{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9991031390134529,
  "eval_steps": 500,
  "global_step": 557,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002810468997013877,
      "grad_norm": 0.08777584135532379,
      "learning_rate": 1.111111111111111e-06,
      "loss": 1.6284,
      "step": 1
    },
    {
      "epoch": 0.005620937994027754,
      "grad_norm": 0.08739772439002991,
      "learning_rate": 2.222222222222222e-06,
      "loss": 1.6386,
      "step": 2
    },
    {
      "epoch": 0.00843140699104163,
      "grad_norm": 0.07890280336141586,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.6366,
      "step": 3
    },
    {
      "epoch": 0.011241875988055507,
      "grad_norm": 0.050849832594394684,
      "learning_rate": 4.444444444444444e-06,
      "loss": 1.6428,
      "step": 4
    },
    {
      "epoch": 0.014052344985069383,
      "grad_norm": 0.026262454688549042,
      "learning_rate": 5.555555555555557e-06,
      "loss": 1.6335,
      "step": 5
    },
    {
      "epoch": 0.01686281398208326,
      "grad_norm": 0.017007650807499886,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.6298,
      "step": 6
    },
    {
      "epoch": 0.019673282979097137,
      "grad_norm": 0.021000387147068977,
      "learning_rate": 7.77777777777778e-06,
      "loss": 1.6349,
      "step": 7
    },
    {
      "epoch": 0.022483751976111015,
      "grad_norm": 0.039422884583473206,
      "learning_rate": 8.888888888888888e-06,
      "loss": 1.6161,
      "step": 8
    },
    {
      "epoch": 0.02529422097312489,
      "grad_norm": 0.03416357561945915,
      "learning_rate": 1e-05,
      "loss": 1.6133,
      "step": 9
    },
    {
      "epoch": 0.028104689970138767,
      "grad_norm": 0.023096244782209396,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 1.6017,
      "step": 10
    },
    {
      "epoch": 0.030915158967152644,
      "grad_norm": 0.028580009937286377,
      "learning_rate": 1.2222222222222224e-05,
      "loss": 1.6117,
      "step": 11
    },
    {
      "epoch": 0.03372562796416652,
      "grad_norm": 0.02169407717883587,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.605,
      "step": 12
    },
    {
      "epoch": 0.036536096961180396,
      "grad_norm": 0.02079104818403721,
      "learning_rate": 1.4444444444444446e-05,
      "loss": 1.6006,
      "step": 13
    },
    {
      "epoch": 0.039346565958194274,
      "grad_norm": 0.02266255021095276,
      "learning_rate": 1.555555555555556e-05,
      "loss": 1.6076,
      "step": 14
    },
    {
      "epoch": 0.04215703495520815,
      "grad_norm": 0.023310977965593338,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.6051,
      "step": 15
    },
    {
      "epoch": 0.04496750395222203,
      "grad_norm": 0.02227545529603958,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 1.5976,
      "step": 16
    },
    {
      "epoch": 0.04777797294923591,
      "grad_norm": 0.02008313126862049,
      "learning_rate": 1.888888888888889e-05,
      "loss": 1.6048,
      "step": 17
    },
    {
      "epoch": 0.05058844194624978,
      "grad_norm": 0.020431598648428917,
      "learning_rate": 2e-05,
      "loss": 1.6033,
      "step": 18
    },
    {
      "epoch": 0.053398910943263656,
      "grad_norm": 0.019984332844614983,
      "learning_rate": 1.999956548296958e-05,
      "loss": 1.5915,
      "step": 19
    },
    {
      "epoch": 0.05620937994027753,
      "grad_norm": 0.020108483731746674,
      "learning_rate": 1.9998261969639324e-05,
      "loss": 1.5938,
      "step": 20
    },
    {
      "epoch": 0.05901984893729141,
      "grad_norm": 0.016152869910001755,
      "learning_rate": 1.9996089573288985e-05,
      "loss": 1.5931,
      "step": 21
    },
    {
      "epoch": 0.06183031793430529,
      "grad_norm": 0.016522705554962158,
      "learning_rate": 1.99930484827072e-05,
      "loss": 1.5953,
      "step": 22
    },
    {
      "epoch": 0.06464078693131917,
      "grad_norm": 0.017014645040035248,
      "learning_rate": 1.9989138962175105e-05,
      "loss": 1.5898,
      "step": 23
    },
    {
      "epoch": 0.06745125592833304,
      "grad_norm": 0.016037074849009514,
      "learning_rate": 1.9984361351443343e-05,
      "loss": 1.5904,
      "step": 24
    },
    {
      "epoch": 0.07026172492534692,
      "grad_norm": 0.01631341129541397,
      "learning_rate": 1.9978716065702566e-05,
      "loss": 1.5913,
      "step": 25
    },
    {
      "epoch": 0.07307219392236079,
      "grad_norm": 0.014392613433301449,
      "learning_rate": 1.9972203595547334e-05,
      "loss": 1.5893,
      "step": 26
    },
    {
      "epoch": 0.07588266291937468,
      "grad_norm": 0.015151307918131351,
      "learning_rate": 1.996482450693348e-05,
      "loss": 1.5912,
      "step": 27
    },
    {
      "epoch": 0.07869313191638855,
      "grad_norm": 0.015482740476727486,
      "learning_rate": 1.9956579441128942e-05,
      "loss": 1.5847,
      "step": 28
    },
    {
      "epoch": 0.08150360091340242,
      "grad_norm": 0.01364535465836525,
      "learning_rate": 1.994746911465802e-05,
      "loss": 1.585,
      "step": 29
    },
    {
      "epoch": 0.0843140699104163,
      "grad_norm": 0.014720232225954533,
      "learning_rate": 1.9937494319239112e-05,
      "loss": 1.5773,
      "step": 30
    },
    {
      "epoch": 0.08712453890743017,
      "grad_norm": 0.012587211094796658,
      "learning_rate": 1.9926655921715924e-05,
      "loss": 1.5766,
      "step": 31
    },
    {
      "epoch": 0.08993500790444406,
      "grad_norm": 0.013409728184342384,
      "learning_rate": 1.9914954863982106e-05,
      "loss": 1.5764,
      "step": 32
    },
    {
      "epoch": 0.09274547690145793,
      "grad_norm": 0.013652725145220757,
      "learning_rate": 1.990239216289944e-05,
      "loss": 1.5818,
      "step": 33
    },
    {
      "epoch": 0.09555594589847181,
      "grad_norm": 0.015371325425803661,
      "learning_rate": 1.9888968910209433e-05,
      "loss": 1.5834,
      "step": 34
    },
    {
      "epoch": 0.09836641489548568,
      "grad_norm": 0.013734452426433563,
      "learning_rate": 1.9874686272438467e-05,
      "loss": 1.5719,
      "step": 35
    },
    {
      "epoch": 0.10117688389249956,
      "grad_norm": 0.012110204435884953,
      "learning_rate": 1.9859545490796414e-05,
      "loss": 1.5797,
      "step": 36
    },
    {
      "epoch": 0.10398735288951344,
      "grad_norm": 0.013209005817770958,
      "learning_rate": 1.9843547881068763e-05,
      "loss": 1.5789,
      "step": 37
    },
    {
      "epoch": 0.10679782188652731,
      "grad_norm": 0.01261916570365429,
      "learning_rate": 1.9826694833502295e-05,
      "loss": 1.5716,
      "step": 38
    },
    {
      "epoch": 0.1096082908835412,
      "grad_norm": 0.012995808385312557,
      "learning_rate": 1.9808987812684247e-05,
      "loss": 1.5682,
      "step": 39
    },
    {
      "epoch": 0.11241875988055507,
      "grad_norm": 0.013399253599345684,
      "learning_rate": 1.979042835741503e-05,
      "loss": 1.5711,
      "step": 40
    },
    {
      "epoch": 0.11522922887756895,
      "grad_norm": 0.012696914374828339,
      "learning_rate": 1.9771018080574534e-05,
      "loss": 1.5742,
      "step": 41
    },
    {
      "epoch": 0.11803969787458282,
      "grad_norm": 0.011595464311540127,
      "learning_rate": 1.9750758668981925e-05,
      "loss": 1.581,
      "step": 42
    },
    {
      "epoch": 0.12085016687159669,
      "grad_norm": 0.01301959715783596,
      "learning_rate": 1.9729651883249075e-05,
      "loss": 1.5821,
      "step": 43
    },
    {
      "epoch": 0.12366063586861058,
      "grad_norm": 0.012804425321519375,
      "learning_rate": 1.9707699557627554e-05,
      "loss": 1.5921,
      "step": 44
    },
    {
      "epoch": 0.12647110486562446,
      "grad_norm": 0.012021846137940884,
      "learning_rate": 1.968490359984923e-05,
      "loss": 1.5795,
      "step": 45
    },
    {
      "epoch": 0.12928157386263833,
      "grad_norm": 0.013476583175361156,
      "learning_rate": 1.9661265990960486e-05,
      "loss": 1.5755,
      "step": 46
    },
    {
      "epoch": 0.1320920428596522,
      "grad_norm": 0.011448990553617477,
      "learning_rate": 1.9636788785150037e-05,
      "loss": 1.5688,
      "step": 47
    },
    {
      "epoch": 0.13490251185666607,
      "grad_norm": 0.014088047668337822,
      "learning_rate": 1.9611474109570446e-05,
      "loss": 1.5785,
      "step": 48
    },
    {
      "epoch": 0.13771298085367994,
      "grad_norm": 0.013553822413086891,
      "learning_rate": 1.9585324164153236e-05,
      "loss": 1.5812,
      "step": 49
    },
    {
      "epoch": 0.14052344985069384,
      "grad_norm": 0.011343201622366905,
      "learning_rate": 1.9558341221417744e-05,
      "loss": 1.5778,
      "step": 50
    },
    {
      "epoch": 0.14333391884770771,
      "grad_norm": 0.01123133860528469,
      "learning_rate": 1.9530527626273592e-05,
      "loss": 1.5758,
      "step": 51
    },
    {
      "epoch": 0.14614438784472158,
      "grad_norm": 0.01259944774210453,
      "learning_rate": 1.9501885795816937e-05,
      "loss": 1.5731,
      "step": 52
    },
    {
      "epoch": 0.14895485684173546,
      "grad_norm": 0.011086889542639256,
      "learning_rate": 1.9472418219120403e-05,
      "loss": 1.576,
      "step": 53
    },
    {
      "epoch": 0.15176532583874935,
      "grad_norm": 0.011971558444201946,
      "learning_rate": 1.9442127457016768e-05,
      "loss": 1.571,
      "step": 54
    },
    {
      "epoch": 0.15457579483576322,
      "grad_norm": 0.012076422572135925,
      "learning_rate": 1.9411016141876438e-05,
      "loss": 1.5743,
      "step": 55
    },
    {
      "epoch": 0.1573862638327771,
      "grad_norm": 0.01147051528096199,
      "learning_rate": 1.9379086977378664e-05,
      "loss": 1.5701,
      "step": 56
    },
    {
      "epoch": 0.16019673282979097,
      "grad_norm": 0.013713942840695381,
      "learning_rate": 1.9346342738276593e-05,
      "loss": 1.5632,
      "step": 57
    },
    {
      "epoch": 0.16300720182680484,
      "grad_norm": 0.012207642197608948,
      "learning_rate": 1.9312786270156135e-05,
      "loss": 1.5684,
      "step": 58
    },
    {
      "epoch": 0.16581767082381874,
      "grad_norm": 0.01164440531283617,
      "learning_rate": 1.927842048918867e-05,
      "loss": 1.572,
      "step": 59
    },
    {
      "epoch": 0.1686281398208326,
      "grad_norm": 0.014730863273143768,
      "learning_rate": 1.9243248381877605e-05,
      "loss": 1.5737,
      "step": 60
    },
    {
      "epoch": 0.17143860881784648,
      "grad_norm": 0.01201706100255251,
      "learning_rate": 1.9207273004798873e-05,
      "loss": 1.5708,
      "step": 61
    },
    {
      "epoch": 0.17424907781486035,
      "grad_norm": 0.012422306463122368,
      "learning_rate": 1.9170497484335276e-05,
      "loss": 1.5666,
      "step": 62
    },
    {
      "epoch": 0.17705954681187422,
      "grad_norm": 0.014150385744869709,
      "learning_rate": 1.9132925016404805e-05,
      "loss": 1.5652,
      "step": 63
    },
    {
      "epoch": 0.17987001580888812,
      "grad_norm": 0.01331349741667509,
      "learning_rate": 1.9094558866182892e-05,
      "loss": 1.5613,
      "step": 64
    },
    {
      "epoch": 0.182680484805902,
      "grad_norm": 0.014323413372039795,
      "learning_rate": 1.9055402367818673e-05,
      "loss": 1.5643,
      "step": 65
    },
    {
      "epoch": 0.18549095380291586,
      "grad_norm": 0.010913478210568428,
      "learning_rate": 1.901545892414523e-05,
      "loss": 1.5768,
      "step": 66
    },
    {
      "epoch": 0.18830142279992973,
      "grad_norm": 0.013672353699803352,
