss": 0.4952, "step": 12622 }, { "epoch": 0.4497212177351836, "grad_norm": 1.1970874091093253, "learning_rate": 1.20988652626819e-05, "loss": 0.4527, "step": 12623 }, { "epoch": 0.44975684486167766, "grad_norm": 1.150444961816928, "learning_rate": 1.2097737015087094e-05, "loss": 0.4191, "step": 12624 }, { "epoch": 0.4497924719881718, "grad_norm": 1.5194943459630599, "learning_rate": 1.2096608739559454e-05, "loss": 0.5542, "step": 12625 }, { "epoch": 0.4498280991146659, "grad_norm": 1.1108545273422288, "learning_rate": 1.2095480436114e-05, "loss": 0.4471, "step": 12626 }, { "epoch": 0.44986372624116, "grad_norm": 1.2871290301386056, "learning_rate": 1.2094352104765762e-05, "loss": 0.5204, "step": 12627 }, { "epoch": 0.44989935336765413, "grad_norm": 1.5272977769738847, "learning_rate": 1.2093223745529763e-05, "loss": 0.5534, "step": 12628 }, { "epoch": 0.44993498049414826, "grad_norm": 1.8985325544405536, "learning_rate": 1.2092095358421023e-05, "loss": 0.5762, "step": 12629 }, { "epoch": 0.44997060762064234, "grad_norm": 1.0791813545456375, "learning_rate": 1.2090966943454574e-05, "loss": 0.5065, "step": 12630 }, { "epoch": 0.45000623474713647, "grad_norm": 1.0706395281295245, "learning_rate": 1.2089838500645438e-05, "loss": 0.4347, "step": 12631 }, { "epoch": 0.4500418618736306, "grad_norm": 1.438487530445111, "learning_rate": 1.2088710030008642e-05, "loss": 0.5333, "step": 12632 }, { "epoch": 0.4500774890001247, "grad_norm": 1.5933110924863958, "learning_rate": 1.2087581531559208e-05, "loss": 0.5843, "step": 12633 }, { "epoch": 0.4501131161266188, "grad_norm": 1.2079028904070719, "learning_rate": 1.208645300531217e-05, "loss": 0.4803, "step": 12634 }, { "epoch": 0.45014874325311294, "grad_norm": 1.5990173011562994, "learning_rate": 1.2085324451282554e-05, "loss": 0.5754, "step": 12635 }, { "epoch": 0.450184370379607, "grad_norm": 1.0738073295109192, "learning_rate": 1.2084195869485385e-05, "loss": 0.4788, "step": 12636 }, { "epoch": 0.45021999750610114, "grad_norm": 1.3698806106956372, "learning_rate": 1.2083067259935688e-05, "loss": 0.5427, "step": 12637 }, { "epoch": 0.4502556246325953, "grad_norm": 1.2076224217762543, "learning_rate": 1.2081938622648494e-05, "loss": 0.4604, "step": 12638 }, { "epoch": 0.45029125175908935, "grad_norm": 1.149230346679637, "learning_rate": 1.2080809957638837e-05, "loss": 0.472, "step": 12639 }, { "epoch": 0.4503268788855835, "grad_norm": 1.3712174485809516, "learning_rate": 1.2079681264921737e-05, "loss": 0.5525, "step": 12640 }, { "epoch": 0.4503625060120776, "grad_norm": 1.1339846186479836, "learning_rate": 1.2078552544512227e-05, "loss": 0.45, "step": 12641 }, { "epoch": 0.4503981331385717, "grad_norm": 1.8661937802195894, "learning_rate": 1.2077423796425341e-05, "loss": 0.5887, "step": 12642 }, { "epoch": 0.4504337602650658, "grad_norm": 1.4251748861261202, "learning_rate": 1.2076295020676103e-05, "loss": 0.5493, "step": 12643 }, { "epoch": 0.45046938739155995, "grad_norm": 1.3727008928397577, "learning_rate": 1.2075166217279544e-05, "loss": 0.5177, "step": 12644 }, { "epoch": 0.450505014518054, "grad_norm": 1.629098704159011, "learning_rate": 1.2074037386250696e-05, "loss": 0.6003, "step": 12645 }, { "epoch": 0.45054064164454816, "grad_norm": 1.082307758150341, "learning_rate": 1.2072908527604595e-05, "loss": 0.4657, "step": 12646 }, { "epoch": 0.4505762687710423, "grad_norm": 1.460614763290681, "learning_rate": 1.2071779641356264e-05, "loss": 0.6051, "step": 12647 }, { "epoch": 0.45061189589753636, "grad_norm": 1.2188898089807605, "learning_rate": 1.2070650727520738e-05, "loss": 0.5002, "step": 12648 }, { "epoch": 0.4506475230240305, "grad_norm": 1.4449465477816026, "learning_rate": 1.2069521786113055e-05, "loss": 0.5507, "step": 12649 }, { "epoch": 0.4506831501505246, "grad_norm": 1.5462310066711478, "learning_rate": 1.206839281714824e-05, "loss": 0.6297, "step": 12650 }, { "epoch": 0.4507187772770187, "grad_norm": 1.1776943975439098, "learning_rate": 1.2067263820641331e-05, "loss": 0.4936, "step": 12651 }, { "epoch": 0.45075440440351283, "grad_norm": 1.2692072933344325, "learning_rate": 1.2066134796607356e-05, "loss": 0.4517, "step": 12652 }, { "epoch": 0.45079003153000696, "grad_norm": 1.2203799096020762, "learning_rate": 1.2065005745061354e-05, "loss": 0.4945, "step": 12653 }, { "epoch": 0.45082565865650104, "grad_norm": 1.162982348513631, "learning_rate": 1.2063876666018357e-05, "loss": 0.4525, "step": 12654 }, { "epoch": 0.45086128578299517, "grad_norm": 1.2931454679821233, "learning_rate": 1.20627475594934e-05, "loss": 0.5097, "step": 12655 }, { "epoch": 0.4508969129094893, "grad_norm": 1.4410735445212899, "learning_rate": 1.2061618425501513e-05, "loss": 0.4991, "step": 12656 }, { "epoch": 0.4509325400359834, "grad_norm": 1.3179122324379107, "learning_rate": 1.2060489264057741e-05, "loss": 0.5475, "step": 12657 }, { "epoch": 0.4509681671624775, "grad_norm": 1.15290534474093, "learning_rate": 1.2059360075177114e-05, "loss": 0.4545, "step": 12658 }, { "epoch": 0.45100379428897164, "grad_norm": 1.528435323280311, "learning_rate": 1.2058230858874667e-05, "loss": 0.4329, "step": 12659 }, { "epoch": 0.4510394214154657, "grad_norm": 2.74561591049107, "learning_rate": 1.2057101615165436e-05, "loss": 0.6508, "step": 12660 }, { "epoch": 0.45107504854195984, "grad_norm": 1.8623598135299457, "learning_rate": 1.205597234406446e-05, "loss": 0.5553, "step": 12661 }, { "epoch": 0.451110675668454, "grad_norm": 1.1877333322205432, "learning_rate": 1.2054843045586779e-05, "loss": 0.4142, "step": 12662 }, { "epoch": 0.45114630279494805, "grad_norm": 1.1909593981250848, "learning_rate": 1.2053713719747423e-05, "loss": 0.5103, "step": 12663 }, { "epoch": 0.4511819299214422, "grad_norm": 1.0735909845533094, "learning_rate": 1.2052584366561433e-05, "loss": 0.4675, "step": 12664 }, { "epoch": 0.4512175570479363, "grad_norm": 1.1799305854581597, "learning_rate": 1.205145498604385e-05, "loss": 0.5086, "step": 12665 }, { "epoch": 0.4512531841744304, "grad_norm": 1.2238093079870838, "learning_rate": 1.2050325578209707e-05, "loss": 0.4756, "step": 12666 }, { "epoch": 0.4512888113009245, "grad_norm": 1.654662392478758, "learning_rate": 1.2049196143074045e-05, "loss": 0.5046, "step": 12667 }, { "epoch": 0.45132443842741865, "grad_norm": 1.1902572486297365, "learning_rate": 1.2048066680651908e-05, "loss": 0.4896, "step": 12668 }, { "epoch": 0.4513600655539127, "grad_norm": 1.1155730306299083, "learning_rate": 1.2046937190958329e-05, "loss": 0.463, "step": 12669 }, { "epoch": 0.45139569268040686, "grad_norm": 1.508851773835289, "learning_rate": 1.2045807674008355e-05, "loss": 0.5409, "step": 12670 }, { "epoch": 0.451431319806901, "grad_norm": 1.073424043666469, "learning_rate": 1.2044678129817016e-05, "loss": 0.4546, "step": 12671 }, { "epoch": 0.45146694693339506, "grad_norm": 1.521924255327492, "learning_rate": 1.2043548558399361e-05, "loss": 0.5027, "step": 12672 }, { "epoch": 0.4515025740598892, "grad_norm": 1.03391646816434, "learning_rate": 1.204241895977043e-05, "loss": 0.4612, "step": 12673 }, { "epoch": 0.4515382011863833, "grad_norm": 1.0286481409264419, "learning_rate": 1.2041289333945258e-05, "loss": 0.4337, "step": 12674 }, { "epoch": 0.4515738283128774, "grad_norm": 1.5106393265850973, "learning_rate": 1.2040159680938897e-05, "loss": 0.5789, "step": 12675 }, { "epoch": 0.45160945543937153, "grad_norm": 1.66294182414031, "learning_rate": 1.2039030000766383e-05, "loss": 0.5548, "step": 12676 }, { "epoch": 0.45164508256586566, "grad_norm": 1.7685624257020218, "learning_rate": 1.2037900293442759e-05, "loss": 0.5159, "step": 12677 }, { "epoch": 0.45168070969235974, "grad_norm": 1.7706280453367071, "learning_rate": 1.2036770558983067e-05, "loss": 0.6132, "step": 12678 }, { "epoch": 0.45171633681885387, "grad_norm": 1.104239099021547, "learning_rate": 1.2035640797402351e-05, "loss": 0.4734, "step": 12679 }, { "epoch": 0.451751963945348, "grad_norm": 1.6839877389089442, "learning_rate": 1.2034511008715655e-05, "loss": 0.5852, "step": 12680 }, { "epoch": 0.4517875910718421, "grad_norm": 1.792998685267753, "learning_rate": 1.2033381192938026e-05, "loss": 0.5504, "step": 12681 }, { "epoch": 0.4518232181983362, "grad_norm": 1.3702702496822565, "learning_rate": 1.20322513500845e-05, "loss": 0.639, "step": 12682 }, { "epoch": 0.45185884532483034, "grad_norm": 1.0562778250949136, "learning_rate": 1.2031121480170132e-05, "loss": 0.4996, "step": 12683 }, { "epoch": 0.4518944724513244, "grad_norm": 1.1206437431777738, "learning_rate": 1.2029991583209959e-05, "loss": 0.4095, "step": 12684 }, { "epoch": 0.45193009957781854, "grad_norm": 1.1318781112086003, "learning_rate": 1.2028861659219028e-05, "loss": 0.4883, "step": 12685 }, { "epoch": 0.4519657267043127, "grad_norm": 1.3565469732817115, "learning_rate": 1.2027731708212389e-05, "loss": 0.5704, "step": 12686 }, { "epoch": 0.45200135383080675, "grad_norm": 1.0622616626131542, "learning_rate": 1.2026601730205085e-05, "loss": 0.4507, "step": 12687 }, { "epoch": 0.4520369809573009, "grad_norm": 1.3571074942964543, "learning_rate": 1.2025471725212162e-05, "loss": 0.4827, "step": 12688 }, { "epoch": 0.452072608083795, "grad_norm": 1.4153216655227276, "learning_rate": 1.2024341693248664e-05, "loss": 0.5117, "step": 12689 }, { "epoch": 0.4521082352102891, "grad_norm": 1.7834413779068572, "learning_rate": 1.2023211634329644e-05, "loss": 0.5543, "step": 12690 }, { "epoch": 0.4521438623367832, "grad_norm": 1.0733618018742568, "learning_rate": 1.2022081548470147e-05, "loss": 0.3947, "step": 12691 }, { "epoch": 0.45217948946327735, "grad_norm": 1.0697310289644448, "learning_rate": 1.202095143568522e-05, "loss": 0.4546, "step": 12692 }, { "epoch": 0.4522151165897714, "grad_norm": 1.250187313247261, "learning_rate": 1.2019821295989913e-05, "loss": 0.4328, "step": 12693 }, { "epoch": 0.45225074371626556, "grad_norm": 1.5762164133109893, "learning_rate": 1.2018691129399273e-05, "loss": 0.5834, "step": 12694 }, { "epoch": 0.4522863708427597, "grad_norm": 1.470489707290999, "learning_rate": 1.201756093592835e-05, "loss": 0.5722, "step": 12695 }, { "epoch": 0.45232199796925376, "grad_norm": 0.9862813794596178, "learning_rate": 1.2016430715592193e-05, "loss": 0.3715, "step": 12696 }, { "epoch": 0.4523576250957479, "grad_norm": 1.2621505785846123, "learning_rate": 1.2015300468405852e-05, "loss": 0.4684, "step": 12697 }, { "epoch": 0.452393252222242, "grad_norm": 1.086280446501127, "learning_rate": 1.2014170194384375e-05, "loss": 0.5032, "step": 12698 }, { "epoch": 0.4524288793487361, "grad_norm": 1.0656622520591688, "learning_rate": 1.2013039893542817e-05, "loss": 0.4608, "step": 12699 }, { "epoch": 0.45246450647523023, "grad_norm": 1.1692833850653257, "learning_rate": 1.2011909565896223e-05, "loss": 0.4214, "step": 12700 }, { "epoch": 0.45250013360172436, "grad_norm": 1.4016836515034183, "learning_rate": 1.2010779211459649e-05, "loss": 0.4828, "step": 12701 }, { "epoch": 0.45253576072821844, "grad_norm": 1.0470775242813894, "learning_rate": 1.2009648830248141e-05, "loss": 0.5028, "step": 12702 }, { "epoch": 0.45257138785471257, "grad_norm": 1.0249468995086406, "learning_rate": 1.2008518422276757e-05, "loss": 0.4443, "step": 12703 }, { "epoch": 0.4526070149812067, "grad_norm": 1.060506975812908, "learning_rate": 1.2007387987560545e-05, "loss": 0.4489, "step": 12704 }, { "epoch": 0.4526426421077008, "grad_norm": 1.1678182215706243, "learning_rate": 1.200625752611456e-05, "loss": 0.4482, "step": 12705 }, { "epoch": 0.4526782692341949, "grad_norm": 1.1297543454342958, "learning_rate": 1.2005127037953854e-05, "loss": 0.4867, "step": 12706 }, { "epoch": 0.45271389636068904, "grad_norm": 1.5340471337848418, "learning_rate": 1.2003996523093482e-05, "loss": 0.6035, "step": 12707 }, { "epoch": 0.4527495234871831, "grad_norm": 1.122562117249674, "learning_rate": 1.200286598154849e-05, "loss": 0.4708, "step": 12708 }, { "epoch": 0.45278515061367725, "grad_norm": 1.4388383877538706, "learning_rate": 1.2001735413333941e-05, "loss": 0.5935, "step": 12709 }, { "epoch": 0.4528207777401714, "grad_norm": 1.845212188416157, "learning_rate": 1.2000604818464887e-05, "loss": 0.5912, "step": 12710 }, { "epoch": 0.45285640486666545, "grad_norm": 1.5187388938078303, "learning_rate": 1.1999474196956378e-05, "loss": 0.5338, "step": 12711 }, { "epoch": 0.4528920319931596, "grad_norm": 1.6272040207638137, "learning_rate": 1.1998343548823474e-05, "loss": 0.5684, "step": 12712 }, { "epoch": 0.4529276591196537, "grad_norm": 1.1057101797676763, "learning_rate": 1.199721287408123e-05, "loss": 0.463, "step": 12713 }, { "epoch": 0.4529632862461478, "grad_norm": 1.5113344271225921, "learning_rate": 1.1996082172744702e-05, "loss": 0.5628, "step": 12714 }, { "epoch": 0.4529989133726419, "grad_norm": 1.5414113906429965, "learning_rate": 1.1994951444828941e-05, "loss": 0.5341, "step": 12715 }, { "epoch": 0.45303454049913605, "grad_norm": 1.104656288874509, "learning_rate": 1.1993820690349008e-05, "loss": 0.4655, "step": 12716 }, { "epoch": 0.4530701676256302, "grad_norm": 1.5618731528867882, "learning_rate": 1.1992689909319958e-05, "loss": 0.5404, "step": 12717 }, { "epoch": 0.45310579475212426, "grad_norm": 1.5505024463213974, "learning_rate": 1.1991559101756853e-05, "loss": 0.6161, "step": 12718 }, { "epoch": 0.4531414218786184, "grad_norm": 1.99290416402033, "learning_rate": 1.1990428267674744e-05, "loss": 0.4993, "step": 12719 }, { "epoch": 0.4531770490051125, "grad_norm": 2.023619207703448, "learning_rate": 1.1989297407088692e-05, "loss": 0.5733, "step": 12720 }, { "epoch": 0.4532126761316066, "grad_norm": 1.1892427807572337, "learning_rate": 1.1988166520013755e-05, "loss": 0.4879, "step": 12721 }, { "epoch": 0.4532483032581007, "grad_norm": 1.091785586847503, "learning_rate": 1.1987035606464987e-05, "loss": 0.4574, "step": 12722 }, { "epoch": 0.45328393038459486, "grad_norm": 1.2651715369379408, "learning_rate": 1.1985904666457455e-05, "loss": 0.4356, "step": 12723 }, { "epoch": 0.45331955751108893, "grad_norm": 1.685390979464221, "learning_rate": 1.1984773700006215e-05, "loss": 0.5315, "step": 12724 }, { "epoch": 0.45335518463758306, "grad_norm": 1.2510746701239517, "learning_rate": 1.1983642707126325e-05, "loss": 0.4609, "step": 12725 }, { "epoch": 0.4533908117640772, "grad_norm": 1.122649353687892, "learning_rate": 1.1982511687832845e-05, "loss": 0.4232, "step": 12726 }, { "epoch": 0.45342643889057127, "grad_norm": 1.2413118783616939, "learning_rate": 1.1981380642140836e-05, "loss": 0.4749, "step": 12727 }, { "epoch": 0.4534620660170654, "grad_norm": 1.4135820509698318, "learning_rate": 1.1980249570065359e-05, "loss": 0.6678, "step": 12728 }, { "epoch": 0.45349769314355953, "grad_norm": 1.5342634606101542, "learning_rate": 1.1979118471621477e-05, "loss": 0.5835, "step": 12729 }, { "epoch": 0.4535333202700536, "grad_norm": 1.6500537140200253, "learning_rate": 1.1977987346824244e-05, "loss": 0.525, "step": 12730 }, { "epoch": 0.45356894739654774, "grad_norm": 1.7717792926766656, "learning_rate": 1.1976856195688733e-05, "loss": 0.473, "step": 12731 }, { "epoch": 0.45360457452304187, "grad_norm": 1.0649927957129068, "learning_rate": 1.1975725018229999e-05, "loss": 0.4454, "step": 12732 }, { "epoch": 0.45364020164953595, "grad_norm": 1.6545745402384666, "learning_rate": 1.1974593814463105e-05, "loss": 0.5625, "step": 12733 }, { "epoch": 0.4536758287760301, "grad_norm": 1.453736669420384, "learning_rate": 1.197346258440311e-05, "loss": 0.603, "step": 12734 }, { "epoch": 0.4537114559025242, "grad_norm": 1.7439911897869247, "learning_rate": 1.1972331328065085e-05, "loss": 0.5863, "step": 12735 }, { "epoch": 0.4537470830290183, "grad_norm": 1.2129073357627032, "learning_rate": 1.1971200045464088e-05, "loss": 0.4862, "step": 12736 }, { "epoch": 0.4537827101555124, "grad_norm": 1.7690335802991666, "learning_rate": 1.1970068736615187e-05, "loss": 0.6002, "step": 12737 }, { "epoch": 0.45381833728200655, "grad_norm": 1.4267181524186316, "learning_rate": 1.1968937401533442e-05, "loss": 0.4969, "step": 12738 }, { "epoch": 0.4538539644085006, "grad_norm": 1.1218768649491888, "learning_rate": 1.196780604023392e-05, "loss": 0.4538, "step": 12739 }, { "epoch": 0.45388959153499475, "grad_norm": 1.6380933262776107, "learning_rate": 1.1966674652731687e-05, "loss": 0.4862, "step": 12740 }, { "epoch": 0.4539252186614889, "grad_norm": 1.7061376766306224, "learning_rate": 1.1965543239041803e-05, "loss": 0.5878, "step": 12741 }, { "epoch": 0.45396084578798296, "grad_norm": 1.2126287946288112, "learning_rate": 1.1964411799179338e-05, "loss": 0.5084, "step": 12742 }, { "epoch": 0.4539964729144771, "grad_norm": 1.253298638864551, "learning_rate": 1.1963280333159359e-05, "loss": 0.5448, "step": 12743 }, { "epoch": 0.4540321000409712, "grad_norm": 1.0390215105976863, "learning_rate": 1.196214884099693e-05, "loss": 0.4644, "step": 12744 }, { "epoch": 0.4540677271674653, "grad_norm": 1.5528095015130268, "learning_rate": 1.1961017322707114e-05, "loss": 0.5464, "step": 12745 }, { "epoch": 0.4541033542939594, "grad_norm": 1.231631354256893, "learning_rate": 1.1959885778304985e-05, "loss": 0.5345, "step": 12746 }, { "epoch": 0.45413898142045356, "grad_norm": 1.0581940102012224, "learning_rate": 1.1958754207805606e-05, "loss": 0.43, "step": 12747 }, { "epoch": 0.45417460854694763, "grad_norm": 1.5207126093629526, "learning_rate": 1.1957622611224044e-05, "loss": 0.5214, "step": 12748 }, { "epoch": 0.45421023567344176, "grad_norm": 1.0980978899635108, "learning_rate": 1.1956490988575369e-05, "loss": 0.4788, "step": 12749 }, { "epoch": 0.4542458627999359, "grad_norm": 1.2308395513082662, "learning_rate": 1.195535933987465e-05, "loss": 0.5098, "step": 12750 }, { "epoch": 0.45428148992642997, "grad_norm": 1.2073030241732448, "learning_rate": 1.1954227665136956e-05, "loss": 0.4994, "step": 12751 }, { "epoch": 0.4543171170529241, "grad_norm": 1.070757588110351, "learning_rate": 1.1953095964377353e-05, "loss": 0.4393, "step": 12752 }, { "epoch": 0.45435274417941823, "grad_norm": 1.0918898616661017, "learning_rate": 1.195196423761091e-05, "loss": 0.4471, "step": 12753 }, { "epoch": 0.4543883713059123, "grad_norm": 2.0910950493545406, "learning_rate": 1.1950832484852704e-05, "loss": 0.6195, "step": 12754 }, { "epoch": 0.45442399843240644, "grad_norm": 1.4966247242855966, "learning_rate": 1.1949700706117797e-05, "loss": 0.5203, "step": 12755 }, { "epoch": 0.45445962555890057, "grad_norm": 1.5141140390823575, "learning_rate": 1.1948568901421262e-05, "loss": 0.553, "step": 12756 }, { "epoch": 0.45449525268539465, "grad_norm": 1.5607029387347167, "learning_rate": 1.1947437070778171e-05, "loss": 0.5619, "step": 12757 }, { "epoch": 0.4545308798118888, "grad_norm": 1.173133790339938, "learning_rate": 1.1946305214203595e-05, "loss": 0.4642, "step": 12758 }, { "epoch": 0.4545665069383829, "grad_norm": 1.1377714831017485, "learning_rate": 1.1945173331712603e-05, "loss": 0.4688, "step": 12759 }, { "epoch": 0.454602134064877, "grad_norm": 1.9402824431095926, "learning_rate": 1.1944041423320267e-05, "loss": 0.5874, "step": 12760 }, { "epoch": 0.4546377611913711, "grad_norm": 1.0411339764584866, "learning_rate": 1.1942909489041663e-05, "loss": 0.4865, "step": 12761 }, { "epoch": 0.45467338831786525, "grad_norm": 1.1748230556339396, "learning_rate": 1.1941777528891862e-05, "loss": 0.4461, "step": 12762 }, { "epoch": 0.4547090154443593, "grad_norm": 1.3263399026285825, "learning_rate": 1.1940645542885935e-05, "loss": 0.4706, "step": 12763 }, { "epoch": 0.45474464257085345, "grad_norm": 3.009826588481647, "learning_rate": 1.1939513531038955e-05, "loss": 0.6169, "step": 12764 }, { "epoch": 0.4547802696973476, "grad_norm": 1.5776834918076763, "learning_rate": 1.1938381493366e-05, "loss": 0.5743, "step": 12765 }, { "epoch": 0.45481589682384166, "grad_norm": 1.4183908785396278, "learning_rate": 1.1937249429882139e-05, "loss": 0.5077, "step": 12766 }, { "epoch": 0.4548515239503358, "grad_norm": 1.4478643267068771, "learning_rate": 1.1936117340602445e-05, "loss": 0.5311, "step": 12767 }, { "epoch": 0.4548871510768299, "grad_norm": 1.6196764808374713, "learning_rate": 1.1934985225541998e-05, "loss": 0.6003, "step": 12768 }, { "epoch": 0.454922778203324, "grad_norm": 1.000707484037727, "learning_rate": 1.1933853084715872e-05, "loss": 0.4548, "step": 12769 }, { "epoch": 0.4549584053298181, "grad_norm": 1.3601292938505523, "learning_rate": 1.1932720918139139e-05, "loss": 0.4744, "step": 12770 }, { "epoch": 0.45499403245631226, "grad_norm": 2.0858258143705166, "learning_rate": 1.1931588725826873e-05, "loss": 0.5593, "step": 12771 }, { "epoch": 0.45502965958280633, "grad_norm": 1.5880128014065311, "learning_rate": 1.1930456507794153e-05, "loss": 0.5524, "step": 12772 }, { "epoch": 0.45506528670930046, "grad_norm": 1.7325063082937353, "learning_rate": 1.1929324264056059e-05, "loss": 0.6023, "step": 12773 }, { "epoch": 0.4551009138357946, "grad_norm": 1.809400826908096, "learning_rate": 1.1928191994627663e-05, "loss": 0.5942, "step": 12774 }, { "epoch": 0.45513654096228867, "grad_norm": 1.1028088435318284, "learning_rate": 1.1927059699524042e-05, "loss": 0.4864, "step": 12775 }, { "epoch": 0.4551721680887828, "grad_norm": 1.1767530948974567, "learning_rate": 1.1925927378760274e-05, "loss": 0.46, "step": 12776 }, { "epoch": 0.45520779521527693, "grad_norm": 1.0670717389722304, "learning_rate": 1.192479503235144e-05, "loss": 0.4529, "step": 12777 }, { "epoch": 0.455243422341771, "grad_norm": 1.4684139877121754, "learning_rate": 1.1923662660312611e-05, "loss": 0.6977, "step": 12778 }, { "epoch": 0.45527904946826514, "grad_norm": 1.1303678767699847, "learning_rate": 1.192253026265887e-05, "loss": 0.4858, "step": 12779 }, { "epoch": 0.45531467659475927, "grad_norm": 1.0777978368881107, "learning_rate": 1.1921397839405296e-05, "loss": 0.4714, "step": 12780 }, { "epoch": 0.45535030372125335, "grad_norm": 1.930622910189697, "learning_rate": 1.1920265390566966e-05, "loss": 0.6055, "step": 12781 }, { "epoch": 0.4553859308477475, "grad_norm": 1.5437115475578052, "learning_rate": 1.1919132916158957e-05, "loss": 0.5925, "step": 12782 }, { "epoch": 0.4554215579742416, "grad_norm": 1.2773484704505667, "learning_rate": 1.1918000416196354e-05, "loss": 0.5033, "step": 12783 }, { "epoch": 0.4554571851007357, "grad_norm": 1.0608674690415612, "learning_rate": 1.1916867890694237e-05, "loss": 0.4731, "step": 12784 }, { "epoch": 0.4554928122272298, "grad_norm": 1.488447792546952, "learning_rate": 1.1915735339667682e-05, "loss": 0.5459, "step": 12785 }, { "epoch": 0.45552843935372395, "grad_norm": 1.0884974879361784, "learning_rate": 1.1914602763131772e-05, "loss": 0.3801, "step": 12786 }, { "epoch": 0.455564066480218, "grad_norm": 1.544495481049023, "learning_rate": 1.191347016110159e-05, "loss": 0.585, "step": 12787 }, { "epoch": 0.45559969360671215, "grad_norm": 1.1540168592048141, "learning_rate": 1.1912337533592216e-05, "loss": 0.4856, "step": 12788 }, { "epoch": 0.4556353207332063, "grad_norm": 1.5669171178222525, "learning_rate": 1.191120488061873e-05, "loss": 0.571, "step": 12789 }, { "epoch": 0.45567094785970036, "grad_norm": 1.593757660311212, "learning_rate": 1.1910072202196212e-05, "loss": 0.5184, "step": 12790 }, { "epoch": 0.4557065749861945, "grad_norm": 1.5868605907630997, "learning_rate": 1.1908939498339751e-05, "loss": 0.5618, "step": 12791 }, { "epoch": 0.4557422021126886, "grad_norm": 1.2524862606545324, "learning_rate": 1.1907806769064425e-05, "loss": 0.4912, "step": 12792 }, { "epoch": 0.4557778292391827, "grad_norm": 1.0564362173024597, "learning_rate": 1.1906674014385318e-05, "loss": 0.4298, "step": 12793 }, { "epoch": 0.45581345636567683, "grad_norm": 2.4211678805136145, "learning_rate": 1.1905541234317513e-05, "loss": 0.6796, "step": 12794 }, { "epoch": 0.45584908349217096, "grad_norm": 1.425491560114086, "learning_rate": 1.1904408428876098e-05, "loss": 0.588, "step": 12795 }, { "epoch": 0.45588471061866503, "grad_norm": 1.0847273540686038, "learning_rate": 1.1903275598076152e-05, "loss": 0.4679, "step": 12796 }, { "epoch": 0.45592033774515917, "grad_norm": 1.3319767147927584, "learning_rate": 1.1902142741932761e-05, "loss": 0.4954, "step": 12797 }, { "epoch": 0.4559559648716533, "grad_norm": 1.427571511156442, "learning_rate": 1.1901009860461008e-05, "loss": 0.5629, "step": 12798 }, { "epoch": 0.45599159199814737, "grad_norm": 1.7003818191561162, "learning_rate": 1.1899876953675981e-05, "loss": 0.6257, "step": 12799 }, { "epoch": 0.4560272191246415, "grad_norm": 1.6308642362817005, "learning_rate": 1.1898744021592768e-05, "loss": 0.5096, "step": 12800 }, { "epoch": 0.45606284625113563, "grad_norm": 1.187521032941639, "learning_rate": 1.1897611064226447e-05, "loss": 0.4484, "step": 12801 }, { "epoch": 0.4560984733776297, "grad_norm": 1.6721096205160322, "learning_rate": 1.189647808159211e-05, "loss": 0.539, "step": 12802 }, { "epoch": 0.45613410050412384, "grad_norm": 1.2179990763518473, "learning_rate": 1.1895345073704842e-05, "loss": 0.4903, "step": 12803 }, { "epoch": 0.45616972763061797, "grad_norm": 1.0613637148557, "learning_rate": 1.1894212040579727e-05, "loss": 0.4778, "step": 12804 }, { "epoch": 0.45620535475711205, "grad_norm": 1.1101994022026436, "learning_rate": 1.1893078982231858e-05, "loss": 0.4547, "step": 12805 }, { "epoch": 0.4562409818836062, "grad_norm": 2.104748559295212, "learning_rate": 1.1891945898676316e-05, "loss": 0.5369, "step": 12806 }, { "epoch": 0.4562766090101003, "grad_norm": 1.9608578224831268, "learning_rate": 1.1890812789928194e-05, "loss": 0.6378, "step": 12807 }, { "epoch": 0.4563122361365944, "grad_norm": 1.5204545815947463, "learning_rate": 1.1889679656002579e-05, "loss": 0.5848, "step": 12808 }, { "epoch": 0.4563478632630885, "grad_norm": 1.1259309468513485, "learning_rate": 1.1888546496914557e-05, "loss": 0.4354, "step": 12809 }, { "epoch": 0.45638349038958265, "grad_norm": 1.2875955634933485, "learning_rate": 1.1887413312679219e-05, "loss": 0.456, "step": 12810 }, { "epoch": 0.4564191175160767, "grad_norm": 1.1579787941422397, "learning_rate": 1.1886280103311655e-05, "loss": 0.4847, "step": 12811 }, { "epoch": 0.45645474464257085, "grad_norm": 1.6318993367564976, "learning_rate": 1.1885146868826949e-05, "loss": 0.5775, "step": 12812 }, { "epoch": 0.456490371769065, "grad_norm": 1.4764991092522235, "learning_rate": 1.18840136092402e-05, "loss": 0.576, "step": 12813 }, { "epoch": 0.45652599889555906, "grad_norm": 1.6275492977155763, "learning_rate": 1.1882880324566491e-05, "loss": 0.6341, "step": 12814 }, { "epoch": 0.4565616260220532, "grad_norm": 1.1742961069445026, "learning_rate": 1.1881747014820912e-05, "loss": 0.4714, "step": 12815 }, { "epoch": 0.4565972531485473, "grad_norm": 1.4075770929335862, "learning_rate": 1.1880613680018558e-05, "loss": 0.576, "step": 12816 }, { "epoch": 0.4566328802750414, "grad_norm": 1.2961041625560688, "learning_rate": 1.1879480320174516e-05, "loss": 0.5195, "step": 12817 }, { "epoch": 0.45666850740153553, "grad_norm": 1.0603484853952692, "learning_rate": 1.1878346935303884e-05, "loss": 0.4283, "step": 12818 }, { "epoch": 0.45670413452802966, "grad_norm": 1.352614559589759, "learning_rate": 1.1877213525421746e-05, "loss": 0.4993, "step": 12819 }, { "epoch": 0.45673976165452373, "grad_norm": 1.6420404513702256, "learning_rate": 1.18760800905432e-05, "loss": 0.5442, "step": 12820 }, { "epoch": 0.45677538878101787, "grad_norm": 1.3110124326737211, "learning_rate": 1.1874946630683337e-05, "loss": 0.4569, "step": 12821 }, { "epoch": 0.456811015907512, "grad_norm": 1.6241899552279007, "learning_rate": 1.187381314585725e-05, "loss": 0.5593, "step": 12822 }, { "epoch": 0.4568466430340061, "grad_norm": 1.2042496576328483, "learning_rate": 1.1872679636080027e-05, "loss": 0.4935, "step": 12823 }, { "epoch": 0.4568822701605002, "grad_norm": 1.229947329916517, "learning_rate": 1.187154610136677e-05, "loss": 0.4622, "step": 12824 }, { "epoch": 0.45691789728699433, "grad_norm": 1.1383518936223924, "learning_rate": 1.1870412541732566e-05, "loss": 0.4234, "step": 12825 }, { "epoch": 0.4569535244134884, "grad_norm": 1.2344296121366303, "learning_rate": 1.1869278957192511e-05, "loss": 0.4923, "step": 12826 }, { "epoch": 0.45698915153998254, "grad_norm": 1.2846958910955109, "learning_rate": 1.1868145347761701e-05, "loss": 0.5093, "step": 12827 }, { "epoch": 0.45702477866647667, "grad_norm": 1.6121823076842108, "learning_rate": 1.186701171345523e-05, "loss": 0.4937, "step": 12828 }, { "epoch": 0.45706040579297075, "grad_norm": 1.4789861305175762, "learning_rate": 1.186587805428819e-05, "loss": 0.5699, "step": 12829 }, { "epoch": 0.4570960329194649, "grad_norm": 1.2670391987970901, "learning_rate": 1.1864744370275685e-05, "loss": 0.5, "step": 12830 }, { "epoch": 0.457131660045959, "grad_norm": 1.4185680576983015, "learning_rate": 1.1863610661432802e-05, "loss": 0.5956, "step": 12831 }, { "epoch": 0.4571672871724531, "grad_norm": 1.1677278654239651, "learning_rate": 1.1862476927774641e-05, "loss": 0.4586, "step": 12832 }, { "epoch": 0.4572029142989472, "grad_norm": 1.4922120504600906, "learning_rate": 1.1861343169316301e-05, "loss": 0.5798, "step": 12833 }, { "epoch": 0.45723854142544135, "grad_norm": 1.1763417226690789, "learning_rate": 1.1860209386072874e-05, "loss": 0.4655, "step": 12834 }, { "epoch": 0.4572741685519354, "grad_norm": 1.0491087942326627, "learning_rate": 1.1859075578059457e-05, "loss": 0.4807, "step": 12835 }, { "epoch": 0.45730979567842955, "grad_norm": 1.6555141267815439, "learning_rate": 1.185794174529115e-05, "loss": 0.6588, "step": 12836 }, { "epoch": 0.4573454228049237, "grad_norm": 1.1950294913260457, "learning_rate": 1.1856807887783052e-05, "loss": 0.5252, "step": 12837 }, { "epoch": 0.45738104993141776, "grad_norm": 1.6172996527642254, "learning_rate": 1.1855674005550256e-05, "loss": 0.6116, "step": 12838 }, { "epoch": 0.4574166770579119, "grad_norm": 1.3350057822967507, "learning_rate": 1.1854540098607865e-05, "loss": 0.6014, "step": 12839 }, { "epoch": 0.457452304184406, "grad_norm": 1.2003842093702046, "learning_rate": 1.1853406166970975e-05, "loss": 0.4777, "step": 12840 }, { "epoch": 0.4574879313109001, "grad_norm": 1.0433365138709987, "learning_rate": 1.185227221065469e-05, "loss": 0.4596, "step": 12841 }, { "epoch": 0.45752355843739423, "grad_norm": 1.0941175243567764, "learning_rate": 1.1851138229674101e-05, "loss": 0.4653, "step": 12842 }, { "epoch": 0.45755918556388836, "grad_norm": 1.060636176221606, "learning_rate": 1.1850004224044315e-05, "loss": 0.4986, "step": 12843 }, { "epoch": 0.45759481269038244, "grad_norm": 1.1279715344800867, "learning_rate": 1.1848870193780432e-05, "loss": 0.4835, "step": 12844 }, { "epoch": 0.45763043981687657, "grad_norm": 1.2226466895417123, "learning_rate": 1.184773613889755e-05, "loss": 0.4234, "step": 12845 }, { "epoch": 0.4576660669433707, "grad_norm": 1.9390692370480607, "learning_rate": 1.1846602059410765e-05, "loss": 0.626, "step": 12846 }, { "epoch": 0.4577016940698648, "grad_norm": 1.507415956171331, "learning_rate": 1.1845467955335188e-05, "loss": 0.518, "step": 12847 }, { "epoch": 0.4577373211963589, "grad_norm": 1.0620458639226293, "learning_rate": 1.1844333826685913e-05, "loss": 0.4943, "step": 12848 }, { "epoch": 0.45777294832285303, "grad_norm": 1.443020574920382, "learning_rate": 1.1843199673478044e-05, "loss": 0.5609, "step": 12849 }, { "epoch": 0.4578085754493471, "grad_norm": 0.9572383966593747, "learning_rate": 1.1842065495726682e-05, "loss": 0.4146, "step": 12850 }, { "epoch": 0.45784420257584124, "grad_norm": 1.459290136514113, "learning_rate": 1.1840931293446931e-05, "loss": 0.5026, "step": 12851 }, { "epoch": 0.4578798297023354, "grad_norm": 1.3165060210992454, "learning_rate": 1.1839797066653893e-05, "loss": 0.5713, "step": 12852 }, { "epoch": 0.45791545682882945, "grad_norm": 1.3859736316881923, "learning_rate": 1.1838662815362673e-05, "loss": 0.6006, "step": 12853 }, { "epoch": 0.4579510839553236, "grad_norm": 1.1462406630453528, "learning_rate": 1.183752853958837e-05, "loss": 0.4618, "step": 12854 }, { "epoch": 0.4579867110818177, "grad_norm": 1.2072133076280878, "learning_rate": 1.1836394239346092e-05, "loss": 0.4569, "step": 12855 }, { "epoch": 0.4580223382083118, "grad_norm": 1.0686338617662443, "learning_rate": 1.183525991465094e-05, "loss": 0.4388, "step": 12856 }, { "epoch": 0.4580579653348059, "grad_norm": 1.1178831962742575, "learning_rate": 1.183412556551802e-05, "loss": 0.4706, "step": 12857 }, { "epoch": 0.45809359246130005, "grad_norm": 1.7074659507040637, "learning_rate": 1.1832991191962435e-05, "loss": 0.6076, "step": 12858 }, { "epoch": 0.4581292195877941, "grad_norm": 1.2119248173558534, "learning_rate": 1.1831856793999292e-05, "loss": 0.4831, "step": 12859 }, { "epoch": 0.45816484671428825, "grad_norm": 1.5472051675376717, "learning_rate": 1.1830722371643697e-05, "loss": 0.5766, "step": 12860 }, { "epoch": 0.4582004738407824, "grad_norm": 1.091811409610869, "learning_rate": 1.1829587924910753e-05, "loss": 0.4066, "step": 12861 }, { "epoch": 0.45823610096727646, "grad_norm": 1.0929304371206394, "learning_rate": 1.1828453453815567e-05, "loss": 0.4589, "step": 12862 }, { "epoch": 0.4582717280937706, "grad_norm": 1.2522291102964005, "learning_rate": 1.1827318958373241e-05, "loss": 0.4902, "step": 12863 }, { "epoch": 0.4583073552202647, "grad_norm": 1.5138503377051038, "learning_rate": 1.1826184438598891e-05, "loss": 0.6293, "step": 12864 }, { "epoch": 0.4583429823467588, "grad_norm": 1.387343366586419, "learning_rate": 1.1825049894507616e-05, "loss": 0.538, "step": 12865 }, { "epoch": 0.45837860947325293, "grad_norm": 1.178971618885453, "learning_rate": 1.1823915326114528e-05, "loss": 0.5091, "step": 12866 }, { "epoch": 0.45841423659974706, "grad_norm": 1.8556423671271076, "learning_rate": 1.1822780733434733e-05, "loss": 0.6093, "step": 12867 }, { "epoch": 0.45844986372624114, "grad_norm": 1.1290794215666204, "learning_rate": 1.1821646116483337e-05, "loss": 0.4668, "step": 12868 }, { "epoch": 0.45848549085273527, "grad_norm": 1.2974333025216247, "learning_rate": 1.1820511475275448e-05, "loss": 0.6077, "step": 12869 }, { "epoch": 0.4585211179792294, "grad_norm": 0.9875655994468127, "learning_rate": 1.1819376809826181e-05, "loss": 0.4271, "step": 12870 }, { "epoch": 0.4585567451057235, "grad_norm": 1.8272936708146947, "learning_rate": 1.1818242120150637e-05, "loss": 0.6294, "step": 12871 }, { "epoch": 0.4585923722322176, "grad_norm": 1.035619996704924, "learning_rate": 1.1817107406263929e-05, "loss": 0.4666, "step": 12872 }, { "epoch": 0.45862799935871174, "grad_norm": 1.5724743681859183, "learning_rate": 1.1815972668181166e-05, "loss": 0.4908, "step": 12873 }, { "epoch": 0.4586636264852058, "grad_norm": 1.5001510704311973, "learning_rate": 1.1814837905917458e-05, "loss": 0.5301, "step": 12874 }, { "epoch": 0.45869925361169994, "grad_norm": 1.835698275518439, "learning_rate": 1.1813703119487912e-05, "loss": 0.5857, "step": 12875 }, { "epoch": 0.4587348807381941, "grad_norm": 1.1063180356925781, "learning_rate": 1.181256830890764e-05, "loss": 0.4251, "step": 12876 }, { "epoch": 0.45877050786468815, "grad_norm": 1.172499257334547, "learning_rate": 1.181143347419176e-05, "loss": 0.4873, "step": 12877 }, { "epoch": 0.4588061349911823, "grad_norm": 1.4938758741514626, "learning_rate": 1.1810298615355376e-05, "loss": 0.5309, "step": 12878 }, { "epoch": 0.4588417621176764, "grad_norm": 0.9845303462458439, "learning_rate": 1.18091637324136e-05, "loss": 0.4785, "step": 12879 }, { "epoch": 0.4588773892441705, "grad_norm": 1.579060565606712, "learning_rate": 1.1808028825381541e-05, "loss": 0.5994, "step": 12880 }, { "epoch": 0.4589130163706646, "grad_norm": 1.3614251126643926, "learning_rate": 1.1806893894274319e-05, "loss": 0.6357, "step": 12881 }, { "epoch": 0.45894864349715875, "grad_norm": 1.0088932082602882, "learning_rate": 1.180575893910704e-05, "loss": 0.4228, "step": 12882 }, { "epoch": 0.4589842706236528, "grad_norm": 1.1107093479900592, "learning_rate": 1.1804623959894817e-05, "loss": 0.5135, "step": 12883 }, { "epoch": 0.45901989775014695, "grad_norm": 1.1814083233102597, "learning_rate": 1.1803488956652767e-05, "loss": 0.5078, "step": 12884 }, { "epoch": 0.4590555248766411, "grad_norm": 1.4207545483604616, "learning_rate": 1.1802353929396e-05, "loss": 0.5605, "step": 12885 }, { "epoch": 0.45909115200313516, "grad_norm": 3.531046718693557, "learning_rate": 1.1801218878139628e-05, "loss": 0.5276, "step": 12886 }, { "epoch": 0.4591267791296293, "grad_norm": 1.4907679536872436, "learning_rate": 1.1800083802898769e-05, "loss": 0.6224, "step": 12887 }, { "epoch": 0.4591624062561234, "grad_norm": 1.4916345661907615, "learning_rate": 1.179894870368854e-05, "loss": 0.5534, "step": 12888 }, { "epoch": 0.4591980333826175, "grad_norm": 1.1401779870862065, "learning_rate": 1.1797813580524049e-05, "loss": 0.3933, "step": 12889 }, { "epoch": 0.45923366050911163, "grad_norm": 1.2116500511134214, "learning_rate": 1.1796678433420415e-05, "loss": 0.4752, "step": 12890 }, { "epoch": 0.45926928763560576, "grad_norm": 1.067080337608058, "learning_rate": 1.1795543262392748e-05, "loss": 0.4807, "step": 12891 }, { "epoch": 0.45930491476209984, "grad_norm": 1.1572953998105564, "learning_rate": 1.1794408067456171e-05, "loss": 0.5057, "step": 12892 }, { "epoch": 0.45934054188859397, "grad_norm": 1.240689962613267, "learning_rate": 1.1793272848625796e-05, "loss": 0.5858, "step": 12893 }, { "epoch": 0.4593761690150881, "grad_norm": 1.4946226092332342, "learning_rate": 1.1792137605916736e-05, "loss": 0.5768, "step": 12894 }, { "epoch": 0.4594117961415822, "grad_norm": 1.4558378545914947, "learning_rate": 1.1791002339344116e-05, "loss": 0.5651, "step": 12895 }, { "epoch": 0.4594474232680763, "grad_norm": 1.5198869502843315, "learning_rate": 1.1789867048923048e-05, "loss": 0.5468, "step": 12896 }, { "epoch": 0.45948305039457044, "grad_norm": 1.4047389167137458, "learning_rate": 1.1788731734668646e-05, "loss": 0.6025, "step": 12897 }, { "epoch": 0.4595186775210645, "grad_norm": 1.1555557710756037, "learning_rate": 1.1787596396596028e-05, "loss": 0.5097, "step": 12898 }, { "epoch": 0.45955430464755864, "grad_norm": 1.8125134274650563, "learning_rate": 1.178646103472032e-05, "loss": 0.5954, "step": 12899 }, { "epoch": 0.4595899317740528, "grad_norm": 1.2071016102203393, "learning_rate": 1.1785325649056633e-05, "loss": 0.4775, "step": 12900 }, { "epoch": 0.45962555890054685, "grad_norm": 1.796496097567623, "learning_rate": 1.1784190239620089e-05, "loss": 0.5165, "step": 12901 }, { "epoch": 0.459661186027041, "grad_norm": 1.4517460741550257, "learning_rate": 1.17830548064258e-05, "loss": 0.5276, "step": 12902 }, { "epoch": 0.4596968131535351, "grad_norm": 1.173867137349121, "learning_rate": 1.1781919349488894e-05, "loss": 0.4826, "step": 12903 }, { "epoch": 0.45973244028002924, "grad_norm": 1.3965637975790999, "learning_rate": 1.1780783868824485e-05, "loss": 0.5216, "step": 12904 }, { "epoch": 0.4597680674065233, "grad_norm": 1.1871608142178793, "learning_rate": 1.1779648364447694e-05, "loss": 0.4616, "step": 12905 }, { "epoch": 0.45980369453301745, "grad_norm": 1.295345324019497, "learning_rate": 1.1778512836373643e-05, "loss": 0.5388, "step": 12906 }, { "epoch": 0.4598393216595116, "grad_norm": 1.4441698279150976, "learning_rate": 1.1777377284617448e-05, "loss": 0.6222, "step": 12907 }, { "epoch": 0.45987494878600566, "grad_norm": 1.7107073271137678, "learning_rate": 1.1776241709194235e-05, "loss": 0.5924, "step": 12908 }, { "epoch": 0.4599105759124998, "grad_norm": 1.5801576694227852, "learning_rate": 1.177510611011912e-05, "loss": 0.5499, "step": 12909 }, { "epoch": 0.4599462030389939, "grad_norm": 1.6446541342859862, "learning_rate": 1.1773970487407224e-05, "loss": 0.5243, "step": 12910 }, { "epoch": 0.459981830165488, "grad_norm": 1.7653934163935243, "learning_rate": 1.1772834841073676e-05, "loss": 0.5404, "step": 12911 }, { "epoch": 0.4600174572919821, "grad_norm": 1.04756961235892, "learning_rate": 1.177169917113359e-05, "loss": 0.433, "step": 12912 }, { "epoch": 0.46005308441847625, "grad_norm": 1.8068298730155437, "learning_rate": 1.1770563477602091e-05, "loss": 0.6109, "step": 12913 }, { "epoch": 0.46008871154497033, "grad_norm": 1.1137317406105969, "learning_rate": 1.1769427760494303e-05, "loss": 0.4652, "step": 12914 }, { "epoch": 0.46012433867146446, "grad_norm": 1.2094265703894163, "learning_rate": 1.1768292019825348e-05, "loss": 0.5172, "step": 12915 }, { "epoch": 0.4601599657979586, "grad_norm": 1.4361277617008825, "learning_rate": 1.1767156255610348e-05, "loss": 0.4827, "step": 12916 }, { "epoch": 0.46019559292445267, "grad_norm": 1.0329746385883924, "learning_rate": 1.1766020467864427e-05, "loss": 0.4268, "step": 12917 }, { "epoch": 0.4602312200509468, "grad_norm": 1.2144289871611837, "learning_rate": 1.1764884656602711e-05, "loss": 0.4662, "step": 12918 }, { "epoch": 0.46026684717744093, "grad_norm": 1.0666677387226564, "learning_rate": 1.1763748821840319e-05, "loss": 0.4446, "step": 12919 }, { "epoch": 0.460302474303935, "grad_norm": 1.1792180749530785, "learning_rate": 1.176261296359238e-05, "loss": 0.4645, "step": 12920 }, { "epoch": 0.46033810143042914, "grad_norm": 1.1469605205018802, "learning_rate": 1.1761477081874016e-05, "loss": 0.47, "step": 12921 }, { "epoch": 0.46037372855692327, "grad_norm": 1.1083516657165982, "learning_rate": 1.1760341176700357e-05, "loss": 0.5091, "step": 12922 }, { "epoch": 0.46040935568341734, "grad_norm": 1.6220342811656658, "learning_rate": 1.1759205248086522e-05, "loss": 0.5749, "step": 12923 }, { "epoch": 0.4604449828099115, "grad_norm": 1.1458667218560012, "learning_rate": 1.1758069296047641e-05, "loss": 0.4311, "step": 12924 }, { "epoch": 0.4604806099364056, "grad_norm": 1.5455512684192432, "learning_rate": 1.1756933320598838e-05, "loss": 0.6049, "step": 12925 }, { "epoch": 0.4605162370628997, "grad_norm": 1.0180648562753525, "learning_rate": 1.175579732175524e-05, "loss": 0.443, "step": 12926 }, { "epoch": 0.4605518641893938, "grad_norm": 1.3454208680765303, "learning_rate": 1.1754661299531975e-05, "loss": 0.5896, "step": 12927 }, { "epoch": 0.46058749131588794, "grad_norm": 1.1561573174913453, "learning_rate": 1.1753525253944164e-05, "loss": 0.4912, "step": 12928 }, { "epoch": 0.460623118442382, "grad_norm": 1.6087890286085258, "learning_rate": 1.1752389185006944e-05, "loss": 0.512, "step": 12929 }, { "epoch": 0.46065874556887615, "grad_norm": 1.3401757022981893, "learning_rate": 1.1751253092735432e-05, "loss": 0.4819, "step": 12930 }, { "epoch": 0.4606943726953703, "grad_norm": 1.087636751363516, "learning_rate": 1.1750116977144763e-05, "loss": 0.4533, "step": 12931 }, { "epoch": 0.46072999982186436, "grad_norm": 1.0410136113542878, "learning_rate": 1.1748980838250065e-05, "loss": 0.4361, "step": 12932 }, { "epoch": 0.4607656269483585, "grad_norm": 1.4969417830932559, "learning_rate": 1.1747844676066462e-05, "loss": 0.6445, "step": 12933 }, { "epoch": 0.4608012540748526, "grad_norm": 1.7678050639692682, "learning_rate": 1.1746708490609087e-05, "loss": 0.4803, "step": 12934 }, { "epoch": 0.4608368812013467, "grad_norm": 1.007847663168268, "learning_rate": 1.1745572281893068e-05, "loss": 0.4114, "step": 12935 }, { "epoch": 0.4608725083278408, "grad_norm": 1.448729712716718, "learning_rate": 1.1744436049933531e-05, "loss": 0.5833, "step": 12936 }, { "epoch": 0.46090813545433496, "grad_norm": 1.654508286357905, "learning_rate": 1.1743299794745612e-05, "loss": 0.5932, "step": 12937 }, { "epoch": 0.46094376258082903, "grad_norm": 1.1487343522047635, "learning_rate": 1.1742163516344439e-05, "loss": 0.4902, "step": 12938 }, { "epoch": 0.46097938970732316, "grad_norm": 1.2662753715396446, "learning_rate": 1.1741027214745138e-05, "loss": 0.557, "step": 12939 }, { "epoch": 0.4610150168338173, "grad_norm": 1.4710788808650166, "learning_rate": 1.1739890889962844e-05, "loss": 0.4938, "step": 12940 }, { "epoch": 0.46105064396031137, "grad_norm": 1.1536338572844538, "learning_rate": 1.1738754542012688e-05, "loss": 0.4964, "step": 12941 }, { "epoch": 0.4610862710868055, "grad_norm": 1.52671671188424, "learning_rate": 1.1737618170909797e-05, "loss": 0.5344, "step": 12942 }, { "epoch": 0.46112189821329963, "grad_norm": 1.6148845944060617, "learning_rate": 1.1736481776669307e-05, "loss": 0.6639, "step": 12943 }, { "epoch": 0.4611575253397937, "grad_norm": 1.3882413609501048, "learning_rate": 1.1735345359306346e-05, "loss": 0.4966, "step": 12944 }, { "epoch": 0.46119315246628784, "grad_norm": 1.1256356109920254, "learning_rate": 1.173420891883605e-05, "loss": 0.4662, "step": 12945 }, { "epoch": 0.46122877959278197, "grad_norm": 2.9399060598768356, "learning_rate": 1.1733072455273553e-05, "loss": 0.5547, "step": 12946 }, { "epoch": 0.46126440671927604, "grad_norm": 1.4978171997086611, "learning_rate": 1.173193596863398e-05, "loss": 0.5881, "step": 12947 }, { "epoch": 0.4613000338457702, "grad_norm": 1.7179254010401412, "learning_rate": 1.1730799458932473e-05, "loss": 0.6009, "step": 12948 }, { "epoch": 0.4613356609722643, "grad_norm": 1.6382089587801534, "learning_rate": 1.172966292618416e-05, "loss": 0.5766, "step": 12949 }, { "epoch": 0.4613712880987584, "grad_norm": 1.1042475648511194, "learning_rate": 1.1728526370404176e-05, "loss": 0.448, "step": 12950 }, { "epoch": 0.4614069152252525, "grad_norm": 1.3437574031435988, "learning_rate": 1.1727389791607655e-05, "loss": 0.5227, "step": 12951 }, { "epoch": 0.46144254235174664, "grad_norm": 1.1502572946305722, "learning_rate": 1.1726253189809732e-05, "loss": 0.4216, "step": 12952 }, { "epoch": 0.4614781694782407, "grad_norm": 1.0332693274520373, "learning_rate": 1.1725116565025544e-05, "loss": 0.4118, "step": 12953 }, { "epoch": 0.46151379660473485, "grad_norm": 1.0227009378459164, "learning_rate": 1.1723979917270218e-05, "loss": 0.4786, "step": 12954 }, { "epoch": 0.461549423731229, "grad_norm": 1.4897923417560528, "learning_rate": 1.1722843246558899e-05, "loss": 0.5079, "step": 12955 }, { "epoch": 0.46158505085772306, "grad_norm": 1.1953854098058465, "learning_rate": 1.1721706552906716e-05, "loss": 0.4724, "step": 12956 }, { "epoch": 0.4616206779842172, "grad_norm": 1.119994467686597, "learning_rate": 1.1720569836328805e-05, "loss": 0.4519, "step": 12957 }, { "epoch": 0.4616563051107113, "grad_norm": 1.5838118160015842, "learning_rate": 1.1719433096840305e-05, "loss": 0.5895, "step": 12958 }, { "epoch": 0.4616919322372054, "grad_norm": 3.007104719576156, "learning_rate": 1.1718296334456353e-05, "loss": 0.5725, "step": 12959 }, { "epoch": 0.4617275593636995, "grad_norm": 1.0814406021135277, "learning_rate": 1.1717159549192085e-05, "loss": 0.4216, "step": 12960 }, { "epoch": 0.46176318649019366, "grad_norm": 1.2825498730296896, "learning_rate": 1.1716022741062637e-05, "loss": 0.5848, "step": 12961 }, { "epoch": 0.46179881361668773, "grad_norm": 1.3667681829909795, "learning_rate": 1.1714885910083143e-05, "loss": 0.5988, "step": 12962 }, { "epoch": 0.46183444074318186, "grad_norm": 1.1527497792413597, "learning_rate": 1.1713749056268748e-05, "loss": 0.511, "step": 12963 }, { "epoch": 0.461870067869676, "grad_norm": 1.4068577180237933, "learning_rate": 1.1712612179634587e-05, "loss": 0.5935, "step": 12964 }, { "epoch": 0.46190569499617007, "grad_norm": 1.1342793160517795, "learning_rate": 1.1711475280195795e-05, "loss": 0.4568, "step": 12965 }, { "epoch": 0.4619413221226642, "grad_norm": 1.1906377216293655, "learning_rate": 1.1710338357967514e-05, "loss": 0.4239, "step": 12966 }, { "epoch": 0.46197694924915833, "grad_norm": 1.5340528034267908, "learning_rate": 1.1709201412964883e-05, "loss": 0.5856, "step": 12967 }, { "epoch": 0.4620125763756524, "grad_norm": 1.1659620350162438, "learning_rate": 1.1708064445203042e-05, "loss": 0.4806, "step": 12968 }, { "epoch": 0.46204820350214654, "grad_norm": 1.2370399011253004, "learning_rate": 1.1706927454697125e-05, "loss": 0.4383, "step": 12969 }, { "epoch": 0.46208383062864067, "grad_norm": 1.8003511855420347, "learning_rate": 1.1705790441462281e-05, "loss": 0.5475, "step": 12970 }, { "epoch": 0.46211945775513474, "grad_norm": 1.110327240135971, "learning_rate": 1.1704653405513642e-05, "loss": 0.4648, "step": 12971 }, { "epoch": 0.4621550848816289, "grad_norm": 1.1185788570138406, "learning_rate": 1.1703516346866353e-05, "loss": 0.4547, "step": 12972 }, { "epoch": 0.462190712008123, "grad_norm": 1.7508178548371298, "learning_rate": 1.1702379265535551e-05, "loss": 0.6509, "step": 12973 }, { "epoch": 0.4622263391346171, "grad_norm": 1.1076968563511902, "learning_rate": 1.1701242161536382e-05, "loss": 0.4269, "step": 12974 }, { "epoch": 0.4622619662611112, "grad_norm": 1.1542327079569918, "learning_rate": 1.1700105034883982e-05, "loss": 0.4486, "step": 12975 }, { "epoch": 0.46229759338760534, "grad_norm": 1.832871849891704, "learning_rate": 1.1698967885593494e-05, "loss": 0.5693, "step": 12976 }, { "epoch": 0.4623332205140994, "grad_norm": 1.096921262248887, "learning_rate": 1.1697830713680065e-05, "loss": 0.4613, "step": 12977 }, { "epoch": 0.46236884764059355, "grad_norm": 1.1049545429629737, "learning_rate": 1.1696693519158833e-05, "loss": 0.4775, "step": 12978 }, { "epoch": 0.4624044747670877, "grad_norm": 1.7812359629409389, "learning_rate": 1.1695556302044937e-05, "loss": 0.5681, "step": 12979 }, { "epoch": 0.46244010189358176, "grad_norm": 1.2649923143755453, "learning_rate": 1.1694419062353525e-05, "loss": 0.4384, "step": 12980 }, { "epoch": 0.4624757290200759, "grad_norm": 2.7581061159019002, "learning_rate": 1.1693281800099741e-05, "loss": 0.5308, "step": 12981 }, { "epoch": 0.46251135614657, "grad_norm": 1.5464211993086234, "learning_rate": 1.1692144515298724e-05, "loss": 0.5589, "step": 12982 }, { "epoch": 0.4625469832730641, "grad_norm": 1.756418241787127, "learning_rate": 1.1691007207965623e-05, "loss": 0.6061, "step": 12983 }, { "epoch": 0.4625826103995582, "grad_norm": 1.1070977874860475, "learning_rate": 1.1689869878115577e-05, "loss": 0.4732, "step": 12984 }, { "epoch": 0.46261823752605236, "grad_norm": 1.5166530330161248, "learning_rate": 1.1688732525763733e-05, "loss": 0.5293, "step": 12985 }, { "epoch": 0.46265386465254643, "grad_norm": 0.9601311871634807, "learning_rate": 1.1687595150925235e-05, "loss": 0.4107, "step": 12986 }, { "epoch": 0.46268949177904056, "grad_norm": 1.634010382616614, "learning_rate": 1.1686457753615228e-05, "loss": 0.6194, "step": 12987 }, { "epoch": 0.4627251189055347, "grad_norm": 1.2094962300313963, "learning_rate": 1.1685320333848857e-05, "loss": 0.4708, "step": 12988 }, { "epoch": 0.46276074603202877, "grad_norm": 1.760399281672153, "learning_rate": 1.168418289164127e-05, "loss": 0.5595, "step": 12989 }, { "epoch": 0.4627963731585229, "grad_norm": 1.1287732148325968, "learning_rate": 1.1683045427007604e-05, "loss": 0.4086, "step": 12990 }, { "epoch": 0.46283200028501703, "grad_norm": 2.0781013376022677, "learning_rate": 1.1681907939963017e-05, "loss": 0.5173, "step": 12991 }, { "epoch": 0.4628676274115111, "grad_norm": 1.1487769117914288, "learning_rate": 1.168077043052265e-05, "loss": 0.4298, "step": 12992 }, { "epoch": 0.46290325453800524, "grad_norm": 1.4433756431453286, "learning_rate": 1.1679632898701649e-05, "loss": 0.6587, "step": 12993 }, { "epoch": 0.46293888166449937, "grad_norm": 2.1163712470369913, "learning_rate": 1.1678495344515164e-05, "loss": 0.5959, "step": 12994 }, { "epoch": 0.46297450879099344, "grad_norm": 1.2993379421353297, "learning_rate": 1.1677357767978337e-05, "loss": 0.4778, "step": 12995 }, { "epoch": 0.4630101359174876, "grad_norm": 4.905514273881456, "learning_rate": 1.1676220169106321e-05, "loss": 0.4445, "step": 12996 }, { "epoch": 0.4630457630439817, "grad_norm": 1.252116609029403, "learning_rate": 1.1675082547914263e-05, "loss": 0.4531, "step": 12997 }, { "epoch": 0.4630813901704758, "grad_norm": 1.1367418738592179, "learning_rate": 1.1673944904417309e-05, "loss": 0.4204, "step": 12998 }, { "epoch": 0.4631170172969699, "grad_norm": 1.6705785377737108, "learning_rate": 1.1672807238630606e-05, "loss": 0.5814, "step": 12999 }, { "epoch": 0.46315264442346404, "grad_norm": 1.1227508057505997, "learning_rate": 1.1671669550569309e-05, "loss": 0.4465, "step": 13000 }, { "epoch": 0.4631882715499581, "grad_norm": 1.3921540490255615, "learning_rate": 1.1670531840248562e-05, "loss": 0.5272, "step": 13001 }, { "epoch": 0.46322389867645225, "grad_norm": 1.3123952615382233, "learning_rate": 1.1669394107683514e-05, "loss": 0.6426, "step": 13002 }, { "epoch": 0.4632595258029464, "grad_norm": 1.2066097596711505, "learning_rate": 1.1668256352889317e-05, "loss": 0.4684, "step": 13003 }, { "epoch": 0.46329515292944046, "grad_norm": 1.0638800059848021, "learning_rate": 1.1667118575881122e-05, "loss": 0.4429, "step": 13004 }, { "epoch": 0.4633307800559346, "grad_norm": 1.6985546463422971, "learning_rate": 1.166598077667408e-05, "loss": 0.6244, "step": 13005 }, { "epoch": 0.4633664071824287, "grad_norm": 1.0987643652269978, "learning_rate": 1.1664842955283335e-05, "loss": 0.4189, "step": 13006 }, { "epoch": 0.4634020343089228, "grad_norm": 1.4628645334239485, "learning_rate": 1.1663705111724046e-05, "loss": 0.5674, "step": 13007 }, { "epoch": 0.4634376614354169, "grad_norm": 1.5923191594224326, "learning_rate": 1.166256724601136e-05, "loss": 0.6486, "step": 13008 }, { "epoch": 0.46347328856191106, "grad_norm": 1.165261170667161, "learning_rate": 1.166142935816043e-05, "loss": 0.438, "step": 13009 }, { "epoch": 0.46350891568840513, "grad_norm": 1.030988324533097, "learning_rate": 1.1660291448186404e-05, "loss": 0.4382, "step": 13010 }, { "epoch": 0.46354454281489926, "grad_norm": 1.417371204890494, "learning_rate": 1.165915351610444e-05, "loss": 0.5174, "step": 13011 }, { "epoch": 0.4635801699413934, "grad_norm": 1.4711375863382965, "learning_rate": 1.1658015561929686e-05, "loss": 0.5837, "step": 13012 }, { "epoch": 0.46361579706788747, "grad_norm": 1.5870059899911038, "learning_rate": 1.1656877585677293e-05, "loss": 0.5274, "step": 13013 }, { "epoch": 0.4636514241943816, "grad_norm": 1.1612919472208039, "learning_rate": 1.1655739587362418e-05, "loss": 0.4918, "step": 13014 }, { "epoch": 0.46368705132087573, "grad_norm": 1.3448355098459899, "learning_rate": 1.1654601567000214e-05, "loss": 0.5239, "step": 13015 }, { "epoch": 0.4637226784473698, "grad_norm": 1.161967630400623, "learning_rate": 1.1653463524605834e-05, "loss": 0.4685, "step": 13016 }, { "epoch": 0.46375830557386394, "grad_norm": 1.1093085475483895, "learning_rate": 1.1652325460194434e-05, "loss": 0.4842, "step": 13017 }, { "epoch": 0.46379393270035807, "grad_norm": 1.5822949727468054, "learning_rate": 1.1651187373781161e-05, "loss": 0.5792, "step": 13018 }, { "epoch": 0.46382955982685214, "grad_norm": 1.487277998030732, "learning_rate": 1.1650049265381178e-05, "loss": 0.5519, "step": 13019 }, { "epoch": 0.4638651869533463, "grad_norm": 1.2880433516844074, "learning_rate": 1.1648911135009635e-05, "loss": 0.5917, "step": 13020 }, { "epoch": 0.4639008140798404, "grad_norm": 1.444670567316729, "learning_rate": 1.1647772982681685e-05, "loss": 0.5574, "step": 13021 }, { "epoch": 0.4639364412063345, "grad_norm": 1.2154265929572188, "learning_rate": 1.1646634808412488e-05, "loss": 0.5316, "step": 13022 }, { "epoch": 0.4639720683328286, "grad_norm": 1.1300867215201824, "learning_rate": 1.1645496612217198e-05, "loss": 0.4383, "step": 13023 }, { "epoch": 0.46400769545932274, "grad_norm": 1.1516694943538528, "learning_rate": 1.1644358394110971e-05, "loss": 0.493, "step": 13024 }, { "epoch": 0.4640433225858168, "grad_norm": 1.163381358558479, "learning_rate": 1.1643220154108956e-05, "loss": 0.4723, "step": 13025 }, { "epoch": 0.46407894971231095, "grad_norm": 1.216325266837087, "learning_rate": 1.1642081892226322e-05, "loss": 0.4618, "step": 13026 }, { "epoch": 0.4641145768388051, "grad_norm": 1.171489290516457, "learning_rate": 1.1640943608478222e-05, "loss": 0.4457, "step": 13027 }, { "epoch": 0.46415020396529916, "grad_norm": 1.071991321391638, "learning_rate": 1.1639805302879807e-05, "loss": 0.4792, "step": 13028 }, { "epoch": 0.4641858310917933, "grad_norm": 1.3911248068726683, "learning_rate": 1.1638666975446238e-05, "loss": 0.5621, "step": 13029 }, { "epoch": 0.4642214582182874, "grad_norm": 1.2533733572800703, "learning_rate": 1.1637528626192676e-05, "loss": 0.5277, "step": 13030 }, { "epoch": 0.4642570853447815, "grad_norm": 1.0706594786921215, "learning_rate": 1.1636390255134273e-05, "loss": 0.4063, "step": 13031 }, { "epoch": 0.4642927124712756, "grad_norm": 1.8730871447586304, "learning_rate": 1.1635251862286189e-05, "loss": 0.5594, "step": 13032 }, { "epoch": 0.46432833959776976, "grad_norm": 1.1116393622097147, "learning_rate": 1.1634113447663587e-05, "loss": 0.4203, "step": 13033 }, { "epoch": 0.46436396672426383, "grad_norm": 1.2167198987455152, "learning_rate": 1.1632975011281622e-05, "loss": 0.4511, "step": 13034 }, { "epoch": 0.46439959385075796, "grad_norm": 0.9820885269078121, "learning_rate": 1.1631836553155454e-05, "loss": 0.4277, "step": 13035 }, { "epoch": 0.4644352209772521, "grad_norm": 1.5570299129164402, "learning_rate": 1.1630698073300237e-05, "loss": 0.6427, "step": 13036 }, { "epoch": 0.46447084810374617, "grad_norm": 1.1398253830510656, "learning_rate": 1.1629559571731136e-05, "loss": 0.4761, "step": 13037 }, { "epoch": 0.4645064752302403, "grad_norm": 1.605658102610919, "learning_rate": 1.1628421048463315e-05, "loss": 0.5552, "step": 13038 }, { "epoch": 0.46454210235673443, "grad_norm": 1.603262626991311, "learning_rate": 1.1627282503511927e-05, "loss": 0.5614, "step": 13039 }, { "epoch": 0.4645777294832285, "grad_norm": 1.065133396821144, "learning_rate": 1.1626143936892133e-05, "loss": 0.422, "step": 13040 }, { "epoch": 0.46461335660972264, "grad_norm": 1.722251033184295, "learning_rate": 1.1625005348619098e-05, "loss": 0.6222, "step": 13041 }, { "epoch": 0.46464898373621677, "grad_norm": 1.2685095155152908, "learning_rate": 1.1623866738707983e-05, "loss": 0.6201, "step": 13042 }, { "epoch": 0.46468461086271085, "grad_norm": 1.5819888862127862, "learning_rate": 1.1622728107173947e-05, "loss": 0.549, "step": 13043 }, { "epoch": 0.464720237989205, "grad_norm": 1.1115871149186318, "learning_rate": 1.1621589454032147e-05, "loss": 0.457, "step": 13044 }, { "epoch": 0.4647558651156991, "grad_norm": 1.7509878298979227, "learning_rate": 1.1620450779297753e-05, "loss": 0.5505, "step": 13045 }, { "epoch": 0.4647914922421932, "grad_norm": 1.16519510209003, "learning_rate": 1.1619312082985927e-05, "loss": 0.4868, "step": 13046 }, { "epoch": 0.4648271193686873, "grad_norm": 1.5457070851962191, "learning_rate": 1.1618173365111825e-05, "loss": 0.5282, "step": 13047 }, { "epoch": 0.46486274649518144, "grad_norm": 1.3999110525773595, "learning_rate": 1.1617034625690613e-05, "loss": 0.5433, "step": 13048 }, { "epoch": 0.4648983736216755, "grad_norm": 1.5069117054840868, "learning_rate": 1.1615895864737459e-05, "loss": 0.5892, "step": 13049 }, { "epoch": 0.46493400074816965, "grad_norm": 1.2854545694827881, "learning_rate": 1.1614757082267518e-05, "loss": 0.4953, "step": 13050 }, { "epoch": 0.4649696278746638, "grad_norm": 1.4114054427206333, "learning_rate": 1.1613618278295959e-05, "loss": 0.5723, "step": 13051 }, { "epoch": 0.46500525500115786, "grad_norm": 1.0576144710576747, "learning_rate": 1.1612479452837946e-05, "loss": 0.4413, "step": 13052 }, { "epoch": 0.465040882127652, "grad_norm": 1.1004987659676995, "learning_rate": 1.1611340605908643e-05, "loss": 0.5326, "step": 13053 }, { "epoch": 0.4650765092541461, "grad_norm": 1.1394633627090296, "learning_rate": 1.1610201737523212e-05, "loss": 0.459, "step": 13054 }, { "epoch": 0.4651121363806402, "grad_norm": 1.0443824322141115, "learning_rate": 1.160906284769682e-05, "loss": 0.439, "step": 13055 }, { "epoch": 0.4651477635071343, "grad_norm": 1.3597741613833598, "learning_rate": 1.160792393644463e-05, "loss": 0.4674, "step": 13056 }, { "epoch": 0.46518339063362846, "grad_norm": 1.4197373701014948, "learning_rate": 1.160678500378181e-05, "loss": 0.544, "step": 13057 }, { "epoch": 0.46521901776012253, "grad_norm": 1.4563150348779041, "learning_rate": 1.1605646049723523e-05, "loss": 0.4736, "step": 13058 }, { "epoch": 0.46525464488661666, "grad_norm": 1.4228313554802343, "learning_rate": 1.1604507074284938e-05, "loss": 0.5631, "step": 13059 }, { "epoch": 0.4652902720131108, "grad_norm": 1.1262376286121736, "learning_rate": 1.1603368077481218e-05, "loss": 0.4463, "step": 13060 }, { "epoch": 0.46532589913960487, "grad_norm": 1.1268082264763586, "learning_rate": 1.1602229059327534e-05, "loss": 0.4767, "step": 13061 }, { "epoch": 0.465361526266099, "grad_norm": 1.2161015777616349, "learning_rate": 1.1601090019839049e-05, "loss": 0.5315, "step": 13062 }, { "epoch": 0.46539715339259313, "grad_norm": 1.0362698555663343, "learning_rate": 1.1599950959030932e-05, "loss": 0.4434, "step": 13063 }, { "epoch": 0.4654327805190872, "grad_norm": 1.1763078397089526, "learning_rate": 1.159881187691835e-05, "loss": 0.456, "step": 13064 }, { "epoch": 0.46546840764558134, "grad_norm": 2.260459395373298, "learning_rate": 1.159767277351647e-05, "loss": 0.4737, "step": 13065 }, { "epoch": 0.46550403477207547, "grad_norm": 1.0274107816501017, "learning_rate": 1.1596533648840458e-05, "loss": 0.4506, "step": 13066 }, { "epoch": 0.46553966189856955, "grad_norm": 1.4163334653764965, "learning_rate": 1.1595394502905488e-05, "loss": 0.5578, "step": 13067 }, { "epoch": 0.4655752890250637, "grad_norm": 1.6493525048423425, "learning_rate": 1.1594255335726725e-05, "loss": 0.5674, "step": 13068 }, { "epoch": 0.4656109161515578, "grad_norm": 2.401610461310696, "learning_rate": 1.1593116147319334e-05, "loss": 0.5771, "step": 13069 }, { "epoch": 0.4656465432780519, "grad_norm": 1.3518014606008184, "learning_rate": 1.1591976937698493e-05, "loss": 0.5246, "step": 13070 }, { "epoch": 0.465682170404546, "grad_norm": 1.4876560200334514, "learning_rate": 1.1590837706879361e-05, "loss": 0.5321, "step": 13071 }, { "epoch": 0.46571779753104015, "grad_norm": 1.1022967719042571, "learning_rate": 1.1589698454877116e-05, "loss": 0.4512, "step": 13072 }, { "epoch": 0.4657534246575342, "grad_norm": 1.0733351471288555, "learning_rate": 1.1588559181706926e-05, "loss": 0.424, "step": 13073 }, { "epoch": 0.46578905178402835, "grad_norm": 1.7718754363405145, "learning_rate": 1.1587419887383958e-05, "loss": 0.6215, "step": 13074 }, { "epoch": 0.4658246789105225, "grad_norm": 1.103153576501538, "learning_rate": 1.1586280571923385e-05, "loss": 0.4384, "step": 13075 }, { "epoch": 0.46586030603701656, "grad_norm": 1.0348386168055363, "learning_rate": 1.158514123534038e-05, "loss": 0.5039, "step": 13076 }, { "epoch": 0.4658959331635107, "grad_norm": 1.432029728035361, "learning_rate": 1.1584001877650109e-05, "loss": 0.5147, "step": 13077 }, { "epoch": 0.4659315602900048, "grad_norm": 2.252322092911258, "learning_rate": 1.1582862498867747e-05, "loss": 0.6287, "step": 13078 }, { "epoch": 0.4659671874164989, "grad_norm": 2.4438102839108975, "learning_rate": 1.1581723099008464e-05, "loss": 0.5842, "step": 13079 }, { "epoch": 0.466002814542993, "grad_norm": 1.1746723342943042, "learning_rate": 1.158058367808743e-05, "loss": 0.4748, "step": 13080 }, { "epoch": 0.46603844166948716, "grad_norm": 1.6771406977101866, "learning_rate": 1.1579444236119823e-05, "loss": 0.5131, "step": 13081 }, { "epoch": 0.46607406879598123, "grad_norm": 1.0241902325681687, "learning_rate": 1.1578304773120811e-05, "loss": 0.4426, "step": 13082 }, { "epoch": 0.46610969592247536, "grad_norm": 1.0462876134516752, "learning_rate": 1.1577165289105565e-05, "loss": 0.482, "step": 13083 }, { "epoch": 0.4661453230489695, "grad_norm": 1.0365391025232016, "learning_rate": 1.1576025784089264e-05, "loss": 0.4723, "step": 13084 }, { "epoch": 0.46618095017546357, "grad_norm": 1.0485320810135534, "learning_rate": 1.1574886258087077e-05, "loss": 0.4404, "step": 13085 }, { "epoch": 0.4662165773019577, "grad_norm": 1.297809592721457, "learning_rate": 1.157374671111418e-05, "loss": 0.4656, "step": 13086 }, { "epoch": 0.46625220442845183, "grad_norm": 1.0748605849084583, "learning_rate": 1.1572607143185745e-05, "loss": 0.4423, "step": 13087 }, { "epoch": 0.4662878315549459, "grad_norm": 1.0987398983237477, "learning_rate": 1.1571467554316947e-05, "loss": 0.5125, "step": 13088 }, { "epoch": 0.46632345868144004, "grad_norm": 1.615911960596381, "learning_rate": 1.1570327944522959e-05, "loss": 0.5674, "step": 13089 }, { "epoch": 0.46635908580793417, "grad_norm": 1.0882917180029712, "learning_rate": 1.1569188313818958e-05, "loss": 0.4163, "step": 13090 }, { "epoch": 0.46639471293442825, "grad_norm": 1.1248381810146584, "learning_rate": 1.1568048662220117e-05, "loss": 0.4545, "step": 13091 }, { "epoch": 0.4664303400609224, "grad_norm": 1.7600967988511467, "learning_rate": 1.1566908989741612e-05, "loss": 0.5669, "step": 13092 }, { "epoch": 0.4664659671874165, "grad_norm": 1.046095764479137, "learning_rate": 1.156576929639862e-05, "loss": 0.4752, "step": 13093 }, { "epoch": 0.46650159431391064, "grad_norm": 1.478036589382555, "learning_rate": 1.1564629582206313e-05, "loss": 0.5171, "step": 13094 }, { "epoch": 0.4665372214404047, "grad_norm": 1.052923526605997, "learning_rate": 1.1563489847179872e-05, "loss": 0.4545, "step": 13095 }, { "epoch": 0.46657284856689885, "grad_norm": 1.0546904730129387, "learning_rate": 1.1562350091334466e-05, "loss": 0.4693, "step": 13096 }, { "epoch": 0.466608475693393, "grad_norm": 1.0857651925896874, "learning_rate": 1.1561210314685282e-05, "loss": 0.4405, "step": 13097 }, { "epoch": 0.46664410281988705, "grad_norm": 1.4552108280309164, "learning_rate": 1.156007051724749e-05, "loss": 0.5539, "step": 13098 }, { "epoch": 0.4666797299463812, "grad_norm": 1.4339878643109207, "learning_rate": 1.155893069903627e-05, "loss": 0.5625, "step": 13099 }, { "epoch": 0.4667153570728753, "grad_norm": 1.5783750980237838, "learning_rate": 1.1557790860066796e-05, "loss": 0.5665, "step": 13100 }, { "epoch": 0.4667509841993694, "grad_norm": 1.1508378201177214, "learning_rate": 1.1556651000354246e-05, "loss": 0.4151, "step": 13101 }, { "epoch": 0.4667866113258635, "grad_norm": 1.9240161767979593, "learning_rate": 1.1555511119913805e-05, "loss": 0.6209, "step": 13102 }, { "epoch": 0.46682223845235765, "grad_norm": 1.3630561371638081, "learning_rate": 1.1554371218760641e-05, "loss": 0.466, "step": 13103 }, { "epoch": 0.4668578655788517, "grad_norm": 1.2329810870277098, "learning_rate": 1.1553231296909942e-05, "loss": 0.4764, "step": 13104 }, { "epoch": 0.46689349270534586, "grad_norm": 1.5514208053590646, "learning_rate": 1.155209135437688e-05, "loss": 0.5882, "step": 13105 }, { "epoch": 0.46692911983184, "grad_norm": 1.0834236434059257, "learning_rate": 1.1550951391176636e-05, "loss": 0.5129, "step": 13106 }, { "epoch": 0.46696474695833406, "grad_norm": 1.8331164587257198, "learning_rate": 1.154981140732439e-05, "loss": 0.6361, "step": 13107 }, { "epoch": 0.4670003740848282, "grad_norm": 1.2008543145447992, "learning_rate": 1.1548671402835325e-05, "loss": 0.5623, "step": 13108 }, { "epoch": 0.4670360012113223, "grad_norm": 1.2371604470737922, "learning_rate": 1.1547531377724616e-05, "loss": 0.4807, "step": 13109 }, { "epoch": 0.4670716283378164, "grad_norm": 1.1350446035205362, "learning_rate": 1.1546391332007448e-05, "loss": 0.4823, "step": 13110 }, { "epoch": 0.46710725546431053, "grad_norm": 1.7453829500969869, "learning_rate": 1.1545251265698993e-05, "loss": 0.5632, "step": 13111 }, { "epoch": 0.46714288259080466, "grad_norm": 1.2119213770586867, "learning_rate": 1.154411117881444e-05, "loss": 0.5263, "step": 13112 }, { "epoch": 0.46717850971729874, "grad_norm": 1.4961110398250979, "learning_rate": 1.1542971071368968e-05, "loss": 0.5409, "step": 13113 }, { "epoch": 0.46721413684379287, "grad_norm": 1.1015335940007922, "learning_rate": 1.1541830943377758e-05, "loss": 0.4602, "step": 13114 }, { "epoch": 0.467249763970287, "grad_norm": 1.3944250299372685, "learning_rate": 1.1540690794855991e-05, "loss": 0.569, "step": 13115 }, { "epoch": 0.4672853910967811, "grad_norm": 1.425522737914776, "learning_rate": 1.153955062581885e-05, "loss": 0.5476, "step": 13116 }, { "epoch": 0.4673210182232752, "grad_norm": 1.8907107484289536, "learning_rate": 1.1538410436281513e-05, "loss": 0.5791, "step": 13117 }, { "epoch": 0.46735664534976934, "grad_norm": 1.1737349996491513, "learning_rate": 1.153727022625917e-05, "loss": 0.4703, "step": 13118 }, { "epoch": 0.4673922724762634, "grad_norm": 1.2266206186662068, "learning_rate": 1.1536129995766995e-05, "loss": 0.6139, "step": 13119 }, { "epoch": 0.46742789960275755, "grad_norm": 1.1280165413923005, "learning_rate": 1.153498974482018e-05, "loss": 0.4832, "step": 13120 }, { "epoch": 0.4674635267292517, "grad_norm": 1.5463286963596918, "learning_rate": 1.1533849473433901e-05, "loss": 0.6281, "step": 13121 }, { "epoch": 0.46749915385574575, "grad_norm": 1.482756476160222, "learning_rate": 1.1532709181623342e-05, "loss": 0.6276, "step": 13122 }, { "epoch": 0.4675347809822399, "grad_norm": 1.7138275675447188, "learning_rate": 1.1531568869403693e-05, "loss": 0.5859, "step": 13123 }, { "epoch": 0.467570408108734, "grad_norm": 1.0157909248618109, "learning_rate": 1.1530428536790134e-05, "loss": 0.3958, "step": 13124 }, { "epoch": 0.4676060352352281, "grad_norm": 1.102500868194023, "learning_rate": 1.1529288183797849e-05, "loss": 0.4401, "step": 13125 }, { "epoch": 0.4676416623617222, "grad_norm": 1.624989324306841, "learning_rate": 1.1528147810442021e-05, "loss": 0.5916, "step": 13126 }, { "epoch": 0.46767728948821635, "grad_norm": 1.195337969503088, "learning_rate": 1.1527007416737836e-05, "loss": 0.4198, "step": 13127 }, { "epoch": 0.46771291661471043, "grad_norm": 1.399688971386943, "learning_rate": 1.1525867002700484e-05, "loss": 0.605, "step": 13128 }, { "epoch": 0.46774854374120456, "grad_norm": 1.174664451006212, "learning_rate": 1.1524726568345145e-05, "loss": 0.4803, "step": 13129 }, { "epoch": 0.4677841708676987, "grad_norm": 1.8568852442435988, "learning_rate": 1.1523586113687001e-05, "loss": 0.5669, "step": 13130 }, { "epoch": 0.46781979799419277, "grad_norm": 2.2517162879111408, "learning_rate": 1.1522445638741248e-05, "loss": 0.5, "step": 13131 }, { "epoch": 0.4678554251206869, "grad_norm": 1.1587383331349785, "learning_rate": 1.1521305143523066e-05, "loss": 0.4867, "step": 13132 }, { "epoch": 0.467891052247181, "grad_norm": 1.681104714068593, "learning_rate": 1.152016462804764e-05, "loss": 0.6108, "step": 13133 }, { "epoch": 0.4679266793736751, "grad_norm": 1.434350044459641, "learning_rate": 1.1519024092330164e-05, "loss": 0.4316, "step": 13134 }, { "epoch": 0.46796230650016923, "grad_norm": 2.129440971145206, "learning_rate": 1.1517883536385822e-05, "loss": 0.5653, "step": 13135 }, { "epoch": 0.46799793362666336, "grad_norm": 1.1352867761078411, "learning_rate": 1.1516742960229796e-05, "loss": 0.4813, "step": 13136 }, { "epoch": 0.46803356075315744, "grad_norm": 1.0841219317159916, "learning_rate": 1.1515602363877276e-05, "loss": 0.4224, "step": 13137 }, { "epoch": 0.46806918787965157, "grad_norm": 1.6359964570341357, "learning_rate": 1.1514461747343452e-05, "loss": 0.5591, "step": 13138 }, { "epoch": 0.4681048150061457, "grad_norm": 1.6285744791794945, "learning_rate": 1.1513321110643514e-05, "loss": 0.48, "step": 13139 }, { "epoch": 0.4681404421326398, "grad_norm": 1.1795173756152875, "learning_rate": 1.1512180453792645e-05, "loss": 0.4734, "step": 13140 }, { "epoch": 0.4681760692591339, "grad_norm": 1.09817718220232, "learning_rate": 1.1511039776806037e-05, "loss": 0.4694, "step": 13141 }, { "epoch": 0.46821169638562804, "grad_norm": 1.9204915903564923, "learning_rate": 1.1509899079698879e-05, "loss": 0.5511, "step": 13142 }, { "epoch": 0.4682473235121221, "grad_norm": 1.813174042140505, "learning_rate": 1.1508758362486358e-05, "loss": 0.5719, "step": 13143 }, { "epoch": 0.46828295063861625, "grad_norm": 1.5341073877567242, "learning_rate": 1.1507617625183664e-05, "loss": 0.5351, "step": 13144 }, { "epoch": 0.4683185777651104, "grad_norm": 1.8023466793296665, "learning_rate": 1.1506476867805992e-05, "loss": 0.655, "step": 13145 }, { "epoch": 0.46835420489160445, "grad_norm": 1.1872790374508442, "learning_rate": 1.1505336090368524e-05, "loss": 0.5407, "step": 13146 }, { "epoch": 0.4683898320180986, "grad_norm": 1.1752429789086758, "learning_rate": 1.1504195292886455e-05, "loss": 0.4741, "step": 13147 }, { "epoch": 0.4684254591445927, "grad_norm": 1.1764346954815321, "learning_rate": 1.1503054475374972e-05, "loss": 0.4745, "step": 13148 }, { "epoch": 0.4684610862710868, "grad_norm": 1.1062453749464236, "learning_rate": 1.1501913637849272e-05, "loss": 0.4234, "step": 13149 }, { "epoch": 0.4684967133975809, "grad_norm": 1.4741371137287016, "learning_rate": 1.150077278032454e-05, "loss": 0.5855, "step": 13150 }, { "epoch": 0.46853234052407505, "grad_norm": 1.130647024461449, "learning_rate": 1.1499631902815968e-05, "loss": 0.5205, "step": 13151 }, { "epoch": 0.46856796765056913, "grad_norm": 2.5968599611282204, "learning_rate": 1.149849100533875e-05, "loss": 0.5807, "step": 13152 }, { "epoch": 0.46860359477706326, "grad_norm": 1.663565101597746, "learning_rate": 1.1497350087908079e-05, "loss": 0.6038, "step": 13153 }, { "epoch": 0.4686392219035574, "grad_norm": 1.6334622939751486, "learning_rate": 1.1496209150539144e-05, "loss": 0.6049, "step": 13154 }, { "epoch": 0.46867484903005147, "grad_norm": 1.259509386945312, "learning_rate": 1.1495068193247138e-05, "loss": 0.4469, "step": 13155 }, { "epoch": 0.4687104761565456, "grad_norm": 1.217883612637135, "learning_rate": 1.1493927216047253e-05, "loss": 0.4932, "step": 13156 }, { "epoch": 0.46874610328303973, "grad_norm": 1.393658493161998, "learning_rate": 1.1492786218954686e-05, "loss": 0.5087, "step": 13157 }, { "epoch": 0.4687817304095338, "grad_norm": 1.0388736895563333, "learning_rate": 1.1491645201984625e-05, "loss": 0.4437, "step": 13158 }, { "epoch": 0.46881735753602793, "grad_norm": 1.0279832107903528, "learning_rate": 1.1490504165152264e-05, "loss": 0.4113, "step": 13159 }, { "epoch": 0.46885298466252207, "grad_norm": 1.4969222507124933, "learning_rate": 1.14893631084728e-05, "loss": 0.5212, "step": 13160 }, { "epoch": 0.46888861178901614, "grad_norm": 1.8599250721067129, "learning_rate": 1.1488222031961428e-05, "loss": 0.5493, "step": 13161 }, { "epoch": 0.46892423891551027, "grad_norm": 1.5829653575443625, "learning_rate": 1.1487080935633335e-05, "loss": 0.6204, "step": 13162 }, { "epoch": 0.4689598660420044, "grad_norm": 1.609768785705119, "learning_rate": 1.1485939819503717e-05, "loss": 0.5703, "step": 13163 }, { "epoch": 0.4689954931684985, "grad_norm": 1.4632736619516382, "learning_rate": 1.1484798683587777e-05, "loss": 0.6066, "step": 13164 }, { "epoch": 0.4690311202949926, "grad_norm": 1.5031976010612713, "learning_rate": 1.1483657527900705e-05, "loss": 0.4799, "step": 13165 }, { "epoch": 0.46906674742148674, "grad_norm": 1.5606468915678569, "learning_rate": 1.1482516352457695e-05, "loss": 0.6083, "step": 13166 }, { "epoch": 0.4691023745479808, "grad_norm": 2.0903516527726476, "learning_rate": 1.1481375157273942e-05, "loss": 0.5528, "step": 13167 }, { "epoch": 0.46913800167447495, "grad_norm": 1.13195680949924, "learning_rate": 1.1480233942364645e-05, "loss": 0.4276, "step": 13168 }, { "epoch": 0.4691736288009691, "grad_norm": 1.933067891789076, "learning_rate": 1.1479092707745e-05, "loss": 0.5475, "step": 13169 }, { "epoch": 0.46920925592746315, "grad_norm": 1.2821694860526256, "learning_rate": 1.1477951453430198e-05, "loss": 0.537, "step": 13170 }, { "epoch": 0.4692448830539573, "grad_norm": 1.5184977703518208, "learning_rate": 1.147681017943544e-05, "loss": 0.6061, "step": 13171 }, { "epoch": 0.4692805101804514, "grad_norm": 1.6215260303126566, "learning_rate": 1.1475668885775926e-05, "loss": 0.5581, "step": 13172 }, { "epoch": 0.4693161373069455, "grad_norm": 1.2762097837374875, "learning_rate": 1.1474527572466847e-05, "loss": 0.6151, "step": 13173 }, { "epoch": 0.4693517644334396, "grad_norm": 1.107501512680031, "learning_rate": 1.14733862395234e-05, "loss": 0.4248, "step": 13174 }, { "epoch": 0.46938739155993375, "grad_norm": 1.5654028311161328, "learning_rate": 1.1472244886960788e-05, "loss": 0.6256, "step": 13175 }, { "epoch": 0.46942301868642783, "grad_norm": 1.16790089587428, "learning_rate": 1.1471103514794207e-05, "loss": 0.4796, "step": 13176 }, { "epoch": 0.46945864581292196, "grad_norm": 1.6456829430979643, "learning_rate": 1.1469962123038854e-05, "loss": 0.5063, "step": 13177 }, { "epoch": 0.4694942729394161, "grad_norm": 1.5956611560865424, "learning_rate": 1.1468820711709928e-05, "loss": 0.5953, "step": 13178 }, { "epoch": 0.46952990006591017, "grad_norm": 1.7628265895230772, "learning_rate": 1.1467679280822626e-05, "loss": 0.6088, "step": 13179 }, { "epoch": 0.4695655271924043, "grad_norm": 1.1322868963986956, "learning_rate": 1.146653783039215e-05, "loss": 0.4284, "step": 13180 }, { "epoch": 0.46960115431889843, "grad_norm": 1.41806700133848, "learning_rate": 1.1465396360433698e-05, "loss": 0.5691, "step": 13181 }, { "epoch": 0.4696367814453925, "grad_norm": 2.082782492732279, "learning_rate": 1.1464254870962467e-05, "loss": 0.6066, "step": 13182 }, { "epoch": 0.46967240857188663, "grad_norm": 1.8897397693683604, "learning_rate": 1.1463113361993663e-05, "loss": 0.6249, "step": 13183 }, { "epoch": 0.46970803569838077, "grad_norm": 2.3888092724825896, "learning_rate": 1.1461971833542481e-05, "loss": 0.5574, "step": 13184 }, { "epoch": 0.46974366282487484, "grad_norm": 1.1124140115399634, "learning_rate": 1.1460830285624119e-05, "loss": 0.4269, "step": 13185 }, { "epoch": 0.469779289951369, "grad_norm": 2.059425306513908, "learning_rate": 1.145968871825378e-05, "loss": 0.506, "step": 13186 }, { "epoch": 0.4698149170778631, "grad_norm": 1.9379323009592673, "learning_rate": 1.145854713144667e-05, "loss": 0.5026, "step": 13187 }, { "epoch": 0.4698505442043572, "grad_norm": 1.4015698906646574, "learning_rate": 1.1457405525217982e-05, "loss": 0.5336, "step": 13188 }, { "epoch": 0.4698861713308513, "grad_norm": 1.0887974808683774, "learning_rate": 1.145626389958292e-05, "loss": 0.4543, "step": 13189 }, { "epoch": 0.46992179845734544, "grad_norm": 1.0705013957737386, "learning_rate": 1.1455122254556689e-05, "loss": 0.4605, "step": 13190 }, { "epoch": 0.4699574255838395, "grad_norm": 1.0771621241622118, "learning_rate": 1.145398059015449e-05, "loss": 0.4655, "step": 13191 }, { "epoch": 0.46999305271033365, "grad_norm": 1.1760333617057455, "learning_rate": 1.1452838906391519e-05, "loss": 0.4494, "step": 13192 }, { "epoch": 0.4700286798368278, "grad_norm": 1.2491981365967038, "learning_rate": 1.1451697203282984e-05, "loss": 0.5153, "step": 13193 }, { "epoch": 0.47006430696332185, "grad_norm": 1.071351429866734, "learning_rate": 1.1450555480844087e-05, "loss": 0.4734, "step": 13194 }, { "epoch": 0.470099934089816, "grad_norm": 1.4295576864284705, "learning_rate": 1.144941373909003e-05, "loss": 0.5376, "step": 13195 }, { "epoch": 0.4701355612163101, "grad_norm": 1.0224416885235652, "learning_rate": 1.1448271978036014e-05, "loss": 0.3956, "step": 13196 }, { "epoch": 0.4701711883428042, "grad_norm": 1.3381540225486352, "learning_rate": 1.1447130197697246e-05, "loss": 0.4667, "step": 13197 }, { "epoch": 0.4702068154692983, "grad_norm": 1.479985058670759, "learning_rate": 1.1445988398088925e-05, "loss": 0.466, "step": 13198 }, { "epoch": 0.47024244259579245, "grad_norm": 1.5219025917429099, "learning_rate": 1.1444846579226261e-05, "loss": 0.5037, "step": 13199 }, { "epoch": 0.47027806972228653, "grad_norm": 1.7583386915646684, "learning_rate": 1.1443704741124455e-05, "loss": 0.5379, "step": 13200 }, { "epoch": 0.47031369684878066, "grad_norm": 1.9569132585388929, "learning_rate": 1.144256288379871e-05, "loss": 0.5991, "step": 13201 }, { "epoch": 0.4703493239752748, "grad_norm": 1.0334005617295723, "learning_rate": 1.1441421007264234e-05, "loss": 0.4424, "step": 13202 }, { "epoch": 0.47038495110176887, "grad_norm": 1.0872963923478098, "learning_rate": 1.1440279111536228e-05, "loss": 0.4062, "step": 13203 }, { "epoch": 0.470420578228263, "grad_norm": 1.0380623424243243, "learning_rate": 1.1439137196629896e-05, "loss": 0.462, "step": 13204 }, { "epoch": 0.47045620535475713, "grad_norm": 1.4627443674544311, "learning_rate": 1.143799526256045e-05, "loss": 0.5676, "step": 13205 }, { "epoch": 0.4704918324812512, "grad_norm": 1.4292885095205987, "learning_rate": 1.1436853309343092e-05, "loss": 0.519, "step": 13206 }, { "epoch": 0.47052745960774534, "grad_norm": 1.4321425534362688, "learning_rate": 1.1435711336993027e-05, "loss": 0.5894, "step": 13207 }, { "epoch": 0.47056308673423947, "grad_norm": 1.12680132901417, "learning_rate": 1.1434569345525462e-05, "loss": 0.4687, "step": 13208 }, { "epoch": 0.47059871386073354, "grad_norm": 1.6434885233562027, "learning_rate": 1.1433427334955605e-05, "loss": 0.5707, "step": 13209 }, { "epoch": 0.4706343409872277, "grad_norm": 1.4071378371762682, "learning_rate": 1.1432285305298658e-05, "loss": 0.5138, "step": 13210 }, { "epoch": 0.4706699681137218, "grad_norm": 1.2968793431736496, "learning_rate": 1.1431143256569835e-05, "loss": 0.5062, "step": 13211 }, { "epoch": 0.4707055952402159, "grad_norm": 1.6006875487951486, "learning_rate": 1.1430001188784335e-05, "loss": 0.5151, "step": 13212 }, { "epoch": 0.47074122236671, "grad_norm": 1.0594384892479596, "learning_rate": 1.1428859101957372e-05, "loss": 0.4707, "step": 13213 }, { "epoch": 0.47077684949320414, "grad_norm": 1.1101216872127944, "learning_rate": 1.1427716996104152e-05, "loss": 0.4867, "step": 13214 }, { "epoch": 0.4708124766196982, "grad_norm": 1.4719512115008884, "learning_rate": 1.1426574871239879e-05, "loss": 0.5194, "step": 13215 }, { "epoch": 0.47084810374619235, "grad_norm": 1.446704662527336, "learning_rate": 1.1425432727379769e-05, "loss": 0.5977, "step": 13216 }, { "epoch": 0.4708837308726865, "grad_norm": 1.8631843188883805, "learning_rate": 1.1424290564539022e-05, "loss": 0.5667, "step": 13217 }, { "epoch": 0.47091935799918055, "grad_norm": 1.0493010586635767, "learning_rate": 1.1423148382732854e-05, "loss": 0.4821, "step": 13218 }, { "epoch": 0.4709549851256747, "grad_norm": 0.9909534564562957, "learning_rate": 1.1422006181976468e-05, "loss": 0.4568, "step": 13219 }, { "epoch": 0.4709906122521688, "grad_norm": 1.1874354441442647, "learning_rate": 1.1420863962285078e-05, "loss": 0.4809, "step": 13220 }, { "epoch": 0.4710262393786629, "grad_norm": 1.3536019958195744, "learning_rate": 1.1419721723673889e-05, "loss": 0.5502, "step": 13221 }, { "epoch": 0.471061866505157, "grad_norm": 2.2724942878256775, "learning_rate": 1.1418579466158113e-05, "loss": 0.5318, "step": 13222 }, { "epoch": 0.47109749363165115, "grad_norm": 1.3623858026784679, "learning_rate": 1.1417437189752961e-05, "loss": 0.5561, "step": 13223 }, { "epoch": 0.47113312075814523, "grad_norm": 1.0604782560671848, "learning_rate": 1.1416294894473641e-05, "loss": 0.4596, "step": 13224 }, { "epoch": 0.47116874788463936, "grad_norm": 1.0179204462948939, "learning_rate": 1.1415152580335367e-05, "loss": 0.4452, "step": 13225 }, { "epoch": 0.4712043750111335, "grad_norm": 1.566208377008144, "learning_rate": 1.1414010247353345e-05, "loss": 0.5993, "step": 13226 }, { "epoch": 0.47124000213762757, "grad_norm": 1.7293154146615637, "learning_rate": 1.1412867895542787e-05, "loss": 0.5485, "step": 13227 }, { "epoch": 0.4712756292641217, "grad_norm": 1.715492215469372, "learning_rate": 1.1411725524918909e-05, "loss": 0.5995, "step": 13228 }, { "epoch": 0.47131125639061583, "grad_norm": 1.4949006838792078, "learning_rate": 1.1410583135496917e-05, "loss": 0.5263, "step": 13229 }, { "epoch": 0.4713468835171099, "grad_norm": 1.6063864054202213, "learning_rate": 1.1409440727292025e-05, "loss": 0.5612, "step": 13230 }, { "epoch": 0.47138251064360404, "grad_norm": 1.3409638738504088, "learning_rate": 1.1408298300319445e-05, "loss": 0.5618, "step": 13231 }, { "epoch": 0.47141813777009817, "grad_norm": 1.9865246790564024, "learning_rate": 1.1407155854594387e-05, "loss": 0.5408, "step": 13232 }, { "epoch": 0.47145376489659224, "grad_norm": 1.1374989014496175, "learning_rate": 1.1406013390132067e-05, "loss": 0.5087, "step": 13233 }, { "epoch": 0.4714893920230864, "grad_norm": 1.1263768654598492, "learning_rate": 1.1404870906947693e-05, "loss": 0.4905, "step": 13234 }, { "epoch": 0.4715250191495805, "grad_norm": 1.2290161035397489, "learning_rate": 1.1403728405056484e-05, "loss": 0.5292, "step": 13235 }, { "epoch": 0.4715606462760746, "grad_norm": 1.2776577802808893, "learning_rate": 1.1402585884473649e-05, "loss": 0.5042, "step": 13236 }, { "epoch": 0.4715962734025687, "grad_norm": 1.6072473604852482, "learning_rate": 1.1401443345214405e-05, "loss": 0.4934, "step": 13237 }, { "epoch": 0.47163190052906284, "grad_norm": 1.4501120667280307, "learning_rate": 1.1400300787293961e-05, "loss": 0.5856, "step": 13238 }, { "epoch": 0.4716675276555569, "grad_norm": 1.2002909152440215, "learning_rate": 1.1399158210727533e-05, "loss": 0.4535, "step": 13239 }, { "epoch": 0.47170315478205105, "grad_norm": 1.7690884695095814, "learning_rate": 1.1398015615530338e-05, "loss": 0.658, "step": 13240 }, { "epoch": 0.4717387819085452, "grad_norm": 1.8267145722822788, "learning_rate": 1.1396873001717583e-05, "loss": 0.5941, "step": 13241 }, { "epoch": 0.47177440903503925, "grad_norm": 1.4888506588822574, "learning_rate": 1.1395730369304492e-05, "loss": 0.6282, "step": 13242 }, { "epoch": 0.4718100361615334, "grad_norm": 1.2452481674120297, "learning_rate": 1.1394587718306276e-05, "loss": 0.4558, "step": 13243 }, { "epoch": 0.4718456632880275, "grad_norm": 1.468467517918979, "learning_rate": 1.1393445048738145e-05, "loss": 0.5363, "step": 13244 }, { "epoch": 0.4718812904145216, "grad_norm": 1.5675547702861248, "learning_rate": 1.1392302360615322e-05, "loss": 0.557, "step": 13245 }, { "epoch": 0.4719169175410157, "grad_norm": 1.2037454564987884, "learning_rate": 1.1391159653953018e-05, "loss": 0.5124, "step": 13246 }, { "epoch": 0.47195254466750985, "grad_norm": 1.4250035421161518, "learning_rate": 1.1390016928766456e-05, "loss": 0.5618, "step": 13247 }, { "epoch": 0.47198817179400393, "grad_norm": 1.1533541015949835, "learning_rate": 1.1388874185070843e-05, "loss": 0.4646, "step": 13248 }, { "epoch": 0.47202379892049806, "grad_norm": 1.2131449229342264, "learning_rate": 1.1387731422881401e-05, "loss": 0.4606, "step": 13249 }, { "epoch": 0.4720594260469922, "grad_norm": 1.5054721635205082, "learning_rate": 1.1386588642213346e-05, "loss": 0.5084, "step": 13250 }, { "epoch": 0.47209505317348627, "grad_norm": 1.188524813635264, "learning_rate": 1.1385445843081892e-05, "loss": 0.4687, "step": 13251 }, { "epoch": 0.4721306802999804, "grad_norm": 1.1006901900459394, "learning_rate": 1.1384303025502257e-05, "loss": 0.4852, "step": 13252 }, { "epoch": 0.47216630742647453, "grad_norm": 1.4870177288164736, "learning_rate": 1.1383160189489663e-05, "loss": 0.6112, "step": 13253 }, { "epoch": 0.4722019345529686, "grad_norm": 1.0865163044615003, "learning_rate": 1.1382017335059324e-05, "loss": 0.482, "step": 13254 }, { "epoch": 0.47223756167946274, "grad_norm": 1.617163599247192, "learning_rate": 1.1380874462226457e-05, "loss": 0.5325, "step": 13255 }, { "epoch": 0.47227318880595687, "grad_norm": 1.0319348036699902, "learning_rate": 1.137973157100628e-05, "loss": 0.3986, "step": 13256 }, { "epoch": 0.47230881593245094, "grad_norm": 1.2261714415382385, "learning_rate": 1.1378588661414014e-05, "loss": 0.4807, "step": 13257 }, { "epoch": 0.4723444430589451, "grad_norm": 1.5984662008027883, "learning_rate": 1.1377445733464876e-05, "loss": 0.534, "step": 13258 }, { "epoch": 0.4723800701854392, "grad_norm": 1.752074581654811, "learning_rate": 1.1376302787174088e-05, "loss": 0.5117, "step": 13259 }, { "epoch": 0.4724156973119333, "grad_norm": 1.5640016441460607, "learning_rate": 1.1375159822556863e-05, "loss": 0.6285, "step": 13260 }, { "epoch": 0.4724513244384274, "grad_norm": 1.5677428465944432, "learning_rate": 1.1374016839628423e-05, "loss": 0.6053, "step": 13261 }, { "epoch": 0.47248695156492154, "grad_norm": 2.332759917228837, "learning_rate": 1.1372873838403993e-05, "loss": 0.5851, "step": 13262 }, { "epoch": 0.4725225786914156, "grad_norm": 2.204171526217113, "learning_rate": 1.1371730818898785e-05, "loss": 0.5753, "step": 13263 }, { "epoch": 0.47255820581790975, "grad_norm": 1.1030020461203134, "learning_rate": 1.1370587781128021e-05, "loss": 0.4465, "step": 13264 }, { "epoch": 0.4725938329444039, "grad_norm": 1.09165740569424, "learning_rate": 1.1369444725106925e-05, "loss": 0.464, "step": 13265 }, { "epoch": 0.47262946007089796, "grad_norm": 1.1705226079836042, "learning_rate": 1.1368301650850717e-05, "loss": 0.4627, "step": 13266 }, { "epoch": 0.4726650871973921, "grad_norm": 1.1388175805188216, "learning_rate": 1.1367158558374614e-05, "loss": 0.4251, "step": 13267 }, { "epoch": 0.4727007143238862, "grad_norm": 1.3746713703533064, "learning_rate": 1.1366015447693837e-05, "loss": 0.554, "step": 13268 }, { "epoch": 0.4727363414503803, "grad_norm": 1.0417722699667302, "learning_rate": 1.1364872318823612e-05, "loss": 0.4656, "step": 13269 }, { "epoch": 0.4727719685768744, "grad_norm": 1.2757219805876425, "learning_rate": 1.1363729171779161e-05, "loss": 0.4642, "step": 13270 }, { "epoch": 0.47280759570336856, "grad_norm": 1.0995401104202263, "learning_rate": 1.1362586006575698e-05, "loss": 0.4648, "step": 13271 }, { "epoch": 0.47284322282986263, "grad_norm": 1.1311998679681727, "learning_rate": 1.1361442823228453e-05, "loss": 0.486, "step": 13272 }, { "epoch": 0.47287884995635676, "grad_norm": 1.613090040351862, "learning_rate": 1.1360299621752644e-05, "loss": 0.5596, "step": 13273 }, { "epoch": 0.4729144770828509, "grad_norm": 1.0832819210924027, "learning_rate": 1.1359156402163496e-05, "loss": 0.4926, "step": 13274 }, { "epoch": 0.47295010420934497, "grad_norm": 1.4796411415811213, "learning_rate": 1.1358013164476227e-05, "loss": 0.5698, "step": 13275 }, { "epoch": 0.4729857313358391, "grad_norm": 1.1530352791200078, "learning_rate": 1.1356869908706068e-05, "loss": 0.461, "step": 13276 }, { "epoch": 0.47302135846233323, "grad_norm": 1.068013610313762, "learning_rate": 1.1355726634868236e-05, "loss": 0.4654, "step": 13277 }, { "epoch": 0.4730569855888273, "grad_norm": 1.8427576322168195, "learning_rate": 1.1354583342977953e-05, "loss": 0.5854, "step": 13278 }, { "epoch": 0.47309261271532144, "grad_norm": 1.152529505064698, "learning_rate": 1.1353440033050446e-05, "loss": 0.4878, "step": 13279 }, { "epoch": 0.47312823984181557, "grad_norm": 2.034032107327342, "learning_rate": 1.1352296705100944e-05, "loss": 0.561, "step": 13280 }, { "epoch": 0.4731638669683097, "grad_norm": 1.3692865429448116, "learning_rate": 1.1351153359144664e-05, "loss": 0.6113, "step": 13281 }, { "epoch": 0.4731994940948038, "grad_norm": 1.61575620574061, "learning_rate": 1.1350009995196834e-05, "loss": 0.4204, "step": 13282 }, { "epoch": 0.4732351212212979, "grad_norm": 1.0478581562196387, "learning_rate": 1.1348866613272675e-05, "loss": 0.4935, "step": 13283 }, { "epoch": 0.47327074834779204, "grad_norm": 1.5642020714428553, "learning_rate": 1.1347723213387416e-05, "loss": 0.6001, "step": 13284 }, { "epoch": 0.4733063754742861, "grad_norm": 1.9638329910089785, "learning_rate": 1.134657979555628e-05, "loss": 0.6639, "step": 13285 }, { "epoch": 0.47334200260078024, "grad_norm": 1.0935046101694614, "learning_rate": 1.134543635979449e-05, "loss": 0.4464, "step": 13286 }, { "epoch": 0.4733776297272744, "grad_norm": 2.1726075869271844, "learning_rate": 1.1344292906117279e-05, "loss": 0.5363, "step": 13287 }, { "epoch": 0.47341325685376845, "grad_norm": 1.3851294180280636, "learning_rate": 1.1343149434539866e-05, "loss": 0.5698, "step": 13288 }, { "epoch": 0.4734488839802626, "grad_norm": 2.2724411816900623, "learning_rate": 1.1342005945077481e-05, "loss": 0.5847, "step": 13289 }, { "epoch": 0.4734845111067567, "grad_norm": 1.4497653548204776, "learning_rate": 1.1340862437745344e-05, "loss": 0.5901, "step": 13290 }, { "epoch": 0.4735201382332508, "grad_norm": 1.0995671648823608, "learning_rate": 1.1339718912558691e-05, "loss": 0.4358, "step": 13291 }, { "epoch": 0.4735557653597449, "grad_norm": 1.0353582984980725, "learning_rate": 1.1338575369532743e-05, "loss": 0.4125, "step": 13292 }, { "epoch": 0.47359139248623905, "grad_norm": 1.1096592200433737, "learning_rate": 1.133743180868273e-05, "loss": 0.4478, "step": 13293 }, { "epoch": 0.4736270196127331, "grad_norm": 1.5687555585187136, "learning_rate": 1.1336288230023875e-05, "loss": 0.5833, "step": 13294 }, { "epoch": 0.47366264673922726, "grad_norm": 1.7725728744793663, "learning_rate": 1.1335144633571412e-05, "loss": 0.5345, "step": 13295 }, { "epoch": 0.4736982738657214, "grad_norm": 1.4635811045002853, "learning_rate": 1.1334001019340562e-05, "loss": 0.5559, "step": 13296 }, { "epoch": 0.47373390099221546, "grad_norm": 1.4816976648835722, "learning_rate": 1.1332857387346554e-05, "loss": 0.556, "step": 13297 }, { "epoch": 0.4737695281187096, "grad_norm": 1.0958861494183034, "learning_rate": 1.1331713737604621e-05, "loss": 0.4329, "step": 13298 }, { "epoch": 0.4738051552452037, "grad_norm": 1.1219629763786436, "learning_rate": 1.1330570070129989e-05, "loss": 0.4522, "step": 13299 }, { "epoch": 0.4738407823716978, "grad_norm": 1.1638633779390246, "learning_rate": 1.1329426384937887e-05, "loss": 0.4468, "step": 13300 }, { "epoch": 0.47387640949819193, "grad_norm": 1.642386231884606, "learning_rate": 1.132828268204354e-05, "loss": 0.5796, "step": 13301 }, { "epoch": 0.47391203662468606, "grad_norm": 1.5030886311999643, "learning_rate": 1.1327138961462183e-05, "loss": 0.5908, "step": 13302 }, { "epoch": 0.47394766375118014, "grad_norm": 1.2014413848570238, "learning_rate": 1.1325995223209042e-05, "loss": 0.5158, "step": 13303 }, { "epoch": 0.47398329087767427, "grad_norm": 1.8403925559746097, "learning_rate": 1.132485146729935e-05, "loss": 0.6417, "step": 13304 }, { "epoch": 0.4740189180041684, "grad_norm": 1.759636542607933, "learning_rate": 1.1323707693748335e-05, "loss": 0.5767, "step": 13305 }, { "epoch": 0.4740545451306625, "grad_norm": 1.341955667371225, "learning_rate": 1.1322563902571227e-05, "loss": 0.5406, "step": 13306 }, { "epoch": 0.4740901722571566, "grad_norm": 1.7339551808616673, "learning_rate": 1.1321420093783254e-05, "loss": 0.6146, "step": 13307 }, { "epoch": 0.47412579938365074, "grad_norm": 1.2762062293821417, "learning_rate": 1.132027626739965e-05, "loss": 0.5475, "step": 13308 }, { "epoch": 0.4741614265101448, "grad_norm": 1.4447853828949722, "learning_rate": 1.1319132423435644e-05, "loss": 0.5158, "step": 13309 }, { "epoch": 0.47419705363663894, "grad_norm": 1.226346164875145, "learning_rate": 1.1317988561906468e-05, "loss": 0.4232, "step": 13310 }, { "epoch": 0.4742326807631331, "grad_norm": 2.8236491233869674, "learning_rate": 1.1316844682827355e-05, "loss": 0.5707, "step": 13311 }, { "epoch": 0.47426830788962715, "grad_norm": 1.740204033560387, "learning_rate": 1.1315700786213531e-05, "loss": 0.5084, "step": 13312 }, { "epoch": 0.4743039350161213, "grad_norm": 1.153613726356393, "learning_rate": 1.1314556872080231e-05, "loss": 0.4842, "step": 13313 }, { "epoch": 0.4743395621426154, "grad_norm": 1.6202086087441931, "learning_rate": 1.1313412940442689e-05, "loss": 0.6197, "step": 13314 }, { "epoch": 0.4743751892691095, "grad_norm": 1.4750577384043122, "learning_rate": 1.1312268991316137e-05, "loss": 0.57, "step": 13315 }, { "epoch": 0.4744108163956036, "grad_norm": 1.2469938073730988, "learning_rate": 1.1311125024715806e-05, "loss": 0.4278, "step": 13316 }, { "epoch": 0.47444644352209775, "grad_norm": 1.2204647490246956, "learning_rate": 1.130998104065693e-05, "loss": 0.4463, "step": 13317 }, { "epoch": 0.4744820706485918, "grad_norm": 1.7789959107769993, "learning_rate": 1.1308837039154739e-05, "loss": 0.4954, "step": 13318 }, { "epoch": 0.47451769777508596, "grad_norm": 1.0658948387867555, "learning_rate": 1.1307693020224468e-05, "loss": 0.4659, "step": 13319 }, { "epoch": 0.4745533249015801, "grad_norm": 1.8953376728762965, "learning_rate": 1.130654898388135e-05, "loss": 0.6226, "step": 13320 }, { "epoch": 0.47458895202807416, "grad_norm": 1.4475131366670078, "learning_rate": 1.130540493014062e-05, "loss": 0.5778, "step": 13321 }, { "epoch": 0.4746245791545683, "grad_norm": 1.7082077539700404, "learning_rate": 1.1304260859017511e-05, "loss": 0.4995, "step": 13322 }, { "epoch": 0.4746602062810624, "grad_norm": 1.104002886915707, "learning_rate": 1.1303116770527257e-05, "loss": 0.4359, "step": 13323 }, { "epoch": 0.4746958334075565, "grad_norm": 3.149734320046385, "learning_rate": 1.130197266468509e-05, "loss": 0.5702, "step": 13324 }, { "epoch": 0.47473146053405063, "grad_norm": 1.83511279598059, "learning_rate": 1.1300828541506247e-05, "loss": 0.5674, "step": 13325 }, { "epoch": 0.47476708766054476, "grad_norm": 1.1002426292359395, "learning_rate": 1.1299684401005967e-05, "loss": 0.4372, "step": 13326 }, { "epoch": 0.47480271478703884, "grad_norm": 1.834626458127904, "learning_rate": 1.1298540243199476e-05, "loss": 0.5214, "step": 13327 }, { "epoch": 0.47483834191353297, "grad_norm": 1.9351845862139017, "learning_rate": 1.1297396068102019e-05, "loss": 0.5868, "step": 13328 }, { "epoch": 0.4748739690400271, "grad_norm": 1.6145057098227944, "learning_rate": 1.1296251875728822e-05, "loss": 0.5764, "step": 13329 }, { "epoch": 0.4749095961665212, "grad_norm": 1.8510631100332655, "learning_rate": 1.1295107666095131e-05, "loss": 0.5654, "step": 13330 }, { "epoch": 0.4749452232930153, "grad_norm": 1.1375098816432203, "learning_rate": 1.129396343921617e-05, "loss": 0.4693, "step": 13331 }, { "epoch": 0.47498085041950944, "grad_norm": 1.0654827422005684, "learning_rate": 1.1292819195107185e-05, "loss": 0.4754, "step": 13332 }, { "epoch": 0.4750164775460035, "grad_norm": 1.2581239841839211, "learning_rate": 1.1291674933783408e-05, "loss": 0.4449, "step": 13333 }, { "epoch": 0.47505210467249764, "grad_norm": 1.1081463531838018, "learning_rate": 1.1290530655260075e-05, "loss": 0.4735, "step": 13334 }, { "epoch": 0.4750877317989918, "grad_norm": 1.0966073582196358, "learning_rate": 1.1289386359552426e-05, "loss": 0.4812, "step": 13335 }, { "epoch": 0.47512335892548585, "grad_norm": 1.5133865034539886, "learning_rate": 1.1288242046675698e-05, "loss": 0.4667, "step": 13336 }, { "epoch": 0.47515898605198, "grad_norm": 1.0902270000930698, "learning_rate": 1.1287097716645121e-05, "loss": 0.4646, "step": 13337 }, { "epoch": 0.4751946131784741, "grad_norm": 1.690093326985989, "learning_rate": 1.128595336947594e-05, "loss": 0.5425, "step": 13338 }, { "epoch": 0.4752302403049682, "grad_norm": 1.2532340618749742, "learning_rate": 1.1284809005183395e-05, "loss": 0.4539, "step": 13339 }, { "epoch": 0.4752658674314623, "grad_norm": 1.0183619855202548, "learning_rate": 1.1283664623782718e-05, "loss": 0.391, "step": 13340 }, { "epoch": 0.47530149455795645, "grad_norm": 1.7308543566993542, "learning_rate": 1.1282520225289152e-05, "loss": 0.5549, "step": 13341 }, { "epoch": 0.4753371216844505, "grad_norm": 1.2027455973931307, "learning_rate": 1.1281375809717927e-05, "loss": 0.4888, "step": 13342 }, { "epoch": 0.47537274881094466, "grad_norm": 1.2146357511965782, "learning_rate": 1.128023137708429e-05, "loss": 0.5062, "step": 13343 }, { "epoch": 0.4754083759374388, "grad_norm": 2.072744175427442, "learning_rate": 1.127908692740348e-05, "loss": 0.5636, "step": 13344 }, { "epoch": 0.47544400306393286, "grad_norm": 1.0668509482810808, "learning_rate": 1.1277942460690731e-05, "loss": 0.4647, "step": 13345 }, { "epoch": 0.475479630190427, "grad_norm": 1.445688390904227, "learning_rate": 1.1276797976961285e-05, "loss": 0.5276, "step": 13346 }, { "epoch": 0.4755152573169211, "grad_norm": 1.1172354146700767, "learning_rate": 1.127565347623038e-05, "loss": 0.4668, "step": 13347 }, { "epoch": 0.4755508844434152, "grad_norm": 1.0560004879267688, "learning_rate": 1.127450895851326e-05, "loss": 0.4359, "step": 13348 }, { "epoch": 0.47558651156990933, "grad_norm": 1.624437349563424, "learning_rate": 1.1273364423825162e-05, "loss": 0.6108, "step": 13349 }, { "epoch": 0.47562213869640346, "grad_norm": 1.0845669382950927, "learning_rate": 1.1272219872181322e-05, "loss": 0.4752, "step": 13350 }, { "epoch": 0.47565776582289754, "grad_norm": 1.6303923013617077, "learning_rate": 1.1271075303596992e-05, "loss": 0.6263, "step": 13351 }, { "epoch": 0.47569339294939167, "grad_norm": 1.1967139237574411, "learning_rate": 1.1269930718087403e-05, "loss": 0.4626, "step": 13352 }, { "epoch": 0.4757290200758858, "grad_norm": 2.2165868958793435, "learning_rate": 1.1268786115667798e-05, "loss": 0.6302, "step": 13353 }, { "epoch": 0.4757646472023799, "grad_norm": 1.2462821867893605, "learning_rate": 1.126764149635342e-05, "loss": 0.475, "step": 13354 }, { "epoch": 0.475800274328874, "grad_norm": 1.862455572852709, "learning_rate": 1.1266496860159512e-05, "loss": 0.5229, "step": 13355 }, { "epoch": 0.47583590145536814, "grad_norm": 1.406264849710652, "learning_rate": 1.1265352207101312e-05, "loss": 0.5763, "step": 13356 }, { "epoch": 0.4758715285818622, "grad_norm": 1.5477012656703202, "learning_rate": 1.1264207537194059e-05, "loss": 0.5405, "step": 13357 }, { "epoch": 0.47590715570835634, "grad_norm": 1.4950473194777016, "learning_rate": 1.1263062850453002e-05, "loss": 0.5774, "step": 13358 }, { "epoch": 0.4759427828348505, "grad_norm": 1.4961561909536796, "learning_rate": 1.1261918146893381e-05, "loss": 0.6003, "step": 13359 }, { "epoch": 0.47597840996134455, "grad_norm": 1.5691545535154414, "learning_rate": 1.1260773426530436e-05, "loss": 0.5416, "step": 13360 }, { "epoch": 0.4760140370878387, "grad_norm": 2.334141813618228, "learning_rate": 1.1259628689379412e-05, "loss": 0.5498, "step": 13361 }, { "epoch": 0.4760496642143328, "grad_norm": 1.5674791280468494, "learning_rate": 1.1258483935455551e-05, "loss": 0.5513, "step": 13362 }, { "epoch": 0.4760852913408269, "grad_norm": 1.0306860930311754, "learning_rate": 1.12573391647741e-05, "loss": 0.4206, "step": 13363 }, { "epoch": 0.476120918467321, "grad_norm": 1.4326546216331284, "learning_rate": 1.1256194377350296e-05, "loss": 0.5708, "step": 13364 }, { "epoch": 0.47615654559381515, "grad_norm": 1.6957506308570651, "learning_rate": 1.1255049573199386e-05, "loss": 0.59, "step": 13365 }, { "epoch": 0.4761921727203092, "grad_norm": 1.209605303857851, "learning_rate": 1.1253904752336615e-05, "loss": 0.4328, "step": 13366 }, { "epoch": 0.47622779984680336, "grad_norm": 1.5820713669085364, "learning_rate": 1.1252759914777224e-05, "loss": 0.612, "step": 13367 }, { "epoch": 0.4762634269732975, "grad_norm": 1.1244028274105178, "learning_rate": 1.125161506053646e-05, "loss": 0.4361, "step": 13368 }, { "epoch": 0.47629905409979156, "grad_norm": 1.0537043308311225, "learning_rate": 1.1250470189629567e-05, "loss": 0.4998, "step": 13369 }, { "epoch": 0.4763346812262857, "grad_norm": 1.0668465674851564, "learning_rate": 1.1249325302071793e-05, "loss": 0.4393, "step": 13370 }, { "epoch": 0.4763703083527798, "grad_norm": 1.1371574443156247, "learning_rate": 1.1248180397878372e-05, "loss": 0.4795, "step": 13371 }, { "epoch": 0.4764059354792739, "grad_norm": 1.665611410703216, "learning_rate": 1.124703547706456e-05, "loss": 0.6027, "step": 13372 }, { "epoch": 0.47644156260576803, "grad_norm": 1.292453192367605, "learning_rate": 1.1245890539645598e-05, "loss": 0.5275, "step": 13373 }, { "epoch": 0.47647718973226216, "grad_norm": 1.0287768085685356, "learning_rate": 1.1244745585636736e-05, "loss": 0.4199, "step": 13374 }, { "epoch": 0.47651281685875624, "grad_norm": 1.302249640040375, "learning_rate": 1.1243600615053217e-05, "loss": 0.5235, "step": 13375 }, { "epoch": 0.47654844398525037, "grad_norm": 1.1364778209206015, "learning_rate": 1.124245562791028e-05, "loss": 0.4798, "step": 13376 }, { "epoch": 0.4765840711117445, "grad_norm": 1.0682449288100613, "learning_rate": 1.1241310624223183e-05, "loss": 0.4847, "step": 13377 }, { "epoch": 0.4766196982382386, "grad_norm": 1.0791738084118159, "learning_rate": 1.1240165604007167e-05, "loss": 0.5217, "step": 13378 }, { "epoch": 0.4766553253647327, "grad_norm": 1.337256373907861, "learning_rate": 1.1239020567277479e-05, "loss": 0.6357, "step": 13379 }, { "epoch": 0.47669095249122684, "grad_norm": 1.0171845943167959, "learning_rate": 1.1237875514049365e-05, "loss": 0.4708, "step": 13380 }, { "epoch": 0.4767265796177209, "grad_norm": 1.5378677937326666, "learning_rate": 1.1236730444338076e-05, "loss": 0.6427, "step": 13381 }, { "epoch": 0.47676220674421504, "grad_norm": 1.1959979465506612, "learning_rate": 1.1235585358158854e-05, "loss": 0.4617, "step": 13382 }, { "epoch": 0.4767978338707092, "grad_norm": 1.060380750044065, "learning_rate": 1.1234440255526948e-05, "loss": 0.482, "step": 13383 }, { "epoch": 0.47683346099720325, "grad_norm": 2.2816898064730258, "learning_rate": 1.1233295136457609e-05, "loss": 0.5656, "step": 13384 }, { "epoch": 0.4768690881236974, "grad_norm": 1.8498950982959488, "learning_rate": 1.1232150000966084e-05, "loss": 0.505, "step": 13385 }, { "epoch": 0.4769047152501915, "grad_norm": 1.5330054701331521, "learning_rate": 1.1231004849067621e-05, "loss": 0.5381, "step": 13386 }, { "epoch": 0.4769403423766856, "grad_norm": 1.5294013064613916, "learning_rate": 1.1229859680777465e-05, "loss": 0.5187, "step": 13387 }, { "epoch": 0.4769759695031797, "grad_norm": 1.6656245597893624, "learning_rate": 1.1228714496110871e-05, "loss": 0.5342, "step": 13388 }, { "epoch": 0.47701159662967385, "grad_norm": 2.5409805853978265, "learning_rate": 1.1227569295083083e-05, "loss": 0.5587, "step": 13389 }, { "epoch": 0.4770472237561679, "grad_norm": 1.6544983190530635, "learning_rate": 1.1226424077709353e-05, "loss": 0.53, "step": 13390 }, { "epoch": 0.47708285088266206, "grad_norm": 1.5508937516958512, "learning_rate": 1.1225278844004927e-05, "loss": 0.6125, "step": 13391 }, { "epoch": 0.4771184780091562, "grad_norm": 1.3654368707627333, "learning_rate": 1.1224133593985061e-05, "loss": 0.4486, "step": 13392 }, { "epoch": 0.47715410513565026, "grad_norm": 1.218425884541049, "learning_rate": 1.1222988327664997e-05, "loss": 0.4235, "step": 13393 }, { "epoch": 0.4771897322621444, "grad_norm": 1.2140196319770677, "learning_rate": 1.1221843045059989e-05, "loss": 0.4333, "step": 13394 }, { "epoch": 0.4772253593886385, "grad_norm": 1.1806696118029854, "learning_rate": 1.1220697746185285e-05, "loss": 0.4513, "step": 13395 }, { "epoch": 0.4772609865151326, "grad_norm": 1.7269917445471723, "learning_rate": 1.121955243105614e-05, "loss": 0.6388, "step": 13396 }, { "epoch": 0.47729661364162673, "grad_norm": 1.4767595862170029, "learning_rate": 1.1218407099687802e-05, "loss": 0.5655, "step": 13397 }, { "epoch": 0.47733224076812086, "grad_norm": 1.182882205697169, "learning_rate": 1.1217261752095518e-05, "loss": 0.4674, "step": 13398 }, { "epoch": 0.47736786789461494, "grad_norm": 1.1864650877536451, "learning_rate": 1.121611638829455e-05, "loss": 0.4852, "step": 13399 }, { "epoch": 0.47740349502110907, "grad_norm": 1.2894586981999037, "learning_rate": 1.1214971008300138e-05, "loss": 0.4817, "step": 13400 }, { "epoch": 0.4774391221476032, "grad_norm": 1.168728162543127, "learning_rate": 1.1213825612127537e-05, "loss": 0.4669, "step": 13401 }, { "epoch": 0.4774747492740973, "grad_norm": 1.098780004288188, "learning_rate": 1.1212680199792001e-05, "loss": 0.4844, "step": 13402 }, { "epoch": 0.4775103764005914, "grad_norm": 2.4726851884495975, "learning_rate": 1.121153477130878e-05, "loss": 0.6262, "step": 13403 }, { "epoch": 0.47754600352708554, "grad_norm": 1.0916197466955195, "learning_rate": 1.121038932669313e-05, "loss": 0.4456, "step": 13404 }, { "epoch": 0.4775816306535796, "grad_norm": 1.4317532887849502, "learning_rate": 1.1209243865960295e-05, "loss": 0.5052, "step": 13405 }, { "epoch": 0.47761725778007375, "grad_norm": 1.3425007921899819, "learning_rate": 1.1208098389125533e-05, "loss": 0.4878, "step": 13406 }, { "epoch": 0.4776528849065679, "grad_norm": 1.325136888158206, "learning_rate": 1.12069528962041e-05, "loss": 0.5857, "step": 13407 }, { "epoch": 0.47768851203306195, "grad_norm": 1.2568056248404007, "learning_rate": 1.1205807387211245e-05, "loss": 0.446, "step": 13408 }, { "epoch": 0.4777241391595561, "grad_norm": 1.0840479904884992, "learning_rate": 1.1204661862162219e-05, "loss": 0.4159, "step": 13409 }, { "epoch": 0.4777597662860502, "grad_norm": 1.7127619381624455, "learning_rate": 1.1203516321072281e-05, "loss": 0.549, "step": 13410 }, { "epoch": 0.4777953934125443, "grad_norm": 1.0427520474273546, "learning_rate": 1.120237076395668e-05, "loss": 0.446, "step": 13411 }, { "epoch": 0.4778310205390384, "grad_norm": 1.6229631592735976, "learning_rate": 1.1201225190830676e-05, "loss": 0.5452, "step": 13412 }, { "epoch": 0.47786664766553255, "grad_norm": 1.1544106428881047, "learning_rate": 1.1200079601709513e-05, "loss": 0.4725, "step": 13413 }, { "epoch": 0.4779022747920266, "grad_norm": 1.6235205153732992, "learning_rate": 1.1198933996608456e-05, "loss": 0.5545, "step": 13414 }, { "epoch": 0.47793790191852076, "grad_norm": 1.092521921107512, "learning_rate": 1.1197788375542754e-05, "loss": 0.4714, "step": 13415 }, { "epoch": 0.4779735290450149, "grad_norm": 1.411271149869775, "learning_rate": 1.119664273852766e-05, "loss": 0.5218, "step": 13416 }, { "epoch": 0.47800915617150896, "grad_norm": 1.0424690425075183, "learning_rate": 1.119549708557843e-05, "loss": 0.4039, "step": 13417 }, { "epoch": 0.4780447832980031, "grad_norm": 2.2573088510754156, "learning_rate": 1.1194351416710325e-05, "loss": 0.4921, "step": 13418 }, { "epoch": 0.4780804104244972, "grad_norm": 1.1320852612903674, "learning_rate": 1.1193205731938597e-05, "loss": 0.4322, "step": 13419 }, { "epoch": 0.4781160375509913, "grad_norm": 1.1217743802398907, "learning_rate": 1.1192060031278497e-05, "loss": 0.4241, "step": 13420 }, { "epoch": 0.47815166467748543, "grad_norm": 1.2656991068923296, "learning_rate": 1.1190914314745286e-05, "loss": 0.4837, "step": 13421 }, { "epoch": 0.47818729180397956, "grad_norm": 1.8159234151655141, "learning_rate": 1.118976858235422e-05, "loss": 0.5849, "step": 13422 }, { "epoch": 0.47822291893047364, "grad_norm": 1.084499765635335, "learning_rate": 1.1188622834120553e-05, "loss": 0.493, "step": 13423 }, { "epoch": 0.47825854605696777, "grad_norm": 1.1628206042712559, "learning_rate": 1.118747707005954e-05, "loss": 0.3886, "step": 13424 }, { "epoch": 0.4782941731834619, "grad_norm": 1.1835845457934806, "learning_rate": 1.1186331290186444e-05, "loss": 0.5071, "step": 13425 }, { "epoch": 0.478329800309956, "grad_norm": 1.4528517406053303, "learning_rate": 1.1185185494516515e-05, "loss": 0.6284, "step": 13426 }, { "epoch": 0.4783654274364501, "grad_norm": 1.775950939029063, "learning_rate": 1.1184039683065014e-05, "loss": 0.5523, "step": 13427 }, { "epoch": 0.47840105456294424, "grad_norm": 1.200924305563294, "learning_rate": 1.1182893855847193e-05, "loss": 0.4295, "step": 13428 }, { "epoch": 0.4784366816894383, "grad_norm": 1.1760000588199417, "learning_rate": 1.1181748012878316e-05, "loss": 0.4106, "step": 13429 }, { "epoch": 0.47847230881593245, "grad_norm": 1.3030618629705328, "learning_rate": 1.1180602154173638e-05, "loss": 0.513, "step": 13430 }, { "epoch": 0.4785079359424266, "grad_norm": 1.4586509152541693, "learning_rate": 1.1179456279748418e-05, "loss": 0.5974, "step": 13431 }, { "epoch": 0.47854356306892065, "grad_norm": 1.1609741144596488, "learning_rate": 1.1178310389617912e-05, "loss": 0.4213, "step": 13432 }, { "epoch": 0.4785791901954148, "grad_norm": 1.3397642072331128, "learning_rate": 1.1177164483797381e-05, "loss": 0.4667, "step": 13433 }, { "epoch": 0.4786148173219089, "grad_norm": 1.5193189406653567, "learning_rate": 1.1176018562302082e-05, "loss": 0.5083, "step": 13434 }, { "epoch": 0.478650444448403, "grad_norm": 1.1480258633484548, "learning_rate": 1.1174872625147273e-05, "loss": 0.5123, "step": 13435 }, { "epoch": 0.4786860715748971, "grad_norm": 1.5091269860536192, "learning_rate": 1.1173726672348215e-05, "loss": 0.6181, "step": 13436 }, { "epoch": 0.47872169870139125, "grad_norm": 1.2213448123732773, "learning_rate": 1.1172580703920165e-05, "loss": 0.5005, "step": 13437 }, { "epoch": 0.4787573258278853, "grad_norm": 1.3723226733782583, "learning_rate": 1.1171434719878385e-05, "loss": 0.5636, "step": 13438 }, { "epoch": 0.47879295295437946, "grad_norm": 1.3509411544092391, "learning_rate": 1.117028872023813e-05, "loss": 0.5767, "step": 13439 }, { "epoch": 0.4788285800808736, "grad_norm": 1.1852616069451951, "learning_rate": 1.1169142705014661e-05, "loss": 0.4537, "step": 13440 }, { "epoch": 0.47886420720736766, "grad_norm": 1.3581541563622468, "learning_rate": 1.1167996674223245e-05, "loss": 0.5231, "step": 13441 }, { "epoch": 0.4788998343338618, "grad_norm": 1.8051398353092138, "learning_rate": 1.1166850627879135e-05, "loss": 0.5539, "step": 13442 }, { "epoch": 0.4789354614603559, "grad_norm": 1.5373792346110566, "learning_rate": 1.1165704565997593e-05, "loss": 0.5907, "step": 13443 }, { "epoch": 0.47897108858685, "grad_norm": 1.063591258877975, "learning_rate": 1.116455848859388e-05, "loss": 0.4665, "step": 13444 }, { "epoch": 0.47900671571334413, "grad_norm": 1.0247001712148962, "learning_rate": 1.1163412395683257e-05, "loss": 0.4528, "step": 13445 }, { "epoch": 0.47904234283983826, "grad_norm": 1.52097829607267, "learning_rate": 1.1162266287280985e-05, "loss": 0.5561, "step": 13446 }, { "epoch": 0.47907796996633234, "grad_norm": 1.3338289078278962, "learning_rate": 1.1161120163402324e-05, "loss": 0.4436, "step": 13447 }, { "epoch": 0.47911359709282647, "grad_norm": 1.5825242441102532, "learning_rate": 1.115997402406254e-05, "loss": 0.6212, "step": 13448 }, { "epoch": 0.4791492242193206, "grad_norm": 1.0716331652641924, "learning_rate": 1.1158827869276887e-05, "loss": 0.4104, "step": 13449 }, { "epoch": 0.4791848513458147, "grad_norm": 1.6581368096317826, "learning_rate": 1.115768169906063e-05, "loss": 0.572, "step": 13450 }, { "epoch": 0.4792204784723088, "grad_norm": 1.1037048882087053, "learning_rate": 1.1156535513429034e-05, "loss": 0.5047, "step": 13451 }, { "epoch": 0.47925610559880294, "grad_norm": 1.3292824137833692, "learning_rate": 1.115538931239736e-05, "loss": 0.5694, "step": 13452 }, { "epoch": 0.479291732725297, "grad_norm": 1.6840778520014799, "learning_rate": 1.1154243095980868e-05, "loss": 0.5176, "step": 13453 }, { "epoch": 0.47932735985179115, "grad_norm": 1.0985886219691796, "learning_rate": 1.1153096864194823e-05, "loss": 0.4315, "step": 13454 }, { "epoch": 0.4793629869782853, "grad_norm": 1.2122883455587299, "learning_rate": 1.1151950617054487e-05, "loss": 0.4652, "step": 13455 }, { "epoch": 0.47939861410477935, "grad_norm": 1.1883801859080387, "learning_rate": 1.1150804354575126e-05, "loss": 0.4501, "step": 13456 }, { "epoch": 0.4794342412312735, "grad_norm": 1.6057190957960068, "learning_rate": 1.1149658076772001e-05, "loss": 0.555, "step": 13457 }, { "epoch": 0.4794698683577676, "grad_norm": 1.3745031240412748, "learning_rate": 1.114851178366037e-05, "loss": 0.5229, "step": 13458 }, { "epoch": 0.4795054954842617, "grad_norm": 1.0925326058496705, "learning_rate": 1.1147365475255505e-05, "loss": 0.377, "step": 13459 }, { "epoch": 0.4795411226107558, "grad_norm": 1.6420319623507162, "learning_rate": 1.1146219151572669e-05, "loss": 0.6133, "step": 13460 }, { "epoch": 0.47957674973724995, "grad_norm": 1.4608024822032728, "learning_rate": 1.1145072812627117e-05, "loss": 0.4969, "step": 13461 }, { "epoch": 0.479612376863744, "grad_norm": 1.7480195114052077, "learning_rate": 1.1143926458434127e-05, "loss": 0.5666, "step": 13462 }, { "epoch": 0.47964800399023816, "grad_norm": 1.3580680240791703, "learning_rate": 1.1142780089008953e-05, "loss": 0.5116, "step": 13463 }, { "epoch": 0.4796836311167323, "grad_norm": 1.6091927794937937, "learning_rate": 1.1141633704366865e-05, "loss": 0.5504, "step": 13464 }, { "epoch": 0.47971925824322637, "grad_norm": 1.7087667020535589, "learning_rate": 1.1140487304523126e-05, "loss": 0.682, "step": 13465 }, { "epoch": 0.4797548853697205, "grad_norm": 1.4394868607959632, "learning_rate": 1.1139340889493e-05, "loss": 0.5507, "step": 13466 }, { "epoch": 0.4797905124962146, "grad_norm": 1.1499194725194344, "learning_rate": 1.1138194459291755e-05, "loss": 0.4603, "step": 13467 }, { "epoch": 0.4798261396227087, "grad_norm": 1.516924593854471, "learning_rate": 1.1137048013934657e-05, "loss": 0.4605, "step": 13468 }, { "epoch": 0.47986176674920283, "grad_norm": 1.3275968282806427, "learning_rate": 1.1135901553436964e-05, "loss": 0.6294, "step": 13469 }, { "epoch": 0.47989739387569696, "grad_norm": 1.0949739694351177, "learning_rate": 1.1134755077813955e-05, "loss": 0.4159, "step": 13470 }, { "epoch": 0.4799330210021911, "grad_norm": 1.1108138598941035, "learning_rate": 1.1133608587080887e-05, "loss": 0.4543, "step": 13471 }, { "epoch": 0.47996864812868517, "grad_norm": 1.3402190561256113, "learning_rate": 1.1132462081253028e-05, "loss": 0.5799, "step": 13472 }, { "epoch": 0.4800042752551793, "grad_norm": 1.1350902156787308, "learning_rate": 1.1131315560345641e-05, "loss": 0.5091, "step": 13473 }, { "epoch": 0.48003990238167343, "grad_norm": 1.557693755321773, "learning_rate": 1.1130169024374002e-05, "loss": 0.4778, "step": 13474 }, { "epoch": 0.4800755295081675, "grad_norm": 1.4150421737366248, "learning_rate": 1.1129022473353368e-05, "loss": 0.4569, "step": 13475 }, { "epoch": 0.48011115663466164, "grad_norm": 1.7820187616163479, "learning_rate": 1.1127875907299013e-05, "loss": 0.5515, "step": 13476 }, { "epoch": 0.48014678376115577, "grad_norm": 1.1528547817085488, "learning_rate": 1.11267293262262e-05, "loss": 0.4054, "step": 13477 }, { "epoch": 0.48018241088764985, "grad_norm": 1.256914494334983, "learning_rate": 1.1125582730150201e-05, "loss": 0.5889, "step": 13478 }, { "epoch": 0.480218038014144, "grad_norm": 1.0044042324710625, "learning_rate": 1.1124436119086281e-05, "loss": 0.4222, "step": 13479 }, { "epoch": 0.4802536651406381, "grad_norm": 1.1734185492442506, "learning_rate": 1.1123289493049707e-05, "loss": 0.5052, "step": 13480 }, { "epoch": 0.4802892922671322, "grad_norm": 1.0164976018375014, "learning_rate": 1.112214285205575e-05, "loss": 0.5032, "step": 13481 }, { "epoch": 0.4803249193936263, "grad_norm": 1.1703063909897617, "learning_rate": 1.1120996196119676e-05, "loss": 0.4534, "step": 13482 }, { "epoch": 0.48036054652012045, "grad_norm": 1.3311589790193503, "learning_rate": 1.1119849525256754e-05, "loss": 0.455, "step": 13483 }, { "epoch": 0.4803961736466145, "grad_norm": 1.0548235340834289, "learning_rate": 1.1118702839482251e-05, "loss": 0.4489, "step": 13484 }, { "epoch": 0.48043180077310865, "grad_norm": 1.7104764571962077, "learning_rate": 1.111755613881144e-05, "loss": 0.5482, "step": 13485 }, { "epoch": 0.4804674278996028, "grad_norm": 1.5691002285002693, "learning_rate": 1.1116409423259586e-05, "loss": 0.6183, "step": 13486 }, { "epoch": 0.48050305502609686, "grad_norm": 1.0444778780001727, "learning_rate": 1.111526269284196e-05, "loss": 0.4532, "step": 13487 }, { "epoch": 0.480538682152591, "grad_norm": 1.564742896952827, "learning_rate": 1.1114115947573834e-05, "loss": 0.5415, "step": 13488 }, { "epoch": 0.4805743092790851, "grad_norm": 1.0703422538262837, "learning_rate": 1.1112969187470475e-05, "loss": 0.4678, "step": 13489 }, { "epoch": 0.4806099364055792, "grad_norm": 1.1135381320617268, "learning_rate": 1.1111822412547152e-05, "loss": 0.4637, "step": 13490 }, { "epoch": 0.4806455635320733, "grad_norm": 1.345811179148382, "learning_rate": 1.1110675622819135e-05, "loss": 0.5718, "step": 13491 }, { "epoch": 0.48068119065856746, "grad_norm": 1.1315765812876033, "learning_rate": 1.11095288183017e-05, "loss": 0.4972, "step": 13492 }, { "epoch": 0.48071681778506153, "grad_norm": 1.4478844270930493, "learning_rate": 1.1108381999010111e-05, "loss": 0.55, "step": 13493 }, { "epoch": 0.48075244491155567, "grad_norm": 1.5012924634959404, "learning_rate": 1.1107235164959643e-05, "loss": 0.6291, "step": 13494 }, { "epoch": 0.4807880720380498, "grad_norm": 1.2125407121873695, "learning_rate": 1.1106088316165562e-05, "loss": 0.4327, "step": 13495 }, { "epoch": 0.48082369916454387, "grad_norm": 1.6290017186717933, "learning_rate": 1.1104941452643143e-05, "loss": 0.5599, "step": 13496 }, { "epoch": 0.480859326291038, "grad_norm": 1.037264197273333, "learning_rate": 1.1103794574407659e-05, "loss": 0.4202, "step": 13497 }, { "epoch": 0.48089495341753213, "grad_norm": 1.1717750828602802, "learning_rate": 1.1102647681474372e-05, "loss": 0.4589, "step": 13498 }, { "epoch": 0.4809305805440262, "grad_norm": 1.5857525296378003, "learning_rate": 1.1101500773858564e-05, "loss": 0.5013, "step": 13499 }, { "epoch": 0.48096620767052034, "grad_norm": 1.0702000502910887, "learning_rate": 1.1100353851575505e-05, "loss": 0.4393, "step": 13500 }, { "epoch": 0.48100183479701447, "grad_norm": 1.0883205812044487, "learning_rate": 1.1099206914640464e-05, "loss": 0.4868, "step": 13501 }, { "epoch": 0.48103746192350855, "grad_norm": 1.1319237978264882, "learning_rate": 1.1098059963068715e-05, "loss": 0.4566, "step": 13502 }, { "epoch": 0.4810730890500027, "grad_norm": 1.032692727347632, "learning_rate": 1.1096912996875529e-05, "loss": 0.463, "step": 13503 }, { "epoch": 0.4811087161764968, "grad_norm": 1.4658781378519332, "learning_rate": 1.1095766016076182e-05, "loss": 0.5943, "step": 13504 }, { "epoch": 0.4811443433029909, "grad_norm": 1.5239451771706243, "learning_rate": 1.1094619020685945e-05, "loss": 0.5647, "step": 13505 }, { "epoch": 0.481179970429485, "grad_norm": 1.7447010142170412, "learning_rate": 1.1093472010720086e-05, "loss": 0.4386, "step": 13506 }, { "epoch": 0.48121559755597915, "grad_norm": 1.8443913576871702, "learning_rate": 1.1092324986193885e-05, "loss": 0.4776, "step": 13507 }, { "epoch": 0.4812512246824732, "grad_norm": 1.0147790808455637, "learning_rate": 1.1091177947122616e-05, "loss": 0.4742, "step": 13508 }, { "epoch": 0.48128685180896735, "grad_norm": 1.403582600875863, "learning_rate": 1.1090030893521549e-05, "loss": 0.4934, "step": 13509 }, { "epoch": 0.4813224789354615, "grad_norm": 1.1768662963205165, "learning_rate": 1.1088883825405953e-05, "loss": 0.4745, "step": 13510 }, { "epoch": 0.48135810606195556, "grad_norm": 1.0439857264149397, "learning_rate": 1.1087736742791114e-05, "loss": 0.4436, "step": 13511 }, { "epoch": 0.4813937331884497, "grad_norm": 1.3782026604893907, "learning_rate": 1.1086589645692299e-05, "loss": 0.6135, "step": 13512 }, { "epoch": 0.4814293603149438, "grad_norm": 1.4528827875132513, "learning_rate": 1.1085442534124784e-05, "loss": 0.5064, "step": 13513 }, { "epoch": 0.4814649874414379, "grad_norm": 1.094695185515817, "learning_rate": 1.108429540810384e-05, "loss": 0.5056, "step": 13514 }, { "epoch": 0.48150061456793203, "grad_norm": 1.1056022350183234, "learning_rate": 1.1083148267644748e-05, "loss": 0.4261, "step": 13515 }, { "epoch": 0.48153624169442616, "grad_norm": 1.6716922302295778, "learning_rate": 1.1082001112762778e-05, "loss": 0.6197, "step": 13516 }, { "epoch": 0.48157186882092023, "grad_norm": 1.8166720504233709, "learning_rate": 1.1080853943473207e-05, "loss": 0.5642, "step": 13517 }, { "epoch": 0.48160749594741437, "grad_norm": 0.9516663919882349, "learning_rate": 1.1079706759791311e-05, "loss": 0.4761, "step": 13518 }, { "epoch": 0.4816431230739085, "grad_norm": 1.1375790346302823, "learning_rate": 1.1078559561732364e-05, "loss": 0.4382, "step": 13519 }, { "epoch": 0.4816787502004026, "grad_norm": 1.0548152748276334, "learning_rate": 1.1077412349311645e-05, "loss": 0.4559, "step": 13520 }, { "epoch": 0.4817143773268967, "grad_norm": 1.3863557615428594, "learning_rate": 1.1076265122544422e-05, "loss": 0.5875, "step": 13521 }, { "epoch": 0.48175000445339083, "grad_norm": 1.6203891586091874, "learning_rate": 1.107511788144598e-05, "loss": 0.5561, "step": 13522 }, { "epoch": 0.4817856315798849, "grad_norm": 1.3912708880225977, "learning_rate": 1.1073970626031594e-05, "loss": 0.596, "step": 13523 }, { "epoch": 0.48182125870637904, "grad_norm": 1.1673683402824506, "learning_rate": 1.1072823356316535e-05, "loss": 0.4691, "step": 13524 }, { "epoch": 0.48185688583287317, "grad_norm": 1.1491547196793188, "learning_rate": 1.1071676072316084e-05, "loss": 0.4616, "step": 13525 }, { "epoch": 0.48189251295936725, "grad_norm": 1.6343360406646719, "learning_rate": 1.107052877404552e-05, "loss": 0.5491, "step": 13526 }, { "epoch": 0.4819281400858614, "grad_norm": 1.0620066975154412, "learning_rate": 1.1069381461520115e-05, "loss": 0.4831, "step": 13527 }, { "epoch": 0.4819637672123555, "grad_norm": 1.3761645503819016, "learning_rate": 1.1068234134755149e-05, "loss": 0.5157, "step": 13528 }, { "epoch": 0.4819993943388496, "grad_norm": 1.5493271882618969, "learning_rate": 1.1067086793765896e-05, "loss": 0.5783, "step": 13529 }, { "epoch": 0.4820350214653437, "grad_norm": 1.1765900861825702, "learning_rate": 1.106593943856764e-05, "loss": 0.5051, "step": 13530 }, { "epoch": 0.48207064859183785, "grad_norm": 1.4657944982372246, "learning_rate": 1.1064792069175654e-05, "loss": 0.5549, "step": 13531 }, { "epoch": 0.4821062757183319, "grad_norm": 1.4783071578923297, "learning_rate": 1.1063644685605216e-05, "loss": 0.5192, "step": 13532 }, { "epoch": 0.48214190284482605, "grad_norm": 1.0189860125660943, "learning_rate": 1.1062497287871606e-05, "loss": 0.398, "step": 13533 }, { "epoch": 0.4821775299713202, "grad_norm": 1.092246732111254, "learning_rate": 1.1061349875990101e-05, "loss": 0.4685, "step": 13534 }, { "epoch": 0.48221315709781426, "grad_norm": 1.391169246045385, "learning_rate": 1.1060202449975984e-05, "loss": 0.5575, "step": 13535 }, { "epoch": 0.4822487842243084, "grad_norm": 1.322968484969797, "learning_rate": 1.1059055009844528e-05, "loss": 0.6191, "step": 13536 }, { "epoch": 0.4822844113508025, "grad_norm": 1.4887236597207516, "learning_rate": 1.1057907555611017e-05, "loss": 0.5101, "step": 13537 }, { "epoch": 0.4823200384772966, "grad_norm": 1.526034566422353, "learning_rate": 1.1056760087290724e-05, "loss": 0.5229, "step": 13538 }, { "epoch": 0.48235566560379073, "grad_norm": 1.0610253533234928, "learning_rate": 1.1055612604898932e-05, "loss": 0.42, "step": 13539 }, { "epoch": 0.48239129273028486, "grad_norm": 1.480365173471103, "learning_rate": 1.1054465108450922e-05, "loss": 0.5804, "step": 13540 }, { "epoch": 0.48242691985677894, "grad_norm": 1.0843378604379748, "learning_rate": 1.1053317597961972e-05, "loss": 0.4409, "step": 13541 }, { "epoch": 0.48246254698327307, "grad_norm": 1.4303171559049306, "learning_rate": 1.1052170073447364e-05, "loss": 0.5283, "step": 13542 }, { "epoch": 0.4824981741097672, "grad_norm": 2.0544648989914482, "learning_rate": 1.1051022534922372e-05, "loss": 0.6769, "step": 13543 }, { "epoch": 0.4825338012362613, "grad_norm": 1.313063157220422, "learning_rate": 1.104987498240228e-05, "loss": 0.6291, "step": 13544 }, { "epoch": 0.4825694283627554, "grad_norm": 1.1275619062859819, "learning_rate": 1.1048727415902371e-05, "loss": 0.4816, "step": 13545 }, { "epoch": 0.48260505548924953, "grad_norm": 1.0535559186487233, "learning_rate": 1.1047579835437925e-05, "loss": 0.4413, "step": 13546 }, { "epoch": 0.4826406826157436, "grad_norm": 1.206334403648428, "learning_rate": 1.1046432241024219e-05, "loss": 0.4501, "step": 13547 }, { "epoch": 0.48267630974223774, "grad_norm": 1.6311942303370766, "learning_rate": 1.1045284632676535e-05, "loss": 0.5696, "step": 13548 }, { "epoch": 0.4827119368687319, "grad_norm": 1.3884008875024927, "learning_rate": 1.1044137010410158e-05, "loss": 0.5327, "step": 13549 }, { "epoch": 0.48274756399522595, "grad_norm": 1.1759213359459753, "learning_rate": 1.1042989374240367e-05, "loss": 0.4901, "step": 13550 }, { "epoch": 0.4827831911217201, "grad_norm": 1.6103473833865574, "learning_rate": 1.104184172418244e-05, "loss": 0.5741, "step": 13551 }, { "epoch": 0.4828188182482142, "grad_norm": 1.0825546450566088, "learning_rate": 1.1040694060251667e-05, "loss": 0.4448, "step": 13552 }, { "epoch": 0.4828544453747083, "grad_norm": 1.0939931149903965, "learning_rate": 1.1039546382463323e-05, "loss": 0.4625, "step": 13553 }, { "epoch": 0.4828900725012024, "grad_norm": 1.1177621060626954, "learning_rate": 1.1038398690832695e-05, "loss": 0.4309, "step": 13554 }, { "epoch": 0.48292569962769655, "grad_norm": 2.268652301877622, "learning_rate": 1.1037250985375056e-05, "loss": 0.5453, "step": 13555 }, { "epoch": 0.4829613267541906, "grad_norm": 1.058091707914879, "learning_rate": 1.10361032661057e-05, "loss": 0.4389, "step": 13556 }, { "epoch": 0.48299695388068475, "grad_norm": 1.9050484556752205, "learning_rate": 1.1034955533039904e-05, "loss": 0.5076, "step": 13557 }, { "epoch": 0.4830325810071789, "grad_norm": 1.0810168502987905, "learning_rate": 1.1033807786192951e-05, "loss": 0.4499, "step": 13558 }, { "epoch": 0.48306820813367296, "grad_norm": 1.750616258449735, "learning_rate": 1.1032660025580127e-05, "loss": 0.5519, "step": 13559 }, { "epoch": 0.4831038352601671, "grad_norm": 1.0356393703657665, "learning_rate": 1.103151225121671e-05, "loss": 0.4116, "step": 13560 }, { "epoch": 0.4831394623866612, "grad_norm": 1.5384140739667602, "learning_rate": 1.1030364463117988e-05, "loss": 0.4727, "step": 13561 }, { "epoch": 0.4831750895131553, "grad_norm": 1.2967192101874279, "learning_rate": 1.1029216661299242e-05, "loss": 0.5512, "step": 13562 }, { "epoch": 0.48321071663964943, "grad_norm": 1.242052073801777, "learning_rate": 1.1028068845775758e-05, "loss": 0.4912, "step": 13563 }, { "epoch": 0.48324634376614356, "grad_norm": 1.1342062148022047, "learning_rate": 1.102692101656282e-05, "loss": 0.4599, "step": 13564 }, { "epoch": 0.48328197089263764, "grad_norm": 1.4968936442455318, "learning_rate": 1.1025773173675708e-05, "loss": 0.5086, "step": 13565 }, { "epoch": 0.48331759801913177, "grad_norm": 1.4691329178501034, "learning_rate": 1.1024625317129711e-05, "loss": 0.5858, "step": 13566 }, { "epoch": 0.4833532251456259, "grad_norm": 1.1562027418907717, "learning_rate": 1.102347744694011e-05, "loss": 0.4816, "step": 13567 }, { "epoch": 0.48338885227212, "grad_norm": 1.3098685398088368, "learning_rate": 1.1022329563122191e-05, "loss": 0.5222, "step": 13568 }, { "epoch": 0.4834244793986141, "grad_norm": 1.4202622957884485, "learning_rate": 1.1021181665691244e-05, "loss": 0.5609, "step": 13569 }, { "epoch": 0.48346010652510824, "grad_norm": 1.1047924389420611, "learning_rate": 1.1020033754662542e-05, "loss": 0.457, "step": 13570 }, { "epoch": 0.4834957336516023, "grad_norm": 1.1344012240504824, "learning_rate": 1.1018885830051385e-05, "loss": 0.4665, "step": 13571 }, { "epoch": 0.48353136077809644, "grad_norm": 1.305992274404751, "learning_rate": 1.1017737891873048e-05, "loss": 0.4575, "step": 13572 }, { "epoch": 0.4835669879045906, "grad_norm": 1.1300277333109368, "learning_rate": 1.1016589940142817e-05, "loss": 0.4235, "step": 13573 }, { "epoch": 0.48360261503108465, "grad_norm": 1.134213484494927, "learning_rate": 1.1015441974875984e-05, "loss": 0.4785, "step": 13574 }, { "epoch": 0.4836382421575788, "grad_norm": 1.5055030744880626, "learning_rate": 1.101429399608783e-05, "loss": 0.5662, "step": 13575 }, { "epoch": 0.4836738692840729, "grad_norm": 1.0561465947339608, "learning_rate": 1.1013146003793644e-05, "loss": 0.4452, "step": 13576 }, { "epoch": 0.483709496410567, "grad_norm": 1.0830902921382652, "learning_rate": 1.1011997998008707e-05, "loss": 0.4397, "step": 13577 }, { "epoch": 0.4837451235370611, "grad_norm": 1.185824562243027, "learning_rate": 1.1010849978748314e-05, "loss": 0.4614, "step": 13578 }, { "epoch": 0.48378075066355525, "grad_norm": 1.1922976391656224, "learning_rate": 1.1009701946027743e-05, "loss": 0.4402, "step": 13579 }, { "epoch": 0.4838163777900493, "grad_norm": 1.9870284969403864, "learning_rate": 1.1008553899862287e-05, "loss": 0.6556, "step": 13580 }, { "epoch": 0.48385200491654345, "grad_norm": 1.344747780205767, "learning_rate": 1.1007405840267229e-05, "loss": 0.6121, "step": 13581 }, { "epoch": 0.4838876320430376, "grad_norm": 1.1422192538712768, "learning_rate": 1.100625776725786e-05, "loss": 0.4472, "step": 13582 }, { "epoch": 0.48392325916953166, "grad_norm": 1.3723344122668846, "learning_rate": 1.1005109680849466e-05, "loss": 0.4834, "step": 13583 }, { "epoch": 0.4839588862960258, "grad_norm": 1.0634946096122493, "learning_rate": 1.1003961581057334e-05, "loss": 0.3966, "step": 13584 }, { "epoch": 0.4839945134225199, "grad_norm": 1.0514230060593848, "learning_rate": 1.100281346789675e-05, "loss": 0.432, "step": 13585 }, { "epoch": 0.484030140549014, "grad_norm": 1.1719114461179625, "learning_rate": 1.1001665341383005e-05, "loss": 0.5205, "step": 13586 }, { "epoch": 0.48406576767550813, "grad_norm": 1.3331159074759775, "learning_rate": 1.1000517201531387e-05, "loss": 0.5681, "step": 13587 }, { "epoch": 0.48410139480200226, "grad_norm": 1.421720222758286, "learning_rate": 1.099936904835718e-05, "loss": 0.5765, "step": 13588 }, { "epoch": 0.48413702192849634, "grad_norm": 1.318063128313927, "learning_rate": 1.0998220881875679e-05, "loss": 0.5829, "step": 13589 }, { "epoch": 0.48417264905499047, "grad_norm": 1.096911495541403, "learning_rate": 1.0997072702102164e-05, "loss": 0.4741, "step": 13590 }, { "epoch": 0.4842082761814846, "grad_norm": 1.4918845250260269, "learning_rate": 1.0995924509051934e-05, "loss": 0.5845, "step": 13591 }, { "epoch": 0.4842439033079787, "grad_norm": 1.1485555559297589, "learning_rate": 1.0994776302740271e-05, "loss": 0.4846, "step": 13592 }, { "epoch": 0.4842795304344728, "grad_norm": 1.7090678916375925, "learning_rate": 1.0993628083182468e-05, "loss": 0.542, "step": 13593 }, { "epoch": 0.48431515756096694, "grad_norm": 1.3561830478842898, "learning_rate": 1.099247985039381e-05, "loss": 0.5283, "step": 13594 }, { "epoch": 0.484350784687461, "grad_norm": 1.1798485961620544, "learning_rate": 1.0991331604389593e-05, "loss": 0.4825, "step": 13595 }, { "epoch": 0.48438641181395514, "grad_norm": 1.1654224323674491, "learning_rate": 1.09901833451851e-05, "loss": 0.4445, "step": 13596 }, { "epoch": 0.4844220389404493, "grad_norm": 2.432619988539202, "learning_rate": 1.0989035072795624e-05, "loss": 0.6264, "step": 13597 }, { "epoch": 0.48445766606694335, "grad_norm": 1.0915995190833137, "learning_rate": 1.0987886787236455e-05, "loss": 0.4612, "step": 13598 }, { "epoch": 0.4844932931934375, "grad_norm": 1.61824710234456, "learning_rate": 1.0986738488522881e-05, "loss": 0.6151, "step": 13599 }, { "epoch": 0.4845289203199316, "grad_norm": 1.1132476014187984, "learning_rate": 1.0985590176670198e-05, "loss": 0.4863, "step": 13600 }, { "epoch": 0.4845645474464257, "grad_norm": 1.1199840588114593, "learning_rate": 1.098444185169369e-05, "loss": 0.4598, "step": 13601 }, { "epoch": 0.4846001745729198, "grad_norm": 1.0757548279429014, "learning_rate": 1.098329351360865e-05, "loss": 0.4886, "step": 13602 }, { "epoch": 0.48463580169941395, "grad_norm": 1.0754742579588559, "learning_rate": 1.0982145162430373e-05, "loss": 0.4979, "step": 13603 }, { "epoch": 0.484671428825908, "grad_norm": 1.11074183292673, "learning_rate": 1.0980996798174143e-05, "loss": 0.5041, "step": 13604 }, { "epoch": 0.48470705595240215, "grad_norm": 1.0521976338826837, "learning_rate": 1.0979848420855256e-05, "loss": 0.4758, "step": 13605 }, { "epoch": 0.4847426830788963, "grad_norm": 1.0742862539683957, "learning_rate": 1.0978700030489007e-05, "loss": 0.4577, "step": 13606 }, { "epoch": 0.48477831020539036, "grad_norm": 1.135452134639113, "learning_rate": 1.0977551627090677e-05, "loss": 0.3998, "step": 13607 }, { "epoch": 0.4848139373318845, "grad_norm": 1.120404578686, "learning_rate": 1.0976403210675566e-05, "loss": 0.4626, "step": 13608 }, { "epoch": 0.4848495644583786, "grad_norm": 1.2927457061957823, "learning_rate": 1.0975254781258966e-05, "loss": 0.4376, "step": 13609 }, { "epoch": 0.4848851915848727, "grad_norm": 1.082307334347906, "learning_rate": 1.0974106338856166e-05, "loss": 0.3881, "step": 13610 }, { "epoch": 0.48492081871136683, "grad_norm": 1.84356978071922, "learning_rate": 1.0972957883482456e-05, "loss": 0.5363, "step": 13611 }, { "epoch": 0.48495644583786096, "grad_norm": 1.6974954121720285, "learning_rate": 1.0971809415153135e-05, "loss": 0.5159, "step": 13612 }, { "epoch": 0.48499207296435504, "grad_norm": 1.411438741060544, "learning_rate": 1.097066093388349e-05, "loss": 0.525, "step": 13613 }, { "epoch": 0.48502770009084917, "grad_norm": 1.8273112227220991, "learning_rate": 1.0969512439688816e-05, "loss": 0.5442, "step": 13614 }, { "epoch": 0.4850633272173433, "grad_norm": 1.5755107466554439, "learning_rate": 1.0968363932584407e-05, "loss": 0.5853, "step": 13615 }, { "epoch": 0.4850989543438374, "grad_norm": 1.876895804868688, "learning_rate": 1.0967215412585557e-05, "loss": 0.6585, "step": 13616 }, { "epoch": 0.4851345814703315, "grad_norm": 1.3866936535281982, "learning_rate": 1.0966066879707555e-05, "loss": 0.5707, "step": 13617 }, { "epoch": 0.48517020859682564, "grad_norm": 1.1508413589044155, "learning_rate": 1.0964918333965699e-05, "loss": 0.4429, "step": 13618 }, { "epoch": 0.4852058357233197, "grad_norm": 1.1517951055470383, "learning_rate": 1.0963769775375281e-05, "loss": 0.4436, "step": 13619 }, { "epoch": 0.48524146284981384, "grad_norm": 1.0829372107561992, "learning_rate": 1.0962621203951595e-05, "loss": 0.4332, "step": 13620 }, { "epoch": 0.485277089976308, "grad_norm": 1.0283275591206444, "learning_rate": 1.0961472619709936e-05, "loss": 0.4549, "step": 13621 }, { "epoch": 0.48531271710280205, "grad_norm": 1.4461501257723481, "learning_rate": 1.0960324022665593e-05, "loss": 0.5849, "step": 13622 }, { "epoch": 0.4853483442292962, "grad_norm": 1.3676712044300632, "learning_rate": 1.0959175412833869e-05, "loss": 0.5591, "step": 13623 }, { "epoch": 0.4853839713557903, "grad_norm": 1.4164666471320908, "learning_rate": 1.0958026790230052e-05, "loss": 0.6107, "step": 13624 }, { "epoch": 0.4854195984822844, "grad_norm": 1.1166531481491293, "learning_rate": 1.0956878154869437e-05, "loss": 0.4833, "step": 13625 }, { "epoch": 0.4854552256087785, "grad_norm": 1.763741562980301, "learning_rate": 1.095572950676732e-05, "loss": 0.5267, "step": 13626 }, { "epoch": 0.48549085273527265, "grad_norm": 1.554286396789902, "learning_rate": 1.0954580845939e-05, "loss": 0.5524, "step": 13627 }, { "epoch": 0.4855264798617667, "grad_norm": 1.201086735665731, "learning_rate": 1.095343217239977e-05, "loss": 0.5405, "step": 13628 }, { "epoch": 0.48556210698826086, "grad_norm": 1.121086641444253, "learning_rate": 1.0952283486164923e-05, "loss": 0.4689, "step": 13629 }, { "epoch": 0.485597734114755, "grad_norm": 1.938292560338395, "learning_rate": 1.0951134787249754e-05, "loss": 0.5504, "step": 13630 }, { "epoch": 0.48563336124124906, "grad_norm": 1.0838845196858033, "learning_rate": 1.0949986075669558e-05, "loss": 0.4423, "step": 13631 }, { "epoch": 0.4856689883677432, "grad_norm": 1.367675258662435, "learning_rate": 1.0948837351439639e-05, "loss": 0.4752, "step": 13632 }, { "epoch": 0.4857046154942373, "grad_norm": 1.2353251546080246, "learning_rate": 1.0947688614575283e-05, "loss": 0.4888, "step": 13633 }, { "epoch": 0.4857402426207314, "grad_norm": 1.133999417478091, "learning_rate": 1.0946539865091794e-05, "loss": 0.462, "step": 13634 }, { "epoch": 0.48577586974722553, "grad_norm": 1.4786613848874677, "learning_rate": 1.0945391103004464e-05, "loss": 0.5547, "step": 13635 }, { "epoch": 0.48581149687371966, "grad_norm": 1.0591948439288297, "learning_rate": 1.094424232832859e-05, "loss": 0.451, "step": 13636 }, { "epoch": 0.48584712400021374, "grad_norm": 1.282680975277269, "learning_rate": 1.0943093541079467e-05, "loss": 0.496, "step": 13637 }, { "epoch": 0.48588275112670787, "grad_norm": 1.5809568000669703, "learning_rate": 1.09419447412724e-05, "loss": 0.5603, "step": 13638 }, { "epoch": 0.485918378253202, "grad_norm": 2.8017021179711037, "learning_rate": 1.0940795928922676e-05, "loss": 0.627, "step": 13639 }, { "epoch": 0.4859540053796961, "grad_norm": 1.4976199556933438, "learning_rate": 1.0939647104045598e-05, "loss": 0.5988, "step": 13640 }, { "epoch": 0.4859896325061902, "grad_norm": 1.2282178666197323, "learning_rate": 1.0938498266656463e-05, "loss": 0.4988, "step": 13641 }, { "epoch": 0.48602525963268434, "grad_norm": 1.5997498380098245, "learning_rate": 1.0937349416770566e-05, "loss": 0.5899, "step": 13642 }, { "epoch": 0.4860608867591784, "grad_norm": 3.3851960768006943, "learning_rate": 1.0936200554403209e-05, "loss": 0.5715, "step": 13643 }, { "epoch": 0.48609651388567254, "grad_norm": 1.1033099205708685, "learning_rate": 1.0935051679569684e-05, "loss": 0.4851, "step": 13644 }, { "epoch": 0.4861321410121667, "grad_norm": 1.6147773094992215, "learning_rate": 1.0933902792285294e-05, "loss": 0.5979, "step": 13645 }, { "epoch": 0.48616776813866075, "grad_norm": 1.8008940817829469, "learning_rate": 1.0932753892565335e-05, "loss": 0.6178, "step": 13646 }, { "epoch": 0.4862033952651549, "grad_norm": 1.6309270577461938, "learning_rate": 1.0931604980425108e-05, "loss": 0.597, "step": 13647 }, { "epoch": 0.486239022391649, "grad_norm": 1.0970956378166778, "learning_rate": 1.0930456055879905e-05, "loss": 0.4894, "step": 13648 }, { "epoch": 0.4862746495181431, "grad_norm": 1.2551154707079872, "learning_rate": 1.0929307118945032e-05, "loss": 0.4514, "step": 13649 }, { "epoch": 0.4863102766446372, "grad_norm": 1.0369941487614884, "learning_rate": 1.0928158169635785e-05, "loss": 0.4138, "step": 13650 }, { "epoch": 0.48634590377113135, "grad_norm": 1.1768452761745238, "learning_rate": 1.0927009207967465e-05, "loss": 0.4808, "step": 13651 }, { "epoch": 0.4863815308976254, "grad_norm": 1.5209955457233442, "learning_rate": 1.0925860233955368e-05, "loss": 0.607, "step": 13652 }, { "epoch": 0.48641715802411956, "grad_norm": 2.096477130934544, "learning_rate": 1.0924711247614794e-05, "loss": 0.5138, "step": 13653 }, { "epoch": 0.4864527851506137, "grad_norm": 2.0361346654525825, "learning_rate": 1.0923562248961046e-05, "loss": 0.6143, "step": 13654 }, { "epoch": 0.48648841227710776, "grad_norm": 0.9915881349075998, "learning_rate": 1.0922413238009417e-05, "loss": 0.4952, "step": 13655 }, { "epoch": 0.4865240394036019, "grad_norm": 1.4480112241406198, "learning_rate": 1.0921264214775214e-05, "loss": 0.5123, "step": 13656 }, { "epoch": 0.486559666530096, "grad_norm": 1.5760415898311413, "learning_rate": 1.0920115179273734e-05, "loss": 0.5642, "step": 13657 }, { "epoch": 0.48659529365659016, "grad_norm": 1.2284255672412538, "learning_rate": 1.0918966131520276e-05, "loss": 0.4466, "step": 13658 }, { "epoch": 0.48663092078308423, "grad_norm": 1.116459916897394, "learning_rate": 1.091781707153014e-05, "loss": 0.4495, "step": 13659 }, { "epoch": 0.48666654790957836, "grad_norm": 1.1127461764848228, "learning_rate": 1.091666799931863e-05, "loss": 0.4354, "step": 13660 }, { "epoch": 0.4867021750360725, "grad_norm": 1.068970094402089, "learning_rate": 1.0915518914901047e-05, "loss": 0.4476, "step": 13661 }, { "epoch": 0.48673780216256657, "grad_norm": 1.1900708279809162, "learning_rate": 1.091436981829269e-05, "loss": 0.4822, "step": 13662 }, { "epoch": 0.4867734292890607, "grad_norm": 1.0758461936982413, "learning_rate": 1.0913220709508856e-05, "loss": 0.4456, "step": 13663 }, { "epoch": 0.48680905641555483, "grad_norm": 1.165403127412763, "learning_rate": 1.0912071588564852e-05, "loss": 0.4877, "step": 13664 }, { "epoch": 0.4868446835420489, "grad_norm": 1.0929924509988855, "learning_rate": 1.0910922455475978e-05, "loss": 0.4506, "step": 13665 }, { "epoch": 0.48688031066854304, "grad_norm": 1.2353804208693988, "learning_rate": 1.0909773310257535e-05, "loss": 0.4255, "step": 13666 }, { "epoch": 0.48691593779503717, "grad_norm": 1.1662568438455363, "learning_rate": 1.090862415292482e-05, "loss": 0.4298, "step": 13667 }, { "epoch": 0.48695156492153124, "grad_norm": 1.3981819524141799, "learning_rate": 1.0907474983493144e-05, "loss": 0.5402, "step": 13668 }, { "epoch": 0.4869871920480254, "grad_norm": 1.2509346510516408, "learning_rate": 1.0906325801977804e-05, "loss": 0.6442, "step": 13669 }, { "epoch": 0.4870228191745195, "grad_norm": 1.3813942479932868, "learning_rate": 1.09051766083941e-05, "loss": 0.5077, "step": 13670 }, { "epoch": 0.4870584463010136, "grad_norm": 1.9260886771315398, "learning_rate": 1.0904027402757337e-05, "loss": 0.601, "step": 13671 }, { "epoch": 0.4870940734275077, "grad_norm": 1.5328278671869986, "learning_rate": 1.090287818508282e-05, "loss": 0.5841, "step": 13672 }, { "epoch": 0.48712970055400184, "grad_norm": 1.6022036640676094, "learning_rate": 1.0901728955385847e-05, "loss": 0.5252, "step": 13673 }, { "epoch": 0.4871653276804959, "grad_norm": 1.5760798275064998, "learning_rate": 1.0900579713681722e-05, "loss": 0.5913, "step": 13674 }, { "epoch": 0.48720095480699005, "grad_norm": 1.435862772987304, "learning_rate": 1.089943045998575e-05, "loss": 0.5508, "step": 13675 }, { "epoch": 0.4872365819334842, "grad_norm": 1.2640626024586292, "learning_rate": 1.0898281194313234e-05, "loss": 0.4229, "step": 13676 }, { "epoch": 0.48727220905997826, "grad_norm": 1.518249159693579, "learning_rate": 1.0897131916679475e-05, "loss": 0.6385, "step": 13677 }, { "epoch": 0.4873078361864724, "grad_norm": 1.6432618171161382, "learning_rate": 1.0895982627099775e-05, "loss": 0.5652, "step": 13678 }, { "epoch": 0.4873434633129665, "grad_norm": 2.0063408055019267, "learning_rate": 1.0894833325589443e-05, "loss": 0.5754, "step": 13679 }, { "epoch": 0.4873790904394606, "grad_norm": 1.4050287468756744, "learning_rate": 1.089368401216378e-05, "loss": 0.5301, "step": 13680 }, { "epoch": 0.4874147175659547, "grad_norm": 1.9442654833638646, "learning_rate": 1.0892534686838086e-05, "loss": 0.5152, "step": 13681 }, { "epoch": 0.48745034469244886, "grad_norm": 1.0809551724186848, "learning_rate": 1.0891385349627671e-05, "loss": 0.4655, "step": 13682 }, { "epoch": 0.48748597181894293, "grad_norm": 1.4236921408823466, "learning_rate": 1.0890236000547839e-05, "loss": 0.5149, "step": 13683 }, { "epoch": 0.48752159894543706, "grad_norm": 1.4995924542472934, "learning_rate": 1.0889086639613891e-05, "loss": 0.5162, "step": 13684 }, { "epoch": 0.4875572260719312, "grad_norm": 2.293674921412678, "learning_rate": 1.0887937266841132e-05, "loss": 0.5713, "step": 13685 }, { "epoch": 0.48759285319842527, "grad_norm": 1.058366193238119, "learning_rate": 1.0886787882244867e-05, "loss": 0.4459, "step": 13686 }, { "epoch": 0.4876284803249194, "grad_norm": 2.1724881674676673, "learning_rate": 1.0885638485840404e-05, "loss": 0.5683, "step": 13687 }, { "epoch": 0.48766410745141353, "grad_norm": 1.5608810294635511, "learning_rate": 1.0884489077643045e-05, "loss": 0.4803, "step": 13688 }, { "epoch": 0.4876997345779076, "grad_norm": 1.488551361701732, "learning_rate": 1.0883339657668095e-05, "loss": 0.5091, "step": 13689 }, { "epoch": 0.48773536170440174, "grad_norm": 1.0700210962939327, "learning_rate": 1.088219022593086e-05, "loss": 0.4576, "step": 13690 }, { "epoch": 0.48777098883089587, "grad_norm": 1.7145865390709678, "learning_rate": 1.0881040782446645e-05, "loss": 0.5715, "step": 13691 }, { "epoch": 0.48780661595738994, "grad_norm": 1.871889629921311, "learning_rate": 1.087989132723076e-05, "loss": 0.551, "step": 13692 }, { "epoch": 0.4878422430838841, "grad_norm": 1.7330447124807855, "learning_rate": 1.0878741860298503e-05, "loss": 0.5544, "step": 13693 }, { "epoch": 0.4878778702103782, "grad_norm": 1.1938738792873034, "learning_rate": 1.0877592381665183e-05, "loss": 0.4331, "step": 13694 }, { "epoch": 0.4879134973368723, "grad_norm": 1.3874338109381494, "learning_rate": 1.0876442891346109e-05, "loss": 0.5232, "step": 13695 }, { "epoch": 0.4879491244633664, "grad_norm": 1.2868404925922179, "learning_rate": 1.0875293389356585e-05, "loss": 0.4766, "step": 13696 }, { "epoch": 0.48798475158986054, "grad_norm": 2.1505641424134585, "learning_rate": 1.0874143875711918e-05, "loss": 0.4513, "step": 13697 }, { "epoch": 0.4880203787163546, "grad_norm": 1.3972959933912616, "learning_rate": 1.0872994350427414e-05, "loss": 0.6142, "step": 13698 }, { "epoch": 0.48805600584284875, "grad_norm": 1.1175568651754235, "learning_rate": 1.0871844813518381e-05, "loss": 0.4193, "step": 13699 }, { "epoch": 0.4880916329693429, "grad_norm": 1.0997636335399614, "learning_rate": 1.0870695265000121e-05, "loss": 0.4586, "step": 13700 }, { "epoch": 0.48812726009583696, "grad_norm": 1.752581616935346, "learning_rate": 1.0869545704887952e-05, "loss": 0.4936, "step": 13701 }, { "epoch": 0.4881628872223311, "grad_norm": 1.0142206364186177, "learning_rate": 1.086839613319717e-05, "loss": 0.468, "step": 13702 }, { "epoch": 0.4881985143488252, "grad_norm": 1.7249525164478652, "learning_rate": 1.0867246549943087e-05, "loss": 0.5585, "step": 13703 }, { "epoch": 0.4882341414753193, "grad_norm": 1.8829633080523578, "learning_rate": 1.0866096955141007e-05, "loss": 0.6005, "step": 13704 }, { "epoch": 0.4882697686018134, "grad_norm": 1.6189566649458889, "learning_rate": 1.0864947348806245e-05, "loss": 0.5395, "step": 13705 }, { "epoch": 0.48830539572830756, "grad_norm": 1.2986698060196362, "learning_rate": 1.0863797730954101e-05, "loss": 0.4658, "step": 13706 }, { "epoch": 0.48834102285480163, "grad_norm": 1.125720848023016, "learning_rate": 1.086264810159989e-05, "loss": 0.4503, "step": 13707 }, { "epoch": 0.48837664998129576, "grad_norm": 1.6835256090720117, "learning_rate": 1.0861498460758915e-05, "loss": 0.6323, "step": 13708 }, { "epoch": 0.4884122771077899, "grad_norm": 1.1301191334571943, "learning_rate": 1.0860348808446486e-05, "loss": 0.4908, "step": 13709 }, { "epoch": 0.48844790423428397, "grad_norm": 1.7959738512512406, "learning_rate": 1.0859199144677911e-05, "loss": 0.6629, "step": 13710 }, { "epoch": 0.4884835313607781, "grad_norm": 1.750359118899796, "learning_rate": 1.0858049469468501e-05, "loss": 0.5806, "step": 13711 }, { "epoch": 0.48851915848727223, "grad_norm": 1.4182304006990931, "learning_rate": 1.0856899782833559e-05, "loss": 0.4953, "step": 13712 }, { "epoch": 0.4885547856137663, "grad_norm": 2.0647976170773115, "learning_rate": 1.08557500847884e-05, "loss": 0.4373, "step": 13713 }, { "epoch": 0.48859041274026044, "grad_norm": 1.488717425678141, "learning_rate": 1.085460037534833e-05, "loss": 0.5306, "step": 13714 }, { "epoch": 0.48862603986675457, "grad_norm": 1.3608150960116931, "learning_rate": 1.0853450654528657e-05, "loss": 0.5854, "step": 13715 }, { "epoch": 0.48866166699324864, "grad_norm": 1.6546449372875875, "learning_rate": 1.0852300922344694e-05, "loss": 0.5828, "step": 13716 }, { "epoch": 0.4886972941197428, "grad_norm": 1.598677007784674, "learning_rate": 1.0851151178811745e-05, "loss": 0.5481, "step": 13717 }, { "epoch": 0.4887329212462369, "grad_norm": 1.2530518886219622, "learning_rate": 1.0850001423945127e-05, "loss": 0.4712, "step": 13718 }, { "epoch": 0.488768548372731, "grad_norm": 1.6754224173240633, "learning_rate": 1.0848851657760142e-05, "loss": 0.5155, "step": 13719 }, { "epoch": 0.4888041754992251, "grad_norm": 1.4276115542822854, "learning_rate": 1.0847701880272108e-05, "loss": 0.4872, "step": 13720 }, { "epoch": 0.48883980262571924, "grad_norm": 1.2600155258574466, "learning_rate": 1.0846552091496332e-05, "loss": 0.5372, "step": 13721 }, { "epoch": 0.4888754297522133, "grad_norm": 1.6457983100584537, "learning_rate": 1.084540229144812e-05, "loss": 0.545, "step": 13722 }, { "epoch": 0.48891105687870745, "grad_norm": 1.1270912734564575, "learning_rate": 1.0844252480142785e-05, "loss": 0.4464, "step": 13723 }, { "epoch": 0.4889466840052016, "grad_norm": 1.1211180925841744, "learning_rate": 1.0843102657595641e-05, "loss": 0.4637, "step": 13724 }, { "epoch": 0.48898231113169566, "grad_norm": 1.1432507083157062, "learning_rate": 1.0841952823821996e-05, "loss": 0.446, "step": 13725 }, { "epoch": 0.4890179382581898, "grad_norm": 1.509138429649417, "learning_rate": 1.0840802978837156e-05, "loss": 0.5539, "step": 13726 }, { "epoch": 0.4890535653846839, "grad_norm": 1.0553219824448383, "learning_rate": 1.083965312265644e-05, "loss": 0.4108, "step": 13727 }, { "epoch": 0.489089192511178, "grad_norm": 1.239192895460986, "learning_rate": 1.0838503255295156e-05, "loss": 0.4772, "step": 13728 }, { "epoch": 0.4891248196376721, "grad_norm": 1.5370149442065082, "learning_rate": 1.0837353376768614e-05, "loss": 0.5954, "step": 13729 }, { "epoch": 0.48916044676416626, "grad_norm": 1.3030045249862263, "learning_rate": 1.0836203487092127e-05, "loss": 0.4419, "step": 13730 }, { "epoch": 0.48919607389066033, "grad_norm": 1.8309715620491822, "learning_rate": 1.0835053586281004e-05, "loss": 0.5216, "step": 13731 }, { "epoch": 0.48923170101715446, "grad_norm": 1.6976728522294262, "learning_rate": 1.0833903674350563e-05, "loss": 0.5488, "step": 13732 }, { "epoch": 0.4892673281436486, "grad_norm": 1.1428394082036564, "learning_rate": 1.0832753751316109e-05, "loss": 0.4473, "step": 13733 }, { "epoch": 0.48930295527014267, "grad_norm": 1.087055558086406, "learning_rate": 1.0831603817192957e-05, "loss": 0.4383, "step": 13734 }, { "epoch": 0.4893385823966368, "grad_norm": 1.6417085544808512, "learning_rate": 1.083045387199642e-05, "loss": 0.6353, "step": 13735 }, { "epoch": 0.48937420952313093, "grad_norm": 1.364114714467907, "learning_rate": 1.0829303915741808e-05, "loss": 0.5432, "step": 13736 }, { "epoch": 0.489409836649625, "grad_norm": 1.1073668872723732, "learning_rate": 1.0828153948444435e-05, "loss": 0.4352, "step": 13737 }, { "epoch": 0.48944546377611914, "grad_norm": 1.709226263374295, "learning_rate": 1.0827003970119614e-05, "loss": 0.5122, "step": 13738 }, { "epoch": 0.48948109090261327, "grad_norm": 1.4405021954561599, "learning_rate": 1.0825853980782657e-05, "loss": 0.5417, "step": 13739 }, { "epoch": 0.48951671802910734, "grad_norm": 1.1166967402923638, "learning_rate": 1.0824703980448876e-05, "loss": 0.4441, "step": 13740 }, { "epoch": 0.4895523451556015, "grad_norm": 1.2407762768969985, "learning_rate": 1.0823553969133587e-05, "loss": 0.4646, "step": 13741 }, { "epoch": 0.4895879722820956, "grad_norm": 1.4005938931414117, "learning_rate": 1.0822403946852099e-05, "loss": 0.5508, "step": 13742 }, { "epoch": 0.4896235994085897, "grad_norm": 1.6921536193265105, "learning_rate": 1.0821253913619727e-05, "loss": 0.5237, "step": 13743 }, { "epoch": 0.4896592265350838, "grad_norm": 1.0163245564790122, "learning_rate": 1.0820103869451788e-05, "loss": 0.4299, "step": 13744 }, { "epoch": 0.48969485366157794, "grad_norm": 1.0166519252405282, "learning_rate": 1.081895381436359e-05, "loss": 0.4423, "step": 13745 }, { "epoch": 0.489730480788072, "grad_norm": 1.1010688502870405, "learning_rate": 1.0817803748370453e-05, "loss": 0.4427, "step": 13746 }, { "epoch": 0.48976610791456615, "grad_norm": 2.1783078811687133, "learning_rate": 1.0816653671487685e-05, "loss": 0.5898, "step": 13747 }, { "epoch": 0.4898017350410603, "grad_norm": 1.8037050961772738, "learning_rate": 1.0815503583730604e-05, "loss": 0.5803, "step": 13748 }, { "epoch": 0.48983736216755436, "grad_norm": 1.3442240500351117, "learning_rate": 1.0814353485114521e-05, "loss": 0.642, "step": 13749 }, { "epoch": 0.4898729892940485, "grad_norm": 1.4084216283332198, "learning_rate": 1.0813203375654753e-05, "loss": 0.4112, "step": 13750 }, { "epoch": 0.4899086164205426, "grad_norm": 1.415205336733916, "learning_rate": 1.0812053255366614e-05, "loss": 0.5843, "step": 13751 }, { "epoch": 0.4899442435470367, "grad_norm": 1.4064520088628174, "learning_rate": 1.0810903124265417e-05, "loss": 0.5346, "step": 13752 }, { "epoch": 0.4899798706735308, "grad_norm": 1.4190509038535983, "learning_rate": 1.0809752982366479e-05, "loss": 0.6363, "step": 13753 }, { "epoch": 0.49001549780002496, "grad_norm": 1.1689689341505756, "learning_rate": 1.0808602829685113e-05, "loss": 0.4685, "step": 13754 }, { "epoch": 0.49005112492651903, "grad_norm": 1.537079398586994, "learning_rate": 1.0807452666236638e-05, "loss": 0.6026, "step": 13755 }, { "epoch": 0.49008675205301316, "grad_norm": 1.4208117654793992, "learning_rate": 1.0806302492036363e-05, "loss": 0.4927, "step": 13756 }, { "epoch": 0.4901223791795073, "grad_norm": 1.1162472291326286, "learning_rate": 1.080515230709961e-05, "loss": 0.455, "step": 13757 }, { "epoch": 0.49015800630600137, "grad_norm": 1.1674058212119698, "learning_rate": 1.080400211144169e-05, "loss": 0.514, "step": 13758 }, { "epoch": 0.4901936334324955, "grad_norm": 1.1928160310646212, "learning_rate": 1.080285190507792e-05, "loss": 0.5037, "step": 13759 }, { "epoch": 0.49022926055898963, "grad_norm": 1.0568223423676175, "learning_rate": 1.0801701688023614e-05, "loss": 0.4637, "step": 13760 }, { "epoch": 0.4902648876854837, "grad_norm": 1.1496601821878818, "learning_rate": 1.0800551460294092e-05, "loss": 0.4266, "step": 13761 }, { "epoch": 0.49030051481197784, "grad_norm": 1.6445261610093533, "learning_rate": 1.0799401221904668e-05, "loss": 0.5965, "step": 13762 }, { "epoch": 0.49033614193847197, "grad_norm": 1.1603872013171863, "learning_rate": 1.0798250972870653e-05, "loss": 0.4536, "step": 13763 }, { "epoch": 0.49037176906496605, "grad_norm": 1.078254661296214, "learning_rate": 1.0797100713207371e-05, "loss": 0.5023, "step": 13764 }, { "epoch": 0.4904073961914602, "grad_norm": 1.682795912130557, "learning_rate": 1.0795950442930138e-05, "loss": 0.5274, "step": 13765 }, { "epoch": 0.4904430233179543, "grad_norm": 1.8249827863254071, "learning_rate": 1.079480016205427e-05, "loss": 0.4787, "step": 13766 }, { "epoch": 0.4904786504444484, "grad_norm": 1.1591832980975292, "learning_rate": 1.079364987059508e-05, "loss": 0.4486, "step": 13767 }, { "epoch": 0.4905142775709425, "grad_norm": 2.43244080420016, "learning_rate": 1.0792499568567885e-05, "loss": 0.5548, "step": 13768 }, { "epoch": 0.49054990469743665, "grad_norm": 1.1188989382165546, "learning_rate": 1.0791349255988009e-05, "loss": 0.4476, "step": 13769 }, { "epoch": 0.4905855318239307, "grad_norm": 1.1332060503319266, "learning_rate": 1.0790198932870761e-05, "loss": 0.4827, "step": 13770 }, { "epoch": 0.49062115895042485, "grad_norm": 1.3868902687544409, "learning_rate": 1.0789048599231462e-05, "loss": 0.4817, "step": 13771 }, { "epoch": 0.490656786076919, "grad_norm": 1.0190848868022344, "learning_rate": 1.0787898255085429e-05, "loss": 0.4479, "step": 13772 }, { "epoch": 0.49069241320341306, "grad_norm": 1.589169408749894, "learning_rate": 1.0786747900447983e-05, "loss": 0.5437, "step": 13773 }, { "epoch": 0.4907280403299072, "grad_norm": 1.7232022749484182, "learning_rate": 1.0785597535334437e-05, "loss": 0.5825, "step": 13774 }, { "epoch": 0.4907636674564013, "grad_norm": 1.4789536484838608, "learning_rate": 1.0784447159760107e-05, "loss": 0.608, "step": 13775 }, { "epoch": 0.4907992945828954, "grad_norm": 1.1737668997905404, "learning_rate": 1.078329677374032e-05, "loss": 0.4542, "step": 13776 }, { "epoch": 0.4908349217093895, "grad_norm": 1.1793462168389424, "learning_rate": 1.0782146377290388e-05, "loss": 0.4441, "step": 13777 }, { "epoch": 0.49087054883588366, "grad_norm": 1.6432429066205192, "learning_rate": 1.0780995970425633e-05, "loss": 0.5822, "step": 13778 }, { "epoch": 0.49090617596237773, "grad_norm": 1.0369225711472176, "learning_rate": 1.0779845553161364e-05, "loss": 0.4334, "step": 13779 }, { "epoch": 0.49094180308887186, "grad_norm": 1.5740232818356155, "learning_rate": 1.077869512551291e-05, "loss": 0.6282, "step": 13780 }, { "epoch": 0.490977430215366, "grad_norm": 1.087671844714553, "learning_rate": 1.0777544687495588e-05, "loss": 0.4578, "step": 13781 }, { "epoch": 0.49101305734186007, "grad_norm": 1.713685436069992, "learning_rate": 1.077639423912471e-05, "loss": 0.5892, "step": 13782 }, { "epoch": 0.4910486844683542, "grad_norm": 1.1077737635922957, "learning_rate": 1.0775243780415606e-05, "loss": 0.5186, "step": 13783 }, { "epoch": 0.49108431159484833, "grad_norm": 1.1596861452506373, "learning_rate": 1.0774093311383587e-05, "loss": 0.4407, "step": 13784 }, { "epoch": 0.4911199387213424, "grad_norm": 1.584359066306663, "learning_rate": 1.0772942832043974e-05, "loss": 0.4583, "step": 13785 }, { "epoch": 0.49115556584783654, "grad_norm": 1.1541130532624344, "learning_rate": 1.0771792342412084e-05, "loss": 0.4757, "step": 13786 }, { "epoch": 0.49119119297433067, "grad_norm": 1.5927646622156326, "learning_rate": 1.0770641842503242e-05, "loss": 0.6014, "step": 13787 }, { "epoch": 0.49122682010082475, "grad_norm": 1.8119147953540176, "learning_rate": 1.0769491332332765e-05, "loss": 0.5117, "step": 13788 }, { "epoch": 0.4912624472273189, "grad_norm": 1.1384687072173356, "learning_rate": 1.0768340811915974e-05, "loss": 0.4683, "step": 13789 }, { "epoch": 0.491298074353813, "grad_norm": 1.0858417878455335, "learning_rate": 1.0767190281268187e-05, "loss": 0.4591, "step": 13790 }, { "epoch": 0.4913337014803071, "grad_norm": 1.178685988173801, "learning_rate": 1.0766039740404726e-05, "loss": 0.5066, "step": 13791 }, { "epoch": 0.4913693286068012, "grad_norm": 1.4941691068380434, "learning_rate": 1.0764889189340911e-05, "loss": 0.5101, "step": 13792 }, { "epoch": 0.49140495573329535, "grad_norm": 1.3507277987067665, "learning_rate": 1.0763738628092061e-05, "loss": 0.5262, "step": 13793 }, { "epoch": 0.4914405828597894, "grad_norm": 1.1320084035797748, "learning_rate": 1.0762588056673497e-05, "loss": 0.5081, "step": 13794 }, { "epoch": 0.49147620998628355, "grad_norm": 1.7885340625941855, "learning_rate": 1.076143747510054e-05, "loss": 0.5443, "step": 13795 }, { "epoch": 0.4915118371127777, "grad_norm": 1.529013555313998, "learning_rate": 1.0760286883388514e-05, "loss": 0.5136, "step": 13796 }, { "epoch": 0.49154746423927176, "grad_norm": 1.10411861520138, "learning_rate": 1.075913628155273e-05, "loss": 0.5011, "step": 13797 }, { "epoch": 0.4915830913657659, "grad_norm": 2.3130790627164517, "learning_rate": 1.075798566960852e-05, "loss": 0.6227, "step": 13798 }, { "epoch": 0.49161871849226, "grad_norm": 1.1333209905511161, "learning_rate": 1.07568350475712e-05, "loss": 0.4019, "step": 13799 }, { "epoch": 0.4916543456187541, "grad_norm": 1.3773122512191767, "learning_rate": 1.0755684415456093e-05, "loss": 0.473, "step": 13800 }, { "epoch": 0.4916899727452482, "grad_norm": 1.0276546317587436, "learning_rate": 1.075453377327852e-05, "loss": 0.4491, "step": 13801 }, { "epoch": 0.49172559987174236, "grad_norm": 1.2260862385670583, "learning_rate": 1.0753383121053801e-05, "loss": 0.4381, "step": 13802 }, { "epoch": 0.49176122699823643, "grad_norm": 1.9212453536465732, "learning_rate": 1.0752232458797262e-05, "loss": 0.5138, "step": 13803 }, { "epoch": 0.49179685412473056, "grad_norm": 1.326223300821452, "learning_rate": 1.0751081786524223e-05, "loss": 0.571, "step": 13804 }, { "epoch": 0.4918324812512247, "grad_norm": 1.463061209139283, "learning_rate": 1.074993110425e-05, "loss": 0.5096, "step": 13805 }, { "epoch": 0.49186810837771877, "grad_norm": 2.4578489363861804, "learning_rate": 1.0748780411989924e-05, "loss": 0.478, "step": 13806 }, { "epoch": 0.4919037355042129, "grad_norm": 1.1658103514549438, "learning_rate": 1.0747629709759313e-05, "loss": 0.4952, "step": 13807 }, { "epoch": 0.49193936263070703, "grad_norm": 1.3587351251480964, "learning_rate": 1.0746478997573486e-05, "loss": 0.5545, "step": 13808 }, { "epoch": 0.4919749897572011, "grad_norm": 1.0595526174976375, "learning_rate": 1.0745328275447774e-05, "loss": 0.4049, "step": 13809 }, { "epoch": 0.49201061688369524, "grad_norm": 1.0579667395552577, "learning_rate": 1.0744177543397492e-05, "loss": 0.4225, "step": 13810 }, { "epoch": 0.49204624401018937, "grad_norm": 1.0443263836416479, "learning_rate": 1.0743026801437968e-05, "loss": 0.4558, "step": 13811 }, { "epoch": 0.49208187113668345, "grad_norm": 1.3571099557728499, "learning_rate": 1.0741876049584523e-05, "loss": 0.44, "step": 13812 }, { "epoch": 0.4921174982631776, "grad_norm": 1.701675360273991, "learning_rate": 1.0740725287852479e-05, "loss": 0.6089, "step": 13813 }, { "epoch": 0.4921531253896717, "grad_norm": 1.6612477991640444, "learning_rate": 1.0739574516257159e-05, "loss": 0.6243, "step": 13814 }, { "epoch": 0.4921887525161658, "grad_norm": 1.6349684457639142, "learning_rate": 1.073842373481389e-05, "loss": 0.5923, "step": 13815 }, { "epoch": 0.4922243796426599, "grad_norm": 2.2410826772688397, "learning_rate": 1.0737272943537989e-05, "loss": 0.536, "step": 13816 }, { "epoch": 0.49226000676915405, "grad_norm": 1.185822373100997, "learning_rate": 1.0736122142444788e-05, "loss": 0.4526, "step": 13817 }, { "epoch": 0.4922956338956481, "grad_norm": 1.1586760802697487, "learning_rate": 1.0734971331549604e-05, "loss": 0.4661, "step": 13818 }, { "epoch": 0.49233126102214225, "grad_norm": 1.1595886220114393, "learning_rate": 1.0733820510867763e-05, "loss": 0.5113, "step": 13819 }, { "epoch": 0.4923668881486364, "grad_norm": 1.0628254616392687, "learning_rate": 1.0732669680414587e-05, "loss": 0.4472, "step": 13820 }, { "epoch": 0.49240251527513046, "grad_norm": 1.0927815845162407, "learning_rate": 1.0731518840205402e-05, "loss": 0.4462, "step": 13821 }, { "epoch": 0.4924381424016246, "grad_norm": 1.4119441041086298, "learning_rate": 1.0730367990255535e-05, "loss": 0.5407, "step": 13822 }, { "epoch": 0.4924737695281187, "grad_norm": 2.4255198789017633, "learning_rate": 1.0729217130580309e-05, "loss": 0.4679, "step": 13823 }, { "epoch": 0.4925093966546128, "grad_norm": 1.1637173411471473, "learning_rate": 1.0728066261195042e-05, "loss": 0.4663, "step": 13824 }, { "epoch": 0.4925450237811069, "grad_norm": 1.3278852152207439, "learning_rate": 1.0726915382115068e-05, "loss": 0.4933, "step": 13825 }, { "epoch": 0.49258065090760106, "grad_norm": 1.1171578232914148, "learning_rate": 1.0725764493355709e-05, "loss": 0.4377, "step": 13826 }, { "epoch": 0.49261627803409513, "grad_norm": 2.1211283620811114, "learning_rate": 1.0724613594932282e-05, "loss": 0.5815, "step": 13827 }, { "epoch": 0.49265190516058927, "grad_norm": 1.098367144414133, "learning_rate": 1.0723462686860124e-05, "loss": 0.4499, "step": 13828 }, { "epoch": 0.4926875322870834, "grad_norm": 1.7084400519380072, "learning_rate": 1.0722311769154557e-05, "loss": 0.5411, "step": 13829 }, { "epoch": 0.49272315941357747, "grad_norm": 1.131283618597988, "learning_rate": 1.07211608418309e-05, "loss": 0.4597, "step": 13830 }, { "epoch": 0.4927587865400716, "grad_norm": 1.188322824950921, "learning_rate": 1.072000990490448e-05, "loss": 0.4915, "step": 13831 }, { "epoch": 0.49279441366656573, "grad_norm": 1.3723355197836304, "learning_rate": 1.0718858958390627e-05, "loss": 0.5307, "step": 13832 }, { "epoch": 0.4928300407930598, "grad_norm": 1.2606369175656897, "learning_rate": 1.0717708002304666e-05, "loss": 0.4517, "step": 13833 }, { "epoch": 0.49286566791955394, "grad_norm": 1.2710963098385584, "learning_rate": 1.071655703666192e-05, "loss": 0.4734, "step": 13834 }, { "epoch": 0.49290129504604807, "grad_norm": 1.6535415284577057, "learning_rate": 1.0715406061477716e-05, "loss": 0.5499, "step": 13835 }, { "epoch": 0.49293692217254215, "grad_norm": 1.520521941204363, "learning_rate": 1.071425507676738e-05, "loss": 0.5012, "step": 13836 }, { "epoch": 0.4929725492990363, "grad_norm": 1.0416249240513573, "learning_rate": 1.0713104082546239e-05, "loss": 0.4962, "step": 13837 }, { "epoch": 0.4930081764255304, "grad_norm": 1.5929499353931411, "learning_rate": 1.0711953078829617e-05, "loss": 0.5222, "step": 13838 }, { "epoch": 0.4930438035520245, "grad_norm": 1.2213975134410588, "learning_rate": 1.0710802065632846e-05, "loss": 0.4565, "step": 13839 }, { "epoch": 0.4930794306785186, "grad_norm": 1.7051376745147582, "learning_rate": 1.0709651042971248e-05, "loss": 0.6045, "step": 13840 }, { "epoch": 0.49311505780501275, "grad_norm": 1.117465210192946, "learning_rate": 1.070850001086015e-05, "loss": 0.4656, "step": 13841 }, { "epoch": 0.4931506849315068, "grad_norm": 1.401123697770492, "learning_rate": 1.0707348969314877e-05, "loss": 0.529, "step": 13842 }, { "epoch": 0.49318631205800095, "grad_norm": 1.1302530030995575, "learning_rate": 1.0706197918350758e-05, "loss": 0.4437, "step": 13843 }, { "epoch": 0.4932219391844951, "grad_norm": 1.107420169061744, "learning_rate": 1.070504685798312e-05, "loss": 0.47, "step": 13844 }, { "epoch": 0.49325756631098916, "grad_norm": 1.2221292780359767, "learning_rate": 1.0703895788227292e-05, "loss": 0.5297, "step": 13845 }, { "epoch": 0.4932931934374833, "grad_norm": 1.0643681348097822, "learning_rate": 1.0702744709098598e-05, "loss": 0.4504, "step": 13846 }, { "epoch": 0.4933288205639774, "grad_norm": 1.6382111571231996, "learning_rate": 1.0701593620612369e-05, "loss": 0.5064, "step": 13847 }, { "epoch": 0.49336444769047155, "grad_norm": 1.3580578226909488, "learning_rate": 1.070044252278393e-05, "loss": 0.5182, "step": 13848 }, { "epoch": 0.49340007481696563, "grad_norm": 1.3931264599094229, "learning_rate": 1.069929141562861e-05, "loss": 0.5079, "step": 13849 }, { "epoch": 0.49343570194345976, "grad_norm": 1.4661511795017663, "learning_rate": 1.0698140299161735e-05, "loss": 0.5906, "step": 13850 }, { "epoch": 0.4934713290699539, "grad_norm": 1.2017821887448015, "learning_rate": 1.0696989173398633e-05, "loss": 0.4191, "step": 13851 }, { "epoch": 0.49350695619644797, "grad_norm": 1.1601976012506223, "learning_rate": 1.0695838038354635e-05, "loss": 0.4859, "step": 13852 }, { "epoch": 0.4935425833229421, "grad_norm": 1.1146723441507063, "learning_rate": 1.0694686894045066e-05, "loss": 0.4724, "step": 13853 }, { "epoch": 0.4935782104494362, "grad_norm": 1.0780470355784877, "learning_rate": 1.0693535740485255e-05, "loss": 0.5067, "step": 13854 }, { "epoch": 0.4936138375759303, "grad_norm": 1.1230804595932118, "learning_rate": 1.0692384577690533e-05, "loss": 0.4678, "step": 13855 }, { "epoch": 0.49364946470242443, "grad_norm": 1.6113601362325902, "learning_rate": 1.0691233405676222e-05, "loss": 0.6735, "step": 13856 }, { "epoch": 0.49368509182891857, "grad_norm": 1.129392212082295, "learning_rate": 1.0690082224457656e-05, "loss": 0.4631, "step": 13857 }, { "epoch": 0.49372071895541264, "grad_norm": 1.6289450451929732, "learning_rate": 1.0688931034050167e-05, "loss": 0.6027, "step": 13858 }, { "epoch": 0.49375634608190677, "grad_norm": 1.131063510629948, "learning_rate": 1.0687779834469079e-05, "loss": 0.4814, "step": 13859 }, { "epoch": 0.4937919732084009, "grad_norm": 1.47169312616936, "learning_rate": 1.068662862572972e-05, "loss": 0.5574, "step": 13860 }, { "epoch": 0.493827600334895, "grad_norm": 1.4560360703641821, "learning_rate": 1.068547740784742e-05, "loss": 0.6047, "step": 13861 }, { "epoch": 0.4938632274613891, "grad_norm": 1.424702912537925, "learning_rate": 1.068432618083751e-05, "loss": 0.5742, "step": 13862 }, { "epoch": 0.49389885458788324, "grad_norm": 1.332261244804147, "learning_rate": 1.068317494471532e-05, "loss": 0.5435, "step": 13863 }, { "epoch": 0.4939344817143773, "grad_norm": 1.1673317849577707, "learning_rate": 1.0682023699496176e-05, "loss": 0.5377, "step": 13864 }, { "epoch": 0.49397010884087145, "grad_norm": 1.0628839309971718, "learning_rate": 1.0680872445195411e-05, "loss": 0.4632, "step": 13865 }, { "epoch": 0.4940057359673656, "grad_norm": 1.3246590328272831, "learning_rate": 1.0679721181828356e-05, "loss": 0.4658, "step": 13866 }, { "epoch": 0.49404136309385965, "grad_norm": 1.1278588175869022, "learning_rate": 1.0678569909410332e-05, "loss": 0.4412, "step": 13867 }, { "epoch": 0.4940769902203538, "grad_norm": 1.2385793955716904, "learning_rate": 1.067741862795668e-05, "loss": 0.4603, "step": 13868 }, { "epoch": 0.4941126173468479, "grad_norm": 1.1895141342777293, "learning_rate": 1.0676267337482723e-05, "loss": 0.4778, "step": 13869 }, { "epoch": 0.494148244473342, "grad_norm": 1.4704906109028828, "learning_rate": 1.0675116038003796e-05, "loss": 0.5487, "step": 13870 }, { "epoch": 0.4941838715998361, "grad_norm": 0.9943012436183243, "learning_rate": 1.0673964729535227e-05, "loss": 0.3936, "step": 13871 }, { "epoch": 0.49421949872633025, "grad_norm": 1.520015765962532, "learning_rate": 1.0672813412092345e-05, "loss": 0.6052, "step": 13872 }, { "epoch": 0.49425512585282433, "grad_norm": 1.6506429691158624, "learning_rate": 1.0671662085690484e-05, "loss": 0.555, "step": 13873 }, { "epoch": 0.49429075297931846, "grad_norm": 1.6054944799146678, "learning_rate": 1.0670510750344971e-05, "loss": 0.5806, "step": 13874 }, { "epoch": 0.4943263801058126, "grad_norm": 1.5401662872595254, "learning_rate": 1.066935940607114e-05, "loss": 0.546, "step": 13875 }, { "epoch": 0.49436200723230667, "grad_norm": 1.1321279930885915, "learning_rate": 1.0668208052884318e-05, "loss": 0.4167, "step": 13876 }, { "epoch": 0.4943976343588008, "grad_norm": 2.300473686473449, "learning_rate": 1.0667056690799844e-05, "loss": 0.5046, "step": 13877 }, { "epoch": 0.49443326148529493, "grad_norm": 1.1699595718867923, "learning_rate": 1.066590531983304e-05, "loss": 0.4852, "step": 13878 }, { "epoch": 0.494468888611789, "grad_norm": 1.0260641386845268, "learning_rate": 1.0664753939999241e-05, "loss": 0.4052, "step": 13879 }, { "epoch": 0.49450451573828313, "grad_norm": 1.5630899720870763, "learning_rate": 1.0663602551313779e-05, "loss": 0.5181, "step": 13880 }, { "epoch": 0.49454014286477727, "grad_norm": 1.455227341248972, "learning_rate": 1.0662451153791987e-05, "loss": 0.5814, "step": 13881 }, { "epoch": 0.49457576999127134, "grad_norm": 1.0550412330419638, "learning_rate": 1.0661299747449194e-05, "loss": 0.4446, "step": 13882 }, { "epoch": 0.49461139711776547, "grad_norm": 1.1188772582818316, "learning_rate": 1.066014833230073e-05, "loss": 0.489, "step": 13883 }, { "epoch": 0.4946470242442596, "grad_norm": 1.6309495024703728, "learning_rate": 1.0658996908361934e-05, "loss": 0.5568, "step": 13884 }, { "epoch": 0.4946826513707537, "grad_norm": 1.1826331281087235, "learning_rate": 1.0657845475648132e-05, "loss": 0.4677, "step": 13885 }, { "epoch": 0.4947182784972478, "grad_norm": 1.2203675728478454, "learning_rate": 1.0656694034174657e-05, "loss": 0.4715, "step": 13886 }, { "epoch": 0.49475390562374194, "grad_norm": 1.1925175631439604, "learning_rate": 1.0655542583956841e-05, "loss": 0.4683, "step": 13887 }, { "epoch": 0.494789532750236, "grad_norm": 1.2255524466501575, "learning_rate": 1.065439112501002e-05, "loss": 0.5176, "step": 13888 }, { "epoch": 0.49482515987673015, "grad_norm": 1.03774602888525, "learning_rate": 1.0653239657349524e-05, "loss": 0.4527, "step": 13889 }, { "epoch": 0.4948607870032243, "grad_norm": 1.110154250828174, "learning_rate": 1.0652088180990681e-05, "loss": 0.4495, "step": 13890 }, { "epoch": 0.49489641412971835, "grad_norm": 1.5901883354658402, "learning_rate": 1.0650936695948827e-05, "loss": 0.6109, "step": 13891 }, { "epoch": 0.4949320412562125, "grad_norm": 1.05848470883509, "learning_rate": 1.0649785202239302e-05, "loss": 0.5076, "step": 13892 }, { "epoch": 0.4949676683827066, "grad_norm": 1.2112876648741762, "learning_rate": 1.064863369987743e-05, "loss": 0.5107, "step": 13893 }, { "epoch": 0.4950032955092007, "grad_norm": 1.4740208844977922, "learning_rate": 1.0647482188878549e-05, "loss": 0.5748, "step": 13894 }, { "epoch": 0.4950389226356948, "grad_norm": 1.6356389924599861, "learning_rate": 1.0646330669257988e-05, "loss": 0.5908, "step": 13895 }, { "epoch": 0.49507454976218895, "grad_norm": 1.4183601572056228, "learning_rate": 1.0645179141031084e-05, "loss": 0.5368, "step": 13896 }, { "epoch": 0.49511017688868303, "grad_norm": 1.0738481337031782, "learning_rate": 1.0644027604213168e-05, "loss": 0.4573, "step": 13897 }, { "epoch": 0.49514580401517716, "grad_norm": 1.473398994694381, "learning_rate": 1.064287605881957e-05, "loss": 0.6015, "step": 13898 }, { "epoch": 0.4951814311416713, "grad_norm": 1.0394087530623912, "learning_rate": 1.0641724504865634e-05, "loss": 0.4414, "step": 13899 }, { "epoch": 0.49521705826816537, "grad_norm": 1.0783773724068226, "learning_rate": 1.0640572942366686e-05, "loss": 0.4636, "step": 13900 }, { "epoch": 0.4952526853946595, "grad_norm": 1.2166058727134168, "learning_rate": 1.0639421371338061e-05, "loss": 0.5653, "step": 13901 }, { "epoch": 0.49528831252115363, "grad_norm": 1.129234885080624, "learning_rate": 1.063826979179509e-05, "loss": 0.4932, "step": 13902 }, { "epoch": 0.4953239396476477, "grad_norm": 1.5565107141477754, "learning_rate": 1.0637118203753116e-05, "loss": 0.4897, "step": 13903 }, { "epoch": 0.49535956677414184, "grad_norm": 1.807245081834955, "learning_rate": 1.0635966607227466e-05, "loss": 0.4996, "step": 13904 }, { "epoch": 0.49539519390063597, "grad_norm": 1.4694707635126905, "learning_rate": 1.0634815002233478e-05, "loss": 0.5458, "step": 13905 }, { "epoch": 0.49543082102713004, "grad_norm": 1.16070291274661, "learning_rate": 1.0633663388786481e-05, "loss": 0.449, "step": 13906 }, { "epoch": 0.4954664481536242, "grad_norm": 0.9915278268856282, "learning_rate": 1.0632511766901815e-05, "loss": 0.4261, "step": 13907 }, { "epoch": 0.4955020752801183, "grad_norm": 1.6064201505415256, "learning_rate": 1.0631360136594813e-05, "loss": 0.5707, "step": 13908 }, { "epoch": 0.4955377024066124, "grad_norm": 1.0782461904219205, "learning_rate": 1.0630208497880807e-05, "loss": 0.4768, "step": 13909 }, { "epoch": 0.4955733295331065, "grad_norm": 1.4652401266341029, "learning_rate": 1.0629056850775136e-05, "loss": 0.5139, "step": 13910 }, { "epoch": 0.49560895665960064, "grad_norm": 1.2919759087813043, "learning_rate": 1.0627905195293135e-05, "loss": 0.4973, "step": 13911 }, { "epoch": 0.4956445837860947, "grad_norm": 1.4296077701129117, "learning_rate": 1.0626753531450138e-05, "loss": 0.5835, "step": 13912 }, { "epoch": 0.49568021091258885, "grad_norm": 1.0834097106774059, "learning_rate": 1.0625601859261477e-05, "loss": 0.4107, "step": 13913 }, { "epoch": 0.495715838039083, "grad_norm": 1.5265403485793545, "learning_rate": 1.062445017874249e-05, "loss": 0.5116, "step": 13914 }, { "epoch": 0.49575146516557705, "grad_norm": 1.6822054145168988, "learning_rate": 1.0623298489908512e-05, "loss": 0.5759, "step": 13915 }, { "epoch": 0.4957870922920712, "grad_norm": 1.4786017928669672, "learning_rate": 1.0622146792774882e-05, "loss": 0.514, "step": 13916 }, { "epoch": 0.4958227194185653, "grad_norm": 1.1134804183803524, "learning_rate": 1.0620995087356929e-05, "loss": 0.4355, "step": 13917 }, { "epoch": 0.4958583465450594, "grad_norm": 1.2061859699647322, "learning_rate": 1.0619843373669993e-05, "loss": 0.4631, "step": 13918 }, { "epoch": 0.4958939736715535, "grad_norm": 1.2161561916311983, "learning_rate": 1.0618691651729412e-05, "loss": 0.4779, "step": 13919 }, { "epoch": 0.49592960079804765, "grad_norm": 1.1326769432883161, "learning_rate": 1.0617539921550515e-05, "loss": 0.4523, "step": 13920 }, { "epoch": 0.49596522792454173, "grad_norm": 1.0543169000309862, "learning_rate": 1.0616388183148646e-05, "loss": 0.4307, "step": 13921 }, { "epoch": 0.49600085505103586, "grad_norm": 1.4521363902190163, "learning_rate": 1.0615236436539137e-05, "loss": 0.5145, "step": 13922 }, { "epoch": 0.49603648217753, "grad_norm": 1.6202499206508794, "learning_rate": 1.0614084681737324e-05, "loss": 0.4163, "step": 13923 }, { "epoch": 0.49607210930402407, "grad_norm": 1.141348911042462, "learning_rate": 1.0612932918758546e-05, "loss": 0.4279, "step": 13924 }, { "epoch": 0.4961077364305182, "grad_norm": 1.5668718276150773, "learning_rate": 1.0611781147618133e-05, "loss": 0.541, "step": 13925 }, { "epoch": 0.49614336355701233, "grad_norm": 1.131582922551901, "learning_rate": 1.061062936833143e-05, "loss": 0.4631, "step": 13926 }, { "epoch": 0.4961789906835064, "grad_norm": 1.1939864055367706, "learning_rate": 1.060947758091377e-05, "loss": 0.4352, "step": 13927 }, { "epoch": 0.49621461781000054, "grad_norm": 1.6624300578737965, "learning_rate": 1.0608325785380488e-05, "loss": 0.5956, "step": 13928 }, { "epoch": 0.49625024493649467, "grad_norm": 1.8060698497744245, "learning_rate": 1.0607173981746926e-05, "loss": 0.4932, "step": 13929 }, { "epoch": 0.49628587206298874, "grad_norm": 1.37223489365675, "learning_rate": 1.0606022170028418e-05, "loss": 0.5958, "step": 13930 }, { "epoch": 0.4963214991894829, "grad_norm": 1.398145716331383, "learning_rate": 1.0604870350240299e-05, "loss": 0.6325, "step": 13931 }, { "epoch": 0.496357126315977, "grad_norm": 1.4646263656631948, "learning_rate": 1.060371852239791e-05, "loss": 0.5937, "step": 13932 }, { "epoch": 0.4963927534424711, "grad_norm": 1.6436765425125022, "learning_rate": 1.0602566686516586e-05, "loss": 0.6678, "step": 13933 }, { "epoch": 0.4964283805689652, "grad_norm": 1.13302509862507, "learning_rate": 1.0601414842611665e-05, "loss": 0.4491, "step": 13934 }, { "epoch": 0.49646400769545934, "grad_norm": 2.2358645858727373, "learning_rate": 1.0600262990698487e-05, "loss": 0.5453, "step": 13935 }, { "epoch": 0.4964996348219534, "grad_norm": 1.2112145083514487, "learning_rate": 1.0599111130792385e-05, "loss": 0.4356, "step": 13936 }, { "epoch": 0.49653526194844755, "grad_norm": 1.317369275155256, "learning_rate": 1.0597959262908702e-05, "loss": 0.456, "step": 13937 }, { "epoch": 0.4965708890749417, "grad_norm": 1.1829319836695447, "learning_rate": 1.0596807387062772e-05, "loss": 0.4557, "step": 13938 }, { "epoch": 0.49660651620143575, "grad_norm": 1.2995461089032658, "learning_rate": 1.0595655503269933e-05, "loss": 0.5079, "step": 13939 }, { "epoch": 0.4966421433279299, "grad_norm": 1.1128237383269264, "learning_rate": 1.0594503611545528e-05, "loss": 0.4842, "step": 13940 }, { "epoch": 0.496677770454424, "grad_norm": 1.0109955192125957, "learning_rate": 1.0593351711904892e-05, "loss": 0.4964, "step": 13941 }, { "epoch": 0.4967133975809181, "grad_norm": 1.5229203893104475, "learning_rate": 1.0592199804363362e-05, "loss": 0.592, "step": 13942 }, { "epoch": 0.4967490247074122, "grad_norm": 1.338608932607628, "learning_rate": 1.0591047888936275e-05, "loss": 0.5246, "step": 13943 }, { "epoch": 0.49678465183390635, "grad_norm": 1.300575892428337, "learning_rate": 1.0589895965638976e-05, "loss": 0.4821, "step": 13944 }, { "epoch": 0.49682027896040043, "grad_norm": 1.4958553954881246, "learning_rate": 1.05887440344868e-05, "loss": 0.5743, "step": 13945 }, { "epoch": 0.49685590608689456, "grad_norm": 1.571385067541201, "learning_rate": 1.058759209549508e-05, "loss": 0.5444, "step": 13946 }, { "epoch": 0.4968915332133887, "grad_norm": 1.1104127120591205, "learning_rate": 1.0586440148679167e-05, "loss": 0.4821, "step": 13947 }, { "epoch": 0.49692716033988277, "grad_norm": 1.115458040498154, "learning_rate": 1.058528819405439e-05, "loss": 0.4409, "step": 13948 }, { "epoch": 0.4969627874663769, "grad_norm": 2.7673317283377257, "learning_rate": 1.0584136231636094e-05, "loss": 0.5174, "step": 13949 }, { "epoch": 0.49699841459287103, "grad_norm": 1.1211676644762338, "learning_rate": 1.0582984261439616e-05, "loss": 0.4806, "step": 13950 }, { "epoch": 0.4970340417193651, "grad_norm": 1.3375797324117134, "learning_rate": 1.0581832283480291e-05, "loss": 0.4336, "step": 13951 }, { "epoch": 0.49706966884585924, "grad_norm": 1.741416019541645, "learning_rate": 1.0580680297773467e-05, "loss": 0.6017, "step": 13952 }, { "epoch": 0.49710529597235337, "grad_norm": 1.159211784732213, "learning_rate": 1.0579528304334477e-05, "loss": 0.4494, "step": 13953 }, { "epoch": 0.49714092309884744, "grad_norm": 1.51752940939705, "learning_rate": 1.0578376303178662e-05, "loss": 0.6017, "step": 13954 }, { "epoch": 0.4971765502253416, "grad_norm": 1.2136878425877744, "learning_rate": 1.0577224294321362e-05, "loss": 0.4464, "step": 13955 }, { "epoch": 0.4972121773518357, "grad_norm": 1.446917970158136, "learning_rate": 1.057607227777792e-05, "loss": 0.528, "step": 13956 }, { "epoch": 0.4972478044783298, "grad_norm": 1.5991353089758025, "learning_rate": 1.0574920253563672e-05, "loss": 0.5866, "step": 13957 }, { "epoch": 0.4972834316048239, "grad_norm": 1.2899345431769904, "learning_rate": 1.0573768221693956e-05, "loss": 0.4732, "step": 13958 }, { "epoch": 0.49731905873131804, "grad_norm": 1.126505028057728, "learning_rate": 1.0572616182184119e-05, "loss": 0.4428, "step": 13959 }, { "epoch": 0.4973546858578121, "grad_norm": 1.8994906761904864, "learning_rate": 1.0571464135049493e-05, "loss": 0.612, "step": 13960 }, { "epoch": 0.49739031298430625, "grad_norm": 1.1551888803417436, "learning_rate": 1.0570312080305426e-05, "loss": 0.4732, "step": 13961 }, { "epoch": 0.4974259401108004, "grad_norm": 2.226786499142823, "learning_rate": 1.0569160017967252e-05, "loss": 0.5527, "step": 13962 }, { "epoch": 0.49746156723729446, "grad_norm": 1.468536314985212, "learning_rate": 1.0568007948050319e-05, "loss": 0.5533, "step": 13963 }, { "epoch": 0.4974971943637886, "grad_norm": 1.0437574538198453, "learning_rate": 1.056685587056996e-05, "loss": 0.3842, "step": 13964 }, { "epoch": 0.4975328214902827, "grad_norm": 1.201186588210285, "learning_rate": 1.056570378554152e-05, "loss": 0.4683, "step": 13965 }, { "epoch": 0.4975684486167768, "grad_norm": 1.6575163136966813, "learning_rate": 1.0564551692980337e-05, "loss": 0.5968, "step": 13966 }, { "epoch": 0.4976040757432709, "grad_norm": 1.2405752494652602, "learning_rate": 1.0563399592901755e-05, "loss": 0.4542, "step": 13967 }, { "epoch": 0.49763970286976505, "grad_norm": 1.748055776832764, "learning_rate": 1.0562247485321116e-05, "loss": 0.5814, "step": 13968 }, { "epoch": 0.49767532999625913, "grad_norm": 1.1951804397783754, "learning_rate": 1.0561095370253754e-05, "loss": 0.5327, "step": 13969 }, { "epoch": 0.49771095712275326, "grad_norm": 1.275623976414679, "learning_rate": 1.0559943247715019e-05, "loss": 0.4844, "step": 13970 }, { "epoch": 0.4977465842492474, "grad_norm": 1.500147117788437, "learning_rate": 1.0558791117720244e-05, "loss": 0.497, "step": 13971 }, { "epoch": 0.49778221137574147, "grad_norm": 1.1579219820348252, "learning_rate": 1.0557638980284778e-05, "loss": 0.3981, "step": 13972 }, { "epoch": 0.4978178385022356, "grad_norm": 1.428930960942949, "learning_rate": 1.0556486835423957e-05, "loss": 0.6172, "step": 13973 }, { "epoch": 0.49785346562872973, "grad_norm": 1.1042253575384053, "learning_rate": 1.0555334683153126e-05, "loss": 0.4694, "step": 13974 }, { "epoch": 0.4978890927552238, "grad_norm": 1.1270881976335394, "learning_rate": 1.0554182523487627e-05, "loss": 0.4541, "step": 13975 }, { "epoch": 0.49792471988171794, "grad_norm": 1.3607535157779076, "learning_rate": 1.0553030356442802e-05, "loss": 0.5026, "step": 13976 }, { "epoch": 0.49796034700821207, "grad_norm": 1.2659637254946023, "learning_rate": 1.0551878182033985e-05, "loss": 0.3969, "step": 13977 }, { "epoch": 0.49799597413470614, "grad_norm": 1.182990532670424, "learning_rate": 1.055072600027653e-05, "loss": 0.4355, "step": 13978 }, { "epoch": 0.4980316012612003, "grad_norm": 1.5452413783084693, "learning_rate": 1.0549573811185772e-05, "loss": 0.5683, "step": 13979 }, { "epoch": 0.4980672283876944, "grad_norm": 1.0836732827747004, "learning_rate": 1.0548421614777051e-05, "loss": 0.4241, "step": 13980 }, { "epoch": 0.4981028555141885, "grad_norm": 1.1031902808693188, "learning_rate": 1.0547269411065718e-05, "loss": 0.473, "step": 13981 }, { "epoch": 0.4981384826406826, "grad_norm": 1.0505355541553862, "learning_rate": 1.0546117200067107e-05, "loss": 0.4756, "step": 13982 }, { "epoch": 0.49817410976717674, "grad_norm": 1.040177714055117, "learning_rate": 1.0544964981796563e-05, "loss": 0.4279, "step": 13983 }, { "epoch": 0.4982097368936708, "grad_norm": 1.279776419026242, "learning_rate": 1.054381275626943e-05, "loss": 0.5857, "step": 13984 }, { "epoch": 0.49824536402016495, "grad_norm": 1.5287431633735105, "learning_rate": 1.0542660523501051e-05, "loss": 0.4781, "step": 13985 }, { "epoch": 0.4982809911466591, "grad_norm": 1.2297464706312842, "learning_rate": 1.054150828350677e-05, "loss": 0.5048, "step": 13986 }, { "epoch": 0.49831661827315316, "grad_norm": 1.363605315113161, "learning_rate": 1.0540356036301924e-05, "loss": 0.4774, "step": 13987 }, { "epoch": 0.4983522453996473, "grad_norm": 1.3560896589771738, "learning_rate": 1.053920378190186e-05, "loss": 0.6231, "step": 13988 }, { "epoch": 0.4983878725261414, "grad_norm": 1.5149471537169201, "learning_rate": 1.0538051520321924e-05, "loss": 0.5859, "step": 13989 }, { "epoch": 0.4984234996526355, "grad_norm": 1.637804989158637, "learning_rate": 1.0536899251577455e-05, "loss": 0.5209, "step": 13990 }, { "epoch": 0.4984591267791296, "grad_norm": 1.638837653116958, "learning_rate": 1.0535746975683793e-05, "loss": 0.666, "step": 13991 }, { "epoch": 0.49849475390562376, "grad_norm": 1.5631786268921946, "learning_rate": 1.0534594692656291e-05, "loss": 0.5701, "step": 13992 }, { "epoch": 0.49853038103211783, "grad_norm": 1.4508388394967768, "learning_rate": 1.0533442402510283e-05, "loss": 0.6204, "step": 13993 }, { "epoch": 0.49856600815861196, "grad_norm": 1.4922174934608883, "learning_rate": 1.0532290105261118e-05, "loss": 0.5478, "step": 13994 }, { "epoch": 0.4986016352851061, "grad_norm": 1.6795699047444645, "learning_rate": 1.0531137800924136e-05, "loss": 0.588, "step": 13995 }, { "epoch": 0.49863726241160017, "grad_norm": 1.5643980492022387, "learning_rate": 1.0529985489514687e-05, "loss": 0.5211, "step": 13996 }, { "epoch": 0.4986728895380943, "grad_norm": 1.0012923898084765, "learning_rate": 1.0528833171048109e-05, "loss": 0.4254, "step": 13997 }, { "epoch": 0.49870851666458843, "grad_norm": 1.0696594629054208, "learning_rate": 1.052768084553975e-05, "loss": 0.4985, "step": 13998 }, { "epoch": 0.4987441437910825, "grad_norm": 1.4252751596014936, "learning_rate": 1.0526528513004947e-05, "loss": 0.5971, "step": 13999 }, { "epoch": 0.49877977091757664, "grad_norm": 1.6595052082086172, "learning_rate": 1.0525376173459053e-05, "loss": 0.5287, "step": 14000 }, { "epoch": 0.49881539804407077, "grad_norm": 1.703101516103301, "learning_rate": 1.0524223826917405e-05, "loss": 0.5482, "step": 14001 }, { "epoch": 0.49885102517056484, "grad_norm": 1.7962989062663544, "learning_rate": 1.0523071473395352e-05, "loss": 0.5917, "step": 14002 }, { "epoch": 0.498886652297059, "grad_norm": 1.101253289220233, "learning_rate": 1.0521919112908237e-05, "loss": 0.4563, "step": 14003 }, { "epoch": 0.4989222794235531, "grad_norm": 1.4049688565724077, "learning_rate": 1.0520766745471404e-05, "loss": 0.4925, "step": 14004 }, { "epoch": 0.4989579065500472, "grad_norm": 1.1103496831419362, "learning_rate": 1.0519614371100198e-05, "loss": 0.4645, "step": 14005 }, { "epoch": 0.4989935336765413, "grad_norm": 1.1615276383215123, "learning_rate": 1.0518461989809962e-05, "loss": 0.4591, "step": 14006 }, { "epoch": 0.49902916080303544, "grad_norm": 1.72954131137137, "learning_rate": 1.0517309601616042e-05, "loss": 0.5964, "step": 14007 }, { "epoch": 0.4990647879295295, "grad_norm": 1.1133861196154733, "learning_rate": 1.0516157206533783e-05, "loss": 0.4381, "step": 14008 }, { "epoch": 0.49910041505602365, "grad_norm": 1.8028997120027257, "learning_rate": 1.0515004804578532e-05, "loss": 0.5118, "step": 14009 }, { "epoch": 0.4991360421825178, "grad_norm": 1.952876718169862, "learning_rate": 1.0513852395765631e-05, "loss": 0.5501, "step": 14010 }, { "epoch": 0.49917166930901186, "grad_norm": 1.3372882872529313, "learning_rate": 1.0512699980110428e-05, "loss": 0.4893, "step": 14011 }, { "epoch": 0.499207296435506, "grad_norm": 1.1710860880408496, "learning_rate": 1.0511547557628265e-05, "loss": 0.4461, "step": 14012 }, { "epoch": 0.4992429235620001, "grad_norm": 1.0904307239766053, "learning_rate": 1.051039512833449e-05, "loss": 0.464, "step": 14013 }, { "epoch": 0.4992785506884942, "grad_norm": 1.2112182784839989, "learning_rate": 1.0509242692244443e-05, "loss": 0.4445, "step": 14014 }, { "epoch": 0.4993141778149883, "grad_norm": 1.107628181181382, "learning_rate": 1.0508090249373477e-05, "loss": 0.4278, "step": 14015 }, { "epoch": 0.49934980494148246, "grad_norm": 1.8083779208778383, "learning_rate": 1.0506937799736935e-05, "loss": 0.5512, "step": 14016 }, { "epoch": 0.49938543206797653, "grad_norm": 1.0364463204705403, "learning_rate": 1.050578534335016e-05, "loss": 0.4481, "step": 14017 }, { "epoch": 0.49942105919447066, "grad_norm": 1.1672786015407053, "learning_rate": 1.0504632880228497e-05, "loss": 0.437, "step": 14018 }, { "epoch": 0.4994566863209648, "grad_norm": 1.7716239761898769, "learning_rate": 1.05034804103873e-05, "loss": 0.5847, "step": 14019 }, { "epoch": 0.49949231344745887, "grad_norm": 1.399480388252268, "learning_rate": 1.0502327933841907e-05, "loss": 0.5388, "step": 14020 }, { "epoch": 0.499527940573953, "grad_norm": 1.0465634618651387, "learning_rate": 1.0501175450607664e-05, "loss": 0.3994, "step": 14021 }, { "epoch": 0.49956356770044713, "grad_norm": 1.5951056835546944, "learning_rate": 1.0500022960699923e-05, "loss": 0.5399, "step": 14022 }, { "epoch": 0.4995991948269412, "grad_norm": 1.192576455622685, "learning_rate": 1.0498870464134026e-05, "loss": 0.4833, "step": 14023 }, { "epoch": 0.49963482195343534, "grad_norm": 1.133116163838276, "learning_rate": 1.0497717960925322e-05, "loss": 0.426, "step": 14024 }, { "epoch": 0.49967044907992947, "grad_norm": 1.0700824580264527, "learning_rate": 1.049656545108915e-05, "loss": 0.4533, "step": 14025 }, { "epoch": 0.49970607620642354, "grad_norm": 1.251622317857934, "learning_rate": 1.0495412934640868e-05, "loss": 0.4774, "step": 14026 }, { "epoch": 0.4997417033329177, "grad_norm": 1.101187779777075, "learning_rate": 1.0494260411595814e-05, "loss": 0.4609, "step": 14027 }, { "epoch": 0.4997773304594118, "grad_norm": 2.2269708629225895, "learning_rate": 1.0493107881969335e-05, "loss": 0.5755, "step": 14028 }, { "epoch": 0.4998129575859059, "grad_norm": 1.1483645261023758, "learning_rate": 1.0491955345776782e-05, "loss": 0.4692, "step": 14029 }, { "epoch": 0.4998485847124, "grad_norm": 1.2389068206252627, "learning_rate": 1.0490802803033497e-05, "loss": 0.4831, "step": 14030 }, { "epoch": 0.49988421183889414, "grad_norm": 1.7917209418670905, "learning_rate": 1.0489650253754836e-05, "loss": 0.5439, "step": 14031 }, { "epoch": 0.4999198389653882, "grad_norm": 1.193936367605213, "learning_rate": 1.0488497697956134e-05, "loss": 0.4746, "step": 14032 }, { "epoch": 0.49995546609188235, "grad_norm": 1.5414095756235147, "learning_rate": 1.0487345135652745e-05, "loss": 0.523, "step": 14033 }, { "epoch": 0.4999910932183765, "grad_norm": 1.1991680953613577, "learning_rate": 1.0486192566860016e-05, "loss": 0.5951, "step": 14034 }, { "epoch": 0.5000267203448706, "grad_norm": 1.0961628916916024, "learning_rate": 1.0485039991593293e-05, "loss": 0.4911, "step": 14035 }, { "epoch": 0.5000623474713647, "grad_norm": 1.1857423451943527, "learning_rate": 1.0483887409867921e-05, "loss": 0.5054, "step": 14036 }, { "epoch": 0.5000979745978588, "grad_norm": 1.0416604081869998, "learning_rate": 1.0482734821699254e-05, "loss": 0.4779, "step": 14037 }, { "epoch": 0.500133601724353, "grad_norm": 1.3253330838942048, "learning_rate": 1.0481582227102634e-05, "loss": 0.5219, "step": 14038 }, { "epoch": 0.500169228850847, "grad_norm": 1.1413383952433735, "learning_rate": 1.0480429626093408e-05, "loss": 0.4247, "step": 14039 }, { "epoch": 0.5002048559773411, "grad_norm": 1.2093265273410334, "learning_rate": 1.0479277018686926e-05, "loss": 0.4565, "step": 14040 }, { "epoch": 0.5002404831038353, "grad_norm": 1.4455269597351839, "learning_rate": 1.0478124404898535e-05, "loss": 0.5155, "step": 14041 }, { "epoch": 0.5002761102303294, "grad_norm": 2.098932433036943, "learning_rate": 1.0476971784743586e-05, "loss": 0.4907, "step": 14042 }, { "epoch": 0.5003117373568234, "grad_norm": 1.659325868681806, "learning_rate": 1.0475819158237426e-05, "loss": 0.6278, "step": 14043 }, { "epoch": 0.5003473644833176, "grad_norm": 1.1793826290766274, "learning_rate": 1.0474666525395396e-05, "loss": 0.4573, "step": 14044 }, { "epoch": 0.5003829916098117, "grad_norm": 1.4995985697116594, "learning_rate": 1.0473513886232854e-05, "loss": 0.5795, "step": 14045 }, { "epoch": 0.5004186187363058, "grad_norm": 1.2817330987659912, "learning_rate": 1.0472361240765141e-05, "loss": 0.4835, "step": 14046 }, { "epoch": 0.5004542458628, "grad_norm": 1.6065091347845368, "learning_rate": 1.0471208589007607e-05, "loss": 0.4256, "step": 14047 }, { "epoch": 0.500489872989294, "grad_norm": 1.1417586204961498, "learning_rate": 1.0470055930975605e-05, "loss": 0.4951, "step": 14048 }, { "epoch": 0.5005255001157881, "grad_norm": 1.157490432242979, "learning_rate": 1.0468903266684478e-05, "loss": 0.4711, "step": 14049 }, { "epoch": 0.5005611272422823, "grad_norm": 1.2438309274611528, "learning_rate": 1.0467750596149576e-05, "loss": 0.5756, "step": 14050 }, { "epoch": 0.5005967543687764, "grad_norm": 1.116391368262272, "learning_rate": 1.0466597919386245e-05, "loss": 0.4034, "step": 14051 }, { "epoch": 0.5006323814952705, "grad_norm": 1.3275069142820857, "learning_rate": 1.046544523640984e-05, "loss": 0.4363, "step": 14052 }, { "epoch": 0.5006680086217646, "grad_norm": 1.2697389740355787, "learning_rate": 1.0464292547235706e-05, "loss": 0.445, "step": 14053 }, { "epoch": 0.5007036357482587, "grad_norm": 1.7162756192244775, "learning_rate": 1.0463139851879192e-05, "loss": 0.579, "step": 14054 }, { "epoch": 0.5007392628747528, "grad_norm": 1.1926014024818063, "learning_rate": 1.0461987150355646e-05, "loss": 0.4708, "step": 14055 }, { "epoch": 0.500774890001247, "grad_norm": 1.548375062213018, "learning_rate": 1.0460834442680419e-05, "loss": 0.5839, "step": 14056 }, { "epoch": 0.500810517127741, "grad_norm": 1.2425933626220527, "learning_rate": 1.045968172886886e-05, "loss": 0.4765, "step": 14057 }, { "epoch": 0.5008461442542351, "grad_norm": 1.619027428081004, "learning_rate": 1.0458529008936317e-05, "loss": 0.4207, "step": 14058 }, { "epoch": 0.5008817713807293, "grad_norm": 1.3529696471021497, "learning_rate": 1.0457376282898139e-05, "loss": 0.4353, "step": 14059 }, { "epoch": 0.5009173985072234, "grad_norm": 1.777395533488121, "learning_rate": 1.0456223550769677e-05, "loss": 0.5022, "step": 14060 }, { "epoch": 0.5009530256337175, "grad_norm": 1.4726558671817986, "learning_rate": 1.0455070812566281e-05, "loss": 0.5237, "step": 14061 }, { "epoch": 0.5009886527602117, "grad_norm": 1.319528491535305, "learning_rate": 1.0453918068303298e-05, "loss": 0.4566, "step": 14062 }, { "epoch": 0.5010242798867057, "grad_norm": 1.435866583780284, "learning_rate": 1.0452765317996076e-05, "loss": 0.5361, "step": 14063 }, { "epoch": 0.5010599070131998, "grad_norm": 1.582781437111165, "learning_rate": 1.045161256165997e-05, "loss": 0.5842, "step": 14064 }, { "epoch": 0.501095534139694, "grad_norm": 1.49985515924028, "learning_rate": 1.0450459799310325e-05, "loss": 0.5595, "step": 14065 }, { "epoch": 0.5011311612661881, "grad_norm": 1.2993220554191076, "learning_rate": 1.0449307030962494e-05, "loss": 0.4501, "step": 14066 }, { "epoch": 0.5011667883926821, "grad_norm": 3.3734791313897947, "learning_rate": 1.0448154256631828e-05, "loss": 0.6054, "step": 14067 }, { "epoch": 0.5012024155191763, "grad_norm": 1.475213560037187, "learning_rate": 1.0447001476333673e-05, "loss": 0.4804, "step": 14068 }, { "epoch": 0.5012380426456704, "grad_norm": 1.3153667893755763, "learning_rate": 1.0445848690083382e-05, "loss": 0.4812, "step": 14069 }, { "epoch": 0.5012736697721645, "grad_norm": 5.894417773298889, "learning_rate": 1.0444695897896301e-05, "loss": 0.5706, "step": 14070 }, { "epoch": 0.5013092968986587, "grad_norm": 1.2823973583315906, "learning_rate": 1.0443543099787788e-05, "loss": 0.4274, "step": 14071 }, { "epoch": 0.5013449240251527, "grad_norm": 1.730878252753401, "learning_rate": 1.0442390295773186e-05, "loss": 0.4864, "step": 14072 }, { "epoch": 0.5013805511516468, "grad_norm": 1.146110714499905, "learning_rate": 1.0441237485867845e-05, "loss": 0.48, "step": 14073 }, { "epoch": 0.501416178278141, "grad_norm": 1.3980703590069778, "learning_rate": 1.0440084670087122e-05, "loss": 0.5235, "step": 14074 }, { "epoch": 0.5014518054046351, "grad_norm": 1.6644709536186542, "learning_rate": 1.0438931848446363e-05, "loss": 0.6187, "step": 14075 }, { "epoch": 0.5014874325311292, "grad_norm": 1.5621166989773088, "learning_rate": 1.043777902096092e-05, "loss": 0.5744, "step": 14076 }, { "epoch": 0.5015230596576233, "grad_norm": 1.1132629136805123, "learning_rate": 1.0436626187646143e-05, "loss": 0.4721, "step": 14077 }, { "epoch": 0.5015586867841174, "grad_norm": 1.421038591774257, "learning_rate": 1.0435473348517384e-05, "loss": 0.5428, "step": 14078 }, { "epoch": 0.5015943139106115, "grad_norm": 1.2652365572561395, "learning_rate": 1.0434320503589994e-05, "loss": 0.4842, "step": 14079 }, { "epoch": 0.5016299410371057, "grad_norm": 1.1398452282129903, "learning_rate": 1.0433167652879321e-05, "loss": 0.3867, "step": 14080 }, { "epoch": 0.5016655681635998, "grad_norm": 1.4978142913035346, "learning_rate": 1.0432014796400719e-05, "loss": 0.4974, "step": 14081 }, { "epoch": 0.5017011952900938, "grad_norm": 1.4052248284461002, "learning_rate": 1.0430861934169536e-05, "loss": 0.5487, "step": 14082 }, { "epoch": 0.501736822416588, "grad_norm": 2.007532521885934, "learning_rate": 1.0429709066201128e-05, "loss": 0.5368, "step": 14083 }, { "epoch": 0.5017724495430821, "grad_norm": 1.3594985841074991, "learning_rate": 1.0428556192510841e-05, "loss": 0.5389, "step": 14084 }, { "epoch": 0.5018080766695762, "grad_norm": 1.2861645671067743, "learning_rate": 1.0427403313114031e-05, "loss": 0.4487, "step": 14085 }, { "epoch": 0.5018437037960704, "grad_norm": 1.1704045846357727, "learning_rate": 1.0426250428026045e-05, "loss": 0.4463, "step": 14086 }, { "epoch": 0.5018793309225644, "grad_norm": 1.7332855172092128, "learning_rate": 1.0425097537262236e-05, "loss": 0.4815, "step": 14087 }, { "epoch": 0.5019149580490585, "grad_norm": 1.5924924451322235, "learning_rate": 1.0423944640837959e-05, "loss": 0.5771, "step": 14088 }, { "epoch": 0.5019505851755527, "grad_norm": 1.3857198229955117, "learning_rate": 1.042279173876856e-05, "loss": 0.63, "step": 14089 }, { "epoch": 0.5019862123020468, "grad_norm": 1.15875891574526, "learning_rate": 1.0421638831069396e-05, "loss": 0.4349, "step": 14090 }, { "epoch": 0.5020218394285408, "grad_norm": 1.2420984538170512, "learning_rate": 1.0420485917755816e-05, "loss": 0.4873, "step": 14091 }, { "epoch": 0.502057466555035, "grad_norm": 1.1589001847467386, "learning_rate": 1.041933299884317e-05, "loss": 0.4433, "step": 14092 }, { "epoch": 0.5020930936815291, "grad_norm": 1.096169887698623, "learning_rate": 1.0418180074346816e-05, "loss": 0.4059, "step": 14093 }, { "epoch": 0.5021287208080232, "grad_norm": 1.120331476152419, "learning_rate": 1.0417027144282099e-05, "loss": 0.4472, "step": 14094 }, { "epoch": 0.5021643479345174, "grad_norm": 1.4872798816687978, "learning_rate": 1.0415874208664376e-05, "loss": 0.55, "step": 14095 }, { "epoch": 0.5021999750610114, "grad_norm": 1.514539081785631, "learning_rate": 1.0414721267508995e-05, "loss": 0.6148, "step": 14096 }, { "epoch": 0.5022356021875055, "grad_norm": 1.05778267721067, "learning_rate": 1.041356832083131e-05, "loss": 0.4359, "step": 14097 }, { "epoch": 0.5022712293139997, "grad_norm": 1.4757074893717226, "learning_rate": 1.0412415368646674e-05, "loss": 0.4885, "step": 14098 }, { "epoch": 0.5023068564404938, "grad_norm": 1.5893077159228364, "learning_rate": 1.041126241097044e-05, "loss": 0.5396, "step": 14099 }, { "epoch": 0.5023424835669879, "grad_norm": 1.1785741488585315, "learning_rate": 1.0410109447817959e-05, "loss": 0.4581, "step": 14100 }, { "epoch": 0.502378110693482, "grad_norm": 1.1845458489081209, "learning_rate": 1.0408956479204586e-05, "loss": 0.4774, "step": 14101 }, { "epoch": 0.5024137378199761, "grad_norm": 1.5060834261435532, "learning_rate": 1.0407803505145671e-05, "loss": 0.5414, "step": 14102 }, { "epoch": 0.5024493649464702, "grad_norm": 1.3752331746964295, "learning_rate": 1.0406650525656565e-05, "loss": 0.6144, "step": 14103 }, { "epoch": 0.5024849920729644, "grad_norm": 2.0141722041911425, "learning_rate": 1.0405497540752624e-05, "loss": 0.6231, "step": 14104 }, { "epoch": 0.5025206191994585, "grad_norm": 2.086055661352783, "learning_rate": 1.04043445504492e-05, "loss": 0.592, "step": 14105 }, { "epoch": 0.5025562463259525, "grad_norm": 1.3221673999742287, "learning_rate": 1.040319155476165e-05, "loss": 0.4553, "step": 14106 }, { "epoch": 0.5025918734524467, "grad_norm": 1.1378215278232422, "learning_rate": 1.0402038553705316e-05, "loss": 0.5406, "step": 14107 }, { "epoch": 0.5026275005789408, "grad_norm": 1.0929661395020798, "learning_rate": 1.0400885547295561e-05, "loss": 0.4304, "step": 14108 }, { "epoch": 0.5026631277054349, "grad_norm": 1.4040717591710261, "learning_rate": 1.0399732535547735e-05, "loss": 0.5537, "step": 14109 }, { "epoch": 0.502698754831929, "grad_norm": 1.194282071423964, "learning_rate": 1.0398579518477188e-05, "loss": 0.4553, "step": 14110 }, { "epoch": 0.5027343819584231, "grad_norm": 1.0665564560460523, "learning_rate": 1.0397426496099277e-05, "loss": 0.4347, "step": 14111 }, { "epoch": 0.5027700090849172, "grad_norm": 1.3925513985831832, "learning_rate": 1.0396273468429356e-05, "loss": 0.5698, "step": 14112 }, { "epoch": 0.5028056362114114, "grad_norm": 1.1586163362904263, "learning_rate": 1.0395120435482777e-05, "loss": 0.4257, "step": 14113 }, { "epoch": 0.5028412633379055, "grad_norm": 1.256416006214548, "learning_rate": 1.0393967397274894e-05, "loss": 0.4227, "step": 14114 }, { "epoch": 0.5028768904643995, "grad_norm": 1.3056066688078154, "learning_rate": 1.0392814353821058e-05, "loss": 0.5136, "step": 14115 }, { "epoch": 0.5029125175908937, "grad_norm": 1.1379185176709095, "learning_rate": 1.0391661305136625e-05, "loss": 0.4493, "step": 14116 }, { "epoch": 0.5029481447173878, "grad_norm": 1.7226701838030374, "learning_rate": 1.0390508251236948e-05, "loss": 0.5744, "step": 14117 }, { "epoch": 0.5029837718438819, "grad_norm": 0.9891773265016727, "learning_rate": 1.0389355192137379e-05, "loss": 0.4286, "step": 14118 }, { "epoch": 0.5030193989703761, "grad_norm": 1.052464717821669, "learning_rate": 1.0388202127853274e-05, "loss": 0.471, "step": 14119 }, { "epoch": 0.5030550260968701, "grad_norm": 1.0586579929165176, "learning_rate": 1.0387049058399989e-05, "loss": 0.4613, "step": 14120 }, { "epoch": 0.5030906532233642, "grad_norm": 1.6664606930018777, "learning_rate": 1.038589598379287e-05, "loss": 0.6448, "step": 14121 }, { "epoch": 0.5031262803498584, "grad_norm": 1.6741077017280959, "learning_rate": 1.0384742904047281e-05, "loss": 0.6017, "step": 14122 }, { "epoch": 0.5031619074763525, "grad_norm": 1.1104468267537986, "learning_rate": 1.038358981917857e-05, "loss": 0.4095, "step": 14123 }, { "epoch": 0.5031975346028466, "grad_norm": 1.5553688651346138, "learning_rate": 1.0382436729202092e-05, "loss": 0.5097, "step": 14124 }, { "epoch": 0.5032331617293407, "grad_norm": 1.1362657761917236, "learning_rate": 1.0381283634133203e-05, "loss": 0.443, "step": 14125 }, { "epoch": 0.5032687888558348, "grad_norm": 1.1612773060682087, "learning_rate": 1.0380130533987254e-05, "loss": 0.4618, "step": 14126 }, { "epoch": 0.5033044159823289, "grad_norm": 1.5921672920773178, "learning_rate": 1.0378977428779601e-05, "loss": 0.6033, "step": 14127 }, { "epoch": 0.5033400431088231, "grad_norm": 1.5543831690962586, "learning_rate": 1.0377824318525603e-05, "loss": 0.5024, "step": 14128 }, { "epoch": 0.5033756702353172, "grad_norm": 1.237337279158276, "learning_rate": 1.0376671203240605e-05, "loss": 0.4202, "step": 14129 }, { "epoch": 0.5034112973618113, "grad_norm": 1.6757563093858698, "learning_rate": 1.0375518082939969e-05, "loss": 0.5497, "step": 14130 }, { "epoch": 0.5034469244883054, "grad_norm": 1.0160265609545749, "learning_rate": 1.0374364957639047e-05, "loss": 0.4162, "step": 14131 }, { "epoch": 0.5034825516147995, "grad_norm": 1.1794428292227845, "learning_rate": 1.0373211827353195e-05, "loss": 0.4521, "step": 14132 }, { "epoch": 0.5035181787412937, "grad_norm": 1.1788137436669979, "learning_rate": 1.0372058692097764e-05, "loss": 0.4448, "step": 14133 }, { "epoch": 0.5035538058677878, "grad_norm": 1.2639924968315206, "learning_rate": 1.0370905551888111e-05, "loss": 0.4723, "step": 14134 }, { "epoch": 0.5035894329942818, "grad_norm": 1.1251608953484722, "learning_rate": 1.0369752406739594e-05, "loss": 0.4529, "step": 14135 }, { "epoch": 0.503625060120776, "grad_norm": 1.5598481364527084, "learning_rate": 1.0368599256667564e-05, "loss": 0.5843, "step": 14136 }, { "epoch": 0.5036606872472701, "grad_norm": 1.735013468343572, "learning_rate": 1.0367446101687377e-05, "loss": 0.6636, "step": 14137 }, { "epoch": 0.5036963143737642, "grad_norm": 1.7034389308446873, "learning_rate": 1.0366292941814392e-05, "loss": 0.6262, "step": 14138 }, { "epoch": 0.5037319415002584, "grad_norm": 4.009109569116351, "learning_rate": 1.0365139777063957e-05, "loss": 0.5557, "step": 14139 }, { "epoch": 0.5037675686267524, "grad_norm": 1.4769850271942264, "learning_rate": 1.0363986607451431e-05, "loss": 0.5707, "step": 14140 }, { "epoch": 0.5038031957532465, "grad_norm": 1.3637064261399212, "learning_rate": 1.0362833432992167e-05, "loss": 0.4865, "step": 14141 }, { "epoch": 0.5038388228797407, "grad_norm": 1.6007607775674948, "learning_rate": 1.0361680253701528e-05, "loss": 0.5544, "step": 14142 }, { "epoch": 0.5038744500062348, "grad_norm": 1.098065461772462, "learning_rate": 1.0360527069594859e-05, "loss": 0.4052, "step": 14143 }, { "epoch": 0.5039100771327288, "grad_norm": 1.3128985692131308, "learning_rate": 1.035937388068752e-05, "loss": 0.4615, "step": 14144 }, { "epoch": 0.503945704259223, "grad_norm": 2.1314694477877185, "learning_rate": 1.0358220686994864e-05, "loss": 0.5562, "step": 14145 }, { "epoch": 0.5039813313857171, "grad_norm": 1.4395748058683835, "learning_rate": 1.0357067488532255e-05, "loss": 0.5422, "step": 14146 }, { "epoch": 0.5040169585122112, "grad_norm": 1.2946612920728933, "learning_rate": 1.035591428531504e-05, "loss": 0.4499, "step": 14147 }, { "epoch": 0.5040525856387054, "grad_norm": 1.1110207274871482, "learning_rate": 1.0354761077358577e-05, "loss": 0.458, "step": 14148 }, { "epoch": 0.5040882127651994, "grad_norm": 1.9389722112767198, "learning_rate": 1.0353607864678225e-05, "loss": 0.5651, "step": 14149 }, { "epoch": 0.5041238398916935, "grad_norm": 1.439020577547995, "learning_rate": 1.0352454647289336e-05, "loss": 0.605, "step": 14150 }, { "epoch": 0.5041594670181877, "grad_norm": 1.181610468859919, "learning_rate": 1.0351301425207268e-05, "loss": 0.4393, "step": 14151 }, { "epoch": 0.5041950941446818, "grad_norm": 1.510341759480572, "learning_rate": 1.0350148198447372e-05, "loss": 0.545, "step": 14152 }, { "epoch": 0.5042307212711759, "grad_norm": 2.1325092838704522, "learning_rate": 1.0348994967025012e-05, "loss": 0.5256, "step": 14153 }, { "epoch": 0.50426634839767, "grad_norm": 1.9584907004183028, "learning_rate": 1.0347841730955539e-05, "loss": 0.585, "step": 14154 }, { "epoch": 0.5043019755241641, "grad_norm": 1.2426713315702251, "learning_rate": 1.0346688490254309e-05, "loss": 0.4324, "step": 14155 }, { "epoch": 0.5043376026506582, "grad_norm": 1.6401704486372861, "learning_rate": 1.0345535244936677e-05, "loss": 0.5048, "step": 14156 }, { "epoch": 0.5043732297771524, "grad_norm": 1.2291485942999232, "learning_rate": 1.0344381995018008e-05, "loss": 0.4808, "step": 14157 }, { "epoch": 0.5044088569036465, "grad_norm": 1.126546044157853, "learning_rate": 1.034322874051365e-05, "loss": 0.4634, "step": 14158 }, { "epoch": 0.5044444840301405, "grad_norm": 1.8736425084978268, "learning_rate": 1.0342075481438959e-05, "loss": 0.4695, "step": 14159 }, { "epoch": 0.5044801111566347, "grad_norm": 1.6529784720449512, "learning_rate": 1.0340922217809294e-05, "loss": 0.5048, "step": 14160 }, { "epoch": 0.5045157382831288, "grad_norm": 2.0238700431984786, "learning_rate": 1.0339768949640014e-05, "loss": 0.511, "step": 14161 }, { "epoch": 0.5045513654096229, "grad_norm": 1.1773926754157502, "learning_rate": 1.0338615676946472e-05, "loss": 0.433, "step": 14162 }, { "epoch": 0.504586992536117, "grad_norm": 2.2543843982265543, "learning_rate": 1.0337462399744025e-05, "loss": 0.5004, "step": 14163 }, { "epoch": 0.5046226196626111, "grad_norm": 1.405078652658911, "learning_rate": 1.0336309118048028e-05, "loss": 0.4449, "step": 14164 }, { "epoch": 0.5046582467891052, "grad_norm": 1.2583642622811269, "learning_rate": 1.0335155831873845e-05, "loss": 0.4635, "step": 14165 }, { "epoch": 0.5046938739155994, "grad_norm": 1.9943964283021842, "learning_rate": 1.0334002541236825e-05, "loss": 0.5715, "step": 14166 }, { "epoch": 0.5047295010420935, "grad_norm": 1.2870558865458468, "learning_rate": 1.0332849246152326e-05, "loss": 0.4697, "step": 14167 }, { "epoch": 0.5047651281685875, "grad_norm": 1.1582705342471993, "learning_rate": 1.0331695946635708e-05, "loss": 0.4816, "step": 14168 }, { "epoch": 0.5048007552950817, "grad_norm": 1.2420350297409015, "learning_rate": 1.0330542642702327e-05, "loss": 0.4767, "step": 14169 }, { "epoch": 0.5048363824215758, "grad_norm": 1.1660552756790719, "learning_rate": 1.0329389334367543e-05, "loss": 0.4676, "step": 14170 }, { "epoch": 0.5048720095480699, "grad_norm": 3.0036231159106883, "learning_rate": 1.0328236021646705e-05, "loss": 0.5769, "step": 14171 }, { "epoch": 0.5049076366745641, "grad_norm": 1.2099226362447204, "learning_rate": 1.0327082704555177e-05, "loss": 0.4482, "step": 14172 }, { "epoch": 0.5049432638010581, "grad_norm": 1.1372429512045086, "learning_rate": 1.0325929383108314e-05, "loss": 0.4734, "step": 14173 }, { "epoch": 0.5049788909275522, "grad_norm": 1.2078621599347075, "learning_rate": 1.0324776057321473e-05, "loss": 0.4708, "step": 14174 }, { "epoch": 0.5050145180540464, "grad_norm": 1.0983845893311706, "learning_rate": 1.0323622727210012e-05, "loss": 0.3961, "step": 14175 }, { "epoch": 0.5050501451805405, "grad_norm": 1.38051933348836, "learning_rate": 1.032246939278929e-05, "loss": 0.5911, "step": 14176 }, { "epoch": 0.5050857723070346, "grad_norm": 2.0850401083946424, "learning_rate": 1.0321316054074662e-05, "loss": 0.482, "step": 14177 }, { "epoch": 0.5051213994335287, "grad_norm": 1.0669130816441752, "learning_rate": 1.0320162711081484e-05, "loss": 0.45, "step": 14178 }, { "epoch": 0.5051570265600228, "grad_norm": 1.2284164672978302, "learning_rate": 1.0319009363825115e-05, "loss": 0.4434, "step": 14179 }, { "epoch": 0.5051926536865169, "grad_norm": 1.5619320582507834, "learning_rate": 1.0317856012320915e-05, "loss": 0.5371, "step": 14180 }, { "epoch": 0.5052282808130111, "grad_norm": 1.4405558603117403, "learning_rate": 1.0316702656584241e-05, "loss": 0.5602, "step": 14181 }, { "epoch": 0.5052639079395052, "grad_norm": 1.0768186487264078, "learning_rate": 1.0315549296630447e-05, "loss": 0.4824, "step": 14182 }, { "epoch": 0.5052995350659992, "grad_norm": 1.165807706263114, "learning_rate": 1.0314395932474896e-05, "loss": 0.4515, "step": 14183 }, { "epoch": 0.5053351621924934, "grad_norm": 1.4784034213377948, "learning_rate": 1.0313242564132943e-05, "loss": 0.5866, "step": 14184 }, { "epoch": 0.5053707893189875, "grad_norm": 1.7208902943746234, "learning_rate": 1.0312089191619948e-05, "loss": 0.4766, "step": 14185 }, { "epoch": 0.5054064164454816, "grad_norm": 1.107931201594779, "learning_rate": 1.0310935814951263e-05, "loss": 0.4284, "step": 14186 }, { "epoch": 0.5054420435719758, "grad_norm": 1.0708079848601826, "learning_rate": 1.0309782434142254e-05, "loss": 0.4904, "step": 14187 }, { "epoch": 0.5054776706984698, "grad_norm": 1.102806777565633, "learning_rate": 1.0308629049208277e-05, "loss": 0.4655, "step": 14188 }, { "epoch": 0.5055132978249639, "grad_norm": 1.6466500207796908, "learning_rate": 1.0307475660164683e-05, "loss": 0.5497, "step": 14189 }, { "epoch": 0.5055489249514581, "grad_norm": 1.6628611393518724, "learning_rate": 1.030632226702684e-05, "loss": 0.4289, "step": 14190 }, { "epoch": 0.5055845520779522, "grad_norm": 1.2318345695227477, "learning_rate": 1.0305168869810099e-05, "loss": 0.5035, "step": 14191 }, { "epoch": 0.5056201792044462, "grad_norm": 1.6298679114560108, "learning_rate": 1.0304015468529821e-05, "loss": 0.5653, "step": 14192 }, { "epoch": 0.5056558063309404, "grad_norm": 1.286711459732781, "learning_rate": 1.0302862063201367e-05, "loss": 0.4865, "step": 14193 }, { "epoch": 0.5056914334574345, "grad_norm": 1.058147024792624, "learning_rate": 1.0301708653840092e-05, "loss": 0.4311, "step": 14194 }, { "epoch": 0.5057270605839286, "grad_norm": 1.3270847631458647, "learning_rate": 1.0300555240461356e-05, "loss": 0.4362, "step": 14195 }, { "epoch": 0.5057626877104228, "grad_norm": 1.9938664857246688, "learning_rate": 1.0299401823080517e-05, "loss": 0.6044, "step": 14196 }, { "epoch": 0.5057983148369168, "grad_norm": 1.740933844705857, "learning_rate": 1.0298248401712931e-05, "loss": 0.6148, "step": 14197 }, { "epoch": 0.5058339419634109, "grad_norm": 1.0963866031792522, "learning_rate": 1.0297094976373962e-05, "loss": 0.4296, "step": 14198 }, { "epoch": 0.5058695690899051, "grad_norm": 1.1111685740444452, "learning_rate": 1.0295941547078965e-05, "loss": 0.502, "step": 14199 }, { "epoch": 0.5059051962163992, "grad_norm": 1.1117590699173847, "learning_rate": 1.0294788113843297e-05, "loss": 0.4387, "step": 14200 }, { "epoch": 0.5059408233428933, "grad_norm": 1.9370866416707566, "learning_rate": 1.0293634676682323e-05, "loss": 0.5995, "step": 14201 }, { "epoch": 0.5059764504693874, "grad_norm": 1.6889640826999857, "learning_rate": 1.0292481235611394e-05, "loss": 0.523, "step": 14202 }, { "epoch": 0.5060120775958815, "grad_norm": 1.3125767821417247, "learning_rate": 1.0291327790645876e-05, "loss": 0.4433, "step": 14203 }, { "epoch": 0.5060477047223756, "grad_norm": 1.4798511826248222, "learning_rate": 1.0290174341801122e-05, "loss": 0.5152, "step": 14204 }, { "epoch": 0.5060833318488698, "grad_norm": 1.2216691807034425, "learning_rate": 1.0289020889092496e-05, "loss": 0.4534, "step": 14205 }, { "epoch": 0.5061189589753639, "grad_norm": 1.7016137230843607, "learning_rate": 1.0287867432535357e-05, "loss": 0.5907, "step": 14206 }, { "epoch": 0.5061545861018579, "grad_norm": 1.8535255346181903, "learning_rate": 1.028671397214506e-05, "loss": 0.5543, "step": 14207 }, { "epoch": 0.5061902132283521, "grad_norm": 1.619748098824871, "learning_rate": 1.0285560507936962e-05, "loss": 0.4254, "step": 14208 }, { "epoch": 0.5062258403548462, "grad_norm": 1.8550839783221365, "learning_rate": 1.0284407039926431e-05, "loss": 0.5779, "step": 14209 }, { "epoch": 0.5062614674813403, "grad_norm": 1.0588192622047916, "learning_rate": 1.0283253568128821e-05, "loss": 0.4307, "step": 14210 }, { "epoch": 0.5062970946078345, "grad_norm": 1.3592668654408866, "learning_rate": 1.028210009255949e-05, "loss": 0.4939, "step": 14211 }, { "epoch": 0.5063327217343285, "grad_norm": 1.832106093354132, "learning_rate": 1.0280946613233803e-05, "loss": 0.5458, "step": 14212 }, { "epoch": 0.5063683488608226, "grad_norm": 1.2809254182472996, "learning_rate": 1.0279793130167112e-05, "loss": 0.4872, "step": 14213 }, { "epoch": 0.5064039759873168, "grad_norm": 1.2263866656985856, "learning_rate": 1.027863964337478e-05, "loss": 0.5508, "step": 14214 }, { "epoch": 0.5064396031138109, "grad_norm": 1.11002473219139, "learning_rate": 1.0277486152872167e-05, "loss": 0.4436, "step": 14215 }, { "epoch": 0.5064752302403049, "grad_norm": 1.032337429063445, "learning_rate": 1.027633265867463e-05, "loss": 0.4299, "step": 14216 }, { "epoch": 0.5065108573667991, "grad_norm": 1.3706574874162851, "learning_rate": 1.0275179160797533e-05, "loss": 0.4418, "step": 14217 }, { "epoch": 0.5065464844932932, "grad_norm": 1.9762027075913662, "learning_rate": 1.0274025659256233e-05, "loss": 0.547, "step": 14218 }, { "epoch": 0.5065821116197873, "grad_norm": 1.143709403821868, "learning_rate": 1.0272872154066089e-05, "loss": 0.4579, "step": 14219 }, { "epoch": 0.5066177387462815, "grad_norm": 1.5378723018514733, "learning_rate": 1.0271718645242462e-05, "loss": 0.6029, "step": 14220 }, { "epoch": 0.5066533658727755, "grad_norm": 1.1151531672719506, "learning_rate": 1.0270565132800711e-05, "loss": 0.4825, "step": 14221 }, { "epoch": 0.5066889929992696, "grad_norm": 1.9793399143805257, "learning_rate": 1.0269411616756197e-05, "loss": 0.5599, "step": 14222 }, { "epoch": 0.5067246201257638, "grad_norm": 1.6146608406273946, "learning_rate": 1.0268258097124277e-05, "loss": 0.5777, "step": 14223 }, { "epoch": 0.5067602472522579, "grad_norm": 1.0018164346627245, "learning_rate": 1.0267104573920317e-05, "loss": 0.4354, "step": 14224 }, { "epoch": 0.506795874378752, "grad_norm": 1.158191373292194, "learning_rate": 1.0265951047159668e-05, "loss": 0.4938, "step": 14225 }, { "epoch": 0.5068315015052461, "grad_norm": 1.4221750400686706, "learning_rate": 1.0264797516857696e-05, "loss": 0.5307, "step": 14226 }, { "epoch": 0.5068671286317402, "grad_norm": 1.1732138280885942, "learning_rate": 1.026364398302976e-05, "loss": 0.45, "step": 14227 }, { "epoch": 0.5069027557582343, "grad_norm": 1.3215094860737198, "learning_rate": 1.0262490445691222e-05, "loss": 0.59, "step": 14228 }, { "epoch": 0.5069383828847285, "grad_norm": 1.6526993431480759, "learning_rate": 1.0261336904857442e-05, "loss": 0.5228, "step": 14229 }, { "epoch": 0.5069740100112226, "grad_norm": 1.1427418533615612, "learning_rate": 1.0260183360543774e-05, "loss": 0.4758, "step": 14230 }, { "epoch": 0.5070096371377166, "grad_norm": 1.3535554996372572, "learning_rate": 1.0259029812765584e-05, "loss": 0.5131, "step": 14231 }, { "epoch": 0.5070452642642108, "grad_norm": 1.6722658881395975, "learning_rate": 1.0257876261538233e-05, "loss": 0.5396, "step": 14232 }, { "epoch": 0.5070808913907049, "grad_norm": 1.5177851676512124, "learning_rate": 1.0256722706877078e-05, "loss": 0.5936, "step": 14233 }, { "epoch": 0.507116518517199, "grad_norm": 2.158243122282618, "learning_rate": 1.0255569148797479e-05, "loss": 0.5964, "step": 14234 }, { "epoch": 0.5071521456436932, "grad_norm": 1.1388388667209235, "learning_rate": 1.0254415587314801e-05, "loss": 0.4974, "step": 14235 }, { "epoch": 0.5071877727701872, "grad_norm": 1.5009890824170449, "learning_rate": 1.0253262022444401e-05, "loss": 0.6182, "step": 14236 }, { "epoch": 0.5072233998966813, "grad_norm": 1.1589889929824564, "learning_rate": 1.025210845420164e-05, "loss": 0.4884, "step": 14237 }, { "epoch": 0.5072590270231755, "grad_norm": 2.198497235822591, "learning_rate": 1.0250954882601876e-05, "loss": 0.4763, "step": 14238 }, { "epoch": 0.5072946541496696, "grad_norm": 1.572185419263217, "learning_rate": 1.0249801307660473e-05, "loss": 0.4693, "step": 14239 }, { "epoch": 0.5073302812761636, "grad_norm": 1.6449575261555751, "learning_rate": 1.0248647729392796e-05, "loss": 0.5446, "step": 14240 }, { "epoch": 0.5073659084026578, "grad_norm": 1.6858210506418974, "learning_rate": 1.0247494147814197e-05, "loss": 0.5338, "step": 14241 }, { "epoch": 0.5074015355291519, "grad_norm": 1.0847318257139487, "learning_rate": 1.0246340562940038e-05, "loss": 0.4665, "step": 14242 }, { "epoch": 0.507437162655646, "grad_norm": 1.3277523571048602, "learning_rate": 1.0245186974785685e-05, "loss": 0.5831, "step": 14243 }, { "epoch": 0.5074727897821402, "grad_norm": 1.665835585083209, "learning_rate": 1.0244033383366498e-05, "loss": 0.5572, "step": 14244 }, { "epoch": 0.5075084169086342, "grad_norm": 1.3874830717687043, "learning_rate": 1.0242879788697833e-05, "loss": 0.4856, "step": 14245 }, { "epoch": 0.5075440440351283, "grad_norm": 1.318809309094334, "learning_rate": 1.0241726190795056e-05, "loss": 0.4308, "step": 14246 }, { "epoch": 0.5075796711616225, "grad_norm": 1.2307994675216936, "learning_rate": 1.0240572589673526e-05, "loss": 0.47, "step": 14247 }, { "epoch": 0.5076152982881166, "grad_norm": 1.3498843602690167, "learning_rate": 1.02394189853486e-05, "loss": 0.5247, "step": 14248 }, { "epoch": 0.5076509254146107, "grad_norm": 1.6959045148164673, "learning_rate": 1.0238265377835644e-05, "loss": 0.5272, "step": 14249 }, { "epoch": 0.5076865525411048, "grad_norm": 1.195285806765651, "learning_rate": 1.0237111767150018e-05, "loss": 0.5913, "step": 14250 }, { "epoch": 0.5077221796675989, "grad_norm": 1.7602601346567253, "learning_rate": 1.0235958153307087e-05, "loss": 0.5106, "step": 14251 }, { "epoch": 0.507757806794093, "grad_norm": 1.0596011206494176, "learning_rate": 1.0234804536322206e-05, "loss": 0.4107, "step": 14252 }, { "epoch": 0.5077934339205872, "grad_norm": 1.0686980648244968, "learning_rate": 1.0233650916210736e-05, "loss": 0.49, "step": 14253 }, { "epoch": 0.5078290610470813, "grad_norm": 1.9206493823160606, "learning_rate": 1.0232497292988043e-05, "loss": 0.5157, "step": 14254 }, { "epoch": 0.5078646881735753, "grad_norm": 1.3556249789163468, "learning_rate": 1.0231343666669486e-05, "loss": 0.4838, "step": 14255 }, { "epoch": 0.5079003153000695, "grad_norm": 1.432976050829391, "learning_rate": 1.0230190037270427e-05, "loss": 0.5475, "step": 14256 }, { "epoch": 0.5079359424265636, "grad_norm": 1.3862344557455335, "learning_rate": 1.0229036404806226e-05, "loss": 0.514, "step": 14257 }, { "epoch": 0.5079715695530577, "grad_norm": 1.4673148887520118, "learning_rate": 1.0227882769292246e-05, "loss": 0.5634, "step": 14258 }, { "epoch": 0.5080071966795519, "grad_norm": 1.505045899716318, "learning_rate": 1.0226729130743846e-05, "loss": 0.5301, "step": 14259 }, { "epoch": 0.5080428238060459, "grad_norm": 1.0423010742484973, "learning_rate": 1.0225575489176386e-05, "loss": 0.4258, "step": 14260 }, { "epoch": 0.50807845093254, "grad_norm": 2.1185273911088602, "learning_rate": 1.0224421844605236e-05, "loss": 0.573, "step": 14261 }, { "epoch": 0.5081140780590342, "grad_norm": 1.4505088768057148, "learning_rate": 1.0223268197045752e-05, "loss": 0.4404, "step": 14262 }, { "epoch": 0.5081497051855283, "grad_norm": 1.5985831456091042, "learning_rate": 1.0222114546513296e-05, "loss": 0.5903, "step": 14263 }, { "epoch": 0.5081853323120223, "grad_norm": 1.2416253641042552, "learning_rate": 1.0220960893023226e-05, "loss": 0.4601, "step": 14264 }, { "epoch": 0.5082209594385165, "grad_norm": 1.0549526800030435, "learning_rate": 1.0219807236590911e-05, "loss": 0.4381, "step": 14265 }, { "epoch": 0.5082565865650106, "grad_norm": 1.0867130014904658, "learning_rate": 1.0218653577231709e-05, "loss": 0.4258, "step": 14266 }, { "epoch": 0.5082922136915047, "grad_norm": 1.7631945411504484, "learning_rate": 1.0217499914960977e-05, "loss": 0.5382, "step": 14267 }, { "epoch": 0.5083278408179989, "grad_norm": 1.1203108469020087, "learning_rate": 1.0216346249794087e-05, "loss": 0.4527, "step": 14268 }, { "epoch": 0.5083634679444929, "grad_norm": 1.3824048667345619, "learning_rate": 1.0215192581746394e-05, "loss": 0.4326, "step": 14269 }, { "epoch": 0.508399095070987, "grad_norm": 1.6015700438671956, "learning_rate": 1.0214038910833261e-05, "loss": 0.5749, "step": 14270 }, { "epoch": 0.5084347221974812, "grad_norm": 1.11079680786122, "learning_rate": 1.0212885237070049e-05, "loss": 0.4777, "step": 14271 }, { "epoch": 0.5084703493239753, "grad_norm": 1.226651722936367, "learning_rate": 1.0211731560472122e-05, "loss": 0.4882, "step": 14272 }, { "epoch": 0.5085059764504694, "grad_norm": 1.1825665704293833, "learning_rate": 1.0210577881054841e-05, "loss": 0.4827, "step": 14273 }, { "epoch": 0.5085416035769635, "grad_norm": 1.5401473371807526, "learning_rate": 1.0209424198833571e-05, "loss": 0.5228, "step": 14274 }, { "epoch": 0.5085772307034576, "grad_norm": 1.4103265332248467, "learning_rate": 1.0208270513823669e-05, "loss": 0.5092, "step": 14275 }, { "epoch": 0.5086128578299517, "grad_norm": 1.6187128353036309, "learning_rate": 1.0207116826040499e-05, "loss": 0.5016, "step": 14276 }, { "epoch": 0.5086484849564459, "grad_norm": 1.4197945074409168, "learning_rate": 1.0205963135499426e-05, "loss": 0.5609, "step": 14277 }, { "epoch": 0.50868411208294, "grad_norm": 1.3747576866983853, "learning_rate": 1.020480944221581e-05, "loss": 0.5504, "step": 14278 }, { "epoch": 0.508719739209434, "grad_norm": 1.1136759888887047, "learning_rate": 1.020365574620501e-05, "loss": 0.4871, "step": 14279 }, { "epoch": 0.5087553663359282, "grad_norm": 1.9248212707100072, "learning_rate": 1.0202502047482395e-05, "loss": 0.5128, "step": 14280 }, { "epoch": 0.5087909934624223, "grad_norm": 1.0203062234700737, "learning_rate": 1.0201348346063322e-05, "loss": 0.4467, "step": 14281 }, { "epoch": 0.5088266205889164, "grad_norm": 1.1588450737463276, "learning_rate": 1.0200194641963153e-05, "loss": 0.5099, "step": 14282 }, { "epoch": 0.5088622477154106, "grad_norm": 2.279334304760399, "learning_rate": 1.0199040935197254e-05, "loss": 0.4553, "step": 14283 }, { "epoch": 0.5088978748419046, "grad_norm": 2.203485318939645, "learning_rate": 1.0197887225780985e-05, "loss": 0.5809, "step": 14284 }, { "epoch": 0.5089335019683987, "grad_norm": 1.6846385542156463, "learning_rate": 1.0196733513729711e-05, "loss": 0.5989, "step": 14285 }, { "epoch": 0.5089691290948929, "grad_norm": 1.8468567185675213, "learning_rate": 1.019557979905879e-05, "loss": 0.5556, "step": 14286 }, { "epoch": 0.509004756221387, "grad_norm": 1.5051818080483064, "learning_rate": 1.019442608178359e-05, "loss": 0.5979, "step": 14287 }, { "epoch": 0.509040383347881, "grad_norm": 1.0073352774629718, "learning_rate": 1.0193272361919473e-05, "loss": 0.4155, "step": 14288 }, { "epoch": 0.5090760104743752, "grad_norm": 1.9217055126602949, "learning_rate": 1.0192118639481795e-05, "loss": 0.5407, "step": 14289 }, { "epoch": 0.5091116376008693, "grad_norm": 1.5670322725621006, "learning_rate": 1.0190964914485926e-05, "loss": 0.5651, "step": 14290 }, { "epoch": 0.5091472647273634, "grad_norm": 1.5907111550037736, "learning_rate": 1.0189811186947224e-05, "loss": 0.4797, "step": 14291 }, { "epoch": 0.5091828918538576, "grad_norm": 1.4455557061670365, "learning_rate": 1.0188657456881054e-05, "loss": 0.6135, "step": 14292 }, { "epoch": 0.5092185189803516, "grad_norm": 1.0754910574883485, "learning_rate": 1.0187503724302776e-05, "loss": 0.4853, "step": 14293 }, { "epoch": 0.5092541461068457, "grad_norm": 1.7274799668644025, "learning_rate": 1.0186349989227755e-05, "loss": 0.5862, "step": 14294 }, { "epoch": 0.5092897732333399, "grad_norm": 1.1016018457336392, "learning_rate": 1.0185196251671357e-05, "loss": 0.4386, "step": 14295 }, { "epoch": 0.509325400359834, "grad_norm": 3.0073986992794115, "learning_rate": 1.0184042511648942e-05, "loss": 0.581, "step": 14296 }, { "epoch": 0.509361027486328, "grad_norm": 1.2244750124955595, "learning_rate": 1.018288876917587e-05, "loss": 0.6083, "step": 14297 }, { "epoch": 0.5093966546128222, "grad_norm": 1.1323509671899028, "learning_rate": 1.0181735024267504e-05, "loss": 0.4785, "step": 14298 }, { "epoch": 0.5094322817393163, "grad_norm": 1.889295906454214, "learning_rate": 1.0180581276939213e-05, "loss": 0.5795, "step": 14299 }, { "epoch": 0.5094679088658104, "grad_norm": 1.7028558496849824, "learning_rate": 1.0179427527206354e-05, "loss": 0.5342, "step": 14300 }, { "epoch": 0.5095035359923046, "grad_norm": 1.1044367288404862, "learning_rate": 1.0178273775084292e-05, "loss": 0.4524, "step": 14301 }, { "epoch": 0.5095391631187987, "grad_norm": 1.1475313616683893, "learning_rate": 1.0177120020588392e-05, "loss": 0.475, "step": 14302 }, { "epoch": 0.5095747902452927, "grad_norm": 0.9695580035838247, "learning_rate": 1.0175966263734014e-05, "loss": 0.3971, "step": 14303 }, { "epoch": 0.5096104173717869, "grad_norm": 1.0368464917778344, "learning_rate": 1.0174812504536523e-05, "loss": 0.4681, "step": 14304 }, { "epoch": 0.509646044498281, "grad_norm": 1.6629360927760575, "learning_rate": 1.017365874301128e-05, "loss": 0.5824, "step": 14305 }, { "epoch": 0.5096816716247751, "grad_norm": 1.352604005650048, "learning_rate": 1.0172504979173647e-05, "loss": 0.569, "step": 14306 }, { "epoch": 0.5097172987512693, "grad_norm": 1.1064204981089347, "learning_rate": 1.0171351213038993e-05, "loss": 0.44, "step": 14307 }, { "epoch": 0.5097529258777633, "grad_norm": 1.5988497662345924, "learning_rate": 1.0170197444622678e-05, "loss": 0.5662, "step": 14308 }, { "epoch": 0.5097885530042574, "grad_norm": 1.177333803856439, "learning_rate": 1.0169043673940062e-05, "loss": 0.4256, "step": 14309 }, { "epoch": 0.5098241801307516, "grad_norm": 1.439225480226727, "learning_rate": 1.0167889901006516e-05, "loss": 0.4569, "step": 14310 }, { "epoch": 0.5098598072572457, "grad_norm": 1.6820475897449967, "learning_rate": 1.0166736125837395e-05, "loss": 0.5523, "step": 14311 }, { "epoch": 0.5098954343837397, "grad_norm": 1.3615733556216538, "learning_rate": 1.0165582348448066e-05, "loss": 0.501, "step": 14312 }, { "epoch": 0.5099310615102339, "grad_norm": 1.2357014588002704, "learning_rate": 1.0164428568853894e-05, "loss": 0.4604, "step": 14313 }, { "epoch": 0.509966688636728, "grad_norm": 1.1599199402058449, "learning_rate": 1.016327478707024e-05, "loss": 0.4481, "step": 14314 }, { "epoch": 0.5100023157632221, "grad_norm": 1.3237797834343852, "learning_rate": 1.0162121003112471e-05, "loss": 0.5258, "step": 14315 }, { "epoch": 0.5100379428897163, "grad_norm": 1.2580750046446199, "learning_rate": 1.0160967216995942e-05, "loss": 0.5517, "step": 14316 }, { "epoch": 0.5100735700162103, "grad_norm": 1.0915962766930394, "learning_rate": 1.0159813428736027e-05, "loss": 0.4598, "step": 14317 }, { "epoch": 0.5101091971427044, "grad_norm": 1.1842905094467941, "learning_rate": 1.015865963834808e-05, "loss": 0.4367, "step": 14318 }, { "epoch": 0.5101448242691986, "grad_norm": 1.047919223495374, "learning_rate": 1.0157505845847473e-05, "loss": 0.4522, "step": 14319 }, { "epoch": 0.5101804513956927, "grad_norm": 2.0414220253594872, "learning_rate": 1.0156352051249561e-05, "loss": 0.5509, "step": 14320 }, { "epoch": 0.5102160785221868, "grad_norm": 1.9398736787484565, "learning_rate": 1.0155198254569716e-05, "loss": 0.5094, "step": 14321 }, { "epoch": 0.5102517056486809, "grad_norm": 1.860728474270024, "learning_rate": 1.0154044455823298e-05, "loss": 0.5329, "step": 14322 }, { "epoch": 0.510287332775175, "grad_norm": 1.6278423402928468, "learning_rate": 1.0152890655025669e-05, "loss": 0.5476, "step": 14323 }, { "epoch": 0.5103229599016691, "grad_norm": 1.261751396066924, "learning_rate": 1.0151736852192193e-05, "loss": 0.4626, "step": 14324 }, { "epoch": 0.5103585870281633, "grad_norm": 1.4171852770761688, "learning_rate": 1.0150583047338236e-05, "loss": 0.5705, "step": 14325 }, { "epoch": 0.5103942141546574, "grad_norm": 1.618146937267313, "learning_rate": 1.0149429240479163e-05, "loss": 0.5138, "step": 14326 }, { "epoch": 0.5104298412811514, "grad_norm": 2.842660887774898, "learning_rate": 1.014827543163033e-05, "loss": 0.4877, "step": 14327 }, { "epoch": 0.5104654684076456, "grad_norm": 1.0974102589015111, "learning_rate": 1.014712162080711e-05, "loss": 0.435, "step": 14328 }, { "epoch": 0.5105010955341397, "grad_norm": 1.2803649168109383, "learning_rate": 1.014596780802486e-05, "loss": 0.4538, "step": 14329 }, { "epoch": 0.5105367226606338, "grad_norm": 1.0767906926802957, "learning_rate": 1.014481399329895e-05, "loss": 0.4023, "step": 14330 }, { "epoch": 0.510572349787128, "grad_norm": 1.43765979383623, "learning_rate": 1.0143660176644736e-05, "loss": 0.4378, "step": 14331 }, { "epoch": 0.510607976913622, "grad_norm": 1.6504655323666397, "learning_rate": 1.0142506358077588e-05, "loss": 0.5284, "step": 14332 }, { "epoch": 0.5106436040401161, "grad_norm": 1.2874240663703105, "learning_rate": 1.0141352537612872e-05, "loss": 0.5533, "step": 14333 }, { "epoch": 0.5106792311666103, "grad_norm": 1.7791074371655047, "learning_rate": 1.0140198715265945e-05, "loss": 0.616, "step": 14334 }, { "epoch": 0.5107148582931044, "grad_norm": 1.2529643117899583, "learning_rate": 1.0139044891052171e-05, "loss": 0.5807, "step": 14335 }, { "epoch": 0.5107504854195984, "grad_norm": 1.1105062717171472, "learning_rate": 1.0137891064986923e-05, "loss": 0.462, "step": 14336 }, { "epoch": 0.5107861125460926, "grad_norm": 1.1418264973220287, "learning_rate": 1.0136737237085557e-05, "loss": 0.4678, "step": 14337 }, { "epoch": 0.5108217396725867, "grad_norm": 1.014464549219314, "learning_rate": 1.0135583407363439e-05, "loss": 0.4712, "step": 14338 }, { "epoch": 0.5108573667990808, "grad_norm": 1.6326567805974412, "learning_rate": 1.0134429575835932e-05, "loss": 0.501, "step": 14339 }, { "epoch": 0.510892993925575, "grad_norm": 1.6957049992173707, "learning_rate": 1.0133275742518402e-05, "loss": 0.5594, "step": 14340 }, { "epoch": 0.510928621052069, "grad_norm": 2.6165157684660807, "learning_rate": 1.0132121907426212e-05, "loss": 0.5627, "step": 14341 }, { "epoch": 0.5109642481785631, "grad_norm": 1.2042757590254354, "learning_rate": 1.0130968070574725e-05, "loss": 0.4859, "step": 14342 }, { "epoch": 0.5109998753050573, "grad_norm": 1.419330337111618, "learning_rate": 1.012981423197931e-05, "loss": 0.5249, "step": 14343 }, { "epoch": 0.5110355024315514, "grad_norm": 5.833516830978396, "learning_rate": 1.0128660391655327e-05, "loss": 0.5955, "step": 14344 }, { "epoch": 0.5110711295580455, "grad_norm": 1.584646996712696, "learning_rate": 1.012750654961814e-05, "loss": 0.5577, "step": 14345 }, { "epoch": 0.5111067566845396, "grad_norm": 1.4918759457769344, "learning_rate": 1.0126352705883113e-05, "loss": 0.5237, "step": 14346 }, { "epoch": 0.5111423838110337, "grad_norm": 1.5442365810412828, "learning_rate": 1.0125198860465615e-05, "loss": 0.6279, "step": 14347 }, { "epoch": 0.5111780109375278, "grad_norm": 1.4177205892507159, "learning_rate": 1.0124045013381003e-05, "loss": 0.5567, "step": 14348 }, { "epoch": 0.511213638064022, "grad_norm": 1.848258340528992, "learning_rate": 1.0122891164644646e-05, "loss": 0.5365, "step": 14349 }, { "epoch": 0.5112492651905161, "grad_norm": 1.237861482888037, "learning_rate": 1.0121737314271908e-05, "loss": 0.5287, "step": 14350 }, { "epoch": 0.5112848923170101, "grad_norm": 1.197210788761847, "learning_rate": 1.0120583462278153e-05, "loss": 0.4742, "step": 14351 }, { "epoch": 0.5113205194435043, "grad_norm": 1.1433054101078677, "learning_rate": 1.0119429608678744e-05, "loss": 0.4839, "step": 14352 }, { "epoch": 0.5113561465699984, "grad_norm": 1.1141208650433259, "learning_rate": 1.0118275753489047e-05, "loss": 0.4441, "step": 14353 }, { "epoch": 0.5113917736964925, "grad_norm": 1.0565323159780657, "learning_rate": 1.0117121896724422e-05, "loss": 0.4578, "step": 14354 }, { "epoch": 0.5114274008229867, "grad_norm": 2.5348084986666084, "learning_rate": 1.011596803840024e-05, "loss": 0.5515, "step": 14355 }, { "epoch": 0.5114630279494807, "grad_norm": 2.5922254958775883, "learning_rate": 1.0114814178531864e-05, "loss": 0.5274, "step": 14356 }, { "epoch": 0.5114986550759748, "grad_norm": 1.613346727464248, "learning_rate": 1.0113660317134655e-05, "loss": 0.5956, "step": 14357 }, { "epoch": 0.511534282202469, "grad_norm": 1.05909904012709, "learning_rate": 1.0112506454223981e-05, "loss": 0.4479, "step": 14358 }, { "epoch": 0.5115699093289631, "grad_norm": 1.0574844088023523, "learning_rate": 1.0111352589815204e-05, "loss": 0.4707, "step": 14359 }, { "epoch": 0.5116055364554571, "grad_norm": 1.308460885035603, "learning_rate": 1.011019872392369e-05, "loss": 0.4588, "step": 14360 }, { "epoch": 0.5116411635819513, "grad_norm": 1.519474508890716, "learning_rate": 1.01090448565648e-05, "loss": 0.5871, "step": 14361 }, { "epoch": 0.5116767907084454, "grad_norm": 1.5270246350554877, "learning_rate": 1.0107890987753905e-05, "loss": 0.5279, "step": 14362 }, { "epoch": 0.5117124178349395, "grad_norm": 1.0766003591164603, "learning_rate": 1.0106737117506367e-05, "loss": 0.4438, "step": 14363 }, { "epoch": 0.5117480449614337, "grad_norm": 1.8541663105507182, "learning_rate": 1.0105583245837547e-05, "loss": 0.4879, "step": 14364 }, { "epoch": 0.5117836720879277, "grad_norm": 1.2110238416169712, "learning_rate": 1.010442937276281e-05, "loss": 0.4046, "step": 14365 }, { "epoch": 0.5118192992144218, "grad_norm": 1.265914356750003, "learning_rate": 1.0103275498297526e-05, "loss": 0.47, "step": 14366 }, { "epoch": 0.511854926340916, "grad_norm": 1.4960457886375256, "learning_rate": 1.0102121622457057e-05, "loss": 0.5313, "step": 14367 }, { "epoch": 0.5118905534674101, "grad_norm": 1.1204631747352474, "learning_rate": 1.0100967745256767e-05, "loss": 0.4188, "step": 14368 }, { "epoch": 0.5119261805939042, "grad_norm": 1.0689900252462994, "learning_rate": 1.009981386671202e-05, "loss": 0.4661, "step": 14369 }, { "epoch": 0.5119618077203983, "grad_norm": 1.1339184179253952, "learning_rate": 1.0098659986838185e-05, "loss": 0.4665, "step": 14370 }, { "epoch": 0.5119974348468924, "grad_norm": 1.2735594085693585, "learning_rate": 1.0097506105650619e-05, "loss": 0.4346, "step": 14371 }, { "epoch": 0.5120330619733865, "grad_norm": 1.0612595291661155, "learning_rate": 1.0096352223164692e-05, "loss": 0.4487, "step": 14372 }, { "epoch": 0.5120686890998807, "grad_norm": 1.0374763283928954, "learning_rate": 1.0095198339395769e-05, "loss": 0.4198, "step": 14373 }, { "epoch": 0.5121043162263748, "grad_norm": 1.4814096892334103, "learning_rate": 1.0094044454359214e-05, "loss": 0.4619, "step": 14374 }, { "epoch": 0.5121399433528688, "grad_norm": 1.6844476062166633, "learning_rate": 1.0092890568070388e-05, "loss": 0.5508, "step": 14375 }, { "epoch": 0.512175570479363, "grad_norm": 1.4239498825099752, "learning_rate": 1.009173668054466e-05, "loss": 0.5286, "step": 14376 }, { "epoch": 0.5122111976058571, "grad_norm": 1.1754676620416535, "learning_rate": 1.0090582791797397e-05, "loss": 0.4337, "step": 14377 }, { "epoch": 0.5122468247323512, "grad_norm": 1.1923589712551967, "learning_rate": 1.0089428901843958e-05, "loss": 0.4169, "step": 14378 }, { "epoch": 0.5122824518588454, "grad_norm": 1.7490506138586615, "learning_rate": 1.0088275010699713e-05, "loss": 0.5571, "step": 14379 }, { "epoch": 0.5123180789853394, "grad_norm": 1.3000403002897345, "learning_rate": 1.0087121118380021e-05, "loss": 0.5633, "step": 14380 }, { "epoch": 0.5123537061118335, "grad_norm": 1.212243148856673, "learning_rate": 1.0085967224900256e-05, "loss": 0.5313, "step": 14381 }, { "epoch": 0.5123893332383277, "grad_norm": 1.1472936763637296, "learning_rate": 1.0084813330275774e-05, "loss": 0.4891, "step": 14382 }, { "epoch": 0.5124249603648218, "grad_norm": 1.1904387734846062, "learning_rate": 1.0083659434521941e-05, "loss": 0.4444, "step": 14383 }, { "epoch": 0.5124605874913158, "grad_norm": 1.1436269953419143, "learning_rate": 1.0082505537654128e-05, "loss": 0.4788, "step": 14384 }, { "epoch": 0.51249621461781, "grad_norm": 1.2515665165598628, "learning_rate": 1.0081351639687695e-05, "loss": 0.5204, "step": 14385 }, { "epoch": 0.5125318417443041, "grad_norm": 1.07149687375513, "learning_rate": 1.0080197740638008e-05, "loss": 0.4047, "step": 14386 }, { "epoch": 0.5125674688707982, "grad_norm": 1.4658704812863461, "learning_rate": 1.0079043840520428e-05, "loss": 0.6395, "step": 14387 }, { "epoch": 0.5126030959972924, "grad_norm": 1.0034345758623255, "learning_rate": 1.007788993935033e-05, "loss": 0.4684, "step": 14388 }, { "epoch": 0.5126387231237864, "grad_norm": 1.152283623475405, "learning_rate": 1.0076736037143069e-05, "loss": 0.4548, "step": 14389 }, { "epoch": 0.5126743502502805, "grad_norm": 1.2219328397117013, "learning_rate": 1.0075582133914016e-05, "loss": 0.4585, "step": 14390 }, { "epoch": 0.5127099773767747, "grad_norm": 1.564628067981974, "learning_rate": 1.0074428229678535e-05, "loss": 0.5375, "step": 14391 }, { "epoch": 0.5127456045032688, "grad_norm": 1.5225946954662233, "learning_rate": 1.0073274324451989e-05, "loss": 0.6138, "step": 14392 }, { "epoch": 0.5127812316297629, "grad_norm": 2.0954261222399806, "learning_rate": 1.0072120418249744e-05, "loss": 0.4523, "step": 14393 }, { "epoch": 0.512816858756257, "grad_norm": 1.0311843621557002, "learning_rate": 1.0070966511087164e-05, "loss": 0.4105, "step": 14394 }, { "epoch": 0.5128524858827511, "grad_norm": 1.1517642664174381, "learning_rate": 1.0069812602979617e-05, "loss": 0.4307, "step": 14395 }, { "epoch": 0.5128881130092452, "grad_norm": 1.528713067637942, "learning_rate": 1.0068658693942466e-05, "loss": 0.5231, "step": 14396 }, { "epoch": 0.5129237401357394, "grad_norm": 1.4532964437834952, "learning_rate": 1.0067504783991078e-05, "loss": 0.5384, "step": 14397 }, { "epoch": 0.5129593672622335, "grad_norm": 1.1907194574108952, "learning_rate": 1.0066350873140814e-05, "loss": 0.4719, "step": 14398 }, { "epoch": 0.5129949943887275, "grad_norm": 1.267590554928695, "learning_rate": 1.006519696140704e-05, "loss": 0.4904, "step": 14399 }, { "epoch": 0.5130306215152217, "grad_norm": 1.587218353528774, "learning_rate": 1.0064043048805129e-05, "loss": 0.6285, "step": 14400 }, { "epoch": 0.5130662486417158, "grad_norm": 1.5342594874179178, "learning_rate": 1.0062889135350435e-05, "loss": 0.5955, "step": 14401 }, { "epoch": 0.5131018757682099, "grad_norm": 2.1380470515746923, "learning_rate": 1.0061735221058328e-05, "loss": 0.568, "step": 14402 }, { "epoch": 0.5131375028947041, "grad_norm": 1.0441169389435567, "learning_rate": 1.0060581305944177e-05, "loss": 0.4559, "step": 14403 }, { "epoch": 0.5131731300211981, "grad_norm": 1.006376193685528, "learning_rate": 1.0059427390023342e-05, "loss": 0.4299, "step": 14404 }, { "epoch": 0.5132087571476922, "grad_norm": 1.8987774537794606, "learning_rate": 1.005827347331119e-05, "loss": 0.5969, "step": 14405 }, { "epoch": 0.5132443842741864, "grad_norm": 1.380293023091899, "learning_rate": 1.0057119555823085e-05, "loss": 0.5356, "step": 14406 }, { "epoch": 0.5132800114006805, "grad_norm": 1.3900514548152672, "learning_rate": 1.0055965637574394e-05, "loss": 0.5297, "step": 14407 }, { "epoch": 0.5133156385271745, "grad_norm": 1.0744254764073473, "learning_rate": 1.0054811718580485e-05, "loss": 0.4022, "step": 14408 }, { "epoch": 0.5133512656536687, "grad_norm": 1.1545146569891975, "learning_rate": 1.0053657798856712e-05, "loss": 0.4304, "step": 14409 }, { "epoch": 0.5133868927801628, "grad_norm": 1.591597773563471, "learning_rate": 1.0052503878418451e-05, "loss": 0.6171, "step": 14410 }, { "epoch": 0.5134225199066569, "grad_norm": 1.4552013296547275, "learning_rate": 1.0051349957281064e-05, "loss": 0.5066, "step": 14411 }, { "epoch": 0.5134581470331511, "grad_norm": 1.3944156723797176, "learning_rate": 1.0050196035459921e-05, "loss": 0.6154, "step": 14412 }, { "epoch": 0.5134937741596451, "grad_norm": 1.068656911536963, "learning_rate": 1.0049042112970375e-05, "loss": 0.472, "step": 14413 }, { "epoch": 0.5135294012861392, "grad_norm": 1.538836115092013, "learning_rate": 1.0047888189827806e-05, "loss": 0.5354, "step": 14414 }, { "epoch": 0.5135650284126334, "grad_norm": 1.201976053523104, "learning_rate": 1.0046734266047568e-05, "loss": 0.4862, "step": 14415 }, { "epoch": 0.5136006555391275, "grad_norm": 1.2927039212481375, "learning_rate": 1.0045580341645033e-05, "loss": 0.4074, "step": 14416 }, { "epoch": 0.5136362826656216, "grad_norm": 1.0334549470181333, "learning_rate": 1.0044426416635562e-05, "loss": 0.3966, "step": 14417 }, { "epoch": 0.5136719097921157, "grad_norm": 1.097966086005052, "learning_rate": 1.0043272491034523e-05, "loss": 0.4439, "step": 14418 }, { "epoch": 0.5137075369186098, "grad_norm": 1.4401889748819032, "learning_rate": 1.0042118564857281e-05, "loss": 0.5549, "step": 14419 }, { "epoch": 0.5137431640451039, "grad_norm": 1.2854940761588005, "learning_rate": 1.0040964638119199e-05, "loss": 0.46, "step": 14420 }, { "epoch": 0.5137787911715981, "grad_norm": 1.493027916609999, "learning_rate": 1.0039810710835644e-05, "loss": 0.552, "step": 14421 }, { "epoch": 0.5138144182980922, "grad_norm": 1.58746244380208, "learning_rate": 1.0038656783021985e-05, "loss": 0.5186, "step": 14422 }, { "epoch": 0.5138500454245862, "grad_norm": 1.1240687842694035, "learning_rate": 1.0037502854693583e-05, "loss": 0.4592, "step": 14423 }, { "epoch": 0.5138856725510804, "grad_norm": 1.1710373560800809, "learning_rate": 1.0036348925865803e-05, "loss": 0.4578, "step": 14424 }, { "epoch": 0.5139212996775745, "grad_norm": 1.1024107662975406, "learning_rate": 1.0035194996554012e-05, "loss": 0.4889, "step": 14425 }, { "epoch": 0.5139569268040686, "grad_norm": 1.670348289648039, "learning_rate": 1.0034041066773577e-05, "loss": 0.6366, "step": 14426 }, { "epoch": 0.5139925539305628, "grad_norm": 1.178693841930067, "learning_rate": 1.0032887136539859e-05, "loss": 0.4374, "step": 14427 }, { "epoch": 0.5140281810570568, "grad_norm": 1.498554218135986, "learning_rate": 1.0031733205868223e-05, "loss": 0.5742, "step": 14428 }, { "epoch": 0.5140638081835509, "grad_norm": 1.9307687422867386, "learning_rate": 1.0030579274774043e-05, "loss": 0.5683, "step": 14429 }, { "epoch": 0.5140994353100451, "grad_norm": 1.6382998010944234, "learning_rate": 1.0029425343272676e-05, "loss": 0.5365, "step": 14430 }, { "epoch": 0.5141350624365392, "grad_norm": 1.104812251427106, "learning_rate": 1.0028271411379488e-05, "loss": 0.4615, "step": 14431 }, { "epoch": 0.5141706895630332, "grad_norm": 1.779996505204765, "learning_rate": 1.002711747910985e-05, "loss": 0.5431, "step": 14432 }, { "epoch": 0.5142063166895274, "grad_norm": 1.5717146201209227, "learning_rate": 1.002596354647912e-05, "loss": 0.551, "step": 14433 }, { "epoch": 0.5142419438160215, "grad_norm": 1.7428552268910413, "learning_rate": 1.0024809613502671e-05, "loss": 0.5473, "step": 14434 }, { "epoch": 0.5142775709425156, "grad_norm": 1.101070275415959, "learning_rate": 1.0023655680195865e-05, "loss": 0.4398, "step": 14435 }, { "epoch": 0.5143131980690098, "grad_norm": 1.2799710952008365, "learning_rate": 1.0022501746574063e-05, "loss": 0.5156, "step": 14436 }, { "epoch": 0.5143488251955038, "grad_norm": 1.4795326808783285, "learning_rate": 1.0021347812652638e-05, "loss": 0.577, "step": 14437 }, { "epoch": 0.5143844523219979, "grad_norm": 1.2033710022030482, "learning_rate": 1.0020193878446952e-05, "loss": 0.4679, "step": 14438 }, { "epoch": 0.5144200794484921, "grad_norm": 1.4524366685622387, "learning_rate": 1.0019039943972367e-05, "loss": 0.475, "step": 14439 }, { "epoch": 0.5144557065749862, "grad_norm": 1.1003489991168085, "learning_rate": 1.0017886009244253e-05, "loss": 0.4811, "step": 14440 }, { "epoch": 0.5144913337014803, "grad_norm": 1.8588711831234574, "learning_rate": 1.0016732074277977e-05, "loss": 0.506, "step": 14441 }, { "epoch": 0.5145269608279744, "grad_norm": 1.1745144154379745, "learning_rate": 1.0015578139088901e-05, "loss": 0.4771, "step": 14442 }, { "epoch": 0.5145625879544685, "grad_norm": 1.1111906389840285, "learning_rate": 1.0014424203692388e-05, "loss": 0.4491, "step": 14443 }, { "epoch": 0.5145982150809626, "grad_norm": 1.175645799720894, "learning_rate": 1.0013270268103808e-05, "loss": 0.4351, "step": 14444 }, { "epoch": 0.5146338422074568, "grad_norm": 1.1380547787955753, "learning_rate": 1.0012116332338526e-05, "loss": 0.4217, "step": 14445 }, { "epoch": 0.5146694693339509, "grad_norm": 1.2748318792764453, "learning_rate": 1.0010962396411908e-05, "loss": 0.4751, "step": 14446 }, { "epoch": 0.5147050964604449, "grad_norm": 1.0888781389158935, "learning_rate": 1.0009808460339314e-05, "loss": 0.423, "step": 14447 }, { "epoch": 0.5147407235869391, "grad_norm": 1.4456236102688618, "learning_rate": 1.0008654524136116e-05, "loss": 0.5871, "step": 14448 }, { "epoch": 0.5147763507134332, "grad_norm": 1.526064907214291, "learning_rate": 1.0007500587817677e-05, "loss": 0.592, "step": 14449 }, { "epoch": 0.5148119778399273, "grad_norm": 1.246580717316474, "learning_rate": 1.0006346651399363e-05, "loss": 0.4151, "step": 14450 }, { "epoch": 0.5148476049664215, "grad_norm": 1.7427408778466738, "learning_rate": 1.000519271489654e-05, "loss": 0.5136, "step": 14451 }, { "epoch": 0.5148832320929155, "grad_norm": 1.1011632714333952, "learning_rate": 1.000403877832457e-05, "loss": 0.4705, "step": 14452 }, { "epoch": 0.5149188592194096, "grad_norm": 2.223338112158117, "learning_rate": 1.0002884841698822e-05, "loss": 0.6262, "step": 14453 }, { "epoch": 0.5149544863459038, "grad_norm": 1.7417549576753202, "learning_rate": 1.0001730905034658e-05, "loss": 0.5768, "step": 14454 }, { "epoch": 0.5149901134723979, "grad_norm": 1.0569469880108517, "learning_rate": 1.0000576968347449e-05, "loss": 0.4372, "step": 14455 }, { "epoch": 0.515025740598892, "grad_norm": 1.2068794469935649, "learning_rate": 9.999423031652556e-06, "loss": 0.3924, "step": 14456 }, { "epoch": 0.5150613677253861, "grad_norm": 1.4271121656200796, "learning_rate": 9.998269094965345e-06, "loss": 0.619, "step": 14457 }, { "epoch": 0.5150969948518802, "grad_norm": 1.7832915694656428, "learning_rate": 9.997115158301183e-06, "loss": 0.5223, "step": 14458 }, { "epoch": 0.5151326219783743, "grad_norm": 1.049743710570416, "learning_rate": 9.995961221675433e-06, "loss": 0.4075, "step": 14459 }, { "epoch": 0.5151682491048685, "grad_norm": 1.161136228493806, "learning_rate": 9.994807285103466e-06, "loss": 0.4475, "step": 14460 }, { "epoch": 0.5152038762313625, "grad_norm": 1.674756445886917, "learning_rate": 9.99365334860064e-06, "loss": 0.572, "step": 14461 }, { "epoch": 0.5152395033578566, "grad_norm": 1.1165861124908423, "learning_rate": 9.992499412182321e-06, "loss": 0.4534, "step": 14462 }, { "epoch": 0.5152751304843508, "grad_norm": 1.0788255059018912, "learning_rate": 9.991345475863886e-06, "loss": 0.4698, "step": 14463 }, { "epoch": 0.5153107576108449, "grad_norm": 0.9953628290481973, "learning_rate": 9.990191539660686e-06, "loss": 0.4454, "step": 14464 }, { "epoch": 0.515346384737339, "grad_norm": 1.8526908378626592, "learning_rate": 9.989037603588096e-06, "loss": 0.5199, "step": 14465 }, { "epoch": 0.5153820118638331, "grad_norm": 1.0224486719979242, "learning_rate": 9.987883667661474e-06, "loss": 0.4103, "step": 14466 }, { "epoch": 0.5154176389903272, "grad_norm": 1.870081876324483, "learning_rate": 9.986729731896196e-06, "loss": 0.5662, "step": 14467 }, { "epoch": 0.5154532661168213, "grad_norm": 1.0316154905050658, "learning_rate": 9.985575796307615e-06, "loss": 0.4598, "step": 14468 }, { "epoch": 0.5154888932433155, "grad_norm": 1.2095605524082327, "learning_rate": 9.984421860911104e-06, "loss": 0.4947, "step": 14469 }, { "epoch": 0.5155245203698096, "grad_norm": 1.5166061106928848, "learning_rate": 9.983267925722026e-06, "loss": 0.6287, "step": 14470 }, { "epoch": 0.5155601474963036, "grad_norm": 1.1552409793005727, "learning_rate": 9.98211399075575e-06, "loss": 0.4056, "step": 14471 }, { "epoch": 0.5155957746227978, "grad_norm": 1.3586972739539065, "learning_rate": 9.980960056027637e-06, "loss": 0.492, "step": 14472 }, { "epoch": 0.5156314017492919, "grad_norm": 1.1841525402780233, "learning_rate": 9.979806121553052e-06, "loss": 0.4627, "step": 14473 }, { "epoch": 0.515667028875786, "grad_norm": 0.9601030862662738, "learning_rate": 9.978652187347366e-06, "loss": 0.449, "step": 14474 }, { "epoch": 0.5157026560022802, "grad_norm": 1.0789114992145223, "learning_rate": 9.977498253425937e-06, "loss": 0.4322, "step": 14475 }, { "epoch": 0.5157382831287742, "grad_norm": 1.2198498439602492, "learning_rate": 9.976344319804138e-06, "loss": 0.441, "step": 14476 }, { "epoch": 0.5157739102552683, "grad_norm": 1.0981934944568654, "learning_rate": 9.975190386497329e-06, "loss": 0.4891, "step": 14477 }, { "epoch": 0.5158095373817625, "grad_norm": 1.9629400526265475, "learning_rate": 9.974036453520881e-06, "loss": 0.6014, "step": 14478 }, { "epoch": 0.5158451645082566, "grad_norm": 1.7465431758804075, "learning_rate": 9.972882520890153e-06, "loss": 0.5466, "step": 14479 }, { "epoch": 0.5158807916347506, "grad_norm": 1.1011682793728739, "learning_rate": 9.971728588620513e-06, "loss": 0.4356, "step": 14480 }, { "epoch": 0.5159164187612448, "grad_norm": 1.856245100009851, "learning_rate": 9.970574656727327e-06, "loss": 0.596, "step": 14481 }, { "epoch": 0.5159520458877389, "grad_norm": 1.154222759234161, "learning_rate": 9.969420725225962e-06, "loss": 0.4401, "step": 14482 }, { "epoch": 0.515987673014233, "grad_norm": 1.136893007303546, "learning_rate": 9.968266794131778e-06, "loss": 0.4659, "step": 14483 }, { "epoch": 0.5160233001407272, "grad_norm": 1.121967426246532, "learning_rate": 9.967112863460146e-06, "loss": 0.4258, "step": 14484 }, { "epoch": 0.5160589272672212, "grad_norm": 1.7864262698153963, "learning_rate": 9.965958933226428e-06, "loss": 0.5116, "step": 14485 }, { "epoch": 0.5160945543937153, "grad_norm": 1.193498303649841, "learning_rate": 9.96480500344599e-06, "loss": 0.4431, "step": 14486 }, { "epoch": 0.5161301815202095, "grad_norm": 1.8521684665108762, "learning_rate": 9.9636510741342e-06, "loss": 0.5456, "step": 14487 }, { "epoch": 0.5161658086467036, "grad_norm": 1.5771752623817068, "learning_rate": 9.962497145306418e-06, "loss": 0.5324, "step": 14488 }, { "epoch": 0.5162014357731977, "grad_norm": 1.0886439191316453, "learning_rate": 9.961343216978018e-06, "loss": 0.4626, "step": 14489 }, { "epoch": 0.5162370628996918, "grad_norm": 1.0315192137111822, "learning_rate": 9.960189289164354e-06, "loss": 0.4436, "step": 14490 }, { "epoch": 0.5162726900261859, "grad_norm": 0.9963348893621911, "learning_rate": 9.959035361880803e-06, "loss": 0.4258, "step": 14491 }, { "epoch": 0.51630831715268, "grad_norm": 1.1251536346754545, "learning_rate": 9.95788143514272e-06, "loss": 0.5184, "step": 14492 }, { "epoch": 0.5163439442791742, "grad_norm": 1.0383625235181904, "learning_rate": 9.956727508965482e-06, "loss": 0.4056, "step": 14493 }, { "epoch": 0.5163795714056683, "grad_norm": 1.3349675586077834, "learning_rate": 9.955573583364441e-06, "loss": 0.4403, "step": 14494 }, { "epoch": 0.5164151985321623, "grad_norm": 1.0904832436627483, "learning_rate": 9.954419658354972e-06, "loss": 0.4383, "step": 14495 }, { "epoch": 0.5164508256586565, "grad_norm": 1.0246602352560434, "learning_rate": 9.953265733952434e-06, "loss": 0.4408, "step": 14496 }, { "epoch": 0.5164864527851506, "grad_norm": 1.7246574307908447, "learning_rate": 9.952111810172195e-06, "loss": 0.5726, "step": 14497 }, { "epoch": 0.5165220799116447, "grad_norm": 1.3912741529878545, "learning_rate": 9.950957887029627e-06, "loss": 0.5673, "step": 14498 }, { "epoch": 0.5165577070381389, "grad_norm": 1.1133955707708645, "learning_rate": 9.94980396454008e-06, "loss": 0.4742, "step": 14499 }, { "epoch": 0.5165933341646329, "grad_norm": 1.7091553898923375, "learning_rate": 9.948650042718938e-06, "loss": 0.5662, "step": 14500 }, { "epoch": 0.516628961291127, "grad_norm": 1.0257087929930655, "learning_rate": 9.94749612158155e-06, "loss": 0.4416, "step": 14501 }, { "epoch": 0.5166645884176212, "grad_norm": 1.7369848255682292, "learning_rate": 9.946342201143291e-06, "loss": 0.5431, "step": 14502 }, { "epoch": 0.5167002155441153, "grad_norm": 1.0536137287674818, "learning_rate": 9.94518828141952e-06, "loss": 0.4712, "step": 14503 }, { "epoch": 0.5167358426706093, "grad_norm": 1.0649556013662933, "learning_rate": 9.94403436242561e-06, "loss": 0.4501, "step": 14504 }, { "epoch": 0.5167714697971035, "grad_norm": 1.0632404524894996, "learning_rate": 9.942880444176918e-06, "loss": 0.434, "step": 14505 }, { "epoch": 0.5168070969235976, "grad_norm": 1.113804870369158, "learning_rate": 9.941726526688815e-06, "loss": 0.4597, "step": 14506 }, { "epoch": 0.5168427240500918, "grad_norm": 1.1441024038659504, "learning_rate": 9.94057260997666e-06, "loss": 0.4763, "step": 14507 }, { "epoch": 0.5168783511765859, "grad_norm": 1.6473262191679996, "learning_rate": 9.939418694055823e-06, "loss": 0.5489, "step": 14508 }, { "epoch": 0.51691397830308, "grad_norm": 1.233868282347144, "learning_rate": 9.938264778941674e-06, "loss": 0.5312, "step": 14509 }, { "epoch": 0.5169496054295741, "grad_norm": 1.162662001522702, "learning_rate": 9.937110864649568e-06, "loss": 0.4198, "step": 14510 }, { "epoch": 0.5169852325560682, "grad_norm": 1.217423873147854, "learning_rate": 9.935956951194876e-06, "loss": 0.5231, "step": 14511 }, { "epoch": 0.5170208596825623, "grad_norm": 1.7246071280643518, "learning_rate": 9.934803038592959e-06, "loss": 0.5969, "step": 14512 }, { "epoch": 0.5170564868090565, "grad_norm": 1.3982634140007355, "learning_rate": 9.93364912685919e-06, "loss": 0.5345, "step": 14513 }, { "epoch": 0.5170921139355505, "grad_norm": 1.0569572113686572, "learning_rate": 9.932495216008926e-06, "loss": 0.4245, "step": 14514 }, { "epoch": 0.5171277410620446, "grad_norm": 1.322562419800743, "learning_rate": 9.931341306057539e-06, "loss": 0.4635, "step": 14515 }, { "epoch": 0.5171633681885388, "grad_norm": 1.2630346571001578, "learning_rate": 9.930187397020385e-06, "loss": 0.4769, "step": 14516 }, { "epoch": 0.5171989953150329, "grad_norm": 1.0952456680514926, "learning_rate": 9.92903348891284e-06, "loss": 0.413, "step": 14517 }, { "epoch": 0.517234622441527, "grad_norm": 1.370601092867061, "learning_rate": 9.92787958175026e-06, "loss": 0.5785, "step": 14518 }, { "epoch": 0.5172702495680211, "grad_norm": 1.6238856190493816, "learning_rate": 9.926725675548013e-06, "loss": 0.5258, "step": 14519 }, { "epoch": 0.5173058766945152, "grad_norm": 1.1570849491422293, "learning_rate": 9.925571770321469e-06, "loss": 0.4454, "step": 14520 }, { "epoch": 0.5173415038210093, "grad_norm": 1.0363632488888883, "learning_rate": 9.924417866085985e-06, "loss": 0.4443, "step": 14521 }, { "epoch": 0.5173771309475035, "grad_norm": 1.8491939811175584, "learning_rate": 9.923263962856933e-06, "loss": 0.5977, "step": 14522 }, { "epoch": 0.5174127580739976, "grad_norm": 1.3962808750724023, "learning_rate": 9.922110060649672e-06, "loss": 0.5651, "step": 14523 }, { "epoch": 0.5174483852004916, "grad_norm": 1.2402694385068163, "learning_rate": 9.920956159479575e-06, "loss": 0.5252, "step": 14524 }, { "epoch": 0.5174840123269858, "grad_norm": 1.259230706550836, "learning_rate": 9.919802259361995e-06, "loss": 0.4802, "step": 14525 }, { "epoch": 0.5175196394534799, "grad_norm": 1.2432552299248427, "learning_rate": 9.918648360312312e-06, "loss": 0.503, "step": 14526 }, { "epoch": 0.517555266579974, "grad_norm": 1.1168034171714976, "learning_rate": 9.917494462345877e-06, "loss": 0.4715, "step": 14527 }, { "epoch": 0.5175908937064682, "grad_norm": 1.1094363069359179, "learning_rate": 9.916340565478062e-06, "loss": 0.4338, "step": 14528 }, { "epoch": 0.5176265208329622, "grad_norm": 1.927069949649558, "learning_rate": 9.91518666972423e-06, "loss": 0.57, "step": 14529 }, { "epoch": 0.5176621479594563, "grad_norm": 1.0036015186931568, "learning_rate": 9.91403277509975e-06, "loss": 0.4365, "step": 14530 }, { "epoch": 0.5176977750859505, "grad_norm": 1.3514012707109182, "learning_rate": 9.91287888161998e-06, "loss": 0.5559, "step": 14531 }, { "epoch": 0.5177334022124446, "grad_norm": 1.2566658330312384, "learning_rate": 9.91172498930029e-06, "loss": 0.5274, "step": 14532 }, { "epoch": 0.5177690293389386, "grad_norm": 1.2863102686716275, "learning_rate": 9.910571098156045e-06, "loss": 0.4797, "step": 14533 }, { "epoch": 0.5178046564654328, "grad_norm": 1.297676536927504, "learning_rate": 9.909417208202604e-06, "loss": 0.5033, "step": 14534 }, { "epoch": 0.5178402835919269, "grad_norm": 1.9724713794581499, "learning_rate": 9.908263319455344e-06, "loss": 0.4681, "step": 14535 }, { "epoch": 0.517875910718421, "grad_norm": 1.0823852676085914, "learning_rate": 9.907109431929615e-06, "loss": 0.4317, "step": 14536 }, { "epoch": 0.5179115378449152, "grad_norm": 1.0111088118572686, "learning_rate": 9.905955545640791e-06, "loss": 0.4409, "step": 14537 }, { "epoch": 0.5179471649714092, "grad_norm": 1.7925919383471756, "learning_rate": 9.904801660604234e-06, "loss": 0.6034, "step": 14538 }, { "epoch": 0.5179827920979033, "grad_norm": 1.3409332351412016, "learning_rate": 9.903647776835312e-06, "loss": 0.5158, "step": 14539 }, { "epoch": 0.5180184192243975, "grad_norm": 1.2264857903439872, "learning_rate": 9.902493894349383e-06, "loss": 0.4088, "step": 14540 }, { "epoch": 0.5180540463508916, "grad_norm": 1.2759856278383508, "learning_rate": 9.901340013161822e-06, "loss": 0.4881, "step": 14541 }, { "epoch": 0.5180896734773857, "grad_norm": 2.4821726365751946, "learning_rate": 