      "learning_rate": 1.897473200638386e-05,
      "loss": 1.5707,
      "step": 67
    },
    {
      "epoch": 0.19111189179694363,
      "grad_norm": 0.011649075895547867,
      "learning_rate": 1.8933225153842446e-05,
      "loss": 1.5678,
      "step": 68
    },
    {
      "epoch": 0.1939223607939575,
      "grad_norm": 0.012527153827250004,
      "learning_rate": 1.8890941973607843e-05,
      "loss": 1.5637,
      "step": 69
    },
    {
      "epoch": 0.19673282979097137,
      "grad_norm": 0.013441166840493679,
      "learning_rate": 1.8847886140232438e-05,
      "loss": 1.5543,
      "step": 70
    },
    {
      "epoch": 0.19954329878798524,
      "grad_norm": 0.014678705483675003,
      "learning_rate": 1.8804061395414795e-05,
      "loss": 1.5594,
      "step": 71
    },
    {
      "epoch": 0.2023537677849991,
      "grad_norm": 0.015578769147396088,
      "learning_rate": 1.875947154767452e-05,
      "loss": 1.5766,
      "step": 72
    },
    {
      "epoch": 0.205164236782013,
      "grad_norm": 0.013635441660881042,
      "learning_rate": 1.8714120472021252e-05,
      "loss": 1.5622,
      "step": 73
    },
    {
      "epoch": 0.20797470577902688,
      "grad_norm": 0.011884918436408043,
      "learning_rate": 1.8668012109617933e-05,
      "loss": 1.5539,
      "step": 74
    },
    {
      "epoch": 0.21078517477604075,
      "grad_norm": 0.013244202360510826,
      "learning_rate": 1.862115046743831e-05,
      "loss": 1.5641,
      "step": 75
    },
    {
      "epoch": 0.21359564377305462,
      "grad_norm": 0.011851202696561813,
      "learning_rate": 1.85735396179187e-05,
      "loss": 1.554,
      "step": 76
    },
    {
      "epoch": 0.2164061127700685,
      "grad_norm": 0.012839299626648426,
      "learning_rate": 1.8525183698604098e-05,
      "loss": 1.5663,
      "step": 77
    },
    {
      "epoch": 0.2192165817670824,
      "grad_norm": 0.013278639875352383,
      "learning_rate": 1.8476086911788588e-05,
      "loss": 1.5674,
      "step": 78
    },
    {
      "epoch": 0.22202705076409626,
      "grad_norm": 0.011714858934283257,
      "learning_rate": 1.8426253524150176e-05,
      "loss": 1.5628,
      "step": 79
    },
    {
      "epoch": 0.22483751976111013,
      "grad_norm": 0.01387378666549921,
      "learning_rate": 1.8375687866379988e-05,
      "loss": 1.5746,
      "step": 80
    },
    {
      "epoch": 0.227647988758124,
      "grad_norm": 0.014316901564598083,
      "learning_rate": 1.8324394332805913e-05,
      "loss": 1.56,
      "step": 81
    },
    {
      "epoch": 0.2304584577551379,
      "grad_norm": 0.013246218673884869,
      "learning_rate": 1.8272377381010726e-05,
      "loss": 1.5604,
      "step": 82
    },
    {
      "epoch": 0.23326892675215177,
      "grad_norm": 0.014117361046373844,
      "learning_rate": 1.8219641531444713e-05,
      "loss": 1.5535,
      "step": 83
    },
    {
      "epoch": 0.23607939574916564,
      "grad_norm": 0.013312868773937225,
      "learning_rate": 1.8166191367032828e-05,
      "loss": 1.5571,
      "step": 84
    },
    {
      "epoch": 0.23888986474617951,
      "grad_norm": 0.012200911529362202,
      "learning_rate": 1.811203153277641e-05,
      "loss": 1.5636,
      "step": 85
    },
    {
      "epoch": 0.24170033374319339,
      "grad_norm": 0.012060770764946938,
      "learning_rate": 1.8057166735349533e-05,
      "loss": 1.5548,
      "step": 86
    },
    {
      "epoch": 0.24451080274020728,
      "grad_norm": 0.01396957691758871,
      "learning_rate": 1.800160174268996e-05,
      "loss": 1.566,
      "step": 87
    },
    {
      "epoch": 0.24732127173722115,
      "grad_norm": 0.017935393378138542,
      "learning_rate": 1.7945341383584818e-05,
      "loss": 1.5705,
      "step": 88
    },
    {
      "epoch": 0.25013174073423505,
      "grad_norm": 0.01408262737095356,
      "learning_rate": 1.7888390547250944e-05,
      "loss": 1.5701,
      "step": 89
    },
    {
      "epoch": 0.2529422097312489,
      "grad_norm": 0.014612286351621151,
      "learning_rate": 1.7830754182909985e-05,
      "loss": 1.56,
      "step": 90
    },
    {
      "epoch": 0.2557526787282628,
      "grad_norm": 0.016053492203354836,
      "learning_rate": 1.7772437299358324e-05,
      "loss": 1.5655,
      "step": 91
    },
    {
      "epoch": 0.25856314772527667,
      "grad_norm": 0.014997400343418121,
      "learning_rate": 1.771344496453177e-05,
      "loss": 1.5692,
      "step": 92
    },
    {
      "epoch": 0.26137361672229054,
      "grad_norm": 0.013961107470095158,
      "learning_rate": 1.7653782305065158e-05,
      "loss": 1.56,
      "step": 93
    },
    {
      "epoch": 0.2641840857193044,
      "grad_norm": 0.015686677768826485,
      "learning_rate": 1.7593454505846807e-05,
      "loss": 1.5561,
      "step": 94
    },
    {
      "epoch": 0.2669945547163183,
      "grad_norm": 0.013397088274359703,
      "learning_rate": 1.753246680956795e-05,
      "loss": 1.5651,
      "step": 95
    },
    {
      "epoch": 0.26980502371333215,
      "grad_norm": 0.014148705638945103,
      "learning_rate": 1.7470824516267125e-05,
      "loss": 1.5538,
      "step": 96
    },
    {
      "epoch": 0.272615492710346,
      "grad_norm": 0.012879609130322933,
      "learning_rate": 1.7408532982869573e-05,
      "loss": 1.5592,
      "step": 97
    },
    {
      "epoch": 0.2754259617073599,
      "grad_norm": 0.012351011857390404,
      "learning_rate": 1.7345597622721727e-05,
      "loss": 1.5633,
      "step": 98
    },
    {
      "epoch": 0.2782364307043738,
      "grad_norm": 0.010927798226475716,
      "learning_rate": 1.7282023905120743e-05,
      "loss": 1.5623,
      "step": 99
    },
    {
      "epoch": 0.2810468997013877,
      "grad_norm": 0.015191658399999142,
      "learning_rate": 1.721781735483921e-05,
      "loss": 1.5585,
      "step": 100
    },
    {
      "epoch": 0.28385736869840156,
      "grad_norm": 0.01361748855561018,
      "learning_rate": 1.7152983551645054e-05,
      "loss": 1.5553,
      "step": 101
    },
    {
      "epoch": 0.28666783769541543,
      "grad_norm": 0.01628255844116211,
      "learning_rate": 1.708752812981659e-05,
      "loss": 1.5681,
      "step": 102
    },
    {
      "epoch": 0.2894783066924293,
      "grad_norm": 0.011448734439909458,
      "learning_rate": 1.702145677765293e-05,
      "loss": 1.5619,
      "step": 103
    },
    {
      "epoch": 0.29228877568944317,
      "grad_norm": 0.0159872155636549,
      "learning_rate": 1.6954775236979616e-05,
      "loss": 1.5528,
      "step": 104
    },
    {
      "epoch": 0.29509924468645704,
      "grad_norm": 0.014284605160355568,
      "learning_rate": 1.6887489302649657e-05,
      "loss": 1.547,
      "step": 105
    },
    {
      "epoch": 0.2979097136834709,
      "grad_norm": 0.014883718453347683,
      "learning_rate": 1.6819604822039924e-05,
      "loss": 1.5608,
      "step": 106
    },
    {
      "epoch": 0.3007201826804848,
      "grad_norm": 0.01631166599690914,
      "learning_rate": 1.6751127694543012e-05,
      "loss": 1.561,
      "step": 107
    },
    {
      "epoch": 0.3035306516774987,
      "grad_norm": 0.014855324290692806,
      "learning_rate": 1.6682063871054534e-05,
      "loss": 1.557,
      "step": 108
    },
    {
      "epoch": 0.3063411206745126,
      "grad_norm": 0.014950740151107311,
      "learning_rate": 1.661241935345599e-05,
      "loss": 1.5686,
      "step": 109
    },
    {
      "epoch": 0.30915158967152645,
      "grad_norm": 0.015889016911387444,
      "learning_rate": 1.654220019409317e-05,
      "loss": 1.5525,
      "step": 110
    },
    {
      "epoch": 0.3119620586685403,
      "grad_norm": 0.012233362533152103,
      "learning_rate": 1.6471412495250195e-05,
      "loss": 1.5586,
      "step": 111
    },
    {
      "epoch": 0.3147725276655542,
      "grad_norm": 0.013401877135038376,
      "learning_rate": 1.640006240861921e-05,
      "loss": 1.5645,
      "step": 112
    },
    {
      "epoch": 0.31758299666256806,
      "grad_norm": 0.012674611061811447,
      "learning_rate": 1.632815613476576e-05,
      "loss": 1.5564,
      "step": 113
    },
    {
      "epoch": 0.32039346565958193,
      "grad_norm": 0.013244451023638248,
      "learning_rate": 1.6255699922589968e-05,
      "loss": 1.5554,
      "step": 114
    },
    {
      "epoch": 0.3232039346565958,
      "grad_norm": 0.013073718175292015,
      "learning_rate": 1.6182700068783463e-05,
      "loss": 1.5561,
      "step": 115
    },
    {
      "epoch": 0.3260144036536097,
      "grad_norm": 0.012644422240555286,
      "learning_rate": 1.610916291728218e-05,
      "loss": 1.5538,
      "step": 116
    },
    {
      "epoch": 0.32882487265062355,
      "grad_norm": 0.013174659572541714,
      "learning_rate": 1.6035094858715065e-05,
      "loss": 1.5509,
      "step": 117
    },
    {
      "epoch": 0.33163534164763747,
      "grad_norm": 0.015327691100537777,
      "learning_rate": 1.5960502329848683e-05,
      "loss": 1.558,
      "step": 118
    },
    {
      "epoch": 0.33444581064465134,
      "grad_norm": 0.011017784476280212,
      "learning_rate": 1.588539181302786e-05,
      "loss": 1.5614,
      "step": 119
    },
    {
      "epoch": 0.3372562796416652,
      "grad_norm": 0.015872148796916008,
      "learning_rate": 1.580976983561235e-05,
      "loss": 1.5557,
      "step": 120
    },
    {
      "epoch": 0.3400667486386791,
      "grad_norm": 0.01222875714302063,
      "learning_rate": 1.5733642969409553e-05,
      "loss": 1.5603,
      "step": 121
    },
    {
      "epoch": 0.34287721763569295,
      "grad_norm": 0.015869170427322388,
      "learning_rate": 1.5657017830103448e-05,
      "loss": 1.5581,
      "step": 122
    },
    {
      "epoch": 0.3456876866327068,
      "grad_norm": 0.01336450781673193,
      "learning_rate": 1.5579901076679625e-05,
      "loss": 1.5663,
      "step": 123
    },
    {
      "epoch": 0.3484981556297207,
      "grad_norm": 0.015786062926054,
      "learning_rate": 1.5502299410846626e-05,
      "loss": 1.5518,
      "step": 124
    },
    {
      "epoch": 0.35130862462673457,
      "grad_norm": 0.01679217629134655,
      "learning_rate": 1.5424219576453526e-05,
      "loss": 1.5535,
      "step": 125
    },
    {
      "epoch": 0.35411909362374844,
      "grad_norm": 0.011585685424506664,
      "learning_rate": 1.5345668358903886e-05,
      "loss": 1.5715,
      "step": 126
    },
    {
      "epoch": 0.35692956262076236,
      "grad_norm": 0.015714086592197418,
      "learning_rate": 1.5266652584566056e-05,
      "loss": 1.5489,
      "step": 127
    },
    {
      "epoch": 0.35974003161777623,
      "grad_norm": 0.013761353679001331,
      "learning_rate": 1.5187179120179969e-05,
      "loss": 1.5515,
      "step": 128
    },
    {
      "epoch": 0.3625505006147901,
      "grad_norm": 0.013305050320923328,
      "learning_rate": 1.5107254872260366e-05,
      "loss": 1.5596,
      "step": 129
    },
    {
      "epoch": 0.365360969611804,
      "grad_norm": 0.013329329900443554,
      "learning_rate": 1.5026886786496624e-05,
      "loss": 1.5488,
      "step": 130
    },
    {
      "epoch": 0.36817143860881785,
      "grad_norm": 0.014221809804439545,
      "learning_rate": 1.4946081847149134e-05,
      "loss": 1.5536,
      "step": 131
    },
    {
      "epoch": 0.3709819076058317,
      "grad_norm": 0.011903174221515656,
      "learning_rate": 1.4864847076442358e-05,
      "loss": 1.5612,
      "step": 132
    },
    {
      "epoch": 0.3737923766028456,
      "grad_norm": 0.012618706561625004,
      "learning_rate": 1.4783189533954555e-05,
      "loss": 1.5578,
      "step": 133
    },
    {
      "epoch": 0.37660284559985946,
      "grad_norm": 0.012923737987875938,
      "learning_rate": 1.4701116316004307e-05,
      "loss": 1.5619,
      "step": 134
    },
    {
      "epoch": 0.37941331459687333,
      "grad_norm": 0.012171992100775242,
      "learning_rate": 1.46186345550338e-05,
      "loss": 1.5596,
      "step": 135
    },
    {
      "epoch": 0.38222378359388726,
      "grad_norm": 0.014030322432518005,
      "learning_rate": 1.4535751418989e-05,
      "loss": 1.5541,
      "step": 136
    },
    {
      "epoch": 0.3850342525909011,
      "grad_norm": 0.012220495380461216,
      "learning_rate": 1.4452474110696738e-05,
      "loss": 1.55,
      "step": 137
    },
    {
      "epoch": 0.387844721587915,
      "grad_norm": 0.013919586315751076,
      "learning_rate": 1.4368809867238754e-05,
      "loss": 1.5607,
      "step": 138
    },
    {
      "epoch": 0.39065519058492887,
      "grad_norm": 0.011982360854744911,
      "learning_rate": 1.4284765959322772e-05,
      "loss": 1.5464,
      "step": 139
    },
    {
      "epoch": 0.39346565958194274,
      "grad_norm": 0.012515905313193798,
      "learning_rate": 1.4200349690650654e-05,
      "loss": 1.5645,
      "step": 140
    },
    {
      "epoch": 0.3962761285789566,
      "grad_norm": 0.011968614533543587,
      "learning_rate": 1.411556839728367e-05,
      "loss": 1.5559,
      "step": 141
    },
    {
      "epoch": 0.3990865975759705,
      "grad_norm": 0.010865562595427036,
      "learning_rate": 1.4030429447004992e-05,
      "loss": 1.5617,
      "step": 142
    },
    {
      "epoch": 0.40189706657298435,
      "grad_norm": 0.013639074750244617,
      "learning_rate": 1.3944940238679384e-05,
      "loss": 1.5469,
      "step": 143
    },
    {
      "epoch": 0.4047075355699982,
      "grad_norm": 0.0128669748082757,
      "learning_rate": 1.3859108201610236e-05,
      "loss": 1.5466,
      "step": 144
    },
    {
      "epoch": 0.4075180045670121,
      "grad_norm": 0.011727554723620415,
      "learning_rate": 1.3772940794893916e-05,
      "loss": 1.5523,
      "step": 145
    },
    {
      "epoch": 0.410328473564026,
      "grad_norm": 0.015109344385564327,
      "learning_rate": 1.368644550677157e-05,
      "loss": 1.5554,
      "step": 146
    },
    {
      "epoch": 0.4131389425610399,
      "grad_norm": 0.013233068399131298,
      "learning_rate": 1.3599629853978342e-05,
      "loss": 1.5594,
      "step": 147
    },
    {
      "epoch": 0.41594941155805376,
      "grad_norm": 0.014392644166946411,
      "learning_rate": 1.3512501381090158e-05,
      "loss": 1.5603,
      "step": 148
    },
    {
      "epoch": 0.41875988055506763,
      "grad_norm": 0.011739206500351429,
      "learning_rate": 1.3425067659868084e-05,
      "loss": 1.5508,
      "step": 149
    },
    {
      "epoch": 0.4215703495520815,
      "grad_norm": 0.016567695885896683,
      "learning_rate": 1.3337336288600297e-05,
      "loss": 1.5529,
      "step": 150
    },
    {
      "epoch": 0.4243808185490954,
      "grad_norm": 0.01213662326335907,
      "learning_rate": 1.324931489144178e-05,
      "loss": 1.5468,
      "step": 151
    },
    {
      "epoch": 0.42719128754610924,
      "grad_norm": 0.01404926273971796,
      "learning_rate": 1.3161011117751756e-05,
      "loss": 1.5564,
      "step": 152
    },
    {
      "epoch": 0.4300017565431231,
      "grad_norm": 0.013470892794430256,
      "learning_rate": 1.3072432641428931e-05,
      "loss": 1.5541,
      "step": 153
    },
    {
      "epoch": 0.432812225540137,
      "grad_norm": 0.01260072086006403,
      "learning_rate": 1.2983587160244602e-05,
      "loss": 1.5453,
      "step": 154
    },
    {
      "epoch": 0.4356226945371509,
      "grad_norm": 0.012238179333508015,
      "learning_rate": 1.2894482395173695e-05,
      "loss": 1.5541,
      "step": 155
    },
    {
      "epoch": 0.4384331635341648,
      "grad_norm": 0.014110546559095383,
      "learning_rate": 1.2805126089723798e-05,
      "loss": 1.5606,
      "step": 156
    },
    {
      "epoch": 0.44124363253117865,
      "grad_norm": 0.011262672953307629,
      "learning_rate": 1.2715526009262209e-05,
      "loss": 1.5603,
      "step": 157
    },
    {
      "epoch": 0.4440541015281925,
      "grad_norm": 0.014026992954313755,
      "learning_rate": 1.2625689940341102e-05,
      "loss": 1.5462,
      "step": 158
    },
    {
      "epoch": 0.4468645705252064,
      "grad_norm": 0.011127503588795662,
      "learning_rate": 1.2535625690020861e-05,
      "loss": 1.5497,
      "step": 159
    },
    {
      "epoch": 0.44967503952222027,
      "grad_norm": 0.014035338535904884,
      "learning_rate": 1.24453410851916e-05,
      "loss": 1.5551,
      "step": 160
    },
    {
      "epoch": 0.45248550851923414,
      "grad_norm": 0.01174125261604786,
      "learning_rate": 1.2354843971892998e-05,
      "loss": 1.5494,
      "step": 161
    },
    {
      "epoch": 0.455295977516248,
      "grad_norm": 0.013476992025971413,
      "learning_rate": 1.2264142214632441e-05,
      "loss": 1.5622,
      "step": 162
    },
    {
      "epoch": 0.4581064465132619,
      "grad_norm": 0.012524358928203583,
      "learning_rate": 1.2173243695701575e-05,
      "loss": 1.5512,
      "step": 163
    },
    {
      "epoch": 0.4609169155102758,
      "grad_norm": 0.011893150396645069,
      "learning_rate": 1.2082156314491298e-05,
      "loss": 1.5601,
      "step": 164
    },
    {
      "epoch": 0.4637273845072897,
      "grad_norm": 0.011717031709849834,
      "learning_rate": 1.1990887986805295e-05,
      "loss": 1.5401,
      "step": 165
    },
    {
      "epoch": 0.46653785350430355,
      "grad_norm": 0.01049530878663063,
      "learning_rate": 1.1899446644172106e-05,
      "loss": 1.5434,
      "step": 166
    },
    {
      "epoch": 0.4693483225013174,
      "grad_norm": 0.013684769161045551,
      "learning_rate": 1.1807840233155863e-05,
      "loss": 1.5473,
      "step": 167
    },
    {
      "epoch": 0.4721587914983313,
      "grad_norm": 0.010303209535777569,
      "learning_rate": 1.1716076714665701e-05,
      "loss": 1.5534,
      "step": 168
    },
    {
      "epoch": 0.47496926049534516,
      "grad_norm": 0.013134041801095009,
      "learning_rate": 1.1624164063263931e-05,
      "loss": 1.54,
      "step": 169
    },
    {
      "epoch": 0.47777972949235903,
      "grad_norm": 0.011170011013746262,
      "learning_rate": 1.1532110266473026e-05,
      "loss": 1.5519,
      "step": 170
    },
    {
      "epoch": 0.4805901984893729,
      "grad_norm": 0.01223431620746851,
      "learning_rate": 1.1439923324081465e-05,
      "loss": 1.5461,
      "step": 171
    },
    {
      "epoch": 0.48340066748638677,
      "grad_norm": 0.01103509496897459,
      "learning_rate": 1.1347611247448544e-05,
      "loss": 1.5469,
      "step": 172
    },
    {
      "epoch": 0.48621113648340064,
      "grad_norm": 0.012365833856165409,
      "learning_rate": 1.1255182058808143e-05,
      "loss": 1.5546,
      "step": 173
    },
    {
      "epoch": 0.48902160548041457,
      "grad_norm": 0.011856972239911556,
      "learning_rate": 1.1162643790571574e-05,
      "loss": 1.5493,
      "step": 174
    },
    {
      "epoch": 0.49183207447742844,
      "grad_norm": 0.011247235350310802,
      "learning_rate": 1.1070004484629543e-05,
      "loss": 1.5485,
      "step": 175
    },
    {
      "epoch": 0.4946425434744423,
      "grad_norm": 0.010246271267533302,
      "learning_rate": 1.0977272191653272e-05,
      "loss": 1.5385,
      "step": 176
    },
    {
      "epoch": 0.4974530124714562,
      "grad_norm": 0.011663992889225483,
      "learning_rate": 1.0884454970394871e-05,
      "loss": 1.5543,
      "step": 177
    },
    {
      "epoch": 0.5002634814684701,
      "grad_norm": 0.011617216281592846,
      "learning_rate": 1.0791560886987016e-05,
      "loss": 1.5621,
      "step": 178
    },
    {
      "epoch": 0.503073950465484,
      "grad_norm": 0.010127879679203033,
      "learning_rate": 1.069859801424196e-05,
      "loss": 1.547,
      "step": 179
    },
    {
      "epoch": 0.5058844194624978,
      "grad_norm": 0.01151992380619049,
      "learning_rate": 1.0605574430949983e-05,
      "loss": 1.5496,
      "step": 180
    },
    {
      "epoch": 0.5086948884595117,
      "grad_norm": 0.012575927190482616,
      "learning_rate": 1.0512498221177319e-05,
      "loss": 1.5523,
      "step": 181
    },
    {
      "epoch": 0.5115053574565256,
      "grad_norm": 0.011552478186786175,
      "learning_rate": 1.0419377473563621e-05,
      "loss": 1.5537,
      "step": 182
    },
    {
      "epoch": 0.5143158264535395,
      "grad_norm": 0.010599706321954727,
      "learning_rate": 1.0326220280619036e-05,
      "loss": 1.5448,
      "step": 183
    },
    {
      "epoch": 0.5171262954505533,
      "grad_norm": 0.010817637667059898,
      "learning_rate": 1.0233034738020933e-05,
      "loss": 1.5503,
      "step": 184
    },
    {
      "epoch": 0.5199367644475672,
      "grad_norm": 0.012649599462747574,
      "learning_rate": 1.0139828943910358e-05,
      "loss": 1.5486,
      "step": 185
    },
    {
      "epoch": 0.5227472334445811,
      "grad_norm": 0.012253242544829845,
      "learning_rate": 1.004661099818829e-05,
      "loss": 1.5514,
      "step": 186
    },
    {
      "epoch": 0.5255577024415949,
      "grad_norm": 0.011790635995566845,
      "learning_rate": 9.953389001811716e-06,
      "loss": 1.5518,
      "step": 187
    },
    {
      "epoch": 0.5283681714386088,
      "grad_norm": 0.013327023014426231,
      "learning_rate": 9.860171056089646e-06,
      "loss": 1.5424,
      "step": 188
    },
    {
      "epoch": 0.5311786404356227,
      "grad_norm": 0.0112162996083498,
      "learning_rate": 9.766965261979072e-06,
      "loss": 1.5488,
      "step": 189
    },
    {
      "epoch": 0.5339891094326366,
      "grad_norm": 0.012566052377223969,
      "learning_rate": 9.673779719380967e-06,
      "loss": 1.5452,
      "step": 190
    },
    {
      "epoch": 0.5367995784296504,
      "grad_norm": 0.012440058402717113,
      "learning_rate": 9.580622526436382e-06,
      "loss": 1.5457,
      "step": 191
    },
    {
      "epoch": 0.5396100474266643,
      "grad_norm": 0.011921185068786144,
      "learning_rate": 9.487501778822685e-06,
      "loss": 1.5413,
      "step": 192
    },
    {
      "epoch": 0.5424205164236782,
      "grad_norm": 0.011896064504981041,
      "learning_rate": 9.394425569050018e-06,
      "loss": 1.5423,
      "step": 193
    },
    {
      "epoch": 0.545230985420692,
      "grad_norm": 0.01354946568608284,
      "learning_rate": 9.30140198575804e-06,
      "loss": 1.5523,
      "step": 194
    },
    {
      "epoch": 0.5480414544177059,
      "grad_norm": 0.011999576352536678,
      "learning_rate": 9.208439113012984e-06,
      "loss": 1.5367,
      "step": 195
    },
    {
      "epoch": 0.5508519234147198,
      "grad_norm": 0.012224970385432243,
      "learning_rate": 9.115545029605129e-06,
      "loss": 1.5564,
      "step": 196
    },
    {
      "epoch": 0.5536623924117338,
      "grad_norm": 0.012914243154227734,
      "learning_rate": 9.022727808346731e-06,
      "loss": 1.5426,
      "step": 197
    },
    {
      "epoch": 0.5564728614087476,
      "grad_norm": 0.012274730019271374,
      "learning_rate": 8.92999551537046e-06,
      "loss": 1.5454,
      "step": 198
    },
    {
      "epoch": 0.5592833304057615,
      "grad_norm": 0.010419441387057304,
      "learning_rate": 8.837356209428428e-06,
      "loss": 1.5388,
      "step": 199
    },
    {
      "epoch": 0.5620937994027754,
      "grad_norm": 0.012639000080525875,
      "learning_rate": 8.744817941191862e-06,
      "loss": 1.5451,
      "step": 200
    },
    {
      "epoch": 0.5649042683997892,