9.900186133287982e-06, "loss": 0.6138, "step": 14542 }, { "epoch": 0.5181253006038798, "grad_norm": 9.193639047548281, "learning_rate": 9.899032254743235e-06, "loss": 0.4729, "step": 14543 }, { "epoch": 0.5181609277303739, "grad_norm": 1.1587651377335546, "learning_rate": 9.897878377542944e-06, "loss": 0.4637, "step": 14544 }, { "epoch": 0.518196554856868, "grad_norm": 1.4112812363214302, "learning_rate": 9.896724501702474e-06, "loss": 0.5811, "step": 14545 }, { "epoch": 0.5182321819833622, "grad_norm": 1.5786189094874086, "learning_rate": 9.895570627237191e-06, "loss": 0.5958, "step": 14546 }, { "epoch": 0.5182678091098563, "grad_norm": 1.472824940504933, "learning_rate": 9.894416754162458e-06, "loss": 0.6139, "step": 14547 }, { "epoch": 0.5183034362363503, "grad_norm": 1.5985298304208035, "learning_rate": 9.893262882493638e-06, "loss": 0.5359, "step": 14548 }, { "epoch": 0.5183390633628445, "grad_norm": 1.5681783685298853, "learning_rate": 9.892109012246097e-06, "loss": 0.5596, "step": 14549 }, { "epoch": 0.5183746904893386, "grad_norm": 1.043806382088803, "learning_rate": 9.890955143435203e-06, "loss": 0.4538, "step": 14550 }, { "epoch": 0.5184103176158327, "grad_norm": 1.0661361986269766, "learning_rate": 9.889801276076312e-06, "loss": 0.4908, "step": 14551 }, { "epoch": 0.5184459447423269, "grad_norm": 1.51448700962699, "learning_rate": 9.888647410184801e-06, "loss": 0.5801, "step": 14552 }, { "epoch": 0.5184815718688209, "grad_norm": 1.3677344925301658, "learning_rate": 9.887493545776022e-06, "loss": 0.6158, "step": 14553 }, { "epoch": 0.518517198995315, "grad_norm": 1.5994461699130367, "learning_rate": 9.886339682865345e-06, "loss": 0.6481, "step": 14554 }, { "epoch": 0.5185528261218092, "grad_norm": 2.195513606528609, "learning_rate": 9.885185821468138e-06, "loss": 0.5899, "step": 14555 }, { "epoch": 0.5185884532483033, "grad_norm": 1.4648687560963523, "learning_rate": 9.88403196159976e-06, "loss": 0.5875, "step": 14556 }, { "epoch": 0.5186240803747973, "grad_norm": 1.4834954315417954, "learning_rate": 9.882878103275581e-06, "loss": 0.5418, "step": 14557 }, { "epoch": 0.5186597075012915, "grad_norm": 1.0454559057285449, "learning_rate": 9.881724246510958e-06, "loss": 0.4467, "step": 14558 }, { "epoch": 0.5186953346277856, "grad_norm": 1.126189138489128, "learning_rate": 9.88057039132126e-06, "loss": 0.4289, "step": 14559 }, { "epoch": 0.5187309617542797, "grad_norm": 1.2298421370292383, "learning_rate": 9.879416537721849e-06, "loss": 0.5072, "step": 14560 }, { "epoch": 0.5187665888807739, "grad_norm": 1.1564248245068076, "learning_rate": 9.878262685728097e-06, "loss": 0.4635, "step": 14561 }, { "epoch": 0.518802216007268, "grad_norm": 1.341231511276872, "learning_rate": 9.877108835355357e-06, "loss": 0.5583, "step": 14562 }, { "epoch": 0.518837843133762, "grad_norm": 1.1489570070284427, "learning_rate": 9.875954986619e-06, "loss": 0.5122, "step": 14563 }, { "epoch": 0.5188734702602562, "grad_norm": 1.0536388421867398, "learning_rate": 9.874801139534389e-06, "loss": 0.4591, "step": 14564 }, { "epoch": 0.5189090973867503, "grad_norm": 1.0948821327137, "learning_rate": 9.873647294116888e-06, "loss": 0.4727, "step": 14565 }, { "epoch": 0.5189447245132444, "grad_norm": 1.9109274174363553, "learning_rate": 9.872493450381863e-06, "loss": 0.5065, "step": 14566 }, { "epoch": 0.5189803516397385, "grad_norm": 1.0344424803500278, "learning_rate": 9.871339608344675e-06, "loss": 0.4538, "step": 14567 }, { "epoch": 0.5190159787662326, "grad_norm": 1.5852494906805645, "learning_rate": 9.870185768020694e-06, "loss": 0.609, "step": 14568 }, { "epoch": 0.5190516058927267, "grad_norm": 1.5290524785308877, "learning_rate": 9.869031929425275e-06, "loss": 0.5039, "step": 14569 }, { "epoch": 0.5190872330192209, "grad_norm": 1.7734385776322057, "learning_rate": 9.86787809257379e-06, "loss": 0.583, "step": 14570 }, { "epoch": 0.519122860145715, "grad_norm": 1.654350019197479, "learning_rate": 9.866724257481601e-06, "loss": 0.506, "step": 14571 }, { "epoch": 0.519158487272209, "grad_norm": 1.0868768149155745, "learning_rate": 9.865570424164073e-06, "loss": 0.4956, "step": 14572 }, { "epoch": 0.5191941143987032, "grad_norm": 1.1113615791949147, "learning_rate": 9.864416592636565e-06, "loss": 0.4412, "step": 14573 }, { "epoch": 0.5192297415251973, "grad_norm": 1.5856253278591819, "learning_rate": 9.863262762914448e-06, "loss": 0.5563, "step": 14574 }, { "epoch": 0.5192653686516914, "grad_norm": 1.0725722649466314, "learning_rate": 9.862108935013079e-06, "loss": 0.4491, "step": 14575 }, { "epoch": 0.5193009957781856, "grad_norm": 1.096708584515708, "learning_rate": 9.860955108947832e-06, "loss": 0.5366, "step": 14576 }, { "epoch": 0.5193366229046796, "grad_norm": 0.9809588854921627, "learning_rate": 9.859801284734057e-06, "loss": 0.4171, "step": 14577 }, { "epoch": 0.5193722500311737, "grad_norm": 1.3990257356252207, "learning_rate": 9.85864746238713e-06, "loss": 0.5414, "step": 14578 }, { "epoch": 0.5194078771576679, "grad_norm": 0.9965953890804103, "learning_rate": 9.857493641922414e-06, "loss": 0.4756, "step": 14579 }, { "epoch": 0.519443504284162, "grad_norm": 1.0110225184723085, "learning_rate": 9.856339823355265e-06, "loss": 0.4508, "step": 14580 }, { "epoch": 0.519479131410656, "grad_norm": 0.9633928460145036, "learning_rate": 9.855186006701054e-06, "loss": 0.4057, "step": 14581 }, { "epoch": 0.5195147585371502, "grad_norm": 1.3050511680203858, "learning_rate": 9.85403219197514e-06, "loss": 0.6113, "step": 14582 }, { "epoch": 0.5195503856636443, "grad_norm": 1.6795441056517464, "learning_rate": 9.852878379192895e-06, "loss": 0.543, "step": 14583 }, { "epoch": 0.5195860127901384, "grad_norm": 1.536272602554404, "learning_rate": 9.851724568369673e-06, "loss": 0.5522, "step": 14584 }, { "epoch": 0.5196216399166326, "grad_norm": 1.3821642822718543, "learning_rate": 9.850570759520844e-06, "loss": 0.587, "step": 14585 }, { "epoch": 0.5196572670431266, "grad_norm": 1.4245537879692, "learning_rate": 9.849416952661765e-06, "loss": 0.5523, "step": 14586 }, { "epoch": 0.5196928941696207, "grad_norm": 1.049981478071578, "learning_rate": 9.848263147807812e-06, "loss": 0.4074, "step": 14587 }, { "epoch": 0.5197285212961149, "grad_norm": 1.07811359837589, "learning_rate": 9.847109344974333e-06, "loss": 0.4662, "step": 14588 }, { "epoch": 0.519764148422609, "grad_norm": 1.284733822370401, "learning_rate": 9.845955544176704e-06, "loss": 0.507, "step": 14589 }, { "epoch": 0.5197997755491031, "grad_norm": 1.5087033451300544, "learning_rate": 9.844801745430287e-06, "loss": 0.5932, "step": 14590 }, { "epoch": 0.5198354026755972, "grad_norm": 1.1691442696052687, "learning_rate": 9.843647948750439e-06, "loss": 0.4711, "step": 14591 }, { "epoch": 0.5198710298020913, "grad_norm": 1.695172201599601, "learning_rate": 9.84249415415253e-06, "loss": 0.536, "step": 14592 }, { "epoch": 0.5199066569285854, "grad_norm": 1.105996193989332, "learning_rate": 9.841340361651921e-06, "loss": 0.4588, "step": 14593 }, { "epoch": 0.5199422840550796, "grad_norm": 1.731563155880241, "learning_rate": 9.84018657126398e-06, "loss": 0.5853, "step": 14594 }, { "epoch": 0.5199779111815737, "grad_norm": 1.0178399477578655, "learning_rate": 9.83903278300406e-06, "loss": 0.4537, "step": 14595 }, { "epoch": 0.5200135383080677, "grad_norm": 1.2342587089426886, "learning_rate": 9.837878996887534e-06, "loss": 0.4158, "step": 14596 }, { "epoch": 0.5200491654345619, "grad_norm": 2.128024981150204, "learning_rate": 9.836725212929762e-06, "loss": 0.5848, "step": 14597 }, { "epoch": 0.520084792561056, "grad_norm": 1.1507953432109563, "learning_rate": 9.83557143114611e-06, "loss": 0.4951, "step": 14598 }, { "epoch": 0.5201204196875501, "grad_norm": 2.318808945758182, "learning_rate": 9.834417651551936e-06, "loss": 0.5897, "step": 14599 }, { "epoch": 0.5201560468140443, "grad_norm": 1.2258482742176908, "learning_rate": 9.833263874162604e-06, "loss": 0.5738, "step": 14600 }, { "epoch": 0.5201916739405383, "grad_norm": 1.9579934675934856, "learning_rate": 9.832110098993489e-06, "loss": 0.5072, "step": 14601 }, { "epoch": 0.5202273010670324, "grad_norm": 1.4210441217053573, "learning_rate": 9.830956326059938e-06, "loss": 0.5495, "step": 14602 }, { "epoch": 0.5202629281935266, "grad_norm": 1.7360204385335047, "learning_rate": 9.829802555377326e-06, "loss": 0.4816, "step": 14603 }, { "epoch": 0.5202985553200207, "grad_norm": 1.3454068417569671, "learning_rate": 9.828648786961009e-06, "loss": 0.5431, "step": 14604 }, { "epoch": 0.5203341824465147, "grad_norm": 1.0596556259476688, "learning_rate": 9.827495020826357e-06, "loss": 0.3989, "step": 14605 }, { "epoch": 0.5203698095730089, "grad_norm": 1.4554780094187976, "learning_rate": 9.826341256988725e-06, "loss": 0.5865, "step": 14606 }, { "epoch": 0.520405436699503, "grad_norm": 1.4828711059297075, "learning_rate": 9.825187495463482e-06, "loss": 0.5777, "step": 14607 }, { "epoch": 0.5204410638259971, "grad_norm": 1.07329180663419, "learning_rate": 9.82403373626599e-06, "loss": 0.4885, "step": 14608 }, { "epoch": 0.5204766909524913, "grad_norm": 1.128223537747089, "learning_rate": 9.822879979411613e-06, "loss": 0.4399, "step": 14609 }, { "epoch": 0.5205123180789853, "grad_norm": 1.1371776811458194, "learning_rate": 9.82172622491571e-06, "loss": 0.5019, "step": 14610 }, { "epoch": 0.5205479452054794, "grad_norm": 1.449473743721314, "learning_rate": 9.820572472793651e-06, "loss": 0.5166, "step": 14611 }, { "epoch": 0.5205835723319736, "grad_norm": 1.102058921816048, "learning_rate": 9.81941872306079e-06, "loss": 0.5267, "step": 14612 }, { "epoch": 0.5206191994584677, "grad_norm": 1.05762907654523, "learning_rate": 9.818264975732497e-06, "loss": 0.4503, "step": 14613 }, { "epoch": 0.5206548265849618, "grad_norm": 1.097434739374761, "learning_rate": 9.817111230824133e-06, "loss": 0.4707, "step": 14614 }, { "epoch": 0.520690453711456, "grad_norm": 1.0045913638094228, "learning_rate": 9.81595748835106e-06, "loss": 0.4333, "step": 14615 }, { "epoch": 0.52072608083795, "grad_norm": 1.6080155602548531, "learning_rate": 9.814803748328646e-06, "loss": 0.5391, "step": 14616 }, { "epoch": 0.5207617079644441, "grad_norm": 1.6582539977008235, "learning_rate": 9.813650010772244e-06, "loss": 0.6238, "step": 14617 }, { "epoch": 0.5207973350909383, "grad_norm": 1.0948076368952968, "learning_rate": 9.812496275697226e-06, "loss": 0.4405, "step": 14618 }, { "epoch": 0.5208329622174324, "grad_norm": 1.691903942479527, "learning_rate": 9.811342543118947e-06, "loss": 0.5471, "step": 14619 }, { "epoch": 0.5208685893439264, "grad_norm": 0.9804540585604568, "learning_rate": 9.81018881305278e-06, "loss": 0.4303, "step": 14620 }, { "epoch": 0.5209042164704206, "grad_norm": 1.1608402851239725, "learning_rate": 9.809035085514078e-06, "loss": 0.4275, "step": 14621 }, { "epoch": 0.5209398435969147, "grad_norm": 0.9719590939971128, "learning_rate": 9.80788136051821e-06, "loss": 0.4018, "step": 14622 }, { "epoch": 0.5209754707234088, "grad_norm": 1.0163672939635597, "learning_rate": 9.80672763808053e-06, "loss": 0.4031, "step": 14623 }, { "epoch": 0.521011097849903, "grad_norm": 1.141934444803042, "learning_rate": 9.80557391821641e-06, "loss": 0.4914, "step": 14624 }, { "epoch": 0.521046724976397, "grad_norm": 1.1096048003504961, "learning_rate": 9.804420200941211e-06, "loss": 0.4733, "step": 14625 }, { "epoch": 0.5210823521028911, "grad_norm": 1.0741136981236177, "learning_rate": 9.803266486270289e-06, "loss": 0.4984, "step": 14626 }, { "epoch": 0.5211179792293853, "grad_norm": 1.673580606698766, "learning_rate": 9.802112774219017e-06, "loss": 0.5592, "step": 14627 }, { "epoch": 0.5211536063558794, "grad_norm": 1.40827733250975, "learning_rate": 9.800959064802748e-06, "loss": 0.5214, "step": 14628 }, { "epoch": 0.5211892334823734, "grad_norm": 1.1162483807662673, "learning_rate": 9.79980535803685e-06, "loss": 0.4747, "step": 14629 }, { "epoch": 0.5212248606088676, "grad_norm": 1.1650141188141219, "learning_rate": 9.798651653936682e-06, "loss": 0.4487, "step": 14630 }, { "epoch": 0.5212604877353617, "grad_norm": 1.50091574451976, "learning_rate": 9.79749795251761e-06, "loss": 0.4057, "step": 14631 }, { "epoch": 0.5212961148618558, "grad_norm": 2.480044437119571, "learning_rate": 9.796344253794993e-06, "loss": 0.5674, "step": 14632 }, { "epoch": 0.52133174198835, "grad_norm": 1.1486549746033614, "learning_rate": 9.795190557784196e-06, "loss": 0.4078, "step": 14633 }, { "epoch": 0.521367369114844, "grad_norm": 1.436927327404257, "learning_rate": 9.794036864500577e-06, "loss": 0.5882, "step": 14634 }, { "epoch": 0.5214029962413381, "grad_norm": 1.2723104401305894, "learning_rate": 9.7928831739595e-06, "loss": 0.5926, "step": 14635 }, { "epoch": 0.5214386233678323, "grad_norm": 1.0770381007141885, "learning_rate": 9.791729486176335e-06, "loss": 0.4074, "step": 14636 }, { "epoch": 0.5214742504943264, "grad_norm": 1.4057938125383829, "learning_rate": 9.790575801166432e-06, "loss": 0.5494, "step": 14637 }, { "epoch": 0.5215098776208205, "grad_norm": 1.0300587579506584, "learning_rate": 9.78942211894516e-06, "loss": 0.4837, "step": 14638 }, { "epoch": 0.5215455047473146, "grad_norm": 1.0683233698793941, "learning_rate": 9.788268439527878e-06, "loss": 0.4083, "step": 14639 }, { "epoch": 0.5215811318738087, "grad_norm": 1.3319675534344266, "learning_rate": 9.787114762929953e-06, "loss": 0.6145, "step": 14640 }, { "epoch": 0.5216167590003028, "grad_norm": 1.8313025746372331, "learning_rate": 9.78596108916674e-06, "loss": 0.4848, "step": 14641 }, { "epoch": 0.521652386126797, "grad_norm": 1.0125407675509477, "learning_rate": 9.78480741825361e-06, "loss": 0.4415, "step": 14642 }, { "epoch": 0.5216880132532911, "grad_norm": 1.1859834536281728, "learning_rate": 9.783653750205916e-06, "loss": 0.4109, "step": 14643 }, { "epoch": 0.5217236403797851, "grad_norm": 1.620356608621874, "learning_rate": 9.782500085039025e-06, "loss": 0.6142, "step": 14644 }, { "epoch": 0.5217592675062793, "grad_norm": 1.970344580556742, "learning_rate": 9.781346422768294e-06, "loss": 0.5948, "step": 14645 }, { "epoch": 0.5217948946327734, "grad_norm": 1.2526899538959348, "learning_rate": 9.780192763409089e-06, "loss": 0.5284, "step": 14646 }, { "epoch": 0.5218305217592675, "grad_norm": 2.3176650219832404, "learning_rate": 9.779039106976776e-06, "loss": 0.5921, "step": 14647 }, { "epoch": 0.5218661488857617, "grad_norm": 1.0851237293579148, "learning_rate": 9.777885453486706e-06, "loss": 0.4005, "step": 14648 }, { "epoch": 0.5219017760122557, "grad_norm": 1.2166411586687267, "learning_rate": 9.77673180295425e-06, "loss": 0.3944, "step": 14649 }, { "epoch": 0.5219374031387498, "grad_norm": 1.1761965547161417, "learning_rate": 9.775578155394764e-06, "loss": 0.4628, "step": 14650 }, { "epoch": 0.521973030265244, "grad_norm": 1.1855060163348299, "learning_rate": 9.774424510823615e-06, "loss": 0.502, "step": 14651 }, { "epoch": 0.5220086573917381, "grad_norm": 1.3229776317244002, "learning_rate": 9.773270869256156e-06, "loss": 0.5826, "step": 14652 }, { "epoch": 0.5220442845182321, "grad_norm": 1.6789292318820022, "learning_rate": 9.772117230707759e-06, "loss": 0.6006, "step": 14653 }, { "epoch": 0.5220799116447263, "grad_norm": 1.1471122274992307, "learning_rate": 9.770963595193778e-06, "loss": 0.4448, "step": 14654 }, { "epoch": 0.5221155387712204, "grad_norm": 1.2412409695743678, "learning_rate": 9.769809962729578e-06, "loss": 0.4747, "step": 14655 }, { "epoch": 0.5221511658977145, "grad_norm": 1.157841271119803, "learning_rate": 9.768656333330516e-06, "loss": 0.4642, "step": 14656 }, { "epoch": 0.5221867930242087, "grad_norm": 1.224988924417455, "learning_rate": 9.767502707011962e-06, "loss": 0.4408, "step": 14657 }, { "epoch": 0.5222224201507027, "grad_norm": 1.0585561223701712, "learning_rate": 9.766349083789266e-06, "loss": 0.4552, "step": 14658 }, { "epoch": 0.5222580472771968, "grad_norm": 1.204214878528315, "learning_rate": 9.765195463677796e-06, "loss": 0.4629, "step": 14659 }, { "epoch": 0.522293674403691, "grad_norm": 1.5029364719590264, "learning_rate": 9.764041846692917e-06, "loss": 0.5401, "step": 14660 }, { "epoch": 0.5223293015301851, "grad_norm": 1.3629728936698757, "learning_rate": 9.76288823284998e-06, "loss": 0.4587, "step": 14661 }, { "epoch": 0.5223649286566792, "grad_norm": 1.0982074274982854, "learning_rate": 9.76173462216436e-06, "loss": 0.4726, "step": 14662 }, { "epoch": 0.5224005557831733, "grad_norm": 1.7913733291986942, "learning_rate": 9.760581014651404e-06, "loss": 0.6494, "step": 14663 }, { "epoch": 0.5224361829096674, "grad_norm": 1.0543352010420362, "learning_rate": 9.75942741032648e-06, "loss": 0.471, "step": 14664 }, { "epoch": 0.5224718100361615, "grad_norm": 1.6215659805027518, "learning_rate": 9.758273809204948e-06, "loss": 0.5068, "step": 14665 }, { "epoch": 0.5225074371626557, "grad_norm": 1.1609933004952633, "learning_rate": 9.75712021130217e-06, "loss": 0.4177, "step": 14666 }, { "epoch": 0.5225430642891498, "grad_norm": 1.6214185224257045, "learning_rate": 9.755966616633504e-06, "loss": 0.554, "step": 14667 }, { "epoch": 0.5225786914156438, "grad_norm": 1.2074838196427666, "learning_rate": 9.754813025214318e-06, "loss": 0.4861, "step": 14668 }, { "epoch": 0.522614318542138, "grad_norm": 1.128906503611062, "learning_rate": 9.753659437059964e-06, "loss": 0.45, "step": 14669 }, { "epoch": 0.5226499456686321, "grad_norm": 1.3632906177180115, "learning_rate": 9.752505852185807e-06, "loss": 0.5383, "step": 14670 }, { "epoch": 0.5226855727951262, "grad_norm": 1.0452354479602322, "learning_rate": 9.751352270607208e-06, "loss": 0.4316, "step": 14671 }, { "epoch": 0.5227211999216204, "grad_norm": 1.2352191314865144, "learning_rate": 9.750198692339527e-06, "loss": 0.5307, "step": 14672 }, { "epoch": 0.5227568270481144, "grad_norm": 1.5241783331990069, "learning_rate": 9.749045117398129e-06, "loss": 0.5979, "step": 14673 }, { "epoch": 0.5227924541746085, "grad_norm": 1.0892653229369909, "learning_rate": 9.747891545798365e-06, "loss": 0.4321, "step": 14674 }, { "epoch": 0.5228280813011027, "grad_norm": 1.067988966167497, "learning_rate": 9.746737977555604e-06, "loss": 0.4475, "step": 14675 }, { "epoch": 0.5228637084275968, "grad_norm": 2.031011832491235, "learning_rate": 9.745584412685202e-06, "loss": 0.5883, "step": 14676 }, { "epoch": 0.5228993355540908, "grad_norm": 1.7250298356882496, "learning_rate": 9.744430851202523e-06, "loss": 0.5931, "step": 14677 }, { "epoch": 0.522934962680585, "grad_norm": 1.353294486616944, "learning_rate": 9.743277293122925e-06, "loss": 0.5549, "step": 14678 }, { "epoch": 0.5229705898070791, "grad_norm": 1.1506859622776469, "learning_rate": 9.742123738461773e-06, "loss": 0.4757, "step": 14679 }, { "epoch": 0.5230062169335732, "grad_norm": 1.074389494416746, "learning_rate": 9.740970187234419e-06, "loss": 0.4654, "step": 14680 }, { "epoch": 0.5230418440600674, "grad_norm": 1.7463751076054, "learning_rate": 9.739816639456227e-06, "loss": 0.5955, "step": 14681 }, { "epoch": 0.5230774711865614, "grad_norm": 1.1282269617070473, "learning_rate": 9.738663095142562e-06, "loss": 0.5015, "step": 14682 }, { "epoch": 0.5231130983130555, "grad_norm": 1.3977388025978617, "learning_rate": 9.737509554308778e-06, "loss": 0.4562, "step": 14683 }, { "epoch": 0.5231487254395497, "grad_norm": 1.1124993630599096, "learning_rate": 9.736356016970243e-06, "loss": 0.4716, "step": 14684 }, { "epoch": 0.5231843525660438, "grad_norm": 1.7045045266068557, "learning_rate": 9.735202483142306e-06, "loss": 0.6138, "step": 14685 }, { "epoch": 0.5232199796925379, "grad_norm": 2.0447616529223964, "learning_rate": 9.734048952840335e-06, "loss": 0.4987, "step": 14686 }, { "epoch": 0.523255606819032, "grad_norm": 1.1756289688778399, "learning_rate": 9.732895426079688e-06, "loss": 0.4359, "step": 14687 }, { "epoch": 0.5232912339455261, "grad_norm": 1.0382890400351288, "learning_rate": 9.731741902875728e-06, "loss": 0.4648, "step": 14688 }, { "epoch": 0.5233268610720202, "grad_norm": 1.4121671328666066, "learning_rate": 9.730588383243806e-06, "loss": 0.5574, "step": 14689 }, { "epoch": 0.5233624881985144, "grad_norm": 1.2256857427142318, "learning_rate": 9.729434867199294e-06, "loss": 0.5155, "step": 14690 }, { "epoch": 0.5233981153250085, "grad_norm": 1.4156191402016798, "learning_rate": 9.72828135475754e-06, "loss": 0.5528, "step": 14691 }, { "epoch": 0.5234337424515025, "grad_norm": 1.3112443972678776, "learning_rate": 9.727127845933915e-06, "loss": 0.5238, "step": 14692 }, { "epoch": 0.5234693695779967, "grad_norm": 1.7677228401267568, "learning_rate": 9.725974340743769e-06, "loss": 0.521, "step": 14693 }, { "epoch": 0.5235049967044908, "grad_norm": 1.8181836664145112, "learning_rate": 9.724820839202468e-06, "loss": 0.5121, "step": 14694 }, { "epoch": 0.5235406238309849, "grad_norm": 1.03648344356819, "learning_rate": 9.723667341325373e-06, "loss": 0.4152, "step": 14695 }, { "epoch": 0.5235762509574791, "grad_norm": 1.8871468831320348, "learning_rate": 9.722513847127836e-06, "loss": 0.5565, "step": 14696 }, { "epoch": 0.5236118780839731, "grad_norm": 1.175956590911629, "learning_rate": 9.721360356625223e-06, "loss": 0.4462, "step": 14697 }, { "epoch": 0.5236475052104672, "grad_norm": 1.025973707223183, "learning_rate": 9.72020686983289e-06, "loss": 0.4507, "step": 14698 }, { "epoch": 0.5236831323369614, "grad_norm": 1.163746334987742, "learning_rate": 9.719053386766204e-06, "loss": 0.4612, "step": 14699 }, { "epoch": 0.5237187594634555, "grad_norm": 1.0923608591173484, "learning_rate": 9.717899907440513e-06, "loss": 0.4569, "step": 14700 }, { "epoch": 0.5237543865899495, "grad_norm": 1.4267427890195286, "learning_rate": 9.716746431871184e-06, "loss": 0.5358, "step": 14701 }, { "epoch": 0.5237900137164437, "grad_norm": 1.4642683370991563, "learning_rate": 9.71559296007357e-06, "loss": 0.5341, "step": 14702 }, { "epoch": 0.5238256408429378, "grad_norm": 1.3692370098353996, "learning_rate": 9.71443949206304e-06, "loss": 0.5576, "step": 14703 }, { "epoch": 0.5238612679694319, "grad_norm": 1.0766645088163453, "learning_rate": 9.713286027854943e-06, "loss": 0.4542, "step": 14704 }, { "epoch": 0.5238968950959261, "grad_norm": 1.7308696149450655, "learning_rate": 9.712132567464645e-06, "loss": 0.5632, "step": 14705 }, { "epoch": 0.5239325222224201, "grad_norm": 1.396321477799517, "learning_rate": 9.710979110907506e-06, "loss": 0.514, "step": 14706 }, { "epoch": 0.5239681493489142, "grad_norm": 1.2893979344484017, "learning_rate": 9.709825658198878e-06, "loss": 0.5633, "step": 14707 }, { "epoch": 0.5240037764754084, "grad_norm": 1.106917395920499, "learning_rate": 9.708672209354128e-06, "loss": 0.4373, "step": 14708 }, { "epoch": 0.5240394036019025, "grad_norm": 1.624606002292941, "learning_rate": 9.707518764388607e-06, "loss": 0.5168, "step": 14709 }, { "epoch": 0.5240750307283966, "grad_norm": 1.7847858430033208, "learning_rate": 9.706365323317684e-06, "loss": 0.6067, "step": 14710 }, { "epoch": 0.5241106578548907, "grad_norm": 1.5290112926045323, "learning_rate": 9.705211886156706e-06, "loss": 0.5198, "step": 14711 }, { "epoch": 0.5241462849813848, "grad_norm": 1.443514716075719, "learning_rate": 9.70405845292104e-06, "loss": 0.5737, "step": 14712 }, { "epoch": 0.5241819121078789, "grad_norm": 1.0505077144154276, "learning_rate": 9.702905023626041e-06, "loss": 0.4658, "step": 14713 }, { "epoch": 0.5242175392343731, "grad_norm": 1.0435350237304082, "learning_rate": 9.701751598287074e-06, "loss": 0.4697, "step": 14714 }, { "epoch": 0.5242531663608672, "grad_norm": 1.115498079635092, "learning_rate": 9.700598176919485e-06, "loss": 0.452, "step": 14715 }, { "epoch": 0.5242887934873612, "grad_norm": 1.1489841710563915, "learning_rate": 9.699444759538644e-06, "loss": 0.4775, "step": 14716 }, { "epoch": 0.5243244206138554, "grad_norm": 1.7751210730645748, "learning_rate": 9.698291346159911e-06, "loss": 0.5428, "step": 14717 }, { "epoch": 0.5243600477403495, "grad_norm": 1.4744276467891966, "learning_rate": 9.697137936798635e-06, "loss": 0.5557, "step": 14718 }, { "epoch": 0.5243956748668436, "grad_norm": 1.2872718400808667, "learning_rate": 9.69598453147018e-06, "loss": 0.501, "step": 14719 }, { "epoch": 0.5244313019933378, "grad_norm": 1.5350898162775852, "learning_rate": 9.694831130189903e-06, "loss": 0.6669, "step": 14720 }, { "epoch": 0.5244669291198318, "grad_norm": 0.9734147485092685, "learning_rate": 9.693677732973166e-06, "loss": 0.4273, "step": 14721 }, { "epoch": 0.5245025562463259, "grad_norm": 1.1418680805440022, "learning_rate": 9.69252433983532e-06, "loss": 0.4771, "step": 14722 }, { "epoch": 0.5245381833728201, "grad_norm": 1.0515616909437318, "learning_rate": 9.691370950791728e-06, "loss": 0.4094, "step": 14723 }, { "epoch": 0.5245738104993142, "grad_norm": 1.3662070587159665, "learning_rate": 9.690217565857747e-06, "loss": 0.6269, "step": 14724 }, { "epoch": 0.5246094376258083, "grad_norm": 1.2568908349282057, "learning_rate": 9.68906418504874e-06, "loss": 0.5228, "step": 14725 }, { "epoch": 0.5246450647523024, "grad_norm": 1.1579089077456153, "learning_rate": 9.687910808380057e-06, "loss": 0.4363, "step": 14726 }, { "epoch": 0.5246806918787965, "grad_norm": 1.4356083890925897, "learning_rate": 9.686757435867055e-06, "loss": 0.4482, "step": 14727 }, { "epoch": 0.5247163190052906, "grad_norm": 1.3904770141354883, "learning_rate": 9.685604067525106e-06, "loss": 0.5854, "step": 14728 }, { "epoch": 0.5247519461317848, "grad_norm": 1.049616580294635, "learning_rate": 9.684450703369553e-06, "loss": 0.5026, "step": 14729 }, { "epoch": 0.5247875732582788, "grad_norm": 1.2916571434685287, "learning_rate": 9.683297343415762e-06, "loss": 0.5147, "step": 14730 }, { "epoch": 0.5248232003847729, "grad_norm": 1.6584324022891814, "learning_rate": 9.682143987679086e-06, "loss": 0.5697, "step": 14731 }, { "epoch": 0.5248588275112671, "grad_norm": 1.655676645464235, "learning_rate": 9.680990636174888e-06, "loss": 0.4916, "step": 14732 }, { "epoch": 0.5248944546377612, "grad_norm": 1.1143386538117999, "learning_rate": 9.67983728891852e-06, "loss": 0.4677, "step": 14733 }, { "epoch": 0.5249300817642553, "grad_norm": 1.9965131593268477, "learning_rate": 9.678683945925343e-06, "loss": 0.5007, "step": 14734 }, { "epoch": 0.5249657088907494, "grad_norm": 1.787952228837085, "learning_rate": 9.677530607210714e-06, "loss": 0.5738, "step": 14735 }, { "epoch": 0.5250013360172435, "grad_norm": 1.222971936847798, "learning_rate": 9.676377272789991e-06, "loss": 0.4863, "step": 14736 }, { "epoch": 0.5250369631437376, "grad_norm": 1.0663471106468143, "learning_rate": 9.67522394267853e-06, "loss": 0.433, "step": 14737 }, { "epoch": 0.5250725902702318, "grad_norm": 1.1632955773811233, "learning_rate": 9.67407061689169e-06, "loss": 0.444, "step": 14738 }, { "epoch": 0.5251082173967259, "grad_norm": 2.0608696001723374, "learning_rate": 9.672917295444826e-06, "loss": 0.5039, "step": 14739 }, { "epoch": 0.5251438445232199, "grad_norm": 1.4748784527934713, "learning_rate": 9.671763978353296e-06, "loss": 0.5933, "step": 14740 }, { "epoch": 0.5251794716497141, "grad_norm": 1.511671126530539, "learning_rate": 9.67061066563246e-06, "loss": 0.6219, "step": 14741 }, { "epoch": 0.5252150987762082, "grad_norm": 1.7074917556549867, "learning_rate": 9.66945735729767e-06, "loss": 0.6644, "step": 14742 }, { "epoch": 0.5252507259027023, "grad_norm": 1.3628306339047436, "learning_rate": 9.668304053364293e-06, "loss": 0.5894, "step": 14743 }, { "epoch": 0.5252863530291965, "grad_norm": 1.0699486872478843, "learning_rate": 9.667150753847676e-06, "loss": 0.4026, "step": 14744 }, { "epoch": 0.5253219801556905, "grad_norm": 2.2268206174043184, "learning_rate": 9.665997458763179e-06, "loss": 0.4964, "step": 14745 }, { "epoch": 0.5253576072821846, "grad_norm": 1.2067279300685911, "learning_rate": 9.664844168126157e-06, "loss": 0.4996, "step": 14746 }, { "epoch": 0.5253932344086788, "grad_norm": 1.117476600527927, "learning_rate": 9.663690881951975e-06, "loss": 0.4488, "step": 14747 }, { "epoch": 0.5254288615351729, "grad_norm": 1.0462169031802961, "learning_rate": 9.662537600255979e-06, "loss": 0.4471, "step": 14748 }, { "epoch": 0.525464488661667, "grad_norm": 1.203576468812109, "learning_rate": 9.661384323053533e-06, "loss": 0.4419, "step": 14749 }, { "epoch": 0.5255001157881611, "grad_norm": 1.2421678902536248, "learning_rate": 9.660231050359989e-06, "loss": 0.4972, "step": 14750 }, { "epoch": 0.5255357429146552, "grad_norm": 1.0352209247064017, "learning_rate": 9.659077782190707e-06, "loss": 0.4502, "step": 14751 }, { "epoch": 0.5255713700411493, "grad_norm": 1.1896070353630703, "learning_rate": 9.657924518561046e-06, "loss": 0.4619, "step": 14752 }, { "epoch": 0.5256069971676435, "grad_norm": 1.2053583613325731, "learning_rate": 9.656771259486352e-06, "loss": 0.4464, "step": 14753 }, { "epoch": 0.5256426242941376, "grad_norm": 1.5413049872227675, "learning_rate": 9.655618004981996e-06, "loss": 0.5666, "step": 14754 }, { "epoch": 0.5256782514206316, "grad_norm": 3.195656579647206, "learning_rate": 9.654464755063323e-06, "loss": 0.4606, "step": 14755 }, { "epoch": 0.5257138785471258, "grad_norm": 1.5247297393335566, "learning_rate": 9.653311509745694e-06, "loss": 0.6428, "step": 14756 }, { "epoch": 0.5257495056736199, "grad_norm": 1.4661592424507397, "learning_rate": 9.652158269044463e-06, "loss": 0.5118, "step": 14757 }, { "epoch": 0.525785132800114, "grad_norm": 1.1893416165763053, "learning_rate": 9.651005032974994e-06, "loss": 0.4871, "step": 14758 }, { "epoch": 0.5258207599266082, "grad_norm": 1.2311050085592767, "learning_rate": 9.64985180155263e-06, "loss": 0.4775, "step": 14759 }, { "epoch": 0.5258563870531022, "grad_norm": 1.2925779485512372, "learning_rate": 9.648698574792737e-06, "loss": 0.4648, "step": 14760 }, { "epoch": 0.5258920141795963, "grad_norm": 1.0705309481047678, "learning_rate": 9.647545352710667e-06, "loss": 0.4387, "step": 14761 }, { "epoch": 0.5259276413060905, "grad_norm": 1.5044936026442959, "learning_rate": 9.646392135321776e-06, "loss": 0.5462, "step": 14762 }, { "epoch": 0.5259632684325846, "grad_norm": 1.1173044846106401, "learning_rate": 9.645238922641425e-06, "loss": 0.468, "step": 14763 }, { "epoch": 0.5259988955590786, "grad_norm": 1.5910637194536443, "learning_rate": 9.64408571468496e-06, "loss": 0.5607, "step": 14764 }, { "epoch": 0.5260345226855728, "grad_norm": 1.9825674605215216, "learning_rate": 9.642932511467748e-06, "loss": 0.5461, "step": 14765 }, { "epoch": 0.5260701498120669, "grad_norm": 1.4658349436802927, "learning_rate": 9.641779313005136e-06, "loss": 0.5538, "step": 14766 }, { "epoch": 0.526105776938561, "grad_norm": 1.1058329227100945, "learning_rate": 9.640626119312483e-06, "loss": 0.4231, "step": 14767 }, { "epoch": 0.5261414040650552, "grad_norm": 1.1800846130070284, "learning_rate": 9.639472930405143e-06, "loss": 0.4722, "step": 14768 }, { "epoch": 0.5261770311915492, "grad_norm": 2.1062674602198252, "learning_rate": 9.638319746298479e-06, "loss": 0.5346, "step": 14769 }, { "epoch": 0.5262126583180433, "grad_norm": 0.9659059505867289, "learning_rate": 9.637166567007834e-06, "loss": 0.4133, "step": 14770 }, { "epoch": 0.5262482854445375, "grad_norm": 1.1654241964803187, "learning_rate": 9.636013392548572e-06, "loss": 0.4953, "step": 14771 }, { "epoch": 0.5262839125710316, "grad_norm": 1.085395534905116, "learning_rate": 9.634860222936047e-06, "loss": 0.4397, "step": 14772 }, { "epoch": 0.5263195396975257, "grad_norm": 1.3591258685331042, "learning_rate": 9.63370705818561e-06, "loss": 0.4831, "step": 14773 }, { "epoch": 0.5263551668240198, "grad_norm": 1.440708797135823, "learning_rate": 9.632553898312625e-06, "loss": 0.4865, "step": 14774 }, { "epoch": 0.5263907939505139, "grad_norm": 1.1845682336523788, "learning_rate": 9.631400743332437e-06, "loss": 0.429, "step": 14775 }, { "epoch": 0.526426421077008, "grad_norm": 1.5199672832334403, "learning_rate": 9.630247593260409e-06, "loss": 0.5072, "step": 14776 }, { "epoch": 0.5264620482035022, "grad_norm": 1.048862749900683, "learning_rate": 9.62909444811189e-06, "loss": 0.4475, "step": 14777 }, { "epoch": 0.5264976753299963, "grad_norm": 1.3462684221938572, "learning_rate": 9.627941307902238e-06, "loss": 0.5392, "step": 14778 }, { "epoch": 0.5265333024564903, "grad_norm": 1.5272881808660224, "learning_rate": 9.626788172646809e-06, "loss": 0.646, "step": 14779 }, { "epoch": 0.5265689295829845, "grad_norm": 1.2945953878246654, "learning_rate": 9.625635042360958e-06, "loss": 0.5549, "step": 14780 }, { "epoch": 0.5266045567094786, "grad_norm": 1.0619168449340504, "learning_rate": 9.624481917060035e-06, "loss": 0.4546, "step": 14781 }, { "epoch": 0.5266401838359727, "grad_norm": 1.6925287190705276, "learning_rate": 9.623328796759399e-06, "loss": 0.5866, "step": 14782 }, { "epoch": 0.5266758109624669, "grad_norm": 1.210194572500989, "learning_rate": 9.6221756814744e-06, "loss": 0.4792, "step": 14783 }, { "epoch": 0.5267114380889609, "grad_norm": 1.0666797674406152, "learning_rate": 9.621022571220404e-06, "loss": 0.4207, "step": 14784 }, { "epoch": 0.526747065215455, "grad_norm": 1.1257549317697617, "learning_rate": 9.61986946601275e-06, "loss": 0.4581, "step": 14785 }, { "epoch": 0.5267826923419492, "grad_norm": 1.1148568118016224, "learning_rate": 9.6187163658668e-06, "loss": 0.4697, "step": 14786 }, { "epoch": 0.5268183194684433, "grad_norm": 1.0908018782610582, "learning_rate": 9.61756327079791e-06, "loss": 0.4745, "step": 14787 }, { "epoch": 0.5268539465949373, "grad_norm": 1.0631737707555975, "learning_rate": 9.616410180821431e-06, "loss": 0.442, "step": 14788 }, { "epoch": 0.5268895737214315, "grad_norm": 1.0839472123791487, "learning_rate": 9.615257095952724e-06, "loss": 0.4648, "step": 14789 }, { "epoch": 0.5269252008479256, "grad_norm": 1.106699812669011, "learning_rate": 9.61410401620713e-06, "loss": 0.4455, "step": 14790 }, { "epoch": 0.5269608279744197, "grad_norm": 1.1242367226017158, "learning_rate": 9.612950941600016e-06, "loss": 0.4398, "step": 14791 }, { "epoch": 0.5269964551009139, "grad_norm": 1.4885831149478705, "learning_rate": 9.611797872146729e-06, "loss": 0.6305, "step": 14792 }, { "epoch": 0.5270320822274079, "grad_norm": 1.1024212643159716, "learning_rate": 9.610644807862625e-06, "loss": 0.463, "step": 14793 }, { "epoch": 0.527067709353902, "grad_norm": 1.8354116006997958, "learning_rate": 9.609491748763055e-06, "loss": 0.5687, "step": 14794 }, { "epoch": 0.5271033364803962, "grad_norm": 1.0876038428608192, "learning_rate": 9.60833869486338e-06, "loss": 0.4279, "step": 14795 }, { "epoch": 0.5271389636068903, "grad_norm": 2.2673834684736565, "learning_rate": 9.607185646178946e-06, "loss": 0.5675, "step": 14796 }, { "epoch": 0.5271745907333844, "grad_norm": 1.4523399552076492, "learning_rate": 9.60603260272511e-06, "loss": 0.5187, "step": 14797 }, { "epoch": 0.5272102178598785, "grad_norm": 1.103183365428499, "learning_rate": 9.604879564517225e-06, "loss": 0.4284, "step": 14798 }, { "epoch": 0.5272458449863726, "grad_norm": 2.357900736777757, "learning_rate": 9.603726531570644e-06, "loss": 0.5551, "step": 14799 }, { "epoch": 0.5272814721128667, "grad_norm": 1.7875057148047928, "learning_rate": 9.602573503900727e-06, "loss": 0.5401, "step": 14800 }, { "epoch": 0.5273170992393609, "grad_norm": 1.0955130139147855, "learning_rate": 9.601420481522813e-06, "loss": 0.442, "step": 14801 }, { "epoch": 0.527352726365855, "grad_norm": 1.0739854976223917, "learning_rate": 9.60026746445227e-06, "loss": 0.4647, "step": 14802 }, { "epoch": 0.527388353492349, "grad_norm": 3.398372200024941, "learning_rate": 9.599114452704442e-06, "loss": 0.4356, "step": 14803 }, { "epoch": 0.5274239806188432, "grad_norm": 1.3659648988477848, "learning_rate": 9.597961446294687e-06, "loss": 0.5106, "step": 14804 }, { "epoch": 0.5274596077453373, "grad_norm": 1.4155159895284326, "learning_rate": 9.596808445238354e-06, "loss": 0.4463, "step": 14805 }, { "epoch": 0.5274952348718314, "grad_norm": 1.7268873464271313, "learning_rate": 9.595655449550804e-06, "loss": 0.5641, "step": 14806 }, { "epoch": 0.5275308619983256, "grad_norm": 1.6144223682752483, "learning_rate": 9.594502459247378e-06, "loss": 0.615, "step": 14807 }, { "epoch": 0.5275664891248196, "grad_norm": 1.7680129826210074, "learning_rate": 9.593349474343437e-06, "loss": 0.6089, "step": 14808 }, { "epoch": 0.5276021162513137, "grad_norm": 1.2337763239769446, "learning_rate": 9.592196494854332e-06, "loss": 0.4826, "step": 14809 }, { "epoch": 0.5276377433778079, "grad_norm": 1.158072590893158, "learning_rate": 9.591043520795415e-06, "loss": 0.4357, "step": 14810 }, { "epoch": 0.527673370504302, "grad_norm": 2.145655056079341, "learning_rate": 9.589890552182044e-06, "loss": 0.648, "step": 14811 }, { "epoch": 0.527708997630796, "grad_norm": 1.4525442638314467, "learning_rate": 9.588737589029562e-06, "loss": 0.5887, "step": 14812 }, { "epoch": 0.5277446247572902, "grad_norm": 1.1851285357728272, "learning_rate": 9.58758463135333e-06, "loss": 0.4694, "step": 14813 }, { "epoch": 0.5277802518837843, "grad_norm": 1.4786931381423738, "learning_rate": 9.586431679168693e-06, "loss": 0.4648, "step": 14814 }, { "epoch": 0.5278158790102784, "grad_norm": 1.4303416742447415, "learning_rate": 9.58527873249101e-06, "loss": 0.5717, "step": 14815 }, { "epoch": 0.5278515061367726, "grad_norm": 1.1028792815959505, "learning_rate": 9.584125791335627e-06, "loss": 0.4587, "step": 14816 }, { "epoch": 0.5278871332632666, "grad_norm": 1.3209190460598699, "learning_rate": 9.582972855717906e-06, "loss": 0.5505, "step": 14817 }, { "epoch": 0.5279227603897607, "grad_norm": 1.4968590296455695, "learning_rate": 9.581819925653189e-06, "loss": 0.4987, "step": 14818 }, { "epoch": 0.5279583875162549, "grad_norm": 1.2788318055436365, "learning_rate": 9.580667001156833e-06, "loss": 0.5559, "step": 14819 }, { "epoch": 0.527994014642749, "grad_norm": 1.256349787611076, "learning_rate": 9.579514082244186e-06, "loss": 0.4616, "step": 14820 }, { "epoch": 0.528029641769243, "grad_norm": 1.5962929147454903, "learning_rate": 9.578361168930604e-06, "loss": 0.5579, "step": 14821 }, { "epoch": 0.5280652688957372, "grad_norm": 1.5090783445683473, "learning_rate": 9.577208261231441e-06, "loss": 0.6158, "step": 14822 }, { "epoch": 0.5281008960222313, "grad_norm": 1.3350690567824002, "learning_rate": 9.576055359162043e-06, "loss": 0.4978, "step": 14823 }, { "epoch": 0.5281365231487254, "grad_norm": 1.2945456661336285, "learning_rate": 9.574902462737766e-06, "loss": 0.5197, "step": 14824 }, { "epoch": 0.5281721502752196, "grad_norm": 1.5394060146961235, "learning_rate": 9.573749571973956e-06, "loss": 0.6156, "step": 14825 }, { "epoch": 0.5282077774017137, "grad_norm": 1.2411693338349252, "learning_rate": 9.572596686885974e-06, "loss": 0.4569, "step": 14826 }, { "epoch": 0.5282434045282077, "grad_norm": 1.2061647756452767, "learning_rate": 9.571443807489162e-06, "loss": 0.4768, "step": 14827 }, { "epoch": 0.5282790316547019, "grad_norm": 1.294362856974811, "learning_rate": 9.570290933798877e-06, "loss": 0.4489, "step": 14828 }, { "epoch": 0.528314658781196, "grad_norm": 1.292487992964239, "learning_rate": 9.569138065830466e-06, "loss": 0.4823, "step": 14829 }, { "epoch": 0.5283502859076901, "grad_norm": 1.7737863829868723, "learning_rate": 9.567985203599285e-06, "loss": 0.5066, "step": 14830 }, { "epoch": 0.5283859130341843, "grad_norm": 1.1335902756137937, "learning_rate": 9.566832347120682e-06, "loss": 0.482, "step": 14831 }, { "epoch": 0.5284215401606783, "grad_norm": 2.352350581196811, "learning_rate": 9.565679496410008e-06, "loss": 0.57, "step": 14832 }, { "epoch": 0.5284571672871724, "grad_norm": 1.1026260752861987, "learning_rate": 9.564526651482619e-06, "loss": 0.4284, "step": 14833 }, { "epoch": 0.5284927944136666, "grad_norm": 1.2219854278791127, "learning_rate": 9.563373812353858e-06, "loss": 0.4676, "step": 14834 }, { "epoch": 0.5285284215401607, "grad_norm": 1.417477615482422, "learning_rate": 9.562220979039083e-06, "loss": 0.4861, "step": 14835 }, { "epoch": 0.5285640486666547, "grad_norm": 1.2012929648819706, "learning_rate": 9.561068151553637e-06, "loss": 0.5008, "step": 14836 }, { "epoch": 0.5285996757931489, "grad_norm": 1.112001855027694, "learning_rate": 9.559915329912883e-06, "loss": 0.4619, "step": 14837 }, { "epoch": 0.528635302919643, "grad_norm": 1.4310771011000947, "learning_rate": 9.558762514132157e-06, "loss": 0.4395, "step": 14838 }, { "epoch": 0.5286709300461371, "grad_norm": 1.2162369880830683, "learning_rate": 9.55760970422682e-06, "loss": 0.455, "step": 14839 }, { "epoch": 0.5287065571726313, "grad_norm": 1.5995079352185078, "learning_rate": 9.556456900212217e-06, "loss": 0.5928, "step": 14840 }, { "epoch": 0.5287421842991253, "grad_norm": 1.334233367400602, "learning_rate": 9.5553041021037e-06, "loss": 0.5778, "step": 14841 }, { "epoch": 0.5287778114256194, "grad_norm": 1.1218507930700605, "learning_rate": 9.554151309916621e-06, "loss": 0.4757, "step": 14842 }, { "epoch": 0.5288134385521136, "grad_norm": 1.3870018671405469, "learning_rate": 9.552998523666327e-06, "loss": 0.5566, "step": 14843 }, { "epoch": 0.5288490656786077, "grad_norm": 1.1048672095821113, "learning_rate": 9.551845743368175e-06, "loss": 0.4352, "step": 14844 }, { "epoch": 0.5288846928051018, "grad_norm": 1.038659392989875, "learning_rate": 9.550692969037506e-06, "loss": 0.4115, "step": 14845 }, { "epoch": 0.5289203199315959, "grad_norm": 1.3535045976686901, "learning_rate": 9.549540200689676e-06, "loss": 0.6156, "step": 14846 }, { "epoch": 0.52895594705809, "grad_norm": 2.168187618413594, "learning_rate": 9.54838743834003e-06, "loss": 0.6432, "step": 14847 }, { "epoch": 0.5289915741845841, "grad_norm": 1.2383221497930423, "learning_rate": 9.547234682003928e-06, "loss": 0.5379, "step": 14848 }, { "epoch": 0.5290272013110783, "grad_norm": 0.9800850838980929, "learning_rate": 9.546081931696707e-06, "loss": 0.4345, "step": 14849 }, { "epoch": 0.5290628284375724, "grad_norm": 1.6775351075868146, "learning_rate": 9.544929187433724e-06, "loss": 0.5597, "step": 14850 }, { "epoch": 0.5290984555640664, "grad_norm": 1.0921292306020856, "learning_rate": 9.543776449230325e-06, "loss": 0.4067, "step": 14851 }, { "epoch": 0.5291340826905606, "grad_norm": 1.5957363826759874, "learning_rate": 9.542623717101866e-06, "loss": 0.5828, "step": 14852 }, { "epoch": 0.5291697098170547, "grad_norm": 1.673482369142735, "learning_rate": 9.541470991063685e-06, "loss": 0.5321, "step": 14853 }, { "epoch": 0.5292053369435488, "grad_norm": 1.0755429025239178, "learning_rate": 9.54031827113114e-06, "loss": 0.454, "step": 14854 }, { "epoch": 0.529240964070043, "grad_norm": 1.648670199363812, "learning_rate": 9.539165557319583e-06, "loss": 0.6499, "step": 14855 }, { "epoch": 0.529276591196537, "grad_norm": 1.0690442417733257, "learning_rate": 9.538012849644355e-06, "loss": 0.425, "step": 14856 }, { "epoch": 0.5293122183230311, "grad_norm": 0.9450354955387466, "learning_rate": 9.536860148120811e-06, "loss": 0.3991, "step": 14857 }, { "epoch": 0.5293478454495253, "grad_norm": 1.1250678997379382, "learning_rate": 9.535707452764296e-06, "loss": 0.4765, "step": 14858 }, { "epoch": 0.5293834725760194, "grad_norm": 1.2231131599486373, "learning_rate": 9.534554763590164e-06, "loss": 0.4208, "step": 14859 }, { "epoch": 0.5294190997025134, "grad_norm": 1.8449461388172623, "learning_rate": 9.533402080613757e-06, "loss": 0.6247, "step": 14860 }, { "epoch": 0.5294547268290076, "grad_norm": 1.2404368434939628, "learning_rate": 9.53224940385043e-06, "loss": 0.4368, "step": 14861 }, { "epoch": 0.5294903539555017, "grad_norm": 1.1490020288466465, "learning_rate": 9.531096733315527e-06, "loss": 0.4824, "step": 14862 }, { "epoch": 0.5295259810819958, "grad_norm": 1.535832380914519, "learning_rate": 9.529944069024402e-06, "loss": 0.5581, "step": 14863 }, { "epoch": 0.52956160820849, "grad_norm": 1.6699569417129698, "learning_rate": 9.528791410992395e-06, "loss": 0.5419, "step": 14864 }, { "epoch": 0.529597235334984, "grad_norm": 1.1712018336669465, "learning_rate": 9.527638759234864e-06, "loss": 0.4563, "step": 14865 }, { "epoch": 0.5296328624614781, "grad_norm": 1.6441481165929517, "learning_rate": 9.526486113767151e-06, "loss": 0.5524, "step": 14866 }, { "epoch": 0.5296684895879723, "grad_norm": 1.0789847496675569, "learning_rate": 9.525333474604604e-06, "loss": 0.4715, "step": 14867 }, { "epoch": 0.5297041167144664, "grad_norm": 1.0768192628223587, "learning_rate": 9.524180841762577e-06, "loss": 0.4504, "step": 14868 }, { "epoch": 0.5297397438409605, "grad_norm": 1.2147887261991244, "learning_rate": 9.523028215256413e-06, "loss": 0.4799, "step": 14869 }, { "epoch": 0.5297753709674546, "grad_norm": 1.15199135952744, "learning_rate": 9.521875595101467e-06, "loss": 0.4447, "step": 14870 }, { "epoch": 0.5298109980939487, "grad_norm": 1.0931721454650891, "learning_rate": 9.520722981313076e-06, "loss": 0.4449, "step": 14871 }, { "epoch": 0.5298466252204428, "grad_norm": 0.9861046771415327, "learning_rate": 9.519570373906597e-06, "loss": 0.4213, "step": 14872 }, { "epoch": 0.529882252346937, "grad_norm": 1.6353035308198283, "learning_rate": 9.51841777289737e-06, "loss": 0.538, "step": 14873 }, { "epoch": 0.529917879473431, "grad_norm": 1.224849328575831, "learning_rate": 9.517265178300753e-06, "loss": 0.4974, "step": 14874 }, { "epoch": 0.5299535065999251, "grad_norm": 1.8317077402638546, "learning_rate": 9.516112590132082e-06, "loss": 0.5639, "step": 14875 }, { "epoch": 0.5299891337264193, "grad_norm": 1.032720047135473, "learning_rate": 9.514960008406712e-06, "loss": 0.4146, "step": 14876 }, { "epoch": 0.5300247608529134, "grad_norm": 1.1466225971611967, "learning_rate": 9.513807433139987e-06, "loss": 0.4693, "step": 14877 }, { "epoch": 0.5300603879794075, "grad_norm": 1.7748163896720595, "learning_rate": 9.512654864347257e-06, "loss": 0.4714, "step": 14878 }, { "epoch": 0.5300960151059017, "grad_norm": 1.608584801245785, "learning_rate": 9.511502302043867e-06, "loss": 0.448, "step": 14879 }, { "epoch": 0.5301316422323957, "grad_norm": 2.0117843269627964, "learning_rate": 9.510349746245166e-06, "loss": 0.5258, "step": 14880 }, { "epoch": 0.5301672693588898, "grad_norm": 1.4232388045063542, "learning_rate": 9.509197196966504e-06, "loss": 0.5443, "step": 14881 }, { "epoch": 0.530202896485384, "grad_norm": 1.1624826462480693, "learning_rate": 9.50804465422322e-06, "loss": 0.4501, "step": 14882 }, { "epoch": 0.5302385236118781, "grad_norm": 1.5791805907664795, "learning_rate": 9.506892118030668e-06, "loss": 0.5394, "step": 14883 }, { "epoch": 0.5302741507383723, "grad_norm": 1.038783170648464, "learning_rate": 9.50573958840419e-06, "loss": 0.4346, "step": 14884 }, { "epoch": 0.5303097778648663, "grad_norm": 1.3033435601858494, "learning_rate": 9.504587065359137e-06, "loss": 0.4865, "step": 14885 }, { "epoch": 0.5303454049913604, "grad_norm": 1.1583716076589248, "learning_rate": 9.50343454891085e-06, "loss": 0.4168, "step": 14886 }, { "epoch": 0.5303810321178546, "grad_norm": 1.664762261268676, "learning_rate": 9.502282039074683e-06, "loss": 0.5167, "step": 14887 }, { "epoch": 0.5304166592443487, "grad_norm": 1.0910701687154898, "learning_rate": 9.501129535865976e-06, "loss": 0.4331, "step": 14888 }, { "epoch": 0.5304522863708427, "grad_norm": 0.97591435421113, "learning_rate": 9.499977039300078e-06, "loss": 0.4079, "step": 14889 }, { "epoch": 0.5304879134973369, "grad_norm": 1.0841662788055482, "learning_rate": 9.498824549392338e-06, "loss": 0.4534, "step": 14890 }, { "epoch": 0.530523540623831, "grad_norm": 1.7049188497607046, "learning_rate": 9.497672066158095e-06, "loss": 0.5587, "step": 14891 }, { "epoch": 0.5305591677503251, "grad_norm": 1.5177558154305673, "learning_rate": 9.496519589612704e-06, "loss": 0.4878, "step": 14892 }, { "epoch": 0.5305947948768193, "grad_norm": 1.2990783081706336, "learning_rate": 9.495367119771504e-06, "loss": 0.4581, "step": 14893 }, { "epoch": 0.5306304220033133, "grad_norm": 1.277813526310328, "learning_rate": 9.494214656649844e-06, "loss": 0.4461, "step": 14894 }, { "epoch": 0.5306660491298074, "grad_norm": 1.2217496863946589, "learning_rate": 9.493062200263068e-06, "loss": 0.4804, "step": 14895 }, { "epoch": 0.5307016762563016, "grad_norm": 1.6787843053524003, "learning_rate": 9.491909750626526e-06, "loss": 0.5367, "step": 14896 }, { "epoch": 0.5307373033827957, "grad_norm": 2.42563223679711, "learning_rate": 9.490757307755559e-06, "loss": 0.6115, "step": 14897 }, { "epoch": 0.5307729305092898, "grad_norm": 0.9758182158520253, "learning_rate": 9.489604871665515e-06, "loss": 0.419, "step": 14898 }, { "epoch": 0.5308085576357839, "grad_norm": 1.0004706150030356, "learning_rate": 9.488452442371736e-06, "loss": 0.4559, "step": 14899 }, { "epoch": 0.530844184762278, "grad_norm": 1.4071435935253211, "learning_rate": 9.487300019889573e-06, "loss": 0.5372, "step": 14900 }, { "epoch": 0.5308798118887721, "grad_norm": 1.401305740623149, "learning_rate": 9.48614760423437e-06, "loss": 0.5701, "step": 14901 }, { "epoch": 0.5309154390152663, "grad_norm": 1.0727178050012338, "learning_rate": 9.48499519542147e-06, "loss": 0.4291, "step": 14902 }, { "epoch": 0.5309510661417604, "grad_norm": 1.3234904474829547, "learning_rate": 9.483842793466218e-06, "loss": 0.5385, "step": 14903 }, { "epoch": 0.5309866932682544, "grad_norm": 1.2333360205427852, "learning_rate": 9.482690398383958e-06, "loss": 0.4685, "step": 14904 }, { "epoch": 0.5310223203947486, "grad_norm": 1.3076504760157635, "learning_rate": 9.481538010190041e-06, "loss": 0.4574, "step": 14905 }, { "epoch": 0.5310579475212427, "grad_norm": 1.0301685735812014, "learning_rate": 9.480385628899804e-06, "loss": 0.4064, "step": 14906 }, { "epoch": 0.5310935746477368, "grad_norm": 1.430831297407381, "learning_rate": 9.4792332545286e-06, "loss": 0.4962, "step": 14907 }, { "epoch": 0.531129201774231, "grad_norm": 1.200637001126011, "learning_rate": 9.478080887091766e-06, "loss": 0.4615, "step": 14908 }, { "epoch": 0.531164828900725, "grad_norm": 1.1290723431850442, "learning_rate": 9.476928526604651e-06, "loss": 0.4163, "step": 14909 }, { "epoch": 0.5312004560272191, "grad_norm": 1.9559481239516956, "learning_rate": 9.475776173082597e-06, "loss": 0.6045, "step": 14910 }, { "epoch": 0.5312360831537133, "grad_norm": 1.2430274229300253, "learning_rate": 9.474623826540952e-06, "loss": 0.4654, "step": 14911 }, { "epoch": 0.5312717102802074, "grad_norm": 1.14022064568394, "learning_rate": 9.473471486995056e-06, "loss": 0.4601, "step": 14912 }, { "epoch": 0.5313073374067014, "grad_norm": 2.1954132072076047, "learning_rate": 9.472319154460254e-06, "loss": 0.5387, "step": 14913 }, { "epoch": 0.5313429645331956, "grad_norm": 1.3914098408053568, "learning_rate": 9.471166828951893e-06, "loss": 0.5912, "step": 14914 }, { "epoch": 0.5313785916596897, "grad_norm": 1.14202263968434, "learning_rate": 9.470014510485313e-06, "loss": 0.4562, "step": 14915 }, { "epoch": 0.5314142187861838, "grad_norm": 1.0866599441312457, "learning_rate": 9.468862199075866e-06, "loss": 0.4632, "step": 14916 }, { "epoch": 0.531449845912678, "grad_norm": 1.0530734035951954, "learning_rate": 9.467709894738882e-06, "loss": 0.4402, "step": 14917 }, { "epoch": 0.531485473039172, "grad_norm": 1.0692461729026808, "learning_rate": 9.46655759748972e-06, "loss": 0.445, "step": 14918 }, { "epoch": 0.5315211001656661, "grad_norm": 1.3484534211900288, "learning_rate": 9.465405307343714e-06, "loss": 0.5834, "step": 14919 }, { "epoch": 0.5315567272921603, "grad_norm": 1.6200017095273032, "learning_rate": 9.46425302431621e-06, "loss": 0.4754, "step": 14920 }, { "epoch": 0.5315923544186544, "grad_norm": 1.3087145673974583, "learning_rate": 9.46310074842255e-06, "loss": 0.4737, "step": 14921 }, { "epoch": 0.5316279815451485, "grad_norm": 1.5319931381491103, "learning_rate": 9.461948479678083e-06, "loss": 0.5507, "step": 14922 }, { "epoch": 0.5316636086716426, "grad_norm": 1.2996593239993826, "learning_rate": 9.460796218098143e-06, "loss": 0.4717, "step": 14923 }, { "epoch": 0.5316992357981367, "grad_norm": 1.3326938487553721, "learning_rate": 9.45964396369808e-06, "loss": 0.5178, "step": 14924 }, { "epoch": 0.5317348629246308, "grad_norm": 1.0475170541835601, "learning_rate": 9.458491716493235e-06, "loss": 0.4231, "step": 14925 }, { "epoch": 0.531770490051125, "grad_norm": 1.098085487008458, "learning_rate": 9.45733947649895e-06, "loss": 0.4895, "step": 14926 }, { "epoch": 0.531806117177619, "grad_norm": 1.1236405209403988, "learning_rate": 9.456187243730573e-06, "loss": 0.4464, "step": 14927 }, { "epoch": 0.5318417443041131, "grad_norm": 1.2065177079468086, "learning_rate": 9.455035018203439e-06, "loss": 0.435, "step": 14928 }, { "epoch": 0.5318773714306073, "grad_norm": 1.0769111801029152, "learning_rate": 9.4538827999329e-06, "loss": 0.4265, "step": 14929 }, { "epoch": 0.5319129985571014, "grad_norm": 1.8773981582895187, "learning_rate": 9.452730588934287e-06, "loss": 0.5375, "step": 14930 }, { "epoch": 0.5319486256835955, "grad_norm": 1.1865350529846685, "learning_rate": 9.451578385222952e-06, "loss": 0.4432, "step": 14931 }, { "epoch": 0.5319842528100897, "grad_norm": 1.0367506401885491, "learning_rate": 9.450426188814233e-06, "loss": 0.4086, "step": 14932 }, { "epoch": 0.5320198799365837, "grad_norm": 1.66739069320828, "learning_rate": 9.449273999723476e-06, "loss": 0.6172, "step": 14933 }, { "epoch": 0.5320555070630778, "grad_norm": 1.2910029534616345, "learning_rate": 9.448121817966017e-06, "loss": 0.4966, "step": 14934 }, { "epoch": 0.532091134189572, "grad_norm": 1.5837298759309149, "learning_rate": 9.446969643557202e-06, "loss": 0.5222, "step": 14935 }, { "epoch": 0.5321267613160661, "grad_norm": 1.7131169281422003, "learning_rate": 9.445817476512376e-06, "loss": 0.5704, "step": 14936 }, { "epoch": 0.5321623884425601, "grad_norm": 1.1433838466591244, "learning_rate": 9.444665316846874e-06, "loss": 0.4239, "step": 14937 }, { "epoch": 0.5321980155690543, "grad_norm": 2.1306787628368404, "learning_rate": 9.443513164576046e-06, "loss": 0.4116, "step": 14938 }, { "epoch": 0.5322336426955484, "grad_norm": 1.3201406307169885, "learning_rate": 9.442361019715225e-06, "loss": 0.6154, "step": 14939 }, { "epoch": 0.5322692698220425, "grad_norm": 1.3639632831720312, "learning_rate": 9.441208882279759e-06, "loss": 0.5402, "step": 14940 }, { "epoch": 0.5323048969485367, "grad_norm": 1.2125751537042864, "learning_rate": 9.440056752284985e-06, "loss": 0.4286, "step": 14941 }, { "epoch": 0.5323405240750307, "grad_norm": 1.512044438109926, "learning_rate": 9.438904629746248e-06, "loss": 0.5674, "step": 14942 }, { "epoch": 0.5323761512015248, "grad_norm": 1.3446113329717126, "learning_rate": 9.437752514678888e-06, "loss": 0.5332, "step": 14943 }, { "epoch": 0.532411778328019, "grad_norm": 1.4652725650156695, "learning_rate": 9.436600407098248e-06, "loss": 0.5647, "step": 14944 }, { "epoch": 0.5324474054545131, "grad_norm": 1.1989185741880202, "learning_rate": 9.435448307019666e-06, "loss": 0.467, "step": 14945 }, { "epoch": 0.5324830325810072, "grad_norm": 1.2259791150892054, "learning_rate": 9.434296214458482e-06, "loss": 0.4741, "step": 14946 }, { "epoch": 0.5325186597075013, "grad_norm": 1.0620452932425992, "learning_rate": 9.433144129430043e-06, "loss": 0.4324, "step": 14947 }, { "epoch": 0.5325542868339954, "grad_norm": 1.5248023202498, "learning_rate": 9.431992051949683e-06, "loss": 0.5973, "step": 14948 }, { "epoch": 0.5325899139604895, "grad_norm": 1.4721601250345566, "learning_rate": 9.43083998203275e-06, "loss": 0.5645, "step": 14949 }, { "epoch": 0.5326255410869837, "grad_norm": 1.3827286685069538, "learning_rate": 9.429687919694576e-06, "loss": 0.6191, "step": 14950 }, { "epoch": 0.5326611682134778, "grad_norm": 1.1591276889823772, "learning_rate": 9.42853586495051e-06, "loss": 0.4643, "step": 14951 }, { "epoch": 0.5326967953399718, "grad_norm": 1.5838392662497796, "learning_rate": 9.427383817815885e-06, "loss": 0.5611, "step": 14952 }, { "epoch": 0.532732422466466, "grad_norm": 1.1676100083605805, "learning_rate": 9.426231778306049e-06, "loss": 0.4078, "step": 14953 }, { "epoch": 0.5327680495929601, "grad_norm": 1.2030753813702941, "learning_rate": 9.425079746436331e-06, "loss": 0.4754, "step": 14954 }, { "epoch": 0.5328036767194542, "grad_norm": 1.0958445811861863, "learning_rate": 9.423927722222085e-06, "loss": 0.4474, "step": 14955 }, { "epoch": 0.5328393038459484, "grad_norm": 1.4148358070905562, "learning_rate": 9.42277570567864e-06, "loss": 0.5569, "step": 14956 }, { "epoch": 0.5328749309724424, "grad_norm": 1.1335410221997164, "learning_rate": 9.421623696821342e-06, "loss": 0.468, "step": 14957 }, { "epoch": 0.5329105580989365, "grad_norm": 1.5870823086847008, "learning_rate": 9.420471695665527e-06, "loss": 0.6499, "step": 14958 }, { "epoch": 0.5329461852254307, "grad_norm": 1.338496268807368, "learning_rate": 9.419319702226534e-06, "loss": 0.5833, "step": 14959 }, { "epoch": 0.5329818123519248, "grad_norm": 1.156009860092755, "learning_rate": 9.418167716519712e-06, "loss": 0.4522, "step": 14960 }, { "epoch": 0.5330174394784188, "grad_norm": 1.0928367248408701, "learning_rate": 9.417015738560388e-06, "loss": 0.4227, "step": 14961 }, { "epoch": 0.533053066604913, "grad_norm": 1.54982627404831, "learning_rate": 9.41586376836391e-06, "loss": 0.5825, "step": 14962 }, { "epoch": 0.5330886937314071, "grad_norm": 1.1169676648129214, "learning_rate": 9.414711805945612e-06, "loss": 0.4249, "step": 14963 }, { "epoch": 0.5331243208579012, "grad_norm": 1.5233563137565358, "learning_rate": 9.413559851320838e-06, "loss": 0.514, "step": 14964 }, { "epoch": 0.5331599479843954, "grad_norm": 1.003607823956063, "learning_rate": 9.412407904504921e-06, "loss": 0.4579, "step": 14965 }, { "epoch": 0.5331955751108894, "grad_norm": 1.2296301949983648, "learning_rate": 9.411255965513206e-06, "loss": 0.4659, "step": 14966 }, { "epoch": 0.5332312022373835, "grad_norm": 1.4337971547679786, "learning_rate": 9.410104034361027e-06, "loss": 0.4979, "step": 14967 }, { "epoch": 0.5332668293638777, "grad_norm": 1.4556068435601184, "learning_rate": 9.408952111063727e-06, "loss": 0.4824, "step": 14968 }, { "epoch": 0.5333024564903718, "grad_norm": 1.09167409429018, "learning_rate": 9.407800195636642e-06, "loss": 0.4504, "step": 14969 }, { "epoch": 0.5333380836168659, "grad_norm": 1.1838101288618739, "learning_rate": 9.40664828809511e-06, "loss": 0.4672, "step": 14970 }, { "epoch": 0.53337371074336, "grad_norm": 1.2564304905227617, "learning_rate": 9.405496388454476e-06, "loss": 0.4399, "step": 14971 }, { "epoch": 0.5334093378698541, "grad_norm": 1.0445560391933846, "learning_rate": 9.404344496730067e-06, "loss": 0.4318, "step": 14972 }, { "epoch": 0.5334449649963482, "grad_norm": 1.213726409497286, "learning_rate": 9.40319261293723e-06, "loss": 0.4526, "step": 14973 }, { "epoch": 0.5334805921228424, "grad_norm": 1.1907543596355465, "learning_rate": 9.4020407370913e-06, "loss": 0.4519, "step": 14974 }, { "epoch": 0.5335162192493365, "grad_norm": 1.9485040463067218, "learning_rate": 9.40088886920762e-06, "loss": 0.6566, "step": 14975 }, { "epoch": 0.5335518463758305, "grad_norm": 1.454380197912399, "learning_rate": 9.399737009301516e-06, "loss": 0.542, "step": 14976 }, { "epoch": 0.5335874735023247, "grad_norm": 1.4937555450084021, "learning_rate": 9.398585157388338e-06, "loss": 0.5604, "step": 14977 }, { "epoch": 0.5336231006288188, "grad_norm": 1.159975877135793, "learning_rate": 9.397433313483417e-06, "loss": 0.4338, "step": 14978 }, { "epoch": 0.5336587277553129, "grad_norm": 1.417959826958728, "learning_rate": 9.396281477602095e-06, "loss": 0.4949, "step": 14979 }, { "epoch": 0.533694354881807, "grad_norm": 1.1376909046997885, "learning_rate": 9.395129649759703e-06, "loss": 0.4886, "step": 14980 }, { "epoch": 0.5337299820083011, "grad_norm": 1.094338455901277, "learning_rate": 9.393977829971583e-06, "loss": 0.4889, "step": 14981 }, { "epoch": 0.5337656091347952, "grad_norm": 1.2188514436775326, "learning_rate": 9.392826018253078e-06, "loss": 0.4188, "step": 14982 }, { "epoch": 0.5338012362612894, "grad_norm": 0.9840961712320844, "learning_rate": 9.391674214619511e-06, "loss": 0.4068, "step": 14983 }, { "epoch": 0.5338368633877835, "grad_norm": 1.7284142938935443, "learning_rate": 9.390522419086232e-06, "loss": 0.5607, "step": 14984 }, { "epoch": 0.5338724905142775, "grad_norm": 1.0481805186024526, "learning_rate": 9.389370631668571e-06, "loss": 0.4198, "step": 14985 }, { "epoch": 0.5339081176407717, "grad_norm": 1.5701020933144407, "learning_rate": 9.38821885238187e-06, "loss": 0.589, "step": 14986 }, { "epoch": 0.5339437447672658, "grad_norm": 1.2593208299559233, "learning_rate": 9.38706708124146e-06, "loss": 0.4094, "step": 14987 }, { "epoch": 0.5339793718937599, "grad_norm": 1.2658929690535627, "learning_rate": 9.38591531826268e-06, "loss": 0.4941, "step": 14988 }, { "epoch": 0.5340149990202541, "grad_norm": 1.5579521608541707, "learning_rate": 9.384763563460866e-06, "loss": 0.5861, "step": 14989 }, { "epoch": 0.5340506261467481, "grad_norm": 1.1597698347424272, "learning_rate": 9.383611816851359e-06, "loss": 0.4259, "step": 14990 }, { "epoch": 0.5340862532732422, "grad_norm": 1.1595657057720292, "learning_rate": 9.382460078449487e-06, "loss": 0.4365, "step": 14991 }, { "epoch": 0.5341218803997364, "grad_norm": 1.2106137552300975, "learning_rate": 9.381308348270593e-06, "loss": 0.4316, "step": 14992 }, { "epoch": 0.5341575075262305, "grad_norm": 1.1666270588203165, "learning_rate": 9.380156626330009e-06, "loss": 0.4215, "step": 14993 }, { "epoch": 0.5341931346527246, "grad_norm": 1.3071292315832288, "learning_rate": 9.379004912643073e-06, "loss": 0.5238, "step": 14994 }, { "epoch": 0.5342287617792187, "grad_norm": 1.3696628764154717, "learning_rate": 9.377853207225122e-06, "loss": 0.4918, "step": 14995 }, { "epoch": 0.5342643889057128, "grad_norm": 1.053585646142613, "learning_rate": 9.376701510091488e-06, "loss": 0.4438, "step": 14996 }, { "epoch": 0.5343000160322069, "grad_norm": 1.1575993753976321, "learning_rate": 9.375549821257515e-06, "loss": 0.4229, "step": 14997 }, { "epoch": 0.5343356431587011, "grad_norm": 1.7346146402915612, "learning_rate": 9.374398140738526e-06, "loss": 0.575, "step": 14998 }, { "epoch": 0.5343712702851952, "grad_norm": 1.4242544883654424, "learning_rate": 9.373246468549867e-06, "loss": 0.5774, "step": 14999 }, { "epoch": 0.5344068974116892, "grad_norm": 1.4605962641514143, "learning_rate": 9.372094804706867e-06, "loss": 0.5713, "step": 15000 }, { "epoch": 0.5344425245381834, "grad_norm": 1.6774597980488561, "learning_rate": 9.370943149224867e-06, "loss": 0.4896, "step": 15001 }, { "epoch": 0.5344781516646775, "grad_norm": 1.6031293430202787, "learning_rate": 9.369791502119195e-06, "loss": 0.5281, "step": 15002 }, { "epoch": 0.5345137787911716, "grad_norm": 1.283403717664779, "learning_rate": 9.368639863405192e-06, "loss": 0.4624, "step": 15003 }, { "epoch": 0.5345494059176658, "grad_norm": 1.4739170948848679, "learning_rate": 9.367488233098187e-06, "loss": 0.5175, "step": 15004 }, { "epoch": 0.5345850330441598, "grad_norm": 1.368082973486437, "learning_rate": 9.36633661121352e-06, "loss": 0.6142, "step": 15005 }, { "epoch": 0.5346206601706539, "grad_norm": 1.187961752299765, "learning_rate": 9.365184997766525e-06, "loss": 0.4791, "step": 15006 }, { "epoch": 0.5346562872971481, "grad_norm": 1.0502836116252152, "learning_rate": 9.364033392772534e-06, "loss": 0.4248, "step": 15007 }, { "epoch": 0.5346919144236422, "grad_norm": 1.1875492198424686, "learning_rate": 9.362881796246886e-06, "loss": 0.4607, "step": 15008 }, { "epoch": 0.5347275415501362, "grad_norm": 1.0049621292968607, "learning_rate": 9.36173020820491e-06, "loss": 0.4313, "step": 15009 }, { "epoch": 0.5347631686766304, "grad_norm": 1.1225779766754393, "learning_rate": 9.360578628661942e-06, "loss": 0.4296, "step": 15010 }, { "epoch": 0.5347987958031245, "grad_norm": 1.1590719808368208, "learning_rate": 9.359427057633317e-06, "loss": 0.449, "step": 15011 }, { "epoch": 0.5348344229296186, "grad_norm": 1.5009588005629653, "learning_rate": 9.358275495134371e-06, "loss": 0.644, "step": 15012 }, { "epoch": 0.5348700500561128, "grad_norm": 1.081977996315348, "learning_rate": 9.35712394118043e-06, "loss": 0.4884, "step": 15013 }, { "epoch": 0.5349056771826068, "grad_norm": 1.4825774309314426, "learning_rate": 9.355972395786837e-06, "loss": 0.4528, "step": 15014 }, { "epoch": 0.5349413043091009, "grad_norm": 1.0588543732102509, "learning_rate": 9.35482085896892e-06, "loss": 0.4068, "step": 15015 }, { "epoch": 0.5349769314355951, "grad_norm": 1.1322672700960985, "learning_rate": 9.353669330742014e-06, "loss": 0.5056, "step": 15016 }, { "epoch": 0.5350125585620892, "grad_norm": 1.1542590027889212, "learning_rate": 9.352517811121453e-06, "loss": 0.4165, "step": 15017 }, { "epoch": 0.5350481856885833, "grad_norm": 1.6621713277432262, "learning_rate": 9.351366300122569e-06, "loss": 0.5387, "step": 15018 }, { "epoch": 0.5350838128150774, "grad_norm": 1.1676802675623925, "learning_rate": 9.350214797760701e-06, "loss": 0.4599, "step": 15019 }, { "epoch": 0.5351194399415715, "grad_norm": 2.6173442174131534, "learning_rate": 9.34906330405117e-06, "loss": 0.5062, "step": 15020 }, { "epoch": 0.5351550670680656, "grad_norm": 1.6664065615101056, "learning_rate": 9.347911819009322e-06, "loss": 0.5097, "step": 15021 }, { "epoch": 0.5351906941945598, "grad_norm": 1.4634718014730574, "learning_rate": 9.346760342650481e-06, "loss": 0.5424, "step": 15022 }, { "epoch": 0.5352263213210539, "grad_norm": 1.2454698253613556, "learning_rate": 9.345608874989986e-06, "loss": 0.544, "step": 15023 }, { "epoch": 0.5352619484475479, "grad_norm": 1.4235335107002876, "learning_rate": 9.34445741604316e-06, "loss": 0.568, "step": 15024 }, { "epoch": 0.5352975755740421, "grad_norm": 1.1287551436277967, "learning_rate": 9.343305965825348e-06, "loss": 0.4559, "step": 15025 }, { "epoch": 0.5353332027005362, "grad_norm": 1.0746354482215599, "learning_rate": 9.342154524351871e-06, "loss": 0.4181, "step": 15026 }, { "epoch": 0.5353688298270303, "grad_norm": 1.1097718082691292, "learning_rate": 9.341003091638066e-06, "loss": 0.4566, "step": 15027 }, { "epoch": 0.5354044569535245, "grad_norm": 1.4819790672879851, "learning_rate": 9.339851667699272e-06, "loss": 0.5444, "step": 15028 }, { "epoch": 0.5354400840800185, "grad_norm": 1.451631635692948, "learning_rate": 9.338700252550806e-06, "loss": 0.4822, "step": 15029 }, { "epoch": 0.5354757112065126, "grad_norm": 1.6113881394235814, "learning_rate": 9.337548846208016e-06, "loss": 0.6217, "step": 15030 }, { "epoch": 0.5355113383330068, "grad_norm": 1.4709980884093474, "learning_rate": 9.336397448686221e-06, "loss": 0.5229, "step": 15031 }, { "epoch": 0.5355469654595009, "grad_norm": 1.157436918221467, "learning_rate": 9.33524606000076e-06, "loss": 0.4202, "step": 15032 }, { "epoch": 0.5355825925859949, "grad_norm": 1.112923722942055, "learning_rate": 9.334094680166962e-06, "loss": 0.4377, "step": 15033 }, { "epoch": 0.5356182197124891, "grad_norm": 2.0457499527885394, "learning_rate": 9.332943309200163e-06, "loss": 0.6285, "step": 15034 }, { "epoch": 0.5356538468389832, "grad_norm": 1.1623198187450476, "learning_rate": 9.331791947115683e-06, "loss": 0.4687, "step": 15035 }, { "epoch": 0.5356894739654773, "grad_norm": 1.4031748090809926, "learning_rate": 9.330640593928863e-06, "loss": 0.4522, "step": 15036 }, { "epoch": 0.5357251010919715, "grad_norm": 1.3505782987334707, "learning_rate": 9.329489249655032e-06, "loss": 0.4586, "step": 15037 }, { "epoch": 0.5357607282184655, "grad_norm": 1.3735389358948298, "learning_rate": 9.328337914309521e-06, "loss": 0.4528, "step": 15038 }, { "epoch": 0.5357963553449596, "grad_norm": 1.538228362125571, "learning_rate": 9.327186587907658e-06, "loss": 0.5098, "step": 15039 }, { "epoch": 0.5358319824714538, "grad_norm": 1.4759748962231836, "learning_rate": 9.326035270464777e-06, "loss": 0.4285, "step": 15040 }, { "epoch": 0.5358676095979479, "grad_norm": 2.1233791989481303, "learning_rate": 9.324883961996207e-06, "loss": 0.5078, "step": 15041 }, { "epoch": 0.535903236724442, "grad_norm": 1.0631552907701896, "learning_rate": 9.323732662517278e-06, "loss": 0.3858, "step": 15042 }, { "epoch": 0.5359388638509361, "grad_norm": 1.2978120709748984, "learning_rate": 9.322581372043322e-06, "loss": 0.483, "step": 15043 }, { "epoch": 0.5359744909774302, "grad_norm": 1.3993708947651948, "learning_rate": 9.321430090589668e-06, "loss": 0.4985, "step": 15044 }, { "epoch": 0.5360101181039243, "grad_norm": 1.2847261059473982, "learning_rate": 9.32027881817165e-06, "loss": 0.6118, "step": 15045 }, { "epoch": 0.5360457452304185, "grad_norm": 1.2225138654670664, "learning_rate": 9.31912755480459e-06, "loss": 0.4823, "step": 15046 }, { "epoch": 0.5360813723569126, "grad_norm": 1.5880931934985019, "learning_rate": 9.317976300503827e-06, "loss": 0.6184, "step": 15047 }, { "epoch": 0.5361169994834066, "grad_norm": 1.1405786241769655, "learning_rate": 9.316825055284682e-06, "loss": 0.45, "step": 15048 }, { "epoch": 0.5361526266099008, "grad_norm": 1.1106834668287158, "learning_rate": 9.315673819162494e-06, "loss": 0.4303, "step": 15049 }, { "epoch": 0.5361882537363949, "grad_norm": 1.4283051525109958, "learning_rate": 9.314522592152582e-06, "loss": 0.5739, "step": 15050 }, { "epoch": 0.536223880862889, "grad_norm": 1.1777987043558134, "learning_rate": 9.313371374270283e-06, "loss": 0.463, "step": 15051 }, { "epoch": 0.5362595079893832, "grad_norm": 1.0703172560987746, "learning_rate": 9.312220165530926e-06, "loss": 0.4385, "step": 15052 }, { "epoch": 0.5362951351158772, "grad_norm": 1.1896157472082582, "learning_rate": 9.311068965949834e-06, "loss": 0.4653, "step": 15053 }, { "epoch": 0.5363307622423713, "grad_norm": 1.638833788696407, "learning_rate": 9.309917775542345e-06, "loss": 0.587, "step": 15054 }, { "epoch": 0.5363663893688655, "grad_norm": 1.0223189791461542, "learning_rate": 9.308766594323778e-06, "loss": 0.4778, "step": 15055 }, { "epoch": 0.5364020164953596, "grad_norm": 1.2405762804381537, "learning_rate": 9.307615422309474e-06, "loss": 0.4541, "step": 15056 }, { "epoch": 0.5364376436218536, "grad_norm": 1.4265924605744587, "learning_rate": 9.306464259514747e-06, "loss": 0.599, "step": 15057 }, { "epoch": 0.5364732707483478, "grad_norm": 1.0430013999081666, "learning_rate": 9.305313105954937e-06, "loss": 0.4323, "step": 15058 }, { "epoch": 0.5365088978748419, "grad_norm": 1.0338495223717767, "learning_rate": 9.304161961645366e-06, "loss": 0.4071, "step": 15059 }, { "epoch": 0.536544525001336, "grad_norm": 1.2562547110426387, "learning_rate": 9.30301082660137e-06, "loss": 0.4638, "step": 15060 }, { "epoch": 0.5365801521278302, "grad_norm": 1.1286995707245053, "learning_rate": 9.301859700838268e-06, "loss": 0.465, "step": 15061 }, { "epoch": 0.5366157792543242, "grad_norm": 1.372729429621644, "learning_rate": 9.300708584371392e-06, "loss": 0.4579, "step": 15062 }, { "epoch": 0.5366514063808183, "grad_norm": 1.7734760620107342, "learning_rate": 9.299557477216073e-06, "loss": 0.579, "step": 15063 }, { "epoch": 0.5366870335073125, "grad_norm": 1.6353952085679835, "learning_rate": 9.298406379387631e-06, "loss": 0.5039, "step": 15064 }, { "epoch": 0.5367226606338066, "grad_norm": 1.077484867258152, "learning_rate": 9.297255290901404e-06, "loss": 0.4276, "step": 15065 }, { "epoch": 0.5367582877603007, "grad_norm": 1.437457187405407, "learning_rate": 9.296104211772709e-06, "loss": 0.5505, "step": 15066 }, { "epoch": 0.5367939148867948, "grad_norm": 0.9972919948877823, "learning_rate": 9.294953142016883e-06, "loss": 0.4535, "step": 15067 }, { "epoch": 0.5368295420132889, "grad_norm": 2.5597464775494516, "learning_rate": 9.293802081649244e-06, "loss": 0.5029, "step": 15068 }, { "epoch": 0.536865169139783, "grad_norm": 1.0841340388348117, "learning_rate": 9.292651030685126e-06, "loss": 0.4241, "step": 15069 }, { "epoch": 0.5369007962662772, "grad_norm": 1.1543128525012827, "learning_rate": 9.291499989139854e-06, "loss": 0.4134, "step": 15070 }, { "epoch": 0.5369364233927713, "grad_norm": 1.1784766011682166, "learning_rate": 9.290348957028757e-06, "loss": 0.4119, "step": 15071 }, { "epoch": 0.5369720505192653, "grad_norm": 1.632506155333053, "learning_rate": 9.289197934367156e-06, "loss": 0.6376, "step": 15072 }, { "epoch": 0.5370076776457595, "grad_norm": 1.4752823202617042, "learning_rate": 9.288046921170382e-06, "loss": 0.5923, "step": 15073 }, { "epoch": 0.5370433047722536, "grad_norm": 1.0212613206088523, "learning_rate": 9.286895917453763e-06, "loss": 0.4609, "step": 15074 }, { "epoch": 0.5370789318987477, "grad_norm": 1.3318943880854726, "learning_rate": 9.285744923232621e-06, "loss": 0.4786, "step": 15075 }, { "epoch": 0.5371145590252419, "grad_norm": 1.0063819881730391, "learning_rate": 9.284593938522289e-06, "loss": 0.4227, "step": 15076 }, { "epoch": 0.5371501861517359, "grad_norm": 1.1347901995830574, "learning_rate": 9.283442963338082e-06, "loss": 0.4514, "step": 15077 }, { "epoch": 0.53718581327823, "grad_norm": 1.2908437017627037, "learning_rate": 9.28229199769534e-06, "loss": 0.4404, "step": 15078 }, { "epoch": 0.5372214404047242, "grad_norm": 1.4560493520166107, "learning_rate": 9.281141041609374e-06, "loss": 0.5113, "step": 15079 }, { "epoch": 0.5372570675312183, "grad_norm": 1.25586457005638, "learning_rate": 9.279990095095525e-06, "loss": 0.4719, "step": 15080 }, { "epoch": 0.5372926946577123, "grad_norm": 1.018729774089867, "learning_rate": 9.278839158169103e-06, "loss": 0.454, "step": 15081 }, { "epoch": 0.5373283217842065, "grad_norm": 1.6141441646508923, "learning_rate": 9.27768823084545e-06, "loss": 0.6259, "step": 15082 }, { "epoch": 0.5373639489107006, "grad_norm": 1.3668313406520094, "learning_rate": 9.276537313139878e-06, "loss": 0.5844, "step": 15083 }, { "epoch": 0.5373995760371947, "grad_norm": 1.1101034867444703, "learning_rate": 9.27538640506772e-06, "loss": 0.4575, "step": 15084 }, { "epoch": 0.5374352031636889, "grad_norm": 1.3185700632681254, "learning_rate": 9.274235506644295e-06, "loss": 0.6168, "step": 15085 }, { "epoch": 0.5374708302901829, "grad_norm": 1.7281086247085782, "learning_rate": 9.273084617884932e-06, "loss": 0.5102, "step": 15086 }, { "epoch": 0.537506457416677, "grad_norm": 1.6127913791102504, "learning_rate": 9.27193373880496e-06, "loss": 0.5484, "step": 15087 }, { "epoch": 0.5375420845431712, "grad_norm": 1.7526134793598067, "learning_rate": 9.270782869419694e-06, "loss": 0.4287, "step": 15088 }, { "epoch": 0.5375777116696653, "grad_norm": 1.128164433236371, "learning_rate": 9.269632009744466e-06, "loss": 0.4107, "step": 15089 }, { "epoch": 0.5376133387961594, "grad_norm": 1.085843941356486, "learning_rate": 9.268481159794598e-06, "loss": 0.4275, "step": 15090 }, { "epoch": 0.5376489659226535, "grad_norm": 1.0749554556009697, "learning_rate": 9.267330319585418e-06, "loss": 0.4579, "step": 15091 }, { "epoch": 0.5376845930491476, "grad_norm": 1.1742447284008124, "learning_rate": 9.26617948913224e-06, "loss": 0.4418, "step": 15092 }, { "epoch": 0.5377202201756417, "grad_norm": 1.0638528171198653, "learning_rate": 9.265028668450403e-06, "loss": 0.46, "step": 15093 }, { "epoch": 0.5377558473021359, "grad_norm": 0.983402363474817, "learning_rate": 9.263877857555216e-06, "loss": 0.4428, "step": 15094 }, { "epoch": 0.53779147442863, "grad_norm": 1.7465602731292733, "learning_rate": 9.262727056462013e-06, "loss": 0.5219, "step": 15095 }, { "epoch": 0.537827101555124, "grad_norm": 1.100321163864178, "learning_rate": 9.261576265186113e-06, "loss": 0.4029, "step": 15096 }, { "epoch": 0.5378627286816182, "grad_norm": 1.294281847765296, "learning_rate": 9.26042548374284e-06, "loss": 0.4468, "step": 15097 }, { "epoch": 0.5378983558081123, "grad_norm": 1.0909040514912713, "learning_rate": 9.259274712147526e-06, "loss": 0.4771, "step": 15098 }, { "epoch": 0.5379339829346064, "grad_norm": 1.1158161394607444, "learning_rate": 9.25812395041548e-06, "loss": 0.4084, "step": 15099 }, { "epoch": 0.5379696100611006, "grad_norm": 1.5107711220607927, "learning_rate": 9.256973198562035e-06, "loss": 0.6578, "step": 15100 }, { "epoch": 0.5380052371875946, "grad_norm": 1.1457865954063338, "learning_rate": 9.255822456602508e-06, "loss": 0.5013, "step": 15101 }, { "epoch": 0.5380408643140887, "grad_norm": 1.135581929781423, "learning_rate": 9.25467172455223e-06, "loss": 0.4885, "step": 15102 }, { "epoch": 0.5380764914405829, "grad_norm": 1.1844555735538835, "learning_rate": 9.253521002426515e-06, "loss": 0.5445, "step": 15103 }, { "epoch": 0.538112118567077, "grad_norm": 1.2052938262257031, "learning_rate": 9.252370290240692e-06, "loss": 0.4739, "step": 15104 }, { "epoch": 0.538147745693571, "grad_norm": 1.0703482931812627, "learning_rate": 9.251219588010078e-06, "loss": 0.4614, "step": 15105 }, { "epoch": 0.5381833728200652, "grad_norm": 1.1222924816839472, "learning_rate": 9.250068895750004e-06, "loss": 0.4597, "step": 15106 }, { "epoch": 0.5382189999465593, "grad_norm": 1.049659774469016, "learning_rate": 9.24891821347578e-06, "loss": 0.4687, "step": 15107 }, { "epoch": 0.5382546270730534, "grad_norm": 1.1146585965397051, "learning_rate": 9.247767541202738e-06, "loss": 0.4511, "step": 15108 }, { "epoch": 0.5382902541995476, "grad_norm": 1.26720174268633, "learning_rate": 9.2466168789462e-06, "loss": 0.5008, "step": 15109 }, { "epoch": 0.5383258813260416, "grad_norm": 1.1213586773864823, "learning_rate": 9.24546622672148e-06, "loss": 0.4557, "step": 15110 }, { "epoch": 0.5383615084525357, "grad_norm": 0.9854753278364801, "learning_rate": 9.244315584543908e-06, "loss": 0.393, "step": 15111 }, { "epoch": 0.5383971355790299, "grad_norm": 1.784048366180004, "learning_rate": 9.243164952428801e-06, "loss": 0.6229, "step": 15112 }, { "epoch": 0.538432762705524, "grad_norm": 1.5862399629025, "learning_rate": 9.242014330391483e-06, "loss": 0.6199, "step": 15113 }, { "epoch": 0.5384683898320181, "grad_norm": 1.1080788554553958, "learning_rate": 9.240863718447271e-06, "loss": 0.4824, "step": 15114 }, { "epoch": 0.5385040169585122, "grad_norm": 1.092750734880486, "learning_rate": 9.239713116611493e-06, "loss": 0.4018, "step": 15115 }, { "epoch": 0.5385396440850063, "grad_norm": 1.5698026297953418, "learning_rate": 9.23856252489946e-06, "loss": 0.5248, "step": 15116 }, { "epoch": 0.5385752712115004, "grad_norm": 0.9879612454098692, "learning_rate": 9.237411943326508e-06, "loss": 0.395, "step": 15117 }, { "epoch": 0.5386108983379946, "grad_norm": 1.6547222013316554, "learning_rate": 9.23626137190794e-06, "loss": 0.5061, "step": 15118 }, { "epoch": 0.5386465254644887, "grad_norm": 1.230248548457113, "learning_rate": 9.235110810659094e-06, "loss": 0.4237, "step": 15119 }, { "epoch": 0.5386821525909827, "grad_norm": 1.1057123588604862, "learning_rate": 9.233960259595277e-06, "loss": 0.4098, "step": 15120 }, { "epoch": 0.5387177797174769, "grad_norm": 1.135785885729737, "learning_rate": 9.232809718731815e-06, "loss": 0.475, "step": 15121 }, { "epoch": 0.538753406843971, "grad_norm": 1.0679436698371385, "learning_rate": 9.231659188084028e-06, "loss": 0.4111, "step": 15122 }, { "epoch": 0.5387890339704651, "grad_norm": 1.0460734124823796, "learning_rate": 9.230508667667235e-06, "loss": 0.4313, "step": 15123 }, { "epoch": 0.5388246610969593, "grad_norm": 1.0945931022849271, "learning_rate": 9.229358157496761e-06, "loss": 0.4152, "step": 15124 }, { "epoch": 0.5388602882234533, "grad_norm": 1.8525663810181996, "learning_rate": 9.228207657587918e-06, "loss": 0.5952, "step": 15125 }, { "epoch": 0.5388959153499474, "grad_norm": 1.044025078329156, "learning_rate": 9.227057167956031e-06, "loss": 0.4374, "step": 15126 }, { "epoch": 0.5389315424764416, "grad_norm": 1.8626448099683741, "learning_rate": 9.225906688616418e-06, "loss": 0.5769, "step": 15127 }, { "epoch": 0.5389671696029357, "grad_norm": 1.4921642036236982, "learning_rate": 9.224756219584401e-06, "loss": 0.6464, "step": 15128 }, { "epoch": 0.5390027967294297, "grad_norm": 1.0489142575874995, "learning_rate": 9.223605760875291e-06, "loss": 0.4303, "step": 15129 }, { "epoch": 0.5390384238559239, "grad_norm": 1.4093705680087576, "learning_rate": 9.222455312504417e-06, "loss": 0.5414, "step": 15130 }, { "epoch": 0.539074050982418, "grad_norm": 1.0589977059562323, "learning_rate": 9.221304874487093e-06, "loss": 0.4389, "step": 15131 }, { "epoch": 0.5391096781089121, "grad_norm": 1.8007325343403007, "learning_rate": 9.220154446838638e-06, "loss": 0.6091, "step": 15132 }, { "epoch": 0.5391453052354063, "grad_norm": 1.4227026533098408, "learning_rate": 9.219004029574372e-06, "loss": 0.6531, "step": 15133 }, { "epoch": 0.5391809323619003, "grad_norm": 2.3922925728050153, "learning_rate": 9.217853622709612e-06, "loss": 0.6288, "step": 15134 }, { "epoch": 0.5392165594883944, "grad_norm": 2.2639292527641737, "learning_rate": 9.216703226259682e-06, "loss": 0.4947, "step": 15135 }, { "epoch": 0.5392521866148886, "grad_norm": 1.5672815438004652, "learning_rate": 9.215552840239893e-06, "loss": 0.4964, "step": 15136 }, { "epoch": 0.5392878137413827, "grad_norm": 1.0872699663328207, "learning_rate": 9.214402464665568e-06, "loss": 0.4423, "step": 15137 }, { "epoch": 0.5393234408678768, "grad_norm": 1.1182076283834923, "learning_rate": 9.213252099552019e-06, "loss": 0.4479, "step": 15138 }, { "epoch": 0.5393590679943709, "grad_norm": 1.4190865741791163, "learning_rate": 9.212101744914574e-06, "loss": 0.5506, "step": 15139 }, { "epoch": 0.539394695120865, "grad_norm": 1.5497098227720885, "learning_rate": 9.210951400768541e-06, "loss": 0.5367, "step": 15140 }, { "epoch": 0.5394303222473591, "grad_norm": 1.1820977639145742, "learning_rate": 9.209801067129244e-06, "loss": 0.4506, "step": 15141 }, { "epoch": 0.5394659493738533, "grad_norm": 1.3002732575318279, "learning_rate": 9.208650744011996e-06, "loss": 0.4721, "step": 15142 }, { "epoch": 0.5395015765003474, "grad_norm": 1.6148186687928996, "learning_rate": 9.207500431432115e-06, "loss": 0.5246, "step": 15143 }, { "epoch": 0.5395372036268414, "grad_norm": 1.6598713068739448, "learning_rate": 9.206350129404924e-06, "loss": 0.6589, "step": 15144 }, { "epoch": 0.5395728307533356, "grad_norm": 1.1386249488000104, "learning_rate": 9.20519983794573e-06, "loss": 0.4341, "step": 15145 }, { "epoch": 0.5396084578798297, "grad_norm": 1.2599179015109305, "learning_rate": 9.204049557069864e-06, "loss": 0.4898, "step": 15146 }, { "epoch": 0.5396440850063238, "grad_norm": 1.233442871317388, "learning_rate": 9.202899286792629e-06, "loss": 0.4529, "step": 15147 }, { "epoch": 0.539679712132818, "grad_norm": 1.380406503230074, "learning_rate": 9.201749027129348e-06, "loss": 0.5896, "step": 15148 }, { "epoch": 0.539715339259312, "grad_norm": 1.4141603950310864, "learning_rate": 9.200598778095335e-06, "loss": 0.5435, "step": 15149 }, { "epoch": 0.5397509663858061, "grad_norm": 1.2253326944426803, "learning_rate": 9.199448539705913e-06, "loss": 0.4546, "step": 15150 }, { "epoch": 0.5397865935123003, "grad_norm": 1.68766452333274, "learning_rate": 9.19829831197639e-06, "loss": 0.4826, "step": 15151 }, { "epoch": 0.5398222206387944, "grad_norm": 1.0827718921669367, "learning_rate": 9.197148094922085e-06, "loss": 0.4688, "step": 15152 }, { "epoch": 0.5398578477652884, "grad_norm": 1.324963500613475, "learning_rate": 9.195997888558312e-06, "loss": 0.5596, "step": 15153 }, { "epoch": 0.5398934748917826, "grad_norm": 1.0487270884727065, "learning_rate": 9.194847692900392e-06, "loss": 0.44, "step": 15154 }, { "epoch": 0.5399291020182767, "grad_norm": 1.3663649181151372, "learning_rate": 9.193697507963639e-06, "loss": 0.5414, "step": 15155 }, { "epoch": 0.5399647291447708, "grad_norm": 1.0314068028285972, "learning_rate": 9.192547333763362e-06, "loss": 0.4195, "step": 15156 }, { "epoch": 0.540000356271265, "grad_norm": 1.3713443441915274, "learning_rate": 9.191397170314888e-06, "loss": 0.6237, "step": 15157 }, { "epoch": 0.540035983397759, "grad_norm": 1.4090103272050143, "learning_rate": 9.190247017633523e-06, "loss": 0.5684, "step": 15158 }, { "epoch": 0.5400716105242531, "grad_norm": 1.0451137521430154, "learning_rate": 9.189096875734585e-06, "loss": 0.4206, "step": 15159 }, { "epoch": 0.5401072376507473, "grad_norm": 1.2626828589820538, "learning_rate": 9.187946744633387e-06, "loss": 0.4802, "step": 15160 }, { "epoch": 0.5401428647772414, "grad_norm": 1.4876741407173344, "learning_rate": 9.186796624345252e-06, "loss": 0.5643, "step": 15161 }, { "epoch": 0.5401784919037355, "grad_norm": 1.6482978390295617, "learning_rate": 9.185646514885482e-06, "loss": 0.55, "step": 15162 }, { "epoch": 0.5402141190302296, "grad_norm": 1.5439744984192991, "learning_rate": 9.184496416269401e-06, "loss": 0.5782, "step": 15163 }, { "epoch": 0.5402497461567237, "grad_norm": 1.11634071049077, "learning_rate": 9.183346328512316e-06, "loss": 0.4435, "step": 15164 }, { "epoch": 0.5402853732832178, "grad_norm": 1.0601364087652254, "learning_rate": 9.182196251629552e-06, "loss": 0.4289, "step": 15165 }, { "epoch": 0.540321000409712, "grad_norm": 1.2417376456297162, "learning_rate": 9.181046185636411e-06, "loss": 0.4685, "step": 15166 }, { "epoch": 0.5403566275362061, "grad_norm": 1.0563646146157568, "learning_rate": 9.179896130548215e-06, "loss": 0.4448, "step": 15167 }, { "epoch": 0.5403922546627001, "grad_norm": 1.5540861602621747, "learning_rate": 9.178746086380274e-06, "loss": 0.5537, "step": 15168 }, { "epoch": 0.5404278817891943, "grad_norm": 1.0685482940911757, "learning_rate": 9.177596053147903e-06, "loss": 0.4323, "step": 15169 }, { "epoch": 0.5404635089156884, "grad_norm": 1.5806887825737435, "learning_rate": 9.176446030866418e-06, "loss": 0.6264, "step": 15170 }, { "epoch": 0.5404991360421825, "grad_norm": 1.0834645137057968, "learning_rate": 9.175296019551125e-06, "loss": 0.4562, "step": 15171 }, { "epoch": 0.5405347631686767, "grad_norm": 2.246269266855306, "learning_rate": 9.174146019217348e-06, "loss": 0.5565, "step": 15172 }, { "epoch": 0.5405703902951707, "grad_norm": 1.095184284710025, "learning_rate": 9.172996029880389e-06, "loss": 0.4257, "step": 15173 }, { "epoch": 0.5406060174216648, "grad_norm": 1.5118220271354927, "learning_rate": 9.171846051555568e-06, "loss": 0.5239, "step": 15174 }, { "epoch": 0.540641644548159, "grad_norm": 1.0903723490832935, "learning_rate": 9.170696084258193e-06, "loss": 0.4362, "step": 15175 }, { "epoch": 0.5406772716746531, "grad_norm": 1.7540009384327253, "learning_rate": 9.169546128003585e-06, "loss": 0.6304, "step": 15176 }, { "epoch": 0.5407128988011471, "grad_norm": 1.1266166104889546, "learning_rate": 9.168396182807046e-06, "loss": 0.4516, "step": 15177 }, { "epoch": 0.5407485259276413, "grad_norm": 1.8634746329987864, "learning_rate": 9.167246248683893e-06, "loss": 0.5142, "step": 15178 }, { "epoch": 0.5407841530541354, "grad_norm": 1.3222559701253358, "learning_rate": 9.16609632564944e-06, "loss": 0.5561, "step": 15179 }, { "epoch": 0.5408197801806295, "grad_norm": 1.1274721717497491, "learning_rate": 9.164946413718996e-06, "loss": 0.3889, "step": 15180 }, { "epoch": 0.5408554073071237, "grad_norm": 1.432677301081006, "learning_rate": 9.163796512907875e-06, "loss": 0.5437, "step": 15181 }, { "epoch": 0.5408910344336177, "grad_norm": 1.504361466494516, "learning_rate": 9.162646623231388e-06, "loss": 0.6391, "step": 15182 }, { "epoch": 0.5409266615601118, "grad_norm": 1.3757766513935719, "learning_rate": 9.161496744704848e-06, "loss": 0.4987, "step": 15183 }, { "epoch": 0.540962288686606, "grad_norm": 1.034558734831663, "learning_rate": 9.160346877343562e-06, "loss": 0.4657, "step": 15184 }, { "epoch": 0.5409979158131001, "grad_norm": 1.4496516820396312, "learning_rate": 9.159197021162846e-06, "loss": 0.5094, "step": 15185 }, { "epoch": 0.5410335429395942, "grad_norm": 1.1619118083469024, "learning_rate": 9.15804717617801e-06, "loss": 0.5144, "step": 15186 }, { "epoch": 0.5410691700660883, "grad_norm": 1.1748531456320699, "learning_rate": 9.156897342404365e-06, "loss": 0.4628, "step": 15187 }, { "epoch": 0.5411047971925824, "grad_norm": 1.2214121557707691, "learning_rate": 9.155747519857218e-06, "loss": 0.5003, "step": 15188 }, { "epoch": 0.5411404243190765, "grad_norm": 1.137318139792515, "learning_rate": 9.154597708551882e-06, "loss": 0.4391, "step": 15189 }, { "epoch": 0.5411760514455707, "grad_norm": 1.2645215834211885, "learning_rate": 9.153447908503671e-06, "loss": 0.4545, "step": 15190 }, { "epoch": 0.5412116785720648, "grad_norm": 1.3262858947305434, "learning_rate": 9.152298119727892e-06, "loss": 0.5394, "step": 15191 }, { "epoch": 0.5412473056985588, "grad_norm": 1.1047252434765429, "learning_rate": 9.15114834223986e-06, "loss": 0.4224, "step": 15192 }, { "epoch": 0.541282932825053, "grad_norm": 1.1157281896374849, "learning_rate": 9.149998576054875e-06, "loss": 0.4621, "step": 15193 }, { "epoch": 0.5413185599515471, "grad_norm": 1.1149197495044567, "learning_rate": 9.148848821188257e-06, "loss": 0.4904, "step": 15194 }, { "epoch": 0.5413541870780412, "grad_norm": 1.2300973707096177, "learning_rate": 9.14769907765531e-06, "loss": 0.4922, "step": 15195 }, { "epoch": 0.5413898142045354, "grad_norm": 1.1058703167803312, "learning_rate": 9.146549345471348e-06, "loss": 0.4194, "step": 15196 }, { "epoch": 0.5414254413310294, "grad_norm": 1.4132808021441925, "learning_rate": 9.145399624651673e-06, "loss": 0.6407, "step": 15197 }, { "epoch": 0.5414610684575235, "grad_norm": 0.9278855584681754, "learning_rate": 9.144249915211605e-06, "loss": 0.432, "step": 15198 }, { "epoch": 0.5414966955840177, "grad_norm": 1.2281696240415354, "learning_rate": 9.143100217166445e-06, "loss": 0.4614, "step": 15199 }, { "epoch": 0.5415323227105118, "grad_norm": 1.7023909656389415, "learning_rate": 9.141950530531502e-06, "loss": 0.5921, "step": 15200 }, { "epoch": 0.5415679498370058, "grad_norm": 1.4172840556697213, "learning_rate": 9.140800855322092e-06, "loss": 0.5025, "step": 15201 }, { "epoch": 0.5416035769635, "grad_norm": 1.0681866970577905, "learning_rate": 9.139651191553515e-06, "loss": 0.4658, "step": 15202 }, { "epoch": 0.5416392040899941, "grad_norm": 0.9782043632298799, "learning_rate": 9.13850153924109e-06, "loss": 0.3912, "step": 15203 }, { "epoch": 0.5416748312164882, "grad_norm": 1.178785752153328, "learning_rate": 9.137351898400113e-06, "loss": 0.4485, "step": 15204 }, { "epoch": 0.5417104583429824, "grad_norm": 1.078366501894992, "learning_rate": 9.1362022690459e-06, "loss": 0.4081, "step": 15205 }, { "epoch": 0.5417460854694764, "grad_norm": 1.416200050092231, "learning_rate": 9.135052651193757e-06, "loss": 0.5211, "step": 15206 }, { "epoch": 0.5417817125959705, "grad_norm": 1.030898091974743, "learning_rate": 9.133903044858994e-06, "loss": 0.4036, "step": 15207 }, { "epoch": 0.5418173397224647, "grad_norm": 1.4095064220105624, "learning_rate": 9.132753450056916e-06, "loss": 0.5452, "step": 15208 }, { "epoch": 0.5418529668489588, "grad_norm": 1.083361789180702, "learning_rate": 9.131603866802836e-06, "loss": 0.4086, "step": 15209 }, { "epoch": 0.5418885939754529, "grad_norm": 1.6850510703531754, "learning_rate": 9.130454295112051e-06, "loss": 0.5749, "step": 15210 }, { "epoch": 0.541924221101947, "grad_norm": 1.345854262838176, "learning_rate": 9.12930473499988e-06, "loss": 0.5519, "step": 15211 }, { "epoch": 0.5419598482284411, "grad_norm": 1.691171579173816, "learning_rate": 9.128155186481622e-06, "loss": 0.5838, "step": 15212 }, { "epoch": 0.5419954753549352, "grad_norm": 1.0690086049922773, "learning_rate": 9.127005649572587e-06, "loss": 0.4179, "step": 15213 }, { "epoch": 0.5420311024814294, "grad_norm": 1.0958290075941235, "learning_rate": 9.125856124288087e-06, "loss": 0.4391, "step": 15214 }, { "epoch": 0.5420667296079235, "grad_norm": 1.1584153737111982, "learning_rate": 9.124706610643417e-06, "loss": 0.4561, "step": 15215 }, { "epoch": 0.5421023567344175, "grad_norm": 1.5980927628139663, "learning_rate": 9.123557108653894e-06, "loss": 0.5502, "step": 15216 }, { "epoch": 0.5421379838609117, "grad_norm": 1.0457853620639979, "learning_rate": 9.122407618334819e-06, "loss": 0.4492, "step": 15217 }, { "epoch": 0.5421736109874058, "grad_norm": 2.199473441531321, "learning_rate": 9.121258139701503e-06, "loss": 0.5921, "step": 15218 }, { "epoch": 0.5422092381138999, "grad_norm": 1.0738485729554352, "learning_rate": 9.120108672769243e-06, "loss": 0.4996, "step": 15219 }, { "epoch": 0.5422448652403941, "grad_norm": 1.5974171505464108, "learning_rate": 9.118959217553358e-06, "loss": 0.6227, "step": 15220 }, { "epoch": 0.5422804923668881, "grad_norm": 1.1713947404624239, "learning_rate": 9.117809774069144e-06, "loss": 0.5352, "step": 15221 }, { "epoch": 0.5423161194933822, "grad_norm": 1.0484742999809697, "learning_rate": 9.116660342331909e-06, "loss": 0.4557, "step": 15222 }, { "epoch": 0.5423517466198764, "grad_norm": 1.08605553134521, "learning_rate": 9.115510922356956e-06, "loss": 0.4216, "step": 15223 }, { "epoch": 0.5423873737463705, "grad_norm": 1.1689957664369313, "learning_rate": 9.114361514159596e-06, "loss": 0.4552, "step": 15224 }, { "epoch": 0.5424230008728645, "grad_norm": 1.6420257237635432, "learning_rate": 9.113212117755134e-06, "loss": 0.5796, "step": 15225 }, { "epoch": 0.5424586279993587, "grad_norm": 1.2115550347536124, "learning_rate": 9.112062733158871e-06, "loss": 0.4658, "step": 15226 }, { "epoch": 0.5424942551258528, "grad_norm": 1.1549512649806315, "learning_rate": 9.110913360386114e-06, "loss": 0.4963, "step": 15227 }, { "epoch": 0.5425298822523469, "grad_norm": 1.2228410624303347, "learning_rate": 9.109763999452163e-06, "loss": 0.4472, "step": 15228 }, { "epoch": 0.5425655093788411, "grad_norm": 2.4607733873418063, "learning_rate": 9.108614650372334e-06, "loss": 0.5334, "step": 15229 }, { "epoch": 0.5426011365053351, "grad_norm": 1.2834556341385615, "learning_rate": 9.107465313161916e-06, "loss": 0.558, "step": 15230 }, { "epoch": 0.5426367636318292, "grad_norm": 1.2045878731736244, "learning_rate": 9.106315987836226e-06, "loss": 0.5983, "step": 15231 }, { "epoch": 0.5426723907583234, "grad_norm": 1.9850765322929693, "learning_rate": 9.105166674410559e-06, "loss": 0.4976, "step": 15232 }, { "epoch": 0.5427080178848175, "grad_norm": 1.025939101240509, "learning_rate": 9.104017372900229e-06, "loss": 0.4162, "step": 15233 }, { "epoch": 0.5427436450113116, "grad_norm": 1.0186425661395995, "learning_rate": 9.102868083320528e-06, "loss": 0.4255, "step": 15234 }, { "epoch": 0.5427792721378057, "grad_norm": 1.0580633236783528, "learning_rate": 9.101718805686767e-06, "loss": 0.4877, "step": 15235 }, { "epoch": 0.5428148992642998, "grad_norm": 0.9381139288363329, "learning_rate": 9.100569540014252e-06, "loss": 0.4105, "step": 15236 }, { "epoch": 0.5428505263907939, "grad_norm": 1.0350635354761974, "learning_rate": 9.09942028631828e-06, "loss": 0.4538, "step": 15237 }, { "epoch": 0.5428861535172881, "grad_norm": 1.2938533950379987, "learning_rate": 9.098271044614155e-06, "loss": 0.5759, "step": 15238 }, { "epoch": 0.5429217806437822, "grad_norm": 1.8284706098279786, "learning_rate": 9.097121814917182e-06, "loss": 0.6193, "step": 15239 }, { "epoch": 0.5429574077702762, "grad_norm": 2.2616063927769594, "learning_rate": 9.095972597242666e-06, "loss": 0.5488, "step": 15240 }, { "epoch": 0.5429930348967704, "grad_norm": 1.4957847522153924, "learning_rate": 9.094823391605903e-06, "loss": 0.6195, "step": 15241 }, { "epoch": 0.5430286620232645, "grad_norm": 1.0438794045415487, "learning_rate": 9.093674198022201e-06, "loss": 0.4998, "step": 15242 }, { "epoch": 0.5430642891497586, "grad_norm": 1.1806976142782335, "learning_rate": 9.092525016506858e-06, "loss": 0.5229, "step": 15243 }, { "epoch": 0.5430999162762528, "grad_norm": 1.434328475753607, "learning_rate": 9.091375847075183e-06, "loss": 0.4907, "step": 15244 }, { "epoch": 0.5431355434027468, "grad_norm": 0.9938094733315168, "learning_rate": 9.090226689742469e-06, "loss": 0.4286, "step": 15245 }, { "epoch": 0.5431711705292409, "grad_norm": 1.1394332143781425, "learning_rate": 9.089077544524027e-06, "loss": 0.4233, "step": 15246 }, { "epoch": 0.5432067976557351, "grad_norm": 1.0864471635113278, "learning_rate": 9.087928411435151e-06, "loss": 0.4771, "step": 15247 }, { "epoch": 0.5432424247822292, "grad_norm": 2.176964443929206, "learning_rate": 9.086779290491145e-06, "loss": 0.52, "step": 15248 }, { "epoch": 0.5432780519087232, "grad_norm": 1.288989821682672, "learning_rate": 9.085630181707314e-06, "loss": 0.5226, "step": 15249 }, { "epoch": 0.5433136790352174, "grad_norm": 1.500926139667671, "learning_rate": 9.084481085098953e-06, "loss": 0.6225, "step": 15250 }, { "epoch": 0.5433493061617115, "grad_norm": 1.1419470750159357, "learning_rate": 9.083332000681371e-06, "loss": 0.4304, "step": 15251 }, { "epoch": 0.5433849332882056, "grad_norm": 1.5243421992166417, "learning_rate": 9.082182928469861e-06, "loss": 0.5165, "step": 15252 }, { "epoch": 0.5434205604146998, "grad_norm": 1.7390842469597534, "learning_rate": 9.081033868479727e-06, "loss": 0.4914, "step": 15253 }, { "epoch": 0.5434561875411938, "grad_norm": 1.082031016900354, "learning_rate": 9.07988482072627e-06, "loss": 0.4524, "step": 15254 }, { "epoch": 0.5434918146676879, "grad_norm": 1.6722321174979105, "learning_rate": 9.078735785224791e-06, "loss": 0.5956, "step": 15255 }, { "epoch": 0.5435274417941821, "grad_norm": 1.4495236412099841, "learning_rate": 9.077586761990586e-06, "loss": 0.5652, "step": 15256 }, { "epoch": 0.5435630689206762, "grad_norm": 1.0999140165206072, "learning_rate": 9.07643775103896e-06, "loss": 0.4602, "step": 15257 }, { "epoch": 0.5435986960471703, "grad_norm": 1.1448254610963051, "learning_rate": 9.075288752385208e-06, "loss": 0.43, "step": 15258 }, { "epoch": 0.5436343231736644, "grad_norm": 1.1004898617833463, "learning_rate": 9.074139766044634e-06, "loss": 0.4401, "step": 15259 }, { "epoch": 0.5436699503001585, "grad_norm": 1.3968799524864595, "learning_rate": 9.072990792032537e-06, "loss": 0.519, "step": 15260 }, { "epoch": 0.5437055774266527, "grad_norm": 1.1330008106409544, "learning_rate": 9.071841830364214e-06, "loss": 0.4446, "step": 15261 }, { "epoch": 0.5437412045531468, "grad_norm": 1.0425943975952545, "learning_rate": 9.070692881054972e-06, "loss": 0.4217, "step": 15262 }, { "epoch": 0.5437768316796409, "grad_norm": 1.0654769242304234, "learning_rate": 9.069543944120097e-06, "loss": 0.4416, "step": 15263 }, { "epoch": 0.543812458806135, "grad_norm": 1.1666372054469292, "learning_rate": 9.068395019574897e-06, "loss": 0.4819, "step": 15264 }, { "epoch": 0.5438480859326291, "grad_norm": 1.0107172963124367, "learning_rate": 9.067246107434667e-06, "loss": 0.4175, "step": 15265 }, { "epoch": 0.5438837130591232, "grad_norm": 1.1121897100572073, "learning_rate": 9.06609720771471e-06, "loss": 0.46, "step": 15266 }, { "epoch": 0.5439193401856174, "grad_norm": 1.135719055631394, "learning_rate": 9.06494832043032e-06, "loss": 0.4494, "step": 15267 }, { "epoch": 0.5439549673121115, "grad_norm": 1.2655381891953825, "learning_rate": 9.063799445596796e-06, "loss": 0.5118, "step": 15268 }, { "epoch": 0.5439905944386055, "grad_norm": 1.8616454260211532, "learning_rate": 9.062650583229435e-06, "loss": 0.5289, "step": 15269 }, { "epoch": 0.5440262215650997, "grad_norm": 1.4352827153851266, "learning_rate": 9.061501733343539e-06, "loss": 0.5529, "step": 15270 }, { "epoch": 0.5440618486915938, "grad_norm": 1.098878337206042, "learning_rate": 9.060352895954403e-06, "loss": 0.4752, "step": 15271 }, { "epoch": 0.5440974758180879, "grad_norm": 1.1078392625005204, "learning_rate": 9.059204071077324e-06, "loss": 0.4168, "step": 15272 }, { "epoch": 0.5441331029445821, "grad_norm": 1.0685000319537004, "learning_rate": 9.058055258727605e-06, "loss": 0.4741, "step": 15273 }, { "epoch": 0.5441687300710761, "grad_norm": 1.7603056455147739, "learning_rate": 9.056906458920534e-06, "loss": 0.5737, "step": 15274 }, { "epoch": 0.5442043571975702, "grad_norm": 1.2019679109283725, "learning_rate": 9.055757671671414e-06, "loss": 0.4443, "step": 15275 }, { "epoch": 0.5442399843240644, "grad_norm": 1.0025938580160263, "learning_rate": 9.054608896995538e-06, "loss": 0.422, "step": 15276 }, { "epoch": 0.5442756114505585, "grad_norm": 0.9804169610895811, "learning_rate": 9.053460134908213e-06, "loss": 0.4252, "step": 15277 }, { "epoch": 0.5443112385770525, "grad_norm": 1.25067528220698, "learning_rate": 9.052311385424719e-06, "loss": 0.5015, "step": 15278 }, { "epoch": 0.5443468657035467, "grad_norm": 1.1568570074284026, "learning_rate": 9.051162648560366e-06, "loss": 0.4851, "step": 15279 }, { "epoch": 0.5443824928300408, "grad_norm": 1.4811359341684254, "learning_rate": 9.050013924330444e-06, "loss": 0.5788, "step": 15280 }, { "epoch": 0.5444181199565349, "grad_norm": 1.395876864332479, "learning_rate": 9.048865212750248e-06, "loss": 0.5812, "step": 15281 }, { "epoch": 0.5444537470830291, "grad_norm": 1.1479163559372574, "learning_rate": 9.04771651383508e-06, "loss": 0.4382, "step": 15282 }, { "epoch": 0.5444893742095231, "grad_norm": 1.4595412972209068, "learning_rate": 9.046567827600231e-06, "loss": 0.4448, "step": 15283 }, { "epoch": 0.5445250013360172, "grad_norm": 1.0401999045615906, "learning_rate": 9.045419154061002e-06, "loss": 0.3927, "step": 15284 }, { "epoch": 0.5445606284625114, "grad_norm": 1.578723943259804, "learning_rate": 9.044270493232678e-06, "loss": 0.5939, "step": 15285 }, { "epoch": 0.5445962555890055, "grad_norm": 1.6921088719244481, "learning_rate": 9.043121845130564e-06, "loss": 0.5228, "step": 15286 }, { "epoch": 0.5446318827154996, "grad_norm": 1.0077911921763734, "learning_rate": 9.04197320976995e-06, "loss": 0.4236, "step": 15287 }, { "epoch": 0.5446675098419937, "grad_norm": 1.217282835917277, "learning_rate": 9.040824587166136e-06, "loss": 0.5179, "step": 15288 }, { "epoch": 0.5447031369684878, "grad_norm": 1.988836225054412, "learning_rate": 9.039675977334408e-06, "loss": 0.4908, "step": 15289 }, { "epoch": 0.5447387640949819, "grad_norm": 1.1599055668199971, "learning_rate": 9.038527380290069e-06, "loss": 0.4434, "step": 15290 }, { "epoch": 0.5447743912214761, "grad_norm": 1.0334388795927716, "learning_rate": 9.037378796048407e-06, "loss": 0.367, "step": 15291 }, { "epoch": 0.5448100183479702, "grad_norm": 1.5218119899246736, "learning_rate": 9.036230224624724e-06, "loss": 0.5727, "step": 15292 }, { "epoch": 0.5448456454744642, "grad_norm": 1.1952001604494358, "learning_rate": 9.035081666034304e-06, "loss": 0.5595, "step": 15293 }, { "epoch": 0.5448812726009584, "grad_norm": 1.4563402009273179, "learning_rate": 9.033933120292445e-06, "loss": 0.5388, "step": 15294 }, { "epoch": 0.5449168997274525, "grad_norm": 1.0810212017649432, "learning_rate": 9.032784587414448e-06, "loss": 0.4303, "step": 15295 }, { "epoch": 0.5449525268539466, "grad_norm": 1.2758092441914637, "learning_rate": 9.031636067415595e-06, "loss": 0.4167, "step": 15296 }, { "epoch": 0.5449881539804408, "grad_norm": 1.446691139423093, "learning_rate": 9.030487560311187e-06, "loss": 0.4983, "step": 15297 }, { "epoch": 0.5450237811069348, "grad_norm": 1.7351435275395624, "learning_rate": 9.029339066116512e-06, "loss": 0.5864, "step": 15298 }, { "epoch": 0.5450594082334289, "grad_norm": 1.204837218664031, "learning_rate": 9.028190584846872e-06, "loss": 0.5048, "step": 15299 }, { "epoch": 0.5450950353599231, "grad_norm": 1.5337125906092328, "learning_rate": 9.027042116517547e-06, "loss": 0.6119, "step": 15300 }, { "epoch": 0.5451306624864172, "grad_norm": 1.1639495620160518, "learning_rate": 9.02589366114384e-06, "loss": 0.4906, "step": 15301 }, { "epoch": 0.5451662896129112, "grad_norm": 0.9697519249535119, "learning_rate": 9.024745218741037e-06, "loss": 0.4361, "step": 15302 }, { "epoch": 0.5452019167394054, "grad_norm": 1.1291713907068366, "learning_rate": 9.023596789324439e-06, "loss": 0.4429, "step": 15303 }, { "epoch": 0.5452375438658995, "grad_norm": 1.0805168132817269, "learning_rate": 9.022448372909326e-06, "loss": 0.4462, "step": 15304 }, { "epoch": 0.5452731709923936, "grad_norm": 1.536436291171707, "learning_rate": 9.021299969510998e-06, "loss": 0.4907, "step": 15305 }, { "epoch": 0.5453087981188878, "grad_norm": 1.3779683274617922, "learning_rate": 9.020151579144745e-06, "loss": 0.5115, "step": 15306 }, { "epoch": 0.5453444252453818, "grad_norm": 1.04562436405903, "learning_rate": 9.019003201825858e-06, "loss": 0.434, "step": 15307 }, { "epoch": 0.5453800523718759, "grad_norm": 1.8380002563916011, "learning_rate": 9.017854837569629e-06, "loss": 0.5333, "step": 15308 }, { "epoch": 0.5454156794983701, "grad_norm": 1.0290935868734619, "learning_rate": 9.016706486391349e-06, "loss": 0.4172, "step": 15309 }, { "epoch": 0.5454513066248642, "grad_norm": 1.3402544685812137, "learning_rate": 9.015558148306313e-06, "loss": 0.5706, "step": 15310 }, { "epoch": 0.5454869337513583, "grad_norm": 1.057308664771511, "learning_rate": 9.014409823329807e-06, "loss": 0.4438, "step": 15311 }, { "epoch": 0.5455225608778524, "grad_norm": 1.5410920532936343, "learning_rate": 9.013261511477122e-06, "loss": 0.5722, "step": 15312 }, { "epoch": 0.5455581880043465, "grad_norm": 1.0994209800700216, "learning_rate": 9.012113212763547e-06, "loss": 0.4769, "step": 15313 }, { "epoch": 0.5455938151308406, "grad_norm": 1.7607803057967337, "learning_rate": 9.01096492720438e-06, "loss": 0.5328, "step": 15314 }, { "epoch": 0.5456294422573348, "grad_norm": 1.3422211168889964, "learning_rate": 9.009816654814903e-06, "loss": 0.5648, "step": 15315 }, { "epoch": 0.5456650693838289, "grad_norm": 1.0652691944988821, "learning_rate": 9.008668395610409e-06, "loss": 0.4017, "step": 15316 }, { "epoch": 0.5457006965103229, "grad_norm": 1.065927126312941, "learning_rate": 9.007520149606191e-06, "loss": 0.4305, "step": 15317 }, { "epoch": 0.5457363236368171, "grad_norm": 1.175517093209937, "learning_rate": 9.006371916817533e-06, "loss": 0.4505, "step": 15318 }, { "epoch": 0.5457719507633112, "grad_norm": 1.4871448027308949, "learning_rate": 9.005223697259732e-06, "loss": 0.5687, "step": 15319 }, { "epoch": 0.5458075778898053, "grad_norm": 1.8781447753526002, "learning_rate": 9.004075490948066e-06, "loss": 0.6041, "step": 15320 }, { "epoch": 0.5458432050162995, "grad_norm": 1.1772287509854493, "learning_rate": 9.002927297897837e-06, "loss": 0.4446, "step": 15321 }, { "epoch": 0.5458788321427935, "grad_norm": 0.9878673475401973, "learning_rate": 9.001779118124325e-06, "loss": 0.3962, "step": 15322 }, { "epoch": 0.5459144592692876, "grad_norm": 1.0297245470362606, "learning_rate": 9.000630951642823e-06, "loss": 0.4096, "step": 15323 }, { "epoch": 0.5459500863957818, "grad_norm": 1.1498450880211228, "learning_rate": 8.999482798468616e-06, "loss": 0.4867, "step": 15324 }, { "epoch": 0.5459857135222759, "grad_norm": 1.3446144275428953, "learning_rate": 8.998334658617e-06, "loss": 0.5687, "step": 15325 }, { "epoch": 0.54602134064877, "grad_norm": 1.558827553200032, "learning_rate": 8.997186532103252e-06, "loss": 0.5885, "step": 15326 }, { "epoch": 0.5460569677752641, "grad_norm": 1.1321911179410817, "learning_rate": 8.996038418942669e-06, "loss": 0.4443, "step": 15327 }, { "epoch": 0.5460925949017582, "grad_norm": 1.205928154318621, "learning_rate": 8.994890319150536e-06, "loss": 0.5188, "step": 15328 }, { "epoch": 0.5461282220282523, "grad_norm": 1.4601497994470227, "learning_rate": 8.99374223274214e-06, "loss": 0.5352, "step": 15329 }, { "epoch": 0.5461638491547465, "grad_norm": 1.3301103621698802, "learning_rate": 8.992594159732773e-06, "loss": 0.4466, "step": 15330 }, { "epoch": 0.5461994762812405, "grad_norm": 1.1346396328203714, "learning_rate": 8.991446100137715e-06, "loss": 0.4848, "step": 15331 }, { "epoch": 0.5462351034077346, "grad_norm": 1.535220395391364, "learning_rate": 8.990298053972259e-06, "loss": 0.4843, "step": 15332 }, { "epoch": 0.5462707305342288, "grad_norm": 2.0634188815697363, "learning_rate": 8.98915002125169e-06, "loss": 0.5162, "step": 15333 }, { "epoch": 0.5463063576607229, "grad_norm": 1.4004536734237576, "learning_rate": 8.988002001991295e-06, "loss": 0.515, "step": 15334 }, { "epoch": 0.546341984787217, "grad_norm": 1.0250633377070535, "learning_rate": 8.986853996206359e-06, "loss": 0.409, "step": 15335 }, { "epoch": 0.5463776119137111, "grad_norm": 2.0518121070266435, "learning_rate": 8.985706003912175e-06, "loss": 0.4911, "step": 15336 }, { "epoch": 0.5464132390402052, "grad_norm": 1.3334755399678286, "learning_rate": 8.98455802512402e-06, "loss": 0.4401, "step": 15337 }, { "epoch": 0.5464488661666993, "grad_norm": 1.1337934429380991, "learning_rate": 8.983410059857184e-06, "loss": 0.4163, "step": 15338 }, { "epoch": 0.5464844932931935, "grad_norm": 1.0502725318483679, "learning_rate": 8.982262108126956e-06, "loss": 0.4597, "step": 15339 }, { "epoch": 0.5465201204196876, "grad_norm": 1.175344230278277, "learning_rate": 8.981114169948617e-06, "loss": 0.4456, "step": 15340 }, { "epoch": 0.5465557475461816, "grad_norm": 1.429620398506493, "learning_rate": 8.97996624533746e-06, "loss": 0.5041, "step": 15341 }, { "epoch": 0.5465913746726758, "grad_norm": 1.2932433213422723, "learning_rate": 8.97881833430876e-06, "loss": 0.4924, "step": 15342 }, { "epoch": 0.5466270017991699, "grad_norm": 1.4451481336770544, "learning_rate": 8.977670436877812e-06, "loss": 0.5141, "step": 15343 }, { "epoch": 0.546662628925664, "grad_norm": 1.436760223381693, "learning_rate": 8.976522553059891e-06, "loss": 0.5878, "step": 15344 }, { "epoch": 0.5466982560521582, "grad_norm": 1.1058376888398322, "learning_rate": 8.975374682870294e-06, "loss": 0.4324, "step": 15345 }, { "epoch": 0.5467338831786522, "grad_norm": 1.6487908376001736, "learning_rate": 8.974226826324293e-06, "loss": 0.5442, "step": 15346 }, { "epoch": 0.5467695103051463, "grad_norm": 2.566593580403113, "learning_rate": 8.973078983437185e-06, "loss": 0.5493, "step": 15347 }, { "epoch": 0.5468051374316405, "grad_norm": 1.5815792019866481, "learning_rate": 8.971931154224245e-06, "loss": 0.4794, "step": 15348 }, { "epoch": 0.5468407645581346, "grad_norm": 1.0879934266560427, "learning_rate": 8.97078333870076e-06, "loss": 0.4342, "step": 15349 }, { "epoch": 0.5468763916846286, "grad_norm": 1.0919210472794523, "learning_rate": 8.969635536882014e-06, "loss": 0.45, "step": 15350 }, { "epoch": 0.5469120188111228, "grad_norm": 1.1495951583758908, "learning_rate": 8.96848774878329e-06, "loss": 0.4366, "step": 15351 }, { "epoch": 0.5469476459376169, "grad_norm": 1.4296008274620327, "learning_rate": 8.967339974419877e-06, "loss": 0.5511, "step": 15352 }, { "epoch": 0.546983273064111, "grad_norm": 1.2291922193860614, "learning_rate": 8.96619221380705e-06, "loss": 0.4483, "step": 15353 }, { "epoch": 0.5470189001906052, "grad_norm": 1.1993008803261245, "learning_rate": 8.965044466960099e-06, "loss": 0.4741, "step": 15354 }, { "epoch": 0.5470545273170992, "grad_norm": 1.0499036773188337, "learning_rate": 8.963896733894302e-06, "loss": 0.4947, "step": 15355 }, { "epoch": 0.5470901544435933, "grad_norm": 1.12691959190285, "learning_rate": 8.962749014624948e-06, "loss": 0.5011, "step": 15356 }, { "epoch": 0.5471257815700875, "grad_norm": 1.0486991702230568, "learning_rate": 8.961601309167309e-06, "loss": 0.4308, "step": 15357 }, { "epoch": 0.5471614086965816, "grad_norm": 1.544116523837091, "learning_rate": 8.960453617536682e-06, "loss": 0.5473, "step": 15358 }, { "epoch": 0.5471970358230757, "grad_norm": 1.423192093557694, "learning_rate": 8.959305939748336e-06, "loss": 0.5579, "step": 15359 }, { "epoch": 0.5472326629495698, "grad_norm": 1.4067123515430677, "learning_rate": 8.958158275817562e-06, "loss": 0.533, "step": 15360 }, { "epoch": 0.5472682900760639, "grad_norm": 1.0939143571838703, "learning_rate": 8.957010625759635e-06, "loss": 0.4871, "step": 15361 }, { "epoch": 0.547303917202558, "grad_norm": 1.1045242364412153, "learning_rate": 8.955862989589842e-06, "loss": 0.4267, "step": 15362 }, { "epoch": 0.5473395443290522, "grad_norm": 1.4210863141314547, "learning_rate": 8.954715367323468e-06, "loss": 0.5472, "step": 15363 }, { "epoch": 0.5473751714555463, "grad_norm": 1.4055619949915383, "learning_rate": 8.953567758975785e-06, "loss": 0.6048, "step": 15364 }, { "epoch": 0.5474107985820403, "grad_norm": 1.0327997269228222, "learning_rate": 8.952420164562078e-06, "loss": 0.4407, "step": 15365 }, { "epoch": 0.5474464257085345, "grad_norm": 1.1674763508626598, "learning_rate": 8.95127258409763e-06, "loss": 0.4717, "step": 15366 }, { "epoch": 0.5474820528350286, "grad_norm": 1.95192006058346, "learning_rate": 8.950125017597724e-06, "loss": 0.5969, "step": 15367 }, { "epoch": 0.5475176799615227, "grad_norm": 1.0262822264320106, "learning_rate": 8.948977465077632e-06, "loss": 0.475, "step": 15368 }, { "epoch": 0.5475533070880169, "grad_norm": 1.4657443155202707, "learning_rate": 8.947829926552643e-06, "loss": 0.5296, "step": 15369 }, { "epoch": 0.5475889342145109, "grad_norm": 1.2459046892795265, "learning_rate": 8.94668240203803e-06, "loss": 0.4959, "step": 15370 }, { "epoch": 0.547624561341005, "grad_norm": 1.4123041106832728, "learning_rate": 8.94553489154908e-06, "loss": 0.4768, "step": 15371 }, { "epoch": 0.5476601884674992, "grad_norm": 1.4723902384224563, "learning_rate": 8.94438739510107e-06, "loss": 0.468, "step": 15372 }, { "epoch": 0.5476958155939933, "grad_norm": 1.0561375597380303, "learning_rate": 8.943239912709281e-06, "loss": 0.4683, "step": 15373 }, { "epoch": 0.5477314427204873, "grad_norm": 1.1138195381756069, "learning_rate": 8.942092444388988e-06, "loss": 0.4161, "step": 15374 }, { "epoch": 0.5477670698469815, "grad_norm": 1.0845066209513816, "learning_rate": 8.940944990155474e-06, "loss": 0.4431, "step": 15375 }, { "epoch": 0.5478026969734756, "grad_norm": 1.1470340653334887, "learning_rate": 8.939797550024018e-06, "loss": 0.4275, "step": 15376 }, { "epoch": 0.5478383240999697, "grad_norm": 2.7560577551976198, "learning_rate": 8.938650124009899e-06, "loss": 0.61, "step": 15377 }, { "epoch": 0.5478739512264639, "grad_norm": 1.9869324617269633, "learning_rate": 8.9375027121284e-06, "loss": 0.5429, "step": 15378 }, { "epoch": 0.547909578352958, "grad_norm": 1.3467319442033243, "learning_rate": 8.936355314394788e-06, "loss": 0.5334, "step": 15379 }, { "epoch": 0.547945205479452, "grad_norm": 1.0592896077509717, "learning_rate": 8.935207930824351e-06, "loss": 0.441, "step": 15380 }, { "epoch": 0.5479808326059462, "grad_norm": 0.9946670030674966, "learning_rate": 8.934060561432364e-06, "loss": 0.397, "step": 15381 }, { "epoch": 0.5480164597324403, "grad_norm": 1.60699894234748, "learning_rate": 8.93291320623411e-06, "loss": 0.6149, "step": 15382 }, { "epoch": 0.5480520868589344, "grad_norm": 1.099850730311809, "learning_rate": 8.931765865244854e-06, "loss": 0.4732, "step": 15383 }, { "epoch": 0.5480877139854285, "grad_norm": 1.0319362296996788, "learning_rate": 8.93061853847989e-06, "loss": 0.4031, "step": 15384 }, { "epoch": 0.5481233411119226, "grad_norm": 1.1186046541051933, "learning_rate": 8.929471225954484e-06, "loss": 0.4835, "step": 15385 }, { "epoch": 0.5481589682384167, "grad_norm": 1.1170763159263313, "learning_rate": 8.928323927683916e-06, "loss": 0.468, "step": 15386 }, { "epoch": 0.5481945953649109, "grad_norm": 1.5098256981499028, "learning_rate": 8.927176643683467e-06, "loss": 0.5545, "step": 15387 }, { "epoch": 0.548230222491405, "grad_norm": 1.2287045699678074, "learning_rate": 8.926029373968407e-06, "loss": 0.4471, "step": 15388 }, { "epoch": 0.548265849617899, "grad_norm": 1.1614083169368128, "learning_rate": 8.924882118554022e-06, "loss": 0.4317, "step": 15389 }, { "epoch": 0.5483014767443932, "grad_norm": 1.4358321637410685, "learning_rate": 8.92373487745558e-06, "loss": 0.6233, "step": 15390 }, { "epoch": 0.5483371038708873, "grad_norm": 1.1088745534093367, "learning_rate": 8.92258765068836e-06, "loss": 0.4554, "step": 15391 }, { "epoch": 0.5483727309973814, "grad_norm": 1.0633983743045818, "learning_rate": 8.921440438267637e-06, "loss": 0.4241, "step": 15392 }, { "epoch": 0.5484083581238756, "grad_norm": 1.5821656626617353, "learning_rate": 8.920293240208694e-06, "loss": 0.5151, "step": 15393 }, { "epoch": 0.5484439852503696, "grad_norm": 1.2940452869946701, "learning_rate": 8.919146056526796e-06, "loss": 0.5233, "step": 15394 }, { "epoch": 0.5484796123768637, "grad_norm": 1.6546955604576556, "learning_rate": 8.917998887237227e-06, "loss": 0.5633, "step": 15395 }, { "epoch": 0.5485152395033579, "grad_norm": 1.0725824417346335, "learning_rate": 8.916851732355256e-06, "loss": 0.4241, "step": 15396 }, { "epoch": 0.548550866629852, "grad_norm": 1.3104809724360589, "learning_rate": 8.91570459189616e-06, "loss": 0.5046, "step": 15397 }, { "epoch": 0.548586493756346, "grad_norm": 1.1159554084075596, "learning_rate": 8.91455746587522e-06, "loss": 0.4463, "step": 15398 }, { "epoch": 0.5486221208828402, "grad_norm": 1.5601424966608208, "learning_rate": 8.913410354307701e-06, "loss": 0.5056, "step": 15399 }, { "epoch": 0.5486577480093343, "grad_norm": 1.4544413436008221, "learning_rate": 8.912263257208888e-06, "loss": 0.6225, "step": 15400 }, { "epoch": 0.5486933751358284, "grad_norm": 1.0562314113750757, "learning_rate": 8.911116174594047e-06, "loss": 0.4161, "step": 15401 }, { "epoch": 0.5487290022623226, "grad_norm": 1.0802512202351597, "learning_rate": 8.909969106478456e-06, "loss": 0.4598, "step": 15402 }, { "epoch": 0.5487646293888166, "grad_norm": 1.6439111653861784, "learning_rate": 8.908822052877387e-06, "loss": 0.5375, "step": 15403 }, { "epoch": 0.5488002565153107, "grad_norm": 2.0140526597209814, "learning_rate": 8.907675013806118e-06, "loss": 0.5744, "step": 15404 }, { "epoch": 0.5488358836418049, "grad_norm": 1.7903239483726023, "learning_rate": 8.906527989279917e-06, "loss": 0.6432, "step": 15405 }, { "epoch": 0.548871510768299, "grad_norm": 1.1279593967369816, "learning_rate": 8.905380979314062e-06, "loss": 0.5096, "step": 15406 }, { "epoch": 0.5489071378947931, "grad_norm": 1.9769316734661246, "learning_rate": 8.904233983923821e-06, "loss": 0.4711, "step": 15407 }, { "epoch": 0.5489427650212872, "grad_norm": 2.5761310164535454, "learning_rate": 8.903087003124471e-06, "loss": 0.5078, "step": 15408 }, { "epoch": 0.5489783921477813, "grad_norm": 1.1320884144735597, "learning_rate": 8.901940036931287e-06, "loss": 0.44, "step": 15409 }, { "epoch": 0.5490140192742754, "grad_norm": 1.0575728552978716, "learning_rate": 8.900793085359536e-06, "loss": 0.4173, "step": 15410 }, { "epoch": 0.5490496464007696, "grad_norm": 1.0011318139894358, "learning_rate": 8.899646148424499e-06, "loss": 0.4422, "step": 15411 }, { "epoch": 0.5490852735272637, "grad_norm": 1.0778658882680465, "learning_rate": 8.898499226141438e-06, "loss": 0.4129, "step": 15412 }, { "epoch": 0.5491209006537577, "grad_norm": 1.6165656435726978, "learning_rate": 8.89735231852563e-06, "loss": 0.5712, "step": 15413 }, { "epoch": 0.5491565277802519, "grad_norm": 1.6087348154010073, "learning_rate": 8.896205425592346e-06, "loss": 0.4491, "step": 15414 }, { "epoch": 0.549192154906746, "grad_norm": 1.193934355666672, "learning_rate": 8.895058547356862e-06, "loss": 0.5119, "step": 15415 }, { "epoch": 0.5492277820332401, "grad_norm": 1.1050034560508897, "learning_rate": 8.893911683834442e-06, "loss": 0.4561, "step": 15416 }, { "epoch": 0.5492634091597343, "grad_norm": 1.0206745215737558, "learning_rate": 8.892764835040362e-06, "loss": 0.444, "step": 15417 }, { "epoch": 0.5492990362862283, "grad_norm": 1.01965936316802, "learning_rate": 8.89161800098989e-06, "loss": 0.3899, "step": 15418 }, { "epoch": 0.5493346634127224, "grad_norm": 1.81962459440897, "learning_rate": 8.890471181698304e-06, "loss": 0.528, "step": 15419 }, { "epoch": 0.5493702905392166, "grad_norm": 1.1318614251948864, "learning_rate": 8.889324377180866e-06, "loss": 0.4329, "step": 15420 }, { "epoch": 0.5494059176657107, "grad_norm": 1.1621786949258572, "learning_rate": 8.88817758745285e-06, "loss": 0.4988, "step": 15421 }, { "epoch": 0.5494415447922048, "grad_norm": 1.7911791427352048, "learning_rate": 8.88703081252953e-06, "loss": 0.5462, "step": 15422 }, { "epoch": 0.5494771719186989, "grad_norm": 1.2942547147489147, "learning_rate": 8.885884052426168e-06, "loss": 0.4728, "step": 15423 }, { "epoch": 0.549512799045193, "grad_norm": 1.7834269774810456, "learning_rate": 8.88473730715804e-06, "loss": 0.5432, "step": 15424 }, { "epoch": 0.5495484261716871, "grad_norm": 1.1301966174014755, "learning_rate": 8.883590576740416e-06, "loss": 0.4221, "step": 15425 }, { "epoch": 0.5495840532981813, "grad_norm": 1.2844845945704488, "learning_rate": 8.882443861188566e-06, "loss": 0.5055, "step": 15426 }, { "epoch": 0.5496196804246753, "grad_norm": 1.0643215319473245, "learning_rate": 8.881297160517752e-06, "loss": 0.4415, "step": 15427 }, { "epoch": 0.5496553075511694, "grad_norm": 1.2295793515144766, "learning_rate": 8.880150474743251e-06, "loss": 0.6379, "step": 15428 }, { "epoch": 0.5496909346776636, "grad_norm": 1.9080799103246926, "learning_rate": 8.879003803880328e-06, "loss": 0.5475, "step": 15429 }, { "epoch": 0.5497265618041577, "grad_norm": 1.1131527944950579, "learning_rate": 8.877857147944256e-06, "loss": 0.477, "step": 15430 }, { "epoch": 0.5497621889306518, "grad_norm": 1.400332156231177, "learning_rate": 8.876710506950296e-06, "loss": 0.569, "step": 15431 }, { "epoch": 0.549797816057146, "grad_norm": 1.0190678736169352, "learning_rate": 8.875563880913719e-06, "loss": 0.435, "step": 15432 }, { "epoch": 0.54983344318364, "grad_norm": 1.452437886030167, "learning_rate": 8.8744172698498e-06, "loss": 0.5841, "step": 15433 }, { "epoch": 0.5498690703101341, "grad_norm": 1.5346113885112775, "learning_rate": 8.8732706737738e-06, "loss": 0.5652, "step": 15434 }, { "epoch": 0.5499046974366283, "grad_norm": 1.0253057343359044, "learning_rate": 8.872124092700988e-06, "loss": 0.4371, "step": 15435 }, { "epoch": 0.5499403245631224, "grad_norm": 1.0469605397902642, "learning_rate": 8.870977526646632e-06, "loss": 0.3997, "step": 15436 }, { "epoch": 0.5499759516896164, "grad_norm": 1.5506791885510274, "learning_rate": 8.869830975626004e-06, "loss": 0.5142, "step": 15437 }, { "epoch": 0.5500115788161106, "grad_norm": 1.0376837215386694, "learning_rate": 8.86868443965436e-06, "loss": 0.4081, "step": 15438 }, { "epoch": 0.5500472059426047, "grad_norm": 1.5381759263340902, "learning_rate": 8.867537918746977e-06, "loss": 0.6686, "step": 15439 }, { "epoch": 0.5500828330690988, "grad_norm": 1.1100592766037582, "learning_rate": 8.866391412919116e-06, "loss": 0.431, "step": 15440 }, { "epoch": 0.550118460195593, "grad_norm": 1.1581073553644274, "learning_rate": 8.86524492218605e-06, "loss": 0.4548, "step": 15441 }, { "epoch": 0.550154087322087, "grad_norm": 1.336175406876651, "learning_rate": 8.864098446563037e-06, "loss": 0.518, "step": 15442 }, { "epoch": 0.5501897144485811, "grad_norm": 1.183868540156283, "learning_rate": 8.862951986065346e-06, "loss": 0.4459, "step": 15443 }, { "epoch": 0.5502253415750753, "grad_norm": 0.9949198476141989, "learning_rate": 8.861805540708246e-06, "loss": 0.4234, "step": 15444 }, { "epoch": 0.5502609687015694, "grad_norm": 1.4784524614009855, "learning_rate": 8.860659110507e-06, "loss": 0.5813, "step": 15445 }, { "epoch": 0.5502965958280635, "grad_norm": 1.1407306877073478, "learning_rate": 8.859512695476876e-06, "loss": 0.4298, "step": 15446 }, { "epoch": 0.5503322229545576, "grad_norm": 1.7372359305069163, "learning_rate": 8.858366295633137e-06, "loss": 0.4855, "step": 15447 }, { "epoch": 0.5503678500810517, "grad_norm": 1.0561755602572513, "learning_rate": 8.85721991099105e-06, "loss": 0.4075, "step": 15448 }, { "epoch": 0.5504034772075458, "grad_norm": 1.0759883746847219, "learning_rate": 8.856073541565876e-06, "loss": 0.4844, "step": 15449 }, { "epoch": 0.55043910433404, "grad_norm": 1.1955997261351778, "learning_rate": 8.854927187372884e-06, "loss": 0.4247, "step": 15450 }, { "epoch": 0.550474731460534, "grad_norm": 1.7342740567868216, "learning_rate": 8.853780848427335e-06, "loss": 0.5837, "step": 15451 }, { "epoch": 0.5505103585870281, "grad_norm": 1.1272829363067973, "learning_rate": 8.852634524744499e-06, "loss": 0.4961, "step": 15452 }, { "epoch": 0.5505459857135223, "grad_norm": 1.0837276034172634, "learning_rate": 8.851488216339632e-06, "loss": 0.4352, "step": 15453 }, { "epoch": 0.5505816128400164, "grad_norm": 1.2479229438509647, "learning_rate": 8.850341923228004e-06, "loss": 0.4142, "step": 15454 }, { "epoch": 0.5506172399665105, "grad_norm": 1.126805739171727, "learning_rate": 8.849195645424877e-06, "loss": 0.4336, "step": 15455 }, { "epoch": 0.5506528670930046, "grad_norm": 1.2020155344436827, "learning_rate": 8.848049382945513e-06, "loss": 0.4549, "step": 15456 }, { "epoch": 0.5506884942194987, "grad_norm": 1.1756232986126833, "learning_rate": 8.84690313580518e-06, "loss": 0.4086, "step": 15457 }, { "epoch": 0.5507241213459928, "grad_norm": 1.1663448581128921, "learning_rate": 8.845756904019132e-06, "loss": 0.4886, "step": 15458 }, { "epoch": 0.550759748472487, "grad_norm": 1.4483435499546127, "learning_rate": 8.844610687602643e-06, "loss": 0.5832, "step": 15459 }, { "epoch": 0.5507953755989811, "grad_norm": 1.6289033719813095, "learning_rate": 8.843464486570967e-06, "loss": 0.6494, "step": 15460 }, { "epoch": 0.5508310027254751, "grad_norm": 1.0467889209000494, "learning_rate": 8.842318300939372e-06, "loss": 0.4363, "step": 15461 }, { "epoch": 0.5508666298519693, "grad_norm": 1.4634203641487067, "learning_rate": 8.841172130723116e-06, "loss": 0.6047, "step": 15462 }, { "epoch": 0.5509022569784634, "grad_norm": 1.0249099449970702, "learning_rate": 8.840025975937467e-06, "loss": 0.4044, "step": 15463 }, { "epoch": 0.5509378841049575, "grad_norm": 1.1691292248701155, "learning_rate": 8.83887983659768e-06, "loss": 0.4274, "step": 15464 }, { "epoch": 0.5509735112314517, "grad_norm": 1.7717615428853337, "learning_rate": 8.83773371271902e-06, "loss": 0.5174, "step": 15465 }, { "epoch": 0.5510091383579457, "grad_norm": 1.1687866880210656, "learning_rate": 8.836587604316746e-06, "loss": 0.4821, "step": 15466 }, { "epoch": 0.5510447654844398, "grad_norm": 1.440615101891584, "learning_rate": 8.83544151140612e-06, "loss": 0.5214, "step": 15467 }, { "epoch": 0.551080392610934, "grad_norm": 1.2229537433036328, "learning_rate": 8.83429543400241e-06, "loss": 0.437, "step": 15468 }, { "epoch": 0.5511160197374281, "grad_norm": 1.1231177313343665, "learning_rate": 8.833149372120867e-06, "loss": 0.453, "step": 15469 }, { "epoch": 0.5511516468639222, "grad_norm": 1.5095074636066448, "learning_rate": 8.832003325776758e-06, "loss": 0.4821, "step": 15470 }, { "epoch": 0.5511872739904163, "grad_norm": 1.5010817734337847, "learning_rate": 8.830857294985337e-06, "loss": 0.4811, "step": 15471 }, { "epoch": 0.5512229011169104, "grad_norm": 1.4181848683431184, "learning_rate": 8.829711279761873e-06, "loss": 0.54, "step": 15472 }, { "epoch": 0.5512585282434045, "grad_norm": 2.1672481906513275, "learning_rate": 8.828565280121619e-06, "loss": 0.6277, "step": 15473 }, { "epoch": 0.5512941553698987, "grad_norm": 1.081813081490037, "learning_rate": 8.82741929607984e-06, "loss": 0.442, "step": 15474 }, { "epoch": 0.5513297824963928, "grad_norm": 1.4126538074015287, "learning_rate": 8.826273327651788e-06, "loss": 0.5625, "step": 15475 }, { "epoch": 0.5513654096228868, "grad_norm": 1.34068454083809, "learning_rate": 8.82512737485273e-06, "loss": 0.4794, "step": 15476 }, { "epoch": 0.551401036749381, "grad_norm": 1.3745588077600066, "learning_rate": 8.823981437697921e-06, "loss": 0.5656, "step": 15477 }, { "epoch": 0.5514366638758751, "grad_norm": 1.2748433988665084, "learning_rate": 8.822835516202619e-06, "loss": 0.4571, "step": 15478 }, { "epoch": 0.5514722910023692, "grad_norm": 1.1218770160646698, "learning_rate": 8.821689610382091e-06, "loss": 0.4328, "step": 15479 }, { "epoch": 0.5515079181288634, "grad_norm": 1.2065981690600034, "learning_rate": 8.820543720251584e-06, "loss": 0.4384, "step": 15480 }, { "epoch": 0.5515435452553574, "grad_norm": 1.1299167961615209, "learning_rate": 8.819397845826365e-06, "loss": 0.5072, "step": 15481 }, { "epoch": 0.5515791723818515, "grad_norm": 1.2093044671402549, "learning_rate": 8.818251987121686e-06, "loss": 0.4437, "step": 15482 }, { "epoch": 0.5516147995083457, "grad_norm": 1.073272920039841, "learning_rate": 8.817106144152812e-06, "loss": 0.4488, "step": 15483 }, { "epoch": 0.5516504266348398, "grad_norm": 1.5858138699463016, "learning_rate": 8.815960316934991e-06, "loss": 0.4552, "step": 15484 }, { "epoch": 0.5516860537613338, "grad_norm": 1.4073598446745892, "learning_rate": 8.814814505483492e-06, "loss": 0.5305, "step": 15485 }, { "epoch": 0.551721680887828, "grad_norm": 1.1819567127269717, "learning_rate": 8.813668709813561e-06, "loss": 0.4406, "step": 15486 }, { "epoch": 0.5517573080143221, "grad_norm": 1.4563475231727012, "learning_rate": 8.812522929940462e-06, "loss": 0.6096, "step": 15487 }, { "epoch": 0.5517929351408162, "grad_norm": 1.1367098125673234, "learning_rate": 8.81137716587945e-06, "loss": 0.4857, "step": 15488 }, { "epoch": 0.5518285622673104, "grad_norm": 1.0547672111116573, "learning_rate": 8.81023141764578e-06, "loss": 0.3909, "step": 15489 }, { "epoch": 0.5518641893938044, "grad_norm": 1.0304631376126707, "learning_rate": 8.809085685254717e-06, "loss": 0.4316, "step": 15490 }, { "epoch": 0.5518998165202985, "grad_norm": 1.471907083740072, "learning_rate": 8.807939968721504e-06, "loss": 0.5346, "step": 15491 }, { "epoch": 0.5519354436467927, "grad_norm": 1.145592493901421, "learning_rate": 8.806794268061407e-06, "loss": 0.4749, "step": 15492 }, { "epoch": 0.5519710707732868, "grad_norm": 1.5159034407242096, "learning_rate": 8.805648583289675e-06, "loss": 0.6028, "step": 15493 }, { "epoch": 0.5520066978997809, "grad_norm": 1.0874316489240181, "learning_rate": 8.804502914421573e-06, "loss": 0.4385, "step": 15494 }, { "epoch": 0.552042325026275, "grad_norm": 1.1510290701782486, "learning_rate": 8.803357261472344e-06, "loss": 0.4031, "step": 15495 }, { "epoch": 0.5520779521527691, "grad_norm": 1.1277414460874453, "learning_rate": 8.802211624457251e-06, "loss": 0.4662, "step": 15496 }, { "epoch": 0.5521135792792632, "grad_norm": 1.3710385790791928, "learning_rate": 8.801066003391547e-06, "loss": 0.5056, "step": 15497 }, { "epoch": 0.5521492064057574, "grad_norm": 1.093602096910449, "learning_rate": 8.799920398290489e-06, "loss": 0.4619, "step": 15498 }, { "epoch": 0.5521848335322515, "grad_norm": 1.6054321114244208, "learning_rate": 8.798774809169328e-06, "loss": 0.5503, "step": 15499 }, { "epoch": 0.5522204606587455, "grad_norm": 1.4632655987844467, "learning_rate": 8.79762923604332e-06, "loss": 0.5434, "step": 15500 }, { "epoch": 0.5522560877852397, "grad_norm": 1.5835808866038408, "learning_rate": 8.796483678927724e-06, "loss": 0.5736, "step": 15501 }, { "epoch": 0.5522917149117338, "grad_norm": 1.0730148302877143, "learning_rate": 8.795338137837783e-06, "loss": 0.4193, "step": 15502 }, { "epoch": 0.5523273420382279, "grad_norm": 1.4101027310367302, "learning_rate": 8.794192612788758e-06, "loss": 0.6451, "step": 15503 }, { "epoch": 0.552362969164722, "grad_norm": 1.4131259796413482, "learning_rate": 8.793047103795902e-06, "loss": 0.5621, "step": 15504 }, { "epoch": 0.5523985962912161, "grad_norm": 1.255932017234222, "learning_rate": 8.791901610874469e-06, "loss": 0.4717, "step": 15505 }, { "epoch": 0.5524342234177102, "grad_norm": 1.1068921708881128, "learning_rate": 8.790756134039708e-06, "loss": 0.4681, "step": 15506 }, { "epoch": 0.5524698505442044, "grad_norm": 1.3974256235775349, "learning_rate": 8.789610673306876e-06, "loss": 0.5137, "step": 15507 }, { "epoch": 0.5525054776706985, "grad_norm": 1.3910568669415802, "learning_rate": 8.788465228691222e-06, "loss": 0.4804, "step": 15508 }, { "epoch": 0.5525411047971925, "grad_norm": 1.1470270338652007, "learning_rate": 8.787319800208004e-06, "loss": 0.4553, "step": 15509 }, { "epoch": 0.5525767319236867, "grad_norm": 1.7725355088008192, "learning_rate": 8.786174387872464e-06, "loss": 0.546, "step": 15510 }, { "epoch": 0.5526123590501808, "grad_norm": 1.2063472743732409, "learning_rate": 8.785028991699867e-06, "loss": 0.4435, "step": 15511 }, { "epoch": 0.5526479861766749, "grad_norm": 1.0035009939043151, "learning_rate": 8.783883611705455e-06, "loss": 0.4598, "step": 15512 }, { "epoch": 0.5526836133031691, "grad_norm": 1.145972938323583, "learning_rate": 8.78273824790448e-06, "loss": 0.4653, "step": 15513 }, { "epoch": 0.5527192404296631, "grad_norm": 1.018283429050377, "learning_rate": 8.781592900312201e-06, "loss": 0.4724, "step": 15514 }, { "epoch": 0.5527548675561572, "grad_norm": 1.3744059400255317, "learning_rate": 8.780447568943862e-06, "loss": 0.5, "step": 15515 }, { "epoch": 0.5527904946826514, "grad_norm": 1.1210018039765783, "learning_rate": 8.779302253814718e-06, "loss": 0.4736, "step": 15516 }, { "epoch": 0.5528261218091455, "grad_norm": 1.654789905979175, "learning_rate": 8.778156954940016e-06, "loss": 0.511, "step": 15517 }, { "epoch": 0.5528617489356396, "grad_norm": 1.75003411295781, "learning_rate": 8.777011672335008e-06, "loss": 0.5377, "step": 15518 }, { "epoch": 0.5528973760621337, "grad_norm": 1.0369154756428272, "learning_rate": 8.775866406014944e-06, "loss": 0.4652, "step": 15519 }, { "epoch": 0.5529330031886278, "grad_norm": 1.8267904157103612, "learning_rate": 8.774721155995078e-06, "loss": 0.5258, "step": 15520 }, { "epoch": 0.5529686303151219, "grad_norm": 1.1640707134474773, "learning_rate": 8.773575922290649e-06, "loss": 0.4366, "step": 15521 }, { "epoch": 0.5530042574416161, "grad_norm": 1.488984678187916, "learning_rate": 8.77243070491692e-06, "loss": 0.5702, "step": 15522 }, { "epoch": 0.5530398845681102, "grad_norm": 1.1909546401951971, "learning_rate": 8.771285503889132e-06, "loss": 0.4764, "step": 15523 }, { "epoch": 0.5530755116946042, "grad_norm": 1.5138900104810242, "learning_rate": 8.770140319222535e-06, "loss": 0.5536, "step": 15524 }, { "epoch": 0.5531111388210984, "grad_norm": 1.496036603331134, "learning_rate": 8.768995150932382e-06, "loss": 0.5606, "step": 15525 }, { "epoch": 0.5531467659475925, "grad_norm": 1.6298607626383956, "learning_rate": 8.767849999033918e-06, "loss": 0.4869, "step": 15526 }, { "epoch": 0.5531823930740866, "grad_norm": 1.136354792360886, "learning_rate": 8.766704863542395e-06, "loss": 0.4441, "step": 15527 }, { "epoch": 0.5532180202005808, "grad_norm": 1.5136857097597227, "learning_rate": 8.765559744473054e-06, "loss": 0.6204, "step": 15528 }, { "epoch": 0.5532536473270748, "grad_norm": 1.798302290988303, "learning_rate": 8.76441464184115e-06, "loss": 0.4542, "step": 15529 }, { "epoch": 0.5532892744535689, "grad_norm": 1.128212468345789, "learning_rate": 8.763269555661928e-06, "loss": 0.4671, "step": 15530 }, { "epoch": 0.5533249015800631, "grad_norm": 2.28915227542171, "learning_rate": 8.76212448595064e-06, "loss": 0.5229, "step": 15531 }, { "epoch": 0.5533605287065572, "grad_norm": 1.0475354700925774, "learning_rate": 8.760979432722525e-06, "loss": 0.492, "step": 15532 }, { "epoch": 0.5533961558330512, "grad_norm": 0.9714769709436142, "learning_rate": 8.759834395992836e-06, "loss": 0.404, "step": 15533 }, { "epoch": 0.5534317829595454, "grad_norm": 2.114699593072292, "learning_rate": 8.758689375776819e-06, "loss": 0.5861, "step": 15534 }, { "epoch": 0.5534674100860395, "grad_norm": 1.0942397437809002, "learning_rate": 8.75754437208972e-06, "loss": 0.4165, "step": 15535 }, { "epoch": 0.5535030372125336, "grad_norm": 1.1461037796941889, "learning_rate": 8.756399384946788e-06, "loss": 0.4478, "step": 15536 }, { "epoch": 0.5535386643390278, "grad_norm": 1.0024861162171264, "learning_rate": 8.755254414363264e-06, "loss": 0.4681, "step": 15537 }, { "epoch": 0.5535742914655218, "grad_norm": 1.6914664088046183, "learning_rate": 8.754109460354403e-06, "loss": 0.52, "step": 15538 }, { "epoch": 0.5536099185920159, "grad_norm": 1.480874617600462, "learning_rate": 8.752964522935442e-06, "loss": 0.5622, "step": 15539 }, { "epoch": 0.5536455457185101, "grad_norm": 0.9899310021555612, "learning_rate": 8.751819602121631e-06, "loss": 0.4512, "step": 15540 }, { "epoch": 0.5536811728450042, "grad_norm": 1.8499953280995893, "learning_rate": 8.750674697928213e-06, "loss": 0.6674, "step": 15541 }, { "epoch": 0.5537167999714983, "grad_norm": 1.829913663853894, "learning_rate": 8.749529810370436e-06, "loss": 0.5435, "step": 15542 }, { "epoch": 0.5537524270979924, "grad_norm": 4.685747897763852, "learning_rate": 8.748384939463543e-06, "loss": 0.5667, "step": 15543 }, { "epoch": 0.5537880542244865, "grad_norm": 1.1331904224478069, "learning_rate": 8.74724008522278e-06, "loss": 0.4831, "step": 15544 }, { "epoch": 0.5538236813509806, "grad_norm": 1.1583197301849506, "learning_rate": 8.746095247663388e-06, "loss": 0.4505, "step": 15545 }, { "epoch": 0.5538593084774748, "grad_norm": 0.9876082566888225, "learning_rate": 8.74495042680062e-06, "loss": 0.4499, "step": 15546 }, { "epoch": 0.5538949356039689, "grad_norm": 1.4161322251238535, "learning_rate": 8.743805622649708e-06, "loss": 0.6257, "step": 15547 }, { "epoch": 0.5539305627304629, "grad_norm": 1.791300706345059, "learning_rate": 8.742660835225902e-06, "loss": 0.5211, "step": 15548 }, { "epoch": 0.5539661898569571, "grad_norm": 1.0707244489229113, "learning_rate": 8.741516064544452e-06, "loss": 0.47, "step": 15549 }, { "epoch": 0.5540018169834512, "grad_norm": 1.7444206158405828, "learning_rate": 8.74037131062059e-06, "loss": 0.5516, "step": 15550 }, { "epoch": 0.5540374441099453, "grad_norm": 1.4500537466827246, "learning_rate": 8.739226573469568e-06, "loss": 0.586, "step": 15551 }, { "epoch": 0.5540730712364395, "grad_norm": 3.3075469526535697, "learning_rate": 8.73808185310662e-06, "loss": 0.5576, "step": 15552 }, { "epoch": 0.5541086983629335, "grad_norm": 1.6722127036389838, "learning_rate": 8.736937149547003e-06, "loss": 0.6096, "step": 15553 }, { "epoch": 0.5541443254894276, "grad_norm": 1.801812335594622, "learning_rate": 8.735792462805943e-06, "loss": 0.5501, "step": 15554 }, { "epoch": 0.5541799526159218, "grad_norm": 1.6213690682472723, "learning_rate": 8.734647792898693e-06, "loss": 0.5655, "step": 15555 }, { "epoch": 0.5542155797424159, "grad_norm": 1.6516065876572437, "learning_rate": 8.733503139840491e-06, "loss": 0.4472, "step": 15556 }, { "epoch": 0.5542512068689099, "grad_norm": 1.198493513956235, "learning_rate": 8.732358503646585e-06, "loss": 0.4765, "step": 15557 }, { "epoch": 0.5542868339954041, "grad_norm": 1.1216435507652283, "learning_rate": 8.731213884332205e-06, "loss": 0.426, "step": 15558 }, { "epoch": 0.5543224611218982, "grad_norm": 1.087887027645326, "learning_rate": 8.730069281912597e-06, "loss": 0.4706, "step": 15559 }, { "epoch": 0.5543580882483923, "grad_norm": 1.1357746834609184, "learning_rate": 8.728924696403011e-06, "loss": 0.4396, "step": 15560 }, { "epoch": 0.5543937153748865, "grad_norm": 1.3838387443440705, "learning_rate": 8.727780127818676e-06, "loss": 0.5537, "step": 15561 }, { "epoch": 0.5544293425013805, "grad_norm": 1.501857181319507, "learning_rate": 8.726635576174843e-06, "loss": 0.5073, "step": 15562 }, { "epoch": 0.5544649696278746, "grad_norm": 1.107235796759687, "learning_rate": 8.725491041486742e-06, "loss": 0.4401, "step": 15563 }, { "epoch": 0.5545005967543688, "grad_norm": 1.5562779847597055, "learning_rate": 8.724346523769624e-06, "loss": 0.5882, "step": 15564 }, { "epoch": 0.5545362238808629, "grad_norm": 3.314684891687375, "learning_rate": 8.723202023038719e-06, "loss": 0.5783, "step": 15565 }, { "epoch": 0.554571851007357, "grad_norm": 1.6455563245982998, "learning_rate": 8.722057539309274e-06, "loss": 0.6675, "step": 15566 }, { "epoch": 0.5546074781338511, "grad_norm": 1.4424869268136356, "learning_rate": 8.720913072596524e-06, "loss": 0.5685, "step": 15567 }, { "epoch": 0.5546431052603452, "grad_norm": 1.414190523891958, "learning_rate": 8.719768622915714e-06, "loss": 0.4646, "step": 15568 }, { "epoch": 0.5546787323868393, "grad_norm": 1.5886081611287433, "learning_rate": 8.718624190282076e-06, "loss": 0.5804, "step": 15569 }, { "epoch": 0.5547143595133335, "grad_norm": 1.5516443359498127, "learning_rate": 8.717479774710853e-06, "loss": 0.5233, "step": 15570 }, { "epoch": 0.5547499866398276, "grad_norm": 1.0735801830606195, "learning_rate": 8.716335376217285e-06, "loss": 0.4424, "step": 15571 }, { "epoch": 0.5547856137663216, "grad_norm": 1.5447825129782637, "learning_rate": 8.715190994816607e-06, "loss": 0.6333, "step": 15572 }, { "epoch": 0.5548212408928158, "grad_norm": 1.426030700360859, "learning_rate": 8.714046630524061e-06, "loss": 0.5338, "step": 15573 }, { "epoch": 0.5548568680193099, "grad_norm": 1.6584048923960881, "learning_rate": 8.712902283354879e-06, "loss": 0.5564, "step": 15574 }, { "epoch": 0.554892495145804, "grad_norm": 1.3322205046129285, "learning_rate": 8.711757953324309e-06, "loss": 0.5218, "step": 15575 }, { "epoch": 0.5549281222722982, "grad_norm": 1.357766789131558, "learning_rate": 8.710613640447578e-06, "loss": 0.4935, "step": 15576 }, { "epoch": 0.5549637493987922, "grad_norm": 1.5036527639971944, "learning_rate": 8.709469344739928e-06, "loss": 0.5242, "step": 15577 }, { "epoch": 0.5549993765252863, "grad_norm": 1.0354043345773947, "learning_rate": 8.708325066216595e-06, "loss": 0.4817, "step": 15578 }, { "epoch": 0.5550350036517805, "grad_norm": 1.8084997274628158, "learning_rate": 8.707180804892822e-06, "loss": 0.5504, "step": 15579 }, { "epoch": 0.5550706307782746, "grad_norm": 1.1348558639623396, "learning_rate": 8.706036560783832e-06, "loss": 0.4496, "step": 15580 }, { "epoch": 0.5551062579047686, "grad_norm": 1.2401982416104658, "learning_rate": 8.704892333904874e-06, "loss": 0.4795, "step": 15581 }, { "epoch": 0.5551418850312628, "grad_norm": 1.1132047781381502, "learning_rate": 8.703748124271179e-06, "loss": 0.4472, "step": 15582 }, { "epoch": 0.5551775121577569, "grad_norm": 1.6203850748647688, "learning_rate": 8.702603931897983e-06, "loss": 0.4424, "step": 15583 }, { "epoch": 0.555213139284251, "grad_norm": 1.201597892982015, "learning_rate": 8.701459756800525e-06, "loss": 0.4644, "step": 15584 }, { "epoch": 0.5552487664107452, "grad_norm": 1.0976784083105346, "learning_rate": 8.700315598994034e-06, "loss": 0.4207, "step": 15585 }, { "epoch": 0.5552843935372392, "grad_norm": 1.0567239990840533, "learning_rate": 8.699171458493754e-06, "loss": 0.4606, "step": 15586 }, { "epoch": 0.5553200206637333, "grad_norm": 1.0833285604725218, "learning_rate": 8.698027335314912e-06, "loss": 0.4263, "step": 15587 }, { "epoch": 0.5553556477902275, "grad_norm": 1.1631517762745927, "learning_rate": 8.696883229472748e-06, "loss": 0.4677, "step": 15588 }, { "epoch": 0.5553912749167216, "grad_norm": 1.4714417833266638, "learning_rate": 8.69573914098249e-06, "loss": 0.55, "step": 15589 }, { "epoch": 0.5554269020432157, "grad_norm": 1.5240085557732732, "learning_rate": 8.694595069859386e-06, "loss": 0.5941, "step": 15590 }, { "epoch": 0.5554625291697098, "grad_norm": 1.1405204343254896, "learning_rate": 8.693451016118654e-06, "loss": 0.4326, "step": 15591 }, { "epoch": 0.5554981562962039, "grad_norm": 2.232201522843417, "learning_rate": 8.692306979775537e-06, "loss": 0.5363, "step": 15592 }, { "epoch": 0.555533783422698, "grad_norm": 1.1187201312817543, "learning_rate": 8.691162960845265e-06, "loss": 0.449, "step": 15593 }, { "epoch": 0.5555694105491922, "grad_norm": 1.3754891412219385, "learning_rate": 8.690018959343071e-06, "loss": 0.4398, "step": 15594 }, { "epoch": 0.5556050376756863, "grad_norm": 1.2212102284234723, "learning_rate": 8.688874975284198e-06, "loss": 0.4707, "step": 15595 }, { "epoch": 0.5556406648021803, "grad_norm": 1.3152858192969001, "learning_rate": 8.687731008683864e-06, "loss": 0.4982, "step": 15596 }, { "epoch": 0.5556762919286745, "grad_norm": 1.1303589971514316, "learning_rate": 8.686587059557313e-06, "loss": 0.4822, "step": 15597 }, { "epoch": 0.5557119190551686, "grad_norm": 1.0783390168272977, "learning_rate": 8.685443127919769e-06, "loss": 0.3933, "step": 15598 }, { "epoch": 0.5557475461816627, "grad_norm": 1.1531646460864349, "learning_rate": 8.684299213786472e-06, "loss": 0.4257, "step": 15599 }, { "epoch": 0.5557831733081569, "grad_norm": 1.1391539473398045, "learning_rate": 8.68315531717265e-06, "loss": 0.478, "step": 15600 }, { "epoch": 0.5558188004346509, "grad_norm": 1.6113360918585768, "learning_rate": 8.682011438093537e-06, "loss": 0.4972, "step": 15601 }, { "epoch": 0.555854427561145, "grad_norm": 1.4028426042086866, "learning_rate": 8.68086757656436e-06, "loss": 0.5018, "step": 15602 }, { "epoch": 0.5558900546876392, "grad_norm": 1.4351207240640451, "learning_rate": 8.679723732600355e-06, "loss": 0.5443, "step": 15603 }, { "epoch": 0.5559256818141333, "grad_norm": 1.1574293858776477, "learning_rate": 8.678579906216748e-06, "loss": 0.3964, "step": 15604 }, { "epoch": 0.5559613089406273, "grad_norm": 0.978534299202155, "learning_rate": 8.677436097428775e-06, "loss": 0.4278, "step": 15605 }, { "epoch": 0.5559969360671215, "grad_norm": 1.343694278187911, "learning_rate": 8.676292306251667e-06, "loss": 0.6038, "step": 15606 }, { "epoch": 0.5560325631936156, "grad_norm": 1.2576953918919553, "learning_rate": 8.675148532700651e-06, "loss": 0.4676, "step": 15607 }, { "epoch": 0.5560681903201097, "grad_norm": 1.377830210560447, "learning_rate": 8.67400477679096e-06, "loss": 0.5486, "step": 15608 }, { "epoch": 0.5561038174466039, "grad_norm": 1.513615500306141, "learning_rate": 8.672861038537818e-06, "loss": 0.6271, "step": 15609 }, { "epoch": 0.5561394445730979, "grad_norm": 1.4278174909512502, "learning_rate": 8.671717317956462e-06, "loss": 0.4835, "step": 15610 }, { "epoch": 0.556175071699592, "grad_norm": 1.0345407518372525, "learning_rate": 8.670573615062115e-06, "loss": 0.4167, "step": 15611 }, { "epoch": 0.5562106988260862, "grad_norm": 1.1489012672381682, "learning_rate": 8.669429929870016e-06, "loss": 0.4706, "step": 15612 }, { "epoch": 0.5562463259525803, "grad_norm": 1.5664977318185551, "learning_rate": 8.66828626239538e-06, "loss": 0.6262, "step": 15613 }, { "epoch": 0.5562819530790744, "grad_norm": 1.1236998975845225, "learning_rate": 8.667142612653449e-06, "loss": 0.4334, "step": 15614 }, { "epoch": 0.5563175802055685, "grad_norm": 1.1712343152778053, "learning_rate": 8.665998980659443e-06, "loss": 0.4213, "step": 15615 }, { "epoch": 0.5563532073320626, "grad_norm": 1.350575315452386, "learning_rate": 8.66485536642859e-06, "loss": 0.5538, "step": 15616 }, { "epoch": 0.5563888344585567, "grad_norm": 1.828019527048667, "learning_rate": 8.663711769976129e-06, "loss": 0.577, "step": 15617 }, { "epoch": 0.5564244615850509, "grad_norm": 1.3883515291657176, "learning_rate": 8.662568191317273e-06, "loss": 0.4582, "step": 15618 }, { "epoch": 0.556460088711545, "grad_norm": 1.4512120493535998, "learning_rate": 8.661424630467258e-06, "loss": 0.5966, "step": 15619 }, { "epoch": 0.556495715838039, "grad_norm": 1.0582637506922916, "learning_rate": 8.66028108744131e-06, "loss": 0.4795, "step": 15620 }, { "epoch": 0.5565313429645332, "grad_norm": 2.638633471845958, "learning_rate": 8.659137562254659e-06, "loss": 0.5772, "step": 15621 }, { "epoch": 0.5565669700910273, "grad_norm": 1.306602968778012, "learning_rate": 8.657994054922522e-06, "loss": 0.4592, "step": 15622 }, { "epoch": 0.5566025972175214, "grad_norm": 1.5911411774010553, "learning_rate": 8.656850565460139e-06, "loss": 0.5559, "step": 15623 }, { "epoch": 0.5566382243440156, "grad_norm": 1.456763858520825, "learning_rate": 8.655707093882723e-06, "loss": 0.5339, "step": 15624 }, { "epoch": 0.5566738514705096, "grad_norm": 1.4466140640956722, "learning_rate": 8.654563640205512e-06, "loss": 0.5603, "step": 15625 }, { "epoch": 0.5567094785970037, "grad_norm": 1.2149318236620847, "learning_rate": 8.653420204443723e-06, "loss": 0.4186, "step": 15626 }, { "epoch": 0.5567451057234979, "grad_norm": 1.4990032002042184, "learning_rate": 8.652276786612584e-06, "loss": 0.5225, "step": 15627 }, { "epoch": 0.556780732849992, "grad_norm": 0.9754792973819401, "learning_rate": 8.651133386727328e-06, "loss": 0.4345, "step": 15628 }, { "epoch": 0.556816359976486, "grad_norm": 1.0721510084626074, "learning_rate": 8.649990004803168e-06, "loss": 0.4634, "step": 15629 }, { "epoch": 0.5568519871029802, "grad_norm": 1.6746928330879058, "learning_rate": 8.648846640855338e-06, "loss": 0.5537, "step": 15630 }, { "epoch": 0.5568876142294743, "grad_norm": 2.1138752044405433, "learning_rate": 8.647703294899057e-06, "loss": 0.55, "step": 15631 }, { "epoch": 0.5569232413559684, "grad_norm": 1.8828442734886242, "learning_rate": 8.646559966949555e-06, "loss": 0.6025, "step": 15632 }, { "epoch": 0.5569588684824626, "grad_norm": 1.6124620443361606, "learning_rate": 8.64541665702205e-06, "loss": 0.5141, "step": 15633 }, { "epoch": 0.5569944956089566, "grad_norm": 1.2171456359369062, "learning_rate": 8.64427336513177e-06, "loss": 0.4664, "step": 15634 }, { "epoch": 0.5570301227354507, "grad_norm": 1.5447696008020622, "learning_rate": 8.643130091293937e-06, "loss": 0.523, "step": 15635 }, { "epoch": 0.5570657498619449, "grad_norm": 1.5289616999164766, "learning_rate": 8.641986835523775e-06, "loss": 0.6167, "step": 15636 }, { "epoch": 0.557101376988439, "grad_norm": 1.156046859877601, "learning_rate": 8.640843597836508e-06, "loss": 0.4496, "step": 15637 }, { "epoch": 0.5571370041149332, "grad_norm": 1.483882444326075, "learning_rate": 8.639700378247362e-06, "loss": 0.5101, "step": 15638 }, { "epoch": 0.5571726312414272, "grad_norm": 1.1301999784371044, "learning_rate": 8.63855717677155e-06, "loss": 0.4441, "step": 15639 }, { "epoch": 0.5572082583679213, "grad_norm": 2.442459454217726, "learning_rate": 8.637413993424303e-06, "loss": 0.5407, "step": 15640 }, { "epoch": 0.5572438854944155, "grad_norm": 1.0772621575846866, "learning_rate": 8.636270828220844e-06, "loss": 0.4398, "step": 15641 }, { "epoch": 0.5572795126209096, "grad_norm": 1.2031127837920228, "learning_rate": 8.635127681176387e-06, "loss": 0.5779, "step": 15642 }, { "epoch": 0.5573151397474037, "grad_norm": 1.1038249446548085, "learning_rate": 8.633984552306165e-06, "loss": 0.4026, "step": 15643 }, { "epoch": 0.5573507668738978, "grad_norm": 1.4802586499479367, "learning_rate": 8.63284144162539e-06, "loss": 0.576, "step": 15644 }, { "epoch": 0.5573863940003919, "grad_norm": 1.9944755434801393, "learning_rate": 8.631698349149287e-06, "loss": 0.4577, "step": 15645 }, { "epoch": 0.557422021126886, "grad_norm": 1.1270789973643063, "learning_rate": 8.630555274893076e-06, "loss": 0.4759, "step": 15646 }, { "epoch": 0.5574576482533802, "grad_norm": 1.07977266580389, "learning_rate": 8.629412218871982e-06, "loss": 0.3849, "step": 15647 }, { "epoch": 0.5574932753798743, "grad_norm": 1.3773167945920557, "learning_rate": 8.628269181101216e-06, "loss": 0.5152, "step": 15648 }, { "epoch": 0.5575289025063683, "grad_norm": 1.4616076171216446, "learning_rate": 8.627126161596012e-06, "loss": 0.5414, "step": 15649 }, { "epoch": 0.5575645296328625, "grad_norm": 1.6917478050640302, "learning_rate": 8.625983160371579e-06, "loss": 0.5707, "step": 15650 }, { "epoch": 0.5576001567593566, "grad_norm": 1.1035740517594737, "learning_rate": 8.624840177443138e-06, "loss": 0.4338, "step": 15651 }, { "epoch": 0.5576357838858507, "grad_norm": 1.0429580659755078, "learning_rate": 8.623697212825917e-06, "loss": 0.4504, "step": 15652 }, { "epoch": 0.5576714110123449, "grad_norm": 1.2836156603967062, "learning_rate": 8.622554266535124e-06, "loss": 0.4789, "step": 15653 }, { "epoch": 0.5577070381388389, "grad_norm": 1.339296232604492, "learning_rate": 8.62141133858599e-06, "loss": 0.5598, "step": 15654 }, { "epoch": 0.557742665265333, "grad_norm": 1.110322080983372, "learning_rate": 8.620268428993723e-06, "loss": 0.4794, "step": 15655 }, { "epoch": 0.5577782923918272, "grad_norm": 1.9513354475118865, "learning_rate": 8.619125537773548e-06, "loss": 0.6094, "step": 15656 }, { "epoch": 0.5578139195183213, "grad_norm": 1.8245546546759384, "learning_rate": 8.61798266494068e-06, "loss": 0.5416, "step": 15657 }, { "epoch": 0.5578495466448153, "grad_norm": 1.1421253625878618, "learning_rate": 8.616839810510342e-06, "loss": 0.4496, "step": 15658 }, { "epoch": 0.5578851737713095, "grad_norm": 1.0798746977540612, "learning_rate": 8.615696974497745e-06, "loss": 0.4122, "step": 15659 }, { "epoch": 0.5579208008978036, "grad_norm": 1.527321348954088, "learning_rate": 8.614554156918113e-06, "loss": 0.5937, "step": 15660 }, { "epoch": 0.5579564280242977, "grad_norm": 1.0543853754594257, "learning_rate": 8.613411357786659e-06, "loss": 0.439, "step": 15661 }, { "epoch": 0.5579920551507919, "grad_norm": 1.3941236722606594, "learning_rate": 8.6122685771186e-06, "loss": 0.5395, "step": 15662 }, { "epoch": 0.5580276822772859, "grad_norm": 1.0356638551841149, "learning_rate": 8.611125814929158e-06, "loss": 0.4123, "step": 15663 }, { "epoch": 0.55806330940378, "grad_norm": 1.1624113539481242, "learning_rate": 8.609983071233545e-06, "loss": 0.4472, "step": 15664 }, { "epoch": 0.5580989365302742, "grad_norm": 1.199231924541965, "learning_rate": 8.608840346046983e-06, "loss": 0.4291, "step": 15665 }, { "epoch": 0.5581345636567683, "grad_norm": 1.106361037395401, "learning_rate": 8.60769763938468e-06, "loss": 0.4181, "step": 15666 }, { "epoch": 0.5581701907832624, "grad_norm": 1.0547420256573712, "learning_rate": 8.606554951261858e-06, "loss": 0.4789, "step": 15667 }, { "epoch": 0.5582058179097565, "grad_norm": 1.1840351610598654, "learning_rate": 8.605412281693727e-06, "loss": 0.434, "step": 15668 }, { "epoch": 0.5582414450362506, "grad_norm": 1.2766169766134858, "learning_rate": 8.604269630695513e-06, "loss": 0.4541, "step": 15669 }, { "epoch": 0.5582770721627447, "grad_norm": 1.727767453639328, "learning_rate": 8.60312699828242e-06, "loss": 0.538, "step": 15670 }, { "epoch": 0.5583126992892389, "grad_norm": 1.451803361903897, "learning_rate": 8.601984384469669e-06, "loss": 0.5086, "step": 15671 }, { "epoch": 0.558348326415733, "grad_norm": 1.2685984400858556, "learning_rate": 8.600841789272469e-06, "loss": 0.592, "step": 15672 }, { "epoch": 0.558383953542227, "grad_norm": 1.0658339085535262, "learning_rate": 8.599699212706045e-06, "loss": 0.4465, "step": 15673 }, { "epoch": 0.5584195806687212, "grad_norm": 1.5774006658671966, "learning_rate": 8.598556654785598e-06, "loss": 0.5183, "step": 15674 }, { "epoch": 0.5584552077952153, "grad_norm": 1.170172584022626, "learning_rate": 8.59741411552635e-06, "loss": 0.4538, "step": 15675 }, { "epoch": 0.5584908349217094, "grad_norm": 1.525711912072386, "learning_rate": 8.59627159494352e-06, "loss": 0.5294, "step": 15676 }, { "epoch": 0.5585264620482036, "grad_norm": 1.2521482419369192, "learning_rate": 8.595129093052308e-06, "loss": 0.4309, "step": 15677 }, { "epoch": 0.5585620891746976, "grad_norm": 1.125747542310743, "learning_rate": 8.593986609867936e-06, "loss": 0.4094, "step": 15678 }, { "epoch": 0.5585977163011917, "grad_norm": 1.1377794357758186, "learning_rate": 8.592844145405615e-06, "loss": 0.4441, "step": 15679 }, { "epoch": 0.5586333434276859, "grad_norm": 1.382965201430341, "learning_rate": 8.591701699680562e-06, "loss": 0.444, "step": 15680 }, { "epoch": 0.55866897055418, "grad_norm": 1.6325980861703397, "learning_rate": 8.590559272707978e-06, "loss": 0.5515, "step": 15681 }, { "epoch": 0.558704597680674, "grad_norm": 1.2457906881111205, "learning_rate": 8.589416864503088e-06, "loss": 0.4657, "step": 15682 }, { "epoch": 0.5587402248071682, "grad_norm": 1.1367914551300513, "learning_rate": 8.588274475081094e-06, "loss": 0.4684, "step": 15683 }, { "epoch": 0.5587758519336623, "grad_norm": 1.4258402134217403, "learning_rate": 8.587132104457216e-06, "loss": 0.488, "step": 15684 }, { "epoch": 0.5588114790601564, "grad_norm": 1.8687520754602907, "learning_rate": 8.58598975264666e-06, "loss": 0.5623, "step": 15685 }, { "epoch": 0.5588471061866506, "grad_norm": 1.7682069470298478, "learning_rate": 8.584847419664635e-06, "loss": 0.5785, "step": 15686 }, { "epoch": 0.5588827333131446, "grad_norm": 1.2225126539721083, "learning_rate": 8.583705105526362e-06, "loss": 0.4497, "step": 15687 }, { "epoch": 0.5589183604396387, "grad_norm": 1.5420148203367283, "learning_rate": 8.58256281024704e-06, "loss": 0.6037, "step": 15688 }, { "epoch": 0.5589539875661329, "grad_norm": 1.4738795777384537, "learning_rate": 8.58142053384189e-06, "loss": 0.6065, "step": 15689 }, { "epoch": 0.558989614692627, "grad_norm": 1.1805537680157678, "learning_rate": 8.580278276326113e-06, "loss": 0.4801, "step": 15690 }, { "epoch": 0.559025241819121, "grad_norm": 1.1850096840191533, "learning_rate": 8.579136037714929e-06, "loss": 0.471, "step": 15691 }, { "epoch": 0.5590608689456152, "grad_norm": 1.4180832520568118, "learning_rate": 8.577993818023535e-06, "loss": 0.5127, "step": 15692 }, { "epoch": 0.5590964960721093, "grad_norm": 1.7142463796474832, "learning_rate": 8.576851617267151e-06, "loss": 0.5864, "step": 15693 }, { "epoch": 0.5591321231986034, "grad_norm": 1.0921648350652602, "learning_rate": 8.57570943546098e-06, "loss": 0.4613, "step": 15694 }, { "epoch": 0.5591677503250976, "grad_norm": 1.1776375188308195, "learning_rate": 8.574567272620238e-06, "loss": 0.3838, "step": 15695 }, { "epoch": 0.5592033774515917, "grad_norm": 1.1029252842948016, "learning_rate": 8.573425128760124e-06, "loss": 0.4275, "step": 15696 }, { "epoch": 0.5592390045780857, "grad_norm": 1.3526346514231355, "learning_rate": 8.57228300389585e-06, "loss": 0.4671, "step": 15697 }, { "epoch": 0.5592746317045799, "grad_norm": 1.5690135172807333, "learning_rate": 8.571140898042631e-06, "loss": 0.4596, "step": 15698 }, { "epoch": 0.559310258831074, "grad_norm": 1.6598610887433358, "learning_rate": 8.569998811215667e-06, "loss": 0.5661, "step": 15699 }, { "epoch": 0.5593458859575681, "grad_norm": 1.300219419032182, "learning_rate": 8.568856743430169e-06, "loss": 0.4955, "step": 15700 }, { "epoch": 0.5593815130840623, "grad_norm": 1.034917124210832, "learning_rate": 8.567714694701344e-06, "loss": 0.3999, "step": 15701 }, { "epoch": 0.5594171402105563, "grad_norm": 1.4310040129737638, "learning_rate": 8.566572665044402e-06, "loss": 0.5982, "step": 15702 }, { "epoch": 0.5594527673370504, "grad_norm": 1.030170571458625, "learning_rate": 8.565430654474542e-06, "loss": 0.4127, "step": 15703 }, { "epoch": 0.5594883944635446, "grad_norm": 1.7634477349677509, "learning_rate": 8.564288663006977e-06, "loss": 0.4886, "step": 15704 }, { "epoch": 0.5595240215900387, "grad_norm": 1.364019556712691, "learning_rate": 8.563146690656911e-06, "loss": 0.5811, "step": 15705 }, { "epoch": 0.5595596487165327, "grad_norm": 1.6937347129817182, "learning_rate": 8.562004737439555e-06, "loss": 0.4913, "step": 15706 }, { "epoch": 0.5595952758430269, "grad_norm": 1.2577049015576265, "learning_rate": 8.560862803370107e-06, "loss": 0.4777, "step": 15707 }, { "epoch": 0.559630902969521, "grad_norm": 1.4529605895418434, "learning_rate": 8.559720888463777e-06, "loss": 0.5259, "step": 15708 }, { "epoch": 0.5596665300960151, "grad_norm": 1.3309431763794612, "learning_rate": 8.558578992735771e-06, "loss": 0.5039, "step": 15709 }, { "epoch": 0.5597021572225093, "grad_norm": 1.167932309628412, "learning_rate": 8.557437116201292e-06, "loss": 0.5018, "step": 15710 }, { "epoch": 0.5597377843490033, "grad_norm": 1.2755479486523849, "learning_rate": 8.556295258875548e-06, "loss": 0.436, "step": 15711 }, { "epoch": 0.5597734114754974, "grad_norm": 1.022543873521741, "learning_rate": 8.555153420773739e-06, "loss": 0.4211, "step": 15712 }, { "epoch": 0.5598090386019916, "grad_norm": 1.4784056079482597, "learning_rate": 8.554011601911077e-06, "loss": 0.5603, "step": 15713 }, { "epoch": 0.5598446657284857, "grad_norm": 1.0803012424238325, "learning_rate": 8.552869802302756e-06, "loss": 0.4749, "step": 15714 }, { "epoch": 0.5598802928549798, "grad_norm": 1.2200616749883744, "learning_rate": 8.55172802196399e-06, "loss": 0.4826, "step": 15715 }, { "epoch": 0.5599159199814739, "grad_norm": 1.4463292444396556, "learning_rate": 8.550586260909974e-06, "loss": 0.5278, "step": 15716 }, { "epoch": 0.559951547107968, "grad_norm": 1.1731011211786757, "learning_rate": 8.549444519155918e-06, "loss": 0.4645, "step": 15717 }, { "epoch": 0.5599871742344621, "grad_norm": 1.059084260519894, "learning_rate": 8.548302796717018e-06, "loss": 0.398, "step": 15718 }, { "epoch": 0.5600228013609563, "grad_norm": 1.456132800492755, "learning_rate": 8.547161093608484e-06, "loss": 0.4672, "step": 15719 }, { "epoch": 0.5600584284874504, "grad_norm": 1.0484206184294593, "learning_rate": 8.546019409845514e-06, "loss": 0.4188, "step": 15720 }, { "epoch": 0.5600940556139444, "grad_norm": 1.0296675328029674, "learning_rate": 8.544877745443311e-06, "loss": 0.4742, "step": 15721 }, { "epoch": 0.5601296827404386, "grad_norm": 1.2419507815534643, "learning_rate": 8.543736100417083e-06, "loss": 0.4342, "step": 15722 }, { "epoch": 0.5601653098669327, "grad_norm": 1.6475539833296422, "learning_rate": 8.542594474782018e-06, "loss": 0.6061, "step": 15723 }, { "epoch": 0.5602009369934268, "grad_norm": 1.3795725340618348, "learning_rate": 8.541452868553335e-06, "loss": 0.5236, "step": 15724 }, { "epoch": 0.560236564119921, "grad_norm": 2.017624005033094, "learning_rate": 8.540311281746222e-06, "loss": 0.5647, "step": 15725 }, { "epoch": 0.560272191246415, "grad_norm": 1.4700985328216125, "learning_rate": 8.539169714375885e-06, "loss": 0.5078, "step": 15726 }, { "epoch": 0.5603078183729091, "grad_norm": 1.44257847022993, "learning_rate": 8.538028166457524e-06, "loss": 0.5224, "step": 15727 }, { "epoch": 0.5603434454994033, "grad_norm": 1.8813423665418378, "learning_rate": 8.536886638006344e-06, "loss": 0.5536, "step": 15728 }, { "epoch": 0.5603790726258974, "grad_norm": 1.248974264589586, "learning_rate": 8.535745129037535e-06, "loss": 0.4338, "step": 15729 }, { "epoch": 0.5604146997523914, "grad_norm": 1.9662302688271953, "learning_rate": 8.534603639566307e-06, "loss": 0.5332, "step": 15730 }, { "epoch": 0.5604503268788856, "grad_norm": 1.8931597602538408, "learning_rate": 8.533462169607853e-06, "loss": 0.4866, "step": 15731 }, { "epoch": 0.5604859540053797, "grad_norm": 1.0491641543040258, "learning_rate": 8.532320719177375e-06, "loss": 0.4671, "step": 15732 }, { "epoch": 0.5605215811318738, "grad_norm": 1.3001853368606862, "learning_rate": 8.531179288290077e-06, "loss": 0.4898, "step": 15733 }, { "epoch": 0.560557208258368, "grad_norm": 1.4006954617807403, "learning_rate": 8.53003787696115e-06, "loss": 0.4888, "step": 15734 }, { "epoch": 0.560592835384862, "grad_norm": 1.1243691316317033, "learning_rate": 8.528896485205796e-06, "loss": 0.4498, "step": 15735 }, { "epoch": 0.5606284625113561, "grad_norm": 1.5028971306296206, "learning_rate": 8.527755113039213e-06, "loss": 0.6337, "step": 15736 }, { "epoch": 0.5606640896378503, "grad_norm": 1.7894881817747774, "learning_rate": 8.526613760476601e-06, "loss": 0.601, "step": 15737 }, { "epoch": 0.5606997167643444, "grad_norm": 1.3683427410121525, "learning_rate": 8.525472427533156e-06, "loss": 0.5426, "step": 15738 }, { "epoch": 0.5607353438908385, "grad_norm": 1.2618442436001838, "learning_rate": 8.524331114224079e-06, "loss": 0.4538, "step": 15739 }, { "epoch": 0.5607709710173326, "grad_norm": 1.054838949137111, "learning_rate": 8.523189820564562e-06, "loss": 0.4518, "step": 15740 }, { "epoch": 0.5608065981438267, "grad_norm": 1.3513240799317905, "learning_rate": 8.522048546569805e-06, "loss": 0.5707, "step": 15741 }, { "epoch": 0.5608422252703208, "grad_norm": 1.9021836893501012, "learning_rate": 8.520907292255004e-06, "loss": 0.5633, "step": 15742 }, { "epoch": 0.560877852396815, "grad_norm": 1.1871203093548688, "learning_rate": 8.519766057635355e-06, "loss": 0.4527, "step": 15743 }, { "epoch": 0.560913479523309, "grad_norm": 1.0957833062064044, "learning_rate": 8.518624842726062e-06, "loss": 0.4327, "step": 15744 }, { "epoch": 0.5609491066498031, "grad_norm": 1.548802832317476, "learning_rate": 8.517483647542307e-06, "loss": 0.5689, "step": 15745 }, { "epoch": 0.5609847337762973, "grad_norm": 1.4573493138149107, "learning_rate": 8.516342472099299e-06, "loss": 0.5486, "step": 15746 }, { "epoch": 0.5610203609027914, "grad_norm": 1.135743108740294, "learning_rate": 8.515201316412222e-06, "loss": 0.4131, "step": 15747 }, { "epoch": 0.5610559880292855, "grad_norm": 1.0946443065383848, "learning_rate": 8.514060180496285e-06, "loss": 0.3957, "step": 15748 }, { "epoch": 0.5610916151557797, "grad_norm": 1.3454750689358816, "learning_rate": 8.512919064366669e-06, "loss": 0.4962, "step": 15749 }, { "epoch": 0.5611272422822737, "grad_norm": 1.1059402899929305, "learning_rate": 8.511777968038579e-06, "loss": 0.448, "step": 15750 }, { "epoch": 0.5611628694087678, "grad_norm": 1.0088286066985577, "learning_rate": 8.510636891527203e-06, "loss": 0.4102, "step": 15751 }, { "epoch": 0.561198496535262, "grad_norm": 1.436017485233007, "learning_rate": 8.509495834847738e-06, "loss": 0.4641, "step": 15752 }, { "epoch": 0.5612341236617561, "grad_norm": 1.219662730217976, "learning_rate": 8.508354798015378e-06, "loss": 0.4969, "step": 15753 }, { "epoch": 0.5612697507882501, "grad_norm": 1.0970703158716832, "learning_rate": 8.507213781045316e-06, "loss": 0.4233, "step": 15754 }, { "epoch": 0.5613053779147443, "grad_norm": 1.4909091864265542, "learning_rate": 8.50607278395275e-06, "loss": 0.5401, "step": 15755 }, { "epoch": 0.5613410050412384, "grad_norm": 1.7781991977892453, "learning_rate": 8.504931806752865e-06, "loss": 0.5283, "step": 15756 }, { "epoch": 0.5613766321677325, "grad_norm": 1.246691333924612, "learning_rate": 8.50379084946086e-06, "loss": 0.4397, "step": 15757 }, { "epoch": 0.5614122592942267, "grad_norm": 1.0889542621495365, "learning_rate": 8.502649912091921e-06, "loss": 0.45, "step": 15758 }, { "epoch": 0.5614478864207207, "grad_norm": 1.681029025392066, "learning_rate": 8.501508994661252e-06, "loss": 0.4796, "step": 15759 }, { "epoch": 0.5614835135472148, "grad_norm": 1.2154463941555216, "learning_rate": 8.500368097184034e-06, "loss": 0.4103, "step": 15760 }, { "epoch": 0.561519140673709, "grad_norm": 1.6330714433517706, "learning_rate": 8.499227219675463e-06, "loss": 0.5359, "step": 15761 }, { "epoch": 0.5615547678002031, "grad_norm": 1.0392565014653272, "learning_rate": 8.49808636215073e-06, "loss": 0.4096, "step": 15762 }, { "epoch": 0.5615903949266972, "grad_norm": 1.4744725840309658, "learning_rate": 8.49694552462503e-06, "loss": 0.5458, "step": 15763 }, { "epoch": 0.5616260220531913, "grad_norm": 1.0139701451538732, "learning_rate": 8.495804707113547e-06, "loss": 0.3905, "step": 15764 }, { "epoch": 0.5616616491796854, "grad_norm": 1.4930509451587248, "learning_rate": 8.49466390963148e-06, "loss": 0.6015, "step": 15765 }, { "epoch": 0.5616972763061795, "grad_norm": 1.4608207386273069, "learning_rate": 8.493523132194012e-06, "loss": 0.4587, "step": 15766 }, { "epoch": 0.5617329034326737, "grad_norm": 0.981069982233121, "learning_rate": 8.492382374816334e-06, "loss": 0.3978, "step": 15767 }, { "epoch": 0.5617685305591678, "grad_norm": 1.1870913606585343, "learning_rate": 8.491241637513644e-06, "loss": 0.4465, "step": 15768 }, { "epoch": 0.5618041576856618, "grad_norm": 1.2580928438108394, "learning_rate": 8.490100920301121e-06, "loss": 0.4956, "step": 15769 }, { "epoch": 0.561839784812156, "grad_norm": 2.4951872004974596, "learning_rate": 8.488960223193968e-06, "loss": 0.4351, "step": 15770 }, { "epoch": 0.5618754119386501, "grad_norm": 1.370664832772056, "learning_rate": 8.487819546207357e-06, "loss": 0.5353, "step": 15771 }, { "epoch": 0.5619110390651442, "grad_norm": 1.6071222206986653, "learning_rate": 8.48667888935649e-06, "loss": 0.5205, "step": 15772 }, { "epoch": 0.5619466661916384, "grad_norm": 1.1259387818205193, "learning_rate": 8.48553825265655e-06, "loss": 0.4891, "step": 15773 }, { "epoch": 0.5619822933181324, "grad_norm": 1.5911127940069565, "learning_rate": 8.484397636122727e-06, "loss": 0.4702, "step": 15774 }, { "epoch": 0.5620179204446265, "grad_norm": 1.585203345518728, "learning_rate": 8.483257039770207e-06, "loss": 0.5118, "step": 15775 }, { "epoch": 0.5620535475711207, "grad_norm": 1.5778903249333274, "learning_rate": 8.482116463614185e-06, "loss": 0.5716, "step": 15776 }, { "epoch": 0.5620891746976148, "grad_norm": 1.1368435286664593, "learning_rate": 8.480975907669838e-06, "loss": 0.4675, "step": 15777 }, { "epoch": 0.5621248018241088, "grad_norm": 1.1832166330669498, "learning_rate": 8.479835371952359e-06, "loss": 0.4659, "step": 15778 }, { "epoch": 0.562160428950603, "grad_norm": 1.5126272219803414, "learning_rate": 8.478694856476937e-06, "loss": 0.5824, "step": 15779 }, { "epoch": 0.5621960560770971, "grad_norm": 1.6878305105930371, "learning_rate": 8.477554361258753e-06, "loss": 0.5584, "step": 15780 }, { "epoch": 0.5622316832035912, "grad_norm": 1.5572851050189624, "learning_rate": 8.476413886313002e-06, "loss": 0.525, "step": 15781 }, { "epoch": 0.5622673103300854, "grad_norm": 1.1247441159951448, "learning_rate": 8.47527343165486e-06, "loss": 0.4352, "step": 15782 }, { "epoch": 0.5623029374565794, "grad_norm": 1.1567359257659828, "learning_rate": 8.474132997299521e-06, "loss": 0.407, "step": 15783 }, { "epoch": 0.5623385645830735, "grad_norm": 1.751107680430348, "learning_rate": 8.472992583262166e-06, "loss": 0.559, "step": 15784 }, { "epoch": 0.5623741917095677, "grad_norm": 1.046753718815462, "learning_rate": 8.471852189557986e-06, "loss": 0.4538, "step": 15785 }, { "epoch": 0.5624098188360618, "grad_norm": 1.3813530805187622, "learning_rate": 8.470711816202154e-06, "loss": 0.5307, "step": 15786 }, { "epoch": 0.5624454459625559, "grad_norm": 1.6391092078194174, "learning_rate": 8.46957146320987e-06, "loss": 0.5485, "step": 15787 }, { "epoch": 0.56248107308905, "grad_norm": 1.5335548724746846, "learning_rate": 8.468431130596309e-06, "loss": 0.642, "step": 15788 }, { "epoch": 0.5625167002155441, "grad_norm": 1.2470282948965918, "learning_rate": 8.467290818376658e-06, "loss": 0.4581, "step": 15789 }, { "epoch": 0.5625523273420382, "grad_norm": 1.1507225778587415, "learning_rate": 8.4661505265661e-06, "loss": 0.4462, "step": 15790 }, { "epoch": 0.5625879544685324, "grad_norm": 1.142513492104062, "learning_rate": 8.465010255179822e-06, "loss": 0.4151, "step": 15791 }, { "epoch": 0.5626235815950265, "grad_norm": 1.6999259780457514, "learning_rate": 8.463870004233008e-06, "loss": 0.5459, "step": 15792 }, { "epoch": 0.5626592087215205, "grad_norm": 2.305342153482282, "learning_rate": 8.462729773740833e-06, "loss": 0.5784, "step": 15793 }, { "epoch": 0.5626948358480147, "grad_norm": 1.2111442770221272, "learning_rate": 8.461589563718489e-06, "loss": 0.4009, "step": 15794 }, { "epoch": 0.5627304629745088, "grad_norm": 1.5145346056358, "learning_rate": 8.460449374181152e-06, "loss": 0.4864, "step": 15795 }, { "epoch": 0.5627660901010029, "grad_norm": 1.4125569273805936, "learning_rate": 8.459309205144014e-06, "loss": 0.5021, "step": 15796 }, { "epoch": 0.5628017172274971, "grad_norm": 1.1495817477601953, "learning_rate": 8.458169056622245e-06, "loss": 0.4545, "step": 15797 }, { "epoch": 0.5628373443539911, "grad_norm": 1.1876671196188282, "learning_rate": 8.457028928631035e-06, "loss": 0.4473, "step": 15798 }, { "epoch": 0.5628729714804852, "grad_norm": 1.1766453395026615, "learning_rate": 8.455888821185561e-06, "loss": 0.4536, "step": 15799 }, { "epoch": 0.5629085986069794, "grad_norm": 1.1825839998127452, "learning_rate": 8.454748734301009e-06, "loss": 0.4101, "step": 15800 }, { "epoch": 0.5629442257334735, "grad_norm": 1.5209144634574774, "learning_rate": 8.453608667992556e-06, "loss": 0.4647, "step": 15801 }, { "epoch": 0.5629798528599675, "grad_norm": 1.5070177135699214, "learning_rate": 8.452468622275384e-06, "loss": 0.5689, "step": 15802 }, { "epoch": 0.5630154799864617, "grad_norm": 1.7007049753143635, "learning_rate": 8.451328597164679e-06, "loss": 0.571, "step": 15803 }, { "epoch": 0.5630511071129558, "grad_norm": 1.1319643377093451, "learning_rate": 8.45018859267561e-06, "loss": 0.4416, "step": 15804 }, { "epoch": 0.5630867342394499, "grad_norm": 1.3450727298639145, "learning_rate": 8.449048608823367e-06, "loss": 0.492, "step": 15805 }, { "epoch": 0.5631223613659441, "grad_norm": 2.3394076752650084, "learning_rate": 8.447908645623121e-06, "loss": 0.529, "step": 15806 }, { "epoch": 0.5631579884924381, "grad_norm": 1.4365356989186897, "learning_rate": 8.446768703090065e-06, "loss": 0.5372, "step": 15807 }, { "epoch": 0.5631936156189322, "grad_norm": 1.4409122071304636, "learning_rate": 8.44562878123936e-06, "loss": 0.5152, "step": 15808 }, { "epoch": 0.5632292427454264, "grad_norm": 1.1046403107490361, "learning_rate": 8.444488880086201e-06, "loss": 0.4512, "step": 15809 }, { "epoch": 0.5632648698719205, "grad_norm": 1.5457641045782025, "learning_rate": 8.443348999645756e-06, "loss": 0.5414, "step": 15810 }, { "epoch": 0.5633004969984146, "grad_norm": 1.3627311756647948, "learning_rate": 8.44220913993321e-06, "loss": 0.5084, "step": 15811 }, { "epoch": 0.5633361241249087, "grad_norm": 1.2145947395179901, "learning_rate": 8.441069300963733e-06, "loss": 0.4646, "step": 15812 }, { "epoch": 0.5633717512514028, "grad_norm": 1.704299723807779, "learning_rate": 8.43992948275251e-06, "loss": 0.5102, "step": 15813 }, { "epoch": 0.5634073783778969, "grad_norm": 1.4878267255630675, "learning_rate": 8.438789685314721e-06, "loss": 0.6081, "step": 15814 }, { "epoch": 0.5634430055043911, "grad_norm": 1.962370876924343, "learning_rate": 8.437649908665532e-06, "loss": 0.5962, "step": 15815 }, { "epoch": 0.5634786326308852, "grad_norm": 2.0764173617673176, "learning_rate": 8.436510152820133e-06, "loss": 0.5086, "step": 15816 }, { "epoch": 0.5635142597573792, "grad_norm": 2.4267520782944154, "learning_rate": 8.435370417793687e-06, "loss": 0.5566, "step": 15817 }, { "epoch": 0.5635498868838734, "grad_norm": 1.7966719427123856, "learning_rate": 8.434230703601386e-06, "loss": 0.6091, "step": 15818 }, { "epoch": 0.5635855140103675, "grad_norm": 1.2231818370568885, "learning_rate": 8.43309101025839e-06, "loss": 0.4161, "step": 15819 }, { "epoch": 0.5636211411368616, "grad_norm": 1.171826118441964, "learning_rate": 8.431951337779886e-06, "loss": 0.4614, "step": 15820 }, { "epoch": 0.5636567682633558, "grad_norm": 1.1316929906386792, "learning_rate": 8.430811686181044e-06, "loss": 0.4835, "step": 15821 }, { "epoch": 0.5636923953898498, "grad_norm": 1.0493338376101373, "learning_rate": 8.429672055477046e-06, "loss": 0.4682, "step": 15822 }, { "epoch": 0.5637280225163439, "grad_norm": 0.9653321833676065, "learning_rate": 8.428532445683058e-06, "loss": 0.3767, "step": 15823 }, { "epoch": 0.5637636496428381, "grad_norm": 1.3491579402568532, "learning_rate": 8.427392856814255e-06, "loss": 0.5599, "step": 15824 }, { "epoch": 0.5637992767693322, "grad_norm": 1.0454819778486077, "learning_rate": 8.426253288885822e-06, "loss": 0.4444, "step": 15825 }, { "epoch": 0.5638349038958262, "grad_norm": 1.161953643891603, "learning_rate": 8.425113741912922e-06, "loss": 0.4738, "step": 15826 }, { "epoch": 0.5638705310223204, "grad_norm": 1.1089352580396117, "learning_rate": 8.423974215910738e-06, "loss": 0.4251, "step": 15827 }, { "epoch": 0.5639061581488145, "grad_norm": 1.1040666663574026, "learning_rate": 8.422834710894434e-06, "loss": 0.4526, "step": 15828 }, { "epoch": 0.5639417852753086, "grad_norm": 1.177477333281314, "learning_rate": 8.421695226879194e-06, "loss": 0.4108, "step": 15829 }, { "epoch": 0.5639774124018028, "grad_norm": 1.7598615680845522, "learning_rate": 8.42055576388018e-06, "loss": 0.4958, "step": 15830 }, { "epoch": 0.5640130395282968, "grad_norm": 1.533913624295905, "learning_rate": 8.419416321912573e-06, "loss": 0.5851, "step": 15831 }, { "epoch": 0.5640486666547909, "grad_norm": 1.103491639889848, "learning_rate": 8.41827690099154e-06, "loss": 0.4786, "step": 15832 }, { "epoch": 0.5640842937812851, "grad_norm": 1.1733707247588838, "learning_rate": 8.41713750113226e-06, "loss": 0.4519, "step": 15833 }, { "epoch": 0.5641199209077792, "grad_norm": 1.3447243736503471, "learning_rate": 8.415998122349896e-06, "loss": 0.5615, "step": 15834 }, { "epoch": 0.5641555480342733, "grad_norm": 1.133652767509468, "learning_rate": 8.414858764659623e-06, "loss": 0.4813, "step": 15835 }, { "epoch": 0.5641911751607674, "grad_norm": 1.2488299045327351, "learning_rate": 8.413719428076617e-06, "loss": 0.4668, "step": 15836 }, { "epoch": 0.5642268022872615, "grad_norm": 1.1309844328462562, "learning_rate": 8.412580112616044e-06, "loss": 0.4197, "step": 15837 }, { "epoch": 0.5642624294137556, "grad_norm": 1.5004387416213187, "learning_rate": 8.411440818293077e-06, "loss": 0.5573, "step": 15838 }, { "epoch": 0.5642980565402498, "grad_norm": 1.1714643408848011, "learning_rate": 8.410301545122884e-06, "loss": 0.4574, "step": 15839 }, { "epoch": 0.5643336836667439, "grad_norm": 1.4235911273125308, "learning_rate": 8.409162293120642e-06, "loss": 0.5521, "step": 15840 }, { "epoch": 0.5643693107932379, "grad_norm": 1.455313189867909, "learning_rate": 8.40802306230151e-06, "loss": 0.563, "step": 15841 }, { "epoch": 0.5644049379197321, "grad_norm": 1.0399153705484414, "learning_rate": 8.406883852680669e-06, "loss": 0.4676, "step": 15842 }, { "epoch": 0.5644405650462262, "grad_norm": 1.227846500941661, "learning_rate": 8.405744664273278e-06, "loss": 0.4808, "step": 15843 }, { "epoch": 0.5644761921727203, "grad_norm": 1.1524035283500562, "learning_rate": 8.404605497094517e-06, "loss": 0.4312, "step": 15844 }, { "epoch": 0.5645118192992145, "grad_norm": 1.4934336473817653, "learning_rate": 8.403466351159543e-06, "loss": 0.5633, "step": 15845 }, { "epoch": 0.5645474464257085, "grad_norm": 1.1300244482297195, "learning_rate": 8.402327226483535e-06, "loss": 0.4135, "step": 15846 }, { "epoch": 0.5645830735522026, "grad_norm": 1.495598654902092, "learning_rate": 8.401188123081653e-06, "loss": 0.479, "step": 15847 }, { "epoch": 0.5646187006786968, "grad_norm": 1.5582143121402914, "learning_rate": 8.400049040969068e-06, "loss": 0.5112, "step": 15848 }, { "epoch": 0.5646543278051909, "grad_norm": 1.1196614641447846, "learning_rate": 8.398909980160954e-06, "loss": 0.4704, "step": 15849 }, { "epoch": 0.564689954931685, "grad_norm": 1.8573463067642486, "learning_rate": 8.397770940672466e-06, "loss": 0.573, "step": 15850 }, { "epoch": 0.5647255820581791, "grad_norm": 1.0771278812412914, "learning_rate": 8.396631922518784e-06, "loss": 0.4358, "step": 15851 }, { "epoch": 0.5647612091846732, "grad_norm": 1.611046294710295, "learning_rate": 8.395492925715064e-06, "loss": 0.4965, "step": 15852 }, { "epoch": 0.5647968363111673, "grad_norm": 1.5434922338993329, "learning_rate": 8.394353950276479e-06, "loss": 0.512, "step": 15853 }, { "epoch": 0.5648324634376615, "grad_norm": 1.0850168695269484, "learning_rate": 8.393214996218193e-06, "loss": 0.4922, "step": 15854 }, { "epoch": 0.5648680905641555, "grad_norm": 1.0912993668450186, "learning_rate": 8.392076063555376e-06, "loss": 0.421, "step": 15855 }, { "epoch": 0.5649037176906496, "grad_norm": 1.0666339685358905, "learning_rate": 8.390937152303184e-06, "loss": 0.4477, "step": 15856 }, { "epoch": 0.5649393448171438, "grad_norm": 1.479591752263213, "learning_rate": 8.389798262476793e-06, "loss": 0.6108, "step": 15857 }, { "epoch": 0.5649749719436379, "grad_norm": 1.410049631420418, "learning_rate": 8.388659394091362e-06, "loss": 0.5164, "step": 15858 }, { "epoch": 0.565010599070132, "grad_norm": 1.1409561225386777, "learning_rate": 8.387520547162054e-06, "loss": 0.4879, "step": 15859 }, { "epoch": 0.5650462261966261, "grad_norm": 1.010196475212686, "learning_rate": 8.386381721704043e-06, "loss": 0.3845, "step": 15860 }, { "epoch": 0.5650818533231202, "grad_norm": 1.391505257396199, "learning_rate": 8.38524291773248e-06, "loss": 0.5625, "step": 15861 }, { "epoch": 0.5651174804496143, "grad_norm": 1.3465456341386188, "learning_rate": 8.384104135262545e-06, "loss": 0.5517, "step": 15862 }, { "epoch": 0.5651531075761085, "grad_norm": 1.7036970053941929, "learning_rate": 8.382965374309387e-06, "loss": 0.559, "step": 15863 }, { "epoch": 0.5651887347026026, "grad_norm": 1.2719455892790252, "learning_rate": 8.381826634888178e-06, "loss": 0.4312, "step": 15864 }, { "epoch": 0.5652243618290966, "grad_norm": 1.2281317008815014, "learning_rate": 8.380687917014076e-06, "loss": 0.432, "step": 15865 }, { "epoch": 0.5652599889555908, "grad_norm": 1.4597966747043605, "learning_rate": 8.37954922070225e-06, "loss": 0.6006, "step": 15866 }, { "epoch": 0.5652956160820849, "grad_norm": 1.381159810809271, "learning_rate": 8.378410545967856e-06, "loss": 0.58, "step": 15867 }, { "epoch": 0.565331243208579, "grad_norm": 1.0998242405042398, "learning_rate": 8.37727189282606e-06, "loss": 0.4733, "step": 15868 }, { "epoch": 0.5653668703350732, "grad_norm": 1.1076241161927842, "learning_rate": 8.37613326129202e-06, "loss": 0.4546, "step": 15869 }, { "epoch": 0.5654024974615672, "grad_norm": 1.37769538424461, "learning_rate": 8.374994651380902e-06, "loss": 0.5203, "step": 15870 }, { "epoch": 0.5654381245880613, "grad_norm": 1.1983435405573288, "learning_rate": 8.373856063107869e-06, "loss": 0.4536, "step": 15871 }, { "epoch": 0.5654737517145555, "grad_norm": 1.758827853229911, "learning_rate": 8.372717496488076e-06, "loss": 0.5782, "step": 15872 }, { "epoch": 0.5655093788410496, "grad_norm": 1.0670157847600465, "learning_rate": 8.371578951536689e-06, "loss": 0.4373, "step": 15873 }, { "epoch": 0.5655450059675436, "grad_norm": 1.0987084115994616, "learning_rate": 8.370440428268864e-06, "loss": 0.4439, "step": 15874 }, { "epoch": 0.5655806330940378, "grad_norm": 1.0770297386542103, "learning_rate": 8.369301926699766e-06, "loss": 0.4724, "step": 15875 }, { "epoch": 0.5656162602205319, "grad_norm": 1.215317484630889, "learning_rate": 8.36816344684455e-06, "loss": 0.4595, "step": 15876 }, { "epoch": 0.565651887347026, "grad_norm": 1.219444209697725, "learning_rate": 8.367024988718383e-06, "loss": 0.4293, "step": 15877 }, { "epoch": 0.5656875144735202, "grad_norm": 1.7605301885178326, "learning_rate": 8.365886552336415e-06, "loss": 0.5494, "step": 15878 }, { "epoch": 0.5657231416000142, "grad_norm": 1.2234365770810185, "learning_rate": 8.364748137713813e-06, "loss": 0.4757, "step": 15879 }, { "epoch": 0.5657587687265083, "grad_norm": 1.674808233931467, "learning_rate": 8.363609744865729e-06, "loss": 0.5584, "step": 15880 }, { "epoch": 0.5657943958530025, "grad_norm": 1.506311318080171, "learning_rate": 8.362471373807326e-06, "loss": 0.5524, "step": 15881 }, { "epoch": 0.5658300229794966, "grad_norm": 1.1100128901051882, "learning_rate": 8.361333024553765e-06, "loss": 0.4126, "step": 15882 }, { "epoch": 0.5658656501059907, "grad_norm": 1.1061768363111975, "learning_rate": 8.360194697120195e-06, "loss": 0.4279, "step": 15883 }, { "epoch": 0.5659012772324848, "grad_norm": 1.1300278589453905, "learning_rate": 8.359056391521783e-06, "loss": 0.452, "step": 15884 }, { "epoch": 0.5659369043589789, "grad_norm": 1.4810347810510396, "learning_rate": 8.357918107773678e-06, "loss": 0.5164, "step": 15885 }, { "epoch": 0.565972531485473, "grad_norm": 1.344657559751985, "learning_rate": 8.356779845891047e-06, "loss": 0.4759, "step": 15886 }, { "epoch": 0.5660081586119672, "grad_norm": 1.2017135505303116, "learning_rate": 8.355641605889034e-06, "loss": 0.5228, "step": 15887 }, { "epoch": 0.5660437857384613, "grad_norm": 2.033828159104594, "learning_rate": 8.354503387782807e-06, "loss": 0.6175, "step": 15888 }, { "epoch": 0.5660794128649553, "grad_norm": 1.1035330506185905, "learning_rate": 8.353365191587514e-06, "loss": 0.4047, "step": 15889 }, { "epoch": 0.5661150399914495, "grad_norm": 4.729912033551369, "learning_rate": 8.352227017318319e-06, "loss": 0.5449, "step": 15890 }, { "epoch": 0.5661506671179436, "grad_norm": 1.8864666351680592, "learning_rate": 8.351088864990369e-06, "loss": 0.5652, "step": 15891 }, { "epoch": 0.5661862942444377, "grad_norm": 1.1841139578950657, "learning_rate": 8.349950734618829e-06, "loss": 0.4152, "step": 15892 }, { "epoch": 0.5662219213709319, "grad_norm": 1.1854452376424494, "learning_rate": 8.34881262621884e-06, "loss": 0.4772, "step": 15893 }, { "epoch": 0.5662575484974259, "grad_norm": 1.0325522430485152, "learning_rate": 8.347674539805568e-06, "loss": 0.4275, "step": 15894 }, { "epoch": 0.56629317562392, "grad_norm": 1.8563062730743058, "learning_rate": 8.346536475394168e-06, "loss": 0.5622, "step": 15895 }, { "epoch": 0.5663288027504142, "grad_norm": 1.0178045520446732, "learning_rate": 8.345398432999784e-06, "loss": 0.4489, "step": 15896 }, { "epoch": 0.5663644298769083, "grad_norm": 1.6950976990633437, "learning_rate": 8.344260412637585e-06, "loss": 0.5291, "step": 15897 }, { "epoch": 0.5664000570034023, "grad_norm": 1.6123099131081071, "learning_rate": 8.34312241432271e-06, "loss": 0.4722, "step": 15898 }, { "epoch": 0.5664356841298965, "grad_norm": 1.1390402357098763, "learning_rate": 8.34198443807032e-06, "loss": 0.3994, "step": 15899 }, { "epoch": 0.5664713112563906, "grad_norm": 1.5648370893886907, "learning_rate": 8.340846483895565e-06, "loss": 0.545, "step": 15900 }, { "epoch": 0.5665069383828847, "grad_norm": 1.1263035703643545, "learning_rate": 8.339708551813599e-06, "loss": 0.443, "step": 15901 }, { "epoch": 0.5665425655093789, "grad_norm": 1.6426692196632755, "learning_rate": 8.338570641839573e-06, "loss": 0.5448, "step": 15902 }, { "epoch": 0.566578192635873, "grad_norm": 1.7877913857791048, "learning_rate": 8.337432753988645e-06, "loss": 0.457, "step": 15903 }, { "epoch": 0.566613819762367, "grad_norm": 1.056215837157236, "learning_rate": 8.336294888275956e-06, "loss": 0.3908, "step": 15904 }, { "epoch": 0.5666494468888612, "grad_norm": 1.2081864391366812, "learning_rate": 8.335157044716663e-06, "loss": 0.4436, "step": 15905 }, { "epoch": 0.5666850740153553, "grad_norm": 1.2417820857473494, "learning_rate": 8.334019223325924e-06, "loss": 0.4364, "step": 15906 }, { "epoch": 0.5667207011418494, "grad_norm": 1.0315581440140782, "learning_rate": 8.332881424118878e-06, "loss": 0.4349, "step": 15907 }, { "epoch": 0.5667563282683435, "grad_norm": 1.0326248547142558, "learning_rate": 8.331743647110686e-06, "loss": 0.4333, "step": 15908 }, { "epoch": 0.5667919553948376, "grad_norm": 1.480803688192906, "learning_rate": 8.330605892316489e-06, "loss": 0.5235, "step": 15909 }, { "epoch": 0.5668275825213317, "grad_norm": 1.1124583035747504, "learning_rate": 8.329468159751443e-06, "loss": 0.3974, "step": 15910 }, { "epoch": 0.5668632096478259, "grad_norm": 1.7480127414970055, "learning_rate": 8.328330449430694e-06, "loss": 0.548, "step": 15911 }, { "epoch": 0.56689883677432, "grad_norm": 1.8535873321834184, "learning_rate": 8.327192761369399e-06, "loss": 0.5805, "step": 15912 }, { "epoch": 0.566934463900814, "grad_norm": 1.2837463733747698, "learning_rate": 8.326055095582694e-06, "loss": 0.4789, "step": 15913 }, { "epoch": 0.5669700910273082, "grad_norm": 1.5945489702297437, "learning_rate": 8.324917452085743e-06, "loss": 0.5879, "step": 15914 }, { "epoch": 0.5670057181538023, "grad_norm": 1.6928962485046506, "learning_rate": 8.32377983089368e-06, "loss": 0.6089, "step": 15915 }, { "epoch": 0.5670413452802964, "grad_norm": 1.0834910989809148, "learning_rate": 8.322642232021664e-06, "loss": 0.394, "step": 15916 }, { "epoch": 0.5670769724067906, "grad_norm": 1.096498205793334, "learning_rate": 8.32150465548484e-06, "loss": 0.3798, "step": 15917 }, { "epoch": 0.5671125995332846, "grad_norm": 1.7824277521949548, "learning_rate": 8.320367101298351e-06, "loss": 0.5394, "step": 15918 }, { "epoch": 0.5671482266597787, "grad_norm": 1.1350848947344927, "learning_rate": 8.319229569477355e-06, "loss": 0.4532, "step": 15919 }, { "epoch": 0.5671838537862729, "grad_norm": 1.7862218900686526, "learning_rate": 8.318092060036984e-06, "loss": 0.5344, "step": 15920 }, { "epoch": 0.567219480912767, "grad_norm": 1.674633150917384, "learning_rate": 8.316954572992397e-06, "loss": 0.5552, "step": 15921 }, { "epoch": 0.567255108039261, "grad_norm": 1.382852864457777, "learning_rate": 8.315817108358735e-06, "loss": 0.475, "step": 15922 }, { "epoch": 0.5672907351657552, "grad_norm": 1.5147504806350542, "learning_rate": 8.314679666151148e-06, "loss": 0.4966, "step": 15923 }, { "epoch": 0.5673263622922493, "grad_norm": 1.1750810305409962, "learning_rate": 8.313542246384775e-06, "loss": 0.44, "step": 15924 }, { "epoch": 0.5673619894187434, "grad_norm": 1.548423137277526, "learning_rate": 8.312404849074769e-06, "loss": 0.5323, "step": 15925 }, { "epoch": 0.5673976165452376, "grad_norm": 1.311080800752646, "learning_rate": 8.31126747423627e-06, "loss": 0.4807, "step": 15926 }, { "epoch": 0.5674332436717316, "grad_norm": 1.6198610509364582, "learning_rate": 8.310130121884425e-06, "loss": 0.5703, "step": 15927 }, { "epoch": 0.5674688707982257, "grad_norm": 1.6987824854103215, "learning_rate": 8.308992792034379e-06, "loss": 0.5329, "step": 15928 }, { "epoch": 0.5675044979247199, "grad_norm": 1.4718350234563178, "learning_rate": 8.307855484701274e-06, "loss": 0.5361, "step": 15929 }, { "epoch": 0.567540125051214, "grad_norm": 1.137472494432864, "learning_rate": 8.306718199900262e-06, "loss": 0.4328, "step": 15930 }, { "epoch": 0.5675757521777081, "grad_norm": 1.7183245731026988, "learning_rate": 8.305580937646476e-06, "loss": 0.5153, "step": 15931 }, { "epoch": 0.5676113793042022, "grad_norm": 1.0875690319218865, "learning_rate": 8.304443697955066e-06, "loss": 0.4238, "step": 15932 }, { "epoch": 0.5676470064306963, "grad_norm": 1.188968964437583, "learning_rate": 8.30330648084117e-06, "loss": 0.3706, "step": 15933 }, { "epoch": 0.5676826335571904, "grad_norm": 1.3835394094028743, "learning_rate": 8.30216928631994e-06, "loss": 0.5612, "step": 15934 }, { "epoch": 0.5677182606836846, "grad_norm": 2.7630627981673572, "learning_rate": 8.301032114406507e-06, "loss": 0.4903, "step": 15935 }, { "epoch": 0.5677538878101787, "grad_norm": 1.514853797109068, "learning_rate": 8.299894965116022e-06, "loss": 0.5239, "step": 15936 }, { "epoch": 0.5677895149366727, "grad_norm": 1.2221799189264269, "learning_rate": 8.298757838463622e-06, "loss": 0.4657, "step": 15937 }, { "epoch": 0.5678251420631669, "grad_norm": 1.7042674749004767, "learning_rate": 8.297620734464454e-06, "loss": 0.5008, "step": 15938 }, { "epoch": 0.567860769189661, "grad_norm": 1.4769711748686423, "learning_rate": 8.296483653133649e-06, "loss": 0.6253, "step": 15939 }, { "epoch": 0.5678963963161551, "grad_norm": 1.2419703080563653, "learning_rate": 8.295346594486358e-06, "loss": 0.4268, "step": 15940 }, { "epoch": 0.5679320234426493, "grad_norm": 1.5578186750546057, "learning_rate": 8.294209558537722e-06, "loss": 0.6051, "step": 15941 }, { "epoch": 0.5679676505691433, "grad_norm": 1.5433596627219521, "learning_rate": 8.293072545302875e-06, "loss": 0.6181, "step": 15942 }, { "epoch": 0.5680032776956374, "grad_norm": 1.043491471192921, "learning_rate": 8.291935554796962e-06, "loss": 0.3995, "step": 15943 }, { "epoch": 0.5680389048221316, "grad_norm": 1.1176411675341826, "learning_rate": 8.290798587035118e-06, "loss": 0.3979, "step": 15944 }, { "epoch": 0.5680745319486257, "grad_norm": 1.1050501326200663, "learning_rate": 8.28966164203249e-06, "loss": 0.4427, "step": 15945 }, { "epoch": 0.5681101590751197, "grad_norm": 1.1169230374760708, "learning_rate": 8.288524719804209e-06, "loss": 0.433, "step": 15946 }, { "epoch": 0.5681457862016139, "grad_norm": 1.1236341684284046, "learning_rate": 8.287387820365418e-06, "loss": 0.4123, "step": 15947 }, { "epoch": 0.568181413328108, "grad_norm": 1.2573032270036275, "learning_rate": 8.286250943731254e-06, "loss": 0.4027, "step": 15948 }, { "epoch": 0.5682170404546021, "grad_norm": 1.1918371695497607, "learning_rate": 8.285114089916862e-06, "loss": 0.4509, "step": 15949 }, { "epoch": 0.5682526675810963, "grad_norm": 1.1117808639018576, "learning_rate": 8.283977258937367e-06, "loss": 0.3891, "step": 15950 }, { "epoch": 0.5682882947075903, "grad_norm": 1.5498627858494611, "learning_rate": 8.282840450807917e-06, "loss": 0.534, "step": 15951 }, { "epoch": 0.5683239218340844, "grad_norm": 1.761488438158156, "learning_rate": 8.28170366554365e-06, "loss": 0.5637, "step": 15952 }, { "epoch": 0.5683595489605786, "grad_norm": 1.574574328937426, "learning_rate": 8.280566903159695e-06, "loss": 0.5946, "step": 15953 }, { "epoch": 0.5683951760870727, "grad_norm": 1.068864870843848, "learning_rate": 8.279430163671197e-06, "loss": 0.4132, "step": 15954 }, { "epoch": 0.5684308032135668, "grad_norm": 1.8941073685157297, "learning_rate": 8.278293447093286e-06, "loss": 0.5745, "step": 15955 }, { "epoch": 0.568466430340061, "grad_norm": 1.3146108473167304, "learning_rate": 8.277156753441106e-06, "loss": 0.4704, "step": 15956 }, { "epoch": 0.568502057466555, "grad_norm": 0.9534873485895566, "learning_rate": 8.276020082729784e-06, "loss": 0.3914, "step": 15957 }, { "epoch": 0.5685376845930491, "grad_norm": 1.0137064174717814, "learning_rate": 8.27488343497446e-06, "loss": 0.4396, "step": 15958 }, { "epoch": 0.5685733117195433, "grad_norm": 1.4356079594384508, "learning_rate": 8.27374681019027e-06, "loss": 0.6209, "step": 15959 }, { "epoch": 0.5686089388460374, "grad_norm": 1.0510958621664352, "learning_rate": 8.272610208392348e-06, "loss": 0.4008, "step": 15960 }, { "epoch": 0.5686445659725314, "grad_norm": 1.0702117600619785, "learning_rate": 8.271473629595827e-06, "loss": 0.3941, "step": 15961 }, { "epoch": 0.5686801930990256, "grad_norm": 1.1599787964199584, "learning_rate": 8.270337073815839e-06, "loss": 0.4365, "step": 15962 }, { "epoch": 0.5687158202255197, "grad_norm": 1.1906439225912822, "learning_rate": 8.26920054106753e-06, "loss": 0.4534, "step": 15963 }, { "epoch": 0.5687514473520138, "grad_norm": 1.5453760779452665, "learning_rate": 8.268064031366021e-06, "loss": 0.5379, "step": 15964 }, { "epoch": 0.568787074478508, "grad_norm": 1.160569433198766, "learning_rate": 8.26692754472645e-06, "loss": 0.4652, "step": 15965 }, { "epoch": 0.568822701605002, "grad_norm": 2.294596643805722, "learning_rate": 8.265791081163949e-06, "loss": 0.5157, "step": 15966 }, { "epoch": 0.5688583287314961, "grad_norm": 1.051296461966516, "learning_rate": 8.264654640693658e-06, "loss": 0.427, "step": 15967 }, { "epoch": 0.5688939558579903, "grad_norm": 1.0414890443227747, "learning_rate": 8.263518223330698e-06, "loss": 0.4474, "step": 15968 }, { "epoch": 0.5689295829844844, "grad_norm": 1.0657447231377206, "learning_rate": 8.262381829090208e-06, "loss": 0.4429, "step": 15969 }, { "epoch": 0.5689652101109784, "grad_norm": 1.4108996441078436, "learning_rate": 8.261245457987318e-06, "loss": 0.5006, "step": 15970 }, { "epoch": 0.5690008372374726, "grad_norm": 1.3314421967500767, "learning_rate": 8.260109110037163e-06, "loss": 0.4633, "step": 15971 }, { "epoch": 0.5690364643639667, "grad_norm": 1.1279139956571997, "learning_rate": 8.258972785254867e-06, "loss": 0.4515, "step": 15972 }, { "epoch": 0.5690720914904608, "grad_norm": 1.0576639219680395, "learning_rate": 8.257836483655566e-06, "loss": 0.385, "step": 15973 }, { "epoch": 0.569107718616955, "grad_norm": 1.5375375327468055, "learning_rate": 8.25670020525439e-06, "loss": 0.5141, "step": 15974 }, { "epoch": 0.569143345743449, "grad_norm": 1.090181889482954, "learning_rate": 8.255563950066468e-06, "loss": 0.4371, "step": 15975 }, { "epoch": 0.5691789728699431, "grad_norm": 1.3174198015178786, "learning_rate": 8.254427718106934e-06, "loss": 0.529, "step": 15976 }, { "epoch": 0.5692145999964373, "grad_norm": 1.0892759942036487, "learning_rate": 8.253291509390913e-06, "loss": 0.4421, "step": 15977 }, { "epoch": 0.5692502271229314, "grad_norm": 1.3997829137837559, "learning_rate": 8.25215532393354e-06, "loss": 0.603, "step": 15978 }, { "epoch": 0.5692858542494255, "grad_norm": 1.8688640954626412, "learning_rate": 8.251019161749938e-06, "loss": 0.5539, "step": 15979 }, { "epoch": 0.5693214813759196, "grad_norm": 1.1146344184090624, "learning_rate": 8.249883022855238e-06, "loss": 0.4026, "step": 15980 }, { "epoch": 0.5693571085024137, "grad_norm": 1.473163315311263, "learning_rate": 8.24874690726457e-06, "loss": 0.4591, "step": 15981 }, { "epoch": 0.5693927356289078, "grad_norm": 1.7262250576093467, "learning_rate": 8.247610814993062e-06, "loss": 0.5245, "step": 15982 }, { "epoch": 0.569428362755402, "grad_norm": 1.076111672789674, "learning_rate": 8.246474746055837e-06, "loss": 0.4129, "step": 15983 }, { "epoch": 0.5694639898818961, "grad_norm": 1.1653441651085723, "learning_rate": 8.24533870046803e-06, "loss": 0.4619, "step": 15984 }, { "epoch": 0.5694996170083901, "grad_norm": 1.380313562779259, "learning_rate": 8.244202678244763e-06, "loss": 0.5727, "step": 15985 }, { "epoch": 0.5695352441348843, "grad_norm": 1.2221698995807315, "learning_rate": 8.243066679401162e-06, "loss": 0.4399, "step": 15986 }, { "epoch": 0.5695708712613784, "grad_norm": 1.0347891596388203, "learning_rate": 8.241930703952362e-06, "loss": 0.407, "step": 15987 }, { "epoch": 0.5696064983878725, "grad_norm": 1.250661627369952, "learning_rate": 8.240794751913476e-06, "loss": 0.5191, "step": 15988 }, { "epoch": 0.5696421255143667, "grad_norm": 1.5146072442302425, "learning_rate": 8.239658823299647e-06, "loss": 0.4884, "step": 15989 }, { "epoch": 0.5696777526408607, "grad_norm": 1.172663745942411, "learning_rate": 8.238522918125984e-06, "loss": 0.4266, "step": 15990 }, { "epoch": 0.5697133797673548, "grad_norm": 1.420242886981093, "learning_rate": 8.237387036407622e-06, "loss": 0.5023, "step": 15991 }, { "epoch": 0.569749006893849, "grad_norm": 1.3752074302724622, "learning_rate": 8.236251178159683e-06, "loss": 0.5577, "step": 15992 }, { "epoch": 0.5697846340203431, "grad_norm": 1.068069078765613, "learning_rate": 8.235115343397295e-06, "loss": 0.4634, "step": 15993 }, { "epoch": 0.5698202611468371, "grad_norm": 1.466781715915168, "learning_rate": 8.233979532135576e-06, "loss": 0.5727, "step": 15994 }, { "epoch": 0.5698558882733313, "grad_norm": 1.5786264756930826, "learning_rate": 8.232843744389655e-06, "loss": 0.5473, "step": 15995 }, { "epoch": 0.5698915153998254, "grad_norm": 1.1212949563769095, "learning_rate": 8.231707980174655e-06, "loss": 0.4283, "step": 15996 }, { "epoch": 0.5699271425263195, "grad_norm": 1.429242563516667, "learning_rate": 8.230572239505697e-06, "loss": 0.6267, "step": 15997 }, { "epoch": 0.5699627696528137, "grad_norm": 1.1036934839721966, "learning_rate": 8.229436522397912e-06, "loss": 0.4322, "step": 15998 }, { "epoch": 0.5699983967793077, "grad_norm": 1.0608353167412388, "learning_rate": 8.228300828866412e-06, "loss": 0.4463, "step": 15999 }, { "epoch": 0.5700340239058018, "grad_norm": 2.066711995438182, "learning_rate": 8.227165158926328e-06, "loss": 0.5575, "step": 16000 }, { "epoch": 0.570069651032296, "grad_norm": 1.512536887929951, "learning_rate": 8.226029512592777e-06, "loss": 0.5096, "step": 16001 }, { "epoch": 0.5701052781587901, "grad_norm": 1.1416274390834822, "learning_rate": 8.224893889880884e-06, "loss": 0.4335, "step": 16002 }, { "epoch": 0.5701409052852842, "grad_norm": 1.56946319513221, "learning_rate": 8.223758290805768e-06, "loss": 0.5695, "step": 16003 }, { "epoch": 0.5701765324117783, "grad_norm": 1.730657505177299, "learning_rate": 8.222622715382555e-06, "loss": 0.5253, "step": 16004 }, { "epoch": 0.5702121595382724, "grad_norm": 1.2153218630076834, "learning_rate": 8.22148716362636e-06, "loss": 0.4708, "step": 16005 }, { "epoch": 0.5702477866647665, "grad_norm": 2.049401861339247, "learning_rate": 8.220351635552308e-06, "loss": 0.6004, "step": 16006 }, { "epoch": 0.5702834137912607, "grad_norm": 1.6358037308781954, "learning_rate": 8.219216131175516e-06, "loss": 0.4539, "step": 16007 }, { "epoch": 0.5703190409177548, "grad_norm": 1.1357540841492841, "learning_rate": 8.218080650511107e-06, "loss": 0.4412, "step": 16008 }, { "epoch": 0.5703546680442488, "grad_norm": 1.0515639337435934, "learning_rate": 8.216945193574204e-06, "loss": 0.4, "step": 16009 }, { "epoch": 0.570390295170743, "grad_norm": 1.0722801861556832, "learning_rate": 8.215809760379915e-06, "loss": 0.4224, "step": 16010 }, { "epoch": 0.5704259222972371, "grad_norm": 1.4556688314976483, "learning_rate": 8.21467435094337e-06, "loss": 0.5624, "step": 16011 }, { "epoch": 0.5704615494237312, "grad_norm": 1.1567398358891798, "learning_rate": 8.213538965279681e-06, "loss": 0.3811, "step": 16012 }, { "epoch": 0.5704971765502254, "grad_norm": 1.3032819465456058, "learning_rate": 8.212403603403974e-06, "loss": 0.5532, "step": 16013 }, { "epoch": 0.5705328036767194, "grad_norm": 1.0719809194383316, "learning_rate": 8.211268265331358e-06, "loss": 0.4325, "step": 16014 }, { "epoch": 0.5705684308032136, "grad_norm": 1.2156974439516997, "learning_rate": 8.210132951076959e-06, "loss": 0.5594, "step": 16015 }, { "epoch": 0.5706040579297077, "grad_norm": 1.9877469321441248, "learning_rate": 8.208997660655887e-06, "loss": 0.5641, "step": 16016 }, { "epoch": 0.5706396850562018, "grad_norm": 1.3615668925693654, "learning_rate": 8.207862394083265e-06, "loss": 0.5294, "step": 16017 }, { "epoch": 0.570675312182696, "grad_norm": 1.173768278095939, "learning_rate": 8.206727151374208e-06, "loss": 0.48, "step": 16018 }, { "epoch": 0.57071093930919, "grad_norm": 1.4040434436743108, "learning_rate": 8.205591932543836e-06, "loss": 0.6257, "step": 16019 }, { "epoch": 0.5707465664356841, "grad_norm": 1.0312916219132053, "learning_rate": 8.204456737607255e-06, "loss": 0.4186, "step": 16020 }, { "epoch": 0.5707821935621783, "grad_norm": 1.0394144580900504, "learning_rate": 8.203321566579589e-06, "loss": 0.4657, "step": 16021 }, { "epoch": 0.5708178206886724, "grad_norm": 1.1440934567698775, "learning_rate": 8.202186419475955e-06, "loss": 0.4417, "step": 16022 }, { "epoch": 0.5708534478151664, "grad_norm": 1.6346551226912098, "learning_rate": 8.201051296311462e-06, "loss": 0.4728, "step": 16023 }, { "epoch": 0.5708890749416606, "grad_norm": 1.4765244672500621, "learning_rate": 8.199916197101233e-06, "loss": 0.4606, "step": 16024 }, { "epoch": 0.5709247020681547, "grad_norm": 1.1882569465945634, "learning_rate": 8.19878112186037e-06, "loss": 0.4851, "step": 16025 }, { "epoch": 0.5709603291946488, "grad_norm": 1.3650432302969786, "learning_rate": 8.197646070604004e-06, "loss": 0.5493, "step": 16026 }, { "epoch": 0.570995956321143, "grad_norm": 1.2107167569499104, "learning_rate": 8.196511043347237e-06, "loss": 0.4699, "step": 16027 }, { "epoch": 0.571031583447637, "grad_norm": 1.6799471741653158, "learning_rate": 8.195376040105186e-06, "loss": 0.4928, "step": 16028 }, { "epoch": 0.5710672105741311, "grad_norm": 1.4196223005763093, "learning_rate": 8.194241060892962e-06, "loss": 0.5164, "step": 16029 }, { "epoch": 0.5711028377006253, "grad_norm": 1.0720183256504223, "learning_rate": 8.193106105725686e-06, "loss": 0.4512, "step": 16030 }, { "epoch": 0.5711384648271194, "grad_norm": 1.3566719685888242, "learning_rate": 8.19197117461846e-06, "loss": 0.5255, "step": 16031 }, { "epoch": 0.5711740919536135, "grad_norm": 1.194209070319305, "learning_rate": 8.190836267586403e-06, "loss": 0.4586, "step": 16032 }, { "epoch": 0.5712097190801076, "grad_norm": 1.188283983244399, "learning_rate": 8.189701384644627e-06, "loss": 0.5118, "step": 16033 }, { "epoch": 0.5712453462066017, "grad_norm": 1.14966622517495, "learning_rate": 8.18856652580824e-06, "loss": 0.4599, "step": 16034 }, { "epoch": 0.5712809733330958, "grad_norm": 1.5440732211147976, "learning_rate": 8.187431691092361e-06, "loss": 0.4793, "step": 16035 }, { "epoch": 0.57131660045959, "grad_norm": 1.0447148693232617, "learning_rate": 8.18629688051209e-06, "loss": 0.3474, "step": 16036 }, { "epoch": 0.5713522275860841, "grad_norm": 1.8693145384874907, "learning_rate": 8.185162094082547e-06, "loss": 0.5164, "step": 16037 }, { "epoch": 0.5713878547125781, "grad_norm": 1.3951166167183162, "learning_rate": 8.184027331818837e-06, "loss": 0.4763, "step": 16038 }, { "epoch": 0.5714234818390723, "grad_norm": 1.2580745207926318, "learning_rate": 8.182892593736073e-06, "loss": 0.4727, "step": 16039 }, { "epoch": 0.5714591089655664, "grad_norm": 1.0955051264537368, "learning_rate": 8.181757879849364e-06, "loss": 0.4241, "step": 16040 }, { "epoch": 0.5714947360920605, "grad_norm": 1.0874335485284714, "learning_rate": 8.180623190173824e-06, "loss": 0.4306, "step": 16041 }, { "epoch": 0.5715303632185547, "grad_norm": 1.2212752958034931, "learning_rate": 8.179488524724553e-06, "loss": 0.4603, "step": 16042 }, { "epoch": 0.5715659903450487, "grad_norm": 1.0804469891268722, "learning_rate": 8.178353883516665e-06, "loss": 0.462, "step": 16043 }, { "epoch": 0.5716016174715428, "grad_norm": 1.4421282478162987, "learning_rate": 8.177219266565269e-06, "loss": 0.5377, "step": 16044 }, { "epoch": 0.571637244598037, "grad_norm": 1.2714355472407082, "learning_rate": 8.176084673885471e-06, "loss": 0.4511, "step": 16045 }, { "epoch": 0.5716728717245311, "grad_norm": 1.1666873146388537, "learning_rate": 8.174950105492386e-06, "loss": 0.4554, "step": 16046 }, { "epoch": 0.5717084988510251, "grad_norm": 1.221985375797667, "learning_rate": 8.173815561401112e-06, "loss": 0.4137, "step": 16047 }, { "epoch": 0.5717441259775193, "grad_norm": 1.0911344488881616, "learning_rate": 8.17268104162676e-06, "loss": 0.4229, "step": 16048 }, { "epoch": 0.5717797531040134, "grad_norm": 1.4081258842366449, "learning_rate": 8.171546546184436e-06, "loss": 0.5879, "step": 16049 }, { "epoch": 0.5718153802305075, "grad_norm": 1.3699629896740595, "learning_rate": 8.170412075089254e-06, "loss": 0.5762, "step": 16050 }, { "epoch": 0.5718510073570017, "grad_norm": 1.0615110328855057, "learning_rate": 8.169277628356305e-06, "loss": 0.467, "step": 16051 }, { "epoch": 0.5718866344834957, "grad_norm": 1.1489779689880557, "learning_rate": 8.168143206000712e-06, "loss": 0.4259, "step": 16052 }, { "epoch": 0.5719222616099898, "grad_norm": 1.0939489181367614, "learning_rate": 8.167008808037568e-06, "loss": 0.4736, "step": 16053 }, { "epoch": 0.571957888736484, "grad_norm": 1.4960756682999705, "learning_rate": 8.165874434481982e-06, "loss": 0.5427, "step": 16054 }, { "epoch": 0.5719935158629781, "grad_norm": 1.0971648199819903, "learning_rate": 8.164740085349062e-06, "loss": 0.4033, "step": 16055 }, { "epoch": 0.5720291429894722, "grad_norm": 1.6081032580287287, "learning_rate": 8.16360576065391e-06, "loss": 0.5066, "step": 16056 }, { "epoch": 0.5720647701159663, "grad_norm": 1.1837123524113435, "learning_rate": 8.162471460411634e-06, "loss": 0.4615, "step": 16057 }, { "epoch": 0.5721003972424604, "grad_norm": 1.248334242858655, "learning_rate": 8.16133718463733e-06, "loss": 0.4332, "step": 16058 }, { "epoch": 0.5721360243689545, "grad_norm": 1.3254558782929728, "learning_rate": 8.16020293334611e-06, "loss": 0.4731, "step": 16059 }, { "epoch": 0.5721716514954487, "grad_norm": 1.4755473891364896, "learning_rate": 8.15906870655307e-06, "loss": 0.4886, "step": 16060 }, { "epoch": 0.5722072786219428, "grad_norm": 1.1745351390208683, "learning_rate": 8.157934504273323e-06, "loss": 0.4419, "step": 16061 }, { "epoch": 0.5722429057484368, "grad_norm": 1.0664146627381441, "learning_rate": 8.156800326521961e-06, "loss": 0.3924, "step": 16062 }, { "epoch": 0.572278532874931, "grad_norm": 1.425920730707555, "learning_rate": 8.155666173314092e-06, "loss": 0.5884, "step": 16063 }, { "epoch": 0.5723141600014251, "grad_norm": 1.1821317751612088, "learning_rate": 8.154532044664815e-06, "loss": 0.4465, "step": 16064 }, { "epoch": 0.5723497871279192, "grad_norm": 1.1518758828126963, "learning_rate": 8.153397940589238e-06, "loss": 0.4297, "step": 16065 }, { "epoch": 0.5723854142544134, "grad_norm": 1.2253792880996663, "learning_rate": 8.152263861102455e-06, "loss": 0.4685, "step": 16066 }, { "epoch": 0.5724210413809074, "grad_norm": 1.1357561463597405, "learning_rate": 8.15112980621957e-06, "loss": 0.4179, "step": 16067 }, { "epoch": 0.5724566685074015, "grad_norm": 1.8410152705081997, "learning_rate": 8.149995775955686e-06, "loss": 0.5633, "step": 16068 }, { "epoch": 0.5724922956338957, "grad_norm": 1.2729133618454913, "learning_rate": 8.148861770325899e-06, "loss": 0.4494, "step": 16069 }, { "epoch": 0.5725279227603898, "grad_norm": 1.566094536994152, "learning_rate": 8.147727789345313e-06, "loss": 0.5896, "step": 16070 }, { "epoch": 0.5725635498868838, "grad_norm": 1.157526499985111, "learning_rate": 8.146593833029024e-06, "loss": 0.4088, "step": 16071 }, { "epoch": 0.572599177013378, "grad_norm": 1.506331951145776, "learning_rate": 8.145459901392138e-06, "loss": 0.5828, "step": 16072 }, { "epoch": 0.5726348041398721, "grad_norm": 1.1087565203817924, "learning_rate": 8.144325994449746e-06, "loss": 0.4594, "step": 16073 }, { "epoch": 0.5726704312663662, "grad_norm": 1.0672613266423672, "learning_rate": 8.143192112216953e-06, "loss": 0.3898, "step": 16074 }, { "epoch": 0.5727060583928604, "grad_norm": 1.3159128240456082, "learning_rate": 8.142058254708853e-06, "loss": 0.5607, "step": 16075 }, { "epoch": 0.5727416855193544, "grad_norm": 1.139050378806974, "learning_rate": 8.140924421940548e-06, "loss": 0.4281, "step": 16076 }, { "epoch": 0.5727773126458485, "grad_norm": 1.0529671278249015, "learning_rate": 8.13979061392713e-06, "loss": 0.4264, "step": 16077 }, { "epoch": 0.5728129397723427, "grad_norm": 1.1994485373594796, "learning_rate": 8.1386568306837e-06, "loss": 0.4262, "step": 16078 }, { "epoch": 0.5728485668988368, "grad_norm": 1.5616507275686735, "learning_rate": 8.13752307222536e-06, "loss": 0.5769, "step": 16079 }, { "epoch": 0.5728841940253309, "grad_norm": 1.5558498983393902, "learning_rate": 8.136389338567198e-06, "loss": 0.45, "step": 16080 }, { "epoch": 0.572919821151825, "grad_norm": 2.4224547112193227, "learning_rate": 8.135255629724317e-06, "loss": 0.5041, "step": 16081 }, { "epoch": 0.5729554482783191, "grad_norm": 1.3279851118625552, "learning_rate": 8.134121945711808e-06, "loss": 0.5691, "step": 16082 }, { "epoch": 0.5729910754048132, "grad_norm": 1.8454178937455614, "learning_rate": 8.132988286544773e-06, "loss": 0.6147, "step": 16083 }, { "epoch": 0.5730267025313074, "grad_norm": 1.1056540552803737, "learning_rate": 8.131854652238302e-06, "loss": 0.4177, "step": 16084 }, { "epoch": 0.5730623296578015, "grad_norm": 1.7801600350475724, "learning_rate": 8.130721042807494e-06, "loss": 0.5689, "step": 16085 }, { "epoch": 0.5730979567842955, "grad_norm": 1.286147485053572, "learning_rate": 8.129587458267439e-06, "loss": 0.4465, "step": 16086 }, { "epoch": 0.5731335839107897, "grad_norm": 1.3318149339584395, "learning_rate": 8.128453898633237e-06, "loss": 0.4483, "step": 16087 }, { "epoch": 0.5731692110372838, "grad_norm": 1.062144499611894, "learning_rate": 8.127320363919976e-06, "loss": 0.4252, "step": 16088 }, { "epoch": 0.5732048381637779, "grad_norm": 1.5841840891076457, "learning_rate": 8.126186854142752e-06, "loss": 0.5721, "step": 16089 }, { "epoch": 0.5732404652902721, "grad_norm": 1.493110426188103, "learning_rate": 8.125053369316666e-06, "loss": 0.493, "step": 16090 }, { "epoch": 0.5732760924167661, "grad_norm": 1.4112932493361905, "learning_rate": 8.123919909456801e-06, "loss": 0.4785, "step": 16091 }, { "epoch": 0.5733117195432602, "grad_norm": 1.6558790057910855, "learning_rate": 8.122786474578255e-06, "loss": 0.559, "step": 16092 }, { "epoch": 0.5733473466697544, "grad_norm": 1.1706832073869458, "learning_rate": 8.121653064696118e-06, "loss": 0.4537, "step": 16093 }, { "epoch": 0.5733829737962485, "grad_norm": 1.8397504159034106, "learning_rate": 8.120519679825487e-06, "loss": 0.5165, "step": 16094 }, { "epoch": 0.5734186009227425, "grad_norm": 1.2213497084841298, "learning_rate": 8.119386319981445e-06, "loss": 0.4811, "step": 16095 }, { "epoch": 0.5734542280492367, "grad_norm": 1.094019446161371, "learning_rate": 8.118252985179091e-06, "loss": 0.4196, "step": 16096 }, { "epoch": 0.5734898551757308, "grad_norm": 1.8406463951099465, "learning_rate": 8.117119675433514e-06, "loss": 0.6025, "step": 16097 }, { "epoch": 0.5735254823022249, "grad_norm": 1.8648703885577707, "learning_rate": 8.115986390759805e-06, "loss": 0.5735, "step": 16098 }, { "epoch": 0.5735611094287191, "grad_norm": 1.204136992999731, "learning_rate": 8.114853131173053e-06, "loss": 0.4694, "step": 16099 }, { "epoch": 0.5735967365552131, "grad_norm": 1.7396168961455274, "learning_rate": 8.11371989668835e-06, "loss": 0.6089, "step": 16100 }, { "epoch": 0.5736323636817072, "grad_norm": 1.0088006331885464, "learning_rate": 8.112586687320783e-06, "loss": 0.44, "step": 16101 }, { "epoch": 0.5736679908082014, "grad_norm": 1.6512099982018813, "learning_rate": 8.111453503085444e-06, "loss": 0.5474, "step": 16102 }, { "epoch": 0.5737036179346955, "grad_norm": 1.2693664427101157, "learning_rate": 8.110320343997423e-06, "loss": 0.4103, "step": 16103 }, { "epoch": 0.5737392450611896, "grad_norm": 1.1749063992205484, "learning_rate": 8.109187210071806e-06, "loss": 0.4859, "step": 16104 }, { "epoch": 0.5737748721876837, "grad_norm": 1.632479788768247, "learning_rate": 8.108054101323686e-06, "loss": 0.553, "step": 16105 }, { "epoch": 0.5738104993141778, "grad_norm": 1.8903577476474829, "learning_rate": 8.106921017768145e-06, "loss": 0.5504, "step": 16106 }, { "epoch": 0.5738461264406719, "grad_norm": 1.0721225825617149, "learning_rate": 8.105787959420274e-06, "loss": 0.492, "step": 16107 }, { "epoch": 0.5738817535671661, "grad_norm": 1.7622031670188674, "learning_rate": 8.104654926295161e-06, "loss": 0.6204, "step": 16108 }, { "epoch": 0.5739173806936602, "grad_norm": 1.0298261045021198, "learning_rate": 8.103521918407894e-06, "loss": 0.456, "step": 16109 }, { "epoch": 0.5739530078201542, "grad_norm": 1.1413635486195044, "learning_rate": 8.102388935773556e-06, "loss": 0.4327, "step": 16110 }, { "epoch": 0.5739886349466484, "grad_norm": 1.3328561576779414, "learning_rate": 8.101255978407237e-06, "loss": 0.5464, "step": 16111 }, { "epoch": 0.5740242620731425, "grad_norm": 1.083336573039359, "learning_rate": 8.10012304632402e-06, "loss": 0.4796, "step": 16112 }, { "epoch": 0.5740598891996366, "grad_norm": 1.0463814723789802, "learning_rate": 8.098990139538992e-06, "loss": 0.4281, "step": 16113 }, { "epoch": 0.5740955163261308, "grad_norm": 1.1955191364922184, "learning_rate": 8.097857258067242e-06, "loss": 0.4431, "step": 16114 }, { "epoch": 0.5741311434526248, "grad_norm": 1.7398994635585396, "learning_rate": 8.09672440192385e-06, "loss": 0.5867, "step": 16115 }, { "epoch": 0.5741667705791189, "grad_norm": 1.6399012002418842, "learning_rate": 8.095591571123906e-06, "loss": 0.4962, "step": 16116 }, { "epoch": 0.5742023977056131, "grad_norm": 1.4353678919840507, "learning_rate": 8.094458765682487e-06, "loss": 0.4913, "step": 16117 }, { "epoch": 0.5742380248321072, "grad_norm": 1.02707540764845, "learning_rate": 8.093325985614686e-06, "loss": 0.388, "step": 16118 }, { "epoch": 0.5742736519586013, "grad_norm": 1.153659535477971, "learning_rate": 8.092193230935577e-06, "loss": 0.4351, "step": 16119 }, { "epoch": 0.5743092790850954, "grad_norm": 1.071840246596091, "learning_rate": 8.091060501660254e-06, "loss": 0.3991, "step": 16120 }, { "epoch": 0.5743449062115895, "grad_norm": 1.1493419291111335, "learning_rate": 8.089927797803791e-06, "loss": 0.4138, "step": 16121 }, { "epoch": 0.5743805333380836, "grad_norm": 1.6523591090327383, "learning_rate": 8.088795119381277e-06, "loss": 0.5225, "step": 16122 }, { "epoch": 0.5744161604645778, "grad_norm": 1.497656795254285, "learning_rate": 8.087662466407788e-06, "loss": 0.5878, "step": 16123 }, { "epoch": 0.5744517875910718, "grad_norm": 1.6483223551522055, "learning_rate": 8.086529838898411e-06, "loss": 0.5894, "step": 16124 }, { "epoch": 0.5744874147175659, "grad_norm": 1.1256072828910517, "learning_rate": 8.08539723686823e-06, "loss": 0.4308, "step": 16125 }, { "epoch": 0.5745230418440601, "grad_norm": 1.349098134218714, "learning_rate": 8.084264660332316e-06, "loss": 0.5807, "step": 16126 }, { "epoch": 0.5745586689705542, "grad_norm": 1.125336899163279, "learning_rate": 8.083132109305766e-06, "loss": 0.4872, "step": 16127 }, { "epoch": 0.5745942960970483, "grad_norm": 1.522625242705393, "learning_rate": 8.081999583803646e-06, "loss": 0.5629, "step": 16128 }, { "epoch": 0.5746299232235424, "grad_norm": 1.528831838705428, "learning_rate": 8.080867083841045e-06, "loss": 0.5744, "step": 16129 }, { "epoch": 0.5746655503500365, "grad_norm": 1.4641929137051062, "learning_rate": 8.079734609433038e-06, "loss": 0.5725, "step": 16130 }, { "epoch": 0.5747011774765306, "grad_norm": 1.2418339508234386, "learning_rate": 8.07860216059471e-06, "loss": 0.4368, "step": 16131 }, { "epoch": 0.5747368046030248, "grad_norm": 2.5765731974972064, "learning_rate": 8.077469737341134e-06, "loss": 0.6262, "step": 16132 }, { "epoch": 0.5747724317295189, "grad_norm": 1.3486646198819094, "learning_rate": 8.076337339687395e-06, "loss": 0.4993, "step": 16133 }, { "epoch": 0.5748080588560129, "grad_norm": 1.4866664138564933, "learning_rate": 8.075204967648565e-06, "loss": 0.4946, "step": 16134 }, { "epoch": 0.5748436859825071, "grad_norm": 1.5818179203074612, "learning_rate": 8.074072621239726e-06, "loss": 0.497, "step": 16135 }, { "epoch": 0.5748793131090012, "grad_norm": 1.1439103632453198, "learning_rate": 8.072940300475963e-06, "loss": 0.4494, "step": 16136 }, { "epoch": 0.5749149402354953, "grad_norm": 1.4345075527019282, "learning_rate": 8.07180800537234e-06, "loss": 0.5013, "step": 16137 }, { "epoch": 0.5749505673619895, "grad_norm": 1.0506883913212428, "learning_rate": 8.070675735943945e-06, "loss": 0.4797, "step": 16138 }, { "epoch": 0.5749861944884835, "grad_norm": 1.5565881566445776, "learning_rate": 8.069543492205847e-06, "loss": 0.6044, "step": 16139 }, { "epoch": 0.5750218216149776, "grad_norm": 1.5836427965473852, "learning_rate": 8.06841127417313e-06, "loss": 0.5364, "step": 16140 }, { "epoch": 0.5750574487414718, "grad_norm": 1.1332242147641676, "learning_rate": 8.067279081860866e-06, "loss": 0.4848, "step": 16141 }, { "epoch": 0.5750930758679659, "grad_norm": 1.4953152987997942, "learning_rate": 8.066146915284134e-06, "loss": 0.5597, "step": 16142 }, { "epoch": 0.57512870299446, "grad_norm": 1.0919588858304576, "learning_rate": 8.065014774458004e-06, "loss": 0.4165, "step": 16143 }, { "epoch": 0.5751643301209541, "grad_norm": 1.6322560732402014, "learning_rate": 8.063882659397556e-06, "loss": 0.5489, "step": 16144 }, { "epoch": 0.5751999572474482, "grad_norm": 1.0994247425659234, "learning_rate": 8.062750570117865e-06, "loss": 0.4277, "step": 16145 }, { "epoch": 0.5752355843739423, "grad_norm": 1.2275022586600783, "learning_rate": 8.061618506634006e-06, "loss": 0.4358, "step": 16146 }, { "epoch": 0.5752712115004365, "grad_norm": 1.1064895750064085, "learning_rate": 8.060486468961047e-06, "loss": 0.4256, "step": 16147 }, { "epoch": 0.5753068386269306, "grad_norm": 1.1822457400122657, "learning_rate": 8.059354457114066e-06, "loss": 0.4561, "step": 16148 }, { "epoch": 0.5753424657534246, "grad_norm": 1.1646482600190455, "learning_rate": 8.05822247110814e-06, "loss": 0.463, "step": 16149 }, { "epoch": 0.5753780928799188, "grad_norm": 1.1356661306975961, "learning_rate": 8.057090510958336e-06, "loss": 0.4285, "step": 16150 }, { "epoch": 0.5754137200064129, "grad_norm": 1.203783884501484, "learning_rate": 8.055958576679735e-06, "loss": 0.3867, "step": 16151 }, { "epoch": 0.575449347132907, "grad_norm": 1.1739067695156609, "learning_rate": 8.054826668287398e-06, "loss": 0.4628, "step": 16152 }, { "epoch": 0.5754849742594011, "grad_norm": 1.185255320761644, "learning_rate": 8.05369478579641e-06, "loss": 0.4199, "step": 16153 }, { "epoch": 0.5755206013858952, "grad_norm": 1.5638491151196259, "learning_rate": 8.05256292922183e-06, "loss": 0.5769, "step": 16154 }, { "epoch": 0.5755562285123893, "grad_norm": 2.036398441154315, "learning_rate": 8.051431098578742e-06, "loss": 0.5934, "step": 16155 }, { "epoch": 0.5755918556388835, "grad_norm": 1.0809319763766585, "learning_rate": 8.050299293882205e-06, "loss": 0.4673, "step": 16156 }, { "epoch": 0.5756274827653776, "grad_norm": 1.4657462047101326, "learning_rate": 8.0491675151473e-06, "loss": 0.4612, "step": 16157 }, { "epoch": 0.5756631098918716, "grad_norm": 1.1450292440461343, "learning_rate": 8.048035762389091e-06, "loss": 0.4059, "step": 16158 }, { "epoch": 0.5756987370183658, "grad_norm": 1.097308130481176, "learning_rate": 8.046904035622648e-06, "loss": 0.4098, "step": 16159 }, { "epoch": 0.5757343641448599, "grad_norm": 1.5054664321421398, "learning_rate": 8.045772334863046e-06, "loss": 0.4819, "step": 16160 }, { "epoch": 0.575769991271354, "grad_norm": 1.6633060237160502, "learning_rate": 8.044640660125351e-06, "loss": 0.6059, "step": 16161 }, { "epoch": 0.5758056183978482, "grad_norm": 1.174713644467061, "learning_rate": 8.043509011424634e-06, "loss": 0.4458, "step": 16162 }, { "epoch": 0.5758412455243422, "grad_norm": 1.2933076683849507, "learning_rate": 8.04237738877596e-06, "loss": 0.437, "step": 16163 }, { "epoch": 0.5758768726508363, "grad_norm": 1.5018330037629977, "learning_rate": 8.0412457921944e-06, "loss": 0.5599, "step": 16164 }, { "epoch": 0.5759124997773305, "grad_norm": 1.5051527124263107, "learning_rate": 8.040114221695019e-06, "loss": 0.5579, "step": 16165 }, { "epoch": 0.5759481269038246, "grad_norm": 1.306030964753684, "learning_rate": 8.038982677292889e-06, "loss": 0.5267, "step": 16166 }, { "epoch": 0.5759837540303187, "grad_norm": 1.1104691333178458, "learning_rate": 8.037851159003074e-06, "loss": 0.4448, "step": 16167 }, { "epoch": 0.5760193811568128, "grad_norm": 1.0826910164458243, "learning_rate": 8.036719666840646e-06, "loss": 0.473, "step": 16168 }, { "epoch": 0.5760550082833069, "grad_norm": 1.1601674271041924, "learning_rate": 8.035588200820665e-06, "loss": 0.4018, "step": 16169 }, { "epoch": 0.576090635409801, "grad_norm": 1.6524383049298181, "learning_rate": 8.034456760958198e-06, "loss": 0.5613, "step": 16170 }, { "epoch": 0.5761262625362952, "grad_norm": 1.0667280601878377, "learning_rate": 8.033325347268318e-06, "loss": 0.4404, "step": 16171 }, { "epoch": 0.5761618896627893, "grad_norm": 2.191035909319672, "learning_rate": 8.03219395976608e-06, "loss": 0.4371, "step": 16172 }, { "epoch": 0.5761975167892833, "grad_norm": 1.233065039993554, "learning_rate": 8.031062598466561e-06, "loss": 0.4687, "step": 16173 }, { "epoch": 0.5762331439157775, "grad_norm": 1.0049685815593354, "learning_rate": 8.029931263384816e-06, "loss": 0.4461, "step": 16174 }, { "epoch": 0.5762687710422716, "grad_norm": 1.3127023490399075, "learning_rate": 8.028799954535915e-06, "loss": 0.4756, "step": 16175 }, { "epoch": 0.5763043981687657, "grad_norm": 1.2227777571164988, "learning_rate": 8.027668671934917e-06, "loss": 0.461, "step": 16176 }, { "epoch": 0.5763400252952599, "grad_norm": 2.412509613456851, "learning_rate": 8.026537415596895e-06, "loss": 0.5881, "step": 16177 }, { "epoch": 0.5763756524217539, "grad_norm": 2.0275798366422277, "learning_rate": 8.0254061855369e-06, "loss": 0.5337, "step": 16178 }, { "epoch": 0.576411279548248, "grad_norm": 1.0182790781912956, "learning_rate": 8.024274981770006e-06, "loss": 0.3904, "step": 16179 }, { "epoch": 0.5764469066747422, "grad_norm": 1.6117501865300377, "learning_rate": 8.023143804311269e-06, "loss": 0.6309, "step": 16180 }, { "epoch": 0.5764825338012363, "grad_norm": 1.2249627216574348, "learning_rate": 8.022012653175754e-06, "loss": 0.5101, "step": 16181 }, { "epoch": 0.5765181609277303, "grad_norm": 1.4825241871637616, "learning_rate": 8.020881528378525e-06, "loss": 0.5134, "step": 16182 }, { "epoch": 0.5765537880542245, "grad_norm": 1.7497735548603437, "learning_rate": 8.019750429934641e-06, "loss": 0.5956, "step": 16183 }, { "epoch": 0.5765894151807186, "grad_norm": 1.824457995089903, "learning_rate": 8.018619357859167e-06, "loss": 0.4044, "step": 16184 }, { "epoch": 0.5766250423072127, "grad_norm": 1.3889425389073229, "learning_rate": 8.017488312167158e-06, "loss": 0.5731, "step": 16185 }, { "epoch": 0.5766606694337069, "grad_norm": 1.4727846368586401, "learning_rate": 8.016357292873679e-06, "loss": 0.586, "step": 16186 }, { "epoch": 0.5766962965602009, "grad_norm": 1.3923840252829511, "learning_rate": 8.015226299993786e-06, "loss": 0.6253, "step": 16187 }, { "epoch": 0.576731923686695, "grad_norm": 1.4617414069314587, "learning_rate": 8.014095333542548e-06, "loss": 0.5104, "step": 16188 }, { "epoch": 0.5767675508131892, "grad_norm": 1.1348142776424877, "learning_rate": 8.012964393535015e-06, "loss": 0.4122, "step": 16189 }, { "epoch": 0.5768031779396833, "grad_norm": 1.157919605784236, "learning_rate": 8.011833479986251e-06, "loss": 0.4653, "step": 16190 }, { "epoch": 0.5768388050661774, "grad_norm": 1.1187674364866733, "learning_rate": 8.010702592911311e-06, "loss": 0.4714, "step": 16191 }, { "epoch": 0.5768744321926715, "grad_norm": 1.5044411774860296, "learning_rate": 8.00957173232526e-06, "loss": 0.5944, "step": 16192 }, { "epoch": 0.5769100593191656, "grad_norm": 1.117745453176757, "learning_rate": 8.008440898243149e-06, "loss": 0.4355, "step": 16193 }, { "epoch": 0.5769456864456597, "grad_norm": 1.5366997769823165, "learning_rate": 8.00731009068004e-06, "loss": 0.6365, "step": 16194 }, { "epoch": 0.5769813135721539, "grad_norm": 1.0564427116876198, "learning_rate": 8.006179309650996e-06, "loss": 0.4716, "step": 16195 }, { "epoch": 0.577016940698648, "grad_norm": 1.1734128959964603, "learning_rate": 8.005048555171062e-06, "loss": 0.4827, "step": 16196 }, { "epoch": 0.577052567825142, "grad_norm": 1.5597496012273733, "learning_rate": 8.003917827255303e-06, "loss": 0.5855, "step": 16197 }, { "epoch": 0.5770881949516362, "grad_norm": 1.586164344138524, "learning_rate": 8.002787125918771e-06, "loss": 0.5636, "step": 16198 }, { "epoch": 0.5771238220781303, "grad_norm": 1.6525632810472, "learning_rate": 8.001656451176529e-06, "loss": 0.5311, "step": 16199 }, { "epoch": 0.5771594492046244, "grad_norm": 1.0840663175825909, "learning_rate": 8.000525803043623e-06, "loss": 0.4541, "step": 16200 }, { "epoch": 0.5771950763311186, "grad_norm": 1.2166889500646962, "learning_rate": 7.999395181535118e-06, "loss": 0.449, "step": 16201 }, { "epoch": 0.5772307034576126, "grad_norm": 1.1234283611936844, "learning_rate": 7.998264586666062e-06, "loss": 0.4419, "step": 16202 }, { "epoch": 0.5772663305841067, "grad_norm": 1.2228187620109112, "learning_rate": 7.997134018451513e-06, "loss": 0.4916, "step": 16203 }, { "epoch": 0.5773019577106009, "grad_norm": 1.2105434015578518, "learning_rate": 7.996003476906521e-06, "loss": 0.4848, "step": 16204 }, { "epoch": 0.577337584837095, "grad_norm": 3.131629565565117, "learning_rate": 7.994872962046146e-06, "loss": 0.4892, "step": 16205 }, { "epoch": 0.577373211963589, "grad_norm": 1.8189709389732098, "learning_rate": 7.993742473885441e-06, "loss": 0.5711, "step": 16206 }, { "epoch": 0.5774088390900832, "grad_norm": 4.627885483641914, "learning_rate": 7.992612012439456e-06, "loss": 0.5104, "step": 16207 }, { "epoch": 0.5774444662165773, "grad_norm": 1.2513069955776723, "learning_rate": 7.991481577723244e-06, "loss": 0.4533, "step": 16208 }, { "epoch": 0.5774800933430714, "grad_norm": 1.5357775121627972, "learning_rate": 7.990351169751859e-06, "loss": 0.4891, "step": 16209 }, { "epoch": 0.5775157204695656, "grad_norm": 1.0941883883981534, "learning_rate": 7.989220788540356e-06, "loss": 0.4167, "step": 16210 }, { "epoch": 0.5775513475960596, "grad_norm": 1.1021240065380749, "learning_rate": 7.988090434103779e-06, "loss": 0.426, "step": 16211 }, { "epoch": 0.5775869747225537, "grad_norm": 1.2068457451933146, "learning_rate": 7.986960106457188e-06, "loss": 0.4561, "step": 16212 }, { "epoch": 0.5776226018490479, "grad_norm": 1.4229860604081979, "learning_rate": 7.985829805615626e-06, "loss": 0.524, "step": 16213 }, { "epoch": 0.577658228975542, "grad_norm": 1.2402818618565765, "learning_rate": 7.984699531594153e-06, "loss": 0.533, "step": 16214 }, { "epoch": 0.577693856102036, "grad_norm": 1.1422100091356162, "learning_rate": 7.983569284407808e-06, "loss": 0.416, "step": 16215 }, { "epoch": 0.5777294832285302, "grad_norm": 1.3418010367501882, "learning_rate": 7.982439064071649e-06, "loss": 0.4311, "step": 16216 }, { "epoch": 0.5777651103550243, "grad_norm": 1.182024741656301, "learning_rate": 7.98130887060073e-06, "loss": 0.4537, "step": 16217 }, { "epoch": 0.5778007374815184, "grad_norm": 1.108077173276332, "learning_rate": 7.980178704010089e-06, "loss": 0.4348, "step": 16218 }, { "epoch": 0.5778363646080126, "grad_norm": 1.0609164842949699, "learning_rate": 7.979048564314782e-06, "loss": 0.4401, "step": 16219 }, { "epoch": 0.5778719917345067, "grad_norm": 1.1881947778931026, "learning_rate": 7.977918451529855e-06, "loss": 0.4383, "step": 16220 }, { "epoch": 0.5779076188610007, "grad_norm": 1.17544797672899, "learning_rate": 7.976788365670361e-06, "loss": 0.4329, "step": 16221 }, { "epoch": 0.5779432459874949, "grad_norm": 1.554766139796284, "learning_rate": 7.97565830675134e-06, "loss": 0.5096, "step": 16222 }, { "epoch": 0.577978873113989, "grad_norm": 1.2829661011590014, "learning_rate": 7.974528274787843e-06, "loss": 0.4704, "step": 16223 }, { "epoch": 0.5780145002404831, "grad_norm": 1.1369118664112132, "learning_rate": 7.973398269794919e-06, "loss": 0.4495, "step": 16224 }, { "epoch": 0.5780501273669773, "grad_norm": 1.4514566010074865, "learning_rate": 7.972268291787616e-06, "loss": 0.5669, "step": 16225 }, { "epoch": 0.5780857544934713, "grad_norm": 1.112669475396207, "learning_rate": 7.971138340780975e-06, "loss": 0.3964, "step": 16226 }, { "epoch": 0.5781213816199654, "grad_norm": 1.765274987005289, "learning_rate": 7.970008416790046e-06, "loss": 0.5518, "step": 16227 }, { "epoch": 0.5781570087464596, "grad_norm": 1.2570382533711852, "learning_rate": 7.968878519829872e-06, "loss": 0.4985, "step": 16228 }, { "epoch": 0.5781926358729537, "grad_norm": 1.496279811723435, "learning_rate": 7.967748649915502e-06, "loss": 0.5479, "step": 16229 }, { "epoch": 0.5782282629994477, "grad_norm": 1.4393071139825397, "learning_rate": 7.96661880706198e-06, "loss": 0.5171, "step": 16230 }, { "epoch": 0.5782638901259419, "grad_norm": 1.5039167846920043, "learning_rate": 7.965488991284346e-06, "loss": 0.4756, "step": 16231 }, { "epoch": 0.578299517252436, "grad_norm": 1.7546156244586433, "learning_rate": 7.964359202597654e-06, "loss": 0.4997, "step": 16232 }, { "epoch": 0.5783351443789301, "grad_norm": 1.447709923038968, "learning_rate": 7.963229441016938e-06, "loss": 0.5695, "step": 16233 }, { "epoch": 0.5783707715054243, "grad_norm": 1.567371354380285, "learning_rate": 7.962099706557248e-06, "loss": 0.4726, "step": 16234 }, { "epoch": 0.5784063986319183, "grad_norm": 1.2011656470800485, "learning_rate": 7.960969999233622e-06, "loss": 0.4234, "step": 16235 }, { "epoch": 0.5784420257584124, "grad_norm": 1.092466607096412, "learning_rate": 7.959840319061108e-06, "loss": 0.4471, "step": 16236 }, { "epoch": 0.5784776528849066, "grad_norm": 2.0740633197456733, "learning_rate": 7.958710666054744e-06, "loss": 0.545, "step": 16237 }, { "epoch": 0.5785132800114007, "grad_norm": 1.3577151917531414, "learning_rate": 7.957581040229577e-06, "loss": 0.5958, "step": 16238 }, { "epoch": 0.5785489071378948, "grad_norm": 1.2702084490100605, "learning_rate": 7.956451441600642e-06, "loss": 0.4163, "step": 16239 }, { "epoch": 0.5785845342643889, "grad_norm": 2.878061465976712, "learning_rate": 7.955321870182985e-06, "loss": 0.5283, "step": 16240 }, { "epoch": 0.578620161390883, "grad_norm": 1.1931912520155974, "learning_rate": 7.95419232599165e-06, "loss": 0.4491, "step": 16241 }, { "epoch": 0.5786557885173771, "grad_norm": 1.4638681387800647, "learning_rate": 7.953062809041671e-06, "loss": 0.5075, "step": 16242 }, { "epoch": 0.5786914156438713, "grad_norm": 1.6610638868438325, "learning_rate": 7.951933319348095e-06, "loss": 0.5579, "step": 16243 }, { "epoch": 0.5787270427703654, "grad_norm": 1.1079056783091341, "learning_rate": 7.950803856925955e-06, "loss": 0.4859, "step": 16244 }, { "epoch": 0.5787626698968594, "grad_norm": 1.4982195167434926, "learning_rate": 7.949674421790297e-06, "loss": 0.5773, "step": 16245 }, { "epoch": 0.5787982970233536, "grad_norm": 4.31988893577488, "learning_rate": 7.948545013956152e-06, "loss": 0.6367, "step": 16246 }, { "epoch": 0.5788339241498477, "grad_norm": 2.0784218457755728, "learning_rate": 7.94741563343857e-06, "loss": 0.5996, "step": 16247 }, { "epoch": 0.5788695512763418, "grad_norm": 1.1046427288925063, "learning_rate": 7.94628628025258e-06, "loss": 0.4261, "step": 16248 }, { "epoch": 0.578905178402836, "grad_norm": 1.4599730524561385, "learning_rate": 7.945156954413226e-06, "loss": 0.6089, "step": 16249 }, { "epoch": 0.57894080552933, "grad_norm": 1.0710525622463682, "learning_rate": 7.94402765593554e-06, "loss": 0.4627, "step": 16250 }, { "epoch": 0.5789764326558241, "grad_norm": 1.2621501630712633, "learning_rate": 7.942898384834564e-06, "loss": 0.5456, "step": 16251 }, { "epoch": 0.5790120597823183, "grad_norm": 1.0573789350035203, "learning_rate": 7.941769141125338e-06, "loss": 0.4375, "step": 16252 }, { "epoch": 0.5790476869088124, "grad_norm": 1.427876161935778, "learning_rate": 7.940639924822886e-06, "loss": 0.4695, "step": 16253 }, { "epoch": 0.5790833140353064, "grad_norm": 1.4300724813628123, "learning_rate": 7.939510735942262e-06, "loss": 0.449, "step": 16254 }, { "epoch": 0.5791189411618006, "grad_norm": 1.6565128676625183, "learning_rate": 7.938381574498485e-06, "loss": 0.4954, "step": 16255 }, { "epoch": 0.5791545682882947, "grad_norm": 0.980374735832495, "learning_rate": 7.937252440506604e-06, "loss": 0.4346, "step": 16256 }, { "epoch": 0.5791901954147888, "grad_norm": 1.119933533482284, "learning_rate": 7.936123333981645e-06, "loss": 0.4061, "step": 16257 }, { "epoch": 0.579225822541283, "grad_norm": 1.411707802700412, "learning_rate": 7.93499425493865e-06, "loss": 0.5868, "step": 16258 }, { "epoch": 0.579261449667777, "grad_norm": 1.2339453484726635, "learning_rate": 7.933865203392646e-06, "loss": 0.5189, "step": 16259 }, { "epoch": 0.5792970767942711, "grad_norm": 1.113968667789272, "learning_rate": 7.932736179358674e-06, "loss": 0.453, "step": 16260 }, { "epoch": 0.5793327039207653, "grad_norm": 1.9457686648988, "learning_rate": 7.931607182851762e-06, "loss": 0.5023, "step": 16261 }, { "epoch": 0.5793683310472594, "grad_norm": 1.3827226582384864, "learning_rate": 7.930478213886947e-06, "loss": 0.4997, "step": 16262 }, { "epoch": 0.5794039581737535, "grad_norm": 1.690043642366866, "learning_rate": 7.929349272479264e-06, "loss": 0.5698, "step": 16263 }, { "epoch": 0.5794395853002476, "grad_norm": 1.122469531806391, "learning_rate": 7.92822035864374e-06, "loss": 0.4319, "step": 16264 }, { "epoch": 0.5794752124267417, "grad_norm": 1.3935518046759576, "learning_rate": 7.927091472395409e-06, "loss": 0.5818, "step": 16265 }, { "epoch": 0.5795108395532358, "grad_norm": 1.042965038145878, "learning_rate": 7.925962613749304e-06, "loss": 0.428, "step": 16266 }, { "epoch": 0.57954646667973, "grad_norm": 1.154899824838467, "learning_rate": 7.924833782720457e-06, "loss": 0.4738, "step": 16267 }, { "epoch": 0.579582093806224, "grad_norm": 1.703864970919487, "learning_rate": 7.923704979323899e-06, "loss": 0.5529, "step": 16268 }, { "epoch": 0.5796177209327181, "grad_norm": 1.535427652886784, "learning_rate": 7.922576203574664e-06, "loss": 0.6412, "step": 16269 }, { "epoch": 0.5796533480592123, "grad_norm": 1.5824262638387963, "learning_rate": 7.921447455487775e-06, "loss": 0.5893, "step": 16270 }, { "epoch": 0.5796889751857064, "grad_norm": 1.2589106012022238, "learning_rate": 7.920318735078267e-06, "loss": 0.4934, "step": 16271 }, { "epoch": 0.5797246023122005, "grad_norm": 1.2434555912747096, "learning_rate": 7.919190042361166e-06, "loss": 0.4756, "step": 16272 }, { "epoch": 0.5797602294386947, "grad_norm": 1.244529042712781, "learning_rate": 7.918061377351509e-06, "loss": 0.5401, "step": 16273 }, { "epoch": 0.5797958565651887, "grad_norm": 1.874592716206313, "learning_rate": 7.916932740064315e-06, "loss": 0.6096, "step": 16274 }, { "epoch": 0.5798314836916828, "grad_norm": 1.1987499164921747, "learning_rate": 7.91580413051462e-06, "loss": 0.4702, "step": 16275 }, { "epoch": 0.579867110818177, "grad_norm": 2.834365635294571, "learning_rate": 7.91467554871745e-06, "loss": 0.5513, "step": 16276 }, { "epoch": 0.5799027379446711, "grad_norm": 1.05405290351783, "learning_rate": 7.91354699468783e-06, "loss": 0.4382, "step": 16277 }, { "epoch": 0.5799383650711651, "grad_norm": 1.1125653849669854, "learning_rate": 7.912418468440794e-06, "loss": 0.4467, "step": 16278 }, { "epoch": 0.5799739921976593, "grad_norm": 1.7340046299596337, "learning_rate": 7.91128996999136e-06, "loss": 0.477, "step": 16279 }, { "epoch": 0.5800096193241534, "grad_norm": 1.8195235549317004, "learning_rate": 7.910161499354567e-06, "loss": 0.5348, "step": 16280 }, { "epoch": 0.5800452464506475, "grad_norm": 1.2877366686533132, "learning_rate": 7.90903305654543e-06, "loss": 0.5124, "step": 16281 }, { "epoch": 0.5800808735771417, "grad_norm": 1.4570331260081764, "learning_rate": 7.90790464157898e-06, "loss": 0.4711, "step": 16282 }, { "epoch": 0.5801165007036357, "grad_norm": 1.2785091287693464, "learning_rate": 7.90677625447024e-06, "loss": 0.5849, "step": 16283 }, { "epoch": 0.5801521278301298, "grad_norm": 1.1721461571990548, "learning_rate": 7.905647895234243e-06, "loss": 0.4884, "step": 16284 }, { "epoch": 0.580187754956624, "grad_norm": 1.4936498755848389, "learning_rate": 7.904519563886003e-06, "loss": 0.4715, "step": 16285 }, { "epoch": 0.5802233820831181, "grad_norm": 1.5617355053210207, "learning_rate": 7.903391260440549e-06, "loss": 0.5562, "step": 16286 }, { "epoch": 0.5802590092096122, "grad_norm": 1.0024101506752283, "learning_rate": 7.90226298491291e-06, "loss": 0.3887, "step": 16287 }, { "epoch": 0.5802946363361063, "grad_norm": 1.015318454093576, "learning_rate": 7.901134737318102e-06, "loss": 0.4077, "step": 16288 }, { "epoch": 0.5803302634626004, "grad_norm": 1.141059305913819, "learning_rate": 7.900006517671155e-06, "loss": 0.4392, "step": 16289 }, { "epoch": 0.5803658905890945, "grad_norm": 1.609300423004133, "learning_rate": 7.898878325987086e-06, "loss": 0.5551, "step": 16290 }, { "epoch": 0.5804015177155887, "grad_norm": 1.240769509360796, "learning_rate": 7.897750162280924e-06, "loss": 0.5099, "step": 16291 }, { "epoch": 0.5804371448420828, "grad_norm": 1.1573562431783484, "learning_rate": 7.896622026567685e-06, "loss": 0.4418, "step": 16292 }, { "epoch": 0.5804727719685768, "grad_norm": 1.1157465961613424, "learning_rate": 7.895493918862395e-06, "loss": 0.4184, "step": 16293 }, { "epoch": 0.580508399095071, "grad_norm": 1.0618165647653866, "learning_rate": 7.894365839180074e-06, "loss": 0.426, "step": 16294 }, { "epoch": 0.5805440262215651, "grad_norm": 1.382068203695658, "learning_rate": 7.893237787535746e-06, "loss": 0.5184, "step": 16295 }, { "epoch": 0.5805796533480592, "grad_norm": 1.1175803007080596, "learning_rate": 7.892109763944426e-06, "loss": 0.4266, "step": 16296 }, { "epoch": 0.5806152804745534, "grad_norm": 1.3946274795123592, "learning_rate": 7.890981768421138e-06, "loss": 0.4878, "step": 16297 }, { "epoch": 0.5806509076010474, "grad_norm": 1.227067164300097, "learning_rate": 7.889853800980905e-06, "loss": 0.5474, "step": 16298 }, { "epoch": 0.5806865347275415, "grad_norm": 1.1708206927104403, "learning_rate": 7.888725861638739e-06, "loss": 0.4376, "step": 16299 }, { "epoch": 0.5807221618540357, "grad_norm": 1.2032088452772889, "learning_rate": 7.88759795040967e-06, "loss": 0.4214, "step": 16300 }, { "epoch": 0.5807577889805298, "grad_norm": 1.5963162017525567, "learning_rate": 7.886470067308705e-06, "loss": 0.5769, "step": 16301 }, { "epoch": 0.5807934161070238, "grad_norm": 1.7484082392234346, "learning_rate": 7.885342212350872e-06, "loss": 0.6169, "step": 16302 }, { "epoch": 0.580829043233518, "grad_norm": 1.8383715687485394, "learning_rate": 7.884214385551183e-06, "loss": 0.5801, "step": 16303 }, { "epoch": 0.5808646703600121, "grad_norm": 1.156593093658328, "learning_rate": 7.88308658692466e-06, "loss": 0.4108, "step": 16304 }, { "epoch": 0.5809002974865062, "grad_norm": 1.199771486544874, "learning_rate": 7.881958816486317e-06, "loss": 0.4099, "step": 16305 }, { "epoch": 0.5809359246130004, "grad_norm": 1.1833615218021567, "learning_rate": 7.880831074251176e-06, "loss": 0.4413, "step": 16306 }, { "epoch": 0.5809715517394944, "grad_norm": 1.1429861334639244, "learning_rate": 7.879703360234247e-06, "loss": 0.448, "step": 16307 }, { "epoch": 0.5810071788659885, "grad_norm": 1.1087124671507615, "learning_rate": 7.878575674450551e-06, "loss": 0.4032, "step": 16308 }, { "epoch": 0.5810428059924827, "grad_norm": 1.2002614340068543, "learning_rate": 7.877448016915104e-06, "loss": 0.4429, "step": 16309 }, { "epoch": 0.5810784331189768, "grad_norm": 1.41342421366521, "learning_rate": 7.876320387642916e-06, "loss": 0.5549, "step": 16310 }, { "epoch": 0.5811140602454709, "grad_norm": 1.1985669215419772, "learning_rate": 7.875192786649012e-06, "loss": 0.4196, "step": 16311 }, { "epoch": 0.581149687371965, "grad_norm": 1.5446435357776598, "learning_rate": 7.874065213948399e-06, "loss": 0.5522, "step": 16312 }, { "epoch": 0.5811853144984591, "grad_norm": 1.2675794270134813, "learning_rate": 7.872937669556094e-06, "loss": 0.5384, "step": 16313 }, { "epoch": 0.5812209416249532, "grad_norm": 1.380620574072357, "learning_rate": 7.871810153487109e-06, "loss": 0.5626, "step": 16314 }, { "epoch": 0.5812565687514474, "grad_norm": 1.4258812181131768, "learning_rate": 7.870682665756461e-06, "loss": 0.5449, "step": 16315 }, { "epoch": 0.5812921958779415, "grad_norm": 1.1265855501826831, "learning_rate": 7.869555206379158e-06, "loss": 0.501, "step": 16316 }, { "epoch": 0.5813278230044355, "grad_norm": 1.5941783158067657, "learning_rate": 7.868427775370222e-06, "loss": 0.5189, "step": 16317 }, { "epoch": 0.5813634501309297, "grad_norm": 1.4965633580780673, "learning_rate": 7.867300372744658e-06, "loss": 0.5598, "step": 16318 }, { "epoch": 0.5813990772574238, "grad_norm": 1.6371872047648288, "learning_rate": 7.86617299851748e-06, "loss": 0.5518, "step": 16319 }, { "epoch": 0.5814347043839179, "grad_norm": 1.3461955482025898, "learning_rate": 7.865045652703697e-06, "loss": 0.5764, "step": 16320 }, { "epoch": 0.581470331510412, "grad_norm": 1.4808099325706192, "learning_rate": 7.863918335318325e-06, "loss": 0.5387, "step": 16321 }, { "epoch": 0.5815059586369061, "grad_norm": 1.1851224615426592, "learning_rate": 7.862791046376376e-06, "loss": 0.4777, "step": 16322 }, { "epoch": 0.5815415857634002, "grad_norm": 1.1587551978247157, "learning_rate": 7.861663785892857e-06, "loss": 0.4199, "step": 16323 }, { "epoch": 0.5815772128898944, "grad_norm": 1.6320831069319781, "learning_rate": 7.860536553882778e-06, "loss": 0.5656, "step": 16324 }, { "epoch": 0.5816128400163885, "grad_norm": 1.7303186493785283, "learning_rate": 7.85940935036115e-06, "loss": 0.5578, "step": 16325 }, { "epoch": 0.5816484671428825, "grad_norm": 1.1147942551541379, "learning_rate": 7.858282175342987e-06, "loss": 0.4197, "step": 16326 }, { "epoch": 0.5816840942693767, "grad_norm": 1.4136553784357393, "learning_rate": 7.857155028843288e-06, "loss": 0.5406, "step": 16327 }, { "epoch": 0.5817197213958708, "grad_norm": 2.9552707518976518, "learning_rate": 7.856027910877073e-06, "loss": 0.5471, "step": 16328 }, { "epoch": 0.5817553485223649, "grad_norm": 1.04909457500893, "learning_rate": 7.854900821459339e-06, "loss": 0.4597, "step": 16329 }, { "epoch": 0.5817909756488591, "grad_norm": 1.1615836736018816, "learning_rate": 7.853773760605104e-06, "loss": 0.4737, "step": 16330 }, { "epoch": 0.5818266027753531, "grad_norm": 1.1078275216166824, "learning_rate": 7.852646728329368e-06, "loss": 0.4558, "step": 16331 }, { "epoch": 0.5818622299018472, "grad_norm": 1.119374545418416, "learning_rate": 7.851519724647143e-06, "loss": 0.4653, "step": 16332 }, { "epoch": 0.5818978570283414, "grad_norm": 1.1413463281938157, "learning_rate": 7.85039274957344e-06, "loss": 0.4284, "step": 16333 }, { "epoch": 0.5819334841548355, "grad_norm": 1.2754232138840682, "learning_rate": 7.849265803123253e-06, "loss": 0.4877, "step": 16334 }, { "epoch": 0.5819691112813296, "grad_norm": 1.6664821861929346, "learning_rate": 7.848138885311599e-06, "loss": 0.5918, "step": 16335 }, { "epoch": 0.5820047384078237, "grad_norm": 1.0569964788099635, "learning_rate": 7.847011996153477e-06, "loss": 0.4098, "step": 16336 }, { "epoch": 0.5820403655343178, "grad_norm": 1.2127118839549762, "learning_rate": 7.845885135663898e-06, "loss": 0.4765, "step": 16337 }, { "epoch": 0.5820759926608119, "grad_norm": 1.224651683382736, "learning_rate": 7.844758303857862e-06, "loss": 0.4439, "step": 16338 }, { "epoch": 0.5821116197873061, "grad_norm": 1.4789881279913808, "learning_rate": 7.843631500750376e-06, "loss": 0.5578, "step": 16339 }, { "epoch": 0.5821472469138002, "grad_norm": 1.057886462154638, "learning_rate": 7.842504726356444e-06, "loss": 0.41, "step": 16340 }, { "epoch": 0.5821828740402942, "grad_norm": 1.0354250865147303, "learning_rate": 7.841377980691071e-06, "loss": 0.3897, "step": 16341 }, { "epoch": 0.5822185011667884, "grad_norm": 1.449238336149045, "learning_rate": 7.840251263769256e-06, "loss": 0.4952, "step": 16342 }, { "epoch": 0.5822541282932825, "grad_norm": 1.86725229093435, "learning_rate": 7.839124575606005e-06, "loss": 0.4671, "step": 16343 }, { "epoch": 0.5822897554197766, "grad_norm": 1.563306728257856, "learning_rate": 7.837997916216325e-06, "loss": 0.5379, "step": 16344 }, { "epoch": 0.5823253825462708, "grad_norm": 1.5220064745625175, "learning_rate": 7.83687128561521e-06, "loss": 0.5635, "step": 16345 }, { "epoch": 0.5823610096727648, "grad_norm": 1.1429669486840137, "learning_rate": 7.835744683817668e-06, "loss": 0.4485, "step": 16346 }, { "epoch": 0.5823966367992589, "grad_norm": 1.3745630669276918, "learning_rate": 7.834618110838695e-06, "loss": 0.4774, "step": 16347 }, { "epoch": 0.5824322639257531, "grad_norm": 1.6459634385075512, "learning_rate": 7.833491566693301e-06, "loss": 0.5972, "step": 16348 }, { "epoch": 0.5824678910522472, "grad_norm": 1.3131368171039897, "learning_rate": 7.832365051396476e-06, "loss": 0.4472, "step": 16349 }, { "epoch": 0.5825035181787412, "grad_norm": 1.3758438769465837, "learning_rate": 7.831238564963227e-06, "loss": 0.547, "step": 16350 }, { "epoch": 0.5825391453052354, "grad_norm": 1.174969668145006, "learning_rate": 7.830112107408553e-06, "loss": 0.4089, "step": 16351 }, { "epoch": 0.5825747724317295, "grad_norm": 1.4523962568481532, "learning_rate": 7.828985678747453e-06, "loss": 0.553, "step": 16352 }, { "epoch": 0.5826103995582236, "grad_norm": 1.961744023845563, "learning_rate": 7.827859278994924e-06, "loss": 0.5873, "step": 16353 }, { "epoch": 0.5826460266847178, "grad_norm": 1.028367241218876, "learning_rate": 7.826732908165966e-06, "loss": 0.4075, "step": 16354 }, { "epoch": 0.5826816538112118, "grad_norm": 1.0121730570817928, "learning_rate": 7.825606566275583e-06, "loss": 0.4057, "step": 16355 }, { "epoch": 0.5827172809377059, "grad_norm": 1.096472436338547, "learning_rate": 7.824480253338765e-06, "loss": 0.46, "step": 16356 }, { "epoch": 0.5827529080642001, "grad_norm": 1.3121074782305207, "learning_rate": 7.823353969370516e-06, "loss": 0.4519, "step": 16357 }, { "epoch": 0.5827885351906942, "grad_norm": 1.498857622275728, "learning_rate": 7.822227714385826e-06, "loss": 0.5437, "step": 16358 }, { "epoch": 0.5828241623171883, "grad_norm": 1.9603601330677962, "learning_rate": 7.821101488399702e-06, "loss": 0.5207, "step": 16359 }, { "epoch": 0.5828597894436824, "grad_norm": 1.0684635801403877, "learning_rate": 7.819975291427128e-06, "loss": 0.416, "step": 16360 }, { "epoch": 0.5828954165701765, "grad_norm": 1.1541417686745277, "learning_rate": 7.818849123483112e-06, "loss": 0.4411, "step": 16361 }, { "epoch": 0.5829310436966706, "grad_norm": 1.326245964517938, "learning_rate": 7.817722984582641e-06, "loss": 0.5298, "step": 16362 }, { "epoch": 0.5829666708231648, "grad_norm": 1.219540536430737, "learning_rate": 7.816596874740718e-06, "loss": 0.4283, "step": 16363 }, { "epoch": 0.5830022979496589, "grad_norm": 1.6387071196823795, "learning_rate": 7.815470793972331e-06, "loss": 0.5592, "step": 16364 }, { "epoch": 0.5830379250761529, "grad_norm": 1.0907586767231683, "learning_rate": 7.814344742292476e-06, "loss": 0.4003, "step": 16365 }, { "epoch": 0.5830735522026471, "grad_norm": 1.5238961208675619, "learning_rate": 7.81321871971615e-06, "loss": 0.6533, "step": 16366 }, { "epoch": 0.5831091793291412, "grad_norm": 1.0705670700315508, "learning_rate": 7.812092726258342e-06, "loss": 0.4296, "step": 16367 }, { "epoch": 0.5831448064556353, "grad_norm": 1.0898289211170888, "learning_rate": 7.810966761934053e-06, "loss": 0.4158, "step": 16368 }, { "epoch": 0.5831804335821295, "grad_norm": 1.1276569271248358, "learning_rate": 7.809840826758269e-06, "loss": 0.4376, "step": 16369 }, { "epoch": 0.5832160607086235, "grad_norm": 1.0235546400820827, "learning_rate": 7.80871492074599e-06, "loss": 0.4239, "step": 16370 }, { "epoch": 0.5832516878351176, "grad_norm": 1.4737273351607645, "learning_rate": 7.807589043912197e-06, "loss": 0.5652, "step": 16371 }, { "epoch": 0.5832873149616118, "grad_norm": 1.0927684131185744, "learning_rate": 7.806463196271892e-06, "loss": 0.4226, "step": 16372 }, { "epoch": 0.5833229420881059, "grad_norm": 1.0074176661417562, "learning_rate": 7.80533737784006e-06, "loss": 0.4446, "step": 16373 }, { "epoch": 0.5833585692145999, "grad_norm": 1.961768075776942, "learning_rate": 7.8042115886317e-06, "loss": 0.5893, "step": 16374 }, { "epoch": 0.5833941963410941, "grad_norm": 1.117996625720557, "learning_rate": 7.803085828661792e-06, "loss": 0.4423, "step": 16375 }, { "epoch": 0.5834298234675882, "grad_norm": 1.067744163224275, "learning_rate": 7.801960097945337e-06, "loss": 0.4025, "step": 16376 }, { "epoch": 0.5834654505940823, "grad_norm": 1.4342955757717162, "learning_rate": 7.800834396497313e-06, "loss": 0.5803, "step": 16377 }, { "epoch": 0.5835010777205765, "grad_norm": 1.5133905038980326, "learning_rate": 7.799708724332718e-06, "loss": 0.5244, "step": 16378 }, { "epoch": 0.5835367048470705, "grad_norm": 1.1094070059164713, "learning_rate": 7.798583081466542e-06, "loss": 0.4655, "step": 16379 }, { "epoch": 0.5835723319735646, "grad_norm": 1.0578766464363192, "learning_rate": 7.797457467913768e-06, "loss": 0.4511, "step": 16380 }, { "epoch": 0.5836079591000588, "grad_norm": 1.4623748141604591, "learning_rate": 7.796331883689392e-06, "loss": 0.4731, "step": 16381 }, { "epoch": 0.5836435862265529, "grad_norm": 1.351027430111369, "learning_rate": 7.795206328808394e-06, "loss": 0.5711, "step": 16382 }, { "epoch": 0.583679213353047, "grad_norm": 1.4834887868919338, "learning_rate": 7.794080803285765e-06, "loss": 0.598, "step": 16383 }, { "epoch": 0.5837148404795411, "grad_norm": 1.451087878246381, "learning_rate": 7.792955307136491e-06, "loss": 0.5587, "step": 16384 }, { "epoch": 0.5837504676060352, "grad_norm": 1.1391977470865144, "learning_rate": 7.791829840375563e-06, "loss": 0.419, "step": 16385 }, { "epoch": 0.5837860947325293, "grad_norm": 1.5951171756436273, "learning_rate": 7.790704403017962e-06, "loss": 0.5789, "step": 16386 }, { "epoch": 0.5838217218590235, "grad_norm": 1.0217076066421378, "learning_rate": 7.789578995078674e-06, "loss": 0.4298, "step": 16387 }, { "epoch": 0.5838573489855176, "grad_norm": 1.5061075299187954, "learning_rate": 7.788453616572689e-06, "loss": 0.5911, "step": 16388 }, { "epoch": 0.5838929761120117, "grad_norm": 1.211160660068318, "learning_rate": 7.787328267514988e-06, "loss": 0.4872, "step": 16389 }, { "epoch": 0.5839286032385058, "grad_norm": 1.4196942998049187, "learning_rate": 7.786202947920562e-06, "loss": 0.6114, "step": 16390 }, { "epoch": 0.5839642303649999, "grad_norm": 1.1597991811900548, "learning_rate": 7.785077657804386e-06, "loss": 0.4591, "step": 16391 }, { "epoch": 0.5839998574914941, "grad_norm": 1.5745437723193851, "learning_rate": 7.783952397181454e-06, "loss": 0.5268, "step": 16392 }, { "epoch": 0.5840354846179882, "grad_norm": 1.4792787586370524, "learning_rate": 7.78282716606674e-06, "loss": 0.5512, "step": 16393 }, { "epoch": 0.5840711117444822, "grad_norm": 1.1996028250905262, "learning_rate": 7.781701964475234e-06, "loss": 0.4339, "step": 16394 }, { "epoch": 0.5841067388709764, "grad_norm": 1.1831180714094753, "learning_rate": 7.780576792421917e-06, "loss": 0.478, "step": 16395 }, { "epoch": 0.5841423659974705, "grad_norm": 1.3328448409206586, "learning_rate": 7.779451649921774e-06, "loss": 0.5396, "step": 16396 }, { "epoch": 0.5841779931239646, "grad_norm": 1.162183438069957, "learning_rate": 7.77832653698978e-06, "loss": 0.4753, "step": 16397 }, { "epoch": 0.5842136202504588, "grad_norm": 1.6239827769347273, "learning_rate": 7.777201453640923e-06, "loss": 0.5429, "step": 16398 }, { "epoch": 0.5842492473769528, "grad_norm": 1.8714708057724512, "learning_rate": 7.77607639989018e-06, "loss": 0.5694, "step": 16399 }, { "epoch": 0.5842848745034469, "grad_norm": 1.6432585687438035, "learning_rate": 7.77495137575254e-06, "loss": 0.62, "step": 16400 }, { "epoch": 0.5843205016299411, "grad_norm": 1.5033467504649225, "learning_rate": 7.773826381242973e-06, "loss": 0.5438, "step": 16401 }, { "epoch": 0.5843561287564352, "grad_norm": 1.5828967980763118, "learning_rate": 7.772701416376462e-06, "loss": 0.5169, "step": 16402 }, { "epoch": 0.5843917558829292, "grad_norm": 1.3900399529949512, "learning_rate": 7.771576481167992e-06, "loss": 0.4359, "step": 16403 }, { "epoch": 0.5844273830094234, "grad_norm": 1.0918306873202632, "learning_rate": 7.770451575632536e-06, "loss": 0.4508, "step": 16404 }, { "epoch": 0.5844630101359175, "grad_norm": 1.5217514916951411, "learning_rate": 7.769326699785078e-06, "loss": 0.5763, "step": 16405 }, { "epoch": 0.5844986372624116, "grad_norm": 1.3611573843052014, "learning_rate": 7.768201853640593e-06, "loss": 0.413, "step": 16406 }, { "epoch": 0.5845342643889058, "grad_norm": 1.1079586812859026, "learning_rate": 7.767077037214063e-06, "loss": 0.4368, "step": 16407 }, { "epoch": 0.5845698915153998, "grad_norm": 1.3701894223244706, "learning_rate": 7.765952250520459e-06, "loss": 0.5395, "step": 16408 }, { "epoch": 0.5846055186418939, "grad_norm": 1.6012896298436055, "learning_rate": 7.764827493574765e-06, "loss": 0.5586, "step": 16409 }, { "epoch": 0.5846411457683881, "grad_norm": 1.0383179804111853, "learning_rate": 7.763702766391952e-06, "loss": 0.4706, "step": 16410 }, { "epoch": 0.5846767728948822, "grad_norm": 1.503169457244898, "learning_rate": 7.762578068987006e-06, "loss": 0.6271, "step": 16411 }, { "epoch": 0.5847124000213763, "grad_norm": 1.3539606548544294, "learning_rate": 7.76145340137489e-06, "loss": 0.5598, "step": 16412 }, { "epoch": 0.5847480271478704, "grad_norm": 1.1734871637483777, "learning_rate": 7.760328763570589e-06, "loss": 0.412, "step": 16413 }, { "epoch": 0.5847836542743645, "grad_norm": 1.1052822269724079, "learning_rate": 7.759204155589076e-06, "loss": 0.4013, "step": 16414 }, { "epoch": 0.5848192814008586, "grad_norm": 1.5626449800855828, "learning_rate": 7.758079577445325e-06, "loss": 0.5302, "step": 16415 }, { "epoch": 0.5848549085273528, "grad_norm": 2.189092815325578, "learning_rate": 7.756955029154316e-06, "loss": 0.5165, "step": 16416 }, { "epoch": 0.5848905356538469, "grad_norm": 1.1655960926371458, "learning_rate": 7.755830510731011e-06, "loss": 0.4472, "step": 16417 }, { "epoch": 0.5849261627803409, "grad_norm": 1.076212834702742, "learning_rate": 7.7547060221904e-06, "loss": 0.4358, "step": 16418 }, { "epoch": 0.5849617899068351, "grad_norm": 1.4498028264668152, "learning_rate": 7.753581563547442e-06, "loss": 0.5732, "step": 16419 }, { "epoch": 0.5849974170333292, "grad_norm": 1.1683381424810855, "learning_rate": 7.752457134817116e-06, "loss": 0.4402, "step": 16420 }, { "epoch": 0.5850330441598233, "grad_norm": 1.165972798961312, "learning_rate": 7.751332736014394e-06, "loss": 0.4454, "step": 16421 }, { "epoch": 0.5850686712863175, "grad_norm": 1.1486807631475415, "learning_rate": 7.750208367154253e-06, "loss": 0.4357, "step": 16422 }, { "epoch": 0.5851042984128115, "grad_norm": 1.0803154012258986, "learning_rate": 7.749084028251654e-06, "loss": 0.4422, "step": 16423 }, { "epoch": 0.5851399255393056, "grad_norm": 1.2801007758734968, "learning_rate": 7.747959719321575e-06, "loss": 0.4376, "step": 16424 }, { "epoch": 0.5851755526657998, "grad_norm": 1.0312502161538264, "learning_rate": 7.74683544037899e-06, "loss": 0.3931, "step": 16425 }, { "epoch": 0.5852111797922939, "grad_norm": 1.40436701478158, "learning_rate": 7.745711191438863e-06, "loss": 0.609, "step": 16426 }, { "epoch": 0.5852468069187879, "grad_norm": 2.2751584489430097, "learning_rate": 7.744586972516171e-06, "loss": 0.6038, "step": 16427 }, { "epoch": 0.5852824340452821, "grad_norm": 1.1792520729027671, "learning_rate": 7.743462783625878e-06, "loss": 0.4465, "step": 16428 }, { "epoch": 0.5853180611717762, "grad_norm": 1.1852287241194883, "learning_rate": 7.742338624782955e-06, "loss": 0.4421, "step": 16429 }, { "epoch": 0.5853536882982703, "grad_norm": 1.7535761577780828, "learning_rate": 7.741214496002368e-06, "loss": 0.5288, "step": 16430 }, { "epoch": 0.5853893154247645, "grad_norm": 1.3688604600362984, "learning_rate": 7.740090397299093e-06, "loss": 0.5985, "step": 16431 }, { "epoch": 0.5854249425512585, "grad_norm": 1.2935180112377265, "learning_rate": 7.738966328688091e-06, "loss": 0.4572, "step": 16432 }, { "epoch": 0.5854605696777526, "grad_norm": 1.518072714343376, "learning_rate": 7.737842290184337e-06, "loss": 0.5213, "step": 16433 }, { "epoch": 0.5854961968042468, "grad_norm": 1.4028606181080319, "learning_rate": 7.73671828180279e-06, "loss": 0.5609, "step": 16434 }, { "epoch": 0.5855318239307409, "grad_norm": 1.1879954544895166, "learning_rate": 7.735594303558422e-06, "loss": 0.4236, "step": 16435 }, { "epoch": 0.585567451057235, "grad_norm": 0.9911211978316614, "learning_rate": 7.734470355466199e-06, "loss": 0.442, "step": 16436 }, { "epoch": 0.5856030781837291, "grad_norm": 2.0247054443785024, "learning_rate": 7.733346437541085e-06, "loss": 0.5846, "step": 16437 }, { "epoch": 0.5856387053102232, "grad_norm": 1.539183070898216, "learning_rate": 7.732222549798051e-06, "loss": 0.4249, "step": 16438 }, { "epoch": 0.5856743324367173, "grad_norm": 1.4115230236882605, "learning_rate": 7.731098692252058e-06, "loss": 0.5432, "step": 16439 }, { "epoch": 0.5857099595632115, "grad_norm": 1.3299497174814703, "learning_rate": 7.729974864918071e-06, "loss": 0.4363, "step": 16440 }, { "epoch": 0.5857455866897056, "grad_norm": 1.162645722849245, "learning_rate": 7.728851067811055e-06, "loss": 0.4313, "step": 16441 }, { "epoch": 0.5857812138161996, "grad_norm": 1.4931237016923162, "learning_rate": 7.727727300945976e-06, "loss": 0.5069, "step": 16442 }, { "epoch": 0.5858168409426938, "grad_norm": 1.3522013067672165, "learning_rate": 7.726603564337791e-06, "loss": 0.5083, "step": 16443 }, { "epoch": 0.5858524680691879, "grad_norm": 2.1435494243038824, "learning_rate": 7.725479858001475e-06, "loss": 0.4739, "step": 16444 }, { "epoch": 0.585888095195682, "grad_norm": 1.4274512733781877, "learning_rate": 7.72435618195198e-06, "loss": 0.4788, "step": 16445 }, { "epoch": 0.5859237223221762, "grad_norm": 1.1432643327437093, "learning_rate": 7.723232536204277e-06, "loss": 0.435, "step": 16446 }, { "epoch": 0.5859593494486702, "grad_norm": 1.4434967229987148, "learning_rate": 7.72210892077332e-06, "loss": 0.5055, "step": 16447 }, { "epoch": 0.5859949765751643, "grad_norm": 1.7365597957429952, "learning_rate": 7.720985335674077e-06, "loss": 0.5314, "step": 16448 }, { "epoch": 0.5860306037016585, "grad_norm": 1.4026187852765328, "learning_rate": 7.719861780921508e-06, "loss": 0.5165, "step": 16449 }, { "epoch": 0.5860662308281526, "grad_norm": 1.9885744312960643, "learning_rate": 7.71873825653057e-06, "loss": 0.5929, "step": 16450 }, { "epoch": 0.5861018579546466, "grad_norm": 1.377483103020886, "learning_rate": 7.71761476251623e-06, "loss": 0.5137, "step": 16451 }, { "epoch": 0.5861374850811408, "grad_norm": 1.13675625391569, "learning_rate": 7.716491298893443e-06, "loss": 0.4362, "step": 16452 }, { "epoch": 0.5861731122076349, "grad_norm": 1.2476076391039874, "learning_rate": 7.715367865677173e-06, "loss": 0.4225, "step": 16453 }, { "epoch": 0.586208739334129, "grad_norm": 1.2948778773970044, "learning_rate": 7.714244462882372e-06, "loss": 0.5095, "step": 16454 }, { "epoch": 0.5862443664606232, "grad_norm": 1.4782099993400148, "learning_rate": 7.71312109052401e-06, "loss": 0.5748, "step": 16455 }, { "epoch": 0.5862799935871172, "grad_norm": 1.5119036152327041, "learning_rate": 7.711997748617033e-06, "loss": 0.5332, "step": 16456 }, { "epoch": 0.5863156207136113, "grad_norm": 1.2286874028905597, "learning_rate": 7.71087443717641e-06, "loss": 0.4194, "step": 16457 }, { "epoch": 0.5863512478401055, "grad_norm": 1.5235875938524055, "learning_rate": 7.709751156217088e-06, "loss": 0.5419, "step": 16458 }, { "epoch": 0.5863868749665996, "grad_norm": 1.5189193615766718, "learning_rate": 7.708627905754033e-06, "loss": 0.4613, "step": 16459 }, { "epoch": 0.5864225020930937, "grad_norm": 1.119692275497441, "learning_rate": 7.707504685802202e-06, "loss": 0.4425, "step": 16460 }, { "epoch": 0.5864581292195878, "grad_norm": 1.2021432253368038, "learning_rate": 7.706381496376545e-06, "loss": 0.4517, "step": 16461 }, { "epoch": 0.5864937563460819, "grad_norm": 1.2001362060368181, "learning_rate": 7.705258337492023e-06, "loss": 0.4436, "step": 16462 }, { "epoch": 0.586529383472576, "grad_norm": 2.007003289976611, "learning_rate": 7.704135209163589e-06, "loss": 0.6303, "step": 16463 }, { "epoch": 0.5865650105990702, "grad_norm": 1.186431905120314, "learning_rate": 7.703012111406203e-06, "loss": 0.4701, "step": 16464 }, { "epoch": 0.5866006377255643, "grad_norm": 1.744319392132354, "learning_rate": 7.701889044234811e-06, "loss": 0.5486, "step": 16465 }, { "epoch": 0.5866362648520583, "grad_norm": 1.1081883451850556, "learning_rate": 7.700766007664377e-06, "loss": 0.4702, "step": 16466 }, { "epoch": 0.5866718919785525, "grad_norm": 1.121690879804071, "learning_rate": 7.699643001709847e-06, "loss": 0.4175, "step": 16467 }, { "epoch": 0.5867075191050466, "grad_norm": 1.1855289260917805, "learning_rate": 7.69852002638618e-06, "loss": 0.4471, "step": 16468 }, { "epoch": 0.5867431462315407, "grad_norm": 1.7872515778103462, "learning_rate": 7.697397081708327e-06, "loss": 0.5815, "step": 16469 }, { "epoch": 0.5867787733580349, "grad_norm": 1.5432871734514904, "learning_rate": 7.69627416769124e-06, "loss": 0.5405, "step": 16470 }, { "epoch": 0.5868144004845289, "grad_norm": 1.3979431630482524, "learning_rate": 7.695151284349878e-06, "loss": 0.5619, "step": 16471 }, { "epoch": 0.586850027611023, "grad_norm": 1.1980279081736818, "learning_rate": 7.694028431699182e-06, "loss": 0.3924, "step": 16472 }, { "epoch": 0.5868856547375172, "grad_norm": 1.7166886583493515, "learning_rate": 7.692905609754113e-06, "loss": 0.4987, "step": 16473 }, { "epoch": 0.5869212818640113, "grad_norm": 1.5036132207064756, "learning_rate": 7.691782818529615e-06, "loss": 0.5383, "step": 16474 }, { "epoch": 0.5869569089905053, "grad_norm": 1.9294227132137338, "learning_rate": 7.690660058040647e-06, "loss": 0.5193, "step": 16475 }, { "epoch": 0.5869925361169995, "grad_norm": 3.091123509516063, "learning_rate": 7.68953732830215e-06, "loss": 0.6747, "step": 16476 }, { "epoch": 0.5870281632434936, "grad_norm": 1.0977576213830518, "learning_rate": 7.68841462932908e-06, "loss": 0.4684, "step": 16477 }, { "epoch": 0.5870637903699877, "grad_norm": 1.3324610738926845, "learning_rate": 7.687291961136383e-06, "loss": 0.416, "step": 16478 }, { "epoch": 0.5870994174964819, "grad_norm": 1.5397246992285283, "learning_rate": 7.686169323739014e-06, "loss": 0.6066, "step": 16479 }, { "epoch": 0.5871350446229759, "grad_norm": 1.0832543867092022, "learning_rate": 7.685046717151913e-06, "loss": 0.4408, "step": 16480 }, { "epoch": 0.58717067174947, "grad_norm": 1.461929044381304, "learning_rate": 