      "grad_norm": 0.01181803084909916,
      "learning_rate": 8.652388752551458e-06,
      "loss": 1.548,
      "step": 201
    },
    {
      "epoch": 0.5677147373968031,
      "grad_norm": 0.01122727058827877,
      "learning_rate": 8.560076675918537e-06,
      "loss": 1.5531,
      "step": 202
    },
    {
      "epoch": 0.570525206393817,
      "grad_norm": 0.011986492201685905,
      "learning_rate": 8.467889733526977e-06,
      "loss": 1.5334,
      "step": 203
    },
    {
      "epoch": 0.5733356753908309,
      "grad_norm": 0.01114217285066843,
      "learning_rate": 8.375835936736072e-06,
      "loss": 1.539,
      "step": 204
    },
    {
      "epoch": 0.5761461443878447,
      "grad_norm": 0.011127611622214317,
      "learning_rate": 8.283923285334304e-06,
      "loss": 1.546,
      "step": 205
    },
    {
      "epoch": 0.5789566133848586,
      "grad_norm": 0.011980608105659485,
      "learning_rate": 8.19215976684414e-06,
      "loss": 1.5556,
      "step": 206
    },
    {
      "epoch": 0.5817670823818725,
      "grad_norm": 0.011237296275794506,
      "learning_rate": 8.100553355827897e-06,
      "loss": 1.5368,
      "step": 207
    },
    {
      "epoch": 0.5845775513788863,
      "grad_norm": 0.01083777192980051,
      "learning_rate": 8.009112013194707e-06,
      "loss": 1.5404,
      "step": 208
    },
    {
      "epoch": 0.5873880203759002,
      "grad_norm": 0.01118643768131733,
      "learning_rate": 7.917843685508702e-06,
      "loss": 1.5405,
      "step": 209
    },
    {
      "epoch": 0.5901984893729141,
      "grad_norm": 0.012426355853676796,
      "learning_rate": 7.826756304298428e-06,
      "loss": 1.5497,
      "step": 210
    },
    {
      "epoch": 0.593008958369928,
      "grad_norm": 0.011632665991783142,
      "learning_rate": 7.73585778536756e-06,
      "loss": 1.5395,
      "step": 211
    },
    {
      "epoch": 0.5958194273669418,
      "grad_norm": 0.010038619861006737,
      "learning_rate": 7.645156028107005e-06,
      "loss": 1.5388,
      "step": 212
    },
    {
      "epoch": 0.5986298963639557,
      "grad_norm": 0.01042498741298914,
      "learning_rate": 7.554658914808404e-06,
      "loss": 1.5446,
      "step": 213
    },
    {
      "epoch": 0.6014403653609696,
      "grad_norm": 0.009714511223137379,
      "learning_rate": 7.464374309979143e-06,
      "loss": 1.545,
      "step": 214
    },
    {
      "epoch": 0.6042508343579834,
      "grad_norm": 0.011047974228858948,
      "learning_rate": 7.3743100596589e-06,
      "loss": 1.5496,
      "step": 215
    },
    {
      "epoch": 0.6070613033549974,
      "grad_norm": 0.009808655828237534,
      "learning_rate": 7.284473990737795e-06,
      "loss": 1.542,
      "step": 216
    },
    {
      "epoch": 0.6098717723520113,
      "grad_norm": 0.010092736221849918,
      "learning_rate": 7.194873910276205e-06,
      "loss": 1.5517,
      "step": 217
    },
    {
      "epoch": 0.6126822413490252,
      "grad_norm": 0.01010197214782238,
      "learning_rate": 7.1055176048263085e-06,
      "loss": 1.541,
      "step": 218
    },
    {
      "epoch": 0.615492710346039,
      "grad_norm": 0.011126919649541378,
      "learning_rate": 7.0164128397554e-06,
      "loss": 1.5582,
      "step": 219
    },
    {
      "epoch": 0.6183031793430529,
      "grad_norm": 0.01078337524086237,
      "learning_rate": 6.92756735857107e-06,
      "loss": 1.5438,
      "step": 220
    },
    {
      "epoch": 0.6211136483400668,
      "grad_norm": 0.010446320287883282,
      "learning_rate": 6.838988882248243e-06,
      "loss": 1.5448,
      "step": 221
    },
    {
      "epoch": 0.6239241173370806,
      "grad_norm": 0.011228025890886784,
      "learning_rate": 6.750685108558221e-06,
      "loss": 1.5347,
      "step": 222
    },
    {
      "epoch": 0.6267345863340945,
      "grad_norm": 0.012325327843427658,
      "learning_rate": 6.662663711399705e-06,
      "loss": 1.5537,
      "step": 223
    },
    {
      "epoch": 0.6295450553311084,
      "grad_norm": 0.01042882353067398,
      "learning_rate": 6.574932340131917e-06,
      "loss": 1.55,
      "step": 224
    },
    {
      "epoch": 0.6323555243281223,
      "grad_norm": 0.012044312432408333,
      "learning_rate": 6.487498618909845e-06,
      "loss": 1.5485,
      "step": 225
    },
    {
      "epoch": 0.6351659933251361,
      "grad_norm": 0.01117675006389618,
      "learning_rate": 6.400370146021662e-06,
      "loss": 1.5488,
      "step": 226
    },
    {
      "epoch": 0.63797646232215,
      "grad_norm": 0.010941598564386368,
      "learning_rate": 6.313554493228431e-06,
      "loss": 1.5456,
      "step": 227
    },
    {
      "epoch": 0.6407869313191639,
      "grad_norm": 0.011893962509930134,
      "learning_rate": 6.227059205106085e-06,
      "loss": 1.543,
      "step": 228
    },
    {
      "epoch": 0.6435974003161777,
      "grad_norm": 0.010805683210492134,
      "learning_rate": 6.14089179838977e-06,
      "loss": 1.5479,
      "step": 229
    },
    {
      "epoch": 0.6464078693131916,
      "grad_norm": 0.011175237596035004,
      "learning_rate": 6.0550597613206205e-06,
      "loss": 1.5399,
      "step": 230
    },
    {
      "epoch": 0.6492183383102055,
      "grad_norm": 0.011044224724173546,
      "learning_rate": 5.969570552995014e-06,
      "loss": 1.547,
      "step": 231
    },
    {
      "epoch": 0.6520288073072193,
      "grad_norm": 0.010453018359839916,
      "learning_rate": 5.8844316027163315e-06,
      "loss": 1.5468,
      "step": 232
    },
    {
      "epoch": 0.6548392763042332,
      "grad_norm": 0.010159061290323734,
      "learning_rate": 5.799650309349348e-06,
      "loss": 1.536,
      "step": 233
    },
    {
      "epoch": 0.6576497453012471,
      "grad_norm": 0.010464577935636044,
      "learning_rate": 5.715234040677229e-06,
      "loss": 1.5529,
      "step": 234
    },
    {
      "epoch": 0.6604602142982611,
      "grad_norm": 0.01091894879937172,
      "learning_rate": 5.631190132761247e-06,
      "loss": 1.5445,
      "step": 235
    },
    {
      "epoch": 0.6632706832952749,
      "grad_norm": 0.011955167166888714,
      "learning_rate": 5.547525889303265e-06,
      "loss": 1.54,
      "step": 236
    },
    {
      "epoch": 0.6660811522922888,
      "grad_norm": 0.010030088014900684,
      "learning_rate": 5.464248581011002e-06,
      "loss": 1.5453,
      "step": 237
    },
    {
      "epoch": 0.6688916212893027,
      "grad_norm": 0.009817596524953842,
      "learning_rate": 5.381365444966205e-06,
      "loss": 1.5367,
      "step": 238
    },
    {
      "epoch": 0.6717020902863166,
      "grad_norm": 0.010095087811350822,
      "learning_rate": 5.298883683995697e-06,
      "loss": 1.5559,
      "step": 239
    },
    {
      "epoch": 0.6745125592833304,
      "grad_norm": 0.010223811492323875,
      "learning_rate": 5.216810466045448e-06,
      "loss": 1.5406,
      "step": 240
    },
    {
      "epoch": 0.6773230282803443,
      "grad_norm": 0.009899747557938099,
      "learning_rate": 5.135152923557647e-06,
      "loss": 1.5395,
      "step": 241
    },
    {
      "epoch": 0.6801334972773582,
      "grad_norm": 0.009352842345833778,
      "learning_rate": 5.053918152850868e-06,
      "loss": 1.5423,
      "step": 242
    },
    {
      "epoch": 0.682943966274372,
      "grad_norm": 0.0106782466173172,
      "learning_rate": 4.973113213503379e-06,
      "loss": 1.5397,
      "step": 243
    },
    {
      "epoch": 0.6857544352713859,
      "grad_norm": 0.010458718985319138,
      "learning_rate": 4.8927451277396365e-06,
      "loss": 1.5264,
      "step": 244
    },
    {
      "epoch": 0.6885649042683998,
      "grad_norm": 0.01009244006127119,
      "learning_rate": 4.812820879820034e-06,
      "loss": 1.5512,
      "step": 245
    },
    {
      "epoch": 0.6913753732654137,
      "grad_norm": 0.009827625937759876,
      "learning_rate": 4.733347415433946e-06,
      "loss": 1.5421,
      "step": 246
    },
    {
      "epoch": 0.6941858422624275,
      "grad_norm": 0.010725892148911953,
      "learning_rate": 4.654331641096118e-06,
      "loss": 1.5447,
      "step": 247
    },
    {
      "epoch": 0.6969963112594414,
      "grad_norm": 0.009530246257781982,
      "learning_rate": 4.575780423546476e-06,
      "loss": 1.5429,
      "step": 248
    },
    {
      "epoch": 0.6998067802564553,
      "grad_norm": 0.009370286948978901,
      "learning_rate": 4.497700589153379e-06,
      "loss": 1.54,
      "step": 249
    },
    {
      "epoch": 0.7026172492534691,
      "grad_norm": 0.010170893743634224,
      "learning_rate": 4.420098923320378e-06,
      "loss": 1.5448,
      "step": 250
    },
    {
      "epoch": 0.705427718250483,
      "grad_norm": 0.01000099815428257,
      "learning_rate": 4.342982169896555e-06,
      "loss": 1.5377,
      "step": 251
    },
    {
      "epoch": 0.7082381872474969,
      "grad_norm": 0.010095086880028248,
      "learning_rate": 4.266357030590449e-06,
      "loss": 1.5321,
      "step": 252
    },
    {
      "epoch": 0.7110486562445109,
      "grad_norm": 0.009943116456270218,
      "learning_rate": 4.1902301643876555e-06,
      "loss": 1.543,
      "step": 253
    },
    {
      "epoch": 0.7138591252415247,
      "grad_norm": 0.009373177774250507,
      "learning_rate": 4.114608186972143e-06,
      "loss": 1.5432,
      "step": 254
    },
    {
      "epoch": 0.7166695942385386,
      "grad_norm": 0.009264045394957066,
      "learning_rate": 4.0394976701513235e-06,
      "loss": 1.5501,
      "step": 255
    },
    {
      "epoch": 0.7194800632355525,
      "grad_norm": 0.00987312849611044,
      "learning_rate": 3.96490514128494e-06,
      "loss": 1.5376,
      "step": 256
    },
    {
      "epoch": 0.7222905322325663,
      "grad_norm": 0.010003172792494297,
      "learning_rate": 3.890837082717822e-06,
      "loss": 1.5424,
      "step": 257
    },
    {
      "epoch": 0.7251010012295802,
      "grad_norm": 0.009270971640944481,
      "learning_rate": 3.817299931216537e-06,
      "loss": 1.5453,
      "step": 258
    },
    {
      "epoch": 0.7279114702265941,
      "grad_norm": 0.009572354145348072,
      "learning_rate": 3.74430007741003e-06,
      "loss": 1.5325,
      "step": 259
    },
    {
      "epoch": 0.730721939223608,
      "grad_norm": 0.009282928891479969,
      "learning_rate": 3.671843865234238e-06,
      "loss": 1.5447,
      "step": 260
    },
    {
      "epoch": 0.7335324082206218,
      "grad_norm": 0.0098671093583107,
      "learning_rate": 3.599937591380791e-06,
      "loss": 1.5466,
      "step": 261
    },
    {
      "epoch": 0.7363428772176357,
      "grad_norm": 0.010173565708100796,
      "learning_rate": 3.5285875047498075e-06,
      "loss": 1.5347,
      "step": 262
    },
    {
      "epoch": 0.7391533462146496,
      "grad_norm": 0.0094713494181633,
      "learning_rate": 3.4577998059068354e-06,
      "loss": 1.542,
      "step": 263
    },
    {
      "epoch": 0.7419638152116634,
      "grad_norm": 0.009647979401051998,
      "learning_rate": 3.3875806465440152e-06,
      "loss": 1.5437,
      "step": 264
    },
    {
      "epoch": 0.7447742842086773,
      "grad_norm": 0.009664146229624748,
      "learning_rate": 3.3179361289454694e-06,
      "loss": 1.5406,
      "step": 265
    },
    {
      "epoch": 0.7475847532056912,
      "grad_norm": 0.010614863596856594,
      "learning_rate": 3.2488723054569905e-06,
      "loss": 1.5512,
      "step": 266
    },
    {
      "epoch": 0.750395222202705,
      "grad_norm": 0.009987459518015385,
      "learning_rate": 3.1803951779600774e-06,
      "loss": 1.5341,
      "step": 267
    },
    {
      "epoch": 0.7532056911997189,
      "grad_norm": 0.009676804766058922,
      "learning_rate": 3.112510697350348e-06,
      "loss": 1.5457,
      "step": 268
    },
    {
      "epoch": 0.7560161601967328,
      "grad_norm": 0.010411749593913555,
      "learning_rate": 3.04522476302039e-06,
      "loss": 1.5456,
      "step": 269
    },
    {
      "epoch": 0.7588266291937467,
      "grad_norm": 0.010124873369932175,
      "learning_rate": 2.978543222347076e-06,
      "loss": 1.5452,
      "step": 270
    },
    {
      "epoch": 0.7616370981907605,
      "grad_norm": 0.009868764318525791,
      "learning_rate": 2.912471870183411e-06,
      "loss": 1.555,
      "step": 271
    },
    {
      "epoch": 0.7644475671877745,
      "grad_norm": 0.009340250864624977,
      "learning_rate": 2.847016448354948e-06,
      "loss": 1.5419,
      "step": 272
    },
    {
      "epoch": 0.7672580361847884,
      "grad_norm": 0.00967357400804758,
      "learning_rate": 2.782182645160789e-06,
      "loss": 1.5433,
      "step": 273
    },
    {
      "epoch": 0.7700685051818023,
      "grad_norm": 0.009733307175338268,
      "learning_rate": 2.71797609487926e-06,
      "loss": 1.5378,
      "step": 274
    },
    {
      "epoch": 0.7728789741788161,
      "grad_norm": 0.009725394658744335,
      "learning_rate": 2.6544023772782736e-06,
      "loss": 1.5373,
      "step": 275
    },
    {
      "epoch": 0.77568944317583,
      "grad_norm": 0.009589461609721184,
      "learning_rate": 2.591467017130426e-06,
      "loss": 1.5331,
      "step": 276
    },
    {
      "epoch": 0.7784999121728439,
      "grad_norm": 0.009730999357998371,
      "learning_rate": 2.5291754837328787e-06,
      "loss": 1.5374,
      "step": 277
    },
    {
      "epoch": 0.7813103811698577,
      "grad_norm": 0.009977961890399456,
      "learning_rate": 2.4675331904320533e-06,
      "loss": 1.5386,
      "step": 278
    },
    {
      "epoch": 0.7841208501668716,
      "grad_norm": 0.009760740213096142,
      "learning_rate": 2.4065454941531963e-06,
      "loss": 1.5521,
      "step": 279
    },
    {
      "epoch": 0.7869313191638855,
      "grad_norm": 0.009903214871883392,
      "learning_rate": 2.346217694934847e-06,
      "loss": 1.5408,
      "step": 280
    },
    {
      "epoch": 0.7897417881608993,
      "grad_norm": 0.010044459253549576,
      "learning_rate": 2.286555035468233e-06,
      "loss": 1.54,
      "step": 281
    },
    {
      "epoch": 0.7925522571579132,
      "grad_norm": 0.009294719435274601,
      "learning_rate": 2.22756270064168e-06,
      "loss": 1.5417,
      "step": 282
    },
    {
      "epoch": 0.7953627261549271,
      "grad_norm": 0.009228096343576908,
      "learning_rate": 2.16924581709002e-06,
      "loss": 1.5285,
      "step": 283
    },
    {
      "epoch": 0.798173195151941,
      "grad_norm": 0.009288915432989597,
      "learning_rate": 2.1116094527490594e-06,
      "loss": 1.5431,
      "step": 284
    },
    {
      "epoch": 0.8009836641489548,
      "grad_norm": 0.010039190761744976,
      "learning_rate": 2.0546586164151827e-06,
      "loss": 1.5324,
      "step": 285
    },
    {
      "epoch": 0.8037941331459687,
      "grad_norm": 0.009471042081713676,
      "learning_rate": 1.9983982573100413e-06,
      "loss": 1.5375,
      "step": 286
    },
    {
      "epoch": 0.8066046021429826,
      "grad_norm": 0.009732691571116447,
      "learning_rate": 1.94283326465047e-06,
      "loss": 1.5425,
      "step": 287
    },
    {
      "epoch": 0.8094150711399964,
      "grad_norm": 0.00961157213896513,
      "learning_rate": 1.887968467223591e-06,
      "loss": 1.5378,
      "step": 288
    },
    {
      "epoch": 0.8122255401370103,
      "grad_norm": 0.009570587426424026,
      "learning_rate": 1.8338086329671734e-06,
      "loss": 1.546,
      "step": 289
    },
    {
      "epoch": 0.8150360091340242,
      "grad_norm": 0.009907620027661324,
      "learning_rate": 1.7803584685552877e-06,
      "loss": 1.545,
      "step": 290
    },
    {
      "epoch": 0.8178464781310382,
      "grad_norm": 0.009468463249504566,
      "learning_rate": 1.7276226189892763e-06,
      "loss": 1.5387,
      "step": 291
    },
    {
      "epoch": 0.820656947128052,
      "grad_norm": 0.008901305496692657,
      "learning_rate": 1.6756056671940902e-06,
      "loss": 1.541,
      "step": 292
    },
    {
      "epoch": 0.8234674161250659,
      "grad_norm": 0.009180068038403988,
      "learning_rate": 1.624312133620013e-06,
      "loss": 1.543,
      "step": 293
    },
    {
      "epoch": 0.8262778851220798,
      "grad_norm": 0.0097258435562253,
      "learning_rate": 1.5737464758498243e-06,
      "loss": 1.5323,
      "step": 294
    },
    {
      "epoch": 0.8290883541190937,
      "grad_norm": 0.00945055577903986,
      "learning_rate": 1.523913088211415e-06,
      "loss": 1.5366,
      "step": 295
    },
    {
      "epoch": 0.8318988231161075,
      "grad_norm": 0.009719719178974628,
      "learning_rate": 1.474816301395906e-06,
      "loss": 1.5489,
      "step": 296
    },
    {
      "epoch": 0.8347092921131214,
      "grad_norm": 0.00904281809926033,
      "learning_rate": 1.4264603820813006e-06,
      "loss": 1.5413,
      "step": 297
    },
    {
      "epoch": 0.8375197611101353,
      "grad_norm": 0.009636901319026947,
      "learning_rate": 1.3788495325616912e-06,
      "loss": 1.5493,
      "step": 298
    },
    {
      "epoch": 0.8403302301071491,
      "grad_norm": 0.00919989962130785,
      "learning_rate": 1.3319878903820682e-06,
      "loss": 1.5387,
      "step": 299
    },
    {
      "epoch": 0.843140699104163,
      "grad_norm": 0.009369692765176296,
      "learning_rate": 1.2858795279787517e-06,
      "loss": 1.54,
      "step": 300
    },
    {
      "epoch": 0.8459511681011769,
      "grad_norm": 0.009285050444304943,
      "learning_rate": 1.2405284523254823e-06,
      "loss": 1.5496,
      "step": 301
    },
    {
      "epoch": 0.8487616370981907,
      "grad_norm": 0.009485352784395218,
      "learning_rate": 1.195938604585205e-06,
      "loss": 1.5274,
      "step": 302
    },
    {
      "epoch": 0.8515721060952046,
      "grad_norm": 0.00928215030580759,
      "learning_rate": 1.152113859767565e-06,
      "loss": 1.5437,
      "step": 303
    },
    {
      "epoch": 0.8543825750922185,
      "grad_norm": 0.009152165614068508,
      "learning_rate": 1.109058026392158e-06,
      "loss": 1.5395,
      "step": 304
    },
    {
      "epoch": 0.8571930440892324,
      "grad_norm": 0.009171172976493835,
      "learning_rate": 1.0667748461575544e-06,
      "loss": 1.5487,
      "step": 305
    },
    {
      "epoch": 0.8600035130862462,
      "grad_norm": 0.00959504209458828,
      "learning_rate": 1.0252679936161392e-06,
      "loss": 1.5374,
      "step": 306
    },
    {
      "epoch": 0.8628139820832601,
      "grad_norm": 0.00906702782958746,
      "learning_rate": 9.845410758547724e-07,
      "loss": 1.5412,
      "step": 307
    },
    {
      "epoch": 0.865624451080274,
      "grad_norm": 0.008921015076339245,
      "learning_rate": 9.445976321813277e-07,
      "loss": 1.5295,
      "step": 308
    },
    {
      "epoch": 0.868434920077288,
      "grad_norm": 0.009434668347239494,
      "learning_rate": 9.054411338171099e-07,
      "loss": 1.5474,
      "step": 309
    },
    {
      "epoch": 0.8712453890743018,
      "grad_norm": 0.009549295529723167,
      "learning_rate": 8.670749835951964e-07,
      "loss": 1.5425,
      "step": 310
    },
    {
      "epoch": 0.8740558580713157,
      "grad_norm": 0.009337035939097404,
      "learning_rate": 8.29502515664723e-07,
      "loss": 1.5412,
      "step": 311
    },
    {
      "epoch": 0.8768663270683296,
      "grad_norm": 0.008889071643352509,
      "learning_rate": 7.927269952011285e-07,
      "loss": 1.5589,
      "step": 312
    },
    {
      "epoch": 0.8796767960653434,
      "grad_norm": 0.008982825092971325,
      "learning_rate": 7.567516181223966e-07,
      "loss": 1.5504,
      "step": 313
    },
    {
      "epoch": 0.8824872650623573,
      "grad_norm": 0.009038092568516731,
      "learning_rate": 7.215795108113343e-07,
      "loss": 1.5391,
      "step": 314
    },
    {
      "epoch": 0.8852977340593712,
      "grad_norm": 0.0092889703810215,
      "learning_rate": 6.872137298438653e-07,
      "loss": 1.5354,
      "step": 315
    },
    {
      "epoch": 0.888108203056385,
      "grad_norm": 0.009272675029933453,
      "learning_rate": 6.536572617234082e-07,
      "loss": 1.5449,
      "step": 316
    },
    {
      "epoch": 0.8909186720533989,
      "grad_norm": 0.00896420981734991,
      "learning_rate": 6.209130226213378e-07,
      "loss": 1.5544,
      "step": 317
    },
    {
      "epoch": 0.8937291410504128,
      "grad_norm": 0.00960630364716053,
      "learning_rate": 5.889838581235641e-07,
      "loss": 1.5425,
      "step": 318
    },
    {
      "epoch": 0.8965396100474267,
      "grad_norm": 0.009224542416632175,
      "learning_rate": 5.578725429832344e-07,
      "loss": 1.5287,
      "step": 319
    },
    {
      "epoch": 0.8993500790444405,
      "grad_norm": 0.009561055339872837,
      "learning_rate": 5.275817808796013e-07,
      "loss": 1.5399,
      "step": 320
    },
    {
      "epoch": 0.9021605480414544,
      "grad_norm": 0.00883413664996624,
      "learning_rate": 4.981142041830645e-07,
      "loss": 1.5373,
      "step": 321
    },
    {
      "epoch": 0.9049710170384683,
      "grad_norm": 0.0092536062002182,
      "learning_rate": 4.6947237372640954e-07,
      "loss": 1.5372,
      "step": 322
    },
    {
      "epoch": 0.9077814860354821,
      "grad_norm": 0.008912206627428532,
      "learning_rate": 4.416587785822568e-07,
      "loss": 1.5412,
      "step": 323
    },
    {
      "epoch": 0.910591955032496,
      "grad_norm": 0.009592842310667038,
      "learning_rate": 4.1467583584676395e-07,
      "loss": 1.541,
      "step": 324
    },
    {
      "epoch": 0.9134024240295099,
      "grad_norm": 0.00898828823119402,
      "learning_rate": 3.885258904295575e-07,
      "loss": 1.539,
      "step": 325
    },
    {
      "epoch": 0.9162128930265238,
      "grad_norm": 0.009104624390602112,
      "learning_rate": 3.6321121484996447e-07,
      "loss": 1.5313,
      "step": 326
    },
    {
      "epoch": 0.9190233620235376,
      "grad_norm": 0.00905083492398262,
      "learning_rate": 3.3873400903951636e-07,
      "loss": 1.5396,
      "step": 327
    },
    {
      "epoch": 0.9218338310205516,
      "grad_norm": 0.009055250324308872,
      "learning_rate": 3.1509640015076946e-07,
      "loss": 1.5471,
      "step": 328
    },
    {
      "epoch": 0.9246443000175655,
      "grad_norm": 0.00856536254286766,
      "learning_rate": 2.923004423724474e-07,
      "loss": 1.5302,
      "step": 329
    },
    {
      "epoch": 0.9274547690145793,
      "grad_norm": 0.0095223942771554,
      "learning_rate": 2.703481167509281e-07,
      "loss": 1.5387,
      "step": 330
    },
    {
      "epoch": 0.9302652380115932,
      "grad_norm": 0.009137586690485477,
      "learning_rate": 2.4924133101807636e-07,
      "loss": 1.5379,
      "step": 331
    },
    {
      "epoch": 0.9330757070086071,
      "grad_norm": 0.009256749413907528,
      "learning_rate": 2.289819194254661e-07,
      "loss": 1.54,
      "step": 332
    },
    {
      "epoch": 0.935886176005621,
      "grad_norm": 0.00952975358814001,
      "learning_rate": 2.0957164258497031e-07,
      "loss": 1.5413,
      "step": 333
    },
    {
      "epoch": 0.9386966450026348,
      "grad_norm": 0.009139897301793098,
      "learning_rate": 1.9101218731575777e-07,
      "loss": 1.5547,
      "step": 334
    },
    {
      "epoch": 0.9415071139996487,
      "grad_norm": 0.009397204965353012,