7.683924141390033e-06, "loss": 0.5926, "step": 16481 }, { "epoch": 0.5872062988759642, "grad_norm": 1.1755521080575084, "learning_rate": 7.682801596468328e-06, "loss": 0.4724, "step": 16482 }, { "epoch": 0.5872419260024583, "grad_norm": 1.3325083494422898, "learning_rate": 7.681679082401734e-06, "loss": 0.4492, "step": 16483 }, { "epoch": 0.5872775531289524, "grad_norm": 1.2134651908159388, "learning_rate": 7.680556599205205e-06, "loss": 0.3996, "step": 16484 }, { "epoch": 0.5873131802554465, "grad_norm": 1.5107127974175254, "learning_rate": 7.679434146893685e-06, "loss": 0.5767, "step": 16485 }, { "epoch": 0.5873488073819406, "grad_norm": 1.1447242484092552, "learning_rate": 7.678311725482126e-06, "loss": 0.4343, "step": 16486 }, { "epoch": 0.5873844345084347, "grad_norm": 1.6282681927238567, "learning_rate": 7.677189334985464e-06, "loss": 0.5258, "step": 16487 }, { "epoch": 0.5874200616349289, "grad_norm": 1.066010425228029, "learning_rate": 7.676066975418652e-06, "loss": 0.42, "step": 16488 }, { "epoch": 0.587455688761423, "grad_norm": 1.4492830458864416, "learning_rate": 7.674944646796629e-06, "loss": 0.5687, "step": 16489 }, { "epoch": 0.587491315887917, "grad_norm": 1.690307288727854, "learning_rate": 7.67382234913435e-06, "loss": 0.5692, "step": 16490 }, { "epoch": 0.5875269430144112, "grad_norm": 1.101145152894645, "learning_rate": 7.672700082446747e-06, "loss": 0.4214, "step": 16491 }, { "epoch": 0.5875625701409053, "grad_norm": 1.534350915956591, "learning_rate": 7.67157784674877e-06, "loss": 0.5443, "step": 16492 }, { "epoch": 0.5875981972673994, "grad_norm": 1.4316109664453416, "learning_rate": 7.67045564205536e-06, "loss": 0.5289, "step": 16493 }, { "epoch": 0.5876338243938936, "grad_norm": 1.6592155039857392, "learning_rate": 7.669333468381462e-06, "loss": 0.5461, "step": 16494 }, { "epoch": 0.5876694515203876, "grad_norm": 1.33717205648226, "learning_rate": 7.66821132574202e-06, "loss": 0.4592, "step": 16495 }, { "epoch": 0.5877050786468817, "grad_norm": 1.3395707636054146, "learning_rate": 7.667089214151972e-06, "loss": 0.4778, "step": 16496 }, { "epoch": 0.5877407057733759, "grad_norm": 1.7538009184448207, "learning_rate": 7.665967133626265e-06, "loss": 0.556, "step": 16497 }, { "epoch": 0.58777633289987, "grad_norm": 1.119457512557398, "learning_rate": 7.664845084179833e-06, "loss": 0.4476, "step": 16498 }, { "epoch": 0.587811960026364, "grad_norm": 1.58936738242808, "learning_rate": 7.663723065827623e-06, "loss": 0.5579, "step": 16499 }, { "epoch": 0.5878475871528582, "grad_norm": 1.5166845805263505, "learning_rate": 7.662601078584572e-06, "loss": 0.4905, "step": 16500 }, { "epoch": 0.5878832142793523, "grad_norm": 1.3543447813316005, "learning_rate": 7.661479122465625e-06, "loss": 0.5658, "step": 16501 }, { "epoch": 0.5879188414058464, "grad_norm": 1.105986719753715, "learning_rate": 7.660357197485714e-06, "loss": 0.4394, "step": 16502 }, { "epoch": 0.5879544685323406, "grad_norm": 1.4599809127327912, "learning_rate": 7.659235303659784e-06, "loss": 0.5629, "step": 16503 }, { "epoch": 0.5879900956588346, "grad_norm": 1.4710286954234182, "learning_rate": 7.65811344100277e-06, "loss": 0.6064, "step": 16504 }, { "epoch": 0.5880257227853287, "grad_norm": 1.0836953217076586, "learning_rate": 7.656991609529613e-06, "loss": 0.4766, "step": 16505 }, { "epoch": 0.5880613499118229, "grad_norm": 1.4248125603725479, "learning_rate": 7.655869809255252e-06, "loss": 0.4972, "step": 16506 }, { "epoch": 0.588096977038317, "grad_norm": 1.1681348448969877, "learning_rate": 7.654748040194621e-06, "loss": 0.443, "step": 16507 }, { "epoch": 0.5881326041648111, "grad_norm": 1.3626450511836194, "learning_rate": 7.653626302362665e-06, "loss": 0.595, "step": 16508 }, { "epoch": 0.5881682312913052, "grad_norm": 1.0321927947597467, "learning_rate": 7.652504595774309e-06, "loss": 0.4058, "step": 16509 }, { "epoch": 0.5882038584177993, "grad_norm": 1.4086754314035592, "learning_rate": 7.651382920444495e-06, "loss": 0.5196, "step": 16510 }, { "epoch": 0.5882394855442934, "grad_norm": 1.2328168293536896, "learning_rate": 7.65026127638816e-06, "loss": 0.5011, "step": 16511 }, { "epoch": 0.5882751126707876, "grad_norm": 1.1137679311488766, "learning_rate": 7.64913966362024e-06, "loss": 0.4653, "step": 16512 }, { "epoch": 0.5883107397972817, "grad_norm": 1.498301836118986, "learning_rate": 7.648018082155666e-06, "loss": 0.5139, "step": 16513 }, { "epoch": 0.5883463669237757, "grad_norm": 1.0597182694003493, "learning_rate": 7.646896532009378e-06, "loss": 0.4439, "step": 16514 }, { "epoch": 0.5883819940502699, "grad_norm": 1.1367154866927673, "learning_rate": 7.645775013196305e-06, "loss": 0.4389, "step": 16515 }, { "epoch": 0.588417621176764, "grad_norm": 1.309583974632839, "learning_rate": 7.644653525731383e-06, "loss": 0.5393, "step": 16516 }, { "epoch": 0.5884532483032581, "grad_norm": 1.322122343912176, "learning_rate": 7.643532069629549e-06, "loss": 0.5288, "step": 16517 }, { "epoch": 0.5884888754297523, "grad_norm": 1.1094136660176916, "learning_rate": 7.642410644905726e-06, "loss": 0.4444, "step": 16518 }, { "epoch": 0.5885245025562463, "grad_norm": 1.4135818818515806, "learning_rate": 7.641289251574861e-06, "loss": 0.5402, "step": 16519 }, { "epoch": 0.5885601296827404, "grad_norm": 1.3529663636396985, "learning_rate": 7.640167889651872e-06, "loss": 0.5544, "step": 16520 }, { "epoch": 0.5885957568092346, "grad_norm": 1.412093754300019, "learning_rate": 7.639046559151702e-06, "loss": 0.5122, "step": 16521 }, { "epoch": 0.5886313839357287, "grad_norm": 1.384427342561659, "learning_rate": 7.637925260089273e-06, "loss": 0.4259, "step": 16522 }, { "epoch": 0.5886670110622227, "grad_norm": 1.0453489192480065, "learning_rate": 7.636803992479525e-06, "loss": 0.416, "step": 16523 }, { "epoch": 0.5887026381887169, "grad_norm": 1.0813172762967482, "learning_rate": 7.635682756337378e-06, "loss": 0.3978, "step": 16524 }, { "epoch": 0.588738265315211, "grad_norm": 1.055991647090643, "learning_rate": 7.634561551677773e-06, "loss": 0.4478, "step": 16525 }, { "epoch": 0.5887738924417051, "grad_norm": 1.0945163376139542, "learning_rate": 7.633440378515629e-06, "loss": 0.418, "step": 16526 }, { "epoch": 0.5888095195681993, "grad_norm": 2.2615360462478744, "learning_rate": 7.632319236865886e-06, "loss": 0.4334, "step": 16527 }, { "epoch": 0.5888451466946933, "grad_norm": 1.443557650933704, "learning_rate": 7.631198126743463e-06, "loss": 0.5941, "step": 16528 }, { "epoch": 0.5888807738211874, "grad_norm": 1.2712217984774268, "learning_rate": 7.630077048163292e-06, "loss": 0.5987, "step": 16529 }, { "epoch": 0.5889164009476816, "grad_norm": 2.011306589945476, "learning_rate": 7.628956001140305e-06, "loss": 0.5724, "step": 16530 }, { "epoch": 0.5889520280741757, "grad_norm": 1.4588480470864413, "learning_rate": 7.627834985689423e-06, "loss": 0.5511, "step": 16531 }, { "epoch": 0.5889876552006698, "grad_norm": 1.282479401737483, "learning_rate": 7.626714001825579e-06, "loss": 0.4303, "step": 16532 }, { "epoch": 0.5890232823271639, "grad_norm": 1.4797169755321278, "learning_rate": 7.6255930495636945e-06, "loss": 0.5755, "step": 16533 }, { "epoch": 0.589058909453658, "grad_norm": 1.0020189794588328, "learning_rate": 7.624472128918702e-06, "loss": 0.4111, "step": 16534 }, { "epoch": 0.5890945365801521, "grad_norm": 1.4349764260425768, "learning_rate": 7.62335123990552e-06, "loss": 0.5075, "step": 16535 }, { "epoch": 0.5891301637066463, "grad_norm": 1.7005742892659133, "learning_rate": 7.622230382539079e-06, "loss": 0.5064, "step": 16536 }, { "epoch": 0.5891657908331404, "grad_norm": 1.5261755498954617, "learning_rate": 7.6211095568343e-06, "loss": 0.5611, "step": 16537 }, { "epoch": 0.5892014179596344, "grad_norm": 1.522648960979548, "learning_rate": 7.619988762806114e-06, "loss": 0.574, "step": 16538 }, { "epoch": 0.5892370450861286, "grad_norm": 1.1514231195018592, "learning_rate": 7.618868000469438e-06, "loss": 0.4612, "step": 16539 }, { "epoch": 0.5892726722126227, "grad_norm": 1.3816856963209423, "learning_rate": 7.617747269839199e-06, "loss": 0.4791, "step": 16540 }, { "epoch": 0.5893082993391168, "grad_norm": 1.0889042300332525, "learning_rate": 7.616626570930321e-06, "loss": 0.4373, "step": 16541 }, { "epoch": 0.589343926465611, "grad_norm": 1.0836830795667816, "learning_rate": 7.615505903757725e-06, "loss": 0.4533, "step": 16542 }, { "epoch": 0.589379553592105, "grad_norm": 1.2586547720819472, "learning_rate": 7.614385268336336e-06, "loss": 0.4903, "step": 16543 }, { "epoch": 0.5894151807185991, "grad_norm": 1.5494683058021312, "learning_rate": 7.613264664681072e-06, "loss": 0.5543, "step": 16544 }, { "epoch": 0.5894508078450933, "grad_norm": 1.14731280693261, "learning_rate": 7.612144092806864e-06, "loss": 0.4466, "step": 16545 }, { "epoch": 0.5894864349715874, "grad_norm": 1.3439007116431798, "learning_rate": 7.611023552728621e-06, "loss": 0.5182, "step": 16546 }, { "epoch": 0.5895220620980814, "grad_norm": 1.5101208057865756, "learning_rate": 7.60990304446127e-06, "loss": 0.5311, "step": 16547 }, { "epoch": 0.5895576892245756, "grad_norm": 1.1501431261152208, "learning_rate": 7.608782568019729e-06, "loss": 0.4865, "step": 16548 }, { "epoch": 0.5895933163510697, "grad_norm": 1.067804092562068, "learning_rate": 7.607662123418924e-06, "loss": 0.4359, "step": 16549 }, { "epoch": 0.5896289434775638, "grad_norm": 1.3442323208107263, "learning_rate": 7.6065417106737646e-06, "loss": 0.5154, "step": 16550 }, { "epoch": 0.589664570604058, "grad_norm": 1.1694272465827835, "learning_rate": 7.605421329799176e-06, "loss": 0.4612, "step": 16551 }, { "epoch": 0.589700197730552, "grad_norm": 1.848039144332931, "learning_rate": 7.60430098081008e-06, "loss": 0.6108, "step": 16552 }, { "epoch": 0.5897358248570461, "grad_norm": 1.2041721507861123, "learning_rate": 7.603180663721387e-06, "loss": 0.418, "step": 16553 }, { "epoch": 0.5897714519835403, "grad_norm": 1.3323274823743974, "learning_rate": 7.602060378548023e-06, "loss": 0.5613, "step": 16554 }, { "epoch": 0.5898070791100344, "grad_norm": 1.5501782908552069, "learning_rate": 7.6009401253048956e-06, "loss": 0.4715, "step": 16555 }, { "epoch": 0.5898427062365285, "grad_norm": 1.0938210593720679, "learning_rate": 7.599819904006933e-06, "loss": 0.3916, "step": 16556 }, { "epoch": 0.5898783333630226, "grad_norm": 1.3908339935229854, "learning_rate": 7.598699714669042e-06, "loss": 0.5539, "step": 16557 }, { "epoch": 0.5899139604895167, "grad_norm": 1.4620489739108986, "learning_rate": 7.597579557306144e-06, "loss": 0.5068, "step": 16558 }, { "epoch": 0.5899495876160108, "grad_norm": 1.0478947577960291, "learning_rate": 7.5964594319331516e-06, "loss": 0.4135, "step": 16559 }, { "epoch": 0.589985214742505, "grad_norm": 1.1540453026559465, "learning_rate": 7.595339338564986e-06, "loss": 0.4429, "step": 16560 }, { "epoch": 0.5900208418689991, "grad_norm": 1.0724323591497478, "learning_rate": 7.5942192772165546e-06, "loss": 0.4608, "step": 16561 }, { "epoch": 0.5900564689954931, "grad_norm": 1.3103335840206396, "learning_rate": 7.593099247902775e-06, "loss": 0.4714, "step": 16562 }, { "epoch": 0.5900920961219873, "grad_norm": 2.0007431576999966, "learning_rate": 7.591979250638562e-06, "loss": 0.5636, "step": 16563 }, { "epoch": 0.5901277232484814, "grad_norm": 1.1857744739076173, "learning_rate": 7.5908592854388275e-06, "loss": 0.4454, "step": 16564 }, { "epoch": 0.5901633503749755, "grad_norm": 1.0963618229897063, "learning_rate": 7.58973935231849e-06, "loss": 0.4459, "step": 16565 }, { "epoch": 0.5901989775014697, "grad_norm": 1.2176793172757718, "learning_rate": 7.5886194512924525e-06, "loss": 0.4875, "step": 16566 }, { "epoch": 0.5902346046279637, "grad_norm": 1.1399963173859544, "learning_rate": 7.587499582375635e-06, "loss": 0.4433, "step": 16567 }, { "epoch": 0.5902702317544578, "grad_norm": 1.1894670090268131, "learning_rate": 7.586379745582945e-06, "loss": 0.4244, "step": 16568 }, { "epoch": 0.590305858880952, "grad_norm": 1.47182057610598, "learning_rate": 7.585259940929297e-06, "loss": 0.5618, "step": 16569 }, { "epoch": 0.5903414860074461, "grad_norm": 1.4465766457744327, "learning_rate": 7.584140168429599e-06, "loss": 0.5421, "step": 16570 }, { "epoch": 0.5903771131339401, "grad_norm": 1.2934009859425568, "learning_rate": 7.5830204280987675e-06, "loss": 0.5242, "step": 16571 }, { "epoch": 0.5904127402604343, "grad_norm": 1.161793259783757, "learning_rate": 7.581900719951705e-06, "loss": 0.4519, "step": 16572 }, { "epoch": 0.5904483673869284, "grad_norm": 1.1075549394388904, "learning_rate": 7.580781044003324e-06, "loss": 0.4308, "step": 16573 }, { "epoch": 0.5904839945134225, "grad_norm": 1.3231392785870195, "learning_rate": 7.579661400268535e-06, "loss": 0.4757, "step": 16574 }, { "epoch": 0.5905196216399167, "grad_norm": 1.1926025125834245, "learning_rate": 7.578541788762245e-06, "loss": 0.4943, "step": 16575 }, { "epoch": 0.5905552487664107, "grad_norm": 1.6282077682976672, "learning_rate": 7.5774222094993675e-06, "loss": 0.5519, "step": 16576 }, { "epoch": 0.5905908758929048, "grad_norm": 1.5318780941245402, "learning_rate": 7.576302662494803e-06, "loss": 0.4878, "step": 16577 }, { "epoch": 0.590626503019399, "grad_norm": 1.8495636825588984, "learning_rate": 7.575183147763465e-06, "loss": 0.5494, "step": 16578 }, { "epoch": 0.5906621301458931, "grad_norm": 1.1846142554267791, "learning_rate": 7.574063665320256e-06, "loss": 0.4509, "step": 16579 }, { "epoch": 0.5906977572723872, "grad_norm": 1.5250622056957817, "learning_rate": 7.572944215180089e-06, "loss": 0.5561, "step": 16580 }, { "epoch": 0.5907333843988813, "grad_norm": 1.19309265832593, "learning_rate": 7.571824797357859e-06, "loss": 0.4403, "step": 16581 }, { "epoch": 0.5907690115253754, "grad_norm": 1.1822662946861395, "learning_rate": 7.570705411868486e-06, "loss": 0.4615, "step": 16582 }, { "epoch": 0.5908046386518695, "grad_norm": 1.515124073539339, "learning_rate": 7.569586058726865e-06, "loss": 0.5399, "step": 16583 }, { "epoch": 0.5908402657783637, "grad_norm": 1.6876891667596674, "learning_rate": 7.568466737947905e-06, "loss": 0.5422, "step": 16584 }, { "epoch": 0.5908758929048578, "grad_norm": 1.0764832521846532, "learning_rate": 7.567347449546508e-06, "loss": 0.471, "step": 16585 }, { "epoch": 0.5909115200313518, "grad_norm": 1.1082288979258927, "learning_rate": 7.566228193537581e-06, "loss": 0.4693, "step": 16586 }, { "epoch": 0.590947147157846, "grad_norm": 1.242062992875173, "learning_rate": 7.56510896993603e-06, "loss": 0.4461, "step": 16587 }, { "epoch": 0.5909827742843401, "grad_norm": 1.083984820453624, "learning_rate": 7.563989778756752e-06, "loss": 0.395, "step": 16588 }, { "epoch": 0.5910184014108342, "grad_norm": 1.4359239126092653, "learning_rate": 7.562870620014654e-06, "loss": 0.5652, "step": 16589 }, { "epoch": 0.5910540285373284, "grad_norm": 1.3297903874372756, "learning_rate": 7.561751493724635e-06, "loss": 0.5145, "step": 16590 }, { "epoch": 0.5910896556638224, "grad_norm": 1.3082454871289313, "learning_rate": 7.560632399901604e-06, "loss": 0.4729, "step": 16591 }, { "epoch": 0.5911252827903165, "grad_norm": 1.0964268186325081, "learning_rate": 7.559513338560452e-06, "loss": 0.4547, "step": 16592 }, { "epoch": 0.5911609099168107, "grad_norm": 1.5089045941883423, "learning_rate": 7.558394309716088e-06, "loss": 0.5501, "step": 16593 }, { "epoch": 0.5911965370433048, "grad_norm": 1.6116341552087603, "learning_rate": 7.55727531338341e-06, "loss": 0.5116, "step": 16594 }, { "epoch": 0.5912321641697988, "grad_norm": 1.436952816406669, "learning_rate": 7.5561563495773196e-06, "loss": 0.5631, "step": 16595 }, { "epoch": 0.591267791296293, "grad_norm": 1.079406024588278, "learning_rate": 7.555037418312713e-06, "loss": 0.4342, "step": 16596 }, { "epoch": 0.5913034184227871, "grad_norm": 1.1887017505422104, "learning_rate": 7.553918519604493e-06, "loss": 0.4089, "step": 16597 }, { "epoch": 0.5913390455492812, "grad_norm": 1.1276887848361719, "learning_rate": 7.552799653467561e-06, "loss": 0.428, "step": 16598 }, { "epoch": 0.5913746726757754, "grad_norm": 1.1131177818387763, "learning_rate": 7.551680819916808e-06, "loss": 0.449, "step": 16599 }, { "epoch": 0.5914102998022694, "grad_norm": 1.1734768775795088, "learning_rate": 7.550562018967138e-06, "loss": 0.4309, "step": 16600 }, { "epoch": 0.5914459269287635, "grad_norm": 1.2321182195176523, "learning_rate": 7.549443250633446e-06, "loss": 0.4807, "step": 16601 }, { "epoch": 0.5914815540552577, "grad_norm": 1.1738281538300823, "learning_rate": 7.5483245149306326e-06, "loss": 0.4575, "step": 16602 }, { "epoch": 0.5915171811817518, "grad_norm": 1.8561914065263896, "learning_rate": 7.5472058118735895e-06, "loss": 0.4778, "step": 16603 }, { "epoch": 0.5915528083082459, "grad_norm": 1.1749565739240073, "learning_rate": 7.546087141477216e-06, "loss": 0.4484, "step": 16604 }, { "epoch": 0.59158843543474, "grad_norm": 1.1389422013384176, "learning_rate": 7.544968503756406e-06, "loss": 0.4627, "step": 16605 }, { "epoch": 0.5916240625612341, "grad_norm": 1.620632586421612, "learning_rate": 7.54384989872606e-06, "loss": 0.4704, "step": 16606 }, { "epoch": 0.5916596896877282, "grad_norm": 1.3396288085062484, "learning_rate": 7.542731326401063e-06, "loss": 0.5136, "step": 16607 }, { "epoch": 0.5916953168142224, "grad_norm": 1.4071657438997698, "learning_rate": 7.541612786796318e-06, "loss": 0.5772, "step": 16608 }, { "epoch": 0.5917309439407165, "grad_norm": 1.095789346220653, "learning_rate": 7.540494279926722e-06, "loss": 0.4343, "step": 16609 }, { "epoch": 0.5917665710672105, "grad_norm": 1.6143187178296414, "learning_rate": 7.5393758058071584e-06, "loss": 0.5548, "step": 16610 }, { "epoch": 0.5918021981937047, "grad_norm": 1.7648967765612322, "learning_rate": 7.538257364452529e-06, "loss": 0.5514, "step": 16611 }, { "epoch": 0.5918378253201988, "grad_norm": 1.3800943633473193, "learning_rate": 7.537138955877721e-06, "loss": 0.5525, "step": 16612 }, { "epoch": 0.5918734524466929, "grad_norm": 2.5799625621063806, "learning_rate": 7.536020580097634e-06, "loss": 0.5247, "step": 16613 }, { "epoch": 0.5919090795731871, "grad_norm": 1.710730520965953, "learning_rate": 7.53490223712715e-06, "loss": 0.4696, "step": 16614 }, { "epoch": 0.5919447066996811, "grad_norm": 1.7568289721252957, "learning_rate": 7.533783926981168e-06, "loss": 0.4843, "step": 16615 }, { "epoch": 0.5919803338261752, "grad_norm": 1.0560603875751837, "learning_rate": 7.5326656496745755e-06, "loss": 0.3921, "step": 16616 }, { "epoch": 0.5920159609526694, "grad_norm": 1.7368068658914475, "learning_rate": 7.531547405222268e-06, "loss": 0.5685, "step": 16617 }, { "epoch": 0.5920515880791635, "grad_norm": 1.4139985757528448, "learning_rate": 7.530429193639128e-06, "loss": 0.4405, "step": 16618 }, { "epoch": 0.5920872152056575, "grad_norm": 1.1212043309701245, "learning_rate": 7.52931101494005e-06, "loss": 0.419, "step": 16619 }, { "epoch": 0.5921228423321517, "grad_norm": 1.1047838600721662, "learning_rate": 7.528192869139923e-06, "loss": 0.4495, "step": 16620 }, { "epoch": 0.5921584694586458, "grad_norm": 1.2231380114806742, "learning_rate": 7.527074756253635e-06, "loss": 0.4525, "step": 16621 }, { "epoch": 0.5921940965851399, "grad_norm": 1.4709510668675652, "learning_rate": 7.5259566762960765e-06, "loss": 0.6782, "step": 16622 }, { "epoch": 0.5922297237116341, "grad_norm": 1.4353000965454015, "learning_rate": 7.524838629282133e-06, "loss": 0.5302, "step": 16623 }, { "epoch": 0.5922653508381281, "grad_norm": 1.3785266522462836, "learning_rate": 7.523720615226697e-06, "loss": 0.5748, "step": 16624 }, { "epoch": 0.5923009779646222, "grad_norm": 1.3183421675215559, "learning_rate": 7.522602634144648e-06, "loss": 0.4233, "step": 16625 }, { "epoch": 0.5923366050911164, "grad_norm": 1.0683947185982483, "learning_rate": 7.5214846860508775e-06, "loss": 0.4559, "step": 16626 }, { "epoch": 0.5923722322176105, "grad_norm": 1.427805340616508, "learning_rate": 7.52036677096027e-06, "loss": 0.5761, "step": 16627 }, { "epoch": 0.5924078593441046, "grad_norm": 1.1932836482791074, "learning_rate": 7.519248888887715e-06, "loss": 0.4257, "step": 16628 }, { "epoch": 0.5924434864705987, "grad_norm": 1.1545752175940494, "learning_rate": 7.5181310398480914e-06, "loss": 0.4125, "step": 16629 }, { "epoch": 0.5924791135970928, "grad_norm": 1.4496330689095385, "learning_rate": 7.51701322385629e-06, "loss": 0.6077, "step": 16630 }, { "epoch": 0.5925147407235869, "grad_norm": 1.8029652980898767, "learning_rate": 7.51589544092719e-06, "loss": 0.5653, "step": 16631 }, { "epoch": 0.5925503678500811, "grad_norm": 1.0235581530803242, "learning_rate": 7.51477769107568e-06, "loss": 0.3998, "step": 16632 }, { "epoch": 0.5925859949765752, "grad_norm": 1.1553011072830461, "learning_rate": 7.5136599743166425e-06, "loss": 0.4336, "step": 16633 }, { "epoch": 0.5926216221030692, "grad_norm": 2.2363864260764084, "learning_rate": 7.51254229066496e-06, "loss": 0.574, "step": 16634 }, { "epoch": 0.5926572492295634, "grad_norm": 1.4810034138301245, "learning_rate": 7.511424640135519e-06, "loss": 0.5958, "step": 16635 }, { "epoch": 0.5926928763560575, "grad_norm": 1.3948401798972552, "learning_rate": 7.510307022743194e-06, "loss": 0.4492, "step": 16636 }, { "epoch": 0.5927285034825516, "grad_norm": 1.7335415701819166, "learning_rate": 7.509189438502873e-06, "loss": 0.4818, "step": 16637 }, { "epoch": 0.5927641306090458, "grad_norm": 1.194538022912224, "learning_rate": 7.508071887429433e-06, "loss": 0.4585, "step": 16638 }, { "epoch": 0.5927997577355398, "grad_norm": 1.4388783649258992, "learning_rate": 7.506954369537764e-06, "loss": 0.5052, "step": 16639 }, { "epoch": 0.5928353848620339, "grad_norm": 1.0819872033911686, "learning_rate": 7.505836884842734e-06, "loss": 0.45, "step": 16640 }, { "epoch": 0.5928710119885281, "grad_norm": 1.2917829515771657, "learning_rate": 7.504719433359232e-06, "loss": 0.4279, "step": 16641 }, { "epoch": 0.5929066391150222, "grad_norm": 1.7068732057947769, "learning_rate": 7.5036020151021325e-06, "loss": 0.4939, "step": 16642 }, { "epoch": 0.5929422662415162, "grad_norm": 1.1789300247218997, "learning_rate": 7.502484630086318e-06, "loss": 0.4229, "step": 16643 }, { "epoch": 0.5929778933680104, "grad_norm": 1.5176254799162203, "learning_rate": 7.501367278326667e-06, "loss": 0.5788, "step": 16644 }, { "epoch": 0.5930135204945045, "grad_norm": 1.6148109072307537, "learning_rate": 7.5002499598380555e-06, "loss": 0.582, "step": 16645 }, { "epoch": 0.5930491476209986, "grad_norm": 1.5376588740240804, "learning_rate": 7.499132674635367e-06, "loss": 0.5749, "step": 16646 }, { "epoch": 0.5930847747474928, "grad_norm": 1.1722662513191955, "learning_rate": 7.4980154227334715e-06, "loss": 0.4752, "step": 16647 }, { "epoch": 0.5931204018739868, "grad_norm": 1.1567327344816758, "learning_rate": 7.496898204147249e-06, "loss": 0.424, "step": 16648 }, { "epoch": 0.5931560290004809, "grad_norm": 1.14978432122479, "learning_rate": 7.4957810188915766e-06, "loss": 0.4674, "step": 16649 }, { "epoch": 0.5931916561269751, "grad_norm": 1.192790616211701, "learning_rate": 7.4946638669813355e-06, "loss": 0.4499, "step": 16650 }, { "epoch": 0.5932272832534692, "grad_norm": 1.1392268863543962, "learning_rate": 7.493546748431391e-06, "loss": 0.4519, "step": 16651 }, { "epoch": 0.5932629103799633, "grad_norm": 1.8033354180967387, "learning_rate": 7.492429663256624e-06, "loss": 0.5849, "step": 16652 }, { "epoch": 0.5932985375064574, "grad_norm": 1.6162552367407192, "learning_rate": 7.491312611471908e-06, "loss": 0.514, "step": 16653 }, { "epoch": 0.5933341646329515, "grad_norm": 1.427723221057942, "learning_rate": 7.490195593092122e-06, "loss": 0.6143, "step": 16654 }, { "epoch": 0.5933697917594456, "grad_norm": 1.076620167026828, "learning_rate": 7.489078608132132e-06, "loss": 0.3946, "step": 16655 }, { "epoch": 0.5934054188859398, "grad_norm": 1.248148026107497, "learning_rate": 7.487961656606813e-06, "loss": 0.4265, "step": 16656 }, { "epoch": 0.5934410460124339, "grad_norm": 1.2010072292845924, "learning_rate": 7.486844738531047e-06, "loss": 0.4201, "step": 16657 }, { "epoch": 0.5934766731389279, "grad_norm": 2.032944880985481, "learning_rate": 7.485727853919695e-06, "loss": 0.5134, "step": 16658 }, { "epoch": 0.5935123002654221, "grad_norm": 1.1409973466031913, "learning_rate": 7.484611002787636e-06, "loss": 0.4319, "step": 16659 }, { "epoch": 0.5935479273919162, "grad_norm": 1.7161231900504328, "learning_rate": 7.483494185149738e-06, "loss": 0.5061, "step": 16660 }, { "epoch": 0.5935835545184103, "grad_norm": 1.1244134730944508, "learning_rate": 7.48237740102088e-06, "loss": 0.4802, "step": 16661 }, { "epoch": 0.5936191816449045, "grad_norm": 1.3444681275766295, "learning_rate": 7.4812606504159195e-06, "loss": 0.5609, "step": 16662 }, { "epoch": 0.5936548087713985, "grad_norm": 1.1620695630681042, "learning_rate": 7.480143933349737e-06, "loss": 0.402, "step": 16663 }, { "epoch": 0.5936904358978926, "grad_norm": 1.3872602282107407, "learning_rate": 7.479027249837196e-06, "loss": 0.4887, "step": 16664 }, { "epoch": 0.5937260630243868, "grad_norm": 1.0971636612870426, "learning_rate": 7.477910599893176e-06, "loss": 0.4597, "step": 16665 }, { "epoch": 0.5937616901508809, "grad_norm": 2.455903849272746, "learning_rate": 7.476793983532533e-06, "loss": 0.5391, "step": 16666 }, { "epoch": 0.593797317277375, "grad_norm": 1.325027405660416, "learning_rate": 7.475677400770142e-06, "loss": 0.4762, "step": 16667 }, { "epoch": 0.5938329444038691, "grad_norm": 1.212642360327939, "learning_rate": 7.474560851620873e-06, "loss": 0.4379, "step": 16668 }, { "epoch": 0.5938685715303632, "grad_norm": 1.1214388155197834, "learning_rate": 7.473444336099591e-06, "loss": 0.4114, "step": 16669 }, { "epoch": 0.5939041986568573, "grad_norm": 1.1882320997709748, "learning_rate": 7.472327854221164e-06, "loss": 0.4028, "step": 16670 }, { "epoch": 0.5939398257833515, "grad_norm": 1.1098793838081835, "learning_rate": 7.471211406000457e-06, "loss": 0.4333, "step": 16671 }, { "epoch": 0.5939754529098455, "grad_norm": 1.2762063746634649, "learning_rate": 7.470094991452341e-06, "loss": 0.4393, "step": 16672 }, { "epoch": 0.5940110800363396, "grad_norm": 1.0961901930987177, "learning_rate": 7.468978610591675e-06, "loss": 0.4024, "step": 16673 }, { "epoch": 0.5940467071628338, "grad_norm": 1.1073882491438636, "learning_rate": 7.4678622634333295e-06, "loss": 0.3875, "step": 16674 }, { "epoch": 0.5940823342893279, "grad_norm": 1.3377256123080632, "learning_rate": 7.466745949992167e-06, "loss": 0.3889, "step": 16675 }, { "epoch": 0.594117961415822, "grad_norm": 1.108653352608834, "learning_rate": 7.465629670283055e-06, "loss": 0.4185, "step": 16676 }, { "epoch": 0.5941535885423161, "grad_norm": 1.0660450697386286, "learning_rate": 7.464513424320852e-06, "loss": 0.4326, "step": 16677 }, { "epoch": 0.5941892156688102, "grad_norm": 1.4996655609572496, "learning_rate": 7.463397212120426e-06, "loss": 0.5472, "step": 16678 }, { "epoch": 0.5942248427953043, "grad_norm": 1.1786348433967113, "learning_rate": 7.462281033696639e-06, "loss": 0.4547, "step": 16679 }, { "epoch": 0.5942604699217985, "grad_norm": 1.1335173696793106, "learning_rate": 7.461164889064354e-06, "loss": 0.4148, "step": 16680 }, { "epoch": 0.5942960970482926, "grad_norm": 1.0355803675273152, "learning_rate": 7.460048778238436e-06, "loss": 0.41, "step": 16681 }, { "epoch": 0.5943317241747866, "grad_norm": 1.058501288122202, "learning_rate": 7.458932701233738e-06, "loss": 0.4308, "step": 16682 }, { "epoch": 0.5943673513012808, "grad_norm": 1.4150347515932458, "learning_rate": 7.4578166580651335e-06, "loss": 0.7115, "step": 16683 }, { "epoch": 0.5944029784277749, "grad_norm": 1.7230968736044707, "learning_rate": 7.456700648747473e-06, "loss": 0.5, "step": 16684 }, { "epoch": 0.594438605554269, "grad_norm": 1.2239697397512552, "learning_rate": 7.455584673295621e-06, "loss": 0.4645, "step": 16685 }, { "epoch": 0.5944742326807632, "grad_norm": 1.1678531065619613, "learning_rate": 7.454468731724438e-06, "loss": 0.4496, "step": 16686 }, { "epoch": 0.5945098598072572, "grad_norm": 1.1489116465092337, "learning_rate": 7.453352824048786e-06, "loss": 0.4711, "step": 16687 }, { "epoch": 0.5945454869337513, "grad_norm": 1.6786994291058008, "learning_rate": 7.452236950283518e-06, "loss": 0.549, "step": 16688 }, { "epoch": 0.5945811140602455, "grad_norm": 1.7769969657900964, "learning_rate": 7.451121110443496e-06, "loss": 0.5605, "step": 16689 }, { "epoch": 0.5946167411867396, "grad_norm": 1.3089372304195621, "learning_rate": 7.450005304543578e-06, "loss": 0.426, "step": 16690 }, { "epoch": 0.5946523683132336, "grad_norm": 1.1196094681993534, "learning_rate": 7.448889532598621e-06, "loss": 0.4841, "step": 16691 }, { "epoch": 0.5946879954397278, "grad_norm": 1.4177079545995657, "learning_rate": 7.447773794623487e-06, "loss": 0.4961, "step": 16692 }, { "epoch": 0.5947236225662219, "grad_norm": 1.6922771844293218, "learning_rate": 7.4466580906330254e-06, "loss": 0.6372, "step": 16693 }, { "epoch": 0.594759249692716, "grad_norm": 1.5543666081619503, "learning_rate": 7.445542420642097e-06, "loss": 0.5457, "step": 16694 }, { "epoch": 0.5947948768192102, "grad_norm": 1.1852107646409327, "learning_rate": 7.444426784665555e-06, "loss": 0.5036, "step": 16695 }, { "epoch": 0.5948305039457042, "grad_norm": 1.3058699284479622, "learning_rate": 7.443311182718258e-06, "loss": 0.5213, "step": 16696 }, { "epoch": 0.5948661310721983, "grad_norm": 1.8007578127292636, "learning_rate": 7.442195614815058e-06, "loss": 0.5344, "step": 16697 }, { "epoch": 0.5949017581986925, "grad_norm": 1.2461255025556979, "learning_rate": 7.441080080970816e-06, "loss": 0.4125, "step": 16698 }, { "epoch": 0.5949373853251866, "grad_norm": 1.369145001108453, "learning_rate": 7.439964581200376e-06, "loss": 0.5317, "step": 16699 }, { "epoch": 0.5949730124516807, "grad_norm": 1.9844275823207398, "learning_rate": 7.4388491155186e-06, "loss": 0.4729, "step": 16700 }, { "epoch": 0.5950086395781748, "grad_norm": 1.7580259860544647, "learning_rate": 7.437733683940336e-06, "loss": 0.4749, "step": 16701 }, { "epoch": 0.5950442667046689, "grad_norm": 1.1097009373714601, "learning_rate": 7.436618286480439e-06, "loss": 0.4281, "step": 16702 }, { "epoch": 0.595079893831163, "grad_norm": 1.4859736286568594, "learning_rate": 7.435502923153766e-06, "loss": 0.5412, "step": 16703 }, { "epoch": 0.5951155209576572, "grad_norm": 1.8514280008510635, "learning_rate": 7.434387593975158e-06, "loss": 0.4579, "step": 16704 }, { "epoch": 0.5951511480841513, "grad_norm": 1.1349173937305805, "learning_rate": 7.4332722989594754e-06, "loss": 0.4432, "step": 16705 }, { "epoch": 0.5951867752106453, "grad_norm": 1.3172692529950107, "learning_rate": 7.4321570381215645e-06, "loss": 0.4428, "step": 16706 }, { "epoch": 0.5952224023371395, "grad_norm": 1.7112827703013684, "learning_rate": 7.43104181147628e-06, "loss": 0.5491, "step": 16707 }, { "epoch": 0.5952580294636336, "grad_norm": 1.211117177108497, "learning_rate": 7.429926619038465e-06, "loss": 0.4138, "step": 16708 }, { "epoch": 0.5952936565901277, "grad_norm": 1.865728306043688, "learning_rate": 7.42881146082298e-06, "loss": 0.5355, "step": 16709 }, { "epoch": 0.5953292837166219, "grad_norm": 1.4453528414842947, "learning_rate": 7.427696336844663e-06, "loss": 0.5625, "step": 16710 }, { "epoch": 0.5953649108431159, "grad_norm": 1.5684140072655786, "learning_rate": 7.426581247118368e-06, "loss": 0.5723, "step": 16711 }, { "epoch": 0.59540053796961, "grad_norm": 1.2542971381619983, "learning_rate": 7.425466191658941e-06, "loss": 0.4018, "step": 16712 }, { "epoch": 0.5954361650961042, "grad_norm": 1.1417838796294935, "learning_rate": 7.424351170481232e-06, "loss": 0.4075, "step": 16713 }, { "epoch": 0.5954717922225983, "grad_norm": 1.628085930837596, "learning_rate": 7.423236183600091e-06, "loss": 0.5636, "step": 16714 }, { "epoch": 0.5955074193490923, "grad_norm": 1.7046115963257717, "learning_rate": 7.422121231030357e-06, "loss": 0.4622, "step": 16715 }, { "epoch": 0.5955430464755865, "grad_norm": 1.1946928025608576, "learning_rate": 7.421006312786882e-06, "loss": 0.4405, "step": 16716 }, { "epoch": 0.5955786736020806, "grad_norm": 1.0565105068461496, "learning_rate": 7.419891428884509e-06, "loss": 0.4506, "step": 16717 }, { "epoch": 0.5956143007285747, "grad_norm": 1.4471792300567508, "learning_rate": 7.41877657933809e-06, "loss": 0.5527, "step": 16718 }, { "epoch": 0.5956499278550689, "grad_norm": 1.2139889993955024, "learning_rate": 7.417661764162458e-06, "loss": 0.4535, "step": 16719 }, { "epoch": 0.595685554981563, "grad_norm": 1.363716940904289, "learning_rate": 7.416546983372472e-06, "loss": 0.5412, "step": 16720 }, { "epoch": 0.595721182108057, "grad_norm": 1.4079151773914933, "learning_rate": 7.415432236982962e-06, "loss": 0.5096, "step": 16721 }, { "epoch": 0.5957568092345512, "grad_norm": 1.4678806508223883, "learning_rate": 7.414317525008783e-06, "loss": 0.5011, "step": 16722 }, { "epoch": 0.5957924363610453, "grad_norm": 1.2555191496043911, "learning_rate": 7.413202847464769e-06, "loss": 0.4605, "step": 16723 }, { "epoch": 0.5958280634875394, "grad_norm": 1.7368001780431512, "learning_rate": 7.412088204365768e-06, "loss": 0.5698, "step": 16724 }, { "epoch": 0.5958636906140335, "grad_norm": 1.163355141816109, "learning_rate": 7.410973595726627e-06, "loss": 0.4512, "step": 16725 }, { "epoch": 0.5958993177405276, "grad_norm": 1.0878518933097152, "learning_rate": 7.409859021562177e-06, "loss": 0.4429, "step": 16726 }, { "epoch": 0.5959349448670217, "grad_norm": 1.9702150723891572, "learning_rate": 7.408744481887267e-06, "loss": 0.5817, "step": 16727 }, { "epoch": 0.5959705719935159, "grad_norm": 1.0701394514033686, "learning_rate": 7.4076299767167325e-06, "loss": 0.4158, "step": 16728 }, { "epoch": 0.59600619912001, "grad_norm": 1.5334678456034367, "learning_rate": 7.406515506065422e-06, "loss": 0.5171, "step": 16729 }, { "epoch": 0.596041826246504, "grad_norm": 1.1741671013122383, "learning_rate": 7.405401069948163e-06, "loss": 0.4305, "step": 16730 }, { "epoch": 0.5960774533729982, "grad_norm": 1.5844687075943742, "learning_rate": 7.404286668379808e-06, "loss": 0.5415, "step": 16731 }, { "epoch": 0.5961130804994923, "grad_norm": 2.2471484130841124, "learning_rate": 7.403172301375188e-06, "loss": 0.5042, "step": 16732 }, { "epoch": 0.5961487076259864, "grad_norm": 1.3939762708646728, "learning_rate": 7.4020579689491434e-06, "loss": 0.4778, "step": 16733 }, { "epoch": 0.5961843347524806, "grad_norm": 1.5370145127214303, "learning_rate": 7.400943671116513e-06, "loss": 0.4552, "step": 16734 }, { "epoch": 0.5962199618789746, "grad_norm": 1.6140611157165436, "learning_rate": 7.399829407892134e-06, "loss": 0.576, "step": 16735 }, { "epoch": 0.5962555890054687, "grad_norm": 1.0317617269243609, "learning_rate": 7.3987151792908475e-06, "loss": 0.4043, "step": 16736 }, { "epoch": 0.5962912161319629, "grad_norm": 1.0275700820408098, "learning_rate": 7.397600985327482e-06, "loss": 0.4933, "step": 16737 }, { "epoch": 0.596326843258457, "grad_norm": 1.0778964446808184, "learning_rate": 7.39648682601688e-06, "loss": 0.4389, "step": 16738 }, { "epoch": 0.596362470384951, "grad_norm": 1.3051709130779392, "learning_rate": 7.395372701373876e-06, "loss": 0.4092, "step": 16739 }, { "epoch": 0.5963980975114452, "grad_norm": 2.0438752722466234, "learning_rate": 7.394258611413308e-06, "loss": 0.551, "step": 16740 }, { "epoch": 0.5964337246379393, "grad_norm": 1.4041439515768372, "learning_rate": 7.393144556150003e-06, "loss": 0.4989, "step": 16741 }, { "epoch": 0.5964693517644334, "grad_norm": 1.232171996712589, "learning_rate": 7.392030535598802e-06, "loss": 0.4678, "step": 16742 }, { "epoch": 0.5965049788909276, "grad_norm": 1.1833813045821764, "learning_rate": 7.390916549774536e-06, "loss": 0.4207, "step": 16743 }, { "epoch": 0.5965406060174216, "grad_norm": 1.1317495748978397, "learning_rate": 7.389802598692044e-06, "loss": 0.4438, "step": 16744 }, { "epoch": 0.5965762331439157, "grad_norm": 1.6447668188455289, "learning_rate": 7.388688682366147e-06, "loss": 0.4777, "step": 16745 }, { "epoch": 0.5966118602704099, "grad_norm": 1.103583975405797, "learning_rate": 7.387574800811694e-06, "loss": 0.4287, "step": 16746 }, { "epoch": 0.596647487396904, "grad_norm": 1.1679966353491376, "learning_rate": 7.3864609540435036e-06, "loss": 0.4495, "step": 16747 }, { "epoch": 0.5966831145233981, "grad_norm": 1.178333736682724, "learning_rate": 7.385347142076412e-06, "loss": 0.4389, "step": 16748 }, { "epoch": 0.5967187416498922, "grad_norm": 1.1868312187471002, "learning_rate": 7.384233364925253e-06, "loss": 0.4791, "step": 16749 }, { "epoch": 0.5967543687763863, "grad_norm": 1.5981128918199476, "learning_rate": 7.383119622604853e-06, "loss": 0.4857, "step": 16750 }, { "epoch": 0.5967899959028804, "grad_norm": 1.6347237646209378, "learning_rate": 7.3820059151300485e-06, "loss": 0.5565, "step": 16751 }, { "epoch": 0.5968256230293746, "grad_norm": 1.382129988675124, "learning_rate": 7.3808922425156605e-06, "loss": 0.5219, "step": 16752 }, { "epoch": 0.5968612501558687, "grad_norm": 1.297301593560983, "learning_rate": 7.379778604776527e-06, "loss": 0.555, "step": 16753 }, { "epoch": 0.5968968772823627, "grad_norm": 1.1827697025910295, "learning_rate": 7.378665001927468e-06, "loss": 0.3888, "step": 16754 }, { "epoch": 0.5969325044088569, "grad_norm": 1.2017570832095896, "learning_rate": 7.377551433983322e-06, "loss": 0.4026, "step": 16755 }, { "epoch": 0.596968131535351, "grad_norm": 1.483703282381933, "learning_rate": 7.376437900958908e-06, "loss": 0.5816, "step": 16756 }, { "epoch": 0.5970037586618451, "grad_norm": 1.128796249707039, "learning_rate": 7.375324402869057e-06, "loss": 0.3959, "step": 16757 }, { "epoch": 0.5970393857883393, "grad_norm": 1.1749580078561512, "learning_rate": 7.374210939728596e-06, "loss": 0.4235, "step": 16758 }, { "epoch": 0.5970750129148333, "grad_norm": 1.5008913026687036, "learning_rate": 7.373097511552352e-06, "loss": 0.5723, "step": 16759 }, { "epoch": 0.5971106400413274, "grad_norm": 1.1564716610528718, "learning_rate": 7.3719841183551525e-06, "loss": 0.426, "step": 16760 }, { "epoch": 0.5971462671678216, "grad_norm": 1.4082182971867847, "learning_rate": 7.370870760151818e-06, "loss": 0.5044, "step": 16761 }, { "epoch": 0.5971818942943157, "grad_norm": 1.4346839310166386, "learning_rate": 7.369757436957183e-06, "loss": 0.4216, "step": 16762 }, { "epoch": 0.5972175214208097, "grad_norm": 1.5151820529448674, "learning_rate": 7.3686441487860615e-06, "loss": 0.5014, "step": 16763 }, { "epoch": 0.5972531485473039, "grad_norm": 1.1737762512695242, "learning_rate": 7.367530895653283e-06, "loss": 0.4262, "step": 16764 }, { "epoch": 0.597288775673798, "grad_norm": 1.4424739927647887, "learning_rate": 7.36641767757367e-06, "loss": 0.4533, "step": 16765 }, { "epoch": 0.5973244028002922, "grad_norm": 1.6392588749262893, "learning_rate": 7.36530449456205e-06, "loss": 0.4859, "step": 16766 }, { "epoch": 0.5973600299267863, "grad_norm": 1.5180657184919886, "learning_rate": 7.364191346633238e-06, "loss": 0.4864, "step": 16767 }, { "epoch": 0.5973956570532803, "grad_norm": 1.1577157297925864, "learning_rate": 7.363078233802063e-06, "loss": 0.4038, "step": 16768 }, { "epoch": 0.5974312841797745, "grad_norm": 2.392931935994168, "learning_rate": 7.3619651560833415e-06, "loss": 0.5053, "step": 16769 }, { "epoch": 0.5974669113062686, "grad_norm": 1.9106517052516458, "learning_rate": 7.360852113491899e-06, "loss": 0.5431, "step": 16770 }, { "epoch": 0.5975025384327627, "grad_norm": 1.4430900141161127, "learning_rate": 7.359739106042556e-06, "loss": 0.5696, "step": 16771 }, { "epoch": 0.5975381655592569, "grad_norm": 1.097993896202148, "learning_rate": 7.358626133750131e-06, "loss": 0.3864, "step": 16772 }, { "epoch": 0.597573792685751, "grad_norm": 1.2385698458265726, "learning_rate": 7.35751319662945e-06, "loss": 0.46, "step": 16773 }, { "epoch": 0.597609419812245, "grad_norm": 1.1219999181705906, "learning_rate": 7.3564002946953226e-06, "loss": 0.3738, "step": 16774 }, { "epoch": 0.5976450469387392, "grad_norm": 1.3911342515375231, "learning_rate": 7.355287427962577e-06, "loss": 0.4951, "step": 16775 }, { "epoch": 0.5976806740652333, "grad_norm": 1.0876567985276482, "learning_rate": 7.354174596446022e-06, "loss": 0.4598, "step": 16776 }, { "epoch": 0.5977163011917274, "grad_norm": 1.197734892285943, "learning_rate": 7.3530618001604885e-06, "loss": 0.4144, "step": 16777 }, { "epoch": 0.5977519283182215, "grad_norm": 1.157355420958329, "learning_rate": 7.351949039120782e-06, "loss": 0.4174, "step": 16778 }, { "epoch": 0.5977875554447156, "grad_norm": 1.3322377776937706, "learning_rate": 7.3508363133417286e-06, "loss": 0.3979, "step": 16779 }, { "epoch": 0.5978231825712097, "grad_norm": 1.3625028828984778, "learning_rate": 7.349723622838139e-06, "loss": 0.4704, "step": 16780 }, { "epoch": 0.5978588096977039, "grad_norm": 1.2555981450628626, "learning_rate": 7.34861096762483e-06, "loss": 0.4593, "step": 16781 }, { "epoch": 0.597894436824198, "grad_norm": 1.5535150610035557, "learning_rate": 7.347498347716624e-06, "loss": 0.5374, "step": 16782 }, { "epoch": 0.597930063950692, "grad_norm": 1.3030388520417358, "learning_rate": 7.346385763128325e-06, "loss": 0.4485, "step": 16783 }, { "epoch": 0.5979656910771862, "grad_norm": 1.5251292575394098, "learning_rate": 7.345273213874762e-06, "loss": 0.5866, "step": 16784 }, { "epoch": 0.5980013182036803, "grad_norm": 1.4528275835529727, "learning_rate": 7.344160699970736e-06, "loss": 0.5788, "step": 16785 }, { "epoch": 0.5980369453301744, "grad_norm": 1.3932434131253117, "learning_rate": 7.34304822143107e-06, "loss": 0.4385, "step": 16786 }, { "epoch": 0.5980725724566686, "grad_norm": 2.4357785834671595, "learning_rate": 7.34193577827057e-06, "loss": 0.506, "step": 16787 }, { "epoch": 0.5981081995831626, "grad_norm": 1.4683626851360598, "learning_rate": 7.34082337050406e-06, "loss": 0.4401, "step": 16788 }, { "epoch": 0.5981438267096567, "grad_norm": 1.6384672419244557, "learning_rate": 7.339710998146339e-06, "loss": 0.4888, "step": 16789 }, { "epoch": 0.5981794538361509, "grad_norm": 1.3005182994754712, "learning_rate": 7.338598661212229e-06, "loss": 0.4343, "step": 16790 }, { "epoch": 0.598215080962645, "grad_norm": 1.1527647800569396, "learning_rate": 7.337486359716535e-06, "loss": 0.4546, "step": 16791 }, { "epoch": 0.598250708089139, "grad_norm": 1.2436110327599614, "learning_rate": 7.3363740936740744e-06, "loss": 0.4095, "step": 16792 }, { "epoch": 0.5982863352156332, "grad_norm": 1.423269920574072, "learning_rate": 7.335261863099652e-06, "loss": 0.5184, "step": 16793 }, { "epoch": 0.5983219623421273, "grad_norm": 1.2400336150365434, "learning_rate": 7.334149668008078e-06, "loss": 0.4534, "step": 16794 }, { "epoch": 0.5983575894686214, "grad_norm": 1.1514031187359537, "learning_rate": 7.333037508414171e-06, "loss": 0.4757, "step": 16795 }, { "epoch": 0.5983932165951156, "grad_norm": 0.9741661732923637, "learning_rate": 7.331925384332729e-06, "loss": 0.3949, "step": 16796 }, { "epoch": 0.5984288437216096, "grad_norm": 1.2121408725044176, "learning_rate": 7.330813295778567e-06, "loss": 0.4901, "step": 16797 }, { "epoch": 0.5984644708481037, "grad_norm": 1.1924478713485087, "learning_rate": 7.329701242766491e-06, "loss": 0.4331, "step": 16798 }, { "epoch": 0.5985000979745979, "grad_norm": 1.4524177646722187, "learning_rate": 7.3285892253113115e-06, "loss": 0.5271, "step": 16799 }, { "epoch": 0.598535725101092, "grad_norm": 1.1159303253415227, "learning_rate": 7.3274772434278315e-06, "loss": 0.4077, "step": 16800 }, { "epoch": 0.5985713522275861, "grad_norm": 1.6402820942133889, "learning_rate": 7.326365297130861e-06, "loss": 0.5796, "step": 16801 }, { "epoch": 0.5986069793540802, "grad_norm": 1.3772295665517458, "learning_rate": 7.325253386435203e-06, "loss": 0.5172, "step": 16802 }, { "epoch": 0.5986426064805743, "grad_norm": 1.2316514064614372, "learning_rate": 7.324141511355671e-06, "loss": 0.4263, "step": 16803 }, { "epoch": 0.5986782336070684, "grad_norm": 1.116193862241954, "learning_rate": 7.32302967190706e-06, "loss": 0.4287, "step": 16804 }, { "epoch": 0.5987138607335626, "grad_norm": 1.4431085287190049, "learning_rate": 7.32191786810418e-06, "loss": 0.4776, "step": 16805 }, { "epoch": 0.5987494878600567, "grad_norm": 1.04895881993267, "learning_rate": 7.320806099961837e-06, "loss": 0.4088, "step": 16806 }, { "epoch": 0.5987851149865507, "grad_norm": 2.25067914289078, "learning_rate": 7.319694367494833e-06, "loss": 0.5829, "step": 16807 }, { "epoch": 0.5988207421130449, "grad_norm": 1.4936543834243066, "learning_rate": 7.318582670717973e-06, "loss": 0.5191, "step": 16808 }, { "epoch": 0.598856369239539, "grad_norm": 1.4744166411783792, "learning_rate": 7.317471009646057e-06, "loss": 0.5001, "step": 16809 }, { "epoch": 0.5988919963660331, "grad_norm": 1.1117709763461008, "learning_rate": 7.316359384293894e-06, "loss": 0.4034, "step": 16810 }, { "epoch": 0.5989276234925273, "grad_norm": 1.7502756623777393, "learning_rate": 7.315247794676276e-06, "loss": 0.4766, "step": 16811 }, { "epoch": 0.5989632506190213, "grad_norm": 1.5970156901230035, "learning_rate": 7.314136240808014e-06, "loss": 0.5553, "step": 16812 }, { "epoch": 0.5989988777455154, "grad_norm": 1.0738551946861181, "learning_rate": 7.313024722703902e-06, "loss": 0.4153, "step": 16813 }, { "epoch": 0.5990345048720096, "grad_norm": 1.186529735620152, "learning_rate": 7.311913240378748e-06, "loss": 0.4626, "step": 16814 }, { "epoch": 0.5990701319985037, "grad_norm": 1.3336058960596182, "learning_rate": 7.310801793847344e-06, "loss": 0.4278, "step": 16815 }, { "epoch": 0.5991057591249977, "grad_norm": 1.4740577166893845, "learning_rate": 7.309690383124494e-06, "loss": 0.5626, "step": 16816 }, { "epoch": 0.5991413862514919, "grad_norm": 1.629777779618167, "learning_rate": 7.308579008225001e-06, "loss": 0.5265, "step": 16817 }, { "epoch": 0.599177013377986, "grad_norm": 1.6029631575796068, "learning_rate": 7.307467669163655e-06, "loss": 0.5522, "step": 16818 }, { "epoch": 0.5992126405044801, "grad_norm": 1.397615074037426, "learning_rate": 7.3063563659552635e-06, "loss": 0.5714, "step": 16819 }, { "epoch": 0.5992482676309743, "grad_norm": 1.4883550871259539, "learning_rate": 7.305245098614614e-06, "loss": 0.5286, "step": 16820 }, { "epoch": 0.5992838947574683, "grad_norm": 1.250511100657375, "learning_rate": 7.304133867156516e-06, "loss": 0.39, "step": 16821 }, { "epoch": 0.5993195218839624, "grad_norm": 2.279644092808714, "learning_rate": 7.3030226715957555e-06, "loss": 0.5661, "step": 16822 }, { "epoch": 0.5993551490104566, "grad_norm": 1.1967928775121175, "learning_rate": 7.301911511947135e-06, "loss": 0.4433, "step": 16823 }, { "epoch": 0.5993907761369507, "grad_norm": 1.4954535136772227, "learning_rate": 7.300800388225446e-06, "loss": 0.5356, "step": 16824 }, { "epoch": 0.5994264032634448, "grad_norm": 1.7331653414001589, "learning_rate": 7.299689300445491e-06, "loss": 0.6909, "step": 16825 }, { "epoch": 0.599462030389939, "grad_norm": 1.1359655283340269, "learning_rate": 7.2985782486220565e-06, "loss": 0.4229, "step": 16826 }, { "epoch": 0.599497657516433, "grad_norm": 2.109784591533286, "learning_rate": 7.297467232769943e-06, "loss": 0.394, "step": 16827 }, { "epoch": 0.5995332846429271, "grad_norm": 1.4406114292358674, "learning_rate": 7.29635625290394e-06, "loss": 0.4782, "step": 16828 }, { "epoch": 0.5995689117694213, "grad_norm": 1.7118425937378918, "learning_rate": 7.295245309038843e-06, "loss": 0.5231, "step": 16829 }, { "epoch": 0.5996045388959154, "grad_norm": 1.54373344810293, "learning_rate": 7.29413440118945e-06, "loss": 0.5813, "step": 16830 }, { "epoch": 0.5996401660224094, "grad_norm": 1.126242130929383, "learning_rate": 7.293023529370544e-06, "loss": 0.3916, "step": 16831 }, { "epoch": 0.5996757931489036, "grad_norm": 1.664895218012449, "learning_rate": 7.291912693596924e-06, "loss": 0.5446, "step": 16832 }, { "epoch": 0.5997114202753977, "grad_norm": 1.1653446079099292, "learning_rate": 7.290801893883377e-06, "loss": 0.4277, "step": 16833 }, { "epoch": 0.5997470474018918, "grad_norm": 1.469549415774676, "learning_rate": 7.289691130244699e-06, "loss": 0.5191, "step": 16834 }, { "epoch": 0.599782674528386, "grad_norm": 1.60244353620695, "learning_rate": 7.288580402695676e-06, "loss": 0.4966, "step": 16835 }, { "epoch": 0.59981830165488, "grad_norm": 1.760853491046219, "learning_rate": 7.287469711251104e-06, "loss": 0.5215, "step": 16836 }, { "epoch": 0.5998539287813741, "grad_norm": 1.065433742641165, "learning_rate": 7.286359055925764e-06, "loss": 0.4052, "step": 16837 }, { "epoch": 0.5998895559078683, "grad_norm": 1.457254431758134, "learning_rate": 7.285248436734453e-06, "loss": 0.5844, "step": 16838 }, { "epoch": 0.5999251830343624, "grad_norm": 1.5051020376801214, "learning_rate": 7.284137853691954e-06, "loss": 0.5623, "step": 16839 }, { "epoch": 0.5999608101608565, "grad_norm": 1.1445107616212054, "learning_rate": 7.283027306813058e-06, "loss": 0.4466, "step": 16840 }, { "epoch": 0.5999964372873506, "grad_norm": 1.1478043774968252, "learning_rate": 7.2819167961125555e-06, "loss": 0.428, "step": 16841 }, { "epoch": 0.6000320644138447, "grad_norm": 1.4630792811252558, "learning_rate": 7.280806321605227e-06, "loss": 0.5461, "step": 16842 }, { "epoch": 0.6000676915403388, "grad_norm": 1.785143678620943, "learning_rate": 7.2796958833058665e-06, "loss": 0.5967, "step": 16843 }, { "epoch": 0.600103318666833, "grad_norm": 1.5071923017832938, "learning_rate": 7.278585481229253e-06, "loss": 0.546, "step": 16844 }, { "epoch": 0.600138945793327, "grad_norm": 1.7873277712765705, "learning_rate": 7.2774751153901805e-06, "loss": 0.5298, "step": 16845 }, { "epoch": 0.6001745729198211, "grad_norm": 1.581313943108299, "learning_rate": 7.276364785803425e-06, "loss": 0.5578, "step": 16846 }, { "epoch": 0.6002102000463153, "grad_norm": 1.7690609303160223, "learning_rate": 7.27525449248378e-06, "loss": 0.5395, "step": 16847 }, { "epoch": 0.6002458271728094, "grad_norm": 1.1530381378872452, "learning_rate": 7.274144235446024e-06, "loss": 0.4216, "step": 16848 }, { "epoch": 0.6002814542993035, "grad_norm": 1.6761067808638113, "learning_rate": 7.273034014704943e-06, "loss": 0.5744, "step": 16849 }, { "epoch": 0.6003170814257976, "grad_norm": 1.5307850789315214, "learning_rate": 7.271923830275318e-06, "loss": 0.5574, "step": 16850 }, { "epoch": 0.6003527085522917, "grad_norm": 1.153977684207854, "learning_rate": 7.270813682171934e-06, "loss": 0.4784, "step": 16851 }, { "epoch": 0.6003883356787858, "grad_norm": 1.6625123803145518, "learning_rate": 7.269703570409579e-06, "loss": 0.5268, "step": 16852 }, { "epoch": 0.60042396280528, "grad_norm": 1.1810346442347932, "learning_rate": 7.268593495003025e-06, "loss": 0.3796, "step": 16853 }, { "epoch": 0.6004595899317741, "grad_norm": 1.2146224938047, "learning_rate": 7.2674834559670585e-06, "loss": 0.4305, "step": 16854 }, { "epoch": 0.6004952170582681, "grad_norm": 1.2128888211472197, "learning_rate": 7.266373453316458e-06, "loss": 0.4495, "step": 16855 }, { "epoch": 0.6005308441847623, "grad_norm": 1.313506947276623, "learning_rate": 7.26526348706601e-06, "loss": 0.5691, "step": 16856 }, { "epoch": 0.6005664713112564, "grad_norm": 1.5871228714211791, "learning_rate": 7.264153557230487e-06, "loss": 0.5264, "step": 16857 }, { "epoch": 0.6006020984377505, "grad_norm": 1.1745696756032074, "learning_rate": 7.2630436638246715e-06, "loss": 0.4276, "step": 16858 }, { "epoch": 0.6006377255642447, "grad_norm": 1.1338329559723694, "learning_rate": 7.2619338068633425e-06, "loss": 0.4548, "step": 16859 }, { "epoch": 0.6006733526907387, "grad_norm": 1.566424741059214, "learning_rate": 7.26082398636128e-06, "loss": 0.5524, "step": 16860 }, { "epoch": 0.6007089798172328, "grad_norm": 1.4921307789617215, "learning_rate": 7.259714202333259e-06, "loss": 0.4886, "step": 16861 }, { "epoch": 0.600744606943727, "grad_norm": 1.577783816649398, "learning_rate": 7.258604454794058e-06, "loss": 0.5833, "step": 16862 }, { "epoch": 0.6007802340702211, "grad_norm": 1.3493824139118755, "learning_rate": 7.25749474375846e-06, "loss": 0.5982, "step": 16863 }, { "epoch": 0.6008158611967152, "grad_norm": 1.0114563953674955, "learning_rate": 7.256385069241232e-06, "loss": 0.3779, "step": 16864 }, { "epoch": 0.6008514883232093, "grad_norm": 1.0772658101920873, "learning_rate": 7.255275431257156e-06, "loss": 0.4395, "step": 16865 }, { "epoch": 0.6008871154497034, "grad_norm": 1.3400038977094932, "learning_rate": 7.254165829821004e-06, "loss": 0.5587, "step": 16866 }, { "epoch": 0.6009227425761975, "grad_norm": 1.5241323146100336, "learning_rate": 7.253056264947558e-06, "loss": 0.5386, "step": 16867 }, { "epoch": 0.6009583697026917, "grad_norm": 1.105881343878443, "learning_rate": 7.251946736651583e-06, "loss": 0.4207, "step": 16868 }, { "epoch": 0.6009939968291858, "grad_norm": 1.122929764401309, "learning_rate": 7.2508372449478605e-06, "loss": 0.3976, "step": 16869 }, { "epoch": 0.6010296239556798, "grad_norm": 1.1090416482801178, "learning_rate": 7.249727789851159e-06, "loss": 0.4402, "step": 16870 }, { "epoch": 0.601065251082174, "grad_norm": 1.090645573163632, "learning_rate": 7.248618371376259e-06, "loss": 0.405, "step": 16871 }, { "epoch": 0.6011008782086681, "grad_norm": 1.2078430556298658, "learning_rate": 7.247508989537922e-06, "loss": 0.4323, "step": 16872 }, { "epoch": 0.6011365053351622, "grad_norm": 1.1987797171589356, "learning_rate": 7.246399644350932e-06, "loss": 0.4658, "step": 16873 }, { "epoch": 0.6011721324616563, "grad_norm": 1.5593116564468037, "learning_rate": 7.2452903358300534e-06, "loss": 0.583, "step": 16874 }, { "epoch": 0.6012077595881504, "grad_norm": 1.0499062483328345, "learning_rate": 7.2441810639900575e-06, "loss": 0.4041, "step": 16875 }, { "epoch": 0.6012433867146445, "grad_norm": 0.9595733190822321, "learning_rate": 7.24307182884572e-06, "loss": 0.4249, "step": 16876 }, { "epoch": 0.6012790138411387, "grad_norm": 1.736639070657585, "learning_rate": 7.241962630411805e-06, "loss": 0.5333, "step": 16877 }, { "epoch": 0.6013146409676328, "grad_norm": 1.7343271818436412, "learning_rate": 7.2408534687030905e-06, "loss": 0.4726, "step": 16878 }, { "epoch": 0.6013502680941268, "grad_norm": 1.0895672328641126, "learning_rate": 7.239744343734336e-06, "loss": 0.4631, "step": 16879 }, { "epoch": 0.601385895220621, "grad_norm": 1.2305670474903279, "learning_rate": 7.238635255520318e-06, "loss": 0.577, "step": 16880 }, { "epoch": 0.6014215223471151, "grad_norm": 1.2500952131530858, "learning_rate": 7.237526204075798e-06, "loss": 0.4459, "step": 16881 }, { "epoch": 0.6014571494736092, "grad_norm": 1.2564107552733794, "learning_rate": 7.236417189415552e-06, "loss": 0.4766, "step": 16882 }, { "epoch": 0.6014927766001034, "grad_norm": 1.5957099260320564, "learning_rate": 7.2353082115543376e-06, "loss": 0.509, "step": 16883 }, { "epoch": 0.6015284037265974, "grad_norm": 1.5408330708111797, "learning_rate": 7.234199270506932e-06, "loss": 0.6055, "step": 16884 }, { "epoch": 0.6015640308530915, "grad_norm": 1.6348687687374737, "learning_rate": 7.233090366288092e-06, "loss": 0.5753, "step": 16885 }, { "epoch": 0.6015996579795857, "grad_norm": 1.5397560653179574, "learning_rate": 7.231981498912588e-06, "loss": 0.5796, "step": 16886 }, { "epoch": 0.6016352851060798, "grad_norm": 1.332636243097376, "learning_rate": 7.230872668395188e-06, "loss": 0.4233, "step": 16887 }, { "epoch": 0.6016709122325739, "grad_norm": 1.2621595574869942, "learning_rate": 7.229763874750651e-06, "loss": 0.4791, "step": 16888 }, { "epoch": 0.601706539359068, "grad_norm": 1.1394490076029014, "learning_rate": 7.228655117993748e-06, "loss": 0.4868, "step": 16889 }, { "epoch": 0.6017421664855621, "grad_norm": 1.1084720815469262, "learning_rate": 7.227546398139236e-06, "loss": 0.4484, "step": 16890 }, { "epoch": 0.6017777936120562, "grad_norm": 1.0584772649675689, "learning_rate": 7.226437715201882e-06, "loss": 0.4311, "step": 16891 }, { "epoch": 0.6018134207385504, "grad_norm": 1.6346503294312824, "learning_rate": 7.2253290691964485e-06, "loss": 0.5335, "step": 16892 }, { "epoch": 0.6018490478650445, "grad_norm": 1.4494978653112698, "learning_rate": 7.224220460137701e-06, "loss": 0.5327, "step": 16893 }, { "epoch": 0.6018846749915385, "grad_norm": 1.2422959456904743, "learning_rate": 7.223111888040394e-06, "loss": 0.4506, "step": 16894 }, { "epoch": 0.6019203021180327, "grad_norm": 1.5782666585087175, "learning_rate": 7.222003352919296e-06, "loss": 0.523, "step": 16895 }, { "epoch": 0.6019559292445268, "grad_norm": 1.1210927174474095, "learning_rate": 7.220894854789161e-06, "loss": 0.4317, "step": 16896 }, { "epoch": 0.6019915563710209, "grad_norm": 1.069607807671841, "learning_rate": 7.219786393664755e-06, "loss": 0.4319, "step": 16897 }, { "epoch": 0.602027183497515, "grad_norm": 1.1675745328203933, "learning_rate": 7.2186779695608375e-06, "loss": 0.4733, "step": 16898 }, { "epoch": 0.6020628106240091, "grad_norm": 1.198212553008149, "learning_rate": 7.217569582492165e-06, "loss": 0.4523, "step": 16899 }, { "epoch": 0.6020984377505032, "grad_norm": 1.7200451298841555, "learning_rate": 7.2164612324735035e-06, "loss": 0.5825, "step": 16900 }, { "epoch": 0.6021340648769974, "grad_norm": 2.5770538055654084, "learning_rate": 7.215352919519603e-06, "loss": 0.5408, "step": 16901 }, { "epoch": 0.6021696920034915, "grad_norm": 1.1031783281438319, "learning_rate": 7.214244643645224e-06, "loss": 0.4256, "step": 16902 }, { "epoch": 0.6022053191299855, "grad_norm": 1.6342321911368594, "learning_rate": 7.213136404865124e-06, "loss": 0.5513, "step": 16903 }, { "epoch": 0.6022409462564797, "grad_norm": 1.698254019446769, "learning_rate": 7.212028203194066e-06, "loss": 0.4856, "step": 16904 }, { "epoch": 0.6022765733829738, "grad_norm": 1.187107834861053, "learning_rate": 7.210920038646795e-06, "loss": 0.459, "step": 16905 }, { "epoch": 0.6023122005094679, "grad_norm": 1.399499500859072, "learning_rate": 7.209811911238076e-06, "loss": 0.5249, "step": 16906 }, { "epoch": 0.6023478276359621, "grad_norm": 1.3879208229501363, "learning_rate": 7.208703820982659e-06, "loss": 0.5967, "step": 16907 }, { "epoch": 0.6023834547624561, "grad_norm": 15.067928600545702, "learning_rate": 7.207595767895303e-06, "loss": 0.5096, "step": 16908 }, { "epoch": 0.6024190818889502, "grad_norm": 1.1422135285259507, "learning_rate": 7.206487751990763e-06, "loss": 0.4546, "step": 16909 }, { "epoch": 0.6024547090154444, "grad_norm": 1.5328378048727795, "learning_rate": 7.205379773283787e-06, "loss": 0.5442, "step": 16910 }, { "epoch": 0.6024903361419385, "grad_norm": 1.1212571842795305, "learning_rate": 7.204271831789138e-06, "loss": 0.4349, "step": 16911 }, { "epoch": 0.6025259632684326, "grad_norm": 1.2036559429687992, "learning_rate": 7.2031639275215606e-06, "loss": 0.4, "step": 16912 }, { "epoch": 0.6025615903949267, "grad_norm": 1.661458363790356, "learning_rate": 7.202056060495811e-06, "loss": 0.4744, "step": 16913 }, { "epoch": 0.6025972175214208, "grad_norm": 0.9679117456109521, "learning_rate": 7.200948230726639e-06, "loss": 0.4092, "step": 16914 }, { "epoch": 0.6026328446479149, "grad_norm": 1.4361329803629812, "learning_rate": 7.199840438228803e-06, "loss": 0.5299, "step": 16915 }, { "epoch": 0.6026684717744091, "grad_norm": 1.2231047800245938, "learning_rate": 7.1987326830170434e-06, "loss": 0.4943, "step": 16916 }, { "epoch": 0.6027040989009032, "grad_norm": 1.411130928910769, "learning_rate": 7.197624965106118e-06, "loss": 0.5263, "step": 16917 }, { "epoch": 0.6027397260273972, "grad_norm": 1.5953936214586162, "learning_rate": 7.196517284510774e-06, "loss": 0.5218, "step": 16918 }, { "epoch": 0.6027753531538914, "grad_norm": 1.114161187889017, "learning_rate": 7.195409641245765e-06, "loss": 0.4507, "step": 16919 }, { "epoch": 0.6028109802803855, "grad_norm": 1.1446911312452932, "learning_rate": 7.194302035325833e-06, "loss": 0.4278, "step": 16920 }, { "epoch": 0.6028466074068796, "grad_norm": 1.1046642668939974, "learning_rate": 7.193194466765727e-06, "loss": 0.4526, "step": 16921 }, { "epoch": 0.6028822345333738, "grad_norm": 1.1669269115086112, "learning_rate": 7.192086935580207e-06, "loss": 0.4382, "step": 16922 }, { "epoch": 0.6029178616598678, "grad_norm": 1.1252208272929434, "learning_rate": 7.1909794417840075e-06, "loss": 0.4492, "step": 16923 }, { "epoch": 0.6029534887863619, "grad_norm": 1.4573043073834369, "learning_rate": 7.1898719853918805e-06, "loss": 0.6727, "step": 16924 }, { "epoch": 0.6029891159128561, "grad_norm": 1.2614957018680932, "learning_rate": 7.1887645664185724e-06, "loss": 0.4255, "step": 16925 }, { "epoch": 0.6030247430393502, "grad_norm": 1.1508925588511485, "learning_rate": 7.187657184878833e-06, "loss": 0.44, "step": 16926 }, { "epoch": 0.6030603701658442, "grad_norm": 1.7059631096781216, "learning_rate": 7.186549840787398e-06, "loss": 0.5241, "step": 16927 }, { "epoch": 0.6030959972923384, "grad_norm": 1.061724465961239, "learning_rate": 7.185442534159021e-06, "loss": 0.3995, "step": 16928 }, { "epoch": 0.6031316244188325, "grad_norm": 1.511309533392406, "learning_rate": 7.184335265008442e-06, "loss": 0.5174, "step": 16929 }, { "epoch": 0.6031672515453266, "grad_norm": 1.443425971174541, "learning_rate": 7.183228033350411e-06, "loss": 0.5058, "step": 16930 }, { "epoch": 0.6032028786718208, "grad_norm": 1.1536710364734806, "learning_rate": 7.1821208391996644e-06, "loss": 0.4428, "step": 16931 }, { "epoch": 0.6032385057983148, "grad_norm": 1.121358564410649, "learning_rate": 7.1810136825709474e-06, "loss": 0.442, "step": 16932 }, { "epoch": 0.6032741329248089, "grad_norm": 1.243343783389603, "learning_rate": 7.179906563479006e-06, "loss": 0.4381, "step": 16933 }, { "epoch": 0.6033097600513031, "grad_norm": 1.3360978677976396, "learning_rate": 7.1787994819385785e-06, "loss": 0.5848, "step": 16934 }, { "epoch": 0.6033453871777972, "grad_norm": 1.123705381592639, "learning_rate": 7.177692437964408e-06, "loss": 0.418, "step": 16935 }, { "epoch": 0.6033810143042913, "grad_norm": 1.9485643690795715, "learning_rate": 7.176585431571235e-06, "loss": 0.5822, "step": 16936 }, { "epoch": 0.6034166414307854, "grad_norm": 1.3063882628785741, "learning_rate": 7.175478462773804e-06, "loss": 0.5579, "step": 16937 }, { "epoch": 0.6034522685572795, "grad_norm": 1.1727331225861317, "learning_rate": 7.174371531586848e-06, "loss": 0.4199, "step": 16938 }, { "epoch": 0.6034878956837736, "grad_norm": 1.7703174977484146, "learning_rate": 7.1732646380251106e-06, "loss": 0.5539, "step": 16939 }, { "epoch": 0.6035235228102678, "grad_norm": 5.999614595054659, "learning_rate": 7.172157782103328e-06, "loss": 0.5699, "step": 16940 }, { "epoch": 0.6035591499367619, "grad_norm": 1.2033898209822012, "learning_rate": 7.171050963836248e-06, "loss": 0.4528, "step": 16941 }, { "epoch": 0.6035947770632559, "grad_norm": 1.4295315922525258, "learning_rate": 7.169944183238595e-06, "loss": 0.5414, "step": 16942 }, { "epoch": 0.6036304041897501, "grad_norm": 1.590931328484128, "learning_rate": 7.1688374403251134e-06, "loss": 0.4882, "step": 16943 }, { "epoch": 0.6036660313162442, "grad_norm": 1.0652463230364175, "learning_rate": 7.167730735110544e-06, "loss": 0.4502, "step": 16944 }, { "epoch": 0.6037016584427383, "grad_norm": 1.1129832652334717, "learning_rate": 7.166624067609616e-06, "loss": 0.3928, "step": 16945 }, { "epoch": 0.6037372855692325, "grad_norm": 1.4591983350475517, "learning_rate": 7.165517437837074e-06, "loss": 0.5471, "step": 16946 }, { "epoch": 0.6037729126957265, "grad_norm": 1.764563309576208, "learning_rate": 7.1644108458076415e-06, "loss": 0.5348, "step": 16947 }, { "epoch": 0.6038085398222206, "grad_norm": 1.657945325505943, "learning_rate": 7.163304291536067e-06, "loss": 0.459, "step": 16948 }, { "epoch": 0.6038441669487148, "grad_norm": 1.2653431564297497, "learning_rate": 7.1621977750370745e-06, "loss": 0.4362, "step": 16949 }, { "epoch": 0.6038797940752089, "grad_norm": 1.195034980941008, "learning_rate": 7.161091296325404e-06, "loss": 0.3895, "step": 16950 }, { "epoch": 0.6039154212017029, "grad_norm": 1.0814154418016138, "learning_rate": 7.1599848554157855e-06, "loss": 0.3671, "step": 16951 }, { "epoch": 0.6039510483281971, "grad_norm": 1.2404585654513742, "learning_rate": 7.158878452322957e-06, "loss": 0.4548, "step": 16952 }, { "epoch": 0.6039866754546912, "grad_norm": 1.1896943299452984, "learning_rate": 7.157772087061645e-06, "loss": 0.4144, "step": 16953 }, { "epoch": 0.6040223025811853, "grad_norm": 1.1210719788773582, "learning_rate": 7.156665759646586e-06, "loss": 0.421, "step": 16954 }, { "epoch": 0.6040579297076795, "grad_norm": 1.182199903868033, "learning_rate": 7.1555594700925065e-06, "loss": 0.423, "step": 16955 }, { "epoch": 0.6040935568341735, "grad_norm": 2.0687683435657833, "learning_rate": 7.154453218414142e-06, "loss": 0.5262, "step": 16956 }, { "epoch": 0.6041291839606676, "grad_norm": 1.478204280383127, "learning_rate": 7.1533470046262255e-06, "loss": 0.4591, "step": 16957 }, { "epoch": 0.6041648110871618, "grad_norm": 1.096551266016024, "learning_rate": 7.1522408287434774e-06, "loss": 0.4061, "step": 16958 }, { "epoch": 0.6042004382136559, "grad_norm": 1.384575894700726, "learning_rate": 7.1511346907806386e-06, "loss": 0.5609, "step": 16959 }, { "epoch": 0.60423606534015, "grad_norm": 1.4852865003483664, "learning_rate": 7.150028590752429e-06, "loss": 0.4287, "step": 16960 }, { "epoch": 0.6042716924666441, "grad_norm": 1.3493653579366436, "learning_rate": 7.148922528673583e-06, "loss": 0.4142, "step": 16961 }, { "epoch": 0.6043073195931382, "grad_norm": 1.218636124468817, "learning_rate": 7.147816504558824e-06, "loss": 0.4173, "step": 16962 }, { "epoch": 0.6043429467196323, "grad_norm": 1.860317564424691, "learning_rate": 7.146710518422885e-06, "loss": 0.5359, "step": 16963 }, { "epoch": 0.6043785738461265, "grad_norm": 1.1574166932846623, "learning_rate": 7.145604570280487e-06, "loss": 0.4135, "step": 16964 }, { "epoch": 0.6044142009726206, "grad_norm": 1.3289717362718392, "learning_rate": 7.144498660146362e-06, "loss": 0.4447, "step": 16965 }, { "epoch": 0.6044498280991146, "grad_norm": 1.1090060006339653, "learning_rate": 7.143392788035229e-06, "loss": 0.417, "step": 16966 }, { "epoch": 0.6044854552256088, "grad_norm": 1.2739439848897458, "learning_rate": 7.142286953961819e-06, "loss": 0.445, "step": 16967 }, { "epoch": 0.6045210823521029, "grad_norm": 1.15846003153129, "learning_rate": 7.141181157940859e-06, "loss": 0.4532, "step": 16968 }, { "epoch": 0.604556709478597, "grad_norm": 1.937796338752674, "learning_rate": 7.140075399987068e-06, "loss": 0.5238, "step": 16969 }, { "epoch": 0.6045923366050912, "grad_norm": 2.694927422504509, "learning_rate": 7.138969680115171e-06, "loss": 0.5542, "step": 16970 }, { "epoch": 0.6046279637315852, "grad_norm": 1.128139133100174, "learning_rate": 7.137863998339892e-06, "loss": 0.4185, "step": 16971 }, { "epoch": 0.6046635908580793, "grad_norm": 1.1837665052360131, "learning_rate": 7.136758354675956e-06, "loss": 0.4709, "step": 16972 }, { "epoch": 0.6046992179845735, "grad_norm": 1.27577879854499, "learning_rate": 7.135652749138082e-06, "loss": 0.4592, "step": 16973 }, { "epoch": 0.6047348451110676, "grad_norm": 1.360208914046306, "learning_rate": 7.134547181740998e-06, "loss": 0.4645, "step": 16974 }, { "epoch": 0.6047704722375616, "grad_norm": 1.7768872872787094, "learning_rate": 7.1334416524994175e-06, "loss": 0.5469, "step": 16975 }, { "epoch": 0.6048060993640558, "grad_norm": 1.3910521402255305, "learning_rate": 7.132336161428066e-06, "loss": 0.5367, "step": 16976 }, { "epoch": 0.6048417264905499, "grad_norm": 1.6738282595297522, "learning_rate": 7.131230708541661e-06, "loss": 0.5502, "step": 16977 }, { "epoch": 0.604877353617044, "grad_norm": 1.1164184323579596, "learning_rate": 7.130125293854925e-06, "loss": 0.3815, "step": 16978 }, { "epoch": 0.6049129807435382, "grad_norm": 1.5063536887378848, "learning_rate": 7.12901991738258e-06, "loss": 0.6423, "step": 16979 }, { "epoch": 0.6049486078700322, "grad_norm": 1.1277542278221795, "learning_rate": 7.127914579139338e-06, "loss": 0.4106, "step": 16980 }, { "epoch": 0.6049842349965263, "grad_norm": 1.4057862077335348, "learning_rate": 7.126809279139923e-06, "loss": 0.3977, "step": 16981 }, { "epoch": 0.6050198621230205, "grad_norm": 1.1425967962469037, "learning_rate": 7.125704017399048e-06, "loss": 0.4648, "step": 16982 }, { "epoch": 0.6050554892495146, "grad_norm": 1.13338080181644, "learning_rate": 7.124598793931436e-06, "loss": 0.4314, "step": 16983 }, { "epoch": 0.6050911163760087, "grad_norm": 1.4941539667764718, "learning_rate": 7.123493608751797e-06, "loss": 0.5699, "step": 16984 }, { "epoch": 0.6051267435025028, "grad_norm": 1.0813006857998655, "learning_rate": 7.1223884618748564e-06, "loss": 0.4399, "step": 16985 }, { "epoch": 0.6051623706289969, "grad_norm": 1.0793971509394509, "learning_rate": 7.121283353315321e-06, "loss": 0.4637, "step": 16986 }, { "epoch": 0.605197997755491, "grad_norm": 1.8890880451809828, "learning_rate": 7.12017828308791e-06, "loss": 0.4888, "step": 16987 }, { "epoch": 0.6052336248819852, "grad_norm": 1.1253055612372795, "learning_rate": 7.119073251207338e-06, "loss": 0.4681, "step": 16988 }, { "epoch": 0.6052692520084793, "grad_norm": 1.4995369530369473, "learning_rate": 7.117968257688319e-06, "loss": 0.6024, "step": 16989 }, { "epoch": 0.6053048791349733, "grad_norm": 0.9966204221071037, "learning_rate": 7.116863302545571e-06, "loss": 0.4325, "step": 16990 }, { "epoch": 0.6053405062614675, "grad_norm": 1.2194396664537013, "learning_rate": 7.115758385793799e-06, "loss": 0.4683, "step": 16991 }, { "epoch": 0.6053761333879616, "grad_norm": 1.167472787580116, "learning_rate": 7.114653507447723e-06, "loss": 0.4608, "step": 16992 }, { "epoch": 0.6054117605144557, "grad_norm": 1.4384281008086148, "learning_rate": 7.11354866752205e-06, "loss": 0.5503, "step": 16993 }, { "epoch": 0.6054473876409499, "grad_norm": 1.3434438565313171, "learning_rate": 7.112443866031497e-06, "loss": 0.4171, "step": 16994 }, { "epoch": 0.6054830147674439, "grad_norm": 1.510172048325213, "learning_rate": 7.111339102990768e-06, "loss": 0.5351, "step": 16995 }, { "epoch": 0.605518641893938, "grad_norm": 1.4787399320498682, "learning_rate": 7.110234378414581e-06, "loss": 0.6068, "step": 16996 }, { "epoch": 0.6055542690204322, "grad_norm": 1.377683829500918, "learning_rate": 7.1091296923176416e-06, "loss": 0.4942, "step": 16997 }, { "epoch": 0.6055898961469263, "grad_norm": 1.4276006688889882, "learning_rate": 7.108025044714661e-06, "loss": 0.4481, "step": 16998 }, { "epoch": 0.6056255232734203, "grad_norm": 1.1054453028336613, "learning_rate": 7.106920435620348e-06, "loss": 0.4362, "step": 16999 }, { "epoch": 0.6056611503999145, "grad_norm": 1.3933658269330182, "learning_rate": 7.105815865049414e-06, "loss": 0.5501, "step": 17000 }, { "epoch": 0.6056967775264086, "grad_norm": 1.4032895780270103, "learning_rate": 7.104711333016563e-06, "loss": 0.5082, "step": 17001 }, { "epoch": 0.6057324046529027, "grad_norm": 1.5796384024754682, "learning_rate": 7.103606839536502e-06, "loss": 0.5252, "step": 17002 }, { "epoch": 0.6057680317793969, "grad_norm": 1.1960701836931027, "learning_rate": 7.1025023846239435e-06, "loss": 0.4506, "step": 17003 }, { "epoch": 0.6058036589058909, "grad_norm": 1.1993870649448117, "learning_rate": 7.101397968293588e-06, "loss": 0.451, "step": 17004 }, { "epoch": 0.605839286032385, "grad_norm": 1.5425772982276917, "learning_rate": 7.10029359056015e-06, "loss": 0.4666, "step": 17005 }, { "epoch": 0.6058749131588792, "grad_norm": 1.153453706720648, "learning_rate": 7.099189251438326e-06, "loss": 0.4214, "step": 17006 }, { "epoch": 0.6059105402853733, "grad_norm": 1.496201038530644, "learning_rate": 7.098084950942825e-06, "loss": 0.5342, "step": 17007 }, { "epoch": 0.6059461674118674, "grad_norm": 1.675260048500017, "learning_rate": 7.096980689088349e-06, "loss": 0.5965, "step": 17008 }, { "epoch": 0.6059817945383615, "grad_norm": 1.450690446219234, "learning_rate": 7.09587646588961e-06, "loss": 0.4783, "step": 17009 }, { "epoch": 0.6060174216648556, "grad_norm": 1.0966468147194686, "learning_rate": 7.094772281361298e-06, "loss": 0.432, "step": 17010 }, { "epoch": 0.6060530487913497, "grad_norm": 1.321771401126622, "learning_rate": 7.0936681355181306e-06, "loss": 0.496, "step": 17011 }, { "epoch": 0.6060886759178439, "grad_norm": 1.5528637125521008, "learning_rate": 7.0925640283747986e-06, "loss": 0.602, "step": 17012 }, { "epoch": 0.606124303044338, "grad_norm": 1.2785053688889354, "learning_rate": 7.0914599599460095e-06, "loss": 0.5791, "step": 17013 }, { "epoch": 0.606159930170832, "grad_norm": 1.4985406380096717, "learning_rate": 7.090355930246466e-06, "loss": 0.4968, "step": 17014 }, { "epoch": 0.6061955572973262, "grad_norm": 1.516835522386926, "learning_rate": 7.089251939290865e-06, "loss": 0.4675, "step": 17015 }, { "epoch": 0.6062311844238203, "grad_norm": 1.5039051973364153, "learning_rate": 7.088147987093912e-06, "loss": 0.5359, "step": 17016 }, { "epoch": 0.6062668115503144, "grad_norm": 1.3781142930811119, "learning_rate": 7.087044073670301e-06, "loss": 0.503, "step": 17017 }, { "epoch": 0.6063024386768086, "grad_norm": 1.1907666653752818, "learning_rate": 7.085940199034735e-06, "loss": 0.4446, "step": 17018 }, { "epoch": 0.6063380658033026, "grad_norm": 1.1977350547163363, "learning_rate": 7.084836363201911e-06, "loss": 0.4962, "step": 17019 }, { "epoch": 0.6063736929297967, "grad_norm": 1.4374735680954083, "learning_rate": 7.083732566186533e-06, "loss": 0.4545, "step": 17020 }, { "epoch": 0.6064093200562909, "grad_norm": 1.1702081968099067, "learning_rate": 7.082628808003289e-06, "loss": 0.4625, "step": 17021 }, { "epoch": 0.606444947182785, "grad_norm": 1.0906762294810253, "learning_rate": 7.081525088666884e-06, "loss": 0.4556, "step": 17022 }, { "epoch": 0.606480574309279, "grad_norm": 1.5034401871776095, "learning_rate": 7.080421408192011e-06, "loss": 0.5036, "step": 17023 }, { "epoch": 0.6065162014357732, "grad_norm": 1.1303423176169518, "learning_rate": 7.079317766593366e-06, "loss": 0.4135, "step": 17024 }, { "epoch": 0.6065518285622673, "grad_norm": 1.041095529660283, "learning_rate": 7.078214163885649e-06, "loss": 0.4496, "step": 17025 }, { "epoch": 0.6065874556887614, "grad_norm": 1.1550586159873055, "learning_rate": 7.077110600083552e-06, "loss": 0.4498, "step": 17026 }, { "epoch": 0.6066230828152556, "grad_norm": 1.3396034272322814, "learning_rate": 7.076007075201773e-06, "loss": 0.5175, "step": 17027 }, { "epoch": 0.6066587099417496, "grad_norm": 1.2532460071395979, "learning_rate": 7.074903589255e-06, "loss": 0.537, "step": 17028 }, { "epoch": 0.6066943370682437, "grad_norm": 1.5175694557280364, "learning_rate": 7.0738001422579315e-06, "loss": 0.5338, "step": 17029 }, { "epoch": 0.6067299641947379, "grad_norm": 1.157180740488151, "learning_rate": 7.072696734225258e-06, "loss": 0.4071, "step": 17030 }, { "epoch": 0.606765591321232, "grad_norm": 1.2537508090443785, "learning_rate": 7.071593365171677e-06, "loss": 0.4607, "step": 17031 }, { "epoch": 0.606801218447726, "grad_norm": 1.7530679429654954, "learning_rate": 7.070490035111874e-06, "loss": 0.5865, "step": 17032 }, { "epoch": 0.6068368455742202, "grad_norm": 1.211356853239377, "learning_rate": 7.069386744060546e-06, "loss": 0.4074, "step": 17033 }, { "epoch": 0.6068724727007143, "grad_norm": 1.0426175078736, "learning_rate": 7.068283492032379e-06, "loss": 0.421, "step": 17034 }, { "epoch": 0.6069080998272084, "grad_norm": 1.4532118954527427, "learning_rate": 7.067180279042066e-06, "loss": 0.5692, "step": 17035 }, { "epoch": 0.6069437269537026, "grad_norm": 1.5862560290000762, "learning_rate": 7.0660771051043005e-06, "loss": 0.5209, "step": 17036 }, { "epoch": 0.6069793540801967, "grad_norm": 1.2960089309023612, "learning_rate": 7.064973970233766e-06, "loss": 0.5032, "step": 17037 }, { "epoch": 0.6070149812066907, "grad_norm": 1.1606475899865563, "learning_rate": 7.06387087444516e-06, "loss": 0.44, "step": 17038 }, { "epoch": 0.6070506083331849, "grad_norm": 1.0332737701129902, "learning_rate": 7.062767817753161e-06, "loss": 0.396, "step": 17039 }, { "epoch": 0.607086235459679, "grad_norm": 1.3851743448530163, "learning_rate": 7.061664800172463e-06, "loss": 0.5473, "step": 17040 }, { "epoch": 0.6071218625861731, "grad_norm": 1.5424017883203613, "learning_rate": 7.06056182171775e-06, "loss": 0.5643, "step": 17041 }, { "epoch": 0.6071574897126673, "grad_norm": 1.2252174242709786, "learning_rate": 7.059458882403716e-06, "loss": 0.4126, "step": 17042 }, { "epoch": 0.6071931168391613, "grad_norm": 1.0874136774770757, "learning_rate": 7.058355982245038e-06, "loss": 0.4295, "step": 17043 }, { "epoch": 0.6072287439656554, "grad_norm": 1.1737005208305993, "learning_rate": 7.057253121256407e-06, "loss": 0.5085, "step": 17044 }, { "epoch": 0.6072643710921496, "grad_norm": 1.023378164825798, "learning_rate": 7.056150299452507e-06, "loss": 0.4021, "step": 17045 }, { "epoch": 0.6072999982186437, "grad_norm": 2.1957575804286042, "learning_rate": 7.055047516848026e-06, "loss": 0.604, "step": 17046 }, { "epoch": 0.6073356253451377, "grad_norm": 1.6625805761397474, "learning_rate": 7.053944773457641e-06, "loss": 0.5327, "step": 17047 }, { "epoch": 0.6073712524716319, "grad_norm": 1.170046398897862, "learning_rate": 7.052842069296042e-06, "loss": 0.4084, "step": 17048 }, { "epoch": 0.607406879598126, "grad_norm": 1.519564293724531, "learning_rate": 7.051739404377914e-06, "loss": 0.533, "step": 17049 }, { "epoch": 0.6074425067246201, "grad_norm": 1.6079998402390885, "learning_rate": 7.050636778717933e-06, "loss": 0.5002, "step": 17050 }, { "epoch": 0.6074781338511143, "grad_norm": 1.718057449514891, "learning_rate": 7.049534192330787e-06, "loss": 0.5166, "step": 17051 }, { "epoch": 0.6075137609776083, "grad_norm": 1.104430140973258, "learning_rate": 7.048431645231153e-06, "loss": 0.4215, "step": 17052 }, { "epoch": 0.6075493881041024, "grad_norm": 1.7265873671674126, "learning_rate": 7.04732913743372e-06, "loss": 0.5622, "step": 17053 }, { "epoch": 0.6075850152305966, "grad_norm": 1.7239303420875287, "learning_rate": 7.046226668953158e-06, "loss": 0.5086, "step": 17054 }, { "epoch": 0.6076206423570907, "grad_norm": 1.1265460302033203, "learning_rate": 7.045124239804155e-06, "loss": 0.4557, "step": 17055 }, { "epoch": 0.6076562694835848, "grad_norm": 1.5734397337176942, "learning_rate": 7.044021850001386e-06, "loss": 0.5354, "step": 17056 }, { "epoch": 0.6076918966100789, "grad_norm": 1.3351018931972136, "learning_rate": 7.042919499559538e-06, "loss": 0.4861, "step": 17057 }, { "epoch": 0.607727523736573, "grad_norm": 1.8229989975696201, "learning_rate": 7.0418171884932766e-06, "loss": 0.5231, "step": 17058 }, { "epoch": 0.6077631508630671, "grad_norm": 1.2712371110574054, "learning_rate": 7.0407149168172865e-06, "loss": 0.4867, "step": 17059 }, { "epoch": 0.6077987779895613, "grad_norm": 1.142479412140074, "learning_rate": 7.039612684546251e-06, "loss": 0.4418, "step": 17060 }, { "epoch": 0.6078344051160554, "grad_norm": 1.5638186322566774, "learning_rate": 7.0385104916948386e-06, "loss": 0.5974, "step": 17061 }, { "epoch": 0.6078700322425494, "grad_norm": 1.258699085878732, "learning_rate": 7.03740833827773e-06, "loss": 0.4534, "step": 17062 }, { "epoch": 0.6079056593690436, "grad_norm": 1.4540970147808723, "learning_rate": 7.0363062243095994e-06, "loss": 0.4843, "step": 17063 }, { "epoch": 0.6079412864955377, "grad_norm": 1.2082410269654345, "learning_rate": 7.035204149805126e-06, "loss": 0.4971, "step": 17064 }, { "epoch": 0.6079769136220318, "grad_norm": 1.321283343708408, "learning_rate": 7.034102114778978e-06, "loss": 0.4917, "step": 17065 }, { "epoch": 0.608012540748526, "grad_norm": 1.0969504647627641, "learning_rate": 7.033000119245836e-06, "loss": 0.3863, "step": 17066 }, { "epoch": 0.60804816787502, "grad_norm": 2.2220913609075037, "learning_rate": 7.031898163220367e-06, "loss": 0.5342, "step": 17067 }, { "epoch": 0.6080837950015141, "grad_norm": 1.6177140483061743, "learning_rate": 7.0307962467172555e-06, "loss": 0.5775, "step": 17068 }, { "epoch": 0.6081194221280083, "grad_norm": 1.133002042926261, "learning_rate": 7.029694369751161e-06, "loss": 0.4473, "step": 17069 }, { "epoch": 0.6081550492545024, "grad_norm": 1.1509047019000158, "learning_rate": 7.028592532336762e-06, "loss": 0.4274, "step": 17070 }, { "epoch": 0.6081906763809964, "grad_norm": 1.164153962137273, "learning_rate": 7.0274907344887335e-06, "loss": 0.4381, "step": 17071 }, { "epoch": 0.6082263035074906, "grad_norm": 1.1976967753615309, "learning_rate": 7.026388976221742e-06, "loss": 0.4485, "step": 17072 }, { "epoch": 0.6082619306339847, "grad_norm": 1.6830731553564553, "learning_rate": 7.0252872575504614e-06, "loss": 0.5923, "step": 17073 }, { "epoch": 0.6082975577604788, "grad_norm": 1.3797583001240987, "learning_rate": 7.0241855784895575e-06, "loss": 0.5735, "step": 17074 }, { "epoch": 0.608333184886973, "grad_norm": 1.2113152411004646, "learning_rate": 7.023083939053706e-06, "loss": 0.4559, "step": 17075 }, { "epoch": 0.608368812013467, "grad_norm": 1.4740462841703699, "learning_rate": 7.0219823392575694e-06, "loss": 0.6045, "step": 17076 }, { "epoch": 0.6084044391399611, "grad_norm": 1.587760892211599, "learning_rate": 7.020880779115821e-06, "loss": 0.5786, "step": 17077 }, { "epoch": 0.6084400662664553, "grad_norm": 1.2259283932233767, "learning_rate": 7.019779258643127e-06, "loss": 0.4413, "step": 17078 }, { "epoch": 0.6084756933929494, "grad_norm": 1.6432125343234554, "learning_rate": 7.018677777854158e-06, "loss": 0.5625, "step": 17079 }, { "epoch": 0.6085113205194435, "grad_norm": 3.55073148078713, "learning_rate": 7.017576336763574e-06, "loss": 0.5221, "step": 17080 }, { "epoch": 0.6085469476459376, "grad_norm": 1.2747129517777882, "learning_rate": 7.016474935386049e-06, "loss": 0.4372, "step": 17081 }, { "epoch": 0.6085825747724317, "grad_norm": 1.4195118407264684, "learning_rate": 7.015373573736241e-06, "loss": 0.5402, "step": 17082 }, { "epoch": 0.6086182018989258, "grad_norm": 1.727627617787551, "learning_rate": 7.014272251828822e-06, "loss": 0.6071, "step": 17083 }, { "epoch": 0.60865382902542, "grad_norm": 1.4307073978259597, "learning_rate": 7.013170969678457e-06, "loss": 0.4722, "step": 17084 }, { "epoch": 0.608689456151914, "grad_norm": 1.0893821488001045, "learning_rate": 7.012069727299805e-06, "loss": 0.4593, "step": 17085 }, { "epoch": 0.6087250832784081, "grad_norm": 1.1098881479637692, "learning_rate": 7.010968524707538e-06, "loss": 0.4401, "step": 17086 }, { "epoch": 0.6087607104049023, "grad_norm": 1.0934208067188127, "learning_rate": 7.009867361916309e-06, "loss": 0.4645, "step": 17087 }, { "epoch": 0.6087963375313964, "grad_norm": 1.3095989709551756, "learning_rate": 7.00876623894079e-06, "loss": 0.4489, "step": 17088 }, { "epoch": 0.6088319646578905, "grad_norm": 1.0290214885813795, "learning_rate": 7.007665155795636e-06, "loss": 0.399, "step": 17089 }, { "epoch": 0.6088675917843847, "grad_norm": 1.205607281740253, "learning_rate": 7.006564112495515e-06, "loss": 0.4473, "step": 17090 }, { "epoch": 0.6089032189108787, "grad_norm": 1.5587629763665234, "learning_rate": 7.005463109055084e-06, "loss": 0.6152, "step": 17091 }, { "epoch": 0.6089388460373728, "grad_norm": 1.120249024023, "learning_rate": 7.004362145489004e-06, "loss": 0.4365, "step": 17092 }, { "epoch": 0.608974473163867, "grad_norm": 1.1319376255985882, "learning_rate": 7.003261221811935e-06, "loss": 0.4222, "step": 17093 }, { "epoch": 0.6090101002903611, "grad_norm": 1.2796015503533478, "learning_rate": 7.002160338038537e-06, "loss": 0.5091, "step": 17094 }, { "epoch": 0.6090457274168551, "grad_norm": 1.123165033923153, "learning_rate": 7.0010594941834745e-06, "loss": 0.3945, "step": 17095 }, { "epoch": 0.6090813545433493, "grad_norm": 1.0989471026632862, "learning_rate": 6.999958690261396e-06, "loss": 0.4028, "step": 17096 }, { "epoch": 0.6091169816698434, "grad_norm": 1.4427996960061278, "learning_rate": 6.998857926286966e-06, "loss": 0.5349, "step": 17097 }, { "epoch": 0.6091526087963375, "grad_norm": 1.200842404118271, "learning_rate": 6.99775720227484e-06, "loss": 0.4349, "step": 17098 }, { "epoch": 0.6091882359228317, "grad_norm": 1.3761020181993526, "learning_rate": 6.996656518239676e-06, "loss": 0.5576, "step": 17099 }, { "epoch": 0.6092238630493257, "grad_norm": 1.2524214931696303, "learning_rate": 6.995555874196128e-06, "loss": 0.512, "step": 17100 }, { "epoch": 0.6092594901758198, "grad_norm": 1.2047593741747764, "learning_rate": 6.994455270158858e-06, "loss": 0.4654, "step": 17101 }, { "epoch": 0.609295117302314, "grad_norm": 1.2314043034540316, "learning_rate": 6.993354706142512e-06, "loss": 0.4798, "step": 17102 }, { "epoch": 0.6093307444288081, "grad_norm": 1.0840668315831203, "learning_rate": 6.992254182161753e-06, "loss": 0.4157, "step": 17103 }, { "epoch": 0.6093663715553022, "grad_norm": 1.1774453145987127, "learning_rate": 6.991153698231228e-06, "loss": 0.4628, "step": 17104 }, { "epoch": 0.6094019986817963, "grad_norm": 1.6173078472294289, "learning_rate": 6.990053254365597e-06, "loss": 0.6123, "step": 17105 }, { "epoch": 0.6094376258082904, "grad_norm": 1.0821628372643322, "learning_rate": 6.988952850579512e-06, "loss": 0.3982, "step": 17106 }, { "epoch": 0.6094732529347845, "grad_norm": 1.1719764528523327, "learning_rate": 6.987852486887622e-06, "loss": 0.4753, "step": 17107 }, { "epoch": 0.6095088800612787, "grad_norm": 1.4009608937748277, "learning_rate": 6.986752163304582e-06, "loss": 0.5407, "step": 17108 }, { "epoch": 0.6095445071877728, "grad_norm": 1.0989364971022288, "learning_rate": 6.985651879845044e-06, "loss": 0.4364, "step": 17109 }, { "epoch": 0.6095801343142668, "grad_norm": 1.5651511158922644, "learning_rate": 6.98455163652366e-06, "loss": 0.5482, "step": 17110 }, { "epoch": 0.609615761440761, "grad_norm": 1.238775223661362, "learning_rate": 6.983451433355072e-06, "loss": 0.4598, "step": 17111 }, { "epoch": 0.6096513885672551, "grad_norm": 1.1556167653551246, "learning_rate": 6.9823512703539445e-06, "loss": 0.4207, "step": 17112 }, { "epoch": 0.6096870156937492, "grad_norm": 3.0506075730212334, "learning_rate": 6.9812511475349135e-06, "loss": 0.5752, "step": 17113 }, { "epoch": 0.6097226428202434, "grad_norm": 1.2005075523356548, "learning_rate": 6.980151064912636e-06, "loss": 0.4108, "step": 17114 }, { "epoch": 0.6097582699467374, "grad_norm": 1.7775278009642816, "learning_rate": 6.979051022501755e-06, "loss": 0.3507, "step": 17115 }, { "epoch": 0.6097938970732315, "grad_norm": 1.3990272281403318, "learning_rate": 6.977951020316922e-06, "loss": 0.4401, "step": 17116 }, { "epoch": 0.6098295241997257, "grad_norm": 1.117817071644375, "learning_rate": 6.976851058372789e-06, "loss": 0.4444, "step": 17117 }, { "epoch": 0.6098651513262198, "grad_norm": 1.090535845737201, "learning_rate": 6.97575113668399e-06, "loss": 0.3899, "step": 17118 }, { "epoch": 0.6099007784527138, "grad_norm": 1.4360926361714854, "learning_rate": 6.974651255265183e-06, "loss": 0.503, "step": 17119 }, { "epoch": 0.609936405579208, "grad_norm": 1.105755394357395, "learning_rate": 6.973551414131005e-06, "loss": 0.4914, "step": 17120 }, { "epoch": 0.6099720327057021, "grad_norm": 1.1182160816630158, "learning_rate": 6.972451613296112e-06, "loss": 0.4404, "step": 17121 }, { "epoch": 0.6100076598321962, "grad_norm": 1.3297293136266428, "learning_rate": 6.971351852775137e-06, "loss": 0.5364, "step": 17122 }, { "epoch": 0.6100432869586904, "grad_norm": 1.3730441597613854, "learning_rate": 6.970252132582729e-06, "loss": 0.5127, "step": 17123 }, { "epoch": 0.6100789140851844, "grad_norm": 1.0474605421399013, "learning_rate": 6.969152452733532e-06, "loss": 0.4146, "step": 17124 }, { "epoch": 0.6101145412116785, "grad_norm": 1.0873368802996446, "learning_rate": 6.968052813242189e-06, "loss": 0.4168, "step": 17125 }, { "epoch": 0.6101501683381727, "grad_norm": 1.531145327200498, "learning_rate": 6.96695321412334e-06, "loss": 0.5134, "step": 17126 }, { "epoch": 0.6101857954646668, "grad_norm": 1.1499130582942647, "learning_rate": 6.965853655391633e-06, "loss": 0.43, "step": 17127 }, { "epoch": 0.6102214225911609, "grad_norm": 1.097941032563542, "learning_rate": 6.964754137061702e-06, "loss": 0.4239, "step": 17128 }, { "epoch": 0.610257049717655, "grad_norm": 1.5276107363755809, "learning_rate": 6.963654659148192e-06, "loss": 0.5169, "step": 17129 }, { "epoch": 0.6102926768441491, "grad_norm": 1.0515800057813827, "learning_rate": 6.962555221665742e-06, "loss": 0.4246, "step": 17130 }, { "epoch": 0.6103283039706432, "grad_norm": 1.066932937334619, "learning_rate": 6.961455824628992e-06, "loss": 0.4068, "step": 17131 }, { "epoch": 0.6103639310971374, "grad_norm": 1.7006850501462238, "learning_rate": 6.960356468052585e-06, "loss": 0.555, "step": 17132 }, { "epoch": 0.6103995582236315, "grad_norm": 1.2663385588463465, "learning_rate": 6.959257151951153e-06, "loss": 0.5918, "step": 17133 }, { "epoch": 0.6104351853501255, "grad_norm": 1.2944897745996162, "learning_rate": 6.958157876339337e-06, "loss": 0.5894, "step": 17134 }, { "epoch": 0.6104708124766197, "grad_norm": 1.3361684689789046, "learning_rate": 6.957058641231775e-06, "loss": 0.4592, "step": 17135 }, { "epoch": 0.6105064396031138, "grad_norm": 1.6555359598354555, "learning_rate": 6.955959446643107e-06, "loss": 0.4994, "step": 17136 }, { "epoch": 0.6105420667296079, "grad_norm": 1.143502559739481, "learning_rate": 6.954860292587964e-06, "loss": 0.4673, "step": 17137 }, { "epoch": 0.610577693856102, "grad_norm": 1.4105308262617016, "learning_rate": 6.953761179080987e-06, "loss": 0.5691, "step": 17138 }, { "epoch": 0.6106133209825961, "grad_norm": 1.5119923387709424, "learning_rate": 6.952662106136806e-06, "loss": 0.4176, "step": 17139 }, { "epoch": 0.6106489481090902, "grad_norm": 1.2170386554293666, "learning_rate": 6.95156307377006e-06, "loss": 0.4198, "step": 17140 }, { "epoch": 0.6106845752355844, "grad_norm": 1.124967389289161, "learning_rate": 6.9504640819953825e-06, "loss": 0.3894, "step": 17141 }, { "epoch": 0.6107202023620785, "grad_norm": 1.2121192041993365, "learning_rate": 6.949365130827407e-06, "loss": 0.3992, "step": 17142 }, { "epoch": 0.6107558294885727, "grad_norm": 2.183925621949129, "learning_rate": 6.94826622028077e-06, "loss": 0.5261, "step": 17143 }, { "epoch": 0.6107914566150667, "grad_norm": 1.3761824601365809, "learning_rate": 6.947167350370098e-06, "loss": 0.584, "step": 17144 }, { "epoch": 0.6108270837415608, "grad_norm": 1.2486816326219805, "learning_rate": 6.946068521110028e-06, "loss": 0.4179, "step": 17145 }, { "epoch": 0.610862710868055, "grad_norm": 1.1516276638259475, "learning_rate": 6.944969732515188e-06, "loss": 0.4248, "step": 17146 }, { "epoch": 0.6108983379945491, "grad_norm": 1.5618761444762987, "learning_rate": 6.943870984600216e-06, "loss": 0.5306, "step": 17147 }, { "epoch": 0.6109339651210431, "grad_norm": 1.7120993067688246, "learning_rate": 6.9427722773797325e-06, "loss": 0.5472, "step": 17148 }, { "epoch": 0.6109695922475373, "grad_norm": 1.4409672737488262, "learning_rate": 6.941673610868378e-06, "loss": 0.5378, "step": 17149 }, { "epoch": 0.6110052193740314, "grad_norm": 1.7760964201455904, "learning_rate": 6.9405749850807725e-06, "loss": 0.612, "step": 17150 }, { "epoch": 0.6110408465005255, "grad_norm": 1.3233199069793835, "learning_rate": 6.9394764000315504e-06, "loss": 0.5056, "step": 17151 }, { "epoch": 0.6110764736270197, "grad_norm": 1.7861821307090187, "learning_rate": 6.938377855735341e-06, "loss": 0.5004, "step": 17152 }, { "epoch": 0.6111121007535137, "grad_norm": 1.1263691749579428, "learning_rate": 6.9372793522067685e-06, "loss": 0.4121, "step": 17153 }, { "epoch": 0.6111477278800078, "grad_norm": 1.202055583425508, "learning_rate": 6.936180889460467e-06, "loss": 0.4651, "step": 17154 }, { "epoch": 0.611183355006502, "grad_norm": 1.1987237089371119, "learning_rate": 6.935082467511053e-06, "loss": 0.4373, "step": 17155 }, { "epoch": 0.6112189821329961, "grad_norm": 1.1223229681838816, "learning_rate": 6.93398408637316e-06, "loss": 0.4652, "step": 17156 }, { "epoch": 0.6112546092594902, "grad_norm": 1.094260612905748, "learning_rate": 6.932885746061411e-06, "loss": 0.4296, "step": 17157 }, { "epoch": 0.6112902363859843, "grad_norm": 1.4947645707969932, "learning_rate": 6.931787446590435e-06, "loss": 0.5786, "step": 17158 }, { "epoch": 0.6113258635124784, "grad_norm": 1.4485848741594822, "learning_rate": 6.9306891879748505e-06, "loss": 0.5563, "step": 17159 }, { "epoch": 0.6113614906389725, "grad_norm": 1.3472795225265886, "learning_rate": 6.929590970229287e-06, "loss": 0.5212, "step": 17160 }, { "epoch": 0.6113971177654667, "grad_norm": 1.4611518995639161, "learning_rate": 6.928492793368365e-06, "loss": 0.5386, "step": 17161 }, { "epoch": 0.6114327448919608, "grad_norm": 1.246110229198788, "learning_rate": 6.927394657406705e-06, "loss": 0.4302, "step": 17162 }, { "epoch": 0.6114683720184548, "grad_norm": 1.400513599895791, "learning_rate": 6.926296562358936e-06, "loss": 0.5867, "step": 17163 }, { "epoch": 0.611503999144949, "grad_norm": 1.6068362215534875, "learning_rate": 6.925198508239677e-06, "loss": 0.5441, "step": 17164 }, { "epoch": 0.6115396262714431, "grad_norm": 1.0634445434395203, "learning_rate": 6.9241004950635505e-06, "loss": 0.3912, "step": 17165 }, { "epoch": 0.6115752533979372, "grad_norm": 1.490258075504414, "learning_rate": 6.923002522845172e-06, "loss": 0.5266, "step": 17166 }, { "epoch": 0.6116108805244314, "grad_norm": 1.917211663972591, "learning_rate": 6.921904591599169e-06, "loss": 0.5439, "step": 17167 }, { "epoch": 0.6116465076509254, "grad_norm": 17.80925927709177, "learning_rate": 6.920806701340154e-06, "loss": 0.5869, "step": 17168 }, { "epoch": 0.6116821347774195, "grad_norm": 1.1362223287154338, "learning_rate": 6.919708852082756e-06, "loss": 0.3896, "step": 17169 }, { "epoch": 0.6117177619039137, "grad_norm": 1.1593982780966072, "learning_rate": 6.918611043841582e-06, "loss": 0.4843, "step": 17170 }, { "epoch": 0.6117533890304078, "grad_norm": 1.5290175199244973, "learning_rate": 6.917513276631258e-06, "loss": 0.5777, "step": 17171 }, { "epoch": 0.6117890161569018, "grad_norm": 1.1609428657696896, "learning_rate": 6.916415550466398e-06, "loss": 0.4578, "step": 17172 }, { "epoch": 0.611824643283396, "grad_norm": 1.261793008597713, "learning_rate": 6.915317865361623e-06, "loss": 0.5106, "step": 17173 }, { "epoch": 0.6118602704098901, "grad_norm": 1.1572592368638188, "learning_rate": 6.9142202213315425e-06, "loss": 0.4572, "step": 17174 }, { "epoch": 0.6118958975363842, "grad_norm": 1.2086177123227178, "learning_rate": 6.913122618390777e-06, "loss": 0.4554, "step": 17175 }, { "epoch": 0.6119315246628784, "grad_norm": 1.376684450980859, "learning_rate": 6.912025056553946e-06, "loss": 0.5579, "step": 17176 }, { "epoch": 0.6119671517893724, "grad_norm": 1.837114464171939, "learning_rate": 6.910927535835656e-06, "loss": 0.5074, "step": 17177 }, { "epoch": 0.6120027789158665, "grad_norm": 1.5157102248226686, "learning_rate": 6.909830056250527e-06, "loss": 0.5735, "step": 17178 }, { "epoch": 0.6120384060423607, "grad_norm": 1.4126122774792, "learning_rate": 6.908732617813168e-06, "loss": 0.4476, "step": 17179 }, { "epoch": 0.6120740331688548, "grad_norm": 1.3698778514168528, "learning_rate": 6.9076352205382005e-06, "loss": 0.587, "step": 17180 }, { "epoch": 0.6121096602953489, "grad_norm": 1.4576667882599885, "learning_rate": 6.906537864440226e-06, "loss": 0.5604, "step": 17181 }, { "epoch": 0.612145287421843, "grad_norm": 1.5900994090095852, "learning_rate": 6.905440549533864e-06, "loss": 0.5397, "step": 17182 }, { "epoch": 0.6121809145483371, "grad_norm": 1.164545133169882, "learning_rate": 6.904343275833722e-06, "loss": 0.4082, "step": 17183 }, { "epoch": 0.6122165416748312, "grad_norm": 1.2662870414377638, "learning_rate": 6.903246043354418e-06, "loss": 0.5096, "step": 17184 }, { "epoch": 0.6122521688013254, "grad_norm": 1.177969908006942, "learning_rate": 6.902148852110553e-06, "loss": 0.4156, "step": 17185 }, { "epoch": 0.6122877959278195, "grad_norm": 1.8104475202263668, "learning_rate": 6.901051702116739e-06, "loss": 0.5163, "step": 17186 }, { "epoch": 0.6123234230543135, "grad_norm": 2.0330875772181782, "learning_rate": 6.899954593387594e-06, "loss": 0.5897, "step": 17187 }, { "epoch": 0.6123590501808077, "grad_norm": 1.0825622731759972, "learning_rate": 6.8988575259377156e-06, "loss": 0.4815, "step": 17188 }, { "epoch": 0.6123946773073018, "grad_norm": 2.4388393044809, "learning_rate": 6.897760499781719e-06, "loss": 0.5362, "step": 17189 }, { "epoch": 0.6124303044337959, "grad_norm": 1.2796838893819422, "learning_rate": 6.896663514934208e-06, "loss": 0.4272, "step": 17190 }, { "epoch": 0.6124659315602901, "grad_norm": 1.1825891099802057, "learning_rate": 6.895566571409795e-06, "loss": 0.4353, "step": 17191 }, { "epoch": 0.6125015586867841, "grad_norm": 1.087065410222922, "learning_rate": 6.89446966922308e-06, "loss": 0.4334, "step": 17192 }, { "epoch": 0.6125371858132782, "grad_norm": 1.068861185448374, "learning_rate": 6.893372808388674e-06, "loss": 0.4456, "step": 17193 }, { "epoch": 0.6125728129397724, "grad_norm": 1.0864827658946474, "learning_rate": 6.892275988921178e-06, "loss": 0.4412, "step": 17194 }, { "epoch": 0.6126084400662665, "grad_norm": 1.5533892260662856, "learning_rate": 6.891179210835202e-06, "loss": 0.4769, "step": 17195 }, { "epoch": 0.6126440671927605, "grad_norm": 1.100406188365024, "learning_rate": 6.890082474145345e-06, "loss": 0.4283, "step": 17196 }, { "epoch": 0.6126796943192547, "grad_norm": 1.4845607379800303, "learning_rate": 6.888985778866214e-06, "loss": 0.5248, "step": 17197 }, { "epoch": 0.6127153214457488, "grad_norm": 1.1892224078645344, "learning_rate": 6.887889125012413e-06, "loss": 0.4732, "step": 17198 }, { "epoch": 0.6127509485722429, "grad_norm": 1.141420784986733, "learning_rate": 6.886792512598542e-06, "loss": 0.4592, "step": 17199 }, { "epoch": 0.6127865756987371, "grad_norm": 1.2099148665888435, "learning_rate": 6.885695941639208e-06, "loss": 0.4573, "step": 17200 }, { "epoch": 0.6128222028252311, "grad_norm": 1.227377287974881, "learning_rate": 6.884599412149004e-06, "loss": 0.4541, "step": 17201 }, { "epoch": 0.6128578299517252, "grad_norm": 2.0028190701025266, "learning_rate": 6.883502924142543e-06, "loss": 0.4761, "step": 17202 }, { "epoch": 0.6128934570782194, "grad_norm": 1.6982592843664095, "learning_rate": 6.882406477634413e-06, "loss": 0.5255, "step": 17203 }, { "epoch": 0.6129290842047135, "grad_norm": 1.229097632003929, "learning_rate": 6.881310072639224e-06, "loss": 0.4716, "step": 17204 }, { "epoch": 0.6129647113312076, "grad_norm": 1.6915267148176578, "learning_rate": 6.8802137091715685e-06, "loss": 0.5587, "step": 17205 }, { "epoch": 0.6130003384577017, "grad_norm": 1.111982797142257, "learning_rate": 6.879117387246052e-06, "loss": 0.4483, "step": 17206 }, { "epoch": 0.6130359655841958, "grad_norm": 1.8451692436752287, "learning_rate": 6.878021106877264e-06, "loss": 0.507, "step": 17207 }, { "epoch": 0.6130715927106899, "grad_norm": 1.6296452357057951, "learning_rate": 6.876924868079811e-06, "loss": 0.5177, "step": 17208 }, { "epoch": 0.6131072198371841, "grad_norm": 1.4845242159419823, "learning_rate": 6.875828670868282e-06, "loss": 0.5644, "step": 17209 }, { "epoch": 0.6131428469636782, "grad_norm": 1.3950828685738508, "learning_rate": 6.8747325152572785e-06, "loss": 0.4214, "step": 17210 }, { "epoch": 0.6131784740901722, "grad_norm": 1.1769298290054722, "learning_rate": 6.873636401261401e-06, "loss": 0.4667, "step": 17211 }, { "epoch": 0.6132141012166664, "grad_norm": 1.1441543855973664, "learning_rate": 6.872540328895233e-06, "loss": 0.4831, "step": 17212 }, { "epoch": 0.6132497283431605, "grad_norm": 1.5537903073055974, "learning_rate": 6.871444298173382e-06, "loss": 0.5157, "step": 17213 }, { "epoch": 0.6132853554696546, "grad_norm": 1.3765347719469228, "learning_rate": 6.870348309110435e-06, "loss": 0.5353, "step": 17214 }, { "epoch": 0.6133209825961488, "grad_norm": 1.339060115043183, "learning_rate": 6.869252361720988e-06, "loss": 0.4318, "step": 17215 }, { "epoch": 0.6133566097226428, "grad_norm": 0.9847716907995356, "learning_rate": 6.868156456019633e-06, "loss": 0.4005, "step": 17216 }, { "epoch": 0.6133922368491369, "grad_norm": 1.0979939031899546, "learning_rate": 6.867060592020967e-06, "loss": 0.4814, "step": 17217 }, { "epoch": 0.6134278639756311, "grad_norm": 1.4031069361276471, "learning_rate": 6.8659647697395745e-06, "loss": 0.4636, "step": 17218 }, { "epoch": 0.6134634911021252, "grad_norm": 1.3248816540429675, "learning_rate": 6.864868989190056e-06, "loss": 0.5314, "step": 17219 }, { "epoch": 0.6134991182286192, "grad_norm": 1.2970381410446017, "learning_rate": 6.863773250386994e-06, "loss": 0.4329, "step": 17220 }, { "epoch": 0.6135347453551134, "grad_norm": 1.1338003105329169, "learning_rate": 6.862677553344984e-06, "loss": 0.4246, "step": 17221 }, { "epoch": 0.6135703724816075, "grad_norm": 1.4086781358249456, "learning_rate": 6.86158189807862e-06, "loss": 0.5211, "step": 17222 }, { "epoch": 0.6136059996081016, "grad_norm": 1.312697760500205, "learning_rate": 6.860486284602479e-06, "loss": 0.4381, "step": 17223 }, { "epoch": 0.6136416267345958, "grad_norm": 1.5487393725687666, "learning_rate": 6.859390712931165e-06, "loss": 0.5131, "step": 17224 }, { "epoch": 0.6136772538610898, "grad_norm": 1.2369813698751897, "learning_rate": 6.858295183079254e-06, "loss": 0.5537, "step": 17225 }, { "epoch": 0.6137128809875839, "grad_norm": 1.12608650665934, "learning_rate": 6.857199695061341e-06, "loss": 0.416, "step": 17226 }, { "epoch": 0.6137485081140781, "grad_norm": 1.9478522871827373, "learning_rate": 6.856104248892008e-06, "loss": 0.5265, "step": 17227 }, { "epoch": 0.6137841352405722, "grad_norm": 1.2189309915197661, "learning_rate": 6.855008844585849e-06, "loss": 0.4326, "step": 17228 }, { "epoch": 0.6138197623670663, "grad_norm": 1.2166981309972247, "learning_rate": 6.8539134821574415e-06, "loss": 0.4194, "step": 17229 }, { "epoch": 0.6138553894935604, "grad_norm": 1.165034906944892, "learning_rate": 6.852818161621377e-06, "loss": 0.4667, "step": 17230 }, { "epoch": 0.6138910166200545, "grad_norm": 1.6223718934530746, "learning_rate": 6.851722882992237e-06, "loss": 0.5306, "step": 17231 }, { "epoch": 0.6139266437465486, "grad_norm": 1.1397404972871623, "learning_rate": 6.850627646284606e-06, "loss": 0.4126, "step": 17232 }, { "epoch": 0.6139622708730428, "grad_norm": 1.750315154230828, "learning_rate": 6.8495324515130744e-06, "loss": 0.5344, "step": 17233 }, { "epoch": 0.6139978979995369, "grad_norm": 1.4893444271141025, "learning_rate": 6.848437298692216e-06, "loss": 0.6171, "step": 17234 }, { "epoch": 0.6140335251260309, "grad_norm": 1.2388287879007691, "learning_rate": 6.847342187836619e-06, "loss": 0.4461, "step": 17235 }, { "epoch": 0.6140691522525251, "grad_norm": 1.280299735914224, "learning_rate": 6.846247118960863e-06, "loss": 0.5853, "step": 17236 }, { "epoch": 0.6141047793790192, "grad_norm": 1.6517977618202506, "learning_rate": 6.8451520920795326e-06, "loss": 0.5502, "step": 17237 }, { "epoch": 0.6141404065055133, "grad_norm": 1.2216581666587873, "learning_rate": 6.844057107207205e-06, "loss": 0.4234, "step": 17238 }, { "epoch": 0.6141760336320075, "grad_norm": 1.1092910187997402, "learning_rate": 6.842962164358467e-06, "loss": 0.3829, "step": 17239 }, { "epoch": 0.6142116607585015, "grad_norm": 1.4215185875833722, "learning_rate": 6.84186726354789e-06, "loss": 0.4492, "step": 17240 }, { "epoch": 0.6142472878849956, "grad_norm": 1.4802319883490094, "learning_rate": 6.8407724047900616e-06, "loss": 0.5089, "step": 17241 }, { "epoch": 0.6142829150114898, "grad_norm": 1.1574783483999207, "learning_rate": 6.839677588099554e-06, "loss": 0.418, "step": 17242 }, { "epoch": 0.6143185421379839, "grad_norm": 1.5316611279824386, "learning_rate": 6.838582813490947e-06, "loss": 0.4692, "step": 17243 }, { "epoch": 0.614354169264478, "grad_norm": 1.1414814978744359, "learning_rate": 6.837488080978825e-06, "loss": 0.4326, "step": 17244 }, { "epoch": 0.6143897963909721, "grad_norm": 1.1850173353266555, "learning_rate": 6.836393390577755e-06, "loss": 0.4774, "step": 17245 }, { "epoch": 0.6144254235174662, "grad_norm": 1.043641681340983, "learning_rate": 6.83529874230232e-06, "loss": 0.4495, "step": 17246 }, { "epoch": 0.6144610506439603, "grad_norm": 1.1649784824877942, "learning_rate": 6.834204136167093e-06, "loss": 0.4274, "step": 17247 }, { "epoch": 0.6144966777704545, "grad_norm": 1.716392807012266, "learning_rate": 6.833109572186653e-06, "loss": 0.53, "step": 17248 }, { "epoch": 0.6145323048969485, "grad_norm": 1.439780968962979, "learning_rate": 6.832015050375569e-06, "loss": 0.4959, "step": 17249 }, { "epoch": 0.6145679320234426, "grad_norm": 1.8282543270978984, "learning_rate": 6.830920570748423e-06, "loss": 0.5429, "step": 17250 }, { "epoch": 0.6146035591499368, "grad_norm": 1.144094424571825, "learning_rate": 6.829826133319781e-06, "loss": 0.4807, "step": 17251 }, { "epoch": 0.6146391862764309, "grad_norm": 1.4853063302576577, "learning_rate": 6.8287317381042216e-06, "loss": 0.6097, "step": 17252 }, { "epoch": 0.614674813402925, "grad_norm": 1.1498388333758314, "learning_rate": 6.827637385116315e-06, "loss": 0.4582, "step": 17253 }, { "epoch": 0.6147104405294191, "grad_norm": 1.8045446217008239, "learning_rate": 6.826543074370637e-06, "loss": 0.5914, "step": 17254 }, { "epoch": 0.6147460676559132, "grad_norm": 1.3351539156414385, "learning_rate": 6.825448805881753e-06, "loss": 0.5852, "step": 17255 }, { "epoch": 0.6147816947824073, "grad_norm": 1.3765097377612896, "learning_rate": 6.824354579664236e-06, "loss": 0.5278, "step": 17256 }, { "epoch": 0.6148173219089015, "grad_norm": 1.1089231834077786, "learning_rate": 6.823260395732661e-06, "loss": 0.4377, "step": 17257 }, { "epoch": 0.6148529490353956, "grad_norm": 1.5628453505472997, "learning_rate": 6.82216625410159e-06, "loss": 0.549, "step": 17258 }, { "epoch": 0.6148885761618896, "grad_norm": 1.9953072717221387, "learning_rate": 6.821072154785602e-06, "loss": 0.5175, "step": 17259 }, { "epoch": 0.6149242032883838, "grad_norm": 1.2513425438661585, "learning_rate": 6.819978097799256e-06, "loss": 0.4531, "step": 17260 }, { "epoch": 0.6149598304148779, "grad_norm": 1.5469285235706398, "learning_rate": 6.818884083157125e-06, "loss": 0.5415, "step": 17261 }, { "epoch": 0.614995457541372, "grad_norm": 1.2767984978099325, "learning_rate": 6.817790110873774e-06, "loss": 0.4262, "step": 17262 }, { "epoch": 0.6150310846678662, "grad_norm": 1.4795731869153217, "learning_rate": 6.816696180963776e-06, "loss": 0.4761, "step": 17263 }, { "epoch": 0.6150667117943602, "grad_norm": 1.2456962112609156, "learning_rate": 6.81560229344169e-06, "loss": 0.4803, "step": 17264 }, { "epoch": 0.6151023389208543, "grad_norm": 1.4748538864173568, "learning_rate": 6.81450844832209e-06, "loss": 0.5367, "step": 17265 }, { "epoch": 0.6151379660473485, "grad_norm": 1.1985250835805799, "learning_rate": 6.813414645619531e-06, "loss": 0.4297, "step": 17266 }, { "epoch": 0.6151735931738426, "grad_norm": 1.447245215935316, "learning_rate": 6.812320885348585e-06, "loss": 0.5383, "step": 17267 }, { "epoch": 0.6152092203003366, "grad_norm": 1.0908629174605848, "learning_rate": 6.8112271675238154e-06, "loss": 0.4099, "step": 17268 }, { "epoch": 0.6152448474268308, "grad_norm": 1.5526942335303822, "learning_rate": 6.810133492159783e-06, "loss": 0.5507, "step": 17269 }, { "epoch": 0.6152804745533249, "grad_norm": 1.74354786508786, "learning_rate": 6.8090398592710575e-06, "loss": 0.5548, "step": 17270 }, { "epoch": 0.615316101679819, "grad_norm": 1.5846966305918448, "learning_rate": 6.807946268872192e-06, "loss": 0.5491, "step": 17271 }, { "epoch": 0.6153517288063132, "grad_norm": 1.1055222852182922, "learning_rate": 6.806852720977754e-06, "loss": 0.4209, "step": 17272 }, { "epoch": 0.6153873559328072, "grad_norm": 1.4639386831501071, "learning_rate": 6.805759215602303e-06, "loss": 0.5788, "step": 17273 }, { "epoch": 0.6154229830593013, "grad_norm": 1.540952264188743, "learning_rate": 6.804665752760406e-06, "loss": 0.5966, "step": 17274 }, { "epoch": 0.6154586101857955, "grad_norm": 1.2683742484592546, "learning_rate": 6.8035723324666105e-06, "loss": 0.5302, "step": 17275 }, { "epoch": 0.6154942373122896, "grad_norm": 1.6372669563591602, "learning_rate": 6.80247895473549e-06, "loss": 0.5617, "step": 17276 }, { "epoch": 0.6155298644387837, "grad_norm": 1.1448893831936306, "learning_rate": 6.801385619581592e-06, "loss": 0.3925, "step": 17277 }, { "epoch": 0.6155654915652778, "grad_norm": 1.4696013608507836, "learning_rate": 6.800292327019481e-06, "loss": 0.5376, "step": 17278 }, { "epoch": 0.6156011186917719, "grad_norm": 2.060037698839311, "learning_rate": 6.799199077063716e-06, "loss": 0.5685, "step": 17279 }, { "epoch": 0.615636745818266, "grad_norm": 1.3003006396998775, "learning_rate": 6.798105869728851e-06, "loss": 0.4757, "step": 17280 }, { "epoch": 0.6156723729447602, "grad_norm": 1.1025217413322075, "learning_rate": 6.797012705029447e-06, "loss": 0.4533, "step": 17281 }, { "epoch": 0.6157080000712543, "grad_norm": 1.6659108201711001, "learning_rate": 6.795919582980055e-06, "loss": 0.5332, "step": 17282 }, { "epoch": 0.6157436271977483, "grad_norm": 1.3347653558699795, "learning_rate": 6.794826503595235e-06, "loss": 0.4922, "step": 17283 }, { "epoch": 0.6157792543242425, "grad_norm": 1.253399862488873, "learning_rate": 6.793733466889539e-06, "loss": 0.4812, "step": 17284 }, { "epoch": 0.6158148814507366, "grad_norm": 1.5816104033253506, "learning_rate": 6.7926404728775255e-06, "loss": 0.5673, "step": 17285 }, { "epoch": 0.6158505085772307, "grad_norm": 1.4706135570497325, "learning_rate": 6.791547521573743e-06, "loss": 0.5549, "step": 17286 }, { "epoch": 0.6158861357037249, "grad_norm": 1.3357326775239495, "learning_rate": 6.790454612992749e-06, "loss": 0.4034, "step": 17287 }, { "epoch": 0.6159217628302189, "grad_norm": 1.3306081973894064, "learning_rate": 6.789361747149092e-06, "loss": 0.4701, "step": 17288 }, { "epoch": 0.615957389956713, "grad_norm": 1.2809610817703188, "learning_rate": 6.78826892405733e-06, "loss": 0.4716, "step": 17289 }, { "epoch": 0.6159930170832072, "grad_norm": 1.1975535985493906, "learning_rate": 6.787176143732013e-06, "loss": 0.5026, "step": 17290 }, { "epoch": 0.6160286442097013, "grad_norm": 1.3607736601799632, "learning_rate": 6.786083406187689e-06, "loss": 0.6742, "step": 17291 }, { "epoch": 0.6160642713361953, "grad_norm": 1.5312918597991656, "learning_rate": 6.784990711438916e-06, "loss": 0.5045, "step": 17292 }, { "epoch": 0.6160998984626895, "grad_norm": 1.1308110484981058, "learning_rate": 6.783898059500233e-06, "loss": 0.4655, "step": 17293 }, { "epoch": 0.6161355255891836, "grad_norm": 1.3523035414800548, "learning_rate": 6.7828054503861985e-06, "loss": 0.5618, "step": 17294 }, { "epoch": 0.6161711527156777, "grad_norm": 1.4334572881685756, "learning_rate": 6.7817128841113556e-06, "loss": 0.4428, "step": 17295 }, { "epoch": 0.6162067798421719, "grad_norm": 1.582475800814748, "learning_rate": 6.78062036069026e-06, "loss": 0.5185, "step": 17296 }, { "epoch": 0.616242406968666, "grad_norm": 1.5034696745506688, "learning_rate": 6.779527880137451e-06, "loss": 0.5244, "step": 17297 }, { "epoch": 0.61627803409516, "grad_norm": 1.5706134235118172, "learning_rate": 6.778435442467479e-06, "loss": 0.5416, "step": 17298 }, { "epoch": 0.6163136612216542, "grad_norm": 1.6762311435558999, "learning_rate": 6.777343047694891e-06, "loss": 0.5222, "step": 17299 }, { "epoch": 0.6163492883481483, "grad_norm": 1.0640275614162285, "learning_rate": 6.776250695834234e-06, "loss": 0.3827, "step": 17300 }, { "epoch": 0.6163849154746424, "grad_norm": 1.1550830612487113, "learning_rate": 6.775158386900049e-06, "loss": 0.4377, "step": 17301 }, { "epoch": 0.6164205426011365, "grad_norm": 1.3964858489710663, "learning_rate": 6.774066120906886e-06, "loss": 0.5089, "step": 17302 }, { "epoch": 0.6164561697276306, "grad_norm": 1.386067265112325, "learning_rate": 6.77297389786929e-06, "loss": 0.5296, "step": 17303 }, { "epoch": 0.6164917968541247, "grad_norm": 1.1959197073708814, "learning_rate": 6.771881717801798e-06, "loss": 0.4252, "step": 17304 }, { "epoch": 0.6165274239806189, "grad_norm": 1.1319823639300601, "learning_rate": 6.77078958071896e-06, "loss": 0.4161, "step": 17305 }, { "epoch": 0.616563051107113, "grad_norm": 1.4710947438638986, "learning_rate": 6.769697486635314e-06, "loss": 0.5691, "step": 17306 }, { "epoch": 0.616598678233607, "grad_norm": 1.2559638331194598, "learning_rate": 6.768605435565407e-06, "loss": 0.4374, "step": 17307 }, { "epoch": 0.6166343053601012, "grad_norm": 1.8385867261588145, "learning_rate": 6.767513427523775e-06, "loss": 0.5335, "step": 17308 }, { "epoch": 0.6166699324865953, "grad_norm": 1.3247052614605461, "learning_rate": 6.766421462524962e-06, "loss": 0.5209, "step": 17309 }, { "epoch": 0.6167055596130894, "grad_norm": 1.59373062788248, "learning_rate": 6.765329540583505e-06, "loss": 0.4983, "step": 17310 }, { "epoch": 0.6167411867395836, "grad_norm": 1.4621291191858856, "learning_rate": 6.76423766171395e-06, "loss": 0.6013, "step": 17311 }, { "epoch": 0.6167768138660776, "grad_norm": 1.2648194996009754, "learning_rate": 6.763145825930826e-06, "loss": 0.4687, "step": 17312 }, { "epoch": 0.6168124409925717, "grad_norm": 1.6747073572576778, "learning_rate": 6.762054033248681e-06, "loss": 0.5123, "step": 17313 }, { "epoch": 0.6168480681190659, "grad_norm": 1.3175295717686324, "learning_rate": 6.760962283682054e-06, "loss": 0.614, "step": 17314 }, { "epoch": 0.61688369524556, "grad_norm": 1.210913932966468, "learning_rate": 6.7598705772454724e-06, "loss": 0.4389, "step": 17315 }, { "epoch": 0.616919322372054, "grad_norm": 1.205524297158502, "learning_rate": 6.758778913953481e-06, "loss": 0.4184, "step": 17316 }, { "epoch": 0.6169549494985482, "grad_norm": 1.3747921454499357, "learning_rate": 6.7576872938206115e-06, "loss": 0.5423, "step": 17317 }, { "epoch": 0.6169905766250423, "grad_norm": 1.0826156171457706, "learning_rate": 6.756595716861407e-06, "loss": 0.3833, "step": 17318 }, { "epoch": 0.6170262037515364, "grad_norm": 1.4081579838719733, "learning_rate": 6.755504183090393e-06, "loss": 0.5696, "step": 17319 }, { "epoch": 0.6170618308780306, "grad_norm": 1.5287811992930407, "learning_rate": 6.754412692522112e-06, "loss": 0.5357, "step": 17320 }, { "epoch": 0.6170974580045246, "grad_norm": 1.1723129016494476, "learning_rate": 6.7533212451710914e-06, "loss": 0.4079, "step": 17321 }, { "epoch": 0.6171330851310187, "grad_norm": 1.0512381818598366, "learning_rate": 6.752229841051872e-06, "loss": 0.4045, "step": 17322 }, { "epoch": 0.6171687122575129, "grad_norm": 1.904126180899226, "learning_rate": 6.751138480178978e-06, "loss": 0.4894, "step": 17323 }, { "epoch": 0.617204339384007, "grad_norm": 1.429705706226225, "learning_rate": 6.750047162566944e-06, "loss": 0.4981, "step": 17324 }, { "epoch": 0.6172399665105011, "grad_norm": 2.1958891291882776, "learning_rate": 6.7489558882303084e-06, "loss": 0.4316, "step": 17325 }, { "epoch": 0.6172755936369952, "grad_norm": 1.5342320096572137, "learning_rate": 6.747864657183596e-06, "loss": 0.5246, "step": 17326 }, { "epoch": 0.6173112207634893, "grad_norm": 1.2182637262240916, "learning_rate": 6.746773469441339e-06, "loss": 0.4564, "step": 17327 }, { "epoch": 0.6173468478899834, "grad_norm": 1.1252637666771408, "learning_rate": 6.745682325018066e-06, "loss": 0.436, "step": 17328 }, { "epoch": 0.6173824750164776, "grad_norm": 1.5161569430757886, "learning_rate": 6.744591223928311e-06, "loss": 0.4926, "step": 17329 }, { "epoch": 0.6174181021429717, "grad_norm": 1.5059104563477015, "learning_rate": 6.743500166186595e-06, "loss": 0.5669, "step": 17330 }, { "epoch": 0.6174537292694657, "grad_norm": 1.3308926562657077, "learning_rate": 6.742409151807453e-06, "loss": 0.4397, "step": 17331 }, { "epoch": 0.6174893563959599, "grad_norm": 1.1940831283477178, "learning_rate": 6.741318180805408e-06, "loss": 0.4231, "step": 17332 }, { "epoch": 0.617524983522454, "grad_norm": 1.4434308973986205, "learning_rate": 6.740227253194993e-06, "loss": 0.5254, "step": 17333 }, { "epoch": 0.6175606106489481, "grad_norm": 1.0364532127914812, "learning_rate": 6.739136368990726e-06, "loss": 0.4349, "step": 17334 }, { "epoch": 0.6175962377754423, "grad_norm": 1.2170946016752267, "learning_rate": 6.738045528207137e-06, "loss": 0.5041, "step": 17335 }, { "epoch": 0.6176318649019363, "grad_norm": 1.4166959969616628, "learning_rate": 6.736954730858756e-06, "loss": 0.4985, "step": 17336 }, { "epoch": 0.6176674920284304, "grad_norm": 1.1152970384384209, "learning_rate": 6.7358639769601e-06, "loss": 0.4034, "step": 17337 }, { "epoch": 0.6177031191549246, "grad_norm": 1.2873258564130612, "learning_rate": 6.7347732665257e-06, "loss": 0.5415, "step": 17338 }, { "epoch": 0.6177387462814187, "grad_norm": 1.677692868071374, "learning_rate": 6.733682599570073e-06, "loss": 0.5528, "step": 17339 }, { "epoch": 0.6177743734079127, "grad_norm": 2.398060719218352, "learning_rate": 6.732591976107749e-06, "loss": 0.5142, "step": 17340 }, { "epoch": 0.6178100005344069, "grad_norm": 1.2904490360050078, "learning_rate": 6.731501396153243e-06, "loss": 0.3891, "step": 17341 }, { "epoch": 0.617845627660901, "grad_norm": 1.137825964966336, "learning_rate": 6.730410859721081e-06, "loss": 0.483, "step": 17342 }, { "epoch": 0.6178812547873951, "grad_norm": 1.210865947150071, "learning_rate": 6.729320366825785e-06, "loss": 0.4674, "step": 17343 }, { "epoch": 0.6179168819138893, "grad_norm": 1.4561100447274162, "learning_rate": 6.7282299174818756e-06, "loss": 0.669, "step": 17344 }, { "epoch": 0.6179525090403833, "grad_norm": 1.6125485978396161, "learning_rate": 6.7271395117038685e-06, "loss": 0.5331, "step": 17345 }, { "epoch": 0.6179881361668774, "grad_norm": 1.4715188989159658, "learning_rate": 6.7260491495062885e-06, "loss": 0.5518, "step": 17346 }, { "epoch": 0.6180237632933716, "grad_norm": 1.0542350076021023, "learning_rate": 6.724958830903649e-06, "loss": 0.4361, "step": 17347 }, { "epoch": 0.6180593904198657, "grad_norm": 1.3176627067840887, "learning_rate": 6.7238685559104745e-06, "loss": 0.5535, "step": 17348 }, { "epoch": 0.6180950175463598, "grad_norm": 1.5958354192778053, "learning_rate": 6.722778324541283e-06, "loss": 0.4724, "step": 17349 }, { "epoch": 0.618130644672854, "grad_norm": 1.168992522183253, "learning_rate": 6.721688136810582e-06, "loss": 0.3988, "step": 17350 }, { "epoch": 0.618166271799348, "grad_norm": 1.274601073769081, "learning_rate": 6.720597992732901e-06, "loss": 0.5942, "step": 17351 }, { "epoch": 0.6182018989258421, "grad_norm": 1.1481900050669873, "learning_rate": 6.719507892322747e-06, "loss": 0.4088, "step": 17352 }, { "epoch": 0.6182375260523363, "grad_norm": 1.1946130191636912, "learning_rate": 6.7184178355946395e-06, "loss": 0.4403, "step": 17353 }, { "epoch": 0.6182731531788304, "grad_norm": 1.2954533800177102, "learning_rate": 6.717327822563089e-06, "loss": 0.4304, "step": 17354 }, { "epoch": 0.6183087803053244, "grad_norm": 1.1156577589019243, "learning_rate": 6.71623785324262e-06, "loss": 0.4231, "step": 17355 }, { "epoch": 0.6183444074318186, "grad_norm": 1.573297588735148, "learning_rate": 6.715147927647733e-06, "loss": 0.491, "step": 17356 }, { "epoch": 0.6183800345583127, "grad_norm": 1.4937810764586321, "learning_rate": 6.714058045792949e-06, "loss": 0.543, "step": 17357 }, { "epoch": 0.6184156616848068, "grad_norm": 1.2253377734919795, "learning_rate": 6.712968207692778e-06, "loss": 0.4424, "step": 17358 }, { "epoch": 0.618451288811301, "grad_norm": 1.0917422162948807, "learning_rate": 6.711878413361732e-06, "loss": 0.4417, "step": 17359 }, { "epoch": 0.618486915937795, "grad_norm": 1.5459588126404171, "learning_rate": 6.710788662814328e-06, "loss": 0.5221, "step": 17360 }, { "epoch": 0.6185225430642891, "grad_norm": 1.1092717947029007, "learning_rate": 6.709698956065067e-06, "loss": 0.4441, "step": 17361 }, { "epoch": 0.6185581701907833, "grad_norm": 1.6224619346467337, "learning_rate": 6.708609293128467e-06, "loss": 0.5413, "step": 17362 }, { "epoch": 0.6185937973172774, "grad_norm": 1.0257806735823471, "learning_rate": 6.7075196740190326e-06, "loss": 0.44, "step": 17363 }, { "epoch": 0.6186294244437714, "grad_norm": 1.1813753015071553, "learning_rate": 6.706430098751278e-06, "loss": 0.4261, "step": 17364 }, { "epoch": 0.6186650515702656, "grad_norm": 1.437392586998466, "learning_rate": 6.7053405673397045e-06, "loss": 0.4654, "step": 17365 }, { "epoch": 0.6187006786967597, "grad_norm": 1.2218689414301813, "learning_rate": 6.7042510797988304e-06, "loss": 0.4271, "step": 17366 }, { "epoch": 0.6187363058232538, "grad_norm": 1.2239589333699907, "learning_rate": 6.703161636143152e-06, "loss": 0.4549, "step": 17367 }, { "epoch": 0.618771932949748, "grad_norm": 1.8894243728986793, "learning_rate": 6.702072236387182e-06, "loss": 0.5099, "step": 17368 }, { "epoch": 0.618807560076242, "grad_norm": 1.1434425661670056, "learning_rate": 6.700982880545424e-06, "loss": 0.4089, "step": 17369 }, { "epoch": 0.6188431872027361, "grad_norm": 1.156719241009993, "learning_rate": 6.699893568632384e-06, "loss": 0.3986, "step": 17370 }, { "epoch": 0.6188788143292303, "grad_norm": 1.058439551792892, "learning_rate": 6.698804300662574e-06, "loss": 0.387, "step": 17371 }, { "epoch": 0.6189144414557244, "grad_norm": 1.1243604631732367, "learning_rate": 6.697715076650486e-06, "loss": 0.4405, "step": 17372 }, { "epoch": 0.6189500685822185, "grad_norm": 1.4983595060144443, "learning_rate": 6.6966258966106314e-06, "loss": 0.5744, "step": 17373 }, { "epoch": 0.6189856957087126, "grad_norm": 1.1791791388704462, "learning_rate": 6.69553676055751e-06, "loss": 0.4086, "step": 17374 }, { "epoch": 0.6190213228352067, "grad_norm": 1.3624818683869686, "learning_rate": 6.6944476685056324e-06, "loss": 0.4179, "step": 17375 }, { "epoch": 0.6190569499617008, "grad_norm": 1.3910415654953727, "learning_rate": 6.693358620469488e-06, "loss": 0.5488, "step": 17376 }, { "epoch": 0.619092577088195, "grad_norm": 1.270717526744955, "learning_rate": 6.692269616463589e-06, "loss": 0.4617, "step": 17377 }, { "epoch": 0.6191282042146891, "grad_norm": 1.3517797201909136, "learning_rate": 6.691180656502429e-06, "loss": 0.5494, "step": 17378 }, { "epoch": 0.6191638313411831, "grad_norm": 1.116286262053526, "learning_rate": 6.690091740600515e-06, "loss": 0.4224, "step": 17379 }, { "epoch": 0.6191994584676773, "grad_norm": 1.2395351522398201, "learning_rate": 6.689002868772339e-06, "loss": 0.4202, "step": 17380 }, { "epoch": 0.6192350855941714, "grad_norm": 1.4267057578627678, "learning_rate": 6.687914041032409e-06, "loss": 0.6141, "step": 17381 }, { "epoch": 0.6192707127206655, "grad_norm": 1.1275034815013651, "learning_rate": 6.686825257395213e-06, "loss": 0.3845, "step": 17382 }, { "epoch": 0.6193063398471597, "grad_norm": 1.4317138148385227, "learning_rate": 6.6857365178752565e-06, "loss": 0.5842, "step": 17383 }, { "epoch": 0.6193419669736537, "grad_norm": 1.5698773940297743, "learning_rate": 6.684647822487036e-06, "loss": 0.4634, "step": 17384 }, { "epoch": 0.6193775941001478, "grad_norm": 1.1850357004100736, "learning_rate": 6.683559171245044e-06, "loss": 0.4661, "step": 17385 }, { "epoch": 0.619413221226642, "grad_norm": 1.2235323884361096, "learning_rate": 6.682470564163785e-06, "loss": 0.4276, "step": 17386 }, { "epoch": 0.6194488483531361, "grad_norm": 1.246286764635738, "learning_rate": 6.681382001257743e-06, "loss": 0.4416, "step": 17387 }, { "epoch": 0.6194844754796301, "grad_norm": 1.4517576824919685, "learning_rate": 6.680293482541424e-06, "loss": 0.5598, "step": 17388 }, { "epoch": 0.6195201026061243, "grad_norm": 1.2348982191190392, "learning_rate": 6.679205008029315e-06, "loss": 0.4726, "step": 17389 }, { "epoch": 0.6195557297326184, "grad_norm": 1.6744758273303346, "learning_rate": 6.678116577735915e-06, "loss": 0.569, "step": 17390 }, { "epoch": 0.6195913568591125, "grad_norm": 1.498765598753787, "learning_rate": 6.677028191675712e-06, "loss": 0.4493, "step": 17391 }, { "epoch": 0.6196269839856067, "grad_norm": 1.476085056504301, "learning_rate": 6.6759398498632044e-06, "loss": 0.5163, "step": 17392 }, { "epoch": 0.6196626111121007, "grad_norm": 1.2361505321512387, "learning_rate": 6.674851552312878e-06, "loss": 0.4943, "step": 17393 }, { "epoch": 0.6196982382385948, "grad_norm": 1.6520139902434379, "learning_rate": 6.6737632990392264e-06, "loss": 0.5165, "step": 17394 }, { "epoch": 0.619733865365089, "grad_norm": 1.332622299002557, "learning_rate": 6.672675090056744e-06, "loss": 0.5316, "step": 17395 }, { "epoch": 0.6197694924915831, "grad_norm": 1.1404460246387274, "learning_rate": 6.671586925379916e-06, "loss": 0.4588, "step": 17396 }, { "epoch": 0.6198051196180772, "grad_norm": 1.2448983310968438, "learning_rate": 6.67049880502324e-06, "loss": 0.3899, "step": 17397 }, { "epoch": 0.6198407467445713, "grad_norm": 1.0745295763002072, "learning_rate": 6.669410729001193e-06, "loss": 0.4493, "step": 17398 }, { "epoch": 0.6198763738710654, "grad_norm": 1.2065307327037431, "learning_rate": 6.668322697328273e-06, "loss": 0.4441, "step": 17399 }, { "epoch": 0.6199120009975595, "grad_norm": 1.85549491926752, "learning_rate": 6.667234710018964e-06, "loss": 0.5936, "step": 17400 }, { "epoch": 0.6199476281240537, "grad_norm": 1.7165319762269455, "learning_rate": 6.666146767087756e-06, "loss": 0.4753, "step": 17401 }, { "epoch": 0.6199832552505478, "grad_norm": 1.2029704900621967, "learning_rate": 6.6650588685491325e-06, "loss": 0.4241, "step": 17402 }, { "epoch": 0.6200188823770418, "grad_norm": 1.2537183722427387, "learning_rate": 6.663971014417585e-06, "loss": 0.4524, "step": 17403 }, { "epoch": 0.620054509503536, "grad_norm": 1.161497239790343, "learning_rate": 6.66288320470759e-06, "loss": 0.4479, "step": 17404 }, { "epoch": 0.6200901366300301, "grad_norm": 1.3946784157202694, "learning_rate": 6.661795439433638e-06, "loss": 0.5128, "step": 17405 }, { "epoch": 0.6201257637565242, "grad_norm": 1.3143210773542615, "learning_rate": 6.660707718610215e-06, "loss": 0.4165, "step": 17406 }, { "epoch": 0.6201613908830184, "grad_norm": 1.174267895919773, "learning_rate": 6.659620042251802e-06, "loss": 0.4423, "step": 17407 }, { "epoch": 0.6201970180095124, "grad_norm": 1.1230556900633417, "learning_rate": 6.658532410372886e-06, "loss": 0.4106, "step": 17408 }, { "epoch": 0.6202326451360065, "grad_norm": 1.747315859308274, "learning_rate": 6.657444822987942e-06, "loss": 0.4653, "step": 17409 }, { "epoch": 0.6202682722625007, "grad_norm": 1.6309906561901348, "learning_rate": 6.656357280111458e-06, "loss": 0.5952, "step": 17410 }, { "epoch": 0.6203038993889948, "grad_norm": 1.6494114148915093, "learning_rate": 6.655269781757914e-06, "loss": 0.598, "step": 17411 }, { "epoch": 0.6203395265154888, "grad_norm": 1.2174841763119786, "learning_rate": 6.654182327941794e-06, "loss": 0.4367, "step": 17412 }, { "epoch": 0.620375153641983, "grad_norm": 1.6207491895770323, "learning_rate": 6.653094918677567e-06, "loss": 0.5332, "step": 17413 }, { "epoch": 0.6204107807684771, "grad_norm": 1.702643597274322, "learning_rate": 6.652007553979729e-06, "loss": 0.4814, "step": 17414 }, { "epoch": 0.6204464078949712, "grad_norm": 1.559043137373421, "learning_rate": 6.6509202338627456e-06, "loss": 0.5228, "step": 17415 }, { "epoch": 0.6204820350214654, "grad_norm": 1.2969789360097252, "learning_rate": 6.6498329583411e-06, "loss": 0.4573, "step": 17416 }, { "epoch": 0.6205176621479594, "grad_norm": 1.2735700760081143, "learning_rate": 6.648745727429272e-06, "loss": 0.4569, "step": 17417 }, { "epoch": 0.6205532892744535, "grad_norm": 1.226225873584861, "learning_rate": 6.647658541141735e-06, "loss": 0.4444, "step": 17418 }, { "epoch": 0.6205889164009477, "grad_norm": 1.5147679464509445, "learning_rate": 6.646571399492972e-06, "loss": 0.4928, "step": 17419 }, { "epoch": 0.6206245435274418, "grad_norm": 1.637993455970354, "learning_rate": 6.645484302497452e-06, "loss": 0.5045, "step": 17420 }, { "epoch": 0.6206601706539359, "grad_norm": 1.6491159776227828, "learning_rate": 6.644397250169652e-06, "loss": 0.5349, "step": 17421 }, { "epoch": 0.62069579778043, "grad_norm": 1.179034248454816, "learning_rate": 6.643310242524048e-06, "loss": 0.4497, "step": 17422 }, { "epoch": 0.6207314249069241, "grad_norm": 1.4806253047220126, "learning_rate": 6.642223279575118e-06, "loss": 0.6046, "step": 17423 }, { "epoch": 0.6207670520334182, "grad_norm": 1.5830693904983202, "learning_rate": 6.641136361337328e-06, "loss": 0.612, "step": 17424 }, { "epoch": 0.6208026791599124, "grad_norm": 1.4839844346236104, "learning_rate": 6.640049487825158e-06, "loss": 0.5148, "step": 17425 }, { "epoch": 0.6208383062864065, "grad_norm": 1.1669906696240246, "learning_rate": 6.638962659053074e-06, "loss": 0.4077, "step": 17426 }, { "epoch": 0.6208739334129005, "grad_norm": 1.4743637287471372, "learning_rate": 6.637875875035555e-06, "loss": 0.5502, "step": 17427 }, { "epoch": 0.6209095605393947, "grad_norm": 1.3769505696039126, "learning_rate": 6.636789135787065e-06, "loss": 0.545, "step": 17428 }, { "epoch": 0.6209451876658888, "grad_norm": 1.6278610966570317, "learning_rate": 6.635702441322081e-06, "loss": 0.607, "step": 17429 }, { "epoch": 0.6209808147923829, "grad_norm": 1.820076200218645, "learning_rate": 6.634615791655071e-06, "loss": 0.5467, "step": 17430 }, { "epoch": 0.6210164419188771, "grad_norm": 1.3734126458600402, "learning_rate": 6.633529186800503e-06, "loss": 0.5245, "step": 17431 }, { "epoch": 0.6210520690453711, "grad_norm": 1.7062710831501884, "learning_rate": 6.632442626772847e-06, "loss": 0.608, "step": 17432 }, { "epoch": 0.6210876961718652, "grad_norm": 1.6194200528325973, "learning_rate": 6.63135611158657e-06, "loss": 0.5532, "step": 17433 }, { "epoch": 0.6211233232983594, "grad_norm": 1.1967749781665022, "learning_rate": 6.630269641256144e-06, "loss": 0.4936, "step": 17434 }, { "epoch": 0.6211589504248535, "grad_norm": 1.893744610613656, "learning_rate": 6.62918321579603e-06, "loss": 0.4697, "step": 17435 }, { "epoch": 0.6211945775513475, "grad_norm": 1.5011017570340193, "learning_rate": 6.6280968352207e-06, "loss": 0.5611, "step": 17436 }, { "epoch": 0.6212302046778417, "grad_norm": 1.4725482086112645, "learning_rate": 6.6270104995446146e-06, "loss": 0.4968, "step": 17437 }, { "epoch": 0.6212658318043358, "grad_norm": 1.6017976468507484, "learning_rate": 6.625924208782246e-06, "loss": 0.5795, "step": 17438 }, { "epoch": 0.6213014589308299, "grad_norm": 1.2658173210283306, "learning_rate": 6.62483796294805e-06, "loss": 0.4373, "step": 17439 }, { "epoch": 0.6213370860573241, "grad_norm": 1.1315317175296866, "learning_rate": 6.623751762056496e-06, "loss": 0.4035, "step": 17440 }, { "epoch": 0.6213727131838181, "grad_norm": 1.3230502923538796, "learning_rate": 6.622665606122052e-06, "loss": 0.4556, "step": 17441 }, { "epoch": 0.6214083403103122, "grad_norm": 1.4033205959353776, "learning_rate": 6.621579495159172e-06, "loss": 0.5266, "step": 17442 }, { "epoch": 0.6214439674368064, "grad_norm": 1.1009798288044836, "learning_rate": 6.620493429182324e-06, "loss": 0.4332, "step": 17443 }, { "epoch": 0.6214795945633005, "grad_norm": 1.436708543788379, "learning_rate": 6.6194074082059665e-06, "loss": 0.4791, "step": 17444 }, { "epoch": 0.6215152216897946, "grad_norm": 1.767027498205745, "learning_rate": 6.618321432244565e-06, "loss": 0.5729, "step": 17445 }, { "epoch": 0.6215508488162887, "grad_norm": 1.395502790354695, "learning_rate": 6.617235501312575e-06, "loss": 0.5046, "step": 17446 }, { "epoch": 0.6215864759427828, "grad_norm": 1.2997123926594611, "learning_rate": 6.616149615424461e-06, "loss": 0.4835, "step": 17447 }, { "epoch": 0.6216221030692769, "grad_norm": 1.3089884837697525, "learning_rate": 6.615063774594677e-06, "loss": 0.4766, "step": 17448 }, { "epoch": 0.6216577301957711, "grad_norm": 1.4344339741866905, "learning_rate": 6.613977978837689e-06, "loss": 0.5754, "step": 17449 }, { "epoch": 0.6216933573222652, "grad_norm": 2.675499869194355, "learning_rate": 6.612892228167946e-06, "loss": 0.5401, "step": 17450 }, { "epoch": 0.6217289844487592, "grad_norm": 1.1976118461520373, "learning_rate": 6.61180652259991e-06, "loss": 0.4871, "step": 17451 }, { "epoch": 0.6217646115752534, "grad_norm": 1.6458061119361753, "learning_rate": 6.610720862148043e-06, "loss": 0.5988, "step": 17452 }, { "epoch": 0.6218002387017475, "grad_norm": 1.1441359605700454, "learning_rate": 6.6096352468267935e-06, "loss": 0.4541, "step": 17453 }, { "epoch": 0.6218358658282416, "grad_norm": 1.6442216622027013, "learning_rate": 6.608549676650622e-06, "loss": 0.4712, "step": 17454 }, { "epoch": 0.6218714929547358, "grad_norm": 1.1861560051681874, "learning_rate": 6.607464151633981e-06, "loss": 0.4455, "step": 17455 }, { "epoch": 0.6219071200812298, "grad_norm": 1.455630399011961, "learning_rate": 6.606378671791329e-06, "loss": 0.4728, "step": 17456 }, { "epoch": 0.6219427472077239, "grad_norm": 1.681169916344775, "learning_rate": 6.6052932371371135e-06, "loss": 0.5277, "step": 17457 }, { "epoch": 0.6219783743342181, "grad_norm": 1.2293400292429009, "learning_rate": 6.604207847685793e-06, "loss": 0.4348, "step": 17458 }, { "epoch": 0.6220140014607122, "grad_norm": 1.1816459010736355, "learning_rate": 6.603122503451818e-06, "loss": 0.4335, "step": 17459 }, { "epoch": 0.6220496285872062, "grad_norm": 1.2630399661510572, "learning_rate": 6.602037204449643e-06, "loss": 0.4483, "step": 17460 }, { "epoch": 0.6220852557137004, "grad_norm": 1.455712478045089, "learning_rate": 6.6009519506937144e-06, "loss": 0.4211, "step": 17461 }, { "epoch": 0.6221208828401945, "grad_norm": 1.1064223891992317, "learning_rate": 6.599866742198486e-06, "loss": 0.4627, "step": 17462 }, { "epoch": 0.6221565099666886, "grad_norm": 1.3063608545185474, "learning_rate": 6.59878157897841e-06, "loss": 0.4348, "step": 17463 }, { "epoch": 0.6221921370931828, "grad_norm": 1.3549254403536886, "learning_rate": 6.597696461047935e-06, "loss": 0.4489, "step": 17464 }, { "epoch": 0.6222277642196768, "grad_norm": 1.3755447454560565, "learning_rate": 6.59661138842151e-06, "loss": 0.5379, "step": 17465 }, { "epoch": 0.6222633913461709, "grad_norm": 1.2285992430191992, "learning_rate": 6.595526361113581e-06, "loss": 0.4276, "step": 17466 }, { "epoch": 0.6222990184726651, "grad_norm": 1.1776658461202163, "learning_rate": 6.594441379138603e-06, "loss": 0.4493, "step": 17467 }, { "epoch": 0.6223346455991592, "grad_norm": 1.0323198575685435, "learning_rate": 6.593356442511014e-06, "loss": 0.4309, "step": 17468 }, { "epoch": 0.6223702727256533, "grad_norm": 1.5209886648879887, "learning_rate": 6.592271551245266e-06, "loss": 0.5259, "step": 17469 }, { "epoch": 0.6224058998521474, "grad_norm": 1.7507468430729312, "learning_rate": 6.591186705355803e-06, "loss": 0.6232, "step": 17470 }, { "epoch": 0.6224415269786415, "grad_norm": 2.2634064803515717, "learning_rate": 6.590101904857076e-06, "loss": 0.5203, "step": 17471 }, { "epoch": 0.6224771541051356, "grad_norm": 1.6327822051456644, "learning_rate": 6.589017149763522e-06, "loss": 0.5498, "step": 17472 }, { "epoch": 0.6225127812316298, "grad_norm": 1.56441961435652, "learning_rate": 6.58793244008959e-06, "loss": 0.611, "step": 17473 }, { "epoch": 0.6225484083581239, "grad_norm": 1.6322675155909716, "learning_rate": 6.586847775849721e-06, "loss": 0.5276, "step": 17474 }, { "epoch": 0.6225840354846179, "grad_norm": 1.271778219431791, "learning_rate": 6.585763157058358e-06, "loss": 0.4779, "step": 17475 }, { "epoch": 0.6226196626111121, "grad_norm": 1.5312383229785, "learning_rate": 6.584678583729947e-06, "loss": 0.6399, "step": 17476 }, { "epoch": 0.6226552897376062, "grad_norm": 1.1372940930255047, "learning_rate": 6.583594055878928e-06, "loss": 0.4228, "step": 17477 }, { "epoch": 0.6226909168641003, "grad_norm": 1.698301544099474, "learning_rate": 6.5825095735197445e-06, "loss": 0.5142, "step": 17478 }, { "epoch": 0.6227265439905945, "grad_norm": 1.3694354303565708, "learning_rate": 6.58142513666683e-06, "loss": 0.5539, "step": 17479 }, { "epoch": 0.6227621711170885, "grad_norm": 1.640833002797523, "learning_rate": 6.580340745334633e-06, "loss": 0.6404, "step": 17480 }, { "epoch": 0.6227977982435826, "grad_norm": 1.67195421281987, "learning_rate": 6.579256399537587e-06, "loss": 0.5702, "step": 17481 }, { "epoch": 0.6228334253700768, "grad_norm": 1.1917817381175808, "learning_rate": 6.578172099290136e-06, "loss": 0.4375, "step": 17482 }, { "epoch": 0.6228690524965709, "grad_norm": 1.2258858727552995, "learning_rate": 6.577087844606712e-06, "loss": 0.465, "step": 17483 }, { "epoch": 0.622904679623065, "grad_norm": 1.1984490320409524, "learning_rate": 6.576003635501758e-06, "loss": 0.442, "step": 17484 }, { "epoch": 0.6229403067495591, "grad_norm": 1.5172359769455854, "learning_rate": 6.574919471989706e-06, "loss": 0.5605, "step": 17485 }, { "epoch": 0.6229759338760532, "grad_norm": 1.1664690391068728, "learning_rate": 6.573835354084995e-06, "loss": 0.3898, "step": 17486 }, { "epoch": 0.6230115610025473, "grad_norm": 1.3254214993253945, "learning_rate": 6.572751281802065e-06, "loss": 0.5543, "step": 17487 }, { "epoch": 0.6230471881290415, "grad_norm": 1.3476536778878323, "learning_rate": 6.571667255155343e-06, "loss": 0.4018, "step": 17488 }, { "epoch": 0.6230828152555355, "grad_norm": 1.23187823104801, "learning_rate": 6.570583274159272e-06, "loss": 0.4351, "step": 17489 }, { "epoch": 0.6231184423820296, "grad_norm": 1.2182930313229252, "learning_rate": 6.569499338828278e-06, "loss": 0.4005, "step": 17490 }, { "epoch": 0.6231540695085238, "grad_norm": 1.1137467238752221, "learning_rate": 6.5684154491768005e-06, "loss": 0.4017, "step": 17491 }, { "epoch": 0.6231896966350179, "grad_norm": 1.2778581769941353, "learning_rate": 6.567331605219267e-06, "loss": 0.4383, "step": 17492 }, { "epoch": 0.623225323761512, "grad_norm": 1.403074370173268, "learning_rate": 6.566247806970119e-06, "loss": 0.55, "step": 17493 }, { "epoch": 0.6232609508880061, "grad_norm": 1.4259999268183439, "learning_rate": 6.565164054443775e-06, "loss": 0.555, "step": 17494 }, { "epoch": 0.6232965780145002, "grad_norm": 1.567314570002324, "learning_rate": 6.564080347654675e-06, "loss": 0.5028, "step": 17495 }, { "epoch": 0.6233322051409943, "grad_norm": 1.2552418599818431, "learning_rate": 6.5629966866172445e-06, "loss": 0.5757, "step": 17496 }, { "epoch": 0.6233678322674885, "grad_norm": 1.11148473238622, "learning_rate": 6.561913071345916e-06, "loss": 0.4659, "step": 17497 }, { "epoch": 0.6234034593939826, "grad_norm": 1.2736068553917084, "learning_rate": 6.560829501855121e-06, "loss": 0.5671, "step": 17498 }, { "epoch": 0.6234390865204766, "grad_norm": 1.4618249968313302, "learning_rate": 6.559745978159282e-06, "loss": 0.5084, "step": 17499 }, { "epoch": 0.6234747136469708, "grad_norm": 1.4687402347198828, "learning_rate": 6.558662500272832e-06, "loss": 0.5192, "step": 17500 }, { "epoch": 0.6235103407734649, "grad_norm": 1.3639043737357315, "learning_rate": 6.557579068210193e-06, "loss": 0.5736, "step": 17501 }, { "epoch": 0.623545967899959, "grad_norm": 1.4572935857943052, "learning_rate": 6.556495681985797e-06, "loss": 0.5451, "step": 17502 }, { "epoch": 0.6235815950264532, "grad_norm": 1.1346856834722436, "learning_rate": 6.555412341614067e-06, "loss": 0.4008, "step": 17503 }, { "epoch": 0.6236172221529472, "grad_norm": 1.7921428518453315, "learning_rate": 6.554329047109433e-06, "loss": 0.4958, "step": 17504 }, { "epoch": 0.6236528492794413, "grad_norm": 1.205574477868127, "learning_rate": 6.553245798486312e-06, "loss": 0.4106, "step": 17505 }, { "epoch": 0.6236884764059355, "grad_norm": 1.2830356048864668, "learning_rate": 6.552162595759135e-06, "loss": 0.5449, "step": 17506 }, { "epoch": 0.6237241035324296, "grad_norm": 1.170113111427578, "learning_rate": 6.551079438942319e-06, "loss": 0.4036, "step": 17507 }, { "epoch": 0.6237597306589237, "grad_norm": 1.2091509056940122, "learning_rate": 6.549996328050296e-06, "loss": 0.4398, "step": 17508 }, { "epoch": 0.6237953577854178, "grad_norm": 1.2797568306853853, "learning_rate": 6.54891326309748e-06, "loss": 0.5259, "step": 17509 }, { "epoch": 0.6238309849119119, "grad_norm": 1.0839530794040362, "learning_rate": 6.547830244098296e-06, "loss": 0.44, "step": 17510 }, { "epoch": 0.623866612038406, "grad_norm": 1.3232462787840007, "learning_rate": 6.546747271067167e-06, "loss": 0.4469, "step": 17511 }, { "epoch": 0.6239022391649002, "grad_norm": 1.8316403656133655, "learning_rate": 6.545664344018509e-06, "loss": 0.6426, "step": 17512 }, { "epoch": 0.6239378662913942, "grad_norm": 2.007003195786579, "learning_rate": 6.544581462966748e-06, "loss": 0.5518, "step": 17513 }, { "epoch": 0.6239734934178883, "grad_norm": 1.444251112881079, "learning_rate": 6.543498627926296e-06, "loss": 0.6038, "step": 17514 }, { "epoch": 0.6240091205443825, "grad_norm": 1.6431703172856267, "learning_rate": 6.54241583891158e-06, "loss": 0.7313, "step": 17515 }, { "epoch": 0.6240447476708766, "grad_norm": 1.2437870932023796, "learning_rate": 6.541333095937011e-06, "loss": 0.4867, "step": 17516 }, { "epoch": 0.6240803747973707, "grad_norm": 1.1955085025751115, "learning_rate": 6.540250399017012e-06, "loss": 0.4569, "step": 17517 }, { "epoch": 0.6241160019238648, "grad_norm": 1.304337144916472, "learning_rate": 6.539167748165993e-06, "loss": 0.4465, "step": 17518 }, { "epoch": 0.6241516290503589, "grad_norm": 1.9138200669490384, "learning_rate": 6.538085143398381e-06, "loss": 0.5506, "step": 17519 }, { "epoch": 0.6241872561768531, "grad_norm": 1.3556486630041857, "learning_rate": 6.537002584728578e-06, "loss": 0.4445, "step": 17520 }, { "epoch": 0.6242228833033472, "grad_norm": 1.2396866004207592, "learning_rate": 6.535920072171008e-06, "loss": 0.444, "step": 17521 }, { "epoch": 0.6242585104298413, "grad_norm": 1.5174745616803134, "learning_rate": 6.534837605740085e-06, "loss": 0.5229, "step": 17522 }, { "epoch": 0.6242941375563354, "grad_norm": 1.1624426777031667, "learning_rate": 6.533755185450219e-06, "loss": 0.4196, "step": 17523 }, { "epoch": 0.6243297646828295, "grad_norm": 2.2230098819843525, "learning_rate": 6.53267281131583e-06, "loss": 0.5171, "step": 17524 }, { "epoch": 0.6243653918093236, "grad_norm": 1.0968296049267967, "learning_rate": 6.531590483351322e-06, "loss": 0.3883, "step": 17525 }, { "epoch": 0.6244010189358178, "grad_norm": 1.2164985778584607, "learning_rate": 6.530508201571113e-06, "loss": 0.4206, "step": 17526 }, { "epoch": 0.6244366460623119, "grad_norm": 1.0275562559159455, "learning_rate": 6.529425965989611e-06, "loss": 0.3957, "step": 17527 }, { "epoch": 0.6244722731888059, "grad_norm": 1.0063746971358911, "learning_rate": 6.528343776621229e-06, "loss": 0.3936, "step": 17528 }, { "epoch": 0.6245079003153001, "grad_norm": 1.210361929893198, "learning_rate": 6.5272616334803755e-06, "loss": 0.5413, "step": 17529 }, { "epoch": 0.6245435274417942, "grad_norm": 1.718411246786957, "learning_rate": 6.5261795365814625e-06, "loss": 0.5374, "step": 17530 }, { "epoch": 0.6245791545682883, "grad_norm": 1.6539074566442165, "learning_rate": 6.525097485938894e-06, "loss": 0.4638, "step": 17531 }, { "epoch": 0.6246147816947825, "grad_norm": 1.1897104549729531, "learning_rate": 6.524015481567081e-06, "loss": 0.4227, "step": 17532 }, { "epoch": 0.6246504088212765, "grad_norm": 1.692527626796873, "learning_rate": 6.522933523480435e-06, "loss": 0.6036, "step": 17533 }, { "epoch": 0.6246860359477706, "grad_norm": 1.6004133303460633, "learning_rate": 6.521851611693355e-06, "loss": 0.5417, "step": 17534 }, { "epoch": 0.6247216630742648, "grad_norm": 1.3495200748133, "learning_rate": 6.520769746220258e-06, "loss": 0.5235, "step": 17535 }, { "epoch": 0.6247572902007589, "grad_norm": 1.3894343257701678, "learning_rate": 6.519687927075539e-06, "loss": 0.5409, "step": 17536 }, { "epoch": 0.624792917327253, "grad_norm": 1.0922549633005931, "learning_rate": 6.51860615427361e-06, "loss": 0.4417, "step": 17537 }, { "epoch": 0.6248285444537471, "grad_norm": 1.923685042633817, "learning_rate": 6.5175244278288705e-06, "loss": 0.5588, "step": 17538 }, { "epoch": 0.6248641715802412, "grad_norm": 1.4682095356989973, "learning_rate": 6.516442747755731e-06, "loss": 0.5127, "step": 17539 }, { "epoch": 0.6248997987067353, "grad_norm": 1.1073409992978738, "learning_rate": 6.5153611140685865e-06, "loss": 0.426, "step": 17540 }, { "epoch": 0.6249354258332295, "grad_norm": 1.5218161669496761, "learning_rate": 6.5142795267818505e-06, "loss": 0.4974, "step": 17541 }, { "epoch": 0.6249710529597235, "grad_norm": 1.2349182926680844, "learning_rate": 6.513197985909914e-06, "loss": 0.5302, "step": 17542 }, { "epoch": 0.6250066800862176, "grad_norm": 1.2340364757706035, "learning_rate": 6.512116491467184e-06, "loss": 0.4379, "step": 17543 }, { "epoch": 0.6250423072127118, "grad_norm": 1.23903757319165, "learning_rate": 6.511035043468064e-06, "loss": 0.4715, "step": 17544 }, { "epoch": 0.6250779343392059, "grad_norm": 1.1316164256493408, "learning_rate": 6.509953641926947e-06, "loss": 0.4296, "step": 17545 }, { "epoch": 0.6251135614657, "grad_norm": 1.3575381133554945, "learning_rate": 6.508872286858241e-06, "loss": 0.5241, "step": 17546 }, { "epoch": 0.6251491885921941, "grad_norm": 1.6305708413262594, "learning_rate": 6.507790978276337e-06, "loss": 0.5566, "step": 17547 }, { "epoch": 0.6251848157186882, "grad_norm": 1.4002908951664768, "learning_rate": 6.506709716195639e-06, "loss": 0.578, "step": 17548 }, { "epoch": 0.6252204428451823, "grad_norm": 1.7361657975445577, "learning_rate": 6.505628500630541e-06, "loss": 0.5586, "step": 17549 }, { "epoch": 0.6252560699716765, "grad_norm": 1.4126810284005467, "learning_rate": 6.5045473315954455e-06, "loss": 0.5519, "step": 17550 }, { "epoch": 0.6252916970981706, "grad_norm": 1.5328383269083785, "learning_rate": 6.503466209104743e-06, "loss": 0.4417, "step": 17551 }, { "epoch": 0.6253273242246646, "grad_norm": 1.1952263544232435, "learning_rate": 6.502385133172833e-06, "loss": 0.4734, "step": 17552 }, { "epoch": 0.6253629513511588, "grad_norm": 1.0847353406913363, "learning_rate": 6.5013041038141055e-06, "loss": 0.4469, "step": 17553 }, { "epoch": 0.6253985784776529, "grad_norm": 1.1126543685485653, "learning_rate": 6.5002231210429636e-06, "loss": 0.4195, "step": 17554 }, { "epoch": 0.625434205604147, "grad_norm": 1.4840892677245863, "learning_rate": 6.499142184873795e-06, "loss": 0.507, "step": 17555 }, { "epoch": 0.6254698327306412, "grad_norm": 1.1847665652386137, "learning_rate": 6.4980612953209946e-06, "loss": 0.482, "step": 17556 }, { "epoch": 0.6255054598571352, "grad_norm": 1.128470515255291, "learning_rate": 6.49698045239896e-06, "loss": 0.4117, "step": 17557 }, { "epoch": 0.6255410869836293, "grad_norm": 1.291168295769262, "learning_rate": 6.495899656122075e-06, "loss": 0.3841, "step": 17558 }, { "epoch": 0.6255767141101235, "grad_norm": 1.3805918393098422, "learning_rate": 6.494818906504737e-06, "loss": 0.675, "step": 17559 }, { "epoch": 0.6256123412366176, "grad_norm": 1.1604676558899696, "learning_rate": 6.493738203561333e-06, "loss": 0.4127, "step": 17560 }, { "epoch": 0.6256479683631117, "grad_norm": 1.233256834578595, "learning_rate": 6.49265754730626e-06, "loss": 0.4845, "step": 17561 }, { "epoch": 0.6256835954896058, "grad_norm": 1.1829065520501436, "learning_rate": 6.4915769377538995e-06, "loss": 0.4246, "step": 17562 }, { "epoch": 0.6257192226160999, "grad_norm": 1.1341140794099858, "learning_rate": 6.490496374918647e-06, "loss": 0.4655, "step": 17563 }, { "epoch": 0.625754849742594, "grad_norm": 1.7805993249064376, "learning_rate": 6.4894158588148844e-06, "loss": 0.5583, "step": 17564 }, { "epoch": 0.6257904768690882, "grad_norm": 1.0955535047822287, "learning_rate": 6.488335389457007e-06, "loss": 0.4328, "step": 17565 }, { "epoch": 0.6258261039955823, "grad_norm": 1.157388143297795, "learning_rate": 6.4872549668593954e-06, "loss": 0.423, "step": 17566 }, { "epoch": 0.6258617311220763, "grad_norm": 1.6543181361450179, "learning_rate": 6.486174591036439e-06, "loss": 0.5536, "step": 17567 }, { "epoch": 0.6258973582485705, "grad_norm": 1.2759845431123538, "learning_rate": 6.485094262002529e-06, "loss": 0.4483, "step": 17568 }, { "epoch": 0.6259329853750646, "grad_norm": 1.1024943532124645, "learning_rate": 6.484013979772041e-06, "loss": 0.3731, "step": 17569 }, { "epoch": 0.6259686125015587, "grad_norm": 2.3300072175298965, "learning_rate": 6.482933744359366e-06, "loss": 0.431, "step": 17570 }, { "epoch": 0.6260042396280528, "grad_norm": 1.1553805854137245, "learning_rate": 6.481853555778886e-06, "loss": 0.4272, "step": 17571 }, { "epoch": 0.6260398667545469, "grad_norm": 1.2182882785072005, "learning_rate": 6.480773414044988e-06, "loss": 0.4259, "step": 17572 }, { "epoch": 0.626075493881041, "grad_norm": 1.293898550163563, "learning_rate": 6.479693319172047e-06, "loss": 0.414, "step": 17573 }, { "epoch": 0.6261111210075352, "grad_norm": 1.2127417632785702, "learning_rate": 6.478613271174453e-06, "loss": 0.427, "step": 17574 }, { "epoch": 0.6261467481340293, "grad_norm": 1.1057917585412982, "learning_rate": 6.477533270066583e-06, "loss": 0.4113, "step": 17575 }, { "epoch": 0.6261823752605233, "grad_norm": 1.0295990809030926, "learning_rate": 6.476453315862823e-06, "loss": 0.448, "step": 17576 }, { "epoch": 0.6262180023870175, "grad_norm": 1.110155001019691, "learning_rate": 6.4753734085775435e-06, "loss": 0.4193, "step": 17577 }, { "epoch": 0.6262536295135116, "grad_norm": 1.131091890572411, "learning_rate": 6.474293548225132e-06, "loss": 0.3715, "step": 17578 }, { "epoch": 0.6262892566400057, "grad_norm": 2.6734200667995616, "learning_rate": 6.473213734819972e-06, "loss": 0.5839, "step": 17579 }, { "epoch": 0.6263248837664999, "grad_norm": 1.6168419028446188, "learning_rate": 6.4721339683764306e-06, "loss": 0.3982, "step": 17580 }, { "epoch": 0.6263605108929939, "grad_norm": 1.0751985094563024, "learning_rate": 6.471054248908895e-06, "loss": 0.431, "step": 17581 }, { "epoch": 0.626396138019488, "grad_norm": 1.5844004771802178, "learning_rate": 6.469974576431734e-06, "loss": 0.5178, "step": 17582 }, { "epoch": 0.6264317651459822, "grad_norm": 1.0361960843899372, "learning_rate": 6.468894950959336e-06, "loss": 0.3966, "step": 17583 }, { "epoch": 0.6264673922724763, "grad_norm": 1.7670198045982222, "learning_rate": 6.467815372506064e-06, "loss": 0.5204, "step": 17584 }, { "epoch": 0.6265030193989704, "grad_norm": 1.1113800280324797, "learning_rate": 6.466735841086303e-06, "loss": 0.4064, "step": 17585 }, { "epoch": 0.6265386465254645, "grad_norm": 1.6057508026124665, "learning_rate": 6.465656356714419e-06, "loss": 0.5195, "step": 17586 }, { "epoch": 0.6265742736519586, "grad_norm": 1.5601348468294978, "learning_rate": 6.464576919404797e-06, "loss": 0.5049, "step": 17587 }, { "epoch": 0.6266099007784527, "grad_norm": 2.5111401001461764, "learning_rate": 6.463497529171799e-06, "loss": 0.5484, "step": 17588 }, { "epoch": 0.6266455279049469, "grad_norm": 1.4574301710173299, "learning_rate": 6.462418186029802e-06, "loss": 0.5873, "step": 17589 }, { "epoch": 0.626681155031441, "grad_norm": 1.2448115654027698, "learning_rate": 6.461338889993185e-06, "loss": 0.4469, "step": 17590 }, { "epoch": 0.626716782157935, "grad_norm": 1.528818211560243, "learning_rate": 6.460259641076312e-06, "loss": 0.6226, "step": 17591 }, { "epoch": 0.6267524092844292, "grad_norm": 1.6677223513014574, "learning_rate": 6.459180439293555e-06, "loss": 0.518, "step": 17592 }, { "epoch": 0.6267880364109233, "grad_norm": 1.5779154210188269, "learning_rate": 6.4581012846592865e-06, "loss": 0.4985, "step": 17593 }, { "epoch": 0.6268236635374174, "grad_norm": 1.3794030500758419, "learning_rate": 6.4570221771878775e-06, "loss": 0.4891, "step": 17594 }, { "epoch": 0.6268592906639116, "grad_norm": 1.1518244949480576, "learning_rate": 6.4559431168936905e-06, "loss": 0.3899, "step": 17595 }, { "epoch": 0.6268949177904056, "grad_norm": 1.2260754872560475, "learning_rate": 6.4548641037911e-06, "loss": 0.3933, "step": 17596 }, { "epoch": 0.6269305449168997, "grad_norm": 1.112926486287939, "learning_rate": 6.453785137894469e-06, "loss": 0.4661, "step": 17597 }, { "epoch": 0.6269661720433939, "grad_norm": 1.1254451424921303, "learning_rate": 6.4527062192181735e-06, "loss": 0.3811, "step": 17598 }, { "epoch": 0.627001799169888, "grad_norm": 1.2992909910240429, "learning_rate": 6.451627347776569e-06, "loss": 0.457, "step": 17599 }, { "epoch": 0.627037426296382, "grad_norm": 1.1673393621184929, "learning_rate": 6.450548523584028e-06, "loss": 0.3762, "step": 17600 }, { "epoch": 0.6270730534228762, "grad_norm": 1.6688982982764342, "learning_rate": 6.449469746654913e-06, "loss": 0.5051, "step": 17601 }, { "epoch": 0.6271086805493703, "grad_norm": 1.140683394433019, "learning_rate": 6.4483910170035905e-06, "loss": 0.446, "step": 17602 }, { "epoch": 0.6271443076758644, "grad_norm": 1.1757761595758127, "learning_rate": 6.447312334644425e-06, "loss": 0.4529, "step": 17603 }, { "epoch": 0.6271799348023586, "grad_norm": 1.1564730434755908, "learning_rate": 6.446233699591778e-06, "loss": 0.4387, "step": 17604 }, { "epoch": 0.6272155619288526, "grad_norm": 1.1490947669051828, "learning_rate": 6.445155111860016e-06, "loss": 0.4533, "step": 17605 }, { "epoch": 0.6272511890553467, "grad_norm": 1.486085341942529, "learning_rate": 6.444076571463495e-06, "loss": 0.6024, "step": 17606 }, { "epoch": 0.6272868161818409, "grad_norm": 1.2035209630656392, "learning_rate": 6.442998078416583e-06, "loss": 0.3993, "step": 17607 }, { "epoch": 0.627322443308335, "grad_norm": 1.1602430508423693, "learning_rate": 6.441919632733634e-06, "loss": 0.4598, "step": 17608 }, { "epoch": 0.627358070434829, "grad_norm": 1.1846822888382327, "learning_rate": 6.440841234429018e-06, "loss": 0.418, "step": 17609 }, { "epoch": 0.6273936975613232, "grad_norm": 1.061829902289941, "learning_rate": 6.439762883517082e-06, "loss": 0.4073, "step": 17610 }, { "epoch": 0.6274293246878173, "grad_norm": 1.3399710323742653, "learning_rate": 6.438684580012196e-06, "loss": 0.6003, "step": 17611 }, { "epoch": 0.6274649518143114, "grad_norm": 1.0711798020563355, "learning_rate": 6.437606323928711e-06, "loss": 0.431, "step": 17612 }, { "epoch": 0.6275005789408056, "grad_norm": 1.3827173510654585, "learning_rate": 6.436528115280988e-06, "loss": 0.5887, "step": 17613 }, { "epoch": 0.6275362060672997, "grad_norm": 1.0403707012200911, "learning_rate": 6.435449954083388e-06, "loss": 0.4187, "step": 17614 }, { "epoch": 0.6275718331937937, "grad_norm": 1.0984894513757875, "learning_rate": 6.434371840350257e-06, "loss": 0.3814, "step": 17615 }, { "epoch": 0.6276074603202879, "grad_norm": 1.077048224046782, "learning_rate": 6.433293774095963e-06, "loss": 0.4332, "step": 17616 }, { "epoch": 0.627643087446782, "grad_norm": 1.31500020781373, "learning_rate": 6.432215755334852e-06, "loss": 0.4432, "step": 17617 }, { "epoch": 0.6276787145732761, "grad_norm": 1.533124310829434, "learning_rate": 6.431137784081283e-06, "loss": 0.5394, "step": 17618 }, { "epoch": 0.6277143416997703, "grad_norm": 1.5674852165265816, "learning_rate": 6.430059860349607e-06, "loss": 0.4731, "step": 17619 }, { "epoch": 0.6277499688262643, "grad_norm": 1.190245956720443, "learning_rate": 6.428981984154183e-06, "loss": 0.3892, "step": 17620 }, { "epoch": 0.6277855959527584, "grad_norm": 1.2306920552608256, "learning_rate": 6.427904155509357e-06, "loss": 0.4632, "step": 17621 }, { "epoch": 0.6278212230792526, "grad_norm": 1.7919983579067362, "learning_rate": 6.4268263744294866e-06, "loss": 0.5037, "step": 17622 }, { "epoch": 0.6278568502057467, "grad_norm": 2.4344725402856295, "learning_rate": 6.425748640928917e-06, "loss": 0.5402, "step": 17623 }, { "epoch": 0.6278924773322407, "grad_norm": 1.4458817632223786, "learning_rate": 6.424670955022003e-06, "loss": 0.4328, "step": 17624 }, { "epoch": 0.6279281044587349, "grad_norm": 1.7629068275477258, "learning_rate": 6.423593316723098e-06, "loss": 0.5097, "step": 17625 }, { "epoch": 0.627963731585229, "grad_norm": 1.2671801195614687, "learning_rate": 6.422515726046543e-06, "loss": 0.4928, "step": 17626 }, { "epoch": 0.6279993587117231, "grad_norm": 1.0957921815313938, "learning_rate": 6.421438183006695e-06, "loss": 0.3929, "step": 17627 }, { "epoch": 0.6280349858382173, "grad_norm": 1.105095172796409, "learning_rate": 6.420360687617897e-06, "loss": 0.4366, "step": 17628 }, { "epoch": 0.6280706129647113, "grad_norm": 1.0907246640966657, "learning_rate": 6.419283239894499e-06, "loss": 0.4307, "step": 17629 }, { "epoch": 0.6281062400912054, "grad_norm": 1.5603509487267126, "learning_rate": 6.418205839850847e-06, "loss": 0.5855, "step": 17630 }, { "epoch": 0.6281418672176996, "grad_norm": 1.2272775909975915, "learning_rate": 6.417128487501291e-06, "loss": 0.4005, "step": 17631 }, { "epoch": 0.6281774943441937, "grad_norm": 1.5538384903776594, "learning_rate": 6.416051182860169e-06, "loss": 0.5556, "step": 17632 }, { "epoch": 0.6282131214706878, "grad_norm": 1.1670076827723692, "learning_rate": 6.4149739259418335e-06, "loss": 0.4668, "step": 17633 }, { "epoch": 0.6282487485971819, "grad_norm": 1.0993705865260195, "learning_rate": 6.413896716760624e-06, "loss": 0.3912, "step": 17634 }, { "epoch": 0.628284375723676, "grad_norm": 1.3592167261015171, "learning_rate": 6.412819555330889e-06, "loss": 0.4519, "step": 17635 }, { "epoch": 0.6283200028501701, "grad_norm": 1.781379580902378, "learning_rate": 6.411742441666966e-06, "loss": 0.5403, "step": 17636 }, { "epoch": 0.6283556299766643, "grad_norm": 1.2174486072024782, "learning_rate": 6.4106653757831994e-06, "loss": 0.4594, "step": 17637 }, { "epoch": 0.6283912571031584, "grad_norm": 1.328063819317792, "learning_rate": 6.409588357693935e-06, "loss": 0.4908, "step": 17638 }, { "epoch": 0.6284268842296524, "grad_norm": 1.1355325453406813, "learning_rate": 6.408511387413508e-06, "loss": 0.4543, "step": 17639 }, { "epoch": 0.6284625113561466, "grad_norm": 1.316292112404342, "learning_rate": 6.407434464956266e-06, "loss": 0.5271, "step": 17640 }, { "epoch": 0.6284981384826407, "grad_norm": 1.065548667674041, "learning_rate": 6.406357590336543e-06, "loss": 0.4384, "step": 17641 }, { "epoch": 0.6285337656091348, "grad_norm": 1.112002470650606, "learning_rate": 6.405280763568683e-06, "loss": 0.3908, "step": 17642 }, { "epoch": 0.628569392735629, "grad_norm": 1.1857582528846509, "learning_rate": 6.404203984667019e-06, "loss": 0.4262, "step": 17643 }, { "epoch": 0.628605019862123, "grad_norm": 1.7300930414920348, "learning_rate": 6.403127253645893e-06, "loss": 0.5058, "step": 17644 }, { "epoch": 0.6286406469886171, "grad_norm": 1.6909933684983198, "learning_rate": 6.402050570519642e-06, "loss": 0.5283, "step": 17645 }, { "epoch": 0.6286762741151113, "grad_norm": 1.3033644001493998, "learning_rate": 6.400973935302605e-06, "loss": 0.499, "step": 17646 }, { "epoch": 0.6287119012416054, "grad_norm": 1.0979642267842193, "learning_rate": 6.399897348009112e-06, "loss": 0.3949, "step": 17647 }, { "epoch": 0.6287475283680994, "grad_norm": 1.1370561951438465, "learning_rate": 6.398820808653503e-06, "loss": 0.4535, "step": 17648 }, { "epoch": 0.6287831554945936, "grad_norm": 1.6128120118937677, "learning_rate": 6.397744317250112e-06, "loss": 0.5514, "step": 17649 }, { "epoch": 0.6288187826210877, "grad_norm": 1.4955531206938664, "learning_rate": 6.396667873813273e-06, "loss": 0.487, "step": 17650 }, { "epoch": 0.6288544097475818, "grad_norm": 1.603876945469275, "learning_rate": 6.395591478357323e-06, "loss": 0.4463, "step": 17651 }, { "epoch": 0.628890036874076, "grad_norm": 1.7582026951012928, "learning_rate": 6.394515130896586e-06, "loss": 0.5836, "step": 17652 }, { "epoch": 0.62892566400057, "grad_norm": 1.0407502972419431, "learning_rate": 6.393438831445407e-06, "loss": 0.3855, "step": 17653 }, { "epoch": 0.6289612911270641, "grad_norm": 1.6625284631175294, "learning_rate": 6.3923625800181055e-06, "loss": 0.6072, "step": 17654 }, { "epoch": 0.6289969182535583, "grad_norm": 1.853375202819413, "learning_rate": 6.39128637662902e-06, "loss": 0.6115, "step": 17655 }, { "epoch": 0.6290325453800524, "grad_norm": 1.3322541437615727, "learning_rate": 6.390210221292476e-06, "loss": 0.5576, "step": 17656 }, { "epoch": 0.6290681725065465, "grad_norm": 1.1285231053353195, "learning_rate": 6.3891341140228125e-06, "loss": 0.3927, "step": 17657 }, { "epoch": 0.6291037996330406, "grad_norm": 1.3099567283830598, "learning_rate": 6.388058054834345e-06, "loss": 0.4671, "step": 17658 }, { "epoch": 0.6291394267595347, "grad_norm": 1.4977746319086607, "learning_rate": 6.386982043741412e-06, "loss": 0.571, "step": 17659 }, { "epoch": 0.6291750538860288, "grad_norm": 1.2186138010770524, "learning_rate": 6.385906080758339e-06, "loss": 0.6046, "step": 17660 }, { "epoch": 0.629210681012523, "grad_norm": 1.0410112412298613, "learning_rate": 6.38483016589945e-06, "loss": 0.4145, "step": 17661 }, { "epoch": 0.629246308139017, "grad_norm": 1.457184462159132, "learning_rate": 6.383754299179079e-06, "loss": 0.4913, "step": 17662 }, { "epoch": 0.6292819352655111, "grad_norm": 1.368806493235888, "learning_rate": 6.382678480611544e-06, "loss": 0.5997, "step": 17663 }, { "epoch": 0.6293175623920053, "grad_norm": 1.3354269527311255, "learning_rate": 6.3816027102111745e-06, "loss": 0.5348, "step": 17664 }, { "epoch": 0.6293531895184994, "grad_norm": 1.3439847101726687, "learning_rate": 6.380526987992291e-06, "loss": 0.5202, "step": 17665 }, { "epoch": 0.6293888166449935, "grad_norm": 1.3733725153884981, "learning_rate": 6.379451313969224e-06, "loss": 0.4922, "step": 17666 }, { "epoch": 0.6294244437714877, "grad_norm": 1.0742575885200116, "learning_rate": 6.378375688156291e-06, "loss": 0.3816, "step": 17667 }, { "epoch": 0.6294600708979817, "grad_norm": 1.1056269219742685, "learning_rate": 6.377300110567821e-06, "loss": 0.4362, "step": 17668 }, { "epoch": 0.6294956980244758, "grad_norm": 1.2256816904237968, "learning_rate": 6.376224581218128e-06, "loss": 0.4234, "step": 17669 }, { "epoch": 0.62953132515097, "grad_norm": 1.620458958608428, "learning_rate": 6.375149100121537e-06, "loss": 0.5565, "step": 17670 }, { "epoch": 0.6295669522774641, "grad_norm": 1.4132341428138826, "learning_rate": 6.374073667292373e-06, "loss": 0.4708, "step": 17671 }, { "epoch": 0.6296025794039581, "grad_norm": 1.4952774507269997, "learning_rate": 6.372998282744949e-06, "loss": 0.4942, "step": 17672 }, { "epoch": 0.6296382065304523, "grad_norm": 1.1183120310263226, "learning_rate": 6.3719229464935915e-06, "loss": 0.4008, "step": 17673 }, { "epoch": 0.6296738336569464, "grad_norm": 1.0377356257966932, "learning_rate": 6.3708476585526135e-06, "loss": 0.3904, "step": 17674 }, { "epoch": 0.6297094607834405, "grad_norm": 1.2568365144015847, "learning_rate": 6.369772418936337e-06, "loss": 0.4438, "step": 17675 }, { "epoch": 0.6297450879099347, "grad_norm": 1.3877162590752514, "learning_rate": 6.368697227659075e-06, "loss": 0.4608, "step": 17676 }, { "epoch": 0.6297807150364287, "grad_norm": 1.0936573683431494, "learning_rate": 6.367622084735153e-06, "loss": 0.4351, "step": 17677 }, { "epoch": 0.6298163421629228, "grad_norm": 1.1428129948248797, "learning_rate": 6.366546990178876e-06, "loss": 0.4149, "step": 17678 }, { "epoch": 0.629851969289417, "grad_norm": 1.351124579233286, "learning_rate": 6.365471944004571e-06, "loss": 0.5453, "step": 17679 }, { "epoch": 0.6298875964159111, "grad_norm": 1.3564085446969627, "learning_rate": 6.364396946226543e-06, "loss": 0.4878, "step": 17680 }, { "epoch": 0.6299232235424052, "grad_norm": 1.694931429678805, "learning_rate": 6.363321996859113e-06, "loss": 0.4379, "step": 17681 }, { "epoch": 0.6299588506688993, "grad_norm": 1.100187653569028, "learning_rate": 6.3622470959165905e-06, "loss": 0.3951, "step": 17682 }, { "epoch": 0.6299944777953934, "grad_norm": 1.5668335115104544, "learning_rate": 6.361172243413291e-06, "loss": 0.503, "step": 17683 }, { "epoch": 0.6300301049218875, "grad_norm": 1.1679801429568346, "learning_rate": 6.36009743936353e-06, "loss": 0.4118, "step": 17684 }, { "epoch": 0.6300657320483817, "grad_norm": 1.623374868539265, "learning_rate": 6.359022683781611e-06, "loss": 0.5279, "step": 17685 }, { "epoch": 0.6301013591748758, "grad_norm": 1.0929206616537508, "learning_rate": 6.357947976681853e-06, "loss": 0.4408, "step": 17686 }, { "epoch": 0.6301369863013698, "grad_norm": 1.1327309092815996, "learning_rate": 6.35687331807856e-06, "loss": 0.4227, "step": 17687 }, { "epoch": 0.630172613427864, "grad_norm": 1.32226159373601, "learning_rate": 6.355798707986052e-06, "loss": 0.512, "step": 17688 }, { "epoch": 0.6302082405543581, "grad_norm": 1.0722466305496654, "learning_rate": 6.3547241464186245e-06, "loss": 0.4102, "step": 17689 }, { "epoch": 0.6302438676808522, "grad_norm": 1.3797783745049839, "learning_rate": 6.353649633390596e-06, "loss": 0.543, "step": 17690 }, { "epoch": 0.6302794948073464, "grad_norm": 1.462659741243072, "learning_rate": 6.352575168916268e-06, "loss": 0.4213, "step": 17691 }, { "epoch": 0.6303151219338404, "grad_norm": 1.4132171648465466, "learning_rate": 6.351500753009956e-06, "loss": 0.5246, "step": 17692 }, { "epoch": 0.6303507490603345, "grad_norm": 1.1290994911406549, "learning_rate": 6.3504263856859575e-06, "loss": 0.437, "step": 17693 }, { "epoch": 0.6303863761868287, "grad_norm": 1.1315667123901614, "learning_rate": 6.349352066958583e-06, "loss": 0.409, "step": 17694 }, { "epoch": 0.6304220033133228, "grad_norm": 1.223777758701639, "learning_rate": 6.3482777968421415e-06, "loss": 0.4467, "step": 17695 }, { "epoch": 0.6304576304398168, "grad_norm": 1.2520772270185272, "learning_rate": 6.347203575350929e-06, "loss": 0.4743, "step": 17696 }, { "epoch": 0.630493257566311, "grad_norm": 1.1539882225875664, "learning_rate": 6.346129402499256e-06, "loss": 0.4051, "step": 17697 }, { "epoch": 0.6305288846928051, "grad_norm": 1.5262098741854135, "learning_rate": 6.345055278301422e-06, "loss": 0.5497, "step": 17698 }, { "epoch": 0.6305645118192992, "grad_norm": 1.2624027468181305, "learning_rate": 6.343981202771736e-06, "loss": 0.4354, "step": 17699 }, { "epoch": 0.6306001389457934, "grad_norm": 1.2678395110856444, "learning_rate": 6.342907175924492e-06, "loss": 0.4337, "step": 17700 }, { "epoch": 0.6306357660722874, "grad_norm": 1.3024799284385826, "learning_rate": 6.341833197773997e-06, "loss": 0.5295, "step": 17701 }, { "epoch": 0.6306713931987815, "grad_norm": 1.2864929911794607, "learning_rate": 6.340759268334546e-06, "loss": 0.4532, "step": 17702 }, { "epoch": 0.6307070203252757, "grad_norm": 1.4710118529388496, "learning_rate": 6.339685387620449e-06, "loss": 0.5321, "step": 17703 }, { "epoch": 0.6307426474517698, "grad_norm": 1.764413121298423, "learning_rate": 6.338611555645993e-06, "loss": 0.5649, "step": 17704 }, { "epoch": 0.6307782745782639, "grad_norm": 1.1314624612078112, "learning_rate": 6.337537772425486e-06, "loss": 0.4104, "step": 17705 }, { "epoch": 0.630813901704758, "grad_norm": 1.3968883165843935, "learning_rate": 6.3364640379732266e-06, "loss": 0.604, "step": 17706 }, { "epoch": 0.6308495288312521, "grad_norm": 1.2373672235571231, "learning_rate": 6.3353903523035056e-06, "loss": 0.475, "step": 17707 }, { "epoch": 0.6308851559577462, "grad_norm": 1.5483865603988363, "learning_rate": 6.334316715430625e-06, "loss": 0.5173, "step": 17708 }, { "epoch": 0.6309207830842404, "grad_norm": 1.420147907925028, "learning_rate": 6.333243127368878e-06, "loss": 0.5545, "step": 17709 }, { "epoch": 0.6309564102107345, "grad_norm": 1.5426221093795867, "learning_rate": 6.332169588132565e-06, "loss": 0.4856, "step": 17710 }, { "epoch": 0.6309920373372285, "grad_norm": 1.4963545039834723, "learning_rate": 6.331096097735973e-06, "loss": 0.4823, "step": 17711 }, { "epoch": 0.6310276644637227, "grad_norm": 1.4044350733327238, "learning_rate": 6.330022656193405e-06, "loss": 0.5619, "step": 17712 }, { "epoch": 0.6310632915902168, "grad_norm": 1.1672375814483147, "learning_rate": 6.3289492635191475e-06, "loss": 0.4706, "step": 17713 }, { "epoch": 0.6310989187167109, "grad_norm": 1.1208635117263936, "learning_rate": 6.3278759197275e-06, "loss": 0.4256, "step": 17714 }, { "epoch": 0.631134545843205, "grad_norm": 1.2976998590875042, "learning_rate": 6.326802624832749e-06, "loss": 0.564, "step": 17715 }, { "epoch": 0.6311701729696991, "grad_norm": 1.240377957992138, "learning_rate": 6.3257293788491835e-06, "loss": 0.4387, "step": 17716 }, { "epoch": 0.6312058000961932, "grad_norm": 1.0828884082384893, "learning_rate": 6.3246561817911076e-06, "loss": 0.4275, "step": 17717 }, { "epoch": 0.6312414272226874, "grad_norm": 1.4685343730506646, "learning_rate": 6.323583033672799e-06, "loss": 0.5461, "step": 17718 }, { "epoch": 0.6312770543491815, "grad_norm": 1.7087548272791504, "learning_rate": 6.322509934508554e-06, "loss": 0.59, "step": 17719 }, { "epoch": 0.6313126814756755, "grad_norm": 1.8679904860229415, "learning_rate": 6.321436884312657e-06, "loss": 0.4716, "step": 17720 }, { "epoch": 0.6313483086021697, "grad_norm": 1.2450828828046576, "learning_rate": 6.3203638830994044e-06, "loss": 0.4096, "step": 17721 }, { "epoch": 0.6313839357286638, "grad_norm": 1.1154933867363628, "learning_rate": 6.319290930883073e-06, "loss": 0.4202, "step": 17722 }, { "epoch": 0.6314195628551579, "grad_norm": 1.1513857879883207, "learning_rate": 6.3182180276779594e-06, "loss": 0.3921, "step": 17723 }, { "epoch": 0.6314551899816521, "grad_norm": 1.897936107639369, "learning_rate": 6.317145173498343e-06, "loss": 0.5245, "step": 17724 }, { "epoch": 0.6314908171081461, "grad_norm": 1.7648319997005921, "learning_rate": 6.316072368358517e-06, "loss": 0.5917, "step": 17725 }, { "epoch": 0.6315264442346402, "grad_norm": 1.55062151879568, "learning_rate": 6.3149996122727585e-06, "loss": 0.5441, "step": 17726 }, { "epoch": 0.6315620713611344, "grad_norm": 1.5847889393779793, "learning_rate": 6.313926905255358e-06, "loss": 0.501, "step": 17727 }, { "epoch": 0.6315976984876285, "grad_norm": 1.6072994543569301, "learning_rate": 6.312854247320594e-06, "loss": 0.5943, "step": 17728 }, { "epoch": 0.6316333256141226, "grad_norm": 1.52766997815188, "learning_rate": 6.311781638482754e-06, "loss": 0.4322, "step": 17729 }, { "epoch": 0.6316689527406167, "grad_norm": 1.4331106303377812, "learning_rate": 6.3107090787561206e-06, "loss": 0.5084, "step": 17730 }, { "epoch": 0.6317045798671108, "grad_norm": 1.573490007283394, "learning_rate": 6.309636568154973e-06, "loss": 0.577, "step": 17731 }, { "epoch": 0.6317402069936049, "grad_norm": 1.6652011503824542, "learning_rate": 6.308564106693597e-06, "loss": 0.6061, "step": 17732 }, { "epoch": 0.6317758341200991, "grad_norm": 1.212557369858097, "learning_rate": 6.307491694386267e-06, "loss": 0.4727, "step": 17733 }, { "epoch": 0.6318114612465932, "grad_norm": 1.0940462447180197, "learning_rate": 6.306419331247269e-06, "loss": 0.4376, "step": 17734 }, { "epoch": 0.6318470883730872, "grad_norm": 1.3634715817575256, "learning_rate": 6.305347017290875e-06, "loss": 0.4211, "step": 17735 }, { "epoch": 0.6318827154995814, "grad_norm": 1.157949588345842, "learning_rate": 6.304274752531374e-06, "loss": 0.427, "step": 17736 }, { "epoch": 0.6319183426260755, "grad_norm": 1.1487259081247951, "learning_rate": 6.303202536983033e-06, "loss": 0.4483, "step": 17737 }, { "epoch": 0.6319539697525696, "grad_norm": 1.3328321183839926, "learning_rate": 6.302130370660135e-06, "loss": 0.43, "step": 17738 }, { "epoch": 0.6319895968790638, "grad_norm": 1.6101955159479449, "learning_rate": 6.3010582535769546e-06, "loss": 0.5027, "step": 17739 }, { "epoch": 0.6320252240055578, "grad_norm": 1.1639887320386717, "learning_rate": 6.299986185747769e-06, "loss": 0.4415, "step": 17740 }, { "epoch": 0.6320608511320519, "grad_norm": 1.9484945288182707, "learning_rate": 6.298914167186856e-06, "loss": 0.5138, "step": 17741 }, { "epoch": 0.6320964782585461, "grad_norm": 1.194912080045596, "learning_rate": 6.297842197908486e-06, "loss": 0.4309, "step": 17742 }, { "epoch": 0.6321321053850402, "grad_norm": 1.2002706005471724, "learning_rate": 6.296770277926938e-06, "loss": 0.4799, "step": 17743 }, { "epoch": 0.6321677325115342, "grad_norm": 1.0978264328163048, "learning_rate": 6.295698407256479e-06, "loss": 0.4153, "step": 17744 }, { "epoch": 0.6322033596380284, "grad_norm": 1.4256148276297504, "learning_rate": 6.2946265859113855e-06, "loss": 0.5405, "step": 17745 }, { "epoch": 0.6322389867645225, "grad_norm": 1.8385702667779875, "learning_rate": 6.293554813905927e-06, "loss": 0.4271, "step": 17746 }, { "epoch": 0.6322746138910166, "grad_norm": 0.9858344444349306, "learning_rate": 6.292483091254382e-06, "loss": 0.4035, "step": 17747 }, { "epoch": 0.6323102410175108, "grad_norm": 3.1554253758018187, "learning_rate": 6.291411417971012e-06, "loss": 0.5901, "step": 17748 }, { "epoch": 0.6323458681440048, "grad_norm": 1.0155966981855633, "learning_rate": 6.290339794070092e-06, "loss": 0.4228, "step": 17749 }, { "epoch": 0.6323814952704989, "grad_norm": 1.335894283727091, "learning_rate": 6.289268219565889e-06, "loss": 0.5126, "step": 17750 }, { "epoch": 0.6324171223969931, "grad_norm": 2.0867935323992293, "learning_rate": 6.288196694472673e-06, "loss": 0.5968, "step": 17751 }, { "epoch": 0.6324527495234872, "grad_norm": 1.1513564100654383, "learning_rate": 6.287125218804717e-06, "loss": 0.4347, "step": 17752 }, { "epoch": 0.6324883766499813, "grad_norm": 1.2171936703565343, "learning_rate": 6.286053792576276e-06, "loss": 0.4118, "step": 17753 }, { "epoch": 0.6325240037764754, "grad_norm": 1.2423749515519102, "learning_rate": 6.284982415801634e-06, "loss": 0.4279, "step": 17754 }, { "epoch": 0.6325596309029695, "grad_norm": 1.1642667865323812, "learning_rate": 6.283911088495041e-06, "loss": 0.4993, "step": 17755 }, { "epoch": 0.6325952580294636, "grad_norm": 1.2380063107245762, "learning_rate": 6.2828398106707734e-06, "loss": 0.4734, "step": 17756 }, { "epoch": 0.6326308851559578, "grad_norm": 1.0827685167454069, "learning_rate": 6.2817685823430885e-06, "loss": 0.3844, "step": 17757 }, { "epoch": 0.6326665122824519, "grad_norm": 1.291130793125896, "learning_rate": 6.280697403526258e-06, "loss": 0.427, "step": 17758 }, { "epoch": 0.6327021394089459, "grad_norm": 1.113764123383938, "learning_rate": 6.279626274234537e-06, "loss": 0.43, "step": 17759 }, { "epoch": 0.6327377665354401, "grad_norm": 1.3843118950557187, "learning_rate": 6.278555194482194e-06, "loss": 0.5069, "step": 17760 }, { "epoch": 0.6327733936619342, "grad_norm": 1.3375156377470665, "learning_rate": 6.277484164283489e-06, "loss": 0.5869, "step": 17761 }, { "epoch": 0.6328090207884283, "grad_norm": 1.1447191838145137, "learning_rate": 6.2764131836526844e-06, "loss": 0.3993, "step": 17762 }, { "epoch": 0.6328446479149225, "grad_norm": 1.5032024978847855, "learning_rate": 6.275342252604044e-06, "loss": 0.5453, "step": 17763 }, { "epoch": 0.6328802750414165, "grad_norm": 1.3988695354716996, "learning_rate": 6.274271371151821e-06, "loss": 0.5021, "step": 17764 }, { "epoch": 0.6329159021679106, "grad_norm": 1.0295211331996819, "learning_rate": 6.273200539310282e-06, "loss": 0.4593, "step": 17765 }, { "epoch": 0.6329515292944048, "grad_norm": 1.110716578306811, "learning_rate": 6.272129757093679e-06, "loss": 0.427, "step": 17766 }, { "epoch": 0.6329871564208989, "grad_norm": 1.5248817730732647, "learning_rate": 6.271059024516277e-06, "loss": 0.5465, "step": 17767 }, { "epoch": 0.6330227835473929, "grad_norm": 1.1748587778016364, "learning_rate": 6.269988341592328e-06, "loss": 0.4362, "step": 17768 }, { "epoch": 0.6330584106738871, "grad_norm": 2.348154990970128, "learning_rate": 6.2689177083360956e-06, "loss": 0.5094, "step": 17769 }, { "epoch": 0.6330940378003812, "grad_norm": 1.7319128748165942, "learning_rate": 6.267847124761828e-06, "loss": 0.5733, "step": 17770 }, { "epoch": 0.6331296649268753, "grad_norm": 1.0695822068064582, "learning_rate": 6.2667765908837856e-06, "loss": 0.4475, "step": 17771 }, { "epoch": 0.6331652920533695, "grad_norm": 1.056959013026191, "learning_rate": 6.2657061067162205e-06, "loss": 0.4566, "step": 17772 }, { "epoch": 0.6332009191798635, "grad_norm": 1.2345814094217196, "learning_rate": 6.264635672273393e-06, "loss": 0.394, "step": 17773 }, { "epoch": 0.6332365463063576, "grad_norm": 1.3044155490228326, "learning_rate": 6.263565287569549e-06, "loss": 0.4959, "step": 17774 }, { "epoch": 0.6332721734328518, "grad_norm": 1.4548590714132306, "learning_rate": 6.262494952618944e-06, "loss": 0.4679, "step": 17775 }, { "epoch": 0.6333078005593459, "grad_norm": 1.5054968847385324, "learning_rate": 6.261424667435833e-06, "loss": 0.5234, "step": 17776 }, { "epoch": 0.63334342768584, "grad_norm": 1.1359360084266465, "learning_rate": 6.260354432034463e-06, "loss": 0.4389, "step": 17777 }, { "epoch": 0.6333790548123341, "grad_norm": 1.0524918280418742, "learning_rate": 6.2592842464290935e-06, "loss": 0.3839, "step": 17778 }, { "epoch": 0.6334146819388282, "grad_norm": 1.1951547563347455, "learning_rate": 6.258214110633962e-06, "loss": 0.5332, "step": 17779 }, { "epoch": 0.6334503090653223, "grad_norm": 1.1476126808969576, "learning_rate": 6.257144024663331e-06, "loss": 0.4491, "step": 17780 }, { "epoch": 0.6334859361918165, "grad_norm": 1.5859053284542415, "learning_rate": 6.256073988531439e-06, "loss": 0.5431, "step": 17781 }, { "epoch": 0.6335215633183106, "grad_norm": 1.296391368956908, "learning_rate": 6.2550040022525405e-06, "loss": 0.4351, "step": 17782 }, { "epoch": 0.6335571904448046, "grad_norm": 1.0872705612808442, "learning_rate": 6.25393406584088e-06, "loss": 0.3984, "step": 17783 }, { "epoch": 0.6335928175712988, "grad_norm": 4.090819313827805, "learning_rate": 6.252864179310709e-06, "loss": 0.5983, "step": 17784 }, { "epoch": 0.6336284446977929, "grad_norm": 1.6205115132943024, "learning_rate": 6.251794342676267e-06, "loss": 0.5966, "step": 17785 }, { "epoch": 0.633664071824287, "grad_norm": 1.3424392789896575, "learning_rate": 6.250724555951804e-06, "loss": 0.573, "step": 17786 }, { "epoch": 0.6336996989507812, "grad_norm": 1.2956701273606517, "learning_rate": 6.249654819151565e-06, "loss": 0.4813, "step": 17787 }, { "epoch": 0.6337353260772752, "grad_norm": 1.1417041697643362, "learning_rate": 6.248585132289792e-06, "loss": 0.4076, "step": 17788 }, { "epoch": 0.6337709532037693, "grad_norm": 1.1887035000818602, "learning_rate": 6.2475154953807336e-06, "loss": 0.4297, "step": 17789 }, { "epoch": 0.6338065803302635, "grad_norm": 1.1162770705748988, "learning_rate": 6.246445908438625e-06, "loss": 0.4231, "step": 17790 }, { "epoch": 0.6338422074567576, "grad_norm": 1.3876160549859446, "learning_rate": 6.245376371477716e-06, "loss": 0.5306, "step": 17791 }, { "epoch": 0.6338778345832516, "grad_norm": 1.1023571309475124, "learning_rate": 6.244306884512241e-06, "loss": 0.3968, "step": 17792 }, { "epoch": 0.6339134617097458, "grad_norm": 1.1221862640276041, "learning_rate": 6.24323744755645e-06, "loss": 0.4045, "step": 17793 }, { "epoch": 0.6339490888362399, "grad_norm": 1.0891812849868563, "learning_rate": 6.242168060624572e-06, "loss": 0.3716, "step": 17794 }, { "epoch": 0.633984715962734, "grad_norm": 3.8385974638363694, "learning_rate": 6.241098723730861e-06, "loss": 0.5244, "step": 17795 }, { "epoch": 0.6340203430892282, "grad_norm": 1.3853984523406329, "learning_rate": 6.2400294368895405e-06, "loss": 0.5625, "step": 17796 }, { "epoch": 0.6340559702157222, "grad_norm": 1.458655394876462, "learning_rate": 6.238960200114857e-06, "loss": 0.4647, "step": 17797 }, { "epoch": 0.6340915973422163, "grad_norm": 1.011235051725914, "learning_rate": 6.23789101342105e-06, "loss": 0.4007, "step": 17798 }, { "epoch": 0.6341272244687105, "grad_norm": 1.1847286629342773, "learning_rate": 6.236821876822352e-06, "loss": 0.412, "step": 17799 }, { "epoch": 0.6341628515952046, "grad_norm": 1.39430304219253, "learning_rate": 6.235752790333004e-06, "loss": 0.5061, "step": 17800 }, { "epoch": 0.6341984787216987, "grad_norm": 1.512126142591589, "learning_rate": 6.234683753967234e-06, "loss": 0.5867, "step": 17801 }, { "epoch": 0.6342341058481928, "grad_norm": 1.1707013110911266, "learning_rate": 6.233614767739287e-06, "loss": 0.4202, "step": 17802 }, { "epoch": 0.6342697329746869, "grad_norm": 1.0977662464054505, "learning_rate": 6.232545831663386e-06, "loss": 0.4703, "step": 17803 }, { "epoch": 0.634305360101181, "grad_norm": 2.0089828057742216, "learning_rate": 6.231476945753776e-06, "loss": 0.5084, "step": 17804 }, { "epoch": 0.6343409872276752, "grad_norm": 1.233966234472493, "learning_rate": 6.23040811002468e-06, "loss": 0.4319, "step": 17805 }, { "epoch": 0.6343766143541693, "grad_norm": 1.2007746931403267, "learning_rate": 6.229339324490339e-06, "loss": 0.4316, "step": 17806 }, { "epoch": 0.6344122414806633, "grad_norm": 1.4962282827305695, "learning_rate": 6.228270589164977e-06, "loss": 0.5234, "step": 17807 }, { "epoch": 0.6344478686071575, "grad_norm": 1.8427264099293743, "learning_rate": 6.22720190406283e-06, "loss": 0.6011, "step": 17808 }, { "epoch": 0.6344834957336516, "grad_norm": 1.653770816654628, "learning_rate": 6.226133269198124e-06, "loss": 0.5436, "step": 17809 }, { "epoch": 0.6345191228601457, "grad_norm": 1.3944899644748783, "learning_rate": 6.225064684585094e-06, "loss": 0.5283, "step": 17810 }, { "epoch": 0.6345547499866399, "grad_norm": 1.549934202109522, "learning_rate": 6.223996150237968e-06, "loss": 0.5037, "step": 17811 }, { "epoch": 0.6345903771131339, "grad_norm": 1.1782241415290056, "learning_rate": 6.222927666170968e-06, "loss": 0.4243, "step": 17812 }, { "epoch": 0.634626004239628, "grad_norm": 1.4332770629534555, "learning_rate": 6.22185923239833e-06, "loss": 0.578, "step": 17813 }, { "epoch": 0.6346616313661222, "grad_norm": 1.1515668865313726, "learning_rate": 6.220790848934274e-06, "loss": 0.4078, "step": 17814 }, { "epoch": 0.6346972584926163, "grad_norm": 1.137007550524521, "learning_rate": 6.219722515793034e-06, "loss": 0.3994, "step": 17815 }, { "epoch": 0.6347328856191103, "grad_norm": 3.3997788889002263, "learning_rate": 6.218654232988823e-06, "loss": 0.5856, "step": 17816 }, { "epoch": 0.6347685127456045, "grad_norm": 1.2455268909907924, "learning_rate": 6.217586000535883e-06, "loss": 0.4293, "step": 17817 }, { "epoch": 0.6348041398720986, "grad_norm": 1.6465487252852629, "learning_rate": 6.216517818448423e-06, "loss": 0.4709, "step": 17818 }, { "epoch": 0.6348397669985927, "grad_norm": 1.2822316204627129, "learning_rate": 6.215449686740675e-06, "loss": 0.486, "step": 17819 }, { "epoch": 0.6348753941250869, "grad_norm": 1.5392325428061078, "learning_rate": 6.214381605426858e-06, "loss": 0.6133, "step": 17820 }, { "epoch": 0.6349110212515809, "grad_norm": 1.5287915551765894, "learning_rate": 6.213313574521196e-06, "loss": 0.5185, "step": 17821 }, { "epoch": 0.634946648378075, "grad_norm": 1.0801518079155132, "learning_rate": 6.2122455940379145e-06, "loss": 0.419, "step": 17822 }, { "epoch": 0.6349822755045692, "grad_norm": 1.095267661980538, "learning_rate": 6.211177663991226e-06, "loss": 0.4636, "step": 17823 }, { "epoch": 0.6350179026310633, "grad_norm": 1.3972751919849065, "learning_rate": 6.210109784395358e-06, "loss": 0.5141, "step": 17824 }, { "epoch": 0.6350535297575574, "grad_norm": 1.2678035176783518, "learning_rate": 6.209041955264524e-06, "loss": 0.4226, "step": 17825 }, { "epoch": 0.6350891568840515, "grad_norm": 1.1900631689760612, "learning_rate": 6.207974176612952e-06, "loss": 0.3898, "step": 17826 }, { "epoch": 0.6351247840105456, "grad_norm": 1.7491735156990353, "learning_rate": 6.206906448454848e-06, "loss": 0.5165, "step": 17827 }, { "epoch": 0.6351604111370397, "grad_norm": 1.3657099630862393, "learning_rate": 6.20583877080444e-06, "loss": 0.5418, "step": 17828 }, { "epoch": 0.6351960382635339, "grad_norm": 1.4825568129673696, "learning_rate": 6.204771143675936e-06, "loss": 0.5552, "step": 17829 }, { "epoch": 0.635231665390028, "grad_norm": 1.14749341682033, "learning_rate": 6.203703567083561e-06, "loss": 0.4647, "step": 17830 }, { "epoch": 0.635267292516522, "grad_norm": 1.1044676045080586, "learning_rate": 6.202636041041524e-06, "loss": 0.3979, "step": 17831 }, { "epoch": 0.6353029196430162, "grad_norm": 1.2428712693768527, "learning_rate": 6.201568565564042e-06, "loss": 0.4921, "step": 17832 }, { "epoch": 0.6353385467695103, "grad_norm": 1.296793090669694, "learning_rate": 6.200501140665334e-06, "loss": 0.5609, "step": 17833 }, { "epoch": 0.6353741738960044, "grad_norm": 1.5352807713859316, "learning_rate": 6.199433766359605e-06, "loss": 0.5309, "step": 17834 }, { "epoch": 0.6354098010224986, "grad_norm": 1.3726215268334283, "learning_rate": 6.1983664426610735e-06, "loss": 0.51, "step": 17835 }, { "epoch": 0.6354454281489926, "grad_norm": 1.9900931934970685, "learning_rate": 6.197299169583948e-06, "loss": 0.4748, "step": 17836 }, { "epoch": 0.6354810552754867, "grad_norm": 1.4126619995632197, "learning_rate": 6.1962319471424455e-06, "loss": 0.5365, "step": 17837 }, { "epoch": 0.6355166824019809, "grad_norm": 1.3197815069379712, "learning_rate": 6.19516477535077e-06, "loss": 0.528, "step": 17838 }, { "epoch": 0.635552309528475, "grad_norm": 1.4161926642297207, "learning_rate": 6.194097654223137e-06, "loss": 0.5386, "step": 17839 }, { "epoch": 0.635587936654969, "grad_norm": 1.7518818928986581, "learning_rate": 6.193030583773752e-06, "loss": 0.4914, "step": 17840 }, { "epoch": 0.6356235637814632, "grad_norm": 1.1709323539645893, "learning_rate": 6.191963564016829e-06, "loss": 0.4212, "step": 17841 }, { "epoch": 0.6356591909079573, "grad_norm": 1.0531117415960398, "learning_rate": 6.1908965949665675e-06, "loss": 0.3985, "step": 17842 }, { "epoch": 0.6356948180344514, "grad_norm": 1.3480524156691878, "learning_rate": 6.189829676637182e-06, "loss": 0.5306, "step": 17843 }, { "epoch": 0.6357304451609456, "grad_norm": 1.4581326473981298, "learning_rate": 6.188762809042882e-06, "loss": 0.4788, "step": 17844 }, { "epoch": 0.6357660722874396, "grad_norm": 1.416498194913261, "learning_rate": 6.187695992197864e-06, "loss": 0.461, "step": 17845 }, { "epoch": 0.6358016994139337, "grad_norm": 1.1085179224267738, "learning_rate": 6.186629226116342e-06, "loss": 0.418, "step": 17846 }, { "epoch": 0.6358373265404279, "grad_norm": 1.0381072303378558, "learning_rate": 6.185562510812514e-06, "loss": 0.3891, "step": 17847 }, { "epoch": 0.635872953666922, "grad_norm": 1.5009883856779689, "learning_rate": 6.184495846300591e-06, "loss": 0.4406, "step": 17848 }, { "epoch": 0.6359085807934161, "grad_norm": 1.1281577242924765, "learning_rate": 6.18342923259477e-06, "loss": 0.4575, "step": 17849 }, { "epoch": 0.6359442079199102, "grad_norm": 2.8705726289296947, "learning_rate": 6.182362669709257e-06, "loss": 0.534, "step": 17850 }, { "epoch": 0.6359798350464043, "grad_norm": 1.0346715105563733, "learning_rate": 6.181296157658252e-06, "loss": 0.4049, "step": 17851 }, { "epoch": 0.6360154621728984, "grad_norm": 2.4200858427305953, "learning_rate": 6.180229696455963e-06, "loss": 0.5739, "step": 17852 }, { "epoch": 0.6360510892993926, "grad_norm": 1.7884807856175469, "learning_rate": 6.179163286116581e-06, "loss": 0.6003, "step": 17853 }, { "epoch": 0.6360867164258867, "grad_norm": 1.7814082078082258, "learning_rate": 6.178096926654311e-06, "loss": 0.5239, "step": 17854 }, { "epoch": 0.6361223435523807, "grad_norm": 1.141228928752073, "learning_rate": 6.177030618083351e-06, "loss": 0.4247, "step": 17855 }, { "epoch": 0.6361579706788749, "grad_norm": 1.1175503422700215, "learning_rate": 6.175964360417899e-06, "loss": 0.4206, "step": 17856 }, { "epoch": 0.636193597805369, "grad_norm": 1.2002598427371978, "learning_rate": 6.174898153672157e-06, "loss": 0.4044, "step": 17857 }, { "epoch": 0.6362292249318631, "grad_norm": 1.536903357248274, "learning_rate": 6.173831997860319e-06, "loss": 0.4272, "step": 17858 }, { "epoch": 0.6362648520583573, "grad_norm": 1.3202266825739846, "learning_rate": 6.172765892996585e-06, "loss": 0.4719, "step": 17859 }, { "epoch": 0.6363004791848513, "grad_norm": 1.3412108997132033, "learning_rate": 6.171699839095144e-06, "loss": 0.4145, "step": 17860 }, { "epoch": 0.6363361063113454, "grad_norm": 1.1792520274423004, "learning_rate": 6.170633836170199e-06, "loss": 0.4083, "step": 17861 }, { "epoch": 0.6363717334378396, "grad_norm": 1.962187713906343, "learning_rate": 6.169567884235938e-06, "loss": 0.4718, "step": 17862 }, { "epoch": 0.6364073605643337, "grad_norm": 1.563681923603431, "learning_rate": 6.168501983306562e-06, "loss": 0.576, "step": 17863 }, { "epoch": 0.6364429876908277, "grad_norm": 1.0585750855714517, "learning_rate": 6.1674361333962565e-06, "loss": 0.4201, "step": 17864 }, { "epoch": 0.6364786148173219, "grad_norm": 1.3485438359803799, "learning_rate": 6.166370334519219e-06, "loss": 0.5709, "step": 17865 }, { "epoch": 0.636514241943816, "grad_norm": 1.3093404869642324, "learning_rate": 6.165304586689638e-06, "loss": 0.5181, "step": 17866 }, { "epoch": 0.6365498690703101, "grad_norm": 1.1143205746754066, "learning_rate": 6.1642388899217075e-06, "loss": 0.4262, "step": 17867 }, { "epoch": 0.6365854961968043, "grad_norm": 1.5733027558100896, "learning_rate": 6.163173244229618e-06, "loss": 0.501, "step": 17868 }, { "epoch": 0.6366211233232983, "grad_norm": 1.1477995392463018, "learning_rate": 6.162107649627558e-06, "loss": 0.4414, "step": 17869 }, { "epoch": 0.6366567504497924, "grad_norm": 1.3328044675974706, "learning_rate": 6.16104210612972e-06, "loss": 0.5343, "step": 17870 }, { "epoch": 0.6366923775762866, "grad_norm": 1.9263628135650182, "learning_rate": 6.159976613750287e-06, "loss": 0.5165, "step": 17871 }, { "epoch": 0.6367280047027807, "grad_norm": 1.8433106397788916, "learning_rate": 6.15891117250345e-06, "loss": 0.581, "step": 17872 }, { "epoch": 0.6367636318292748, "grad_norm": 1.1234041692845496, "learning_rate": 6.1578457824033935e-06, "loss": 0.4554, "step": 17873 }, { "epoch": 0.6367992589557689, "grad_norm": 1.375005258057349, "learning_rate": 6.156780443464311e-06, "loss": 0.4343, "step": 17874 }, { "epoch": 0.636834886082263, "grad_norm": 1.0316972137170897, "learning_rate": 6.155715155700378e-06, "loss": 0.4071, "step": 17875 }, { "epoch": 0.6368705132087571, "grad_norm": 1.450212124182609, "learning_rate": 6.154649919125788e-06, "loss": 0.4676, "step": 17876 }, { "epoch": 0.6369061403352513, "grad_norm": 1.716613195279915, "learning_rate": 6.1535847337547185e-06, "loss": 0.4161, "step": 17877 }, { "epoch": 0.6369417674617454, "grad_norm": 1.6051475355333442, "learning_rate": 6.152519599601359e-06, "loss": 0.5569, "step": 17878 }, { "epoch": 0.6369773945882394, "grad_norm": 1.2343513869215357, "learning_rate": 6.151454516679891e-06, "loss": 0.446, "step": 17879 }, { "epoch": 0.6370130217147336, "grad_norm": 1.6042992351288683, "learning_rate": 6.150389485004493e-06, "loss": 0.4568, "step": 17880 }, { "epoch": 0.6370486488412277, "grad_norm": 1.0948791890633767, "learning_rate": 6.149324504589354e-06, "loss": 0.4228, "step": 17881 }, { "epoch": 0.6370842759677218, "grad_norm": 1.5852273066549658, "learning_rate": 6.148259575448648e-06, "loss": 0.6666, "step": 17882 }, { "epoch": 0.637119903094216, "grad_norm": 1.5194433303791894, "learning_rate": 6.147194697596559e-06, "loss": 0.4215, "step": 17883 }, { "epoch": 0.63715553022071, "grad_norm": 1.6296850007583468, "learning_rate": 6.1461298710472635e-06, "loss": 0.5547, "step": 17884 }, { "epoch": 0.6371911573472041, "grad_norm": 1.3812024932944724, "learning_rate": 6.145065095814947e-06, "loss": 0.4523, "step": 17885 }, { "epoch": 0.6372267844736983, "grad_norm": 1.1025122203298474, "learning_rate": 6.144000371913779e-06, "loss": 0.4288, "step": 17886 }, { "epoch": 0.6372624116001924, "grad_norm": 1.786503740732536, "learning_rate": 6.1429356993579435e-06, "loss": 0.5498, "step": 17887 }, { "epoch": 0.6372980387266864, "grad_norm": 1.1327893463258074, "learning_rate": 6.141871078161612e-06, "loss": 0.4257, "step": 17888 }, { "epoch": 0.6373336658531806, "grad_norm": 1.4764138891390417, "learning_rate": 6.140806508338966e-06, "loss": 0.5903, "step": 17889 }, { "epoch": 0.6373692929796747, "grad_norm": 1.1171434330225867, "learning_rate": 6.139741989904181e-06, "loss": 0.3983, "step": 17890 }, { "epoch": 0.6374049201061688, "grad_norm": 1.3630081052581406, "learning_rate": 6.138677522871424e-06, "loss": 0.5978, "step": 17891 }, { "epoch": 0.637440547232663, "grad_norm": 5.28464781309136, "learning_rate": 6.13761310725488e-06, "loss": 0.4615, "step": 17892 }, { "epoch": 0.637476174359157, "grad_norm": 1.0379979744893344, "learning_rate": 6.136548743068713e-06, "loss": 0.3769, "step": 17893 }, { "epoch": 0.6375118014856511, "grad_norm": 1.6801086451949268, "learning_rate": 6.135484430327103e-06, "loss": 0.5818, "step": 17894 }, { "epoch": 0.6375474286121453, "grad_norm": 1.2591320244899706, "learning_rate": 6.1344201690442175e-06, "loss": 0.533, "step": 17895 }, { "epoch": 0.6375830557386394, "grad_norm": 1.120098281698462, "learning_rate": 6.133355959234232e-06, "loss": 0.4199, "step": 17896 }, { "epoch": 0.6376186828651336, "grad_norm": 1.5603827581093452, "learning_rate": 6.1322918009113105e-06, "loss": 0.5422, "step": 17897 }, { "epoch": 0.6376543099916276, "grad_norm": 1.4687811186027058, "learning_rate": 6.13122769408963e-06, "loss": 0.6222, "step": 17898 }, { "epoch": 0.6376899371181217, "grad_norm": 1.1586251546576145, "learning_rate": 6.1301636387833545e-06, "loss": 0.4262, "step": 17899 }, { "epoch": 0.6377255642446159, "grad_norm": 1.0898257284997848, "learning_rate": 6.129099635006659e-06, "loss": 0.4006, "step": 17900 }, { "epoch": 0.63776119137111, "grad_norm": 1.0896022085705186, "learning_rate": 6.128035682773703e-06, "loss": 0.4107, "step": 17901 }, { "epoch": 0.6377968184976041, "grad_norm": 2.0872218079224987, "learning_rate": 6.1269717820986585e-06, "loss": 0.5684, "step": 17902 }, { "epoch": 0.6378324456240982, "grad_norm": 1.5986132633120969, "learning_rate": 6.125907932995694e-06, "loss": 0.5376, "step": 17903 }, { "epoch": 0.6378680727505923, "grad_norm": 1.216924450191897, "learning_rate": 6.124844135478971e-06, "loss": 0.428, "step": 17904 }, { "epoch": 0.6379036998770864, "grad_norm": 1.2516846174779304, "learning_rate": 6.123780389562659e-06, "loss": 0.4108, "step": 17905 }, { "epoch": 0.6379393270035806, "grad_norm": 1.331092271697223, "learning_rate": 6.1227166952609174e-06, "loss": 0.5208, "step": 17906 }, { "epoch": 0.6379749541300747, "grad_norm": 1.0343728177035791, "learning_rate": 6.1216530525879185e-06, "loss": 0.4101, "step": 17907 }, { "epoch": 0.6380105812565687, "grad_norm": 1.1608140548007664, "learning_rate": 6.120589461557815e-06, "loss": 0.4357, "step": 17908 }, { "epoch": 0.6380462083830629, "grad_norm": 1.5008228134325383, "learning_rate": 6.119525922184777e-06, "loss": 0.5391, "step": 17909 }, { "epoch": 0.638081835509557, "grad_norm": 1.255122562005201, "learning_rate": 6.118462434482961e-06, "loss": 0.4804, "step": 17910 }, { "epoch": 0.6381174626360511, "grad_norm": 1.2749074170644548, "learning_rate": 6.1173989984665365e-06, "loss": 0.4579, "step": 17911 }, { "epoch": 0.6381530897625453, "grad_norm": 1.1094236326105635, "learning_rate": 6.116335614149652e-06, "loss": 0.3854, "step": 17912 }, { "epoch": 0.6381887168890393, "grad_norm": 1.130840035368593, "learning_rate": 6.115272281546474e-06, "loss": 0.4394, "step": 17913 }, { "epoch": 0.6382243440155334, "grad_norm": 1.1489720602390707, "learning_rate": 6.114209000671163e-06, "loss": 0.4489, "step": 17914 }, { "epoch": 0.6382599711420276, "grad_norm": 1.3171536957203749, "learning_rate": 6.113145771537874e-06, "loss": 0.5189, "step": 17915 }, { "epoch": 0.6382955982685217, "grad_norm": 1.4979191184472844, "learning_rate": 6.112082594160769e-06, "loss": 0.5107, "step": 17916 }, { "epoch": 0.6383312253950157, "grad_norm": 1.4275336859608956, "learning_rate": 6.111019468553995e-06, "loss": 0.5744, "step": 17917 }, { "epoch": 0.6383668525215099, "grad_norm": 1.6422758581686159, "learning_rate": 6.109956394731722e-06, "loss": 0.5386, "step": 17918 }, { "epoch": 0.638402479648004, "grad_norm": 1.1383093474212935, "learning_rate": 6.1088933727080935e-06, "loss": 0.4249, "step": 17919 }, { "epoch": 0.6384381067744981, "grad_norm": 1.4687652761341863, "learning_rate": 6.107830402497272e-06, "loss": 0.5217, "step": 17920 }, { "epoch": 0.6384737339009923, "grad_norm": 1.1353831037027682, "learning_rate": 6.1067674841134075e-06, "loss": 0.4205, "step": 17921 }, { "epoch": 0.6385093610274863, "grad_norm": 1.5181097429923602, "learning_rate": 6.1057046175706605e-06, "loss": 0.5336, "step": 17922 }, { "epoch": 0.6385449881539804, "grad_norm": 1.052628192197619, "learning_rate": 6.104641802883172e-06, "loss": 0.447, "step": 17923 }, { "epoch": 0.6385806152804746, "grad_norm": 1.6117753778732529, "learning_rate": 6.103579040065102e-06, "loss": 0.5191, "step": 17924 }, { "epoch": 0.6386162424069687, "grad_norm": 1.1712692248803835, "learning_rate": 6.102516329130601e-06, "loss": 0.433, "step": 17925 }, { "epoch": 0.6386518695334628, "grad_norm": 1.0870822280002213, "learning_rate": 6.10145367009382e-06, "loss": 0.4055, "step": 17926 }, { "epoch": 0.6386874966599569, "grad_norm": 2.196084305492867, "learning_rate": 6.10039106296891e-06, "loss": 0.567, "step": 17927 }, { "epoch": 0.638723123786451, "grad_norm": 1.2671697139977618, "learning_rate": 6.099328507770015e-06, "loss": 0.4487, "step": 17928 }, { "epoch": 0.6387587509129451, "grad_norm": 1.7964017860202273, "learning_rate": 6.098266004511291e-06, "loss": 0.5118, "step": 17929 }, { "epoch": 0.6387943780394393, "grad_norm": 1.1388701337634146, "learning_rate": 6.097203553206879e-06, "loss": 0.4166, "step": 17930 }, { "epoch": 0.6388300051659334, "grad_norm": 1.2807853037347925, "learning_rate": 6.09614115387093e-06, "loss": 0.3893, "step": 17931 }, { "epoch": 0.6388656322924274, "grad_norm": 1.2163162251514723, "learning_rate": 6.09507880651759e-06, "loss": 0.4668, "step": 17932 }, { "epoch": 0.6389012594189216, "grad_norm": 1.4205194754561143, "learning_rate": 6.09401651116101e-06, "loss": 0.4918, "step": 17933 }, { "epoch": 0.6389368865454157, "grad_norm": 1.7061445566392233, "learning_rate": 6.0929542678153244e-06, "loss": 0.5327, "step": 17934 }, { "epoch": 0.6389725136719098, "grad_norm": 1.1603752723515324, "learning_rate": 6.0918920764946856e-06, "loss": 0.422, "step": 17935 }, { "epoch": 0.639008140798404, "grad_norm": 1.1431272425311554, "learning_rate": 6.090829937213234e-06, "loss": 0.4256, "step": 17936 }, { "epoch": 0.639043767924898, "grad_norm": 1.0660932711924025, "learning_rate": 6.089767849985114e-06, "loss": 0.4001, "step": 17937 }, { "epoch": 0.6390793950513921, "grad_norm": 1.0799411854384593, "learning_rate": 6.088705814824473e-06, "loss": 0.3915, "step": 17938 }, { "epoch": 0.6391150221778863, "grad_norm": 1.67322563495518, "learning_rate": 6.0876438317454434e-06, "loss": 0.5291, "step": 17939 }, { "epoch": 0.6391506493043804, "grad_norm": 2.009170983971519, "learning_rate": 6.0865819007621716e-06, "loss": 0.5873, "step": 17940 }, { "epoch": 0.6391862764308744, "grad_norm": 1.1144081887203658, "learning_rate": 6.085520021888797e-06, "loss": 0.4294, "step": 17941 }, { "epoch": 0.6392219035573686, "grad_norm": 1.5634461219826559, "learning_rate": 6.084458195139462e-06, "loss": 0.4479, "step": 17942 }, { "epoch": 0.6392575306838627, "grad_norm": 1.236001011297852, "learning_rate": 6.083396420528298e-06, "loss": 0.4545, "step": 17943 }, { "epoch": 0.6392931578103568, "grad_norm": 1.2903618467518836, "learning_rate": 6.082334698069454e-06, "loss": 0.4768, "step": 17944 }, { "epoch": 0.639328784936851, "grad_norm": 1.4501861359896457, "learning_rate": 6.081273027777057e-06, "loss": 0.5216, "step": 17945 }, { "epoch": 0.639364412063345, "grad_norm": 2.2313024655384126, "learning_rate": 6.080211409665251e-06, "loss": 0.5069, "step": 17946 }, { "epoch": 0.6394000391898391, "grad_norm": 1.3506095704420733, "learning_rate": 6.079149843748168e-06, "loss": 0.4304, "step": 17947 }, { "epoch": 0.6394356663163333, "grad_norm": 1.1990126172622984, "learning_rate": 6.078088330039945e-06, "loss": 0.4837, "step": 17948 }, { "epoch": 0.6394712934428274, "grad_norm": 1.490234212009401, "learning_rate": 6.077026868554721e-06, "loss": 0.5, "step": 17949 }, { "epoch": 0.6395069205693215, "grad_norm": 1.855089494920462, "learning_rate": 6.0759654593066235e-06, "loss": 0.5873, "step": 17950 }, { "epoch": 0.6395425476958156, "grad_norm": 1.2121129857769375, "learning_rate": 6.074904102309789e-06, "loss": 0.455, "step": 17951 }, { "epoch": 0.6395781748223097, "grad_norm": 1.9532268142865281, "learning_rate": 6.0738427975783485e-06, "loss": 0.5748, "step": 17952 }, { "epoch": 0.6396138019488038, "grad_norm": 1.5382028081001244, "learning_rate": 6.072781545126441e-06, "loss": 0.4967, "step": 17953 }, { "epoch": 0.639649429075298, "grad_norm": 1.707483572375368, "learning_rate": 6.071720344968186e-06, "loss": 0.5011, "step": 17954 }, { "epoch": 0.6396850562017921, "grad_norm": 1.4465215888516922, "learning_rate": 6.0706591971177235e-06, "loss": 0.4942, "step": 17955 }, { "epoch": 0.6397206833282861, "grad_norm": 1.1609249673503519, "learning_rate": 6.069598101589177e-06, "loss": 0.4301, "step": 17956 }, { "epoch": 0.6397563104547803, "grad_norm": 1.0931674175442374, "learning_rate": 6.068537058396682e-06, "loss": 0.3982, "step": 17957 }, { "epoch": 0.6397919375812744, "grad_norm": 1.5289314724639989, "learning_rate": 6.067476067554361e-06, "loss": 0.5195, "step": 17958 }, { "epoch": 0.6398275647077685, "grad_norm": 1.6906461156380181, "learning_rate": 6.066415129076346e-06, "loss": 0.5049, "step": 17959 }, { "epoch": 0.6398631918342627, "grad_norm": 1.1934731434223436, "learning_rate": 6.0653542429767664e-06, "loss": 0.4128, "step": 17960 }, { "epoch": 0.6398988189607567, "grad_norm": 1.8332783319946284, "learning_rate": 6.064293409269742e-06, "loss": 0.5493, "step": 17961 }, { "epoch": 0.6399344460872508, "grad_norm": 1.3211669581569172, "learning_rate": 6.063232627969402e-06, "loss": 0.4877, "step": 17962 }, { "epoch": 0.639970073213745, "grad_norm": 1.5538227584857296, "learning_rate": 6.06217189908987e-06, "loss": 0.5569, "step": 17963 }, { "epoch": 0.6400057003402391, "grad_norm": 1.1731261646219746, "learning_rate": 6.061111222645276e-06, "loss": 0.4395, "step": 17964 }, { "epoch": 0.6400413274667331, "grad_norm": 1.5211245168125407, "learning_rate": 6.060050598649735e-06, "loss": 0.4957, "step": 17965 }, { "epoch": 0.6400769545932273, "grad_norm": 1.3609440341494459, "learning_rate": 6.058990027117376e-06, "loss": 0.5541, "step": 17966 }, { "epoch": 0.6401125817197214, "grad_norm": 1.485301487260748, "learning_rate": 6.057929508062318e-06, "loss": 0.5544, "step": 17967 }, { "epoch": 0.6401482088462155, "grad_norm": 1.880221068060303, "learning_rate": 6.0568690414986875e-06, "loss": 0.4771, "step": 17968 }, { "epoch": 0.6401838359727097, "grad_norm": 1.3978928394647412, "learning_rate": 6.055808627440596e-06, "loss": 0.5152, "step": 17969 }, { "epoch": 0.6402194630992037, "grad_norm": 1.0505490238952684, "learning_rate": 6.0547482659021704e-06, "loss": 0.3872, "step": 17970 }, { "epoch": 0.6402550902256978, "grad_norm": 1.5707403987943915, "learning_rate": 6.053687956897534e-06, "loss": 0.6384, "step": 17971 }, { "epoch": 0.640290717352192, "grad_norm": 1.8497415649293873, "learning_rate": 6.052627700440796e-06, "loss": 0.4964, "step": 17972 }, { "epoch": 0.6403263444786861, "grad_norm": 1.9634686227116465, "learning_rate": 6.051567496546083e-06, "loss": 0.5279, "step": 17973 }, { "epoch": 0.6403619716051802, "grad_norm": 1.182996582398908, "learning_rate": 6.050507345227505e-06, "loss": 0.4135, "step": 17974 }, { "epoch": 0.6403975987316743, "grad_norm": 1.0817983618076001, "learning_rate": 6.049447246499187e-06, "loss": 0.4565, "step": 17975 }, { "epoch": 0.6404332258581684, "grad_norm": 1.1089901825905228, "learning_rate": 6.048387200375236e-06, "loss": 0.4272, "step": 17976 }, { "epoch": 0.6404688529846625, "grad_norm": 1.22459358840062, "learning_rate": 6.047327206869773e-06, "loss": 0.4539, "step": 17977 }, { "epoch": 0.6405044801111567, "grad_norm": 1.5945843809841302, "learning_rate": 6.0462672659969105e-06, "loss": 0.5887, "step": 17978 }, { "epoch": 0.6405401072376508, "grad_norm": 1.1623762571175662, "learning_rate": 6.045207377770765e-06, "loss": 0.4161, "step": 17979 }, { "epoch": 0.6405757343641448, "grad_norm": 1.213181454186307, "learning_rate": 6.044147542205443e-06, "loss": 0.442, "step": 17980 }, { "epoch": 0.640611361490639, "grad_norm": 1.2932600575906943, "learning_rate": 6.043087759315066e-06, "loss": 0.5111, "step": 17981 }, { "epoch": 0.6406469886171331, "grad_norm": 1.2089960981125754, "learning_rate": 6.042028029113738e-06, "loss": 0.4634, "step": 17982 }, { "epoch": 0.6406826157436272, "grad_norm": 1.1948001916911573, "learning_rate": 6.040968351615572e-06, "loss": 0.4309, "step": 17983 }, { "epoch": 0.6407182428701214, "grad_norm": 1.9861465624313206, "learning_rate": 6.039908726834684e-06, "loss": 0.5022, "step": 17984 }, { "epoch": 0.6407538699966154, "grad_norm": 1.2972424909851463, "learning_rate": 6.038849154785176e-06, "loss": 0.4014, "step": 17985 }, { "epoch": 0.6407894971231095, "grad_norm": 1.1076138579199726, "learning_rate": 6.037789635481164e-06, "loss": 0.4422, "step": 17986 }, { "epoch": 0.6408251242496037, "grad_norm": 1.1524039138587763, "learning_rate": 6.0367301689367485e-06, "loss": 0.3967, "step": 17987 }, { "epoch": 0.6408607513760978, "grad_norm": 1.1691726744658002, "learning_rate": 6.0356707551660434e-06, "loss": 0.4307, "step": 17988 }, { "epoch": 0.6408963785025918, "grad_norm": 1.0127591186529532, "learning_rate": 6.034611394183151e-06, "loss": 0.4158, "step": 17989 }, { "epoch": 0.640932005629086, "grad_norm": 1.0385497351675783, "learning_rate": 6.033552086002184e-06, "loss": 0.451, "step": 17990 }, { "epoch": 0.6409676327555801, "grad_norm": 1.1269660844178475, "learning_rate": 6.032492830637238e-06, "loss": 0.4238, "step": 17991 }, { "epoch": 0.6410032598820742, "grad_norm": 1.14698118556143, "learning_rate": 6.031433628102426e-06, "loss": 0.4057, "step": 17992 }, { "epoch": 0.6410388870085684, "grad_norm": 1.5431940525359678, "learning_rate": 6.030374478411847e-06, "loss": 0.5945, "step": 17993 }, { "epoch": 0.6410745141350624, "grad_norm": 1.2604198694474569, "learning_rate": 6.029315381579606e-06, "loss": 0.4903, "step": 17994 }, { "epoch": 0.6411101412615565, "grad_norm": 1.0624786737561005, "learning_rate": 6.028256337619807e-06, "loss": 0.4498, "step": 17995 }, { "epoch": 0.6411457683880507, "grad_norm": 1.1883834713300243, "learning_rate": 6.0271973465465505e-06, "loss": 0.4243, "step": 17996 }, { "epoch": 0.6411813955145448, "grad_norm": 1.4182463414112796, "learning_rate": 6.026138408373942e-06, "loss": 0.5142, "step": 17997 }, { "epoch": 0.6412170226410389, "grad_norm": 1.150956382363798, "learning_rate": 6.0250795231160745e-06, "loss": 0.436, "step": 17998 }, { "epoch": 0.641252649767533, "grad_norm": 1.0690713266068412, "learning_rate": 6.0240206907870526e-06, "loss": 0.4176, "step": 17999 }, { "epoch": 0.6412882768940271, "grad_norm": 2.3076146408480076, "learning_rate": 6.022961911400973e-06, "loss": 0.5423, "step": 18000 }, { "epoch": 0.6413239040205212, "grad_norm": 1.5775404914122653, "learning_rate": 6.021903184971938e-06, "loss": 0.5314, "step": 18001 }, { "epoch": 0.6413595311470154, "grad_norm": 1.2404298636632407, "learning_rate": 6.02084451151404e-06, "loss": 0.3997, "step": 18002 }, { "epoch": 0.6413951582735095, "grad_norm": 1.1613124030381343, "learning_rate": 6.019785891041381e-06, "loss": 0.4365, "step": 18003 }, { "epoch": 0.6414307854000035, "grad_norm": 1.1969511241972663, "learning_rate": 6.018727323568052e-06, "loss": 0.4743, "step": 18004 }, { "epoch": 0.6414664125264977, "grad_norm": 1.41082709520977, "learning_rate": 6.017668809108151e-06, "loss": 0.5768, "step": 18005 }, { "epoch": 0.6415020396529918, "grad_norm": 1.1665064386344481, "learning_rate": 6.016610347675775e-06, "loss": 0.4175, "step": 18006 }, { "epoch": 0.6415376667794859, "grad_norm": 1.235655450392062, "learning_rate": 6.015551939285016e-06, "loss": 0.4446, "step": 18007 }, { "epoch": 0.6415732939059801, "grad_norm": 1.1181135138162148, "learning_rate": 6.014493583949971e-06, "loss": 0.4099, "step": 18008 }, { "epoch": 0.6416089210324741, "grad_norm": 1.1920502494960559, "learning_rate": 6.013435281684726e-06, "loss": 0.4255, "step": 18009 }, { "epoch": 0.6416445481589682, "grad_norm": 1.1752421742888723, "learning_rate": 6.012377032503378e-06, "loss": 0.4264, "step": 18010 }, { "epoch": 0.6416801752854624, "grad_norm": 1.3293000028870001, "learning_rate": 6.0113188364200145e-06, "loss": 0.4632, "step": 18011 }, { "epoch": 0.6417158024119565, "grad_norm": 1.5486562503530286, "learning_rate": 6.010260693448735e-06, "loss": 0.5397, "step": 18012 }, { "epoch": 0.6417514295384505, "grad_norm": 1.3325801971276545, "learning_rate": 6.009202603603616e-06, "loss": 0.4705, "step": 18013 }, { "epoch": 0.6417870566649447, "grad_norm": 1.8041562819943535, "learning_rate": 6.008144566898758e-06, "loss": 0.5804, "step": 18014 }, { "epoch": 0.6418226837914388, "grad_norm": 1.0910371110876196, "learning_rate": 6.007086583348243e-06, "loss": 0.438, "step": 18015 }, { "epoch": 0.6418583109179329, "grad_norm": 1.5451954403023567, "learning_rate": 6.0060286529661604e-06, "loss": 0.5163, "step": 18016 }, { "epoch": 0.6418939380444271, "grad_norm": 1.5035889143398962, "learning_rate": 6.004970775766601e-06, "loss": 0.4598, "step": 18017 }, { "epoch": 0.6419295651709211, "grad_norm": 1.2155334648677747, "learning_rate": 6.003912951763644e-06, "loss": 0.4382, "step": 18018 }, { "epoch": 0.6419651922974152, "grad_norm": 1.1824582841174514, "learning_rate": 6.002855180971384e-06, "loss": 0.4356, "step": 18019 }, { "epoch": 0.6420008194239094, "grad_norm": 2.201991292199303, "learning_rate": 6.001797463403898e-06, "loss": 0.3746, "step": 18020 }, { "epoch": 0.6420364465504035, "grad_norm": 1.1327042976833066, "learning_rate": 6.000739799075275e-06, "loss": 0.4061, "step": 18021 }, { "epoch": 0.6420720736768976, "grad_norm": 1.3029952709879051, "learning_rate": 5.999682187999594e-06, "loss": 0.4042, "step": 18022 }, { "epoch": 0.6421077008033917, "grad_norm": 1.4833440602359547, "learning_rate": 5.998624630190946e-06, "loss": 0.5497, "step": 18023 }, { "epoch": 0.6421433279298858, "grad_norm": 1.2702056259676813, "learning_rate": 5.997567125663401e-06, "loss": 0.4842, "step": 18024 }, { "epoch": 0.6421789550563799, "grad_norm": 1.4467491633485787, "learning_rate": 5.996509674431053e-06, "loss": 0.471, "step": 18025 }, { "epoch": 0.6422145821828741, "grad_norm": 2.0144625602470994, "learning_rate": 5.995452276507973e-06, "loss": 0.5109, "step": 18026 }, { "epoch": 0.6422502093093682, "grad_norm": 1.2792428567434682, "learning_rate": 5.99439493190825e-06, "loss": 0.4351, "step": 18027 }, { "epoch": 0.6422858364358622, "grad_norm": 1.6125559567850236, "learning_rate": 5.993337640645953e-06, "loss": 0.5534, "step": 18028 }, { "epoch": 0.6423214635623564, "grad_norm": 1.1927692352651231, "learning_rate": 5.992280402735167e-06, "loss": 0.4291, "step": 18029 }, { "epoch": 0.6423570906888505, "grad_norm": 1.3163180155495764, "learning_rate": 5.991223218189969e-06, "loss": 0.4622, "step": 18030 }, { "epoch": 0.6423927178153446, "grad_norm": 1.3450897857666098, "learning_rate": 5.9901660870244345e-06, "loss": 0.462, "step": 18031 }, { "epoch": 0.6424283449418388, "grad_norm": 1.3552664521478264, "learning_rate": 5.989109009252644e-06, "loss": 0.461, "step": 18032 }, { "epoch": 0.6424639720683328, "grad_norm": 1.152907771151692, "learning_rate": 5.988051984888668e-06, "loss": 0.4419, "step": 18033 }, { "epoch": 0.6424995991948269, "grad_norm": 1.2946774580249787, "learning_rate": 5.986995013946588e-06, "loss": 0.4452, "step": 18034 }, { "epoch": 0.6425352263213211, "grad_norm": 1.4496062134317258, "learning_rate": 5.985938096440471e-06, "loss": 0.3987, "step": 18035 }, { "epoch": 0.6425708534478152, "grad_norm": 1.2573764466184811, "learning_rate": 5.984881232384394e-06, "loss": 0.4491, "step": 18036 }, { "epoch": 0.6426064805743092, "grad_norm": 1.2292072676488819, "learning_rate": 5.98382442179243e-06, "loss": 0.4144, "step": 18037 }, { "epoch": 0.6426421077008034, "grad_norm": 1.1173893503417878, "learning_rate": 5.982767664678654e-06, "loss": 0.4374, "step": 18038 }, { "epoch": 0.6426777348272975, "grad_norm": 1.5254102115208816, "learning_rate": 5.98171096105713e-06, "loss": 0.4542, "step": 18039 }, { "epoch": 0.6427133619537916, "grad_norm": 1.7520225148797106, "learning_rate": 5.9806543109419335e-06, "loss": 0.5153, "step": 18040 }, { "epoch": 0.6427489890802858, "grad_norm": 1.5256616325804646, "learning_rate": 5.979597714347136e-06, "loss": 0.5077, "step": 18041 }, { "epoch": 0.6427846162067798, "grad_norm": 1.180056363372927, "learning_rate": 5.9785411712868045e-06, "loss": 0.4314, "step": 18042 }, { "epoch": 0.6428202433332739, "grad_norm": 1.6171628952102588, "learning_rate": 5.977484681775011e-06, "loss": 0.5369, "step": 18043 }, { "epoch": 0.6428558704597681, "grad_norm": 1.7349026763761164, "learning_rate": 5.976428245825814e-06, "loss": 0.5257, "step": 18044 }, { "epoch": 0.6428914975862622, "grad_norm": 1.4609441457223564, "learning_rate": 5.9753718634532944e-06, "loss": 0.5524, "step": 18045 }, { "epoch": 0.6429271247127563, "grad_norm": 1.371255912942918, "learning_rate": 5.974315534671507e-06, "loss": 0.4519, "step": 18046 }, { "epoch": 0.6429627518392504, "grad_norm": 1.1338440595365347, "learning_rate": 5.973259259494524e-06, "loss": 0.4182, "step": 18047 }, { "epoch": 0.6429983789657445, "grad_norm": 1.5212872646522066, "learning_rate": 5.972203037936406e-06, "loss": 0.6113, "step": 18048 }, { "epoch": 0.6430340060922386, "grad_norm": 1.173240619211958, "learning_rate": 5.971146870011224e-06, "loss": 0.4475, "step": 18049 }, { "epoch": 0.6430696332187328, "grad_norm": 1.4719996609559638, "learning_rate": 5.970090755733034e-06, "loss": 0.4827, "step": 18050 }, { "epoch": 0.6431052603452269, "grad_norm": 1.1168608972111058, "learning_rate": 5.9690346951159005e-06, "loss": 0.4331, "step": 18051 }, { "epoch": 0.6431408874717209, "grad_norm": 1.3486258831722913, "learning_rate": 5.9679786881738925e-06, "loss": 0.4173, "step": 18052 }, { "epoch": 0.6431765145982151, "grad_norm": 1.2502716346555627, "learning_rate": 5.966922734921062e-06, "loss": 0.3935, "step": 18053 }, { "epoch": 0.6432121417247092, "grad_norm": 1.4260543887372863, "learning_rate": 5.965866835371479e-06, "loss": 0.6434, "step": 18054 }, { "epoch": 0.6432477688512033, "grad_norm": 1.2742806596219967, "learning_rate": 5.964810989539196e-06, "loss": 0.4374, "step": 18055 }, { "epoch": 0.6432833959776975, "grad_norm": 1.5245866538922546, "learning_rate": 5.963755197438275e-06, "loss": 0.5259, "step": 18056 }, { "epoch": 0.6433190231041915, "grad_norm": 1.5949089680556179, "learning_rate": 5.962699459082775e-06, "loss": 0.5767, "step": 18057 }, { "epoch": 0.6433546502306856, "grad_norm": 1.177027731427983, "learning_rate": 5.961643774486754e-06, "loss": 0.4144, "step": 18058 }, { "epoch": 0.6433902773571798, "grad_norm": 1.1907363434984846, "learning_rate": 5.960588143664266e-06, "loss": 0.4219, "step": 18059 }, { "epoch": 0.6434259044836739, "grad_norm": 1.1808142181612091, "learning_rate": 5.959532566629376e-06, "loss": 0.4327, "step": 18060 }, { "epoch": 0.643461531610168, "grad_norm": 1.1157827989825277, "learning_rate": 5.958477043396129e-06, "loss": 0.4162, "step": 18061 }, { "epoch": 0.6434971587366621, "grad_norm": 2.3590223563792687, "learning_rate": 5.9574215739785875e-06, "loss": 0.4818, "step": 18062 }, { "epoch": 0.6435327858631562, "grad_norm": 1.5826277703754381, "learning_rate": 5.956366158390802e-06, "loss": 0.5495, "step": 18063 }, { "epoch": 0.6435684129896503, "grad_norm": 1.2428332387625418, "learning_rate": 5.955310796646826e-06, "loss": 0.5096, "step": 18064 }, { "epoch": 0.6436040401161445, "grad_norm": 1.5406631533349635, "learning_rate": 5.954255488760719e-06, "loss": 0.4996, "step": 18065 }, { "epoch": 0.6436396672426385, "grad_norm": 1.189194510992666, "learning_rate": 5.953200234746524e-06, "loss": 0.4359, "step": 18066 }, { "epoch": 0.6436752943691326, "grad_norm": 1.11258093744402, "learning_rate": 5.9521450346182975e-06, "loss": 0.4173, "step": 18067 }, { "epoch": 0.6437109214956268, "grad_norm": 1.1055007910580439, "learning_rate": 5.951089888390087e-06, "loss": 0.3867, "step": 18068 }, { "epoch": 0.6437465486221209, "grad_norm": 1.112541799577213, "learning_rate": 5.950034796075948e-06, "loss": 0.4313, "step": 18069 }, { "epoch": 0.643782175748615, "grad_norm": 1.8168983986999383, "learning_rate": 5.948979757689923e-06, "loss": 0.4456, "step": 18070 }, { "epoch": 0.6438178028751091, "grad_norm": 1.1237432710855384, "learning_rate": 5.947924773246068e-06, "loss": 0.4386, "step": 18071 }, { "epoch": 0.6438534300016032, "grad_norm": 1.2834780021815893, "learning_rate": 5.946869842758425e-06, "loss": 0.3805, "step": 18072 }, { "epoch": 0.6438890571280973, "grad_norm": 1.8575791528819101, "learning_rate": 5.945814966241043e-06, "loss": 0.5415, "step": 18073 }, { "epoch": 0.6439246842545915, "grad_norm": 1.4377494025214494, "learning_rate": 5.944760143707966e-06, "loss": 0.4623, "step": 18074 }, { "epoch": 0.6439603113810856, "grad_norm": 1.0939369303536086, "learning_rate": 5.943705375173243e-06, "loss": 0.4192, "step": 18075 }, { "epoch": 0.6439959385075796, "grad_norm": 1.183170633610951, "learning_rate": 5.942650660650923e-06, "loss": 0.4222, "step": 18076 }, { "epoch": 0.6440315656340738, "grad_norm": 1.1454159545608056, "learning_rate": 5.941596000155041e-06, "loss": 0.403, "step": 18077 }, { "epoch": 0.6440671927605679, "grad_norm": 1.6717314011123168, "learning_rate": 5.940541393699646e-06, "loss": 0.5538, "step": 18078 }, { "epoch": 0.644102819887062, "grad_norm": 1.3454304211574488, "learning_rate": 5.939486841298779e-06, "loss": 0.5033, "step": 18079 }, { "epoch": 0.6441384470135562, "grad_norm": 1.4347047860793645, "learning_rate": 5.9384323429664845e-06, "loss": 0.5388, "step": 18080 }, { "epoch": 0.6441740741400502, "grad_norm": 2.500136340059567, "learning_rate": 5.9373778987167984e-06, "loss": 0.575, "step": 18081 }, { "epoch": 0.6442097012665443, "grad_norm": 1.5436494624374864, "learning_rate": 5.93632350856377e-06, "loss": 0.5775, "step": 18082 }, { "epoch": 0.6442453283930385, "grad_norm": 1.2207006906734075, "learning_rate": 5.935269172521432e-06, "loss": 0.4578, "step": 18083 }, { "epoch": 0.6442809555195326, "grad_norm": 1.3622669776452518, "learning_rate": 5.934214890603826e-06, "loss": 0.4364, "step": 18084 }, { "epoch": 0.6443165826460266, "grad_norm": 1.6340660043359643, "learning_rate": 5.933160662824991e-06, "loss": 0.5833, "step": 18085 }, { "epoch": 0.6443522097725208, "grad_norm": 1.1486768585019413, "learning_rate": 5.932106489198962e-06, "loss": 0.461, "step": 18086 }, { "epoch": 0.6443878368990149, "grad_norm": 1.2522500009295214, "learning_rate": 5.931052369739783e-06, "loss": 0.5232, "step": 18087 }, { "epoch": 0.644423464025509, "grad_norm": 1.6491571040619217, "learning_rate": 5.929998304461482e-06, "loss": 0.5229, "step": 18088 }, { "epoch": 0.6444590911520032, "grad_norm": 1.1880169706288823, "learning_rate": 5.928944293378101e-06, "loss": 0.4258, "step": 18089 }, { "epoch": 0.6444947182784972, "grad_norm": 1.161078101332024, "learning_rate": 5.92789033650367e-06, "loss": 0.3913, "step": 18090 }, { "epoch": 0.6445303454049913, "grad_norm": 1.5388446003797889, "learning_rate": 5.926836433852228e-06, "loss": 0.4153, "step": 18091 }, { "epoch": 0.6445659725314855, "grad_norm": 1.2454597765621147, "learning_rate": 5.9257825854378024e-06, "loss": 0.4605, "step": 18092 }, { "epoch": 0.6446015996579796, "grad_norm": 1.217817093727738, "learning_rate": 5.924728791274432e-06, "loss": 0.5129, "step": 18093 }, { "epoch": 0.6446372267844737, "grad_norm": 1.2019498669671498, "learning_rate": 5.923675051376144e-06, "loss": 0.4535, "step": 18094 }, { "epoch": 0.6446728539109678, "grad_norm": 1.8409996275656706, "learning_rate": 5.922621365756974e-06, "loss": 0.5352, "step": 18095 }, { "epoch": 0.6447084810374619, "grad_norm": 1.88753109455142, "learning_rate": 5.921567734430949e-06, "loss": 0.5509, "step": 18096 }, { "epoch": 0.644744108163956, "grad_norm": 1.2596115803455536, "learning_rate": 5.920514157412099e-06, "loss": 0.4573, "step": 18097 }, { "epoch": 0.6447797352904502, "grad_norm": 1.4922741880937371, "learning_rate": 5.919460634714459e-06, "loss": 0.5399, "step": 18098 }, { "epoch": 0.6448153624169443, "grad_norm": 1.6027058630161006, "learning_rate": 5.91840716635205e-06, "loss": 0.5609, "step": 18099 }, { "epoch": 0.6448509895434383, "grad_norm": 1.5624850807950372, "learning_rate": 5.917353752338902e-06, "loss": 0.5382, "step": 18100 }, { "epoch": 0.6448866166699325, "grad_norm": 1.134641491648463, "learning_rate": 5.916300392689042e-06, "loss": 0.4269, "step": 18101 }, { "epoch": 0.6449222437964266, "grad_norm": 1.1846158187705143, "learning_rate": 5.9152470874165004e-06, "loss": 0.4512, "step": 18102 }, { "epoch": 0.6449578709229207, "grad_norm": 1.4135560770913596, "learning_rate": 5.914193836535294e-06, "loss": 0.5506, "step": 18103 }, { "epoch": 0.6449934980494149, "grad_norm": 1.5356665080448575, "learning_rate": 5.913140640059456e-06, "loss": 0.5711, "step": 18104 }, { "epoch": 0.6450291251759089, "grad_norm": 1.3871706540181303, "learning_rate": 5.912087498003003e-06, "loss": 0.4975, "step": 18105 }, { "epoch": 0.645064752302403, "grad_norm": 1.1764364597763228, "learning_rate": 5.911034410379968e-06, "loss": 0.4218, "step": 18106 }, { "epoch": 0.6451003794288972, "grad_norm": 1.0961101519860803, "learning_rate": 5.909981377204361e-06, "loss": 0.4061, "step": 18107 }, { "epoch": 0.6451360065553913, "grad_norm": 1.2166018007357358, "learning_rate": 5.908928398490217e-06, "loss": 0.4336, "step": 18108 }, { "epoch": 0.6451716336818853, "grad_norm": 1.6578411484461288, "learning_rate": 5.907875474251546e-06, "loss": 0.5787, "step": 18109 }, { "epoch": 0.6452072608083795, "grad_norm": 1.1415742178471424, "learning_rate": 5.906822604502375e-06, "loss": 0.379, "step": 18110 }, { "epoch": 0.6452428879348736, "grad_norm": 1.5948696925092496, "learning_rate": 5.905769789256722e-06, "loss": 0.5818, "step": 18111 }, { "epoch": 0.6452785150613677, "grad_norm": 1.3200006671136169, "learning_rate": 5.904717028528606e-06, "loss": 0.4204, "step": 18112 }, { "epoch": 0.6453141421878619, "grad_norm": 1.6129627764467482, "learning_rate": 5.903664322332048e-06, "loss": 0.5643, "step": 18113 }, { "epoch": 0.645349769314356, "grad_norm": 1.141923450930049, "learning_rate": 5.902611670681058e-06, "loss": 0.4167, "step": 18114 }, { "epoch": 0.64538539644085, "grad_norm": 1.194511971817733, "learning_rate": 5.901559073589661e-06, "loss": 0.4181, "step": 18115 }, { "epoch": 0.6454210235673442, "grad_norm": 1.5310279981433608, "learning_rate": 5.900506531071867e-06, "loss": 0.5295, "step": 18116 }, { "epoch": 0.6454566506938383, "grad_norm": 1.1581784000649344, "learning_rate": 5.899454043141697e-06, "loss": 0.4363, "step": 18117 }, { "epoch": 0.6454922778203324, "grad_norm": 1.519478853596827, "learning_rate": 5.8984016098131605e-06, "loss": 0.506, "step": 18118 }, { "epoch": 0.6455279049468265, "grad_norm": 1.2840972979021863, "learning_rate": 5.897349231100273e-06, "loss": 0.396, "step": 18119 }, { "epoch": 0.6455635320733206, "grad_norm": 1.3845136435333563, "learning_rate": 5.896296907017049e-06, "loss": 0.4527, "step": 18120 }, { "epoch": 0.6455991591998147, "grad_norm": 1.198168284827385, "learning_rate": 5.895244637577498e-06, "loss": 0.4536, "step": 18121 }, { "epoch": 0.6456347863263089, "grad_norm": 1.1050293530105668, "learning_rate": 5.894192422795635e-06, "loss": 0.4838, "step": 18122 }, { "epoch": 0.645670413452803, "grad_norm": 1.5797489994626988, "learning_rate": 5.893140262685469e-06, "loss": 0.6182, "step": 18123 }, { "epoch": 0.645706040579297, "grad_norm": 1.4235115039613875, "learning_rate": 5.892088157261016e-06, "loss": 0.576, "step": 18124 }, { "epoch": 0.6457416677057912, "grad_norm": 1.0746239010132226, "learning_rate": 5.891036106536274e-06, "loss": 0.4096, "step": 18125 }, { "epoch": 0.6457772948322853, "grad_norm": 1.5899732107658615, "learning_rate": 5.889984110525262e-06, "loss": 0.4626, "step": 18126 }, { "epoch": 0.6458129219587794, "grad_norm": 1.1923802917525095, "learning_rate": 5.888932169241981e-06, "loss": 0.4301, "step": 18127 }, { "epoch": 0.6458485490852736, "grad_norm": 1.1391985292924358, "learning_rate": 5.8878802827004465e-06, "loss": 0.4707, "step": 18128 }, { "epoch": 0.6458841762117676, "grad_norm": 1.054824613524108, "learning_rate": 5.8868284509146564e-06, "loss": 0.4086, "step": 18129 }, { "epoch": 0.6459198033382617, "grad_norm": 1.8976901599848044, "learning_rate": 5.885776673898622e-06, "loss": 0.4546, "step": 18130 }, { "epoch": 0.6459554304647559, "grad_norm": 1.6929182344144165, "learning_rate": 5.884724951666345e-06, "loss": 0.529, "step": 18131 }, { "epoch": 0.64599105759125, "grad_norm": 1.7328298625454366, "learning_rate": 5.883673284231831e-06, "loss": 0.5123, "step": 18132 }, { "epoch": 0.646026684717744, "grad_norm": 3.0125224775796493, "learning_rate": 5.882621671609087e-06, "loss": 0.4246, "step": 18133 }, { "epoch": 0.6460623118442382, "grad_norm": 1.3773118073267399, "learning_rate": 5.8815701138121115e-06, "loss": 0.4368, "step": 18134 }, { "epoch": 0.6460979389707323, "grad_norm": 1.2831525492148628, "learning_rate": 5.8805186108549114e-06, "loss": 0.4251, "step": 18135 }, { "epoch": 0.6461335660972264, "grad_norm": 1.255243901799783, "learning_rate": 5.879467162751482e-06, "loss": 0.4599, "step": 18136 }, { "epoch": 0.6461691932237206, "grad_norm": 1.57634499230495, "learning_rate": 5.878415769515831e-06, "loss": 0.5344, "step": 18137 }, { "epoch": 0.6462048203502146, "grad_norm": 1.4477251645664242, "learning_rate": 5.877364431161952e-06, "loss": 0.4269, "step": 18138 }, { "epoch": 0.6462404474767087, "grad_norm": 5.100450578605403, "learning_rate": 5.876313147703849e-06, "loss": 0.5593, "step": 18139 }, { "epoch": 0.6462760746032029, "grad_norm": 1.7297888727598654, "learning_rate": 5.875261919155517e-06, "loss": 0.5375, "step": 18140 }, { "epoch": 0.646311701729697, "grad_norm": 1.1098730052241452, "learning_rate": 5.8742107455309574e-06, "loss": 0.3901, "step": 18141 }, { "epoch": 0.6463473288561911, "grad_norm": 1.7857191529342935, "learning_rate": 5.873159626844163e-06, "loss": 0.5215, "step": 18142 }, { "epoch": 0.6463829559826852, "grad_norm": 1.3215559869360742, "learning_rate": 5.872108563109132e-06, "loss": 0.4306, "step": 18143 }, { "epoch": 0.6464185831091793, "grad_norm": 1.350348368353454, "learning_rate": 5.871057554339865e-06, "loss": 0.4629, "step": 18144 }, { "epoch": 0.6464542102356734, "grad_norm": 1.5436409838698952, "learning_rate": 5.870006600550348e-06, "loss": 0.5117, "step": 18145 }, { "epoch": 0.6464898373621676, "grad_norm": 1.5400311811003364, "learning_rate": 5.868955701754584e-06, "loss": 0.4751, "step": 18146 }, { "epoch": 0.6465254644886617, "grad_norm": 1.331567496425032, "learning_rate": 5.86790485796656e-06, "loss": 0.422, "step": 18147 }, { "epoch": 0.6465610916151557, "grad_norm": 1.4495463885000974, "learning_rate": 5.8668540692002714e-06, "loss": 0.4972, "step": 18148 }, { "epoch": 0.6465967187416499, "grad_norm": 1.4183156624387474, "learning_rate": 5.865803335469708e-06, "loss": 0.6732, "step": 18149 }, { "epoch": 0.646632345868144, "grad_norm": 1.1831451228083238, "learning_rate": 5.8647526567888676e-06, "loss": 0.4542, "step": 18150 }, { "epoch": 0.6466679729946381, "grad_norm": 1.3292972342131986, "learning_rate": 5.863702033171732e-06, "loss": 0.5688, "step": 18151 }, { "epoch": 0.6467036001211323, "grad_norm": 1.6223836070648683, "learning_rate": 5.8626514646322965e-06, "loss": 0.5689, "step": 18152 }, { "epoch": 0.6467392272476263, "grad_norm": 1.0923840211221307, "learning_rate": 5.861600951184545e-06, "loss": 0.4173, "step": 18153 }, { "epoch": 0.6467748543741204, "grad_norm": 1.6333673850204249, "learning_rate": 5.860550492842476e-06, "loss": 0.3826, "step": 18154 }, { "epoch": 0.6468104815006146, "grad_norm": 1.3538248439635567, "learning_rate": 5.859500089620065e-06, "loss": 0.5385, "step": 18155 }, { "epoch": 0.6468461086271087, "grad_norm": 1.4180332103611113, "learning_rate": 5.858449741531303e-06, "loss": 0.5195, "step": 18156 }, { "epoch": 0.6468817357536027, "grad_norm": 1.1482812103587778, "learning_rate": 5.8573994485901835e-06, "loss": 0.4145, "step": 18157 }, { "epoch": 0.6469173628800969, "grad_norm": 2.0567982446333866, "learning_rate": 5.856349210810681e-06, "loss": 0.585, "step": 18158 }, { "epoch": 0.646952990006591, "grad_norm": 1.1075110794359773, "learning_rate": 5.855299028206788e-06, "loss": 0.4254, "step": 18159 }, { "epoch": 0.6469886171330851, "grad_norm": 1.550219961415485, "learning_rate": 5.854248900792483e-06, "loss": 0.5403, "step": 18160 }, { "epoch": 0.6470242442595793, "grad_norm": 1.3975836512990205, "learning_rate": 5.8531988285817555e-06, "loss": 0.5066, "step": 18161 }, { "epoch": 0.6470598713860733, "grad_norm": 1.5545957118568143, "learning_rate": 5.852148811588581e-06, "loss": 0.451, "step": 18162 }, { "epoch": 0.6470954985125674, "grad_norm": 1.553392217505892, "learning_rate": 5.851098849826944e-06, "loss": 0.6153, "step": 18163 }, { "epoch": 0.6471311256390616, "grad_norm": 1.2335452618828877, "learning_rate": 5.8500489433108265e-06, "loss": 0.4549, "step": 18164 }, { "epoch": 0.6471667527655557, "grad_norm": 1.736399489922387, "learning_rate": 5.848999092054211e-06, "loss": 0.5402, "step": 18165 }, { "epoch": 0.6472023798920498, "grad_norm": 1.8913782378164385, "learning_rate": 5.847949296071068e-06, "loss": 0.4944, "step": 18166 }, { "epoch": 0.647238007018544, "grad_norm": 1.26866052178632, "learning_rate": 5.846899555375385e-06, "loss": 0.4772, "step": 18167 }, { "epoch": 0.647273634145038, "grad_norm": 1.3662929005988247, "learning_rate": 5.845849869981137e-06, "loss": 0.5468, "step": 18168 }, { "epoch": 0.6473092612715321, "grad_norm": 1.1409790118952474, "learning_rate": 5.844800239902301e-06, "loss": 0.3555, "step": 18169 }, { "epoch": 0.6473448883980263, "grad_norm": 1.247880785029767, "learning_rate": 5.8437506651528555e-06, "loss": 0.4758, "step": 18170 }, { "epoch": 0.6473805155245204, "grad_norm": 1.2568887983237524, "learning_rate": 5.842701145746773e-06, "loss": 0.4243, "step": 18171 }, { "epoch": 0.6474161426510144, "grad_norm": 1.0665039360692585, "learning_rate": 5.8416516816980355e-06, "loss": 0.3856, "step": 18172 }, { "epoch": 0.6474517697775086, "grad_norm": 1.362677600885952, "learning_rate": 5.840602273020607e-06, "loss": 0.5123, "step": 18173 }, { "epoch": 0.6474873969040027, "grad_norm": 1.7157559435385241, "learning_rate": 5.839552919728469e-06, "loss": 0.5035, "step": 18174 }, { "epoch": 0.6475230240304968, "grad_norm": 1.1404869592154863, "learning_rate": 5.838503621835592e-06, "loss": 0.4877, "step": 18175 }, { "epoch": 0.647558651156991, "grad_norm": 1.185239001645769, "learning_rate": 5.8374543793559455e-06, "loss": 0.3985, "step": 18176 }, { "epoch": 0.647594278283485, "grad_norm": 1.3752654852577624, "learning_rate": 5.836405192303504e-06, "loss": 0.4421, "step": 18177 }, { "epoch": 0.6476299054099791, "grad_norm": 1.2341902009431456, "learning_rate": 5.835356060692236e-06, "loss": 0.4337, "step": 18178 }, { "epoch": 0.6476655325364733, "grad_norm": 1.3184260903041984, "learning_rate": 5.83430698453612e-06, "loss": 0.4071, "step": 18179 }, { "epoch": 0.6477011596629674, "grad_norm": 1.1588391611399844, "learning_rate": 5.833257963849111e-06, "loss": 0.4168, "step": 18180 }, { "epoch": 0.6477367867894614, "grad_norm": 1.3050876913792078, "learning_rate": 5.8322089986451906e-06, "loss": 0.4258, "step": 18181 }, { "epoch": 0.6477724139159556, "grad_norm": 1.4865717004350516, "learning_rate": 5.831160088938315e-06, "loss": 0.5372, "step": 18182 }, { "epoch": 0.6478080410424497, "grad_norm": 1.1528765625061927, "learning_rate": 5.830111234742459e-06, "loss": 0.4137, "step": 18183 }, { "epoch": 0.6478436681689438, "grad_norm": 1.7244158074248244, "learning_rate": 5.829062436071584e-06, "loss": 0.5474, "step": 18184 }, { "epoch": 0.647879295295438, "grad_norm": 1.1261300338452576, "learning_rate": 5.828013692939664e-06, "loss": 0.4204, "step": 18185 }, { "epoch": 0.647914922421932, "grad_norm": 1.8578657302520687, "learning_rate": 5.826965005360652e-06, "loss": 0.4852, "step": 18186 }, { "epoch": 0.6479505495484261, "grad_norm": 1.2786386475872673, "learning_rate": 5.825916373348524e-06, "loss": 0.385, "step": 18187 }, { "epoch": 0.6479861766749203, "grad_norm": 1.1969280422203108, "learning_rate": 5.82486779691723e-06, "loss": 0.4373, "step": 18188 }, { "epoch": 0.6480218038014144, "grad_norm": 1.3642199834500301, "learning_rate": 5.823819276080741e-06, "loss": 0.5128, "step": 18189 }, { "epoch": 0.6480574309279085, "grad_norm": 1.8442774069309276, "learning_rate": 5.8227708108530225e-06, "loss": 0.5887, "step": 18190 }, { "epoch": 0.6480930580544026, "grad_norm": 1.4845653309282092, "learning_rate": 5.821722401248025e-06, "loss": 0.5091, "step": 18191 }, { "epoch": 0.6481286851808967, "grad_norm": 1.455654972548027, "learning_rate": 5.820674047279713e-06, "loss": 0.4832, "step": 18192 }, { "epoch": 0.6481643123073908, "grad_norm": 1.1524763596285255, "learning_rate": 5.819625748962049e-06, "loss": 0.4602, "step": 18193 }, { "epoch": 0.648199939433885, "grad_norm": 1.3538966085689947, "learning_rate": 5.818577506308993e-06, "loss": 0.4419, "step": 18194 }, { "epoch": 0.6482355665603791, "grad_norm": 1.3618648720771653, "learning_rate": 5.8175293193344975e-06, "loss": 0.5233, "step": 18195 }, { "epoch": 0.6482711936868731, "grad_norm": 1.4676745096283714, "learning_rate": 5.8164811880525254e-06, "loss": 0.4781, "step": 18196 }, { "epoch": 0.6483068208133673, "grad_norm": 1.234138335155823, "learning_rate": 5.815433112477027e-06, "loss": 0.5309, "step": 18197 }, { "epoch": 0.6483424479398614, "grad_norm": 1.2117680548358112, "learning_rate": 5.814385092621961e-06, "loss": 0.4159, "step": 18198 }, { "epoch": 0.6483780750663555, "grad_norm": 1.15519575734063, "learning_rate": 5.813337128501284e-06, "loss": 0.4237, "step": 18199 }, { "epoch": 0.6484137021928497, "grad_norm": 1.4116292965176014, "learning_rate": 5.812289220128953e-06, "loss": 0.5282, "step": 18200 }, { "epoch": 0.6484493293193437, "grad_norm": 1.6858314592682462, "learning_rate": 5.811241367518914e-06, "loss": 0.4958, "step": 18201 }, { "epoch": 0.6484849564458378, "grad_norm": 2.5315846047134563, "learning_rate": 5.8101935706851235e-06, "loss": 0.5777, "step": 18202 }, { "epoch": 0.648520583572332, "grad_norm": 1.2821718119259187, "learning_rate": 5.809145829641538e-06, "loss": 0.4702, "step": 18203 }, { "epoch": 0.6485562106988261, "grad_norm": 1.880349331415296, "learning_rate": 5.808098144402101e-06, "loss": 0.6078, "step": 18204 }, { "epoch": 0.6485918378253201, "grad_norm": 1.1136450777888849, "learning_rate": 5.807050514980772e-06, "loss": 0.438, "step": 18205 }, { "epoch": 0.6486274649518143, "grad_norm": 1.2632765439731855, "learning_rate": 5.8060029413914905e-06, "loss": 0.4285, "step": 18206 }, { "epoch": 0.6486630920783084, "grad_norm": 1.1194678744440247, "learning_rate": 5.804955423648213e-06, "loss": 0.3979, "step": 18207 }, { "epoch": 0.6486987192048025, "grad_norm": 1.5146102230980518, "learning_rate": 5.803907961764886e-06, "loss": 0.574, "step": 18208 }, { "epoch": 0.6487343463312967, "grad_norm": 1.1580142445124813, "learning_rate": 5.80286055575546e-06, "loss": 0.4125, "step": 18209 }, { "epoch": 0.6487699734577907, "grad_norm": 1.5117870131017905, "learning_rate": 5.8018132056338776e-06, "loss": 0.4718, "step": 18210 }, { "epoch": 0.6488056005842848, "grad_norm": 1.4536723704583736, "learning_rate": 5.80076591141409e-06, "loss": 0.4606, "step": 18211 }, { "epoch": 0.648841227710779, "grad_norm": 1.4750555353866601, "learning_rate": 5.7997186731100355e-06, "loss": 0.472, "step": 18212 }, { "epoch": 0.6488768548372731, "grad_norm": 1.2463573722218677, "learning_rate": 5.7986714907356614e-06, "loss": 0.4399, "step": 18213 }, { "epoch": 0.6489124819637672, "grad_norm": 1.14828239524102, "learning_rate": 5.7976243643049124e-06, "loss": 0.4636, "step": 18214 }, { "epoch": 0.6489481090902613, "grad_norm": 1.572380126441876, "learning_rate": 5.796577293831734e-06, "loss": 0.5777, "step": 18215 }, { "epoch": 0.6489837362167554, "grad_norm": 1.4933358997181618, "learning_rate": 5.79553027933007e-06, "loss": 0.5993, "step": 18216 }, { "epoch": 0.6490193633432495, "grad_norm": 1.5924512510374182, "learning_rate": 5.794483320813856e-06, "loss": 0.5125, "step": 18217 }, { "epoch": 0.6490549904697437, "grad_norm": 1.4379565112638504, "learning_rate": 5.79343641829704e-06, "loss": 0.4646, "step": 18218 }, { "epoch": 0.6490906175962378, "grad_norm": 1.472045381045377, "learning_rate": 5.792389571793553e-06, "loss": 0.506, "step": 18219 }, { "epoch": 0.6491262447227318, "grad_norm": 1.1713311644943594, "learning_rate": 5.791342781317345e-06, "loss": 0.4423, "step": 18220 }, { "epoch": 0.649161871849226, "grad_norm": 1.243720227019355, "learning_rate": 5.790296046882342e-06, "loss": 0.4369, "step": 18221 }, { "epoch": 0.6491974989757201, "grad_norm": 1.0848645065088105, "learning_rate": 5.789249368502496e-06, "loss": 0.3985, "step": 18222 }, { "epoch": 0.6492331261022142, "grad_norm": 1.2088711584959841, "learning_rate": 5.788202746191735e-06, "loss": 0.4193, "step": 18223 }, { "epoch": 0.6492687532287084, "grad_norm": 1.208560579639405, "learning_rate": 5.787156179963998e-06, "loss": 0.4698, "step": 18224 }, { "epoch": 0.6493043803552024, "grad_norm": 1.2333065467143152, "learning_rate": 5.786109669833224e-06, "loss": 0.4905, "step": 18225 }, { "epoch": 0.6493400074816965, "grad_norm": 1.0288931204185785, "learning_rate": 5.785063215813342e-06, "loss": 0.3595, "step": 18226 }, { "epoch": 0.6493756346081907, "grad_norm": 1.2453675903045012, "learning_rate": 5.784016817918294e-06, "loss": 0.4239, "step": 18227 }, { "epoch": 0.6494112617346848, "grad_norm": 1.60985080938222, "learning_rate": 5.782970476162004e-06, "loss": 0.5325, "step": 18228 }, { "epoch": 0.6494468888611789, "grad_norm": 1.2243486280872147, "learning_rate": 5.78192419055841e-06, "loss": 0.4441, "step": 18229 }, { "epoch": 0.649482515987673, "grad_norm": 1.2268841912128279, "learning_rate": 5.7808779611214425e-06, "loss": 0.4228, "step": 18230 }, { "epoch": 0.6495181431141671, "grad_norm": 1.1733664241938127, "learning_rate": 5.77983178786504e-06, "loss": 0.4351, "step": 18231 }, { "epoch": 0.6495537702406612, "grad_norm": 1.6280266499482867, "learning_rate": 5.778785670803122e-06, "loss": 0.5047, "step": 18232 }, { "epoch": 0.6495893973671554, "grad_norm": 1.1508554433622173, "learning_rate": 5.777739609949626e-06, "loss": 0.427, "step": 18233 }, { "epoch": 0.6496250244936495, "grad_norm": 1.776689683349941, "learning_rate": 5.7766936053184764e-06, "loss": 0.5701, "step": 18234 }, { "epoch": 0.6496606516201435, "grad_norm": 1.3165804513693085, "learning_rate": 5.775647656923602e-06, "loss": 0.404, "step": 18235 }, { "epoch": 0.6496962787466377, "grad_norm": 1.3303262422761242, "learning_rate": 5.774601764778933e-06, "loss": 0.4224, "step": 18236 }, { "epoch": 0.6497319058731318, "grad_norm": 1.5886744887206503, "learning_rate": 5.773555928898393e-06, "loss": 0.5188, "step": 18237 }, { "epoch": 0.6497675329996259, "grad_norm": 1.467444050884767, "learning_rate": 5.772510149295916e-06, "loss": 0.4887, "step": 18238 }, { "epoch": 0.64980316012612, "grad_norm": 1.2378035994230814, "learning_rate": 5.771464425985416e-06, "loss": 0.4053, "step": 18239 }, { "epoch": 0.6498387872526141, "grad_norm": 1.0967291632905778, "learning_rate": 5.770418758980827e-06, "loss": 0.4185, "step": 18240 }, { "epoch": 0.6498744143791082, "grad_norm": 1.7508016888715683, "learning_rate": 5.769373148296063e-06, "loss": 0.5228, "step": 18241 }, { "epoch": 0.6499100415056024, "grad_norm": 1.1973930842672162, "learning_rate": 5.7683275939450575e-06, "loss": 0.4326, "step": 18242 }, { "epoch": 0.6499456686320965, "grad_norm": 1.609251098788361, "learning_rate": 5.767282095941725e-06, "loss": 0.4888, "step": 18243 }, { "epoch": 0.6499812957585905, "grad_norm": 1.1042632539856734, "learning_rate": 5.7662366542999896e-06, "loss": 0.4114, "step": 18244 }, { "epoch": 0.6500169228850847, "grad_norm": 1.9517554615409818, "learning_rate": 5.765191269033772e-06, "loss": 0.5475, "step": 18245 }, { "epoch": 0.6500525500115788, "grad_norm": 1.7938453141386699, "learning_rate": 5.764145940156996e-06, "loss": 0.5677, "step": 18246 }, { "epoch": 0.6500881771380729, "grad_norm": 1.1687046074387528, "learning_rate": 5.763100667683574e-06, "loss": 0.3995, "step": 18247 }, { "epoch": 0.6501238042645671, "grad_norm": 1.2635490165530707, "learning_rate": 5.762055451627428e-06, "loss": 0.4363, "step": 18248 }, { "epoch": 0.6501594313910611, "grad_norm": 1.753099010789798, "learning_rate": 5.76101029200248e-06, "loss": 0.5303, "step": 18249 }, { "epoch": 0.6501950585175552, "grad_norm": 1.9003366169957292, "learning_rate": 5.759965188822638e-06, "loss": 0.4863, "step": 18250 }, { "epoch": 0.6502306856440494, "grad_norm": 1.1204092991381676, "learning_rate": 5.758920142101825e-06, "loss": 0.4216, "step": 18251 }, { "epoch": 0.6502663127705435, "grad_norm": 1.2908315313399374, "learning_rate": 5.757875151853953e-06, "loss": 0.4163, "step": 18252 }, { "epoch": 0.6503019398970376, "grad_norm": 1.1200530602073449, "learning_rate": 5.756830218092942e-06, "loss": 0.4372, "step": 18253 }, { "epoch": 0.6503375670235317, "grad_norm": 1.4630707255910902, "learning_rate": 5.7557853408327e-06, "loss": 0.4972, "step": 18254 }, { "epoch": 0.6503731941500258, "grad_norm": 1.303712921651174, "learning_rate": 5.754740520087145e-06, "loss": 0.4913, "step": 18255 }, { "epoch": 0.6504088212765199, "grad_norm": 3.8300185515113294, "learning_rate": 5.753695755870185e-06, "loss": 0.5008, "step": 18256 }, { "epoch": 0.6504444484030141, "grad_norm": 1.2677218141515016, "learning_rate": 5.752651048195738e-06, "loss": 0.5076, "step": 18257 }, { "epoch": 0.6504800755295082, "grad_norm": 1.1709057984445155, "learning_rate": 5.751606397077703e-06, "loss": 0.4825, "step": 18258 }, { "epoch": 0.6505157026560022, "grad_norm": 1.1172377349338942, "learning_rate": 5.75056180253e-06, "loss": 0.4222, "step": 18259 }, { "epoch": 0.6505513297824964, "grad_norm": 1.7685461544986552, "learning_rate": 5.749517264566542e-06, "loss": 0.4719, "step": 18260 }, { "epoch": 0.6505869569089905, "grad_norm": 1.700382747105966, "learning_rate": 5.748472783201229e-06, "loss": 0.5785, "step": 18261 }, { "epoch": 0.6506225840354846, "grad_norm": 1.5196363874523617, "learning_rate": 5.747428358447977e-06, "loss": 0.5473, "step": 18262 }, { "epoch": 0.6506582111619788, "grad_norm": 1.9935325184747599, "learning_rate": 5.7463839903206844e-06, "loss": 0.4873, "step": 18263 }, { "epoch": 0.6506938382884728, "grad_norm": 1.283317619579702, "learning_rate": 5.745339678833266e-06, "loss": 0.3923, "step": 18264 }, { "epoch": 0.6507294654149669, "grad_norm": 1.2206403007935338, "learning_rate": 5.744295423999619e-06, "loss": 0.415, "step": 18265 }, { "epoch": 0.6507650925414611, "grad_norm": 1.1271685532142512, "learning_rate": 5.743251225833653e-06, "loss": 0.3832, "step": 18266 }, { "epoch": 0.6508007196679552, "grad_norm": 1.3556834311422392, "learning_rate": 5.742207084349274e-06, "loss": 0.4565, "step": 18267 }, { "epoch": 0.6508363467944492, "grad_norm": 1.1878622028556405, "learning_rate": 5.7411629995603855e-06, "loss": 0.4296, "step": 18268 }, { "epoch": 0.6508719739209434, "grad_norm": 1.680112428551218, "learning_rate": 5.740118971480886e-06, "loss": 0.5026, "step": 18269 }, { "epoch": 0.6509076010474375, "grad_norm": 1.456483218530932, "learning_rate": 5.739075000124679e-06, "loss": 0.4288, "step": 18270 }, { "epoch": 0.6509432281739316, "grad_norm": 1.125985742303259, "learning_rate": 5.73803108550567e-06, "loss": 0.4149, "step": 18271 }, { "epoch": 0.6509788553004258, "grad_norm": 1.3835826273208778, "learning_rate": 5.736987227637752e-06, "loss": 0.6243, "step": 18272 }, { "epoch": 0.6510144824269198, "grad_norm": 1.2962049404401685, "learning_rate": 5.73594342653483e-06, "loss": 0.466, "step": 18273 }, { "epoch": 0.651050109553414, "grad_norm": 1.4617531057893516, "learning_rate": 5.734899682210801e-06, "loss": 0.554, "step": 18274 }, { "epoch": 0.6510857366799081, "grad_norm": 1.2667746730098364, "learning_rate": 5.733855994679569e-06, "loss": 0.4384, "step": 18275 }, { "epoch": 0.6511213638064022, "grad_norm": 1.5133849919184519, "learning_rate": 5.732812363955021e-06, "loss": 0.5435, "step": 18276 }, { "epoch": 0.6511569909328964, "grad_norm": 1.1894573297647009, "learning_rate": 5.731768790051062e-06, "loss": 0.4369, "step": 18277 }, { "epoch": 0.6511926180593904, "grad_norm": 2.4316734367365056, "learning_rate": 5.7307252729815835e-06, "loss": 0.5229, "step": 18278 }, { "epoch": 0.6512282451858845, "grad_norm": 1.5958378104030084, "learning_rate": 5.729681812760484e-06, "loss": 0.5768, "step": 18279 }, { "epoch": 0.6512638723123787, "grad_norm": 1.2256659624305557, "learning_rate": 5.728638409401652e-06, "loss": 0.392, "step": 18280 }, { "epoch": 0.6512994994388728, "grad_norm": 1.4325002998761962, "learning_rate": 5.727595062918987e-06, "loss": 0.4748, "step": 18281 }, { "epoch": 0.6513351265653669, "grad_norm": 1.495617819303647, "learning_rate": 5.7265517733263785e-06, "loss": 0.5083, "step": 18282 }, { "epoch": 0.651370753691861, "grad_norm": 1.250162603761243, "learning_rate": 5.72550854063772e-06, "loss": 0.4265, "step": 18283 }, { "epoch": 0.6514063808183551, "grad_norm": 1.6193413312277478, "learning_rate": 5.724465364866907e-06, "loss": 0.5348, "step": 18284 }, { "epoch": 0.6514420079448492, "grad_norm": 1.352834329296292, "learning_rate": 5.723422246027822e-06, "loss": 0.5057, "step": 18285 }, { "epoch": 0.6514776350713434, "grad_norm": 1.266924688502637, "learning_rate": 5.7223791841343635e-06, "loss": 0.36, "step": 18286 }, { "epoch": 0.6515132621978375, "grad_norm": 1.580912074168042, "learning_rate": 5.721336179200412e-06, "loss": 0.5109, "step": 18287 }, { "epoch": 0.6515488893243315, "grad_norm": 1.3637854088334793, "learning_rate": 5.720293231239859e-06, "loss": 0.4246, "step": 18288 }, { "epoch": 0.6515845164508257, "grad_norm": 1.1463656532887518, "learning_rate": 5.719250340266595e-06, "loss": 0.3874, "step": 18289 }, { "epoch": 0.6516201435773198, "grad_norm": 1.2303384581670216, "learning_rate": 5.718207506294506e-06, "loss": 0.4213, "step": 18290 }, { "epoch": 0.6516557707038139, "grad_norm": 1.6914822754910521, "learning_rate": 5.717164729337474e-06, "loss": 0.5368, "step": 18291 }, { "epoch": 0.651691397830308, "grad_norm": 1.752877095676758, "learning_rate": 5.716122009409393e-06, "loss": 0.6207, "step": 18292 }, { "epoch": 0.6517270249568021, "grad_norm": 1.5821830429512744, "learning_rate": 5.715079346524135e-06, "loss": 0.4853, "step": 18293 }, { "epoch": 0.6517626520832962, "grad_norm": 1.453030032930966, "learning_rate": 5.7140367406955925e-06, "loss": 0.4659, "step": 18294 }, { "epoch": 0.6517982792097904, "grad_norm": 1.2310147439143164, "learning_rate": 5.7129941919376485e-06, "loss": 0.4066, "step": 18295 }, { "epoch": 0.6518339063362845, "grad_norm": 1.5706367401810222, "learning_rate": 5.711951700264176e-06, "loss": 0.4977, "step": 18296 }, { "epoch": 0.6518695334627785, "grad_norm": 1.5439042850571636, "learning_rate": 5.710909265689073e-06, "loss": 0.447, "step": 18297 }, { "epoch": 0.6519051605892727, "grad_norm": 1.1991339294588845, "learning_rate": 5.709866888226206e-06, "loss": 0.4149, "step": 18298 }, { "epoch": 0.6519407877157668, "grad_norm": 1.5599065904035059, "learning_rate": 5.7088245678894635e-06, "loss": 0.4854, "step": 18299 }, { "epoch": 0.6519764148422609, "grad_norm": 1.2098022624335891, "learning_rate": 5.70778230469272e-06, "loss": 0.4159, "step": 18300 }, { "epoch": 0.6520120419687551, "grad_norm": 1.4696942081356008, "learning_rate": 5.706740098649857e-06, "loss": 0.4245, "step": 18301 }, { "epoch": 0.6520476690952491, "grad_norm": 1.2901255079448826, "learning_rate": 5.705697949774746e-06, "loss": 0.4377, "step": 18302 }, { "epoch": 0.6520832962217432, "grad_norm": 1.371519887922527, "learning_rate": 5.704655858081268e-06, "loss": 0.5525, "step": 18303 }, { "epoch": 0.6521189233482374, "grad_norm": 1.3753547083021653, "learning_rate": 5.7036138235833005e-06, "loss": 0.4482, "step": 18304 }, { "epoch": 0.6521545504747315, "grad_norm": 1.1399290067680026, "learning_rate": 5.7025718462947175e-06, "loss": 0.3869, "step": 18305 }, { "epoch": 0.6521901776012256, "grad_norm": 1.3984182854704246, "learning_rate": 5.7015299262293984e-06, "loss": 0.5169, "step": 18306 }, { "epoch": 0.6522258047277197, "grad_norm": 1.5583577244759472, "learning_rate": 5.700488063401209e-06, "loss": 0.5483, "step": 18307 }, { "epoch": 0.6522614318542138, "grad_norm": 1.2885652084400405, "learning_rate": 5.6994462578240305e-06, "loss": 0.4797, "step": 18308 }, { "epoch": 0.6522970589807079, "grad_norm": 1.1954709912847252, "learning_rate": 5.698404509511726e-06, "loss": 0.4178, "step": 18309 }, { "epoch": 0.6523326861072021, "grad_norm": 1.162298380281653, "learning_rate": 5.697362818478172e-06, "loss": 0.4096, "step": 18310 }, { "epoch": 0.6523683132336962, "grad_norm": 1.7415534233334447, "learning_rate": 5.696321184737241e-06, "loss": 0.5384, "step": 18311 }, { "epoch": 0.6524039403601902, "grad_norm": 1.379538151270146, "learning_rate": 5.695279608302804e-06, "loss": 0.5394, "step": 18312 }, { "epoch": 0.6524395674866844, "grad_norm": 1.169653962176157, "learning_rate": 5.694238089188724e-06, "loss": 0.3994, "step": 18313 }, { "epoch": 0.6524751946131785, "grad_norm": 1.7649590482992696, "learning_rate": 5.693196627408879e-06, "loss": 0.6249, "step": 18314 }, { "epoch": 0.6525108217396726, "grad_norm": 1.2379802222904843, "learning_rate": 5.692155222977125e-06, "loss": 0.4278, "step": 18315 }, { "epoch": 0.6525464488661668, "grad_norm": 1.6409695943599902, "learning_rate": 5.691113875907337e-06, "loss": 0.4597, "step": 18316 }, { "epoch": 0.6525820759926608, "grad_norm": 2.0364212096757677, "learning_rate": 5.690072586213382e-06, "loss": 0.5047, "step": 18317 }, { "epoch": 0.6526177031191549, "grad_norm": 1.331226066089957, "learning_rate": 5.68903135390912e-06, "loss": 0.4351, "step": 18318 }, { "epoch": 0.6526533302456491, "grad_norm": 1.2685167218122662, "learning_rate": 5.687990179008418e-06, "loss": 0.4357, "step": 18319 }, { "epoch": 0.6526889573721432, "grad_norm": 1.564746675210876, "learning_rate": 5.68694906152514e-06, "loss": 0.5482, "step": 18320 }, { "epoch": 0.6527245844986372, "grad_norm": 1.0899051027767563, "learning_rate": 5.685908001473155e-06, "loss": 0.3922, "step": 18321 }, { "epoch": 0.6527602116251314, "grad_norm": 1.1137307280654176, "learning_rate": 5.684866998866316e-06, "loss": 0.33, "step": 18322 }, { "epoch": 0.6527958387516255, "grad_norm": 1.8537110287431549, "learning_rate": 5.683826053718493e-06, "loss": 0.5827, "step": 18323 }, { "epoch": 0.6528314658781196, "grad_norm": 1.1265641270040263, "learning_rate": 5.682785166043541e-06, "loss": 0.4039, "step": 18324 }, { "epoch": 0.6528670930046138, "grad_norm": 1.5839558604799873, "learning_rate": 5.681744335855319e-06, "loss": 0.4263, "step": 18325 }, { "epoch": 0.6529027201311078, "grad_norm": 1.6328668982466594, "learning_rate": 5.680703563167691e-06, "loss": 0.5952, "step": 18326 }, { "epoch": 0.6529383472576019, "grad_norm": 1.2757287130074944, "learning_rate": 5.679662847994518e-06, "loss": 0.4145, "step": 18327 }, { "epoch": 0.6529739743840961, "grad_norm": 1.4563274972586708, "learning_rate": 5.67862219034965e-06, "loss": 0.5353, "step": 18328 }, { "epoch": 0.6530096015105902, "grad_norm": 1.3830810487505443, "learning_rate": 5.6775815902469465e-06, "loss": 0.4246, "step": 18329 }, { "epoch": 0.6530452286370843, "grad_norm": 4.717399878788894, "learning_rate": 5.67654104770027e-06, "loss": 0.5326, "step": 18330 }, { "epoch": 0.6530808557635784, "grad_norm": 1.3585674340521374, "learning_rate": 5.675500562723468e-06, "loss": 0.5095, "step": 18331 }, { "epoch": 0.6531164828900725, "grad_norm": 1.4257835238559438, "learning_rate": 5.674460135330403e-06, "loss": 0.4887, "step": 18332 }, { "epoch": 0.6531521100165666, "grad_norm": 1.3785028258787837, "learning_rate": 5.673419765534915e-06, "loss": 0.4373, "step": 18333 }, { "epoch": 0.6531877371430608, "grad_norm": 1.3340158466709915, "learning_rate": 5.672379453350877e-06, "loss": 0.4222, "step": 18334 }, { "epoch": 0.6532233642695549, "grad_norm": 1.5780012752474422, "learning_rate": 5.671339198792126e-06, "loss": 0.4728, "step": 18335 }, { "epoch": 0.6532589913960489, "grad_norm": 1.4154828260194772, "learning_rate": 5.670299001872523e-06, "loss": 0.4566, "step": 18336 }, { "epoch": 0.6532946185225431, "grad_norm": 1.1878262130750967, "learning_rate": 5.669258862605913e-06, "loss": 0.4249, "step": 18337 }, { "epoch": 0.6533302456490372, "grad_norm": 1.9330592867717407, "learning_rate": 5.66821878100615e-06, "loss": 0.5708, "step": 18338 }, { "epoch": 0.6533658727755313, "grad_norm": 1.2574947441531164, "learning_rate": 5.66717875708708e-06, "loss": 0.4567, "step": 18339 }, { "epoch": 0.6534014999020255, "grad_norm": 1.1703111416865557, "learning_rate": 5.66613879086255e-06, "loss": 0.4472, "step": 18340 }, { "epoch": 0.6534371270285195, "grad_norm": 1.6765946905289235, "learning_rate": 5.665098882346414e-06, "loss": 0.4457, "step": 18341 }, { "epoch": 0.6534727541550136, "grad_norm": 1.236430242728098, "learning_rate": 5.664059031552516e-06, "loss": 0.5036, "step": 18342 }, { "epoch": 0.6535083812815078, "grad_norm": 1.276349874074527, "learning_rate": 5.663019238494705e-06, "loss": 0.4399, "step": 18343 }, { "epoch": 0.6535440084080019, "grad_norm": 1.2378640011524968, "learning_rate": 5.661979503186821e-06, "loss": 0.4311, "step": 18344 }, { "epoch": 0.6535796355344959, "grad_norm": 1.647785329841711, "learning_rate": 5.660939825642716e-06, "loss": 0.523, "step": 18345 }, { "epoch": 0.6536152626609901, "grad_norm": 1.6405448352600371, "learning_rate": 5.659900205876226e-06, "loss": 0.5262, "step": 18346 }, { "epoch": 0.6536508897874842, "grad_norm": 1.2122481039572521, "learning_rate": 5.658860643901199e-06, "loss": 0.3983, "step": 18347 }, { "epoch": 0.6536865169139783, "grad_norm": 1.3654413481216408, "learning_rate": 5.6578211397314765e-06, "loss": 0.3884, "step": 18348 }, { "epoch": 0.6537221440404725, "grad_norm": 1.266707017586804, "learning_rate": 5.656781693380904e-06, "loss": 0.3932, "step": 18349 }, { "epoch": 0.6537577711669665, "grad_norm": 1.5385195482925007, "learning_rate": 5.6557423048633155e-06, "loss": 0.5596, "step": 18350 }, { "epoch": 0.6537933982934606, "grad_norm": 1.2139009192313546, "learning_rate": 5.654702974192554e-06, "loss": 0.4024, "step": 18351 }, { "epoch": 0.6538290254199548, "grad_norm": 1.5531992757124506, "learning_rate": 5.653663701382464e-06, "loss": 0.5665, "step": 18352 }, { "epoch": 0.6538646525464489, "grad_norm": 1.2850567338672485, "learning_rate": 5.652624486446876e-06, "loss": 0.4648, "step": 18353 }, { "epoch": 0.653900279672943, "grad_norm": 1.3158182331643715, "learning_rate": 5.651585329399636e-06, "loss": 0.4763, "step": 18354 }, { "epoch": 0.6539359067994371, "grad_norm": 1.2178914449015252, "learning_rate": 5.650546230254572e-06, "loss": 0.4702, "step": 18355 }, { "epoch": 0.6539715339259312, "grad_norm": 1.590504264663933, "learning_rate": 5.649507189025525e-06, "loss": 0.5792, "step": 18356 }, { "epoch": 0.6540071610524253, "grad_norm": 1.429973937009545, "learning_rate": 5.648468205726331e-06, "loss": 0.5111, "step": 18357 }, { "epoch": 0.6540427881789195, "grad_norm": 1.2125968236722204, "learning_rate": 5.647429280370828e-06, "loss": 0.404, "step": 18358 }, { "epoch": 0.6540784153054136, "grad_norm": 1.0839378861255757, "learning_rate": 5.646390412972842e-06, "loss": 0.4443, "step": 18359 }, { "epoch": 0.6541140424319076, "grad_norm": 1.2260991891952533, "learning_rate": 5.645351603546216e-06, "loss": 0.3921, "step": 18360 }, { "epoch": 0.6541496695584018, "grad_norm": 1.4032720152606186, "learning_rate": 5.644312852104772e-06, "loss": 0.545, "step": 18361 }, { "epoch": 0.6541852966848959, "grad_norm": 1.1103100845538325, "learning_rate": 5.643274158662346e-06, "loss": 0.3783, "step": 18362 }, { "epoch": 0.65422092381139, "grad_norm": 1.3251861249786783, "learning_rate": 5.642235523232772e-06, "loss": 0.385, "step": 18363 }, { "epoch": 0.6542565509378842, "grad_norm": 1.4981147953673333, "learning_rate": 5.641196945829877e-06, "loss": 0.4915, "step": 18364 }, { "epoch": 0.6542921780643782, "grad_norm": 1.227473966708027, "learning_rate": 5.640158426467495e-06, "loss": 0.4227, "step": 18365 }, { "epoch": 0.6543278051908723, "grad_norm": 2.156801062446599, "learning_rate": 5.639119965159447e-06, "loss": 0.5594, "step": 18366 }, { "epoch": 0.6543634323173665, "grad_norm": 1.4835016187556735, "learning_rate": 5.638081561919568e-06, "loss": 0.5195, "step": 18367 }, { "epoch": 0.6543990594438606, "grad_norm": 1.3220927744639295, "learning_rate": 5.637043216761679e-06, "loss": 0.4514, "step": 18368 }, { "epoch": 0.6544346865703546, "grad_norm": 1.1861057984452614, "learning_rate": 5.636004929699612e-06, "loss": 0.4332, "step": 18369 }, { "epoch": 0.6544703136968488, "grad_norm": 1.462860536277027, "learning_rate": 5.6349667007471866e-06, "loss": 0.5064, "step": 18370 }, { "epoch": 0.6545059408233429, "grad_norm": 1.5964270960787166, "learning_rate": 5.633928529918229e-06, "loss": 0.4997, "step": 18371 }, { "epoch": 0.654541567949837, "grad_norm": 1.4354735154199871, "learning_rate": 5.632890417226566e-06, "loss": 0.5056, "step": 18372 }, { "epoch": 0.6545771950763312, "grad_norm": 1.3221267760335085, "learning_rate": 5.631852362686023e-06, "loss": 0.44, "step": 18373 }, { "epoch": 0.6546128222028252, "grad_norm": 1.4056461634092958, "learning_rate": 5.6308143663104166e-06, "loss": 0.5326, "step": 18374 }, { "epoch": 0.6546484493293193, "grad_norm": 1.3644000341740816, "learning_rate": 5.629776428113569e-06, "loss": 0.4073, "step": 18375 }, { "epoch": 0.6546840764558135, "grad_norm": 1.6178885635445963, "learning_rate": 5.628738548109307e-06, "loss": 0.4661, "step": 18376 }, { "epoch": 0.6547197035823076, "grad_norm": 1.1950278066732016, "learning_rate": 5.627700726311444e-06, "loss": 0.3941, "step": 18377 }, { "epoch": 0.6547553307088017, "grad_norm": 1.594563757778149, "learning_rate": 5.6266629627338e-06, "loss": 0.5022, "step": 18378 }, { "epoch": 0.6547909578352958, "grad_norm": 1.5880089576741998, "learning_rate": 5.625625257390197e-06, "loss": 0.5656, "step": 18379 }, { "epoch": 0.6548265849617899, "grad_norm": 1.4740660628187952, "learning_rate": 5.624587610294454e-06, "loss": 0.5003, "step": 18380 }, { "epoch": 0.654862212088284, "grad_norm": 1.8806305364733815, "learning_rate": 5.623550021460382e-06, "loss": 0.5871, "step": 18381 }, { "epoch": 0.6548978392147782, "grad_norm": 1.4630751551472745, "learning_rate": 5.622512490901804e-06, "loss": 0.4976, "step": 18382 }, { "epoch": 0.6549334663412723, "grad_norm": 1.8163446511092127, "learning_rate": 5.621475018632527e-06, "loss": 0.5665, "step": 18383 }, { "epoch": 0.6549690934677663, "grad_norm": 1.1863378974583614, "learning_rate": 5.620437604666375e-06, "loss": 0.4523, "step": 18384 }, { "epoch": 0.6550047205942605, "grad_norm": 1.3389482502870773, "learning_rate": 5.619400249017151e-06, "loss": 0.3847, "step": 18385 }, { "epoch": 0.6550403477207546, "grad_norm": 1.1784023586392176, "learning_rate": 5.618362951698677e-06, "loss": 0.4512, "step": 18386 }, { "epoch": 0.6550759748472487, "grad_norm": 1.2047581565241015, "learning_rate": 5.6173257127247685e-06, "loss": 0.4452, "step": 18387 }, { "epoch": 0.6551116019737429, "grad_norm": 1.3468514179751472, "learning_rate": 5.616288532109225e-06, "loss": 0.4764, "step": 18388 }, { "epoch": 0.6551472291002369, "grad_norm": 1.6343469630565575, "learning_rate": 5.61525140986587e-06, "loss": 0.5129, "step": 18389 }, { "epoch": 0.655182856226731, "grad_norm": 1.5497818761588127, "learning_rate": 5.614214346008502e-06, "loss": 0.6173, "step": 18390 }, { "epoch": 0.6552184833532252, "grad_norm": 1.148881070906254, "learning_rate": 5.613177340550939e-06, "loss": 0.4205, "step": 18391 }, { "epoch": 0.6552541104797193, "grad_norm": 1.4858199892373969, "learning_rate": 5.612140393506982e-06, "loss": 0.5217, "step": 18392 }, { "epoch": 0.6552897376062133, "grad_norm": 1.2688203495220125, "learning_rate": 5.611103504890444e-06, "loss": 0.4209, "step": 18393 }, { "epoch": 0.6553253647327075, "grad_norm": 1.416153030476695, "learning_rate": 5.610066674715129e-06, "loss": 0.4743, "step": 18394 }, { "epoch": 0.6553609918592016, "grad_norm": 1.1970503985703136, "learning_rate": 5.609029902994849e-06, "loss": 0.3907, "step": 18395 }, { "epoch": 0.6553966189856957, "grad_norm": 1.1001908775518376, "learning_rate": 5.607993189743401e-06, "loss": 0.3865, "step": 18396 }, { "epoch": 0.6554322461121899, "grad_norm": 2.0437326501675948, "learning_rate": 5.606956534974594e-06, "loss": 0.421, "step": 18397 }, { "epoch": 0.6554678732386839, "grad_norm": 1.2971138146767052, "learning_rate": 5.605919938702235e-06, "loss": 0.4545, "step": 18398 }, { "epoch": 0.655503500365178, "grad_norm": 1.5096993365901066, "learning_rate": 5.6048834009401196e-06, "loss": 0.5442, "step": 18399 }, { "epoch": 0.6555391274916722, "grad_norm": 1.417734306982996, "learning_rate": 5.603846921702054e-06, "loss": 0.5184, "step": 18400 }, { "epoch": 0.6555747546181663, "grad_norm": 1.0772862980791704, "learning_rate": 5.602810501001839e-06, "loss": 0.3982, "step": 18401 }, { "epoch": 0.6556103817446604, "grad_norm": 1.4855541538920598, "learning_rate": 5.601774138853278e-06, "loss": 0.5143, "step": 18402 }, { "epoch": 0.6556460088711545, "grad_norm": 1.412868665073559, "learning_rate": 5.600737835270167e-06, "loss": 0.5377, "step": 18403 }, { "epoch": 0.6556816359976486, "grad_norm": 1.0597325832458278, "learning_rate": 5.599701590266309e-06, "loss": 0.3641, "step": 18404 }, { "epoch": 0.6557172631241427, "grad_norm": 1.9401167815903846, "learning_rate": 5.598665403855497e-06, "loss": 0.4547, "step": 18405 }, { "epoch": 0.6557528902506369, "grad_norm": 2.0467871929506694, "learning_rate": 5.597629276051535e-06, "loss": 0.4859, "step": 18406 }, { "epoch": 0.655788517377131, "grad_norm": 1.2735411476652805, "learning_rate": 5.596593206868213e-06, "loss": 0.5496, "step": 18407 }, { "epoch": 0.655824144503625, "grad_norm": 1.4967248304966434, "learning_rate": 5.595557196319329e-06, "loss": 0.5279, "step": 18408 }, { "epoch": 0.6558597716301192, "grad_norm": 1.0906469648152206, "learning_rate": 5.59452124441868e-06, "loss": 0.41, "step": 18409 }, { "epoch": 0.6558953987566133, "grad_norm": 1.1831928511692338, "learning_rate": 5.593485351180059e-06, "loss": 0.4503, "step": 18410 }, { "epoch": 0.6559310258831074, "grad_norm": 1.1461299108471457, "learning_rate": 5.5924495166172664e-06, "loss": 0.4286, "step": 18411 }, { "epoch": 0.6559666530096016, "grad_norm": 1.3837988101218996, "learning_rate": 5.5914137407440835e-06, "loss": 0.5016, "step": 18412 }, { "epoch": 0.6560022801360956, "grad_norm": 1.8402330435183012, "learning_rate": 5.590378023574312e-06, "loss": 0.4946, "step": 18413 }, { "epoch": 0.6560379072625897, "grad_norm": 1.2134716647997517, "learning_rate": 5.5893423651217375e-06, "loss": 0.4265, "step": 18414 }, { "epoch": 0.6560735343890839, "grad_norm": 1.8298689209202283, "learning_rate": 5.58830676540015e-06, "loss": 0.5263, "step": 18415 }, { "epoch": 0.656109161515578, "grad_norm": 1.1499387348694106, "learning_rate": 