      "learning_rate": 1.73305166497707e-07,
      "loss": 1.5374,
      "step": 335
    },
    {
      "epoch": 0.9443175829966626,
      "grad_norm": 0.009158180095255375,
      "learning_rate": 1.5645211893123846e-07,
      "loss": 1.5449,
      "step": 336
    },
    {
      "epoch": 0.9471280519936764,
      "grad_norm": 0.008892880752682686,
      "learning_rate": 1.4045450920358917e-07,
      "loss": 1.5418,
      "step": 337
    },
    {
      "epoch": 0.9499385209906903,
      "grad_norm": 0.008902773261070251,
      "learning_rate": 1.2531372756153458e-07,
      "loss": 1.5451,
      "step": 338
    },
    {
      "epoch": 0.9527489899877042,
      "grad_norm": 0.00923394039273262,
      "learning_rate": 1.1103108979056865e-07,
      "loss": 1.5329,
      "step": 339
    },
    {
      "epoch": 0.9555594589847181,
      "grad_norm": 0.009119280613958836,
      "learning_rate": 9.760783710056176e-08,
      "loss": 1.5399,
      "step": 340
    },
    {
      "epoch": 0.9583699279817319,
      "grad_norm": 0.009382793679833412,
      "learning_rate": 8.504513601789388e-08,
      "loss": 1.5372,
      "step": 341
    },
    {
      "epoch": 0.9611803969787458,
      "grad_norm": 0.0087265744805336,
      "learning_rate": 7.334407828407885e-08,
      "loss": 1.5415,
      "step": 342
    },
    {
      "epoch": 0.9639908659757597,
      "grad_norm": 0.008734731003642082,
      "learning_rate": 6.250568076088814e-08,
      "loss": 1.5388,
      "step": 343
    },
    {
      "epoch": 0.9668013349727735,
      "grad_norm": 0.00877598486840725,
      "learning_rate": 5.2530885341982586e-08,
      "loss": 1.5385,
      "step": 344
    },
    {
      "epoch": 0.9696118039697874,
      "grad_norm": 0.008809634484350681,
      "learning_rate": 4.3420558871060116e-08,
      "loss": 1.5447,
      "step": 345
    },
    {
      "epoch": 0.9724222729668013,
      "grad_norm": 0.00890128593891859,
      "learning_rate": 3.517549306652157e-08,
      "loss": 1.538,
      "step": 346
    },
    {
      "epoch": 0.9752327419638153,
      "grad_norm": 0.008788186125457287,
      "learning_rate": 2.7796404452666847e-08,
      "loss": 1.5506,
      "step": 347
    },
    {
      "epoch": 0.9780432109608291,
      "grad_norm": 0.009096509777009487,
      "learning_rate": 2.1283934297432472e-08,
      "loss": 1.5365,
      "step": 348
    },
    {
      "epoch": 0.980853679957843,
      "grad_norm": 0.00890890508890152,
      "learning_rate": 1.5638648556656198e-08,
      "loss": 1.5406,
      "step": 349
    },
    {
      "epoch": 0.9836641489548569,
      "grad_norm": 0.008733571507036686,
      "learning_rate": 1.0861037824896337e-08,
      "loss": 1.5474,
      "step": 350
    },
    {
      "epoch": 0.9864746179518707,
      "grad_norm": 0.00961330346763134,
      "learning_rate": 6.951517292800303e-09,
      "loss": 1.5417,
      "step": 351
    },
    {
      "epoch": 0.9892850869488846,
      "grad_norm": 0.009068558923900127,
      "learning_rate": 3.9104267110168235e-09,
      "loss": 1.5369,
      "step": 352
    },
    {
      "epoch": 0.9920955559458985,
      "grad_norm": 0.008888053707778454,
      "learning_rate": 1.738030360677323e-09,
      "loss": 1.5488,
      "step": 353
    },
    {
      "epoch": 0.9949060249429124,
      "grad_norm": 0.009277201257646084,
      "learning_rate": 4.3451703042207694e-10,
      "loss": 1.5375,
      "step": 354
    },
    {
      "epoch": 0.9977164939399262,
      "grad_norm": 0.009334351867437363,
      "learning_rate": 0.0,
      "loss": 1.5352,
      "step": 355
    },
    {
      "epoch": 0.6385650224215247,
      "grad_norm": 0.010235507041215897,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 356
    },
    {
      "epoch": 0.6403587443946188,
      "grad_norm": 0.010560178197920322,
      "learning_rate": 2e-05,
      "loss": 1.5118,
      "step": 357
    },
    {
      "epoch": 0.6421524663677131,
      "grad_norm": 0.010353959165513515,
      "learning_rate": 2e-05,
      "loss": 1.5226,
      "step": 358
    },
    {
      "epoch": 0.6439461883408072,
      "grad_norm": 0.010382940992712975,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 359
    },
    {
      "epoch": 0.6457399103139013,
      "grad_norm": 0.009856803342700005,
      "learning_rate": 2e-05,
      "loss": 1.5167,
      "step": 360
    },
    {
      "epoch": 0.6475336322869956,
      "grad_norm": 0.010195410810410976,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 361
    },
    {
      "epoch": 0.6493273542600897,
      "grad_norm": 0.010302864946424961,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 362
    },
    {
      "epoch": 0.6511210762331838,
      "grad_norm": 0.010046405717730522,
      "learning_rate": 2e-05,
      "loss": 1.5112,
      "step": 363
    },
    {
      "epoch": 0.6529147982062781,
      "grad_norm": 0.010849208571016788,
      "learning_rate": 2e-05,
      "loss": 1.5114,
      "step": 364
    },
    {
      "epoch": 0.6547085201793722,
      "grad_norm": 0.010421674698591232,
      "learning_rate": 2e-05,
      "loss": 1.5173,
      "step": 365
    },
    {
      "epoch": 0.6565022421524663,
      "grad_norm": 0.00989589188247919,
      "learning_rate": 2e-05,
      "loss": 1.5063,
      "step": 366
    },
    {
      "epoch": 0.6582959641255606,
      "grad_norm": 0.010465629398822784,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 367
    },
    {
      "epoch": 0.6600896860986547,
      "grad_norm": 0.009964341297745705,
      "learning_rate": 2e-05,
      "loss": 1.5207,
      "step": 368
    },
    {
      "epoch": 0.6618834080717488,
      "grad_norm": 0.01189314667135477,
      "learning_rate": 2e-05,
      "loss": 1.5361,
      "step": 369
    },
    {
      "epoch": 0.6636771300448431,
      "grad_norm": 0.01012677513062954,
      "learning_rate": 2e-05,
      "loss": 1.5215,
      "step": 370
    },
    {
      "epoch": 0.6654708520179372,
      "grad_norm": 0.009877102449536324,
      "learning_rate": 2e-05,
      "loss": 1.5262,
      "step": 371
    },
    {
      "epoch": 0.6672645739910313,
      "grad_norm": 0.01000463031232357,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 372
    },
    {
      "epoch": 0.6690582959641256,
      "grad_norm": 0.010188892483711243,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 373
    },
    {
      "epoch": 0.6708520179372197,
      "grad_norm": 0.010129815898835659,
      "learning_rate": 2e-05,
      "loss": 1.5245,
      "step": 374
    },
    {
      "epoch": 0.672645739910314,
      "grad_norm": 0.010608335956931114,
      "learning_rate": 2e-05,
      "loss": 1.5169,
      "step": 375
    },
    {
      "epoch": 0.6744394618834081,
      "grad_norm": 0.010223207995295525,
      "learning_rate": 2e-05,
      "loss": 1.5185,
      "step": 376
    },
    {
      "epoch": 0.6762331838565022,
      "grad_norm": 0.010141369886696339,
      "learning_rate": 2e-05,
      "loss": 1.5161,
      "step": 377
    },
    {
      "epoch": 0.6780269058295965,
      "grad_norm": 0.01027351152151823,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 378
    },
    {
      "epoch": 0.6798206278026906,
      "grad_norm": 0.010362266562879086,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 379
    },
    {
      "epoch": 0.6816143497757847,
      "grad_norm": 0.010336722247302532,
      "learning_rate": 2e-05,
      "loss": 1.5173,
      "step": 380
    },
    {
      "epoch": 0.683408071748879,
      "grad_norm": 0.01007298193871975,
      "learning_rate": 2e-05,
      "loss": 1.5075,
      "step": 381
    },
    {
      "epoch": 0.6852017937219731,
      "grad_norm": 0.010275410488247871,
      "learning_rate": 2e-05,
      "loss": 1.5123,
      "step": 382
    },
    {
      "epoch": 0.6869955156950672,
      "grad_norm": 0.010203160345554352,
      "learning_rate": 2e-05,
      "loss": 1.5151,
      "step": 383
    },
    {
      "epoch": 0.6887892376681615,
      "grad_norm": 0.010127630084753036,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 384
    },
    {
      "epoch": 0.6905829596412556,
      "grad_norm": 0.009799284860491753,
      "learning_rate": 2e-05,
      "loss": 1.5191,
      "step": 385
    },
    {
      "epoch": 0.6923766816143497,
      "grad_norm": 0.01014394499361515,
      "learning_rate": 2e-05,
      "loss": 1.5261,
      "step": 386
    },
    {
      "epoch": 0.694170403587444,
      "grad_norm": 0.010567774064838886,
      "learning_rate": 2e-05,
      "loss": 1.5232,
      "step": 387
    },
    {
      "epoch": 0.6959641255605381,
      "grad_norm": 0.010051852092146873,
      "learning_rate": 2e-05,
      "loss": 1.5212,
      "step": 388
    },
    {
      "epoch": 0.6977578475336322,
      "grad_norm": 0.010241293348371983,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 389
    },
    {
      "epoch": 0.6995515695067265,
      "grad_norm": 0.0095717404037714,
      "learning_rate": 2e-05,
      "loss": 1.5115,
      "step": 390
    },
    {
      "epoch": 0.7013452914798206,
      "grad_norm": 0.00974031537771225,
      "learning_rate": 2e-05,
      "loss": 1.5195,
      "step": 391
    },
    {
      "epoch": 0.7031390134529149,
      "grad_norm": 0.010140657424926758,
      "learning_rate": 2e-05,
      "loss": 1.5048,
      "step": 392
    },
    {
      "epoch": 0.704932735426009,
      "grad_norm": 0.010055477730929852,
      "learning_rate": 2e-05,
      "loss": 1.5162,
      "step": 393
    },
    {
      "epoch": 0.7067264573991031,
      "grad_norm": 0.01005468424409628,
      "learning_rate": 2e-05,
      "loss": 1.5258,
      "step": 394
    },
    {
      "epoch": 0.7085201793721974,
      "grad_norm": 0.010284669697284698,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 395
    },
    {
      "epoch": 0.7103139013452915,
      "grad_norm": 0.010200968012213707,
      "learning_rate": 2e-05,
      "loss": 1.5172,
      "step": 396
    },
    {
      "epoch": 0.7121076233183856,
      "grad_norm": 0.01015354972332716,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 397
    },
    {
      "epoch": 0.7139013452914799,
      "grad_norm": 0.009913373738527298,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 398
    },
    {
      "epoch": 0.715695067264574,
      "grad_norm": 0.010287330485880375,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 399
    },
    {
      "epoch": 0.7174887892376681,
      "grad_norm": 0.01057345885783434,
      "learning_rate": 2e-05,
      "loss": 1.5199,
      "step": 400
    },
    {
      "epoch": 0.7192825112107624,
      "grad_norm": 0.010113878175616264,
      "learning_rate": 2e-05,
      "loss": 1.5168,
      "step": 401
    },
    {
      "epoch": 0.7210762331838565,
      "grad_norm": 0.009940318763256073,
      "learning_rate": 2e-05,
      "loss": 1.5175,
      "step": 402
    },
    {
      "epoch": 0.7228699551569506,
      "grad_norm": 0.010180394165217876,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 403
    },
    {
      "epoch": 0.7246636771300449,
      "grad_norm": 0.00961736124008894,
      "learning_rate": 2e-05,
      "loss": 1.5228,
      "step": 404
    },
    {
      "epoch": 0.726457399103139,
      "grad_norm": 0.010378845036029816,
      "learning_rate": 2e-05,
      "loss": 1.522,
      "step": 405
    },
    {
      "epoch": 0.7282511210762331,
      "grad_norm": 0.010189516469836235,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 406
    },
    {
      "epoch": 0.7300448430493274,
      "grad_norm": 0.010004358366131783,
      "learning_rate": 2e-05,
      "loss": 1.5172,
      "step": 407
    },
    {
      "epoch": 0.7318385650224215,
      "grad_norm": 0.010387993417680264,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 408