5.587271224423343e-06, "loss": 0.4432, "step": 18416 }, { "epoch": 0.656144788642072, "grad_norm": 1.0138915389898582, "learning_rate": 5.586235742205106e-06, "loss": 0.3873, "step": 18417 }, { "epoch": 0.6561804157685662, "grad_norm": 1.1063294066040545, "learning_rate": 5.585200318759223e-06, "loss": 0.3927, "step": 18418 }, { "epoch": 0.6562160428950603, "grad_norm": 1.543123317838699, "learning_rate": 5.584164954099487e-06, "loss": 0.5611, "step": 18419 }, { "epoch": 0.6562516700215544, "grad_norm": 1.3088164022787805, "learning_rate": 5.583129648239677e-06, "loss": 0.4477, "step": 18420 }, { "epoch": 0.6562872971480486, "grad_norm": 1.5734508978705952, "learning_rate": 5.582094401193583e-06, "loss": 0.5345, "step": 18421 }, { "epoch": 0.6563229242745426, "grad_norm": 1.1830804297783633, "learning_rate": 5.58105921297499e-06, "loss": 0.4255, "step": 18422 }, { "epoch": 0.6563585514010367, "grad_norm": 1.5221830988777498, "learning_rate": 5.580024083597682e-06, "loss": 0.5641, "step": 18423 }, { "epoch": 0.6563941785275309, "grad_norm": 1.131022987069655, "learning_rate": 5.578989013075446e-06, "loss": 0.414, "step": 18424 }, { "epoch": 0.656429805654025, "grad_norm": 1.8715449492758534, "learning_rate": 5.577954001422059e-06, "loss": 0.5162, "step": 18425 }, { "epoch": 0.656465432780519, "grad_norm": 1.2072083560981774, "learning_rate": 5.576919048651308e-06, "loss": 0.4252, "step": 18426 }, { "epoch": 0.6565010599070132, "grad_norm": 1.4621117974765032, "learning_rate": 5.575884154776968e-06, "loss": 0.5014, "step": 18427 }, { "epoch": 0.6565366870335073, "grad_norm": 1.2070543437495473, "learning_rate": 5.5748493198128275e-06, "loss": 0.4418, "step": 18428 }, { "epoch": 0.6565723141600014, "grad_norm": 1.1967134800241197, "learning_rate": 5.573814543772656e-06, "loss": 0.4228, "step": 18429 }, { "epoch": 0.6566079412864956, "grad_norm": 1.1264148494714006, "learning_rate": 5.572779826670238e-06, "loss": 0.4261, "step": 18430 }, { "epoch": 0.6566435684129897, "grad_norm": 1.1392910908635001, "learning_rate": 5.571745168519351e-06, "loss": 0.4261, "step": 18431 }, { "epoch": 0.6566791955394837, "grad_norm": 1.149834335512481, "learning_rate": 5.570710569333773e-06, "loss": 0.4779, "step": 18432 }, { "epoch": 0.6567148226659779, "grad_norm": 1.8494677583791552, "learning_rate": 5.569676029127282e-06, "loss": 0.4136, "step": 18433 }, { "epoch": 0.656750449792472, "grad_norm": 1.664175806918462, "learning_rate": 5.568641547913649e-06, "loss": 0.4542, "step": 18434 }, { "epoch": 0.6567860769189661, "grad_norm": 1.657171183389557, "learning_rate": 5.5676071257066535e-06, "loss": 0.5074, "step": 18435 }, { "epoch": 0.6568217040454603, "grad_norm": 1.1296221978401377, "learning_rate": 5.566572762520064e-06, "loss": 0.5565, "step": 18436 }, { "epoch": 0.6568573311719543, "grad_norm": 1.1865245069376582, "learning_rate": 5.565538458367658e-06, "loss": 0.4556, "step": 18437 }, { "epoch": 0.6568929582984484, "grad_norm": 1.1490503838700066, "learning_rate": 5.564504213263205e-06, "loss": 0.4521, "step": 18438 }, { "epoch": 0.6569285854249426, "grad_norm": 1.2299719487635379, "learning_rate": 5.5634700272204835e-06, "loss": 0.4374, "step": 18439 }, { "epoch": 0.6569642125514367, "grad_norm": 1.7437587111021557, "learning_rate": 5.562435900253256e-06, "loss": 0.5569, "step": 18440 }, { "epoch": 0.6569998396779307, "grad_norm": 1.1654262784142704, "learning_rate": 5.5614018323753e-06, "loss": 0.462, "step": 18441 }, { "epoch": 0.6570354668044249, "grad_norm": 1.7068720559916268, "learning_rate": 5.560367823600377e-06, "loss": 0.5843, "step": 18442 }, { "epoch": 0.657071093930919, "grad_norm": 1.1176298083031087, "learning_rate": 5.559333873942259e-06, "loss": 0.4226, "step": 18443 }, { "epoch": 0.6571067210574131, "grad_norm": 1.144105926837691, "learning_rate": 5.558299983414719e-06, "loss": 0.4214, "step": 18444 }, { "epoch": 0.6571423481839073, "grad_norm": 1.0644236493532921, "learning_rate": 5.557266152031515e-06, "loss": 0.4384, "step": 18445 }, { "epoch": 0.6571779753104013, "grad_norm": 1.3128336427486331, "learning_rate": 5.556232379806417e-06, "loss": 0.5021, "step": 18446 }, { "epoch": 0.6572136024368954, "grad_norm": 1.1155514042928787, "learning_rate": 5.555198666753192e-06, "loss": 0.4578, "step": 18447 }, { "epoch": 0.6572492295633896, "grad_norm": 1.1828941262665031, "learning_rate": 5.554165012885606e-06, "loss": 0.4461, "step": 18448 }, { "epoch": 0.6572848566898837, "grad_norm": 1.2618918392512415, "learning_rate": 5.553131418217417e-06, "loss": 0.4622, "step": 18449 }, { "epoch": 0.6573204838163778, "grad_norm": 1.3182362084581751, "learning_rate": 5.552097882762395e-06, "loss": 0.4247, "step": 18450 }, { "epoch": 0.6573561109428719, "grad_norm": 1.1812292728840879, "learning_rate": 5.551064406534296e-06, "loss": 0.3732, "step": 18451 }, { "epoch": 0.657391738069366, "grad_norm": 1.7685861566106138, "learning_rate": 5.550030989546883e-06, "loss": 0.6055, "step": 18452 }, { "epoch": 0.6574273651958601, "grad_norm": 1.5824361584311653, "learning_rate": 5.5489976318139175e-06, "loss": 0.5412, "step": 18453 }, { "epoch": 0.6574629923223543, "grad_norm": 1.444665943264576, "learning_rate": 5.547964333349164e-06, "loss": 0.5034, "step": 18454 }, { "epoch": 0.6574986194488484, "grad_norm": 1.6654225455669822, "learning_rate": 5.546931094166372e-06, "loss": 0.5622, "step": 18455 }, { "epoch": 0.6575342465753424, "grad_norm": 1.5973361279573342, "learning_rate": 5.545897914279307e-06, "loss": 0.6282, "step": 18456 }, { "epoch": 0.6575698737018366, "grad_norm": 1.4070358027974885, "learning_rate": 5.5448647937017275e-06, "loss": 0.4931, "step": 18457 }, { "epoch": 0.6576055008283307, "grad_norm": 1.4222550206651763, "learning_rate": 5.543831732447383e-06, "loss": 0.5647, "step": 18458 }, { "epoch": 0.6576411279548248, "grad_norm": 1.1226115836773756, "learning_rate": 5.542798730530038e-06, "loss": 0.4201, "step": 18459 }, { "epoch": 0.657676755081319, "grad_norm": 1.2000344738393032, "learning_rate": 5.541765787963435e-06, "loss": 0.4484, "step": 18460 }, { "epoch": 0.657712382207813, "grad_norm": 1.889022092517678, "learning_rate": 5.5407329047613455e-06, "loss": 0.5747, "step": 18461 }, { "epoch": 0.6577480093343071, "grad_norm": 1.3888027681626458, "learning_rate": 5.539700080937509e-06, "loss": 0.5089, "step": 18462 }, { "epoch": 0.6577836364608013, "grad_norm": 1.2174240970657964, "learning_rate": 5.5386673165056884e-06, "loss": 0.437, "step": 18463 }, { "epoch": 0.6578192635872954, "grad_norm": 1.1384312276179198, "learning_rate": 5.537634611479626e-06, "loss": 0.4123, "step": 18464 }, { "epoch": 0.6578548907137894, "grad_norm": 1.0589747021916065, "learning_rate": 5.536601965873083e-06, "loss": 0.3877, "step": 18465 }, { "epoch": 0.6578905178402836, "grad_norm": 1.1375105745818186, "learning_rate": 5.535569379699799e-06, "loss": 0.43, "step": 18466 }, { "epoch": 0.6579261449667777, "grad_norm": 1.2169911571261174, "learning_rate": 5.53453685297353e-06, "loss": 0.4058, "step": 18467 }, { "epoch": 0.6579617720932718, "grad_norm": 1.3135776043427199, "learning_rate": 5.533504385708024e-06, "loss": 0.4498, "step": 18468 }, { "epoch": 0.657997399219766, "grad_norm": 1.3860759656680297, "learning_rate": 5.532471977917029e-06, "loss": 0.5935, "step": 18469 }, { "epoch": 0.65803302634626, "grad_norm": 2.61348650730092, "learning_rate": 5.531439629614296e-06, "loss": 0.5911, "step": 18470 }, { "epoch": 0.6580686534727541, "grad_norm": 1.1058471288359295, "learning_rate": 5.530407340813564e-06, "loss": 0.4423, "step": 18471 }, { "epoch": 0.6581042805992483, "grad_norm": 1.2271550749383588, "learning_rate": 5.529375111528586e-06, "loss": 0.4263, "step": 18472 }, { "epoch": 0.6581399077257424, "grad_norm": 1.6666933606590102, "learning_rate": 5.5283429417730996e-06, "loss": 0.524, "step": 18473 }, { "epoch": 0.6581755348522365, "grad_norm": 1.27351037808964, "learning_rate": 5.5273108315608514e-06, "loss": 0.4867, "step": 18474 }, { "epoch": 0.6582111619787306, "grad_norm": 1.1232351863349335, "learning_rate": 5.526278780905587e-06, "loss": 0.4272, "step": 18475 }, { "epoch": 0.6582467891052247, "grad_norm": 1.3033153897256664, "learning_rate": 5.5252467898210505e-06, "loss": 0.5069, "step": 18476 }, { "epoch": 0.6582824162317188, "grad_norm": 1.1624466773924635, "learning_rate": 5.524214858320978e-06, "loss": 0.4057, "step": 18477 }, { "epoch": 0.658318043358213, "grad_norm": 1.1716342621870932, "learning_rate": 5.523182986419111e-06, "loss": 0.4489, "step": 18478 }, { "epoch": 0.658353670484707, "grad_norm": 1.378892918431882, "learning_rate": 5.5221511741291975e-06, "loss": 0.4952, "step": 18479 }, { "epoch": 0.6583892976112011, "grad_norm": 1.2173295988141095, "learning_rate": 5.5211194214649665e-06, "loss": 0.4303, "step": 18480 }, { "epoch": 0.6584249247376953, "grad_norm": 1.8939341535615937, "learning_rate": 5.520087728440164e-06, "loss": 0.5409, "step": 18481 }, { "epoch": 0.6584605518641894, "grad_norm": 1.9117364895639797, "learning_rate": 5.5190560950685215e-06, "loss": 0.5401, "step": 18482 }, { "epoch": 0.6584961789906835, "grad_norm": 1.0844226617493173, "learning_rate": 5.5180245213637785e-06, "loss": 0.3878, "step": 18483 }, { "epoch": 0.6585318061171777, "grad_norm": 1.2847715726613607, "learning_rate": 5.5169930073396725e-06, "loss": 0.4192, "step": 18484 }, { "epoch": 0.6585674332436717, "grad_norm": 1.14433668008346, "learning_rate": 5.515961553009942e-06, "loss": 0.4138, "step": 18485 }, { "epoch": 0.6586030603701658, "grad_norm": 1.1036167954796843, "learning_rate": 5.514930158388313e-06, "loss": 0.3851, "step": 18486 }, { "epoch": 0.65863868749666, "grad_norm": 1.1440824707611814, "learning_rate": 5.513898823488528e-06, "loss": 0.4004, "step": 18487 }, { "epoch": 0.6586743146231541, "grad_norm": 1.2551436426566656, "learning_rate": 5.512867548324312e-06, "loss": 0.4323, "step": 18488 }, { "epoch": 0.6587099417496481, "grad_norm": 1.4188139530879633, "learning_rate": 5.511836332909401e-06, "loss": 0.5133, "step": 18489 }, { "epoch": 0.6587455688761423, "grad_norm": 1.1255823325040903, "learning_rate": 5.510805177257525e-06, "loss": 0.3831, "step": 18490 }, { "epoch": 0.6587811960026364, "grad_norm": 1.1474010146012197, "learning_rate": 5.509774081382417e-06, "loss": 0.4435, "step": 18491 }, { "epoch": 0.6588168231291305, "grad_norm": 1.5396819491845999, "learning_rate": 5.5087430452978105e-06, "loss": 0.5094, "step": 18492 }, { "epoch": 0.6588524502556247, "grad_norm": 1.3411879059665572, "learning_rate": 5.507712069017425e-06, "loss": 0.5134, "step": 18493 }, { "epoch": 0.6588880773821187, "grad_norm": 1.3391310032463373, "learning_rate": 5.506681152554996e-06, "loss": 0.5558, "step": 18494 }, { "epoch": 0.6589237045086128, "grad_norm": 1.3826200585211497, "learning_rate": 5.505650295924245e-06, "loss": 0.4458, "step": 18495 }, { "epoch": 0.658959331635107, "grad_norm": 1.7568141975576093, "learning_rate": 5.504619499138906e-06, "loss": 0.4761, "step": 18496 }, { "epoch": 0.6589949587616011, "grad_norm": 1.7632929793088206, "learning_rate": 5.5035887622126915e-06, "loss": 0.5371, "step": 18497 }, { "epoch": 0.6590305858880952, "grad_norm": 1.232514279344479, "learning_rate": 5.502558085159344e-06, "loss": 0.4483, "step": 18498 }, { "epoch": 0.6590662130145893, "grad_norm": 1.4552197384473353, "learning_rate": 5.501527467992576e-06, "loss": 0.5733, "step": 18499 }, { "epoch": 0.6591018401410834, "grad_norm": 1.6963959456681201, "learning_rate": 5.5004969107261166e-06, "loss": 0.5427, "step": 18500 }, { "epoch": 0.6591374672675775, "grad_norm": 1.3612376571139275, "learning_rate": 5.499466413373682e-06, "loss": 0.4068, "step": 18501 }, { "epoch": 0.6591730943940717, "grad_norm": 1.451381565810405, "learning_rate": 5.4984359759489966e-06, "loss": 0.5722, "step": 18502 }, { "epoch": 0.6592087215205658, "grad_norm": 1.3808049356035876, "learning_rate": 5.49740559846579e-06, "loss": 0.4412, "step": 18503 }, { "epoch": 0.6592443486470598, "grad_norm": 1.348492459851258, "learning_rate": 5.496375280937768e-06, "loss": 0.5255, "step": 18504 }, { "epoch": 0.659279975773554, "grad_norm": 1.4706843376247718, "learning_rate": 5.495345023378658e-06, "loss": 0.6156, "step": 18505 }, { "epoch": 0.6593156029000481, "grad_norm": 1.753197734368973, "learning_rate": 5.494314825802178e-06, "loss": 0.5372, "step": 18506 }, { "epoch": 0.6593512300265422, "grad_norm": 1.4883140900142868, "learning_rate": 5.493284688222049e-06, "loss": 0.4204, "step": 18507 }, { "epoch": 0.6593868571530364, "grad_norm": 1.3736178957935605, "learning_rate": 5.492254610651979e-06, "loss": 0.421, "step": 18508 }, { "epoch": 0.6594224842795304, "grad_norm": 1.711954226414557, "learning_rate": 5.491224593105695e-06, "loss": 0.428, "step": 18509 }, { "epoch": 0.6594581114060245, "grad_norm": 2.4304838181175743, "learning_rate": 5.490194635596904e-06, "loss": 0.4059, "step": 18510 }, { "epoch": 0.6594937385325187, "grad_norm": 1.620193581632694, "learning_rate": 5.489164738139322e-06, "loss": 0.4579, "step": 18511 }, { "epoch": 0.6595293656590128, "grad_norm": 1.3100659047342609, "learning_rate": 5.488134900746665e-06, "loss": 0.4775, "step": 18512 }, { "epoch": 0.6595649927855068, "grad_norm": 1.4323405062016659, "learning_rate": 5.487105123432645e-06, "loss": 0.5007, "step": 18513 }, { "epoch": 0.659600619912001, "grad_norm": 1.6716913894233665, "learning_rate": 5.486075406210979e-06, "loss": 0.4082, "step": 18514 }, { "epoch": 0.6596362470384951, "grad_norm": 2.034604260620384, "learning_rate": 5.48504574909537e-06, "loss": 0.4954, "step": 18515 }, { "epoch": 0.6596718741649892, "grad_norm": 1.4582525598079912, "learning_rate": 5.484016152099536e-06, "loss": 0.5043, "step": 18516 }, { "epoch": 0.6597075012914834, "grad_norm": 1.8412363154028843, "learning_rate": 5.48298661523718e-06, "loss": 0.4501, "step": 18517 }, { "epoch": 0.6597431284179774, "grad_norm": 1.4453342664612192, "learning_rate": 5.481957138522018e-06, "loss": 0.4887, "step": 18518 }, { "epoch": 0.6597787555444715, "grad_norm": 1.961604464703254, "learning_rate": 5.480927721967752e-06, "loss": 0.5442, "step": 18519 }, { "epoch": 0.6598143826709657, "grad_norm": 1.602389277837438, "learning_rate": 5.47989836558809e-06, "loss": 0.5249, "step": 18520 }, { "epoch": 0.6598500097974598, "grad_norm": 1.5647445824796402, "learning_rate": 5.4788690693967414e-06, "loss": 0.4075, "step": 18521 }, { "epoch": 0.6598856369239539, "grad_norm": 1.4895094225515921, "learning_rate": 5.477839833407416e-06, "loss": 0.406, "step": 18522 }, { "epoch": 0.659921264050448, "grad_norm": 1.561336146369323, "learning_rate": 5.47681065763381e-06, "loss": 0.4148, "step": 18523 }, { "epoch": 0.6599568911769421, "grad_norm": 1.5613954028038666, "learning_rate": 5.475781542089631e-06, "loss": 0.3951, "step": 18524 }, { "epoch": 0.6599925183034362, "grad_norm": 1.4082918052451165, "learning_rate": 5.474752486788587e-06, "loss": 0.4841, "step": 18525 }, { "epoch": 0.6600281454299304, "grad_norm": 1.915880509025183, "learning_rate": 5.473723491744373e-06, "loss": 0.4513, "step": 18526 }, { "epoch": 0.6600637725564245, "grad_norm": 1.5224010906426535, "learning_rate": 5.472694556970693e-06, "loss": 0.5185, "step": 18527 }, { "epoch": 0.6600993996829185, "grad_norm": 1.3999062350091516, "learning_rate": 5.4716656824812505e-06, "loss": 0.4864, "step": 18528 }, { "epoch": 0.6601350268094127, "grad_norm": 2.130546575656017, "learning_rate": 5.470636868289749e-06, "loss": 0.4189, "step": 18529 }, { "epoch": 0.6601706539359068, "grad_norm": 1.904633439190295, "learning_rate": 5.469608114409879e-06, "loss": 0.4888, "step": 18530 }, { "epoch": 0.6602062810624009, "grad_norm": 1.4611501977444068, "learning_rate": 5.4685794208553465e-06, "loss": 0.5216, "step": 18531 }, { "epoch": 0.660241908188895, "grad_norm": 1.7657491059890436, "learning_rate": 5.467550787639844e-06, "loss": 0.373, "step": 18532 }, { "epoch": 0.6602775353153891, "grad_norm": 1.8192451100065234, "learning_rate": 5.466522214777074e-06, "loss": 0.4092, "step": 18533 }, { "epoch": 0.6603131624418832, "grad_norm": 1.5407521332219205, "learning_rate": 5.465493702280725e-06, "loss": 0.5928, "step": 18534 }, { "epoch": 0.6603487895683774, "grad_norm": 1.8200646061837789, "learning_rate": 5.464465250164498e-06, "loss": 0.4329, "step": 18535 }, { "epoch": 0.6603844166948715, "grad_norm": 1.5450331316647663, "learning_rate": 5.463436858442085e-06, "loss": 0.4473, "step": 18536 }, { "epoch": 0.6604200438213655, "grad_norm": 2.068726034102741, "learning_rate": 5.462408527127182e-06, "loss": 0.4327, "step": 18537 }, { "epoch": 0.6604556709478597, "grad_norm": 1.9422500610981712, "learning_rate": 5.461380256233484e-06, "loss": 0.565, "step": 18538 }, { "epoch": 0.6604912980743538, "grad_norm": 1.6432221060711587, "learning_rate": 5.4603520457746775e-06, "loss": 0.6191, "step": 18539 }, { "epoch": 0.6605269252008479, "grad_norm": 2.7165708626404808, "learning_rate": 5.459323895764459e-06, "loss": 0.4363, "step": 18540 }, { "epoch": 0.6605625523273421, "grad_norm": 1.935692737541058, "learning_rate": 5.458295806216514e-06, "loss": 0.7164, "step": 18541 }, { "epoch": 0.6605981794538361, "grad_norm": 2.470617264562458, "learning_rate": 5.457267777144535e-06, "loss": 0.4382, "step": 18542 }, { "epoch": 0.6606338065803302, "grad_norm": 1.6470330503838235, "learning_rate": 5.45623980856221e-06, "loss": 0.5278, "step": 18543 }, { "epoch": 0.6606694337068244, "grad_norm": 1.7126988257606648, "learning_rate": 5.455211900483232e-06, "loss": 0.4315, "step": 18544 }, { "epoch": 0.6607050608333185, "grad_norm": 2.244001782145433, "learning_rate": 5.454184052921278e-06, "loss": 0.4031, "step": 18545 }, { "epoch": 0.6607406879598126, "grad_norm": 2.461218982718604, "learning_rate": 5.453156265890048e-06, "loss": 0.4016, "step": 18546 }, { "epoch": 0.6607763150863067, "grad_norm": 1.7722164313747542, "learning_rate": 5.452128539403215e-06, "loss": 0.4461, "step": 18547 }, { "epoch": 0.6608119422128008, "grad_norm": 1.7381063219360224, "learning_rate": 5.451100873474468e-06, "loss": 0.5708, "step": 18548 }, { "epoch": 0.6608475693392949, "grad_norm": 1.6934733881782442, "learning_rate": 5.450073268117495e-06, "loss": 0.423, "step": 18549 }, { "epoch": 0.6608831964657891, "grad_norm": 1.4926232570155933, "learning_rate": 5.449045723345976e-06, "loss": 0.6001, "step": 18550 }, { "epoch": 0.6609188235922832, "grad_norm": 1.4670992370812945, "learning_rate": 5.448018239173597e-06, "loss": 0.4866, "step": 18551 }, { "epoch": 0.6609544507187772, "grad_norm": 1.665808919282987, "learning_rate": 5.446990815614033e-06, "loss": 0.6487, "step": 18552 }, { "epoch": 0.6609900778452714, "grad_norm": 1.9049078866447056, "learning_rate": 5.445963452680974e-06, "loss": 0.551, "step": 18553 }, { "epoch": 0.6610257049717655, "grad_norm": 1.6379040481568519, "learning_rate": 5.4449361503880895e-06, "loss": 0.4753, "step": 18554 }, { "epoch": 0.6610613320982596, "grad_norm": 3.884665312130936, "learning_rate": 5.443908908749067e-06, "loss": 0.4964, "step": 18555 }, { "epoch": 0.6610969592247538, "grad_norm": 1.9001898370765251, "learning_rate": 5.442881727777579e-06, "loss": 0.4128, "step": 18556 }, { "epoch": 0.6611325863512478, "grad_norm": 1.9358704370228805, "learning_rate": 5.441854607487305e-06, "loss": 0.3955, "step": 18557 }, { "epoch": 0.6611682134777419, "grad_norm": 2.354615726810754, "learning_rate": 5.4408275478919246e-06, "loss": 0.352, "step": 18558 }, { "epoch": 0.6612038406042361, "grad_norm": 1.5133922916926807, "learning_rate": 5.439800549005112e-06, "loss": 0.4177, "step": 18559 }, { "epoch": 0.6612394677307302, "grad_norm": 1.5410318760577395, "learning_rate": 5.438773610840546e-06, "loss": 0.476, "step": 18560 }, { "epoch": 0.6612750948572242, "grad_norm": 1.4060886549640295, "learning_rate": 5.437746733411893e-06, "loss": 0.5399, "step": 18561 }, { "epoch": 0.6613107219837184, "grad_norm": 1.6832383070601116, "learning_rate": 5.436719916732836e-06, "loss": 0.4999, "step": 18562 }, { "epoch": 0.6613463491102125, "grad_norm": 1.6211932062399947, "learning_rate": 5.435693160817037e-06, "loss": 0.4181, "step": 18563 }, { "epoch": 0.6613819762367066, "grad_norm": 1.883410148026755, "learning_rate": 5.434666465678176e-06, "loss": 0.425, "step": 18564 }, { "epoch": 0.6614176033632008, "grad_norm": 1.5061341866449522, "learning_rate": 5.43363983132992e-06, "loss": 0.426, "step": 18565 }, { "epoch": 0.6614532304896948, "grad_norm": 1.5772337714990055, "learning_rate": 5.432613257785946e-06, "loss": 0.4479, "step": 18566 }, { "epoch": 0.6614888576161889, "grad_norm": 1.6399054552740093, "learning_rate": 5.431586745059916e-06, "loss": 0.4452, "step": 18567 }, { "epoch": 0.6615244847426831, "grad_norm": 1.6948550092496975, "learning_rate": 5.430560293165504e-06, "loss": 0.4324, "step": 18568 }, { "epoch": 0.6615601118691772, "grad_norm": 1.9031091101204631, "learning_rate": 5.429533902116373e-06, "loss": 0.4797, "step": 18569 }, { "epoch": 0.6615957389956713, "grad_norm": 1.7981209237611118, "learning_rate": 5.428507571926192e-06, "loss": 0.5646, "step": 18570 }, { "epoch": 0.6616313661221654, "grad_norm": 1.4953308611581073, "learning_rate": 5.427481302608632e-06, "loss": 0.4384, "step": 18571 }, { "epoch": 0.6616669932486595, "grad_norm": 1.5430529731382947, "learning_rate": 5.426455094177346e-06, "loss": 0.4221, "step": 18572 }, { "epoch": 0.6617026203751536, "grad_norm": 1.9383347412542915, "learning_rate": 5.425428946646016e-06, "loss": 0.4117, "step": 18573 }, { "epoch": 0.6617382475016478, "grad_norm": 1.6219079831787144, "learning_rate": 5.4244028600282925e-06, "loss": 0.5838, "step": 18574 }, { "epoch": 0.6617738746281419, "grad_norm": 1.8336655049464352, "learning_rate": 5.423376834337848e-06, "loss": 0.5556, "step": 18575 }, { "epoch": 0.6618095017546359, "grad_norm": 1.649684099794687, "learning_rate": 5.422350869588334e-06, "loss": 0.3852, "step": 18576 }, { "epoch": 0.6618451288811301, "grad_norm": 1.9473384521065025, "learning_rate": 5.421324965793423e-06, "loss": 0.4437, "step": 18577 }, { "epoch": 0.6618807560076242, "grad_norm": 1.5081320252285806, "learning_rate": 5.420299122966766e-06, "loss": 0.3994, "step": 18578 }, { "epoch": 0.6619163831341183, "grad_norm": 1.533477421562055, "learning_rate": 5.4192733411220275e-06, "loss": 0.4365, "step": 18579 }, { "epoch": 0.6619520102606125, "grad_norm": 1.7736946071638606, "learning_rate": 5.418247620272866e-06, "loss": 0.5742, "step": 18580 }, { "epoch": 0.6619876373871065, "grad_norm": 1.8868466175420624, "learning_rate": 5.4172219604329436e-06, "loss": 0.4524, "step": 18581 }, { "epoch": 0.6620232645136006, "grad_norm": 1.571622763022118, "learning_rate": 5.4161963616159104e-06, "loss": 0.4442, "step": 18582 }, { "epoch": 0.6620588916400948, "grad_norm": 2.157287794123377, "learning_rate": 5.415170823835427e-06, "loss": 0.4137, "step": 18583 }, { "epoch": 0.6620945187665889, "grad_norm": 1.9879392330240888, "learning_rate": 5.414145347105152e-06, "loss": 0.5096, "step": 18584 }, { "epoch": 0.6621301458930829, "grad_norm": 2.0257882783518317, "learning_rate": 5.413119931438733e-06, "loss": 0.4152, "step": 18585 }, { "epoch": 0.6621657730195771, "grad_norm": 1.5360887649388166, "learning_rate": 5.412094576849829e-06, "loss": 0.4274, "step": 18586 }, { "epoch": 0.6622014001460712, "grad_norm": 2.102087209284168, "learning_rate": 5.411069283352091e-06, "loss": 0.5303, "step": 18587 }, { "epoch": 0.6622370272725653, "grad_norm": 1.4541641240407073, "learning_rate": 5.410044050959178e-06, "loss": 0.5275, "step": 18588 }, { "epoch": 0.6622726543990595, "grad_norm": 1.7751086851807643, "learning_rate": 5.409018879684732e-06, "loss": 0.4415, "step": 18589 }, { "epoch": 0.6623082815255535, "grad_norm": 2.084656961744457, "learning_rate": 5.407993769542412e-06, "loss": 0.4466, "step": 18590 }, { "epoch": 0.6623439086520476, "grad_norm": 1.3082399758441898, "learning_rate": 5.406968720545862e-06, "loss": 0.5335, "step": 18591 }, { "epoch": 0.6623795357785418, "grad_norm": 2.047895500122751, "learning_rate": 5.405943732708738e-06, "loss": 0.5576, "step": 18592 }, { "epoch": 0.6624151629050359, "grad_norm": 1.521019563015136, "learning_rate": 5.404918806044679e-06, "loss": 0.4977, "step": 18593 }, { "epoch": 0.66245079003153, "grad_norm": 1.7790416905087456, "learning_rate": 5.403893940567338e-06, "loss": 0.532, "step": 18594 }, { "epoch": 0.6624864171580241, "grad_norm": 1.467275488589459, "learning_rate": 5.402869136290363e-06, "loss": 0.4628, "step": 18595 }, { "epoch": 0.6625220442845182, "grad_norm": 2.0011528431031746, "learning_rate": 5.401844393227398e-06, "loss": 0.4606, "step": 18596 }, { "epoch": 0.6625576714110123, "grad_norm": 1.408438622905239, "learning_rate": 5.400819711392092e-06, "loss": 0.4442, "step": 18597 }, { "epoch": 0.6625932985375065, "grad_norm": 1.5395048857201818, "learning_rate": 5.399795090798082e-06, "loss": 0.4407, "step": 18598 }, { "epoch": 0.6626289256640006, "grad_norm": 1.7394644124789855, "learning_rate": 5.398770531459021e-06, "loss": 0.4116, "step": 18599 }, { "epoch": 0.6626645527904946, "grad_norm": 1.7955984839933998, "learning_rate": 5.397746033388542e-06, "loss": 0.4986, "step": 18600 }, { "epoch": 0.6627001799169888, "grad_norm": 2.053799687860805, "learning_rate": 5.396721596600292e-06, "loss": 0.4026, "step": 18601 }, { "epoch": 0.6627358070434829, "grad_norm": 1.4118342748107642, "learning_rate": 5.395697221107909e-06, "loss": 0.4106, "step": 18602 }, { "epoch": 0.662771434169977, "grad_norm": 1.8006682260994462, "learning_rate": 5.394672906925043e-06, "loss": 0.4741, "step": 18603 }, { "epoch": 0.6628070612964712, "grad_norm": 1.429563819972579, "learning_rate": 5.39364865406532e-06, "loss": 0.4038, "step": 18604 }, { "epoch": 0.6628426884229652, "grad_norm": 1.6384808746761248, "learning_rate": 5.3926244625423855e-06, "loss": 0.4001, "step": 18605 }, { "epoch": 0.6628783155494593, "grad_norm": 1.5055385793041354, "learning_rate": 5.39160033236988e-06, "loss": 0.5344, "step": 18606 }, { "epoch": 0.6629139426759535, "grad_norm": 1.6439354438325384, "learning_rate": 5.3905762635614334e-06, "loss": 0.5325, "step": 18607 }, { "epoch": 0.6629495698024476, "grad_norm": 2.576317616242927, "learning_rate": 5.38955225613069e-06, "loss": 0.4437, "step": 18608 }, { "epoch": 0.6629851969289416, "grad_norm": 1.8040661881062348, "learning_rate": 5.388528310091277e-06, "loss": 0.4638, "step": 18609 }, { "epoch": 0.6630208240554358, "grad_norm": 1.5414637922237233, "learning_rate": 5.387504425456833e-06, "loss": 0.4033, "step": 18610 }, { "epoch": 0.6630564511819299, "grad_norm": 1.3931429567335483, "learning_rate": 5.386480602240991e-06, "loss": 0.4438, "step": 18611 }, { "epoch": 0.663092078308424, "grad_norm": 1.5963005476135548, "learning_rate": 5.385456840457388e-06, "loss": 0.5739, "step": 18612 }, { "epoch": 0.6631277054349182, "grad_norm": 1.377291469338426, "learning_rate": 5.384433140119649e-06, "loss": 0.3693, "step": 18613 }, { "epoch": 0.6631633325614122, "grad_norm": 1.501219974370287, "learning_rate": 5.3834095012414125e-06, "loss": 0.5608, "step": 18614 }, { "epoch": 0.6631989596879063, "grad_norm": 1.5399301702250547, "learning_rate": 5.382385923836302e-06, "loss": 0.4151, "step": 18615 }, { "epoch": 0.6632345868144005, "grad_norm": 1.4139021615580043, "learning_rate": 5.3813624079179495e-06, "loss": 0.4344, "step": 18616 }, { "epoch": 0.6632702139408946, "grad_norm": 1.7555493389219141, "learning_rate": 5.380338953499986e-06, "loss": 0.4801, "step": 18617 }, { "epoch": 0.6633058410673887, "grad_norm": 1.4943977655932352, "learning_rate": 5.379315560596038e-06, "loss": 0.4107, "step": 18618 }, { "epoch": 0.6633414681938828, "grad_norm": 1.5242282518581864, "learning_rate": 5.378292229219736e-06, "loss": 0.4119, "step": 18619 }, { "epoch": 0.6633770953203769, "grad_norm": 1.585454244480689, "learning_rate": 5.3772689593847e-06, "loss": 0.3731, "step": 18620 }, { "epoch": 0.663412722446871, "grad_norm": 1.4364222865323548, "learning_rate": 5.376245751104563e-06, "loss": 0.3822, "step": 18621 }, { "epoch": 0.6634483495733652, "grad_norm": 1.623895041497824, "learning_rate": 5.375222604392942e-06, "loss": 0.3976, "step": 18622 }, { "epoch": 0.6634839766998593, "grad_norm": 1.7759085202135132, "learning_rate": 5.374199519263468e-06, "loss": 0.5755, "step": 18623 }, { "epoch": 0.6635196038263533, "grad_norm": 1.716777556674158, "learning_rate": 5.373176495729752e-06, "loss": 0.6323, "step": 18624 }, { "epoch": 0.6635552309528475, "grad_norm": 2.0771939560541877, "learning_rate": 5.372153533805433e-06, "loss": 0.4264, "step": 18625 }, { "epoch": 0.6635908580793416, "grad_norm": 4.435509345003235, "learning_rate": 5.3711306335041224e-06, "loss": 0.4084, "step": 18626 }, { "epoch": 0.6636264852058357, "grad_norm": 2.3759930525459407, "learning_rate": 5.370107794839445e-06, "loss": 0.4132, "step": 18627 }, { "epoch": 0.6636621123323299, "grad_norm": 1.5246169409352597, "learning_rate": 5.369085017825015e-06, "loss": 0.4547, "step": 18628 }, { "epoch": 0.6636977394588239, "grad_norm": 1.618786660377114, "learning_rate": 5.368062302474455e-06, "loss": 0.5534, "step": 18629 }, { "epoch": 0.663733366585318, "grad_norm": 1.320269313631991, "learning_rate": 5.367039648801386e-06, "loss": 0.4832, "step": 18630 }, { "epoch": 0.6637689937118122, "grad_norm": 1.6366830184299004, "learning_rate": 5.366017056819419e-06, "loss": 0.4089, "step": 18631 }, { "epoch": 0.6638046208383063, "grad_norm": 1.3737470717386864, "learning_rate": 5.364994526542173e-06, "loss": 0.5777, "step": 18632 }, { "epoch": 0.6638402479648003, "grad_norm": 1.8332228604804437, "learning_rate": 5.363972057983266e-06, "loss": 0.4652, "step": 18633 }, { "epoch": 0.6638758750912945, "grad_norm": 1.7774028689996282, "learning_rate": 5.3629496511563154e-06, "loss": 0.3873, "step": 18634 }, { "epoch": 0.6639115022177886, "grad_norm": 1.6098293004932036, "learning_rate": 5.361927306074925e-06, "loss": 0.4483, "step": 18635 }, { "epoch": 0.6639471293442827, "grad_norm": 1.4210316345248324, "learning_rate": 5.3609050227527205e-06, "loss": 0.6168, "step": 18636 }, { "epoch": 0.6639827564707769, "grad_norm": 1.3093989729664537, "learning_rate": 5.359882801203305e-06, "loss": 0.445, "step": 18637 }, { "epoch": 0.664018383597271, "grad_norm": 1.5766900111098143, "learning_rate": 5.3588606414402914e-06, "loss": 0.4506, "step": 18638 }, { "epoch": 0.664054010723765, "grad_norm": 1.4775546857393427, "learning_rate": 5.3578385434772916e-06, "loss": 0.4767, "step": 18639 }, { "epoch": 0.6640896378502592, "grad_norm": 1.9251313753016146, "learning_rate": 5.356816507327918e-06, "loss": 0.5605, "step": 18640 }, { "epoch": 0.6641252649767533, "grad_norm": 1.4874577738946846, "learning_rate": 5.355794533005782e-06, "loss": 0.4374, "step": 18641 }, { "epoch": 0.6641608921032474, "grad_norm": 1.4352369116145594, "learning_rate": 5.354772620524482e-06, "loss": 0.5363, "step": 18642 }, { "epoch": 0.6641965192297415, "grad_norm": 1.4928714307523383, "learning_rate": 5.353750769897636e-06, "loss": 0.5807, "step": 18643 }, { "epoch": 0.6642321463562356, "grad_norm": 1.2339001896799475, "learning_rate": 5.3527289811388435e-06, "loss": 0.4473, "step": 18644 }, { "epoch": 0.6642677734827297, "grad_norm": 1.099288313728827, "learning_rate": 5.3517072542617164e-06, "loss": 0.3773, "step": 18645 }, { "epoch": 0.6643034006092239, "grad_norm": 1.4064772656084685, "learning_rate": 5.350685589279852e-06, "loss": 0.5098, "step": 18646 }, { "epoch": 0.664339027735718, "grad_norm": 1.5271286808105917, "learning_rate": 5.349663986206859e-06, "loss": 0.4863, "step": 18647 }, { "epoch": 0.664374654862212, "grad_norm": 1.2966543130074932, "learning_rate": 5.34864244505634e-06, "loss": 0.3883, "step": 18648 }, { "epoch": 0.6644102819887062, "grad_norm": 1.366358456568256, "learning_rate": 5.3476209658419024e-06, "loss": 0.4289, "step": 18649 }, { "epoch": 0.6644459091152003, "grad_norm": 1.4523084479163344, "learning_rate": 5.346599548577138e-06, "loss": 0.6003, "step": 18650 }, { "epoch": 0.6644815362416945, "grad_norm": 1.5788497315924521, "learning_rate": 5.345578193275654e-06, "loss": 0.4941, "step": 18651 }, { "epoch": 0.6645171633681886, "grad_norm": 1.3584385260162228, "learning_rate": 5.344556899951055e-06, "loss": 0.4092, "step": 18652 }, { "epoch": 0.6645527904946826, "grad_norm": 1.2793539031881693, "learning_rate": 5.343535668616928e-06, "loss": 0.4767, "step": 18653 }, { "epoch": 0.6645884176211768, "grad_norm": 1.5109322047790534, "learning_rate": 5.34251449928688e-06, "loss": 0.5395, "step": 18654 }, { "epoch": 0.6646240447476709, "grad_norm": 1.423360467349319, "learning_rate": 5.341493391974507e-06, "loss": 0.5333, "step": 18655 }, { "epoch": 0.664659671874165, "grad_norm": 1.1273036898236277, "learning_rate": 5.340472346693409e-06, "loss": 0.4438, "step": 18656 }, { "epoch": 0.6646952990006592, "grad_norm": 1.2180680003937934, "learning_rate": 5.339451363457175e-06, "loss": 0.4084, "step": 18657 }, { "epoch": 0.6647309261271532, "grad_norm": 1.4528620267443337, "learning_rate": 5.338430442279408e-06, "loss": 0.3941, "step": 18658 }, { "epoch": 0.6647665532536473, "grad_norm": 1.3161368569305736, "learning_rate": 5.337409583173693e-06, "loss": 0.4974, "step": 18659 }, { "epoch": 0.6648021803801415, "grad_norm": 2.5735078764588564, "learning_rate": 5.336388786153632e-06, "loss": 0.559, "step": 18660 }, { "epoch": 0.6648378075066356, "grad_norm": 1.1260435268392157, "learning_rate": 5.335368051232808e-06, "loss": 0.4035, "step": 18661 }, { "epoch": 0.6648734346331296, "grad_norm": 1.8434751553561053, "learning_rate": 5.334347378424825e-06, "loss": 0.5696, "step": 18662 }, { "epoch": 0.6649090617596238, "grad_norm": 1.599314448031589, "learning_rate": 5.333326767743263e-06, "loss": 0.5047, "step": 18663 }, { "epoch": 0.6649446888861179, "grad_norm": 1.2395128745403174, "learning_rate": 5.332306219201718e-06, "loss": 0.4401, "step": 18664 }, { "epoch": 0.664980316012612, "grad_norm": 1.5786626027592905, "learning_rate": 5.331285732813782e-06, "loss": 0.5012, "step": 18665 }, { "epoch": 0.6650159431391062, "grad_norm": 1.1558866820731306, "learning_rate": 5.330265308593037e-06, "loss": 0.4302, "step": 18666 }, { "epoch": 0.6650515702656002, "grad_norm": 1.635856920958621, "learning_rate": 5.329244946553076e-06, "loss": 0.5671, "step": 18667 }, { "epoch": 0.6650871973920943, "grad_norm": 1.0805653761662362, "learning_rate": 5.328224646707479e-06, "loss": 0.4441, "step": 18668 }, { "epoch": 0.6651228245185885, "grad_norm": 1.0978695428676672, "learning_rate": 5.327204409069837e-06, "loss": 0.3832, "step": 18669 }, { "epoch": 0.6651584516450826, "grad_norm": 1.6673655327352623, "learning_rate": 5.326184233653735e-06, "loss": 0.4863, "step": 18670 }, { "epoch": 0.6651940787715767, "grad_norm": 1.208017146601627, "learning_rate": 5.325164120472759e-06, "loss": 0.4013, "step": 18671 }, { "epoch": 0.6652297058980708, "grad_norm": 1.7052091813293033, "learning_rate": 5.324144069540487e-06, "loss": 0.378, "step": 18672 }, { "epoch": 0.6652653330245649, "grad_norm": 1.4606457066252099, "learning_rate": 5.32312408087051e-06, "loss": 0.5028, "step": 18673 }, { "epoch": 0.665300960151059, "grad_norm": 3.2555836093080766, "learning_rate": 5.3221041544764e-06, "loss": 0.4377, "step": 18674 }, { "epoch": 0.6653365872775532, "grad_norm": 1.7677578848337918, "learning_rate": 5.321084290371742e-06, "loss": 0.55, "step": 18675 }, { "epoch": 0.6653722144040473, "grad_norm": 1.3805545051130177, "learning_rate": 5.3200644885701184e-06, "loss": 0.511, "step": 18676 }, { "epoch": 0.6654078415305413, "grad_norm": 1.5860296844055792, "learning_rate": 5.319044749085105e-06, "loss": 0.5731, "step": 18677 }, { "epoch": 0.6654434686570355, "grad_norm": 1.1635482007959435, "learning_rate": 5.318025071930288e-06, "loss": 0.4283, "step": 18678 }, { "epoch": 0.6654790957835296, "grad_norm": 1.3018245936633195, "learning_rate": 5.317005457119237e-06, "loss": 0.4646, "step": 18679 }, { "epoch": 0.6655147229100237, "grad_norm": 1.55650822431671, "learning_rate": 5.315985904665533e-06, "loss": 0.5516, "step": 18680 }, { "epoch": 0.6655503500365179, "grad_norm": 1.4608161140497207, "learning_rate": 5.31496641458275e-06, "loss": 0.5403, "step": 18681 }, { "epoch": 0.6655859771630119, "grad_norm": 1.450147455880843, "learning_rate": 5.313946986884464e-06, "loss": 0.5622, "step": 18682 }, { "epoch": 0.665621604289506, "grad_norm": 1.4862063859638919, "learning_rate": 5.3129276215842475e-06, "loss": 0.5361, "step": 18683 }, { "epoch": 0.6656572314160002, "grad_norm": 1.5299337842556129, "learning_rate": 5.311908318695676e-06, "loss": 0.5566, "step": 18684 }, { "epoch": 0.6656928585424943, "grad_norm": 1.0754974239455224, "learning_rate": 5.310889078232321e-06, "loss": 0.4103, "step": 18685 }, { "epoch": 0.6657284856689883, "grad_norm": 1.6105197195688654, "learning_rate": 5.3098699002077554e-06, "loss": 0.5076, "step": 18686 }, { "epoch": 0.6657641127954825, "grad_norm": 1.8867354926798008, "learning_rate": 5.308850784635555e-06, "loss": 0.5535, "step": 18687 }, { "epoch": 0.6657997399219766, "grad_norm": 1.2601616000955265, "learning_rate": 5.307831731529281e-06, "loss": 0.4603, "step": 18688 }, { "epoch": 0.6658353670484707, "grad_norm": 1.6733133287957458, "learning_rate": 5.306812740902511e-06, "loss": 0.4774, "step": 18689 }, { "epoch": 0.6658709941749649, "grad_norm": 1.1450053905901594, "learning_rate": 5.3057938127688045e-06, "loss": 0.4019, "step": 18690 }, { "epoch": 0.665906621301459, "grad_norm": 1.3088050665592819, "learning_rate": 5.304774947141737e-06, "loss": 0.6058, "step": 18691 }, { "epoch": 0.665942248427953, "grad_norm": 1.345265125508131, "learning_rate": 5.303756144034871e-06, "loss": 0.51, "step": 18692 }, { "epoch": 0.6659778755544472, "grad_norm": 1.1788388414045976, "learning_rate": 5.302737403461778e-06, "loss": 0.4486, "step": 18693 }, { "epoch": 0.6660135026809413, "grad_norm": 1.316216595652273, "learning_rate": 5.301718725436018e-06, "loss": 0.4322, "step": 18694 }, { "epoch": 0.6660491298074354, "grad_norm": 1.557476052102065, "learning_rate": 5.30070010997116e-06, "loss": 0.47, "step": 18695 }, { "epoch": 0.6660847569339295, "grad_norm": 1.4828768475835452, "learning_rate": 5.299681557080759e-06, "loss": 0.5121, "step": 18696 }, { "epoch": 0.6661203840604236, "grad_norm": 1.1285824646199736, "learning_rate": 5.2986630667783846e-06, "loss": 0.4476, "step": 18697 }, { "epoch": 0.6661560111869177, "grad_norm": 1.2650493143421022, "learning_rate": 5.2976446390776015e-06, "loss": 0.4469, "step": 18698 }, { "epoch": 0.6661916383134119, "grad_norm": 1.3315792040779104, "learning_rate": 5.29662627399196e-06, "loss": 0.4415, "step": 18699 }, { "epoch": 0.666227265439906, "grad_norm": 1.3164745495152932, "learning_rate": 5.2956079715350326e-06, "loss": 0.4521, "step": 18700 }, { "epoch": 0.6662628925664, "grad_norm": 1.5506045176575862, "learning_rate": 5.294589731720372e-06, "loss": 0.6106, "step": 18701 }, { "epoch": 0.6662985196928942, "grad_norm": 1.344696215618841, "learning_rate": 5.293571554561541e-06, "loss": 0.5298, "step": 18702 }, { "epoch": 0.6663341468193883, "grad_norm": 1.224258126329782, "learning_rate": 5.292553440072089e-06, "loss": 0.4327, "step": 18703 }, { "epoch": 0.6663697739458824, "grad_norm": 1.7437997255683972, "learning_rate": 5.291535388265583e-06, "loss": 0.4472, "step": 18704 }, { "epoch": 0.6664054010723766, "grad_norm": 1.561132031899896, "learning_rate": 5.29051739915557e-06, "loss": 0.518, "step": 18705 }, { "epoch": 0.6664410281988706, "grad_norm": 1.6517011985513514, "learning_rate": 5.289499472755611e-06, "loss": 0.5909, "step": 18706 }, { "epoch": 0.6664766553253647, "grad_norm": 1.4024404295879085, "learning_rate": 5.288481609079259e-06, "loss": 0.4738, "step": 18707 }, { "epoch": 0.6665122824518589, "grad_norm": 1.467071248240889, "learning_rate": 5.287463808140069e-06, "loss": 0.5161, "step": 18708 }, { "epoch": 0.666547909578353, "grad_norm": 2.3856513823617416, "learning_rate": 5.286446069951591e-06, "loss": 0.592, "step": 18709 }, { "epoch": 0.666583536704847, "grad_norm": 1.642211654483052, "learning_rate": 5.285428394527377e-06, "loss": 0.5622, "step": 18710 }, { "epoch": 0.6666191638313412, "grad_norm": 1.4953018966886742, "learning_rate": 5.2844107818809835e-06, "loss": 0.5254, "step": 18711 }, { "epoch": 0.6666547909578353, "grad_norm": 1.4291571038010749, "learning_rate": 5.2833932320259515e-06, "loss": 0.409, "step": 18712 }, { "epoch": 0.6666904180843294, "grad_norm": 1.4320836981559906, "learning_rate": 5.282375744975836e-06, "loss": 0.6313, "step": 18713 }, { "epoch": 0.6667260452108236, "grad_norm": 1.5422017777486143, "learning_rate": 5.281358320744186e-06, "loss": 0.4251, "step": 18714 }, { "epoch": 0.6667616723373176, "grad_norm": 1.5679292548800674, "learning_rate": 5.28034095934455e-06, "loss": 0.4152, "step": 18715 }, { "epoch": 0.6667972994638117, "grad_norm": 1.6638904856203875, "learning_rate": 5.27932366079047e-06, "loss": 0.4772, "step": 18716 }, { "epoch": 0.6668329265903059, "grad_norm": 1.4821650796979517, "learning_rate": 5.2783064250954986e-06, "loss": 0.5825, "step": 18717 }, { "epoch": 0.6668685537168, "grad_norm": 1.351383433009835, "learning_rate": 5.277289252273175e-06, "loss": 0.5075, "step": 18718 }, { "epoch": 0.6669041808432941, "grad_norm": 1.3830958104583166, "learning_rate": 5.276272142337048e-06, "loss": 0.4209, "step": 18719 }, { "epoch": 0.6669398079697882, "grad_norm": 1.7920076194452303, "learning_rate": 5.275255095300656e-06, "loss": 0.5666, "step": 18720 }, { "epoch": 0.6669754350962823, "grad_norm": 1.6483457135034958, "learning_rate": 5.274238111177545e-06, "loss": 0.4471, "step": 18721 }, { "epoch": 0.6670110622227764, "grad_norm": 1.574394667038276, "learning_rate": 5.273221189981257e-06, "loss": 0.5009, "step": 18722 }, { "epoch": 0.6670466893492706, "grad_norm": 1.487116313650884, "learning_rate": 5.272204331725331e-06, "loss": 0.4896, "step": 18723 }, { "epoch": 0.6670823164757647, "grad_norm": 1.6452426526330852, "learning_rate": 5.271187536423314e-06, "loss": 0.5998, "step": 18724 }, { "epoch": 0.6671179436022587, "grad_norm": 1.4231167459688105, "learning_rate": 5.270170804088736e-06, "loss": 0.4671, "step": 18725 }, { "epoch": 0.6671535707287529, "grad_norm": 1.0950423656286272, "learning_rate": 5.2691541347351434e-06, "loss": 0.3803, "step": 18726 }, { "epoch": 0.667189197855247, "grad_norm": 1.8146490888128353, "learning_rate": 5.2681375283760676e-06, "loss": 0.4666, "step": 18727 }, { "epoch": 0.6672248249817411, "grad_norm": 1.2995415736363274, "learning_rate": 5.267120985025047e-06, "loss": 0.4319, "step": 18728 }, { "epoch": 0.6672604521082353, "grad_norm": 1.454545753077182, "learning_rate": 5.266104504695617e-06, "loss": 0.5434, "step": 18729 }, { "epoch": 0.6672960792347293, "grad_norm": 1.2548298689544255, "learning_rate": 5.26508808740132e-06, "loss": 0.412, "step": 18730 }, { "epoch": 0.6673317063612234, "grad_norm": 1.1509439289408165, "learning_rate": 5.264071733155681e-06, "loss": 0.4624, "step": 18731 }, { "epoch": 0.6673673334877176, "grad_norm": 1.1283447312495278, "learning_rate": 5.263055441972237e-06, "loss": 0.4208, "step": 18732 }, { "epoch": 0.6674029606142117, "grad_norm": 1.0662685020653584, "learning_rate": 5.262039213864524e-06, "loss": 0.4093, "step": 18733 }, { "epoch": 0.6674385877407057, "grad_norm": 1.168747024895633, "learning_rate": 5.261023048846066e-06, "loss": 0.4223, "step": 18734 }, { "epoch": 0.6674742148671999, "grad_norm": 1.6813136503169137, "learning_rate": 5.260006946930404e-06, "loss": 0.5262, "step": 18735 }, { "epoch": 0.667509841993694, "grad_norm": 1.0612422137046909, "learning_rate": 5.258990908131058e-06, "loss": 0.3846, "step": 18736 }, { "epoch": 0.6675454691201881, "grad_norm": 1.1383056882092082, "learning_rate": 5.2579749324615605e-06, "loss": 0.4079, "step": 18737 }, { "epoch": 0.6675810962466823, "grad_norm": 1.2278952143639859, "learning_rate": 5.256959019935444e-06, "loss": 0.4096, "step": 18738 }, { "epoch": 0.6676167233731763, "grad_norm": 1.4926821542360007, "learning_rate": 5.255943170566234e-06, "loss": 0.54, "step": 18739 }, { "epoch": 0.6676523504996704, "grad_norm": 3.4917227765514394, "learning_rate": 5.2549273843674544e-06, "loss": 0.5019, "step": 18740 }, { "epoch": 0.6676879776261646, "grad_norm": 1.3594559714307506, "learning_rate": 5.253911661352637e-06, "loss": 0.4797, "step": 18741 }, { "epoch": 0.6677236047526587, "grad_norm": 1.8260939196882207, "learning_rate": 5.252896001535299e-06, "loss": 0.5546, "step": 18742 }, { "epoch": 0.6677592318791528, "grad_norm": 1.5211305448232808, "learning_rate": 5.251880404928971e-06, "loss": 0.5156, "step": 18743 }, { "epoch": 0.667794859005647, "grad_norm": 1.5510489633576947, "learning_rate": 5.250864871547171e-06, "loss": 0.5527, "step": 18744 }, { "epoch": 0.667830486132141, "grad_norm": 1.1713073509549399, "learning_rate": 5.249849401403427e-06, "loss": 0.4402, "step": 18745 }, { "epoch": 0.6678661132586351, "grad_norm": 1.1706620198690048, "learning_rate": 5.24883399451126e-06, "loss": 0.4162, "step": 18746 }, { "epoch": 0.6679017403851293, "grad_norm": 1.6375459290340058, "learning_rate": 5.247818650884187e-06, "loss": 0.5335, "step": 18747 }, { "epoch": 0.6679373675116234, "grad_norm": 1.115713574146156, "learning_rate": 5.2468033705357336e-06, "loss": 0.4588, "step": 18748 }, { "epoch": 0.6679729946381174, "grad_norm": 1.4820829733229595, "learning_rate": 5.245788153479412e-06, "loss": 0.5262, "step": 18749 }, { "epoch": 0.6680086217646116, "grad_norm": 1.671593941869656, "learning_rate": 5.244772999728743e-06, "loss": 0.5078, "step": 18750 }, { "epoch": 0.6680442488911057, "grad_norm": 1.1316893609906096, "learning_rate": 5.243757909297247e-06, "loss": 0.4477, "step": 18751 }, { "epoch": 0.6680798760175998, "grad_norm": 1.381965771660657, "learning_rate": 5.242742882198441e-06, "loss": 0.5173, "step": 18752 }, { "epoch": 0.668115503144094, "grad_norm": 1.6079982411028784, "learning_rate": 5.241727918445836e-06, "loss": 0.4262, "step": 18753 }, { "epoch": 0.668151130270588, "grad_norm": 1.0836411430768118, "learning_rate": 5.240713018052953e-06, "loss": 0.4357, "step": 18754 }, { "epoch": 0.6681867573970821, "grad_norm": 1.0155362572315763, "learning_rate": 5.2396981810332995e-06, "loss": 0.3941, "step": 18755 }, { "epoch": 0.6682223845235763, "grad_norm": 1.129839503578134, "learning_rate": 5.238683407400391e-06, "loss": 0.436, "step": 18756 }, { "epoch": 0.6682580116500704, "grad_norm": 1.1629414663513467, "learning_rate": 5.237668697167746e-06, "loss": 0.4354, "step": 18757 }, { "epoch": 0.6682936387765644, "grad_norm": 1.4182955438785183, "learning_rate": 5.236654050348866e-06, "loss": 0.4605, "step": 18758 }, { "epoch": 0.6683292659030586, "grad_norm": 1.5582178537164617, "learning_rate": 5.235639466957266e-06, "loss": 0.5638, "step": 18759 }, { "epoch": 0.6683648930295527, "grad_norm": 1.727188487207464, "learning_rate": 5.234624947006458e-06, "loss": 0.525, "step": 18760 }, { "epoch": 0.6684005201560468, "grad_norm": 1.6178424489082117, "learning_rate": 5.233610490509953e-06, "loss": 0.6113, "step": 18761 }, { "epoch": 0.668436147282541, "grad_norm": 1.171025574965167, "learning_rate": 5.232596097481252e-06, "loss": 0.3982, "step": 18762 }, { "epoch": 0.668471774409035, "grad_norm": 1.5080218122795541, "learning_rate": 5.2315817679338695e-06, "loss": 0.5173, "step": 18763 }, { "epoch": 0.6685074015355291, "grad_norm": 2.1813806205192106, "learning_rate": 5.230567501881305e-06, "loss": 0.5648, "step": 18764 }, { "epoch": 0.6685430286620233, "grad_norm": 1.5777986244719573, "learning_rate": 5.229553299337069e-06, "loss": 0.532, "step": 18765 }, { "epoch": 0.6685786557885174, "grad_norm": 1.68056707300592, "learning_rate": 5.228539160314665e-06, "loss": 0.5965, "step": 18766 }, { "epoch": 0.6686142829150115, "grad_norm": 1.1657504893434731, "learning_rate": 5.227525084827596e-06, "loss": 0.4005, "step": 18767 }, { "epoch": 0.6686499100415056, "grad_norm": 3.667147989766074, "learning_rate": 5.226511072889371e-06, "loss": 0.4851, "step": 18768 }, { "epoch": 0.6686855371679997, "grad_norm": 1.1505824679534709, "learning_rate": 5.225497124513484e-06, "loss": 0.3941, "step": 18769 }, { "epoch": 0.6687211642944938, "grad_norm": 1.3962142875206782, "learning_rate": 5.224483239713443e-06, "loss": 0.4708, "step": 18770 }, { "epoch": 0.668756791420988, "grad_norm": 1.7289542293398212, "learning_rate": 5.22346941850274e-06, "loss": 0.5261, "step": 18771 }, { "epoch": 0.6687924185474821, "grad_norm": 1.1672874876280235, "learning_rate": 5.222455660894887e-06, "loss": 0.4412, "step": 18772 }, { "epoch": 0.6688280456739761, "grad_norm": 1.1068800137436665, "learning_rate": 5.221441966903371e-06, "loss": 0.379, "step": 18773 }, { "epoch": 0.6688636728004703, "grad_norm": 1.6453488998905894, "learning_rate": 5.220428336541695e-06, "loss": 0.52, "step": 18774 }, { "epoch": 0.6688992999269644, "grad_norm": 1.8339171638017506, "learning_rate": 5.219414769823357e-06, "loss": 0.5449, "step": 18775 }, { "epoch": 0.6689349270534585, "grad_norm": 1.104374549608372, "learning_rate": 5.218401266761857e-06, "loss": 0.3961, "step": 18776 }, { "epoch": 0.6689705541799527, "grad_norm": 1.4980917177595088, "learning_rate": 5.217387827370681e-06, "loss": 0.609, "step": 18777 }, { "epoch": 0.6690061813064467, "grad_norm": 1.475288441034602, "learning_rate": 5.2163744516633305e-06, "loss": 0.5815, "step": 18778 }, { "epoch": 0.6690418084329408, "grad_norm": 2.195291156864099, "learning_rate": 5.215361139653301e-06, "loss": 0.5274, "step": 18779 }, { "epoch": 0.669077435559435, "grad_norm": 1.438900491187912, "learning_rate": 5.214347891354079e-06, "loss": 0.5497, "step": 18780 }, { "epoch": 0.6691130626859291, "grad_norm": 1.2598038114322334, "learning_rate": 5.2133347067791585e-06, "loss": 0.514, "step": 18781 }, { "epoch": 0.6691486898124231, "grad_norm": 2.0832384811221147, "learning_rate": 5.212321585942033e-06, "loss": 0.5449, "step": 18782 }, { "epoch": 0.6691843169389173, "grad_norm": 1.1167291357154816, "learning_rate": 5.211308528856196e-06, "loss": 0.4353, "step": 18783 }, { "epoch": 0.6692199440654114, "grad_norm": 1.42465201402701, "learning_rate": 5.210295535535128e-06, "loss": 0.513, "step": 18784 }, { "epoch": 0.6692555711919055, "grad_norm": 1.393587340926649, "learning_rate": 5.209282605992328e-06, "loss": 0.4801, "step": 18785 }, { "epoch": 0.6692911983183997, "grad_norm": 1.4805385081962399, "learning_rate": 5.208269740241273e-06, "loss": 0.5727, "step": 18786 }, { "epoch": 0.6693268254448937, "grad_norm": 1.3564057287510443, "learning_rate": 5.207256938295461e-06, "loss": 0.512, "step": 18787 }, { "epoch": 0.6693624525713878, "grad_norm": 1.2268290098347887, "learning_rate": 5.206244200168364e-06, "loss": 0.4245, "step": 18788 }, { "epoch": 0.669398079697882, "grad_norm": 1.6120840133018404, "learning_rate": 5.2052315258734846e-06, "loss": 0.5441, "step": 18789 }, { "epoch": 0.6694337068243761, "grad_norm": 1.6579179866421916, "learning_rate": 5.204218915424295e-06, "loss": 0.5189, "step": 18790 }, { "epoch": 0.6694693339508702, "grad_norm": 1.4858240827241203, "learning_rate": 5.2032063688342815e-06, "loss": 0.4422, "step": 18791 }, { "epoch": 0.6695049610773643, "grad_norm": 1.132211315476186, "learning_rate": 5.202193886116933e-06, "loss": 0.4559, "step": 18792 }, { "epoch": 0.6695405882038584, "grad_norm": 1.2344937295043004, "learning_rate": 5.201181467285723e-06, "loss": 0.4808, "step": 18793 }, { "epoch": 0.6695762153303525, "grad_norm": 1.39137238495649, "learning_rate": 5.200169112354139e-06, "loss": 0.5017, "step": 18794 }, { "epoch": 0.6696118424568467, "grad_norm": 1.096267727611073, "learning_rate": 5.199156821335654e-06, "loss": 0.3846, "step": 18795 }, { "epoch": 0.6696474695833408, "grad_norm": 1.6214231538168835, "learning_rate": 5.1981445942437515e-06, "loss": 0.5475, "step": 18796 }, { "epoch": 0.6696830967098348, "grad_norm": 1.1931518170008968, "learning_rate": 5.19713243109191e-06, "loss": 0.4295, "step": 18797 }, { "epoch": 0.669718723836329, "grad_norm": 1.1089125368564943, "learning_rate": 5.1961203318936116e-06, "loss": 0.4214, "step": 18798 }, { "epoch": 0.6697543509628231, "grad_norm": 1.0812280690399179, "learning_rate": 5.195108296662325e-06, "loss": 0.4076, "step": 18799 }, { "epoch": 0.6697899780893172, "grad_norm": 1.278344989858223, "learning_rate": 5.194096325411533e-06, "loss": 0.4228, "step": 18800 }, { "epoch": 0.6698256052158114, "grad_norm": 1.1518529374836175, "learning_rate": 5.193084418154703e-06, "loss": 0.4007, "step": 18801 }, { "epoch": 0.6698612323423054, "grad_norm": 2.5415191987627126, "learning_rate": 5.192072574905316e-06, "loss": 0.4107, "step": 18802 }, { "epoch": 0.6698968594687995, "grad_norm": 1.4702481744541152, "learning_rate": 5.191060795676841e-06, "loss": 0.5276, "step": 18803 }, { "epoch": 0.6699324865952937, "grad_norm": 1.5943422472847846, "learning_rate": 5.190049080482754e-06, "loss": 0.5411, "step": 18804 }, { "epoch": 0.6699681137217878, "grad_norm": 1.5773017158929976, "learning_rate": 5.1890374293365295e-06, "loss": 0.4678, "step": 18805 }, { "epoch": 0.6700037408482818, "grad_norm": 1.532007449222639, "learning_rate": 5.188025842251629e-06, "loss": 0.5729, "step": 18806 }, { "epoch": 0.670039367974776, "grad_norm": 1.0469249808329513, "learning_rate": 5.187014319241533e-06, "loss": 0.4126, "step": 18807 }, { "epoch": 0.6700749951012701, "grad_norm": 1.714349380040181, "learning_rate": 5.186002860319701e-06, "loss": 0.5618, "step": 18808 }, { "epoch": 0.6701106222277642, "grad_norm": 1.3000412741947078, "learning_rate": 5.184991465499609e-06, "loss": 0.4467, "step": 18809 }, { "epoch": 0.6701462493542584, "grad_norm": 1.3719899123662527, "learning_rate": 5.183980134794719e-06, "loss": 0.6077, "step": 18810 }, { "epoch": 0.6701818764807524, "grad_norm": 1.604014503042193, "learning_rate": 5.182968868218498e-06, "loss": 0.5038, "step": 18811 }, { "epoch": 0.6702175036072465, "grad_norm": 1.1320926301070973, "learning_rate": 5.181957665784414e-06, "loss": 0.4412, "step": 18812 }, { "epoch": 0.6702531307337407, "grad_norm": 1.1275097917738854, "learning_rate": 5.1809465275059325e-06, "loss": 0.3936, "step": 18813 }, { "epoch": 0.6702887578602348, "grad_norm": 1.4383915193096497, "learning_rate": 5.179935453396519e-06, "loss": 0.5873, "step": 18814 }, { "epoch": 0.6703243849867289, "grad_norm": 1.1425778373621702, "learning_rate": 5.178924443469629e-06, "loss": 0.4104, "step": 18815 }, { "epoch": 0.670360012113223, "grad_norm": 1.3906488394879475, "learning_rate": 5.177913497738736e-06, "loss": 0.5202, "step": 18816 }, { "epoch": 0.6703956392397171, "grad_norm": 1.0648169579538154, "learning_rate": 5.17690261621729e-06, "loss": 0.4428, "step": 18817 }, { "epoch": 0.6704312663662112, "grad_norm": 1.2004384746433532, "learning_rate": 5.175891798918757e-06, "loss": 0.4645, "step": 18818 }, { "epoch": 0.6704668934927054, "grad_norm": 1.1769045303909584, "learning_rate": 5.174881045856597e-06, "loss": 0.4414, "step": 18819 }, { "epoch": 0.6705025206191995, "grad_norm": 1.2486294894654717, "learning_rate": 5.173870357044271e-06, "loss": 0.4525, "step": 18820 }, { "epoch": 0.6705381477456935, "grad_norm": 1.2600193098991033, "learning_rate": 5.1728597324952315e-06, "loss": 0.4442, "step": 18821 }, { "epoch": 0.6705737748721877, "grad_norm": 1.4420265941103265, "learning_rate": 5.171849172222943e-06, "loss": 0.4582, "step": 18822 }, { "epoch": 0.6706094019986818, "grad_norm": 1.5272365606693719, "learning_rate": 5.170838676240853e-06, "loss": 0.4631, "step": 18823 }, { "epoch": 0.6706450291251759, "grad_norm": 1.4114900385408493, "learning_rate": 5.16982824456242e-06, "loss": 0.4621, "step": 18824 }, { "epoch": 0.6706806562516701, "grad_norm": 1.5961866843069241, "learning_rate": 5.1688178772011025e-06, "loss": 0.5352, "step": 18825 }, { "epoch": 0.6707162833781641, "grad_norm": 1.6338927663236122, "learning_rate": 5.167807574170349e-06, "loss": 0.5427, "step": 18826 }, { "epoch": 0.6707519105046582, "grad_norm": 1.88873892085142, "learning_rate": 5.166797335483622e-06, "loss": 0.5372, "step": 18827 }, { "epoch": 0.6707875376311524, "grad_norm": 1.6664112978860806, "learning_rate": 5.165787161154361e-06, "loss": 0.5503, "step": 18828 }, { "epoch": 0.6708231647576465, "grad_norm": 1.5219311965010578, "learning_rate": 5.164777051196027e-06, "loss": 0.4814, "step": 18829 }, { "epoch": 0.6708587918841405, "grad_norm": 1.4624450559244038, "learning_rate": 5.163767005622063e-06, "loss": 0.4743, "step": 18830 }, { "epoch": 0.6708944190106347, "grad_norm": 1.4730906865174267, "learning_rate": 5.162757024445925e-06, "loss": 0.5555, "step": 18831 }, { "epoch": 0.6709300461371288, "grad_norm": 1.2565486050454142, "learning_rate": 5.161747107681054e-06, "loss": 0.47, "step": 18832 }, { "epoch": 0.6709656732636229, "grad_norm": 1.1214358235838495, "learning_rate": 5.160737255340902e-06, "loss": 0.4239, "step": 18833 }, { "epoch": 0.6710013003901171, "grad_norm": 1.1802189643798502, "learning_rate": 5.159727467438918e-06, "loss": 0.4274, "step": 18834 }, { "epoch": 0.6710369275166111, "grad_norm": 1.40640723264097, "learning_rate": 5.1587177439885475e-06, "loss": 0.5589, "step": 18835 }, { "epoch": 0.6710725546431052, "grad_norm": 1.321874938765802, "learning_rate": 5.15770808500323e-06, "loss": 0.4704, "step": 18836 }, { "epoch": 0.6711081817695994, "grad_norm": 1.144926990692852, "learning_rate": 5.156698490496414e-06, "loss": 0.4235, "step": 18837 }, { "epoch": 0.6711438088960935, "grad_norm": 1.2691396545174563, "learning_rate": 5.155688960481549e-06, "loss": 0.403, "step": 18838 }, { "epoch": 0.6711794360225876, "grad_norm": 1.271249842669244, "learning_rate": 5.1546794949720655e-06, "loss": 0.4141, "step": 18839 }, { "epoch": 0.6712150631490817, "grad_norm": 1.627891960852835, "learning_rate": 5.153670093981411e-06, "loss": 0.5803, "step": 18840 }, { "epoch": 0.6712506902755758, "grad_norm": 1.5000130754614243, "learning_rate": 5.152660757523028e-06, "loss": 0.5056, "step": 18841 }, { "epoch": 0.6712863174020699, "grad_norm": 1.9825368434778687, "learning_rate": 5.151651485610359e-06, "loss": 0.4999, "step": 18842 }, { "epoch": 0.6713219445285641, "grad_norm": 1.1336783966033466, "learning_rate": 5.1506422782568345e-06, "loss": 0.4136, "step": 18843 }, { "epoch": 0.6713575716550582, "grad_norm": 1.2920538764779284, "learning_rate": 5.1496331354759015e-06, "loss": 0.4217, "step": 18844 }, { "epoch": 0.6713931987815522, "grad_norm": 1.2880306925888352, "learning_rate": 5.148624057280988e-06, "loss": 0.4407, "step": 18845 }, { "epoch": 0.6714288259080464, "grad_norm": 1.2447413010194135, "learning_rate": 5.147615043685543e-06, "loss": 0.5032, "step": 18846 }, { "epoch": 0.6714644530345405, "grad_norm": 1.4896563785293524, "learning_rate": 5.14660609470299e-06, "loss": 0.503, "step": 18847 }, { "epoch": 0.6715000801610346, "grad_norm": 1.2228132089858528, "learning_rate": 5.145597210346769e-06, "loss": 0.3855, "step": 18848 }, { "epoch": 0.6715357072875288, "grad_norm": 1.21896712271182, "learning_rate": 5.144588390630315e-06, "loss": 0.4057, "step": 18849 }, { "epoch": 0.6715713344140228, "grad_norm": 1.731365207657294, "learning_rate": 5.14357963556706e-06, "loss": 0.6112, "step": 18850 }, { "epoch": 0.6716069615405169, "grad_norm": 1.587349283884632, "learning_rate": 5.14257094517044e-06, "loss": 0.5113, "step": 18851 }, { "epoch": 0.6716425886670111, "grad_norm": 1.2498558098328865, "learning_rate": 5.141562319453879e-06, "loss": 0.4085, "step": 18852 }, { "epoch": 0.6716782157935052, "grad_norm": 1.9569912406481629, "learning_rate": 5.140553758430816e-06, "loss": 0.5409, "step": 18853 }, { "epoch": 0.6717138429199992, "grad_norm": 1.1840896546255892, "learning_rate": 5.139545262114673e-06, "loss": 0.4241, "step": 18854 }, { "epoch": 0.6717494700464934, "grad_norm": 1.2673151584201006, "learning_rate": 5.138536830518881e-06, "loss": 0.456, "step": 18855 }, { "epoch": 0.6717850971729875, "grad_norm": 1.523377902641713, "learning_rate": 5.137528463656869e-06, "loss": 0.4973, "step": 18856 }, { "epoch": 0.6718207242994816, "grad_norm": 1.2483196752694856, "learning_rate": 5.136520161542069e-06, "loss": 0.4268, "step": 18857 }, { "epoch": 0.6718563514259758, "grad_norm": 1.205482860612317, "learning_rate": 5.135511924187897e-06, "loss": 0.4179, "step": 18858 }, { "epoch": 0.6718919785524698, "grad_norm": 1.3155329855258988, "learning_rate": 5.1345037516077846e-06, "loss": 0.4559, "step": 18859 }, { "epoch": 0.6719276056789639, "grad_norm": 1.2847780942736333, "learning_rate": 5.13349564381516e-06, "loss": 0.4149, "step": 18860 }, { "epoch": 0.6719632328054581, "grad_norm": 1.4260243271583708, "learning_rate": 5.132487600823438e-06, "loss": 0.3722, "step": 18861 }, { "epoch": 0.6719988599319522, "grad_norm": 1.4189830827661727, "learning_rate": 5.13147962264605e-06, "loss": 0.4978, "step": 18862 }, { "epoch": 0.6720344870584463, "grad_norm": 1.0510477788022146, "learning_rate": 5.130471709296405e-06, "loss": 0.4058, "step": 18863 }, { "epoch": 0.6720701141849404, "grad_norm": 1.2928805211396999, "learning_rate": 5.129463860787942e-06, "loss": 0.4209, "step": 18864 }, { "epoch": 0.6721057413114345, "grad_norm": 1.1427871399243976, "learning_rate": 5.1284560771340674e-06, "loss": 0.4511, "step": 18865 }, { "epoch": 0.6721413684379286, "grad_norm": 1.5283111883123974, "learning_rate": 5.127448358348208e-06, "loss": 0.4926, "step": 18866 }, { "epoch": 0.6721769955644228, "grad_norm": 1.425725766481441, "learning_rate": 5.126440704443777e-06, "loss": 0.4504, "step": 18867 }, { "epoch": 0.6722126226909169, "grad_norm": 2.127379162953062, "learning_rate": 5.125433115434198e-06, "loss": 0.5789, "step": 18868 }, { "epoch": 0.6722482498174109, "grad_norm": 1.0706312144499015, "learning_rate": 5.12442559133288e-06, "loss": 0.4078, "step": 18869 }, { "epoch": 0.6722838769439051, "grad_norm": 1.2282407374183968, "learning_rate": 5.1234181321532436e-06, "loss": 0.4619, "step": 18870 }, { "epoch": 0.6723195040703992, "grad_norm": 1.8483632111837425, "learning_rate": 5.122410737908703e-06, "loss": 0.5471, "step": 18871 }, { "epoch": 0.6723551311968933, "grad_norm": 1.4153897920184773, "learning_rate": 5.121403408612672e-06, "loss": 0.4275, "step": 18872 }, { "epoch": 0.6723907583233875, "grad_norm": 1.7295120460721434, "learning_rate": 5.1203961442785685e-06, "loss": 0.4422, "step": 18873 }, { "epoch": 0.6724263854498815, "grad_norm": 1.3830670411758241, "learning_rate": 5.119388944919798e-06, "loss": 0.5199, "step": 18874 }, { "epoch": 0.6724620125763756, "grad_norm": 1.9657904576073721, "learning_rate": 5.1183818105497785e-06, "loss": 0.5017, "step": 18875 }, { "epoch": 0.6724976397028698, "grad_norm": 1.1125053222684356, "learning_rate": 5.117374741181913e-06, "loss": 0.4218, "step": 18876 }, { "epoch": 0.6725332668293639, "grad_norm": 1.1837083967819695, "learning_rate": 5.116367736829616e-06, "loss": 0.4168, "step": 18877 }, { "epoch": 0.672568893955858, "grad_norm": 1.3335035741496903, "learning_rate": 5.115360797506297e-06, "loss": 0.4808, "step": 18878 }, { "epoch": 0.6726045210823521, "grad_norm": 1.2981959622091193, "learning_rate": 5.114353923225366e-06, "loss": 0.4022, "step": 18879 }, { "epoch": 0.6726401482088462, "grad_norm": 1.3150414827990158, "learning_rate": 5.113347114000222e-06, "loss": 0.4709, "step": 18880 }, { "epoch": 0.6726757753353403, "grad_norm": 1.4150173197327607, "learning_rate": 5.112340369844283e-06, "loss": 0.5616, "step": 18881 }, { "epoch": 0.6727114024618345, "grad_norm": 1.3611738207022703, "learning_rate": 5.111333690770943e-06, "loss": 0.5359, "step": 18882 }, { "epoch": 0.6727470295883285, "grad_norm": 2.102695681096745, "learning_rate": 5.110327076793613e-06, "loss": 0.5784, "step": 18883 }, { "epoch": 0.6727826567148226, "grad_norm": 1.256355083840714, "learning_rate": 5.1093205279256986e-06, "loss": 0.3949, "step": 18884 }, { "epoch": 0.6728182838413168, "grad_norm": 1.4004102315294567, "learning_rate": 5.1083140441805955e-06, "loss": 0.4377, "step": 18885 }, { "epoch": 0.6728539109678109, "grad_norm": 1.4520058619904128, "learning_rate": 5.107307625571711e-06, "loss": 0.5687, "step": 18886 }, { "epoch": 0.672889538094305, "grad_norm": 1.2095881000862394, "learning_rate": 5.1063012721124446e-06, "loss": 0.4063, "step": 18887 }, { "epoch": 0.6729251652207991, "grad_norm": 1.5696307500280104, "learning_rate": 5.105294983816203e-06, "loss": 0.4938, "step": 18888 }, { "epoch": 0.6729607923472932, "grad_norm": 1.5150426888331645, "learning_rate": 5.104288760696373e-06, "loss": 0.4898, "step": 18889 }, { "epoch": 0.6729964194737873, "grad_norm": 1.432976656379896, "learning_rate": 5.103282602766366e-06, "loss": 0.4295, "step": 18890 }, { "epoch": 0.6730320466002815, "grad_norm": 1.2111434476328156, "learning_rate": 5.102276510039569e-06, "loss": 0.4626, "step": 18891 }, { "epoch": 0.6730676737267756, "grad_norm": 1.280194241038514, "learning_rate": 5.101270482529384e-06, "loss": 0.5619, "step": 18892 }, { "epoch": 0.6731033008532696, "grad_norm": 1.5070605732874505, "learning_rate": 5.100264520249205e-06, "loss": 0.5564, "step": 18893 }, { "epoch": 0.6731389279797638, "grad_norm": 1.3679948673144082, "learning_rate": 5.099258623212431e-06, "loss": 0.5621, "step": 18894 }, { "epoch": 0.6731745551062579, "grad_norm": 1.4155833900618346, "learning_rate": 5.098252791432456e-06, "loss": 0.4423, "step": 18895 }, { "epoch": 0.673210182232752, "grad_norm": 1.1771888240498005, "learning_rate": 5.097247024922669e-06, "loss": 0.4681, "step": 18896 }, { "epoch": 0.6732458093592462, "grad_norm": 1.6352020005727688, "learning_rate": 5.096241323696467e-06, "loss": 0.6512, "step": 18897 }, { "epoch": 0.6732814364857402, "grad_norm": 1.28168388341718, "learning_rate": 5.095235687767236e-06, "loss": 0.4381, "step": 18898 }, { "epoch": 0.6733170636122343, "grad_norm": 1.1286564196787254, "learning_rate": 5.094230117148373e-06, "loss": 0.4256, "step": 18899 }, { "epoch": 0.6733526907387285, "grad_norm": 1.1770817441562929, "learning_rate": 5.093224611853262e-06, "loss": 0.3915, "step": 18900 }, { "epoch": 0.6733883178652226, "grad_norm": 1.1380786282296267, "learning_rate": 5.092219171895296e-06, "loss": 0.3822, "step": 18901 }, { "epoch": 0.6734239449917166, "grad_norm": 1.121754567218867, "learning_rate": 5.09121379728786e-06, "loss": 0.3682, "step": 18902 }, { "epoch": 0.6734595721182108, "grad_norm": 1.3463125480678881, "learning_rate": 5.090208488044348e-06, "loss": 0.4982, "step": 18903 }, { "epoch": 0.6734951992447049, "grad_norm": 1.1393594882733404, "learning_rate": 5.089203244178138e-06, "loss": 0.4158, "step": 18904 }, { "epoch": 0.673530826371199, "grad_norm": 2.1025291624076474, "learning_rate": 5.08819806570262e-06, "loss": 0.5799, "step": 18905 }, { "epoch": 0.6735664534976932, "grad_norm": 1.3008946932789567, "learning_rate": 5.087192952631181e-06, "loss": 0.4011, "step": 18906 }, { "epoch": 0.6736020806241872, "grad_norm": 1.2628291853240512, "learning_rate": 5.086187904977198e-06, "loss": 0.408, "step": 18907 }, { "epoch": 0.6736377077506813, "grad_norm": 1.5411902509376008, "learning_rate": 5.085182922754057e-06, "loss": 0.5417, "step": 18908 }, { "epoch": 0.6736733348771755, "grad_norm": 1.165576599390079, "learning_rate": 5.084178005975141e-06, "loss": 0.4018, "step": 18909 }, { "epoch": 0.6737089620036696, "grad_norm": 1.4680719254399073, "learning_rate": 5.0831731546538354e-06, "loss": 0.4424, "step": 18910 }, { "epoch": 0.6737445891301637, "grad_norm": 1.216691743915766, "learning_rate": 5.0821683688035105e-06, "loss": 0.4309, "step": 18911 }, { "epoch": 0.6737802162566578, "grad_norm": 1.1349086764011553, "learning_rate": 5.081163648437556e-06, "loss": 0.4148, "step": 18912 }, { "epoch": 0.6738158433831519, "grad_norm": 1.555301357950502, "learning_rate": 5.0801589935693415e-06, "loss": 0.544, "step": 18913 }, { "epoch": 0.673851470509646, "grad_norm": 1.5797492130314716, "learning_rate": 5.0791544042122484e-06, "loss": 0.5013, "step": 18914 }, { "epoch": 0.6738870976361402, "grad_norm": 1.2292282761846582, "learning_rate": 5.078149880379655e-06, "loss": 0.4464, "step": 18915 }, { "epoch": 0.6739227247626343, "grad_norm": 1.2481073015308524, "learning_rate": 5.0771454220849395e-06, "loss": 0.5045, "step": 18916 }, { "epoch": 0.6739583518891283, "grad_norm": 1.1542055766942356, "learning_rate": 5.076141029341468e-06, "loss": 0.3903, "step": 18917 }, { "epoch": 0.6739939790156225, "grad_norm": 1.5196233166273876, "learning_rate": 5.075136702162622e-06, "loss": 0.5768, "step": 18918 }, { "epoch": 0.6740296061421166, "grad_norm": 1.5133172639962247, "learning_rate": 5.074132440561778e-06, "loss": 0.5891, "step": 18919 }, { "epoch": 0.6740652332686107, "grad_norm": 1.1808822213048409, "learning_rate": 5.073128244552298e-06, "loss": 0.3817, "step": 18920 }, { "epoch": 0.6741008603951049, "grad_norm": 1.4230424957998429, "learning_rate": 5.072124114147563e-06, "loss": 0.4367, "step": 18921 }, { "epoch": 0.6741364875215989, "grad_norm": 1.5254379516109615, "learning_rate": 5.071120049360936e-06, "loss": 0.5313, "step": 18922 }, { "epoch": 0.674172114648093, "grad_norm": 1.4873163581276883, "learning_rate": 5.070116050205791e-06, "loss": 0.5498, "step": 18923 }, { "epoch": 0.6742077417745872, "grad_norm": 1.6540869802533256, "learning_rate": 5.069112116695496e-06, "loss": 0.4878, "step": 18924 }, { "epoch": 0.6742433689010813, "grad_norm": 1.5219629778384016, "learning_rate": 5.068108248843424e-06, "loss": 0.5131, "step": 18925 }, { "epoch": 0.6742789960275754, "grad_norm": 1.6042388778350878, "learning_rate": 5.067104446662934e-06, "loss": 0.499, "step": 18926 }, { "epoch": 0.6743146231540695, "grad_norm": 1.222588769613498, "learning_rate": 5.066100710167401e-06, "loss": 0.4313, "step": 18927 }, { "epoch": 0.6743502502805636, "grad_norm": 1.1768931157035407, "learning_rate": 5.0650970393701805e-06, "loss": 0.4452, "step": 18928 }, { "epoch": 0.6743858774070577, "grad_norm": 1.578727467047696, "learning_rate": 5.064093434284643e-06, "loss": 0.5578, "step": 18929 }, { "epoch": 0.6744215045335519, "grad_norm": 1.1714227040303344, "learning_rate": 5.06308989492415e-06, "loss": 0.4268, "step": 18930 }, { "epoch": 0.674457131660046, "grad_norm": 1.513051131467644, "learning_rate": 5.062086421302066e-06, "loss": 0.4012, "step": 18931 }, { "epoch": 0.67449275878654, "grad_norm": 2.075243499529373, "learning_rate": 5.061083013431758e-06, "loss": 0.5028, "step": 18932 }, { "epoch": 0.6745283859130342, "grad_norm": 1.342998412157522, "learning_rate": 5.060079671326577e-06, "loss": 0.4506, "step": 18933 }, { "epoch": 0.6745640130395283, "grad_norm": 1.2720121118725485, "learning_rate": 5.059076394999891e-06, "loss": 0.4029, "step": 18934 }, { "epoch": 0.6745996401660224, "grad_norm": 1.1805291699511198, "learning_rate": 5.058073184465054e-06, "loss": 0.3717, "step": 18935 }, { "epoch": 0.6746352672925165, "grad_norm": 1.5781987415147791, "learning_rate": 5.05707003973543e-06, "loss": 0.5558, "step": 18936 }, { "epoch": 0.6746708944190106, "grad_norm": 1.2194716869435862, "learning_rate": 5.05606696082437e-06, "loss": 0.4735, "step": 18937 }, { "epoch": 0.6747065215455047, "grad_norm": 1.596026303848913, "learning_rate": 5.055063947745234e-06, "loss": 0.5767, "step": 18938 }, { "epoch": 0.6747421486719989, "grad_norm": 2.747967225348584, "learning_rate": 5.054061000511376e-06, "loss": 0.4862, "step": 18939 }, { "epoch": 0.674777775798493, "grad_norm": 1.0945012054518481, "learning_rate": 5.053058119136154e-06, "loss": 0.4048, "step": 18940 }, { "epoch": 0.674813402924987, "grad_norm": 1.2393826158502017, "learning_rate": 5.0520553036329255e-06, "loss": 0.4241, "step": 18941 }, { "epoch": 0.6748490300514812, "grad_norm": 1.3760289046093774, "learning_rate": 5.051052554015034e-06, "loss": 0.5066, "step": 18942 }, { "epoch": 0.6748846571779753, "grad_norm": 1.1867821731890262, "learning_rate": 5.050049870295841e-06, "loss": 0.4381, "step": 18943 }, { "epoch": 0.6749202843044694, "grad_norm": 1.029263377222359, "learning_rate": 5.04904725248869e-06, "loss": 0.4172, "step": 18944 }, { "epoch": 0.6749559114309636, "grad_norm": 1.416279792891906, "learning_rate": 5.048044700606935e-06, "loss": 0.5284, "step": 18945 }, { "epoch": 0.6749915385574576, "grad_norm": 1.3658579704659972, "learning_rate": 5.0470422146639275e-06, "loss": 0.4909, "step": 18946 }, { "epoch": 0.6750271656839517, "grad_norm": 1.5105851378281823, "learning_rate": 5.0460397946730174e-06, "loss": 0.5147, "step": 18947 }, { "epoch": 0.6750627928104459, "grad_norm": 1.1604732974974754, "learning_rate": 5.045037440647547e-06, "loss": 0.3858, "step": 18948 }, { "epoch": 0.67509841993694, "grad_norm": 1.478669547453807, "learning_rate": 5.044035152600869e-06, "loss": 0.511, "step": 18949 }, { "epoch": 0.675134047063434, "grad_norm": 1.3253846575607586, "learning_rate": 5.043032930546323e-06, "loss": 0.4554, "step": 18950 }, { "epoch": 0.6751696741899282, "grad_norm": 1.2601677511777725, "learning_rate": 5.042030774497259e-06, "loss": 0.3829, "step": 18951 }, { "epoch": 0.6752053013164223, "grad_norm": 1.131590081843459, "learning_rate": 5.041028684467022e-06, "loss": 0.4679, "step": 18952 }, { "epoch": 0.6752409284429164, "grad_norm": 1.1314572613267888, "learning_rate": 5.040026660468953e-06, "loss": 0.3884, "step": 18953 }, { "epoch": 0.6752765555694106, "grad_norm": 8.589596213072397, "learning_rate": 5.0390247025164e-06, "loss": 0.476, "step": 18954 }, { "epoch": 0.6753121826959047, "grad_norm": 1.299335307054546, "learning_rate": 5.038022810622696e-06, "loss": 0.4695, "step": 18955 }, { "epoch": 0.6753478098223987, "grad_norm": 1.4310533633123994, "learning_rate": 5.037020984801191e-06, "loss": 0.5107, "step": 18956 }, { "epoch": 0.6753834369488929, "grad_norm": 1.1562065004019921, "learning_rate": 5.036019225065215e-06, "loss": 0.4724, "step": 18957 }, { "epoch": 0.675419064075387, "grad_norm": 1.1327631044012163, "learning_rate": 5.035017531428118e-06, "loss": 0.3709, "step": 18958 }, { "epoch": 0.6754546912018811, "grad_norm": 1.5628165274511396, "learning_rate": 5.034015903903229e-06, "loss": 0.5342, "step": 18959 }, { "epoch": 0.6754903183283752, "grad_norm": 1.3019314133012025, "learning_rate": 5.033014342503889e-06, "loss": 0.4605, "step": 18960 }, { "epoch": 0.6755259454548693, "grad_norm": 1.1819873327255743, "learning_rate": 5.032012847243435e-06, "loss": 0.3989, "step": 18961 }, { "epoch": 0.6755615725813634, "grad_norm": 1.1142297691273066, "learning_rate": 5.031011418135205e-06, "loss": 0.4197, "step": 18962 }, { "epoch": 0.6755971997078576, "grad_norm": 1.297255312949376, "learning_rate": 5.030010055192529e-06, "loss": 0.4063, "step": 18963 }, { "epoch": 0.6756328268343517, "grad_norm": 1.6353981709182632, "learning_rate": 5.029008758428742e-06, "loss": 0.4886, "step": 18964 }, { "epoch": 0.6756684539608457, "grad_norm": 1.1454581153916483, "learning_rate": 5.028007527857182e-06, "loss": 0.3965, "step": 18965 }, { "epoch": 0.6757040810873399, "grad_norm": 1.9084474543636363, "learning_rate": 5.027006363491174e-06, "loss": 0.4589, "step": 18966 }, { "epoch": 0.675739708213834, "grad_norm": 1.1304738105502254, "learning_rate": 5.02600526534405e-06, "loss": 0.4402, "step": 18967 }, { "epoch": 0.6757753353403281, "grad_norm": 1.112763910316492, "learning_rate": 5.025004233429145e-06, "loss": 0.4456, "step": 18968 }, { "epoch": 0.6758109624668223, "grad_norm": 1.2407637153787392, "learning_rate": 5.0240032677597875e-06, "loss": 0.3741, "step": 18969 }, { "epoch": 0.6758465895933163, "grad_norm": 1.3387578289775925, "learning_rate": 5.023002368349301e-06, "loss": 0.4953, "step": 18970 }, { "epoch": 0.6758822167198104, "grad_norm": 1.1497030117921847, "learning_rate": 5.022001535211022e-06, "loss": 0.3541, "step": 18971 }, { "epoch": 0.6759178438463046, "grad_norm": 1.2472782720957925, "learning_rate": 5.021000768358267e-06, "loss": 0.4052, "step": 18972 }, { "epoch": 0.6759534709727987, "grad_norm": 1.4646874073618328, "learning_rate": 5.020000067804372e-06, "loss": 0.4255, "step": 18973 }, { "epoch": 0.6759890980992928, "grad_norm": 1.7413489013961865, "learning_rate": 5.018999433562652e-06, "loss": 0.5182, "step": 18974 }, { "epoch": 0.6760247252257869, "grad_norm": 1.4006854194169835, "learning_rate": 5.0179988656464366e-06, "loss": 0.4766, "step": 18975 }, { "epoch": 0.676060352352281, "grad_norm": 1.496749482892645, "learning_rate": 5.0169983640690476e-06, "loss": 0.5605, "step": 18976 }, { "epoch": 0.6760959794787751, "grad_norm": 1.261544272688767, "learning_rate": 5.0159979288438075e-06, "loss": 0.3899, "step": 18977 }, { "epoch": 0.6761316066052693, "grad_norm": 1.1193820150208516, "learning_rate": 5.014997559984045e-06, "loss": 0.4179, "step": 18978 }, { "epoch": 0.6761672337317634, "grad_norm": 1.2816267879635268, "learning_rate": 5.013997257503069e-06, "loss": 0.4366, "step": 18979 }, { "epoch": 0.6762028608582574, "grad_norm": 1.3193722711135198, "learning_rate": 5.012997021414208e-06, "loss": 0.4858, "step": 18980 }, { "epoch": 0.6762384879847516, "grad_norm": 1.206297012962852, "learning_rate": 5.011996851730774e-06, "loss": 0.4117, "step": 18981 }, { "epoch": 0.6762741151112457, "grad_norm": 1.7320471014660428, "learning_rate": 5.010996748466088e-06, "loss": 0.5043, "step": 18982 }, { "epoch": 0.6763097422377398, "grad_norm": 1.6907142365963235, "learning_rate": 5.009996711633469e-06, "loss": 0.5659, "step": 18983 }, { "epoch": 0.676345369364234, "grad_norm": 1.0987166302582942, "learning_rate": 5.008996741246234e-06, "loss": 0.4099, "step": 18984 }, { "epoch": 0.676380996490728, "grad_norm": 1.5709096764539332, "learning_rate": 5.007996837317692e-06, "loss": 0.5509, "step": 18985 }, { "epoch": 0.6764166236172221, "grad_norm": 1.4393938819287033, "learning_rate": 5.006996999861163e-06, "loss": 0.5296, "step": 18986 }, { "epoch": 0.6764522507437163, "grad_norm": 1.6582970430933608, "learning_rate": 5.00599722888996e-06, "loss": 0.5296, "step": 18987 }, { "epoch": 0.6764878778702104, "grad_norm": 1.1908854142228325, "learning_rate": 5.004997524417392e-06, "loss": 0.4359, "step": 18988 }, { "epoch": 0.6765235049967044, "grad_norm": 1.332135635823354, "learning_rate": 5.003997886456777e-06, "loss": 0.4292, "step": 18989 }, { "epoch": 0.6765591321231986, "grad_norm": 1.2038419758653247, "learning_rate": 5.002998315021415e-06, "loss": 0.4399, "step": 18990 }, { "epoch": 0.6765947592496927, "grad_norm": 1.4653377359072608, "learning_rate": 5.00199881012463e-06, "loss": 0.5675, "step": 18991 }, { "epoch": 0.6766303863761868, "grad_norm": 1.807508960999355, "learning_rate": 5.00099937177972e-06, "loss": 0.5682, "step": 18992 }, { "epoch": 0.676666013502681, "grad_norm": 1.1760242484316943, "learning_rate": 5.000000000000003e-06, "loss": 0.4268, "step": 18993 }, { "epoch": 0.676701640629175, "grad_norm": 2.04156906084583, "learning_rate": 4.999000694798776e-06, "loss": 0.538, "step": 18994 }, { "epoch": 0.6767372677556691, "grad_norm": 1.5713945232410653, "learning_rate": 4.998001456189353e-06, "loss": 0.6101, "step": 18995 }, { "epoch": 0.6767728948821633, "grad_norm": 1.2661103464757797, "learning_rate": 4.997002284185034e-06, "loss": 0.4382, "step": 18996 }, { "epoch": 0.6768085220086574, "grad_norm": 1.2387587414752885, "learning_rate": 4.996003178799127e-06, "loss": 0.4683, "step": 18997 }, { "epoch": 0.6768441491351515, "grad_norm": 1.160418547676233, "learning_rate": 4.995004140044935e-06, "loss": 0.5029, "step": 18998 }, { "epoch": 0.6768797762616456, "grad_norm": 2.1974562520181444, "learning_rate": 4.99400516793576e-06, "loss": 0.5251, "step": 18999 }, { "epoch": 0.6769154033881397, "grad_norm": 1.3311588350015624, "learning_rate": 4.993006262484911e-06, "loss": 0.4547, "step": 19000 }, { "epoch": 0.6769510305146338, "grad_norm": 1.324033042963558, "learning_rate": 4.992007423705677e-06, "loss": 0.4465, "step": 19001 }, { "epoch": 0.676986657641128, "grad_norm": 1.9844083849916347, "learning_rate": 4.99100865161137e-06, "loss": 0.4317, "step": 19002 }, { "epoch": 0.677022284767622, "grad_norm": 1.1264860846059646, "learning_rate": 4.990009946215279e-06, "loss": 0.3908, "step": 19003 }, { "epoch": 0.6770579118941161, "grad_norm": 2.051692512719627, "learning_rate": 4.989011307530709e-06, "loss": 0.5772, "step": 19004 }, { "epoch": 0.6770935390206103, "grad_norm": 1.4197540590136108, "learning_rate": 4.988012735570955e-06, "loss": 0.5923, "step": 19005 }, { "epoch": 0.6771291661471044, "grad_norm": 1.1513453857333087, "learning_rate": 4.987014230349319e-06, "loss": 0.4252, "step": 19006 }, { "epoch": 0.6771647932735985, "grad_norm": 1.411687746203949, "learning_rate": 4.986015791879088e-06, "loss": 0.458, "step": 19007 }, { "epoch": 0.6772004204000927, "grad_norm": 1.2850687194608863, "learning_rate": 4.9850174201735645e-06, "loss": 0.4847, "step": 19008 }, { "epoch": 0.6772360475265867, "grad_norm": 1.2342396310615507, "learning_rate": 4.984019115246036e-06, "loss": 0.4759, "step": 19009 }, { "epoch": 0.6772716746530808, "grad_norm": 1.565445949447377, "learning_rate": 4.983020877109799e-06, "loss": 0.4927, "step": 19010 }, { "epoch": 0.677307301779575, "grad_norm": 1.711437130453092, "learning_rate": 4.98202270577815e-06, "loss": 0.5345, "step": 19011 }, { "epoch": 0.6773429289060691, "grad_norm": 1.1840036892295753, "learning_rate": 4.981024601264372e-06, "loss": 0.3832, "step": 19012 }, { "epoch": 0.6773785560325631, "grad_norm": 2.3764558200406025, "learning_rate": 4.980026563581759e-06, "loss": 0.449, "step": 19013 }, { "epoch": 0.6774141831590573, "grad_norm": 1.658216517701185, "learning_rate": 4.979028592743601e-06, "loss": 0.4406, "step": 19014 }, { "epoch": 0.6774498102855514, "grad_norm": 1.5876612585364833, "learning_rate": 4.978030688763191e-06, "loss": 0.5028, "step": 19015 }, { "epoch": 0.6774854374120455, "grad_norm": 1.2162607587117138, "learning_rate": 4.977032851653808e-06, "loss": 0.4335, "step": 19016 }, { "epoch": 0.6775210645385397, "grad_norm": 1.7611603445753155, "learning_rate": 4.9760350814287486e-06, "loss": 0.4725, "step": 19017 }, { "epoch": 0.6775566916650337, "grad_norm": 1.1393718629045835, "learning_rate": 4.975037378101289e-06, "loss": 0.4206, "step": 19018 }, { "epoch": 0.6775923187915278, "grad_norm": 1.2822207495863776, "learning_rate": 4.974039741684719e-06, "loss": 0.459, "step": 19019 }, { "epoch": 0.677627945918022, "grad_norm": 1.5643195978658566, "learning_rate": 4.973042172192323e-06, "loss": 0.4941, "step": 19020 }, { "epoch": 0.6776635730445161, "grad_norm": 1.2045449011057585, "learning_rate": 4.972044669637384e-06, "loss": 0.4342, "step": 19021 }, { "epoch": 0.6776992001710102, "grad_norm": 1.57534150047861, "learning_rate": 4.971047234033187e-06, "loss": 0.4746, "step": 19022 }, { "epoch": 0.6777348272975043, "grad_norm": 1.6191183701599055, "learning_rate": 4.970049865393009e-06, "loss": 0.5676, "step": 19023 }, { "epoch": 0.6777704544239984, "grad_norm": 1.266262383652788, "learning_rate": 4.969052563730136e-06, "loss": 0.4034, "step": 19024 }, { "epoch": 0.6778060815504925, "grad_norm": 1.8571465267841931, "learning_rate": 4.9680553290578405e-06, "loss": 0.451, "step": 19025 }, { "epoch": 0.6778417086769867, "grad_norm": 1.4559513541206894, "learning_rate": 4.9670581613894094e-06, "loss": 0.5241, "step": 19026 }, { "epoch": 0.6778773358034808, "grad_norm": 1.48887653766732, "learning_rate": 4.96606106073811e-06, "loss": 0.491, "step": 19027 }, { "epoch": 0.6779129629299749, "grad_norm": 1.3309451663319396, "learning_rate": 4.965064027117232e-06, "loss": 0.5102, "step": 19028 }, { "epoch": 0.677948590056469, "grad_norm": 1.1781393986182715, "learning_rate": 4.964067060540043e-06, "loss": 0.4873, "step": 19029 }, { "epoch": 0.6779842171829631, "grad_norm": 1.0677962233380376, "learning_rate": 4.963070161019825e-06, "loss": 0.4242, "step": 19030 }, { "epoch": 0.6780198443094573, "grad_norm": 1.0313452117716329, "learning_rate": 4.962073328569844e-06, "loss": 0.4034, "step": 19031 }, { "epoch": 0.6780554714359514, "grad_norm": 2.3403061512166996, "learning_rate": 4.961076563203379e-06, "loss": 0.5581, "step": 19032 }, { "epoch": 0.6780910985624454, "grad_norm": 1.5250516193555672, "learning_rate": 4.960079864933705e-06, "loss": 0.5329, "step": 19033 }, { "epoch": 0.6781267256889396, "grad_norm": 1.7334142060150568, "learning_rate": 4.959083233774085e-06, "loss": 0.562, "step": 19034 }, { "epoch": 0.6781623528154337, "grad_norm": 1.2568842216419935, "learning_rate": 4.958086669737796e-06, "loss": 0.4262, "step": 19035 }, { "epoch": 0.6781979799419278, "grad_norm": 1.1583677393312803, "learning_rate": 4.957090172838107e-06, "loss": 0.4155, "step": 19036 }, { "epoch": 0.678233607068422, "grad_norm": 1.167801475321994, "learning_rate": 4.956093743088292e-06, "loss": 0.5012, "step": 19037 }, { "epoch": 0.678269234194916, "grad_norm": 1.2169383456067395, "learning_rate": 4.955097380501609e-06, "loss": 0.4387, "step": 19038 }, { "epoch": 0.6783048613214101, "grad_norm": 1.235901318054301, "learning_rate": 4.954101085091335e-06, "loss": 0.4078, "step": 19039 }, { "epoch": 0.6783404884479043, "grad_norm": 1.478079510928204, "learning_rate": 4.953104856870729e-06, "loss": 0.5835, "step": 19040 }, { "epoch": 0.6783761155743984, "grad_norm": 1.2593923467925803, "learning_rate": 4.952108695853057e-06, "loss": 0.4604, "step": 19041 }, { "epoch": 0.6784117427008924, "grad_norm": 1.8624590546442623, "learning_rate": 4.951112602051587e-06, "loss": 0.5584, "step": 19042 }, { "epoch": 0.6784473698273866, "grad_norm": 1.1782237635073753, "learning_rate": 4.950116575479587e-06, "loss": 0.3985, "step": 19043 }, { "epoch": 0.6784829969538807, "grad_norm": 1.3818981506011092, "learning_rate": 4.949120616150309e-06, "loss": 0.427, "step": 19044 }, { "epoch": 0.6785186240803748, "grad_norm": 1.455808561973934, "learning_rate": 4.948124724077021e-06, "loss": 0.5497, "step": 19045 }, { "epoch": 0.678554251206869, "grad_norm": 1.3602825784184338, "learning_rate": 4.947128899272988e-06, "loss": 0.4847, "step": 19046 }, { "epoch": 0.678589878333363, "grad_norm": 1.3582796273869877, "learning_rate": 4.94613314175146e-06, "loss": 0.5403, "step": 19047 }, { "epoch": 0.6786255054598571, "grad_norm": 1.4047043575023463, "learning_rate": 4.945137451525707e-06, "loss": 0.4537, "step": 19048 }, { "epoch": 0.6786611325863513, "grad_norm": 1.2110182333780544, "learning_rate": 4.944141828608977e-06, "loss": 0.3877, "step": 19049 }, { "epoch": 0.6786967597128454, "grad_norm": 1.1018600336064643, "learning_rate": 4.943146273014533e-06, "loss": 0.4237, "step": 19050 }, { "epoch": 0.6787323868393395, "grad_norm": 1.3350460689720678, "learning_rate": 4.9421507847556305e-06, "loss": 0.398, "step": 19051 }, { "epoch": 0.6787680139658336, "grad_norm": 1.2017381101348306, "learning_rate": 4.94115536384553e-06, "loss": 0.4653, "step": 19052 }, { "epoch": 0.6788036410923277, "grad_norm": 1.6899867670958613, "learning_rate": 4.940160010297476e-06, "loss": 0.4805, "step": 19053 }, { "epoch": 0.6788392682188218, "grad_norm": 1.1963230351136231, "learning_rate": 4.9391647241247345e-06, "loss": 0.4235, "step": 19054 }, { "epoch": 0.678874895345316, "grad_norm": 1.1791206434786103, "learning_rate": 4.9381695053405464e-06, "loss": 0.4528, "step": 19055 }, { "epoch": 0.67891052247181, "grad_norm": 1.325577832148518, "learning_rate": 4.937174353958171e-06, "loss": 0.4828, "step": 19056 }, { "epoch": 0.6789461495983041, "grad_norm": 1.3248842944757582, "learning_rate": 4.936179269990856e-06, "loss": 0.4274, "step": 19057 }, { "epoch": 0.6789817767247983, "grad_norm": 1.1932578031544567, "learning_rate": 4.935184253451855e-06, "loss": 0.4221, "step": 19058 }, { "epoch": 0.6790174038512924, "grad_norm": 1.2613595454095337, "learning_rate": 4.934189304354418e-06, "loss": 0.3933, "step": 19059 }, { "epoch": 0.6790530309777865, "grad_norm": 1.3417343001508528, "learning_rate": 4.933194422711788e-06, "loss": 0.4091, "step": 19060 }, { "epoch": 0.6790886581042807, "grad_norm": 1.2239359198284163, "learning_rate": 4.932199608537221e-06, "loss": 0.4124, "step": 19061 }, { "epoch": 0.6791242852307747, "grad_norm": 1.4746697119869128, "learning_rate": 4.931204861843952e-06, "loss": 0.4713, "step": 19062 }, { "epoch": 0.6791599123572688, "grad_norm": 1.3294376598599358, "learning_rate": 4.930210182645239e-06, "loss": 0.4763, "step": 19063 }, { "epoch": 0.679195539483763, "grad_norm": 1.3728358481300853, "learning_rate": 4.929215570954315e-06, "loss": 0.4704, "step": 19064 }, { "epoch": 0.6792311666102571, "grad_norm": 1.3212770372568183, "learning_rate": 4.9282210267844325e-06, "loss": 0.4564, "step": 19065 }, { "epoch": 0.6792667937367511, "grad_norm": 1.3382563780292738, "learning_rate": 4.9272265501488304e-06, "loss": 0.4383, "step": 19066 }, { "epoch": 0.6793024208632453, "grad_norm": 1.2417739965579373, "learning_rate": 4.9262321410607526e-06, "loss": 0.4058, "step": 19067 }, { "epoch": 0.6793380479897394, "grad_norm": 1.2048193586843128, "learning_rate": 4.925237799533445e-06, "loss": 0.4692, "step": 19068 }, { "epoch": 0.6793736751162335, "grad_norm": 1.2225310315709568, "learning_rate": 4.9242435255801376e-06, "loss": 0.4269, "step": 19069 }, { "epoch": 0.6794093022427277, "grad_norm": 1.3659502329741007, "learning_rate": 4.92324931921408e-06, "loss": 0.4187, "step": 19070 }, { "epoch": 0.6794449293692217, "grad_norm": 1.317962275442871, "learning_rate": 4.922255180448504e-06, "loss": 0.5335, "step": 19071 }, { "epoch": 0.6794805564957158, "grad_norm": 1.3399473378033868, "learning_rate": 4.921261109296648e-06, "loss": 0.4602, "step": 19072 }, { "epoch": 0.67951618362221, "grad_norm": 1.8636948433553597, "learning_rate": 4.92026710577175e-06, "loss": 0.5919, "step": 19073 }, { "epoch": 0.6795518107487041, "grad_norm": 1.5165941612112446, "learning_rate": 4.91927316988705e-06, "loss": 0.42, "step": 19074 }, { "epoch": 0.6795874378751982, "grad_norm": 1.6159902244740356, "learning_rate": 4.918279301655775e-06, "loss": 0.5036, "step": 19075 }, { "epoch": 0.6796230650016923, "grad_norm": 1.5797047701650935, "learning_rate": 4.917285501091167e-06, "loss": 0.4535, "step": 19076 }, { "epoch": 0.6796586921281864, "grad_norm": 1.3956542366469447, "learning_rate": 4.916291768206451e-06, "loss": 0.5099, "step": 19077 }, { "epoch": 0.6796943192546805, "grad_norm": 1.3873528839426725, "learning_rate": 4.9152981030148635e-06, "loss": 0.5454, "step": 19078 }, { "epoch": 0.6797299463811747, "grad_norm": 1.2063243709840807, "learning_rate": 4.9143045055296356e-06, "loss": 0.3899, "step": 19079 }, { "epoch": 0.6797655735076688, "grad_norm": 1.2156473117736688, "learning_rate": 4.913310975763999e-06, "loss": 0.4371, "step": 19080 }, { "epoch": 0.6798012006341628, "grad_norm": 1.735920840762787, "learning_rate": 4.912317513731185e-06, "loss": 0.5865, "step": 19081 }, { "epoch": 0.679836827760657, "grad_norm": 1.402969451212804, "learning_rate": 4.911324119444416e-06, "loss": 0.489, "step": 19082 }, { "epoch": 0.6798724548871511, "grad_norm": 1.2598899652718223, "learning_rate": 4.910330792916926e-06, "loss": 0.5019, "step": 19083 }, { "epoch": 0.6799080820136452, "grad_norm": 1.394638352651157, "learning_rate": 4.909337534161936e-06, "loss": 0.5404, "step": 19084 }, { "epoch": 0.6799437091401394, "grad_norm": 1.1533536148253944, "learning_rate": 4.908344343192679e-06, "loss": 0.386, "step": 19085 }, { "epoch": 0.6799793362666334, "grad_norm": 3.2625378735042294, "learning_rate": 4.907351220022371e-06, "loss": 0.4263, "step": 19086 }, { "epoch": 0.6800149633931275, "grad_norm": 1.6315464222554352, "learning_rate": 4.906358164664242e-06, "loss": 0.5731, "step": 19087 }, { "epoch": 0.6800505905196217, "grad_norm": 1.501995027610568, "learning_rate": 4.905365177131514e-06, "loss": 0.4405, "step": 19088 }, { "epoch": 0.6800862176461158, "grad_norm": 1.570543795124663, "learning_rate": 4.904372257437413e-06, "loss": 0.5763, "step": 19089 }, { "epoch": 0.6801218447726098, "grad_norm": 1.44011609599122, "learning_rate": 4.903379405595154e-06, "loss": 0.5297, "step": 19090 }, { "epoch": 0.680157471899104, "grad_norm": 1.3072095030737942, "learning_rate": 4.902386621617961e-06, "loss": 0.3966, "step": 19091 }, { "epoch": 0.6801930990255981, "grad_norm": 1.3920909279352205, "learning_rate": 4.9013939055190554e-06, "loss": 0.4657, "step": 19092 }, { "epoch": 0.6802287261520922, "grad_norm": 1.4548134392611927, "learning_rate": 4.90040125731165e-06, "loss": 0.5052, "step": 19093 }, { "epoch": 0.6802643532785864, "grad_norm": 1.3349194718428792, "learning_rate": 4.899408677008968e-06, "loss": 0.3905, "step": 19094 }, { "epoch": 0.6802999804050804, "grad_norm": 1.5741248015907179, "learning_rate": 4.898416164624222e-06, "loss": 0.458, "step": 19095 }, { "epoch": 0.6803356075315745, "grad_norm": 1.4006553750031956, "learning_rate": 4.897423720170637e-06, "loss": 0.4457, "step": 19096 }, { "epoch": 0.6803712346580687, "grad_norm": 1.6318531538723606, "learning_rate": 4.896431343661416e-06, "loss": 0.5254, "step": 19097 }, { "epoch": 0.6804068617845628, "grad_norm": 1.4479559201886545, "learning_rate": 4.895439035109784e-06, "loss": 0.3985, "step": 19098 }, { "epoch": 0.6804424889110569, "grad_norm": 1.2843480926743662, "learning_rate": 4.894446794528944e-06, "loss": 0.4246, "step": 19099 }, { "epoch": 0.680478116037551, "grad_norm": 1.2954225236095174, "learning_rate": 4.893454621932118e-06, "loss": 0.4144, "step": 19100 }, { "epoch": 0.6805137431640451, "grad_norm": 1.5774342296620065, "learning_rate": 4.892462517332509e-06, "loss": 0.5446, "step": 19101 }, { "epoch": 0.6805493702905392, "grad_norm": 1.2250454081973818, "learning_rate": 4.8914704807433275e-06, "loss": 0.4222, "step": 19102 }, { "epoch": 0.6805849974170334, "grad_norm": 1.0505459015385938, "learning_rate": 4.890478512177796e-06, "loss": 0.373, "step": 19103 }, { "epoch": 0.6806206245435275, "grad_norm": 1.3276582391180851, "learning_rate": 4.8894866116491095e-06, "loss": 0.4252, "step": 19104 }, { "epoch": 0.6806562516700215, "grad_norm": 1.290349152333242, "learning_rate": 4.888494779170484e-06, "loss": 0.4682, "step": 19105 }, { "epoch": 0.6806918787965157, "grad_norm": 1.6764422535845016, "learning_rate": 4.887503014755121e-06, "loss": 0.4815, "step": 19106 }, { "epoch": 0.6807275059230098, "grad_norm": 1.7477814816287365, "learning_rate": 4.8865113184162315e-06, "loss": 0.514, "step": 19107 }, { "epoch": 0.6807631330495039, "grad_norm": 1.6420840730088893, "learning_rate": 4.885519690167015e-06, "loss": 0.6459, "step": 19108 }, { "epoch": 0.680798760175998, "grad_norm": 1.1981696117101759, "learning_rate": 4.884528130020677e-06, "loss": 0.4434, "step": 19109 }, { "epoch": 0.6808343873024921, "grad_norm": 1.5804090675565523, "learning_rate": 4.8835366379904235e-06, "loss": 0.5217, "step": 19110 }, { "epoch": 0.6808700144289862, "grad_norm": 1.1782299635213893, "learning_rate": 4.88254521408946e-06, "loss": 0.4067, "step": 19111 }, { "epoch": 0.6809056415554804, "grad_norm": 1.9713541867406295, "learning_rate": 4.88155385833098e-06, "loss": 0.4966, "step": 19112 }, { "epoch": 0.6809412686819745, "grad_norm": 1.2889852293091708, "learning_rate": 4.880562570728188e-06, "loss": 0.452, "step": 19113 }, { "epoch": 0.6809768958084685, "grad_norm": 1.3952001245068397, "learning_rate": 4.879571351294287e-06, "loss": 0.4843, "step": 19114 }, { "epoch": 0.6810125229349627, "grad_norm": 1.8200269737180097, "learning_rate": 4.878580200042469e-06, "loss": 0.4217, "step": 19115 }, { "epoch": 0.6810481500614568, "grad_norm": 1.3124439954421177, "learning_rate": 4.877589116985935e-06, "loss": 0.4621, "step": 19116 }, { "epoch": 0.6810837771879509, "grad_norm": 1.835063198067583, "learning_rate": 4.876598102137882e-06, "loss": 0.5775, "step": 19117 }, { "epoch": 0.6811194043144451, "grad_norm": 1.7244788200111918, "learning_rate": 4.875607155511511e-06, "loss": 0.5659, "step": 19118 }, { "epoch": 0.6811550314409391, "grad_norm": 1.4098828210735994, "learning_rate": 4.874616277120008e-06, "loss": 0.4825, "step": 19119 }, { "epoch": 0.6811906585674332, "grad_norm": 1.3532132516335995, "learning_rate": 4.873625466976575e-06, "loss": 0.3941, "step": 19120 }, { "epoch": 0.6812262856939274, "grad_norm": 1.4948220923021691, "learning_rate": 4.8726347250943995e-06, "loss": 0.512, "step": 19121 }, { "epoch": 0.6812619128204215, "grad_norm": 1.2232547064043278, "learning_rate": 4.871644051486679e-06, "loss": 0.4318, "step": 19122 }, { "epoch": 0.6812975399469156, "grad_norm": 1.0289512501640259, "learning_rate": 4.8706534461666e-06, "loss": 0.3902, "step": 19123 }, { "epoch": 0.6813331670734097, "grad_norm": 1.1687759955476615, "learning_rate": 4.869662909147354e-06, "loss": 0.3749, "step": 19124 }, { "epoch": 0.6813687941999038, "grad_norm": 1.3297692804979304, "learning_rate": 4.868672440442134e-06, "loss": 0.4431, "step": 19125 }, { "epoch": 0.6814044213263979, "grad_norm": 1.330455782568953, "learning_rate": 4.8676820400641255e-06, "loss": 0.4167, "step": 19126 }, { "epoch": 0.6814400484528921, "grad_norm": 1.4007445577717956, "learning_rate": 4.866691708026522e-06, "loss": 0.4979, "step": 19127 }, { "epoch": 0.6814756755793862, "grad_norm": 1.444290307353439, "learning_rate": 4.865701444342503e-06, "loss": 0.5042, "step": 19128 }, { "epoch": 0.6815113027058802, "grad_norm": 1.4322640895227896, "learning_rate": 4.864711249025261e-06, "loss": 0.487, "step": 19129 }, { "epoch": 0.6815469298323744, "grad_norm": 1.2239554692930952, "learning_rate": 4.863721122087975e-06, "loss": 0.4444, "step": 19130 }, { "epoch": 0.6815825569588685, "grad_norm": 1.2243857657768085, "learning_rate": 4.862731063543833e-06, "loss": 0.4278, "step": 19131 }, { "epoch": 0.6816181840853626, "grad_norm": 1.3710959158749574, "learning_rate": 4.8617410734060166e-06, "loss": 0.4301, "step": 19132 }, { "epoch": 0.6816538112118568, "grad_norm": 1.6454211490126518, "learning_rate": 4.860751151687713e-06, "loss": 0.514, "step": 19133 }, { "epoch": 0.6816894383383508, "grad_norm": 1.3803122562557473, "learning_rate": 4.859761298402097e-06, "loss": 0.4874, "step": 19134 }, { "epoch": 0.6817250654648449, "grad_norm": 1.491984133032852, "learning_rate": 4.858771513562355e-06, "loss": 0.4844, "step": 19135 }, { "epoch": 0.6817606925913391, "grad_norm": 1.8781619267706726, "learning_rate": 4.857781797181661e-06, "loss": 0.4786, "step": 19136 }, { "epoch": 0.6817963197178332, "grad_norm": 1.3683065209651697, "learning_rate": 4.856792149273195e-06, "loss": 0.4304, "step": 19137 }, { "epoch": 0.6818319468443272, "grad_norm": 1.192168609216742, "learning_rate": 4.855802569850141e-06, "loss": 0.3719, "step": 19138 }, { "epoch": 0.6818675739708214, "grad_norm": 1.2717818682324062, "learning_rate": 4.854813058925668e-06, "loss": 0.4147, "step": 19139 }, { "epoch": 0.6819032010973155, "grad_norm": 1.9907806171401898, "learning_rate": 4.853823616512955e-06, "loss": 0.5384, "step": 19140 }, { "epoch": 0.6819388282238096, "grad_norm": 1.2804355093068491, "learning_rate": 4.852834242625177e-06, "loss": 0.4405, "step": 19141 }, { "epoch": 0.6819744553503038, "grad_norm": 1.4849471338146611, "learning_rate": 4.851844937275514e-06, "loss": 0.4631, "step": 19142 }, { "epoch": 0.6820100824767978, "grad_norm": 1.510687827824639, "learning_rate": 4.85085570047713e-06, "loss": 0.4695, "step": 19143 }, { "epoch": 0.6820457096032919, "grad_norm": 1.2034071052720028, "learning_rate": 4.8498665322432036e-06, "loss": 0.4882, "step": 19144 }, { "epoch": 0.6820813367297861, "grad_norm": 1.2337280861107192, "learning_rate": 4.848877432586901e-06, "loss": 0.389, "step": 19145 }, { "epoch": 0.6821169638562802, "grad_norm": 1.1273475374173763, "learning_rate": 4.847888401521396e-06, "loss": 0.3994, "step": 19146 }, { "epoch": 0.6821525909827743, "grad_norm": 1.6013594409696803, "learning_rate": 4.846899439059858e-06, "loss": 0.4425, "step": 19147 }, { "epoch": 0.6821882181092684, "grad_norm": 1.6077725964677414, "learning_rate": 4.845910545215456e-06, "loss": 0.4735, "step": 19148 }, { "epoch": 0.6822238452357625, "grad_norm": 1.4038388607483805, "learning_rate": 4.844921720001361e-06, "loss": 0.5555, "step": 19149 }, { "epoch": 0.6822594723622566, "grad_norm": 1.3618430703892461, "learning_rate": 4.843932963430733e-06, "loss": 0.4578, "step": 19150 }, { "epoch": 0.6822950994887508, "grad_norm": 1.23805549954611, "learning_rate": 4.842944275516745e-06, "loss": 0.409, "step": 19151 }, { "epoch": 0.6823307266152449, "grad_norm": 1.3288158218355823, "learning_rate": 4.8419556562725545e-06, "loss": 0.4456, "step": 19152 }, { "epoch": 0.6823663537417389, "grad_norm": 1.3889166355388531, "learning_rate": 4.8409671057113335e-06, "loss": 0.5554, "step": 19153 }, { "epoch": 0.6824019808682331, "grad_norm": 1.4011680242176097, "learning_rate": 4.8399786238462345e-06, "loss": 0.4745, "step": 19154 }, { "epoch": 0.6824376079947272, "grad_norm": 1.2333940713798135, "learning_rate": 4.838990210690434e-06, "loss": 0.4259, "step": 19155 }, { "epoch": 0.6824732351212213, "grad_norm": 1.1730994245190827, "learning_rate": 4.838001866257083e-06, "loss": 0.4063, "step": 19156 }, { "epoch": 0.6825088622477155, "grad_norm": 2.486103862909447, "learning_rate": 4.8370135905593475e-06, "loss": 0.5754, "step": 19157 }, { "epoch": 0.6825444893742095, "grad_norm": 1.395434232423286, "learning_rate": 4.836025383610382e-06, "loss": 0.4187, "step": 19158 }, { "epoch": 0.6825801165007036, "grad_norm": 1.6179249798934716, "learning_rate": 4.835037245423348e-06, "loss": 0.4368, "step": 19159 }, { "epoch": 0.6826157436271978, "grad_norm": 1.5878916462219839, "learning_rate": 4.8340491760114065e-06, "loss": 0.6151, "step": 19160 }, { "epoch": 0.6826513707536919, "grad_norm": 1.528376699694311, "learning_rate": 4.833061175387708e-06, "loss": 0.4721, "step": 19161 }, { "epoch": 0.6826869978801859, "grad_norm": 1.7365127343722129, "learning_rate": 4.832073243565411e-06, "loss": 0.5264, "step": 19162 }, { "epoch": 0.6827226250066801, "grad_norm": 1.31031293641627, "learning_rate": 4.83108538055767e-06, "loss": 0.3963, "step": 19163 }, { "epoch": 0.6827582521331742, "grad_norm": 1.5883657104866524, "learning_rate": 4.830097586377645e-06, "loss": 0.5362, "step": 19164 }, { "epoch": 0.6827938792596683, "grad_norm": 1.372426291132334, "learning_rate": 4.82910986103848e-06, "loss": 0.4481, "step": 19165 }, { "epoch": 0.6828295063861625, "grad_norm": 1.5746618344902756, "learning_rate": 4.8281222045533345e-06, "loss": 0.5555, "step": 19166 }, { "epoch": 0.6828651335126565, "grad_norm": 1.517539961413678, "learning_rate": 4.827134616935354e-06, "loss": 0.4362, "step": 19167 }, { "epoch": 0.6829007606391506, "grad_norm": 1.49938936682592, "learning_rate": 4.826147098197691e-06, "loss": 0.5327, "step": 19168 }, { "epoch": 0.6829363877656448, "grad_norm": 1.5897922801812754, "learning_rate": 4.825159648353497e-06, "loss": 0.5195, "step": 19169 }, { "epoch": 0.6829720148921389, "grad_norm": 1.3289069339965316, "learning_rate": 4.8241722674159175e-06, "loss": 0.4908, "step": 19170 }, { "epoch": 0.683007642018633, "grad_norm": 1.2085495457043256, "learning_rate": 4.823184955398107e-06, "loss": 0.4591, "step": 19171 }, { "epoch": 0.6830432691451271, "grad_norm": 1.2916498773830511, "learning_rate": 4.822197712313203e-06, "loss": 0.416, "step": 19172 }, { "epoch": 0.6830788962716212, "grad_norm": 1.573554094712097, "learning_rate": 4.821210538174359e-06, "loss": 0.6189, "step": 19173 }, { "epoch": 0.6831145233981153, "grad_norm": 1.2823814058079432, "learning_rate": 4.820223432994713e-06, "loss": 0.4197, "step": 19174 }, { "epoch": 0.6831501505246095, "grad_norm": 1.4304159687636995, "learning_rate": 4.819236396787417e-06, "loss": 0.4719, "step": 19175 }, { "epoch": 0.6831857776511036, "grad_norm": 1.6744027967543762, "learning_rate": 4.818249429565605e-06, "loss": 0.5409, "step": 19176 }, { "epoch": 0.6832214047775976, "grad_norm": 3.1733110593294245, "learning_rate": 4.8172625313424225e-06, "loss": 0.5321, "step": 19177 }, { "epoch": 0.6832570319040918, "grad_norm": 1.489088858964312, "learning_rate": 4.816275702131013e-06, "loss": 0.4227, "step": 19178 }, { "epoch": 0.6832926590305859, "grad_norm": 1.5080301306929043, "learning_rate": 4.8152889419445195e-06, "loss": 0.5444, "step": 19179 }, { "epoch": 0.68332828615708, "grad_norm": 1.120792883977875, "learning_rate": 4.814302250796074e-06, "loss": 0.3861, "step": 19180 }, { "epoch": 0.6833639132835742, "grad_norm": 1.3918448242108614, "learning_rate": 4.813315628698822e-06, "loss": 0.3843, "step": 19181 }, { "epoch": 0.6833995404100682, "grad_norm": 1.316302944030456, "learning_rate": 4.812329075665894e-06, "loss": 0.3958, "step": 19182 }, { "epoch": 0.6834351675365623, "grad_norm": 1.4801492438481858, "learning_rate": 4.81134259171043e-06, "loss": 0.5525, "step": 19183 }, { "epoch": 0.6834707946630565, "grad_norm": 1.1767332335217315, "learning_rate": 4.8103561768455676e-06, "loss": 0.3989, "step": 19184 }, { "epoch": 0.6835064217895506, "grad_norm": 1.2754045434659687, "learning_rate": 4.809369831084439e-06, "loss": 0.4137, "step": 19185 }, { "epoch": 0.6835420489160446, "grad_norm": 1.6444521571036215, "learning_rate": 4.8083835544401825e-06, "loss": 0.5016, "step": 19186 }, { "epoch": 0.6835776760425388, "grad_norm": 1.4742014781649078, "learning_rate": 4.807397346925925e-06, "loss": 0.422, "step": 19187 }, { "epoch": 0.6836133031690329, "grad_norm": 1.178859050791637, "learning_rate": 4.806411208554804e-06, "loss": 0.4074, "step": 19188 }, { "epoch": 0.683648930295527, "grad_norm": 2.2537265724872992, "learning_rate": 4.805425139339944e-06, "loss": 0.5616, "step": 19189 }, { "epoch": 0.6836845574220212, "grad_norm": 1.5557286006736477, "learning_rate": 4.804439139294484e-06, "loss": 0.5051, "step": 19190 }, { "epoch": 0.6837201845485152, "grad_norm": 1.5853299594222845, "learning_rate": 4.80345320843154e-06, "loss": 0.5711, "step": 19191 }, { "epoch": 0.6837558116750093, "grad_norm": 1.6185749635960298, "learning_rate": 4.802467346764256e-06, "loss": 0.5386, "step": 19192 }, { "epoch": 0.6837914388015035, "grad_norm": 2.415564814317033, "learning_rate": 4.801481554305748e-06, "loss": 0.5457, "step": 19193 }, { "epoch": 0.6838270659279976, "grad_norm": 1.9923014167817645, "learning_rate": 4.800495831069147e-06, "loss": 0.4891, "step": 19194 }, { "epoch": 0.6838626930544917, "grad_norm": 1.4672776382779535, "learning_rate": 4.799510177067583e-06, "loss": 0.5419, "step": 19195 }, { "epoch": 0.6838983201809858, "grad_norm": 1.261693049611951, "learning_rate": 4.798524592314171e-06, "loss": 0.4231, "step": 19196 }, { "epoch": 0.6839339473074799, "grad_norm": 1.3232044419383493, "learning_rate": 4.797539076822044e-06, "loss": 0.4026, "step": 19197 }, { "epoch": 0.683969574433974, "grad_norm": 1.2748863787371696, "learning_rate": 4.796553630604317e-06, "loss": 0.3753, "step": 19198 }, { "epoch": 0.6840052015604682, "grad_norm": 1.2696577099115531, "learning_rate": 4.795568253674115e-06, "loss": 0.4217, "step": 19199 }, { "epoch": 0.6840408286869623, "grad_norm": 1.0688575861923104, "learning_rate": 4.794582946044559e-06, "loss": 0.4226, "step": 19200 }, { "epoch": 0.6840764558134563, "grad_norm": 1.3989015438951433, "learning_rate": 4.793597707728774e-06, "loss": 0.3588, "step": 19201 }, { "epoch": 0.6841120829399505, "grad_norm": 1.2099585831661197, "learning_rate": 4.792612538739871e-06, "loss": 0.4492, "step": 19202 }, { "epoch": 0.6841477100664446, "grad_norm": 1.2562892790652898, "learning_rate": 4.791627439090975e-06, "loss": 0.4474, "step": 19203 }, { "epoch": 0.6841833371929387, "grad_norm": 1.437000023482954, "learning_rate": 4.790642408795195e-06, "loss": 0.4285, "step": 19204 }, { "epoch": 0.6842189643194329, "grad_norm": 1.2015984261885326, "learning_rate": 4.7896574478656545e-06, "loss": 0.4369, "step": 19205 }, { "epoch": 0.6842545914459269, "grad_norm": 1.69480429428242, "learning_rate": 4.788672556315466e-06, "loss": 0.552, "step": 19206 }, { "epoch": 0.684290218572421, "grad_norm": 2.344769931354971, "learning_rate": 4.787687734157745e-06, "loss": 0.4973, "step": 19207 }, { "epoch": 0.6843258456989152, "grad_norm": 1.647937710356485, "learning_rate": 4.7867029814056085e-06, "loss": 0.6361, "step": 19208 }, { "epoch": 0.6843614728254093, "grad_norm": 1.556711801593532, "learning_rate": 4.785718298072163e-06, "loss": 0.5318, "step": 19209 }, { "epoch": 0.6843970999519033, "grad_norm": 1.1976375142255178, "learning_rate": 4.784733684170525e-06, "loss": 0.4081, "step": 19210 }, { "epoch": 0.6844327270783975, "grad_norm": 1.3729795022481677, "learning_rate": 4.783749139713801e-06, "loss": 0.4538, "step": 19211 }, { "epoch": 0.6844683542048916, "grad_norm": 1.3317213294693024, "learning_rate": 4.782764664715106e-06, "loss": 0.5181, "step": 19212 }, { "epoch": 0.6845039813313857, "grad_norm": 1.5776180944870968, "learning_rate": 4.781780259187543e-06, "loss": 0.481, "step": 19213 }, { "epoch": 0.6845396084578799, "grad_norm": 1.5417479281619557, "learning_rate": 4.780795923144222e-06, "loss": 0.5758, "step": 19214 }, { "epoch": 0.6845752355843739, "grad_norm": 1.4807440891138028, "learning_rate": 4.779811656598253e-06, "loss": 0.5935, "step": 19215 }, { "epoch": 0.684610862710868, "grad_norm": 1.5196127594947848, "learning_rate": 4.778827459562743e-06, "loss": 0.5671, "step": 19216 }, { "epoch": 0.6846464898373622, "grad_norm": 1.200722044685049, "learning_rate": 4.777843332050791e-06, "loss": 0.4083, "step": 19217 }, { "epoch": 0.6846821169638563, "grad_norm": 1.323166322257036, "learning_rate": 4.776859274075506e-06, "loss": 0.5529, "step": 19218 }, { "epoch": 0.6847177440903504, "grad_norm": 1.4026554759072891, "learning_rate": 4.775875285649993e-06, "loss": 0.4311, "step": 19219 }, { "epoch": 0.6847533712168445, "grad_norm": 1.4203209733895048, "learning_rate": 4.774891366787349e-06, "loss": 0.3956, "step": 19220 }, { "epoch": 0.6847889983433386, "grad_norm": 1.5123904278812157, "learning_rate": 4.77390751750068e-06, "loss": 0.4683, "step": 19221 }, { "epoch": 0.6848246254698327, "grad_norm": 1.4840923783152709, "learning_rate": 4.7729237378030835e-06, "loss": 0.4447, "step": 19222 }, { "epoch": 0.6848602525963269, "grad_norm": 1.5133045507559457, "learning_rate": 4.771940027707664e-06, "loss": 0.578, "step": 19223 }, { "epoch": 0.684895879722821, "grad_norm": 1.1154396738629162, "learning_rate": 4.770956387227516e-06, "loss": 0.4149, "step": 19224 }, { "epoch": 0.684931506849315, "grad_norm": 1.6117035343745187, "learning_rate": 4.76997281637574e-06, "loss": 0.5151, "step": 19225 }, { "epoch": 0.6849671339758092, "grad_norm": 1.3377767947458783, "learning_rate": 4.768989315165429e-06, "loss": 0.3863, "step": 19226 }, { "epoch": 0.6850027611023033, "grad_norm": 1.2395500333629998, "learning_rate": 4.768005883609684e-06, "loss": 0.4321, "step": 19227 }, { "epoch": 0.6850383882287974, "grad_norm": 1.1698462469358302, "learning_rate": 4.767022521721595e-06, "loss": 0.3867, "step": 19228 }, { "epoch": 0.6850740153552916, "grad_norm": 1.2606590664601345, "learning_rate": 4.766039229514255e-06, "loss": 0.4667, "step": 19229 }, { "epoch": 0.6851096424817856, "grad_norm": 1.3836973962576642, "learning_rate": 4.765056007000768e-06, "loss": 0.5997, "step": 19230 }, { "epoch": 0.6851452696082797, "grad_norm": 1.3591304502966037, "learning_rate": 4.764072854194216e-06, "loss": 0.4406, "step": 19231 }, { "epoch": 0.6851808967347739, "grad_norm": 1.7608158169509962, "learning_rate": 4.763089771107697e-06, "loss": 0.5377, "step": 19232 }, { "epoch": 0.685216523861268, "grad_norm": 1.7100518506720785, "learning_rate": 4.762106757754294e-06, "loss": 0.432, "step": 19233 }, { "epoch": 0.685252150987762, "grad_norm": 1.4330407200784239, "learning_rate": 4.761123814147105e-06, "loss": 0.4714, "step": 19234 }, { "epoch": 0.6852877781142562, "grad_norm": 3.994531978221016, "learning_rate": 4.76014094029921e-06, "loss": 0.4724, "step": 19235 }, { "epoch": 0.6853234052407503, "grad_norm": 1.2912545010106837, "learning_rate": 4.759158136223701e-06, "loss": 0.4343, "step": 19236 }, { "epoch": 0.6853590323672444, "grad_norm": 1.8176289214623014, "learning_rate": 4.758175401933665e-06, "loss": 0.4677, "step": 19237 }, { "epoch": 0.6853946594937386, "grad_norm": 1.4528269337986555, "learning_rate": 4.757192737442191e-06, "loss": 0.572, "step": 19238 }, { "epoch": 0.6854302866202326, "grad_norm": 1.5690548366359172, "learning_rate": 4.756210142762357e-06, "loss": 0.522, "step": 19239 }, { "epoch": 0.6854659137467267, "grad_norm": 1.1705035960017023, "learning_rate": 4.755227617907249e-06, "loss": 0.4252, "step": 19240 }, { "epoch": 0.6855015408732209, "grad_norm": 1.5597368448090387, "learning_rate": 4.754245162889954e-06, "loss": 0.495, "step": 19241 }, { "epoch": 0.685537167999715, "grad_norm": 1.2317802314298536, "learning_rate": 4.75326277772355e-06, "loss": 0.3944, "step": 19242 }, { "epoch": 0.6855727951262091, "grad_norm": 1.5834676052433767, "learning_rate": 4.7522804624211175e-06, "loss": 0.437, "step": 19243 }, { "epoch": 0.6856084222527032, "grad_norm": 1.2558359115543252, "learning_rate": 4.751298216995739e-06, "loss": 0.4291, "step": 19244 }, { "epoch": 0.6856440493791973, "grad_norm": 1.443452474227076, "learning_rate": 4.750316041460496e-06, "loss": 0.4245, "step": 19245 }, { "epoch": 0.6856796765056914, "grad_norm": 1.1541006164583663, "learning_rate": 4.749333935828463e-06, "loss": 0.4314, "step": 19246 }, { "epoch": 0.6857153036321856, "grad_norm": 1.2388033146122803, "learning_rate": 4.748351900112721e-06, "loss": 0.4289, "step": 19247 }, { "epoch": 0.6857509307586797, "grad_norm": 1.3569510404519072, "learning_rate": 4.74736993432634e-06, "loss": 0.3919, "step": 19248 }, { "epoch": 0.6857865578851737, "grad_norm": 1.4422161956745772, "learning_rate": 4.746388038482403e-06, "loss": 0.4994, "step": 19249 }, { "epoch": 0.6858221850116679, "grad_norm": 2.1046203628183897, "learning_rate": 4.745406212593979e-06, "loss": 0.3944, "step": 19250 }, { "epoch": 0.685857812138162, "grad_norm": 1.443593711462913, "learning_rate": 4.744424456674144e-06, "loss": 0.4274, "step": 19251 }, { "epoch": 0.6858934392646561, "grad_norm": 1.6111807538889658, "learning_rate": 4.74344277073597e-06, "loss": 0.534, "step": 19252 }, { "epoch": 0.6859290663911503, "grad_norm": 1.9946485234451843, "learning_rate": 4.74246115479253e-06, "loss": 0.4979, "step": 19253 }, { "epoch": 0.6859646935176443, "grad_norm": 1.5530108955635922, "learning_rate": 4.741479608856898e-06, "loss": 0.5265, "step": 19254 }, { "epoch": 0.6860003206441384, "grad_norm": 1.556685561259528, "learning_rate": 4.740498132942137e-06, "loss": 0.4145, "step": 19255 }, { "epoch": 0.6860359477706326, "grad_norm": 2.1217217538618742, "learning_rate": 4.739516727061323e-06, "loss": 0.505, "step": 19256 }, { "epoch": 0.6860715748971267, "grad_norm": 1.836795472829063, "learning_rate": 4.738535391227517e-06, "loss": 0.5476, "step": 19257 }, { "epoch": 0.6861072020236207, "grad_norm": 1.8534503586886852, "learning_rate": 4.737554125453791e-06, "loss": 0.4121, "step": 19258 }, { "epoch": 0.6861428291501149, "grad_norm": 1.3280476108597283, "learning_rate": 4.736572929753208e-06, "loss": 0.4463, "step": 19259 }, { "epoch": 0.686178456276609, "grad_norm": 1.3322625655298568, "learning_rate": 4.73559180413884e-06, "loss": 0.4495, "step": 19260 }, { "epoch": 0.6862140834031031, "grad_norm": 1.5162191762668396, "learning_rate": 4.734610748623743e-06, "loss": 0.4927, "step": 19261 }, { "epoch": 0.6862497105295973, "grad_norm": 1.4277230535677141, "learning_rate": 4.733629763220988e-06, "loss": 0.4345, "step": 19262 }, { "epoch": 0.6862853376560913, "grad_norm": 1.4839136765347258, "learning_rate": 4.7326488479436295e-06, "loss": 0.5043, "step": 19263 }, { "epoch": 0.6863209647825854, "grad_norm": 1.6690391095476993, "learning_rate": 4.7316680028047335e-06, "loss": 0.5236, "step": 19264 }, { "epoch": 0.6863565919090796, "grad_norm": 1.4784879637499169, "learning_rate": 4.730687227817365e-06, "loss": 0.3875, "step": 19265 }, { "epoch": 0.6863922190355737, "grad_norm": 1.3920706382763863, "learning_rate": 4.72970652299457e-06, "loss": 0.4262, "step": 19266 }, { "epoch": 0.6864278461620678, "grad_norm": 1.4931661548788144, "learning_rate": 4.728725888349425e-06, "loss": 0.4636, "step": 19267 }, { "epoch": 0.6864634732885619, "grad_norm": 1.2149626305643868, "learning_rate": 4.727745323894976e-06, "loss": 0.428, "step": 19268 }, { "epoch": 0.686499100415056, "grad_norm": 1.4730702343259534, "learning_rate": 4.726764829644286e-06, "loss": 0.4007, "step": 19269 }, { "epoch": 0.6865347275415501, "grad_norm": 1.4254920963331654, "learning_rate": 4.725784405610404e-06, "loss": 0.5601, "step": 19270 }, { "epoch": 0.6865703546680443, "grad_norm": 1.332565215291366, "learning_rate": 4.724804051806393e-06, "loss": 0.4263, "step": 19271 }, { "epoch": 0.6866059817945384, "grad_norm": 1.4928477854613857, "learning_rate": 4.723823768245299e-06, "loss": 0.4803, "step": 19272 }, { "epoch": 0.6866416089210324, "grad_norm": 1.5776553633317634, "learning_rate": 4.7228435549401805e-06, "loss": 0.5281, "step": 19273 }, { "epoch": 0.6866772360475266, "grad_norm": 1.3477966014481548, "learning_rate": 4.721863411904087e-06, "loss": 0.4868, "step": 19274 }, { "epoch": 0.6867128631740207, "grad_norm": 1.5218952250397104, "learning_rate": 4.720883339150073e-06, "loss": 0.6238, "step": 19275 }, { "epoch": 0.6867484903005148, "grad_norm": 1.741253503737326, "learning_rate": 4.7199033366911905e-06, "loss": 0.4108, "step": 19276 }, { "epoch": 0.686784117427009, "grad_norm": 1.4897111133822218, "learning_rate": 4.718923404540482e-06, "loss": 0.4579, "step": 19277 }, { "epoch": 0.686819744553503, "grad_norm": 1.3872602076495404, "learning_rate": 4.717943542711003e-06, "loss": 0.6036, "step": 19278 }, { "epoch": 0.6868553716799971, "grad_norm": 1.3680197191459371, "learning_rate": 4.716963751215794e-06, "loss": 0.4344, "step": 19279 }, { "epoch": 0.6868909988064913, "grad_norm": 1.3035121924614863, "learning_rate": 4.715984030067905e-06, "loss": 0.5103, "step": 19280 }, { "epoch": 0.6869266259329854, "grad_norm": 1.4528224509157392, "learning_rate": 4.715004379280382e-06, "loss": 0.4306, "step": 19281 }, { "epoch": 0.6869622530594794, "grad_norm": 1.5591319739351828, "learning_rate": 4.714024798866275e-06, "loss": 0.4447, "step": 19282 }, { "epoch": 0.6869978801859736, "grad_norm": 1.520816544411345, "learning_rate": 4.713045288838617e-06, "loss": 0.5716, "step": 19283 }, { "epoch": 0.6870335073124677, "grad_norm": 1.5483808745558598, "learning_rate": 4.71206584921046e-06, "loss": 0.4683, "step": 19284 }, { "epoch": 0.6870691344389618, "grad_norm": 1.3883956806354014, "learning_rate": 4.711086479994839e-06, "loss": 0.5998, "step": 19285 }, { "epoch": 0.687104761565456, "grad_norm": 1.8182861412800193, "learning_rate": 4.710107181204798e-06, "loss": 0.504, "step": 19286 }, { "epoch": 0.68714038869195, "grad_norm": 1.713580459297018, "learning_rate": 4.7091279528533816e-06, "loss": 0.5377, "step": 19287 }, { "epoch": 0.6871760158184441, "grad_norm": 1.8618783599495379, "learning_rate": 4.708148794953621e-06, "loss": 0.3949, "step": 19288 }, { "epoch": 0.6872116429449383, "grad_norm": 2.320635227606405, "learning_rate": 4.707169707518558e-06, "loss": 0.4546, "step": 19289 }, { "epoch": 0.6872472700714324, "grad_norm": 1.9124626613677262, "learning_rate": 4.706190690561229e-06, "loss": 0.4671, "step": 19290 }, { "epoch": 0.6872828971979265, "grad_norm": 1.7680917911648788, "learning_rate": 4.705211744094674e-06, "loss": 0.5439, "step": 19291 }, { "epoch": 0.6873185243244206, "grad_norm": 1.5056284114416558, "learning_rate": 4.704232868131924e-06, "loss": 0.5025, "step": 19292 }, { "epoch": 0.6873541514509147, "grad_norm": 1.339756635825532, "learning_rate": 4.703254062686017e-06, "loss": 0.5748, "step": 19293 }, { "epoch": 0.6873897785774088, "grad_norm": 1.7468304265234293, "learning_rate": 4.7022753277699816e-06, "loss": 0.4519, "step": 19294 }, { "epoch": 0.687425405703903, "grad_norm": 3.946115664344709, "learning_rate": 4.701296663396852e-06, "loss": 0.4235, "step": 19295 }, { "epoch": 0.6874610328303971, "grad_norm": 1.6488716608053482, "learning_rate": 4.700318069579663e-06, "loss": 0.4991, "step": 19296 }, { "epoch": 0.6874966599568911, "grad_norm": 5.084304894309338, "learning_rate": 4.699339546331441e-06, "loss": 0.4459, "step": 19297 }, { "epoch": 0.6875322870833853, "grad_norm": 1.5487618395269338, "learning_rate": 4.698361093665222e-06, "loss": 0.5661, "step": 19298 }, { "epoch": 0.6875679142098794, "grad_norm": 2.0504888138180624, "learning_rate": 4.697382711594027e-06, "loss": 0.4367, "step": 19299 }, { "epoch": 0.6876035413363735, "grad_norm": 1.5375858780455283, "learning_rate": 4.6964044001308915e-06, "loss": 0.5624, "step": 19300 }, { "epoch": 0.6876391684628677, "grad_norm": 1.8281664347471427, "learning_rate": 4.695426159288835e-06, "loss": 0.4745, "step": 19301 }, { "epoch": 0.6876747955893617, "grad_norm": 1.8582034377456713, "learning_rate": 4.694447989080891e-06, "loss": 0.5236, "step": 19302 }, { "epoch": 0.6877104227158558, "grad_norm": 3.2153680235611635, "learning_rate": 4.693469889520074e-06, "loss": 0.55, "step": 19303 }, { "epoch": 0.68774604984235, "grad_norm": 3.2977092053883656, "learning_rate": 4.692491860619418e-06, "loss": 0.4197, "step": 19304 }, { "epoch": 0.6877816769688441, "grad_norm": 1.4165900364106994, "learning_rate": 4.691513902391941e-06, "loss": 0.5173, "step": 19305 }, { "epoch": 0.6878173040953381, "grad_norm": 1.4956259991313285, "learning_rate": 4.690536014850671e-06, "loss": 0.4583, "step": 19306 }, { "epoch": 0.6878529312218323, "grad_norm": 1.7468722215536236, "learning_rate": 4.68955819800862e-06, "loss": 0.5128, "step": 19307 }, { "epoch": 0.6878885583483264, "grad_norm": 1.4239422161261042, "learning_rate": 4.688580451878818e-06, "loss": 0.4406, "step": 19308 }, { "epoch": 0.6879241854748205, "grad_norm": 1.9004506386410445, "learning_rate": 4.687602776474276e-06, "loss": 0.623, "step": 19309 }, { "epoch": 0.6879598126013147, "grad_norm": 2.997832308297578, "learning_rate": 4.686625171808016e-06, "loss": 0.4268, "step": 19310 }, { "epoch": 0.6879954397278087, "grad_norm": 1.482296695421401, "learning_rate": 4.685647637893056e-06, "loss": 0.4869, "step": 19311 }, { "epoch": 0.6880310668543028, "grad_norm": 1.7906419247703729, "learning_rate": 4.6846701747424115e-06, "loss": 0.474, "step": 19312 }, { "epoch": 0.688066693980797, "grad_norm": 1.526608257964116, "learning_rate": 4.683692782369104e-06, "loss": 0.5102, "step": 19313 }, { "epoch": 0.6881023211072911, "grad_norm": 1.7458700431171397, "learning_rate": 4.682715460786138e-06, "loss": 0.4806, "step": 19314 }, { "epoch": 0.6881379482337852, "grad_norm": 2.5522813701508005, "learning_rate": 4.681738210006536e-06, "loss": 0.4182, "step": 19315 }, { "epoch": 0.6881735753602793, "grad_norm": 2.6732756046750428, "learning_rate": 4.680761030043303e-06, "loss": 0.4596, "step": 19316 }, { "epoch": 0.6882092024867734, "grad_norm": 3.1015304721612873, "learning_rate": 4.67978392090946e-06, "loss": 0.4533, "step": 19317 }, { "epoch": 0.6882448296132675, "grad_norm": 1.382558416698264, "learning_rate": 4.678806882618003e-06, "loss": 0.5886, "step": 19318 }, { "epoch": 0.6882804567397617, "grad_norm": 2.855416832022978, "learning_rate": 4.677829915181961e-06, "loss": 0.4564, "step": 19319 }, { "epoch": 0.6883160838662558, "grad_norm": 3.645237797939564, "learning_rate": 4.676853018614329e-06, "loss": 0.4811, "step": 19320 }, { "epoch": 0.6883517109927498, "grad_norm": 2.8466696986534354, "learning_rate": 4.675876192928121e-06, "loss": 0.405, "step": 19321 }, { "epoch": 0.688387338119244, "grad_norm": 3.2255847192055533, "learning_rate": 4.674899438136346e-06, "loss": 0.4055, "step": 19322 }, { "epoch": 0.6884229652457381, "grad_norm": 2.9476060719976664, "learning_rate": 4.673922754252001e-06, "loss": 0.4352, "step": 19323 }, { "epoch": 0.6884585923722322, "grad_norm": 3.56231092991628, "learning_rate": 4.672946141288104e-06, "loss": 0.4346, "step": 19324 }, { "epoch": 0.6884942194987264, "grad_norm": 1.4419219224692748, "learning_rate": 4.671969599257648e-06, "loss": 0.5797, "step": 19325 }, { "epoch": 0.6885298466252204, "grad_norm": 2.555772464757312, "learning_rate": 4.6709931281736395e-06, "loss": 0.4268, "step": 19326 }, { "epoch": 0.6885654737517145, "grad_norm": 1.4669253261892516, "learning_rate": 4.670016728049084e-06, "loss": 0.4474, "step": 19327 }, { "epoch": 0.6886011008782087, "grad_norm": 2.1736598780404663, "learning_rate": 4.669040398896983e-06, "loss": 0.3893, "step": 19328 }, { "epoch": 0.6886367280047028, "grad_norm": 1.586169088158225, "learning_rate": 4.668064140730333e-06, "loss": 0.5779, "step": 19329 }, { "epoch": 0.6886723551311968, "grad_norm": 1.8195792573263458, "learning_rate": 4.66708795356214e-06, "loss": 0.3974, "step": 19330 }, { "epoch": 0.688707982257691, "grad_norm": 1.706885921478655, "learning_rate": 4.666111837405392e-06, "loss": 0.4546, "step": 19331 }, { "epoch": 0.6887436093841851, "grad_norm": 1.442896599023603, "learning_rate": 4.665135792273095e-06, "loss": 0.554, "step": 19332 }, { "epoch": 0.6887792365106792, "grad_norm": 1.4709821559321734, "learning_rate": 4.664159818178244e-06, "loss": 0.4053, "step": 19333 }, { "epoch": 0.6888148636371734, "grad_norm": 1.3384238945070843, "learning_rate": 4.6631839151338345e-06, "loss": 0.4195, "step": 19334 }, { "epoch": 0.6888504907636674, "grad_norm": 1.5421163789769903, "learning_rate": 4.662208083152865e-06, "loss": 0.4418, "step": 19335 }, { "epoch": 0.6888861178901615, "grad_norm": 2.21915623926898, "learning_rate": 4.661232322248321e-06, "loss": 0.5502, "step": 19336 }, { "epoch": 0.6889217450166557, "grad_norm": 1.2921428097170438, "learning_rate": 4.660256632433205e-06, "loss": 0.4853, "step": 19337 }, { "epoch": 0.6889573721431498, "grad_norm": 1.560396850199858, "learning_rate": 4.6592810137205e-06, "loss": 0.4142, "step": 19338 }, { "epoch": 0.6889929992696439, "grad_norm": 1.5831563876034456, "learning_rate": 4.658305466123205e-06, "loss": 0.5226, "step": 19339 }, { "epoch": 0.689028626396138, "grad_norm": 1.671110132202348, "learning_rate": 4.657329989654303e-06, "loss": 0.5547, "step": 19340 }, { "epoch": 0.6890642535226321, "grad_norm": 1.288655016235779, "learning_rate": 4.656354584326787e-06, "loss": 0.4314, "step": 19341 }, { "epoch": 0.6890998806491262, "grad_norm": 1.2776974626866533, "learning_rate": 4.6553792501536436e-06, "loss": 0.4368, "step": 19342 }, { "epoch": 0.6891355077756204, "grad_norm": 1.3998179846592307, "learning_rate": 4.654403987147865e-06, "loss": 0.5756, "step": 19343 }, { "epoch": 0.6891711349021145, "grad_norm": 1.3639942844522681, "learning_rate": 4.65342879532243e-06, "loss": 0.4778, "step": 19344 }, { "epoch": 0.6892067620286085, "grad_norm": 1.3026520750655386, "learning_rate": 4.652453674690328e-06, "loss": 0.5519, "step": 19345 }, { "epoch": 0.6892423891551027, "grad_norm": 1.198384342759288, "learning_rate": 4.651478625264547e-06, "loss": 0.4618, "step": 19346 }, { "epoch": 0.6892780162815968, "grad_norm": 1.397717844886898, "learning_rate": 4.650503647058062e-06, "loss": 0.5108, "step": 19347 }, { "epoch": 0.6893136434080909, "grad_norm": 1.4937582759328238, "learning_rate": 4.64952874008386e-06, "loss": 0.3952, "step": 19348 }, { "epoch": 0.6893492705345851, "grad_norm": 1.7319257354340374, "learning_rate": 4.648553904354923e-06, "loss": 0.5279, "step": 19349 }, { "epoch": 0.6893848976610791, "grad_norm": 1.4581356786252968, "learning_rate": 4.6475791398842355e-06, "loss": 0.4228, "step": 19350 }, { "epoch": 0.6894205247875732, "grad_norm": 1.391778122654355, "learning_rate": 4.6466044466847685e-06, "loss": 0.4874, "step": 19351 }, { "epoch": 0.6894561519140674, "grad_norm": 1.4765839752334935, "learning_rate": 4.645629824769509e-06, "loss": 0.422, "step": 19352 }, { "epoch": 0.6894917790405615, "grad_norm": 1.3324678238557752, "learning_rate": 4.6446552741514266e-06, "loss": 0.4372, "step": 19353 }, { "epoch": 0.6895274061670555, "grad_norm": 1.6173149112140683, "learning_rate": 4.643680794843506e-06, "loss": 0.5077, "step": 19354 }, { "epoch": 0.6895630332935497, "grad_norm": 1.2131108033576496, "learning_rate": 4.642706386858713e-06, "loss": 0.3855, "step": 19355 }, { "epoch": 0.6895986604200438, "grad_norm": 1.4763238451829905, "learning_rate": 4.641732050210032e-06, "loss": 0.4244, "step": 19356 }, { "epoch": 0.6896342875465379, "grad_norm": 1.3416689223608438, "learning_rate": 4.640757784910438e-06, "loss": 0.4884, "step": 19357 }, { "epoch": 0.6896699146730321, "grad_norm": 1.3035991195742642, "learning_rate": 4.639783590972896e-06, "loss": 0.4334, "step": 19358 }, { "epoch": 0.6897055417995261, "grad_norm": 1.5414566720122354, "learning_rate": 4.638809468410385e-06, "loss": 0.5359, "step": 19359 }, { "epoch": 0.6897411689260202, "grad_norm": 1.639065378596146, "learning_rate": 4.637835417235871e-06, "loss": 0.5526, "step": 19360 }, { "epoch": 0.6897767960525144, "grad_norm": 1.2924134704101846, "learning_rate": 4.63686143746233e-06, "loss": 0.4262, "step": 19361 }, { "epoch": 0.6898124231790085, "grad_norm": 1.236167040924671, "learning_rate": 4.635887529102723e-06, "loss": 0.4207, "step": 19362 }, { "epoch": 0.6898480503055026, "grad_norm": 1.518864539663065, "learning_rate": 4.634913692170024e-06, "loss": 0.5914, "step": 19363 }, { "epoch": 0.6898836774319967, "grad_norm": 1.3010161525858046, "learning_rate": 4.633939926677197e-06, "loss": 0.3938, "step": 19364 }, { "epoch": 0.6899193045584908, "grad_norm": 1.2406009054815612, "learning_rate": 4.632966232637216e-06, "loss": 0.3675, "step": 19365 }, { "epoch": 0.6899549316849849, "grad_norm": 1.396113316338701, "learning_rate": 4.631992610063036e-06, "loss": 0.523, "step": 19366 }, { "epoch": 0.6899905588114791, "grad_norm": 1.3262547327894998, "learning_rate": 4.631019058967627e-06, "loss": 0.5912, "step": 19367 }, { "epoch": 0.6900261859379732, "grad_norm": 1.444883035387326, "learning_rate": 4.6300455793639565e-06, "loss": 0.5513, "step": 19368 }, { "epoch": 0.6900618130644672, "grad_norm": 1.2361850939339771, "learning_rate": 4.6290721712649776e-06, "loss": 0.4673, "step": 19369 }, { "epoch": 0.6900974401909614, "grad_norm": 1.2610474263213354, "learning_rate": 4.628098834683658e-06, "loss": 0.4902, "step": 19370 }, { "epoch": 0.6901330673174555, "grad_norm": 1.2044957460734889, "learning_rate": 4.6271255696329556e-06, "loss": 0.4054, "step": 19371 }, { "epoch": 0.6901686944439496, "grad_norm": 1.5311514906816839, "learning_rate": 4.626152376125837e-06, "loss": 0.5525, "step": 19372 }, { "epoch": 0.6902043215704438, "grad_norm": 1.186580029856427, "learning_rate": 4.625179254175251e-06, "loss": 0.436, "step": 19373 }, { "epoch": 0.6902399486969378, "grad_norm": 1.7895350897738138, "learning_rate": 4.624206203794164e-06, "loss": 0.5678, "step": 19374 }, { "epoch": 0.6902755758234319, "grad_norm": 1.2491185529806654, "learning_rate": 4.623233224995524e-06, "loss": 0.4018, "step": 19375 }, { "epoch": 0.6903112029499261, "grad_norm": 1.1483908186313412, "learning_rate": 4.6222603177922965e-06, "loss": 0.3973, "step": 19376 }, { "epoch": 0.6903468300764202, "grad_norm": 1.3059676229004034, "learning_rate": 4.6212874821974275e-06, "loss": 0.4214, "step": 19377 }, { "epoch": 0.6903824572029142, "grad_norm": 1.4308206040947702, "learning_rate": 4.620314718223876e-06, "loss": 0.5951, "step": 19378 }, { "epoch": 0.6904180843294084, "grad_norm": 1.2663004451071802, "learning_rate": 4.619342025884593e-06, "loss": 0.4321, "step": 19379 }, { "epoch": 0.6904537114559025, "grad_norm": 1.1614414216363145, "learning_rate": 4.6183694051925335e-06, "loss": 0.3862, "step": 19380 }, { "epoch": 0.6904893385823966, "grad_norm": 1.209460131301544, "learning_rate": 4.61739685616065e-06, "loss": 0.374, "step": 19381 }, { "epoch": 0.6905249657088908, "grad_norm": 1.485749849303203, "learning_rate": 4.616424378801885e-06, "loss": 0.441, "step": 19382 }, { "epoch": 0.6905605928353848, "grad_norm": 1.2931392378651863, "learning_rate": 4.615451973129196e-06, "loss": 0.3935, "step": 19383 }, { "epoch": 0.6905962199618789, "grad_norm": 1.5368110877838064, "learning_rate": 4.6144796391555245e-06, "loss": 0.4179, "step": 19384 }, { "epoch": 0.6906318470883731, "grad_norm": 1.7810108200436985, "learning_rate": 4.613507376893821e-06, "loss": 0.5751, "step": 19385 }, { "epoch": 0.6906674742148672, "grad_norm": 1.6639931204195886, "learning_rate": 4.612535186357031e-06, "loss": 0.4606, "step": 19386 }, { "epoch": 0.6907031013413613, "grad_norm": 1.1720513813395086, "learning_rate": 4.611563067558105e-06, "loss": 0.4038, "step": 19387 }, { "epoch": 0.6907387284678554, "grad_norm": 1.485918177145048, "learning_rate": 4.6105910205099795e-06, "loss": 0.507, "step": 19388 }, { "epoch": 0.6907743555943495, "grad_norm": 1.4097432029124037, "learning_rate": 4.609619045225604e-06, "loss": 0.5762, "step": 19389 }, { "epoch": 0.6908099827208436, "grad_norm": 1.207059351152083, "learning_rate": 4.6086471417179155e-06, "loss": 0.386, "step": 19390 }, { "epoch": 0.6908456098473378, "grad_norm": 1.2739793716761147, "learning_rate": 4.60767530999986e-06, "loss": 0.4769, "step": 19391 }, { "epoch": 0.6908812369738319, "grad_norm": 1.3649148634412693, "learning_rate": 4.606703550084378e-06, "loss": 0.4177, "step": 19392 }, { "epoch": 0.6909168641003259, "grad_norm": 1.2631032900388848, "learning_rate": 4.6057318619844014e-06, "loss": 0.4304, "step": 19393 }, { "epoch": 0.6909524912268201, "grad_norm": 1.6976366160862848, "learning_rate": 4.604760245712883e-06, "loss": 0.5438, "step": 19394 }, { "epoch": 0.6909881183533142, "grad_norm": 1.3019630144569256, "learning_rate": 4.60378870128275e-06, "loss": 0.4311, "step": 19395 }, { "epoch": 0.6910237454798083, "grad_norm": 1.433565156841657, "learning_rate": 4.602817228706945e-06, "loss": 0.4815, "step": 19396 }, { "epoch": 0.6910593726063025, "grad_norm": 1.338460502576054, "learning_rate": 4.601845827998398e-06, "loss": 0.4324, "step": 19397 }, { "epoch": 0.6910949997327965, "grad_norm": 1.497004387365702, "learning_rate": 4.6008744991700505e-06, "loss": 0.5662, "step": 19398 }, { "epoch": 0.6911306268592906, "grad_norm": 1.279087595969602, "learning_rate": 4.599903242234829e-06, "loss": 0.406, "step": 19399 }, { "epoch": 0.6911662539857848, "grad_norm": 2.239879435652221, "learning_rate": 4.59893205720567e-06, "loss": 0.5115, "step": 19400 }, { "epoch": 0.6912018811122789, "grad_norm": 1.5847353961813762, "learning_rate": 4.597960944095505e-06, "loss": 0.4077, "step": 19401 }, { "epoch": 0.691237508238773, "grad_norm": 1.5438816480308548, "learning_rate": 4.5969899029172674e-06, "loss": 0.6006, "step": 19402 }, { "epoch": 0.6912731353652671, "grad_norm": 1.2886127599311101, "learning_rate": 4.596018933683889e-06, "loss": 0.5043, "step": 19403 }, { "epoch": 0.6913087624917612, "grad_norm": 1.2965166396170265, "learning_rate": 4.595048036408291e-06, "loss": 0.3672, "step": 19404 }, { "epoch": 0.6913443896182554, "grad_norm": 1.2021290208019997, "learning_rate": 4.59407721110341e-06, "loss": 0.3777, "step": 19405 }, { "epoch": 0.6913800167447495, "grad_norm": 1.2868806299804272, "learning_rate": 4.593106457782167e-06, "loss": 0.424, "step": 19406 }, { "epoch": 0.6914156438712435, "grad_norm": 1.1933730200607042, "learning_rate": 4.59213577645749e-06, "loss": 0.4294, "step": 19407 }, { "epoch": 0.6914512709977377, "grad_norm": 1.3687515909206605, "learning_rate": 4.5911651671423055e-06, "loss": 0.5283, "step": 19408 }, { "epoch": 0.6914868981242318, "grad_norm": 1.173137041908191, "learning_rate": 4.590194629849541e-06, "loss": 0.3899, "step": 19409 }, { "epoch": 0.6915225252507259, "grad_norm": 2.105136194614054, "learning_rate": 4.5892241645921125e-06, "loss": 0.5189, "step": 19410 }, { "epoch": 0.6915581523772201, "grad_norm": 1.097871725452058, "learning_rate": 4.5882537713829486e-06, "loss": 0.4292, "step": 19411 }, { "epoch": 0.6915937795037141, "grad_norm": 1.4350238620047526, "learning_rate": 4.587283450234965e-06, "loss": 0.4459, "step": 19412 }, { "epoch": 0.6916294066302082, "grad_norm": 1.1915789878007008, "learning_rate": 4.5863132011610866e-06, "loss": 0.4021, "step": 19413 }, { "epoch": 0.6916650337567024, "grad_norm": 1.2544349039711826, "learning_rate": 4.585343024174235e-06, "loss": 0.4341, "step": 19414 }, { "epoch": 0.6917006608831965, "grad_norm": 1.1588097046120318, "learning_rate": 4.584372919287322e-06, "loss": 0.4202, "step": 19415 }, { "epoch": 0.6917362880096906, "grad_norm": 2.6329290038793696, "learning_rate": 4.583402886513269e-06, "loss": 0.5436, "step": 19416 }, { "epoch": 0.6917719151361847, "grad_norm": 1.2678955729576409, "learning_rate": 4.582432925864993e-06, "loss": 0.3943, "step": 19417 }, { "epoch": 0.6918075422626788, "grad_norm": 1.546158699183657, "learning_rate": 4.581463037355411e-06, "loss": 0.5047, "step": 19418 }, { "epoch": 0.6918431693891729, "grad_norm": 1.6413694075507759, "learning_rate": 4.580493220997434e-06, "loss": 0.5755, "step": 19419 }, { "epoch": 0.6918787965156671, "grad_norm": 1.4230011140454488, "learning_rate": 4.579523476803981e-06, "loss": 0.5344, "step": 19420 }, { "epoch": 0.6919144236421612, "grad_norm": 1.36325889811496, "learning_rate": 4.5785538047879586e-06, "loss": 0.5481, "step": 19421 }, { "epoch": 0.6919500507686552, "grad_norm": 1.202670762771281, "learning_rate": 4.577584204962281e-06, "loss": 0.4471, "step": 19422 }, { "epoch": 0.6919856778951494, "grad_norm": 1.6077777226249923, "learning_rate": 4.576614677339859e-06, "loss": 0.5481, "step": 19423 }, { "epoch": 0.6920213050216435, "grad_norm": 1.1808803907416756, "learning_rate": 4.575645221933604e-06, "loss": 0.385, "step": 19424 }, { "epoch": 0.6920569321481376, "grad_norm": 1.5044720299137477, "learning_rate": 4.574675838756429e-06, "loss": 0.6038, "step": 19425 }, { "epoch": 0.6920925592746318, "grad_norm": 1.112011808877266, "learning_rate": 4.573706527821232e-06, "loss": 0.4057, "step": 19426 }, { "epoch": 0.6921281864011258, "grad_norm": 1.2078136835717521, "learning_rate": 4.57273728914093e-06, "loss": 0.4606, "step": 19427 }, { "epoch": 0.6921638135276199, "grad_norm": 1.1800429580205702, "learning_rate": 4.571768122728421e-06, "loss": 0.4241, "step": 19428 }, { "epoch": 0.6921994406541141, "grad_norm": 1.424819760439193, "learning_rate": 4.570799028596617e-06, "loss": 0.4586, "step": 19429 }, { "epoch": 0.6922350677806082, "grad_norm": 1.5506638584487888, "learning_rate": 4.56983000675841e-06, "loss": 0.5477, "step": 19430 }, { "epoch": 0.6922706949071022, "grad_norm": 1.2772253977220667, "learning_rate": 4.56886105722672e-06, "loss": 0.3924, "step": 19431 }, { "epoch": 0.6923063220335964, "grad_norm": 1.157891251315408, "learning_rate": 4.5678921800144385e-06, "loss": 0.365, "step": 19432 }, { "epoch": 0.6923419491600905, "grad_norm": 1.3912088144297918, "learning_rate": 4.5669233751344725e-06, "loss": 0.4447, "step": 19433 }, { "epoch": 0.6923775762865846, "grad_norm": 1.6587068220616257, "learning_rate": 4.565954642599715e-06, "loss": 0.5246, "step": 19434 }, { "epoch": 0.6924132034130788, "grad_norm": 1.3204286709164514, "learning_rate": 4.564985982423073e-06, "loss": 0.4248, "step": 19435 }, { "epoch": 0.6924488305395728, "grad_norm": 1.1067276442357532, "learning_rate": 4.56401739461744e-06, "loss": 0.3695, "step": 19436 }, { "epoch": 0.6924844576660669, "grad_norm": 1.3203565571890434, "learning_rate": 4.563048879195713e-06, "loss": 0.4607, "step": 19437 }, { "epoch": 0.6925200847925611, "grad_norm": 1.570796125729268, "learning_rate": 4.56208043617079e-06, "loss": 0.4613, "step": 19438 }, { "epoch": 0.6925557119190552, "grad_norm": 2.1326501197052687, "learning_rate": 4.5611120655555675e-06, "loss": 0.5092, "step": 19439 }, { "epoch": 0.6925913390455493, "grad_norm": 1.4443467788984043, "learning_rate": 4.560143767362943e-06, "loss": 0.5955, "step": 19440 }, { "epoch": 0.6926269661720434, "grad_norm": 1.0365900665811842, "learning_rate": 4.5591755416058035e-06, "loss": 0.4356, "step": 19441 }, { "epoch": 0.6926625932985375, "grad_norm": 3.14357394527176, "learning_rate": 4.558207388297048e-06, "loss": 0.5223, "step": 19442 }, { "epoch": 0.6926982204250316, "grad_norm": 1.2939712693222927, "learning_rate": 4.557239307449562e-06, "loss": 0.4031, "step": 19443 }, { "epoch": 0.6927338475515258, "grad_norm": 1.3885661267433145, "learning_rate": 4.556271299076237e-06, "loss": 0.4433, "step": 19444 }, { "epoch": 0.6927694746780199, "grad_norm": 1.5697343547881668, "learning_rate": 4.555303363189965e-06, "loss": 0.5548, "step": 19445 }, { "epoch": 0.6928051018045139, "grad_norm": 1.3634406677270923, "learning_rate": 4.5543354998036385e-06, "loss": 0.5497, "step": 19446 }, { "epoch": 0.6928407289310081, "grad_norm": 1.8133720627396153, "learning_rate": 4.5533677089301375e-06, "loss": 0.518, "step": 19447 }, { "epoch": 0.6928763560575022, "grad_norm": 1.290981260585103, "learning_rate": 4.552399990582351e-06, "loss": 0.4462, "step": 19448 }, { "epoch": 0.6929119831839963, "grad_norm": 1.180032949710584, "learning_rate": 4.551432344773172e-06, "loss": 0.4116, "step": 19449 }, { "epoch": 0.6929476103104905, "grad_norm": 1.3906097008398064, "learning_rate": 4.550464771515475e-06, "loss": 0.4278, "step": 19450 }, { "epoch": 0.6929832374369845, "grad_norm": 1.6888716456671677, "learning_rate": 4.549497270822153e-06, "loss": 0.5979, "step": 19451 }, { "epoch": 0.6930188645634786, "grad_norm": 1.4701777510285279, "learning_rate": 4.54852984270608e-06, "loss": 0.5341, "step": 19452 }, { "epoch": 0.6930544916899728, "grad_norm": 1.743875142792457, "learning_rate": 4.547562487180143e-06, "loss": 0.5492, "step": 19453 }, { "epoch": 0.6930901188164669, "grad_norm": 1.133944132430457, "learning_rate": 4.546595204257222e-06, "loss": 0.4022, "step": 19454 }, { "epoch": 0.693125745942961, "grad_norm": 1.3989770984147714, "learning_rate": 4.545627993950201e-06, "loss": 0.4402, "step": 19455 }, { "epoch": 0.6931613730694551, "grad_norm": 1.4011239952535568, "learning_rate": 4.544660856271952e-06, "loss": 0.5398, "step": 19456 }, { "epoch": 0.6931970001959492, "grad_norm": 1.2148584759592198, "learning_rate": 4.543693791235361e-06, "loss": 0.43, "step": 19457 }, { "epoch": 0.6932326273224433, "grad_norm": 1.3600590040810574, "learning_rate": 4.542726798853297e-06, "loss": 0.4958, "step": 19458 }, { "epoch": 0.6932682544489375, "grad_norm": 1.4607722902233262, "learning_rate": 4.541759879138641e-06, "loss": 0.4353, "step": 19459 }, { "epoch": 0.6933038815754315, "grad_norm": 1.61877363312515, "learning_rate": 4.540793032104266e-06, "loss": 0.6054, "step": 19460 }, { "epoch": 0.6933395087019256, "grad_norm": 1.2694715955298095, "learning_rate": 4.539826257763048e-06, "loss": 0.3992, "step": 19461 }, { "epoch": 0.6933751358284198, "grad_norm": 1.1884916512774466, "learning_rate": 4.538859556127864e-06, "loss": 0.4235, "step": 19462 }, { "epoch": 0.6934107629549139, "grad_norm": 1.4914521435073018, "learning_rate": 4.537892927211577e-06, "loss": 0.3952, "step": 19463 }, { "epoch": 0.693446390081408, "grad_norm": 1.281240592711213, "learning_rate": 4.536926371027068e-06, "loss": 0.3855, "step": 19464 }, { "epoch": 0.6934820172079021, "grad_norm": 1.3315660142248655, "learning_rate": 4.535959887587199e-06, "loss": 0.4314, "step": 19465 }, { "epoch": 0.6935176443343962, "grad_norm": 1.3171728446046103, "learning_rate": 4.5349934769048474e-06, "loss": 0.4662, "step": 19466 }, { "epoch": 0.6935532714608903, "grad_norm": 1.550388884625969, "learning_rate": 4.5340271389928735e-06, "loss": 0.5498, "step": 19467 }, { "epoch": 0.6935888985873845, "grad_norm": 1.7536697658037135, "learning_rate": 4.5330608738641495e-06, "loss": 0.5383, "step": 19468 }, { "epoch": 0.6936245257138786, "grad_norm": 1.4206552075090253, "learning_rate": 4.532094681531539e-06, "loss": 0.4966, "step": 19469 }, { "epoch": 0.6936601528403726, "grad_norm": 1.322085474357144, "learning_rate": 4.5311285620079144e-06, "loss": 0.4904, "step": 19470 }, { "epoch": 0.6936957799668668, "grad_norm": 1.7581505465506182, "learning_rate": 4.530162515306132e-06, "loss": 0.5557, "step": 19471 }, { "epoch": 0.6937314070933609, "grad_norm": 1.5298831294615105, "learning_rate": 4.529196541439059e-06, "loss": 0.571, "step": 19472 }, { "epoch": 0.693767034219855, "grad_norm": 1.2582831391554898, "learning_rate": 4.528230640419562e-06, "loss": 0.4694, "step": 19473 }, { "epoch": 0.6938026613463492, "grad_norm": 1.1505598786121034, "learning_rate": 4.5272648122604945e-06, "loss": 0.4164, "step": 19474 }, { "epoch": 0.6938382884728432, "grad_norm": 1.3542474847876753, "learning_rate": 4.526299056974721e-06, "loss": 0.4009, "step": 19475 }, { "epoch": 0.6938739155993373, "grad_norm": 1.558876306870577, "learning_rate": 4.525333374575103e-06, "loss": 0.5154, "step": 19476 }, { "epoch": 0.6939095427258315, "grad_norm": 1.369707354914756, "learning_rate": 4.524367765074499e-06, "loss": 0.5634, "step": 19477 }, { "epoch": 0.6939451698523256, "grad_norm": 1.5215080250334356, "learning_rate": 4.523402228485763e-06, "loss": 0.4934, "step": 19478 }, { "epoch": 0.6939807969788196, "grad_norm": 1.3942547081816183, "learning_rate": 4.522436764821758e-06, "loss": 0.4411, "step": 19479 }, { "epoch": 0.6940164241053138, "grad_norm": 1.390911882505803, "learning_rate": 4.521471374095332e-06, "loss": 0.43, "step": 19480 }, { "epoch": 0.6940520512318079, "grad_norm": 1.2852204179796995, "learning_rate": 4.520506056319347e-06, "loss": 0.4111, "step": 19481 }, { "epoch": 0.694087678358302, "grad_norm": 1.499075832453069, "learning_rate": 4.519540811506647e-06, "loss": 0.4421, "step": 19482 }, { "epoch": 0.6941233054847962, "grad_norm": 1.3466631274353416, "learning_rate": 4.518575639670097e-06, "loss": 0.4982, "step": 19483 }, { "epoch": 0.6941589326112902, "grad_norm": 1.4647808251867342, "learning_rate": 4.517610540822545e-06, "loss": 0.4471, "step": 19484 }, { "epoch": 0.6941945597377843, "grad_norm": 1.33897559034241, "learning_rate": 4.516645514976837e-06, "loss": 0.4242, "step": 19485 }, { "epoch": 0.6942301868642785, "grad_norm": 1.1578781324257905, "learning_rate": 4.515680562145832e-06, "loss": 0.3893, "step": 19486 }, { "epoch": 0.6942658139907726, "grad_norm": 1.4646481006162329, "learning_rate": 4.514715682342369e-06, "loss": 0.554, "step": 19487 }, { "epoch": 0.6943014411172667, "grad_norm": 1.5696925923322531, "learning_rate": 4.513750875579303e-06, "loss": 0.4154, "step": 19488 }, { "epoch": 0.6943370682437608, "grad_norm": 1.4062250405890342, "learning_rate": 4.5127861418694775e-06, "loss": 0.4846, "step": 19489 }, { "epoch": 0.6943726953702549, "grad_norm": 1.3721697961810637, "learning_rate": 4.5118214812257386e-06, "loss": 0.4171, "step": 19490 }, { "epoch": 0.694408322496749, "grad_norm": 1.216655925981192, "learning_rate": 4.5108568936609336e-06, "loss": 0.4618, "step": 19491 }, { "epoch": 0.6944439496232432, "grad_norm": 1.459614329240997, "learning_rate": 4.509892379187908e-06, "loss": 0.5271, "step": 19492 }, { "epoch": 0.6944795767497373, "grad_norm": 1.532741041786577, "learning_rate": 4.5089279378194995e-06, "loss": 0.5961, "step": 19493 }, { "epoch": 0.6945152038762313, "grad_norm": 1.6971092998339952, "learning_rate": 4.507963569568554e-06, "loss": 0.5272, "step": 19494 }, { "epoch": 0.6945508310027255, "grad_norm": 1.3245533644849785, "learning_rate": 4.506999274447915e-06, "loss": 0.4501, "step": 19495 }, { "epoch": 0.6945864581292196, "grad_norm": 1.1514775275739986, "learning_rate": 4.506035052470418e-06, "loss": 0.4819, "step": 19496 }, { "epoch": 0.6946220852557137, "grad_norm": 1.3615681628934149, "learning_rate": 4.505070903648903e-06, "loss": 0.4405, "step": 19497 }, { "epoch": 0.6946577123822079, "grad_norm": 1.494297944174898, "learning_rate": 4.50410682799621e-06, "loss": 0.5305, "step": 19498 }, { "epoch": 0.6946933395087019, "grad_norm": 1.3078107989999972, "learning_rate": 4.50314282552518e-06, "loss": 0.472, "step": 19499 }, { "epoch": 0.694728966635196, "grad_norm": 1.2197418652747756, "learning_rate": 4.502178896248641e-06, "loss": 0.443, "step": 19500 }, { "epoch": 0.6947645937616902, "grad_norm": 1.4104362550830318, "learning_rate": 4.501215040179438e-06, "loss": 0.5299, "step": 19501 }, { "epoch": 0.6948002208881843, "grad_norm": 1.3725029610980555, "learning_rate": 4.500251257330395e-06, "loss": 0.4675, "step": 19502 }, { "epoch": 0.6948358480146783, "grad_norm": 1.422304161064932, "learning_rate": 4.499287547714354e-06, "loss": 0.5152, "step": 19503 }, { "epoch": 0.6948714751411725, "grad_norm": 1.2978616078479448, "learning_rate": 4.498323911344142e-06, "loss": 0.3955, "step": 19504 }, { "epoch": 0.6949071022676666, "grad_norm": 1.432036282285959, "learning_rate": 4.497360348232587e-06, "loss": 0.5618, "step": 19505 }, { "epoch": 0.6949427293941607, "grad_norm": 1.566657721622752, "learning_rate": 4.496396858392533e-06, "loss": 0.4305, "step": 19506 }, { "epoch": 0.6949783565206549, "grad_norm": 1.3601465374819217, "learning_rate": 4.495433441836798e-06, "loss": 0.4085, "step": 19507 }, { "epoch": 0.695013983647149, "grad_norm": 2.5075451862414275, "learning_rate": 4.494470098578218e-06, "loss": 0.4871, "step": 19508 }, { "epoch": 0.695049610773643, "grad_norm": 1.176529372268668, "learning_rate": 4.493506828629613e-06, "loss": 0.4069, "step": 19509 }, { "epoch": 0.6950852379001372, "grad_norm": 1.3504514232483578, "learning_rate": 4.492543632003817e-06, "loss": 0.4841, "step": 19510 }, { "epoch": 0.6951208650266313, "grad_norm": 1.3611533122023054, "learning_rate": 4.4915805087136485e-06, "loss": 0.5354, "step": 19511 }, { "epoch": 0.6951564921531254, "grad_norm": 1.1854781919523716, "learning_rate": 4.490617458771936e-06, "loss": 0.437, "step": 19512 }, { "epoch": 0.6951921192796195, "grad_norm": 1.3260905993032444, "learning_rate": 4.489654482191503e-06, "loss": 0.4294, "step": 19513 }, { "epoch": 0.6952277464061136, "grad_norm": 1.6350885013093297, "learning_rate": 4.488691578985176e-06, "loss": 0.5231, "step": 19514 }, { "epoch": 0.6952633735326077, "grad_norm": 1.7782569417769185, "learning_rate": 4.487728749165768e-06, "loss": 0.531, "step": 19515 }, { "epoch": 0.6952990006591019, "grad_norm": 1.5048464715862042, "learning_rate": 4.486765992746108e-06, "loss": 0.5694, "step": 19516 }, { "epoch": 0.695334627785596, "grad_norm": 3.081622319824436, "learning_rate": 4.4858033097390095e-06, "loss": 0.39, "step": 19517 }, { "epoch": 0.69537025491209, "grad_norm": 1.52902047686153, "learning_rate": 4.4848407001572945e-06, "loss": 0.4827, "step": 19518 }, { "epoch": 0.6954058820385842, "grad_norm": 1.6920110605692071, "learning_rate": 4.48387816401378e-06, "loss": 0.4728, "step": 19519 }, { "epoch": 0.6954415091650783, "grad_norm": 1.4318972807955042, "learning_rate": 4.482915701321283e-06, "loss": 0.5637, "step": 19520 }, { "epoch": 0.6954771362915724, "grad_norm": 1.3146639289506896, "learning_rate": 4.4819533120926236e-06, "loss": 0.4151, "step": 19521 }, { "epoch": 0.6955127634180666, "grad_norm": 1.9102738592494195, "learning_rate": 4.48099099634061e-06, "loss": 0.509, "step": 19522 }, { "epoch": 0.6955483905445606, "grad_norm": 1.4369260954449263, "learning_rate": 4.480028754078061e-06, "loss": 0.4412, "step": 19523 }, { "epoch": 0.6955840176710547, "grad_norm": 1.4389670069984073, "learning_rate": 4.479066585317784e-06, "loss": 0.4003, "step": 19524 }, { "epoch": 0.6956196447975489, "grad_norm": 1.3389814006294247, "learning_rate": 4.478104490072598e-06, "loss": 0.501, "step": 19525 }, { "epoch": 0.695655271924043, "grad_norm": 1.2468578457467534, "learning_rate": 4.477142468355307e-06, "loss": 0.3894, "step": 19526 }, { "epoch": 0.695690899050537, "grad_norm": 1.4277703580444856, "learning_rate": 4.476180520178725e-06, "loss": 0.5645, "step": 19527 }, { "epoch": 0.6957265261770312, "grad_norm": 2.0038383411846277, "learning_rate": 4.475218645555659e-06, "loss": 0.5426, "step": 19528 }, { "epoch": 0.6957621533035253, "grad_norm": 1.3732642852922559, "learning_rate": 4.474256844498918e-06, "loss": 0.612, "step": 19529 }, { "epoch": 0.6957977804300194, "grad_norm": 1.731984334699192, "learning_rate": 4.473295117021314e-06, "loss": 0.5082, "step": 19530 }, { "epoch": 0.6958334075565136, "grad_norm": 1.3126615089240938, "learning_rate": 4.472333463135644e-06, "loss": 0.3729, "step": 19531 }, { "epoch": 0.6958690346830076, "grad_norm": 1.3128234850048839, "learning_rate": 4.471371882854723e-06, "loss": 0.4686, "step": 19532 }, { "epoch": 0.6959046618095017, "grad_norm": 1.594360627196822, "learning_rate": 4.470410376191344e-06, "loss": 0.4391, "step": 19533 }, { "epoch": 0.6959402889359959, "grad_norm": 1.2928817979259126, "learning_rate": 4.4694489431583165e-06, "loss": 0.4402, "step": 19534 }, { "epoch": 0.69597591606249, "grad_norm": 1.9526248906407164, "learning_rate": 4.468487583768442e-06, "loss": 0.4166, "step": 19535 }, { "epoch": 0.6960115431889841, "grad_norm": 1.3891617535296847, "learning_rate": 4.467526298034525e-06, "loss": 0.5706, "step": 19536 }, { "epoch": 0.6960471703154782, "grad_norm": 1.8102434536494378, "learning_rate": 4.466565085969359e-06, "loss": 0.4273, "step": 19537 }, { "epoch": 0.6960827974419723, "grad_norm": 1.2326894641086774, "learning_rate": 4.4656039475857505e-06, "loss": 0.4104, "step": 19538 }, { "epoch": 0.6961184245684664, "grad_norm": 1.3323780451878176, "learning_rate": 4.464642882896489e-06, "loss": 0.4916, "step": 19539 }, { "epoch": 0.6961540516949606, "grad_norm": 1.991637331916546, "learning_rate": 4.463681891914378e-06, "loss": 0.6035, "step": 19540 }, { "epoch": 0.6961896788214547, "grad_norm": 1.4630060488437144, "learning_rate": 4.462720974652215e-06, "loss": 0.5004, "step": 19541 }, { "epoch": 0.6962253059479487, "grad_norm": 1.4747736312181936, "learning_rate": 4.461760131122789e-06, "loss": 0.435, "step": 19542 }, { "epoch": 0.6962609330744429, "grad_norm": 2.0202329184874173, "learning_rate": 4.460799361338898e-06, "loss": 0.6391, "step": 19543 }, { "epoch": 0.696296560200937, "grad_norm": 1.6273881725772317, "learning_rate": 4.459838665313335e-06, "loss": 0.4232, "step": 19544 }, { "epoch": 0.6963321873274311, "grad_norm": 1.6509588530857064, "learning_rate": 4.458878043058897e-06, "loss": 0.6345, "step": 19545 }, { "epoch": 0.6963678144539253, "grad_norm": 1.501960704727765, "learning_rate": 4.4579174945883654e-06, "loss": 0.4943, "step": 19546 }, { "epoch": 0.6964034415804193, "grad_norm": 1.7536877121833627, "learning_rate": 4.4569570199145405e-06, "loss": 0.5306, "step": 19547 }, { "epoch": 0.6964390687069134, "grad_norm": 1.4978035959604727, "learning_rate": 4.455996619050204e-06, "loss": 0.478, "step": 19548 }, { "epoch": 0.6964746958334076, "grad_norm": 1.2553466974744882, "learning_rate": 4.455036292008148e-06, "loss": 0.4359, "step": 19549 }, { "epoch": 0.6965103229599017, "grad_norm": 1.9657196015301779, "learning_rate": 4.4540760388011584e-06, "loss": 0.4367, "step": 19550 }, { "epoch": 0.6965459500863957, "grad_norm": 1.5484750559921514, "learning_rate": 4.453115859442023e-06, "loss": 0.4354, "step": 19551 }, { "epoch": 0.6965815772128899, "grad_norm": 1.633687427947, "learning_rate": 4.452155753943531e-06, "loss": 0.4378, "step": 19552 }, { "epoch": 0.696617204339384, "grad_norm": 1.6608494389278694, "learning_rate": 4.451195722318459e-06, "loss": 0.4509, "step": 19553 }, { "epoch": 0.6966528314658781, "grad_norm": 1.47899923579101, "learning_rate": 4.450235764579598e-06, "loss": 0.4331, "step": 19554 }, { "epoch": 0.6966884585923723, "grad_norm": 1.4701641513377528, "learning_rate": 4.449275880739723e-06, "loss": 0.5003, "step": 19555 }, { "epoch": 0.6967240857188663, "grad_norm": 1.3524074290455332, "learning_rate": 4.448316070811622e-06, "loss": 0.4856, "step": 19556 }, { "epoch": 0.6967597128453604, "grad_norm": 1.3866721618291977, "learning_rate": 4.447356334808067e-06, "loss": 0.5217, "step": 19557 }, { "epoch": 0.6967953399718546, "grad_norm": 1.3504416266328256, "learning_rate": 4.446396672741852e-06, "loss": 0.5358, "step": 19558 }, { "epoch": 0.6968309670983487, "grad_norm": 1.7355267730303436, "learning_rate": 4.4454370846257414e-06, "loss": 0.492, "step": 19559 }, { "epoch": 0.6968665942248428, "grad_norm": 1.2937088017491793, "learning_rate": 4.4444775704725225e-06, "loss": 0.3794, "step": 19560 }, { "epoch": 0.696902221351337, "grad_norm": 1.3822488096204286, "learning_rate": 4.443518130294963e-06, "loss": 0.3966, "step": 19561 }, { "epoch": 0.696937848477831, "grad_norm": 1.3474125989467534, "learning_rate": 4.44255876410585e-06, "loss": 0.4508, "step": 19562 }, { "epoch": 0.6969734756043251, "grad_norm": 1.8041802667451041, "learning_rate": 4.441599471917946e-06, "loss": 0.5748, "step": 19563 }, { "epoch": 0.6970091027308193, "grad_norm": 1.5699895461494937, "learning_rate": 4.44064025374403e-06, "loss": 0.5697, "step": 19564 }, { "epoch": 0.6970447298573134, "grad_norm": 1.3870723494114452, "learning_rate": 4.439681109596875e-06, "loss": 0.4686, "step": 19565 }, { "epoch": 0.6970803569838074, "grad_norm": 1.536254900516391, "learning_rate": 4.4387220394892525e-06, "loss": 0.5344, "step": 19566 }, { "epoch": 0.6971159841103016, "grad_norm": 1.4897119425080652, "learning_rate": 4.437763043433937e-06, "loss": 0.5362, "step": 19567 }, { "epoch": 0.6971516112367957, "grad_norm": 1.5419393632756884, "learning_rate": 4.43680412144369e-06, "loss": 0.4023, "step": 19568 }, { "epoch": 0.6971872383632898, "grad_norm": 1.638827423155441, "learning_rate": 4.435845273531289e-06, "loss": 0.4548, "step": 19569 }, { "epoch": 0.697222865489784, "grad_norm": 1.5006304767216143, "learning_rate": 4.434886499709493e-06, "loss": 0.415, "step": 19570 }, { "epoch": 0.697258492616278, "grad_norm": 1.429849308910057, "learning_rate": 4.433927799991073e-06, "loss": 0.5358, "step": 19571 }, { "epoch": 0.6972941197427721, "grad_norm": 1.5155122317453416, "learning_rate": 4.432969174388795e-06, "loss": 0.4133, "step": 19572 }, { "epoch": 0.6973297468692663, "grad_norm": 1.5163509031923867, "learning_rate": 4.432010622915428e-06, "loss": 0.4617, "step": 19573 }, { "epoch": 0.6973653739957604, "grad_norm": 1.3642770446499968, "learning_rate": 4.431052145583726e-06, "loss": 0.5324, "step": 19574 }, { "epoch": 0.6974010011222544, "grad_norm": 1.3290438221864633, "learning_rate": 4.4300937424064595e-06, "loss": 0.5721, "step": 19575 }, { "epoch": 0.6974366282487486, "grad_norm": 1.768971666485914, "learning_rate": 4.4291354133963895e-06, "loss": 0.4111, "step": 19576 }, { "epoch": 0.6974722553752427, "grad_norm": 1.467295768611268, "learning_rate": 4.428177158566273e-06, "loss": 0.5584, "step": 19577 }, { "epoch": 0.6975078825017368, "grad_norm": 1.2815261388992254, "learning_rate": 4.427218977928875e-06, "loss": 0.3961, "step": 19578 }, { "epoch": 0.697543509628231, "grad_norm": 1.8061611903405734, "learning_rate": 4.426260871496948e-06, "loss": 0.4352, "step": 19579 }, { "epoch": 0.697579136754725, "grad_norm": 1.9983072941426248, "learning_rate": 4.425302839283253e-06, "loss": 0.4971, "step": 19580 }, { "epoch": 0.6976147638812191, "grad_norm": 2.2734591807706, "learning_rate": 4.424344881300547e-06, "loss": 0.4093, "step": 19581 }, { "epoch": 0.6976503910077133, "grad_norm": 1.2468554215638568, "learning_rate": 4.4233869975615905e-06, "loss": 0.3891, "step": 19582 }, { "epoch": 0.6976860181342074, "grad_norm": 1.5209193446621732, "learning_rate": 4.4224291880791305e-06, "loss": 0.41, "step": 19583 }, { "epoch": 0.6977216452607015, "grad_norm": 1.7724503051656075, "learning_rate": 4.421471452865928e-06, "loss": 0.4667, "step": 19584 }, { "epoch": 0.6977572723871956, "grad_norm": 1.5077692477317475, "learning_rate": 4.420513791934727e-06, "loss": 0.4368, "step": 19585 }, { "epoch": 0.6977928995136897, "grad_norm": 1.2930070197815893, "learning_rate": 4.419556205298286e-06, "loss": 0.4593, "step": 19586 }, { "epoch": 0.6978285266401838, "grad_norm": 1.4726007642007575, "learning_rate": 4.418598692969355e-06, "loss": 0.599, "step": 19587 }, { "epoch": 0.697864153766678, "grad_norm": 1.4887339762036513, "learning_rate": 4.4176412549606825e-06, "loss": 0.4015, "step": 19588 }, { "epoch": 0.6978997808931721, "grad_norm": 1.3962495388276073, "learning_rate": 4.416683891285023e-06, "loss": 0.4227, "step": 19589 }, { "epoch": 0.6979354080196661, "grad_norm": 1.590466796536675, "learning_rate": 4.415726601955116e-06, "loss": 0.3817, "step": 19590 }, { "epoch": 0.6979710351461603, "grad_norm": 1.8113994762897438, "learning_rate": 4.414769386983716e-06, "loss": 0.5354, "step": 19591 }, { "epoch": 0.6980066622726544, "grad_norm": 1.612261875162297, "learning_rate": 4.413812246383562e-06, "loss": 0.4733, "step": 19592 }, { "epoch": 0.6980422893991485, "grad_norm": 1.686051300548389, "learning_rate": 4.412855180167406e-06, "loss": 0.46, "step": 19593 }, { "epoch": 0.6980779165256427, "grad_norm": 1.3847258580191213, "learning_rate": 4.4118981883479805e-06, "loss": 0.4426, "step": 19594 }, { "epoch": 0.6981135436521367, "grad_norm": 1.3751687539435151, "learning_rate": 4.410941270938045e-06, "loss": 0.4327, "step": 19595 }, { "epoch": 0.6981491707786308, "grad_norm": 1.74204219986881, "learning_rate": 4.409984427950328e-06, "loss": 0.4739, "step": 19596 }, { "epoch": 0.698184797905125, "grad_norm": 1.4067212625590368, "learning_rate": 4.409027659397577e-06, "loss": 0.4102, "step": 19597 }, { "epoch": 0.6982204250316191, "grad_norm": 1.2767037930698355, "learning_rate": 4.408070965292534e-06, "loss": 0.5342, "step": 19598 }, { "epoch": 0.6982560521581131, "grad_norm": 1.2931796303769443, "learning_rate": 4.407114345647929e-06, "loss": 0.4032, "step": 19599 }, { "epoch": 0.6982916792846073, "grad_norm": 1.695335866757161, "learning_rate": 4.406157800476512e-06, "loss": 0.5107, "step": 19600 }, { "epoch": 0.6983273064111014, "grad_norm": 1.3257527208199176, "learning_rate": 4.405201329791008e-06, "loss": 0.4462, "step": 19601 }, { "epoch": 0.6983629335375955, "grad_norm": 1.3260182413568216, "learning_rate": 4.40424493360416e-06, "loss": 0.3983, "step": 19602 }, { "epoch": 0.6983985606640897, "grad_norm": 1.1890077278559557, "learning_rate": 4.403288611928702e-06, "loss": 0.3946, "step": 19603 }, { "epoch": 0.6984341877905837, "grad_norm": 1.325823703139743, "learning_rate": 4.402332364777371e-06, "loss": 0.5472, "step": 19604 }, { "epoch": 0.6984698149170778, "grad_norm": 2.00506260973862, "learning_rate": 4.401376192162894e-06, "loss": 0.4169, "step": 19605 }, { "epoch": 0.698505442043572, "grad_norm": 1.4108735806951456, "learning_rate": 4.400420094098009e-06, "loss": 0.4577, "step": 19606 }, { "epoch": 0.6985410691700661, "grad_norm": 1.4714229828216563, "learning_rate": 4.399464070595442e-06, "loss": 0.4103, "step": 19607 }, { "epoch": 0.6985766962965602, "grad_norm": 1.7055783304844572, "learning_rate": 4.398508121667925e-06, "loss": 0.4116, "step": 19608 }, { "epoch": 0.6986123234230543, "grad_norm": 1.6181839712831925, "learning_rate": 4.397552247328187e-06, "loss": 0.5393, "step": 19609 }, { "epoch": 0.6986479505495484, "grad_norm": 1.56665400747381, "learning_rate": 4.396596447588956e-06, "loss": 0.564, "step": 19610 }, { "epoch": 0.6986835776760425, "grad_norm": 1.4008364555545685, "learning_rate": 4.3956407224629645e-06, "loss": 0.5227, "step": 19611 }, { "epoch": 0.6987192048025367, "grad_norm": 1.5472584438648285, "learning_rate": 4.394685071962931e-06, "loss": 0.5191, "step": 19612 }, { "epoch": 0.6987548319290308, "grad_norm": 1.5752603298396899, "learning_rate": 4.393729496101587e-06, "loss": 0.5228, "step": 19613 }, { "epoch": 0.6987904590555248, "grad_norm": 2.9920232832616147, "learning_rate": 4.3927739948916496e-06, "loss": 0.5271, "step": 19614 }, { "epoch": 0.698826086182019, "grad_norm": 1.2026995750784595, "learning_rate": 4.391818568345848e-06, "loss": 0.41, "step": 19615 }, { "epoch": 0.6988617133085131, "grad_norm": 1.3865163856587004, "learning_rate": 4.390863216476899e-06, "loss": 0.4354, "step": 19616 }, { "epoch": 0.6988973404350072, "grad_norm": 1.6455880030243484, "learning_rate": 4.389907939297528e-06, "loss": 0.5048, "step": 19617 }, { "epoch": 0.6989329675615014, "grad_norm": 1.4264421541792265, "learning_rate": 4.388952736820453e-06, "loss": 0.4639, "step": 19618 }, { "epoch": 0.6989685946879954, "grad_norm": 1.149080017209253, "learning_rate": 4.387997609058399e-06, "loss": 0.3827, "step": 19619 }, { "epoch": 0.6990042218144895, "grad_norm": 1.7561462442591393, "learning_rate": 4.387042556024074e-06, "loss": 0.4317, "step": 19620 }, { "epoch": 0.6990398489409837, "grad_norm": 1.6587654676197687, "learning_rate": 4.386087577730201e-06, "loss": 0.468, "step": 19621 }, { "epoch": 0.6990754760674778, "grad_norm": 1.6016771483014371, "learning_rate": 4.3851326741895e-06, "loss": 0.522, "step": 19622 }, { "epoch": 0.6991111031939719, "grad_norm": 1.2509744224622075, "learning_rate": 4.384177845414679e-06, "loss": 0.408, "step": 19623 }, { "epoch": 0.699146730320466, "grad_norm": 1.178355333495242, "learning_rate": 4.383223091418454e-06, "loss": 0.4167, "step": 19624 }, { "epoch": 0.6991823574469601, "grad_norm": 1.3874723510007252, "learning_rate": 4.382268412213539e-06, "loss": 0.4248, "step": 19625 }, { "epoch": 0.6992179845734542, "grad_norm": 1.2286527877645834, "learning_rate": 4.381313807812652e-06, "loss": 0.4188, "step": 19626 }, { "epoch": 0.6992536116999484, "grad_norm": 1.3184327970268814, "learning_rate": 4.380359278228494e-06, "loss": 0.4031, "step": 19627 }, { "epoch": 0.6992892388264424, "grad_norm": 1.541220398102953, "learning_rate": 4.3794048234737855e-06, "loss": 0.5397, "step": 19628 }, { "epoch": 0.6993248659529365, "grad_norm": 1.3211012119584804, "learning_rate": 4.378450443561224e-06, "loss": 0.4219, "step": 19629 }, { "epoch": 0.6993604930794307, "grad_norm": 1.3286762903569433, "learning_rate": 4.37749613850353e-06, "loss": 0.4747, "step": 19630 }, { "epoch": 0.6993961202059248, "grad_norm": 1.0756881473398725, "learning_rate": 4.3765419083134e-06, "loss": 0.3414, "step": 19631 }, { "epoch": 0.6994317473324189, "grad_norm": 2.619808970440853, "learning_rate": 4.3755877530035415e-06, "loss": 0.4895, "step": 19632 }, { "epoch": 0.699467374458913, "grad_norm": 1.5165337437990674, "learning_rate": 4.37463367258667e-06, "loss": 0.4278, "step": 19633 }, { "epoch": 0.6995030015854071, "grad_norm": 1.2948495022426125, "learning_rate": 4.373679667075481e-06, "loss": 0.4074, "step": 19634 }, { "epoch": 0.6995386287119012, "grad_norm": 1.4782525559333073, "learning_rate": 4.372725736482682e-06, "loss": 0.5641, "step": 19635 }, { "epoch": 0.6995742558383954, "grad_norm": 1.2111630987668454, "learning_rate": 4.37177188082097e-06, "loss": 0.4031, "step": 19636 }, { "epoch": 0.6996098829648895, "grad_norm": 1.6734275625435804, "learning_rate": 4.370818100103054e-06, "loss": 0.568, "step": 19637 }, { "epoch": 0.6996455100913835, "grad_norm": 1.2511143042876103, "learning_rate": 4.369864394341624e-06, "loss": 0.4282, "step": 19638 }, { "epoch": 0.6996811372178777, "grad_norm": 1.249998269390981, "learning_rate": 4.3689107635493855e-06, "loss": 0.4561, "step": 19639 }, { "epoch": 0.6997167643443718, "grad_norm": 1.5365176809887038, "learning_rate": 4.367957207739036e-06, "loss": 0.6419, "step": 19640 }, { "epoch": 0.6997523914708659, "grad_norm": 1.4126069138626667, "learning_rate": 4.367003726923277e-06, "loss": 0.4088, "step": 19641 }, { "epoch": 0.6997880185973601, "grad_norm": 1.2450175772452523, "learning_rate": 4.366050321114797e-06, "loss": 0.4222, "step": 19642 }, { "epoch": 0.6998236457238541, "grad_norm": 1.3525885210173054, "learning_rate": 4.365096990326298e-06, "loss": 0.5347, "step": 19643 }, { "epoch": 0.6998592728503482, "grad_norm": 1.2652235894710264, "learning_rate": 4.364143734570467e-06, "loss": 0.4107, "step": 19644 }, { "epoch": 0.6998948999768424, "grad_norm": 1.12746043511387, "learning_rate": 4.363190553860003e-06, "loss": 0.4329, "step": 19645 }, { "epoch": 0.6999305271033365, "grad_norm": 1.4426348944164264, "learning_rate": 4.362237448207596e-06, "loss": 0.5909, "step": 19646 }, { "epoch": 0.6999661542298306, "grad_norm": 2.0734793874492867, "learning_rate": 4.361284417625938e-06, "loss": 0.5692, "step": 19647 }, { "epoch": 0.7000017813563247, "grad_norm": 1.288473474700667, "learning_rate": 4.360331462127721e-06, "loss": 0.4116, "step": 19648 }, { "epoch": 0.7000374084828188, "grad_norm": 1.4527870455238112, "learning_rate": 4.3593785817256305e-06, "loss": 0.5312, "step": 19649 }, { "epoch": 0.7000730356093129, "grad_norm": 1.5397390550416017, "learning_rate": 4.35842577643236e-06, "loss": 0.4288, "step": 19650 }, { "epoch": 0.7001086627358071, "grad_norm": 1.5653842088048353, "learning_rate": 4.357473046260591e-06, "loss": 0.5564, "step": 19651 }, { "epoch": 0.7001442898623012, "grad_norm": 1.5437749268181116, "learning_rate": 4.356520391223015e-06, "loss": 0.5899, "step": 19652 }, { "epoch": 0.7001799169887952, "grad_norm": 1.5875768551843734, "learning_rate": 4.355567811332311e-06, "loss": 0.4716, "step": 19653 }, { "epoch": 0.7002155441152894, "grad_norm": 1.3719317736424064, "learning_rate": 4.354615306601166e-06, "loss": 0.4037, "step": 19654 }, { "epoch": 0.7002511712417835, "grad_norm": 1.5184726003290774, "learning_rate": 4.3536628770422655e-06, "loss": 0.4908, "step": 19655 }, { "epoch": 0.7002867983682776, "grad_norm": 1.2065997401785755, "learning_rate": 4.35271052266829e-06, "loss": 0.3959, "step": 19656 }, { "epoch": 0.7003224254947717, "grad_norm": 1.4584490794803333, "learning_rate": 4.3517582434919235e-06, "loss": 0.4428, "step": 19657 }, { "epoch": 0.7003580526212658, "grad_norm": 1.516095344885826, "learning_rate": 4.350806039525841e-06, "loss": 0.5709, "step": 19658 }, { "epoch": 0.7003936797477599, "grad_norm": 1.5149485391212174, "learning_rate": 4.349853910782728e-06, "loss": 0.5338, "step": 19659 }, { "epoch": 0.7004293068742541, "grad_norm": 1.4315294643150729, "learning_rate": 4.348901857275257e-06, "loss": 0.419, "step": 19660 }, { "epoch": 0.7004649340007482, "grad_norm": 1.416593034330685, "learning_rate": 4.347949879016107e-06, "loss": 0.5241, "step": 19661 }, { "epoch": 0.7005005611272422, "grad_norm": 1.1659113636392449, "learning_rate": 4.346997976017955e-06, "loss": 0.4104, "step": 19662 }, { "epoch": 0.7005361882537364, "grad_norm": 1.3555484653378502, "learning_rate": 4.346046148293479e-06, "loss": 0.397, "step": 19663 }, { "epoch": 0.7005718153802305, "grad_norm": 1.2071918493380431, "learning_rate": 4.345094395855347e-06, "loss": 0.4405, "step": 19664 }, { "epoch": 0.7006074425067246, "grad_norm": 1.3604100891751407, "learning_rate": 4.34414271871624e-06, "loss": 0.4295, "step": 19665 }, { "epoch": 0.7006430696332188, "grad_norm": 1.30970130303646, "learning_rate": 4.343191116888822e-06, "loss": 0.4943, "step": 19666 }, { "epoch": 0.7006786967597128, "grad_norm": 1.1896822114677275, "learning_rate": 4.342239590385768e-06, "loss": 0.4, "step": 19667 }, { "epoch": 0.7007143238862069, "grad_norm": 1.2275370187627834, "learning_rate": 4.341288139219752e-06, "loss": 0.3725, "step": 19668 }, { "epoch": 0.7007499510127011, "grad_norm": 1.5341327861242877, "learning_rate": 4.340336763403437e-06, "loss": 0.4453, "step": 19669 }, { "epoch": 0.7007855781391952, "grad_norm": 1.417516368557334, "learning_rate": 4.339385462949492e-06, "loss": 0.3526, "step": 19670 }, { "epoch": 0.7008212052656893, "grad_norm": 1.2610258768647649, "learning_rate": 4.338434237870587e-06, "loss": 0.4093, "step": 19671 }, { "epoch": 0.7008568323921834, "grad_norm": 1.3474303243485055, "learning_rate": 4.337483088179391e-06, "loss": 0.4134, "step": 19672 }, { "epoch": 0.7008924595186775, "grad_norm": 1.7208616159912777, "learning_rate": 4.336532013888563e-06, "loss": 0.5565, "step": 19673 }, { "epoch": 0.7009280866451716, "grad_norm": 1.410610690932047, "learning_rate": 4.335581015010772e-06, "loss": 0.4795, "step": 19674 }, { "epoch": 0.7009637137716658, "grad_norm": 1.2500273685138514, "learning_rate": 4.334630091558675e-06, "loss": 0.4512, "step": 19675 }, { "epoch": 0.7009993408981599, "grad_norm": 1.212685791751139, "learning_rate": 4.3336792435449385e-06, "loss": 0.3898, "step": 19676 }, { "epoch": 0.7010349680246539, "grad_norm": 1.2855661580094833, "learning_rate": 4.332728470982224e-06, "loss": 0.4094, "step": 19677 }, { "epoch": 0.7010705951511481, "grad_norm": 1.3122792841553896, "learning_rate": 4.3317777738831914e-06, "loss": 0.3677, "step": 19678 }, { "epoch": 0.7011062222776422, "grad_norm": 1.9308968315929782, "learning_rate": 4.330827152260502e-06, "loss": 0.4512, "step": 19679 }, { "epoch": 0.7011418494041363, "grad_norm": 1.2692773903400167, "learning_rate": 4.329876606126808e-06, "loss": 0.4724, "step": 19680 }, { "epoch": 0.7011774765306305, "grad_norm": 1.163211251181026, "learning_rate": 4.3289261354947755e-06, "loss": 0.3978, "step": 19681 }, { "epoch": 0.7012131036571245, "grad_norm": 1.3003489932500818, "learning_rate": 4.3279757403770495e-06, "loss": 0.4318, "step": 19682 }, { "epoch": 0.7012487307836186, "grad_norm": 1.344698132993909, "learning_rate": 4.327025420786293e-06, "loss": 0.4486, "step": 19683 }, { "epoch": 0.7012843579101128, "grad_norm": 1.4729165142251015, "learning_rate": 4.326075176735157e-06, "loss": 0.5067, "step": 19684 }, { "epoch": 0.7013199850366069, "grad_norm": 1.1014522556941517, "learning_rate": 4.3251250082363e-06, "loss": 0.4307, "step": 19685 }, { "epoch": 0.7013556121631009, "grad_norm": 1.4285566975761967, "learning_rate": 4.324174915302366e-06, "loss": 0.6287, "step": 19686 }, { "epoch": 0.7013912392895951, "grad_norm": 1.100032737798762, "learning_rate": 4.323224897946013e-06, "loss": 0.44, "step": 19687 }, { "epoch": 0.7014268664160892, "grad_norm": 1.549695820398836, "learning_rate": 4.322274956179887e-06, "loss": 0.5673, "step": 19688 }, { "epoch": 0.7014624935425833, "grad_norm": 1.4907605438532026, "learning_rate": 4.321325090016641e-06, "loss": 0.5356, "step": 19689 }, { "epoch": 0.7014981206690775, "grad_norm": 1.4992887872608842, "learning_rate": 4.320375299468916e-06, "loss": 0.5368, "step": 19690 }, { "epoch": 0.7015337477955715, "grad_norm": 1.331316962303856, "learning_rate": 4.3194255845493645e-06, "loss": 0.4392, "step": 19691 }, { "epoch": 0.7015693749220656, "grad_norm": 1.0674243726701147, "learning_rate": 4.318475945270632e-06, "loss": 0.3932, "step": 19692 }, { "epoch": 0.7016050020485598, "grad_norm": 1.6786372472571873, "learning_rate": 4.3175263816453625e-06, "loss": 0.5306, "step": 19693 }, { "epoch": 0.7016406291750539, "grad_norm": 1.497521357977074, "learning_rate": 4.316576893686206e-06, "loss": 0.5281, "step": 19694 }, { "epoch": 0.701676256301548, "grad_norm": 1.329201449072648, "learning_rate": 4.315627481405796e-06, "loss": 0.408, "step": 19695 }, { "epoch": 0.7017118834280421, "grad_norm": 1.4952201960908629, "learning_rate": 4.314678144816783e-06, "loss": 0.5657, "step": 19696 }, { "epoch": 0.7017475105545362, "grad_norm": 1.6417401418835107, "learning_rate": 4.313728883931801e-06, "loss": 0.4789, "step": 19697 }, { "epoch": 0.7017831376810303, "grad_norm": 1.3163657729949103, "learning_rate": 4.312779698763493e-06, "loss": 0.5395, "step": 19698 }, { "epoch": 0.7018187648075245, "grad_norm": 1.5253187655809834, "learning_rate": 4.311830589324499e-06, "loss": 0.555, "step": 19699 }, { "epoch": 0.7018543919340186, "grad_norm": 1.558281325204672, "learning_rate": 4.31088155562746e-06, "loss": 0.601, "step": 19700 }, { "epoch": 0.7018900190605126, "grad_norm": 1.3565102868193422, "learning_rate": 4.309932597685006e-06, "loss": 0.6042, "step": 19701 }, { "epoch": 0.7019256461870068, "grad_norm": 1.2802513880877848, "learning_rate": 4.308983715509776e-06, "loss": 0.42, "step": 19702 }, { "epoch": 0.7019612733135009, "grad_norm": 1.224042600419192, "learning_rate": 4.308034909114409e-06, "loss": 0.4471, "step": 19703 }, { "epoch": 0.701996900439995, "grad_norm": 1.3035976003075092, "learning_rate": 4.307086178511533e-06, "loss": 0.4643, "step": 19704 }, { "epoch": 0.7020325275664892, "grad_norm": 1.280398599248605, "learning_rate": 4.306137523713787e-06, "loss": 0.4316, "step": 19705 }, { "epoch": 0.7020681546929832, "grad_norm": 1.6492349486097093, "learning_rate": 4.305188944733796e-06, "loss": 0.4919, "step": 19706 }, { "epoch": 0.7021037818194773, "grad_norm": 1.26465000680409, "learning_rate": 4.304240441584194e-06, "loss": 0.3909, "step": 19707 }, { "epoch": 0.7021394089459715, "grad_norm": 1.2874603458301912, "learning_rate": 4.3032920142776125e-06, "loss": 0.4569, "step": 19708 }, { "epoch": 0.7021750360724656, "grad_norm": 1.4979244619344827, "learning_rate": 4.302343662826683e-06, "loss": 0.5461, "step": 19709 }, { "epoch": 0.7022106631989596, "grad_norm": 1.943800207446012, "learning_rate": 4.301395387244026e-06, "loss": 0.4961, "step": 19710 }, { "epoch": 0.7022462903254538, "grad_norm": 1.5777045537662362, "learning_rate": 4.300447187542276e-06, "loss": 0.5039, "step": 19711 }, { "epoch": 0.7022819174519479, "grad_norm": 1.2539037497621466, "learning_rate": 4.299499063734053e-06, "loss": 0.4076, "step": 19712 }, { "epoch": 0.702317544578442, "grad_norm": 1.162672831444936, "learning_rate": 4.298551015831983e-06, "loss": 0.3851, "step": 19713 }, { "epoch": 0.7023531717049362, "grad_norm": 1.5149142477154351, "learning_rate": 4.297603043848692e-06, "loss": 0.4904, "step": 19714 }, { "epoch": 0.7023887988314302, "grad_norm": 1.9028921194485262, "learning_rate": 4.296655147796802e-06, "loss": 0.5006, "step": 19715 }, { "epoch": 0.7024244259579243, "grad_norm": 1.4482063378113665, "learning_rate": 4.295707327688938e-06, "loss": 0.4721, "step": 19716 }, { "epoch": 0.7024600530844185, "grad_norm": 1.491814170305108, "learning_rate": 4.2947595835377155e-06, "loss": 0.4466, "step": 19717 }, { "epoch": 0.7024956802109126, "grad_norm": 1.8384253610236, "learning_rate": 4.293811915355761e-06, "loss": 0.5041, "step": 19718 }, { "epoch": 0.7025313073374067, "grad_norm": 1.4583304110634405, "learning_rate": 4.292864323155684e-06, "loss": 0.4337, "step": 19719 }, { "epoch": 0.7025669344639008, "grad_norm": 1.277454666144159, "learning_rate": 4.291916806950112e-06, "loss": 0.4218, "step": 19720 }, { "epoch": 0.7026025615903949, "grad_norm": 1.2783194931662032, "learning_rate": 4.29096936675165e-06, "loss": 0.4388, "step": 19721 }, { "epoch": 0.702638188716889, "grad_norm": 1.4547856768924348, "learning_rate": 4.290022002572929e-06, "loss": 0.5244, "step": 19722 }, { "epoch": 0.7026738158433832, "grad_norm": 1.6241977460230903, "learning_rate": 4.289074714426552e-06, "loss": 0.5208, "step": 19723 }, { "epoch": 0.7027094429698773, "grad_norm": 1.2127427693413384, "learning_rate": 4.2881275023251375e-06, "loss": 0.5183, "step": 19724 }, { "epoch": 0.7027450700963713, "grad_norm": 1.4082608106568826, "learning_rate": 4.287180366281301e-06, "loss": 0.4533, "step": 19725 }, { "epoch": 0.7027806972228655, "grad_norm": 1.3747679242020001, "learning_rate": 4.286233306307647e-06, "loss": 0.4818, "step": 19726 }, { "epoch": 0.7028163243493596, "grad_norm": 1.1529029806959339, "learning_rate": 4.285286322416794e-06, "loss": 0.4225, "step": 19727 }, { "epoch": 0.7028519514758537, "grad_norm": 1.2275141773710159, "learning_rate": 4.284339414621343e-06, "loss": 0.393, "step": 19728 }, { "epoch": 0.7028875786023479, "grad_norm": 1.37215675236067, "learning_rate": 4.28339258293391e-06, "loss": 0.434, "step": 19729 }, { "epoch": 0.7029232057288419, "grad_norm": 1.7123083088860962, "learning_rate": 4.2824458273671e-06, "loss": 0.6126, "step": 19730 }, { "epoch": 0.702958832855336, "grad_norm": 1.4858902921265351, "learning_rate": 4.281499147933522e-06, "loss": 0.5216, "step": 19731 }, { "epoch": 0.7029944599818302, "grad_norm": 1.3036996052596483, "learning_rate": 4.280552544645779e-06, "loss": 0.5789, "step": 19732 }, { "epoch": 0.7030300871083243, "grad_norm": 1.581862490009571, "learning_rate": 4.279606017516478e-06, "loss": 0.4814, "step": 19733 }, { "epoch": 0.7030657142348183, "grad_norm": 1.5243542943737503, "learning_rate": 4.278659566558218e-06, "loss": 0.467, "step": 19734 }, { "epoch": 0.7031013413613125, "grad_norm": 1.443834592892088, "learning_rate": 4.277713191783604e-06, "loss": 0.4733, "step": 19735 }, { "epoch": 0.7031369684878066, "grad_norm": 1.4063865467800518, "learning_rate": 4.27676689320524e-06, "loss": 0.5326, "step": 19736 }, { "epoch": 0.7031725956143007, "grad_norm": 1.2049063058548417, "learning_rate": 4.275820670835724e-06, "loss": 0.4174, "step": 19737 }, { "epoch": 0.7032082227407949, "grad_norm": 1.2125798699784869, "learning_rate": 4.274874524687661e-06, "loss": 0.4287, "step": 19738 }, { "epoch": 0.7032438498672889, "grad_norm": 1.1889707502522555, "learning_rate": 4.2739284547736426e-06, "loss": 0.443, "step": 19739 }, { "epoch": 0.703279476993783, "grad_norm": 1.6483960545621086, "learning_rate": 4.272982461106271e-06, "loss": 0.5907, "step": 19740 }, { "epoch": 0.7033151041202772, "grad_norm": 1.1899588446936118, "learning_rate": 4.272036543698138e-06, "loss": 0.4224, "step": 19741 }, { "epoch": 0.7033507312467713, "grad_norm": 1.2134628222595047, "learning_rate": 4.271090702561846e-06, "loss": 0.4334, "step": 19742 }, { "epoch": 0.7033863583732654, "grad_norm": 1.3457951641003225, "learning_rate": 4.270144937709981e-06, "loss": 0.5383, "step": 19743 }, { "epoch": 0.7034219854997595, "grad_norm": 1.5914560167735479, "learning_rate": 4.269199249155142e-06, "loss": 0.5973, "step": 19744 }, { "epoch": 0.7034576126262536, "grad_norm": 1.2158901418649557, "learning_rate": 4.26825363690992e-06, "loss": 0.4192, "step": 19745 }, { "epoch": 0.7034932397527477, "grad_norm": 1.3574404770490898, "learning_rate": 4.267308100986911e-06, "loss": 0.5436, "step": 19746 }, { "epoch": 0.7035288668792419, "grad_norm": 1.3242411302477588, "learning_rate": 4.266362641398698e-06, "loss": 0.4351, "step": 19747 }, { "epoch": 0.703564494005736, "grad_norm": 1.2530221957254568, "learning_rate": 4.265417258157875e-06, "loss": 0.4242, "step": 19748 }, { "epoch": 0.70360012113223, "grad_norm": 1.8433338818170364, "learning_rate": 4.2644719512770314e-06, "loss": 0.5103, "step": 19749 }, { "epoch": 0.7036357482587242, "grad_norm": 1.5331536304551208, "learning_rate": 4.26352672076875e-06, "loss": 0.5669, "step": 19750 }, { "epoch": 0.7036713753852183, "grad_norm": 5.609917799240021, "learning_rate": 4.2625815666456194e-06, "loss": 0.417, "step": 19751 }, { "epoch": 0.7037070025117124, "grad_norm": 1.6634010323376636, "learning_rate": 4.261636488920225e-06, "loss": 0.5926, "step": 19752 }, { "epoch": 0.7037426296382066, "grad_norm": 1.3113037544990982, "learning_rate": 4.260691487605157e-06, "loss": 0.4106, "step": 19753 }, { "epoch": 0.7037782567647006, "grad_norm": 1.4368834571241393, "learning_rate": 4.259746562712988e-06, "loss": 0.594, "step": 19754 }, { "epoch": 0.7038138838911947, "grad_norm": 1.6187544333515735, "learning_rate": 4.258801714256309e-06, "loss": 0.5599, "step": 19755 }, { "epoch": 0.7038495110176889, "grad_norm": 1.1345179870746949, "learning_rate": 4.2578569422476955e-06, "loss": 0.3667, "step": 19756 }, { "epoch": 0.703885138144183, "grad_norm": 1.1611501465414547, "learning_rate": 4.2569122466997335e-06, "loss": 0.4002, "step": 19757 }, { "epoch": 0.703920765270677, "grad_norm": 1.131656770543113, "learning_rate": 4.2559676276249926e-06, "loss": 0.3427, "step": 19758 }, { "epoch": 0.7039563923971712, "grad_norm": 1.1788391066053971, "learning_rate": 4.2550230850360604e-06, "loss": 0.4302, "step": 19759 }, { "epoch": 0.7039920195236653, "grad_norm": 1.7311275938871042, "learning_rate": 4.2540786189455156e-06, "loss": 0.6267, "step": 19760 }, { "epoch": 0.7040276466501594, "grad_norm": 4.196842709591192, "learning_rate": 4.253134229365926e-06, "loss": 0.4933, "step": 19761 }, { "epoch": 0.7040632737766536, "grad_norm": 1.4607792374802036, "learning_rate": 4.252189916309874e-06, "loss": 0.6157, "step": 19762 }, { "epoch": 0.7040989009031476, "grad_norm": 1.5115131075122545, "learning_rate": 4.251245679789928e-06, "loss": 0.5094, "step": 19763 }, { "epoch": 0.7041345280296417, "grad_norm": 1.4505061986764514, "learning_rate": 4.250301519818668e-06, "loss": 0.3943, "step": 19764 }, { "epoch": 0.7041701551561359, "grad_norm": 1.197310021215595, "learning_rate": 4.2493574364086575e-06, "loss": 0.3874, "step": 19765 }, { "epoch": 0.70420578228263, "grad_norm": 1.1945223115994419, "learning_rate": 4.248413429572473e-06, "loss": 0.4288, "step": 19766 }, { "epoch": 0.704241409409124, "grad_norm": 1.5665752135795563, "learning_rate": 4.247469499322683e-06, "loss": 0.4916, "step": 19767 }, { "epoch": 0.7042770365356182, "grad_norm": 1.2637481460112812, "learning_rate": 4.246525645671862e-06, "loss": 0.4991, "step": 19768 }, { "epoch": 0.7043126636621123, "grad_norm": 2.117414967665907, "learning_rate": 4.245581868632569e-06, "loss": 0.5018, "step": 19769 }, { "epoch": 0.7043482907886064, "grad_norm": 1.2299602687915816, "learning_rate": 4.244638168217379e-06, "loss": 0.4319, "step": 19770 }, { "epoch": 0.7043839179151006, "grad_norm": 1.1795101977840743, "learning_rate": 4.243694544438851e-06, "loss": 0.4162, "step": 19771 }, { "epoch": 0.7044195450415947, "grad_norm": 1.2645585925619918, "learning_rate": 4.2427509973095536e-06, "loss": 0.404, "step": 19772 }, { "epoch": 0.7044551721680887, "grad_norm": 1.1774560365277078, "learning_rate": 4.241807526842051e-06, "loss": 0.361, "step": 19773 }, { "epoch": 0.7044907992945829, "grad_norm": 1.5252759166696663, "learning_rate": 4.2408641330489066e-06, "loss": 0.5057, "step": 19774 }, { "epoch": 0.704526426421077, "grad_norm": 1.6508390669246138, "learning_rate": 4.239920815942684e-06, "loss": 0.4487, "step": 19775 }, { "epoch": 0.7045620535475711, "grad_norm": 1.5857252935029056, "learning_rate": 4.238977575535939e-06, "loss": 0.5405, "step": 19776 }, { "epoch": 0.7045976806740653, "grad_norm": 1.2656718265003877, "learning_rate": 4.238034411841238e-06, "loss": 0.3695, "step": 19777 }, { "epoch": 0.7046333078005593, "grad_norm": 1.7026063704626624, "learning_rate": 4.237091324871132e-06, "loss": 0.4598, "step": 19778 }, { "epoch": 0.7046689349270534, "grad_norm": 1.7497779588589948, "learning_rate": 4.2361483146381865e-06, "loss": 0.5101, "step": 19779 }, { "epoch": 0.7047045620535476, "grad_norm": 1.2153563742433355, "learning_rate": 4.2352053811549525e-06, "loss": 0.4205, "step": 19780 }, { "epoch": 0.7047401891800417, "grad_norm": 1.6061130043608676, "learning_rate": 4.234262524433986e-06, "loss": 0.5595, "step": 19781 }, { "epoch": 0.7047758163065359, "grad_norm": 1.484861213792928, "learning_rate": 4.233319744487846e-06, "loss": 0.5434, "step": 19782 }, { "epoch": 0.7048114434330299, "grad_norm": 1.3370959501936437, "learning_rate": 4.2323770413290845e-06, "loss": 0.425, "step": 19783 }, { "epoch": 0.704847070559524, "grad_norm": 1.8534076457494266, "learning_rate": 4.2314344149702554e-06, "loss": 0.5554, "step": 19784 }, { "epoch": 0.7048826976860182, "grad_norm": 1.2930721170957724, "learning_rate": 4.230491865423908e-06, "loss": 0.4618, "step": 19785 }, { "epoch": 0.7049183248125123, "grad_norm": 1.4619198322324582, "learning_rate": 4.2295493927025955e-06, "loss": 0.4762, "step": 19786 }, { "epoch": 0.7049539519390063, "grad_norm": 1.5163334807808315, "learning_rate": 4.228606996818864e-06, "loss": 0.515, "step": 19787 }, { "epoch": 0.7049895790655005, "grad_norm": 1.1961847755944306, "learning_rate": 4.227664677785264e-06, "loss": 0.3751, "step": 19788 }, { "epoch": 0.7050252061919946, "grad_norm": 1.176620987769919, "learning_rate": 4.2267224356143436e-06, "loss": 0.4292, "step": 19789 }, { "epoch": 0.7050608333184887, "grad_norm": 1.1118334772036242, "learning_rate": 4.225780270318652e-06, "loss": 0.3892, "step": 19790 }, { "epoch": 0.7050964604449829, "grad_norm": 1.789938225159123, "learning_rate": 4.224838181910729e-06, "loss": 0.5295, "step": 19791 }, { "epoch": 0.7051320875714769, "grad_norm": 1.3408838996749708, "learning_rate": 4.223896170403126e-06, "loss": 0.5114, "step": 19792 }, { "epoch": 0.705167714697971, "grad_norm": 1.1350877672315507, "learning_rate": 4.222954235808379e-06, "loss": 0.4349, "step": 19793 }, { "epoch": 0.7052033418244652, "grad_norm": 1.2602269775087636, "learning_rate": 4.222012378139033e-06, "loss": 0.444, "step": 19794 }, { "epoch": 0.7052389689509593, "grad_norm": 1.2352924335664839, "learning_rate": 4.221070597407636e-06, "loss": 0.4276, "step": 19795 }, { "epoch": 0.7052745960774534, "grad_norm": 1.1588525611383211, "learning_rate": 4.220128893626715e-06, "loss": 0.4204, "step": 19796 }, { "epoch": 0.7053102232039475, "grad_norm": 1.1328609276918091, "learning_rate": 4.219187266808826e-06, "loss": 0.4207, "step": 19797 }, { "epoch": 0.7053458503304416, "grad_norm": 1.127540876157809, "learning_rate": 4.218245716966495e-06, "loss": 0.4318, "step": 19798 }, { "epoch": 0.7053814774569357, "grad_norm": 1.5173172444511676, "learning_rate": 4.217304244112267e-06, "loss": 0.5453, "step": 19799 }, { "epoch": 0.7054171045834299, "grad_norm": 1.3549530895207411, "learning_rate": 4.21636284825867e-06, "loss": 0.3735, "step": 19800 }, { "epoch": 0.705452731709924, "grad_norm": 1.2061264980119286, "learning_rate": 4.21542152941825e-06, "loss": 0.4439, "step": 19801 }, { "epoch": 0.705488358836418, "grad_norm": 1.3591250988633092, "learning_rate": 4.214480287603531e-06, "loss": 0.5083, "step": 19802 }, { "epoch": 0.7055239859629122, "grad_norm": 1.2608703406715829, "learning_rate": 4.21353912282705e-06, "loss": 0.3932, "step": 19803 }, { "epoch": 0.7055596130894063, "grad_norm": 1.485444630423135, "learning_rate": 4.212598035101341e-06, "loss": 0.5165, "step": 19804 }, { "epoch": 0.7055952402159004, "grad_norm": 1.5936742262823735, "learning_rate": 4.211657024438934e-06, "loss": 0.5707, "step": 19805 }, { "epoch": 0.7056308673423946, "grad_norm": 1.1235840554370244, "learning_rate": 4.210716090852364e-06, "loss": 0.4475, "step": 19806 }, { "epoch": 0.7056664944688886, "grad_norm": 1.7440174359408236, "learning_rate": 4.209775234354151e-06, "loss": 0.4817, "step": 19807 }, { "epoch": 0.7057021215953827, "grad_norm": 1.4111385220796606, "learning_rate": 4.208834454956831e-06, "loss": 0.4918, "step": 19808 }, { "epoch": 0.7057377487218769, "grad_norm": 1.7900739690123646, "learning_rate": 4.2078937526729255e-06, "loss": 0.541, "step": 19809 }, { "epoch": 0.705773375848371, "grad_norm": 1.162390828484291, "learning_rate": 4.2069531275149634e-06, "loss": 0.4465, "step": 19810 }, { "epoch": 0.705809002974865, "grad_norm": 1.19907930254944, "learning_rate": 4.206012579495469e-06, "loss": 0.3853, "step": 19811 }, { "epoch": 0.7058446301013592, "grad_norm": 1.249965403294438, "learning_rate": 4.205072108626971e-06, "loss": 0.4378, "step": 19812 }, { "epoch": 0.7058802572278533, "grad_norm": 1.1054645455869714, "learning_rate": 4.204131714921984e-06, "loss": 0.4107, "step": 19813 }, { "epoch": 0.7059158843543474, "grad_norm": 1.1899736499200861, "learning_rate": 4.203191398393037e-06, "loss": 0.3851, "step": 19814 }, { "epoch": 0.7059515114808416, "grad_norm": 1.0542727273425014, "learning_rate": 4.2022511590526464e-06, "loss": 0.3904, "step": 19815 }, { "epoch": 0.7059871386073356, "grad_norm": 1.287524122457776, "learning_rate": 4.201310996913337e-06, "loss": 0.4568, "step": 19816 }, { "epoch": 0.7060227657338297, "grad_norm": 1.2728559736153309, "learning_rate": 4.20037091198762e-06, "loss": 0.4476, "step": 19817 }, { "epoch": 0.7060583928603239, "grad_norm": 1.222894860603678, "learning_rate": 4.19943090428802e-06, "loss": 0.3611, "step": 19818 }, { "epoch": 0.706094019986818, "grad_norm": 1.1707651292780943, "learning_rate": 4.1984909738270505e-06, "loss": 0.4124, "step": 19819 }, { "epoch": 0.706129647113312, "grad_norm": 1.8782342474138543, "learning_rate": 4.1975511206172295e-06, "loss": 0.4414, "step": 19820 }, { "epoch": 0.7061652742398062, "grad_norm": 1.844261200091598, "learning_rate": 4.196611344671073e-06, "loss": 0.5499, "step": 19821 }, { "epoch": 0.7062009013663003, "grad_norm": 1.7909499967035938, "learning_rate": 4.195671646001091e-06, "loss": 0.4692, "step": 19822 }, { "epoch": 0.7062365284927944, "grad_norm": 1.2953424534408926, "learning_rate": 4.194732024619801e-06, "loss": 0.4572, "step": 19823 }, { "epoch": 0.7062721556192886, "grad_norm": 1.1372847365905276, "learning_rate": 4.193792480539708e-06, "loss": 0.4166, "step": 19824 }, { "epoch": 0.7063077827457827, "grad_norm": 1.2794374158687989, "learning_rate": 4.192853013773327e-06, "loss": 0.3947, "step": 19825 }, { "epoch": 0.7063434098722767, "grad_norm": 1.3763391426336968, "learning_rate": 4.191913624333166e-06, "loss": 0.4695, "step": 19826 }, { "epoch": 0.7063790369987709, "grad_norm": 1.593039843453583, "learning_rate": 4.190974312231739e-06, "loss": 0.5224, "step": 19827 }, { "epoch": 0.706414664125265, "grad_norm": 1.2676452577585122, "learning_rate": 4.190035077481546e-06, "loss": 0.458, "step": 19828 }, { "epoch": 0.7064502912517591, "grad_norm": 1.3310385679872299, "learning_rate": 4.189095920095096e-06, "loss": 0.47, "step": 19829 }, { "epoch": 0.7064859183782533, "grad_norm": 1.6640541519008285, "learning_rate": 4.188156840084899e-06, "loss": 0.4553, "step": 19830 }, { "epoch": 0.7065215455047473, "grad_norm": 1.5510826172879213, "learning_rate": 4.187217837463453e-06, "loss": 0.5109, "step": 19831 }, { "epoch": 0.7065571726312414, "grad_norm": 1.3511417034612176, "learning_rate": 4.186278912243268e-06, "loss": 0.5278, "step": 19832 }, { "epoch": 0.7065927997577356, "grad_norm": 1.4337039353490448, "learning_rate": 4.1853400644368395e-06, "loss": 0.4797, "step": 19833 }, { "epoch": 0.7066284268842297, "grad_norm": 1.7651988170849697, "learning_rate": 4.1844012940566715e-06, "loss": 0.5681, "step": 19834 }, { "epoch": 0.7066640540107237, "grad_norm": 1.1079692919997144, "learning_rate": 4.183462601115264e-06, "loss": 0.3855, "step": 19835 }, { "epoch": 0.7066996811372179, "grad_norm": 1.183003343894596, "learning_rate": 4.182523985625123e-06, "loss": 0.4791, "step": 19836 }, { "epoch": 0.706735308263712, "grad_norm": 1.1536228461781637, "learning_rate": 4.181585447598737e-06, "loss": 0.4233, "step": 19837 }, { "epoch": 0.7067709353902061, "grad_norm": 1.7137087161251958, "learning_rate": 4.180646987048612e-06, "loss": 0.574, "step": 19838 }, { "epoch": 0.7068065625167003, "grad_norm": 1.2991885276491444, "learning_rate": 4.1797086039872355e-06, "loss": 0.4076, "step": 19839 }, { "epoch": 0.7068421896431943, "grad_norm": 1.0905449369577647, "learning_rate": 4.178770298427107e-06, "loss": 0.4313, "step": 19840 }, { "epoch": 0.7068778167696884, "grad_norm": 1.3049546600560247, "learning_rate": 4.177832070380721e-06, "loss": 0.448, "step": 19841 }, { "epoch": 0.7069134438961826, "grad_norm": 1.1011943417363628, "learning_rate": 4.1768939198605705e-06, "loss": 0.4261, "step": 19842 }, { "epoch": 0.7069490710226767, "grad_norm": 1.8292521012557297, "learning_rate": 4.17595584687915e-06, "loss": 0.483, "step": 19843 }, { "epoch": 0.7069846981491708, "grad_norm": 1.165059174163667, "learning_rate": 4.1750178514489456e-06, "loss": 0.4348, "step": 19844 }, { "epoch": 0.7070203252756649, "grad_norm": 1.1960776699834548, "learning_rate": 4.174079933582454e-06, "loss": 0.4082, "step": 19845 }, { "epoch": 0.707055952402159, "grad_norm": 1.2716265035063656, "learning_rate": 4.173142093292156e-06, "loss": 0.4057, "step": 19846 }, { "epoch": 0.7070915795286531, "grad_norm": 1.711923420463026, "learning_rate": 4.1722043305905445e-06, "loss": 0.5059, "step": 19847 }, { "epoch": 0.7071272066551473, "grad_norm": 1.6692322050376691, "learning_rate": 4.171266645490105e-06, "loss": 0.5592, "step": 19848 }, { "epoch": 0.7071628337816414, "grad_norm": 1.3252564633372752, "learning_rate": 4.170329038003328e-06, "loss": 0.4319, "step": 19849 }, { "epoch": 0.7071984609081354, "grad_norm": 1.090521796326907, "learning_rate": 4.169391508142691e-06, "loss": 0.4116, "step": 19850 }, { "epoch": 0.7072340880346296, "grad_norm": 1.222275570509867, "learning_rate": 4.168454055920681e-06, "loss": 0.4796, "step": 19851 }, { "epoch": 0.7072697151611237, "grad_norm": 1.8589079129764738, "learning_rate": 4.167516681349785e-06, "loss": 0.4268, "step": 19852 }, { "epoch": 0.7073053422876178, "grad_norm": 1.1446337511361773, "learning_rate": 4.166579384442477e-06, "loss": 0.4397, "step": 19853 }, { "epoch": 0.707340969414112, "grad_norm": 1.2922321690902199, "learning_rate": 4.165642165211246e-06, "loss": 0.5345, "step": 19854 }, { "epoch": 0.707376596540606, "grad_norm": 1.1162614225608152, "learning_rate": 4.164705023668564e-06, "loss": 0.3692, "step": 19855 }, { "epoch": 0.7074122236671001, "grad_norm": 1.5719257771720239, "learning_rate": 4.1637679598269136e-06, "loss": 0.5935, "step": 19856 }, { "epoch": 0.7074478507935943, "grad_norm": 1.6813341166747728, "learning_rate": 4.162830973698771e-06, "loss": 0.5527, "step": 19857 }, { "epoch": 0.7074834779200884, "grad_norm": 1.4551183692881653, "learning_rate": 4.161894065296618e-06, "loss": 0.4145, "step": 19858 }, { "epoch": 0.7075191050465824, "grad_norm": 1.2502519063814734, "learning_rate": 4.160957234632923e-06, "loss": 0.4453, "step": 19859 }, { "epoch": 0.7075547321730766, "grad_norm": 1.4633894519742883, "learning_rate": 4.160020481720166e-06, "loss": 0.538, "step": 19860 }, { "epoch": 0.7075903592995707, "grad_norm": 1.5185177037272342, "learning_rate": 4.159083806570816e-06, "loss": 0.5087, "step": 19861 }, { "epoch": 0.7076259864260648, "grad_norm": 1.4543831450814637, "learning_rate": 4.158147209197347e-06, "loss": 0.5963, "step": 19862 }, { "epoch": 0.707661613552559, "grad_norm": 1.2101297429652216, "learning_rate": 4.157210689612231e-06, "loss": 0.4223, "step": 19863 }, { "epoch": 0.707697240679053, "grad_norm": 1.7175349591567273, "learning_rate": 4.15627424782794e-06, "loss": 0.5015, "step": 19864 }, { "epoch": 0.7077328678055471, "grad_norm": 1.5173472449070298, "learning_rate": 4.155337883856943e-06, "loss": 0.3986, "step": 19865 }, { "epoch": 0.7077684949320413, "grad_norm": 1.4801278842047325, "learning_rate": 4.154401597711706e-06, "loss": 0.5098, "step": 19866 }, { "epoch": 0.7078041220585354, "grad_norm": 1.0821425891657146, "learning_rate": 4.1534653894046995e-06, "loss": 0.386, "step": 19867 }, { "epoch": 0.7078397491850295, "grad_norm": 1.805243766078897, "learning_rate": 4.152529258948384e-06, "loss": 0.502, "step": 19868 }, { "epoch": 0.7078753763115236, "grad_norm": 1.3823771293472424, "learning_rate": 4.151593206355233e-06, "loss": 0.6041, "step": 19869 }, { "epoch": 0.7079110034380177, "grad_norm": 1.5148492830844327, "learning_rate": 4.1506572316377035e-06, "loss": 0.5399, "step": 19870 }, { "epoch": 0.7079466305645118, "grad_norm": 1.1249989556098117, "learning_rate": 4.14972133480826e-06, "loss": 0.4401, "step": 19871 }, { "epoch": 0.707982257691006, "grad_norm": 1.746174559742241, "learning_rate": 4.148785515879367e-06, "loss": 0.4983, "step": 19872 }, { "epoch": 0.7080178848175, "grad_norm": 1.2399587576908813, "learning_rate": 4.147849774863488e-06, "loss": 0.4403, "step": 19873 }, { "epoch": 0.7080535119439941, "grad_norm": 1.258063193519878, "learning_rate": 4.146914111773076e-06, "loss": 0.4147, "step": 19874 }, { "epoch": 0.7080891390704883, "grad_norm": 1.1862130362245358, "learning_rate": 4.145978526620593e-06, "loss": 0.414, "step": 19875 }, { "epoch": 0.7081247661969824, "grad_norm": 1.5023565297300345, "learning_rate": 4.1450430194185025e-06, "loss": 0.5418, "step": 19876 }, { "epoch": 0.7081603933234765, "grad_norm": 1.4219257441543722, "learning_rate": 4.144107590179251e-06, "loss": 0.5042, "step": 19877 }, { "epoch": 0.7081960204499707, "grad_norm": 1.114720327027747, "learning_rate": 4.143172238915302e-06, "loss": 0.3883, "step": 19878 }, { "epoch": 0.7082316475764647, "grad_norm": 1.3217010561948868, "learning_rate": 4.142236965639107e-06, "loss": 0.4673, "step": 19879 }, { "epoch": 0.7082672747029588, "grad_norm": 1.7816606487950595, "learning_rate": 4.1413017703631255e-06, "loss": 0.4701, "step": 19880 }, { "epoch": 0.708302901829453, "grad_norm": 1.1589917980342375, "learning_rate": 4.140366653099802e-06, "loss": 0.4049, "step": 19881 }, { "epoch": 0.7083385289559471, "grad_norm": 1.0614697947419607, "learning_rate": 4.139431613861595e-06, "loss": 0.4199, "step": 19882 }, { "epoch": 0.7083741560824411, "grad_norm": 1.178879393421, "learning_rate": 4.13849665266095e-06, "loss": 0.3743, "step": 19883 }, { "epoch": 0.7084097832089353, "grad_norm": 1.149939739683469, "learning_rate": 4.137561769510322e-06, "loss": 0.3917, "step": 19884 }, { "epoch": 0.7084454103354294, "grad_norm": 1.1860195553333621, "learning_rate": 4.1366269644221505e-06, "loss": 0.3747, "step": 19885 }, { "epoch": 0.7084810374619235, "grad_norm": 1.183409494149629, "learning_rate": 4.135692237408891e-06, "loss": 0.4244, "step": 19886 }, { "epoch": 0.7085166645884177, "grad_norm": 1.2093291220823879, "learning_rate": 4.134757588482993e-06, "loss": 0.4234, "step": 19887 }, { "epoch": 0.7085522917149117, "grad_norm": 1.4644881063674582, "learning_rate": 4.133823017656894e-06, "loss": 0.4924, "step": 19888 }, { "epoch": 0.7085879188414058, "grad_norm": 1.3349110688249024, "learning_rate": 4.132888524943044e-06, "loss": 0.5261, "step": 19889 }, { "epoch": 0.7086235459679, "grad_norm": 1.164226167698236, "learning_rate": 4.131954110353882e-06, "loss": 0.4223, "step": 19890 }, { "epoch": 0.7086591730943941, "grad_norm": 1.1957753572196648, "learning_rate": 4.131019773901855e-06, "loss": 0.4345, "step": 19891 }, { "epoch": 0.7086948002208882, "grad_norm": 1.5435367172433996, "learning_rate": 4.130085515599399e-06, "loss": 0.4675, "step": 19892 }, { "epoch": 0.7087304273473823, "grad_norm": 1.2321401981310458, "learning_rate": 4.1291513354589576e-06, "loss": 0.4411, "step": 19893 }, { "epoch": 0.7087660544738764, "grad_norm": 1.2078171963297366, "learning_rate": 4.128217233492968e-06, "loss": 0.394, "step": 19894 }, { "epoch": 0.7088016816003705, "grad_norm": 1.5391719986346288, "learning_rate": 4.127283209713875e-06, "loss": 0.4315, "step": 19895 }, { "epoch": 0.7088373087268647, "grad_norm": 1.535770054197973, "learning_rate": 4.126349264134106e-06, "loss": 0.5786, "step": 19896 }, { "epoch": 0.7088729358533588, "grad_norm": 1.738037502485621, "learning_rate": 4.125415396766107e-06, "loss": 0.5632, "step": 19897 }, { "epoch": 0.7089085629798528, "grad_norm": 1.2065295431329535, "learning_rate": 4.124481607622304e-06, "loss": 0.4074, "step": 19898 }, { "epoch": 0.708944190106347, "grad_norm": 1.3266422682352204, "learning_rate": 4.123547896715135e-06, "loss": 0.4638, "step": 19899 }, { "epoch": 0.7089798172328411, "grad_norm": 1.1928146058180653, "learning_rate": 4.1226142640570335e-06, "loss": 0.4419, "step": 19900 }, { "epoch": 0.7090154443593352, "grad_norm": 1.3166255920183438, "learning_rate": 4.121680709660431e-06, "loss": 0.4087, "step": 19901 }, { "epoch": 0.7090510714858294, "grad_norm": 3.39988449019103, "learning_rate": 4.120747233537762e-06, "loss": 0.5253, "step": 19902 }, { "epoch": 0.7090866986123234, "grad_norm": 1.2196882611044102, "learning_rate": 4.119813835701449e-06, "loss": 0.4136, "step": 19903 }, { "epoch": 0.7091223257388175, "grad_norm": 1.1854815317439291, "learning_rate": 4.1188805161639285e-06, "loss": 0.4151, "step": 19904 }, { "epoch": 0.7091579528653117, "grad_norm": 1.4127836135578808, "learning_rate": 4.117947274937622e-06, "loss": 0.4082, "step": 19905 }, { "epoch": 0.7091935799918058, "grad_norm": 1.1783386791797914, "learning_rate": 4.117014112034961e-06, "loss": 0.4543, "step": 19906 }, { "epoch": 0.7092292071182998, "grad_norm": 1.895637898511812, "learning_rate": 4.116081027468367e-06, "loss": 0.5329, "step": 19907 }, { "epoch": 0.709264834244794, "grad_norm": 1.148794662453971, "learning_rate": 4.115148021250266e-06, "loss": 0.4288, "step": 19908 }, { "epoch": 0.7093004613712881, "grad_norm": 1.1308085374044587, "learning_rate": 4.114215093393082e-06, "loss": 0.3858, "step": 19909 }, { "epoch": 0.7093360884977822, "grad_norm": 2.7662781074661598, "learning_rate": 4.1132822439092375e-06, "loss": 0.4165, "step": 19910 }, { "epoch": 0.7093717156242764, "grad_norm": 1.7816413595747, "learning_rate": 4.112349472811159e-06, "loss": 0.5395, "step": 19911 }, { "epoch": 0.7094073427507704, "grad_norm": 1.3567049492736925, "learning_rate": 4.111416780111259e-06, "loss": 0.533, "step": 19912 }, { "epoch": 0.7094429698772645, "grad_norm": 1.195431503572976, "learning_rate": 4.110484165821963e-06, "loss": 0.4394, "step": 19913 }, { "epoch": 0.7094785970037587, "grad_norm": 1.2579201358796772, "learning_rate": 4.109551629955685e-06, "loss": 0.4248, "step": 19914 }, { "epoch": 0.7095142241302528, "grad_norm": 1.735781536317537, "learning_rate": 4.108619172524842e-06, "loss": 0.4697, "step": 19915 }, { "epoch": 0.7095498512567469, "grad_norm": 1.257037558425907, "learning_rate": 4.107686793541854e-06, "loss": 0.4257, "step": 19916 }, { "epoch": 0.709585478383241, "grad_norm": 1.1149742859448708, "learning_rate": 4.106754493019138e-06, "loss": 0.4136, "step": 19917 }, { "epoch": 0.7096211055097351, "grad_norm": 1.1136277423354977, "learning_rate": 4.105822270969102e-06, "loss": 0.386, "step": 19918 }, { "epoch": 0.7096567326362292, "grad_norm": 1.066954540548313, "learning_rate": 4.104890127404165e-06, "loss": 0.3791, "step": 19919 }, { "epoch": 0.7096923597627234, "grad_norm": 1.4314723794489586, "learning_rate": 4.103958062336733e-06, "loss": 0.4222, "step": 19920 }, { "epoch": 0.7097279868892175, "grad_norm": 1.4366247568738406, "learning_rate": 4.103026075779221e-06, "loss": 0.4325, "step": 19921 }, { "epoch": 0.7097636140157115, "grad_norm": 1.2352406332676726, "learning_rate": 4.102094167744043e-06, "loss": 0.3974, "step": 19922 }, { "epoch": 0.7097992411422057, "grad_norm": 1.5865347482593595, "learning_rate": 4.101162338243595e-06, "loss": 0.5446, "step": 19923 }, { "epoch": 0.7098348682686998, "grad_norm": 1.5958088707557692, "learning_rate": 4.1002305872903005e-06, "loss": 0.5448, "step": 19924 }, { "epoch": 0.7098704953951939, "grad_norm": 1.5475169461279958, "learning_rate": 4.099298914896556e-06, "loss": 0.5695, "step": 19925 }, { "epoch": 0.709906122521688, "grad_norm": 1.3634078551170903, "learning_rate": 4.098367321074775e-06, "loss": 0.5292, "step": 19926 }, { "epoch": 0.7099417496481821, "grad_norm": 2.278525546872258, "learning_rate": 4.097435805837354e-06, "loss": 0.5049, "step": 19927 }, { "epoch": 0.7099773767746762, "grad_norm": 1.2430611983916688, "learning_rate": 4.0965043691967045e-06, "loss": 0.4438, "step": 19928 }, { "epoch": 0.7100130039011704, "grad_norm": 1.1476685738096482, "learning_rate": 4.095573011165223e-06, "loss": 0.3807, "step": 19929 }, { "epoch": 0.7100486310276645, "grad_norm": 1.4036482728274386, "learning_rate": 4.0946417317553124e-06, "loss": 0.5837, "step": 19930 }, { "epoch": 0.7100842581541585, "grad_norm": 1.3339769431521407, "learning_rate": 4.093710530979375e-06, "loss": 0.4771, "step": 19931 }, { "epoch": 0.7101198852806527, "grad_norm": 1.3262617717541374, "learning_rate": 4.092779408849811e-06, "loss": 0.3738, "step": 19932 }, { "epoch": 0.7101555124071468, "grad_norm": 1.2968250867824294, "learning_rate": 4.0918483653790205e-06, "loss": 0.4506, "step": 19933 }, { "epoch": 0.7101911395336409, "grad_norm": 1.3376081541801137, "learning_rate": 4.0909174005793965e-06, "loss": 0.5291, "step": 19934 }, { "epoch": 0.7102267666601351, "grad_norm": 1.1391491491824244, "learning_rate": 4.0899865144633395e-06, "loss": 0.4266, "step": 19935 }, { "epoch": 0.7102623937866291, "grad_norm": 1.1670229750325616, "learning_rate": 4.08905570704324e-06, "loss": 0.4058, "step": 19936 }, { "epoch": 0.7102980209131232, "grad_norm": 1.392677004164262, "learning_rate": 4.088124978331496e-06, "loss": 0.45, "step": 19937 }, { "epoch": 0.7103336480396174, "grad_norm": 1.7088709099948036, "learning_rate": 4.087194328340499e-06, "loss": 0.4186, "step": 19938 }, { "epoch": 0.7103692751661115, "grad_norm": 1.596421205361862, "learning_rate": 4.086263757082646e-06, "loss": 0.4464, "step": 19939 }, { "epoch": 0.7104049022926056, "grad_norm": 1.7116914955430638, "learning_rate": 4.085333264570322e-06, "loss": 0.4748, "step": 19940 }, { "epoch": 0.7104405294190997, "grad_norm": 1.1740333324579288, "learning_rate": 4.084402850815922e-06, "loss": 0.4542, "step": 19941 }, { "epoch": 0.7104761565455938, "grad_norm": 1.9324108758598657, "learning_rate": 4.083472515831829e-06, "loss": 0.5673, "step": 19942 }, { "epoch": 0.7105117836720879, "grad_norm": 1.4926885703330874, "learning_rate": 4.08254225963044e-06, "loss": 0.4708, "step": 19943 }, { "epoch": 0.7105474107985821, "grad_norm": 1.274645637814348, "learning_rate": 4.081612082224132e-06, "loss": 0.4746, "step": 19944 }, { "epoch": 0.7105830379250762, "grad_norm": 1.1304798858364338, "learning_rate": 4.080681983625297e-06, "loss": 0.4674, "step": 19945 }, { "epoch": 0.7106186650515702, "grad_norm": 1.2150272008040723, "learning_rate": 4.079751963846317e-06, "loss": 0.4575, "step": 19946 }, { "epoch": 0.7106542921780644, "grad_norm": 1.5624376064650394, "learning_rate": 4.078822022899579e-06, "loss": 0.5166, "step": 19947 }, { "epoch": 0.7106899193045585, "grad_norm": 1.5433330902688285, "learning_rate": 4.077892160797468e-06, "loss": 0.5035, "step": 19948 }, { "epoch": 0.7107255464310526, "grad_norm": 1.1268109492160616, "learning_rate": 4.076962377552357e-06, "loss": 0.3867, "step": 19949 }, { "epoch": 0.7107611735575468, "grad_norm": 1.2673632533704753, "learning_rate": 4.0760326731766376e-06, "loss": 0.417, "step": 19950 }, { "epoch": 0.7107968006840408, "grad_norm": 1.234648245818022, "learning_rate": 4.075103047682679e-06, "loss": 0.4122, "step": 19951 }, { "epoch": 0.7108324278105349, "grad_norm": 1.6425288741786916, "learning_rate": 4.074173501082865e-06, "loss": 0.4544, "step": 19952 }, { "epoch": 0.7108680549370291, "grad_norm": 2.5407812976673667, "learning_rate": 4.073244033389572e-06, "loss": 0.5452, "step": 19953 }, { "epoch": 0.7109036820635232, "grad_norm": 1.2386824861718757, "learning_rate": 4.0723146446151815e-06, "loss": 0.4448, "step": 19954 }, { "epoch": 0.7109393091900172, "grad_norm": 1.4596312087345797, "learning_rate": 4.071385334772061e-06, "loss": 0.5412, "step": 19955 }, { "epoch": 0.7109749363165114, "grad_norm": 1.1940669990754622, "learning_rate": 4.070456103872589e-06, "loss": 0.3936, "step": 19956 }, { "epoch": 0.7110105634430055, "grad_norm": 1.2626456805344324, "learning_rate": 4.0695269519291416e-06, "loss": 0.4737, "step": 19957 }, { "epoch": 0.7110461905694996, "grad_norm": 1.1910899589637611, "learning_rate": 4.068597878954085e-06, "loss": 0.4083, "step": 19958 }, { "epoch": 0.7110818176959938, "grad_norm": 1.2928116839474253, "learning_rate": 4.067668884959795e-06, "loss": 0.4039, "step": 19959 }, { "epoch": 0.7111174448224878, "grad_norm": 1.206273866940081, "learning_rate": 4.0667399699586355e-06, "loss": 0.4518, "step": 19960 }, { "epoch": 0.7111530719489819, "grad_norm": 1.3491459790744909, "learning_rate": 4.065811133962987e-06, "loss": 0.4003, "step": 19961 }, { "epoch": 0.7111886990754761, "grad_norm": 1.3702643347606838, "learning_rate": 4.064882376985206e-06, "loss": 0.5142, "step": 19962 }, { "epoch": 0.7112243262019702, "grad_norm": 1.4542667317541718, "learning_rate": 4.06395369903767e-06, "loss": 0.5179, "step": 19963 }, { "epoch": 0.7112599533284643, "grad_norm": 1.416387386322391, "learning_rate": 4.063025100132735e-06, "loss": 0.5383, "step": 19964 }, { "epoch": 0.7112955804549584, "grad_norm": 1.545905364823352, "learning_rate": 4.062096580282775e-06, "loss": 0.4857, "step": 19965 }, { "epoch": 0.7113312075814525, "grad_norm": 1.1534176236864917, "learning_rate": 4.061168139500145e-06, "loss": 0.3981, "step": 19966 }, { "epoch": 0.7113668347079466, "grad_norm": 1.0899233261543033, "learning_rate": 4.0602397777972125e-06, "loss": 0.426, "step": 19967 }, { "epoch": 0.7114024618344408, "grad_norm": 1.3695529155003023, "learning_rate": 4.059311495186338e-06, "loss": 0.5075, "step": 19968 }, { "epoch": 0.7114380889609349, "grad_norm": 1.743475901353651, "learning_rate": 4.058383291679884e-06, "loss": 0.5524, "step": 19969 }, { "epoch": 0.7114737160874289, "grad_norm": 1.1001762756851963, "learning_rate": 4.057455167290213e-06, "loss": 0.3773, "step": 19970 }, { "epoch": 0.7115093432139231, "grad_norm": 1.289958932589086, "learning_rate": 4.056527122029677e-06, "loss": 0.4575, "step": 19971 }, { "epoch": 0.7115449703404172, "grad_norm": 1.7901395180723216, "learning_rate": 4.0555991559106385e-06, "loss": 0.4947, "step": 19972 }, { "epoch": 0.7115805974669113, "grad_norm": 1.3567783006472742, "learning_rate": 4.05467126894545e-06, "loss": 0.4803, "step": 19973 }, { "epoch": 0.7116162245934055, "grad_norm": 1.6486061082620322, "learning_rate": 4.05374346114647e-06, "loss": 0.5258, "step": 19974 }, { "epoch": 0.7116518517198995, "grad_norm": 1.479253430671032, "learning_rate": 4.05281573252605e-06, "loss": 0.5429, "step": 19975 }, { "epoch": 0.7116874788463936, "grad_norm": 1.4245226162477367, "learning_rate": 4.05188808309655e-06, "loss": 0.4089, "step": 19976 }, { "epoch": 0.7117231059728878, "grad_norm": 1.545992835112707, "learning_rate": 4.050960512870313e-06, "loss": 0.521, "step": 19977 }, { "epoch": 0.7117587330993819, "grad_norm": 1.4683166921048938, "learning_rate": 4.0500330218596965e-06, "loss": 0.5532, "step": 19978 }, { "epoch": 0.7117943602258759, "grad_norm": 1.1430897901988053, "learning_rate": 4.049105610077052e-06, "loss": 0.3589, "step": 19979 }, { "epoch": 0.7118299873523701, "grad_norm": 1.6401805506307632, "learning_rate": 4.048178277534722e-06, "loss": 0.5342, "step": 19980 }, { "epoch": 0.7118656144788642, "grad_norm": 1.2647497404317172, "learning_rate": 4.0472510242450614e-06, "loss": 0.4067, "step": 19981 }, { "epoch": 0.7119012416053583, "grad_norm": 1.703861091319673, "learning_rate": 4.046323850220412e-06, "loss": 0.5136, "step": 19982 }, { "epoch": 0.7119368687318525, "grad_norm": 1.5907626694299593, "learning_rate": 4.045396755473121e-06, "loss": 0.5668, "step": 19983 }, { "epoch": 0.7119724958583465, "grad_norm": 1.7176363379096389, "learning_rate": 4.0444697400155345e-06, "loss": 0.4435, "step": 19984 }, { "epoch": 0.7120081229848406, "grad_norm": 1.287141016595197, "learning_rate": 4.043542803859999e-06, "loss": 0.4399, "step": 19985 }, { "epoch": 0.7120437501113348, "grad_norm": 1.4062862068578914, "learning_rate": 4.042615947018851e-06, "loss": 0.5433, "step": 19986 }, { "epoch": 0.7120793772378289, "grad_norm": 1.190565963474987, "learning_rate": 4.041689169504439e-06, "loss": 0.3743, "step": 19987 }, { "epoch": 0.712115004364323, "grad_norm": 1.3072739049668334, "learning_rate": 4.040762471329097e-06, "loss": 0.4235, "step": 19988 }, { "epoch": 0.7121506314908171, "grad_norm": 1.6513268348109116, "learning_rate": 4.039835852505169e-06, "loss": 0.5253, "step": 19989 }, { "epoch": 0.7121862586173112, "grad_norm": 1.5884034753590464, "learning_rate": 4.038909313044991e-06, "loss": 0.4592, "step": 19990 }, { "epoch": 0.7122218857438053, "grad_norm": 1.2342113141576525, "learning_rate": 4.037982852960902e-06, "loss": 0.4178, "step": 19991 }, { "epoch": 0.7122575128702995, "grad_norm": 1.3017194388139655, "learning_rate": 4.037056472265244e-06, "loss": 0.4247, "step": 19992 }, { "epoch": 0.7122931399967936, "grad_norm": 1.4388796723769384, "learning_rate": 4.036130170970342e-06, "loss": 0.5013, "step": 19993 }, { "epoch": 0.7123287671232876, "grad_norm": 1.0406211131341567, "learning_rate": 4.035203949088538e-06, "loss": 0.4206, "step": 19994 }, { "epoch": 0.7123643942497818, "grad_norm": 1.7367883122793715, "learning_rate": 4.03427780663216e-06, "loss": 0.5128, "step": 19995 }, { "epoch": 0.7124000213762759, "grad_norm": 1.943457337577756, "learning_rate": 4.033351743613546e-06, "loss": 0.4842, "step": 19996 }, { "epoch": 0.71243564850277, "grad_norm": 1.297351777556697, "learning_rate": 4.0324257600450215e-06, "loss": 0.3775, "step": 19997 }, { "epoch": 0.7124712756292642, "grad_norm": 1.228713161110135, "learning_rate": 4.0314998559389185e-06, "loss": 0.4108, "step": 19998 }, { "epoch": 0.7125069027557582, "grad_norm": 1.2143100347042537, "learning_rate": 4.030574031307566e-06, "loss": 0.4241, "step": 19999 }, { "epoch": 0.7125425298822523, "grad_norm": 1.1636697995162362, "learning_rate": 4.0296482861632965e-06, "loss": 0.4281, "step": 20000 }, { "epoch": 0.7125781570087465, "grad_norm": 1.633026864801619, "learning_rate": 4.0287226205184294e-06, "loss": 0.5613, "step": 20001 }, { "epoch": 0.7126137841352406, "grad_norm": 1.3438210548218912, "learning_rate": 4.027797034385295e-06, "loss": 0.4134, "step": 20002 }, { "epoch": 0.7126494112617346, "grad_norm": 1.1193848393038643, "learning_rate": 4.02687152777622e-06, "loss": 0.3653, "step": 20003 }, { "epoch": 0.7126850383882288, "grad_norm": 1.605681382033145, "learning_rate": 4.0259461007035225e-06, "loss": 0.5933, "step": 20004 }, { "epoch": 0.7127206655147229, "grad_norm": 1.2550822884965733, "learning_rate": 4.0250207531795284e-06, "loss": 0.4905, "step": 20005 }, { "epoch": 0.712756292641217, "grad_norm": 1.3504153183984073, "learning_rate": 4.024095485216559e-06, "loss": 0.4075, "step": 20006 }, { "epoch": 0.7127919197677112, "grad_norm": 1.533367271976344, "learning_rate": 4.023170296826939e-06, "loss": 0.4687, "step": 20007 }, { "epoch": 0.7128275468942052, "grad_norm": 1.2964643129951103, "learning_rate": 4.022245188022981e-06, "loss": 0.489, "step": 20008 }, { "epoch": 0.7128631740206993, "grad_norm": 1.313427127260295, "learning_rate": 4.0213201588170094e-06, "loss": 0.4245, "step": 20009 }, { "epoch": 0.7128988011471935, "grad_norm": 1.3807695239491204, "learning_rate": 4.020395209221335e-06, "loss": 0.544, "step": 20010 }, { "epoch": 0.7129344282736876, "grad_norm": 1.6744247183694454, "learning_rate": 4.019470339248278e-06, "loss": 0.5302, "step": 20011 }, { "epoch": 0.7129700554001817, "grad_norm": 1.0671709045853204, "learning_rate": 4.018545548910153e-06, "loss": 0.3765, "step": 20012 }, { "epoch": 0.7130056825266758, "grad_norm": 1.260023633879235, "learning_rate": 4.017620838219276e-06, "loss": 0.4154, "step": 20013 }, { "epoch": 0.7130413096531699, "grad_norm": 1.6112990565794738, "learning_rate": 4.016696207187962e-06, "loss": 0.5352, "step": 20014 }, { "epoch": 0.713076936779664, "grad_norm": 1.2999654774074454, "learning_rate": 4.015771655828517e-06, "loss": 0.4106, "step": 20015 }, { "epoch": 0.7131125639061582, "grad_norm": 1.4614780061994572, "learning_rate": 4.014847184153258e-06, "loss": 0.4932, "step": 20016 }, { "epoch": 0.7131481910326523, "grad_norm": 1.3922493980881954, "learning_rate": 4.013922792174488e-06, "loss": 0.4807, "step": 20017 }, { "epoch": 0.7131838181591463, "grad_norm": 3.247841288513733, "learning_rate": 4.012998479904525e-06, "loss": 0.3991, "step": 20018 }, { "epoch": 0.7132194452856405, "grad_norm": 1.6757046328576917, "learning_rate": 4.012074247355667e-06, "loss": 0.4734, "step": 20019 }, { "epoch": 0.7132550724121346, "grad_norm": 1.4701715268203224, "learning_rate": 4.011150094540227e-06, "loss": 0.532, "step": 20020 }, { "epoch": 0.7132906995386287, "grad_norm": 1.2678384834050116, "learning_rate": 4.0102260214705094e-06, "loss": 0.3894, "step": 20021 }, { "epoch": 0.7133263266651229, "grad_norm": 1.7610573345173721, "learning_rate": 4.009302028158821e-06, "loss": 0.5227, "step": 20022 }, { "epoch": 0.7133619537916169, "grad_norm": 1.2931123982529713, "learning_rate": 4.00837811461746e-06, "loss": 0.4294, "step": 20023 }, { "epoch": 0.713397580918111, "grad_norm": 1.6368410964977376, "learning_rate": 4.007454280858733e-06, "loss": 0.4944, "step": 20024 }, { "epoch": 0.7134332080446052, "grad_norm": 1.3651292065045968, "learning_rate": 4.006530526894945e-06, "loss": 0.4411, "step": 20025 }, { "epoch": 0.7134688351710993, "grad_norm": 1.2023651384046312, "learning_rate": 4.005606852738388e-06, "loss": 0.4417, "step": 20026 }, { "epoch": 0.7135044622975933, "grad_norm": 1.4593058104714574, "learning_rate": 4.0046832584013666e-06, "loss": 0.54, "step": 20027 }, { "epoch": 0.7135400894240875, "grad_norm": 1.3296478441810888, "learning_rate": 4.003759743896176e-06, "loss": 0.3986, "step": 20028 }, { "epoch": 0.7135757165505816, "grad_norm": 1.543997566812207, "learning_rate": 4.002836309235121e-06, "loss": 0.4719, "step": 20029 }, { "epoch": 0.7136113436770757, "grad_norm": 1.2862685591597398, "learning_rate": 4.001912954430487e-06, "loss": 0.4066, "step": 20030 }, { "epoch": 0.7136469708035699, "grad_norm": 1.5390674101941566, "learning_rate": 4.000989679494579e-06, "loss": 0.4654, "step": 20031 }, { "epoch": 0.7136825979300639, "grad_norm": 1.1771511807128563, "learning_rate": 4.000066484439683e-06, "loss": 0.4633, "step": 20032 }, { "epoch": 0.713718225056558, "grad_norm": 1.7768536626932236, "learning_rate": 3.999143369278099e-06, "loss": 0.5777, "step": 20033 }, { "epoch": 0.7137538521830522, "grad_norm": 1.3029918860684087, "learning_rate": 3.998220334022111e-06, "loss": 0.4156, "step": 20034 }, { "epoch": 0.7137894793095463, "grad_norm": 1.1116089087443306, "learning_rate": 3.997297378684015e-06, "loss": 0.4394, "step": 20035 }, { "epoch": 0.7138251064360404, "grad_norm": 1.8890442792613302, "learning_rate": 3.9963745032761005e-06, "loss": 0.5006, "step": 20036 }, { "epoch": 0.7138607335625345, "grad_norm": 1.4922519780890644, "learning_rate": 3.995451707810655e-06, "loss": 0.4824, "step": 20037 }, { "epoch": 0.7138963606890286, "grad_norm": 1.2878069884612513, "learning_rate": 3.9945289922999705e-06, "loss": 0.462, "step": 20038 }, { "epoch": 0.7139319878155227, "grad_norm": 1.2180319524213832, "learning_rate": 3.9936063567563275e-06, "loss": 0.4287, "step": 20039 }, { "epoch": 0.7139676149420169, "grad_norm": 1.5741370242366666, "learning_rate": 3.992683801192016e-06, "loss": 0.5651, "step": 20040 }, { "epoch": 0.714003242068511, "grad_norm": 1.0140941417852363, "learning_rate": 3.9917613256193164e-06, "loss": 0.3709, "step": 20041 }, { "epoch": 0.714038869195005, "grad_norm": 1.4889332651829017, "learning_rate": 3.990838930050513e-06, "loss": 0.4764, "step": 20042 }, { "epoch": 0.7140744963214992, "grad_norm": 1.9995328825925045, "learning_rate": 3.989916614497891e-06, "loss": 0.5072, "step": 20043 }, { "epoch": 0.7141101234479933, "grad_norm": 2.740642061845559, "learning_rate": 3.988994378973733e-06, "loss": 0.5125, "step": 20044 }, { "epoch": 0.7141457505744874, "grad_norm": 1.1152881272756572, "learning_rate": 3.988072223490312e-06, "loss": 0.3765, "step": 20045 }, { "epoch": 0.7141813777009816, "grad_norm": 1.0352182174938436, "learning_rate": 3.9871501480599154e-06, "loss": 0.4179, "step": 20046 }, { "epoch": 0.7142170048274756, "grad_norm": 1.2073602927976153, "learning_rate": 3.986228152694813e-06, "loss": 0.441, "step": 20047 }, { "epoch": 0.7142526319539697, "grad_norm": 1.4456539938710926, "learning_rate": 3.9853062374072884e-06, "loss": 0.4449, "step": 20048 }, { "epoch": 0.7142882590804639, "grad_norm": 1.6939430989834865, "learning_rate": 3.984384402209613e-06, "loss": 0.489, "step": 20049 }, { "epoch": 0.714323886206958, "grad_norm": 1.204764801421919, "learning_rate": 3.983462647114065e-06, "loss": 0.425, "step": 20050 }, { "epoch": 0.714359513333452, "grad_norm": 2.340780888160537, "learning_rate": 3.98254097213292e-06, "loss": 0.6032, "step": 20051 }, { "epoch": 0.7143951404599462, "grad_norm": 1.148872385787534, "learning_rate": 3.981619377278445e-06, "loss": 0.4051, "step": 20052 }, { "epoch": 0.7144307675864403, "grad_norm": 1.4935272226822685, "learning_rate": 3.980697862562918e-06, "loss": 0.5258, "step": 20053 }, { "epoch": 0.7144663947129344, "grad_norm": 1.4735348272153754, "learning_rate": 3.979776427998603e-06, "loss": 0.4988, "step": 20054 }, { "epoch": 0.7145020218394286, "grad_norm": 1.1320520034258101, "learning_rate": 3.978855073597776e-06, "loss": 0.4034, "step": 20055 }, { "epoch": 0.7145376489659226, "grad_norm": 2.0631551459888486, "learning_rate": 3.977933799372699e-06, "loss": 0.5979, "step": 20056 }, { "epoch": 0.7145732760924167, "grad_norm": 1.4579549250203225, "learning_rate": 3.977012605335642e-06, "loss": 0.6394, "step": 20057 }, { "epoch": 0.7146089032189109, "grad_norm": 1.343256718197015, "learning_rate": 3.9760914914988716e-06, "loss": 0.4105, "step": 20058 }, { "epoch": 0.714644530345405, "grad_norm": 1.1907948690953636, "learning_rate": 3.975170457874654e-06, "loss": 0.4112, "step": 20059 }, { "epoch": 0.7146801574718991, "grad_norm": 1.4165504258724564, "learning_rate": 3.974249504475256e-06, "loss": 0.5244, "step": 20060 }, { "epoch": 0.7147157845983932, "grad_norm": 1.5798531787618597, "learning_rate": 3.9733286313129346e-06, "loss": 0.5577, "step": 20061 }, { "epoch": 0.7147514117248873, "grad_norm": 1.7177814486837168, "learning_rate": 3.972407838399959e-06, "loss": 0.4565, "step": 20062 }, { "epoch": 0.7147870388513814, "grad_norm": 1.5617059542437473, "learning_rate": 3.971487125748581e-06, "loss": 0.5634, "step": 20063 }, { "epoch": 0.7148226659778756, "grad_norm": 1.5578856276142692, "learning_rate": 3.9705664933710665e-06, "loss": 0.5442, "step": 20064 }, { "epoch": 0.7148582931043697, "grad_norm": 1.6336893581621015, "learning_rate": 3.969645941279674e-06, "loss": 0.5397, "step": 20065 }, { "epoch": 0.7148939202308637, "grad_norm": 1.1894830557363942, "learning_rate": 3.9687254694866625e-06, "loss": 0.4347, "step": 20066 }, { "epoch": 0.7149295473573579, "grad_norm": 1.5767752350376645, "learning_rate": 3.9678050780042855e-06, "loss": 0.5257, "step": 20067 }, { "epoch": 0.714965174483852, "grad_norm": 1.1743617240968867, "learning_rate": 3.966884766844803e-06, "loss": 0.3753, "step": 20068 }, { "epoch": 0.7150008016103461, "grad_norm": 1.3008946854251415, "learning_rate": 3.9659645360204625e-06, "loss": 0.535, "step": 20069 }, { "epoch": 0.7150364287368403, "grad_norm": 1.043328421540554, "learning_rate": 3.965044385543526e-06, "loss": 0.3544, "step": 20070 }, { "epoch": 0.7150720558633343, "grad_norm": 1.2554048859068676, "learning_rate": 3.964124315426237e-06, "loss": 0.4273, "step": 20071 }, { "epoch": 0.7151076829898284, "grad_norm": 1.2227231923319979, "learning_rate": 3.963204325680853e-06, "loss": 0.4107, "step": 20072 }, { "epoch": 0.7151433101163226, "grad_norm": 1.2629399685166416, "learning_rate": 3.962284416319623e-06, "loss": 0.4438, "step": 20073 }, { "epoch": 0.7151789372428167, "grad_norm": 1.862076333770391, "learning_rate": 3.9613645873547944e-06, "loss": 0.5597, "step": 20074 }, { "epoch": 0.7152145643693107, "grad_norm": 1.2033323726651666, "learning_rate": 3.960444838798623e-06, "loss": 0.4015, "step": 20075 }, { "epoch": 0.7152501914958049, "grad_norm": 1.0878548581260932, "learning_rate": 3.959525170663344e-06, "loss": 0.3423, "step": 20076 }, { "epoch": 0.715285818622299, "grad_norm": 1.2635926479582764, "learning_rate": 3.9586055829612145e-06, "loss": 0.4276, "step": 20077 }, { "epoch": 0.7153214457487931, "grad_norm": 1.1583933446345347, "learning_rate": 3.95768607570447e-06, "loss": 0.4293, "step": 20078 }, { "epoch": 0.7153570728752873, "grad_norm": 1.245605707055397, "learning_rate": 3.956766648905359e-06, "loss": 0.4112, "step": 20079 }, { "epoch": 0.7153927000017813, "grad_norm": 1.1908901267734846, "learning_rate": 3.955847302576125e-06, "loss": 0.3774, "step": 20080 }, { "epoch": 0.7154283271282754, "grad_norm": 2.1111014053020636, "learning_rate": 3.954928036729011e-06, "loss": 0.4275, "step": 20081 }, { "epoch": 0.7154639542547696, "grad_norm": 1.3564459721803406, "learning_rate": 3.954008851376252e-06, "loss": 0.4211, "step": 20082 }, { "epoch": 0.7154995813812637, "grad_norm": 1.2663661725868773, "learning_rate": 3.953089746530091e-06, "loss": 0.3941, "step": 20083 }, { "epoch": 0.7155352085077578, "grad_norm": 1.0652956369763191, "learning_rate": 3.952170722202771e-06, "loss": 0.4121, "step": 20084 }, { "epoch": 0.715570835634252, "grad_norm": 1.8983269993137912, "learning_rate": 3.951251778406521e-06, "loss": 0.5347, "step": 20085 }, { "epoch": 0.715606462760746, "grad_norm": 1.6903289623198108, "learning_rate": 3.950332915153585e-06, "loss": 0.551, "step": 20086 }, { "epoch": 0.7156420898872401, "grad_norm": 1.5554124869354167, "learning_rate": 3.949414132456187e-06, "loss": 0.5041, "step": 20087 }, { "epoch": 0.7156777170137343, "grad_norm": 1.2455082488797176, "learning_rate": 3.948495430326576e-06, "loss": 0.4425, "step": 20088 }, { "epoch": 0.7157133441402284, "grad_norm": 1.7799063848589394, "learning_rate": 3.947576808776976e-06, "loss": 0.4964, "step": 20089 }, { "epoch": 0.7157489712667224, "grad_norm": 1.315335363155529, "learning_rate": 3.946658267819624e-06, "loss": 0.4025, "step": 20090 }, { "epoch": 0.7157845983932166, "grad_norm": 1.5184557241513537, "learning_rate": 3.945739807466744e-06, "loss": 0.5479, "step": 20091 }, { "epoch": 0.7158202255197107, "grad_norm": 1.1688953664355595, "learning_rate": 3.944821427730574e-06, "loss": 0.4201, "step": 20092 }, { "epoch": 0.7158558526462048, "grad_norm": 1.3993294498286786, "learning_rate": 3.943903128623336e-06, "loss": 0.4432, "step": 20093 }, { "epoch": 0.715891479772699, "grad_norm": 1.483335791900044, "learning_rate": 3.94298491015726e-06, "loss": 0.5037, "step": 20094 }, { "epoch": 0.715927106899193, "grad_norm": 1.400889292982978, "learning_rate": 3.942066772344574e-06, "loss": 0.5102, "step": 20095 }, { "epoch": 0.7159627340256871, "grad_norm": 1.153207876744057, "learning_rate": 3.941148715197503e-06, "loss": 0.3945, "step": 20096 }, { "epoch": 0.7159983611521813, "grad_norm": 1.5608288339358096, "learning_rate": 3.940230738728276e-06, "loss": 0.4867, "step": 20097 }, { "epoch": 0.7160339882786754, "grad_norm": 1.8474540220737283, "learning_rate": 3.9393128429491075e-06, "loss": 0.5838, "step": 20098 }, { "epoch": 0.7160696154051694, "grad_norm": 1.2154779232156123, "learning_rate": 3.938395027872228e-06, "loss": 0.4041, "step": 20099 }, { "epoch": 0.7161052425316636, "grad_norm": 1.2112666291172685, "learning_rate": 3.937477293509853e-06, "loss": 0.3937, "step": 20100 }, { "epoch": 0.7161408696581577, "grad_norm": 1.2086980126553948, "learning_rate": 3.936559639874205e-06, "loss": 0.4476, "step": 20101 }, { "epoch": 0.7161764967846518, "grad_norm": 1.4440394663926417, "learning_rate": 3.935642066977504e-06, "loss": 0.4539, "step": 20102 }, { "epoch": 0.716212123911146, "grad_norm": 1.1486558350284695, "learning_rate": 3.93472457483197e-06, "loss": 0.4148, "step": 20103 }, { "epoch": 0.71624775103764, "grad_norm": 1.163633886952962, "learning_rate": 3.933807163449815e-06, "loss": 0.4022, "step": 20104 }, { "epoch": 0.7162833781641341, "grad_norm": 1.3627932362189674, "learning_rate": 3.932889832843257e-06, "loss": 0.4513, "step": 20105 }, { "epoch": 0.7163190052906283, "grad_norm": 1.6892480210814789, "learning_rate": 3.931972583024516e-06, "loss": 0.5066, "step": 20106 }, { "epoch": 0.7163546324171224, "grad_norm": 1.48858474643774, "learning_rate": 3.931055414005798e-06, "loss": 0.4917, "step": 20107 }, { "epoch": 0.7163902595436165, "grad_norm": 1.3162803827966822, "learning_rate": 3.930138325799322e-06, "loss": 0.4147, "step": 20108 }, { "epoch": 0.7164258866701106, "grad_norm": 1.9517778007301123, "learning_rate": 3.9292213184172925e-06, "loss": 0.5109, "step": 20109 }, { "epoch": 0.7164615137966047, "grad_norm": 1.1393587925299173, "learning_rate": 3.928304391871925e-06, "loss": 0.3923, "step": 20110 }, { "epoch": 0.7164971409230988, "grad_norm": 4.837474271348256, "learning_rate": 3.927387546175428e-06, "loss": 0.4446, "step": 20111 }, { "epoch": 0.716532768049593, "grad_norm": 1.7292400793920482, "learning_rate": 3.9264707813400145e-06, "loss": 0.6128, "step": 20112 }, { "epoch": 0.7165683951760871, "grad_norm": 1.3623038776717464, "learning_rate": 3.925554097377884e-06, "loss": 0.4092, "step": 20113 }, { "epoch": 0.7166040223025811, "grad_norm": 1.6610762806100563, "learning_rate": 3.92463749430125e-06, "loss": 0.5872, "step": 20114 }, { "epoch": 0.7166396494290753, "grad_norm": 1.3053401591863156, "learning_rate": 3.923720972122311e-06, "loss": 0.4133, "step": 20115 }, { "epoch": 0.7166752765555694, "grad_norm": 1.6781582679954714, "learning_rate": 3.922804530853274e-06, "loss": 0.4552, "step": 20116 }, { "epoch": 0.7167109036820635, "grad_norm": 1.695461269010825, "learning_rate": 3.921888170506342e-06, "loss": 0.47, "step": 20117 }, { "epoch": 0.7167465308085577, "grad_norm": 1.2624268990080998, "learning_rate": 3.9209718910937174e-06, "loss": 0.481, "step": 20118 }, { "epoch": 0.7167821579350517, "grad_norm": 1.2107462313087198, "learning_rate": 3.9200556926276046e-06, "loss": 0.3793, "step": 20119 }, { "epoch": 0.7168177850615458, "grad_norm": 1.6763987425521019, "learning_rate": 3.919139575120197e-06, "loss": 0.4908, "step": 20120 }, { "epoch": 0.71685341218804, "grad_norm": 1.406059971323562, "learning_rate": 3.918223538583699e-06, "loss": 0.4882, "step": 20121 }, { "epoch": 0.7168890393145341, "grad_norm": 1.4122225581242294, "learning_rate": 3.917307583030303e-06, "loss": 0.4604, "step": 20122 }, { "epoch": 0.7169246664410281, "grad_norm": 1.4027058126763632, "learning_rate": 3.91639170847221e-06, "loss": 0.4998, "step": 20123 }, { "epoch": 0.7169602935675223, "grad_norm": 1.4405284725209475, "learning_rate": 3.915475914921608e-06, "loss": 0.3997, "step": 20124 }, { "epoch": 0.7169959206940164, "grad_norm": 1.1959286291869755, "learning_rate": 3.914560202390703e-06, "loss": 0.4084, "step": 20125 }, { "epoch": 0.7170315478205105, "grad_norm": 1.3723636366053884, "learning_rate": 3.913644570891681e-06, "loss": 0.52, "step": 20126 }, { "epoch": 0.7170671749470047, "grad_norm": 1.0492281644225763, "learning_rate": 3.912729020436737e-06, "loss": 0.3746, "step": 20127 }, { "epoch": 0.7171028020734987, "grad_norm": 1.7190937640082957, "learning_rate": 3.911813551038057e-06, "loss": 0.5992, "step": 20128 }, { "epoch": 0.7171384291999928, "grad_norm": 1.4308211644861972, "learning_rate": 3.910898162707836e-06, "loss": 0.4753, "step": 20129 }, { "epoch": 0.717174056326487, "grad_norm": 1.1037171626559414, "learning_rate": 3.909982855458265e-06, "loss": 0.4119, "step": 20130 }, { "epoch": 0.7172096834529811, "grad_norm": 1.5790019882032094, "learning_rate": 3.9090676293015265e-06, "loss": 0.5955, "step": 20131 }, { "epoch": 0.7172453105794752, "grad_norm": 1.1684827247000629, "learning_rate": 3.908152484249809e-06, "loss": 0.3978, "step": 20132 }, { "epoch": 0.7172809377059693, "grad_norm": 1.3767117654826615, "learning_rate": 3.907237420315298e-06, "loss": 0.5463, "step": 20133 }, { "epoch": 0.7173165648324634, "grad_norm": 1.1301276018691881, "learning_rate": 3.906322437510185e-06, "loss": 0.4073, "step": 20134 }, { "epoch": 0.7173521919589575, "grad_norm": 1.4091453677286747, "learning_rate": 3.905407535846644e-06, "loss": 0.4559, "step": 20135 }, { "epoch": 0.7173878190854517, "grad_norm": 1.4166067810960419, "learning_rate": 3.904492715336864e-06, "loss": 0.4942, "step": 20136 }, { "epoch": 0.7174234462119458, "grad_norm": 1.262747561933598, "learning_rate": 3.903577975993021e-06, "loss": 0.3898, "step": 20137 }, { "epoch": 0.7174590733384398, "grad_norm": 1.3420588264230917, "learning_rate": 3.9026633178272985e-06, "loss": 0.4489, "step": 20138 }, { "epoch": 0.717494700464934, "grad_norm": 1.1030403588932671, "learning_rate": 3.901748740851876e-06, "loss": 0.3971, "step": 20139 }, { "epoch": 0.7175303275914281, "grad_norm": 1.0933005245398226, "learning_rate": 3.90083424507893e-06, "loss": 0.4066, "step": 20140 }, { "epoch": 0.7175659547179222, "grad_norm": 1.0803695351383784, "learning_rate": 3.899919830520644e-06, "loss": 0.4083, "step": 20141 }, { "epoch": 0.7176015818444164, "grad_norm": 1.4019929865210192, "learning_rate": 3.899005497189186e-06, "loss": 0.4934, "step": 20142 }, { "epoch": 0.7176372089709104, "grad_norm": 1.4505344017290966, "learning_rate": 3.8980912450967366e-06, "loss": 0.5011, "step": 20143 }, { "epoch": 0.7176728360974045, "grad_norm": 1.3275003363236784, "learning_rate": 3.897177074255464e-06, "loss": 0.3968, "step": 20144 }, { "epoch": 0.7177084632238987, "grad_norm": 1.2459264140014021, "learning_rate": 3.896262984677548e-06, "loss": 0.3861, "step": 20145 }, { "epoch": 0.7177440903503928, "grad_norm": 1.6230060466920346, "learning_rate": 3.895348976375153e-06, "loss": 0.5164, "step": 20146 }, { "epoch": 0.7177797174768868, "grad_norm": 1.6259126302935751, "learning_rate": 3.894435049360453e-06, "loss": 0.4731, "step": 20147 }, { "epoch": 0.717815344603381, "grad_norm": 1.505877402990364, "learning_rate": 3.893521203645618e-06, "loss": 0.5414, "step": 20148 }, { "epoch": 0.7178509717298751, "grad_norm": 1.2023830813846097, "learning_rate": 3.89260743924282e-06, "loss": 0.4528, "step": 20149 }, { "epoch": 0.7178865988563692, "grad_norm": 1.4782443820392281, "learning_rate": 3.8916937561642195e-06, "loss": 0.5407, "step": 20150 }, { "epoch": 0.7179222259828634, "grad_norm": 1.2737780288843608, "learning_rate": 3.890780154421986e-06, "loss": 0.4366, "step": 20151 }, { "epoch": 0.7179578531093574, "grad_norm": 1.4736962475573718, "learning_rate": 3.889866634028287e-06, "loss": 0.5661, "step": 20152 }, { "epoch": 0.7179934802358515, "grad_norm": 1.2206274422815961, "learning_rate": 3.888953194995283e-06, "loss": 0.4241, "step": 20153 }, { "epoch": 0.7180291073623457, "grad_norm": 1.466213670515007, "learning_rate": 3.888039837335138e-06, "loss": 0.4756, "step": 20154 }, { "epoch": 0.7180647344888398, "grad_norm": 1.7163847718282126, "learning_rate": 3.887126561060014e-06, "loss": 0.4994, "step": 20155 }, { "epoch": 0.7181003616153339, "grad_norm": 1.4261389036721892, "learning_rate": 3.886213366182077e-06, "loss": 0.5637, "step": 20156 }, { "epoch": 0.718135988741828, "grad_norm": 1.2590799222968752, "learning_rate": 3.885300252713477e-06, "loss": 0.3805, "step": 20157 }, { "epoch": 0.7181716158683221, "grad_norm": 1.4276395962124884, "learning_rate": 3.884387220666383e-06, "loss": 0.5757, "step": 20158 }, { "epoch": 0.7182072429948163, "grad_norm": 1.2115326499676653, "learning_rate": 3.883474270052944e-06, "loss": 0.4332, "step": 20159 }, { "epoch": 0.7182428701213104, "grad_norm": 1.5214076662911757, "learning_rate": 3.882561400885324e-06, "loss": 0.518, "step": 20160 }, { "epoch": 0.7182784972478045, "grad_norm": 1.1532212429961903, "learning_rate": 3.881648613175671e-06, "loss": 0.4578, "step": 20161 }, { "epoch": 0.7183141243742986, "grad_norm": 1.189278514211172, "learning_rate": 3.880735906936143e-06, "loss": 0.4255, "step": 20162 }, { "epoch": 0.7183497515007927, "grad_norm": 1.3103745447230286, "learning_rate": 3.879823282178894e-06, "loss": 0.4553, "step": 20163 }, { "epoch": 0.7183853786272868, "grad_norm": 1.389273172996013, "learning_rate": 3.8789107389160755e-06, "loss": 0.4659, "step": 20164 }, { "epoch": 0.718421005753781, "grad_norm": 1.06828962348104, "learning_rate": 3.8779982771598414e-06, "loss": 0.3915, "step": 20165 }, { "epoch": 0.7184566328802751, "grad_norm": 1.7085928816142673, "learning_rate": 3.877085896922337e-06, "loss": 0.5187, "step": 20166 }, { "epoch": 0.7184922600067691, "grad_norm": 1.170695039680777, "learning_rate": 3.876173598215717e-06, "loss": 0.4284, "step": 20167 }, { "epoch": 0.7185278871332633, "grad_norm": 1.7461228690728736, "learning_rate": 3.875261381052121e-06, "loss": 0.5591, "step": 20168 }, { "epoch": 0.7185635142597574, "grad_norm": 1.481718130817291, "learning_rate": 3.874349245443702e-06, "loss": 0.5193, "step": 20169 }, { "epoch": 0.7185991413862515, "grad_norm": 1.8086530868265345, "learning_rate": 3.873437191402605e-06, "loss": 0.4821, "step": 20170 }, { "epoch": 0.7186347685127457, "grad_norm": 1.4840337464519362, "learning_rate": 3.872525218940976e-06, "loss": 0.5562, "step": 20171 }, { "epoch": 0.7186703956392397, "grad_norm": 1.3496913820752046, "learning_rate": 3.871613328070953e-06, "loss": 0.3847, "step": 20172 }, { "epoch": 0.7187060227657338, "grad_norm": 1.2945267978027346, "learning_rate": 3.870701518804687e-06, "loss": 0.4173, "step": 20173 }, { "epoch": 0.718741649892228, "grad_norm": 2.155569591661722, "learning_rate": 3.869789791154309e-06, "loss": 0.4888, "step": 20174 }, { "epoch": 0.7187772770187221, "grad_norm": 1.3085698341944172, "learning_rate": 3.868878145131965e-06, "loss": 0.4235, "step": 20175 }, { "epoch": 0.7188129041452161, "grad_norm": 1.2633149266783734, "learning_rate": 3.867966580749796e-06, "loss": 0.4308, "step": 20176 }, { "epoch": 0.7188485312717103, "grad_norm": 1.5036628534226353, "learning_rate": 3.867055098019936e-06, "loss": 0.5493, "step": 20177 }, { "epoch": 0.7188841583982044, "grad_norm": 1.2135187068435225, "learning_rate": 3.8661436969545285e-06, "loss": 0.4586, "step": 20178 }, { "epoch": 0.7189197855246985, "grad_norm": 1.2961918028701083, "learning_rate": 3.865232377565701e-06, "loss": 0.5738, "step": 20179 }, { "epoch": 0.7189554126511927, "grad_norm": 1.6109854337676768, "learning_rate": 3.864321139865597e-06, "loss": 0.5091, "step": 20180 }, { "epoch": 0.7189910397776867, "grad_norm": 1.7514562303826593, "learning_rate": 3.863409983866341e-06, "loss": 0.4852, "step": 20181 }, { "epoch": 0.7190266669041808, "grad_norm": 1.1861787385918114, "learning_rate": 3.862498909580074e-06, "loss": 0.4154, "step": 20182 }, { "epoch": 0.719062294030675, "grad_norm": 1.542923005899317, "learning_rate": 3.86158791701892e-06, "loss": 0.4658, "step": 20183 }, { "epoch": 0.7190979211571691, "grad_norm": 1.5416597340073044, "learning_rate": 3.860677006195012e-06, "loss": 0.4868, "step": 20184 }, { "epoch": 0.7191335482836632, "grad_norm": 1.2300129582424186, "learning_rate": 3.859766177120483e-06, "loss": 0.4256, "step": 20185 }, { "epoch": 0.7191691754101573, "grad_norm": 2.2598910742039724, "learning_rate": 3.858855429807459e-06, "loss": 0.4525, "step": 20186 }, { "epoch": 0.7192048025366514, "grad_norm": 1.242584608185906, "learning_rate": 3.85794476426807e-06, "loss": 0.4604, "step": 20187 }, { "epoch": 0.7192404296631455, "grad_norm": 1.682302735338136, "learning_rate": 3.857034180514437e-06, "loss": 0.5341, "step": 20188 }, { "epoch": 0.7192760567896397, "grad_norm": 2.3386672064912943, "learning_rate": 3.856123678558691e-06, "loss": 0.5417, "step": 20189 }, { "epoch": 0.7193116839161338, "grad_norm": 1.161702319945731, "learning_rate": 3.855213258412948e-06, "loss": 0.3818, "step": 20190 }, { "epoch": 0.7193473110426278, "grad_norm": 1.1785565691412792, "learning_rate": 3.854302920089336e-06, "loss": 0.4156, "step": 20191 }, { "epoch": 0.719382938169122, "grad_norm": 1.5065782801484464, "learning_rate": 3.853392663599977e-06, "loss": 0.5161, "step": 20192 }, { "epoch": 0.7194185652956161, "grad_norm": 1.286555863870673, "learning_rate": 3.852482488956992e-06, "loss": 0.4505, "step": 20193 }, { "epoch": 0.7194541924221102, "grad_norm": 1.2411477726773494, "learning_rate": 3.851572396172498e-06, "loss": 0.4308, "step": 20194 }, { "epoch": 0.7194898195486044, "grad_norm": 1.4537678752957064, "learning_rate": 3.8506623852586174e-06, "loss": 0.4718, "step": 20195 }, { "epoch": 0.7195254466750984, "grad_norm": 1.2874137153286893, "learning_rate": 3.849752456227463e-06, "loss": 0.4358, "step": 20196 }, { "epoch": 0.7195610738015925, "grad_norm": 1.3207757614844624, "learning_rate": 3.848842609091157e-06, "loss": 0.3984, "step": 20197 }, { "epoch": 0.7195967009280867, "grad_norm": 1.1699115091438141, "learning_rate": 3.847932843861807e-06, "loss": 0.3734, "step": 20198 }, { "epoch": 0.7196323280545808, "grad_norm": 1.1888695130415854, "learning_rate": 3.847023160551528e-06, "loss": 0.4785, "step": 20199 }, { "epoch": 0.7196679551810748, "grad_norm": 1.1841740764886362, "learning_rate": 3.8461135591724416e-06, "loss": 0.486, "step": 20200 }, { "epoch": 0.719703582307569, "grad_norm": 1.6123119522656981, "learning_rate": 3.845204039736653e-06, "loss": 0.51, "step": 20201 }, { "epoch": 0.7197392094340631, "grad_norm": 1.612628523775182, "learning_rate": 3.844294602256277e-06, "loss": 0.5927, "step": 20202 }, { "epoch": 0.7197748365605572, "grad_norm": 1.6271933343586567, "learning_rate": 3.8433852467434175e-06, "loss": 0.4383, "step": 20203 }, { "epoch": 0.7198104636870514, "grad_norm": 1.8018910881173553, "learning_rate": 3.842475973210191e-06, "loss": 0.4904, "step": 20204 }, { "epoch": 0.7198460908135454, "grad_norm": 1.549226854874988, "learning_rate": 3.841566781668696e-06, "loss": 0.532, "step": 20205 }, { "epoch": 0.7198817179400395, "grad_norm": 1.4728763035075956, "learning_rate": 3.8406576721310454e-06, "loss": 0.5148, "step": 20206 }, { "epoch": 0.7199173450665337, "grad_norm": 1.1643896498819455, "learning_rate": 3.839748644609342e-06, "loss": 0.4146, "step": 20207 }, { "epoch": 0.7199529721930278, "grad_norm": 1.2407752619141876, "learning_rate": 3.838839699115695e-06, "loss": 0.437, "step": 20208 }, { "epoch": 0.7199885993195219, "grad_norm": 1.1629691404420885, "learning_rate": 3.837930835662199e-06, "loss": 0.4359, "step": 20209 }, { "epoch": 0.720024226446016, "grad_norm": 2.1531484808550587, "learning_rate": 3.837022054260963e-06, "loss": 0.5456, "step": 20210 }, { "epoch": 0.7200598535725101, "grad_norm": 1.6862677399690051, "learning_rate": 3.836113354924088e-06, "loss": 0.52, "step": 20211 }, { "epoch": 0.7200954806990042, "grad_norm": 1.3560038802822754, "learning_rate": 3.835204737663669e-06, "loss": 0.4955, "step": 20212 }, { "epoch": 0.7201311078254984, "grad_norm": 1.3492789465880388, "learning_rate": 3.834296202491809e-06, "loss": 0.4327, "step": 20213 }, { "epoch": 0.7201667349519925, "grad_norm": 1.3217516668820806, "learning_rate": 3.833387749420603e-06, "loss": 0.4438, "step": 20214 }, { "epoch": 0.7202023620784865, "grad_norm": 1.3654841267265316, "learning_rate": 3.832479378462153e-06, "loss": 0.4269, "step": 20215 }, { "epoch": 0.7202379892049807, "grad_norm": 1.6006031903677365, "learning_rate": 3.831571089628547e-06, "loss": 0.5443, "step": 20216 }, { "epoch": 0.7202736163314748, "grad_norm": 1.469457129067349, "learning_rate": 3.830662882931888e-06, "loss": 0.5433, "step": 20217 }, { "epoch": 0.7203092434579689, "grad_norm": 1.46431869514363, "learning_rate": 3.829754758384262e-06, "loss": 0.4897, "step": 20218 }, { "epoch": 0.7203448705844631, "grad_norm": 1.1811933129454195, "learning_rate": 3.828846715997766e-06, "loss": 0.4182, "step": 20219 }, { "epoch": 0.7203804977109571, "grad_norm": 1.3011360358618798, "learning_rate": 3.8279387557844875e-06, "loss": 0.5973, "step": 20220 }, { "epoch": 0.7204161248374512, "grad_norm": 2.0571870031498443, "learning_rate": 3.8270308777565165e-06, "loss": 0.4883, "step": 20221 }, { "epoch": 0.7204517519639454, "grad_norm": 1.1466729648796852, "learning_rate": 3.826123081925946e-06, "loss": 0.4094, "step": 20222 }, { "epoch": 0.7204873790904395, "grad_norm": 1.1059175422714873, "learning_rate": 3.82521536830486e-06, "loss": 0.4118, "step": 20223 }, { "epoch": 0.7205230062169335, "grad_norm": 1.7261167165953628, "learning_rate": 3.824307736905351e-06, "loss": 0.5466, "step": 20224 }, { "epoch": 0.7205586333434277, "grad_norm": 1.5573438485153452, "learning_rate": 3.823400187739499e-06, "loss": 0.4569, "step": 20225 }, { "epoch": 0.7205942604699218, "grad_norm": 1.2928603960322402, "learning_rate": 3.822492720819393e-06, "loss": 0.4915, "step": 20226 }, { "epoch": 0.7206298875964159, "grad_norm": 1.3486011134644598, "learning_rate": 3.821585336157112e-06, "loss": 0.4253, "step": 20227 }, { "epoch": 0.7206655147229101, "grad_norm": 1.424466035620312, "learning_rate": 3.820678033764739e-06, "loss": 0.4438, "step": 20228 }, { "epoch": 0.7207011418494041, "grad_norm": 1.540249919375399, "learning_rate": 3.8197708136543585e-06, "loss": 0.4521, "step": 20229 }, { "epoch": 0.7207367689758982, "grad_norm": 1.500948382414862, "learning_rate": 3.818863675838052e-06, "loss": 0.5751, "step": 20230 }, { "epoch": 0.7207723961023924, "grad_norm": 1.4493631938777505, "learning_rate": 3.817956620327893e-06, "loss": 0.5281, "step": 20231 }, { "epoch": 0.7208080232288865, "grad_norm": 1.3298579536463861, "learning_rate": 3.817049647135962e-06, "loss": 0.4349, "step": 20232 }, { "epoch": 0.7208436503553806, "grad_norm": 1.3755444056952888, "learning_rate": 3.8161427562743415e-06, "loss": 0.4453, "step": 20233 }, { "epoch": 0.7208792774818747, "grad_norm": 1.438333487931875, "learning_rate": 3.815235947755097e-06, "loss": 0.5746, "step": 20234 }, { "epoch": 0.7209149046083688, "grad_norm": 1.7156240302676007, "learning_rate": 3.814329221590314e-06, "loss": 0.4559, "step": 20235 }, { "epoch": 0.7209505317348629, "grad_norm": 1.254021257758223, "learning_rate": 3.8134225777920565e-06, "loss": 0.4561, "step": 20236 }, { "epoch": 0.7209861588613571, "grad_norm": 1.4952946099863145, "learning_rate": 3.812516016372402e-06, "loss": 0.4841, "step": 20237 }, { "epoch": 0.7210217859878512, "grad_norm": 1.3736389953572476, "learning_rate": 3.8116095373434204e-06, "loss": 0.3821, "step": 20238 }, { "epoch": 0.7210574131143452, "grad_norm": 2.069551847074942, "learning_rate": 3.810703140717188e-06, "loss": 0.5577, "step": 20239 }, { "epoch": 0.7210930402408394, "grad_norm": 1.4517673682440557, "learning_rate": 3.809796826505765e-06, "loss": 0.4213, "step": 20240 }, { "epoch": 0.7211286673673335, "grad_norm": 1.284439179074661, "learning_rate": 3.8088905947212274e-06, "loss": 0.4742, "step": 20241 }, { "epoch": 0.7211642944938276, "grad_norm": 1.6071711994807159, "learning_rate": 3.8079844453756367e-06, "loss": 0.5198, "step": 20242 }, { "epoch": 0.7211999216203218, "grad_norm": 1.1879208085149209, "learning_rate": 3.8070783784810593e-06, "loss": 0.4126, "step": 20243 }, { "epoch": 0.7212355487468158, "grad_norm": 1.3637206511539972, "learning_rate": 3.8061723940495634e-06, "loss": 0.4001, "step": 20244 }, { "epoch": 0.7212711758733099, "grad_norm": 1.267481304705957, "learning_rate": 3.8052664920932103e-06, "loss": 0.4162, "step": 20245 }, { "epoch": 0.7213068029998041, "grad_norm": 1.634735956339245, "learning_rate": 3.8043606726240676e-06, "loss": 0.4734, "step": 20246 }, { "epoch": 0.7213424301262982, "grad_norm": 1.8740981110457509, "learning_rate": 3.8034549356541896e-06, "loss": 0.5296, "step": 20247 }, { "epoch": 0.7213780572527922, "grad_norm": 1.927372328103066, "learning_rate": 3.8025492811956434e-06, "loss": 0.5025, "step": 20248 }, { "epoch": 0.7214136843792864, "grad_norm": 1.3328018172614002, "learning_rate": 3.801643709260482e-06, "loss": 0.4339, "step": 20249 }, { "epoch": 0.7214493115057805, "grad_norm": 1.7008985501272567, "learning_rate": 3.8007382198607703e-06, "loss": 0.5576, "step": 20250 }, { "epoch": 0.7214849386322746, "grad_norm": 1.4484154585497555, "learning_rate": 3.799832813008556e-06, "loss": 0.4251, "step": 20251 }, { "epoch": 0.7215205657587688, "grad_norm": 1.2493255075987688, "learning_rate": 3.798927488715908e-06, "loss": 0.4462, "step": 20252 }, { "epoch": 0.7215561928852628, "grad_norm": 1.6102018082535463, "learning_rate": 3.79802224699487e-06, "loss": 0.5869, "step": 20253 }, { "epoch": 0.7215918200117569, "grad_norm": 1.5478501173428079, "learning_rate": 3.797117087857506e-06, "loss": 0.5562, "step": 20254 }, { "epoch": 0.7216274471382511, "grad_norm": 2.1076225614431308, "learning_rate": 3.7962120113158597e-06, "loss": 0.3928, "step": 20255 }, { "epoch": 0.7216630742647452, "grad_norm": 1.6537794647913755, "learning_rate": 3.7953070173819863e-06, "loss": 0.5245, "step": 20256 }, { "epoch": 0.7216987013912393, "grad_norm": 1.1735308663459707, "learning_rate": 3.7944021060679404e-06, "loss": 0.4133, "step": 20257 }, { "epoch": 0.7217343285177334, "grad_norm": 1.349699574252794, "learning_rate": 3.7934972773857637e-06, "loss": 0.4555, "step": 20258 }, { "epoch": 0.7217699556442275, "grad_norm": 1.5262990605785742, "learning_rate": 3.7925925313475097e-06, "loss": 0.5144, "step": 20259 }, { "epoch": 0.7218055827707216, "grad_norm": 1.4809439320833346, "learning_rate": 3.7916878679652247e-06, "loss": 0.5006, "step": 20260 }, { "epoch": 0.7218412098972158, "grad_norm": 2.0407747378670127, "learning_rate": 3.7907832872509587e-06, "loss": 0.5705, "step": 20261 }, { "epoch": 0.7218768370237099, "grad_norm": 2.5035019765080877, "learning_rate": 3.7898787892167497e-06, "loss": 0.4879, "step": 20262 }, { "epoch": 0.7219124641502039, "grad_norm": 1.4392360905740238, "learning_rate": 3.788974373874649e-06, "loss": 0.4223, "step": 20263 }, { "epoch": 0.7219480912766981, "grad_norm": 1.7538608098764112, "learning_rate": 3.788070041236691e-06, "loss": 0.6412, "step": 20264 }, { "epoch": 0.7219837184031922, "grad_norm": 1.4415269692064834, "learning_rate": 3.787165791314923e-06, "loss": 0.4363, "step": 20265 }, { "epoch": 0.7220193455296863, "grad_norm": 1.3504064193626164, "learning_rate": 3.7862616241213855e-06, "loss": 0.4573, "step": 20266 }, { "epoch": 0.7220549726561805, "grad_norm": 1.376715248072153, "learning_rate": 3.7853575396681176e-06, "loss": 0.4021, "step": 20267 }, { "epoch": 0.7220905997826745, "grad_norm": 1.6740132950631548, "learning_rate": 3.784453537967161e-06, "loss": 0.4529, "step": 20268 }, { "epoch": 0.7221262269091686, "grad_norm": 1.7311688334844177, "learning_rate": 3.7835496190305465e-06, "loss": 0.5055, "step": 20269 }, { "epoch": 0.7221618540356628, "grad_norm": 1.4966624130999588, "learning_rate": 3.782645782870317e-06, "loss": 0.4351, "step": 20270 }, { "epoch": 0.7221974811621569, "grad_norm": 1.677292093267962, "learning_rate": 3.781742029498501e-06, "loss": 0.5046, "step": 20271 }, { "epoch": 0.722233108288651, "grad_norm": 1.6296169808141816, "learning_rate": 3.78083835892714e-06, "loss": 0.5279, "step": 20272 }, { "epoch": 0.7222687354151451, "grad_norm": 1.7979894215870778, "learning_rate": 3.7799347711682597e-06, "loss": 0.5455, "step": 20273 }, { "epoch": 0.7223043625416392, "grad_norm": 1.7957636793588336, "learning_rate": 3.7790312662338957e-06, "loss": 0.5407, "step": 20274 }, { "epoch": 0.7223399896681333, "grad_norm": 1.442241006705735, "learning_rate": 3.778127844136078e-06, "loss": 0.5164, "step": 20275 }, { "epoch": 0.7223756167946275, "grad_norm": 1.4176968688282692, "learning_rate": 3.7772245048868418e-06, "loss": 0.3922, "step": 20276 }, { "epoch": 0.7224112439211215, "grad_norm": 1.286262551946688, "learning_rate": 3.776321248498206e-06, "loss": 0.4123, "step": 20277 }, { "epoch": 0.7224468710476156, "grad_norm": 1.8127618113602386, "learning_rate": 3.7754180749822033e-06, "loss": 0.5215, "step": 20278 }, { "epoch": 0.7224824981741098, "grad_norm": 1.4866642839463782, "learning_rate": 3.774514984350862e-06, "loss": 0.4959, "step": 20279 }, { "epoch": 0.7225181253006039, "grad_norm": 1.3664304785764234, "learning_rate": 3.773611976616203e-06, "loss": 0.5372, "step": 20280 }, { "epoch": 0.722553752427098, "grad_norm": 1.513727929055616, "learning_rate": 3.772709051790252e-06, "loss": 0.6087, "step": 20281 }, { "epoch": 0.7225893795535921, "grad_norm": 1.4673611353705012, "learning_rate": 3.771806209885033e-06, "loss": 0.4756, "step": 20282 }, { "epoch": 0.7226250066800862, "grad_norm": 1.2699727644887142, "learning_rate": 3.7709034509125706e-06, "loss": 0.4108, "step": 20283 }, { "epoch": 0.7226606338065803, "grad_norm": 1.2420510445692816, "learning_rate": 3.7700007748848788e-06, "loss": 0.4174, "step": 20284 }, { "epoch": 0.7226962609330745, "grad_norm": 1.515157179601548, "learning_rate": 3.769098181813986e-06, "loss": 0.5079, "step": 20285 }, { "epoch": 0.7227318880595686, "grad_norm": 1.7548380964021566, "learning_rate": 3.7681956717119005e-06, "loss": 0.5097, "step": 20286 }, { "epoch": 0.7227675151860626, "grad_norm": 1.4304923111502268, "learning_rate": 3.767293244590651e-06, "loss": 0.4123, "step": 20287 }, { "epoch": 0.7228031423125568, "grad_norm": 1.4885925050148592, "learning_rate": 3.76639090046224e-06, "loss": 0.5092, "step": 20288 }, { "epoch": 0.7228387694390509, "grad_norm": 1.4635643410035937, "learning_rate": 3.765488639338699e-06, "loss": 0.4372, "step": 20289 }, { "epoch": 0.722874396565545, "grad_norm": 1.1332200497567806, "learning_rate": 3.7645864612320305e-06, "loss": 0.363, "step": 20290 }, { "epoch": 0.7229100236920392, "grad_norm": 1.2262499924107522, "learning_rate": 3.763684366154251e-06, "loss": 0.3745, "step": 20291 }, { "epoch": 0.7229456508185332, "grad_norm": 1.76423373160814, "learning_rate": 3.762782354117378e-06, "loss": 0.4405, "step": 20292 }, { "epoch": 0.7229812779450273, "grad_norm": 1.6959391997062232, "learning_rate": 3.7618804251334127e-06, "loss": 0.5861, "step": 20293 }, { "epoch": 0.7230169050715215, "grad_norm": 1.4199630494759017, "learning_rate": 3.760978579214374e-06, "loss": 0.5248, "step": 20294 }, { "epoch": 0.7230525321980156, "grad_norm": 1.3465340265667483, "learning_rate": 3.7600768163722625e-06, "loss": 0.4161, "step": 20295 }, { "epoch": 0.7230881593245096, "grad_norm": 1.514023724904603, "learning_rate": 3.75917513661909e-06, "loss": 0.5531, "step": 20296 }, { "epoch": 0.7231237864510038, "grad_norm": 1.513264689058045, "learning_rate": 3.758273539966862e-06, "loss": 0.5963, "step": 20297 }, { "epoch": 0.7231594135774979, "grad_norm": 2.362109932578666, "learning_rate": 3.7573720264275883e-06, "loss": 0.4081, "step": 20298 }, { "epoch": 0.723195040703992, "grad_norm": 1.3466012703137245, "learning_rate": 3.7564705960132662e-06, "loss": 0.4624, "step": 20299 }, { "epoch": 0.7232306678304862, "grad_norm": 1.5761271368297427, "learning_rate": 3.755569248735905e-06, "loss": 0.5011, "step": 20300 }, { "epoch": 0.7232662949569802, "grad_norm": 1.3864962332594508, "learning_rate": 3.7546679846075006e-06, "loss": 0.5316, "step": 20301 }, { "epoch": 0.7233019220834743, "grad_norm": 1.2564678632810224, "learning_rate": 3.753766803640058e-06, "loss": 0.5248, "step": 20302 }, { "epoch": 0.7233375492099685, "grad_norm": 1.2256781338272285, "learning_rate": 3.7528657058455753e-06, "loss": 0.406, "step": 20303 }, { "epoch": 0.7233731763364626, "grad_norm": 1.2095268412448963, "learning_rate": 3.7519646912360517e-06, "loss": 0.4406, "step": 20304 }, { "epoch": 0.7234088034629567, "grad_norm": 1.3169323710597707, "learning_rate": 3.75106375982349e-06, "loss": 0.4104, "step": 20305 }, { "epoch": 0.7234444305894508, "grad_norm": 1.9300303981536637, "learning_rate": 3.7501629116198777e-06, "loss": 0.5497, "step": 20306 }, { "epoch": 0.7234800577159449, "grad_norm": 1.5214035702065538, "learning_rate": 3.749262146637219e-06, "loss": 0.4986, "step": 20307 }, { "epoch": 0.723515684842439, "grad_norm": 1.375296497042231, "learning_rate": 3.7483614648875e-06, "loss": 0.4399, "step": 20308 }, { "epoch": 0.7235513119689332, "grad_norm": 1.214767390190274, "learning_rate": 3.7474608663827215e-06, "loss": 0.3723, "step": 20309 }, { "epoch": 0.7235869390954273, "grad_norm": 1.4374450874426579, "learning_rate": 3.746560351134868e-06, "loss": 0.5191, "step": 20310 }, { "epoch": 0.7236225662219213, "grad_norm": 1.270342712006484, "learning_rate": 3.745659919155934e-06, "loss": 0.4302, "step": 20311 }, { "epoch": 0.7236581933484155, "grad_norm": 1.2458206545826067, "learning_rate": 3.74475957045791e-06, "loss": 0.439, "step": 20312 }, { "epoch": 0.7236938204749096, "grad_norm": 1.281303383580518, "learning_rate": 3.743859305052785e-06, "loss": 0.413, "step": 20313 }, { "epoch": 0.7237294476014037, "grad_norm": 1.5386075182422858, "learning_rate": 3.7429591229525484e-06, "loss": 0.4917, "step": 20314 }, { "epoch": 0.7237650747278979, "grad_norm": 1.7650090848935263, "learning_rate": 3.7420590241691823e-06, "loss": 0.5612, "step": 20315 }, { "epoch": 0.7238007018543919, "grad_norm": 1.6692486768184105, "learning_rate": 3.7411590087146767e-06, "loss": 0.5164, "step": 20316 }, { "epoch": 0.723836328980886, "grad_norm": 1.2480453346746074, "learning_rate": 3.7402590766010103e-06, "loss": 0.415, "step": 20317 }, { "epoch": 0.7238719561073802, "grad_norm": 1.403205650915261, "learning_rate": 3.7393592278401704e-06, "loss": 0.4664, "step": 20318 }, { "epoch": 0.7239075832338743, "grad_norm": 1.234129566589113, "learning_rate": 3.7384594624441383e-06, "loss": 0.4415, "step": 20319 }, { "epoch": 0.7239432103603683, "grad_norm": 1.339093384652317, "learning_rate": 3.7375597804248974e-06, "loss": 0.5322, "step": 20320 }, { "epoch": 0.7239788374868625, "grad_norm": 1.3686153772489433, "learning_rate": 3.7366601817944225e-06, "loss": 0.38, "step": 20321 }, { "epoch": 0.7240144646133566, "grad_norm": 1.3682486622160634, "learning_rate": 3.7357606665646984e-06, "loss": 0.5034, "step": 20322 }, { "epoch": 0.7240500917398507, "grad_norm": 1.4247774855157123, "learning_rate": 3.7348612347476965e-06, "loss": 0.4522, "step": 20323 }, { "epoch": 0.7240857188663449, "grad_norm": 2.6533478553002263, "learning_rate": 3.7339618863553983e-06, "loss": 0.4836, "step": 20324 }, { "epoch": 0.724121345992839, "grad_norm": 1.3079943573996233, "learning_rate": 3.7330626213997755e-06, "loss": 0.4166, "step": 20325 }, { "epoch": 0.724156973119333, "grad_norm": 1.1868635783239119, "learning_rate": 3.7321634398928e-06, "loss": 0.4097, "step": 20326 }, { "epoch": 0.7241926002458272, "grad_norm": 1.4170144925049701, "learning_rate": 3.731264341846457e-06, "loss": 0.5455, "step": 20327 }, { "epoch": 0.7242282273723213, "grad_norm": 1.268627348903494, "learning_rate": 3.7303653272727057e-06, "loss": 0.4086, "step": 20328 }, { "epoch": 0.7242638544988154, "grad_norm": 1.6184071189917004, "learning_rate": 3.7294663961835263e-06, "loss": 0.5016, "step": 20329 }, { "epoch": 0.7242994816253095, "grad_norm": 1.3054484401525073, "learning_rate": 3.7285675485908823e-06, "loss": 0.5049, "step": 20330 }, { "epoch": 0.7243351087518036, "grad_norm": 1.6679087523605856, "learning_rate": 3.7276687845067474e-06, "loss": 0.4711, "step": 20331 }, { "epoch": 0.7243707358782977, "grad_norm": 1.2599650053216875, "learning_rate": 3.7267701039430836e-06, "loss": 0.4806, "step": 20332 }, { "epoch": 0.7244063630047919, "grad_norm": 1.1756026117831122, "learning_rate": 3.72587150691186e-06, "loss": 0.3923, "step": 20333 }, { "epoch": 0.724441990131286, "grad_norm": 1.9042162081483838, "learning_rate": 3.7249729934250433e-06, "loss": 0.5903, "step": 20334 }, { "epoch": 0.72447761725778, "grad_norm": 1.1745862919131107, "learning_rate": 3.7240745634946005e-06, "loss": 0.4074, "step": 20335 }, { "epoch": 0.7245132443842742, "grad_norm": 1.0594810621380977, "learning_rate": 3.723176217132488e-06, "loss": 0.4266, "step": 20336 }, { "epoch": 0.7245488715107683, "grad_norm": 1.1479596021989877, "learning_rate": 3.7222779543506714e-06, "loss": 0.4384, "step": 20337 }, { "epoch": 0.7245844986372624, "grad_norm": 1.2694896344680342, "learning_rate": 3.721379775161116e-06, "loss": 0.4106, "step": 20338 }, { "epoch": 0.7246201257637566, "grad_norm": 1.2904631269178304, "learning_rate": 3.7204816795757727e-06, "loss": 0.4142, "step": 20339 }, { "epoch": 0.7246557528902506, "grad_norm": 1.9454404150667146, "learning_rate": 3.7195836676066057e-06, "loss": 0.535, "step": 20340 }, { "epoch": 0.7246913800167447, "grad_norm": 1.4390493306736616, "learning_rate": 3.7186857392655708e-06, "loss": 0.4557, "step": 20341 }, { "epoch": 0.7247270071432389, "grad_norm": 1.2542009708809116, "learning_rate": 3.7177878945646306e-06, "loss": 0.3601, "step": 20342 }, { "epoch": 0.724762634269733, "grad_norm": 1.4265527757528804, "learning_rate": 3.7168901335157313e-06, "loss": 0.5012, "step": 20343 }, { "epoch": 0.724798261396227, "grad_norm": 1.3694677663435044, "learning_rate": 3.7159924561308346e-06, "loss": 0.4474, "step": 20344 }, { "epoch": 0.7248338885227212, "grad_norm": 1.2220241511829133, "learning_rate": 3.7150948624218876e-06, "loss": 0.3812, "step": 20345 }, { "epoch": 0.7248695156492153, "grad_norm": 1.390392184206241, "learning_rate": 3.714197352400849e-06, "loss": 0.4192, "step": 20346 }, { "epoch": 0.7249051427757094, "grad_norm": 1.1528201442664974, "learning_rate": 3.713299926079663e-06, "loss": 0.4551, "step": 20347 }, { "epoch": 0.7249407699022036, "grad_norm": 1.14178200145387, "learning_rate": 3.7124025834702816e-06, "loss": 0.3976, "step": 20348 }, { "epoch": 0.7249763970286976, "grad_norm": 2.5655080163921378, "learning_rate": 3.7115053245846554e-06, "loss": 0.4972, "step": 20349 }, { "epoch": 0.7250120241551917, "grad_norm": 1.6574843424378174, "learning_rate": 3.710608149434731e-06, "loss": 0.5344, "step": 20350 }, { "epoch": 0.7250476512816859, "grad_norm": 1.3059315195294403, "learning_rate": 3.709711058032459e-06, "loss": 0.4093, "step": 20351 }, { "epoch": 0.72508327840818, "grad_norm": 1.502942040518161, "learning_rate": 3.708814050389776e-06, "loss": 0.5022, "step": 20352 }, { "epoch": 0.7251189055346741, "grad_norm": 1.803384691192281, "learning_rate": 3.7079171265186363e-06, "loss": 0.4684, "step": 20353 }, { "epoch": 0.7251545326611682, "grad_norm": 1.1589650965380425, "learning_rate": 3.7070202864309733e-06, "loss": 0.4825, "step": 20354 }, { "epoch": 0.7251901597876623, "grad_norm": 1.5976366540208689, "learning_rate": 3.7061235301387342e-06, "loss": 0.5457, "step": 20355 }, { "epoch": 0.7252257869141564, "grad_norm": 1.854790094814895, "learning_rate": 3.70522685765386e-06, "loss": 0.5022, "step": 20356 }, { "epoch": 0.7252614140406506, "grad_norm": 1.2275230692840835, "learning_rate": 3.704330268988293e-06, "loss": 0.4045, "step": 20357 }, { "epoch": 0.7252970411671447, "grad_norm": 1.2167917693288748, "learning_rate": 3.7034337641539652e-06, "loss": 0.4263, "step": 20358 }, { "epoch": 0.7253326682936387, "grad_norm": 2.507395039656096, "learning_rate": 3.7025373431628185e-06, "loss": 0.3933, "step": 20359 }, { "epoch": 0.7253682954201329, "grad_norm": 1.3895018552759533, "learning_rate": 3.7016410060267914e-06, "loss": 0.5906, "step": 20360 }, { "epoch": 0.725403922546627, "grad_norm": 1.655183587520659, "learning_rate": 3.7007447527578134e-06, "loss": 0.4736, "step": 20361 }, { "epoch": 0.7254395496731211, "grad_norm": 1.3450583606035726, "learning_rate": 3.6998485833678256e-06, "loss": 0.5412, "step": 20362 }, { "epoch": 0.7254751767996153, "grad_norm": 1.5097292871086856, "learning_rate": 3.698952497868751e-06, "loss": 0.5793, "step": 20363 }, { "epoch": 0.7255108039261093, "grad_norm": 1.2298945787790656, "learning_rate": 3.698056496272534e-06, "loss": 0.5653, "step": 20364 }, { "epoch": 0.7255464310526034, "grad_norm": 1.162583929390708, "learning_rate": 3.6971605785910967e-06, "loss": 0.4332, "step": 20365 }, { "epoch": 0.7255820581790976, "grad_norm": 1.4224901963441163, "learning_rate": 3.6962647448363753e-06, "loss": 0.4148, "step": 20366 }, { "epoch": 0.7256176853055917, "grad_norm": 1.2573730522580773, "learning_rate": 3.6953689950202908e-06, "loss": 0.4285, "step": 20367 }, { "epoch": 0.7256533124320858, "grad_norm": 1.2367841571023848, "learning_rate": 3.6944733291547784e-06, "loss": 0.3689, "step": 20368 }, { "epoch": 0.7256889395585799, "grad_norm": 1.745151311998869, "learning_rate": 3.6935777472517576e-06, "loss": 0.4786, "step": 20369 }, { "epoch": 0.725724566685074, "grad_norm": 1.5082836559698751, "learning_rate": 3.692682249323157e-06, "loss": 0.5065, "step": 20370 }, { "epoch": 0.7257601938115681, "grad_norm": 1.4977403067535389, "learning_rate": 3.6917868353809007e-06, "loss": 0.4838, "step": 20371 }, { "epoch": 0.7257958209380623, "grad_norm": 1.8637537399370487, "learning_rate": 3.690891505436912e-06, "loss": 0.5031, "step": 20372 }, { "epoch": 0.7258314480645564, "grad_norm": 1.4403834116736438, "learning_rate": 3.689996259503116e-06, "loss": 0.466, "step": 20373 }, { "epoch": 0.7258670751910504, "grad_norm": 1.6412960960599396, "learning_rate": 3.6891010975914264e-06, "loss": 0.5489, "step": 20374 }, { "epoch": 0.7259027023175446, "grad_norm": 1.4369506298130514, "learning_rate": 3.6882060197137714e-06, "loss": 0.5119, "step": 20375 }, { "epoch": 0.7259383294440387, "grad_norm": 1.544176178860966, "learning_rate": 3.687311025882061e-06, "loss": 0.5487, "step": 20376 }, { "epoch": 0.7259739565705328, "grad_norm": 1.2589807254930467, "learning_rate": 3.6864161161082156e-06, "loss": 0.4044, "step": 20377 }, { "epoch": 0.726009583697027, "grad_norm": 1.2349668363818518, "learning_rate": 3.6855212904041538e-06, "loss": 0.4324, "step": 20378 }, { "epoch": 0.726045210823521, "grad_norm": 1.677292795860415, "learning_rate": 3.6846265487817924e-06, "loss": 0.5387, "step": 20379 }, { "epoch": 0.7260808379500151, "grad_norm": 1.4212627822397328, "learning_rate": 3.6837318912530394e-06, "loss": 0.4891, "step": 20380 }, { "epoch": 0.7261164650765093, "grad_norm": 1.3493619083185586, "learning_rate": 3.6828373178298147e-06, "loss": 0.4683, "step": 20381 }, { "epoch": 0.7261520922030034, "grad_norm": 1.1993856961354987, "learning_rate": 3.6819428285240235e-06, "loss": 0.4517, "step": 20382 }, { "epoch": 0.7261877193294974, "grad_norm": 1.5179412884907773, "learning_rate": 3.6810484233475787e-06, "loss": 0.4893, "step": 20383 }, { "epoch": 0.7262233464559916, "grad_norm": 1.1257033278554922, "learning_rate": 3.6801541023123943e-06, "loss": 0.3418, "step": 20384 }, { "epoch": 0.7262589735824857, "grad_norm": 1.2739441269152572, "learning_rate": 3.6792598654303724e-06, "loss": 0.4549, "step": 20385 }, { "epoch": 0.7262946007089798, "grad_norm": 1.5348398785363755, "learning_rate": 3.678365712713423e-06, "loss": 0.5651, "step": 20386 }, { "epoch": 0.726330227835474, "grad_norm": 1.3175791352047235, "learning_rate": 3.6774716441734524e-06, "loss": 0.3707, "step": 20387 }, { "epoch": 0.726365854961968, "grad_norm": 1.2497587362366334, "learning_rate": 3.67657765982237e-06, "loss": 0.4057, "step": 20388 }, { "epoch": 0.7264014820884621, "grad_norm": 1.8943041107538412, "learning_rate": 3.675683759672073e-06, "loss": 0.5021, "step": 20389 }, { "epoch": 0.7264371092149563, "grad_norm": 1.386061023703761, "learning_rate": 3.674789943734469e-06, "loss": 0.4985, "step": 20390 }, { "epoch": 0.7264727363414504, "grad_norm": 1.3723384812181703, "learning_rate": 3.673896212021456e-06, "loss": 0.4365, "step": 20391 }, { "epoch": 0.7265083634679445, "grad_norm": 1.5427407122851982, "learning_rate": 3.6730025645449363e-06, "loss": 0.5339, "step": 20392 }, { "epoch": 0.7265439905944386, "grad_norm": 1.2864686868189066, "learning_rate": 3.6721090013168093e-06, "loss": 0.4294, "step": 20393 }, { "epoch": 0.7265796177209327, "grad_norm": 1.6329125772668616, "learning_rate": 3.6712155223489745e-06, "loss": 0.5273, "step": 20394 }, { "epoch": 0.7266152448474268, "grad_norm": 1.4770963360062521, "learning_rate": 3.6703221276533306e-06, "loss": 0.5552, "step": 20395 }, { "epoch": 0.726650871973921, "grad_norm": 1.1004908340867006, "learning_rate": 3.66942881724177e-06, "loss": 0.4021, "step": 20396 }, { "epoch": 0.726686499100415, "grad_norm": 1.7412783183249136, "learning_rate": 3.668535591126192e-06, "loss": 0.5072, "step": 20397 }, { "epoch": 0.7267221262269091, "grad_norm": 1.404931287633241, "learning_rate": 3.6676424493184848e-06, "loss": 0.4465, "step": 20398 }, { "epoch": 0.7267577533534033, "grad_norm": 1.5102988877207213, "learning_rate": 3.6667493918305485e-06, "loss": 0.5242, "step": 20399 }, { "epoch": 0.7267933804798974, "grad_norm": 1.6851466861967515, "learning_rate": 3.665856418674266e-06, "loss": 0.5373, "step": 20400 }, { "epoch": 0.7268290076063915, "grad_norm": 1.229774266019771, "learning_rate": 3.664963529861534e-06, "loss": 0.3457, "step": 20401 }, { "epoch": 0.7268646347328857, "grad_norm": 1.4838100374445031, "learning_rate": 3.6640707254042395e-06, "loss": 0.4805, "step": 20402 }, { "epoch": 0.7269002618593797, "grad_norm": 1.2879887120283773, "learning_rate": 3.663178005314275e-06, "loss": 0.3978, "step": 20403 }, { "epoch": 0.7269358889858738, "grad_norm": 1.0573864115070843, "learning_rate": 3.6622853696035222e-06, "loss": 0.3797, "step": 20404 }, { "epoch": 0.726971516112368, "grad_norm": 1.3617055882969276, "learning_rate": 3.661392818283869e-06, "loss": 0.3948, "step": 20405 }, { "epoch": 0.7270071432388621, "grad_norm": 1.1949751050028081, "learning_rate": 3.6605003513672045e-06, "loss": 0.374, "step": 20406 }, { "epoch": 0.7270427703653561, "grad_norm": 3.0796062718347295, "learning_rate": 3.6596079688654053e-06, "loss": 0.4336, "step": 20407 }, { "epoch": 0.7270783974918503, "grad_norm": 1.3008252763759889, "learning_rate": 3.6587156707903572e-06, "loss": 0.3881, "step": 20408 }, { "epoch": 0.7271140246183444, "grad_norm": 1.1737104719252203, "learning_rate": 3.657823457153943e-06, "loss": 0.382, "step": 20409 }, { "epoch": 0.7271496517448385, "grad_norm": 1.4397400213717706, "learning_rate": 3.656931327968045e-06, "loss": 0.5416, "step": 20410 }, { "epoch": 0.7271852788713327, "grad_norm": 1.4279826432475695, "learning_rate": 3.656039283244537e-06, "loss": 0.376, "step": 20411 }, { "epoch": 0.7272209059978267, "grad_norm": 1.9779538242102401, "learning_rate": 3.655147322995304e-06, "loss": 0.5537, "step": 20412 }, { "epoch": 0.7272565331243208, "grad_norm": 1.5985989966128276, "learning_rate": 3.6542554472322147e-06, "loss": 0.5232, "step": 20413 }, { "epoch": 0.727292160250815, "grad_norm": 1.4186339429684718, "learning_rate": 3.6533636559671537e-06, "loss": 0.4153, "step": 20414 }, { "epoch": 0.7273277873773091, "grad_norm": 1.3723957820210628, "learning_rate": 3.6524719492119844e-06, "loss": 0.4065, "step": 20415 }, { "epoch": 0.7273634145038032, "grad_norm": 1.2904391308639078, "learning_rate": 3.6515803269785944e-06, "loss": 0.3864, "step": 20416 }, { "epoch": 0.7273990416302973, "grad_norm": 1.6346142471935112, "learning_rate": 3.6506887892788466e-06, "loss": 0.5302, "step": 20417 }, { "epoch": 0.7274346687567914, "grad_norm": 1.2861851955166586, "learning_rate": 3.6497973361246153e-06, "loss": 0.4334, "step": 20418 }, { "epoch": 0.7274702958832855, "grad_norm": 1.1759425057847552, "learning_rate": 3.648905967527775e-06, "loss": 0.3989, "step": 20419 }, { "epoch": 0.7275059230097797, "grad_norm": 1.1745811408238545, "learning_rate": 3.6480146835001874e-06, "loss": 0.4018, "step": 20420 }, { "epoch": 0.7275415501362738, "grad_norm": 1.1875644706840824, "learning_rate": 3.647123484053727e-06, "loss": 0.378, "step": 20421 }, { "epoch": 0.7275771772627678, "grad_norm": 1.2141745680946836, "learning_rate": 3.646232369200254e-06, "loss": 0.4043, "step": 20422 }, { "epoch": 0.727612804389262, "grad_norm": 1.214070966289432, "learning_rate": 3.645341338951639e-06, "loss": 0.3888, "step": 20423 }, { "epoch": 0.7276484315157561, "grad_norm": 1.4810147983952695, "learning_rate": 3.644450393319745e-06, "loss": 0.4719, "step": 20424 }, { "epoch": 0.7276840586422502, "grad_norm": 1.4111214407758126, "learning_rate": 3.6435595323164395e-06, "loss": 0.457, "step": 20425 }, { "epoch": 0.7277196857687444, "grad_norm": 1.8545029287348553, "learning_rate": 3.642668755953579e-06, "loss": 0.5686, "step": 20426 }, { "epoch": 0.7277553128952384, "grad_norm": 1.183485311560025, "learning_rate": 3.641778064243031e-06, "loss": 0.3978, "step": 20427 }, { "epoch": 0.7277909400217325, "grad_norm": 1.66858488823345, "learning_rate": 3.640887457196648e-06, "loss": 0.4388, "step": 20428 }, { "epoch": 0.7278265671482267, "grad_norm": 1.5526473926006612, "learning_rate": 3.639996934826294e-06, "loss": 0.5002, "step": 20429 }, { "epoch": 0.7278621942747208, "grad_norm": 1.3344651689829055, "learning_rate": 3.639106497143826e-06, "loss": 0.4051, "step": 20430 }, { "epoch": 0.7278978214012148, "grad_norm": 1.2747096658194925, "learning_rate": 3.638216144161101e-06, "loss": 0.4463, "step": 20431 }, { "epoch": 0.727933448527709, "grad_norm": 1.140929369842831, "learning_rate": 3.6373258758899778e-06, "loss": 0.4178, "step": 20432 }, { "epoch": 0.7279690756542031, "grad_norm": 1.3107679112829957, "learning_rate": 3.636435692342305e-06, "loss": 0.4593, "step": 20433 }, { "epoch": 0.7280047027806972, "grad_norm": 1.3366478334898688, "learning_rate": 3.635545593529941e-06, "loss": 0.4878, "step": 20434 }, { "epoch": 0.7280403299071914, "grad_norm": 1.2352115484589323, "learning_rate": 3.634655579464733e-06, "loss": 0.4114, "step": 20435 }, { "epoch": 0.7280759570336854, "grad_norm": 1.924211624280252, "learning_rate": 3.6337656501585395e-06, "loss": 0.5652, "step": 20436 }, { "epoch": 0.7281115841601795, "grad_norm": 1.3777269584282137, "learning_rate": 3.6328758056232017e-06, "loss": 0.4378, "step": 20437 }, { "epoch": 0.7281472112866737, "grad_norm": 1.4777157854974412, "learning_rate": 3.6319860458705735e-06, "loss": 0.561, "step": 20438 }, { "epoch": 0.7281828384131678, "grad_norm": 1.4257203657400441, "learning_rate": 3.6310963709125014e-06, "loss": 0.4077, "step": 20439 }, { "epoch": 0.7282184655396619, "grad_norm": 1.3750147297599695, "learning_rate": 3.6302067807608334e-06, "loss": 0.4969, "step": 20440 }, { "epoch": 0.728254092666156, "grad_norm": 1.5481125086392677, "learning_rate": 3.6293172754274177e-06, "loss": 0.5588, "step": 20441 }, { "epoch": 0.7282897197926501, "grad_norm": 4.185468450936361, "learning_rate": 3.6284278549240912e-06, "loss": 0.4477, "step": 20442 }, { "epoch": 0.7283253469191442, "grad_norm": 1.2215092625672759, "learning_rate": 3.6275385192627056e-06, "loss": 0.4303, "step": 20443 }, { "epoch": 0.7283609740456384, "grad_norm": 1.4267464624657145, "learning_rate": 3.626649268455095e-06, "loss": 0.5077, "step": 20444 }, { "epoch": 0.7283966011721325, "grad_norm": 1.7498918243415202, "learning_rate": 3.625760102513103e-06, "loss": 0.5407, "step": 20445 }, { "epoch": 0.7284322282986265, "grad_norm": 1.6043610554706356, "learning_rate": 3.6248710214485717e-06, "loss": 0.4413, "step": 20446 }, { "epoch": 0.7284678554251207, "grad_norm": 1.262985407118442, "learning_rate": 3.6239820252733416e-06, "loss": 0.3876, "step": 20447 }, { "epoch": 0.7285034825516148, "grad_norm": 1.2324108121580906, "learning_rate": 3.6230931139992443e-06, "loss": 0.3819, "step": 20448 }, { "epoch": 0.7285391096781089, "grad_norm": 1.3787325598469888, "learning_rate": 3.6222042876381227e-06, "loss": 0.5047, "step": 20449 }, { "epoch": 0.728574736804603, "grad_norm": 1.417094390431845, "learning_rate": 3.621315546201806e-06, "loss": 0.4683, "step": 20450 }, { "epoch": 0.7286103639310971, "grad_norm": 1.8383705304663052, "learning_rate": 3.6204268897021333e-06, "loss": 0.5745, "step": 20451 }, { "epoch": 0.7286459910575912, "grad_norm": 1.1599137570926916, "learning_rate": 3.6195383181509304e-06, "loss": 0.4098, "step": 20452 }, { "epoch": 0.7286816181840854, "grad_norm": 1.311029807760034, "learning_rate": 3.618649831560037e-06, "loss": 0.4466, "step": 20453 }, { "epoch": 0.7287172453105795, "grad_norm": 1.1438803601135092, "learning_rate": 3.6177614299412845e-06, "loss": 0.4124, "step": 20454 }, { "epoch": 0.7287528724370735, "grad_norm": 1.3280742477476317, "learning_rate": 3.616873113306496e-06, "loss": 0.433, "step": 20455 }, { "epoch": 0.7287884995635677, "grad_norm": 1.171315802455937, "learning_rate": 3.615984881667508e-06, "loss": 0.4036, "step": 20456 }, { "epoch": 0.7288241266900618, "grad_norm": 1.4312189898372158, "learning_rate": 3.61509673503614e-06, "loss": 0.5088, "step": 20457 }, { "epoch": 0.7288597538165559, "grad_norm": 1.613260474044887, "learning_rate": 3.6142086734242242e-06, "loss": 0.4963, "step": 20458 }, { "epoch": 0.7288953809430501, "grad_norm": 1.5020343166657442, "learning_rate": 3.6133206968435795e-06, "loss": 0.6079, "step": 20459 }, { "epoch": 0.7289310080695441, "grad_norm": 1.44216841059357, "learning_rate": 3.6124328053060354e-06, "loss": 0.6325, "step": 20460 }, { "epoch": 0.7289666351960382, "grad_norm": 1.7338457045729265, "learning_rate": 3.6115449988234118e-06, "loss": 0.5098, "step": 20461 }, { "epoch": 0.7290022623225324, "grad_norm": 2.026779466818275, "learning_rate": 3.6106572774075354e-06, "loss": 0.5352, "step": 20462 }, { "epoch": 0.7290378894490265, "grad_norm": 1.4917092198358564, "learning_rate": 3.609769641070221e-06, "loss": 0.4018, "step": 20463 }, { "epoch": 0.7290735165755206, "grad_norm": 1.4343090516523385, "learning_rate": 3.608882089823289e-06, "loss": 0.5003, "step": 20464 }, { "epoch": 0.7291091437020147, "grad_norm": 1.0567432634108271, "learning_rate": 3.6079946236785633e-06, "loss": 0.3944, "step": 20465 }, { "epoch": 0.7291447708285088, "grad_norm": 1.230612350026262, "learning_rate": 3.6071072426478527e-06, "loss": 0.4083, "step": 20466 }, { "epoch": 0.7291803979550029, "grad_norm": 1.5150378659156574, "learning_rate": 3.6062199467429784e-06, "loss": 0.5448, "step": 20467 }, { "epoch": 0.7292160250814971, "grad_norm": 1.1587805452424043, "learning_rate": 3.605332735975754e-06, "loss": 0.3851, "step": 20468 }, { "epoch": 0.7292516522079912, "grad_norm": 1.498749534705231, "learning_rate": 3.604445610357997e-06, "loss": 0.4914, "step": 20469 }, { "epoch": 0.7292872793344852, "grad_norm": 1.1333186003135904, "learning_rate": 3.603558569901514e-06, "loss": 0.392, "step": 20470 }, { "epoch": 0.7293229064609794, "grad_norm": 1.1671316934185767, "learning_rate": 3.6026716146181227e-06, "loss": 0.468, "step": 20471 }, { "epoch": 0.7293585335874735, "grad_norm": 1.5601961217523332, "learning_rate": 3.6017847445196273e-06, "loss": 0.5237, "step": 20472 }, { "epoch": 0.7293941607139676, "grad_norm": 1.7478196981441425, "learning_rate": 3.600897959617843e-06, "loss": 0.5375, "step": 20473 }, { "epoch": 0.7294297878404618, "grad_norm": 1.4903806441992158, "learning_rate": 3.6000112599245717e-06, "loss": 0.5449, "step": 20474 }, { "epoch": 0.7294654149669558, "grad_norm": 1.358889717615623, "learning_rate": 3.599124645451624e-06, "loss": 0.5325, "step": 20475 }, { "epoch": 0.7295010420934499, "grad_norm": 8.338163856694225, "learning_rate": 3.598238116210806e-06, "loss": 0.5091, "step": 20476 }, { "epoch": 0.7295366692199441, "grad_norm": 1.346638669761694, "learning_rate": 3.5973516722139212e-06, "loss": 0.475, "step": 20477 }, { "epoch": 0.7295722963464382, "grad_norm": 1.587039040154296, "learning_rate": 3.596465313472778e-06, "loss": 0.4836, "step": 20478 }, { "epoch": 0.7296079234729322, "grad_norm": 2.9631834342352668, "learning_rate": 3.5955790399991717e-06, "loss": 0.5747, "step": 20479 }, { "epoch": 0.7296435505994264, "grad_norm": 1.7264541535215137, "learning_rate": 3.5946928518049097e-06, "loss": 0.5743, "step": 20480 }, { "epoch": 0.7296791777259205, "grad_norm": 1.3115382122388222, "learning_rate": 3.5938067489017857e-06, "loss": 0.4584, "step": 20481 }, { "epoch": 0.7297148048524146, "grad_norm": 1.1571850818897307, "learning_rate": 3.5929207313016033e-06, "loss": 0.404, "step": 20482 }, { "epoch": 0.7297504319789088, "grad_norm": 1.200004237140403, "learning_rate": 3.59203479901616e-06, "loss": 0.3846, "step": 20483 }, { "epoch": 0.7297860591054028, "grad_norm": 1.4465138147583372, "learning_rate": 3.591148952057255e-06, "loss": 0.4515, "step": 20484 }, { "epoch": 0.7298216862318969, "grad_norm": 1.5322821277915641, "learning_rate": 3.590263190436678e-06, "loss": 0.4986, "step": 20485 }, { "epoch": 0.7298573133583911, "grad_norm": 1.6529928958888072, "learning_rate": 3.5893775141662267e-06, "loss": 0.6087, "step": 20486 }, { "epoch": 0.7298929404848852, "grad_norm": 1.286048147320866, "learning_rate": 3.5884919232576976e-06, "loss": 0.4399, "step": 20487 }, { "epoch": 0.7299285676113793, "grad_norm": 1.3081454206186345, "learning_rate": 3.5876064177228775e-06, "loss": 0.4489, "step": 20488 }, { "epoch": 0.7299641947378734, "grad_norm": 1.2395521147441284, "learning_rate": 3.586720997573563e-06, "loss": 0.4531, "step": 20489 }, { "epoch": 0.7299998218643675, "grad_norm": 1.348316324285519, "learning_rate": 3.585835662821534e-06, "loss": 0.4715, "step": 20490 }, { "epoch": 0.7300354489908616, "grad_norm": 1.1769790629723615, "learning_rate": 3.5849504134785927e-06, "loss": 0.433, "step": 20491 }, { "epoch": 0.7300710761173558, "grad_norm": 1.3031462931174895, "learning_rate": 3.5840652495565178e-06, "loss": 0.4381, "step": 20492 }, { "epoch": 0.7301067032438499, "grad_norm": 1.4547843587520206, "learning_rate": 3.583180171067101e-06, "loss": 0.5321, "step": 20493 }, { "epoch": 0.7301423303703439, "grad_norm": 1.1845821606703384, "learning_rate": 3.582295178022123e-06, "loss": 0.4064, "step": 20494 }, { "epoch": 0.7301779574968381, "grad_norm": 1.2529200163354983, "learning_rate": 3.581410270433373e-06, "loss": 0.425, "step": 20495 }, { "epoch": 0.7302135846233322, "grad_norm": 1.2356623470372894, "learning_rate": 3.5805254483126295e-06, "loss": 0.3758, "step": 20496 }, { "epoch": 0.7302492117498263, "grad_norm": 1.1872559384484482, "learning_rate": 3.579640711671676e-06, "loss": 0.3794, "step": 20497 }, { "epoch": 0.7302848388763205, "grad_norm": 1.1634824672510848, "learning_rate": 3.5787560605222926e-06, "loss": 0.3476, "step": 20498 }, { "epoch": 0.7303204660028145, "grad_norm": 1.5998419586698513, "learning_rate": 3.5778714948762617e-06, "loss": 0.5688, "step": 20499 }, { "epoch": 0.7303560931293086, "grad_norm": 1.0851545620361316, "learning_rate": 3.5769870147453643e-06, "loss": 0.3665, "step": 20500 }, { "epoch": 0.7303917202558028, "grad_norm": 1.2846761377429892, "learning_rate": 3.5761026201413708e-06, "loss": 0.4481, "step": 20501 }, { "epoch": 0.7304273473822969, "grad_norm": 1.4000289738222376, "learning_rate": 3.575218311076063e-06, "loss": 0.5272, "step": 20502 }, { "epoch": 0.7304629745087909, "grad_norm": 1.2136104523946885, "learning_rate": 3.574334087561212e-06, "loss": 0.4025, "step": 20503 }, { "epoch": 0.7304986016352851, "grad_norm": 1.6384363634920618, "learning_rate": 3.573449949608594e-06, "loss": 0.5562, "step": 20504 }, { "epoch": 0.7305342287617792, "grad_norm": 1.7920647603387732, "learning_rate": 3.57256589722998e-06, "loss": 0.3911, "step": 20505 }, { "epoch": 0.7305698558882733, "grad_norm": 1.1588734956828428, "learning_rate": 3.5716819304371485e-06, "loss": 0.382, "step": 20506 }, { "epoch": 0.7306054830147675, "grad_norm": 1.8163630343375898, "learning_rate": 3.5707980492418624e-06, "loss": 0.4537, "step": 20507 }, { "epoch": 0.7306411101412615, "grad_norm": 1.3400986416313863, "learning_rate": 3.569914253655896e-06, "loss": 0.4077, "step": 20508 }, { "epoch": 0.7306767372677556, "grad_norm": 1.356476751626817, "learning_rate": 3.5690305436910133e-06, "loss": 0.4161, "step": 20509 }, { "epoch": 0.7307123643942498, "grad_norm": 1.5875401703251377, "learning_rate": 3.5681469193589834e-06, "loss": 0.4746, "step": 20510 }, { "epoch": 0.7307479915207439, "grad_norm": 1.5614501464633552, "learning_rate": 3.5672633806715763e-06, "loss": 0.5222, "step": 20511 }, { "epoch": 0.730783618647238, "grad_norm": 1.3537121672204309, "learning_rate": 3.5663799276405496e-06, "loss": 0.4238, "step": 20512 }, { "epoch": 0.7308192457737321, "grad_norm": 1.6302371496756236, "learning_rate": 3.5654965602776724e-06, "loss": 0.5225, "step": 20513 }, { "epoch": 0.7308548729002262, "grad_norm": 1.711621588904219, "learning_rate": 3.564613278594705e-06, "loss": 0.5282, "step": 20514 }, { "epoch": 0.7308905000267203, "grad_norm": 1.6835976717255958, "learning_rate": 3.5637300826034126e-06, "loss": 0.5612, "step": 20515 }, { "epoch": 0.7309261271532145, "grad_norm": 1.3084102904986206, "learning_rate": 3.5628469723155513e-06, "loss": 0.445, "step": 20516 }, { "epoch": 0.7309617542797086, "grad_norm": 1.2688775385546458, "learning_rate": 3.5619639477428836e-06, "loss": 0.4036, "step": 20517 }, { "epoch": 0.7309973814062026, "grad_norm": 1.2105875682209342, "learning_rate": 3.5610810088971627e-06, "loss": 0.4056, "step": 20518 }, { "epoch": 0.7310330085326968, "grad_norm": 1.5654030278164957, "learning_rate": 3.5601981557901488e-06, "loss": 0.5076, "step": 20519 }, { "epoch": 0.7310686356591909, "grad_norm": 1.1751259913011933, "learning_rate": 3.5593153884335986e-06, "loss": 0.4471, "step": 20520 }, { "epoch": 0.731104262785685, "grad_norm": 1.6332071858598072, "learning_rate": 3.5584327068392643e-06, "loss": 0.5508, "step": 20521 }, { "epoch": 0.7311398899121792, "grad_norm": 1.1967371317743254, "learning_rate": 3.557550111018906e-06, "loss": 0.43, "step": 20522 }, { "epoch": 0.7311755170386732, "grad_norm": 1.4765677761247207, "learning_rate": 3.556667600984266e-06, "loss": 0.5089, "step": 20523 }, { "epoch": 0.7312111441651673, "grad_norm": 1.1465196809233458, "learning_rate": 3.555785176747104e-06, "loss": 0.3976, "step": 20524 }, { "epoch": 0.7312467712916615, "grad_norm": 1.3092904929131604, "learning_rate": 3.554902838319164e-06, "loss": 0.4941, "step": 20525 }, { "epoch": 0.7312823984181556, "grad_norm": 1.3461024412067828, "learning_rate": 3.554020585712201e-06, "loss": 0.5239, "step": 20526 }, { "epoch": 0.7313180255446496, "grad_norm": 1.3522059031643587, "learning_rate": 3.5531384189379515e-06, "loss": 0.4843, "step": 20527 }, { "epoch": 0.7313536526711438, "grad_norm": 1.4679775686665293, "learning_rate": 3.5522563380081778e-06, "loss": 0.5784, "step": 20528 }, { "epoch": 0.7313892797976379, "grad_norm": 2.097407911510545, "learning_rate": 3.5513743429346135e-06, "loss": 0.5993, "step": 20529 }, { "epoch": 0.731424906924132, "grad_norm": 1.3840798458595533, "learning_rate": 3.5504924337290113e-06, "loss": 0.4391, "step": 20530 }, { "epoch": 0.7314605340506262, "grad_norm": 1.9963172812118801, "learning_rate": 3.549610610403107e-06, "loss": 0.4968, "step": 20531 }, { "epoch": 0.7314961611771202, "grad_norm": 1.3418606392047916, "learning_rate": 3.5487288729686454e-06, "loss": 0.398, "step": 20532 }, { "epoch": 0.7315317883036143, "grad_norm": 1.2177442481017844, "learning_rate": 3.5478472214373716e-06, "loss": 0.4242, "step": 20533 }, { "epoch": 0.7315674154301085, "grad_norm": 1.8558268906106878, "learning_rate": 3.5469656558210186e-06, "loss": 0.481, "step": 20534 }, { "epoch": 0.7316030425566026, "grad_norm": 1.254422593044476, "learning_rate": 3.5460841761313282e-06, "loss": 0.4193, "step": 20535 }, { "epoch": 0.7316386696830968, "grad_norm": 1.2387676160892818, "learning_rate": 3.5452027823800384e-06, "loss": 0.4224, "step": 20536 }, { "epoch": 0.7316742968095908, "grad_norm": 1.6720764148285538, "learning_rate": 3.544321474578888e-06, "loss": 0.5292, "step": 20537 }, { "epoch": 0.7317099239360849, "grad_norm": 1.4131557233470742, "learning_rate": 3.543440252739606e-06, "loss": 0.5141, "step": 20538 }, { "epoch": 0.7317455510625791, "grad_norm": 1.2573734145512587, "learning_rate": 3.5425591168739336e-06, "loss": 0.4285, "step": 20539 }, { "epoch": 0.7317811781890732, "grad_norm": 1.705117179618804, "learning_rate": 3.541678066993597e-06, "loss": 0.4939, "step": 20540 }, { "epoch": 0.7318168053155673, "grad_norm": 1.3195454034732934, "learning_rate": 3.5407971031103305e-06, "loss": 0.4477, "step": 20541 }, { "epoch": 0.7318524324420614, "grad_norm": 1.6627798541715237, "learning_rate": 3.5399162252358653e-06, "loss": 0.5232, "step": 20542 }, { "epoch": 0.7318880595685555, "grad_norm": 1.5361375250974239, "learning_rate": 3.5390354333819344e-06, "loss": 0.5225, "step": 20543 }, { "epoch": 0.7319236866950496, "grad_norm": 1.5824344093140355, "learning_rate": 3.538154727560259e-06, "loss": 0.4937, "step": 20544 }, { "epoch": 0.7319593138215438, "grad_norm": 1.2859807450799268, "learning_rate": 3.53727410778257e-06, "loss": 0.389, "step": 20545 }, { "epoch": 0.7319949409480379, "grad_norm": 1.447122632749927, "learning_rate": 3.536393574060597e-06, "loss": 0.4737, "step": 20546 }, { "epoch": 0.7320305680745319, "grad_norm": 1.7698755413439182, "learning_rate": 3.5355131264060573e-06, "loss": 0.5923, "step": 20547 }, { "epoch": 0.7320661952010261, "grad_norm": 1.4766939424635548, "learning_rate": 3.5346327648306834e-06, "loss": 0.5878, "step": 20548 }, { "epoch": 0.7321018223275202, "grad_norm": 1.467170409516091, "learning_rate": 3.5337524893461894e-06, "loss": 0.4687, "step": 20549 }, { "epoch": 0.7321374494540143, "grad_norm": 1.3255432672165373, "learning_rate": 3.5328722999642997e-06, "loss": 0.5179, "step": 20550 }, { "epoch": 0.7321730765805085, "grad_norm": 1.4263177276314754, "learning_rate": 3.531992196696736e-06, "loss": 0.4551, "step": 20551 }, { "epoch": 0.7322087037070025, "grad_norm": 1.7473276326950193, "learning_rate": 3.531112179555222e-06, "loss": 0.5914, "step": 20552 }, { "epoch": 0.7322443308334966, "grad_norm": 1.2420614988479337, "learning_rate": 3.530232248551466e-06, "loss": 0.4403, "step": 20553 }, { "epoch": 0.7322799579599908, "grad_norm": 1.4020937303466237, "learning_rate": 3.5293524036971927e-06, "loss": 0.4169, "step": 20554 }, { "epoch": 0.7323155850864849, "grad_norm": 1.5286390545229465, "learning_rate": 3.528472645004112e-06, "loss": 0.4959, "step": 20555 }, { "epoch": 0.7323512122129789, "grad_norm": 1.7212097533185433, "learning_rate": 3.527592972483942e-06, "loss": 0.5063, "step": 20556 }, { "epoch": 0.7323868393394731, "grad_norm": 1.1123514264178933, "learning_rate": 3.5267133861483948e-06, "loss": 0.4259, "step": 20557 }, { "epoch": 0.7324224664659672, "grad_norm": 1.1128197094754464, "learning_rate": 3.5258338860091836e-06, "loss": 0.3809, "step": 20558 }, { "epoch": 0.7324580935924613, "grad_norm": 1.1265766393005685, "learning_rate": 3.5249544720780225e-06, "loss": 0.4273, "step": 20559 }, { "epoch": 0.7324937207189555, "grad_norm": 1.3251298816204917, "learning_rate": 3.5240751443666157e-06, "loss": 0.425, "step": 20560 }, { "epoch": 0.7325293478454495, "grad_norm": 1.3834814540715636, "learning_rate": 3.523195902886677e-06, "loss": 0.4459, "step": 20561 }, { "epoch": 0.7325649749719436, "grad_norm": 1.6325620892573256, "learning_rate": 3.5223167476499098e-06, "loss": 0.3766, "step": 20562 }, { "epoch": 0.7326006020984378, "grad_norm": 1.3570312628520576, "learning_rate": 3.5214376786680248e-06, "loss": 0.5101, "step": 20563 }, { "epoch": 0.7326362292249319, "grad_norm": 1.169283359386423, "learning_rate": 3.5205586959527228e-06, "loss": 0.388, "step": 20564 }, { "epoch": 0.732671856351426, "grad_norm": 1.712263256060726, "learning_rate": 3.5196797995157105e-06, "loss": 0.5084, "step": 20565 }, { "epoch": 0.7327074834779201, "grad_norm": 1.3461022992197087, "learning_rate": 3.5188009893686916e-06, "loss": 0.4523, "step": 20566 }, { "epoch": 0.7327431106044142, "grad_norm": 1.3152689981640653, "learning_rate": 3.517922265523367e-06, "loss": 0.4659, "step": 20567 }, { "epoch": 0.7327787377309083, "grad_norm": 1.1798718036192468, "learning_rate": 3.517043627991441e-06, "loss": 0.3829, "step": 20568 }, { "epoch": 0.7328143648574025, "grad_norm": 1.267475403343176, "learning_rate": 3.5161650767846077e-06, "loss": 0.4431, "step": 20569 }, { "epoch": 0.7328499919838966, "grad_norm": 1.2745976464177782, "learning_rate": 3.5152866119145722e-06, "loss": 0.447, "step": 20570 }, { "epoch": 0.7328856191103906, "grad_norm": 1.74203523768355, "learning_rate": 3.514408233393024e-06, "loss": 0.4755, "step": 20571 }, { "epoch": 0.7329212462368848, "grad_norm": 1.883164663148191, "learning_rate": 3.5135299412316635e-06, "loss": 0.4775, "step": 20572 }, { "epoch": 0.7329568733633789, "grad_norm": 1.2164317692385085, "learning_rate": 3.5126517354421853e-06, "loss": 0.4606, "step": 20573 }, { "epoch": 0.732992500489873, "grad_norm": 1.9618367155036962, "learning_rate": 3.511773616036287e-06, "loss": 0.4571, "step": 20574 }, { "epoch": 0.7330281276163672, "grad_norm": 1.5433804802392417, "learning_rate": 3.5108955830256552e-06, "loss": 0.5377, "step": 20575 }, { "epoch": 0.7330637547428612, "grad_norm": 1.2092101025766122, "learning_rate": 3.5100176364219875e-06, "loss": 0.4316, "step": 20576 }, { "epoch": 0.7330993818693553, "grad_norm": 1.3037743882729502, "learning_rate": 3.509139776236967e-06, "loss": 0.4083, "step": 20577 }, { "epoch": 0.7331350089958495, "grad_norm": 1.5372100638982313, "learning_rate": 3.508262002482289e-06, "loss": 0.5477, "step": 20578 }, { "epoch": 0.7331706361223436, "grad_norm": 1.1790253578732184, "learning_rate": 3.5073843151696387e-06, "loss": 0.3811, "step": 20579 }, { "epoch": 0.7332062632488376, "grad_norm": 1.180530221310967, "learning_rate": 3.5065067143107046e-06, "loss": 0.4566, "step": 20580 }, { "epoch": 0.7332418903753318, "grad_norm": 1.3005716304599224, "learning_rate": 3.5056291999171754e-06, "loss": 0.4369, "step": 20581 }, { "epoch": 0.7332775175018259, "grad_norm": 1.2583233939394352, "learning_rate": 3.5047517720007283e-06, "loss": 0.4503, "step": 20582 }, { "epoch": 0.73331314462832, "grad_norm": 1.3184833633414086, "learning_rate": 3.5038744305730566e-06, "loss": 0.5314, "step": 20583 }, { "epoch": 0.7333487717548142, "grad_norm": 2.1449962831507055, "learning_rate": 3.5029971756458323e-06, "loss": 0.5279, "step": 20584 }, { "epoch": 0.7333843988813082, "grad_norm": 1.5291751163285832, "learning_rate": 3.502120007230747e-06, "loss": 0.581, "step": 20585 }, { "epoch": 0.7334200260078023, "grad_norm": 1.242505449655018, "learning_rate": 3.5012429253394707e-06, "loss": 0.4209, "step": 20586 }, { "epoch": 0.7334556531342965, "grad_norm": 1.2034816464362375, "learning_rate": 3.5003659299836877e-06, "loss": 0.4135, "step": 20587 }, { "epoch": 0.7334912802607906, "grad_norm": 1.372293638677655, "learning_rate": 3.4994890211750754e-06, "loss": 0.485, "step": 20588 }, { "epoch": 0.7335269073872847, "grad_norm": 1.1748276143218377, "learning_rate": 3.498612198925313e-06, "loss": 0.45, "step": 20589 }, { "epoch": 0.7335625345137788, "grad_norm": 1.3734505012639535, "learning_rate": 3.4977354632460715e-06, "loss": 0.5444, "step": 20590 }, { "epoch": 0.7335981616402729, "grad_norm": 1.2744138316712517, "learning_rate": 3.4968588141490255e-06, "loss": 0.4268, "step": 20591 }, { "epoch": 0.733633788766767, "grad_norm": 1.5478964814745004, "learning_rate": 3.4959822516458543e-06, "loss": 0.5431, "step": 20592 }, { "epoch": 0.7336694158932612, "grad_norm": 1.2924004019359243, "learning_rate": 3.495105775748221e-06, "loss": 0.4341, "step": 20593 }, { "epoch": 0.7337050430197553, "grad_norm": 1.4528324310157392, "learning_rate": 3.4942293864678023e-06, "loss": 0.5526, "step": 20594 }, { "epoch": 0.7337406701462493, "grad_norm": 1.5087091749740755, "learning_rate": 3.4933530838162654e-06, "loss": 0.5627, "step": 20595 }, { "epoch": 0.7337762972727435, "grad_norm": 1.3064724362883302, "learning_rate": 3.4924768678052845e-06, "loss": 0.4081, "step": 20596 }, { "epoch": 0.7338119243992376, "grad_norm": 1.5491093534788067, "learning_rate": 3.491600738446519e-06, "loss": 0.4707, "step": 20597 }, { "epoch": 0.7338475515257317, "grad_norm": 1.254898447230612, "learning_rate": 3.4907246957516416e-06, "loss": 0.4301, "step": 20598 }, { "epoch": 0.7338831786522259, "grad_norm": 1.5140512655367449, "learning_rate": 3.489848739732311e-06, "loss": 0.5469, "step": 20599 }, { "epoch": 0.7339188057787199, "grad_norm": 1.6153532307418512, "learning_rate": 3.488972870400199e-06, "loss": 0.4922, "step": 20600 }, { "epoch": 0.733954432905214, "grad_norm": 1.1346606427275003, "learning_rate": 3.4880970877669594e-06, "loss": 0.401, "step": 20601 }, { "epoch": 0.7339900600317082, "grad_norm": 1.3745130628098587, "learning_rate": 3.48722139184426e-06, "loss": 0.4013, "step": 20602 }, { "epoch": 0.7340256871582023, "grad_norm": 1.2053580170169433, "learning_rate": 3.4863457826437597e-06, "loss": 0.3931, "step": 20603 }, { "epoch": 0.7340613142846963, "grad_norm": 1.6566192822384342, "learning_rate": 3.485470260177117e-06, "loss": 0.574, "step": 20604 }, { "epoch": 0.7340969414111905, "grad_norm": 1.1318132398041794, "learning_rate": 3.484594824455996e-06, "loss": 0.4117, "step": 20605 }, { "epoch": 0.7341325685376846, "grad_norm": 1.2026122965138677, "learning_rate": 3.4837194754920446e-06, "loss": 0.3876, "step": 20606 }, { "epoch": 0.7341681956641787, "grad_norm": 1.159743670159072, "learning_rate": 3.4828442132969275e-06, "loss": 0.439, "step": 20607 }, { "epoch": 0.7342038227906729, "grad_norm": 1.36190319970625, "learning_rate": 3.48196903788229e-06, "loss": 0.487, "step": 20608 }, { "epoch": 0.7342394499171669, "grad_norm": 1.4769760082804793, "learning_rate": 3.4810939492597918e-06, "loss": 0.4046, "step": 20609 }, { "epoch": 0.734275077043661, "grad_norm": 1.160855396167829, "learning_rate": 3.480218947441083e-06, "loss": 0.401, "step": 20610 }, { "epoch": 0.7343107041701552, "grad_norm": 1.222023492388433, "learning_rate": 3.4793440324378203e-06, "loss": 0.4058, "step": 20611 }, { "epoch": 0.7343463312966493, "grad_norm": 1.1324701632039782, "learning_rate": 3.4784692042616463e-06, "loss": 0.4324, "step": 20612 }, { "epoch": 0.7343819584231434, "grad_norm": 1.1347323828770974, "learning_rate": 3.477594462924212e-06, "loss": 0.4085, "step": 20613 }, { "epoch": 0.7344175855496375, "grad_norm": 1.1688725214511289, "learning_rate": 3.476719808437171e-06, "loss": 0.359, "step": 20614 }, { "epoch": 0.7344532126761316, "grad_norm": 1.6899342113505256, "learning_rate": 3.475845240812161e-06, "loss": 0.5259, "step": 20615 }, { "epoch": 0.7344888398026257, "grad_norm": 2.1751211133284487, "learning_rate": 3.4749707600608318e-06, "loss": 0.4721, "step": 20616 }, { "epoch": 0.7345244669291199, "grad_norm": 1.1168275771911336, "learning_rate": 3.474096366194828e-06, "loss": 0.4369, "step": 20617 }, { "epoch": 0.734560094055614, "grad_norm": 1.4446296984421312, "learning_rate": 3.473222059225795e-06, "loss": 0.515, "step": 20618 }, { "epoch": 0.734595721182108, "grad_norm": 1.2351325213788074, "learning_rate": 3.47234783916537e-06, "loss": 0.4417, "step": 20619 }, { "epoch": 0.7346313483086022, "grad_norm": 1.899696325495679, "learning_rate": 3.471473706025199e-06, "loss": 0.4765, "step": 20620 }, { "epoch": 0.7346669754350963, "grad_norm": 1.4655642677115668, "learning_rate": 3.4705996598169146e-06, "loss": 0.407, "step": 20621 }, { "epoch": 0.7347026025615904, "grad_norm": 1.4146090531252116, "learning_rate": 3.4697257005521635e-06, "loss": 0.5532, "step": 20622 }, { "epoch": 0.7347382296880846, "grad_norm": 1.2119214102423155, "learning_rate": 3.4688518282425764e-06, "loss": 0.4268, "step": 20623 }, { "epoch": 0.7347738568145786, "grad_norm": 1.4826951265066755, "learning_rate": 3.467978042899791e-06, "loss": 0.5334, "step": 20624 }, { "epoch": 0.7348094839410727, "grad_norm": 1.8420135932029413, "learning_rate": 3.4671043445354446e-06, "loss": 0.5291, "step": 20625 }, { "epoch": 0.7348451110675669, "grad_norm": 1.353469370661183, "learning_rate": 3.4662307331611687e-06, "loss": 0.552, "step": 20626 }, { "epoch": 0.734880738194061, "grad_norm": 1.2665785822413451, "learning_rate": 3.465357208788602e-06, "loss": 0.4133, "step": 20627 }, { "epoch": 0.734916365320555, "grad_norm": 1.82050352231987, "learning_rate": 3.464483771429368e-06, "loss": 0.5173, "step": 20628 }, { "epoch": 0.7349519924470492, "grad_norm": 1.1763394720277718, "learning_rate": 3.4636104210951038e-06, "loss": 0.4396, "step": 20629 }, { "epoch": 0.7349876195735433, "grad_norm": 1.4085623026512504, "learning_rate": 3.4627371577974323e-06, "loss": 0.592, "step": 20630 }, { "epoch": 0.7350232467000374, "grad_norm": 1.2728725260314113, "learning_rate": 3.461863981547985e-06, "loss": 0.4218, "step": 20631 }, { "epoch": 0.7350588738265316, "grad_norm": 1.2285755575348034, "learning_rate": 3.460990892358388e-06, "loss": 0.4519, "step": 20632 }, { "epoch": 0.7350945009530256, "grad_norm": 1.3474146834521048, "learning_rate": 3.460117890240271e-06, "loss": 0.4027, "step": 20633 }, { "epoch": 0.7351301280795197, "grad_norm": 1.3549902025548115, "learning_rate": 3.4592449752052526e-06, "loss": 0.5318, "step": 20634 }, { "epoch": 0.7351657552060139, "grad_norm": 1.1581738055829134, "learning_rate": 3.458372147264962e-06, "loss": 0.4107, "step": 20635 }, { "epoch": 0.735201382332508, "grad_norm": 1.5191792694501944, "learning_rate": 3.457499406431016e-06, "loss": 0.5067, "step": 20636 }, { "epoch": 0.735237009459002, "grad_norm": 1.5505428061298727, "learning_rate": 3.4566267527150387e-06, "loss": 0.5762, "step": 20637 }, { "epoch": 0.7352726365854962, "grad_norm": 1.237279124474846, "learning_rate": 3.4557541861286512e-06, "loss": 0.381, "step": 20638 }, { "epoch": 0.7353082637119903, "grad_norm": 1.0514029004901326, "learning_rate": 3.4548817066834693e-06, "loss": 0.4194, "step": 20639 }, { "epoch": 0.7353438908384844, "grad_norm": 1.3745127653960914, "learning_rate": 3.454009314391111e-06, "loss": 0.4826, "step": 20640 }, { "epoch": 0.7353795179649786, "grad_norm": 1.5232429423841365, "learning_rate": 3.4531370092631955e-06, "loss": 0.5275, "step": 20641 }, { "epoch": 0.7354151450914727, "grad_norm": 1.406454203627445, "learning_rate": 3.4522647913113382e-06, "loss": 0.5121, "step": 20642 }, { "epoch": 0.7354507722179667, "grad_norm": 1.1586474772603814, "learning_rate": 3.4513926605471504e-06, "loss": 0.4152, "step": 20643 }, { "epoch": 0.7354863993444609, "grad_norm": 1.458039758058009, "learning_rate": 3.450520616982248e-06, "loss": 0.4794, "step": 20644 }, { "epoch": 0.735522026470955, "grad_norm": 1.2317166498764154, "learning_rate": 3.44964866062824e-06, "loss": 0.4216, "step": 20645 }, { "epoch": 0.7355576535974491, "grad_norm": 1.0952287102761766, "learning_rate": 3.4487767914967375e-06, "loss": 0.3879, "step": 20646 }, { "epoch": 0.7355932807239433, "grad_norm": 1.5866849361368525, "learning_rate": 3.4479050095993507e-06, "loss": 0.5445, "step": 20647 }, { "epoch": 0.7356289078504373, "grad_norm": 1.1882753528664118, "learning_rate": 3.4470333149476885e-06, "loss": 0.3928, "step": 20648 }, { "epoch": 0.7356645349769314, "grad_norm": 1.533699826334013, "learning_rate": 3.4461617075533614e-06, "loss": 0.4015, "step": 20649 }, { "epoch": 0.7357001621034256, "grad_norm": 1.565480933889819, "learning_rate": 3.445290187427969e-06, "loss": 0.5657, "step": 20650 }, { "epoch": 0.7357357892299197, "grad_norm": 1.2180432953614062, "learning_rate": 3.444418754583122e-06, "loss": 0.3911, "step": 20651 }, { "epoch": 0.7357714163564137, "grad_norm": 1.4429897608252784, "learning_rate": 3.443547409030418e-06, "loss": 0.4637, "step": 20652 }, { "epoch": 0.7358070434829079, "grad_norm": 1.3063267572974453, "learning_rate": 3.4426761507814664e-06, "loss": 0.4227, "step": 20653 }, { "epoch": 0.735842670609402, "grad_norm": 1.3971223557901156, "learning_rate": 3.4418049798478594e-06, "loss": 0.5457, "step": 20654 }, { "epoch": 0.7358782977358961, "grad_norm": 1.467754988881637, "learning_rate": 3.440933896241209e-06, "loss": 0.4604, "step": 20655 }, { "epoch": 0.7359139248623903, "grad_norm": 1.3956490850662453, "learning_rate": 3.4400628999731046e-06, "loss": 0.5863, "step": 20656 }, { "epoch": 0.7359495519888843, "grad_norm": 1.6270674167797057, "learning_rate": 3.4391919910551506e-06, "loss": 0.5347, "step": 20657 }, { "epoch": 0.7359851791153784, "grad_norm": 1.3710386954993412, "learning_rate": 3.4383211694989395e-06, "loss": 0.5079, "step": 20658 }, { "epoch": 0.7360208062418726, "grad_norm": 1.2953112069223147, "learning_rate": 3.4374504353160665e-06, "loss": 0.4295, "step": 20659 }, { "epoch": 0.7360564333683667, "grad_norm": 1.4475544399571572, "learning_rate": 3.436579788518132e-06, "loss": 0.4987, "step": 20660 }, { "epoch": 0.7360920604948608, "grad_norm": 2.7457403380613936, "learning_rate": 3.4357092291167227e-06, "loss": 0.4587, "step": 20661 }, { "epoch": 0.7361276876213549, "grad_norm": 1.5015892030499403, "learning_rate": 3.434838757123432e-06, "loss": 0.4344, "step": 20662 }, { "epoch": 0.736163314747849, "grad_norm": 1.3315581280119855, "learning_rate": 3.433968372549852e-06, "loss": 0.4366, "step": 20663 }, { "epoch": 0.7361989418743431, "grad_norm": 1.1482170370901486, "learning_rate": 3.433098075407576e-06, "loss": 0.4184, "step": 20664 }, { "epoch": 0.7362345690008373, "grad_norm": 1.4283666194480513, "learning_rate": 3.4322278657081866e-06, "loss": 0.3974, "step": 20665 }, { "epoch": 0.7362701961273314, "grad_norm": 1.6656848618145117, "learning_rate": 3.431357743463276e-06, "loss": 0.4709, "step": 20666 }, { "epoch": 0.7363058232538254, "grad_norm": 1.1729440735844008, "learning_rate": 3.430487708684426e-06, "loss": 0.4579, "step": 20667 }, { "epoch": 0.7363414503803196, "grad_norm": 1.6549482354218898, "learning_rate": 3.429617761383223e-06, "loss": 0.495, "step": 20668 }, { "epoch": 0.7363770775068137, "grad_norm": 1.5040060446741785, "learning_rate": 3.428747901571252e-06, "loss": 0.5664, "step": 20669 }, { "epoch": 0.7364127046333078, "grad_norm": 1.1758702962036702, "learning_rate": 3.4278781292601005e-06, "loss": 0.4095, "step": 20670 }, { "epoch": 0.736448331759802, "grad_norm": 1.471079995129168, "learning_rate": 3.4270084444613404e-06, "loss": 0.5783, "step": 20671 }, { "epoch": 0.736483958886296, "grad_norm": 1.3757349185377208, "learning_rate": 3.4261388471865574e-06, "loss": 0.5334, "step": 20672 }, { "epoch": 0.7365195860127901, "grad_norm": 1.2664896633961018, "learning_rate": 3.4252693374473343e-06, "loss": 0.5023, "step": 20673 }, { "epoch": 0.7365552131392843, "grad_norm": 1.2475645638579886, "learning_rate": 3.424399915255242e-06, "loss": 0.4528, "step": 20674 }, { "epoch": 0.7365908402657784, "grad_norm": 1.2618483500677513, "learning_rate": 3.423530580621863e-06, "loss": 0.4301, "step": 20675 }, { "epoch": 0.7366264673922724, "grad_norm": 1.5666890530326525, "learning_rate": 3.42266133355877e-06, "loss": 0.5166, "step": 20676 }, { "epoch": 0.7366620945187666, "grad_norm": 1.3083451232895342, "learning_rate": 3.421792174077537e-06, "loss": 0.4283, "step": 20677 }, { "epoch": 0.7366977216452607, "grad_norm": 1.288757349091695, "learning_rate": 3.4209231021897395e-06, "loss": 0.469, "step": 20678 }, { "epoch": 0.7367333487717548, "grad_norm": 1.5164822162353953, "learning_rate": 3.4200541179069525e-06, "loss": 0.4292, "step": 20679 }, { "epoch": 0.736768975898249, "grad_norm": 1.0753888109099876, "learning_rate": 3.4191852212407406e-06, "loss": 0.4203, "step": 20680 }, { "epoch": 0.736804603024743, "grad_norm": 1.1491154160387869, "learning_rate": 3.418316412202681e-06, "loss": 0.3886, "step": 20681 }, { "epoch": 0.7368402301512371, "grad_norm": 1.3557617255388867, "learning_rate": 3.4174476908043354e-06, "loss": 0.4516, "step": 20682 }, { "epoch": 0.7368758572777313, "grad_norm": 1.211080190189355, "learning_rate": 3.4165790570572744e-06, "loss": 0.4074, "step": 20683 }, { "epoch": 0.7369114844042254, "grad_norm": 1.1885597725401869, "learning_rate": 3.4157105109730646e-06, "loss": 0.3756, "step": 20684 }, { "epoch": 0.7369471115307195, "grad_norm": 1.4603540362027285, "learning_rate": 3.414842052563272e-06, "loss": 0.5113, "step": 20685 }, { "epoch": 0.7369827386572136, "grad_norm": 1.3007437376104871, "learning_rate": 3.413973681839463e-06, "loss": 0.4211, "step": 20686 }, { "epoch": 0.7370183657837077, "grad_norm": 1.845390831730388, "learning_rate": 3.4131053988131947e-06, "loss": 0.4478, "step": 20687 }, { "epoch": 0.7370539929102018, "grad_norm": 1.228767387972545, "learning_rate": 3.412237203496036e-06, "loss": 0.4436, "step": 20688 }, { "epoch": 0.737089620036696, "grad_norm": 1.8552913151807597, "learning_rate": 3.4113690958995395e-06, "loss": 0.5598, "step": 20689 }, { "epoch": 0.7371252471631901, "grad_norm": 1.5501706466720997, "learning_rate": 3.410501076035272e-06, "loss": 0.5644, "step": 20690 }, { "epoch": 0.7371608742896841, "grad_norm": 1.2467136484071384, "learning_rate": 3.409633143914781e-06, "loss": 0.4176, "step": 20691 }, { "epoch": 0.7371965014161783, "grad_norm": 1.1970715697152987, "learning_rate": 3.408765299549639e-06, "loss": 0.3721, "step": 20692 }, { "epoch": 0.7372321285426724, "grad_norm": 1.5855680140128192, "learning_rate": 3.4078975429513906e-06, "loss": 0.5457, "step": 20693 }, { "epoch": 0.7372677556691665, "grad_norm": 1.2170308262279133, "learning_rate": 3.4070298741315932e-06, "loss": 0.3946, "step": 20694 }, { "epoch": 0.7373033827956607, "grad_norm": 1.25808766708294, "learning_rate": 3.4061622931018057e-06, "loss": 0.387, "step": 20695 }, { "epoch": 0.7373390099221547, "grad_norm": 1.1571489750538713, "learning_rate": 3.405294799873572e-06, "loss": 0.3828, "step": 20696 }, { "epoch": 0.7373746370486488, "grad_norm": 1.2579674748335838, "learning_rate": 3.4044273944584507e-06, "loss": 0.4156, "step": 20697 }, { "epoch": 0.737410264175143, "grad_norm": 1.2668049954911096, "learning_rate": 3.4035600768679855e-06, "loss": 0.4625, "step": 20698 }, { "epoch": 0.7374458913016371, "grad_norm": 1.9133555040328878, "learning_rate": 3.4026928471137286e-06, "loss": 0.4321, "step": 20699 }, { "epoch": 0.7374815184281311, "grad_norm": 1.2160761795593216, "learning_rate": 3.401825705207228e-06, "loss": 0.3841, "step": 20700 }, { "epoch": 0.7375171455546253, "grad_norm": 1.1407506874542794, "learning_rate": 3.4009586511600335e-06, "loss": 0.3815, "step": 20701 }, { "epoch": 0.7375527726811194, "grad_norm": 1.4533390739798762, "learning_rate": 3.400091684983683e-06, "loss": 0.4915, "step": 20702 }, { "epoch": 0.7375883998076135, "grad_norm": 1.6860013607363844, "learning_rate": 3.3992248066897283e-06, "loss": 0.52, "step": 20703 }, { "epoch": 0.7376240269341077, "grad_norm": 1.1975311275234877, "learning_rate": 3.3983580162897057e-06, "loss": 0.4315, "step": 20704 }, { "epoch": 0.7376596540606017, "grad_norm": 1.142726869428551, "learning_rate": 3.3974913137951605e-06, "loss": 0.3817, "step": 20705 }, { "epoch": 0.7376952811870958, "grad_norm": 1.5905329302682762, "learning_rate": 3.396624699217632e-06, "loss": 0.5252, "step": 20706 }, { "epoch": 0.73773090831359, "grad_norm": 1.2627351772016417, "learning_rate": 3.395758172568663e-06, "loss": 0.4999, "step": 20707 }, { "epoch": 0.7377665354400841, "grad_norm": 1.5739133161646441, "learning_rate": 3.394891733859792e-06, "loss": 0.5033, "step": 20708 }, { "epoch": 0.7378021625665782, "grad_norm": 1.58467709299884, "learning_rate": 3.3940253831025514e-06, "loss": 0.519, "step": 20709 }, { "epoch": 0.7378377896930723, "grad_norm": 1.5131029333395751, "learning_rate": 3.3931591203084834e-06, "loss": 0.5917, "step": 20710 }, { "epoch": 0.7378734168195664, "grad_norm": 1.1543298702354619, "learning_rate": 3.3922929454891173e-06, "loss": 0.4027, "step": 20711 }, { "epoch": 0.7379090439460605, "grad_norm": 1.4238766409579595, "learning_rate": 3.391426858655992e-06, "loss": 0.4417, "step": 20712 }, { "epoch": 0.7379446710725547, "grad_norm": 1.458415082731673, "learning_rate": 3.390560859820633e-06, "loss": 0.5457, "step": 20713 }, { "epoch": 0.7379802981990488, "grad_norm": 1.54367061734661, "learning_rate": 3.389694948994576e-06, "loss": 0.5221, "step": 20714 }, { "epoch": 0.7380159253255428, "grad_norm": 1.3523442943119692, "learning_rate": 3.3888291261893513e-06, "loss": 0.4299, "step": 20715 }, { "epoch": 0.738051552452037, "grad_norm": 1.5483509392439854, "learning_rate": 3.387963391416491e-06, "loss": 0.4704, "step": 20716 }, { "epoch": 0.7380871795785311, "grad_norm": 1.9227541986341716, "learning_rate": 3.3870977446875176e-06, "loss": 0.5544, "step": 20717 }, { "epoch": 0.7381228067050252, "grad_norm": 1.1814205999986422, "learning_rate": 3.3862321860139578e-06, "loss": 0.4048, "step": 20718 }, { "epoch": 0.7381584338315194, "grad_norm": 1.2778170882786792, "learning_rate": 3.3853667154073444e-06, "loss": 0.4008, "step": 20719 }, { "epoch": 0.7381940609580134, "grad_norm": 1.7850371932746008, "learning_rate": 3.384501332879192e-06, "loss": 0.4141, "step": 20720 }, { "epoch": 0.7382296880845075, "grad_norm": 1.209415778780422, "learning_rate": 3.383636038441028e-06, "loss": 0.4216, "step": 20721 }, { "epoch": 0.7382653152110017, "grad_norm": 1.5623323828833497, "learning_rate": 3.382770832104376e-06, "loss": 0.5561, "step": 20722 }, { "epoch": 0.7383009423374958, "grad_norm": 1.6929775154652906, "learning_rate": 3.381905713880759e-06, "loss": 0.5098, "step": 20723 }, { "epoch": 0.7383365694639898, "grad_norm": 1.2599071796933672, "learning_rate": 3.381040683781689e-06, "loss": 0.39, "step": 20724 }, { "epoch": 0.738372196590484, "grad_norm": 1.2488733255470341, "learning_rate": 3.380175741818692e-06, "loss": 0.4417, "step": 20725 }, { "epoch": 0.7384078237169781, "grad_norm": 1.2599743531041978, "learning_rate": 3.379310888003279e-06, "loss": 0.5371, "step": 20726 }, { "epoch": 0.7384434508434722, "grad_norm": 1.4516234592360373, "learning_rate": 3.3784461223469713e-06, "loss": 0.4797, "step": 20727 }, { "epoch": 0.7384790779699664, "grad_norm": 1.2092471429348495, "learning_rate": 3.3775814448612797e-06, "loss": 0.4426, "step": 20728 }, { "epoch": 0.7385147050964604, "grad_norm": 1.4513224216471083, "learning_rate": 3.3767168555577158e-06, "loss": 0.5135, "step": 20729 }, { "epoch": 0.7385503322229545, "grad_norm": 1.2224405186802831, "learning_rate": 3.3758523544478038e-06, "loss": 0.3938, "step": 20730 }, { "epoch": 0.7385859593494487, "grad_norm": 1.0562112853946146, "learning_rate": 3.374987941543043e-06, "loss": 0.3736, "step": 20731 }, { "epoch": 0.7386215864759428, "grad_norm": 1.2290303963314526, "learning_rate": 3.374123616854953e-06, "loss": 0.408, "step": 20732 }, { "epoch": 0.7386572136024369, "grad_norm": 1.1792308283850959, "learning_rate": 3.3732593803950354e-06, "loss": 0.4586, "step": 20733 }, { "epoch": 0.738692840728931, "grad_norm": 1.2994173066682326, "learning_rate": 3.372395232174803e-06, "loss": 0.4475, "step": 20734 }, { "epoch": 0.7387284678554251, "grad_norm": 2.450000198315332, "learning_rate": 3.3715311722057577e-06, "loss": 0.4416, "step": 20735 }, { "epoch": 0.7387640949819192, "grad_norm": 1.241740816454003, "learning_rate": 3.370667200499408e-06, "loss": 0.5252, "step": 20736 }, { "epoch": 0.7387997221084134, "grad_norm": 1.448546904355114, "learning_rate": 3.3698033170672574e-06, "loss": 0.4948, "step": 20737 }, { "epoch": 0.7388353492349075, "grad_norm": 1.4534098475045367, "learning_rate": 3.3689395219208143e-06, "loss": 0.5145, "step": 20738 }, { "epoch": 0.7388709763614015, "grad_norm": 1.7073837487495462, "learning_rate": 3.368075815071572e-06, "loss": 0.5383, "step": 20739 }, { "epoch": 0.7389066034878957, "grad_norm": 1.4031434509662144, "learning_rate": 3.3672121965310358e-06, "loss": 0.469, "step": 20740 }, { "epoch": 0.7389422306143898, "grad_norm": 1.4581867526870798, "learning_rate": 3.366348666310709e-06, "loss": 0.4403, "step": 20741 }, { "epoch": 0.7389778577408839, "grad_norm": 1.5315623390960251, "learning_rate": 3.3654852244220824e-06, "loss": 0.4286, "step": 20742 }, { "epoch": 0.7390134848673781, "grad_norm": 1.2663916642398008, "learning_rate": 3.3646218708766586e-06, "loss": 0.4206, "step": 20743 }, { "epoch": 0.7390491119938721, "grad_norm": 1.0514232985438603, "learning_rate": 3.363758605685932e-06, "loss": 0.3999, "step": 20744 }, { "epoch": 0.7390847391203662, "grad_norm": 1.5006000644226092, "learning_rate": 3.3628954288614003e-06, "loss": 0.5282, "step": 20745 }, { "epoch": 0.7391203662468604, "grad_norm": 1.2584598133461407, "learning_rate": 3.362032340414553e-06, "loss": 0.4104, "step": 20746 }, { "epoch": 0.7391559933733545, "grad_norm": 1.5480963917342536, "learning_rate": 3.3611693403568866e-06, "loss": 0.448, "step": 20747 }, { "epoch": 0.7391916204998485, "grad_norm": 1.4536122587870477, "learning_rate": 3.3603064286998888e-06, "loss": 0.5088, "step": 20748 }, { "epoch": 0.7392272476263427, "grad_norm": 1.178285252124513, "learning_rate": 3.3594436054550537e-06, "loss": 0.4083, "step": 20749 }, { "epoch": 0.7392628747528368, "grad_norm": 1.3442960909539332, "learning_rate": 3.3585808706338663e-06, "loss": 0.3919, "step": 20750 }, { "epoch": 0.7392985018793309, "grad_norm": 1.2710247876230987, "learning_rate": 3.357718224247817e-06, "loss": 0.4104, "step": 20751 }, { "epoch": 0.7393341290058251, "grad_norm": 1.299341395038129, "learning_rate": 3.3568556663083906e-06, "loss": 0.4198, "step": 20752 }, { "epoch": 0.7393697561323191, "grad_norm": 1.3410198723257383, "learning_rate": 3.355993196827075e-06, "loss": 0.4349, "step": 20753 }, { "epoch": 0.7394053832588132, "grad_norm": 1.2550170463814863, "learning_rate": 3.3551308158153585e-06, "loss": 0.4393, "step": 20754 }, { "epoch": 0.7394410103853074, "grad_norm": 1.4223238755151546, "learning_rate": 3.354268523284715e-06, "loss": 0.4471, "step": 20755 }, { "epoch": 0.7394766375118015, "grad_norm": 1.3517714821190396, "learning_rate": 3.353406319246635e-06, "loss": 0.4291, "step": 20756 }, { "epoch": 0.7395122646382956, "grad_norm": 1.282574049685611, "learning_rate": 3.352544203712591e-06, "loss": 0.4114, "step": 20757 }, { "epoch": 0.7395478917647897, "grad_norm": 1.2361792391660058, "learning_rate": 3.3516821766940687e-06, "loss": 0.3936, "step": 20758 }, { "epoch": 0.7395835188912838, "grad_norm": 1.7378784977963337, "learning_rate": 3.3508202382025447e-06, "loss": 0.5478, "step": 20759 }, { "epoch": 0.7396191460177779, "grad_norm": 1.3495946578343325, "learning_rate": 3.3499583882495003e-06, "loss": 0.4695, "step": 20760 }, { "epoch": 0.7396547731442721, "grad_norm": 2.3682623683650816, "learning_rate": 3.349096626846404e-06, "loss": 0.5387, "step": 20761 }, { "epoch": 0.7396904002707662, "grad_norm": 1.1720771981604137, "learning_rate": 3.3482349540047387e-06, "loss": 0.4299, "step": 20762 }, { "epoch": 0.7397260273972602, "grad_norm": 1.6600350370064378, "learning_rate": 3.347373369735971e-06, "loss": 0.514, "step": 20763 }, { "epoch": 0.7397616545237544, "grad_norm": 1.1735265590752786, "learning_rate": 3.346511874051577e-06, "loss": 0.3553, "step": 20764 }, { "epoch": 0.7397972816502485, "grad_norm": 1.7422119832283596, "learning_rate": 3.345650466963031e-06, "loss": 0.5407, "step": 20765 }, { "epoch": 0.7398329087767426, "grad_norm": 1.4397209860231257, "learning_rate": 3.3447891484817963e-06, "loss": 0.5566, "step": 20766 }, { "epoch": 0.7398685359032368, "grad_norm": 1.1982592085819188, "learning_rate": 3.3439279186193464e-06, "loss": 0.3845, "step": 20767 }, { "epoch": 0.7399041630297308, "grad_norm": 1.1841648153060116, "learning_rate": 3.343066777387148e-06, "loss": 0.4311, "step": 20768 }, { "epoch": 0.7399397901562249, "grad_norm": 1.2591709359218866, "learning_rate": 3.342205724796672e-06, "loss": 0.4143, "step": 20769 }, { "epoch": 0.7399754172827191, "grad_norm": 1.1752844731189058, "learning_rate": 3.3413447608593774e-06, "loss": 0.3896, "step": 20770 }, { "epoch": 0.7400110444092132, "grad_norm": 1.414911816065704, "learning_rate": 3.3404838855867337e-06, "loss": 0.4195, "step": 20771 }, { "epoch": 0.7400466715357072, "grad_norm": 1.542724336589234, "learning_rate": 3.3396230989901988e-06, "loss": 0.533, "step": 20772 }, { "epoch": 0.7400822986622014, "grad_norm": 1.250025116779742, "learning_rate": 3.338762401081238e-06, "loss": 0.4396, "step": 20773 }, { "epoch": 0.7401179257886955, "grad_norm": 1.4594626680752392, "learning_rate": 3.337901791871312e-06, "loss": 0.5431, "step": 20774 }, { "epoch": 0.7401535529151896, "grad_norm": 1.2389552403951969, "learning_rate": 3.3370412713718804e-06, "loss": 0.4333, "step": 20775 }, { "epoch": 0.7401891800416838, "grad_norm": 1.274806390782422, "learning_rate": 3.336180839594404e-06, "loss": 0.471, "step": 20776 }, { "epoch": 0.7402248071681778, "grad_norm": 1.4000819712189472, "learning_rate": 3.335320496550335e-06, "loss": 0.4544, "step": 20777 }, { "epoch": 0.7402604342946719, "grad_norm": 1.300658050876774, "learning_rate": 3.3344602422511343e-06, "loss": 0.4018, "step": 20778 }, { "epoch": 0.7402960614211661, "grad_norm": 1.0952575860373768, "learning_rate": 3.333600076708251e-06, "loss": 0.3807, "step": 20779 }, { "epoch": 0.7403316885476602, "grad_norm": 1.3991873901104621, "learning_rate": 3.3327399999331434e-06, "loss": 0.5254, "step": 20780 }, { "epoch": 0.7403673156741543, "grad_norm": 1.2662685587099862, "learning_rate": 3.3318800119372617e-06, "loss": 0.4917, "step": 20781 }, { "epoch": 0.7404029428006484, "grad_norm": 1.1931580397502803, "learning_rate": 3.331020112732062e-06, "loss": 0.3916, "step": 20782 }, { "epoch": 0.7404385699271425, "grad_norm": 1.4171929824250096, "learning_rate": 3.3301603023289873e-06, "loss": 0.4411, "step": 20783 }, { "epoch": 0.7404741970536366, "grad_norm": 1.2029989106791592, "learning_rate": 3.329300580739494e-06, "loss": 0.4411, "step": 20784 }, { "epoch": 0.7405098241801308, "grad_norm": 1.1904750589609168, "learning_rate": 3.3284409479750223e-06, "loss": 0.4588, "step": 20785 }, { "epoch": 0.7405454513066249, "grad_norm": 3.930513559580793, "learning_rate": 3.3275814040470235e-06, "loss": 0.5528, "step": 20786 }, { "epoch": 0.7405810784331189, "grad_norm": 1.647409598907826, "learning_rate": 3.3267219489669434e-06, "loss": 0.5351, "step": 20787 }, { "epoch": 0.7406167055596131, "grad_norm": 1.3184835338970478, "learning_rate": 3.325862582746223e-06, "loss": 0.4478, "step": 20788 }, { "epoch": 0.7406523326861072, "grad_norm": 1.3721007949369064, "learning_rate": 3.325003305396307e-06, "loss": 0.4601, "step": 20789 }, { "epoch": 0.7406879598126013, "grad_norm": 1.3580199655686636, "learning_rate": 3.324144116928637e-06, "loss": 0.4135, "step": 20790 }, { "epoch": 0.7407235869390955, "grad_norm": 1.1706936722150254, "learning_rate": 3.3232850173546584e-06, "loss": 0.3822, "step": 20791 }, { "epoch": 0.7407592140655895, "grad_norm": 1.3827015784052534, "learning_rate": 3.3224260066858026e-06, "loss": 0.4524, "step": 20792 }, { "epoch": 0.7407948411920836, "grad_norm": 1.2094590928174243, "learning_rate": 3.3215670849335156e-06, "loss": 0.4482, "step": 20793 }, { "epoch": 0.7408304683185778, "grad_norm": 1.5864135807577262, "learning_rate": 3.320708252109227e-06, "loss": 0.5106, "step": 20794 }, { "epoch": 0.7408660954450719, "grad_norm": 1.8310870885589743, "learning_rate": 3.3198495082243777e-06, "loss": 0.5108, "step": 20795 }, { "epoch": 0.740901722571566, "grad_norm": 1.7001252419159343, "learning_rate": 3.3189908532904002e-06, "loss": 0.5525, "step": 20796 }, { "epoch": 0.7409373496980601, "grad_norm": 1.4197856837302103, "learning_rate": 3.3181322873187325e-06, "loss": 0.4406, "step": 20797 }, { "epoch": 0.7409729768245542, "grad_norm": 1.2942525216045258, "learning_rate": 3.3172738103208015e-06, "loss": 0.3883, "step": 20798 }, { "epoch": 0.7410086039510483, "grad_norm": 1.895909857164053, "learning_rate": 3.316415422308039e-06, "loss": 0.5464, "step": 20799 }, { "epoch": 0.7410442310775425, "grad_norm": 1.566940428733085, "learning_rate": 3.3155571232918817e-06, "loss": 0.4454, "step": 20800 }, { "epoch": 0.7410798582040365, "grad_norm": 1.2907634189037827, "learning_rate": 3.3146989132837492e-06, "loss": 0.375, "step": 20801 }, { "epoch": 0.7411154853305306, "grad_norm": 1.2271743064958596, "learning_rate": 3.3138407922950767e-06, "loss": 0.3919, "step": 20802 }, { "epoch": 0.7411511124570248, "grad_norm": 1.2728790103328356, "learning_rate": 3.3129827603372843e-06, "loss": 0.4238, "step": 20803 }, { "epoch": 0.7411867395835189, "grad_norm": 1.2263611247859527, "learning_rate": 3.3121248174218e-06, "loss": 0.3879, "step": 20804 }, { "epoch": 0.741222366710013, "grad_norm": 1.6584407122330695, "learning_rate": 3.3112669635600494e-06, "loss": 0.6493, "step": 20805 }, { "epoch": 0.7412579938365071, "grad_norm": 1.3976294001903182, "learning_rate": 3.310409198763457e-06, "loss": 0.5316, "step": 20806 }, { "epoch": 0.7412936209630012, "grad_norm": 1.2879558905132633, "learning_rate": 3.3095515230434384e-06, "loss": 0.4075, "step": 20807 }, { "epoch": 0.7413292480894953, "grad_norm": 1.2504302935378495, "learning_rate": 3.308693936411421e-06, "loss": 0.3749, "step": 20808 }, { "epoch": 0.7413648752159895, "grad_norm": 1.4644808658626889, "learning_rate": 3.3078364388788176e-06, "loss": 0.5414, "step": 20809 }, { "epoch": 0.7414005023424836, "grad_norm": 1.5493728055723501, "learning_rate": 3.3069790304570495e-06, "loss": 0.4801, "step": 20810 }, { "epoch": 0.7414361294689776, "grad_norm": 1.506453108206119, "learning_rate": 3.306121711157534e-06, "loss": 0.5259, "step": 20811 }, { "epoch": 0.7414717565954718, "grad_norm": 1.5329345954178335, "learning_rate": 3.3052644809916857e-06, "loss": 0.6179, "step": 20812 }, { "epoch": 0.7415073837219659, "grad_norm": 1.2420744251056628, "learning_rate": 3.3044073399709243e-06, "loss": 0.3793, "step": 20813 }, { "epoch": 0.74154301084846, "grad_norm": 1.31903223315054, "learning_rate": 3.303550288106656e-06, "loss": 0.4245, "step": 20814 }, { "epoch": 0.7415786379749542, "grad_norm": 1.2877911987274888, "learning_rate": 3.302693325410299e-06, "loss": 0.4318, "step": 20815 }, { "epoch": 0.7416142651014482, "grad_norm": 1.5015749067338036, "learning_rate": 3.3018364518932577e-06, "loss": 0.481, "step": 20816 }, { "epoch": 0.7416498922279423, "grad_norm": 1.2734339851678556, "learning_rate": 3.3009796675669504e-06, "loss": 0.4418, "step": 20817 }, { "epoch": 0.7416855193544365, "grad_norm": 1.166428005897933, "learning_rate": 3.300122972442773e-06, "loss": 0.4032, "step": 20818 }, { "epoch": 0.7417211464809306, "grad_norm": 1.6240929717061159, "learning_rate": 3.299266366532149e-06, "loss": 0.4994, "step": 20819 }, { "epoch": 0.7417567736074246, "grad_norm": 1.3952343852435334, "learning_rate": 3.2984098498464736e-06, "loss": 0.4137, "step": 20820 }, { "epoch": 0.7417924007339188, "grad_norm": 1.4748673466608753, "learning_rate": 3.297553422397154e-06, "loss": 0.4516, "step": 20821 }, { "epoch": 0.7418280278604129, "grad_norm": 1.2521033278892924, "learning_rate": 3.2966970841956004e-06, "loss": 0.4361, "step": 20822 }, { "epoch": 0.741863654986907, "grad_norm": 1.4928904568811163, "learning_rate": 3.2958408352532055e-06, "loss": 0.4932, "step": 20823 }, { "epoch": 0.7418992821134012, "grad_norm": 1.849182893001984, "learning_rate": 3.2949846755813795e-06, "loss": 0.5697, "step": 20824 }, { "epoch": 0.7419349092398952, "grad_norm": 1.2172198286072067, "learning_rate": 3.2941286051915165e-06, "loss": 0.4055, "step": 20825 }, { "epoch": 0.7419705363663893, "grad_norm": 1.196154928325037, "learning_rate": 3.2932726240950173e-06, "loss": 0.4248, "step": 20826 }, { "epoch": 0.7420061634928835, "grad_norm": 1.3148570990300428, "learning_rate": 3.292416732303282e-06, "loss": 0.383, "step": 20827 }, { "epoch": 0.7420417906193776, "grad_norm": 1.927493608062567, "learning_rate": 3.2915609298277083e-06, "loss": 0.5072, "step": 20828 }, { "epoch": 0.7420774177458717, "grad_norm": 1.6351195207120122, "learning_rate": 3.290705216679687e-06, "loss": 0.5204, "step": 20829 }, { "epoch": 0.7421130448723658, "grad_norm": 1.242115794537332, "learning_rate": 3.2898495928706187e-06, "loss": 0.403, "step": 20830 }, { "epoch": 0.7421486719988599, "grad_norm": 1.9803445383378948, "learning_rate": 3.2889940584118897e-06, "loss": 0.599, "step": 20831 }, { "epoch": 0.742184299125354, "grad_norm": 1.2573364819461466, "learning_rate": 3.288138613314895e-06, "loss": 0.4419, "step": 20832 }, { "epoch": 0.7422199262518482, "grad_norm": 1.7752158978852695, "learning_rate": 3.287283257591026e-06, "loss": 0.4508, "step": 20833 }, { "epoch": 0.7422555533783423, "grad_norm": 1.439231191521346, "learning_rate": 3.286427991251674e-06, "loss": 0.5771, "step": 20834 }, { "epoch": 0.7422911805048363, "grad_norm": 1.231327510557187, "learning_rate": 3.285572814308228e-06, "loss": 0.4219, "step": 20835 }, { "epoch": 0.7423268076313305, "grad_norm": 1.2177111182446365, "learning_rate": 3.2847177267720686e-06, "loss": 0.3933, "step": 20836 }, { "epoch": 0.7423624347578246, "grad_norm": 1.533703938267516, "learning_rate": 3.283862728654591e-06, "loss": 0.5108, "step": 20837 }, { "epoch": 0.7423980618843187, "grad_norm": 1.4558932318758488, "learning_rate": 3.283007819967172e-06, "loss": 0.5441, "step": 20838 }, { "epoch": 0.7424336890108129, "grad_norm": 1.4163057259961827, "learning_rate": 3.2821530007212023e-06, "loss": 0.4704, "step": 20839 }, { "epoch": 0.7424693161373069, "grad_norm": 1.233569433430194, "learning_rate": 3.2812982709280573e-06, "loss": 0.3922, "step": 20840 }, { "epoch": 0.742504943263801, "grad_norm": 1.3478801769564617, "learning_rate": 3.2804436305991217e-06, "loss": 0.4082, "step": 20841 }, { "epoch": 0.7425405703902952, "grad_norm": 3.666661143625572, "learning_rate": 3.2795890797457763e-06, "loss": 0.5648, "step": 20842 }, { "epoch": 0.7425761975167893, "grad_norm": 1.3760104593403892, "learning_rate": 3.278734618379402e-06, "loss": 0.4848, "step": 20843 }, { "epoch": 0.7426118246432833, "grad_norm": 1.3407589495424916, "learning_rate": 3.2778802465113725e-06, "loss": 0.3983, "step": 20844 }, { "epoch": 0.7426474517697775, "grad_norm": 1.63729658301691, "learning_rate": 3.2770259641530642e-06, "loss": 0.4857, "step": 20845 }, { "epoch": 0.7426830788962716, "grad_norm": 1.9602908082576753, "learning_rate": 3.276171771315858e-06, "loss": 0.5785, "step": 20846 }, { "epoch": 0.7427187060227657, "grad_norm": 1.3627181748971444, "learning_rate": 3.275317668011122e-06, "loss": 0.4429, "step": 20847 }, { "epoch": 0.7427543331492599, "grad_norm": 1.5078825923772992, "learning_rate": 3.2744636542502307e-06, "loss": 0.5463, "step": 20848 }, { "epoch": 0.742789960275754, "grad_norm": 1.8894912512180866, "learning_rate": 3.2736097300445567e-06, "loss": 0.6013, "step": 20849 }, { "epoch": 0.742825587402248, "grad_norm": 1.7558614392167597, "learning_rate": 3.272755895405474e-06, "loss": 0.5605, "step": 20850 }, { "epoch": 0.7428612145287422, "grad_norm": 1.3953332254466309, "learning_rate": 3.271902150344346e-06, "loss": 0.5114, "step": 20851 }, { "epoch": 0.7428968416552363, "grad_norm": 1.7992708367657584, "learning_rate": 3.2710484948725465e-06, "loss": 0.4915, "step": 20852 }, { "epoch": 0.7429324687817304, "grad_norm": 3.375408809166429, "learning_rate": 3.2701949290014355e-06, "loss": 0.4796, "step": 20853 }, { "epoch": 0.7429680959082245, "grad_norm": 1.3239342517688124, "learning_rate": 3.2693414527423873e-06, "loss": 0.4485, "step": 20854 }, { "epoch": 0.7430037230347186, "grad_norm": 1.2856787854819953, "learning_rate": 3.2684880661067573e-06, "loss": 0.3908, "step": 20855 }, { "epoch": 0.7430393501612127, "grad_norm": 1.4272314618295643, "learning_rate": 3.267634769105912e-06, "loss": 0.4907, "step": 20856 }, { "epoch": 0.7430749772877069, "grad_norm": 1.2987547789267448, "learning_rate": 3.266781561751221e-06, "loss": 0.4246, "step": 20857 }, { "epoch": 0.743110604414201, "grad_norm": 1.5644584295258959, "learning_rate": 3.2659284440540375e-06, "loss": 0.5365, "step": 20858 }, { "epoch": 0.743146231540695, "grad_norm": 1.1698624670266942, "learning_rate": 3.2650754160257257e-06, "loss": 0.4549, "step": 20859 }, { "epoch": 0.7431818586671892, "grad_norm": 1.1379265858175982, "learning_rate": 3.2642224776776388e-06, "loss": 0.3794, "step": 20860 }, { "epoch": 0.7432174857936833, "grad_norm": 1.3957331824851116, "learning_rate": 3.2633696290211413e-06, "loss": 0.3908, "step": 20861 }, { "epoch": 0.7432531129201774, "grad_norm": 1.519582685989031, "learning_rate": 3.2625168700675836e-06, "loss": 0.58, "step": 20862 }, { "epoch": 0.7432887400466716, "grad_norm": 1.2450334698158816, "learning_rate": 3.2616642008283218e-06, "loss": 0.424, "step": 20863 }, { "epoch": 0.7433243671731656, "grad_norm": 1.2021670046739417, "learning_rate": 3.260811621314711e-06, "loss": 0.4219, "step": 20864 }, { "epoch": 0.7433599942996597, "grad_norm": 3.4999854639923647, "learning_rate": 3.2599591315381075e-06, "loss": 0.4018, "step": 20865 }, { "epoch": 0.7433956214261539, "grad_norm": 1.403596422216659, "learning_rate": 3.2591067315098556e-06, "loss": 0.4235, "step": 20866 }, { "epoch": 0.743431248552648, "grad_norm": 1.3533214736473613, "learning_rate": 3.2582544212413103e-06, "loss": 0.4957, "step": 20867 }, { "epoch": 0.743466875679142, "grad_norm": 1.7147000560331662, "learning_rate": 3.257402200743821e-06, "loss": 0.5068, "step": 20868 }, { "epoch": 0.7435025028056362, "grad_norm": 1.516832565964157, "learning_rate": 3.2565500700287324e-06, "loss": 0.5092, "step": 20869 }, { "epoch": 0.7435381299321303, "grad_norm": 1.6387937572712177, "learning_rate": 3.2556980291073927e-06, "loss": 0.518, "step": 20870 }, { "epoch": 0.7435737570586244, "grad_norm": 1.1647906190215453, "learning_rate": 3.2548460779911475e-06, "loss": 0.3885, "step": 20871 }, { "epoch": 0.7436093841851186, "grad_norm": 1.4134612252201129, "learning_rate": 3.253994216691344e-06, "loss": 0.5698, "step": 20872 }, { "epoch": 0.7436450113116126, "grad_norm": 1.414907112578116, "learning_rate": 3.2531424452193205e-06, "loss": 0.5193, "step": 20873 }, { "epoch": 0.7436806384381067, "grad_norm": 1.3749285173386403, "learning_rate": 3.2522907635864244e-06, "loss": 0.4178, "step": 20874 }, { "epoch": 0.7437162655646009, "grad_norm": 1.3060128965607085, "learning_rate": 3.2514391718039882e-06, "loss": 0.431, "step": 20875 }, { "epoch": 0.743751892691095, "grad_norm": 1.6880145936130393, "learning_rate": 3.2505876698833617e-06, "loss": 0.5132, "step": 20876 }, { "epoch": 0.7437875198175891, "grad_norm": 1.9310275994734993, "learning_rate": 3.249736257835874e-06, "loss": 0.518, "step": 20877 }, { "epoch": 0.7438231469440832, "grad_norm": 1.476617235330223, "learning_rate": 3.2488849356728656e-06, "loss": 0.5423, "step": 20878 }, { "epoch": 0.7438587740705773, "grad_norm": 1.6283319242832939, "learning_rate": 3.248033703405673e-06, "loss": 0.501, "step": 20879 }, { "epoch": 0.7438944011970714, "grad_norm": 1.3157366137794013, "learning_rate": 3.247182561045632e-06, "loss": 0.5162, "step": 20880 }, { "epoch": 0.7439300283235656, "grad_norm": 1.3371761304871737, "learning_rate": 3.246331508604078e-06, "loss": 0.4689, "step": 20881 }, { "epoch": 0.7439656554500597, "grad_norm": 1.4245601010909177, "learning_rate": 3.245480546092337e-06, "loss": 0.4219, "step": 20882 }, { "epoch": 0.7440012825765537, "grad_norm": 1.279966150970182, "learning_rate": 3.2446296735217462e-06, "loss": 0.4509, "step": 20883 }, { "epoch": 0.7440369097030479, "grad_norm": 1.3668127829108714, "learning_rate": 3.2437788909036317e-06, "loss": 0.4154, "step": 20884 }, { "epoch": 0.744072536829542, "grad_norm": 1.3656965758944295, "learning_rate": 3.242928198249322e-06, "loss": 0.5291, "step": 20885 }, { "epoch": 0.7441081639560361, "grad_norm": 1.1901888788247474, "learning_rate": 3.2420775955701468e-06, "loss": 0.443, "step": 20886 }, { "epoch": 0.7441437910825303, "grad_norm": 1.4402729645623826, "learning_rate": 3.241227082877435e-06, "loss": 0.46, "step": 20887 }, { "epoch": 0.7441794182090243, "grad_norm": 1.3210193245650468, "learning_rate": 3.2403766601825046e-06, "loss": 0.4588, "step": 20888 }, { "epoch": 0.7442150453355184, "grad_norm": 1.6800243683958513, "learning_rate": 3.2395263274966882e-06, "loss": 0.428, "step": 20889 }, { "epoch": 0.7442506724620126, "grad_norm": 1.630940227713243, "learning_rate": 3.2386760848313002e-06, "loss": 0.5334, "step": 20890 }, { "epoch": 0.7442862995885067, "grad_norm": 1.6781125627652422, "learning_rate": 3.2378259321976668e-06, "loss": 0.4907, "step": 20891 }, { "epoch": 0.7443219267150007, "grad_norm": 1.40400149422712, "learning_rate": 3.2369758696071097e-06, "loss": 0.5111, "step": 20892 }, { "epoch": 0.7443575538414949, "grad_norm": 1.252494230753371, "learning_rate": 3.23612589707094e-06, "loss": 0.4684, "step": 20893 }, { "epoch": 0.744393180967989, "grad_norm": 1.224734452960226, "learning_rate": 3.235276014600488e-06, "loss": 0.4059, "step": 20894 }, { "epoch": 0.7444288080944831, "grad_norm": 1.5076188598057285, "learning_rate": 3.234426222207061e-06, "loss": 0.5327, "step": 20895 }, { "epoch": 0.7444644352209773, "grad_norm": 1.3231762567626837, "learning_rate": 3.2335765199019807e-06, "loss": 0.4253, "step": 20896 }, { "epoch": 0.7445000623474713, "grad_norm": 1.4680532200805527, "learning_rate": 3.232726907696556e-06, "loss": 0.4812, "step": 20897 }, { "epoch": 0.7445356894739654, "grad_norm": 1.1226735283729512, "learning_rate": 3.2318773856021057e-06, "loss": 0.3959, "step": 20898 }, { "epoch": 0.7445713166004596, "grad_norm": 1.2511874799659004, "learning_rate": 3.2310279536299347e-06, "loss": 0.4089, "step": 20899 }, { "epoch": 0.7446069437269537, "grad_norm": 1.7605756434454705, "learning_rate": 3.230178611791358e-06, "loss": 0.5595, "step": 20900 }, { "epoch": 0.7446425708534478, "grad_norm": 1.9849383038437902, "learning_rate": 3.229329360097685e-06, "loss": 0.5444, "step": 20901 }, { "epoch": 0.744678197979942, "grad_norm": 1.2151523713820818, "learning_rate": 3.2284801985602242e-06, "loss": 0.406, "step": 20902 }, { "epoch": 0.744713825106436, "grad_norm": 1.208152147386437, "learning_rate": 3.227631127190285e-06, "loss": 0.4049, "step": 20903 }, { "epoch": 0.7447494522329301, "grad_norm": 1.605992702006605, "learning_rate": 3.2267821459991686e-06, "loss": 0.4297, "step": 20904 }, { "epoch": 0.7447850793594243, "grad_norm": 1.6658627400268884, "learning_rate": 3.2259332549981847e-06, "loss": 0.6008, "step": 20905 }, { "epoch": 0.7448207064859184, "grad_norm": 1.1957470891713309, "learning_rate": 3.2250844541986313e-06, "loss": 0.4019, "step": 20906 }, { "epoch": 0.7448563336124124, "grad_norm": 1.5539390908181667, "learning_rate": 3.224235743611814e-06, "loss": 0.4876, "step": 20907 }, { "epoch": 0.7448919607389066, "grad_norm": 1.4687453763495688, "learning_rate": 3.2233871232490344e-06, "loss": 0.5278, "step": 20908 }, { "epoch": 0.7449275878654007, "grad_norm": 1.6514578453464608, "learning_rate": 3.2225385931215935e-06, "loss": 0.5553, "step": 20909 }, { "epoch": 0.7449632149918948, "grad_norm": 1.343492390514608, "learning_rate": 3.2216901532407862e-06, "loss": 0.3971, "step": 20910 }, { "epoch": 0.744998842118389, "grad_norm": 1.0996200869803812, "learning_rate": 3.2208418036179155e-06, "loss": 0.391, "step": 20911 }, { "epoch": 0.745034469244883, "grad_norm": 1.3707574479940738, "learning_rate": 3.2199935442642704e-06, "loss": 0.5095, "step": 20912 }, { "epoch": 0.7450700963713772, "grad_norm": 1.2419396770902391, "learning_rate": 3.2191453751911505e-06, "loss": 0.4749, "step": 20913 }, { "epoch": 0.7451057234978713, "grad_norm": 1.2985074244306725, "learning_rate": 3.2182972964098535e-06, "loss": 0.4222, "step": 20914 }, { "epoch": 0.7451413506243654, "grad_norm": 1.2375837636066425, "learning_rate": 3.217449307931665e-06, "loss": 0.4334, "step": 20915 }, { "epoch": 0.7451769777508596, "grad_norm": 1.146626055154258, "learning_rate": 3.2166014097678787e-06, "loss": 0.4578, "step": 20916 }, { "epoch": 0.7452126048773536, "grad_norm": 1.3325119681822688, "learning_rate": 3.2157536019297863e-06, "loss": 0.4706, "step": 20917 }, { "epoch": 0.7452482320038477, "grad_norm": 1.3063497324009872, "learning_rate": 3.2149058844286796e-06, "loss": 0.434, "step": 20918 }, { "epoch": 0.7452838591303419, "grad_norm": 1.229290040043449, "learning_rate": 3.2140582572758405e-06, "loss": 0.4373, "step": 20919 }, { "epoch": 0.745319486256836, "grad_norm": 1.4955096551365386, "learning_rate": 3.2132107204825626e-06, "loss": 0.56, "step": 20920 }, { "epoch": 0.74535511338333, "grad_norm": 1.4138989512542302, "learning_rate": 3.212363274060124e-06, "loss": 0.3888, "step": 20921 }, { "epoch": 0.7453907405098242, "grad_norm": 1.3174881554760935, "learning_rate": 3.2115159180198117e-06, "loss": 0.4711, "step": 20922 }, { "epoch": 0.7454263676363183, "grad_norm": 1.363027937655725, "learning_rate": 3.2106686523729116e-06, "loss": 0.579, "step": 20923 }, { "epoch": 0.7454619947628124, "grad_norm": 1.2200019189969469, "learning_rate": 3.2098214771307057e-06, "loss": 0.3889, "step": 20924 }, { "epoch": 0.7454976218893066, "grad_norm": 1.3137427813950078, "learning_rate": 3.20897439230447e-06, "loss": 0.4887, "step": 20925 }, { "epoch": 0.7455332490158006, "grad_norm": 1.4554472470484507, "learning_rate": 3.208127397905487e-06, "loss": 0.5464, "step": 20926 }, { "epoch": 0.7455688761422947, "grad_norm": 1.1207488989792713, "learning_rate": 3.2072804939450385e-06, "loss": 0.4076, "step": 20927 }, { "epoch": 0.7456045032687889, "grad_norm": 1.352799395652862, "learning_rate": 3.206433680434394e-06, "loss": 0.4377, "step": 20928 }, { "epoch": 0.745640130395283, "grad_norm": 1.6318733012394395, "learning_rate": 3.2055869573848374e-06, "loss": 0.5142, "step": 20929 }, { "epoch": 0.7456757575217771, "grad_norm": 1.1729177762874055, "learning_rate": 3.204740324807636e-06, "loss": 0.3852, "step": 20930 }, { "epoch": 0.7457113846482712, "grad_norm": 1.6095180224778705, "learning_rate": 3.203893782714067e-06, "loss": 0.5235, "step": 20931 }, { "epoch": 0.7457470117747653, "grad_norm": 1.7540578009487338, "learning_rate": 3.203047331115402e-06, "loss": 0.6021, "step": 20932 }, { "epoch": 0.7457826389012594, "grad_norm": 1.2094387900945949, "learning_rate": 3.202200970022916e-06, "loss": 0.4693, "step": 20933 }, { "epoch": 0.7458182660277536, "grad_norm": 1.3262794700915999, "learning_rate": 3.201354699447873e-06, "loss": 0.3908, "step": 20934 }, { "epoch": 0.7458538931542477, "grad_norm": 1.2374286767027205, "learning_rate": 3.2005085194015463e-06, "loss": 0.4338, "step": 20935 }, { "epoch": 0.7458895202807417, "grad_norm": 1.2796329855264406, "learning_rate": 3.1996624298951982e-06, "loss": 0.4004, "step": 20936 }, { "epoch": 0.7459251474072359, "grad_norm": 1.2409360956294806, "learning_rate": 3.1988164309400993e-06, "loss": 0.4119, "step": 20937 }, { "epoch": 0.74596077453373, "grad_norm": 1.2069849101101113, "learning_rate": 3.197970522547512e-06, "loss": 0.4496, "step": 20938 }, { "epoch": 0.7459964016602241, "grad_norm": 1.7366517186256958, "learning_rate": 3.1971247047287013e-06, "loss": 0.5439, "step": 20939 }, { "epoch": 0.7460320287867183, "grad_norm": 1.483607714936178, "learning_rate": 3.196278977494934e-06, "loss": 0.5371, "step": 20940 }, { "epoch": 0.7460676559132123, "grad_norm": 1.218203388162057, "learning_rate": 3.1954333408574643e-06, "loss": 0.4398, "step": 20941 }, { "epoch": 0.7461032830397064, "grad_norm": 1.5552421381769899, "learning_rate": 3.1945877948275595e-06, "loss": 0.4572, "step": 20942 }, { "epoch": 0.7461389101662006, "grad_norm": 1.8108769450882154, "learning_rate": 3.19374233941647e-06, "loss": 0.5009, "step": 20943 }, { "epoch": 0.7461745372926947, "grad_norm": 1.255005101665949, "learning_rate": 3.19289697463546e-06, "loss": 0.4326, "step": 20944 }, { "epoch": 0.7462101644191887, "grad_norm": 1.879876368582751, "learning_rate": 3.192051700495784e-06, "loss": 0.5, "step": 20945 }, { "epoch": 0.7462457915456829, "grad_norm": 1.1956554537201358, "learning_rate": 3.1912065170087027e-06, "loss": 0.3931, "step": 20946 }, { "epoch": 0.746281418672177, "grad_norm": 2.7447410406886923, "learning_rate": 3.1903614241854606e-06, "loss": 0.5222, "step": 20947 }, { "epoch": 0.7463170457986711, "grad_norm": 1.3004252792731763, "learning_rate": 3.1895164220373167e-06, "loss": 0.4293, "step": 20948 }, { "epoch": 0.7463526729251653, "grad_norm": 1.6694520270876028, "learning_rate": 3.188671510575525e-06, "loss": 0.5429, "step": 20949 }, { "epoch": 0.7463883000516593, "grad_norm": 1.5147733708039004, "learning_rate": 3.1878266898113296e-06, "loss": 0.4113, "step": 20950 }, { "epoch": 0.7464239271781534, "grad_norm": 1.7168478991393752, "learning_rate": 3.186981959755987e-06, "loss": 0.4763, "step": 20951 }, { "epoch": 0.7464595543046476, "grad_norm": 1.3452817856726142, "learning_rate": 3.186137320420738e-06, "loss": 0.4449, "step": 20952 }, { "epoch": 0.7464951814311417, "grad_norm": 1.8091616542225433, "learning_rate": 3.1852927718168324e-06, "loss": 0.5636, "step": 20953 }, { "epoch": 0.7465308085576358, "grad_norm": 1.707854870843988, "learning_rate": 3.184448313955518e-06, "loss": 0.4723, "step": 20954 }, { "epoch": 0.74656643568413, "grad_norm": 1.198647182347232, "learning_rate": 3.1836039468480405e-06, "loss": 0.4104, "step": 20955 }, { "epoch": 0.746602062810624, "grad_norm": 1.9306036701828055, "learning_rate": 3.1827596705056383e-06, "loss": 0.4371, "step": 20956 }, { "epoch": 0.7466376899371181, "grad_norm": 1.302884000043577, "learning_rate": 3.1819154849395597e-06, "loss": 0.4115, "step": 20957 }, { "epoch": 0.7466733170636123, "grad_norm": 1.3735473575344035, "learning_rate": 3.1810713901610367e-06, "loss": 0.4046, "step": 20958 }, { "epoch": 0.7467089441901064, "grad_norm": 1.297155386473038, "learning_rate": 3.180227386181316e-06, "loss": 0.4436, "step": 20959 }, { "epoch": 0.7467445713166004, "grad_norm": 1.650692039997479, "learning_rate": 3.1793834730116337e-06, "loss": 0.4802, "step": 20960 }, { "epoch": 0.7467801984430946, "grad_norm": 1.204854228127575, "learning_rate": 3.1785396506632283e-06, "loss": 0.4266, "step": 20961 }, { "epoch": 0.7468158255695887, "grad_norm": 1.2033036367107959, "learning_rate": 3.177695919147339e-06, "loss": 0.4623, "step": 20962 }, { "epoch": 0.7468514526960828, "grad_norm": 1.1893804854167764, "learning_rate": 3.176852278475192e-06, "loss": 0.4463, "step": 20963 }, { "epoch": 0.746887079822577, "grad_norm": 1.6066487132581717, "learning_rate": 3.176008728658031e-06, "loss": 0.5506, "step": 20964 }, { "epoch": 0.746922706949071, "grad_norm": 1.4354372562832667, "learning_rate": 3.1751652697070788e-06, "loss": 0.444, "step": 20965 }, { "epoch": 0.7469583340755651, "grad_norm": 1.7239992888785622, "learning_rate": 3.174321901633576e-06, "loss": 0.5965, "step": 20966 }, { "epoch": 0.7469939612020593, "grad_norm": 1.5874896414017838, "learning_rate": 3.173478624448744e-06, "loss": 0.5765, "step": 20967 }, { "epoch": 0.7470295883285534, "grad_norm": 1.2444910418972428, "learning_rate": 3.1726354381638157e-06, "loss": 0.4084, "step": 20968 }, { "epoch": 0.7470652154550474, "grad_norm": 1.1059853754933517, "learning_rate": 3.171792342790019e-06, "loss": 0.4172, "step": 20969 }, { "epoch": 0.7471008425815416, "grad_norm": 1.6817476932282343, "learning_rate": 3.1709493383385824e-06, "loss": 0.5135, "step": 20970 }, { "epoch": 0.7471364697080357, "grad_norm": 1.3022107027366532, "learning_rate": 3.170106424820725e-06, "loss": 0.4267, "step": 20971 }, { "epoch": 0.7471720968345298, "grad_norm": 1.230429283850074, "learning_rate": 3.169263602247675e-06, "loss": 0.5415, "step": 20972 }, { "epoch": 0.747207723961024, "grad_norm": 1.3514959137466125, "learning_rate": 3.1684208706306572e-06, "loss": 0.3854, "step": 20973 }, { "epoch": 0.747243351087518, "grad_norm": 1.3632617248560897, "learning_rate": 3.167578229980889e-06, "loss": 0.5319, "step": 20974 }, { "epoch": 0.7472789782140121, "grad_norm": 1.1906773236359733, "learning_rate": 3.1667356803095905e-06, "loss": 0.4081, "step": 20975 }, { "epoch": 0.7473146053405063, "grad_norm": 1.206977082146156, "learning_rate": 3.1658932216279825e-06, "loss": 0.4054, "step": 20976 }, { "epoch": 0.7473502324670004, "grad_norm": 1.4336270023612319, "learning_rate": 3.165050853947287e-06, "loss": 0.4141, "step": 20977 }, { "epoch": 0.7473858595934945, "grad_norm": 1.295164193413551, "learning_rate": 3.1642085772787134e-06, "loss": 0.4209, "step": 20978 }, { "epoch": 0.7474214867199886, "grad_norm": 2.058851363779514, "learning_rate": 3.1633663916334843e-06, "loss": 0.5479, "step": 20979 }, { "epoch": 0.7474571138464827, "grad_norm": 1.2476206943060157, "learning_rate": 3.1625242970228065e-06, "loss": 0.3815, "step": 20980 }, { "epoch": 0.7474927409729768, "grad_norm": 1.4245863154301976, "learning_rate": 3.1616822934578994e-06, "loss": 0.41, "step": 20981 }, { "epoch": 0.747528368099471, "grad_norm": 1.1959904674057, "learning_rate": 3.1608403809499667e-06, "loss": 0.4035, "step": 20982 }, { "epoch": 0.7475639952259651, "grad_norm": 1.4755306762244789, "learning_rate": 3.1599985595102266e-06, "loss": 0.5426, "step": 20983 }, { "epoch": 0.7475996223524591, "grad_norm": 1.2343214300604586, "learning_rate": 3.1591568291498907e-06, "loss": 0.3719, "step": 20984 }, { "epoch": 0.7476352494789533, "grad_norm": 1.2338954638398891, "learning_rate": 3.1583151898801602e-06, "loss": 0.4424, "step": 20985 }, { "epoch": 0.7476708766054474, "grad_norm": 1.2245465338158286, "learning_rate": 3.1574736417122474e-06, "loss": 0.3887, "step": 20986 }, { "epoch": 0.7477065037319415, "grad_norm": 1.4627493374135314, "learning_rate": 3.1566321846573533e-06, "loss": 0.5344, "step": 20987 }, { "epoch": 0.7477421308584357, "grad_norm": 1.2941601984631863, "learning_rate": 3.1557908187266872e-06, "loss": 0.5394, "step": 20988 }, { "epoch": 0.7477777579849297, "grad_norm": 1.3651837556720123, "learning_rate": 3.154949543931448e-06, "loss": 0.4593, "step": 20989 }, { "epoch": 0.7478133851114238, "grad_norm": 1.5488998353165742, "learning_rate": 3.1541083602828394e-06, "loss": 0.5852, "step": 20990 }, { "epoch": 0.747849012237918, "grad_norm": 1.3975702959026135, "learning_rate": 3.153267267792063e-06, "loss": 0.525, "step": 20991 }, { "epoch": 0.7478846393644121, "grad_norm": 1.8371413216127237, "learning_rate": 3.152426266470322e-06, "loss": 0.5369, "step": 20992 }, { "epoch": 0.7479202664909061, "grad_norm": 1.2815769674053543, "learning_rate": 3.151585356328808e-06, "loss": 0.4551, "step": 20993 }, { "epoch": 0.7479558936174003, "grad_norm": 1.589249232029958, "learning_rate": 3.1507445373787214e-06, "loss": 0.5358, "step": 20994 }, { "epoch": 0.7479915207438944, "grad_norm": 1.2510249158292983, "learning_rate": 3.1499038096312617e-06, "loss": 0.4094, "step": 20995 }, { "epoch": 0.7480271478703885, "grad_norm": 1.2252013807480147, "learning_rate": 3.1490631730976186e-06, "loss": 0.3941, "step": 20996 }, { "epoch": 0.7480627749968827, "grad_norm": 1.2805561442733895, "learning_rate": 3.1482226277889873e-06, "loss": 0.4137, "step": 20997 }, { "epoch": 0.7480984021233767, "grad_norm": 1.3923445823655403, "learning_rate": 3.14738217371656e-06, "loss": 0.5118, "step": 20998 }, { "epoch": 0.7481340292498708, "grad_norm": 1.217968371788553, "learning_rate": 3.1465418108915336e-06, "loss": 0.4347, "step": 20999 }, { "epoch": 0.748169656376365, "grad_norm": 1.1134580196616328, "learning_rate": 3.145701539325089e-06, "loss": 0.3972, "step": 21000 }, { "epoch": 0.7482052835028591, "grad_norm": 1.2120182392493954, "learning_rate": 3.144861359028423e-06, "loss": 0.4291, "step": 21001 }, { "epoch": 0.7482409106293532, "grad_norm": 1.2953298266615119, "learning_rate": 3.144021270012715e-06, "loss": 0.4211, "step": 21002 }, { "epoch": 0.7482765377558473, "grad_norm": 1.1633365638281938, "learning_rate": 3.1431812722891598e-06, "loss": 0.4095, "step": 21003 }, { "epoch": 0.7483121648823414, "grad_norm": 1.6060917346426826, "learning_rate": 3.142341365868936e-06, "loss": 0.5201, "step": 21004 }, { "epoch": 0.7483477920088355, "grad_norm": 1.4693963443654938, "learning_rate": 3.1415015507632294e-06, "loss": 0.4897, "step": 21005 }, { "epoch": 0.7483834191353297, "grad_norm": 1.1865193707540294, "learning_rate": 3.1406618269832232e-06, "loss": 0.4152, "step": 21006 }, { "epoch": 0.7484190462618238, "grad_norm": 1.5570371400632317, "learning_rate": 3.1398221945401e-06, "loss": 0.4532, "step": 21007 }, { "epoch": 0.7484546733883178, "grad_norm": 1.5325244772205855, "learning_rate": 3.1389826534450417e-06, "loss": 0.5413, "step": 21008 }, { "epoch": 0.748490300514812, "grad_norm": 1.2622720638052183, "learning_rate": 3.138143203709223e-06, "loss": 0.4165, "step": 21009 }, { "epoch": 0.7485259276413061, "grad_norm": 1.2042278665087687, "learning_rate": 3.1373038453438254e-06, "loss": 0.3868, "step": 21010 }, { "epoch": 0.7485615547678002, "grad_norm": 1.4262929614050996, "learning_rate": 3.136464578360021e-06, "loss": 0.4751, "step": 21011 }, { "epoch": 0.7485971818942944, "grad_norm": 1.381419221062674, "learning_rate": 3.135625402768989e-06, "loss": 0.3861, "step": 21012 }, { "epoch": 0.7486328090207884, "grad_norm": 1.9743898863155906, "learning_rate": 3.1347863185819016e-06, "loss": 0.5228, "step": 21013 }, { "epoch": 0.7486684361472825, "grad_norm": 1.2031851761111962, "learning_rate": 3.1339473258099374e-06, "loss": 0.3736, "step": 21014 }, { "epoch": 0.7487040632737767, "grad_norm": 1.300055081964533, "learning_rate": 3.1331084244642585e-06, "loss": 0.3834, "step": 21015 }, { "epoch": 0.7487396904002708, "grad_norm": 1.4580724520427168, "learning_rate": 3.132269614556045e-06, "loss": 0.5428, "step": 21016 }, { "epoch": 0.7487753175267648, "grad_norm": 1.1590468669109215, "learning_rate": 3.131430896096459e-06, "loss": 0.3907, "step": 21017 }, { "epoch": 0.748810944653259, "grad_norm": 1.3323525531708083, "learning_rate": 3.1305922690966705e-06, "loss": 0.3972, "step": 21018 }, { "epoch": 0.7488465717797531, "grad_norm": 1.2834684669711036, "learning_rate": 3.1297537335678507e-06, "loss": 0.4338, "step": 21019 }, { "epoch": 0.7488821989062472, "grad_norm": 1.4639370105891563, "learning_rate": 3.128915289521156e-06, "loss": 0.5123, "step": 21020 }, { "epoch": 0.7489178260327414, "grad_norm": 1.4970178728405619, "learning_rate": 3.1280769369677635e-06, "loss": 0.5449, "step": 21021 }, { "epoch": 0.7489534531592354, "grad_norm": 1.1116453837959845, "learning_rate": 3.1272386759188254e-06, "loss": 0.3841, "step": 21022 }, { "epoch": 0.7489890802857295, "grad_norm": 1.0588255071072972, "learning_rate": 3.126400506385513e-06, "loss": 0.3771, "step": 21023 }, { "epoch": 0.7490247074122237, "grad_norm": 1.217343158287542, "learning_rate": 3.1255624283789776e-06, "loss": 0.4435, "step": 21024 }, { "epoch": 0.7490603345387178, "grad_norm": 1.2150881786508754, "learning_rate": 3.124724441910387e-06, "loss": 0.4321, "step": 21025 }, { "epoch": 0.7490959616652119, "grad_norm": 1.6431600242214361, "learning_rate": 3.123886546990893e-06, "loss": 0.5159, "step": 21026 }, { "epoch": 0.749131588791706, "grad_norm": 1.2329130823921444, "learning_rate": 3.123048743631656e-06, "loss": 0.4137, "step": 21027 }, { "epoch": 0.7491672159182001, "grad_norm": 1.2063233476805468, "learning_rate": 3.12221103184383e-06, "loss": 0.3874, "step": 21028 }, { "epoch": 0.7492028430446942, "grad_norm": 2.1457558595804036, "learning_rate": 3.121373411638573e-06, "loss": 0.5486, "step": 21029 }, { "epoch": 0.7492384701711884, "grad_norm": 1.8552543827668986, "learning_rate": 3.1205358830270405e-06, "loss": 0.5207, "step": 21030 }, { "epoch": 0.7492740972976825, "grad_norm": 1.229159989532858, "learning_rate": 3.119698446020377e-06, "loss": 0.4557, "step": 21031 }, { "epoch": 0.7493097244241765, "grad_norm": 1.1862536081810597, "learning_rate": 3.118861100629741e-06, "loss": 0.4701, "step": 21032 }, { "epoch": 0.7493453515506707, "grad_norm": 1.1839535735611737, "learning_rate": 3.118023846866276e-06, "loss": 0.4157, "step": 21033 }, { "epoch": 0.7493809786771648, "grad_norm": 1.5330684114430393, "learning_rate": 3.1171866847411335e-06, "loss": 0.5037, "step": 21034 }, { "epoch": 0.7494166058036589, "grad_norm": 1.4482733442597597, "learning_rate": 3.1163496142654613e-06, "loss": 0.5679, "step": 21035 }, { "epoch": 0.7494522329301531, "grad_norm": 1.1694857680554254, "learning_rate": 3.1155126354504096e-06, "loss": 0.3935, "step": 21036 }, { "epoch": 0.7494878600566471, "grad_norm": 1.4716480951784772, "learning_rate": 3.1146757483071144e-06, "loss": 0.5191, "step": 21037 }, { "epoch": 0.7495234871831412, "grad_norm": 1.2270545211849138, "learning_rate": 3.1138389528467284e-06, "loss": 0.4184, "step": 21038 }, { "epoch": 0.7495591143096354, "grad_norm": 1.207700134505145, "learning_rate": 3.1130022490803856e-06, "loss": 0.3592, "step": 21039 }, { "epoch": 0.7495947414361295, "grad_norm": 1.7143425363000173, "learning_rate": 3.112165637019232e-06, "loss": 0.5305, "step": 21040 }, { "epoch": 0.7496303685626236, "grad_norm": 1.2737432426079094, "learning_rate": 3.111329116674411e-06, "loss": 0.4494, "step": 21041 }, { "epoch": 0.7496659956891177, "grad_norm": 1.2711379313308164, "learning_rate": 3.110492688057054e-06, "loss": 0.4425, "step": 21042 }, { "epoch": 0.7497016228156118, "grad_norm": 1.5056719466671928, "learning_rate": 3.109656351178302e-06, "loss": 0.5134, "step": 21043 }, { "epoch": 0.7497372499421059, "grad_norm": 1.0700349519013013, "learning_rate": 3.1088201060492927e-06, "loss": 0.3743, "step": 21044 }, { "epoch": 0.7497728770686001, "grad_norm": 1.1161676699550513, "learning_rate": 3.1079839526811627e-06, "loss": 0.3892, "step": 21045 }, { "epoch": 0.7498085041950941, "grad_norm": 1.5459171565594005, "learning_rate": 3.1071478910850405e-06, "loss": 0.5001, "step": 21046 }, { "epoch": 0.7498441313215882, "grad_norm": 1.6061777733259786, "learning_rate": 3.106311921272066e-06, "loss": 0.5141, "step": 21047 }, { "epoch": 0.7498797584480824, "grad_norm": 1.6101476529352199, "learning_rate": 3.1054760432533626e-06, "loss": 0.5487, "step": 21048 }, { "epoch": 0.7499153855745765, "grad_norm": 1.2427747075572897, "learning_rate": 3.1046402570400648e-06, "loss": 0.4172, "step": 21049 }, { "epoch": 0.7499510127010706, "grad_norm": 2.0496189885817206, "learning_rate": 3.103804562643302e-06, "loss": 0.5338, "step": 21050 }, { "epoch": 0.7499866398275647, "grad_norm": 1.397323445404193, "learning_rate": 3.102968960074204e-06, "loss": 0.4983, "step": 21051 }, { "epoch": 0.7500222669540588, "grad_norm": 1.5088935268405974, "learning_rate": 3.1021334493438924e-06, "loss": 0.5244, "step": 21052 }, { "epoch": 0.7500578940805529, "grad_norm": 1.4605930169818235, "learning_rate": 3.1012980304634953e-06, "loss": 0.4775, "step": 21053 }, { "epoch": 0.7500935212070471, "grad_norm": 1.312031909562286, "learning_rate": 3.1004627034441393e-06, "loss": 0.417, "step": 21054 }, { "epoch": 0.7501291483335412, "grad_norm": 1.1956494368795576, "learning_rate": 3.099627468296943e-06, "loss": 0.3906, "step": 21055 }, { "epoch": 0.7501647754600352, "grad_norm": 1.0682739530737746, "learning_rate": 3.098792325033032e-06, "loss": 0.3884, "step": 21056 }, { "epoch": 0.7502004025865294, "grad_norm": 2.004975676776453, "learning_rate": 3.0979572736635187e-06, "loss": 0.484, "step": 21057 }, { "epoch": 0.7502360297130235, "grad_norm": 1.453310301825049, "learning_rate": 3.0971223141995354e-06, "loss": 0.4604, "step": 21058 }, { "epoch": 0.7502716568395176, "grad_norm": 1.186655637440828, "learning_rate": 3.09628744665219e-06, "loss": 0.4391, "step": 21059 }, { "epoch": 0.7503072839660118, "grad_norm": 1.7915897339748819, "learning_rate": 3.095452671032606e-06, "loss": 0.4799, "step": 21060 }, { "epoch": 0.7503429110925058, "grad_norm": 1.4236488992485636, "learning_rate": 3.0946179873518923e-06, "loss": 0.5399, "step": 21061 }, { "epoch": 0.7503785382189999, "grad_norm": 1.4741133617968116, "learning_rate": 3.0937833956211694e-06, "loss": 0.5054, "step": 21062 }, { "epoch": 0.7504141653454941, "grad_norm": 1.2077441417526733, "learning_rate": 3.092948895851545e-06, "loss": 0.4029, "step": 21063 }, { "epoch": 0.7504497924719882, "grad_norm": 1.5212544439965996, "learning_rate": 3.0921144880541343e-06, "loss": 0.5126, "step": 21064 }, { "epoch": 0.7504854195984823, "grad_norm": 1.1863291685957575, "learning_rate": 3.0912801722400466e-06, "loss": 0.4179, "step": 21065 }, { "epoch": 0.7505210467249764, "grad_norm": 1.7220856596312557, "learning_rate": 3.090445948420392e-06, "loss": 0.5086, "step": 21066 }, { "epoch": 0.7505566738514705, "grad_norm": 1.3103807758073238, "learning_rate": 3.089611816606284e-06, "loss": 0.431, "step": 21067 }, { "epoch": 0.7505923009779646, "grad_norm": 1.3395837126393078, "learning_rate": 3.08877777680882e-06, "loss": 0.4681, "step": 21068 }, { "epoch": 0.7506279281044588, "grad_norm": 1.2531869730998049, "learning_rate": 3.087943829039114e-06, "loss": 0.4035, "step": 21069 }, { "epoch": 0.7506635552309529, "grad_norm": 1.3685909329602601, "learning_rate": 3.0871099733082642e-06, "loss": 0.4464, "step": 21070 }, { "epoch": 0.7506991823574469, "grad_norm": 1.3925488178054342, "learning_rate": 3.086276209627377e-06, "loss": 0.4564, "step": 21071 }, { "epoch": 0.7507348094839411, "grad_norm": 1.3438573468106092, "learning_rate": 3.0854425380075548e-06, "loss": 0.4129, "step": 21072 }, { "epoch": 0.7507704366104352, "grad_norm": 1.688722422194701, "learning_rate": 3.0846089584599005e-06, "loss": 0.4941, "step": 21073 }, { "epoch": 0.7508060637369293, "grad_norm": 1.7948413723305412, "learning_rate": 3.083775470995509e-06, "loss": 0.5511, "step": 21074 }, { "epoch": 0.7508416908634234, "grad_norm": 1.5544558840091802, "learning_rate": 3.0829420756254813e-06, "loss": 0.5316, "step": 21075 }, { "epoch": 0.7508773179899175, "grad_norm": 1.293872959835777, "learning_rate": 3.0821087723609167e-06, "loss": 0.4124, "step": 21076 }, { "epoch": 0.7509129451164116, "grad_norm": 1.3534690042714443, "learning_rate": 3.0812755612129075e-06, "loss": 0.5322, "step": 21077 }, { "epoch": 0.7509485722429058, "grad_norm": 1.2177437107101212, "learning_rate": 3.0804424421925515e-06, "loss": 0.4061, "step": 21078 }, { "epoch": 0.7509841993693999, "grad_norm": 1.3791498816921017, "learning_rate": 3.079609415310939e-06, "loss": 0.6356, "step": 21079 }, { "epoch": 0.7510198264958939, "grad_norm": 1.1771958645571778, "learning_rate": 3.078776480579164e-06, "loss": 0.4261, "step": 21080 }, { "epoch": 0.7510554536223881, "grad_norm": 1.3114082829944775, "learning_rate": 3.077943638008318e-06, "loss": 0.43, "step": 21081 }, { "epoch": 0.7510910807488822, "grad_norm": 1.2859932968035706, "learning_rate": 3.0771108876094935e-06, "loss": 0.3842, "step": 21082 }, { "epoch": 0.7511267078753763, "grad_norm": 1.2617585077241058, "learning_rate": 3.076278229393773e-06, "loss": 0.3424, "step": 21083 }, { "epoch": 0.7511623350018705, "grad_norm": 1.553884473545649, "learning_rate": 3.0754456633722506e-06, "loss": 0.5144, "step": 21084 }, { "epoch": 0.7511979621283645, "grad_norm": 1.1126690335179577, "learning_rate": 3.074613189556006e-06, "loss": 0.3649, "step": 21085 }, { "epoch": 0.7512335892548586, "grad_norm": 1.7388579938366786, "learning_rate": 3.0737808079561283e-06, "loss": 0.573, "step": 21086 }, { "epoch": 0.7512692163813528, "grad_norm": 1.683751293206187, "learning_rate": 3.0729485185836983e-06, "loss": 0.5356, "step": 21087 }, { "epoch": 0.7513048435078469, "grad_norm": 1.0905262289471065, "learning_rate": 3.0721163214498016e-06, "loss": 0.4236, "step": 21088 }, { "epoch": 0.751340470634341, "grad_norm": 1.3128204060023512, "learning_rate": 3.071284216565522e-06, "loss": 0.4325, "step": 21089 }, { "epoch": 0.7513760977608351, "grad_norm": 1.3203997349506853, "learning_rate": 3.070452203941933e-06, "loss": 0.4351, "step": 21090 }, { "epoch": 0.7514117248873292, "grad_norm": 1.1711545283104292, "learning_rate": 3.069620283590118e-06, "loss": 0.4142, "step": 21091 }, { "epoch": 0.7514473520138233, "grad_norm": 1.3060737945032403, "learning_rate": 3.0687884555211513e-06, "loss": 0.4246, "step": 21092 }, { "epoch": 0.7514829791403175, "grad_norm": 1.2811170660945115, "learning_rate": 3.0679567197461135e-06, "loss": 0.4267, "step": 21093 }, { "epoch": 0.7515186062668116, "grad_norm": 1.3696348306034267, "learning_rate": 3.0671250762760728e-06, "loss": 0.4356, "step": 21094 }, { "epoch": 0.7515542333933056, "grad_norm": 1.5721487053219931, "learning_rate": 3.0662935251221094e-06, "loss": 0.5223, "step": 21095 }, { "epoch": 0.7515898605197998, "grad_norm": 1.6053337796505813, "learning_rate": 3.0654620662952926e-06, "loss": 0.3614, "step": 21096 }, { "epoch": 0.7516254876462939, "grad_norm": 1.3624450172284877, "learning_rate": 3.0646306998066997e-06, "loss": 0.4135, "step": 21097 }, { "epoch": 0.751661114772788, "grad_norm": 1.1678040399039082, "learning_rate": 3.063799425667393e-06, "loss": 0.3924, "step": 21098 }, { "epoch": 0.7516967418992822, "grad_norm": 1.6149190677471037, "learning_rate": 3.0629682438884455e-06, "loss": 0.5397, "step": 21099 }, { "epoch": 0.7517323690257762, "grad_norm": 1.5232908351629741, "learning_rate": 3.0621371544809264e-06, "loss": 0.4749, "step": 21100 }, { "epoch": 0.7517679961522703, "grad_norm": 1.059182246756644, "learning_rate": 3.061306157455899e-06, "loss": 0.396, "step": 21101 }, { "epoch": 0.7518036232787645, "grad_norm": 1.433723279772112, "learning_rate": 3.060475252824429e-06, "loss": 0.386, "step": 21102 }, { "epoch": 0.7518392504052586, "grad_norm": 1.372784819915144, "learning_rate": 3.059644440597581e-06, "loss": 0.4269, "step": 21103 }, { "epoch": 0.7518748775317526, "grad_norm": 1.4794734162201546, "learning_rate": 3.0588137207864223e-06, "loss": 0.5273, "step": 21104 }, { "epoch": 0.7519105046582468, "grad_norm": 1.2735714733852204, "learning_rate": 3.0579830934020062e-06, "loss": 0.4048, "step": 21105 }, { "epoch": 0.7519461317847409, "grad_norm": 1.1646875923629818, "learning_rate": 3.057152558455402e-06, "loss": 0.4515, "step": 21106 }, { "epoch": 0.751981758911235, "grad_norm": 1.4741262876800174, "learning_rate": 3.0563221159576594e-06, "loss": 0.4764, "step": 21107 }, { "epoch": 0.7520173860377292, "grad_norm": 3.257616998721542, "learning_rate": 3.055491765919846e-06, "loss": 0.5402, "step": 21108 }, { "epoch": 0.7520530131642232, "grad_norm": 1.6779190397961532, "learning_rate": 3.0546615083530064e-06, "loss": 0.4915, "step": 21109 }, { "epoch": 0.7520886402907173, "grad_norm": 1.7521105786782687, "learning_rate": 3.0538313432682064e-06, "loss": 0.4891, "step": 21110 }, { "epoch": 0.7521242674172115, "grad_norm": 1.5133754334537826, "learning_rate": 3.053001270676501e-06, "loss": 0.4841, "step": 21111 }, { "epoch": 0.7521598945437056, "grad_norm": 1.206017278529195, "learning_rate": 3.0521712905889357e-06, "loss": 0.4201, "step": 21112 }, { "epoch": 0.7521955216701997, "grad_norm": 1.1925163814193729, "learning_rate": 3.051341403016569e-06, "loss": 0.4346, "step": 21113 }, { "epoch": 0.7522311487966938, "grad_norm": 1.5253357189851342, "learning_rate": 3.0505116079704454e-06, "loss": 0.4582, "step": 21114 }, { "epoch": 0.7522667759231879, "grad_norm": 1.4715115598693629, "learning_rate": 3.04968190546162e-06, "loss": 0.5357, "step": 21115 }, { "epoch": 0.752302403049682, "grad_norm": 1.630866386107665, "learning_rate": 3.0488522955011356e-06, "loss": 0.4635, "step": 21116 }, { "epoch": 0.7523380301761762, "grad_norm": 1.2111147588152753, "learning_rate": 3.0480227781000406e-06, "loss": 0.3889, "step": 21117 }, { "epoch": 0.7523736573026703, "grad_norm": 1.3342079512847602, "learning_rate": 3.047193353269382e-06, "loss": 0.3914, "step": 21118 }, { "epoch": 0.7524092844291643, "grad_norm": 1.2566923964166263, "learning_rate": 3.046364021020206e-06, "loss": 0.4394, "step": 21119 }, { "epoch": 0.7524449115556585, "grad_norm": 1.1658968634046114, "learning_rate": 3.045534781363552e-06, "loss": 0.4137, "step": 21120 }, { "epoch": 0.7524805386821526, "grad_norm": 1.20883747418117, "learning_rate": 3.044705634310462e-06, "loss": 0.4031, "step": 21121 }, { "epoch": 0.7525161658086467, "grad_norm": 1.519225297417703, "learning_rate": 3.0438765798719807e-06, "loss": 0.4251, "step": 21122 }, { "epoch": 0.7525517929351409, "grad_norm": 1.0844995811138705, "learning_rate": 3.0430476180591416e-06, "loss": 0.4294, "step": 21123 }, { "epoch": 0.7525874200616349, "grad_norm": 1.3552945643665086, "learning_rate": 3.042218748882987e-06, "loss": 0.4274, "step": 21124 }, { "epoch": 0.752623047188129, "grad_norm": 1.609749494164575, "learning_rate": 3.0413899723545526e-06, "loss": 0.5826, "step": 21125 }, { "epoch": 0.7526586743146232, "grad_norm": 1.9065075299049372, "learning_rate": 3.040561288484878e-06, "loss": 0.4923, "step": 21126 }, { "epoch": 0.7526943014411173, "grad_norm": 1.3252842441272443, "learning_rate": 3.0397326972849895e-06, "loss": 0.4563, "step": 21127 }, { "epoch": 0.7527299285676113, "grad_norm": 1.5243613188929457, "learning_rate": 3.0389041987659296e-06, "loss": 0.4855, "step": 21128 }, { "epoch": 0.7527655556941055, "grad_norm": 1.0970588647609625, "learning_rate": 3.0380757929387226e-06, "loss": 0.3527, "step": 21129 }, { "epoch": 0.7528011828205996, "grad_norm": 1.1097799785694842, "learning_rate": 3.037247479814406e-06, "loss": 0.4183, "step": 21130 }, { "epoch": 0.7528368099470937, "grad_norm": 1.180201013458172, "learning_rate": 3.036419259404002e-06, "loss": 0.3861, "step": 21131 }, { "epoch": 0.7528724370735879, "grad_norm": 1.2831817157548695, "learning_rate": 3.0355911317185392e-06, "loss": 0.4355, "step": 21132 }, { "epoch": 0.7529080642000819, "grad_norm": 1.4527513115491653, "learning_rate": 3.0347630967690556e-06, "loss": 0.5465, "step": 21133 }, { "epoch": 0.752943691326576, "grad_norm": 1.3248169624046962, "learning_rate": 3.0339351545665665e-06, "loss": 0.3911, "step": 21134 }, { "epoch": 0.7529793184530702, "grad_norm": 1.4894142786353235, "learning_rate": 3.0331073051221027e-06, "loss": 0.6136, "step": 21135 }, { "epoch": 0.7530149455795643, "grad_norm": 2.000110643099056, "learning_rate": 3.0322795484466817e-06, "loss": 0.4252, "step": 21136 }, { "epoch": 0.7530505727060584, "grad_norm": 1.1605098062862387, "learning_rate": 3.031451884551331e-06, "loss": 0.4469, "step": 21137 }, { "epoch": 0.7530861998325525, "grad_norm": 1.1177734961368433, "learning_rate": 3.0306243134470668e-06, "loss": 0.4057, "step": 21138 }, { "epoch": 0.7531218269590466, "grad_norm": 1.8360704719192704, "learning_rate": 3.0297968351449115e-06, "loss": 0.4717, "step": 21139 }, { "epoch": 0.7531574540855407, "grad_norm": 1.81683841939414, "learning_rate": 3.028969449655883e-06, "loss": 0.4983, "step": 21140 }, { "epoch": 0.7531930812120349, "grad_norm": 1.1766108902311287, "learning_rate": 3.028142156991001e-06, "loss": 0.4579, "step": 21141 }, { "epoch": 0.753228708338529, "grad_norm": 1.1864519607727453, "learning_rate": 3.027314957161277e-06, "loss": 0.4389, "step": 21142 }, { "epoch": 0.753264335465023, "grad_norm": 1.5083459773548797, "learning_rate": 3.0264878501777307e-06, "loss": 0.5411, "step": 21143 }, { "epoch": 0.7532999625915172, "grad_norm": 2.36097802573371, "learning_rate": 3.0256608360513708e-06, "loss": 0.4864, "step": 21144 }, { "epoch": 0.7533355897180113, "grad_norm": 1.212730063785661, "learning_rate": 3.0248339147932106e-06, "loss": 0.4037, "step": 21145 }, { "epoch": 0.7533712168445054, "grad_norm": 1.6708254735702956, "learning_rate": 3.024007086414261e-06, "loss": 0.4904, "step": 21146 }, { "epoch": 0.7534068439709996, "grad_norm": 1.2920086723779542, "learning_rate": 3.0231803509255353e-06, "loss": 0.4148, "step": 21147 }, { "epoch": 0.7534424710974936, "grad_norm": 1.380244403430446, "learning_rate": 3.022353708338042e-06, "loss": 0.5385, "step": 21148 }, { "epoch": 0.7534780982239877, "grad_norm": 1.1950179291931282, "learning_rate": 3.0215271586627824e-06, "loss": 0.4056, "step": 21149 }, { "epoch": 0.7535137253504819, "grad_norm": 1.0531772682415748, "learning_rate": 3.0207007019107705e-06, "loss": 0.3474, "step": 21150 }, { "epoch": 0.753549352476976, "grad_norm": 1.2681909122903199, "learning_rate": 3.019874338093003e-06, "loss": 0.4155, "step": 21151 }, { "epoch": 0.75358497960347, "grad_norm": 1.239459600799663, "learning_rate": 3.01904806722049e-06, "loss": 0.4298, "step": 21152 }, { "epoch": 0.7536206067299642, "grad_norm": 1.4310367011256204, "learning_rate": 3.0182218893042304e-06, "loss": 0.5391, "step": 21153 }, { "epoch": 0.7536562338564583, "grad_norm": 1.382803981264993, "learning_rate": 3.017395804355224e-06, "loss": 0.4524, "step": 21154 }, { "epoch": 0.7536918609829524, "grad_norm": 1.9107143213511868, "learning_rate": 3.0165698123844744e-06, "loss": 0.5116, "step": 21155 }, { "epoch": 0.7537274881094466, "grad_norm": 1.2829126070169619, "learning_rate": 3.0157439134029786e-06, "loss": 0.5102, "step": 21156 }, { "epoch": 0.7537631152359406, "grad_norm": 1.1499843679729167, "learning_rate": 3.014918107421738e-06, "loss": 0.4527, "step": 21157 }, { "epoch": 0.7537987423624347, "grad_norm": 1.2829247705545699, "learning_rate": 3.0140923944517407e-06, "loss": 0.5115, "step": 21158 }, { "epoch": 0.7538343694889289, "grad_norm": 1.6325214549375415, "learning_rate": 3.01326677450399e-06, "loss": 0.5174, "step": 21159 }, { "epoch": 0.753869996615423, "grad_norm": 1.1206519487692714, "learning_rate": 3.0124412475894726e-06, "loss": 0.3539, "step": 21160 }, { "epoch": 0.753905623741917, "grad_norm": 1.1824993900038552, "learning_rate": 3.011615813719184e-06, "loss": 0.4095, "step": 21161 }, { "epoch": 0.7539412508684112, "grad_norm": 1.2100318234202723, "learning_rate": 3.0107904729041147e-06, "loss": 0.4285, "step": 21162 }, { "epoch": 0.7539768779949053, "grad_norm": 1.2404826134318088, "learning_rate": 3.009965225155259e-06, "loss": 0.4146, "step": 21163 }, { "epoch": 0.7540125051213994, "grad_norm": 1.2994247104184848, "learning_rate": 3.0091400704835995e-06, "loss": 0.4747, "step": 21164 }, { "epoch": 0.7540481322478936, "grad_norm": 1.6378099293799233, "learning_rate": 3.008315008900129e-06, "loss": 0.6061, "step": 21165 }, { "epoch": 0.7540837593743877, "grad_norm": 1.5423982065752846, "learning_rate": 3.007490040415827e-06, "loss": 0.5396, "step": 21166 }, { "epoch": 0.7541193865008817, "grad_norm": 1.2345627960874148, "learning_rate": 3.0066651650416834e-06, "loss": 0.3801, "step": 21167 }, { "epoch": 0.7541550136273759, "grad_norm": 1.4141601996298407, "learning_rate": 3.005840382788685e-06, "loss": 0.4344, "step": 21168 }, { "epoch": 0.75419064075387, "grad_norm": 1.2584399055870046, "learning_rate": 3.0050156936678064e-06, "loss": 0.4144, "step": 21169 }, { "epoch": 0.7542262678803641, "grad_norm": 1.5766458547594777, "learning_rate": 3.0041910976900335e-06, "loss": 0.5595, "step": 21170 }, { "epoch": 0.7542618950068583, "grad_norm": 1.8076923287940365, "learning_rate": 3.003366594866345e-06, "loss": 0.5524, "step": 21171 }, { "epoch": 0.7542975221333523, "grad_norm": 1.6507048783966132, "learning_rate": 3.002542185207725e-06, "loss": 0.4406, "step": 21172 }, { "epoch": 0.7543331492598464, "grad_norm": 1.5088443791028805, "learning_rate": 3.0017178687251424e-06, "loss": 0.5224, "step": 21173 }, { "epoch": 0.7543687763863406, "grad_norm": 1.321701400422692, "learning_rate": 3.000893645429582e-06, "loss": 0.3971, "step": 21174 }, { "epoch": 0.7544044035128347, "grad_norm": 1.1719544212884443, "learning_rate": 3.0000695153320114e-06, "loss": 0.4061, "step": 21175 }, { "epoch": 0.7544400306393287, "grad_norm": 1.2189543291164033, "learning_rate": 2.999245478443408e-06, "loss": 0.4254, "step": 21176 }, { "epoch": 0.7544756577658229, "grad_norm": 1.2706920875036458, "learning_rate": 2.9984215347747437e-06, "loss": 0.3769, "step": 21177 }, { "epoch": 0.754511284892317, "grad_norm": 1.8564254022243292, "learning_rate": 2.9975976843369946e-06, "loss": 0.5444, "step": 21178 }, { "epoch": 0.7545469120188111, "grad_norm": 1.6703264145452217, "learning_rate": 2.9967739271411235e-06, "loss": 0.4985, "step": 21179 }, { "epoch": 0.7545825391453053, "grad_norm": 1.8084732669994155, "learning_rate": 2.995950263198102e-06, "loss": 0.4927, "step": 21180 }, { "epoch": 0.7546181662717993, "grad_norm": 1.5388208797586491, "learning_rate": 2.9951266925189006e-06, "loss": 0.5437, "step": 21181 }, { "epoch": 0.7546537933982934, "grad_norm": 1.4023474009400745, "learning_rate": 2.9943032151144814e-06, "loss": 0.5008, "step": 21182 }, { "epoch": 0.7546894205247876, "grad_norm": 1.4264562251715418, "learning_rate": 2.993479830995815e-06, "loss": 0.4363, "step": 21183 }, { "epoch": 0.7547250476512817, "grad_norm": 1.8119465056674293, "learning_rate": 2.992656540173855e-06, "loss": 0.498, "step": 21184 }, { "epoch": 0.7547606747777758, "grad_norm": 1.139523727557821, "learning_rate": 2.991833342659577e-06, "loss": 0.407, "step": 21185 }, { "epoch": 0.7547963019042699, "grad_norm": 1.2576831235136552, "learning_rate": 2.991010238463934e-06, "loss": 0.4043, "step": 21186 }, { "epoch": 0.754831929030764, "grad_norm": 1.5895876794749142, "learning_rate": 2.9901872275978915e-06, "loss": 0.6255, "step": 21187 }, { "epoch": 0.7548675561572581, "grad_norm": 1.667601232945883, "learning_rate": 2.9893643100724023e-06, "loss": 0.5541, "step": 21188 }, { "epoch": 0.7549031832837523, "grad_norm": 1.3220602860508572, "learning_rate": 2.9885414858984308e-06, "loss": 0.4184, "step": 21189 }, { "epoch": 0.7549388104102464, "grad_norm": 1.3879182250489763, "learning_rate": 2.987718755086927e-06, "loss": 0.4807, "step": 21190 }, { "epoch": 0.7549744375367404, "grad_norm": 1.3594605748862323, "learning_rate": 2.9868961176488488e-06, "loss": 0.4183, "step": 21191 }, { "epoch": 0.7550100646632346, "grad_norm": 1.2505766108927345, "learning_rate": 2.986073573595152e-06, "loss": 0.5369, "step": 21192 }, { "epoch": 0.7550456917897287, "grad_norm": 1.1381156490984827, "learning_rate": 2.9852511229367862e-06, "loss": 0.3929, "step": 21193 }, { "epoch": 0.7550813189162228, "grad_norm": 1.6989190761882882, "learning_rate": 2.9844287656847095e-06, "loss": 0.5345, "step": 21194 }, { "epoch": 0.755116946042717, "grad_norm": 1.1254153447999466, "learning_rate": 2.9836065018498628e-06, "loss": 0.4246, "step": 21195 }, { "epoch": 0.755152573169211, "grad_norm": 1.2460830599540178, "learning_rate": 2.9827843314432037e-06, "loss": 0.4362, "step": 21196 }, { "epoch": 0.7551882002957051, "grad_norm": 1.1974984795917005, "learning_rate": 2.981962254475673e-06, "loss": 0.4231, "step": 21197 }, { "epoch": 0.7552238274221993, "grad_norm": 1.4442098381985764, "learning_rate": 2.9811402709582195e-06, "loss": 0.5029, "step": 21198 }, { "epoch": 0.7552594545486934, "grad_norm": 1.8046418792007153, "learning_rate": 2.98031838090179e-06, "loss": 0.4885, "step": 21199 }, { "epoch": 0.7552950816751874, "grad_norm": 1.4600822288178898, "learning_rate": 2.9794965843173294e-06, "loss": 0.4239, "step": 21200 }, { "epoch": 0.7553307088016816, "grad_norm": 1.6489996486632008, "learning_rate": 2.978674881215777e-06, "loss": 0.5242, "step": 21201 }, { "epoch": 0.7553663359281757, "grad_norm": 2.01880355171496, "learning_rate": 2.9778532716080764e-06, "loss": 0.5423, "step": 21202 }, { "epoch": 0.7554019630546698, "grad_norm": 1.4677160880086506, "learning_rate": 2.9770317555051697e-06, "loss": 0.4453, "step": 21203 }, { "epoch": 0.755437590181164, "grad_norm": 1.1850044315111852, "learning_rate": 2.9762103329179916e-06, "loss": 0.4142, "step": 21204 }, { "epoch": 0.755473217307658, "grad_norm": 1.4418361259734582, "learning_rate": 2.9753890038574853e-06, "loss": 0.4215, "step": 21205 }, { "epoch": 0.7555088444341521, "grad_norm": 1.5151682626890537, "learning_rate": 2.974567768334581e-06, "loss": 0.5178, "step": 21206 }, { "epoch": 0.7555444715606463, "grad_norm": 1.7147703717480507, "learning_rate": 2.9737466263602176e-06, "loss": 0.526, "step": 21207 }, { "epoch": 0.7555800986871404, "grad_norm": 1.464007278109741, "learning_rate": 2.972925577945328e-06, "loss": 0.5174, "step": 21208 }, { "epoch": 0.7556157258136345, "grad_norm": 1.5562918411082798, "learning_rate": 2.97210462310085e-06, "loss": 0.4873, "step": 21209 }, { "epoch": 0.7556513529401286, "grad_norm": 1.2803962284015762, "learning_rate": 2.9712837618377076e-06, "loss": 0.3878, "step": 21210 }, { "epoch": 0.7556869800666227, "grad_norm": 1.1859592215322345, "learning_rate": 2.9704629941668384e-06, "loss": 0.4196, "step": 21211 }, { "epoch": 0.7557226071931168, "grad_norm": 1.4824011617648396, "learning_rate": 2.9696423200991644e-06, "loss": 0.5138, "step": 21212 }, { "epoch": 0.755758234319611, "grad_norm": 1.6859559501173516, "learning_rate": 2.9688217396456166e-06, "loss": 0.4887, "step": 21213 }, { "epoch": 0.755793861446105, "grad_norm": 1.410542639302035, "learning_rate": 2.968001252817122e-06, "loss": 0.5931, "step": 21214 }, { "epoch": 0.7558294885725991, "grad_norm": 1.311955938242515, "learning_rate": 2.9671808596246053e-06, "loss": 0.4435, "step": 21215 }, { "epoch": 0.7558651156990933, "grad_norm": 1.5877071762225794, "learning_rate": 2.9663605600789957e-06, "loss": 0.4969, "step": 21216 }, { "epoch": 0.7559007428255874, "grad_norm": 1.2447223812704566, "learning_rate": 2.9655403541912065e-06, "loss": 0.396, "step": 21217 }, { "epoch": 0.7559363699520815, "grad_norm": 1.2953430161761335, "learning_rate": 2.9647202419721687e-06, "loss": 0.4712, "step": 21218 }, { "epoch": 0.7559719970785757, "grad_norm": 1.9921394363748774, "learning_rate": 2.963900223432794e-06, "loss": 0.5013, "step": 21219 }, { "epoch": 0.7560076242050697, "grad_norm": 1.348340842747615, "learning_rate": 2.96308029858401e-06, "loss": 0.4075, "step": 21220 }, { "epoch": 0.7560432513315638, "grad_norm": 1.7320913843174746, "learning_rate": 2.962260467436724e-06, "loss": 0.503, "step": 21221 }, { "epoch": 0.756078878458058, "grad_norm": 1.316464638439146, "learning_rate": 2.961440730001864e-06, "loss": 0.4443, "step": 21222 }, { "epoch": 0.7561145055845521, "grad_norm": 1.5003639887034694, "learning_rate": 2.9606210862903385e-06, "loss": 0.5249, "step": 21223 }, { "epoch": 0.7561501327110461, "grad_norm": 1.180802982722605, "learning_rate": 2.959801536313066e-06, "loss": 0.3727, "step": 21224 }, { "epoch": 0.7561857598375403, "grad_norm": 1.2489780444564909, "learning_rate": 2.9589820800809545e-06, "loss": 0.3831, "step": 21225 }, { "epoch": 0.7562213869640344, "grad_norm": 2.0990452612924653, "learning_rate": 2.9581627176049167e-06, "loss": 0.4279, "step": 21226 }, { "epoch": 0.7562570140905285, "grad_norm": 1.5271071639523708, "learning_rate": 2.9573434488958687e-06, "loss": 0.4285, "step": 21227 }, { "epoch": 0.7562926412170227, "grad_norm": 1.580692531637937, "learning_rate": 2.9565242739647115e-06, "loss": 0.5007, "step": 21228 }, { "epoch": 0.7563282683435167, "grad_norm": 1.1738272722088814, "learning_rate": 2.9557051928223556e-06, "loss": 0.3493, "step": 21229 }, { "epoch": 0.7563638954700108, "grad_norm": 1.7474441643741778, "learning_rate": 2.95488620547971e-06, "loss": 0.5442, "step": 21230 }, { "epoch": 0.756399522596505, "grad_norm": 1.3841062132065534, "learning_rate": 2.9540673119476816e-06, "loss": 0.425, "step": 21231 }, { "epoch": 0.7564351497229991, "grad_norm": 1.7030569069571497, "learning_rate": 2.953248512237169e-06, "loss": 0.4592, "step": 21232 }, { "epoch": 0.7564707768494932, "grad_norm": 1.8356416481131028, "learning_rate": 2.9524298063590808e-06, "loss": 0.4596, "step": 21233 }, { "epoch": 0.7565064039759873, "grad_norm": 1.1353157098710036, "learning_rate": 2.951611194324312e-06, "loss": 0.4053, "step": 21234 }, { "epoch": 0.7565420311024814, "grad_norm": 1.206188044401408, "learning_rate": 2.9507926761437676e-06, "loss": 0.425, "step": 21235 }, { "epoch": 0.7565776582289755, "grad_norm": 1.1962427939306024, "learning_rate": 2.949974251828345e-06, "loss": 0.3912, "step": 21236 }, { "epoch": 0.7566132853554697, "grad_norm": 1.1173677577120813, "learning_rate": 2.9491559213889433e-06, "loss": 0.4089, "step": 21237 }, { "epoch": 0.7566489124819638, "grad_norm": 1.1653685061208812, "learning_rate": 2.948337684836462e-06, "loss": 0.3778, "step": 21238 }, { "epoch": 0.7566845396084578, "grad_norm": 1.2629830909786433, "learning_rate": 2.9475195421817895e-06, "loss": 0.3895, "step": 21239 }, { "epoch": 0.756720166734952, "grad_norm": 1.4370751420914343, "learning_rate": 2.946701493435827e-06, "loss": 0.5647, "step": 21240 }, { "epoch": 0.7567557938614461, "grad_norm": 1.4843721891276096, "learning_rate": 2.945883538609462e-06, "loss": 0.5394, "step": 21241 }, { "epoch": 0.7567914209879402, "grad_norm": 1.3593900394171692, "learning_rate": 2.94506567771359e-06, "loss": 0.5727, "step": 21242 }, { "epoch": 0.7568270481144344, "grad_norm": 1.1650211104400146, "learning_rate": 2.944247910759097e-06, "loss": 0.4046, "step": 21243 }, { "epoch": 0.7568626752409284, "grad_norm": 1.1870564510444914, "learning_rate": 2.9434302377568747e-06, "loss": 0.3307, "step": 21244 }, { "epoch": 0.7568983023674225, "grad_norm": 1.202943358517899, "learning_rate": 2.9426126587178117e-06, "loss": 0.4507, "step": 21245 }, { "epoch": 0.7569339294939167, "grad_norm": 1.5057194113676384, "learning_rate": 2.941795173652796e-06, "loss": 0.485, "step": 21246 }, { "epoch": 0.7569695566204108, "grad_norm": 1.2467770243360572, "learning_rate": 2.940977782572708e-06, "loss": 0.423, "step": 21247 }, { "epoch": 0.7570051837469048, "grad_norm": 1.193937949400886, "learning_rate": 2.940160485488436e-06, "loss": 0.4071, "step": 21248 }, { "epoch": 0.757040810873399, "grad_norm": 1.3958138078320639, "learning_rate": 2.9393432824108636e-06, "loss": 0.5044, "step": 21249 }, { "epoch": 0.7570764379998931, "grad_norm": 1.5236053987702411, "learning_rate": 2.9385261733508676e-06, "loss": 0.5237, "step": 21250 }, { "epoch": 0.7571120651263872, "grad_norm": 1.2270318632100383, "learning_rate": 2.9377091583193317e-06, "loss": 0.4157, "step": 21251 }, { "epoch": 0.7571476922528814, "grad_norm": 1.3039953594258307, "learning_rate": 2.9368922373271338e-06, "loss": 0.3952, "step": 21252 }, { "epoch": 0.7571833193793754, "grad_norm": 1.5679544420230933, "learning_rate": 2.936075410385156e-06, "loss": 0.4648, "step": 21253 }, { "epoch": 0.7572189465058695, "grad_norm": 1.3193866603737818, "learning_rate": 2.9352586775042692e-06, "loss": 0.4526, "step": 21254 }, { "epoch": 0.7572545736323637, "grad_norm": 1.6048230025083832, "learning_rate": 2.934442038695353e-06, "loss": 0.4886, "step": 21255 }, { "epoch": 0.7572902007588578, "grad_norm": 1.367629497706519, "learning_rate": 2.9336254939692765e-06, "loss": 0.426, "step": 21256 }, { "epoch": 0.7573258278853519, "grad_norm": 1.6931803722270509, "learning_rate": 2.932809043336919e-06, "loss": 0.5952, "step": 21257 }, { "epoch": 0.757361455011846, "grad_norm": 2.124978577149309, "learning_rate": 2.9319926868091464e-06, "loss": 0.5184, "step": 21258 }, { "epoch": 0.7573970821383401, "grad_norm": 1.3549362664119577, "learning_rate": 2.9311764243968278e-06, "loss": 0.4505, "step": 21259 }, { "epoch": 0.7574327092648342, "grad_norm": 1.6419830795860566, "learning_rate": 2.9303602561108425e-06, "loss": 0.5791, "step": 21260 }, { "epoch": 0.7574683363913284, "grad_norm": 1.5678946723456193, "learning_rate": 2.9295441819620483e-06, "loss": 0.48, "step": 21261 }, { "epoch": 0.7575039635178225, "grad_norm": 1.407444953938376, "learning_rate": 2.928728201961318e-06, "loss": 0.5225, "step": 21262 }, { "epoch": 0.7575395906443165, "grad_norm": 1.7614696203688525, "learning_rate": 2.927912316119511e-06, "loss": 0.5236, "step": 21263 }, { "epoch": 0.7575752177708107, "grad_norm": 1.5256545202049494, "learning_rate": 2.9270965244474994e-06, "loss": 0.4041, "step": 21264 }, { "epoch": 0.7576108448973048, "grad_norm": 1.3126521855943176, "learning_rate": 2.926280826956136e-06, "loss": 0.4749, "step": 21265 }, { "epoch": 0.7576464720237989, "grad_norm": 1.360839856743629, "learning_rate": 2.9254652236562887e-06, "loss": 0.3628, "step": 21266 }, { "epoch": 0.757682099150293, "grad_norm": 2.3759591141620633, "learning_rate": 2.924649714558817e-06, "loss": 0.5267, "step": 21267 }, { "epoch": 0.7577177262767871, "grad_norm": 1.2578904429617987, "learning_rate": 2.9238342996745816e-06, "loss": 0.4919, "step": 21268 }, { "epoch": 0.7577533534032812, "grad_norm": 1.5840371872682621, "learning_rate": 2.923018979014437e-06, "loss": 0.5246, "step": 21269 }, { "epoch": 0.7577889805297754, "grad_norm": 1.568007223664147, "learning_rate": 2.9222037525892433e-06, "loss": 0.429, "step": 21270 }, { "epoch": 0.7578246076562695, "grad_norm": 1.1524200497845223, "learning_rate": 2.9213886204098508e-06, "loss": 0.4, "step": 21271 }, { "epoch": 0.7578602347827635, "grad_norm": 2.2372985154665463, "learning_rate": 2.920573582487116e-06, "loss": 0.5226, "step": 21272 }, { "epoch": 0.7578958619092577, "grad_norm": 1.5729304031446065, "learning_rate": 2.919758638831893e-06, "loss": 0.4338, "step": 21273 }, { "epoch": 0.7579314890357518, "grad_norm": 1.3881165358473933, "learning_rate": 2.918943789455032e-06, "loss": 0.5718, "step": 21274 }, { "epoch": 0.7579671161622459, "grad_norm": 1.7035056607184689, "learning_rate": 2.9181290343673876e-06, "loss": 0.5269, "step": 21275 }, { "epoch": 0.7580027432887401, "grad_norm": 1.2581514969851533, "learning_rate": 2.9173143735798016e-06, "loss": 0.4078, "step": 21276 }, { "epoch": 0.7580383704152341, "grad_norm": 1.497339941709168, "learning_rate": 2.916499807103128e-06, "loss": 0.5487, "step": 21277 }, { "epoch": 0.7580739975417282, "grad_norm": 1.1813775634693728, "learning_rate": 2.9156853349482075e-06, "loss": 0.4355, "step": 21278 }, { "epoch": 0.7581096246682224, "grad_norm": 1.2165247737296054, "learning_rate": 2.9148709571258914e-06, "loss": 0.4121, "step": 21279 }, { "epoch": 0.7581452517947165, "grad_norm": 1.5867302817410214, "learning_rate": 2.9140566736470176e-06, "loss": 0.4973, "step": 21280 }, { "epoch": 0.7581808789212106, "grad_norm": 1.2073435010897173, "learning_rate": 2.9132424845224316e-06, "loss": 0.3816, "step": 21281 }, { "epoch": 0.7582165060477047, "grad_norm": 1.203584917621161, "learning_rate": 2.912428389762977e-06, "loss": 0.4397, "step": 21282 }, { "epoch": 0.7582521331741988, "grad_norm": 1.7431664288282023, "learning_rate": 2.91161438937949e-06, "loss": 0.5695, "step": 21283 }, { "epoch": 0.7582877603006929, "grad_norm": 1.3283795912198904, "learning_rate": 2.9108004833828164e-06, "loss": 0.4515, "step": 21284 }, { "epoch": 0.7583233874271871, "grad_norm": 1.5148203798709048, "learning_rate": 2.9099866717837864e-06, "loss": 0.544, "step": 21285 }, { "epoch": 0.7583590145536812, "grad_norm": 1.8718060287898726, "learning_rate": 2.9091729545932424e-06, "loss": 0.5272, "step": 21286 }, { "epoch": 0.7583946416801753, "grad_norm": 1.440890433450467, "learning_rate": 2.908359331822013e-06, "loss": 0.4889, "step": 21287 }, { "epoch": 0.7584302688066694, "grad_norm": 1.2994335757419968, "learning_rate": 2.9075458034809368e-06, "loss": 0.3669, "step": 21288 }, { "epoch": 0.7584658959331635, "grad_norm": 2.5012332421973302, "learning_rate": 2.906732369580845e-06, "loss": 0.5504, "step": 21289 }, { "epoch": 0.7585015230596577, "grad_norm": 1.236424107006287, "learning_rate": 2.9059190301325723e-06, "loss": 0.3953, "step": 21290 }, { "epoch": 0.7585371501861518, "grad_norm": 1.3291533812082805, "learning_rate": 2.9051057851469433e-06, "loss": 0.4393, "step": 21291 }, { "epoch": 0.7585727773126458, "grad_norm": 1.2832639228413787, "learning_rate": 2.9042926346347932e-06, "loss": 0.5687, "step": 21292 }, { "epoch": 0.75860840443914, "grad_norm": 1.4110481968223778, "learning_rate": 2.9034795786069436e-06, "loss": 0.5317, "step": 21293 }, { "epoch": 0.7586440315656341, "grad_norm": 1.648666910560695, "learning_rate": 2.902666617074222e-06, "loss": 0.4992, "step": 21294 }, { "epoch": 0.7586796586921282, "grad_norm": 1.5226396662902828, "learning_rate": 2.9018537500474595e-06, "loss": 0.4928, "step": 21295 }, { "epoch": 0.7587152858186224, "grad_norm": 1.2070143754245095, "learning_rate": 2.9010409775374692e-06, "loss": 0.4062, "step": 21296 }, { "epoch": 0.7587509129451164, "grad_norm": 1.1597914270362448, "learning_rate": 2.9002282995550866e-06, "loss": 0.4187, "step": 21297 }, { "epoch": 0.7587865400716105, "grad_norm": 1.4094497178453655, "learning_rate": 2.8994157161111235e-06, "loss": 0.5535, "step": 21298 }, { "epoch": 0.7588221671981047, "grad_norm": 1.7243471520479972, "learning_rate": 2.8986032272164068e-06, "loss": 0.5294, "step": 21299 }, { "epoch": 0.7588577943245988, "grad_norm": 1.448608359542312, "learning_rate": 2.897790832881747e-06, "loss": 0.4299, "step": 21300 }, { "epoch": 0.7588934214510928, "grad_norm": 1.1257748221171577, "learning_rate": 2.89697853311797e-06, "loss": 0.3577, "step": 21301 }, { "epoch": 0.758929048577587, "grad_norm": 1.5148133688080923, "learning_rate": 2.8961663279358865e-06, "loss": 0.4858, "step": 21302 }, { "epoch": 0.7589646757040811, "grad_norm": 1.2644268604884386, "learning_rate": 2.8953542173463133e-06, "loss": 0.4024, "step": 21303 }, { "epoch": 0.7590003028305752, "grad_norm": 1.1135872743129689, "learning_rate": 2.894542201360063e-06, "loss": 0.3796, "step": 21304 }, { "epoch": 0.7590359299570694, "grad_norm": 1.1754505729691604, "learning_rate": 2.8937302799879553e-06, "loss": 0.3366, "step": 21305 }, { "epoch": 0.7590715570835634, "grad_norm": 1.2086807152167955, "learning_rate": 2.8929184532407905e-06, "loss": 0.4213, "step": 21306 }, { "epoch": 0.7591071842100575, "grad_norm": 1.5789179564879876, "learning_rate": 2.892106721129385e-06, "loss": 0.4822, "step": 21307 }, { "epoch": 0.7591428113365517, "grad_norm": 1.35192851469904, "learning_rate": 2.8912950836645505e-06, "loss": 0.3693, "step": 21308 }, { "epoch": 0.7591784384630458, "grad_norm": 1.5250748850323106, "learning_rate": 2.8904835408570865e-06, "loss": 0.4898, "step": 21309 }, { "epoch": 0.7592140655895399, "grad_norm": 1.3610664679331106, "learning_rate": 2.8896720927178033e-06, "loss": 0.4198, "step": 21310 }, { "epoch": 0.759249692716034, "grad_norm": 1.3414853368462416, "learning_rate": 2.8888607392575073e-06, "loss": 0.3968, "step": 21311 }, { "epoch": 0.7592853198425281, "grad_norm": 1.238081206226992, "learning_rate": 2.888049480487003e-06, "loss": 0.4058, "step": 21312 }, { "epoch": 0.7593209469690222, "grad_norm": 0.9824754660278723, "learning_rate": 2.887238316417088e-06, "loss": 0.346, "step": 21313 }, { "epoch": 0.7593565740955164, "grad_norm": 1.921729060467987, "learning_rate": 2.8864272470585696e-06, "loss": 0.5447, "step": 21314 }, { "epoch": 0.7593922012220105, "grad_norm": 1.61774901614314, "learning_rate": 2.8856162724222414e-06, "loss": 0.5123, "step": 21315 }, { "epoch": 0.7594278283485045, "grad_norm": 1.5746104654853017, "learning_rate": 2.884805392518908e-06, "loss": 0.5115, "step": 21316 }, { "epoch": 0.7594634554749987, "grad_norm": 1.9935691093753523, "learning_rate": 2.8839946073593604e-06, "loss": 0.518, "step": 21317 }, { "epoch": 0.7594990826014928, "grad_norm": 1.2541589735529983, "learning_rate": 2.8831839169543998e-06, "loss": 0.3696, "step": 21318 }, { "epoch": 0.7595347097279869, "grad_norm": 1.3050286336157657, "learning_rate": 2.8823733213148187e-06, "loss": 0.3977, "step": 21319 }, { "epoch": 0.759570336854481, "grad_norm": 1.630065340099112, "learning_rate": 2.8815628204514113e-06, "loss": 0.4684, "step": 21320 }, { "epoch": 0.7596059639809751, "grad_norm": 1.1717111310932338, "learning_rate": 2.8807524143749743e-06, "loss": 0.4315, "step": 21321 }, { "epoch": 0.7596415911074692, "grad_norm": 1.4338891017428683, "learning_rate": 2.8799421030962917e-06, "loss": 0.5836, "step": 21322 }, { "epoch": 0.7596772182339634, "grad_norm": 1.2191194030970887, "learning_rate": 2.8791318866261597e-06, "loss": 0.4378, "step": 21323 }, { "epoch": 0.7597128453604575, "grad_norm": 1.5570402415510325, "learning_rate": 2.8783217649753616e-06, "loss": 0.5574, "step": 21324 }, { "epoch": 0.7597484724869515, "grad_norm": 1.870607326724912, "learning_rate": 2.8775117381546856e-06, "loss": 0.5654, "step": 21325 }, { "epoch": 0.7597840996134457, "grad_norm": 1.271880478509151, "learning_rate": 2.87670180617492e-06, "loss": 0.4184, "step": 21326 }, { "epoch": 0.7598197267399398, "grad_norm": 1.339188326934561, "learning_rate": 2.8758919690468514e-06, "loss": 0.3869, "step": 21327 }, { "epoch": 0.7598553538664339, "grad_norm": 1.3808800290900876, "learning_rate": 2.8750822267812574e-06, "loss": 0.42, "step": 21328 }, { "epoch": 0.7598909809929281, "grad_norm": 1.1613502393849175, "learning_rate": 2.8742725793889235e-06, "loss": 0.3828, "step": 21329 }, { "epoch": 0.7599266081194221, "grad_norm": 1.3616415474681003, "learning_rate": 2.873463026880634e-06, "loss": 0.5201, "step": 21330 }, { "epoch": 0.7599622352459162, "grad_norm": 1.5547248846060011, "learning_rate": 2.872653569267163e-06, "loss": 0.5429, "step": 21331 }, { "epoch": 0.7599978623724104, "grad_norm": 1.3778869043777262, "learning_rate": 2.8718442065592944e-06, "loss": 0.5074, "step": 21332 }, { "epoch": 0.7600334894989045, "grad_norm": 1.4807589399733225, "learning_rate": 2.871034938767798e-06, "loss": 0.5465, "step": 21333 }, { "epoch": 0.7600691166253986, "grad_norm": 1.15815731960701, "learning_rate": 2.870225765903456e-06, "loss": 0.3843, "step": 21334 }, { "epoch": 0.7601047437518927, "grad_norm": 1.5735988304341653, "learning_rate": 2.869416687977041e-06, "loss": 0.4539, "step": 21335 }, { "epoch": 0.7601403708783868, "grad_norm": 1.401093604694752, "learning_rate": 2.868607704999329e-06, "loss": 0.4955, "step": 21336 }, { "epoch": 0.7601759980048809, "grad_norm": 1.2957812546406633, "learning_rate": 2.8677988169810876e-06, "loss": 0.5105, "step": 21337 }, { "epoch": 0.7602116251313751, "grad_norm": 1.4459538733439954, "learning_rate": 2.866990023933093e-06, "loss": 0.5301, "step": 21338 }, { "epoch": 0.7602472522578692, "grad_norm": 1.0650984174371452, "learning_rate": 2.8661813258661087e-06, "loss": 0.3815, "step": 21339 }, { "epoch": 0.7602828793843632, "grad_norm": 1.2046793554151265, "learning_rate": 2.865372722790907e-06, "loss": 0.3915, "step": 21340 }, { "epoch": 0.7603185065108574, "grad_norm": 1.9909584858768206, "learning_rate": 2.8645642147182527e-06, "loss": 0.4965, "step": 21341 }, { "epoch": 0.7603541336373515, "grad_norm": 1.6070157344916118, "learning_rate": 2.863755801658915e-06, "loss": 0.5191, "step": 21342 }, { "epoch": 0.7603897607638456, "grad_norm": 1.9192673754372698, "learning_rate": 2.862947483623659e-06, "loss": 0.5114, "step": 21343 }, { "epoch": 0.7604253878903398, "grad_norm": 1.7486956650393635, "learning_rate": 2.8621392606232424e-06, "loss": 0.4793, "step": 21344 }, { "epoch": 0.7604610150168338, "grad_norm": 1.4472924414312762, "learning_rate": 2.8613311326684347e-06, "loss": 0.5129, "step": 21345 }, { "epoch": 0.7604966421433279, "grad_norm": 1.7158901302249685, "learning_rate": 2.8605230997699885e-06, "loss": 0.5044, "step": 21346 }, { "epoch": 0.7605322692698221, "grad_norm": 1.9633192173561593, "learning_rate": 2.859715161938671e-06, "loss": 0.5457, "step": 21347 }, { "epoch": 0.7605678963963162, "grad_norm": 1.4157636646153209, "learning_rate": 2.8589073191852303e-06, "loss": 0.5537, "step": 21348 }, { "epoch": 0.7606035235228102, "grad_norm": 1.3970602892346093, "learning_rate": 2.858099571520436e-06, "loss": 0.5106, "step": 21349 }, { "epoch": 0.7606391506493044, "grad_norm": 1.7381779033694555, "learning_rate": 2.8572919189550354e-06, "loss": 0.4284, "step": 21350 }, { "epoch": 0.7606747777757985, "grad_norm": 1.1925907233398638, "learning_rate": 2.8564843614997884e-06, "loss": 0.4038, "step": 21351 }, { "epoch": 0.7607104049022926, "grad_norm": 1.2360709059656287, "learning_rate": 2.8556768991654414e-06, "loss": 0.4112, "step": 21352 }, { "epoch": 0.7607460320287868, "grad_norm": 1.3955917302347571, "learning_rate": 2.8548695319627504e-06, "loss": 0.4652, "step": 21353 }, { "epoch": 0.7607816591552808, "grad_norm": 2.251910970154445, "learning_rate": 2.854062259902469e-06, "loss": 0.5059, "step": 21354 }, { "epoch": 0.7608172862817749, "grad_norm": 1.3389439306349633, "learning_rate": 2.8532550829953397e-06, "loss": 0.4878, "step": 21355 }, { "epoch": 0.7608529134082691, "grad_norm": 1.7352467147460846, "learning_rate": 2.852448001252115e-06, "loss": 0.5769, "step": 21356 }, { "epoch": 0.7608885405347632, "grad_norm": 1.8231660279460147, "learning_rate": 2.8516410146835406e-06, "loss": 0.4942, "step": 21357 }, { "epoch": 0.7609241676612573, "grad_norm": 1.6813492739995621, "learning_rate": 2.8508341233003656e-06, "loss": 0.5388, "step": 21358 }, { "epoch": 0.7609597947877514, "grad_norm": 1.2862012602614596, "learning_rate": 2.8500273271133283e-06, "loss": 0.4282, "step": 21359 }, { "epoch": 0.7609954219142455, "grad_norm": 1.1836225519246713, "learning_rate": 2.849220626133178e-06, "loss": 0.3917, "step": 21360 }, { "epoch": 0.7610310490407396, "grad_norm": 1.181014737548657, "learning_rate": 2.848414020370649e-06, "loss": 0.4388, "step": 21361 }, { "epoch": 0.7610666761672338, "grad_norm": 1.3694014223489108, "learning_rate": 2.8476075098364875e-06, "loss": 0.4158, "step": 21362 }, { "epoch": 0.7611023032937279, "grad_norm": 1.1647406297136536, "learning_rate": 2.84680109454143e-06, "loss": 0.4142, "step": 21363 }, { "epoch": 0.7611379304202219, "grad_norm": 1.2442822371419626, "learning_rate": 2.845994774496216e-06, "loss": 0.4163, "step": 21364 }, { "epoch": 0.7611735575467161, "grad_norm": 1.2367471567209936, "learning_rate": 2.845188549711585e-06, "loss": 0.4017, "step": 21365 }, { "epoch": 0.7612091846732102, "grad_norm": 1.2582537160808411, "learning_rate": 2.8443824201982663e-06, "loss": 0.4381, "step": 21366 }, { "epoch": 0.7612448117997043, "grad_norm": 1.1814713808875474, "learning_rate": 2.8435763859670007e-06, "loss": 0.4156, "step": 21367 }, { "epoch": 0.7612804389261985, "grad_norm": 1.7287717738827446, "learning_rate": 2.8427704470285143e-06, "loss": 0.4741, "step": 21368 }, { "epoch": 0.7613160660526925, "grad_norm": 1.3458323197214104, "learning_rate": 2.8419646033935443e-06, "loss": 0.4264, "step": 21369 }, { "epoch": 0.7613516931791866, "grad_norm": 1.7779567138864791, "learning_rate": 2.8411588550728164e-06, "loss": 0.4862, "step": 21370 }, { "epoch": 0.7613873203056808, "grad_norm": 1.21546419796723, "learning_rate": 2.840353202077063e-06, "loss": 0.366, "step": 21371 }, { "epoch": 0.7614229474321749, "grad_norm": 2.3039084534360623, "learning_rate": 2.83954764441701e-06, "loss": 0.489, "step": 21372 }, { "epoch": 0.7614585745586689, "grad_norm": 1.6769203272317659, "learning_rate": 2.838742182103388e-06, "loss": 0.4945, "step": 21373 }, { "epoch": 0.7614942016851631, "grad_norm": 1.507294786495549, "learning_rate": 2.8379368151469155e-06, "loss": 0.5233, "step": 21374 }, { "epoch": 0.7615298288116572, "grad_norm": 1.2463669336869365, "learning_rate": 2.8371315435583214e-06, "loss": 0.4483, "step": 21375 }, { "epoch": 0.7615654559381513, "grad_norm": 1.2282992879194228, "learning_rate": 2.83632636734833e-06, "loss": 0.4085, "step": 21376 }, { "epoch": 0.7616010830646455, "grad_norm": 1.449551901171474, "learning_rate": 2.835521286527657e-06, "loss": 0.4174, "step": 21377 }, { "epoch": 0.7616367101911395, "grad_norm": 1.5354126722724137, "learning_rate": 2.8347163011070255e-06, "loss": 0.4787, "step": 21378 }, { "epoch": 0.7616723373176336, "grad_norm": 1.3118651544591724, "learning_rate": 2.8339114110971556e-06, "loss": 0.4536, "step": 21379 }, { "epoch": 0.7617079644441278, "grad_norm": 1.5485877794233422, "learning_rate": 2.833106616508766e-06, "loss": 0.5297, "step": 21380 }, { "epoch": 0.7617435915706219, "grad_norm": 1.696219048691764, "learning_rate": 2.832301917352569e-06, "loss": 0.539, "step": 21381 }, { "epoch": 0.761779218697116, "grad_norm": 1.514368127574329, "learning_rate": 2.8314973136392844e-06, "loss": 0.5131, "step": 21382 }, { "epoch": 0.7618148458236101, "grad_norm": 1.489087718181288, "learning_rate": 2.830692805379621e-06, "loss": 0.5525, "step": 21383 }, { "epoch": 0.7618504729501042, "grad_norm": 1.5058815518652717, "learning_rate": 2.829888392584297e-06, "loss": 0.5866, "step": 21384 }, { "epoch": 0.7618861000765983, "grad_norm": 1.1360744944029622, "learning_rate": 2.8290840752640135e-06, "loss": 0.4056, "step": 21385 }, { "epoch": 0.7619217272030925, "grad_norm": 1.6805716694354622, "learning_rate": 2.828279853429492e-06, "loss": 0.5355, "step": 21386 }, { "epoch": 0.7619573543295866, "grad_norm": 1.470764333716952, "learning_rate": 2.82747572709144e-06, "loss": 0.4783, "step": 21387 }, { "epoch": 0.7619929814560806, "grad_norm": 1.2824524757655686, "learning_rate": 2.8266716962605577e-06, "loss": 0.5304, "step": 21388 }, { "epoch": 0.7620286085825748, "grad_norm": 1.1494287744452008, "learning_rate": 2.8258677609475595e-06, "loss": 0.4229, "step": 21389 }, { "epoch": 0.7620642357090689, "grad_norm": 1.2228612607785667, "learning_rate": 2.825063921163144e-06, "loss": 0.4172, "step": 21390 }, { "epoch": 0.762099862835563, "grad_norm": 1.105915443657035, "learning_rate": 2.82426017691802e-06, "loss": 0.376, "step": 21391 }, { "epoch": 0.7621354899620572, "grad_norm": 1.2178644661705647, "learning_rate": 2.823456528222883e-06, "loss": 0.423, "step": 21392 }, { "epoch": 0.7621711170885512, "grad_norm": 1.1917713445128335, "learning_rate": 2.8226529750884403e-06, "loss": 0.4089, "step": 21393 }, { "epoch": 0.7622067442150453, "grad_norm": 1.4962260904687574, "learning_rate": 2.821849517525389e-06, "loss": 0.512, "step": 21394 }, { "epoch": 0.7622423713415395, "grad_norm": 2.2224682386001233, "learning_rate": 2.8210461555444313e-06, "loss": 0.5276, "step": 21395 }, { "epoch": 0.7622779984680336, "grad_norm": 1.5361393856136565, "learning_rate": 2.8202428891562595e-06, "loss": 0.5412, "step": 21396 }, { "epoch": 0.7623136255945276, "grad_norm": 1.7669075936923417, "learning_rate": 2.819439718371575e-06, "loss": 0.5514, "step": 21397 }, { "epoch": 0.7623492527210218, "grad_norm": 1.3351442179107091, "learning_rate": 2.8186366432010658e-06, "loss": 0.4525, "step": 21398 }, { "epoch": 0.7623848798475159, "grad_norm": 1.257049429630588, "learning_rate": 2.817833663655429e-06, "loss": 0.375, "step": 21399 }, { "epoch": 0.76242050697401, "grad_norm": 1.4174352268834431, "learning_rate": 2.817030779745358e-06, "loss": 0.4176, "step": 21400 }, { "epoch": 0.7624561341005042, "grad_norm": 2.0408073788726777, "learning_rate": 2.8162279914815427e-06, "loss": 0.4274, "step": 21401 }, { "epoch": 0.7624917612269982, "grad_norm": 1.700846735049696, "learning_rate": 2.8154252988746757e-06, "loss": 0.5837, "step": 21402 }, { "epoch": 0.7625273883534923, "grad_norm": 1.2252781733316132, "learning_rate": 2.814622701935439e-06, "loss": 0.4073, "step": 21403 }, { "epoch": 0.7625630154799865, "grad_norm": 1.1901073893721188, "learning_rate": 2.8138202006745276e-06, "loss": 0.402, "step": 21404 }, { "epoch": 0.7625986426064806, "grad_norm": 1.1255500496112476, "learning_rate": 2.8130177951026183e-06, "loss": 0.4051, "step": 21405 }, { "epoch": 0.7626342697329747, "grad_norm": 1.3468405442068132, "learning_rate": 2.812215485230405e-06, "loss": 0.4071, "step": 21406 }, { "epoch": 0.7626698968594688, "grad_norm": 1.183899373727605, "learning_rate": 2.8114132710685625e-06, "loss": 0.3818, "step": 21407 }, { "epoch": 0.7627055239859629, "grad_norm": 1.1067489142820541, "learning_rate": 2.810611152627777e-06, "loss": 0.3996, "step": 21408 }, { "epoch": 0.762741151112457, "grad_norm": 1.2170423673748296, "learning_rate": 2.809809129918729e-06, "loss": 0.4228, "step": 21409 }, { "epoch": 0.7627767782389512, "grad_norm": 1.3276882768038964, "learning_rate": 2.8090072029520987e-06, "loss": 0.5072, "step": 21410 }, { "epoch": 0.7628124053654453, "grad_norm": 2.5684559087575014, "learning_rate": 2.808205371738566e-06, "loss": 0.5117, "step": 21411 }, { "epoch": 0.7628480324919393, "grad_norm": 1.1920355116046608, "learning_rate": 2.8074036362888036e-06, "loss": 0.4302, "step": 21412 }, { "epoch": 0.7628836596184335, "grad_norm": 1.564881953252621, "learning_rate": 2.8066019966134907e-06, "loss": 0.5863, "step": 21413 }, { "epoch": 0.7629192867449276, "grad_norm": 1.7929224249606246, "learning_rate": 2.8058004527232974e-06, "loss": 0.5926, "step": 21414 }, { "epoch": 0.7629549138714217, "grad_norm": 1.175896995448579, "learning_rate": 2.8049990046289e-06, "loss": 0.418, "step": 21415 }, { "epoch": 0.7629905409979159, "grad_norm": 1.2643169460521309, "learning_rate": 2.80419765234097e-06, "loss": 0.4122, "step": 21416 }, { "epoch": 0.7630261681244099, "grad_norm": 1.592193331372629, "learning_rate": 2.8033963958701805e-06, "loss": 0.5363, "step": 21417 }, { "epoch": 0.763061795250904, "grad_norm": 1.2729983338232074, "learning_rate": 2.8025952352271957e-06, "loss": 0.3363, "step": 21418 }, { "epoch": 0.7630974223773982, "grad_norm": 1.299833232454793, "learning_rate": 2.801794170422688e-06, "loss": 0.386, "step": 21419 }, { "epoch": 0.7631330495038923, "grad_norm": 1.4379681369214012, "learning_rate": 2.8009932014673193e-06, "loss": 0.5501, "step": 21420 }, { "epoch": 0.7631686766303863, "grad_norm": 1.447270378223266, "learning_rate": 2.800192328371759e-06, "loss": 0.4433, "step": 21421 }, { "epoch": 0.7632043037568805, "grad_norm": 1.161638441307324, "learning_rate": 2.799391551146672e-06, "loss": 0.3781, "step": 21422 }, { "epoch": 0.7632399308833746, "grad_norm": 1.2610935574024762, "learning_rate": 2.798590869802714e-06, "loss": 0.4138, "step": 21423 }, { "epoch": 0.7632755580098687, "grad_norm": 1.036100072295347, "learning_rate": 2.7977902843505578e-06, "loss": 0.3941, "step": 21424 }, { "epoch": 0.7633111851363629, "grad_norm": 1.452062594083104, "learning_rate": 2.796989794800855e-06, "loss": 0.485, "step": 21425 }, { "epoch": 0.7633468122628569, "grad_norm": 1.1913916018444792, "learning_rate": 2.796189401164271e-06, "loss": 0.4193, "step": 21426 }, { "epoch": 0.763382439389351, "grad_norm": 1.4251353550912256, "learning_rate": 2.795389103451457e-06, "loss": 0.5125, "step": 21427 }, { "epoch": 0.7634180665158452, "grad_norm": 1.2427944127529447, "learning_rate": 2.7945889016730767e-06, "loss": 0.4146, "step": 21428 }, { "epoch": 0.7634536936423393, "grad_norm": 1.2859509736933235, "learning_rate": 2.7937887958397768e-06, "loss": 0.5079, "step": 21429 }, { "epoch": 0.7634893207688334, "grad_norm": 1.5674758564857831, "learning_rate": 2.792988785962217e-06, "loss": 0.468, "step": 21430 }, { "epoch": 0.7635249478953275, "grad_norm": 1.189645514585024, "learning_rate": 2.7921888720510494e-06, "loss": 0.3546, "step": 21431 }, { "epoch": 0.7635605750218216, "grad_norm": 2.144026261789837, "learning_rate": 2.791389054116924e-06, "loss": 0.4886, "step": 21432 }, { "epoch": 0.7635962021483157, "grad_norm": 1.2827143911015846, "learning_rate": 2.7905893321704945e-06, "loss": 0.4082, "step": 21433 }, { "epoch": 0.7636318292748099, "grad_norm": 2.1268185322881425, "learning_rate": 2.7897897062224055e-06, "loss": 0.5928, "step": 21434 }, { "epoch": 0.763667456401304, "grad_norm": 1.6607947120914055, "learning_rate": 2.7889901762833082e-06, "loss": 0.3684, "step": 21435 }, { "epoch": 0.763703083527798, "grad_norm": 1.2161991406615227, "learning_rate": 2.788190742363844e-06, "loss": 0.4363, "step": 21436 }, { "epoch": 0.7637387106542922, "grad_norm": 1.6328178746095547, "learning_rate": 2.7873914044746607e-06, "loss": 0.5554, "step": 21437 }, { "epoch": 0.7637743377807863, "grad_norm": 1.4947140088318105, "learning_rate": 2.7865921626264023e-06, "loss": 0.5024, "step": 21438 }, { "epoch": 0.7638099649072804, "grad_norm": 4.648533243959248, "learning_rate": 2.7857930168297155e-06, "loss": 0.5797, "step": 21439 }, { "epoch": 0.7638455920337746, "grad_norm": 1.2282928242828763, "learning_rate": 2.7849939670952318e-06, "loss": 0.4234, "step": 21440 }, { "epoch": 0.7638812191602686, "grad_norm": 1.2923967214038787, "learning_rate": 2.7841950134336004e-06, "loss": 0.3954, "step": 21441 }, { "epoch": 0.7639168462867627, "grad_norm": 1.1054155903923626, "learning_rate": 2.7833961558554533e-06, "loss": 0.4177, "step": 21442 }, { "epoch": 0.7639524734132569, "grad_norm": 1.280487420471853, "learning_rate": 2.7825973943714336e-06, "loss": 0.3897, "step": 21443 }, { "epoch": 0.763988100539751, "grad_norm": 1.538898486900458, "learning_rate": 2.781798728992171e-06, "loss": 0.5236, "step": 21444 }, { "epoch": 0.764023727666245, "grad_norm": 1.3712486131653845, "learning_rate": 2.781000159728303e-06, "loss": 0.4914, "step": 21445 }, { "epoch": 0.7640593547927392, "grad_norm": 1.3975392171609484, "learning_rate": 2.7802016865904646e-06, "loss": 0.4822, "step": 21446 }, { "epoch": 0.7640949819192333, "grad_norm": 1.2905424620042656, "learning_rate": 2.779403309589286e-06, "loss": 0.4431, "step": 21447 }, { "epoch": 0.7641306090457274, "grad_norm": 1.1612352102955037, "learning_rate": 2.778605028735403e-06, "loss": 0.3909, "step": 21448 }, { "epoch": 0.7641662361722216, "grad_norm": 1.2256515725771069, "learning_rate": 2.777806844039438e-06, "loss": 0.4234, "step": 21449 }, { "epoch": 0.7642018632987156, "grad_norm": 1.2778650074910003, "learning_rate": 2.7770087555120263e-06, "loss": 0.415, "step": 21450 }, { "epoch": 0.7642374904252097, "grad_norm": 1.173978487998688, "learning_rate": 2.776210763163788e-06, "loss": 0.4186, "step": 21451 }, { "epoch": 0.7642731175517039, "grad_norm": 1.157454417869381, "learning_rate": 2.7754128670053536e-06, "loss": 0.4522, "step": 21452 }, { "epoch": 0.764308744678198, "grad_norm": 1.6322632610785124, "learning_rate": 2.774615067047346e-06, "loss": 0.5644, "step": 21453 }, { "epoch": 0.7643443718046921, "grad_norm": 1.4615492617843335, "learning_rate": 2.7738173633003918e-06, "loss": 0.5101, "step": 21454 }, { "epoch": 0.7643799989311862, "grad_norm": 1.184481928412922, "learning_rate": 2.773019755775107e-06, "loss": 0.4134, "step": 21455 }, { "epoch": 0.7644156260576803, "grad_norm": 1.5860596721052775, "learning_rate": 2.7722222444821156e-06, "loss": 0.6153, "step": 21456 }, { "epoch": 0.7644512531841744, "grad_norm": 1.6552854478365282, "learning_rate": 2.771424829432041e-06, "loss": 0.5252, "step": 21457 }, { "epoch": 0.7644868803106686, "grad_norm": 1.094846361932075, "learning_rate": 2.770627510635493e-06, "loss": 0.3631, "step": 21458 }, { "epoch": 0.7645225074371627, "grad_norm": 1.5463559504200093, "learning_rate": 2.7698302881030957e-06, "loss": 0.5599, "step": 21459 }, { "epoch": 0.7645581345636567, "grad_norm": 1.2454931307589208, "learning_rate": 2.769033161845457e-06, "loss": 0.4282, "step": 21460 }, { "epoch": 0.7645937616901509, "grad_norm": 1.2333007099125572, "learning_rate": 2.768236131873201e-06, "loss": 0.4214, "step": 21461 }, { "epoch": 0.764629388816645, "grad_norm": 1.3483890444209998, "learning_rate": 2.7674391981969327e-06, "loss": 0.4268, "step": 21462 }, { "epoch": 0.7646650159431391, "grad_norm": 1.2137217868423917, "learning_rate": 2.76664236082727e-06, "loss": 0.4393, "step": 21463 }, { "epoch": 0.7647006430696333, "grad_norm": 1.2097147505690504, "learning_rate": 2.7658456197748165e-06, "loss": 0.4299, "step": 21464 }, { "epoch": 0.7647362701961273, "grad_norm": 1.3912821954425267, "learning_rate": 2.7650489750501885e-06, "loss": 0.5754, "step": 21465 }, { "epoch": 0.7647718973226214, "grad_norm": 1.3416410303319324, "learning_rate": 2.764252426663987e-06, "loss": 0.4154, "step": 21466 }, { "epoch": 0.7648075244491156, "grad_norm": 2.471052335193522, "learning_rate": 2.763455974626822e-06, "loss": 0.4254, "step": 21467 }, { "epoch": 0.7648431515756097, "grad_norm": 1.6819247118921996, "learning_rate": 2.7626596189492983e-06, "loss": 0.5075, "step": 21468 }, { "epoch": 0.7648787787021037, "grad_norm": 1.1031553923663713, "learning_rate": 2.761863359642021e-06, "loss": 0.3802, "step": 21469 }, { "epoch": 0.7649144058285979, "grad_norm": 2.296837586622901, "learning_rate": 2.7610671967155943e-06, "loss": 0.5759, "step": 21470 }, { "epoch": 0.764950032955092, "grad_norm": 1.5305498269196727, "learning_rate": 2.7602711301806164e-06, "loss": 0.4578, "step": 21471 }, { "epoch": 0.7649856600815861, "grad_norm": 1.302294513463769, "learning_rate": 2.7594751600476898e-06, "loss": 0.5137, "step": 21472 }, { "epoch": 0.7650212872080803, "grad_norm": 1.720331363130525, "learning_rate": 2.7586792863274104e-06, "loss": 0.5681, "step": 21473 }, { "epoch": 0.7650569143345743, "grad_norm": 1.2545137211701975, "learning_rate": 2.757883509030377e-06, "loss": 0.4049, "step": 21474 }, { "epoch": 0.7650925414610684, "grad_norm": 1.1728914914162725, "learning_rate": 2.7570878281671875e-06, "loss": 0.4047, "step": 21475 }, { "epoch": 0.7651281685875626, "grad_norm": 1.4018944112936325, "learning_rate": 2.7562922437484384e-06, "loss": 0.5499, "step": 21476 }, { "epoch": 0.7651637957140567, "grad_norm": 1.2848067783144155, "learning_rate": 2.755496755784718e-06, "loss": 0.4425, "step": 21477 }, { "epoch": 0.7651994228405508, "grad_norm": 1.1449334383427936, "learning_rate": 2.754701364286625e-06, "loss": 0.3945, "step": 21478 }, { "epoch": 0.765235049967045, "grad_norm": 1.1295191697958777, "learning_rate": 2.7539060692647447e-06, "loss": 0.3697, "step": 21479 }, { "epoch": 0.765270677093539, "grad_norm": 1.2944890558383864, "learning_rate": 2.7531108707296695e-06, "loss": 0.4316, "step": 21480 }, { "epoch": 0.7653063042200331, "grad_norm": 1.0795536575864189, "learning_rate": 2.752315768691992e-06, "loss": 0.377, "step": 21481 }, { "epoch": 0.7653419313465273, "grad_norm": 1.5406159770934402, "learning_rate": 2.751520763162292e-06, "loss": 0.5094, "step": 21482 }, { "epoch": 0.7653775584730214, "grad_norm": 1.4720902487326863, "learning_rate": 2.7507258541511604e-06, "loss": 0.4546, "step": 21483 }, { "epoch": 0.7654131855995154, "grad_norm": 1.192943052004459, "learning_rate": 2.7499310416691806e-06, "loss": 0.4505, "step": 21484 }, { "epoch": 0.7654488127260096, "grad_norm": 1.623284493622015, "learning_rate": 2.749136325726939e-06, "loss": 0.472, "step": 21485 }, { "epoch": 0.7654844398525037, "grad_norm": 1.0969356792118432, "learning_rate": 2.7483417063350124e-06, "loss": 0.3971, "step": 21486 }, { "epoch": 0.7655200669789978, "grad_norm": 1.1275647010758107, "learning_rate": 2.747547183503988e-06, "loss": 0.4228, "step": 21487 }, { "epoch": 0.765555694105492, "grad_norm": 1.1735769529010647, "learning_rate": 2.7467527572444386e-06, "loss": 0.3948, "step": 21488 }, { "epoch": 0.765591321231986, "grad_norm": 1.4676810550026433, "learning_rate": 2.745958427566946e-06, "loss": 0.5801, "step": 21489 }, { "epoch": 0.7656269483584801, "grad_norm": 1.28271477054431, "learning_rate": 2.7451641944820874e-06, "loss": 0.4041, "step": 21490 }, { "epoch": 0.7656625754849743, "grad_norm": 1.4534740262703616, "learning_rate": 2.7443700580004383e-06, "loss": 0.5786, "step": 21491 }, { "epoch": 0.7656982026114684, "grad_norm": 1.5322830852808205, "learning_rate": 2.7435760181325767e-06, "loss": 0.525, "step": 21492 }, { "epoch": 0.7657338297379624, "grad_norm": 1.2754234788027627, "learning_rate": 2.742782074889069e-06, "loss": 0.5166, "step": 21493 }, { "epoch": 0.7657694568644566, "grad_norm": 1.3940295087597594, "learning_rate": 2.741988228280493e-06, "loss": 0.4469, "step": 21494 }, { "epoch": 0.7658050839909507, "grad_norm": 1.4141825741615877, "learning_rate": 2.741194478317414e-06, "loss": 0.4834, "step": 21495 }, { "epoch": 0.7658407111174448, "grad_norm": 1.1240044605427937, "learning_rate": 2.7404008250104075e-06, "loss": 0.3789, "step": 21496 }, { "epoch": 0.765876338243939, "grad_norm": 1.2174344902589156, "learning_rate": 2.739607268370035e-06, "loss": 0.4065, "step": 21497 }, { "epoch": 0.765911965370433, "grad_norm": 1.1683455782115673, "learning_rate": 2.738813808406866e-06, "loss": 0.3961, "step": 21498 }, { "epoch": 0.7659475924969271, "grad_norm": 1.5410125121452058, "learning_rate": 2.7380204451314674e-06, "loss": 0.3844, "step": 21499 }, { "epoch": 0.7659832196234213, "grad_norm": 1.6865728171672065, "learning_rate": 2.7372271785544046e-06, "loss": 0.5004, "step": 21500 }, { "epoch": 0.7660188467499154, "grad_norm": 1.6565985512663228, "learning_rate": 2.736434008686235e-06, "loss": 0.5855, "step": 21501 }, { "epoch": 0.7660544738764095, "grad_norm": 1.177007229539065, "learning_rate": 2.7356409355375246e-06, "loss": 0.3648, "step": 21502 }, { "epoch": 0.7660901010029036, "grad_norm": 2.4910213595954716, "learning_rate": 2.734847959118835e-06, "loss": 0.4809, "step": 21503 }, { "epoch": 0.7661257281293977, "grad_norm": 2.9522524274153352, "learning_rate": 2.7340550794407206e-06, "loss": 0.3863, "step": 21504 }, { "epoch": 0.7661613552558918, "grad_norm": 1.543607613238474, "learning_rate": 2.73326229651374e-06, "loss": 0.4967, "step": 21505 }, { "epoch": 0.766196982382386, "grad_norm": 1.7991577759076758, "learning_rate": 2.7324696103484517e-06, "loss": 0.5297, "step": 21506 }, { "epoch": 0.7662326095088801, "grad_norm": 1.3983140546334891, "learning_rate": 2.7316770209554135e-06, "loss": 0.4068, "step": 21507 }, { "epoch": 0.7662682366353741, "grad_norm": 1.1863983211499116, "learning_rate": 2.7308845283451725e-06, "loss": 0.4376, "step": 21508 }, { "epoch": 0.7663038637618683, "grad_norm": 1.3644924370940492, "learning_rate": 2.7300921325282883e-06, "loss": 0.5682, "step": 21509 }, { "epoch": 0.7663394908883624, "grad_norm": 1.441416685108155, "learning_rate": 2.729299833515305e-06, "loss": 0.5326, "step": 21510 }, { "epoch": 0.7663751180148565, "grad_norm": 1.1466140872629422, "learning_rate": 2.7285076313167815e-06, "loss": 0.4146, "step": 21511 }, { "epoch": 0.7664107451413507, "grad_norm": 1.2688207843437476, "learning_rate": 2.727715525943253e-06, "loss": 0.4504, "step": 21512 }, { "epoch": 0.7664463722678447, "grad_norm": 1.3977266179938241, "learning_rate": 2.7269235174052788e-06, "loss": 0.5278, "step": 21513 }, { "epoch": 0.7664819993943388, "grad_norm": 1.21048519058792, "learning_rate": 2.726131605713405e-06, "loss": 0.4437, "step": 21514 }, { "epoch": 0.766517626520833, "grad_norm": 1.4381984598977833, "learning_rate": 2.72533979087817e-06, "loss": 0.5872, "step": 21515 }, { "epoch": 0.7665532536473271, "grad_norm": 1.7813818604573763, "learning_rate": 2.7245480729101227e-06, "loss": 0.5234, "step": 21516 }, { "epoch": 0.7665888807738211, "grad_norm": 1.8794038598042633, "learning_rate": 2.723756451819801e-06, "loss": 0.5299, "step": 21517 }, { "epoch": 0.7666245079003153, "grad_norm": 1.2299367302196103, "learning_rate": 2.7229649276177504e-06, "loss": 0.4419, "step": 21518 }, { "epoch": 0.7666601350268094, "grad_norm": 1.2023716081707243, "learning_rate": 2.7221735003145054e-06, "loss": 0.4351, "step": 21519 }, { "epoch": 0.7666957621533035, "grad_norm": 1.454941593603134, "learning_rate": 2.7213821699206066e-06, "loss": 0.591, "step": 21520 }, { "epoch": 0.7667313892797977, "grad_norm": 1.1575022664612413, "learning_rate": 2.720590936446592e-06, "loss": 0.3952, "step": 21521 }, { "epoch": 0.7667670164062917, "grad_norm": 1.3924223155371007, "learning_rate": 2.7197997999030002e-06, "loss": 0.4745, "step": 21522 }, { "epoch": 0.7668026435327858, "grad_norm": 1.2611599834245188, "learning_rate": 2.719008760300359e-06, "loss": 0.4321, "step": 21523 }, { "epoch": 0.76683827065928, "grad_norm": 1.5072235585834393, "learning_rate": 2.718217817649209e-06, "loss": 0.5522, "step": 21524 }, { "epoch": 0.7668738977857741, "grad_norm": 1.769955897988482, "learning_rate": 2.7174269719600756e-06, "loss": 0.4971, "step": 21525 }, { "epoch": 0.7669095249122682, "grad_norm": 2.0123836513465108, "learning_rate": 2.7166362232434917e-06, "loss": 0.4905, "step": 21526 }, { "epoch": 0.7669451520387623, "grad_norm": 1.158319366460691, "learning_rate": 2.7158455715099872e-06, "loss": 0.355, "step": 21527 }, { "epoch": 0.7669807791652564, "grad_norm": 1.3730567968456515, "learning_rate": 2.7150550167700905e-06, "loss": 0.4719, "step": 21528 }, { "epoch": 0.7670164062917505, "grad_norm": 1.4844300489013609, "learning_rate": 2.714264559034331e-06, "loss": 0.4622, "step": 21529 }, { "epoch": 0.7670520334182447, "grad_norm": 1.7050785402000612, "learning_rate": 2.7134741983132283e-06, "loss": 0.552, "step": 21530 }, { "epoch": 0.7670876605447388, "grad_norm": 1.5332090929584241, "learning_rate": 2.7126839346173128e-06, "loss": 0.5406, "step": 21531 }, { "epoch": 0.7671232876712328, "grad_norm": 1.473695300068901, "learning_rate": 2.7118937679571013e-06, "loss": 0.55, "step": 21532 }, { "epoch": 0.767158914797727, "grad_norm": 1.4145686612407162, "learning_rate": 2.7111036983431214e-06, "loss": 0.3888, "step": 21533 }, { "epoch": 0.7671945419242211, "grad_norm": 1.3359831953509393, "learning_rate": 2.7103137257858867e-06, "loss": 0.4017, "step": 21534 }, { "epoch": 0.7672301690507152, "grad_norm": 1.9958369599315209, "learning_rate": 2.709523850295921e-06, "loss": 0.5654, "step": 21535 }, { "epoch": 0.7672657961772094, "grad_norm": 1.6283317395832735, "learning_rate": 2.7087340718837396e-06, "loss": 0.3926, "step": 21536 }, { "epoch": 0.7673014233037034, "grad_norm": 1.4819063425506112, "learning_rate": 2.707944390559861e-06, "loss": 0.5258, "step": 21537 }, { "epoch": 0.7673370504301975, "grad_norm": 1.174227539025942, "learning_rate": 2.707154806334803e-06, "loss": 0.42, "step": 21538 }, { "epoch": 0.7673726775566917, "grad_norm": 1.2208332291650417, "learning_rate": 2.706365319219073e-06, "loss": 0.4223, "step": 21539 }, { "epoch": 0.7674083046831858, "grad_norm": 1.5459489422562087, "learning_rate": 2.7055759292231896e-06, "loss": 0.4477, "step": 21540 }, { "epoch": 0.7674439318096798, "grad_norm": 1.423779362970767, "learning_rate": 2.704786636357659e-06, "loss": 0.5063, "step": 21541 }, { "epoch": 0.767479558936174, "grad_norm": 1.6043969443200001, "learning_rate": 2.7039974406329927e-06, "loss": 0.4568, "step": 21542 }, { "epoch": 0.7675151860626681, "grad_norm": 1.2593179163135468, "learning_rate": 2.7032083420597e-06, "loss": 0.4144, "step": 21543 }, { "epoch": 0.7675508131891622, "grad_norm": 1.2378810872677752, "learning_rate": 2.702419340648291e-06, "loss": 0.3906, "step": 21544 }, { "epoch": 0.7675864403156564, "grad_norm": 1.8611626754337367, "learning_rate": 2.7016304364092672e-06, "loss": 0.5154, "step": 21545 }, { "epoch": 0.7676220674421504, "grad_norm": 1.3204443044692264, "learning_rate": 2.700841629353137e-06, "loss": 0.4169, "step": 21546 }, { "epoch": 0.7676576945686445, "grad_norm": 1.3129178210002035, "learning_rate": 2.7000529194904e-06, "loss": 0.4703, "step": 21547 }, { "epoch": 0.7676933216951387, "grad_norm": 1.3725593591241732, "learning_rate": 2.6992643068315615e-06, "loss": 0.3962, "step": 21548 }, { "epoch": 0.7677289488216328, "grad_norm": 1.3871342770102786, "learning_rate": 2.6984757913871208e-06, "loss": 0.3915, "step": 21549 }, { "epoch": 0.7677645759481269, "grad_norm": 1.4604342056717028, "learning_rate": 2.6976873731675777e-06, "loss": 0.4781, "step": 21550 }, { "epoch": 0.767800203074621, "grad_norm": 1.5123293790222303, "learning_rate": 2.696899052183436e-06, "loss": 0.4192, "step": 21551 }, { "epoch": 0.7678358302011151, "grad_norm": 1.3785122079727907, "learning_rate": 2.6961108284451842e-06, "loss": 0.4233, "step": 21552 }, { "epoch": 0.7678714573276092, "grad_norm": 1.2182950958208372, "learning_rate": 2.695322701963324e-06, "loss": 0.4408, "step": 21553 }, { "epoch": 0.7679070844541034, "grad_norm": 1.2184934591112795, "learning_rate": 2.6945346727483458e-06, "loss": 0.3781, "step": 21554 }, { "epoch": 0.7679427115805975, "grad_norm": 1.1813723521206754, "learning_rate": 2.6937467408107476e-06, "loss": 0.3882, "step": 21555 }, { "epoch": 0.7679783387070915, "grad_norm": 1.4387927535684317, "learning_rate": 2.692958906161015e-06, "loss": 0.4953, "step": 21556 }, { "epoch": 0.7680139658335857, "grad_norm": 1.5618488441404195, "learning_rate": 2.692171168809642e-06, "loss": 0.5714, "step": 21557 }, { "epoch": 0.7680495929600798, "grad_norm": 1.2169293410692044, "learning_rate": 2.6913835287671177e-06, "loss": 0.3805, "step": 21558 }, { "epoch": 0.7680852200865739, "grad_norm": 1.5703941122282277, "learning_rate": 2.6905959860439303e-06, "loss": 0.4853, "step": 21559 }, { "epoch": 0.7681208472130681, "grad_norm": 1.3516313404821285, "learning_rate": 2.689808540650568e-06, "loss": 0.5025, "step": 21560 }, { "epoch": 0.7681564743395621, "grad_norm": 1.178356023967142, "learning_rate": 2.689021192597513e-06, "loss": 0.5302, "step": 21561 }, { "epoch": 0.7681921014660562, "grad_norm": 1.2464392497819148, "learning_rate": 2.6882339418952542e-06, "loss": 0.5245, "step": 21562 }, { "epoch": 0.7682277285925504, "grad_norm": 1.1488656306856904, "learning_rate": 2.6874467885542664e-06, "loss": 0.3843, "step": 21563 }, { "epoch": 0.7682633557190445, "grad_norm": 1.563361645516565, "learning_rate": 2.6866597325850373e-06, "loss": 0.5033, "step": 21564 }, { "epoch": 0.7682989828455385, "grad_norm": 2.3205415733266777, "learning_rate": 2.685872773998045e-06, "loss": 0.4215, "step": 21565 }, { "epoch": 0.7683346099720327, "grad_norm": 1.6349512101362746, "learning_rate": 2.685085912803772e-06, "loss": 0.498, "step": 21566 }, { "epoch": 0.7683702370985268, "grad_norm": 1.3082337765289636, "learning_rate": 2.6842991490126903e-06, "loss": 0.3776, "step": 21567 }, { "epoch": 0.7684058642250209, "grad_norm": 1.1732568086218953, "learning_rate": 2.683512482635281e-06, "loss": 0.4019, "step": 21568 }, { "epoch": 0.7684414913515151, "grad_norm": 1.1846530074199122, "learning_rate": 2.682725913682014e-06, "loss": 0.4209, "step": 21569 }, { "epoch": 0.7684771184780091, "grad_norm": 1.2773286942516142, "learning_rate": 2.681939442163369e-06, "loss": 0.4086, "step": 21570 }, { "epoch": 0.7685127456045032, "grad_norm": 1.4880900904307526, "learning_rate": 2.6811530680898133e-06, "loss": 0.4787, "step": 21571 }, { "epoch": 0.7685483727309974, "grad_norm": 1.5588916333956027, "learning_rate": 2.680366791471819e-06, "loss": 0.5413, "step": 21572 }, { "epoch": 0.7685839998574915, "grad_norm": 1.1242224383983777, "learning_rate": 2.6795806123198575e-06, "loss": 0.3909, "step": 21573 }, { "epoch": 0.7686196269839856, "grad_norm": 1.4372779493940746, "learning_rate": 2.6787945306443963e-06, "loss": 0.5574, "step": 21574 }, { "epoch": 0.7686552541104797, "grad_norm": 1.2420738650647816, "learning_rate": 2.678008546455907e-06, "loss": 0.4745, "step": 21575 }, { "epoch": 0.7686908812369738, "grad_norm": 1.585199591324763, "learning_rate": 2.6772226597648476e-06, "loss": 0.5601, "step": 21576 }, { "epoch": 0.7687265083634679, "grad_norm": 1.3841133790381688, "learning_rate": 2.67643687058169e-06, "loss": 0.4658, "step": 21577 }, { "epoch": 0.7687621354899621, "grad_norm": 1.8414849210525404, "learning_rate": 2.6756511789168926e-06, "loss": 0.4727, "step": 21578 }, { "epoch": 0.7687977626164562, "grad_norm": 1.7423064355437665, "learning_rate": 2.674865584780918e-06, "loss": 0.3859, "step": 21579 }, { "epoch": 0.7688333897429502, "grad_norm": 1.545073386763291, "learning_rate": 2.6740800881842277e-06, "loss": 0.4995, "step": 21580 }, { "epoch": 0.7688690168694444, "grad_norm": 1.8292404454390374, "learning_rate": 2.673294689137286e-06, "loss": 0.6008, "step": 21581 }, { "epoch": 0.7689046439959385, "grad_norm": 1.1590143614998853, "learning_rate": 2.6725093876505426e-06, "loss": 0.4089, "step": 21582 }, { "epoch": 0.7689402711224326, "grad_norm": 1.5907840139904812, "learning_rate": 2.671724183734459e-06, "loss": 0.6039, "step": 21583 }, { "epoch": 0.7689758982489268, "grad_norm": 1.5255429188026082, "learning_rate": 2.6709390773994915e-06, "loss": 0.4933, "step": 21584 }, { "epoch": 0.7690115253754208, "grad_norm": 1.5342830892634645, "learning_rate": 2.670154068656091e-06, "loss": 0.509, "step": 21585 }, { "epoch": 0.7690471525019149, "grad_norm": 1.5179452221331364, "learning_rate": 2.6693691575147153e-06, "loss": 0.5586, "step": 21586 }, { "epoch": 0.7690827796284091, "grad_norm": 1.1643986931507768, "learning_rate": 2.6685843439858072e-06, "loss": 0.4027, "step": 21587 }, { "epoch": 0.7691184067549032, "grad_norm": 1.0712209168351075, "learning_rate": 2.667799628079829e-06, "loss": 0.3948, "step": 21588 }, { "epoch": 0.7691540338813972, "grad_norm": 1.3315304651023185, "learning_rate": 2.66701500980722e-06, "loss": 0.451, "step": 21589 }, { "epoch": 0.7691896610078914, "grad_norm": 1.2350766305631622, "learning_rate": 2.666230489178435e-06, "loss": 0.4112, "step": 21590 }, { "epoch": 0.7692252881343855, "grad_norm": 1.2585804911084773, "learning_rate": 2.665446066203914e-06, "loss": 0.4196, "step": 21591 }, { "epoch": 0.7692609152608796, "grad_norm": 1.7720246151166985, "learning_rate": 2.6646617408941087e-06, "loss": 0.5739, "step": 21592 }, { "epoch": 0.7692965423873738, "grad_norm": 1.6532881739117002, "learning_rate": 2.663877513259455e-06, "loss": 0.6261, "step": 21593 }, { "epoch": 0.7693321695138678, "grad_norm": 1.6613537460832681, "learning_rate": 2.663093383310401e-06, "loss": 0.5044, "step": 21594 }, { "epoch": 0.7693677966403619, "grad_norm": 1.2918292333608354, "learning_rate": 2.6623093510573874e-06, "loss": 0.3738, "step": 21595 }, { "epoch": 0.7694034237668561, "grad_norm": 1.4608288600573949, "learning_rate": 2.661525416510853e-06, "loss": 0.5217, "step": 21596 }, { "epoch": 0.7694390508933502, "grad_norm": 1.6234825020698576, "learning_rate": 2.6607415796812385e-06, "loss": 0.5808, "step": 21597 }, { "epoch": 0.7694746780198443, "grad_norm": 1.2324190205768177, "learning_rate": 2.6599578405789783e-06, "loss": 0.4252, "step": 21598 }, { "epoch": 0.7695103051463384, "grad_norm": 1.5031837133635095, "learning_rate": 2.6591741992145128e-06, "loss": 0.5152, "step": 21599 }, { "epoch": 0.7695459322728325, "grad_norm": 1.8189025079572245, "learning_rate": 2.65839065559827e-06, "loss": 0.5129, "step": 21600 }, { "epoch": 0.7695815593993266, "grad_norm": 1.4593457342246374, "learning_rate": 2.657607209740688e-06, "loss": 0.5469, "step": 21601 }, { "epoch": 0.7696171865258208, "grad_norm": 1.4981023656043566, "learning_rate": 2.656823861652197e-06, "loss": 0.4268, "step": 21602 }, { "epoch": 0.7696528136523149, "grad_norm": 1.3064481668003745, "learning_rate": 2.6560406113432323e-06, "loss": 0.3977, "step": 21603 }, { "epoch": 0.7696884407788089, "grad_norm": 1.3532811361403154, "learning_rate": 2.6552574588242175e-06, "loss": 0.3776, "step": 21604 }, { "epoch": 0.7697240679053031, "grad_norm": 1.92024093512963, "learning_rate": 2.6544744041055857e-06, "loss": 0.5125, "step": 21605 }, { "epoch": 0.7697596950317972, "grad_norm": 1.1953247559910143, "learning_rate": 2.653691447197758e-06, "loss": 0.4123, "step": 21606 }, { "epoch": 0.7697953221582913, "grad_norm": 1.65417345900349, "learning_rate": 2.6529085881111636e-06, "loss": 0.5128, "step": 21607 }, { "epoch": 0.7698309492847855, "grad_norm": 1.1990183433834911, "learning_rate": 2.652125826856231e-06, "loss": 0.4558, "step": 21608 }, { "epoch": 0.7698665764112795, "grad_norm": 1.5693883871290772, "learning_rate": 2.651343163443375e-06, "loss": 0.3979, "step": 21609 }, { "epoch": 0.7699022035377736, "grad_norm": 1.5863762157155212, "learning_rate": 2.6505605978830206e-06, "loss": 0.5307, "step": 21610 }, { "epoch": 0.7699378306642678, "grad_norm": 1.4586296161197037, "learning_rate": 2.6497781301855897e-06, "loss": 0.5143, "step": 21611 }, { "epoch": 0.7699734577907619, "grad_norm": 1.2916074940168243, "learning_rate": 2.6489957603615025e-06, "loss": 0.4284, "step": 21612 }, { "epoch": 0.770009084917256, "grad_norm": 1.2594286227712657, "learning_rate": 2.648213488421173e-06, "loss": 0.4258, "step": 21613 }, { "epoch": 0.7700447120437501, "grad_norm": 1.3630409007030637, "learning_rate": 2.647431314375022e-06, "loss": 0.6433, "step": 21614 }, { "epoch": 0.7700803391702442, "grad_norm": 1.560009676862841, "learning_rate": 2.6466492382334587e-06, "loss": 0.4876, "step": 21615 }, { "epoch": 0.7701159662967383, "grad_norm": 1.274618857865421, "learning_rate": 2.6458672600069014e-06, "loss": 0.4179, "step": 21616 }, { "epoch": 0.7701515934232325, "grad_norm": 1.5331412092086592, "learning_rate": 2.645085379705761e-06, "loss": 0.5304, "step": 21617 }, { "epoch": 0.7701872205497265, "grad_norm": 1.180826989868072, "learning_rate": 2.6443035973404497e-06, "loss": 0.4299, "step": 21618 }, { "epoch": 0.7702228476762206, "grad_norm": 1.4853607619734635, "learning_rate": 2.6435219129213796e-06, "loss": 0.4515, "step": 21619 }, { "epoch": 0.7702584748027148, "grad_norm": 1.5999718053062184, "learning_rate": 2.6427403264589556e-06, "loss": 0.4927, "step": 21620 }, { "epoch": 0.7702941019292089, "grad_norm": 1.2119458544108175, "learning_rate": 2.6419588379635884e-06, "loss": 0.4288, "step": 21621 }, { "epoch": 0.770329729055703, "grad_norm": 1.3163585053242841, "learning_rate": 2.6411774474456798e-06, "loss": 0.4389, "step": 21622 }, { "epoch": 0.7703653561821971, "grad_norm": 1.3058409430633393, "learning_rate": 2.6403961549156396e-06, "loss": 0.4343, "step": 21623 }, { "epoch": 0.7704009833086912, "grad_norm": 1.3054138515187088, "learning_rate": 2.639614960383864e-06, "loss": 0.4131, "step": 21624 }, { "epoch": 0.7704366104351853, "grad_norm": 1.2152085065412466, "learning_rate": 2.638833863860765e-06, "loss": 0.3753, "step": 21625 }, { "epoch": 0.7704722375616795, "grad_norm": 1.5801398724515907, "learning_rate": 2.638052865356734e-06, "loss": 0.5043, "step": 21626 }, { "epoch": 0.7705078646881736, "grad_norm": 1.3900655189070827, "learning_rate": 2.6372719648821785e-06, "loss": 0.5153, "step": 21627 }, { "epoch": 0.7705434918146676, "grad_norm": 1.220666669824554, "learning_rate": 2.636491162447491e-06, "loss": 0.4078, "step": 21628 }, { "epoch": 0.7705791189411618, "grad_norm": 1.0993132645631805, "learning_rate": 2.6357104580630697e-06, "loss": 0.3854, "step": 21629 }, { "epoch": 0.7706147460676559, "grad_norm": 1.4331766058472724, "learning_rate": 2.634929851739314e-06, "loss": 0.4313, "step": 21630 }, { "epoch": 0.77065037319415, "grad_norm": 1.3153485393901976, "learning_rate": 2.634149343486612e-06, "loss": 0.3894, "step": 21631 }, { "epoch": 0.7706860003206442, "grad_norm": 1.298680342874813, "learning_rate": 2.63336893331536e-06, "loss": 0.4413, "step": 21632 }, { "epoch": 0.7707216274471382, "grad_norm": 1.2348084833142572, "learning_rate": 2.6325886212359496e-06, "loss": 0.4209, "step": 21633 }, { "epoch": 0.7707572545736323, "grad_norm": 1.1882933618835327, "learning_rate": 2.6318084072587745e-06, "loss": 0.4133, "step": 21634 }, { "epoch": 0.7707928817001265, "grad_norm": 1.257891256926966, "learning_rate": 2.6310282913942177e-06, "loss": 0.4046, "step": 21635 }, { "epoch": 0.7708285088266206, "grad_norm": 1.2456997465081414, "learning_rate": 2.6302482736526723e-06, "loss": 0.457, "step": 21636 }, { "epoch": 0.7708641359531146, "grad_norm": 1.2053610533157464, "learning_rate": 2.62946835404452e-06, "loss": 0.4752, "step": 21637 }, { "epoch": 0.7708997630796088, "grad_norm": 1.3965574670178689, "learning_rate": 2.628688532580148e-06, "loss": 0.4885, "step": 21638 }, { "epoch": 0.7709353902061029, "grad_norm": 1.1427159757557714, "learning_rate": 2.62790880926994e-06, "loss": 0.3999, "step": 21639 }, { "epoch": 0.770971017332597, "grad_norm": 1.6045779322330116, "learning_rate": 2.6271291841242786e-06, "loss": 0.4625, "step": 21640 }, { "epoch": 0.7710066444590912, "grad_norm": 1.5858100297717113, "learning_rate": 2.6263496571535495e-06, "loss": 0.6304, "step": 21641 }, { "epoch": 0.7710422715855852, "grad_norm": 1.4511860472189297, "learning_rate": 2.625570228368125e-06, "loss": 0.4097, "step": 21642 }, { "epoch": 0.7710778987120793, "grad_norm": 1.6446695540555343, "learning_rate": 2.624790897778391e-06, "loss": 0.5576, "step": 21643 }, { "epoch": 0.7711135258385735, "grad_norm": 1.9474021226849454, "learning_rate": 2.6240116653947166e-06, "loss": 0.4971, "step": 21644 }, { "epoch": 0.7711491529650676, "grad_norm": 1.4652016169599549, "learning_rate": 2.6232325312274865e-06, "loss": 0.4567, "step": 21645 }, { "epoch": 0.7711847800915617, "grad_norm": 1.2885567802439557, "learning_rate": 2.622453495287068e-06, "loss": 0.4212, "step": 21646 }, { "epoch": 0.7712204072180558, "grad_norm": 1.6728724263180046, "learning_rate": 2.621674557583839e-06, "loss": 0.4813, "step": 21647 }, { "epoch": 0.7712560343445499, "grad_norm": 1.1047799150814426, "learning_rate": 2.6208957181281693e-06, "loss": 0.4152, "step": 21648 }, { "epoch": 0.771291661471044, "grad_norm": 1.5614616163410524, "learning_rate": 2.6201169769304334e-06, "loss": 0.6081, "step": 21649 }, { "epoch": 0.7713272885975382, "grad_norm": 1.1896810071874615, "learning_rate": 2.619338334000996e-06, "loss": 0.4153, "step": 21650 }, { "epoch": 0.7713629157240323, "grad_norm": 1.516480714709986, "learning_rate": 2.6185597893502303e-06, "loss": 0.4773, "step": 21651 }, { "epoch": 0.7713985428505263, "grad_norm": 1.3390725798401786, "learning_rate": 2.617781342988497e-06, "loss": 0.4173, "step": 21652 }, { "epoch": 0.7714341699770205, "grad_norm": 1.5847472104464, "learning_rate": 2.6170029949261655e-06, "loss": 0.5934, "step": 21653 }, { "epoch": 0.7714697971035146, "grad_norm": 1.3441101308852725, "learning_rate": 2.616224745173599e-06, "loss": 0.4115, "step": 21654 }, { "epoch": 0.7715054242300087, "grad_norm": 1.5339603762318912, "learning_rate": 2.615446593741161e-06, "loss": 0.5596, "step": 21655 }, { "epoch": 0.7715410513565029, "grad_norm": 1.9052861341583245, "learning_rate": 2.614668540639216e-06, "loss": 0.4871, "step": 21656 }, { "epoch": 0.7715766784829969, "grad_norm": 1.2007503222267697, "learning_rate": 2.6138905858781183e-06, "loss": 0.3806, "step": 21657 }, { "epoch": 0.771612305609491, "grad_norm": 1.5143319777506716, "learning_rate": 2.6131127294682334e-06, "loss": 0.4364, "step": 21658 }, { "epoch": 0.7716479327359852, "grad_norm": 1.21364730853512, "learning_rate": 2.612334971419913e-06, "loss": 0.4143, "step": 21659 }, { "epoch": 0.7716835598624793, "grad_norm": 1.1930828898895574, "learning_rate": 2.6115573117435187e-06, "loss": 0.3907, "step": 21660 }, { "epoch": 0.7717191869889733, "grad_norm": 1.7470737147095534, "learning_rate": 2.6107797504494e-06, "loss": 0.5026, "step": 21661 }, { "epoch": 0.7717548141154675, "grad_norm": 1.7381556785281227, "learning_rate": 2.6100022875479138e-06, "loss": 0.5156, "step": 21662 }, { "epoch": 0.7717904412419616, "grad_norm": 1.3014480534977682, "learning_rate": 2.6092249230494128e-06, "loss": 0.4331, "step": 21663 }, { "epoch": 0.7718260683684558, "grad_norm": 1.2580095204611035, "learning_rate": 2.608447656964247e-06, "loss": 0.3947, "step": 21664 }, { "epoch": 0.7718616954949499, "grad_norm": 1.2920055504979093, "learning_rate": 2.6076704893027703e-06, "loss": 0.403, "step": 21665 }, { "epoch": 0.771897322621444, "grad_norm": 1.3460135754379343, "learning_rate": 2.6068934200753248e-06, "loss": 0.4319, "step": 21666 }, { "epoch": 0.7719329497479381, "grad_norm": 1.2150108031401556, "learning_rate": 2.6061164492922642e-06, "loss": 0.3641, "step": 21667 }, { "epoch": 0.7719685768744322, "grad_norm": 1.6808917058556265, "learning_rate": 2.605339576963929e-06, "loss": 0.5114, "step": 21668 }, { "epoch": 0.7720042040009263, "grad_norm": 1.3093904197720778, "learning_rate": 2.6045628031006642e-06, "loss": 0.4449, "step": 21669 }, { "epoch": 0.7720398311274205, "grad_norm": 1.6768130650177486, "learning_rate": 2.603786127712815e-06, "loss": 0.578, "step": 21670 }, { "epoch": 0.7720754582539145, "grad_norm": 1.560244023337313, "learning_rate": 2.603009550810728e-06, "loss": 0.524, "step": 21671 }, { "epoch": 0.7721110853804086, "grad_norm": 1.2116282608970381, "learning_rate": 2.602233072404734e-06, "loss": 0.385, "step": 21672 }, { "epoch": 0.7721467125069028, "grad_norm": 1.1181127713818824, "learning_rate": 2.601456692505181e-06, "loss": 0.3574, "step": 21673 }, { "epoch": 0.7721823396333969, "grad_norm": 1.2114121441019063, "learning_rate": 2.6006804111224006e-06, "loss": 0.48, "step": 21674 }, { "epoch": 0.772217966759891, "grad_norm": 1.4875727662742646, "learning_rate": 2.599904228266732e-06, "loss": 0.4781, "step": 21675 }, { "epoch": 0.7722535938863851, "grad_norm": 1.0967234590494308, "learning_rate": 2.5991281439485117e-06, "loss": 0.3806, "step": 21676 }, { "epoch": 0.7722892210128792, "grad_norm": 1.3161420865235258, "learning_rate": 2.5983521581780724e-06, "loss": 0.4437, "step": 21677 }, { "epoch": 0.7723248481393733, "grad_norm": 1.0678350995469785, "learning_rate": 2.5975762709657506e-06, "loss": 0.401, "step": 21678 }, { "epoch": 0.7723604752658675, "grad_norm": 1.1919292621109563, "learning_rate": 2.596800482321872e-06, "loss": 0.4811, "step": 21679 }, { "epoch": 0.7723961023923616, "grad_norm": 1.3432865664049882, "learning_rate": 2.5960247922567716e-06, "loss": 0.3805, "step": 21680 }, { "epoch": 0.7724317295188556, "grad_norm": 1.1350681408548446, "learning_rate": 2.595249200780774e-06, "loss": 0.393, "step": 21681 }, { "epoch": 0.7724673566453498, "grad_norm": 1.4553101209726134, "learning_rate": 2.5944737079042114e-06, "loss": 0.5393, "step": 21682 }, { "epoch": 0.7725029837718439, "grad_norm": 1.3240860166841328, "learning_rate": 2.593698313637405e-06, "loss": 0.447, "step": 21683 }, { "epoch": 0.772538610898338, "grad_norm": 1.42311119073296, "learning_rate": 2.5929230179906827e-06, "loss": 0.4688, "step": 21684 }, { "epoch": 0.7725742380248322, "grad_norm": 1.0656434261005256, "learning_rate": 2.5921478209743665e-06, "loss": 0.4041, "step": 21685 }, { "epoch": 0.7726098651513262, "grad_norm": 1.6750844366467124, "learning_rate": 2.5913727225987805e-06, "loss": 0.4568, "step": 21686 }, { "epoch": 0.7726454922778203, "grad_norm": 1.328125736489162, "learning_rate": 2.5905977228742483e-06, "loss": 0.3876, "step": 21687 }, { "epoch": 0.7726811194043145, "grad_norm": 1.6127506621731666, "learning_rate": 2.5898228218110834e-06, "loss": 0.6029, "step": 21688 }, { "epoch": 0.7727167465308086, "grad_norm": 1.2623405179887135, "learning_rate": 2.58904801941961e-06, "loss": 0.4756, "step": 21689 }, { "epoch": 0.7727523736573026, "grad_norm": 1.172863921959009, "learning_rate": 2.588273315710138e-06, "loss": 0.4207, "step": 21690 }, { "epoch": 0.7727880007837968, "grad_norm": 1.5894162466862072, "learning_rate": 2.587498710692988e-06, "loss": 0.4871, "step": 21691 }, { "epoch": 0.7728236279102909, "grad_norm": 1.1275210658264148, "learning_rate": 2.5867242043784745e-06, "loss": 0.4013, "step": 21692 }, { "epoch": 0.772859255036785, "grad_norm": 1.5294491023774104, "learning_rate": 2.585949796776912e-06, "loss": 0.5719, "step": 21693 }, { "epoch": 0.7728948821632792, "grad_norm": 1.5615576905025965, "learning_rate": 2.585175487898607e-06, "loss": 0.5141, "step": 21694 }, { "epoch": 0.7729305092897732, "grad_norm": 1.5563090871663328, "learning_rate": 2.5844012777538763e-06, "loss": 0.5534, "step": 21695 }, { "epoch": 0.7729661364162673, "grad_norm": 1.2845932404122127, "learning_rate": 2.583627166353022e-06, "loss": 0.4368, "step": 21696 }, { "epoch": 0.7730017635427615, "grad_norm": 1.4094998254120494, "learning_rate": 2.5828531537063595e-06, "loss": 0.4136, "step": 21697 }, { "epoch": 0.7730373906692556, "grad_norm": 1.2396183716341125, "learning_rate": 2.582079239824188e-06, "loss": 0.4163, "step": 21698 }, { "epoch": 0.7730730177957497, "grad_norm": 1.2633720910897435, "learning_rate": 2.581305424716817e-06, "loss": 0.3872, "step": 21699 }, { "epoch": 0.7731086449222438, "grad_norm": 1.670804672440489, "learning_rate": 2.580531708394549e-06, "loss": 0.5348, "step": 21700 }, { "epoch": 0.7731442720487379, "grad_norm": 1.250385446042982, "learning_rate": 2.579758090867688e-06, "loss": 0.3755, "step": 21701 }, { "epoch": 0.773179899175232, "grad_norm": 1.6064764029474377, "learning_rate": 2.578984572146537e-06, "loss": 0.4204, "step": 21702 }, { "epoch": 0.7732155263017262, "grad_norm": 1.09372964125538, "learning_rate": 2.57821115224139e-06, "loss": 0.4086, "step": 21703 }, { "epoch": 0.7732511534282203, "grad_norm": 1.470870472347976, "learning_rate": 2.5774378311625526e-06, "loss": 0.4799, "step": 21704 }, { "epoch": 0.7732867805547143, "grad_norm": 1.1568442743244312, "learning_rate": 2.576664608920315e-06, "loss": 0.3638, "step": 21705 }, { "epoch": 0.7733224076812085, "grad_norm": 1.4720796481649545, "learning_rate": 2.575891485524977e-06, "loss": 0.5856, "step": 21706 }, { "epoch": 0.7733580348077026, "grad_norm": 1.3846481364705412, "learning_rate": 2.575118460986833e-06, "loss": 0.5293, "step": 21707 }, { "epoch": 0.7733936619341967, "grad_norm": 1.1586042398359897, "learning_rate": 2.5743455353161794e-06, "loss": 0.3943, "step": 21708 }, { "epoch": 0.7734292890606909, "grad_norm": 1.1654003712030214, "learning_rate": 2.573572708523302e-06, "loss": 0.4271, "step": 21709 }, { "epoch": 0.7734649161871849, "grad_norm": 1.511123434146626, "learning_rate": 2.5727999806184947e-06, "loss": 0.4884, "step": 21710 }, { "epoch": 0.773500543313679, "grad_norm": 1.20017394194489, "learning_rate": 2.5720273516120507e-06, "loss": 0.3789, "step": 21711 }, { "epoch": 0.7735361704401732, "grad_norm": 1.30383962433921, "learning_rate": 2.571254821514251e-06, "loss": 0.5011, "step": 21712 }, { "epoch": 0.7735717975666673, "grad_norm": 1.1765880576365992, "learning_rate": 2.570482390335385e-06, "loss": 0.4192, "step": 21713 }, { "epoch": 0.7736074246931613, "grad_norm": 1.180250912994889, "learning_rate": 2.5697100580857392e-06, "loss": 0.4527, "step": 21714 }, { "epoch": 0.7736430518196555, "grad_norm": 1.3685307151101365, "learning_rate": 2.5689378247756015e-06, "loss": 0.523, "step": 21715 }, { "epoch": 0.7736786789461496, "grad_norm": 1.35681269628412, "learning_rate": 2.568165690415246e-06, "loss": 0.4247, "step": 21716 }, { "epoch": 0.7737143060726437, "grad_norm": 1.3283029814320315, "learning_rate": 2.567393655014964e-06, "loss": 0.5078, "step": 21717 }, { "epoch": 0.7737499331991379, "grad_norm": 1.621594069713787, "learning_rate": 2.566621718585026e-06, "loss": 0.5422, "step": 21718 }, { "epoch": 0.773785560325632, "grad_norm": 1.1164595104776849, "learning_rate": 2.565849881135719e-06, "loss": 0.39, "step": 21719 }, { "epoch": 0.773821187452126, "grad_norm": 1.1897450315619311, "learning_rate": 2.565078142677313e-06, "loss": 0.3815, "step": 21720 }, { "epoch": 0.7738568145786202, "grad_norm": 1.1772994694887684, "learning_rate": 2.56430650322009e-06, "loss": 0.3911, "step": 21721 }, { "epoch": 0.7738924417051143, "grad_norm": 1.1368623363913175, "learning_rate": 2.563534962774321e-06, "loss": 0.3726, "step": 21722 }, { "epoch": 0.7739280688316084, "grad_norm": 1.898120793304696, "learning_rate": 2.5627635213502832e-06, "loss": 0.5886, "step": 21723 }, { "epoch": 0.7739636959581025, "grad_norm": 1.1868294231484688, "learning_rate": 2.56199217895825e-06, "loss": 0.3921, "step": 21724 }, { "epoch": 0.7739993230845966, "grad_norm": 1.1463669803864724, "learning_rate": 2.5612209356084873e-06, "loss": 0.3812, "step": 21725 }, { "epoch": 0.7740349502110907, "grad_norm": 1.4515287589442596, "learning_rate": 2.56044979131127e-06, "loss": 0.5193, "step": 21726 }, { "epoch": 0.7740705773375849, "grad_norm": 1.4009980003928377, "learning_rate": 2.55967874607686e-06, "loss": 0.5004, "step": 21727 }, { "epoch": 0.774106204464079, "grad_norm": 1.5777824490521524, "learning_rate": 2.558907799915529e-06, "loss": 0.4838, "step": 21728 }, { "epoch": 0.774141831590573, "grad_norm": 1.2108156575567632, "learning_rate": 2.558136952837541e-06, "loss": 0.4231, "step": 21729 }, { "epoch": 0.7741774587170672, "grad_norm": 1.0817146580120345, "learning_rate": 2.5573662048531643e-06, "loss": 0.437, "step": 21730 }, { "epoch": 0.7742130858435613, "grad_norm": 1.4030834839933062, "learning_rate": 2.556595555972655e-06, "loss": 0.4555, "step": 21731 }, { "epoch": 0.7742487129700554, "grad_norm": 1.2912701587500837, "learning_rate": 2.5558250062062826e-06, "loss": 0.3766, "step": 21732 }, { "epoch": 0.7742843400965496, "grad_norm": 1.1511749731073133, "learning_rate": 2.555054555564299e-06, "loss": 0.4294, "step": 21733 }, { "epoch": 0.7743199672230436, "grad_norm": 1.5945228450356168, "learning_rate": 2.5542842040569695e-06, "loss": 0.5945, "step": 21734 }, { "epoch": 0.7743555943495377, "grad_norm": 1.5446836246454811, "learning_rate": 2.553513951694552e-06, "loss": 0.492, "step": 21735 }, { "epoch": 0.7743912214760319, "grad_norm": 1.1455378468296784, "learning_rate": 2.552743798487298e-06, "loss": 0.4048, "step": 21736 }, { "epoch": 0.774426848602526, "grad_norm": 1.395916924749015, "learning_rate": 2.551973744445466e-06, "loss": 0.3427, "step": 21737 }, { "epoch": 0.77446247572902, "grad_norm": 1.8916419901304755, "learning_rate": 2.5512037895793086e-06, "loss": 0.6368, "step": 21738 }, { "epoch": 0.7744981028555142, "grad_norm": 1.3941201286552989, "learning_rate": 2.5504339338990835e-06, "loss": 0.4322, "step": 21739 }, { "epoch": 0.7745337299820083, "grad_norm": 1.5308067789169924, "learning_rate": 2.5496641774150333e-06, "loss": 0.5186, "step": 21740 }, { "epoch": 0.7745693571085024, "grad_norm": 1.204591639921633, "learning_rate": 2.548894520137416e-06, "loss": 0.4372, "step": 21741 }, { "epoch": 0.7746049842349966, "grad_norm": 1.5096565367693147, "learning_rate": 2.5481249620764726e-06, "loss": 0.5274, "step": 21742 }, { "epoch": 0.7746406113614906, "grad_norm": 1.6593267657570798, "learning_rate": 2.5473555032424534e-06, "loss": 0.5784, "step": 21743 }, { "epoch": 0.7746762384879847, "grad_norm": 1.444878119728114, "learning_rate": 2.5465861436456053e-06, "loss": 0.4214, "step": 21744 }, { "epoch": 0.7747118656144789, "grad_norm": 1.4129260809111164, "learning_rate": 2.545816883296173e-06, "loss": 0.466, "step": 21745 }, { "epoch": 0.774747492740973, "grad_norm": 1.1287245819250147, "learning_rate": 2.545047722204401e-06, "loss": 0.3796, "step": 21746 }, { "epoch": 0.7747831198674671, "grad_norm": 1.6183526573930227, "learning_rate": 2.544278660380527e-06, "loss": 0.5819, "step": 21747 }, { "epoch": 0.7748187469939612, "grad_norm": 1.4779957719538284, "learning_rate": 2.543509697834796e-06, "loss": 0.5082, "step": 21748 }, { "epoch": 0.7748543741204553, "grad_norm": 1.42690735073314, "learning_rate": 2.542740834577443e-06, "loss": 0.4139, "step": 21749 }, { "epoch": 0.7748900012469494, "grad_norm": 1.5509925807361074, "learning_rate": 2.541972070618711e-06, "loss": 0.4542, "step": 21750 }, { "epoch": 0.7749256283734436, "grad_norm": 1.434833641464831, "learning_rate": 2.5412034059688273e-06, "loss": 0.5346, "step": 21751 }, { "epoch": 0.7749612554999377, "grad_norm": 1.3851702937581356, "learning_rate": 2.54043484063804e-06, "loss": 0.5289, "step": 21752 }, { "epoch": 0.7749968826264317, "grad_norm": 1.320923173797415, "learning_rate": 2.5396663746365734e-06, "loss": 0.4159, "step": 21753 }, { "epoch": 0.7750325097529259, "grad_norm": 2.9583281464139937, "learning_rate": 2.5388980079746673e-06, "loss": 0.4834, "step": 21754 }, { "epoch": 0.77506813687942, "grad_norm": 1.7678418129768092, "learning_rate": 2.5381297406625447e-06, "loss": 0.4735, "step": 21755 }, { "epoch": 0.7751037640059141, "grad_norm": 1.3321690912284374, "learning_rate": 2.5373615727104417e-06, "loss": 0.3909, "step": 21756 }, { "epoch": 0.7751393911324083, "grad_norm": 1.303470502071322, "learning_rate": 2.5365935041285882e-06, "loss": 0.4488, "step": 21757 }, { "epoch": 0.7751750182589023, "grad_norm": 1.193648331214674, "learning_rate": 2.5358255349272055e-06, "loss": 0.3933, "step": 21758 }, { "epoch": 0.7752106453853964, "grad_norm": 1.471234438518668, "learning_rate": 2.5350576651165227e-06, "loss": 0.4353, "step": 21759 }, { "epoch": 0.7752462725118906, "grad_norm": 1.535902299680132, "learning_rate": 2.534289894706765e-06, "loss": 0.5098, "step": 21760 }, { "epoch": 0.7752818996383847, "grad_norm": 1.2438042022780906, "learning_rate": 2.533522223708158e-06, "loss": 0.4256, "step": 21761 }, { "epoch": 0.7753175267648788, "grad_norm": 1.1709253166268245, "learning_rate": 2.5327546521309186e-06, "loss": 0.3857, "step": 21762 }, { "epoch": 0.7753531538913729, "grad_norm": 1.154986293614266, "learning_rate": 2.5319871799852736e-06, "loss": 0.3737, "step": 21763 }, { "epoch": 0.775388781017867, "grad_norm": 1.1525620105184222, "learning_rate": 2.5312198072814364e-06, "loss": 0.3975, "step": 21764 }, { "epoch": 0.7754244081443611, "grad_norm": 1.3758407870712985, "learning_rate": 2.5304525340296272e-06, "loss": 0.4671, "step": 21765 }, { "epoch": 0.7754600352708553, "grad_norm": 1.3612066087893173, "learning_rate": 2.529685360240064e-06, "loss": 0.4185, "step": 21766 }, { "epoch": 0.7754956623973494, "grad_norm": 1.5364356935907673, "learning_rate": 2.528918285922961e-06, "loss": 0.491, "step": 21767 }, { "epoch": 0.7755312895238434, "grad_norm": 1.6191531064520135, "learning_rate": 2.528151311088537e-06, "loss": 0.4922, "step": 21768 }, { "epoch": 0.7755669166503376, "grad_norm": 1.4933001633445426, "learning_rate": 2.5273844357469968e-06, "loss": 0.5294, "step": 21769 }, { "epoch": 0.7756025437768317, "grad_norm": 1.5045868799155009, "learning_rate": 2.526617659908559e-06, "loss": 0.578, "step": 21770 }, { "epoch": 0.7756381709033258, "grad_norm": 1.163711950899907, "learning_rate": 2.525850983583428e-06, "loss": 0.4069, "step": 21771 }, { "epoch": 0.77567379802982, "grad_norm": 1.1907571488227289, "learning_rate": 2.525084406781818e-06, "loss": 0.4075, "step": 21772 }, { "epoch": 0.775709425156314, "grad_norm": 1.1512938088940645, "learning_rate": 2.5243179295139307e-06, "loss": 0.4124, "step": 21773 }, { "epoch": 0.7757450522828081, "grad_norm": 1.4574824018524963, "learning_rate": 2.5235515517899768e-06, "loss": 0.455, "step": 21774 }, { "epoch": 0.7757806794093023, "grad_norm": 1.6698105102576606, "learning_rate": 2.5227852736201573e-06, "loss": 0.5408, "step": 21775 }, { "epoch": 0.7758163065357964, "grad_norm": 1.497794764212727, "learning_rate": 2.522019095014683e-06, "loss": 0.5009, "step": 21776 }, { "epoch": 0.7758519336622904, "grad_norm": 1.3905792206373735, "learning_rate": 2.5212530159837477e-06, "loss": 0.5179, "step": 21777 }, { "epoch": 0.7758875607887846, "grad_norm": 1.2241017880811764, "learning_rate": 2.5204870365375587e-06, "loss": 0.3939, "step": 21778 }, { "epoch": 0.7759231879152787, "grad_norm": 1.5508569316358098, "learning_rate": 2.51972115668631e-06, "loss": 0.5576, "step": 21779 }, { "epoch": 0.7759588150417728, "grad_norm": 1.7194633582535417, "learning_rate": 2.518955376440202e-06, "loss": 0.4975, "step": 21780 }, { "epoch": 0.775994442168267, "grad_norm": 1.348837866337491, "learning_rate": 2.518189695809432e-06, "loss": 0.514, "step": 21781 }, { "epoch": 0.776030069294761, "grad_norm": 1.3681278602284512, "learning_rate": 2.5174241148041958e-06, "loss": 0.3934, "step": 21782 }, { "epoch": 0.7760656964212551, "grad_norm": 1.3852684937586162, "learning_rate": 2.516658633434691e-06, "loss": 0.476, "step": 21783 }, { "epoch": 0.7761013235477493, "grad_norm": 1.3136192961927184, "learning_rate": 2.5158932517111033e-06, "loss": 0.434, "step": 21784 }, { "epoch": 0.7761369506742434, "grad_norm": 1.112219469859719, "learning_rate": 2.5151279696436315e-06, "loss": 0.3417, "step": 21785 }, { "epoch": 0.7761725778007375, "grad_norm": 1.4628747216357445, "learning_rate": 2.514362787242459e-06, "loss": 0.5433, "step": 21786 }, { "epoch": 0.7762082049272316, "grad_norm": 2.084620608307679, "learning_rate": 2.5135977045177816e-06, "loss": 0.4799, "step": 21787 }, { "epoch": 0.7762438320537257, "grad_norm": 1.3165708511993233, "learning_rate": 2.512832721479781e-06, "loss": 0.4681, "step": 21788 }, { "epoch": 0.7762794591802198, "grad_norm": 1.2240596881471935, "learning_rate": 2.5120678381386454e-06, "loss": 0.4209, "step": 21789 }, { "epoch": 0.776315086306714, "grad_norm": 1.7659066781930974, "learning_rate": 2.5113030545045604e-06, "loss": 0.5668, "step": 21790 }, { "epoch": 0.776350713433208, "grad_norm": 1.4538702970552833, "learning_rate": 2.5105383705877083e-06, "loss": 0.5082, "step": 21791 }, { "epoch": 0.7763863405597021, "grad_norm": 1.28944252220226, "learning_rate": 2.5097737863982775e-06, "loss": 0.4097, "step": 21792 }, { "epoch": 0.7764219676861963, "grad_norm": 1.4906547254523312, "learning_rate": 2.50900930194644e-06, "loss": 0.4937, "step": 21793 }, { "epoch": 0.7764575948126904, "grad_norm": 1.5861885843496564, "learning_rate": 2.508244917242383e-06, "loss": 0.5036, "step": 21794 }, { "epoch": 0.7764932219391845, "grad_norm": 1.1095418211069779, "learning_rate": 2.5074806322962775e-06, "loss": 0.4159, "step": 21795 }, { "epoch": 0.7765288490656787, "grad_norm": 1.252832926818789, "learning_rate": 2.506716447118305e-06, "loss": 0.4476, "step": 21796 }, { "epoch": 0.7765644761921727, "grad_norm": 1.7247814399713117, "learning_rate": 2.505952361718641e-06, "loss": 0.5532, "step": 21797 }, { "epoch": 0.7766001033186668, "grad_norm": 1.2667145663081134, "learning_rate": 2.5051883761074613e-06, "loss": 0.4012, "step": 21798 }, { "epoch": 0.776635730445161, "grad_norm": 1.656051549652444, "learning_rate": 2.5044244902949346e-06, "loss": 0.5054, "step": 21799 }, { "epoch": 0.7766713575716551, "grad_norm": 1.4622926413112447, "learning_rate": 2.5036607042912375e-06, "loss": 0.4549, "step": 21800 }, { "epoch": 0.7767069846981491, "grad_norm": 1.0701277760350232, "learning_rate": 2.502897018106535e-06, "loss": 0.4018, "step": 21801 }, { "epoch": 0.7767426118246433, "grad_norm": 1.346716129365497, "learning_rate": 2.502133431750998e-06, "loss": 0.5286, "step": 21802 }, { "epoch": 0.7767782389511374, "grad_norm": 2.005733334437097, "learning_rate": 2.5013699452347963e-06, "loss": 0.5163, "step": 21803 }, { "epoch": 0.7768138660776315, "grad_norm": 1.252459145623996, "learning_rate": 2.500606558568095e-06, "loss": 0.4092, "step": 21804 }, { "epoch": 0.7768494932041257, "grad_norm": 1.1519730823228316, "learning_rate": 2.499843271761061e-06, "loss": 0.3964, "step": 21805 }, { "epoch": 0.7768851203306197, "grad_norm": 1.3765228533186749, "learning_rate": 2.499080084823854e-06, "loss": 0.5194, "step": 21806 }, { "epoch": 0.7769207474571138, "grad_norm": 1.135290164007098, "learning_rate": 2.498316997766641e-06, "loss": 0.3989, "step": 21807 }, { "epoch": 0.776956374583608, "grad_norm": 1.5469499148126298, "learning_rate": 2.497554010599579e-06, "loss": 0.4937, "step": 21808 }, { "epoch": 0.7769920017101021, "grad_norm": 1.5334086179642343, "learning_rate": 2.4967911233328302e-06, "loss": 0.4844, "step": 21809 }, { "epoch": 0.7770276288365962, "grad_norm": 1.832455918427317, "learning_rate": 2.49602833597655e-06, "loss": 0.5626, "step": 21810 }, { "epoch": 0.7770632559630903, "grad_norm": 1.5494557364422987, "learning_rate": 2.495265648540898e-06, "loss": 0.3957, "step": 21811 }, { "epoch": 0.7770988830895844, "grad_norm": 1.9158011417086125, "learning_rate": 2.494503061036029e-06, "loss": 0.4743, "step": 21812 }, { "epoch": 0.7771345102160785, "grad_norm": 1.3802422772576266, "learning_rate": 2.4937405734720964e-06, "loss": 0.5245, "step": 21813 }, { "epoch": 0.7771701373425727, "grad_norm": 1.9405455156603966, "learning_rate": 2.4929781858592595e-06, "loss": 0.4931, "step": 21814 }, { "epoch": 0.7772057644690668, "grad_norm": 1.62663594845496, "learning_rate": 2.4922158982076615e-06, "loss": 0.5441, "step": 21815 }, { "epoch": 0.7772413915955608, "grad_norm": 1.3011950918694861, "learning_rate": 2.4914537105274595e-06, "loss": 0.4092, "step": 21816 }, { "epoch": 0.777277018722055, "grad_norm": 1.1790334830254507, "learning_rate": 2.490691622828796e-06, "loss": 0.4096, "step": 21817 }, { "epoch": 0.7773126458485491, "grad_norm": 1.1681332343169202, "learning_rate": 2.489929635121823e-06, "loss": 0.4267, "step": 21818 }, { "epoch": 0.7773482729750432, "grad_norm": 1.4854502898607387, "learning_rate": 2.4891677474166854e-06, "loss": 0.4974, "step": 21819 }, { "epoch": 0.7773839001015374, "grad_norm": 1.2477124961763366, "learning_rate": 2.488405959723532e-06, "loss": 0.4116, "step": 21820 }, { "epoch": 0.7774195272280314, "grad_norm": 1.3493152445100955, "learning_rate": 2.4876442720525006e-06, "loss": 0.4504, "step": 21821 }, { "epoch": 0.7774551543545255, "grad_norm": 1.7984559013551018, "learning_rate": 2.4868826844137406e-06, "loss": 0.5273, "step": 21822 }, { "epoch": 0.7774907814810197, "grad_norm": 1.6311462958915184, "learning_rate": 2.4861211968173847e-06, "loss": 0.5518, "step": 21823 }, { "epoch": 0.7775264086075138, "grad_norm": 1.334379279344637, "learning_rate": 2.485359809273581e-06, "loss": 0.5161, "step": 21824 }, { "epoch": 0.7775620357340078, "grad_norm": 1.2034929753428003, "learning_rate": 2.4845985217924594e-06, "loss": 0.4106, "step": 21825 }, { "epoch": 0.777597662860502, "grad_norm": 1.6347931923631753, "learning_rate": 2.4838373343841603e-06, "loss": 0.5104, "step": 21826 }, { "epoch": 0.7776332899869961, "grad_norm": 1.4131562662713022, "learning_rate": 2.4830762470588255e-06, "loss": 0.4463, "step": 21827 }, { "epoch": 0.7776689171134902, "grad_norm": 1.2815390617201943, "learning_rate": 2.482315259826581e-06, "loss": 0.4221, "step": 21828 }, { "epoch": 0.7777045442399844, "grad_norm": 1.1883613632134056, "learning_rate": 2.481554372697568e-06, "loss": 0.4536, "step": 21829 }, { "epoch": 0.7777401713664784, "grad_norm": 1.187905789254903, "learning_rate": 2.480793585681909e-06, "loss": 0.4186, "step": 21830 }, { "epoch": 0.7777757984929725, "grad_norm": 1.076472737566573, "learning_rate": 2.4800328987897427e-06, "loss": 0.4073, "step": 21831 }, { "epoch": 0.7778114256194667, "grad_norm": 1.1009918336072624, "learning_rate": 2.479272312031191e-06, "loss": 0.4087, "step": 21832 }, { "epoch": 0.7778470527459608, "grad_norm": 1.2105323225203661, "learning_rate": 2.4785118254163867e-06, "loss": 0.4078, "step": 21833 }, { "epoch": 0.7778826798724549, "grad_norm": 1.1302939487639574, "learning_rate": 2.4777514389554524e-06, "loss": 0.3623, "step": 21834 }, { "epoch": 0.777918306998949, "grad_norm": 1.3400581932526767, "learning_rate": 2.47699115265852e-06, "loss": 0.4525, "step": 21835 }, { "epoch": 0.7779539341254431, "grad_norm": 1.6737295990235437, "learning_rate": 2.4762309665357056e-06, "loss": 0.4204, "step": 21836 }, { "epoch": 0.7779895612519372, "grad_norm": 1.7698004809541088, "learning_rate": 2.4754708805971338e-06, "loss": 0.5491, "step": 21837 }, { "epoch": 0.7780251883784314, "grad_norm": 1.2903287154208367, "learning_rate": 2.4747108948529297e-06, "loss": 0.4178, "step": 21838 }, { "epoch": 0.7780608155049255, "grad_norm": 1.6351741873377608, "learning_rate": 2.473951009313208e-06, "loss": 0.5491, "step": 21839 }, { "epoch": 0.7780964426314195, "grad_norm": 1.5001082589600303, "learning_rate": 2.473191223988087e-06, "loss": 0.4995, "step": 21840 }, { "epoch": 0.7781320697579137, "grad_norm": 1.4431141887040884, "learning_rate": 2.4724315388876872e-06, "loss": 0.4825, "step": 21841 }, { "epoch": 0.7781676968844078, "grad_norm": 1.1397478159505992, "learning_rate": 2.4716719540221267e-06, "loss": 0.4113, "step": 21842 }, { "epoch": 0.7782033240109019, "grad_norm": 1.5142329407982391, "learning_rate": 2.470912469401512e-06, "loss": 0.5623, "step": 21843 }, { "epoch": 0.778238951137396, "grad_norm": 1.282903950241295, "learning_rate": 2.4701530850359634e-06, "loss": 0.4813, "step": 21844 }, { "epoch": 0.7782745782638901, "grad_norm": 1.3230952984732465, "learning_rate": 2.469393800935587e-06, "loss": 0.4951, "step": 21845 }, { "epoch": 0.7783102053903842, "grad_norm": 1.4363998308516426, "learning_rate": 2.468634617110498e-06, "loss": 0.4891, "step": 21846 }, { "epoch": 0.7783458325168784, "grad_norm": 1.7743453461824221, "learning_rate": 2.4678755335708005e-06, "loss": 0.5301, "step": 21847 }, { "epoch": 0.7783814596433725, "grad_norm": 1.6180384604736606, "learning_rate": 2.4671165503266046e-06, "loss": 0.5701, "step": 21848 }, { "epoch": 0.7784170867698665, "grad_norm": 1.7991604266963321, "learning_rate": 2.4663576673880174e-06, "loss": 0.5605, "step": 21849 }, { "epoch": 0.7784527138963607, "grad_norm": 1.283145591745877, "learning_rate": 2.4655988847651434e-06, "loss": 0.4053, "step": 21850 }, { "epoch": 0.7784883410228548, "grad_norm": 1.9429262764851907, "learning_rate": 2.4648402024680895e-06, "loss": 0.4622, "step": 21851 }, { "epoch": 0.7785239681493489, "grad_norm": 1.2630706438049506, "learning_rate": 2.464081620506952e-06, "loss": 0.3924, "step": 21852 }, { "epoch": 0.7785595952758431, "grad_norm": 1.3259815698244564, "learning_rate": 2.4633231388918377e-06, "loss": 0.491, "step": 21853 }, { "epoch": 0.7785952224023371, "grad_norm": 1.5415959696297548, "learning_rate": 2.4625647576328415e-06, "loss": 0.5318, "step": 21854 }, { "epoch": 0.7786308495288312, "grad_norm": 1.4587685534530956, "learning_rate": 2.461806476740063e-06, "loss": 0.5418, "step": 21855 }, { "epoch": 0.7786664766553254, "grad_norm": 1.2086234652468761, "learning_rate": 2.4610482962236003e-06, "loss": 0.4046, "step": 21856 }, { "epoch": 0.7787021037818195, "grad_norm": 1.1795556073481888, "learning_rate": 2.4602902160935527e-06, "loss": 0.3993, "step": 21857 }, { "epoch": 0.7787377309083136, "grad_norm": 1.5955131047089184, "learning_rate": 2.459532236360007e-06, "loss": 0.5263, "step": 21858 }, { "epoch": 0.7787733580348077, "grad_norm": 1.6212234743512384, "learning_rate": 2.4587743570330615e-06, "loss": 0.5528, "step": 21859 }, { "epoch": 0.7788089851613018, "grad_norm": 1.4607885308092388, "learning_rate": 2.458016578122808e-06, "loss": 0.3854, "step": 21860 }, { "epoch": 0.7788446122877959, "grad_norm": 1.3286329124867085, "learning_rate": 2.4572588996393323e-06, "loss": 0.4681, "step": 21861 }, { "epoch": 0.7788802394142901, "grad_norm": 1.446228395615573, "learning_rate": 2.45650132159273e-06, "loss": 0.4752, "step": 21862 }, { "epoch": 0.7789158665407842, "grad_norm": 1.2886184800932203, "learning_rate": 2.4557438439930813e-06, "loss": 0.4135, "step": 21863 }, { "epoch": 0.7789514936672782, "grad_norm": 1.471974461651286, "learning_rate": 2.4549864668504774e-06, "loss": 0.5426, "step": 21864 }, { "epoch": 0.7789871207937724, "grad_norm": 2.031403827357667, "learning_rate": 2.4542291901750025e-06, "loss": 0.541, "step": 21865 }, { "epoch": 0.7790227479202665, "grad_norm": 1.2591769338425804, "learning_rate": 2.453472013976742e-06, "loss": 0.3937, "step": 21866 }, { "epoch": 0.7790583750467606, "grad_norm": 1.4927308590088124, "learning_rate": 2.4527149382657744e-06, "loss": 0.4183, "step": 21867 }, { "epoch": 0.7790940021732548, "grad_norm": 1.1783771694987213, "learning_rate": 2.451957963052185e-06, "loss": 0.3903, "step": 21868 }, { "epoch": 0.7791296292997488, "grad_norm": 1.5522593298756684, "learning_rate": 2.451201088346048e-06, "loss": 0.5609, "step": 21869 }, { "epoch": 0.7791652564262429, "grad_norm": 1.2730407263327563, "learning_rate": 2.450444314157445e-06, "loss": 0.3985, "step": 21870 }, { "epoch": 0.7792008835527371, "grad_norm": 1.2212511014968133, "learning_rate": 2.4496876404964533e-06, "loss": 0.4093, "step": 21871 }, { "epoch": 0.7792365106792312, "grad_norm": 1.2347349432309769, "learning_rate": 2.448931067373147e-06, "loss": 0.4367, "step": 21872 }, { "epoch": 0.7792721378057252, "grad_norm": 1.164051663043594, "learning_rate": 2.4481745947976053e-06, "loss": 0.4317, "step": 21873 }, { "epoch": 0.7793077649322194, "grad_norm": 1.4417949825171612, "learning_rate": 2.4474182227798936e-06, "loss": 0.5089, "step": 21874 }, { "epoch": 0.7793433920587135, "grad_norm": 1.163318606497753, "learning_rate": 2.4466619513300917e-06, "loss": 0.4238, "step": 21875 }, { "epoch": 0.7793790191852076, "grad_norm": 1.2606354343396684, "learning_rate": 2.445905780458261e-06, "loss": 0.4173, "step": 21876 }, { "epoch": 0.7794146463117018, "grad_norm": 1.7719823107195682, "learning_rate": 2.4451497101744746e-06, "loss": 0.5584, "step": 21877 }, { "epoch": 0.7794502734381958, "grad_norm": 1.9205831505618822, "learning_rate": 2.4443937404888007e-06, "loss": 0.6019, "step": 21878 }, { "epoch": 0.7794859005646899, "grad_norm": 1.1984624219622977, "learning_rate": 2.4436378714113087e-06, "loss": 0.3983, "step": 21879 }, { "epoch": 0.7795215276911841, "grad_norm": 1.1538622727753882, "learning_rate": 2.4428821029520566e-06, "loss": 0.3963, "step": 21880 }, { "epoch": 0.7795571548176782, "grad_norm": 1.4236668456267332, "learning_rate": 2.442126435121115e-06, "loss": 0.5108, "step": 21881 }, { "epoch": 0.7795927819441723, "grad_norm": 1.7802436721465775, "learning_rate": 2.4413708679285386e-06, "loss": 0.5173, "step": 21882 }, { "epoch": 0.7796284090706664, "grad_norm": 1.8334869500979936, "learning_rate": 2.440615401384393e-06, "loss": 0.5681, "step": 21883 }, { "epoch": 0.7796640361971605, "grad_norm": 1.2849773958418735, "learning_rate": 2.439860035498739e-06, "loss": 0.4104, "step": 21884 }, { "epoch": 0.7796996633236546, "grad_norm": 1.4064209246778854, "learning_rate": 2.4391047702816306e-06, "loss": 0.4643, "step": 21885 }, { "epoch": 0.7797352904501488, "grad_norm": 1.2154274177276554, "learning_rate": 2.4383496057431264e-06, "loss": 0.4196, "step": 21886 }, { "epoch": 0.7797709175766429, "grad_norm": 1.442589226943207, "learning_rate": 2.4375945418932832e-06, "loss": 0.4435, "step": 21887 }, { "epoch": 0.7798065447031369, "grad_norm": 1.1356317248006564, "learning_rate": 2.4368395787421575e-06, "loss": 0.4004, "step": 21888 }, { "epoch": 0.7798421718296311, "grad_norm": 1.1315909997959321, "learning_rate": 2.4360847162997958e-06, "loss": 0.3612, "step": 21889 }, { "epoch": 0.7798777989561252, "grad_norm": 1.448493187887049, "learning_rate": 2.4353299545762567e-06, "loss": 0.4453, "step": 21890 }, { "epoch": 0.7799134260826193, "grad_norm": 1.1907299750313645, "learning_rate": 2.4345752935815826e-06, "loss": 0.4251, "step": 21891 }, { "epoch": 0.7799490532091135, "grad_norm": 1.1591463732280232, "learning_rate": 2.433820733325827e-06, "loss": 0.3898, "step": 21892 }, { "epoch": 0.7799846803356075, "grad_norm": 1.6983612173971345, "learning_rate": 2.433066273819037e-06, "loss": 0.4573, "step": 21893 }, { "epoch": 0.7800203074621016, "grad_norm": 1.7628374813596441, "learning_rate": 2.432311915071257e-06, "loss": 0.491, "step": 21894 }, { "epoch": 0.7800559345885958, "grad_norm": 1.4262845305466307, "learning_rate": 2.431557657092538e-06, "loss": 0.5394, "step": 21895 }, { "epoch": 0.7800915617150899, "grad_norm": 1.1277712562144266, "learning_rate": 2.430803499892915e-06, "loss": 0.3968, "step": 21896 }, { "epoch": 0.7801271888415839, "grad_norm": 1.2895547432101762, "learning_rate": 2.4300494434824373e-06, "loss": 0.4026, "step": 21897 }, { "epoch": 0.7801628159680781, "grad_norm": 1.4786217635890784, "learning_rate": 2.4292954878711395e-06, "loss": 0.4633, "step": 21898 }, { "epoch": 0.7801984430945722, "grad_norm": 1.1297749727407598, "learning_rate": 2.428541633069066e-06, "loss": 0.4144, "step": 21899 }, { "epoch": 0.7802340702210663, "grad_norm": 1.8338508332709977, "learning_rate": 2.427787879086251e-06, "loss": 0.5264, "step": 21900 }, { "epoch": 0.7802696973475605, "grad_norm": 1.1099003980697044, "learning_rate": 2.427034225932733e-06, "loss": 0.4222, "step": 21901 }, { "epoch": 0.7803053244740545, "grad_norm": 1.50255634450279, "learning_rate": 2.4262806736185463e-06, "loss": 0.5152, "step": 21902 }, { "epoch": 0.7803409516005486, "grad_norm": 1.892216210524647, "learning_rate": 2.4255272221537295e-06, "loss": 0.4939, "step": 21903 }, { "epoch": 0.7803765787270428, "grad_norm": 1.2027000975487494, "learning_rate": 2.4247738715483092e-06, "loss": 0.4272, "step": 21904 }, { "epoch": 0.7804122058535369, "grad_norm": 1.6858543708862974, "learning_rate": 2.424020621812323e-06, "loss": 0.5716, "step": 21905 }, { "epoch": 0.780447832980031, "grad_norm": 3.782771740250925, "learning_rate": 2.4232674729557936e-06, "loss": 0.4738, "step": 21906 }, { "epoch": 0.7804834601065251, "grad_norm": 1.274259383997368, "learning_rate": 2.4225144249887546e-06, "loss": 0.4684, "step": 21907 }, { "epoch": 0.7805190872330192, "grad_norm": 1.0862825745150742, "learning_rate": 2.421761477921232e-06, "loss": 0.4022, "step": 21908 }, { "epoch": 0.7805547143595133, "grad_norm": 1.2977696013653877, "learning_rate": 2.421008631763252e-06, "loss": 0.401, "step": 21909 }, { "epoch": 0.7805903414860075, "grad_norm": 1.6630745848553046, "learning_rate": 2.4202558865248426e-06, "loss": 0.4294, "step": 21910 }, { "epoch": 0.7806259686125016, "grad_norm": 1.0839989213207126, "learning_rate": 2.4195032422160213e-06, "loss": 0.3799, "step": 21911 }, { "epoch": 0.7806615957389956, "grad_norm": 1.4350036644941884, "learning_rate": 2.418750698846816e-06, "loss": 0.5256, "step": 21912 }, { "epoch": 0.7806972228654898, "grad_norm": 1.6282018010655344, "learning_rate": 2.4179982564272396e-06, "loss": 0.5661, "step": 21913 }, { "epoch": 0.7807328499919839, "grad_norm": 1.8025785001806005, "learning_rate": 2.41724591496732e-06, "loss": 0.5708, "step": 21914 }, { "epoch": 0.780768477118478, "grad_norm": 1.750946734825256, "learning_rate": 2.4164936744770653e-06, "loss": 0.4962, "step": 21915 }, { "epoch": 0.7808041042449722, "grad_norm": 1.118618111359636, "learning_rate": 2.4157415349665037e-06, "loss": 0.3939, "step": 21916 }, { "epoch": 0.7808397313714662, "grad_norm": 1.1909304849579532, "learning_rate": 2.414989496445641e-06, "loss": 0.4247, "step": 21917 }, { "epoch": 0.7808753584979603, "grad_norm": 1.4030951872109423, "learning_rate": 2.414237558924496e-06, "loss": 0.3985, "step": 21918 }, { "epoch": 0.7809109856244545, "grad_norm": 1.147689355027032, "learning_rate": 2.4134857224130814e-06, "loss": 0.4367, "step": 21919 }, { "epoch": 0.7809466127509486, "grad_norm": 1.508086839665904, "learning_rate": 2.412733986921405e-06, "loss": 0.5305, "step": 21920 }, { "epoch": 0.7809822398774426, "grad_norm": 1.4455058488199783, "learning_rate": 2.4119823524594822e-06, "loss": 0.4401, "step": 21921 }, { "epoch": 0.7810178670039368, "grad_norm": 1.5522107706096053, "learning_rate": 2.4112308190373147e-06, "loss": 0.4782, "step": 21922 }, { "epoch": 0.7810534941304309, "grad_norm": 1.572563588459737, "learning_rate": 2.4104793866649124e-06, "loss": 0.3606, "step": 21923 }, { "epoch": 0.781089121256925, "grad_norm": 1.1954525709344046, "learning_rate": 2.409728055352283e-06, "loss": 0.3934, "step": 21924 }, { "epoch": 0.7811247483834192, "grad_norm": 1.7062354864304816, "learning_rate": 2.4089768251094314e-06, "loss": 0.5812, "step": 21925 }, { "epoch": 0.7811603755099132, "grad_norm": 1.8751746448244158, "learning_rate": 2.408225695946358e-06, "loss": 0.4085, "step": 21926 }, { "epoch": 0.7811960026364073, "grad_norm": 1.5331267593775348, "learning_rate": 2.4074746678730677e-06, "loss": 0.4997, "step": 21927 }, { "epoch": 0.7812316297629015, "grad_norm": 1.2581436753386142, "learning_rate": 2.4067237408995557e-06, "loss": 0.4238, "step": 21928 }, { "epoch": 0.7812672568893956, "grad_norm": 1.6931022203275403, "learning_rate": 2.4059729150358244e-06, "loss": 0.5075, "step": 21929 }, { "epoch": 0.7813028840158897, "grad_norm": 1.565582740040876, "learning_rate": 2.405222190291873e-06, "loss": 0.5751, "step": 21930 }, { "epoch": 0.7813385111423838, "grad_norm": 1.1025735778867838, "learning_rate": 2.4044715666776954e-06, "loss": 0.3702, "step": 21931 }, { "epoch": 0.7813741382688779, "grad_norm": 1.5786713373216317, "learning_rate": 2.403721044203291e-06, "loss": 0.3501, "step": 21932 }, { "epoch": 0.781409765395372, "grad_norm": 1.1079851266172203, "learning_rate": 2.4029706228786477e-06, "loss": 0.3588, "step": 21933 }, { "epoch": 0.7814453925218662, "grad_norm": 1.890365237110708, "learning_rate": 2.4022203027137635e-06, "loss": 0.5301, "step": 21934 }, { "epoch": 0.7814810196483603, "grad_norm": 1.4654374496997198, "learning_rate": 2.4014700837186234e-06, "loss": 0.5025, "step": 21935 }, { "epoch": 0.7815166467748543, "grad_norm": 1.3777152941766662, "learning_rate": 2.400719965903223e-06, "loss": 0.4625, "step": 21936 }, { "epoch": 0.7815522739013485, "grad_norm": 1.1350035473496336, "learning_rate": 2.399969949277545e-06, "loss": 0.4025, "step": 21937 }, { "epoch": 0.7815879010278426, "grad_norm": 1.4103355431933364, "learning_rate": 2.3992200338515794e-06, "loss": 0.4784, "step": 21938 }, { "epoch": 0.7816235281543367, "grad_norm": 1.203456430294895, "learning_rate": 2.3984702196353116e-06, "loss": 0.3929, "step": 21939 }, { "epoch": 0.7816591552808309, "grad_norm": 2.378664378386477, "learning_rate": 2.3977205066387255e-06, "loss": 0.4972, "step": 21940 }, { "epoch": 0.7816947824073249, "grad_norm": 1.1153912067513025, "learning_rate": 2.3969708948718084e-06, "loss": 0.3892, "step": 21941 }, { "epoch": 0.781730409533819, "grad_norm": 1.5857799494253577, "learning_rate": 2.396221384344535e-06, "loss": 0.5396, "step": 21942 }, { "epoch": 0.7817660366603132, "grad_norm": 1.5063782044654441, "learning_rate": 2.395471975066891e-06, "loss": 0.5476, "step": 21943 }, { "epoch": 0.7818016637868073, "grad_norm": 1.7029043219979196, "learning_rate": 2.39472266704885e-06, "loss": 0.5126, "step": 21944 }, { "epoch": 0.7818372909133013, "grad_norm": 1.23343432294182, "learning_rate": 2.3939734603003928e-06, "loss": 0.3922, "step": 21945 }, { "epoch": 0.7818729180397955, "grad_norm": 1.2329814458911474, "learning_rate": 2.393224354831495e-06, "loss": 0.4032, "step": 21946 }, { "epoch": 0.7819085451662896, "grad_norm": 1.1961255598379112, "learning_rate": 2.3924753506521346e-06, "loss": 0.4616, "step": 21947 }, { "epoch": 0.7819441722927837, "grad_norm": 1.180478206807865, "learning_rate": 2.391726447772279e-06, "loss": 0.4194, "step": 21948 }, { "epoch": 0.7819797994192779, "grad_norm": 1.4103261673674035, "learning_rate": 2.3909776462019073e-06, "loss": 0.5323, "step": 21949 }, { "epoch": 0.7820154265457719, "grad_norm": 1.1041445223852266, "learning_rate": 2.3902289459509832e-06, "loss": 0.372, "step": 21950 }, { "epoch": 0.782051053672266, "grad_norm": 1.2338469553777416, "learning_rate": 2.3894803470294824e-06, "loss": 0.4381, "step": 21951 }, { "epoch": 0.7820866807987602, "grad_norm": 1.1780152326569282, "learning_rate": 2.3887318494473676e-06, "loss": 0.3627, "step": 21952 }, { "epoch": 0.7821223079252543, "grad_norm": 1.8237507255743324, "learning_rate": 2.387983453214605e-06, "loss": 0.5227, "step": 21953 }, { "epoch": 0.7821579350517484, "grad_norm": 1.4253590724571013, "learning_rate": 2.38723515834117e-06, "loss": 0.5325, "step": 21954 }, { "epoch": 0.7821935621782425, "grad_norm": 1.2984023778245533, "learning_rate": 2.386486964837016e-06, "loss": 0.3919, "step": 21955 }, { "epoch": 0.7822291893047366, "grad_norm": 1.2671627119588715, "learning_rate": 2.3857388727121123e-06, "loss": 0.4059, "step": 21956 }, { "epoch": 0.7822648164312307, "grad_norm": 1.6050592948963467, "learning_rate": 2.3849908819764154e-06, "loss": 0.4666, "step": 21957 }, { "epoch": 0.7823004435577249, "grad_norm": 1.265905670322566, "learning_rate": 2.3842429926398903e-06, "loss": 0.451, "step": 21958 }, { "epoch": 0.782336070684219, "grad_norm": 1.175792987516147, "learning_rate": 2.3834952047124895e-06, "loss": 0.4033, "step": 21959 }, { "epoch": 0.782371697810713, "grad_norm": 1.4785699242450052, "learning_rate": 2.3827475182041738e-06, "loss": 0.5523, "step": 21960 }, { "epoch": 0.7824073249372072, "grad_norm": 1.3027845683340826, "learning_rate": 2.3819999331248987e-06, "loss": 0.5606, "step": 21961 }, { "epoch": 0.7824429520637013, "grad_norm": 1.1224946635149333, "learning_rate": 2.3812524494846224e-06, "loss": 0.3972, "step": 21962 }, { "epoch": 0.7824785791901954, "grad_norm": 1.530962476376688, "learning_rate": 2.380505067293293e-06, "loss": 0.4593, "step": 21963 }, { "epoch": 0.7825142063166896, "grad_norm": 1.2047240476916008, "learning_rate": 2.3797577865608633e-06, "loss": 0.4119, "step": 21964 }, { "epoch": 0.7825498334431836, "grad_norm": 1.1682010786320391, "learning_rate": 2.3790106072972874e-06, "loss": 0.4115, "step": 21965 }, { "epoch": 0.7825854605696777, "grad_norm": 1.1400608147485007, "learning_rate": 2.37826352951251e-06, "loss": 0.3818, "step": 21966 }, { "epoch": 0.7826210876961719, "grad_norm": 1.0297948425095431, "learning_rate": 2.37751655321648e-06, "loss": 0.4305, "step": 21967 }, { "epoch": 0.782656714822666, "grad_norm": 1.8675806909700208, "learning_rate": 2.3767696784191464e-06, "loss": 0.6005, "step": 21968 }, { "epoch": 0.78269234194916, "grad_norm": 1.1332795529995563, "learning_rate": 2.376022905130454e-06, "loss": 0.3513, "step": 21969 }, { "epoch": 0.7827279690756542, "grad_norm": 1.668115687424401, "learning_rate": 2.375276233360344e-06, "loss": 0.5265, "step": 21970 }, { "epoch": 0.7827635962021483, "grad_norm": 1.4259127712099, "learning_rate": 2.374529663118762e-06, "loss": 0.513, "step": 21971 }, { "epoch": 0.7827992233286424, "grad_norm": 1.1350641430593498, "learning_rate": 2.3737831944156453e-06, "loss": 0.4232, "step": 21972 }, { "epoch": 0.7828348504551366, "grad_norm": 1.522402082610251, "learning_rate": 2.3730368272609394e-06, "loss": 0.4899, "step": 21973 }, { "epoch": 0.7828704775816306, "grad_norm": 1.6889017007138323, "learning_rate": 2.3722905616645754e-06, "loss": 0.5296, "step": 21974 }, { "epoch": 0.7829061047081247, "grad_norm": 1.2042003593563833, "learning_rate": 2.3715443976364937e-06, "loss": 0.4875, "step": 21975 }, { "epoch": 0.7829417318346189, "grad_norm": 1.2773885176642545, "learning_rate": 2.3707983351866304e-06, "loss": 0.393, "step": 21976 }, { "epoch": 0.782977358961113, "grad_norm": 1.2177496658085007, "learning_rate": 2.370052374324919e-06, "loss": 0.425, "step": 21977 }, { "epoch": 0.783012986087607, "grad_norm": 1.2451844745404423, "learning_rate": 2.369306515061298e-06, "loss": 0.418, "step": 21978 }, { "epoch": 0.7830486132141012, "grad_norm": 1.4318773265822304, "learning_rate": 2.36856075740569e-06, "loss": 0.4901, "step": 21979 }, { "epoch": 0.7830842403405953, "grad_norm": 1.6210609996441774, "learning_rate": 2.367815101368033e-06, "loss": 0.547, "step": 21980 }, { "epoch": 0.7831198674670894, "grad_norm": 1.4615280046907926, "learning_rate": 2.3670695469582492e-06, "loss": 0.5658, "step": 21981 }, { "epoch": 0.7831554945935836, "grad_norm": 1.549606857105924, "learning_rate": 2.36632409418627e-06, "loss": 0.5056, "step": 21982 }, { "epoch": 0.7831911217200777, "grad_norm": 1.6701196818862807, "learning_rate": 2.3655787430620214e-06, "loss": 0.5558, "step": 21983 }, { "epoch": 0.7832267488465717, "grad_norm": 1.3829244854801732, "learning_rate": 2.36483349359543e-06, "loss": 0.4513, "step": 21984 }, { "epoch": 0.7832623759730659, "grad_norm": 1.186009018883587, "learning_rate": 2.3640883457964146e-06, "loss": 0.353, "step": 21985 }, { "epoch": 0.78329800309956, "grad_norm": 1.298951526651692, "learning_rate": 2.3633432996749005e-06, "loss": 0.4136, "step": 21986 }, { "epoch": 0.7833336302260541, "grad_norm": 1.4025022519658492, "learning_rate": 2.3625983552408103e-06, "loss": 0.4937, "step": 21987 }, { "epoch": 0.7833692573525483, "grad_norm": 1.1131899063486241, "learning_rate": 2.361853512504059e-06, "loss": 0.4365, "step": 21988 }, { "epoch": 0.7834048844790423, "grad_norm": 1.1593299999285207, "learning_rate": 2.3611087714745696e-06, "loss": 0.4219, "step": 21989 }, { "epoch": 0.7834405116055364, "grad_norm": 1.3100725039109873, "learning_rate": 2.3603641321622506e-06, "loss": 0.4779, "step": 21990 }, { "epoch": 0.7834761387320306, "grad_norm": 1.4538504942625428, "learning_rate": 2.3596195945770284e-06, "loss": 0.5001, "step": 21991 }, { "epoch": 0.7835117658585247, "grad_norm": 1.8359649199381718, "learning_rate": 2.358875158728808e-06, "loss": 0.5372, "step": 21992 }, { "epoch": 0.7835473929850187, "grad_norm": 1.2349759207542172, "learning_rate": 2.3581308246275103e-06, "loss": 0.3687, "step": 21993 }, { "epoch": 0.7835830201115129, "grad_norm": 1.2592113359715955, "learning_rate": 2.3573865922830387e-06, "loss": 0.4223, "step": 21994 }, { "epoch": 0.783618647238007, "grad_norm": 1.139787701747006, "learning_rate": 2.3566424617053084e-06, "loss": 0.364, "step": 21995 }, { "epoch": 0.7836542743645011, "grad_norm": 1.5221678299358634, "learning_rate": 2.3558984329042233e-06, "loss": 0.4939, "step": 21996 }, { "epoch": 0.7836899014909953, "grad_norm": 1.2547292505457186, "learning_rate": 2.355154505889694e-06, "loss": 0.4292, "step": 21997 }, { "epoch": 0.7837255286174893, "grad_norm": 1.2673498898845825, "learning_rate": 2.354410680671626e-06, "loss": 0.4299, "step": 21998 }, { "epoch": 0.7837611557439834, "grad_norm": 1.3397402791664956, "learning_rate": 2.3536669572599234e-06, "loss": 0.5025, "step": 21999 }, { "epoch": 0.7837967828704776, "grad_norm": 1.203824631846567, "learning_rate": 2.352923335664493e-06, "loss": 0.424, "step": 22000 }, { "epoch": 0.7838324099969717, "grad_norm": 1.602386498442976, "learning_rate": 2.3521798158952302e-06, "loss": 0.4619, "step": 22001 }, { "epoch": 0.7838680371234658, "grad_norm": 1.3395195454338502, "learning_rate": 2.3514363979620412e-06, "loss": 0.4084, "step": 22002 }, { "epoch": 0.7839036642499599, "grad_norm": 1.1852336279271092, "learning_rate": 2.3506930818748207e-06, "loss": 0.3895, "step": 22003 }, { "epoch": 0.783939291376454, "grad_norm": 1.1807168613629675, "learning_rate": 2.3499498676434675e-06, "loss": 0.3719, "step": 22004 }, { "epoch": 0.7839749185029481, "grad_norm": 1.3610861215660868, "learning_rate": 2.3492067552778784e-06, "loss": 0.558, "step": 22005 }, { "epoch": 0.7840105456294423, "grad_norm": 1.4892692444014266, "learning_rate": 2.348463744787953e-06, "loss": 0.3714, "step": 22006 }, { "epoch": 0.7840461727559364, "grad_norm": 1.2137999203097538, "learning_rate": 2.347720836183578e-06, "loss": 0.3983, "step": 22007 }, { "epoch": 0.7840817998824304, "grad_norm": 1.4723310246917813, "learning_rate": 2.34697802947465e-06, "loss": 0.5629, "step": 22008 }, { "epoch": 0.7841174270089246, "grad_norm": 1.2171417025880746, "learning_rate": 2.346235324671058e-06, "loss": 0.4177, "step": 22009 }, { "epoch": 0.7841530541354187, "grad_norm": 1.9192527034036893, "learning_rate": 2.3454927217826906e-06, "loss": 0.5092, "step": 22010 }, { "epoch": 0.7841886812619128, "grad_norm": 1.6051221415998522, "learning_rate": 2.3447502208194415e-06, "loss": 0.577, "step": 22011 }, { "epoch": 0.784224308388407, "grad_norm": 1.2719231344827864, "learning_rate": 2.344007821791191e-06, "loss": 0.3876, "step": 22012 }, { "epoch": 0.784259935514901, "grad_norm": 3.0278274171342563, "learning_rate": 2.3432655247078273e-06, "loss": 0.4585, "step": 22013 }, { "epoch": 0.7842955626413951, "grad_norm": 1.7089852045112626, "learning_rate": 2.342523329579235e-06, "loss": 0.6332, "step": 22014 }, { "epoch": 0.7843311897678893, "grad_norm": 1.5673201853581495, "learning_rate": 2.3417812364153e-06, "loss": 0.5595, "step": 22015 }, { "epoch": 0.7843668168943834, "grad_norm": 1.246997487023527, "learning_rate": 2.341039245225898e-06, "loss": 0.3904, "step": 22016 }, { "epoch": 0.7844024440208774, "grad_norm": 1.0940936061369253, "learning_rate": 2.340297356020913e-06, "loss": 0.3784, "step": 22017 }, { "epoch": 0.7844380711473716, "grad_norm": 1.485784238935625, "learning_rate": 2.339555568810221e-06, "loss": 0.5378, "step": 22018 }, { "epoch": 0.7844736982738657, "grad_norm": 1.5210261629178978, "learning_rate": 2.338813883603701e-06, "loss": 0.4743, "step": 22019 }, { "epoch": 0.7845093254003598, "grad_norm": 1.4602030215180046, "learning_rate": 2.338072300411228e-06, "loss": 0.5434, "step": 22020 }, { "epoch": 0.784544952526854, "grad_norm": 1.4734578078475546, "learning_rate": 2.3373308192426793e-06, "loss": 0.4504, "step": 22021 }, { "epoch": 0.784580579653348, "grad_norm": 1.908620033129707, "learning_rate": 2.336589440107928e-06, "loss": 0.4998, "step": 22022 }, { "epoch": 0.7846162067798421, "grad_norm": 1.3308351924937376, "learning_rate": 2.3358481630168416e-06, "loss": 0.4205, "step": 22023 }, { "epoch": 0.7846518339063363, "grad_norm": 1.3858554182133, "learning_rate": 2.3351069879792975e-06, "loss": 0.3932, "step": 22024 }, { "epoch": 0.7846874610328304, "grad_norm": 1.5930934293534502, "learning_rate": 2.334365915005159e-06, "loss": 0.5125, "step": 22025 }, { "epoch": 0.7847230881593245, "grad_norm": 1.2703470312284824, "learning_rate": 2.333624944104298e-06, "loss": 0.3774, "step": 22026 }, { "epoch": 0.7847587152858186, "grad_norm": 1.6693273289651116, "learning_rate": 2.3328840752865776e-06, "loss": 0.4902, "step": 22027 }, { "epoch": 0.7847943424123127, "grad_norm": 1.5611923413822693, "learning_rate": 2.3321433085618637e-06, "loss": 0.5098, "step": 22028 }, { "epoch": 0.7848299695388068, "grad_norm": 1.452412009090838, "learning_rate": 2.3314026439400217e-06, "loss": 0.4876, "step": 22029 }, { "epoch": 0.784865596665301, "grad_norm": 1.2062684502702816, "learning_rate": 2.3306620814309165e-06, "loss": 0.42, "step": 22030 }, { "epoch": 0.784901223791795, "grad_norm": 1.496541608134421, "learning_rate": 2.329921621044403e-06, "loss": 0.5891, "step": 22031 }, { "epoch": 0.7849368509182891, "grad_norm": 1.5643835954269727, "learning_rate": 2.3291812627903475e-06, "loss": 0.5197, "step": 22032 }, { "epoch": 0.7849724780447833, "grad_norm": 1.1930681950132191, "learning_rate": 2.3284410066786025e-06, "loss": 0.3815, "step": 22033 }, { "epoch": 0.7850081051712774, "grad_norm": 2.896222296118363, "learning_rate": 2.327700852719026e-06, "loss": 0.47, "step": 22034 }, { "epoch": 0.7850437322977715, "grad_norm": 1.3592993244829967, "learning_rate": 2.326960800921476e-06, "loss": 0.4343, "step": 22035 }, { "epoch": 0.7850793594242657, "grad_norm": 1.1767540234171046, "learning_rate": 2.3262208512958074e-06, "loss": 0.366, "step": 22036 }, { "epoch": 0.7851149865507597, "grad_norm": 1.4113001744364875, "learning_rate": 2.3254810038518727e-06, "loss": 0.4617, "step": 22037 }, { "epoch": 0.7851506136772538, "grad_norm": 1.451051312426855, "learning_rate": 2.324741258599521e-06, "loss": 0.4731, "step": 22038 }, { "epoch": 0.785186240803748, "grad_norm": 1.2156460565940892, "learning_rate": 2.324001615548607e-06, "loss": 0.4255, "step": 22039 }, { "epoch": 0.7852218679302421, "grad_norm": 1.1810557182671828, "learning_rate": 2.323262074708973e-06, "loss": 0.4279, "step": 22040 }, { "epoch": 0.7852574950567363, "grad_norm": 1.3428130421599869, "learning_rate": 2.322522636090474e-06, "loss": 0.423, "step": 22041 }, { "epoch": 0.7852931221832303, "grad_norm": 1.2793469765920122, "learning_rate": 2.3217832997029467e-06, "loss": 0.36, "step": 22042 }, { "epoch": 0.7853287493097244, "grad_norm": 1.2057696669429685, "learning_rate": 2.321044065556246e-06, "loss": 0.3977, "step": 22043 }, { "epoch": 0.7853643764362186, "grad_norm": 1.5502097324913315, "learning_rate": 2.320304933660209e-06, "loss": 0.4917, "step": 22044 }, { "epoch": 0.7854000035627127, "grad_norm": 1.546537872950001, "learning_rate": 2.319565904024679e-06, "loss": 0.5458, "step": 22045 }, { "epoch": 0.7854356306892067, "grad_norm": 1.4426664116739352, "learning_rate": 2.3188269766595004e-06, "loss": 0.5696, "step": 22046 }, { "epoch": 0.7854712578157009, "grad_norm": 1.1951690006553375, "learning_rate": 2.3180881515745067e-06, "loss": 0.3906, "step": 22047 }, { "epoch": 0.785506884942195, "grad_norm": 1.2042490549402427, "learning_rate": 2.3173494287795405e-06, "loss": 0.3508, "step": 22048 }, { "epoch": 0.7855425120686891, "grad_norm": 1.740773281145591, "learning_rate": 2.316610808284435e-06, "loss": 0.5626, "step": 22049 }, { "epoch": 0.7855781391951833, "grad_norm": 1.6702811619157634, "learning_rate": 2.315872290099026e-06, "loss": 0.4908, "step": 22050 }, { "epoch": 0.7856137663216773, "grad_norm": 1.3573859080703632, "learning_rate": 2.3151338742331496e-06, "loss": 0.5255, "step": 22051 }, { "epoch": 0.7856493934481714, "grad_norm": 1.5247909606165284, "learning_rate": 2.3143955606966382e-06, "loss": 0.5068, "step": 22052 }, { "epoch": 0.7856850205746656, "grad_norm": 1.1861515400032188, "learning_rate": 2.313657349499321e-06, "loss": 0.4244, "step": 22053 }, { "epoch": 0.7857206477011597, "grad_norm": 1.501284214637964, "learning_rate": 2.31291924065103e-06, "loss": 0.4773, "step": 22054 }, { "epoch": 0.7857562748276538, "grad_norm": 1.241743761635516, "learning_rate": 2.31218123416159e-06, "loss": 0.4247, "step": 22055 }, { "epoch": 0.7857919019541479, "grad_norm": 1.571892116499664, "learning_rate": 2.3114433300408302e-06, "loss": 0.4919, "step": 22056 }, { "epoch": 0.785827529080642, "grad_norm": 1.1815890107504514, "learning_rate": 2.3107055282985756e-06, "loss": 0.4057, "step": 22057 }, { "epoch": 0.7858631562071361, "grad_norm": 1.2738971283162943, "learning_rate": 2.3099678289446527e-06, "loss": 0.4283, "step": 22058 }, { "epoch": 0.7858987833336303, "grad_norm": 1.2146228278243592, "learning_rate": 2.309230231988885e-06, "loss": 0.4567, "step": 22059 }, { "epoch": 0.7859344104601244, "grad_norm": 1.4064182877294293, "learning_rate": 2.3084927374410905e-06, "loss": 0.5065, "step": 22060 }, { "epoch": 0.7859700375866184, "grad_norm": 1.615753682581119, "learning_rate": 2.307755345311092e-06, "loss": 0.5354, "step": 22061 }, { "epoch": 0.7860056647131126, "grad_norm": 1.386641289549652, "learning_rate": 2.3070180556087076e-06, "loss": 0.5132, "step": 22062 }, { "epoch": 0.7860412918396067, "grad_norm": 1.290382055401278, "learning_rate": 2.3062808683437555e-06, "loss": 0.3861, "step": 22063 }, { "epoch": 0.7860769189661008, "grad_norm": 2.2054766642024393, "learning_rate": 2.3055437835260497e-06, "loss": 0.5321, "step": 22064 }, { "epoch": 0.786112546092595, "grad_norm": 2.1738265558182785, "learning_rate": 2.3048068011654056e-06, "loss": 0.5394, "step": 22065 }, { "epoch": 0.786148173219089, "grad_norm": 1.170650060126965, "learning_rate": 2.304069921271639e-06, "loss": 0.4164, "step": 22066 }, { "epoch": 0.7861838003455831, "grad_norm": 1.3842044789324306, "learning_rate": 2.3033331438545593e-06, "loss": 0.4435, "step": 22067 }, { "epoch": 0.7862194274720773, "grad_norm": 1.5331152116348197, "learning_rate": 2.302596468923981e-06, "loss": 0.5266, "step": 22068 }, { "epoch": 0.7862550545985714, "grad_norm": 1.3865723053164258, "learning_rate": 2.301859896489709e-06, "loss": 0.4495, "step": 22069 }, { "epoch": 0.7862906817250654, "grad_norm": 1.3251174649560455, "learning_rate": 2.3011234265615557e-06, "loss": 0.394, "step": 22070 }, { "epoch": 0.7863263088515596, "grad_norm": 2.122480283171364, "learning_rate": 2.3003870591493215e-06, "loss": 0.4708, "step": 22071 }, { "epoch": 0.7863619359780537, "grad_norm": 1.4613621502242986, "learning_rate": 2.2996507942628167e-06, "loss": 0.4614, "step": 22072 }, { "epoch": 0.7863975631045478, "grad_norm": 2.0242581977863927, "learning_rate": 2.2989146319118428e-06, "loss": 0.3968, "step": 22073 }, { "epoch": 0.786433190231042, "grad_norm": 1.1873949645064303, "learning_rate": 2.2981785721062066e-06, "loss": 0.4353, "step": 22074 }, { "epoch": 0.786468817357536, "grad_norm": 1.296562503011234, "learning_rate": 2.2974426148557028e-06, "loss": 0.4046, "step": 22075 }, { "epoch": 0.7865044444840301, "grad_norm": 1.2471077930903756, "learning_rate": 2.2967067601701375e-06, "loss": 0.453, "step": 22076 }, { "epoch": 0.7865400716105243, "grad_norm": 1.1584009992806374, "learning_rate": 2.2959710080593033e-06, "loss": 0.4014, "step": 22077 }, { "epoch": 0.7865756987370184, "grad_norm": 1.1773420272009734, "learning_rate": 2.2952353585330025e-06, "loss": 0.401, "step": 22078 }, { "epoch": 0.7866113258635125, "grad_norm": 1.8703222650739952, "learning_rate": 2.294499811601022e-06, "loss": 0.4385, "step": 22079 }, { "epoch": 0.7866469529900066, "grad_norm": 1.2553083298232668, "learning_rate": 2.2937643672731656e-06, "loss": 0.4072, "step": 22080 }, { "epoch": 0.7866825801165007, "grad_norm": 1.1542330352523353, "learning_rate": 2.2930290255592256e-06, "loss": 0.425, "step": 22081 }, { "epoch": 0.7867182072429948, "grad_norm": 1.1440154747978175, "learning_rate": 2.2922937864689888e-06, "loss": 0.3756, "step": 22082 }, { "epoch": 0.786753834369489, "grad_norm": 1.1307230296685533, "learning_rate": 2.29155865001225e-06, "loss": 0.3503, "step": 22083 }, { "epoch": 0.7867894614959831, "grad_norm": 1.3711441350546667, "learning_rate": 2.290823616198793e-06, "loss": 0.3761, "step": 22084 }, { "epoch": 0.7868250886224771, "grad_norm": 1.3598263365259173, "learning_rate": 2.2900886850384107e-06, "loss": 0.4651, "step": 22085 }, { "epoch": 0.7868607157489713, "grad_norm": 1.8864375339699033, "learning_rate": 2.2893538565408834e-06, "loss": 0.5557, "step": 22086 }, { "epoch": 0.7868963428754654, "grad_norm": 1.5987748248748317, "learning_rate": 2.2886191307159988e-06, "loss": 0.5001, "step": 22087 }, { "epoch": 0.7869319700019595, "grad_norm": 1.6080154242802025, "learning_rate": 2.2878845075735412e-06, "loss": 0.4643, "step": 22088 }, { "epoch": 0.7869675971284537, "grad_norm": 1.1232161291980538, "learning_rate": 2.287149987123294e-06, "loss": 0.3884, "step": 22089 }, { "epoch": 0.7870032242549477, "grad_norm": 1.5265116768445246, "learning_rate": 2.2864155693750334e-06, "loss": 0.5225, "step": 22090 }, { "epoch": 0.7870388513814418, "grad_norm": 1.6317674728195433, "learning_rate": 2.28568125433854e-06, "loss": 0.4686, "step": 22091 }, { "epoch": 0.787074478507936, "grad_norm": 1.1116130656616798, "learning_rate": 2.2849470420235963e-06, "loss": 0.373, "step": 22092 }, { "epoch": 0.7871101056344301, "grad_norm": 1.292659868851566, "learning_rate": 2.284212932439972e-06, "loss": 0.4354, "step": 22093 }, { "epoch": 0.7871457327609241, "grad_norm": 1.570600709468112, "learning_rate": 2.2834789255974466e-06, "loss": 0.4744, "step": 22094 }, { "epoch": 0.7871813598874183, "grad_norm": 1.5999845658650353, "learning_rate": 2.2827450215057912e-06, "loss": 0.5234, "step": 22095 }, { "epoch": 0.7872169870139124, "grad_norm": 1.4706170605988298, "learning_rate": 2.2820112201747835e-06, "loss": 0.4965, "step": 22096 }, { "epoch": 0.7872526141404065, "grad_norm": 1.110789528980393, "learning_rate": 2.281277521614188e-06, "loss": 0.3937, "step": 22097 }, { "epoch": 0.7872882412669007, "grad_norm": 1.5022584658994114, "learning_rate": 2.2805439258337803e-06, "loss": 0.5038, "step": 22098 }, { "epoch": 0.7873238683933947, "grad_norm": 1.2011425062487682, "learning_rate": 2.2798104328433237e-06, "loss": 0.3845, "step": 22099 }, { "epoch": 0.7873594955198888, "grad_norm": 1.169298788727641, "learning_rate": 2.2790770426525888e-06, "loss": 0.3985, "step": 22100 }, { "epoch": 0.787395122646383, "grad_norm": 1.2177811300914354, "learning_rate": 2.278343755271337e-06, "loss": 0.3743, "step": 22101 }, { "epoch": 0.7874307497728771, "grad_norm": 1.1842219742794904, "learning_rate": 2.2776105707093368e-06, "loss": 0.397, "step": 22102 }, { "epoch": 0.7874663768993712, "grad_norm": 1.1824462934088926, "learning_rate": 2.276877488976348e-06, "loss": 0.3886, "step": 22103 }, { "epoch": 0.7875020040258653, "grad_norm": 1.3307334365074424, "learning_rate": 2.2761445100821334e-06, "loss": 0.4061, "step": 22104 }, { "epoch": 0.7875376311523594, "grad_norm": 1.2599472902913378, "learning_rate": 2.275411634036456e-06, "loss": 0.4143, "step": 22105 }, { "epoch": 0.7875732582788535, "grad_norm": 1.6385533670596422, "learning_rate": 2.27467886084907e-06, "loss": 0.5277, "step": 22106 }, { "epoch": 0.7876088854053477, "grad_norm": 1.39330136060021, "learning_rate": 2.273946190529738e-06, "loss": 0.4493, "step": 22107 }, { "epoch": 0.7876445125318418, "grad_norm": 1.4324244475820052, "learning_rate": 2.2732136230882084e-06, "loss": 0.4097, "step": 22108 }, { "epoch": 0.7876801396583358, "grad_norm": 1.8601938719705595, "learning_rate": 2.272481158534241e-06, "loss": 0.5016, "step": 22109 }, { "epoch": 0.78771576678483, "grad_norm": 2.2087224590979875, "learning_rate": 2.2717487968775877e-06, "loss": 0.4567, "step": 22110 }, { "epoch": 0.7877513939113241, "grad_norm": 1.2952349567964856, "learning_rate": 2.2710165381280036e-06, "loss": 0.477, "step": 22111 }, { "epoch": 0.7877870210378182, "grad_norm": 1.3364534296014077, "learning_rate": 2.2702843822952337e-06, "loss": 0.5145, "step": 22112 }, { "epoch": 0.7878226481643124, "grad_norm": 1.6973159503968311, "learning_rate": 2.269552329389031e-06, "loss": 0.4839, "step": 22113 }, { "epoch": 0.7878582752908064, "grad_norm": 1.9830955437703572, "learning_rate": 2.268820379419144e-06, "loss": 0.4642, "step": 22114 }, { "epoch": 0.7878939024173005, "grad_norm": 1.2447740557756872, "learning_rate": 2.268088532395315e-06, "loss": 0.4029, "step": 22115 }, { "epoch": 0.7879295295437947, "grad_norm": 1.8501350140834043, "learning_rate": 2.267356788327294e-06, "loss": 0.5662, "step": 22116 }, { "epoch": 0.7879651566702888, "grad_norm": 1.7468379828477927, "learning_rate": 2.266625147224817e-06, "loss": 0.5471, "step": 22117 }, { "epoch": 0.7880007837967828, "grad_norm": 1.2155932852765443, "learning_rate": 2.265893609097637e-06, "loss": 0.3873, "step": 22118 }, { "epoch": 0.788036410923277, "grad_norm": 2.670049069319864, "learning_rate": 2.2651621739554864e-06, "loss": 0.4101, "step": 22119 }, { "epoch": 0.7880720380497711, "grad_norm": 1.2878906068561469, "learning_rate": 2.2644308418081108e-06, "loss": 0.3866, "step": 22120 }, { "epoch": 0.7881076651762652, "grad_norm": 1.1590260761309792, "learning_rate": 2.263699612665242e-06, "loss": 0.3938, "step": 22121 }, { "epoch": 0.7881432923027594, "grad_norm": 1.4379979750165441, "learning_rate": 2.262968486536624e-06, "loss": 0.5163, "step": 22122 }, { "epoch": 0.7881789194292534, "grad_norm": 1.2685976648289712, "learning_rate": 2.2622374634319853e-06, "loss": 0.3849, "step": 22123 }, { "epoch": 0.7882145465557475, "grad_norm": 1.4131093841325666, "learning_rate": 2.2615065433610626e-06, "loss": 0.5278, "step": 22124 }, { "epoch": 0.7882501736822417, "grad_norm": 1.1869241457692044, "learning_rate": 2.2607757263335894e-06, "loss": 0.4261, "step": 22125 }, { "epoch": 0.7882858008087358, "grad_norm": 1.6248138871590923, "learning_rate": 2.2600450123592976e-06, "loss": 0.4929, "step": 22126 }, { "epoch": 0.7883214279352299, "grad_norm": 1.4184193597460601, "learning_rate": 2.2593144014479184e-06, "loss": 0.571, "step": 22127 }, { "epoch": 0.788357055061724, "grad_norm": 1.9113524575951697, "learning_rate": 2.2585838936091753e-06, "loss": 0.538, "step": 22128 }, { "epoch": 0.7883926821882181, "grad_norm": 1.1790405884892812, "learning_rate": 2.2578534888528024e-06, "loss": 0.3836, "step": 22129 }, { "epoch": 0.7884283093147122, "grad_norm": 1.542128983669624, "learning_rate": 2.257123187188518e-06, "loss": 0.4658, "step": 22130 }, { "epoch": 0.7884639364412064, "grad_norm": 1.4311720462256519, "learning_rate": 2.2563929886260506e-06, "loss": 0.4365, "step": 22131 }, { "epoch": 0.7884995635677005, "grad_norm": 1.3563520172372592, "learning_rate": 2.255662893175122e-06, "loss": 0.4428, "step": 22132 }, { "epoch": 0.7885351906941945, "grad_norm": 1.2980712715298606, "learning_rate": 2.25493290084546e-06, "loss": 0.4498, "step": 22133 }, { "epoch": 0.7885708178206887, "grad_norm": 1.177390778845592, "learning_rate": 2.254203011646775e-06, "loss": 0.3765, "step": 22134 }, { "epoch": 0.7886064449471828, "grad_norm": 1.2951010318425764, "learning_rate": 2.253473225588795e-06, "loss": 0.5011, "step": 22135 }, { "epoch": 0.7886420720736769, "grad_norm": 1.5980331787036433, "learning_rate": 2.2527435426812307e-06, "loss": 0.5876, "step": 22136 }, { "epoch": 0.7886776992001711, "grad_norm": 1.2727982296856206, "learning_rate": 2.2520139629338e-06, "loss": 0.431, "step": 22137 }, { "epoch": 0.7887133263266651, "grad_norm": 1.376713317285257, "learning_rate": 2.251284486356222e-06, "loss": 0.4334, "step": 22138 }, { "epoch": 0.7887489534531592, "grad_norm": 1.5689470117243083, "learning_rate": 2.2505551129582047e-06, "loss": 0.5783, "step": 22139 }, { "epoch": 0.7887845805796534, "grad_norm": 1.3485177013575156, "learning_rate": 2.2498258427494636e-06, "loss": 0.458, "step": 22140 }, { "epoch": 0.7888202077061475, "grad_norm": 1.1579066796105888, "learning_rate": 2.249096675739707e-06, "loss": 0.3997, "step": 22141 }, { "epoch": 0.7888558348326415, "grad_norm": 1.823623572700964, "learning_rate": 2.2483676119386487e-06, "loss": 0.5058, "step": 22142 }, { "epoch": 0.7888914619591357, "grad_norm": 1.2350301738356029, "learning_rate": 2.247638651355991e-06, "loss": 0.4078, "step": 22143 }, { "epoch": 0.7889270890856298, "grad_norm": 1.6758919725951427, "learning_rate": 2.2469097940014474e-06, "loss": 0.5685, "step": 22144 }, { "epoch": 0.7889627162121239, "grad_norm": 1.1598318475632985, "learning_rate": 2.2461810398847138e-06, "loss": 0.3946, "step": 22145 }, { "epoch": 0.7889983433386181, "grad_norm": 1.669518793947272, "learning_rate": 2.245452389015501e-06, "loss": 0.5313, "step": 22146 }, { "epoch": 0.7890339704651121, "grad_norm": 1.185001394197789, "learning_rate": 2.244723841403509e-06, "loss": 0.3884, "step": 22147 }, { "epoch": 0.7890695975916062, "grad_norm": 1.287092610443473, "learning_rate": 2.2439953970584404e-06, "loss": 0.4289, "step": 22148 }, { "epoch": 0.7891052247181004, "grad_norm": 1.385529003567518, "learning_rate": 2.2432670559899962e-06, "loss": 0.4994, "step": 22149 }, { "epoch": 0.7891408518445945, "grad_norm": 1.339562128788026, "learning_rate": 2.24253881820787e-06, "loss": 0.438, "step": 22150 }, { "epoch": 0.7891764789710886, "grad_norm": 1.604486177545814, "learning_rate": 2.2418106837217646e-06, "loss": 0.5188, "step": 22151 }, { "epoch": 0.7892121060975827, "grad_norm": 1.2220409005914292, "learning_rate": 2.24108265254137e-06, "loss": 0.3925, "step": 22152 }, { "epoch": 0.7892477332240768, "grad_norm": 1.2277236146739339, "learning_rate": 2.2403547246763857e-06, "loss": 0.4259, "step": 22153 }, { "epoch": 0.7892833603505709, "grad_norm": 1.2572711091169726, "learning_rate": 2.239626900136496e-06, "loss": 0.4002, "step": 22154 }, { "epoch": 0.7893189874770651, "grad_norm": 1.4591685761654027, "learning_rate": 2.2388991789314043e-06, "loss": 0.4738, "step": 22155 }, { "epoch": 0.7893546146035592, "grad_norm": 1.2639831559433918, "learning_rate": 2.238171561070792e-06, "loss": 0.412, "step": 22156 }, { "epoch": 0.7893902417300532, "grad_norm": 1.297363039531197, "learning_rate": 2.2374440465643533e-06, "loss": 0.3956, "step": 22157 }, { "epoch": 0.7894258688565474, "grad_norm": 1.3076376004292658, "learning_rate": 2.2367166354217695e-06, "loss": 0.4408, "step": 22158 }, { "epoch": 0.7894614959830415, "grad_norm": 1.204416425388041, "learning_rate": 2.2359893276527335e-06, "loss": 0.3886, "step": 22159 }, { "epoch": 0.7894971231095356, "grad_norm": 1.3121396402221899, "learning_rate": 2.2352621232669225e-06, "loss": 0.4457, "step": 22160 }, { "epoch": 0.7895327502360298, "grad_norm": 1.5057062432748345, "learning_rate": 2.2345350222740246e-06, "loss": 0.507, "step": 22161 }, { "epoch": 0.7895683773625238, "grad_norm": 1.2295434551629392, "learning_rate": 2.2338080246837203e-06, "loss": 0.4169, "step": 22162 }, { "epoch": 0.7896040044890179, "grad_norm": 1.1529591005062607, "learning_rate": 2.23308113050569e-06, "loss": 0.3916, "step": 22163 }, { "epoch": 0.7896396316155121, "grad_norm": 1.3252876397248594, "learning_rate": 2.232354339749616e-06, "loss": 0.4504, "step": 22164 }, { "epoch": 0.7896752587420062, "grad_norm": 1.4714686328335527, "learning_rate": 2.231627652425171e-06, "loss": 0.4908, "step": 22165 }, { "epoch": 0.7897108858685002, "grad_norm": 1.2257136754382922, "learning_rate": 2.2309010685420365e-06, "loss": 0.351, "step": 22166 }, { "epoch": 0.7897465129949944, "grad_norm": 1.291933662203304, "learning_rate": 2.23017458810988e-06, "loss": 0.5052, "step": 22167 }, { "epoch": 0.7897821401214885, "grad_norm": 1.4123737751345506, "learning_rate": 2.2294482111383817e-06, "loss": 0.5804, "step": 22168 }, { "epoch": 0.7898177672479826, "grad_norm": 1.8101964534253079, "learning_rate": 2.2287219376372104e-06, "loss": 0.5577, "step": 22169 }, { "epoch": 0.7898533943744768, "grad_norm": 1.1796101394807068, "learning_rate": 2.2279957676160425e-06, "loss": 0.4267, "step": 22170 }, { "epoch": 0.7898890215009708, "grad_norm": 1.2193784391774742, "learning_rate": 2.227269701084539e-06, "loss": 0.4174, "step": 22171 }, { "epoch": 0.7899246486274649, "grad_norm": 1.3451180605942474, "learning_rate": 2.2265437380523736e-06, "loss": 0.4653, "step": 22172 }, { "epoch": 0.7899602757539591, "grad_norm": 1.7415624009416693, "learning_rate": 2.225817878529214e-06, "loss": 0.4774, "step": 22173 }, { "epoch": 0.7899959028804532, "grad_norm": 1.5358626799395152, "learning_rate": 2.2250921225247192e-06, "loss": 0.5062, "step": 22174 }, { "epoch": 0.7900315300069473, "grad_norm": 1.089489070244644, "learning_rate": 2.224366470048561e-06, "loss": 0.3752, "step": 22175 }, { "epoch": 0.7900671571334414, "grad_norm": 1.6721975381261414, "learning_rate": 2.223640921110395e-06, "loss": 0.5349, "step": 22176 }, { "epoch": 0.7901027842599355, "grad_norm": 1.1745782858792473, "learning_rate": 2.2229154757198855e-06, "loss": 0.384, "step": 22177 }, { "epoch": 0.7901384113864296, "grad_norm": 1.2173044588447863, "learning_rate": 2.2221901338866925e-06, "loss": 0.4185, "step": 22178 }, { "epoch": 0.7901740385129238, "grad_norm": 1.151563468945928, "learning_rate": 2.2214648956204775e-06, "loss": 0.4117, "step": 22179 }, { "epoch": 0.7902096656394179, "grad_norm": 1.206174915153149, "learning_rate": 2.2207397609308913e-06, "loss": 0.4061, "step": 22180 }, { "epoch": 0.7902452927659119, "grad_norm": 1.6255720600204828, "learning_rate": 2.2200147298275943e-06, "loss": 0.5171, "step": 22181 }, { "epoch": 0.7902809198924061, "grad_norm": 1.5006891346176856, "learning_rate": 2.219289802320237e-06, "loss": 0.4322, "step": 22182 }, { "epoch": 0.7903165470189002, "grad_norm": 1.3077811034510052, "learning_rate": 2.218564978418475e-06, "loss": 0.4257, "step": 22183 }, { "epoch": 0.7903521741453943, "grad_norm": 1.4844063436737167, "learning_rate": 2.217840258131958e-06, "loss": 0.4699, "step": 22184 }, { "epoch": 0.7903878012718885, "grad_norm": 1.6395949077053344, "learning_rate": 2.2171156414703377e-06, "loss": 0.4911, "step": 22185 }, { "epoch": 0.7904234283983825, "grad_norm": 1.9896331974576518, "learning_rate": 2.216391128443266e-06, "loss": 0.4912, "step": 22186 }, { "epoch": 0.7904590555248766, "grad_norm": 1.2779071163262572, "learning_rate": 2.215666719060383e-06, "loss": 0.481, "step": 22187 }, { "epoch": 0.7904946826513708, "grad_norm": 1.718769888042034, "learning_rate": 2.2149424133313425e-06, "loss": 0.5019, "step": 22188 }, { "epoch": 0.7905303097778649, "grad_norm": 1.6000365493770965, "learning_rate": 2.2142182112657816e-06, "loss": 0.5443, "step": 22189 }, { "epoch": 0.790565936904359, "grad_norm": 1.4403558966326266, "learning_rate": 2.21349411287335e-06, "loss": 0.491, "step": 22190 }, { "epoch": 0.7906015640308531, "grad_norm": 1.235167201707985, "learning_rate": 2.2127701181636853e-06, "loss": 0.4405, "step": 22191 }, { "epoch": 0.7906371911573472, "grad_norm": 1.7192879842441462, "learning_rate": 2.2120462271464273e-06, "loss": 0.4661, "step": 22192 }, { "epoch": 0.7906728182838413, "grad_norm": 1.4769433406540042, "learning_rate": 2.2113224398312185e-06, "loss": 0.4297, "step": 22193 }, { "epoch": 0.7907084454103355, "grad_norm": 1.4922876885247327, "learning_rate": 2.2105987562276954e-06, "loss": 0.4613, "step": 22194 }, { "epoch": 0.7907440725368295, "grad_norm": 1.4136535647557067, "learning_rate": 2.2098751763454962e-06, "loss": 0.4753, "step": 22195 }, { "epoch": 0.7907796996633236, "grad_norm": 1.8458851100924138, "learning_rate": 2.209151700194252e-06, "loss": 0.4757, "step": 22196 }, { "epoch": 0.7908153267898178, "grad_norm": 1.4288122291378407, "learning_rate": 2.2084283277836015e-06, "loss": 0.5243, "step": 22197 }, { "epoch": 0.7908509539163119, "grad_norm": 1.3868693315547507, "learning_rate": 2.2077050591231695e-06, "loss": 0.4626, "step": 22198 }, { "epoch": 0.790886581042806, "grad_norm": 1.2577295706895912, "learning_rate": 2.206981894222593e-06, "loss": 0.4798, "step": 22199 }, { "epoch": 0.7909222081693001, "grad_norm": 1.4144552393578418, "learning_rate": 2.206258833091498e-06, "loss": 0.5823, "step": 22200 }, { "epoch": 0.7909578352957942, "grad_norm": 1.6901536682788239, "learning_rate": 2.2055358757395183e-06, "loss": 0.5662, "step": 22201 }, { "epoch": 0.7909934624222883, "grad_norm": 1.403588040733832, "learning_rate": 2.204813022176273e-06, "loss": 0.4002, "step": 22202 }, { "epoch": 0.7910290895487825, "grad_norm": 1.1633822133418912, "learning_rate": 2.2040902724113934e-06, "loss": 0.3984, "step": 22203 }, { "epoch": 0.7910647166752766, "grad_norm": 1.5691259607777703, "learning_rate": 2.203367626454498e-06, "loss": 0.5619, "step": 22204 }, { "epoch": 0.7911003438017706, "grad_norm": 1.5059108112232757, "learning_rate": 2.2026450843152156e-06, "loss": 0.6225, "step": 22205 }, { "epoch": 0.7911359709282648, "grad_norm": 1.236045103430089, "learning_rate": 2.201922646003157e-06, "loss": 0.3921, "step": 22206 }, { "epoch": 0.7911715980547589, "grad_norm": 1.1638095359252023, "learning_rate": 2.2012003115279533e-06, "loss": 0.3575, "step": 22207 }, { "epoch": 0.791207225181253, "grad_norm": 1.9015293185218947, "learning_rate": 2.200478080899221e-06, "loss": 0.5329, "step": 22208 }, { "epoch": 0.7912428523077472, "grad_norm": 1.2427250279909379, "learning_rate": 2.199755954126571e-06, "loss": 0.3877, "step": 22209 }, { "epoch": 0.7912784794342412, "grad_norm": 1.6765993732276474, "learning_rate": 2.1990339312196263e-06, "loss": 0.518, "step": 22210 }, { "epoch": 0.7913141065607353, "grad_norm": 1.2976066096879102, "learning_rate": 2.198312012187993e-06, "loss": 0.4474, "step": 22211 }, { "epoch": 0.7913497336872295, "grad_norm": 1.5027737218485926, "learning_rate": 2.1975901970412927e-06, "loss": 0.4374, "step": 22212 }, { "epoch": 0.7913853608137236, "grad_norm": 1.2407540116780256, "learning_rate": 2.1968684857891285e-06, "loss": 0.4211, "step": 22213 }, { "epoch": 0.7914209879402176, "grad_norm": 1.2891611486316954, "learning_rate": 2.1961468784411153e-06, "loss": 0.4636, "step": 22214 }, { "epoch": 0.7914566150667118, "grad_norm": 1.4495953619920066, "learning_rate": 2.195425375006861e-06, "loss": 0.4568, "step": 22215 }, { "epoch": 0.7914922421932059, "grad_norm": 1.1944819465667165, "learning_rate": 2.1947039754959755e-06, "loss": 0.434, "step": 22216 }, { "epoch": 0.7915278693197, "grad_norm": 1.272908103971974, "learning_rate": 2.1939826799180584e-06, "loss": 0.4508, "step": 22217 }, { "epoch": 0.7915634964461942, "grad_norm": 1.4185909573676327, "learning_rate": 2.1932614882827196e-06, "loss": 0.5106, "step": 22218 }, { "epoch": 0.7915991235726882, "grad_norm": 1.2858895931540955, "learning_rate": 2.192540400599562e-06, "loss": 0.4483, "step": 22219 }, { "epoch": 0.7916347506991823, "grad_norm": 1.253336624523869, "learning_rate": 2.1918194168781847e-06, "loss": 0.4163, "step": 22220 }, { "epoch": 0.7916703778256765, "grad_norm": 1.442386029598376, "learning_rate": 2.191098537128189e-06, "loss": 0.4984, "step": 22221 }, { "epoch": 0.7917060049521706, "grad_norm": 1.542562736715187, "learning_rate": 2.1903777613591737e-06, "loss": 0.5275, "step": 22222 }, { "epoch": 0.7917416320786647, "grad_norm": 1.526731092734528, "learning_rate": 2.1896570895807403e-06, "loss": 0.5011, "step": 22223 }, { "epoch": 0.7917772592051588, "grad_norm": 1.4854793281733272, "learning_rate": 2.1889365218024796e-06, "loss": 0.4507, "step": 22224 }, { "epoch": 0.7918128863316529, "grad_norm": 1.1928570049913627, "learning_rate": 2.1882160580339907e-06, "loss": 0.4079, "step": 22225 }, { "epoch": 0.791848513458147, "grad_norm": 1.1587828690987236, "learning_rate": 2.1874956982848628e-06, "loss": 0.3744, "step": 22226 }, { "epoch": 0.7918841405846412, "grad_norm": 1.5964230141663527, "learning_rate": 2.1867754425646923e-06, "loss": 0.5913, "step": 22227 }, { "epoch": 0.7919197677111353, "grad_norm": 1.2976431626175184, "learning_rate": 2.1860552908830657e-06, "loss": 0.4082, "step": 22228 }, { "epoch": 0.7919553948376293, "grad_norm": 1.193259575377468, "learning_rate": 2.1853352432495724e-06, "loss": 0.4595, "step": 22229 }, { "epoch": 0.7919910219641235, "grad_norm": 1.2697998864943858, "learning_rate": 2.184615299673807e-06, "loss": 0.4711, "step": 22230 }, { "epoch": 0.7920266490906176, "grad_norm": 1.1514053150939674, "learning_rate": 2.1838954601653485e-06, "loss": 0.4116, "step": 22231 }, { "epoch": 0.7920622762171117, "grad_norm": 1.2460878031552367, "learning_rate": 2.1831757247337895e-06, "loss": 0.4682, "step": 22232 }, { "epoch": 0.7920979033436059, "grad_norm": 1.508149556193032, "learning_rate": 2.1824560933887053e-06, "loss": 0.463, "step": 22233 }, { "epoch": 0.7921335304700999, "grad_norm": 1.1324601854884855, "learning_rate": 2.181736566139685e-06, "loss": 0.371, "step": 22234 }, { "epoch": 0.792169157596594, "grad_norm": 1.5301035477379803, "learning_rate": 2.1810171429963044e-06, "loss": 0.469, "step": 22235 }, { "epoch": 0.7922047847230882, "grad_norm": 1.5172528710927298, "learning_rate": 2.180297823968146e-06, "loss": 0.4301, "step": 22236 }, { "epoch": 0.7922404118495823, "grad_norm": 1.546438240289221, "learning_rate": 2.1795786090647875e-06, "loss": 0.5226, "step": 22237 }, { "epoch": 0.7922760389760763, "grad_norm": 1.2671229312826267, "learning_rate": 2.1788594982958087e-06, "loss": 0.4116, "step": 22238 }, { "epoch": 0.7923116661025705, "grad_norm": 1.5691913004465718, "learning_rate": 2.1781404916707793e-06, "loss": 0.4153, "step": 22239 }, { "epoch": 0.7923472932290646, "grad_norm": 1.4708281615518757, "learning_rate": 2.1774215891992767e-06, "loss": 0.498, "step": 22240 }, { "epoch": 0.7923829203555587, "grad_norm": 1.4934772537786662, "learning_rate": 2.1767027908908767e-06, "loss": 0.4854, "step": 22241 }, { "epoch": 0.7924185474820529, "grad_norm": 1.3128787169388105, "learning_rate": 2.1759840967551438e-06, "loss": 0.4172, "step": 22242 }, { "epoch": 0.792454174608547, "grad_norm": 1.3200554708616363, "learning_rate": 2.1752655068016516e-06, "loss": 0.4336, "step": 22243 }, { "epoch": 0.792489801735041, "grad_norm": 1.3741490391448874, "learning_rate": 2.1745470210399676e-06, "loss": 0.4249, "step": 22244 }, { "epoch": 0.7925254288615352, "grad_norm": 2.194475991810288, "learning_rate": 2.1738286394796625e-06, "loss": 0.4544, "step": 22245 }, { "epoch": 0.7925610559880293, "grad_norm": 1.4589976817910282, "learning_rate": 2.1731103621302975e-06, "loss": 0.4917, "step": 22246 }, { "epoch": 0.7925966831145234, "grad_norm": 1.4581678626540588, "learning_rate": 2.172392189001441e-06, "loss": 0.5904, "step": 22247 }, { "epoch": 0.7926323102410175, "grad_norm": 1.2632684198966226, "learning_rate": 2.1716741201026503e-06, "loss": 0.4158, "step": 22248 }, { "epoch": 0.7926679373675116, "grad_norm": 1.1950066940719462, "learning_rate": 2.1709561554434933e-06, "loss": 0.4103, "step": 22249 }, { "epoch": 0.7927035644940057, "grad_norm": 1.318039936045414, "learning_rate": 2.1702382950335257e-06, "loss": 0.4284, "step": 22250 }, { "epoch": 0.7927391916204999, "grad_norm": 1.114782747321264, "learning_rate": 2.169520538882307e-06, "loss": 0.3907, "step": 22251 }, { "epoch": 0.792774818746994, "grad_norm": 1.215165653090527, "learning_rate": 2.168802886999395e-06, "loss": 0.4137, "step": 22252 }, { "epoch": 0.792810445873488, "grad_norm": 1.3662553457862652, "learning_rate": 2.1680853393943467e-06, "loss": 0.4271, "step": 22253 }, { "epoch": 0.7928460729999822, "grad_norm": 1.6198443425937368, "learning_rate": 2.16736789607672e-06, "loss": 0.5523, "step": 22254 }, { "epoch": 0.7928817001264763, "grad_norm": 1.8118670511992114, "learning_rate": 2.1666505570560605e-06, "loss": 0.5545, "step": 22255 }, { "epoch": 0.7929173272529704, "grad_norm": 1.8328190425462727, "learning_rate": 2.1659333223419267e-06, "loss": 0.4656, "step": 22256 }, { "epoch": 0.7929529543794646, "grad_norm": 1.3352208860633679, "learning_rate": 2.165216191943864e-06, "loss": 0.3995, "step": 22257 }, { "epoch": 0.7929885815059586, "grad_norm": 1.1633144003123361, "learning_rate": 2.1644991658714233e-06, "loss": 0.3621, "step": 22258 }, { "epoch": 0.7930242086324527, "grad_norm": 1.4658888461081523, "learning_rate": 2.163782244134154e-06, "loss": 0.5596, "step": 22259 }, { "epoch": 0.7930598357589469, "grad_norm": 1.4218806730631, "learning_rate": 2.163065426741603e-06, "loss": 0.3885, "step": 22260 }, { "epoch": 0.793095462885441, "grad_norm": 1.361167133600951, "learning_rate": 2.162348713703312e-06, "loss": 0.4538, "step": 22261 }, { "epoch": 0.793131090011935, "grad_norm": 1.1828047980096685, "learning_rate": 2.161632105028828e-06, "loss": 0.4025, "step": 22262 }, { "epoch": 0.7931667171384292, "grad_norm": 1.3331047808736292, "learning_rate": 2.160915600727688e-06, "loss": 0.4724, "step": 22263 }, { "epoch": 0.7932023442649233, "grad_norm": 1.282370637029382, "learning_rate": 2.1601992008094363e-06, "loss": 0.4997, "step": 22264 }, { "epoch": 0.7932379713914174, "grad_norm": 1.2047549756202203, "learning_rate": 2.1594829052836154e-06, "loss": 0.3766, "step": 22265 }, { "epoch": 0.7932735985179116, "grad_norm": 1.1246582572905894, "learning_rate": 2.158766714159757e-06, "loss": 0.3351, "step": 22266 }, { "epoch": 0.7933092256444056, "grad_norm": 1.3572265513868156, "learning_rate": 2.1580506274474e-06, "loss": 0.4446, "step": 22267 }, { "epoch": 0.7933448527708997, "grad_norm": 1.2995255237969896, "learning_rate": 2.1573346451560797e-06, "loss": 0.48, "step": 22268 }, { "epoch": 0.7933804798973939, "grad_norm": 1.2508335582781642, "learning_rate": 2.156618767295333e-06, "loss": 0.4049, "step": 22269 }, { "epoch": 0.793416107023888, "grad_norm": 1.1688775382960355, "learning_rate": 2.155902993874688e-06, "loss": 0.4099, "step": 22270 }, { "epoch": 0.7934517341503821, "grad_norm": 1.762244552294041, "learning_rate": 2.155187324903679e-06, "loss": 0.5176, "step": 22271 }, { "epoch": 0.7934873612768762, "grad_norm": 1.9623228427599346, "learning_rate": 2.154471760391832e-06, "loss": 0.4479, "step": 22272 }, { "epoch": 0.7935229884033703, "grad_norm": 1.4559323866704874, "learning_rate": 2.153756300348677e-06, "loss": 0.442, "step": 22273 }, { "epoch": 0.7935586155298644, "grad_norm": 1.5700513517144028, "learning_rate": 2.153040944783741e-06, "loss": 0.5494, "step": 22274 }, { "epoch": 0.7935942426563586, "grad_norm": 1.2734132835970133, "learning_rate": 2.1523256937065496e-06, "loss": 0.465, "step": 22275 }, { "epoch": 0.7936298697828527, "grad_norm": 1.3467916487140634, "learning_rate": 2.151610547126629e-06, "loss": 0.3661, "step": 22276 }, { "epoch": 0.7936654969093467, "grad_norm": 1.1546065501386997, "learning_rate": 2.1508955050534975e-06, "loss": 0.4262, "step": 22277 }, { "epoch": 0.7937011240358409, "grad_norm": 1.7976185362487231, "learning_rate": 2.150180567496681e-06, "loss": 0.5161, "step": 22278 }, { "epoch": 0.793736751162335, "grad_norm": 1.5075684110659449, "learning_rate": 2.1494657344656944e-06, "loss": 