    },
    {
      "epoch": 0.7336322869955157,
      "grad_norm": 0.010004810988903046,
      "learning_rate": 2e-05,
      "loss": 1.5132,
      "step": 409
    },
    {
      "epoch": 0.7354260089686099,
      "grad_norm": 0.009845850057899952,
      "learning_rate": 2e-05,
      "loss": 1.5248,
      "step": 410
    },
    {
      "epoch": 0.737219730941704,
      "grad_norm": 0.010015097446739674,
      "learning_rate": 2e-05,
      "loss": 1.5196,
      "step": 411
    },
    {
      "epoch": 0.7390134529147983,
      "grad_norm": 0.009975203312933445,
      "learning_rate": 2e-05,
      "loss": 1.5096,
      "step": 412
    },
    {
      "epoch": 0.7408071748878924,
      "grad_norm": 0.010078891180455685,
      "learning_rate": 2e-05,
      "loss": 1.5162,
      "step": 413
    },
    {
      "epoch": 0.7426008968609865,
      "grad_norm": 0.011885426007211208,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 414
    },
    {
      "epoch": 0.7443946188340808,
      "grad_norm": 0.009693853557109833,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 415
    },
    {
      "epoch": 0.7461883408071749,
      "grad_norm": 0.010337116196751595,
      "learning_rate": 2e-05,
      "loss": 1.5191,
      "step": 416
    },
    {
      "epoch": 0.747982062780269,
      "grad_norm": 0.00993486400693655,
      "learning_rate": 2e-05,
      "loss": 1.5177,
      "step": 417
    },
    {
      "epoch": 0.7497757847533633,
      "grad_norm": 0.010143253020942211,
      "learning_rate": 2e-05,
      "loss": 1.514,
      "step": 418
    },
    {
      "epoch": 0.7515695067264574,
      "grad_norm": 0.010233073495328426,
      "learning_rate": 2e-05,
      "loss": 1.5154,
      "step": 419
    },
    {
      "epoch": 0.7533632286995515,
      "grad_norm": 0.009982983581721783,
      "learning_rate": 2e-05,
      "loss": 1.5223,
      "step": 420
    },
    {
      "epoch": 0.7551569506726458,
      "grad_norm": 0.010409766808152199,
      "learning_rate": 2e-05,
      "loss": 1.5152,
      "step": 421
    },
    {
      "epoch": 0.7569506726457399,
      "grad_norm": 0.0099264495074749,
      "learning_rate": 2e-05,
      "loss": 1.5185,
      "step": 422
    },
    {
      "epoch": 0.758744394618834,
      "grad_norm": 0.009928545914590359,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 423
    },
    {
      "epoch": 0.7605381165919283,
      "grad_norm": 0.009940563701093197,
      "learning_rate": 2e-05,
      "loss": 1.5071,
      "step": 424
    },
    {
      "epoch": 0.7623318385650224,
      "grad_norm": 0.010767797008156776,
      "learning_rate": 2e-05,
      "loss": 1.5006,
      "step": 425
    },
    {
      "epoch": 0.7641255605381166,
      "grad_norm": 0.010551121085882187,
      "learning_rate": 2e-05,
      "loss": 1.5201,
      "step": 426
    },
    {
      "epoch": 0.7659192825112108,
      "grad_norm": 0.010118665173649788,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 427
    },
    {
      "epoch": 0.7677130044843049,
      "grad_norm": 0.010247626341879368,
      "learning_rate": 2e-05,
      "loss": 1.5178,
      "step": 428
    },
    {
      "epoch": 0.7695067264573991,
      "grad_norm": 0.010188435204327106,
      "learning_rate": 2e-05,
      "loss": 1.5085,
      "step": 429
    },
    {
      "epoch": 0.7713004484304933,
      "grad_norm": 0.010428003035485744,
      "learning_rate": 2e-05,
      "loss": 1.5124,
      "step": 430
    },
    {
      "epoch": 0.7730941704035874,
      "grad_norm": 0.01012035645544529,
      "learning_rate": 2e-05,
      "loss": 1.5299,
      "step": 431
    },
    {
      "epoch": 0.7748878923766817,
      "grad_norm": 0.010584665462374687,
      "learning_rate": 2e-05,
      "loss": 1.5095,
      "step": 432
    },
    {
      "epoch": 0.7766816143497758,
      "grad_norm": 0.009979243390262127,
      "learning_rate": 2e-05,
      "loss": 1.5193,
      "step": 433
    },
    {
      "epoch": 0.7784753363228699,
      "grad_norm": 0.00958004966378212,
      "learning_rate": 2e-05,
      "loss": 1.5214,
      "step": 434
    },
    {
      "epoch": 0.7802690582959642,
      "grad_norm": 0.00973733700811863,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 435
    },
    {
      "epoch": 0.7820627802690583,
      "grad_norm": 0.010465665720403194,
      "learning_rate": 2e-05,
      "loss": 1.5227,
      "step": 436
    },
    {
      "epoch": 0.7838565022421524,
      "grad_norm": 0.010098133236169815,
      "learning_rate": 2e-05,
      "loss": 1.5248,
      "step": 437
    },
    {
      "epoch": 0.7856502242152467,
      "grad_norm": 0.10259313136339188,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 438
    },
    {
      "epoch": 0.7874439461883408,
      "grad_norm": 0.01040815282613039,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 439
    },
    {
      "epoch": 0.7892376681614349,
      "grad_norm": 0.010325520299375057,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 440
    },
    {
      "epoch": 0.7910313901345292,
      "grad_norm": 0.010079775005578995,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 441
    },
    {
      "epoch": 0.7928251121076233,
      "grad_norm": 0.010167201980948448,
      "learning_rate": 2e-05,
      "loss": 1.5116,
      "step": 442
    },
    {
      "epoch": 0.7946188340807175,
      "grad_norm": 0.010806124657392502,
      "learning_rate": 2e-05,
      "loss": 1.5153,
      "step": 443
    },
    {
      "epoch": 0.7964125560538117,
      "grad_norm": 0.010324080474674702,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 444
    },
    {
      "epoch": 0.7982062780269058,
      "grad_norm": 0.010092305950820446,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 445
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.01007048413157463,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 446
    },
    {
      "epoch": 0.8017937219730942,
      "grad_norm": 0.010184276849031448,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 447
    },
    {
      "epoch": 0.8035874439461883,
      "grad_norm": 0.010521662421524525,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 448
    },
    {
      "epoch": 0.8053811659192825,
      "grad_norm": 0.010600044392049313,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 449
    },
    {
      "epoch": 0.8071748878923767,
      "grad_norm": 0.009714100509881973,
      "learning_rate": 2e-05,
      "loss": 1.5122,
      "step": 450
    },
    {
      "epoch": 0.8089686098654708,
      "grad_norm": 0.010295005515217781,
      "learning_rate": 2e-05,
      "loss": 1.52,
      "step": 451
    },
    {
      "epoch": 0.810762331838565,
      "grad_norm": 0.010034569539129734,
      "learning_rate": 2e-05,
      "loss": 1.5197,
      "step": 452
    },
    {
      "epoch": 0.8125560538116592,
      "grad_norm": 0.010086962021887302,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 453
    },
    {
      "epoch": 0.8143497757847533,
      "grad_norm": 0.010277335532009602,
      "learning_rate": 2e-05,
      "loss": 1.5033,
      "step": 454
    },
    {
      "epoch": 0.8161434977578476,
      "grad_norm": 0.010540721006691456,
      "learning_rate": 2e-05,
      "loss": 1.5166,
      "step": 455
    },
    {
      "epoch": 0.8179372197309417,
      "grad_norm": 0.009755424223840237,
      "learning_rate": 2e-05,
      "loss": 1.5149,
      "step": 456
    },
    {
      "epoch": 0.8197309417040358,
      "grad_norm": 0.00984253827482462,
      "learning_rate": 2e-05,
      "loss": 1.5093,
      "step": 457
    },
    {
      "epoch": 0.8215246636771301,
      "grad_norm": 0.009836334735155106,
      "learning_rate": 2e-05,
      "loss": 1.5141,
      "step": 458
    },
    {
      "epoch": 0.8233183856502242,
      "grad_norm": 0.01032332144677639,
      "learning_rate": 2e-05,
      "loss": 1.5241,
      "step": 459
    },
    {
      "epoch": 0.8251121076233184,
      "grad_norm": 0.010635129176080227,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 460
    },
    {
      "epoch": 0.8269058295964126,
      "grad_norm": 0.009664127603173256,
      "learning_rate": 2e-05,
      "loss": 1.5052,
      "step": 461
    },
    {
      "epoch": 0.8286995515695067,
      "grad_norm": 0.010554889217019081,
      "learning_rate": 2e-05,
      "loss": 1.5071,
      "step": 462
    },
    {
      "epoch": 0.8304932735426009,
      "grad_norm": 0.009871057234704494,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 463
    },
    {
      "epoch": 0.8322869955156951,
      "grad_norm": 0.010431516915559769,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 464
    },
    {
      "epoch": 0.8340807174887892,
      "grad_norm": 0.009860005229711533,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 465
    },
    {
      "epoch": 0.8358744394618834,
      "grad_norm": 0.010233579203486443,
      "learning_rate": 2e-05,
      "loss": 1.5182,
      "step": 466
    },
    {
      "epoch": 0.8376681614349776,
      "grad_norm": 0.010311591438949108,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 467
    },
    {
      "epoch": 0.8394618834080717,
      "grad_norm": 0.010733729228377342,
      "learning_rate": 2e-05,
      "loss": 1.5186,
      "step": 468
    },
    {
      "epoch": 0.841255605381166,
      "grad_norm": 0.009951340965926647,
      "learning_rate": 2e-05,
      "loss": 1.5097,
      "step": 469
    },
    {
      "epoch": 0.8430493273542601,
      "grad_norm": 0.01003777701407671,
      "learning_rate": 2e-05,
      "loss": 1.5173,
      "step": 470
    },
    {
      "epoch": 0.8448430493273542,
      "grad_norm": 0.009939250536262989,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 471
    },
    {
      "epoch": 0.8466367713004485,
      "grad_norm": 0.009835812263190746,
      "learning_rate": 2e-05,
      "loss": 1.5272,
      "step": 472
    },
    {
      "epoch": 0.8484304932735426,
      "grad_norm": 0.010321546345949173,
      "learning_rate": 2e-05,
      "loss": 1.5193,
      "step": 473
    },
    {
      "epoch": 0.8502242152466367,
      "grad_norm": 0.01006554439663887,
      "learning_rate": 2e-05,
      "loss": 1.5165,
      "step": 474
    },
    {
      "epoch": 0.852017937219731,
      "grad_norm": 0.009972809813916683,
      "learning_rate": 2e-05,
      "loss": 1.5228,
      "step": 475
    },
    {
      "epoch": 0.8538116591928251,
      "grad_norm": 0.010388972237706184,
      "learning_rate": 2e-05,
      "loss": 1.5188,
      "step": 476
    },
    {
      "epoch": 0.8556053811659193,
      "grad_norm": 0.010111154057085514,
      "learning_rate": 2e-05,
      "loss": 1.5199,
      "step": 477
    },
    {
      "epoch": 0.8573991031390135,
      "grad_norm": 0.01029327604919672,
      "learning_rate": 2e-05,
      "loss": 1.516,
      "step": 478
    },
    {
      "epoch": 0.8591928251121076,
      "grad_norm": 0.010400544852018356,
      "learning_rate": 2e-05,
      "loss": 1.5218,
      "step": 479
    },
    {
      "epoch": 0.8609865470852018,
      "grad_norm": 0.0099885743111372,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 480
    },
    {
      "epoch": 0.862780269058296,
      "grad_norm": 0.010007279925048351,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 481
    },
    {
      "epoch": 0.8645739910313901,
      "grad_norm": 0.01053563691675663,
      "learning_rate": 2e-05,
      "loss": 1.5019,
      "step": 482
    },
    {
      "epoch": 0.8663677130044843,
      "grad_norm": 0.01031608134508133,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 483
    },
    {
      "epoch": 0.8681614349775785,
      "grad_norm": 0.010082092136144638,
      "learning_rate": 2e-05,
      "loss": 1.5073,