0.5145, "step": 22279 }, { "epoch": 0.7937723782888291, "grad_norm": 1.2008737064689814, "learning_rate": 2.1487510059700612e-06, "loss": 0.4287, "step": 22280 }, { "epoch": 0.7938080054153233, "grad_norm": 1.215306755546772, "learning_rate": 2.148036382019291e-06, "loss": 0.449, "step": 22281 }, { "epoch": 0.7938436325418173, "grad_norm": 1.388250461280845, "learning_rate": 2.14732186262291e-06, "loss": 0.3959, "step": 22282 }, { "epoch": 0.7938792596683114, "grad_norm": 1.2567245362608561, "learning_rate": 2.1466074477904244e-06, "loss": 0.43, "step": 22283 }, { "epoch": 0.7939148867948056, "grad_norm": 1.5050318942385732, "learning_rate": 2.145893137531352e-06, "loss": 0.4689, "step": 22284 }, { "epoch": 0.7939505139212997, "grad_norm": 1.6660568847213053, "learning_rate": 2.1451789318551998e-06, "loss": 0.5908, "step": 22285 }, { "epoch": 0.7939861410477937, "grad_norm": 1.2476280475671813, "learning_rate": 2.144464830771483e-06, "loss": 0.4019, "step": 22286 }, { "epoch": 0.7940217681742879, "grad_norm": 1.2671628902721341, "learning_rate": 2.143750834289705e-06, "loss": 0.4209, "step": 22287 }, { "epoch": 0.794057395300782, "grad_norm": 1.1485570548732216, "learning_rate": 2.1430369424193754e-06, "loss": 0.4091, "step": 22288 }, { "epoch": 0.7940930224272761, "grad_norm": 1.4131500288282217, "learning_rate": 2.142323155170001e-06, "loss": 0.4679, "step": 22289 }, { "epoch": 0.7941286495537703, "grad_norm": 1.3637405712046198, "learning_rate": 2.141609472551086e-06, "loss": 0.498, "step": 22290 }, { "epoch": 0.7941642766802643, "grad_norm": 1.7285971749892548, "learning_rate": 2.140895894572136e-06, "loss": 0.5084, "step": 22291 }, { "epoch": 0.7941999038067584, "grad_norm": 1.2213815031167368, "learning_rate": 2.140182421242648e-06, "loss": 0.4257, "step": 22292 }, { "epoch": 0.7942355309332526, "grad_norm": 1.2222696385335157, "learning_rate": 2.1394690525721275e-06, "loss": 0.4236, "step": 22293 }, { "epoch": 0.7942711580597467, "grad_norm": 1.0898832922174921, "learning_rate": 2.1387557885700683e-06, "loss": 0.3938, "step": 22294 }, { "epoch": 0.7943067851862408, "grad_norm": 1.2260568440586836, "learning_rate": 2.13804262924597e-06, "loss": 0.4057, "step": 22295 }, { "epoch": 0.794342412312735, "grad_norm": 1.3327625366742695, "learning_rate": 2.137329574609329e-06, "loss": 0.4007, "step": 22296 }, { "epoch": 0.794378039439229, "grad_norm": 1.2440483208544022, "learning_rate": 2.136616624669643e-06, "loss": 0.3797, "step": 22297 }, { "epoch": 0.7944136665657231, "grad_norm": 1.4296641831842318, "learning_rate": 2.1359037794364003e-06, "loss": 0.5618, "step": 22298 }, { "epoch": 0.7944492936922173, "grad_norm": 1.6404298957173709, "learning_rate": 2.1351910389190953e-06, "loss": 0.5413, "step": 22299 }, { "epoch": 0.7944849208187114, "grad_norm": 1.493704922192756, "learning_rate": 2.1344784031272224e-06, "loss": 0.4118, "step": 22300 }, { "epoch": 0.7945205479452054, "grad_norm": 1.1182012587781576, "learning_rate": 2.1337658720702636e-06, "loss": 0.3686, "step": 22301 }, { "epoch": 0.7945561750716996, "grad_norm": 1.2138618098376022, "learning_rate": 2.1330534457577133e-06, "loss": 0.3836, "step": 22302 }, { "epoch": 0.7945918021981937, "grad_norm": 1.6147478288177137, "learning_rate": 2.132341124199053e-06, "loss": 0.5534, "step": 22303 }, { "epoch": 0.7946274293246878, "grad_norm": 1.749785477472121, "learning_rate": 2.1316289074037687e-06, "loss": 0.5356, "step": 22304 }, { "epoch": 0.794663056451182, "grad_norm": 1.237607793375414, "learning_rate": 2.130916795381346e-06, "loss": 0.3791, "step": 22305 }, { "epoch": 0.794698683577676, "grad_norm": 1.4947478116850919, "learning_rate": 2.130204788141269e-06, "loss": 0.4736, "step": 22306 }, { "epoch": 0.7947343107041701, "grad_norm": 1.2415617402836658, "learning_rate": 2.129492885693013e-06, "loss": 0.4136, "step": 22307 }, { "epoch": 0.7947699378306643, "grad_norm": 1.2858886604070707, "learning_rate": 2.1287810880460636e-06, "loss": 0.4089, "step": 22308 }, { "epoch": 0.7948055649571584, "grad_norm": 1.6180871369771797, "learning_rate": 2.128069395209892e-06, "loss": 0.4412, "step": 22309 }, { "epoch": 0.7948411920836524, "grad_norm": 1.175194758794801, "learning_rate": 2.127357807193979e-06, "loss": 0.3794, "step": 22310 }, { "epoch": 0.7948768192101466, "grad_norm": 1.2379832370316304, "learning_rate": 2.1266463240077995e-06, "loss": 0.4582, "step": 22311 }, { "epoch": 0.7949124463366407, "grad_norm": 1.1795526979023423, "learning_rate": 2.125934945660827e-06, "loss": 0.4233, "step": 22312 }, { "epoch": 0.7949480734631348, "grad_norm": 1.1560210214620201, "learning_rate": 2.1252236721625375e-06, "loss": 0.4352, "step": 22313 }, { "epoch": 0.794983700589629, "grad_norm": 1.2489980579476334, "learning_rate": 2.1245125035223956e-06, "loss": 0.4058, "step": 22314 }, { "epoch": 0.795019327716123, "grad_norm": 1.1749654109113212, "learning_rate": 2.1238014397498776e-06, "loss": 0.4216, "step": 22315 }, { "epoch": 0.7950549548426171, "grad_norm": 1.6054060039805473, "learning_rate": 2.123090480854446e-06, "loss": 0.5278, "step": 22316 }, { "epoch": 0.7950905819691113, "grad_norm": 1.082366641784289, "learning_rate": 2.1223796268455733e-06, "loss": 0.3926, "step": 22317 }, { "epoch": 0.7951262090956054, "grad_norm": 2.863217761076777, "learning_rate": 2.1216688777327153e-06, "loss": 0.4024, "step": 22318 }, { "epoch": 0.7951618362220995, "grad_norm": 1.221990892272369, "learning_rate": 2.1209582335253486e-06, "loss": 0.401, "step": 22319 }, { "epoch": 0.7951974633485936, "grad_norm": 1.211290464777862, "learning_rate": 2.1202476942329286e-06, "loss": 0.4144, "step": 22320 }, { "epoch": 0.7952330904750877, "grad_norm": 1.221218896076252, "learning_rate": 2.1195372598649167e-06, "loss": 0.4369, "step": 22321 }, { "epoch": 0.7952687176015818, "grad_norm": 1.296035216384598, "learning_rate": 2.1188269304307785e-06, "loss": 0.3786, "step": 22322 }, { "epoch": 0.795304344728076, "grad_norm": 1.277865809708174, "learning_rate": 2.1181167059399643e-06, "loss": 0.4737, "step": 22323 }, { "epoch": 0.7953399718545701, "grad_norm": 1.4721741073181784, "learning_rate": 2.117406586401939e-06, "loss": 0.5365, "step": 22324 }, { "epoch": 0.7953755989810641, "grad_norm": 1.1965765456520052, "learning_rate": 2.116696571826151e-06, "loss": 0.4125, "step": 22325 }, { "epoch": 0.7954112261075583, "grad_norm": 1.6744594363231313, "learning_rate": 2.115986662222058e-06, "loss": 0.5698, "step": 22326 }, { "epoch": 0.7954468532340524, "grad_norm": 1.7679692234987732, "learning_rate": 2.1152768575991146e-06, "loss": 0.5248, "step": 22327 }, { "epoch": 0.7954824803605465, "grad_norm": 1.2025324231865768, "learning_rate": 2.1145671579667727e-06, "loss": 0.3943, "step": 22328 }, { "epoch": 0.7955181074870407, "grad_norm": 2.4638783143940772, "learning_rate": 2.113857563334478e-06, "loss": 0.4599, "step": 22329 }, { "epoch": 0.7955537346135347, "grad_norm": 1.7935236772371765, "learning_rate": 2.113148073711685e-06, "loss": 0.5356, "step": 22330 }, { "epoch": 0.7955893617400288, "grad_norm": 1.5736731953368368, "learning_rate": 2.112438689107834e-06, "loss": 0.4671, "step": 22331 }, { "epoch": 0.795624988866523, "grad_norm": 1.6588394230941046, "learning_rate": 2.111729409532377e-06, "loss": 0.5555, "step": 22332 }, { "epoch": 0.7956606159930171, "grad_norm": 1.432048978743875, "learning_rate": 2.111020234994755e-06, "loss": 0.3963, "step": 22333 }, { "epoch": 0.7956962431195111, "grad_norm": 1.2346246842367234, "learning_rate": 2.110311165504413e-06, "loss": 0.4135, "step": 22334 }, { "epoch": 0.7957318702460053, "grad_norm": 1.5432470709182602, "learning_rate": 2.109602201070795e-06, "loss": 0.5431, "step": 22335 }, { "epoch": 0.7957674973724994, "grad_norm": 1.8767809872618095, "learning_rate": 2.1088933417033365e-06, "loss": 0.5067, "step": 22336 }, { "epoch": 0.7958031244989935, "grad_norm": 1.1843191130203121, "learning_rate": 2.1081845874114815e-06, "loss": 0.3895, "step": 22337 }, { "epoch": 0.7958387516254877, "grad_norm": 1.598567593312593, "learning_rate": 2.1074759382046617e-06, "loss": 0.4603, "step": 22338 }, { "epoch": 0.7958743787519817, "grad_norm": 1.7527850789045427, "learning_rate": 2.1067673940923185e-06, "loss": 0.4287, "step": 22339 }, { "epoch": 0.7959100058784758, "grad_norm": 1.3985994628608704, "learning_rate": 2.106058955083883e-06, "loss": 0.5096, "step": 22340 }, { "epoch": 0.79594563300497, "grad_norm": 1.593174375836929, "learning_rate": 2.1053506211887885e-06, "loss": 0.5029, "step": 22341 }, { "epoch": 0.7959812601314641, "grad_norm": 1.8517623494540345, "learning_rate": 2.1046423924164693e-06, "loss": 0.526, "step": 22342 }, { "epoch": 0.7960168872579582, "grad_norm": 1.6181394902685942, "learning_rate": 2.1039342687763587e-06, "loss": 0.5343, "step": 22343 }, { "epoch": 0.7960525143844523, "grad_norm": 1.1530514161736154, "learning_rate": 2.1032262502778787e-06, "loss": 0.3678, "step": 22344 }, { "epoch": 0.7960881415109464, "grad_norm": 1.405842490184276, "learning_rate": 2.1025183369304604e-06, "loss": 0.4338, "step": 22345 }, { "epoch": 0.7961237686374405, "grad_norm": 1.3716275327974408, "learning_rate": 2.101810528743534e-06, "loss": 0.431, "step": 22346 }, { "epoch": 0.7961593957639347, "grad_norm": 1.3479451352479817, "learning_rate": 2.1011028257265176e-06, "loss": 0.4128, "step": 22347 }, { "epoch": 0.7961950228904288, "grad_norm": 1.503003962422423, "learning_rate": 2.1003952278888382e-06, "loss": 0.5269, "step": 22348 }, { "epoch": 0.7962306500169228, "grad_norm": 1.339380238122967, "learning_rate": 2.099687735239918e-06, "loss": 0.4589, "step": 22349 }, { "epoch": 0.796266277143417, "grad_norm": 1.1694135468674127, "learning_rate": 2.0989803477891803e-06, "loss": 0.4169, "step": 22350 }, { "epoch": 0.7963019042699111, "grad_norm": 1.525075324489433, "learning_rate": 2.0982730655460393e-06, "loss": 0.5195, "step": 22351 }, { "epoch": 0.7963375313964052, "grad_norm": 1.665341466507466, "learning_rate": 2.097565888519917e-06, "loss": 0.4376, "step": 22352 }, { "epoch": 0.7963731585228994, "grad_norm": 1.4306055453071425, "learning_rate": 2.0968588167202265e-06, "loss": 0.4924, "step": 22353 }, { "epoch": 0.7964087856493934, "grad_norm": 1.210627159785064, "learning_rate": 2.096151850156388e-06, "loss": 0.3558, "step": 22354 }, { "epoch": 0.7964444127758875, "grad_norm": 1.891949795493627, "learning_rate": 2.095444988837808e-06, "loss": 0.5269, "step": 22355 }, { "epoch": 0.7964800399023817, "grad_norm": 1.547689484143512, "learning_rate": 2.094738232773902e-06, "loss": 0.4784, "step": 22356 }, { "epoch": 0.7965156670288758, "grad_norm": 1.683503507640416, "learning_rate": 2.0940315819740865e-06, "loss": 0.5809, "step": 22357 }, { "epoch": 0.7965512941553698, "grad_norm": 1.4754410162419949, "learning_rate": 2.093325036447763e-06, "loss": 0.5764, "step": 22358 }, { "epoch": 0.796586921281864, "grad_norm": 1.4700264009694728, "learning_rate": 2.0926185962043465e-06, "loss": 0.4396, "step": 22359 }, { "epoch": 0.7966225484083581, "grad_norm": 1.3101742356804915, "learning_rate": 2.091912261253237e-06, "loss": 0.4646, "step": 22360 }, { "epoch": 0.7966581755348522, "grad_norm": 1.456901964857739, "learning_rate": 2.0912060316038463e-06, "loss": 0.4158, "step": 22361 }, { "epoch": 0.7966938026613464, "grad_norm": 1.2690068154040803, "learning_rate": 2.090499907265573e-06, "loss": 0.4234, "step": 22362 }, { "epoch": 0.7967294297878404, "grad_norm": 1.4790553705719938, "learning_rate": 2.089793888247822e-06, "loss": 0.5056, "step": 22363 }, { "epoch": 0.7967650569143345, "grad_norm": 1.3106349320364779, "learning_rate": 2.0890879745599936e-06, "loss": 0.4441, "step": 22364 }, { "epoch": 0.7968006840408287, "grad_norm": 1.309929413133634, "learning_rate": 2.0883821662114902e-06, "loss": 0.3913, "step": 22365 }, { "epoch": 0.7968363111673228, "grad_norm": 1.7772386232925985, "learning_rate": 2.0876764632117064e-06, "loss": 0.4125, "step": 22366 }, { "epoch": 0.7968719382938169, "grad_norm": 1.2509457961608725, "learning_rate": 2.0869708655700406e-06, "loss": 0.3713, "step": 22367 }, { "epoch": 0.796907565420311, "grad_norm": 2.689438663859388, "learning_rate": 2.0862653732958914e-06, "loss": 0.4384, "step": 22368 }, { "epoch": 0.7969431925468051, "grad_norm": 1.5761520240467948, "learning_rate": 2.0855599863986475e-06, "loss": 0.5274, "step": 22369 }, { "epoch": 0.7969788196732992, "grad_norm": 1.7444638398843084, "learning_rate": 2.0848547048877042e-06, "loss": 0.5419, "step": 22370 }, { "epoch": 0.7970144467997934, "grad_norm": 1.540298693309749, "learning_rate": 2.084149528772453e-06, "loss": 0.5223, "step": 22371 }, { "epoch": 0.7970500739262875, "grad_norm": 1.4082097821634076, "learning_rate": 2.083444458062286e-06, "loss": 0.5329, "step": 22372 }, { "epoch": 0.7970857010527815, "grad_norm": 1.2629184810822875, "learning_rate": 2.0827394927665865e-06, "loss": 0.4218, "step": 22373 }, { "epoch": 0.7971213281792757, "grad_norm": 1.3295908484046706, "learning_rate": 2.0820346328947484e-06, "loss": 0.3821, "step": 22374 }, { "epoch": 0.7971569553057698, "grad_norm": 1.1357703757468953, "learning_rate": 2.0813298784561497e-06, "loss": 0.3606, "step": 22375 }, { "epoch": 0.7971925824322639, "grad_norm": 1.4331533404164718, "learning_rate": 2.0806252294601815e-06, "loss": 0.5368, "step": 22376 }, { "epoch": 0.7972282095587581, "grad_norm": 1.2824258814824716, "learning_rate": 2.0799206859162213e-06, "loss": 0.4564, "step": 22377 }, { "epoch": 0.7972638366852521, "grad_norm": 1.663178225530952, "learning_rate": 2.079216247833652e-06, "loss": 0.5293, "step": 22378 }, { "epoch": 0.7972994638117462, "grad_norm": 1.4902470578682026, "learning_rate": 2.078511915221856e-06, "loss": 0.4032, "step": 22379 }, { "epoch": 0.7973350909382404, "grad_norm": 9.606561872731636, "learning_rate": 2.0778076880902096e-06, "loss": 0.3844, "step": 22380 }, { "epoch": 0.7973707180647345, "grad_norm": 1.25905066599098, "learning_rate": 2.0771035664480944e-06, "loss": 0.4482, "step": 22381 }, { "epoch": 0.7974063451912285, "grad_norm": 1.4006746280233078, "learning_rate": 2.0763995503048793e-06, "loss": 0.4999, "step": 22382 }, { "epoch": 0.7974419723177227, "grad_norm": 2.626453647053628, "learning_rate": 2.0756956396699457e-06, "loss": 0.5579, "step": 22383 }, { "epoch": 0.7974775994442168, "grad_norm": 1.2341950477182153, "learning_rate": 2.074991834552661e-06, "loss": 0.3931, "step": 22384 }, { "epoch": 0.7975132265707109, "grad_norm": 1.404950188204915, "learning_rate": 2.0742881349624e-06, "loss": 0.4043, "step": 22385 }, { "epoch": 0.7975488536972051, "grad_norm": 1.3724137193496249, "learning_rate": 2.073584540908531e-06, "loss": 0.5246, "step": 22386 }, { "epoch": 0.7975844808236991, "grad_norm": 1.190447219557691, "learning_rate": 2.072881052400427e-06, "loss": 0.4308, "step": 22387 }, { "epoch": 0.7976201079501932, "grad_norm": 1.320675620340305, "learning_rate": 2.0721776694474503e-06, "loss": 0.4401, "step": 22388 }, { "epoch": 0.7976557350766874, "grad_norm": 1.1126702280776104, "learning_rate": 2.0714743920589707e-06, "loss": 0.3751, "step": 22389 }, { "epoch": 0.7976913622031815, "grad_norm": 2.1133016511377134, "learning_rate": 2.07077122024435e-06, "loss": 0.541, "step": 22390 }, { "epoch": 0.7977269893296756, "grad_norm": 1.2530559369033445, "learning_rate": 2.0700681540129517e-06, "loss": 0.4002, "step": 22391 }, { "epoch": 0.7977626164561697, "grad_norm": 1.2599263435313621, "learning_rate": 2.069365193374142e-06, "loss": 0.3842, "step": 22392 }, { "epoch": 0.7977982435826638, "grad_norm": 1.4162395691497747, "learning_rate": 2.0686623383372717e-06, "loss": 0.4098, "step": 22393 }, { "epoch": 0.7978338707091579, "grad_norm": 1.5187512986558749, "learning_rate": 2.0679595889117123e-06, "loss": 0.4295, "step": 22394 }, { "epoch": 0.7978694978356521, "grad_norm": 1.5656589694313163, "learning_rate": 2.0672569451068126e-06, "loss": 0.5125, "step": 22395 }, { "epoch": 0.7979051249621462, "grad_norm": 1.839650708393819, "learning_rate": 2.066554406931933e-06, "loss": 0.4691, "step": 22396 }, { "epoch": 0.7979407520886402, "grad_norm": 1.4542111910212208, "learning_rate": 2.0658519743964246e-06, "loss": 0.5051, "step": 22397 }, { "epoch": 0.7979763792151344, "grad_norm": 1.4809935206215539, "learning_rate": 2.0651496475096455e-06, "loss": 0.5048, "step": 22398 }, { "epoch": 0.7980120063416285, "grad_norm": 1.1649873825191284, "learning_rate": 2.064447426280942e-06, "loss": 0.43, "step": 22399 }, { "epoch": 0.7980476334681226, "grad_norm": 1.0949582041812957, "learning_rate": 2.063745310719668e-06, "loss": 0.4139, "step": 22400 }, { "epoch": 0.7980832605946168, "grad_norm": 1.6239409498109056, "learning_rate": 2.0630433008351723e-06, "loss": 0.4644, "step": 22401 }, { "epoch": 0.7981188877211108, "grad_norm": 1.2150459282990678, "learning_rate": 2.0623413966368023e-06, "loss": 0.4089, "step": 22402 }, { "epoch": 0.7981545148476049, "grad_norm": 1.1792435181015102, "learning_rate": 2.0616395981339076e-06, "loss": 0.4026, "step": 22403 }, { "epoch": 0.7981901419740991, "grad_norm": 1.407306313477246, "learning_rate": 2.0609379053358268e-06, "loss": 0.39, "step": 22404 }, { "epoch": 0.7982257691005932, "grad_norm": 1.7066930702340728, "learning_rate": 2.06023631825191e-06, "loss": 0.4951, "step": 22405 }, { "epoch": 0.7982613962270872, "grad_norm": 1.5432592691907172, "learning_rate": 2.0595348368914946e-06, "loss": 0.5466, "step": 22406 }, { "epoch": 0.7982970233535814, "grad_norm": 1.4260389084899057, "learning_rate": 2.0588334612639217e-06, "loss": 0.5012, "step": 22407 }, { "epoch": 0.7983326504800755, "grad_norm": 1.1561727262670476, "learning_rate": 2.0581321913785325e-06, "loss": 0.3951, "step": 22408 }, { "epoch": 0.7983682776065696, "grad_norm": 1.371879118037241, "learning_rate": 2.0574310272446663e-06, "loss": 0.508, "step": 22409 }, { "epoch": 0.7984039047330638, "grad_norm": 1.3363290111760049, "learning_rate": 2.0567299688716545e-06, "loss": 0.3788, "step": 22410 }, { "epoch": 0.7984395318595578, "grad_norm": 1.2433270079299377, "learning_rate": 2.056029016268839e-06, "loss": 0.3732, "step": 22411 }, { "epoch": 0.7984751589860519, "grad_norm": 1.5824617994425048, "learning_rate": 2.055328169445546e-06, "loss": 0.498, "step": 22412 }, { "epoch": 0.7985107861125461, "grad_norm": 1.4157340075846796, "learning_rate": 2.0546274284111123e-06, "loss": 0.3801, "step": 22413 }, { "epoch": 0.7985464132390402, "grad_norm": 1.3168143439846278, "learning_rate": 2.0539267931748696e-06, "loss": 0.4436, "step": 22414 }, { "epoch": 0.7985820403655343, "grad_norm": 2.3702790409612065, "learning_rate": 2.053226263746143e-06, "loss": 0.6122, "step": 22415 }, { "epoch": 0.7986176674920284, "grad_norm": 1.4608327667342882, "learning_rate": 2.052525840134264e-06, "loss": 0.5396, "step": 22416 }, { "epoch": 0.7986532946185225, "grad_norm": 1.1979389205719602, "learning_rate": 2.051825522348557e-06, "loss": 0.4269, "step": 22417 }, { "epoch": 0.7986889217450167, "grad_norm": 1.6500187637578105, "learning_rate": 2.051125310398353e-06, "loss": 0.5367, "step": 22418 }, { "epoch": 0.7987245488715108, "grad_norm": 1.3333608589006607, "learning_rate": 2.0504252042929686e-06, "loss": 0.4136, "step": 22419 }, { "epoch": 0.7987601759980049, "grad_norm": 1.5308607974646233, "learning_rate": 2.0497252040417303e-06, "loss": 0.529, "step": 22420 }, { "epoch": 0.798795803124499, "grad_norm": 1.6018048001192466, "learning_rate": 2.049025309653956e-06, "loss": 0.4417, "step": 22421 }, { "epoch": 0.7988314302509931, "grad_norm": 1.3145777940953522, "learning_rate": 2.0483255211389673e-06, "loss": 0.3926, "step": 22422 }, { "epoch": 0.7988670573774872, "grad_norm": 1.3986979938298, "learning_rate": 2.0476258385060823e-06, "loss": 0.485, "step": 22423 }, { "epoch": 0.7989026845039814, "grad_norm": 1.225999977022805, "learning_rate": 2.046926261764619e-06, "loss": 0.3865, "step": 22424 }, { "epoch": 0.7989383116304755, "grad_norm": 1.495962014576399, "learning_rate": 2.0462267909238897e-06, "loss": 0.5035, "step": 22425 }, { "epoch": 0.7989739387569695, "grad_norm": 1.456409411199196, "learning_rate": 2.04552742599321e-06, "loss": 0.5104, "step": 22426 }, { "epoch": 0.7990095658834637, "grad_norm": 1.370531403720448, "learning_rate": 2.0448281669818948e-06, "loss": 0.5766, "step": 22427 }, { "epoch": 0.7990451930099578, "grad_norm": 1.5642528560644549, "learning_rate": 2.044129013899251e-06, "loss": 0.586, "step": 22428 }, { "epoch": 0.7990808201364519, "grad_norm": 1.2388982555898256, "learning_rate": 2.043429966754592e-06, "loss": 0.42, "step": 22429 }, { "epoch": 0.7991164472629461, "grad_norm": 1.4029716413144164, "learning_rate": 2.042731025557222e-06, "loss": 0.4974, "step": 22430 }, { "epoch": 0.7991520743894401, "grad_norm": 1.1974181452593047, "learning_rate": 2.042032190316451e-06, "loss": 0.408, "step": 22431 }, { "epoch": 0.7991877015159342, "grad_norm": 1.6624914953039827, "learning_rate": 2.0413334610415825e-06, "loss": 0.5359, "step": 22432 }, { "epoch": 0.7992233286424284, "grad_norm": 1.384741306195173, "learning_rate": 2.040634837741925e-06, "loss": 0.4245, "step": 22433 }, { "epoch": 0.7992589557689225, "grad_norm": 1.2795037285659727, "learning_rate": 2.039936320426774e-06, "loss": 0.3778, "step": 22434 }, { "epoch": 0.7992945828954165, "grad_norm": 1.3258145347093337, "learning_rate": 2.039237909105439e-06, "loss": 0.5888, "step": 22435 }, { "epoch": 0.7993302100219107, "grad_norm": 1.5829330533156967, "learning_rate": 2.038539603787212e-06, "loss": 0.5046, "step": 22436 }, { "epoch": 0.7993658371484048, "grad_norm": 1.8008602648438, "learning_rate": 2.0378414044813943e-06, "loss": 0.5385, "step": 22437 }, { "epoch": 0.7994014642748989, "grad_norm": 1.3054434555035141, "learning_rate": 2.037143311197285e-06, "loss": 0.5394, "step": 22438 }, { "epoch": 0.7994370914013931, "grad_norm": 1.3723735487877438, "learning_rate": 2.0364453239441773e-06, "loss": 0.446, "step": 22439 }, { "epoch": 0.7994727185278871, "grad_norm": 1.2423956380748766, "learning_rate": 2.0357474427313695e-06, "loss": 0.4176, "step": 22440 }, { "epoch": 0.7995083456543812, "grad_norm": 1.1088671176255103, "learning_rate": 2.0350496675681476e-06, "loss": 0.3821, "step": 22441 }, { "epoch": 0.7995439727808754, "grad_norm": 1.5982498238972271, "learning_rate": 2.0343519984638104e-06, "loss": 0.4353, "step": 22442 }, { "epoch": 0.7995795999073695, "grad_norm": 1.1642134339331656, "learning_rate": 2.03365443542764e-06, "loss": 0.3642, "step": 22443 }, { "epoch": 0.7996152270338636, "grad_norm": 1.3651764086709715, "learning_rate": 2.032956978468933e-06, "loss": 0.604, "step": 22444 }, { "epoch": 0.7996508541603577, "grad_norm": 1.158578811459464, "learning_rate": 2.0322596275969666e-06, "loss": 0.4029, "step": 22445 }, { "epoch": 0.7996864812868518, "grad_norm": 1.4559921059352086, "learning_rate": 2.0315623828210374e-06, "loss": 0.5325, "step": 22446 }, { "epoch": 0.7997221084133459, "grad_norm": 1.69994473408646, "learning_rate": 2.0308652441504216e-06, "loss": 0.5387, "step": 22447 }, { "epoch": 0.7997577355398401, "grad_norm": 1.423235333123829, "learning_rate": 2.0301682115944054e-06, "loss": 0.3859, "step": 22448 }, { "epoch": 0.7997933626663342, "grad_norm": 1.2152346912649172, "learning_rate": 2.0294712851622734e-06, "loss": 0.4181, "step": 22449 }, { "epoch": 0.7998289897928282, "grad_norm": 1.2332987847196426, "learning_rate": 2.028774464863299e-06, "loss": 0.3889, "step": 22450 }, { "epoch": 0.7998646169193224, "grad_norm": 1.5140909194693892, "learning_rate": 2.0280777507067673e-06, "loss": 0.5292, "step": 22451 }, { "epoch": 0.7999002440458165, "grad_norm": 1.1611168197005617, "learning_rate": 2.0273811427019496e-06, "loss": 0.3952, "step": 22452 }, { "epoch": 0.7999358711723106, "grad_norm": 1.6270497513794078, "learning_rate": 2.026684640858124e-06, "loss": 0.468, "step": 22453 }, { "epoch": 0.7999714982988048, "grad_norm": 1.1044185422057773, "learning_rate": 2.0259882451845657e-06, "loss": 0.4105, "step": 22454 }, { "epoch": 0.8000071254252988, "grad_norm": 1.2196342707237786, "learning_rate": 2.0252919556905504e-06, "loss": 0.4273, "step": 22455 }, { "epoch": 0.8000427525517929, "grad_norm": 1.099849466939351, "learning_rate": 2.0245957723853436e-06, "loss": 0.4119, "step": 22456 }, { "epoch": 0.8000783796782871, "grad_norm": 1.5333968971802192, "learning_rate": 2.023899695278222e-06, "loss": 0.5402, "step": 22457 }, { "epoch": 0.8001140068047812, "grad_norm": 1.1973434319127914, "learning_rate": 2.0232037243784475e-06, "loss": 0.4054, "step": 22458 }, { "epoch": 0.8001496339312753, "grad_norm": 1.2800832491631082, "learning_rate": 2.0225078596952907e-06, "loss": 0.4675, "step": 22459 }, { "epoch": 0.8001852610577694, "grad_norm": 1.2744993124172082, "learning_rate": 2.021812101238019e-06, "loss": 0.371, "step": 22460 }, { "epoch": 0.8002208881842635, "grad_norm": 1.675846297860582, "learning_rate": 2.021116449015894e-06, "loss": 0.5526, "step": 22461 }, { "epoch": 0.8002565153107576, "grad_norm": 2.6094258288103194, "learning_rate": 2.0204209030381837e-06, "loss": 0.4659, "step": 22462 }, { "epoch": 0.8002921424372518, "grad_norm": 1.3161486322396705, "learning_rate": 2.019725463314144e-06, "loss": 0.4143, "step": 22463 }, { "epoch": 0.8003277695637458, "grad_norm": 1.2110838607263206, "learning_rate": 2.01903012985304e-06, "loss": 0.4009, "step": 22464 }, { "epoch": 0.8003633966902399, "grad_norm": 1.6296390805782803, "learning_rate": 2.018334902664125e-06, "loss": 0.5363, "step": 22465 }, { "epoch": 0.8003990238167341, "grad_norm": 1.4413994676491173, "learning_rate": 2.017639781756664e-06, "loss": 0.413, "step": 22466 }, { "epoch": 0.8004346509432282, "grad_norm": 1.7312831119653531, "learning_rate": 2.0169447671399044e-06, "loss": 0.4527, "step": 22467 }, { "epoch": 0.8004702780697223, "grad_norm": 1.3891708345299771, "learning_rate": 2.016249858823106e-06, "loss": 0.5411, "step": 22468 }, { "epoch": 0.8005059051962164, "grad_norm": 1.2025231629445954, "learning_rate": 2.01555505681552e-06, "loss": 0.4007, "step": 22469 }, { "epoch": 0.8005415323227105, "grad_norm": 1.134713932269933, "learning_rate": 2.014860361126403e-06, "loss": 0.3877, "step": 22470 }, { "epoch": 0.8005771594492046, "grad_norm": 1.2786235138138253, "learning_rate": 2.0141657717649975e-06, "loss": 0.3841, "step": 22471 }, { "epoch": 0.8006127865756988, "grad_norm": 1.7484189990182044, "learning_rate": 2.0134712887405583e-06, "loss": 0.487, "step": 22472 }, { "epoch": 0.8006484137021929, "grad_norm": 1.3025735988570601, "learning_rate": 2.012776912062332e-06, "loss": 0.4761, "step": 22473 }, { "epoch": 0.8006840408286869, "grad_norm": 2.251978052245117, "learning_rate": 2.012082641739562e-06, "loss": 0.4841, "step": 22474 }, { "epoch": 0.8007196679551811, "grad_norm": 1.5641160123367714, "learning_rate": 2.011388477781495e-06, "loss": 0.488, "step": 22475 }, { "epoch": 0.8007552950816752, "grad_norm": 1.6199013186129525, "learning_rate": 2.0106944201973745e-06, "loss": 0.5334, "step": 22476 }, { "epoch": 0.8007909222081693, "grad_norm": 1.4076469582333984, "learning_rate": 2.0100004689964434e-06, "loss": 0.4723, "step": 22477 }, { "epoch": 0.8008265493346635, "grad_norm": 1.6014150003620276, "learning_rate": 2.00930662418794e-06, "loss": 0.5944, "step": 22478 }, { "epoch": 0.8008621764611575, "grad_norm": 1.969622349622236, "learning_rate": 2.0086128857811047e-06, "loss": 0.5027, "step": 22479 }, { "epoch": 0.8008978035876516, "grad_norm": 1.5127542235220612, "learning_rate": 2.0079192537851734e-06, "loss": 0.5432, "step": 22480 }, { "epoch": 0.8009334307141458, "grad_norm": 1.3973702384963258, "learning_rate": 2.0072257282093855e-06, "loss": 0.4912, "step": 22481 }, { "epoch": 0.8009690578406399, "grad_norm": 1.2576970208294702, "learning_rate": 2.0065323090629675e-06, "loss": 0.4089, "step": 22482 }, { "epoch": 0.801004684967134, "grad_norm": 1.305020582585304, "learning_rate": 2.005838996355163e-06, "loss": 0.4222, "step": 22483 }, { "epoch": 0.8010403120936281, "grad_norm": 1.502856692307607, "learning_rate": 2.005145790095201e-06, "loss": 0.5294, "step": 22484 }, { "epoch": 0.8010759392201222, "grad_norm": 1.6091276761333158, "learning_rate": 2.004452690292309e-06, "loss": 0.4614, "step": 22485 }, { "epoch": 0.8011115663466163, "grad_norm": 1.4716833890528698, "learning_rate": 2.0037596969557205e-06, "loss": 0.5161, "step": 22486 }, { "epoch": 0.8011471934731105, "grad_norm": 1.565701300325099, "learning_rate": 2.003066810094658e-06, "loss": 0.6083, "step": 22487 }, { "epoch": 0.8011828205996046, "grad_norm": 1.1832185162292965, "learning_rate": 2.0023740297183536e-06, "loss": 0.419, "step": 22488 }, { "epoch": 0.8012184477260986, "grad_norm": 1.4922579984995716, "learning_rate": 2.001681355836026e-06, "loss": 0.5644, "step": 22489 }, { "epoch": 0.8012540748525928, "grad_norm": 1.3167594341968891, "learning_rate": 2.0009887884569014e-06, "loss": 0.4411, "step": 22490 }, { "epoch": 0.8012897019790869, "grad_norm": 1.5365711411972949, "learning_rate": 2.000296327590202e-06, "loss": 0.5382, "step": 22491 }, { "epoch": 0.801325329105581, "grad_norm": 1.4693692145437223, "learning_rate": 1.999603973245151e-06, "loss": 0.5148, "step": 22492 }, { "epoch": 0.8013609562320752, "grad_norm": 1.7941930135690505, "learning_rate": 1.9989117254309633e-06, "loss": 0.5335, "step": 22493 }, { "epoch": 0.8013965833585692, "grad_norm": 1.1494692386712666, "learning_rate": 1.9982195841568575e-06, "loss": 0.4068, "step": 22494 }, { "epoch": 0.8014322104850633, "grad_norm": 1.2789104258076234, "learning_rate": 1.9975275494320535e-06, "loss": 0.4498, "step": 22495 }, { "epoch": 0.8014678376115575, "grad_norm": 1.5896732098038797, "learning_rate": 1.9968356212657613e-06, "loss": 0.4459, "step": 22496 }, { "epoch": 0.8015034647380516, "grad_norm": 1.2191260703980122, "learning_rate": 1.9961437996671973e-06, "loss": 0.4328, "step": 22497 }, { "epoch": 0.8015390918645456, "grad_norm": 1.312083423001589, "learning_rate": 1.9954520846455717e-06, "loss": 0.5327, "step": 22498 }, { "epoch": 0.8015747189910398, "grad_norm": 1.7366609922591414, "learning_rate": 1.9947604762101003e-06, "loss": 0.5447, "step": 22499 }, { "epoch": 0.8016103461175339, "grad_norm": 1.4757674350925132, "learning_rate": 1.994068974369985e-06, "loss": 0.4847, "step": 22500 }, { "epoch": 0.801645973244028, "grad_norm": 1.5379752472834791, "learning_rate": 1.99337757913444e-06, "loss": 0.4967, "step": 22501 }, { "epoch": 0.8016816003705222, "grad_norm": 1.486146966635514, "learning_rate": 1.9926862905126663e-06, "loss": 0.5188, "step": 22502 }, { "epoch": 0.8017172274970162, "grad_norm": 1.2305486822440301, "learning_rate": 1.9919951085138735e-06, "loss": 0.3643, "step": 22503 }, { "epoch": 0.8017528546235103, "grad_norm": 1.3485442505484329, "learning_rate": 1.991304033147261e-06, "loss": 0.5089, "step": 22504 }, { "epoch": 0.8017884817500045, "grad_norm": 1.2354931054910991, "learning_rate": 1.9906130644220333e-06, "loss": 0.4082, "step": 22505 }, { "epoch": 0.8018241088764986, "grad_norm": 1.5800829460795018, "learning_rate": 1.9899222023473895e-06, "loss": 0.5377, "step": 22506 }, { "epoch": 0.8018597360029927, "grad_norm": 1.625403582899112, "learning_rate": 1.9892314469325315e-06, "loss": 0.4777, "step": 22507 }, { "epoch": 0.8018953631294868, "grad_norm": 1.6007783633752526, "learning_rate": 1.988540798186659e-06, "loss": 0.5255, "step": 22508 }, { "epoch": 0.8019309902559809, "grad_norm": 1.5967521682285177, "learning_rate": 1.9878502561189616e-06, "loss": 0.5223, "step": 22509 }, { "epoch": 0.801966617382475, "grad_norm": 1.1987953432612157, "learning_rate": 1.987159820738642e-06, "loss": 0.3962, "step": 22510 }, { "epoch": 0.8020022445089692, "grad_norm": 1.2901048394684649, "learning_rate": 1.9864694920548867e-06, "loss": 0.3708, "step": 22511 }, { "epoch": 0.8020378716354633, "grad_norm": 1.2387158664339666, "learning_rate": 1.9857792700768908e-06, "loss": 0.3799, "step": 22512 }, { "epoch": 0.8020734987619573, "grad_norm": 1.2374867565323755, "learning_rate": 1.9850891548138463e-06, "loss": 0.3954, "step": 22513 }, { "epoch": 0.8021091258884515, "grad_norm": 1.6497328208195796, "learning_rate": 1.9843991462749444e-06, "loss": 0.4746, "step": 22514 }, { "epoch": 0.8021447530149456, "grad_norm": 1.2735463356219976, "learning_rate": 1.9837092444693675e-06, "loss": 0.4255, "step": 22515 }, { "epoch": 0.8021803801414397, "grad_norm": 1.1331342022122333, "learning_rate": 1.983019449406308e-06, "loss": 0.3652, "step": 22516 }, { "epoch": 0.8022160072679339, "grad_norm": 1.4705311278010227, "learning_rate": 1.9823297610949455e-06, "loss": 0.4904, "step": 22517 }, { "epoch": 0.8022516343944279, "grad_norm": 1.4488301668829235, "learning_rate": 1.9816401795444664e-06, "loss": 0.479, "step": 22518 }, { "epoch": 0.802287261520922, "grad_norm": 1.370297798957977, "learning_rate": 1.9809507047640543e-06, "loss": 0.4174, "step": 22519 }, { "epoch": 0.8023228886474162, "grad_norm": 1.548956846821589, "learning_rate": 1.9802613367628843e-06, "loss": 0.5089, "step": 22520 }, { "epoch": 0.8023585157739103, "grad_norm": 1.1699782193938868, "learning_rate": 1.9795720755501446e-06, "loss": 0.3767, "step": 22521 }, { "epoch": 0.8023941429004043, "grad_norm": 1.1961666812900134, "learning_rate": 1.978882921135007e-06, "loss": 0.4046, "step": 22522 }, { "epoch": 0.8024297700268985, "grad_norm": 1.0909971269509764, "learning_rate": 1.9781938735266513e-06, "loss": 0.3807, "step": 22523 }, { "epoch": 0.8024653971533926, "grad_norm": 1.4697617029828156, "learning_rate": 1.9775049327342486e-06, "loss": 0.6127, "step": 22524 }, { "epoch": 0.8025010242798867, "grad_norm": 1.5070646871433913, "learning_rate": 1.976816098766978e-06, "loss": 0.5591, "step": 22525 }, { "epoch": 0.8025366514063809, "grad_norm": 1.7888375181749514, "learning_rate": 1.976127371634007e-06, "loss": 0.5082, "step": 22526 }, { "epoch": 0.8025722785328749, "grad_norm": 1.7112829431249965, "learning_rate": 1.9754387513445085e-06, "loss": 0.5003, "step": 22527 }, { "epoch": 0.802607905659369, "grad_norm": 1.276623767224989, "learning_rate": 1.97475023790765e-06, "loss": 0.3939, "step": 22528 }, { "epoch": 0.8026435327858632, "grad_norm": 1.867885889428164, "learning_rate": 1.974061831332603e-06, "loss": 0.4866, "step": 22529 }, { "epoch": 0.8026791599123573, "grad_norm": 1.0423901520646748, "learning_rate": 1.973373531628535e-06, "loss": 0.3593, "step": 22530 }, { "epoch": 0.8027147870388514, "grad_norm": 1.172107501683449, "learning_rate": 1.9726853388046052e-06, "loss": 0.4257, "step": 22531 }, { "epoch": 0.8027504141653455, "grad_norm": 1.228313287239546, "learning_rate": 1.9719972528699847e-06, "loss": 0.4139, "step": 22532 }, { "epoch": 0.8027860412918396, "grad_norm": 1.8037327395681229, "learning_rate": 1.971309273833828e-06, "loss": 0.568, "step": 22533 }, { "epoch": 0.8028216684183337, "grad_norm": 1.801783793199351, "learning_rate": 1.9706214017053017e-06, "loss": 0.6174, "step": 22534 }, { "epoch": 0.8028572955448279, "grad_norm": 1.3656237995087737, "learning_rate": 1.969933636493562e-06, "loss": 0.5414, "step": 22535 }, { "epoch": 0.802892922671322, "grad_norm": 1.2765641465453437, "learning_rate": 1.9692459782077724e-06, "loss": 0.4069, "step": 22536 }, { "epoch": 0.802928549797816, "grad_norm": 1.1811011970256171, "learning_rate": 1.9685584268570824e-06, "loss": 0.4029, "step": 22537 }, { "epoch": 0.8029641769243102, "grad_norm": 1.190142058860131, "learning_rate": 1.9678709824506535e-06, "loss": 0.427, "step": 22538 }, { "epoch": 0.8029998040508043, "grad_norm": 2.1665697641105224, "learning_rate": 1.9671836449976345e-06, "loss": 0.5221, "step": 22539 }, { "epoch": 0.8030354311772984, "grad_norm": 1.230766928701865, "learning_rate": 1.9664964145071784e-06, "loss": 0.4251, "step": 22540 }, { "epoch": 0.8030710583037926, "grad_norm": 1.547701083283876, "learning_rate": 1.965809290988442e-06, "loss": 0.5254, "step": 22541 }, { "epoch": 0.8031066854302866, "grad_norm": 1.1785269192241128, "learning_rate": 1.965122274450567e-06, "loss": 0.4374, "step": 22542 }, { "epoch": 0.8031423125567807, "grad_norm": 1.1829419836635542, "learning_rate": 1.9644353649027048e-06, "loss": 0.4264, "step": 22543 }, { "epoch": 0.8031779396832749, "grad_norm": 1.5137702135602347, "learning_rate": 1.963748562354002e-06, "loss": 0.4819, "step": 22544 }, { "epoch": 0.803213566809769, "grad_norm": 1.5974303100830327, "learning_rate": 1.963061866813608e-06, "loss": 0.5607, "step": 22545 }, { "epoch": 0.803249193936263, "grad_norm": 1.1680441421271597, "learning_rate": 1.9623752782906593e-06, "loss": 0.3659, "step": 22546 }, { "epoch": 0.8032848210627572, "grad_norm": 1.2501741917344158, "learning_rate": 1.9616887967943045e-06, "loss": 0.4995, "step": 22547 }, { "epoch": 0.8033204481892513, "grad_norm": 1.4541906454658415, "learning_rate": 1.961002422333679e-06, "loss": 0.4929, "step": 22548 }, { "epoch": 0.8033560753157454, "grad_norm": 1.2384419794943353, "learning_rate": 1.960316154917925e-06, "loss": 0.4042, "step": 22549 }, { "epoch": 0.8033917024422396, "grad_norm": 1.2263581136869655, "learning_rate": 1.9596299945561805e-06, "loss": 0.3645, "step": 22550 }, { "epoch": 0.8034273295687336, "grad_norm": 1.2718848746035045, "learning_rate": 1.958943941257586e-06, "loss": 0.4123, "step": 22551 }, { "epoch": 0.8034629566952277, "grad_norm": 1.4309600556782018, "learning_rate": 1.9582579950312696e-06, "loss": 0.5135, "step": 22552 }, { "epoch": 0.8034985838217219, "grad_norm": 2.0355048980315464, "learning_rate": 1.9575721558863704e-06, "loss": 0.3578, "step": 22553 }, { "epoch": 0.803534210948216, "grad_norm": 2.7798794432026233, "learning_rate": 1.956886423832021e-06, "loss": 0.4155, "step": 22554 }, { "epoch": 0.80356983807471, "grad_norm": 1.3206328513345735, "learning_rate": 1.956200798877348e-06, "loss": 0.416, "step": 22555 }, { "epoch": 0.8036054652012042, "grad_norm": 1.6488658983994635, "learning_rate": 1.955515281031486e-06, "loss": 0.499, "step": 22556 }, { "epoch": 0.8036410923276983, "grad_norm": 1.2431724921161496, "learning_rate": 1.954829870303555e-06, "loss": 0.4112, "step": 22557 }, { "epoch": 0.8036767194541924, "grad_norm": 1.3467728154582848, "learning_rate": 1.9541445667026936e-06, "loss": 0.4304, "step": 22558 }, { "epoch": 0.8037123465806866, "grad_norm": 1.1268802558832864, "learning_rate": 1.953459370238019e-06, "loss": 0.3831, "step": 22559 }, { "epoch": 0.8037479737071807, "grad_norm": 1.6119341262582423, "learning_rate": 1.952774280918659e-06, "loss": 0.4687, "step": 22560 }, { "epoch": 0.8037836008336747, "grad_norm": 1.3168799611981987, "learning_rate": 1.952089298753732e-06, "loss": 0.4348, "step": 22561 }, { "epoch": 0.8038192279601689, "grad_norm": 1.3074753286165426, "learning_rate": 1.951404423752363e-06, "loss": 0.4805, "step": 22562 }, { "epoch": 0.803854855086663, "grad_norm": 1.828585439950879, "learning_rate": 1.9507196559236675e-06, "loss": 0.4335, "step": 22563 }, { "epoch": 0.8038904822131571, "grad_norm": 1.1558152732160565, "learning_rate": 1.9500349952767663e-06, "loss": 0.4095, "step": 22564 }, { "epoch": 0.8039261093396513, "grad_norm": 1.6186663053372372, "learning_rate": 1.9493504418207754e-06, "loss": 0.4945, "step": 22565 }, { "epoch": 0.8039617364661453, "grad_norm": 1.3897596467239721, "learning_rate": 1.94866599556481e-06, "loss": 0.5895, "step": 22566 }, { "epoch": 0.8039973635926394, "grad_norm": 1.3473255210590604, "learning_rate": 1.9479816565179875e-06, "loss": 0.529, "step": 22567 }, { "epoch": 0.8040329907191336, "grad_norm": 1.084062670795034, "learning_rate": 1.947297424689414e-06, "loss": 0.3625, "step": 22568 }, { "epoch": 0.8040686178456277, "grad_norm": 1.5020994409928314, "learning_rate": 1.946613300088207e-06, "loss": 0.4961, "step": 22569 }, { "epoch": 0.8041042449721217, "grad_norm": 1.5887417388757241, "learning_rate": 1.94592928272347e-06, "loss": 0.5864, "step": 22570 }, { "epoch": 0.8041398720986159, "grad_norm": 1.4224057506718737, "learning_rate": 1.9452453726043143e-06, "loss": 0.4108, "step": 22571 }, { "epoch": 0.80417549922511, "grad_norm": 1.770336760473117, "learning_rate": 1.9445615697398467e-06, "loss": 0.4781, "step": 22572 }, { "epoch": 0.8042111263516041, "grad_norm": 1.4111188213675137, "learning_rate": 1.943877874139174e-06, "loss": 0.4297, "step": 22573 }, { "epoch": 0.8042467534780983, "grad_norm": 1.1473886924801588, "learning_rate": 1.943194285811396e-06, "loss": 0.3988, "step": 22574 }, { "epoch": 0.8042823806045923, "grad_norm": 1.4063477409328722, "learning_rate": 1.942510804765618e-06, "loss": 0.4659, "step": 22575 }, { "epoch": 0.8043180077310864, "grad_norm": 1.3846616941779706, "learning_rate": 1.941827431010943e-06, "loss": 0.494, "step": 22576 }, { "epoch": 0.8043536348575806, "grad_norm": 1.325171974792982, "learning_rate": 1.9411441645564655e-06, "loss": 0.4245, "step": 22577 }, { "epoch": 0.8043892619840747, "grad_norm": 1.348483863306, "learning_rate": 1.940461005411288e-06, "loss": 0.4461, "step": 22578 }, { "epoch": 0.8044248891105688, "grad_norm": 1.5370010241694887, "learning_rate": 1.939777953584503e-06, "loss": 0.4879, "step": 22579 }, { "epoch": 0.8044605162370629, "grad_norm": 1.1696319244089242, "learning_rate": 1.939095009085209e-06, "loss": 0.4487, "step": 22580 }, { "epoch": 0.804496143363557, "grad_norm": 1.690648855384772, "learning_rate": 1.938412171922499e-06, "loss": 0.4934, "step": 22581 }, { "epoch": 0.8045317704900511, "grad_norm": 1.2104336310072419, "learning_rate": 1.9377294421054683e-06, "loss": 0.4142, "step": 22582 }, { "epoch": 0.8045673976165453, "grad_norm": 1.610492842565051, "learning_rate": 1.9370468196432025e-06, "loss": 0.5174, "step": 22583 }, { "epoch": 0.8046030247430394, "grad_norm": 1.5368845930534987, "learning_rate": 1.9363643045447967e-06, "loss": 0.5016, "step": 22584 }, { "epoch": 0.8046386518695334, "grad_norm": 1.6061982746396148, "learning_rate": 1.9356818968193336e-06, "loss": 0.6106, "step": 22585 }, { "epoch": 0.8046742789960276, "grad_norm": 1.2944533769773836, "learning_rate": 1.934999596475903e-06, "loss": 0.3916, "step": 22586 }, { "epoch": 0.8047099061225217, "grad_norm": 1.1491704998089158, "learning_rate": 1.934317403523589e-06, "loss": 0.4051, "step": 22587 }, { "epoch": 0.8047455332490158, "grad_norm": 1.4185324142023161, "learning_rate": 1.9336353179714774e-06, "loss": 0.5035, "step": 22588 }, { "epoch": 0.80478116037551, "grad_norm": 1.2258569543621884, "learning_rate": 1.9329533398286517e-06, "loss": 0.3945, "step": 22589 }, { "epoch": 0.804816787502004, "grad_norm": 1.2123046478479869, "learning_rate": 1.932271469104188e-06, "loss": 0.4107, "step": 22590 }, { "epoch": 0.8048524146284981, "grad_norm": 1.3928421706254126, "learning_rate": 1.931589705807172e-06, "loss": 0.4408, "step": 22591 }, { "epoch": 0.8048880417549923, "grad_norm": 1.3297533025128097, "learning_rate": 1.930908049946675e-06, "loss": 0.3881, "step": 22592 }, { "epoch": 0.8049236688814864, "grad_norm": 1.2490590783224815, "learning_rate": 1.93022650153178e-06, "loss": 0.446, "step": 22593 }, { "epoch": 0.8049592960079804, "grad_norm": 1.1462835847140098, "learning_rate": 1.929545060571557e-06, "loss": 0.4345, "step": 22594 }, { "epoch": 0.8049949231344746, "grad_norm": 1.14834726697752, "learning_rate": 1.9288637270750832e-06, "loss": 0.3949, "step": 22595 }, { "epoch": 0.8050305502609687, "grad_norm": 1.1959793638764808, "learning_rate": 1.92818250105143e-06, "loss": 0.4186, "step": 22596 }, { "epoch": 0.8050661773874628, "grad_norm": 1.3156401035916692, "learning_rate": 1.927501382509671e-06, "loss": 0.4813, "step": 22597 }, { "epoch": 0.805101804513957, "grad_norm": 1.7438907510516202, "learning_rate": 1.9268203714588706e-06, "loss": 0.4749, "step": 22598 }, { "epoch": 0.805137431640451, "grad_norm": 1.203851343891167, "learning_rate": 1.9261394679081004e-06, "loss": 0.3901, "step": 22599 }, { "epoch": 0.8051730587669451, "grad_norm": 1.8263938035292218, "learning_rate": 1.9254586718664293e-06, "loss": 0.3828, "step": 22600 }, { "epoch": 0.8052086858934393, "grad_norm": 1.3686447573937681, "learning_rate": 1.924777983342916e-06, "loss": 0.4172, "step": 22601 }, { "epoch": 0.8052443130199334, "grad_norm": 1.1377757957484615, "learning_rate": 1.9240974023466284e-06, "loss": 0.3838, "step": 22602 }, { "epoch": 0.8052799401464275, "grad_norm": 1.2311712842238567, "learning_rate": 1.923416928886629e-06, "loss": 0.4139, "step": 22603 }, { "epoch": 0.8053155672729216, "grad_norm": 1.684745326797374, "learning_rate": 1.9227365629719806e-06, "loss": 0.5504, "step": 22604 }, { "epoch": 0.8053511943994157, "grad_norm": 1.1414709999266088, "learning_rate": 1.922056304611739e-06, "loss": 0.4233, "step": 22605 }, { "epoch": 0.8053868215259098, "grad_norm": 1.4946308983212315, "learning_rate": 1.9213761538149657e-06, "loss": 0.5336, "step": 22606 }, { "epoch": 0.805422448652404, "grad_norm": 1.3177768112454504, "learning_rate": 1.920696110590713e-06, "loss": 0.5127, "step": 22607 }, { "epoch": 0.805458075778898, "grad_norm": 1.412044378720387, "learning_rate": 1.9200161749480406e-06, "loss": 0.5246, "step": 22608 }, { "epoch": 0.8054937029053921, "grad_norm": 1.4371233263323295, "learning_rate": 1.9193363468959956e-06, "loss": 0.4198, "step": 22609 }, { "epoch": 0.8055293300318863, "grad_norm": 1.5884504668798456, "learning_rate": 1.918656626443638e-06, "loss": 0.4851, "step": 22610 }, { "epoch": 0.8055649571583804, "grad_norm": 1.4260019436836924, "learning_rate": 1.9179770136000187e-06, "loss": 0.5163, "step": 22611 }, { "epoch": 0.8056005842848745, "grad_norm": 1.1635490693330328, "learning_rate": 1.917297508374182e-06, "loss": 0.3724, "step": 22612 }, { "epoch": 0.8056362114113687, "grad_norm": 1.2548157085982774, "learning_rate": 1.9166181107751802e-06, "loss": 0.3979, "step": 22613 }, { "epoch": 0.8056718385378627, "grad_norm": 1.4877454424026022, "learning_rate": 1.9159388208120568e-06, "loss": 0.4669, "step": 22614 }, { "epoch": 0.8057074656643568, "grad_norm": 1.3668301514144212, "learning_rate": 1.915259638493859e-06, "loss": 0.456, "step": 22615 }, { "epoch": 0.805743092790851, "grad_norm": 1.2134501664091233, "learning_rate": 1.9145805638296287e-06, "loss": 0.4179, "step": 22616 }, { "epoch": 0.8057787199173451, "grad_norm": 1.780030271006896, "learning_rate": 1.9139015968284104e-06, "loss": 0.5325, "step": 22617 }, { "epoch": 0.8058143470438391, "grad_norm": 1.3295827243819764, "learning_rate": 1.913222737499243e-06, "loss": 0.4894, "step": 22618 }, { "epoch": 0.8058499741703333, "grad_norm": 1.283251461163335, "learning_rate": 1.912543985851171e-06, "loss": 0.3919, "step": 22619 }, { "epoch": 0.8058856012968274, "grad_norm": 1.8535044140596657, "learning_rate": 1.9118653418932244e-06, "loss": 0.5202, "step": 22620 }, { "epoch": 0.8059212284233215, "grad_norm": 1.260907384186037, "learning_rate": 1.911186805634445e-06, "loss": 0.4843, "step": 22621 }, { "epoch": 0.8059568555498157, "grad_norm": 1.2183097880695262, "learning_rate": 1.9105083770838694e-06, "loss": 0.368, "step": 22622 }, { "epoch": 0.8059924826763097, "grad_norm": 1.4346055722743356, "learning_rate": 1.9098300562505266e-06, "loss": 0.4276, "step": 22623 }, { "epoch": 0.8060281098028038, "grad_norm": 1.3923664790323773, "learning_rate": 1.9091518431434507e-06, "loss": 0.4658, "step": 22624 }, { "epoch": 0.806063736929298, "grad_norm": 1.2069715280332667, "learning_rate": 1.9084737377716732e-06, "loss": 0.4593, "step": 22625 }, { "epoch": 0.8060993640557921, "grad_norm": 1.6600084714660899, "learning_rate": 1.9077957401442272e-06, "loss": 0.527, "step": 22626 }, { "epoch": 0.8061349911822862, "grad_norm": 1.6961983248499475, "learning_rate": 1.9071178502701337e-06, "loss": 0.5148, "step": 22627 }, { "epoch": 0.8061706183087803, "grad_norm": 1.4060773787516931, "learning_rate": 1.906440068158425e-06, "loss": 0.4425, "step": 22628 }, { "epoch": 0.8062062454352744, "grad_norm": 1.6095839358814816, "learning_rate": 1.9057623938181213e-06, "loss": 0.4981, "step": 22629 }, { "epoch": 0.8062418725617685, "grad_norm": 1.5531072808442405, "learning_rate": 1.9050848272582522e-06, "loss": 0.5672, "step": 22630 }, { "epoch": 0.8062774996882627, "grad_norm": 1.1637148207899095, "learning_rate": 1.904407368487834e-06, "loss": 0.3487, "step": 22631 }, { "epoch": 0.8063131268147568, "grad_norm": 1.531687365140522, "learning_rate": 1.903730017515889e-06, "loss": 0.5982, "step": 22632 }, { "epoch": 0.8063487539412508, "grad_norm": 1.3978323305538363, "learning_rate": 1.9030527743514382e-06, "loss": 0.492, "step": 22633 }, { "epoch": 0.806384381067745, "grad_norm": 1.333909105104198, "learning_rate": 1.9023756390034998e-06, "loss": 0.399, "step": 22634 }, { "epoch": 0.8064200081942391, "grad_norm": 1.2277402979680276, "learning_rate": 1.901698611481092e-06, "loss": 0.4016, "step": 22635 }, { "epoch": 0.8064556353207332, "grad_norm": 1.5584634000294972, "learning_rate": 1.9010216917932244e-06, "loss": 0.531, "step": 22636 }, { "epoch": 0.8064912624472274, "grad_norm": 1.3671554861274853, "learning_rate": 1.9003448799489165e-06, "loss": 0.3996, "step": 22637 }, { "epoch": 0.8065268895737214, "grad_norm": 1.4576718102107264, "learning_rate": 1.8996681759571755e-06, "loss": 0.4782, "step": 22638 }, { "epoch": 0.8065625167002155, "grad_norm": 1.2595943772568345, "learning_rate": 1.8989915798270152e-06, "loss": 0.4019, "step": 22639 }, { "epoch": 0.8065981438267097, "grad_norm": 1.1281679387944592, "learning_rate": 1.8983150915674431e-06, "loss": 0.3849, "step": 22640 }, { "epoch": 0.8066337709532038, "grad_norm": 1.6628382451810486, "learning_rate": 1.8976387111874717e-06, "loss": 0.4625, "step": 22641 }, { "epoch": 0.8066693980796978, "grad_norm": 1.199500687222301, "learning_rate": 1.8969624386961015e-06, "loss": 0.4265, "step": 22642 }, { "epoch": 0.806705025206192, "grad_norm": 2.339810258872434, "learning_rate": 1.8962862741023425e-06, "loss": 0.4958, "step": 22643 }, { "epoch": 0.8067406523326861, "grad_norm": 1.3247612693668211, "learning_rate": 1.8956102174151935e-06, "loss": 0.4308, "step": 22644 }, { "epoch": 0.8067762794591802, "grad_norm": 1.6142852999024775, "learning_rate": 1.894934268643659e-06, "loss": 0.539, "step": 22645 }, { "epoch": 0.8068119065856744, "grad_norm": 1.8146320892872738, "learning_rate": 1.894258427796739e-06, "loss": 0.5088, "step": 22646 }, { "epoch": 0.8068475337121684, "grad_norm": 1.3116045945436283, "learning_rate": 1.8935826948834347e-06, "loss": 0.4116, "step": 22647 }, { "epoch": 0.8068831608386625, "grad_norm": 1.9274185904913783, "learning_rate": 1.892907069912745e-06, "loss": 0.5533, "step": 22648 }, { "epoch": 0.8069187879651567, "grad_norm": 1.5145909177616466, "learning_rate": 1.8922315528936619e-06, "loss": 0.4748, "step": 22649 }, { "epoch": 0.8069544150916508, "grad_norm": 1.2743159035219698, "learning_rate": 1.8915561438351847e-06, "loss": 0.4228, "step": 22650 }, { "epoch": 0.8069900422181449, "grad_norm": 1.1778211553275295, "learning_rate": 1.890880842746302e-06, "loss": 0.4073, "step": 22651 }, { "epoch": 0.807025669344639, "grad_norm": 1.444571023504508, "learning_rate": 1.8902056496360122e-06, "loss": 0.4779, "step": 22652 }, { "epoch": 0.8070612964711331, "grad_norm": 1.2530946741430113, "learning_rate": 1.8895305645133e-06, "loss": 0.4267, "step": 22653 }, { "epoch": 0.8070969235976272, "grad_norm": 1.379149743786249, "learning_rate": 1.8888555873871573e-06, "loss": 0.447, "step": 22654 }, { "epoch": 0.8071325507241214, "grad_norm": 1.564834283050134, "learning_rate": 1.8881807182665711e-06, "loss": 0.5229, "step": 22655 }, { "epoch": 0.8071681778506155, "grad_norm": 1.8797625883051048, "learning_rate": 1.8875059571605293e-06, "loss": 0.5465, "step": 22656 }, { "epoch": 0.8072038049771095, "grad_norm": 1.6657876528132627, "learning_rate": 1.8868313040780184e-06, "loss": 0.5178, "step": 22657 }, { "epoch": 0.8072394321036037, "grad_norm": 1.7028057514789319, "learning_rate": 1.886156759028016e-06, "loss": 0.4948, "step": 22658 }, { "epoch": 0.8072750592300978, "grad_norm": 1.7265448544885642, "learning_rate": 1.8854823220195106e-06, "loss": 0.5085, "step": 22659 }, { "epoch": 0.8073106863565919, "grad_norm": 1.278547325203254, "learning_rate": 1.8848079930614783e-06, "loss": 0.4119, "step": 22660 }, { "epoch": 0.807346313483086, "grad_norm": 1.1231872214266694, "learning_rate": 1.8841337721628982e-06, "loss": 0.3831, "step": 22661 }, { "epoch": 0.8073819406095801, "grad_norm": 1.1825184143449146, "learning_rate": 1.8834596593327503e-06, "loss": 0.3962, "step": 22662 }, { "epoch": 0.8074175677360742, "grad_norm": 1.2120319043591607, "learning_rate": 1.8827856545800126e-06, "loss": 0.4012, "step": 22663 }, { "epoch": 0.8074531948625684, "grad_norm": 1.4986188871383703, "learning_rate": 1.882111757913655e-06, "loss": 0.4249, "step": 22664 }, { "epoch": 0.8074888219890625, "grad_norm": 1.2013385713044034, "learning_rate": 1.8814379693426554e-06, "loss": 0.3974, "step": 22665 }, { "epoch": 0.8075244491155565, "grad_norm": 1.2193323945625836, "learning_rate": 1.8807642888759825e-06, "loss": 0.4182, "step": 22666 }, { "epoch": 0.8075600762420507, "grad_norm": 1.3481147610701456, "learning_rate": 1.8800907165226068e-06, "loss": 0.4016, "step": 22667 }, { "epoch": 0.8075957033685448, "grad_norm": 1.3833457903373279, "learning_rate": 1.8794172522915022e-06, "loss": 0.3814, "step": 22668 }, { "epoch": 0.8076313304950389, "grad_norm": 1.2668175384759974, "learning_rate": 1.8787438961916292e-06, "loss": 0.422, "step": 22669 }, { "epoch": 0.8076669576215331, "grad_norm": 1.2088188940063567, "learning_rate": 1.8780706482319588e-06, "loss": 0.3523, "step": 22670 }, { "epoch": 0.8077025847480271, "grad_norm": 1.234192714399172, "learning_rate": 1.8773975084214547e-06, "loss": 0.3604, "step": 22671 }, { "epoch": 0.8077382118745212, "grad_norm": 1.8138427348411377, "learning_rate": 1.8767244767690818e-06, "loss": 0.5789, "step": 22672 }, { "epoch": 0.8077738390010154, "grad_norm": 1.2049297674841941, "learning_rate": 1.8760515532837976e-06, "loss": 0.3773, "step": 22673 }, { "epoch": 0.8078094661275095, "grad_norm": 1.2318673270168086, "learning_rate": 1.8753787379745691e-06, "loss": 0.4329, "step": 22674 }, { "epoch": 0.8078450932540036, "grad_norm": 1.4250861189523463, "learning_rate": 1.8747060308503473e-06, "loss": 0.4722, "step": 22675 }, { "epoch": 0.8078807203804977, "grad_norm": 1.4016316437764162, "learning_rate": 1.8740334319200948e-06, "loss": 0.4725, "step": 22676 }, { "epoch": 0.8079163475069918, "grad_norm": 1.5507227989022567, "learning_rate": 1.8733609411927678e-06, "loss": 0.4664, "step": 22677 }, { "epoch": 0.8079519746334859, "grad_norm": 1.0676688996863009, "learning_rate": 1.8726885586773213e-06, "loss": 0.389, "step": 22678 }, { "epoch": 0.8079876017599801, "grad_norm": 1.1499262043423844, "learning_rate": 1.872016284382705e-06, "loss": 0.3987, "step": 22679 }, { "epoch": 0.8080232288864742, "grad_norm": 1.746106562300953, "learning_rate": 1.8713441183178728e-06, "loss": 0.4903, "step": 22680 }, { "epoch": 0.8080588560129682, "grad_norm": 1.659378282712023, "learning_rate": 1.8706720604917782e-06, "loss": 0.4471, "step": 22681 }, { "epoch": 0.8080944831394624, "grad_norm": 1.2575313597832323, "learning_rate": 1.8700001109133648e-06, "loss": 0.4336, "step": 22682 }, { "epoch": 0.8081301102659565, "grad_norm": 1.6396714864226625, "learning_rate": 1.869328269591585e-06, "loss": 0.3858, "step": 22683 }, { "epoch": 0.8081657373924506, "grad_norm": 1.2134701908092116, "learning_rate": 1.868656536535376e-06, "loss": 0.3895, "step": 22684 }, { "epoch": 0.8082013645189448, "grad_norm": 1.4012789000367394, "learning_rate": 1.8679849117536952e-06, "loss": 0.5053, "step": 22685 }, { "epoch": 0.8082369916454388, "grad_norm": 1.4518240084130116, "learning_rate": 1.8673133952554756e-06, "loss": 0.3748, "step": 22686 }, { "epoch": 0.8082726187719329, "grad_norm": 1.5476410013015423, "learning_rate": 1.8666419870496666e-06, "loss": 0.486, "step": 22687 }, { "epoch": 0.8083082458984271, "grad_norm": 1.2803567516996823, "learning_rate": 1.865970687145201e-06, "loss": 0.4446, "step": 22688 }, { "epoch": 0.8083438730249212, "grad_norm": 1.693818569753502, "learning_rate": 1.8652994955510228e-06, "loss": 0.5412, "step": 22689 }, { "epoch": 0.8083795001514152, "grad_norm": 1.3093043297808253, "learning_rate": 1.864628412276066e-06, "loss": 0.463, "step": 22690 }, { "epoch": 0.8084151272779094, "grad_norm": 1.1416985065714547, "learning_rate": 1.863957437329268e-06, "loss": 0.3716, "step": 22691 }, { "epoch": 0.8084507544044035, "grad_norm": 1.3676025866575554, "learning_rate": 1.8632865707195636e-06, "loss": 0.3852, "step": 22692 }, { "epoch": 0.8084863815308976, "grad_norm": 1.2121932595426212, "learning_rate": 1.8626158124558858e-06, "loss": 0.411, "step": 22693 }, { "epoch": 0.8085220086573918, "grad_norm": 1.22794783990135, "learning_rate": 1.8619451625471685e-06, "loss": 0.4139, "step": 22694 }, { "epoch": 0.8085576357838858, "grad_norm": 1.2519073487230448, "learning_rate": 1.861274621002337e-06, "loss": 0.4793, "step": 22695 }, { "epoch": 0.8085932629103799, "grad_norm": 1.206716000894195, "learning_rate": 1.8606041878303249e-06, "loss": 0.3345, "step": 22696 }, { "epoch": 0.8086288900368741, "grad_norm": 1.2843393724464056, "learning_rate": 1.8599338630400544e-06, "loss": 0.4023, "step": 22697 }, { "epoch": 0.8086645171633682, "grad_norm": 1.6132704521137549, "learning_rate": 1.8592636466404545e-06, "loss": 0.344, "step": 22698 }, { "epoch": 0.8087001442898623, "grad_norm": 1.757908546621938, "learning_rate": 1.8585935386404496e-06, "loss": 0.5743, "step": 22699 }, { "epoch": 0.8087357714163564, "grad_norm": 1.2867644949850252, "learning_rate": 1.8579235390489636e-06, "loss": 0.4837, "step": 22700 }, { "epoch": 0.8087713985428505, "grad_norm": 1.430351139190214, "learning_rate": 1.8572536478749148e-06, "loss": 0.3808, "step": 22701 }, { "epoch": 0.8088070256693446, "grad_norm": 1.268022792398949, "learning_rate": 1.856583865127226e-06, "loss": 0.4177, "step": 22702 }, { "epoch": 0.8088426527958388, "grad_norm": 1.2292275809996875, "learning_rate": 1.855914190814816e-06, "loss": 0.3917, "step": 22703 }, { "epoch": 0.8088782799223329, "grad_norm": 1.6834636697659953, "learning_rate": 1.8552446249465995e-06, "loss": 0.4862, "step": 22704 }, { "epoch": 0.8089139070488269, "grad_norm": 1.5137063557386117, "learning_rate": 1.854575167531496e-06, "loss": 0.3481, "step": 22705 }, { "epoch": 0.8089495341753211, "grad_norm": 1.1439106993870525, "learning_rate": 1.8539058185784143e-06, "loss": 0.3484, "step": 22706 }, { "epoch": 0.8089851613018152, "grad_norm": 1.4680679158717627, "learning_rate": 1.8532365780962702e-06, "loss": 0.4719, "step": 22707 }, { "epoch": 0.8090207884283093, "grad_norm": 1.1918005473548574, "learning_rate": 1.8525674460939757e-06, "loss": 0.3801, "step": 22708 }, { "epoch": 0.8090564155548035, "grad_norm": 1.668566269736329, "learning_rate": 1.8518984225804437e-06, "loss": 0.5434, "step": 22709 }, { "epoch": 0.8090920426812975, "grad_norm": 1.0931135859389696, "learning_rate": 1.8512295075645758e-06, "loss": 0.395, "step": 22710 }, { "epoch": 0.8091276698077916, "grad_norm": 1.1886579761814764, "learning_rate": 1.8505607010552862e-06, "loss": 0.3771, "step": 22711 }, { "epoch": 0.8091632969342858, "grad_norm": 1.2759942581978678, "learning_rate": 1.8498920030614732e-06, "loss": 0.4239, "step": 22712 }, { "epoch": 0.8091989240607799, "grad_norm": 1.0904503408676178, "learning_rate": 1.849223413592046e-06, "loss": 0.3519, "step": 22713 }, { "epoch": 0.8092345511872739, "grad_norm": 1.6500667255082988, "learning_rate": 1.848554932655905e-06, "loss": 0.4962, "step": 22714 }, { "epoch": 0.8092701783137681, "grad_norm": 1.685260424645163, "learning_rate": 1.8478865602619533e-06, "loss": 0.5068, "step": 22715 }, { "epoch": 0.8093058054402622, "grad_norm": 1.2042214974804069, "learning_rate": 1.8472182964190922e-06, "loss": 0.4118, "step": 22716 }, { "epoch": 0.8093414325667563, "grad_norm": 1.4775075458928812, "learning_rate": 1.8465501411362153e-06, "loss": 0.4533, "step": 22717 }, { "epoch": 0.8093770596932505, "grad_norm": 1.2188303310795012, "learning_rate": 1.8458820944222256e-06, "loss": 0.419, "step": 22718 }, { "epoch": 0.8094126868197445, "grad_norm": 1.2301671443082292, "learning_rate": 1.8452141562860115e-06, "loss": 0.4341, "step": 22719 }, { "epoch": 0.8094483139462386, "grad_norm": 1.824836421619771, "learning_rate": 1.8445463267364738e-06, "loss": 0.5817, "step": 22720 }, { "epoch": 0.8094839410727328, "grad_norm": 1.2777396285890181, "learning_rate": 1.8438786057824999e-06, "loss": 0.4148, "step": 22721 }, { "epoch": 0.8095195681992269, "grad_norm": 1.6190038354166825, "learning_rate": 1.8432109934329833e-06, "loss": 0.4845, "step": 22722 }, { "epoch": 0.809555195325721, "grad_norm": 1.3394703180586953, "learning_rate": 1.8425434896968132e-06, "loss": 0.4185, "step": 22723 }, { "epoch": 0.8095908224522151, "grad_norm": 1.3297237269294537, "learning_rate": 1.8418760945828806e-06, "loss": 0.3868, "step": 22724 }, { "epoch": 0.8096264495787092, "grad_norm": 1.2676453545093933, "learning_rate": 1.8412088081000668e-06, "loss": 0.4364, "step": 22725 }, { "epoch": 0.8096620767052033, "grad_norm": 1.4083010967138652, "learning_rate": 1.8405416302572621e-06, "loss": 0.6047, "step": 22726 }, { "epoch": 0.8096977038316975, "grad_norm": 1.0685261714910523, "learning_rate": 1.8398745610633495e-06, "loss": 0.3522, "step": 22727 }, { "epoch": 0.8097333309581916, "grad_norm": 1.2022748298079453, "learning_rate": 1.8392076005272098e-06, "loss": 0.4181, "step": 22728 }, { "epoch": 0.8097689580846856, "grad_norm": 1.3636632760543475, "learning_rate": 1.8385407486577233e-06, "loss": 0.5335, "step": 22729 }, { "epoch": 0.8098045852111798, "grad_norm": 1.1106198882498879, "learning_rate": 1.8378740054637722e-06, "loss": 0.4152, "step": 22730 }, { "epoch": 0.8098402123376739, "grad_norm": 1.160012540704274, "learning_rate": 1.837207370954236e-06, "loss": 0.4085, "step": 22731 }, { "epoch": 0.809875839464168, "grad_norm": 1.8393012247741944, "learning_rate": 1.8365408451379873e-06, "loss": 0.5393, "step": 22732 }, { "epoch": 0.8099114665906622, "grad_norm": 1.8352520953005396, "learning_rate": 1.8358744280239048e-06, "loss": 0.5064, "step": 22733 }, { "epoch": 0.8099470937171562, "grad_norm": 1.3754047325164618, "learning_rate": 1.8352081196208581e-06, "loss": 0.4302, "step": 22734 }, { "epoch": 0.8099827208436503, "grad_norm": 1.3114572242548592, "learning_rate": 1.8345419199377223e-06, "loss": 0.4449, "step": 22735 }, { "epoch": 0.8100183479701445, "grad_norm": 1.2704147456813972, "learning_rate": 1.8338758289833692e-06, "loss": 0.3653, "step": 22736 }, { "epoch": 0.8100539750966386, "grad_norm": 1.1585181601745507, "learning_rate": 1.8332098467666658e-06, "loss": 0.4424, "step": 22737 }, { "epoch": 0.8100896022231326, "grad_norm": 1.111584301904645, "learning_rate": 1.8325439732964844e-06, "loss": 0.3653, "step": 22738 }, { "epoch": 0.8101252293496268, "grad_norm": 1.5321275889167574, "learning_rate": 1.8318782085816867e-06, "loss": 0.5204, "step": 22739 }, { "epoch": 0.8101608564761209, "grad_norm": 1.193655890928733, "learning_rate": 1.831212552631142e-06, "loss": 0.4079, "step": 22740 }, { "epoch": 0.810196483602615, "grad_norm": 1.4000400689548105, "learning_rate": 1.830547005453709e-06, "loss": 0.5367, "step": 22741 }, { "epoch": 0.8102321107291092, "grad_norm": 1.2839023601430952, "learning_rate": 1.829881567058256e-06, "loss": 0.4451, "step": 22742 }, { "epoch": 0.8102677378556032, "grad_norm": 1.171078565113334, "learning_rate": 1.829216237453637e-06, "loss": 0.419, "step": 22743 }, { "epoch": 0.8103033649820973, "grad_norm": 1.2748452655752185, "learning_rate": 1.8285510166487154e-06, "loss": 0.3554, "step": 22744 }, { "epoch": 0.8103389921085915, "grad_norm": 1.8819717893913936, "learning_rate": 1.827885904652349e-06, "loss": 0.438, "step": 22745 }, { "epoch": 0.8103746192350856, "grad_norm": 1.639351881462556, "learning_rate": 1.8272209014733955e-06, "loss": 0.5356, "step": 22746 }, { "epoch": 0.8104102463615797, "grad_norm": 1.198304405755652, "learning_rate": 1.8265560071207056e-06, "loss": 0.3846, "step": 22747 }, { "epoch": 0.8104458734880738, "grad_norm": 1.1272147690053698, "learning_rate": 1.8258912216031355e-06, "loss": 0.4278, "step": 22748 }, { "epoch": 0.8104815006145679, "grad_norm": 1.6839291343462617, "learning_rate": 1.8252265449295404e-06, "loss": 0.5379, "step": 22749 }, { "epoch": 0.810517127741062, "grad_norm": 1.1358766586677278, "learning_rate": 1.8245619771087642e-06, "loss": 0.4055, "step": 22750 }, { "epoch": 0.8105527548675562, "grad_norm": 1.2697487479027887, "learning_rate": 1.8238975181496598e-06, "loss": 0.4111, "step": 22751 }, { "epoch": 0.8105883819940503, "grad_norm": 1.5904786407813043, "learning_rate": 1.8232331680610749e-06, "loss": 0.4775, "step": 22752 }, { "epoch": 0.8106240091205443, "grad_norm": 1.1161448714344135, "learning_rate": 1.8225689268518575e-06, "loss": 0.3866, "step": 22753 }, { "epoch": 0.8106596362470385, "grad_norm": 1.2014824599445564, "learning_rate": 1.8219047945308488e-06, "loss": 0.4272, "step": 22754 }, { "epoch": 0.8106952633735326, "grad_norm": 1.7209915638657485, "learning_rate": 1.8212407711068958e-06, "loss": 0.5528, "step": 22755 }, { "epoch": 0.8107308905000267, "grad_norm": 1.4180482576121671, "learning_rate": 1.820576856588837e-06, "loss": 0.4198, "step": 22756 }, { "epoch": 0.8107665176265209, "grad_norm": 1.201630940297046, "learning_rate": 1.8199130509855157e-06, "loss": 0.4159, "step": 22757 }, { "epoch": 0.8108021447530149, "grad_norm": 1.3211229202618202, "learning_rate": 1.8192493543057676e-06, "loss": 0.4154, "step": 22758 }, { "epoch": 0.810837771879509, "grad_norm": 1.5564533294613785, "learning_rate": 1.8185857665584328e-06, "loss": 0.5474, "step": 22759 }, { "epoch": 0.8108733990060032, "grad_norm": 1.3511013868824775, "learning_rate": 1.8179222877523472e-06, "loss": 0.4145, "step": 22760 }, { "epoch": 0.8109090261324973, "grad_norm": 1.1569254669691622, "learning_rate": 1.817258917896344e-06, "loss": 0.3696, "step": 22761 }, { "epoch": 0.8109446532589913, "grad_norm": 1.5068521937950863, "learning_rate": 1.8165956569992615e-06, "loss": 0.5764, "step": 22762 }, { "epoch": 0.8109802803854855, "grad_norm": 1.3481822256924219, "learning_rate": 1.815932505069925e-06, "loss": 0.43, "step": 22763 }, { "epoch": 0.8110159075119796, "grad_norm": 1.251173475989965, "learning_rate": 1.8152694621171696e-06, "loss": 0.4207, "step": 22764 }, { "epoch": 0.8110515346384737, "grad_norm": 1.4013735653414021, "learning_rate": 1.8146065281498193e-06, "loss": 0.4006, "step": 22765 }, { "epoch": 0.8110871617649679, "grad_norm": 1.5065328601549333, "learning_rate": 1.8139437031767048e-06, "loss": 0.5223, "step": 22766 }, { "epoch": 0.8111227888914619, "grad_norm": 1.693483151644578, "learning_rate": 1.8132809872066515e-06, "loss": 0.5181, "step": 22767 }, { "epoch": 0.811158416017956, "grad_norm": 1.2988002643431806, "learning_rate": 1.8126183802484864e-06, "loss": 0.407, "step": 22768 }, { "epoch": 0.8111940431444502, "grad_norm": 1.118518245362807, "learning_rate": 1.8119558823110284e-06, "loss": 0.3636, "step": 22769 }, { "epoch": 0.8112296702709443, "grad_norm": 1.189167068492457, "learning_rate": 1.8112934934031023e-06, "loss": 0.3849, "step": 22770 }, { "epoch": 0.8112652973974384, "grad_norm": 1.5585448810600109, "learning_rate": 1.8106312135335246e-06, "loss": 0.5121, "step": 22771 }, { "epoch": 0.8113009245239325, "grad_norm": 1.705907234358232, "learning_rate": 1.8099690427111172e-06, "loss": 0.5949, "step": 22772 }, { "epoch": 0.8113365516504266, "grad_norm": 1.6829442324958617, "learning_rate": 1.8093069809446962e-06, "loss": 0.4694, "step": 22773 }, { "epoch": 0.8113721787769207, "grad_norm": 1.4807541993461732, "learning_rate": 1.8086450282430766e-06, "loss": 0.5489, "step": 22774 }, { "epoch": 0.8114078059034149, "grad_norm": 1.348369091317856, "learning_rate": 1.8079831846150775e-06, "loss": 0.4283, "step": 22775 }, { "epoch": 0.811443433029909, "grad_norm": 1.2242124033497481, "learning_rate": 1.8073214500695058e-06, "loss": 0.4432, "step": 22776 }, { "epoch": 0.811479060156403, "grad_norm": 1.6899370250066212, "learning_rate": 1.8066598246151767e-06, "loss": 0.4812, "step": 22777 }, { "epoch": 0.8115146872828972, "grad_norm": 1.3477726509784147, "learning_rate": 1.8059983082608977e-06, "loss": 0.4442, "step": 22778 }, { "epoch": 0.8115503144093913, "grad_norm": 1.6636964316336589, "learning_rate": 1.8053369010154808e-06, "loss": 0.5334, "step": 22779 }, { "epoch": 0.8115859415358854, "grad_norm": 1.4426581207237923, "learning_rate": 1.8046756028877277e-06, "loss": 0.5208, "step": 22780 }, { "epoch": 0.8116215686623796, "grad_norm": 1.7593364952761745, "learning_rate": 1.8040144138864468e-06, "loss": 0.4979, "step": 22781 }, { "epoch": 0.8116571957888736, "grad_norm": 1.5525940130300573, "learning_rate": 1.8033533340204434e-06, "loss": 0.5243, "step": 22782 }, { "epoch": 0.8116928229153677, "grad_norm": 1.683822921798738, "learning_rate": 1.8026923632985195e-06, "loss": 0.4858, "step": 22783 }, { "epoch": 0.8117284500418619, "grad_norm": 1.2960801144444554, "learning_rate": 1.8020315017294776e-06, "loss": 0.3832, "step": 22784 }, { "epoch": 0.811764077168356, "grad_norm": 1.3857261347007068, "learning_rate": 1.8013707493221155e-06, "loss": 0.4394, "step": 22785 }, { "epoch": 0.81179970429485, "grad_norm": 1.5517437623605368, "learning_rate": 1.800710106085234e-06, "loss": 0.5112, "step": 22786 }, { "epoch": 0.8118353314213442, "grad_norm": 1.9013871400509461, "learning_rate": 1.8000495720276257e-06, "loss": 0.5554, "step": 22787 }, { "epoch": 0.8118709585478383, "grad_norm": 1.4156490858964987, "learning_rate": 1.7993891471580894e-06, "loss": 0.3977, "step": 22788 }, { "epoch": 0.8119065856743324, "grad_norm": 1.2035442027788206, "learning_rate": 1.798728831485418e-06, "loss": 0.4137, "step": 22789 }, { "epoch": 0.8119422128008266, "grad_norm": 1.1970825561900065, "learning_rate": 1.7980686250184077e-06, "loss": 0.4215, "step": 22790 }, { "epoch": 0.8119778399273206, "grad_norm": 1.3674558060101194, "learning_rate": 1.7974085277658426e-06, "loss": 0.4114, "step": 22791 }, { "epoch": 0.8120134670538147, "grad_norm": 1.367270748774574, "learning_rate": 1.7967485397365202e-06, "loss": 0.3625, "step": 22792 }, { "epoch": 0.8120490941803089, "grad_norm": 1.7595403741686528, "learning_rate": 1.7960886609392215e-06, "loss": 0.4892, "step": 22793 }, { "epoch": 0.812084721306803, "grad_norm": 1.910702434779764, "learning_rate": 1.7954288913827366e-06, "loss": 0.4747, "step": 22794 }, { "epoch": 0.8121203484332972, "grad_norm": 1.2561682202349935, "learning_rate": 1.7947692310758535e-06, "loss": 0.4082, "step": 22795 }, { "epoch": 0.8121559755597912, "grad_norm": 1.9576470513012765, "learning_rate": 1.7941096800273495e-06, "loss": 0.5356, "step": 22796 }, { "epoch": 0.8121916026862853, "grad_norm": 1.4652878490187662, "learning_rate": 1.7934502382460118e-06, "loss": 0.478, "step": 22797 }, { "epoch": 0.8122272298127795, "grad_norm": 1.611547025671158, "learning_rate": 1.7927909057406201e-06, "loss": 0.3996, "step": 22798 }, { "epoch": 0.8122628569392736, "grad_norm": 1.3651214941865564, "learning_rate": 1.7921316825199564e-06, "loss": 0.393, "step": 22799 }, { "epoch": 0.8122984840657677, "grad_norm": 1.2489800566996236, "learning_rate": 1.7914725685927936e-06, "loss": 0.4257, "step": 22800 }, { "epoch": 0.8123341111922618, "grad_norm": 2.225142882256766, "learning_rate": 1.7908135639679137e-06, "loss": 0.5785, "step": 22801 }, { "epoch": 0.8123697383187559, "grad_norm": 1.5483599970816606, "learning_rate": 1.7901546686540872e-06, "loss": 0.5338, "step": 22802 }, { "epoch": 0.81240536544525, "grad_norm": 1.2100345197919928, "learning_rate": 1.7894958826600884e-06, "loss": 0.38, "step": 22803 }, { "epoch": 0.8124409925717442, "grad_norm": 1.4946890713161511, "learning_rate": 1.7888372059946912e-06, "loss": 0.4738, "step": 22804 }, { "epoch": 0.8124766196982383, "grad_norm": 1.5152748260968933, "learning_rate": 1.7881786386666688e-06, "loss": 0.5101, "step": 22805 }, { "epoch": 0.8125122468247323, "grad_norm": 1.2437380206149289, "learning_rate": 1.7875201806847842e-06, "loss": 0.4207, "step": 22806 }, { "epoch": 0.8125478739512265, "grad_norm": 1.495302876541795, "learning_rate": 1.786861832057809e-06, "loss": 0.3936, "step": 22807 }, { "epoch": 0.8125835010777206, "grad_norm": 1.1926293490322308, "learning_rate": 1.786203592794512e-06, "loss": 0.4173, "step": 22808 }, { "epoch": 0.8126191282042147, "grad_norm": 1.2566741874529967, "learning_rate": 1.7855454629036516e-06, "loss": 0.3575, "step": 22809 }, { "epoch": 0.8126547553307089, "grad_norm": 1.1866897087772736, "learning_rate": 1.7848874423939966e-06, "loss": 0.4533, "step": 22810 }, { "epoch": 0.8126903824572029, "grad_norm": 1.686353046682948, "learning_rate": 1.7842295312743064e-06, "loss": 0.5161, "step": 22811 }, { "epoch": 0.812726009583697, "grad_norm": 1.3182547654448273, "learning_rate": 1.783571729553345e-06, "loss": 0.3764, "step": 22812 }, { "epoch": 0.8127616367101912, "grad_norm": 1.4446563812946815, "learning_rate": 1.7829140372398667e-06, "loss": 0.4072, "step": 22813 }, { "epoch": 0.8127972638366853, "grad_norm": 1.526668033457118, "learning_rate": 1.7822564543426336e-06, "loss": 0.5023, "step": 22814 }, { "epoch": 0.8128328909631793, "grad_norm": 1.2206498090672586, "learning_rate": 1.781598980870397e-06, "loss": 0.4175, "step": 22815 }, { "epoch": 0.8128685180896735, "grad_norm": 1.3537516835019434, "learning_rate": 1.7809416168319172e-06, "loss": 0.4322, "step": 22816 }, { "epoch": 0.8129041452161676, "grad_norm": 1.2231213306869486, "learning_rate": 1.7802843622359422e-06, "loss": 0.4179, "step": 22817 }, { "epoch": 0.8129397723426617, "grad_norm": 1.417847889410005, "learning_rate": 1.7796272170912255e-06, "loss": 0.4853, "step": 22818 }, { "epoch": 0.8129753994691559, "grad_norm": 1.4329568997569453, "learning_rate": 1.7789701814065186e-06, "loss": 0.5592, "step": 22819 }, { "epoch": 0.8130110265956499, "grad_norm": 1.2422717473754232, "learning_rate": 1.7783132551905702e-06, "loss": 0.4356, "step": 22820 }, { "epoch": 0.813046653722144, "grad_norm": 2.343557624633204, "learning_rate": 1.777656438452129e-06, "loss": 0.5607, "step": 22821 }, { "epoch": 0.8130822808486382, "grad_norm": 1.3160818772074812, "learning_rate": 1.7769997311999375e-06, "loss": 0.4657, "step": 22822 }, { "epoch": 0.8131179079751323, "grad_norm": 1.8590007775111024, "learning_rate": 1.7763431334427438e-06, "loss": 0.5206, "step": 22823 }, { "epoch": 0.8131535351016264, "grad_norm": 1.5240450766511215, "learning_rate": 1.7756866451892885e-06, "loss": 0.4612, "step": 22824 }, { "epoch": 0.8131891622281205, "grad_norm": 1.2156554887362154, "learning_rate": 1.7750302664483121e-06, "loss": 0.378, "step": 22825 }, { "epoch": 0.8132247893546146, "grad_norm": 1.5892300285079806, "learning_rate": 1.7743739972285578e-06, "loss": 0.6689, "step": 22826 }, { "epoch": 0.8132604164811087, "grad_norm": 1.865314911438003, "learning_rate": 1.7737178375387653e-06, "loss": 0.5411, "step": 22827 }, { "epoch": 0.8132960436076029, "grad_norm": 1.6240178334608133, "learning_rate": 1.7730617873876666e-06, "loss": 0.5108, "step": 22828 }, { "epoch": 0.813331670734097, "grad_norm": 1.3466053295038551, "learning_rate": 1.772405846784001e-06, "loss": 0.4337, "step": 22829 }, { "epoch": 0.813367297860591, "grad_norm": 1.3824353073235096, "learning_rate": 1.7717500157365053e-06, "loss": 0.4616, "step": 22830 }, { "epoch": 0.8134029249870852, "grad_norm": 1.318360340849355, "learning_rate": 1.7710942942539055e-06, "loss": 0.3758, "step": 22831 }, { "epoch": 0.8134385521135793, "grad_norm": 1.4853221350025887, "learning_rate": 1.7704386823449404e-06, "loss": 0.405, "step": 22832 }, { "epoch": 0.8134741792400734, "grad_norm": 1.451436466092532, "learning_rate": 1.769783180018334e-06, "loss": 0.5303, "step": 22833 }, { "epoch": 0.8135098063665676, "grad_norm": 1.3550766379348, "learning_rate": 1.769127787282816e-06, "loss": 0.5105, "step": 22834 }, { "epoch": 0.8135454334930616, "grad_norm": 1.165502757911356, "learning_rate": 1.7684725041471162e-06, "loss": 0.3806, "step": 22835 }, { "epoch": 0.8135810606195557, "grad_norm": 1.1448286658493705, "learning_rate": 1.7678173306199597e-06, "loss": 0.4007, "step": 22836 }, { "epoch": 0.8136166877460499, "grad_norm": 1.6725062929313799, "learning_rate": 1.7671622667100674e-06, "loss": 0.5302, "step": 22837 }, { "epoch": 0.813652314872544, "grad_norm": 1.284700544773978, "learning_rate": 1.7665073124261666e-06, "loss": 0.5216, "step": 22838 }, { "epoch": 0.813687941999038, "grad_norm": 1.4372557656595066, "learning_rate": 1.765852467776973e-06, "loss": 0.5148, "step": 22839 }, { "epoch": 0.8137235691255322, "grad_norm": 1.268806503352803, "learning_rate": 1.765197732771209e-06, "loss": 0.397, "step": 22840 }, { "epoch": 0.8137591962520263, "grad_norm": 1.2912451587316611, "learning_rate": 1.7645431074175923e-06, "loss": 0.4165, "step": 22841 }, { "epoch": 0.8137948233785204, "grad_norm": 1.5745567256659052, "learning_rate": 1.7638885917248416e-06, "loss": 0.53, "step": 22842 }, { "epoch": 0.8138304505050146, "grad_norm": 1.61701806283258, "learning_rate": 1.7632341857016733e-06, "loss": 0.5057, "step": 22843 }, { "epoch": 0.8138660776315086, "grad_norm": 1.1703144668936092, "learning_rate": 1.7625798893567958e-06, "loss": 0.391, "step": 22844 }, { "epoch": 0.8139017047580027, "grad_norm": 1.5148253984068238, "learning_rate": 1.7619257026989278e-06, "loss": 0.528, "step": 22845 }, { "epoch": 0.8139373318844969, "grad_norm": 1.492965477809129, "learning_rate": 1.7612716257367745e-06, "loss": 0.4999, "step": 22846 }, { "epoch": 0.813972959010991, "grad_norm": 1.420330808387517, "learning_rate": 1.76061765847905e-06, "loss": 0.4775, "step": 22847 }, { "epoch": 0.8140085861374851, "grad_norm": 1.2326180565493692, "learning_rate": 1.7599638009344566e-06, "loss": 0.4222, "step": 22848 }, { "epoch": 0.8140442132639792, "grad_norm": 1.159082974892221, "learning_rate": 1.759310053111709e-06, "loss": 0.4522, "step": 22849 }, { "epoch": 0.8140798403904733, "grad_norm": 1.5031652714357684, "learning_rate": 1.7586564150195052e-06, "loss": 0.5031, "step": 22850 }, { "epoch": 0.8141154675169674, "grad_norm": 1.686913173605172, "learning_rate": 1.7580028866665544e-06, "loss": 0.5261, "step": 22851 }, { "epoch": 0.8141510946434616, "grad_norm": 1.321257799929563, "learning_rate": 1.7573494680615532e-06, "loss": 0.4144, "step": 22852 }, { "epoch": 0.8141867217699557, "grad_norm": 1.2231205490151495, "learning_rate": 1.7566961592132047e-06, "loss": 0.3782, "step": 22853 }, { "epoch": 0.8142223488964497, "grad_norm": 1.2239659831214762, "learning_rate": 1.756042960130212e-06, "loss": 0.4453, "step": 22854 }, { "epoch": 0.8142579760229439, "grad_norm": 1.6561450915453326, "learning_rate": 1.7553898708212658e-06, "loss": 0.5402, "step": 22855 }, { "epoch": 0.814293603149438, "grad_norm": 1.2389789844527235, "learning_rate": 1.7547368912950669e-06, "loss": 0.3817, "step": 22856 }, { "epoch": 0.8143292302759321, "grad_norm": 1.4643261472865534, "learning_rate": 1.7540840215603094e-06, "loss": 0.476, "step": 22857 }, { "epoch": 0.8143648574024263, "grad_norm": 1.5670589739588838, "learning_rate": 1.7534312616256889e-06, "loss": 0.5257, "step": 22858 }, { "epoch": 0.8144004845289203, "grad_norm": 1.1405550136711908, "learning_rate": 1.7527786114998934e-06, "loss": 0.3999, "step": 22859 }, { "epoch": 0.8144361116554144, "grad_norm": 1.5583201787756185, "learning_rate": 1.7521260711916166e-06, "loss": 0.5472, "step": 22860 }, { "epoch": 0.8144717387819086, "grad_norm": 1.3242769457151118, "learning_rate": 1.7514736407095444e-06, "loss": 0.4021, "step": 22861 }, { "epoch": 0.8145073659084027, "grad_norm": 1.2629155753954528, "learning_rate": 1.7508213200623658e-06, "loss": 0.3898, "step": 22862 }, { "epoch": 0.8145429930348967, "grad_norm": 1.1912333216787858, "learning_rate": 1.750169109258767e-06, "loss": 0.3924, "step": 22863 }, { "epoch": 0.8145786201613909, "grad_norm": 1.2180450463756025, "learning_rate": 1.7495170083074332e-06, "loss": 0.4484, "step": 22864 }, { "epoch": 0.814614247287885, "grad_norm": 2.4315664482122443, "learning_rate": 1.7488650172170496e-06, "loss": 0.5381, "step": 22865 }, { "epoch": 0.8146498744143791, "grad_norm": 1.129522189139551, "learning_rate": 1.7482131359962917e-06, "loss": 0.3858, "step": 22866 }, { "epoch": 0.8146855015408733, "grad_norm": 1.4664515412711698, "learning_rate": 1.747561364653848e-06, "loss": 0.4734, "step": 22867 }, { "epoch": 0.8147211286673673, "grad_norm": 1.150681188427511, "learning_rate": 1.7469097031983894e-06, "loss": 0.4109, "step": 22868 }, { "epoch": 0.8147567557938614, "grad_norm": 1.5165042790210792, "learning_rate": 1.746258151638599e-06, "loss": 0.4841, "step": 22869 }, { "epoch": 0.8147923829203556, "grad_norm": 2.353714231773565, "learning_rate": 1.7456067099831475e-06, "loss": 0.4965, "step": 22870 }, { "epoch": 0.8148280100468497, "grad_norm": 1.2269114040103883, "learning_rate": 1.7449553782407136e-06, "loss": 0.4216, "step": 22871 }, { "epoch": 0.8148636371733438, "grad_norm": 1.4215356930947791, "learning_rate": 1.744304156419967e-06, "loss": 0.5286, "step": 22872 }, { "epoch": 0.8148992642998379, "grad_norm": 1.295005624389132, "learning_rate": 1.7436530445295852e-06, "loss": 0.3874, "step": 22873 }, { "epoch": 0.814934891426332, "grad_norm": 1.4263683680542023, "learning_rate": 1.74300204257823e-06, "loss": 0.4724, "step": 22874 }, { "epoch": 0.8149705185528261, "grad_norm": 1.8940837248837683, "learning_rate": 1.742351150574575e-06, "loss": 0.6136, "step": 22875 }, { "epoch": 0.8150061456793203, "grad_norm": 1.2836591755798574, "learning_rate": 1.7417003685272882e-06, "loss": 0.3972, "step": 22876 }, { "epoch": 0.8150417728058144, "grad_norm": 1.2552764193346317, "learning_rate": 1.7410496964450296e-06, "loss": 0.3536, "step": 22877 }, { "epoch": 0.8150773999323084, "grad_norm": 1.1935922765417426, "learning_rate": 1.740399134336469e-06, "loss": 0.4017, "step": 22878 }, { "epoch": 0.8151130270588026, "grad_norm": 1.668337199170449, "learning_rate": 1.739748682210266e-06, "loss": 0.5719, "step": 22879 }, { "epoch": 0.8151486541852967, "grad_norm": 2.610409948659797, "learning_rate": 1.739098340075085e-06, "loss": 0.3345, "step": 22880 }, { "epoch": 0.8151842813117908, "grad_norm": 1.5539995409797571, "learning_rate": 1.738448107939581e-06, "loss": 0.4994, "step": 22881 }, { "epoch": 0.815219908438285, "grad_norm": 1.5329819149017279, "learning_rate": 1.7377979858124184e-06, "loss": 0.4958, "step": 22882 }, { "epoch": 0.815255535564779, "grad_norm": 1.2542157259124276, "learning_rate": 1.7371479737022478e-06, "loss": 0.4032, "step": 22883 }, { "epoch": 0.8152911626912731, "grad_norm": 1.416363979774927, "learning_rate": 1.7364980716177294e-06, "loss": 0.5018, "step": 22884 }, { "epoch": 0.8153267898177673, "grad_norm": 1.5482386817828269, "learning_rate": 1.7358482795675124e-06, "loss": 0.5333, "step": 22885 }, { "epoch": 0.8153624169442614, "grad_norm": 1.3220676684490689, "learning_rate": 1.7351985975602526e-06, "loss": 0.3966, "step": 22886 }, { "epoch": 0.8153980440707554, "grad_norm": 1.3469277371187909, "learning_rate": 1.7345490256045993e-06, "loss": 0.3854, "step": 22887 }, { "epoch": 0.8154336711972496, "grad_norm": 1.1713181698729986, "learning_rate": 1.7338995637092038e-06, "loss": 0.3934, "step": 22888 }, { "epoch": 0.8154692983237437, "grad_norm": 1.609598022970934, "learning_rate": 1.7332502118827154e-06, "loss": 0.5269, "step": 22889 }, { "epoch": 0.8155049254502378, "grad_norm": 1.4361555256774867, "learning_rate": 1.7326009701337764e-06, "loss": 0.5233, "step": 22890 }, { "epoch": 0.815540552576732, "grad_norm": 1.487353526091633, "learning_rate": 1.7319518384710353e-06, "loss": 0.5789, "step": 22891 }, { "epoch": 0.815576179703226, "grad_norm": 1.3163432592230098, "learning_rate": 1.731302816903133e-06, "loss": 0.4457, "step": 22892 }, { "epoch": 0.8156118068297201, "grad_norm": 1.652019441825837, "learning_rate": 1.730653905438714e-06, "loss": 0.5235, "step": 22893 }, { "epoch": 0.8156474339562143, "grad_norm": 1.230637932612441, "learning_rate": 1.7300051040864163e-06, "loss": 0.4345, "step": 22894 }, { "epoch": 0.8156830610827084, "grad_norm": 1.1876602166263979, "learning_rate": 1.7293564128548845e-06, "loss": 0.4148, "step": 22895 }, { "epoch": 0.8157186882092025, "grad_norm": 1.2783379456517938, "learning_rate": 1.7287078317527505e-06, "loss": 0.4213, "step": 22896 }, { "epoch": 0.8157543153356966, "grad_norm": 1.5458032371534758, "learning_rate": 1.7280593607886554e-06, "loss": 0.5104, "step": 22897 }, { "epoch": 0.8157899424621907, "grad_norm": 1.2903726383282736, "learning_rate": 1.7274109999712295e-06, "loss": 0.3878, "step": 22898 }, { "epoch": 0.8158255695886848, "grad_norm": 1.1489111704278197, "learning_rate": 1.7267627493091077e-06, "loss": 0.3762, "step": 22899 }, { "epoch": 0.815861196715179, "grad_norm": 1.55102518973541, "learning_rate": 1.7261146088109226e-06, "loss": 0.5191, "step": 22900 }, { "epoch": 0.8158968238416731, "grad_norm": 1.842242224048164, "learning_rate": 1.7254665784853053e-06, "loss": 0.5162, "step": 22901 }, { "epoch": 0.8159324509681671, "grad_norm": 2.0457391040740625, "learning_rate": 1.7248186583408856e-06, "loss": 0.5077, "step": 22902 }, { "epoch": 0.8159680780946613, "grad_norm": 1.696625177495974, "learning_rate": 1.7241708483862873e-06, "loss": 0.4758, "step": 22903 }, { "epoch": 0.8160037052211554, "grad_norm": 1.2181816050461647, "learning_rate": 1.723523148630142e-06, "loss": 0.4025, "step": 22904 }, { "epoch": 0.8160393323476495, "grad_norm": 1.7697307760743184, "learning_rate": 1.7228755590810674e-06, "loss": 0.4994, "step": 22905 }, { "epoch": 0.8160749594741437, "grad_norm": 1.3507789373486803, "learning_rate": 1.722228079747692e-06, "loss": 0.431, "step": 22906 }, { "epoch": 0.8161105866006377, "grad_norm": 1.4777381280824031, "learning_rate": 1.721580710638634e-06, "loss": 0.4918, "step": 22907 }, { "epoch": 0.8161462137271318, "grad_norm": 1.2074538125677778, "learning_rate": 1.7209334517625154e-06, "loss": 0.3948, "step": 22908 }, { "epoch": 0.816181840853626, "grad_norm": 1.3417203900292318, "learning_rate": 1.7202863031279537e-06, "loss": 0.4175, "step": 22909 }, { "epoch": 0.8162174679801201, "grad_norm": 1.1863615406897756, "learning_rate": 1.7196392647435678e-06, "loss": 0.407, "step": 22910 }, { "epoch": 0.8162530951066141, "grad_norm": 1.5056217368891964, "learning_rate": 1.718992336617975e-06, "loss": 0.4268, "step": 22911 }, { "epoch": 0.8162887222331083, "grad_norm": 1.434640804354733, "learning_rate": 1.7183455187597842e-06, "loss": 0.5068, "step": 22912 }, { "epoch": 0.8163243493596024, "grad_norm": 1.23677904210253, "learning_rate": 1.7176988111776149e-06, "loss": 0.3634, "step": 22913 }, { "epoch": 0.8163599764860965, "grad_norm": 1.3456921991302673, "learning_rate": 1.7170522138800717e-06, "loss": 0.4003, "step": 22914 }, { "epoch": 0.8163956036125907, "grad_norm": 1.587066997472202, "learning_rate": 1.7164057268757683e-06, "loss": 0.5342, "step": 22915 }, { "epoch": 0.8164312307390847, "grad_norm": 1.4892580006081746, "learning_rate": 1.7157593501733117e-06, "loss": 0.4132, "step": 22916 }, { "epoch": 0.8164668578655788, "grad_norm": 1.7404948825568365, "learning_rate": 1.7151130837813123e-06, "loss": 0.5004, "step": 22917 }, { "epoch": 0.816502484992073, "grad_norm": 1.8783530241858473, "learning_rate": 1.7144669277083714e-06, "loss": 0.5535, "step": 22918 }, { "epoch": 0.8165381121185671, "grad_norm": 1.2284158866311976, "learning_rate": 1.7138208819630964e-06, "loss": 0.3864, "step": 22919 }, { "epoch": 0.8165737392450612, "grad_norm": 1.2205564224206862, "learning_rate": 1.7131749465540859e-06, "loss": 0.38, "step": 22920 }, { "epoch": 0.8166093663715553, "grad_norm": 1.198945972825193, "learning_rate": 1.7125291214899431e-06, "loss": 0.4223, "step": 22921 }, { "epoch": 0.8166449934980494, "grad_norm": 1.4412177268591748, "learning_rate": 1.7118834067792688e-06, "loss": 0.5102, "step": 22922 }, { "epoch": 0.8166806206245435, "grad_norm": 1.2404123780744085, "learning_rate": 1.7112378024306565e-06, "loss": 0.4029, "step": 22923 }, { "epoch": 0.8167162477510377, "grad_norm": 1.1386302371704138, "learning_rate": 1.710592308452711e-06, "loss": 0.4058, "step": 22924 }, { "epoch": 0.8167518748775318, "grad_norm": 1.5676788912826196, "learning_rate": 1.7099469248540202e-06, "loss": 0.4709, "step": 22925 }, { "epoch": 0.8167875020040258, "grad_norm": 1.6091594255545694, "learning_rate": 1.7093016516431837e-06, "loss": 0.497, "step": 22926 }, { "epoch": 0.81682312913052, "grad_norm": 1.5213123571433285, "learning_rate": 1.7086564888287872e-06, "loss": 0.4807, "step": 22927 }, { "epoch": 0.8168587562570141, "grad_norm": 1.1791525238573006, "learning_rate": 1.7080114364194278e-06, "loss": 0.3898, "step": 22928 }, { "epoch": 0.8168943833835082, "grad_norm": 1.7501044442885543, "learning_rate": 1.707366494423689e-06, "loss": 0.4897, "step": 22929 }, { "epoch": 0.8169300105100024, "grad_norm": 1.1452777097031777, "learning_rate": 1.7067216628501616e-06, "loss": 0.3578, "step": 22930 }, { "epoch": 0.8169656376364964, "grad_norm": 1.1150650530226862, "learning_rate": 1.706076941707432e-06, "loss": 0.3806, "step": 22931 }, { "epoch": 0.8170012647629905, "grad_norm": 1.2303700344925028, "learning_rate": 1.7054323310040877e-06, "loss": 0.4288, "step": 22932 }, { "epoch": 0.8170368918894847, "grad_norm": 1.1605237059423357, "learning_rate": 1.7047878307487065e-06, "loss": 0.4102, "step": 22933 }, { "epoch": 0.8170725190159788, "grad_norm": 1.294947672358455, "learning_rate": 1.7041434409498737e-06, "loss": 0.3918, "step": 22934 }, { "epoch": 0.8171081461424728, "grad_norm": 1.372255949234931, "learning_rate": 1.7034991616161723e-06, "loss": 0.4322, "step": 22935 }, { "epoch": 0.817143773268967, "grad_norm": 1.1431942847214733, "learning_rate": 1.7028549927561756e-06, "loss": 0.3768, "step": 22936 }, { "epoch": 0.8171794003954611, "grad_norm": 1.1424740926706964, "learning_rate": 1.7022109343784643e-06, "loss": 0.3999, "step": 22937 }, { "epoch": 0.8172150275219552, "grad_norm": 1.1220866963431413, "learning_rate": 1.701566986491614e-06, "loss": 0.391, "step": 22938 }, { "epoch": 0.8172506546484494, "grad_norm": 1.2724612139104938, "learning_rate": 1.700923149104202e-06, "loss": 0.4218, "step": 22939 }, { "epoch": 0.8172862817749434, "grad_norm": 1.4203598096928498, "learning_rate": 1.7002794222247966e-06, "loss": 0.3957, "step": 22940 }, { "epoch": 0.8173219089014375, "grad_norm": 1.114730911582194, "learning_rate": 1.6996358058619733e-06, "loss": 0.3989, "step": 22941 }, { "epoch": 0.8173575360279317, "grad_norm": 1.4596776559310032, "learning_rate": 1.6989923000243003e-06, "loss": 0.4536, "step": 22942 }, { "epoch": 0.8173931631544258, "grad_norm": 1.4489452037476642, "learning_rate": 1.6983489047203482e-06, "loss": 0.5392, "step": 22943 }, { "epoch": 0.8174287902809199, "grad_norm": 1.2017675753733907, "learning_rate": 1.6977056199586806e-06, "loss": 0.4048, "step": 22944 }, { "epoch": 0.817464417407414, "grad_norm": 1.2260868781882641, "learning_rate": 1.6970624457478658e-06, "loss": 0.4386, "step": 22945 }, { "epoch": 0.8175000445339081, "grad_norm": 1.268909090449392, "learning_rate": 1.6964193820964669e-06, "loss": 0.436, "step": 22946 }, { "epoch": 0.8175356716604022, "grad_norm": 1.2316132180143746, "learning_rate": 1.6957764290130484e-06, "loss": 0.4763, "step": 22947 }, { "epoch": 0.8175712987868964, "grad_norm": 1.4927196318496216, "learning_rate": 1.6951335865061746e-06, "loss": 0.5406, "step": 22948 }, { "epoch": 0.8176069259133905, "grad_norm": 1.3744745858454614, "learning_rate": 1.6944908545843986e-06, "loss": 0.4664, "step": 22949 }, { "epoch": 0.8176425530398845, "grad_norm": 1.324069659087754, "learning_rate": 1.6938482332562844e-06, "loss": 0.4071, "step": 22950 }, { "epoch": 0.8176781801663787, "grad_norm": 1.5933539867718491, "learning_rate": 1.6932057225303844e-06, "loss": 0.4582, "step": 22951 }, { "epoch": 0.8177138072928728, "grad_norm": 1.27670681465823, "learning_rate": 1.692563322415256e-06, "loss": 0.4415, "step": 22952 }, { "epoch": 0.8177494344193669, "grad_norm": 1.4679179116344712, "learning_rate": 1.6919210329194535e-06, "loss": 0.4927, "step": 22953 }, { "epoch": 0.8177850615458611, "grad_norm": 1.5014482572358911, "learning_rate": 1.6912788540515324e-06, "loss": 0.5156, "step": 22954 }, { "epoch": 0.8178206886723551, "grad_norm": 1.5428454033259835, "learning_rate": 1.6906367858200379e-06, "loss": 0.5383, "step": 22955 }, { "epoch": 0.8178563157988492, "grad_norm": 1.402869281531224, "learning_rate": 1.6899948282335221e-06, "loss": 0.4105, "step": 22956 }, { "epoch": 0.8178919429253434, "grad_norm": 1.200252346220041, "learning_rate": 1.6893529813005371e-06, "loss": 0.4358, "step": 22957 }, { "epoch": 0.8179275700518375, "grad_norm": 1.5318293356829547, "learning_rate": 1.6887112450296227e-06, "loss": 0.506, "step": 22958 }, { "epoch": 0.8179631971783315, "grad_norm": 1.6458211465625554, "learning_rate": 1.68806961942933e-06, "loss": 0.5233, "step": 22959 }, { "epoch": 0.8179988243048257, "grad_norm": 1.266139903823725, "learning_rate": 1.6874281045081976e-06, "loss": 0.5502, "step": 22960 }, { "epoch": 0.8180344514313198, "grad_norm": 1.4435538233242642, "learning_rate": 1.6867867002747696e-06, "loss": 0.4976, "step": 22961 }, { "epoch": 0.8180700785578139, "grad_norm": 1.282981188098478, "learning_rate": 1.6861454067375882e-06, "loss": 0.4115, "step": 22962 }, { "epoch": 0.8181057056843081, "grad_norm": 1.554077629535275, "learning_rate": 1.6855042239051934e-06, "loss": 0.3815, "step": 22963 }, { "epoch": 0.8181413328108021, "grad_norm": 1.4005789654595877, "learning_rate": 1.6848631517861192e-06, "loss": 0.5402, "step": 22964 }, { "epoch": 0.8181769599372962, "grad_norm": 1.218366868794018, "learning_rate": 1.6842221903889066e-06, "loss": 0.3914, "step": 22965 }, { "epoch": 0.8182125870637904, "grad_norm": 1.8046673596993412, "learning_rate": 1.6835813397220857e-06, "loss": 0.4896, "step": 22966 }, { "epoch": 0.8182482141902845, "grad_norm": 1.2409858310549267, "learning_rate": 1.6829405997941917e-06, "loss": 0.3746, "step": 22967 }, { "epoch": 0.8182838413167786, "grad_norm": 1.6252322750503594, "learning_rate": 1.6822999706137565e-06, "loss": 0.3872, "step": 22968 }, { "epoch": 0.8183194684432727, "grad_norm": 1.217936057920155, "learning_rate": 1.6816594521893114e-06, "loss": 0.3901, "step": 22969 }, { "epoch": 0.8183550955697668, "grad_norm": 1.3621410947174029, "learning_rate": 1.6810190445293873e-06, "loss": 0.5275, "step": 22970 }, { "epoch": 0.8183907226962609, "grad_norm": 1.6748359597074536, "learning_rate": 1.680378747642507e-06, "loss": 0.5532, "step": 22971 }, { "epoch": 0.8184263498227551, "grad_norm": 1.1462660229057964, "learning_rate": 1.6797385615372008e-06, "loss": 0.389, "step": 22972 }, { "epoch": 0.8184619769492492, "grad_norm": 1.9295433340618662, "learning_rate": 1.6790984862219895e-06, "loss": 0.503, "step": 22973 }, { "epoch": 0.8184976040757432, "grad_norm": 1.3844789099795953, "learning_rate": 1.6784585217053994e-06, "loss": 0.5298, "step": 22974 }, { "epoch": 0.8185332312022374, "grad_norm": 1.3631264267565086, "learning_rate": 1.6778186679959453e-06, "loss": 0.422, "step": 22975 }, { "epoch": 0.8185688583287315, "grad_norm": 1.2063286945374536, "learning_rate": 1.6771789251021586e-06, "loss": 0.422, "step": 22976 }, { "epoch": 0.8186044854552256, "grad_norm": 1.3752805731951188, "learning_rate": 1.6765392930325486e-06, "loss": 0.4999, "step": 22977 }, { "epoch": 0.8186401125817198, "grad_norm": 1.6743642022817034, "learning_rate": 1.6758997717956382e-06, "loss": 0.5531, "step": 22978 }, { "epoch": 0.8186757397082138, "grad_norm": 1.2743046395823552, "learning_rate": 1.6752603613999384e-06, "loss": 0.4012, "step": 22979 }, { "epoch": 0.8187113668347079, "grad_norm": 1.282572914483351, "learning_rate": 1.674621061853965e-06, "loss": 0.3827, "step": 22980 }, { "epoch": 0.8187469939612021, "grad_norm": 1.208927343284305, "learning_rate": 1.6739818731662338e-06, "loss": 0.405, "step": 22981 }, { "epoch": 0.8187826210876962, "grad_norm": 1.1895859151190866, "learning_rate": 1.673342795345251e-06, "loss": 0.4156, "step": 22982 }, { "epoch": 0.8188182482141902, "grad_norm": 1.6050832982018466, "learning_rate": 1.672703828399529e-06, "loss": 0.4965, "step": 22983 }, { "epoch": 0.8188538753406844, "grad_norm": 1.1959094256804648, "learning_rate": 1.672064972337576e-06, "loss": 0.3545, "step": 22984 }, { "epoch": 0.8188895024671785, "grad_norm": 1.8764216199441173, "learning_rate": 1.6714262271679016e-06, "loss": 0.5554, "step": 22985 }, { "epoch": 0.8189251295936726, "grad_norm": 1.1014604411913005, "learning_rate": 1.6707875928990059e-06, "loss": 0.3915, "step": 22986 }, { "epoch": 0.8189607567201668, "grad_norm": 1.468234503529911, "learning_rate": 1.6701490695393973e-06, "loss": 0.4842, "step": 22987 }, { "epoch": 0.8189963838466608, "grad_norm": 1.548041769043861, "learning_rate": 1.6695106570975728e-06, "loss": 0.5732, "step": 22988 }, { "epoch": 0.8190320109731549, "grad_norm": 2.036306928756787, "learning_rate": 1.6688723555820374e-06, "loss": 0.4736, "step": 22989 }, { "epoch": 0.8190676380996491, "grad_norm": 2.2138020065940003, "learning_rate": 1.6682341650012902e-06, "loss": 0.515, "step": 22990 }, { "epoch": 0.8191032652261432, "grad_norm": 1.672454721675552, "learning_rate": 1.6675960853638273e-06, "loss": 0.4536, "step": 22991 }, { "epoch": 0.8191388923526373, "grad_norm": 1.8177880065025307, "learning_rate": 1.66695811667815e-06, "loss": 0.5219, "step": 22992 }, { "epoch": 0.8191745194791314, "grad_norm": 1.278542865303994, "learning_rate": 1.6663202589527471e-06, "loss": 0.4076, "step": 22993 }, { "epoch": 0.8192101466056255, "grad_norm": 1.6492252936126919, "learning_rate": 1.6656825121961172e-06, "loss": 0.5146, "step": 22994 }, { "epoch": 0.8192457737321196, "grad_norm": 1.3999929410451208, "learning_rate": 1.6650448764167481e-06, "loss": 0.4001, "step": 22995 }, { "epoch": 0.8192814008586138, "grad_norm": 1.11304949053715, "learning_rate": 1.664407351623134e-06, "loss": 0.3996, "step": 22996 }, { "epoch": 0.8193170279851079, "grad_norm": 1.1430819900772022, "learning_rate": 1.6637699378237605e-06, "loss": 0.3916, "step": 22997 }, { "epoch": 0.8193526551116019, "grad_norm": 1.3827874090622367, "learning_rate": 1.663132635027116e-06, "loss": 0.3655, "step": 22998 }, { "epoch": 0.8193882822380961, "grad_norm": 1.4832428458677023, "learning_rate": 1.6624954432416884e-06, "loss": 0.4145, "step": 22999 }, { "epoch": 0.8194239093645902, "grad_norm": 1.340708563965806, "learning_rate": 1.6618583624759631e-06, "loss": 0.4414, "step": 23000 }, { "epoch": 0.8194595364910843, "grad_norm": 2.063717776291432, "learning_rate": 1.6612213927384202e-06, "loss": 0.4425, "step": 23001 }, { "epoch": 0.8194951636175785, "grad_norm": 1.3983731704397868, "learning_rate": 1.6605845340375427e-06, "loss": 0.4767, "step": 23002 }, { "epoch": 0.8195307907440725, "grad_norm": 1.8813020972541439, "learning_rate": 1.6599477863818125e-06, "loss": 0.5591, "step": 23003 }, { "epoch": 0.8195664178705666, "grad_norm": 1.5313395710994777, "learning_rate": 1.6593111497797055e-06, "loss": 0.4733, "step": 23004 }, { "epoch": 0.8196020449970608, "grad_norm": 1.6927793578877575, "learning_rate": 1.6586746242397001e-06, "loss": 0.5192, "step": 23005 }, { "epoch": 0.8196376721235549, "grad_norm": 1.6995069311716515, "learning_rate": 1.658038209770273e-06, "loss": 0.5432, "step": 23006 }, { "epoch": 0.819673299250049, "grad_norm": 1.2591303989130835, "learning_rate": 1.6574019063798997e-06, "loss": 0.381, "step": 23007 }, { "epoch": 0.8197089263765431, "grad_norm": 1.6140478344824796, "learning_rate": 1.6567657140770477e-06, "loss": 0.5684, "step": 23008 }, { "epoch": 0.8197445535030372, "grad_norm": 1.4216715553673922, "learning_rate": 1.656129632870196e-06, "loss": 0.5555, "step": 23009 }, { "epoch": 0.8197801806295313, "grad_norm": 1.176634108962201, "learning_rate": 1.6554936627678065e-06, "loss": 0.3994, "step": 23010 }, { "epoch": 0.8198158077560255, "grad_norm": 1.2204811113146226, "learning_rate": 1.6548578037783536e-06, "loss": 0.4157, "step": 23011 }, { "epoch": 0.8198514348825195, "grad_norm": 1.4812422822978413, "learning_rate": 1.6542220559102983e-06, "loss": 0.492, "step": 23012 }, { "epoch": 0.8198870620090136, "grad_norm": 1.5124336601257065, "learning_rate": 1.653586419172114e-06, "loss": 0.5121, "step": 23013 }, { "epoch": 0.8199226891355078, "grad_norm": 1.265517666469891, "learning_rate": 1.652950893572257e-06, "loss": 0.3801, "step": 23014 }, { "epoch": 0.8199583162620019, "grad_norm": 1.4608700941113113, "learning_rate": 1.6523154791191942e-06, "loss": 0.5297, "step": 23015 }, { "epoch": 0.819993943388496, "grad_norm": 1.4138467829084083, "learning_rate": 1.6516801758213873e-06, "loss": 0.4511, "step": 23016 }, { "epoch": 0.8200295705149901, "grad_norm": 1.3412797967148613, "learning_rate": 1.6510449836872922e-06, "loss": 0.4483, "step": 23017 }, { "epoch": 0.8200651976414842, "grad_norm": 1.539615848629006, "learning_rate": 1.6504099027253707e-06, "loss": 0.4875, "step": 23018 }, { "epoch": 0.8201008247679783, "grad_norm": 1.4810582571273447, "learning_rate": 1.649774932944075e-06, "loss": 0.5581, "step": 23019 }, { "epoch": 0.8201364518944725, "grad_norm": 1.3024358785983303, "learning_rate": 1.6491400743518627e-06, "loss": 0.5061, "step": 23020 }, { "epoch": 0.8201720790209666, "grad_norm": 1.4494857201829079, "learning_rate": 1.648505326957187e-06, "loss": 0.5131, "step": 23021 }, { "epoch": 0.8202077061474606, "grad_norm": 1.3050414029620934, "learning_rate": 1.6478706907685027e-06, "loss": 0.4111, "step": 23022 }, { "epoch": 0.8202433332739548, "grad_norm": 1.2813479656583908, "learning_rate": 1.647236165794256e-06, "loss": 0.3809, "step": 23023 }, { "epoch": 0.8202789604004489, "grad_norm": 1.1882957816629152, "learning_rate": 1.6466017520429001e-06, "loss": 0.3927, "step": 23024 }, { "epoch": 0.820314587526943, "grad_norm": 1.6128391403501452, "learning_rate": 1.6459674495228772e-06, "loss": 0.5165, "step": 23025 }, { "epoch": 0.8203502146534372, "grad_norm": 1.7673543593003584, "learning_rate": 1.6453332582426384e-06, "loss": 0.3888, "step": 23026 }, { "epoch": 0.8203858417799312, "grad_norm": 1.773444457596672, "learning_rate": 1.6446991782106258e-06, "loss": 0.5464, "step": 23027 }, { "epoch": 0.8204214689064253, "grad_norm": 1.3431477969373784, "learning_rate": 1.6440652094352838e-06, "loss": 0.4254, "step": 23028 }, { "epoch": 0.8204570960329195, "grad_norm": 1.186091736719743, "learning_rate": 1.6434313519250567e-06, "loss": 0.379, "step": 23029 }, { "epoch": 0.8204927231594136, "grad_norm": 1.3874452597318239, "learning_rate": 1.6427976056883787e-06, "loss": 0.4742, "step": 23030 }, { "epoch": 0.8205283502859076, "grad_norm": 1.1056178206582945, "learning_rate": 1.6421639707336956e-06, "loss": 0.4042, "step": 23031 }, { "epoch": 0.8205639774124018, "grad_norm": 1.2354167528641402, "learning_rate": 1.6415304470694382e-06, "loss": 0.4304, "step": 23032 }, { "epoch": 0.8205996045388959, "grad_norm": 1.284113234758541, "learning_rate": 1.6408970347040475e-06, "loss": 0.4101, "step": 23033 }, { "epoch": 0.82063523166539, "grad_norm": 1.698926604615148, "learning_rate": 1.6402637336459527e-06, "loss": 0.612, "step": 23034 }, { "epoch": 0.8206708587918842, "grad_norm": 1.5898266961092602, "learning_rate": 1.6396305439035898e-06, "loss": 0.466, "step": 23035 }, { "epoch": 0.8207064859183782, "grad_norm": 1.1109113433481417, "learning_rate": 1.6389974654853892e-06, "loss": 0.3622, "step": 23036 }, { "epoch": 0.8207421130448723, "grad_norm": 1.859285608356962, "learning_rate": 1.6383644983997816e-06, "loss": 0.5387, "step": 23037 }, { "epoch": 0.8207777401713665, "grad_norm": 1.7107622728758334, "learning_rate": 1.637731642655197e-06, "loss": 0.4781, "step": 23038 }, { "epoch": 0.8208133672978606, "grad_norm": 1.432863046400261, "learning_rate": 1.6370988982600588e-06, "loss": 0.4595, "step": 23039 }, { "epoch": 0.8208489944243547, "grad_norm": 1.203746665737681, "learning_rate": 1.636466265222797e-06, "loss": 0.41, "step": 23040 }, { "epoch": 0.8208846215508488, "grad_norm": 1.7024782641075786, "learning_rate": 1.635833743551829e-06, "loss": 0.5612, "step": 23041 }, { "epoch": 0.8209202486773429, "grad_norm": 1.0546772711727967, "learning_rate": 1.6352013332555817e-06, "loss": 0.4221, "step": 23042 }, { "epoch": 0.820955875803837, "grad_norm": 1.342961011629017, "learning_rate": 1.6345690343424758e-06, "loss": 0.4196, "step": 23043 }, { "epoch": 0.8209915029303312, "grad_norm": 1.3623182784965249, "learning_rate": 1.6339368468209316e-06, "loss": 0.4257, "step": 23044 }, { "epoch": 0.8210271300568253, "grad_norm": 1.5892140999629794, "learning_rate": 1.6333047706993632e-06, "loss": 0.583, "step": 23045 }, { "epoch": 0.8210627571833193, "grad_norm": 1.1694097575861384, "learning_rate": 1.632672805986193e-06, "loss": 0.4425, "step": 23046 }, { "epoch": 0.8210983843098135, "grad_norm": 1.5249084213062978, "learning_rate": 1.6320409526898307e-06, "loss": 0.4793, "step": 23047 }, { "epoch": 0.8211340114363076, "grad_norm": 1.1608152404126075, "learning_rate": 1.6314092108186907e-06, "loss": 0.3552, "step": 23048 }, { "epoch": 0.8211696385628017, "grad_norm": 1.4082254606642541, "learning_rate": 1.63077758038119e-06, "loss": 0.5081, "step": 23049 }, { "epoch": 0.8212052656892959, "grad_norm": 1.3470602156605669, "learning_rate": 1.63014606138573e-06, "loss": 0.3992, "step": 23050 }, { "epoch": 0.8212408928157899, "grad_norm": 1.1287635348684226, "learning_rate": 1.6295146538407303e-06, "loss": 0.3788, "step": 23051 }, { "epoch": 0.821276519942284, "grad_norm": 1.3966000412139892, "learning_rate": 1.6288833577545914e-06, "loss": 0.4751, "step": 23052 }, { "epoch": 0.8213121470687782, "grad_norm": 1.4993075085049838, "learning_rate": 1.6282521731357237e-06, "loss": 0.5829, "step": 23053 }, { "epoch": 0.8213477741952723, "grad_norm": 1.2969573224031152, "learning_rate": 1.6276210999925279e-06, "loss": 0.3806, "step": 23054 }, { "epoch": 0.8213834013217663, "grad_norm": 1.5906855680501435, "learning_rate": 1.626990138333412e-06, "loss": 0.4847, "step": 23055 }, { "epoch": 0.8214190284482605, "grad_norm": 1.094671865087031, "learning_rate": 1.6263592881667712e-06, "loss": 0.3959, "step": 23056 }, { "epoch": 0.8214546555747546, "grad_norm": 1.3894425992193684, "learning_rate": 1.62572854950101e-06, "loss": 0.4534, "step": 23057 }, { "epoch": 0.8214902827012487, "grad_norm": 1.2043313389054102, "learning_rate": 1.6250979223445262e-06, "loss": 0.3966, "step": 23058 }, { "epoch": 0.8215259098277429, "grad_norm": 1.6507963153521634, "learning_rate": 1.6244674067057197e-06, "loss": 0.4915, "step": 23059 }, { "epoch": 0.821561536954237, "grad_norm": 1.415947983816295, "learning_rate": 1.6238370025929807e-06, "loss": 0.5096, "step": 23060 }, { "epoch": 0.821597164080731, "grad_norm": 1.428136510233719, "learning_rate": 1.6232067100147076e-06, "loss": 0.4168, "step": 23061 }, { "epoch": 0.8216327912072252, "grad_norm": 2.043565442790881, "learning_rate": 1.6225765289792939e-06, "loss": 0.5106, "step": 23062 }, { "epoch": 0.8216684183337193, "grad_norm": 1.4204062244309017, "learning_rate": 1.6219464594951273e-06, "loss": 0.5585, "step": 23063 }, { "epoch": 0.8217040454602134, "grad_norm": 1.1775671399172443, "learning_rate": 1.6213165015706e-06, "loss": 0.4309, "step": 23064 }, { "epoch": 0.8217396725867075, "grad_norm": 1.3356890711155063, "learning_rate": 1.6206866552140988e-06, "loss": 0.4156, "step": 23065 }, { "epoch": 0.8217752997132016, "grad_norm": 1.2299647413677013, "learning_rate": 1.6200569204340143e-06, "loss": 0.3761, "step": 23066 }, { "epoch": 0.8218109268396957, "grad_norm": 2.115424269611484, "learning_rate": 1.619427297238727e-06, "loss": 0.511, "step": 23067 }, { "epoch": 0.8218465539661899, "grad_norm": 2.5644726561885554, "learning_rate": 1.6187977856366255e-06, "loss": 0.4908, "step": 23068 }, { "epoch": 0.821882181092684, "grad_norm": 1.2606581121238463, "learning_rate": 1.6181683856360864e-06, "loss": 0.3235, "step": 23069 }, { "epoch": 0.821917808219178, "grad_norm": 1.184460745763151, "learning_rate": 1.6175390972454962e-06, "loss": 0.4359, "step": 23070 }, { "epoch": 0.8219534353456722, "grad_norm": 1.4782923348887642, "learning_rate": 1.6169099204732309e-06, "loss": 0.4843, "step": 23071 }, { "epoch": 0.8219890624721663, "grad_norm": 1.3715826838854692, "learning_rate": 1.6162808553276677e-06, "loss": 0.4465, "step": 23072 }, { "epoch": 0.8220246895986604, "grad_norm": 1.1681205163472181, "learning_rate": 1.6156519018171856e-06, "loss": 0.3657, "step": 23073 }, { "epoch": 0.8220603167251546, "grad_norm": 1.1723055251366363, "learning_rate": 1.6150230599501593e-06, "loss": 0.4376, "step": 23074 }, { "epoch": 0.8220959438516486, "grad_norm": 1.8086817830815958, "learning_rate": 1.6143943297349629e-06, "loss": 0.5761, "step": 23075 }, { "epoch": 0.8221315709781427, "grad_norm": 2.3065805506452075, "learning_rate": 1.6137657111799654e-06, "loss": 0.4694, "step": 23076 }, { "epoch": 0.8221671981046369, "grad_norm": 1.2728193176344926, "learning_rate": 1.613137204293541e-06, "loss": 0.4441, "step": 23077 }, { "epoch": 0.822202825231131, "grad_norm": 1.3257884719712667, "learning_rate": 1.6125088090840556e-06, "loss": 0.3919, "step": 23078 }, { "epoch": 0.822238452357625, "grad_norm": 1.588169164446347, "learning_rate": 1.6118805255598769e-06, "loss": 0.4972, "step": 23079 }, { "epoch": 0.8222740794841192, "grad_norm": 1.324589227398047, "learning_rate": 1.6112523537293722e-06, "loss": 0.3619, "step": 23080 }, { "epoch": 0.8223097066106133, "grad_norm": 1.451929730690782, "learning_rate": 1.6106242936009087e-06, "loss": 0.5426, "step": 23081 }, { "epoch": 0.8223453337371074, "grad_norm": 1.2166041889672592, "learning_rate": 1.6099963451828427e-06, "loss": 0.4156, "step": 23082 }, { "epoch": 0.8223809608636016, "grad_norm": 1.575109224741532, "learning_rate": 1.609368508483541e-06, "loss": 0.5046, "step": 23083 }, { "epoch": 0.8224165879900956, "grad_norm": 1.3574729602025701, "learning_rate": 1.6087407835113644e-06, "loss": 0.3755, "step": 23084 }, { "epoch": 0.8224522151165897, "grad_norm": 1.6389870746343764, "learning_rate": 1.6081131702746666e-06, "loss": 0.5444, "step": 23085 }, { "epoch": 0.8224878422430839, "grad_norm": 1.291582319509674, "learning_rate": 1.6074856687818107e-06, "loss": 0.4468, "step": 23086 }, { "epoch": 0.822523469369578, "grad_norm": 1.324389948017073, "learning_rate": 1.6068582790411435e-06, "loss": 0.4007, "step": 23087 }, { "epoch": 0.8225590964960721, "grad_norm": 1.1683101225143213, "learning_rate": 1.6062310010610294e-06, "loss": 0.3893, "step": 23088 }, { "epoch": 0.8225947236225662, "grad_norm": 1.1934784035860364, "learning_rate": 1.6056038348498149e-06, "loss": 0.3961, "step": 23089 }, { "epoch": 0.8226303507490603, "grad_norm": 1.208046246373908, "learning_rate": 1.6049767804158546e-06, "loss": 0.4273, "step": 23090 }, { "epoch": 0.8226659778755544, "grad_norm": 1.3586747269290813, "learning_rate": 1.6043498377674937e-06, "loss": 0.5039, "step": 23091 }, { "epoch": 0.8227016050020486, "grad_norm": 1.7937592142269048, "learning_rate": 1.603723006913086e-06, "loss": 0.5665, "step": 23092 }, { "epoch": 0.8227372321285427, "grad_norm": 1.2253697529356333, "learning_rate": 1.6030962878609723e-06, "loss": 0.3856, "step": 23093 }, { "epoch": 0.8227728592550367, "grad_norm": 1.117609178169066, "learning_rate": 1.6024696806195007e-06, "loss": 0.4058, "step": 23094 }, { "epoch": 0.8228084863815309, "grad_norm": 1.9855164006426829, "learning_rate": 1.601843185197014e-06, "loss": 0.5115, "step": 23095 }, { "epoch": 0.822844113508025, "grad_norm": 2.145579690694027, "learning_rate": 1.601216801601857e-06, "loss": 0.5467, "step": 23096 }, { "epoch": 0.8228797406345191, "grad_norm": 1.5785069619091097, "learning_rate": 1.6005905298423696e-06, "loss": 0.4801, "step": 23097 }, { "epoch": 0.8229153677610133, "grad_norm": 1.178802180267138, "learning_rate": 1.5999643699268886e-06, "loss": 0.4211, "step": 23098 }, { "epoch": 0.8229509948875073, "grad_norm": 1.2107194859947694, "learning_rate": 1.5993383218637547e-06, "loss": 0.4371, "step": 23099 }, { "epoch": 0.8229866220140014, "grad_norm": 1.6420406025247545, "learning_rate": 1.5987123856613006e-06, "loss": 0.5877, "step": 23100 }, { "epoch": 0.8230222491404956, "grad_norm": 1.353105354680838, "learning_rate": 1.598086561327864e-06, "loss": 0.3846, "step": 23101 }, { "epoch": 0.8230578762669897, "grad_norm": 1.262509271944262, "learning_rate": 1.597460848871777e-06, "loss": 0.3778, "step": 23102 }, { "epoch": 0.8230935033934837, "grad_norm": 1.642564505324309, "learning_rate": 1.5968352483013738e-06, "loss": 0.494, "step": 23103 }, { "epoch": 0.8231291305199779, "grad_norm": 1.292131023505941, "learning_rate": 1.5962097596249816e-06, "loss": 0.4129, "step": 23104 }, { "epoch": 0.823164757646472, "grad_norm": 1.6494060884931014, "learning_rate": 1.595584382850931e-06, "loss": 0.4655, "step": 23105 }, { "epoch": 0.8232003847729661, "grad_norm": 1.3710000583409694, "learning_rate": 1.5949591179875479e-06, "loss": 0.498, "step": 23106 }, { "epoch": 0.8232360118994603, "grad_norm": 1.4403578105900687, "learning_rate": 1.5943339650431578e-06, "loss": 0.4441, "step": 23107 }, { "epoch": 0.8232716390259543, "grad_norm": 1.2125088586429784, "learning_rate": 1.5937089240260883e-06, "loss": 0.5019, "step": 23108 }, { "epoch": 0.8233072661524484, "grad_norm": 1.2117872169984114, "learning_rate": 1.5930839949446586e-06, "loss": 0.4657, "step": 23109 }, { "epoch": 0.8233428932789426, "grad_norm": 1.287926021942413, "learning_rate": 1.5924591778071917e-06, "loss": 0.4353, "step": 23110 }, { "epoch": 0.8233785204054367, "grad_norm": 1.6031481126506875, "learning_rate": 1.591834472622007e-06, "loss": 0.4831, "step": 23111 }, { "epoch": 0.8234141475319308, "grad_norm": 1.4936837662392364, "learning_rate": 1.5912098793974252e-06, "loss": 0.4346, "step": 23112 }, { "epoch": 0.823449774658425, "grad_norm": 2.416516099150746, "learning_rate": 1.5905853981417597e-06, "loss": 0.4272, "step": 23113 }, { "epoch": 0.823485401784919, "grad_norm": 1.1284292044490745, "learning_rate": 1.5899610288633293e-06, "loss": 0.4208, "step": 23114 }, { "epoch": 0.8235210289114131, "grad_norm": 1.9231418884797629, "learning_rate": 1.5893367715704443e-06, "loss": 0.5084, "step": 23115 }, { "epoch": 0.8235566560379073, "grad_norm": 1.1571277555079444, "learning_rate": 1.5887126262714181e-06, "loss": 0.43, "step": 23116 }, { "epoch": 0.8235922831644014, "grad_norm": 1.4815126568149533, "learning_rate": 1.5880885929745627e-06, "loss": 0.4423, "step": 23117 }, { "epoch": 0.8236279102908954, "grad_norm": 1.274958635045961, "learning_rate": 1.587464671688187e-06, "loss": 0.4114, "step": 23118 }, { "epoch": 0.8236635374173896, "grad_norm": 1.2430292409121297, "learning_rate": 1.5868408624206022e-06, "loss": 0.4099, "step": 23119 }, { "epoch": 0.8236991645438837, "grad_norm": 1.1185592081538545, "learning_rate": 1.5862171651801096e-06, "loss": 0.4051, "step": 23120 }, { "epoch": 0.8237347916703778, "grad_norm": 1.5242279197676503, "learning_rate": 1.5855935799750177e-06, "loss": 0.4891, "step": 23121 }, { "epoch": 0.823770418796872, "grad_norm": 1.620423379499024, "learning_rate": 1.5849701068136269e-06, "loss": 0.4455, "step": 23122 }, { "epoch": 0.823806045923366, "grad_norm": 1.1858712123589856, "learning_rate": 1.5843467457042427e-06, "loss": 0.4109, "step": 23123 }, { "epoch": 0.8238416730498601, "grad_norm": 1.239037739324119, "learning_rate": 1.5837234966551619e-06, "loss": 0.3962, "step": 23124 }, { "epoch": 0.8238773001763543, "grad_norm": 1.175455615092537, "learning_rate": 1.5831003596746853e-06, "loss": 0.426, "step": 23125 }, { "epoch": 0.8239129273028484, "grad_norm": 1.4409359684369083, "learning_rate": 1.5824773347711098e-06, "loss": 0.4785, "step": 23126 }, { "epoch": 0.8239485544293425, "grad_norm": 1.1330161609633378, "learning_rate": 1.5818544219527354e-06, "loss": 0.4042, "step": 23127 }, { "epoch": 0.8239841815558366, "grad_norm": 1.632662984542331, "learning_rate": 1.581231621227851e-06, "loss": 0.5009, "step": 23128 }, { "epoch": 0.8240198086823307, "grad_norm": 1.674691058526897, "learning_rate": 1.5806089326047514e-06, "loss": 0.5433, "step": 23129 }, { "epoch": 0.8240554358088248, "grad_norm": 1.22415262981295, "learning_rate": 1.5799863560917307e-06, "loss": 0.4282, "step": 23130 }, { "epoch": 0.824091062935319, "grad_norm": 1.1811740674824733, "learning_rate": 1.5793638916970756e-06, "loss": 0.4142, "step": 23131 }, { "epoch": 0.824126690061813, "grad_norm": 1.2107544894297684, "learning_rate": 1.578741539429075e-06, "loss": 0.4107, "step": 23132 }, { "epoch": 0.8241623171883071, "grad_norm": 1.3682228651097579, "learning_rate": 1.578119299296017e-06, "loss": 0.4331, "step": 23133 }, { "epoch": 0.8241979443148013, "grad_norm": 1.4144632869676683, "learning_rate": 1.5774971713061905e-06, "loss": 0.4646, "step": 23134 }, { "epoch": 0.8242335714412954, "grad_norm": 1.322121512485952, "learning_rate": 1.5768751554678719e-06, "loss": 0.4649, "step": 23135 }, { "epoch": 0.8242691985677895, "grad_norm": 1.6436592743609202, "learning_rate": 1.5762532517893515e-06, "loss": 0.5565, "step": 23136 }, { "epoch": 0.8243048256942836, "grad_norm": 1.3958523702037668, "learning_rate": 1.5756314602789046e-06, "loss": 0.3917, "step": 23137 }, { "epoch": 0.8243404528207777, "grad_norm": 1.2199038280230858, "learning_rate": 1.5750097809448162e-06, "loss": 0.4195, "step": 23138 }, { "epoch": 0.8243760799472718, "grad_norm": 1.8951705048474108, "learning_rate": 1.5743882137953549e-06, "loss": 0.4159, "step": 23139 }, { "epoch": 0.824411707073766, "grad_norm": 1.1440725893458925, "learning_rate": 1.5737667588388095e-06, "loss": 0.3776, "step": 23140 }, { "epoch": 0.8244473342002601, "grad_norm": 1.5995010554775677, "learning_rate": 1.573145416083447e-06, "loss": 0.5702, "step": 23141 }, { "epoch": 0.8244829613267541, "grad_norm": 1.165197147244079, "learning_rate": 1.572524185537543e-06, "loss": 0.3864, "step": 23142 }, { "epoch": 0.8245185884532483, "grad_norm": 1.173850870190025, "learning_rate": 1.5719030672093717e-06, "loss": 0.41, "step": 23143 }, { "epoch": 0.8245542155797424, "grad_norm": 1.5617746582975116, "learning_rate": 1.5712820611072e-06, "loss": 0.5965, "step": 23144 }, { "epoch": 0.8245898427062365, "grad_norm": 1.1406069202467455, "learning_rate": 1.5706611672393013e-06, "loss": 0.4236, "step": 23145 }, { "epoch": 0.8246254698327307, "grad_norm": 1.1999781359538024, "learning_rate": 1.570040385613939e-06, "loss": 0.4074, "step": 23146 }, { "epoch": 0.8246610969592247, "grad_norm": 1.117850688431599, "learning_rate": 1.569419716239381e-06, "loss": 0.4036, "step": 23147 }, { "epoch": 0.8246967240857188, "grad_norm": 1.383743277510929, "learning_rate": 1.5687991591238916e-06, "loss": 0.5099, "step": 23148 }, { "epoch": 0.824732351212213, "grad_norm": 1.6693860336731916, "learning_rate": 1.5681787142757375e-06, "loss": 0.4925, "step": 23149 }, { "epoch": 0.8247679783387071, "grad_norm": 1.4201691365182334, "learning_rate": 1.5675583817031747e-06, "loss": 0.4227, "step": 23150 }, { "epoch": 0.8248036054652012, "grad_norm": 1.6036727834758142, "learning_rate": 1.5669381614144685e-06, "loss": 0.4825, "step": 23151 }, { "epoch": 0.8248392325916953, "grad_norm": 1.3117675033025706, "learning_rate": 1.5663180534178724e-06, "loss": 0.4198, "step": 23152 }, { "epoch": 0.8248748597181894, "grad_norm": 1.6108034490754477, "learning_rate": 1.5656980577216464e-06, "loss": 0.4985, "step": 23153 }, { "epoch": 0.8249104868446835, "grad_norm": 1.0971200591396102, "learning_rate": 1.5650781743340449e-06, "loss": 0.412, "step": 23154 }, { "epoch": 0.8249461139711777, "grad_norm": 1.279086859949679, "learning_rate": 1.5644584032633248e-06, "loss": 0.4596, "step": 23155 }, { "epoch": 0.8249817410976718, "grad_norm": 1.6986904877989, "learning_rate": 1.5638387445177384e-06, "loss": 0.5154, "step": 23156 }, { "epoch": 0.8250173682241658, "grad_norm": 1.1485352428219544, "learning_rate": 1.5632191981055334e-06, "loss": 0.3851, "step": 23157 }, { "epoch": 0.82505299535066, "grad_norm": 1.2027903258570545, "learning_rate": 1.5625997640349643e-06, "loss": 0.3965, "step": 23158 }, { "epoch": 0.8250886224771541, "grad_norm": 1.1642641991040426, "learning_rate": 1.5619804423142749e-06, "loss": 0.3839, "step": 23159 }, { "epoch": 0.8251242496036482, "grad_norm": 1.660656744860796, "learning_rate": 1.5613612329517158e-06, "loss": 0.4563, "step": 23160 }, { "epoch": 0.8251598767301423, "grad_norm": 1.1708058689850371, "learning_rate": 1.5607421359555275e-06, "loss": 0.3999, "step": 23161 }, { "epoch": 0.8251955038566364, "grad_norm": 1.6890145541287966, "learning_rate": 1.5601231513339566e-06, "loss": 0.5345, "step": 23162 }, { "epoch": 0.8252311309831305, "grad_norm": 1.0619468760806632, "learning_rate": 1.5595042790952442e-06, "loss": 0.3737, "step": 23163 }, { "epoch": 0.8252667581096247, "grad_norm": 1.2734901671499894, "learning_rate": 1.5588855192476327e-06, "loss": 0.4229, "step": 23164 }, { "epoch": 0.8253023852361188, "grad_norm": 1.4345981047690335, "learning_rate": 1.5582668717993632e-06, "loss": 0.4501, "step": 23165 }, { "epoch": 0.8253380123626128, "grad_norm": 1.4448031351535822, "learning_rate": 1.557648336758668e-06, "loss": 0.4593, "step": 23166 }, { "epoch": 0.825373639489107, "grad_norm": 1.1695219574207867, "learning_rate": 1.5570299141337886e-06, "loss": 0.4018, "step": 23167 }, { "epoch": 0.8254092666156011, "grad_norm": 1.2493423282813125, "learning_rate": 1.5564116039329546e-06, "loss": 0.4234, "step": 23168 }, { "epoch": 0.8254448937420952, "grad_norm": 1.4988745610284062, "learning_rate": 1.5557934061644008e-06, "loss": 0.426, "step": 23169 }, { "epoch": 0.8254805208685894, "grad_norm": 1.6278448538940666, "learning_rate": 1.5551753208363618e-06, "loss": 0.4754, "step": 23170 }, { "epoch": 0.8255161479950834, "grad_norm": 1.2127545737836953, "learning_rate": 1.5545573479570675e-06, "loss": 0.3934, "step": 23171 }, { "epoch": 0.8255517751215776, "grad_norm": 1.1639208525501896, "learning_rate": 1.5539394875347425e-06, "loss": 0.4101, "step": 23172 }, { "epoch": 0.8255874022480717, "grad_norm": 1.5435916644159857, "learning_rate": 1.553321739577619e-06, "loss": 0.5024, "step": 23173 }, { "epoch": 0.8256230293745658, "grad_norm": 1.632271099322004, "learning_rate": 1.5527041040939184e-06, "loss": 0.6248, "step": 23174 }, { "epoch": 0.82565865650106, "grad_norm": 1.646408118208332, "learning_rate": 1.5520865810918673e-06, "loss": 0.5219, "step": 23175 }, { "epoch": 0.825694283627554, "grad_norm": 1.2591876027835551, "learning_rate": 1.551469170579687e-06, "loss": 0.415, "step": 23176 }, { "epoch": 0.8257299107540481, "grad_norm": 1.5199792375172054, "learning_rate": 1.5508518725656007e-06, "loss": 0.4997, "step": 23177 }, { "epoch": 0.8257655378805423, "grad_norm": 1.6692331202742252, "learning_rate": 1.55023468705783e-06, "loss": 0.4329, "step": 23178 }, { "epoch": 0.8258011650070364, "grad_norm": 1.492426784732675, "learning_rate": 1.5496176140645868e-06, "loss": 0.3767, "step": 23179 }, { "epoch": 0.8258367921335305, "grad_norm": 1.2754167979398836, "learning_rate": 1.549000653594095e-06, "loss": 0.4128, "step": 23180 }, { "epoch": 0.8258724192600246, "grad_norm": 1.2070127389798115, "learning_rate": 1.548383805654564e-06, "loss": 0.3795, "step": 23181 }, { "epoch": 0.8259080463865187, "grad_norm": 1.3803066241413395, "learning_rate": 1.5477670702542113e-06, "loss": 0.5358, "step": 23182 }, { "epoch": 0.8259436735130128, "grad_norm": 1.7114564903577796, "learning_rate": 1.5471504474012466e-06, "loss": 0.4781, "step": 23183 }, { "epoch": 0.825979300639507, "grad_norm": 1.1962524505257985, "learning_rate": 1.5465339371038813e-06, "loss": 0.4005, "step": 23184 }, { "epoch": 0.826014927766001, "grad_norm": 1.1041707047101021, "learning_rate": 1.5459175393703263e-06, "loss": 0.3916, "step": 23185 }, { "epoch": 0.8260505548924951, "grad_norm": 1.4446630584085407, "learning_rate": 1.5453012542087898e-06, "loss": 0.5205, "step": 23186 }, { "epoch": 0.8260861820189893, "grad_norm": 1.489337888585349, "learning_rate": 1.544685081627474e-06, "loss": 0.4425, "step": 23187 }, { "epoch": 0.8261218091454834, "grad_norm": 1.132778443335249, "learning_rate": 1.5440690216345866e-06, "loss": 0.4074, "step": 23188 }, { "epoch": 0.8261574362719775, "grad_norm": 1.234850258183415, "learning_rate": 1.5434530742383324e-06, "loss": 0.4259, "step": 23189 }, { "epoch": 0.8261930633984716, "grad_norm": 2.056167956669272, "learning_rate": 1.5428372394469094e-06, "loss": 0.4708, "step": 23190 }, { "epoch": 0.8262286905249657, "grad_norm": 1.440966536073844, "learning_rate": 1.5422215172685194e-06, "loss": 0.4765, "step": 23191 }, { "epoch": 0.8262643176514598, "grad_norm": 1.1210184133579326, "learning_rate": 1.5416059077113621e-06, "loss": 0.3643, "step": 23192 }, { "epoch": 0.826299944777954, "grad_norm": 1.2902926459809547, "learning_rate": 1.540990410783636e-06, "loss": 0.3815, "step": 23193 }, { "epoch": 0.8263355719044481, "grad_norm": 1.6944490877439788, "learning_rate": 1.5403750264935336e-06, "loss": 0.486, "step": 23194 }, { "epoch": 0.8263711990309421, "grad_norm": 1.2595184656746028, "learning_rate": 1.5397597548492515e-06, "loss": 0.3931, "step": 23195 }, { "epoch": 0.8264068261574363, "grad_norm": 1.5752835658404982, "learning_rate": 1.53914459585898e-06, "loss": 0.5015, "step": 23196 }, { "epoch": 0.8264424532839304, "grad_norm": 1.1411419514240717, "learning_rate": 1.5385295495309149e-06, "loss": 0.4605, "step": 23197 }, { "epoch": 0.8264780804104245, "grad_norm": 1.2548292152098688, "learning_rate": 1.5379146158732405e-06, "loss": 0.3982, "step": 23198 }, { "epoch": 0.8265137075369187, "grad_norm": 1.326855863029599, "learning_rate": 1.537299794894147e-06, "loss": 0.3949, "step": 23199 }, { "epoch": 0.8265493346634127, "grad_norm": 1.1958912957818804, "learning_rate": 1.5366850866018224e-06, "loss": 0.4056, "step": 23200 }, { "epoch": 0.8265849617899068, "grad_norm": 1.2867380300604658, "learning_rate": 1.536070491004451e-06, "loss": 0.4022, "step": 23201 }, { "epoch": 0.826620588916401, "grad_norm": 1.1330702425290136, "learning_rate": 1.53545600811022e-06, "loss": 0.3798, "step": 23202 }, { "epoch": 0.8266562160428951, "grad_norm": 1.0972153757279872, "learning_rate": 1.5348416379273057e-06, "loss": 0.4323, "step": 23203 }, { "epoch": 0.8266918431693892, "grad_norm": 1.2237971211619474, "learning_rate": 1.5342273804638942e-06, "loss": 0.4141, "step": 23204 }, { "epoch": 0.8267274702958833, "grad_norm": 1.6575868490681454, "learning_rate": 1.5336132357281598e-06, "loss": 0.4931, "step": 23205 }, { "epoch": 0.8267630974223774, "grad_norm": 1.46790406969558, "learning_rate": 1.5329992037282827e-06, "loss": 0.4696, "step": 23206 }, { "epoch": 0.8267987245488715, "grad_norm": 1.181972261791865, "learning_rate": 1.5323852844724395e-06, "loss": 0.4313, "step": 23207 }, { "epoch": 0.8268343516753657, "grad_norm": 2.6766093276753513, "learning_rate": 1.5317714779688076e-06, "loss": 0.5228, "step": 23208 }, { "epoch": 0.8268699788018598, "grad_norm": 1.4602614841681647, "learning_rate": 1.5311577842255543e-06, "loss": 0.5273, "step": 23209 }, { "epoch": 0.8269056059283538, "grad_norm": 1.6228750176611184, "learning_rate": 1.5305442032508555e-06, "loss": 0.5079, "step": 23210 }, { "epoch": 0.826941233054848, "grad_norm": 1.0819234561527324, "learning_rate": 1.5299307350528813e-06, "loss": 0.3788, "step": 23211 }, { "epoch": 0.8269768601813421, "grad_norm": 1.3029009831512957, "learning_rate": 1.5293173796397987e-06, "loss": 0.4035, "step": 23212 }, { "epoch": 0.8270124873078362, "grad_norm": 1.5477444749105969, "learning_rate": 1.5287041370197775e-06, "loss": 0.5304, "step": 23213 }, { "epoch": 0.8270481144343304, "grad_norm": 1.2127682239892927, "learning_rate": 1.528091007200977e-06, "loss": 0.5189, "step": 23214 }, { "epoch": 0.8270837415608244, "grad_norm": 1.609358444387146, "learning_rate": 1.5274779901915715e-06, "loss": 0.5453, "step": 23215 }, { "epoch": 0.8271193686873185, "grad_norm": 1.5001053380743736, "learning_rate": 1.5268650859997147e-06, "loss": 0.4435, "step": 23216 }, { "epoch": 0.8271549958138127, "grad_norm": 1.1132488446507283, "learning_rate": 1.5262522946335756e-06, "loss": 0.4075, "step": 23217 }, { "epoch": 0.8271906229403068, "grad_norm": 1.285108918958288, "learning_rate": 1.5256396161013077e-06, "loss": 0.4273, "step": 23218 }, { "epoch": 0.8272262500668008, "grad_norm": 1.1918307345866526, "learning_rate": 1.5250270504110732e-06, "loss": 0.4182, "step": 23219 }, { "epoch": 0.827261877193295, "grad_norm": 1.2428817609386569, "learning_rate": 1.5244145975710246e-06, "loss": 0.4012, "step": 23220 }, { "epoch": 0.8272975043197891, "grad_norm": 1.7343102844835094, "learning_rate": 1.5238022575893197e-06, "loss": 0.5113, "step": 23221 }, { "epoch": 0.8273331314462832, "grad_norm": 1.5233133944023112, "learning_rate": 1.523190030474112e-06, "loss": 0.4523, "step": 23222 }, { "epoch": 0.8273687585727774, "grad_norm": 1.2936891605877379, "learning_rate": 1.5225779162335551e-06, "loss": 0.4171, "step": 23223 }, { "epoch": 0.8274043856992714, "grad_norm": 1.1671390036639249, "learning_rate": 1.5219659148758003e-06, "loss": 0.3838, "step": 23224 }, { "epoch": 0.8274400128257655, "grad_norm": 1.3699061891920152, "learning_rate": 1.5213540264089933e-06, "loss": 0.403, "step": 23225 }, { "epoch": 0.8274756399522597, "grad_norm": 1.6038629362636434, "learning_rate": 1.520742250841285e-06, "loss": 0.5113, "step": 23226 }, { "epoch": 0.8275112670787538, "grad_norm": 1.9067329603929766, "learning_rate": 1.520130588180818e-06, "loss": 0.4522, "step": 23227 }, { "epoch": 0.8275468942052479, "grad_norm": 1.3243651531051277, "learning_rate": 1.5195190384357405e-06, "loss": 0.4601, "step": 23228 }, { "epoch": 0.827582521331742, "grad_norm": 1.5171821843457483, "learning_rate": 1.5189076016141935e-06, "loss": 0.5318, "step": 23229 }, { "epoch": 0.8276181484582361, "grad_norm": 1.3939715294261357, "learning_rate": 1.5182962777243227e-06, "loss": 0.3878, "step": 23230 }, { "epoch": 0.8276537755847302, "grad_norm": 1.400558507774866, "learning_rate": 1.5176850667742626e-06, "loss": 0.5622, "step": 23231 }, { "epoch": 0.8276894027112244, "grad_norm": 1.4517078142276332, "learning_rate": 1.517073968772157e-06, "loss": 0.5658, "step": 23232 }, { "epoch": 0.8277250298377185, "grad_norm": 1.5685139417856142, "learning_rate": 1.516462983726138e-06, "loss": 0.5627, "step": 23233 }, { "epoch": 0.8277606569642125, "grad_norm": 1.2789536952879443, "learning_rate": 1.5158521116443447e-06, "loss": 0.3877, "step": 23234 }, { "epoch": 0.8277962840907067, "grad_norm": 1.1455583724450564, "learning_rate": 1.515241352534913e-06, "loss": 0.4215, "step": 23235 }, { "epoch": 0.8278319112172008, "grad_norm": 1.2900802534628033, "learning_rate": 1.5146307064059696e-06, "loss": 0.4455, "step": 23236 }, { "epoch": 0.8278675383436949, "grad_norm": 1.3798670260898942, "learning_rate": 1.5140201732656501e-06, "loss": 0.4052, "step": 23237 }, { "epoch": 0.827903165470189, "grad_norm": 1.3669009898673135, "learning_rate": 1.5134097531220838e-06, "loss": 0.4364, "step": 23238 }, { "epoch": 0.8279387925966831, "grad_norm": 2.7232456714268567, "learning_rate": 1.5127994459834005e-06, "loss": 0.4487, "step": 23239 }, { "epoch": 0.8279744197231772, "grad_norm": 1.5249029653837918, "learning_rate": 1.5121892518577218e-06, "loss": 0.5636, "step": 23240 }, { "epoch": 0.8280100468496714, "grad_norm": 1.2362550887484534, "learning_rate": 1.511579170753179e-06, "loss": 0.3989, "step": 23241 }, { "epoch": 0.8280456739761655, "grad_norm": 1.476961630752818, "learning_rate": 1.5109692026778899e-06, "loss": 0.451, "step": 23242 }, { "epoch": 0.8280813011026595, "grad_norm": 1.1417318129254515, "learning_rate": 1.5103593476399791e-06, "loss": 0.4064, "step": 23243 }, { "epoch": 0.8281169282291537, "grad_norm": 1.493380646492991, "learning_rate": 1.509749605647568e-06, "loss": 0.4046, "step": 23244 }, { "epoch": 0.8281525553556478, "grad_norm": 1.1364927218586331, "learning_rate": 1.5091399767087743e-06, "loss": 0.4506, "step": 23245 }, { "epoch": 0.8281881824821419, "grad_norm": 1.2291021846259051, "learning_rate": 1.5085304608317196e-06, "loss": 0.4631, "step": 23246 }, { "epoch": 0.8282238096086361, "grad_norm": 1.480351602693406, "learning_rate": 1.5079210580245152e-06, "loss": 0.5283, "step": 23247 }, { "epoch": 0.8282594367351301, "grad_norm": 1.3568532351979061, "learning_rate": 1.5073117682952787e-06, "loss": 0.3856, "step": 23248 }, { "epoch": 0.8282950638616242, "grad_norm": 1.6003763910341353, "learning_rate": 1.5067025916521206e-06, "loss": 0.4747, "step": 23249 }, { "epoch": 0.8283306909881184, "grad_norm": 1.1927134247151399, "learning_rate": 1.5060935281031564e-06, "loss": 0.3855, "step": 23250 }, { "epoch": 0.8283663181146125, "grad_norm": 1.5148069706098832, "learning_rate": 1.5054845776564885e-06, "loss": 0.4595, "step": 23251 }, { "epoch": 0.8284019452411066, "grad_norm": 1.187876950131503, "learning_rate": 1.5048757403202353e-06, "loss": 0.3499, "step": 23252 }, { "epoch": 0.8284375723676007, "grad_norm": 1.4127939648073875, "learning_rate": 1.5042670161024975e-06, "loss": 0.4907, "step": 23253 }, { "epoch": 0.8284731994940948, "grad_norm": 1.863503845802675, "learning_rate": 1.5036584050113844e-06, "loss": 0.459, "step": 23254 }, { "epoch": 0.8285088266205889, "grad_norm": 1.5934406073574248, "learning_rate": 1.5030499070549965e-06, "loss": 0.4743, "step": 23255 }, { "epoch": 0.8285444537470831, "grad_norm": 1.351302578735155, "learning_rate": 1.5024415222414369e-06, "loss": 0.5287, "step": 23256 }, { "epoch": 0.8285800808735772, "grad_norm": 1.792571506694148, "learning_rate": 1.5018332505788114e-06, "loss": 0.5104, "step": 23257 }, { "epoch": 0.8286157080000712, "grad_norm": 1.5982591307682137, "learning_rate": 1.5012250920752137e-06, "loss": 0.4857, "step": 23258 }, { "epoch": 0.8286513351265654, "grad_norm": 1.2254282861337222, "learning_rate": 1.5006170467387437e-06, "loss": 0.4387, "step": 23259 }, { "epoch": 0.8286869622530595, "grad_norm": 1.7428089488434084, "learning_rate": 1.5000091145774987e-06, "loss": 0.5343, "step": 23260 }, { "epoch": 0.8287225893795536, "grad_norm": 1.3439048347295373, "learning_rate": 1.4994012955995752e-06, "loss": 0.4488, "step": 23261 }, { "epoch": 0.8287582165060478, "grad_norm": 1.3074521423316807, "learning_rate": 1.4987935898130634e-06, "loss": 0.3976, "step": 23262 }, { "epoch": 0.8287938436325418, "grad_norm": 1.172967279318878, "learning_rate": 1.4981859972260594e-06, "loss": 0.4097, "step": 23263 }, { "epoch": 0.8288294707590359, "grad_norm": 1.9237373730579177, "learning_rate": 1.4975785178466484e-06, "loss": 0.5224, "step": 23264 }, { "epoch": 0.8288650978855301, "grad_norm": 1.241460549539925, "learning_rate": 1.496971151682922e-06, "loss": 0.4446, "step": 23265 }, { "epoch": 0.8289007250120242, "grad_norm": 1.7700933150160347, "learning_rate": 1.4963638987429685e-06, "loss": 0.5217, "step": 23266 }, { "epoch": 0.8289363521385182, "grad_norm": 1.2303866700825994, "learning_rate": 1.495756759034873e-06, "loss": 0.3997, "step": 23267 }, { "epoch": 0.8289719792650124, "grad_norm": 1.1822750496636738, "learning_rate": 1.495149732566723e-06, "loss": 0.4218, "step": 23268 }, { "epoch": 0.8290076063915065, "grad_norm": 1.2633779929859792, "learning_rate": 1.4945428193465972e-06, "loss": 0.4048, "step": 23269 }, { "epoch": 0.8290432335180006, "grad_norm": 1.839264121802928, "learning_rate": 1.4939360193825803e-06, "loss": 0.5163, "step": 23270 }, { "epoch": 0.8290788606444948, "grad_norm": 1.3340263511915476, "learning_rate": 1.493329332682748e-06, "loss": 0.3711, "step": 23271 }, { "epoch": 0.8291144877709888, "grad_norm": 1.1215817118237992, "learning_rate": 1.4927227592551841e-06, "loss": 0.3623, "step": 23272 }, { "epoch": 0.8291501148974829, "grad_norm": 1.1949896151477517, "learning_rate": 1.4921162991079607e-06, "loss": 0.3867, "step": 23273 }, { "epoch": 0.8291857420239771, "grad_norm": 1.10866513809912, "learning_rate": 1.4915099522491561e-06, "loss": 0.4166, "step": 23274 }, { "epoch": 0.8292213691504712, "grad_norm": 1.1779810712639862, "learning_rate": 1.4909037186868424e-06, "loss": 0.3816, "step": 23275 }, { "epoch": 0.8292569962769653, "grad_norm": 1.5798846136603688, "learning_rate": 1.4902975984290958e-06, "loss": 0.5089, "step": 23276 }, { "epoch": 0.8292926234034594, "grad_norm": 1.613155749703154, "learning_rate": 1.4896915914839828e-06, "loss": 0.5322, "step": 23277 }, { "epoch": 0.8293282505299535, "grad_norm": 1.2734946367350697, "learning_rate": 1.4890856978595768e-06, "loss": 0.4473, "step": 23278 }, { "epoch": 0.8293638776564476, "grad_norm": 1.4381768223785039, "learning_rate": 1.4884799175639409e-06, "loss": 0.4544, "step": 23279 }, { "epoch": 0.8293995047829418, "grad_norm": 1.2607378986771423, "learning_rate": 1.4878742506051436e-06, "loss": 0.4383, "step": 23280 }, { "epoch": 0.8294351319094359, "grad_norm": 1.1712693113262187, "learning_rate": 1.4872686969912508e-06, "loss": 0.433, "step": 23281 }, { "epoch": 0.8294707590359299, "grad_norm": 1.2251008898566083, "learning_rate": 1.4866632567303252e-06, "loss": 0.4158, "step": 23282 }, { "epoch": 0.8295063861624241, "grad_norm": 1.260720083384309, "learning_rate": 1.4860579298304311e-06, "loss": 0.4565, "step": 23283 }, { "epoch": 0.8295420132889182, "grad_norm": 1.7603129658955008, "learning_rate": 1.4854527162996246e-06, "loss": 0.5392, "step": 23284 }, { "epoch": 0.8295776404154123, "grad_norm": 1.5124453457904754, "learning_rate": 1.484847616145969e-06, "loss": 0.5811, "step": 23285 }, { "epoch": 0.8296132675419065, "grad_norm": 1.1526641250068639, "learning_rate": 1.484242629377517e-06, "loss": 0.367, "step": 23286 }, { "epoch": 0.8296488946684005, "grad_norm": 1.7726931824106866, "learning_rate": 1.4836377560023275e-06, "loss": 0.5214, "step": 23287 }, { "epoch": 0.8296845217948946, "grad_norm": 1.1781833042275833, "learning_rate": 1.4830329960284528e-06, "loss": 0.4167, "step": 23288 }, { "epoch": 0.8297201489213888, "grad_norm": 1.2368153742263757, "learning_rate": 1.4824283494639458e-06, "loss": 0.4377, "step": 23289 }, { "epoch": 0.8297557760478829, "grad_norm": 1.3852334591814393, "learning_rate": 1.4818238163168597e-06, "loss": 0.5129, "step": 23290 }, { "epoch": 0.8297914031743769, "grad_norm": 1.7427830367939634, "learning_rate": 1.4812193965952426e-06, "loss": 0.5452, "step": 23291 }, { "epoch": 0.8298270303008711, "grad_norm": 1.778041388265387, "learning_rate": 1.480615090307146e-06, "loss": 0.4764, "step": 23292 }, { "epoch": 0.8298626574273652, "grad_norm": 1.6299498571814457, "learning_rate": 1.480010897460612e-06, "loss": 0.4993, "step": 23293 }, { "epoch": 0.8298982845538593, "grad_norm": 1.5820641744049753, "learning_rate": 1.479406818063691e-06, "loss": 0.4847, "step": 23294 }, { "epoch": 0.8299339116803535, "grad_norm": 1.1008356108847486, "learning_rate": 1.4788028521244201e-06, "loss": 0.4108, "step": 23295 }, { "epoch": 0.8299695388068475, "grad_norm": 1.4751067506794358, "learning_rate": 1.478198999650846e-06, "loss": 0.4536, "step": 23296 }, { "epoch": 0.8300051659333416, "grad_norm": 1.2561808087986157, "learning_rate": 1.477595260651008e-06, "loss": 0.4872, "step": 23297 }, { "epoch": 0.8300407930598358, "grad_norm": 1.1063786801406212, "learning_rate": 1.4769916351329495e-06, "loss": 0.388, "step": 23298 }, { "epoch": 0.8300764201863299, "grad_norm": 1.7625514700561542, "learning_rate": 1.4763881231047007e-06, "loss": 0.4763, "step": 23299 }, { "epoch": 0.830112047312824, "grad_norm": 1.58992503856525, "learning_rate": 1.475784724574305e-06, "loss": 0.5131, "step": 23300 }, { "epoch": 0.8301476744393181, "grad_norm": 1.2007417056059138, "learning_rate": 1.475181439549791e-06, "loss": 0.439, "step": 23301 }, { "epoch": 0.8301833015658122, "grad_norm": 1.2137263963872484, "learning_rate": 1.4745782680391952e-06, "loss": 0.3867, "step": 23302 }, { "epoch": 0.8302189286923063, "grad_norm": 1.1780154219052879, "learning_rate": 1.4739752100505478e-06, "loss": 0.3985, "step": 23303 }, { "epoch": 0.8302545558188005, "grad_norm": 1.5859342271624577, "learning_rate": 1.4733722655918804e-06, "loss": 0.5047, "step": 23304 }, { "epoch": 0.8302901829452946, "grad_norm": 1.1568839815349579, "learning_rate": 1.472769434671224e-06, "loss": 0.3778, "step": 23305 }, { "epoch": 0.8303258100717886, "grad_norm": 1.16701214548538, "learning_rate": 1.4721667172965993e-06, "loss": 0.4022, "step": 23306 }, { "epoch": 0.8303614371982828, "grad_norm": 1.0643926845811593, "learning_rate": 1.4715641134760384e-06, "loss": 0.3958, "step": 23307 }, { "epoch": 0.8303970643247769, "grad_norm": 1.2555092515822057, "learning_rate": 1.4709616232175593e-06, "loss": 0.3973, "step": 23308 }, { "epoch": 0.830432691451271, "grad_norm": 1.2576229579056577, "learning_rate": 1.4703592465291905e-06, "loss": 0.4155, "step": 23309 }, { "epoch": 0.8304683185777652, "grad_norm": 1.4914436509612845, "learning_rate": 1.4697569834189474e-06, "loss": 0.4807, "step": 23310 }, { "epoch": 0.8305039457042592, "grad_norm": 1.6437498604408167, "learning_rate": 1.4691548338948537e-06, "loss": 0.6046, "step": 23311 }, { "epoch": 0.8305395728307533, "grad_norm": 1.1479224150268654, "learning_rate": 1.4685527979649238e-06, "loss": 0.4393, "step": 23312 }, { "epoch": 0.8305751999572475, "grad_norm": 1.2561338448201527, "learning_rate": 1.4679508756371807e-06, "loss": 0.4662, "step": 23313 }, { "epoch": 0.8306108270837416, "grad_norm": 1.1583100279566174, "learning_rate": 1.4673490669196322e-06, "loss": 0.3666, "step": 23314 }, { "epoch": 0.8306464542102356, "grad_norm": 1.450404779049364, "learning_rate": 1.4667473718202952e-06, "loss": 0.4749, "step": 23315 }, { "epoch": 0.8306820813367298, "grad_norm": 1.1636233610823221, "learning_rate": 1.4661457903471833e-06, "loss": 0.4413, "step": 23316 }, { "epoch": 0.8307177084632239, "grad_norm": 1.125538892752245, "learning_rate": 1.4655443225083032e-06, "loss": 0.3573, "step": 23317 }, { "epoch": 0.830753335589718, "grad_norm": 1.7115499919207047, "learning_rate": 1.4649429683116645e-06, "loss": 0.473, "step": 23318 }, { "epoch": 0.8307889627162122, "grad_norm": 1.457569371378506, "learning_rate": 1.464341727765277e-06, "loss": 0.4874, "step": 23319 }, { "epoch": 0.8308245898427062, "grad_norm": 1.689775210880217, "learning_rate": 1.463740600877147e-06, "loss": 0.4489, "step": 23320 }, { "epoch": 0.8308602169692003, "grad_norm": 1.5622577206292738, "learning_rate": 1.4631395876552756e-06, "loss": 0.5765, "step": 23321 }, { "epoch": 0.8308958440956945, "grad_norm": 1.5438630268246736, "learning_rate": 1.4625386881076697e-06, "loss": 0.5236, "step": 23322 }, { "epoch": 0.8309314712221886, "grad_norm": 1.62050766232277, "learning_rate": 1.4619379022423264e-06, "loss": 0.4799, "step": 23323 }, { "epoch": 0.8309670983486827, "grad_norm": 1.4783196245157164, "learning_rate": 1.4613372300672502e-06, "loss": 0.4956, "step": 23324 }, { "epoch": 0.8310027254751768, "grad_norm": 1.2323228783764715, "learning_rate": 1.4607366715904336e-06, "loss": 0.4184, "step": 23325 }, { "epoch": 0.8310383526016709, "grad_norm": 1.1003261624068565, "learning_rate": 1.4601362268198749e-06, "loss": 0.3891, "step": 23326 }, { "epoch": 0.831073979728165, "grad_norm": 1.0565873629143292, "learning_rate": 1.4595358957635763e-06, "loss": 0.3913, "step": 23327 }, { "epoch": 0.8311096068546592, "grad_norm": 1.2663418421715043, "learning_rate": 1.458935678429525e-06, "loss": 0.4282, "step": 23328 }, { "epoch": 0.8311452339811533, "grad_norm": 1.5190374824573123, "learning_rate": 1.4583355748257156e-06, "loss": 0.5479, "step": 23329 }, { "epoch": 0.8311808611076473, "grad_norm": 1.6765846302998964, "learning_rate": 1.4577355849601372e-06, "loss": 0.4821, "step": 23330 }, { "epoch": 0.8312164882341415, "grad_norm": 1.4020231513728871, "learning_rate": 1.4571357088407822e-06, "loss": 0.4621, "step": 23331 }, { "epoch": 0.8312521153606356, "grad_norm": 2.050592714112613, "learning_rate": 1.4565359464756345e-06, "loss": 0.5742, "step": 23332 }, { "epoch": 0.8312877424871297, "grad_norm": 1.2345934718428258, "learning_rate": 1.455936297872681e-06, "loss": 0.402, "step": 23333 }, { "epoch": 0.8313233696136239, "grad_norm": 1.4144302456161406, "learning_rate": 1.4553367630399085e-06, "loss": 0.4831, "step": 23334 }, { "epoch": 0.8313589967401179, "grad_norm": 1.4967699728231376, "learning_rate": 1.4547373419853016e-06, "loss": 0.5141, "step": 23335 }, { "epoch": 0.831394623866612, "grad_norm": 1.5801845542430022, "learning_rate": 1.4541380347168377e-06, "loss": 0.5489, "step": 23336 }, { "epoch": 0.8314302509931062, "grad_norm": 1.709765097602196, "learning_rate": 1.4535388412424989e-06, "loss": 0.554, "step": 23337 }, { "epoch": 0.8314658781196003, "grad_norm": 1.6941497013252442, "learning_rate": 1.4529397615702656e-06, "loss": 0.5419, "step": 23338 }, { "epoch": 0.8315015052460943, "grad_norm": 1.626348039205537, "learning_rate": 1.4523407957081114e-06, "loss": 0.5509, "step": 23339 }, { "epoch": 0.8315371323725885, "grad_norm": 1.4590737172182269, "learning_rate": 1.4517419436640145e-06, "loss": 0.5289, "step": 23340 }, { "epoch": 0.8315727594990826, "grad_norm": 1.339142533175173, "learning_rate": 1.4511432054459485e-06, "loss": 0.4136, "step": 23341 }, { "epoch": 0.8316083866255767, "grad_norm": 1.2102418586058767, "learning_rate": 1.4505445810618878e-06, "loss": 0.4185, "step": 23342 }, { "epoch": 0.8316440137520709, "grad_norm": 1.049358111877244, "learning_rate": 1.4499460705198e-06, "loss": 0.4128, "step": 23343 }, { "epoch": 0.8316796408785649, "grad_norm": 2.217686985115404, "learning_rate": 1.449347673827659e-06, "loss": 0.57, "step": 23344 }, { "epoch": 0.831715268005059, "grad_norm": 1.2193133615996323, "learning_rate": 1.448749390993427e-06, "loss": 0.4058, "step": 23345 }, { "epoch": 0.8317508951315532, "grad_norm": 1.5869697698424723, "learning_rate": 1.448151222025076e-06, "loss": 0.5204, "step": 23346 }, { "epoch": 0.8317865222580473, "grad_norm": 1.0177148129531004, "learning_rate": 1.4475531669305676e-06, "loss": 0.3726, "step": 23347 }, { "epoch": 0.8318221493845414, "grad_norm": 1.1542677097374265, "learning_rate": 1.4469552257178655e-06, "loss": 0.3959, "step": 23348 }, { "epoch": 0.8318577765110355, "grad_norm": 1.2321442811357162, "learning_rate": 1.446357398394934e-06, "loss": 0.4271, "step": 23349 }, { "epoch": 0.8318934036375296, "grad_norm": 1.787685669652776, "learning_rate": 1.4457596849697319e-06, "loss": 0.5178, "step": 23350 }, { "epoch": 0.8319290307640237, "grad_norm": 1.1415467640510375, "learning_rate": 1.4451620854502202e-06, "loss": 0.4001, "step": 23351 }, { "epoch": 0.8319646578905179, "grad_norm": 1.5225311642999257, "learning_rate": 1.4445645998443537e-06, "loss": 0.5507, "step": 23352 }, { "epoch": 0.832000285017012, "grad_norm": 1.2431408366153283, "learning_rate": 1.4439672281600903e-06, "loss": 0.4405, "step": 23353 }, { "epoch": 0.832035912143506, "grad_norm": 2.9139741351385124, "learning_rate": 1.4433699704053827e-06, "loss": 0.4458, "step": 23354 }, { "epoch": 0.8320715392700002, "grad_norm": 1.2134499146553905, "learning_rate": 1.4427728265881836e-06, "loss": 0.3985, "step": 23355 }, { "epoch": 0.8321071663964943, "grad_norm": 1.3623240532955572, "learning_rate": 1.4421757967164463e-06, "loss": 0.4199, "step": 23356 }, { "epoch": 0.8321427935229884, "grad_norm": 1.2865314689890848, "learning_rate": 1.4415788807981212e-06, "loss": 0.3501, "step": 23357 }, { "epoch": 0.8321784206494826, "grad_norm": 1.7613547028055618, "learning_rate": 1.4409820788411544e-06, "loss": 0.5124, "step": 23358 }, { "epoch": 0.8322140477759766, "grad_norm": 1.3375381522729244, "learning_rate": 1.440385390853495e-06, "loss": 0.4461, "step": 23359 }, { "epoch": 0.8322496749024707, "grad_norm": 1.3481216758747954, "learning_rate": 1.4397888168430841e-06, "loss": 0.4154, "step": 23360 }, { "epoch": 0.8322853020289649, "grad_norm": 1.2295158793957615, "learning_rate": 1.439192356817869e-06, "loss": 0.3405, "step": 23361 }, { "epoch": 0.832320929155459, "grad_norm": 1.3209063247672286, "learning_rate": 1.4385960107857933e-06, "loss": 0.416, "step": 23362 }, { "epoch": 0.832356556281953, "grad_norm": 1.142869395056001, "learning_rate": 1.4379997787547939e-06, "loss": 0.3788, "step": 23363 }, { "epoch": 0.8323921834084472, "grad_norm": 1.5025120449751712, "learning_rate": 1.4374036607328113e-06, "loss": 0.6039, "step": 23364 }, { "epoch": 0.8324278105349413, "grad_norm": 1.2535075778586986, "learning_rate": 1.4368076567277845e-06, "loss": 0.38, "step": 23365 }, { "epoch": 0.8324634376614354, "grad_norm": 1.557066688301154, "learning_rate": 1.4362117667476517e-06, "loss": 0.5495, "step": 23366 }, { "epoch": 0.8324990647879296, "grad_norm": 1.16207424086705, "learning_rate": 1.435615990800342e-06, "loss": 0.3903, "step": 23367 }, { "epoch": 0.8325346919144236, "grad_norm": 1.2503615785278557, "learning_rate": 1.4350203288937937e-06, "loss": 0.4345, "step": 23368 }, { "epoch": 0.8325703190409177, "grad_norm": 1.2124677666638677, "learning_rate": 1.4344247810359347e-06, "loss": 0.4296, "step": 23369 }, { "epoch": 0.8326059461674119, "grad_norm": 1.6712657602073928, "learning_rate": 1.4338293472346954e-06, "loss": 0.4848, "step": 23370 }, { "epoch": 0.832641573293906, "grad_norm": 1.8709751127896908, "learning_rate": 1.4332340274980073e-06, "loss": 0.5782, "step": 23371 }, { "epoch": 0.8326772004204, "grad_norm": 1.0964841477657459, "learning_rate": 1.432638821833795e-06, "loss": 0.3766, "step": 23372 }, { "epoch": 0.8327128275468942, "grad_norm": 1.591989747551255, "learning_rate": 1.4320437302499878e-06, "loss": 0.5507, "step": 23373 }, { "epoch": 0.8327484546733883, "grad_norm": 1.2717424133116157, "learning_rate": 1.431448752754504e-06, "loss": 0.4813, "step": 23374 }, { "epoch": 0.8327840817998824, "grad_norm": 1.5145672689052059, "learning_rate": 1.4308538893552715e-06, "loss": 0.526, "step": 23375 }, { "epoch": 0.8328197089263766, "grad_norm": 1.5251483500907093, "learning_rate": 1.4302591400602072e-06, "loss": 0.4973, "step": 23376 }, { "epoch": 0.8328553360528707, "grad_norm": 1.4002303622151158, "learning_rate": 1.429664504877234e-06, "loss": 0.4434, "step": 23377 }, { "epoch": 0.8328909631793647, "grad_norm": 1.4672807909006556, "learning_rate": 1.429069983814264e-06, "loss": 0.532, "step": 23378 }, { "epoch": 0.8329265903058589, "grad_norm": 2.0703195056620918, "learning_rate": 1.4284755768792224e-06, "loss": 0.5611, "step": 23379 }, { "epoch": 0.832962217432353, "grad_norm": 1.2250841561056625, "learning_rate": 1.4278812840800184e-06, "loss": 0.3916, "step": 23380 }, { "epoch": 0.8329978445588471, "grad_norm": 1.752091051603169, "learning_rate": 1.4272871054245674e-06, "loss": 0.4731, "step": 23381 }, { "epoch": 0.8330334716853413, "grad_norm": 1.1663399377342945, "learning_rate": 1.426693040920779e-06, "loss": 0.4051, "step": 23382 }, { "epoch": 0.8330690988118353, "grad_norm": 1.2809531936949656, "learning_rate": 1.4260990905765648e-06, "loss": 0.4472, "step": 23383 }, { "epoch": 0.8331047259383294, "grad_norm": 1.383763961744121, "learning_rate": 1.4255052543998372e-06, "loss": 0.5455, "step": 23384 }, { "epoch": 0.8331403530648236, "grad_norm": 1.3006568963164666, "learning_rate": 1.4249115323984975e-06, "loss": 0.4233, "step": 23385 }, { "epoch": 0.8331759801913177, "grad_norm": 1.5387387407421607, "learning_rate": 1.424317924580454e-06, "loss": 0.5297, "step": 23386 }, { "epoch": 0.8332116073178117, "grad_norm": 1.5342121483664064, "learning_rate": 1.4237244309536114e-06, "loss": 0.5237, "step": 23387 }, { "epoch": 0.8332472344443059, "grad_norm": 1.2927900085350288, "learning_rate": 1.4231310515258745e-06, "loss": 0.461, "step": 23388 }, { "epoch": 0.8332828615708, "grad_norm": 1.519532368466669, "learning_rate": 1.4225377863051415e-06, "loss": 0.4988, "step": 23389 }, { "epoch": 0.8333184886972941, "grad_norm": 1.5335872932165096, "learning_rate": 1.4219446352993138e-06, "loss": 0.6198, "step": 23390 }, { "epoch": 0.8333541158237883, "grad_norm": 1.6659587048950864, "learning_rate": 1.4213515985162874e-06, "loss": 0.457, "step": 23391 }, { "epoch": 0.8333897429502823, "grad_norm": 1.0873391716952618, "learning_rate": 1.4207586759639613e-06, "loss": 0.3396, "step": 23392 }, { "epoch": 0.8334253700767764, "grad_norm": 1.3017701877658487, "learning_rate": 1.4201658676502294e-06, "loss": 0.4422, "step": 23393 }, { "epoch": 0.8334609972032706, "grad_norm": 1.5977778634312192, "learning_rate": 1.4195731735829854e-06, "loss": 0.4979, "step": 23394 }, { "epoch": 0.8334966243297647, "grad_norm": 1.5880592075384325, "learning_rate": 1.418980593770125e-06, "loss": 0.4994, "step": 23395 }, { "epoch": 0.8335322514562588, "grad_norm": 1.3507551684427936, "learning_rate": 1.4183881282195333e-06, "loss": 0.4285, "step": 23396 }, { "epoch": 0.8335678785827529, "grad_norm": 1.2510588587739797, "learning_rate": 1.4177957769391037e-06, "loss": 0.4184, "step": 23397 }, { "epoch": 0.833603505709247, "grad_norm": 1.1664158043968507, "learning_rate": 1.4172035399367202e-06, "loss": 0.4, "step": 23398 }, { "epoch": 0.8336391328357411, "grad_norm": 1.2035466312415173, "learning_rate": 1.4166114172202728e-06, "loss": 0.4173, "step": 23399 }, { "epoch": 0.8336747599622353, "grad_norm": 1.1779116134322443, "learning_rate": 1.416019408797642e-06, "loss": 0.4621, "step": 23400 }, { "epoch": 0.8337103870887294, "grad_norm": 1.2641220157807294, "learning_rate": 1.4154275146767117e-06, "loss": 0.4144, "step": 23401 }, { "epoch": 0.8337460142152234, "grad_norm": 1.5229890829828931, "learning_rate": 1.414835734865364e-06, "loss": 0.465, "step": 23402 }, { "epoch": 0.8337816413417176, "grad_norm": 1.1790143122599455, "learning_rate": 1.414244069371482e-06, "loss": 0.4203, "step": 23403 }, { "epoch": 0.8338172684682117, "grad_norm": 1.78413999420345, "learning_rate": 1.4136525182029392e-06, "loss": 0.512, "step": 23404 }, { "epoch": 0.8338528955947058, "grad_norm": 1.3675314511128207, "learning_rate": 1.4130610813676172e-06, "loss": 0.4396, "step": 23405 }, { "epoch": 0.8338885227212, "grad_norm": 1.2807974871480534, "learning_rate": 1.412469758873386e-06, "loss": 0.4225, "step": 23406 }, { "epoch": 0.833924149847694, "grad_norm": 1.562551125582308, "learning_rate": 1.411878550728123e-06, "loss": 0.5949, "step": 23407 }, { "epoch": 0.8339597769741881, "grad_norm": 1.4355370210829992, "learning_rate": 1.4112874569396983e-06, "loss": 0.5532, "step": 23408 }, { "epoch": 0.8339954041006823, "grad_norm": 1.2934659168500766, "learning_rate": 1.410696477515986e-06, "loss": 0.3864, "step": 23409 }, { "epoch": 0.8340310312271764, "grad_norm": 1.6455790816644311, "learning_rate": 1.410105612464855e-06, "loss": 0.4974, "step": 23410 }, { "epoch": 0.8340666583536704, "grad_norm": 1.6164128769835644, "learning_rate": 1.409514861794169e-06, "loss": 0.6056, "step": 23411 }, { "epoch": 0.8341022854801646, "grad_norm": 1.2799311979510928, "learning_rate": 1.4089242255118007e-06, "loss": 0.4263, "step": 23412 }, { "epoch": 0.8341379126066587, "grad_norm": 1.1335358021034567, "learning_rate": 1.4083337036256072e-06, "loss": 0.4104, "step": 23413 }, { "epoch": 0.8341735397331528, "grad_norm": 1.2020207409689865, "learning_rate": 1.407743296143459e-06, "loss": 0.3873, "step": 23414 }, { "epoch": 0.834209166859647, "grad_norm": 1.1508671865086355, "learning_rate": 1.4071530030732094e-06, "loss": 0.4426, "step": 23415 }, { "epoch": 0.834244793986141, "grad_norm": 1.2111215343427555, "learning_rate": 1.4065628244227292e-06, "loss": 0.3896, "step": 23416 }, { "epoch": 0.8342804211126351, "grad_norm": 1.6455384563219932, "learning_rate": 1.4059727601998685e-06, "loss": 0.524, "step": 23417 }, { "epoch": 0.8343160482391293, "grad_norm": 1.4603376538151371, "learning_rate": 1.4053828104124867e-06, "loss": 0.4238, "step": 23418 }, { "epoch": 0.8343516753656234, "grad_norm": 1.4176308197252123, "learning_rate": 1.4047929750684431e-06, "loss": 0.5302, "step": 23419 }, { "epoch": 0.8343873024921175, "grad_norm": 1.497378073687611, "learning_rate": 1.4042032541755867e-06, "loss": 0.486, "step": 23420 }, { "epoch": 0.8344229296186116, "grad_norm": 1.347325557939335, "learning_rate": 1.403613647741774e-06, "loss": 0.4549, "step": 23421 }, { "epoch": 0.8344585567451057, "grad_norm": 1.8859823013415797, "learning_rate": 1.4030241557748514e-06, "loss": 0.4675, "step": 23422 }, { "epoch": 0.8344941838715998, "grad_norm": 1.1740057685125445, "learning_rate": 1.4024347782826708e-06, "loss": 0.4077, "step": 23423 }, { "epoch": 0.834529810998094, "grad_norm": 1.49885986378755, "learning_rate": 1.4018455152730814e-06, "loss": 0.5328, "step": 23424 }, { "epoch": 0.834565438124588, "grad_norm": 1.3590149303935297, "learning_rate": 1.4012563667539302e-06, "loss": 0.4205, "step": 23425 }, { "epoch": 0.8346010652510821, "grad_norm": 1.6487673126438596, "learning_rate": 1.4006673327330588e-06, "loss": 0.5091, "step": 23426 }, { "epoch": 0.8346366923775763, "grad_norm": 1.603817067623387, "learning_rate": 1.4000784132183154e-06, "loss": 0.6098, "step": 23427 }, { "epoch": 0.8346723195040704, "grad_norm": 1.474006701082223, "learning_rate": 1.3994896082175357e-06, "loss": 0.4436, "step": 23428 }, { "epoch": 0.8347079466305645, "grad_norm": 1.2038439778274077, "learning_rate": 1.3989009177385637e-06, "loss": 0.4006, "step": 23429 }, { "epoch": 0.8347435737570587, "grad_norm": 1.081874100256693, "learning_rate": 1.3983123417892376e-06, "loss": 0.342, "step": 23430 }, { "epoch": 0.8347792008835527, "grad_norm": 1.2168433249731458, "learning_rate": 1.397723880377395e-06, "loss": 0.3689, "step": 23431 }, { "epoch": 0.8348148280100468, "grad_norm": 1.7920455478848134, "learning_rate": 1.397135533510874e-06, "loss": 0.5191, "step": 23432 }, { "epoch": 0.834850455136541, "grad_norm": 1.2199833380786471, "learning_rate": 1.396547301197504e-06, "loss": 0.39, "step": 23433 }, { "epoch": 0.8348860822630351, "grad_norm": 1.1738747075115032, "learning_rate": 1.3959591834451225e-06, "loss": 0.3623, "step": 23434 }, { "epoch": 0.8349217093895291, "grad_norm": 1.511180181220471, "learning_rate": 1.3953711802615556e-06, "loss": 0.5409, "step": 23435 }, { "epoch": 0.8349573365160233, "grad_norm": 1.5117776294859562, "learning_rate": 1.3947832916546377e-06, "loss": 0.461, "step": 23436 }, { "epoch": 0.8349929636425174, "grad_norm": 1.3858707469130571, "learning_rate": 1.394195517632193e-06, "loss": 0.4682, "step": 23437 }, { "epoch": 0.8350285907690115, "grad_norm": 1.2527721950880284, "learning_rate": 1.3936078582020506e-06, "loss": 0.4217, "step": 23438 }, { "epoch": 0.8350642178955057, "grad_norm": 1.245483337999967, "learning_rate": 1.3930203133720354e-06, "loss": 0.3951, "step": 23439 }, { "epoch": 0.8350998450219997, "grad_norm": 1.5079504182637364, "learning_rate": 1.3924328831499722e-06, "loss": 0.4872, "step": 23440 }, { "epoch": 0.8351354721484938, "grad_norm": 1.0855942309952649, "learning_rate": 1.3918455675436793e-06, "loss": 0.3806, "step": 23441 }, { "epoch": 0.835171099274988, "grad_norm": 1.1444259159924377, "learning_rate": 1.3912583665609792e-06, "loss": 0.4084, "step": 23442 }, { "epoch": 0.8352067264014821, "grad_norm": 1.6496725691605736, "learning_rate": 1.3906712802096933e-06, "loss": 0.4784, "step": 23443 }, { "epoch": 0.8352423535279762, "grad_norm": 1.1115593499535301, "learning_rate": 1.3900843084976357e-06, "loss": 0.3761, "step": 23444 }, { "epoch": 0.8352779806544703, "grad_norm": 1.2879237178103164, "learning_rate": 1.389497451432622e-06, "loss": 0.4126, "step": 23445 }, { "epoch": 0.8353136077809644, "grad_norm": 1.1811229974791035, "learning_rate": 1.3889107090224696e-06, "loss": 0.3986, "step": 23446 }, { "epoch": 0.8353492349074585, "grad_norm": 1.2448214576634873, "learning_rate": 1.388324081274991e-06, "loss": 0.4395, "step": 23447 }, { "epoch": 0.8353848620339527, "grad_norm": 2.848270527619677, "learning_rate": 1.3877375681979944e-06, "loss": 0.5203, "step": 23448 }, { "epoch": 0.8354204891604468, "grad_norm": 1.5548982260632023, "learning_rate": 1.3871511697992946e-06, "loss": 0.449, "step": 23449 }, { "epoch": 0.8354561162869408, "grad_norm": 1.7095308895336332, "learning_rate": 1.3865648860866943e-06, "loss": 0.509, "step": 23450 }, { "epoch": 0.835491743413435, "grad_norm": 1.2181115039453259, "learning_rate": 1.385978717068005e-06, "loss": 0.3778, "step": 23451 }, { "epoch": 0.8355273705399291, "grad_norm": 1.426538257663306, "learning_rate": 1.3853926627510283e-06, "loss": 0.5251, "step": 23452 }, { "epoch": 0.8355629976664232, "grad_norm": 1.1861449781519424, "learning_rate": 1.384806723143567e-06, "loss": 0.4069, "step": 23453 }, { "epoch": 0.8355986247929174, "grad_norm": 1.498529763270231, "learning_rate": 1.38422089825343e-06, "loss": 0.5081, "step": 23454 }, { "epoch": 0.8356342519194114, "grad_norm": 1.5008106512905173, "learning_rate": 1.3836351880884113e-06, "loss": 0.5063, "step": 23455 }, { "epoch": 0.8356698790459055, "grad_norm": 1.4398664916946202, "learning_rate": 1.383049592656316e-06, "loss": 0.4792, "step": 23456 }, { "epoch": 0.8357055061723997, "grad_norm": 1.818527493101006, "learning_rate": 1.382464111964935e-06, "loss": 0.5324, "step": 23457 }, { "epoch": 0.8357411332988938, "grad_norm": 1.5059786880426638, "learning_rate": 1.3818787460220695e-06, "loss": 0.5216, "step": 23458 }, { "epoch": 0.8357767604253878, "grad_norm": 1.4816199611497956, "learning_rate": 1.3812934948355117e-06, "loss": 0.5666, "step": 23459 }, { "epoch": 0.835812387551882, "grad_norm": 1.7709115057001708, "learning_rate": 1.380708358413053e-06, "loss": 0.5012, "step": 23460 }, { "epoch": 0.8358480146783761, "grad_norm": 1.8484852985906113, "learning_rate": 1.380123336762489e-06, "loss": 0.5541, "step": 23461 }, { "epoch": 0.8358836418048702, "grad_norm": 1.2710258468591775, "learning_rate": 1.3795384298916092e-06, "loss": 0.4183, "step": 23462 }, { "epoch": 0.8359192689313644, "grad_norm": 1.3630006762065416, "learning_rate": 1.378953637808198e-06, "loss": 0.4714, "step": 23463 }, { "epoch": 0.8359548960578584, "grad_norm": 1.51948301530657, "learning_rate": 1.3783689605200446e-06, "loss": 0.5204, "step": 23464 }, { "epoch": 0.8359905231843525, "grad_norm": 1.2054561896930718, "learning_rate": 1.377784398034937e-06, "loss": 0.4034, "step": 23465 }, { "epoch": 0.8360261503108467, "grad_norm": 1.7345110469988159, "learning_rate": 1.3771999503606547e-06, "loss": 0.5224, "step": 23466 }, { "epoch": 0.8360617774373408, "grad_norm": 1.184504934237753, "learning_rate": 1.3766156175049827e-06, "loss": 0.3972, "step": 23467 }, { "epoch": 0.8360974045638349, "grad_norm": 1.1684030367088436, "learning_rate": 1.3760313994757002e-06, "loss": 0.427, "step": 23468 }, { "epoch": 0.836133031690329, "grad_norm": 1.1643647392866336, "learning_rate": 1.3754472962805899e-06, "loss": 0.4086, "step": 23469 }, { "epoch": 0.8361686588168231, "grad_norm": 1.5291628216399409, "learning_rate": 1.3748633079274254e-06, "loss": 0.4753, "step": 23470 }, { "epoch": 0.8362042859433172, "grad_norm": 1.2519834188938206, "learning_rate": 1.3742794344239862e-06, "loss": 0.4225, "step": 23471 }, { "epoch": 0.8362399130698114, "grad_norm": 1.6472521024344795, "learning_rate": 1.373695675778043e-06, "loss": 0.5143, "step": 23472 }, { "epoch": 0.8362755401963055, "grad_norm": 1.1664850064563832, "learning_rate": 1.3731120319973745e-06, "loss": 0.3991, "step": 23473 }, { "epoch": 0.8363111673227995, "grad_norm": 1.2146492243303713, "learning_rate": 1.3725285030897463e-06, "loss": 0.3714, "step": 23474 }, { "epoch": 0.8363467944492937, "grad_norm": 1.3888021864908322, "learning_rate": 1.3719450890629304e-06, "loss": 0.4409, "step": 23475 }, { "epoch": 0.8363824215757878, "grad_norm": 1.4784151608278155, "learning_rate": 1.3713617899246977e-06, "loss": 0.6054, "step": 23476 }, { "epoch": 0.8364180487022819, "grad_norm": 1.7747380927733982, "learning_rate": 1.3707786056828132e-06, "loss": 0.4674, "step": 23477 }, { "epoch": 0.836453675828776, "grad_norm": 1.160383316141756, "learning_rate": 1.3701955363450447e-06, "loss": 0.3601, "step": 23478 }, { "epoch": 0.8364893029552701, "grad_norm": 1.7358475134950808, "learning_rate": 1.3696125819191531e-06, "loss": 0.5211, "step": 23479 }, { "epoch": 0.8365249300817642, "grad_norm": 1.3032862731115726, "learning_rate": 1.369029742412904e-06, "loss": 0.4268, "step": 23480 }, { "epoch": 0.8365605572082584, "grad_norm": 1.2850580325735044, "learning_rate": 1.3684470178340548e-06, "loss": 0.4207, "step": 23481 }, { "epoch": 0.8365961843347525, "grad_norm": 1.4653264324013657, "learning_rate": 1.367864408190367e-06, "loss": 0.5081, "step": 23482 }, { "epoch": 0.8366318114612465, "grad_norm": 1.1679657764537115, "learning_rate": 1.3672819134895964e-06, "loss": 0.3833, "step": 23483 }, { "epoch": 0.8366674385877407, "grad_norm": 1.4108577746038071, "learning_rate": 1.366699533739505e-06, "loss": 0.5235, "step": 23484 }, { "epoch": 0.8367030657142348, "grad_norm": 1.294509421499332, "learning_rate": 1.3661172689478408e-06, "loss": 0.4371, "step": 23485 }, { "epoch": 0.8367386928407289, "grad_norm": 1.1485494623150447, "learning_rate": 1.365535119122362e-06, "loss": 0.3742, "step": 23486 }, { "epoch": 0.8367743199672231, "grad_norm": 2.0559258626112786, "learning_rate": 1.3649530842708158e-06, "loss": 0.5154, "step": 23487 }, { "epoch": 0.8368099470937171, "grad_norm": 1.1466009011939602, "learning_rate": 1.364371164400955e-06, "loss": 0.3561, "step": 23488 }, { "epoch": 0.8368455742202112, "grad_norm": 1.2063506971578657, "learning_rate": 1.363789359520531e-06, "loss": 0.4137, "step": 23489 }, { "epoch": 0.8368812013467054, "grad_norm": 1.2600660372539687, "learning_rate": 1.3632076696372831e-06, "loss": 0.3855, "step": 23490 }, { "epoch": 0.8369168284731995, "grad_norm": 1.227489488601099, "learning_rate": 1.3626260947589664e-06, "loss": 0.4234, "step": 23491 }, { "epoch": 0.8369524555996936, "grad_norm": 1.1422112666230297, "learning_rate": 1.362044634893318e-06, "loss": 0.4079, "step": 23492 }, { "epoch": 0.8369880827261877, "grad_norm": 1.7636578302594197, "learning_rate": 1.361463290048085e-06, "loss": 0.4934, "step": 23493 }, { "epoch": 0.8370237098526818, "grad_norm": 1.4564454096163342, "learning_rate": 1.3608820602310036e-06, "loss": 0.521, "step": 23494 }, { "epoch": 0.8370593369791759, "grad_norm": 1.1302962859752477, "learning_rate": 1.3603009454498184e-06, "loss": 0.3919, "step": 23495 }, { "epoch": 0.8370949641056701, "grad_norm": 1.137979103308273, "learning_rate": 1.3597199457122633e-06, "loss": 0.4087, "step": 23496 }, { "epoch": 0.8371305912321642, "grad_norm": 1.1178020335352195, "learning_rate": 1.3591390610260757e-06, "loss": 0.4063, "step": 23497 }, { "epoch": 0.8371662183586582, "grad_norm": 1.2883251507505835, "learning_rate": 1.3585582913989915e-06, "loss": 0.4307, "step": 23498 }, { "epoch": 0.8372018454851524, "grad_norm": 1.2015477179921248, "learning_rate": 1.3579776368387442e-06, "loss": 0.4327, "step": 23499 }, { "epoch": 0.8372374726116465, "grad_norm": 1.3870164673501342, "learning_rate": 1.3573970973530659e-06, "loss": 0.4922, "step": 23500 }, { "epoch": 0.8372730997381406, "grad_norm": 1.3338303607078887, "learning_rate": 1.3568166729496847e-06, "loss": 0.4235, "step": 23501 }, { "epoch": 0.8373087268646348, "grad_norm": 1.5135612661364324, "learning_rate": 1.356236363636333e-06, "loss": 0.4538, "step": 23502 }, { "epoch": 0.8373443539911288, "grad_norm": 1.5178245787969653, "learning_rate": 1.3556561694207337e-06, "loss": 0.5094, "step": 23503 }, { "epoch": 0.8373799811176229, "grad_norm": 1.80908693142684, "learning_rate": 1.3550760903106142e-06, "loss": 0.6493, "step": 23504 }, { "epoch": 0.8374156082441171, "grad_norm": 1.7288657986405274, "learning_rate": 1.3544961263136991e-06, "loss": 0.4306, "step": 23505 }, { "epoch": 0.8374512353706112, "grad_norm": 1.3065234856246009, "learning_rate": 1.3539162774377125e-06, "loss": 0.4175, "step": 23506 }, { "epoch": 0.8374868624971052, "grad_norm": 1.2345462342653883, "learning_rate": 1.3533365436903722e-06, "loss": 0.4075, "step": 23507 }, { "epoch": 0.8375224896235994, "grad_norm": 1.166755120804709, "learning_rate": 1.3527569250794014e-06, "loss": 0.393, "step": 23508 }, { "epoch": 0.8375581167500935, "grad_norm": 1.451204429427217, "learning_rate": 1.352177421612515e-06, "loss": 0.4752, "step": 23509 }, { "epoch": 0.8375937438765876, "grad_norm": 1.5548780957287336, "learning_rate": 1.35159803329743e-06, "loss": 0.5028, "step": 23510 }, { "epoch": 0.8376293710030818, "grad_norm": 1.3414725060057187, "learning_rate": 1.3510187601418646e-06, "loss": 0.4125, "step": 23511 }, { "epoch": 0.8376649981295758, "grad_norm": 1.2296572281549718, "learning_rate": 1.3504396021535259e-06, "loss": 0.4072, "step": 23512 }, { "epoch": 0.8377006252560699, "grad_norm": 1.2723546648714412, "learning_rate": 1.3498605593401315e-06, "loss": 0.4349, "step": 23513 }, { "epoch": 0.8377362523825641, "grad_norm": 1.156383655124836, "learning_rate": 1.3492816317093894e-06, "loss": 0.4256, "step": 23514 }, { "epoch": 0.8377718795090582, "grad_norm": 1.0487643015395032, "learning_rate": 1.348702819269011e-06, "loss": 0.4291, "step": 23515 }, { "epoch": 0.8378075066355523, "grad_norm": 1.2126684600121227, "learning_rate": 1.3481241220266984e-06, "loss": 0.4296, "step": 23516 }, { "epoch": 0.8378431337620464, "grad_norm": 1.4978157044839417, "learning_rate": 1.3475455399901639e-06, "loss": 0.4561, "step": 23517 }, { "epoch": 0.8378787608885405, "grad_norm": 1.4018366815802383, "learning_rate": 1.3469670731671046e-06, "loss": 0.562, "step": 23518 }, { "epoch": 0.8379143880150346, "grad_norm": 1.2013996179260893, "learning_rate": 1.3463887215652283e-06, "loss": 0.4346, "step": 23519 }, { "epoch": 0.8379500151415288, "grad_norm": 1.5767159393640249, "learning_rate": 1.3458104851922338e-06, "loss": 0.5766, "step": 23520 }, { "epoch": 0.8379856422680229, "grad_norm": 1.9324403049790755, "learning_rate": 1.3452323640558217e-06, "loss": 0.5072, "step": 23521 }, { "epoch": 0.8380212693945169, "grad_norm": 1.7361365003859537, "learning_rate": 1.3446543581636928e-06, "loss": 0.5447, "step": 23522 }, { "epoch": 0.8380568965210111, "grad_norm": 1.2114734853399942, "learning_rate": 1.3440764675235384e-06, "loss": 0.3756, "step": 23523 }, { "epoch": 0.8380925236475052, "grad_norm": 1.1010971774294604, "learning_rate": 1.3434986921430582e-06, "loss": 0.3889, "step": 23524 }, { "epoch": 0.8381281507739993, "grad_norm": 1.4690502610000025, "learning_rate": 1.3429210320299402e-06, "loss": 0.479, "step": 23525 }, { "epoch": 0.8381637779004935, "grad_norm": 1.154718871015324, "learning_rate": 1.3423434871918828e-06, "loss": 0.3875, "step": 23526 }, { "epoch": 0.8381994050269875, "grad_norm": 1.3008896533620007, "learning_rate": 1.341766057636571e-06, "loss": 0.4037, "step": 23527 }, { "epoch": 0.8382350321534816, "grad_norm": 1.196953238373869, "learning_rate": 1.3411887433716953e-06, "loss": 0.4136, "step": 23528 }, { "epoch": 0.8382706592799758, "grad_norm": 1.6968991261965818, "learning_rate": 1.3406115444049439e-06, "loss": 0.5337, "step": 23529 }, { "epoch": 0.8383062864064699, "grad_norm": 1.1832026907336552, "learning_rate": 1.3400344607440053e-06, "loss": 0.3938, "step": 23530 }, { "epoch": 0.838341913532964, "grad_norm": 1.043549557249111, "learning_rate": 1.3394574923965575e-06, "loss": 0.412, "step": 23531 }, { "epoch": 0.8383775406594581, "grad_norm": 1.2557266934797755, "learning_rate": 1.3388806393702891e-06, "loss": 0.4485, "step": 23532 }, { "epoch": 0.8384131677859522, "grad_norm": 1.2602742640569962, "learning_rate": 1.338303901672877e-06, "loss": 0.4374, "step": 23533 }, { "epoch": 0.8384487949124463, "grad_norm": 1.1354415524521553, "learning_rate": 1.3377272793120021e-06, "loss": 0.4029, "step": 23534 }, { "epoch": 0.8384844220389405, "grad_norm": 1.4535711817412116, "learning_rate": 1.3371507722953437e-06, "loss": 0.4691, "step": 23535 }, { "epoch": 0.8385200491654345, "grad_norm": 1.2257458292982057, "learning_rate": 1.3365743806305776e-06, "loss": 0.4161, "step": 23536 }, { "epoch": 0.8385556762919286, "grad_norm": 1.7045299408264272, "learning_rate": 1.33599810432538e-06, "loss": 0.5155, "step": 23537 }, { "epoch": 0.8385913034184228, "grad_norm": 1.1834501217929234, "learning_rate": 1.3354219433874227e-06, "loss": 0.4198, "step": 23538 }, { "epoch": 0.8386269305449169, "grad_norm": 1.3889715910845881, "learning_rate": 1.3348458978243795e-06, "loss": 0.5058, "step": 23539 }, { "epoch": 0.838662557671411, "grad_norm": 1.6678991715376645, "learning_rate": 1.3342699676439186e-06, "loss": 0.5629, "step": 23540 }, { "epoch": 0.8386981847979051, "grad_norm": 1.454605112114026, "learning_rate": 1.3336941528537118e-06, "loss": 0.4679, "step": 23541 }, { "epoch": 0.8387338119243992, "grad_norm": 1.2063993033833926, "learning_rate": 1.3331184534614194e-06, "loss": 0.4115, "step": 23542 }, { "epoch": 0.8387694390508933, "grad_norm": 1.7200923485711048, "learning_rate": 1.3325428694747178e-06, "loss": 0.465, "step": 23543 }, { "epoch": 0.8388050661773875, "grad_norm": 1.346804956565993, "learning_rate": 1.331967400901264e-06, "loss": 0.4193, "step": 23544 }, { "epoch": 0.8388406933038816, "grad_norm": 1.4472842150730993, "learning_rate": 1.331392047748723e-06, "loss": 0.5222, "step": 23545 }, { "epoch": 0.8388763204303756, "grad_norm": 1.2644739624550563, "learning_rate": 1.3308168100247575e-06, "loss": 0.3696, "step": 23546 }, { "epoch": 0.8389119475568698, "grad_norm": 1.501540201163032, "learning_rate": 1.3302416877370238e-06, "loss": 0.4981, "step": 23547 }, { "epoch": 0.8389475746833639, "grad_norm": 1.1538206817058916, "learning_rate": 1.3296666808931846e-06, "loss": 0.4284, "step": 23548 }, { "epoch": 0.8389832018098581, "grad_norm": 1.2056463874473833, "learning_rate": 1.3290917895008903e-06, "loss": 0.3729, "step": 23549 }, { "epoch": 0.8390188289363522, "grad_norm": 1.2955306136552769, "learning_rate": 1.3285170135678005e-06, "loss": 0.3927, "step": 23550 }, { "epoch": 0.8390544560628462, "grad_norm": 1.3653411472787496, "learning_rate": 1.3279423531015667e-06, "loss": 0.4491, "step": 23551 }, { "epoch": 0.8390900831893404, "grad_norm": 1.1231798068865178, "learning_rate": 1.327367808109845e-06, "loss": 0.3943, "step": 23552 }, { "epoch": 0.8391257103158345, "grad_norm": 1.2167466159359972, "learning_rate": 1.3267933786002796e-06, "loss": 0.4106, "step": 23553 }, { "epoch": 0.8391613374423286, "grad_norm": 1.5483807592389915, "learning_rate": 1.326219064580525e-06, "loss": 0.3764, "step": 23554 }, { "epoch": 0.8391969645688228, "grad_norm": 1.2471425941707035, "learning_rate": 1.3256448660582244e-06, "loss": 0.3884, "step": 23555 }, { "epoch": 0.8392325916953168, "grad_norm": 1.1490696162854008, "learning_rate": 1.3250707830410259e-06, "loss": 0.3692, "step": 23556 }, { "epoch": 0.8392682188218109, "grad_norm": 1.2371050811282038, "learning_rate": 1.3244968155365723e-06, "loss": 0.3907, "step": 23557 }, { "epoch": 0.8393038459483051, "grad_norm": 1.5613961278300235, "learning_rate": 1.3239229635525074e-06, "loss": 0.5973, "step": 23558 }, { "epoch": 0.8393394730747992, "grad_norm": 1.7346080719179273, "learning_rate": 1.3233492270964754e-06, "loss": 0.5002, "step": 23559 }, { "epoch": 0.8393751002012932, "grad_norm": 1.165630186330682, "learning_rate": 1.322775606176111e-06, "loss": 0.3939, "step": 23560 }, { "epoch": 0.8394107273277874, "grad_norm": 1.577132398097558, "learning_rate": 1.322202100799056e-06, "loss": 0.5495, "step": 23561 }, { "epoch": 0.8394463544542815, "grad_norm": 1.1633479716269708, "learning_rate": 1.321628710972943e-06, "loss": 0.3666, "step": 23562 }, { "epoch": 0.8394819815807756, "grad_norm": 1.1958737993473476, "learning_rate": 1.3210554367054119e-06, "loss": 0.4114, "step": 23563 }, { "epoch": 0.8395176087072698, "grad_norm": 1.1658687762742785, "learning_rate": 1.3204822780040916e-06, "loss": 0.4024, "step": 23564 }, { "epoch": 0.8395532358337638, "grad_norm": 1.7065071794238162, "learning_rate": 1.3199092348766174e-06, "loss": 0.5288, "step": 23565 }, { "epoch": 0.8395888629602579, "grad_norm": 1.162291667822481, "learning_rate": 1.3193363073306176e-06, "loss": 0.361, "step": 23566 }, { "epoch": 0.8396244900867521, "grad_norm": 1.1832833743298556, "learning_rate": 1.318763495373725e-06, "loss": 0.3653, "step": 23567 }, { "epoch": 0.8396601172132462, "grad_norm": 1.1716308359371537, "learning_rate": 1.3181907990135624e-06, "loss": 0.3686, "step": 23568 }, { "epoch": 0.8396957443397403, "grad_norm": 1.304533973026876, "learning_rate": 1.317618218257757e-06, "loss": 0.5134, "step": 23569 }, { "epoch": 0.8397313714662344, "grad_norm": 1.4414780440403319, "learning_rate": 1.3170457531139358e-06, "loss": 0.5192, "step": 23570 }, { "epoch": 0.8397669985927285, "grad_norm": 1.5651193753564003, "learning_rate": 1.3164734035897175e-06, "loss": 0.5204, "step": 23571 }, { "epoch": 0.8398026257192226, "grad_norm": 1.5228068514919992, "learning_rate": 1.3159011696927248e-06, "loss": 0.4739, "step": 23572 }, { "epoch": 0.8398382528457168, "grad_norm": 1.4111071329714886, "learning_rate": 1.3153290514305784e-06, "loss": 0.5259, "step": 23573 }, { "epoch": 0.8398738799722109, "grad_norm": 1.1792986227264561, "learning_rate": 1.3147570488108985e-06, "loss": 0.4103, "step": 23574 }, { "epoch": 0.8399095070987049, "grad_norm": 1.2690821057173287, "learning_rate": 1.314185161841296e-06, "loss": 0.4402, "step": 23575 }, { "epoch": 0.8399451342251991, "grad_norm": 1.121686609847981, "learning_rate": 1.3136133905293912e-06, "loss": 0.4273, "step": 23576 }, { "epoch": 0.8399807613516932, "grad_norm": 1.1826785443570058, "learning_rate": 1.313041734882794e-06, "loss": 0.4161, "step": 23577 }, { "epoch": 0.8400163884781873, "grad_norm": 1.2406972937370313, "learning_rate": 1.31247019490912e-06, "loss": 0.3812, "step": 23578 }, { "epoch": 0.8400520156046815, "grad_norm": 1.6643627167682677, "learning_rate": 1.3118987706159724e-06, "loss": 0.4578, "step": 23579 }, { "epoch": 0.8400876427311755, "grad_norm": 1.1982364677775237, "learning_rate": 1.3113274620109684e-06, "loss": 0.3699, "step": 23580 }, { "epoch": 0.8401232698576696, "grad_norm": 1.3386569784455375, "learning_rate": 1.310756269101714e-06, "loss": 0.3929, "step": 23581 }, { "epoch": 0.8401588969841638, "grad_norm": 1.130330217399844, "learning_rate": 1.310185191895812e-06, "loss": 0.3757, "step": 23582 }, { "epoch": 0.8401945241106579, "grad_norm": 1.1726667671522124, "learning_rate": 1.3096142304008696e-06, "loss": 0.3941, "step": 23583 }, { "epoch": 0.840230151237152, "grad_norm": 1.1930106304929944, "learning_rate": 1.3090433846244853e-06, "loss": 0.3887, "step": 23584 }, { "epoch": 0.8402657783636461, "grad_norm": 1.437125828247876, "learning_rate": 1.3084726545742654e-06, "loss": 0.5584, "step": 23585 }, { "epoch": 0.8403014054901402, "grad_norm": 1.2625363107901928, "learning_rate": 1.3079020402578046e-06, "loss": 0.4001, "step": 23586 }, { "epoch": 0.8403370326166343, "grad_norm": 1.8514024988517908, "learning_rate": 1.3073315416827036e-06, "loss": 0.4875, "step": 23587 }, { "epoch": 0.8403726597431285, "grad_norm": 1.5525996354759404, "learning_rate": 1.3067611588565598e-06, "loss": 0.593, "step": 23588 }, { "epoch": 0.8404082868696225, "grad_norm": 1.1158222652239562, "learning_rate": 1.3061908917869692e-06, "loss": 0.3336, "step": 23589 }, { "epoch": 0.8404439139961166, "grad_norm": 1.2712838258102268, "learning_rate": 1.3056207404815214e-06, "loss": 0.4815, "step": 23590 }, { "epoch": 0.8404795411226108, "grad_norm": 1.5864622513923186, "learning_rate": 1.30505070494781e-06, "loss": 0.4896, "step": 23591 }, { "epoch": 0.8405151682491049, "grad_norm": 1.776005338896676, "learning_rate": 1.3044807851934283e-06, "loss": 0.5346, "step": 23592 }, { "epoch": 0.840550795375599, "grad_norm": 1.6016587828774698, "learning_rate": 1.3039109812259598e-06, "loss": 0.4485, "step": 23593 }, { "epoch": 0.8405864225020931, "grad_norm": 1.9045180094691447, "learning_rate": 1.3033412930529953e-06, "loss": 0.4552, "step": 23594 }, { "epoch": 0.8406220496285872, "grad_norm": 1.7278955280083825, "learning_rate": 1.302771720682121e-06, "loss": 0.4519, "step": 23595 }, { "epoch": 0.8406576767550813, "grad_norm": 1.649244236854687, "learning_rate": 1.302202264120922e-06, "loss": 0.4579, "step": 23596 }, { "epoch": 0.8406933038815755, "grad_norm": 1.7940041878682305, "learning_rate": 1.3016329233769765e-06, "loss": 0.5333, "step": 23597 }, { "epoch": 0.8407289310080696, "grad_norm": 1.278470941980172, "learning_rate": 1.3010636984578706e-06, "loss": 0.4064, "step": 23598 }, { "epoch": 0.8407645581345636, "grad_norm": 1.7198580039504758, "learning_rate": 1.3004945893711796e-06, "loss": 0.4957, "step": 23599 }, { "epoch": 0.8408001852610578, "grad_norm": 1.4789619174012774, "learning_rate": 1.299925596124486e-06, "loss": 0.516, "step": 23600 }, { "epoch": 0.8408358123875519, "grad_norm": 1.2795876463536406, "learning_rate": 1.2993567187253619e-06, "loss": 0.4303, "step": 23601 }, { "epoch": 0.840871439514046, "grad_norm": 1.2364197644367185, "learning_rate": 1.2987879571813855e-06, "loss": 0.4405, "step": 23602 }, { "epoch": 0.8409070666405402, "grad_norm": 3.561138412117606, "learning_rate": 1.2982193115001285e-06, "loss": 0.592, "step": 23603 }, { "epoch": 0.8409426937670342, "grad_norm": 1.325301685513608, "learning_rate": 1.2976507816891627e-06, "loss": 0.4493, "step": 23604 }, { "epoch": 0.8409783208935283, "grad_norm": 1.4164496996329434, "learning_rate": 1.2970823677560628e-06, "loss": 0.4226, "step": 23605 }, { "epoch": 0.8410139480200225, "grad_norm": 1.1746126425741852, "learning_rate": 1.2965140697083911e-06, "loss": 0.3909, "step": 23606 }, { "epoch": 0.8410495751465166, "grad_norm": 1.4299246677750026, "learning_rate": 1.2959458875537201e-06, "loss": 0.4773, "step": 23607 }, { "epoch": 0.8410852022730106, "grad_norm": 1.1434300990288166, "learning_rate": 1.2953778212996115e-06, "loss": 0.3978, "step": 23608 }, { "epoch": 0.8411208293995048, "grad_norm": 1.1883026176503526, "learning_rate": 1.2948098709536317e-06, "loss": 0.3847, "step": 23609 }, { "epoch": 0.8411564565259989, "grad_norm": 1.383330707605681, "learning_rate": 1.2942420365233421e-06, "loss": 0.4461, "step": 23610 }, { "epoch": 0.841192083652493, "grad_norm": 1.3870782658336023, "learning_rate": 1.2936743180163081e-06, "loss": 0.4683, "step": 23611 }, { "epoch": 0.8412277107789872, "grad_norm": 1.6320994429440194, "learning_rate": 1.2931067154400834e-06, "loss": 0.4871, "step": 23612 }, { "epoch": 0.8412633379054812, "grad_norm": 1.5709028469677309, "learning_rate": 1.2925392288022299e-06, "loss": 0.5297, "step": 23613 }, { "epoch": 0.8412989650319753, "grad_norm": 1.3024565245928468, "learning_rate": 1.2919718581103013e-06, "loss": 0.4052, "step": 23614 }, { "epoch": 0.8413345921584695, "grad_norm": 1.1761365966314512, "learning_rate": 1.291404603371853e-06, "loss": 0.4107, "step": 23615 }, { "epoch": 0.8413702192849636, "grad_norm": 1.5257018534655318, "learning_rate": 1.2908374645944432e-06, "loss": 0.5398, "step": 23616 }, { "epoch": 0.8414058464114577, "grad_norm": 1.570244124181383, "learning_rate": 1.2902704417856137e-06, "loss": 0.562, "step": 23617 }, { "epoch": 0.8414414735379518, "grad_norm": 1.1091627375552346, "learning_rate": 1.2897035349529263e-06, "loss": 0.3915, "step": 23618 }, { "epoch": 0.8414771006644459, "grad_norm": 1.3191098125899452, "learning_rate": 1.2891367441039226e-06, "loss": 0.449, "step": 23619 }, { "epoch": 0.84151272779094, "grad_norm": 1.1997455854593864, "learning_rate": 1.2885700692461533e-06, "loss": 0.415, "step": 23620 }, { "epoch": 0.8415483549174342, "grad_norm": 1.1630457647495798, "learning_rate": 1.2880035103871614e-06, "loss": 0.3921, "step": 23621 }, { "epoch": 0.8415839820439283, "grad_norm": 1.4471589824418407, "learning_rate": 1.2874370675344927e-06, "loss": 0.548, "step": 23622 }, { "epoch": 0.8416196091704223, "grad_norm": 1.2678573345141306, "learning_rate": 1.2868707406956881e-06, "loss": 0.3884, "step": 23623 }, { "epoch": 0.8416552362969165, "grad_norm": 1.9203786362528283, "learning_rate": 1.2863045298782894e-06, "loss": 0.4782, "step": 23624 }, { "epoch": 0.8416908634234106, "grad_norm": 1.6690951485501893, "learning_rate": 1.2857384350898361e-06, "loss": 0.5379, "step": 23625 }, { "epoch": 0.8417264905499047, "grad_norm": 1.2260340605146542, "learning_rate": 1.2851724563378664e-06, "loss": 0.4188, "step": 23626 }, { "epoch": 0.8417621176763989, "grad_norm": 1.4535652858732746, "learning_rate": 1.284606593629919e-06, "loss": 0.5114, "step": 23627 }, { "epoch": 0.8417977448028929, "grad_norm": 1.3259455957054767, "learning_rate": 1.2840408469735245e-06, "loss": 0.4262, "step": 23628 }, { "epoch": 0.841833371929387, "grad_norm": 1.2646807861338194, "learning_rate": 1.2834752163762198e-06, "loss": 0.4268, "step": 23629 }, { "epoch": 0.8418689990558812, "grad_norm": 1.6164365114160313, "learning_rate": 1.2829097018455339e-06, "loss": 0.4921, "step": 23630 }, { "epoch": 0.8419046261823753, "grad_norm": 1.6355977756444573, "learning_rate": 1.2823443033889983e-06, "loss": 0.4678, "step": 23631 }, { "epoch": 0.8419402533088693, "grad_norm": 1.699619279532864, "learning_rate": 1.2817790210141401e-06, "loss": 0.5906, "step": 23632 }, { "epoch": 0.8419758804353635, "grad_norm": 1.17881589486835, "learning_rate": 1.2812138547284913e-06, "loss": 0.4359, "step": 23633 }, { "epoch": 0.8420115075618576, "grad_norm": 1.5601523379812599, "learning_rate": 1.2806488045395715e-06, "loss": 0.4868, "step": 23634 }, { "epoch": 0.8420471346883517, "grad_norm": 1.424847606667926, "learning_rate": 1.28008387045491e-06, "loss": 0.4611, "step": 23635 }, { "epoch": 0.8420827618148459, "grad_norm": 1.1235026841106497, "learning_rate": 1.2795190524820245e-06, "loss": 0.3912, "step": 23636 }, { "epoch": 0.84211838894134, "grad_norm": 1.4316171467684558, "learning_rate": 1.2789543506284374e-06, "loss": 0.4268, "step": 23637 }, { "epoch": 0.842154016067834, "grad_norm": 1.980635693366241, "learning_rate": 1.2783897649016709e-06, "loss": 0.5437, "step": 23638 }, { "epoch": 0.8421896431943282, "grad_norm": 1.2450419041634186, "learning_rate": 1.2778252953092384e-06, "loss": 0.389, "step": 23639 }, { "epoch": 0.8422252703208223, "grad_norm": 1.2504283899576485, "learning_rate": 1.277260941858659e-06, "loss": 0.4146, "step": 23640 }, { "epoch": 0.8422608974473164, "grad_norm": 1.1828844438402388, "learning_rate": 1.2766967045574475e-06, "loss": 0.4066, "step": 23641 }, { "epoch": 0.8422965245738105, "grad_norm": 1.681834354829094, "learning_rate": 1.2761325834131177e-06, "loss": 0.5461, "step": 23642 }, { "epoch": 0.8423321517003046, "grad_norm": 1.2997800222375324, "learning_rate": 1.2755685784331784e-06, "loss": 0.4245, "step": 23643 }, { "epoch": 0.8423677788267987, "grad_norm": 1.6485747141589433, "learning_rate": 1.2750046896251444e-06, "loss": 0.5556, "step": 23644 }, { "epoch": 0.8424034059532929, "grad_norm": 1.2654646748930456, "learning_rate": 1.2744409169965188e-06, "loss": 0.4942, "step": 23645 }, { "epoch": 0.842439033079787, "grad_norm": 1.5469890968163114, "learning_rate": 1.27387726055481e-06, "loss": 0.5342, "step": 23646 }, { "epoch": 0.842474660206281, "grad_norm": 1.6969979179998544, "learning_rate": 1.2733137203075263e-06, "loss": 0.5098, "step": 23647 }, { "epoch": 0.8425102873327752, "grad_norm": 1.7438520242428381, "learning_rate": 1.2727502962621696e-06, "loss": 0.4701, "step": 23648 }, { "epoch": 0.8425459144592693, "grad_norm": 1.5871509062355291, "learning_rate": 1.2721869884262438e-06, "loss": 0.4184, "step": 23649 }, { "epoch": 0.8425815415857634, "grad_norm": 1.6325306447890016, "learning_rate": 1.2716237968072476e-06, "loss": 0.5607, "step": 23650 }, { "epoch": 0.8426171687122576, "grad_norm": 1.2951032534457678, "learning_rate": 1.2710607214126835e-06, "loss": 0.417, "step": 23651 }, { "epoch": 0.8426527958387516, "grad_norm": 1.274373136842416, "learning_rate": 1.2704977622500436e-06, "loss": 0.4134, "step": 23652 }, { "epoch": 0.8426884229652457, "grad_norm": 1.5189985527586523, "learning_rate": 1.2699349193268307e-06, "loss": 0.5395, "step": 23653 }, { "epoch": 0.8427240500917399, "grad_norm": 1.1408058093444267, "learning_rate": 1.2693721926505342e-06, "loss": 0.3857, "step": 23654 }, { "epoch": 0.842759677218234, "grad_norm": 1.2132489140581133, "learning_rate": 1.2688095822286494e-06, "loss": 0.4314, "step": 23655 }, { "epoch": 0.842795304344728, "grad_norm": 1.235972802723387, "learning_rate": 1.2682470880686682e-06, "loss": 0.4261, "step": 23656 }, { "epoch": 0.8428309314712222, "grad_norm": 1.2967323836388591, "learning_rate": 1.267684710178081e-06, "loss": 0.4581, "step": 23657 }, { "epoch": 0.8428665585977163, "grad_norm": 1.5916589691580996, "learning_rate": 1.267122448564374e-06, "loss": 0.57, "step": 23658 }, { "epoch": 0.8429021857242104, "grad_norm": 1.1437487667993742, "learning_rate": 1.266560303235037e-06, "loss": 0.3932, "step": 23659 }, { "epoch": 0.8429378128507046, "grad_norm": 1.472785009911156, "learning_rate": 1.265998274197553e-06, "loss": 0.4863, "step": 23660 }, { "epoch": 0.8429734399771986, "grad_norm": 1.5806717741926128, "learning_rate": 1.2654363614594055e-06, "loss": 0.4754, "step": 23661 }, { "epoch": 0.8430090671036927, "grad_norm": 1.2196993554324878, "learning_rate": 1.264874565028079e-06, "loss": 0.3664, "step": 23662 }, { "epoch": 0.8430446942301869, "grad_norm": 1.805374432656534, "learning_rate": 1.264312884911052e-06, "loss": 0.5196, "step": 23663 }, { "epoch": 0.843080321356681, "grad_norm": 1.5912221649301375, "learning_rate": 1.263751321115808e-06, "loss": 0.5161, "step": 23664 }, { "epoch": 0.8431159484831751, "grad_norm": 1.187424416699319, "learning_rate": 1.2631898736498183e-06, "loss": 0.3961, "step": 23665 }, { "epoch": 0.8431515756096692, "grad_norm": 1.6489094562924862, "learning_rate": 1.2626285425205652e-06, "loss": 0.5542, "step": 23666 }, { "epoch": 0.8431872027361633, "grad_norm": 1.165993385959055, "learning_rate": 1.2620673277355178e-06, "loss": 0.4578, "step": 23667 }, { "epoch": 0.8432228298626574, "grad_norm": 1.6842770982360957, "learning_rate": 1.2615062293021508e-06, "loss": 0.5141, "step": 23668 }, { "epoch": 0.8432584569891516, "grad_norm": 1.1391980357556064, "learning_rate": 1.260945247227936e-06, "loss": 0.4007, "step": 23669 }, { "epoch": 0.8432940841156457, "grad_norm": 1.4808215077188993, "learning_rate": 1.2603843815203466e-06, "loss": 0.5297, "step": 23670 }, { "epoch": 0.8433297112421397, "grad_norm": 1.8392286837178102, "learning_rate": 1.259823632186845e-06, "loss": 0.5265, "step": 23671 }, { "epoch": 0.8433653383686339, "grad_norm": 1.2694537570872122, "learning_rate": 1.2592629992349003e-06, "loss": 0.3609, "step": 23672 }, { "epoch": 0.843400965495128, "grad_norm": 2.0588069244485423, "learning_rate": 1.2587024826719817e-06, "loss": 0.4524, "step": 23673 }, { "epoch": 0.8434365926216221, "grad_norm": 1.348814510268725, "learning_rate": 1.2581420825055457e-06, "loss": 0.3827, "step": 23674 }, { "epoch": 0.8434722197481163, "grad_norm": 1.581836786515886, "learning_rate": 1.2575817987430616e-06, "loss": 0.5131, "step": 23675 }, { "epoch": 0.8435078468746103, "grad_norm": 1.333170300773651, "learning_rate": 1.2570216313919836e-06, "loss": 0.4043, "step": 23676 }, { "epoch": 0.8435434740011044, "grad_norm": 1.2942136706930003, "learning_rate": 1.2564615804597734e-06, "loss": 0.4848, "step": 23677 }, { "epoch": 0.8435791011275986, "grad_norm": 1.7619280700678943, "learning_rate": 1.2559016459538885e-06, "loss": 0.5097, "step": 23678 }, { "epoch": 0.8436147282540927, "grad_norm": 1.2431661027449004, "learning_rate": 1.2553418278817874e-06, "loss": 0.4126, "step": 23679 }, { "epoch": 0.8436503553805867, "grad_norm": 1.3146780627377077, "learning_rate": 1.2547821262509197e-06, "loss": 0.3937, "step": 23680 }, { "epoch": 0.8436859825070809, "grad_norm": 1.3313916220639246, "learning_rate": 1.2542225410687436e-06, "loss": 0.5112, "step": 23681 }, { "epoch": 0.843721609633575, "grad_norm": 1.2703568341918545, "learning_rate": 1.2536630723427045e-06, "loss": 0.4122, "step": 23682 }, { "epoch": 0.8437572367600691, "grad_norm": 1.528421253002082, "learning_rate": 1.2531037200802543e-06, "loss": 0.4476, "step": 23683 }, { "epoch": 0.8437928638865633, "grad_norm": 1.5708252506680178, "learning_rate": 1.2525444842888423e-06, "loss": 0.5371, "step": 23684 }, { "epoch": 0.8438284910130573, "grad_norm": 1.6751938954111385, "learning_rate": 1.251985364975915e-06, "loss": 0.4623, "step": 23685 }, { "epoch": 0.8438641181395514, "grad_norm": 1.205858001745147, "learning_rate": 1.2514263621489187e-06, "loss": 0.4083, "step": 23686 }, { "epoch": 0.8438997452660456, "grad_norm": 1.5145947843799412, "learning_rate": 1.2508674758152927e-06, "loss": 0.4951, "step": 23687 }, { "epoch": 0.8439353723925397, "grad_norm": 1.4634955866588049, "learning_rate": 1.2503087059824848e-06, "loss": 0.5068, "step": 23688 }, { "epoch": 0.8439709995190338, "grad_norm": 1.729341420806276, "learning_rate": 1.2497500526579288e-06, "loss": 0.4554, "step": 23689 }, { "epoch": 0.844006626645528, "grad_norm": 1.2853281217512433, "learning_rate": 1.2491915158490697e-06, "loss": 0.4329, "step": 23690 }, { "epoch": 0.844042253772022, "grad_norm": 1.1365888626063534, "learning_rate": 1.2486330955633397e-06, "loss": 0.4112, "step": 23691 }, { "epoch": 0.8440778808985161, "grad_norm": 1.1436754720277522, "learning_rate": 1.248074791808177e-06, "loss": 0.3832, "step": 23692 }, { "epoch": 0.8441135080250103, "grad_norm": 1.6194159114713667, "learning_rate": 1.2475166045910158e-06, "loss": 0.5901, "step": 23693 }, { "epoch": 0.8441491351515044, "grad_norm": 1.2156897317980597, "learning_rate": 1.2469585339192892e-06, "loss": 0.4019, "step": 23694 }, { "epoch": 0.8441847622779984, "grad_norm": 1.7495064018158646, "learning_rate": 1.2464005798004287e-06, "loss": 0.5085, "step": 23695 }, { "epoch": 0.8442203894044926, "grad_norm": 1.3334428360790989, "learning_rate": 1.2458427422418618e-06, "loss": 0.4212, "step": 23696 }, { "epoch": 0.8442560165309867, "grad_norm": 1.2559557936979413, "learning_rate": 1.2452850212510182e-06, "loss": 0.3758, "step": 23697 }, { "epoch": 0.8442916436574808, "grad_norm": 1.3690938242025124, "learning_rate": 1.244727416835323e-06, "loss": 0.4294, "step": 23698 }, { "epoch": 0.844327270783975, "grad_norm": 1.0733006540444783, "learning_rate": 1.2441699290022014e-06, "loss": 0.3979, "step": 23699 }, { "epoch": 0.844362897910469, "grad_norm": 1.449588592527381, "learning_rate": 1.2436125577590774e-06, "loss": 0.4264, "step": 23700 }, { "epoch": 0.8443985250369631, "grad_norm": 1.3336803843190044, "learning_rate": 1.243055303113374e-06, "loss": 0.4228, "step": 23701 }, { "epoch": 0.8444341521634573, "grad_norm": 1.316119686806604, "learning_rate": 1.2424981650725077e-06, "loss": 0.4352, "step": 23702 }, { "epoch": 0.8444697792899514, "grad_norm": 1.7459384504010584, "learning_rate": 1.2419411436439021e-06, "loss": 0.5306, "step": 23703 }, { "epoch": 0.8445054064164454, "grad_norm": 1.2153014994288156, "learning_rate": 1.2413842388349696e-06, "loss": 0.4506, "step": 23704 }, { "epoch": 0.8445410335429396, "grad_norm": 1.3557016136554576, "learning_rate": 1.2408274506531305e-06, "loss": 0.587, "step": 23705 }, { "epoch": 0.8445766606694337, "grad_norm": 1.826547729902852, "learning_rate": 1.240270779105791e-06, "loss": 0.5434, "step": 23706 }, { "epoch": 0.8446122877959278, "grad_norm": 1.2339193829216966, "learning_rate": 1.2397142242003713e-06, "loss": 0.3917, "step": 23707 }, { "epoch": 0.844647914922422, "grad_norm": 1.1511262182323365, "learning_rate": 1.2391577859442816e-06, "loss": 0.4062, "step": 23708 }, { "epoch": 0.844683542048916, "grad_norm": 1.5004828294116022, "learning_rate": 1.2386014643449284e-06, "loss": 0.4639, "step": 23709 }, { "epoch": 0.8447191691754101, "grad_norm": 1.2035787123206714, "learning_rate": 1.2380452594097215e-06, "loss": 0.385, "step": 23710 }, { "epoch": 0.8447547963019043, "grad_norm": 1.2262070969088668, "learning_rate": 1.2374891711460645e-06, "loss": 0.3823, "step": 23711 }, { "epoch": 0.8447904234283984, "grad_norm": 1.254811985142387, "learning_rate": 1.2369331995613664e-06, "loss": 0.373, "step": 23712 }, { "epoch": 0.8448260505548925, "grad_norm": 1.5384189520396516, "learning_rate": 1.2363773446630245e-06, "loss": 0.4566, "step": 23713 }, { "epoch": 0.8448616776813866, "grad_norm": 1.4236486748257984, "learning_rate": 1.235821606458445e-06, "loss": 0.4613, "step": 23714 }, { "epoch": 0.8448973048078807, "grad_norm": 1.2118118255136978, "learning_rate": 1.2352659849550253e-06, "loss": 0.4221, "step": 23715 }, { "epoch": 0.8449329319343748, "grad_norm": 1.4862818712581112, "learning_rate": 1.2347104801601683e-06, "loss": 0.4865, "step": 23716 }, { "epoch": 0.844968559060869, "grad_norm": 1.1938844469267547, "learning_rate": 1.234155092081265e-06, "loss": 0.3913, "step": 23717 }, { "epoch": 0.8450041861873631, "grad_norm": 1.5431894221671443, "learning_rate": 1.2335998207257138e-06, "loss": 0.548, "step": 23718 }, { "epoch": 0.8450398133138571, "grad_norm": 1.3278690215269084, "learning_rate": 1.233044666100911e-06, "loss": 0.3691, "step": 23719 }, { "epoch": 0.8450754404403513, "grad_norm": 3.103704722036557, "learning_rate": 1.232489628214244e-06, "loss": 0.4658, "step": 23720 }, { "epoch": 0.8451110675668454, "grad_norm": 1.2981988865084977, "learning_rate": 1.2319347070731047e-06, "loss": 0.399, "step": 23721 }, { "epoch": 0.8451466946933395, "grad_norm": 1.7449199372704656, "learning_rate": 1.2313799026848839e-06, "loss": 0.4724, "step": 23722 }, { "epoch": 0.8451823218198337, "grad_norm": 1.1871971301390027, "learning_rate": 1.230825215056971e-06, "loss": 0.4346, "step": 23723 }, { "epoch": 0.8452179489463277, "grad_norm": 2.041271770254566, "learning_rate": 1.230270644196747e-06, "loss": 0.5048, "step": 23724 }, { "epoch": 0.8452535760728218, "grad_norm": 1.2330431811410196, "learning_rate": 1.2297161901116018e-06, "loss": 0.4264, "step": 23725 }, { "epoch": 0.845289203199316, "grad_norm": 1.2487133879621812, "learning_rate": 1.2291618528089134e-06, "loss": 0.4618, "step": 23726 }, { "epoch": 0.8453248303258101, "grad_norm": 1.570844220850831, "learning_rate": 1.2286076322960672e-06, "loss": 0.4271, "step": 23727 }, { "epoch": 0.8453604574523041, "grad_norm": 1.153430711807803, "learning_rate": 1.2280535285804396e-06, "loss": 0.3613, "step": 23728 }, { "epoch": 0.8453960845787983, "grad_norm": 1.311829952581329, "learning_rate": 1.2274995416694102e-06, "loss": 0.4015, "step": 23729 }, { "epoch": 0.8454317117052924, "grad_norm": 1.3603160093932583, "learning_rate": 1.2269456715703554e-06, "loss": 0.4797, "step": 23730 }, { "epoch": 0.8454673388317865, "grad_norm": 1.1217931728737076, "learning_rate": 1.2263919182906514e-06, "loss": 0.3557, "step": 23731 }, { "epoch": 0.8455029659582807, "grad_norm": 1.182553214728561, "learning_rate": 1.2258382818376735e-06, "loss": 0.4154, "step": 23732 }, { "epoch": 0.8455385930847747, "grad_norm": 2.0547166364523095, "learning_rate": 1.2252847622187902e-06, "loss": 0.4865, "step": 23733 }, { "epoch": 0.8455742202112688, "grad_norm": 1.225987973363999, "learning_rate": 1.2247313594413745e-06, "loss": 0.3815, "step": 23734 }, { "epoch": 0.845609847337763, "grad_norm": 1.6951973357575385, "learning_rate": 1.2241780735127927e-06, "loss": 0.4651, "step": 23735 }, { "epoch": 0.8456454744642571, "grad_norm": 1.4559664912721566, "learning_rate": 1.2236249044404137e-06, "loss": 0.5496, "step": 23736 }, { "epoch": 0.8456811015907512, "grad_norm": 1.3204143831315736, "learning_rate": 1.2230718522316031e-06, "loss": 0.3959, "step": 23737 }, { "epoch": 0.8457167287172453, "grad_norm": 1.111803973877564, "learning_rate": 1.222518916893728e-06, "loss": 0.3511, "step": 23738 }, { "epoch": 0.8457523558437394, "grad_norm": 1.4505906766184837, "learning_rate": 1.2219660984341463e-06, "loss": 0.4719, "step": 23739 }, { "epoch": 0.8457879829702335, "grad_norm": 1.4645851307684647, "learning_rate": 1.2214133968602238e-06, "loss": 0.4634, "step": 23740 }, { "epoch": 0.8458236100967277, "grad_norm": 1.4224704339020136, "learning_rate": 1.2208608121793164e-06, "loss": 0.5156, "step": 23741 }, { "epoch": 0.8458592372232218, "grad_norm": 1.6141926297931768, "learning_rate": 1.2203083443987818e-06, "loss": 0.5007, "step": 23742 }, { "epoch": 0.8458948643497158, "grad_norm": 1.453107915542468, "learning_rate": 1.2197559935259796e-06, "loss": 0.4373, "step": 23743 }, { "epoch": 0.84593049147621, "grad_norm": 1.2622808365276033, "learning_rate": 1.2192037595682626e-06, "loss": 0.3973, "step": 23744 }, { "epoch": 0.8459661186027041, "grad_norm": 1.283752301242764, "learning_rate": 1.2186516425329874e-06, "loss": 0.3979, "step": 23745 }, { "epoch": 0.8460017457291982, "grad_norm": 1.5869559542279341, "learning_rate": 1.2180996424275016e-06, "loss": 0.5006, "step": 23746 }, { "epoch": 0.8460373728556924, "grad_norm": 1.2518202047287605, "learning_rate": 1.217547759259159e-06, "loss": 0.3949, "step": 23747 }, { "epoch": 0.8460729999821864, "grad_norm": 1.2423001969832477, "learning_rate": 1.2169959930353049e-06, "loss": 0.424, "step": 23748 }, { "epoch": 0.8461086271086805, "grad_norm": 1.288990242692192, "learning_rate": 1.2164443437632901e-06, "loss": 0.4443, "step": 23749 }, { "epoch": 0.8461442542351747, "grad_norm": 1.1990081554512242, "learning_rate": 1.2158928114504552e-06, "loss": 0.3731, "step": 23750 }, { "epoch": 0.8461798813616688, "grad_norm": 1.184889535457467, "learning_rate": 1.2153413961041483e-06, "loss": 0.3722, "step": 23751 }, { "epoch": 0.8462155084881628, "grad_norm": 1.671146617576943, "learning_rate": 1.2147900977317096e-06, "loss": 0.5228, "step": 23752 }, { "epoch": 0.846251135614657, "grad_norm": 1.1332459872633116, "learning_rate": 1.2142389163404821e-06, "loss": 0.3792, "step": 23753 }, { "epoch": 0.8462867627411511, "grad_norm": 1.4686005035098686, "learning_rate": 1.213687851937807e-06, "loss": 0.4965, "step": 23754 }, { "epoch": 0.8463223898676452, "grad_norm": 1.5584817919864755, "learning_rate": 1.2131369045310159e-06, "loss": 0.4529, "step": 23755 }, { "epoch": 0.8463580169941394, "grad_norm": 1.5351763213529075, "learning_rate": 1.212586074127452e-06, "loss": 0.5251, "step": 23756 }, { "epoch": 0.8463936441206334, "grad_norm": 1.1627608982170654, "learning_rate": 1.2120353607344438e-06, "loss": 0.3985, "step": 23757 }, { "epoch": 0.8464292712471275, "grad_norm": 1.1223310475190338, "learning_rate": 1.2114847643593264e-06, "loss": 0.3996, "step": 23758 }, { "epoch": 0.8464648983736217, "grad_norm": 2.181859846440208, "learning_rate": 1.210934285009433e-06, "loss": 0.5032, "step": 23759 }, { "epoch": 0.8465005255001158, "grad_norm": 1.1406491634122025, "learning_rate": 1.2103839226920954e-06, "loss": 0.3942, "step": 23760 }, { "epoch": 0.8465361526266099, "grad_norm": 1.1672471647612588, "learning_rate": 1.209833677414637e-06, "loss": 0.3722, "step": 23761 }, { "epoch": 0.846571779753104, "grad_norm": 1.506290424488162, "learning_rate": 1.209283549184389e-06, "loss": 0.5529, "step": 23762 }, { "epoch": 0.8466074068795981, "grad_norm": 1.2751362445766061, "learning_rate": 1.2087335380086728e-06, "loss": 0.3984, "step": 23763 }, { "epoch": 0.8466430340060922, "grad_norm": 1.1626611109762006, "learning_rate": 1.2081836438948147e-06, "loss": 0.4127, "step": 23764 }, { "epoch": 0.8466786611325864, "grad_norm": 1.3002831794452407, "learning_rate": 1.2076338668501374e-06, "loss": 0.4923, "step": 23765 }, { "epoch": 0.8467142882590805, "grad_norm": 1.715076543912799, "learning_rate": 1.2070842068819599e-06, "loss": 0.5639, "step": 23766 }, { "epoch": 0.8467499153855745, "grad_norm": 1.3421972365007115, "learning_rate": 1.2065346639976016e-06, "loss": 0.3982, "step": 23767 }, { "epoch": 0.8467855425120687, "grad_norm": 1.3004774210663053, "learning_rate": 1.2059852382043802e-06, "loss": 0.4202, "step": 23768 }, { "epoch": 0.8468211696385628, "grad_norm": 1.34711987530281, "learning_rate": 1.2054359295096141e-06, "loss": 0.409, "step": 23769 }, { "epoch": 0.8468567967650569, "grad_norm": 1.1117960573793777, "learning_rate": 1.2048867379206142e-06, "loss": 0.3506, "step": 23770 }, { "epoch": 0.8468924238915511, "grad_norm": 1.7551006129016555, "learning_rate": 1.2043376634446967e-06, "loss": 0.4358, "step": 23771 }, { "epoch": 0.8469280510180451, "grad_norm": 1.3405764115696828, "learning_rate": 1.2037887060891683e-06, "loss": 0.4011, "step": 23772 }, { "epoch": 0.8469636781445392, "grad_norm": 1.4975061468764632, "learning_rate": 1.203239865861342e-06, "loss": 0.4859, "step": 23773 }, { "epoch": 0.8469993052710334, "grad_norm": 1.2201902491733956, "learning_rate": 1.202691142768525e-06, "loss": 0.4061, "step": 23774 }, { "epoch": 0.8470349323975275, "grad_norm": 1.610130829977248, "learning_rate": 1.202142536818025e-06, "loss": 0.4345, "step": 23775 }, { "epoch": 0.8470705595240215, "grad_norm": 1.548894286710987, "learning_rate": 1.2015940480171485e-06, "loss": 0.4699, "step": 23776 }, { "epoch": 0.8471061866505157, "grad_norm": 1.273366516219027, "learning_rate": 1.2010456763731948e-06, "loss": 0.4085, "step": 23777 }, { "epoch": 0.8471418137770098, "grad_norm": 1.3406897703796652, "learning_rate": 1.2004974218934695e-06, "loss": 0.3897, "step": 23778 }, { "epoch": 0.8471774409035039, "grad_norm": 1.757041341456946, "learning_rate": 1.1999492845852701e-06, "loss": 0.5755, "step": 23779 }, { "epoch": 0.8472130680299981, "grad_norm": 1.4407807574382219, "learning_rate": 1.1994012644558984e-06, "loss": 0.4832, "step": 23780 }, { "epoch": 0.8472486951564921, "grad_norm": 1.3511485250375863, "learning_rate": 1.1988533615126463e-06, "loss": 0.4385, "step": 23781 }, { "epoch": 0.8472843222829862, "grad_norm": 1.8310466422124527, "learning_rate": 1.198305575762817e-06, "loss": 0.5275, "step": 23782 }, { "epoch": 0.8473199494094804, "grad_norm": 1.1967644776500925, "learning_rate": 1.1977579072137001e-06, "loss": 0.3948, "step": 23783 }, { "epoch": 0.8473555765359745, "grad_norm": 1.1590606908507404, "learning_rate": 1.19721035587259e-06, "loss": 0.4225, "step": 23784 }, { "epoch": 0.8473912036624686, "grad_norm": 1.567029290050627, "learning_rate": 1.1966629217467751e-06, "loss": 0.4194, "step": 23785 }, { "epoch": 0.8474268307889627, "grad_norm": 1.2463760770159273, "learning_rate": 1.1961156048435484e-06, "loss": 0.4246, "step": 23786 }, { "epoch": 0.8474624579154568, "grad_norm": 1.3872197715140853, "learning_rate": 1.1955684051701944e-06, "loss": 0.4669, "step": 23787 }, { "epoch": 0.8474980850419509, "grad_norm": 1.1709623455682097, "learning_rate": 1.1950213227340013e-06, "loss": 0.4021, "step": 23788 }, { "epoch": 0.8475337121684451, "grad_norm": 1.6007975327294712, "learning_rate": 1.1944743575422523e-06, "loss": 0.5462, "step": 23789 }, { "epoch": 0.8475693392949392, "grad_norm": 1.3911174039244392, "learning_rate": 1.1939275096022329e-06, "loss": 0.4401, "step": 23790 }, { "epoch": 0.8476049664214332, "grad_norm": 1.2122559593189204, "learning_rate": 1.1933807789212259e-06, "loss": 0.4221, "step": 23791 }, { "epoch": 0.8476405935479274, "grad_norm": 1.1978651565766194, "learning_rate": 1.1928341655065068e-06, "loss": 0.3849, "step": 23792 }, { "epoch": 0.8476762206744215, "grad_norm": 1.1050404456809675, "learning_rate": 1.1922876693653584e-06, "loss": 0.3793, "step": 23793 }, { "epoch": 0.8477118478009156, "grad_norm": 1.511991258122478, "learning_rate": 1.191741290505055e-06, "loss": 0.5084, "step": 23794 }, { "epoch": 0.8477474749274098, "grad_norm": 1.1433683560273462, "learning_rate": 1.1911950289328722e-06, "loss": 0.3826, "step": 23795 }, { "epoch": 0.8477831020539038, "grad_norm": 1.4158840401553858, "learning_rate": 1.1906488846560837e-06, "loss": 0.5383, "step": 23796 }, { "epoch": 0.8478187291803979, "grad_norm": 1.6497388696082425, "learning_rate": 1.1901028576819663e-06, "loss": 0.5514, "step": 23797 }, { "epoch": 0.8478543563068921, "grad_norm": 1.4233616976901906, "learning_rate": 1.1895569480177838e-06, "loss": 0.5939, "step": 23798 }, { "epoch": 0.8478899834333862, "grad_norm": 1.145987141716955, "learning_rate": 1.1890111556708095e-06, "loss": 0.3958, "step": 23799 }, { "epoch": 0.8479256105598802, "grad_norm": 1.187432228082641, "learning_rate": 1.188465480648312e-06, "loss": 0.3697, "step": 23800 }, { "epoch": 0.8479612376863744, "grad_norm": 1.2781058943635322, "learning_rate": 1.1879199229575533e-06, "loss": 0.3677, "step": 23801 }, { "epoch": 0.8479968648128685, "grad_norm": 1.2391193766424822, "learning_rate": 1.187374482605802e-06, "loss": 0.3758, "step": 23802 }, { "epoch": 0.8480324919393626, "grad_norm": 1.3564356983215364, "learning_rate": 1.1868291596003179e-06, "loss": 0.481, "step": 23803 }, { "epoch": 0.8480681190658568, "grad_norm": 1.569203101401124, "learning_rate": 1.1862839539483628e-06, "loss": 0.6551, "step": 23804 }, { "epoch": 0.8481037461923508, "grad_norm": 1.2535374453532881, "learning_rate": 1.1857388656571978e-06, "loss": 0.4227, "step": 23805 }, { "epoch": 0.8481393733188449, "grad_norm": 1.1980042558797648, "learning_rate": 1.1851938947340824e-06, "loss": 0.4271, "step": 23806 }, { "epoch": 0.8481750004453391, "grad_norm": 1.3057695562580969, "learning_rate": 1.1846490411862689e-06, "loss": 0.481, "step": 23807 }, { "epoch": 0.8482106275718332, "grad_norm": 1.5150302952385233, "learning_rate": 1.1841043050210177e-06, "loss": 0.5065, "step": 23808 }, { "epoch": 0.8482462546983273, "grad_norm": 1.0767162556141554, "learning_rate": 1.1835596862455767e-06, "loss": 0.3455, "step": 23809 }, { "epoch": 0.8482818818248214, "grad_norm": 1.2470562966559573, "learning_rate": 1.1830151848672022e-06, "loss": 0.358, "step": 23810 }, { "epoch": 0.8483175089513155, "grad_norm": 2.2881119287079352, "learning_rate": 1.1824708008931419e-06, "loss": 0.4881, "step": 23811 }, { "epoch": 0.8483531360778096, "grad_norm": 1.5762009125163636, "learning_rate": 1.1819265343306463e-06, "loss": 0.5145, "step": 23812 }, { "epoch": 0.8483887632043038, "grad_norm": 1.3344147641488309, "learning_rate": 1.1813823851869656e-06, "loss": 0.4042, "step": 23813 }, { "epoch": 0.8484243903307979, "grad_norm": 1.1043395831729044, "learning_rate": 1.1808383534693391e-06, "loss": 0.4285, "step": 23814 }, { "epoch": 0.8484600174572919, "grad_norm": 1.3286162811471878, "learning_rate": 1.1802944391850157e-06, "loss": 0.4329, "step": 23815 }, { "epoch": 0.8484956445837861, "grad_norm": 1.759909086262119, "learning_rate": 1.1797506423412353e-06, "loss": 0.5512, "step": 23816 }, { "epoch": 0.8485312717102802, "grad_norm": 1.67061446885753, "learning_rate": 1.1792069629452419e-06, "loss": 0.4835, "step": 23817 }, { "epoch": 0.8485668988367743, "grad_norm": 1.2812362032091236, "learning_rate": 1.178663401004272e-06, "loss": 0.4204, "step": 23818 }, { "epoch": 0.8486025259632685, "grad_norm": 1.3389820769219942, "learning_rate": 1.178119956525563e-06, "loss": 0.4276, "step": 23819 }, { "epoch": 0.8486381530897625, "grad_norm": 1.52339643143096, "learning_rate": 1.1775766295163538e-06, "loss": 0.4141, "step": 23820 }, { "epoch": 0.8486737802162566, "grad_norm": 1.492443898879459, "learning_rate": 1.1770334199838784e-06, "loss": 0.4957, "step": 23821 }, { "epoch": 0.8487094073427508, "grad_norm": 1.215279029409632, "learning_rate": 1.1764903279353711e-06, "loss": 0.3566, "step": 23822 }, { "epoch": 0.8487450344692449, "grad_norm": 1.5856513056694177, "learning_rate": 1.175947353378062e-06, "loss": 0.5164, "step": 23823 }, { "epoch": 0.848780661595739, "grad_norm": 1.43462430692095, "learning_rate": 1.1754044963191824e-06, "loss": 0.4484, "step": 23824 }, { "epoch": 0.8488162887222331, "grad_norm": 1.2179104133579082, "learning_rate": 1.174861756765957e-06, "loss": 0.4196, "step": 23825 }, { "epoch": 0.8488519158487272, "grad_norm": 1.1365339004037116, "learning_rate": 1.1743191347256178e-06, "loss": 0.4137, "step": 23826 }, { "epoch": 0.8488875429752213, "grad_norm": 1.6413563354418936, "learning_rate": 1.1737766302053867e-06, "loss": 0.5312, "step": 23827 }, { "epoch": 0.8489231701017155, "grad_norm": 1.3848488045574072, "learning_rate": 1.173234243212491e-06, "loss": 0.471, "step": 23828 }, { "epoch": 0.8489587972282095, "grad_norm": 1.6710496233374657, "learning_rate": 1.1726919737541499e-06, "loss": 0.4836, "step": 23829 }, { "epoch": 0.8489944243547036, "grad_norm": 1.643815711839589, "learning_rate": 1.172149821837586e-06, "loss": 0.5567, "step": 23830 }, { "epoch": 0.8490300514811978, "grad_norm": 1.1231856551926358, "learning_rate": 1.171607787470016e-06, "loss": 0.3916, "step": 23831 }, { "epoch": 0.8490656786076919, "grad_norm": 1.284868776360616, "learning_rate": 1.1710658706586587e-06, "loss": 0.3854, "step": 23832 }, { "epoch": 0.849101305734186, "grad_norm": 1.3107127547968636, "learning_rate": 1.1705240714107301e-06, "loss": 0.4215, "step": 23833 }, { "epoch": 0.8491369328606801, "grad_norm": 1.584630444324055, "learning_rate": 1.169982389733445e-06, "loss": 0.466, "step": 23834 }, { "epoch": 0.8491725599871742, "grad_norm": 1.5724385894618358, "learning_rate": 1.1694408256340196e-06, "loss": 0.4355, "step": 23835 }, { "epoch": 0.8492081871136683, "grad_norm": 1.2162564165771084, "learning_rate": 1.168899379119658e-06, "loss": 0.4125, "step": 23836 }, { "epoch": 0.8492438142401625, "grad_norm": 1.670924243408391, "learning_rate": 1.1683580501975766e-06, "loss": 0.5838, "step": 23837 }, { "epoch": 0.8492794413666566, "grad_norm": 1.5517863571918071, "learning_rate": 1.1678168388749788e-06, "loss": 0.4998, "step": 23838 }, { "epoch": 0.8493150684931506, "grad_norm": 1.4295146349735801, "learning_rate": 1.1672757451590756e-06, "loss": 0.5272, "step": 23839 }, { "epoch": 0.8493506956196448, "grad_norm": 1.1770277204467938, "learning_rate": 1.1667347690570663e-06, "loss": 0.4005, "step": 23840 }, { "epoch": 0.8493863227461389, "grad_norm": 1.228221140400019, "learning_rate": 1.1661939105761576e-06, "loss": 0.4244, "step": 23841 }, { "epoch": 0.849421949872633, "grad_norm": 1.0880698494087686, "learning_rate": 1.1656531697235529e-06, "loss": 0.3689, "step": 23842 }, { "epoch": 0.8494575769991272, "grad_norm": 1.297284895572292, "learning_rate": 1.1651125465064516e-06, "loss": 0.4649, "step": 23843 }, { "epoch": 0.8494932041256212, "grad_norm": 1.191627863895662, "learning_rate": 1.1645720409320505e-06, "loss": 0.4093, "step": 23844 }, { "epoch": 0.8495288312521153, "grad_norm": 1.4222976825428315, "learning_rate": 1.1640316530075491e-06, "loss": 0.4946, "step": 23845 }, { "epoch": 0.8495644583786095, "grad_norm": 1.2363954217878967, "learning_rate": 1.163491382740143e-06, "loss": 0.3975, "step": 23846 }, { "epoch": 0.8496000855051036, "grad_norm": 1.4645756369271694, "learning_rate": 1.162951230137024e-06, "loss": 0.5043, "step": 23847 }, { "epoch": 0.8496357126315977, "grad_norm": 1.247971535005702, "learning_rate": 1.1624111952053862e-06, "loss": 0.4527, "step": 23848 }, { "epoch": 0.8496713397580918, "grad_norm": 1.467569405980959, "learning_rate": 1.1618712779524199e-06, "loss": 0.4921, "step": 23849 }, { "epoch": 0.8497069668845859, "grad_norm": 1.3041033089820275, "learning_rate": 1.161331478385318e-06, "loss": 0.4008, "step": 23850 }, { "epoch": 0.84974259401108, "grad_norm": 1.2563384085110703, "learning_rate": 1.1607917965112626e-06, "loss": 0.3943, "step": 23851 }, { "epoch": 0.8497782211375742, "grad_norm": 1.118940617396282, "learning_rate": 1.1602522323374455e-06, "loss": 0.3751, "step": 23852 }, { "epoch": 0.8498138482640683, "grad_norm": 1.3262094161680908, "learning_rate": 1.1597127858710455e-06, "loss": 0.4024, "step": 23853 }, { "epoch": 0.8498494753905623, "grad_norm": 1.2866697475325426, "learning_rate": 1.1591734571192515e-06, "loss": 0.387, "step": 23854 }, { "epoch": 0.8498851025170565, "grad_norm": 1.3738082878547995, "learning_rate": 1.1586342460892397e-06, "loss": 0.4941, "step": 23855 }, { "epoch": 0.8499207296435506, "grad_norm": 1.107053696109491, "learning_rate": 1.1580951527881902e-06, "loss": 0.4353, "step": 23856 }, { "epoch": 0.8499563567700447, "grad_norm": 1.6052699823430927, "learning_rate": 1.157556177223289e-06, "loss": 0.4722, "step": 23857 }, { "epoch": 0.8499919838965388, "grad_norm": 1.3174349447252884, "learning_rate": 1.1570173194017053e-06, "loss": 0.4384, "step": 23858 }, { "epoch": 0.8500276110230329, "grad_norm": 1.7812712311900802, "learning_rate": 1.1564785793306187e-06, "loss": 0.5088, "step": 23859 }, { "epoch": 0.850063238149527, "grad_norm": 1.1668615034933312, "learning_rate": 1.1559399570171992e-06, "loss": 0.4395, "step": 23860 }, { "epoch": 0.8500988652760212, "grad_norm": 1.410502902720788, "learning_rate": 1.155401452468623e-06, "loss": 0.4172, "step": 23861 }, { "epoch": 0.8501344924025153, "grad_norm": 1.3180228990637823, "learning_rate": 1.1548630656920567e-06, "loss": 0.4005, "step": 23862 }, { "epoch": 0.8501701195290093, "grad_norm": 1.1416374757338525, "learning_rate": 1.1543247966946702e-06, "loss": 0.3829, "step": 23863 }, { "epoch": 0.8502057466555035, "grad_norm": 1.1581902444918524, "learning_rate": 1.1537866454836321e-06, "loss": 0.4309, "step": 23864 }, { "epoch": 0.8502413737819976, "grad_norm": 1.3693939769114873, "learning_rate": 1.1532486120661103e-06, "loss": 0.5257, "step": 23865 }, { "epoch": 0.8502770009084917, "grad_norm": 2.2118822808915737, "learning_rate": 1.152710696449264e-06, "loss": 0.4717, "step": 23866 }, { "epoch": 0.8503126280349859, "grad_norm": 1.4598351440543593, "learning_rate": 1.1521728986402613e-06, "loss": 0.4406, "step": 23867 }, { "epoch": 0.8503482551614799, "grad_norm": 1.2315600565164704, "learning_rate": 1.1516352186462588e-06, "loss": 0.3992, "step": 23868 }, { "epoch": 0.850383882287974, "grad_norm": 1.1415291282011522, "learning_rate": 1.1510976564744181e-06, "loss": 0.393, "step": 23869 }, { "epoch": 0.8504195094144682, "grad_norm": 1.3942947931279346, "learning_rate": 1.1505602121318971e-06, "loss": 0.3895, "step": 23870 }, { "epoch": 0.8504551365409623, "grad_norm": 1.39449194545089, "learning_rate": 1.1500228856258522e-06, "loss": 0.4107, "step": 23871 }, { "epoch": 0.8504907636674564, "grad_norm": 1.2685128345094419, "learning_rate": 1.14948567696344e-06, "loss": 0.4161, "step": 23872 }, { "epoch": 0.8505263907939505, "grad_norm": 1.5143552357911405, "learning_rate": 1.14894858615181e-06, "loss": 0.5634, "step": 23873 }, { "epoch": 0.8505620179204446, "grad_norm": 1.633264881382683, "learning_rate": 1.1484116131981182e-06, "loss": 0.5604, "step": 23874 }, { "epoch": 0.8505976450469387, "grad_norm": 1.5850686636343165, "learning_rate": 1.1478747581095107e-06, "loss": 0.4411, "step": 23875 }, { "epoch": 0.8506332721734329, "grad_norm": 1.3097807690909127, "learning_rate": 1.1473380208931406e-06, "loss": 0.4182, "step": 23876 }, { "epoch": 0.850668899299927, "grad_norm": 1.2329047213176163, "learning_rate": 1.1468014015561503e-06, "loss": 0.4499, "step": 23877 }, { "epoch": 0.850704526426421, "grad_norm": 1.3521542873019148, "learning_rate": 1.146264900105687e-06, "loss": 0.4131, "step": 23878 }, { "epoch": 0.8507401535529152, "grad_norm": 1.4833692724224397, "learning_rate": 1.1457285165488951e-06, "loss": 0.5093, "step": 23879 }, { "epoch": 0.8507757806794093, "grad_norm": 1.2797762256014724, "learning_rate": 1.145192250892917e-06, "loss": 0.4005, "step": 23880 }, { "epoch": 0.8508114078059034, "grad_norm": 1.3734356830706655, "learning_rate": 1.1446561031448944e-06, "loss": 0.4177, "step": 23881 }, { "epoch": 0.8508470349323976, "grad_norm": 1.6266571097714493, "learning_rate": 1.144120073311964e-06, "loss": 0.5418, "step": 23882 }, { "epoch": 0.8508826620588916, "grad_norm": 1.5769174816119607, "learning_rate": 1.1435841614012666e-06, "loss": 0.5269, "step": 23883 }, { "epoch": 0.8509182891853857, "grad_norm": 1.3691839295483714, "learning_rate": 1.1430483674199345e-06, "loss": 0.4418, "step": 23884 }, { "epoch": 0.8509539163118799, "grad_norm": 1.212040084352433, "learning_rate": 1.1425126913751029e-06, "loss": 0.4103, "step": 23885 }, { "epoch": 0.850989543438374, "grad_norm": 1.2132510522383138, "learning_rate": 1.1419771332739072e-06, "loss": 0.4002, "step": 23886 }, { "epoch": 0.851025170564868, "grad_norm": 1.3196167407390798, "learning_rate": 1.1414416931234785e-06, "loss": 0.4571, "step": 23887 }, { "epoch": 0.8510607976913622, "grad_norm": 1.7186639938702695, "learning_rate": 1.1409063709309442e-06, "loss": 0.5761, "step": 23888 }, { "epoch": 0.8510964248178563, "grad_norm": 1.5038550671808086, "learning_rate": 1.1403711667034356e-06, "loss": 0.477, "step": 23889 }, { "epoch": 0.8511320519443504, "grad_norm": 1.2443559674785327, "learning_rate": 1.1398360804480746e-06, "loss": 0.4144, "step": 23890 }, { "epoch": 0.8511676790708446, "grad_norm": 2.190114287644248, "learning_rate": 1.1393011121719899e-06, "loss": 0.5368, "step": 23891 }, { "epoch": 0.8512033061973386, "grad_norm": 1.259593528010642, "learning_rate": 1.1387662618823058e-06, "loss": 0.4049, "step": 23892 }, { "epoch": 0.8512389333238327, "grad_norm": 1.3413739450547275, "learning_rate": 1.1382315295861401e-06, "loss": 0.4234, "step": 23893 }, { "epoch": 0.8512745604503269, "grad_norm": 1.327947942533025, "learning_rate": 1.137696915290617e-06, "loss": 0.4503, "step": 23894 }, { "epoch": 0.851310187576821, "grad_norm": 1.2068119560545478, "learning_rate": 1.137162419002853e-06, "loss": 0.4067, "step": 23895 }, { "epoch": 0.851345814703315, "grad_norm": 1.186247399046983, "learning_rate": 1.1366280407299678e-06, "loss": 0.4288, "step": 23896 }, { "epoch": 0.8513814418298092, "grad_norm": 1.69840822085792, "learning_rate": 1.136093780479074e-06, "loss": 0.474, "step": 23897 }, { "epoch": 0.8514170689563033, "grad_norm": 1.0551690223352035, "learning_rate": 1.1355596382572887e-06, "loss": 0.3833, "step": 23898 }, { "epoch": 0.8514526960827974, "grad_norm": 1.2042251826046768, "learning_rate": 1.1350256140717196e-06, "loss": 0.4045, "step": 23899 }, { "epoch": 0.8514883232092916, "grad_norm": 1.272456408040733, "learning_rate": 1.1344917079294827e-06, "loss": 0.4161, "step": 23900 }, { "epoch": 0.8515239503357857, "grad_norm": 3.8615168692703206, "learning_rate": 1.1339579198376838e-06, "loss": 0.4762, "step": 23901 }, { "epoch": 0.8515595774622797, "grad_norm": 1.2548422372925043, "learning_rate": 1.1334242498034332e-06, "loss": 0.3794, "step": 23902 }, { "epoch": 0.8515952045887739, "grad_norm": 1.4285304273014736, "learning_rate": 1.1328906978338372e-06, "loss": 0.476, "step": 23903 }, { "epoch": 0.851630831715268, "grad_norm": 1.4934757832490178, "learning_rate": 1.1323572639359981e-06, "loss": 0.543, "step": 23904 }, { "epoch": 0.8516664588417621, "grad_norm": 1.524014610602978, "learning_rate": 1.1318239481170224e-06, "loss": 0.5346, "step": 23905 }, { "epoch": 0.8517020859682563, "grad_norm": 1.382819751646156, "learning_rate": 1.1312907503840064e-06, "loss": 0.4814, "step": 23906 }, { "epoch": 0.8517377130947503, "grad_norm": 1.6545529915474366, "learning_rate": 1.130757670744055e-06, "loss": 0.5046, "step": 23907 }, { "epoch": 0.8517733402212444, "grad_norm": 1.4573175917533119, "learning_rate": 1.1302247092042607e-06, "loss": 0.5434, "step": 23908 }, { "epoch": 0.8518089673477386, "grad_norm": 1.1401942509995362, "learning_rate": 1.1296918657717281e-06, "loss": 0.417, "step": 23909 }, { "epoch": 0.8518445944742327, "grad_norm": 1.1064789356745681, "learning_rate": 1.1291591404535463e-06, "loss": 0.4196, "step": 23910 }, { "epoch": 0.8518802216007267, "grad_norm": 1.0490292997731636, "learning_rate": 1.1286265332568124e-06, "loss": 0.4086, "step": 23911 }, { "epoch": 0.8519158487272209, "grad_norm": 1.602306969048113, "learning_rate": 1.1280940441886156e-06, "loss": 0.4748, "step": 23912 }, { "epoch": 0.851951475853715, "grad_norm": 1.286548164846191, "learning_rate": 1.127561673256049e-06, "loss": 0.4106, "step": 23913 }, { "epoch": 0.8519871029802091, "grad_norm": 1.230975353466492, "learning_rate": 1.127029420466199e-06, "loss": 0.4222, "step": 23914 }, { "epoch": 0.8520227301067033, "grad_norm": 1.7439020195977957, "learning_rate": 1.1264972858261535e-06, "loss": 0.5766, "step": 23915 }, { "epoch": 0.8520583572331973, "grad_norm": 1.5163362886228537, "learning_rate": 1.125965269342999e-06, "loss": 0.491, "step": 23916 }, { "epoch": 0.8520939843596914, "grad_norm": 1.4627570731393025, "learning_rate": 1.1254333710238197e-06, "loss": 0.533, "step": 23917 }, { "epoch": 0.8521296114861856, "grad_norm": 1.2252354606370282, "learning_rate": 1.1249015908757e-06, "loss": 0.3799, "step": 23918 }, { "epoch": 0.8521652386126797, "grad_norm": 1.5781161822777032, "learning_rate": 1.1243699289057164e-06, "loss": 0.4461, "step": 23919 }, { "epoch": 0.8522008657391738, "grad_norm": 1.522056459635772, "learning_rate": 1.1238383851209533e-06, "loss": 0.5307, "step": 23920 }, { "epoch": 0.8522364928656679, "grad_norm": 1.4744812697178624, "learning_rate": 1.123306959528484e-06, "loss": 0.5472, "step": 23921 }, { "epoch": 0.852272119992162, "grad_norm": 1.5795620841435964, "learning_rate": 1.1227756521353873e-06, "loss": 0.4761, "step": 23922 }, { "epoch": 0.8523077471186561, "grad_norm": 1.7603894847621837, "learning_rate": 1.1222444629487372e-06, "loss": 0.5341, "step": 23923 }, { "epoch": 0.8523433742451503, "grad_norm": 1.2640799930714504, "learning_rate": 1.1217133919756096e-06, "loss": 0.4214, "step": 23924 }, { "epoch": 0.8523790013716444, "grad_norm": 1.4667677586023793, "learning_rate": 1.121182439223072e-06, "loss": 0.4858, "step": 23925 }, { "epoch": 0.8524146284981385, "grad_norm": 1.134626038400927, "learning_rate": 1.1206516046981952e-06, "loss": 0.3871, "step": 23926 }, { "epoch": 0.8524502556246326, "grad_norm": 1.162473770042145, "learning_rate": 1.1201208884080516e-06, "loss": 0.4119, "step": 23927 }, { "epoch": 0.8524858827511267, "grad_norm": 1.1734563823629915, "learning_rate": 1.1195902903597023e-06, "loss": 0.3621, "step": 23928 }, { "epoch": 0.8525215098776209, "grad_norm": 1.0707806064703427, "learning_rate": 1.119059810560218e-06, "loss": 0.3933, "step": 23929 }, { "epoch": 0.852557137004115, "grad_norm": 1.5966806981922972, "learning_rate": 1.1185294490166564e-06, "loss": 0.4919, "step": 23930 }, { "epoch": 0.852592764130609, "grad_norm": 1.4855278238126433, "learning_rate": 1.1179992057360844e-06, "loss": 0.5127, "step": 23931 }, { "epoch": 0.8526283912571032, "grad_norm": 1.2146153993059778, "learning_rate": 1.1174690807255605e-06, "loss": 0.3835, "step": 23932 }, { "epoch": 0.8526640183835973, "grad_norm": 1.4827379982065594, "learning_rate": 1.116939073992146e-06, "loss": 0.4217, "step": 23933 }, { "epoch": 0.8526996455100914, "grad_norm": 1.1977094044946823, "learning_rate": 1.116409185542896e-06, "loss": 0.4107, "step": 23934 }, { "epoch": 0.8527352726365856, "grad_norm": 1.5397166596864755, "learning_rate": 1.1158794153848674e-06, "loss": 0.5937, "step": 23935 }, { "epoch": 0.8527708997630796, "grad_norm": 1.4798479897829324, "learning_rate": 1.1153497635251131e-06, "loss": 0.4867, "step": 23936 }, { "epoch": 0.8528065268895737, "grad_norm": 1.3105217351215044, "learning_rate": 1.1148202299706857e-06, "loss": 0.4145, "step": 23937 }, { "epoch": 0.8528421540160679, "grad_norm": 1.2587112966061977, "learning_rate": 1.114290814728638e-06, "loss": 0.4048, "step": 23938 }, { "epoch": 0.852877781142562, "grad_norm": 1.4145062717233212, "learning_rate": 1.113761517806019e-06, "loss": 0.4134, "step": 23939 }, { "epoch": 0.852913408269056, "grad_norm": 1.6705189659496127, "learning_rate": 1.1132323392098777e-06, "loss": 0.5514, "step": 23940 }, { "epoch": 0.8529490353955502, "grad_norm": 1.5448580381384844, "learning_rate": 1.112703278947258e-06, "loss": 0.478, "step": 23941 }, { "epoch": 0.8529846625220443, "grad_norm": 1.1824893248605397, "learning_rate": 1.112174337025207e-06, "loss": 0.412, "step": 23942 }, { "epoch": 0.8530202896485384, "grad_norm": 1.5770130252240109, "learning_rate": 1.1116455134507665e-06, "loss": 0.5585, "step": 23943 }, { "epoch": 0.8530559167750326, "grad_norm": 1.2261483953936947, "learning_rate": 1.1111168082309798e-06, "loss": 0.3998, "step": 23944 }, { "epoch": 0.8530915439015266, "grad_norm": 1.6077781217310432, "learning_rate": 1.1105882213728814e-06, "loss": 0.477, "step": 23945 }, { "epoch": 0.8531271710280207, "grad_norm": 1.2142496501154851, "learning_rate": 1.1100597528835177e-06, "loss": 0.3839, "step": 23946 }, { "epoch": 0.8531627981545149, "grad_norm": 1.4671839780703193, "learning_rate": 1.109531402769921e-06, "loss": 0.4515, "step": 23947 }, { "epoch": 0.853198425281009, "grad_norm": 1.1935366186562895, "learning_rate": 1.1090031710391281e-06, "loss": 0.4233, "step": 23948 }, { "epoch": 0.853234052407503, "grad_norm": 1.5247432127469556, "learning_rate": 1.108475057698174e-06, "loss": 0.5388, "step": 23949 }, { "epoch": 0.8532696795339972, "grad_norm": 1.368509918924408, "learning_rate": 1.1079470627540866e-06, "loss": 0.5751, "step": 23950 }, { "epoch": 0.8533053066604913, "grad_norm": 1.218816469243992, "learning_rate": 1.1074191862139016e-06, "loss": 0.4344, "step": 23951 }, { "epoch": 0.8533409337869854, "grad_norm": 1.2291190564278076, "learning_rate": 1.1068914280846444e-06, "loss": 0.4397, "step": 23952 }, { "epoch": 0.8533765609134796, "grad_norm": 1.431161076868823, "learning_rate": 1.1063637883733436e-06, "loss": 0.4695, "step": 23953 }, { "epoch": 0.8534121880399737, "grad_norm": 1.6469556249530553, "learning_rate": 1.1058362670870248e-06, "loss": 0.5731, "step": 23954 }, { "epoch": 0.8534478151664677, "grad_norm": 1.5148646579605995, "learning_rate": 1.1053088642327147e-06, "loss": 0.466, "step": 23955 }, { "epoch": 0.8534834422929619, "grad_norm": 1.303978002668739, "learning_rate": 1.104781579817432e-06, "loss": 0.3869, "step": 23956 }, { "epoch": 0.853519069419456, "grad_norm": 1.273989399939745, "learning_rate": 1.1042544138482026e-06, "loss": 0.4412, "step": 23957 }, { "epoch": 0.8535546965459501, "grad_norm": 1.2411404415285237, "learning_rate": 1.1037273663320402e-06, "loss": 0.3795, "step": 23958 }, { "epoch": 0.8535903236724443, "grad_norm": 1.3122652259542673, "learning_rate": 1.1032004372759674e-06, "loss": 0.4243, "step": 23959 }, { "epoch": 0.8536259507989383, "grad_norm": 1.89786292620159, "learning_rate": 1.1026736266869987e-06, "loss": 0.4925, "step": 23960 }, { "epoch": 0.8536615779254324, "grad_norm": 1.482539588032561, "learning_rate": 1.1021469345721492e-06, "loss": 0.5578, "step": 23961 }, { "epoch": 0.8536972050519266, "grad_norm": 1.5079989435578567, "learning_rate": 1.1016203609384347e-06, "loss": 0.5535, "step": 23962 }, { "epoch": 0.8537328321784207, "grad_norm": 1.1656046202681956, "learning_rate": 1.1010939057928626e-06, "loss": 0.4012, "step": 23963 }, { "epoch": 0.8537684593049147, "grad_norm": 1.4936097318674226, "learning_rate": 1.1005675691424477e-06, "loss": 0.5157, "step": 23964 }, { "epoch": 0.8538040864314089, "grad_norm": 1.5237332759326097, "learning_rate": 1.1000413509941932e-06, "loss": 0.4647, "step": 23965 }, { "epoch": 0.853839713557903, "grad_norm": 1.2636527151122374, "learning_rate": 1.0995152513551111e-06, "loss": 0.4424, "step": 23966 }, { "epoch": 0.8538753406843971, "grad_norm": 1.3877367671272323, "learning_rate": 1.0989892702322025e-06, "loss": 0.4277, "step": 23967 }, { "epoch": 0.8539109678108913, "grad_norm": 1.6282856118651516, "learning_rate": 1.098463407632474e-06, "loss": 0.5216, "step": 23968 }, { "epoch": 0.8539465949373853, "grad_norm": 1.3256530305557397, "learning_rate": 1.0979376635629256e-06, "loss": 0.3763, "step": 23969 }, { "epoch": 0.8539822220638794, "grad_norm": 1.1966183038218463, "learning_rate": 1.097412038030563e-06, "loss": 0.3856, "step": 23970 }, { "epoch": 0.8540178491903736, "grad_norm": 1.4834395330339893, "learning_rate": 1.096886531042378e-06, "loss": 0.5058, "step": 23971 }, { "epoch": 0.8540534763168677, "grad_norm": 1.2296382551679759, "learning_rate": 1.0963611426053722e-06, "loss": 0.4093, "step": 23972 }, { "epoch": 0.8540891034433618, "grad_norm": 1.3337649932279267, "learning_rate": 1.0958358727265438e-06, "loss": 0.4667, "step": 23973 }, { "epoch": 0.8541247305698559, "grad_norm": 1.5541027714674807, "learning_rate": 1.0953107214128821e-06, "loss": 0.5035, "step": 23974 }, { "epoch": 0.85416035769635, "grad_norm": 1.491458450516399, "learning_rate": 1.0947856886713814e-06, "loss": 0.474, "step": 23975 }, { "epoch": 0.8541959848228441, "grad_norm": 1.7379511923295707, "learning_rate": 1.094260774509035e-06, "loss": 0.6055, "step": 23976 }, { "epoch": 0.8542316119493383, "grad_norm": 1.1375476240164282, "learning_rate": 1.0937359789328317e-06, "loss": 0.3592, "step": 23977 }, { "epoch": 0.8542672390758324, "grad_norm": 1.3201867321174452, "learning_rate": 1.0932113019497581e-06, "loss": 0.4246, "step": 23978 }, { "epoch": 0.8543028662023264, "grad_norm": 1.5055671931176908, "learning_rate": 1.0926867435668031e-06, "loss": 0.4781, "step": 23979 }, { "epoch": 0.8543384933288206, "grad_norm": 1.2080828237334835, "learning_rate": 1.092162303790948e-06, "loss": 0.418, "step": 23980 }, { "epoch": 0.8543741204553147, "grad_norm": 1.52568746822858, "learning_rate": 1.091637982629179e-06, "loss": 0.5126, "step": 23981 }, { "epoch": 0.8544097475818088, "grad_norm": 1.4405663365691135, "learning_rate": 1.0911137800884764e-06, "loss": 0.5177, "step": 23982 }, { "epoch": 0.854445374708303, "grad_norm": 1.737090452650933, "learning_rate": 1.0905896961758179e-06, "loss": 0.5047, "step": 23983 }, { "epoch": 0.854481001834797, "grad_norm": 1.0913495495167267, "learning_rate": 1.0900657308981888e-06, "loss": 0.3695, "step": 23984 }, { "epoch": 0.8545166289612911, "grad_norm": 1.446110935178294, "learning_rate": 1.0895418842625605e-06, "loss": 0.4781, "step": 23985 }, { "epoch": 0.8545522560877853, "grad_norm": 1.2851236579853174, "learning_rate": 1.0890181562759117e-06, "loss": 0.4147, "step": 23986 }, { "epoch": 0.8545878832142794, "grad_norm": 1.3416350451720136, "learning_rate": 1.0884945469452125e-06, "loss": 0.3786, "step": 23987 }, { "epoch": 0.8546235103407734, "grad_norm": 1.6625778109737064, "learning_rate": 1.0879710562774404e-06, "loss": 0.5295, "step": 23988 }, { "epoch": 0.8546591374672676, "grad_norm": 1.9911702344503548, "learning_rate": 1.087447684279559e-06, "loss": 0.5906, "step": 23989 }, { "epoch": 0.8546947645937617, "grad_norm": 1.153588889086357, "learning_rate": 1.0869244309585425e-06, "loss": 0.3724, "step": 23990 }, { "epoch": 0.8547303917202558, "grad_norm": 1.4357717897111428, "learning_rate": 1.0864012963213566e-06, "loss": 0.4966, "step": 23991 }, { "epoch": 0.85476601884675, "grad_norm": 1.7722566416016872, "learning_rate": 1.08587828037497e-06, "loss": 0.5169, "step": 23992 }, { "epoch": 0.854801645973244, "grad_norm": 1.8107131071602554, "learning_rate": 1.085355383126342e-06, "loss": 0.44, "step": 23993 }, { "epoch": 0.8548372730997381, "grad_norm": 1.2120307533541679, "learning_rate": 1.084832604582441e-06, "loss": 0.4209, "step": 23994 }, { "epoch": 0.8548729002262323, "grad_norm": 1.328164795875631, "learning_rate": 1.0843099447502226e-06, "loss": 0.3257, "step": 23995 }, { "epoch": 0.8549085273527264, "grad_norm": 1.2243026593199235, "learning_rate": 1.083787403636649e-06, "loss": 0.379, "step": 23996 }, { "epoch": 0.8549441544792205, "grad_norm": 1.2104097588515117, "learning_rate": 1.0832649812486785e-06, "loss": 0.4393, "step": 23997 }, { "epoch": 0.8549797816057146, "grad_norm": 1.2955276359224626, "learning_rate": 1.0827426775932658e-06, "loss": 0.4164, "step": 23998 }, { "epoch": 0.8550154087322087, "grad_norm": 5.295767683421274, "learning_rate": 1.0822204926773706e-06, "loss": 0.6132, "step": 23999 }, { "epoch": 0.8550510358587028, "grad_norm": 1.3011314212224396, "learning_rate": 1.081698426507941e-06, "loss": 0.3772, "step": 24000 }, { "epoch": 0.855086662985197, "grad_norm": 1.3607601323402272, "learning_rate": 1.081176479091931e-06, "loss": 0.5503, "step": 24001 }, { "epoch": 0.855122290111691, "grad_norm": 1.6279836555565412, "learning_rate": 1.0806546504362892e-06, "loss": 0.4774, "step": 24002 }, { "epoch": 0.8551579172381851, "grad_norm": 1.1079284985936577, "learning_rate": 1.0801329405479667e-06, "loss": 0.3872, "step": 24003 }, { "epoch": 0.8551935443646793, "grad_norm": 1.2965914234753273, "learning_rate": 1.0796113494339066e-06, "loss": 0.3809, "step": 24004 }, { "epoch": 0.8552291714911734, "grad_norm": 1.20789973332079, "learning_rate": 1.079089877101056e-06, "loss": 0.3838, "step": 24005 }, { "epoch": 0.8552647986176675, "grad_norm": 1.5477238932810502, "learning_rate": 1.0785685235563592e-06, "loss": 0.4868, "step": 24006 }, { "epoch": 0.8553004257441617, "grad_norm": 1.580834830553292, "learning_rate": 1.0780472888067583e-06, "loss": 0.3791, "step": 24007 }, { "epoch": 0.8553360528706557, "grad_norm": 1.2659581119564365, "learning_rate": 1.077526172859197e-06, "loss": 0.3979, "step": 24008 }, { "epoch": 0.8553716799971498, "grad_norm": 1.3320606115639837, "learning_rate": 1.0770051757206078e-06, "loss": 0.4763, "step": 24009 }, { "epoch": 0.855407307123644, "grad_norm": 1.2376645540158508, "learning_rate": 1.0764842973979338e-06, "loss": 0.4192, "step": 24010 }, { "epoch": 0.8554429342501381, "grad_norm": 1.7763140594406166, "learning_rate": 1.075963537898106e-06, "loss": 0.5254, "step": 24011 }, { "epoch": 0.8554785613766321, "grad_norm": 1.3468794689705161, "learning_rate": 1.0754428972280617e-06, "loss": 0.4151, "step": 24012 }, { "epoch": 0.8555141885031263, "grad_norm": 1.6523697189068438, "learning_rate": 1.0749223753947335e-06, "loss": 0.433, "step": 24013 }, { "epoch": 0.8555498156296204, "grad_norm": 2.2930462272903163, "learning_rate": 1.0744019724050537e-06, "loss": 0.5482, "step": 24014 }, { "epoch": 0.8555854427561145, "grad_norm": 1.2718895491906426, "learning_rate": 1.0738816882659487e-06, "loss": 0.3816, "step": 24015 }, { "epoch": 0.8556210698826087, "grad_norm": 1.1656722215743318, "learning_rate": 1.0733615229843486e-06, "loss": 0.4089, "step": 24016 }, { "epoch": 0.8556566970091027, "grad_norm": 1.5946177546891427, "learning_rate": 1.072841476567179e-06, "loss": 0.5131, "step": 24017 }, { "epoch": 0.8556923241355968, "grad_norm": 1.1051542347151835, "learning_rate": 1.0723215490213635e-06, "loss": 0.3865, "step": 24018 }, { "epoch": 0.855727951262091, "grad_norm": 1.2632165757019964, "learning_rate": 1.0718017403538294e-06, "loss": 0.4357, "step": 24019 }, { "epoch": 0.8557635783885851, "grad_norm": 1.5878993147737461, "learning_rate": 1.0712820505714916e-06, "loss": 0.4666, "step": 24020 }, { "epoch": 0.8557992055150792, "grad_norm": 1.1407002295365902, "learning_rate": 1.0707624796812776e-06, "loss": 0.3831, "step": 24021 }, { "epoch": 0.8558348326415733, "grad_norm": 1.1715090159041568, "learning_rate": 1.070243027690101e-06, "loss": 0.4271, "step": 24022 }, { "epoch": 0.8558704597680674, "grad_norm": 1.1587878931929148, "learning_rate": 1.0697236946048805e-06, "loss": 0.4214, "step": 24023 }, { "epoch": 0.8559060868945615, "grad_norm": 1.8229650198629699, "learning_rate": 1.0692044804325308e-06, "loss": 0.5004, "step": 24024 }, { "epoch": 0.8559417140210557, "grad_norm": 1.509563740269606, "learning_rate": 1.0686853851799672e-06, "loss": 0.4227, "step": 24025 }, { "epoch": 0.8559773411475498, "grad_norm": 1.4871312519450466, "learning_rate": 1.0681664088540977e-06, "loss": 0.5268, "step": 24026 }, { "epoch": 0.8560129682740438, "grad_norm": 1.2749378725621265, "learning_rate": 1.0676475514618356e-06, "loss": 0.4062, "step": 24027 }, { "epoch": 0.856048595400538, "grad_norm": 1.2109319383872412, "learning_rate": 1.0671288130100898e-06, "loss": 0.4079, "step": 24028 }, { "epoch": 0.8560842225270321, "grad_norm": 1.2587136133683876, "learning_rate": 1.066610193505767e-06, "loss": 0.4113, "step": 24029 }, { "epoch": 0.8561198496535262, "grad_norm": 1.1164448256208115, "learning_rate": 1.0660916929557763e-06, "loss": 0.3824, "step": 24030 }, { "epoch": 0.8561554767800204, "grad_norm": 1.4874879396373393, "learning_rate": 1.0655733113670175e-06, "loss": 0.5647, "step": 24031 }, { "epoch": 0.8561911039065144, "grad_norm": 1.152656556133211, "learning_rate": 1.0650550487463963e-06, "loss": 0.4188, "step": 24032 }, { "epoch": 0.8562267310330085, "grad_norm": 2.2563287332663755, "learning_rate": 1.0645369051008103e-06, "loss": 0.5841, "step": 24033 }, { "epoch": 0.8562623581595027, "grad_norm": 1.498761440223337, "learning_rate": 1.0640188804371609e-06, "loss": 0.5737, "step": 24034 }, { "epoch": 0.8562979852859968, "grad_norm": 1.6316323584485897, "learning_rate": 1.063500974762346e-06, "loss": 0.4576, "step": 24035 }, { "epoch": 0.8563336124124908, "grad_norm": 1.4919373083156047, "learning_rate": 1.0629831880832642e-06, "loss": 0.4493, "step": 24036 }, { "epoch": 0.856369239538985, "grad_norm": 1.6143459591906395, "learning_rate": 1.0624655204068056e-06, "loss": 0.3958, "step": 24037 }, { "epoch": 0.8564048666654791, "grad_norm": 1.25193557401423, "learning_rate": 1.0619479717398674e-06, "loss": 0.4121, "step": 24038 }, { "epoch": 0.8564404937919732, "grad_norm": 1.7623770609345943, "learning_rate": 1.0614305420893367e-06, "loss": 0.4811, "step": 24039 }, { "epoch": 0.8564761209184674, "grad_norm": 1.2677949515729432, "learning_rate": 1.0609132314621095e-06, "loss": 0.4136, "step": 24040 }, { "epoch": 0.8565117480449614, "grad_norm": 1.491078615693372, "learning_rate": 1.060396039865067e-06, "loss": 0.5203, "step": 24041 }, { "epoch": 0.8565473751714555, "grad_norm": 1.39191815812791, "learning_rate": 1.0598789673051013e-06, "loss": 0.5473, "step": 24042 }, { "epoch": 0.8565830022979497, "grad_norm": 1.7104933525801598, "learning_rate": 1.0593620137890948e-06, "loss": 0.5258, "step": 24043 }, { "epoch": 0.8566186294244438, "grad_norm": 1.608974570645298, "learning_rate": 1.058845179323933e-06, "loss": 0.567, "step": 24044 }, { "epoch": 0.8566542565509379, "grad_norm": 1.6168032202483857, "learning_rate": 1.058328463916498e-06, "loss": 0.4996, "step": 24045 }, { "epoch": 0.856689883677432, "grad_norm": 1.2066569457254068, "learning_rate": 1.0578118675736681e-06, "loss": 0.388, "step": 24046 }, { "epoch": 0.8567255108039261, "grad_norm": 1.2261283837154646, "learning_rate": 1.0572953903023254e-06, "loss": 0.3854, "step": 24047 }, { "epoch": 0.8567611379304202, "grad_norm": 1.532630045251666, "learning_rate": 1.056779032109343e-06, "loss": 0.4775, "step": 24048 }, { "epoch": 0.8567967650569144, "grad_norm": 1.212833280514819, "learning_rate": 1.056262793001599e-06, "loss": 0.4022, "step": 24049 }, { "epoch": 0.8568323921834085, "grad_norm": 1.5188407682121812, "learning_rate": 1.0557466729859667e-06, "loss": 0.5414, "step": 24050 }, { "epoch": 0.8568680193099025, "grad_norm": 1.8453042634293098, "learning_rate": 1.0552306720693218e-06, "loss": 0.5918, "step": 24051 }, { "epoch": 0.8569036464363967, "grad_norm": 1.362075683982896, "learning_rate": 1.0547147902585298e-06, "loss": 0.526, "step": 24052 }, { "epoch": 0.8569392735628908, "grad_norm": 1.3012988039215336, "learning_rate": 1.054199027560463e-06, "loss": 0.3523, "step": 24053 }, { "epoch": 0.8569749006893849, "grad_norm": 1.429397510244516, "learning_rate": 1.0536833839819915e-06, "loss": 0.4701, "step": 24054 }, { "epoch": 0.857010527815879, "grad_norm": 1.7977163727198782, "learning_rate": 1.0531678595299766e-06, "loss": 0.5131, "step": 24055 }, { "epoch": 0.8570461549423731, "grad_norm": 1.3683332327658633, "learning_rate": 1.052652454211286e-06, "loss": 0.4244, "step": 24056 }, { "epoch": 0.8570817820688672, "grad_norm": 1.2069985231430973, "learning_rate": 1.0521371680327808e-06, "loss": 0.4026, "step": 24057 }, { "epoch": 0.8571174091953614, "grad_norm": 1.7195929377881636, "learning_rate": 1.0516220010013234e-06, "loss": 0.5262, "step": 24058 }, { "epoch": 0.8571530363218555, "grad_norm": 1.4397159431107804, "learning_rate": 1.0511069531237739e-06, "loss": 0.4056, "step": 24059 }, { "epoch": 0.8571886634483495, "grad_norm": 1.4973187956293421, "learning_rate": 1.050592024406991e-06, "loss": 0.4816, "step": 24060 }, { "epoch": 0.8572242905748437, "grad_norm": 1.2510262320501404, "learning_rate": 1.0500772148578297e-06, "loss": 0.39, "step": 24061 }, { "epoch": 0.8572599177013378, "grad_norm": 1.2386682348086437, "learning_rate": 1.0495625244831475e-06, "loss": 0.4161, "step": 24062 }, { "epoch": 0.8572955448278319, "grad_norm": 1.3493832082196822, "learning_rate": 1.0490479532897946e-06, "loss": 0.4166, "step": 24063 }, { "epoch": 0.8573311719543261, "grad_norm": 1.099042949418802, "learning_rate": 1.0485335012846243e-06, "loss": 0.377, "step": 24064 }, { "epoch": 0.8573667990808201, "grad_norm": 1.244105069080103, "learning_rate": 1.0480191684744867e-06, "loss": 0.4439, "step": 24065 }, { "epoch": 0.8574024262073142, "grad_norm": 1.571975615808509, "learning_rate": 1.0475049548662308e-06, "loss": 0.5327, "step": 24066 }, { "epoch": 0.8574380533338084, "grad_norm": 3.38542782596482, "learning_rate": 1.0469908604667067e-06, "loss": 0.5095, "step": 24067 }, { "epoch": 0.8574736804603025, "grad_norm": 1.3053953654364647, "learning_rate": 1.0464768852827544e-06, "loss": 0.4113, "step": 24068 }, { "epoch": 0.8575093075867966, "grad_norm": 1.3392878067033407, "learning_rate": 1.0459630293212231e-06, "loss": 0.3693, "step": 24069 }, { "epoch": 0.8575449347132907, "grad_norm": 1.2267705314743216, "learning_rate": 1.0454492925889503e-06, "loss": 0.4301, "step": 24070 }, { "epoch": 0.8575805618397848, "grad_norm": 1.6443916516136976, "learning_rate": 1.0449356750927809e-06, "loss": 0.5232, "step": 24071 }, { "epoch": 0.8576161889662789, "grad_norm": 1.2528139564177532, "learning_rate": 1.044422176839549e-06, "loss": 0.3662, "step": 24072 }, { "epoch": 0.8576518160927731, "grad_norm": 1.833473805233498, "learning_rate": 1.0439087978360985e-06, "loss": 0.4432, "step": 24073 }, { "epoch": 0.8576874432192672, "grad_norm": 1.2014987317153907, "learning_rate": 1.0433955380892603e-06, "loss": 0.3655, "step": 24074 }, { "epoch": 0.8577230703457612, "grad_norm": 1.4471617708449849, "learning_rate": 1.042882397605871e-06, "loss": 0.4987, "step": 24075 }, { "epoch": 0.8577586974722554, "grad_norm": 1.398219970826837, "learning_rate": 1.0423693763927656e-06, "loss": 0.5499, "step": 24076 }, { "epoch": 0.8577943245987495, "grad_norm": 1.324858938201175, "learning_rate": 1.0418564744567706e-06, "loss": 0.3915, "step": 24077 }, { "epoch": 0.8578299517252436, "grad_norm": 1.8498635522308493, "learning_rate": 1.0413436918047193e-06, "loss": 0.551, "step": 24078 }, { "epoch": 0.8578655788517378, "grad_norm": 1.2057810353159877, "learning_rate": 1.0408310284434376e-06, "loss": 0.3666, "step": 24079 }, { "epoch": 0.8579012059782318, "grad_norm": 1.3422777959571037, "learning_rate": 1.0403184843797521e-06, "loss": 0.3944, "step": 24080 }, { "epoch": 0.8579368331047259, "grad_norm": 1.2224022964943668, "learning_rate": 1.0398060596204885e-06, "loss": 0.365, "step": 24081 }, { "epoch": 0.8579724602312201, "grad_norm": 1.18935927901641, "learning_rate": 1.0392937541724724e-06, "loss": 0.3666, "step": 24082 }, { "epoch": 0.8580080873577142, "grad_norm": 1.2044965970415031, "learning_rate": 1.0387815680425206e-06, "loss": 0.3788, "step": 24083 }, { "epoch": 0.8580437144842082, "grad_norm": 1.2601439972559856, "learning_rate": 1.0382695012374588e-06, "loss": 0.3871, "step": 24084 }, { "epoch": 0.8580793416107024, "grad_norm": 1.5143563006314078, "learning_rate": 1.0377575537640994e-06, "loss": 0.4927, "step": 24085 }, { "epoch": 0.8581149687371965, "grad_norm": 1.3789607183992902, "learning_rate": 1.0372457256292635e-06, "loss": 0.4256, "step": 24086 }, { "epoch": 0.8581505958636906, "grad_norm": 1.5970299945849462, "learning_rate": 1.0367340168397645e-06, "loss": 0.4994, "step": 24087 }, { "epoch": 0.8581862229901848, "grad_norm": 1.2803492229106086, "learning_rate": 1.036222427402417e-06, "loss": 0.3996, "step": 24088 }, { "epoch": 0.8582218501166788, "grad_norm": 1.9256386722202439, "learning_rate": 1.0357109573240353e-06, "loss": 0.5163, "step": 24089 }, { "epoch": 0.8582574772431729, "grad_norm": 1.5244781550220199, "learning_rate": 1.0351996066114256e-06, "loss": 0.5051, "step": 24090 }, { "epoch": 0.8582931043696671, "grad_norm": 1.3885241860066706, "learning_rate": 1.0346883752714009e-06, "loss": 0.504, "step": 24091 }, { "epoch": 0.8583287314961612, "grad_norm": 1.2069962143114943, "learning_rate": 1.034177263310766e-06, "loss": 0.3793, "step": 24092 }, { "epoch": 0.8583643586226553, "grad_norm": 1.1718810861965674, "learning_rate": 1.0336662707363287e-06, "loss": 0.4172, "step": 24093 }, { "epoch": 0.8583999857491494, "grad_norm": 1.5175598898163454, "learning_rate": 1.0331553975548902e-06, "loss": 0.4719, "step": 24094 }, { "epoch": 0.8584356128756435, "grad_norm": 1.4278205898817766, "learning_rate": 1.0326446437732552e-06, "loss": 0.5327, "step": 24095 }, { "epoch": 0.8584712400021376, "grad_norm": 1.2563341029333037, "learning_rate": 1.0321340093982245e-06, "loss": 0.396, "step": 24096 }, { "epoch": 0.8585068671286318, "grad_norm": 1.4527811063496572, "learning_rate": 1.0316234944365998e-06, "loss": 0.4605, "step": 24097 }, { "epoch": 0.8585424942551259, "grad_norm": 1.4265298673792495, "learning_rate": 1.0311130988951756e-06, "loss": 0.4829, "step": 24098 }, { "epoch": 0.8585781213816199, "grad_norm": 1.7749194648636157, "learning_rate": 1.0306028227807486e-06, "loss": 0.5501, "step": 24099 }, { "epoch": 0.8586137485081141, "grad_norm": 1.184794655226084, "learning_rate": 1.0300926661001154e-06, "loss": 0.41, "step": 24100 }, { "epoch": 0.8586493756346082, "grad_norm": 1.4219787571442195, "learning_rate": 1.0295826288600664e-06, "loss": 0.5222, "step": 24101 }, { "epoch": 0.8586850027611023, "grad_norm": 1.3024994992863212, "learning_rate": 1.0290727110673948e-06, "loss": 0.4593, "step": 24102 }, { "epoch": 0.8587206298875965, "grad_norm": 1.8065184070427236, "learning_rate": 1.028562912728891e-06, "loss": 0.4677, "step": 24103 }, { "epoch": 0.8587562570140905, "grad_norm": 1.3569673979228467, "learning_rate": 1.0280532338513437e-06, "loss": 0.3917, "step": 24104 }, { "epoch": 0.8587918841405846, "grad_norm": 1.2135686846508273, "learning_rate": 1.0275436744415379e-06, "loss": 0.447, "step": 24105 }, { "epoch": 0.8588275112670788, "grad_norm": 1.6215106416149279, "learning_rate": 1.0270342345062601e-06, "loss": 0.4958, "step": 24106 }, { "epoch": 0.8588631383935729, "grad_norm": 1.1632563104899416, "learning_rate": 1.0265249140522916e-06, "loss": 0.3677, "step": 24107 }, { "epoch": 0.8588987655200669, "grad_norm": 1.1834889657802279, "learning_rate": 1.0260157130864178e-06, "loss": 0.3983, "step": 24108 }, { "epoch": 0.8589343926465611, "grad_norm": 1.2703104387038058, "learning_rate": 1.0255066316154139e-06, "loss": 0.3605, "step": 24109 }, { "epoch": 0.8589700197730552, "grad_norm": 1.1787147107609928, "learning_rate": 1.024997669646064e-06, "loss": 0.4358, "step": 24110 }, { "epoch": 0.8590056468995493, "grad_norm": 1.7728830697021547, "learning_rate": 1.024488827185144e-06, "loss": 0.5005, "step": 24111 }, { "epoch": 0.8590412740260435, "grad_norm": 1.2456123236631313, "learning_rate": 1.0239801042394282e-06, "loss": 0.4015, "step": 24112 }, { "epoch": 0.8590769011525375, "grad_norm": 1.2148315833758756, "learning_rate": 1.0234715008156925e-06, "loss": 0.376, "step": 24113 }, { "epoch": 0.8591125282790316, "grad_norm": 1.2631421486725187, "learning_rate": 1.022963016920706e-06, "loss": 0.4328, "step": 24114 }, { "epoch": 0.8591481554055258, "grad_norm": 1.5842746802251144, "learning_rate": 1.0224546525612434e-06, "loss": 0.4602, "step": 24115 }, { "epoch": 0.8591837825320199, "grad_norm": 1.5629123870002295, "learning_rate": 1.02194640774407e-06, "loss": 0.4698, "step": 24116 }, { "epoch": 0.859219409658514, "grad_norm": 1.2375574695205318, "learning_rate": 1.0214382824759561e-06, "loss": 0.4222, "step": 24117 }, { "epoch": 0.8592550367850081, "grad_norm": 1.439783281582952, "learning_rate": 1.0209302767636664e-06, "loss": 0.4544, "step": 24118 }, { "epoch": 0.8592906639115022, "grad_norm": 1.1888250209869042, "learning_rate": 1.0204223906139688e-06, "loss": 0.3553, "step": 24119 }, { "epoch": 0.8593262910379963, "grad_norm": 1.48764470508313, "learning_rate": 1.0199146240336211e-06, "loss": 0.5163, "step": 24120 }, { "epoch": 0.8593619181644905, "grad_norm": 1.6620708413255194, "learning_rate": 1.019406977029388e-06, "loss": 0.5047, "step": 24121 }, { "epoch": 0.8593975452909846, "grad_norm": 1.2304208313127143, "learning_rate": 1.0188994496080273e-06, "loss": 0.3669, "step": 24122 }, { "epoch": 0.8594331724174786, "grad_norm": 1.9095445447073012, "learning_rate": 1.0183920417762971e-06, "loss": 0.5021, "step": 24123 }, { "epoch": 0.8594687995439728, "grad_norm": 1.5137286983810974, "learning_rate": 1.0178847535409554e-06, "loss": 0.528, "step": 24124 }, { "epoch": 0.8595044266704669, "grad_norm": 1.28202858251091, "learning_rate": 1.0173775849087552e-06, "loss": 0.418, "step": 24125 }, { "epoch": 0.859540053796961, "grad_norm": 1.3019803144322628, "learning_rate": 1.016870535886454e-06, "loss": 0.392, "step": 24126 }, { "epoch": 0.8595756809234552, "grad_norm": 1.416318425035131, "learning_rate": 1.016363606480798e-06, "loss": 0.4737, "step": 24127 }, { "epoch": 0.8596113080499492, "grad_norm": 1.1764654156075602, "learning_rate": 1.0158567966985412e-06, "loss": 0.4263, "step": 24128 }, { "epoch": 0.8596469351764433, "grad_norm": 1.2422181957077665, "learning_rate": 1.01535010654643e-06, "loss": 0.428, "step": 24129 }, { "epoch": 0.8596825623029375, "grad_norm": 1.2093267002470265, "learning_rate": 1.0148435360312125e-06, "loss": 0.4374, "step": 24130 }, { "epoch": 0.8597181894294316, "grad_norm": 1.2025394443541044, "learning_rate": 1.0143370851596324e-06, "loss": 0.433, "step": 24131 }, { "epoch": 0.8597538165559256, "grad_norm": 1.692705663838652, "learning_rate": 1.0138307539384352e-06, "loss": 0.4591, "step": 24132 }, { "epoch": 0.8597894436824198, "grad_norm": 1.3539356454362326, "learning_rate": 1.013324542374361e-06, "loss": 0.4141, "step": 24133 }, { "epoch": 0.8598250708089139, "grad_norm": 1.4499663668551692, "learning_rate": 1.0128184504741535e-06, "loss": 0.5368, "step": 24134 }, { "epoch": 0.859860697935408, "grad_norm": 1.2170845247955686, "learning_rate": 1.0123124782445505e-06, "loss": 0.4349, "step": 24135 }, { "epoch": 0.8598963250619022, "grad_norm": 1.23158346529714, "learning_rate": 1.0118066256922876e-06, "loss": 0.407, "step": 24136 }, { "epoch": 0.8599319521883962, "grad_norm": 1.2601233696056868, "learning_rate": 1.0113008928241042e-06, "loss": 0.4413, "step": 24137 }, { "epoch": 0.8599675793148903, "grad_norm": 1.470090274096328, "learning_rate": 1.010795279646729e-06, "loss": 0.4983, "step": 24138 }, { "epoch": 0.8600032064413845, "grad_norm": 1.6885398121726973, "learning_rate": 1.0102897861668993e-06, "loss": 0.404, "step": 24139 }, { "epoch": 0.8600388335678786, "grad_norm": 1.613738189259774, "learning_rate": 1.0097844123913437e-06, "loss": 0.4948, "step": 24140 }, { "epoch": 0.8600744606943727, "grad_norm": 1.4244705877333674, "learning_rate": 1.0092791583267936e-06, "loss": 0.4207, "step": 24141 }, { "epoch": 0.8601100878208668, "grad_norm": 1.3048226736944617, "learning_rate": 1.0087740239799749e-06, "loss": 0.4119, "step": 24142 }, { "epoch": 0.8601457149473609, "grad_norm": 1.3867571119426692, "learning_rate": 1.0082690093576163e-06, "loss": 0.4627, "step": 24143 }, { "epoch": 0.860181342073855, "grad_norm": 1.621911182504184, "learning_rate": 1.0077641144664386e-06, "loss": 0.5011, "step": 24144 }, { "epoch": 0.8602169692003492, "grad_norm": 1.6903695973717716, "learning_rate": 1.0072593393131668e-06, "loss": 0.6156, "step": 24145 }, { "epoch": 0.8602525963268433, "grad_norm": 2.1175766417383737, "learning_rate": 1.006754683904524e-06, "loss": 0.4853, "step": 24146 }, { "epoch": 0.8602882234533373, "grad_norm": 1.179455877870952, "learning_rate": 1.0062501482472254e-06, "loss": 0.3986, "step": 24147 }, { "epoch": 0.8603238505798315, "grad_norm": 1.8964382168123284, "learning_rate": 1.0057457323479957e-06, "loss": 0.4797, "step": 24148 }, { "epoch": 0.8603594777063256, "grad_norm": 1.8608506439871386, "learning_rate": 1.0052414362135466e-06, "loss": 0.4664, "step": 24149 }, { "epoch": 0.8603951048328197, "grad_norm": 1.3495675212645437, "learning_rate": 1.0047372598505956e-06, "loss": 0.4164, "step": 24150 }, { "epoch": 0.8604307319593139, "grad_norm": 2.085826219643515, "learning_rate": 1.0042332032658541e-06, "loss": 0.5101, "step": 24151 }, { "epoch": 0.8604663590858079, "grad_norm": 1.1966276797712558, "learning_rate": 1.003729266466038e-06, "loss": 0.3948, "step": 24152 }, { "epoch": 0.860501986212302, "grad_norm": 1.701621232894912, "learning_rate": 1.0032254494578519e-06, "loss": 0.5515, "step": 24153 }, { "epoch": 0.8605376133387962, "grad_norm": 1.269655077890373, "learning_rate": 1.002721752248007e-06, "loss": 0.3707, "step": 24154 }, { "epoch": 0.8605732404652903, "grad_norm": 1.1814241537016559, "learning_rate": 1.0022181748432113e-06, "loss": 0.4144, "step": 24155 }, { "epoch": 0.8606088675917843, "grad_norm": 1.1896800428053889, "learning_rate": 1.0017147172501695e-06, "loss": 0.4184, "step": 24156 }, { "epoch": 0.8606444947182785, "grad_norm": 1.4740505835912514, "learning_rate": 1.0012113794755873e-06, "loss": 0.5213, "step": 24157 }, { "epoch": 0.8606801218447726, "grad_norm": 1.6217194869494975, "learning_rate": 1.0007081615261638e-06, "loss": 0.4712, "step": 24158 }, { "epoch": 0.8607157489712667, "grad_norm": 1.4077656778536234, "learning_rate": 1.0002050634086036e-06, "loss": 0.4164, "step": 24159 }, { "epoch": 0.8607513760977609, "grad_norm": 1.1811841871028823, "learning_rate": 9.997020851296003e-07, "loss": 0.41, "step": 24160 }, { "epoch": 0.8607870032242549, "grad_norm": 1.2335412119583582, "learning_rate": 9.99199226695856e-07, "loss": 0.3965, "step": 24161 }, { "epoch": 0.860822630350749, "grad_norm": 1.7173517782097016, "learning_rate": 9.986964881140638e-07, "loss": 0.652, "step": 24162 }, { "epoch": 0.8608582574772432, "grad_norm": 1.1941301659674746, "learning_rate": 9.981938693909221e-07, "loss": 0.3515, "step": 24163 }, { "epoch": 0.8608938846037373, "grad_norm": 1.5929461636664821, "learning_rate": 9.976913705331192e-07, "loss": 0.509, "step": 24164 }, { "epoch": 0.8609295117302314, "grad_norm": 1.6478054138168654, "learning_rate": 9.971889915473499e-07, "loss": 0.4112, "step": 24165 }, { "epoch": 0.8609651388567255, "grad_norm": 1.218311086484585, "learning_rate": 9.966867324402997e-07, "loss": 0.4086, "step": 24166 }, { "epoch": 0.8610007659832196, "grad_norm": 1.4245139150134247, "learning_rate": 9.961845932186599e-07, "loss": 0.3974, "step": 24167 }, { "epoch": 0.8610363931097137, "grad_norm": 1.4847487687692171, "learning_rate": 9.95682573889114e-07, "loss": 0.5347, "step": 24168 }, { "epoch": 0.8610720202362079, "grad_norm": 1.5202254985579535, "learning_rate": 9.951806744583482e-07, "loss": 0.5098, "step": 24169 }, { "epoch": 0.861107647362702, "grad_norm": 1.5137890872063624, "learning_rate": 9.946788949330454e-07, "loss": 0.4089, "step": 24170 }, { "epoch": 0.861143274489196, "grad_norm": 1.476461470493257, "learning_rate": 9.941772353198886e-07, "loss": 0.5449, "step": 24171 }, { "epoch": 0.8611789016156902, "grad_norm": 1.366041677013989, "learning_rate": 9.936756956255577e-07, "loss": 0.4507, "step": 24172 }, { "epoch": 0.8612145287421843, "grad_norm": 1.3113760246122206, "learning_rate": 9.931742758567275e-07, "loss": 0.4308, "step": 24173 }, { "epoch": 0.8612501558686784, "grad_norm": 1.7444016151838082, "learning_rate": 9.926729760200805e-07, "loss": 0.4723, "step": 24174 }, { "epoch": 0.8612857829951726, "grad_norm": 1.7008601213199248, "learning_rate": 9.921717961222855e-07, "loss": 0.4945, "step": 24175 }, { "epoch": 0.8613214101216666, "grad_norm": 1.3139312837152493, "learning_rate": 9.916707361700195e-07, "loss": 0.4216, "step": 24176 }, { "epoch": 0.8613570372481607, "grad_norm": 1.6101369495435442, "learning_rate": 9.91169796169954e-07, "loss": 0.514, "step": 24177 }, { "epoch": 0.8613926643746549, "grad_norm": 1.0893997708082759, "learning_rate": 9.906689761287613e-07, "loss": 0.3467, "step": 24178 }, { "epoch": 0.861428291501149, "grad_norm": 1.3295096135331617, "learning_rate": 9.901682760531072e-07, "loss": 0.4253, "step": 24179 }, { "epoch": 0.861463918627643, "grad_norm": 1.2187863860526562, "learning_rate": 9.896676959496587e-07, "loss": 0.3922, "step": 24180 }, { "epoch": 0.8614995457541372, "grad_norm": 1.1971841245214088, "learning_rate": 9.89167235825086e-07, "loss": 0.4211, "step": 24181 }, { "epoch": 0.8615351728806313, "grad_norm": 1.7855823534564237, "learning_rate": 9.886668956860467e-07, "loss": 0.5104, "step": 24182 }, { "epoch": 0.8615708000071254, "grad_norm": 1.6229778672098971, "learning_rate": 9.881666755392083e-07, "loss": 0.4521, "step": 24183 }, { "epoch": 0.8616064271336196, "grad_norm": 2.0182537313664874, "learning_rate": 9.876665753912262e-07, "loss": 0.5782, "step": 24184 }, { "epoch": 0.8616420542601136, "grad_norm": 1.797077494178957, "learning_rate": 9.871665952487663e-07, "loss": 0.4071, "step": 24185 }, { "epoch": 0.8616776813866077, "grad_norm": 1.338212013969534, "learning_rate": 9.86666735118481e-07, "loss": 0.4344, "step": 24186 }, { "epoch": 0.8617133085131019, "grad_norm": 1.2966471908989572, "learning_rate": 9.861669950070308e-07, "loss": 0.4057, "step": 24187 }, { "epoch": 0.861748935639596, "grad_norm": 1.2771390014335624, "learning_rate": 9.856673749210644e-07, "loss": 0.4032, "step": 24188 }, { "epoch": 0.8617845627660901, "grad_norm": 1.3582919689813604, "learning_rate": 9.8516787486724e-07, "loss": 0.4494, "step": 24189 }, { "epoch": 0.8618201898925842, "grad_norm": 1.5401525809428585, "learning_rate": 9.846684948522034e-07, "loss": 0.5839, "step": 24190 }, { "epoch": 0.8618558170190783, "grad_norm": 1.8855894987673234, "learning_rate": 9.841692348826082e-07, "loss": 0.54, "step": 24191 }, { "epoch": 0.8618914441455724, "grad_norm": 1.4286046829014347, "learning_rate": 9.836700949651012e-07, "loss": 0.4455, "step": 24192 }, { "epoch": 0.8619270712720666, "grad_norm": 1.334467896842323, "learning_rate": 9.831710751063284e-07, "loss": 0.3839, "step": 24193 }, { "epoch": 0.8619626983985607, "grad_norm": 1.5418900432645757, "learning_rate": 9.826721753129375e-07, "loss": 0.4356, "step": 24194 }, { "epoch": 0.8619983255250547, "grad_norm": 1.1575520962952786, "learning_rate": 9.82173395591568e-07, "loss": 0.3797, "step": 24195 }, { "epoch": 0.8620339526515489, "grad_norm": 1.164483178488704, "learning_rate": 9.816747359488632e-07, "loss": 0.3868, "step": 24196 }, { "epoch": 0.862069579778043, "grad_norm": 1.1505394231991803, "learning_rate": 9.811761963914623e-07, "loss": 0.4447, "step": 24197 }, { "epoch": 0.8621052069045371, "grad_norm": 1.4013757325305811, "learning_rate": 9.806777769260034e-07, "loss": 0.4258, "step": 24198 }, { "epoch": 0.8621408340310313, "grad_norm": 1.2166488455746043, "learning_rate": 9.801794775591234e-07, "loss": 0.397, "step": 24199 }, { "epoch": 0.8621764611575253, "grad_norm": 1.5256423459418833, "learning_rate": 9.796812982974613e-07, "loss": 0.4369, "step": 24200 }, { "epoch": 0.8622120882840194, "grad_norm": 1.2606979088291923, "learning_rate": 9.791832391476452e-07, "loss": 0.4334, "step": 24201 }, { "epoch": 0.8622477154105136, "grad_norm": 1.1635702125591916, "learning_rate": 9.786853001163087e-07, "loss": 0.4436, "step": 24202 }, { "epoch": 0.8622833425370077, "grad_norm": 2.4045089908748767, "learning_rate": 9.781874812100856e-07, "loss": 0.5374, "step": 24203 }, { "epoch": 0.8623189696635017, "grad_norm": 1.6635691255313616, "learning_rate": 9.776897824356002e-07, "loss": 0.491, "step": 24204 }, { "epoch": 0.8623545967899959, "grad_norm": 1.1770007375540137, "learning_rate": 9.77192203799483e-07, "loss": 0.3722, "step": 24205 }, { "epoch": 0.86239022391649, "grad_norm": 1.2922363650678075, "learning_rate": 9.766947453083564e-07, "loss": 0.3571, "step": 24206 }, { "epoch": 0.8624258510429841, "grad_norm": 1.923404165387163, "learning_rate": 9.761974069688462e-07, "loss": 0.5332, "step": 24207 }, { "epoch": 0.8624614781694783, "grad_norm": 1.1710355027615384, "learning_rate": 9.75700188787575e-07, "loss": 0.4174, "step": 24208 }, { "epoch": 0.8624971052959723, "grad_norm": 1.3755765873485417, "learning_rate": 9.752030907711652e-07, "loss": 0.4286, "step": 24209 }, { "epoch": 0.8625327324224664, "grad_norm": 1.7433110111721486, "learning_rate": 9.747061129262325e-07, "loss": 0.4481, "step": 24210 }, { "epoch": 0.8625683595489606, "grad_norm": 1.2184979374722136, "learning_rate": 9.742092552593985e-07, "loss": 0.4482, "step": 24211 }, { "epoch": 0.8626039866754547, "grad_norm": 1.2382249972492139, "learning_rate": 9.737125177772744e-07, "loss": 0.3549, "step": 24212 }, { "epoch": 0.8626396138019488, "grad_norm": 1.4369528027780083, "learning_rate": 9.732159004864772e-07, "loss": 0.5137, "step": 24213 }, { "epoch": 0.8626752409284429, "grad_norm": 1.1975071213927821, "learning_rate": 9.727194033936194e-07, "loss": 0.394, "step": 24214 }, { "epoch": 0.862710868054937, "grad_norm": 1.2949537701127554, "learning_rate": 9.722230265053124e-07, "loss": 0.3976, "step": 24215 }, { "epoch": 0.8627464951814311, "grad_norm": 1.1116437797849954, "learning_rate": 9.717267698281684e-07, "loss": 0.3843, "step": 24216 }, { "epoch": 0.8627821223079253, "grad_norm": 1.2693798549986506, "learning_rate": 9.712306333687905e-07, "loss": 0.4117, "step": 24217 }, { "epoch": 0.8628177494344194, "grad_norm": 1.2006592583869544, "learning_rate": 9.707346171337895e-07, "loss": 0.3865, "step": 24218 }, { "epoch": 0.8628533765609134, "grad_norm": 1.2129649823904245, "learning_rate": 9.702387211297659e-07, "loss": 0.3807, "step": 24219 }, { "epoch": 0.8628890036874076, "grad_norm": 1.453651485698196, "learning_rate": 9.697429453633268e-07, "loss": 0.3966, "step": 24220 }, { "epoch": 0.8629246308139017, "grad_norm": 1.4707563149278948, "learning_rate": 9.69247289841071e-07, "loss": 0.5489, "step": 24221 }, { "epoch": 0.8629602579403958, "grad_norm": 1.2505345611869072, "learning_rate": 9.68751754569599e-07, "loss": 0.4277, "step": 24222 }, { "epoch": 0.86299588506689, "grad_norm": 1.3918131414191723, "learning_rate": 9.6825633955551e-07, "loss": 0.5633, "step": 24223 }, { "epoch": 0.863031512193384, "grad_norm": 1.26550618632959, "learning_rate": 9.677610448054032e-07, "loss": 0.4247, "step": 24224 }, { "epoch": 0.8630671393198781, "grad_norm": 1.3547370228432383, "learning_rate": 9.672658703258685e-07, "loss": 0.5022, "step": 24225 }, { "epoch": 0.8631027664463723, "grad_norm": 1.445537088015265, "learning_rate": 9.667708161235033e-07, "loss": 0.4531, "step": 24226 }, { "epoch": 0.8631383935728664, "grad_norm": 1.2247180746788704, "learning_rate": 9.662758822049002e-07, "loss": 0.385, "step": 24227 }, { "epoch": 0.8631740206993604, "grad_norm": 1.6543072738747886, "learning_rate": 9.657810685766455e-07, "loss": 0.4754, "step": 24228 }, { "epoch": 0.8632096478258546, "grad_norm": 1.5045511581730884, "learning_rate": 9.6528637524533e-07, "loss": 0.5596, "step": 24229 }, { "epoch": 0.8632452749523487, "grad_norm": 1.6711381473751334, "learning_rate": 9.647918022175417e-07, "loss": 0.5815, "step": 24230 }, { "epoch": 0.8632809020788428, "grad_norm": 1.5136621217441317, "learning_rate": 9.642973494998687e-07, "loss": 0.4846, "step": 24231 }, { "epoch": 0.863316529205337, "grad_norm": 1.287818776737971, "learning_rate": 9.638030170988888e-07, "loss": 0.3673, "step": 24232 }, { "epoch": 0.863352156331831, "grad_norm": 1.1651192989145656, "learning_rate": 9.633088050211902e-07, "loss": 0.407, "step": 24233 }, { "epoch": 0.8633877834583251, "grad_norm": 2.171189539364837, "learning_rate": 9.62814713273349e-07, "loss": 0.5061, "step": 24234 }, { "epoch": 0.8634234105848193, "grad_norm": 1.4402031761245224, "learning_rate": 9.623207418619484e-07, "loss": 0.4882, "step": 24235 }, { "epoch": 0.8634590377113134, "grad_norm": 2.1641490972139916, "learning_rate": 9.618268907935602e-07, "loss": 0.4692, "step": 24236 }, { "epoch": 0.8634946648378075, "grad_norm": 1.2476559947189774, "learning_rate": 9.613331600747666e-07, "loss": 0.4231, "step": 24237 }, { "epoch": 0.8635302919643016, "grad_norm": 1.2226215755576555, "learning_rate": 9.608395497121414e-07, "loss": 0.3904, "step": 24238 }, { "epoch": 0.8635659190907957, "grad_norm": 1.4913979257778698, "learning_rate": 9.603460597122527e-07, "loss": 0.5006, "step": 24239 }, { "epoch": 0.8636015462172898, "grad_norm": 1.1386198059320793, "learning_rate": 9.598526900816773e-07, "loss": 0.4069, "step": 24240 }, { "epoch": 0.863637173343784, "grad_norm": 1.5531749544286864, "learning_rate": 9.593594408269802e-07, "loss": 0.4352, "step": 24241 }, { "epoch": 0.8636728004702781, "grad_norm": 1.404914269870055, "learning_rate": 9.588663119547337e-07, "loss": 0.5162, "step": 24242 }, { "epoch": 0.8637084275967721, "grad_norm": 1.179657890480839, "learning_rate": 9.583733034714982e-07, "loss": 0.3941, "step": 24243 }, { "epoch": 0.8637440547232663, "grad_norm": 1.343693973765736, "learning_rate": 9.57880415383843e-07, "loss": 0.4232, "step": 24244 }, { "epoch": 0.8637796818497604, "grad_norm": 1.4877511603922473, "learning_rate": 9.5738764769833e-07, "loss": 0.5053, "step": 24245 }, { "epoch": 0.8638153089762545, "grad_norm": 1.7020575826902813, "learning_rate": 9.568950004215227e-07, "loss": 0.5584, "step": 24246 }, { "epoch": 0.8638509361027487, "grad_norm": 1.4721605058715292, "learning_rate": 9.564024735599775e-07, "loss": 0.5076, "step": 24247 }, { "epoch": 0.8638865632292427, "grad_norm": 1.1982986375907765, "learning_rate": 9.559100671202537e-07, "loss": 0.4038, "step": 24248 }, { "epoch": 0.8639221903557368, "grad_norm": 1.5624732802093837, "learning_rate": 9.554177811089115e-07, "loss": 0.4562, "step": 24249 }, { "epoch": 0.863957817482231, "grad_norm": 1.550967435647717, "learning_rate": 9.549256155325015e-07, "loss": 0.5289, "step": 24250 }, { "epoch": 0.8639934446087251, "grad_norm": 1.227938233316196, "learning_rate": 9.544335703975781e-07, "loss": 0.3833, "step": 24251 }, { "epoch": 0.8640290717352191, "grad_norm": 1.234288911190575, "learning_rate": 9.539416457106954e-07, "loss": 0.3842, "step": 24252 }, { "epoch": 0.8640646988617133, "grad_norm": 1.161642142293042, "learning_rate": 9.534498414784033e-07, "loss": 0.4288, "step": 24253 }, { "epoch": 0.8641003259882074, "grad_norm": 2.6982118706219107, "learning_rate": 9.529581577072478e-07, "loss": 0.5095, "step": 24254 }, { "epoch": 0.8641359531147015, "grad_norm": 1.3730879470643294, "learning_rate": 9.524665944037803e-07, "loss": 0.4645, "step": 24255 }, { "epoch": 0.8641715802411957, "grad_norm": 1.522560140688763, "learning_rate": 9.519751515745424e-07, "loss": 0.4999, "step": 24256 }, { "epoch": 0.8642072073676897, "grad_norm": 1.4153903503671648, "learning_rate": 9.514838292260808e-07, "loss": 0.4763, "step": 24257 }, { "epoch": 0.8642428344941838, "grad_norm": 1.5838034001983212, "learning_rate": 9.509926273649351e-07, "loss": 0.5267, "step": 24258 }, { "epoch": 0.864278461620678, "grad_norm": 2.1608098578086237, "learning_rate": 9.505015459976474e-07, "loss": 0.5183, "step": 24259 }, { "epoch": 0.8643140887471721, "grad_norm": 1.378759886648714, "learning_rate": 9.500105851307573e-07, "loss": 0.4782, "step": 24260 }, { "epoch": 0.8643497158736662, "grad_norm": 1.2447362740889056, "learning_rate": 9.495197447708004e-07, "loss": 0.4461, "step": 24261 }, { "epoch": 0.8643853430001603, "grad_norm": 1.7419052767009857, "learning_rate": 9.490290249243173e-07, "loss": 0.5486, "step": 24262 }, { "epoch": 0.8644209701266544, "grad_norm": 1.4471695816004309, "learning_rate": 9.48538425597837e-07, "loss": 0.4793, "step": 24263 }, { "epoch": 0.8644565972531485, "grad_norm": 1.58098982075625, "learning_rate": 9.480479467978953e-07, "loss": 0.4795, "step": 24264 }, { "epoch": 0.8644922243796427, "grad_norm": 1.2863741292458033, "learning_rate": 9.475575885310207e-07, "loss": 0.4234, "step": 24265 }, { "epoch": 0.8645278515061368, "grad_norm": 1.5379387547443037, "learning_rate": 9.470673508037431e-07, "loss": 0.631, "step": 24266 }, { "epoch": 0.8645634786326308, "grad_norm": 1.2811608503856078, "learning_rate": 9.465772336225931e-07, "loss": 0.4147, "step": 24267 }, { "epoch": 0.864599105759125, "grad_norm": 1.1612901029069609, "learning_rate": 9.460872369940954e-07, "loss": 0.4045, "step": 24268 }, { "epoch": 0.8646347328856191, "grad_norm": 1.3372283857469727, "learning_rate": 9.455973609247737e-07, "loss": 0.416, "step": 24269 }, { "epoch": 0.8646703600121132, "grad_norm": 1.1554806163903972, "learning_rate": 9.451076054211527e-07, "loss": 0.4124, "step": 24270 }, { "epoch": 0.8647059871386074, "grad_norm": 1.6331569404520074, "learning_rate": 9.446179704897518e-07, "loss": 0.4757, "step": 24271 }, { "epoch": 0.8647416142651014, "grad_norm": 1.451563246324616, "learning_rate": 9.441284561370923e-07, "loss": 0.4495, "step": 24272 }, { "epoch": 0.8647772413915955, "grad_norm": 1.4642660234480156, "learning_rate": 9.436390623696911e-07, "loss": 0.4513, "step": 24273 }, { "epoch": 0.8648128685180897, "grad_norm": 1.2034833627314978, "learning_rate": 9.431497891940667e-07, "loss": 0.4001, "step": 24274 }, { "epoch": 0.8648484956445838, "grad_norm": 2.5987490834314144, "learning_rate": 9.426606366167346e-07, "loss": 0.5199, "step": 24275 }, { "epoch": 0.8648841227710778, "grad_norm": 1.45505816967152, "learning_rate": 9.421716046442054e-07, "loss": 0.4898, "step": 24276 }, { "epoch": 0.864919749897572, "grad_norm": 1.196595295155138, "learning_rate": 9.416826932829936e-07, "loss": 0.3964, "step": 24277 }, { "epoch": 0.8649553770240661, "grad_norm": 1.2460973644492457, "learning_rate": 9.411939025396067e-07, "loss": 0.389, "step": 24278 }, { "epoch": 0.8649910041505602, "grad_norm": 1.1769433370211824, "learning_rate": 9.407052324205556e-07, "loss": 0.3696, "step": 24279 }, { "epoch": 0.8650266312770544, "grad_norm": 1.496912812652255, "learning_rate": 9.402166829323455e-07, "loss": 0.4646, "step": 24280 }, { "epoch": 0.8650622584035484, "grad_norm": 1.3235689271974262, "learning_rate": 9.397282540814823e-07, "loss": 0.3914, "step": 24281 }, { "epoch": 0.8650978855300425, "grad_norm": 1.7364245047256905, "learning_rate": 9.392399458744706e-07, "loss": 0.5039, "step": 24282 }, { "epoch": 0.8651335126565367, "grad_norm": 1.3918302007557914, "learning_rate": 9.387517583178119e-07, "loss": 0.4555, "step": 24283 }, { "epoch": 0.8651691397830308, "grad_norm": 1.561658700528088, "learning_rate": 9.382636914180087e-07, "loss": 0.498, "step": 24284 }, { "epoch": 0.8652047669095249, "grad_norm": 1.7854623784883672, "learning_rate": 9.377757451815562e-07, "loss": 0.4659, "step": 24285 }, { "epoch": 0.865240394036019, "grad_norm": 1.5025719632602716, "learning_rate": 9.372879196149564e-07, "loss": 0.4574, "step": 24286 }, { "epoch": 0.8652760211625131, "grad_norm": 1.1963034976275548, "learning_rate": 9.368002147247002e-07, "loss": 0.3828, "step": 24287 }, { "epoch": 0.8653116482890072, "grad_norm": 1.124657499620606, "learning_rate": 9.363126305172831e-07, "loss": 0.4408, "step": 24288 }, { "epoch": 0.8653472754155014, "grad_norm": 1.3009009324837013, "learning_rate": 9.358251669991992e-07, "loss": 0.4305, "step": 24289 }, { "epoch": 0.8653829025419955, "grad_norm": 1.1999636991516387, "learning_rate": 9.353378241769407e-07, "loss": 0.4049, "step": 24290 }, { "epoch": 0.8654185296684895, "grad_norm": 1.266357909443185, "learning_rate": 9.348506020569925e-07, "loss": 0.4125, "step": 24291 }, { "epoch": 0.8654541567949837, "grad_norm": 1.8526909829894729, "learning_rate": 9.343635006458473e-07, "loss": 0.4819, "step": 24292 }, { "epoch": 0.8654897839214778, "grad_norm": 1.2048898306375873, "learning_rate": 9.338765199499855e-07, "loss": 0.4567, "step": 24293 }, { "epoch": 0.8655254110479719, "grad_norm": 2.806015089910347, "learning_rate": 9.333896599758974e-07, "loss": 0.4193, "step": 24294 }, { "epoch": 0.8655610381744661, "grad_norm": 1.5028710746927743, "learning_rate": 9.329029207300611e-07, "loss": 0.5244, "step": 24295 }, { "epoch": 0.8655966653009601, "grad_norm": 1.4013297256872754, "learning_rate": 9.324163022189603e-07, "loss": 0.5021, "step": 24296 }, { "epoch": 0.8656322924274542, "grad_norm": 1.4632335335971838, "learning_rate": 9.319298044490744e-07, "loss": 0.5312, "step": 24297 }, { "epoch": 0.8656679195539484, "grad_norm": 1.2567022970049682, "learning_rate": 9.314434274268802e-07, "loss": 0.3763, "step": 24298 }, { "epoch": 0.8657035466804425, "grad_norm": 2.260235587061821, "learning_rate": 9.309571711588583e-07, "loss": 0.5579, "step": 24299 }, { "epoch": 0.8657391738069365, "grad_norm": 1.9059229342772839, "learning_rate": 9.304710356514779e-07, "loss": 0.4645, "step": 24300 }, { "epoch": 0.8657748009334307, "grad_norm": 1.2288197048389133, "learning_rate": 9.299850209112182e-07, "loss": 0.4053, "step": 24301 }, { "epoch": 0.8658104280599248, "grad_norm": 1.5554076945363688, "learning_rate": 9.294991269445442e-07, "loss": 0.5467, "step": 24302 }, { "epoch": 0.865846055186419, "grad_norm": 1.4390170016719162, "learning_rate": 9.290133537579294e-07, "loss": 0.4615, "step": 24303 }, { "epoch": 0.8658816823129131, "grad_norm": 1.2011505985396724, "learning_rate": 9.28527701357842e-07, "loss": 0.3829, "step": 24304 }, { "epoch": 0.8659173094394071, "grad_norm": 1.1659654446069692, "learning_rate": 9.280421697507514e-07, "loss": 0.3809, "step": 24305 }, { "epoch": 0.8659529365659013, "grad_norm": 1.3338732494817582, "learning_rate": 9.275567589431178e-07, "loss": 0.427, "step": 24306 }, { "epoch": 0.8659885636923954, "grad_norm": 1.5518281961995841, "learning_rate": 9.270714689414073e-07, "loss": 0.4933, "step": 24307 }, { "epoch": 0.8660241908188895, "grad_norm": 1.3021479724398946, "learning_rate": 9.265862997520836e-07, "loss": 0.3581, "step": 24308 }, { "epoch": 0.8660598179453837, "grad_norm": 1.2105722978118987, "learning_rate": 9.261012513816037e-07, "loss": 0.3943, "step": 24309 }, { "epoch": 0.8660954450718777, "grad_norm": 1.5928388168126555, "learning_rate": 9.256163238364302e-07, "loss": 0.4363, "step": 24310 }, { "epoch": 0.8661310721983718, "grad_norm": 1.6383385247739812, "learning_rate": 9.251315171230136e-07, "loss": 0.5129, "step": 24311 }, { "epoch": 0.866166699324866, "grad_norm": 1.4846564045983508, "learning_rate": 9.246468312478174e-07, "loss": 0.521, "step": 24312 }, { "epoch": 0.8662023264513601, "grad_norm": 2.9132521908767783, "learning_rate": 9.2416226621729e-07, "loss": 0.542, "step": 24313 }, { "epoch": 0.8662379535778542, "grad_norm": 1.4385933101823363, "learning_rate": 9.236778220378872e-07, "loss": 0.5374, "step": 24314 }, { "epoch": 0.8662735807043483, "grad_norm": 1.3906712553304224, "learning_rate": 9.231934987160573e-07, "loss": 0.3832, "step": 24315 }, { "epoch": 0.8663092078308424, "grad_norm": 1.1303524440005908, "learning_rate": 9.227092962582518e-07, "loss": 0.3737, "step": 24316 }, { "epoch": 0.8663448349573365, "grad_norm": 1.300658724576635, "learning_rate": 9.222252146709143e-07, "loss": 0.4201, "step": 24317 }, { "epoch": 0.8663804620838307, "grad_norm": 1.4759586810334022, "learning_rate": 9.217412539604942e-07, "loss": 0.3883, "step": 24318 }, { "epoch": 0.8664160892103248, "grad_norm": 1.6691685469684603, "learning_rate": 9.21257414133433e-07, "loss": 0.5188, "step": 24319 }, { "epoch": 0.8664517163368188, "grad_norm": 1.1980657633062273, "learning_rate": 9.207736951961766e-07, "loss": 0.3741, "step": 24320 }, { "epoch": 0.866487343463313, "grad_norm": 1.7338607051846222, "learning_rate": 9.202900971551654e-07, "loss": 0.5448, "step": 24321 }, { "epoch": 0.8665229705898071, "grad_norm": 1.5105863679897058, "learning_rate": 9.198066200168365e-07, "loss": 0.4568, "step": 24322 }, { "epoch": 0.8665585977163012, "grad_norm": 1.1340723029780613, "learning_rate": 9.193232637876303e-07, "loss": 0.3728, "step": 24323 }, { "epoch": 0.8665942248427954, "grad_norm": 1.484622525856464, "learning_rate": 9.188400284739807e-07, "loss": 0.527, "step": 24324 }, { "epoch": 0.8666298519692894, "grad_norm": 1.1774800202474929, "learning_rate": 9.183569140823223e-07, "loss": 0.4071, "step": 24325 }, { "epoch": 0.8666654790957835, "grad_norm": 1.087242905907099, "learning_rate": 9.178739206190912e-07, "loss": 0.3743, "step": 24326 }, { "epoch": 0.8667011062222777, "grad_norm": 1.2153008623038137, "learning_rate": 9.173910480907166e-07, "loss": 0.4064, "step": 24327 }, { "epoch": 0.8667367333487718, "grad_norm": 1.1559987412711303, "learning_rate": 9.16908296503628e-07, "loss": 0.3697, "step": 24328 }, { "epoch": 0.8667723604752658, "grad_norm": 1.446264267320692, "learning_rate": 9.164256658642546e-07, "loss": 0.6052, "step": 24329 }, { "epoch": 0.86680798760176, "grad_norm": 1.331120694842177, "learning_rate": 9.159431561790233e-07, "loss": 0.4343, "step": 24330 }, { "epoch": 0.8668436147282541, "grad_norm": 1.2471455383939831, "learning_rate": 9.15460767454357e-07, "loss": 0.3762, "step": 24331 }, { "epoch": 0.8668792418547482, "grad_norm": 1.5090571753520288, "learning_rate": 9.149784996966815e-07, "loss": 0.4874, "step": 24332 }, { "epoch": 0.8669148689812424, "grad_norm": 1.3525033173895755, "learning_rate": 9.144963529124163e-07, "loss": 0.4296, "step": 24333 }, { "epoch": 0.8669504961077364, "grad_norm": 1.459468458430209, "learning_rate": 9.140143271079815e-07, "loss": 0.5147, "step": 24334 }, { "epoch": 0.8669861232342305, "grad_norm": 1.365042244747129, "learning_rate": 9.135324222897979e-07, "loss": 0.4712, "step": 24335 }, { "epoch": 0.8670217503607247, "grad_norm": 1.2769653978051736, "learning_rate": 9.130506384642823e-07, "loss": 0.4233, "step": 24336 }, { "epoch": 0.8670573774872188, "grad_norm": 1.4384670372722703, "learning_rate": 9.125689756378464e-07, "loss": 0.5735, "step": 24337 }, { "epoch": 0.8670930046137129, "grad_norm": 1.463623707048267, "learning_rate": 9.120874338169095e-07, "loss": 0.4897, "step": 24338 }, { "epoch": 0.867128631740207, "grad_norm": 1.3136765226165752, "learning_rate": 9.116060130078774e-07, "loss": 0.4044, "step": 24339 }, { "epoch": 0.8671642588667011, "grad_norm": 1.302857905403541, "learning_rate": 9.111247132171652e-07, "loss": 0.5003, "step": 24340 }, { "epoch": 0.8671998859931952, "grad_norm": 1.6789928387860127, "learning_rate": 9.106435344511788e-07, "loss": 0.4546, "step": 24341 }, { "epoch": 0.8672355131196894, "grad_norm": 1.2402160070430501, "learning_rate": 9.101624767163275e-07, "loss": 0.4195, "step": 24342 }, { "epoch": 0.8672711402461835, "grad_norm": 1.2142947588204989, "learning_rate": 9.096815400190173e-07, "loss": 0.4411, "step": 24343 }, { "epoch": 0.8673067673726775, "grad_norm": 1.22513530669005, "learning_rate": 9.092007243656497e-07, "loss": 0.4279, "step": 24344 }, { "epoch": 0.8673423944991717, "grad_norm": 1.2881329073200511, "learning_rate": 9.087200297626309e-07, "loss": 0.4262, "step": 24345 }, { "epoch": 0.8673780216256658, "grad_norm": 1.569354155671362, "learning_rate": 9.082394562163576e-07, "loss": 0.671, "step": 24346 }, { "epoch": 0.8674136487521599, "grad_norm": 2.3003360552635406, "learning_rate": 9.077590037332318e-07, "loss": 0.5309, "step": 24347 }, { "epoch": 0.8674492758786541, "grad_norm": 1.3953652697693764, "learning_rate": 9.07278672319647e-07, "loss": 0.4144, "step": 24348 }, { "epoch": 0.8674849030051481, "grad_norm": 1.2575000979014186, "learning_rate": 9.067984619820048e-07, "loss": 0.4047, "step": 24349 }, { "epoch": 0.8675205301316422, "grad_norm": 1.2714159146448205, "learning_rate": 9.063183727266956e-07, "loss": 0.4036, "step": 24350 }, { "epoch": 0.8675561572581364, "grad_norm": 1.442799002214776, "learning_rate": 9.058384045601154e-07, "loss": 0.4485, "step": 24351 }, { "epoch": 0.8675917843846305, "grad_norm": 1.2769885474839238, "learning_rate": 9.053585574886503e-07, "loss": 0.3994, "step": 24352 }, { "epoch": 0.8676274115111245, "grad_norm": 1.239699437673889, "learning_rate": 9.048788315186941e-07, "loss": 0.3696, "step": 24353 }, { "epoch": 0.8676630386376187, "grad_norm": 1.1748699591552492, "learning_rate": 9.043992266566348e-07, "loss": 0.4155, "step": 24354 }, { "epoch": 0.8676986657641128, "grad_norm": 1.1707777717040038, "learning_rate": 9.039197429088553e-07, "loss": 0.3836, "step": 24355 }, { "epoch": 0.8677342928906069, "grad_norm": 1.1652447181359795, "learning_rate": 9.034403802817426e-07, "loss": 0.4042, "step": 24356 }, { "epoch": 0.8677699200171011, "grad_norm": 1.064730704043733, "learning_rate": 9.029611387816794e-07, "loss": 0.3725, "step": 24357 }, { "epoch": 0.8678055471435951, "grad_norm": 1.2424637530968217, "learning_rate": 9.024820184150485e-07, "loss": 0.4007, "step": 24358 }, { "epoch": 0.8678411742700892, "grad_norm": 1.214963705005116, "learning_rate": 9.020030191882256e-07, "loss": 0.3808, "step": 24359 }, { "epoch": 0.8678768013965834, "grad_norm": 1.2411420503489616, "learning_rate": 9.015241411075948e-07, "loss": 0.3915, "step": 24360 }, { "epoch": 0.8679124285230775, "grad_norm": 1.1241575235027736, "learning_rate": 9.010453841795275e-07, "loss": 0.3991, "step": 24361 }, { "epoch": 0.8679480556495716, "grad_norm": 1.226532163743003, "learning_rate": 9.005667484104008e-07, "loss": 0.3862, "step": 24362 }, { "epoch": 0.8679836827760657, "grad_norm": 1.1564285435280475, "learning_rate": 9.000882338065875e-07, "loss": 0.4054, "step": 24363 }, { "epoch": 0.8680193099025598, "grad_norm": 1.2483839274023938, "learning_rate": 8.996098403744602e-07, "loss": 0.4246, "step": 24364 }, { "epoch": 0.8680549370290539, "grad_norm": 1.6028545189987153, "learning_rate": 8.991315681203894e-07, "loss": 0.5734, "step": 24365 }, { "epoch": 0.8680905641555481, "grad_norm": 1.660179618165375, "learning_rate": 8.986534170507421e-07, "loss": 0.5216, "step": 24366 }, { "epoch": 0.8681261912820422, "grad_norm": 1.3412635546177514, "learning_rate": 8.981753871718879e-07, "loss": 0.3916, "step": 24367 }, { "epoch": 0.8681618184085362, "grad_norm": 1.328628567812547, "learning_rate": 8.976974784901882e-07, "loss": 0.385, "step": 24368 }, { "epoch": 0.8681974455350304, "grad_norm": 1.1789518598292423, "learning_rate": 8.972196910120101e-07, "loss": 0.4102, "step": 24369 }, { "epoch": 0.8682330726615245, "grad_norm": 1.3150265704347093, "learning_rate": 8.96742024743713e-07, "loss": 0.3934, "step": 24370 }, { "epoch": 0.8682686997880186, "grad_norm": 1.2318916699529208, "learning_rate": 8.962644796916586e-07, "loss": 0.3748, "step": 24371 }, { "epoch": 0.8683043269145128, "grad_norm": 1.6365746119807065, "learning_rate": 8.957870558622062e-07, "loss": 0.5632, "step": 24372 }, { "epoch": 0.8683399540410068, "grad_norm": 2.2357368537345437, "learning_rate": 8.953097532617138e-07, "loss": 0.4728, "step": 24373 }, { "epoch": 0.8683755811675009, "grad_norm": 1.8271101808848822, "learning_rate": 8.948325718965334e-07, "loss": 0.4387, "step": 24374 }, { "epoch": 0.8684112082939951, "grad_norm": 1.4316366855913192, "learning_rate": 8.94355511773023e-07, "loss": 0.389, "step": 24375 }, { "epoch": 0.8684468354204892, "grad_norm": 1.2463269204334473, "learning_rate": 8.938785728975341e-07, "loss": 0.4106, "step": 24376 }, { "epoch": 0.8684824625469832, "grad_norm": 1.2333310813671308, "learning_rate": 8.934017552764151e-07, "loss": 0.4118, "step": 24377 }, { "epoch": 0.8685180896734774, "grad_norm": 3.0328021564771173, "learning_rate": 8.929250589160166e-07, "loss": 0.5292, "step": 24378 }, { "epoch": 0.8685537167999715, "grad_norm": 1.1924543759122566, "learning_rate": 8.924484838226877e-07, "loss": 0.3778, "step": 24379 }, { "epoch": 0.8685893439264656, "grad_norm": 1.3965196233940482, "learning_rate": 8.919720300027735e-07, "loss": 0.4391, "step": 24380 }, { "epoch": 0.8686249710529598, "grad_norm": 1.328046340894831, "learning_rate": 8.914956974626166e-07, "loss": 0.4749, "step": 24381 }, { "epoch": 0.8686605981794538, "grad_norm": 1.2714374689526446, "learning_rate": 8.91019486208563e-07, "loss": 0.4504, "step": 24382 }, { "epoch": 0.8686962253059479, "grad_norm": 1.6864801692884868, "learning_rate": 8.905433962469489e-07, "loss": 0.4632, "step": 24383 }, { "epoch": 0.8687318524324421, "grad_norm": 1.4208409281622487, "learning_rate": 8.900674275841193e-07, "loss": 0.5526, "step": 24384 }, { "epoch": 0.8687674795589362, "grad_norm": 2.133877952556197, "learning_rate": 8.895915802264076e-07, "loss": 0.5473, "step": 24385 }, { "epoch": 0.8688031066854303, "grad_norm": 2.457991710072098, "learning_rate": 8.891158541801514e-07, "loss": 0.5071, "step": 24386 }, { "epoch": 0.8688387338119244, "grad_norm": 1.4854327684433684, "learning_rate": 8.886402494516866e-07, "loss": 0.4459, "step": 24387 }, { "epoch": 0.8688743609384185, "grad_norm": 1.3733479351848459, "learning_rate": 8.881647660473447e-07, "loss": 0.4128, "step": 24388 }, { "epoch": 0.8689099880649126, "grad_norm": 1.3807423655860769, "learning_rate": 8.876894039734585e-07, "loss": 0.418, "step": 24389 }, { "epoch": 0.8689456151914068, "grad_norm": 1.5972913992329574, "learning_rate": 8.872141632363563e-07, "loss": 0.5161, "step": 24390 }, { "epoch": 0.8689812423179009, "grad_norm": 3.0783313717842304, "learning_rate": 8.867390438423695e-07, "loss": 0.5684, "step": 24391 }, { "epoch": 0.8690168694443949, "grad_norm": 1.5385198947134682, "learning_rate": 8.862640457978189e-07, "loss": 0.5119, "step": 24392 }, { "epoch": 0.8690524965708891, "grad_norm": 1.1255793063823376, "learning_rate": 8.857891691090336e-07, "loss": 0.3956, "step": 24393 }, { "epoch": 0.8690881236973832, "grad_norm": 1.8047827890681645, "learning_rate": 8.853144137823355e-07, "loss": 0.4811, "step": 24394 }, { "epoch": 0.8691237508238773, "grad_norm": 1.046921391127157, "learning_rate": 8.848397798240483e-07, "loss": 0.3716, "step": 24395 }, { "epoch": 0.8691593779503715, "grad_norm": 1.953909951879699, "learning_rate": 8.84365267240489e-07, "loss": 0.4786, "step": 24396 }, { "epoch": 0.8691950050768655, "grad_norm": 1.219096598823343, "learning_rate": 8.838908760379805e-07, "loss": 0.4085, "step": 24397 }, { "epoch": 0.8692306322033596, "grad_norm": 1.5803720640498609, "learning_rate": 8.834166062228333e-07, "loss": 0.4786, "step": 24398 }, { "epoch": 0.8692662593298538, "grad_norm": 1.8209690848663187, "learning_rate": 8.829424578013668e-07, "loss": 0.5858, "step": 24399 }, { "epoch": 0.8693018864563479, "grad_norm": 1.7140665309879075, "learning_rate": 8.824684307798937e-07, "loss": 0.4494, "step": 24400 }, { "epoch": 0.869337513582842, "grad_norm": 1.491901993318748, "learning_rate": 8.819945251647267e-07, "loss": 0.3992, "step": 24401 }, { "epoch": 0.8693731407093361, "grad_norm": 1.5297764894439625, "learning_rate": 8.815207409621773e-07, "loss": 0.5647, "step": 24402 }, { "epoch": 0.8694087678358302, "grad_norm": 1.563816415975026, "learning_rate": 8.810470781785507e-07, "loss": 0.5299, "step": 24403 }, { "epoch": 0.8694443949623243, "grad_norm": 1.2184691765938584, "learning_rate": 8.805735368201584e-07, "loss": 0.3684, "step": 24404 }, { "epoch": 0.8694800220888185, "grad_norm": 1.1972919564973157, "learning_rate": 8.80100116893301e-07, "loss": 0.4081, "step": 24405 }, { "epoch": 0.8695156492153125, "grad_norm": 1.2353446025690962, "learning_rate": 8.796268184042867e-07, "loss": 0.3749, "step": 24406 }, { "epoch": 0.8695512763418066, "grad_norm": 1.6440263138997795, "learning_rate": 8.791536413594149e-07, "loss": 0.5361, "step": 24407 }, { "epoch": 0.8695869034683008, "grad_norm": 1.4485291837213574, "learning_rate": 8.786805857649871e-07, "loss": 0.4202, "step": 24408 }, { "epoch": 0.8696225305947949, "grad_norm": 1.2728728214952925, "learning_rate": 8.78207651627303e-07, "loss": 0.449, "step": 24409 }, { "epoch": 0.869658157721289, "grad_norm": 1.3223976869629788, "learning_rate": 8.777348389526597e-07, "loss": 0.4059, "step": 24410 }, { "epoch": 0.8696937848477831, "grad_norm": 1.4184735198267788, "learning_rate": 8.772621477473553e-07, "loss": 0.5598, "step": 24411 }, { "epoch": 0.8697294119742772, "grad_norm": 1.15987004485052, "learning_rate": 8.767895780176794e-07, "loss": 0.3984, "step": 24412 }, { "epoch": 0.8697650391007713, "grad_norm": 1.3791202916012602, "learning_rate": 8.76317129769928e-07, "loss": 0.536, "step": 24413 }, { "epoch": 0.8698006662272655, "grad_norm": 1.6127222906043963, "learning_rate": 8.758448030103905e-07, "loss": 0.503, "step": 24414 }, { "epoch": 0.8698362933537596, "grad_norm": 1.241526662215398, "learning_rate": 8.753725977453564e-07, "loss": 0.3935, "step": 24415 }, { "epoch": 0.8698719204802536, "grad_norm": 1.3574915310440323, "learning_rate": 8.74900513981114e-07, "loss": 0.3978, "step": 24416 }, { "epoch": 0.8699075476067478, "grad_norm": 2.1575077760219936, "learning_rate": 8.744285517239515e-07, "loss": 0.517, "step": 24417 }, { "epoch": 0.8699431747332419, "grad_norm": 1.3611321602718969, "learning_rate": 8.739567109801495e-07, "loss": 0.5526, "step": 24418 }, { "epoch": 0.869978801859736, "grad_norm": 2.925413840212906, "learning_rate": 8.73484991755994e-07, "loss": 0.5033, "step": 24419 }, { "epoch": 0.8700144289862302, "grad_norm": 1.2334755458483218, "learning_rate": 8.730133940577623e-07, "loss": 0.3906, "step": 24420 }, { "epoch": 0.8700500561127242, "grad_norm": 1.4653748484912645, "learning_rate": 8.725419178917394e-07, "loss": 0.5462, "step": 24421 }, { "epoch": 0.8700856832392183, "grad_norm": 1.2651861115698013, "learning_rate": 8.72070563264199e-07, "loss": 0.4199, "step": 24422 }, { "epoch": 0.8701213103657125, "grad_norm": 1.3557571713983798, "learning_rate": 8.715993301814174e-07, "loss": 0.4534, "step": 24423 }, { "epoch": 0.8701569374922066, "grad_norm": 1.1809159768313342, "learning_rate": 8.711282186496739e-07, "loss": 0.3993, "step": 24424 }, { "epoch": 0.8701925646187006, "grad_norm": 1.349975269824301, "learning_rate": 8.706572286752368e-07, "loss": 0.5103, "step": 24425 }, { "epoch": 0.8702281917451948, "grad_norm": 1.4824136515706634, "learning_rate": 8.701863602643823e-07, "loss": 0.5628, "step": 24426 }, { "epoch": 0.8702638188716889, "grad_norm": 1.2172453974462416, "learning_rate": 8.697156134233764e-07, "loss": 0.4053, "step": 24427 }, { "epoch": 0.870299445998183, "grad_norm": 1.3338107951429468, "learning_rate": 8.692449881584897e-07, "loss": 0.4431, "step": 24428 }, { "epoch": 0.8703350731246772, "grad_norm": 1.296316281747065, "learning_rate": 8.68774484475986e-07, "loss": 0.4392, "step": 24429 }, { "epoch": 0.8703707002511712, "grad_norm": 1.670409789071195, "learning_rate": 8.683041023821326e-07, "loss": 0.5152, "step": 24430 }, { "epoch": 0.8704063273776653, "grad_norm": 1.5213122563941366, "learning_rate": 8.678338418831933e-07, "loss": 0.5143, "step": 24431 }, { "epoch": 0.8704419545041595, "grad_norm": 1.597828086022258, "learning_rate": 8.673637029854309e-07, "loss": 0.38, "step": 24432 }, { "epoch": 0.8704775816306536, "grad_norm": 1.314119946090441, "learning_rate": 8.668936856951026e-07, "loss": 0.4067, "step": 24433 }, { "epoch": 0.8705132087571477, "grad_norm": 1.2372439825064607, "learning_rate": 8.66423790018468e-07, "loss": 0.4352, "step": 24434 }, { "epoch": 0.8705488358836418, "grad_norm": 1.6107399426033095, "learning_rate": 8.659540159617863e-07, "loss": 0.5532, "step": 24435 }, { "epoch": 0.8705844630101359, "grad_norm": 1.4369693448318825, "learning_rate": 8.654843635313104e-07, "loss": 0.3954, "step": 24436 }, { "epoch": 0.87062009013663, "grad_norm": 1.2283841189611078, "learning_rate": 8.650148327332941e-07, "loss": 0.4027, "step": 24437 }, { "epoch": 0.8706557172631242, "grad_norm": 1.4959187679453916, "learning_rate": 8.645454235739903e-07, "loss": 0.4613, "step": 24438 }, { "epoch": 0.8706913443896183, "grad_norm": 1.2557810733022863, "learning_rate": 8.640761360596517e-07, "loss": 0.3895, "step": 24439 }, { "epoch": 0.8707269715161123, "grad_norm": 1.3849120998909643, "learning_rate": 8.636069701965222e-07, "loss": 0.3942, "step": 24440 }, { "epoch": 0.8707625986426065, "grad_norm": 1.2791561302421037, "learning_rate": 8.631379259908546e-07, "loss": 0.4448, "step": 24441 }, { "epoch": 0.8707982257691006, "grad_norm": 1.201039343899243, "learning_rate": 8.626690034488894e-07, "loss": 0.3887, "step": 24442 }, { "epoch": 0.8708338528955947, "grad_norm": 1.6667767454380265, "learning_rate": 8.622002025768749e-07, "loss": 0.4759, "step": 24443 }, { "epoch": 0.8708694800220889, "grad_norm": 2.4742566607787393, "learning_rate": 8.617315233810508e-07, "loss": 0.519, "step": 24444 }, { "epoch": 0.8709051071485829, "grad_norm": 1.7073524856201072, "learning_rate": 8.612629658676575e-07, "loss": 0.4192, "step": 24445 }, { "epoch": 0.870940734275077, "grad_norm": 1.3133778639867806, "learning_rate": 8.607945300429366e-07, "loss": 0.3869, "step": 24446 }, { "epoch": 0.8709763614015712, "grad_norm": 2.0403964747536265, "learning_rate": 8.603262159131232e-07, "loss": 0.5743, "step": 24447 }, { "epoch": 0.8710119885280653, "grad_norm": 1.2588104769627817, "learning_rate": 8.598580234844567e-07, "loss": 0.3692, "step": 24448 }, { "epoch": 0.8710476156545593, "grad_norm": 1.2895891417310048, "learning_rate": 8.593899527631677e-07, "loss": 0.3909, "step": 24449 }, { "epoch": 0.8710832427810535, "grad_norm": 1.4076997173398127, "learning_rate": 8.589220037554924e-07, "loss": 0.4204, "step": 24450 }, { "epoch": 0.8711188699075476, "grad_norm": 1.2472365772565117, "learning_rate": 8.584541764676579e-07, "loss": 0.4079, "step": 24451 }, { "epoch": 0.8711544970340417, "grad_norm": 1.8793235937731836, "learning_rate": 8.579864709058949e-07, "loss": 0.5007, "step": 24452 }, { "epoch": 0.8711901241605359, "grad_norm": 1.7148849678102354, "learning_rate": 8.575188870764328e-07, "loss": 0.4308, "step": 24453 }, { "epoch": 0.87122575128703, "grad_norm": 1.3395268199315802, "learning_rate": 8.570514249854977e-07, "loss": 0.3879, "step": 24454 }, { "epoch": 0.871261378413524, "grad_norm": 2.0629414284973095, "learning_rate": 8.565840846393126e-07, "loss": 0.56, "step": 24455 }, { "epoch": 0.8712970055400182, "grad_norm": 1.4069675817195273, "learning_rate": 8.56116866044101e-07, "loss": 0.4884, "step": 24456 }, { "epoch": 0.8713326326665123, "grad_norm": 1.4906383344596772, "learning_rate": 8.556497692060861e-07, "loss": 0.5088, "step": 24457 }, { "epoch": 0.8713682597930064, "grad_norm": 1.4664541096118604, "learning_rate": 8.55182794131485e-07, "loss": 0.4074, "step": 24458 }, { "epoch": 0.8714038869195005, "grad_norm": 1.2992837180264316, "learning_rate": 8.547159408265171e-07, "loss": 0.3844, "step": 24459 }, { "epoch": 0.8714395140459946, "grad_norm": 1.5434248412044258, "learning_rate": 8.542492092973987e-07, "loss": 0.5561, "step": 24460 }, { "epoch": 0.8714751411724887, "grad_norm": 1.288702734015972, "learning_rate": 8.537825995503435e-07, "loss": 0.4094, "step": 24461 }, { "epoch": 0.8715107682989829, "grad_norm": 1.7520820841412101, "learning_rate": 8.533161115915655e-07, "loss": 0.4869, "step": 24462 }, { "epoch": 0.871546395425477, "grad_norm": 1.0759852138530497, "learning_rate": 8.528497454272788e-07, "loss": 0.348, "step": 24463 }, { "epoch": 0.871582022551971, "grad_norm": 2.042065234025859, "learning_rate": 8.523835010636894e-07, "loss": 0.4745, "step": 24464 }, { "epoch": 0.8716176496784652, "grad_norm": 1.0949151614655224, "learning_rate": 8.519173785070101e-07, "loss": 0.3754, "step": 24465 }, { "epoch": 0.8716532768049593, "grad_norm": 1.401457418881561, "learning_rate": 8.514513777634426e-07, "loss": 0.5353, "step": 24466 }, { "epoch": 0.8716889039314534, "grad_norm": 1.1464162359695342, "learning_rate": 8.509854988391941e-07, "loss": 0.3769, "step": 24467 }, { "epoch": 0.8717245310579476, "grad_norm": 1.558396530138657, "learning_rate": 8.505197417404687e-07, "loss": 0.5498, "step": 24468 }, { "epoch": 0.8717601581844416, "grad_norm": 1.501261087790866, "learning_rate": 8.50054106473468e-07, "loss": 0.5121, "step": 24469 }, { "epoch": 0.8717957853109357, "grad_norm": 1.3872121258403383, "learning_rate": 8.495885930443936e-07, "loss": 0.4847, "step": 24470 }, { "epoch": 0.8718314124374299, "grad_norm": 1.2299436419890717, "learning_rate": 8.491232014594419e-07, "loss": 0.3774, "step": 24471 }, { "epoch": 0.871867039563924, "grad_norm": 1.5622367199753324, "learning_rate": 8.486579317248122e-07, "loss": 0.5786, "step": 24472 }, { "epoch": 0.871902666690418, "grad_norm": 1.6090851354090292, "learning_rate": 8.481927838466975e-07, "loss": 0.5581, "step": 24473 }, { "epoch": 0.8719382938169122, "grad_norm": 1.1249762301671822, "learning_rate": 8.477277578312936e-07, "loss": 0.3536, "step": 24474 }, { "epoch": 0.8719739209434063, "grad_norm": 1.3829172057623689, "learning_rate": 8.472628536847882e-07, "loss": 0.4614, "step": 24475 }, { "epoch": 0.8720095480699004, "grad_norm": 1.6058996606271814, "learning_rate": 8.467980714133795e-07, "loss": 0.531, "step": 24476 }, { "epoch": 0.8720451751963946, "grad_norm": 1.1529154637010122, "learning_rate": 8.463334110232491e-07, "loss": 0.3414, "step": 24477 }, { "epoch": 0.8720808023228886, "grad_norm": 1.2896786059946312, "learning_rate": 8.458688725205899e-07, "loss": 0.4273, "step": 24478 }, { "epoch": 0.8721164294493827, "grad_norm": 1.7751819340911204, "learning_rate": 8.454044559115837e-07, "loss": 0.495, "step": 24479 }, { "epoch": 0.8721520565758769, "grad_norm": 1.273555854274529, "learning_rate": 8.449401612024155e-07, "loss": 0.3735, "step": 24480 }, { "epoch": 0.872187683702371, "grad_norm": 1.7530552804068582, "learning_rate": 8.444759883992693e-07, "loss": 0.5991, "step": 24481 }, { "epoch": 0.8722233108288651, "grad_norm": 1.4821678362909396, "learning_rate": 8.440119375083233e-07, "loss": 0.5074, "step": 24482 }, { "epoch": 0.8722589379553592, "grad_norm": 1.5694107322265438, "learning_rate": 8.435480085357583e-07, "loss": 0.5076, "step": 24483 }, { "epoch": 0.8722945650818533, "grad_norm": 1.159282201141508, "learning_rate": 8.430842014877505e-07, "loss": 0.347, "step": 24484 }, { "epoch": 0.8723301922083474, "grad_norm": 1.661222015184143, "learning_rate": 8.426205163704793e-07, "loss": 0.5353, "step": 24485 }, { "epoch": 0.8723658193348416, "grad_norm": 1.2695216122871518, "learning_rate": 8.421569531901153e-07, "loss": 0.4784, "step": 24486 }, { "epoch": 0.8724014464613357, "grad_norm": 1.2277875321974399, "learning_rate": 8.416935119528336e-07, "loss": 0.3845, "step": 24487 }, { "epoch": 0.8724370735878297, "grad_norm": 1.4227144994780516, "learning_rate": 8.412301926648025e-07, "loss": 0.5229, "step": 24488 }, { "epoch": 0.8724727007143239, "grad_norm": 1.7941365758017878, "learning_rate": 8.407669953321929e-07, "loss": 0.515, "step": 24489 }, { "epoch": 0.872508327840818, "grad_norm": 1.1663278308833596, "learning_rate": 8.403039199611729e-07, "loss": 0.4087, "step": 24490 }, { "epoch": 0.8725439549673121, "grad_norm": 1.3733015168182714, "learning_rate": 8.398409665579077e-07, "loss": 0.4681, "step": 24491 }, { "epoch": 0.8725795820938063, "grad_norm": 1.9394767679240341, "learning_rate": 8.393781351285657e-07, "loss": 0.5102, "step": 24492 }, { "epoch": 0.8726152092203003, "grad_norm": 1.1512535833658255, "learning_rate": 8.389154256793042e-07, "loss": 0.3658, "step": 24493 }, { "epoch": 0.8726508363467944, "grad_norm": 1.5646440534401396, "learning_rate": 8.384528382162882e-07, "loss": 0.6004, "step": 24494 }, { "epoch": 0.8726864634732886, "grad_norm": 1.242696987311207, "learning_rate": 8.379903727456751e-07, "loss": 0.41, "step": 24495 }, { "epoch": 0.8727220905997827, "grad_norm": 1.6742266851105354, "learning_rate": 8.375280292736254e-07, "loss": 0.4859, "step": 24496 }, { "epoch": 0.8727577177262767, "grad_norm": 1.642395328121479, "learning_rate": 8.370658078062933e-07, "loss": 0.4806, "step": 24497 }, { "epoch": 0.8727933448527709, "grad_norm": 1.6872228199115487, "learning_rate": 8.366037083498335e-07, "loss": 0.4858, "step": 24498 }, { "epoch": 0.872828971979265, "grad_norm": 1.5141468850356516, "learning_rate": 8.361417309104003e-07, "loss": 0.47, "step": 24499 }, { "epoch": 0.8728645991057591, "grad_norm": 1.3129257960514162, "learning_rate": 8.356798754941475e-07, "loss": 0.4361, "step": 24500 }, { "epoch": 0.8729002262322533, "grad_norm": 1.3162255244767698, "learning_rate": 8.352181421072203e-07, "loss": 0.3931, "step": 24501 }, { "epoch": 0.8729358533587473, "grad_norm": 1.2037326959414463, "learning_rate": 8.347565307557693e-07, "loss": 0.3792, "step": 24502 }, { "epoch": 0.8729714804852414, "grad_norm": 1.134075830952552, "learning_rate": 8.342950414459439e-07, "loss": 0.366, "step": 24503 }, { "epoch": 0.8730071076117356, "grad_norm": 1.184439407692561, "learning_rate": 8.338336741838837e-07, "loss": 0.3746, "step": 24504 }, { "epoch": 0.8730427347382297, "grad_norm": 1.3104959873355464, "learning_rate": 8.33372428975735e-07, "loss": 0.3944, "step": 24505 }, { "epoch": 0.8730783618647238, "grad_norm": 1.703814947666402, "learning_rate": 8.329113058276394e-07, "loss": 0.4387, "step": 24506 }, { "epoch": 0.873113988991218, "grad_norm": 1.17387502890598, "learning_rate": 8.3245030474574e-07, "loss": 0.3972, "step": 24507 }, { "epoch": 0.873149616117712, "grad_norm": 1.6470351897467332, "learning_rate": 8.319894257361694e-07, "loss": 0.5088, "step": 24508 }, { "epoch": 0.8731852432442061, "grad_norm": 1.5547999145795948, "learning_rate": 8.315286688050705e-07, "loss": 0.4921, "step": 24509 }, { "epoch": 0.8732208703707003, "grad_norm": 1.6100947107638306, "learning_rate": 8.310680339585731e-07, "loss": 0.4682, "step": 24510 }, { "epoch": 0.8732564974971944, "grad_norm": 1.4221814721221666, "learning_rate": 8.306075212028153e-07, "loss": 0.4442, "step": 24511 }, { "epoch": 0.8732921246236884, "grad_norm": 1.2173777494356592, "learning_rate": 8.301471305439246e-07, "loss": 0.3839, "step": 24512 }, { "epoch": 0.8733277517501826, "grad_norm": 1.6145738051059544, "learning_rate": 8.296868619880372e-07, "loss": 0.4796, "step": 24513 }, { "epoch": 0.8733633788766767, "grad_norm": 1.3267069958352011, "learning_rate": 8.29226715541277e-07, "loss": 0.4537, "step": 24514 }, { "epoch": 0.8733990060031708, "grad_norm": 1.7529321511485212, "learning_rate": 8.287666912097736e-07, "loss": 0.5516, "step": 24515 }, { "epoch": 0.873434633129665, "grad_norm": 1.4515538788470008, "learning_rate": 8.283067889996543e-07, "loss": 0.5558, "step": 24516 }, { "epoch": 0.873470260256159, "grad_norm": 1.1574380292716189, "learning_rate": 8.278470089170377e-07, "loss": 0.3938, "step": 24517 }, { "epoch": 0.8735058873826531, "grad_norm": 1.2827111070272497, "learning_rate": 8.273873509680519e-07, "loss": 0.3828, "step": 24518 }, { "epoch": 0.8735415145091473, "grad_norm": 1.6295272611330023, "learning_rate": 8.269278151588123e-07, "loss": 0.5207, "step": 24519 }, { "epoch": 0.8735771416356414, "grad_norm": 1.7598335805450347, "learning_rate": 8.264684014954405e-07, "loss": 0.4917, "step": 24520 }, { "epoch": 0.8736127687621354, "grad_norm": 1.3000958256807824, "learning_rate": 8.260091099840539e-07, "loss": 0.4025, "step": 24521 }, { "epoch": 0.8736483958886296, "grad_norm": 1.711163161975585, "learning_rate": 8.255499406307698e-07, "loss": 0.4152, "step": 24522 }, { "epoch": 0.8736840230151237, "grad_norm": 1.180672502512435, "learning_rate": 8.250908934417001e-07, "loss": 0.3632, "step": 24523 }, { "epoch": 0.8737196501416178, "grad_norm": 1.0693010886306382, "learning_rate": 8.246319684229586e-07, "loss": 0.3605, "step": 24524 }, { "epoch": 0.873755277268112, "grad_norm": 1.070571716962699, "learning_rate": 8.241731655806539e-07, "loss": 0.3511, "step": 24525 }, { "epoch": 0.873790904394606, "grad_norm": 1.3641402388406256, "learning_rate": 8.237144849208978e-07, "loss": 0.4799, "step": 24526 }, { "epoch": 0.8738265315211001, "grad_norm": 1.5381591227029534, "learning_rate": 8.232559264497975e-07, "loss": 0.5426, "step": 24527 }, { "epoch": 0.8738621586475943, "grad_norm": 1.1805434037144449, "learning_rate": 8.22797490173457e-07, "loss": 0.4015, "step": 24528 }, { "epoch": 0.8738977857740884, "grad_norm": 1.3641441788665662, "learning_rate": 8.22339176097986e-07, "loss": 0.5183, "step": 24529 }, { "epoch": 0.8739334129005825, "grad_norm": 1.626421201788427, "learning_rate": 8.218809842294806e-07, "loss": 0.3768, "step": 24530 }, { "epoch": 0.8739690400270766, "grad_norm": 1.5287432378459447, "learning_rate": 8.214229145740471e-07, "loss": 0.4938, "step": 24531 }, { "epoch": 0.8740046671535707, "grad_norm": 1.1194318474934541, "learning_rate": 8.209649671377817e-07, "loss": 0.3949, "step": 24532 }, { "epoch": 0.8740402942800648, "grad_norm": 1.24926440532452, "learning_rate": 8.205071419267851e-07, "loss": 0.3871, "step": 24533 }, { "epoch": 0.874075921406559, "grad_norm": 1.580296755966684, "learning_rate": 8.200494389471503e-07, "loss": 0.434, "step": 24534 }, { "epoch": 0.8741115485330531, "grad_norm": 1.232276562937335, "learning_rate": 8.195918582049734e-07, "loss": 0.5001, "step": 24535 }, { "epoch": 0.8741471756595471, "grad_norm": 1.277183650215407, "learning_rate": 8.191343997063472e-07, "loss": 0.4181, "step": 24536 }, { "epoch": 0.8741828027860413, "grad_norm": 1.3559720331246488, "learning_rate": 8.186770634573637e-07, "loss": 0.4088, "step": 24537 }, { "epoch": 0.8742184299125354, "grad_norm": 1.210237235928576, "learning_rate": 8.182198494641147e-07, "loss": 0.4217, "step": 24538 }, { "epoch": 0.8742540570390295, "grad_norm": 1.478751757863255, "learning_rate": 8.177627577326841e-07, "loss": 0.5483, "step": 24539 }, { "epoch": 0.8742896841655237, "grad_norm": 1.1702169193811705, "learning_rate": 8.173057882691615e-07, "loss": 0.3921, "step": 24540 }, { "epoch": 0.8743253112920177, "grad_norm": 1.195607118975837, "learning_rate": 8.168489410796298e-07, "loss": 0.3936, "step": 24541 }, { "epoch": 0.8743609384185118, "grad_norm": 1.6597687577187823, "learning_rate": 8.163922161701732e-07, "loss": 0.5218, "step": 24542 }, { "epoch": 0.874396565545006, "grad_norm": 2.1989735364154375, "learning_rate": 8.159356135468721e-07, "loss": 0.5185, "step": 24543 }, { "epoch": 0.8744321926715001, "grad_norm": 1.6891370037953517, "learning_rate": 8.154791332158107e-07, "loss": 0.5382, "step": 24544 }, { "epoch": 0.8744678197979942, "grad_norm": 1.3199751282140582, "learning_rate": 8.15022775183062e-07, "loss": 0.3751, "step": 24545 }, { "epoch": 0.8745034469244883, "grad_norm": 2.222603451850376, "learning_rate": 8.145665394547064e-07, "loss": 0.53, "step": 24546 }, { "epoch": 0.8745390740509824, "grad_norm": 1.4982255311125636, "learning_rate": 8.14110426036816e-07, "loss": 0.523, "step": 24547 }, { "epoch": 0.8745747011774765, "grad_norm": 1.231494945460337, "learning_rate": 8.136544349354669e-07, "loss": 0.3849, "step": 24548 }, { "epoch": 0.8746103283039707, "grad_norm": 1.1504634474829745, "learning_rate": 8.131985661567287e-07, "loss": 0.4154, "step": 24549 }, { "epoch": 0.8746459554304647, "grad_norm": 1.1962048289682607, "learning_rate": 8.127428197066711e-07, "loss": 0.3798, "step": 24550 }, { "epoch": 0.8746815825569588, "grad_norm": 1.4479322599338516, "learning_rate": 8.122871955913681e-07, "loss": 0.4969, "step": 24551 }, { "epoch": 0.874717209683453, "grad_norm": 1.6135749477574994, "learning_rate": 8.118316938168813e-07, "loss": 0.5232, "step": 24552 }, { "epoch": 0.8747528368099471, "grad_norm": 1.2737849942199364, "learning_rate": 8.113763143892784e-07, "loss": 0.4504, "step": 24553 }, { "epoch": 0.8747884639364412, "grad_norm": 1.3026694095629179, "learning_rate": 8.109210573146209e-07, "loss": 0.4064, "step": 24554 }, { "epoch": 0.8748240910629353, "grad_norm": 1.4457541628090007, "learning_rate": 8.104659225989753e-07, "loss": 0.4122, "step": 24555 }, { "epoch": 0.8748597181894294, "grad_norm": 1.7728626797387999, "learning_rate": 8.100109102483955e-07, "loss": 0.5351, "step": 24556 }, { "epoch": 0.8748953453159235, "grad_norm": 1.1940281920912246, "learning_rate": 8.095560202689446e-07, "loss": 0.4189, "step": 24557 }, { "epoch": 0.8749309724424177, "grad_norm": 1.8427108847542728, "learning_rate": 8.091012526666797e-07, "loss": 0.5258, "step": 24558 }, { "epoch": 0.8749665995689118, "grad_norm": 1.2190618713192325, "learning_rate": 8.086466074476562e-07, "loss": 0.4426, "step": 24559 }, { "epoch": 0.8750022266954058, "grad_norm": 1.545404782170534, "learning_rate": 8.08192084617927e-07, "loss": 0.4836, "step": 24560 }, { "epoch": 0.8750378538219, "grad_norm": 1.6718485373582368, "learning_rate": 8.077376841835438e-07, "loss": 0.5545, "step": 24561 }, { "epoch": 0.8750734809483941, "grad_norm": 1.4218388309077437, "learning_rate": 8.072834061505608e-07, "loss": 0.4312, "step": 24562 }, { "epoch": 0.8751091080748882, "grad_norm": 1.5254450485029136, "learning_rate": 8.068292505250231e-07, "loss": 0.4755, "step": 24563 }, { "epoch": 0.8751447352013824, "grad_norm": 1.4758352928091394, "learning_rate": 8.063752173129791e-07, "loss": 0.5012, "step": 24564 }, { "epoch": 0.8751803623278764, "grad_norm": 1.530977950985027, "learning_rate": 8.059213065204741e-07, "loss": 0.478, "step": 24565 }, { "epoch": 0.8752159894543705, "grad_norm": 1.1020988961044673, "learning_rate": 8.054675181535565e-07, "loss": 0.4083, "step": 24566 }, { "epoch": 0.8752516165808647, "grad_norm": 1.4236010697782961, "learning_rate": 8.050138522182627e-07, "loss": 0.4834, "step": 24567 }, { "epoch": 0.8752872437073588, "grad_norm": 1.1846537879617143, "learning_rate": 8.045603087206388e-07, "loss": 0.359, "step": 24568 }, { "epoch": 0.8753228708338529, "grad_norm": 1.1581736400402, "learning_rate": 8.041068876667191e-07, "loss": 0.3563, "step": 24569 }, { "epoch": 0.875358497960347, "grad_norm": 1.2623014375096173, "learning_rate": 8.036535890625463e-07, "loss": 0.4359, "step": 24570 }, { "epoch": 0.8753941250868411, "grad_norm": 1.1702683563799896, "learning_rate": 8.032004129141513e-07, "loss": 0.3952, "step": 24571 }, { "epoch": 0.8754297522133352, "grad_norm": 1.1711143528406573, "learning_rate": 8.027473592275714e-07, "loss": 0.3782, "step": 24572 }, { "epoch": 0.8754653793398294, "grad_norm": 1.1566523296622953, "learning_rate": 8.022944280088385e-07, "loss": 0.3715, "step": 24573 }, { "epoch": 0.8755010064663235, "grad_norm": 1.4941502728376548, "learning_rate": 8.018416192639844e-07, "loss": 0.5049, "step": 24574 }, { "epoch": 0.8755366335928175, "grad_norm": 1.2746843451254073, "learning_rate": 8.01388932999041e-07, "loss": 0.4124, "step": 24575 }, { "epoch": 0.8755722607193117, "grad_norm": 1.5233799649378115, "learning_rate": 8.009363692200301e-07, "loss": 0.4926, "step": 24576 }, { "epoch": 0.8756078878458058, "grad_norm": 1.3188085435018964, "learning_rate": 8.004839279329834e-07, "loss": 0.3779, "step": 24577 }, { "epoch": 0.8756435149722999, "grad_norm": 1.3884585409683108, "learning_rate": 8.00031609143922e-07, "loss": 0.4262, "step": 24578 }, { "epoch": 0.875679142098794, "grad_norm": 1.6842530137001344, "learning_rate": 7.995794128588696e-07, "loss": 0.5428, "step": 24579 }, { "epoch": 0.8757147692252881, "grad_norm": 1.5960992321932987, "learning_rate": 7.991273390838483e-07, "loss": 0.5658, "step": 24580 }, { "epoch": 0.8757503963517822, "grad_norm": 1.12632725193001, "learning_rate": 7.986753878248798e-07, "loss": 0.3968, "step": 24581 }, { "epoch": 0.8757860234782764, "grad_norm": 1.58832341786982, "learning_rate": 7.982235590879772e-07, "loss": 0.4763, "step": 24582 }, { "epoch": 0.8758216506047705, "grad_norm": 1.3298520979076958, "learning_rate": 7.9777185287916e-07, "loss": 0.4385, "step": 24583 }, { "epoch": 0.8758572777312645, "grad_norm": 1.3360627718777571, "learning_rate": 7.973202692044447e-07, "loss": 0.4231, "step": 24584 }, { "epoch": 0.8758929048577587, "grad_norm": 1.1305468786963253, "learning_rate": 7.968688080698394e-07, "loss": 0.368, "step": 24585 }, { "epoch": 0.8759285319842528, "grad_norm": 1.5301929118767568, "learning_rate": 7.964174694813609e-07, "loss": 0.5074, "step": 24586 }, { "epoch": 0.8759641591107469, "grad_norm": 1.5046541191824634, "learning_rate": 7.959662534450152e-07, "loss": 0.5041, "step": 24587 }, { "epoch": 0.8759997862372411, "grad_norm": 2.073153929691388, "learning_rate": 7.955151599668121e-07, "loss": 0.4742, "step": 24588 }, { "epoch": 0.8760354133637351, "grad_norm": 1.530689146100897, "learning_rate": 7.950641890527577e-07, "loss": 0.5607, "step": 24589 }, { "epoch": 0.8760710404902292, "grad_norm": 1.4387574659691067, "learning_rate": 7.946133407088596e-07, "loss": 0.3954, "step": 24590 }, { "epoch": 0.8761066676167234, "grad_norm": 1.407502682860419, "learning_rate": 7.941626149411174e-07, "loss": 0.4771, "step": 24591 }, { "epoch": 0.8761422947432175, "grad_norm": 1.6364430184131966, "learning_rate": 7.937120117555363e-07, "loss": 0.5206, "step": 24592 }, { "epoch": 0.8761779218697116, "grad_norm": 1.7286222536533447, "learning_rate": 7.932615311581127e-07, "loss": 0.5977, "step": 24593 }, { "epoch": 0.8762135489962057, "grad_norm": 1.8390754039802342, "learning_rate": 7.92811173154846e-07, "loss": 0.4828, "step": 24594 }, { "epoch": 0.8762491761226998, "grad_norm": 1.4899685516037622, "learning_rate": 7.92360937751735e-07, "loss": 0.5305, "step": 24595 }, { "epoch": 0.8762848032491939, "grad_norm": 2.0837561870342824, "learning_rate": 7.919108249547736e-07, "loss": 0.5323, "step": 24596 }, { "epoch": 0.8763204303756881, "grad_norm": 1.1419520402019112, "learning_rate": 7.914608347699571e-07, "loss": 0.4424, "step": 24597 }, { "epoch": 0.8763560575021822, "grad_norm": 1.172874174667396, "learning_rate": 7.910109672032751e-07, "loss": 0.3956, "step": 24598 }, { "epoch": 0.8763916846286762, "grad_norm": 1.2305208093461513, "learning_rate": 7.905612222607195e-07, "loss": 0.3618, "step": 24599 }, { "epoch": 0.8764273117551704, "grad_norm": 1.6469098354032943, "learning_rate": 7.901115999482767e-07, "loss": 0.5142, "step": 24600 }, { "epoch": 0.8764629388816645, "grad_norm": 1.4464279698609124, "learning_rate": 7.896621002719362e-07, "loss": 0.5255, "step": 24601 }, { "epoch": 0.8764985660081586, "grad_norm": 1.323510818162325, "learning_rate": 7.892127232376811e-07, "loss": 0.4004, "step": 24602 }, { "epoch": 0.8765341931346528, "grad_norm": 1.2704435477536318, "learning_rate": 7.887634688515e-07, "loss": 0.3834, "step": 24603 }, { "epoch": 0.8765698202611468, "grad_norm": 1.1299178160602514, "learning_rate": 7.88314337119368e-07, "loss": 0.4015, "step": 24604 }, { "epoch": 0.8766054473876409, "grad_norm": 1.3455423947332341, "learning_rate": 7.878653280472726e-07, "loss": 0.4132, "step": 24605 }, { "epoch": 0.8766410745141351, "grad_norm": 1.3132705736386467, "learning_rate": 7.874164416411856e-07, "loss": 0.4348, "step": 24606 }, { "epoch": 0.8766767016406292, "grad_norm": 1.4268188498263463, "learning_rate": 7.86967677907089e-07, "loss": 0.5431, "step": 24607 }, { "epoch": 0.8767123287671232, "grad_norm": 1.072301491429762, "learning_rate": 7.86519036850959e-07, "loss": 0.4109, "step": 24608 }, { "epoch": 0.8767479558936174, "grad_norm": 1.6830407452476155, "learning_rate": 7.860705184787665e-07, "loss": 0.6506, "step": 24609 }, { "epoch": 0.8767835830201115, "grad_norm": 1.4815551744919795, "learning_rate": 7.856221227964844e-07, "loss": 0.5474, "step": 24610 }, { "epoch": 0.8768192101466056, "grad_norm": 1.1578466599348445, "learning_rate": 7.851738498100848e-07, "loss": 0.3767, "step": 24611 }, { "epoch": 0.8768548372730998, "grad_norm": 1.845274002114819, "learning_rate": 7.847256995255381e-07, "loss": 0.5302, "step": 24612 }, { "epoch": 0.8768904643995938, "grad_norm": 1.7402401151728206, "learning_rate": 7.842776719488076e-07, "loss": 0.4874, "step": 24613 }, { "epoch": 0.8769260915260879, "grad_norm": 1.2663319826927566, "learning_rate": 7.83829767085863e-07, "loss": 0.3892, "step": 24614 }, { "epoch": 0.8769617186525821, "grad_norm": 1.256297254779624, "learning_rate": 7.833819849426649e-07, "loss": 0.4012, "step": 24615 }, { "epoch": 0.8769973457790762, "grad_norm": 1.4533106380234082, "learning_rate": 7.829343255251787e-07, "loss": 0.4993, "step": 24616 }, { "epoch": 0.8770329729055703, "grad_norm": 1.2628549581368231, "learning_rate": 7.824867888393628e-07, "loss": 0.4271, "step": 24617 }, { "epoch": 0.8770686000320644, "grad_norm": 1.56339019807995, "learning_rate": 7.820393748911792e-07, "loss": 0.5149, "step": 24618 }, { "epoch": 0.8771042271585585, "grad_norm": 1.5628892871135265, "learning_rate": 7.815920836865854e-07, "loss": 0.5139, "step": 24619 }, { "epoch": 0.8771398542850526, "grad_norm": 1.1281177396954734, "learning_rate": 7.811449152315342e-07, "loss": 0.3576, "step": 24620 }, { "epoch": 0.8771754814115468, "grad_norm": 1.5788293815354977, "learning_rate": 7.806978695319844e-07, "loss": 0.5167, "step": 24621 }, { "epoch": 0.8772111085380409, "grad_norm": 1.433450310942877, "learning_rate": 7.802509465938846e-07, "loss": 0.3457, "step": 24622 }, { "epoch": 0.8772467356645349, "grad_norm": 1.3447739487587151, "learning_rate": 7.798041464231898e-07, "loss": 0.4263, "step": 24623 }, { "epoch": 0.8772823627910291, "grad_norm": 1.2352663344391444, "learning_rate": 7.793574690258454e-07, "loss": 0.4455, "step": 24624 }, { "epoch": 0.8773179899175232, "grad_norm": 1.2551308067243856, "learning_rate": 7.789109144078e-07, "loss": 0.4319, "step": 24625 }, { "epoch": 0.8773536170440173, "grad_norm": 2.2858660694385398, "learning_rate": 7.78464482575002e-07, "loss": 0.498, "step": 24626 }, { "epoch": 0.8773892441705115, "grad_norm": 1.620743459332183, "learning_rate": 7.780181735333969e-07, "loss": 0.4956, "step": 24627 }, { "epoch": 0.8774248712970055, "grad_norm": 1.1354629930181344, "learning_rate": 7.775719872889232e-07, "loss": 0.4051, "step": 24628 }, { "epoch": 0.8774604984234996, "grad_norm": 1.8324437908767541, "learning_rate": 7.771259238475248e-07, "loss": 0.5434, "step": 24629 }, { "epoch": 0.8774961255499938, "grad_norm": 1.1617553545783799, "learning_rate": 7.766799832151428e-07, "loss": 0.3821, "step": 24630 }, { "epoch": 0.8775317526764879, "grad_norm": 2.2051278121997147, "learning_rate": 7.762341653977112e-07, "loss": 0.5733, "step": 24631 }, { "epoch": 0.8775673798029819, "grad_norm": 1.8341444354580334, "learning_rate": 7.757884704011698e-07, "loss": 0.4885, "step": 24632 }, { "epoch": 0.8776030069294761, "grad_norm": 1.587146278358631, "learning_rate": 7.753428982314515e-07, "loss": 0.5648, "step": 24633 }, { "epoch": 0.8776386340559702, "grad_norm": 1.4241118872792364, "learning_rate": 7.748974488944927e-07, "loss": 0.4728, "step": 24634 }, { "epoch": 0.8776742611824643, "grad_norm": 1.1597833682126102, "learning_rate": 7.744521223962198e-07, "loss": 0.39, "step": 24635 }, { "epoch": 0.8777098883089585, "grad_norm": 1.4377746168134515, "learning_rate": 7.740069187425669e-07, "loss": 0.4626, "step": 24636 }, { "epoch": 0.8777455154354525, "grad_norm": 1.7826442805118248, "learning_rate": 7.735618379394594e-07, "loss": 0.492, "step": 24637 }, { "epoch": 0.8777811425619466, "grad_norm": 1.534207357781366, "learning_rate": 7.731168799928257e-07, "loss": 0.4949, "step": 24638 }, { "epoch": 0.8778167696884408, "grad_norm": 5.502055885836358, "learning_rate": 7.726720449085878e-07, "loss": 0.4468, "step": 24639 }, { "epoch": 0.8778523968149349, "grad_norm": 1.334646504080341, "learning_rate": 7.722273326926733e-07, "loss": 0.4473, "step": 24640 }, { "epoch": 0.877888023941429, "grad_norm": 1.3065643490745842, "learning_rate": 7.717827433510005e-07, "loss": 0.3804, "step": 24641 }, { "epoch": 0.8779236510679231, "grad_norm": 1.5545433085442735, "learning_rate": 7.713382768894905e-07, "loss": 0.5325, "step": 24642 }, { "epoch": 0.8779592781944172, "grad_norm": 1.1664306886717142, "learning_rate": 7.708939333140642e-07, "loss": 0.4248, "step": 24643 }, { "epoch": 0.8779949053209113, "grad_norm": 1.228277625903114, "learning_rate": 7.704497126306332e-07, "loss": 0.416, "step": 24644 }, { "epoch": 0.8780305324474055, "grad_norm": 1.0357707464749746, "learning_rate": 7.700056148451174e-07, "loss": 0.3787, "step": 24645 }, { "epoch": 0.8780661595738996, "grad_norm": 1.448397443088073, "learning_rate": 7.695616399634265e-07, "loss": 0.5723, "step": 24646 }, { "epoch": 0.8781017867003936, "grad_norm": 2.191614841505224, "learning_rate": 7.691177879914746e-07, "loss": 0.5004, "step": 24647 }, { "epoch": 0.8781374138268878, "grad_norm": 1.729255051201821, "learning_rate": 7.686740589351704e-07, "loss": 0.4891, "step": 24648 }, { "epoch": 0.8781730409533819, "grad_norm": 1.2183988501864995, "learning_rate": 7.682304528004258e-07, "loss": 0.4063, "step": 24649 }, { "epoch": 0.878208668079876, "grad_norm": 1.7624821754691156, "learning_rate": 7.67786969593144e-07, "loss": 0.5443, "step": 24650 }, { "epoch": 0.8782442952063702, "grad_norm": 1.416180623480366, "learning_rate": 7.673436093192322e-07, "loss": 0.4857, "step": 24651 }, { "epoch": 0.8782799223328642, "grad_norm": 1.3621814480413292, "learning_rate": 7.669003719845925e-07, "loss": 0.4216, "step": 24652 }, { "epoch": 0.8783155494593583, "grad_norm": 1.2411364378530083, "learning_rate": 7.664572575951279e-07, "loss": 0.4261, "step": 24653 }, { "epoch": 0.8783511765858525, "grad_norm": 1.3210474011670188, "learning_rate": 7.660142661567383e-07, "loss": 0.3612, "step": 24654 }, { "epoch": 0.8783868037123466, "grad_norm": 1.6724679872580177, "learning_rate": 7.655713976753232e-07, "loss": 0.5303, "step": 24655 }, { "epoch": 0.8784224308388406, "grad_norm": 1.6282622757538863, "learning_rate": 7.651286521567813e-07, "loss": 0.4187, "step": 24656 }, { "epoch": 0.8784580579653348, "grad_norm": 1.3138222377305382, "learning_rate": 7.646860296070047e-07, "loss": 0.4358, "step": 24657 }, { "epoch": 0.8784936850918289, "grad_norm": 1.227694132469873, "learning_rate": 7.642435300318906e-07, "loss": 0.4181, "step": 24658 }, { "epoch": 0.878529312218323, "grad_norm": 1.4325591543574183, "learning_rate": 7.638011534373269e-07, "loss": 0.5573, "step": 24659 }, { "epoch": 0.8785649393448172, "grad_norm": 1.4831170896369186, "learning_rate": 7.633588998292086e-07, "loss": 0.42, "step": 24660 }, { "epoch": 0.8786005664713112, "grad_norm": 1.4544462451451505, "learning_rate": 7.629167692134199e-07, "loss": 0.5289, "step": 24661 }, { "epoch": 0.8786361935978053, "grad_norm": 3.591528950234765, "learning_rate": 7.624747615958516e-07, "loss": 0.4405, "step": 24662 }, { "epoch": 0.8786718207242995, "grad_norm": 1.6849503369254037, "learning_rate": 7.620328769823892e-07, "loss": 0.4926, "step": 24663 }, { "epoch": 0.8787074478507936, "grad_norm": 1.558820402160586, "learning_rate": 7.615911153789146e-07, "loss": 0.587, "step": 24664 }, { "epoch": 0.8787430749772877, "grad_norm": 1.2058951086843206, "learning_rate": 7.611494767913141e-07, "loss": 0.4119, "step": 24665 }, { "epoch": 0.8787787021037818, "grad_norm": 1.2299473368321243, "learning_rate": 7.607079612254642e-07, "loss": 0.4261, "step": 24666 }, { "epoch": 0.8788143292302759, "grad_norm": 1.4517402936256874, "learning_rate": 7.602665686872468e-07, "loss": 0.5078, "step": 24667 }, { "epoch": 0.87884995635677, "grad_norm": 1.2189778342176993, "learning_rate": 7.598252991825372e-07, "loss": 0.4183, "step": 24668 }, { "epoch": 0.8788855834832642, "grad_norm": 1.546528093995757, "learning_rate": 7.593841527172119e-07, "loss": 0.4971, "step": 24669 }, { "epoch": 0.8789212106097583, "grad_norm": 1.3053319674184125, "learning_rate": 7.589431292971461e-07, "loss": 0.4019, "step": 24670 }, { "epoch": 0.8789568377362523, "grad_norm": 1.7152073646313633, "learning_rate": 7.585022289282129e-07, "loss": 0.5221, "step": 24671 }, { "epoch": 0.8789924648627465, "grad_norm": 1.2572561788714522, "learning_rate": 7.580614516162799e-07, "loss": 0.409, "step": 24672 }, { "epoch": 0.8790280919892406, "grad_norm": 1.6517709994251975, "learning_rate": 7.5762079736722e-07, "loss": 0.5121, "step": 24673 }, { "epoch": 0.8790637191157347, "grad_norm": 1.418605668457572, "learning_rate": 7.571802661868976e-07, "loss": 0.5521, "step": 24674 }, { "epoch": 0.8790993462422289, "grad_norm": 1.553765544707057, "learning_rate": 7.567398580811802e-07, "loss": 0.4398, "step": 24675 }, { "epoch": 0.8791349733687229, "grad_norm": 1.1503615003074414, "learning_rate": 7.562995730559331e-07, "loss": 0.4179, "step": 24676 }, { "epoch": 0.879170600495217, "grad_norm": 1.7066621885977142, "learning_rate": 7.55859411117017e-07, "loss": 0.4875, "step": 24677 }, { "epoch": 0.8792062276217112, "grad_norm": 1.2141336664624132, "learning_rate": 7.554193722702962e-07, "loss": 0.3606, "step": 24678 }, { "epoch": 0.8792418547482053, "grad_norm": 1.2895280003738088, "learning_rate": 7.549794565216273e-07, "loss": 0.3907, "step": 24679 }, { "epoch": 0.8792774818746995, "grad_norm": 1.7132048852136492, "learning_rate": 7.545396638768698e-07, "loss": 0.5657, "step": 24680 }, { "epoch": 0.8793131090011935, "grad_norm": 1.1930982870433515, "learning_rate": 7.54099994341877e-07, "loss": 0.4277, "step": 24681 }, { "epoch": 0.8793487361276876, "grad_norm": 1.1405425023218778, "learning_rate": 7.536604479225074e-07, "loss": 0.3871, "step": 24682 }, { "epoch": 0.8793843632541818, "grad_norm": 1.1217487719612615, "learning_rate": 7.532210246246097e-07, "loss": 0.3818, "step": 24683 }, { "epoch": 0.8794199903806759, "grad_norm": 1.3386815390784033, "learning_rate": 7.52781724454037e-07, "loss": 0.4383, "step": 24684 }, { "epoch": 0.8794556175071699, "grad_norm": 1.6528083601235595, "learning_rate": 7.52342547416639e-07, "loss": 0.5188, "step": 24685 }, { "epoch": 0.8794912446336641, "grad_norm": 1.257425605398843, "learning_rate": 7.519034935182656e-07, "loss": 0.4093, "step": 24686 }, { "epoch": 0.8795268717601582, "grad_norm": 1.4740282121036505, "learning_rate": 7.514645627647599e-07, "loss": 0.5101, "step": 24687 }, { "epoch": 0.8795624988866523, "grad_norm": 1.454259992106302, "learning_rate": 7.51025755161967e-07, "loss": 0.4203, "step": 24688 }, { "epoch": 0.8795981260131465, "grad_norm": 1.1551212275737, "learning_rate": 7.505870707157337e-07, "loss": 0.4029, "step": 24689 }, { "epoch": 0.8796337531396405, "grad_norm": 1.1946404258911463, "learning_rate": 7.501485094318961e-07, "loss": 0.4094, "step": 24690 }, { "epoch": 0.8796693802661346, "grad_norm": 1.19253913102053, "learning_rate": 7.497100713162964e-07, "loss": 0.3782, "step": 24691 }, { "epoch": 0.8797050073926288, "grad_norm": 1.3621658181107803, "learning_rate": 7.492717563747721e-07, "loss": 0.4334, "step": 24692 }, { "epoch": 0.8797406345191229, "grad_norm": 1.3091865229574315, "learning_rate": 7.488335646131628e-07, "loss": 0.4116, "step": 24693 }, { "epoch": 0.879776261645617, "grad_norm": 1.9531580900563685, "learning_rate": 7.483954960372985e-07, "loss": 0.4855, "step": 24694 }, { "epoch": 0.8798118887721111, "grad_norm": 1.2736170664234352, "learning_rate": 7.479575506530168e-07, "loss": 0.4135, "step": 24695 }, { "epoch": 0.8798475158986052, "grad_norm": 1.444835776651226, "learning_rate": 7.47519728466145e-07, "loss": 0.4684, "step": 24696 }, { "epoch": 0.8798831430250993, "grad_norm": 1.95560820671049, "learning_rate": 7.470820294825177e-07, "loss": 0.5117, "step": 24697 }, { "epoch": 0.8799187701515935, "grad_norm": 2.5599501680318744, "learning_rate": 7.466444537079575e-07, "loss": 0.4664, "step": 24698 }, { "epoch": 0.8799543972780876, "grad_norm": 1.3109606129435833, "learning_rate": 7.462070011482958e-07, "loss": 0.3903, "step": 24699 }, { "epoch": 0.8799900244045816, "grad_norm": 1.4644058593414029, "learning_rate": 7.457696718093543e-07, "loss": 0.5382, "step": 24700 }, { "epoch": 0.8800256515310758, "grad_norm": 1.2173048374325015, "learning_rate": 7.453324656969585e-07, "loss": 0.4021, "step": 24701 }, { "epoch": 0.8800612786575699, "grad_norm": 2.1702870874104874, "learning_rate": 7.448953828169314e-07, "loss": 0.4959, "step": 24702 }, { "epoch": 0.880096905784064, "grad_norm": 1.3290177353369246, "learning_rate": 7.444584231750895e-07, "loss": 0.508, "step": 24703 }, { "epoch": 0.8801325329105582, "grad_norm": 1.763510143345917, "learning_rate": 7.440215867772548e-07, "loss": 0.6015, "step": 24704 }, { "epoch": 0.8801681600370522, "grad_norm": 1.32574606459312, "learning_rate": 7.435848736292406e-07, "loss": 0.4159, "step": 24705 }, { "epoch": 0.8802037871635463, "grad_norm": 1.3002811628734126, "learning_rate": 7.431482837368631e-07, "loss": 0.4113, "step": 24706 }, { "epoch": 0.8802394142900405, "grad_norm": 1.2947093383274422, "learning_rate": 7.427118171059367e-07, "loss": 0.4133, "step": 24707 }, { "epoch": 0.8802750414165346, "grad_norm": 1.5756008854154409, "learning_rate": 7.422754737422744e-07, "loss": 0.5195, "step": 24708 }, { "epoch": 0.8803106685430286, "grad_norm": 1.4671474609434538, "learning_rate": 7.418392536516839e-07, "loss": 0.5341, "step": 24709 }, { "epoch": 0.8803462956695228, "grad_norm": 1.8860880553489234, "learning_rate": 7.414031568399738e-07, "loss": 0.6072, "step": 24710 }, { "epoch": 0.8803819227960169, "grad_norm": 1.8368798531529893, "learning_rate": 7.409671833129539e-07, "loss": 0.4877, "step": 24711 }, { "epoch": 0.880417549922511, "grad_norm": 1.5529089412831363, "learning_rate": 7.405313330764263e-07, "loss": 0.5294, "step": 24712 }, { "epoch": 0.8804531770490052, "grad_norm": 1.6589711925117558, "learning_rate": 7.400956061361975e-07, "loss": 0.4989, "step": 24713 }, { "epoch": 0.8804888041754992, "grad_norm": 1.1835011194181908, "learning_rate": 7.39660002498065e-07, "loss": 0.3779, "step": 24714 }, { "epoch": 0.8805244313019933, "grad_norm": 1.1979508680818887, "learning_rate": 7.392245221678351e-07, "loss": 0.3879, "step": 24715 }, { "epoch": 0.8805600584284875, "grad_norm": 1.4228536309509168, "learning_rate": 7.387891651513024e-07, "loss": 0.5186, "step": 24716 }, { "epoch": 0.8805956855549816, "grad_norm": 1.5121540701003515, "learning_rate": 7.383539314542665e-07, "loss": 0.4265, "step": 24717 }, { "epoch": 0.8806313126814757, "grad_norm": 1.1731149186762446, "learning_rate": 7.379188210825195e-07, "loss": 0.3741, "step": 24718 }, { "epoch": 0.8806669398079698, "grad_norm": 1.3014112052290963, "learning_rate": 7.374838340418588e-07, "loss": 0.4394, "step": 24719 }, { "epoch": 0.8807025669344639, "grad_norm": 3.315167856541825, "learning_rate": 7.370489703380745e-07, "loss": 0.4692, "step": 24720 }, { "epoch": 0.880738194060958, "grad_norm": 1.3773887696857796, "learning_rate": 7.366142299769563e-07, "loss": 0.3695, "step": 24721 }, { "epoch": 0.8807738211874522, "grad_norm": 1.1715663532615026, "learning_rate": 7.361796129642951e-07, "loss": 0.4077, "step": 24722 }, { "epoch": 0.8808094483139463, "grad_norm": 1.5734022558265452, "learning_rate": 7.357451193058773e-07, "loss": 0.5186, "step": 24723 }, { "epoch": 0.8808450754404403, "grad_norm": 1.9648633373857198, "learning_rate": 7.353107490074896e-07, "loss": 0.4847, "step": 24724 }, { "epoch": 0.8808807025669345, "grad_norm": 1.4886893821474305, "learning_rate": 7.348765020749138e-07, "loss": 0.5085, "step": 24725 }, { "epoch": 0.8809163296934286, "grad_norm": 1.2493085452468942, "learning_rate": 7.344423785139343e-07, "loss": 0.4633, "step": 24726 }, { "epoch": 0.8809519568199227, "grad_norm": 1.457783193897299, "learning_rate": 7.340083783303298e-07, "loss": 0.5469, "step": 24727 }, { "epoch": 0.8809875839464169, "grad_norm": 1.240412630762148, "learning_rate": 7.335745015298801e-07, "loss": 0.4458, "step": 24728 }, { "epoch": 0.8810232110729109, "grad_norm": 1.2737986942328348, "learning_rate": 7.331407481183628e-07, "loss": 0.3992, "step": 24729 }, { "epoch": 0.881058838199405, "grad_norm": 1.2627626592150263, "learning_rate": 7.327071181015555e-07, "loss": 0.4181, "step": 24730 }, { "epoch": 0.8810944653258992, "grad_norm": 1.1866353445794655, "learning_rate": 7.322736114852281e-07, "loss": 0.3798, "step": 24731 }, { "epoch": 0.8811300924523933, "grad_norm": 1.2453822645899693, "learning_rate": 7.31840228275158e-07, "loss": 0.3638, "step": 24732 }, { "epoch": 0.8811657195788873, "grad_norm": 1.4308694435941511, "learning_rate": 7.314069684771108e-07, "loss": 0.4113, "step": 24733 }, { "epoch": 0.8812013467053815, "grad_norm": 1.5160138567828434, "learning_rate": 7.309738320968584e-07, "loss": 0.4948, "step": 24734 }, { "epoch": 0.8812369738318756, "grad_norm": 1.6650267862371226, "learning_rate": 7.305408191401697e-07, "loss": 0.5682, "step": 24735 }, { "epoch": 0.8812726009583697, "grad_norm": 1.6540167790800462, "learning_rate": 7.301079296128066e-07, "loss": 0.4866, "step": 24736 }, { "epoch": 0.8813082280848639, "grad_norm": 1.4838948928398923, "learning_rate": 7.296751635205368e-07, "loss": 0.5384, "step": 24737 }, { "epoch": 0.8813438552113579, "grad_norm": 1.507974317227368, "learning_rate": 7.292425208691212e-07, "loss": 0.5242, "step": 24738 }, { "epoch": 0.881379482337852, "grad_norm": 1.4590307815103052, "learning_rate": 7.288100016643229e-07, "loss": 0.5057, "step": 24739 }, { "epoch": 0.8814151094643462, "grad_norm": 1.272566329491339, "learning_rate": 7.283776059118986e-07, "loss": 0.3702, "step": 24740 }, { "epoch": 0.8814507365908403, "grad_norm": 1.5110641115426284, "learning_rate": 7.279453336176079e-07, "loss": 0.4517, "step": 24741 }, { "epoch": 0.8814863637173344, "grad_norm": 1.4506989582332548, "learning_rate": 7.275131847872041e-07, "loss": 0.4842, "step": 24742 }, { "epoch": 0.8815219908438285, "grad_norm": 1.5870929568835417, "learning_rate": 7.270811594264438e-07, "loss": 0.5314, "step": 24743 }, { "epoch": 0.8815576179703226, "grad_norm": 1.1795042472217712, "learning_rate": 7.2664925754108e-07, "loss": 0.3446, "step": 24744 }, { "epoch": 0.8815932450968167, "grad_norm": 1.4456220495633656, "learning_rate": 7.262174791368615e-07, "loss": 0.5191, "step": 24745 }, { "epoch": 0.8816288722233109, "grad_norm": 1.2460911446585432, "learning_rate": 7.257858242195414e-07, "loss": 0.3965, "step": 24746 }, { "epoch": 0.881664499349805, "grad_norm": 1.4504453484562854, "learning_rate": 7.253542927948643e-07, "loss": 0.499, "step": 24747 }, { "epoch": 0.881700126476299, "grad_norm": 1.3170597138519204, "learning_rate": 7.249228848685785e-07, "loss": 0.3906, "step": 24748 }, { "epoch": 0.8817357536027932, "grad_norm": 1.2395046831048226, "learning_rate": 7.244916004464253e-07, "loss": 0.4138, "step": 24749 }, { "epoch": 0.8817713807292873, "grad_norm": 1.6019102108816758, "learning_rate": 7.240604395341521e-07, "loss": 0.4478, "step": 24750 }, { "epoch": 0.8818070078557814, "grad_norm": 1.6420514713012702, "learning_rate": 7.236294021374957e-07, "loss": 0.4689, "step": 24751 }, { "epoch": 0.8818426349822756, "grad_norm": 1.254158661022365, "learning_rate": 7.231984882621978e-07, "loss": 0.4087, "step": 24752 }, { "epoch": 0.8818782621087696, "grad_norm": 1.4162220326753898, "learning_rate": 7.227676979139953e-07, "loss": 0.3947, "step": 24753 }, { "epoch": 0.8819138892352637, "grad_norm": 1.1896938409863185, "learning_rate": 7.223370310986277e-07, "loss": 0.4121, "step": 24754 }, { "epoch": 0.8819495163617579, "grad_norm": 1.7927560793062052, "learning_rate": 7.219064878218263e-07, "loss": 0.4898, "step": 24755 }, { "epoch": 0.881985143488252, "grad_norm": 1.201715690546046, "learning_rate": 7.214760680893241e-07, "loss": 0.3778, "step": 24756 }, { "epoch": 0.882020770614746, "grad_norm": 1.2172418525069473, "learning_rate": 7.210457719068542e-07, "loss": 0.4454, "step": 24757 }, { "epoch": 0.8820563977412402, "grad_norm": 1.2918693899177556, "learning_rate": 7.206155992801455e-07, "loss": 0.3921, "step": 24758 }, { "epoch": 0.8820920248677343, "grad_norm": 1.559458829781246, "learning_rate": 7.201855502149247e-07, "loss": 0.5885, "step": 24759 }, { "epoch": 0.8821276519942284, "grad_norm": 1.1485833374481293, "learning_rate": 7.197556247169202e-07, "loss": 0.3776, "step": 24760 }, { "epoch": 0.8821632791207226, "grad_norm": 1.5164796193129357, "learning_rate": 7.193258227918575e-07, "loss": 0.5105, "step": 24761 }, { "epoch": 0.8821989062472166, "grad_norm": 2.2804238893268893, "learning_rate": 7.188961444454567e-07, "loss": 0.4084, "step": 24762 }, { "epoch": 0.8822345333737107, "grad_norm": 1.3055148659401792, "learning_rate": 7.18466589683442e-07, "loss": 0.3853, "step": 24763 }, { "epoch": 0.8822701605002049, "grad_norm": 1.1785142719558777, "learning_rate": 7.180371585115309e-07, "loss": 0.3571, "step": 24764 }, { "epoch": 0.882305787626699, "grad_norm": 1.2642226766047044, "learning_rate": 7.176078509354422e-07, "loss": 0.3749, "step": 24765 }, { "epoch": 0.882341414753193, "grad_norm": 1.1473791281158539, "learning_rate": 7.171786669608938e-07, "loss": 0.3834, "step": 24766 }, { "epoch": 0.8823770418796872, "grad_norm": 1.3836386217812036, "learning_rate": 7.167496065936008e-07, "loss": 0.4277, "step": 24767 }, { "epoch": 0.8824126690061813, "grad_norm": 1.30189483644349, "learning_rate": 7.163206698392744e-07, "loss": 0.4384, "step": 24768 }, { "epoch": 0.8824482961326754, "grad_norm": 1.2587041173183466, "learning_rate": 7.158918567036266e-07, "loss": 0.3746, "step": 24769 }, { "epoch": 0.8824839232591696, "grad_norm": 1.2435670843539088, "learning_rate": 7.154631671923695e-07, "loss": 0.3914, "step": 24770 }, { "epoch": 0.8825195503856637, "grad_norm": 1.271036422024489, "learning_rate": 7.150346013112086e-07, "loss": 0.3448, "step": 24771 }, { "epoch": 0.8825551775121577, "grad_norm": 1.334097272636724, "learning_rate": 7.146061590658526e-07, "loss": 0.3963, "step": 24772 }, { "epoch": 0.8825908046386519, "grad_norm": 1.257345875856196, "learning_rate": 7.141778404620048e-07, "loss": 0.4127, "step": 24773 }, { "epoch": 0.882626431765146, "grad_norm": 1.6318287404104663, "learning_rate": 7.137496455053694e-07, "loss": 0.4385, "step": 24774 }, { "epoch": 0.8826620588916401, "grad_norm": 1.630082299971281, "learning_rate": 7.133215742016486e-07, "loss": 0.4341, "step": 24775 }, { "epoch": 0.8826976860181343, "grad_norm": 1.476960362268095, "learning_rate": 7.128936265565434e-07, "loss": 0.4823, "step": 24776 }, { "epoch": 0.8827333131446283, "grad_norm": 1.1644072549035474, "learning_rate": 7.124658025757481e-07, "loss": 0.4179, "step": 24777 }, { "epoch": 0.8827689402711224, "grad_norm": 1.7164999150207176, "learning_rate": 7.120381022649658e-07, "loss": 0.5183, "step": 24778 }, { "epoch": 0.8828045673976166, "grad_norm": 1.6308266343236268, "learning_rate": 7.116105256298845e-07, "loss": 0.4815, "step": 24779 }, { "epoch": 0.8828401945241107, "grad_norm": 1.2598734346445397, "learning_rate": 7.111830726762026e-07, "loss": 0.4074, "step": 24780 }, { "epoch": 0.8828758216506047, "grad_norm": 1.4223402148838244, "learning_rate": 7.107557434096102e-07, "loss": 0.4494, "step": 24781 }, { "epoch": 0.8829114487770989, "grad_norm": 1.2109329931311519, "learning_rate": 7.103285378357983e-07, "loss": 0.3882, "step": 24782 }, { "epoch": 0.882947075903593, "grad_norm": 1.845285832374053, "learning_rate": 7.099014559604556e-07, "loss": 0.5576, "step": 24783 }, { "epoch": 0.8829827030300871, "grad_norm": 1.3415174394444875, "learning_rate": 7.094744977892676e-07, "loss": 0.3813, "step": 24784 }, { "epoch": 0.8830183301565813, "grad_norm": 1.5733322140579293, "learning_rate": 7.090476633279209e-07, "loss": 0.4991, "step": 24785 }, { "epoch": 0.8830539572830753, "grad_norm": 2.0276178720246056, "learning_rate": 7.086209525820986e-07, "loss": 0.512, "step": 24786 }, { "epoch": 0.8830895844095694, "grad_norm": 1.2382499923316672, "learning_rate": 7.08194365557483e-07, "loss": 0.4258, "step": 24787 }, { "epoch": 0.8831252115360636, "grad_norm": 1.3244524388196064, "learning_rate": 7.077679022597528e-07, "loss": 0.4106, "step": 24788 }, { "epoch": 0.8831608386625577, "grad_norm": 1.6468079805180706, "learning_rate": 7.073415626945879e-07, "loss": 0.4448, "step": 24789 }, { "epoch": 0.8831964657890518, "grad_norm": 1.287959181477156, "learning_rate": 7.069153468676648e-07, "loss": 0.4117, "step": 24790 }, { "epoch": 0.8832320929155459, "grad_norm": 1.6086770491973024, "learning_rate": 7.064892547846602e-07, "loss": 0.5893, "step": 24791 }, { "epoch": 0.88326772004204, "grad_norm": 1.585530962892179, "learning_rate": 7.060632864512473e-07, "loss": 0.4498, "step": 24792 }, { "epoch": 0.8833033471685341, "grad_norm": 1.602593896293357, "learning_rate": 7.056374418730971e-07, "loss": 0.5734, "step": 24793 }, { "epoch": 0.8833389742950283, "grad_norm": 1.2753924057816353, "learning_rate": 7.052117210558829e-07, "loss": 0.4156, "step": 24794 }, { "epoch": 0.8833746014215224, "grad_norm": 1.2238349835767222, "learning_rate": 7.047861240052689e-07, "loss": 0.3894, "step": 24795 }, { "epoch": 0.8834102285480164, "grad_norm": 1.5380484081826042, "learning_rate": 7.043606507269241e-07, "loss": 0.4599, "step": 24796 }, { "epoch": 0.8834458556745106, "grad_norm": 1.3029849792162869, "learning_rate": 7.03935301226516e-07, "loss": 0.4314, "step": 24797 }, { "epoch": 0.8834814828010047, "grad_norm": 1.4373464738832642, "learning_rate": 7.035100755097069e-07, "loss": 0.5144, "step": 24798 }, { "epoch": 0.8835171099274988, "grad_norm": 1.1872388458916023, "learning_rate": 7.030849735821588e-07, "loss": 0.3912, "step": 24799 }, { "epoch": 0.883552737053993, "grad_norm": 1.234307164211148, "learning_rate": 7.026599954495328e-07, "loss": 0.3936, "step": 24800 }, { "epoch": 0.883588364180487, "grad_norm": 1.2200873232712002, "learning_rate": 7.022351411174866e-07, "loss": 0.4049, "step": 24801 }, { "epoch": 0.8836239913069811, "grad_norm": 1.5238785895194071, "learning_rate": 7.018104105916779e-07, "loss": 0.6236, "step": 24802 }, { "epoch": 0.8836596184334753, "grad_norm": 1.3383884731603712, "learning_rate": 7.013858038777632e-07, "loss": 0.4311, "step": 24803 }, { "epoch": 0.8836952455599694, "grad_norm": 1.18438791038108, "learning_rate": 7.009613209813948e-07, "loss": 0.3759, "step": 24804 }, { "epoch": 0.8837308726864634, "grad_norm": 1.5423595997581085, "learning_rate": 7.005369619082292e-07, "loss": 0.5508, "step": 24805 }, { "epoch": 0.8837664998129576, "grad_norm": 1.523377287191363, "learning_rate": 7.001127266639108e-07, "loss": 0.4033, "step": 24806 }, { "epoch": 0.8838021269394517, "grad_norm": 1.405493572459386, "learning_rate": 6.996886152540938e-07, "loss": 0.471, "step": 24807 }, { "epoch": 0.8838377540659458, "grad_norm": 1.5394930692475122, "learning_rate": 6.992646276844217e-07, "loss": 0.5194, "step": 24808 }, { "epoch": 0.88387338119244, "grad_norm": 1.5239287475306764, "learning_rate": 6.988407639605443e-07, "loss": 0.5667, "step": 24809 }, { "epoch": 0.883909008318934, "grad_norm": 1.1568385160583696, "learning_rate": 6.984170240881005e-07, "loss": 0.3894, "step": 24810 }, { "epoch": 0.8839446354454281, "grad_norm": 1.6825577709596051, "learning_rate": 6.979934080727369e-07, "loss": 0.5044, "step": 24811 }, { "epoch": 0.8839802625719223, "grad_norm": 1.8012092760395455, "learning_rate": 6.97569915920091e-07, "loss": 0.4504, "step": 24812 }, { "epoch": 0.8840158896984164, "grad_norm": 1.4120036639369957, "learning_rate": 6.971465476358064e-07, "loss": 0.4892, "step": 24813 }, { "epoch": 0.8840515168249105, "grad_norm": 1.2042532548797205, "learning_rate": 6.96723303225516e-07, "loss": 0.3995, "step": 24814 }, { "epoch": 0.8840871439514046, "grad_norm": 1.1170701702617913, "learning_rate": 6.963001826948579e-07, "loss": 0.3599, "step": 24815 }, { "epoch": 0.8841227710778987, "grad_norm": 1.589925302073224, "learning_rate": 6.958771860494662e-07, "loss": 0.4936, "step": 24816 }, { "epoch": 0.8841583982043928, "grad_norm": 1.6661010794017095, "learning_rate": 6.95454313294972e-07, "loss": 0.5045, "step": 24817 }, { "epoch": 0.884194025330887, "grad_norm": 1.44384475154817, "learning_rate": 6.950315644370075e-07, "loss": 0.4519, "step": 24818 }, { "epoch": 0.884229652457381, "grad_norm": 1.411334936665019, "learning_rate": 6.946089394812006e-07, "loss": 0.5298, "step": 24819 }, { "epoch": 0.8842652795838751, "grad_norm": 1.1557633221503405, "learning_rate": 6.94186438433182e-07, "loss": 0.3617, "step": 24820 }, { "epoch": 0.8843009067103693, "grad_norm": 1.2245737111433175, "learning_rate": 6.937640612985741e-07, "loss": 0.3825, "step": 24821 }, { "epoch": 0.8843365338368634, "grad_norm": 1.3633936968632678, "learning_rate": 6.933418080830034e-07, "loss": 0.4645, "step": 24822 }, { "epoch": 0.8843721609633575, "grad_norm": 1.239689234326501, "learning_rate": 6.9291967879209e-07, "loss": 0.3707, "step": 24823 }, { "epoch": 0.8844077880898517, "grad_norm": 1.7182324761511691, "learning_rate": 6.924976734314581e-07, "loss": 0.537, "step": 24824 }, { "epoch": 0.8844434152163457, "grad_norm": 1.6297249550620936, "learning_rate": 6.920757920067234e-07, "loss": 0.5943, "step": 24825 }, { "epoch": 0.8844790423428398, "grad_norm": 1.6463811924364655, "learning_rate": 6.916540345235056e-07, "loss": 0.4814, "step": 24826 }, { "epoch": 0.884514669469334, "grad_norm": 1.2121292076522585, "learning_rate": 6.912324009874205e-07, "loss": 0.3995, "step": 24827 }, { "epoch": 0.8845502965958281, "grad_norm": 1.1911611555362551, "learning_rate": 6.908108914040823e-07, "loss": 0.3889, "step": 24828 }, { "epoch": 0.8845859237223221, "grad_norm": 1.5692829478324921, "learning_rate": 6.903895057791043e-07, "loss": 0.5324, "step": 24829 }, { "epoch": 0.8846215508488163, "grad_norm": 1.2060385269865488, "learning_rate": 6.899682441180966e-07, "loss": 0.4599, "step": 24830 }, { "epoch": 0.8846571779753104, "grad_norm": 1.513389970118905, "learning_rate": 6.895471064266713e-07, "loss": 0.4981, "step": 24831 }, { "epoch": 0.8846928051018045, "grad_norm": 1.7762058188959753, "learning_rate": 6.891260927104315e-07, "loss": 0.5205, "step": 24832 }, { "epoch": 0.8847284322282987, "grad_norm": 1.235481066348753, "learning_rate": 6.887052029749852e-07, "loss": 0.3902, "step": 24833 }, { "epoch": 0.8847640593547927, "grad_norm": 1.283474211637876, "learning_rate": 6.882844372259378e-07, "loss": 0.4082, "step": 24834 }, { "epoch": 0.8847996864812868, "grad_norm": 1.4920138695742804, "learning_rate": 6.878637954688938e-07, "loss": 0.5289, "step": 24835 }, { "epoch": 0.884835313607781, "grad_norm": 1.0487686041815056, "learning_rate": 6.874432777094498e-07, "loss": 0.3843, "step": 24836 }, { "epoch": 0.8848709407342751, "grad_norm": 1.4140094837001382, "learning_rate": 6.870228839532078e-07, "loss": 0.4047, "step": 24837 }, { "epoch": 0.8849065678607692, "grad_norm": 1.2527149538855011, "learning_rate": 6.86602614205768e-07, "loss": 0.4009, "step": 24838 }, { "epoch": 0.8849421949872633, "grad_norm": 1.446814661059691, "learning_rate": 6.861824684727214e-07, "loss": 0.3707, "step": 24839 }, { "epoch": 0.8849778221137574, "grad_norm": 1.5095211450225519, "learning_rate": 6.857624467596669e-07, "loss": 0.5035, "step": 24840 }, { "epoch": 0.8850134492402515, "grad_norm": 2.0200660198734965, "learning_rate": 6.853425490721921e-07, "loss": 0.5507, "step": 24841 }, { "epoch": 0.8850490763667457, "grad_norm": 1.3077989049434842, "learning_rate": 6.849227754158961e-07, "loss": 0.3834, "step": 24842 }, { "epoch": 0.8850847034932398, "grad_norm": 1.4942677937557858, "learning_rate": 6.845031257963619e-07, "loss": 0.5652, "step": 24843 }, { "epoch": 0.8851203306197338, "grad_norm": 1.324371373140062, "learning_rate": 6.84083600219182e-07, "loss": 0.3849, "step": 24844 }, { "epoch": 0.885155957746228, "grad_norm": 1.1974285561219424, "learning_rate": 6.836641986899384e-07, "loss": 0.3986, "step": 24845 }, { "epoch": 0.8851915848727221, "grad_norm": 1.225610991901041, "learning_rate": 6.83244921214219e-07, "loss": 0.4262, "step": 24846 }, { "epoch": 0.8852272119992162, "grad_norm": 1.5075895647571056, "learning_rate": 6.828257677976035e-07, "loss": 0.4703, "step": 24847 }, { "epoch": 0.8852628391257104, "grad_norm": 1.4910617032638143, "learning_rate": 6.824067384456756e-07, "loss": 0.5102, "step": 24848 }, { "epoch": 0.8852984662522044, "grad_norm": 1.388311025929222, "learning_rate": 6.81987833164015e-07, "loss": 0.4655, "step": 24849 }, { "epoch": 0.8853340933786985, "grad_norm": 1.2373847278929355, "learning_rate": 6.815690519581997e-07, "loss": 0.3812, "step": 24850 }, { "epoch": 0.8853697205051927, "grad_norm": 1.4520374824545126, "learning_rate": 6.811503948338061e-07, "loss": 0.3937, "step": 24851 }, { "epoch": 0.8854053476316868, "grad_norm": 1.1155615683302085, "learning_rate": 6.807318617964077e-07, "loss": 0.4038, "step": 24852 }, { "epoch": 0.8854409747581808, "grad_norm": 1.3934995789486986, "learning_rate": 6.803134528515798e-07, "loss": 0.4686, "step": 24853 }, { "epoch": 0.885476601884675, "grad_norm": 1.2615438692547114, "learning_rate": 6.798951680048904e-07, "loss": 0.405, "step": 24854 }, { "epoch": 0.8855122290111691, "grad_norm": 1.640728995740897, "learning_rate": 6.794770072619116e-07, "loss": 0.4725, "step": 24855 }, { "epoch": 0.8855478561376632, "grad_norm": 1.103310733617747, "learning_rate": 6.790589706282113e-07, "loss": 0.3644, "step": 24856 }, { "epoch": 0.8855834832641574, "grad_norm": 1.2457659041624443, "learning_rate": 6.786410581093572e-07, "loss": 0.3706, "step": 24857 }, { "epoch": 0.8856191103906514, "grad_norm": 1.3099072138439647, "learning_rate": 6.782232697109104e-07, "loss": 0.4082, "step": 24858 }, { "epoch": 0.8856547375171455, "grad_norm": 1.2779524118898846, "learning_rate": 6.778056054384385e-07, "loss": 0.4091, "step": 24859 }, { "epoch": 0.8856903646436397, "grad_norm": 1.3304601800920852, "learning_rate": 6.773880652974985e-07, "loss": 0.4135, "step": 24860 }, { "epoch": 0.8857259917701338, "grad_norm": 1.518043772746907, "learning_rate": 6.769706492936534e-07, "loss": 0.5341, "step": 24861 }, { "epoch": 0.8857616188966279, "grad_norm": 1.6547237799824062, "learning_rate": 6.765533574324623e-07, "loss": 0.4142, "step": 24862 }, { "epoch": 0.885797246023122, "grad_norm": 1.3462496492153275, "learning_rate": 6.761361897194774e-07, "loss": 0.4212, "step": 24863 }, { "epoch": 0.8858328731496161, "grad_norm": 1.1741693299567473, "learning_rate": 6.757191461602563e-07, "loss": 0.3984, "step": 24864 }, { "epoch": 0.8858685002761102, "grad_norm": 1.59157513532259, "learning_rate": 6.753022267603526e-07, "loss": 0.5731, "step": 24865 }, { "epoch": 0.8859041274026044, "grad_norm": 1.3680438971539526, "learning_rate": 6.748854315253184e-07, "loss": 0.3921, "step": 24866 }, { "epoch": 0.8859397545290985, "grad_norm": 1.370403374985758, "learning_rate": 6.744687604607015e-07, "loss": 0.4038, "step": 24867 }, { "epoch": 0.8859753816555925, "grad_norm": 1.2246600071327711, "learning_rate": 6.740522135720518e-07, "loss": 0.4153, "step": 24868 }, { "epoch": 0.8860110087820867, "grad_norm": 1.2792625974881477, "learning_rate": 6.736357908649149e-07, "loss": 0.4134, "step": 24869 }, { "epoch": 0.8860466359085808, "grad_norm": 1.325810544554596, "learning_rate": 6.732194923448354e-07, "loss": 0.4128, "step": 24870 }, { "epoch": 0.8860822630350749, "grad_norm": 1.3230877845189506, "learning_rate": 6.728033180173576e-07, "loss": 0.4054, "step": 24871 }, { "epoch": 0.886117890161569, "grad_norm": 1.1813337256300074, "learning_rate": 6.723872678880228e-07, "loss": 0.3885, "step": 24872 }, { "epoch": 0.8861535172880631, "grad_norm": 1.1642621614367457, "learning_rate": 6.71971341962373e-07, "loss": 0.3579, "step": 24873 }, { "epoch": 0.8861891444145572, "grad_norm": 1.4705831393585271, "learning_rate": 6.715555402459439e-07, "loss": 0.4817, "step": 24874 }, { "epoch": 0.8862247715410514, "grad_norm": 1.578704508235179, "learning_rate": 6.711398627442733e-07, "loss": 0.5309, "step": 24875 }, { "epoch": 0.8862603986675455, "grad_norm": 1.3913760264754735, "learning_rate": 6.707243094628946e-07, "loss": 0.4641, "step": 24876 }, { "epoch": 0.8862960257940395, "grad_norm": 1.4630998923413798, "learning_rate": 6.703088804073443e-07, "loss": 0.3832, "step": 24877 }, { "epoch": 0.8863316529205337, "grad_norm": 1.104572768314447, "learning_rate": 6.698935755831493e-07, "loss": 0.4001, "step": 24878 }, { "epoch": 0.8863672800470278, "grad_norm": 1.2331115904094305, "learning_rate": 6.69478394995845e-07, "loss": 0.4476, "step": 24879 }, { "epoch": 0.8864029071735219, "grad_norm": 1.2224844503763002, "learning_rate": 6.690633386509571e-07, "loss": 0.3532, "step": 24880 }, { "epoch": 0.8864385343000161, "grad_norm": 1.4093561643257428, "learning_rate": 6.686484065540133e-07, "loss": 0.3536, "step": 24881 }, { "epoch": 0.8864741614265101, "grad_norm": 1.2783377276247656, "learning_rate": 6.682335987105371e-07, "loss": 0.3721, "step": 24882 }, { "epoch": 0.8865097885530042, "grad_norm": 1.4952658140888655, "learning_rate": 6.67818915126054e-07, "loss": 0.4988, "step": 24883 }, { "epoch": 0.8865454156794984, "grad_norm": 1.6569838124252743, "learning_rate": 6.67404355806085e-07, "loss": 0.4795, "step": 24884 }, { "epoch": 0.8865810428059925, "grad_norm": 1.4197256278054151, "learning_rate": 6.669899207561492e-07, "loss": 0.4566, "step": 24885 }, { "epoch": 0.8866166699324866, "grad_norm": 1.407845398165534, "learning_rate": 6.665756099817655e-07, "loss": 0.4218, "step": 24886 }, { "epoch": 0.8866522970589807, "grad_norm": 1.3005777644145327, "learning_rate": 6.661614234884517e-07, "loss": 0.5066, "step": 24887 }, { "epoch": 0.8866879241854748, "grad_norm": 1.284339210175696, "learning_rate": 6.657473612817244e-07, "loss": 0.4032, "step": 24888 }, { "epoch": 0.8867235513119689, "grad_norm": 1.9423330935063767, "learning_rate": 6.653334233670928e-07, "loss": 0.4961, "step": 24889 }, { "epoch": 0.8867591784384631, "grad_norm": 1.2486014400522272, "learning_rate": 6.649196097500732e-07, "loss": 0.3892, "step": 24890 }, { "epoch": 0.8867948055649572, "grad_norm": 1.5638056496235115, "learning_rate": 6.645059204361726e-07, "loss": 0.4765, "step": 24891 }, { "epoch": 0.8868304326914512, "grad_norm": 1.1668071000888725, "learning_rate": 6.640923554308998e-07, "loss": 0.3804, "step": 24892 }, { "epoch": 0.8868660598179454, "grad_norm": 1.5980588073189435, "learning_rate": 6.636789147397638e-07, "loss": 0.544, "step": 24893 }, { "epoch": 0.8869016869444395, "grad_norm": 1.5037792185514403, "learning_rate": 6.63265598368269e-07, "loss": 0.5488, "step": 24894 }, { "epoch": 0.8869373140709336, "grad_norm": 2.0776359543503324, "learning_rate": 6.628524063219189e-07, "loss": 0.5341, "step": 24895 }, { "epoch": 0.8869729411974278, "grad_norm": 1.2247051623831544, "learning_rate": 6.624393386062145e-07, "loss": 0.353, "step": 24896 }, { "epoch": 0.8870085683239218, "grad_norm": 1.4854960571461706, "learning_rate": 6.620263952266581e-07, "loss": 0.3826, "step": 24897 }, { "epoch": 0.8870441954504159, "grad_norm": 1.3621652887819657, "learning_rate": 6.616135761887455e-07, "loss": 0.3986, "step": 24898 }, { "epoch": 0.8870798225769101, "grad_norm": 1.3330545970675203, "learning_rate": 6.612008814979776e-07, "loss": 0.3841, "step": 24899 }, { "epoch": 0.8871154497034042, "grad_norm": 1.342578089493762, "learning_rate": 6.607883111598446e-07, "loss": 0.3855, "step": 24900 }, { "epoch": 0.8871510768298982, "grad_norm": 1.4697652424050942, "learning_rate": 6.603758651798442e-07, "loss": 0.5265, "step": 24901 }, { "epoch": 0.8871867039563924, "grad_norm": 1.1866194367575273, "learning_rate": 6.599635435634666e-07, "loss": 0.41, "step": 24902 }, { "epoch": 0.8872223310828865, "grad_norm": 1.2900572608193241, "learning_rate": 6.59551346316204e-07, "loss": 0.3814, "step": 24903 }, { "epoch": 0.8872579582093806, "grad_norm": 1.2206893578839049, "learning_rate": 6.591392734435431e-07, "loss": 0.3718, "step": 24904 }, { "epoch": 0.8872935853358748, "grad_norm": 1.2766905746263657, "learning_rate": 6.587273249509718e-07, "loss": 0.4477, "step": 24905 }, { "epoch": 0.8873292124623688, "grad_norm": 1.248250863212601, "learning_rate": 6.583155008439745e-07, "loss": 0.4026, "step": 24906 }, { "epoch": 0.8873648395888629, "grad_norm": 1.6471732043319447, "learning_rate": 6.57903801128037e-07, "loss": 0.4977, "step": 24907 }, { "epoch": 0.8874004667153571, "grad_norm": 1.3342193521565855, "learning_rate": 6.57492225808638e-07, "loss": 0.3987, "step": 24908 }, { "epoch": 0.8874360938418512, "grad_norm": 3.063106108363112, "learning_rate": 6.570807748912611e-07, "loss": 0.4361, "step": 24909 }, { "epoch": 0.8874717209683453, "grad_norm": 1.2244042356261584, "learning_rate": 6.566694483813862e-07, "loss": 0.3784, "step": 24910 }, { "epoch": 0.8875073480948394, "grad_norm": 1.4482467948862066, "learning_rate": 6.562582462844858e-07, "loss": 0.4651, "step": 24911 }, { "epoch": 0.8875429752213335, "grad_norm": 1.4784178536876764, "learning_rate": 6.558471686060397e-07, "loss": 0.4818, "step": 24912 }, { "epoch": 0.8875786023478276, "grad_norm": 2.3613470416719613, "learning_rate": 6.554362153515171e-07, "loss": 0.4718, "step": 24913 }, { "epoch": 0.8876142294743218, "grad_norm": 1.3716174410569075, "learning_rate": 6.550253865263956e-07, "loss": 0.3735, "step": 24914 }, { "epoch": 0.8876498566008159, "grad_norm": 1.4092200689653998, "learning_rate": 6.54614682136141e-07, "loss": 0.3774, "step": 24915 }, { "epoch": 0.8876854837273099, "grad_norm": 1.9165561109978242, "learning_rate": 6.542041021862233e-07, "loss": 0.5438, "step": 24916 }, { "epoch": 0.8877211108538041, "grad_norm": 1.3764380349212133, "learning_rate": 6.537936466821105e-07, "loss": 0.3852, "step": 24917 }, { "epoch": 0.8877567379802982, "grad_norm": 1.2973548442540788, "learning_rate": 6.53383315629268e-07, "loss": 0.3958, "step": 24918 }, { "epoch": 0.8877923651067923, "grad_norm": 1.4055084993234046, "learning_rate": 6.529731090331615e-07, "loss": 0.4222, "step": 24919 }, { "epoch": 0.8878279922332865, "grad_norm": 1.8283400749406207, "learning_rate": 6.5256302689925e-07, "loss": 0.502, "step": 24920 }, { "epoch": 0.8878636193597805, "grad_norm": 1.761273921484333, "learning_rate": 6.521530692329958e-07, "loss": 0.5606, "step": 24921 }, { "epoch": 0.8878992464862746, "grad_norm": 2.460667971905289, "learning_rate": 6.517432360398556e-07, "loss": 0.4062, "step": 24922 }, { "epoch": 0.8879348736127688, "grad_norm": 1.550347900898012, "learning_rate": 6.513335273252896e-07, "loss": 0.5488, "step": 24923 }, { "epoch": 0.8879705007392629, "grad_norm": 1.6781774430528151, "learning_rate": 6.509239430947511e-07, "loss": 0.4801, "step": 24924 }, { "epoch": 0.888006127865757, "grad_norm": 1.6389587352802628, "learning_rate": 6.505144833536958e-07, "loss": 0.4718, "step": 24925 }, { "epoch": 0.8880417549922511, "grad_norm": 1.2662868904194307, "learning_rate": 6.501051481075748e-07, "loss": 0.3741, "step": 24926 }, { "epoch": 0.8880773821187452, "grad_norm": 1.476696786567984, "learning_rate": 6.496959373618394e-07, "loss": 0.392, "step": 24927 }, { "epoch": 0.8881130092452393, "grad_norm": 1.0983562374639808, "learning_rate": 6.492868511219374e-07, "loss": 0.3457, "step": 24928 }, { "epoch": 0.8881486363717335, "grad_norm": 1.532951664391248, "learning_rate": 6.488778893933168e-07, "loss": 0.5067, "step": 24929 }, { "epoch": 0.8881842634982275, "grad_norm": 1.3824802989475897, "learning_rate": 6.484690521814218e-07, "loss": 0.4444, "step": 24930 }, { "epoch": 0.8882198906247216, "grad_norm": 1.257428316311544, "learning_rate": 6.480603394916996e-07, "loss": 0.3906, "step": 24931 }, { "epoch": 0.8882555177512158, "grad_norm": 1.2834988168885593, "learning_rate": 6.476517513295911e-07, "loss": 0.3701, "step": 24932 }, { "epoch": 0.8882911448777099, "grad_norm": 1.3283544089150572, "learning_rate": 6.472432877005341e-07, "loss": 0.4281, "step": 24933 }, { "epoch": 0.888326772004204, "grad_norm": 1.2016892919413147, "learning_rate": 6.468349486099723e-07, "loss": 0.3975, "step": 24934 }, { "epoch": 0.8883623991306981, "grad_norm": 1.2603352114091109, "learning_rate": 6.464267340633401e-07, "loss": 0.3874, "step": 24935 }, { "epoch": 0.8883980262571922, "grad_norm": 1.986595375373375, "learning_rate": 6.460186440660743e-07, "loss": 0.5327, "step": 24936 }, { "epoch": 0.8884336533836863, "grad_norm": 1.729297372903426, "learning_rate": 6.456106786236061e-07, "loss": 0.5312, "step": 24937 }, { "epoch": 0.8884692805101805, "grad_norm": 1.59931967103997, "learning_rate": 6.452028377413711e-07, "loss": 0.5397, "step": 24938 }, { "epoch": 0.8885049076366746, "grad_norm": 1.450603050358215, "learning_rate": 6.447951214247983e-07, "loss": 0.4274, "step": 24939 }, { "epoch": 0.8885405347631686, "grad_norm": 2.099890543714522, "learning_rate": 6.443875296793189e-07, "loss": 0.562, "step": 24940 }, { "epoch": 0.8885761618896628, "grad_norm": 1.3465837228193271, "learning_rate": 6.439800625103576e-07, "loss": 0.4218, "step": 24941 }, { "epoch": 0.8886117890161569, "grad_norm": 1.2115521581585533, "learning_rate": 6.43572719923341e-07, "loss": 0.4313, "step": 24942 }, { "epoch": 0.888647416142651, "grad_norm": 1.6226517796585809, "learning_rate": 6.431655019236949e-07, "loss": 0.5635, "step": 24943 }, { "epoch": 0.8886830432691452, "grad_norm": 1.6386450179981566, "learning_rate": 6.427584085168381e-07, "loss": 0.5309, "step": 24944 }, { "epoch": 0.8887186703956392, "grad_norm": 1.1838693281678971, "learning_rate": 6.423514397081942e-07, "loss": 0.4292, "step": 24945 }, { "epoch": 0.8887542975221333, "grad_norm": 1.2808734361798064, "learning_rate": 6.419445955031811e-07, "loss": 0.3865, "step": 24946 }, { "epoch": 0.8887899246486275, "grad_norm": 1.6656538729845067, "learning_rate": 6.415378759072177e-07, "loss": 0.544, "step": 24947 }, { "epoch": 0.8888255517751216, "grad_norm": 1.1127096371031584, "learning_rate": 6.411312809257187e-07, "loss": 0.3431, "step": 24948 }, { "epoch": 0.8888611789016156, "grad_norm": 1.2432689688671106, "learning_rate": 6.407248105640984e-07, "loss": 0.4095, "step": 24949 }, { "epoch": 0.8888968060281098, "grad_norm": 1.429258852555141, "learning_rate": 6.403184648277672e-07, "loss": 0.4387, "step": 24950 }, { "epoch": 0.8889324331546039, "grad_norm": 1.721420172451695, "learning_rate": 6.399122437221405e-07, "loss": 0.3628, "step": 24951 }, { "epoch": 0.888968060281098, "grad_norm": 1.241110123467424, "learning_rate": 6.39506147252622e-07, "loss": 0.3657, "step": 24952 }, { "epoch": 0.8890036874075922, "grad_norm": 1.6086981642609783, "learning_rate": 6.391001754246206e-07, "loss": 0.4343, "step": 24953 }, { "epoch": 0.8890393145340862, "grad_norm": 1.481068232517546, "learning_rate": 6.386943282435454e-07, "loss": 0.4138, "step": 24954 }, { "epoch": 0.8890749416605803, "grad_norm": 1.4321291391877815, "learning_rate": 6.382886057147986e-07, "loss": 0.4108, "step": 24955 }, { "epoch": 0.8891105687870745, "grad_norm": 1.207342382158443, "learning_rate": 6.378830078437825e-07, "loss": 0.4074, "step": 24956 }, { "epoch": 0.8891461959135686, "grad_norm": 1.184759176311465, "learning_rate": 6.374775346358975e-07, "loss": 0.4212, "step": 24957 }, { "epoch": 0.8891818230400627, "grad_norm": 1.4286045665937253, "learning_rate": 6.370721860965435e-07, "loss": 0.5053, "step": 24958 }, { "epoch": 0.8892174501665568, "grad_norm": 2.1298777867614085, "learning_rate": 6.366669622311161e-07, "loss": 0.4102, "step": 24959 }, { "epoch": 0.8892530772930509, "grad_norm": 1.231272438481515, "learning_rate": 6.362618630450134e-07, "loss": 0.4381, "step": 24960 }, { "epoch": 0.889288704419545, "grad_norm": 1.4435071471610623, "learning_rate": 6.358568885436289e-07, "loss": 0.397, "step": 24961 }, { "epoch": 0.8893243315460392, "grad_norm": 1.4278086573904027, "learning_rate": 6.35452038732357e-07, "loss": 0.4, "step": 24962 }, { "epoch": 0.8893599586725333, "grad_norm": 1.5985966928131514, "learning_rate": 6.350473136165836e-07, "loss": 0.5424, "step": 24963 }, { "epoch": 0.8893955857990273, "grad_norm": 1.3187745055582, "learning_rate": 6.346427132017019e-07, "loss": 0.4362, "step": 24964 }, { "epoch": 0.8894312129255215, "grad_norm": 1.1825586353298285, "learning_rate": 6.342382374931e-07, "loss": 0.3949, "step": 24965 }, { "epoch": 0.8894668400520156, "grad_norm": 1.5473195467111909, "learning_rate": 6.338338864961613e-07, "loss": 0.5618, "step": 24966 }, { "epoch": 0.8895024671785097, "grad_norm": 1.180838802670109, "learning_rate": 6.334296602162703e-07, "loss": 0.4167, "step": 24967 }, { "epoch": 0.8895380943050039, "grad_norm": 1.324381336304884, "learning_rate": 6.330255586588096e-07, "loss": 0.4234, "step": 24968 }, { "epoch": 0.8895737214314979, "grad_norm": 1.2925161260222353, "learning_rate": 6.326215818291636e-07, "loss": 0.4003, "step": 24969 }, { "epoch": 0.889609348557992, "grad_norm": 1.6166448355583904, "learning_rate": 6.322177297327059e-07, "loss": 0.5033, "step": 24970 }, { "epoch": 0.8896449756844862, "grad_norm": 1.3062592794746022, "learning_rate": 6.318140023748188e-07, "loss": 0.4031, "step": 24971 }, { "epoch": 0.8896806028109803, "grad_norm": 1.333128067419662, "learning_rate": 6.314103997608745e-07, "loss": 0.4088, "step": 24972 }, { "epoch": 0.8897162299374743, "grad_norm": 1.297495257260138, "learning_rate": 6.310069218962512e-07, "loss": 0.3811, "step": 24973 }, { "epoch": 0.8897518570639685, "grad_norm": 1.1866493348230622, "learning_rate": 6.306035687863165e-07, "loss": 0.3875, "step": 24974 }, { "epoch": 0.8897874841904626, "grad_norm": 1.2448237632178918, "learning_rate": 6.302003404364454e-07, "loss": 0.4125, "step": 24975 }, { "epoch": 0.8898231113169567, "grad_norm": 1.4024286415089253, "learning_rate": 6.297972368520056e-07, "loss": 0.4299, "step": 24976 }, { "epoch": 0.8898587384434509, "grad_norm": 1.3809201257304295, "learning_rate": 6.29394258038365e-07, "loss": 0.3863, "step": 24977 }, { "epoch": 0.889894365569945, "grad_norm": 1.2039185707638582, "learning_rate": 6.289914040008904e-07, "loss": 0.412, "step": 24978 }, { "epoch": 0.889929992696439, "grad_norm": 1.7314099335656918, "learning_rate": 6.285886747449443e-07, "loss": 0.4572, "step": 24979 }, { "epoch": 0.8899656198229332, "grad_norm": 1.306952829960407, "learning_rate": 6.281860702758913e-07, "loss": 0.4138, "step": 24980 }, { "epoch": 0.8900012469494273, "grad_norm": 1.3126946537865385, "learning_rate": 6.277835905990892e-07, "loss": 0.3566, "step": 24981 }, { "epoch": 0.8900368740759214, "grad_norm": 1.2401824757241737, "learning_rate": 6.273812357199005e-07, "loss": 0.3707, "step": 24982 }, { "epoch": 0.8900725012024155, "grad_norm": 1.669062400308954, "learning_rate": 6.269790056436808e-07, "loss": 0.5558, "step": 24983 }, { "epoch": 0.8901081283289096, "grad_norm": 1.212015230201098, "learning_rate": 6.265769003757894e-07, "loss": 0.3907, "step": 24984 }, { "epoch": 0.8901437554554037, "grad_norm": 1.6987925307309697, "learning_rate": 6.261749199215761e-07, "loss": 0.428, "step": 24985 }, { "epoch": 0.8901793825818979, "grad_norm": 1.6083584439984115, "learning_rate": 6.25773064286398e-07, "loss": 0.5929, "step": 24986 }, { "epoch": 0.890215009708392, "grad_norm": 1.6141013089777922, "learning_rate": 6.253713334756018e-07, "loss": 0.5547, "step": 24987 }, { "epoch": 0.890250636834886, "grad_norm": 1.2240629612733058, "learning_rate": 6.249697274945377e-07, "loss": 0.4245, "step": 24988 }, { "epoch": 0.8902862639613802, "grad_norm": 2.0085345873295886, "learning_rate": 6.245682463485569e-07, "loss": 0.5413, "step": 24989 }, { "epoch": 0.8903218910878743, "grad_norm": 1.3849564571667954, "learning_rate": 6.24166890043002e-07, "loss": 0.4119, "step": 24990 }, { "epoch": 0.8903575182143684, "grad_norm": 1.4801809250357472, "learning_rate": 6.237656585832174e-07, "loss": 0.5219, "step": 24991 }, { "epoch": 0.8903931453408626, "grad_norm": 1.797687427526518, "learning_rate": 6.233645519745469e-07, "loss": 0.5374, "step": 24992 }, { "epoch": 0.8904287724673566, "grad_norm": 1.222667380996931, "learning_rate": 6.229635702223325e-07, "loss": 0.3912, "step": 24993 }, { "epoch": 0.8904643995938507, "grad_norm": 1.6154856075566628, "learning_rate": 6.225627133319112e-07, "loss": 0.5311, "step": 24994 }, { "epoch": 0.8905000267203449, "grad_norm": 1.285792297582091, "learning_rate": 6.221619813086232e-07, "loss": 0.3768, "step": 24995 }, { "epoch": 0.890535653846839, "grad_norm": 1.378760150397774, "learning_rate": 6.217613741578011e-07, "loss": 0.5034, "step": 24996 }, { "epoch": 0.890571280973333, "grad_norm": 1.503969551143636, "learning_rate": 6.213608918847825e-07, "loss": 0.4969, "step": 24997 }, { "epoch": 0.8906069080998272, "grad_norm": 1.6580852437241973, "learning_rate": 6.209605344948977e-07, "loss": 0.5022, "step": 24998 }, { "epoch": 0.8906425352263213, "grad_norm": 1.2390333712264698, "learning_rate": 6.205603019934791e-07, "loss": 0.4206, "step": 24999 }, { "epoch": 0.8906781623528154, "grad_norm": 1.535386686831499, "learning_rate": 6.201601943858582e-07, "loss": 0.4864, "step": 25000 }, { "epoch": 0.8907137894793096, "grad_norm": 1.2806485413036384, "learning_rate": 6.197602116773582e-07, "loss": 0.4989, "step": 25001 }, { "epoch": 0.8907494166058036, "grad_norm": 1.3058243101719795, "learning_rate": 6.193603538733095e-07, "loss": 0.3853, "step": 25002 }, { "epoch": 0.8907850437322977, "grad_norm": 2.2326920863554394, "learning_rate": 6.189606209790322e-07, "loss": 0.4496, "step": 25003 }, { "epoch": 0.8908206708587919, "grad_norm": 1.2568693434763163, "learning_rate": 6.185610129998532e-07, "loss": 0.43, "step": 25004 }, { "epoch": 0.890856297985286, "grad_norm": 1.564616192365921, "learning_rate": 6.181615299410882e-07, "loss": 0.488, "step": 25005 }, { "epoch": 0.8908919251117801, "grad_norm": 1.155160361829856, "learning_rate": 6.177621718080629e-07, "loss": 0.3673, "step": 25006 }, { "epoch": 0.8909275522382742, "grad_norm": 1.2077775273505604, "learning_rate": 6.173629386060909e-07, "loss": 0.3659, "step": 25007 }, { "epoch": 0.8909631793647683, "grad_norm": 1.2300710109966384, "learning_rate": 6.169638303404912e-07, "loss": 0.3836, "step": 25008 }, { "epoch": 0.8909988064912624, "grad_norm": 1.293775257083895, "learning_rate": 6.165648470165741e-07, "loss": 0.4386, "step": 25009 }, { "epoch": 0.8910344336177566, "grad_norm": 1.302372511485389, "learning_rate": 6.161659886396553e-07, "loss": 0.3626, "step": 25010 }, { "epoch": 0.8910700607442507, "grad_norm": 1.7902143545756593, "learning_rate": 6.157672552150473e-07, "loss": 0.499, "step": 25011 }, { "epoch": 0.8911056878707447, "grad_norm": 1.5649101028928023, "learning_rate": 6.153686467480547e-07, "loss": 0.4763, "step": 25012 }, { "epoch": 0.8911413149972389, "grad_norm": 1.3486495880037002, "learning_rate": 6.149701632439886e-07, "loss": 0.4069, "step": 25013 }, { "epoch": 0.891176942123733, "grad_norm": 1.6664981040052085, "learning_rate": 6.14571804708155e-07, "loss": 0.5249, "step": 25014 }, { "epoch": 0.8912125692502271, "grad_norm": 1.2411850064937606, "learning_rate": 6.141735711458585e-07, "loss": 0.3937, "step": 25015 }, { "epoch": 0.8912481963767213, "grad_norm": 1.548905229462603, "learning_rate": 6.137754625624004e-07, "loss": 0.4338, "step": 25016 }, { "epoch": 0.8912838235032153, "grad_norm": 1.4120827421098539, "learning_rate": 6.133774789630842e-07, "loss": 0.374, "step": 25017 }, { "epoch": 0.8913194506297094, "grad_norm": 1.2155846750909982, "learning_rate": 6.129796203532057e-07, "loss": 0.3989, "step": 25018 }, { "epoch": 0.8913550777562036, "grad_norm": 1.1752791825087852, "learning_rate": 6.12581886738064e-07, "loss": 0.3483, "step": 25019 }, { "epoch": 0.8913907048826977, "grad_norm": 1.8982421359761703, "learning_rate": 6.12184278122957e-07, "loss": 0.5088, "step": 25020 }, { "epoch": 0.8914263320091917, "grad_norm": 1.245265047704199, "learning_rate": 6.117867945131794e-07, "loss": 0.4255, "step": 25021 }, { "epoch": 0.8914619591356859, "grad_norm": 1.4323893364525708, "learning_rate": 6.113894359140204e-07, "loss": 0.4452, "step": 25022 }, { "epoch": 0.89149758626218, "grad_norm": 1.6842365645329096, "learning_rate": 6.109922023307746e-07, "loss": 0.5258, "step": 25023 }, { "epoch": 0.8915332133886741, "grad_norm": 1.488090737404407, "learning_rate": 6.10595093768731e-07, "loss": 0.5014, "step": 25024 }, { "epoch": 0.8915688405151683, "grad_norm": 1.870312429398334, "learning_rate": 6.101981102331744e-07, "loss": 0.5288, "step": 25025 }, { "epoch": 0.8916044676416623, "grad_norm": 1.4615805337865553, "learning_rate": 6.09801251729395e-07, "loss": 0.3495, "step": 25026 }, { "epoch": 0.8916400947681564, "grad_norm": 1.2951199567945932, "learning_rate": 6.09404518262674e-07, "loss": 0.399, "step": 25027 }, { "epoch": 0.8916757218946506, "grad_norm": 1.675080936488601, "learning_rate": 6.090079098382951e-07, "loss": 0.4645, "step": 25028 }, { "epoch": 0.8917113490211447, "grad_norm": 2.0326186278015648, "learning_rate": 6.086114264615406e-07, "loss": 0.5205, "step": 25029 }, { "epoch": 0.8917469761476388, "grad_norm": 3.0968716140831694, "learning_rate": 6.082150681376897e-07, "loss": 0.3871, "step": 25030 }, { "epoch": 0.891782603274133, "grad_norm": 1.2636244978054563, "learning_rate": 6.078188348720183e-07, "loss": 0.4068, "step": 25031 }, { "epoch": 0.891818230400627, "grad_norm": 1.5789642370804435, "learning_rate": 6.074227266698063e-07, "loss": 0.5366, "step": 25032 }, { "epoch": 0.8918538575271211, "grad_norm": 1.3779925335888037, "learning_rate": 6.07026743536323e-07, "loss": 0.4692, "step": 25033 }, { "epoch": 0.8918894846536153, "grad_norm": 1.260450561782193, "learning_rate": 6.066308854768444e-07, "loss": 0.4243, "step": 25034 }, { "epoch": 0.8919251117801094, "grad_norm": 1.4518470556111684, "learning_rate": 6.062351524966414e-07, "loss": 0.548, "step": 25035 }, { "epoch": 0.8919607389066034, "grad_norm": 1.1160947056526063, "learning_rate": 6.058395446009823e-07, "loss": 0.3655, "step": 25036 }, { "epoch": 0.8919963660330976, "grad_norm": 1.6137298029376623, "learning_rate": 6.054440617951385e-07, "loss": 0.5193, "step": 25037 }, { "epoch": 0.8920319931595917, "grad_norm": 1.61998318432545, "learning_rate": 6.050487040843711e-07, "loss": 0.3913, "step": 25038 }, { "epoch": 0.8920676202860858, "grad_norm": 1.297721600585346, "learning_rate": 6.046534714739483e-07, "loss": 0.4102, "step": 25039 }, { "epoch": 0.89210324741258, "grad_norm": 1.694425135157884, "learning_rate": 6.042583639691301e-07, "loss": 0.4837, "step": 25040 }, { "epoch": 0.892138874539074, "grad_norm": 1.1427601027530658, "learning_rate": 6.038633815751815e-07, "loss": 0.4212, "step": 25041 }, { "epoch": 0.8921745016655681, "grad_norm": 1.7113281299856493, "learning_rate": 6.034685242973548e-07, "loss": 0.5031, "step": 25042 }, { "epoch": 0.8922101287920623, "grad_norm": 1.2999769076039223, "learning_rate": 6.030737921409169e-07, "loss": 0.4312, "step": 25043 }, { "epoch": 0.8922457559185564, "grad_norm": 1.5701575412348192, "learning_rate": 6.026791851111169e-07, "loss": 0.4702, "step": 25044 }, { "epoch": 0.8922813830450504, "grad_norm": 1.5730948242336344, "learning_rate": 6.022847032132128e-07, "loss": 0.3657, "step": 25045 }, { "epoch": 0.8923170101715446, "grad_norm": 1.705590128471188, "learning_rate": 6.018903464524573e-07, "loss": 0.5347, "step": 25046 }, { "epoch": 0.8923526372980387, "grad_norm": 1.4638348770859015, "learning_rate": 6.014961148341003e-07, "loss": 0.4406, "step": 25047 }, { "epoch": 0.8923882644245328, "grad_norm": 1.191011685245288, "learning_rate": 6.011020083633923e-07, "loss": 0.4477, "step": 25048 }, { "epoch": 0.892423891551027, "grad_norm": 1.5602364385787935, "learning_rate": 6.00708027045579e-07, "loss": 0.4435, "step": 25049 }, { "epoch": 0.892459518677521, "grad_norm": 1.253193789298646, "learning_rate": 6.003141708859084e-07, "loss": 0.4491, "step": 25050 }, { "epoch": 0.8924951458040151, "grad_norm": 1.2683481219478543, "learning_rate": 5.999204398896241e-07, "loss": 0.4393, "step": 25051 }, { "epoch": 0.8925307729305093, "grad_norm": 1.3572455790627094, "learning_rate": 5.995268340619709e-07, "loss": 0.4364, "step": 25052 }, { "epoch": 0.8925664000570034, "grad_norm": 1.39534598449796, "learning_rate": 5.991333534081878e-07, "loss": 0.4202, "step": 25053 }, { "epoch": 0.8926020271834975, "grad_norm": 1.713107496320807, "learning_rate": 5.987399979335151e-07, "loss": 0.5267, "step": 25054 }, { "epoch": 0.8926376543099916, "grad_norm": 1.178551917521086, "learning_rate": 5.983467676431898e-07, "loss": 0.4289, "step": 25055 }, { "epoch": 0.8926732814364857, "grad_norm": 1.7182894616947975, "learning_rate": 5.979536625424487e-07, "loss": 0.5071, "step": 25056 }, { "epoch": 0.8927089085629799, "grad_norm": 1.420749201340678, "learning_rate": 5.975606826365265e-07, "loss": 0.5423, "step": 25057 }, { "epoch": 0.892744535689474, "grad_norm": 1.0935651976292518, "learning_rate": 5.971678279306547e-07, "loss": 0.3785, "step": 25058 }, { "epoch": 0.8927801628159681, "grad_norm": 1.3163735307152944, "learning_rate": 5.96775098430068e-07, "loss": 0.3781, "step": 25059 }, { "epoch": 0.8928157899424622, "grad_norm": 1.6456625316690652, "learning_rate": 5.96382494139991e-07, "loss": 0.5323, "step": 25060 }, { "epoch": 0.8928514170689563, "grad_norm": 1.6942694952457964, "learning_rate": 5.959900150656562e-07, "loss": 0.5156, "step": 25061 }, { "epoch": 0.8928870441954504, "grad_norm": 1.5679638223190226, "learning_rate": 5.955976612122849e-07, "loss": 0.5635, "step": 25062 }, { "epoch": 0.8929226713219446, "grad_norm": 1.1008852608176045, "learning_rate": 5.952054325851053e-07, "loss": 0.3945, "step": 25063 }, { "epoch": 0.8929582984484387, "grad_norm": 1.674787541985518, "learning_rate": 5.948133291893387e-07, "loss": 0.534, "step": 25064 }, { "epoch": 0.8929939255749327, "grad_norm": 1.533920086157111, "learning_rate": 5.944213510302054e-07, "loss": 0.4762, "step": 25065 }, { "epoch": 0.8930295527014269, "grad_norm": 1.280368979494348, "learning_rate": 5.940294981129269e-07, "loss": 0.4267, "step": 25066 }, { "epoch": 0.893065179827921, "grad_norm": 1.2874728237499262, "learning_rate": 5.93637770442721e-07, "loss": 0.4029, "step": 25067 }, { "epoch": 0.8931008069544151, "grad_norm": 1.3369919752272588, "learning_rate": 5.932461680248014e-07, "loss": 0.4348, "step": 25068 }, { "epoch": 0.8931364340809093, "grad_norm": 1.3606214557446312, "learning_rate": 5.928546908643839e-07, "loss": 0.4376, "step": 25069 }, { "epoch": 0.8931720612074033, "grad_norm": 1.4641811030404475, "learning_rate": 5.924633389666823e-07, "loss": 0.4322, "step": 25070 }, { "epoch": 0.8932076883338974, "grad_norm": 1.4900253374022145, "learning_rate": 5.920721123369067e-07, "loss": 0.3897, "step": 25071 }, { "epoch": 0.8932433154603916, "grad_norm": 1.8453334939238295, "learning_rate": 5.916810109802651e-07, "loss": 0.3932, "step": 25072 }, { "epoch": 0.8932789425868857, "grad_norm": 1.6541621184218636, "learning_rate": 5.912900349019679e-07, "loss": 0.5334, "step": 25073 }, { "epoch": 0.8933145697133797, "grad_norm": 1.6868318064347265, "learning_rate": 5.90899184107221e-07, "loss": 0.5142, "step": 25074 }, { "epoch": 0.8933501968398739, "grad_norm": 1.369040575766243, "learning_rate": 5.905084586012278e-07, "loss": 0.3949, "step": 25075 }, { "epoch": 0.893385823966368, "grad_norm": 1.430705446809214, "learning_rate": 5.901178583891908e-07, "loss": 0.499, "step": 25076 }, { "epoch": 0.8934214510928621, "grad_norm": 1.2057890986678104, "learning_rate": 5.897273834763118e-07, "loss": 0.3981, "step": 25077 }, { "epoch": 0.8934570782193563, "grad_norm": 1.33062461886778, "learning_rate": 5.893370338677917e-07, "loss": 0.4402, "step": 25078 }, { "epoch": 0.8934927053458503, "grad_norm": 1.270403352405324, "learning_rate": 5.889468095688244e-07, "loss": 0.3917, "step": 25079 }, { "epoch": 0.8935283324723444, "grad_norm": 1.834435779800449, "learning_rate": 5.885567105846068e-07, "loss": 0.5237, "step": 25080 }, { "epoch": 0.8935639595988386, "grad_norm": 1.5961606099390213, "learning_rate": 5.881667369203381e-07, "loss": 0.5702, "step": 25081 }, { "epoch": 0.8935995867253327, "grad_norm": 2.593756953623315, "learning_rate": 5.877768885812052e-07, "loss": 0.446, "step": 25082 }, { "epoch": 0.8936352138518268, "grad_norm": 1.930746600388533, "learning_rate": 5.873871655724051e-07, "loss": 0.6078, "step": 25083 }, { "epoch": 0.893670840978321, "grad_norm": 1.206411470795106, "learning_rate": 5.869975678991214e-07, "loss": 0.3781, "step": 25084 }, { "epoch": 0.893706468104815, "grad_norm": 1.6784460007864406, "learning_rate": 5.866080955665455e-07, "loss": 0.4302, "step": 25085 }, { "epoch": 0.8937420952313091, "grad_norm": 1.204859232508136, "learning_rate": 5.862187485798609e-07, "loss": 0.3693, "step": 25086 }, { "epoch": 0.8937777223578033, "grad_norm": 1.8828002839935196, "learning_rate": 5.858295269442537e-07, "loss": 0.622, "step": 25087 }, { "epoch": 0.8938133494842974, "grad_norm": 1.17955456140193, "learning_rate": 5.854404306649075e-07, "loss": 0.3998, "step": 25088 }, { "epoch": 0.8938489766107914, "grad_norm": 1.2250935495542647, "learning_rate": 5.850514597470025e-07, "loss": 0.4231, "step": 25089 }, { "epoch": 0.8938846037372856, "grad_norm": 1.5334863085677408, "learning_rate": 5.846626141957168e-07, "loss": 0.5353, "step": 25090 }, { "epoch": 0.8939202308637797, "grad_norm": 1.5430700730770999, "learning_rate": 5.842738940162296e-07, "loss": 0.4597, "step": 25091 }, { "epoch": 0.8939558579902738, "grad_norm": 1.2904831806572545, "learning_rate": 5.83885299213719e-07, "loss": 0.3831, "step": 25092 }, { "epoch": 0.893991485116768, "grad_norm": 1.2981886502370819, "learning_rate": 5.834968297933541e-07, "loss": 0.46, "step": 25093 }, { "epoch": 0.894027112243262, "grad_norm": 1.4906263864564444, "learning_rate": 5.831084857603109e-07, "loss": 0.5169, "step": 25094 }, { "epoch": 0.8940627393697561, "grad_norm": 1.2384198396494932, "learning_rate": 5.827202671197607e-07, "loss": 0.3715, "step": 25095 }, { "epoch": 0.8940983664962503, "grad_norm": 1.3331706454308458, "learning_rate": 5.823321738768739e-07, "loss": 0.4169, "step": 25096 }, { "epoch": 0.8941339936227444, "grad_norm": 1.165423341101407, "learning_rate": 5.819442060368153e-07, "loss": 0.3504, "step": 25097 }, { "epoch": 0.8941696207492384, "grad_norm": 1.6367773822209757, "learning_rate": 5.815563636047539e-07, "loss": 0.5576, "step": 25098 }, { "epoch": 0.8942052478757326, "grad_norm": 1.5346209014196295, "learning_rate": 5.811686465858513e-07, "loss": 0.4934, "step": 25099 }, { "epoch": 0.8942408750022267, "grad_norm": 1.3031384695289345, "learning_rate": 5.807810549852721e-07, "loss": 0.3795, "step": 25100 }, { "epoch": 0.8942765021287208, "grad_norm": 1.3310503381273315, "learning_rate": 5.803935888081769e-07, "loss": 0.4467, "step": 25101 }, { "epoch": 0.894312129255215, "grad_norm": 1.2821488818309963, "learning_rate": 5.800062480597235e-07, "loss": 0.4173, "step": 25102 }, { "epoch": 0.894347756381709, "grad_norm": 1.2896638796324944, "learning_rate": 5.796190327450713e-07, "loss": 0.4159, "step": 25103 }, { "epoch": 0.8943833835082031, "grad_norm": 1.0712205458956627, "learning_rate": 5.792319428693771e-07, "loss": 0.4044, "step": 25104 }, { "epoch": 0.8944190106346973, "grad_norm": 1.360101471582351, "learning_rate": 5.788449784377948e-07, "loss": 0.3887, "step": 25105 }, { "epoch": 0.8944546377611914, "grad_norm": 1.7458525079327136, "learning_rate": 5.784581394554756e-07, "loss": 0.5073, "step": 25106 }, { "epoch": 0.8944902648876855, "grad_norm": 1.4887362203661059, "learning_rate": 5.780714259275733e-07, "loss": 0.4218, "step": 25107 }, { "epoch": 0.8945258920141796, "grad_norm": 1.603973131640708, "learning_rate": 5.776848378592337e-07, "loss": 0.5596, "step": 25108 }, { "epoch": 0.8945615191406737, "grad_norm": 1.4052751684296263, "learning_rate": 5.772983752556061e-07, "loss": 0.5494, "step": 25109 }, { "epoch": 0.8945971462671678, "grad_norm": 1.199186583633557, "learning_rate": 5.769120381218362e-07, "loss": 0.3667, "step": 25110 }, { "epoch": 0.894632773393662, "grad_norm": 1.6556251389527457, "learning_rate": 5.765258264630713e-07, "loss": 0.4995, "step": 25111 }, { "epoch": 0.8946684005201561, "grad_norm": 1.979608429987658, "learning_rate": 5.761397402844493e-07, "loss": 0.5769, "step": 25112 }, { "epoch": 0.8947040276466501, "grad_norm": 1.2132396779362558, "learning_rate": 5.757537795911161e-07, "loss": 0.3961, "step": 25113 }, { "epoch": 0.8947396547731443, "grad_norm": 1.2296072180546582, "learning_rate": 5.753679443882065e-07, "loss": 0.3459, "step": 25114 }, { "epoch": 0.8947752818996384, "grad_norm": 1.276598964094391, "learning_rate": 5.749822346808598e-07, "loss": 0.3687, "step": 25115 }, { "epoch": 0.8948109090261325, "grad_norm": 1.5447998536278162, "learning_rate": 5.745966504742139e-07, "loss": 0.4519, "step": 25116 }, { "epoch": 0.8948465361526267, "grad_norm": 1.1224521315748137, "learning_rate": 5.742111917733994e-07, "loss": 0.3833, "step": 25117 }, { "epoch": 0.8948821632791207, "grad_norm": 1.3286367896285742, "learning_rate": 5.738258585835532e-07, "loss": 0.4418, "step": 25118 }, { "epoch": 0.8949177904056148, "grad_norm": 1.2709932814629588, "learning_rate": 5.734406509098033e-07, "loss": 0.3944, "step": 25119 }, { "epoch": 0.894953417532109, "grad_norm": 1.3804836943338208, "learning_rate": 5.730555687572803e-07, "loss": 0.4073, "step": 25120 }, { "epoch": 0.8949890446586031, "grad_norm": 1.5513793950190213, "learning_rate": 5.72670612131111e-07, "loss": 0.5099, "step": 25121 }, { "epoch": 0.8950246717850971, "grad_norm": 1.2322089812452313, "learning_rate": 5.722857810364235e-07, "loss": 0.398, "step": 25122 }, { "epoch": 0.8950602989115913, "grad_norm": 1.182675124117538, "learning_rate": 5.719010754783383e-07, "loss": 0.3897, "step": 25123 }, { "epoch": 0.8950959260380854, "grad_norm": 1.387869830061188, "learning_rate": 5.715164954619801e-07, "loss": 0.5453, "step": 25124 }, { "epoch": 0.8951315531645795, "grad_norm": 1.5476984277638686, "learning_rate": 5.711320409924703e-07, "loss": 0.5318, "step": 25125 }, { "epoch": 0.8951671802910737, "grad_norm": 1.3377949743352786, "learning_rate": 5.707477120749272e-07, "loss": 0.4015, "step": 25126 }, { "epoch": 0.8952028074175677, "grad_norm": 2.3808006885941886, "learning_rate": 5.70363508714471e-07, "loss": 0.5639, "step": 25127 }, { "epoch": 0.8952384345440618, "grad_norm": 1.311753308130665, "learning_rate": 5.699794309162132e-07, "loss": 0.3721, "step": 25128 }, { "epoch": 0.895274061670556, "grad_norm": 1.1926021859540081, "learning_rate": 5.695954786852731e-07, "loss": 0.3629, "step": 25129 }, { "epoch": 0.8953096887970501, "grad_norm": 1.592192160791856, "learning_rate": 5.692116520267587e-07, "loss": 0.4993, "step": 25130 }, { "epoch": 0.8953453159235442, "grad_norm": 1.2259323514438627, "learning_rate": 5.688279509457828e-07, "loss": 0.4348, "step": 25131 }, { "epoch": 0.8953809430500383, "grad_norm": 1.8757998552980715, "learning_rate": 5.684443754474544e-07, "loss": 0.4507, "step": 25132 }, { "epoch": 0.8954165701765324, "grad_norm": 1.2397554063315552, "learning_rate": 5.68060925536884e-07, "loss": 0.3449, "step": 25133 }, { "epoch": 0.8954521973030265, "grad_norm": 1.6036048028794916, "learning_rate": 5.67677601219172e-07, "loss": 0.5013, "step": 25134 }, { "epoch": 0.8954878244295207, "grad_norm": 1.5626413697053811, "learning_rate": 5.672944024994287e-07, "loss": 0.5385, "step": 25135 }, { "epoch": 0.8955234515560148, "grad_norm": 1.902992240155018, "learning_rate": 5.66911329382751e-07, "loss": 0.469, "step": 25136 }, { "epoch": 0.8955590786825088, "grad_norm": 1.3350804910105372, "learning_rate": 5.665283818742418e-07, "loss": 0.4889, "step": 25137 }, { "epoch": 0.895594705809003, "grad_norm": 1.6532748911735375, "learning_rate": 5.661455599790033e-07, "loss": 0.4968, "step": 25138 }, { "epoch": 0.8956303329354971, "grad_norm": 1.3572387799801793, "learning_rate": 5.657628637021284e-07, "loss": 0.4187, "step": 25139 }, { "epoch": 0.8956659600619912, "grad_norm": 1.2425356176091198, "learning_rate": 5.653802930487151e-07, "loss": 0.4573, "step": 25140 }, { "epoch": 0.8957015871884854, "grad_norm": 1.2647147710074516, "learning_rate": 5.649978480238583e-07, "loss": 0.4021, "step": 25141 }, { "epoch": 0.8957372143149794, "grad_norm": 1.242389449207545, "learning_rate": 5.646155286326505e-07, "loss": 0.4055, "step": 25142 }, { "epoch": 0.8957728414414735, "grad_norm": 1.9153360531121628, "learning_rate": 5.64233334880181e-07, "loss": 0.5145, "step": 25143 }, { "epoch": 0.8958084685679677, "grad_norm": 1.5551495027564102, "learning_rate": 5.638512667715401e-07, "loss": 0.4396, "step": 25144 }, { "epoch": 0.8958440956944618, "grad_norm": 1.673629686526521, "learning_rate": 5.63469324311815e-07, "loss": 0.5387, "step": 25145 }, { "epoch": 0.8958797228209558, "grad_norm": 1.455726181121766, "learning_rate": 5.630875075060915e-07, "loss": 0.4664, "step": 25146 }, { "epoch": 0.89591534994745, "grad_norm": 1.476154181807628, "learning_rate": 5.627058163594534e-07, "loss": 0.4699, "step": 25147 }, { "epoch": 0.8959509770739441, "grad_norm": 1.293419470303586, "learning_rate": 5.623242508769844e-07, "loss": 0.3742, "step": 25148 }, { "epoch": 0.8959866042004382, "grad_norm": 1.6086946774996695, "learning_rate": 5.619428110637637e-07, "loss": 0.5317, "step": 25149 }, { "epoch": 0.8960222313269324, "grad_norm": 2.2139501997398354, "learning_rate": 5.615614969248717e-07, "loss": 0.4713, "step": 25150 }, { "epoch": 0.8960578584534264, "grad_norm": 1.6062411174847555, "learning_rate": 5.611803084653866e-07, "loss": 0.6203, "step": 25151 }, { "epoch": 0.8960934855799205, "grad_norm": 1.661501639955227, "learning_rate": 5.607992456903822e-07, "loss": 0.4503, "step": 25152 }, { "epoch": 0.8961291127064147, "grad_norm": 1.3036928494192446, "learning_rate": 5.604183086049342e-07, "loss": 0.4331, "step": 25153 }, { "epoch": 0.8961647398329088, "grad_norm": 1.5802176653561646, "learning_rate": 5.60037497214112e-07, "loss": 0.615, "step": 25154 }, { "epoch": 0.8962003669594029, "grad_norm": 1.4302409456544896, "learning_rate": 5.596568115229906e-07, "loss": 0.3895, "step": 25155 }, { "epoch": 0.896235994085897, "grad_norm": 1.0889608443923802, "learning_rate": 5.592762515366356e-07, "loss": 0.3586, "step": 25156 }, { "epoch": 0.8962716212123911, "grad_norm": 1.2141299352033894, "learning_rate": 5.588958172601178e-07, "loss": 0.3939, "step": 25157 }, { "epoch": 0.8963072483388852, "grad_norm": 1.6022930180858848, "learning_rate": 5.585155086985006e-07, "loss": 0.4841, "step": 25158 }, { "epoch": 0.8963428754653794, "grad_norm": 1.190177201638589, "learning_rate": 5.581353258568489e-07, "loss": 0.4051, "step": 25159 }, { "epoch": 0.8963785025918735, "grad_norm": 1.1566067144115089, "learning_rate": 5.577552687402233e-07, "loss": 0.4233, "step": 25160 }, { "epoch": 0.8964141297183675, "grad_norm": 1.5029535926949789, "learning_rate": 5.573753373536872e-07, "loss": 0.4793, "step": 25161 }, { "epoch": 0.8964497568448617, "grad_norm": 1.3879041807081764, "learning_rate": 5.569955317022968e-07, "loss": 0.4829, "step": 25162 }, { "epoch": 0.8964853839713558, "grad_norm": 1.3700419479621235, "learning_rate": 5.566158517911125e-07, "loss": 0.4996, "step": 25163 }, { "epoch": 0.8965210110978499, "grad_norm": 1.2817823070918326, "learning_rate": 5.562362976251901e-07, "loss": 0.4542, "step": 25164 }, { "epoch": 0.8965566382243441, "grad_norm": 1.2873586171277005, "learning_rate": 5.558568692095801e-07, "loss": 0.4212, "step": 25165 }, { "epoch": 0.8965922653508381, "grad_norm": 1.4003778657737218, "learning_rate": 5.554775665493384e-07, "loss": 0.3961, "step": 25166 }, { "epoch": 0.8966278924773322, "grad_norm": 1.8350422708650307, "learning_rate": 5.550983896495132e-07, "loss": 0.4967, "step": 25167 }, { "epoch": 0.8966635196038264, "grad_norm": 1.395217745829991, "learning_rate": 5.547193385151561e-07, "loss": 0.5108, "step": 25168 }, { "epoch": 0.8966991467303205, "grad_norm": 1.4358271751403284, "learning_rate": 5.543404131513097e-07, "loss": 0.4938, "step": 25169 }, { "epoch": 0.8967347738568145, "grad_norm": 1.17931471006901, "learning_rate": 5.539616135630254e-07, "loss": 0.3673, "step": 25170 }, { "epoch": 0.8967704009833087, "grad_norm": 1.394631256342815, "learning_rate": 5.535829397553438e-07, "loss": 0.4489, "step": 25171 }, { "epoch": 0.8968060281098028, "grad_norm": 1.2157037892763698, "learning_rate": 5.532043917333074e-07, "loss": 0.4386, "step": 25172 }, { "epoch": 0.8968416552362969, "grad_norm": 1.560196515452873, "learning_rate": 5.5282596950196e-07, "loss": 0.5294, "step": 25173 }, { "epoch": 0.8968772823627911, "grad_norm": 1.2132530566293462, "learning_rate": 5.524476730663353e-07, "loss": 0.3486, "step": 25174 }, { "epoch": 0.8969129094892851, "grad_norm": 1.5321707181690263, "learning_rate": 5.520695024314749e-07, "loss": 0.4334, "step": 25175 }, { "epoch": 0.8969485366157792, "grad_norm": 1.2884031107219682, "learning_rate": 5.516914576024124e-07, "loss": 0.3979, "step": 25176 }, { "epoch": 0.8969841637422734, "grad_norm": 1.6674999065496723, "learning_rate": 5.513135385841806e-07, "loss": 0.5278, "step": 25177 }, { "epoch": 0.8970197908687675, "grad_norm": 1.3853020199739996, "learning_rate": 5.509357453818142e-07, "loss": 0.43, "step": 25178 }, { "epoch": 0.8970554179952616, "grad_norm": 1.5004206536631974, "learning_rate": 5.505580780003439e-07, "loss": 0.4948, "step": 25179 }, { "epoch": 0.8970910451217557, "grad_norm": 1.5000058675351478, "learning_rate": 5.501805364447965e-07, "loss": 0.4688, "step": 25180 }, { "epoch": 0.8971266722482498, "grad_norm": 1.4134269698116269, "learning_rate": 5.498031207202015e-07, "loss": 0.4349, "step": 25181 }, { "epoch": 0.8971622993747439, "grad_norm": 1.6185858600550709, "learning_rate": 5.494258308315815e-07, "loss": 0.5832, "step": 25182 }, { "epoch": 0.8971979265012381, "grad_norm": 1.8096752229736939, "learning_rate": 5.490486667839623e-07, "loss": 0.4613, "step": 25183 }, { "epoch": 0.8972335536277322, "grad_norm": 1.2682772484266749, "learning_rate": 5.486716285823667e-07, "loss": 0.4101, "step": 25184 }, { "epoch": 0.8972691807542262, "grad_norm": 1.3790056210367694, "learning_rate": 5.482947162318131e-07, "loss": 0.374, "step": 25185 }, { "epoch": 0.8973048078807204, "grad_norm": 1.8008656537369607, "learning_rate": 5.479179297373239e-07, "loss": 0.4629, "step": 25186 }, { "epoch": 0.8973404350072145, "grad_norm": 1.6085993335771132, "learning_rate": 5.475412691039128e-07, "loss": 0.4942, "step": 25187 }, { "epoch": 0.8973760621337086, "grad_norm": 1.5134195341898316, "learning_rate": 5.471647343365982e-07, "loss": 0.4926, "step": 25188 }, { "epoch": 0.8974116892602028, "grad_norm": 1.5160362617651928, "learning_rate": 5.467883254403905e-07, "loss": 0.375, "step": 25189 }, { "epoch": 0.8974473163866968, "grad_norm": 1.2267954199646034, "learning_rate": 5.464120424203046e-07, "loss": 0.3919, "step": 25190 }, { "epoch": 0.8974829435131909, "grad_norm": 1.3283976792590209, "learning_rate": 5.460358852813496e-07, "loss": 0.4641, "step": 25191 }, { "epoch": 0.8975185706396851, "grad_norm": 1.433773589410683, "learning_rate": 5.456598540285341e-07, "loss": 0.3973, "step": 25192 }, { "epoch": 0.8975541977661792, "grad_norm": 1.2411913508844838, "learning_rate": 5.45283948666866e-07, "loss": 0.4183, "step": 25193 }, { "epoch": 0.8975898248926732, "grad_norm": 2.183729871784747, "learning_rate": 5.449081692013513e-07, "loss": 0.5161, "step": 25194 }, { "epoch": 0.8976254520191674, "grad_norm": 1.3124347294220573, "learning_rate": 5.445325156369918e-07, "loss": 0.4331, "step": 25195 }, { "epoch": 0.8976610791456615, "grad_norm": 1.276514129018795, "learning_rate": 5.441569879787911e-07, "loss": 0.3872, "step": 25196 }, { "epoch": 0.8976967062721556, "grad_norm": 1.2891283095344814, "learning_rate": 5.437815862317519e-07, "loss": 0.3719, "step": 25197 }, { "epoch": 0.8977323333986498, "grad_norm": 1.3201349879611755, "learning_rate": 5.43406310400868e-07, "loss": 0.3485, "step": 25198 }, { "epoch": 0.8977679605251438, "grad_norm": 5.429045282984093, "learning_rate": 5.430311604911387e-07, "loss": 0.4854, "step": 25199 }, { "epoch": 0.8978035876516379, "grad_norm": 1.351882300697591, "learning_rate": 5.426561365075589e-07, "loss": 0.5095, "step": 25200 }, { "epoch": 0.8978392147781321, "grad_norm": 1.7961760548402101, "learning_rate": 5.422812384551257e-07, "loss": 0.5242, "step": 25201 }, { "epoch": 0.8978748419046262, "grad_norm": 1.3795862056611918, "learning_rate": 5.419064663388263e-07, "loss": 0.4717, "step": 25202 }, { "epoch": 0.8979104690311203, "grad_norm": 1.2427070947459102, "learning_rate": 5.415318201636543e-07, "loss": 0.4395, "step": 25203 }, { "epoch": 0.8979460961576144, "grad_norm": 1.5677124437089336, "learning_rate": 5.411572999345959e-07, "loss": 0.549, "step": 25204 }, { "epoch": 0.8979817232841085, "grad_norm": 1.192455352898757, "learning_rate": 5.407829056566405e-07, "loss": 0.414, "step": 25205 }, { "epoch": 0.8980173504106026, "grad_norm": 1.2133358565798162, "learning_rate": 5.404086373347705e-07, "loss": 0.3964, "step": 25206 }, { "epoch": 0.8980529775370968, "grad_norm": 1.4748205809827974, "learning_rate": 5.400344949739722e-07, "loss": 0.4855, "step": 25207 }, { "epoch": 0.8980886046635909, "grad_norm": 1.2143218840428927, "learning_rate": 5.396604785792281e-07, "loss": 0.3835, "step": 25208 }, { "epoch": 0.8981242317900849, "grad_norm": 1.4279592852231013, "learning_rate": 5.392865881555154e-07, "loss": 0.6149, "step": 25209 }, { "epoch": 0.8981598589165791, "grad_norm": 3.2703444771080976, "learning_rate": 5.389128237078156e-07, "loss": 0.4711, "step": 25210 }, { "epoch": 0.8981954860430732, "grad_norm": 1.358336400781491, "learning_rate": 5.385391852411038e-07, "loss": 0.4755, "step": 25211 }, { "epoch": 0.8982311131695673, "grad_norm": 1.3335032833596172, "learning_rate": 5.38165672760358e-07, "loss": 0.3669, "step": 25212 }, { "epoch": 0.8982667402960615, "grad_norm": 1.6356550282830247, "learning_rate": 5.377922862705465e-07, "loss": 0.4501, "step": 25213 }, { "epoch": 0.8983023674225555, "grad_norm": 1.3076334041250879, "learning_rate": 5.374190257766454e-07, "loss": 0.4196, "step": 25214 }, { "epoch": 0.8983379945490496, "grad_norm": 1.2971402326158799, "learning_rate": 5.37045891283624e-07, "loss": 0.4415, "step": 25215 }, { "epoch": 0.8983736216755438, "grad_norm": 1.3675227689742087, "learning_rate": 5.366728827964518e-07, "loss": 0.4263, "step": 25216 }, { "epoch": 0.8984092488020379, "grad_norm": 1.5276613312229648, "learning_rate": 5.363000003200936e-07, "loss": 0.5046, "step": 25217 }, { "epoch": 0.898444875928532, "grad_norm": 1.6138420695420872, "learning_rate": 5.359272438595153e-07, "loss": 0.5858, "step": 25218 }, { "epoch": 0.8984805030550261, "grad_norm": 1.3891978836525865, "learning_rate": 5.355546134196821e-07, "loss": 0.5054, "step": 25219 }, { "epoch": 0.8985161301815202, "grad_norm": 1.6999340732810222, "learning_rate": 5.351821090055531e-07, "loss": 0.61, "step": 25220 }, { "epoch": 0.8985517573080143, "grad_norm": 1.2820001800018543, "learning_rate": 5.348097306220901e-07, "loss": 0.4221, "step": 25221 }, { "epoch": 0.8985873844345085, "grad_norm": 2.1282200296741305, "learning_rate": 5.344374782742512e-07, "loss": 0.5046, "step": 25222 }, { "epoch": 0.8986230115610025, "grad_norm": 1.5173181940531957, "learning_rate": 5.340653519669947e-07, "loss": 0.5016, "step": 25223 }, { "epoch": 0.8986586386874966, "grad_norm": 1.3088008287073418, "learning_rate": 5.336933517052734e-07, "loss": 0.462, "step": 25224 }, { "epoch": 0.8986942658139908, "grad_norm": 1.6244622885836904, "learning_rate": 5.33321477494042e-07, "loss": 0.5617, "step": 25225 }, { "epoch": 0.8987298929404849, "grad_norm": 1.4268006998664586, "learning_rate": 5.329497293382513e-07, "loss": 0.4165, "step": 25226 }, { "epoch": 0.898765520066979, "grad_norm": 1.3602334859958674, "learning_rate": 5.325781072428538e-07, "loss": 0.4107, "step": 25227 }, { "epoch": 0.8988011471934731, "grad_norm": 1.2382191174682053, "learning_rate": 5.322066112127932e-07, "loss": 0.4162, "step": 25228 }, { "epoch": 0.8988367743199672, "grad_norm": 1.257331110563545, "learning_rate": 5.318352412530204e-07, "loss": 0.4498, "step": 25229 }, { "epoch": 0.8988724014464613, "grad_norm": 1.2526123824420807, "learning_rate": 5.314639973684788e-07, "loss": 0.3854, "step": 25230 }, { "epoch": 0.8989080285729555, "grad_norm": 1.2431755046939117, "learning_rate": 5.310928795641124e-07, "loss": 0.3847, "step": 25231 }, { "epoch": 0.8989436556994496, "grad_norm": 1.5507647566803564, "learning_rate": 5.307218878448638e-07, "loss": 0.5674, "step": 25232 }, { "epoch": 0.8989792828259436, "grad_norm": 1.399744015424999, "learning_rate": 5.303510222156716e-07, "loss": 0.3775, "step": 25233 }, { "epoch": 0.8990149099524378, "grad_norm": 1.6257139062509662, "learning_rate": 5.299802826814748e-07, "loss": 0.5307, "step": 25234 }, { "epoch": 0.8990505370789319, "grad_norm": 1.2320810165526843, "learning_rate": 5.296096692472086e-07, "loss": 0.3859, "step": 25235 }, { "epoch": 0.899086164205426, "grad_norm": 1.100373065889659, "learning_rate": 5.292391819178089e-07, "loss": 0.3514, "step": 25236 }, { "epoch": 0.8991217913319202, "grad_norm": 1.4912686807443964, "learning_rate": 5.288688206982084e-07, "loss": 0.5841, "step": 25237 }, { "epoch": 0.8991574184584142, "grad_norm": 1.5121657552733834, "learning_rate": 5.284985855933422e-07, "loss": 0.48, "step": 25238 }, { "epoch": 0.8991930455849083, "grad_norm": 1.306406623056415, "learning_rate": 5.281284766081352e-07, "loss": 0.4315, "step": 25239 }, { "epoch": 0.8992286727114025, "grad_norm": 1.3209882350068516, "learning_rate": 5.2775849374752e-07, "loss": 0.3925, "step": 25240 }, { "epoch": 0.8992642998378966, "grad_norm": 1.6158314626328785, "learning_rate": 5.273886370164194e-07, "loss": 0.5392, "step": 25241 }, { "epoch": 0.8992999269643907, "grad_norm": 1.6098234317251547, "learning_rate": 5.270189064197606e-07, "loss": 0.5228, "step": 25242 }, { "epoch": 0.8993355540908848, "grad_norm": 1.229408317563881, "learning_rate": 5.266493019624664e-07, "loss": 0.4022, "step": 25243 }, { "epoch": 0.8993711812173789, "grad_norm": 1.5785416856087648, "learning_rate": 5.26279823649457e-07, "loss": 0.4735, "step": 25244 }, { "epoch": 0.899406808343873, "grad_norm": 1.2993122805415263, "learning_rate": 5.259104714856555e-07, "loss": 0.4557, "step": 25245 }, { "epoch": 0.8994424354703672, "grad_norm": 1.6426471136491727, "learning_rate": 5.255412454759767e-07, "loss": 0.5674, "step": 25246 }, { "epoch": 0.8994780625968612, "grad_norm": 1.8482205690225748, "learning_rate": 5.2517214562534e-07, "loss": 0.5296, "step": 25247 }, { "epoch": 0.8995136897233553, "grad_norm": 1.2585121731236282, "learning_rate": 5.24803171938657e-07, "loss": 0.329, "step": 25248 }, { "epoch": 0.8995493168498495, "grad_norm": 1.3568901022330864, "learning_rate": 5.244343244208439e-07, "loss": 0.4151, "step": 25249 }, { "epoch": 0.8995849439763436, "grad_norm": 1.3711404031894274, "learning_rate": 5.240656030768098e-07, "loss": 0.4866, "step": 25250 }, { "epoch": 0.8996205711028377, "grad_norm": 1.2945020655732413, "learning_rate": 5.236970079114645e-07, "loss": 0.4057, "step": 25251 }, { "epoch": 0.8996561982293318, "grad_norm": 1.3126712055739307, "learning_rate": 5.233285389297182e-07, "loss": 0.4191, "step": 25252 }, { "epoch": 0.8996918253558259, "grad_norm": 1.6458108397008633, "learning_rate": 5.229601961364761e-07, "loss": 0.5301, "step": 25253 }, { "epoch": 0.89972745248232, "grad_norm": 1.584216663608841, "learning_rate": 5.225919795366441e-07, "loss": 0.5108, "step": 25254 }, { "epoch": 0.8997630796088142, "grad_norm": 1.5831293214881064, "learning_rate": 5.222238891351228e-07, "loss": 0.505, "step": 25255 }, { "epoch": 0.8997987067353083, "grad_norm": 1.1712857093344995, "learning_rate": 5.21855924936816e-07, "loss": 0.3637, "step": 25256 }, { "epoch": 0.8998343338618023, "grad_norm": 2.2980250201127936, "learning_rate": 5.214880869466221e-07, "loss": 0.4818, "step": 25257 }, { "epoch": 0.8998699609882965, "grad_norm": 1.2071923170129304, "learning_rate": 5.211203751694382e-07, "loss": 0.4361, "step": 25258 }, { "epoch": 0.8999055881147906, "grad_norm": 1.1745084444978675, "learning_rate": 5.207527896101627e-07, "loss": 0.3867, "step": 25259 }, { "epoch": 0.8999412152412847, "grad_norm": 1.6665265898281927, "learning_rate": 5.203853302736905e-07, "loss": 0.5804, "step": 25260 }, { "epoch": 0.8999768423677789, "grad_norm": 1.6079470875145159, "learning_rate": 5.200179971649122e-07, "loss": 0.4266, "step": 25261 }, { "epoch": 0.9000124694942729, "grad_norm": 1.488418970789573, "learning_rate": 5.196507902887215e-07, "loss": 0.5444, "step": 25262 }, { "epoch": 0.900048096620767, "grad_norm": 1.6117547486107018, "learning_rate": 5.192837096500058e-07, "loss": 0.3725, "step": 25263 }, { "epoch": 0.9000837237472612, "grad_norm": 1.2952227196668045, "learning_rate": 5.189167552536534e-07, "loss": 0.3842, "step": 25264 }, { "epoch": 0.9001193508737553, "grad_norm": 1.3346625685469773, "learning_rate": 5.185499271045536e-07, "loss": 0.4221, "step": 25265 }, { "epoch": 0.9001549780002494, "grad_norm": 1.2450201954509947, "learning_rate": 5.181832252075869e-07, "loss": 0.3804, "step": 25266 }, { "epoch": 0.9001906051267435, "grad_norm": 1.5450094217297108, "learning_rate": 5.178166495676385e-07, "loss": 0.4515, "step": 25267 }, { "epoch": 0.9002262322532376, "grad_norm": 1.636380394452091, "learning_rate": 5.174502001895887e-07, "loss": 0.4646, "step": 25268 }, { "epoch": 0.9002618593797317, "grad_norm": 1.788683890407669, "learning_rate": 5.170838770783182e-07, "loss": 0.4715, "step": 25269 }, { "epoch": 0.9002974865062259, "grad_norm": 1.3384616551442041, "learning_rate": 5.16717680238703e-07, "loss": 0.4184, "step": 25270 }, { "epoch": 0.90033311363272, "grad_norm": 2.147261309777597, "learning_rate": 5.163516096756216e-07, "loss": 0.5093, "step": 25271 }, { "epoch": 0.900368740759214, "grad_norm": 1.686816581034808, "learning_rate": 5.159856653939466e-07, "loss": 0.6005, "step": 25272 }, { "epoch": 0.9004043678857082, "grad_norm": 1.2148049581212912, "learning_rate": 5.156198473985508e-07, "loss": 0.3861, "step": 25273 }, { "epoch": 0.9004399950122023, "grad_norm": 1.2414736502985186, "learning_rate": 5.152541556943058e-07, "loss": 0.394, "step": 25274 }, { "epoch": 0.9004756221386964, "grad_norm": 1.3483048169566225, "learning_rate": 5.148885902860822e-07, "loss": 0.4805, "step": 25275 }, { "epoch": 0.9005112492651905, "grad_norm": 1.2942992529492143, "learning_rate": 5.145231511787463e-07, "loss": 0.4406, "step": 25276 }, { "epoch": 0.9005468763916846, "grad_norm": 2.8249975488936863, "learning_rate": 5.141578383771639e-07, "loss": 0.4001, "step": 25277 }, { "epoch": 0.9005825035181787, "grad_norm": 1.4637816497553235, "learning_rate": 5.137926518862013e-07, "loss": 0.4169, "step": 25278 }, { "epoch": 0.9006181306446729, "grad_norm": 1.9403494009908648, "learning_rate": 5.134275917107189e-07, "loss": 0.4967, "step": 25279 }, { "epoch": 0.900653757771167, "grad_norm": 1.4103729747732565, "learning_rate": 5.130626578555809e-07, "loss": 0.4329, "step": 25280 }, { "epoch": 0.900689384897661, "grad_norm": 1.5468160810309768, "learning_rate": 5.126978503256408e-07, "loss": 0.5168, "step": 25281 }, { "epoch": 0.9007250120241552, "grad_norm": 1.9103753230156597, "learning_rate": 5.123331691257627e-07, "loss": 0.6292, "step": 25282 }, { "epoch": 0.9007606391506493, "grad_norm": 1.1575018338521288, "learning_rate": 5.119686142607993e-07, "loss": 0.3629, "step": 25283 }, { "epoch": 0.9007962662771434, "grad_norm": 1.3871840984375299, "learning_rate": 5.116041857356069e-07, "loss": 0.4219, "step": 25284 }, { "epoch": 0.9008318934036376, "grad_norm": 1.7235139918836553, "learning_rate": 5.112398835550348e-07, "loss": 0.4312, "step": 25285 }, { "epoch": 0.9008675205301316, "grad_norm": 1.8314883487776024, "learning_rate": 5.108757077239379e-07, "loss": 0.5337, "step": 25286 }, { "epoch": 0.9009031476566257, "grad_norm": 1.3443690171087035, "learning_rate": 5.105116582471614e-07, "loss": 0.4829, "step": 25287 }, { "epoch": 0.9009387747831199, "grad_norm": 1.7376535212787647, "learning_rate": 5.101477351295558e-07, "loss": 0.5164, "step": 25288 }, { "epoch": 0.900974401909614, "grad_norm": 1.4088181080231768, "learning_rate": 5.09783938375965e-07, "loss": 0.4557, "step": 25289 }, { "epoch": 0.901010029036108, "grad_norm": 1.6557076280576728, "learning_rate": 5.094202679912352e-07, "loss": 0.4888, "step": 25290 }, { "epoch": 0.9010456561626022, "grad_norm": 1.7939682683706244, "learning_rate": 5.090567239802091e-07, "loss": 0.5144, "step": 25291 }, { "epoch": 0.9010812832890963, "grad_norm": 1.1492173112058728, "learning_rate": 5.086933063477251e-07, "loss": 0.3576, "step": 25292 }, { "epoch": 0.9011169104155904, "grad_norm": 1.182087214232762, "learning_rate": 5.083300150986259e-07, "loss": 0.3722, "step": 25293 }, { "epoch": 0.9011525375420846, "grad_norm": 1.0864113780354407, "learning_rate": 5.079668502377444e-07, "loss": 0.3618, "step": 25294 }, { "epoch": 0.9011881646685787, "grad_norm": 1.4410094732073457, "learning_rate": 5.076038117699189e-07, "loss": 0.5262, "step": 25295 }, { "epoch": 0.9012237917950727, "grad_norm": 1.2202128955253246, "learning_rate": 5.072408996999844e-07, "loss": 0.3835, "step": 25296 }, { "epoch": 0.9012594189215669, "grad_norm": 1.2372872335020644, "learning_rate": 5.068781140327727e-07, "loss": 0.4851, "step": 25297 }, { "epoch": 0.901295046048061, "grad_norm": 1.7759758510960326, "learning_rate": 5.065154547731132e-07, "loss": 0.5517, "step": 25298 }, { "epoch": 0.9013306731745551, "grad_norm": 1.4213028114017858, "learning_rate": 5.061529219258366e-07, "loss": 0.5773, "step": 25299 }, { "epoch": 0.9013663003010493, "grad_norm": 1.1159153630885694, "learning_rate": 5.057905154957699e-07, "loss": 0.3858, "step": 25300 }, { "epoch": 0.9014019274275433, "grad_norm": 1.2860410841374461, "learning_rate": 5.054282354877382e-07, "loss": 0.435, "step": 25301 }, { "epoch": 0.9014375545540374, "grad_norm": 1.5899388514495003, "learning_rate": 5.050660819065667e-07, "loss": 0.5304, "step": 25302 }, { "epoch": 0.9014731816805316, "grad_norm": 1.569843677660047, "learning_rate": 5.047040547570748e-07, "loss": 0.556, "step": 25303 }, { "epoch": 0.9015088088070257, "grad_norm": 1.5207013032759868, "learning_rate": 5.043421540440863e-07, "loss": 0.5175, "step": 25304 }, { "epoch": 0.9015444359335197, "grad_norm": 1.2465924206756407, "learning_rate": 5.039803797724185e-07, "loss": 0.4371, "step": 25305 }, { "epoch": 0.9015800630600139, "grad_norm": 1.2166842786618766, "learning_rate": 5.03618731946891e-07, "loss": 0.4063, "step": 25306 }, { "epoch": 0.901615690186508, "grad_norm": 1.3097810364004263, "learning_rate": 5.032572105723166e-07, "loss": 0.4193, "step": 25307 }, { "epoch": 0.9016513173130021, "grad_norm": 1.3640280251661059, "learning_rate": 5.028958156535113e-07, "loss": 0.4072, "step": 25308 }, { "epoch": 0.9016869444394963, "grad_norm": 1.5281490620254243, "learning_rate": 5.025345471952847e-07, "loss": 0.5194, "step": 25309 }, { "epoch": 0.9017225715659903, "grad_norm": 1.4411392094493987, "learning_rate": 5.021734052024485e-07, "loss": 0.4211, "step": 25310 }, { "epoch": 0.9017581986924844, "grad_norm": 1.1977288828110662, "learning_rate": 5.018123896798132e-07, "loss": 0.3825, "step": 25311 }, { "epoch": 0.9017938258189786, "grad_norm": 1.5217229685507536, "learning_rate": 5.01451500632184e-07, "loss": 0.5111, "step": 25312 }, { "epoch": 0.9018294529454727, "grad_norm": 1.2360902982215092, "learning_rate": 5.010907380643682e-07, "loss": 0.3837, "step": 25313 }, { "epoch": 0.9018650800719668, "grad_norm": 1.2035390207547425, "learning_rate": 5.007301019811683e-07, "loss": 0.3946, "step": 25314 }, { "epoch": 0.9019007071984609, "grad_norm": 1.281184741606341, "learning_rate": 5.003695923873875e-07, "loss": 0.4279, "step": 25315 }, { "epoch": 0.901936334324955, "grad_norm": 1.676233905829993, "learning_rate": 5.000092092878228e-07, "loss": 0.5211, "step": 25316 }, { "epoch": 0.9019719614514491, "grad_norm": 1.2221723652018308, "learning_rate": 4.996489526872783e-07, "loss": 0.4277, "step": 25317 }, { "epoch": 0.9020075885779433, "grad_norm": 1.278748634329963, "learning_rate": 4.992888225905467e-07, "loss": 0.3881, "step": 25318 }, { "epoch": 0.9020432157044374, "grad_norm": 1.194764933770725, "learning_rate": 4.989288190024255e-07, "loss": 0.4096, "step": 25319 }, { "epoch": 0.9020788428309314, "grad_norm": 1.5583719940656475, "learning_rate": 4.985689419277073e-07, "loss": 0.4847, "step": 25320 }, { "epoch": 0.9021144699574256, "grad_norm": 1.1177079094403084, "learning_rate": 4.982091913711862e-07, "loss": 0.3765, "step": 25321 }, { "epoch": 0.9021500970839197, "grad_norm": 1.7405966894110634, "learning_rate": 4.978495673376505e-07, "loss": 0.5159, "step": 25322 }, { "epoch": 0.9021857242104138, "grad_norm": 1.1462250239550769, "learning_rate": 4.974900698318885e-07, "loss": 0.3639, "step": 25323 }, { "epoch": 0.902221351336908, "grad_norm": 1.3168136952596563, "learning_rate": 4.9713069885869e-07, "loss": 0.4005, "step": 25324 }, { "epoch": 0.902256978463402, "grad_norm": 1.2174228880173998, "learning_rate": 4.967714544228364e-07, "loss": 0.3925, "step": 25325 }, { "epoch": 0.9022926055898961, "grad_norm": 1.0710717267585461, "learning_rate": 4.964123365291129e-07, "loss": 0.3662, "step": 25326 }, { "epoch": 0.9023282327163903, "grad_norm": 1.6770970821461606, "learning_rate": 4.960533451823024e-07, "loss": 0.4717, "step": 25327 }, { "epoch": 0.9023638598428844, "grad_norm": 1.128145957928844, "learning_rate": 4.956944803871866e-07, "loss": 0.4008, "step": 25328 }, { "epoch": 0.9023994869693784, "grad_norm": 1.2424117971064268, "learning_rate": 4.953357421485394e-07, "loss": 0.4005, "step": 25329 }, { "epoch": 0.9024351140958726, "grad_norm": 1.2376008982144189, "learning_rate": 4.949771304711415e-07, "loss": 0.4534, "step": 25330 }, { "epoch": 0.9024707412223667, "grad_norm": 1.2339778111509, "learning_rate": 4.946186453597657e-07, "loss": 0.4086, "step": 25331 }, { "epoch": 0.9025063683488608, "grad_norm": 1.2919835662600831, "learning_rate": 4.942602868191881e-07, "loss": 0.3993, "step": 25332 }, { "epoch": 0.902541995475355, "grad_norm": 1.5852889011072215, "learning_rate": 4.939020548541762e-07, "loss": 0.5189, "step": 25333 }, { "epoch": 0.902577622601849, "grad_norm": 1.1784002028266094, "learning_rate": 4.935439494695038e-07, "loss": 0.4078, "step": 25334 }, { "epoch": 0.9026132497283431, "grad_norm": 1.3729907288497496, "learning_rate": 4.931859706699394e-07, "loss": 0.4057, "step": 25335 }, { "epoch": 0.9026488768548373, "grad_norm": 1.5160156866032153, "learning_rate": 4.92828118460248e-07, "loss": 0.4957, "step": 25336 }, { "epoch": 0.9026845039813314, "grad_norm": 1.2952812611257083, "learning_rate": 4.924703928451968e-07, "loss": 0.3992, "step": 25337 }, { "epoch": 0.9027201311078255, "grad_norm": 1.2701867087709358, "learning_rate": 4.921127938295456e-07, "loss": 0.4283, "step": 25338 }, { "epoch": 0.9027557582343196, "grad_norm": 1.6175271781604559, "learning_rate": 4.917553214180603e-07, "loss": 0.4483, "step": 25339 }, { "epoch": 0.9027913853608137, "grad_norm": 1.5144519619254093, "learning_rate": 4.913979756154963e-07, "loss": 0.5315, "step": 25340 }, { "epoch": 0.9028270124873078, "grad_norm": 1.2379289726337521, "learning_rate": 4.910407564266151e-07, "loss": 0.3629, "step": 25341 }, { "epoch": 0.902862639613802, "grad_norm": 1.3205397025238925, "learning_rate": 4.906836638561729e-07, "loss": 0.4632, "step": 25342 }, { "epoch": 0.902898266740296, "grad_norm": 1.2860349798838253, "learning_rate": 4.90326697908925e-07, "loss": 0.4084, "step": 25343 }, { "epoch": 0.9029338938667901, "grad_norm": 2.345847206488401, "learning_rate": 4.899698585896229e-07, "loss": 0.4214, "step": 25344 }, { "epoch": 0.9029695209932843, "grad_norm": 1.4518679698263328, "learning_rate": 4.896131459030196e-07, "loss": 0.507, "step": 25345 }, { "epoch": 0.9030051481197784, "grad_norm": 1.3789176592975583, "learning_rate": 4.892565598538656e-07, "loss": 0.4198, "step": 25346 }, { "epoch": 0.9030407752462725, "grad_norm": 1.5051287016747203, "learning_rate": 4.889001004469063e-07, "loss": 0.4849, "step": 25347 }, { "epoch": 0.9030764023727667, "grad_norm": 1.539237373286826, "learning_rate": 4.885437676868909e-07, "loss": 0.5466, "step": 25348 }, { "epoch": 0.9031120294992607, "grad_norm": 1.364050930694997, "learning_rate": 4.881875615785637e-07, "loss": 0.4142, "step": 25349 }, { "epoch": 0.9031476566257548, "grad_norm": 1.56838473924537, "learning_rate": 4.878314821266672e-07, "loss": 0.4401, "step": 25350 }, { "epoch": 0.903183283752249, "grad_norm": 1.245592601831266, "learning_rate": 4.874755293359435e-07, "loss": 0.3946, "step": 25351 }, { "epoch": 0.9032189108787431, "grad_norm": 1.6906559380121764, "learning_rate": 4.871197032111319e-07, "loss": 0.5091, "step": 25352 }, { "epoch": 0.9032545380052371, "grad_norm": 1.6058958424243834, "learning_rate": 4.867640037569699e-07, "loss": 0.5075, "step": 25353 }, { "epoch": 0.9032901651317313, "grad_norm": 1.3899447316553928, "learning_rate": 4.864084309781958e-07, "loss": 0.3882, "step": 25354 }, { "epoch": 0.9033257922582254, "grad_norm": 1.765631462699297, "learning_rate": 4.860529848795414e-07, "loss": 0.4161, "step": 25355 }, { "epoch": 0.9033614193847195, "grad_norm": 1.3773753642432607, "learning_rate": 4.856976654657408e-07, "loss": 0.4481, "step": 25356 }, { "epoch": 0.9033970465112137, "grad_norm": 1.2434071480370328, "learning_rate": 4.85342472741528e-07, "loss": 0.4016, "step": 25357 }, { "epoch": 0.9034326736377077, "grad_norm": 1.5031139016222292, "learning_rate": 4.849874067116289e-07, "loss": 0.5178, "step": 25358 }, { "epoch": 0.9034683007642018, "grad_norm": 1.335177537973518, "learning_rate": 4.846324673807756e-07, "loss": 0.4471, "step": 25359 }, { "epoch": 0.903503927890696, "grad_norm": 1.2036729555826924, "learning_rate": 4.842776547536898e-07, "loss": 0.3365, "step": 25360 }, { "epoch": 0.9035395550171901, "grad_norm": 1.6442614437404754, "learning_rate": 4.839229688350988e-07, "loss": 0.5729, "step": 25361 }, { "epoch": 0.9035751821436842, "grad_norm": 1.180940889193636, "learning_rate": 4.835684096297244e-07, "loss": 0.4092, "step": 25362 }, { "epoch": 0.9036108092701783, "grad_norm": 1.5834757215198803, "learning_rate": 4.832139771422872e-07, "loss": 0.4693, "step": 25363 }, { "epoch": 0.9036464363966724, "grad_norm": 1.6035185364918494, "learning_rate": 4.828596713775091e-07, "loss": 0.5421, "step": 25364 }, { "epoch": 0.9036820635231665, "grad_norm": 1.3206572374127514, "learning_rate": 4.825054923401074e-07, "loss": 0.3496, "step": 25365 }, { "epoch": 0.9037176906496607, "grad_norm": 2.56416899524613, "learning_rate": 4.821514400347949e-07, "loss": 0.5135, "step": 25366 }, { "epoch": 0.9037533177761548, "grad_norm": 1.3183187239151006, "learning_rate": 4.817975144662912e-07, "loss": 0.3696, "step": 25367 }, { "epoch": 0.9037889449026488, "grad_norm": 1.340829864607116, "learning_rate": 4.814437156393048e-07, "loss": 0.3829, "step": 25368 }, { "epoch": 0.903824572029143, "grad_norm": 1.6216972903811564, "learning_rate": 4.810900435585486e-07, "loss": 0.5144, "step": 25369 }, { "epoch": 0.9038601991556371, "grad_norm": 1.75327643164455, "learning_rate": 4.80736498228731e-07, "loss": 0.5122, "step": 25370 }, { "epoch": 0.9038958262821312, "grad_norm": 1.510019738543672, "learning_rate": 4.803830796545605e-07, "loss": 0.3954, "step": 25371 }, { "epoch": 0.9039314534086254, "grad_norm": 1.4603029717043146, "learning_rate": 4.800297878407456e-07, "loss": 0.5029, "step": 25372 }, { "epoch": 0.9039670805351194, "grad_norm": 1.2844412980237847, "learning_rate": 4.796766227919858e-07, "loss": 0.4006, "step": 25373 }, { "epoch": 0.9040027076616135, "grad_norm": 1.2947277871207126, "learning_rate": 4.793235845129885e-07, "loss": 0.4224, "step": 25374 }, { "epoch": 0.9040383347881077, "grad_norm": 1.5683665336604837, "learning_rate": 4.789706730084498e-07, "loss": 0.5037, "step": 25375 }, { "epoch": 0.9040739619146018, "grad_norm": 1.3314421078788767, "learning_rate": 4.786178882830727e-07, "loss": 0.4109, "step": 25376 }, { "epoch": 0.9041095890410958, "grad_norm": 1.9843887284523607, "learning_rate": 4.782652303415525e-07, "loss": 0.5253, "step": 25377 }, { "epoch": 0.90414521616759, "grad_norm": 1.2141743732732238, "learning_rate": 4.779126991885863e-07, "loss": 0.3668, "step": 25378 }, { "epoch": 0.9041808432940841, "grad_norm": 1.6742523186263503, "learning_rate": 4.775602948288683e-07, "loss": 0.481, "step": 25379 }, { "epoch": 0.9042164704205782, "grad_norm": 1.22210464475455, "learning_rate": 4.772080172670901e-07, "loss": 0.4048, "step": 25380 }, { "epoch": 0.9042520975470724, "grad_norm": 1.2494847853298805, "learning_rate": 4.768558665079448e-07, "loss": 0.3805, "step": 25381 }, { "epoch": 0.9042877246735664, "grad_norm": 1.4872918778607476, "learning_rate": 4.765038425561186e-07, "loss": 0.4826, "step": 25382 }, { "epoch": 0.9043233518000605, "grad_norm": 1.626697720516805, "learning_rate": 4.761519454163011e-07, "loss": 0.4734, "step": 25383 }, { "epoch": 0.9043589789265547, "grad_norm": 1.2829925193726153, "learning_rate": 4.7580017509317623e-07, "loss": 0.3892, "step": 25384 }, { "epoch": 0.9043946060530488, "grad_norm": 1.674234902506027, "learning_rate": 4.754485315914292e-07, "loss": 0.5227, "step": 25385 }, { "epoch": 0.9044302331795429, "grad_norm": 1.272406882071732, "learning_rate": 4.750970149157419e-07, "loss": 0.3812, "step": 25386 }, { "epoch": 0.904465860306037, "grad_norm": 1.48150288375767, "learning_rate": 4.7474562507079713e-07, "loss": 0.4612, "step": 25387 }, { "epoch": 0.9045014874325311, "grad_norm": 1.5512924910743517, "learning_rate": 4.7439436206127007e-07, "loss": 0.5301, "step": 25388 }, { "epoch": 0.9045371145590252, "grad_norm": 1.6617090874007359, "learning_rate": 4.740432258918415e-07, "loss": 0.4603, "step": 25389 }, { "epoch": 0.9045727416855194, "grad_norm": 1.7933912958731169, "learning_rate": 4.736922165671853e-07, "loss": 0.571, "step": 25390 }, { "epoch": 0.9046083688120135, "grad_norm": 1.22151546532834, "learning_rate": 4.733413340919746e-07, "loss": 0.4169, "step": 25391 }, { "epoch": 0.9046439959385075, "grad_norm": 1.6468430990188614, "learning_rate": 4.7299057847088545e-07, "loss": 0.5733, "step": 25392 }, { "epoch": 0.9046796230650017, "grad_norm": 1.3779301073449786, "learning_rate": 4.726399497085832e-07, "loss": 0.409, "step": 25393 }, { "epoch": 0.9047152501914958, "grad_norm": 1.2710198178407344, "learning_rate": 4.722894478097395e-07, "loss": 0.3946, "step": 25394 }, { "epoch": 0.9047508773179899, "grad_norm": 1.309940297783278, "learning_rate": 4.7193907277902186e-07, "loss": 0.4043, "step": 25395 }, { "epoch": 0.904786504444484, "grad_norm": 1.1527193349938976, "learning_rate": 4.7158882462109535e-07, "loss": 0.4114, "step": 25396 }, { "epoch": 0.9048221315709781, "grad_norm": 1.3180005986912753, "learning_rate": 4.7123870334062295e-07, "loss": 0.3746, "step": 25397 }, { "epoch": 0.9048577586974722, "grad_norm": 1.479773710018136, "learning_rate": 4.708887089422687e-07, "loss": 0.4826, "step": 25398 }, { "epoch": 0.9048933858239664, "grad_norm": 1.533686488680284, "learning_rate": 4.7053884143069106e-07, "loss": 0.5319, "step": 25399 }, { "epoch": 0.9049290129504605, "grad_norm": 1.2039045202720764, "learning_rate": 4.7018910081054856e-07, "loss": 0.3771, "step": 25400 }, { "epoch": 0.9049646400769545, "grad_norm": 1.563720323163918, "learning_rate": 4.698394870864997e-07, "loss": 0.5211, "step": 25401 }, { "epoch": 0.9050002672034487, "grad_norm": 1.5902359625134703, "learning_rate": 4.6949000026319966e-07, "loss": 0.4895, "step": 25402 }, { "epoch": 0.9050358943299428, "grad_norm": 1.3398072357669535, "learning_rate": 4.6914064034530137e-07, "loss": 0.4857, "step": 25403 }, { "epoch": 0.9050715214564369, "grad_norm": 1.2849038405666982, "learning_rate": 4.6879140733745663e-07, "loss": 0.4275, "step": 25404 }, { "epoch": 0.9051071485829311, "grad_norm": 1.2906308897690995, "learning_rate": 4.6844230124431735e-07, "loss": 0.4017, "step": 25405 }, { "epoch": 0.9051427757094251, "grad_norm": 1.2553677866586588, "learning_rate": 4.6809332207053083e-07, "loss": 0.3827, "step": 25406 }, { "epoch": 0.9051784028359192, "grad_norm": 1.4959602793374431, "learning_rate": 4.6774446982074453e-07, "loss": 0.3984, "step": 25407 }, { "epoch": 0.9052140299624134, "grad_norm": 1.5111237057552864, "learning_rate": 4.6739574449960024e-07, "loss": 0.4932, "step": 25408 }, { "epoch": 0.9052496570889075, "grad_norm": 1.485698737259755, "learning_rate": 4.6704714611174764e-07, "loss": 0.4522, "step": 25409 }, { "epoch": 0.9052852842154016, "grad_norm": 1.760390463192605, "learning_rate": 4.666986746618241e-07, "loss": 0.537, "step": 25410 }, { "epoch": 0.9053209113418957, "grad_norm": 1.6056639333453748, "learning_rate": 4.663503301544725e-07, "loss": 0.496, "step": 25411 }, { "epoch": 0.9053565384683898, "grad_norm": 1.5909894095181527, "learning_rate": 4.6600211259432813e-07, "loss": 0.4314, "step": 25412 }, { "epoch": 0.9053921655948839, "grad_norm": 1.467043479547936, "learning_rate": 4.656540219860317e-07, "loss": 0.5357, "step": 25413 }, { "epoch": 0.9054277927213781, "grad_norm": 1.5007134202002055, "learning_rate": 4.653060583342139e-07, "loss": 0.5565, "step": 25414 }, { "epoch": 0.9054634198478722, "grad_norm": 1.310697900920313, "learning_rate": 4.6495822164351e-07, "loss": 0.56, "step": 25415 }, { "epoch": 0.9054990469743662, "grad_norm": 1.224819579904685, "learning_rate": 4.646105119185529e-07, "loss": 0.4005, "step": 25416 }, { "epoch": 0.9055346741008604, "grad_norm": 1.3624108738004892, "learning_rate": 4.642629291639711e-07, "loss": 0.4207, "step": 25417 }, { "epoch": 0.9055703012273545, "grad_norm": 1.2647281064342568, "learning_rate": 4.6391547338439533e-07, "loss": 0.3771, "step": 25418 }, { "epoch": 0.9056059283538486, "grad_norm": 1.5379280508468605, "learning_rate": 4.6356814458444866e-07, "loss": 0.4915, "step": 25419 }, { "epoch": 0.9056415554803428, "grad_norm": 1.5099023171749324, "learning_rate": 4.632209427687595e-07, "loss": 0.5122, "step": 25420 }, { "epoch": 0.9056771826068368, "grad_norm": 1.4353231218028093, "learning_rate": 4.6287386794194756e-07, "loss": 0.4101, "step": 25421 }, { "epoch": 0.9057128097333309, "grad_norm": 1.6470375106209234, "learning_rate": 4.6252692010863573e-07, "loss": 0.5269, "step": 25422 }, { "epoch": 0.9057484368598251, "grad_norm": 1.6049653638748695, "learning_rate": 4.6218009927344486e-07, "loss": 0.488, "step": 25423 }, { "epoch": 0.9057840639863192, "grad_norm": 1.4191849018112346, "learning_rate": 4.6183340544099343e-07, "loss": 0.408, "step": 25424 }, { "epoch": 0.9058196911128132, "grad_norm": 2.9140957717849485, "learning_rate": 4.614868386158955e-07, "loss": 0.4248, "step": 25425 }, { "epoch": 0.9058553182393074, "grad_norm": 1.2705161593708185, "learning_rate": 4.611403988027674e-07, "loss": 0.4036, "step": 25426 }, { "epoch": 0.9058909453658015, "grad_norm": 1.2790208253153164, "learning_rate": 4.6079408600622435e-07, "loss": 0.4633, "step": 25427 }, { "epoch": 0.9059265724922956, "grad_norm": 1.37483023918265, "learning_rate": 4.6044790023087373e-07, "loss": 0.3946, "step": 25428 }, { "epoch": 0.9059621996187898, "grad_norm": 1.4205782092193247, "learning_rate": 4.6010184148132854e-07, "loss": 0.4204, "step": 25429 }, { "epoch": 0.9059978267452838, "grad_norm": 1.5606436092342502, "learning_rate": 4.597559097621929e-07, "loss": 0.4352, "step": 25430 }, { "epoch": 0.9060334538717779, "grad_norm": 1.5744090785475593, "learning_rate": 4.594101050780764e-07, "loss": 0.5051, "step": 25431 }, { "epoch": 0.9060690809982721, "grad_norm": 1.4700032219659251, "learning_rate": 4.5906442743358314e-07, "loss": 0.4303, "step": 25432 }, { "epoch": 0.9061047081247662, "grad_norm": 1.3806872212813093, "learning_rate": 4.587188768333173e-07, "loss": 0.5187, "step": 25433 }, { "epoch": 0.9061403352512604, "grad_norm": 1.1381305765025034, "learning_rate": 4.5837345328187623e-07, "loss": 0.3642, "step": 25434 }, { "epoch": 0.9061759623777544, "grad_norm": 1.209087962662271, "learning_rate": 4.5802815678386405e-07, "loss": 0.3398, "step": 25435 }, { "epoch": 0.9062115895042485, "grad_norm": 1.5660790105376399, "learning_rate": 4.576829873438737e-07, "loss": 0.4057, "step": 25436 }, { "epoch": 0.9062472166307427, "grad_norm": 1.523618471621985, "learning_rate": 4.573379449665061e-07, "loss": 0.5694, "step": 25437 }, { "epoch": 0.9062828437572368, "grad_norm": 1.5119530740753622, "learning_rate": 4.5699302965635185e-07, "loss": 0.5299, "step": 25438 }, { "epoch": 0.9063184708837309, "grad_norm": 1.069565617988466, "learning_rate": 4.5664824141800623e-07, "loss": 0.3816, "step": 25439 }, { "epoch": 0.906354098010225, "grad_norm": 1.1642470734858361, "learning_rate": 4.563035802560611e-07, "loss": 0.3747, "step": 25440 }, { "epoch": 0.9063897251367191, "grad_norm": 1.6189042994057334, "learning_rate": 4.559590461751029e-07, "loss": 0.4907, "step": 25441 }, { "epoch": 0.9064253522632132, "grad_norm": 1.3507716001108032, "learning_rate": 4.556146391797234e-07, "loss": 0.4833, "step": 25442 }, { "epoch": 0.9064609793897074, "grad_norm": 1.6747130199405333, "learning_rate": 4.5527035927450337e-07, "loss": 0.5787, "step": 25443 }, { "epoch": 0.9064966065162015, "grad_norm": 1.4099494093601788, "learning_rate": 4.549262064640314e-07, "loss": 0.5156, "step": 25444 }, { "epoch": 0.9065322336426955, "grad_norm": 1.1129570045206199, "learning_rate": 4.545821807528861e-07, "loss": 0.3961, "step": 25445 }, { "epoch": 0.9065678607691897, "grad_norm": 1.6010672246453044, "learning_rate": 4.542382821456537e-07, "loss": 0.4939, "step": 25446 }, { "epoch": 0.9066034878956838, "grad_norm": 1.2850161608516604, "learning_rate": 4.538945106469095e-07, "loss": 0.3752, "step": 25447 }, { "epoch": 0.9066391150221779, "grad_norm": 1.5428001252293195, "learning_rate": 4.535508662612331e-07, "loss": 0.4704, "step": 25448 }, { "epoch": 0.906674742148672, "grad_norm": 1.882108510292843, "learning_rate": 4.5320734899319873e-07, "loss": 0.4917, "step": 25449 }, { "epoch": 0.9067103692751661, "grad_norm": 1.4666245249629428, "learning_rate": 4.5286395884738045e-07, "loss": 0.5082, "step": 25450 }, { "epoch": 0.9067459964016602, "grad_norm": 1.1927146912282296, "learning_rate": 4.5252069582835345e-07, "loss": 0.379, "step": 25451 }, { "epoch": 0.9067816235281544, "grad_norm": 1.2735025271991347, "learning_rate": 4.5217755994068525e-07, "loss": 0.3801, "step": 25452 }, { "epoch": 0.9068172506546485, "grad_norm": 1.7508087226268507, "learning_rate": 4.518345511889466e-07, "loss": 0.5687, "step": 25453 }, { "epoch": 0.9068528777811425, "grad_norm": 1.7183444269356725, "learning_rate": 4.5149166957770495e-07, "loss": 0.4782, "step": 25454 }, { "epoch": 0.9068885049076367, "grad_norm": 1.6698976719225351, "learning_rate": 4.511489151115267e-07, "loss": 0.501, "step": 25455 }, { "epoch": 0.9069241320341308, "grad_norm": 1.4099277392501142, "learning_rate": 4.508062877949737e-07, "loss": 0.3753, "step": 25456 }, { "epoch": 0.9069597591606249, "grad_norm": 1.5256264817376821, "learning_rate": 4.504637876326101e-07, "loss": 0.4603, "step": 25457 }, { "epoch": 0.9069953862871191, "grad_norm": 1.2610624881467138, "learning_rate": 4.501214146289956e-07, "loss": 0.3898, "step": 25458 }, { "epoch": 0.9070310134136131, "grad_norm": 1.4569478836181022, "learning_rate": 4.497791687886888e-07, "loss": 0.4664, "step": 25459 }, { "epoch": 0.9070666405401072, "grad_norm": 1.1665831032036236, "learning_rate": 4.4943705011624704e-07, "loss": 0.3715, "step": 25460 }, { "epoch": 0.9071022676666014, "grad_norm": 1.222350545836245, "learning_rate": 4.4909505861622794e-07, "loss": 0.3987, "step": 25461 }, { "epoch": 0.9071378947930955, "grad_norm": 1.9807519545708452, "learning_rate": 4.4875319429318334e-07, "loss": 0.5472, "step": 25462 }, { "epoch": 0.9071735219195896, "grad_norm": 1.291949571048206, "learning_rate": 4.4841145715166625e-07, "loss": 0.4051, "step": 25463 }, { "epoch": 0.9072091490460837, "grad_norm": 1.4908257031168703, "learning_rate": 4.480698471962264e-07, "loss": 0.5091, "step": 25464 }, { "epoch": 0.9072447761725778, "grad_norm": 1.6158793217106187, "learning_rate": 4.477283644314134e-07, "loss": 0.57, "step": 25465 }, { "epoch": 0.9072804032990719, "grad_norm": 1.3490294406871495, "learning_rate": 4.473870088617738e-07, "loss": 0.4098, "step": 25466 }, { "epoch": 0.9073160304255661, "grad_norm": 1.447126127141214, "learning_rate": 4.4704578049185267e-07, "loss": 0.4868, "step": 25467 }, { "epoch": 0.9073516575520602, "grad_norm": 1.1551986984826754, "learning_rate": 4.467046793261931e-07, "loss": 0.3861, "step": 25468 }, { "epoch": 0.9073872846785542, "grad_norm": 1.4878363037418207, "learning_rate": 4.4636370536933927e-07, "loss": 0.4867, "step": 25469 }, { "epoch": 0.9074229118050484, "grad_norm": 1.1377541381334175, "learning_rate": 4.4602285862583083e-07, "loss": 0.3469, "step": 25470 }, { "epoch": 0.9074585389315425, "grad_norm": 1.1967858408918677, "learning_rate": 4.456821391002042e-07, "loss": 0.3835, "step": 25471 }, { "epoch": 0.9074941660580366, "grad_norm": 1.9984987968537655, "learning_rate": 4.45341546796999e-07, "loss": 0.421, "step": 25472 }, { "epoch": 0.9075297931845308, "grad_norm": 1.143689509679015, "learning_rate": 4.450010817207495e-07, "loss": 0.3545, "step": 25473 }, { "epoch": 0.9075654203110248, "grad_norm": 1.372250350966317, "learning_rate": 4.4466074387598867e-07, "loss": 0.4026, "step": 25474 }, { "epoch": 0.9076010474375189, "grad_norm": 1.4982498841609984, "learning_rate": 4.4432053326724845e-07, "loss": 0.5315, "step": 25475 }, { "epoch": 0.9076366745640131, "grad_norm": 1.4602893366106344, "learning_rate": 4.4398044989905964e-07, "loss": 0.5137, "step": 25476 }, { "epoch": 0.9076723016905072, "grad_norm": 1.582072752524766, "learning_rate": 4.4364049377595196e-07, "loss": 0.4047, "step": 25477 }, { "epoch": 0.9077079288170012, "grad_norm": 1.6358584071721978, "learning_rate": 4.433006649024485e-07, "loss": 0.575, "step": 25478 }, { "epoch": 0.9077435559434954, "grad_norm": 1.8571290457221679, "learning_rate": 4.4296096328307893e-07, "loss": 0.5025, "step": 25479 }, { "epoch": 0.9077791830699895, "grad_norm": 1.5661151250874346, "learning_rate": 4.4262138892236185e-07, "loss": 0.491, "step": 25480 }, { "epoch": 0.9078148101964836, "grad_norm": 1.4384416720863062, "learning_rate": 4.4228194182482254e-07, "loss": 0.514, "step": 25481 }, { "epoch": 0.9078504373229778, "grad_norm": 1.1514876218772987, "learning_rate": 4.419426219949785e-07, "loss": 0.3782, "step": 25482 }, { "epoch": 0.9078860644494718, "grad_norm": 1.5573005444339119, "learning_rate": 4.4160342943734723e-07, "loss": 0.567, "step": 25483 }, { "epoch": 0.9079216915759659, "grad_norm": 1.2057255196433174, "learning_rate": 4.4126436415645066e-07, "loss": 0.3746, "step": 25484 }, { "epoch": 0.9079573187024601, "grad_norm": 1.5153430249000561, "learning_rate": 4.4092542615679855e-07, "loss": 0.4505, "step": 25485 }, { "epoch": 0.9079929458289542, "grad_norm": 1.2029395335829842, "learning_rate": 4.405866154429073e-07, "loss": 0.3379, "step": 25486 }, { "epoch": 0.9080285729554483, "grad_norm": 1.2245984682035587, "learning_rate": 4.402479320192854e-07, "loss": 0.3589, "step": 25487 }, { "epoch": 0.9080642000819424, "grad_norm": 1.6228599820558427, "learning_rate": 4.39909375890446e-07, "loss": 0.5189, "step": 25488 }, { "epoch": 0.9080998272084365, "grad_norm": 1.5325341787416853, "learning_rate": 4.395709470608933e-07, "loss": 0.5152, "step": 25489 }, { "epoch": 0.9081354543349306, "grad_norm": 1.5924939394675197, "learning_rate": 4.392326455351359e-07, "loss": 0.5751, "step": 25490 }, { "epoch": 0.9081710814614248, "grad_norm": 1.337128670602344, "learning_rate": 4.38894471317679e-07, "loss": 0.4294, "step": 25491 }, { "epoch": 0.9082067085879189, "grad_norm": 1.4823732603843562, "learning_rate": 4.3855642441302584e-07, "loss": 0.4789, "step": 25492 }, { "epoch": 0.9082423357144129, "grad_norm": 1.3881664423263504, "learning_rate": 4.3821850482567595e-07, "loss": 0.3994, "step": 25493 }, { "epoch": 0.9082779628409071, "grad_norm": 1.3517145977602796, "learning_rate": 4.3788071256013033e-07, "loss": 0.4381, "step": 25494 }, { "epoch": 0.9083135899674012, "grad_norm": 1.3700728636993227, "learning_rate": 4.375430476208864e-07, "loss": 0.3963, "step": 25495 }, { "epoch": 0.9083492170938953, "grad_norm": 1.6040638180379496, "learning_rate": 4.3720551001243947e-07, "loss": 0.5917, "step": 25496 }, { "epoch": 0.9083848442203895, "grad_norm": 1.4556290529837301, "learning_rate": 4.36868099739286e-07, "loss": 0.5099, "step": 25497 }, { "epoch": 0.9084204713468835, "grad_norm": 1.2702896698807011, "learning_rate": 4.365308168059179e-07, "loss": 0.4141, "step": 25498 }, { "epoch": 0.9084560984733776, "grad_norm": 1.9208845605248106, "learning_rate": 4.361936612168272e-07, "loss": 0.5688, "step": 25499 }, { "epoch": 0.9084917255998718, "grad_norm": 1.2800820796653218, "learning_rate": 4.3585663297650236e-07, "loss": 0.3827, "step": 25500 }, { "epoch": 0.9085273527263659, "grad_norm": 1.2752964915001346, "learning_rate": 4.3551973208943223e-07, "loss": 0.4114, "step": 25501 }, { "epoch": 0.9085629798528599, "grad_norm": 1.2820375158223563, "learning_rate": 4.3518295856009974e-07, "loss": 0.4038, "step": 25502 }, { "epoch": 0.9085986069793541, "grad_norm": 1.5894618464948498, "learning_rate": 4.3484631239299356e-07, "loss": 0.4968, "step": 25503 }, { "epoch": 0.9086342341058482, "grad_norm": 1.2794940870608613, "learning_rate": 4.345097935925935e-07, "loss": 0.3943, "step": 25504 }, { "epoch": 0.9086698612323423, "grad_norm": 1.1918584119339342, "learning_rate": 4.3417340216338147e-07, "loss": 0.392, "step": 25505 }, { "epoch": 0.9087054883588365, "grad_norm": 1.6351953692391867, "learning_rate": 4.3383713810983605e-07, "loss": 0.504, "step": 25506 }, { "epoch": 0.9087411154853305, "grad_norm": 1.5192691844350408, "learning_rate": 4.3350100143643604e-07, "loss": 0.457, "step": 25507 }, { "epoch": 0.9087767426118246, "grad_norm": 1.7326349318374934, "learning_rate": 4.3316499214765775e-07, "loss": 0.5301, "step": 25508 }, { "epoch": 0.9088123697383188, "grad_norm": 1.3408094236601227, "learning_rate": 4.328291102479731e-07, "loss": 0.4667, "step": 25509 }, { "epoch": 0.9088479968648129, "grad_norm": 1.5157755518160103, "learning_rate": 4.324933557418576e-07, "loss": 0.5184, "step": 25510 }, { "epoch": 0.908883623991307, "grad_norm": 1.5446865173750566, "learning_rate": 4.321577286337786e-07, "loss": 0.4774, "step": 25511 }, { "epoch": 0.9089192511178011, "grad_norm": 1.4768734072221108, "learning_rate": 4.31822228928207e-07, "loss": 0.4904, "step": 25512 }, { "epoch": 0.9089548782442952, "grad_norm": 1.2502379337032512, "learning_rate": 4.3148685662961043e-07, "loss": 0.3907, "step": 25513 }, { "epoch": 0.9089905053707893, "grad_norm": 1.38920836410124, "learning_rate": 4.3115161174245526e-07, "loss": 0.3864, "step": 25514 }, { "epoch": 0.9090261324972835, "grad_norm": 1.530120185860312, "learning_rate": 4.308164942712034e-07, "loss": 0.5838, "step": 25515 }, { "epoch": 0.9090617596237776, "grad_norm": 1.2395893368678244, "learning_rate": 4.304815042203203e-07, "loss": 0.4148, "step": 25516 }, { "epoch": 0.9090973867502716, "grad_norm": 1.876796370690064, "learning_rate": 4.3014664159426234e-07, "loss": 0.523, "step": 25517 }, { "epoch": 0.9091330138767658, "grad_norm": 1.4037611222545423, "learning_rate": 4.298119063974915e-07, "loss": 0.4034, "step": 25518 }, { "epoch": 0.9091686410032599, "grad_norm": 1.643610640211199, "learning_rate": 4.294772986344653e-07, "loss": 0.5069, "step": 25519 }, { "epoch": 0.909204268129754, "grad_norm": 1.7253221019219287, "learning_rate": 4.291428183096369e-07, "loss": 0.503, "step": 25520 }, { "epoch": 0.9092398952562482, "grad_norm": 1.3026346994129891, "learning_rate": 4.2880846542746156e-07, "loss": 0.4015, "step": 25521 }, { "epoch": 0.9092755223827422, "grad_norm": 2.0291458041161397, "learning_rate": 4.284742399923925e-07, "loss": 0.4828, "step": 25522 }, { "epoch": 0.9093111495092363, "grad_norm": 1.2593297711578182, "learning_rate": 4.2814014200887935e-07, "loss": 0.3753, "step": 25523 }, { "epoch": 0.9093467766357305, "grad_norm": 1.515347671856487, "learning_rate": 4.278061714813686e-07, "loss": 0.5033, "step": 25524 }, { "epoch": 0.9093824037622246, "grad_norm": 1.1528282566621264, "learning_rate": 4.274723284143112e-07, "loss": 0.3225, "step": 25525 }, { "epoch": 0.9094180308887186, "grad_norm": 1.3857119750286127, "learning_rate": 4.2713861281214905e-07, "loss": 0.4679, "step": 25526 }, { "epoch": 0.9094536580152128, "grad_norm": 1.2551888871966532, "learning_rate": 4.268050246793276e-07, "loss": 0.4182, "step": 25527 }, { "epoch": 0.9094892851417069, "grad_norm": 1.2714697539578559, "learning_rate": 4.2647156402028876e-07, "loss": 0.4213, "step": 25528 }, { "epoch": 0.909524912268201, "grad_norm": 1.4914658803167513, "learning_rate": 4.2613823083947346e-07, "loss": 0.5138, "step": 25529 }, { "epoch": 0.9095605393946952, "grad_norm": 1.4575171232335906, "learning_rate": 4.2580502514131705e-07, "loss": 0.4622, "step": 25530 }, { "epoch": 0.9095961665211892, "grad_norm": 1.5012736547901875, "learning_rate": 4.2547194693026037e-07, "loss": 0.5352, "step": 25531 }, { "epoch": 0.9096317936476833, "grad_norm": 1.44247309659347, "learning_rate": 4.2513899621073774e-07, "loss": 0.4395, "step": 25532 }, { "epoch": 0.9096674207741775, "grad_norm": 1.4884628086491563, "learning_rate": 4.2480617298718e-07, "loss": 0.5114, "step": 25533 }, { "epoch": 0.9097030479006716, "grad_norm": 1.295050287099264, "learning_rate": 4.2447347726402134e-07, "loss": 0.4107, "step": 25534 }, { "epoch": 0.9097386750271657, "grad_norm": 1.6224280888832205, "learning_rate": 4.241409090456905e-07, "loss": 0.5709, "step": 25535 }, { "epoch": 0.9097743021536598, "grad_norm": 1.2398683615557202, "learning_rate": 4.238084683366184e-07, "loss": 0.4113, "step": 25536 }, { "epoch": 0.9098099292801539, "grad_norm": 1.7127128883235587, "learning_rate": 4.234761551412281e-07, "loss": 0.5659, "step": 25537 }, { "epoch": 0.909845556406648, "grad_norm": 1.1932141562044534, "learning_rate": 4.2314396946394833e-07, "loss": 0.3748, "step": 25538 }, { "epoch": 0.9098811835331422, "grad_norm": 2.0585242874277387, "learning_rate": 4.2281191130919885e-07, "loss": 0.5076, "step": 25539 }, { "epoch": 0.9099168106596363, "grad_norm": 1.702658032562377, "learning_rate": 4.2247998068140395e-07, "loss": 0.4784, "step": 25540 }, { "epoch": 0.9099524377861303, "grad_norm": 1.2001085723246987, "learning_rate": 4.221481775849812e-07, "loss": 0.3729, "step": 25541 }, { "epoch": 0.9099880649126245, "grad_norm": 1.2903556134469345, "learning_rate": 4.218165020243503e-07, "loss": 0.4181, "step": 25542 }, { "epoch": 0.9100236920391186, "grad_norm": 1.6252995814883484, "learning_rate": 4.2148495400392675e-07, "loss": 0.5603, "step": 25543 }, { "epoch": 0.9100593191656127, "grad_norm": 1.466170384824323, "learning_rate": 4.2115353352812694e-07, "loss": 0.5073, "step": 25544 }, { "epoch": 0.9100949462921069, "grad_norm": 1.8989317090087585, "learning_rate": 4.20822240601364e-07, "loss": 0.5768, "step": 25545 }, { "epoch": 0.9101305734186009, "grad_norm": 1.3043383178387764, "learning_rate": 4.2049107522804664e-07, "loss": 0.424, "step": 25546 }, { "epoch": 0.910166200545095, "grad_norm": 1.3909958368322832, "learning_rate": 4.2016003741258805e-07, "loss": 0.4128, "step": 25547 }, { "epoch": 0.9102018276715892, "grad_norm": 1.615458018410875, "learning_rate": 4.198291271593924e-07, "loss": 0.3788, "step": 25548 }, { "epoch": 0.9102374547980833, "grad_norm": 1.3376088676650046, "learning_rate": 4.194983444728684e-07, "loss": 0.4081, "step": 25549 }, { "epoch": 0.9102730819245773, "grad_norm": 1.7563796936631144, "learning_rate": 4.191676893574215e-07, "loss": 0.5493, "step": 25550 }, { "epoch": 0.9103087090510715, "grad_norm": 1.2743028033700532, "learning_rate": 4.188371618174536e-07, "loss": 0.3951, "step": 25551 }, { "epoch": 0.9103443361775656, "grad_norm": 1.413463660446959, "learning_rate": 4.185067618573646e-07, "loss": 0.4485, "step": 25552 }, { "epoch": 0.9103799633040597, "grad_norm": 1.6726646261488902, "learning_rate": 4.181764894815543e-07, "loss": 0.5182, "step": 25553 }, { "epoch": 0.9104155904305539, "grad_norm": 1.1970357968380212, "learning_rate": 4.1784634469442363e-07, "loss": 0.4058, "step": 25554 }, { "epoch": 0.9104512175570479, "grad_norm": 3.8731899444543916, "learning_rate": 4.175163275003658e-07, "loss": 0.4039, "step": 25555 }, { "epoch": 0.910486844683542, "grad_norm": 1.2758484196313407, "learning_rate": 4.17186437903776e-07, "loss": 0.4042, "step": 25556 }, { "epoch": 0.9105224718100362, "grad_norm": 1.622526213511379, "learning_rate": 4.1685667590904534e-07, "loss": 0.5295, "step": 25557 }, { "epoch": 0.9105580989365303, "grad_norm": 1.4640378567969796, "learning_rate": 4.165270415205669e-07, "loss": 0.4875, "step": 25558 }, { "epoch": 0.9105937260630244, "grad_norm": 1.1231490281109853, "learning_rate": 4.161975347427294e-07, "loss": 0.3757, "step": 25559 }, { "epoch": 0.9106293531895185, "grad_norm": 1.4978400869549744, "learning_rate": 4.1586815557992043e-07, "loss": 0.4776, "step": 25560 }, { "epoch": 0.9106649803160126, "grad_norm": 1.3122287223563234, "learning_rate": 4.155389040365265e-07, "loss": 0.4042, "step": 25561 }, { "epoch": 0.9107006074425067, "grad_norm": 1.3416900919362234, "learning_rate": 4.152097801169308e-07, "loss": 0.3962, "step": 25562 }, { "epoch": 0.9107362345690009, "grad_norm": 1.657775003344334, "learning_rate": 4.1488078382551646e-07, "loss": 0.491, "step": 25563 }, { "epoch": 0.910771861695495, "grad_norm": 1.5613706690032143, "learning_rate": 4.145519151666633e-07, "loss": 0.51, "step": 25564 }, { "epoch": 0.910807488821989, "grad_norm": 1.4015022229663991, "learning_rate": 4.142231741447511e-07, "loss": 0.5096, "step": 25565 }, { "epoch": 0.9108431159484832, "grad_norm": 1.6729190572034278, "learning_rate": 4.138945607641576e-07, "loss": 0.4731, "step": 25566 }, { "epoch": 0.9108787430749773, "grad_norm": 1.6280512483587077, "learning_rate": 4.135660750292592e-07, "loss": 0.4285, "step": 25567 }, { "epoch": 0.9109143702014714, "grad_norm": 1.6987237281481165, "learning_rate": 4.1323771694442796e-07, "loss": 0.4335, "step": 25568 }, { "epoch": 0.9109499973279656, "grad_norm": 1.3113343258252204, "learning_rate": 4.129094865140382e-07, "loss": 0.39, "step": 25569 }, { "epoch": 0.9109856244544596, "grad_norm": 1.1008348551997227, "learning_rate": 4.125813837424597e-07, "loss": 0.3459, "step": 25570 }, { "epoch": 0.9110212515809537, "grad_norm": 1.9399445875924068, "learning_rate": 4.1225340863406126e-07, "loss": 0.5308, "step": 25571 }, { "epoch": 0.9110568787074479, "grad_norm": 1.3319889047120645, "learning_rate": 4.1192556119320827e-07, "loss": 0.4339, "step": 25572 }, { "epoch": 0.911092505833942, "grad_norm": 1.2197255397442985, "learning_rate": 4.115978414242694e-07, "loss": 0.3736, "step": 25573 }, { "epoch": 0.911128132960436, "grad_norm": 1.3235540065262545, "learning_rate": 4.1127024933160785e-07, "loss": 0.5169, "step": 25574 }, { "epoch": 0.9111637600869302, "grad_norm": 1.500773055113169, "learning_rate": 4.109427849195846e-07, "loss": 0.4923, "step": 25575 }, { "epoch": 0.9111993872134243, "grad_norm": 1.4365374215778743, "learning_rate": 4.106154481925606e-07, "loss": 0.4971, "step": 25576 }, { "epoch": 0.9112350143399184, "grad_norm": 1.7466771403650623, "learning_rate": 4.1028823915489344e-07, "loss": 0.4902, "step": 25577 }, { "epoch": 0.9112706414664126, "grad_norm": 1.5998086174260508, "learning_rate": 4.0996115781094306e-07, "loss": 0.4999, "step": 25578 }, { "epoch": 0.9113062685929066, "grad_norm": 1.1769116151935035, "learning_rate": 4.0963420416506136e-07, "loss": 0.3797, "step": 25579 }, { "epoch": 0.9113418957194007, "grad_norm": 1.469535654425029, "learning_rate": 4.0930737822160285e-07, "loss": 0.494, "step": 25580 }, { "epoch": 0.9113775228458949, "grad_norm": 1.5992978453273097, "learning_rate": 4.0898067998492163e-07, "loss": 0.6287, "step": 25581 }, { "epoch": 0.911413149972389, "grad_norm": 1.5884410314810506, "learning_rate": 4.086541094593666e-07, "loss": 0.5314, "step": 25582 }, { "epoch": 0.9114487770988831, "grad_norm": 1.5262904290835169, "learning_rate": 4.083276666492841e-07, "loss": 0.4959, "step": 25583 }, { "epoch": 0.9114844042253772, "grad_norm": 1.2206161588477231, "learning_rate": 4.080013515590253e-07, "loss": 0.4119, "step": 25584 }, { "epoch": 0.9115200313518713, "grad_norm": 1.3466037687737853, "learning_rate": 4.076751641929322e-07, "loss": 0.3952, "step": 25585 }, { "epoch": 0.9115556584783654, "grad_norm": 1.8035970839044935, "learning_rate": 4.073491045553479e-07, "loss": 0.5673, "step": 25586 }, { "epoch": 0.9115912856048596, "grad_norm": 1.4196088178148498, "learning_rate": 4.0702317265061466e-07, "loss": 0.5225, "step": 25587 }, { "epoch": 0.9116269127313537, "grad_norm": 1.472008729942707, "learning_rate": 4.066973684830744e-07, "loss": 0.5386, "step": 25588 }, { "epoch": 0.9116625398578477, "grad_norm": 1.7252116051459419, "learning_rate": 4.063716920570637e-07, "loss": 0.5788, "step": 25589 }, { "epoch": 0.9116981669843419, "grad_norm": 1.2367290182115962, "learning_rate": 4.060461433769192e-07, "loss": 0.4465, "step": 25590 }, { "epoch": 0.911733794110836, "grad_norm": 1.2041371224156783, "learning_rate": 4.0572072244697726e-07, "loss": 0.3736, "step": 25591 }, { "epoch": 0.9117694212373301, "grad_norm": 1.1544165145115146, "learning_rate": 4.0539542927156893e-07, "loss": 0.3814, "step": 25592 }, { "epoch": 0.9118050483638243, "grad_norm": 1.614734214066312, "learning_rate": 4.0507026385502747e-07, "loss": 0.5453, "step": 25593 }, { "epoch": 0.9118406754903183, "grad_norm": 1.6814207018950875, "learning_rate": 4.0474522620168047e-07, "loss": 0.5148, "step": 25594 }, { "epoch": 0.9118763026168124, "grad_norm": 1.506629364005453, "learning_rate": 4.0442031631585774e-07, "loss": 0.5302, "step": 25595 }, { "epoch": 0.9119119297433066, "grad_norm": 1.2376301844126922, "learning_rate": 4.040955342018849e-07, "loss": 0.3565, "step": 25596 }, { "epoch": 0.9119475568698007, "grad_norm": 1.1651186085237601, "learning_rate": 4.0377087986408937e-07, "loss": 0.3792, "step": 25597 }, { "epoch": 0.9119831839962947, "grad_norm": 1.7097483631828714, "learning_rate": 4.034463533067901e-07, "loss": 0.5645, "step": 25598 }, { "epoch": 0.9120188111227889, "grad_norm": 1.5956995440597694, "learning_rate": 4.0312195453431015e-07, "loss": 0.5135, "step": 25599 }, { "epoch": 0.912054438249283, "grad_norm": 1.6688794048338158, "learning_rate": 4.0279768355097057e-07, "loss": 0.4381, "step": 25600 }, { "epoch": 0.9120900653757771, "grad_norm": 2.2319175875887427, "learning_rate": 4.0247354036108687e-07, "loss": 0.4764, "step": 25601 }, { "epoch": 0.9121256925022713, "grad_norm": 1.3670764738649208, "learning_rate": 4.0214952496897663e-07, "loss": 0.4112, "step": 25602 }, { "epoch": 0.9121613196287653, "grad_norm": 1.230925865340831, "learning_rate": 4.0182563737895306e-07, "loss": 0.4304, "step": 25603 }, { "epoch": 0.9121969467552594, "grad_norm": 1.266204936149763, "learning_rate": 4.015018775953317e-07, "loss": 0.3409, "step": 25604 }, { "epoch": 0.9122325738817536, "grad_norm": 1.2556118436241719, "learning_rate": 4.0117824562242e-07, "loss": 0.3668, "step": 25605 }, { "epoch": 0.9122682010082477, "grad_norm": 1.2667527780079524, "learning_rate": 4.0085474146453027e-07, "loss": 0.4459, "step": 25606 }, { "epoch": 0.9123038281347418, "grad_norm": 1.2132478702988507, "learning_rate": 4.00531365125969e-07, "loss": 0.4122, "step": 25607 }, { "epoch": 0.9123394552612359, "grad_norm": 1.2132403657576587, "learning_rate": 4.0020811661104163e-07, "loss": 0.401, "step": 25608 }, { "epoch": 0.91237508238773, "grad_norm": 1.609743027109478, "learning_rate": 3.998849959240525e-07, "loss": 0.4838, "step": 25609 }, { "epoch": 0.9124107095142241, "grad_norm": 1.4946980363598916, "learning_rate": 3.995620030693048e-07, "loss": 0.5046, "step": 25610 }, { "epoch": 0.9124463366407183, "grad_norm": 1.4796587984064966, "learning_rate": 3.992391380511007e-07, "loss": 0.4162, "step": 25611 }, { "epoch": 0.9124819637672124, "grad_norm": 1.176272575657593, "learning_rate": 3.989164008737367e-07, "loss": 0.4128, "step": 25612 }, { "epoch": 0.9125175908937064, "grad_norm": 1.5789601108262366, "learning_rate": 3.985937915415128e-07, "loss": 0.4697, "step": 25613 }, { "epoch": 0.9125532180202006, "grad_norm": 1.4668920635952853, "learning_rate": 3.982713100587232e-07, "loss": 0.4728, "step": 25614 }, { "epoch": 0.9125888451466947, "grad_norm": 1.4690359521566192, "learning_rate": 3.9794895642966234e-07, "loss": 0.4263, "step": 25615 }, { "epoch": 0.9126244722731888, "grad_norm": 1.4023036053809823, "learning_rate": 3.976267306586223e-07, "loss": 0.4803, "step": 25616 }, { "epoch": 0.912660099399683, "grad_norm": 1.404130715370303, "learning_rate": 3.9730463274989305e-07, "loss": 0.3917, "step": 25617 }, { "epoch": 0.912695726526177, "grad_norm": 1.9573791927721549, "learning_rate": 3.9698266270776556e-07, "loss": 0.5101, "step": 25618 }, { "epoch": 0.9127313536526711, "grad_norm": 1.5488268635417328, "learning_rate": 3.966608205365263e-07, "loss": 0.4308, "step": 25619 }, { "epoch": 0.9127669807791653, "grad_norm": 1.2374499097582439, "learning_rate": 3.963391062404598e-07, "loss": 0.3641, "step": 25620 }, { "epoch": 0.9128026079056594, "grad_norm": 1.4914972505901845, "learning_rate": 3.960175198238525e-07, "loss": 0.4063, "step": 25621 }, { "epoch": 0.9128382350321534, "grad_norm": 1.4623276034664854, "learning_rate": 3.9569606129098216e-07, "loss": 0.465, "step": 25622 }, { "epoch": 0.9128738621586476, "grad_norm": 1.3691529295616172, "learning_rate": 3.9537473064613307e-07, "loss": 0.4501, "step": 25623 }, { "epoch": 0.9129094892851417, "grad_norm": 1.5059594839687112, "learning_rate": 3.9505352789358187e-07, "loss": 0.5741, "step": 25624 }, { "epoch": 0.9129451164116358, "grad_norm": 1.7886386023125789, "learning_rate": 3.947324530376062e-07, "loss": 0.5674, "step": 25625 }, { "epoch": 0.91298074353813, "grad_norm": 1.4630540178717077, "learning_rate": 3.944115060824827e-07, "loss": 0.5282, "step": 25626 }, { "epoch": 0.913016370664624, "grad_norm": 1.39295298633518, "learning_rate": 3.9409068703248346e-07, "loss": 0.5508, "step": 25627 }, { "epoch": 0.9130519977911181, "grad_norm": 1.5031458001480484, "learning_rate": 3.937699958918817e-07, "loss": 0.4454, "step": 25628 }, { "epoch": 0.9130876249176123, "grad_norm": 1.3181004379166827, "learning_rate": 3.934494326649463e-07, "loss": 0.4168, "step": 25629 }, { "epoch": 0.9131232520441064, "grad_norm": 1.4328396127812482, "learning_rate": 3.93128997355946e-07, "loss": 0.4921, "step": 25630 }, { "epoch": 0.9131588791706005, "grad_norm": 1.387863674353744, "learning_rate": 3.9280868996914745e-07, "loss": 0.3978, "step": 25631 }, { "epoch": 0.9131945062970946, "grad_norm": 1.126185325141175, "learning_rate": 3.924885105088172e-07, "loss": 0.3915, "step": 25632 }, { "epoch": 0.9132301334235887, "grad_norm": 1.4179546129207148, "learning_rate": 3.921684589792163e-07, "loss": 0.4032, "step": 25633 }, { "epoch": 0.9132657605500828, "grad_norm": 1.1826310986875608, "learning_rate": 3.91848535384608e-07, "loss": 0.4074, "step": 25634 }, { "epoch": 0.913301387676577, "grad_norm": 1.2424665106124688, "learning_rate": 3.9152873972925444e-07, "loss": 0.3751, "step": 25635 }, { "epoch": 0.9133370148030711, "grad_norm": 1.3854065703734648, "learning_rate": 3.912090720174089e-07, "loss": 0.408, "step": 25636 }, { "epoch": 0.9133726419295651, "grad_norm": 1.531262503230583, "learning_rate": 3.908895322533335e-07, "loss": 0.5136, "step": 25637 }, { "epoch": 0.9134082690560593, "grad_norm": 1.2608812523322943, "learning_rate": 3.9057012044127817e-07, "loss": 0.3639, "step": 25638 }, { "epoch": 0.9134438961825534, "grad_norm": 1.7443629591169652, "learning_rate": 3.9025083658549845e-07, "loss": 0.5092, "step": 25639 }, { "epoch": 0.9134795233090475, "grad_norm": 1.6393690338893518, "learning_rate": 3.8993168069024535e-07, "loss": 0.4829, "step": 25640 }, { "epoch": 0.9135151504355417, "grad_norm": 1.3105423402748664, "learning_rate": 3.896126527597699e-07, "loss": 0.4057, "step": 25641 }, { "epoch": 0.9135507775620357, "grad_norm": 1.28799688579222, "learning_rate": 3.892937527983187e-07, "loss": 0.4465, "step": 25642 }, { "epoch": 0.9135864046885298, "grad_norm": 1.8202757059048007, "learning_rate": 3.889749808101395e-07, "loss": 0.4843, "step": 25643 }, { "epoch": 0.913622031815024, "grad_norm": 1.5016738644805947, "learning_rate": 3.886563367994756e-07, "loss": 0.4553, "step": 25644 }, { "epoch": 0.9136576589415181, "grad_norm": 1.356508124093116, "learning_rate": 3.8833782077057015e-07, "loss": 0.3848, "step": 25645 }, { "epoch": 0.9136932860680121, "grad_norm": 1.2436682074333645, "learning_rate": 3.880194327276654e-07, "loss": 0.4058, "step": 25646 }, { "epoch": 0.9137289131945063, "grad_norm": 1.531753032369979, "learning_rate": 3.8770117267499905e-07, "loss": 0.4363, "step": 25647 }, { "epoch": 0.9137645403210004, "grad_norm": 1.2941685647083212, "learning_rate": 3.8738304061681107e-07, "loss": 0.4153, "step": 25648 }, { "epoch": 0.9138001674474945, "grad_norm": 1.324847180358592, "learning_rate": 3.8706503655733696e-07, "loss": 0.3942, "step": 25649 }, { "epoch": 0.9138357945739887, "grad_norm": 1.2854463712298871, "learning_rate": 3.8674716050081105e-07, "loss": 0.4029, "step": 25650 }, { "epoch": 0.9138714217004827, "grad_norm": 1.1570356983004262, "learning_rate": 3.864294124514656e-07, "loss": 0.3332, "step": 25651 }, { "epoch": 0.9139070488269768, "grad_norm": 1.2378811409540103, "learning_rate": 3.861117924135327e-07, "loss": 0.351, "step": 25652 }, { "epoch": 0.913942675953471, "grad_norm": 1.5241223160701374, "learning_rate": 3.857943003912401e-07, "loss": 0.5766, "step": 25653 }, { "epoch": 0.9139783030799651, "grad_norm": 1.3227396786297259, "learning_rate": 3.854769363888156e-07, "loss": 0.3665, "step": 25654 }, { "epoch": 0.9140139302064592, "grad_norm": 1.8141033807342477, "learning_rate": 3.851597004104868e-07, "loss": 0.5674, "step": 25655 }, { "epoch": 0.9140495573329533, "grad_norm": 1.6658567299025766, "learning_rate": 3.8484259246047705e-07, "loss": 0.4914, "step": 25656 }, { "epoch": 0.9140851844594474, "grad_norm": 1.563471225744291, "learning_rate": 3.8452561254300856e-07, "loss": 0.4875, "step": 25657 }, { "epoch": 0.9141208115859415, "grad_norm": 1.5230027294881914, "learning_rate": 3.842087606623024e-07, "loss": 0.4237, "step": 25658 }, { "epoch": 0.9141564387124357, "grad_norm": 1.2753061678394455, "learning_rate": 3.838920368225785e-07, "loss": 0.4053, "step": 25659 }, { "epoch": 0.9141920658389298, "grad_norm": 1.4319368499518175, "learning_rate": 3.8357544102805233e-07, "loss": 0.5258, "step": 25660 }, { "epoch": 0.9142276929654238, "grad_norm": 1.4813545833635613, "learning_rate": 3.832589732829406e-07, "loss": 0.4358, "step": 25661 }, { "epoch": 0.914263320091918, "grad_norm": 1.6412613780990144, "learning_rate": 3.8294263359145657e-07, "loss": 0.4819, "step": 25662 }, { "epoch": 0.9142989472184121, "grad_norm": 1.2955042403261936, "learning_rate": 3.826264219578157e-07, "loss": 0.3821, "step": 25663 }, { "epoch": 0.9143345743449062, "grad_norm": 1.3013340347443865, "learning_rate": 3.823103383862248e-07, "loss": 0.426, "step": 25664 }, { "epoch": 0.9143702014714004, "grad_norm": 1.3736453591379616, "learning_rate": 3.8199438288089474e-07, "loss": 0.5189, "step": 25665 }, { "epoch": 0.9144058285978944, "grad_norm": 1.8178015107549108, "learning_rate": 3.8167855544603115e-07, "loss": 0.4848, "step": 25666 }, { "epoch": 0.9144414557243885, "grad_norm": 1.3870384998941405, "learning_rate": 3.8136285608584065e-07, "loss": 0.5729, "step": 25667 }, { "epoch": 0.9144770828508827, "grad_norm": 1.5621412295528472, "learning_rate": 3.810472848045266e-07, "loss": 0.4036, "step": 25668 }, { "epoch": 0.9145127099773768, "grad_norm": 1.2543641646094055, "learning_rate": 3.8073184160629105e-07, "loss": 0.3528, "step": 25669 }, { "epoch": 0.9145483371038708, "grad_norm": 1.3523671010869307, "learning_rate": 3.8041652649533413e-07, "loss": 0.4422, "step": 25670 }, { "epoch": 0.914583964230365, "grad_norm": 1.5731094425552816, "learning_rate": 3.801013394758557e-07, "loss": 0.5221, "step": 25671 }, { "epoch": 0.9146195913568591, "grad_norm": 1.2413313506063453, "learning_rate": 3.7978628055205245e-07, "loss": 0.4012, "step": 25672 }, { "epoch": 0.9146552184833532, "grad_norm": 1.2110424061625449, "learning_rate": 3.7947134972811775e-07, "loss": 0.3575, "step": 25673 }, { "epoch": 0.9146908456098474, "grad_norm": 1.3254424696934213, "learning_rate": 3.791565470082481e-07, "loss": 0.438, "step": 25674 }, { "epoch": 0.9147264727363414, "grad_norm": 1.216944801326538, "learning_rate": 3.7884187239663137e-07, "loss": 0.4261, "step": 25675 }, { "epoch": 0.9147620998628355, "grad_norm": 1.6545906040770442, "learning_rate": 3.7852732589746087e-07, "loss": 0.5132, "step": 25676 }, { "epoch": 0.9147977269893297, "grad_norm": 1.5735987489554706, "learning_rate": 3.7821290751492325e-07, "loss": 0.5065, "step": 25677 }, { "epoch": 0.9148333541158238, "grad_norm": 2.0031197614931426, "learning_rate": 3.778986172532073e-07, "loss": 0.5859, "step": 25678 }, { "epoch": 0.9148689812423179, "grad_norm": 1.7610539368128815, "learning_rate": 3.7758445511649645e-07, "loss": 0.4079, "step": 25679 }, { "epoch": 0.914904608368812, "grad_norm": 1.4028561757463793, "learning_rate": 3.7727042110897283e-07, "loss": 0.525, "step": 25680 }, { "epoch": 0.9149402354953061, "grad_norm": 1.371164450054716, "learning_rate": 3.76956515234822e-07, "loss": 0.4442, "step": 25681 }, { "epoch": 0.9149758626218002, "grad_norm": 1.2987365591300761, "learning_rate": 3.766427374982196e-07, "loss": 0.3767, "step": 25682 }, { "epoch": 0.9150114897482944, "grad_norm": 1.4721283972076191, "learning_rate": 3.7632908790334656e-07, "loss": 0.3823, "step": 25683 }, { "epoch": 0.9150471168747885, "grad_norm": 1.1789944433429014, "learning_rate": 3.7601556645437633e-07, "loss": 0.4198, "step": 25684 }, { "epoch": 0.9150827440012825, "grad_norm": 1.4064703874654825, "learning_rate": 3.757021731554866e-07, "loss": 0.4196, "step": 25685 }, { "epoch": 0.9151183711277767, "grad_norm": 1.6314960224277908, "learning_rate": 3.753889080108497e-07, "loss": 0.5341, "step": 25686 }, { "epoch": 0.9151539982542708, "grad_norm": 1.24508703328674, "learning_rate": 3.7507577102463665e-07, "loss": 0.3778, "step": 25687 }, { "epoch": 0.9151896253807649, "grad_norm": 1.5718540253056896, "learning_rate": 3.747627622010164e-07, "loss": 0.5013, "step": 25688 }, { "epoch": 0.9152252525072591, "grad_norm": 1.3467651013832933, "learning_rate": 3.7444988154415997e-07, "loss": 0.477, "step": 25689 }, { "epoch": 0.9152608796337531, "grad_norm": 1.5559796917077908, "learning_rate": 3.7413712905822854e-07, "loss": 0.5373, "step": 25690 }, { "epoch": 0.9152965067602472, "grad_norm": 1.4060353816900961, "learning_rate": 3.7382450474738987e-07, "loss": 0.4107, "step": 25691 }, { "epoch": 0.9153321338867414, "grad_norm": 1.2450175953302716, "learning_rate": 3.735120086158062e-07, "loss": 0.3923, "step": 25692 }, { "epoch": 0.9153677610132355, "grad_norm": 1.2771987417896788, "learning_rate": 3.7319964066763857e-07, "loss": 0.423, "step": 25693 }, { "epoch": 0.9154033881397295, "grad_norm": 1.2756239251330237, "learning_rate": 3.7288740090704824e-07, "loss": 0.3782, "step": 25694 }, { "epoch": 0.9154390152662237, "grad_norm": 1.2115192382855193, "learning_rate": 3.725752893381884e-07, "loss": 0.3823, "step": 25695 }, { "epoch": 0.9154746423927178, "grad_norm": 1.661402792159118, "learning_rate": 3.722633059652203e-07, "loss": 0.4393, "step": 25696 }, { "epoch": 0.9155102695192119, "grad_norm": 1.3550984539277184, "learning_rate": 3.719514507922939e-07, "loss": 0.4745, "step": 25697 }, { "epoch": 0.9155458966457061, "grad_norm": 1.7527897920344235, "learning_rate": 3.7163972382356364e-07, "loss": 0.5277, "step": 25698 }, { "epoch": 0.9155815237722001, "grad_norm": 1.5414621070152308, "learning_rate": 3.713281250631806e-07, "loss": 0.4754, "step": 25699 }, { "epoch": 0.9156171508986942, "grad_norm": 1.8055455079033036, "learning_rate": 3.710166545152949e-07, "loss": 0.4834, "step": 25700 }, { "epoch": 0.9156527780251884, "grad_norm": 2.0491221530971235, "learning_rate": 3.7070531218405095e-07, "loss": 0.417, "step": 25701 }, { "epoch": 0.9156884051516825, "grad_norm": 1.7375558532562494, "learning_rate": 3.703940980735976e-07, "loss": 0.5114, "step": 25702 }, { "epoch": 0.9157240322781766, "grad_norm": 1.5748193309637448, "learning_rate": 3.7008301218807716e-07, "loss": 0.4577, "step": 25703 }, { "epoch": 0.9157596594046707, "grad_norm": 1.6805198839790059, "learning_rate": 3.697720545316308e-07, "loss": 0.5367, "step": 25704 }, { "epoch": 0.9157952865311648, "grad_norm": 2.5196231683069286, "learning_rate": 3.694612251084029e-07, "loss": 0.4259, "step": 25705 }, { "epoch": 0.9158309136576589, "grad_norm": 1.4434928499517783, "learning_rate": 3.6915052392252903e-07, "loss": 0.5017, "step": 25706 }, { "epoch": 0.9158665407841531, "grad_norm": 1.2895860253391949, "learning_rate": 3.6883995097814707e-07, "loss": 0.4365, "step": 25707 }, { "epoch": 0.9159021679106472, "grad_norm": 1.5002970012707073, "learning_rate": 3.6852950627939253e-07, "loss": 0.5084, "step": 25708 }, { "epoch": 0.9159377950371412, "grad_norm": 1.1897603811713837, "learning_rate": 3.6821918983040105e-07, "loss": 0.3526, "step": 25709 }, { "epoch": 0.9159734221636354, "grad_norm": 1.2779584099392887, "learning_rate": 3.6790900163530266e-07, "loss": 0.3757, "step": 25710 }, { "epoch": 0.9160090492901295, "grad_norm": 1.540453292965241, "learning_rate": 3.6759894169822954e-07, "loss": 0.3624, "step": 25711 }, { "epoch": 0.9160446764166236, "grad_norm": 1.2479530924440927, "learning_rate": 3.672890100233073e-07, "loss": 0.4353, "step": 25712 }, { "epoch": 0.9160803035431178, "grad_norm": 1.350350720840899, "learning_rate": 3.669792066146649e-07, "loss": 0.4811, "step": 25713 }, { "epoch": 0.9161159306696118, "grad_norm": 1.3479444561707823, "learning_rate": 3.666695314764279e-07, "loss": 0.3783, "step": 25714 }, { "epoch": 0.9161515577961059, "grad_norm": 1.6121178637733007, "learning_rate": 3.663599846127186e-07, "loss": 0.4759, "step": 25715 }, { "epoch": 0.9161871849226001, "grad_norm": 1.3224653869672116, "learning_rate": 3.660505660276614e-07, "loss": 0.3902, "step": 25716 }, { "epoch": 0.9162228120490942, "grad_norm": 1.329456492816602, "learning_rate": 3.657412757253731e-07, "loss": 0.3916, "step": 25717 }, { "epoch": 0.9162584391755882, "grad_norm": 1.4149504214294217, "learning_rate": 3.6543211370997587e-07, "loss": 0.4717, "step": 25718 }, { "epoch": 0.9162940663020824, "grad_norm": 1.5818214129493435, "learning_rate": 3.6512307998558205e-07, "loss": 0.537, "step": 25719 }, { "epoch": 0.9163296934285765, "grad_norm": 1.6098423720929207, "learning_rate": 3.6481417455631054e-07, "loss": 0.432, "step": 25720 }, { "epoch": 0.9163653205550706, "grad_norm": 1.4679770570362751, "learning_rate": 3.6450539742627135e-07, "loss": 0.4438, "step": 25721 }, { "epoch": 0.9164009476815648, "grad_norm": 1.1780352998936794, "learning_rate": 3.6419674859957677e-07, "loss": 0.4136, "step": 25722 }, { "epoch": 0.9164365748080588, "grad_norm": 1.2074898425785756, "learning_rate": 3.6388822808033905e-07, "loss": 0.3979, "step": 25723 }, { "epoch": 0.9164722019345529, "grad_norm": 1.2473955814641642, "learning_rate": 3.6357983587266497e-07, "loss": 0.3933, "step": 25724 }, { "epoch": 0.9165078290610471, "grad_norm": 1.3368225932354112, "learning_rate": 3.632715719806601e-07, "loss": 0.4098, "step": 25725 }, { "epoch": 0.9165434561875412, "grad_norm": 1.1968940793480285, "learning_rate": 3.6296343640843e-07, "loss": 0.3932, "step": 25726 }, { "epoch": 0.9165790833140353, "grad_norm": 1.576160956868131, "learning_rate": 3.6265542916007814e-07, "loss": 0.4535, "step": 25727 }, { "epoch": 0.9166147104405294, "grad_norm": 1.476653770836412, "learning_rate": 3.6234755023970447e-07, "loss": 0.404, "step": 25728 }, { "epoch": 0.9166503375670235, "grad_norm": 1.3220754902380485, "learning_rate": 3.6203979965141025e-07, "loss": 0.3986, "step": 25729 }, { "epoch": 0.9166859646935176, "grad_norm": 6.459819914797589, "learning_rate": 3.61732177399291e-07, "loss": 0.5601, "step": 25730 }, { "epoch": 0.9167215918200118, "grad_norm": 1.7119813717812364, "learning_rate": 3.614246834874469e-07, "loss": 0.4434, "step": 25731 }, { "epoch": 0.9167572189465059, "grad_norm": 1.883503560146275, "learning_rate": 3.6111731791996894e-07, "loss": 0.5584, "step": 25732 }, { "epoch": 0.9167928460729999, "grad_norm": 1.0859073855958483, "learning_rate": 3.608100807009518e-07, "loss": 0.3754, "step": 25733 }, { "epoch": 0.9168284731994941, "grad_norm": 1.3036593759912096, "learning_rate": 3.6050297183448543e-07, "loss": 0.4049, "step": 25734 }, { "epoch": 0.9168641003259882, "grad_norm": 1.2658807411202002, "learning_rate": 3.6019599132465997e-07, "loss": 0.4197, "step": 25735 }, { "epoch": 0.9168997274524823, "grad_norm": 1.3632261340152323, "learning_rate": 3.59889139175561e-07, "loss": 0.3729, "step": 25736 }, { "epoch": 0.9169353545789765, "grad_norm": 1.2991303398610312, "learning_rate": 3.5958241539127635e-07, "loss": 0.3744, "step": 25737 }, { "epoch": 0.9169709817054705, "grad_norm": 1.3142885111300553, "learning_rate": 3.5927581997589276e-07, "loss": 0.4288, "step": 25738 }, { "epoch": 0.9170066088319646, "grad_norm": 1.18410341490274, "learning_rate": 3.589693529334881e-07, "loss": 0.3933, "step": 25739 }, { "epoch": 0.9170422359584588, "grad_norm": 1.2661650798504704, "learning_rate": 3.5866301426814573e-07, "loss": 0.4159, "step": 25740 }, { "epoch": 0.9170778630849529, "grad_norm": 1.4295875992366667, "learning_rate": 3.583568039839436e-07, "loss": 0.415, "step": 25741 }, { "epoch": 0.917113490211447, "grad_norm": 1.2127818708904838, "learning_rate": 3.580507220849616e-07, "loss": 0.3521, "step": 25742 }, { "epoch": 0.9171491173379411, "grad_norm": 1.3865505304608243, "learning_rate": 3.5774476857527107e-07, "loss": 0.4757, "step": 25743 }, { "epoch": 0.9171847444644352, "grad_norm": 1.3088184811400638, "learning_rate": 3.574389434589487e-07, "loss": 0.4218, "step": 25744 }, { "epoch": 0.9172203715909293, "grad_norm": 1.2468790155709346, "learning_rate": 3.5713324674006787e-07, "loss": 0.3772, "step": 25745 }, { "epoch": 0.9172559987174235, "grad_norm": 1.2968037265369268, "learning_rate": 3.568276784226976e-07, "loss": 0.4146, "step": 25746 }, { "epoch": 0.9172916258439175, "grad_norm": 1.5865090514457727, "learning_rate": 3.5652223851090686e-07, "loss": 0.4738, "step": 25747 }, { "epoch": 0.9173272529704116, "grad_norm": 1.6017652067405925, "learning_rate": 3.562169270087634e-07, "loss": 0.5169, "step": 25748 }, { "epoch": 0.9173628800969058, "grad_norm": 1.4219468639486366, "learning_rate": 3.5591174392033077e-07, "loss": 0.4739, "step": 25749 }, { "epoch": 0.9173985072233999, "grad_norm": 1.432566528349411, "learning_rate": 3.5560668924967455e-07, "loss": 0.5134, "step": 25750 }, { "epoch": 0.917434134349894, "grad_norm": 1.1507177389020278, "learning_rate": 3.5530176300085593e-07, "loss": 0.396, "step": 25751 }, { "epoch": 0.9174697614763881, "grad_norm": 1.2582461818817612, "learning_rate": 3.5499696517793615e-07, "loss": 0.4037, "step": 25752 }, { "epoch": 0.9175053886028822, "grad_norm": 1.8848360396544708, "learning_rate": 3.546922957849741e-07, "loss": 0.5043, "step": 25753 }, { "epoch": 0.9175410157293763, "grad_norm": 1.4464400637352923, "learning_rate": 3.543877548260244e-07, "loss": 0.4775, "step": 25754 }, { "epoch": 0.9175766428558705, "grad_norm": 1.1940615496405964, "learning_rate": 3.5408334230514484e-07, "loss": 0.4021, "step": 25755 }, { "epoch": 0.9176122699823646, "grad_norm": 1.226005338279035, "learning_rate": 3.537790582263867e-07, "loss": 0.3543, "step": 25756 }, { "epoch": 0.9176478971088586, "grad_norm": 1.3625628250161317, "learning_rate": 3.5347490259380334e-07, "loss": 0.4573, "step": 25757 }, { "epoch": 0.9176835242353528, "grad_norm": 2.412586939360608, "learning_rate": 3.531708754114438e-07, "loss": 0.4109, "step": 25758 }, { "epoch": 0.9177191513618469, "grad_norm": 1.253008696253613, "learning_rate": 3.528669766833559e-07, "loss": 0.3798, "step": 25759 }, { "epoch": 0.917754778488341, "grad_norm": 1.3867525071207951, "learning_rate": 3.5256320641358756e-07, "loss": 0.4152, "step": 25760 }, { "epoch": 0.9177904056148352, "grad_norm": 1.7165001444771626, "learning_rate": 3.522595646061844e-07, "loss": 0.423, "step": 25761 }, { "epoch": 0.9178260327413292, "grad_norm": 1.294197287945844, "learning_rate": 3.519560512651887e-07, "loss": 0.3948, "step": 25762 }, { "epoch": 0.9178616598678233, "grad_norm": 1.4474640745710838, "learning_rate": 3.5165266639463957e-07, "loss": 0.4232, "step": 25763 }, { "epoch": 0.9178972869943175, "grad_norm": 1.1860167273510986, "learning_rate": 3.513494099985815e-07, "loss": 0.413, "step": 25764 }, { "epoch": 0.9179329141208116, "grad_norm": 1.1630287721165433, "learning_rate": 3.5104628208104895e-07, "loss": 0.3689, "step": 25765 }, { "epoch": 0.9179685412473056, "grad_norm": 1.6210289299183307, "learning_rate": 3.507432826460788e-07, "loss": 0.5491, "step": 25766 }, { "epoch": 0.9180041683737998, "grad_norm": 1.1528794282997914, "learning_rate": 3.504404116977067e-07, "loss": 0.3271, "step": 25767 }, { "epoch": 0.9180397955002939, "grad_norm": 1.5787555261768513, "learning_rate": 3.501376692399661e-07, "loss": 0.4544, "step": 25768 }, { "epoch": 0.918075422626788, "grad_norm": 1.6421460163622643, "learning_rate": 3.498350552768859e-07, "loss": 0.5285, "step": 25769 }, { "epoch": 0.9181110497532822, "grad_norm": 1.5073155834359027, "learning_rate": 3.495325698124985e-07, "loss": 0.5246, "step": 25770 }, { "epoch": 0.9181466768797762, "grad_norm": 1.4396456767266015, "learning_rate": 3.492302128508285e-07, "loss": 0.5037, "step": 25771 }, { "epoch": 0.9181823040062703, "grad_norm": 1.1605216276462966, "learning_rate": 3.489279843959037e-07, "loss": 0.3861, "step": 25772 }, { "epoch": 0.9182179311327645, "grad_norm": 1.2618986964861654, "learning_rate": 3.4862588445174985e-07, "loss": 0.3726, "step": 25773 }, { "epoch": 0.9182535582592586, "grad_norm": 1.1373620059066665, "learning_rate": 3.48323913022387e-07, "loss": 0.4295, "step": 25774 }, { "epoch": 0.9182891853857527, "grad_norm": 1.518465788862452, "learning_rate": 3.4802207011183864e-07, "loss": 0.4089, "step": 25775 }, { "epoch": 0.9183248125122468, "grad_norm": 1.3683860900582232, "learning_rate": 3.4772035572412154e-07, "loss": 0.4095, "step": 25776 }, { "epoch": 0.9183604396387409, "grad_norm": 1.2215659646652304, "learning_rate": 3.4741876986325586e-07, "loss": 0.3682, "step": 25777 }, { "epoch": 0.918396066765235, "grad_norm": 1.5366693576639199, "learning_rate": 3.47117312533255e-07, "loss": 0.5051, "step": 25778 }, { "epoch": 0.9184316938917292, "grad_norm": 1.2726596859312698, "learning_rate": 3.4681598373813465e-07, "loss": 0.3975, "step": 25779 }, { "epoch": 0.9184673210182233, "grad_norm": 1.3021442225755053, "learning_rate": 3.4651478348190604e-07, "loss": 0.4202, "step": 25780 }, { "epoch": 0.9185029481447173, "grad_norm": 1.5999109084462266, "learning_rate": 3.4621371176858043e-07, "loss": 0.5354, "step": 25781 }, { "epoch": 0.9185385752712115, "grad_norm": 1.7975708044394996, "learning_rate": 3.459127686021657e-07, "loss": 0.5449, "step": 25782 }, { "epoch": 0.9185742023977056, "grad_norm": 1.771958743634794, "learning_rate": 3.45611953986672e-07, "loss": 0.6013, "step": 25783 }, { "epoch": 0.9186098295241997, "grad_norm": 1.28894729390751, "learning_rate": 3.453112679261028e-07, "loss": 0.3838, "step": 25784 }, { "epoch": 0.9186454566506939, "grad_norm": 1.6865310723020677, "learning_rate": 3.4501071042446263e-07, "loss": 0.5406, "step": 25785 }, { "epoch": 0.9186810837771879, "grad_norm": 1.3021397964041266, "learning_rate": 3.447102814857539e-07, "loss": 0.4871, "step": 25786 }, { "epoch": 0.918716710903682, "grad_norm": 1.1992991532932389, "learning_rate": 3.444099811139745e-07, "loss": 0.4206, "step": 25787 }, { "epoch": 0.9187523380301762, "grad_norm": 1.148863703580996, "learning_rate": 3.4410980931312566e-07, "loss": 0.3619, "step": 25788 }, { "epoch": 0.9187879651566703, "grad_norm": 1.6702646778823642, "learning_rate": 3.4380976608720416e-07, "loss": 0.5376, "step": 25789 }, { "epoch": 0.9188235922831643, "grad_norm": 1.3589550738980463, "learning_rate": 3.4350985144020575e-07, "loss": 0.4406, "step": 25790 }, { "epoch": 0.9188592194096585, "grad_norm": 1.3476671220210479, "learning_rate": 3.432100653761217e-07, "loss": 0.4565, "step": 25791 }, { "epoch": 0.9188948465361526, "grad_norm": 1.5124635495302547, "learning_rate": 3.4291040789894647e-07, "loss": 0.5292, "step": 25792 }, { "epoch": 0.9189304736626467, "grad_norm": 1.5772462688761346, "learning_rate": 3.426108790126681e-07, "loss": 0.4662, "step": 25793 }, { "epoch": 0.9189661007891409, "grad_norm": 1.2703186491968093, "learning_rate": 3.423114787212767e-07, "loss": 0.3726, "step": 25794 }, { "epoch": 0.919001727915635, "grad_norm": 1.2324662046932333, "learning_rate": 3.420122070287579e-07, "loss": 0.3976, "step": 25795 }, { "epoch": 0.919037355042129, "grad_norm": 1.727786147276813, "learning_rate": 3.4171306393909643e-07, "loss": 0.5579, "step": 25796 }, { "epoch": 0.9190729821686232, "grad_norm": 1.0735105901921373, "learning_rate": 3.4141404945627675e-07, "loss": 0.4183, "step": 25797 }, { "epoch": 0.9191086092951173, "grad_norm": 1.61746107623002, "learning_rate": 3.411151635842802e-07, "loss": 0.4071, "step": 25798 }, { "epoch": 0.9191442364216114, "grad_norm": 1.4186685095830522, "learning_rate": 3.408164063270869e-07, "loss": 0.4018, "step": 25799 }, { "epoch": 0.9191798635481055, "grad_norm": 1.2564809339376153, "learning_rate": 3.4051777768867364e-07, "loss": 0.496, "step": 25800 }, { "epoch": 0.9192154906745996, "grad_norm": 1.2625771771598444, "learning_rate": 3.4021927767301843e-07, "loss": 0.3962, "step": 25801 }, { "epoch": 0.9192511178010937, "grad_norm": 1.8425878401870417, "learning_rate": 3.399209062840947e-07, "loss": 0.4643, "step": 25802 }, { "epoch": 0.9192867449275879, "grad_norm": 1.481308567469421, "learning_rate": 3.396226635258759e-07, "loss": 0.5009, "step": 25803 }, { "epoch": 0.919322372054082, "grad_norm": 1.1736022147779528, "learning_rate": 3.393245494023345e-07, "loss": 0.4055, "step": 25804 }, { "epoch": 0.919357999180576, "grad_norm": 1.3599567809883915, "learning_rate": 3.390265639174384e-07, "loss": 0.4815, "step": 25805 }, { "epoch": 0.9193936263070702, "grad_norm": 1.2597760649715328, "learning_rate": 3.387287070751566e-07, "loss": 0.4124, "step": 25806 }, { "epoch": 0.9194292534335643, "grad_norm": 1.4995318199394156, "learning_rate": 3.384309788794549e-07, "loss": 0.5197, "step": 25807 }, { "epoch": 0.9194648805600585, "grad_norm": 1.60959119607454, "learning_rate": 3.381333793342978e-07, "loss": 0.5673, "step": 25808 }, { "epoch": 0.9195005076865526, "grad_norm": 1.7077215683322846, "learning_rate": 3.378359084436478e-07, "loss": 0.5229, "step": 25809 }, { "epoch": 0.9195361348130466, "grad_norm": 1.306460909654652, "learning_rate": 3.375385662114661e-07, "loss": 0.3819, "step": 25810 }, { "epoch": 0.9195717619395408, "grad_norm": 1.2950345916653223, "learning_rate": 3.3724135264171063e-07, "loss": 0.4257, "step": 25811 }, { "epoch": 0.9196073890660349, "grad_norm": 1.2180236903662012, "learning_rate": 3.3694426773834276e-07, "loss": 0.4191, "step": 25812 }, { "epoch": 0.919643016192529, "grad_norm": 1.3529404321972676, "learning_rate": 3.3664731150531484e-07, "loss": 0.3966, "step": 25813 }, { "epoch": 0.9196786433190232, "grad_norm": 1.1989641679353817, "learning_rate": 3.3635048394658364e-07, "loss": 0.4366, "step": 25814 }, { "epoch": 0.9197142704455172, "grad_norm": 1.179868299433796, "learning_rate": 3.3605378506609834e-07, "loss": 0.3868, "step": 25815 }, { "epoch": 0.9197498975720113, "grad_norm": 1.3355628231812688, "learning_rate": 3.357572148678134e-07, "loss": 0.3816, "step": 25816 }, { "epoch": 0.9197855246985055, "grad_norm": 1.3323180473439773, "learning_rate": 3.354607733556747e-07, "loss": 0.381, "step": 25817 }, { "epoch": 0.9198211518249996, "grad_norm": 1.3008700519212522, "learning_rate": 3.3516446053363015e-07, "loss": 0.4132, "step": 25818 }, { "epoch": 0.9198567789514936, "grad_norm": 1.739172070275998, "learning_rate": 3.348682764056277e-07, "loss": 0.5083, "step": 25819 }, { "epoch": 0.9198924060779878, "grad_norm": 2.545406640622553, "learning_rate": 3.345722209756086e-07, "loss": 0.5216, "step": 25820 }, { "epoch": 0.9199280332044819, "grad_norm": 1.4051779331429675, "learning_rate": 3.342762942475175e-07, "loss": 0.5078, "step": 25821 }, { "epoch": 0.919963660330976, "grad_norm": 1.2802967177503852, "learning_rate": 3.3398049622529237e-07, "loss": 0.4213, "step": 25822 }, { "epoch": 0.9199992874574702, "grad_norm": 1.584716266295304, "learning_rate": 3.336848269128734e-07, "loss": 0.5131, "step": 25823 }, { "epoch": 0.9200349145839642, "grad_norm": 1.2750349377035293, "learning_rate": 3.3338928631419745e-07, "loss": 0.3769, "step": 25824 }, { "epoch": 0.9200705417104583, "grad_norm": 1.5121794387808487, "learning_rate": 3.3309387443319906e-07, "loss": 0.5254, "step": 25825 }, { "epoch": 0.9201061688369525, "grad_norm": 1.3094985160066617, "learning_rate": 3.3279859127381297e-07, "loss": 0.4026, "step": 25826 }, { "epoch": 0.9201417959634466, "grad_norm": 1.6877654829829407, "learning_rate": 3.3250343683997046e-07, "loss": 0.4996, "step": 25827 }, { "epoch": 0.9201774230899407, "grad_norm": 1.341526892300723, "learning_rate": 3.3220841113560165e-07, "loss": 0.4882, "step": 25828 }, { "epoch": 0.9202130502164348, "grad_norm": 1.1522680862714905, "learning_rate": 3.319135141646368e-07, "loss": 0.4129, "step": 25829 }, { "epoch": 0.9202486773429289, "grad_norm": 1.1812291070123675, "learning_rate": 3.3161874593099943e-07, "loss": 0.3477, "step": 25830 }, { "epoch": 0.920284304469423, "grad_norm": 1.569646972855156, "learning_rate": 3.3132410643861637e-07, "loss": 0.5413, "step": 25831 }, { "epoch": 0.9203199315959172, "grad_norm": 1.4900974429743286, "learning_rate": 3.310295956914122e-07, "loss": 0.3753, "step": 25832 }, { "epoch": 0.9203555587224113, "grad_norm": 1.197747385857007, "learning_rate": 3.3073521369330616e-07, "loss": 0.4068, "step": 25833 }, { "epoch": 0.9203911858489053, "grad_norm": 1.360432382310643, "learning_rate": 3.3044096044821836e-07, "loss": 0.424, "step": 25834 }, { "epoch": 0.9204268129753995, "grad_norm": 1.3451912388502791, "learning_rate": 3.301468359600679e-07, "loss": 0.4332, "step": 25835 }, { "epoch": 0.9204624401018936, "grad_norm": 1.831405018684177, "learning_rate": 3.2985284023277274e-07, "loss": 0.4723, "step": 25836 }, { "epoch": 0.9204980672283877, "grad_norm": 1.6119148952608653, "learning_rate": 3.295589732702453e-07, "loss": 0.4964, "step": 25837 }, { "epoch": 0.9205336943548819, "grad_norm": 1.230940667430554, "learning_rate": 3.292652350763992e-07, "loss": 0.4155, "step": 25838 }, { "epoch": 0.9205693214813759, "grad_norm": 1.638701405487046, "learning_rate": 3.289716256551456e-07, "loss": 0.5175, "step": 25839 }, { "epoch": 0.92060494860787, "grad_norm": 1.5083786014825549, "learning_rate": 3.2867814501039484e-07, "loss": 0.592, "step": 25840 }, { "epoch": 0.9206405757343642, "grad_norm": 1.2990914085244565, "learning_rate": 3.2838479314605373e-07, "loss": 0.3718, "step": 25841 }, { "epoch": 0.9206762028608583, "grad_norm": 1.2878554785623928, "learning_rate": 3.280915700660292e-07, "loss": 0.3882, "step": 25842 }, { "epoch": 0.9207118299873523, "grad_norm": 1.2571447927197203, "learning_rate": 3.2779847577422696e-07, "loss": 0.4086, "step": 25843 }, { "epoch": 0.9207474571138465, "grad_norm": 1.8414365115140836, "learning_rate": 3.2750551027454727e-07, "loss": 0.5083, "step": 25844 }, { "epoch": 0.9207830842403406, "grad_norm": 1.313823941885702, "learning_rate": 3.272126735708936e-07, "loss": 0.4386, "step": 25845 }, { "epoch": 0.9208187113668347, "grad_norm": 1.556085760270356, "learning_rate": 3.269199656671629e-07, "loss": 0.5525, "step": 25846 }, { "epoch": 0.9208543384933289, "grad_norm": 1.4817903527038825, "learning_rate": 3.266273865672553e-07, "loss": 0.5005, "step": 25847 }, { "epoch": 0.920889965619823, "grad_norm": 1.2354010057632052, "learning_rate": 3.263349362750645e-07, "loss": 0.391, "step": 25848 }, { "epoch": 0.920925592746317, "grad_norm": 1.3784328751869002, "learning_rate": 3.2604261479448507e-07, "loss": 0.3885, "step": 25849 }, { "epoch": 0.9209612198728112, "grad_norm": 1.5138435077403234, "learning_rate": 3.2575042212940944e-07, "loss": 0.512, "step": 25850 }, { "epoch": 0.9209968469993053, "grad_norm": 1.6554146812729127, "learning_rate": 3.254583582837312e-07, "loss": 0.4762, "step": 25851 }, { "epoch": 0.9210324741257994, "grad_norm": 1.4964479809713922, "learning_rate": 3.2516642326133496e-07, "loss": 0.5126, "step": 25852 }, { "epoch": 0.9210681012522935, "grad_norm": 1.2396217623449168, "learning_rate": 3.248746170661099e-07, "loss": 0.4452, "step": 25853 }, { "epoch": 0.9211037283787876, "grad_norm": 1.37232643270069, "learning_rate": 3.2458293970194285e-07, "loss": 0.374, "step": 25854 }, { "epoch": 0.9211393555052817, "grad_norm": 1.2808826636468649, "learning_rate": 3.2429139117271634e-07, "loss": 0.4801, "step": 25855 }, { "epoch": 0.9211749826317759, "grad_norm": 1.397347733388023, "learning_rate": 3.239999714823128e-07, "loss": 0.3785, "step": 25856 }, { "epoch": 0.92121060975827, "grad_norm": 1.6670364604450618, "learning_rate": 3.2370868063461234e-07, "loss": 0.4739, "step": 25857 }, { "epoch": 0.921246236884764, "grad_norm": 1.7175407434278178, "learning_rate": 3.234175186334954e-07, "loss": 0.5377, "step": 25858 }, { "epoch": 0.9212818640112582, "grad_norm": 1.283593970088104, "learning_rate": 3.2312648548283643e-07, "loss": 0.3993, "step": 25859 }, { "epoch": 0.9213174911377523, "grad_norm": 1.425059945296258, "learning_rate": 3.228355811865136e-07, "loss": 0.4305, "step": 25860 }, { "epoch": 0.9213531182642464, "grad_norm": 2.0005792669433653, "learning_rate": 3.2254480574839706e-07, "loss": 0.4631, "step": 25861 }, { "epoch": 0.9213887453907406, "grad_norm": 1.7373247293430354, "learning_rate": 3.222541591723627e-07, "loss": 0.6058, "step": 25862 }, { "epoch": 0.9214243725172346, "grad_norm": 1.6688275184137698, "learning_rate": 3.219636414622751e-07, "loss": 0.3772, "step": 25863 }, { "epoch": 0.9214599996437287, "grad_norm": 1.2812724481909228, "learning_rate": 3.216732526220079e-07, "loss": 0.4262, "step": 25864 }, { "epoch": 0.9214956267702229, "grad_norm": 1.4589773355575644, "learning_rate": 3.2138299265542794e-07, "loss": 0.3904, "step": 25865 }, { "epoch": 0.921531253896717, "grad_norm": 1.8320656097213728, "learning_rate": 3.210928615663966e-07, "loss": 0.5039, "step": 25866 }, { "epoch": 0.921566881023211, "grad_norm": 1.8979485535850098, "learning_rate": 3.2080285935877977e-07, "loss": 0.4991, "step": 25867 }, { "epoch": 0.9216025081497052, "grad_norm": 2.216448140261925, "learning_rate": 3.2051298603643754e-07, "loss": 0.4979, "step": 25868 }, { "epoch": 0.9216381352761993, "grad_norm": 1.6617112377295562, "learning_rate": 3.202232416032314e-07, "loss": 0.4375, "step": 25869 }, { "epoch": 0.9216737624026934, "grad_norm": 1.8309390232348897, "learning_rate": 3.1993362606301703e-07, "loss": 0.486, "step": 25870 }, { "epoch": 0.9217093895291876, "grad_norm": 1.5867166531942232, "learning_rate": 3.196441394196537e-07, "loss": 0.5395, "step": 25871 }, { "epoch": 0.9217450166556816, "grad_norm": 1.3458086835629237, "learning_rate": 3.193547816769937e-07, "loss": 0.3843, "step": 25872 }, { "epoch": 0.9217806437821757, "grad_norm": 1.6427844031868575, "learning_rate": 3.19065552838892e-07, "loss": 0.4765, "step": 25873 }, { "epoch": 0.9218162709086699, "grad_norm": 1.2547054595893667, "learning_rate": 3.1877645290919854e-07, "loss": 0.5177, "step": 25874 }, { "epoch": 0.921851898035164, "grad_norm": 1.4536539775695125, "learning_rate": 3.184874818917649e-07, "loss": 0.4174, "step": 25875 }, { "epoch": 0.9218875251616581, "grad_norm": 3.539737595109672, "learning_rate": 3.1819863979043576e-07, "loss": 0.3742, "step": 25876 }, { "epoch": 0.9219231522881522, "grad_norm": 1.131751109481546, "learning_rate": 3.1790992660905906e-07, "loss": 0.3829, "step": 25877 }, { "epoch": 0.9219587794146463, "grad_norm": 1.5017636858124297, "learning_rate": 3.1762134235147844e-07, "loss": 0.4227, "step": 25878 }, { "epoch": 0.9219944065411404, "grad_norm": 1.449075618974651, "learning_rate": 3.1733288702153755e-07, "loss": 0.5552, "step": 25879 }, { "epoch": 0.9220300336676346, "grad_norm": 1.3937690819764674, "learning_rate": 3.1704456062307877e-07, "loss": 0.3989, "step": 25880 }, { "epoch": 0.9220656607941287, "grad_norm": 1.4446060284539055, "learning_rate": 3.1675636315993795e-07, "loss": 0.4237, "step": 25881 }, { "epoch": 0.9221012879206227, "grad_norm": 1.9836183093201796, "learning_rate": 3.164682946359554e-07, "loss": 0.5431, "step": 25882 }, { "epoch": 0.9221369150471169, "grad_norm": 1.325161955580728, "learning_rate": 3.161803550549647e-07, "loss": 0.3895, "step": 25883 }, { "epoch": 0.922172542173611, "grad_norm": 1.2522579768417497, "learning_rate": 3.1589254442080276e-07, "loss": 0.396, "step": 25884 }, { "epoch": 0.9222081693001051, "grad_norm": 1.441890767397139, "learning_rate": 3.156048627372987e-07, "loss": 0.4511, "step": 25885 }, { "epoch": 0.9222437964265993, "grad_norm": 1.7125457904813846, "learning_rate": 3.1531731000828624e-07, "loss": 0.5197, "step": 25886 }, { "epoch": 0.9222794235530933, "grad_norm": 1.5047774852498406, "learning_rate": 3.1502988623759115e-07, "loss": 0.3809, "step": 25887 }, { "epoch": 0.9223150506795874, "grad_norm": 1.3602400394846295, "learning_rate": 3.1474259142904363e-07, "loss": 0.4072, "step": 25888 }, { "epoch": 0.9223506778060816, "grad_norm": 1.4887929545925833, "learning_rate": 3.1445542558646846e-07, "loss": 0.4638, "step": 25889 }, { "epoch": 0.9223863049325757, "grad_norm": 2.5458806864165515, "learning_rate": 3.1416838871368925e-07, "loss": 0.4878, "step": 25890 }, { "epoch": 0.9224219320590697, "grad_norm": 1.7395973501889157, "learning_rate": 3.138814808145274e-07, "loss": 0.5338, "step": 25891 }, { "epoch": 0.9224575591855639, "grad_norm": 1.4725959164355358, "learning_rate": 3.135947018928043e-07, "loss": 0.3913, "step": 25892 }, { "epoch": 0.922493186312058, "grad_norm": 1.3198786422083684, "learning_rate": 3.1330805195233684e-07, "loss": 0.445, "step": 25893 }, { "epoch": 0.9225288134385521, "grad_norm": 1.663617968659001, "learning_rate": 3.130215309969442e-07, "loss": 0.3652, "step": 25894 }, { "epoch": 0.9225644405650463, "grad_norm": 1.212925510402851, "learning_rate": 3.127351390304412e-07, "loss": 0.401, "step": 25895 }, { "epoch": 0.9226000676915403, "grad_norm": 1.3127093148488593, "learning_rate": 3.124488760566402e-07, "loss": 0.5573, "step": 25896 }, { "epoch": 0.9226356948180344, "grad_norm": 1.8106905797991517, "learning_rate": 3.121627420793538e-07, "loss": 0.544, "step": 25897 }, { "epoch": 0.9226713219445286, "grad_norm": 1.3499675392881967, "learning_rate": 3.118767371023912e-07, "loss": 0.3816, "step": 25898 }, { "epoch": 0.9227069490710227, "grad_norm": 1.5941422245890384, "learning_rate": 3.115908611295626e-07, "loss": 0.4686, "step": 25899 }, { "epoch": 0.9227425761975168, "grad_norm": 1.3124808978761884, "learning_rate": 3.113051141646717e-07, "loss": 0.3831, "step": 25900 }, { "epoch": 0.922778203324011, "grad_norm": 1.1310661057457159, "learning_rate": 3.110194962115265e-07, "loss": 0.4269, "step": 25901 }, { "epoch": 0.922813830450505, "grad_norm": 1.4712480623879087, "learning_rate": 3.1073400727392956e-07, "loss": 0.4717, "step": 25902 }, { "epoch": 0.9228494575769991, "grad_norm": 1.1890668870672878, "learning_rate": 3.1044864735568003e-07, "loss": 0.4143, "step": 25903 }, { "epoch": 0.9228850847034933, "grad_norm": 1.5497817381788976, "learning_rate": 3.1016341646058155e-07, "loss": 0.5212, "step": 25904 }, { "epoch": 0.9229207118299874, "grad_norm": 1.5953435420939566, "learning_rate": 3.098783145924278e-07, "loss": 0.5374, "step": 25905 }, { "epoch": 0.9229563389564814, "grad_norm": 1.4151885782256624, "learning_rate": 3.0959334175501897e-07, "loss": 0.4533, "step": 25906 }, { "epoch": 0.9229919660829756, "grad_norm": 1.5264705605716302, "learning_rate": 3.0930849795214657e-07, "loss": 0.5087, "step": 25907 }, { "epoch": 0.9230275932094697, "grad_norm": 1.40829960361515, "learning_rate": 3.090237831876053e-07, "loss": 0.5046, "step": 25908 }, { "epoch": 0.9230632203359638, "grad_norm": 1.9304788052598243, "learning_rate": 3.0873919746518546e-07, "loss": 0.4795, "step": 25909 }, { "epoch": 0.923098847462458, "grad_norm": 1.2992244080761728, "learning_rate": 3.0845474078867733e-07, "loss": 0.4227, "step": 25910 }, { "epoch": 0.923134474588952, "grad_norm": 1.4155308207534052, "learning_rate": 3.081704131618701e-07, "loss": 0.5044, "step": 25911 }, { "epoch": 0.9231701017154461, "grad_norm": 1.230536642980576, "learning_rate": 3.078862145885453e-07, "loss": 0.3956, "step": 25912 }, { "epoch": 0.9232057288419403, "grad_norm": 1.3551090175134914, "learning_rate": 3.0760214507249197e-07, "loss": 0.4675, "step": 25913 }, { "epoch": 0.9232413559684344, "grad_norm": 1.4073217212782767, "learning_rate": 3.073182046174894e-07, "loss": 0.4854, "step": 25914 }, { "epoch": 0.9232769830949284, "grad_norm": 1.124410246797165, "learning_rate": 3.070343932273201e-07, "loss": 0.3411, "step": 25915 }, { "epoch": 0.9233126102214226, "grad_norm": 1.6044029123072412, "learning_rate": 3.0675071090576325e-07, "loss": 0.5393, "step": 25916 }, { "epoch": 0.9233482373479167, "grad_norm": 1.5464948972164263, "learning_rate": 3.0646715765659583e-07, "loss": 0.5197, "step": 25917 }, { "epoch": 0.9233838644744108, "grad_norm": 2.86306119187265, "learning_rate": 3.0618373348359265e-07, "loss": 0.5285, "step": 25918 }, { "epoch": 0.923419491600905, "grad_norm": 1.178600713830026, "learning_rate": 3.0590043839053064e-07, "loss": 0.3757, "step": 25919 }, { "epoch": 0.923455118727399, "grad_norm": 1.4920938904223762, "learning_rate": 3.056172723811779e-07, "loss": 0.4877, "step": 25920 }, { "epoch": 0.9234907458538931, "grad_norm": 1.2067911854550881, "learning_rate": 3.0533423545930807e-07, "loss": 0.4238, "step": 25921 }, { "epoch": 0.9235263729803873, "grad_norm": 1.669778503264346, "learning_rate": 3.050513276286893e-07, "loss": 0.4386, "step": 25922 }, { "epoch": 0.9235620001068814, "grad_norm": 1.4408023887021773, "learning_rate": 3.0476854889308737e-07, "loss": 0.4658, "step": 25923 }, { "epoch": 0.9235976272333755, "grad_norm": 1.404672748181115, "learning_rate": 3.044858992562694e-07, "loss": 0.3663, "step": 25924 }, { "epoch": 0.9236332543598696, "grad_norm": 1.5877251093809992, "learning_rate": 3.0420337872199776e-07, "loss": 0.4807, "step": 25925 }, { "epoch": 0.9236688814863637, "grad_norm": 1.3353899832522358, "learning_rate": 3.039209872940363e-07, "loss": 0.4968, "step": 25926 }, { "epoch": 0.9237045086128578, "grad_norm": 1.6458061660253953, "learning_rate": 3.0363872497614413e-07, "loss": 0.5437, "step": 25927 }, { "epoch": 0.923740135739352, "grad_norm": 1.5090738315261143, "learning_rate": 3.033565917720793e-07, "loss": 0.4298, "step": 25928 }, { "epoch": 0.9237757628658461, "grad_norm": 1.210882485474746, "learning_rate": 3.0307458768559894e-07, "loss": 0.4752, "step": 25929 }, { "epoch": 0.9238113899923401, "grad_norm": 1.5476127355869937, "learning_rate": 3.0279271272045776e-07, "loss": 0.5297, "step": 25930 }, { "epoch": 0.9238470171188343, "grad_norm": 1.6312276740170624, "learning_rate": 3.0251096688040936e-07, "loss": 0.5552, "step": 25931 }, { "epoch": 0.9238826442453284, "grad_norm": 1.2875035346494939, "learning_rate": 3.022293501692064e-07, "loss": 0.4322, "step": 25932 }, { "epoch": 0.9239182713718225, "grad_norm": 1.3754549156581293, "learning_rate": 3.0194786259059806e-07, "loss": 0.4216, "step": 25933 }, { "epoch": 0.9239538984983167, "grad_norm": 1.5231267003717979, "learning_rate": 3.016665041483313e-07, "loss": 0.513, "step": 25934 }, { "epoch": 0.9239895256248107, "grad_norm": 1.4689704813176916, "learning_rate": 3.013852748461554e-07, "loss": 0.4177, "step": 25935 }, { "epoch": 0.9240251527513048, "grad_norm": 1.5277092502849325, "learning_rate": 3.0110417468781184e-07, "loss": 0.4034, "step": 25936 }, { "epoch": 0.924060779877799, "grad_norm": 1.5004864758996794, "learning_rate": 3.0082320367704756e-07, "loss": 0.4682, "step": 25937 }, { "epoch": 0.9240964070042931, "grad_norm": 1.387419066894383, "learning_rate": 3.0054236181759845e-07, "loss": 0.523, "step": 25938 }, { "epoch": 0.9241320341307871, "grad_norm": 1.2460013317626413, "learning_rate": 3.002616491132093e-07, "loss": 0.4159, "step": 25939 }, { "epoch": 0.9241676612572813, "grad_norm": 1.5011475149286833, "learning_rate": 2.9998106556761606e-07, "loss": 0.5787, "step": 25940 }, { "epoch": 0.9242032883837754, "grad_norm": 1.587242318644637, "learning_rate": 2.9970061118455573e-07, "loss": 0.4734, "step": 25941 }, { "epoch": 0.9242389155102695, "grad_norm": 1.3540508339957136, "learning_rate": 2.9942028596775975e-07, "loss": 0.4127, "step": 25942 }, { "epoch": 0.9242745426367637, "grad_norm": 1.9780218146609179, "learning_rate": 2.9914008992096513e-07, "loss": 0.4402, "step": 25943 }, { "epoch": 0.9243101697632577, "grad_norm": 1.452199029863809, "learning_rate": 2.9886002304789886e-07, "loss": 0.4812, "step": 25944 }, { "epoch": 0.9243457968897518, "grad_norm": 1.299278818937764, "learning_rate": 2.9858008535229245e-07, "loss": 0.4592, "step": 25945 }, { "epoch": 0.924381424016246, "grad_norm": 1.3446873408262348, "learning_rate": 2.983002768378729e-07, "loss": 0.5602, "step": 25946 }, { "epoch": 0.9244170511427401, "grad_norm": 1.7468304201331084, "learning_rate": 2.980205975083661e-07, "loss": 0.5702, "step": 25947 }, { "epoch": 0.9244526782692342, "grad_norm": 1.5464077952450663, "learning_rate": 2.97741047367498e-07, "loss": 0.5805, "step": 25948 }, { "epoch": 0.9244883053957283, "grad_norm": 1.7610775897711375, "learning_rate": 2.974616264189878e-07, "loss": 0.4683, "step": 25949 }, { "epoch": 0.9245239325222224, "grad_norm": 1.2487937022399758, "learning_rate": 2.971823346665581e-07, "loss": 0.3734, "step": 25950 }, { "epoch": 0.9245595596487165, "grad_norm": 1.2758108291116688, "learning_rate": 2.9690317211392814e-07, "loss": 0.4353, "step": 25951 }, { "epoch": 0.9245951867752107, "grad_norm": 1.6398137174876, "learning_rate": 2.966241387648128e-07, "loss": 0.5925, "step": 25952 }, { "epoch": 0.9246308139017048, "grad_norm": 1.752659838540849, "learning_rate": 2.9634523462293005e-07, "loss": 0.5387, "step": 25953 }, { "epoch": 0.9246664410281988, "grad_norm": 1.3456806530381844, "learning_rate": 2.960664596919938e-07, "loss": 0.3874, "step": 25954 }, { "epoch": 0.924702068154693, "grad_norm": 1.7664082914641603, "learning_rate": 2.9578781397571534e-07, "loss": 0.6414, "step": 25955 }, { "epoch": 0.9247376952811871, "grad_norm": 1.495632402200383, "learning_rate": 2.955092974778051e-07, "loss": 0.4616, "step": 25956 }, { "epoch": 0.9247733224076812, "grad_norm": 2.1800433876058105, "learning_rate": 2.9523091020197013e-07, "loss": 0.5592, "step": 25957 }, { "epoch": 0.9248089495341754, "grad_norm": 1.1154862239328847, "learning_rate": 2.949526521519197e-07, "loss": 0.4086, "step": 25958 }, { "epoch": 0.9248445766606694, "grad_norm": 1.3605885364651633, "learning_rate": 2.946745233313586e-07, "loss": 0.4356, "step": 25959 }, { "epoch": 0.9248802037871635, "grad_norm": 1.4202947500856402, "learning_rate": 2.943965237439883e-07, "loss": 0.4639, "step": 25960 }, { "epoch": 0.9249158309136577, "grad_norm": 1.5061089444068891, "learning_rate": 2.9411865339351253e-07, "loss": 0.4856, "step": 25961 }, { "epoch": 0.9249514580401518, "grad_norm": 1.5417910045200822, "learning_rate": 2.9384091228363164e-07, "loss": 0.4404, "step": 25962 }, { "epoch": 0.9249870851666459, "grad_norm": 1.5243783003110687, "learning_rate": 2.9356330041804384e-07, "loss": 0.5415, "step": 25963 }, { "epoch": 0.92502271229314, "grad_norm": 1.5495777636618597, "learning_rate": 2.9328581780044383e-07, "loss": 0.4081, "step": 25964 }, { "epoch": 0.9250583394196341, "grad_norm": 1.186528781189558, "learning_rate": 2.930084644345288e-07, "loss": 0.385, "step": 25965 }, { "epoch": 0.9250939665461282, "grad_norm": 1.3616376442807734, "learning_rate": 2.9273124032399015e-07, "loss": 0.4415, "step": 25966 }, { "epoch": 0.9251295936726224, "grad_norm": 1.4484946396563347, "learning_rate": 2.924541454725194e-07, "loss": 0.4135, "step": 25967 }, { "epoch": 0.9251652207991165, "grad_norm": 1.3306338346008306, "learning_rate": 2.921771798838069e-07, "loss": 0.4098, "step": 25968 }, { "epoch": 0.9252008479256105, "grad_norm": 1.329909154044164, "learning_rate": 2.9190034356154085e-07, "loss": 0.4926, "step": 25969 }, { "epoch": 0.9252364750521047, "grad_norm": 1.1377886354348878, "learning_rate": 2.916236365094083e-07, "loss": 0.4241, "step": 25970 }, { "epoch": 0.9252721021785988, "grad_norm": 1.8367600554117773, "learning_rate": 2.913470587310918e-07, "loss": 0.5027, "step": 25971 }, { "epoch": 0.9253077293050929, "grad_norm": 1.2309338247141826, "learning_rate": 2.910706102302763e-07, "loss": 0.4029, "step": 25972 }, { "epoch": 0.925343356431587, "grad_norm": 1.2292071072918422, "learning_rate": 2.9079429101064093e-07, "loss": 0.4246, "step": 25973 }, { "epoch": 0.9253789835580811, "grad_norm": 1.3958922572730992, "learning_rate": 2.905181010758662e-07, "loss": 0.401, "step": 25974 }, { "epoch": 0.9254146106845752, "grad_norm": 1.8637128269347043, "learning_rate": 2.90242040429628e-07, "loss": 0.3864, "step": 25975 }, { "epoch": 0.9254502378110694, "grad_norm": 1.8499325662452413, "learning_rate": 2.8996610907560565e-07, "loss": 0.5518, "step": 25976 }, { "epoch": 0.9254858649375635, "grad_norm": 1.3161397124631533, "learning_rate": 2.896903070174706e-07, "loss": 0.4073, "step": 25977 }, { "epoch": 0.9255214920640575, "grad_norm": 1.3847671162189592, "learning_rate": 2.894146342588977e-07, "loss": 0.4185, "step": 25978 }, { "epoch": 0.9255571191905517, "grad_norm": 1.1779997259008732, "learning_rate": 2.8913909080355405e-07, "loss": 0.4088, "step": 25979 }, { "epoch": 0.9255927463170458, "grad_norm": 2.2922833449435887, "learning_rate": 2.8886367665511116e-07, "loss": 0.4481, "step": 25980 }, { "epoch": 0.9256283734435399, "grad_norm": 1.645423261005532, "learning_rate": 2.885883918172383e-07, "loss": 0.472, "step": 25981 }, { "epoch": 0.9256640005700341, "grad_norm": 1.321868420834432, "learning_rate": 2.88313236293597e-07, "loss": 0.4801, "step": 25982 }, { "epoch": 0.9256996276965281, "grad_norm": 1.4632300740513429, "learning_rate": 2.8803821008785203e-07, "loss": 0.4189, "step": 25983 }, { "epoch": 0.9257352548230222, "grad_norm": 1.3368010082899529, "learning_rate": 2.8776331320366836e-07, "loss": 0.5344, "step": 25984 }, { "epoch": 0.9257708819495164, "grad_norm": 1.2134348174288485, "learning_rate": 2.874885456447041e-07, "loss": 0.3884, "step": 25985 }, { "epoch": 0.9258065090760105, "grad_norm": 1.2971848695337695, "learning_rate": 2.872139074146185e-07, "loss": 0.3784, "step": 25986 }, { "epoch": 0.9258421362025046, "grad_norm": 1.6524353235428555, "learning_rate": 2.8693939851706876e-07, "loss": 0.5359, "step": 25987 }, { "epoch": 0.9258777633289987, "grad_norm": 1.5246775813083513, "learning_rate": 2.8666501895570965e-07, "loss": 0.5262, "step": 25988 }, { "epoch": 0.9259133904554928, "grad_norm": 1.5319711910220724, "learning_rate": 2.863907687341949e-07, "loss": 0.4238, "step": 25989 }, { "epoch": 0.9259490175819869, "grad_norm": 1.4922188173991313, "learning_rate": 2.861166478561761e-07, "loss": 0.5152, "step": 25990 }, { "epoch": 0.9259846447084811, "grad_norm": 1.3504459659204038, "learning_rate": 2.858426563253036e-07, "loss": 0.4241, "step": 25991 }, { "epoch": 0.9260202718349752, "grad_norm": 1.189748101116235, "learning_rate": 2.855687941452279e-07, "loss": 0.3585, "step": 25992 }, { "epoch": 0.9260558989614692, "grad_norm": 1.4619956056919712, "learning_rate": 2.852950613195915e-07, "loss": 0.5186, "step": 25993 }, { "epoch": 0.9260915260879634, "grad_norm": 1.2133042506179905, "learning_rate": 2.850214578520438e-07, "loss": 0.3792, "step": 25994 }, { "epoch": 0.9261271532144575, "grad_norm": 1.467863230241543, "learning_rate": 2.8474798374622416e-07, "loss": 0.4229, "step": 25995 }, { "epoch": 0.9261627803409516, "grad_norm": 1.1852822819491016, "learning_rate": 2.8447463900577623e-07, "loss": 0.4219, "step": 25996 }, { "epoch": 0.9261984074674458, "grad_norm": 1.3971523792425449, "learning_rate": 2.842014236343382e-07, "loss": 0.466, "step": 25997 }, { "epoch": 0.9262340345939398, "grad_norm": 1.5357837055367125, "learning_rate": 2.839283376355506e-07, "loss": 0.5627, "step": 25998 }, { "epoch": 0.9262696617204339, "grad_norm": 1.3899986338430055, "learning_rate": 2.8365538101304713e-07, "loss": 0.4852, "step": 25999 }, { "epoch": 0.9263052888469281, "grad_norm": 1.2694991161107974, "learning_rate": 2.8338255377046375e-07, "loss": 0.3662, "step": 26000 }, { "epoch": 0.9263409159734222, "grad_norm": 1.5947487816910195, "learning_rate": 2.8310985591143313e-07, "loss": 0.4564, "step": 26001 }, { "epoch": 0.9263765430999162, "grad_norm": 1.6391253544251287, "learning_rate": 2.8283728743958793e-07, "loss": 0.6027, "step": 26002 }, { "epoch": 0.9264121702264104, "grad_norm": 1.3967610060741291, "learning_rate": 2.8256484835855526e-07, "loss": 0.4587, "step": 26003 }, { "epoch": 0.9264477973529045, "grad_norm": 1.6274526624437156, "learning_rate": 2.8229253867196216e-07, "loss": 0.3748, "step": 26004 }, { "epoch": 0.9264834244793986, "grad_norm": 1.3614508511104588, "learning_rate": 2.82020358383438e-07, "loss": 0.3763, "step": 26005 }, { "epoch": 0.9265190516058928, "grad_norm": 1.5755557811725311, "learning_rate": 2.817483074966043e-07, "loss": 0.4271, "step": 26006 }, { "epoch": 0.9265546787323868, "grad_norm": 1.311431524099056, "learning_rate": 2.814763860150849e-07, "loss": 0.3913, "step": 26007 }, { "epoch": 0.9265903058588809, "grad_norm": 1.3986070707109155, "learning_rate": 2.812045939425001e-07, "loss": 0.4063, "step": 26008 }, { "epoch": 0.9266259329853751, "grad_norm": 1.1249706948365927, "learning_rate": 2.8093293128247046e-07, "loss": 0.3847, "step": 26009 }, { "epoch": 0.9266615601118692, "grad_norm": 1.2348528953329234, "learning_rate": 2.806613980386108e-07, "loss": 0.4112, "step": 26010 }, { "epoch": 0.9266971872383633, "grad_norm": 1.4129060154205995, "learning_rate": 2.8038999421453827e-07, "loss": 0.4429, "step": 26011 }, { "epoch": 0.9267328143648574, "grad_norm": 1.6291731309332864, "learning_rate": 2.8011871981386664e-07, "loss": 0.4797, "step": 26012 }, { "epoch": 0.9267684414913515, "grad_norm": 1.273015686256515, "learning_rate": 2.798475748402074e-07, "loss": 0.3467, "step": 26013 }, { "epoch": 0.9268040686178456, "grad_norm": 1.8963840899733893, "learning_rate": 2.795765592971722e-07, "loss": 0.4968, "step": 26014 }, { "epoch": 0.9268396957443398, "grad_norm": 1.3152824347623187, "learning_rate": 2.7930567318836923e-07, "loss": 0.5508, "step": 26015 }, { "epoch": 0.9268753228708339, "grad_norm": 1.4769729428863707, "learning_rate": 2.7903491651740555e-07, "loss": 0.4932, "step": 26016 }, { "epoch": 0.9269109499973279, "grad_norm": 1.206216751674406, "learning_rate": 2.787642892878861e-07, "loss": 0.4113, "step": 26017 }, { "epoch": 0.9269465771238221, "grad_norm": 1.6526853409572857, "learning_rate": 2.784937915034169e-07, "loss": 0.4986, "step": 26018 }, { "epoch": 0.9269822042503162, "grad_norm": 1.4043831501922157, "learning_rate": 2.782234231675951e-07, "loss": 0.5045, "step": 26019 }, { "epoch": 0.9270178313768103, "grad_norm": 1.6048690412385318, "learning_rate": 2.779531842840244e-07, "loss": 0.5282, "step": 26020 }, { "epoch": 0.9270534585033045, "grad_norm": 1.5400259512519363, "learning_rate": 2.776830748563031e-07, "loss": 0.5489, "step": 26021 }, { "epoch": 0.9270890856297985, "grad_norm": 1.274747003829109, "learning_rate": 2.7741309488802717e-07, "loss": 0.361, "step": 26022 }, { "epoch": 0.9271247127562926, "grad_norm": 1.3048917548898884, "learning_rate": 2.771432443827904e-07, "loss": 0.3825, "step": 26023 }, { "epoch": 0.9271603398827868, "grad_norm": 1.1942141318872312, "learning_rate": 2.768735233441888e-07, "loss": 0.3846, "step": 26024 }, { "epoch": 0.9271959670092809, "grad_norm": 1.2879264473902994, "learning_rate": 2.766039317758107e-07, "loss": 0.3847, "step": 26025 }, { "epoch": 0.9272315941357749, "grad_norm": 1.4498769279137622, "learning_rate": 2.7633446968124864e-07, "loss": 0.5471, "step": 26026 }, { "epoch": 0.9272672212622691, "grad_norm": 1.2530107626028173, "learning_rate": 2.760651370640888e-07, "loss": 0.4082, "step": 26027 }, { "epoch": 0.9273028483887632, "grad_norm": 1.2489017600095174, "learning_rate": 2.757959339279181e-07, "loss": 0.4018, "step": 26028 }, { "epoch": 0.9273384755152573, "grad_norm": 1.1981012374217939, "learning_rate": 2.755268602763217e-07, "loss": 0.3939, "step": 26029 }, { "epoch": 0.9273741026417515, "grad_norm": 1.6540974938707838, "learning_rate": 2.752579161128821e-07, "loss": 0.4259, "step": 26030 }, { "epoch": 0.9274097297682455, "grad_norm": 1.66070763519154, "learning_rate": 2.7498910144118206e-07, "loss": 0.3802, "step": 26031 }, { "epoch": 0.9274453568947396, "grad_norm": 1.2848659543533258, "learning_rate": 2.7472041626479763e-07, "loss": 0.3843, "step": 26032 }, { "epoch": 0.9274809840212338, "grad_norm": 1.3274850988942877, "learning_rate": 2.744518605873092e-07, "loss": 0.434, "step": 26033 }, { "epoch": 0.9275166111477279, "grad_norm": 1.39869626925417, "learning_rate": 2.741834344122918e-07, "loss": 0.4485, "step": 26034 }, { "epoch": 0.927552238274222, "grad_norm": 1.3644232298409982, "learning_rate": 2.739151377433191e-07, "loss": 0.359, "step": 26035 }, { "epoch": 0.9275878654007161, "grad_norm": 1.6188253882033898, "learning_rate": 2.7364697058396504e-07, "loss": 0.5042, "step": 26036 }, { "epoch": 0.9276234925272102, "grad_norm": 1.4900238089566642, "learning_rate": 2.7337893293780007e-07, "loss": 0.5137, "step": 26037 }, { "epoch": 0.9276591196537043, "grad_norm": 1.3180795565505428, "learning_rate": 2.7311102480839347e-07, "loss": 0.3759, "step": 26038 }, { "epoch": 0.9276947467801985, "grad_norm": 1.1453134883106995, "learning_rate": 2.728432461993114e-07, "loss": 0.3836, "step": 26039 }, { "epoch": 0.9277303739066926, "grad_norm": 1.5960823736497392, "learning_rate": 2.7257559711412195e-07, "loss": 0.5109, "step": 26040 }, { "epoch": 0.9277660010331866, "grad_norm": 1.3606055592598099, "learning_rate": 2.7230807755638576e-07, "loss": 0.4492, "step": 26041 }, { "epoch": 0.9278016281596808, "grad_norm": 1.2412506292664334, "learning_rate": 2.720406875296677e-07, "loss": 0.3741, "step": 26042 }, { "epoch": 0.9278372552861749, "grad_norm": 1.2401078958321141, "learning_rate": 2.717734270375272e-07, "loss": 0.4458, "step": 26043 }, { "epoch": 0.927872882412669, "grad_norm": 1.4924697412111831, "learning_rate": 2.715062960835235e-07, "loss": 0.3896, "step": 26044 }, { "epoch": 0.9279085095391632, "grad_norm": 1.7546973244443902, "learning_rate": 2.712392946712117e-07, "loss": 0.5624, "step": 26045 }, { "epoch": 0.9279441366656572, "grad_norm": 1.689802133930001, "learning_rate": 2.7097242280415105e-07, "loss": 0.476, "step": 26046 }, { "epoch": 0.9279797637921513, "grad_norm": 1.2636242297370321, "learning_rate": 2.7070568048589094e-07, "loss": 0.3693, "step": 26047 }, { "epoch": 0.9280153909186455, "grad_norm": 1.4903327355627536, "learning_rate": 2.704390677199864e-07, "loss": 0.5352, "step": 26048 }, { "epoch": 0.9280510180451396, "grad_norm": 1.9381986575719483, "learning_rate": 2.701725845099856e-07, "loss": 0.455, "step": 26049 }, { "epoch": 0.9280866451716336, "grad_norm": 2.030839730971074, "learning_rate": 2.699062308594358e-07, "loss": 0.4663, "step": 26050 }, { "epoch": 0.9281222722981278, "grad_norm": 1.2661525350875849, "learning_rate": 2.696400067718874e-07, "loss": 0.4178, "step": 26051 }, { "epoch": 0.9281578994246219, "grad_norm": 1.529299932901594, "learning_rate": 2.6937391225088315e-07, "loss": 0.4486, "step": 26052 }, { "epoch": 0.928193526551116, "grad_norm": 1.8143117496358707, "learning_rate": 2.691079472999669e-07, "loss": 0.5395, "step": 26053 }, { "epoch": 0.9282291536776102, "grad_norm": 1.78819466279049, "learning_rate": 2.688421119226792e-07, "loss": 0.4911, "step": 26054 }, { "epoch": 0.9282647808041042, "grad_norm": 1.376534816759723, "learning_rate": 2.685764061225615e-07, "loss": 0.4453, "step": 26055 }, { "epoch": 0.9283004079305983, "grad_norm": 1.4506970056976451, "learning_rate": 2.6831082990315003e-07, "loss": 0.3794, "step": 26056 }, { "epoch": 0.9283360350570925, "grad_norm": 1.8691721289124574, "learning_rate": 2.680453832679808e-07, "loss": 0.5614, "step": 26057 }, { "epoch": 0.9283716621835866, "grad_norm": 1.3864895815188496, "learning_rate": 2.6778006622059095e-07, "loss": 0.4794, "step": 26058 }, { "epoch": 0.9284072893100807, "grad_norm": 1.6205709054210573, "learning_rate": 2.675148787645121e-07, "loss": 0.3859, "step": 26059 }, { "epoch": 0.9284429164365748, "grad_norm": 1.2245731623311051, "learning_rate": 2.6724982090327477e-07, "loss": 0.3837, "step": 26060 }, { "epoch": 0.9284785435630689, "grad_norm": 1.4319653689983336, "learning_rate": 2.669848926404084e-07, "loss": 0.4979, "step": 26061 }, { "epoch": 0.928514170689563, "grad_norm": 1.2343904057703141, "learning_rate": 2.667200939794423e-07, "loss": 0.3939, "step": 26062 }, { "epoch": 0.9285497978160572, "grad_norm": 1.2698003986109165, "learning_rate": 2.664554249238993e-07, "loss": 0.4096, "step": 26063 }, { "epoch": 0.9285854249425513, "grad_norm": 1.4237813959318426, "learning_rate": 2.6619088547730763e-07, "loss": 0.4681, "step": 26064 }, { "epoch": 0.9286210520690453, "grad_norm": 1.2649029175996298, "learning_rate": 2.6592647564318676e-07, "loss": 0.3902, "step": 26065 }, { "epoch": 0.9286566791955395, "grad_norm": 1.2751892720828903, "learning_rate": 2.656621954250593e-07, "loss": 0.4162, "step": 26066 }, { "epoch": 0.9286923063220336, "grad_norm": 1.2898977899171729, "learning_rate": 2.653980448264426e-07, "loss": 0.4549, "step": 26067 }, { "epoch": 0.9287279334485277, "grad_norm": 1.1866874762705186, "learning_rate": 2.6513402385085706e-07, "loss": 0.3601, "step": 26068 }, { "epoch": 0.9287635605750219, "grad_norm": 1.717398085243087, "learning_rate": 2.6487013250181436e-07, "loss": 0.5354, "step": 26069 }, { "epoch": 0.9287991877015159, "grad_norm": 1.7820444985504684, "learning_rate": 2.646063707828317e-07, "loss": 0.501, "step": 26070 }, { "epoch": 0.92883481482801, "grad_norm": 1.2213591410746611, "learning_rate": 2.643427386974184e-07, "loss": 0.4148, "step": 26071 }, { "epoch": 0.9288704419545042, "grad_norm": 1.2592877568869763, "learning_rate": 2.640792362490863e-07, "loss": 0.3974, "step": 26072 }, { "epoch": 0.9289060690809983, "grad_norm": 1.4025056909494316, "learning_rate": 2.638158634413446e-07, "loss": 0.5676, "step": 26073 }, { "epoch": 0.9289416962074923, "grad_norm": 1.2608705792009263, "learning_rate": 2.6355262027769946e-07, "loss": 0.4149, "step": 26074 }, { "epoch": 0.9289773233339865, "grad_norm": 1.2158284974619098, "learning_rate": 2.632895067616581e-07, "loss": 0.441, "step": 26075 }, { "epoch": 0.9290129504604806, "grad_norm": 1.3399191451254493, "learning_rate": 2.6302652289672104e-07, "loss": 0.4937, "step": 26076 }, { "epoch": 0.9290485775869747, "grad_norm": 1.2996596209911517, "learning_rate": 2.627636686863932e-07, "loss": 0.3916, "step": 26077 }, { "epoch": 0.9290842047134689, "grad_norm": 1.5131562270494148, "learning_rate": 2.625009441341708e-07, "loss": 0.4467, "step": 26078 }, { "epoch": 0.9291198318399629, "grad_norm": 1.5936518293179873, "learning_rate": 2.622383492435554e-07, "loss": 0.5284, "step": 26079 }, { "epoch": 0.929155458966457, "grad_norm": 1.2459473876226745, "learning_rate": 2.619758840180431e-07, "loss": 0.423, "step": 26080 }, { "epoch": 0.9291910860929512, "grad_norm": 1.314697210263992, "learning_rate": 2.6171354846112885e-07, "loss": 0.4218, "step": 26081 }, { "epoch": 0.9292267132194453, "grad_norm": 1.2680754978215916, "learning_rate": 2.6145134257630433e-07, "loss": 0.4579, "step": 26082 }, { "epoch": 0.9292623403459394, "grad_norm": 1.2881708699117156, "learning_rate": 2.611892663670612e-07, "loss": 0.399, "step": 26083 }, { "epoch": 0.9292979674724335, "grad_norm": 1.3072217450250838, "learning_rate": 2.609273198368922e-07, "loss": 0.3728, "step": 26084 }, { "epoch": 0.9293335945989276, "grad_norm": 1.7115104709245792, "learning_rate": 2.606655029892813e-07, "loss": 0.4939, "step": 26085 }, { "epoch": 0.9293692217254217, "grad_norm": 1.5388720125042161, "learning_rate": 2.604038158277189e-07, "loss": 0.5308, "step": 26086 }, { "epoch": 0.9294048488519159, "grad_norm": 1.30722057014239, "learning_rate": 2.601422583556845e-07, "loss": 0.3855, "step": 26087 }, { "epoch": 0.92944047597841, "grad_norm": 1.353774406000664, "learning_rate": 2.5988083057666534e-07, "loss": 0.412, "step": 26088 }, { "epoch": 0.929476103104904, "grad_norm": 1.8214093032827, "learning_rate": 2.5961953249413974e-07, "loss": 0.5078, "step": 26089 }, { "epoch": 0.9295117302313982, "grad_norm": 1.7738580285359282, "learning_rate": 2.5935836411158935e-07, "loss": 0.493, "step": 26090 }, { "epoch": 0.9295473573578923, "grad_norm": 1.3526070901704579, "learning_rate": 2.5909732543249023e-07, "loss": 0.406, "step": 26091 }, { "epoch": 0.9295829844843864, "grad_norm": 1.528732499569857, "learning_rate": 2.5883641646031966e-07, "loss": 0.3899, "step": 26092 }, { "epoch": 0.9296186116108806, "grad_norm": 2.2852526232313077, "learning_rate": 2.5857563719854927e-07, "loss": 0.4594, "step": 26093 }, { "epoch": 0.9296542387373746, "grad_norm": 1.4534738458454899, "learning_rate": 2.58314987650653e-07, "loss": 0.4909, "step": 26094 }, { "epoch": 0.9296898658638687, "grad_norm": 1.4091714734945828, "learning_rate": 2.5805446782010245e-07, "loss": 0.4099, "step": 26095 }, { "epoch": 0.9297254929903629, "grad_norm": 1.409940451133081, "learning_rate": 2.577940777103649e-07, "loss": 0.4238, "step": 26096 }, { "epoch": 0.929761120116857, "grad_norm": 1.5558436539593685, "learning_rate": 2.575338173249109e-07, "loss": 0.4468, "step": 26097 }, { "epoch": 0.929796747243351, "grad_norm": 1.4134471692466464, "learning_rate": 2.572736866672021e-07, "loss": 0.4109, "step": 26098 }, { "epoch": 0.9298323743698452, "grad_norm": 1.3659487301331632, "learning_rate": 2.570136857407046e-07, "loss": 0.3953, "step": 26099 }, { "epoch": 0.9298680014963393, "grad_norm": 1.830430949964877, "learning_rate": 2.5675381454887903e-07, "loss": 0.4721, "step": 26100 }, { "epoch": 0.9299036286228334, "grad_norm": 1.8454064124413188, "learning_rate": 2.56494073095187e-07, "loss": 0.4884, "step": 26101 }, { "epoch": 0.9299392557493276, "grad_norm": 1.9118009141596874, "learning_rate": 2.5623446138308584e-07, "loss": 0.4965, "step": 26102 }, { "epoch": 0.9299748828758216, "grad_norm": 1.4222588343846922, "learning_rate": 2.559749794160349e-07, "loss": 0.4107, "step": 26103 }, { "epoch": 0.9300105100023157, "grad_norm": 1.5339963757350592, "learning_rate": 2.5571562719748587e-07, "loss": 0.4924, "step": 26104 }, { "epoch": 0.9300461371288099, "grad_norm": 1.7125778585537363, "learning_rate": 2.5545640473089607e-07, "loss": 0.5113, "step": 26105 }, { "epoch": 0.930081764255304, "grad_norm": 1.660866206570753, "learning_rate": 2.5519731201971374e-07, "loss": 0.4024, "step": 26106 }, { "epoch": 0.930117391381798, "grad_norm": 1.2533809345813967, "learning_rate": 2.5493834906739065e-07, "loss": 0.4264, "step": 26107 }, { "epoch": 0.9301530185082922, "grad_norm": 1.411240079821105, "learning_rate": 2.5467951587737625e-07, "loss": 0.4975, "step": 26108 }, { "epoch": 0.9301886456347863, "grad_norm": 1.293892467929606, "learning_rate": 2.5442081245311446e-07, "loss": 0.3907, "step": 26109 }, { "epoch": 0.9302242727612804, "grad_norm": 1.3166325584684644, "learning_rate": 2.541622387980514e-07, "loss": 0.4055, "step": 26110 }, { "epoch": 0.9302598998877746, "grad_norm": 1.8876076549474938, "learning_rate": 2.5390379491562977e-07, "loss": 0.4986, "step": 26111 }, { "epoch": 0.9302955270142687, "grad_norm": 1.1315532443729894, "learning_rate": 2.536454808092914e-07, "loss": 0.3943, "step": 26112 }, { "epoch": 0.9303311541407627, "grad_norm": 1.4028337521371619, "learning_rate": 2.5338729648247574e-07, "loss": 0.3852, "step": 26113 }, { "epoch": 0.9303667812672569, "grad_norm": 1.6447914924004756, "learning_rate": 2.531292419386222e-07, "loss": 0.5995, "step": 26114 }, { "epoch": 0.930402408393751, "grad_norm": 1.154699265059159, "learning_rate": 2.5287131718116365e-07, "loss": 0.4349, "step": 26115 }, { "epoch": 0.9304380355202451, "grad_norm": 1.3885196013909402, "learning_rate": 2.526135222135362e-07, "loss": 0.3969, "step": 26116 }, { "epoch": 0.9304736626467393, "grad_norm": 1.4635507938247656, "learning_rate": 2.5235585703917265e-07, "loss": 0.5465, "step": 26117 }, { "epoch": 0.9305092897732333, "grad_norm": 4.25795461462246, "learning_rate": 2.520983216615047e-07, "loss": 0.54, "step": 26118 }, { "epoch": 0.9305449168997274, "grad_norm": 1.2387477994460698, "learning_rate": 2.518409160839619e-07, "loss": 0.3363, "step": 26119 }, { "epoch": 0.9305805440262216, "grad_norm": 1.4151025626844165, "learning_rate": 2.515836403099703e-07, "loss": 0.386, "step": 26120 }, { "epoch": 0.9306161711527157, "grad_norm": 1.2708419058122238, "learning_rate": 2.5132649434295607e-07, "loss": 0.4304, "step": 26121 }, { "epoch": 0.9306517982792097, "grad_norm": 1.6388744978438476, "learning_rate": 2.510694781863432e-07, "loss": 0.5468, "step": 26122 }, { "epoch": 0.9306874254057039, "grad_norm": 1.4626889075048437, "learning_rate": 2.5081259184355554e-07, "loss": 0.4416, "step": 26123 }, { "epoch": 0.930723052532198, "grad_norm": 1.2087796651855558, "learning_rate": 2.5055583531801044e-07, "loss": 0.389, "step": 26124 }, { "epoch": 0.9307586796586921, "grad_norm": 1.400085029000186, "learning_rate": 2.502992086131295e-07, "loss": 0.3759, "step": 26125 }, { "epoch": 0.9307943067851863, "grad_norm": 1.3056985870437525, "learning_rate": 2.500427117323301e-07, "loss": 0.4499, "step": 26126 }, { "epoch": 0.9308299339116803, "grad_norm": 1.3323432922791238, "learning_rate": 2.4978634467902607e-07, "loss": 0.4085, "step": 26127 }, { "epoch": 0.9308655610381744, "grad_norm": 1.635155187521176, "learning_rate": 2.4953010745663255e-07, "loss": 0.4916, "step": 26128 }, { "epoch": 0.9309011881646686, "grad_norm": 1.9810794098185938, "learning_rate": 2.492740000685612e-07, "loss": 0.4829, "step": 26129 }, { "epoch": 0.9309368152911627, "grad_norm": 1.160844133383867, "learning_rate": 2.490180225182204e-07, "loss": 0.369, "step": 26130 }, { "epoch": 0.9309724424176568, "grad_norm": 1.4375249081344068, "learning_rate": 2.4876217480902077e-07, "loss": 0.4399, "step": 26131 }, { "epoch": 0.9310080695441509, "grad_norm": 2.0814977005390918, "learning_rate": 2.4850645694436736e-07, "loss": 0.5122, "step": 26132 }, { "epoch": 0.931043696670645, "grad_norm": 1.7682387664869181, "learning_rate": 2.4825086892766745e-07, "loss": 0.5325, "step": 26133 }, { "epoch": 0.9310793237971391, "grad_norm": 1.8143032510568189, "learning_rate": 2.479954107623239e-07, "loss": 0.5481, "step": 26134 }, { "epoch": 0.9311149509236333, "grad_norm": 1.3319515113190934, "learning_rate": 2.477400824517373e-07, "loss": 0.5216, "step": 26135 }, { "epoch": 0.9311505780501274, "grad_norm": 2.172902872804295, "learning_rate": 2.4748488399930826e-07, "loss": 0.3771, "step": 26136 }, { "epoch": 0.9311862051766214, "grad_norm": 1.2372524566845144, "learning_rate": 2.4722981540843403e-07, "loss": 0.3938, "step": 26137 }, { "epoch": 0.9312218323031156, "grad_norm": 1.220874191734237, "learning_rate": 2.46974876682512e-07, "loss": 0.3523, "step": 26138 }, { "epoch": 0.9312574594296097, "grad_norm": 1.2832981735437357, "learning_rate": 2.4672006782493487e-07, "loss": 0.4282, "step": 26139 }, { "epoch": 0.9312930865561038, "grad_norm": 1.5698088060830622, "learning_rate": 2.4646538883909885e-07, "loss": 0.4667, "step": 