      "step": 484
    },
    {
      "epoch": 0.8699551569506726,
      "grad_norm": 0.01012254785746336,
      "learning_rate": 2e-05,
      "loss": 1.5101,
      "step": 485
    },
    {
      "epoch": 0.8717488789237668,
      "grad_norm": 0.010539901442825794,
      "learning_rate": 2e-05,
      "loss": 1.5209,
      "step": 486
    },
    {
      "epoch": 0.873542600896861,
      "grad_norm": 0.009883386082947254,
      "learning_rate": 2e-05,
      "loss": 1.5275,
      "step": 487
    },
    {
      "epoch": 0.8753363228699551,
      "grad_norm": 0.010055874474346638,
      "learning_rate": 2e-05,
      "loss": 1.521,
      "step": 488
    },
    {
      "epoch": 0.8771300448430494,
      "grad_norm": 0.010441599413752556,
      "learning_rate": 2e-05,
      "loss": 1.5253,
      "step": 489
    },
    {
      "epoch": 0.8789237668161435,
      "grad_norm": 0.010321282781660557,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 490
    },
    {
      "epoch": 0.8807174887892377,
      "grad_norm": 0.010404079221189022,
      "learning_rate": 2e-05,
      "loss": 1.5216,
      "step": 491
    },
    {
      "epoch": 0.8825112107623319,
      "grad_norm": 0.010680857114493847,
      "learning_rate": 2e-05,
      "loss": 1.5102,
      "step": 492
    },
    {
      "epoch": 0.884304932735426,
      "grad_norm": 0.009785238653421402,
      "learning_rate": 2e-05,
      "loss": 1.5152,
      "step": 493
    },
    {
      "epoch": 0.8860986547085202,
      "grad_norm": 0.010622934438288212,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 494
    },
    {
      "epoch": 0.8878923766816144,
      "grad_norm": 0.009563595987856388,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 495
    },
    {
      "epoch": 0.8896860986547085,
      "grad_norm": 0.009900403209030628,
      "learning_rate": 2e-05,
      "loss": 1.5254,
      "step": 496
    },
    {
      "epoch": 0.8914798206278027,
      "grad_norm": 0.010441206395626068,
      "learning_rate": 2e-05,
      "loss": 1.5042,
      "step": 497
    },
    {
      "epoch": 0.8932735426008969,
      "grad_norm": 0.010110273025929928,
      "learning_rate": 2e-05,
      "loss": 1.5141,
      "step": 498
    },
    {
      "epoch": 0.895067264573991,
      "grad_norm": 0.00976527575403452,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 499
    },
    {
      "epoch": 0.8968609865470852,
      "grad_norm": 0.010270185768604279,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 500
    },
    {
      "epoch": 0.8986547085201794,
      "grad_norm": 0.010477078147232533,
      "learning_rate": 2e-05,
      "loss": 1.5331,
      "step": 501
    },
    {
      "epoch": 0.9004484304932735,
      "grad_norm": 0.009786723181605339,
      "learning_rate": 2e-05,
      "loss": 1.5143,
      "step": 502
    },
    {
      "epoch": 0.9022421524663677,
      "grad_norm": 0.009838691912591457,
      "learning_rate": 2e-05,
      "loss": 1.5237,
      "step": 503
    },
    {
      "epoch": 0.9040358744394619,
      "grad_norm": 0.010305250994861126,
      "learning_rate": 2e-05,
      "loss": 1.5236,
      "step": 504
    },
    {
      "epoch": 0.905829596412556,
      "grad_norm": 0.010098317638039589,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 505
    },
    {
      "epoch": 0.9076233183856502,
      "grad_norm": 0.010335841216146946,
      "learning_rate": 2e-05,
      "loss": 1.519,
      "step": 506
    },
    {
      "epoch": 0.9094170403587444,
      "grad_norm": 0.009809168055653572,
      "learning_rate": 2e-05,
      "loss": 1.5176,
      "step": 507
    },
    {
      "epoch": 0.9112107623318386,
      "grad_norm": 0.01069081760942936,
      "learning_rate": 2e-05,
      "loss": 1.5055,
      "step": 508
    },
    {
      "epoch": 0.9130044843049328,
      "grad_norm": 0.009927291423082352,
      "learning_rate": 2e-05,
      "loss": 1.5224,
      "step": 509
    },
    {
      "epoch": 0.9147982062780269,
      "grad_norm": 0.010560589842498302,
      "learning_rate": 2e-05,
      "loss": 1.5129,
      "step": 510
    },
    {
      "epoch": 0.9165919282511211,
      "grad_norm": 0.010154438205063343,
      "learning_rate": 2e-05,
      "loss": 1.52,
      "step": 511
    },
    {
      "epoch": 0.9183856502242153,
      "grad_norm": 0.010346156544983387,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 512
    },
    {
      "epoch": 0.9201793721973094,
      "grad_norm": 0.010523281060159206,
      "learning_rate": 2e-05,
      "loss": 1.5187,
      "step": 513
    },
    {
      "epoch": 0.9219730941704036,
      "grad_norm": 0.010443002916872501,
      "learning_rate": 2e-05,
      "loss": 1.5059,
      "step": 514
    },
    {
      "epoch": 0.9237668161434978,
      "grad_norm": 0.010005362331867218,
      "learning_rate": 2e-05,
      "loss": 1.5102,
      "step": 515
    },
    {
      "epoch": 0.9255605381165919,
      "grad_norm": 0.010285025462508202,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 516
    },
    {
      "epoch": 0.9273542600896861,
      "grad_norm": 0.010401098988950253,
      "learning_rate": 2e-05,
      "loss": 1.5243,
      "step": 517
    },
    {
      "epoch": 0.9291479820627803,
      "grad_norm": 0.010455128736793995,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 518
    },
    {
      "epoch": 0.9309417040358744,
      "grad_norm": 0.00987928081303835,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 519
    },
    {
      "epoch": 0.9327354260089686,
      "grad_norm": 0.010212692432105541,
      "learning_rate": 2e-05,
      "loss": 1.5151,
      "step": 520
    },
    {
      "epoch": 0.9345291479820628,
      "grad_norm": 0.010937588289380074,
      "learning_rate": 2e-05,
      "loss": 1.5099,
      "step": 521
    },
    {
      "epoch": 0.9363228699551569,
      "grad_norm": 0.010248001664876938,
      "learning_rate": 2e-05,
      "loss": 1.5256,
      "step": 522
    },
    {
      "epoch": 0.9381165919282511,
      "grad_norm": 0.010430903173983097,
      "learning_rate": 2e-05,
      "loss": 1.5056,
      "step": 523
    },
    {
      "epoch": 0.9399103139013453,
      "grad_norm": 0.0102499695494771,
      "learning_rate": 2e-05,
      "loss": 1.5285,
      "step": 524
    },
    {
      "epoch": 0.9417040358744395,
      "grad_norm": 0.010674213990569115,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 525
    },
    {
      "epoch": 0.9434977578475336,
      "grad_norm": 0.010732615366578102,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 526
    },
    {
      "epoch": 0.9452914798206278,
      "grad_norm": 0.009994648396968842,
      "learning_rate": 2e-05,
      "loss": 1.5228,
      "step": 527
    },
    {
      "epoch": 0.947085201793722,
      "grad_norm": 0.010234368033707142,
      "learning_rate": 2e-05,
      "loss": 1.5258,
      "step": 528
    },
    {
      "epoch": 0.9488789237668162,
      "grad_norm": 0.010327205993235111,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 529
    },
    {
      "epoch": 0.9506726457399103,
      "grad_norm": 0.009836922399699688,
      "learning_rate": 2e-05,
      "loss": 1.5171,
      "step": 530
    },
    {
      "epoch": 0.9524663677130045,
      "grad_norm": 0.009962068870663643,
      "learning_rate": 2e-05,
      "loss": 1.5125,
      "step": 531
    },
    {
      "epoch": 0.9542600896860987,
      "grad_norm": 0.010127882473170757,
      "learning_rate": 2e-05,
      "loss": 1.5182,
      "step": 532
    },
    {
      "epoch": 0.9560538116591928,
      "grad_norm": 0.010251611471176147,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 533
    },
    {
      "epoch": 0.957847533632287,
      "grad_norm": 0.010081682354211807,
      "learning_rate": 2e-05,
      "loss": 1.5239,
      "step": 534
    },
    {
      "epoch": 0.9596412556053812,
      "grad_norm": 0.010235367342829704,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 535
    },
    {
      "epoch": 0.9614349775784753,
      "grad_norm": 0.009694702923297882,
      "learning_rate": 2e-05,
      "loss": 1.5174,
      "step": 536
    },
    {
      "epoch": 0.9632286995515695,
      "grad_norm": 0.010224996134638786,
      "learning_rate": 2e-05,
      "loss": 1.5171,
      "step": 537
    },
    {
      "epoch": 0.9650224215246637,
      "grad_norm": 0.010206632316112518,
      "learning_rate": 2e-05,
      "loss": 1.5223,
      "step": 538
    },
    {
      "epoch": 0.9668161434977578,
      "grad_norm": 0.010011864826083183,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 539
    },
    {
      "epoch": 0.968609865470852,
      "grad_norm": 0.010364921763539314,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 540
    },
    {
      "epoch": 0.9704035874439462,
      "grad_norm": 0.010109508410096169,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 541
    },
    {
      "epoch": 0.9721973094170404,
      "grad_norm": 0.00964987464249134,
      "learning_rate": 2e-05,
      "loss": 1.5089,
      "step": 542
    },
    {
      "epoch": 0.9739910313901345,
      "grad_norm": 0.010244207456707954,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 543
    },
    {
      "epoch": 0.9757847533632287,
      "grad_norm": 0.009797874838113785,
      "learning_rate": 2e-05,
      "loss": 1.5143,
      "step": 544
    },
    {
      "epoch": 0.9775784753363229,
      "grad_norm": 0.010056640952825546,
      "learning_rate": 2e-05,
      "loss": 1.5276,
      "step": 545
    },
    {
      "epoch": 0.979372197309417,
      "grad_norm": 0.009898710064589977,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 546
    },
    {
      "epoch": 0.9811659192825112,
      "grad_norm": 0.0099082225933671,
      "learning_rate": 2e-05,
      "loss": 1.5276,
      "step": 547
    },
    {
      "epoch": 0.9829596412556054,
      "grad_norm": 0.01018478162586689,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 548
    },
    {
      "epoch": 0.9847533632286996,
      "grad_norm": 0.009828625246882439,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 549
    },
    {
      "epoch": 0.9865470852017937,
      "grad_norm": 0.010311014950275421,
      "learning_rate": 2e-05,
      "loss": 1.5138,
      "step": 550
    },
    {
      "epoch": 0.9883408071748879,
      "grad_norm": 0.010840130038559437,
      "learning_rate": 2e-05,
      "loss": 1.5044,
      "step": 551
    },
    {
      "epoch": 0.9901345291479821,
      "grad_norm": 0.009595104493200779,
      "learning_rate": 2e-05,
      "loss": 1.5165,
      "step": 552
    },
    {
      "epoch": 0.9919282511210762,
      "grad_norm": 0.01027593482285738,
      "learning_rate": 2e-05,
      "loss": 1.5291,
      "step": 553
    },
    {
      "epoch": 0.9937219730941704,
      "grad_norm": 0.010394555516541004,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 554
    },
    {
      "epoch": 0.9955156950672646,
      "grad_norm": 0.00996735692024231,
      "learning_rate": 2e-05,
      "loss": 1.5212,
      "step": 555
    },
    {
      "epoch": 0.9973094170403587,
      "grad_norm": 0.010095257312059402,
      "learning_rate": 2e-05,
      "loss": 1.5106,
      "step": 556
    },
    {
      "epoch": 0.9991031390134529,
      "grad_norm": 0.01082176435738802,
      "learning_rate": 2e-05,
      "loss": 1.5099,
      "step": 557
    },
    {
      "epoch": 0.9991031390134529,
      "step": 557,
      "total_flos": 7841700554735616.0,
      "train_loss": 0.0,
      "train_runtime": 1.3623,
      "train_samples_per_second": 104762.356,
      "train_steps_per_second": 408.87
    }
  ],
  "logging_steps": 1,
  "max_steps": 557,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7841700554735616.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}