26140 }, { "epoch": 0.931328713682598, "grad_norm": 1.6798325363983615, "learning_rate": 2.4621083972839244e-07, "loss": 0.4223, "step": 26141 }, { "epoch": 0.931364340809092, "grad_norm": 1.5919847793152866, "learning_rate": 2.4595642049620725e-07, "loss": 0.4953, "step": 26142 }, { "epoch": 0.9313999679355861, "grad_norm": 1.5513341455722804, "learning_rate": 2.4570213114592957e-07, "loss": 0.4291, "step": 26143 }, { "epoch": 0.9314355950620803, "grad_norm": 2.7830774634114186, "learning_rate": 2.4544797168094547e-07, "loss": 0.538, "step": 26144 }, { "epoch": 0.9314712221885744, "grad_norm": 1.420127358236831, "learning_rate": 2.451939421046412e-07, "loss": 0.4281, "step": 26145 }, { "epoch": 0.9315068493150684, "grad_norm": 1.227226433283383, "learning_rate": 2.449400424203963e-07, "loss": 0.3817, "step": 26146 }, { "epoch": 0.9315424764415626, "grad_norm": 1.6197746616904363, "learning_rate": 2.446862726315935e-07, "loss": 0.4814, "step": 26147 }, { "epoch": 0.9315781035680567, "grad_norm": 1.5368010865220867, "learning_rate": 2.444326327416102e-07, "loss": 0.5084, "step": 26148 }, { "epoch": 0.9316137306945508, "grad_norm": 1.3023519018104566, "learning_rate": 2.4417912275382703e-07, "loss": 0.4041, "step": 26149 }, { "epoch": 0.931649357821045, "grad_norm": 1.2411697727728956, "learning_rate": 2.439257426716168e-07, "loss": 0.4277, "step": 26150 }, { "epoch": 0.931684984947539, "grad_norm": 1.244171414853823, "learning_rate": 2.436724924983558e-07, "loss": 0.3673, "step": 26151 }, { "epoch": 0.9317206120740331, "grad_norm": 2.397542458262732, "learning_rate": 2.434193722374134e-07, "loss": 0.5199, "step": 26152 }, { "epoch": 0.9317562392005273, "grad_norm": 1.2493504433776672, "learning_rate": 2.4316638189216146e-07, "loss": 0.3902, "step": 26153 }, { "epoch": 0.9317918663270214, "grad_norm": 1.4305497185076808, "learning_rate": 2.4291352146596946e-07, "loss": 0.3899, "step": 26154 }, { "epoch": 0.9318274934535155, "grad_norm": 1.3970996511924556, "learning_rate": 2.426607909622025e-07, "loss": 0.5286, "step": 26155 }, { "epoch": 0.9318631205800096, "grad_norm": 1.2981401345842172, "learning_rate": 2.4240819038422903e-07, "loss": 0.3993, "step": 26156 }, { "epoch": 0.9318987477065037, "grad_norm": 1.470191541572463, "learning_rate": 2.4215571973540854e-07, "loss": 0.4782, "step": 26157 }, { "epoch": 0.9319343748329978, "grad_norm": 1.4152504160710173, "learning_rate": 2.419033790191072e-07, "loss": 0.3843, "step": 26158 }, { "epoch": 0.931970001959492, "grad_norm": 1.2539978058309083, "learning_rate": 2.416511682386813e-07, "loss": 0.3929, "step": 26159 }, { "epoch": 0.932005629085986, "grad_norm": 1.8697571186477708, "learning_rate": 2.413990873974914e-07, "loss": 0.474, "step": 26160 }, { "epoch": 0.9320412562124801, "grad_norm": 1.3116294119756624, "learning_rate": 2.411471364988938e-07, "loss": 0.4002, "step": 26161 }, { "epoch": 0.9320768833389743, "grad_norm": 1.4227722460455732, "learning_rate": 2.4089531554624233e-07, "loss": 0.423, "step": 26162 }, { "epoch": 0.9321125104654684, "grad_norm": 1.6373440995873867, "learning_rate": 2.4064362454289e-07, "loss": 0.4553, "step": 26163 }, { "epoch": 0.9321481375919625, "grad_norm": 1.509423118493117, "learning_rate": 2.403920634921908e-07, "loss": 0.4347, "step": 26164 }, { "epoch": 0.9321837647184567, "grad_norm": 1.2886520648138855, "learning_rate": 2.401406323974931e-07, "loss": 0.4583, "step": 26165 }, { "epoch": 0.9322193918449507, "grad_norm": 1.5108021496360637, "learning_rate": 2.398893312621431e-07, "loss": 0.5071, "step": 26166 }, { "epoch": 0.9322550189714448, "grad_norm": 1.4528618214826183, "learning_rate": 2.3963816008949034e-07, "loss": 0.5509, "step": 26167 }, { "epoch": 0.932290646097939, "grad_norm": 1.7549378757533742, "learning_rate": 2.393871188828767e-07, "loss": 0.5596, "step": 26168 }, { "epoch": 0.9323262732244331, "grad_norm": 2.1478014143666058, "learning_rate": 2.3913620764564495e-07, "loss": 0.511, "step": 26169 }, { "epoch": 0.9323619003509271, "grad_norm": 1.3062991864174955, "learning_rate": 2.38885426381138e-07, "loss": 0.4097, "step": 26170 }, { "epoch": 0.9323975274774213, "grad_norm": 1.830095672569887, "learning_rate": 2.386347750926943e-07, "loss": 0.5336, "step": 26171 }, { "epoch": 0.9324331546039154, "grad_norm": 1.1657325180073588, "learning_rate": 2.3838425378365116e-07, "loss": 0.3905, "step": 26172 }, { "epoch": 0.9324687817304095, "grad_norm": 1.2568950861408057, "learning_rate": 2.3813386245734483e-07, "loss": 0.3634, "step": 26173 }, { "epoch": 0.9325044088569037, "grad_norm": 1.547641302338445, "learning_rate": 2.3788360111710818e-07, "loss": 0.5263, "step": 26174 }, { "epoch": 0.9325400359833977, "grad_norm": 1.8030889901849638, "learning_rate": 2.3763346976627632e-07, "loss": 0.5233, "step": 26175 }, { "epoch": 0.9325756631098918, "grad_norm": 1.2658642359536758, "learning_rate": 2.3738346840817772e-07, "loss": 0.4116, "step": 26176 }, { "epoch": 0.932611290236386, "grad_norm": 1.4607935222762378, "learning_rate": 2.371335970461408e-07, "loss": 0.4275, "step": 26177 }, { "epoch": 0.9326469173628801, "grad_norm": 1.5386121740014662, "learning_rate": 2.3688385568349515e-07, "loss": 0.5111, "step": 26178 }, { "epoch": 0.9326825444893742, "grad_norm": 1.3817794833034225, "learning_rate": 2.3663424432356363e-07, "loss": 0.4123, "step": 26179 }, { "epoch": 0.9327181716158683, "grad_norm": 1.4048792342191463, "learning_rate": 2.363847629696736e-07, "loss": 0.4033, "step": 26180 }, { "epoch": 0.9327537987423624, "grad_norm": 1.2642023277770649, "learning_rate": 2.3613541162514242e-07, "loss": 0.388, "step": 26181 }, { "epoch": 0.9327894258688565, "grad_norm": 1.5449273543474413, "learning_rate": 2.358861902932952e-07, "loss": 0.5194, "step": 26182 }, { "epoch": 0.9328250529953507, "grad_norm": 1.4989916169223598, "learning_rate": 2.3563709897744592e-07, "loss": 0.4185, "step": 26183 }, { "epoch": 0.9328606801218448, "grad_norm": 1.2824594033911392, "learning_rate": 2.3538813768091418e-07, "loss": 0.4007, "step": 26184 }, { "epoch": 0.9328963072483389, "grad_norm": 1.3570837713680461, "learning_rate": 2.35139306407014e-07, "loss": 0.4132, "step": 26185 }, { "epoch": 0.932931934374833, "grad_norm": 1.3181287046475811, "learning_rate": 2.3489060515906048e-07, "loss": 0.4288, "step": 26186 }, { "epoch": 0.9329675615013271, "grad_norm": 1.585355475765595, "learning_rate": 2.3464203394036322e-07, "loss": 0.4713, "step": 26187 }, { "epoch": 0.9330031886278213, "grad_norm": 1.4654383913478646, "learning_rate": 2.343935927542329e-07, "loss": 0.4149, "step": 26188 }, { "epoch": 0.9330388157543154, "grad_norm": 1.7516127007564284, "learning_rate": 2.3414528160397798e-07, "loss": 0.5845, "step": 26189 }, { "epoch": 0.9330744428808094, "grad_norm": 1.3161941117181561, "learning_rate": 2.338971004929036e-07, "loss": 0.4604, "step": 26190 }, { "epoch": 0.9331100700073036, "grad_norm": 2.674172606753775, "learning_rate": 2.33649049424316e-07, "loss": 0.4141, "step": 26191 }, { "epoch": 0.9331456971337977, "grad_norm": 1.349533370009606, "learning_rate": 2.3340112840151696e-07, "loss": 0.4722, "step": 26192 }, { "epoch": 0.9331813242602918, "grad_norm": 1.3501837412655524, "learning_rate": 2.3315333742780942e-07, "loss": 0.4345, "step": 26193 }, { "epoch": 0.933216951386786, "grad_norm": 1.4050521916195304, "learning_rate": 2.3290567650649187e-07, "loss": 0.3985, "step": 26194 }, { "epoch": 0.93325257851328, "grad_norm": 1.743415031722454, "learning_rate": 2.3265814564086165e-07, "loss": 0.5309, "step": 26195 }, { "epoch": 0.9332882056397741, "grad_norm": 1.3178409422716368, "learning_rate": 2.3241074483421498e-07, "loss": 0.3972, "step": 26196 }, { "epoch": 0.9333238327662683, "grad_norm": 1.2957397206897372, "learning_rate": 2.3216347408984707e-07, "loss": 0.435, "step": 26197 }, { "epoch": 0.9333594598927624, "grad_norm": 1.4245364189621805, "learning_rate": 2.3191633341104859e-07, "loss": 0.4173, "step": 26198 }, { "epoch": 0.9333950870192564, "grad_norm": 1.2724439144187707, "learning_rate": 2.3166932280111242e-07, "loss": 0.3811, "step": 26199 }, { "epoch": 0.9334307141457506, "grad_norm": 1.3178992598821406, "learning_rate": 2.31422442263326e-07, "loss": 0.4121, "step": 26200 }, { "epoch": 0.9334663412722447, "grad_norm": 1.6848939974023152, "learning_rate": 2.3117569180097778e-07, "loss": 0.5135, "step": 26201 }, { "epoch": 0.9335019683987388, "grad_norm": 1.4120285462229492, "learning_rate": 2.309290714173551e-07, "loss": 0.4802, "step": 26202 }, { "epoch": 0.933537595525233, "grad_norm": 1.2231322727970713, "learning_rate": 2.306825811157376e-07, "loss": 0.4237, "step": 26203 }, { "epoch": 0.933573222651727, "grad_norm": 1.2316405581765377, "learning_rate": 2.3043622089941152e-07, "loss": 0.4171, "step": 26204 }, { "epoch": 0.9336088497782211, "grad_norm": 1.2878723927688347, "learning_rate": 2.3018999077165537e-07, "loss": 0.3593, "step": 26205 }, { "epoch": 0.9336444769047153, "grad_norm": 1.2165771743499911, "learning_rate": 2.2994389073574762e-07, "loss": 0.4082, "step": 26206 }, { "epoch": 0.9336801040312094, "grad_norm": 1.6165991268237998, "learning_rate": 2.2969792079496567e-07, "loss": 0.4985, "step": 26207 }, { "epoch": 0.9337157311577035, "grad_norm": 1.5386752470376734, "learning_rate": 2.2945208095258575e-07, "loss": 0.3752, "step": 26208 }, { "epoch": 0.9337513582841976, "grad_norm": 1.36862259352438, "learning_rate": 2.2920637121187972e-07, "loss": 0.4459, "step": 26209 }, { "epoch": 0.9337869854106917, "grad_norm": 1.2231559940210428, "learning_rate": 2.289607915761205e-07, "loss": 0.3702, "step": 26210 }, { "epoch": 0.9338226125371858, "grad_norm": 1.8294563258364873, "learning_rate": 2.287153420485788e-07, "loss": 0.4966, "step": 26211 }, { "epoch": 0.93385823966368, "grad_norm": 1.1948258763786235, "learning_rate": 2.2847002263252093e-07, "loss": 0.4398, "step": 26212 }, { "epoch": 0.933893866790174, "grad_norm": 1.5912099006083067, "learning_rate": 2.2822483333121537e-07, "loss": 0.4577, "step": 26213 }, { "epoch": 0.9339294939166681, "grad_norm": 2.346748605881352, "learning_rate": 2.2797977414792505e-07, "loss": 0.3684, "step": 26214 }, { "epoch": 0.9339651210431623, "grad_norm": 1.4714474253921668, "learning_rate": 2.2773484508591515e-07, "loss": 0.4289, "step": 26215 }, { "epoch": 0.9340007481696564, "grad_norm": 1.7112383730604614, "learning_rate": 2.2749004614844638e-07, "loss": 0.3944, "step": 26216 }, { "epoch": 0.9340363752961505, "grad_norm": 1.284070724416753, "learning_rate": 2.2724537733877838e-07, "loss": 0.427, "step": 26217 }, { "epoch": 0.9340720024226447, "grad_norm": 1.860215367791693, "learning_rate": 2.2700083866016854e-07, "loss": 0.4638, "step": 26218 }, { "epoch": 0.9341076295491387, "grad_norm": 1.4951783492051112, "learning_rate": 2.2675643011587312e-07, "loss": 0.4551, "step": 26219 }, { "epoch": 0.9341432566756328, "grad_norm": 1.1726895654019995, "learning_rate": 2.2651215170914731e-07, "loss": 0.3736, "step": 26220 }, { "epoch": 0.934178883802127, "grad_norm": 1.2573575526418566, "learning_rate": 2.2626800344324295e-07, "loss": 0.3923, "step": 26221 }, { "epoch": 0.9342145109286211, "grad_norm": 1.4461565263989145, "learning_rate": 2.260239853214119e-07, "loss": 0.574, "step": 26222 }, { "epoch": 0.9342501380551151, "grad_norm": 1.7738609516442054, "learning_rate": 2.2578009734690264e-07, "loss": 0.546, "step": 26223 }, { "epoch": 0.9342857651816093, "grad_norm": 1.3910309500738685, "learning_rate": 2.2553633952296484e-07, "loss": 0.4363, "step": 26224 }, { "epoch": 0.9343213923081034, "grad_norm": 1.3203989948451373, "learning_rate": 2.252927118528403e-07, "loss": 0.5136, "step": 26225 }, { "epoch": 0.9343570194345975, "grad_norm": 1.630901500853601, "learning_rate": 2.2504921433977755e-07, "loss": 0.482, "step": 26226 }, { "epoch": 0.9343926465610917, "grad_norm": 1.2427628519642835, "learning_rate": 2.2480584698701513e-07, "loss": 0.435, "step": 26227 }, { "epoch": 0.9344282736875857, "grad_norm": 1.813999153830311, "learning_rate": 2.24562609797796e-07, "loss": 0.5573, "step": 26228 }, { "epoch": 0.9344639008140798, "grad_norm": 1.3811951778023204, "learning_rate": 2.2431950277535753e-07, "loss": 0.4252, "step": 26229 }, { "epoch": 0.934499527940574, "grad_norm": 1.2157864910442568, "learning_rate": 2.2407652592293938e-07, "loss": 0.4209, "step": 26230 }, { "epoch": 0.9345351550670681, "grad_norm": 1.24233265622737, "learning_rate": 2.2383367924377454e-07, "loss": 0.3722, "step": 26231 }, { "epoch": 0.9345707821935622, "grad_norm": 1.1750677269202578, "learning_rate": 2.2359096274109705e-07, "loss": 0.3473, "step": 26232 }, { "epoch": 0.9346064093200563, "grad_norm": 1.3967232951973354, "learning_rate": 2.233483764181399e-07, "loss": 0.4024, "step": 26233 }, { "epoch": 0.9346420364465504, "grad_norm": 1.3632757651365606, "learning_rate": 2.231059202781316e-07, "loss": 0.4582, "step": 26234 }, { "epoch": 0.9346776635730445, "grad_norm": 1.2795539893410033, "learning_rate": 2.2286359432430182e-07, "loss": 0.4029, "step": 26235 }, { "epoch": 0.9347132906995387, "grad_norm": 1.5390928085551923, "learning_rate": 2.2262139855987796e-07, "loss": 0.4265, "step": 26236 }, { "epoch": 0.9347489178260328, "grad_norm": 1.4754383483160616, "learning_rate": 2.2237933298808301e-07, "loss": 0.5013, "step": 26237 }, { "epoch": 0.9347845449525268, "grad_norm": 1.2560237232401235, "learning_rate": 2.2213739761214103e-07, "loss": 0.3667, "step": 26238 }, { "epoch": 0.934820172079021, "grad_norm": 1.2233963770480258, "learning_rate": 2.2189559243527614e-07, "loss": 0.3907, "step": 26239 }, { "epoch": 0.9348557992055151, "grad_norm": 1.396917656903448, "learning_rate": 2.2165391746070352e-07, "loss": 0.4432, "step": 26240 }, { "epoch": 0.9348914263320092, "grad_norm": 1.6003080109127372, "learning_rate": 2.2141237269164507e-07, "loss": 0.4922, "step": 26241 }, { "epoch": 0.9349270534585034, "grad_norm": 1.2756089979375713, "learning_rate": 2.2117095813131485e-07, "loss": 0.3992, "step": 26242 }, { "epoch": 0.9349626805849974, "grad_norm": 1.3211249551919118, "learning_rate": 2.209296737829292e-07, "loss": 0.4269, "step": 26243 }, { "epoch": 0.9349983077114915, "grad_norm": 1.4244357528680585, "learning_rate": 2.2068851964969884e-07, "loss": 0.5067, "step": 26244 }, { "epoch": 0.9350339348379857, "grad_norm": 1.3539239123502615, "learning_rate": 2.2044749573483683e-07, "loss": 0.5227, "step": 26245 }, { "epoch": 0.9350695619644798, "grad_norm": 1.3834088076242672, "learning_rate": 2.2020660204155276e-07, "loss": 0.4374, "step": 26246 }, { "epoch": 0.9351051890909738, "grad_norm": 1.8411273795275234, "learning_rate": 2.19965838573053e-07, "loss": 0.5311, "step": 26247 }, { "epoch": 0.935140816217468, "grad_norm": 1.2582502297099851, "learning_rate": 2.1972520533254383e-07, "loss": 0.3723, "step": 26248 }, { "epoch": 0.9351764433439621, "grad_norm": 1.221362378850048, "learning_rate": 2.1948470232322937e-07, "loss": 0.3645, "step": 26249 }, { "epoch": 0.9352120704704562, "grad_norm": 1.3753450312610676, "learning_rate": 2.1924432954831265e-07, "loss": 0.4896, "step": 26250 }, { "epoch": 0.9352476975969504, "grad_norm": 1.2466476566638631, "learning_rate": 2.190040870109933e-07, "loss": 0.4071, "step": 26251 }, { "epoch": 0.9352833247234444, "grad_norm": 1.4049509300579106, "learning_rate": 2.1876397471447097e-07, "loss": 0.5158, "step": 26252 }, { "epoch": 0.9353189518499385, "grad_norm": 1.609677424020933, "learning_rate": 2.1852399266194312e-07, "loss": 0.4928, "step": 26253 }, { "epoch": 0.9353545789764327, "grad_norm": 1.3570025408619202, "learning_rate": 2.182841408566061e-07, "loss": 0.398, "step": 26254 }, { "epoch": 0.9353902061029268, "grad_norm": 1.3160530785906794, "learning_rate": 2.1804441930165175e-07, "loss": 0.5285, "step": 26255 }, { "epoch": 0.9354258332294209, "grad_norm": 1.434034317314143, "learning_rate": 2.1780482800027425e-07, "loss": 0.4771, "step": 26256 }, { "epoch": 0.935461460355915, "grad_norm": 1.2710476518022207, "learning_rate": 2.175653669556621e-07, "loss": 0.3997, "step": 26257 }, { "epoch": 0.9354970874824091, "grad_norm": 1.3513033634081582, "learning_rate": 2.173260361710039e-07, "loss": 0.3894, "step": 26258 }, { "epoch": 0.9355327146089032, "grad_norm": 1.256128918087453, "learning_rate": 2.170868356494882e-07, "loss": 0.3777, "step": 26259 }, { "epoch": 0.9355683417353974, "grad_norm": 1.4843747801962157, "learning_rate": 2.1684776539429797e-07, "loss": 0.3372, "step": 26260 }, { "epoch": 0.9356039688618915, "grad_norm": 1.2201369840554435, "learning_rate": 2.166088254086196e-07, "loss": 0.4415, "step": 26261 }, { "epoch": 0.9356395959883855, "grad_norm": 1.509526207769272, "learning_rate": 2.1637001569563166e-07, "loss": 0.4171, "step": 26262 }, { "epoch": 0.9356752231148797, "grad_norm": 1.5618986110367266, "learning_rate": 2.16131336258516e-07, "loss": 0.4765, "step": 26263 }, { "epoch": 0.9357108502413738, "grad_norm": 1.70205283103904, "learning_rate": 2.1589278710045015e-07, "loss": 0.4454, "step": 26264 }, { "epoch": 0.9357464773678679, "grad_norm": 1.2219402261025571, "learning_rate": 2.156543682246104e-07, "loss": 0.3582, "step": 26265 }, { "epoch": 0.935782104494362, "grad_norm": 1.4425712451475607, "learning_rate": 2.1541607963416978e-07, "loss": 0.4091, "step": 26266 }, { "epoch": 0.9358177316208561, "grad_norm": 1.3668362440020374, "learning_rate": 2.1517792133230574e-07, "loss": 0.4103, "step": 26267 }, { "epoch": 0.9358533587473502, "grad_norm": 1.4455013208015322, "learning_rate": 2.1493989332218468e-07, "loss": 0.4923, "step": 26268 }, { "epoch": 0.9358889858738444, "grad_norm": 1.2044457596259517, "learning_rate": 2.1470199560697957e-07, "loss": 0.3831, "step": 26269 }, { "epoch": 0.9359246130003385, "grad_norm": 1.2336004937445162, "learning_rate": 2.1446422818985678e-07, "loss": 0.3825, "step": 26270 }, { "epoch": 0.9359602401268325, "grad_norm": 1.8909730590137315, "learning_rate": 2.1422659107398158e-07, "loss": 0.4619, "step": 26271 }, { "epoch": 0.9359958672533267, "grad_norm": 1.240761068439554, "learning_rate": 2.139890842625214e-07, "loss": 0.4025, "step": 26272 }, { "epoch": 0.9360314943798208, "grad_norm": 2.090816058309701, "learning_rate": 2.1375170775863375e-07, "loss": 0.4677, "step": 26273 }, { "epoch": 0.9360671215063149, "grad_norm": 1.9077958952471987, "learning_rate": 2.1351446156548382e-07, "loss": 0.4823, "step": 26274 }, { "epoch": 0.9361027486328091, "grad_norm": 1.3792835482377868, "learning_rate": 2.1327734568622916e-07, "loss": 0.4015, "step": 26275 }, { "epoch": 0.9361383757593031, "grad_norm": 1.649630923905607, "learning_rate": 2.1304036012402719e-07, "loss": 0.5721, "step": 26276 }, { "epoch": 0.9361740028857972, "grad_norm": 1.193932308344663, "learning_rate": 2.1280350488203317e-07, "loss": 0.3459, "step": 26277 }, { "epoch": 0.9362096300122914, "grad_norm": 1.3174355986553048, "learning_rate": 2.1256677996340124e-07, "loss": 0.4209, "step": 26278 }, { "epoch": 0.9362452571387855, "grad_norm": 1.2475093375353692, "learning_rate": 2.1233018537128446e-07, "loss": 0.3848, "step": 26279 }, { "epoch": 0.9362808842652796, "grad_norm": 1.557399988687828, "learning_rate": 2.120937211088314e-07, "loss": 0.5702, "step": 26280 }, { "epoch": 0.9363165113917737, "grad_norm": 1.7453669162352587, "learning_rate": 2.118573871791918e-07, "loss": 0.51, "step": 26281 }, { "epoch": 0.9363521385182678, "grad_norm": 1.8130751908258873, "learning_rate": 2.11621183585512e-07, "loss": 0.4485, "step": 26282 }, { "epoch": 0.9363877656447619, "grad_norm": 1.26328959993492, "learning_rate": 2.1138511033093944e-07, "loss": 0.3925, "step": 26283 }, { "epoch": 0.9364233927712561, "grad_norm": 1.407018966467864, "learning_rate": 2.1114916741861502e-07, "loss": 0.3527, "step": 26284 }, { "epoch": 0.9364590198977502, "grad_norm": 1.2359163363954528, "learning_rate": 2.1091335485168286e-07, "loss": 0.3774, "step": 26285 }, { "epoch": 0.9364946470242442, "grad_norm": 1.4706598053069582, "learning_rate": 2.1067767263327932e-07, "loss": 0.6029, "step": 26286 }, { "epoch": 0.9365302741507384, "grad_norm": 1.4243570771383045, "learning_rate": 2.1044212076654746e-07, "loss": 0.4642, "step": 26287 }, { "epoch": 0.9365659012772325, "grad_norm": 1.2486756308689797, "learning_rate": 2.1020669925461923e-07, "loss": 0.3803, "step": 26288 }, { "epoch": 0.9366015284037266, "grad_norm": 1.3494654769788743, "learning_rate": 2.09971408100631e-07, "loss": 0.4545, "step": 26289 }, { "epoch": 0.9366371555302208, "grad_norm": 1.357342424851572, "learning_rate": 2.0973624730771803e-07, "loss": 0.4197, "step": 26290 }, { "epoch": 0.9366727826567148, "grad_norm": 1.376383197357363, "learning_rate": 2.095012168790078e-07, "loss": 0.491, "step": 26291 }, { "epoch": 0.9367084097832089, "grad_norm": 1.45991104886435, "learning_rate": 2.0926631681763455e-07, "loss": 0.3779, "step": 26292 }, { "epoch": 0.9367440369097031, "grad_norm": 1.5733750946461378, "learning_rate": 2.0903154712672236e-07, "loss": 0.5596, "step": 26293 }, { "epoch": 0.9367796640361972, "grad_norm": 1.346247358997891, "learning_rate": 2.0879690780939876e-07, "loss": 0.3951, "step": 26294 }, { "epoch": 0.9368152911626912, "grad_norm": 1.3741081634247005, "learning_rate": 2.0856239886878793e-07, "loss": 0.4161, "step": 26295 }, { "epoch": 0.9368509182891854, "grad_norm": 1.251287628193053, "learning_rate": 2.083280203080118e-07, "loss": 0.4334, "step": 26296 }, { "epoch": 0.9368865454156795, "grad_norm": 1.2326242434568244, "learning_rate": 2.0809377213019233e-07, "loss": 0.402, "step": 26297 }, { "epoch": 0.9369221725421736, "grad_norm": 2.019343291838811, "learning_rate": 2.0785965433845036e-07, "loss": 0.5368, "step": 26298 }, { "epoch": 0.9369577996686678, "grad_norm": 1.2559145165920556, "learning_rate": 2.0762566693589892e-07, "loss": 0.3963, "step": 26299 }, { "epoch": 0.9369934267951618, "grad_norm": 2.354255956490292, "learning_rate": 2.0739180992565778e-07, "loss": 0.3732, "step": 26300 }, { "epoch": 0.9370290539216559, "grad_norm": 1.4707012491736478, "learning_rate": 2.0715808331083887e-07, "loss": 0.4528, "step": 26301 }, { "epoch": 0.9370646810481501, "grad_norm": 1.309838529236659, "learning_rate": 2.0692448709455527e-07, "loss": 0.464, "step": 26302 }, { "epoch": 0.9371003081746442, "grad_norm": 1.280944172608728, "learning_rate": 2.066910212799156e-07, "loss": 0.3814, "step": 26303 }, { "epoch": 0.9371359353011383, "grad_norm": 1.6346416772473074, "learning_rate": 2.064576858700307e-07, "loss": 0.5263, "step": 26304 }, { "epoch": 0.9371715624276324, "grad_norm": 1.8198605166047948, "learning_rate": 2.0622448086800694e-07, "loss": 0.5014, "step": 26305 }, { "epoch": 0.9372071895541265, "grad_norm": 1.6099516134805332, "learning_rate": 2.059914062769497e-07, "loss": 0.6373, "step": 26306 }, { "epoch": 0.9372428166806206, "grad_norm": 1.420822579692974, "learning_rate": 2.057584620999642e-07, "loss": 0.4702, "step": 26307 }, { "epoch": 0.9372784438071148, "grad_norm": 1.4432534311058804, "learning_rate": 2.0552564834014797e-07, "loss": 0.5076, "step": 26308 }, { "epoch": 0.9373140709336089, "grad_norm": 1.2900502060680512, "learning_rate": 2.0529296500060636e-07, "loss": 0.4, "step": 26309 }, { "epoch": 0.9373496980601029, "grad_norm": 1.5228138662272535, "learning_rate": 2.050604120844324e-07, "loss": 0.4596, "step": 26310 }, { "epoch": 0.9373853251865971, "grad_norm": 1.6698238304085569, "learning_rate": 2.0482798959472584e-07, "loss": 0.3829, "step": 26311 }, { "epoch": 0.9374209523130912, "grad_norm": 1.1489721729094606, "learning_rate": 2.0459569753458198e-07, "loss": 0.3998, "step": 26312 }, { "epoch": 0.9374565794395853, "grad_norm": 1.4277027421732076, "learning_rate": 2.043635359070928e-07, "loss": 0.4276, "step": 26313 }, { "epoch": 0.9374922065660795, "grad_norm": 1.408213673048229, "learning_rate": 2.0413150471534915e-07, "loss": 0.3973, "step": 26314 }, { "epoch": 0.9375278336925735, "grad_norm": 1.5900817035011163, "learning_rate": 2.038996039624419e-07, "loss": 0.4555, "step": 26315 }, { "epoch": 0.9375634608190676, "grad_norm": 1.3850306980771645, "learning_rate": 2.036678336514586e-07, "loss": 0.4423, "step": 26316 }, { "epoch": 0.9375990879455618, "grad_norm": 2.0177268436497204, "learning_rate": 2.0343619378548564e-07, "loss": 0.5319, "step": 26317 }, { "epoch": 0.9376347150720559, "grad_norm": 1.7734816269443372, "learning_rate": 2.032046843676061e-07, "loss": 0.5275, "step": 26318 }, { "epoch": 0.9376703421985499, "grad_norm": 1.8450532037227683, "learning_rate": 2.029733054009042e-07, "loss": 0.5552, "step": 26319 }, { "epoch": 0.9377059693250441, "grad_norm": 1.3963964705877174, "learning_rate": 2.0274205688846082e-07, "loss": 0.4154, "step": 26320 }, { "epoch": 0.9377415964515382, "grad_norm": 1.307080259380463, "learning_rate": 2.025109388333546e-07, "loss": 0.3815, "step": 26321 }, { "epoch": 0.9377772235780323, "grad_norm": 1.4868880230262458, "learning_rate": 2.0227995123866418e-07, "loss": 0.5164, "step": 26322 }, { "epoch": 0.9378128507045265, "grad_norm": 1.3951712709156896, "learning_rate": 2.0204909410746265e-07, "loss": 0.4705, "step": 26323 }, { "epoch": 0.9378484778310205, "grad_norm": 1.908739830332957, "learning_rate": 2.0181836744282758e-07, "loss": 0.5083, "step": 26324 }, { "epoch": 0.9378841049575146, "grad_norm": 1.4386296744035034, "learning_rate": 2.015877712478287e-07, "loss": 0.5122, "step": 26325 }, { "epoch": 0.9379197320840088, "grad_norm": 1.4185624662445382, "learning_rate": 2.013573055255369e-07, "loss": 0.4488, "step": 26326 }, { "epoch": 0.9379553592105029, "grad_norm": 1.5515974131588033, "learning_rate": 2.0112697027902083e-07, "loss": 0.41, "step": 26327 }, { "epoch": 0.937990986336997, "grad_norm": 1.2806985123124643, "learning_rate": 2.0089676551134918e-07, "loss": 0.4516, "step": 26328 }, { "epoch": 0.9380266134634911, "grad_norm": 2.0871280968898467, "learning_rate": 2.0066669122558613e-07, "loss": 0.4336, "step": 26329 }, { "epoch": 0.9380622405899852, "grad_norm": 1.4829525472428802, "learning_rate": 2.004367474247948e-07, "loss": 0.4679, "step": 26330 }, { "epoch": 0.9380978677164793, "grad_norm": 1.3425612400123963, "learning_rate": 2.002069341120394e-07, "loss": 0.3574, "step": 26331 }, { "epoch": 0.9381334948429735, "grad_norm": 1.3959044268195016, "learning_rate": 1.9997725129037637e-07, "loss": 0.371, "step": 26332 }, { "epoch": 0.9381691219694676, "grad_norm": 1.266658358390046, "learning_rate": 1.9974769896286662e-07, "loss": 0.3797, "step": 26333 }, { "epoch": 0.9382047490959616, "grad_norm": 1.4613430834859154, "learning_rate": 1.9951827713256656e-07, "loss": 0.5163, "step": 26334 }, { "epoch": 0.9382403762224558, "grad_norm": 1.6759994593737715, "learning_rate": 1.9928898580253153e-07, "loss": 0.4647, "step": 26335 }, { "epoch": 0.9382760033489499, "grad_norm": 1.294989706710293, "learning_rate": 1.9905982497581246e-07, "loss": 0.4076, "step": 26336 }, { "epoch": 0.938311630475444, "grad_norm": 1.900238344183856, "learning_rate": 1.988307946554624e-07, "loss": 0.4587, "step": 26337 }, { "epoch": 0.9383472576019382, "grad_norm": 1.3645564445086975, "learning_rate": 1.986018948445323e-07, "loss": 0.3522, "step": 26338 }, { "epoch": 0.9383828847284322, "grad_norm": 1.430328776506205, "learning_rate": 1.9837312554606748e-07, "loss": 0.4187, "step": 26339 }, { "epoch": 0.9384185118549263, "grad_norm": 1.365802504955012, "learning_rate": 1.9814448676311548e-07, "loss": 0.4906, "step": 26340 }, { "epoch": 0.9384541389814205, "grad_norm": 1.6608904093700487, "learning_rate": 1.9791597849872057e-07, "loss": 0.4828, "step": 26341 }, { "epoch": 0.9384897661079146, "grad_norm": 1.5144103892585918, "learning_rate": 1.9768760075592698e-07, "loss": 0.4687, "step": 26342 }, { "epoch": 0.9385253932344086, "grad_norm": 1.231937942090366, "learning_rate": 1.9745935353777222e-07, "loss": 0.3905, "step": 26343 }, { "epoch": 0.9385610203609028, "grad_norm": 1.4840087091660616, "learning_rate": 1.972312368472995e-07, "loss": 0.5079, "step": 26344 }, { "epoch": 0.9385966474873969, "grad_norm": 1.3869685476447229, "learning_rate": 1.9700325068754412e-07, "loss": 0.5133, "step": 26345 }, { "epoch": 0.938632274613891, "grad_norm": 1.7877166935036908, "learning_rate": 1.9677539506154363e-07, "loss": 0.4366, "step": 26346 }, { "epoch": 0.9386679017403852, "grad_norm": 1.6422034862277155, "learning_rate": 1.9654766997232898e-07, "loss": 0.4234, "step": 26347 }, { "epoch": 0.9387035288668792, "grad_norm": 1.2976711432996204, "learning_rate": 1.963200754229344e-07, "loss": 0.3982, "step": 26348 }, { "epoch": 0.9387391559933733, "grad_norm": 1.6340023504357872, "learning_rate": 1.960926114163908e-07, "loss": 0.4713, "step": 26349 }, { "epoch": 0.9387747831198675, "grad_norm": 1.2652063909866338, "learning_rate": 1.9586527795572574e-07, "loss": 0.3648, "step": 26350 }, { "epoch": 0.9388104102463616, "grad_norm": 1.3350503759722996, "learning_rate": 1.9563807504396903e-07, "loss": 0.4133, "step": 26351 }, { "epoch": 0.9388460373728557, "grad_norm": 1.2860835813073737, "learning_rate": 1.954110026841427e-07, "loss": 0.3981, "step": 26352 }, { "epoch": 0.9388816644993498, "grad_norm": 1.5538778200821413, "learning_rate": 1.951840608792732e-07, "loss": 0.5342, "step": 26353 }, { "epoch": 0.9389172916258439, "grad_norm": 1.2428246659205586, "learning_rate": 1.9495724963237928e-07, "loss": 0.4111, "step": 26354 }, { "epoch": 0.938952918752338, "grad_norm": 1.240978767575727, "learning_rate": 1.94730568946484e-07, "loss": 0.3577, "step": 26355 }, { "epoch": 0.9389885458788322, "grad_norm": 1.246123537356458, "learning_rate": 1.945040188246039e-07, "loss": 0.4202, "step": 26356 }, { "epoch": 0.9390241730053263, "grad_norm": 1.8895580673432972, "learning_rate": 1.9427759926975764e-07, "loss": 0.5663, "step": 26357 }, { "epoch": 0.9390598001318203, "grad_norm": 1.6138305600008835, "learning_rate": 1.9405131028495838e-07, "loss": 0.363, "step": 26358 }, { "epoch": 0.9390954272583145, "grad_norm": 1.4396286481720388, "learning_rate": 1.9382515187322037e-07, "loss": 0.4734, "step": 26359 }, { "epoch": 0.9391310543848086, "grad_norm": 1.539998935104866, "learning_rate": 1.9359912403755343e-07, "loss": 0.5131, "step": 26360 }, { "epoch": 0.9391666815113027, "grad_norm": 1.4937627227667014, "learning_rate": 1.9337322678096847e-07, "loss": 0.509, "step": 26361 }, { "epoch": 0.9392023086377969, "grad_norm": 1.9634740900834498, "learning_rate": 1.9314746010647533e-07, "loss": 0.531, "step": 26362 }, { "epoch": 0.9392379357642909, "grad_norm": 1.5595264320833107, "learning_rate": 1.9292182401707603e-07, "loss": 0.5862, "step": 26363 }, { "epoch": 0.939273562890785, "grad_norm": 1.623782898575181, "learning_rate": 1.926963185157782e-07, "loss": 0.5498, "step": 26364 }, { "epoch": 0.9393091900172792, "grad_norm": 1.4917491157382163, "learning_rate": 1.9247094360558383e-07, "loss": 0.5372, "step": 26365 }, { "epoch": 0.9393448171437733, "grad_norm": 1.2733109091255712, "learning_rate": 1.9224569928949498e-07, "loss": 0.3778, "step": 26366 }, { "epoch": 0.9393804442702673, "grad_norm": 1.3592719475945008, "learning_rate": 1.920205855705082e-07, "loss": 0.4471, "step": 26367 }, { "epoch": 0.9394160713967615, "grad_norm": 1.8376209185601873, "learning_rate": 1.9179560245162433e-07, "loss": 0.6715, "step": 26368 }, { "epoch": 0.9394516985232556, "grad_norm": 1.5815892717198083, "learning_rate": 1.9157074993583546e-07, "loss": 0.4166, "step": 26369 }, { "epoch": 0.9394873256497497, "grad_norm": 1.2595496394169352, "learning_rate": 1.9134602802613922e-07, "loss": 0.403, "step": 26370 }, { "epoch": 0.9395229527762439, "grad_norm": 1.936217599458948, "learning_rate": 1.911214367255254e-07, "loss": 0.4944, "step": 26371 }, { "epoch": 0.939558579902738, "grad_norm": 1.6101017441291292, "learning_rate": 1.908969760369861e-07, "loss": 0.5832, "step": 26372 }, { "epoch": 0.939594207029232, "grad_norm": 1.6311582121799275, "learning_rate": 1.906726459635111e-07, "loss": 0.5138, "step": 26373 }, { "epoch": 0.9396298341557262, "grad_norm": 1.3384486821286536, "learning_rate": 1.9044844650808468e-07, "loss": 0.4352, "step": 26374 }, { "epoch": 0.9396654612822203, "grad_norm": 1.3779027324198885, "learning_rate": 1.902243776736945e-07, "loss": 0.4363, "step": 26375 }, { "epoch": 0.9397010884087144, "grad_norm": 1.2886451941550683, "learning_rate": 1.900004394633226e-07, "loss": 0.4126, "step": 26376 }, { "epoch": 0.9397367155352085, "grad_norm": 1.3589170278382128, "learning_rate": 1.897766318799521e-07, "loss": 0.4079, "step": 26377 }, { "epoch": 0.9397723426617026, "grad_norm": 1.2473217327721875, "learning_rate": 1.8955295492656068e-07, "loss": 0.4137, "step": 26378 }, { "epoch": 0.9398079697881967, "grad_norm": 1.371917778898925, "learning_rate": 1.8932940860613147e-07, "loss": 0.4272, "step": 26379 }, { "epoch": 0.9398435969146909, "grad_norm": 1.4415247995374938, "learning_rate": 1.8910599292163767e-07, "loss": 0.4002, "step": 26380 }, { "epoch": 0.939879224041185, "grad_norm": 1.4344170320289784, "learning_rate": 1.8888270787605466e-07, "loss": 0.5342, "step": 26381 }, { "epoch": 0.939914851167679, "grad_norm": 2.0188262788892013, "learning_rate": 1.8865955347235564e-07, "loss": 0.4936, "step": 26382 }, { "epoch": 0.9399504782941732, "grad_norm": 1.131762189287941, "learning_rate": 1.884365297135138e-07, "loss": 0.3858, "step": 26383 }, { "epoch": 0.9399861054206673, "grad_norm": 1.501544337170277, "learning_rate": 1.882136366024956e-07, "loss": 0.4764, "step": 26384 }, { "epoch": 0.9400217325471614, "grad_norm": 1.135510831276156, "learning_rate": 1.8799087414227202e-07, "loss": 0.3653, "step": 26385 }, { "epoch": 0.9400573596736556, "grad_norm": 1.28863437654878, "learning_rate": 1.8776824233580737e-07, "loss": 0.4039, "step": 26386 }, { "epoch": 0.9400929868001496, "grad_norm": 1.7513222344072261, "learning_rate": 1.8754574118606706e-07, "loss": 0.5954, "step": 26387 }, { "epoch": 0.9401286139266437, "grad_norm": 1.379209687379488, "learning_rate": 1.8732337069601426e-07, "loss": 0.5216, "step": 26388 }, { "epoch": 0.9401642410531379, "grad_norm": 1.353617864226052, "learning_rate": 1.871011308686088e-07, "loss": 0.3898, "step": 26389 }, { "epoch": 0.940199868179632, "grad_norm": 1.8824203286419108, "learning_rate": 1.868790217068117e-07, "loss": 0.5948, "step": 26390 }, { "epoch": 0.940235495306126, "grad_norm": 1.421198299934843, "learning_rate": 1.8665704321357835e-07, "loss": 0.447, "step": 26391 }, { "epoch": 0.9402711224326202, "grad_norm": 1.4857189348276771, "learning_rate": 1.8643519539186527e-07, "loss": 0.4111, "step": 26392 }, { "epoch": 0.9403067495591143, "grad_norm": 1.4318859571785962, "learning_rate": 1.8621347824462787e-07, "loss": 0.388, "step": 26393 }, { "epoch": 0.9403423766856084, "grad_norm": 1.4434112009622821, "learning_rate": 1.8599189177481713e-07, "loss": 0.4081, "step": 26394 }, { "epoch": 0.9403780038121026, "grad_norm": 1.541294565731793, "learning_rate": 1.8577043598538402e-07, "loss": 0.4258, "step": 26395 }, { "epoch": 0.9404136309385966, "grad_norm": 1.2764763300828543, "learning_rate": 1.8554911087927617e-07, "loss": 0.4084, "step": 26396 }, { "epoch": 0.9404492580650907, "grad_norm": 1.3159706124083255, "learning_rate": 1.8532791645944347e-07, "loss": 0.3784, "step": 26397 }, { "epoch": 0.9404848851915849, "grad_norm": 1.5256569191557694, "learning_rate": 1.851068527288291e-07, "loss": 0.4628, "step": 26398 }, { "epoch": 0.940520512318079, "grad_norm": 1.635878754632975, "learning_rate": 1.8488591969037738e-07, "loss": 0.5154, "step": 26399 }, { "epoch": 0.9405561394445731, "grad_norm": 1.2999811253438345, "learning_rate": 1.8466511734702929e-07, "loss": 0.4616, "step": 26400 }, { "epoch": 0.9405917665710672, "grad_norm": 1.4378683007552915, "learning_rate": 1.8444444570172583e-07, "loss": 0.3972, "step": 26401 }, { "epoch": 0.9406273936975613, "grad_norm": 1.4565228778956132, "learning_rate": 1.842239047574046e-07, "loss": 0.4598, "step": 26402 }, { "epoch": 0.9406630208240554, "grad_norm": 1.3325934412238523, "learning_rate": 1.8400349451700438e-07, "loss": 0.3914, "step": 26403 }, { "epoch": 0.9406986479505496, "grad_norm": 1.260252736431017, "learning_rate": 1.837832149834573e-07, "loss": 0.4047, "step": 26404 }, { "epoch": 0.9407342750770437, "grad_norm": 1.963150654585039, "learning_rate": 1.8356306615969876e-07, "loss": 0.5197, "step": 26405 }, { "epoch": 0.9407699022035377, "grad_norm": 1.3376672814645523, "learning_rate": 1.8334304804865865e-07, "loss": 0.4114, "step": 26406 }, { "epoch": 0.9408055293300319, "grad_norm": 1.5037750983952571, "learning_rate": 1.8312316065326796e-07, "loss": 0.4181, "step": 26407 }, { "epoch": 0.940841156456526, "grad_norm": 1.7640981573903853, "learning_rate": 1.8290340397645324e-07, "loss": 0.4019, "step": 26408 }, { "epoch": 0.9408767835830201, "grad_norm": 1.6163235269974274, "learning_rate": 1.82683778021141e-07, "loss": 0.4295, "step": 26409 }, { "epoch": 0.9409124107095143, "grad_norm": 1.8332713711729856, "learning_rate": 1.8246428279025784e-07, "loss": 0.4294, "step": 26410 }, { "epoch": 0.9409480378360083, "grad_norm": 1.6564941740800239, "learning_rate": 1.8224491828672364e-07, "loss": 0.5206, "step": 26411 }, { "epoch": 0.9409836649625024, "grad_norm": 1.6473192868018964, "learning_rate": 1.820256845134616e-07, "loss": 0.4324, "step": 26412 }, { "epoch": 0.9410192920889966, "grad_norm": 1.8281793593361575, "learning_rate": 1.8180658147338826e-07, "loss": 0.3814, "step": 26413 }, { "epoch": 0.9410549192154907, "grad_norm": 1.498717172006766, "learning_rate": 1.8158760916942465e-07, "loss": 0.5396, "step": 26414 }, { "epoch": 0.9410905463419847, "grad_norm": 2.0316987586935125, "learning_rate": 1.8136876760448286e-07, "loss": 0.5802, "step": 26415 }, { "epoch": 0.9411261734684789, "grad_norm": 1.5357518183687207, "learning_rate": 1.8115005678147945e-07, "loss": 0.4976, "step": 26416 }, { "epoch": 0.941161800594973, "grad_norm": 1.2206812550567527, "learning_rate": 1.8093147670332544e-07, "loss": 0.3746, "step": 26417 }, { "epoch": 0.9411974277214671, "grad_norm": 1.3454414533407957, "learning_rate": 1.8071302737293294e-07, "loss": 0.3849, "step": 26418 }, { "epoch": 0.9412330548479613, "grad_norm": 1.43448908723325, "learning_rate": 1.8049470879320962e-07, "loss": 0.4371, "step": 26419 }, { "epoch": 0.9412686819744553, "grad_norm": 1.2628305264801583, "learning_rate": 1.8027652096706204e-07, "loss": 0.3696, "step": 26420 }, { "epoch": 0.9413043091009494, "grad_norm": 1.5359729183611268, "learning_rate": 1.800584638973979e-07, "loss": 0.4801, "step": 26421 }, { "epoch": 0.9413399362274436, "grad_norm": 1.6827801170162042, "learning_rate": 1.7984053758711707e-07, "loss": 0.4916, "step": 26422 }, { "epoch": 0.9413755633539377, "grad_norm": 1.5074077121872231, "learning_rate": 1.7962274203912388e-07, "loss": 0.5078, "step": 26423 }, { "epoch": 0.9414111904804318, "grad_norm": 1.2711192749553186, "learning_rate": 1.7940507725631717e-07, "loss": 0.3889, "step": 26424 }, { "epoch": 0.941446817606926, "grad_norm": 1.4205004196886757, "learning_rate": 1.7918754324159792e-07, "loss": 0.4966, "step": 26425 }, { "epoch": 0.94148244473342, "grad_norm": 1.7608731853188773, "learning_rate": 1.789701399978594e-07, "loss": 0.4628, "step": 26426 }, { "epoch": 0.9415180718599141, "grad_norm": 2.1465186040513227, "learning_rate": 1.7875286752799926e-07, "loss": 0.4406, "step": 26427 }, { "epoch": 0.9415536989864083, "grad_norm": 1.1334411469093393, "learning_rate": 1.7853572583490853e-07, "loss": 0.3854, "step": 26428 }, { "epoch": 0.9415893261129024, "grad_norm": 1.5413729330745352, "learning_rate": 1.7831871492147935e-07, "loss": 0.4546, "step": 26429 }, { "epoch": 0.9416249532393964, "grad_norm": 1.8477114910735486, "learning_rate": 1.781018347906005e-07, "loss": 0.5955, "step": 26430 }, { "epoch": 0.9416605803658906, "grad_norm": 1.173005960361795, "learning_rate": 1.7788508544516192e-07, "loss": 0.3569, "step": 26431 }, { "epoch": 0.9416962074923847, "grad_norm": 1.4983657188333417, "learning_rate": 1.7766846688804906e-07, "loss": 0.4789, "step": 26432 }, { "epoch": 0.9417318346188788, "grad_norm": 1.658530605561647, "learning_rate": 1.774519791221452e-07, "loss": 0.4946, "step": 26433 }, { "epoch": 0.941767461745373, "grad_norm": 1.3737692822595038, "learning_rate": 1.7723562215033463e-07, "loss": 0.4804, "step": 26434 }, { "epoch": 0.941803088871867, "grad_norm": 1.2345474795896187, "learning_rate": 1.7701939597549622e-07, "loss": 0.4164, "step": 26435 }, { "epoch": 0.9418387159983611, "grad_norm": 1.3530097201012603, "learning_rate": 1.76803300600511e-07, "loss": 0.3766, "step": 26436 }, { "epoch": 0.9418743431248553, "grad_norm": 1.133416455954319, "learning_rate": 1.7658733602825661e-07, "loss": 0.3812, "step": 26437 }, { "epoch": 0.9419099702513494, "grad_norm": 1.198255611380733, "learning_rate": 1.7637150226160637e-07, "loss": 0.4107, "step": 26438 }, { "epoch": 0.9419455973778434, "grad_norm": 1.3007318597850315, "learning_rate": 1.7615579930343684e-07, "loss": 0.4166, "step": 26439 }, { "epoch": 0.9419812245043376, "grad_norm": 1.8430737134149944, "learning_rate": 1.7594022715661907e-07, "loss": 0.4797, "step": 26440 }, { "epoch": 0.9420168516308317, "grad_norm": 1.396885440252674, "learning_rate": 1.7572478582402407e-07, "loss": 0.3635, "step": 26441 }, { "epoch": 0.9420524787573258, "grad_norm": 1.262285147428436, "learning_rate": 1.755094753085196e-07, "loss": 0.3974, "step": 26442 }, { "epoch": 0.94208810588382, "grad_norm": 1.2939338553632689, "learning_rate": 1.752942956129744e-07, "loss": 0.4023, "step": 26443 }, { "epoch": 0.942123733010314, "grad_norm": 1.9719696654673624, "learning_rate": 1.750792467402529e-07, "loss": 0.5756, "step": 26444 }, { "epoch": 0.9421593601368081, "grad_norm": 1.6569894622006327, "learning_rate": 1.7486432869321724e-07, "loss": 0.499, "step": 26445 }, { "epoch": 0.9421949872633023, "grad_norm": 1.2859699295630485, "learning_rate": 1.7464954147473067e-07, "loss": 0.4321, "step": 26446 }, { "epoch": 0.9422306143897964, "grad_norm": 1.3959263021889723, "learning_rate": 1.7443488508765316e-07, "loss": 0.4313, "step": 26447 }, { "epoch": 0.9422662415162905, "grad_norm": 1.192697981216473, "learning_rate": 1.742203595348435e-07, "loss": 0.4114, "step": 26448 }, { "epoch": 0.9423018686427846, "grad_norm": 1.2795767878407815, "learning_rate": 1.740059648191572e-07, "loss": 0.4105, "step": 26449 }, { "epoch": 0.9423374957692787, "grad_norm": 1.2894906438256033, "learning_rate": 1.7379170094344867e-07, "loss": 0.4139, "step": 26450 }, { "epoch": 0.9423731228957728, "grad_norm": 1.2768982945499978, "learning_rate": 1.7357756791057335e-07, "loss": 0.4031, "step": 26451 }, { "epoch": 0.942408750022267, "grad_norm": 1.2885032502303067, "learning_rate": 1.7336356572337897e-07, "loss": 0.3973, "step": 26452 }, { "epoch": 0.9424443771487611, "grad_norm": 1.3372615951390427, "learning_rate": 1.7314969438471775e-07, "loss": 0.4055, "step": 26453 }, { "epoch": 0.9424800042752551, "grad_norm": 1.3650216357134684, "learning_rate": 1.7293595389743733e-07, "loss": 0.4493, "step": 26454 }, { "epoch": 0.9425156314017493, "grad_norm": 1.273589894185178, "learning_rate": 1.727223442643833e-07, "loss": 0.4256, "step": 26455 }, { "epoch": 0.9425512585282434, "grad_norm": 1.812568128775283, "learning_rate": 1.7250886548839995e-07, "loss": 0.4965, "step": 26456 }, { "epoch": 0.9425868856547375, "grad_norm": 1.3579921768444723, "learning_rate": 1.7229551757233065e-07, "loss": 0.4396, "step": 26457 }, { "epoch": 0.9426225127812317, "grad_norm": 1.960650701814736, "learning_rate": 1.720823005190153e-07, "loss": 0.4813, "step": 26458 }, { "epoch": 0.9426581399077257, "grad_norm": 1.526710527968247, "learning_rate": 1.7186921433129277e-07, "loss": 0.5544, "step": 26459 }, { "epoch": 0.9426937670342198, "grad_norm": 1.3836898486774722, "learning_rate": 1.716562590120019e-07, "loss": 0.4162, "step": 26460 }, { "epoch": 0.942729394160714, "grad_norm": 1.9926246941508239, "learning_rate": 1.7144343456397706e-07, "loss": 0.5476, "step": 26461 }, { "epoch": 0.9427650212872081, "grad_norm": 1.4238984256077745, "learning_rate": 1.7123074099005376e-07, "loss": 0.423, "step": 26462 }, { "epoch": 0.9428006484137021, "grad_norm": 1.2576839552905565, "learning_rate": 1.71018178293062e-07, "loss": 0.4092, "step": 26463 }, { "epoch": 0.9428362755401963, "grad_norm": 1.2757930371959123, "learning_rate": 1.708057464758328e-07, "loss": 0.3954, "step": 26464 }, { "epoch": 0.9428719026666904, "grad_norm": 1.5458725422777202, "learning_rate": 1.7059344554119616e-07, "loss": 0.3818, "step": 26465 }, { "epoch": 0.9429075297931845, "grad_norm": 1.4151963527122473, "learning_rate": 1.7038127549197758e-07, "loss": 0.3797, "step": 26466 }, { "epoch": 0.9429431569196787, "grad_norm": 1.7986540044599055, "learning_rate": 1.701692363310037e-07, "loss": 0.5252, "step": 26467 }, { "epoch": 0.9429787840461727, "grad_norm": 1.410799211796159, "learning_rate": 1.6995732806109555e-07, "loss": 0.4525, "step": 26468 }, { "epoch": 0.9430144111726668, "grad_norm": 1.833716693170254, "learning_rate": 1.697455506850776e-07, "loss": 0.4395, "step": 26469 }, { "epoch": 0.943050038299161, "grad_norm": 1.7556334882902271, "learning_rate": 1.6953390420576864e-07, "loss": 0.5537, "step": 26470 }, { "epoch": 0.9430856654256551, "grad_norm": 1.256907668253617, "learning_rate": 1.6932238862598648e-07, "loss": 0.3434, "step": 26471 }, { "epoch": 0.9431212925521492, "grad_norm": 1.3811633136610753, "learning_rate": 1.6911100394854772e-07, "loss": 0.4016, "step": 26472 }, { "epoch": 0.9431569196786433, "grad_norm": 1.32347740736365, "learning_rate": 1.6889975017626902e-07, "loss": 0.3709, "step": 26473 }, { "epoch": 0.9431925468051374, "grad_norm": 1.4456803507958207, "learning_rate": 1.6868862731195922e-07, "loss": 0.3983, "step": 26474 }, { "epoch": 0.9432281739316315, "grad_norm": 1.680616010805124, "learning_rate": 1.6847763535843275e-07, "loss": 0.5297, "step": 26475 }, { "epoch": 0.9432638010581257, "grad_norm": 1.4285330445448883, "learning_rate": 1.6826677431849848e-07, "loss": 0.3668, "step": 26476 }, { "epoch": 0.9432994281846198, "grad_norm": 1.9853958501450022, "learning_rate": 1.6805604419496412e-07, "loss": 0.545, "step": 26477 }, { "epoch": 0.9433350553111138, "grad_norm": 1.5392065249430078, "learning_rate": 1.6784544499063637e-07, "loss": 0.5495, "step": 26478 }, { "epoch": 0.943370682437608, "grad_norm": 1.29879776993458, "learning_rate": 1.676349767083174e-07, "loss": 0.4375, "step": 26479 }, { "epoch": 0.9434063095641021, "grad_norm": 1.5865217858065406, "learning_rate": 1.6742463935081277e-07, "loss": 0.5813, "step": 26480 }, { "epoch": 0.9434419366905962, "grad_norm": 1.2265371365198479, "learning_rate": 1.6721443292092022e-07, "loss": 0.3699, "step": 26481 }, { "epoch": 0.9434775638170904, "grad_norm": 1.2169901611093934, "learning_rate": 1.6700435742144082e-07, "loss": 0.3505, "step": 26482 }, { "epoch": 0.9435131909435844, "grad_norm": 1.2539534848869325, "learning_rate": 1.6679441285517128e-07, "loss": 0.4241, "step": 26483 }, { "epoch": 0.9435488180700785, "grad_norm": 2.3469881756038746, "learning_rate": 1.665845992249071e-07, "loss": 0.4558, "step": 26484 }, { "epoch": 0.9435844451965727, "grad_norm": 1.854405222861916, "learning_rate": 1.6637491653344273e-07, "loss": 0.5355, "step": 26485 }, { "epoch": 0.9436200723230668, "grad_norm": 1.5590757592516205, "learning_rate": 1.6616536478356927e-07, "loss": 0.5364, "step": 26486 }, { "epoch": 0.9436556994495608, "grad_norm": 1.5875198310045366, "learning_rate": 1.6595594397807669e-07, "loss": 0.5077, "step": 26487 }, { "epoch": 0.943691326576055, "grad_norm": 1.1835396391007114, "learning_rate": 1.6574665411975389e-07, "loss": 0.4522, "step": 26488 }, { "epoch": 0.9437269537025491, "grad_norm": 1.2873314610500324, "learning_rate": 1.6553749521138973e-07, "loss": 0.4121, "step": 26489 }, { "epoch": 0.9437625808290432, "grad_norm": 1.4315162847321914, "learning_rate": 1.6532846725576645e-07, "loss": 0.414, "step": 26490 }, { "epoch": 0.9437982079555374, "grad_norm": 1.4128145996128625, "learning_rate": 1.651195702556696e-07, "loss": 0.4519, "step": 26491 }, { "epoch": 0.9438338350820314, "grad_norm": 1.5976965184081702, "learning_rate": 1.6491080421387917e-07, "loss": 0.5123, "step": 26492 }, { "epoch": 0.9438694622085255, "grad_norm": 1.2675995939733808, "learning_rate": 1.6470216913317628e-07, "loss": 0.3652, "step": 26493 }, { "epoch": 0.9439050893350197, "grad_norm": 2.2693036423631825, "learning_rate": 1.6449366501633756e-07, "loss": 0.4814, "step": 26494 }, { "epoch": 0.9439407164615138, "grad_norm": 1.2907061926325551, "learning_rate": 1.6428529186614195e-07, "loss": 0.4192, "step": 26495 }, { "epoch": 0.9439763435880079, "grad_norm": 1.2366028327204475, "learning_rate": 1.6407704968536052e-07, "loss": 0.3932, "step": 26496 }, { "epoch": 0.944011970714502, "grad_norm": 1.1965693198000908, "learning_rate": 1.638689384767689e-07, "loss": 0.3962, "step": 26497 }, { "epoch": 0.9440475978409961, "grad_norm": 1.4759856751553193, "learning_rate": 1.63660958243137e-07, "loss": 0.5097, "step": 26498 }, { "epoch": 0.9440832249674902, "grad_norm": 1.4144303269356742, "learning_rate": 1.6345310898723487e-07, "loss": 0.4158, "step": 26499 }, { "epoch": 0.9441188520939844, "grad_norm": 1.5980576983828383, "learning_rate": 1.6324539071183033e-07, "loss": 0.5004, "step": 26500 }, { "epoch": 0.9441544792204785, "grad_norm": 1.3877503771004913, "learning_rate": 1.630378034196889e-07, "loss": 0.4838, "step": 26501 }, { "epoch": 0.9441901063469725, "grad_norm": 1.4827473144984207, "learning_rate": 1.6283034711357505e-07, "loss": 0.5242, "step": 26502 }, { "epoch": 0.9442257334734667, "grad_norm": 1.391655931052266, "learning_rate": 1.626230217962499e-07, "loss": 0.4175, "step": 26503 }, { "epoch": 0.9442613605999608, "grad_norm": 1.461980459445339, "learning_rate": 1.624158274704757e-07, "loss": 0.4157, "step": 26504 }, { "epoch": 0.9442969877264549, "grad_norm": 1.2637970937325875, "learning_rate": 1.6220876413901021e-07, "loss": 0.3715, "step": 26505 }, { "epoch": 0.9443326148529491, "grad_norm": 1.497720530080222, "learning_rate": 1.6200183180461237e-07, "loss": 0.3867, "step": 26506 }, { "epoch": 0.9443682419794431, "grad_norm": 1.2595591250905769, "learning_rate": 1.6179503047003553e-07, "loss": 0.4066, "step": 26507 }, { "epoch": 0.9444038691059372, "grad_norm": 1.5317509458355982, "learning_rate": 1.6158836013803525e-07, "loss": 0.4665, "step": 26508 }, { "epoch": 0.9444394962324314, "grad_norm": 1.1723963863580777, "learning_rate": 1.6138182081136155e-07, "loss": 0.4049, "step": 26509 }, { "epoch": 0.9444751233589255, "grad_norm": 1.6612565592852067, "learning_rate": 1.6117541249276557e-07, "loss": 0.5281, "step": 26510 }, { "epoch": 0.9445107504854195, "grad_norm": 1.4730201163863554, "learning_rate": 1.6096913518499735e-07, "loss": 0.4091, "step": 26511 }, { "epoch": 0.9445463776119137, "grad_norm": 1.676710079304546, "learning_rate": 1.6076298889080134e-07, "loss": 0.458, "step": 26512 }, { "epoch": 0.9445820047384078, "grad_norm": 1.4474613108186951, "learning_rate": 1.605569736129231e-07, "loss": 0.5307, "step": 26513 }, { "epoch": 0.9446176318649019, "grad_norm": 1.3118760524575666, "learning_rate": 1.6035108935410605e-07, "loss": 0.3865, "step": 26514 }, { "epoch": 0.9446532589913961, "grad_norm": 1.6081098991089091, "learning_rate": 1.601453361170924e-07, "loss": 0.6059, "step": 26515 }, { "epoch": 0.9446888861178901, "grad_norm": 1.268911687040101, "learning_rate": 1.5993971390461994e-07, "loss": 0.4028, "step": 26516 }, { "epoch": 0.9447245132443842, "grad_norm": 1.9443114681529752, "learning_rate": 1.5973422271942985e-07, "loss": 0.5338, "step": 26517 }, { "epoch": 0.9447601403708784, "grad_norm": 1.4232815045206273, "learning_rate": 1.5952886256425547e-07, "loss": 0.4901, "step": 26518 }, { "epoch": 0.9447957674973725, "grad_norm": 1.2827602746518023, "learning_rate": 1.5932363344183245e-07, "loss": 0.4237, "step": 26519 }, { "epoch": 0.9448313946238666, "grad_norm": 1.3394178085790813, "learning_rate": 1.5911853535489298e-07, "loss": 0.369, "step": 26520 }, { "epoch": 0.9448670217503607, "grad_norm": 1.6207433040836625, "learning_rate": 1.5891356830616933e-07, "loss": 0.5547, "step": 26521 }, { "epoch": 0.9449026488768548, "grad_norm": 1.524906865950899, "learning_rate": 1.5870873229838935e-07, "loss": 0.4907, "step": 26522 }, { "epoch": 0.9449382760033489, "grad_norm": 1.458165374267723, "learning_rate": 1.5850402733428194e-07, "loss": 0.4737, "step": 26523 }, { "epoch": 0.9449739031298431, "grad_norm": 1.375805728067033, "learning_rate": 1.582994534165716e-07, "loss": 0.3681, "step": 26524 }, { "epoch": 0.9450095302563372, "grad_norm": 1.2175393859893018, "learning_rate": 1.5809501054798394e-07, "loss": 0.3953, "step": 26525 }, { "epoch": 0.9450451573828312, "grad_norm": 1.5647056197754936, "learning_rate": 1.5789069873124009e-07, "loss": 0.4474, "step": 26526 }, { "epoch": 0.9450807845093254, "grad_norm": 1.2207017619200877, "learning_rate": 1.5768651796906008e-07, "loss": 0.4236, "step": 26527 }, { "epoch": 0.9451164116358195, "grad_norm": 1.3428331580113086, "learning_rate": 1.574824682641629e-07, "loss": 0.413, "step": 26528 }, { "epoch": 0.9451520387623136, "grad_norm": 1.3652810036860228, "learning_rate": 1.5727854961926748e-07, "loss": 0.3986, "step": 26529 }, { "epoch": 0.9451876658888078, "grad_norm": 1.945083245751665, "learning_rate": 1.5707476203708716e-07, "loss": 0.4438, "step": 26530 }, { "epoch": 0.9452232930153018, "grad_norm": 1.4432059064010154, "learning_rate": 1.5687110552033645e-07, "loss": 0.4854, "step": 26531 }, { "epoch": 0.9452589201417959, "grad_norm": 1.294656482264603, "learning_rate": 1.5666758007172767e-07, "loss": 0.3898, "step": 26532 }, { "epoch": 0.9452945472682901, "grad_norm": 1.562179919011619, "learning_rate": 1.564641856939686e-07, "loss": 0.5252, "step": 26533 }, { "epoch": 0.9453301743947842, "grad_norm": 1.378837780512195, "learning_rate": 1.562609223897693e-07, "loss": 0.426, "step": 26534 }, { "epoch": 0.9453658015212782, "grad_norm": 1.564911975383159, "learning_rate": 1.5605779016183654e-07, "loss": 0.3925, "step": 26535 }, { "epoch": 0.9454014286477724, "grad_norm": 1.442975707730672, "learning_rate": 1.5585478901287476e-07, "loss": 0.4006, "step": 26536 }, { "epoch": 0.9454370557742665, "grad_norm": 1.3717993482154092, "learning_rate": 1.556519189455885e-07, "loss": 0.4517, "step": 26537 }, { "epoch": 0.9454726829007606, "grad_norm": 1.2414585115989636, "learning_rate": 1.5544917996267562e-07, "loss": 0.4452, "step": 26538 }, { "epoch": 0.9455083100272548, "grad_norm": 1.2709073797790231, "learning_rate": 1.5524657206683947e-07, "loss": 0.3936, "step": 26539 }, { "epoch": 0.9455439371537488, "grad_norm": 1.2180558971034157, "learning_rate": 1.5504409526077569e-07, "loss": 0.3602, "step": 26540 }, { "epoch": 0.9455795642802429, "grad_norm": 2.0120803734478927, "learning_rate": 1.54841749547181e-07, "loss": 0.5569, "step": 26541 }, { "epoch": 0.9456151914067371, "grad_norm": 1.2636834417393097, "learning_rate": 1.5463953492874883e-07, "loss": 0.4216, "step": 26542 }, { "epoch": 0.9456508185332312, "grad_norm": 1.2930984014155025, "learning_rate": 1.5443745140817368e-07, "loss": 0.4019, "step": 26543 }, { "epoch": 0.9456864456597253, "grad_norm": 1.2630072312749703, "learning_rate": 1.542354989881456e-07, "loss": 0.3978, "step": 26544 }, { "epoch": 0.9457220727862194, "grad_norm": 1.174638085295576, "learning_rate": 1.5403367767135358e-07, "loss": 0.3942, "step": 26545 }, { "epoch": 0.9457576999127135, "grad_norm": 2.2422914768520066, "learning_rate": 1.5383198746048544e-07, "loss": 0.4891, "step": 26546 }, { "epoch": 0.9457933270392076, "grad_norm": 1.7114597636314048, "learning_rate": 1.536304283582246e-07, "loss": 0.5489, "step": 26547 }, { "epoch": 0.9458289541657018, "grad_norm": 1.3942902977757208, "learning_rate": 1.5342900036725894e-07, "loss": 0.3952, "step": 26548 }, { "epoch": 0.9458645812921959, "grad_norm": 1.5503026993189937, "learning_rate": 1.532277034902674e-07, "loss": 0.5362, "step": 26549 }, { "epoch": 0.9459002084186899, "grad_norm": 1.5900331992764787, "learning_rate": 1.5302653772993226e-07, "loss": 0.4806, "step": 26550 }, { "epoch": 0.9459358355451841, "grad_norm": 1.5620706345185484, "learning_rate": 1.528255030889303e-07, "loss": 0.3983, "step": 26551 }, { "epoch": 0.9459714626716782, "grad_norm": 1.567512520679427, "learning_rate": 1.5262459956994047e-07, "loss": 0.4096, "step": 26552 }, { "epoch": 0.9460070897981723, "grad_norm": 1.7339689153905966, "learning_rate": 1.524238271756362e-07, "loss": 0.4699, "step": 26553 }, { "epoch": 0.9460427169246665, "grad_norm": 1.201639228323897, "learning_rate": 1.5222318590869312e-07, "loss": 0.402, "step": 26554 }, { "epoch": 0.9460783440511605, "grad_norm": 1.3444774885305528, "learning_rate": 1.520226757717802e-07, "loss": 0.3986, "step": 26555 }, { "epoch": 0.9461139711776546, "grad_norm": 1.3926367770873032, "learning_rate": 1.5182229676756867e-07, "loss": 0.4663, "step": 26556 }, { "epoch": 0.9461495983041488, "grad_norm": 1.237495832408142, "learning_rate": 1.5162204889872633e-07, "loss": 0.4062, "step": 26557 }, { "epoch": 0.9461852254306429, "grad_norm": 1.236116505491425, "learning_rate": 1.5142193216792111e-07, "loss": 0.3539, "step": 26558 }, { "epoch": 0.946220852557137, "grad_norm": 1.4082570148627802, "learning_rate": 1.512219465778164e-07, "loss": 0.3966, "step": 26559 }, { "epoch": 0.9462564796836311, "grad_norm": 1.2445630368513845, "learning_rate": 1.5102209213107454e-07, "loss": 0.3804, "step": 26560 }, { "epoch": 0.9462921068101252, "grad_norm": 1.6745485765807513, "learning_rate": 1.5082236883035896e-07, "loss": 0.5249, "step": 26561 }, { "epoch": 0.9463277339366194, "grad_norm": 1.4856386442605267, "learning_rate": 1.5062277667832637e-07, "loss": 0.4626, "step": 26562 }, { "epoch": 0.9463633610631135, "grad_norm": 1.8329394401185404, "learning_rate": 1.504233156776369e-07, "loss": 0.4716, "step": 26563 }, { "epoch": 0.9463989881896075, "grad_norm": 1.5752133616310446, "learning_rate": 1.502239858309451e-07, "loss": 0.4049, "step": 26564 }, { "epoch": 0.9464346153161017, "grad_norm": 1.5469766105516725, "learning_rate": 1.500247871409055e-07, "loss": 0.5171, "step": 26565 }, { "epoch": 0.9464702424425958, "grad_norm": 1.618173924230518, "learning_rate": 1.4982571961016934e-07, "loss": 0.4922, "step": 26566 }, { "epoch": 0.9465058695690899, "grad_norm": 1.7259621075671983, "learning_rate": 1.4962678324139002e-07, "loss": 0.6118, "step": 26567 }, { "epoch": 0.9465414966955841, "grad_norm": 1.2846146170488941, "learning_rate": 1.4942797803721543e-07, "loss": 0.4132, "step": 26568 }, { "epoch": 0.9465771238220781, "grad_norm": 1.3470354517582372, "learning_rate": 1.4922930400029123e-07, "loss": 0.4018, "step": 26569 }, { "epoch": 0.9466127509485722, "grad_norm": 1.7123576271264942, "learning_rate": 1.4903076113326532e-07, "loss": 0.493, "step": 26570 }, { "epoch": 0.9466483780750664, "grad_norm": 1.2838956705827065, "learning_rate": 1.4883234943877888e-07, "loss": 0.4059, "step": 26571 }, { "epoch": 0.9466840052015605, "grad_norm": 1.6338654028132253, "learning_rate": 1.486340689194754e-07, "loss": 0.4755, "step": 26572 }, { "epoch": 0.9467196323280546, "grad_norm": 1.2750094193004156, "learning_rate": 1.4843591957799497e-07, "loss": 0.4193, "step": 26573 }, { "epoch": 0.9467552594545487, "grad_norm": 1.3517871796898016, "learning_rate": 1.482379014169777e-07, "loss": 0.4039, "step": 26574 }, { "epoch": 0.9467908865810428, "grad_norm": 1.5252111938910633, "learning_rate": 1.4804001443905703e-07, "loss": 0.4801, "step": 26575 }, { "epoch": 0.9468265137075369, "grad_norm": 1.9544262636607652, "learning_rate": 1.4784225864687085e-07, "loss": 0.4982, "step": 26576 }, { "epoch": 0.9468621408340311, "grad_norm": 1.152105512039777, "learning_rate": 1.4764463404305042e-07, "loss": 0.3727, "step": 26577 }, { "epoch": 0.9468977679605252, "grad_norm": 1.2864788969809364, "learning_rate": 1.4744714063022914e-07, "loss": 0.4025, "step": 26578 }, { "epoch": 0.9469333950870192, "grad_norm": 1.5546315951682856, "learning_rate": 1.4724977841103493e-07, "loss": 0.4964, "step": 26579 }, { "epoch": 0.9469690222135134, "grad_norm": 1.5174730180877782, "learning_rate": 1.470525473880957e-07, "loss": 0.5464, "step": 26580 }, { "epoch": 0.9470046493400075, "grad_norm": 1.6666697403927835, "learning_rate": 1.4685544756403935e-07, "loss": 0.497, "step": 26581 }, { "epoch": 0.9470402764665016, "grad_norm": 1.4820989510153328, "learning_rate": 1.4665847894148933e-07, "loss": 0.4499, "step": 26582 }, { "epoch": 0.9470759035929958, "grad_norm": 1.2189905614478085, "learning_rate": 1.464616415230702e-07, "loss": 0.3805, "step": 26583 }, { "epoch": 0.9471115307194898, "grad_norm": 1.4080180434827947, "learning_rate": 1.4626493531139985e-07, "loss": 0.48, "step": 26584 }, { "epoch": 0.9471471578459839, "grad_norm": 1.282592990557646, "learning_rate": 1.4606836030910066e-07, "loss": 0.4083, "step": 26585 }, { "epoch": 0.9471827849724781, "grad_norm": 1.2534044573776968, "learning_rate": 1.458719165187883e-07, "loss": 0.3753, "step": 26586 }, { "epoch": 0.9472184120989722, "grad_norm": 1.2048130076487418, "learning_rate": 1.4567560394307844e-07, "loss": 0.3931, "step": 26587 }, { "epoch": 0.9472540392254662, "grad_norm": 1.2782986474431144, "learning_rate": 1.454794225845857e-07, "loss": 0.3779, "step": 26588 }, { "epoch": 0.9472896663519604, "grad_norm": 1.365764647292265, "learning_rate": 1.452833724459224e-07, "loss": 0.398, "step": 26589 }, { "epoch": 0.9473252934784545, "grad_norm": 1.7043997563467992, "learning_rate": 1.4508745352969865e-07, "loss": 0.5122, "step": 26590 }, { "epoch": 0.9473609206049486, "grad_norm": 1.432675160499536, "learning_rate": 1.4489166583852466e-07, "loss": 0.5, "step": 26591 }, { "epoch": 0.9473965477314428, "grad_norm": 1.9392428554041867, "learning_rate": 1.4469600937500604e-07, "loss": 0.5878, "step": 26592 }, { "epoch": 0.9474321748579368, "grad_norm": 1.3239950157497329, "learning_rate": 1.4450048414174854e-07, "loss": 0.4252, "step": 26593 }, { "epoch": 0.9474678019844309, "grad_norm": 1.3406072211292992, "learning_rate": 1.4430509014135564e-07, "loss": 0.4051, "step": 26594 }, { "epoch": 0.9475034291109251, "grad_norm": 1.643505979811503, "learning_rate": 1.4410982737642853e-07, "loss": 0.5285, "step": 26595 }, { "epoch": 0.9475390562374192, "grad_norm": 4.76346562879647, "learning_rate": 1.4391469584956852e-07, "loss": 0.3662, "step": 26596 }, { "epoch": 0.9475746833639133, "grad_norm": 1.3913351726656862, "learning_rate": 1.437196955633735e-07, "loss": 0.4699, "step": 26597 }, { "epoch": 0.9476103104904074, "grad_norm": 1.8036963296383293, "learning_rate": 1.435248265204403e-07, "loss": 0.4565, "step": 26598 }, { "epoch": 0.9476459376169015, "grad_norm": 1.5122410378350093, "learning_rate": 1.433300887233624e-07, "loss": 0.5245, "step": 26599 }, { "epoch": 0.9476815647433956, "grad_norm": 1.1735109391898664, "learning_rate": 1.431354821747344e-07, "loss": 0.4052, "step": 26600 }, { "epoch": 0.9477171918698898, "grad_norm": 1.729260173124942, "learning_rate": 1.429410068771464e-07, "loss": 0.5441, "step": 26601 }, { "epoch": 0.9477528189963839, "grad_norm": 1.370455214523715, "learning_rate": 1.427466628331886e-07, "loss": 0.4016, "step": 26602 }, { "epoch": 0.9477884461228779, "grad_norm": 1.286054894928712, "learning_rate": 1.4255245004544894e-07, "loss": 0.4866, "step": 26603 }, { "epoch": 0.9478240732493721, "grad_norm": 1.2868084000761493, "learning_rate": 1.423583685165131e-07, "loss": 0.4178, "step": 26604 }, { "epoch": 0.9478597003758662, "grad_norm": 1.2937058133347261, "learning_rate": 1.4216441824896565e-07, "loss": 0.4346, "step": 26605 }, { "epoch": 0.9478953275023603, "grad_norm": 1.5317864497543792, "learning_rate": 1.4197059924539015e-07, "loss": 0.5278, "step": 26606 }, { "epoch": 0.9479309546288545, "grad_norm": 1.5924619424383928, "learning_rate": 1.417769115083656e-07, "loss": 0.5586, "step": 26607 }, { "epoch": 0.9479665817553485, "grad_norm": 1.508932447828594, "learning_rate": 1.4158335504047216e-07, "loss": 0.4205, "step": 26608 }, { "epoch": 0.9480022088818426, "grad_norm": 1.7038318064384355, "learning_rate": 1.4138992984428669e-07, "loss": 0.5699, "step": 26609 }, { "epoch": 0.9480378360083368, "grad_norm": 1.3315947387287586, "learning_rate": 1.4119663592238486e-07, "loss": 0.4059, "step": 26610 }, { "epoch": 0.9480734631348309, "grad_norm": 1.3410596348765276, "learning_rate": 1.4100347327734244e-07, "loss": 0.5102, "step": 26611 }, { "epoch": 0.948109090261325, "grad_norm": 1.6044732498857104, "learning_rate": 1.4081044191172733e-07, "loss": 0.417, "step": 26612 }, { "epoch": 0.9481447173878191, "grad_norm": 1.1508102581913708, "learning_rate": 1.4061754182811415e-07, "loss": 0.3792, "step": 26613 }, { "epoch": 0.9481803445143132, "grad_norm": 1.573443758237781, "learning_rate": 1.4042477302906976e-07, "loss": 0.4542, "step": 26614 }, { "epoch": 0.9482159716408073, "grad_norm": 1.4682831281237823, "learning_rate": 1.4023213551715985e-07, "loss": 0.5038, "step": 26615 }, { "epoch": 0.9482515987673015, "grad_norm": 1.8180095778254546, "learning_rate": 1.400396292949513e-07, "loss": 0.5213, "step": 26616 }, { "epoch": 0.9482872258937955, "grad_norm": 1.7795559788861117, "learning_rate": 1.3984725436500645e-07, "loss": 0.5508, "step": 26617 }, { "epoch": 0.9483228530202896, "grad_norm": 1.5501042436082628, "learning_rate": 1.3965501072988662e-07, "loss": 0.4555, "step": 26618 }, { "epoch": 0.9483584801467838, "grad_norm": 1.419405092822101, "learning_rate": 1.394628983921531e-07, "loss": 0.5384, "step": 26619 }, { "epoch": 0.9483941072732779, "grad_norm": 1.5291080307272933, "learning_rate": 1.3927091735436382e-07, "loss": 0.5668, "step": 26620 }, { "epoch": 0.948429734399772, "grad_norm": 1.5670677631521068, "learning_rate": 1.390790676190734e-07, "loss": 0.4365, "step": 26621 }, { "epoch": 0.9484653615262661, "grad_norm": 1.351477960249268, "learning_rate": 1.3888734918883873e-07, "loss": 0.4769, "step": 26622 }, { "epoch": 0.9485009886527602, "grad_norm": 1.10838693331915, "learning_rate": 1.3869576206620994e-07, "loss": 0.3965, "step": 26623 }, { "epoch": 0.9485366157792543, "grad_norm": 1.712280357713955, "learning_rate": 1.3850430625374056e-07, "loss": 0.4926, "step": 26624 }, { "epoch": 0.9485722429057485, "grad_norm": 1.453154103382087, "learning_rate": 1.3831298175397856e-07, "loss": 0.5227, "step": 26625 }, { "epoch": 0.9486078700322426, "grad_norm": 1.6086109419466812, "learning_rate": 1.38121788569473e-07, "loss": 0.575, "step": 26626 }, { "epoch": 0.9486434971587366, "grad_norm": 1.4870359044529495, "learning_rate": 1.3793072670276852e-07, "loss": 0.4907, "step": 26627 }, { "epoch": 0.9486791242852308, "grad_norm": 1.6067115089100776, "learning_rate": 1.3773979615640976e-07, "loss": 0.4315, "step": 26628 }, { "epoch": 0.9487147514117249, "grad_norm": 1.5036573442717693, "learning_rate": 1.375489969329391e-07, "loss": 0.5267, "step": 26629 }, { "epoch": 0.948750378538219, "grad_norm": 1.450489084363313, "learning_rate": 1.3735832903489676e-07, "loss": 0.3557, "step": 26630 }, { "epoch": 0.9487860056647132, "grad_norm": 2.0380298277689084, "learning_rate": 1.371677924648207e-07, "loss": 0.4766, "step": 26631 }, { "epoch": 0.9488216327912072, "grad_norm": 1.8107902428850904, "learning_rate": 1.3697738722525e-07, "loss": 0.596, "step": 26632 }, { "epoch": 0.9488572599177013, "grad_norm": 3.287674406798536, "learning_rate": 1.367871133187204e-07, "loss": 0.5737, "step": 26633 }, { "epoch": 0.9488928870441955, "grad_norm": 1.3988551882491467, "learning_rate": 1.3659697074776324e-07, "loss": 0.5526, "step": 26634 }, { "epoch": 0.9489285141706896, "grad_norm": 1.267284887025957, "learning_rate": 1.36406959514912e-07, "loss": 0.3918, "step": 26635 }, { "epoch": 0.9489641412971836, "grad_norm": 1.4935308776574399, "learning_rate": 1.3621707962269582e-07, "loss": 0.3771, "step": 26636 }, { "epoch": 0.9489997684236778, "grad_norm": 1.2597517549185682, "learning_rate": 1.3602733107364373e-07, "loss": 0.433, "step": 26637 }, { "epoch": 0.9490353955501719, "grad_norm": 2.3636485233277695, "learning_rate": 1.3583771387028267e-07, "loss": 0.5132, "step": 26638 }, { "epoch": 0.949071022676666, "grad_norm": 1.401971867064944, "learning_rate": 1.356482280151361e-07, "loss": 0.411, "step": 26639 }, { "epoch": 0.9491066498031602, "grad_norm": 1.622085768542098, "learning_rate": 1.3545887351072872e-07, "loss": 0.4017, "step": 26640 }, { "epoch": 0.9491422769296542, "grad_norm": 1.2624223752281372, "learning_rate": 1.3526965035958073e-07, "loss": 0.4122, "step": 26641 }, { "epoch": 0.9491779040561483, "grad_norm": 1.7627134294377393, "learning_rate": 1.3508055856421452e-07, "loss": 0.4784, "step": 26642 }, { "epoch": 0.9492135311826425, "grad_norm": 1.288659079388181, "learning_rate": 1.348915981271437e-07, "loss": 0.3902, "step": 26643 }, { "epoch": 0.9492491583091366, "grad_norm": 1.1930199130142816, "learning_rate": 1.347027690508873e-07, "loss": 0.3982, "step": 26644 }, { "epoch": 0.9492847854356307, "grad_norm": 1.642281074527148, "learning_rate": 1.345140713379589e-07, "loss": 0.4606, "step": 26645 }, { "epoch": 0.9493204125621248, "grad_norm": 1.5355079428313592, "learning_rate": 1.3432550499087093e-07, "loss": 0.5124, "step": 26646 }, { "epoch": 0.9493560396886189, "grad_norm": 1.6375146432664371, "learning_rate": 1.3413707001213472e-07, "loss": 0.4743, "step": 26647 }, { "epoch": 0.949391666815113, "grad_norm": 1.4241380549520883, "learning_rate": 1.3394876640425937e-07, "loss": 0.3897, "step": 26648 }, { "epoch": 0.9494272939416072, "grad_norm": 1.261036535161184, "learning_rate": 1.3376059416975174e-07, "loss": 0.4013, "step": 26649 }, { "epoch": 0.9494629210681013, "grad_norm": 1.861765447631177, "learning_rate": 1.3357255331111874e-07, "loss": 0.5857, "step": 26650 }, { "epoch": 0.9494985481945953, "grad_norm": 1.235452211844119, "learning_rate": 1.333846438308628e-07, "loss": 0.4154, "step": 26651 }, { "epoch": 0.9495341753210895, "grad_norm": 1.1899139087233408, "learning_rate": 1.331968657314864e-07, "loss": 0.4356, "step": 26652 }, { "epoch": 0.9495698024475836, "grad_norm": 1.4421437125162164, "learning_rate": 1.330092190154919e-07, "loss": 0.4187, "step": 26653 }, { "epoch": 0.9496054295740777, "grad_norm": 1.395814367189242, "learning_rate": 1.3282170368537407e-07, "loss": 0.3977, "step": 26654 }, { "epoch": 0.9496410567005719, "grad_norm": 1.4849964465265795, "learning_rate": 1.3263431974363306e-07, "loss": 0.4508, "step": 26655 }, { "epoch": 0.9496766838270659, "grad_norm": 2.2377588606036323, "learning_rate": 1.3244706719276246e-07, "loss": 0.493, "step": 26656 }, { "epoch": 0.94971231095356, "grad_norm": 2.0555654959114906, "learning_rate": 1.3225994603525694e-07, "loss": 0.5992, "step": 26657 }, { "epoch": 0.9497479380800542, "grad_norm": 1.3114223917262946, "learning_rate": 1.3207295627360672e-07, "loss": 0.4328, "step": 26658 }, { "epoch": 0.9497835652065483, "grad_norm": 1.2018046072121296, "learning_rate": 1.3188609791030316e-07, "loss": 0.3745, "step": 26659 }, { "epoch": 0.9498191923330424, "grad_norm": 1.2661003178741275, "learning_rate": 1.3169937094783314e-07, "loss": 0.4098, "step": 26660 }, { "epoch": 0.9498548194595365, "grad_norm": 1.1689104386517148, "learning_rate": 1.3151277538868245e-07, "loss": 0.4292, "step": 26661 }, { "epoch": 0.9498904465860306, "grad_norm": 1.5726439227571865, "learning_rate": 1.31326311235338e-07, "loss": 0.5, "step": 26662 }, { "epoch": 0.9499260737125247, "grad_norm": 1.6304754982915257, "learning_rate": 1.3113997849028114e-07, "loss": 0.3629, "step": 26663 }, { "epoch": 0.9499617008390189, "grad_norm": 1.2308132842927428, "learning_rate": 1.3095377715599433e-07, "loss": 0.3909, "step": 26664 }, { "epoch": 0.949997327965513, "grad_norm": 1.5513113859773404, "learning_rate": 1.3076770723495446e-07, "loss": 0.4551, "step": 26665 }, { "epoch": 0.950032955092007, "grad_norm": 1.5176835025324695, "learning_rate": 1.3058176872964178e-07, "loss": 0.4531, "step": 26666 }, { "epoch": 0.9500685822185012, "grad_norm": 1.8013671889964848, "learning_rate": 1.30395961642531e-07, "loss": 0.4178, "step": 26667 }, { "epoch": 0.9501042093449953, "grad_norm": 1.3943112104521216, "learning_rate": 1.3021028597609676e-07, "loss": 0.5042, "step": 26668 }, { "epoch": 0.9501398364714894, "grad_norm": 1.3049916385370133, "learning_rate": 1.3002474173280932e-07, "loss": 0.3654, "step": 26669 }, { "epoch": 0.9501754635979835, "grad_norm": 1.3917162481134377, "learning_rate": 1.298393289151434e-07, "loss": 0.5042, "step": 26670 }, { "epoch": 0.9502110907244776, "grad_norm": 1.3508308830046247, "learning_rate": 1.2965404752556477e-07, "loss": 0.5135, "step": 26671 }, { "epoch": 0.9502467178509717, "grad_norm": 1.3231302227693111, "learning_rate": 1.2946889756654146e-07, "loss": 0.4473, "step": 26672 }, { "epoch": 0.9502823449774659, "grad_norm": 1.3807132506958575, "learning_rate": 1.292838790405393e-07, "loss": 0.4993, "step": 26673 }, { "epoch": 0.95031797210396, "grad_norm": 1.7206348558333397, "learning_rate": 1.2909899195002072e-07, "loss": 0.482, "step": 26674 }, { "epoch": 0.950353599230454, "grad_norm": 1.2416321041599783, "learning_rate": 1.2891423629744938e-07, "loss": 0.4032, "step": 26675 }, { "epoch": 0.9503892263569482, "grad_norm": 1.4587793172555243, "learning_rate": 1.2872961208528323e-07, "loss": 0.5069, "step": 26676 }, { "epoch": 0.9504248534834423, "grad_norm": 1.775617569647153, "learning_rate": 1.2854511931598257e-07, "loss": 0.4624, "step": 26677 }, { "epoch": 0.9504604806099364, "grad_norm": 1.6876305112496945, "learning_rate": 1.283607579920032e-07, "loss": 0.5284, "step": 26678 }, { "epoch": 0.9504961077364306, "grad_norm": 1.2709779387784255, "learning_rate": 1.2817652811579984e-07, "loss": 0.4588, "step": 26679 }, { "epoch": 0.9505317348629246, "grad_norm": 1.279023526185687, "learning_rate": 1.2799242968982717e-07, "loss": 0.418, "step": 26680 }, { "epoch": 0.9505673619894187, "grad_norm": 1.421416972782459, "learning_rate": 1.2780846271653435e-07, "loss": 0.4893, "step": 26681 }, { "epoch": 0.9506029891159129, "grad_norm": 2.248462225054512, "learning_rate": 1.2762462719837276e-07, "loss": 0.4071, "step": 26682 }, { "epoch": 0.950638616242407, "grad_norm": 1.6049826144539787, "learning_rate": 1.2744092313778933e-07, "loss": 0.4982, "step": 26683 }, { "epoch": 0.950674243368901, "grad_norm": 1.3600536554074365, "learning_rate": 1.2725735053722986e-07, "loss": 0.4167, "step": 26684 }, { "epoch": 0.9507098704953952, "grad_norm": 1.3782594097471204, "learning_rate": 1.2707390939914022e-07, "loss": 0.3767, "step": 26685 }, { "epoch": 0.9507454976218893, "grad_norm": 1.4869302723339088, "learning_rate": 1.2689059972596284e-07, "loss": 0.5657, "step": 26686 }, { "epoch": 0.9507811247483834, "grad_norm": 1.5497089156000805, "learning_rate": 1.2670742152013693e-07, "loss": 0.4723, "step": 26687 }, { "epoch": 0.9508167518748776, "grad_norm": 1.4566775789344044, "learning_rate": 1.2652437478410272e-07, "loss": 0.3892, "step": 26688 }, { "epoch": 0.9508523790013717, "grad_norm": 1.5686693991076779, "learning_rate": 1.263414595202983e-07, "loss": 0.4691, "step": 26689 }, { "epoch": 0.9508880061278657, "grad_norm": 1.2840977385117884, "learning_rate": 1.2615867573115836e-07, "loss": 0.4558, "step": 26690 }, { "epoch": 0.9509236332543599, "grad_norm": 1.7148643742109742, "learning_rate": 1.2597602341911764e-07, "loss": 0.5092, "step": 26691 }, { "epoch": 0.950959260380854, "grad_norm": 1.4141186558549985, "learning_rate": 1.2579350258660638e-07, "loss": 0.4001, "step": 26692 }, { "epoch": 0.9509948875073481, "grad_norm": 1.287118295118443, "learning_rate": 1.2561111323605714e-07, "loss": 0.4047, "step": 26693 }, { "epoch": 0.9510305146338422, "grad_norm": 1.7963749583671007, "learning_rate": 1.254288553698979e-07, "loss": 0.5406, "step": 26694 }, { "epoch": 0.9510661417603363, "grad_norm": 1.4855781907537366, "learning_rate": 1.2524672899055456e-07, "loss": 0.4602, "step": 26695 }, { "epoch": 0.9511017688868304, "grad_norm": 1.4307771334920936, "learning_rate": 1.2506473410045295e-07, "loss": 0.4671, "step": 26696 }, { "epoch": 0.9511373960133246, "grad_norm": 1.5142887140423917, "learning_rate": 1.2488287070201776e-07, "loss": 0.4846, "step": 26697 }, { "epoch": 0.9511730231398187, "grad_norm": 1.6220599276478542, "learning_rate": 1.247011387976682e-07, "loss": 0.5114, "step": 26698 }, { "epoch": 0.9512086502663127, "grad_norm": 1.151966138260999, "learning_rate": 1.2451953838982566e-07, "loss": 0.3512, "step": 26699 }, { "epoch": 0.9512442773928069, "grad_norm": 1.5332819247639191, "learning_rate": 1.2433806948090822e-07, "loss": 0.4505, "step": 26700 }, { "epoch": 0.951279904519301, "grad_norm": 1.1959729391000546, "learning_rate": 1.2415673207333278e-07, "loss": 0.3733, "step": 26701 }, { "epoch": 0.9513155316457951, "grad_norm": 1.2959001865412572, "learning_rate": 1.2397552616951302e-07, "loss": 0.4166, "step": 26702 }, { "epoch": 0.9513511587722893, "grad_norm": 1.573923576849146, "learning_rate": 1.2379445177186144e-07, "loss": 0.5841, "step": 26703 }, { "epoch": 0.9513867858987833, "grad_norm": 1.2410914648707858, "learning_rate": 1.2361350888279055e-07, "loss": 0.3611, "step": 26704 }, { "epoch": 0.9514224130252774, "grad_norm": 1.2734406287294424, "learning_rate": 1.2343269750470842e-07, "loss": 0.4428, "step": 26705 }, { "epoch": 0.9514580401517716, "grad_norm": 1.3844869065745598, "learning_rate": 1.2325201764002316e-07, "loss": 0.4699, "step": 26706 }, { "epoch": 0.9514936672782657, "grad_norm": 1.6101382789028722, "learning_rate": 1.2307146929113945e-07, "loss": 0.5215, "step": 26707 }, { "epoch": 0.9515292944047598, "grad_norm": 1.5357697787678353, "learning_rate": 1.228910524604654e-07, "loss": 0.5009, "step": 26708 }, { "epoch": 0.9515649215312539, "grad_norm": 1.2911762139138772, "learning_rate": 1.2271076715039909e-07, "loss": 0.411, "step": 26709 }, { "epoch": 0.951600548657748, "grad_norm": 1.581597637913059, "learning_rate": 1.2253061336334304e-07, "loss": 0.5186, "step": 26710 }, { "epoch": 0.9516361757842421, "grad_norm": 1.3265820867073832, "learning_rate": 1.2235059110169534e-07, "loss": 0.4069, "step": 26711 }, { "epoch": 0.9516718029107363, "grad_norm": 1.4965545311430895, "learning_rate": 1.2217070036785406e-07, "loss": 0.3657, "step": 26712 }, { "epoch": 0.9517074300372304, "grad_norm": 1.6883355848330834, "learning_rate": 1.2199094116421395e-07, "loss": 0.4252, "step": 26713 }, { "epoch": 0.9517430571637244, "grad_norm": 1.4994546251553167, "learning_rate": 1.2181131349316866e-07, "loss": 0.5132, "step": 26714 }, { "epoch": 0.9517786842902186, "grad_norm": 1.3715263212773656, "learning_rate": 1.2163181735711072e-07, "loss": 0.4012, "step": 26715 }, { "epoch": 0.9518143114167127, "grad_norm": 1.4004623780412455, "learning_rate": 1.2145245275842932e-07, "loss": 0.3806, "step": 26716 }, { "epoch": 0.9518499385432068, "grad_norm": 1.5311422437441293, "learning_rate": 1.2127321969951367e-07, "loss": 0.4031, "step": 26717 }, { "epoch": 0.951885565669701, "grad_norm": 1.5206661767279195, "learning_rate": 1.2109411818274851e-07, "loss": 0.4867, "step": 26718 }, { "epoch": 0.951921192796195, "grad_norm": 1.6302423577265879, "learning_rate": 1.2091514821052197e-07, "loss": 0.5217, "step": 26719 }, { "epoch": 0.9519568199226891, "grad_norm": 1.4846355287773367, "learning_rate": 1.2073630978521432e-07, "loss": 0.5102, "step": 26720 }, { "epoch": 0.9519924470491833, "grad_norm": 1.3080536070430693, "learning_rate": 1.2055760290920704e-07, "loss": 0.3908, "step": 26721 }, { "epoch": 0.9520280741756774, "grad_norm": 1.6576235044255452, "learning_rate": 1.203790275848815e-07, "loss": 0.5536, "step": 26722 }, { "epoch": 0.9520637013021714, "grad_norm": 1.5689618918020904, "learning_rate": 1.2020058381461586e-07, "loss": 0.5673, "step": 26723 }, { "epoch": 0.9520993284286656, "grad_norm": 1.167596153087222, "learning_rate": 1.2002227160078262e-07, "loss": 0.381, "step": 26724 }, { "epoch": 0.9521349555551597, "grad_norm": 1.6173235149177871, "learning_rate": 1.19844090945761e-07, "loss": 0.5952, "step": 26725 }, { "epoch": 0.9521705826816538, "grad_norm": 1.480488427752435, "learning_rate": 1.196660418519191e-07, "loss": 0.5049, "step": 26726 }, { "epoch": 0.952206209808148, "grad_norm": 1.4614282521728854, "learning_rate": 1.1948812432163172e-07, "loss": 0.4535, "step": 26727 }, { "epoch": 0.952241836934642, "grad_norm": 1.3169196694370824, "learning_rate": 1.193103383572647e-07, "loss": 0.4042, "step": 26728 }, { "epoch": 0.9522774640611361, "grad_norm": 1.1391209023403719, "learning_rate": 1.1913268396118616e-07, "loss": 0.3811, "step": 26729 }, { "epoch": 0.9523130911876303, "grad_norm": 1.8406120120055036, "learning_rate": 1.1895516113576311e-07, "loss": 0.5089, "step": 26730 }, { "epoch": 0.9523487183141244, "grad_norm": 1.8175417184149048, "learning_rate": 1.1877776988335809e-07, "loss": 0.4972, "step": 26731 }, { "epoch": 0.9523843454406185, "grad_norm": 1.379851481039016, "learning_rate": 1.1860051020633368e-07, "loss": 0.4146, "step": 26732 }, { "epoch": 0.9524199725671126, "grad_norm": 1.5090793904434445, "learning_rate": 1.1842338210705018e-07, "loss": 0.5467, "step": 26733 }, { "epoch": 0.9524555996936067, "grad_norm": 1.3003895113235864, "learning_rate": 1.1824638558786682e-07, "loss": 0.3938, "step": 26734 }, { "epoch": 0.9524912268201008, "grad_norm": 1.2668018254853488, "learning_rate": 1.1806952065113952e-07, "loss": 0.3599, "step": 26735 }, { "epoch": 0.952526853946595, "grad_norm": 1.3436572412897463, "learning_rate": 1.1789278729922305e-07, "loss": 0.4226, "step": 26736 }, { "epoch": 0.952562481073089, "grad_norm": 1.7541600566903985, "learning_rate": 1.1771618553447217e-07, "loss": 0.5899, "step": 26737 }, { "epoch": 0.9525981081995831, "grad_norm": 1.332031437799723, "learning_rate": 1.1753971535923725e-07, "loss": 0.3877, "step": 26738 }, { "epoch": 0.9526337353260773, "grad_norm": 1.1304990074610473, "learning_rate": 1.173633767758675e-07, "loss": 0.3698, "step": 26739 }, { "epoch": 0.9526693624525714, "grad_norm": 1.4874924637333404, "learning_rate": 1.1718716978671329e-07, "loss": 0.5102, "step": 26740 }, { "epoch": 0.9527049895790655, "grad_norm": 1.8209345364063458, "learning_rate": 1.170110943941194e-07, "loss": 0.483, "step": 26741 }, { "epoch": 0.9527406167055597, "grad_norm": 1.4561268737525759, "learning_rate": 1.168351506004295e-07, "loss": 0.5518, "step": 26742 }, { "epoch": 0.9527762438320537, "grad_norm": 1.5657815109762876, "learning_rate": 1.1665933840798838e-07, "loss": 0.424, "step": 26743 }, { "epoch": 0.9528118709585478, "grad_norm": 1.5756778029491258, "learning_rate": 1.1648365781913528e-07, "loss": 0.5129, "step": 26744 }, { "epoch": 0.952847498085042, "grad_norm": 1.1872674004408788, "learning_rate": 1.1630810883621168e-07, "loss": 0.3765, "step": 26745 }, { "epoch": 0.9528831252115361, "grad_norm": 1.5967192875422054, "learning_rate": 1.1613269146155348e-07, "loss": 0.4972, "step": 26746 }, { "epoch": 0.9529187523380301, "grad_norm": 1.1139827355251044, "learning_rate": 1.1595740569749658e-07, "loss": 0.4419, "step": 26747 }, { "epoch": 0.9529543794645243, "grad_norm": 1.4869930380295195, "learning_rate": 1.1578225154637579e-07, "loss": 0.4912, "step": 26748 }, { "epoch": 0.9529900065910184, "grad_norm": 1.352437809152991, "learning_rate": 1.1560722901052368e-07, "loss": 0.4432, "step": 26749 }, { "epoch": 0.9530256337175125, "grad_norm": 1.2391832870281747, "learning_rate": 1.1543233809226839e-07, "loss": 0.4644, "step": 26750 }, { "epoch": 0.9530612608440067, "grad_norm": 2.186322540180718, "learning_rate": 1.1525757879394139e-07, "loss": 0.4605, "step": 26751 }, { "epoch": 0.9530968879705007, "grad_norm": 1.299152104502526, "learning_rate": 1.150829511178686e-07, "loss": 0.3646, "step": 26752 }, { "epoch": 0.9531325150969948, "grad_norm": 1.3931904292174329, "learning_rate": 1.1490845506637482e-07, "loss": 0.4301, "step": 26753 }, { "epoch": 0.953168142223489, "grad_norm": 1.294270954198489, "learning_rate": 1.1473409064178487e-07, "loss": 0.4024, "step": 26754 }, { "epoch": 0.9532037693499831, "grad_norm": 1.5213920130235143, "learning_rate": 1.1455985784642021e-07, "loss": 0.5605, "step": 26755 }, { "epoch": 0.9532393964764772, "grad_norm": 1.4396335011741648, "learning_rate": 1.143857566826001e-07, "loss": 0.426, "step": 26756 }, { "epoch": 0.9532750236029713, "grad_norm": 1.4520381193922647, "learning_rate": 1.142117871526438e-07, "loss": 0.464, "step": 26757 }, { "epoch": 0.9533106507294654, "grad_norm": 1.2934554564270941, "learning_rate": 1.1403794925886614e-07, "loss": 0.4103, "step": 26758 }, { "epoch": 0.9533462778559595, "grad_norm": 1.330627542744877, "learning_rate": 1.1386424300358412e-07, "loss": 0.4884, "step": 26759 }, { "epoch": 0.9533819049824537, "grad_norm": 1.7128381006489624, "learning_rate": 1.1369066838910925e-07, "loss": 0.5371, "step": 26760 }, { "epoch": 0.9534175321089478, "grad_norm": 1.725486337213236, "learning_rate": 1.1351722541775411e-07, "loss": 0.5678, "step": 26761 }, { "epoch": 0.9534531592354418, "grad_norm": 2.023618257485628, "learning_rate": 1.1334391409182687e-07, "loss": 0.5338, "step": 26762 }, { "epoch": 0.953488786361936, "grad_norm": 1.1918512962489267, "learning_rate": 1.1317073441363458e-07, "loss": 0.4272, "step": 26763 }, { "epoch": 0.9535244134884301, "grad_norm": 1.3297241909292006, "learning_rate": 1.1299768638548536e-07, "loss": 0.3973, "step": 26764 }, { "epoch": 0.9535600406149242, "grad_norm": 1.7659193360245125, "learning_rate": 1.1282477000968296e-07, "loss": 0.547, "step": 26765 }, { "epoch": 0.9535956677414184, "grad_norm": 1.213583440903087, "learning_rate": 1.1265198528852884e-07, "loss": 0.3599, "step": 26766 }, { "epoch": 0.9536312948679124, "grad_norm": 1.4144340337761032, "learning_rate": 1.124793322243245e-07, "loss": 0.5063, "step": 26767 }, { "epoch": 0.9536669219944065, "grad_norm": 2.492496281366199, "learning_rate": 1.1230681081936922e-07, "loss": 0.5401, "step": 26768 }, { "epoch": 0.9537025491209007, "grad_norm": 2.68125191056931, "learning_rate": 1.1213442107596007e-07, "loss": 0.387, "step": 26769 }, { "epoch": 0.9537381762473948, "grad_norm": 1.3736351956264072, "learning_rate": 1.1196216299639073e-07, "loss": 0.3829, "step": 26770 }, { "epoch": 0.9537738033738888, "grad_norm": 1.375593026149602, "learning_rate": 1.1179003658295828e-07, "loss": 0.4183, "step": 26771 }, { "epoch": 0.953809430500383, "grad_norm": 1.3533596122769533, "learning_rate": 1.1161804183795199e-07, "loss": 0.4088, "step": 26772 }, { "epoch": 0.9538450576268771, "grad_norm": 1.2313075039708763, "learning_rate": 1.1144617876366337e-07, "loss": 0.3656, "step": 26773 }, { "epoch": 0.9538806847533712, "grad_norm": 1.5739804652110079, "learning_rate": 1.1127444736238057e-07, "loss": 0.5017, "step": 26774 }, { "epoch": 0.9539163118798654, "grad_norm": 1.26457666167722, "learning_rate": 1.1110284763639068e-07, "loss": 0.4016, "step": 26775 }, { "epoch": 0.9539519390063594, "grad_norm": 1.6647789320886721, "learning_rate": 1.109313795879774e-07, "loss": 0.5649, "step": 26776 }, { "epoch": 0.9539875661328535, "grad_norm": 1.7578765562773153, "learning_rate": 1.1076004321942557e-07, "loss": 0.5758, "step": 26777 }, { "epoch": 0.9540231932593477, "grad_norm": 1.390229867464432, "learning_rate": 1.105888385330156e-07, "loss": 0.5525, "step": 26778 }, { "epoch": 0.9540588203858418, "grad_norm": 1.332506605193704, "learning_rate": 1.104177655310279e-07, "loss": 0.4267, "step": 26779 }, { "epoch": 0.9540944475123359, "grad_norm": 1.3758560328537963, "learning_rate": 1.1024682421573951e-07, "loss": 0.4775, "step": 26780 }, { "epoch": 0.95413007463883, "grad_norm": 1.3978085396681452, "learning_rate": 1.1007601458942751e-07, "loss": 0.3924, "step": 26781 }, { "epoch": 0.9541657017653241, "grad_norm": 1.5519452600761403, "learning_rate": 1.0990533665436565e-07, "loss": 0.5818, "step": 26782 }, { "epoch": 0.9542013288918182, "grad_norm": 1.1763584607930189, "learning_rate": 1.0973479041282654e-07, "loss": 0.3543, "step": 26783 }, { "epoch": 0.9542369560183124, "grad_norm": 1.1660715492251064, "learning_rate": 1.095643758670828e-07, "loss": 0.4026, "step": 26784 }, { "epoch": 0.9542725831448065, "grad_norm": 1.2578306925177694, "learning_rate": 1.0939409301940151e-07, "loss": 0.4244, "step": 26785 }, { "epoch": 0.9543082102713005, "grad_norm": 1.2134049946993655, "learning_rate": 1.0922394187205198e-07, "loss": 0.3279, "step": 26786 }, { "epoch": 0.9543438373977947, "grad_norm": 1.4069168647572807, "learning_rate": 1.0905392242729796e-07, "loss": 0.4706, "step": 26787 }, { "epoch": 0.9543794645242888, "grad_norm": 1.316111503337276, "learning_rate": 1.0888403468740538e-07, "loss": 0.3901, "step": 26788 }, { "epoch": 0.9544150916507829, "grad_norm": 1.5210189993977685, "learning_rate": 1.0871427865463357e-07, "loss": 0.4711, "step": 26789 }, { "epoch": 0.954450718777277, "grad_norm": 1.477584651693251, "learning_rate": 1.0854465433124628e-07, "loss": 0.4159, "step": 26790 }, { "epoch": 0.9544863459037711, "grad_norm": 1.3128060604426508, "learning_rate": 1.0837516171950057e-07, "loss": 0.3847, "step": 26791 }, { "epoch": 0.9545219730302652, "grad_norm": 1.3047536728023688, "learning_rate": 1.0820580082165244e-07, "loss": 0.3496, "step": 26792 }, { "epoch": 0.9545576001567594, "grad_norm": 1.9307475271729877, "learning_rate": 1.0803657163995896e-07, "loss": 0.5065, "step": 26793 }, { "epoch": 0.9545932272832535, "grad_norm": 1.1912611810398321, "learning_rate": 1.0786747417667165e-07, "loss": 0.4202, "step": 26794 }, { "epoch": 0.9546288544097475, "grad_norm": 1.7118460395946253, "learning_rate": 1.0769850843404428e-07, "loss": 0.5267, "step": 26795 }, { "epoch": 0.9546644815362417, "grad_norm": 1.689304832941683, "learning_rate": 1.0752967441432394e-07, "loss": 0.4678, "step": 26796 }, { "epoch": 0.9547001086627358, "grad_norm": 1.61288655085769, "learning_rate": 1.0736097211976216e-07, "loss": 0.383, "step": 26797 }, { "epoch": 0.9547357357892299, "grad_norm": 7.48070026376599, "learning_rate": 1.0719240155260269e-07, "loss": 0.371, "step": 26798 }, { "epoch": 0.9547713629157241, "grad_norm": 1.478999935855198, "learning_rate": 1.0702396271509153e-07, "loss": 0.4902, "step": 26799 }, { "epoch": 0.9548069900422181, "grad_norm": 1.5178507890718487, "learning_rate": 1.0685565560947131e-07, "loss": 0.4867, "step": 26800 }, { "epoch": 0.9548426171687122, "grad_norm": 1.2906088898486483, "learning_rate": 1.0668748023798137e-07, "loss": 0.3739, "step": 26801 }, { "epoch": 0.9548782442952064, "grad_norm": 1.5750790221734403, "learning_rate": 1.0651943660286434e-07, "loss": 0.5667, "step": 26802 }, { "epoch": 0.9549138714217005, "grad_norm": 1.494546159089485, "learning_rate": 1.0635152470635513e-07, "loss": 0.4059, "step": 26803 }, { "epoch": 0.9549494985481946, "grad_norm": 1.2083222077974038, "learning_rate": 1.061837445506908e-07, "loss": 0.3808, "step": 26804 }, { "epoch": 0.9549851256746887, "grad_norm": 1.3322318912362012, "learning_rate": 1.0601609613810515e-07, "loss": 0.4365, "step": 26805 }, { "epoch": 0.9550207528011828, "grad_norm": 1.2619175032816234, "learning_rate": 1.0584857947083194e-07, "loss": 0.3847, "step": 26806 }, { "epoch": 0.9550563799276769, "grad_norm": 1.4476323677343117, "learning_rate": 1.0568119455109938e-07, "loss": 0.4887, "step": 26807 }, { "epoch": 0.9550920070541711, "grad_norm": 1.6831410217993232, "learning_rate": 1.055139413811379e-07, "loss": 0.5367, "step": 26808 }, { "epoch": 0.9551276341806652, "grad_norm": 1.2940231308120704, "learning_rate": 1.0534681996317353e-07, "loss": 0.4468, "step": 26809 }, { "epoch": 0.9551632613071592, "grad_norm": 1.3600291719396924, "learning_rate": 1.0517983029943223e-07, "loss": 0.3814, "step": 26810 }, { "epoch": 0.9551988884336534, "grad_norm": 1.6214986429564755, "learning_rate": 1.0501297239213781e-07, "loss": 0.4819, "step": 26811 }, { "epoch": 0.9552345155601475, "grad_norm": 1.4476517024763154, "learning_rate": 1.048462462435107e-07, "loss": 0.519, "step": 26812 }, { "epoch": 0.9552701426866416, "grad_norm": 1.453080785508674, "learning_rate": 1.0467965185577356e-07, "loss": 0.461, "step": 26813 }, { "epoch": 0.9553057698131358, "grad_norm": 1.3562094730398302, "learning_rate": 1.0451318923114351e-07, "loss": 0.4028, "step": 26814 }, { "epoch": 0.9553413969396298, "grad_norm": 1.2691137851004128, "learning_rate": 1.0434685837183656e-07, "loss": 0.4266, "step": 26815 }, { "epoch": 0.9553770240661239, "grad_norm": 1.2538215506964119, "learning_rate": 1.041806592800676e-07, "loss": 0.3787, "step": 26816 }, { "epoch": 0.9554126511926181, "grad_norm": 1.9862938875315137, "learning_rate": 1.0401459195805041e-07, "loss": 0.5163, "step": 26817 }, { "epoch": 0.9554482783191122, "grad_norm": 1.2936186546748274, "learning_rate": 1.0384865640799436e-07, "loss": 0.4353, "step": 26818 }, { "epoch": 0.9554839054456062, "grad_norm": 1.8467055143136124, "learning_rate": 1.0368285263211097e-07, "loss": 0.5187, "step": 26819 }, { "epoch": 0.9555195325721004, "grad_norm": 1.830980352889485, "learning_rate": 1.0351718063260852e-07, "loss": 0.4982, "step": 26820 }, { "epoch": 0.9555551596985945, "grad_norm": 1.2440366711336046, "learning_rate": 1.0335164041169188e-07, "loss": 0.3755, "step": 26821 }, { "epoch": 0.9555907868250886, "grad_norm": 1.6662363738279158, "learning_rate": 1.0318623197156486e-07, "loss": 0.5356, "step": 26822 }, { "epoch": 0.9556264139515828, "grad_norm": 1.674360823105068, "learning_rate": 1.0302095531443013e-07, "loss": 0.4167, "step": 26823 }, { "epoch": 0.9556620410780768, "grad_norm": 1.3143046813736552, "learning_rate": 1.0285581044249037e-07, "loss": 0.4025, "step": 26824 }, { "epoch": 0.9556976682045709, "grad_norm": 1.6914273250928056, "learning_rate": 1.0269079735794163e-07, "loss": 0.5219, "step": 26825 }, { "epoch": 0.9557332953310651, "grad_norm": 1.5312986316576143, "learning_rate": 1.0252591606298434e-07, "loss": 0.6, "step": 26826 }, { "epoch": 0.9557689224575592, "grad_norm": 1.625554002137644, "learning_rate": 1.0236116655981121e-07, "loss": 0.5072, "step": 26827 }, { "epoch": 0.9558045495840533, "grad_norm": 1.6620101529390565, "learning_rate": 1.0219654885061825e-07, "loss": 0.4516, "step": 26828 }, { "epoch": 0.9558401767105474, "grad_norm": 1.2950228704448763, "learning_rate": 1.0203206293759482e-07, "loss": 0.389, "step": 26829 }, { "epoch": 0.9558758038370415, "grad_norm": 1.2200923772478183, "learning_rate": 1.0186770882293474e-07, "loss": 0.3975, "step": 26830 }, { "epoch": 0.9559114309635356, "grad_norm": 1.3773513758284883, "learning_rate": 1.0170348650882289e-07, "loss": 0.4278, "step": 26831 }, { "epoch": 0.9559470580900298, "grad_norm": 1.5880721746487045, "learning_rate": 1.0153939599744866e-07, "loss": 0.4405, "step": 26832 }, { "epoch": 0.9559826852165239, "grad_norm": 1.455473173829463, "learning_rate": 1.0137543729099474e-07, "loss": 0.5095, "step": 26833 }, { "epoch": 0.9560183123430179, "grad_norm": 1.2912203199003427, "learning_rate": 1.0121161039164606e-07, "loss": 0.3793, "step": 26834 }, { "epoch": 0.9560539394695121, "grad_norm": 1.49251686652018, "learning_rate": 1.010479153015842e-07, "loss": 0.5497, "step": 26835 }, { "epoch": 0.9560895665960062, "grad_norm": 1.6517302102338676, "learning_rate": 1.0088435202298741e-07, "loss": 0.5206, "step": 26836 }, { "epoch": 0.9561251937225003, "grad_norm": 1.2890305513531906, "learning_rate": 1.0072092055803618e-07, "loss": 0.404, "step": 26837 }, { "epoch": 0.9561608208489945, "grad_norm": 1.853030111472297, "learning_rate": 1.005576209089032e-07, "loss": 0.5032, "step": 26838 }, { "epoch": 0.9561964479754885, "grad_norm": 1.379254280615752, "learning_rate": 1.0039445307776563e-07, "loss": 0.4349, "step": 26839 }, { "epoch": 0.9562320751019826, "grad_norm": 1.2901892652455744, "learning_rate": 1.0023141706679507e-07, "loss": 0.3789, "step": 26840 }, { "epoch": 0.9562677022284768, "grad_norm": 1.5743721531013202, "learning_rate": 1.000685128781631e-07, "loss": 0.4701, "step": 26841 }, { "epoch": 0.9563033293549709, "grad_norm": 1.199491620004799, "learning_rate": 9.9905740514038e-08, "loss": 0.4109, "step": 26842 }, { "epoch": 0.9563389564814649, "grad_norm": 1.7269064620590937, "learning_rate": 9.974309997658915e-08, "loss": 0.526, "step": 26843 }, { "epoch": 0.9563745836079591, "grad_norm": 1.3557025946823322, "learning_rate": 9.958059126797925e-08, "loss": 0.5685, "step": 26844 }, { "epoch": 0.9564102107344532, "grad_norm": 1.6606733083013983, "learning_rate": 9.941821439037436e-08, "loss": 0.5152, "step": 26845 }, { "epoch": 0.9564458378609473, "grad_norm": 2.1918177111451493, "learning_rate": 9.925596934593607e-08, "loss": 0.5618, "step": 26846 }, { "epoch": 0.9564814649874415, "grad_norm": 1.3717670199437793, "learning_rate": 9.909385613682377e-08, "loss": 0.3932, "step": 26847 }, { "epoch": 0.9565170921139355, "grad_norm": 1.6057062300536402, "learning_rate": 9.893187476519794e-08, "loss": 0.3929, "step": 26848 }, { "epoch": 0.9565527192404296, "grad_norm": 1.2322648350005152, "learning_rate": 9.877002523321466e-08, "loss": 0.4447, "step": 26849 }, { "epoch": 0.9565883463669238, "grad_norm": 1.5607382980639117, "learning_rate": 9.860830754302886e-08, "loss": 0.4761, "step": 26850 }, { "epoch": 0.9566239734934179, "grad_norm": 1.1516611873214329, "learning_rate": 9.844672169679437e-08, "loss": 0.4124, "step": 26851 }, { "epoch": 0.956659600619912, "grad_norm": 1.2803876831844674, "learning_rate": 9.82852676966628e-08, "loss": 0.3707, "step": 26852 }, { "epoch": 0.9566952277464061, "grad_norm": 1.3320606120913838, "learning_rate": 9.812394554478355e-08, "loss": 0.4116, "step": 26853 }, { "epoch": 0.9567308548729002, "grad_norm": 1.2889151466480033, "learning_rate": 9.79627552433049e-08, "loss": 0.4304, "step": 26854 }, { "epoch": 0.9567664819993943, "grad_norm": 1.379900560667921, "learning_rate": 9.78016967943729e-08, "loss": 0.3896, "step": 26855 }, { "epoch": 0.9568021091258885, "grad_norm": 1.3308206854031157, "learning_rate": 9.764077020013252e-08, "loss": 0.396, "step": 26856 }, { "epoch": 0.9568377362523826, "grad_norm": 1.4660990117861903, "learning_rate": 9.747997546272647e-08, "loss": 0.4348, "step": 26857 }, { "epoch": 0.9568733633788766, "grad_norm": 1.4391141201251216, "learning_rate": 9.731931258429638e-08, "loss": 0.4217, "step": 26858 }, { "epoch": 0.9569089905053708, "grad_norm": 1.4658935393360226, "learning_rate": 9.715878156698166e-08, "loss": 0.5575, "step": 26859 }, { "epoch": 0.9569446176318649, "grad_norm": 1.5166100180672701, "learning_rate": 9.699838241291948e-08, "loss": 0.5252, "step": 26860 }, { "epoch": 0.956980244758359, "grad_norm": 1.492615564549306, "learning_rate": 9.68381151242448e-08, "loss": 0.5017, "step": 26861 }, { "epoch": 0.9570158718848532, "grad_norm": 1.2630165636692512, "learning_rate": 9.667797970309257e-08, "loss": 0.3318, "step": 26862 }, { "epoch": 0.9570514990113472, "grad_norm": 1.3824530103138182, "learning_rate": 9.651797615159552e-08, "loss": 0.4126, "step": 26863 }, { "epoch": 0.9570871261378413, "grad_norm": 1.3934336136153822, "learning_rate": 9.63581044718831e-08, "loss": 0.4485, "step": 26864 }, { "epoch": 0.9571227532643355, "grad_norm": 1.3729353142456453, "learning_rate": 9.619836466608578e-08, "loss": 0.4439, "step": 26865 }, { "epoch": 0.9571583803908296, "grad_norm": 1.6291864714232869, "learning_rate": 9.603875673632968e-08, "loss": 0.4675, "step": 26866 }, { "epoch": 0.9571940075173236, "grad_norm": 1.306792030741468, "learning_rate": 9.587928068473973e-08, "loss": 0.3976, "step": 26867 }, { "epoch": 0.9572296346438178, "grad_norm": 1.5926253817704836, "learning_rate": 9.571993651343869e-08, "loss": 0.4905, "step": 26868 }, { "epoch": 0.9572652617703119, "grad_norm": 1.3118978433691635, "learning_rate": 9.556072422454932e-08, "loss": 0.4031, "step": 26869 }, { "epoch": 0.957300888896806, "grad_norm": 1.3041044143392895, "learning_rate": 9.540164382019324e-08, "loss": 0.4209, "step": 26870 }, { "epoch": 0.9573365160233002, "grad_norm": 1.4465709451250344, "learning_rate": 9.524269530248542e-08, "loss": 0.4248, "step": 26871 }, { "epoch": 0.9573721431497942, "grad_norm": 1.4786036300852547, "learning_rate": 9.508387867354529e-08, "loss": 0.5248, "step": 26872 }, { "epoch": 0.9574077702762883, "grad_norm": 1.6775314934479217, "learning_rate": 9.492519393548672e-08, "loss": 0.3971, "step": 26873 }, { "epoch": 0.9574433974027825, "grad_norm": 1.5033672601681827, "learning_rate": 9.476664109042134e-08, "loss": 0.452, "step": 26874 }, { "epoch": 0.9574790245292766, "grad_norm": 1.921422376810557, "learning_rate": 9.46082201404619e-08, "loss": 0.5342, "step": 26875 }, { "epoch": 0.9575146516557707, "grad_norm": 1.214812816281348, "learning_rate": 9.444993108771782e-08, "loss": 0.4024, "step": 26876 }, { "epoch": 0.9575502787822648, "grad_norm": 1.4122814769012284, "learning_rate": 9.429177393429634e-08, "loss": 0.4857, "step": 26877 }, { "epoch": 0.9575859059087589, "grad_norm": 1.3651629046079434, "learning_rate": 9.413374868230352e-08, "loss": 0.5074, "step": 26878 }, { "epoch": 0.957621533035253, "grad_norm": 1.2379115575237027, "learning_rate": 9.397585533384324e-08, "loss": 0.3564, "step": 26879 }, { "epoch": 0.9576571601617472, "grad_norm": 1.442821177235768, "learning_rate": 9.381809389101826e-08, "loss": 0.4638, "step": 26880 }, { "epoch": 0.9576927872882413, "grad_norm": 1.1923056861353971, "learning_rate": 9.366046435593024e-08, "loss": 0.4212, "step": 26881 }, { "epoch": 0.9577284144147353, "grad_norm": 2.2445968583733857, "learning_rate": 9.350296673067638e-08, "loss": 0.5576, "step": 26882 }, { "epoch": 0.9577640415412295, "grad_norm": 1.4619877681987867, "learning_rate": 9.334560101735612e-08, "loss": 0.4929, "step": 26883 }, { "epoch": 0.9577996686677236, "grad_norm": 1.936782741737751, "learning_rate": 9.318836721806223e-08, "loss": 0.4376, "step": 26884 }, { "epoch": 0.9578352957942177, "grad_norm": 1.2843037652690208, "learning_rate": 9.303126533488971e-08, "loss": 0.3929, "step": 26885 }, { "epoch": 0.9578709229207119, "grad_norm": 1.3646494612707218, "learning_rate": 9.28742953699302e-08, "loss": 0.492, "step": 26886 }, { "epoch": 0.9579065500472059, "grad_norm": 1.7423848186564106, "learning_rate": 9.271745732527537e-08, "loss": 0.5326, "step": 26887 }, { "epoch": 0.9579421771737, "grad_norm": 1.3300381010210403, "learning_rate": 9.256075120301134e-08, "loss": 0.4744, "step": 26888 }, { "epoch": 0.9579778043001942, "grad_norm": 1.4086404301463182, "learning_rate": 9.240417700522641e-08, "loss": 0.4136, "step": 26889 }, { "epoch": 0.9580134314266883, "grad_norm": 1.1067542768706364, "learning_rate": 9.224773473400451e-08, "loss": 0.4276, "step": 26890 }, { "epoch": 0.9580490585531823, "grad_norm": 1.8716536824919678, "learning_rate": 9.209142439142948e-08, "loss": 0.4485, "step": 26891 }, { "epoch": 0.9580846856796765, "grad_norm": 1.5120668837532782, "learning_rate": 9.193524597958193e-08, "loss": 0.5167, "step": 26892 }, { "epoch": 0.9581203128061706, "grad_norm": 1.7573218219017797, "learning_rate": 9.177919950054237e-08, "loss": 0.5433, "step": 26893 }, { "epoch": 0.9581559399326647, "grad_norm": 1.2365232083374986, "learning_rate": 9.162328495638695e-08, "loss": 0.4253, "step": 26894 }, { "epoch": 0.9581915670591589, "grad_norm": 1.4390671361410572, "learning_rate": 9.1467502349194e-08, "loss": 0.4098, "step": 26895 }, { "epoch": 0.9582271941856529, "grad_norm": 2.034841448480273, "learning_rate": 9.13118516810374e-08, "loss": 0.4612, "step": 26896 }, { "epoch": 0.958262821312147, "grad_norm": 1.8473400775440088, "learning_rate": 9.115633295398884e-08, "loss": 0.5296, "step": 26897 }, { "epoch": 0.9582984484386412, "grad_norm": 1.5086700922749496, "learning_rate": 9.10009461701189e-08, "loss": 0.4752, "step": 26898 }, { "epoch": 0.9583340755651353, "grad_norm": 1.315401908825193, "learning_rate": 9.084569133149812e-08, "loss": 0.3932, "step": 26899 }, { "epoch": 0.9583697026916294, "grad_norm": 1.435830621697913, "learning_rate": 9.069056844019264e-08, "loss": 0.4005, "step": 26900 }, { "epoch": 0.9584053298181235, "grad_norm": 1.395171637515088, "learning_rate": 9.053557749826857e-08, "loss": 0.5235, "step": 26901 }, { "epoch": 0.9584409569446176, "grad_norm": 1.9940075130566224, "learning_rate": 9.038071850778985e-08, "loss": 0.4249, "step": 26902 }, { "epoch": 0.9584765840711117, "grad_norm": 1.6179134190815467, "learning_rate": 9.022599147081813e-08, "loss": 0.4902, "step": 26903 }, { "epoch": 0.9585122111976059, "grad_norm": 1.6136844301073205, "learning_rate": 9.007139638941287e-08, "loss": 0.5729, "step": 26904 }, { "epoch": 0.9585478383241, "grad_norm": 1.4479993404460585, "learning_rate": 8.991693326563466e-08, "loss": 0.4114, "step": 26905 }, { "epoch": 0.958583465450594, "grad_norm": 1.6602241035056597, "learning_rate": 8.976260210153964e-08, "loss": 0.4785, "step": 26906 }, { "epoch": 0.9586190925770882, "grad_norm": 1.3178442491949185, "learning_rate": 8.960840289918172e-08, "loss": 0.4019, "step": 26907 }, { "epoch": 0.9586547197035823, "grad_norm": 1.5712144907210028, "learning_rate": 8.945433566061479e-08, "loss": 0.5371, "step": 26908 }, { "epoch": 0.9586903468300764, "grad_norm": 1.3026215782246475, "learning_rate": 8.930040038789057e-08, "loss": 0.3638, "step": 26909 }, { "epoch": 0.9587259739565706, "grad_norm": 1.139473097967011, "learning_rate": 8.914659708305851e-08, "loss": 0.3936, "step": 26910 }, { "epoch": 0.9587616010830646, "grad_norm": 1.8620873292187168, "learning_rate": 8.899292574816698e-08, "loss": 0.4361, "step": 26911 }, { "epoch": 0.9587972282095587, "grad_norm": 1.2854254874222535, "learning_rate": 8.883938638526213e-08, "loss": 0.4058, "step": 26912 }, { "epoch": 0.9588328553360529, "grad_norm": 1.3664445040925117, "learning_rate": 8.868597899638897e-08, "loss": 0.4294, "step": 26913 }, { "epoch": 0.958868482462547, "grad_norm": 1.345255404811181, "learning_rate": 8.853270358358812e-08, "loss": 0.3999, "step": 26914 }, { "epoch": 0.958904109589041, "grad_norm": 1.770900223707633, "learning_rate": 8.837956014890236e-08, "loss": 0.5502, "step": 26915 }, { "epoch": 0.9589397367155352, "grad_norm": 1.5964520239367748, "learning_rate": 8.82265486943712e-08, "loss": 0.5169, "step": 26916 }, { "epoch": 0.9589753638420293, "grad_norm": 2.1856408396520957, "learning_rate": 8.807366922203075e-08, "loss": 0.616, "step": 26917 }, { "epoch": 0.9590109909685234, "grad_norm": 1.498841966738368, "learning_rate": 8.79209217339183e-08, "loss": 0.45, "step": 26918 }, { "epoch": 0.9590466180950176, "grad_norm": 1.2129212941183345, "learning_rate": 8.776830623206667e-08, "loss": 0.3922, "step": 26919 }, { "epoch": 0.9590822452215116, "grad_norm": 1.3661807839859992, "learning_rate": 8.761582271850865e-08, "loss": 0.4573, "step": 26920 }, { "epoch": 0.9591178723480057, "grad_norm": 1.7100028390171778, "learning_rate": 8.746347119527377e-08, "loss": 0.5122, "step": 26921 }, { "epoch": 0.9591534994744999, "grad_norm": 1.2469876045311714, "learning_rate": 8.731125166439036e-08, "loss": 0.4342, "step": 26922 }, { "epoch": 0.959189126600994, "grad_norm": 1.205743681444053, "learning_rate": 8.715916412788683e-08, "loss": 0.3686, "step": 26923 }, { "epoch": 0.959224753727488, "grad_norm": 1.3451478034712039, "learning_rate": 8.700720858778822e-08, "loss": 0.408, "step": 26924 }, { "epoch": 0.9592603808539822, "grad_norm": 1.236309681188517, "learning_rate": 8.685538504611734e-08, "loss": 0.4383, "step": 26925 }, { "epoch": 0.9592960079804763, "grad_norm": 1.325076560084999, "learning_rate": 8.670369350489482e-08, "loss": 0.3917, "step": 26926 }, { "epoch": 0.9593316351069704, "grad_norm": 1.5882783824265427, "learning_rate": 8.655213396614349e-08, "loss": 0.5154, "step": 26927 }, { "epoch": 0.9593672622334646, "grad_norm": 1.2388165646123368, "learning_rate": 8.640070643187837e-08, "loss": 0.3841, "step": 26928 }, { "epoch": 0.9594028893599587, "grad_norm": 1.2983542617528794, "learning_rate": 8.624941090411676e-08, "loss": 0.4042, "step": 26929 }, { "epoch": 0.9594385164864527, "grad_norm": 1.3473426571771203, "learning_rate": 8.609824738487371e-08, "loss": 0.3958, "step": 26930 }, { "epoch": 0.9594741436129469, "grad_norm": 1.3459112749997626, "learning_rate": 8.594721587616206e-08, "loss": 0.5552, "step": 26931 }, { "epoch": 0.959509770739441, "grad_norm": 1.5424169968098225, "learning_rate": 8.579631637999242e-08, "loss": 0.4153, "step": 26932 }, { "epoch": 0.9595453978659351, "grad_norm": 1.8287406140731428, "learning_rate": 8.56455488983754e-08, "loss": 0.5021, "step": 26933 }, { "epoch": 0.9595810249924293, "grad_norm": 1.2959002851914523, "learning_rate": 8.549491343331718e-08, "loss": 0.3872, "step": 26934 }, { "epoch": 0.9596166521189233, "grad_norm": 1.412621407641157, "learning_rate": 8.534440998682392e-08, "loss": 0.5188, "step": 26935 }, { "epoch": 0.9596522792454174, "grad_norm": 1.5204321132208376, "learning_rate": 8.519403856089959e-08, "loss": 0.4765, "step": 26936 }, { "epoch": 0.9596879063719116, "grad_norm": 2.316284009436368, "learning_rate": 8.504379915754702e-08, "loss": 0.5179, "step": 26937 }, { "epoch": 0.9597235334984057, "grad_norm": 1.2195598891932429, "learning_rate": 8.489369177876571e-08, "loss": 0.4148, "step": 26938 }, { "epoch": 0.9597591606248999, "grad_norm": 1.4432829048643752, "learning_rate": 8.474371642655521e-08, "loss": 0.4065, "step": 26939 }, { "epoch": 0.9597947877513939, "grad_norm": 2.262046886476175, "learning_rate": 8.459387310291389e-08, "loss": 0.4887, "step": 26940 }, { "epoch": 0.959830414877888, "grad_norm": 1.2120469788358992, "learning_rate": 8.44441618098335e-08, "loss": 0.3711, "step": 26941 }, { "epoch": 0.9598660420043822, "grad_norm": 1.4539095531699822, "learning_rate": 8.429458254931133e-08, "loss": 0.5849, "step": 26942 }, { "epoch": 0.9599016691308763, "grad_norm": 1.5502963955353295, "learning_rate": 8.41451353233369e-08, "loss": 0.5007, "step": 26943 }, { "epoch": 0.9599372962573703, "grad_norm": 1.5857153789420373, "learning_rate": 8.399582013390084e-08, "loss": 0.497, "step": 26944 }, { "epoch": 0.9599729233838645, "grad_norm": 1.6484043187203516, "learning_rate": 8.384663698299156e-08, "loss": 0.5315, "step": 26945 }, { "epoch": 0.9600085505103586, "grad_norm": 1.248928039264398, "learning_rate": 8.369758587259413e-08, "loss": 0.4388, "step": 26946 }, { "epoch": 0.9600441776368527, "grad_norm": 1.2401105196176445, "learning_rate": 8.354866680469587e-08, "loss": 0.4106, "step": 26947 }, { "epoch": 0.9600798047633469, "grad_norm": 1.7427206801371324, "learning_rate": 8.33998797812774e-08, "loss": 0.5156, "step": 26948 }, { "epoch": 0.9601154318898409, "grad_norm": 1.4517730911968794, "learning_rate": 8.325122480432046e-08, "loss": 0.5208, "step": 26949 }, { "epoch": 0.960151059016335, "grad_norm": 1.2089015132593903, "learning_rate": 8.310270187580572e-08, "loss": 0.3789, "step": 26950 }, { "epoch": 0.9601866861428292, "grad_norm": 1.5214232928160187, "learning_rate": 8.295431099771045e-08, "loss": 0.4428, "step": 26951 }, { "epoch": 0.9602223132693233, "grad_norm": 1.6108050353736587, "learning_rate": 8.280605217200866e-08, "loss": 0.4857, "step": 26952 }, { "epoch": 0.9602579403958174, "grad_norm": 1.41151017652058, "learning_rate": 8.265792540067653e-08, "loss": 0.3976, "step": 26953 }, { "epoch": 0.9602935675223115, "grad_norm": 1.2115963580242723, "learning_rate": 8.25099306856858e-08, "loss": 0.4022, "step": 26954 }, { "epoch": 0.9603291946488056, "grad_norm": 1.326120068042391, "learning_rate": 8.236206802900826e-08, "loss": 0.3882, "step": 26955 }, { "epoch": 0.9603648217752997, "grad_norm": 1.3782810391364682, "learning_rate": 8.22143374326112e-08, "loss": 0.5304, "step": 26956 }, { "epoch": 0.9604004489017939, "grad_norm": 1.2658768455782425, "learning_rate": 8.206673889846306e-08, "loss": 0.3688, "step": 26957 }, { "epoch": 0.960436076028288, "grad_norm": 1.2454227393221002, "learning_rate": 8.191927242852671e-08, "loss": 0.4092, "step": 26958 }, { "epoch": 0.960471703154782, "grad_norm": 1.2923811750516272, "learning_rate": 8.177193802476945e-08, "loss": 0.4218, "step": 26959 }, { "epoch": 0.9605073302812762, "grad_norm": 1.3781836952770288, "learning_rate": 8.162473568914975e-08, "loss": 0.416, "step": 26960 }, { "epoch": 0.9605429574077703, "grad_norm": 1.2498125566390668, "learning_rate": 8.147766542362934e-08, "loss": 0.3532, "step": 26961 }, { "epoch": 0.9605785845342644, "grad_norm": 1.80917683614938, "learning_rate": 8.133072723016777e-08, "loss": 0.4287, "step": 26962 }, { "epoch": 0.9606142116607586, "grad_norm": 2.7353984512785985, "learning_rate": 8.118392111071905e-08, "loss": 0.4654, "step": 26963 }, { "epoch": 0.9606498387872526, "grad_norm": 3.373310768440992, "learning_rate": 8.103724706724048e-08, "loss": 0.5635, "step": 26964 }, { "epoch": 0.9606854659137467, "grad_norm": 1.325035584079765, "learning_rate": 8.089070510168273e-08, "loss": 0.4125, "step": 26965 }, { "epoch": 0.9607210930402409, "grad_norm": 1.251235545157512, "learning_rate": 8.074429521599869e-08, "loss": 0.3888, "step": 26966 }, { "epoch": 0.960756720166735, "grad_norm": 1.6943712572290228, "learning_rate": 8.059801741213791e-08, "loss": 0.5095, "step": 26967 }, { "epoch": 0.960792347293229, "grad_norm": 1.6189308694774838, "learning_rate": 8.04518716920466e-08, "loss": 0.4936, "step": 26968 }, { "epoch": 0.9608279744197232, "grad_norm": 1.4737792685813509, "learning_rate": 8.030585805767321e-08, "loss": 0.4598, "step": 26969 }, { "epoch": 0.9608636015462173, "grad_norm": 1.4953742504309748, "learning_rate": 8.015997651096064e-08, "loss": 0.3835, "step": 26970 }, { "epoch": 0.9608992286727114, "grad_norm": 1.5197795769892581, "learning_rate": 8.001422705385064e-08, "loss": 0.4105, "step": 26971 }, { "epoch": 0.9609348557992056, "grad_norm": 1.2283232366097023, "learning_rate": 7.986860968828613e-08, "loss": 0.4391, "step": 26972 }, { "epoch": 0.9609704829256996, "grad_norm": 1.1605957343983409, "learning_rate": 7.972312441620444e-08, "loss": 0.3844, "step": 26973 }, { "epoch": 0.9610061100521937, "grad_norm": 1.212570163631924, "learning_rate": 7.95777712395429e-08, "loss": 0.3851, "step": 26974 }, { "epoch": 0.9610417371786879, "grad_norm": 2.101957401075467, "learning_rate": 7.943255016023665e-08, "loss": 0.4523, "step": 26975 }, { "epoch": 0.961077364305182, "grad_norm": 1.265462207346268, "learning_rate": 7.92874611802208e-08, "loss": 0.3971, "step": 26976 }, { "epoch": 0.9611129914316761, "grad_norm": 1.1883396703494946, "learning_rate": 7.914250430142711e-08, "loss": 0.3775, "step": 26977 }, { "epoch": 0.9611486185581702, "grad_norm": 1.3462405331572187, "learning_rate": 7.899767952578407e-08, "loss": 0.406, "step": 26978 }, { "epoch": 0.9611842456846643, "grad_norm": 2.0097546719021286, "learning_rate": 7.885298685522235e-08, "loss": 0.4431, "step": 26979 }, { "epoch": 0.9612198728111584, "grad_norm": 1.450346286177242, "learning_rate": 7.870842629166709e-08, "loss": 0.5122, "step": 26980 }, { "epoch": 0.9612554999376526, "grad_norm": 1.4423695652068012, "learning_rate": 7.856399783704339e-08, "loss": 0.4184, "step": 26981 }, { "epoch": 0.9612911270641467, "grad_norm": 1.3429100409120256, "learning_rate": 7.841970149327527e-08, "loss": 0.3852, "step": 26982 }, { "epoch": 0.9613267541906407, "grad_norm": 1.4076798523297376, "learning_rate": 7.827553726228343e-08, "loss": 0.4392, "step": 26983 }, { "epoch": 0.9613623813171349, "grad_norm": 1.5148232539283168, "learning_rate": 7.813150514598744e-08, "loss": 0.5362, "step": 26984 }, { "epoch": 0.961398008443629, "grad_norm": 1.5920391542026116, "learning_rate": 7.798760514630466e-08, "loss": 0.5455, "step": 26985 }, { "epoch": 0.9614336355701231, "grad_norm": 1.5700620743696911, "learning_rate": 7.784383726515355e-08, "loss": 0.3806, "step": 26986 }, { "epoch": 0.9614692626966173, "grad_norm": 1.414436376076301, "learning_rate": 7.770020150444591e-08, "loss": 0.5199, "step": 26987 }, { "epoch": 0.9615048898231113, "grad_norm": 1.2987695753795088, "learning_rate": 7.755669786609688e-08, "loss": 0.4976, "step": 26988 }, { "epoch": 0.9615405169496054, "grad_norm": 1.3473972349542953, "learning_rate": 7.741332635201381e-08, "loss": 0.3744, "step": 26989 }, { "epoch": 0.9615761440760996, "grad_norm": 1.252657044178667, "learning_rate": 7.727008696410854e-08, "loss": 0.4205, "step": 26990 }, { "epoch": 0.9616117712025937, "grad_norm": 1.9692523181605828, "learning_rate": 7.712697970428729e-08, "loss": 0.4646, "step": 26991 }, { "epoch": 0.9616473983290877, "grad_norm": 1.6702707737075069, "learning_rate": 7.698400457445742e-08, "loss": 0.5274, "step": 26992 }, { "epoch": 0.9616830254555819, "grad_norm": 1.2591076500160252, "learning_rate": 7.684116157651967e-08, "loss": 0.4154, "step": 26993 }, { "epoch": 0.961718652582076, "grad_norm": 1.5368550315295415, "learning_rate": 7.669845071237913e-08, "loss": 0.5488, "step": 26994 }, { "epoch": 0.9617542797085701, "grad_norm": 1.3352186725983293, "learning_rate": 7.655587198393321e-08, "loss": 0.4019, "step": 26995 }, { "epoch": 0.9617899068350643, "grad_norm": 1.4090989852038605, "learning_rate": 7.641342539308261e-08, "loss": 0.501, "step": 26996 }, { "epoch": 0.9618255339615583, "grad_norm": 1.4616150330856517, "learning_rate": 7.627111094172356e-08, "loss": 0.4202, "step": 26997 }, { "epoch": 0.9618611610880524, "grad_norm": 2.391857088778609, "learning_rate": 7.612892863175014e-08, "loss": 0.4824, "step": 26998 }, { "epoch": 0.9618967882145466, "grad_norm": 1.2922953592987287, "learning_rate": 7.598687846505748e-08, "loss": 0.395, "step": 26999 }, { "epoch": 0.9619324153410407, "grad_norm": 1.4788780034599862, "learning_rate": 7.584496044353517e-08, "loss": 0.5365, "step": 27000 }, { "epoch": 0.9619680424675348, "grad_norm": 1.2135246544366876, "learning_rate": 7.570317456907506e-08, "loss": 0.3773, "step": 27001 }, { "epoch": 0.9620036695940289, "grad_norm": 2.0674175112355955, "learning_rate": 7.556152084356227e-08, "loss": 0.5055, "step": 27002 }, { "epoch": 0.962039296720523, "grad_norm": 2.917400874503329, "learning_rate": 7.541999926888643e-08, "loss": 0.5513, "step": 27003 }, { "epoch": 0.9620749238470171, "grad_norm": 1.5878158397823339, "learning_rate": 7.527860984692825e-08, "loss": 0.4181, "step": 27004 }, { "epoch": 0.9621105509735113, "grad_norm": 1.2111686144935352, "learning_rate": 7.513735257957288e-08, "loss": 0.4027, "step": 27005 }, { "epoch": 0.9621461781000054, "grad_norm": 1.299267517196068, "learning_rate": 7.499622746870106e-08, "loss": 0.4018, "step": 27006 }, { "epoch": 0.9621818052264994, "grad_norm": 1.6152703231083998, "learning_rate": 7.485523451619125e-08, "loss": 0.5139, "step": 27007 }, { "epoch": 0.9622174323529936, "grad_norm": 1.487062625552497, "learning_rate": 7.471437372392198e-08, "loss": 0.5218, "step": 27008 }, { "epoch": 0.9622530594794877, "grad_norm": 1.4590839541614429, "learning_rate": 7.457364509376841e-08, "loss": 0.5296, "step": 27009 }, { "epoch": 0.9622886866059818, "grad_norm": 1.22001532376562, "learning_rate": 7.443304862760348e-08, "loss": 0.419, "step": 27010 }, { "epoch": 0.962324313732476, "grad_norm": 1.6005107349550052, "learning_rate": 7.429258432730014e-08, "loss": 0.5359, "step": 27011 }, { "epoch": 0.96235994085897, "grad_norm": 1.1945394454434468, "learning_rate": 7.415225219472799e-08, "loss": 0.3807, "step": 27012 }, { "epoch": 0.9623955679854641, "grad_norm": 1.2654430464744661, "learning_rate": 7.401205223175777e-08, "loss": 0.3912, "step": 27013 }, { "epoch": 0.9624311951119583, "grad_norm": 1.4166625253475178, "learning_rate": 7.387198444025467e-08, "loss": 0.3908, "step": 27014 }, { "epoch": 0.9624668222384524, "grad_norm": 1.4609370462669629, "learning_rate": 7.373204882208273e-08, "loss": 0.58, "step": 27015 }, { "epoch": 0.9625024493649464, "grad_norm": 1.2915089443835206, "learning_rate": 7.359224537910825e-08, "loss": 0.3905, "step": 27016 }, { "epoch": 0.9625380764914406, "grad_norm": 1.557409908412025, "learning_rate": 7.345257411318973e-08, "loss": 0.4939, "step": 27017 }, { "epoch": 0.9625737036179347, "grad_norm": 1.3022378599461912, "learning_rate": 7.331303502618903e-08, "loss": 0.3715, "step": 27018 }, { "epoch": 0.9626093307444288, "grad_norm": 1.4173640515314165, "learning_rate": 7.317362811996465e-08, "loss": 0.4759, "step": 27019 }, { "epoch": 0.962644957870923, "grad_norm": 1.5363315973162015, "learning_rate": 7.303435339637066e-08, "loss": 0.5021, "step": 27020 }, { "epoch": 0.962680584997417, "grad_norm": 1.8559253781734832, "learning_rate": 7.289521085726226e-08, "loss": 0.4821, "step": 27021 }, { "epoch": 0.9627162121239111, "grad_norm": 1.3169481345591516, "learning_rate": 7.27562005044935e-08, "loss": 0.383, "step": 27022 }, { "epoch": 0.9627518392504053, "grad_norm": 1.7839313367822733, "learning_rate": 7.261732233991514e-08, "loss": 0.5119, "step": 27023 }, { "epoch": 0.9627874663768994, "grad_norm": 1.5278940829072194, "learning_rate": 7.24785763653757e-08, "loss": 0.4758, "step": 27024 }, { "epoch": 0.9628230935033935, "grad_norm": 1.2493059132973037, "learning_rate": 7.233996258272258e-08, "loss": 0.3939, "step": 27025 }, { "epoch": 0.9628587206298876, "grad_norm": 1.4202935652313406, "learning_rate": 7.220148099380208e-08, "loss": 0.4592, "step": 27026 }, { "epoch": 0.9628943477563817, "grad_norm": 1.7010801984027801, "learning_rate": 7.206313160045831e-08, "loss": 0.4885, "step": 27027 }, { "epoch": 0.9629299748828758, "grad_norm": 1.2165224965975612, "learning_rate": 7.192491440453197e-08, "loss": 0.4295, "step": 27028 }, { "epoch": 0.96296560200937, "grad_norm": 1.5856567082592314, "learning_rate": 7.178682940786608e-08, "loss": 0.5346, "step": 27029 }, { "epoch": 0.9630012291358641, "grad_norm": 1.2686561065042692, "learning_rate": 7.164887661229691e-08, "loss": 0.4313, "step": 27030 }, { "epoch": 0.9630368562623581, "grad_norm": 1.4541697968007579, "learning_rate": 7.151105601966301e-08, "loss": 0.383, "step": 27031 }, { "epoch": 0.9630724833888523, "grad_norm": 1.2505162890106756, "learning_rate": 7.137336763179958e-08, "loss": 0.4183, "step": 27032 }, { "epoch": 0.9631081105153464, "grad_norm": 1.3196857186528836, "learning_rate": 7.123581145053849e-08, "loss": 0.3933, "step": 27033 }, { "epoch": 0.9631437376418405, "grad_norm": 1.743872240099884, "learning_rate": 7.109838747771269e-08, "loss": 0.5123, "step": 27034 }, { "epoch": 0.9631793647683347, "grad_norm": 1.1291931056579607, "learning_rate": 7.096109571515075e-08, "loss": 0.3833, "step": 27035 }, { "epoch": 0.9632149918948287, "grad_norm": 1.424531346841535, "learning_rate": 7.082393616468342e-08, "loss": 0.4855, "step": 27036 }, { "epoch": 0.9632506190213228, "grad_norm": 1.7213364378685332, "learning_rate": 7.068690882813478e-08, "loss": 0.4096, "step": 27037 }, { "epoch": 0.963286246147817, "grad_norm": 1.5281660719429804, "learning_rate": 7.055001370733117e-08, "loss": 0.4756, "step": 27038 }, { "epoch": 0.9633218732743111, "grad_norm": 1.608677211609847, "learning_rate": 7.041325080409223e-08, "loss": 0.5872, "step": 27039 }, { "epoch": 0.9633575004008051, "grad_norm": 1.2630825324824506, "learning_rate": 7.027662012024317e-08, "loss": 0.45, "step": 27040 }, { "epoch": 0.9633931275272993, "grad_norm": 1.2325968818272777, "learning_rate": 7.014012165760031e-08, "loss": 0.3501, "step": 27041 }, { "epoch": 0.9634287546537934, "grad_norm": 1.650733739397904, "learning_rate": 7.00037554179822e-08, "loss": 0.4869, "step": 27042 }, { "epoch": 0.9634643817802875, "grad_norm": 1.1933279953135827, "learning_rate": 6.986752140320519e-08, "loss": 0.4346, "step": 27043 }, { "epoch": 0.9635000089067817, "grad_norm": 1.141170190235493, "learning_rate": 6.973141961508223e-08, "loss": 0.3829, "step": 27044 }, { "epoch": 0.9635356360332757, "grad_norm": 1.2604084540301637, "learning_rate": 6.959545005542745e-08, "loss": 0.412, "step": 27045 }, { "epoch": 0.9635712631597698, "grad_norm": 2.123691054844115, "learning_rate": 6.94596127260494e-08, "loss": 0.5901, "step": 27046 }, { "epoch": 0.963606890286264, "grad_norm": 1.606282935593938, "learning_rate": 6.932390762875884e-08, "loss": 0.4477, "step": 27047 }, { "epoch": 0.9636425174127581, "grad_norm": 1.214466331252872, "learning_rate": 6.91883347653599e-08, "loss": 0.4506, "step": 27048 }, { "epoch": 0.9636781445392522, "grad_norm": 1.5638388175376838, "learning_rate": 6.905289413766004e-08, "loss": 0.5384, "step": 27049 }, { "epoch": 0.9637137716657463, "grad_norm": 1.2522701118501147, "learning_rate": 6.891758574746221e-08, "loss": 0.368, "step": 27050 }, { "epoch": 0.9637493987922404, "grad_norm": 1.2334454175787315, "learning_rate": 6.878240959656835e-08, "loss": 0.3829, "step": 27051 }, { "epoch": 0.9637850259187345, "grad_norm": 1.4053916699448088, "learning_rate": 6.86473656867781e-08, "loss": 0.3878, "step": 27052 }, { "epoch": 0.9638206530452287, "grad_norm": 1.3106858662491208, "learning_rate": 6.851245401989004e-08, "loss": 0.39, "step": 27053 }, { "epoch": 0.9638562801717228, "grad_norm": 1.3993503516753059, "learning_rate": 6.83776745977005e-08, "loss": 0.4073, "step": 27054 }, { "epoch": 0.9638919072982168, "grad_norm": 1.2491262355775912, "learning_rate": 6.824302742200251e-08, "loss": 0.3851, "step": 27055 }, { "epoch": 0.963927534424711, "grad_norm": 1.428896118698088, "learning_rate": 6.810851249459238e-08, "loss": 0.432, "step": 27056 }, { "epoch": 0.9639631615512051, "grad_norm": 1.4251885650597114, "learning_rate": 6.797412981725871e-08, "loss": 0.3846, "step": 27057 }, { "epoch": 0.9639987886776992, "grad_norm": 1.371003222526076, "learning_rate": 6.783987939179115e-08, "loss": 0.3971, "step": 27058 }, { "epoch": 0.9640344158041934, "grad_norm": 1.273693670369738, "learning_rate": 6.77057612199783e-08, "loss": 0.4408, "step": 27059 }, { "epoch": 0.9640700429306874, "grad_norm": 1.7295328713822178, "learning_rate": 6.757177530360536e-08, "loss": 0.5106, "step": 27060 }, { "epoch": 0.9641056700571815, "grad_norm": 1.6994988306088012, "learning_rate": 6.743792164445651e-08, "loss": 0.5641, "step": 27061 }, { "epoch": 0.9641412971836757, "grad_norm": 1.6151183248947816, "learning_rate": 6.73042002443136e-08, "loss": 0.5139, "step": 27062 }, { "epoch": 0.9641769243101698, "grad_norm": 1.293632374029189, "learning_rate": 6.71706111049586e-08, "loss": 0.3949, "step": 27063 }, { "epoch": 0.9642125514366638, "grad_norm": 1.5158792718100595, "learning_rate": 6.703715422816893e-08, "loss": 0.4945, "step": 27064 }, { "epoch": 0.964248178563158, "grad_norm": 1.6282260044369576, "learning_rate": 6.690382961572317e-08, "loss": 0.4318, "step": 27065 }, { "epoch": 0.9642838056896521, "grad_norm": 1.6313914664521838, "learning_rate": 6.677063726939437e-08, "loss": 0.4635, "step": 27066 }, { "epoch": 0.9643194328161462, "grad_norm": 1.216225467411173, "learning_rate": 6.663757719095887e-08, "loss": 0.4112, "step": 27067 }, { "epoch": 0.9643550599426404, "grad_norm": 1.3264721926864735, "learning_rate": 6.650464938218637e-08, "loss": 0.413, "step": 27068 }, { "epoch": 0.9643906870691344, "grad_norm": 2.0245532750121447, "learning_rate": 6.637185384484767e-08, "loss": 0.5385, "step": 27069 }, { "epoch": 0.9644263141956285, "grad_norm": 1.3922495120029186, "learning_rate": 6.623919058071138e-08, "loss": 0.3986, "step": 27070 }, { "epoch": 0.9644619413221227, "grad_norm": 1.2011053828893377, "learning_rate": 6.610665959154272e-08, "loss": 0.3901, "step": 27071 }, { "epoch": 0.9644975684486168, "grad_norm": 1.2021126291985487, "learning_rate": 6.597426087910697e-08, "loss": 0.4195, "step": 27072 }, { "epoch": 0.9645331955751109, "grad_norm": 1.4035274125248918, "learning_rate": 6.584199444516826e-08, "loss": 0.3844, "step": 27073 }, { "epoch": 0.964568822701605, "grad_norm": 1.6392336249636539, "learning_rate": 6.57098602914874e-08, "loss": 0.3903, "step": 27074 }, { "epoch": 0.9646044498280991, "grad_norm": 1.2131899635637993, "learning_rate": 6.557785841982301e-08, "loss": 0.3928, "step": 27075 }, { "epoch": 0.9646400769545932, "grad_norm": 1.3962238171011703, "learning_rate": 6.544598883193253e-08, "loss": 0.516, "step": 27076 }, { "epoch": 0.9646757040810874, "grad_norm": 1.6162200636378943, "learning_rate": 6.531425152957238e-08, "loss": 0.4895, "step": 27077 }, { "epoch": 0.9647113312075815, "grad_norm": 1.2707445243079436, "learning_rate": 6.51826465144978e-08, "loss": 0.4136, "step": 27078 }, { "epoch": 0.9647469583340755, "grad_norm": 1.0939771915514769, "learning_rate": 6.505117378846071e-08, "loss": 0.3827, "step": 27079 }, { "epoch": 0.9647825854605697, "grad_norm": 1.9371128262911141, "learning_rate": 6.491983335320973e-08, "loss": 0.4495, "step": 27080 }, { "epoch": 0.9648182125870638, "grad_norm": 1.4001799162881854, "learning_rate": 6.478862521049678e-08, "loss": 0.4567, "step": 27081 }, { "epoch": 0.9648538397135579, "grad_norm": 1.5370963104396989, "learning_rate": 6.465754936206715e-08, "loss": 0.4947, "step": 27082 }, { "epoch": 0.9648894668400521, "grad_norm": 1.4877854311442102, "learning_rate": 6.452660580966608e-08, "loss": 0.407, "step": 27083 }, { "epoch": 0.9649250939665461, "grad_norm": 1.7287654532490138, "learning_rate": 6.439579455503886e-08, "loss": 0.5019, "step": 27084 }, { "epoch": 0.9649607210930402, "grad_norm": 1.6384309739895886, "learning_rate": 6.426511559992521e-08, "loss": 0.5059, "step": 27085 }, { "epoch": 0.9649963482195344, "grad_norm": 1.1553125444596455, "learning_rate": 6.413456894606595e-08, "loss": 0.396, "step": 27086 }, { "epoch": 0.9650319753460285, "grad_norm": 2.0038566906123574, "learning_rate": 6.400415459520081e-08, "loss": 0.4224, "step": 27087 }, { "epoch": 0.9650676024725225, "grad_norm": 1.5910548407264855, "learning_rate": 6.387387254906397e-08, "loss": 0.428, "step": 27088 }, { "epoch": 0.9651032295990167, "grad_norm": 1.2026601061448192, "learning_rate": 6.37437228093929e-08, "loss": 0.3922, "step": 27089 }, { "epoch": 0.9651388567255108, "grad_norm": 1.4292727498036202, "learning_rate": 6.361370537791845e-08, "loss": 0.501, "step": 27090 }, { "epoch": 0.9651744838520049, "grad_norm": 1.3588161852378853, "learning_rate": 6.348382025637256e-08, "loss": 0.4008, "step": 27091 }, { "epoch": 0.9652101109784991, "grad_norm": 1.340711179570387, "learning_rate": 6.335406744648498e-08, "loss": 0.3982, "step": 27092 }, { "epoch": 0.9652457381049931, "grad_norm": 1.3168933425434015, "learning_rate": 6.32244469499832e-08, "loss": 0.353, "step": 27093 }, { "epoch": 0.9652813652314872, "grad_norm": 1.3776507233896051, "learning_rate": 6.309495876859362e-08, "loss": 0.4205, "step": 27094 }, { "epoch": 0.9653169923579814, "grad_norm": 1.671243140645493, "learning_rate": 6.29656029040393e-08, "loss": 0.4987, "step": 27095 }, { "epoch": 0.9653526194844755, "grad_norm": 1.4191408694472216, "learning_rate": 6.283637935804443e-08, "loss": 0.3839, "step": 27096 }, { "epoch": 0.9653882466109696, "grad_norm": 1.9243855950044793, "learning_rate": 6.270728813232873e-08, "loss": 0.573, "step": 27097 }, { "epoch": 0.9654238737374637, "grad_norm": 1.216672449146951, "learning_rate": 6.257832922861084e-08, "loss": 0.421, "step": 27098 }, { "epoch": 0.9654595008639578, "grad_norm": 1.3178584887861748, "learning_rate": 6.244950264860828e-08, "loss": 0.3994, "step": 27099 }, { "epoch": 0.9654951279904519, "grad_norm": 1.6769982495690374, "learning_rate": 6.232080839403631e-08, "loss": 0.4713, "step": 27100 }, { "epoch": 0.9655307551169461, "grad_norm": 1.8572098473719472, "learning_rate": 6.219224646660915e-08, "loss": 0.4886, "step": 27101 }, { "epoch": 0.9655663822434402, "grad_norm": 1.4018413139309196, "learning_rate": 6.206381686803876e-08, "loss": 0.4533, "step": 27102 }, { "epoch": 0.9656020093699342, "grad_norm": 1.3571484110584358, "learning_rate": 6.193551960003374e-08, "loss": 0.4055, "step": 27103 }, { "epoch": 0.9656376364964284, "grad_norm": 1.5593166276561967, "learning_rate": 6.180735466430498e-08, "loss": 0.4587, "step": 27104 }, { "epoch": 0.9656732636229225, "grad_norm": 1.2835221441876414, "learning_rate": 6.167932206255666e-08, "loss": 0.4379, "step": 27105 }, { "epoch": 0.9657088907494166, "grad_norm": 1.3791241014185198, "learning_rate": 6.15514217964952e-08, "loss": 0.401, "step": 27106 }, { "epoch": 0.9657445178759108, "grad_norm": 1.841123179945945, "learning_rate": 6.142365386782257e-08, "loss": 0.4068, "step": 27107 }, { "epoch": 0.9657801450024048, "grad_norm": 1.2936061673012595, "learning_rate": 6.129601827824183e-08, "loss": 0.4084, "step": 27108 }, { "epoch": 0.9658157721288989, "grad_norm": 1.2994271983243408, "learning_rate": 6.116851502945054e-08, "loss": 0.3908, "step": 27109 }, { "epoch": 0.9658513992553931, "grad_norm": 1.378079961039467, "learning_rate": 6.10411441231462e-08, "loss": 0.4393, "step": 27110 }, { "epoch": 0.9658870263818872, "grad_norm": 1.6177205414403883, "learning_rate": 6.091390556102749e-08, "loss": 0.4739, "step": 27111 }, { "epoch": 0.9659226535083812, "grad_norm": 1.6542947851094973, "learning_rate": 6.078679934478637e-08, "loss": 0.5469, "step": 27112 }, { "epoch": 0.9659582806348754, "grad_norm": 1.2924846948606552, "learning_rate": 6.065982547611704e-08, "loss": 0.3858, "step": 27113 }, { "epoch": 0.9659939077613695, "grad_norm": 1.3386964737842926, "learning_rate": 6.053298395670814e-08, "loss": 0.3922, "step": 27114 }, { "epoch": 0.9660295348878636, "grad_norm": 1.7670920278658961, "learning_rate": 6.040627478825056e-08, "loss": 0.4684, "step": 27115 }, { "epoch": 0.9660651620143578, "grad_norm": 1.5176246139576053, "learning_rate": 6.027969797243072e-08, "loss": 0.5141, "step": 27116 }, { "epoch": 0.9661007891408518, "grad_norm": 1.7258887357308674, "learning_rate": 6.015325351093393e-08, "loss": 0.4682, "step": 27117 }, { "epoch": 0.9661364162673459, "grad_norm": 1.2844889663592804, "learning_rate": 6.00269414054433e-08, "loss": 0.3722, "step": 27118 }, { "epoch": 0.9661720433938401, "grad_norm": 1.7269695560184064, "learning_rate": 5.990076165764302e-08, "loss": 0.5228, "step": 27119 }, { "epoch": 0.9662076705203342, "grad_norm": 1.5818187822603924, "learning_rate": 5.977471426921067e-08, "loss": 0.4931, "step": 27120 }, { "epoch": 0.9662432976468283, "grad_norm": 1.4572931722202134, "learning_rate": 5.964879924182709e-08, "loss": 0.3772, "step": 27121 }, { "epoch": 0.9662789247733224, "grad_norm": 1.313240298136588, "learning_rate": 5.9523016577166525e-08, "loss": 0.412, "step": 27122 }, { "epoch": 0.9663145518998165, "grad_norm": 1.6026440074294304, "learning_rate": 5.93973662769054e-08, "loss": 0.4535, "step": 27123 }, { "epoch": 0.9663501790263106, "grad_norm": 1.645624263250833, "learning_rate": 5.927184834271571e-08, "loss": 0.4622, "step": 27124 }, { "epoch": 0.9663858061528048, "grad_norm": 2.088704319580813, "learning_rate": 5.9146462776270565e-08, "loss": 0.4646, "step": 27125 }, { "epoch": 0.9664214332792989, "grad_norm": 1.9034217854145332, "learning_rate": 5.9021209579238624e-08, "loss": 0.5057, "step": 27126 }, { "epoch": 0.9664570604057929, "grad_norm": 1.342646158295132, "learning_rate": 5.889608875328634e-08, "loss": 0.4106, "step": 27127 }, { "epoch": 0.9664926875322871, "grad_norm": 1.8773805458011377, "learning_rate": 5.8771100300082375e-08, "loss": 0.5579, "step": 27128 }, { "epoch": 0.9665283146587812, "grad_norm": 1.4620613151066537, "learning_rate": 5.864624422128873e-08, "loss": 0.3994, "step": 27129 }, { "epoch": 0.9665639417852753, "grad_norm": 1.5940508391771064, "learning_rate": 5.852152051856963e-08, "loss": 0.5115, "step": 27130 }, { "epoch": 0.9665995689117695, "grad_norm": 1.198502182079017, "learning_rate": 5.8396929193584864e-08, "loss": 0.4255, "step": 27131 }, { "epoch": 0.9666351960382635, "grad_norm": 1.152788390553018, "learning_rate": 5.8272470247994205e-08, "loss": 0.3312, "step": 27132 }, { "epoch": 0.9666708231647576, "grad_norm": 1.5915711794962244, "learning_rate": 5.814814368345412e-08, "loss": 0.5268, "step": 27133 }, { "epoch": 0.9667064502912518, "grad_norm": 1.4207150129441921, "learning_rate": 5.802394950161994e-08, "loss": 0.4022, "step": 27134 }, { "epoch": 0.9667420774177459, "grad_norm": 1.1890270490961126, "learning_rate": 5.7899887704145897e-08, "loss": 0.3842, "step": 27135 }, { "epoch": 0.96677770454424, "grad_norm": 1.2966307513286826, "learning_rate": 5.777595829268512e-08, "loss": 0.4132, "step": 27136 }, { "epoch": 0.9668133316707341, "grad_norm": 1.4612911912467423, "learning_rate": 5.765216126888629e-08, "loss": 0.541, "step": 27137 }, { "epoch": 0.9668489587972282, "grad_norm": 1.6359964667133517, "learning_rate": 5.752849663439697e-08, "loss": 0.4793, "step": 27138 }, { "epoch": 0.9668845859237223, "grad_norm": 1.6123321093351637, "learning_rate": 5.740496439086585e-08, "loss": 0.5378, "step": 27139 }, { "epoch": 0.9669202130502165, "grad_norm": 1.4036382768887787, "learning_rate": 5.7281564539937165e-08, "loss": 0.5197, "step": 27140 }, { "epoch": 0.9669558401767105, "grad_norm": 1.2029564580386558, "learning_rate": 5.715829708325404e-08, "loss": 0.3726, "step": 27141 }, { "epoch": 0.9669914673032046, "grad_norm": 1.2185698439022432, "learning_rate": 5.703516202245851e-08, "loss": 0.3771, "step": 27142 }, { "epoch": 0.9670270944296988, "grad_norm": 1.6008706283105105, "learning_rate": 5.691215935918815e-08, "loss": 0.5047, "step": 27143 }, { "epoch": 0.9670627215561929, "grad_norm": 1.3899758697451556, "learning_rate": 5.6789289095082746e-08, "loss": 0.4408, "step": 27144 }, { "epoch": 0.967098348682687, "grad_norm": 1.2177120271532793, "learning_rate": 5.666655123177767e-08, "loss": 0.3917, "step": 27145 }, { "epoch": 0.9671339758091811, "grad_norm": 1.2149276252094372, "learning_rate": 5.654394577090827e-08, "loss": 0.3551, "step": 27146 }, { "epoch": 0.9671696029356752, "grad_norm": 1.3905576428571913, "learning_rate": 5.642147271410548e-08, "loss": 0.4698, "step": 27147 }, { "epoch": 0.9672052300621693, "grad_norm": 1.3578387222505406, "learning_rate": 5.629913206300131e-08, "loss": 0.4314, "step": 27148 }, { "epoch": 0.9672408571886635, "grad_norm": 1.29470354597543, "learning_rate": 5.617692381922446e-08, "loss": 0.3995, "step": 27149 }, { "epoch": 0.9672764843151576, "grad_norm": 1.4609636040526266, "learning_rate": 5.605484798440253e-08, "loss": 0.4862, "step": 27150 }, { "epoch": 0.9673121114416516, "grad_norm": 1.5764767720930402, "learning_rate": 5.593290456015976e-08, "loss": 0.5262, "step": 27151 }, { "epoch": 0.9673477385681458, "grad_norm": 1.7078807246566838, "learning_rate": 5.5811093548121534e-08, "loss": 0.5296, "step": 27152 }, { "epoch": 0.9673833656946399, "grad_norm": 1.2964276442279814, "learning_rate": 5.568941494990876e-08, "loss": 0.382, "step": 27153 }, { "epoch": 0.967418992821134, "grad_norm": 1.5076561640091133, "learning_rate": 5.556786876714238e-08, "loss": 0.5921, "step": 27154 }, { "epoch": 0.9674546199476282, "grad_norm": 1.696645176378656, "learning_rate": 5.544645500143997e-08, "loss": 0.4796, "step": 27155 }, { "epoch": 0.9674902470741222, "grad_norm": 2.1464782985851447, "learning_rate": 5.532517365441914e-08, "loss": 0.5116, "step": 27156 }, { "epoch": 0.9675258742006163, "grad_norm": 1.3995987547498077, "learning_rate": 5.520402472769526e-08, "loss": 0.4317, "step": 27157 }, { "epoch": 0.9675615013271105, "grad_norm": 1.5534315654622357, "learning_rate": 5.5083008222880375e-08, "loss": 0.5447, "step": 27158 }, { "epoch": 0.9675971284536046, "grad_norm": 1.4947194120536582, "learning_rate": 5.4962124141586526e-08, "loss": 0.4883, "step": 27159 }, { "epoch": 0.9676327555800986, "grad_norm": 1.2157873981071263, "learning_rate": 5.4841372485422426e-08, "loss": 0.4116, "step": 27160 }, { "epoch": 0.9676683827065928, "grad_norm": 2.039320633992893, "learning_rate": 5.47207532559968e-08, "loss": 0.6189, "step": 27161 }, { "epoch": 0.9677040098330869, "grad_norm": 1.2786902361247692, "learning_rate": 5.4600266454916115e-08, "loss": 0.525, "step": 27162 }, { "epoch": 0.967739636959581, "grad_norm": 1.2402020166658065, "learning_rate": 5.4479912083783556e-08, "loss": 0.3995, "step": 27163 }, { "epoch": 0.9677752640860752, "grad_norm": 1.6195527025192094, "learning_rate": 5.4359690144203394e-08, "loss": 0.436, "step": 27164 }, { "epoch": 0.9678108912125692, "grad_norm": 1.2863458667188028, "learning_rate": 5.423960063777545e-08, "loss": 0.3959, "step": 27165 }, { "epoch": 0.9678465183390633, "grad_norm": 1.4252053767290536, "learning_rate": 5.411964356609844e-08, "loss": 0.4461, "step": 27166 }, { "epoch": 0.9678821454655575, "grad_norm": 1.203508676480049, "learning_rate": 5.399981893077111e-08, "loss": 0.3936, "step": 27167 }, { "epoch": 0.9679177725920516, "grad_norm": 1.6162934019841502, "learning_rate": 5.388012673338661e-08, "loss": 0.4979, "step": 27168 }, { "epoch": 0.9679533997185457, "grad_norm": 1.351916006495632, "learning_rate": 5.376056697554033e-08, "loss": 0.3773, "step": 27169 }, { "epoch": 0.9679890268450398, "grad_norm": 1.195399829040217, "learning_rate": 5.364113965882545e-08, "loss": 0.4005, "step": 27170 }, { "epoch": 0.9680246539715339, "grad_norm": 1.273190071024918, "learning_rate": 5.352184478482958e-08, "loss": 0.4317, "step": 27171 }, { "epoch": 0.968060281098028, "grad_norm": 1.2763449722513953, "learning_rate": 5.340268235514257e-08, "loss": 0.4199, "step": 27172 }, { "epoch": 0.9680959082245222, "grad_norm": 1.3067550765440585, "learning_rate": 5.3283652371352026e-08, "loss": 0.3753, "step": 27173 }, { "epoch": 0.9681315353510163, "grad_norm": 1.6945442100563526, "learning_rate": 5.316475483504113e-08, "loss": 0.5516, "step": 27174 }, { "epoch": 0.9681671624775103, "grad_norm": 1.3908968425877983, "learning_rate": 5.304598974779418e-08, "loss": 0.4203, "step": 27175 }, { "epoch": 0.9682027896040045, "grad_norm": 1.2273652861566093, "learning_rate": 5.2927357111192125e-08, "loss": 0.4148, "step": 27176 }, { "epoch": 0.9682384167304986, "grad_norm": 1.3973623855012594, "learning_rate": 5.280885692681592e-08, "loss": 0.4267, "step": 27177 }, { "epoch": 0.9682740438569927, "grad_norm": 1.3116552550704232, "learning_rate": 5.2690489196242086e-08, "loss": 0.3519, "step": 27178 }, { "epoch": 0.9683096709834869, "grad_norm": 1.2333409832507716, "learning_rate": 5.257225392104603e-08, "loss": 0.3956, "step": 27179 }, { "epoch": 0.9683452981099809, "grad_norm": 1.9399062438885672, "learning_rate": 5.2454151102804276e-08, "loss": 0.4991, "step": 27180 }, { "epoch": 0.968380925236475, "grad_norm": 1.4446714490732528, "learning_rate": 5.233618074308888e-08, "loss": 0.5313, "step": 27181 }, { "epoch": 0.9684165523629692, "grad_norm": 1.4623996928646974, "learning_rate": 5.221834284346861e-08, "loss": 0.3746, "step": 27182 }, { "epoch": 0.9684521794894633, "grad_norm": 1.3094980530250144, "learning_rate": 5.210063740551552e-08, "loss": 0.4096, "step": 27183 }, { "epoch": 0.9684878066159573, "grad_norm": 1.1855202778276255, "learning_rate": 5.198306443079504e-08, "loss": 0.3956, "step": 27184 }, { "epoch": 0.9685234337424515, "grad_norm": 1.4889489280574397, "learning_rate": 5.1865623920873685e-08, "loss": 0.4331, "step": 27185 }, { "epoch": 0.9685590608689456, "grad_norm": 1.5736820304216956, "learning_rate": 5.174831587731466e-08, "loss": 0.5168, "step": 27186 }, { "epoch": 0.9685946879954397, "grad_norm": 1.8578471119710656, "learning_rate": 5.163114030168004e-08, "loss": 0.4678, "step": 27187 }, { "epoch": 0.9686303151219339, "grad_norm": 1.3338693381797153, "learning_rate": 5.15140971955308e-08, "loss": 0.4976, "step": 27188 }, { "epoch": 0.968665942248428, "grad_norm": 1.2404685568121323, "learning_rate": 5.1397186560424586e-08, "loss": 0.3883, "step": 27189 }, { "epoch": 0.968701569374922, "grad_norm": 1.2450278181996386, "learning_rate": 5.1280408397919036e-08, "loss": 0.4149, "step": 27190 }, { "epoch": 0.9687371965014162, "grad_norm": 1.245952126638905, "learning_rate": 5.116376270956735e-08, "loss": 0.3934, "step": 27191 }, { "epoch": 0.9687728236279103, "grad_norm": 1.30801036919938, "learning_rate": 5.1047249496924965e-08, "loss": 0.4067, "step": 27192 }, { "epoch": 0.9688084507544044, "grad_norm": 1.7653927325627456, "learning_rate": 5.093086876154174e-08, "loss": 0.4508, "step": 27193 }, { "epoch": 0.9688440778808985, "grad_norm": 1.4037458582111098, "learning_rate": 5.081462050496866e-08, "loss": 0.4971, "step": 27194 }, { "epoch": 0.9688797050073926, "grad_norm": 1.2059386216678065, "learning_rate": 5.0698504728752264e-08, "loss": 0.4363, "step": 27195 }, { "epoch": 0.9689153321338867, "grad_norm": 1.3648535545864968, "learning_rate": 5.05825214344402e-08, "loss": 0.4346, "step": 27196 }, { "epoch": 0.9689509592603809, "grad_norm": 1.6254847744182177, "learning_rate": 5.046667062357569e-08, "loss": 0.5749, "step": 27197 }, { "epoch": 0.968986586386875, "grad_norm": 1.1735354662621105, "learning_rate": 5.035095229770193e-08, "loss": 0.3684, "step": 27198 }, { "epoch": 0.969022213513369, "grad_norm": 1.7293537507180763, "learning_rate": 5.02353664583588e-08, "loss": 0.5503, "step": 27199 }, { "epoch": 0.9690578406398632, "grad_norm": 1.3098006934062605, "learning_rate": 5.01199131070873e-08, "loss": 0.3926, "step": 27200 }, { "epoch": 0.9690934677663573, "grad_norm": 2.1494259786119643, "learning_rate": 5.0004592245422866e-08, "loss": 0.4681, "step": 27201 }, { "epoch": 0.9691290948928514, "grad_norm": 1.5113447284665507, "learning_rate": 4.9889403874903154e-08, "loss": 0.4768, "step": 27202 }, { "epoch": 0.9691647220193456, "grad_norm": 1.5802899055146165, "learning_rate": 4.977434799705916e-08, "loss": 0.5835, "step": 27203 }, { "epoch": 0.9692003491458396, "grad_norm": 1.320017934893911, "learning_rate": 4.965942461342521e-08, "loss": 0.4116, "step": 27204 }, { "epoch": 0.9692359762723337, "grad_norm": 1.317451380668911, "learning_rate": 4.9544633725531196e-08, "loss": 0.4035, "step": 27205 }, { "epoch": 0.9692716033988279, "grad_norm": 1.342759061982074, "learning_rate": 4.94299753349059e-08, "loss": 0.3665, "step": 27206 }, { "epoch": 0.969307230525322, "grad_norm": 1.8440043126703967, "learning_rate": 4.931544944307476e-08, "loss": 0.4619, "step": 27207 }, { "epoch": 0.969342857651816, "grad_norm": 1.1941419578118138, "learning_rate": 4.9201056051563226e-08, "loss": 0.3653, "step": 27208 }, { "epoch": 0.9693784847783102, "grad_norm": 1.6198610588188467, "learning_rate": 4.908679516189452e-08, "loss": 0.4369, "step": 27209 }, { "epoch": 0.9694141119048043, "grad_norm": 1.4098471757997235, "learning_rate": 4.897266677559187e-08, "loss": 0.542, "step": 27210 }, { "epoch": 0.9694497390312984, "grad_norm": 1.2128432564200626, "learning_rate": 4.885867089417295e-08, "loss": 0.4401, "step": 27211 }, { "epoch": 0.9694853661577926, "grad_norm": 1.5363800714140832, "learning_rate": 4.8744807519155446e-08, "loss": 0.4452, "step": 27212 }, { "epoch": 0.9695209932842866, "grad_norm": 1.3405109267397801, "learning_rate": 4.863107665205702e-08, "loss": 0.4658, "step": 27213 }, { "epoch": 0.9695566204107807, "grad_norm": 1.9419230659854543, "learning_rate": 4.851747829439091e-08, "loss": 0.5259, "step": 27214 }, { "epoch": 0.9695922475372749, "grad_norm": 1.3415096764831296, "learning_rate": 4.8404012447670344e-08, "loss": 0.4952, "step": 27215 }, { "epoch": 0.969627874663769, "grad_norm": 1.1221796244762894, "learning_rate": 4.829067911340635e-08, "loss": 0.4043, "step": 27216 }, { "epoch": 0.9696635017902631, "grad_norm": 1.6193344015860842, "learning_rate": 4.8177478293107707e-08, "loss": 0.5342, "step": 27217 }, { "epoch": 0.9696991289167572, "grad_norm": 1.3895851515762565, "learning_rate": 4.8064409988282104e-08, "loss": 0.4367, "step": 27218 }, { "epoch": 0.9697347560432513, "grad_norm": 1.2170098152158897, "learning_rate": 4.7951474200435e-08, "loss": 0.4372, "step": 27219 }, { "epoch": 0.9697703831697454, "grad_norm": 1.4271008251845776, "learning_rate": 4.7838670931069643e-08, "loss": 0.424, "step": 27220 }, { "epoch": 0.9698060102962396, "grad_norm": 1.3564003367571131, "learning_rate": 4.772600018168816e-08, "loss": 0.523, "step": 27221 }, { "epoch": 0.9698416374227337, "grad_norm": 1.3435107845016643, "learning_rate": 4.761346195379157e-08, "loss": 0.394, "step": 27222 }, { "epoch": 0.9698772645492277, "grad_norm": 2.2142239146708036, "learning_rate": 4.750105624887868e-08, "loss": 0.5539, "step": 27223 }, { "epoch": 0.9699128916757219, "grad_norm": 1.5888361254678651, "learning_rate": 4.738878306844496e-08, "loss": 0.3908, "step": 27224 }, { "epoch": 0.969948518802216, "grad_norm": 1.6270868633451854, "learning_rate": 4.727664241398588e-08, "loss": 0.5371, "step": 27225 }, { "epoch": 0.9699841459287101, "grad_norm": 1.1574788559422269, "learning_rate": 4.7164634286994696e-08, "loss": 0.3848, "step": 27226 }, { "epoch": 0.9700197730552043, "grad_norm": 1.3534642201425207, "learning_rate": 4.705275868896464e-08, "loss": 0.4197, "step": 27227 }, { "epoch": 0.9700554001816983, "grad_norm": 1.9863141944513796, "learning_rate": 4.6941015621382316e-08, "loss": 0.4032, "step": 27228 }, { "epoch": 0.9700910273081924, "grad_norm": 1.3516964263386277, "learning_rate": 4.6829405085736525e-08, "loss": 0.4054, "step": 27229 }, { "epoch": 0.9701266544346866, "grad_norm": 1.4890555967065864, "learning_rate": 4.671792708351497e-08, "loss": 0.5008, "step": 27230 }, { "epoch": 0.9701622815611807, "grad_norm": 1.3330536660812795, "learning_rate": 4.660658161620202e-08, "loss": 0.4255, "step": 27231 }, { "epoch": 0.9701979086876747, "grad_norm": 1.422892477047243, "learning_rate": 4.649536868527871e-08, "loss": 0.3985, "step": 27232 }, { "epoch": 0.9702335358141689, "grad_norm": 1.442895670459743, "learning_rate": 4.6384288292226075e-08, "loss": 0.4858, "step": 27233 }, { "epoch": 0.970269162940663, "grad_norm": 1.7388847182373284, "learning_rate": 4.627334043852405e-08, "loss": 0.4803, "step": 27234 }, { "epoch": 0.9703047900671571, "grad_norm": 1.664073095118221, "learning_rate": 4.6162525125650335e-08, "loss": 0.4235, "step": 27235 }, { "epoch": 0.9703404171936513, "grad_norm": 1.3575263791866135, "learning_rate": 4.605184235507931e-08, "loss": 0.4349, "step": 27236 }, { "epoch": 0.9703760443201453, "grad_norm": 1.3339204906644793, "learning_rate": 4.594129212828535e-08, "loss": 0.4182, "step": 27237 }, { "epoch": 0.9704116714466394, "grad_norm": 1.4190133127477065, "learning_rate": 4.5830874446740616e-08, "loss": 0.4503, "step": 27238 }, { "epoch": 0.9704472985731336, "grad_norm": 1.5233985174318652, "learning_rate": 4.572058931191614e-08, "loss": 0.4552, "step": 27239 }, { "epoch": 0.9704829256996277, "grad_norm": 1.6402132364678923, "learning_rate": 4.561043672527854e-08, "loss": 0.5626, "step": 27240 }, { "epoch": 0.9705185528261218, "grad_norm": 1.2995068647242798, "learning_rate": 4.5500416688296634e-08, "loss": 0.3897, "step": 27241 }, { "epoch": 0.970554179952616, "grad_norm": 2.144899834251319, "learning_rate": 4.53905292024337e-08, "loss": 0.4833, "step": 27242 }, { "epoch": 0.97058980707911, "grad_norm": 1.2438281929793866, "learning_rate": 4.528077426915412e-08, "loss": 0.4003, "step": 27243 }, { "epoch": 0.9706254342056041, "grad_norm": 1.5262578061310577, "learning_rate": 4.517115188991894e-08, "loss": 0.4788, "step": 27244 }, { "epoch": 0.9706610613320983, "grad_norm": 1.5749369212257078, "learning_rate": 4.5061662066187004e-08, "loss": 0.4252, "step": 27245 }, { "epoch": 0.9706966884585924, "grad_norm": 1.4006123868192362, "learning_rate": 4.4952304799418256e-08, "loss": 0.4651, "step": 27246 }, { "epoch": 0.9707323155850864, "grad_norm": 1.2898664838679526, "learning_rate": 4.484308009106819e-08, "loss": 0.4396, "step": 27247 }, { "epoch": 0.9707679427115806, "grad_norm": 1.6616365765526213, "learning_rate": 4.47339879425901e-08, "loss": 0.4751, "step": 27248 }, { "epoch": 0.9708035698380747, "grad_norm": 1.4507952827665667, "learning_rate": 4.462502835543614e-08, "loss": 0.4238, "step": 27249 }, { "epoch": 0.9708391969645688, "grad_norm": 1.4587276670341218, "learning_rate": 4.4516201331059606e-08, "loss": 0.456, "step": 27250 }, { "epoch": 0.970874824091063, "grad_norm": 1.299534877191517, "learning_rate": 4.440750687090822e-08, "loss": 0.384, "step": 27251 }, { "epoch": 0.970910451217557, "grad_norm": 2.042743921436992, "learning_rate": 4.429894497642861e-08, "loss": 0.5968, "step": 27252 }, { "epoch": 0.9709460783440511, "grad_norm": 1.6950626794606731, "learning_rate": 4.419051564906851e-08, "loss": 0.5944, "step": 27253 }, { "epoch": 0.9709817054705453, "grad_norm": 1.4036610950379989, "learning_rate": 4.408221889026898e-08, "loss": 0.4625, "step": 27254 }, { "epoch": 0.9710173325970394, "grad_norm": 3.1040840630924604, "learning_rate": 4.397405470147442e-08, "loss": 0.4985, "step": 27255 }, { "epoch": 0.9710529597235334, "grad_norm": 1.4500103975560006, "learning_rate": 4.386602308412369e-08, "loss": 0.4946, "step": 27256 }, { "epoch": 0.9710885868500276, "grad_norm": 1.2727105032420676, "learning_rate": 4.375812403965673e-08, "loss": 0.4263, "step": 27257 }, { "epoch": 0.9711242139765217, "grad_norm": 1.4435528445226318, "learning_rate": 4.365035756950797e-08, "loss": 0.5331, "step": 27258 }, { "epoch": 0.9711598411030158, "grad_norm": 1.2804366061831514, "learning_rate": 4.354272367511514e-08, "loss": 0.3739, "step": 27259 }, { "epoch": 0.97119546822951, "grad_norm": 1.230329898823187, "learning_rate": 4.34352223579082e-08, "loss": 0.3898, "step": 27260 }, { "epoch": 0.971231095356004, "grad_norm": 1.5498870213410452, "learning_rate": 4.332785361932157e-08, "loss": 0.5488, "step": 27261 }, { "epoch": 0.9712667224824981, "grad_norm": 1.7364349759081412, "learning_rate": 4.322061746078521e-08, "loss": 0.5433, "step": 27262 }, { "epoch": 0.9713023496089923, "grad_norm": 1.3673556848791757, "learning_rate": 4.3113513883723534e-08, "loss": 0.3927, "step": 27263 }, { "epoch": 0.9713379767354864, "grad_norm": 2.0354136577956923, "learning_rate": 4.300654288956652e-08, "loss": 0.5023, "step": 27264 }, { "epoch": 0.9713736038619805, "grad_norm": 1.24350446668421, "learning_rate": 4.2899704479736345e-08, "loss": 0.4092, "step": 27265 }, { "epoch": 0.9714092309884746, "grad_norm": 1.2563844205540076, "learning_rate": 4.279299865565523e-08, "loss": 0.3657, "step": 27266 }, { "epoch": 0.9714448581149687, "grad_norm": 1.3415392315322132, "learning_rate": 4.2686425418745345e-08, "loss": 0.407, "step": 27267 }, { "epoch": 0.9714804852414628, "grad_norm": 1.2049785128349444, "learning_rate": 4.257998477042669e-08, "loss": 0.391, "step": 27268 }, { "epoch": 0.971516112367957, "grad_norm": 1.1907131247996303, "learning_rate": 4.2473676712114776e-08, "loss": 0.4232, "step": 27269 }, { "epoch": 0.9715517394944511, "grad_norm": 1.3072919716140852, "learning_rate": 4.236750124522515e-08, "loss": 0.401, "step": 27270 }, { "epoch": 0.9715873666209451, "grad_norm": 1.2961524802742739, "learning_rate": 4.226145837117335e-08, "loss": 0.39, "step": 27271 }, { "epoch": 0.9716229937474393, "grad_norm": 1.4077782882454506, "learning_rate": 4.215554809136935e-08, "loss": 0.4458, "step": 27272 }, { "epoch": 0.9716586208739334, "grad_norm": 1.223110230307473, "learning_rate": 4.204977040722425e-08, "loss": 0.3811, "step": 27273 }, { "epoch": 0.9716942480004275, "grad_norm": 1.3374220453853012, "learning_rate": 4.1944125320146914e-08, "loss": 0.4065, "step": 27274 }, { "epoch": 0.9717298751269217, "grad_norm": 1.390376446104572, "learning_rate": 4.1838612831544e-08, "loss": 0.4882, "step": 27275 }, { "epoch": 0.9717655022534157, "grad_norm": 1.373940429922066, "learning_rate": 4.173323294281995e-08, "loss": 0.4516, "step": 27276 }, { "epoch": 0.9718011293799098, "grad_norm": 1.3761987619765434, "learning_rate": 4.1627985655378064e-08, "loss": 0.5121, "step": 27277 }, { "epoch": 0.971836756506404, "grad_norm": 1.4250054792138405, "learning_rate": 4.152287097062058e-08, "loss": 0.5134, "step": 27278 }, { "epoch": 0.9718723836328981, "grad_norm": 1.5350780101085113, "learning_rate": 4.1417888889947466e-08, "loss": 0.5298, "step": 27279 }, { "epoch": 0.9719080107593921, "grad_norm": 1.4982715605906938, "learning_rate": 4.1313039414754285e-08, "loss": 0.3694, "step": 27280 }, { "epoch": 0.9719436378858863, "grad_norm": 1.5142418732913687, "learning_rate": 4.1208322546438806e-08, "loss": 0.4806, "step": 27281 }, { "epoch": 0.9719792650123804, "grad_norm": 1.2937661682398216, "learning_rate": 4.110373828639547e-08, "loss": 0.3799, "step": 27282 }, { "epoch": 0.9720148921388745, "grad_norm": 1.6438830653003114, "learning_rate": 4.09992866360176e-08, "loss": 0.3409, "step": 27283 }, { "epoch": 0.9720505192653687, "grad_norm": 1.3584564641797388, "learning_rate": 4.089496759669409e-08, "loss": 0.3965, "step": 27284 }, { "epoch": 0.9720861463918627, "grad_norm": 1.3240480086826047, "learning_rate": 4.0790781169816055e-08, "loss": 0.4275, "step": 27285 }, { "epoch": 0.9721217735183568, "grad_norm": 1.5300219829210076, "learning_rate": 4.068672735676904e-08, "loss": 0.4213, "step": 27286 }, { "epoch": 0.972157400644851, "grad_norm": 1.2242996301163789, "learning_rate": 4.058280615893972e-08, "loss": 0.4434, "step": 27287 }, { "epoch": 0.9721930277713451, "grad_norm": 1.195468599044052, "learning_rate": 4.0479017577711447e-08, "loss": 0.3705, "step": 27288 }, { "epoch": 0.9722286548978392, "grad_norm": 1.2992392216986426, "learning_rate": 4.037536161446642e-08, "loss": 0.3713, "step": 27289 }, { "epoch": 0.9722642820243333, "grad_norm": 1.480659004660459, "learning_rate": 4.0271838270584675e-08, "loss": 0.5178, "step": 27290 }, { "epoch": 0.9722999091508274, "grad_norm": 1.251896989354454, "learning_rate": 4.016844754744509e-08, "loss": 0.342, "step": 27291 }, { "epoch": 0.9723355362773215, "grad_norm": 1.4020028110043377, "learning_rate": 4.0065189446424346e-08, "loss": 0.5231, "step": 27292 }, { "epoch": 0.9723711634038157, "grad_norm": 1.8802434861940593, "learning_rate": 3.99620639688969e-08, "loss": 0.5595, "step": 27293 }, { "epoch": 0.9724067905303098, "grad_norm": 1.7923708872357897, "learning_rate": 3.98590711162361e-08, "loss": 0.5748, "step": 27294 }, { "epoch": 0.9724424176568038, "grad_norm": 1.4805566376165407, "learning_rate": 3.975621088981308e-08, "loss": 0.5376, "step": 27295 }, { "epoch": 0.972478044783298, "grad_norm": 1.3806549477397578, "learning_rate": 3.965348329099894e-08, "loss": 0.4677, "step": 27296 }, { "epoch": 0.9725136719097921, "grad_norm": 1.2566863976818896, "learning_rate": 3.955088832115928e-08, "loss": 0.3867, "step": 27297 }, { "epoch": 0.9725492990362862, "grad_norm": 1.3588353943454778, "learning_rate": 3.9448425981661876e-08, "loss": 0.4101, "step": 27298 }, { "epoch": 0.9725849261627804, "grad_norm": 1.6127930962750148, "learning_rate": 3.934609627387121e-08, "loss": 0.5589, "step": 27299 }, { "epoch": 0.9726205532892744, "grad_norm": 1.3011489923238355, "learning_rate": 3.9243899199149504e-08, "loss": 0.3666, "step": 27300 }, { "epoch": 0.9726561804157685, "grad_norm": 1.81840250341714, "learning_rate": 3.91418347588568e-08, "loss": 0.4252, "step": 27301 }, { "epoch": 0.9726918075422627, "grad_norm": 1.5044902601038774, "learning_rate": 3.903990295435311e-08, "loss": 0.5169, "step": 27302 }, { "epoch": 0.9727274346687568, "grad_norm": 1.3367783352303182, "learning_rate": 3.8938103786995144e-08, "loss": 0.3852, "step": 27303 }, { "epoch": 0.9727630617952508, "grad_norm": 1.4141329780248442, "learning_rate": 3.883643725813846e-08, "loss": 0.3917, "step": 27304 }, { "epoch": 0.972798688921745, "grad_norm": 1.2207375400772342, "learning_rate": 3.8734903369138654e-08, "loss": 0.4116, "step": 27305 }, { "epoch": 0.9728343160482391, "grad_norm": 1.2075531487203226, "learning_rate": 3.8633502121344643e-08, "loss": 0.3673, "step": 27306 }, { "epoch": 0.9728699431747332, "grad_norm": 1.8126893159962911, "learning_rate": 3.853223351610757e-08, "loss": 0.4816, "step": 27307 }, { "epoch": 0.9729055703012274, "grad_norm": 1.405927615680957, "learning_rate": 3.8431097554778585e-08, "loss": 0.5089, "step": 27308 }, { "epoch": 0.9729411974277214, "grad_norm": 1.4142298918609397, "learning_rate": 3.8330094238699936e-08, "loss": 0.4288, "step": 27309 }, { "epoch": 0.9729768245542155, "grad_norm": 1.471172732140236, "learning_rate": 3.822922356922054e-08, "loss": 0.503, "step": 27310 }, { "epoch": 0.9730124516807097, "grad_norm": 1.6988397655698246, "learning_rate": 3.812848554768045e-08, "loss": 0.5005, "step": 27311 }, { "epoch": 0.9730480788072038, "grad_norm": 1.3319625993025033, "learning_rate": 3.802788017542414e-08, "loss": 0.3925, "step": 27312 }, { "epoch": 0.9730837059336979, "grad_norm": 1.350166986594984, "learning_rate": 3.792740745378831e-08, "loss": 0.4303, "step": 27313 }, { "epoch": 0.973119333060192, "grad_norm": 1.5565328859487826, "learning_rate": 3.7827067384113014e-08, "loss": 0.4716, "step": 27314 }, { "epoch": 0.9731549601866861, "grad_norm": 1.2610147223738535, "learning_rate": 3.772685996773273e-08, "loss": 0.4015, "step": 27315 }, { "epoch": 0.9731905873131803, "grad_norm": 1.222367833547283, "learning_rate": 3.7626785205983065e-08, "loss": 0.3772, "step": 27316 }, { "epoch": 0.9732262144396744, "grad_norm": 1.322723195852731, "learning_rate": 3.752684310019628e-08, "loss": 0.3646, "step": 27317 }, { "epoch": 0.9732618415661685, "grad_norm": 3.2308472569047555, "learning_rate": 3.7427033651702414e-08, "loss": 0.4725, "step": 27318 }, { "epoch": 0.9732974686926626, "grad_norm": 1.6291370198415196, "learning_rate": 3.732735686183042e-08, "loss": 0.5033, "step": 27319 }, { "epoch": 0.9733330958191567, "grad_norm": 1.2970123685373665, "learning_rate": 3.722781273190923e-08, "loss": 0.3571, "step": 27320 }, { "epoch": 0.9733687229456508, "grad_norm": 1.4925581767765823, "learning_rate": 3.7128401263262226e-08, "loss": 0.5205, "step": 27321 }, { "epoch": 0.973404350072145, "grad_norm": 1.310626845768912, "learning_rate": 3.702912245721502e-08, "loss": 0.4189, "step": 27322 }, { "epoch": 0.9734399771986391, "grad_norm": 1.381479278337211, "learning_rate": 3.6929976315088765e-08, "loss": 0.4198, "step": 27323 }, { "epoch": 0.9734756043251331, "grad_norm": 1.5812302465588268, "learning_rate": 3.683096283820353e-08, "loss": 0.5741, "step": 27324 }, { "epoch": 0.9735112314516273, "grad_norm": 1.2370884201009664, "learning_rate": 3.6732082027877145e-08, "loss": 0.4108, "step": 27325 }, { "epoch": 0.9735468585781214, "grad_norm": 1.393816326174732, "learning_rate": 3.663333388542745e-08, "loss": 0.4986, "step": 27326 }, { "epoch": 0.9735824857046155, "grad_norm": 1.5604800836521007, "learning_rate": 3.653471841217004e-08, "loss": 0.5194, "step": 27327 }, { "epoch": 0.9736181128311097, "grad_norm": 1.4591319497180386, "learning_rate": 3.643623560941611e-08, "loss": 0.4033, "step": 27328 }, { "epoch": 0.9736537399576037, "grad_norm": 1.4389984611747915, "learning_rate": 3.633788547847683e-08, "loss": 0.498, "step": 27329 }, { "epoch": 0.9736893670840978, "grad_norm": 1.5272547737473645, "learning_rate": 3.623966802066447e-08, "loss": 0.493, "step": 27330 }, { "epoch": 0.973724994210592, "grad_norm": 1.6375737312646652, "learning_rate": 3.6141583237283564e-08, "loss": 0.5004, "step": 27331 }, { "epoch": 0.9737606213370861, "grad_norm": 1.6102379055292813, "learning_rate": 3.6043631129643044e-08, "loss": 0.4397, "step": 27332 }, { "epoch": 0.9737962484635801, "grad_norm": 1.6770704954875693, "learning_rate": 3.5945811699045206e-08, "loss": 0.5166, "step": 27333 }, { "epoch": 0.9738318755900743, "grad_norm": 1.3193466753278484, "learning_rate": 3.584812494679346e-08, "loss": 0.422, "step": 27334 }, { "epoch": 0.9738675027165684, "grad_norm": 1.2333056259568391, "learning_rate": 3.5750570874188985e-08, "loss": 0.392, "step": 27335 }, { "epoch": 0.9739031298430625, "grad_norm": 2.372616539500196, "learning_rate": 3.565314948252962e-08, "loss": 0.4093, "step": 27336 }, { "epoch": 0.9739387569695567, "grad_norm": 1.4615112290070285, "learning_rate": 3.5555860773113235e-08, "loss": 0.5047, "step": 27337 }, { "epoch": 0.9739743840960507, "grad_norm": 1.548985848998838, "learning_rate": 3.545870474723545e-08, "loss": 0.4092, "step": 27338 }, { "epoch": 0.9740100112225448, "grad_norm": 1.2610668450957878, "learning_rate": 3.536168140618968e-08, "loss": 0.3623, "step": 27339 }, { "epoch": 0.974045638349039, "grad_norm": 1.144566428201631, "learning_rate": 3.526479075126821e-08, "loss": 0.3686, "step": 27340 }, { "epoch": 0.9740812654755331, "grad_norm": 1.5094615211436824, "learning_rate": 3.516803278376113e-08, "loss": 0.4685, "step": 27341 }, { "epoch": 0.9741168926020272, "grad_norm": 1.6577523662581946, "learning_rate": 3.50714075049563e-08, "loss": 0.5575, "step": 27342 }, { "epoch": 0.9741525197285213, "grad_norm": 1.2950014727533856, "learning_rate": 3.497491491614158e-08, "loss": 0.427, "step": 27343 }, { "epoch": 0.9741881468550154, "grad_norm": 1.954341798582663, "learning_rate": 3.487855501860038e-08, "loss": 0.4952, "step": 27344 }, { "epoch": 0.9742237739815095, "grad_norm": 1.2874643024668109, "learning_rate": 3.478232781361612e-08, "loss": 0.4607, "step": 27345 }, { "epoch": 0.9742594011080037, "grad_norm": 1.3357300963976335, "learning_rate": 3.468623330247112e-08, "loss": 0.3786, "step": 27346 }, { "epoch": 0.9742950282344978, "grad_norm": 1.3220077014186344, "learning_rate": 3.4590271486444336e-08, "loss": 0.3644, "step": 27347 }, { "epoch": 0.9743306553609918, "grad_norm": 1.757578485827876, "learning_rate": 3.449444236681254e-08, "loss": 0.5446, "step": 27348 }, { "epoch": 0.974366282487486, "grad_norm": 1.3526568190902797, "learning_rate": 3.4398745944853597e-08, "loss": 0.4298, "step": 27349 }, { "epoch": 0.9744019096139801, "grad_norm": 1.2307223049893612, "learning_rate": 3.430318222183981e-08, "loss": 0.5263, "step": 27350 }, { "epoch": 0.9744375367404742, "grad_norm": 1.4105901179705478, "learning_rate": 3.4207751199045736e-08, "loss": 0.4341, "step": 27351 }, { "epoch": 0.9744731638669684, "grad_norm": 1.551976525403658, "learning_rate": 3.411245287774034e-08, "loss": 0.5135, "step": 27352 }, { "epoch": 0.9745087909934624, "grad_norm": 2.0982979927578147, "learning_rate": 3.401728725919373e-08, "loss": 0.5051, "step": 27353 }, { "epoch": 0.9745444181199565, "grad_norm": 1.5291704538072395, "learning_rate": 3.392225434467378e-08, "loss": 0.5387, "step": 27354 }, { "epoch": 0.9745800452464507, "grad_norm": 1.3628294856550165, "learning_rate": 3.382735413544391e-08, "loss": 0.3982, "step": 27355 }, { "epoch": 0.9746156723729448, "grad_norm": 1.2349072089622426, "learning_rate": 3.373258663276868e-08, "loss": 0.4004, "step": 27356 }, { "epoch": 0.9746512994994389, "grad_norm": 2.38305692993343, "learning_rate": 3.363795183791041e-08, "loss": 0.4955, "step": 27357 }, { "epoch": 0.974686926625933, "grad_norm": 1.5862502804622804, "learning_rate": 3.35434497521292e-08, "loss": 0.5394, "step": 27358 }, { "epoch": 0.9747225537524271, "grad_norm": 1.6277024675463532, "learning_rate": 3.344908037668182e-08, "loss": 0.5074, "step": 27359 }, { "epoch": 0.9747581808789212, "grad_norm": 1.2797484543539954, "learning_rate": 3.335484371282727e-08, "loss": 0.3937, "step": 27360 }, { "epoch": 0.9747938080054154, "grad_norm": 1.3140537407149868, "learning_rate": 3.3260739761818986e-08, "loss": 0.3838, "step": 27361 }, { "epoch": 0.9748294351319094, "grad_norm": 1.44856401167264, "learning_rate": 3.316676852491041e-08, "loss": 0.3671, "step": 27362 }, { "epoch": 0.9748650622584035, "grad_norm": 1.653827162083323, "learning_rate": 3.3072930003351655e-08, "loss": 0.5521, "step": 27363 }, { "epoch": 0.9749006893848977, "grad_norm": 1.7763556638614886, "learning_rate": 3.2979224198395054e-08, "loss": 0.406, "step": 27364 }, { "epoch": 0.9749363165113918, "grad_norm": 1.2942318152510226, "learning_rate": 3.2885651111285166e-08, "loss": 0.4144, "step": 27365 }, { "epoch": 0.9749719436378859, "grad_norm": 1.3673548846238177, "learning_rate": 3.2792210743269883e-08, "loss": 0.3786, "step": 27366 }, { "epoch": 0.97500757076438, "grad_norm": 1.324583945265018, "learning_rate": 3.269890309559376e-08, "loss": 0.3933, "step": 27367 }, { "epoch": 0.9750431978908741, "grad_norm": 1.3202984754120435, "learning_rate": 3.260572816949692e-08, "loss": 0.3658, "step": 27368 }, { "epoch": 0.9750788250173682, "grad_norm": 1.6332575087097871, "learning_rate": 3.2512685966222815e-08, "loss": 0.5279, "step": 27369 }, { "epoch": 0.9751144521438624, "grad_norm": 1.6167697163981625, "learning_rate": 3.241977648700934e-08, "loss": 0.4938, "step": 27370 }, { "epoch": 0.9751500792703565, "grad_norm": 1.2392671012302512, "learning_rate": 3.2326999733092165e-08, "loss": 0.3877, "step": 27371 }, { "epoch": 0.9751857063968505, "grad_norm": 1.2131343378034305, "learning_rate": 3.2234355705709206e-08, "loss": 0.4024, "step": 27372 }, { "epoch": 0.9752213335233447, "grad_norm": 1.1784407747624286, "learning_rate": 3.2141844406091696e-08, "loss": 0.401, "step": 27373 }, { "epoch": 0.9752569606498388, "grad_norm": 1.2618924523026238, "learning_rate": 3.2049465835474194e-08, "loss": 0.4284, "step": 27374 }, { "epoch": 0.9752925877763329, "grad_norm": 1.2802680757676932, "learning_rate": 3.195721999508461e-08, "loss": 0.4029, "step": 27375 }, { "epoch": 0.9753282149028271, "grad_norm": 1.6696029531642638, "learning_rate": 3.1865106886150856e-08, "loss": 0.452, "step": 27376 }, { "epoch": 0.9753638420293211, "grad_norm": 1.1891279617849089, "learning_rate": 3.1773126509900834e-08, "loss": 0.4277, "step": 27377 }, { "epoch": 0.9753994691558152, "grad_norm": 1.508386501148512, "learning_rate": 3.168127886756023e-08, "loss": 0.5012, "step": 27378 }, { "epoch": 0.9754350962823094, "grad_norm": 1.6882002023519878, "learning_rate": 3.158956396034918e-08, "loss": 0.497, "step": 27379 }, { "epoch": 0.9754707234088035, "grad_norm": 1.3749409925732474, "learning_rate": 3.149798178949226e-08, "loss": 0.3831, "step": 27380 }, { "epoch": 0.9755063505352976, "grad_norm": 1.9662549487148713, "learning_rate": 3.1406532356206274e-08, "loss": 0.5326, "step": 27381 }, { "epoch": 0.9755419776617917, "grad_norm": 1.515008964413732, "learning_rate": 3.131521566171025e-08, "loss": 0.4809, "step": 27382 }, { "epoch": 0.9755776047882858, "grad_norm": 1.659710844933309, "learning_rate": 3.1224031707219885e-08, "loss": 0.5635, "step": 27383 }, { "epoch": 0.9756132319147799, "grad_norm": 1.5251205510619266, "learning_rate": 3.113298049394975e-08, "loss": 0.5209, "step": 27384 }, { "epoch": 0.9756488590412741, "grad_norm": 1.2840052886386746, "learning_rate": 3.104206202311111e-08, "loss": 0.3905, "step": 27385 }, { "epoch": 0.9756844861677682, "grad_norm": 1.8546104510870227, "learning_rate": 3.0951276295915214e-08, "loss": 0.4882, "step": 27386 }, { "epoch": 0.9757201132942622, "grad_norm": 1.4541019844783598, "learning_rate": 3.086062331357109e-08, "loss": 0.5097, "step": 27387 }, { "epoch": 0.9757557404207564, "grad_norm": 1.1830921565951622, "learning_rate": 3.077010307728556e-08, "loss": 0.3839, "step": 27388 }, { "epoch": 0.9757913675472505, "grad_norm": 1.3024168395197944, "learning_rate": 3.067971558826433e-08, "loss": 0.3956, "step": 27389 }, { "epoch": 0.9758269946737446, "grad_norm": 1.2664001837008678, "learning_rate": 3.058946084771086e-08, "loss": 0.4088, "step": 27390 }, { "epoch": 0.9758626218002387, "grad_norm": 1.2828387707019275, "learning_rate": 3.049933885682754e-08, "loss": 0.3607, "step": 27391 }, { "epoch": 0.9758982489267328, "grad_norm": 1.2308863100405694, "learning_rate": 3.040934961681341e-08, "loss": 0.3804, "step": 27392 }, { "epoch": 0.9759338760532269, "grad_norm": 1.91253899156426, "learning_rate": 3.03194931288664e-08, "loss": 0.5758, "step": 27393 }, { "epoch": 0.9759695031797211, "grad_norm": 1.469805456663251, "learning_rate": 3.0229769394184427e-08, "loss": 0.4874, "step": 27394 }, { "epoch": 0.9760051303062152, "grad_norm": 2.0068807085722953, "learning_rate": 3.014017841396211e-08, "loss": 0.4667, "step": 27395 }, { "epoch": 0.9760407574327092, "grad_norm": 1.443428156152512, "learning_rate": 3.00507201893907e-08, "loss": 0.4123, "step": 27396 }, { "epoch": 0.9760763845592034, "grad_norm": 1.482537914391941, "learning_rate": 2.9961394721663706e-08, "loss": 0.5132, "step": 27397 }, { "epoch": 0.9761120116856975, "grad_norm": 1.4527934003162672, "learning_rate": 2.987220201196794e-08, "loss": 0.4081, "step": 27398 }, { "epoch": 0.9761476388121916, "grad_norm": 1.3408505228566716, "learning_rate": 2.9783142061493575e-08, "loss": 0.5071, "step": 27399 }, { "epoch": 0.9761832659386858, "grad_norm": 2.0331696109378483, "learning_rate": 2.969421487142521e-08, "loss": 0.4567, "step": 27400 }, { "epoch": 0.9762188930651798, "grad_norm": 2.560114114581195, "learning_rate": 2.9605420442946342e-08, "loss": 0.4915, "step": 27401 }, { "epoch": 0.9762545201916739, "grad_norm": 1.263271369328901, "learning_rate": 2.9516758777241583e-08, "loss": 0.3859, "step": 27402 }, { "epoch": 0.9762901473181681, "grad_norm": 1.320460154979649, "learning_rate": 2.9428229875488878e-08, "loss": 0.422, "step": 27403 }, { "epoch": 0.9763257744446622, "grad_norm": 1.5777920440562032, "learning_rate": 2.9339833738869506e-08, "loss": 0.5449, "step": 27404 }, { "epoch": 0.9763614015711563, "grad_norm": 1.3584077690608778, "learning_rate": 2.925157036855919e-08, "loss": 0.4385, "step": 27405 }, { "epoch": 0.9763970286976504, "grad_norm": 1.404713247548786, "learning_rate": 2.916343976573255e-08, "loss": 0.3657, "step": 27406 }, { "epoch": 0.9764326558241445, "grad_norm": 2.2229718779300507, "learning_rate": 2.9075441931564197e-08, "loss": 0.5098, "step": 27407 }, { "epoch": 0.9764682829506386, "grad_norm": 1.9239658438853506, "learning_rate": 2.898757686722542e-08, "loss": 0.4166, "step": 27408 }, { "epoch": 0.9765039100771328, "grad_norm": 1.272421550808297, "learning_rate": 2.8899844573886392e-08, "loss": 0.3885, "step": 27409 }, { "epoch": 0.9765395372036269, "grad_norm": 1.2276463663703483, "learning_rate": 2.881224505271618e-08, "loss": 0.4223, "step": 27410 }, { "epoch": 0.9765751643301209, "grad_norm": 1.3359797793110835, "learning_rate": 2.8724778304879407e-08, "loss": 0.4809, "step": 27411 }, { "epoch": 0.9766107914566151, "grad_norm": 1.3732372410810867, "learning_rate": 2.8637444331541808e-08, "loss": 0.37, "step": 27412 }, { "epoch": 0.9766464185831092, "grad_norm": 1.8332153341226336, "learning_rate": 2.8550243133866896e-08, "loss": 0.5543, "step": 27413 }, { "epoch": 0.9766820457096033, "grad_norm": 1.2768066534285623, "learning_rate": 2.8463174713014852e-08, "loss": 0.4048, "step": 27414 }, { "epoch": 0.9767176728360975, "grad_norm": 1.6293101275314585, "learning_rate": 2.837623907014475e-08, "loss": 0.5761, "step": 27415 }, { "epoch": 0.9767532999625915, "grad_norm": 1.983845482420661, "learning_rate": 2.8289436206415665e-08, "loss": 0.4519, "step": 27416 }, { "epoch": 0.9767889270890856, "grad_norm": 1.2770719100850838, "learning_rate": 2.820276612298223e-08, "loss": 0.4414, "step": 27417 }, { "epoch": 0.9768245542155798, "grad_norm": 1.3032812437329413, "learning_rate": 2.811622882099796e-08, "loss": 0.3713, "step": 27418 }, { "epoch": 0.9768601813420739, "grad_norm": 1.5740575367936336, "learning_rate": 2.802982430161749e-08, "loss": 0.5197, "step": 27419 }, { "epoch": 0.9768958084685679, "grad_norm": 1.4370358097349667, "learning_rate": 2.7943552565988796e-08, "loss": 0.4096, "step": 27420 }, { "epoch": 0.9769314355950621, "grad_norm": 1.6696258270850597, "learning_rate": 2.7857413615262063e-08, "loss": 0.4918, "step": 27421 }, { "epoch": 0.9769670627215562, "grad_norm": 1.5028163662005427, "learning_rate": 2.7771407450584154e-08, "loss": 0.4047, "step": 27422 }, { "epoch": 0.9770026898480503, "grad_norm": 1.8746059073395207, "learning_rate": 2.7685534073099705e-08, "loss": 0.4817, "step": 27423 }, { "epoch": 0.9770383169745445, "grad_norm": 1.5428790012452278, "learning_rate": 2.7599793483952254e-08, "loss": 0.4658, "step": 27424 }, { "epoch": 0.9770739441010385, "grad_norm": 1.2976806232878393, "learning_rate": 2.7514185684284212e-08, "loss": 0.4039, "step": 27425 }, { "epoch": 0.9771095712275326, "grad_norm": 1.5342773502313654, "learning_rate": 2.742871067523578e-08, "loss": 0.5301, "step": 27426 }, { "epoch": 0.9771451983540268, "grad_norm": 1.26470670721267, "learning_rate": 2.734336845794383e-08, "loss": 0.4162, "step": 27427 }, { "epoch": 0.9771808254805209, "grad_norm": 1.5231950052410959, "learning_rate": 2.725815903354634e-08, "loss": 0.4855, "step": 27428 }, { "epoch": 0.977216452607015, "grad_norm": 1.1989236916202497, "learning_rate": 2.7173082403175733e-08, "loss": 0.3343, "step": 27429 }, { "epoch": 0.9772520797335091, "grad_norm": 1.3042561889018454, "learning_rate": 2.708813856796666e-08, "loss": 0.4009, "step": 27430 }, { "epoch": 0.9772877068600032, "grad_norm": 1.2805395135260373, "learning_rate": 2.7003327529050438e-08, "loss": 0.3831, "step": 27431 }, { "epoch": 0.9773233339864973, "grad_norm": 1.3887980690695032, "learning_rate": 2.6918649287555055e-08, "loss": 0.4219, "step": 27432 }, { "epoch": 0.9773589611129915, "grad_norm": 1.4435521624539762, "learning_rate": 2.6834103844607383e-08, "loss": 0.4944, "step": 27433 }, { "epoch": 0.9773945882394856, "grad_norm": 1.2906062582389473, "learning_rate": 2.6749691201335416e-08, "loss": 0.4493, "step": 27434 }, { "epoch": 0.9774302153659796, "grad_norm": 1.2725132703234607, "learning_rate": 2.6665411358862693e-08, "loss": 0.4135, "step": 27435 }, { "epoch": 0.9774658424924738, "grad_norm": 1.9260891805539841, "learning_rate": 2.658126431831054e-08, "loss": 0.5605, "step": 27436 }, { "epoch": 0.9775014696189679, "grad_norm": 1.280673684380376, "learning_rate": 2.6497250080800285e-08, "loss": 0.4107, "step": 27437 }, { "epoch": 0.977537096745462, "grad_norm": 1.6669475170045744, "learning_rate": 2.641336864744992e-08, "loss": 0.4615, "step": 27438 }, { "epoch": 0.9775727238719562, "grad_norm": 1.8400421540594822, "learning_rate": 2.6329620019376335e-08, "loss": 0.4868, "step": 27439 }, { "epoch": 0.9776083509984502, "grad_norm": 1.3942244759124038, "learning_rate": 2.6246004197695296e-08, "loss": 0.5326, "step": 27440 }, { "epoch": 0.9776439781249443, "grad_norm": 1.640887526539839, "learning_rate": 2.6162521183520362e-08, "loss": 0.5665, "step": 27441 }, { "epoch": 0.9776796052514385, "grad_norm": 1.3285183574699648, "learning_rate": 2.6079170977961754e-08, "loss": 0.3739, "step": 27442 }, { "epoch": 0.9777152323779326, "grad_norm": 1.4168791683643738, "learning_rate": 2.5995953582130806e-08, "loss": 0.5599, "step": 27443 }, { "epoch": 0.9777508595044266, "grad_norm": 1.591765922774646, "learning_rate": 2.591286899713552e-08, "loss": 0.5385, "step": 27444 }, { "epoch": 0.9777864866309208, "grad_norm": 1.3323173637344663, "learning_rate": 2.582991722408168e-08, "loss": 0.4062, "step": 27445 }, { "epoch": 0.9778221137574149, "grad_norm": 1.3808901574029069, "learning_rate": 2.5747098264073957e-08, "loss": 0.3691, "step": 27446 }, { "epoch": 0.977857740883909, "grad_norm": 1.596152984878125, "learning_rate": 2.56644121182148e-08, "loss": 0.5125, "step": 27447 }, { "epoch": 0.9778933680104032, "grad_norm": 1.4705548041826386, "learning_rate": 2.558185878760555e-08, "loss": 0.4278, "step": 27448 }, { "epoch": 0.9779289951368972, "grad_norm": 1.8736725396017855, "learning_rate": 2.5499438273346445e-08, "loss": 0.4917, "step": 27449 }, { "epoch": 0.9779646222633913, "grad_norm": 1.3419323278729514, "learning_rate": 2.5417150576533267e-08, "loss": 0.3824, "step": 27450 }, { "epoch": 0.9780002493898855, "grad_norm": 1.5135983149854315, "learning_rate": 2.533499569826292e-08, "loss": 0.43, "step": 27451 }, { "epoch": 0.9780358765163796, "grad_norm": 1.7148574611268461, "learning_rate": 2.5252973639628975e-08, "loss": 0.5816, "step": 27452 }, { "epoch": 0.9780715036428737, "grad_norm": 1.3430070289474654, "learning_rate": 2.517108440172389e-08, "loss": 0.4056, "step": 27453 }, { "epoch": 0.9781071307693678, "grad_norm": 1.7143583389595844, "learning_rate": 2.50893279856379e-08, "loss": 0.5136, "step": 27454 }, { "epoch": 0.9781427578958619, "grad_norm": 1.3021075060880707, "learning_rate": 2.500770439245903e-08, "loss": 0.374, "step": 27455 }, { "epoch": 0.978178385022356, "grad_norm": 1.2932866189858896, "learning_rate": 2.4926213623274185e-08, "loss": 0.4393, "step": 27456 }, { "epoch": 0.9782140121488502, "grad_norm": 1.6924027764894014, "learning_rate": 2.4844855679170276e-08, "loss": 0.4829, "step": 27457 }, { "epoch": 0.9782496392753443, "grad_norm": 1.4314542284515162, "learning_rate": 2.476363056122866e-08, "loss": 0.5195, "step": 27458 }, { "epoch": 0.9782852664018383, "grad_norm": 1.633125032133311, "learning_rate": 2.4682538270531797e-08, "loss": 0.4931, "step": 27459 }, { "epoch": 0.9783208935283325, "grad_norm": 1.4926236020127044, "learning_rate": 2.4601578808158832e-08, "loss": 0.4209, "step": 27460 }, { "epoch": 0.9783565206548266, "grad_norm": 1.202810171776542, "learning_rate": 2.452075217518779e-08, "loss": 0.3868, "step": 27461 }, { "epoch": 0.9783921477813207, "grad_norm": 1.3136846962204762, "learning_rate": 2.4440058372695586e-08, "loss": 0.4237, "step": 27462 }, { "epoch": 0.9784277749078149, "grad_norm": 1.1893328406923172, "learning_rate": 2.4359497401758026e-08, "loss": 0.3848, "step": 27463 }, { "epoch": 0.9784634020343089, "grad_norm": 1.2912198993190984, "learning_rate": 2.4279069263444254e-08, "loss": 0.4017, "step": 27464 }, { "epoch": 0.978499029160803, "grad_norm": 1.5390797003355305, "learning_rate": 2.419877395882897e-08, "loss": 0.4696, "step": 27465 }, { "epoch": 0.9785346562872972, "grad_norm": 1.3009934875830635, "learning_rate": 2.4118611488979095e-08, "loss": 0.3921, "step": 27466 }, { "epoch": 0.9785702834137913, "grad_norm": 1.2436979966326307, "learning_rate": 2.403858185496266e-08, "loss": 0.3869, "step": 27467 }, { "epoch": 0.9786059105402853, "grad_norm": 1.7621731365931177, "learning_rate": 2.3958685057844378e-08, "loss": 0.569, "step": 27468 }, { "epoch": 0.9786415376667795, "grad_norm": 1.3403593189895362, "learning_rate": 2.3878921098690055e-08, "loss": 0.442, "step": 27469 }, { "epoch": 0.9786771647932736, "grad_norm": 1.4136760361338738, "learning_rate": 2.379928997856107e-08, "loss": 0.4366, "step": 27470 }, { "epoch": 0.9787127919197677, "grad_norm": 1.5577158260559736, "learning_rate": 2.371979169851657e-08, "loss": 0.489, "step": 27471 }, { "epoch": 0.9787484190462619, "grad_norm": 1.6489043232419713, "learning_rate": 2.364042625961682e-08, "loss": 0.5279, "step": 27472 }, { "epoch": 0.9787840461727559, "grad_norm": 1.3207789461748445, "learning_rate": 2.3561193662916536e-08, "loss": 0.4123, "step": 27473 }, { "epoch": 0.97881967329925, "grad_norm": 1.4207370717975376, "learning_rate": 2.348209390947376e-08, "loss": 0.519, "step": 27474 }, { "epoch": 0.9788553004257442, "grad_norm": 1.1370951278982, "learning_rate": 2.3403127000338756e-08, "loss": 0.377, "step": 27475 }, { "epoch": 0.9788909275522383, "grad_norm": 1.5881229495401588, "learning_rate": 2.332429293656513e-08, "loss": 0.5319, "step": 27476 }, { "epoch": 0.9789265546787324, "grad_norm": 1.4695250295559028, "learning_rate": 2.3245591719200932e-08, "loss": 0.502, "step": 27477 }, { "epoch": 0.9789621818052265, "grad_norm": 1.349951911665944, "learning_rate": 2.3167023349296435e-08, "loss": 0.3928, "step": 27478 }, { "epoch": 0.9789978089317206, "grad_norm": 1.3021935019046085, "learning_rate": 2.3088587827895247e-08, "loss": 0.3448, "step": 27479 }, { "epoch": 0.9790334360582147, "grad_norm": 1.5963013386897311, "learning_rate": 2.30102851560432e-08, "loss": 0.5271, "step": 27480 }, { "epoch": 0.9790690631847089, "grad_norm": 1.469360884917057, "learning_rate": 2.29321153347839e-08, "loss": 0.5196, "step": 27481 }, { "epoch": 0.979104690311203, "grad_norm": 2.0720059532453465, "learning_rate": 2.2854078365155406e-08, "loss": 0.5206, "step": 27482 }, { "epoch": 0.979140317437697, "grad_norm": 1.457715798291148, "learning_rate": 2.2776174248199114e-08, "loss": 0.4475, "step": 27483 }, { "epoch": 0.9791759445641912, "grad_norm": 1.6278940378983664, "learning_rate": 2.2698402984951962e-08, "loss": 0.4747, "step": 27484 }, { "epoch": 0.9792115716906853, "grad_norm": 1.3770968942369068, "learning_rate": 2.2620764576448685e-08, "loss": 0.4547, "step": 27485 }, { "epoch": 0.9792471988171794, "grad_norm": 1.1929977453974778, "learning_rate": 2.2543259023724005e-08, "loss": 0.3632, "step": 27486 }, { "epoch": 0.9792828259436736, "grad_norm": 1.2948373339810675, "learning_rate": 2.246588632780933e-08, "loss": 0.3818, "step": 27487 }, { "epoch": 0.9793184530701676, "grad_norm": 1.6125322603449748, "learning_rate": 2.2388646489736043e-08, "loss": 0.3778, "step": 27488 }, { "epoch": 0.9793540801966617, "grad_norm": 1.527767970924379, "learning_rate": 2.231153951053111e-08, "loss": 0.4718, "step": 27489 }, { "epoch": 0.9793897073231559, "grad_norm": 1.1801704881035462, "learning_rate": 2.2234565391222595e-08, "loss": 0.4187, "step": 27490 }, { "epoch": 0.97942533444965, "grad_norm": 1.4815903557440788, "learning_rate": 2.2157724132834126e-08, "loss": 0.4939, "step": 27491 }, { "epoch": 0.979460961576144, "grad_norm": 1.5465088167465013, "learning_rate": 2.2081015736390432e-08, "loss": 0.4613, "step": 27492 }, { "epoch": 0.9794965887026382, "grad_norm": 1.3061096174907112, "learning_rate": 2.2004440202911816e-08, "loss": 0.3781, "step": 27493 }, { "epoch": 0.9795322158291323, "grad_norm": 1.4946190996605015, "learning_rate": 2.1927997533418567e-08, "loss": 0.4687, "step": 27494 }, { "epoch": 0.9795678429556264, "grad_norm": 1.305614005237615, "learning_rate": 2.185168772892765e-08, "loss": 0.3792, "step": 27495 }, { "epoch": 0.9796034700821206, "grad_norm": 1.3056638806364544, "learning_rate": 2.1775510790456035e-08, "loss": 0.3821, "step": 27496 }, { "epoch": 0.9796390972086146, "grad_norm": 1.2796022919786643, "learning_rate": 2.1699466719018457e-08, "loss": 0.3518, "step": 27497 }, { "epoch": 0.9796747243351087, "grad_norm": 1.7734675648222804, "learning_rate": 2.1623555515626336e-08, "loss": 0.5491, "step": 27498 }, { "epoch": 0.9797103514616029, "grad_norm": 2.0182576055795924, "learning_rate": 2.1547777181291084e-08, "loss": 0.5127, "step": 27499 }, { "epoch": 0.979745978588097, "grad_norm": 1.4795747080965804, "learning_rate": 2.1472131717021894e-08, "loss": 0.5094, "step": 27500 }, { "epoch": 0.979781605714591, "grad_norm": 1.5155635001649097, "learning_rate": 2.1396619123825735e-08, "loss": 0.4954, "step": 27501 }, { "epoch": 0.9798172328410852, "grad_norm": 1.3652815679657297, "learning_rate": 2.132123940270847e-08, "loss": 0.537, "step": 27502 }, { "epoch": 0.9798528599675793, "grad_norm": 1.4796378621699364, "learning_rate": 2.1245992554673743e-08, "loss": 0.467, "step": 27503 }, { "epoch": 0.9798884870940734, "grad_norm": 1.2851453721790098, "learning_rate": 2.1170878580721867e-08, "loss": 0.3825, "step": 27504 }, { "epoch": 0.9799241142205676, "grad_norm": 1.2621449259754465, "learning_rate": 2.1095897481856476e-08, "loss": 0.3716, "step": 27505 }, { "epoch": 0.9799597413470617, "grad_norm": 1.3049586837626677, "learning_rate": 2.1021049259072335e-08, "loss": 0.4027, "step": 27506 }, { "epoch": 0.9799953684735557, "grad_norm": 1.2547307728420587, "learning_rate": 2.0946333913368645e-08, "loss": 0.4471, "step": 27507 }, { "epoch": 0.9800309956000499, "grad_norm": 1.1418302115197656, "learning_rate": 2.0871751445740163e-08, "loss": 0.3552, "step": 27508 }, { "epoch": 0.980066622726544, "grad_norm": 1.2719661978938477, "learning_rate": 2.079730185717832e-08, "loss": 0.4201, "step": 27509 }, { "epoch": 0.9801022498530381, "grad_norm": 1.1873937134647967, "learning_rate": 2.0722985148675655e-08, "loss": 0.383, "step": 27510 }, { "epoch": 0.9801378769795323, "grad_norm": 2.0483424025545403, "learning_rate": 2.064880132122249e-08, "loss": 0.5412, "step": 27511 }, { "epoch": 0.9801735041060263, "grad_norm": 1.4478252907619353, "learning_rate": 2.0574750375804696e-08, "loss": 0.4349, "step": 27512 }, { "epoch": 0.9802091312325204, "grad_norm": 1.8334635511247857, "learning_rate": 2.050083231340927e-08, "loss": 0.5503, "step": 27513 }, { "epoch": 0.9802447583590146, "grad_norm": 2.021830361935971, "learning_rate": 2.042704713501986e-08, "loss": 0.4957, "step": 27514 }, { "epoch": 0.9802803854855087, "grad_norm": 1.3686663683705202, "learning_rate": 2.0353394841621244e-08, "loss": 0.4281, "step": 27515 }, { "epoch": 0.9803160126120027, "grad_norm": 1.5925879428653182, "learning_rate": 2.027987543419152e-08, "loss": 0.4447, "step": 27516 }, { "epoch": 0.9803516397384969, "grad_norm": 1.593320098321282, "learning_rate": 2.020648891371102e-08, "loss": 0.6058, "step": 27517 }, { "epoch": 0.980387266864991, "grad_norm": 1.990601959767141, "learning_rate": 2.013323528115674e-08, "loss": 0.5917, "step": 27518 }, { "epoch": 0.9804228939914851, "grad_norm": 1.4180390018667872, "learning_rate": 2.006011453750345e-08, "loss": 0.4312, "step": 27519 }, { "epoch": 0.9804585211179793, "grad_norm": 1.1923172388860137, "learning_rate": 1.9987126683724822e-08, "loss": 0.372, "step": 27520 }, { "epoch": 0.9804941482444733, "grad_norm": 1.5847308245954372, "learning_rate": 1.991427172079452e-08, "loss": 0.4686, "step": 27521 }, { "epoch": 0.9805297753709674, "grad_norm": 1.8406000428539442, "learning_rate": 1.9841549649680658e-08, "loss": 0.5723, "step": 27522 }, { "epoch": 0.9805654024974616, "grad_norm": 1.1673778156178658, "learning_rate": 1.976896047135246e-08, "loss": 0.404, "step": 27523 }, { "epoch": 0.9806010296239557, "grad_norm": 2.191193905387645, "learning_rate": 1.9696504186775822e-08, "loss": 0.4424, "step": 27524 }, { "epoch": 0.9806366567504498, "grad_norm": 1.9046622242994427, "learning_rate": 1.9624180796916637e-08, "loss": 0.4333, "step": 27525 }, { "epoch": 0.9806722838769439, "grad_norm": 1.3075572485231235, "learning_rate": 1.9551990302736357e-08, "loss": 0.4111, "step": 27526 }, { "epoch": 0.980707911003438, "grad_norm": 1.4873365489619448, "learning_rate": 1.9479932705198658e-08, "loss": 0.4147, "step": 27527 }, { "epoch": 0.9807435381299321, "grad_norm": 1.6007619905810122, "learning_rate": 1.9408008005260548e-08, "loss": 0.4704, "step": 27528 }, { "epoch": 0.9807791652564263, "grad_norm": 1.1817567934517992, "learning_rate": 1.9336216203881265e-08, "loss": 0.3429, "step": 27529 }, { "epoch": 0.9808147923829204, "grad_norm": 1.2546936732716971, "learning_rate": 1.9264557302016706e-08, "loss": 0.3495, "step": 27530 }, { "epoch": 0.9808504195094144, "grad_norm": 1.5728512936717136, "learning_rate": 1.9193031300620558e-08, "loss": 0.4943, "step": 27531 }, { "epoch": 0.9808860466359086, "grad_norm": 1.6866640671120443, "learning_rate": 1.9121638200644276e-08, "loss": 0.5529, "step": 27532 }, { "epoch": 0.9809216737624027, "grad_norm": 2.130949116759307, "learning_rate": 1.9050378003040438e-08, "loss": 0.4481, "step": 27533 }, { "epoch": 0.9809573008888968, "grad_norm": 1.2632011948138278, "learning_rate": 1.897925070875717e-08, "loss": 0.3707, "step": 27534 }, { "epoch": 0.980992928015391, "grad_norm": 1.281191726495806, "learning_rate": 1.890825631874149e-08, "loss": 0.406, "step": 27535 }, { "epoch": 0.981028555141885, "grad_norm": 1.5355770509647135, "learning_rate": 1.8837394833938207e-08, "loss": 0.5331, "step": 27536 }, { "epoch": 0.9810641822683791, "grad_norm": 1.2530153299966118, "learning_rate": 1.876666625529211e-08, "loss": 0.4159, "step": 27537 }, { "epoch": 0.9810998093948733, "grad_norm": 1.587556774885981, "learning_rate": 1.8696070583744674e-08, "loss": 0.5363, "step": 27538 }, { "epoch": 0.9811354365213674, "grad_norm": 1.534623620035124, "learning_rate": 1.8625607820234037e-08, "loss": 0.4177, "step": 27539 }, { "epoch": 0.9811710636478614, "grad_norm": 1.5721124516937808, "learning_rate": 1.855527796570167e-08, "loss": 0.5325, "step": 27540 }, { "epoch": 0.9812066907743556, "grad_norm": 1.3504418279266428, "learning_rate": 1.8485081021081263e-08, "loss": 0.3932, "step": 27541 }, { "epoch": 0.9812423179008497, "grad_norm": 1.2558177782768276, "learning_rate": 1.8415016987308742e-08, "loss": 0.3754, "step": 27542 }, { "epoch": 0.9812779450273438, "grad_norm": 1.8544038388161646, "learning_rate": 1.83450858653178e-08, "loss": 0.5128, "step": 27543 }, { "epoch": 0.981313572153838, "grad_norm": 1.432218946166878, "learning_rate": 1.82752876560377e-08, "loss": 0.499, "step": 27544 }, { "epoch": 0.981349199280332, "grad_norm": 1.5463703025137119, "learning_rate": 1.820562236039991e-08, "loss": 0.4434, "step": 27545 }, { "epoch": 0.9813848264068261, "grad_norm": 1.2511806824239529, "learning_rate": 1.8136089979331473e-08, "loss": 0.3551, "step": 27546 }, { "epoch": 0.9814204535333203, "grad_norm": 1.6719136062914082, "learning_rate": 1.806669051375609e-08, "loss": 0.5149, "step": 27547 }, { "epoch": 0.9814560806598144, "grad_norm": 1.6899234349535637, "learning_rate": 1.7997423964601913e-08, "loss": 0.5012, "step": 27548 }, { "epoch": 0.9814917077863085, "grad_norm": 1.3848305691490288, "learning_rate": 1.7928290332787092e-08, "loss": 0.4136, "step": 27549 }, { "epoch": 0.9815273349128026, "grad_norm": 1.5020171361153196, "learning_rate": 1.7859289619235332e-08, "loss": 0.4746, "step": 27550 }, { "epoch": 0.9815629620392967, "grad_norm": 1.6732659951857367, "learning_rate": 1.779042182486368e-08, "loss": 0.6131, "step": 27551 }, { "epoch": 0.9815985891657908, "grad_norm": 1.4662655630218717, "learning_rate": 1.7721686950589178e-08, "loss": 0.4164, "step": 27552 }, { "epoch": 0.981634216292285, "grad_norm": 1.2810570580727207, "learning_rate": 1.765308499732887e-08, "loss": 0.3795, "step": 27553 }, { "epoch": 0.981669843418779, "grad_norm": 1.2862483157983986, "learning_rate": 1.7584615965994256e-08, "loss": 0.3912, "step": 27554 }, { "epoch": 0.9817054705452731, "grad_norm": 1.4295705978545334, "learning_rate": 1.7516279857497933e-08, "loss": 0.4626, "step": 27555 }, { "epoch": 0.9817410976717673, "grad_norm": 1.3233705454864746, "learning_rate": 1.744807667275028e-08, "loss": 0.3546, "step": 27556 }, { "epoch": 0.9817767247982614, "grad_norm": 1.6524164663494376, "learning_rate": 1.7380006412657246e-08, "loss": 0.3974, "step": 27557 }, { "epoch": 0.9818123519247555, "grad_norm": 1.4198608289999657, "learning_rate": 1.73120690781281e-08, "loss": 0.4418, "step": 27558 }, { "epoch": 0.9818479790512497, "grad_norm": 1.2529177500600253, "learning_rate": 1.724426467006657e-08, "loss": 0.4202, "step": 27559 }, { "epoch": 0.9818836061777437, "grad_norm": 1.4354538192464155, "learning_rate": 1.7176593189374145e-08, "loss": 0.4981, "step": 27560 }, { "epoch": 0.9819192333042378, "grad_norm": 1.3861526592451976, "learning_rate": 1.710905463695345e-08, "loss": 0.4921, "step": 27561 }, { "epoch": 0.981954860430732, "grad_norm": 1.1842383048235425, "learning_rate": 1.7041649013703754e-08, "loss": 0.3758, "step": 27562 }, { "epoch": 0.9819904875572261, "grad_norm": 1.6605655388576293, "learning_rate": 1.6974376320521014e-08, "loss": 0.4984, "step": 27563 }, { "epoch": 0.9820261146837201, "grad_norm": 1.345949537258071, "learning_rate": 1.6907236558303397e-08, "loss": 0.5606, "step": 27564 }, { "epoch": 0.9820617418102143, "grad_norm": 1.1172755048407936, "learning_rate": 1.6840229727942415e-08, "loss": 0.3926, "step": 27565 }, { "epoch": 0.9820973689367084, "grad_norm": 1.272732098112671, "learning_rate": 1.6773355830332903e-08, "loss": 0.3554, "step": 27566 }, { "epoch": 0.9821329960632025, "grad_norm": 1.5498329769922796, "learning_rate": 1.6706614866363048e-08, "loss": 0.4875, "step": 27567 }, { "epoch": 0.9821686231896967, "grad_norm": 1.6795391758079663, "learning_rate": 1.664000683692324e-08, "loss": 0.5293, "step": 27568 }, { "epoch": 0.9822042503161907, "grad_norm": 1.6510216787664607, "learning_rate": 1.657353174289944e-08, "loss": 0.6122, "step": 27569 }, { "epoch": 0.9822398774426848, "grad_norm": 1.5578936721803105, "learning_rate": 1.6507189585177606e-08, "loss": 0.482, "step": 27570 }, { "epoch": 0.982275504569179, "grad_norm": 1.2940741012423829, "learning_rate": 1.6440980364639258e-08, "loss": 0.4191, "step": 27571 }, { "epoch": 0.9823111316956731, "grad_norm": 1.603488816611168, "learning_rate": 1.637490408216813e-08, "loss": 0.5274, "step": 27572 }, { "epoch": 0.9823467588221672, "grad_norm": 1.413324869510439, "learning_rate": 1.630896073864352e-08, "loss": 0.5114, "step": 27573 }, { "epoch": 0.9823823859486613, "grad_norm": 1.499766077671831, "learning_rate": 1.6243150334942503e-08, "loss": 0.3868, "step": 27574 }, { "epoch": 0.9824180130751554, "grad_norm": 1.4087586760800725, "learning_rate": 1.6177472871943268e-08, "loss": 0.3876, "step": 27575 }, { "epoch": 0.9824536402016495, "grad_norm": 1.358593856869747, "learning_rate": 1.6111928350518445e-08, "loss": 0.5074, "step": 27576 }, { "epoch": 0.9824892673281437, "grad_norm": 1.4269435328587088, "learning_rate": 1.6046516771541787e-08, "loss": 0.5518, "step": 27577 }, { "epoch": 0.9825248944546378, "grad_norm": 1.5596053427557275, "learning_rate": 1.5981238135883705e-08, "loss": 0.5395, "step": 27578 }, { "epoch": 0.9825605215811318, "grad_norm": 1.1465085573640117, "learning_rate": 1.591609244441461e-08, "loss": 0.3826, "step": 27579 }, { "epoch": 0.982596148707626, "grad_norm": 1.8453310199947046, "learning_rate": 1.585107969800048e-08, "loss": 0.4419, "step": 27580 }, { "epoch": 0.9826317758341201, "grad_norm": 1.3383317553664962, "learning_rate": 1.5786199897508402e-08, "loss": 0.3685, "step": 27581 }, { "epoch": 0.9826674029606142, "grad_norm": 1.6108908639897823, "learning_rate": 1.5721453043801014e-08, "loss": 0.5195, "step": 27582 }, { "epoch": 0.9827030300871084, "grad_norm": 1.2808033125916614, "learning_rate": 1.5656839137742073e-08, "loss": 0.4069, "step": 27583 }, { "epoch": 0.9827386572136024, "grad_norm": 1.6164129722266583, "learning_rate": 1.5592358180189782e-08, "loss": 0.5498, "step": 27584 }, { "epoch": 0.9827742843400965, "grad_norm": 1.399237925348711, "learning_rate": 1.552801017200456e-08, "loss": 0.4137, "step": 27585 }, { "epoch": 0.9828099114665907, "grad_norm": 1.165951212266164, "learning_rate": 1.546379511404239e-08, "loss": 0.406, "step": 27586 }, { "epoch": 0.9828455385930848, "grad_norm": 1.2484159661189105, "learning_rate": 1.5399713007159255e-08, "loss": 0.4117, "step": 27587 }, { "epoch": 0.9828811657195788, "grad_norm": 1.3063868463617183, "learning_rate": 1.5335763852206697e-08, "loss": 0.399, "step": 27588 }, { "epoch": 0.982916792846073, "grad_norm": 1.3613694501643956, "learning_rate": 1.5271947650037365e-08, "loss": 0.41, "step": 27589 }, { "epoch": 0.9829524199725671, "grad_norm": 1.4331484107066474, "learning_rate": 1.5208264401501695e-08, "loss": 0.4329, "step": 27590 }, { "epoch": 0.9829880470990612, "grad_norm": 2.8363094588529862, "learning_rate": 1.5144714107446777e-08, "loss": 0.4159, "step": 27591 }, { "epoch": 0.9830236742255554, "grad_norm": 1.269433766412918, "learning_rate": 1.5081296768718613e-08, "loss": 0.3879, "step": 27592 }, { "epoch": 0.9830593013520494, "grad_norm": 1.255810122058001, "learning_rate": 1.5018012386162072e-08, "loss": 0.406, "step": 27593 }, { "epoch": 0.9830949284785435, "grad_norm": 1.4469909142925679, "learning_rate": 1.4954860960619823e-08, "loss": 0.4984, "step": 27594 }, { "epoch": 0.9831305556050377, "grad_norm": 1.3274969673903385, "learning_rate": 1.4891842492933406e-08, "loss": 0.362, "step": 27595 }, { "epoch": 0.9831661827315318, "grad_norm": 1.1011114763194758, "learning_rate": 1.4828956983941046e-08, "loss": 0.3738, "step": 27596 }, { "epoch": 0.9832018098580259, "grad_norm": 1.903003222320386, "learning_rate": 1.4766204434479848e-08, "loss": 0.4316, "step": 27597 }, { "epoch": 0.98323743698452, "grad_norm": 1.3504031784050072, "learning_rate": 1.470358484538581e-08, "loss": 0.3673, "step": 27598 }, { "epoch": 0.9832730641110141, "grad_norm": 1.27924580877623, "learning_rate": 1.4641098217493821e-08, "loss": 0.3801, "step": 27599 }, { "epoch": 0.9833086912375082, "grad_norm": 1.975085109630755, "learning_rate": 1.4578744551633217e-08, "loss": 0.4959, "step": 27600 }, { "epoch": 0.9833443183640024, "grad_norm": 1.6772060893668523, "learning_rate": 1.4516523848637776e-08, "loss": 0.4442, "step": 27601 }, { "epoch": 0.9833799454904965, "grad_norm": 1.4479744675973099, "learning_rate": 1.4454436109333502e-08, "loss": 0.4775, "step": 27602 }, { "epoch": 0.9834155726169905, "grad_norm": 1.3681707445976206, "learning_rate": 1.4392481334547515e-08, "loss": 0.4235, "step": 27603 }, { "epoch": 0.9834511997434847, "grad_norm": 1.674430022668506, "learning_rate": 1.4330659525105817e-08, "loss": 0.577, "step": 27604 }, { "epoch": 0.9834868268699788, "grad_norm": 1.2981816482929558, "learning_rate": 1.4268970681831085e-08, "loss": 0.4292, "step": 27605 }, { "epoch": 0.9835224539964729, "grad_norm": 1.6979306928466447, "learning_rate": 1.4207414805543774e-08, "loss": 0.4675, "step": 27606 }, { "epoch": 0.983558081122967, "grad_norm": 1.5098641558831962, "learning_rate": 1.4145991897064336e-08, "loss": 0.418, "step": 27607 }, { "epoch": 0.9835937082494611, "grad_norm": 1.4363918056129283, "learning_rate": 1.4084701957212122e-08, "loss": 0.4973, "step": 27608 }, { "epoch": 0.9836293353759552, "grad_norm": 1.5122243604672507, "learning_rate": 1.4023544986800918e-08, "loss": 0.375, "step": 27609 }, { "epoch": 0.9836649625024494, "grad_norm": 1.5995011156985313, "learning_rate": 1.3962520986645634e-08, "loss": 0.4428, "step": 27610 }, { "epoch": 0.9837005896289435, "grad_norm": 1.547989890745016, "learning_rate": 1.3901629957558948e-08, "loss": 0.5224, "step": 27611 }, { "epoch": 0.9837362167554375, "grad_norm": 1.3153211311651234, "learning_rate": 1.3840871900353547e-08, "loss": 0.3985, "step": 27612 }, { "epoch": 0.9837718438819317, "grad_norm": 1.7889613228409813, "learning_rate": 1.3780246815835452e-08, "loss": 0.5563, "step": 27613 }, { "epoch": 0.9838074710084258, "grad_norm": 1.429531125622731, "learning_rate": 1.3719754704814015e-08, "loss": 0.4373, "step": 27614 }, { "epoch": 0.9838430981349199, "grad_norm": 1.2905828818620284, "learning_rate": 1.3659395568093037e-08, "loss": 0.4176, "step": 27615 }, { "epoch": 0.9838787252614141, "grad_norm": 1.719585363297837, "learning_rate": 1.359916940647854e-08, "loss": 0.5645, "step": 27616 }, { "epoch": 0.9839143523879081, "grad_norm": 1.4965905876883292, "learning_rate": 1.3539076220769887e-08, "loss": 0.369, "step": 27617 }, { "epoch": 0.9839499795144022, "grad_norm": 1.4685802729405224, "learning_rate": 1.3479116011769766e-08, "loss": 0.4521, "step": 27618 }, { "epoch": 0.9839856066408964, "grad_norm": 1.3715939339913066, "learning_rate": 1.341928878027421e-08, "loss": 0.3836, "step": 27619 }, { "epoch": 0.9840212337673905, "grad_norm": 1.3012610396205604, "learning_rate": 1.3359594527082575e-08, "loss": 0.3535, "step": 27620 }, { "epoch": 0.9840568608938846, "grad_norm": 1.3333512732290562, "learning_rate": 1.3300033252986454e-08, "loss": 0.3963, "step": 27621 }, { "epoch": 0.9840924880203787, "grad_norm": 1.629247114967574, "learning_rate": 1.3240604958781877e-08, "loss": 0.4699, "step": 27622 }, { "epoch": 0.9841281151468728, "grad_norm": 1.629813031486892, "learning_rate": 1.3181309645259321e-08, "loss": 0.4536, "step": 27623 }, { "epoch": 0.9841637422733669, "grad_norm": 1.2945778952858091, "learning_rate": 1.3122147313207046e-08, "loss": 0.4165, "step": 27624 }, { "epoch": 0.9841993693998611, "grad_norm": 1.2004375470729842, "learning_rate": 1.3063117963413307e-08, "loss": 0.4077, "step": 27625 }, { "epoch": 0.9842349965263552, "grad_norm": 1.565816013057186, "learning_rate": 1.3004221596665256e-08, "loss": 0.5203, "step": 27626 }, { "epoch": 0.9842706236528492, "grad_norm": 1.4782352397684844, "learning_rate": 1.294545821374671e-08, "loss": 0.4118, "step": 27627 }, { "epoch": 0.9843062507793434, "grad_norm": 1.803520466095489, "learning_rate": 1.2886827815440373e-08, "loss": 0.4275, "step": 27628 }, { "epoch": 0.9843418779058375, "grad_norm": 1.5389061408829519, "learning_rate": 1.2828330402525624e-08, "loss": 0.4581, "step": 27629 }, { "epoch": 0.9843775050323316, "grad_norm": 1.146382959250779, "learning_rate": 1.276996597578295e-08, "loss": 0.3624, "step": 27630 }, { "epoch": 0.9844131321588258, "grad_norm": 1.23566728692016, "learning_rate": 1.2711734535988396e-08, "loss": 0.3757, "step": 27631 }, { "epoch": 0.9844487592853198, "grad_norm": 1.6165134087507276, "learning_rate": 1.2653636083916898e-08, "loss": 0.5734, "step": 27632 }, { "epoch": 0.9844843864118139, "grad_norm": 2.365260971705302, "learning_rate": 1.2595670620344502e-08, "loss": 0.4763, "step": 27633 }, { "epoch": 0.9845200135383081, "grad_norm": 1.8756299998773305, "learning_rate": 1.2537838146040592e-08, "loss": 0.5867, "step": 27634 }, { "epoch": 0.9845556406648022, "grad_norm": 1.3260866378554894, "learning_rate": 1.2480138661776776e-08, "loss": 0.3964, "step": 27635 }, { "epoch": 0.9845912677912962, "grad_norm": 1.328085286835763, "learning_rate": 1.2422572168320212e-08, "loss": 0.4291, "step": 27636 }, { "epoch": 0.9846268949177904, "grad_norm": 1.4131336439623836, "learning_rate": 1.236513866643807e-08, "loss": 0.389, "step": 27637 }, { "epoch": 0.9846625220442845, "grad_norm": 1.4936261501585666, "learning_rate": 1.230783815689529e-08, "loss": 0.4933, "step": 27638 }, { "epoch": 0.9846981491707786, "grad_norm": 1.574294123330422, "learning_rate": 1.2250670640454599e-08, "loss": 0.5675, "step": 27639 }, { "epoch": 0.9847337762972728, "grad_norm": 1.5421181288481725, "learning_rate": 1.2193636117877604e-08, "loss": 0.4885, "step": 27640 }, { "epoch": 0.9847694034237668, "grad_norm": 1.2031292347216807, "learning_rate": 1.2136734589922594e-08, "loss": 0.4419, "step": 27641 }, { "epoch": 0.9848050305502609, "grad_norm": 1.418602971447387, "learning_rate": 1.2079966057348957e-08, "loss": 0.4849, "step": 27642 }, { "epoch": 0.9848406576767551, "grad_norm": 1.626228604870627, "learning_rate": 1.2023330520911648e-08, "loss": 0.413, "step": 27643 }, { "epoch": 0.9848762848032492, "grad_norm": 1.551048342574762, "learning_rate": 1.1966827981364505e-08, "loss": 0.5004, "step": 27644 }, { "epoch": 0.9849119119297433, "grad_norm": 1.5258875082298267, "learning_rate": 1.191045843946026e-08, "loss": 0.4318, "step": 27645 }, { "epoch": 0.9849475390562374, "grad_norm": 1.2380803349495053, "learning_rate": 1.1854221895949425e-08, "loss": 0.3921, "step": 27646 }, { "epoch": 0.9849831661827315, "grad_norm": 1.535022970542812, "learning_rate": 1.179811835158029e-08, "loss": 0.4365, "step": 27647 }, { "epoch": 0.9850187933092256, "grad_norm": 1.5178792233310006, "learning_rate": 1.1742147807101145e-08, "loss": 0.3702, "step": 27648 }, { "epoch": 0.9850544204357198, "grad_norm": 1.5096037732451801, "learning_rate": 1.168631026325695e-08, "loss": 0.5387, "step": 27649 }, { "epoch": 0.9850900475622139, "grad_norm": 1.665569678288795, "learning_rate": 1.1630605720791555e-08, "loss": 0.5613, "step": 27650 }, { "epoch": 0.9851256746887079, "grad_norm": 1.4226361937710186, "learning_rate": 1.1575034180445477e-08, "loss": 0.4061, "step": 27651 }, { "epoch": 0.9851613018152021, "grad_norm": 1.261353158367049, "learning_rate": 1.1519595642959236e-08, "loss": 0.421, "step": 27652 }, { "epoch": 0.9851969289416962, "grad_norm": 1.5541207900742116, "learning_rate": 1.146429010907002e-08, "loss": 0.5012, "step": 27653 }, { "epoch": 0.9852325560681903, "grad_norm": 1.5816886109910333, "learning_rate": 1.1409117579517237e-08, "loss": 0.5291, "step": 27654 }, { "epoch": 0.9852681831946845, "grad_norm": 1.2043626300116894, "learning_rate": 1.1354078055032525e-08, "loss": 0.3845, "step": 27655 }, { "epoch": 0.9853038103211785, "grad_norm": 4.85026852258548, "learning_rate": 1.129917153634974e-08, "loss": 0.5257, "step": 27656 }, { "epoch": 0.9853394374476726, "grad_norm": 1.2221791882891653, "learning_rate": 1.124439802420052e-08, "loss": 0.4013, "step": 27657 }, { "epoch": 0.9853750645741668, "grad_norm": 1.222508316099362, "learning_rate": 1.1189757519314282e-08, "loss": 0.3913, "step": 27658 }, { "epoch": 0.9854106917006609, "grad_norm": 1.4230846131899821, "learning_rate": 1.113525002241822e-08, "loss": 0.5471, "step": 27659 }, { "epoch": 0.9854463188271549, "grad_norm": 1.2113925458968817, "learning_rate": 1.1080875534237312e-08, "loss": 0.3753, "step": 27660 }, { "epoch": 0.9854819459536491, "grad_norm": 1.6249898803812683, "learning_rate": 1.1026634055497643e-08, "loss": 0.5473, "step": 27661 }, { "epoch": 0.9855175730801432, "grad_norm": 1.34231716138172, "learning_rate": 1.0972525586918636e-08, "loss": 0.3883, "step": 27662 }, { "epoch": 0.9855532002066373, "grad_norm": 1.6473479414287024, "learning_rate": 1.0918550129223049e-08, "loss": 0.4625, "step": 27663 }, { "epoch": 0.9855888273331315, "grad_norm": 1.5809244442028385, "learning_rate": 1.0864707683129194e-08, "loss": 0.5023, "step": 27664 }, { "epoch": 0.9856244544596255, "grad_norm": 1.4174429866863565, "learning_rate": 1.0810998249353167e-08, "loss": 0.4607, "step": 27665 }, { "epoch": 0.9856600815861196, "grad_norm": 1.2303754637977324, "learning_rate": 1.075742182861106e-08, "loss": 0.3738, "step": 27666 }, { "epoch": 0.9856957087126138, "grad_norm": 1.359371676115737, "learning_rate": 1.0703978421616745e-08, "loss": 0.404, "step": 27667 }, { "epoch": 0.9857313358391079, "grad_norm": 1.5730710967939205, "learning_rate": 1.0650668029079657e-08, "loss": 0.4971, "step": 27668 }, { "epoch": 0.985766962965602, "grad_norm": 1.197754458917889, "learning_rate": 1.059749065171145e-08, "loss": 0.3928, "step": 27669 }, { "epoch": 0.9858025900920961, "grad_norm": 1.3489868147974158, "learning_rate": 1.0544446290220445e-08, "loss": 0.4344, "step": 27670 }, { "epoch": 0.9858382172185902, "grad_norm": 1.6364573004691332, "learning_rate": 1.0491534945312742e-08, "loss": 0.4293, "step": 27671 }, { "epoch": 0.9858738443450843, "grad_norm": 1.3948242846973409, "learning_rate": 1.0438756617691115e-08, "loss": 0.3689, "step": 27672 }, { "epoch": 0.9859094714715785, "grad_norm": 1.5151289156736647, "learning_rate": 1.0386111308060554e-08, "loss": 0.5344, "step": 27673 }, { "epoch": 0.9859450985980726, "grad_norm": 1.4261304345058758, "learning_rate": 1.03335990171205e-08, "loss": 0.4925, "step": 27674 }, { "epoch": 0.9859807257245666, "grad_norm": 2.4189702062164677, "learning_rate": 1.0281219745571503e-08, "loss": 0.5089, "step": 27675 }, { "epoch": 0.9860163528510608, "grad_norm": 1.2937500360462049, "learning_rate": 1.0228973494110783e-08, "loss": 0.4149, "step": 27676 }, { "epoch": 0.9860519799775549, "grad_norm": 1.4479271873946256, "learning_rate": 1.0176860263433342e-08, "loss": 0.4747, "step": 27677 }, { "epoch": 0.986087607104049, "grad_norm": 1.4971256305947915, "learning_rate": 1.0124880054233066e-08, "loss": 0.5614, "step": 27678 }, { "epoch": 0.9861232342305432, "grad_norm": 1.4215027582157602, "learning_rate": 1.0073032867203846e-08, "loss": 0.4917, "step": 27679 }, { "epoch": 0.9861588613570372, "grad_norm": 1.5851624547477696, "learning_rate": 1.0021318703034022e-08, "loss": 0.4981, "step": 27680 }, { "epoch": 0.9861944884835313, "grad_norm": 1.4848059471126904, "learning_rate": 9.96973756241193e-09, "loss": 0.3578, "step": 27681 }, { "epoch": 0.9862301156100255, "grad_norm": 1.7441939751066609, "learning_rate": 9.918289446027019e-09, "loss": 0.5231, "step": 27682 }, { "epoch": 0.9862657427365196, "grad_norm": 1.4345118981213694, "learning_rate": 9.866974354560966e-09, "loss": 0.4789, "step": 27683 }, { "epoch": 0.9863013698630136, "grad_norm": 1.310948581152908, "learning_rate": 9.815792288699887e-09, "loss": 0.3281, "step": 27684 }, { "epoch": 0.9863369969895078, "grad_norm": 1.312616733187441, "learning_rate": 9.764743249124353e-09, "loss": 0.3748, "step": 27685 }, { "epoch": 0.9863726241160019, "grad_norm": 1.3538055077336388, "learning_rate": 9.713827236513818e-09, "loss": 0.3672, "step": 27686 }, { "epoch": 0.986408251242496, "grad_norm": 1.25191721597672, "learning_rate": 9.663044251546628e-09, "loss": 0.3896, "step": 27687 }, { "epoch": 0.9864438783689902, "grad_norm": 1.6640978166493763, "learning_rate": 9.612394294898908e-09, "loss": 0.5774, "step": 27688 }, { "epoch": 0.9864795054954842, "grad_norm": 1.1190299228317022, "learning_rate": 9.561877367244565e-09, "loss": 0.3802, "step": 27689 }, { "epoch": 0.9865151326219783, "grad_norm": 1.4406324338451106, "learning_rate": 9.511493469256395e-09, "loss": 0.4, "step": 27690 }, { "epoch": 0.9865507597484725, "grad_norm": 1.6952926375640058, "learning_rate": 9.46124260160608e-09, "loss": 0.544, "step": 27691 }, { "epoch": 0.9865863868749666, "grad_norm": 1.3479773301291418, "learning_rate": 9.411124764963087e-09, "loss": 0.3906, "step": 27692 }, { "epoch": 0.9866220140014608, "grad_norm": 1.4404014194592067, "learning_rate": 9.361139959993549e-09, "loss": 0.5075, "step": 27693 }, { "epoch": 0.9866576411279548, "grad_norm": 1.6974578637063766, "learning_rate": 9.311288187362487e-09, "loss": 0.4603, "step": 27694 }, { "epoch": 0.9866932682544489, "grad_norm": 1.1979922632080768, "learning_rate": 9.261569447736041e-09, "loss": 0.4192, "step": 27695 }, { "epoch": 0.9867288953809431, "grad_norm": 1.3543523624819842, "learning_rate": 9.211983741773678e-09, "loss": 0.4791, "step": 27696 }, { "epoch": 0.9867645225074372, "grad_norm": 1.9913665874077247, "learning_rate": 9.162531070138203e-09, "loss": 0.5312, "step": 27697 }, { "epoch": 0.9868001496339313, "grad_norm": 1.31746948369905, "learning_rate": 9.113211433485758e-09, "loss": 0.4132, "step": 27698 }, { "epoch": 0.9868357767604254, "grad_norm": 1.628595598250984, "learning_rate": 9.064024832474704e-09, "loss": 0.5396, "step": 27699 }, { "epoch": 0.9868714038869195, "grad_norm": 1.2602627727135587, "learning_rate": 9.014971267758965e-09, "loss": 0.3846, "step": 27700 }, { "epoch": 0.9869070310134136, "grad_norm": 1.4592497045975297, "learning_rate": 8.96605073999246e-09, "loss": 0.5174, "step": 27701 }, { "epoch": 0.9869426581399078, "grad_norm": 1.5180397369032468, "learning_rate": 8.917263249825781e-09, "loss": 0.4126, "step": 27702 }, { "epoch": 0.9869782852664019, "grad_norm": 1.28490993702572, "learning_rate": 8.868608797909517e-09, "loss": 0.4353, "step": 27703 }, { "epoch": 0.9870139123928959, "grad_norm": 1.3413834288697495, "learning_rate": 8.82008738489093e-09, "loss": 0.3939, "step": 27704 }, { "epoch": 0.9870495395193901, "grad_norm": 1.25833123029633, "learning_rate": 8.771699011416169e-09, "loss": 0.3337, "step": 27705 }, { "epoch": 0.9870851666458842, "grad_norm": 1.3707635302197991, "learning_rate": 8.723443678130272e-09, "loss": 0.3659, "step": 27706 }, { "epoch": 0.9871207937723783, "grad_norm": 1.1456298982885649, "learning_rate": 8.67532138567495e-09, "loss": 0.3822, "step": 27707 }, { "epoch": 0.9871564208988725, "grad_norm": 2.794723135488776, "learning_rate": 8.627332134690802e-09, "loss": 0.4625, "step": 27708 }, { "epoch": 0.9871920480253665, "grad_norm": 1.3386791906400552, "learning_rate": 8.579475925818426e-09, "loss": 0.3961, "step": 27709 }, { "epoch": 0.9872276751518606, "grad_norm": 1.2646963222060448, "learning_rate": 8.531752759692869e-09, "loss": 0.3987, "step": 27710 }, { "epoch": 0.9872633022783548, "grad_norm": 1.2094970728165044, "learning_rate": 8.4841626369514e-09, "loss": 0.3914, "step": 27711 }, { "epoch": 0.9872989294048489, "grad_norm": 1.7598832380289424, "learning_rate": 8.436705558225733e-09, "loss": 0.5163, "step": 27712 }, { "epoch": 0.9873345565313429, "grad_norm": 1.503881780177042, "learning_rate": 8.389381524149808e-09, "loss": 0.6049, "step": 27713 }, { "epoch": 0.9873701836578371, "grad_norm": 1.169127363248067, "learning_rate": 8.342190535353123e-09, "loss": 0.4165, "step": 27714 }, { "epoch": 0.9874058107843312, "grad_norm": 1.3165579435405739, "learning_rate": 8.295132592464062e-09, "loss": 0.3958, "step": 27715 }, { "epoch": 0.9874414379108253, "grad_norm": 1.225078455745394, "learning_rate": 8.248207696108789e-09, "loss": 0.4105, "step": 27716 }, { "epoch": 0.9874770650373195, "grad_norm": 1.2773003511666794, "learning_rate": 8.201415846912364e-09, "loss": 0.4093, "step": 27717 }, { "epoch": 0.9875126921638135, "grad_norm": 1.1738241281826038, "learning_rate": 8.154757045497618e-09, "loss": 0.3328, "step": 27718 }, { "epoch": 0.9875483192903076, "grad_norm": 1.8676998310159403, "learning_rate": 8.10823129248739e-09, "loss": 0.4658, "step": 27719 }, { "epoch": 0.9875839464168018, "grad_norm": 1.2749577856528906, "learning_rate": 8.06183858849896e-09, "loss": 0.4226, "step": 27720 }, { "epoch": 0.9876195735432959, "grad_norm": 1.375966702477949, "learning_rate": 8.015578934150726e-09, "loss": 0.4094, "step": 27721 }, { "epoch": 0.98765520066979, "grad_norm": 1.4997122544419839, "learning_rate": 7.969452330059969e-09, "loss": 0.461, "step": 27722 }, { "epoch": 0.9876908277962841, "grad_norm": 1.5792702713515363, "learning_rate": 7.923458776839532e-09, "loss": 0.4822, "step": 27723 }, { "epoch": 0.9877264549227782, "grad_norm": 1.2589582847346563, "learning_rate": 7.877598275102261e-09, "loss": 0.3744, "step": 27724 }, { "epoch": 0.9877620820492723, "grad_norm": 1.3957544028732816, "learning_rate": 7.831870825458777e-09, "loss": 0.4095, "step": 27725 }, { "epoch": 0.9877977091757665, "grad_norm": 1.26861884072232, "learning_rate": 7.786276428518592e-09, "loss": 0.4148, "step": 27726 }, { "epoch": 0.9878333363022606, "grad_norm": 1.6658584718061276, "learning_rate": 7.740815084887887e-09, "loss": 0.5964, "step": 27727 }, { "epoch": 0.9878689634287546, "grad_norm": 1.5333389551273653, "learning_rate": 7.695486795171735e-09, "loss": 0.5057, "step": 27728 }, { "epoch": 0.9879045905552488, "grad_norm": 1.2741421390817589, "learning_rate": 7.65029155997521e-09, "loss": 0.4073, "step": 27729 }, { "epoch": 0.9879402176817429, "grad_norm": 1.8102339809659544, "learning_rate": 7.605229379898937e-09, "loss": 0.5794, "step": 27730 }, { "epoch": 0.987975844808237, "grad_norm": 1.2652973727442853, "learning_rate": 7.560300255543551e-09, "loss": 0.4218, "step": 27731 }, { "epoch": 0.9880114719347312, "grad_norm": 1.6372135213762755, "learning_rate": 7.51550418750746e-09, "loss": 0.4625, "step": 27732 }, { "epoch": 0.9880470990612252, "grad_norm": 1.1555879851851396, "learning_rate": 7.470841176385746e-09, "loss": 0.3537, "step": 27733 }, { "epoch": 0.9880827261877193, "grad_norm": 1.3028968241578445, "learning_rate": 7.426311222774596e-09, "loss": 0.3953, "step": 27734 }, { "epoch": 0.9881183533142135, "grad_norm": 1.655239441612196, "learning_rate": 7.38191432726687e-09, "loss": 0.4974, "step": 27735 }, { "epoch": 0.9881539804407076, "grad_norm": 1.7261145295886215, "learning_rate": 7.337650490453208e-09, "loss": 0.4307, "step": 27736 }, { "epoch": 0.9881896075672016, "grad_norm": 1.7398406095254835, "learning_rate": 7.293519712923136e-09, "loss": 0.4613, "step": 27737 }, { "epoch": 0.9882252346936958, "grad_norm": 1.3143157185116425, "learning_rate": 7.2495219952639636e-09, "loss": 0.3862, "step": 27738 }, { "epoch": 0.9882608618201899, "grad_norm": 1.6364519197698204, "learning_rate": 7.205657338062999e-09, "loss": 0.5669, "step": 27739 }, { "epoch": 0.988296488946684, "grad_norm": 1.2422851839976656, "learning_rate": 7.161925741901998e-09, "loss": 0.3634, "step": 27740 }, { "epoch": 0.9883321160731782, "grad_norm": 1.596840863276596, "learning_rate": 7.118327207364939e-09, "loss": 0.4978, "step": 27741 }, { "epoch": 0.9883677431996722, "grad_norm": 1.6158834077196642, "learning_rate": 7.074861735032468e-09, "loss": 0.4942, "step": 27742 }, { "epoch": 0.9884033703261663, "grad_norm": 1.485855911112279, "learning_rate": 7.0315293254819e-09, "loss": 0.3988, "step": 27743 }, { "epoch": 0.9884389974526605, "grad_norm": 2.5031109712639315, "learning_rate": 6.988329979291664e-09, "loss": 0.5446, "step": 27744 }, { "epoch": 0.9884746245791546, "grad_norm": 1.4349598162532309, "learning_rate": 6.945263697036853e-09, "loss": 0.4378, "step": 27745 }, { "epoch": 0.9885102517056487, "grad_norm": 1.3966839666530746, "learning_rate": 6.902330479290342e-09, "loss": 0.398, "step": 27746 }, { "epoch": 0.9885458788321428, "grad_norm": 1.3595727072394208, "learning_rate": 6.859530326623898e-09, "loss": 0.4345, "step": 27747 }, { "epoch": 0.9885815059586369, "grad_norm": 1.4902489207762448, "learning_rate": 6.816863239608173e-09, "loss": 0.5437, "step": 27748 }, { "epoch": 0.988617133085131, "grad_norm": 1.7620818588446452, "learning_rate": 6.774329218810494e-09, "loss": 0.503, "step": 27749 }, { "epoch": 0.9886527602116252, "grad_norm": 1.4574458059079134, "learning_rate": 6.7319282647970715e-09, "loss": 0.4046, "step": 27750 }, { "epoch": 0.9886883873381193, "grad_norm": 1.5288739997925906, "learning_rate": 6.6896603781330115e-09, "loss": 0.5445, "step": 27751 }, { "epoch": 0.9887240144646133, "grad_norm": 1.489283773005479, "learning_rate": 6.647525559381196e-09, "loss": 0.5012, "step": 27752 }, { "epoch": 0.9887596415911075, "grad_norm": 1.4683934544910702, "learning_rate": 6.605523809102288e-09, "loss": 0.4082, "step": 27753 }, { "epoch": 0.9887952687176016, "grad_norm": 1.4180165544707657, "learning_rate": 6.563655127855839e-09, "loss": 0.3783, "step": 27754 }, { "epoch": 0.9888308958440957, "grad_norm": 1.636656279320023, "learning_rate": 6.521919516199182e-09, "loss": 0.4118, "step": 27755 }, { "epoch": 0.9888665229705899, "grad_norm": 1.1959677245421165, "learning_rate": 6.480316974688539e-09, "loss": 0.3808, "step": 27756 }, { "epoch": 0.9889021500970839, "grad_norm": 1.4206560656556448, "learning_rate": 6.4388475038779095e-09, "loss": 0.387, "step": 27757 }, { "epoch": 0.988937777223578, "grad_norm": 1.7129914527670052, "learning_rate": 6.3975111043179665e-09, "loss": 0.5743, "step": 27758 }, { "epoch": 0.9889734043500722, "grad_norm": 1.1555279505709208, "learning_rate": 6.356307776560489e-09, "loss": 0.3877, "step": 27759 }, { "epoch": 0.9890090314765663, "grad_norm": 1.2101845679250098, "learning_rate": 6.315237521155038e-09, "loss": 0.4095, "step": 27760 }, { "epoch": 0.9890446586030603, "grad_norm": 1.3225273773192552, "learning_rate": 6.274300338646733e-09, "loss": 0.3991, "step": 27761 }, { "epoch": 0.9890802857295545, "grad_norm": 1.3827788408381654, "learning_rate": 6.233496229580693e-09, "loss": 0.4041, "step": 27762 }, { "epoch": 0.9891159128560486, "grad_norm": 1.4333419952429691, "learning_rate": 6.192825194502039e-09, "loss": 0.471, "step": 27763 }, { "epoch": 0.9891515399825427, "grad_norm": 1.4046501348257623, "learning_rate": 6.152287233950338e-09, "loss": 0.44, "step": 27764 }, { "epoch": 0.9891871671090369, "grad_norm": 1.5861088066287725, "learning_rate": 6.111882348466269e-09, "loss": 0.5527, "step": 27765 }, { "epoch": 0.9892227942355309, "grad_norm": 1.3177191440554552, "learning_rate": 6.07161053858829e-09, "loss": 0.4674, "step": 27766 }, { "epoch": 0.989258421362025, "grad_norm": 1.4797650655239143, "learning_rate": 6.0314718048515295e-09, "loss": 0.5455, "step": 27767 }, { "epoch": 0.9892940484885192, "grad_norm": 1.5854561084664758, "learning_rate": 5.991466147791114e-09, "loss": 0.4767, "step": 27768 }, { "epoch": 0.9893296756150133, "grad_norm": 2.1413631435614984, "learning_rate": 5.951593567939951e-09, "loss": 0.5592, "step": 27769 }, { "epoch": 0.9893653027415074, "grad_norm": 1.3184318969747961, "learning_rate": 5.911854065828726e-09, "loss": 0.4957, "step": 27770 }, { "epoch": 0.9894009298680015, "grad_norm": 1.9365131219022613, "learning_rate": 5.872247641987017e-09, "loss": 0.5493, "step": 27771 }, { "epoch": 0.9894365569944956, "grad_norm": 1.7402338095846024, "learning_rate": 5.832774296941068e-09, "loss": 0.505, "step": 27772 }, { "epoch": 0.9894721841209897, "grad_norm": 1.57143874467575, "learning_rate": 5.793434031218237e-09, "loss": 0.519, "step": 27773 }, { "epoch": 0.9895078112474839, "grad_norm": 1.9852492103185548, "learning_rate": 5.754226845340327e-09, "loss": 0.5813, "step": 27774 }, { "epoch": 0.989543438373978, "grad_norm": 1.4233155680833036, "learning_rate": 5.7151527398302545e-09, "loss": 0.4914, "step": 27775 }, { "epoch": 0.989579065500472, "grad_norm": 1.5173315841959425, "learning_rate": 5.676211715209823e-09, "loss": 0.4981, "step": 27776 }, { "epoch": 0.9896146926269662, "grad_norm": 1.5660268203246448, "learning_rate": 5.637403771995287e-09, "loss": 0.5523, "step": 27777 }, { "epoch": 0.9896503197534603, "grad_norm": 1.4030679878686991, "learning_rate": 5.59872891070512e-09, "loss": 0.4593, "step": 27778 }, { "epoch": 0.9896859468799544, "grad_norm": 1.314590467221124, "learning_rate": 5.560187131852246e-09, "loss": 0.4062, "step": 27779 }, { "epoch": 0.9897215740064486, "grad_norm": 1.2807543038129439, "learning_rate": 5.521778435952918e-09, "loss": 0.4063, "step": 27780 }, { "epoch": 0.9897572011329426, "grad_norm": 1.4215177028984531, "learning_rate": 5.483502823515618e-09, "loss": 0.4306, "step": 27781 }, { "epoch": 0.9897928282594367, "grad_norm": 2.125100738859952, "learning_rate": 5.4453602950510495e-09, "loss": 0.4692, "step": 27782 }, { "epoch": 0.9898284553859309, "grad_norm": 1.5747518841670272, "learning_rate": 5.407350851067694e-09, "loss": 0.6215, "step": 27783 }, { "epoch": 0.989864082512425, "grad_norm": 1.4008266270121612, "learning_rate": 5.369474492071814e-09, "loss": 0.4929, "step": 27784 }, { "epoch": 0.989899709638919, "grad_norm": 1.4916901048714273, "learning_rate": 5.331731218566338e-09, "loss": 0.5133, "step": 27785 }, { "epoch": 0.9899353367654132, "grad_norm": 1.245110956305376, "learning_rate": 5.2941210310553106e-09, "loss": 0.3983, "step": 27786 }, { "epoch": 0.9899709638919073, "grad_norm": 1.3733296947044822, "learning_rate": 5.2566439300383296e-09, "loss": 0.4095, "step": 27787 }, { "epoch": 0.9900065910184014, "grad_norm": 1.3785694103034136, "learning_rate": 5.2192999160161075e-09, "loss": 0.4797, "step": 27788 }, { "epoch": 0.9900422181448956, "grad_norm": 1.452071937584723, "learning_rate": 5.182088989483802e-09, "loss": 0.4716, "step": 27789 }, { "epoch": 0.9900778452713896, "grad_norm": 1.3352845962542779, "learning_rate": 5.145011150937685e-09, "loss": 0.3634, "step": 27790 }, { "epoch": 0.9901134723978837, "grad_norm": 1.58047186407145, "learning_rate": 5.108066400872913e-09, "loss": 0.5268, "step": 27791 }, { "epoch": 0.9901490995243779, "grad_norm": 1.4723887713953099, "learning_rate": 5.071254739780207e-09, "loss": 0.4485, "step": 27792 }, { "epoch": 0.990184726650872, "grad_norm": 1.230664934557044, "learning_rate": 5.034576168149175e-09, "loss": 0.3547, "step": 27793 }, { "epoch": 0.9902203537773661, "grad_norm": 1.6255202475883248, "learning_rate": 4.998030686468314e-09, "loss": 0.5957, "step": 27794 }, { "epoch": 0.9902559809038602, "grad_norm": 1.36238846966918, "learning_rate": 4.961618295225012e-09, "loss": 0.4435, "step": 27795 }, { "epoch": 0.9902916080303543, "grad_norm": 1.3202959848113012, "learning_rate": 4.925338994904438e-09, "loss": 0.4251, "step": 27796 }, { "epoch": 0.9903272351568484, "grad_norm": 1.673793560968326, "learning_rate": 4.8891927859895384e-09, "loss": 0.5617, "step": 27797 }, { "epoch": 0.9903628622833426, "grad_norm": 1.665349089825527, "learning_rate": 4.853179668959928e-09, "loss": 0.5136, "step": 27798 }, { "epoch": 0.9903984894098367, "grad_norm": 1.3984358202607285, "learning_rate": 4.8172996442974464e-09, "loss": 0.4591, "step": 27799 }, { "epoch": 0.9904341165363307, "grad_norm": 1.4606200934736968, "learning_rate": 4.781552712478377e-09, "loss": 0.5289, "step": 27800 }, { "epoch": 0.9904697436628249, "grad_norm": 1.620140483291895, "learning_rate": 4.745938873979006e-09, "loss": 0.5065, "step": 27801 }, { "epoch": 0.990505370789319, "grad_norm": 1.6527121243264127, "learning_rate": 4.7104581292745085e-09, "loss": 0.4853, "step": 27802 }, { "epoch": 0.9905409979158131, "grad_norm": 1.1964710079501437, "learning_rate": 4.6751104788356205e-09, "loss": 0.39, "step": 27803 }, { "epoch": 0.9905766250423073, "grad_norm": 1.2818756078971751, "learning_rate": 4.639895923134186e-09, "loss": 0.3872, "step": 27804 }, { "epoch": 0.9906122521688013, "grad_norm": 1.444053145093934, "learning_rate": 4.6048144626387185e-09, "loss": 0.4185, "step": 27805 }, { "epoch": 0.9906478792952954, "grad_norm": 1.432709785929949, "learning_rate": 4.569866097816622e-09, "loss": 0.4127, "step": 27806 }, { "epoch": 0.9906835064217896, "grad_norm": 1.9519356288042906, "learning_rate": 4.5350508291341914e-09, "loss": 0.4771, "step": 27807 }, { "epoch": 0.9907191335482837, "grad_norm": 1.2569901391779108, "learning_rate": 4.500368657053278e-09, "loss": 0.3772, "step": 27808 }, { "epoch": 0.9907547606747777, "grad_norm": 1.5182577061563793, "learning_rate": 4.465819582035735e-09, "loss": 0.4901, "step": 27809 }, { "epoch": 0.9907903878012719, "grad_norm": 2.962802710489385, "learning_rate": 4.431403604542306e-09, "loss": 0.485, "step": 27810 }, { "epoch": 0.990826014927766, "grad_norm": 1.3490673255450885, "learning_rate": 4.397120725032622e-09, "loss": 0.3907, "step": 27811 }, { "epoch": 0.9908616420542601, "grad_norm": 1.5534207413560752, "learning_rate": 4.362970943960765e-09, "loss": 0.5073, "step": 27812 }, { "epoch": 0.9908972691807543, "grad_norm": 1.4674144027809097, "learning_rate": 4.328954261781926e-09, "loss": 0.4005, "step": 27813 }, { "epoch": 0.9909328963072483, "grad_norm": 1.702013841707887, "learning_rate": 4.2950706789501865e-09, "loss": 0.5014, "step": 27814 }, { "epoch": 0.9909685234337424, "grad_norm": 1.469649039406418, "learning_rate": 4.2613201959162966e-09, "loss": 0.5304, "step": 27815 }, { "epoch": 0.9910041505602366, "grad_norm": 1.5702234383193614, "learning_rate": 4.227702813129897e-09, "loss": 0.4274, "step": 27816 }, { "epoch": 0.9910397776867307, "grad_norm": 2.2723396404731324, "learning_rate": 4.194218531037298e-09, "loss": 0.5168, "step": 27817 }, { "epoch": 0.9910754048132248, "grad_norm": 1.4911159094072226, "learning_rate": 4.1608673500859175e-09, "loss": 0.5539, "step": 27818 }, { "epoch": 0.991111031939719, "grad_norm": 1.3533137170059395, "learning_rate": 4.1276492707187366e-09, "loss": 0.3985, "step": 27819 }, { "epoch": 0.991146659066213, "grad_norm": 1.7666766994375205, "learning_rate": 4.094564293379843e-09, "loss": 0.5082, "step": 27820 }, { "epoch": 0.9911822861927071, "grad_norm": 1.3979499537323867, "learning_rate": 4.0616124185077765e-09, "loss": 0.465, "step": 27821 }, { "epoch": 0.9912179133192013, "grad_norm": 1.5890765916253575, "learning_rate": 4.028793646542184e-09, "loss": 0.5222, "step": 27822 }, { "epoch": 0.9912535404456954, "grad_norm": 1.2363498169596592, "learning_rate": 3.996107977919383e-09, "loss": 0.3803, "step": 27823 }, { "epoch": 0.9912891675721894, "grad_norm": 1.7127526307966594, "learning_rate": 3.963555413075693e-09, "loss": 0.4525, "step": 27824 }, { "epoch": 0.9913247946986836, "grad_norm": 1.940261125867013, "learning_rate": 3.931135952444099e-09, "loss": 0.4177, "step": 27825 }, { "epoch": 0.9913604218251777, "grad_norm": 1.9823958274556337, "learning_rate": 3.898849596456477e-09, "loss": 0.5017, "step": 27826 }, { "epoch": 0.9913960489516718, "grad_norm": 1.3872672490534412, "learning_rate": 3.866696345542487e-09, "loss": 0.3839, "step": 27827 }, { "epoch": 0.991431676078166, "grad_norm": 1.3553638559450918, "learning_rate": 3.8346762001295616e-09, "loss": 0.497, "step": 27828 }, { "epoch": 0.99146730320466, "grad_norm": 1.741722970040602, "learning_rate": 3.802789160645137e-09, "loss": 0.5099, "step": 27829 }, { "epoch": 0.9915029303311541, "grad_norm": 1.4103483338167657, "learning_rate": 3.77103522751443e-09, "loss": 0.454, "step": 27830 }, { "epoch": 0.9915385574576483, "grad_norm": 1.5234427070930623, "learning_rate": 3.739414401158214e-09, "loss": 0.4629, "step": 27831 }, { "epoch": 0.9915741845841424, "grad_norm": 1.410270480167847, "learning_rate": 3.707926681999485e-09, "loss": 0.4873, "step": 27832 }, { "epoch": 0.9916098117106364, "grad_norm": 1.386347742323606, "learning_rate": 3.676572070455686e-09, "loss": 0.4782, "step": 27833 }, { "epoch": 0.9916454388371306, "grad_norm": 1.388115767297751, "learning_rate": 3.6453505669464817e-09, "loss": 0.4368, "step": 27834 }, { "epoch": 0.9916810659636247, "grad_norm": 1.2271508493423684, "learning_rate": 3.614262171885985e-09, "loss": 0.4099, "step": 27835 }, { "epoch": 0.9917166930901188, "grad_norm": 1.2625254798995857, "learning_rate": 3.583306885688309e-09, "loss": 0.4085, "step": 27836 }, { "epoch": 0.991752320216613, "grad_norm": 1.447377902647571, "learning_rate": 3.552484708766457e-09, "loss": 0.5301, "step": 27837 }, { "epoch": 0.991787947343107, "grad_norm": 1.3825590414617985, "learning_rate": 3.5217956415301013e-09, "loss": 0.4317, "step": 27838 }, { "epoch": 0.9918235744696011, "grad_norm": 1.6243567845483917, "learning_rate": 3.491239684387804e-09, "loss": 0.5095, "step": 27839 }, { "epoch": 0.9918592015960953, "grad_norm": 1.133251310141466, "learning_rate": 3.4608168377470165e-09, "loss": 0.4288, "step": 27840 }, { "epoch": 0.9918948287225894, "grad_norm": 1.4143614069987298, "learning_rate": 3.4305271020129705e-09, "loss": 0.4142, "step": 27841 }, { "epoch": 0.9919304558490835, "grad_norm": 1.2531873340438826, "learning_rate": 3.400370477588677e-09, "loss": 0.4014, "step": 27842 }, { "epoch": 0.9919660829755776, "grad_norm": 1.2006128667875855, "learning_rate": 3.3703469648760367e-09, "loss": 0.394, "step": 27843 }, { "epoch": 0.9920017101020717, "grad_norm": 1.2491298820480299, "learning_rate": 3.3404565642736198e-09, "loss": 0.3982, "step": 27844 }, { "epoch": 0.9920373372285658, "grad_norm": 1.8952150559323333, "learning_rate": 3.3106992761799962e-09, "loss": 0.4802, "step": 27845 }, { "epoch": 0.99207296435506, "grad_norm": 1.7459853804923124, "learning_rate": 3.2810751009926255e-09, "loss": 0.5084, "step": 27846 }, { "epoch": 0.9921085914815541, "grad_norm": 1.351817588256008, "learning_rate": 3.2515840391045272e-09, "loss": 0.4933, "step": 27847 }, { "epoch": 0.9921442186080481, "grad_norm": 1.6142715262828318, "learning_rate": 3.22222609090872e-09, "loss": 0.5684, "step": 27848 }, { "epoch": 0.9921798457345423, "grad_norm": 1.517004164847797, "learning_rate": 3.1930012567971123e-09, "loss": 0.5812, "step": 27849 }, { "epoch": 0.9922154728610364, "grad_norm": 1.333695172111566, "learning_rate": 3.1639095371571727e-09, "loss": 0.4463, "step": 27850 }, { "epoch": 0.9922510999875305, "grad_norm": 2.2558247124393094, "learning_rate": 3.1349509323774784e-09, "loss": 0.3831, "step": 27851 }, { "epoch": 0.9922867271140247, "grad_norm": 1.6332743639503258, "learning_rate": 3.1061254428432773e-09, "loss": 0.5626, "step": 27852 }, { "epoch": 0.9923223542405187, "grad_norm": 1.4278687912081982, "learning_rate": 3.0774330689387066e-09, "loss": 0.4236, "step": 27853 }, { "epoch": 0.9923579813670128, "grad_norm": 1.299379226287957, "learning_rate": 3.0488738110456827e-09, "loss": 0.4857, "step": 27854 }, { "epoch": 0.992393608493507, "grad_norm": 1.4437838392118951, "learning_rate": 3.020447669543902e-09, "loss": 0.4169, "step": 27855 }, { "epoch": 0.9924292356200011, "grad_norm": 1.5621966725417635, "learning_rate": 2.992154644811951e-09, "loss": 0.4848, "step": 27856 }, { "epoch": 0.9924648627464951, "grad_norm": 1.5922916272921461, "learning_rate": 2.963994737227305e-09, "loss": 0.4661, "step": 27857 }, { "epoch": 0.9925004898729893, "grad_norm": 1.2193365872187676, "learning_rate": 2.9359679471652193e-09, "loss": 0.3563, "step": 27858 }, { "epoch": 0.9925361169994834, "grad_norm": 1.2728575212937114, "learning_rate": 2.908074274997619e-09, "loss": 0.4291, "step": 27859 }, { "epoch": 0.9925717441259775, "grad_norm": 1.4339859907671648, "learning_rate": 2.8803137210964283e-09, "loss": 0.4014, "step": 27860 }, { "epoch": 0.9926073712524717, "grad_norm": 1.3883128326506324, "learning_rate": 2.852686285831352e-09, "loss": 0.4705, "step": 27861 }, { "epoch": 0.9926429983789657, "grad_norm": 1.5421014862907714, "learning_rate": 2.825191969570984e-09, "loss": 0.3787, "step": 27862 }, { "epoch": 0.9926786255054598, "grad_norm": 1.448612380990659, "learning_rate": 2.7978307726805875e-09, "loss": 0.4998, "step": 27863 }, { "epoch": 0.992714252631954, "grad_norm": 1.6370080724516267, "learning_rate": 2.770602695524316e-09, "loss": 0.5534, "step": 27864 }, { "epoch": 0.9927498797584481, "grad_norm": 1.3359025642355904, "learning_rate": 2.743507738465212e-09, "loss": 0.4322, "step": 27865 }, { "epoch": 0.9927855068849422, "grad_norm": 1.2810146654849723, "learning_rate": 2.7165459018640983e-09, "loss": 0.3741, "step": 27866 }, { "epoch": 0.9928211340114363, "grad_norm": 1.121287559950289, "learning_rate": 2.689717186079577e-09, "loss": 0.4252, "step": 27867 }, { "epoch": 0.9928567611379304, "grad_norm": 2.0625966627156695, "learning_rate": 2.6630215914702495e-09, "loss": 0.5467, "step": 27868 }, { "epoch": 0.9928923882644245, "grad_norm": 1.5009706076984761, "learning_rate": 2.6364591183891675e-09, "loss": 0.5469, "step": 27869 }, { "epoch": 0.9929280153909187, "grad_norm": 1.251413397051268, "learning_rate": 2.610029767191602e-09, "loss": 0.4183, "step": 27870 }, { "epoch": 0.9929636425174128, "grad_norm": 1.1440880979875951, "learning_rate": 2.5837335382294937e-09, "loss": 0.4309, "step": 27871 }, { "epoch": 0.9929992696439068, "grad_norm": 1.361380461237143, "learning_rate": 2.5575704318536732e-09, "loss": 0.4456, "step": 27872 }, { "epoch": 0.993034896770401, "grad_norm": 1.7654989747770404, "learning_rate": 2.53154044841053e-09, "loss": 0.4654, "step": 27873 }, { "epoch": 0.9930705238968951, "grad_norm": 1.459431811333313, "learning_rate": 2.505643588247564e-09, "loss": 0.4959, "step": 27874 }, { "epoch": 0.9931061510233892, "grad_norm": 1.342345064512169, "learning_rate": 2.4798798517100543e-09, "loss": 0.3832, "step": 27875 }, { "epoch": 0.9931417781498834, "grad_norm": 1.3348186469828676, "learning_rate": 2.45424923914106e-09, "loss": 0.39, "step": 27876 }, { "epoch": 0.9931774052763774, "grad_norm": 1.4819638475989505, "learning_rate": 2.4287517508814197e-09, "loss": 0.3558, "step": 27877 }, { "epoch": 0.9932130324028715, "grad_norm": 1.4456573778648658, "learning_rate": 2.403387387270861e-09, "loss": 0.4112, "step": 27878 }, { "epoch": 0.9932486595293657, "grad_norm": 1.5977155321492769, "learning_rate": 2.378156148648003e-09, "loss": 0.4938, "step": 27879 }, { "epoch": 0.9932842866558598, "grad_norm": 1.5256304382665495, "learning_rate": 2.3530580353470223e-09, "loss": 0.458, "step": 27880 }, { "epoch": 0.9933199137823538, "grad_norm": 1.466890880955782, "learning_rate": 2.328093047703206e-09, "loss": 0.5419, "step": 27881 }, { "epoch": 0.993355540908848, "grad_norm": 1.273840709607688, "learning_rate": 2.303261186048511e-09, "loss": 0.3607, "step": 27882 }, { "epoch": 0.9933911680353421, "grad_norm": 1.2446226607679665, "learning_rate": 2.2785624507137837e-09, "loss": 0.3525, "step": 27883 }, { "epoch": 0.9934267951618362, "grad_norm": 1.7177686188161383, "learning_rate": 2.2539968420276503e-09, "loss": 0.4244, "step": 27884 }, { "epoch": 0.9934624222883304, "grad_norm": 1.5281079009499754, "learning_rate": 2.2295643603187366e-09, "loss": 0.5208, "step": 27885 }, { "epoch": 0.9934980494148244, "grad_norm": 1.917004620654206, "learning_rate": 2.2052650059101176e-09, "loss": 0.5275, "step": 27886 }, { "epoch": 0.9935336765413185, "grad_norm": 1.5983278711079405, "learning_rate": 2.181098779127089e-09, "loss": 0.4606, "step": 27887 }, { "epoch": 0.9935693036678127, "grad_norm": 1.2267223447854136, "learning_rate": 2.1570656802905042e-09, "loss": 0.3579, "step": 27888 }, { "epoch": 0.9936049307943068, "grad_norm": 1.3631077939921903, "learning_rate": 2.1331657097201087e-09, "loss": 0.5045, "step": 27889 }, { "epoch": 0.9936405579208009, "grad_norm": 1.436770244649323, "learning_rate": 2.109398867735646e-09, "loss": 0.4462, "step": 27890 }, { "epoch": 0.993676185047295, "grad_norm": 1.3642212215291984, "learning_rate": 2.0857651546524195e-09, "loss": 0.3847, "step": 27891 }, { "epoch": 0.9937118121737891, "grad_norm": 1.4197685360962073, "learning_rate": 2.0622645707857324e-09, "loss": 0.4764, "step": 27892 }, { "epoch": 0.9937474393002832, "grad_norm": 1.1942603515788213, "learning_rate": 2.038897116447558e-09, "loss": 0.3858, "step": 27893 }, { "epoch": 0.9937830664267774, "grad_norm": 1.420830990138024, "learning_rate": 2.015662791948758e-09, "loss": 0.444, "step": 27894 }, { "epoch": 0.9938186935532715, "grad_norm": 1.2596279775031265, "learning_rate": 1.9925615976013058e-09, "loss": 0.3933, "step": 27895 }, { "epoch": 0.9938543206797655, "grad_norm": 1.596343548302204, "learning_rate": 1.969593533709402e-09, "loss": 0.421, "step": 27896 }, { "epoch": 0.9938899478062597, "grad_norm": 2.2343488651257455, "learning_rate": 1.9467586005816887e-09, "loss": 0.391, "step": 27897 }, { "epoch": 0.9939255749327538, "grad_norm": 1.5628013529709854, "learning_rate": 1.9240567985212565e-09, "loss": 0.4668, "step": 27898 }, { "epoch": 0.9939612020592479, "grad_norm": 1.239032052666202, "learning_rate": 1.901488127830087e-09, "loss": 0.4027, "step": 27899 }, { "epoch": 0.9939968291857421, "grad_norm": 1.2932213545055489, "learning_rate": 1.879052588807939e-09, "loss": 0.4337, "step": 27900 }, { "epoch": 0.9940324563122361, "grad_norm": 1.7404886215906195, "learning_rate": 1.8567501817556842e-09, "loss": 0.53, "step": 27901 }, { "epoch": 0.9940680834387302, "grad_norm": 1.3519856769101455, "learning_rate": 1.8345809069686416e-09, "loss": 0.3881, "step": 27902 }, { "epoch": 0.9941037105652244, "grad_norm": 1.7992466464711567, "learning_rate": 1.8125447647421302e-09, "loss": 0.5307, "step": 27903 }, { "epoch": 0.9941393376917185, "grad_norm": 1.3235231238032685, "learning_rate": 1.7906417553703592e-09, "loss": 0.4686, "step": 27904 }, { "epoch": 0.9941749648182125, "grad_norm": 1.246394398277389, "learning_rate": 1.7688718791442072e-09, "loss": 0.3918, "step": 27905 }, { "epoch": 0.9942105919447067, "grad_norm": 1.6707816932806705, "learning_rate": 1.7472351363545525e-09, "loss": 0.5302, "step": 27906 }, { "epoch": 0.9942462190712008, "grad_norm": 1.8369104771687483, "learning_rate": 1.7257315272878327e-09, "loss": 0.4912, "step": 27907 }, { "epoch": 0.9942818461976949, "grad_norm": 1.5209697182727697, "learning_rate": 1.7043610522315956e-09, "loss": 0.4381, "step": 27908 }, { "epoch": 0.9943174733241891, "grad_norm": 1.3549447136423218, "learning_rate": 1.6831237114700582e-09, "loss": 0.4173, "step": 27909 }, { "epoch": 0.9943531004506831, "grad_norm": 1.928487739261514, "learning_rate": 1.6620195052874378e-09, "loss": 0.4521, "step": 27910 }, { "epoch": 0.9943887275771772, "grad_norm": 1.2720605904819784, "learning_rate": 1.6410484339624e-09, "loss": 0.3686, "step": 27911 }, { "epoch": 0.9944243547036714, "grad_norm": 1.2548196616742162, "learning_rate": 1.6202104977747213e-09, "loss": 0.3901, "step": 27912 }, { "epoch": 0.9944599818301655, "grad_norm": 1.3263961875282286, "learning_rate": 1.5995056970030676e-09, "loss": 0.3956, "step": 27913 }, { "epoch": 0.9944956089566596, "grad_norm": 1.4820557690333225, "learning_rate": 1.5789340319227741e-09, "loss": 0.4867, "step": 27914 }, { "epoch": 0.9945312360831537, "grad_norm": 1.9670560829178925, "learning_rate": 1.5584955028069559e-09, "loss": 0.5409, "step": 27915 }, { "epoch": 0.9945668632096478, "grad_norm": 1.7451376327057333, "learning_rate": 1.5381901099287278e-09, "loss": 0.5799, "step": 27916 }, { "epoch": 0.9946024903361419, "grad_norm": 1.5564551521367505, "learning_rate": 1.5180178535567636e-09, "loss": 0.5032, "step": 27917 }, { "epoch": 0.9946381174626361, "grad_norm": 1.3877627012107174, "learning_rate": 1.497978733961958e-09, "loss": 0.3755, "step": 27918 }, { "epoch": 0.9946737445891302, "grad_norm": 1.146295685962316, "learning_rate": 1.4780727514096538e-09, "loss": 0.393, "step": 27919 }, { "epoch": 0.9947093717156242, "grad_norm": 1.422466923431044, "learning_rate": 1.458299906166305e-09, "loss": 0.3752, "step": 27920 }, { "epoch": 0.9947449988421184, "grad_norm": 1.5281121434332459, "learning_rate": 1.438660198493924e-09, "loss": 0.5255, "step": 27921 }, { "epoch": 0.9947806259686125, "grad_norm": 1.4551616541908095, "learning_rate": 1.4191536286534135e-09, "loss": 0.4997, "step": 27922 }, { "epoch": 0.9948162530951066, "grad_norm": 1.278006521748705, "learning_rate": 1.3997801969056757e-09, "loss": 0.4128, "step": 27923 }, { "epoch": 0.9948518802216008, "grad_norm": 1.648573310861182, "learning_rate": 1.3805399035093926e-09, "loss": 0.5433, "step": 27924 }, { "epoch": 0.9948875073480948, "grad_norm": 1.2198086108191653, "learning_rate": 1.361432748718805e-09, "loss": 0.3852, "step": 27925 }, { "epoch": 0.9949231344745889, "grad_norm": 1.5716381007512545, "learning_rate": 1.342458732790375e-09, "loss": 0.4437, "step": 27926 }, { "epoch": 0.9949587616010831, "grad_norm": 3.140522435194786, "learning_rate": 1.3236178559750124e-09, "loss": 0.4642, "step": 27927 }, { "epoch": 0.9949943887275772, "grad_norm": 1.6151613201741306, "learning_rate": 1.3049101185247382e-09, "loss": 0.4778, "step": 27928 }, { "epoch": 0.9950300158540712, "grad_norm": 1.329491544632615, "learning_rate": 1.2863355206882422e-09, "loss": 0.4219, "step": 27929 }, { "epoch": 0.9950656429805654, "grad_norm": 1.2929785417404887, "learning_rate": 1.2678940627131042e-09, "loss": 0.3801, "step": 27930 }, { "epoch": 0.9951012701070595, "grad_norm": 1.59014395418493, "learning_rate": 1.2495857448435733e-09, "loss": 0.5412, "step": 27931 }, { "epoch": 0.9951368972335536, "grad_norm": 1.4283605469456355, "learning_rate": 1.2314105673250088e-09, "loss": 0.4178, "step": 27932 }, { "epoch": 0.9951725243600478, "grad_norm": 1.2958636664238363, "learning_rate": 1.213368530399439e-09, "loss": 0.4018, "step": 27933 }, { "epoch": 0.9952081514865418, "grad_norm": 1.5864404850533684, "learning_rate": 1.1954596343066726e-09, "loss": 0.5006, "step": 27934 }, { "epoch": 0.9952437786130359, "grad_norm": 1.4149915808407294, "learning_rate": 1.177683879284297e-09, "loss": 0.4865, "step": 27935 }, { "epoch": 0.9952794057395301, "grad_norm": 1.24046059808583, "learning_rate": 1.1600412655710102e-09, "loss": 0.3824, "step": 27936 }, { "epoch": 0.9953150328660242, "grad_norm": 1.5213459450693196, "learning_rate": 1.1425317933988488e-09, "loss": 0.4816, "step": 27937 }, { "epoch": 0.9953506599925183, "grad_norm": 1.7726000787426273, "learning_rate": 1.1251554630042905e-09, "loss": 0.539, "step": 27938 }, { "epoch": 0.9953862871190124, "grad_norm": 1.5077689235106142, "learning_rate": 1.1079122746160408e-09, "loss": 0.5346, "step": 27939 }, { "epoch": 0.9954219142455065, "grad_norm": 1.8414516207881113, "learning_rate": 1.0908022284650266e-09, "loss": 0.4037, "step": 27940 }, { "epoch": 0.9954575413720006, "grad_norm": 1.4603331020977464, "learning_rate": 1.0738253247788433e-09, "loss": 0.4207, "step": 27941 }, { "epoch": 0.9954931684984948, "grad_norm": 1.339790939302868, "learning_rate": 1.0569815637839765e-09, "loss": 0.4174, "step": 27942 }, { "epoch": 0.9955287956249889, "grad_norm": 1.1974033146338592, "learning_rate": 1.040270945703581e-09, "loss": 0.3653, "step": 27943 }, { "epoch": 0.9955644227514829, "grad_norm": 1.5151008025393153, "learning_rate": 1.0236934707608115e-09, "loss": 0.4918, "step": 27944 }, { "epoch": 0.9956000498779771, "grad_norm": 1.6941956545028167, "learning_rate": 1.0072491391766026e-09, "loss": 0.5788, "step": 27945 }, { "epoch": 0.9956356770044712, "grad_norm": 1.3112620944937619, "learning_rate": 9.909379511696682e-10, "loss": 0.3831, "step": 27946 }, { "epoch": 0.9956713041309653, "grad_norm": 1.273336924623399, "learning_rate": 9.74759906957612e-10, "loss": 0.4125, "step": 27947 }, { "epoch": 0.9957069312574595, "grad_norm": 1.5927972241630608, "learning_rate": 9.587150067558172e-10, "loss": 0.5523, "step": 27948 }, { "epoch": 0.9957425583839535, "grad_norm": 1.8624291956248056, "learning_rate": 9.428032507774465e-10, "loss": 0.4956, "step": 27949 }, { "epoch": 0.9957781855104476, "grad_norm": 1.648499028901532, "learning_rate": 9.270246392334426e-10, "loss": 0.5516, "step": 27950 }, { "epoch": 0.9958138126369418, "grad_norm": 1.3427071539709114, "learning_rate": 9.113791723358578e-10, "loss": 0.377, "step": 27951 }, { "epoch": 0.9958494397634359, "grad_norm": 1.448965161347921, "learning_rate": 8.95866850292304e-10, "loss": 0.4135, "step": 27952 }, { "epoch": 0.99588506688993, "grad_norm": 1.876463346381855, "learning_rate": 8.804876733092826e-10, "loss": 0.5121, "step": 27953 }, { "epoch": 0.9959206940164241, "grad_norm": 1.292901758585503, "learning_rate": 8.652416415921849e-10, "loss": 0.451, "step": 27954 }, { "epoch": 0.9959563211429182, "grad_norm": 1.3135672751750136, "learning_rate": 8.501287553419613e-10, "loss": 0.4391, "step": 27955 }, { "epoch": 0.9959919482694123, "grad_norm": 1.3333451057496253, "learning_rate": 8.351490147617824e-10, "loss": 0.4027, "step": 27956 }, { "epoch": 0.9960275753959065, "grad_norm": 1.5094125747435876, "learning_rate": 8.203024200514887e-10, "loss": 0.4601, "step": 27957 }, { "epoch": 0.9960632025224005, "grad_norm": 1.6109297878039355, "learning_rate": 8.05588971406479e-10, "loss": 0.524, "step": 27958 }, { "epoch": 0.9960988296488946, "grad_norm": 1.4018014123285454, "learning_rate": 7.910086690254836e-10, "loss": 0.3965, "step": 27959 }, { "epoch": 0.9961344567753888, "grad_norm": 1.3782740015497055, "learning_rate": 7.76561513100571e-10, "loss": 0.417, "step": 27960 }, { "epoch": 0.9961700839018829, "grad_norm": 1.3105497783399338, "learning_rate": 7.622475038238097e-10, "loss": 0.4052, "step": 27961 }, { "epoch": 0.996205711028377, "grad_norm": 1.7002795666244321, "learning_rate": 7.480666413883786e-10, "loss": 0.5118, "step": 27962 }, { "epoch": 0.9962413381548711, "grad_norm": 1.4314332052589345, "learning_rate": 7.340189259807951e-10, "loss": 0.3769, "step": 27963 }, { "epoch": 0.9962769652813652, "grad_norm": 1.242204315912951, "learning_rate": 7.201043577886868e-10, "loss": 0.4077, "step": 27964 }, { "epoch": 0.9963125924078593, "grad_norm": 1.7189287424032946, "learning_rate": 7.063229369974611e-10, "loss": 0.4879, "step": 27965 }, { "epoch": 0.9963482195343535, "grad_norm": 1.356698030660596, "learning_rate": 6.926746637903048e-10, "loss": 0.4191, "step": 27966 }, { "epoch": 0.9963838466608476, "grad_norm": 1.4697502579619597, "learning_rate": 6.791595383504046e-10, "loss": 0.5334, "step": 27967 }, { "epoch": 0.9964194737873416, "grad_norm": 1.2295652656272225, "learning_rate": 6.657775608553962e-10, "loss": 0.4126, "step": 27968 }, { "epoch": 0.9964551009138358, "grad_norm": 1.4251924564523077, "learning_rate": 6.525287314851358e-10, "loss": 0.41, "step": 27969 }, { "epoch": 0.9964907280403299, "grad_norm": 1.3257821769336322, "learning_rate": 6.394130504161489e-10, "loss": 0.4001, "step": 27970 }, { "epoch": 0.996526355166824, "grad_norm": 1.6080409464327243, "learning_rate": 6.264305178216301e-10, "loss": 0.554, "step": 27971 }, { "epoch": 0.9965619822933182, "grad_norm": 1.2408834129154434, "learning_rate": 6.135811338758846e-10, "loss": 0.4331, "step": 27972 }, { "epoch": 0.9965976094198122, "grad_norm": 1.438259712982687, "learning_rate": 6.008648987487764e-10, "loss": 0.4924, "step": 27973 }, { "epoch": 0.9966332365463063, "grad_norm": 1.1937672210459795, "learning_rate": 5.882818126112799e-10, "loss": 0.362, "step": 27974 }, { "epoch": 0.9966688636728005, "grad_norm": 1.308747703800505, "learning_rate": 5.758318756299286e-10, "loss": 0.4435, "step": 27975 }, { "epoch": 0.9967044907992946, "grad_norm": 1.4870960833872233, "learning_rate": 5.635150879701456e-10, "loss": 0.5289, "step": 27976 }, { "epoch": 0.9967401179257886, "grad_norm": 1.5829991668522518, "learning_rate": 5.513314497962441e-10, "loss": 0.4669, "step": 27977 }, { "epoch": 0.9967757450522828, "grad_norm": 1.5749004276690595, "learning_rate": 5.392809612703165e-10, "loss": 0.5599, "step": 27978 }, { "epoch": 0.9968113721787769, "grad_norm": 1.6134460818046854, "learning_rate": 5.273636225533452e-10, "loss": 0.4835, "step": 27979 }, { "epoch": 0.996846999305271, "grad_norm": 1.3187855133537074, "learning_rate": 5.15579433804092e-10, "loss": 0.38, "step": 27980 }, { "epoch": 0.9968826264317652, "grad_norm": 1.250620914036116, "learning_rate": 5.039283951790985e-10, "loss": 0.4139, "step": 27981 }, { "epoch": 0.9969182535582592, "grad_norm": 1.2521189688169923, "learning_rate": 4.924105068326857e-10, "loss": 0.3977, "step": 27982 }, { "epoch": 0.9969538806847533, "grad_norm": 1.9358319775365365, "learning_rate": 4.810257689202847e-10, "loss": 0.4281, "step": 27983 }, { "epoch": 0.9969895078112475, "grad_norm": 1.273167993507193, "learning_rate": 4.697741815906654e-10, "loss": 0.4111, "step": 27984 }, { "epoch": 0.9970251349377416, "grad_norm": 1.4438638072263452, "learning_rate": 4.5865574499703855e-10, "loss": 0.4242, "step": 27985 }, { "epoch": 0.9970607620642357, "grad_norm": 1.554068281488271, "learning_rate": 4.476704592848435e-10, "loss": 0.4759, "step": 27986 }, { "epoch": 0.9970963891907298, "grad_norm": 1.5392143401272511, "learning_rate": 4.368183246006297e-10, "loss": 0.5149, "step": 27987 }, { "epoch": 0.9971320163172239, "grad_norm": 1.2605416017963764, "learning_rate": 4.260993410898362e-10, "loss": 0.3684, "step": 27988 }, { "epoch": 0.997167643443718, "grad_norm": 1.7645846546221255, "learning_rate": 4.1551350889568185e-10, "loss": 0.507, "step": 27989 }, { "epoch": 0.9972032705702122, "grad_norm": 1.2725873236487741, "learning_rate": 4.050608281569446e-10, "loss": 0.3909, "step": 27990 }, { "epoch": 0.9972388976967063, "grad_norm": 1.7723759869648874, "learning_rate": 3.947412990157329e-10, "loss": 0.5521, "step": 27991 }, { "epoch": 0.9972745248232003, "grad_norm": 1.3731490595926554, "learning_rate": 3.8455492160638374e-10, "loss": 0.3791, "step": 27992 }, { "epoch": 0.9973101519496945, "grad_norm": 1.5182218532461842, "learning_rate": 3.745016960665648e-10, "loss": 0.4707, "step": 27993 }, { "epoch": 0.9973457790761886, "grad_norm": 1.1839611405614305, "learning_rate": 3.6458162252950293e-10, "loss": 0.3519, "step": 27994 }, { "epoch": 0.9973814062026827, "grad_norm": 1.4765407578371927, "learning_rate": 3.547947011273145e-10, "loss": 0.5041, "step": 27995 }, { "epoch": 0.9974170333291769, "grad_norm": 1.097022340931579, "learning_rate": 3.4514093199100597e-10, "loss": 0.4043, "step": 27996 }, { "epoch": 0.9974526604556709, "grad_norm": 1.4548828847128594, "learning_rate": 3.356203152471427e-10, "loss": 0.4925, "step": 27997 }, { "epoch": 0.997488287582165, "grad_norm": 1.6172596897453164, "learning_rate": 3.2623285102451054e-10, "loss": 0.5308, "step": 27998 }, { "epoch": 0.9975239147086592, "grad_norm": 1.8925166956144475, "learning_rate": 3.169785394474545e-10, "loss": 0.6043, "step": 27999 }, { "epoch": 0.9975595418351533, "grad_norm": 1.3699691680711605, "learning_rate": 3.078573806392093e-10, "loss": 0.4145, "step": 28000 }, { "epoch": 0.9975951689616473, "grad_norm": 1.228397033000187, "learning_rate": 2.988693747218996e-10, "loss": 0.4084, "step": 28001 }, { "epoch": 0.9976307960881415, "grad_norm": 1.1976539077491621, "learning_rate": 2.900145218143191e-10, "loss": 0.4207, "step": 28002 }, { "epoch": 0.9976664232146356, "grad_norm": 1.233341804685219, "learning_rate": 2.812928220330413e-10, "loss": 0.4178, "step": 28003 }, { "epoch": 0.9977020503411297, "grad_norm": 1.5993038617600461, "learning_rate": 2.7270427549686006e-10, "loss": 0.6139, "step": 28004 }, { "epoch": 0.9977376774676239, "grad_norm": 1.5652093755871879, "learning_rate": 2.6424888231901813e-10, "loss": 0.4224, "step": 28005 }, { "epoch": 0.997773304594118, "grad_norm": 1.261122257652784, "learning_rate": 2.5592664261275826e-10, "loss": 0.3822, "step": 28006 }, { "epoch": 0.997808931720612, "grad_norm": 1.2800091567214125, "learning_rate": 2.4773755648688226e-10, "loss": 0.412, "step": 28007 }, { "epoch": 0.9978445588471062, "grad_norm": 1.323366139102996, "learning_rate": 2.3968162405352267e-10, "loss": 0.3807, "step": 28008 }, { "epoch": 0.9978801859736003, "grad_norm": 1.1568154162125799, "learning_rate": 2.317588454170405e-10, "loss": 0.3854, "step": 28009 }, { "epoch": 0.9979158131000944, "grad_norm": 1.5972061448731734, "learning_rate": 2.2396922068401717e-10, "loss": 0.4897, "step": 28010 }, { "epoch": 0.9979514402265885, "grad_norm": 1.3270129606494936, "learning_rate": 2.1631274995770336e-10, "loss": 0.3972, "step": 28011 }, { "epoch": 0.9979870673530826, "grad_norm": 1.828646320476949, "learning_rate": 2.0878943334246005e-10, "loss": 0.542, "step": 28012 }, { "epoch": 0.9980226944795767, "grad_norm": 1.921239733834689, "learning_rate": 2.0139927093487666e-10, "loss": 0.4627, "step": 28013 }, { "epoch": 0.9980583216060709, "grad_norm": 1.4016631481485513, "learning_rate": 1.9414226283598348e-10, "loss": 0.4049, "step": 28014 }, { "epoch": 0.998093948732565, "grad_norm": 1.7764834168204935, "learning_rate": 1.8701840914125967e-10, "loss": 0.5573, "step": 28015 }, { "epoch": 0.998129575859059, "grad_norm": 2.8561837939239787, "learning_rate": 1.8002770994618446e-10, "loss": 0.4795, "step": 28016 }, { "epoch": 0.9981652029855532, "grad_norm": 1.4003775910237288, "learning_rate": 1.7317016534290632e-10, "loss": 0.446, "step": 28017 }, { "epoch": 0.9982008301120473, "grad_norm": 1.4400549763690882, "learning_rate": 1.6644577542357376e-10, "loss": 0.4486, "step": 28018 }, { "epoch": 0.9982364572385414, "grad_norm": 1.3599046998747444, "learning_rate": 1.5985454027811488e-10, "loss": 0.3905, "step": 28019 }, { "epoch": 0.9982720843650356, "grad_norm": 1.7840150669265664, "learning_rate": 1.5339645999312701e-10, "loss": 0.5055, "step": 28020 }, { "epoch": 0.9983077114915296, "grad_norm": 1.9414476352846914, "learning_rate": 1.470715346540974e-10, "loss": 0.4149, "step": 28021 }, { "epoch": 0.9983433386180237, "grad_norm": 1.3413352042795847, "learning_rate": 1.408797643476234e-10, "loss": 0.4278, "step": 28022 }, { "epoch": 0.9983789657445179, "grad_norm": 1.2683390880863843, "learning_rate": 1.3482114915475132e-10, "loss": 0.3558, "step": 28023 }, { "epoch": 0.998414592871012, "grad_norm": 1.3126070155976826, "learning_rate": 1.2889568915541716e-10, "loss": 0.4015, "step": 28024 }, { "epoch": 0.998450219997506, "grad_norm": 1.248584895562639, "learning_rate": 1.2310338443066728e-10, "loss": 0.4348, "step": 28025 }, { "epoch": 0.9984858471240002, "grad_norm": 1.7101696186322644, "learning_rate": 1.1744423505488657e-10, "loss": 0.5114, "step": 28026 }, { "epoch": 0.9985214742504943, "grad_norm": 1.2473049620414696, "learning_rate": 1.1191824110579063e-10, "loss": 0.424, "step": 28027 }, { "epoch": 0.9985571013769884, "grad_norm": 1.1826211238408588, "learning_rate": 1.0652540265554401e-10, "loss": 0.3754, "step": 28028 }, { "epoch": 0.9985927285034826, "grad_norm": 2.0753303101810583, "learning_rate": 1.0126571977631117e-10, "loss": 0.4913, "step": 28029 }, { "epoch": 0.9986283556299766, "grad_norm": 1.4713269978537509, "learning_rate": 9.613919253914639e-11, "loss": 0.3845, "step": 28030 }, { "epoch": 0.9986639827564707, "grad_norm": 1.5518570253845532, "learning_rate": 9.114582101066304e-11, "loss": 0.5, "step": 28031 }, { "epoch": 0.9986996098829649, "grad_norm": 1.3123178056689546, "learning_rate": 8.628560525747454e-11, "loss": 0.4752, "step": 28032 }, { "epoch": 0.998735237009459, "grad_norm": 1.4683245274484724, "learning_rate": 8.155854534619424e-11, "loss": 0.5115, "step": 28033 }, { "epoch": 0.9987708641359531, "grad_norm": 1.2250067052425944, "learning_rate": 7.696464133788439e-11, "loss": 0.3573, "step": 28034 }, { "epoch": 0.9988064912624472, "grad_norm": 1.6812687526223344, "learning_rate": 7.250389329471752e-11, "loss": 0.5737, "step": 28035 }, { "epoch": 0.9988421183889413, "grad_norm": 1.629069781813763, "learning_rate": 6.81763012744252e-11, "loss": 0.4524, "step": 28036 }, { "epoch": 0.9988777455154354, "grad_norm": 1.6204652132069557, "learning_rate": 6.39818653369595e-11, "loss": 0.5051, "step": 28037 }, { "epoch": 0.9989133726419296, "grad_norm": 1.5490536183735644, "learning_rate": 5.99205855367213e-11, "loss": 0.526, "step": 28038 }, { "epoch": 0.9989489997684237, "grad_norm": 1.3076598245445874, "learning_rate": 5.599246192922181e-11, "loss": 0.4112, "step": 28039 }, { "epoch": 0.9989846268949177, "grad_norm": 1.3460373428951329, "learning_rate": 5.219749456442102e-11, "loss": 0.4042, "step": 28040 }, { "epoch": 0.9990202540214119, "grad_norm": 1.5956724735898062, "learning_rate": 4.8535683494499443e-11, "loss": 0.3985, "step": 28041 }, { "epoch": 0.999055881147906, "grad_norm": 1.2991188488183836, "learning_rate": 4.500702876830687e-11, "loss": 0.4115, "step": 28042 }, { "epoch": 0.9990915082744001, "grad_norm": 1.4329581913992286, "learning_rate": 4.161153043136246e-11, "loss": 0.5344, "step": 28043 }, { "epoch": 0.9991271354008943, "grad_norm": 1.9539127338155924, "learning_rate": 3.834918853140579e-11, "loss": 0.5473, "step": 28044 }, { "epoch": 0.9991627625273883, "grad_norm": 1.546055893429037, "learning_rate": 3.5220003109515124e-11, "loss": 0.5439, "step": 28045 }, { "epoch": 0.9991983896538824, "grad_norm": 1.585833708675475, "learning_rate": 3.222397420787893e-11, "loss": 0.5867, "step": 28046 }, { "epoch": 0.9992340167803766, "grad_norm": 1.518282998970716, "learning_rate": 2.936110186757546e-11, "loss": 0.4674, "step": 28047 }, { "epoch": 0.9992696439068707, "grad_norm": 1.9103160251267877, "learning_rate": 2.6631386125242077e-11, "loss": 0.5628, "step": 28048 }, { "epoch": 0.9993052710333648, "grad_norm": 1.465962353192507, "learning_rate": 2.403482701862636e-11, "loss": 0.5403, "step": 28049 }, { "epoch": 0.9993408981598589, "grad_norm": 1.2132913466338047, "learning_rate": 2.1571424581035005e-11, "loss": 0.3913, "step": 28050 }, { "epoch": 0.999376525286353, "grad_norm": 1.2724506372702986, "learning_rate": 1.92411788457747e-11, "loss": 0.396, "step": 28051 }, { "epoch": 0.9994121524128471, "grad_norm": 1.7207798077859873, "learning_rate": 1.7044089843931687e-11, "loss": 0.5744, "step": 28052 }, { "epoch": 0.9994477795393413, "grad_norm": 1.4812662045838558, "learning_rate": 1.498015760437177e-11, "loss": 0.4696, "step": 28053 }, { "epoch": 0.9994834066658353, "grad_norm": 1.2424117941917356, "learning_rate": 1.3049382155960744e-11, "loss": 0.4032, "step": 28054 }, { "epoch": 0.9995190337923294, "grad_norm": 1.498170733982071, "learning_rate": 1.1251763522013292e-11, "loss": 0.4968, "step": 28055 }, { "epoch": 0.9995546609188236, "grad_norm": 1.2049894514373196, "learning_rate": 9.587301728064546e-12, "loss": 0.3824, "step": 28056 }, { "epoch": 0.9995902880453177, "grad_norm": 1.2545205199519087, "learning_rate": 8.055996797429189e-12, "loss": 0.4065, "step": 28057 }, { "epoch": 0.9996259151718118, "grad_norm": 1.768990780580661, "learning_rate": 6.657848747870788e-12, "loss": 0.453, "step": 28058 }, { "epoch": 0.999661542298306, "grad_norm": 1.3168817386416531, "learning_rate": 5.392857599373358e-12, "loss": 0.5437, "step": 28059 }, { "epoch": 0.9996971694248, "grad_norm": 1.179438040385728, "learning_rate": 4.261023368590245e-12, "loss": 0.3834, "step": 28060 }, { "epoch": 0.9997327965512941, "grad_norm": 1.6800797005272772, "learning_rate": 3.2623460710645706e-12, "loss": 0.5189, "step": 28061 }, { "epoch": 0.9997684236777883, "grad_norm": 1.3716486142126754, "learning_rate": 2.3968257190087885e-12, "loss": 0.4077, "step": 28062 }, { "epoch": 0.9998040508042824, "grad_norm": 1.1995513295520743, "learning_rate": 1.664462325745575e-12, "loss": 0.3663, "step": 28063 }, { "epoch": 0.9998396779307764, "grad_norm": 1.5936406404583026, "learning_rate": 1.0652558990464913e-12, "loss": 0.5156, "step": 28064 }, { "epoch": 0.9998753050572706, "grad_norm": 1.3025952943284516, "learning_rate": 5.992064477933212e-13, "loss": 0.399, "step": 28065 }, { "epoch": 0.9999109321837647, "grad_norm": 1.491873173436856, "learning_rate": 2.6631397864740336e-13, "loss": 0.4451, "step": 28066 }, { "epoch": 0.9999465593102588, "grad_norm": 1.0451174509811707, "learning_rate": 6.65784949394066e-14, "loss": 0.3636, "step": 28067 }, { "epoch": 0.999982186436753, "grad_norm": 1.377110986485588, "learning_rate": 0.0, "loss": 0.4067, "step": 28068 }, { "epoch": 0.999982186436753, "step": 28068, "total_flos": 5396477423622144.0, "train_loss": 0.0, "train_runtime": 4.0993, "train_samples_per_second": 876438.349, "train_steps_per_second": 6847.06 } ], "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": 5396477423622144.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }